Frontiers of Propulsion Science

Frontiers of Propulsion Science

Edited by
Marc G. Millis
NASA Glenn Research Center
Cleveland, Ohio

Eric W. Davis
Institute for Advanced Studies at Austin
Austin, Texas

Volume 227
PROGRESS IN
ASTRONAUTICS AND AERONAUTICS

Frank K. Lu, Editor-in-Chief

University of Texas at Arlington
Arlington, Texas

Published by the
American Institute of Aeronautics and Astronautics, Inc.
1801 Alexander Bell Drive, Reston, Virginia 20191-4344
American Institute of Aeronautics and Astronautics, Inc., Reston, Virginia
1 2 3 4 5

Copyright # 2009 by the American Institute of Aeronautics and Astronautics, Inc. Printed in the United
States of America. All rights reserved. Reproduction or translation of any part of this work beyond that per-
mitted by Sections 107 and 108 of the U.S. Copyright Law without the permission of the copyright owner is
unlawful. The code following this statement indicates the copyright owner’s consent that copies of articles in
this volume may be made for personal or internal use, on condition that the copier pay the per-copy fee ($2.50)
plus the per-page fee ($0.50) through the Copyright Clearance Center. Inc., 222 Rosewood Drive, Danvers,
Massachusetts 01923. This consent does not extend to other kinds of copying, for which permission requests
should be addressed to the publisher. Users should employ the following code when reporting copying from
the volume to the Copyright Clearance Center:

978-1-56347-956-4=09 $2:50 þ :50

Data and information appearing in this book are for informational purposes only. AIAA is not responsible for
any injury or damage resulting from use or reliance, nor does AIAA warrant that use or reliance will be free
from privately owned rights.

ISBN 978-1-56347-956-4
 HE editors dedicate this book to Dr. Robert L. Forward and Sir Arthur C.
T Clarke. We honor these individuals for both the advances they made in
spaceflight and for the inspirational effect their work had on the rest of us.
Each was competent, visionary, and entertaining. Each risked topics beyond
the comfort zones of their peers and succeeded in making landmark progress.
Their works have become milestones in history. Additionally, their science
fiction helped provoke thought on a deeper level and entertained us in the
process. Without pioneers like these, this book would not exist. Their vision
and passion compels us to carry on, hopefully continuing the progress that will
one day allow humanity to travel amongst the stars.
 Progress in Astronautics and Aeronautics
Editor-in-Chief
Frank K. Lu
University of Texas at Arlington

Editorial Board

David A. Bearden Eswar Josyula

The Aerospace Corporation U.S. Air Force Research Laboratory

John D. Binder Gail A. Klein

viaSolutions Jet Propulsion Laboratory

Steven A. Brandt Konstantinos Kontis

U.S. Air Force Academy University of Manchester

Jose Camberos Richard C. Lind

U.S. Air Force Research Laboratory University of Florida

Richard Curran Ning Qin

Queen’s University of Belfast University of Sheffield

Sanjay Garg Oleg A. Yakimenko

NASA Glenn Research Center U.S. Naval Postgraduate School

Christopher H. Jenkins
Montana State University
 Foreword

S AN aerospace researcher, I appreciate the value and challenge of advan-

A cing revolutionary ideas. While my partners and I are opening near-Earth
spaceflight to citizens, the creators of this book are extending far beyond
that investigating how to enable interstellar flight.
This is the stuff of breakthroughs; those notions that sound crazy at first, but
actually lead to enormous improvements in the human condition. I love that
word, “breakthrough.” It means taking risks to explore what average researchers
consider nonsense, and then persevering until you’ve changed the world. While
most aerospace professionals play it safe with yawning improvements in technol-
ogy, the authors of this book take the risk of seriously considering ideas that pre-
sently look impossible. . . to some. From their own initiative they created this
first-ever technical book on star-drive science.
Although the topics of this book might sound like science fiction to the less
adventurous covering ideas like antigravity, space drives, warp drives, and
faster-than-light travel these goals are dealt with here as a rigorous scientific
inquiry. The authors identify what’s already been investigated with comparisons
to the foundations of physics, cover some dead-end approaches, and show where
next to focus attention to continue systematic, rigorous advancements. This is not
light reading. The details and citations are numerous. In short, this book offers the
seeds for undiscovered breakthroughs.
It is hoped that this book will inspire students and young professionals. In
these pages are the starting materials from which they can begin to make their
mark on history. And to help these future pioneers, the editors saw fit to
include a chapter on how to conduct such visionary work within typically
stodgy establishments. Lessons of prior breakthroughs are contrasted to the
demands of bureaucracies, with specific suggestions on how to make such
high-risk/high-gain research seem downright prudent.
While I continue to open spaceflight to the masses and NASA reaches back for
the moon, it is comforting to know that the scientists and engineers behind this
book are looking beyond current activities to answer “what comes after
that?” the breakthroughs that will take us to the stars.

Burt Rutan
Scaled Composites LLC
August 2008
 Table of Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvii

I. Understanding the Problem

Chapter 1. Recent History of Breakthrough Propulsion Studies . . . . 1

Paul A. Gilster, Tau Zero Foundation, Raleigh, North Carolina

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Challenging Initial Stages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Reflecting on a Time Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Advanced Propulsion Concepts Emerge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Vision 21 and Its Aftermath . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Wormholes and Warp Drive Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Quantum Tunneling at FTL Speeds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Podkletnov “Gravity Shield” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Considering the Quantum Vacuum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Institute for Advanced Studies at Austin . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Breakthrough Propulsion Physics Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Project Greenglow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
ESA’s General Studies Programme and Advanced Concepts Team . . . . . . . . . . . . 19
Broad Strategies for Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Continuing Research Projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Faster than Light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Chapter 2. Limits of Interstellar Flight Technology . . . . . . . . . . . . . 31

Robert H. Frisbee, Jet Propulsion Laboratory, California Institute of
Technology, Pasadena, California

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Challenge of Interstellar Missions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Fundamentals of Interstellar Flight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Rocket Based Propulsion Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Nonrocket (Beamed Momentum) Propulsion Options . . . . . . . . . . . . . . . . . . . . 79
Illustrative Missions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Most Critical Technologies and Feasible Engineering Limits . . . . . . . . . . . . . . . . 95
Closing Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

ix
x

Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Appendix: Derivation of the Classical and Relativistic Rocket Equations . . . . . . . 108
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

II. Propulsion without Rockets

Chapter 3. Prerequisites for Space Drive Science . . . . . . . . . . . . . . 127

Marc G. Millis, NASA Glenn Research Center, Cleveland, Ohio

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Major Objections and Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Estimating Potential Benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Hypothetical Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Next Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170

Chapter 4. Review of Gravity Control Within Newtonian and

General Relativistic Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Eric W. Davis, Institute for Advanced Studies at Austin, Austin, Texas

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Gravity Control Within Newtonian Physics . . . . . . . . . . . . . . . . . . . . . . . . . 176
Gravity Control Within General Relativity . . . . . . . . . . . . . . . . . . . . . . . . . . 184
Miscellaneous Gravity Control Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

Chapter 5. Gravitational Experiments with Superconductors:

History and Lessons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
George D. Hathaway, Hathaway Consulting Services, Toronto, Ontario, Canada

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Experimental Traps and Pitfalls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Historical Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
Summary and Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246

Chapter 6. Nonviable Mechanical “Antigravity” Devices . . . . . . . . . 249

Marc G. Millis, NASA Glenn Research Center, Cleveland, Ohio

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
Oscillation Thrusters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
 xi

Gyroscopic Antigravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254

Unrelated Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260

Chapter 7. Null Findings of Yamishita Electrogravitational Patent . . 263

Kenneth E. Siegenthaler and Timothy J. Lawrence, U.S. Air Force Academy,
Colorado Springs, Colorado

Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
Experiment 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
Experiment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291

Chapter 8. Force Characterization of Asymmetrical Capacitor

Thrusters in Air . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
William M. Miller, Sandia National Laboratories, Albuquerque, New Mexico; Paul
B. Miller, East Mountain Charter High School, Sandia Park, New Mexico; and
Timothy J. Drummond, Sandia National Laboratories, Albuquerque, New Mexico

Introduction . . . . . . . . . . . . ........ . . . . . . . . . . . . . . . . . . . . . . . . . 293

Summary of Theories . . . . . . ........ . . . . . . . . . . . . . . . . . . . . . . . . . 294
Overall Experiment Setup . . . . ........ . . . . . . . . . . . . . . . . . . . . . . . . . 295
Results . . . . . . . . . . . . . . . . ........ . . . . . . . . . . . . . . . . . . . . . . . . . 297
Discussion of Data as it Relates to Theories . . . . . . . . . . . . . . . . . . . . . . . . . 319
Conclusions . . . . . . . . . . . . ........ . . . . . . . . . . . . . . . . . . . . . . . . . 325
Acknowledgments . . . . . . . . ........ . . . . . . . . . . . . . . . . . . . . . . . . . 326
References . . . . . . . . . . . . . ........ . . . . . . . . . . . . . . . . . . . . . . . . . 326

Chapter 9. Experimental Findings of Asymmetrical Capacitor

Thrusters for Various Gasses and Pressures . . . . . . . . . . . . . . . . . . 329
Francis X. Canning, Simply Sparse Technologies, Morgantown, West Virginia

Introduction . . . . . . . . . . . . . . . . . . . ...... . . . . . . . . . . . . . . . . . . . . 329

Experimental Setup . . . . . . . . . . . . . . ...... . . . . . . . . . . . . . . . . . . . . 332
Qualitative Experimental Results . . . . . . ...... . . . . . . . . . . . . . . . . . . . . 333
Numerical Calculations of Electric Fields . ...... . . . . . . . . . . . . . . . . . . . . 336
Theories Versus Quantitative Experimental Results . . . . . . . . . . . . . . . . . . . . . 337
Vacuum Results . . . . . . . . . . . . . . . . . ...... . . . . . . . . . . . . . . . . . . . . 339
Conclusions . . . . . . . . . . . . . . . . . . . ...... . . . . . . . . . . . . . . . . . . . . 339
Acknowledgments . . . . . . . . . . . . . . . ...... . . . . . . . . . . . . . . . . . . . . 340
References . . . . . . . . . . . . . . . . . . . . ...... . . . . . . . . . . . . . . . . . . . . 340
xii

Chapter 10. Propulsive Implications of Photon

Momentum in Media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
Michael R. LaPointe, NASA Marshall Space Flight Center, Huntsville, Alabama

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
Electromagnetic Fields in Dielectric Media . . . . . . . . . . . . . . . . . . . . . . . . . . 345
A Century of Controversy: Theory, Experiment, and Attempts at Resolution . . . . 347
Propulsion Concepts Based on Electromagnetic Momentum Exchange . . . . . . . . 356
Feigel Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368

Chapter 11. Experimental Results of the Woodward Effect

on a Micro-Newton Thrust Balance . . . . . . . . . . . . . . . . . . . . . . . 373
Nembo Buldrini and Martin Tajmar, Austrian Research Centers GmbH ARC,
Seibersdorf, Austria

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
Theoretical Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
Thrust Balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376
Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388

Chapter 12. Thrusting Against the Quantum Vacuum . . . . . . . . . . . 391

G. Jordan Maclay, Quantum Fields LLC, Richland Center, Wisconsin

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Physics of the Quantum Vacuum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
Measurements of Casimir Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399
Space Propulsion Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402
Vibrating Mirror Casimir Drive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
Unresolved Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419

Chapter 13. Inertial Mass from Stochastic Electrodynamics . . . . . . . 423

Jean Luc Cambier, U.S. Air Force Research Laboratory, Edwards Air
Force Base, Edwards, California

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
 xiii

Stochastic Electrodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425

Quantum Electrodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
The Way Forward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
Appendix A: Relativistic Transformation to a Uniformly Accelerated Frame . . . . 442
Appendix B: Estimate of Rest Mass from Classical Interaction with ZPF . . . . . . . 445
Appendix C: Boyer’s Correlation Function
and Poynting Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 446
Appendix D: Unruh Davies and Classical Oscillator . . . . . . . . . . . . . . . .... 451
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 452

III. Faster-than-Light Travel

Chapter 14. Relativistic Limits of Spaceflight . . . . . . . . . . . . . . . . . 455

Brice N. Cassenti, Rensselaer Polytechnic Institute, Hartford, Connecticut

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
Principle of Special Relativity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456
Paradoxes in Special Relativity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
Empirical Foundations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
Relativistic Rockets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469

Chapter 15. Faster-than-Light Approaches in

General Relativity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
Eric W. Davis, Institute for Advanced Studies at Austin, Austin, Texas

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . ............... . . . . . 471

General Relativistic Definition of Exotic Matter and the Energy Conditions . . . . . 472
Brief Review of Faster than Light Spacetimes . . . . ............... . . . . . 484
Make or Break Issues . . . . . . . . . . . . . . . . . . . ............... . . . . . 489
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . ............... . . . . . 500
Acknowledgments . . . . . . . . . . . . . . . . . . . . . ............... . . . . . 502
References . . . . . . . . . . . . . . . . . . . . . . . . . . ............... . . . . . 502

Chapter 16. Faster-than-Light Implications of Quantum

Entanglement and Nonlocality . . . . . . . . . . . . . . . . . . . . . . . . . . . 509
John G. Cramer, University of Washington, Seattle, Washington

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509
Quantum Entanglement, Nonlocality, and EPR Experiments . . . . . . . . . . . . . . . 510
Quantum No Signal Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
Nonlocality vs Special Relativity? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
Momentum Domain Entanglement and EPR Experiments . . . . . . . . . . . . . . . . 514
Coherence Entanglement Complementarity . . . . . . . . . . . . . . . . . . . . . . . . . . 518
Nonlocal Communication vs Signaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
xiv

Superluminal and Retrocausal Nonlocal Communication . . . . . . . . . . . . . . . . . 522

Paradoxes and Nonlocal Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
Nonlinear Quantum Mechanics and Nonlocal Communication. . . . . . . . . . . . . . 524
Issues and Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525
Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
Appendix: Glossary and Description of Key Concepts . . . . . . . . . . . . . . . . . . . 526
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 528

IV. Energy Considerations

Chapter 17. Comparative Space Power Baselines . . . . . . . . . . . . . . 531

Gary L. Bennett, Metaspace Enterprises, Emmett, Idaho

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
Non nuclear Space Power Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535
Nuclear Power Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565

Chapter 18. On Extracting Energy from the

Quantum Vacuum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 569
Eric W. Davis and H. E. Puthoff, Institute for Advanced Studies at Austin,
Austin, Texas

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 569
Early Concepts for Extracting Energy and Thermodynamic Considerations . . . . . 570
Origin of Zero Point Field Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 574
Review of Selected Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577
Additional Considerations and Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 599

Chapter 19. Investigating Sonoluminescence as a Means of

Energy Harvesting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
John D. Wrbanek, Gustave C. Fralick, Susan Y. Wrbanek, and
Nancy R. Hall, NASA Glenn Research Center, Cleveland, Ohio

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 607
Challenges for Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609
Sonoluminescence at NASA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 610
Indications of High Temperature Generation . . . . . . . . . . . . . . . . . . . . . . . . . 617
Energy Harvesting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 625
Summary and Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 634
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 634
 xv

Chapter 20. Null Tests of “Free Energy” Claims . . . . . . . . . . . . . . . 639

Scott R. Little, EarthTech International, Austin, Texas

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 639
Testing of Energy Claims . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 640
Some Tests of Breakthrough Energy Claims . . . . . . . . . . . . . . . . . . . . . . . . . 641
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 648
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 648

V. From This Point Forward

Chapter 21. General Relativity Computational Tools and

Conventions for Propulsion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
Claudio Maccone

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
Recommended Propulsion Computational Conventions . . . . . . . . . . . . . . . . . . 652
Representative Problems in Propulsion Science . . . . . . . . . . . . . . . . . . . . . . . 655
Review of Existing Computational Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . 655
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 660
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 661
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 661

Chapter 22. Prioritizing Pioneering Research . . . . . . . . . . . . . . . . . 663

Marc G. Millis, NASA Glenn Research Center, Cleveland, Ohio

Introduction . . . . . . . . . . . . . . . . . . . . . . . . ........ . . . . . . . . . . . . . 663

Historical Perspectives . . . . . . . . . . . . . . . . . . ........ . . . . . . . . . . . . . 665
Combining Vision and Rigor . . . . . . . . . . . . . . ........ . . . . . . . . . . . . . 673
Research Project Operating Principles . . . . . . . . ........ . . . . . . . . . . . . . 681
Devising Prioritization Criteria . . . . . . . . . . . . . ........ . . . . . . . . . . . . . 686
NASA Breakthrough Propulsion Physics Research Solicitation . . . . . . . . . . . . . 689
From Individual to Overall Progress . . . . . . . . . ........ . . . . . . . . . . . . . 694
Final Lessons and Comparisons . . . . . . . . . . . . ........ . . . . . . . . . . . . . 699
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . ........ . . . . . . . . . . . . . 700
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . ........ . . . . . . . . . . . . . 701
References . . . . . . . . . . . . . . . . . . . . . . . . . ........ . . . . . . . . . . . . . 714

Subject Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 719

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 737

Supporting Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 739

 Preface

“Fortune favors the bold.”

Book X of The Aeneid by Virgil (19 B.C.E. )

I. Purpose
HIS book is the first-ever compilation of emerging science relevant to such
T notions as space drives, warp drives, gravity control, and faster-than-light
travel the kind of breakthroughs that would revolutionize spaceflight and
enable human voyages to other star systems. Although these concepts might
sound like science fiction, they are appearing in increasing numbers in reputable
scientific journals. The intent of this book is to provide managers, scientists,
engineers, and graduate students with enough starting material to comprehend
the status of this research and decide for themselves if and how to pursue this
topic in more depth.
As with any young topic, it can be difficult to comprehend the potential benefits,
the pros and cons of the competing approaches, and then decide what actions are
warranted. To that end, the editors have endeavored to collect impartial overviews
of the best-known and most relevant approaches. In many cases, dead-end lessons
are included to counter recurring claims and offer examples for how to assess such
claims. In addition, the methods for dealing with such pioneering topics are
included, both from the historical perspective and more specifically from the
lessons learned from NASA’s Breakthrough Propulsion Physics Project. It is
hoped that this volume will give future researchers the foundations to eventually
discover the breakthroughs that will allow humanity to thrive beyond Earth.
This research falls within the realm of science instead of technology, with the
distinction that science is about uncovering the laws of nature while technology is
about applying that science to build useful devices. Because existing technology
is inadequate for traversing astronomical distances between neighboring stars
(even if advanced to the limit of its underlying physics), the only way to circum-
vent these limits is to discover new propulsion science. In addition to their utility
for spaceflight, the discovery of any new force-production or energy-exchange
principles would lead to a whole new class of technologies. The implications
of success are profound.
Objectively, the desired breakthroughs might turn out to be impossible, but
progress is not made by conceding defeat. Breakthroughs have a habit of
taking pessimists by surprise, but can equally remain elusive. Although no break-
throughs appear imminent, enough progress has been made to provide the
groundwork for deeper studies, both with the science itself and with the program-
matic methods for tackling such provocative goals. If the history of scientific and
technological revolutions is any indication, this topic could one day eclipse fam-
iliar aerospace technology and enable humanity to travel to habitable planets
around neighboring stars. For now, however, the work is predominantly at
xvii
xviii

stages 1 and 2 of the scientific method; that is, defining the problem and collect-
ing data, with a few approaches already testing hypotheses.
Regardless of whether the breakthroughs are found, this inquiry provides an
additional perspective from which to seek answers to the lingering unknowns
of our universe. While general science continues to assess cosmological data
for its implication to the birth and fate of the universe, a spaceflight focus will
cast these observations in different contexts, offering insights that might other-
wise be overlooked from the curiosity-driven inquiries alone. Therefore, even
if there are no spaceflight breakthroughs to be found, adding the inquiry of space-
flight expands our ability to better understand the universe. The lessons learned in
the attempt will advance science in general.
The editors of this volume optimistically look forward to a time when this
first-ever technical book on space drives and faster-than-light travel becomes out-
dated. It will be interesting to look back decades from now to see what future dis-
coveries transpired and then compare them with the directions examined in these
chapters. Regardless of the book’s inevitable obsolescence, the editors hope that
the methods of continued discovery implicit in this book will have lasting impact.

II. Chapter by Chapter

An introduction to each chapter follows, so that the context of each can be
better understood. Consider each chapter to be a primer to its topic rather than
a definitive last word. There is much that could change with further research,
and in some cases different interim conclusions are found in different chapters.
Such divergent conclusions are a reflection of the embryonic state of this
research, where many unresolved issues still exist. In many cases, chapters
present ideas that are found to not work. From experience with the NASA Break-
through Propulsion Physics Project, it was found that dead-end approaches are
repeatedly researched because the null results were never distributed. Other chap-
ters describe work that is currently under investigation where more definitive
results are imminent. And lastly, much research on this topic has not even begun.
To convey the relative maturity of propulsion studies, a historical perspective
is offered in Chapter 1. Although the notions for breakthrough spaceflight have
been around for quite some time, the actual scientific publications started to
accumulate around 1995, with a few examples predating this by many years.
From 1996 through 2002, NASA funded work on these subjects through the
Breakthrough Propulsion Physics Project. During that same period, several
other organizations sponsored work of their own, including British Aerospace
Systems. Most of this research now continues under the discretionary time and
resources of individual researchers scattered across the globe.
It would not be appropriate to have a book about seeking new propulsion
science without first articulating the edge of interstellar flight technology, both
the foreseeable embodiments and their upper theoretical limits. Chapter 2 pro-
vides a thorough overview of the edge of interstellar technology, along with
details about optimum trajectory and mission planning. Comparisons between
the required mission energies and available terrestrial energy are provided to
convey the scale of the challenge. Methods to minimize trip time and vehicle
system mass are offered, including assessments of the impact of acceleration
 xix

and maximum cruise velocity. A number of technological options are described,

including their performance predictions, spanning the technology of light sails
through antimatter-annihilation rockets.
Presently, the scientific foundations from which to engineer space drives
propulsion that uses only the interactions between the spacecraft and its surround-
ing space do not exist. To help initiate the systematic search and assessment of
possibilities, Chapter 3 transforms the major objections to the notion of a space
drive into a problem statement to guide future research. The major objections are
the scarcity of indigenous reaction mass (for momentum conservation), and the
lack of known methods to impart net forces against such matter. By examining
these issues and the various forms of matter in the universe, and by examining
10 hypothetical space drives, a problem statement is derived. From simple
energy analyses, estimates for potential benefits and various analytical
approaches are identified. It is found that the very definitions of spacetime and
inertial frames warrant deeper research. When viewed in the context of space pro-
pulsion rather than general science, these questions present different research
paths that still have not been explored.
When it comes to moving spacecraft without using rockets or light sails, one
of the most commonly raised approaches is that of manipulating gravity.
Chapter 4 examines several ways of approaching this challenge, from simple
Newtonian concepts, General Relativity Theory, semi-classical Quantum
Gravity Theory, Quantum Field Theory, and others. Although it is possible, in
principle, to create or modify acceleration fields, the present theoretical
approaches require large (kilometer scale) and massive (solar-system scale)
devices operating at extremely high energy levels (relativistic). The chapter
also addresses the cosmological antigravity interpretations of dark energy, to
find that this does not lead to obvious propulsion opportunities. The details of
a wide variety of specific approaches are discussed, and key issues and unex-
plored research paths are identified.
Whereas Chapter 4 dealt with theoretical approaches, Chapter 5 examines
recent experimental approaches to create gravitational-like effects using supercon-
ductors. An introduction to the theoretical basis for the relationship between gravity
and superconductors from General Relativity is presented, followed by descriptions
of a variety of experimental complications that occur when exploring such notions.
These complications are included to better convey the difficulty of accurately dedu-
cing how nature works via low-temperature experiments. And finally, an overview is
provided of earlier and ongoing attempts to experimentally observe gravitational
effects with superconducting and low-temperature devices. Some experiments
have turned out to be dead-ends, while others are still under evaluation.
As much as it is important to identify the promising areas of future research, it
is also important to clearly state why some approaches will not work, especially
when such approaches are repeatedly suggested. This is the case with mechanical
devices that purport to create net forces or antigravity-like effects. Chapter 6
gives examples of very common devices, discusses why they might appear to
be breakthroughs, and explains why they are not. It also offers suggestions for
tests to provide convincing evidence of their operation, including methods that
are written at a level suitable for the independent researchers who frequently
propose similar devices. It is hoped that, by providing this information, other
xx

researchers who are asked to review such submissions can more quickly and
effectively respond.
Chapter 7 presents the findings of the Yamishita electrogravitational patent,
which is another null result that warrants reporting. The patent claim involves
coupling electrical charge with a rotating mass to produce a gravitational-like
force (which was not observed in these independent tests). The chapter also
shows how to inexpensively test such claims in a manner that helps educate
students on the methods of scientific inquiry. These experiments were conducted
as student projects at the U.S. Air Force Academy, in Colorado Springs,
Colorado.
A very common device promoted over the Internet is the “Lifter,” which has
numerous variants (Biefeld Brown, Asymmetrical Capacitor Thrusters, electro-
gravitics, etc.), some of which have existed for more than 80 years. This device
involves high-voltage capacitors that create thrust by interacting with the sur-
rounding air. Perhaps because of the ease of their construction and the scarcity
of rigorous publications on this phenomenon, many jump to the conclusion
that the effect is evidence of “antigravity.” Chapter 8 reports on careful measure-
ments of various capacitor configurations, itemizing the resulting correlations.
The final conclusion is that the observed effects are consistent with corona
wind, which can also be referred to as ion drift.
In addition to the comprehensive in-air tests of the previous chapter,
Chapter 9 reports on tests in nitrogen and argon at atmospheric pressure and
at various partial vacuums. Several other device geometries are examined also
as are thorough variations of polarity and ground connections. The combination
of grounding options, geometry, and polarity are found to affect thrust. Because
the thrust is found to be inversely proportional to the pressure, these findings
support the coronal wind conclusion of Chapter 8. More specifically, this
chapter concludes that the thrust is from the charged ions leaking across the
capacitor that undergo multiple collisions with air, transferring momentum to
neutral air molecules in the process.
Whereas the momentum of a photon in vacuum is well understood, the
momentum of a photon passing through dielectric media has generated signifi-
cant debate, beginning with the Abraham Minkowski controversy of 1908. In
particular, the Abraham formulation of the electromagnetic stress tensor in a
dielectric medium predicts a photon momentum that differs from the Minkowski
formulation, and experimental tests have not yet been able to resolve which per-
spective is correct. This controversy has entered the realm of breakthrough
spaceflight, with more than one device proposed to create net thrust via mechan-
isms embodied by this ambiguity (Corum, Brito). Chapter 10 reviews the under-
lying physics of photon momentum in dielectric media and the potential
propulsion implications. The most critical make-or-break issues are identified,
as are the conceptual and operational difficulties associated with known exper-
iments. In short, it is found that both the Abraham and Minkowski formulations
match experiments, with the caveat that the assumptions and conventions of each
formulation must be applied consistently when analyzing the entire system.
Because some of these details are subtle, it is easy to reach misleading con-
clusions if one errantly uses portions of each formulation to analyze a given
problem.
 xxi

James F. Woodward, a science professor at California State University, has

been experimenting with a technique to induce net thrust using a particular
interpretation of Mach’s principle a principle that deals with the very definition
of an inertial frame. Chapter 11 reports independent tests of this propulsion
concept, comparing it with previous experimental claims. Two devices that
were developed and tested by Woodward were independently tested using a sen-
sitive thrust balance developed for field-emission electric thrusters. The results
do not seem to be in full agreement with the findings claimed by Woodward
and collaborators. Nevertheless, the importance of such a discovery is sufficient
to recommend continuing experimentation to reach a complete understanding of
the phenomenon. Even if this effect is found not to occur, the issues raised by
Woodward’s approach offer several provocative questions for deeper
investigation.
Quantum electrodynamics theory, whose predicted effects have been verified
to 1 part in 10 billion, predicts that the lowest energy state of the electromagnetic
field still contains energy that can produce forces between nearby surfaces.
These Casimir forces have been measured and found to agree with predictions.
Chapter 12 examines whether this quantum vacuum might be exploited to
propel a spacecraft. Restrictions resulting from the conservation of energy and
momentum are discussed. A propulsion system based on an uncharged, conduct-
ing, mirror that vibrates asymmetrically in the vacuum produces real photons
that create thrust. Even though the thrust is even less than a photon rocket,
this action demonstrates that the vacuum can be used for propulsion. Techno-
logical improvements, some of which are proposed, may increase the accelerat-
ing force. Many questions remain about the supporting theory, and further
experiments are needed to probe questions about the quantum vacuum that are
far beyond current theory.
Controversial theories exist in the peer-reviewed literature that assert that
inertia is a side effect of accelerated motion through quantum vacuum fluctu-
ations. Chapter 13 examines the constructs of one of these theories to identify
the critical make-or-break issues and opportunities for discriminating tests.
Although the specific approach examined in this chapter has several shortcom-
ings, the more general notion of the connection between inertia and quantum
fields remains open for deeper investigations.
To open the next section on faster-than-light travel, it is appropriate to have a
chapter on Special Relativity, the theoretical tool for describing the consequences
of hyper-fast motion through spacetime. Experimental observations that support
special relativity are described in Chapter 14. The chapter outlines the basic
groundwork of special relativity and then proceeds to show the paradoxes
created if faster-than-light travel were allowed within spacetime. Paradoxes are
presented and shown to be primarily concerned with our concept of time.
Depending on how time is defined, it may be possible to resolve all of these para-
doxes. The other caveat to faster-than-light travel is the issue of spacetime itself.
Although faster-than-light motion is clearly a problem within spacetime, the situ-
ation is different when toying with manipulating spacetime. Those manipulations
are the subject of the next chapter.
Whereas Special Relativity forbids faster-than-light travel within spacetime
(or at least sets interesting constraints upon motion), the situation is different
xxii

in General Relativity where spacetime itself can be manipulated. Chapter 15

examines the variety of theoretically postulated faster-than-light schemes
(warp drives and wormholes in particular) to identify the energy requirements
and make-or-break issues. Many of the related issues are provocative topics in
their own right in science, and when viewed from the point of view of hyper-
fast travel, present interesting approaches toward their resolution. It is shown
that many of the energy restrictions commonly imposed are conventions rather
than physical absolutes. The possibility of testing some elements of theory is
raised, in particular the use of extreme electric and magnetic field strengths poss-
ible with ultrahigh-intensity lasers.
General Relativity is not the only branch of science with curious issues of
faster-than-light phenomenon. Quantum Theory has its own set of provocative
topics that are commonly, and sometimes errantly, associated with faster-than-
light travel. Chapter 16 reviews such quantum “nonlocality” effects and articu-
lates numerous experiments and their interpretations. One of the prime complex-
ities is having exacting definitions of what is meant by “entanglement” and
“nonlocality” and how these compare to the light-speed constraints of Special
Relativity. These complexities are explained in the chapter, and it is shown
that, in many situations, there are no conflicts. The chapter also shows that the
assumptions that lead to the “no-signal theorems” warrant close inspection
when assessing their applicability to some contemporary experiments. Under
certain circumstances, situations leading to retrocausal paradoxes can be con-
ceived where the effect precedes the cause. These situations are described, but
no resolutions exist at this point.
To open the next section about energy implications, Chapter 17 reviews the
past and projected technology that is based on accrued science. Numerous
devices are reviewed, including multiple power sources and conversion
methods and estimates of their upper performance limits.
The phenomenon of quantum fluctuations and its high-energy density as
calculated from some estimates has led some to ponder if the quantum
vacuum can be tapped as an energy source. Chapter 18 examines a span of
both theoretical and experimental approaches. It concludes that energy con-
versions between vacuum fluctuations and tangible effects are possible in prin-
ciple (albeit small), but that continuous energy extraction appears problematic
on the basis of the current understanding of quantum electrodynamics theory.
Specific experimental approaches are described for investigating these issues
directly.
Sonoluminescence, which is light generated from acoustic cavitation in fluids,
has been associated with claims of energy production. To provide grounding in
this topic, Chapter 19 covers recent experimental methods and findings. Instru-
mentation techniques that measure optical, radiation, and thermal properties of
the phenomenon of sonoluminescence are described. Initial efforts are directed
to the generation and the imaging of sonoluminescence in water and solvents.
Evidence of high-energy generation in the modification of thin films from sono-
luminescence is seen in heavy water but not seen in light water. The attainable
energy output was found to be less than the energy input for the experiments dis-
cussed. Improvements for realizing fusion processes and energy harvesting are
 xxiii

suggested. This phenomenon also serves as an empirical path to further under-

stand the interaction of sound, light, and fluid physics.
In much the same way that it is important to publish the space drive
approaches that produced null findings, Chapter 20 presents experimental find-
ings of a variety of energy conversion devices purported to produce more energy
than they consume. Although such concepts can be summarily dismissed for vio-
lating thermodynamic laws, the prospect that there might be genuine unobvious
energy sources has led many to suggest such devices. This chapter details the
experimental methods and experiences derived from testing such claims to
help other researchers become alert to the complexities and possible pitfalls of
related experiments.
To address challenges of breakthrough spaceflight within the sciences of
General Relativity and Quantum Field Theory, computer computational tools
are required. Calculations in these disciplines are extensive and involve
complex notational conventions. The time to do these calculations manually is
prohibitive, as is the risk of inducing transcription errors. Chapter 21 reviews
the main tensor calculus capabilities of the three most advanced and commer-
cially available “symbolic manipulator” tools. It also examines different conven-
tions in tensor calculus and suggests conventions that would be useful for space
propulsion research.
Based on lessons from history and the NASA Breakthrough Propulsion
Physics Project, suggestions are offered in Chapter 22 for how to comparatively
assess the variety of approaches toward revolutionary spaceflight. Perspectives of
prior scientific revolutions are examined in the context of seeking spaceflight
revolutions. The intent is to identify techniques to improve the management of
such visionary research. Methods tested in the course of the Breakthrough Pro-
pulsion Physics Project are provided, including details from its one formal
cycle of research solicitations. Key recommendations are to rigorously contrast
existing foundations of science with the visionary goals of breakthrough space-
flight, and to dissect the grand challenges into more approachable, short-term
research objectives.

III. Snapshot of the Book

Following the principles of a “traceability map,” as discussed in Chapter 22,
Fig. 1 shows a relational depiction of the contents of this book. In the far left
column of this map, the relevant disciplines of science are listed. In the far
right column, the visionary goals are listed and broken down into categories of
different approaches. From there, these categories branch out into numerous con-
cepts and devices. In the columns between the foundations of science and these
concepts and devices, there are various unknowns, critical issues, and curious
effects.
Because of printing limitations, the connecting lines between these various
facets cannot be shown, but many of the connections can be inferred by noting
the numbers in the parallelograms attached to the blocks. These numbers refer
to the chapters in which the specific items are discussed. From this representation
it is obvious that a given concept is connected to many issues and that there are
xxiv

Fig. 1 Relational depiction of this book’s contents.

issues that pertain to many concepts. In an expanded version of this map it would
be possible to see how all of the items are interconnected. The items having more
connections would be those that are more relevant to other facets. Another detail
of the figure is that the shaded blocks refer to null tests or to concepts that are
shown analytically not to be viable. All of the rest of the subjects are open for
further study.
 xxv

IV. Taking the Next Step

As stated earlier, this book should not be interpreted as the definitive last word
on the topic of seeking spaceflight breakthroughs. Instead, it is a primer to get
started. A variety of known approaches have been reviewed, including numerous
dead ends to avoid. For those researchers who are contemplating investing in this
field, we hope that this book provides you with the foundation necessary to make
genuine progress.

Marc G. Millis
Eric W. Davis
November 2008
 Acknowledgments

HANKS are due to the AIAA Nuclear and Future Flight Propulsion Techni-
T cal Committee for helping support the development of this book, in particular
the participation of the following three authors: Jean-Luc Cambier, Brice
N. Cassenti, and Robert H. Frisbee; and for providing the following reviewers:
Dana Andrews, Brice Cassenti, Allen Goff, Jochem Hauser, and Pavlos
Mikellides. Thanks are due to all the other authors as well; Gary L.
Bennett, Nembo Buldrini, Francis X. Canning, John G. Cramer, Timothy
J. Drummond, Gustav C. Fralick, Paul A. Gilster, Nancy R. Hall, George
D. Hathaway, Michael R. LaPointe, Timothy J. Lawrence, Scott R. Little,
G. Jordan Maclay, Claudio Maccone, Paul B. Miller, William H. Miller,
H. E. Puthoff, Kenneth E. Siegenthaler, Martin Tajmar, John D. Wrbanek, and
Susan Y. Wrbanek; and my co-editor and contributing author, Eric W. Davis.
And what publication is possible without the help of reviewers for their time
and insights? Thanks are owed to the following additional reviewers: Peter
Badzey, Gary Bennett, Dan Cole, John Cole, Steve Cooke, Roger Dyson, Ron
Evans, Larry Ford, Peter Garretson, Bernard Haisch, Nancy Hall, Don
Howard, Andrew Ketsdever, Geoff Landis, Mike Lapointe, Carolyn Mercer,
William Meyer, William Miller, H. E. Puthoff, Alfonso Rueda, William
Saylor, Roy Sullivan, Alexandre Szames, Martin Tajmar, Jeff Wilson, Clive
Woods, Jim Woodward, and Ed Zampino.

Marc G. Millis
Eric W. Davis
November 2008

xxvii
 Chapter 1

Recent History of Breakthrough

Propulsion Studies

Paul A. Gilster
Tau Zero Foundation, Raleigh, North Carolina

I. Introduction
HE knowledge that it is impossible to reach even the nearest stars in a single
T human lifetime using chemical propulsion technologies has launched an
active search for alternatives. Interstellar flight is not beyond reach, as Robert
Forward understood, a fact that prompted the researcher to conduct a series of
studies advocating propulsion methods that worked within the known laws of
physics, such as laser-pushed “lightsails” and rockets driven by antimatter reac-
tions. But beginning in the early 1990s, a second track of studies into what can be
called “breakthrough propulsion” began to emerge in workshops, papers, and
organizational efforts within NASA [Breakthrough Propulsion Physics (BPP)
Project], ESA, and private industry (e.g., BAE Systems Project Greenglow).
Hoping to uncover new physical principles of motion and energy, all have ener-
gized research while, as is common when working at the frontiers of knowledge,
encountering ideas that remain controversial.
This chapter examines the recent history of advanced propulsion as a dis-
cipline. Its goal is to place the contents of this book in the context of the
broader progression of science and technology. The ideas under investigation
are remarkable in their range, and engage the community to establish new
methodologies for dealing with theoretical and experimental work at the edge
of known physics. They include so-called “warp drive” theories involving the
manipulation of space-time itself, as well as wormhole concepts that, like the
warp drive, offer fast transit without violating Einstein’s Special Relativity.
Also under study: vacuum fluctuation energy, quantum tunneling, and the coup-

Copyright # 2008 by the American Institute of Aeronautics and Astronautics, Inc. All rights
reserved.

Lead journalist.

1
2 P. A. GILSTER

ling of gravity and electromagnetism. While the theoretical literature spawned by

these concepts has grown, so has the appreciation that their study must occur
within the rigorous constraints of conventional physics while remaining open
to experimental results that may extend our understanding of those principles.
Thus breakthrough propulsion research requires a balancing act, insisting upon
scientific rigor while examining concepts that are all too easily sensationalized
and misunderstood.
The need for continuing review of recent work in this field becomes apparent
when surveying the resources available to researchers. In the late 1970s and early
1980s, Robert Forward and astronautical engineer Eugene Mallove published
several bibliographies of interstellar studies in the pages of the Journal of the
British Interplanetary Society. Their 1980 effort demonstrates the range of
work being done on these topics even then, including 2700 items filed under
seventy subject categories. Forward and Mallove eventually had to abandon
their updates due to the labor involved; unfortunately, this valuable service has
not been resumed by others and remains unaddressed. Regarding the smaller
topic of emerging propulsion physics literature, no such bibliography has yet
been initiated. In the absence of such a resource, we can begin by surveying
the growth of these studies over the past two decades in the hope of more
methodical updates to come. A précis attempting an overview of concepts now
under active investigation and noting those likely to be discarded follows.

II. Challenging Initial Stages

The broader questions remain: Has enough progress been made from which
starting points for meaningful research exist? And are studies that seek break-
through physics for propulsion justified if they seem to violate commonly
accepted principles of physics such as conservation of momentum? As will be
shown in this historical overview and the content of subsequent chapters,
enough progress has been made to provide starting points for deeper research.
It is too soon, however, to predict whether any breakthroughs actually exist to
be discovered. Regarding the issue of possibly violating accepted physics, the
real challenge is to find those approaches that do not violate well-established
physical laws while also rigorously challenging provisional hypotheses or
extending research to where it has not yet been taken. In other words, this field
of study does not aim to violate physics, but rather to further extend physics.
An issue related to the appearance of violating physics is whether there is
value in pursuing breakthroughs that might not exist. Consider first the value
of curiosity-driven physics, for which no application goals guide the pursuit.
The value is in what is learned along the way. So too is the case of pursuing
breakthroughs that might not exist the minimum value is what is learned
along the way. Granted, if the research does yield the desired breakthroughs
the value will be still higher, but even the pessimistic outcome presents the
same learning opportunities as research that is unbound by application goals.
This caveat remains: For the research to be of value it must adhere to the
same high standards as curiosity-driven physics. Unfortunately, given the
allure of the grand goal of star flight, where the stakes are higher for humanity,
it is common to encounter sensationalistic work where premature claims are
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 3

made without any rigor to back them up. From the statistics compiled in the
course of NASA’s BPP Project [1], roughly 10% of the correspondence came
from proponents claiming breakthroughs with no credible evidence. This beha-
vior can taint the topic and amplify the concerns of other professionals. Again,
to make genuine progress, the emphasis is on the rigor and reliability of the
research, rather than on the magnitude of the claims. With that priority under-
stood, this topic does offer opportunities for learning more about the workings
of our Universe.
Recent developments, particularly those relating to the presence of unseen
dark matter and the dark energy force driving the accelerating expansion of the
universe, illustrate how engineering applications ultimately fall back upon theor-
etical physics for their foundation. But a focus on applications can clarify theory.
Seen through the lens of spaceflight, the anomalous galactic rotations that led to
the inference of dark matter become significant not only because of their perti-
nence toward understanding the origin and fate of the universe, but also
because they suggest the possibility of an indigenous reaction mass in interstellar
space. Dark energy, discovered through anomalous recessional redshifts, raises
the question of manipulating gravity. Thus the definition of a problem, the essen-
tial first step in scientific methodology, provokes different lines of inquiry
depending upon the objective of the study being performed. Multiple approaches
raise the chances of resolving underlying issues in physics.

III. Reflecting on a Time Line

To help convey the stage within which this work is set, Table 1 presents a time
line of events. The left side of the table lists general advances in physics and books
about scientific progress, while the right side lists publications and events that are
focused more specifically on practical interstellar flight. A key point is that, while
general science continues to evolve and reflect on its progress, the earliest works
pointing toward new space propulsion science are just beginning to emerge. There
are no publications indicating certain breakthroughs yet, but several theoretical
constructs are introduced. Also, the time line reflects the fact that surveys of the
options and initial research tasks have commenced. The topic is not, however,
at the point where progress is widespread or focal approaches have emerged.
The discipline of breakthrough propulsion is still in the early stages of emergence,
ripe with opportunities for conducting pioneering work.
It is interesting to note how much time passed before an international confer-
ence on General Relativity took place after the appearance of Einstein’s theory. It
is also interesting to note the timing of the publication of books reflecting on the
progress of science and technology, in particular, Kuhn [2], Foster [3], Horgan
[4], and Dyson [5].

IV. Advanced Propulsion Concepts Emerge

Although speculations on space travel and the demands of interstellar flight
stretch back to the Renaissance and even earlier, the modern study of advanced
propulsion may be said to have begun with a lecture given by Eugen Sänger at the
Fourth International Astronautical Congress [6]. Sänger’s investigations into
 4
Table 1 Time line of pertinent events

General science events Year Propulsion goal-specific events

Einstein’s General Relativity Theory 1916

Term “quantum mechanics” coined by Max Born 1924
Schrödinger develops wave mechanics 1926
Heisenberg’s Uncertainty Principle 1927
Quantum electrodynamics 1947 Ackeret, relativistic rocket analysis
De Broglie, The Revolution in Physics 1953 Sänger, photon rockets studied
1954
First international conference on General Relativity 1955
1956
Bondi, “Negative Mass in General Relativity” 1957
Bohr, Atomic Physics and Human Knowledge 1958
Heisenberg, Physics and Philosophy
1959
Maiman develops the first practical laser 1960
Aharonov-Bohm effect observed
P. A. GILSTER

1961
Kuhn, Structure of Scientific Revolutions (paradigm 1962
shifts)
1963 Forward, “Guidelines to Antigravity”
1964
Cosmic microwave background radiation observed 1965
1966
1967
1968
John Wheeler coins the term “black hole” 1969
String Theory born 1970 British Aircraft Corporation’s Future Concepts Group
1971
1972
Misner, Thorne, Wheeler, Gravitation 1973
1974
1975
1976
1977
1978
1979
1980
1981
Aspect, experimental test of Bell’s inequalities 1982
1983
1984
1985
Foster, Innovation: The Attacker’s Advantage 1986
1987
1988 Morris and Thorne on traversable wormholes
Forward studies negative matter propulsion
Forward, Future Magic
Barbour, Mach, “Absolute or Relative Motion” 1989 Matloff, The Starflight Handbook
1990 BAe Gravity Control Workshop
NASA GRC Vision-21 Workshop No.1
1991 U.S. AFRL: Tally, on Biefeld-Brown effect
1992 Podkletnov claims “gravity shield”
1993
Milloni, The Quantum Vacuum 1994 Alcubierre on warp drive
Haisch studies inertia as ZPE drag
Woodward, propulsion patent no.1
NASA JPL Workshop on quantum and FTL travel

(continued )
RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES
5
 6

Table 1 Time line of pertinent events (continued)

General science events Year Propulsion goal-specific events

1995 Visser, Lorentzian Wormholes

Cramer gives visual clues to search for wormholes
Krauss, Physics of Star Trek
Horgan, The End of Science 1996
Dyson, Imagined Worlds (new tools lead to 1997 Millis provides space drive problem definition
revolutionary discoveries) NASA, Breakthrough Propulsion Physics (BPP) Project workshop
Discovery of accelerated expansion 1998 Krasnikov tubes
1999 NASA BPP research solicitation
2000
2001 NASA BPP Project findings presented at Joint Propulsion
Conference
2002 Puthoff on the polarizable vacuum and propulsion (Journal of the
British Interplanetary Society)
P. A. GILSTER

2003 Hathaway dismisses Podkletnov

Europen Space agency’s Gravity Control study
2004 Gilster, Centauri Dreams
NASA BPP Project contractor reports published
Maclay and Forward, gedanken ZPE propulsion
2005 Matloff, Deep Space Probes
2006 NASA BPP Project final findings published
2007 BIS warp drive symposium
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 7

antimatter rocketry led to a so-called “photon rocket” design that would use the
gamma rays produced by the annihilation of electrons and positrons for thrust.
Although the idea foundered on the inability to direct the gamma ray exhaust
stream, the work led to another Sänger study of interstellar propulsion possibili-
ties in 1956 [7]. Other significant papers of the same era include those of Ackeret
[8], von Hoerner [9], and Shepherd [10], all of which examined the issues raised
by relativistic rocket flight.
The Air Force Office of Scientific Research began a series of symposia in 1957
that investigated the status of basic and applied research into advanced propul-
sion. These gatherings, held every two to three years and lasting until the
1970s, resulted in published proceedings whose papers addressed breakthrough
propulsion concepts considered futuristic but within the current paradigm of
physics. The year 1975 would see NASA’s earliest research into these concepts
and the publication of its ensuing technical report [11]. A major figure in early
studies was Robert Forward, who researched advanced propulsion while
working at the Hughes Research Laboratory in Malibu, California. Much of
Forward’s work was published in the form of research study reports for the Air
Force under contract to Franklin B. Mead Jr. at the Propulsion Directorate of
the Air Force Research Laboratory at Edwards Air Force Base, California. For-
ward’s popular book, Future Magic: How Today’s Science Fiction Will Become
Tomorrow’s Reality, reviewing the breakthrough propulsion concepts he had
examined for interstellar flight, appeared in 1988, bringing these studies to a
general audience [12].

V. Vision-21 and Its Aftermath

One recent attempt to reassess the prospects for significant advances in
spaceflight methods was Vision-21, a volunteer activity within NASA’s Lewis
Research Center (now Glenn Research Center) in Cleveland, Ohio. While indi-
vidual researchers studied such questions, largely in their spare time and
through occasional workshops with formal support [13 17] an organizational
impetus began to emerge at a symposium held at this NASA center in 1990,
dubbed “Vision-21: Space Travel for the Next Millennium.” Roughly half the
contributed papers focused on propulsion, with all papers (other than keynote
addresses) presented in poster format to maximize the interaction between speak-
ers and audience [18]. Contributors were charged with examining possible tech-
nologies as they might emerge, not just within decades but over the course of the
next 1000 years. Although some ideas we would now consider under the heading
of breakthrough propulsion did appear both in papers and panel discussion,
including the possibility of space-coupling (reacting with spacetime itself to
produce propulsive forces) [19], the bulk of these papers reflected current
thinking in physics taken to its engineering extreme through concepts such as
laser-pushed lightsails, fusion, and antimatter propulsion.
The significance of the conference, however, lay in the scientists behind it.
Working under the Vision-21 rubric, a group of researchers at NASA Lewis
had met regularly for several years before the meeting, and were encouraged
to develop ideas outside NASA’s current technological sphere. These scientists
and engineers developed methodologies that would later coalesce in the BPP
8 P. A. GILSTER

Project. But in 1990, the arrival of almost 200 guests from other NASA centers as
well as universities, research facilities, and businesses flagged the growing inter-
est in building a research process that could tackle concepts at the very edge of
physics. These were ideas too speculative to trigger official sponsorship, and thus
were in need of an alternate track that could proceed at least informally.

VI. Wormholes and Warp Drive Physics

Vision-21 occurred in the interval between two publications that galvanized
the nascent field and suggested fundamentally new areas for investigation. Kip
Thorne and (then) graduate students Michael Morris and Uli Yurtsever proposed
a novel means of fast transit through engineered wormholes that could be devel-
oped by sufficiently advanced technological cultures [20]. The wormhole acts as
a hyperspace tunnel through which travelers pass, with no need for faster than
light travel because the wormhole connects different regions within the universe,
different universes, or even different times. To create and maintain the wormhole
geometry, a wormhole with a one-meter radius required an amount of negative
energy comparable to the mass of the planet Jupiter [21,22]. Continuing worm-
hole research examines the stability of such “shortcuts” through spacetime and
the power required to keep wormholes open. But one effect of the early papers
of Thorne et al. was to put into high relief the understanding that fast interstellar
travel would demand propulsion strategies drawn from physical principles only
now being investigated.
Breakthrough propulsion began to emerge as a discipline in its own right,
given further impetus by the work of theoretical physicist Miguel Alcubierre.
Noting that spacetime itself was not bound to the same speed of light constraint
as objects within it (a nod to inflationary theories of the early cosmos), the latter
argued that the problems of special relativity (and inertia) could be resolved by
modifying spacetime so that its volume expanded behind the starship while
being compressed in front of it. The vessel itself, moving in a “bubble” of
normal space and never surpassing the speed of light, would be carried to its des-
tination, much like a surfer on a wave [23]. With Alcubierre’s 1994 paper in
place, the stage was set for the broader institutional investigation of issues that
had until recently seemed like nothing more than escapist television fare. The
physicist’s words on the matter set the tone. Describing the effects on ship and
crew as, never exceeding the speed of light, they are taken to their destination:
“The spaceship will then be able to travel much faster than the speed of light.
However, as we have seen, it will always remain on a timelike trajectory, that
is, inside its local light-cone: Light itself is also being pushed by the distortion
of spacetime. A propulsion mechanism based on such a local distortion of space-
time just begs to be given the familiar name of the ‘warp drive’ of science fiction”
[23]. The science fiction connection has continued to bedevil breakthrough pro-
pulsion studies since, though it must be noted that the genre’s uses in promoting
unfettered looks at imaginative ideas have captivated more than a few physicists.
Although rarely accurate at technical prediction, science fiction is known to have
inspired careers and even particular research paths [24].
Space drive studies now began to take form. Sergei Krasnikov, seeing that the
interior of the Alcubierre “bubble” would be causally disconnected from its
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 9

exterior, introduced an alternative strategy, the creation of a negative energy

“tunnel” at sublight speeds. Reaching their destination, the crew could return
through the same tunnel to the point of origin at speeds faster than light [25].
Such emerging ideas grew within a broader context; i.e., the understanding
that gravity and electromagnetism are related phenomena within a coupled space-
time [26]. This being the case, the possibility of manipulating gravity or inertia
through electromagnetism comes into play [27]. Energy demands placed many
such concepts in deep theoretical waters. Both the Thorne and Alcubierre inves-
tigations showed that manipulation of spacetime required matter with a negative
energy density. The Alcubierre starship would require positive energy to contract
spacetime in front of the vessel and negative energy to expand it behind. As it is
unknown whether negative energy densities can exist, propulsion studies could
only focus on the theoretical, with alarming results. Michael Pfenning and
Larry Ford were soon to calculate that the Alcubierre drive apparently demanded
more energy than was available in the entire universe to make it function [28].
Subsequent work would reduce that energy requirement considerably [29] but
massive amounts of negative energy were still demanded, not only by the Alcu-
bierre warp drive concept but also by wormhole physics, the wormhole’s opening
drawing on its effects to remain stable. The most current estimates of the energy
required for such notions are presented in Chapter 15.

VII. Quantum Tunneling at FTL Speeds

If the warp drive triggered a small burst of media interest in breakthrough
propulsion, the attention of scientists and lay people alike was also drawn to
the odd phenomenon of photons tunneling through barriers at velocities appar-
ently higher than the speed of light in a vacuum. Experiments conducted in
the early 1990s by Aephraim M. Steinberg et al. detected photons tunneling
through a mirror barrier at 1.7 times the speed of light [30]. The researchers
were quick to note that their results did not violate causality because the effect
could not be used to carry information, but explanations of wave packet behavior
could not constrain a growing interest in the phenomenon, which would be the
subject of subsequent investigations (to be discussed later) by both NASA and
the European Space Agency (ESA). Chapter 16 addresses these issues directly.

VIII. Podkletnov “Gravity Shield”

Public interest was also fired by the possibility of manipulating the force
of gravity, the ultimate prize being an “antigravity” effect that might produce
new spacecraft designs, not to mention radically transforming all other
forms of transportation. Two years before Alcubierre’s paper appeared,
E. Podkletnov and R. Nieminen published the results of their work with a rotating
superconductor disk [31]. Podkletnov’s disk (made of yttrium barium copper
oxide, or YBCO) was cooled to superconducting temperatures by liquid
helium. It was then levitated by a magnetic field, with electric current supplied
to coils surrounding the disk, and rotated between 3000 and 5000 revolutions
per minute. Podkletnov asserted that he had discovered a gravity shielding
effect that accounted for a two percent weight reduction in small objects held
10 P. A. GILSTER

above the spinning disk. Naively, such an effect would seem to violate con-
servation of energy, but the apparatus did require an energy input. Podkletnov
continued his work, producing a 1996 follow-up paper that was leaked before
publication to the British Sunday Telegraph, causing unfortunate claims about
“. . . the world’s first antigravity device” and a flurry of tabloid-style speculations
in the press.
Before Podkletnov’s experiments, Ning Li and Douglas Torr had investigated
antigravity effects, predicting in 1991 that superconductors could play a role
in producing them [32]. Li and Torr were influential in convincing NASA’s
Marshall Space Flight Center in Huntsville, Alabama to begin its own study,
which would attempt to recreate Podkletnov’s configuration. The project,
which ran from 1995 to 2002, was unable to complete the needed test hardware
with the resources then available [33]. The experimental configuration was,
however, reproduced in a privately funded test, with results published in 2003.
Using equipment 50 times more sensitive than had been available to Podkletnov,
the group found no evidence of a gravity-like force [34]. A subsequent
Podkletnov claim, that a force beam could be created from the application of
high-voltage discharges near superconductors, appeared on the arXiv physics
preprint site on the Internet in 2001 but remains unsubstantiated by independent
researchers [35]. Nonetheless, the publicity attendant to Podkletnov’s work
infused the mid-1990s with both energy and serious skepticism.

IX. Considering the Quantum Vacuum

A 1994 workshop held at the Jet Propulsion Laboratory shortly before the
appearance of the Alcubierre warp drive paper examined the topic of faster-
than-light travel from a range of perspectives, noting the experimental
approaches that might be used to study the subject [36]. These included resolving
the question of the neutrino’s rest mass to determine whether or not evidence of
imaginary mass existed. Another approach was to measure the speed of light
inside the space between closely spaced conductive plates (a “Casimir cavity”)
to look for evidence of negative energy. Other possibilities were to study
cosmic rays above the atmosphere to look for behavior characteristic of tachyons,
and to search for astronomical evidence of wormholes, which should show a
distinctive visual signature caused by the distortions indicative of negative
mass at the entrance [37].
The informal workshop noted a key issue of faster-than-light travel, the
apparent violation of causality, concluding that while such violations would be
unavoidable in this context, their physical prohibition could not be established.
Also, new theories were emerging on the connection between gravity and
electromagnetism, already known to be coupled phenomena. Gravity’s effects
upon spacetime have been confirmed, mass being known to warp the spacetime
against which electromagnetism is measured. In 1968, Andrei Sakharov had
raised the possibility that gravity could be an induced effect related to changes
in the vacuum energy that pervades the universe [38]. A theory published by
Bernard Haisch et al. early in 1994 suggested that inertia results from the resist-
ance of a particle’s velocity to change as it moves through this vacuum energy
background (dubbed zero-point energy or ZPE), whose electromagnetic
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 11

oscillations remain in the vacuum after all other energy has been removed [39].
Subsequent NASA-sponsored research (1996 1999) on the cosmological impli-
cations of quantum vacuum energy produced five papers, two examining the
emerging view of inertia [40,41]. A subsequent paper looked at the possibility
of engineering such effects [42]. Chapter 13 examines the quantum-vacuum-
inertia hypothesis in detail.

X. Institute for Advanced Studies at Austin

In addition to academic and government-sponsored work, privately sponsored
organizations also investigate such topics. The Institute for Advanced Studies at
Austin, Texas, established in 1985 by H. E. Puthoff, continues to examine ZPE’s
implications through work on gravitation and the nature of inertia, while probing
issues of energy generation and cosmology. The broader theory behind this work
sees all matter as reliant on the existence of the ZPE field, with gravity as a
necessary consequence. Max Planck first introduced a term for ZPE into the
equations for blackbody spectral radiation in 1912. Einstein would recognize
the physical reality of ZPE the following year, while Milliken’s experiments in
1924 provided spectroscopic evidence (see Chapter 18 for more on the back-
ground of ZPE studies). Since then, it has been the subject of investigations
into the Casimir effect, the Lamb shift, photon noise in lasers, van der Waal
forces, spontaneous atomic emission, and the stability of the ground state of
the hydrogen atom from radiative collapse. Active debate continues between cos-
mologists and quantum field theorists over the energy density of the ZPE, there
being no accepted bridge between quantum theory and spacetime physics.
With these subjects in the air, the factors that would lead to the establishment
of NASA’s BPP Project were now in play. They drew as well on earlier work con-
ducted by the U.S. Air Force and other agencies, which had already examined
numerous propulsion concepts that stayed within the realm of classical
physics, including nuclear rocketry, laser-beamed lightsails, laser rockets, and
antimatter propulsion, but also considered more speculative topics [43 47].
Robert Forward’s continuing investigations under Air Force Research Laboratory
contracts and in-house Hughes Research Laboratory programs ranged from anti-
gravity studies to antimatter, leading to his keynote address at the Vision-21 con-
ference and providing inspiration to numerous researchers going forward. Seven
years of academic research culminated in Matt Visser’s graduate-level textbook
on wormhole theory [21], while physicists in New York attended a conference
called “Practical Robotic Interstellar Flight: Are We Ready?” convened by
Edward Belbruno at New York University [48].

XI. Breakthrough Propulsion Physics Project

The year 1996 saw NASA’s decision to create a comprehensive strategy for
advancing the state of the art in space propulsion. Appearing under the rubric
Advanced Space Transportation Program (ASTP), the new plan became the
charge of the Marshall Space Flight Center (MSFC), which was given the task
of making the new 25-year plan more visionary than previous attempts. ASTP
was to examine propulsion concepts ranging from near-term concepts to
12 P. A. GILSTER

genuinely revolutionary ideas that could enable interstellar travel. The latter
could clearly draw on an established context at NASA’s Lewis Research
Center, where individuals had pursued such topics through informal sessions
and the Vision-21 Conference that grew out of them. Accordingly, MSFC
charged the Lewis Center with the task of developing the BPP Project to
address those issues at the most visionary end of the ASTP scale. Rather than
seeking refinements to existing methods, the BPP Project would specifically
target propulsion breakthroughs from physics. Its objective: Near-term, credible,
and measurable progress in attaining these results.†
Under the leadership of Marc Millis, the BPP Project set about its mission
noting the historical pattern common to technological revolutions. When steam
ships displaced sail, an existing technology had reached the limits of its physical
principles, underlining the need for new methods. The BPP Project operated
under the assumption that chemical and solid rocket technologies, as was the
case with aircraft supplanting ground transport, were approaching their perfor-
mance limits. Intended to be application-oriented, the project would tie its discov-
eries to the unique problem of spaceflight. As opposed to broader and more
theoretical research, it would present different lines of inquiry related to the
physics of propulsion. The program chose a “Horizon Mission Methodology”
that set the “impossible” goal of interstellar travel as its focus, hoping to spur think-
ing beyond existing solutions into the realm of entirely new possibilities [49].
The major barriers to interstellar travel were quickly identified in the form of
three “grand challenges”:
1) Mass: Discover fundamentally new propulsion methods that either elimin-
ate or drastically reduce the need for propellant. Such methods might create pro-
pulsive forces through the manipulation of inertia or gravity, or they might
interact with the properties of space itself, or mine the interactions between
matter, energy, and spacetime.
2) Speed: Discover how to achieve transit speeds that could drastically
reduce travel times. This goal entailed looking for ways to move a vehicle at
or near the maximum limits for motion through spacetime, or to exploit the
motion of spacetime itself (warp drive theory).
3) Energy: Discover ways to generate the massive amounts of onboard
energy that theory suggested would be required by propulsion systems like
these. Propulsion goals were thus seen as closely linked to energy physics.
BPP was to be implemented according to a series of conditional steps, each
contingent upon the completion of the previous step. These milestones were orga-
nized as follows [50]:
1) 1996: Determine that sufficient scientific foundations exist to pursue the
objective (completed).
2) 1997: Determine that affordable research candidates exist (completed).

†
The early history of the project is recounted in a thesis by Szames, A.D., “Histoire et enjeux du
projet NASA Breakthrough Propulsion Physics [History and stakes of NASA’s BPP Project],”
(unpublished). Diplôme d’Etudes Approfondies (DEA) Science Technologie Société, Paris: Conser
vatoire National des Arts & Métiers (CNAM), 1999.
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 13

3) 1998: Devise means to prioritize and select research tasks (completed).

4) 1999/2000: Solicit and select first round of research tasks (completed).
5) 2001/2002: Streamline the review and selection process via a research
consortium, to be operated by the Ohio Aerospace Institute (terminated when
funding was cut).

A. Kickoff Workshops
Because it could not be known whether breakthroughs like these were
possible, the BPP Project promised only incremental progress toward
achieving them. Using a system of collaborative networking among individual
researchers one connecting the various NASA centers, government labora-
tories, universities, industries, and individual scientists the program hoped to
create a multidisciplinary team open to collaboration and collateral support.
Although the Internet was seen as a primary communications medium (and sup-
porting Web sites established), a brainstorming session in Austin, Texas took
place in February of 1997. The program’s major event would be a workshop
on propulsion breakthroughs held on 12 14 August 1997 in Cleveland, Ohio.
Fourteen invited presentations, 30 poster papers, and a series of parallel breakout
sessions were designed to generate a list of candidate research tasks, with partici-
pants asked to entertain for the duration of the event that the breakthroughs
sought were indeed achievable within the context of sound and tangible research
approaches. The stated goal: Credible progress toward incredible possibilities
[51]. Between the Austin and Cleveland meetings, some 128 ideas on research
approaches, many of them redundant, were generated.
Below is a list of the invited presentations in the order presented. References
are cited to related or equivalent works:
1) L. Krauss (Case Western Reserve University), “Propellantless Propulsion:
The Most Inefficient Way to Fly?” [52]
2) H. E. Puthoff (Institute for Advanced Studies at Austin), “Can the Vacuum
Be Engineered for Spaceflight Applications?: Overview of Theory and Exper-
iments” [39,53]
3) R. Chiao (University of California at Berkeley) and A. Steinberg,
“Quantum Optical Studies of Tunneling Times and Superluminality” [54]
4) J. Cramer (University of Washington), “Quantum Nonlocality and Poss-
ible Superluminal Effects” [55]
5) R. Koczor and D. Noever (Marshall Space Flight Center), “Experiments
on the Possible Interaction of Rotating Type II YBCO Ceramic Superconductors
and the Local Gravity Field” [33]
6) R. Forward (Forward Unlimited), “Apparent Endless Extraction of Energy
from the Vacuum by Cyclic Manipulation of Casimir Cavity Dimensions” [56]
7) B. Haisch (Lockheed) and A. Rueda, “The Zero-Point Field and the NASA
Challenge to Create the Space Drive” [39]
8) A. Rueda (California State University) and B. Haisch, “Inertial Mass as
Reaction of the Vacuum to Accelerated Motion” [39]
9) D. Cole (IBM Microelectronics), “Calculations on Electromagnetic Zero-
Point Contributions to Mass and Perspectives” [57]
14 P. A. GILSTER

10) P. Milonni (Los Alamos), “Casimir Effect: Evidence and Implications”

[58]
11) H. Yilmaz (Electro-Optics Technical Center), “The New Theory of Grav-
itation and the Fifth Test” [59]
12) A. Kheyfets (North Carolina State University) and W. Miller, “Hyper-
Fast Interstellar Travel via Modification of Spacetime Geometry” [23,28]
13) F. Tipler III (Tulane University), “Ultrarelativistic Rockets and the
Ultimate Future of the Universe” [60]
14) G. Miley (University of Illinois), “Possible Evidence of Anomalous
Energy Effects in H/D-Loaded Solids Low Energy Nuclear Reactions”
Following these presentations, participants divided into six breakout groups,
with each of the three “grand challenge” goals addressed by two of the six.
From the ensuing discussions, ideas for next-step research tasks were solicited
focusing on strategies that would be of short-duration, low-cost, and incremental
in their nature.

B. BPP Project Research Tasks

Although the Cleveland conference was to be the BPP Project’s first and only
workshop, work proceeded actively in other venues. In its short life, the program
sponsored five tasks through competitive selection, two in-house tasks and one
minor grant. Sixteen peer-reviewed journal articles resulted [61,62]. Summaries
of the eight tasks appear below:
1) Define Space Drive Strategy: What research paths lead to the creation of a
space drive, defined as a device creating propulsion without a propellant? The
vehicle must exert external net forces on itself while satisfying conservation of
momentum. This task conceived and assessed seven hypothetical space drives
and analyzed remaining research approaches including 1) investigating new
sources of reaction mass in space; 2) revisiting Mach’s principle to consider
coupling to surrounding mass via inertial frames; and 3) studying the coupling
between gravity, inertia, and controllable electromagnetic phenomena [27].
(Chapter 3 is a follow-on to this initial work.)
2) Test Schlicher Thruster: Experiments conducted in-house were performed
to investigate claims by Rex Schlicher et al. [62] that a specially terminated coax
could create more thrust than could be attributed to photon pressure. No such
thrust was observed [63].
3) Assess Deep Dirac Energy: Examined the question of additional energy
levels and energy transitions in atomic structures [64]. A theoretical assessment
demonstrated that some of these states were not possible, but the topic is not con-
sidered fully resolved.
4) Cavendish Test of Superconductor Claims: This work was performed as a
lower-cost way to examine claims of gravity shielding advanced by Podkletnov
and Niemenen [31]. Not intended as a replication of Podkletnov’s experiments, it
was rather designed to test related consequences. Radio frequency radiation used
in these experiments was found to couple too strongly to supporting instrumenta-
tion to produce a clear resolution of any gravitational effect [65]. No follow-up to
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 15

this approach is expected. (Chapter 5 discusses the experimental challenges in

general with superconductors and provides a brief history of related work.)
5) Test Woodward’s Transient Inertia: Experiments were conducted to test
James Woodward’s claim that transient changes to inertia can be induced by elec-
tromagnetic means, with possible relation to Mach’s principle [66 67]. The test
program was changed when later Woodward publications defined a smaller effect
than originally anticipated. No discernible effect appeared in the revised testing.
Woodward’s experiments and investigations of their theoretical background
continue and the issue of transient inertia is considered unresolved. (Chapter
11 discusses subsequent independent testing.)
6) Test EM Torsion Theory: These experiments studied a torsion analogy to
the coupling between electromagnetism and spacetime [68]. The experiments
were later found to be insufficient to resolve the issue correctly, and the
possibility of asymmetric interactions that might prove valuable in propulsion
[69] is considered unresolved [70].
7) Explore Quantum Tunneling: Studied the question of faster-than-light
information transfer, as raised by the phenomenon of quantum tunneling,
where signals appear to pass through barriers at superluminal speed [71]. The
information transfer rate was found to be only apparently superluminal (the
entire signal remains light-speed limited), producing no causality violations.
The experimental and theoretical work sponsored on this subject examined the
special case where energy is added to the barrier. (Chapter 16 examines the
faster-than-light implications of quantum mechanics.)
8) Explore Vacuum Energy: The quantum vacuum energy, also known as
ZPE, draws on Heisenberg’s Uncertainty Principle, which implies that an absol-
ute zero electromagnetic energy state is not possible in the space vacuum. The
manifestation of such “vacuum energy” in the so-called “Casimir effect” is
known to pull two parallel plates together at dimensions on the micron scale.
BPP Project-sponsored investigations used micro-electro-mechanical (MEM)
rectangular Casimir cavities, showing the possibility of creating net propulsive
forces using this energy, although these forces are exceedingly small [72]. The
quantum vacuum continues to be the subject of active research into the nature
of space itself. (Chapter 12 specifically addresses the notion of using the
quantum vacuum as a reactive medium for space propulsion.)

The BPP Project ran for seven years and the funding devoted to this effort was
$1.6 million in total over those seven years [50]. In October of 2002, the
Advanced Space Transportation Program was reorganized, with all research
deemed at less than Technology Readiness Level 3 (at which active research
and development including laboratory studies is initiated) being canceled. Thus
ended the BPP Project. It was in the very next month that the final report of
the President’s Commission on the Future of the U.S. Aerospace Industry,
which had been charged to assess the health of the aerospace industry in the
United States, appeared. Chapter 9 of that report presented a justification for
the kind of research that the BPP Project had represented: “In the nearer-term,
nuclear fission and plasma sources should be actively pursued for space appli-
cations. In the longer-term, breakthrough energy sources that go beyond our
16 P. A. GILSTER

current understanding of physical laws, such as nuclear fusion and anti-matter,

must be credibly investigated in order for us to practically pursue human explora-
tion of the solar system and beyond. These energy sources should be the topic of a
focused basic research effort . . .” [73]. The same report also emphasized the need
to explore energy sources such as ZPE.

C. BPP Assessment: Nonviable Concepts

Credible investigation, of course, identifies not only research avenues to be
pursued, but those that are unlikely to prove fruitful for energy generation or pro-
pulsive effects. Drawing on the BPP Project’s research tasks and subsequent
investigations, a recent paper clarified those approaches deemed to be nonviable
[74]. These include:
1) Quantum Tunneling as a Faster-than-Light Venue: Because the effect is
only apparent (the entire signal still being limited by the speed of light even if
its leading edge appears to exceed that speed), this phenomenon does not demon-
strate faster-than-light information transfer and therefore has no broader impli-
cations for faster-than-light travel.
2) Asymmetric Capacitor Lifters (Biefeld-Brown Effect): Involving the cre-
ation of thrust by high-voltage capacitors, the effect (sometimes discussed
under the rubric “electrogravitics” or “electrostatic antigravity” was found to
be the result of coronal wind [75]. Chapters 8 and 9 describe experimental
tests of these devices in more detail, showing that asymmetric capacitor lifters
operate according to known physics rather than being evidence of new gravita-
tional physics, as is often claimed.
3) Hooper Antigravity Coils: Although said to reduce the weight of objects
placed beneath them, these self-canceling electromagnetic coils (based on U.S.
Patent 3,610,971 by W. J. Hooper) were shown to produce no such effect [76].
4) Oscillation Thrusters: These are mechanical devices claimed to produce
a net thrust using the motion of internal components. The apparent thrust seen
in such devices is a misinterpretation of mechanical effects. A gyroscopic thrus-
ter, for example, can move in a way that seems to defy gravity, but its motion is
actually caused by gyroscopic precession, the forces involved being torques
around the axes of the gyroscope mounts. No net thrust is thus created.
(Chapter 6 describes these forces and how they are commonly misinterpreted.)
5) Podkletnov Gravity Shield: Privately funded replication of Podkletnov’s
experimental configuration has demonstrated no gravity shielding effect over
spinning superconductors [34].
The BPP Project’s research represents one part of a larger contribution, which
is the demonstration that breakthrough physics related to specific space-related
tasks can be productively studied. An independent review panel of the Space
Transportation Research Program reached this conclusion in 1999, stating:

The BPP approach was unanimously judged to be well thought out, logically
structured and carefully designed to steer clear of the undesirable fringe
claims that are widespread on the Internet. The claim that the timing is
ripe for a modest Project of this sort was agreed to be justified: Clues do
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 17

appear to be emerging within mainstream science of promising new areas of

investigations. The team concurred that the 1997 BPP kickoff workshop did
identify affordable candidate research tasks which can be respectably
pursued, and that the proposed research prioritization criteria were a valid
way to select from amongst these (and future) proposals. The Project
approach was deemed to be sound: emphasizing near term progress toward
long term goals; supporting a diversity of approaches having a credible
scientific basis; holding workshops to exchange ideas; solicit constructive
criticism and assess progress; aiming toward testable concepts [77].

XII. Project Greenglow

While the BPP Project proceeded, a variety of other research projects
emerged, each looking at advanced propulsion capabilities from its own angle.
Prominent among these investigations is Project Greenglow, whose antecedents
actually pre-date the BPP Project by a decade with beginnings in 1986. In some
ways similar to the BPP Project although funded privately, Project Greenglow
was sponsored by Europe’s BAE Systems (created by the merger of British Aero-
space and Marconi Electronics Systems) to study a range of approaches to
advanced propulsion, maintaining the BPP Project’s outlook that such work
could best be advanced in this nascent stage by broad examination rather
than narrowing to particular channels. Project Greenglow’s emphasis was on
near-term aeronautics developments, but the similarities with the BPP Project
in its research interests were made explicit in an online announcement, although
no formal connection existed between the groups:

Our approach to the research programme will be much along the lines of
the newly established NASA Breakthrough Propulsion Physics Programme
which has, as its central theme, the goal of developing propellantless
propulsion. We hope that there will be elements where our two programmes
come together and we will support interaction between the two. Following
the lead set by NASA, it is intended that most of the administrative effort
and day to day communications will be conducted via the Internet. Also,
the aim will be to keep the research programme as open as possible,
within the restrictions imposed by academic ownership and commercial
return on investment, as we recognise that there is a great deal of interest
throughout the world on gravitational research, particularly on the possibility
of the control of gravitational fields. We may wish to harness this support in
our quest for funding [78].

Project Greenglow’s origins actually date as far back as the 1970s, in what was
then the British Aircraft Corporation’s Future Concepts Group, which was estab-
lished to investigate new fighter designs and technologies. What aerospace
engineer and Project Greenglow consultant John E. Allen calls “speculative
propulsion” entered the domain of serious study in 1986 [79]. A 1990 meeting
at Preston (Lancashire) highlighted the company’s interest in gravitational
research, while links to researchers in government and private industry to
determine their level of interest were made in the early 1990s. The funding of
18 P. A. GILSTER

exploratory programs at several British universities began in 1996, their out-

growth being Project Greenglow itself in 1997. While the project would
examine technologies based on conventional physics solar sails, ion drives,
and microwave thrusters a Greenglow “route map” included research vectors
into quantum electrodynamics, ZPE, and quantum gravity. Work on gravitation
was performed at Lancaster University and Sheffield University. Project Green-
glow director R. A. Evans notes:

The existence of gravitomagnetism was predicted more than a century ago, but
has never been detected. From an engineering point of view, being able to gen
erate artificial gravitomagnetic fields offers the prospect of being able to induce
gravity fields. Following on from this would be the control of inertia which would
open up the possibility of field propulsion without expelling mass [79].

With superconductivity research also handled at Sheffield, the University of King-

ston was charged with the study of engineering design issues, Birmingham with the
Casimir force, and Dundee and Strathclyde with microwave thrust technologies.
The project supported a series of incremental research tasks including inves-
tigation of the Podkletnov gravity shield claims, vacuum energy extraction, and
various approaches to the coupling of electromagnetism and gravity [80]. On the
issue of gravity shielding, Project Greenglow commissioned a study performed at
Sheffield University that examined the effect of a rotating superconductor on
the weight of a nearby test mass. No effect was observed. Left undetermined
by the study was the question of whether the methods Podkletnov used in levitat-
ing the superconductor disk (three 1 MHz solenoids) could explain the gravita-
tional shielding effects.
But experiment, theory, and design were seen as only a part of the larger
Greenglow program. The project also examined published concepts, theories,
and patents seemingly germane to advanced propulsion, reaching the conclusion
that most of these avenues were impractical due to the weakness of the forces
alleged to exist. The project thus turned to a method not dissimilar from the
BPP Project’s own “horizon mission methodology,” creating a hypothetical
system that implied the existence of physics breakthroughs and could be used
to design revolutionary aircraft. These “what if ” concepts led to a series of aero-
space vehicle designs; their engineering was explored and consequences on
world transportation were analyzed. As opposed to the BPP Project, Project
Greenglow emphasized global outcomes of physics breakthroughs rather than
the steps needed to achieve those breakthroughs. Assuming a new kind of
force interacting with space, researchers used the term “mass dynamic drive”
as their breakthrough model, incorporating it ultimately into 20 aerospace projects.
A major challenge in such work was noted by one Greenglow consultant:

The fundamental difficulty is the totally different professional worlds of

the advanced physicist and the engineering designer. It is most unlikely that
the key would be discovered by accident. So how can the chance be increased
significantly? The essence must be to bring together the two worlds of project
design and advanced physics in an intense serendipitous initiative [81].
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 19

Whether funding for this privately financed project will continue is not known,
as is the case for several other potential studies. Reports in Jane’s Defense Weekly
that American aerospace giant Boeing was investigating superconductivity
effects related to Podkletnov’s work remain unconfirmed [82]. Meanwhile, an
article in Aviation Week and Space Technology has claimed that a U.S. aerospace
company is working on ZPE research in conjunction with the Department of
Defense [83]. The private nature of such potential work makes the prospects
for disclosure unknown.

XIII. ESA’s General Studies Programme and Advanced

Concepts Team
The ESA General Studies Programme (GSP) acts as a think tank examining
the feasibility of new concepts and methodologies, with studies selected from
proposals submitted by ESA staff. GSP’s feasibility studies last between 12
and 24 months, with 30 to 50 new studies typically initiated during each cycle.
Through its Ariadna program, the GSP serves as an interface between the ESA
and university researchers whose work may have space applications. Ariadna
allows academics to propose projects for investigations that range from two to
six months in duration. An ESA study on the potential use of gravity control
for propulsion identified the “Pioneer effect” as well as gravitomagnetic fields
in quantum materials as an area for continuing research [84]. The same report rec-
ommended ongoing space experiments to search for possible violations of Ein-
stein’s Equivalence Principle.
The related ESA Advanced Concepts Team (ACT) is a multidisciplinary
research group based at the European Space Research and Technology Centre
in Noordwijk, The Netherlands. Created in 2001, ACT includes among its
battery of tasks research into fundamental physics in the fields of gravitation,
quantum physics and high-energy physics. Within the range of its assessments
have come revolutionary concepts that the ACT team has analyzed for potential
benefits, ruling out those concepts found to be nonviable. ACT has also devel-
oped two mission architectures for missions to the outer Solar System that
could be used to test the anomalous acceleration detected in both Pioneer
probes (along with the Galileo and Ulysses spacecrafts) that may involve a
novel physical effect [85].
ACT’s charter sometimes takes it into theoretical realms where fruition might
be decades away, the kind of work that a busy agency faced with launch dates on
upcoming missions seldom has time to investigate. Of particular relevance to
breakthrough propulsion are three studies on what ACT calls “nonissues,”
topics that upon investigation have been declared to be nonviable:
1) Specific Coupling Effects Between Electromagnetism and Gravity: To
quote from ACT’s description of this work:

There are . . . some tricks which might be realized by nature which would
allow the scalar field to be dynamical under some circumstances (e.g., low
gravitation, low temperatures). In this case the scalar could mediate an
enhanced coupling between gravitation and some type of matter. A suggestion
20 P. A. GILSTER

for a particular strong interaction has recently been put forward. It even pos
tulates a measurable effect on gravity from the earth’s magnetic field. Such
a strong gravielectric coupling would however open dramatic technological
possibilities including the possibility of electromagnetic levitation [86].

The study relies on experimental data from work on gravitational physics at the
University of Washington, deriving constraints that yield upper bounds for the
strength of such coupling. ACT’s study argues that the effects are too weak for
technological applications.
2) Faster-than-Light Communications by Quantum Tunneling: As did the
BPP Project, ACT examined tunneling effects, its particular emphasis being to
determine their relevance for space communications. It found no technical appli-
cations in the absence of suitable material for a tunneling barrier in the interpla-
netary medium. ACT also notes that “long tunneling barriers will only allow
faster than light communications with long wavelength signal and hence at low
data rates. Hence in the end one will have to opt for either high data rates or
high transmission speeds” [87].
3) Quantum Vacuum Effects: A study involving the universities of Cologne
and Grenoble examined energy or momentum extraction from the quantum
vacuum for possible propulsive effects, concluding that the inability to amplify
these forces makes them unusable for spacecraft applications [88].

XIV. Broad Strategies for Future Research

Looking to the potential of propellantless propulsion via space drive, two
demands emerge. The drive must satisfy conservation of momentum, and the
vehicle it propels must be able to induce external net forces on itself. The BPP
Project assessed seven hypothetical space drives [27]. From this and later work
[74], candidates for continuing research emerged as follows (discussed in
Chapter 3):
1) Dark Matter and Dark Energy: A space vehicle carrying no fuel of its own
must use a reaction mass indigenous to the universe around it. Recent
astronomical findings related to dark matter and dark energy reveal possible strat-
egies for finding new forms of reaction mass. The anomalous rotation rates of
galaxies suggest the presence of huge amounts of dark matter, perhaps account-
ing for as much as 90% of the visible galaxy. The possible use of dark matter for
propulsion will drive future studies, while the dark energy seemingly responsible
for the continuing acceleration of the expansion of the cosmos is in its earliest
stages of analysis. Dark energy is perhaps linked to an effect similar to antigrav-
ity, and is thus of obvious interest for propulsion theory.
2) Fundamental Force Relationships: Electromagnetism, gravity, and space-
time are known to be coupled phenomena. General relativity suggests that the
amount of electromagnetic energy required to produce gravitational effects
would be enormous. Continuing work on couplings at the atomic scale and
smaller, however, explores the realm of quantum and particle physics in relation
to such coupling. Such work deepens our understanding of general relativity’s
relationship to quantum mechanics, offering the potential of uncovering hitherto
unknown propulsive effects.
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 21

3) Frame Dependent Effects: Mach’s principle sees spatial reference frames

for acceleration as connected to the surrounding mass in the universe. An absol-
ute reference frame, which grows out of a literal interpretation of Mach’s work,
includes all matter in the universe (the cosmic microwave background radiation
is an observable phenomenon suggestive of such a reference frame). Possible
propulsive effects may be investigated that could induce net forces on a space-
craft relative to the surrounding mass. Frame dependent effects depart from the
geometric analysis of general relativity and point to a Euclidean interpretation.
The latter treatment is often referred to as an “optical analogy,” one that views
space as an optical medium with an effective index of refraction that is a function
of gravitational potential. Consistent with observation, such an approach may
prove useful in developing new propulsion concepts [89].
4) Quantum Vacuum Effects: Inertia may be interpreted as an electromag-
netic drag force experienced by accelerating matter as it moves against the
vacuum under some theories. A corollary is that gravity can be interpreted in
terms of the distribution of vacuum energy in the presence of matter. These con-
cepts remain one focus of the Institute for Advanced Studies in Austin, Texas,
including recent experimental work [90]. The California Institute for Physics
and Astrophysics has pursued privately sponsored research into these and
related issues. Given that these theories are subject to much debate, it is not sur-
prising that propulsion applications have been only rarely studied [42]. However,
the broader issue of quantum vacuum energy as demonstrated through the
Casimir effect is more widely in play, even with the understanding that the
latter is appreciable only for the smallest cavities (at the micron level). Recent
experimental work has focused on microelectromechanical (MEMS) structures
in the examination of the production of net forces from this energy [91,92].

XV. Continuing Research Projects

Beyond these broad topics, the following specific projects are among those
undergoing further investigation, their status unresolved:
1) Gravitomagnetic Fields in Rotating Superconductors: The coupling of
electromagnetism, gravity, and spacetime offers ground for continuing interest.
Couplings at the macro-level invoke General Relativity and thus imply
substantial levels of electromagnetic energy to produce a measurable effect.
But the lack of a unified theory incorporating both gravity and quantum
mechanics points to the unresolved issues found at the level of particle physics
that may have implications for propulsion. Growing out of work done for the
ESA on hypothetical gravity control [84], Martin Tajmar and collaborators
have used a spinning superconductor ring to detect an apparent frame-dragging
force considerably stronger than what would have been predicted by theory.
Tajmar’s team at the Austrian Research Center facility in Seibersdorf draws on
the understanding that below a critical temperature, such effects begin to appear
in various materials, including a nonsuperconducting ring of aluminum. Measur-
ing the gravitomagnetic field in the laboratory involves using rings of these
materials, with both accelerometers and ring-laser-gyros positioned near the
ring. As the temperature drops, the accelerometers and ring-laser-gyros registered
signals suggestive of a gravitomagnetic, frame-dragging effect. Recent work by
22 P. A. GILSTER

Tajmar and colleagues describes experimental details of the original observations

[93, 94] and compares their work to that of a New Zealand group that had used a spin-
ning lead disk at superconductor temperatures [95]. The dissimilarity in the exper-
imental setups deployed by the two groups leaves the issue of induced
acceleration fields open.
2) Transient Inertia Phenomena: James Woodward’s experiments on oscil-
latory inertia changes created by electromagnetic means continue in his and
other hands (Woodward has actually taken out a patent U.S. Patent
No. 5,280,864 on the use of the technology for propulsion purposes).
Woodward’s “mass fluctuations” challenge conservation of momentum unless the
Mach Principle be invoked to assert that all matter is in some fashion connected
to all other mass in the universe, even the most distant. Experimental results have
been unable to confirm or refute these theories [96] (see Chapter 11).
3) Abraham-Minkowski Debate: The controversy here concerns electromag-
netic momentum within dielectric media (the equations for electromagnetic
momentum in vacuum are widely accepted). Momentum transfer between
matter and electromagnetic fields is described by contradictory equations
derived by Max Abraham and Hermann Minkowski. The latter’s work suggests
that in materials where light travels at slower velocities, electromagnetic fields
should exert greater momentum (the inverse of Abraham’s findings). Propulsion
concepts emerging from the Minkowski formulation have been studied under
such terms as “electromagnetic stress tensor propulsion,” “Heaviside force,” and
“Slepian drive,” while recent work by Hector Brito suggests a propulsive device
shaped around this theory, though the matter is still considered unresolved [97].
The West Virginia-based Institute for Scientific Research has also conducted
experiments on the propulsive possibilities involved [98], while a study completed
by the U.S. Air Force Academy rules out any net propulsive effects deriving from
internal momenta as described by Minkowski [99]. Another recent paper argues
that net forces are not possible using these principles [42]. Chapter 10 examines
this topic in more detail.

XVI. Faster than Light

Unlike many of these topics, which are susceptible to experimental analysis,
the warp drive and wormhole concepts that did so much to reawaken interest in
these investigations remain purely theoretical constructs. The magnitude of nega-
tive energy required makes creating suitable laboratory experiments unlikely,
although some theorists such as Eric Davis continue to study the possibility
(see Chapter 15). In a study for the Air Force Research Laboratory, Davis
examined using nuclear explosion magnetic compression or ultrahigh-intensity
tabletop lasers to create laboratory wormholes [100]. The likelihood of anything
but theoretical study of both warp drive and wormholes for the near future is
small. There remains, however, the possibility of detecting evidence of worm-
holes through astronomical data by noting their gravitational lensing effects
on distant light [37].
Since Alcubierre’s 1994 paper, some 50 publications have appeared addres-
sing one or another aspect of his “warp drive” concept. In identifying key
physics obstacles, especially the enormous amounts of negative energy needed
to generate and sustain warped spacetime, this work launched international
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 23

debate that continues in conferences and workshops such as the annual Space
Technology and Applications International Forum (STAIF) in Albuquerque,
New Mexico. A faster-than-light symposium hosted by the British Interplanetary
Society met on 15 November 2007 to discuss warp drive, the potential of vacuum
energy, and international standards for the mathematical discussion of these
concepts. Both warp drive and wormhole theories will doubtless remain in the
speculative realm for the indefinite future, but they do present provocative
thought experiments for deeper study of the related physics.

XVII. Conclusions
It is clear from this assessment that much of the most recent work on break-
through propulsion is because of individual efforts, often with scant funding,
scattered among government agencies, research laboratories, universities, and
private organizations. Government-sponsored projects like BPP operated over
a seven-year period and with a total budget of approximately $1.6 million.
This scenario is likely to persist, given that the major space agencies have par-
ticular agendas focusing on near-term missions. In the absence of demonstrated
success, many of the approaches examined here will continue to receive sparse
attention. However, the upsurge in studies within the past 15 years indicates an
appreciation of the implications of success. Should we find alternatives to
costly chemical propulsion, especially in the area of “propellantless” spacecraft,
the entire Solar System would swing into focus as a practical site for a space-
based infrastructure that might one day lead to interstellar missions.
The stakes in such work are high, but the issues under investigation often par-
allel those actively studied in general physics. The coupling of electromagnetism,
gravity, and spacetime is obviously of significance well beyond the realm of pro-
pulsion, as is the physics of inertial frames, but the perspective provided by a
practical, engineering viewpoint may prove beneficial to the theoretical study
of these matters. Similarly, an examination of the quantum vacuum may or
may not lead to propulsive effects, but using that particular filter may reveal
new theoretical possibilities. Propulsion physics, then, may be considered as a
discipline useful in its own right, recognizing that some or all of the hoped for
technologies may prove impractical or impossible to achieve. Even so, the
history of technology offers the prospect that ideas that fail may be replaced
by workable concepts not currently understood. It being the business of the
future to surprise us, we should not be too quick to rule out such scenarios.

Acknowledgments
The author wishes to thank Marc G. Millis, whose assistance throughout the
research and writing of this chapter proved invaluable. Thanks are owed as
well to Eric W. Davis for his insightful comments in reviewing the manuscript.

References
[1] Millis, M. G., and Nicholas, T., “Responding to Mechanical Antigravity,” NASA
TM 2006 214390; AIAA Paper 2006 4913.
[2] Kuhn, T. S., The Structure of Scientific Revolutions, Univ. Chicago Press, Chicago,
1962.
24 P. A. GILSTER

[3] Foster, R. N., Innovation: The Attacker’s Advantage, Summit Books, New York, 1986.
[4] Horgan, J., The End of Science: Facing the Limits of Knowledge in the Twilight of
the Scientific Age, Addison Wesley, Reading, MA, 1996.
[5] Dyson, F., Imagined Worlds, Harvard Univ. Press, Cambridge, MA, 1997.
[6] Sänger, E., “The Theory of Photon Rockets,” Space Flight Problems, Laubscher,
Biel Bienne, Switzerland, 1953, pp. 32 40.
[7] Sänger, E., “On the Attainability of the Fixed Stars,” Proceedings of the 7th
International Astronautical Congress, Rand Corp., Santa Monica, CA, 17 22
Sept. 1956, pp. 89 133.
[8] Ackeret, J., “On the Theory of Rockets,” Journal of the British Interplanetary
Society, Vol. 6, 1947, pp. 116 123.
[9] von Hoerner, S., “The General Limits of Space Travel,” Interstellar Commu
nication, A. G. W. Cameron (ed.), New York, Benjamin, 1963, pp. 144 159.
[10] Shepherd, L. R. “Interstellar Flight,” Journal of the British Interplanetary Society,
Vol. 11, 1952, pp. 149 167.
[11] Papailiou, D. D. (ed.), “Frontiers in Propulsion Research: Laser, Matter Antimatter,
Excited Helium, Energy Exchange, Thermonuclear Fusion,” TM 33 722 (NASA
CR 142707), Jet Propulsion Lab., Cal Tech, Pasadena, CA, 1975.
[12] Forward, R. L., Future Magic: How Today’s Science Fiction Will Become
Tomorrow’s Reality, Avon Books, New York, 1988; updated and republished as
Indistinguishable from Magic, Baen Books, New York, 1995.
[13] Forward, R. L., “Feasibility of Interstellar Travel: A Review,” Acta Astronautica,
Vol. 14, 1986, pp. 243 252.
[14] Evans, R. A. (ed.), BAe University Round Table on Gravitational Research, Rept.
FBS 007, British Aerospace Ltd., Preston, U.K., 26 27 March 1990.
[15] Millis, M., and Williamson, G. S., “Experimental Results of Hooper’s
Gravity Electromagnetic Coupling Concept,” NASA TM 106963, Lewis Research
Center, 1995.
[16] Niedra, J., Myers, I., Fralick, C., and Baldwin, R., “Replication of the Apparent
Excess Heat Effect in a Light Water Potassium Carbonate Nickel Electrolytic
Cell,” NASA TM 107167, Lewis Research Center, 1996.
[17] Mead, F., Jr., “Exotic Concepts for Future Propulsion and Space Travel,”
Advanced Propulsion Concepts, 1989 JPM Specialist Session, (JANNAF), CPIA
Publ. 528:93 99, 1989.
[18] Landis, G. L. (ed.), Vision 21: Space Travel for the Next Millennium. Proceedings,
NASA Lewis Research Center, 3 4 April 1990, NASA CP 10059, 1990.
[19] Millis, M. G., “Exploring the Notion of Space Coupling Propulsion,” Vision 21:
Space Travel for the Next Millennium. Proceedings, NASA Lewis Research
Center, NASA CP 10059, 3 4 Apr. 1990, pp. 313 322.
[20] Morris, M. S., and Thorne, K. S., “Wormholes in Spacetime and Their Use for
Interstellar Travel: A Tool for Teaching General Relativity,” American Journal
of Physics, Vol. 56, 1988, pp. 395 412; also see Morris, M. S., Thorne, K. S.,
and Yurtsever, U., “Wormholes, Time Machines, and the Weak Energy Condition,”
Physical Review Letters, Vol. 61, 1988, pp. 1446 1449.
[21] Visser, M., Lorentzian Wormholes: From Einstein to Hawking, AIP Press,
New York, 1995.
[22] Davis, E. W., “Advanced Propulsion Study,” Air Force Research Lab., Final Report
AFRL PR ED TR 2004 0024, Edwards AFB, CA, April 2004, p. 65.
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 25

[23] Alcubierre, M., “The Warp Drive: Hyper Fast Travel Within General Relativity,”
Classical and Quantum Gravity, Vol. 11, May 1994, pp. L73 L77.
[24] Emme, E. D. (ed.), Science Fiction and Space Futures Past and Present, American
Astronautical Society Historical Series, Vol. 5, American Astronautical Society,
San Diego, CA, 1982.
[25] Krasnikov, S. V., “Hyperfast Interstellar Travel in General Relativity,” Physical
Review D, Vol. 56, 1997, pp. 2100 2108.
[26] Misner, C. W., Thorne, K. W., and Wheeler, J. A., Gravitation, W. H. Freeman &
Co., New York, 1973.
[27] Millis, M., “Challenge to Create the Space Drive,” Journal of Propulsion and
Power, Vol. 13, 1997, pp. 577 582.
[28] Pfenning, M. J., and Ford, L. H. “The Unphysical Nature of Warp Drive,” Classical
and Quantum Gravity, Vol. 14, 1997, pp. 1743 1751.
[29] van Den Broeck, C., “A ‘Warp Drive’ with More Reasonable Total
Energy Requirements,” Classical and Quantum Gravity, Vol. 16, 1999, pp.
3973 3979.
[30] Sternberg, A. M., Kwiat, P. G., and Chiao, R. Y., “Measurement of the Single
Photon Tunneling Time,” Physical Review Letters, Vol. 71, 1993, pp. 708 711.
[31] Podkletnov, E., and Nieminen, R., “A Possibility of Gravitational Force Shielding
by Bulk YBa2Cu3O7-x Superconductor,” Physica C, Vol. 203, 1992, pp. 441 444.
[32] Li, N., and Torr, D. G., “Effects of a Gravitomagnetic Field on Pure Supercon
ductors,” Physical Review D, Vol. 43, 1991, pp. 457 459.
[33] Li, N., Noever, D., Robertson, T., Koczor, R., and Brantley, W., “Static Test for a
Gravitational Force Coupled to Type II YBCO Superconductors,” Physica C, Vol.
281, 1997, pp. 260 267.
[34] Hathaway, G., Cleveland, B., and Bao, Y., “Gravity Modification Experiment Using
a Rotating Superconducting Disk and Radio Frequency Fields,” Physica C, Vol.
385, 2003, pp. 488 500.
[35] Podkletnov, E., and Modanese, G., “Impulse Gravity Generator Based on Charged
YBa2Cu3O7 y Superconductor with Composite Crystal Structure,” arXiv:physics/
0108005 v2, August 2001.
[36] Bennett, G. L., Forward, R. L., and Frisbee, R. H., “Report on the NASA/JPL
Workshop on Advanced Quantum/Relativity Theory Propulsion.” AIAA Paper
1995 2599, 31st ASME/SAE/ASEE Joint Propulsion Conference and Exhibit,
San Diego, CA, 10 12 July 1995.
[37] Cramer, J., Forward, R. L., Morris, M., Visser, M., Benford, G., and Landis, G. L.,
“Natural Wormholes as Gravitational Lenses,” Phyical Review D, Vol. 51, 1995,
pp. 3117 3120.
[38] Sakharov, A., “Vacuum Quantum Fluctuations in Curved Space and the Theory of
Gravitation,” Soviet Physics Doklady, Vol. 12, 1968, pp. 1040 1041.
[39] Haisch, B., Rueda, A., and Puthoff, H. E., “Inertia as a Zero Point Field Lorentz
Force,” Physical Review A, Vol. 49, No. 2, 1994, pp. 678 694.
[40] Rueda, A., and Haisch, B., “Inertial Mass as Reaction of the Vacuum to Accelerated
Motion,” Physics Letters A, Vol. 240, 1998, pp. 115 126.
[41] Rueda, A., and Haisch, B., “Contribution to Inertial Mass by Reaction of
the Vacuum to Accelerated Motion,” Foundations of Physics, Vol. 28, 1998,
pp. 1057 1108.
26 P. A. GILSTER

[42] Puthoff, H. E., Little, S. R., and Ibison, M., “Engineering the Zero Point Field and
Polarizable Vacuum for Interstellar Flight,” Journal of the British Interplanetary
Society, Vol. 55, 2002, pp. 137 144.
[43] Mead, F. B., Jr., Advanced Propulsion Concepts Project Outgrowth, AFRPL TR
72 31, June 1972.
[44] Cravens, D. L., “Electric Propulsion Study,” AL TR 89 040, Final Report on
Contract FO4611 88 C 0014, Air Force Astronautics Lab. (AFSC), August 1990.
[45] Forward, R. L., 21st Century Space Propulsion Study, AL TR 90 030, Final Report
on Contract FO4611 87 C 0029, Air Force Astronautics Lab. (AFSC), Oct. 1990;
see also Forward, R. L., 21st Century Space Propulsion Study (Addendum),
PL TR 91 3022, Final (Addendum), OLAC Phillips Lab., formally known as Air
Force Astronautics Lab. (AFSC), June 1991.
[46] Cravens, D. L., “Electric Propulsion Study,” AL TR 89 040, Air Force
Astronautics Laboratory (AFSC), Edwards AFB, CA, 1990.
[47] Talley, R. L., “Twenty First Century Propulsion Concept,” PL TR 91 3009,
Phillips Laboratory, Air Force Systems Command, Edwards AFB, CA, 1991.
[48] “Practical Robotic Interstellar Flight: Are We Ready?” 29 Aug. 1 Sept. 1994,
New York Univ. and The United Nations, New York. Sponsored by the Planetary
Society.
[49] Anderson, J. L., “Leaps of the Imagination: Interstellar Flight and the Horizon Mission
Methodology,” Journal of the British Interplanetary Society, Vol. 49, 1996, pp. 15 20.
[50] Millis, M., Breakthrough Propulsion Physics Project: Project Management
Methods, NASA TM 2004 213406, 2004.
[51] Millis, M., NASA Breakthrough Propulsion Physics Program, NASA TM 1998 208400,
1998.
[52] Krauss, L. M., The Physics of Star Trek, Basic Books, New York, 1995.
[53] Puthoff, H. E., “Gravity as a Zero Point Fluctuation Force,” Physical Review A,
Vol. 39, 1989, pp. 2333 2342.
[54] Chiao, R. Y., Steinberg, A. M., and Kwiat, P. G., “The Photonic Tunneling Time
and the Superluminal Propagation of Wave Packets,” Proceedings of the Adriatico
Workshop on Quantum Interferometry, World Scientific, Singapore, 1994, p. 258.
[55] Cramer, J. G., “The Transactional Interpretation of Quantum Mechanics,” Reviews
of Modern Physics, American Physical Society, Vol. 58, 1986, pp. 647 688.
[56] Forward, R. L., “Extracting Electrical Energy from the Vacuum by Cohesion of
Charged Foliated Conductors,” Physical Review B, Vol. B30, 1984, pp. 1700
1702.
[57] Cole, D., and Puthoff, H. E., “Extracting Energy and Heat from the Vacuum,”
Physical Review E, Vol. 48, 1993, pp. 1562 1565.
[58] Milonni, P. W., The Quantum Vacuum: An Introduction to Quantum Electro
dynamics, Academic Press, San Diego, CA, 1994.
[59] Yilmaz, H., “Toward a Field Theory of Gravitation,” II Nuovo Cimento, Vol. 107B,
1992, pp. 941 960.
[60] Tipler, Frank, The Physics of Immortality, Doubleday, New York, 1994.
[61] Millis, M. G., “Prospects for Breakthrough Propulsion from Physics,” NASA
TM 2004 213082, May 2004.
[62] Schlicher, R. L., Biggs, A. W., and Tedeschi, W. J., “Mechanical Propulsion from
Unsymmetrical Magnetic Induction Fields,” AIAA Paper 95 2643, Joint Propulsion
Conference, San Diego, CA, 1995.
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 27

[63] Fralick, G., and Niedra, J., “Experimental Results of Schlicher’s Thrusting Antenna,”
Joint Propulsion Conference, AIAA Paper 2001 3657, Salt Lake City, UT, 2001.
[64] Maly, J., and Vavra, J., “Electron Transitions on Deep Dirac Levels I and II,”
Fusion Technology, Vol. 24, 1993, pp. 307 381, and Vol. 27, 1995, pp. 59 70.
[65] Robertson, G. A., “Exploration of Anomalous Gravity Effects by RF Pumped
Magnetized High T Superconducting Oxides,” Joint Propulsion Conference,
AIAA Paper 2001 3364, Salt Lake City, UT, 2001.
[66] Woodward, J. F., “A New Experimental Approach to Mach’s Principle and
Relativistic Gravitation,” Foundations of Physics Letters, Vol. 3, No. 5, 1990,
pp. 497 506.
[67] Woodward, J. F., “Measurements of a Machian Transient Mass Fluctuation,” Foun
dations of Physics Letters, Vol. 4, 1991, pp. 407 423.
[68] Ringermacher, H., “An Electrodynamic Connection,” Classical and Quantum
Gravity, Vol. 11, 1994, pp. 2383 2394.
[69] Cassenti, B. N., and Ringermacher, H., “The How to of Antigravity,” Joint Propul
sion Conference, AIAA Paper 96 2788, Lake Buena Vista, FL, 1996.
[70] Ringermacher, H., Conradi, M. S., Browning, C. D., and Cassenti, B. N., “Search
for Effects of Electric Potentials on Charged Particle Clocks,” Joint Propulsion
Conference, AIAA Paper 2001 3906, Salt Lake City, UT, 2001.
[71] Mojahedi, M., Schamiloglu, E., Hegeler, F., and Malloy, K. J., “Time Domain
Detection of Superluminal Group Velocity for Single Microwave Pulses,” Physical
Review E, Vol. 62, 2000, pp. 5758 5766.
[72] Maclay, J., and Forward, R. L., “A Gedanken Spacecraft that Operates Using the
Quantum Vacuum (Dynamic Casimir Effect),” Foundations of Physics, Vol. 34,
2004, pp. 477 500.
[73] Walker, R., Peters, F. W., Aldrin, B., Bolen, E., Buffenbarger, R. T., Douglass, J.,
Fowler, T. K., Hamre, J., Schneider, W., Stevens, R., Tyson, N., and Wood, H., Pre
sident’s Commission on the Future of the US Aerospace Industry Final Report:
Anyone, Anything, Anytime, Anywhere, 2002, p. 9 6.
[74] Millis, M., “Assessing Potential Propulsion Breakthroughs,” New Trends in Astro
dynamics and Applications, Belbruno, E. ed., Annals of the New York Academy of
Sciences, New York, Vol. 1065, 2005, pp. 441 461.
[75] Tajmar, M., “The Biefeld Brown Effect: Misinterpretation of Corona Wind
Phenomena,” Journal of Propulsion and Power, Vol. 42, 2004, pp. 315 318.
[76] Millis, M., and Williamson, G. S., “Experimental Results of Hooper’s
Gravity Electromagnetic Coupling Concept,” NASA TM 106963, Lewis Research
Center, 1995.
[77] Merkle, C. (ed.), “Ad Astra per Aspera: Reaching for the Stars,” Report of the
Independent Review Panel of the NASA Space Transportation Research
Program, Jan. 1999.
[78] “What Is Project Greenglow?,” Project Greenglow Internet site, URL: http://
www.greenglow.co.uk/ [3 January 2008].
[79] Allen, J. E., “Quest for a Novel Force: A Possible Revolution in Aerospace,”
Progress in Aerospace Sciences, Vol. 39, 2003, pp. 1 60.
[80] Szames, A. D., “Le Projet Greenglow pour Controller la Gravite,” Air & Cosmos, 14
July 2000, pp. 44 45.
[81] Allen, J. E., “Aeronautics 1903; Aerospace 2003; ?? 2103,” Journal of Aerospace
Engineering, Vol. 219 (G3), June 2005, pp. 235 260.
28 P. A. GILSTER

[82] Cook, N., “Anti Gravity Propulsion Comes ‘Out of the Closet,’” Janes Defense
Weekly, 29 July 2002.
[83] Scott, W., “To the Stars: Zero Point Energy Emerges from Realm of Science
Fiction, May Be Key to Deep Space Travel,” Aviation Week & Space Technology,
March 2004, pp. 50 53.
[84] Bertolami, O., and Tajmar, M., “Gravity Control and Possible Influence on Space
Propulsion: A Scientific Study,” ESA CR (P) 4365, on Contract ESTEC 15464/01/
NL/Sfe, 2002.
[85] Rathke, A., and Izzo, D., “Options for a Non dedicated Test of the Pioneer
Anomaly,” Journal of Spacecraft and Rockets, ArXiv: astro ph/0504634, Vol. 43,
No. 4, July Aug. 2006, pp. 806 821.
[86] Rathke, A., “Constraining a Possible Dependence of Newton’s Constant on
the Earth’s Magnetic Field,” ESA Advanced Concepts Team, ESTEC, Noordwijk,
The Netherlands, Aug. 2004.
[87] ESA Advanced Concepts Team, “Fundamental Physics: Faster than light
Communications by Tunneling,” URL: http://www.esa.int/gsp/ACT/phy/pp/
non issues/faster than light.htm [3 January 2008].
[88] Tiggelen, B. A., Rikken, G. L., and Krstic, V., “The Feigel Process: Lorentz Invar
iance, Regularization, and Experimental Feasibility,” Final Report, ACT Web site,
URL: http://www.esa.int/gsp/ACT/ariadna/completed.htm [3 January 2008].
[89] Evans, J. K., Nandi, K. K., and Islam, A., “The Optical Mechanical Analogy in
General Relativity: Exact Newtonian Forms for the Equations of Motion of
Particles and Photons,” General Relativity and Gravitation, Vol. 28, 1996,
pp. 413 439.
[90] Rueda, A., and Haisch, B., “Gravity and the Quantum Vacuum Hypothesis,” Annals
of Physics, Vol. 14, No. 8, 2005, pp. 479 498.
[91] Maclay, G. J., “Analysis of Zero Point Electromagnetic Energy and Casimir Forces
in Conducting Rectangular Cavities,” Physical Review A, Vol. 61, 2000,
pp. 052110 1 to 052110 18.
[92] Esquivel Sirvent, R., Villareal, C., and Mochan, W. L., “Casimir Forces in
Nanostructures,” Physica Status Solidi (b), Vol. 230, 2002, pp. 409 413.
[93] Tajmar, M., Plesescu, F., Seifert, B., and Marhold, K., “Measurement of
Gravitomagnetic and Acceleration Fields Around Rotating Superconductors,”
Proceedings of the STAIF 2007, AIP Conference Proceedings 880, 1071, 2007.
[94] Tajmar, M., Plesescu, F., Seifert, B., Schnitzer, R., and Vasiljevich, I., “Search for
Frame Dragging Like Signals Close to Spinning Superconductors,” Proceedings of
the Time and Matter 2007 Conference, Bled, Slovenia, World Scientific Press,
2007.
[95] Graham, R. D., Hurst, R. B., Thirkettle, R. J., Rowe, C. H., and Butler, P. H.,
“Experiment to Detect Frame Dragging in a Lead Superconductor,” Physica C,
Vol. 468, Issue 5, March 2008, pp. 383 387.
[96] Cramer, J., Fey, C. W., and Casissi, D. P., “Tests of Mach’s Principle with a Mech
anical Oscillator,” NASA CR 2004 213310, 2004.
[97] Brito, H. H., “Experimental Status of Thrusting by Electromagnetic Inertia Manipu
lation,” Acta Astronautica, Vol. 54, pp. 547 558, 2004. Based on paper IAF 01
S.6.02, 52nd International Astronautical Congress, Toulouse, France, 2001.
[98] Corum, J. F., Keech, T. D., Kapin, S. A., Gray, D. A., Pesavento, P. V., Duncan, M. S.,
and Spadaro, J. F. “The Electromagnetic Stress Tensor as a Possible Space Drive
 RECENT HISTORY OF BREAKTHROUGH PROPULSION STUDIES 29

Propulsion Concept,” AIAA Paper 2001 3654, 37th AIAA/ASME/SAE/ASEE

Joint Propulsion Conference, Salt Lake City, UT, July 2001.
[99] Bulmer, J. S., and Lawrence, T., “Interferometer Examination of the Time
Derivative of Electromagnetic Momentum Created by Independent Fields and
Applications to Space Travel,” USAFA TR 2003 03, United States Air Force
Academy, Colorado Springs, CO, 2003.
[100] Davis, E. W., “Teleportation Physics Study,” Air Force Research Lab., Final
Report AFRL PR ED TR 2003 0034, Air Force Materiel Command, Edwards
AFB, CA, 2004.
 Chapter 2

Limits of Interstellar Flight Technology

Robert H. Frisbee
Jet Propulsion Laboratory, California Institute of Technology,
Pasadena, California

I. Introduction
NE of mankind’s oldest dreams has been to visit the tiny pinpoints of light
O visible in the night sky. Much of the science fiction literature in the early
20th century is filled with fantastic stories of galaxy-spanning voyages to
distant planetary shores, much as Verne’s stories of the latter half of the
19th century dealt with fantastic voyages to the Moon (by cannon/rocket),
across Africa (by balloon), and under the sea (by submarine). Elements of
classic “space opera” continue today in which the laws of nature, at least as
we currently know and understand them, are ignored to varying degrees for the
sake of the storyteller’s art. In literary criticism this has been called, using a
tennis metaphor, “playing with the net down,” where all manner of “super-
science” technology is available. Here, the writer can make use of faster-than-
light (FTL) “warp drives,” unlimited power sources (e.g., for powering warp
drives or blowing up planets), “deflector shields,” traversable wormholes,
matter transporters, time machines, and so on. The opposite, “playing with the
net up,” or “mundane” science fiction [1], deals with stories in which the
author explicitly limits the availability of future technologies to relatively
near-term extrapolations of existing technologies, with strict adherence to the
laws of nature as we currently know and understand them. In this case, the

Copyright # 2008 by the American Institute of Aeronautics and Astronautics, Inc. The U.S.
Government has a royalty free license to exercise all rights under the copyright claimed herein for
Governmental purposes. All other rights are reserved by the copyright owner.

Senior Engineer, Propulsion Section.
This work is dedicated to the memory of Robert L. Forward (1932 2002). So much of what we
think of when we consider interstellar propulsion concepts and technologies was first conceptualized
and evaluated by Bob Forward; our debt to him cannot be overemphasized. Isaac Newton said it best
when he said that we see further only because we stand on the shoulders of giants; Bob Forward was
one of those giants.

31
32 R. H. FRISBEE

starships travel much slower than the speed of light, relativistic effects are
important, power is always at a premium, and physical shielding plates are
required to protect against interstellar dust grain impacts.
In the overall context of this book, this chapter takes the “mundane” science
fiction approach of asking what types of technologies are required to perform,
initially, interstellar precursor missions to distances of hundreds or thousands
of astronomical units (AU, 149.6  106 km), and, ultimately, full interstellar
missions with cruise (peak) velocities on the order of 50% of the speed of
light, c. Here, we will first identify representative mission requirements (e.g.,
distance and desired trip time) and then use extrapolations of known science
and engineering to develop “paper” spacecraft that might meet the mission
requirements.

II. Challenge of Interstellar Missions

Over the last 50 years we have visited most of the major bodies in our solar
system, reaching out far beyond the orbit of Pluto with our robotic spacecraft.
However, the leap to interstellar distances represents an unprecedented
challenge. For comparison, more than 30 years after its launch (on 5 September
1977), the Voyager I spacecraft is about 103.6 AU, or 14.3 light-hours, from the
sun, traveling with a solar system escape velocity, Vinfinity, of 17.4 km/s (3.67 AU
per year) or 0.006% of the speed of light. Yet this distance, which strains the
limits of our technology, represents an almost negligible step toward the light-
years that must be traversed to travel to the nearest stars. For example, even
though the Voyager spacecraft is one of the fastest vehicles ever built,† it
would still require almost 75,000 years for it to traverse the 4.3 light-year
(LY) distance to our nearest stellar neighbor Alpha Centauri (if it were pointed
in that direction). Thus, travel to the stars is not impossible but will, however, rep-
resent a major commitment by a civilization simply because of the size and scale
of any technology designed to accelerate a vehicle to speeds of a few tenths of the
speed of light.
In fact, one of the most daunting aspects of comprehending the scale of an
interstellar mission is the sheer size of any transportation system capable of
reaching a significant fraction of the speed of light (e.g., 0.1 c or faster). As a
point of reference, a “payload” mass of 100 metric tons (MT 1000 kg),
roughly the mass of the space shuttle Orbiter, traveling at 0.5 c has a kinetic
energy of 1.29  1021 joules (J), including a relativistic mass correction of
1.15 {1/[1 (V/c)2]}1/2. This energy represents about 2.7 years worth of
the annual energy production of human civilization (4.85  1020 J in 2005).
As we will see, adding the required propulsion system to reach interstellar trans-
portation speeds results in vehicle systems with dimensions on the order of
planetary diameters, masses of hundreds of billions of tons, and power levels
thousands of times that of human civilization (about 15.4 TW in 2005) [2].

†
The New Horizons Pluto spacecraft started with a much higher velocity leaving Earth than
Voyager, but its Solar System escape velocity (Vinfinity) is less than Voyager (e.g., 3 AU/yr versus
3.7 AU/yr). This is because Voyager, initially slower, had the benefit of three gravity assists compared
to Pluto’s one, so Voyager has the higher final Vinfinity.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 33

A. Interstellar Mission Requirements

As a tool for evaluating the limits of technology for interstellar flight, we can
think of a “fast” (0.5 c cruise velocity) interstellar rendezvous mission as our
“Vision Mission” [3,4]. This mission represents a “stretch goal” that is intention-
ally made as difficult as possible so that a simple extrapolation of existing, near-
term technologies will not suffice [4]. Ultimately, this gives us a tool to aid in
structuring future technology development programs and interstellar precursor
missions with a long-range goal of giving us the capability to perform the
Vision Mission. As an historical example, we can consider the Apollo lunar
landing as the Vision Mission and stretch goal in the early 1960s. This led to
development of technologies like large chemical rocket engines, fuel cells, and
0-g cryogenic fluids systems. Similarly, it led to development during the
Gemini Program of space operations techniques like rendezvous and docking.
Finally, to support the Apollo lunar landings, a number of robotic precursor mis-
sions like Ranger, Surveyor, and Lunar Orbiter were flown.
Using our Vision Mission as a guide, we will focus on an evaluation of various
propulsion options for interstellar rendezvous missions ranging from 4.3 LY with
a 10-year trip time goal, to 40 LY with a 100-year trip time goal. The 40 LY limit
was selected to be compatible with the range of future NASA Origins Program tele-
scopes, as will be discussed. Various propulsion candidates are described in more
detail that might fulfill the Vision Mission, as well as interstellar precursor missions
leading up to implementation of the vision interstellar rendezvous mission. These
propulsion candidates [3] include advanced chemical propulsion, electric propul-
sion, nuclear (fission, fusion, antimatter) propulsion, beamed momentum propulsion
(e.g., laser “pushed” sails), electromagnetic catapults, in-situ propellant production
concepts (e.g., the Bussard interstellar ramjet), and hybrid systems (e.g., antimatter-
catalyzed fission/fusion). Interestingly, because of the high DV required for inter-
stellar flyby and rendezvous missions (up to several tenths of the speed of light)
and the large dry masses of the candidate propulsion systems, traditional nonpropul-
sive enhancements (like the gravity assists used by the Voyager spacecraft) are of
limited use, resulting in a need for specific impulses (Isp) on the order of 104 to
107 lbf -s/lbm (102 to 105 km/s), depending on the mission overall DV [3].
Our ultimate mission goal is the ability to visit scientifically interesting planets
circling around other stars. Mission targets, such as planets capable of harboring
life and/or habitable by humans, would be identified by the NASA Origins
Program. This Program has the long-range goal (by ca. 2040) of detection,
remote-sensing spectral analysis, and imaging of potentially habitable planets
around stars out to 40 LY, corresponding to a sphere containing roughly 350
stars [5]. (This total does not include solitary red dwarfs, which are generally
incapable of harboring habitable planets.) This will be accomplished by the
use of progressively more sophisticated, space-based, observational techniques
(e.g., telescopes, interferometers, etc.) to ultimately image Earth-like planets in
the potentially habitable region or the “Goldilocks zone” not too hot, not too
cold around a star. In the near term, however, there are a number of interesting
interstellar precursor scientific opportunities that could serve as intermediate,
near-term technology demonstration missions on the way to developing
propulsion systems capable of achieving the ultimate mission to the stars.
34 R. H. FRISBEE

Some examples of these precursor opportunities are shown in Fig. 1 [6]. These
include missions to the Heliopause (at 100 AU), the sun’s gravitational lens
(550 AU), and the Oort cloud (2000 AU to 1 LY, with 1 LY 63,300
AU). Order-of-magnitude DV and corresponding propulsion system Isp require-
ments for these missions are given in Table 1, where we have assumed that Isp,
expressed as exhaust velocity (Vexhaust) is comparable to the mission DV. (This
assumption is discussed in more detail in the following.) Figure 2 illustrates
the range of DV values typically encountered as a function of the trip time
desired. Figure 2 also illustrates the difficulty of performing even near-term inter-
stellar precursor missions; for example, advanced nuclear electric propulsion
(NEP) requires roughly 50 years to reach the outer edge of the Kuiper belt at
1000 AU. More ambitious missions require the use of fission, fusion, or antimat-
ter reactions as the propellant exhaust, or of photon momentum beamed to the
vehicle by, for example, massive lasers in the solar system.
Robotic interstellar missions can be viewed as a natural follow-on to the
Origins Program, which will tell us where to send the interstellar spacecraft to
provide close-up imaging with a flyby, and detailed in-situ science (“ground
truth”) with a rendezvous mission. Current emphasis is on a fast interstellar ren-
dezvous mission where the spacecraft stops at its destination. Thus, there is a
desire for a high cruise velocity to minimize trip time. For example, to travel
4.3 LY with a 10-year trip time requires an average speed of 0.43 c. However,
a high-speed (0.1 c) flyby is not thought to give significantly more science
return than that provided by Origins Program capability in the time frame of
interest; in effect, virtually as much imaging capability is provided by advanced
telescopes in the solar system as from a rapidly moving spacecraft in a flyby.
Also, telescopes in the solar system could monitor the target for an indefinite
amount of time; by contrast, a 0.1 c flythrough of, for example, our solar
system (with a diameter of 11 light-hours assuming Pluto’s orbit) would only
allow 110 hours of close-up observation of the target solar system. Thus we
see the need for a rendezvous mission, even though this has the effect of doubling
the mission DV. Finally, for eventual crewed missions to the stars, special empha-
sis is placed on minimizing trip time with rendezvous assumed to be a given.

Fig. 1 Scale of the interstellar medium (after Ref. 6).

 Table 1 Sample mission, DV, and Isp requirements

Slow interstellar Fast interstellar

Mission Planetary 100– 1000 AU 10,000 AU flyby rendezvous

Mission examples Inner solar system Heliopause (100 AU) Oort cloud 4.3 LY in 40 yrs 4.3 LY in 10 yrs
orbiters (2000 –60,000 AU)
Outer solar system Gravity lens (550 AU) 40 LY in 100 yrs
flybys
Slow outer solar Kuiper belt
system orbiters (40 – 1000 AU)
Typical DV 10 km/s 100 km/s 1000 km/s 0.1 c 0.5 c for flyby;
doubled for
rendezvous
Typical Isp
(lbf-s/lbm) 103 104 105 3  106 1.5  107
(km/s) 101 102 103 3  104 (0.1 c) 1.5  105 (0.5 c)
(ignores
relativistic
effects)
LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY
35
36 R. H. FRISBEE

Fig. 2 Mission velocity and trip time.

B. “Impossible” Missions and the Importance of Specific Impulse

and Stage Dry Mass
As we will show, the “Rocket Equation” describes the exponential increase in
the ratio of the “wet” mass, Mwet, of a vehicle (fully loaded with propellant)
divided by its “dry” mass without propellant, Mdry, as a function of the ratio
of velocity change, DV, divided by the propulsion system’s specific impulse,
Isp, or more properly, its exhaust velocity, where Vexhaust gc Isp (i.e.,
gc 9.80665 [m/s]/[lbf -s/lbm]).

Mwet =Mdry ¼ EXP[DV=(gc Isp )] ¼ EXP(DV=Vexhaust )

Note that in this chapter, we generally give Isp in “Imperial” or foot-pound-

second (fps) units of lbf-s/lbm because this is how Isp has been (and continues
to be) reported in the literature. This also avoids the potential for inappropriately
increasing or decreasing the number of significant digits in the value of Isp when
converting Isp from lbf-s/lbm to m/s; for example, converting a chemical rocket’s
quoted Isp of 452 lbf-s/lbm to 4,433 m/s implies an additional digit of precision
not evident in the original literature value.
Historically, Isp was defined as a rocket engine’s thrust divided by the rate of
propellant mass flowing into the engine. For example, a chemical rocket’s thrust
would be measured [e.g., in pounds of force (lbf) or newtons (N)] and the propel-
lant mass flow rate [e.g., pounds of mass per second (lbm/s) or kilograms per
second (kg/s)] derived, for example, from propellant volumetric flow rate
(e.g., gallons or liters per second) and the propellant’s density. Interestingly, in
the metric meter-kilogram-second (mks) system, Isp in N-s/kg reduces
directly to Vexhaust in m/s because the Newton is a compound unit, where
N kg-m/s2. Thus we see that Isp in the “right” units of velocity is simply
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 37

(and interchangeable) Vexhaust. To convert Isp in Imperial lbf-s/lbm (fps) units to

velocity, we multiply the value in lbf-s/lbm by a units conversion factor (gc) that
is numerically equal to the standard acceleration of gravity: 9.80665 [m/s]/
[lbf-s/lbm] 32.17405 [ft/s]/[lbf-s/lbm]. Also, note that gc is formally a unit
conversion factor and does not depend on the local gravitational field because
our unit systems are based on a standard Earth gravity. Finally, one often encoun-
ters Isp given in units of “seconds” when it should more properly be expressed as
lbf-s/lbm. This is because of the erroneous tendency to “cancel” units of lbf
and lbm in the lbf-s/lbm unit, leaving only seconds. In fact, if Isp is defined as
thrust divided by propellant weight (not mass) flow rate, then Isp can in fact
be correctly expressed as seconds. To avoid any confusion, we will generally
give Isp in lbf-s/lbm and Vexhaust in velocity units (e.g., km/s or a fraction of
the speed of light).
Examples of the increase in vehicle wet mass as DV increases (with a given Isp
or Vexhaust) are shown in Fig. 3 for a single-stage vehicle, where we can have an
“impossible” mission when DV is several times the propulsion system’s Isp or
Vexhaust. Thus, a chemical propellant like O2/H2, with an Isp on the order of
500 lbf s/lbm corresponding to a Vexhaust of about 5 km/s, would literally
need an ocean’s worth of propellant to achieve an interstellar precursor DV of
about 200 km/s. This is the primary reason why Isp is so strongly emphasized
in determining rocket performance. This is also the driver behind the use of
multi-stage vehicles so as to reduce the DV that must be delivered by each
stage, and thus reduce the mass ratio of each stage.
However, there is a second aspect of “impossible” missions that is a
consequence of the fact that any real propulsion system has a finite dry mass
(Mprop-dry) of engines, propellant tanks, structure, etc. For example, we can

Fig. 3 Increase in vehicle wet mass (Mwet) as a function of DV and Isp (s 5 lbf-s/lbm)
for representative propulsion systems (based on the Relativistic Rocket Equation,
with no loss of propellant). Calculations based on vehicle dry mass (Mdry) 5 1000 MT.
38 R. H. FRISBEE

write the vehicle dry mass as the sum of the payload and propulsion system dry
masses:

Mdry ¼ Mpayload þ Mdry prop

Further, as a simplification, we can also divide the propulsion system’s dry

mass into two components, a fixed mass term, Mfixed, independent of the propel-
lant mass, Mp, and a term dependant on propellant mass through a tankage
factor, TF:

Mdry-prop ¼ Mfixed þ TF  Mp

In the extreme limit of large Mp, the payload and propulsion system fixed
masses become negligible, and we have

TF(max)  Mdry =Mp

This represents the maximum possible value for TF; any nonzero payload or
propulsion system fixed masses would tend to lower the allowable tankage frac-
tion of the vehicle. Thus, we can recast the results shown above for a single-stage
vehicle and calculate the maximum possible TF based on the Rocket Equation as
a function of the propulsion system’s Isp and mission DV, as shown in Fig. 4.
From this we see that as DV grows, the maximum TF allowed by the Rocket
Equation decreases and approaches zero at high DV. In practical terms, it
becomes “impossible” in an engineering sense to build propellant tankage light
enough to allow the vehicle to reach a DV more than a few times its Isp (or
Vexhaust). For comparison, we have shown the TF for the space shuttle external
tank (ET), which holds roughly 700 MT of liquid oxygen (LO2) and liquid hydro-
gen (LH2) at an oxidizer-to-fuel (O/F) ratio of about 6. The ET by itself, an extre-
mely lightweight structure, has a TF of 4%. If we were to include the mass of the
space shuttle Orbiter, the “effective” TF would be about 18%, which not surpris-
ingly matches the curved line for the limit of maximum possible TF. (In this
example, the mass and DV of the solid rocket boosters are omitted.)
We can also see the effect of economy-of-scale by considering the two TF
points for the fission/fusion system. These points are based on the total propul-
sion system masses for the Daedalus two-stage fusion rocket designed by the
British Interplanetary Society for a 0.1 c interstellar flyby mission [7]. For this
vehicle, the fusion fuels were deuterium (D) and the mass-3 isotope of helium
(He3), with a D/He3 mass ratio of 2/3. The second stage had 4000 MT of pro-
pellant with a total propulsion system only (no payload) TF  Mdry-prop/Mp of
24.5%. By contrast, the first stage had 46,000 MT of propellant, with a propulsion
system TF of only 3.7%. Note that these two points are still considerably below
the fission/fusion curves because these data points for TF only consider the dry
propulsion system in each stage. By contrast, the curves correspond to the case of
considering the total Mdry for each stage (i.e., maximum TF  Mdry/Mp) with
each stage having Mdry Mdry-prop þ Mpayload. However, for stage 1, the
“payload” mass is all of the fully-loaded stage 2 with its payload.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 39

Fig. 4 Decrease in vehicle maximum possible tankage factor (TF  Mdry/Mp) as a

function of DV and Isp (“s” 5 lbf-s/lbm) for representative propulsion systems (based
on the Relativistic Rocket Equation with no loss of propellant).

As a final example of the inherent engineering “impossibility” of using a low-

Isp propulsion system for a high-DV mission, we can calculate the ratio of DV/Isp
(or, more formally, DV/Vexhaust) as a function of an individual stage’s total
TF Mdry/Mp. (From the Classical Rocket Equation, DV/Vexhaust ln(Mwet/
Mdry), and Mwet Mdry þ Mp.) As shown in Fig. 5, a total TF of about 15%
(representative of many chemical propulsion systems) gives a DV only twice
the Vexhaust of the propulsion system. This is the practical reason why most
state-of-the-art propulsion stages are limited to a DV of about twice the Vexhaust
(or corresponding Isp) of their propulsion system; for example, the Space
Shuttle has an Earth-to-orbit DV of 9 km/s for O2/H2 with a Vexhaust of
4.5 km/s (Isp 450 lbf-s/lbm). Even if we could somehow build a propulsion
system with a TF of 0.1%, the achievable DV for that system would still only be
6.9 times its Vexhaust.
Finally, note that we have ignored the effect of using multiple stages to deliver
some total DV; in practice, this is used to reduce the DV per stage and to increase
the allowable TF per stage. However, this approach still has its limitations,
because the “upper” stages are the “payload” of the “lower” stages. This produces
the “pyramid” effect seen in multi-stage rockets: the first stage is very large, with
subsequent stages becoming smaller and smaller. For example, if we have a
(DV per stage)/(Vexhaust) of 2, the Classical Rocket Equation mass ratio of
40 R. H. FRISBEE

Fig. 5 Ratio of DV/Vexhaust as a function of stage TF 5 Mdry/Mp (based on the

Classical Rocket Equation).

Mwet/Mdry per stage is exp(2) 7.38; thus, a three-stage rocket would have
an initial total wet mass of at least (7.38)3 403 times the final dry mass of
the third stage (including payload), even if stages one and two had a TF 0.

C. Technology Options Selection Process

Having established the requirements for our fast, interstellar rendezvous
Vision Mission (i.e., Vcruise 0.5 c, corresponding to a total mission DV of 1 c
for a rendezvous mission), we now seek to identify viable propulsion candidates
for interstellar missions [3]. To do this, we ask three questions that are used as
screening filters,‡ as shown in Fig. 6. Note that this screening process is used
to reduce the large possible number of propulsion options down to a manageable
few; in effect, we have used the Horizon Mission methodology [4] where we have
intentionally selected a set of mission requirements (from the fast, rendezvous
interstellar Vision Mission) that are so demanding that only a limited number
of propulsion options are applicable.

1. DV Capability
First, we ask the basic question of whether the propulsion system has the capa-
bility of providing the required DV for the mission. As described previously, the
Rocket Equation suggests that it is desirable to have the mission DV and propul-
sion system specific impulse (Isp) or exhaust velocity (Vexhaust) comparable in size
to prevent excessive propellant requirements. (For our Vision Mission case, with
Vcruise 0.5 c, this implies that DV 0.5 c 1.5  105 km/s  Vexhaust, corre-
sponding to Isp 15  106 lbf s/lbm.)
This evaluation criterion quickly eliminates chemical, advanced electric pro-
pulsion (EP) and electromagnetic (EM) catapult launchers [8] from consideration

‡
The original “filtering” concept was introduced by H. Harris, Jet Propulsion Laboratory, 1997.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 41

Fig. 6 Fast (Vcruise 5 0.5 c) interstellar rendezvous mission propulsion option

screening process.

for fast interstellar rendezvous missions. Similarly, fission fragment [9,10] and
fusion propulsion [3] concepts have a maximum Isp on the order of 1  106
lbf s/lbm (i.e., Vexhaust 0.03 c), which is too low for a fast rendezvous
mission. However, a two-stage fusion rocket has been shown to be a reasonable
choice for a slow, 0.1 c interstellar flyby [7]. The one loophole to fusion Isp or
Vexhaust limitation is the Bussard interstellar fusion ramjet [11] that collects inter-
stellar hydrogen (H) or deuterium (D) for use in a fusion rocket. Because all of the
propellant required for the mission is not carried on board the vehicle, the inter-
stellar ramjet effectively “cheats” the Rocket Equation and is capable of supply-
ing unlimited DV. Solar sails also “cheat” the Isp limitations of the Rocket
Equation; however, even with an ultra-low mass per unit area (areal density),
they cannot achieve the required velocities because of the 1/R 2 drop-off in sun-
light intensity (i.e., decrease in photon momentum “push” per unit area) on the
sail [3]. Light sails overcome the 1/R 2 limitation of a solar sail by using laser
or microwave power (actually momentum) beaming. However, although micro-
wave light sails (e.g., Starwisp [12]) can be used for interstellar flybys, they
cannot be used for rendezvous missions because the long wavelength of
microwaves (compared to near-visible laser wavelengths) results in impossibly
large optics requirements for focusing the microwaves at interstellar distances.
Thus, only laser light sails (with a laser near-visible wavelength on the order
of 1 mm) [2,13], matter antimatter annihilation rockets (with an Isp of 1  107
lbf-s/lbm corresponding to a Vexhaust 0.33 c) [3,14], and fusion ramjets [3]
strongly pass the DV filter, with fission fragment and fusion propulsion (both
with a Vexhaust of 0.03 c) weakly passing the filter.
42 R. H. FRISBEE

2. Infrastructure Requirements
The second evaluation criterion deals with the potential need for a large, poss-
ibly space-based supporting infrastructure that is unique for the propulsion
concept. The assumption here is that this infrastructure would represent a signifi-
cant up-front cost that typically would have limited application beyond the inter-
stellar mission.
For example, the fission fragment propulsion concept would require the
construction of a unique facility (ground- or space-based) to produce large
amounts of short-lived, high-energy, highly-fissionable nuclear fuels such
as americium (Am) or curium (Cm) [9]. Similarly, a relativistic particle
beam that would “push” a magnetic sail (e.g., MagSail) [15], analogously
to the laser-driven light sail, would require an enormous space-based particle
beam facility that would have limited applicability beyond in-space
transportation.
By contrast, “pure” matter antimatter annihilation propulsion, where all
of the propulsive energy comes from the annihilation reaction, will require
major new antiproton production facilities to supply the tons of antimatter
required for interstellar missions. However, it must be noted that there are
a number of dual-use spin-offs of antiproton research, such as medical appli-
cations (e.g., imaging and destruction of cancer tumors in the 1-mm size
range) [16], that could justify the infrastructure investment. Similarly, laser
(or microwave) light sails would require a major space-based infrastructure
consisting of the beam source and the associated optics, but the beamed-
energy infrastructure has the unique capability of multiple use as a
time-shared power and propulsion source as, for example, a “public utilities
in space,” with a grid of laser/microwave beams supplying power in space
analogous to the electric power and natural gas utilities on Earth.
Thus, the fission fragment and particle beam/magsail concepts strongly fail
the infrastructure test. Matter antimatter and beamed-energy light sail propul-
sion concepts only weakly fail this test, either because of the potential for
multiple in-space or spin-off applications. Therefore, only fusion, matter
antimatter annihilation, and light sail propulsion will be carried on to the third
evaluation criterion, technology requirements.

3. Technology Requirements
Our third and final criterion relates to the current technology level and future
technology development needs of the various systems. Not surprisingly for an
interstellar propulsion system, the technology requirements for all of the three
leading candidates will be formidable. Note that all of the concepts have numer-
ous uncertainties and major unresolved feasibility issues; there is no clear winner.
Rather, the challenge is to identify the approach that has the fewest number of
developmental and operational “miracles” required for its implementation. Ironi-
cally, the interstellar fusion ramjet has the greatest performance potential but also
the greatest number of technology challenges. However, from our perspective
today, all three are equally “impossible”; only continued research and analysis
will identify which are less “impossible” than others.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 43

III. Fundamentals of Interstellar Flight

In this section, we will develop the governing equations for interstellar mis-
sions with regard to the propulsion systems. First, we will outline the derivation
of the “Classical” and “Relativistic” Rocket Equation, with particular attention to
the case of a matter antimatter annihilation propulsion system where a large
fraction of the initial propellant mass is “lost” during the annihilation process
and is thus not available to impart forward momentum to the vehicle. Next, we
will describe the equations governing the behavior of laser-pushed sails (light
sails), where photon momentum is used to accelerate the sail. We will also
briefly discuss some of the issues associated with the Bussard interstellar
ramjet. Next, we will discuss the impact of vehicle acceleration (and deceleration
at the target), and how too low an acceleration can dramatically impact the total
mission flight time. Finally, we will outline an approach to ameliorating the
effects on an interstellar vehicle of ultra-high speed (relativistic) impacts with
interstellar dust grains.

A. Rocket Equation
We will present the results of the derivation of four versions of the Rocket
Equation; details of the derivations are given in the Appendix to this chapter.
The first is the Classical Rocket Equation, first derived by Tsiolkovsky [17].
The next is a Relativistic Rocket Equation, in which both the vehicle and the
propellant exhaust velocity (Vexhaust or equivalently Isp) are moving at a large
fraction of the speed of light, such that relativistic corrections must be included.
These derivations will follow Forward’s method [18]; an alternative mathemat-
ical formulation of relativistic systems is discussed in Chapter 14. The next
two versions of the Rocket Equation parallel the Classical and Relativistic
Rocket Equations, but for the situation where some large fraction of the initial
propellant mass or mass-energy is unavailable (e.g., 60% for a proton-antiproton
annihilation rocket) for momentum transfer to the vehicle.
1. Classical Rocket Equation
In the Classical Rocket Equation derivation, a rocket with initial mass M ejects
a mass of reaction mass (propellant) dm at a constant exhaust velocity
(i.e., Vexhaust corresponding to Isp) relative to the rocket with a velocity, as
shown in Fig. 7. In the center of mass of the system, the resultant rocket velocity
is V (i.e., DV), and the reaction mass velocity is u. Note that we have shown the
situation for both acceleration and deceleration; derivation of the Rocket
Equation will be shown for both options. For convenience, we will adopt a
sign convention for acceleration and deceleration using expressions like x
“þ/ ” y or x “ /þ” y, where the “upper” sign is for acceleration and
the “lower” sign is for deceleration. Thus, the propellant velocity in the center
of mass of the system can be written as u w /þ V, corresponding to
u w V for acceleration or u w þ V for deceleration.
To begin the derivation of the Classical Rocket Equation, we first start with the
conservation of mass:

dM ¼ dm
44 R. H. FRISBEE

Fig. 7 Classical rocket mass, velocity, and momentum (with no loss of propellant).

The minus sign indicates the decreasing mass of the rocket. Also, the two
equations are the same for acceleration or deceleration. Next, we conserve
momentum:

d(M  V) ¼ þ=u  dm

Note the sign difference between the acceleration and deceleration forms of the
equation. Finally, we use the classical addition (or subtraction) of velocities:

u ¼ w= þ V

Again, note the sign differences. We then use the conservation of mass equation
to find dm; this identity will be then be used when we expand the conservation
of momentum equation [i.e., d(MV ) MdV þ VdM]. Thus, expanding the deri-
vatives, combining the above equations, and rearranging gives the derivative
form of the equation to be analytically integrated:

dM=M ¼ = þ dV=w

If we were performing a numerical integration, we could write: dM M

[ /þ dV/w]. Analytical integration gives:

‘n(M) ¼ = þ V=w

which, when evaluated for the rocket mass limits of Mwet (initial wet mass)
and Mdry (final “burnout” dry mass), initial and final velocities (Vi and Vf, with
Vf Vi DV), and w Vexhaust gcIsp, [with gc 9.8 (m/s)/(lbf -s/lbm)],
yields the Classical Rocket Equation:

‘n(Mwet =Mdry ) ¼ þ=DV=½gc Isp 

Mwet =Mdry ¼ exp (þ=DV=½gc Isp )
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 45

Note that for acceleration, DV is positive (DV . 0) since Vf . Vi.

However, for deceleration, DV , 0 since Vf , Vi. Thus, both equations give
the expected result that Mwet . Mdry independent of whether the vehicle is
accelerating or decelerating.

2. Relativistic Rocket Equation

For the Relativistic Rocket Equation, we follow essentially the same steps as
for the Classical Rocket Equation, with the difference being the use of relativistic
equations for mass and velocity, and conservation of mass energy content of the
system (as opposed to conservation of mass only in the classical case). Details of
the derivation are given in the appendix to this chapter.
First, we start with the conservation of mass energy; i.e., mass energy, or
mc2, is conserved. Also, we use relativistic mass, such that:
1=2
M ¼ Mr =(1  V 2 =c2 )1=2 and m ¼ mr =(1  u2 =c2 )

where Mr and mr are the rest masses of the vehicle and propellant, respectively.
Note also that the mass energy conservation equation is the same for accelera-
tion or deceleration. Next, we conserve momentum, for example, Mr/(1 V 2/
c 2)1/2 * V. Finally, we use the relativistic addition (or subtraction) of velocities:

u ¼ (w  = þ V)=(1  = þ wV=c2 )

To begin the derivation, we expand the mass energy conservation derivative

and solve for dmr. Next, we expand the conservation of momentum derivative,
and substitute values for dmr and u from above. After considerable algebraic
manipulation (detailed in the appendix), we obtain the derivative form of the
equation to be analytically integrated:

dMr =Mr ¼ dV=½þ=  w(1  V 2 =c2 ) ¼ c2 =(þ=  w)  ½dV=(c2  V 2 )

For numerical integration, we would have:

dMr ¼ Mr {dV=½þ=  w(1  V 2 =c2 )}

dMr ¼ Mr {c2 =(þ=  w)  ½dV=(c2  V 2 )}

Analytical integration gives:

‘n(Mr ) ¼ c=(þ=  2w)  ‘n½(c þ V)=(c  V)

¼ 1=(þ=  2w=c)  ‘n½(1 þ V=c)=(1  V=c)

Evaluating the integrals for the initial and final rest-mass limits of Mwet
and Mdry, initial and final velocities of Vi and Vf, with Vf Vi DV, and
46 R. H. FRISBEE

w Vexhaust gc Isp, yields the Relativistic Rocket Equation:

‘n(Mwet =Mdry )
¼ c=(þ =  2w)  ‘n{½(c þ Vf )=(c  Vf )=½(c þ Vi )=(c  Vi )}
¼ 1=(þ =  2w=c)  ‘n{½(1 þ Vf =c)(1  Vi =c)=½(1 þ Vi =c)(1  Vf =c)}
¼ 1=(þ =  2gc Isp =c)  ‘n{½(1 þ Vf =c)(1  Vi =c)=½(1 þ Vi =c)(1  Vf =c)}
Mwet =Mdry
2gc Isp =c)
¼ {½(1 þ Vf =c)(1  Vi =c)=½(1 þ Vi =c)(1  Vf =c)}½1=(þ=

Note that we have dropped the subscript “r” denoting rest mass for
convenience.

3. Classical Rocket Equation with “Loss” of Propellant

Our goal is to derive a Relativistic Rocket Equation for a matter antimatter
annihilation rocket where a significant fraction of the initial propellant mass,
Mp, is converted into forms that are not usable for thrust production (e.g.,
high-energy gamma ray photons). However, this derivation is complicated by
the need to conserve the mass energy content of the various reactants and
annihilation products. As an introduction to this problem, we will first perform
a derivation of the Classical Rocket Equation with “loss” of propellant where
we only need to conserve mass and momentum; in this case, we assume that a
fraction “a” of the initial propellant mass is used to produce forward motion of
the rocket, as shown in Fig. 8. In effect, a fraction (1 a) of the initial propellant
mass (and momentum) is “lost,” at least insofar as production of forward momen-
tum is concerned. It is as if a quantity (a)Mp is ejected parallel to the direction of
motion (to produce some DV) and a quantity (1 a)Mp is ejected perpendicular

Fig. 8 Classical rocket mass, velocity, and momentum (with “loss” of propellant).
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 47

in such a way that (1 a)Mp does not contribute to the forward DV. Finally, note
that the usual situation for rocket propulsion is for a 1; for example, even in a
fission or fusion rocket, where some of the rest mass of the nuclear fuel is turned
into energy, the amount of mass “lost” (to conversion into energy) is negligible,
so the “standard” Classical or Relativistic Rocket Equations (with a 1) can
be used.
For the Classical Rocket Equation with “loss” of propellant, we follow
essentially the same steps as for the standard (a 1) Classical Rocket Equation,
with the difference being the use of terms in (a) and (1 a) in the momentum
conservation equation. Thus, we again write the equations for conservation
of mass and momentum, and addition (or subtraction) of velocities. (Note
again the use of the “þ/ ” and “ /þ” sign convention for acceleration
and deceleration.)

dM ¼ dm ¼ {(a)dm ½usable þ (1  a)dm ½lost}

Note that a * dm represents the “usable” fraction of the (incremental)

propellant mass, whereas (1 a) * dm represents the fraction “lost” and
unavailable for production of velocity (V). For conservation of momentum,
we have:

dðM  VÞ ¼ þ= u  a dm ½only a dm contributes to V

 Vð1  aÞ dm ½momentum ‘‘lost’’]

As above, only u * a * dm contributes momentum toward changing V. Also, in

defining w and u, we are only interested in the “bulk” or center-of-mass velocity
of the (incremental) propellant mass quantity a * dm moving parallel to V. Thus,
the exhaust velocity (w Vexhaust) or Isp of the rocket engine corresponds to an
“effective” Isp or Vexhaust that takes into account any inefficiencies in the
nozzle in converting the random velocities of the propellant exhaust gas into
directed motion (w and u) parallel to V.
Similarly, the (incremental) propellant mass (1 a) * dm is ejected at some
exhaust velocity (not necessarily w) perpendicular to V (and u) in a symmetric
fashion that results in a net cancellation of the perpendicular velocity of the
center-of-mass of (1 a) * dm. However, this unit of mass is still moving
forward at the rocket’s velocity, V, so the total amount of “lost” momentum is
V * (1 a) * dm. Interestingly, this momentum is totally lost from the system
(its value is negative for either acceleration or deceleration), so it also reduces
the mass and momentum of the rocket that must be accelerated or decelerated
in subsequent differential time steps. However, it also represents a “dead”
weight that must be initially loaded into the rocket, which increases Mwet over
that seen when a 1.
As before, we expand the conservation of mass equation, solve for dm, and
then expand the conservation of momentum equation and substitute dm and u
as appropriate to get the equation to be analytically integrated:

dM=M ¼ = þ dV=(a w)
48 R. H. FRISBEE

If we were performing a numerical integration, we could write:

dM ¼ M½= þ dV=(a w)
Analytical integration gives:

‘n(M) ¼ = þ V=(a w)

and again when evaluated at the integration limits gives:

‘n(Mwet =Mdry ) ¼ þ=  DV=(a gc Isp )

Mwet =Mdry ¼ exp½þ=  DV=(a gc Isp )

Note that the final form of the equation is similar to that of the standard (a 1)
Classical Rocket Equation. The result has been to add a to the Isp term, resulting
in an “effective” Isp that takes into account the “loss” of propellant. In fact, this
result could be anticipated from the classical definition of Isp in terms of thrust (F)
and propellant mass flow rate (ṁ) entering the rocket engine:
_ ¼ Vexhaust
gc Isp ¼ F=m
In this equation, gc 9.8 [m/s]/[lbf-s/lbm], Isp lbf-s/lbm, F Newtons,
ṁ kg/s, and Vexhaust m/s. In the context used here, ṁ is the total mass of
propellant entering the rocket engine. However, only a quantity a * ṁ is available
to produce thrust, so we would have:
_ ½Total)
gc Isp ¼ F=(a m
which rearranges to give the same [a gc Isp] term seen in the Rocket Equation
above:
_ ½Total
a gc Isp ¼ F=m

4. Relativistic Rocket Equation with “Loss” of Propellant

For this derivation, we will combine elements of the derivations of the Rela-
tivistic Rocket Equation and Classical Rocket Equation with “loss” of propellant.
However, unlike the classical version of this equation, where we conserve only
(rest) mass and classical momentum (rest mass multiplied by velocity), the rela-
tivistic version requires that we conserve the total amount of mass energy and
relativistic momentum, including that of the “lost” propellant. Thus, because
of the need to include the relativistic mass energy and momentum of the
“lost” propellant, the algebra will be more complex, resulting in an equation
that must be numerically integrated.
Also, for a matter antimatter annihilation reaction, we need to bookkeep
the rest mass and kinetic energy of the various reactants (protons, antiprotons,
electrons, positrons) and annihilation products (charged and neutral pions,
gamma ray photons). This is needed to determine both the effective value of
“a” in the equation (i.e., how much of the mass energy and momentum are
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 49

usable for propulsion), as well as to determine the mass energy and momen-
tum of the various annihilation products. Table A1 in the appendix summarizes
mass and energy quantities based on the assumption of a reaction between an
atom of hydrogen (H), containing a proton (Pþ) and electron (e ), and an anti-
atom of anti-hydrogen containing an antiproton (P ) and positron (eþ) [14].
The protons and antiprotons annihilate to give neutral and charged pions
(po, pþ, p ); the electrons and positrons annihilate to give high-energy
photons (gamma rays, g). The initial annihilation reaction is given as [19]:

Pþ þ P ! 2:0po þ 1:5pþ þ 1:5p and e þ eþ ! 2 g

The neural pions promptly decay into very high-energy gamma rays (each
about 355 times more energetic than an electron/positron gamma). The
charged pions decay into charged muons (m) and neutrinos (n) after traveling
about 21 meters at 0.93 c [18,19].

2:0p8 ! 4g, 1:5pþ ! 1:5 mþ þ 1:5 nm , 1:5p ! 1:5 m þ 1:5 anti-nm

To determine the value for a, we first determine the fractions (percentages)

of the total initial mass energy contained in the various annihilation species.
For the charged pions (pþ, p ), we can further split the total mass energy
content into rest mass (mr) and kinetic energy, such that:

Rest massenergy ¼ mr c2
Kinetic energy ¼ mr c2 {1=(1  v2 =c2 )1=2  1}
Total massenergy ¼ mr c2 =(1  v2 =c2 )1=2

This results in the charged pions having a rest mass that is 22.3% of the total
mass energy content of the reactants (H and anti-H atoms), and a kinetic
energy that is about 39.9% of the initial mass-energy. However, even though
the total mass energy content of the charged pions is about 62.2% of the
total mass energy content of the reactants (H and anti-H atoms), this is not
the value of a. This is because the electromagnet used to deflect the charged
pions is not perfectly reflective (some charged pions “leak” through the mag-
netic field and travel upstream), and for those that are reflected, the resultant
beam is not perfectly collimated. This is illustrated in Fig. 9 [14,20], where
straight-line particle trajectories are the gamma rays (from neutral pion
decay) and non-linear particle trajectories are the charged pions deflected by
the magnetic field from a single-loop magnet (shown superimposed on the
annihilation region).
Thus, there is an inherent kinetic energy inefficiency in the nozzle that
must also be taken into account. Callas [20], using Monte Carlo simulations
of particle trajectories in proton/antiproton annihilations, estimated that the
magnetic nozzle would be at best 50% kinetic energy efficient, corresponding
to a value of a 42.2%. We have arbitrarily assumed a value of a 40% of
the initial total mass energy content of the propellant, corresponding to a
50 R. H. FRISBEE

Fig. 9 Proton-antiproton annihilation sample Monte Carlo simulation results.

44.37% nozzle kinetic energy efficiency. These values are summarized in

Table A2 in the appendix. Note that the magnetic nozzle does not affect
the fraction of rest mass of the charged pions (22.3%); that remains the
same. Instead, what we have done is take into account the nozzle inefficiency
in converting the kinetic energy of the charged pions, whose 0.9334 c velocity
vectors are pointed in random directions, into a directed, collimated stream of
charged pions with a 0.6729 c kinetic energy velocity vector pointed directly
out the back of the rocket.
Finally, the discussion above has concentrated on the efficiency of the annihil-
ation reaction and electromagnet nozzle in capturing the total mass energy (rest
mass and kinetic energy) of the charged pions. However, there are additional
losses in the magnetic nozzle that result in a reduction in overall Isp. For
example, although the magnetic nozzle can capture 40% of the reactant’s
initial mass energy for thrust production, the “beam” of charged pions leaving
the engine are not perfectly collimated (i.e., parallel to V). Using the Monte
Carlo particle trajectories, Callas [20] used direct momentum calculations
to determine an “effective” Isp or Vexhaust of 1  108 N-s/kg m/s 0.333 c
1.02  107 lbf-s/lbm.
To better understand the distinction between mass energy content and
efficiency (i.e., a) and nozzle efficiency (i.e., Isp), it may be useful to consider
a chemical rocket engine as an analogy. For example, in the chemical
propulsion world there are computer codes that calculate the thermochemical
properties (enthalpy, combustion temperature, mole fractions, etc.) of the
various reaction products from a given set of propellants (e.g., from hydrazine
decomposition or oxygen hydrogen combustion). Sophisticated versions of
these codes can also account for inefficiencies and losses, like incomplete
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 51

combustion due to finite reaction rates (e.g., kinetics), energy losses to the
walls of the reaction chamber, film cooling, and so on. This corresponds to
the calculation of a for the annihilation reaction. For a chemical rocket, the
computer program then calculates the isenthalpic expansion of the reaction
products through a nozzle with a given expansion ratio and pressure con-
ditions (e.g., expansion into sea-level pressure or vacuum at the nozzle
exit). Again, depending on the sophistication of the program, various losses
can be calculated, in addition to the inherent limitations based on expansion
ratio and nozzle back pressure. This corresponds to the calculation of the
magnetic nozzle’s Isp [20].
To begin our derivation, we again begin with conservation of mass energy
and momentum, expand the conservation of mass equation, solve for dmr, and
expand the conservation of momentum equation and substitute dmr and u as
appropriate to get the equation to be integrated. This gives us (after considerable
algebraic manipulation as described in the appendix):

dMr =Mr ¼ dV=(1  V 2 =c2 )  ½(1  V 2 =c2  X)  = þ wV=c2 (1  X)=

½V  (1  X) þ =  w  (X  V 2 =c2 )

where we have combined the various terms containing a into the quantity X:

X ¼ ½a(1  u2 =c2 ) 1=2

 (1  a)=(þ=  u)  (1  V 2 =c2 ) 1=2
 V=
½a(1  u2 =c2 ) 1=2
þ (1  a)(1  V 2 =c2 ) 1=2


Note that when a 1, X 1, and this equation reduces to the form seen for
the standard (a 1) Relativistic Rocket Equation:

dMr =Mr ¼ dV=(1  V 2 =c2 )=½þ=  w ¼ dV=½þ=w(1  V 2 =c2 )

Unfortunately, when a or X are not 1, the various terms in V do not cancel out,
leaving us with an equation that must be numerically integrated. For this, we use
the equation in the form:

dMr ¼ Mr  fdVMr =(1  V 2 =c2 )  ½(1  V 2 =c2  X)

 = þ wV=c2 (1  X)=½V  (1  X) þ =  w  (X  V 2 =c2 )}

where X is defined above and u (w /þ V )/(1 /þ wV/c 2).

5. Comparison of the Rocket Equations

Figure 10 illustrates a comparison of the various Rocket Equations. Not sur-
prisingly, the standard (a 1) Classical and Relativistic versions do not
diverge significantly until the vehicle’s final velocity becomes a significant frac-
tion of the speed of light. We also see the dramatic impact that reductions in a
have to the vehicle mass ratio (Mwet/Mdry). For example, if an antimatter
rocket could have a value of a 1, its mass ratio would only be 2.2 to accelerate
52 R. H. FRISBEE

Fig. 10 Comparison of classical and relativistic rocket equations.

from V 0 to V 0.25 c. However, with a value of a 0.4, the mass ratio rises
dramatically to 6.4. In practical terms, in order to keep the mass ratio to a reason-
able value, a relativistic antimatter rocket, even with an exhaust velocity of 0.3 c,
must limit its DV per stage to no more than 0.25 c, because of the loss of so much
mass energy from the initial mass energy content of the propellant (i.e.,
1 a 0.6). Thus, we would need to have a four-stage vehicle for a fast
(Vcruise 0.5 c) interstellar rendezvous mission, with stages 1 and 4 performing
the “low” speed acceleration (0.00 ! 0.25 c stage 1) or deceleration
(0.25 ! 0.00 c stage 4), and stages 2 and 3 doing the “high” speed acceleration
(0.25 ! 0.50 c stage 2) or deceleration (0.50 ! 0.25 c stage 3). Finally, note
that classical and relativistic acceleration and deceleration are symmetric with
respect to each other; for example, the mass ratio is the same when accelerating
from 0.00 ! 0.25 c or when decelerating from 0.25 ! 0.00 c.§
Figure 11 illustrates a second effect seen for the Relativistic Rocket Equations
where the actual values of Vi and Vf, and not just DV, are important. Because of
relativistic effects, the mass ratio will be different for a relativistic rocket accel-
erating from 0.00 to 0.25 c compared to one accelerating from 0.25 to 0.50 c (i.e.,
both with the same DV 0.25 c); at higher initial and final speeds, the mass ratio
is significantly higher, reflecting the greater impact that relativistic effects have at
higher velocities. This effect is completely absent from the Classical Rocket
Equations where only DV is relevant.
Thus, to have two relativistic stages reach 0.5 c and have the same mass ratio,
the first stage would need to have a DV greater than 0.25 c, and the second stage a
DV less than 0.25 c. For example, to accelerate from 0.0 to 0.5 c, stage 1 would

§
This is exactly true for the standard, a ¼ 1 Relativistic Rocket Equation, as can be see from its
analytically integrated form. This appears to also be true when a ¼ 0.4 because the numerical differ
ence between the mass ratio for acceleration and deceleration is of the same order for the numerically
integrated Relativistic Rocket Equations for both a ¼ 1 and a ¼ 0.4.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 53

Fig. 11 Comparison of Classical and Relativistic Rocket Equations for a two-stage

vehicle with DV 5 0.25 c per stage and Vi 5 0.00 c versus Vi 5 0.25 c.

have a DV of 0.268 c (starting from 0.000 c), stage 2 would have a DV of 0.232 c
(starting from 0.268 c), and both stages would have the same Mwet/Mdry 7.312.
By contrast, classical rocket stages could both have a DV of 0.25 c and have the
same mass ratio. In effect, relativistic effects make acceleration (or deceleration)
from a higher initial velocity “harder” because the relativistic mass increase
applies to the total loaded vehicle, so more propellant is used and the mass
ratio is higher.

B. Relativistic Light Sail Equations

As discussed above, solar sails are unable to provide the performance
(acceleration and peak velocity) needed for interstellar missions. This is
primarily driven by the 1/R 2 drop-off in sunlight intensity, and thus the photon
momentum “push” on the sail, as the solar sail moves away from the sun. The
classic solution to this problem is to use a laser or microwave source to drive
the sail; in this mode, the sail is referred to as a generic “light sail” or laser
sail/microwave sail as appropriate to the wavelength of the source. Note that
the primary advantage of visible or near-visible laser wavelengths is the ability
to focus the laser beam at interstellar distances, given sufficiently large transmit-
ter (laser) and receiver (light sail) optics. In this section we will follow Forward’s
development of the mathematical relationships used to determine light sail size
(based on the assumption of a diffraction-limited optics system), laser power
requirements (based on the light sail mass, acceleration, and optical properties
such as reflectivity and absorbtivity), light sail mass (based on its mass per
unit area or “areal density”), light sail thermally-limited maximum acceleration
(based on sail areal density, optical properties including emissivity, and
maximum operating temperature) [13].
We will also use Forward’s method [13] of using a “two-stage” light sail for an
interstellar rendezvous mission, as shown in Fig. 12. In this approach, a large-
diameter laser transmitter and light sail first stage is used to accelerate the
vehicle (first and second stage) away from Earth. As the vehicle approaches
54 R. H. FRISBEE

Fig. 12 Laser-driven light sail interstellar rendezvous mission.

the target star, the first stage and the smaller-diameter second stage (with one-
tenth the area of the first stage [13]) separate, and the first stage is used to
retro-reflect the laser beam back onto the second stage. This has the effect of
slowly accelerating the first stage (because of its larger mass) out of the target
solar system, but rapidly decelerating the second stage (because of its smaller
mass) and bringing it to a stop in the target solar system. The need to focus
the laser onto the first stage at interstellar distances drives the overall size and
scale of the mission, as is shown next.

1. Light Sail Diameter

If we assume the limiting (best) case of diffraction-limited optics, the diameter
of stage 1 receiving the laser beam, Dr, is determined by the diameter of the laser
transmitting optics, Dt, the wavelength of the laser light, l, and the distance
between the transmitter and receiver, L:

Dr ¼ 2:440lL=Dt
The receiver’s diameter Dr (and the constant 2.440) corresponds to the first
minimum (null) in a circular diffraction pattern; the bright central “spot” inside
this null ring is called the “Airy Disk” and contains 83.78% of the total beam
power (with the remaining power contained in bright diffraction rings outside
Dr). For example, with a laser wavelength of 1 mm, a 1000-km diameter transmit-
ter, Dt has a “spot” size (airy disk) Dr of 1000 km at 43 LY; similarly,
Dt Dr 316 km at 4.3 LY.

2. Laser Beam Power

The laser beam power required to push the light sail (Plight sail) is directly pro-
portional to the light sail’s total mass including payload, Mlight sail; acceleration,
a; and speed of light, c; and inversely proportional to its reflectivity, h, and
absorbtivity, a, at the laser beam wavelength}:

Plight sail ¼ Mlight sail ac=(2h þ a)

}
The absorbtivity term (a) in the denominator of this equation was omitted in Forward’s original
derivation [13]; this was corrected by him in later papers.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 55

The total laser transmitted beam power, Plaser, required to deliver Plight sail to
the light sail is dependent on the fraction of the laser beam intercepted by the sail
(“spot size” efficiency, hspot 83.78% for a diffraction-limited optics system):

Plaser ¼ Plight sail =hspot

3. Light Sail Mass and Areal Density

The sail itself (without payload) is assumed to consist of three major elements,
the sail sheet, sail structure, and sail high-emissivity, 1, coating (to aid in reject-
ing any power absorbed from the laser input power):

Msail only ¼ Msheet þ Mstructure þ Mhigh-1 coating

Dividing each mass term by the total sail reflecting area (Ar p D2r /4) gives a
corresponding areal density (s M/A, typically expressed in g/m2):

ssail only ¼ ssheet þ sstructure þ shigh-1 coating

The total light sail mass and corresponding areal density with payload
are thus:

Mlight sail ¼ Msail only þ Mpayload

slight sail ¼ ssail only þ spayload

For our analyses [2], we have diverged slightly from Forward’s original
assumption that the total light sail areal density (including payload) would be
slight sail 0.1 g/m2 independent of sail size [13]. Instead, we have arbitrarily
assumed a 100 MT robotic spacecraft payload and fixed the interstellar rendez-
vous sail-only areal density at ssail only 0.1 g/m2 (independent of sail size).
With large sails, this assumption results in a payload areal density that
approaches zero; more generally, it allows us to treat the sail by itself and add
an arbitrarily large payload for mission analyses. Also, as discussed in more
detail below, this allows us to add a dedicated high-emissivity layer of heat-
rejecting material to the backside of the sail (the side facing away from the
laser) so as to reject any heat absorbed by the sail and thus allow higher laser
powers and correspondingly higher acceleration. Thus, if we assume a total
sail-only areal density of 0.1 g/m2 for the interstellar rendezvous light sail, we
can then subtract out Forward’s values [13] for the sail sheet areal density
(0.043 g/m2 for a 16-nm thick Al 63 Al atoms thick! sheet) and sail structure
areal density (0.030 g/m2 for a spinning sail), leaving an areal density of
0.027 g/m2 for the high-emissivity backside coating.

4. Thermally Limited Light Sail Acceleration

One of Forward’s important insights was that, although high acceleration is
generally desirable to minimize trip time, high acceleration is absolutely essential
during the rendezvous portion of a light sail mission to minimize the separation
56 R. H. FRISBEE

distance between stage 1 and stage 2. This minimizes the optical requirements on
stage 1 as it is used as a transmitter to retro-reflect laser beam power onto stage 2
[13]. In principle, this could be achieved by simply increasing the laser power
hitting stage 1 of the light sail. Because stage 2 has 1/10 the area and thus
mass of stage 1 [13] (assuming slight sail is the same for both stages), the laser
beam retro-reflected from stage 1 back onto stage 2 will cause 10 times higher
deceleration in stage 2 than acceleration in stage 1. In effect, the high deceleration
of stage 2 allows it to come to rest before the distance between the two stages
becomes so great that stage 1 can no longer effectively retro-reflect the laser
beam back to stage 2.
However, there is a thermal limit to the amount of laser power (or, more pre-
cisely, laser power intensity, i.e., I P/A W/m2) that can be used on the sail.
In effect, the power per unit area absorbed by the sail must be no greater than
what can be radiatively emitted by the sail. (This effect is equally true for
solar sails and limits their closest approach to the sun.) Thus, light sail accelera-
tion is ultimately limited by how much laser power (intensity) can be projected
onto the sail due to the sail sheet’s finite absorbtivity and material temperature
limits. This results in a thermally-limited acceleration that is a complex function
of sail sheet reflectivity h; absorbtivity, a; emissivity, 1; maximum temperature,
Tmax; and the total light sail areal density including payload, slight sail.
To find the thermally-limited maximum acceleration, amax, we first balance
the amount of power or P/A absorbed by the sail with that radiated into space
(from both sides of the sail):

(P=A) Absorbed ¼ (P=A) Radiated away

a (Plight sail =A) ¼ sSB (1 frontside þ 1backside )(Tmax 4  Tsink 4 )

where sSB is the Stefan-Boltzmann constant (5.67  10 8 W/m2K4), 1frontside

and 1backside are the emissivities on the front side (facing the laser) and back
side (facing away from the laser) of the light sail, respectively, and Tsink is the
background temperature of space (assumed to be 0 K). (Also, Tmax is in Kelvin
and slight sail is converted from g/m2 to kg/m2 for unit consistency with the
other quantities.) Substituting the definition of Plight sail into this equation
and solving for the acceleration (and using Mlight sail/A slight sail) gives the
thermally-limited acceleration (amax):

amax (m=s2 ) ¼ (2h þ a)sSB (1frontside þ 1backside )(Tmax 4  Tsink 4 )=(slight sail ca)

Thus, to achieve high acceleration or deceleration, we need a sail material with

low slight sail, high 1, high Tmax, low a, and high h.
Another very important result of Forward’s analyses was the identification of
an optimum sail sheet thickness that yields a maximum thermally-limited accel-
eration [13]. In his analyses, Forward found that for aluminum (Al), the optimum
sail thickness was 16 nm; at this thickness, Al has a h 0.820, a 0.135, and
transmittance t 0.045 (with h þ a þ t 1). Assuming the emissivity of
bare Al (1 0.06) on both sides of the sail sheet, and combining this with a
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 57

conservative maximum operating temperature of 600 K (only 64% of the melting

point of Al at 933 K) and a total light sail areal density of 0.1 g/m2, results in a
thermally limited maximum acceleration of 0.3865 m/s2 (0.03944 g). Under
these conditions, the maximum laser power intensity hitting the sail is 6.532
kW/m2 (4.838 “suns” with one sun 1.350 kW/m2). However, in order to
perform a rendezvous mission, a significantly higher acceleration is required.
For this application, Forward assumed the use of an advanced high-emissivity
coating on the backside of the sail sheet such that 1backside 0.95 (and an
1frontside 0.06 of bare Al); this allowed an increase in the thermally limited
sail acceleration to 3.253 m/s2 (0.3319 g). For our analyses, we will assume a
more conservative 1backside 0.60 (10 times 1frontside) which results in a ther-
mally limited acceleration of 2.126 m/s2 (0.2169 g) and a corresponding laser
power intensity on the sail sheet of 35.925 kW/m2 (26.611 suns).
Note that a high-1 backside coating was a somewhat arbitrary assumption
made by Forward (and this author) in order to have a light sail with a sufficiently
high acceleration to perform a rendezvous mission; thus, one significant
technology challenge for an interstellar rendezvous light sail is the need to
develop a low-s, high-1 coating capable of producing the required emissivity.
Similarly, the need for an extraordinarily thin sail sheet (16 nm), to produce a
low slight sail for either flyby or rendezvous missions is a major challenge in its
own right. One suggested approach to fabricating such an ultra-thin sheet is
to use in-space fabrication where Al vapor is sprayed [e.g., by chemical vapor
deposition (CVD), technologies] onto a relatively thick substrate (e.g., plastic).
The substrate then evaporates (sublimes) or decomposes in sunlight to leave
the Al sheet.
Finally, a diffraction-limited optics system produces a power (intensity) distri-
bution across the surface of the light sail with a central peak intensity 4.38 times
the average intensity [2]. This raises significant issues with respect to the sail
temperature at the center of the sail disk because the total laser beam power
hitting the light sail was based on a thermal limit for the average intensity
across the whole sail, and not the intensity in the center. There are several
approaches to ameliorating the additional intensity in the center of the light
sail [2]. For example, the simplest approach would be to just accept higher temp-
erature at the center of the light sail. In this case, the 4.38-times greater
intensity would increase the sail sheet absolute temperature by a factor of
(4.38)1/4 1.447, corresponding to an increase from the nominal 600 K to
868 K. Although this is below the melting point of Al at 933 K, this may still
not be feasible because, as Forward notes [13], ultra-thin films can fail by
agglomeration (i.e., thin films have a large surface-to-volume ratio, permitting
them to reduce their surface energy by forming droplets). For the case of Al,
this agglomeration temperature may be as low 725 K [13]. Because of these
temperature limitations for Al, other authors have suggested the use of higher-
temperature metals and nonmetals [21].

5. Relativistic Effects
For sails moving at 0.5 c, we need to explicitly consider relativistic effects on
sail acceleration due to changes in both mass and laser power (redshifting of the
58 R. H. FRISBEE

wavelength of the laser’s photons hitting the sail). Generally, these two relativis-
tic effects result in a reduction in acceleration due to an increase in the (relativis-
tic) mass of the light sail and a reduction in effective power (momentum) of the
photons driving the light sail. Also, the increase in wavelength of the redshifted
photons can impact the characteristics of the diffraction-limited retro-reflection
of light from the sail first stage back onto the second stage during the rendezvous
maneuver.
The standard equations for relativistic mass increase and wavelength
redshifting are:

M=Mr ¼ ½(1=(1  V 2 =c2 ))1=2 ¼ 1:15

l=lr ¼ ½(1 þ V=c)=(1  V=c)1=2 ¼ 1:73

where M relativistic mass, Mr rest mass (at V 0), l relativistic (red-

shifted) wavelength, lr wavelength at rest (V 0), and the numerical values
are given for V 0.5 c.
Finally, we have for the photons

E ¼ hn ¼ hc=l

where E energy (e.g., Joules), h Planck’s constant (6.626  10 34 J-s), and

n frequency (s 1 or Hz).
For example, at V 0.5 c, the light sail’s relativistic mass increases by a
factor of 1.15; this has a relatively small effect on overall acceleration perform-
ance. However, relativistic photon redshift represents a much larger impact than
the relativistic mass increase. For example, the wavelength (l) of laser photons
striking the light sail at 0.5 c will be increased (redshifted) by a factor of 1.73,
resulting in a corresponding decrease in the photon frequency, momentum, and
energy (and power) by a factor of 1/1.73. Similarly, during the interstellar
rendezvous, the photons retro-reflected from stage 1 back onto stage 2 experience
a second redshift factor of 1.73; in effect, the photons seen by stage 2 are
“double” redshifted, that is, 1/1.732 1/3.00. Also, the increased wavelength
of photons retro-reflected from stage 1 back onto stage 2 will cause the
diffraction-limited spot size on stage 2 to increase by a factor of 1.73. Finally,
note that for a constant power/constant wavelength transmitter, the overall
sail acceleration at 0.5 c will decrease by a factor of 1/(1.15 . 1.73) 1/2, as
compared to its acceleration at V 0, due to the combined relativistic mass
and wavelength increase. However, in our analyses, we have assumed that
laser power is “cheap,” and sized the laser system for the worst-case power
conditions of acceleration (for a flyby) or deceleration (for stage 2 rendezvous).


Note that relativistic effects were incorrectly ignored in the light sail mission analyses presented
in Frisbee and Leifer [3], even though they are discussed in Forward [13]. These errors were corrected
in Frisbee [2].
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 59

In part, this is based on the assumption that the laser system can be upgraded as
needed relatively easily because it is nearby within our solar system.
An alternative approach would be to vary the wavelength of the laser system to
compensate for relativistic redshifting. Generally, this is not easily implemented
in most lasers, although it is an option for the free electron laser (FEL). Also,
transmissions optics (e.g., adaptive optics) are typically optimized for a single
wavelength or narrow wavelength band. The issue remains, however, as to the
feasibility of large changes in wavelength (e.g., from 1 to 1/3 mm in the
example above) for the laser and its optics, so we have taken the simpler approach
of varying laser power as needed.

C. Bussard Interstellar Fusion Ramjet Mission Performance

and Momentum Drag
The interstellar fusion ramjet is still highly conceptual; however, it has the
potential advantage of providing unlimited range and mission flexibility and so
warrants continued study. Here, we will describe the rather unconventional
mission performance encountered when dealing with the interstellar fusion
ramjet, and describe the impact that the need to minimize momentum drag has
on the design of the interstellar hydrogen ram-scoop [22]. Finally, it is important
to note that the ram-scoop is not a physical structure, but a magnetic field, so even
though the “scoop” has a large frontal cross-sectional area to collect the low-
density interstellar fusion fuel, the mass of the scoop consists only of the
magnets (and their supporting structure) that project the magnetic field out in
front of the vehicle.

1. Interstellar Fusion Ramjet Mission Performance

The mission performance (acceleration, speed, etc.) of an interstellar fusion
ramjet has several unique aspects that are not typically encountered in more
conventional rocket systems (like an antimatter rocket). For example, the
mission can be broken into several steps depending on the propulsion mode
being employed. Assuming a constant-thrust (force) fusion engine, there is an
initial phase where the vehicle accelerates to the onset of ram-scoop operation,
typically 0.06 c, using onboard propellant. During this phase, the mass of the
vehicle gradually decreases as onboard propellant is consumed, so the accelera-
tion increases. Because of the low speed at onset of ram-scoop operation, there is
a negligible relativistic mass increase of the vehicle. Ram-scoop operation is
achieved when the forward speed of the vehicle collects a mass of H or D at
the same rate as that consumed and exhausted by the engine. (Interstellar H
would be used if a H H fusion cycle proves feasible; a fallback would be D,
as is used in contemporary D D fusion cycles, although there is less D than H
in interstellar space.) In this case, the mass flow rate (ṁ) of fusion fuel into the
ram-scoop equals the ṁ of the exhaust of the fusion engine. During this time,
the thrust and rest mass remains the same, so acceleration is nearly constant,
decreasing slightly due to the relativistic mass increase of the vehicle (e.g., by
a factor of 1.15 at 0.5 c).
60 R. H. FRISBEE

The unique feature of the interstellar fusion ramjet is that, once it reaches
ram-scoop operation, it can continue to accelerate to ultra-relativistic speeds
(e.g., .0.9 c) without any mass penalty (other than relativistic mass increase).
For speeds above the onset of ram-scoop operation, the ram-scoop magnetic
field geometry can be adjusted (e.g., the “scoop” collection area reduced) to
maintain the same mass flow (ṁ) of H (or D) into the fusion engine and thus
maintain constant thrust. For a flyby mission, ultra-relativistic speeds can be
especially advantageous for distant stellar targets because there is a long distance
(and time) over which to accelerate, and a long coast distance that is made more
tolerable by a high Vcruise. For a rendezvous mission, the same general argument
can also be made, especially because the ramscoop can act as a momentum
“brake” or “parachute” to slow the vehicle without the need to use the
fusion engine.
In the case of deceleration, we simply leave the ram-scoop magnetic field
turned on, such that it intercepts interstellar H (or D) and brings the H (or D)
to a stop with respect to the vehicle’s forward motion. The H (or D) can be
vented perpendicular to the vehicle’s forward motion so as to produce no side-
ways DV. As long as the vehicle’s forward velocity is greater than the fusion
engine’s Vexhaust (e.g., 0.03 c), the intercepted interstellar H/D will act like a
decelerating rocket engine with a Vexhaust equal to the vehicle’s forward speed.
During this period, acceleration will increase slightly because relativistic mass
is decreasing. Finally, as the forward speed approaches that of the engine’s
Vexhaust, the onboard propellant tanks can be filled from the scooped H (or D).
This will cause a significant decrease in deceleration due to the mass of H (or
D) propellant added to the vehicle. Finally, the vehicle is turned around and
the fusion rocket operated in a normal vehicle-decelerating fashion to bring the
vehicle to rest in the target solar system. This will produce an increase in decel-
eration as onboard propellant is consumed. Note that because the onboard propel-
lant tanks are sized to deliver a DV equal to ram-scoop onset (e.g., 0.06 c), there
will be residual onboard propellant to supply DV (e.g., 0.06 c-Vexhaust 0.03 c in
this example) for exploration of the target solar system.

2. Momentum Drag and Its Impact on Ram-Scoop Design

One major feasibility issue associated with the interstellar fusion ramjet is the
same one encountered in supersonic ramjet (scramjet) systems, where the engine
thrust must be greater than the ram inlet and engine drag. This suggests a
cylindrical-geometry magnetic confinement fusion (MCF) reactor, such as a
tandem-mirror reactor, where solenoid magnets confine the plasma radially.
Momentum drag would also be an issue for the magnetic scoop, because the H
(or D) atoms must be accelerated radially inward toward the centerline of the
scoop without being axially decelerated to produce drag. As mentioned above,
the ram-scoop is not a physical structure, but a magnetic field; however, it is
more complex than a single-coil magnetic nozzle “run backwards.” (A magnetic
nozzle run this way would “choke” and exclude H at the throat of the scoop.)
Another unique requirement of the scoop magnetic field is its immense size
(e.g., dimensions of thousands of kilometers), implying powerful magnetic fields.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 61

Fig. 13 Ram-scoop geometry for 1% axial momentum drag. (Angles are

exaggerated by factor of five for clarity.)

This suggests that the scoop magnetic field would need to have a very large
length-to-diameter (L/D) ratio to minimize the axial contact angle of the ions
with the field. An example of this is shown in Fig. 13, which shows a ram-
scoop with a cone half-angle (i.e., axial contact angle) of 18. After “reflection”
off of the ram-scoop’s magnetic “wall,” the H (or D) atom is traveling at an
angle of 28 (i.e., twice the initial contact angle). For an initial atom speed of
0.5 c (relative to the vehicle) corresponding to the maximum cruise velocity,
the atoms after reflection have an axial (x-axis) velocity of 0.4997 c, or
0.0003 c slower than their initial speed. This velocity change, 0.0003 c, rep-
resents 1% of the Isp (or Vexhaust) of the fusion rocket engine, so the overall
momentum drag is only 1% of the Isp of the engine. In effect, the engine
would now have an effective Isp of 0.0297 c because of the momentum drag of
the ram-scoop. Note, however that in order to produce only 1% drag from the
ram-scoop, the L/D of the ram-scoop is 114.6/1 [i.e., 1/tan(18) for L/R or
2/tan(18) for L/D], resulting in the need for a magnetic “wall” 687,500 km
long (almost 1.8 times the Earth moon distance!) for a 6000 km diameter
scoop. Needless to say, this represents only one of the many technical challenges
facing the interstellar fusion ramjet.
Finally, an alternative approach to ameliorating momentum drag from the ram
scoop that has been suggested by B. Cassenti (personal communication, 30
December 2007) would be to use magnetic fields to re-accelerate the scooped
atoms beck to their original speed (e.g., 0.5 c) prior to entering the fusion
reactor. Energy for this process could come from the fusion reactor and/or
from energy produced in the ram-scoop magnetic field due to the atom’s initial
deflection. Further study is needed to assess the efficiencies and energy flows
(i.e., energy sources and sinks) of the various processes to determine the feasi-
bility of this approach.

D. Effects of Acceleration and Cruise Velocity on Trip Time

In this section, we consider first the general problem of acceleration and cruise
(coast) velocity and their impact on mission trip time for the case of a fast (0.5 c)
interstellar rendezvous mission. Later we will illustrate how acceleration and
cruise velocity impact the transportation (propulsion) system mass for several
options, such as antimatter rockets and laser sails.
For a given travel distance, trip time will be a function of acceleration as well
as cruise velocity. For example, too low an acceleration can adversely impact trip
time, because the vehicle spends too much time in the acceleration/deceleration
62 R. H. FRISBEE

phase and not enough time at peak (cruise) velocity. This problem can be illus-
trated several ways. First, as shown in Fig. 14, we can consider the case of a ren-
dezvous mission with a given distance and total trip time and find Vcruise as a
function of acceleration. For this case, we assume a trip time in years numerically
equal to twice the distance in LY, so that, ignoring acceleration or deceleration,
the average cruise velocity would be 0.5 c. However, in practice, there is some
time spent (and distance traversed) during acceleration from Earth and decelera-
tion at the target solar system. (For these analyses, the acceleration and decelera-
tion phases are assumed equal.) Thus, a minimum acceleration is needed to reach
the target star where the vehicle accelerates to the midpoint in the trajectory, turns
around, and immediately begins to decelerate (i.e., there is no time spent coast-
ing); this limiting case requires a peak velocity approaching the speed of light.
As acceleration increases, some time is spent coasting, and the peak or cruise
velocity approaches 0.5 c as a limiting case for infinite acceleration.
An alternative mission scenario is shown in Fig. 15, where the maximum
cruise velocity (Vmax) is limited to some value (0.5 c in this case) so as to con-
strain the overall mission DV and thus the vehicle wet mass. In this case, the
mission total trip time is a function of acceleration (and deceleration) and dis-
tance traversed. Therefore, the trip time at the limit of infinite acceleration is
numerically twice the distance because the cruise velocity is limited to 0.5 c.
In order to minimize the trip time (by maximizing the time spent at peak
velocity), the vehicle needs to accelerate (and decelerate) at about 0.01 g
(1 g 9.8 m/s2 1.03 LY/yr2) as a minimum. At less than about 0.01 g, the
vehicle does not even reach the maximum allowed cruise velocity and the total
trip time increases dramatically; this effect is seen to become worse as the
total travel distance decreases. As shown earlier, higher acceleration is better,

Fig. 14 Coast (cruise) time and peak (cruise) velocity versus acceleration for an
interstellar rendezvous mission with equal acceleration and deceleration phases.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 63

Fig. 15 Trip time and maximum (cruise) velocity, Vmax, versus acceleration for an
interstellar rendezvous mission with equal acceleration and deceleration phases.

but higher acceleration will typically require more power and thus more system
mass. Interestingly, there is no significant benefit for acceleration .1 g.
Figure 15 also illustrates the somewhat counter-intuitive result that it can actu-
ally be harder to achieve a short trip time for nearby stars (i.e., less distance to
travel) than those farther away. This is because at low acceleration and short
total distance, the vehicle does not have enough distance to accelerate to a
high cruise velocity. For example, performing a 5 LY rendezvous mission at
an acceleration of 0.01 g takes about 45 years because the vehicle can only accel-
erate to less than one-half the nominal Vcruise (0.5 c) before it has to turn around
and begin deceleration. By contrast, at the same 0.01 g, we can travel to 20 LY, 4
times the distance, in about 90 years, twice the time, because the vehicle has more
distance to accelerate to roughly 0.45 c before beginning deceleration.

E. Interstellar Dust Impacts

One often-raised issue concerning interstellar missions is the effect of high-
speed (e.g., 0.5 c) interstellar dust grain impacts on a vehicle. Typically, the
assumption is made that a dedicated dust shield is placed at the front of a
vehicle to protect it from relativistic dust impacts [7,14]. The shield thickness
will be a function of the vehicle velocity, the distance traveled, and the dust par-
ticle number and size (and thus mass) distribution per unit volume in interstellar
space [23]. The calculation methodology we have developed previously [14]
assumes that the kinetic energy of the dust impact is turned into thermal
energy that evaporates (sublimes) the shield material (e.g., graphite). Thus, it
is necessary to take into account the energy per unit mass of dust at a given vel-
ocity (e.g., Vcruise 0.5 c) and then determine the total mass of dust hitting the
64 R. H. FRISBEE

shield. To do this, we calculate the kinetic energy per unit mass of a dust grain
(e.g., J/g), multiply this by the total mass of dust grains encountered along the
mission path per unit of shield frontal area (e.g., g/cm2), and divide by the
heat of vaporization of the shield material (e.g., J/g) to determine the mass of
shield material (and thus thickness) required per unit area of shield area. A
second effect, which might actually be larger, would be to include the effect of
additional shield material that could be “spalled” (mechanically broken) off of
the surface by the shock of impact.
For light sails, the payload and vehicle systems would need a dust shield, but it
is impractical to shield the total sail sheet due to the dust shield’s mass. Ironically,
it is actually better for the sail sheet to fly perpendicular (face-on) to the velocity
vector than fly parallel (edge-on). This is because an edge-on impact could “cut”
through more sheet material than a face-on impact. Fortunately, when flying face-
on, only a small amount of sail area, 2.3%, is lost due to dust impacts over even
a 40-LY path [2]. This rather modest loss of sail sheet area corresponds to the sum
of the cross-sectional areas of all the dust grains hitting the sail. A useful familiar
analogy is to imagine the action of a cookie cutter on a sheet of cookie dough,
where the area of the “hole” in the sheet (i.e., the cookie) is the same as that
of the cookie cutter.
As shown in Fig. 16, the upper end of the size range of the dust particles can be
very large compared to the sail sheet thickness. At relativistic speeds, the kinetic
energy of the dust grain simply vaporizes the sail sheet material in its direct path.
As for the surrounding sail sheet material, there are two effects that tend to limit
the size of the sail sheet hole to that of the incoming dust particle. First, at rela-
tivistic speeds, the interaction time between the sail sheet and the dust particle is
very short (e.g., 16 nm/0.1 c 5  10 16 s), such that the sail sheet material
surrounding the dust sail impact region is inertially frozen in place and unable
to respond mechanically by tearing or stretching under the force of the dust par-
ticle (and be vaporized). Second, the sail sheet is so thin that no significant

Fig. 16 Comparison of sizes of light sail sheet thickness (16 nm), dust particle
diameters (10 500 nm), and laser photon wavelength (1 mm).
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 65

amount of heat can transfer laterally along the sheet and cause vaporization of
additional sail material. In effect, the surrounding material does not have time
to tear/stretch or heat significantly, thus localizing the damage to only that
area punched out by the dust particle.
Note that this is an area where additional research is needed. For example, an
understanding of the physics of dust grain impacts with a shield is not known for
relativistic speeds. This research would allow us to quantify whether vaporization
or spalling is the dominant effect, and more generally estimate the shield thick-
ness needed. Also, in situ measurements of dust particle number, size, and
material during interstellar precursor missions would allow us to benchmark
our theoretical particle distribution models.
One final observation is that an interstellar rendezvous vehicle is only at high
speed in interstellar space. By contrast, flyby mission vehicles encounter thick
solar system dust clouds at high speed. Thus, we have the somewhat paradoxical
situation that the dust shield for an interstellar flyby could be thicker (and thus
heavier) than the one needed for a rendezvous mission.

IV. Rocket-Based Propulsion Options

In this section, we describe some of the general performance limitations of
existing propulsion technologies, and consider the potential performance that
might be seen for future technologies. Note that in this section, we are addressing
“rocket” propulsion systems in the sense that some onboard reaction mass (pro-
pellant) is expelled from the vehicle to produce acceleration. Thus, the rocket
vehicle has an onboard means of producing the momentum that accelerates it.
By contrast, the next section will deal with propulsion technologies that make
use of an external source of momentum (e.g., photon pressure from sunlight or
a laser). In this scenario, the vehicle possesses no intrinsic onboard momentum
source; rather, the external momentum source is manipulated in some fashion
to produce the desired vehicle acceleration.

A. Assumptions Used when Projecting Upper Performance Limits

Candidate propulsion systems are categorized by their energy source (e.g.,
fission, fusion, etc.) and are discussed herein. More detailed discussions can be
found in the technical literature (e.g., review articles [24] and individual
papers on each topic), from which the various performance parameters listed
below were derived.

B. Chemical Propulsion
The highest-Isp SOA chemical propulsion technology is represented by the
O2/H2 combination with a theoretical (ideal) Isp of 528 lbf-s/lbm. In practice,
the achievable Isp is approximately 450 lbf-s/lbm in such engines as the space
shuttle main engine (SSME) or the RL-10 engines in the Centaur upper stage.
If they could be produced in high concentrations, more exotic propellant combi-
nations, such as high energy density matter (HEDM) free-radical or “atomic”
hydrogen (H), electronic metastable atoms (e.g., triplet helium), or metallic
hydrogen have theoretical Isp’s as high as 3,000 lbf-s/lbm.
66 R. H. FRISBEE

C. Electric Propulsion
Advanced nuclear electric propulsion (NEP) systems have been studied for a
variety of interstellar precursor missions (to 1000 AU), although they would have
insufficient exhaust velocity (,200 km/s) for interstellar missions using extra-
polations of existing electric thruster technologies (e.g., ion, magnetoplasmady-
namic, Hall, etc.). Because of the 1/R 2 decrease in sunlight intensity as one
moves away from the Sun, a solar electric propulsion (SEP) vehicle, using
solar photovoltaic cells, is not an option. For a high-power NEP vehicle, we
could use a uranium-based reactor with either a static thermal-to-electrical
power conversion system (with no moving parts; e.g., solid-state thermoelectric
power conversion) or dynamic system (with moving parts; e.g., turbines, genera-
tors, etc.). An alternative, which is better-suited to small, low-power systems is
radioisotope electric propulsion (REP). In this concept, a spacecraft that
already has an onboard radioisotope power system [e.g., a plutonium-based
radioisotope thermoelectric generator (RTG)] due to the need to operate far
from the Sun would be combined with a small, low-power electric propulsion
system.
Finally, note that in principle, a high-energy relativistic particle accelerator
could be used as an electric thruster with an exhaust velocity approaching that
of the speed of light. However, the mass of such a “thruster” would be substantial;
one can imagine the low acceleration produced by taking the thrust from the par-
ticle accelerators at CERN or the Fermi National Accelerator Laboratory (Fermi-
Lab) and dividing this thrust by the mass of these facilities (as “thrusters”) along
with the mass of their power plants.

D. Fission Thermal and Pulsed Propulsion

Fission thermal and pulsed propulsion systems are characterized by high
accelerations (typically on the order of 1 g), but relatively modest Isp’s and
corresponding exhaust velocities. In fission thermal propulsion concepts, heat
from a fission reactor is transferred to a working fluid (typically H2) and the
high-temperature H2 gas expelled out a nozzle to produce thrust. SOA perform-
ance is represented by the nuclear engine for rocket vehicle applications
(NERVA) solid core reactor engine from the early 1970s with an Isp of 825
lbf-s/lbm. A modern version of this engine (NERVA derivative), using contem-
porary technologies in materials, pumps, etc. is projected to have an Isp of 900 lbf-
s/lbm, with advanced versions potentially reaching 1000 lbf-s/lbm. Advanced
gas-core reactor engines might have Isp’s as high as 7000 lbf-s/lbm, although
this would require dedicated external radiators (in addition to regenerative
engine cooling from the hydrogen expellant). These systems are at the lower
end of applicability to interstellar precursor missions because of their inability
to meet the high exhaust velocities and Isp’s needed for these missions (e.g.,
Vexhaust . 100 km/s or Isp . 10,000 lbf-s/lbm).
In the pulsed fission propulsion concept, typified by the Orion vehicle concept
(not to be confused with the NASA crew exploration vehicle), a fission “pulse
unit” is ejected behind the vehicle and exploded; debris and energy from the
explosion impacts a pusher plate that transfers the explosive shock to the
vehicle. Orion was originally intended for missions within the solar system,
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 67

and thus designed with a high thrust and modest Isp around 1800 to 2500 lbf-s/
lbm. To do this, the pusher plate was intentionally designed to ablate during
each pulse to increase the total mass of material ejected as propellant; this had
the effect of increasing thrust at the expense of potential Isp. An interstellar
version of Orion designed by Freeman Dyson [25] used high-energy fission or
fusion pulse units to produce a very high Isp version projected to be capable of
reaching a speed of 10,000 km/s (3.3% of the speed of light) in only 10 days
at an average acceleration of 1.2 g. At this speed (and with a negligible accelera-
tion time), Alpha Centauri could be reached in 130 years. The vehicle’s wet and
dry masses were projected to be 400,000 MT and 100,000 MT, respectively, and
to use 300,000 1-megaton pulse units. This vehicle also had an ablative pusher
plate; however, in this case ablation was used to cool the pusher plate and
protect it from the 1-megaton nuclear pulses. For an acceleration period of 10
days, it would be necessary to use 1 pulse unit every 2.9 seconds. Interestingly,
these values illustrate the extraordinary efficiency of direct use of fission or fusion
energy, as opposed to its use in a thermal rocket engine. For this case, with a mass
ratio (Mwet/Mdry) of 4 and a DV of 10,000 km/s, the equivalent Isp is an amazing
736,068 lbf-s/lbm, roughly 3/4 the Isp of a “pure” fission fragment or fusion
rocket (106 lbf-s/lbm), as discussed below.
Unfortunately, this would only be a flyby mission; to perform a rendezvous
mission we would need an additional “first” stage for acceleration and a
“second stage” (the original flyby stage) for deceleration and rendezvous. The
first stage (only) would have a wet mass of 1,600,000 MT [18]; the total two-
stage vehicle (with a 20,000 MT payload) would have a wet mass of 2,000,000
MT. Keeping the same pulse unit size and delivery rate, it would take 50 days
to consume 1,500,000 pulse units and accelerate at an average 0.2 g to a speed
of 10,000 km/s (and, 130 years later, 10 days at 1.2 g to decelerate).
However, it should be emphasized that the interstellar version of Orion was at
a much less mature level of study than the interplanetary versions. As Dyson
himself says, “The starship was like an existence theorem in math. It was to
prove that you could do it” [26]. Also, any version of Orion must contend with the
issue that the vehicle contains many thousands of nuclear fission (or fusion) pulse
units that could potentially be misused; for comparison, the interstellar flyby
Orion contains an order-of-magnitude more nuclear pulse units than the total
world inventory of nuclear warheads.

E. Fission Fragment Propulsion

Unlike a conventional fission thermal propulsion system (such as NERVA
and others), in which fission energy is used to heat a secondary working fluid
(e.g., H), the fission fragment propulsion concept uses fragments of the fission
process (e.g., Sr90 and Xe136 from U235 fission) directly as the rocket exhaust
gas. Because the high-energy fragments are not “cooled” by the addition of
any working fluid, this concept has the potential for achieving the ultimate
limit of fission energy with an Isp of about 106 lbf-s/lbm or an exhaust velocity
of 0.03 c. However, as discussed previously, this approach would require the cre-
ation of a propulsion-specific “propellant” production infrastructure that might be
difficult to justify. Also, as discussed below with the Daedalus fusion propulsion
68 R. H. FRISBEE

system, which also has an exhaust velocity of 0.03 c, a two-stage fission fragment
propulsion vehicle would be needed to reach a cruise velocity of 0.1 c for a flyby
mission (and four stages for a corresponding rendezvous mission).

1. Fission Fragment Rocket

In the fission fragment rocket, rods containing the fissionable material are
placed in a reactor to control the rate of fission. The rods are made thin to facili-
tate escape of the fission fragments from the rod. The rods are arranged like the
spokes in a rotating bicycle wheel; this allows the rods to be removed periodically
from the reactor to quench the reaction rate and cool before being re-inserted into
the reactor. Electromagnetic fields are used to capture, collimate, and direct the
charged fragment nuclei out the back of the rocket to produce thrust. As with any
propulsion system using electromagnetic fields to control charged particles, we
encounter a space-charge limit to the plasma density, with the result that,
although the Isp is high, the inherent thrust and thus vehicle acceleration will
be low.

2. Fission Fragment “Sail”

In this concept, shown in Fig. 17, a thin sheet (sail) would be doped with a
fissionable material near one side of the sail sheet. The fission fragments
would escape to space on the side of the sail with the fissionable material; on
the other, thicker side, the fragments would be absorbed by the sail sheet to
produce thrust. In this case, there would be no moderating reactor to control
the rate of fission or to allow periodic cooling. Therefore, it would be necessary
to dope the sail such that the rate of fission does not reach a level where waste
heat absorbed by the sail exceeds the ability of the sail to radiate the heat to space.
In principle, this system can have a higher acceleration than a corresponding
solar sail because the fission fragments produce more momentum than photons.
Also, it can be used as a regular solar sail near a star once its fission fuel is
depleted. Note, however, that this system throws away at least half of its potential

Fig. 17 Fission fragment sail concept.

 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 69

propellant (fission fragment) mass, depending on how well the fragments

are absorbed and stopped in the thicker side of the sail. Thus, although it has a
potential for high Isp, it will suffer a degradation in performance due to “loss”
of propellant.

F. Fusion Propulsion
There are two principal schemes for providing the confinement necessary to
sustain a fusion reaction: inertial confinement fusion (ICF), and magnetic con-
finement fusion (MCF). These confinement schemes result in two very different
propulsion system designs. There are literally dozens of different ICF, MCF, and
hybrid-ICF/MCF fusion reactor concepts [27]; regardless of the particular
reactor type, fusion propulsion concepts fall into two broad Isp categories. At
the lower end of Isp (and with correspondingly higher acceleration), fusion
rockets are well-suited to future large-scale exploration and exploitation of the
solar system (e.g., fusion propulsion could enable four-month round trips to
Mars). For these systems, the fusion energy is used to heat a secondary propellant
working fluid (such as H) to yield an Isp of about 104 to 105 lbf-s/lbm. For more
demanding interstellar missions, the fusion products would be directly exhausted
(as with the fission fragment rocket) with a maximum Isp of about 106 lbf-s/lbm or
an exhaust velocity of 0.03 c.

1. Conventional (All-Onboard Propellants) Fusion Propulsion

for Interstellar Missions
In our first example of fusion propulsion for interstellar missions, we will
discuss the “conventional” rocket approach in which all of the propellants to
be used during the mission are loaded into the vehicle at the start of the
mission; this approach is dominated by the strictures of the Rocket Equation
because DV (.0.1 c) is typically much larger than fusion propulsion’s Isp or
Vexhaust (0.03 c). A second approach, discussed below, seeks to “cheat” the
Rocket Equation by continuously refueling the vehicle from interstellar
hydrogen.
For a “conventional” fusion rocket to perform a 0.1 c cruise velocity interstel-
lar flyby mission, a two-stage vehicle would be needed, such as the Daedalus
two-stage ICF fusion rocket designed by the British Interplanetary Society [7].
Table 2 summarizes the performance of this vehicle in its original two-stage,
0.1 c flyby configuration, where stage 1 had a higher DV than stage 2. Also
shown is a hypothetical two-stage rendezvous version with the same total
vehicle mass, with equal DV for both stages.
Finally, we could employ a four-stage vehicle for a rendezvous mission with a
cruise velocity of 0.1 c. We would need a new first and second stage to accelerate
to 0.1 c, and the original two-stage Daedalus for the deceleration and rendezvous.
However, this would result in a very heavy vehicle. For example, if we assume
that the first two stages have the same propulsion system TF (Mdry-prop/Mp) as
the Daedalus’ first stage (3.674%), and the same DV’s and accelerations, then
the total vehicle would weigh over 1.7 million MT and the total trip time to
6 LY would be 61.96 years.
 70

Table 2 Daedalus two-stage fusion rocket performance to 6.0 LY (Barnard’s star)

6.0 LY Flyby 6.0 LY Rendezvous

Mission type (Vcruise ¼ 0.102 c) (Vcruise ¼ 0.045 c)

Mission phase Acceleration Acceleration Total mission Acceleration Deceleration Total mission
Total mission trip time 1.895 1.874 60.311 1.648 4.212 136.276
R. H. FRISBEE

(years)
Average acceleration or 0.0329 0.0197 0.0265 0.0104
deceleration (g)
Total mission distance 0.061 0.156 6.000 0.037 0.095 6.000
(LY)
Vehicle stage Stage 1 Stage 2 Total vehicle Stage 1 Stage 2 Total vehicle
Initial total wet mass, 53,500 5810 53,500 53,500 12,025 53,500
Mwet (MT)
Total propellant mass, 46,000 4000 50,000 40,006 8992 48,998
Mp (MT)
Final total dry mass, 7500 1810 3500 13,494 3033 4503
Mdry (MT)
Vehicle mass ratio, 7.133 3.210 3.965 3.965
Mwet/Mdry
Propellant system dry 1690 980 2670 1470 2203 3673
mass, Mdry-prop (MT)
Propellant system total 3.674% 24.500% 3.674% 24.500%
TF, Mdry-prop/Mp
Payload mass, MPL (MT) 5,810 830 830 12,025 830 830
Velocity change, DV (c) 0.06418 0.03809 0.10228 0.04499 0.04499 0.04499
Propellant exhaust 0.03267 0.03267 0.03267 0.03267
velocity, Vexhaust (c)
Rocket engine thrust, F 7.54E þ 06 6.63E þ 05 7.54E þ 06 6.63E þ 05
(N)
Propellant mass flow 0.7694 0.0677 0.7694 0.0677
rate (kg/s)
“Jet” exhaust power, Pjet 36.95 3.25 36.95 3.25
(TWjet)

Rocket engine run time 1.895 1.874 3.768 1.648 4.212 5.859
(years)
LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY
71
72 R. H. FRISBEE

2. Bussard Interstellar Ramjet: Extraterrestrial Resource Utilization

(ETRU) Fusion Propulsion for Interstellar Missions
Thus far in our discussions of potential interstellar propulsion options, we
have seen how strongly Isp (and the corresponding Vexhaust) impacts performance
for flyby and, especially, rendezvous missions. Even direct use of fission and
fusion is limited to a Vexhaust of 0.03 c, constraining these concepts to rather
modest peak velocities and corresponding long trip times. This level of perform-
ance is ultimately driven by the Rocket Equation, and its exponential increase in
wet mass and propellant mass as (DV/Vexhaust) increases. One way to “cheat” the
Rocket Equation is to not carry onboard all the propellant that would be needed
for the mission. Instead, we “live off the land,” and collect materials along the
way that could be used for propellant during the mission. This approach is
referred to as extraterrestrial resource utilization (ETRU); for example, chemical
propellants needed for a robotic Mars sample return rocket could be produced
from CO2 in the Martian atmosphere and/or water-ice in the soil. We take this
same approach in the ultimate ETRU concept, the Bussard interstellar ramjet
[11], in which interstellar H would be scooped to provide propellant mass for a
fusion propulsion system. Interstellar H would be ionized and then collected
by an electromagnetic field. Onset of ramjet operation is at a velocity of about
0.06 c. Beyond this ram speed, the scooped H is “free” (as far as the Rocket
Equation is concerned), so the vehicle could continue to accelerate even up to
ultrarelativistic speeds (0.5 c). Thus, the only onboard propellant required
would be that needed for the initial acceleration to, e.g., 0.06 c (or corresponding
deceleration from 0.06 c to rest for a rendezvous), resulting in a significantly
lighter vehicle (i.e., Mwet/Mdry exp[0.06/0.03] 7.4).
In operation, the Bussard interstellar ramjet would first accelerate up to ram
onset speed (e.g., 0.06 c) using onboard propellant. At speeds above ram onset,
the fusion ramjet is used to accelerate to the final cruise velocity, which is
constrained by relativity and not the Rocket Equation. To begin a rendezvous
mission’s deceleration phase, the electromagnetic scoop is turned on but set
such that the scooped interstellar H comes to rest (relative to the vehicle)
inside the scoop; this is equivalent to allowing the ram scoop to “choke” the
gas flow. Momentum drag from the scoop (due to interstellar H coming to rest
in the scoop) is actually more efficient than the fusion rocket at speeds above
the Vexhaust of the rocket (0.03 c). As the speed approaches Vexhaust, the scooped
H is used to refill the vehicle’s propellant tanks for the final fusion rocket decel-
eration to rest. Because the tanks are sized for a DV of 0.06 c (but we only need to
decelerate 0.03 c), there would be some propellant remaining in the tanks that can
then be used for normal fusion propulsion within the target solar system.
However, although the Bussard interstellar ramjet is very attractive for inter-
stellar missions because of its unlimited range and potential for ultrarelativistic
speeds, there are several, very major feasibility issues associated with its oper-
ation [3]. For example, there do not appear to be any feasible technological sol-
utions to performing pure H H fusion (short of using the cores of stars).
However, D D fusion appears quite feasible technologically, so it may simply
be necessary to collect interstellar D and discard the H (although there is only
about 1 D per 105 H in interstellar space). A second significant issue is the
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 73

design of the electromagnetic “scoop,” both in terms of its electromagnetic

shaping as well as the structural design required to support the magnets that
produce the magnetic scoop fields. Finally, a potentially major issue is the
need to design a ram-scoop and fusion engine system that minimizes momentum
drag on the collected H. This is because, relative to the vehicle, the collected H
(or D) represents a decelerating thrust from the momentum of the H (or D)
approaching the vehicle (e.g., at 0.5 c) versus accelerating thrust from the
fusion engine at an exhaust velocity of only 0.03 c.

G. Matter – Antimatter Annihilation Propulsion

Matter antimatter annihilation offers the highest possible physical energy
density of any known reaction substance. The ideal energy density (E/M c 2)
of 9  1016 J/kg is orders of magnitude greater than chemical (1  107 J/kg),
fission (8  1013 J/kg), or even fusion (3  1014 J/kg) reactions. Additionally,
the matter antimatter annihilation reaction proceeds spontaneously, and there-
fore does not require massive or complicated reactor systems. These properties
(high energy density and spontaneous annihilation) make antimatter very
attractive for propulsively ambitious space missions like interstellar travel.
This section describes those antimatter propulsion concepts in which matter
antimatter annihilation provides all of the propulsive energy. A related
concept, in which a small amount of antimatter triggers a much larger micro-
fission/fusion reaction, is discussed below.

1. Overview of Matter Antimatter Annihilation Propulsion

Not surprisingly, antimatter production, storage, and utilization represent major
challenges. Numerous fundamental feasibility issues remain to be addressed, such
as scaling up antimatter production rates and efficiencies, storage in a high-density
form suitable for propulsion applications, and design and implementation of a
complete propulsion system containing all the ancillary systems required to
contain the antimatter on the vehicle and ultimately use it in a thruster [14]. Never-
theless, research aimed at addressing these issues is ongoing at a modest level.
As a specific thruster design, we have the “beam-core” antimatter rocket
concept [14], in which equal amounts of matter and antimatter (in the form of
protons and antiprotons) are combined and annihilate each other to produce
high-speed annihilation products (neutral and charged pions). A magnetic
nozzle is used to direct the charged pions out of the nozzle to produce an Isp of
about 107 lbf-s/lbm [20] corresponding to an exhaust velocity of 0.3 c.
Note that in this concept, we have emphasized the use of proton antiproton
annihilation because the charged pion annihilation products could be deflected
and somewhat collimated to produce a directed stream of “exhaust” (particles
with mass and thus momentum) from the engine. In principle, we could have
used electrons and anti-electrons (positrons) as our propellants; however, their
annihilation products are high-energy photons (gamma rays). This results in
two problems with this approach. First, there is no practical way to focus or
deflect gamma rays without the use of massive amounts of shielding material,
resulting in an extremely low acceleration. Second, although photons have
74 R. H. FRISBEE

momentum, their momentum is intrinsically low compared to “normal” particles

that contain actual mass. As a simple example of the sort of acceleration we might
expect from a “pure” photon rocket, we can consider an ordinary household flash-
light with two “D” cells. It weighs on the order of 0.5 kg (1 lb); its “Krypton”
light bulb is rated at 2.4 V (DC) and 0.7 A, for a power of 1.68 W. We can
use the equations developed for light sails to estimate an upper limit to the flash-
light’s acceleration by assuming that all of the beam power goes into producing
thrust (e.g., perfect reflectivity). Thus, we have Pflashlight Mflashlight a c/
(2 h þ a) which, upon rearrangement and simplification becomes:

Acceleration ¼ a ¼ Pflashlight (2h þ a)=ðMflashlight cÞ

¼ 2ð1:68WÞ=ð0:5 kgÞ=c ¼ 2:2  10 8 m=s2

This corresponds to 2.3 nano-g of acceleration, which means that a flashlight-

based photon rocket would require “only” 42 million years to accelerate to
0.1 c. In practice, any photon rocket will have low acceleration due to the
factor of c (speed of light) in the denominator of the acceleration equation.
Given the requirement for propellants in the form of protons and antiprotons,
we find that that actual propellant storage form needs to be that of a condensed
phase of molecules (and anti-molecules). This is driven primarily by the
density of the propellant and its impact on the propellant tankage mass. For
example, we routinely trap, move, and store antiprotons using electromagnetic
fields. However, as a charged particle, an antiproton is subject to the space-
charge limits imposed on any ion plasma; with current magnet technology, this
corresponds to 1010 to 1012 ions/cm3 or 10 14 to 10 12 g/cm3 for (anti)protons.
Thus, it would be wildly impractical to store the tons of antiprotons needed for an
interstellar mission as an ion gas; the tanks would be impossibly large and heavy.
This then drives us to use ordinary (normal-matter) liquid molecular hydrogen
(LH2) and solid molecular anti-hydrogen (anti-SH2). The normal-matter LH2
can be stored as ordinary LH2 at around 20K. However, the anti-SH2 must be
stored as magnetically-suspended/levitated pellets at 1K. In this case, the mag-
netic suspension prevents the antimatter from contacting the propellant tank
walls, feed lines, etc. and annihilating. Furthermore, the pellets must be stored
at 1K to prevent their sublimation to form gaseous anti-H2, which is not contained
by the magnetic field and thus would expand (like any gas) until it reached the
tank walls and annihilated [28].
Note that condensed-phase molecular storage implies conversion of antipro-
tons to anti-atoms (i.e., anti-H, consisting of a hydrogen-like anti-atom contain-
ing an antiproton and positron), and subsequent conversion of anti-atoms into
antimolecules. This has been seen as a significant challenge due to the need to
remove excess energy from the collision partners; for example, an antiproton
positron collision needs to have 13.6 eV (the ionization energy of a hydrogen
atom) removed to form an anti-atom, and an anti-H/anti-H collision 4.5 eV
(the heat of formation of atomic-H from molecular-H2) to form an antimolecule.
However, recent advanced in electromagnetic “cooling” of atoms [29] suggests
that this may be more feasible than the laser cooling method previously proposed
for this process [28].
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 75

Finally, the most serious feasibility issue associated with antimatter propul-
sion is the sheer difficulty of producing the antimatter. Antiprotons do not exist
in nature and currently are produced only by energetic particle collisions con-
ducted at large accelerator facilities [e.g., FermiLab and Brookhaven National
Laboratory (BNL) in the United States; CERN in Switzerland; or IHEP in
Russia]. This process typically involves accelerating protons to relativistic vel-
ocities (very near the speed of light) and slamming them into a metal (e.g., tung-
sten) target. The high-energy protons are slowed or stopped by collisions with
nuclei of the target; the relativistic kinetic energy of the rapidly-moving proton
is converted into matter in the form of various subatomic particles, some of
which are antiprotons. The antiprotons are then electromagnetically separated
from the other particles and stored in electromagnetic rings. Note that antiprotons
annihilate spontaneously when brought into contact with normal matter; thus,
they must be contained by electromagnetic fields in high vacuums. This
greatly complicates the collection, storage, and handling of antimatter. Currently
the highest antiproton production/capture/accumulation level (not optimized for
rate or efficiency) is of the order of 10 nanograms per year, although planned
upgrades to CERN may increase these production rates by a factor of 10 to
100. Additionally, only a much lower level of antiprotons have actually been col-
lected, cooled, and stored after production. Finally, current production/capture/
accumulation technology has an energy efficiency of only about 1 part in 109 (i.e.,
109 units of energy are consumed to produce an amount of antimatter that will
release 1 unit of energy upon annihilation) [28].
To date, only infinitesimal quantities of antiprotons and anti-atoms have been
produced, as shown in Fig. 18. However, it is worth considering the progress
of technology in producing “exotic” forms of matter. For example, it has taken
little more than a century to go from Dewar’s few drops of LH2 to the 100
tons of LH2 used in every Space Shuttle launch. One can imagine Dewar’s
reaction if he were told in 1898 that less than a century later, we would be
using hundreds of tons per year for space exploration missions, given the techno-
logical challenges he faced in 1898.

2. Matter Antimatter Annihilation Propulsion Vehicle Sizing

In previous studies,†† we have developed detailed mass scaling equations for
determining the dry mass of a “beam core” antimatter rocket stage [14].‡‡ As
discussed, the Relativistic Rocket Equation with a 0.4 limits each stage to a

††
The antimatter rocket mission analyses presented in Frisbee and Leifer [3] were based on 0th
order vehicle mass estimates; a much more detailed set of vehicle system mass and mission perform
ance estimates is given in Frisbee [14].
‡‡
For the vehicle mass sizing calculations used here, we assumed the case for a 10 fold
improvement in superconductor magnets and radiators. Also, because we evaluated several different
multi stage vehicle configurations in which the “top” or final two stages (e.g., stages 1 and 2 in a two
stage vehicle, or stages 3 and 4 in a four stage vehicle) could be accelerating or decelerating, we
assumed that these two stages would have a full dust shield. By contrast, we retained the assumption
that “lower” stages (e.g., stages 1 and 2 in a four stage vehicle) could have a central hole in their dust
shields corresponding to the diameter of the upper stage (e.g., stage 3 sitting on “top” of stage 2, or
stage 2 sitting on “top” of stage 1).
76 R. H. FRISBEE

Fig. 18 Comparison of growth in annual production rates of normal-matter LH2

and antimatter.

DV of around 0.25 c. Because the vehicle dry mass is a function of its acceleration
and DV, we want to determine an “optimum” that produces a reasonable compro-
mise between total vehicle wet mass (and especially antimatter mass), where
mass is a function of the acceleration and mass ratio, Mwet/Mdry, and trip time,
where trip time is a function of both acceleration and maximum or cruise
velocity, Vcruise.
This tradeoff is illustrated in Fig. 19 for the case of our Vision Mission, a
40-LY interstellar rendezvous. Generally, as acceleration increases, we first see
a dramatic reduction in trip time; however, for accelerations above about
0.04 g, we see only a modest decrease in trip time and a modest increase in
mass. For this system, the curves for 0.05 and 0.06 g accelerations essentially
overlap in the vicinity of the “knee” of the curves (roughly, 104-year flight
time), with 0.05 g just barely more optimum. At a given acceleration, a much
larger effect on mass is seen in increasing the velocity due to the effect of the
Rocket Equation’s mass ratio. Thus, the best approach seems to use a moderately
high acceleration and limit Vcruise to a value that keeps Mwet/Mdry tolerable.
For the purposes of the mission analyses presented below, we have somewhat
arbitrarily selected an acceleration of 0.05 g and a Vcruise that yields a value
of Mwet/Mdry 5.0. For a single-stage vehicle accelerating to some final
Vcruise, a value of Mwet/Mdry 5.0 occurs for Vcruise DV 0.2186 c; for a
two-stage vehicle accelerating to some final Vcruise with each stage having
Mwet/Mdry 5.0, Vcruise 0.4173 c, with stage 1 accelerating from 0 to
0.2186 c (DV 0.2186 c) and stage 2 accelerating from 0.2186 c to 0.4173 c
(DV 0.1987 c).
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 77

Fig. 19 40-LY interstellar rendezvous antimatter rocket total wet mass (top) and
antimatter mass (bottom) as a function of cruise velocity (Vcruise).

H. Catalyzed Nuclear Fission/Fusion/Antimatter Combinations

Because of the difficulty of “igniting” fusion reactions, several schemes have
been proposed to simplify or reduce the energy requirements of the “drivers” of
the fusion ignition process. For example, fission reactions can be “catalyzed”
by addition of small amounts of antimatter (antiprotons). The energy released
from the fission reaction can then be used to ignite a fusion reaction.
Because the required amounts of antiprotons are relatively small, it becomes
feasible to consider such systems in terms of modest upgrades to existing
production facilities (e.g., CERN, FermiLab) and use of existing storage
technologies (e.g., as space-charge limited plasmas). However, it should be
emphasized that these concepts will have a peak Isp of only 106 lbf-s/lbm,
because the total energy release is still dependent on the much larger fission
and/or fusion reactions. Thus, the overall vehicle may be lighter (because
Mdry-prop is lighter), but the propulsion system’s exhaust velocity will ultimately
be limited to that of the primary energy production mechanism (i.e., fission or
fusion Vexhaust 0.03 c).
78 R. H. FRISBEE

1. Muon Catalyzed Fusion

In muon catalyzed fusion, negative muons (m ) produced by antiproton
annihilation are used to “catalyze” the fusion reaction [30]. The muon is often
considered to be a “heavy electron,” with a mass 207 times that of the electron.
In this concept, the muon replaces the electron in an atom of the fusion fuel (D, T,
He3). The resulting “atom” has a classical Bohr radius 207 times smaller than its
electron counterpart; thus, the nuclei are able to approach each other more
closely. This in turn enhances the probability of overlap between the wave func-
tions of the nuclei of colliding atoms, and thus increases the probability of fusion.
The fusion energy “ionizes” the atom and ejects the muon, which goes on to
attach itself to another nucleus, and the process repeats for the lifetime of the
muon (2 ms).

2. Antimatter Catalyzed Nuclear Fission/Fusion Combinations

An alternative approach to “conventional” fusion propulsion systems is the
Inertial-Confinement Antiproton-catalyzed micro-fission/fusion Nuclear (ICAN)
propulsion concept developed by Pennsylvania State University [31]. In this
approach to ICF propulsion, a pellet containing uranium (U) fission fuel and deu-
terium tritium (D T) fusion fuel is compressed by lasers, ion beams, etc. At the
time of peak compression, the target is bombarded with a small number (108
to 1011) of antiprotons to catalyze the uranium fission process. (For comparison,
ordinary U fission produces 2 to 3 neutrons per fission; by contrast, antiproton-
induced U fission produces 16 neutrons per fission.) The fission energy
release then triggers a high-efficiency fusion burn to heat the propellant, resulting
in an expanding plasma used to produce thrust. Significantly, unlike “pure” anti-
matter propulsion concepts that require large amounts of antimatter (because all
of the propulsive energy is supplied by matter antimatter annihilation), this
concept uses antimatter in amounts that we can produce today with existing tech-
nology and facilities. This technology could enable 100- to 130-day round trip
(with 30-day stop-over) piloted Mars missions, 1.5-year round trip (with 30-
day stop-over) piloted Jupiter missions, and 3-year one-way robotic Pluto
orbiter mission (all with 100 MT payloads).
A recent variation on the ICAN concept is antimatter initiated microfusion
(AIMStar) [32], which uses an electromagnetic trap (rather than laser or particle
beam implosion) to confine a cloud of antiprotons during the antimatter-induced
microfission step. This concept may enable the construction of very small
systems (at least as compared to a conventional ICF fusion rocket) because a
large ICF-type pellet implosion system is not required.

3. Antimatter-Catalyzed Fission Fragment Sail

A variation on the fission fragment sail concept (discussed above) is the use of
antiprotons to induce fission in fissionable material arranged as a sheet, somewhat
similar to a solar sail [33]. As with any fission fragment propulsion concept, the
Isp and Vexhaust are limited to around 0.03 c.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 79

I. Beamed-Energy Propulsion
In beamed-energy propulsion concepts, a remote energy source beams electro-
magnetic radiation (e.g., sunlight, laser, or microwave energy) to the space
vehicle where the energy is used to heat or electromagnetically accelerate on-
board propellant to produce thrust. Note that beamed-energy is distinct from
the beamed-momentum concepts listed below, because the beamed-energy
source only provides energy; the spacecraft still must have onboard propellant
and follow the Rocket Equation. By contrast, a beamed-momentum source
(which can again be sunlight, laser, or microwave, as well as particle beam)
provides actual momentum to the vehicle; in effect, the photon or particle
momentum “pushes” the vehicle and no on-board propellant is required.
The primary advantage of beamed-energy systems is the reduction in vehicle
mass afforded by taking the energy source off the vehicle. However, the ultimate
achievable specific impulse or exhaust velocity is limited by the type of thruster
employed on the vehicle. For example, with a thermal engine [e.g., solar thermal
propulsion (STP)], where the incoming beamed energy is used to thermally heat a
propellant (typically H), materials temperature limits constrain Isp to values
similar to NERVA-type engines (1000 lbf-s/lbm). With an electric thruster
[e.g., solar electric propulsion (SEP)], where the incoming beamed energy is
first converted into electricity for the thruster’s use, the Isp will be representative
of the particular type of electric thruster used. However, an SEP system would
still be limited by the 1/R 2 drop-off in sunlight intensity. To overcome this,
laser electric propulsion (LEP) uses laser light to provide constant power even
far from the Sun. Here, high-performance electric thruster exhaust velocities
(,200 km/s) would be adequate for interstellar precursor missions but not for
interstellar missions.

V. Nonrocket (Beamed-Momentum) Propulsion Options

In beamed-momentum propulsion, the momentum carried by a stream of
particles (e.g., photons or charged particles) is used to push the vehicle; in
effect, the stream of particles becomes the “propellant” that supplies the
momentum to move the spacecraft. This is in contrast to a beamed-energy
system, where the beamed energy (sunlight, laser/microwave beam) provides
thermal energy (or indirectly electricity) that is used to energize onboard propel-
lant. Thus, a beamed-momentum propulsion system represents an example of a
“propellantless” propulsion system, with both the energy and propellant system
taken off the vehicle. As such, it also represents a way to fundamentally
“cheat” the Rocket Equation because there is no onboard propellant; in effect,
Mwet/Mdry 1.
Two general types of beamed-momentum systems are considered. The first
type uses momentum exchange between photons (solar/laser/microwave sails)
and a reflective sheet or “sail.” The second type uses momentum exchange
between charged particles (from the solar wind or a dedicated particle accelera-
tor) and an electromagnetic field (electromagnetic sails).
80 R. H. FRISBEE

A. Assumptions Used when Projecting Upper Performance Limits

Candidate propulsion systems are categorized by their momentum source
(e.g., sunlight, laser, etc.) and are discussed below. As with the rocket-based pro-
pulsion systems, more detailed discussions can be found in the technical
literature.

B. Solar Sails
A solar sail is a propulsion concept that makes use of a flat surface of very thin
reflective material supported by a lightweight, deployable structure [34]. Because
a solar sail uses no propellant, it has an effectively infinite specific impulse;
however, the thrust-to-weight (T/W) ratio is very low, 10 4 to 10 5 for the
9 N/km2 (5.2 lbf/mile2) solar pressure at Earth’s distance from the sun, resulting
in the potential for long trip times in and out of planetary gravity wells.
Solar sails have been considered for interstellar precursor missions; Fig. 20
shows trajectories that make use of intense sunlight pressure near the sun (at
0.25 AU in this example) to “blast” the sail off into a high solar system escape
trajectory. Note that the sails shown here represent fairly aggressive technologies
in terms of their total areal density (mass, including payload, per unit area of sail
sheet, typically in units of g/m2), as compared to near-term solar sails with areal
densities . 10 g/m2. Overall, solar sail performance is similar to nuclear electric
propulsion in terms of achievable solar system escape velocities (Vinfinity) and
thus trip time.

C. Light Sails
One important limitation in solar sails is the 1/R 2 drop in sunlight intensity as
one moves out of the solar system. Nevertheless, as seen above, solar sails could
be used for deep space or interstellar precursor missions by first spiraling in close

Fig. 20 Representative solar sail trajectories for interstellar precursor missions.

 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 81

to the sun (e.g., to 0.25 AU) and using the increased sunlight pressure to drive
them out of the solar system. It would also be possible to perform interstellar mis-
sions with a laser-driven “light” sail. This concept [13] is uniquely suited to inter-
stellar missions since it is one of the few ways that sufficient energy (per unit
mass) can be imparted to a vehicle to achieve the high velocities (.0.1 c)
required for interstellar missions. This is possible because the spacecraft
“engine” (lasers) is left back in Earth’s solar system; a somewhat arbitrarily
large amount of energy (number of photons per unit of sail mass) could be
imparted to the vehicle’s propellant (photons) to accelerate the vehicle. In fact,
because of the light sail’s thermally limited acceleration, input power is ulti-
mately limited by the imperfect reflectivity of the receiver optics; solar or laser
light absorbed by the receiver material must be radiated to space as “waste
heat” so the maximum power that can be received is a function of the sail’s
material reflectivity, emissivity, and maximum temperature limits.
Note, however, that for interstellar distances, very large optics and laser power
levels would be required. For example, a laser operating at 1 mm wavelength
requires a transmitter lens with a diameter of 1000 km to illuminate a 1000-km
diameter receiver (sail) at 43 LY. Similarly, a very high power level (and
ultra-light sail) would be required for reasonable acceleration (typically 0.04 g
for flybys to 0.2 g for rendezvous) of the vehicle. For example, the laser power
required for a robotic flyby mission to 4.3 LY with a maximum cruise velocity
of 0.4 c would be 14 terawatts (TW), which is comparable to the average
power produced by all of human civilization (15 TW in 2005). However, any
interstellar mission, regardless of the propulsion system, will require high
power levels to achieve the high speeds required. Even today we achieve nontri-
vial propulsion power levels for ambitious space missions; for example, the
Saturn V rocket generated a power on liftoff corresponding to about 0.8% of
human civilization’s total power output in 1969.
The microwave sail (Starwisp) [12] concept is the microwave analog to the
laser light sail. This approach has the advantage that the vehicle can be made
ultra-lightweight for robotic interstellar mission flybys, thereby reducing both
the transmitter power requirements and the size of the transmitter optics (because
the microwave sail can be accelerated at high “gs” to its final coast velocity while
still relatively near the Earth). In order to achieve this low mass, the “sail” con-
sists of wire mesh with holes in the mesh less than one-half the wavelength of the
microwaves. Under these conditions, the sail acts like a solid sheet with respect to
the incoming microwave photons. (A related concept, the “perforated” solar light
sail, has also been proposed for visible-light sails.) However, microwave sails are
only feasible for interstellar flybys, and not rendezvous missions, because the
long wavelengths of microwaves (as compared to visible or near-visible light)
makes it impractical to build optics large enough to focus a microwave beam
onto a sail at interstellar distances.
By contrast, laser light sails have a short enough wavelength to make it poss-
ible to focus the beam and enable the use of two-stage light sails for rendezvous
missions, as shown in Fig. 21, and even a three-stage light sail [13] for a round
trip mission. Ultimately, beamed-momentum light sails represent a major devel-
opment challenge, both because of the extraordinarily demanding technologies,
and because of the extraordinarily large scale of the systems. Nevertheless,
82 R. H. FRISBEE

Fig. 21 Beamed-momentum laser light sail missions (after Ref. 13).

they do represent one of the few ways to perform interstellar missions with
reasonable trip times.

D. Electromagnetic Sails
In electromagnetic (EM) sails, charged particles (typically protons) from the
solar wind or a charged-particle accelerator are reflected by a magnetic field, ana-
logous to the reflection of solar or laser photons off of a solar/laser sail’s reflec-
tive sheet. Thus, EM sails are the charged-particle analogs of solar light sails.

1. Electromagnetic Sails
In principle, a solar-wind EM sail could be built using a physical sheet of
material, but the momentum per unit area carried by the solar wind is so much
less than that from photons that it requires an impossibly lightweight sheet;
instead, a (massless) magnetic field, 10s to 100s of kilometers in diameter, sub-
stitutes for the solar sail’s sheet. Interestingly, EM sails provide many of the same
potential benefits and drawbacks of solar sails; for example, sunlight intensity
and solar wind density both drop off as the square of the distance (1/R 2) from
the Sun. (The solar wind maintains a roughly constant velocity of around 300
to 800 km/s throughout the solar system, but the momentum force decreases
due to the expansion of the solar wind, and thus increasing dilution of individual
particles, at increasing distance from the sun.)
The first proposed EM sail was the magnetic sail, or MagSail, concept [35].
The MagSail consists of a cable of superconducting material, millimeters in
diameter, which forms a hoop that is 10s to 100s of kilometers in diameter.
The current loop creates a magnetic dipole that diverts the background flow of
solar wind. This deflection produces a drag force on the MagSail radially
outward from the sun. In addition, proper orientation of the dipole may
produce a lift force that could provide thrust perpendicular to the radial
drag force.
A newer concept, the mini-magnetospheric plasma propulsion (M2P2) sail
[36], uses an artificially generated mini-magnetosphere that is supported by
magnets on the spacecraft and inflated by the injection of low-energy plasma
into the magnets. (Thus, M2P2 is not, strictly speaking, a true “propellantless”
propulsion system; however, the amount of propellant needed to produce the
plasma is small, resulting in an effective Isp of 35,000 lbf-s/lbm.) This plasma
injection allows the deployment of the magnetic field in space over large dis-
tances, comparable to those of the magsail.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 83

Fig. 22 Relativistic particle beam/magsail concept.

2. Particle Beam Driven Electromagnetic Sails and MagOrion

Note that the similarity of EM sails to solar sails extends to their general
mission performance; for example, both EM sails and solar sails could be used
for the more modest interstellar precursor missions. However, like light sails,
EM sails would require a dedicated beamed-momentum source (e.g., charged
particle beam) for interstellar missions. Thus, we could use a linear accelerator
to fire out high-speed particles (e.g., a relativistic particle beam) or pellets that
push against the vehicle [37] as illustrated in Fig. 22. The particles can either
hit a physical “pusher plate,” or charged particles can push against (and be
reflected by) the electromagnetic fields of an EM sail in much the same way
that photons are employed in a laser/microwave sail. In principle, a two-stage
EM sail could be used for an interstellar rendezvous mission by analogy with
the two-stage light sail.
However, one potentially serious problem with the use of a charged ion or
neutral particle beam is that the beam can be deflected by stray magnetic fields
within a solar system and in interstellar space. Because of the speed of light
time delay, it would not be practical to use a feedback control system to
correct for the beam deflections. Also, neutral particles would have to be
ionized before hitting an EM sail’s magnetic fields. Alternatively, neutrals
could impact a physical “pusher plate.” One advantage of light sails is that
their laser/microwave beams are unaffected by magnetic fields, although the
beam could be scattered and defocused by interstellar dust clouds.
To avoid the problem of particle beam deflection from a remote source (e.g.,
from the solar system), we could carry a source of charged particles with the
vehicle; for example, we could use the charged particles produced by a nuclear
explosion. This would be the EM sail analog of the Orion nuclear pulse
concept, with an electromagnetic (rather than physical) “pusher plate,” hence
the name Magnetic Orion or MagOrion [38]. Interestingly, in the original
Orion concept, some of the material of the pusher plate would evaporate
(ablate) with each pulse. This serves to both to keep the pusher plate cool, and
to add additional “propellant” mass. The result is an increase in thrust, although
at the expense of some specific impulse (Isp). By contrast, there is no mass in the
MagOrion’s magnetic pusher plate, so the effective Isp (15,000 to 45,000 lbf-s/
lbm) makes this concept well-suited for interstellar precursor missions.

E. Electromagnetic Catapults
Electromagnetic launchers, such as advanced versions of the rail gun [39] and
mass driver (coil gun) [40], could be used to launch small or microspacecraft.
84 R. H. FRISBEE

However, their demonstrated “muzzle” velocity (,12 km/s) may limit them to
solar system exploration missions [8].

VI. Illustrative Missions

Thus far in this discussion, we have described a number of propulsion options
that have the potential of performing interstellar precursor missions out to thou-
sands of AU, interstellar flybys, and finally our Vision Mission of a fast (0.5 c
Vcruise) interstellar rendezvous mission. In this section, we will illustrate the enor-
mous difficulty, but not impossibility, of performing these missions.

A. Next-Step Teething Mission Candidates

Although often forgotten, we have in fact already launched five interstellar
precursor spacecraft in the two Voyager, two Pioneer, and the New Horizons
Pluto spacecrafts. However, even the fastest (Voyager 1) will barely leave the
solar system during its 40-year lifetime. Historically, studies of interstellar mis-
sions have considered this 40-year time frame an important mission goal because
it represents the “professional lifetime” of scientists and engineers who would
launch the spacecraft and then receive the final science data shortly before
their retirement. Thus, we tend to see 40- to 50-year mission trip time goals
for interstellar precursors and the “easier” interstellar flybys of nearby targets
(e.g., Alpha Centauri).
A wide variety of studies have been performed over the years that considered
moderately near-term interstellar precursors from hundred to several thousand
AU. These missions have included general particle and field measurements of
the interstellar medium beyond the Heliopause (as has just recently been possible
with the Voyager spacecraft), investigation of Kuiper belt objects, stellar parallax
measurements (e.g., with a 100 to 1000 AU triangulation baseline using a
telescope on the spacecraft and one at Earth), and astronomy at the Sun’s grav-
itational lens (e.g., 550 AU and beyond). These missions typically require
propulsion systems with an Isp of 10,000 lbf-s/lbm. Many of these missions
could be performed by NEP and its smaller sibling radioisotope electric
propulsion (REP), fission fragment propulsion, fusion propulsion, various
antimatter-fission-fusion hybrids, and solar sails. From the perspective of the
interstellar goal, these missions could provide “engineering” data on the near-
sun interstellar environment (e.g., dust grain number and size distribution,
density of H and D for fusion ramjets, etc.), as well as highly-autonomous,
long-duration engineering test beds for spacecraft systems (e.g., power, com-
munications, etc.).
As next-steps missions beyond 1000 AU we have Oort cloud missions, nomi-
nally from 2000 to 20,000 AU for the toroidal inner Oort Cloud, and 20,000 to
60,000 AU (1 LY) for the spherical outer Oort Cloud. For reasonable trip
times, these missions would need propulsion systems with an Isp of 100,000
lbf-s/lbm, which implies fission fragment, fusion, or hybrid antimatter-fission-
fusion systems. At this range, first-generation beam-core antimatter rockets and
light sails could also begin to be used; this could allow them to be tested in a
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 85

“small” scale mission before scaling them up to the levels required for a full inter-
stellar mission.
As a “strawman” example, we can imagine an Oort cloud interstellar flyby and
rendezvous precursor mission to 0.1 LY (6000 AU), with an assumption of a
39.9-year flight time, for a total time from launch to signal return to Earth of
40 years. Table 3 summarizes results for a single-stage light sail and antimatter
rocket propulsion system for the flyby mission, and a two-stage light sail and
single-stage antimatter rocket propulsion system for the rendezvous mission.
Here, the light sail transmitter and receiver (sail) have the same diameter.
Also, the acceleration level (gs) fixed by the light sail flyby is used for the anti-
matter rocket flyby and rendezvous. Note that there is an additional subtlety in
calculating the mass ratio (Mwet/Mdry) for the antimatter rocket rendezvous
mission when a single stage is used for acceleration and deceleration. For
example, accelerating from rest to 0.002560 c for a flyby mission results in a
straight-forward calculation of the antimatter rocket’s mass ratio of 1.01868.

Table 3 Comparison of light sail and antimatter rocket for 0.1-LY 39.9-year
flyby and rendezvous mission

System Light sail Antimatter rocket

Payload mass (MT) 100 100

Mission type Flyby Rendezvous Flyby Rendezvous

Average acceleration (g)

Stage 1 0.001487 0.0191/0.0316a 0.001487 0.001487
Stage 2 0.03290
Vcruise (c) 0.002560 0.002513 0.002560 0.002618
Antimatter rocket mass 1.01868 1.03857
ratio (Mwet/Mdry)
Laser sail
Laser diameter ¼ sail 7.03 47.71
diameter (km)
Sail only areal density 0.73 0.100
(ssail only, g/m2)
Total mass with payload 102.8 296.7
(MT)
Laser power (TW) 0.303 11.210
Antimatter rocket
Total wet mass with 357.1 374.0
payload (Mwet, MT)
Total dry mass with 350.5 360.2
payload (Mdry, MT)
Total antimatter mass (MT) 3.44 7.29
Total flight time (years) 39.9 39.9 39.9 39.9
Time from launch until signal 40.0 40.0 40.0 40.0
received at Earth (years)
a
Light sail stage 1 acceleration for Earth departure/target rendezvous.
86 R. H. FRISBEE

The corresponding single-stage antimatter rocket rendezvous mission has an

acceleration and deceleration to or from the same Vcruise (0.002618 c).
However, the single-stage rendezvous antimatter rocket’s mass ratio is not
based on the total DV (0.005236 c, with Mwet/Mdry 1.03862), but rather
based on the product of the mass ratio for each step’s DV (0.005236 c, with
Mwet/Mdry 1.01910), such that the mass ratio is 1.019102 1.03857.
Finally, the Bussard interstellar fusion ramjet has been far less studied than anti-
matter rockets or laser light sails; therefore we must defer mission analyses of the
fusion ramjet until detailed system studies become available.
These results suggest that an inner Oort cloud mission to 0.1 LY in 40 years
would be a reasonable “teething” mission for propulsion systems like light sails
and antimatter rockets, capable of ultimately performing our Vision Mission of a
fast interstellar rendezvous. For example, if we assume that we selected a flyby
mission, as is typically done for interstellar precursors, the light sail and laser
transmitter optics are only 7 km in diameter, and the laser system power is a
modest 300 GW, or about 7 times the power generated by a Saturn V at liftoff.
Also, the sail sheet need not be the high-emissivity version needed for interstellar
rendezvous because this mission has a low acceleration, making the fabrication
of the sail easier. Similarly, the mass of the antimatter rocket is less than four
times the mass of the Space Shuttle Orbiter. However, the issue here would be
to achieve antiproton production rates 16 orders-of-magnitude above today’s
rates (e.g., 100 MT/yr versus 10 ng/yr).
Alternatively, we could choose to perform a rendezvous mission as a means of
demonstrating a two-stage light sail mission with all the complexities of pointing
and tracking the sail at 0.1 LY, and using the sail’s first stage to retro-reflect the
laser beam back onto the second stage. Even though a rendezvous mission might
not be needed for science reasons, this would give us a chance to exercise all the
essential light sail systems required for an interstellar rendezvous.

B. Alpha Centauri (4.3 LY) in 75 Years

The Alpha Centauri star system [41], 4.3 LY away, is a double star (possibly a
triple star) system. The central star, Alpha Centauri A, is very similar to our own
sun; it is slightly larger and hotter, but both the sun and Alpha Centauri A are in
spectral class G2.
The second star, Alpha Centauri B, is smaller and dimmer than our sun; it
orbits far enough away from A (11 to 35 AU) that its orbit would not cause any
planets in Alpha Centauri A’s habitable “Goldilocks” zone to be ejected. For
Alpha Centauri A or B, the “safe” zone for immunity from being gravitationally
ejected stretches out to about 3 AU for planets in the same plane in which A and B
orbit around each other; however, this drops to 0.72 AU for any planets in orbits
inclined at right angle to the A/B orbit plane. Figure 23 shows the Alpha Centauri
A/B system, with the orbits of the planets of our solar system shown for scale.
Finally, there is a small, cool red dwarf star, Alpha Centauri C, that is quite
far from Alpha Centauri A/B (13,000 AU); it may not actually be a gravitat-
ionally connected part of the Alpha Centauri system, but rather a wandering
star that just happens to be near (in stellar terms) to Alpha Centauri A/B.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 87

Fig. 23 Alpha Centauri system with orbits of our solar system planets shown
to scale.

Interestingly, Alpha Centauri C is slightly closer (4.2 LY) to us than A/B, so C is

also called Proxima Centauri.
For this target, we will illustrate a 75-year duration interstellar flyby and ren-
dezvous mission; the results are given in Tables 4 and 5. As discussed previously,
systems like a two-stage fusion rocket (e.g., Daedalus) or an interstellar Orion
could perform the flyby mission or rendezvous, but the Orion trip times are sig-
nificantly longer than those for Daedalus. (Also, the flight times for the Orion are
essentially the same for flyby or rendezvous due to Orion’s high acceleration.)
As with the 0.1-AU interstellar precursor mission, we can determine the
performance of a light sail and antimatter rocket for an Alpha Centauri flyby
and rendezvous mission, with an assumption of a 70.7-year flight time, for a

Table 4 Comparison of Daedalus and Orion for Alpha Centauri 4.3-LY flyby
and rendezvous missions

System Interstellar Orion Daedalus

Mission type Flyby Rendezvous Flyby Rendezvous

Stage 1 Acceleration Acceleration Acceleration Acceleration

Stage 2 Acceleration Deceleration Acceleration Deceleration
Vcruise (c) 0.033 0.033 0.102 0.045
Total wet mass 400,000 2,000,000 53,500 53,500
(Mwet, MT)
Total flight time 129.0 129.1 43.7 98.4
(years)
Time from launch 133.3 133.4 48.0 102.7
until signal
Received at
Earth (years)
88 R. H. FRISBEE

Table 5 Comparison of light sail and antimatter rocket for 4.3-LY 70.7-year
flight time flyby and rendezvous missions

System Light sail Antimatter rocket

Payload mass (MT) 100 100

Mission type Flyby Rendezvous Flyby Rendezvous

Average acceleration (gs)

Stage 1 0.0261 0.0317/0.0326a 0.05 0.05
Stage 2 0.1918 0.05
Vcruise (c) 0.06182 0.06178 0.06134 0.06187
Antimatter rocket mass 1.56031 1.56633
ratio (Mwet/Mdry)
Laser sail
Laser diameter ¼ sail 40.49 312.98
diameter (km)
Sail only areal density 0.073 0.100
(ssail only, g/m2)
Total mass with 194.3 8,563.1
payload (MT)
Laser power (TW) 10.682 552.146
Antimatter rocket
Total wet mass with 1,876 6,518
payload (Mwet, MT)
Total dry mass with 1,202 3,477
payload (Mdry, MT)
Total antimatter mass 354 1,597
(MT)
Total flight time (years) 70.7 70.7 70.7 70.7
Time from launch until 75.0 75.0 75.0 75.0
signal received at Earth
(years)
a
Light sail stage 1 acceleration for Earth departure/target rendezvous.

total time from launch to signal return to Earth of 75 years. For this mission, with
a relatively modest cruise velocity requirement (0.06 c), we could use a single
stage flyby and two-stage rendezvous antimatter rocket. (The light sail always
uses a single stage for a flyby, and two stages for a rendezvous mission.)
Here we begin to see the enormous scale of a “true” interstellar mission as
compared to an interstellar precursor. For the light sail, the laser powers are
10s to 100s of TW (1012 W); for comparison, today all of human civilization
operates at an average power level of about 15 TW! Similarly, the amounts of
antimatter needed have jumped to 100s to 1000s of MT.

C. Gliese 876 (15.2 LY) in 75 Years

Gliese 876 [42] is a cool red dwarf (spectral class M3.5), with about 32% of
the mass of our Sun. It is the closest star to us known to possess planets. The first
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 89

planet to be detected around Gliese 867 (Gliese 876 “b” in June 1998) has a mass
of about 2.5 Jupiter-masses and orbits its stat at 0.21 AU. The next to be detected
(Gliese 876 “c” in January 2001) has an even closer orbit, 0.13 AU, and a mass of
0.8 Jupiters. The most recent to be detected (Gliese 876 “d” in June 2005) is a
potentially “terrestrial” planet (i.e., rocky as opposed to Jupiter-like gas giant)
with a mass of 7.5 Earth-masses (0.024 Jupiter-masses) in a 0.021 AU orbit.
As with the Alpha Centauri mission, the results of a 75-year duration interstel-
lar flyby and rendezvous mission for the light sail and antimatter rocket are sum-
marized in Table 6. Because of the greater distance, flight times for systems like
Daedalus or the interstellar Orion are probably unacceptable. For example, the
flight time for Daedalus for the flyby and rendezvous missions would be 165
and 356 years, respectively; the corresponding value for the interstellar Orion
is 471 years (essentially the same for flyby or rendezvous due to Orion’s high

Table 6 Comparison of light sail and antimatter rocket for 15.2-LY 59.8-year
flight time flyby and rendezvous missions

System Light sail Antimatter rocket

Payload mass (MT) 100 100

Mission type Flyby Rendezvous Flyby Rendezvous

Average acceleration (g)

Stage 1 (or 4) 0.0479 0.0476/0.0359a 0.05 0.05
Stage 2 (or 3) 0.2054 0.05 0.05
Vcruise (c) 0.26618 0.26942 0.26561 0.27953
Antimatter rocket mass 2.67912 2.82915
ratio (Mwet/Mdry)
Laser sail
Laser diameter ¼ sail 128.76 591.56
diameter (km)
Sail only areal density 0.073 0.100
(ssail only, g/m2)
Total mass with 1,053.2 30,332.8
payload (MT)
Laser power (TW) 133.4 3,286.9
Antimatter rocket
Total wet mass with 44,370 1,105,040
payload (Mwet, MT)
Total dry mass with 11,843 179,273
payload (Mdry, MT)
Total antimatter mass 17,077 486,025
(MT)
Total flight time (years) 59.8 59.8 59.8 59.8
Time from launch until 75.0 75.0 75.0 75.0
signal received at Earth
(years)
a
Light sail stage 1 acceleration for Earth departure/target rendezvous.
90 R. H. FRISBEE

acceleration). This also illustrates the classic tradeoff of interstellar missions: Do

we launch now using a slower vehicle with a longer flight time, or do we wait X
years for a faster vehicle to be developed, with a shorter flight time, because by
waiting we might actually reach our target sooner? This is a real issue for “slow”
(Vcruise  0.1 c) fission or fusion systems; however, it becomes a moot point for
relatively “fast” (Vcruise  0.5 c) systems like light sails or antimatter rockets
simply because we bump up against the speed-of-light velocity limit. Thus,
even with an ultra-relativistic (Vcruise . 0.9 c) Bussard fusion ramjet, we
would only have at most a factor of 2 reduction in flight time as compared to a
Vcruise 0.5 c vehicle.
As compared to the Alpha Centauri mission, the greater distance of Gliese 876
has a significant impact on the light sail system, primarily driven by the beam
focusing requirements for the longer distances. This has the effect of increasing
the laser transmitter optics and light sail diameters, which results in a heavier sail.
This heavier sail must also be accelerated to a higher speed in order to cover a
longer distance in the same amount of time as the Alpha Centauri mission. For
example, the beaming distance for the flyby is longer due to the increased
cruise velocity requirement, which in turn drives the laser power level. For the
rendezvous, the laser would need to focus the beam onto the sail at Gliese’s dis-
tance, and provide full power to decelerate stage 2 of the sail at its thermally
limited maximum.
For the antimatter rocket, if we attempt to use a single-stage flyby and two-
stage rendezvous vehicle, the flight time would exceed our 59.8-year target
when we limit the speed to keep Mwet/Mdry to about 5. Thus, we would have
to use a two-stage flyby and four-stage rendezvous vehicle. By using the two-/
four-stage combinations, we would keep the DV per stage to a much more com-
fortable level of Mwet/Mdry  2.7. However, even with these limits on Vcruise and
Mwet/Mdry, the vehicle mass and especially the mass of antimatter grow to
remarkable levels (e.g., half a million tons for the rendezvous mission).

D. Vision Mission: 40 LY
Our Vision Mission to 40 LY was originally selected because a sphere with
this radius contains approximately 1,000 stars. In fact, the binary star system 55
or Rho(1) Canceri [43] lies just outside this range at 40.9 AU. This system is
especially interesting because the A star of the binary has at least 5 planets,
including one inside the star’s habitable “Goldilocks” zone. 55 Canceri A is
a yellow-orange dwarf (spectral class G8) with about 0.94 times the mass of
the Sun. A fifth planet, “f” or “A5,” was discovered in November 2007 at a
distance of 0.781 AU, within the stars habitable zone. Unfortunately, this
planet’s mass (45 to 57 Earth-masses) means that it is probably a gas giant;
for comparison, it is about half the mass of Saturn (95 Earth-masses).
However, looking to our own solar system, 55 Canceri A5 might have a hab-
itable moon analogous to Jupiter and Europa. Finally, star B is a red dwarf
about 1,100 AU from star A.
Our Vision Mission has as its goal a “fast” (Vcruise 0.5 c) rendezvous at 40
LY. As shown in Table 7, the light sail system could achieve this, but the
antimatter rocket is only able to reach 0.4173 c because of the impact on
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 91

Table 7 Comparison of light sail and antimatter rocket for 40-LY flyby
and rendezvous missions

System Light sail Antimatter rocket

Payload mass (MT) 100 100

Mission type Flyby Rendezvous Flyby Rendezvous

Average acceleration (g)

Stage 1 (or 4) 0.0488 0.0850/0.0388a 0.05 0.05
Stage 2 (or 3) 0.1996 0.05 0.05
Vcruise (c) 0.5 0.5 0.4173 0.4173
Antimatter rocket mass 5.000 5.000
ratio (Mwet/Mdry)
Laser sail
Laser diameter ¼ sail 239.61 961.95
diameter (km)
Sail only areal density 0.073 0.100
(ssail only, g/m2)
Total mass with 3,400.8 80,044.0
payload (MT)
Laser power (TW) 608.922 17,889.764
Antimatter rocket
Total wet mass with 1,127,632 83,980,666
payload (Mwet, MT)
Total dry mass with 156,792 7,661,580
payload (Mdry, MT)
Total antimatter mass 509,691 40,067,520
(MT)
Total flight time (years) 84.97 84.07 99.9 103.9
Time from launch until 124.97 124.07 139.9 143.9
signal received at Earth
(years)
a
Light sail stage 1 acceleration for Earth departure/target rendezvous.

Mwet/Mdry of the Relativistic Rocket Equation when a 0.4. As discussed

earlier, we must to limit the speed to keep Mwet/Mdry to about 5.0. Also, stage
2 (or 3) starts at a higher speed than stage 1 (or 4), so the DV for stage 2 or 3
(0.1987 c) would be less than that for stage 1 or 4 (0.2186 c) so as to have the
same Mwet/Mdry. Note also that, although staging reduces the DV per stage and
thus Mwet/Mdry, each stage has a Mwet/Mdry of 5.0, so the total four-stage rendez-
vous vehicle would have a wet mass at least (5.0)4 625 times the dry mass of
the fourth (final) stage. Thus we see the need for truly prodigious amounts of anti-
matter (40 million tons!) for the four-stage rendezvous mission. Finally, because
of the limitation on Vcruise, the trip time is about 20 years longer than for the
light sail.
As discussed previously, there are two additional complications introduced
when using a light sail for a rendezvous mission. These both involve the
92 R. H. FRISBEE

Fig. 24 Comparison between classical and relativistic light sail parameters.

problem of using stage 1 to retro-reflect the laser beam back onto stage 2. This has
the effect of decelerating stage 2 to a stop, but accelerating stage 1. Also, as stage
2 is decelerating, stage 1 is moving out ahead in front of stage 2, thus increasing
the retro-reflection beaming distance between stage 1 and stage 2. The increasing
speed of stage 1 exacerbates the already large amount of laser beam redshifting
due to the relative speeds between stage 1 and 2. This is at its maximum at the end
of the rendezvous maneuver, when stage 2 is at rest and stage 1 is at its maximum
velocity, e.g., 0.597 c for this example. This has the result of dramatically
increasing the required laser system power by a factor of 3.97 at 0.597 c due
to “double” redshifting of the laser’s photons seen by stage 2. Finally, redshifting
of photons beamed from stage 1 onto stage 2 results in a spot size increase of 1.99
times its classical (nonrelativistic) value. The various relativistic effects are illus-
trated in Fig. 24 and Table 8.
The other potentially serious light sail feasibility issue is the ability of a very
lightweight gossamer structure like stage 1 to act as a high-quality optical trans-
mitter for beaming laser light over near-interstellar distances to stage 2. In this
example, the stage 1 2 separation distance at the end of the rendezvous maneu-
ver is 0.725 LY, or roughly one-sixth the distance between us and Alpha
Centauri.

Table 8 Light sail relativistic effects

Mission step Earth departure Rendezvous

Light sail stage Stage 1 Stage 1 Stage 2

Velocity (c) 0.00 ! 0.50 0.50 ! 0.597 0.50 ! 0.00
M/Mrest 1.00 ! 1.15 1.15 ! 1.25 1.15 ! 1.00
l/lrest 1.00 ! 1.73 1.73 ! 1.99 1.73 ! 3.97
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 93

E. Comparison of Light Sails and Antimatter Rockets for Robotic

Interstellar Missions
For our initial mission examples at target distances of less than about 15 LY,
the laser light sail and antimatter rocket were at significantly less than their
maximum performance as required for the 40-LY Vision Mission. In Fig. 25,
we illustrate the performance of these vehicles for 75-year, 40-year, and
minimum total mission time robotic missions, where the total mission time con-
sists of the flight time plus the time required to return the first data (at the speed of
light) back to Earth.
As expected, very “fast” missions, with Vcruise approaching 0.5 c, can yield
short flight times, but only at the cost of enormous laser powers or quantities
of antimatter. (Note the logarithmic power and mass scale in Fig. 25.) Also,
for more distant missions, the time required to return the data signal to Earth
becomes a significant fraction of the total mission time; e.g., the Vision
Mission rendezvous light sail has a flight time of 84 years to 40 LY, plus a 40-
year signal travel time back to Earth. However, even at these limits of perform-
ance and even if laser power and antimatter were “free,” we would ultimately be
constrained by such “mundane” engineering factors as the temperature limits of

Fig. 25 Comparison of light sail laser power and antimatter rocket antimatter mass
as a function of total mission time. Total mission time 5 flight time plus time for first
signal to reach Earth.
94 R. H. FRISBEE

materials (i.e., light sail thermally-limited acceleration) or the Rocket Equation’s

inexorable mass ratio (i.e., Mwet/Mdry as a function of DV per stage).

F. Generational Ships
At the extreme limit of infinite acceleration (i.e., no time spent accelerating)
and a cruise velocity of 0.5 c, a robotic mission could only travel as far as
13.33 LY and have the science data returned within the Earth-bound scientist’s
“professional” lifetime of 40 years, ranging nominally from a person’s early
20s to their early 60s. Any robotic mission beyond this distance would require
multiple generations of humans between launch and final data return. For
piloted missions with the crew on board the vehicle, we are faced with two
options. First, we can assume a one-way mission, where the primary goal is
exploration and not colonization from subsequent generations from the original
crew. In this case, we can assume a 40-year time period, but now with the distinc-
tion that it is a 40-year flight time, as opposed to flight time plus data return time
(at the speed of light) for the robotic mission. In this regard, Forward has explored
the ethical issue of sending a crew on a one-way mission in a fictional setting of a
piloted light sail mission [44]. Here, the hero is asked about his feelings on
embarking on a one-way mission from which there is no “rescue.” He replies
simply that we are all on one-way “missions” through life, with the subtext
that the real question is not how long we live, but rather what we do with
our lives.
In terms of flight time, if we take the “mundane” science fiction assumption
that there are no convenient loopholes for long-duration missions (e.g., no sus-
pended animation/hibernation, no lifespan-extending drugs, etc.), then we are
faced with the need for multigeneration human missions. Multigeneration ships
would also be needed for colonization [45]. In this case, the mission flight
time could be extended indefinitely, but the payloads would become very large
due to the need to carry a sufficient human population to prevent inbreeding
and/or genetic drift after a few generations. Estimates of the population required
have ranged from a low of 50 (a minimum to prevent excessive inbreeding) to
around 500 (a level that balances the rate of gain in genetic variation due to
mutation with the rate of loss due to genetic drift). A value of 150 to 180
would make it possible to have a stable crew size over 60 to 80 generations
(roughly 2,000 years) with each generation having a good selection of potential
mates. These numbers could, in principle, be driven even lower with the use of
stored/frozen sperm, eggs, or embryos [46].
However, these population estimates were based only on the need to minimize
inbreeding and genetic drift effects. Much larger populations may be desirable to
overcome the inherent limitations of small populations. For example, the small
populations described above did not take into account the psychology or
group-dynamics issues of small groups. Also, they made no allowance for the
need for a large skills mix of potential colonists. This might be especially import-
ant on a new world, where the high-tech imports from the original colony ship
would eventually break down, and the colony reduced to a mid-1800s level of
technology (e.g., plows pulled by horses rather than tractors). Finally, it may
be just as important to have a broad range of social and cultural variation as
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 95

genetic variation. For example, it is humanity’s rapid social/cultural adaptability

that has allowed us to survive dramatic changes in environment and fill virtually
every ecological niche on Earth over the last 100,000 years, an eye blink in time
compared to the slow march of genetic evolution.
Thus, to assess the worst-case mass impact of a large population size, we
assumed a vehicle payload of 500,000 MT with a population of 10,000 corre-
sponding to the mass and population of an O’Neill “Island One” (Bernal
sphere) space colony [47]. Ironically, this is one time when the enormous mass
of the antimatter rocket acts in our favor because is would have such a high
dry mass that adding even a large payload (e.g., 500,000 MT) would represent
only a modest increase in the dry mass of the total vehicle. For example,
adding a 500,000 MT payload to our Vision Mission 40-LY rendezvous four-
stage antimatter rocket would only increase its wet mass and antimatter mass
by a factor of 32.5 and 32.8, respectively.
Unfortunately, the situation for the two-stage (rendezvous) light sail would be
more complicated because of the impact that a large payload has on the stage 2
thermally limited acceleration. Here, adding a large payload to stage 2 (7,267 MT
without its 100 MT payload) would decrease its allowable deceleration; we either
have to limit the cruise velocity so that stage 2 could be decelerated (from the
lower Vcruise) without exceeding the thermal limits on stage 1 or 2, or we
would have to make the sail larger, so that the payload again would become a
small fraction of the mass of the sail. For example, keeping the sail diameter
the same as our baseline case (962 km with 100 MT payload), we would have
to limit Vcruise to 0.055 c. Under these conditions, the laser power would grow
by roughly a factor of two, from the baseline 17,890 TW to 32,129 TW.
However, the flight time would dramatically increase to 737 years. At the
other extreme, keeping Vcruise at 0.5 c would require an almost seven-fold
increase in sail diameter, from the baseline 962 km to 6639 km, and an almost
84-fold increase in laser system power (17,890 TW to 1,495,158 TW). This
does, fortunately, keep the flight time to a more reasonable 84.7 years.
Needless to say, these numbers represent prodigious quantities, i.e., “only”
1.3  1017 years to produce the 1.3 billion tons of antimatter required for a multi-
generation colony ship at today’s production rates (10 ng/yr) or a light sail laser
system power level “only” 105 times the total power output of human civilization.
Thus, the physics of the problem says that these missions are not impossible,
although one might quite reasonably question their practicality.

VII. Most Critical Technologies and Feasible Engineering Limits

In this section we will discuss several technologies and their potential limits.
We will also discuss several other issues, such as societal resources that might be
applied to interstellar missions, and the limits of plausibility of performing an
interstellar mission as distinct from its physical and technical feasibility.

A. Specific Impulse
In many space propulsion systems it is necessary to trade thrust (which
directly impacts vehicle acceleration) for Isp in order to reach an acceptable
96 R. H. FRISBEE

flight time. This is because for a given total power level, thrust is inversely pro-
portional to Isp. Thus, in order to achieve a fast Earth Mars round trip time, a
fusion propulsion system would be designed to run at an Isp of 20,000 lbf-s/
lbm [48], even though it is theoretically capable of 106 lbf-s/lbm. In this case,
the selection of Isp depends on the mission needs in terms of both DV and trip
time; too high an Isp can result in too low a thrust and acceleration, and thus
too long a trip time.
We see the same effect in electric propulsion thrusters. For example, Hall
thrusters have an acceptable efficiency (50%) in the relatively low Isp
range (few thousands of lbf-s/lbm) that is optimum for Earth orbit raising
(i.e., DV of a few km/s). By contrast, ion thrusters have very poor efficiency
in this Isp range. Instead, ion engines have much better efficiencies at Isp’s
appropriate for the large DV’s and long distances of interplanetary missions.
In this context, interplanetary distances are long enough that there is an oppor-
tunity to use even a low acceleration vehicle to accelerate to a high cruise
velocity. Ironically, as far as low-acceleration electric propulsion vehicles are
concerned, Mars is too close for high-Isp vehicles to build up enough speed
for a short flight time; depending on the mission, it may be better to use a
lower-Isp option.
Unfortunately, this focus on near-term mission applications, where the ultra-
high Isp’s needed for interstellar missions are not required, has hampered devel-
opment of these systems. Thus, the vast majority of the propulsion technologies
discussed here for interstellar precursor and full-up interstellar missions are
purely conceptual (e.g., fission fragment, fusion, antimatter); their performance
is based on calculations using various theoretical models. Some technologies
have had important proof-of-concept experiments performed for them (e.g.,
in-space deployment of a solar sail-like spinning structure by the Russians,
“noncontact” magnetic levitation of normal-matter solid H2 [49], etc.) but no
propulsion technology with performance required for interstellar missions exists.
Even for the more modest interstellar precursor mission where Isp’s of 10,000
to 20,000 lbf-s/lbm are needed, only electric propulsion using ion thruster tech-
nology is currently even close, with the power system lagging far behind.

B. Specific Power and Specific Energy Density

We can estimate the ideal energy density (energy per unit mass, E/M) that
must be delivered to a propellant by combining the (Classical) Rocket Equation
and the definition of kinetic energy (KE) of the propellant (Ep). Thus:

KE ¼ 1=2MV 2 ¼ 1=2Mp (Vexhaust )2 ¼ 1=2Mp (gc Isp )2 ¼ Ep

Rearranging and solving for (gcIsp) gives:

(gc Isp ) ¼ (2Ep =Mp )1=2

From the Rocket Equation, we have:

‘n(Mwet =Mdry ) ¼ DV=(gc Isp ) ¼ DV=(2Ep =Mp )1=2

 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 97

We rearrange this equation and solve for the propellant energy density (Ep/Mp)
as a function of DV and the assumed vehicle mass ratio:

(2Ep =Mp )1=2 ¼ DV=‘n(Mwet =Mdry )

(Ep =Mp ) ¼ 1=2½DV=‘n(Mwet =Mdry )2 ¼ 1=2½DV  ‘n(Mdry =Mwet )2

Thus, the required energy density grows as the square of the required DV.
Figure 26 illustrates this trend for a range of Mdry/Mwet of 10% (very challenging
to build) to 50%, corresponding to Mwet/Mdry 10 to 2.
From this, we see that a single-stage to low Earth orbit (SSTO LEO) vehicle
is just on the edge of feasibility with chemical (O2/H2) propellants, and then only
if Mdry/Mwet is around 10%, a value that represents a major challenge. For a
single-stage to geosynchronous Earth orbit (SSTO GEO) vehicle, we would
need a high energy density matter (HEDM) propellant like free-radical
(atomic) H. More ambitious missions require progressively higher and higher
energy to be imparted to the propellant, with matter antimatter annihilation
the extreme case. Interestingly, fission and fusion energy densities are very
similar, which is why the Isp or Vexhaust for fission-fragment and fusion propulsion
are essentially the same. Finally, an electron-positron annihilation pair would
have an energy density of c 2, but a proton-antiproton pair is less because not
all of the initial matter is converted into energy.
We can also calculate the required vehicle specific power, e.g., kg/W, based
on the fundamental definition of (classical) rocket engine “jet” or exhaust power

Fig. 26 DV capability as a function of propellant energy density.

98 R. H. FRISBEE

as a function of Vexhaust (gc Isp) and thrust (F):

Pjet ¼ 1=2gc Isp F

We combine this with the definition of force as a function of mass and

acceleration:

F ¼ MA

where in this case M is the total mass of the vehicle and A its acceleration. Com-
bining these two equations and solving for specific power as M/Pjet (e.g., kg of
vehicle/Wjet) gives:

Specific power ¼ M=Pjet ¼ 1=(1=2gc Isp A)

We have seen above that the minimum practical acceleration for interstellar
missions is 0.01 g; any less than this and the flight times become excessive
because so much time is spent in acceleration (or deceleration). By contrast,
acceleration at more than 1 g is not needed because the acceleration (or decelera-
tion) times approach zero. Thus, we can “bracket” the range of total vehicle
specific power needed for interstellar missions, as shown in Fig. 27.
Not surprisingly, interstellar fusion vehicles (Isp  1,000,000 lbf-s/lbm) and
antimatter rockets (Isp  10,000,000 lbf-s/lbm) will require extraordinarily low
specific masses compared to contemporary and near-term vehicles (e.g., NEP
on the order of 10 kg/kWjet at megawatt power levels).

Fig. 27 Total vehicle specific power (kg/kWjet) as a function of vehicle acceleration

and propulsion system Isp.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 99

C. Critical Nonpropulsion Technologies

Although the focus of this chapter has been on propulsion technology, there
are several additional critical technologies that will require major advancements.
For example, because of the finite limit of the speed of light, round-trip communi-
cation times will be measured in decades. Thus, the vehicle will require extre-
mely advanced autonomy (e.g., software) and avionics (e.g., hardware), which
are separate functions in today’s spacecraft but must grow to become a single
function. More generally, robotic spacecraft will require artificial intelligence
on the level of humans if they are to serve as our surrogate explorers at interstellar
distances. Similarly, structures technology requires major advancements due to
the very large size of the various concepts (e.g., dimensions on the order of thou-
sands of kilometers). In fact, there are two major technology paths in structures;
the first involves large, ultra-lightweight, thin-film structures for the light sails;
the second involves large, “heavy” structures for systems like the light sail
laser array, the antimatter rocket, and the fusion ramjet.
One area that will require significant advancement is reliability because of the
long overall mission times. Other critical technologies that will require signifi-
cant (but not major) advancement include communications systems, power
systems, and navigation. Onboard spacecraft power at sub-kilowatt levels
could be met by advanced nuclear power systems like radioisotope thermoelec-
tric generators. However, large propulsion system related, multi-megawatt
power systems, like those envisioned for advanced NEP vehicles, may be
required for the antimatter rocket and fusion ramjet systems. This includes
energy storage systems for startup power, thermal-to-electric power conversion
during engine operation, and housekeeping power during coast (for cryogenic
refrigeration systems, electromagnetic storage of antimatter, etc.). Various
power system options and their limitations are discussed in more detail in
Chapter 17. Finally, navigation will require advancements in position knowledge
(e.g., advanced optical navigation), timing (e.g., advanced, highly accurate, and
stable clocks), and acceleration (changes in position and time). Finally, we will
conclude with recommendations for future systems modeling and basic research
related to assessing the various elements in interstellar propulsion system.

1. Autonomous Reliability Versus Reparability

Any long-duration space system will require a high level of reliability and
system lifetime. With a requirement for systems to operate for decades to
centuries, it may be necessary to rethink our traditional assumptions about
trading performance for lifetime. For example, instead of pursuing the goal of
maximum performance, we may need to design systems for ease of maintenance,
repair, or replacement, even if this means sacrificing some level of performance.
Also, in the context of a highly intelligent robotic spacecraft, or ultimately a
piloted mission, it is possible to imagine a completely autonomous vehicle
where replacement parts are manufactured on the vehicle as needed; in effect,
the vehicle would have its own “machine shop” and robots to perform the
needed work. This also introduces the idea of sacrificing performance for ease
of manufacturability in a completely autonomous robotic environment.
100 R. H. FRISBEE

Fig. 28 Radio and optical communication system transmitter power as a function of

transmission distance. (Data from Ref. 50.)

2. Minimum Communication System Power

Interestingly, Lesh et al. [50] have already shown that optical (laser) com-
munication over interstellar distances is feasible given modest extrapolations
of the technology. This also suggests that programs like SETI (Search for Extra-
terrestrial Intelligence) may have better success when observations are made at
near-visible wavelengths, rather than the current microwave searches.
Figure 28 illustrates the general trends seen in optical and radio communi-
cation systems as a function of transmission distance. However, it should be
noted that the values shown are from a wide variety of point designs with
widely different assumptions [50] about system design (e.g., transmitter and
especially receiver architectures), technology levels, efficiencies, and so on.
However, we do see the general trend that radio is more power-intensive
than optical.

D. Societal Investments in Interstellar Missions

Given the inherent enormous scale of any interstellar mission, one question
that can be asked is what resources will a civilization be willing to expend for
an interstellar mission? Previously, based on U.S. government spending as a frac-
tion of the U.S. gross national product (GNP), we have estimated that about 10%
of the GNP might be available based on a combination of NASA and defense
spending [14]. However, this does not give us a feel for the magnitude of the
gap between today’s total resources and the requirements of an interstellar
mission. Figure 29 attempts to answer this question in terms of the total annual
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 101

Fig. 29 Comparison of energy and power of Human Civilization and Daedalus

fusion rocket.

energy and power output of human civilization as compared to the requirements

of even a “modest” interstellar capability. Not surprisingly, it would take roughly
five years of the total annual energy output of human civilization to equal
the energy required for the minimal interstellar capability represented by the
Daedalus two-stage 0.1 c flyby fusion vehicle. Finally, as a side note, Fig. 29
also shows the change of slope in annual energy production and consumption
due to the energy crisis of the 1970s.
For comparison, the energy content of annihilating the antimatter in the Vision
Mission rendezvous four-stage antimatter rocket is capable of vaporizing the
entire surface of the Earth to a depth on the order of 100 m. Similarly, the
17.9-PW laser system required for the Vision Mission rendezvous light sail has
the capability of delivering the energy equivalent of a 4-megaton nuclear
weapon per second into a 0.7-m diameter spot at a distance of 2 AU from the
laser system. This suggests that it would be prudent to base an antimatter
“factory” or light sail laser in a 1 AU orbit on the side of the Sun opposite
Earth, although the light sail laser would have to contend with zodiacal dust scat-
tering if placed in an orbit in the plane of the ecliptic [51]. More generally, the
ability of an advanced civilization to destroy itself before reaching interstellar
flight capability has been an ongoing issue with estimating the lifetime of a tech-
nological civilization for use in the Drake Equation.

E. Physical Feasibility Versus Engineering Feasibility

We mentioned previously the inherent enormous scale of fast interstellar ren-
dezvous missions. Human civilization is certainly capable of committing enor-
mous resources to projects considered important to the civilization. History
abounds with examples of amazing engineering feats that would be difficult for
us to duplicate, even with 21st century technologies, ranging from the Great
102 R. H. FRISBEE

Pyramid of Egypt to the Apollo Saturn V at liftoff (with a first-stage rocket power
corresponding to 0.8% of the steady-state power level of human civilization in
1969). However, at some point we have to ask about the feasibility of attempting
some engineering task, given its inherent “difficulty.” As an often-quoted
example, we could in principle recreate conditions present during the initial
stages of the Big Bang with some incredible extension of current particle accel-
erator technology; however, to do so would require a particle accelerator ring the
diameter of the galaxy!

1. Antimatter Rocket Issues

Thus, there can be situations where the fundamental physics of the problem
do not preclude it being done, but the engineering technology required to actu-
ally implement a solution are not feasible because of cost, materials, tempera-
tures, etc. For example, we have not identified any physics showstoppers in
operating a matter antimatter annihilation rocket, although the four-stage
40-LY rendezvous vehicle would be enormous in size and require 40-million
MT of antimatter. Similarly, there are no inherent physics issues in producing
antiprotons because this is already done (albeit at 10s of nanograms per year).
Physical feasibility should also extend to the conversion of antiprotons to anti-
solid H2, although the necessary steps have not yet been demonstrated. The
problem here is the engineering of an antimatter “factory” capable of
producing millions of tons of antimatter per year at very high energy efficiency.
In particular, the production efficiency issue may prove to be a potential show-
stopper, because current production technology has an energy efficiency of
only about 1 part in 109 (i.e., 109 units of energy are consumed to produce
an amount of antimatter that will release one unit of energy upon annihilation)
[28]. To illustrate the magnitude of the engineering feasibility issue,
the annihilation (MC2) energy of 40-million MT of antimatter (plus an equal
amount of matter) corresponds to 16 million years of current human civi-
lization energy output. At current production efficiencies (10 9), the energy
required to produce the antiprotons corresponds to 16 quadrillion
(1015) years of current human civilization energy output. For comparison,
this is “only” 601 years of the total energy output of the sun. Forward has ident-
ified a number of potential improvements that could be made to various anti-
proton production facilities around the world, but even at the maximum
predicted energy efficiency of these facilities (0.01%) [28], we would need
160 billion years of current human civilization energy output (2.2 days of
the sun’s total energy output) for production. Thus, the sheer magnitude
of the problem of producing enough energy, in the form of electricity, to make
the antimatter is a major issue. In effect, we might very well have to first build
a Dyson sphere to collect all of the sun’s energy output and convert this into
electricity (e.g., for 10 days at 22% sunlight-to-electricity efficiency) before our
first interstellar rendezvous mission using an antimatter rocket.
However, although the difficulty of producing large quantities of antiprotons
may ultimately prove to be a fatal show-stopper to this approach, it is worth con-
sidering the general trends seen in technological advancement. We have already
illustrated the tremendous improvement in the production of liquid hydrogen
from Dewar’s first few drops in the late 1800s to the space shuttle’s 100 MT
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 103

per flight. A similar trend is seen for the production of aluminum metal that
emphasizes the importance of technological breakthroughs and synergisms
with other technologies. For example, in the mid-1850s, aluminum metal was
so hard to produce that it was more expensive than gold. Emperor Napoleon
III commissioned a set of aluminum dinner cutlery for use by his most
honored and important guests; less important guests dined on “mere” silver.
And yet, the discovery of the Hall Heroult electrolysis process barely 30
years later (1886) would drive the cost of aluminum down to the point that the
Wright brothers could afford to build an aluminum crankcase engine for their
1903 Flyer. However, the breakthrough of the Hall Heroult electrolysis
process would have been of only modest importance if it had not been coupled
with the growth of an electric power industry in the latter third of the 19th
century. Thus, the mass production of inexpensive aluminum metal depended
on the synergism between a breakthrough discovery and the availability of
cheap, reliable, plentiful electric power [52]. Hopefully, between now and our
first interstellar missions, there will be an antiproton production breakthroughs
and synergisms that echo those of aluminum production.

2. Laser Light Sail Issues

In a similar fashion, we have sized the laser light sail assuming diffraction-
limited optics on the laser power-beaming system. In practice, it can be very dif-
ficult to engineer optics systems that approach this limit of performance. In part,
we made the assumption that this would not be an unreasonable goal for the laser
transmitter optics based on the recognition that telescope/interferometer systems
capable of resolving continent-sized features on planets at 40 LY would have
already solved the engineering problems of optics pointing, tracking, and beam
jitter that would be required for the laser light sail rendezvous mission. As a
numerical example, we can estimate the pointing requirement by assuming a
pointing accuracy to within 100 km (e.g., the size of a single pixel when
imaging an Earth-sized planet in a 128  128 image array), corresponding to
roughly 10% of the sail diameter for the Vision Mission 40-LY rendezvous
light sail. The arctangent of the angle of a triangle 100 km high by 40 LY long
is 0.26 pico-radians, which is comparable to picking out a 25 cent piece on
Mars (L 0.6 AU)! Obviously, we cannot do this today, but this is the class
of capability that will be required for the Origins Program to implement space-
based observatories like terrestrial planet imager (TPI). Thus, although this is a
major challenge, it will have already been solved by the astronomical community
before we begin our flights to the stars.
However, a potentially more serious engineering issue with the light sail is its
ability to use an ultra-gossamer first-stage (with a “mirror” film 63-atoms thick!)
to retro-reflect laser light back onto the second stage during the rendezvous. One
approach to solving the potential problem of poor focusing ability in the first
stage is to simply increase the diameter of the second stage to capture more
of the reflected light, although this will have the effect of decreasing the accelera-
tion of the vehicle as it leaves the solar system. More generally, the size of the
second stage is somewhat arbitrary; we assumed a second-stage area 1/10 that
of the first stage, although it may be necessary to increase the size of the
104 R. H. FRISBEE

second stage for nearby targets even when the first stage has diffraction-limited
optics capability [2].
Another issue for the light sail is the enormous size (e.g., physical size and
power level) of its laser (momentum beaming) system. At the least, we can
conduct a “sanity check” on the magnitude of the problem. For example, for a
17.9 PW laser system, the amount of solar power required to run the laser (at
10% efficiency of converting sunlight into laser light) is comparable to “only”
the amount of sunlight hitting the surface of the Earth. Thus, the sunlight collec-
tor is certainly large, but not of the order of a Dyson sphere.
Second, if we assume that the laser system is composed of individual laser þ
optics modules that are assembled in space to form the complete 17.9-PW,
roughly 1000-km diameter laser transmitter system, we can perform a sanity
check on the number of modules required and the optics diameter and laser
power per module. For example, if we assume that there are 109 modules
(comparable to the number of automobiles on Earth), the optics diameter and
laser power per module are on the order of a few 10s of meters and 10s of
MW, respectively. (For comparison to the familiar automobile engine, 100
horsepower 133 kW 0.133 MW.)
Finally, it is possible to estimate the potential mass of the laser transmitter
system based on its individual subsystems [2]. Note that this is only an approxi-
mation because these mass estimates are based on contemporary and near-term
technologies; many of these subsystems may undergo major improvements
before we attempt construction of a high-power laser power/momentum
beaming system. Using the system sizing estimates given in [2], we obtain the
results shown in Figs. 30 and 31.
Thus, the total light sail plus laser system has a mass of 210.8-billion MT, with
most of the mass in the radiators, which is not uncommon for any space-based
power system. The laser system’s mass is roughly 1,000 times the mass of the
antimatter rocket. However, this is not an entirely fair comparison, because we
have not included the mass of an antimatter “factory,” and we have not “amor-
tized” the cost (mass) of the antimatter factory or laser by dividing by the
number of missions each would support.

Fig. 30 17.9-PW laser system for the 40-LY 0.5 c interstellar rendezvous light sail
mission.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 105

Fig. 31 Laser system size compared to Earth.

3. Fusion Ramjet Issues

Finally, as discussed earlier, the Bussard interstellar ramjet has both the great-
est mission potential and the greatest number of unresolved feasibility issues. For
example, we know that stars can “burn” H fuel in a fusion reaction, but there do
not appear to be any practical engineering solutions to pure hydrogen fusion. For-
tunately, an alternative is to use interstellar D, which we currently use in fusion
reactors, although this places great demands on the ram-scoop system. Also, the
ram-scoop and fusion engines have several unique engineering feasibility chal-
lenges. First, we need to design a lightweight (i.e., minimum structure) electro-
magnet system to produce the required field geometry to collect interstellar H
or D [53]. This in itself may be a significant problem given the high L/D required
for a drag-free collection system. Similarly, the fusion engine must be designed in
such a way that the fusion fuel atoms are never slowed down (relative to the
vehicle), otherwise the momentum drag will overcome the rocket exhaust
moving at only 0.03 c. In engineering terms, this means that the usual techniques
of turning or reflecting the fusion plasma in the reactor to increase residence time
cannot be used and instead some other approach must be used to confine the
plasma in the reactor.

F. Recommendations for Future Work

During the course of our studies we have identified several additional research
areas that we recommend be pursued. In several cases, this work would consist of
systems-level studies of various elements of the interstellar propulsion systems.
For example, the Bussard interstellar fusion ramjet has been studied far less than
antimatter rockets or laser light sails. Thus, we would recommend detailed mod-
eling of all aspects of the fusion ramjet, and especially the ram-scoop structure
and electromagnetic field, as well as investigation of fusion reactor concepts
that would lend themselves to the “drag free” fusion engine requirement of the
fusion ramjet.
106 R. H. FRISBEE

Beyond system studies, we recommend experimental research on determining

“engineering” properties of interest to laser light sails, and especially the optical
properties of ultra-thin films (e.g., high-temperature metals, carbon, etc.). Also,
there are several fundamental feasibility issues associated with antimatter
rockets that need to be demonstrated experimentally, although in several cases
normal-matter protons, electrons, hydrogen atoms, and hydrogen molecules
can be used instead of antimatter (in appropriate “noncontact” experiments to
simulate what would be needed to handle antimatter). For example, although a
modest number of anti-H atoms have been produced, the complete process of
“noncontact” conversion of H atoms to H2 molecules to H2 molecular solid ice
needs to be demonstrated. Also, additional work needs to be done in investigating
high-efficiency, high-production rate processes for the production of antiprotons,
and ultimately conversion into anti-H2 ice. We would also like to see an estimate
of the theoretical ideal limit of antiproton production efficiency in terms of elec-
tric energy “in” versus antiproton mass “out,” as a point for comparing various
current and potential antiproton “factory” efficiencies. Additionally, one of the
major inefficiencies of antimatter rockets is the low fraction of proton antiproton
annihilation products with charge and mass (i.e., charged pions) that can be used
to produce thrust. Thus, we also recommend research that would seek to increase
the fraction of usable annihilation products and thereby reduce the effective
“loss” of propellant mass, while still maintaining the extraordinarily high Isp of
the antimatter rocket engine. Finally, cross-cutting technologies such as high
current density high-temperature superconductors and lightweight radiators
that are applicable to many advanced power and propulsion applications
should be pursued.
Lastly, impacts with interstellar dust grains at relativistic speeds introduce a
requirement for a dust impact shield. However, there are currently no data for
particle-shield interactions at the anticipated speeds (e.g., 0.5 c) of interstellar
missions, although this has been investigated for micrometeoroid impacts at
the more modest speeds of solar system velocities (e.g., several 10s of km/s).
Also, interstellar precursors should carry instruments to characterize interstellar
dust particle properties (number, size, material distributions) that could serve to
benchmark current models based on astronomical observations.

VIII. Closing Comments

The most likely candidates for fast (i.e., 0.5 c) interstellar rendezvous
missions laser light sails, antimatter rockets, and fusion ramjets all suffer
from either a need for a large infrastructure or a host of major feasibility issues.
Thus, there is no single concept that is without potentially significant shortcom-
ings and there is no single clear winner. Laser light sails, antimatter rockets, and
fusion ramjets are the “best” choices based only on our current level of under-
standing (i.e., ignorance!).
For example, although fusion propulsion does not require a unique infrastruc-
ture (even an He3 production infrastructure might be shared with terrestrial fusion
users), there remain major feasibility issues associated with all of the fusion
concepts at the performance levels required for interstellar missions.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 107

Similarly, although matter antimatter annihilation propulsion has been recog-

nized for decades as the “ultimate” propulsion capability, a major antiproton pro-
duction infrastructure is required for its implementation. There is however a
potential for significant dual-use of antiprotons for non-propulsive uses that
might “amortize” and justify construction of an antiproton “factory.” Also,
there are a number of major feasibility issues associated with the production
and storage of high-density forms of antimatter (e.g., solid anti-H2).
Beamed-momentum light sail systems also require a large space-based infra-
structure of beam source and optics systems. However, this infrastructure can
lead to a major in-space capability for beamed power to a variety of users, result-
ing in a “public utility” for power in space. Many of the required technologies for
a beamed-momentum light sail system are in the process of being demonstrated
(e.g., high-power lasers, adaptive optics, solar sails, etc.); the major issue is
scaling to the ultra-thin sail sheets, large sizes, and high powers required for inter-
stellar missions. Finally, although a light sail system may be the most near-term
option, it gives the least mission flexibility; every maneuver at interstellar dis-
tances must be planned years in advance because of the speed-of-light time
lag. Therefore, matter antimatter annihilation and fusion systems should also
be pursued because these systems will provide the highest degree of mission
flexibility.
Near-term technology development goals should seek to resolve the major
feasibility issues associated with each concept. For example, the mechanical,
thermal, and optical properties of ultra-thin film materials is largely unknown
and represents a major question that must be answered in order to assess light
sails. For matter antimatter propulsion, experiments with normal matter
should seek to duplicate, in a manner appropriate for antimatter, the steps
required to go from protons and electrons to atoms, then to molecules, and
finally to solid molecular ice. For the fusion ramjet, the H H (or D D) fusion
reaction, magnetic scoop, and “drag-free” reactor concepts should be computa-
tionally modeled. Finally, cross-cutting technologies, such as lightweight radia-
tors or high-temperature, high-field, radiation-resistant superconducting magnets,
should be pursued.
Ultimately, the issues that we worry about today may turn out to be a nonpro-
blem given the technological advances that will occur over the next several cen-
turies. We may very well be in the position of Jules Verne trying to predict the
engineering required to send men to the moon in his classic From the Earth to
the Moon. Verne got the space capsule right [54]; it was very similar to the
Apollo command module in size, weight, crew (three), and material (aluminum).
However, Verne completely missed the mark in designing the launch vehicle; he
used a 274-m long cannon (derived from the “high-tech” weapons of the 1860s)
instead of a Saturn V rocket (derived from the “high-tech” weapons of the 1960s).
Thus, in our case, we may be designing the antimatter rocket like Verne’s cannon,
and ultimately have a vehicle design that is way too big and heavy, simply
because we cannot predict the major technological breakthroughs that will
occur over the next several centuries.
We would like to close this chapter with the often-noted observation that inter-
stellar missions are not impossible [55]; they are just enormously demanding in
terms of size and resources. What is remarkable is that in the brief time we have
108 R. H. FRISBEE

been a space-faring species (barely a half century), we have already identified

three candidate propulsion concepts with the potential for performing the most
propulsion-intensive space missions of all, and allowing us to respond to
Konstantin Tsiolkovsky’s classic challenge [17], “Earth is the cradle of human-
ity, but one cannot live in a cradle forever.”

Acknowledgments
The work described in this chapter was carried out at the Jet Propulsion
Laboratory, California Institute of Technology, under a contract with the
National Aeronautics and Space Administration (NASA).
The author would like to thank John Cole (former Head of the Revolutionary
Propulsion Program at NASA Marshall Spaceflight Center) for providing overall
funding support for this task in previous years. Although the Revolutionary Pro-
pulsion Program has been cancelled, I hope that it will be reinitiated in coming
years to provide an opportunity to continue investigation of advanced propulsion
concepts for interstellar missions.

Appendix: Derivation of the Classical and Relativistic Rocket Equations

In this appendix we describe the derivation of four versions of the Rocket
Equation. The first is the Classical Rocket Equation, first derived by Tsiolkovsky
[17]. The next is a Relativistic Rocket Equation, in which both the vehicle and the
propellant exhaust velocity (Vexhaust or, equivalently, Isp) are moving at a large
fraction of the speed of light, such that relativistic corrections must be included.
These derivations will follow Forward’s method [18]. The next two versions of
the Rocket Equation parallel the Classical and Relativistic Rocket Equations,
but for the situation where some large fraction of the initial propellant mass or
mass energy is unavailable (e.g., 60% for a proton antiproton annihilation
rocket) for momentum transfer to the vehicle.

A1. Classical Rocket Equation

In the Classical Rocket Equation derivation, a rocket with initial mass M ejects
a mass of reaction mass (propellant) dm at a constant exhaust velocity (i.e.,
Vexhaust corresponding to Isp) relative to the rocket with a velocity, as shown in
Fig. A1. (To avoid confusion with the vehicle’s velocity “V,” Vexhaust will be rep-
resented by “w” [18].) In the center of mass of the system, the resultant rocket
velocity is V (i.e., DV), and the reaction mass velocity is u. Note that we have
shown the situation for both acceleration and deceleration; derivation of the
Rocket Equation will be shown in parallel for both options.
To begin the derivation of the Classical Rocket Equation, we first start with the
conservation of mass:

Acceleration: dM ¼ dm Deceleration: dM ¼ dm

(The minus sign indicates the decreasing mass of the rocket. Also, the two
equations are the same for acceleration or deceleration.)
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 109

Fig. A1 Classical rocket mass, velocity, and momentum (with no loss of propellant).

Next, we conserve momentum:

Acceleration: d(M  V) ¼ u  dm Deceleration: d(M  V) ¼ u  dm

Note the sign difference between the acceleration and deceleration forms of the
equation. Finally, we use the classical addition (or subtraction) of velocities:

Acceleration: u ¼ w  V Deceleration: u ¼ w þ V

Again, note the sign differences. We then use the conservation of mass
equation to find dm; this identity will be then be used when we expand the con-
servation of momentum equation (i.e., d[MV] MdV þ VdM). Thus, expanding
the derivatives, combining the above equations, and rearranging gives:

Acceleration: Deceleration:
MdV þ VdM ¼ udm MdV þ VdM ¼ udm
¼ (w  V)dM ¼ þ(w þ V)dM
VdM þ (w  V)dM ¼ MdV VdM  (w þ V)dM ¼ MdV
(V þ w  V)dM ¼ MdV (V  w  V)dM ¼ MdV
dM=M ¼ dV=w dM=M ¼ þdV=w

If we were performing a numerical integration, we could write:

Acceleration: dM ¼ M½dV=w Deceleration: dM ¼ M½þdV=w

Analytical integration gives:

Acceleration: ‘n(M) ¼ V=w Deceleration: ‘n(M) ¼ þV=w

which, when evaluated for the rocket mass limits of Mwet (initial wet mass)
and Mdry (final “burnout” dry mass), initial and final velocities (Vi and Vf, with
110 R. H. FRISBEE

Vf Vi DV ), and w Vexhaust gc Isp [with gc 9.8 (m/s)/(lbf-s/lbm)] yields

the Classical Rocket Equation:

Acceleration: Deceleration:
‘n(Mwet =Mdry ) ¼ þDV=(gc Isp ) ‘n(Mwet =Mdry ) ¼ DV=(gc Isp )
Mwet =Mdry ¼ exp½þDV=(gc Isp ) Mwet =Mdry ¼ exp½DV=(gc Isp )

Note that for acceleration, DV is positive (DV . 0) since Vf . Vi. However, for
deceleration, DV , 0 since Vf , Vi. Thus, both equations give the expected result
that Mwet . Mdry independent of whether the vehicle is accelerating or
decelerating.
For convenience, we will adopt a sign convention for acceleration and decel-
eration using expressions like x “þ/ ” y or x “ /þ” y, where the “upper” sign is
for acceleration and the “lower” sign is for deceleration. Thus, our derivation can
be written as

Mass conservation: dM ¼ dm

Momentum conservation: d(M  V) ¼ þ= u  dm
Velocity addition: u ¼ w =þ V

As before, we first expand the conservation of mass equation and solve for dm.
Next, we expand the conservation of momentum equation and substitute in values
for dm and u found previously:

MdV þ VdM ¼ þ= udm ¼ =þ (w =þ V)dM

VdM þ= (w =þ u)dM ¼ MdV
(V þ=  w  V)dM ¼ MdV

This is rearranged to give forms suitable for analytical or numerical integration:

For analytical integration: dM=M ¼ =þ dV=w

For numerical integration: dM ¼ dM(=þ dV=w)

Analytical integration gives:

Analytical integration: ‘n(M) ¼ =þ V=w

Integration limits: ‘n(Mwet =Mdry ) ¼ þ= DV=(gc Isp )
Mwet =Mdry ¼ exp [ þ= DV=(gc Isp )

A2. Relativistic Rocket Equation

For the Relativistic Rocket Equation, we follow essentially the same steps as
for the Classical Rocket Equation, with the difference being the use of relativistic
equations for mass and velocity, and conservation of mass energy content of the
system (as opposed to conservation of mass only in the classical case). For
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 111

example, we use relativistic mass (M and m), such that:

M ¼ Mr =(1  V 2 =c2 )1=2 and m ¼ mr =(1  u2 =c2 )1=2

where Mr and mr are the rest masses of the vehicle and propellant, respectively.
First, we start with the conservation of mass energy; i.e., mass energy, or
mc 2, is conserved. (Note again the use of the “þ/ ” and “ / þ ” sign conven-
tion for acceleration and deceleration.)

d½Mr =(1  V 2 =c2 )1=2   c2 ¼ d(mr )=(1  u2 =c2 )1=2  c2

Note also that the mass energy conservation equation is the same for accelera-
tion or deceleration.
Next, we conserve momentum§§:

d½Mr =(1  V 2 =c2 )1=2  V ¼ þ=  d(mr )=(1  u2 =c2 )1=2  u

Finally, we use the relativistic addition (or subtraction) of velocities:

u ¼ (w  = þ V)=(1  = þ wV=c2 )

To begin the derivation, we expand the mass energy conservation derivative

and solve for dmr:

Mr d½(1  V 2 =c2 ) 1=2

  c2 þ dMr  (1  V 2 =c2 ) 1=2
 c2
¼ dmr  (1  u2 =c2 ) 1=2
 c2

Mr  (1  V 2 =c2 ) 3=2
V=c2 dV  c2 þ dMr  (1  V 2 =c2 ) 1=2
 c2
¼ dmr  (1  u2 =c2 ) 1=2
 c2

dmr ¼ ½Mr  (1  V 2 =c2 ) 3=2

V=c2 dV þ dMr  (1  V 2 =c2 ) 1=2
=(1  u2 =c2 ) 1=2

(Again, note that the mass energy conservation equation is the same for accel-
eration or deceleration.)

§§
We should note that the equations given above are the correct forms of the mass energy and
momentum conservation equations, respectively, as defined by Forward [18]. In our previous publi
cations [3,14], we had incorrectly included the propellant center of mass relativistic velocity correc
tion (1 2 u 2/c 2)1/2, within the d(mr) term on the right hand side of the equations; e.g., d[mr/(1 2 u 2/
c 2)1/2]. Fortunately, we had used an approximation to produce a form of the differential equation that
could be analytically integrated, so that the net result for the equation for the analytical integration was
the same as that presented.
112 R. H. FRISBEE

Next, we expand the conservation of momentum derivative, and substitute values

for dmr and u from above:

Mr d½V  (1  V 2 =c2 ) 1=2

 þ dMr  V  (1  V 2 =c2 ) 1=2

¼ þ=  (dmr )  (1  u2 =c2 ) 1=2

 (u)
Mr ½dV  (1  V 2 =c2 ) 1=2
þ V  (1  V 2 =c2 ) 3=2
(V=c2 )dV
þ dMr  V  (1  V 2 =c2 ) 1=2

¼ = þ ½Mr  (1  V 2 =c2 ) 3=2

V=c2 dV þ dMr  (1  V 2 =c2 ) 1=2

2
 ½(w  = þ V)=(1  = þ wV=c )

Multiplying through by (1 V 2/c 2)þ1/2 gives:

Mr ½dV þ (V 2 =c2 )  (1  V 2 =c2 ) 1 dV þ dMr  V

¼ = þ ½Mr  (1  V 2 =c2 ) 1 V=c2 dV þ dMr 
 ½(w  = þ V)=(1  = þ wV=c2 )
Rearranging gives:

dMr  ½V þ =  (w  = þ V)=(1  = þ wV=c2 )

¼ dVMr ½1 þ (V 2 =c2 )  (1  V 2 =c2 ) 1
þ =  V=c2  (1  V 2 =c2 ) 1

 (w  = þ V)=(1  = þ wV=c2 )

dMr =Mr
¼ dV½1 þ (V 2 =c2 )  (1  V 2 =c2 ) 1
þ =  V=c2  (1  V 2 =c2 ) 1

 (w  = þ V)=(1  = þ wV=c2 )=½V þ =  (w  = þ V)=(1  = þ wV=c2 )

¼ dV=(1  V 2 =c2 )  [1  V 2 =c2 þ (V 2 =c2 ) þ =  V=c2  (w  = þ V)
=(1  = þ wV=c2 )]=½V þ =  (w  = þ V)=(1  = þ wV=c2 )
¼ dV=(1  V 2 =c2 )  [1 þ =  V=c2  (w  = þ V)=(1  = þ wV=c2 )]=
½V þ =  (w  = þ V)=(1  = þ wV=c2 )

Multiplying the numerator and denominator of the right-hand term by

(1 /þwV/c2) gives:

dMr =Mr ¼ dV=(1  V 2 =c2 )  ½1  = þ wV=c2 þ =  V=c2

 (w  = þ V)=½V(1  = þ wV=c2 ) þ =  (w  = þ V)
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 113

Expanding and rearranging terms gives forms suitable for analytical or numerical
integration:

dMr =Mr ¼ dV=(1  V 2 =c2 )  ½1  = þ wV=c2 þ =  wV=c2

 V 2 =c2 =½V  = þ wV 2 =c2 þ =  w  V)
¼ dV=(1  V 2 =c2 )  ½1  V 2 =c2 =½þ=  w(1  V 2 =c2 )

For analytical integration: dMr =Mr ¼ dV=½þ=  w(1  V 2 =c2 )

¼ ½c2 =( þ =  w)  ½dV=(c2  V 2 )

For numerical integration: dMr ¼ Mr {dV=½þ=  w(1  V 2 =c2 )}

dMr ¼ Mr {½c2 =( þ =  w)  ½dV=(c2  V 2 )}

Analytical integration gives:

Analytical integration: ‘n(Mr ) ¼ c2 =(þ=w)½1=(2c)  ‘n½(c þ V)=(c  V)

¼ c=(þ=2w)  ‘n½(c þ V)=(c  V)
¼ 1=(þ=2w=c)  ‘n½(1 þ V=c)=(1  V=c)

Evaluating the integrals for the initial and final rest-mass limits of Mwet
and Mdry, initial and final velocities of Vi and Vf, with Vf Vi DV, and w
Vexhaust gc Isp, yields the Relativistic Rocket Equation:
Integration limits:
‘nðMwet =Mdry Þ
¼ c=(þ=2w)  ‘n{½(c þ Vf )=(c  Vf )=½(c þ Vi )=(c  Vi )}
¼ 1=(þ=2w=c)  ‘n{½(1 þ Vf =c)(1  Vi =c)=½(1 þ Vi =c)(1  Vf =c)}
¼ 1=(þ=2gc Isp =c)  ‘n{½(1 þ Vf =c)(1  Vi =c)=½(1 þ Vi =c)(1  Vf =c)}
 Mwet =Mdry
2gc Isp =c)
¼ f½(1 þ Vf =c)(1  Vi =c)=½(1 þ Vi =c)(1  Vf =c)g½1=(þ=

Note that we have dropped the subscript “r” denoting rest mass for
convenience.
For the case where the initial (or final) velocity is zero before acceleration (or
after deceleration), we have:

‘n(Mwet =Mdry ) ¼ 1=(þ= 2gc Isp =c)  ‘n½(1 þ= V=c)=(1 =þ V=c)

where V is the final (or initial) peak velocity after acceleration from V 0
(or before deceleration to V 0).
114 R. H. FRISBEE

A3. Classical Rocket Equation with “Loss” of Propellant

In this section, we perform a derivation of the Classical Rocket Equation with
“loss” of propellant where we only need to conserve mass and momentum; the
corresponding case for the Relativistic Rocket Equation is complicated by the
need to conserve relativistic mass energy content as well as relativistic momen-
tum. Thus, in the classical case, we assume that a fraction “a” of the initial pro-
pellant mass is used to produce forward motion of the rocket, as shown in
Fig. A2. In effect, a fraction (1 a) of the initial propellant mass (and momen-
tum) is “lost,” at least insofar as production of forward momentum is concerned.
It is as if a quantity (a)Mp is ejected parallel to the direction of motion (to produce
some DV) and a quantity (1 a)Mp is ejected perpendicular in such a way that
(1 a)Mp does not contribute to the forward DV. Finally, note that the usual situ-
ation for rocket propulsion is for a 1; for example, even in a fission or fusion
rocket, where some of the rest mass of the nuclear fuel is turned into energy, the
amount of mass “lost” (to conversion into energy) is negligible, so the “standard”
Classical or Relativistic Rocket Equations (with a 1) can be used.
For the Classical Rocket Equation with “loss” of propellant, we follow essen-
tially the same steps as for the standard (a 1) Classical Rocket Equation, with
the difference being the use of terms in (a) and (1 a) in the momentum con-
servation equation. Thus, we again write the equations for conservation of
mass and momentum, and addition (or subtraction) of velocities. (Note again
the use of the “þ/ ” and “ /þ” sign convention for acceleration and
deceleration.)

Mass conservation: dM ¼ dm ¼ ½(a)dm(usable) þ (1  a)dm(lost)

Note that a * dm represents the “usable” fraction of the (incremental) propel-

lant mass, whereas (1 a) dm represents the fraction “lost” and unavailable for
production of velocity (V).

Fig. A2 Classical rocket mass, velocity, and momentum (with “loss” of propellant).
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 115

Momentum conservation: d(M  V) ¼ þ=  u  a dm

(only a dm contributes to V)
 V(1  a) dm (momentum ‘‘lost”)

As above, only u * a * dm contributes momentum toward changing V. Also, in

defining w and u, we are only interested in the “bulk” or center-of-mass velocity
of the (incremental) propellant mass quantity a * dm moving parallel to V. Thus,
the exhaust velocity (w Vexhaust) and Isp of the rocket engine corresponds to an
“effective” Isp or Vexhaust that takes into account any inefficiencies in the nozzle in
converting the random velocities of the propellant exhaust gas into directed
motion (w and u) parallel to V.
Similarly, the (incremental) propellant mass (1 a) * dm is ejected at some
exhaust velocity (not necessarily w) perpendicular to V (and u) in a symmetric
fashion that results in a net cancellation of the perpendicular velocity of the
center-of-mass of (1 a) * dm. However, this unit of mass is still moving
forward at the rocket’s velocity V, so the total amount of “lost” momentum is
V * (1 a) * dm. Interestingly, this momentum is totally lost from the system
(its value is negative for either acceleration or deceleration), so it also reduces
the mass and momentum of the rocket that must be accelerated or decelerated
in subsequent differential time steps. However, it also represents a “dead”
weight that must be initially loaded into the rocket, which increases Mwet over
that seen when a 1.

Velocity addition: u¼w=þV

As above, we first expand the conservation of mass equation and solve for dm.
Next, we expand the conservation of momentum equation and substitute in values
for dm and u found above:

MdV þ VdM ¼ þ=  ua dm  V(1  a)dm

¼ = þ a (w  = þ V)dM þ V(1  a)dM
VdM þ =  a (w  = þ V)dM  V(1  a)dM ¼ MdV
½V þ =  a w  a V  V(1  a)dM ¼ MdV
(þ =  a w)dM ¼ MdV

This is rearranged to give forms suitable for analytical or numerical integration:

For analytical integration: dM=M ¼ = þ dV=(a w)

For numerical integration: dM ¼ M½= þ dV=(a w)
Analytical integration: ‘n(M) ¼ = þ V=(a w)
Integration limits: ‘n(Mwet =Mdry ) ¼ þ=  DV=(a gc Isp )
Mwet =Mdry ¼ exp½þ=  DV=(a gc Isp )
116 R. H. FRISBEE

Note that the final form of the equation is similar to that of the standard (a 1)
Classical Rocket Equation. The result has been to add a to the Isp term, resulting
in an “effective” Isp that takes into account the “loss” of propellant. In fact, this
result could be anticipated from the classical definition of Isp in terms of thrust (F)
and propellant mass flow rate (ṁ) entering the rocket engine:

_ ¼ Vexhaust
gc Isp ¼ F=m

In this equation, gc 9.8 (m/s)/(lbf-s/lbm), where Isp lbf-s/lbm, F

Newtons, ṁ kg/s, and Vexhaust m/s. In the context used here, ṁ is the
total mass of propellant entering the rocket engine. However, only a quantity
a * ṁ is available to produce thrust, so we would have:
_ (Total)
gc Isp ¼ F=½a m

which rearranges to give the same (a gc Isp) term seen in the Rocket Equation
above:

_ (Total)
a gc Isp ¼ F=m

A4. Relativistic Rocket Equation with “Loss” of Propellant

For this derivation, we will combine elements of the derivations of the Rela-
tivistic Rocket Equation and Classical Rocket Equation with “loss” of propellant.
However, unlike the classical version of this equation, where we conserve only
(rest) mass and classical momentum (rest mass multiplied by velocity), the rela-
tivistic version requires that we conserve the total amount of mass energy and
relativistic momentum, including that of the “lost” propellant. Thus, because
of the need to include the relativistic mass energy and momentum of the
“lost” propellant, the algebra will be more complex, resulting in an equation
that must be numerically integrated.
Also, for a matter antimatter annihilation reaction, we need to bookkeep the
rest mass and kinetic energy of the various reactants (protons, antiprotons, elec-
trons, positrons) and annihilation products (charged and neutral pions, gamma
ray photons). This is needed to determine both the effective value of “a” in the
equation (i.e., how much of the mass energy and momentum are usable for pro-
pulsion), as well as to determine the mass energy and momentum of the various
annihilation products. Table A1 summarizes mass and energy quantities based on
the assumption of a reaction between an atom of H [containing a proton (Pþ) and
electron (e )] and an anti-atom of anti-hydrogen [containing an antiproton (P )
and positron (eþ)]. The protons and antiprotons annihilate to give neutral and
charged pions (po, pþ, p ); the electrons and positrons annihilate to give high-
energy photons (gamma rays, g). The initial annihilation reaction is given as [19]:

Pþ þ P ! 2:0p8 þ 1:5pþ þ 1:5p and e þ eþ ! 2g

The neural pions promptly decay into very high-energy gamma rays (each about
355 times more energetic than an electron/positron gamma). The charged pions
 Table A1 Rest mass, kinetic energy, and total mass – energy content of matter (H atom) 1 antimatter (anti-H atom) annihilation

Fraction of total Kinetic energy Fraction of total Effective Total mass – Fraction of total
Species Rest mass (MeV) (%) (MeV) (%) velocity (c) energy (MeV) (%)
Initial reactants 1877.6 100.00 0.0 0.00 1877.6 100.00
Pþ 938.3 49.97 0 0.00 0.0000 938.3 49.97
e2 0.5 0.03 0 0.00 0.0000 0.5 0.03
P2 938.3 49.97 0 0.00 0.0000 938.3 49.97
eþ 0.5 0.03 0 0.00 0.0000 0.5 0.03
Initial products 688.7 36.68 1188.7 63.32 1877.4 100.00
2.0 po 269.9 14.38 439.1 23.39 0.9247 709.0 37.76
1.5 pþ 209.4 11.15 374.3 19.94 0.9334 583.7 31.09
1.5 p – 209.4 11.15 374.3 19.94 0.9334 583.7 31.09
e2 þ eþ ! 2g 1.0 0.05 (1.0000) 1.0 0.05
Decay products 317.0 16.89 1559.7 83.11 1876.7 100.00
2.0 po ! 4g 709.1 37.78 (1.0000) 709.1 37.78
1.5 pþ ! 1.5 mþ 158.5 8.45 288.5 15.37 0.9350 447.0 23.82
1.5 pþ ! 1.5 nm 136.8 7.29 (1.0000) 136.8 7.29
1.5 p2 ! 1.5 m2 158.5 8.45 288.5 15.37 0.9350 447.0 23.82
1.5 p2 ! 1.5 anti-nm 136.8 7.29 (1.0000) 136.8 7.29
LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY
117
118 R. H. FRISBEE

decay into charged muons (m) and neutrinos (n) after traveling about 21 m at 0.93 c
[18,19].

2:0p8 ! 4g, 1:5pþ ! 1:5mþ þ 1:5nm , 1:5p ! 1:5m þ 1:5 anti-nm

To determine the value for a, we first determine the fractions (percentages) of

the total initial mass energy contained in the various annihilation species. For the
charged pions (p+), we can further split the total mass energy content into rest
mass (mr) and kinetic energy, such that:

Rest massenergy ¼ mr c2
Kinetic energy ¼ mr c2 ½1=(1  v2 =c2 )1=2  1
Total massenergy ¼ mr c2 =(1  v2 =c2 )1=2

As shown in Table A1, this results in the charged pions having a rest mass that is
22.3% of the total mass energy content of the reactants (H and anti-H atoms), and
a kinetic energy that is about 39.9% of the initial mass energy. However, even
though the total mass energy content of the charged pions is about 62% of the
total mass energy content of the reactants (H and anti-H atoms), this is not the
value of a. This is because the electromagnet used to deflect the charged pions
is not perfectly reflective (some charged pions “leak” through the magnetic field
and travel upstream), and for those that are reflected, the resultant beam is not per-
fectly collimated. Thus, there is an inherent kinetic energy inefficiency in the
nozzle that must also be accounted for. Callas [20], using Monte Carlo simulations
of particle trajectories in proton/antiproton annihilations, estimated that the mag-
netic nozzle would be at best 50% kinetic energy efficient, corresponding to a
value of a 42.2%. We have arbitrarily assumed a value of a 40% of the
initial total mass energy content of the propellant, corresponding to a 44.37%
nozzle kinetic energy efficiency. These values are summarized in Table A2.
Note that the magnetic nozzle does not affect the fraction of rest mass of the
charged pions; that remains the same. Instead, what we have done is take into
account the nozzle inefficiency in converting the kinetic energy of the charged
pions, with 0.9334 c velocity vectors pointed in random directions, into a directed,
collimated stream of charged pions with a 0.6729 c kinetic energy velocity vector
pointed directly out the back of the rocket.
Finally, the discussion above has concentrated on the efficiency of the annihil-
ation reaction and electromagnet nozzle in capturing the total mass energy (rest
mass and kinetic energy) of the charged pions. However, there are additional
losses in the magnetic nozzle that result in a reduction in overall Isp. For
example, although the magnetic nozzle can capture 40% of the reactant’s
initial mass energy for thrust production, the “beam” of charged pions leaving
the engine are not perfectly collimated (i.e., parallel to V). Using the Monte
Carlo particle trajectories, Callas [20] used direct momentum calculations
to determine an “effective” Isp or Vexhaust of 1  108 N-s/kg m/s 0.333
c 1.02  107 lbf-s/lbm.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 119

Table A2 Hydrogen/anti-hydrogen annihilation mass energy breakdown

to determine “a”

Species % of Total Quantity

o
2.0 p ! 4 g 37.76 Total mass energy
e þ eþ ! 2 g 0.05 Total mass energy
1.5 p+ 22.31 Total rest mass
1.5 p+ 39.87 Total kinetic energy
1.5 p+ 62.18 Total mass energy
1.5 p+ 22.31 Usable rest mass (100.00% of charged pions)
1.5 p+ 17.69 Usable kinetic energy (44.37% of charged pions)
1.5 p+ 40.00 Total usable mass energy
1.5 p+ 0.00 Unusable rest mass (0.00% of charged pions)
1.5 p+ 22.18 Unusable kinetic energy (55.63% of charged pions)
1.5 p+ 22.18 Total Unusable mass energy
1.5 p+ 40.00 Net Usable mass energy (a)
po, p+, g 60.00 Net “Lost” mass energy (1 a)

To begin our derivation, we again begin with conservation of mass energy

(and again use the “þ/ ” and “ /þ” nomenclature for acceleration and
deceleration):
Mass conservation:

d½Mr =(1  V 2 =c2 )1=2   c2 ¼ a d(mr )=(1  u2 =c2 )1=2  c2

 (1  a)d(mr )=(1  V 2 =c2 )1=2  c2

Note that unlike the Classical Rocket Equation with “loss” of propellant, where
the (incremental) propellant mass dm was independent of velocity, we now have
to explicitly include the velocities (u and V) because of the relativistic mass
increases. A similar effect is seen for the conservation of momentum:
Momentum conservation:

d½Mr =(1  V 2 =c2 )1=2  V ¼ þ= a d(mr )=(1  u2 =c2 )1=2  u

 (1  a)d(mr )=(1  V 2 =c2 )1=2  V

As with the classical version, the (1 a) * dmr term is always negative inde-
pendent of whether the rocket is accelerating or decelerating. Finally, we again
have the relativistic addition (or subtraction) of velocities:
Velocity addition:

u ¼ (w  = þ V)=(1  = þ wV=c2 )
120 R. H. FRISBEE

To begin the derivation, we expand the mass energy conservation derivative

and solve for dmr:

Mr d½(1  V 2 =c2 ) 1=2

  c2 þ dMr  (1  V 2 =c2 ) 1=2
 c2
¼ a d(mr )(1  u2 =c2 ) 1=2
 c2
 (1  a)d(mr )(1  V 2 =c2 ) 1=2
 c2
Mr  (1  V 2 =c2 ) 3=2
V=c2 dV  c2 þ dMr  (1  V 2 =c2 ) 1=2
 c2
¼ d(mr )½a (1  u2 =c2 ) 1=2

þ (1  a)(1  V 2 =c2 ) 1=2

  c2
dmr ¼ ½Mr  (1  V 2 =c2 ) 3=2
V=c2 dV þ dMr
 (1  V 2 =c2 ) 1=2
=½a (1  u2 =c2 ) 1=2

þ (1  a)(1  V 2 =c2 ) 1=2



(Again, note that the mass energy conservation equation is the same for accel-
eration or deceleration.)
As before, we expand the conservation of momentum derivative, and substi-
tute values for dmr and u from above:

Mr d½V  (1  V 2 =c2 ) 1=2

 þ dMr  V  (1  V 2 =c2 ) 1=2

¼ þ=  a d(mr )(1  u2 =c2 ) 1=2

 (u)
2 2 1=2
 (1  a) d(mr )(1  V =c )  (V)
¼ dmr ½þ=  a (1  u2 =c2 ) 1=2
 (u)
 (1  a)(1  V 2 =c2 ) 1=2
 (V)
Mr ½dV  (1  V 2 =c2 ) 1=2
þV  (1  V 2 =c2 ) 3=2
(V=c2 )dV
þ dMr  V  (1  V 2 =c2 ) 1=2

¼ ½Mr  (1  V 2 =c2 ) 3=2

V=c2 dV
þ dMr  (1  V 2 =c2 ) 1=2
=
½a(1  u2 =c2 ) 1=2
þ (1  a)(1  V 2 =c2 ) 1=2

 ½þ=  a(1  u2 =c2 ) 1=2
 (u)
 (1  a)(1  V 2 =c2 ) 1=2
 (V)
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 121

¼ ½Mr  (1  V 2 =c2 ) 3=2

V=c2 dV
þ dMr  (1  V 2 =c2 ) 1=2
  (þ =  u)  X

where the various terms containing a are combined into the quantity X:

X ¼ ½a(1  u2 =c2 ) 1=2

 (1  a)=(þ =  u)  (1  V 2 =c2 ) 1=2
 V=
½a(1  u2 =c2 ) 1=2
þ (1  a)(1  V 2 =c2 ) 1=2


Note that when a 1, X 1. Also, this substitution now allows us to continue in

a manner analogous to the derivation of the standard (a 1) Relativistic Rocket
Equation. For example, separating terms in dMr and dV on opposite sides of the
equation gives:

dMr ½V  (1  V 2=c2 ) 1=2

þ (1  V 2=c2 ) 1=2
 (þ =  u  X)
¼ Mr dV½(1  V 2 =c2 ) 3=2
V=c2  (þ =  u  X) þ (1  V 2=c2 ) 1=2

þ V 2 =c2  (1  V 2=c2 ) 3=2



We than multiply through by (1 V 2/c 2)þ1/2 to get

dMr ½V þ =  u  X
¼ Mr dV½(1  V 2=c2 ) 1 V=c2  (þ =  u  X) þ (1) þ V 2=c2  (1  V 2=c2 ) 1 
¼ Mr dV=(1  V 2=c2 )  ½V=c2  (þ =  u  X) þ (1  V 2=c2 ) þ V 2=c2 
¼ Mr dV=(1  V 2=c2 ) ½V=c2  (þ =  u  X) þ 1
dMr =Mr ¼ dV=(1  V 2=c2 )  ½V=c2  (þ =  u  X) þ 1=½V þ =  u  X

Substituting for u (w /þ V) / (1 /þ wV/c 2) gives

dMr =Mr ¼ dV=(1  V 2=c2 )  ½V=c2  (þ=  1)(w  = þ V)=

(1  = þ wV=c2 )  X þ 1]=½V þ (þ =  1)(w  = þ V)=
(1  = þ wV=c2 )  X
¼ dV=(1  V 2 =c2 )  f1 þ ½wV=c2  (þ=  X)  = þ V 2 =c2
 (þ=  X)=(1  = þ wV=c2 )g=
{V þ ½w  (þ=  X)  = þ V  (þ=  X)=(1  = þ wV=c2 )}
122 R. H. FRISBEE

¼ dV=(1  V 2 =c2 )  ½(1  = þ wV=c2 ) þ wV=c2

 (þ=  X)  = þ V 2 =c2  (þ=  X)=
½V(1  = þ wV=c2 ) þ w  (þ=  X)  = þ V  (þ=  X)
¼ dV=(1  V 2 =c2 )  ½1  = þ wV=c2 þ =  wV=c2  (X)
 V 2 =c2  (X)=½V  = þ wV 2 =c2 þ =  w  (X)  V  (X)
¼ dV=(1  V 2 =c2 )  ½(1  V 2 =c2  X)  = þ wV=c2 (1  X)=
½V  (1  X) þ =  w  (X  V 2 =c2 )

Note that when a 1, X 1, and this equation reduces to the form seen for
the standard (a 1) Relativistic Rocket Equation:

dMr =Mr ¼ dV=(1  V 2 =c2 )=½þ=  w ¼ dV=½þ=  w(1  V 2 =c2 )

Unfortunately, when a or X are not 1, the various terms in V do not cancel out,
leaving us with an equation that must be numerically integrated. For this, we use
the equation in the form:
For numerical integration:

dMr ¼ Mr  {dV=(1  V 2=c2 )  ½(1  V 2 =c2  X)  = þ wV=c2 (1  X)=

½V  (1  X) þ =  w  (X  V 2 =c2 )}

where X is again defined as

X ¼ ½a(1  u2 =c2 ) 1=2

 (1  a)=( þ =  u)  (1  V 2 =c2 ) 1=2
 V=
½a(1  u2=c2 ) 1=2
þ (1  a)(1  V 2 =c2 ) 1=2


and again u (w /þ V) / (1 /þ wV/c 2)

To begin the numerical integration, we initialize the calculations with a
vehicle initial wet mass (Mwet), which can be arbitrarily set equal to 1, and an
initial velocity (Vi). We then select a (constant) value for dV and use the above
equations to determine dMr for each calculation step. During each step, we use
the values of Mr and V from the previous step to calculate dMr and “new”
values of Mr (previous Mr dMr) and V (previous V þ/ dV), with the
þ/ reflecting either acceleration or deceleration. We continue the process
until we reach the final velocity (Vf) and corresponding vehicle final dry mass
(Mdry) to obtain the Relativistic Rocket Equation mass ratio Mwet/Mdry for a
given DV Vf Vi.
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 123

References
[1] Williams, S., “Trends,” Asimov’s Science Fiction, Oct./Nov., 2007, pp. 4, 6.
[2] Frisbee, R. H., “Beamed Momentum LightSails for Interstellar Missions: Mission
Applications and Technology Requirements,” AIAA Paper 2004 3567, 11 14 July
2004.
[3] Frisbee, R. H., and Leifer, S. D., “Evaluation of Propulsion Options for Interstellar
Missions,” AIAA Paper 98 3403, 13 15 July 1998.
[4] Anderson, J. L., “Leaps of the Imagination: Interstellar Flight and the Horizon Mission
Methodology,” Journal of the British Interplanetary Society, Vol. 49, 1996, pp. 15 20.
[5] Wilcox, R. M., “Internet STELLAR DATABASE,” URL: http://www.stellar
database.com [cited 12 Dec. 2007].
[6] Mewaldt, R., “Interstellar Mission Science Objectives,” Presented at the NASA
Office of Space Access and Technology (OSAT) Ninth Advanced Space Propulsion
Workshop, Pasadena CA, 11 13 March 1998; JPL Internal Document D 15671,
R. H. Frisbee (ed.), 1998.
[7] Martin, A. R. (ed.), “Project Daedalus The Final Report of the BIS Starship Study,”
Journal of the British Interplanetary Society, suppl., 1978.
[8] Jones, R. M., “Electromagnetically Launched Micro Spacecraft for Space Science
Missions,” AIAA Paper 88 0068, 11 Jan. 1988.
[9] Schnitzler, B. G., Jones, J. L., and Chapline, G. F., “Fission Fragment Rocket
Preliminary Feasibility Assessment,” Idaho National Engineering Lab. Contract
No. DEACO7 76IDO1570 and Lawrence Livermore National Lab. Contract No.
W 7405 ENG 88, 1989.
[10] Forward, R. L, “Radioisotope Sails for Deep Space Propulsion and Electrical
Power,” AIAA Paper 95 2596, 10 12 July 1995.
[11] Bussard, R. W., “Galactic Matter and Interstellar Flight,” Astronautica Acta, Vol. 6,
No. 4, 1960, pp. 179 194.
[12] Forward, R. L., “Starwisp: An Ultra Light Interstellar Probe,” Journal of Spacecraft
and Rockets, Vol. 22, No. 3, 1985, pp. 345 350.
[13] Forward, R. L., “Roundtrip Interstellar Travel Using Laser Pushed Lightsails,”
Journal of Spacecraft and Rockets, Vol. 21, No. 2, 1984, pp. 187 195.
[14] Frisbee, R. H., “How to Build an Antimatter Rocket for Interstellar Missions
Systems Level Considerations in Designing Advanced Propulsion Technology
Vehicles,” AIAA Paper 2003 4696, 20 23 July 2003.
[15] Singer, C. E., “Interstellar Propulsion Using a Pellet Stream for Momentum Trans
fer,” Journal of the British Interplanetary Society, Vol. 33, 1980, pp. 107 115;
Zubrin, R. M., and Andrews, D. G., “Magnetic Sails and Interplanetary Travel,”
Journal of Spacecraft and Rockets, Vol. 28, No. 2, 1991, pp. 197 203.
[16] Smith, G. A., “Applications of Trapped Antiprotons,” Hyperfine Interactions, Vol.
81, No. 1 4, August 1993, pp. 189 196; Lewis, R. A., Smith, G. A., and Howe,
S. D., “Antiproton Portable Traps and Medical Applications,” Hyperfine Inter
actions, Vol. 109, No. 1 4, Aug. 1997, pp. 155 164.
[17] URANOS Group, “Konstantin E. Tsiolkovsky [Ciolkowski] (1857 1935),” URL:
http://www.uranos.eu.org/biogr/ciolke.html [last update Aug. 6, 2001, cited
January 22, 2003].
[18] Forward, R. L., “Ad Astra!,” Journal of the British Interplanetary Society, Vol. 49,
1996, pp. 147 149.
124 R. H. FRISBEE

[19] LaPointe, M. R., “Antiproton Powered Propulsion with Magnetically Confined

Plasma Engines,” Journal of Propulsion and Power, Vol. 7, No. 5, Sept. Oct.
1991, pp. 749 759.
[20] Callas, J. L., The Application of Monte Carlo Modeling to Matter Antimatter
Annihilation Propulsion Concepts, Jet Propulsion Laboratory, JPL Internal Docu
ment D 6830, CA, 1 Oct. 1989.
[21] Meyer, T. R., McKay, C. P., McKenna, P. M., and Pyror, W. R, “Rapid Delivery of
Small Payloads to Mars,” AAS Paper 84 172, July 1984, Proceedings of the Case for
Mars II, pp. 419 431, C. P. McKay (ed.), Vol. 62, Science and Technology Series,
American Astronomical Society, 1985; Landis, G. A., “Optics and Materials Con
siderations for a Laser Propelled Lightsail,” IAA Paper 89 684, 7 12 Oct. 1989;
Landis, G. A., “Small Laser Propelled Interstellar Probe,” Journal of the British
Interplanetary Society, Vol. 50, 1997, pp. 149 154.
[22] Frisbee, R. H., “Impact of Interstellar Vehicle Acceleration and Cruise Velocity on
Total Mission Mass and Trip Time,” AIAA Paper 2006 5224, 9 12 July 2006.
[23] Draine, B. T., and Lee, H. M., “Optical Properties of Interstellar Graphite and Silicate
Grains,” Astrophysical Journal, Vol. 285, 1 Oct. 1984, pp. 89 108.
[24] Frisbee, R. H., “Advanced Space Propulsion for the 21st Century,” Journal of Pro
pulsion and Power, Vol. 19, 2003, pp. 1129 1154.
[25] Dyson, F. J., “Interstellar Transport,” Physics Today, Oct., 1968, pp. 41 45.
[26] Brower, K., The Starship and the Canoe, Harper and Row, New York, 1978, p. 149.
[27] Thio, F., “A Summary of the NASA Fusion Propulsion Workshop 2000,” AIAA
Paper 2001 3669, 8 11 July 2001.
[28] Forward, R. L., “Antiproton Annihilation Propulsion,” AIAA Paper 84 1482, 11 13
June 1984; Air Force Rocket Propulsion Laboratory (AFRPL) Technical Report
AFRPL TR 85 034, Sept. 1985.
[29] Narevicius, E., Parthey, C. G., Libson, A., Narevicius, J., Chavez, I., Even, U., and
Raizen, M. G., “An Atomic Coilgun: Using Pulsed Magnetic Fields to Slow a Super
sonic Beam,” New Journal of Physics, Vol. 9, No. 10, Oct. 2007, pp. 358 366;
Narevicius, E., Parthey, C. G., Libson, A., Riedel, M. F., Even, U., and Raizen,
M. G., “Towards Magnetic Slowing of Atoms and Molecules,” New Journal of
Physics, Vol. 9, No 4, April 2007, pp. 96 103.
[30] Takahashi, H., “Application of Muon Catalyzed Fusion, and an Alternative
Approach for Space Propulsion,” NASA Office of Space Access and Technology
(OSAT) Eighth Advanced Space Propulsion Workshop, Pasadena CA, 20 21 May
1997, Proceedings published as JPL Internal Document D 15461, R. H. Frisbee
(ed.), 1997.
[31] Lewis, R. A., Smith, G. A., Cardoff, E., Dundore, B., Fulmer, J., Watson, B. J., and
Chakrabati, S., “Antiproton Catalyzed Micro fission/Fusion Propulsion Systems for
Exploration of the Outer Solar System and Beyond,” AIAA Paper 96 3069, 1 3 July
1996.
[32] Lewis, R. A., Meyer, K., Smith, G. A., and Howe, S. D., “AIMStar: Antimatter
Initiated Microfusion for Precursor Interstellar Missions,” AIAA Paper 99 2700,
20 23 June 1999.
[33] Howe, S., and Jackson, G., “Antimatter Driven Sail for Deep Space Missions,” Phase
I Final Report, Nov. 2002, NASA Institute for Advanced Concepts (NIAC), URL:
http://www.niac.usra.edu [cited Aug. 20, 2003].
 LIMITS OF INTERSTELLAR FLIGHT TECHNOLOGY 125

[34] Garner, C., “Developments and Activities in Solar Sails,” AIAA Paper 2001 3234,
8 11 July 2001.
[35] Zubrin, R. M., and Andrews, D. G., “Magnetic Sails and Interplanetary Travel,”
Journal of Spacecraft and Rockets, Vol. 28, No. 2, 1991, pp. 197 203.
[36] Winglee, R., Slough, J., Ziemba, T., and Goodson, A., “Mini Magnetospheric Plasma
Propulsion (M2P2): High Speed Propulsion Sailing the Solar Wind,” STAIF
Paper CP504, Space Technology and Applications Forum 2000 (STAIF 2000),
M. S. El Genk (ed.), American Institute of Physics, 2000; Winglee, R. M., Slough,
J., Ziemba, T., and Goodson, A., “Mini Magnetospheric Plasma Propulsion:
Trapping the Energy of the Solar Wind for Spacecraft Propulsion,” Journal of
Geophysical Research, Vol. 105, 2000, pp. 21,067 21,077.
[37] Singer, C. E., “Interstellar Propulsion Using a Pellet Stream for Momentum
Transfer,” Journal of the British Interplanetary Society, Vol. 33, 1980, pp.
107 115.
[38] Ewig, R., and Andrews, D., “Microfission Powered Orion Rocket,” NASA JPL/
MSFC Thirteenth Annual Advanced Space Propulsion Workshop (ASPW 2002),
Pasadena CA, 4 6 June 2002; Andrews Space and Technology, “Mini MagOrion
(MMO),” URL: http://www.andrews space.com/en/corporate/MMO.html [last
update 2003, cited June 20, 2003].
[39] Jones, R. M., “Electromagnetically Launched Micro Spacecraft for Space Science
Missions,” AIAA Paper 88 0068, 11 Jan. 1988, pp. 691 695.
[40] Lipinski, R. J., Beard, S., Boyes, J., Cnare, E. C., Cowan, M., Duggin, B. W., Kaye,
R. J., Morgan, R. M., Outka, D., Potter, D., Widner, M. M., and Wong, C., “Space
Applications for Contactless Coilguns,” IEEE Transactions on Magnetics, Vol. 29,
No. 1, Jan. 1993.
[41] Sol Company, “Alpha Centauri 3,” URL: http://www.solstation.com/stars/alp
cent3.htm [cited June 15, 2007].
[42] Sol Company, “Gliese 867/Ross 780,” URL: http://www.solstation.com/stars/
gl876.htm [cited 15 June 2007].
[43] Sol Company, “55 Rho (1) Canceri 2,” URL: http://www.solstation.com/stars2/
55cnc2.htm [cited 15 June 2007].
[44] Forward, R. L., Flight of the Dragonfly, Baen Books, Riverdale, New York, 1985.
[45] Goddard, R. H., “The Ultimate Migration” (manuscript), 14 Jan. 1918, The Goddard
Biblio Log, Friends of the Goddard Library, 11 Nov. 1972.
[46] Bowman, L. “Interstellar Travel: A Family Affair,” URL: http://news.nationalgeogra
phic.com/news/2002/02/0220 0220 wirelifeinspace.html, 20 Feb. 2002, Scripps
Howard News Service, National Geographic News [cited 2 Jan. 2008]; Angier,
N., “Scientists Reach Out to Distant Worlds,” URL: http://query.nytimes.
com/gst/fullpage.html?res ¼9D05EED81630F936A35750C0A9649C8B63&sec¼
&spon¼&pagewanted¼all, 5 March 2002, The New York Times [cited 2 Jan. 2008];
Malik, T., “Sex and Society Aboard the First Starships,” URL: http://www.space.com/s
cienceastronomy/generalscience/star voyage 020319 1.html, 19 March 2002, Space.
Com [cited 2 Jan. 2008].
[47] O’Neill, G. K., “The Colonization of Space,” Physics Today, Vol. 27, No. 9, Sept.
1974, pp. 32 40; O’Neill, G. K., “Space Colonies: The High Frontier,” The Futurist,
February 1976.
126 R. H. FRISBEE

[48] Orth, C., Klein, G., Sercel, J., Hoffman, N., and Murray, K., “Transport Vehicle for
Manned Mars Missions Powered by Inertial Confinement Fusion,” AIAA Paper 87
1904, 29 June 2 July 1987.
[49] Paine, C., and Seidel, G., “Brown University Magnetic Levitation/Supercooling
Research,” NASA Office of Advanced Concepts and Technology (OACT) Third
Annual Workshop on Advanced Propulsion Concepts, Pasadena CA, 30 31 Jan.
1992, JPL Internal Document D 9416, R. H. Frisbee (ed.), 1992.
[50] Lesh, J. R., Rugglen, C., and Ceggarone, R., “Space Communications Techniques for
Interstellar Missions,” Journal of the British Interplanetary Society, Vol. 49, 1996,
pp. 7 14.
[51] Chalsson, E. J., Quoted in Section 4.3.6, Transcripts of Break Groups, Proceedings of
the Humanity 3000 Workshop, Humans and Space: The Next Thousand Years, Foun
dation for the Future, Bellevue, WA, June 2005, p. 182.
[52] Geller, T., “Aluminum: Common Metal, Uncommon Past,” Chemical Heritage
Newsmagazine, Winter 2007/8, Vol. 25, No. 4, URL: http://www.chemheritage.
org/pubs/magazine/feature alum p1.html [cited Jan. 18, 2008].
[53] Cassenti, B., “The Interstellar Ramjet,” AIAA Paper 2004 3568, 11 14 July 2004.
[54] TRW Space Log, W. A. Donop Jr. (ed.), Vol. 9, No. 4, Winter 1969 1970.
[55] Forward, R. L., Indistinguishable from Magic, Baen Books, Riverdale, New York,
1995.
 Chapter 3

Prerequisites for Space Drive Science

Marc G. Millis
NASA Glenn Research Center, Cleveland, Ohio

I. Introduction
O CIRCUMVENT the propellant limits of rockets and the maneuvering
T limits of solar sails, a means to propel spacecraft using only the interactions
between the spacecraft and its surrounding space is sought. A general term for
such a device is “space drive,” which is adopted for convenience from science
fiction. (This term first appeared in John Campbell’s 1932, The Electronic
Siege [1].) At present, the scientific foundations from which to engineer a
space drive have not been discovered. In fact, the issues, unknowns, and oppor-
tunities to seek these discoveries have only recently begun to be articulated [2,3].
To set the stage for further progress, this chapter examines this topic at the level
of the first step of the scientific method, specifically defining the problem.
While this chapter focuses on clarifying the problem to guide future research,
subsequent chapters deal separately with specific ongoing investigations. These
include manipulating gravity, electromagnetic gravitational couplings, photon
momentum in media, inertial modifications, quantum vacuum interactions,
and others.
It is important to stress that space drives might be physically impossible, and
conversely, that the prerequisite discoveries have just not yet been made. To
begin the discussion, the key physics issues, basic energy estimates, hypothetical
propulsion concepts, and relevant topics in science are examined. In several
instances, provocative conjectures are introduced to demonstrate how space
drive goals are distinct from the more general scientific inquiries. The intent is
to provide starting points that future researchers can use to specifically address
the physics of space drives. Eventually, future research will determine if, and
how, space drives are physically possible.

This material is a work of the U.S. Government and is not subject to copyright protection in the
United States.

Propulsion Physicist, Propulsion and Propellants Branch, Research and Technology Directorate.

127
128 M. G. MILLIS

A further assertion is that adding the challenge of breakthrough spaceflight can

enhance the study of the more general lingering unknowns of science. The chal-
lenges of spaceflight offer different lines of inquiry, providing insights that might
otherwise be overlooked from just curiosity-driven science.

II. Methods
Starting with the major objections to the notion of a space drive, namely con-
servation of momentum and the scarcity of indigenous reaction mass, specific
research objectives are articulated. Next, as both a tool for deeper analysis and
to reflect potential benefits, energy comparisons are made between ideal
rockets and space drives. This includes demonstrating possible pitfalls when
conducting such analyses. Finally, 10 hypothetical space drive concepts are
presented to illustrate possibilities and issues. Throughout, relevant physical
effects and lingering unknowns are cited. Where possible, the distinctions
between general science and the perspectives of propulsion-oriented science
are described. The intent is to provide starting points from which to apply scien-
tific progress toward answering the goal of enabling humanity to traverse inter-
stellar space.
Newtonian representations are used predominantly instead of General Relati-
vistic perspectives for several reasons. First, it is easier to introduce space drives
using Newtonian treatments. Newtonian terms are more familiar and can be
easily presented in the meter-kilogram-sec (MKS) units typical of engineering
endeavors; converting General Relativity equations into engineering terms can
be more cumbersome. Although General Relativity has broader validity, Newto-
nian treatments are a valid approximation for the low energy and nonrelativistic
situations dealt with here [4]. And finally, Newtonian perspectives allow inves-
tigating effects that occur within spacetime, while General Relativity deals with
spacetime itself. To convey this in terms of an analogy, consider moving an auto-
mobile across a landscape. General Relativity allows us to consider how to
reshape the landscape so that the automobile (and everything in the vicinity)
rolls passively downhill toward the desired destination. The Newtonian perspec-
tives allow us to consider how to move the automobile under its own power rela-
tive to the landscape. Both perspectives are intellectually provocative. While the
newtonian perspectives are prevalent here, subsequent chapters (Chapters 4 and
15) examine General Relativity perspectives in more depth.

III. Major Objections and Objectives

A. Conservation of Momentum
The most obvious issue facing the prospect of a space drive is conservation of
momentum. When accelerating a vehicle there must be an equal and opposite
change in momentum imparted to a reaction mass. This encompasses
Newton’s basic laws of motion. For land vehicles, the reaction mass is the
Earth’s surface; for rockets, it is their propellant; and for space sails, it is the
light that hits the sails. For space drives, the reaction mass is not obvious, but
options for further inquiry exist.
 PREREQUISITES FOR SPACE DRIVE SCIENCE 129

Even though one could entertain schemes where momentum conservation

does not apply, which would require delving even deeper into undiscovered
science, this chapter imposes conservation of momentum as a working
premise. From that premise, it follows that some form of indigenous reaction
mass must be found in space, and that a means to interact with that mass must
be discovered to create thrust.
Other phenomena that play a role in conservation of momentum are inertia
and the reference frames upon which momentum conservation is described.
Later in this chapter the implications of inertia and the role of reference
frames are explored in more depth. For now, the focus is on interacting with reac-
tion masses.
In preparation for examining the possible sources of indigenous reaction mass,
first consider the basic equation of momentum Eq. (1), where greater impulse
(change in momentum), Dp, can be obtained by either increasing the amount
of reaction mass used, m, or increasing the change in velocity, Dv, imparted to
that reaction mass. For these introductory exercises, it is sufficient to use non-
relativistic equations.

Dp ¼ mDv (1)

Because the space drive relies on interaction with some form of reaction mass
in the surrounding space, it is helpful to introduce terms that address the distri-
bution of that reaction mass. To this end, Eq. (2) presents a modified version of
Eq. (1) where the volume element of the affected space is included and in a
manner that still balances the equation. The reaction mass is then expressed as
a density, r.
To avoid confusion, please be alert to the visual similarity of the terms used to
represent the momentum, p, and mass density, r, and further between volume, V,
and velocity, v. These representations follow common conventions.
m
Dp ¼ V Dv ) Dp ¼ V rDv (2)
V

To allow exploring the implications of different momentum transfer schemes,

the volume element, V, is factored into the product of an area, A, and distance, r,
as shown in Eq. (3).

Dp ¼ rArDv (3)

It should be stressed that this equation is offered to help identify critical

factors, rather than to imply that this would be the exact form of any solutions.
For now, this equation merely identifies which factors pertain, and further, if
the factors are directly or indirectly proportional to the desired effect. When
investigating specific momentum-exchange methods, the equation will likely
change. For example, if momentum transfer is accomplished through a collision
process, analogous to light hitting a solar sail, then the propulsive effect is
directly proportional to the area of the sail, A, and r becomes negligible (r then
represents the perpendicular distance to the sail over which the momentum
130 M. G. MILLIS

transfer takes place). Conversely, if the momentum exchange is through some

intermediary field distributed over space, analogous to how a charged particle
is accelerated by an electric field, then the effective distance, r, of that field
becomes the dominant term. The equation, as presented, is just a first step
toward considering such options and their consequences.
To recap, by starting with a major objection to the notion of a space drive,
namely conservation of momentum, key objectives have now been identified.
From Eq. (3), it can be seen that a greater change in momentum can be achieved
by maximizing any of the following, and that these characteristics would need to
be addressed in future space drive research:
. Density of the natural phenomenon in space, r, which might constitute a reac-
tions mass (kg/m3)
. Change in velocity imparted to the reaction mass, Dv (m/s), (having an upper
limit of light speed, c 3.0  108 m/s)
2
. Surface area of the spacecraft’s space drive effector, A (m )
. Effective range of the space drive effect, r (m)

B. Indigenous Reaction Mass

The notion of interacting with some form of indigenous reaction mass in the
vicinity of the spacecraft presents the next objection: The vacuum of space
appears devoid of tangible matter. There are, however, several phenomena that
serve as starting points for deeper inquiry, as shown in Table 1. Consistent
with importance of mass density, r, as identified by Eq. (3), both the mass
density and equivalent energy density for these phenomena are listed. The
relation between the mass and energy densities is governed by the familiar
E mc 2 relation. The most intriguing of these phenomena are discussed in
subsequent paragraphs.
Estimates for the matter in the Universe are still evolving as astronomical
observations accumulate. The data shown are based on combinations of text-
books [4,5] and more recent data from the Wilkinson Microwave Anisotropy
Probe (WMAP) [6,7]. Table 1 includes the various omegas, Vi, (proportions)
of the known constituents, starting with the generally accepted assumption
that the total mass-density of the Universe satisfies the conditions for a “flat”
Universe [5].
Relative to the goal of space drives, the cosmological implications and uncer-
tainties [6] of these values are not of concern. Instead, in the context of space-
flight, the main question is if any of these phenomena can serve as a reaction
mass for a spacecraft. This perspective presents different lines of inquiry from
the more general cosmological concerns and thereby provides an additional
venue through which to learn about these phenomena.
The first issue for space drives is to find phenomena with the highest mass
density and, second, to find ways to create forces against these phenomena. At
present, none of these phenomena is an obvious candidate for a reaction mass,
but some are so inadequately understood that the opportunities for new discov-
eries are high. These include Dark Matter, Dark Energy, quantum fluctuations,
and the deeper meanings of spacetime itself. Even if the propulsive study of
these phenomena does not lead to a breakthrough, assessing their characteristics
 PREREQUISITES FOR SPACE DRIVE SCIENCE 131

Table 1 Known indigenous space phenomena

Known forms of mass and In terms of mass In terms of energy

energy density, r (kg/m3) density (J/m3)
27a 10
Total matter in the Proportions 9.5  10 8.6  10
universe (critical V ¼ 1.00
density)
“Dark Energy” VL ¼ 0.73b 6.9  10 27
6.2  10 10

“Dark Matter” VDM ¼ 0.22b 2.1  10 27

1.9  10 10

Baryonic matter VB ¼ 0.04b 3.8  10 28

3.4  10 11

(normal matter)
31 14
Photons and relativistic Vrel ¼ 8.3
b
 7.9  10 7.1  10
matter 10 4
Cosmic microwave VCMB ¼ 10 5 10 31c
10 15

background
Quantum vacuum
fluctuations
Inferred as dark energy 10 26d 10 9e
Up to nucleon Compton 1018 1035
frequency (1023 Hz) d
Up to Planck limit 1098 10113
(1043 Hz)
21f 4
Galactic hydrogen 3.3  10 3.0  10
Spacetime itself
27a 10
In terms of total 9.5  10 8.6  10
mass density g
General Relativity 5.3  1025 4.8  1042
analogy to Young’s
modulus

Values in this table that are accompanied by a reference citation are from that reference. Other values
shown are calculated from the cited values.
a
From Ref. 5.
b
From Ref. 6.
c
From Ref. 4.
d
Maclay, Chapter 12 in this book.
e
Davis and Puthoff, Chapter 18 in this book.
f
From Ref. 19.
g
See Eq. 6 to 8.

from the perspective of space drives will lead to a more thorough understand by
providing another perspective from which to analyze the data. In other words,
even if the propulsive goals are not viable, using a propulsion perspective will
improve our understanding of these phenomena.

1. “Dark Energy”
Dark Energy is only a working hypothesis to explain other anomalous obser-
vations. In this case, the original phenomenon that led to the Dark Energy
hypothesis is the anomalous redshifts from the most distant stars. Specifically,
the light coming from the most distant Type 1a supernovae (whose distance
can be more accurately inferred than other distant objects) is even more
132 M. G. MILLIS

redshifted than that attributable to Hubble expansion [8]. These anomalously

higher redshifts are interpreted as indications that the Universe’s expansion is
accelerating [9]. To explain what might be causing this, the term Dark Energy
has become the working hypothesis, specifically asserting that an invisible
form of energy is causing this expansion [10].
One interpretation is that this dark energy is related to quantum vacuum fluc-
tuations, but the values calculated by astronomical observations and from
quantum regimes are mismatched by roughly 120 orders of magnitude (see
Chapter 12 in this book). Clearly, this is a subject where more discoveries
await. The propulsive implications of the Dark Energy hypothesis, or more gen-
erally of the anomalous phenomena that led to the Dark Energy hypothesis, have
not yet been explored.

2. “Dark Matter”
Dark Matter is also a working hypothesis for other anomalous phenomena. The
first confirmed empirical observations that lead to the Dark Matter hypothesis are
the anomalous rotations of galaxies [11]. When considering galactic rotation rates
and newtonian mechanics, the stars at the outer regions of a galaxy should have
been moving away from the center of the galaxy. Instead, galaxies appear to
hold together as if some form of invisible matter is adding to the gravitational
attraction that binds the galaxy’s stars together. More recently, other observations
are showing indications of unseen gravitating matter. Specifically, gravitational
lensing has been observed around clusters of galaxies to a degree beyond that
attributable to the visible matter of these galaxies [12].
Whether it is indeed a new form of matter or indicative of other physics yet to
be discovered is not yet known. Other theories are being explored such as the
MOND hypothesis (modified newtonian dynamics) [13,14]. The utility of such
theories or the actual phenomena that led to the Dark Matter hypothesis have
not been explored in the context of spaceflight, either as a direct reaction mass
or as something indicative of other phenomena.

3. Cosmic Microwave Background Radiation

Considered to be a remnant of the Big Bang, space is filled with photons whose
frequency spectrum matches that of blackbody radiation corresponding to a temp-
erature of 2.7K. While cosmological studies focus on the slight anisotropic details
of this background [6,7], its prominent isotropic nature makes it intriguing for deep
spaceflight. Because the overall spectrum is so isotropic, Doppler shifts due to
motion through this background provide reliable measures of velocity relative to
the mean rest frame of the Universe. For example, the Earth’s motion through
this background has been measured as 365 km/s [15]. This cosmic microwave
background is thus a natural phenomenon for deep spaceflight navigation. Its pro-
spects as a propulsive media are discussed in Section V of this chapter.

4. Quantum Vacuum Fluctuations

As a consequence of the Uncertainty Principle in quantum physics, the energy
state of any system can never be exactly zero. This is true of the electromagnetic
 PREREQUISITES FOR SPACE DRIVE SCIENCE 133

spectrum of empty space. Chapter 12 provides an introduction to both the

theoretical and experimental details of this phenomenon, along with discussion
of its utility as an effective reaction mass for propulsion. Chapter 18 also dis-
cusses the energy implications of this phenomenon. Because subsequent chapters
in this text provide details, only preliminary information is offered here.
One of the main unknowns of this quantum vacuum energy is its fundamental
energy density. Estimates vary widely depending upon the upper cutoff fre-
quency applied to the calculations; Table 1 lists just three possibilities. The
first is the estimate based on the assertion that Dark Energy is identical to
quantum vacuum energy; the energy density is then calculated from estimates
of the total mass of the Universe and other factors [6]. At the other extreme, if
the Planck frequency is used as the cutoff value, an energy density of roughly
120 orders of magnitude higher is calculated [Maclay, Chapter 12]. The Planck
frequency (1043 Hz) is considered the absolute highest possible frequency
allowed by the application of the Uncertainty Principle to the structure of space-
time itself [16]. An intermediate estimate based on the Compton frequency of the
nucleon (1023 Hz) is about 80 orders of magnitude less than the Planck limit and
about 40 orders of magnitude greater than the Dark Energy estimate. Clearly, this
is an area where more discoveries can be made.

5. Virtual Particle Pairs

A collateral phenomenon to quantum vacuum energy is virtual particle pairs,
where particle antiparticle pairs continually appear and then disappear from
space. The prospects for using this phenomenon for space flight date back to at
least 1997 [17].
Regardless of the wide discrepancy in the estimated energy levels of the
quantum vacuum background, there is a fundamental limit to the deliberate pro-
duction of such matter from the vacuum. The creation of mass from energy, is
governed by the familiar E mc 2 relation. When considering imparting a vel-
ocity, v, to the particles, the more complete form of the energy mass equation
applies [18]:
q
E ¼ mc2 = 1  (v=c)2 (4)

where E is the energy required to create a mass, m, with a velocity, v. Considering

that this equation represents 100% efficient conversion, this significant energy
expenditure is only a lower limit on the energy required to impart momentum
to virtual pairs.

6. Galactic Hydrogen
In the voids of our Galaxy, there exist small amounts of hydrogen that have
been considered as a possible reaction mass for spaceflight. One concept, the
Bussard interstellar ramjet, is discussed in Section V. A of this chapter.
Based on analyses of the mass-to-light ratios of our Galaxy [19], it is estimated
that the density of molecular and atomic hydrogen is 3.3  10 21 kg/m3. Other
sources assert that some areas may be as high as 10 20 kg/m3 [20].
134 M. G. MILLIS

7. Spacetime Itself
In addition to the lingering mysteries of Dark Energy and Dark Matter, among
others, there is the deeper unknown of the fundamental phenomenon that
underlies the Universe itself: spacetime. Spacetime is the background against
which all things exist and their interactions are measured. From the Hubble
expansion, it appears that spacetime can change dimensions. As evidenced
from the lightspeed limit, spacetime has properties that govern the motion of
mass and energy. And spacetime even has the property of being an inertial
frame specifically, the property of a space that serves as a reference frame
for acceleration and rotation. But because it is so ubiquitous, it is difficult to
isolate for deeper study.
When faced with the Hubble expansion, an obvious question is: “Expanding
relative to what?” As a reflection of the cosmologist perspective, examine the
following quote from a general relativity textbook [4].

It’s simplest and most elegant to assume that the observed homogeneity and
isotropy extend over the whole universe. In an exactly homogenous model
there can be no center and no boundary. In that context it doesn’t make
sense to talk about the universe expanding into something or from
somewhere.

To a propulsion engineer, knowing the components of what he or she is working

with is of paramount importance. If there is some underlying structure to space-
time itself (against which spacetime expands) then it must also be considered.
Given how much is still not known and considering that the “no center and no
boundary” assertions are working assumptions, it is reasonable to consider that
additional constituents can be hypothesized.
Spacetime is the background against which both electromagnetic and gravita-
tional fields exist. The factors that govern how these fields behave are well
modeled, but how they function and if additional factors await discovery are
unknown. Considering the Dark Energy and Dark Matter anomalies and the
fact that physics on the quantum scale has still not been merged with the large
scale physics of General Relativity, it is clear that there is much left to learn.
Consider the lightspeed limit. The evidence shows that lightspeed is always
the same in any local inertial frame. It is a relation that governs the fundamental
correlation between space and time for all electromagnetic phenomena. But when
viewed more globally and in the presence of matter, the situation is more com-
plicated. Gravity bends light or spacetime itself, and there is more than one
way to model the observations. For example, in the geometric versions of
General Relativity, the speed of light is used as the defining constant, and conse-
quentially, space and time curve in the presence of matter. Within General
Relativity there also exist “optical analogies,” where space is represented as an
optical medium with an effective index of refraction that is a function of gravita-
tional potential [21,22]. In this case, the speed of light is a variable. Although
different from the more common geometric interpretation, this interpretation
has been shown to be consistent with physical observables, and comparisons
between the optical and geometric perspectives have also been published [23,24].
 PREREQUISITES FOR SPACE DRIVE SCIENCE 135

To explain this difference using a simpler analogy, consider the basic equation
of constant velocity motion:

d ¼ vt (5)

where d is the distance covered after a time, t, of moving at velocity, v. From the
geometric versions of General Relativity, the velocity, v, is held as the universal
constant (lightspeed), and thus both space, d, and time, t, have to “curve” when
this relation gets distorted by the presence of matter. Conversely, from the optical
analogy perspective, space, d, is taken to be fixed, and thus both the speed, v, and
time, t, must adjust when this relation gets distorted by the presence of matter.
Both of these options are available for deeper study, and the choice of which
to pursue depends on what problem one is trying to solve.
Little attention is typically focused on the optical analogy because it does not
predict any new effects that are not already covered by the more common geo-
metric perspective and because it raises unanswered issues with coordinate
systems choices (Matt Visser, personal communication, 2 October 2002). But
it is these coordinate system issues that are provocative from the spaceflight
point of view, and to further illustrate this issue it is necessary to address
Mach’s principle.
Mach’s principle asserts that an inertial frame, specifically the property of a
space to be a reference frame for acceleration, is actually created by, and con-
nected to, the surrounding mass in the Universe [25]. A related assertion is
that a literal interpretation of Mach’s Principle implies an absolute reference
frame, coincident with the mean rest frame for all the matter in the Universe
[26]. This is because the matter in the Universe creates the inertial frame and,
hence, the location and intensity of that inertial frame tracks with the position
and distribution of that matter. Such considerations will evoke issues with coor-
dinate transforms.
It can be argued that, even if such an underlying, machian inertial frame did
exist, it could not be detected because inertial forces do not appear until there
is a change in velocity. Conversely, however, it can be argued that this nondetect-
ability is a consequence of how one chooses to model phenomena. A crucial
underlying premise of both Special and General Relativity is that their very foun-
dations are constructed to provide frame-independent representations [4]. This
comes from the desire to find representations that are valid everywhere, with
universal correctness. But if Nature does possess frame-dependent phenomena,
it would be difficult to describe these with theories that are limited to frame-
independent representations.
There have been attempts to incorporate frame-dependent considerations into
General Relativity, such as the Brans-Dicke Theory [16,27]. There have also been
attempts to incorporate Mach’s principle into General Relativity [28], but it
remains an unresolved problem in physics. In the Bondi and Samuel paper
[28], which deals with the connection between the Lense Thirring effect
(frame dragging) and Mach’s principle, no less than 10 variations of Mach’s
principle are articulated, none of which specify propulsive considerations.
As mentioned previously, Nature does possess at least one frame-dependent
phenomenon, namely the cosmic microwave background. While curiosity-
driven science is focused on understanding the minute variations in this
136 M. G. MILLIS

background for its implications to the origins, structures, and fate of the whole
Universe [6,7,29], a propulsion engineer sees a fore aft energy difference
when moving through space. Such an effect also provokes conjectures about
inducing motion by deliberately creating fore aft energy differences around a
spacecraft. (Hypothetical mechanisms are listed in Section V of this chapter.)
If, like the cosmic microwave background, there is a literal machian inertial
frame that is connected to the surrounding mass of the Universe, then it might
serve as a propulsive medium specifically an effective reaction mass to thrust
against. By thrusting against this progenitor inertial frame, the reaction forces
are imparted to the surrounding mass of the Universe. At this point, this is
sheer conjecture, but considering how little is known about the nature of inertial
frames themselves, this notion at least warrants deeper consideration. Continuing
with conjectures, the notion that there might be a connection between the cosmic
microwave background and such a machian frame also warrants consideration.
Pursuing such conjectures leads to a different line of inquiry than the more
common frame-independent, geometric form of General Relativity. The propul-
sion perspective leads to euclidean treatments like the previously mentioned
“optical analogies.” This is where frame-dependence might be more readily
addressed. The difficulties with coordinate transforms could be due to the implicit
connection to frame-dependent inertial frames, but again, at this stage, these are
just conjectures for future research. Situations such as this are what support the
assertion that meaningful and different knowledge will be revealed by the
pursuit of spaceflight breakthroughs, which might otherwise be overlooked by
curiosity-driven physics.
Proceeding now to provide some initial estimates of the effective reaction mass
of spacetime itself, two versions are offered. The first is just to use the average
mass density of the whole Universe to reflect how much mass contributes to
the machian frame. Here the value already cited from Carroll and Ostlie [5]
applies. The second version uses an analogy to Young’s modulus from Einstein’s
field equation.
Young’s modulus is a way of quantifying the property of materials for how
much they distort when under strain. The higher the value, the stiffer the material,
or said another way, the less it will distort when subjected to a given force. The
units for Young’s modulus, Y, are Force/Area, which is equivalent to the units of
Energy/Volume, and is defined as [30]:

Stress
Y¼ (6)
Strain

Compare this to Einstein’s Field Equation (shown with MKS units) where the
major portions are isolated in parentheses to simplify the analogy [4]:
   
1 8p G
Rmn  gmn R  Lgmn ¼ ðT mn Þ
2 c4
  (7)
8p G
ðStrainÞ ¼ ðStressÞ
c4
 PREREQUISITES FOR SPACE DRIVE SCIENCE 137

All the terms on the left-hand side represent the curvature of spacetime, which is
analogous to the strain (distortion) resulting from an imposed stress. The right-
hand side includes the energy momentum stress tensor, T mn, which is analogous
to the stress imposed onto spacetime. The scaling factor (8pG/c 4) is based on the
properties of spacetime and is analogous to the reciprocal of Young’s modulus.
Continuing with the analogy, it follows that spacetime (where G is Newton’s con-
stant and c is lightspeed) has an analogous Young’s modulus per:
 4 
Stress c N
¼ ¼ 4:8  1042 2 (8)
Strain 8pG m

This extremely high value (the units are equivalent to energy density) suggests
that spacetime is very stiff. This means it takes a significant amount of matter or
field energy (represented by the energy momentum stress tensor, T mn) to induce
much of a curvature on spacetime. To what degree, if any, this makes spacetime a
viable reaction mass is completely unknown. This calculation and perspective is
offered to simply provoke deeper inquiries.

8. Summary
At present, none of the indigenous phenomena of space are obvious candidates
for a space drive reaction mass, but some are so inadequately understood that the
opportunity for new discoveries is high. These include Dark Matter, Dark
Energy, quantum fluctuations, and the deeper meanings of spacetime itself.
Even if none of these leads to a desired breakthrough, assessing their character-
istics from the perspective of a space drive increases the chances of more
thoroughly understanding nature by providing another perspective from which
to analyze the data.
In the first step of the scientific method, where the problem is defined, that
definition affects how data are subsequently collected and analyzed. For space
drive inquiries, some basic initial questions include:
. What are the indigenous phenomena in space?
. Can forces be induced by interacting with indigenous phenomena?
. If so, can net forces be created between a device and a phenomenon?
. Are the forces and the amount of accessible reaction mass sufficient to propel a
spacecraft?
. If not directly applicable for propulsion, might the phenomenon be indicative
of other potentially more useful phenomena? In other words, if the phenom-
enon is not suitable as a reaction mass, might it be useful to measure other
phenomena?
To examine the issues and limitations of interacting with any of these candi-
dates, it is helpful to envision a number of hypothetical space drives and then
analyze their viability. Before proceeding to that step, the energy associated
with space drives is examined next. These energy analyses highlight pitfalls of
prior assessments, give a crude estimation of potential benefits, and provide
another means to explore the physics of space drives.
138 M. G. MILLIS

IV. Estimating Potential Benefits

A. Avoiding Pitfalls
The historic tendency when trying to gauge the value of an emerging technol-
ogy is to compare it using the characteristics of the incumbent technology. Such
provisional assessments can be seriously misleading, however, when the emer-
ging technology uses fundamentally different operating principles. For
example, the value of steamships is misleading when judged in terms of sails
and rigging [31]. Although reduced sail area and rigging lines are indeed a con-
sequence of steamships, the true benefit is that shipping can continue regardless
of the wind conditions and with far more maneuvering control. Similarly, the
benefits of a breakthrough space drive would likely surpass the operational con-
ventions of rocketry. Issues such as optimizing specific impulse become mean-
ingless if there is no longer any propellant. Three examples are offered next to
illustrate the pitfalls of using rocket equations to describe the physics of a
space drive.

1. Infinite Specific Impulse Perspective

The first and common misleading practice when describing a hypothetical
space drive is to view it as a rocket with an infinite specific impulse. This
seems reasonable at first because a higher specific impulse leads to less propel-
lant, so an infinite specific impulse should lead to zero propellant. As shown
from Eq. (9), however, specific impulse is a measure of the thrust, F, per propel-
lant weight flow rate (g dm/dt) [32]. For a true space drive, the dm/dt term would
be meaningless, rendering the entire equation inappropriate for assessing propel-
lantless propulsion.
F
I sp ¼ (9)
dm
g
dt
Furthermore, as shown from Eq. (10), which is based on the energy imparted
to the propellant from the rocket’s frame of reference [33], an infinite specific
impulse, Isp, implies that a propellantless space drive would require infinite
energy (substituting Isp 1).
Conversely, this same equation can be used to conclude that a propellantless
space drive would require zero energy if there was no propellant (substituting
mp 0). Neither of these extremes is likely the case. (In the strictest sense when
extrapolating, the squared impulse term would dominate the mass term, resulting
in infinite energy. Regardless, this equation is not appropriate when contemplat-
ing propellantless propulsion.)

1  2
E ¼ mp Isp g (10)
2
2. Modify Rocket Inertia
To illustrate another misleading use of the rocket equations, consider the
notion of manipulating the inertia of a rocket. If such a breakthrough were
 PREREQUISITES FOR SPACE DRIVE SCIENCE 139

Fig. 1 Rocket in field-free space.

ever achieved, the implications and applications would likely extend beyond
rocketry. Even if used on a rocket, there are a number of different ways to envi-
sion applying such an effect, each yielding considerably different conclusions;
i.e., apply the inertial change 1) to the whole rocket system, 2) just to the expelled
propellant, or 3) just to the rocket with its stored propellant. In the latter, a con-
dition of operation is that the propellant resumes its full inertia as it is accelerated
out of the rocket. This notion is similar to the science fiction concept of “impulse
drive with inertial dampers” as presented in the Star Trek television series [34].
It should be noted that, at present, there are no confirmed techniques to affect
such a change in inertia, even though experiments are underway [35]. It is import-
ant to stress that this is only a hypothetical example to illustrate the sensitivity of
the findings to the methods, rather than to suggest that this is a viable break-
through. Numerous variations on this analysis are possible.
Consider a rocket in field-free space (Fig. 1). To derive the rocket equation,
one can start with conservation of momentum, where the rocket expels an
increment of propellant, dm, to produce an incremental change in the rocket’s
velocity, dv.
The standard equation to represent this conservation of momentum has been
slightly modified into Eq. (11), where coefficients are inserted to represent
hypothetical manipulations of the inertia of the expelled propellant, dp, and the
rocket, dm, where the rocket includes the stored propellant. Values of d greater
than one imply an increase, less than one imply a decrease, and equal to one rep-
resents no change. To be clear, when considering such isolated modifications of
inertia, the dp term only affects the propellant as it is ejected from the rocket, and
the dm term only affects the inertia of the stored propellant. In this latter case, it is
assumed that the propellant regains its full inertia as it is accelerated out of
the rocket.
 
ve dp dm ¼ dvðdm Þðm  dmÞ (11)

Proceeding with the normal steps to derive the rocket equation, it can be
shown [36] that the final result for the Dv imparted to the rocket is represented by:
m þ m  d
p p
Dv ¼ ve ln (12)
m dm
Consider now the implications of modifying the inertia of the whole rocket
system, which implies equal changes to dp and dm. In this circumstance there
140 M. G. MILLIS

Table 2 Different ways to modify rocket inertia

Which inertia is modified: Propellant Rocket

From Eq. (12) dp dm Net effect

Unmodified 1 1 Baseline: Dv0 ¼ Dv

Whole rocket system d d No change: Dv0 ¼ Dv
Just rocket with its stored propellant 1 d Dv0 ¼ 1/dDv
Just ejected propellant d 1 Dv0 ¼ dDv

is no change at all in Dv. This null finding was one of the observations reported by
Tajmar and Bertolami [37]. Alternatively, consider that the inertia of the rocket
with its stored propellant is somehow reduced, while the inertia of the expelled
propellant regains its full value. In this case the improvement in Dv tracks inver-
sely to dm. In other words a dm of 0.5, representing a 50% decrease in the rocket’s
inertia, would yield a 50% increase in Dv. Table 2 summarizes how the different
assumptions yield different results.
In addition to the ambiguity and wide span of results when using rocket
equations to predict the benefits of modifying inertia, this approach does not
provoke the questions needed to further explore such conjectures. For
example, the issue of energy conservation is not revealed from the prior
equations; although momentum was conserved, energy conservation is not
addressed. It is presumed that any benefit must come at some expense, and
because energy is a fundamental currency of mechanical transactions, it is
reasonable to expect that such a benefit requires an expenditure of energy.
These equations do not provide the means to calculate the extra energy required
to support this hypothetical change in the rocket’s inertia.

3. Gravitationally Altered Launch Pad

Another misleading use of rocket equations is when considering the impli-
cation of a hypothetical “gravity shield.” This idea was provoked from the
“gravity-shield” claim [38] that was later found not to be reproducible [39],
but this scenario still serves to illustrate key issues. This example highlights
issues associated with the Equivalence Principle. The Equivalence Principle
asserts that gravitational mass is identical to inertial mass. If the gravitational
mass is modified, then the inertial mass would be similarly modified.
Consider placing a launch pad above a hypothetical gravity shield (Fig. 2). A
naive assumption would be that the reduced gravity would make it easier for the
rocket to ascend as if being launched from a smaller planet.
There is more than one way to interpret this situation if one entertains possi-
bilities beyond the initial “gravity shield” conjecture. In addition to considering
that the gravitational field, g, is modified, one can consider that the “gravity
shield” is, instead, altering the mass of the rocket above the device. In the case
where the rocket mass is modified, one can further consider that just its gravita-
tional mass is affected or, if the Equivalence Principle is in effect, that both its
 PREREQUISITES FOR SPACE DRIVE SCIENCE 141

Fig. 2 Rocket over hypothetical gravity shield.

gravitational and inertial mass are equally affected. Figure 3 illustrates these
assessment options.
To explore these options, start with Eq. (13) for a rocket ascending in a grav-
itational field [40]. The term on the left represents the mass and acceleration

Fig. 3 Analysis options for gravitationally shielded launch pad.

142 M. G. MILLIS

of the rocket; the middle term is the force of gravity; and the right term is the
reaction force from the expulsion of an increment of propellant with an
exhaust velocity, ve, relative to the rocket.

dv dm
m ¼ m g  ve (13)
dt dt

To consider the hypothetical modifications, coefficients are inserted next to

reflect modifications to the gravitational field, dgf, and to the rocket’s gravita-
tional mass, dgm, and its inertial mass, dim. As before, values of d greater than
one imply an increase, less than one a decrease, and equal to one represents no
change. Also, the equation is now rearranged to isolate the inertial terms from
the gravitational terms:

    dv dm
 dgm m dgf g ¼ ðdim Þ m þ ve (14)
dt dt

The left side represents the gravitational contributions while the right side rep-
resents the inertial contributions. It can be shown that this equation results in the
following representation for the Dv of the rocket:

 
dgm minitial
Dv ¼  dgf gDt þ ve ln (15)
dim m final

The increment of time during which propellant is expelled is represented by

Dt, and accordingly the two mass terms reflect the initial (higher) and final
(lower) masses of the rocket (including its stored propellant) over this time inter-
val. With the exception of the modification coefficients, this equation is identical
to that of a normal rocket ascent in a gravitational field.
Table 3 shows how the different possible interpretations of the hypothetical
gravity shield (as outlined in Fig. 3) might affect this situation. If it were
assumed that the gravitational field, g, is modified, the result would be as
naively expected; it would be the same as launching in a different gravitational
environment. If, however, it is assumed that the device affects the mass of the
rocket, there are further possibilities. If the Equivalence Principle is in effect,
then both the gravitational and inertial mass are equally affected, resulting in
no change in the rocket’s Dv. If the Equivalence Principle is not in effect, then
only the gravitational or inertial masses are affected, resulting in an analogous
case to launching in a different gravitational environment.
As before, the energy implications of these conjectures are not addressed
in such approaches. These examples illustrate how misleading results become
possible if only rocket equations are used to ponder space drive physics. To
proceed with a more fundamental basis of comparison, energy considerations
are examined next.
 PREREQUISITES FOR SPACE DRIVE SCIENCE 143

Table 3 Different analysis results for gravitationally shielded launchpad

Gravitational Gravitational Inertial

Modified term field mass mass
From Eq. (15) dgf dgm dim Net effect

Unmodified launch 1 1 1 Baseline:

g0 ¼ g
Gravity modified d 1 1 g 0 ¼ dg
Rocket inertial and gravitational 1 d d No change:
mass (Equivalence Principle g0 ¼ g
in effect)
Rocket gravitational mass 1 d 1 g0 ¼ dg
(Equivalence Principle
negated)
Rocket inertial mass 1 1 d g0 ¼ g/d
(Equivalence Principle
negated)

B. Energy as a Basis of Comparison

Although comparisons built on the incumbent methods might be useful for
introductory purposes, a deeper understanding of the benefits and issues are
better illustrated by using a more fundamental basis. When considering
moving a mass from one place to another, energy is the fundamental currency.
Using this basis, three situations will be compared; deep space travel, Earth to
orbit, and levitation. These comparisons are presented for three reasons: 1) to
continue to illustrate the difference between space drive considerations and the
use of rocket equations, 2) to reflect the magnitude of potential benefits compared
to rockets, and 3) to provide starting points for deeper inquiry.

1. Deep Space Travel Energy

To compare the energy requirements of a rocket and a hypothetical space
drive, the following assumptions are used. To more fully understand the chal-
lenges, approaches, and potential benefits of breakthrough propulsion, it would
be fruitful to repeat the analysis using different assumptions.
. The space drive is interpreted to be simply a device that converts potential
energy (regardless of the source of this energy) into kinetic energy.
. Both the rocket and the space drive are assumed to be 100% efficient with their
energy conversions. Absent of any real mechanism, this at least compares the
upper performance limits.
. The thrusting duration is assumed to be much shorter than the trip duration,
which for interstellar travel is reasonable.
. For the rocket, constant exhaust velocity is assumed.
. Nonrelativistic trip velocity and exhaust velocity are assumed.
. The energy requirements for a rendezvous mission are based on equal Dv’s for
acceleration and deceleration.
144 M. G. MILLIS

To compare a rocket to another method that does not require propellant, we

need an equation for rocket energy where the propellant mass is represented in
terms of the vehicle’s empty mass and the Dv of the mission variables shared
by the space drive. A common way to calculate the total kinetic energy of a
rocket system, including both the rocket and the propellant, is just to calculate
the kinetic energy imparted to the propellant from the rocket’s frame of reference
where the rocket has zero velocity (hence a zero contribution to the total kinetic
energy) [32,33]. This is consistent with the previously stated assumptions.

1
E ¼ m p ðve Þ2 (16)
2

Next, to convert this into a form where the rocket’s propellant mass, mp, is
represented in terms of the exhaust velocity and the mission Dv, we apply the
following form of the Rocket Equation, which is an algebraic equivalent
of the Tsiolkovski equation:

 Dv 
m eðve Þ  1 ¼ m p (17)

Substitution of this form of the Rocket Equation into the Kinetic Energy Equation
yields this approximation:

1  Dv 
E ¼ ðve Þ2 m eðve Þ  1 (18)
2

Before proceeding, a limitation must be communicated. For these introductory

exercises, the comparisons are limited to nonrelativistic regimes. For rockets, this
implies limiting the exhaust velocity to 10% lightspeed, where g is less than
about 1%. (g is the factor by which length and time are affected and is a function
of the speed between observers.) The corresponding upper limit to specific
impulse follows easily from the equation relating specific impulse to exhaust
velocity (where g is the Earth’s gravitational acceleration) [32]:

ve ¼ Isp g (19)

Setting the exhaust velocity to 10% of lightspeed (beyond which relativistic

effects must be considered), the limiting specific impulse is found to be:

 m  m
ð10%Þ 3:0  108  Isp 9:8 2 ) Isp  3:0  106 s (20)
s s

Because a space drive has been defined for this exercise as a device that con-
verts potential energy into kinetic energy, the basic equation of kinetic energy is
used to calculate the required energy, where the values of vehicle mass and
 PREREQUISITES FOR SPACE DRIVE SCIENCE 145

mission Dv are the same as with the rocket. For these first-step exercises, the
source of the stored potential energy need not be specified. The first issue to
deal with is the magnitude of energy.
1
E ¼ mðDvÞ2 (21)
2
Two things are important to note regarding the energy differences between a
rocket and a hypothetical space drive. First, the energy for a rocket is an exponen-
tial function of Dv, whereas the ideal energy of a space drive is a squared function
of Dv. This by itself is significant, but it is important to point out that a rocket and
a space drive treat additional maneuvers differently.
For a rocket it is conventional to talk in terms of increases to Dv for additional
maneuvers. For example, a rendezvous mission has twice the Dv (accelerate and
decelerate) than just a flyby (accelerate). For space drives, however, the
additional maneuvers are in terms of additional kinetic energy. To illustrate
this difference, consider a mission consisting of multiple maneuvers, n, each
having the same incremental change in velocity, Dv. Notice in Eqs. (22) and
(23) the location of the term representing the number of repeated maneuvers,
n. In the case of the space drive, additional maneuvers scale linearly, while for
rockets they scale exponentially:

1  Dv

Rocket maneuvers: E ¼ ðve Þ2 m eððnÞve Þ  1 (22)
2
1
Hypothetical space drive maneuvers: E ¼ ðnÞ mðDvÞ2 (23)
2
To put this into perspective, consider a mission to send a 5000 kg probe over a
distance of 5 light-years in a 50-year time frame. This range is representative of
the distance to our nearest neighboring star (4.3 light-years) and the 50-year time
frame is chosen because it is short enough to be within the threshold of a human
career span, yet long enough to be treated with nonrelativistic equations. This
equates to a required trip velocity of 10% light speed. The probe size of
5000 kg is roughly that of the Voyager probe plus the dry mass of the Centaur
Upper Stage (4075 kg) that propelled it out of Earth’s orbit [41]. The comparison
is made for both a flyby mission and a rendezvous mission.
The results are listed in Table 4. The rocket case is calculated for two different
specific impulses, one set at the upper nonrelativistic limit previously described,
and another set at an actual maximum value achieved during electric propulsion
lab tests [42].
Even in the case of the nonrelativistic upper limit to specific impulse
an incredibly high-performance hypothetical rocket the space drive uses a
factor of 2 to 3 less energy. When compared to attainable values of specific
impulse values that are still considerably higher than those currently used in
spacecraft the benefits of a space drive are enormous. For just a flyby
mission, the gain is 72 orders of magnitude. When considering a rendezvous
mission, the gain is almost 150 orders of magnitude. Again, though these
results are intriguing, they should only be interpreted as the magnitude of
146 M. G. MILLIS

Table 4 Deep spaceflight energy comparisons (5000 kg, 5 Ly, 50 yr)

Flyby Compared to Rendezvous Compared to

(n ¼ 1) space drive (n ¼ 2) space drive
(Joules) Eq. (22)/Eq. (23) (Joules) Eq. (22)/Eq. (23)

Space drive, Eq. (23) 2.3  1018 4.5  1018

Theoretical rocket, 3.8  1018 1.7 1.5  1019 3.3
Eq. (22)
Isp ¼ 3,000,000 s
Actual rocket, 1091 1072 10168 10149
Eq. (22)
Isp ¼ 17,200 sa
a
From Ref. 42.

gains sought by breakthrough propulsion research. Other assessments and results

are possible.
In the case of deep space transport, the energy was previously calculated
assuming a constant exhaust velocity for the rocket and thrusting durations that
were negligible compared to trip times. Although reasonable assumptions for
interstellar flight, it would also be instructive to repeat the energy comparisons
with assumptions of constant acceleration, constant thrust, constant power, and
when optimized for minimum trip time. To further explore these notions, it
would also be instructive to repeat all of these comparisons using the relativistic
forms of the equations.
Newtonian equations are not the only way to further explore these notions.
From the formalism of General Relativity, there are a variety of transportation con-
cepts that do not require propellant, including: a gravitational dipole toroid (indu-
cing an acceleration field from frame-dragging effects) [43], warp drives (moving
a section of spacetime faster than light) [44], wormholes (spacetime shortcuts)
[45,46], and Krasnikov tubes (creating a faster-than-light geodesic) [47].
To explore these General Relativity formalisms in the context of creating
space drives requires the introduction of entirely different energy requirements
than with the Newtonian versions explored in this paper. In the General Relativity
approach, one must supply enough energy to manipulate all of the surrounding
spacetime so that the spacecraft naturally falls in the desired direction. Although
such approaches require considerably more energy than the simple Newtonian
concepts, they are nonetheless instructive.

2. Earth to Orbit Energy

Consider next the case of lifting an object off the surface of the Earth and
placing it into orbit. This requires energy expenditures both for the altitude
change and for the speed difference between the Earth’s surface and the orbital
velocity. Again, the source of this energy is not considered. The point explored
first is the amount of energy required. For the hypothetical space drive, this
energy expenditure can be represented as:
ESpace drive ¼ U þ K (24)
 PREREQUISITES FOR SPACE DRIVE SCIENCE 147

where U is the potential energy change associated with the altitude change, and
K is the kinetic energy change associated with different speeds at the Earth’s
surface, rE, and at orbit, rO. The change in potential energy, which requires
expending work to raise a mass in a gravitational field, is represented by:
ð rO
ME
U¼ G 2 m dr (25)
rE r

where G is Newton’s constant (6.67  10 11 m3 . kg 1 . s 2), ME is the mass of

the Earth (5.98  1024 kg), r is the distance from the center of the Earth, and m is
the mass of the vehicle.
The change in kinetic energy requires solving for the orbital velocity (function
of orbital radius, rO) and the velocity of the Earth’s surface (Earth’s circumfer-
ence divided by one rotation period) and can be shown to take this form [36]:
"    #
1 ME 2p rE 2
K¼ m G  (26)
2 rO 24 h

For the case of using a space drive to place the shuttle orbiter (m 9.76  104
kg) into a typical low Earth orbit, (rO 400 km), the energy required is found to
be 3.18  1012 Joules.
To assess the required energy for a rocket to accomplish the same mission, the
following equation is used [33]:
 
1
E¼ F Isp g t (27)
2

The parenthetical term is the rocket power, which is presented in this form for
two reasons: to show this additional detail of the Rocket Equation and to intro-
duce the idea of contemplating power in addition to just energy. While power
implications are not further explored here, they constitute a fertile perspective
for further study.
Entering the values for the Space Shuttle System, as presented in Table 5, into
Eq. (27), the total energy for delivering the Shuttle orbiter via rockets is found to
be 1.16  1013 Joules.
Comparing this rocket energy value to the hypothetical space drive energy,
Eq. (28), where the efficiency of both systems is assumed to be 100%, indicates
that the space drive is 3.65 times more energy efficient.

Rocket 1:16  1013

Gain ¼ ¼ ¼ 3:65 (28)
Space drive 3:18  1012

3. Levitation Energy
Levitation is an excellent example to illustrate how contemplating break-
through propulsion is different from rocketry. Rockets can hover, but not for
148 M. G. MILLIS

Table 5 Data and calculations of Space Shuttle energy to reach low Earth orbita

Space Shuttle Solid rocket Orbital maneuvering

main engines boosters system

Number of engines 3 2 2
Thrust each, F 470  103 lbs 6 lbs 6  103 lbs
(2.1  106 N) (12.9  106 N) (271  103 N)
Specific impulse, Isp 453 s 266 s 313 s
Total power 1.40  1010 W 3.36  1010 W 8.31  108 W
Burn duration, t 514 s 126 s 200 s
Total energy used, E 7.19  1012 J 4.24  1012 J 1.66  1011 J
Total Combined Energy ¼ 1.16  1013 Joules
a
Values taken from STS 3 Thirds Space Shuttle Mission Press Kit, March 1982, Release #82 29.

very long before they run out of propellant. For an ideal breakthrough, some form
of indefinite levitation is desirable, but there is no preferred way to represent the
energy or power to perform this feat. Because physics defines work (energy) as
the product of force acting over distance, no work is performed if there is no
change in altitude. Levitation means hovering with no change in altitude. This
zero-energy expenditure is also obtained in the case of indefinitely levitating a
permanent magnet over an arrangement of other permanent magnets that are,
themselves, supported by the Earth.
Regardless, there are a variety of ways to toy with the notion of indefinite levi-
tation that look beyond this too-good-to-be-true zero-energy requirement. In
addition to the potential energy approach to be examined next, here are a
variety of other ways to contemplate levitation energy:
1) Helicopter analogy: Calculate the energy and power required to sustain a
downward flow of reaction mass to keep the vehicle at a fixed altitude.
2) Normal accelerated motion: Rather than assess levitation energy directly
(where the mass sustains zero velocity in an accelerated frame), calculate the
energy or power required to continuously accelerate a mass at 1 g in an inertial
frame. In this case, the normal Force  Distance formula can be used, but
issues with selecting the integration limits arise.
3) Escape velocity: Calculate the kinetic energy for an object that has
achieved escape velocity. (This approach actually results in the same value as
when calculating the absolute potential energy.)
4) Thermodynamic: Treat levitation analogously to keeping a system in a
nonequilibrium state, where equilibrium is defined as free-fall motion in a grav-
itational field and where the stable nonequilibrium condition is defined as levita-
tion at a given height.
5) Damped oscillation: Calculating the energy of oscillation about a median
hovering height, but where an energy cost is incurred for both the upward and
downward excursions, where damping losses are included.
6) Impulse: Rather than use the Force  Distance formula, use the “impulse”
treatment of Force  Duration. Like the accelerated motion approach, this intro-
duces issues with integration limits.
 PREREQUISITES FOR SPACE DRIVE SCIENCE 149

For now, only one approach is illustrated, specifically the nullification of grav-
itational potential. Usually Eq. (29) is used to compare gravitational potential
energy differences between two relatively short differences in height (the inte-
gration limits), but in our situation we are considering this energy in the more
absolute sense. This equation can be applied to calculate how much energy it
would take to completely remove the object from the gravitational field, as if
moving it to infinity. This is more analogous to nullifying the effect of gravita-
tional energy. This is also the same amount of energy that is required to stop
an object at the levitation height, r, if it were falling in from infinity with an
initial velocity of zero (switching signs and the order of integration limits).
ð1
ME ME
U¼ G m dr ¼ G m (29)
rE r2 rE

where G is Newton’s Constant (6.67  10 11 m3 . kg 1 . s 2), ME is the mass of

the Earth (5.98  1024 kg), r is the distance from the center of the Earth, and m is
the mass of the vehicle.
Using this potential energy equation, it could conceivably require 62 mega-
Joules to levitate 1 kg near the Earth’s surface. As an aside, this is roughly
twice as much as putting 1 kg into low Earth orbit. Again, these assessments
are strictly for illustrative purposes rather than suggesting that such break-
throughs are achievable or if they would even take this form if achievable.
Some starting point for comparisons is needed, and this is just one version.
To avoid confusion, the potential energy calculated here refers to the energy
due to the gravitational field that we want to counteract. How the vehicle achieves
this feat, or where it gets the energy to induce this effect, is not specified. At this
stage, estimating the energy requirements is a necessary prerequisite.

V. Hypothetical Mechanisms
With conservation of momentum and energy examined, the next step is to con-
sider hypothetical space drive mechanisms. These mechanisms are presented to
help identify the key issues needing resolution rather than to imply that any of
these are viable propulsion devices. To interpret these concepts at the level
that they are intended, consider this section as a brainstorming exercise. By
this it is meant that the concepts are intended to show that there are many differ-
ent ways to approach the challenge of creating a space drive and to provoke
deeper inquiry into the many unresolved science questions encountered while
considering them.
Ten different hypothetical space drive concepts are offered for this exercise.
Table 6 lists these compared with a cursory assessment of their applicability
for interacting with the known indigenous space phenomena (taken from
Table 1). Note that rough order of magnitude estimates are provided for the
mass densities of the phenomena. At this early stage it is sufficient to consider
only rough orders of magnitude. Also, many of the phenomena are not known
to a high degree of precision. Within Table 6, where a particular propulsion
concept intersects with a particular phenomenon, a cursory indication of the
applicability or viability is offered. In those cases where a particular option has
 150

Table 6 Hypothetical space drives and candidate phenomena

Candidate phenomena Cosmic

Mass density (kg/m3)a Galactic Dark Dark microwave Quantum Fields Fields
hydrogen Energy Matter background vacuum Spacetime due to due to
Hypothetical concept 10221 10226 10227 10231 1036+62 1021+26 charges masses

Interstellar jet Fig. 4 Db D D D TBD TBD N/A N/A

Differential sail Fig. 5a D TBD TBD D TBD TBD N/A N/A
Induction sail Fig. 5b D TBD TBD D TBD TBD N/A N/A
Diode sail Fig. 5c D TBD TBD D TBD TBD N/A N/A
Inertia modified rocket Fig. 1 N/A N/A N/A N/A N/A Relevant N/A Relevant
Oscillating inertia thruster Fig. 7 N/A N/A N/A N/A N/A Relevantc N/A Relevant
Diametric drive Fig. 9 N/A TBD TBD N/A N/A TBD D Relevantd
Disjunction drive Fig. 10 N/A TBD TBD N/A N/A N/A TBD TBD
M. G. MILLIS

Gradient potential drive Fig. 11 N/A TBD TBD N/A TBD TBD TBD TBD
Bias drive Fig. 12 N/A TBD TBD N/A TBD TBD TBD TBD

D ¼ “Doubtful ” The combination of this propulsion concept and phenomenon is cursorily assessed in this chapter and found to be of doubtful viability
TBD ¼ “To Be Determined ” The combination of this propulsion concept and phenomenon has not been sufficiently studied to reliably determine viability
N/A ¼ “Not Applicable ” No obvious way to apply this phenomena to the hypothetical device is recognized
a
From Table 1
b
From Refs 48–50
c
See Refs 35, 51, 52
d
See Refs 53–56
 PREREQUISITES FOR SPACE DRIVE SCIENCE 151

been studied, a reference is cited. The abbreviations used in the table are
explained below the table. Note that many possibilities remain unexplored.
The analyses of each of the 10 hypothetical mechanisms follow.

A. Hypothetical Interstellar Jet Propulsion

As a first example, consider the well-known Bussard interstellar ramjet
[48 50] illustrated in Fig. 4. As the vehicle moves forward, it scoops up the
indigenous interstellar hydrogen and channels it into a chamber where the hydro-
gen nuclei undergo fusion (while in motion) and then release the resulting energy
and helium nuclei out the rear of the spacecraft. In addition, some means is
required to initially set the spacecraft in motion, such as the method shown in
Fig. 4, where a laser sail is used for both the initial push and to add energy to
the spacecraft (via the concentrator). Even though there are profound
engineering difficulties with this concept [49,50,57], it serves as an example of
the idea for using indigenous matter in space as a reaction mass.
In more general terms, consider being able to interact with indigenous matter
or energy in space analogous to the way a jet engine captures and accelerates air
particles to produce thrust. This would require a sufficient amount of matter to
provide a meaningful reaction mass, and require a means to impart forces on
that matter.
Considering that tangible matter in space is scarce, a large volume of space
would have to be swept to obtain sufficient reaction mass to be practical. To
give some sense of the scale of this challenge, Table 7 shows the volume of
space that might contain a metric ton (MT) or 1000 kg of matter for the
various candidates of indigenous reaction mass. For completeness, equivalent

Fig. 4 Bussard ramjet with laser assist [57].

152 M. G. MILLIS

Table 7 Equivalent volumes to contain a metric ton (MT) of matter

Cosmic
Candidate Galactic Dark Dark microwave Quantum
phenomenon hydrogen energy matter background vacuum Spacetime
21 26 27 31 36+62 1+26
Equivalent mass 10 10 10 10 10 10
density (kg/m3)a
33 + 62
Volume required to 1024 1029 1030 1033 10 104+26
encompass
1 MT (m3)
a
See Table 1.

mass estimates for the indigenous forms of energy, using the familiar E mc 2
relation, are offered.
At this point most of these are impractical even before moving on to the issue
of how to induce forces against this matter. Again, as identified in Section III.B,
the majority of possibilities are with the unknowns of the quantum vacuum and
the nature of spacetime itself. Presently there are no known methods or proposed
concepts that apply the analogy of a “jet” to any phenomenon other than the
Bussard interstellar ramjet.

B. Hypothetical Sail Drives

The next theme from which to envision space drives is by entertaining the
possibility of collisions with some form of indigenous energy in space. To put
this into a more familiar context, recall the basic principle of a solar sail,
where energy (photons) from the Sun impinges on a sail, pushing it away from
the Sun. In the context of space drives, energy sources, like the cosmic micro-
wave background radiation, surround our spacecraft. To obtain a net thrust,
more momentum must be imparted to the rear of the spacecraft than to the front.
As an aside, the possible drag force from the cosmic microwave background
radiation on a laser-driven light sail was recently assessed in terms of a “terminal
velocity.” The terminal velocity is achieved when the drag force becomes large
enough to cancel the accelerating force from the laser. From this analysis, the
terminal velocity was calculated to be 0.99997 of the speed of light [58].
While momentum transfer from photons is familiar, the more general notion
of energy density imparting pressure is considered here. Again it is emphasized
that these are hypothetical examples to illustrate possibilities and issues, rather
than to assert that these are viable mechanisms. In that context, all of the indigen-
ous media listed in Table 1 are considered in terms of an energy density. Some
are already in that form, while the mass densities are converted using the familiar
E mc 2 relation, as shown in Table 8. From there, a rudimentary estimate of the
“pressure” follows directly from the energy density since the units of energy
density (J/m3) are identical to pressure (N/m2).
To illustrate possibilities for how to induce net forces from such surrounding
energy, three variations of hypothetical space drive sails are illustrated in
 PREREQUISITES FOR SPACE DRIVE SCIENCE 153

Table 8 Equivalent “pressure” from indigenous energy densities

Cosmic
Candidate Galactic Dark Dark microwave Quantum
phenomenon hydrogen energy matter background vacuum Spacetime
2 9 9 10 15 52+61
(N/m ) 10 10 10 10 10 1016+26
Equivalent
“pressure” from
energy densitya
5 5 6 11
(psi) (kPa) 10 10 10 10 1056+61 1020+26
(psi and kPa are
approximately
the same order of
magnitude)
a
See Table 1.

Fig. 5a 5c. In these illustrations, the rectangular box represents a cross-sectional

element of “sail” and the small arrows represent the impinging energy or photons.
The large arrows indicate the direction of acceleration. For introductory pur-
poses, simple force equations are provided within Fig. 5, where F represents
the net force, P represents the pressure of the medium acting on the device
(Table 8 values), and A represents the area of the device. The first terms on the
right of each equation are the rear surface pressures; the second terms are the
forward surface pressures. Their difference yields the net force on the sail,
whose equation is shown underneath. In the case of the induction sail
(Fig. 5b), the additional coefficient, d, represents the percentage by which the
energy impinging on the sail has been locally altered, reciprocally across the
front and back of the sail. A d greater than one implies an increase, less than
one a decrease, and equal to one implies no change.

a) Differential sail b) Induction sail c) Diode sail

Analogous to ideal radiometer vane Analogous to a real radiometer Analogous to one-way mirror

P 1 P
F =2
P P
− F = (d ) − F =2
P
−0
A A A d A A
(30) (31) (32)
P d2 −1 P P
F= F= F =2
A d A A

Fig. 5 Hypothetical sail drives. (Graphics courtesy of John MacNeil.)

154 M. G. MILLIS

1. Differential Sail
The first version, the differential sail (Fig. 5a), assumes that the photons
impinging on the rear are reflected (two units of momenta) while the ones on
the front are absorbed (imparting one unit of momentum), resulting in a net
photon pressure in the forward direction. This concept is analogous to an ideal
radiometer. Figure 6 shows an actual radiometer. In both the ideal and real situ-
ations, the photons striking the black side of the paddles are absorbed, while they
are reflected from the white side. The distinction between the ideal and real radio-
meters depends on whether air or vacuum surrounds these paddles. In the ideal
case, as with our differential sail, the paddles are in true vacuum so the forces
are only due to photon momenta. This would move the paddles in the direction
pointing from white to black, as reflection (white) imparts twice the momentum
than absorption (black).
In a real radiometer, some residual air is still present in the bulb. In this case,
the black side of the paddle absorbs more energy than the white, and then heats
the residual air more than on the white side. As the air molecules collide more
energetically on the black side than the white, the paddles are pushed by the
air in the direction from black to white. This is exactly opposite to the direction
than with the ideal radiometer.

Fig. 6 Crookes radiometer. (Courtesy of Timeline.)

 PREREQUISITES FOR SPACE DRIVE SCIENCE 155

Before proceeding to the next concept (which is similar to a real radiometer),

a critical limitation of this idea needs to be explained. Although the effects
described are entirely possible, they cease once the sail has reached thermal equi-
librium with the background. In short, the radiation absorbed will be re-emitted,
making the absorbing side, in effect, a delayed reflector. Once in thermal equili-
brium, both the front and the back undergo the same momentum transfer. A more
detailed explanation of this limitation is presented in Chapter 12, which deals
with using vacuum fluctuations as a propulsive medium.

2. Induction Sail
The operation of a real radiometer is analogous to the next hypothetical sail
drive concept, the induction sail. This concept assumes that the surrounding
medium has an energy density or pressure that can be locally altered in much
the same way that the real radiometer’s paddles affect the air temperature in
the vicinity of the paddles, raising the air temperature near the black side more
so than the white side. In Fig. 5b, this difference of the medium’s energy
density is graphically represented by the line density of the impinging rays. In
the hypothetical induction sail, the pressure of the medium has been altered by
a factor, d, behind the sail and by a reciprocal amount in front of the sail, resulting
in a net pressure difference across the sail of (d2 1)/d. Note that in the case of
d 1, which implies no change to the medium, the net force is zero, as expected.
This concept does not suffer the same equilibrium limitation as the differential
sail, since this induction sail requires that an energy flow be maintained. The sail
is not allowed to reach thermal equilibrium. The kinetic energy acquired by the
sail will be from the energy expended to maintain this nonequilibrium state,
minus conversion losses.

3. Diode Sail
The next hypothetical sail-drive, the diode sail (Fig. 5c), does not have the pro-
blems of removing any absorbed energy, at least in the ideal sense. It uses an
analogy to a diode, or one-way mirror, for its operation. The energy impinging
on the back is perfectly reflected, while the energy impinging on the front passes
through as if the sail were perfectly transparent. This concept is useful to introduce
another level of detail, specifically non-ideal conditions. The diode sail equations
shown in Fig. 5c represent perfect conditions. Consider, instead, if reflection,
absorption, or transparency were not perfect and coefficients had to be inserted
into the equation to account for this. In this case the equation would take the form:

P P
F ¼ 2ðkR Þ  ð1  kT Þ (33)
A A
where kR represents a reflectivity coefficient and kT represents a transparency coef-
ficient. Consider, for example, if both coefficients were only 50%, meaning that
only half of the energy impinging the back were reflected, only half from the
front passed through, and the rest of the impinging energy was absorbed. The per-
formance would only be a quarter of the ideal situation. At this point, the thermal
equilibrium issues raised with the differential sail apply equally to this concept too.
156 M. G. MILLIS

Absent of real sail devices, preliminary upper performance limits can be esti-
mated from the energy density of the various candidate indigenous energy
sources, as listed in Table 8. Consider first the performance in the context of
using the cosmic microwave background radiation. The energy density of this
medium equates to a pressure of only 10 15 N/m2. To put this in context, to
produce the same thrust as an ion engine, specifically around 100 mN (the
approximate maximum thrust of the NSTAR ion engine used for NASA’s
Deep Space 1), a perfect differential sail would need a surface area equivalent
to a square having sides of 10 million kilometers.
At the other extreme, consider interacting with the quantum vacuum energy.
Depending on the upper cutoff frequency used to calculate the energy of this
medium, greatly different values result. If the upper cutoff frequency were the
Compton frequency of a nucleon (1023 Hz), a sail of 1 square meter could
produce 1035 Newtons. As intriguing as this seems, it is important to stress
that these initial assessments are not definitive. While they help bracket the pos-
sibilities, the genuine limits of actual interactions remain to be discovered. One
example of a more thorough investigation into the possibilities of thrusting
against the quantum vacuum energy is given in see Chapter 12, which shows
that net thrust is possible in principle but likely to produce even less thrust per
power than a photon rocket.
As a closing note to this subsection on sail drive concepts: In 1991 a patent
was granted for a “spaceship to harness radiations in interstellar flights,” [59]
which summarized its operation thus:

The resultant force of radiations being absorbed from the rear and reflected
from the front propels the spacecraft forward.

Per the preceding discussions, the thrust direction is opposite to that expected
from its own description unless the device operates analogously to the induction
sail. But given the propeller shown at the front of the cylindrical spacecraft in
the figures of the patent, it is doubtful any rigorous analysis was conducted to
substantiate the claims.

C. Hypothetical Inertial Modifications

1. Inertia Modified Rocket
In Section IV, the notion of modifying the inertia of a rocket or its propellant
was raised as a space drive concept (Fig. 1) and can be included again here, in the
interest of completeness, as an example of hypothetical mechanisms. As already
shown, such notions evoke questions of energy conservation and the deeper
implications of the Equivalence Principle. In particular, the question of the
energy required to produce the desired change in inertia is both an issue and
an approach to deeper inquiries. (For a refresher of the possibilities, refer to
Tables 2 and 3.)
For the interested student, a logical next step is to calculate the energy and
power required to sustain such a hypothetical inertial modification as the
rocket thrusts.
 PREREQUISITES FOR SPACE DRIVE SCIENCE 157

2. Oscillatory Inertia Thruster

A completely different approach that entertains the implications of inertial
modification exists in Patent No. 5,280,864 [51], which is based on the transient
inertia observations reported by Woodward [35,52]. Chapter 11 examines an
experiment of Woodward in more detail. For this chapter, this concept is only
examined at an introductory level to illustrate the implications.
It should be emphasized that this concept is not in the category of the “non-
viable mechanical ‘antigravity’ devices” described in Chapter 6. Whereas those
concepts involve the oscillatory motion of masses, the focal distinction of the
Woodward approach is that the inertia of its component masses change, not
just their position and velocity. This key difference casts it in an entirely different
category.
In the Woodward concept, a device cyclically changes the distance between
two masses, while the inertia of each mass is oscillating about its nominal
mean, so that the system as a whole shifts its position to the right. One full
cycle is shown in Fig. 7, starting with the fully extended position and with
both masses at their nominal and equal value. (The operation is described in
the caption.)
A critical question behind this concept is, again, conservation of momentum.
Here the role played by the inertial frame is nontrivial. Because the inertia of the
masses vary, the role played by the inertial frame, in defining the position of the
center of mass, is no longer clear. For example, this concept evokes the question:
“Is inertia an intrinsic property of only matter, or is does it measure a relationship
between matter and spacetime?” As such, this concept is an excellent venue
through which to revisit deeper meanings of Mach’s principle.
The inventor of this concept evokes a fairly literal interpretation of Mach’s
principle, specifically that an inertial frame is created by, and connected to, the
surrounding matter in the Universe. Using this perspective, the reaction forces
are imparted to the surrounding mass of the Universe to conserve momentum.
This evokes the even deeper question: If the Universe was pushed a little to
the left, while the device moved to the right, what would be the background iner-
tial frame for the motion of this Universe/thruster interaction? Again, pursuing
such concepts introduces deep, provocative questions of basic physics. Even if
the Woodward devices or theories are later dismissed, this approach serves as
the foundation for thought experiments about the connection between inertial
matter and spacetime, the total matter in the Universe, and characteristics of
matter and spacetime that have not yet been discovered.
While the principles of this device are unfamiliar (especially the notion of
variations in inertia and the relation this has to a Machian inertial frame), there
is a somewhat analogous case that can at least illustrate the notion of using a
field for propellantless propulsion. Specifically, consider the concept shown in
Fig. 8 where a satellite uses tethers with masses at their tips for orbital maneuver-
ing [60]. By taking advantage of the nonlinear nature of a gravitational well, an
orbiting satellite can modify its orbit without expenditure of propellant. If the sat-
ellite extends a mass at the end of a tether toward Earth and another such mass
away from Earth, the imbalanced reactions will create a net force toward the
Earth. This is because the downward force on the near-Earth tether increases
158 M. G. MILLIS

Fig. 7 Inertia oscillation thruster. The first part of the cycle, (1) starts with the fully
extended position and with both masses at their nominal and equal value. Note the
bar graphs on each mass, whose arrow pointers indicate their current inertia (in
mid-position at stage 1). The Dm term refers to the difference between the forward
(right-side) mass and the rearward (left-side) mass. At the start of the cycle, the
masses are equal, so Dm is zero. As the device contracts through stages 1 5,
the forward inertial mass increases, peaking to a maximum value at stage 3 of the
cycle, and returning to its nominal value as the contraction is completed at stage 5.
During this time, the mass at the front is greater than the mass at the rear, so the
center position shifts forward. When the device begins to expand, the rearward
mass increases, while the forward mass decreases, again returning to their
nominal, equal values by the time the extension completes. During the extension,
the mass at the rear was higher, and the front lower, shifting the center of the
device again to the right.

more than the outward force on the outer tether as the tethers are deployed. By
alternately deploying and retracting long tethers at different points during the
orbit (apogee and perigee), an orbiting satellite can change its orbital altitude
or eccentricity. The result given is a purely classical result, which depends
only on the 1/r 2 force dependence of Newtonian gravity. It is an example of “pro-
pellantless” propulsion in that it allows orbital motion to be changed without
expenditure of reaction mass, yet it still satisfies Newtonian conservation of
 PREREQUISITES FOR SPACE DRIVE SCIENCE 159

Fig. 8 Taking advantage of nonlinearities: Tether orbital maneuvering [55].

momentum. Here the Earth is the reaction mass and its gravitational field is how it
is connected to the spacecraft. A somewhat different form of propulsion against a
gravitational field can be derived from General Relativity [61].

D. Hypothetical Field Drives

Gravitational fields accelerate masses and electric fields accelerate charges.
To entertain the analogy of using Newtonian field interactions for a space
drive, it is necessary to assume that there is some means for a vehicle to
induce a field around itself that will, in turn, accelerate itself. At first glance,
the expectation is that the induced forces would just act between the vehicle’s
field-inducing device and the rest of the vehicle, like trying to move a car by
pushing on it from the inside. In such cases all the forces act internally and
there would be no net motion of the vehicle. For reference, this issue can be
called the “net external force requirement.” In addition, field drives will also
have to satisfy conservation of momentum, which also appears difficult to
satisfy at first glance.
There is more than one way to explore possibilities of field drives to address
these requirements. Four “what-if” examples follow: the diametric drive (Fig. 9),
disjunction drive (Fig. 10), gradient potential drive (Fig. 11), and the bias drive
(Fig. 12). Each of these offers a different approach to convey the issues faced when
160 M. G. MILLIS

pondering space drives that are based on Newtonian fields. For completeness,
propulsion concepts that are based on Einstein’s geometric General Relativity
are covered in Chapters 4 and 15.

1. Diametric Drive
The first hypothetical field drive is based on the existing concept of negative
mass propulsion [53 56], the basic premise of which is extended here to more
general terms. Chapter 4 also examines this concept and provides a basic
diagram (see Chapter 4, Fig. 2).
Negative mass is a hypothetical entity where it is assumed that all the charac-
teristics of mass behave with opposite sign. The juxtaposition of negative and
positive mass leads to both masses accelerating in the same direction. Because
of their mutual characteristics, a repulsive force is created between the positive
and negative masses, but because the negative mass also has negative inertia,
it accelerates in the direction opposing the force. It has been shown that such
a scheme does not violate conservation of momentum or energy [54], and is
discussed further by Landis [55]. The crucial assumption to the success of this
concept is that negative mass has negative inertia.
Whereas Chapter 4 discusses the original concept of negative mass propul-
sion, more general terms are used here to set the stage for covering additional
possibilities and issues. Rather than just describing this concept as the interaction
of the two point masses, consider this concept in terms of a scalar potential, as
illustrated in Fig. 9, where only the x and y axes are plotted to allow the magni-
tude of the scalar potential to graphically project into the third visual dimension.

Fig. 9 Hypothetical diametric drive. Negative mass propulsion is shown here in a

more general form of a scalar potential created by positive and negative point
sources. (Singularities truncated for clarity.)
 PREREQUISITES FOR SPACE DRIVE SCIENCE 161

This visual analogy and the tools of vector analysis and scalar potentials make
it easier to consider various phenomena of spacetime and matter that exists within
spacetime. In the context of negative mass propulsion, Fig. 9 would represent a
gravitational scalar potential. It could equally represent an electric potential or
the density of any of the space phenomena previously discussed. The remaining
space drives presented here will all be examined in the context of scalar poten-
tials. It should be noted that scalar potentials are a Newtonian perspective
more so than a General Relativistic approach.
To proceed with this broader analogy, it is necessary to more explicitly dis-
tinguish factors that govern how the presence of sources affect the scalar potential,
and then how the scalar potential affects the matter. In the familiar cases of mass
and charge, the cause of a field is the same thing that reacts to a field. Masses create
gravitational fields that affect masses and charges create electric fields that affect
charges. To delve deeper, consider the analysis by Bondi in 1957 in which the
concept of negative mass propulsion was articulated [53]. To explore issues of
the Equivalence Principle and others, Bondi made distinctions between the
various properties of mass and then assessed the consequences of changes to
the signs of each characteristic. He found that all the characteristics must have
the same sign to avoid logical inconsistencies, but that either a positive or negative
sign could be assigned. Table 9 summarizes these various characteristics, the
defining equations, and the signs that define both positive and negative mass.
First, inertial mass is the characteristic that sets the defining relationship
between force and acceleration. The active gravitational mass is the one whose
mass creates the gravitational field. The last characteristic, the passive gravita-
tional mass, is the one that reacts to the presence of a gravitational field.
When considering other natural phenomena for their utility to space drives, it
is helpful to entertain such distinctions. For example, consider the applicability of
electrical fields in the context of this diametric drive. Although opposing charges
do exist, placing them next to each other will not result in a unidirectional accel-
eration. The distinction that renders the notion of the diametric drive incompati-
ble for electric fields is that all charges, regardless if they are positive or negative,
have positive inertia. The magnitude of a charge’s inertia is not directly tied to the
magnitude of its electrical charge.

Table 9 Dissecting the characteristics of mass

Representative Negative Normal

Characteristic equation mass mass

“Inertial mass,” mI (Defines the relation F mI , 0 mI . 0

mI
between force and acceleration) a
r
“Active gravitational mass,” mA mA F mA , 0 mA . 0
(Creates a gravitational field) G
F
“Passive gravitational mass,” mP mP mP , 0 mP . 0
(Responds to the presence of a rF
gravitational field)

F ¼ force, a ¼ acceleration, r ¼ distance from the mass, F ¼ gravitational scalar potential, and
G ¼ Newton’s constant.
162 M. G. MILLIS

To further pursue the general notion of a diametric drive in the context of other
phenomena such as Dark Matter, Dark Energy, or spacetime itself, the various
properties of these phenomena and their relation with spacetime would need
to be carefully distinguished and subsequently analyzed. No such definitions or ana-
lyses yet exist to this author’s knowledge.
To provide another example of separating the features of mass and then exam-
ining the consequences, the disjunction drive is offered next.

2. Disjunction Drive
Another spin-off from the idea of negative mass propulsion is the disjunction
drive. The disjunction drive entertains the possibility that the source of a gravita-
tional field, or the “active gravitational mass” as Bondi called it, mA, and that
which reacts to a field, which Bondi called the “passive gravitational mass,”
mP, can be physically separated. By displacing them across space, the passive
mass is shifted to a point where the field has a slope, thus producing reaction
forces on the passive mass.
Because we have physically separated the active and passive gravitational
masses, we must further distinguish separate inertial characteristics for each to
proceed to examine the consequences, where mIP is the inertial mass of the
Passive mass, and mIA is the inertial mass of the active mass.
Equation (34) and Fig. 10 illustrate the gravitational scalar potential that
results from physically separating the active and passive masses a distance, d,
apart. By displacing them in space, the passive mass is shifted to a point where

Fig. 10 Hypothetical disjunction drive. The active mass, mA, is the only source
affecting this gravitational potential, whereas passive mass, mP, is the only thing
affected by the gravitational potential. (Singularity truncated for clarity.)
 PREREQUISITES FOR SPACE DRIVE SCIENCE 163

the field has a slope, thus producing reaction forces between the source and the
reactant.
GmA
F¼p (34)
ðx  dÞ2 þy2 þ z2
where F is the gravitational scalar potential, G is Newton’s constant, mA is the
active mass that creates the gravitational field, d is distance separating the
location of the active and passive masses, and x, y, and z, are the usual rectilinear
coordinates. This scalar potential is graphically represented in Fig. 10, where
only the x and y axes are plotted to allow the potential, F, to graphically
project into the third visual dimension.
The force, F, on the passive mass, as defined from the gradient of the scalar
potential from the active mass, is found to be:
GmA
F ¼ mIP (35)
d2
where, in this case, a positive sign is assigned to denote the orientation of the
x-axis pointing in the direction from the passive mass toward the active mass.
Also, because the GmA/d 2 term represents the gravitational acceleration due
to the active mass, then the type of mass shown for the passive mass must be
its inertial mass, mIP, to satisfy the F ma form of this equation.
Next, imposing Newton’s action and reaction requires that this same force is
acting on the active mass, but in the opposite direction, and where again the inertial
mass property of the active mass, mIA, is the proper form to use in this example, and
where aA is the acceleration of the active mass corresponding to that force:

F ¼ mIA aA (36)

In this case, the negative sign denotes that its direction is opposite to the passive
mass’s acceleration. By imposing these conditions, momentum conservation is
satisfied. The resulting motion, however, will just bring the masses back together
because both are accelerating toward each other.
If, however, the distance separating them were fixed by using some hypothe-
tical rigid device of length, d, then the acceleration of both masses would have to
be the same. Since the passive mass does not present any gravitational force on
the active mass (and the active mass is defined as not having characteristics to
respond to a gravitational field), then both masses would accelerate due to the
gravitational force of the active mass on the passive mass only, yet tempered
by their inertia. In this case, the acceleration of the combined masses
(mIP þ mIA), and taking into account their inertia, would result in the following
net acceleration, a, of the system:
GmA GmA
aðmIP þ mIA Þ ¼ mIP  2 mIA
d2 d
GmA ðmIP  mIA Þ
a¼ 2 (37)
d ðmIP þ mIA Þ
164 M. G. MILLIS

Note that if the magnitudes of both the active and passive inertial masses have the
same value, then there would be no net acceleration.
There are a variety of possibilities that could be posited and analyzed based on
these introductory notions. For example, the implications of having different
magnitudes of the constituent mass characteristics could be explored, where
the issues of conservation of momentum becomes more complex. Additionally,
point sources other than just mass could be explored, such as electrical charges
and perhaps even provisional characteristics for Dark Matter or Dark Energy.
It is hoped that the disjunction drive example will provide future researchers
with starting points for how to explore these other options.

3. Gradient Potential Drive

The previous two field-drive conjectures dealt with point sources. This next
type of hypothetical field mechanism considers that somehow a small region of
space around the spacecraft is affected without the use of point sources, specifi-
cally where a localized slope in scalar potential is induced across a spacecraft that
then causes forces on the vehicle. It is not yet known if and how such an effect can
be created, but this brainstorming exercise is offered to provide some analytical
foundations for further inquiries and to address issues that arise.
The first obvious issue is the net external force requirement. In general, when
pondering the creation of fields to induce forces, the forces act between the field-
inducing device and the object that the field acts on. To illustrate this, consider
the following equation that represents a gravitational scalar potential for a space-
craft plus an imposed short-range gradient across the vehicle:
   
Gm 2
F¼ þ xAe r (38)
r
where the first parenthetical term is the contribution from the spacecraft’s
own mass, m, G is Newton’s gravitational constant, and r is the distance in
Cartesian coordinates; r (x 2 þ y 2 þ z 2)1/2. The final parenthetical term is the
hypothetical asymmetric effect imposed onto this scalar potential. This
induced gradient is represented by a magnitude, A, with a negative slope 2
in the
positive x direction, and is localized by a Gaussian distribution (e r ) centered
at the origin. This localizing equation was arbitrarily chosen for illustration
purposes only. Again, this example is only the foundation for future thought
experiments where such assumptions can be individually varied.
This scalar potential is shown as a surface plot over an x y plane in Fig. 11,
which is equal to the superposition of the potentials from the vehicle and the
induced local gradient.
By calculating the gradient of the scalar potential at the location of the vehicle,
specifically the derivative of F with respect to r of the induced pitch effect at
r 0, the acceleration for the vehicle is determined to be equal to A, and acts
in the positive x direction.
It is not clear where the reaction forces will be imparted. Consider the force
on the spacecraft from the asymmetric gradient: F mA. On what does the
 PREREQUISITES FOR SPACE DRIVE SCIENCE 165

Fig. 11 Hypothetical gradient potential drive. A localized gradient in the

gravitational potential along the x-axis is superimposed over that of the spacecraft.
(Singularity truncated for clarity.)

counterforce act? Is it the induced gradient itself, or the device that induced that
gradient? This option presents a point of departure for subsequent analysis.
The first and more common option is to consider that the forces act between
the field-inducing device (which is on the vehicle) and the vehicle, which does
not satisfy the net external force requirement. Hence, the force would be
acting within the components of the vehicle, with no resulting motion.
The second option to get around this situation is suppose that the effect that
created the gradient and the effect that this gradient has on the mass of the space-
craft are two, separate interactions. An easier example with which to illustrate
this option is the bias drive, described next.

4. Bias Drive
The final type of hypothetical field mechanism entertains the possibility that
the vehicle creates an asymmetry in the gravitational scalar potential, not by
imposing it directly, but rather by altering the properties of spacetime itself.
As a second step, this gravitational gradient then acts on the mass of the space-
craft in the usual manner, accelerating it in the positive x direction. Again, it must
be emphasized that this hypothetical notion is introduced to illustrate options and
issues rather than to assert that such features can be achieved.
To better understand this concept, consider the analogy of a “soap boat.” The
soap boat is a demonstration of the surface tension of water and how asymmetric
changes in that surface tension can propel a toy boat. The demonstration begins
with a toy boat floating in clean water. Then, a small quantity of detergent is
placed in the water behind the boat and immediately the boat is propelled
166 M. G. MILLIS

forward. The reaction mass is the water including the detergent. The force was
from difference in surface tension behind and in front of the boat.
To apply this analogy to our spacecraft, consider that Newton’s constant, G, is
analogous to the surface tension of the water and that some method has been dis-
covered to asymmetrically alter Newton’s gravitational constant over a small
region of space. As shown by both versions of Eq. (39) and in both versions
of Fig. 12, an asymmetric alteration of Newton’s gravitational constant will
induce a gradient in the scalar potential, F:
Multiplicative Modification
m  
r2
F¼ G 1 þ xBe (39a)
r
Exponential Modification
m  2

1þxbe r
F¼ G (39b)
r
where m is the mass of the spacecraft, G is Newton’s gravitational constant, r is
the distance in Cartesian coordinates; r (x 2 þ y 2 þ z 2)1/2, and both B and b
represent the magnitudes of two different hypothetical modifications of
Newton’s constant. In the version shown in Eq. (39a), G is multiplied by an
asymmetric modifier, similar to the version used in Fig. 11: Gradient potential
drive. The “þ1” identity term is necessary to return the Newtonian gravitational
potential to its original form when the propulsive effects2 are off (B 0), the x
provides the gradient, and the Gaussian distribution (e r ) limits the extension
of the effect. This localizing equation was arbitrarily chosen for illustration
purposes only. In the version shown in Eq. (39b), G is raised to the power of
the local asymmetric affect, of identical form, but now where B is replaced
with b to simply distinguish it from the prior version.

Fig. 12 Hypothetical bias drives. To asymmetrically distort gravitational potentials,

Newton’s gravitational constant, G, is altered to take on a locally confined gradient
along the x-axis, which then creates a gradient in the potential at the location of
the spacecraft. (Singularities truncated for clarity.)
 PREREQUISITES FOR SPACE DRIVE SCIENCE 167

It is interesting to note from the figure, that the asymmetry posited by the
exponential modifier [Eq. (39b) in Fig. 12b] has the characteristics of not allow-
ing the scalar potential to exceed zero. At this stage of exploration, it should again
be emphasized that these possibilities are introduced to demonstrate multiple
approaches as tools for thought experiments, rather than to suggest that these
are viable techniques.
When addressing the question of the reaction mass, it is now easier to suggest
an alternative perspective. Because Newton’s constant is considered a property of
space, rather than a consequence of the field created by the vehicle, it is easier to
consider space itself as the reaction mass.
One of the issues not yet raised with any of these propulsion thought exper-
iments is the time rate of change of a scalar potential when responding to
changes in the intensity or position of source phenomena. In electrostatics,
where net charges can be varied to include a null net value, the consequences
of this time delay are easier to study. This delay suggests that the field resulting
from a charge is not rigidly connected to that charge, but rather is a combined
function of the charge and the properties of the space. When considering the
propagation limits for changes in an electrostatic potential, the phenomena of
magnetism and the formalisms of Special Relativity come into play.
In the case of gravitational scalar potentials, where the source masses cannot
be nullified, considering notions of time-delayed potentials (also called “retarded
potentials”) is less familiar. Although there are theories exploring the time-rate-
of-change limitations of gravitational scalar potentials [62], the implications of
such approaches for space drives have not been explored.
To return to the question at hand, namely the source of reaction mass, consider
if the space itself is the reaction mass. As another analogy, consider the scalar
potential to represent air pressure. A gradient would represent a pressure differ-
ence where the air is the reaction mass. Next, consider the prior train of thought in
the context of Mach’s principle, specifically the literal interpretation that leads to
the possibility that there is a progenitor inertial frame caused by, and coincident
with, all the mass in the Universe. Does this mean that forces induced against
such a Machian frame might translate to the surrounding matter that creates
this frame? At this point these are just conjectures, but the thought experiments
offered by these hypothetical propulsion concepts provide yet another venue with
which to revisit such fundamental and still unresolved questions of physics. If
such explorations also lead to the discovery of a propulsive effect, all the better.

5. Closing Comments on Field Drives

The most critical issues when considering field drives are conservation of
momentum and the net-external force requirement. It is hoped that the examples
and discussions offered here can help future researchers avoid common mistakes
and realize how to further investigate the possibilities and issues. The role
played by spacetime is not trivial in these examples, nor are the distinct character-
istics that make up masses, charges, and any other phenomena to be considered. By
considering these various approaches and details, the requisite features needed to
provide a space drive can eventually be articulated. Although these questions
deal with the same data of nature, it casts them in a different context that can
168 M. G. MILLIS

illuminate possibilities that might otherwise be overlooked. The next section begins
to compile these unknowns into a problem statement to guide future research.

VI. Next Steps

With the indigenous reaction matter, energy considerations, and hypothetical
thrusting mechanisms introduced, this section compiles the critical questions and
unexplored options for space drives.

A. Problem Statement
Step 1 of the scientific method is to define the problem. To that end the critical
issues from the various space drive concepts have been compiled into the follow-
ing problem statement. Simply put, a space drive requires some controllable and
sustainable means to create asymmetric forces on the vehicle without expelling a
reaction mass, and some means to satisfy conservation laws in the process.
Regardless of which concept is explored, a mechanism must exist that can
affect a property, or the relationship between properties, of matter and spacetime
that satisfies these conditions:
Physics issues:
1) Satisfies conservation of momentum.
2) Satisfies conservation of energy.
3) Satisfies the net external force requirement, inducing a unidirectional
acceleration of the vehicle.
4) The effect must be sustainable as the vehicle moves.
5) The physics proposed for the propulsive mechanism, and for the properties
of matter and spacetime used for the propulsive effect, must be completely
consistent with empirical observations.
Practical issues:
6) The effect must be controllable (on/off, throttle-able, and directional).
7) The effect must be effective enough to propel the vehicle at practical
performance levels.
An important detail not elaborated on previously is item 4 in the problem
statement sustainability. Sustainability refers to the ability to continue the
propulsive effect throughout the vehicle’s motion. This implies that the force-
inducing effect must work in both an inertial frame and an accelerated frame.
This provides further mathematical constraints with which to scrutinize
approaches. It also requires that the force-producing field is carried along with,
or propagated with, the vehicle, or at least that the effect can be induced again
after the vehicle has been set in motion. This is yet another level of detail for
future space drive research to explore.

B. Continuing Research
Throughout this chapter, a variety of next steps are mentioned. These are com-
piled here for the reader’s convenience. Addressing any of these in more depth
 PREREQUISITES FOR SPACE DRIVE SCIENCE 169

would add to the overall progress in this field of study. This is not a comprehen-
sive list of possibilities, but rather just those raised in the course of this chapter.
1) Explore the anomalies of cosmology from the point of view of space-
flight, in particular by going deeper to revisit the progenitor observations that
lead to the working hypotheses of Dark Matter and Dark Energy.
2) Explore the connection between the quantum vacuum and spacetime
itself, from the propulsion point of view of reactive media.
3) Continue fundamental research into the nature of spacetime itself, con-
sidering the literal interpretation of Mach’s principle with its preferential inertial
frames, along with other analogues [62 64].
4) Assess the propulsive implications of the “optical analogies” and if those
can incorporate Mach’s principle and address the issues of preferred reference
frames.
5) Explore possible connections between Mach’s principle and the cosmic
microwave background radiation.
6) Repeat space drive energy analyses for relativistic conditions and with
different assumptions from those specified in Sec. IV.B.
7) Analyze space drives from the perspective of power calculations (energy
expended over time). This approach raises provocative questions from inte-
gration limits.
8) Assess sustained levitation using any of the analysis approaches
suggested earlier in Sec. IV.B.
9) From Table 6 entitled Hypothetical Space Drives and Candidate
Phenomena, pick any cell with the entry “TBD” (To Be Determined), and inves-
tigate the correlations further.
10) Repeat bias drive analysis, entertaining different localizing equations,
different hypothetical scalar potential characteristics, and dissecting the individ-
ual characteristics that define things that affect the scalar potential and those
things affected by the scalar potential, including time-rate-of-change factors for
the various characteristics. The intent is to look for actions with different prop-
erties than their reactions.

VII. Conclusions
To provide a framework for assessing emerging science in the context of
creating a space drive, a problem statement is now offered in this chapter. It is
based on considering the key objections to the notion of a space drive: 1) conser-
vation of momentum, 2) scarcity of indigenous reaction mass, and 3) inducing net
forces relative to a reaction mass. Various properties of space are described, with
most prospects residing with the least understood phenomena, specifically the
quantum vacuum, Dark Matter, Dark Energy, and spacetime itself, which
includes the fundamental unknown of the constituents of inertial frames. Of
these, both the quantum vacuum and spacetime itself appear to have the greatest
potential effective reaction mass density, specifically 1036+62 kg/m3 and
10 1+26 kg/m3, respectively. Note the wide uncertainty bands in those values,
reflecting how much is still unknown about these phenomena. As starting
points for future propulsion research, 10 different hypothetical propulsion
170 M. G. MILLIS

concepts are offered, including inertial modification, space sail concepts, and
field drive concepts. To address potential benefits and provide starting points
for continued analyses, energy calculations are offered in several different
contexts, with the most provocative being the case of indefinite levitation in a
gravitational field.
Evidence continues to grow that our Universe is more complex than is yet
explainable; it therefore may be prudent to consider more than one way to
look at this evidence. General Relativity accurately describes the motions of
matter at the scale of solar systems, and it can account for the spacetime
warping from the presence of extreme energy densities, but it is not yet able to
reconcile the anomalous rotation rates of galaxies (dubbed the Dark Matter
problem), the anomalous accelerated redshifts (dubbed the Dark Energy
problem), completely address Mach’s principle, or incorporate quantum mech-
anics. Although quantum theory accurately describes subatomic electromagnetic
energy transitions that cascade to macroscopic effects, it has not yet been con-
verted to also address gravity.
Curiosity-driven physics often seeks a “theory of everything,” one overarch-
ing theory to explain everything from the quantum level to cosmological
scales. Approaches include string theory, quantum gravity, and variations of
General Relativity to account for the new large-scale observations. As an alterna-
tive perspective that builds on the same physical observations and theoretical
foundations, this chapter suggests using the narrower goal of breakthrough pro-
pulsion as a problem around which to initiate deeper investigations. This intro-
duces different contexts for investigating Mach’s principle, the anomalous
rotation rates of galaxies, the anomalous distant redshifts, quantum fluctuations
of the vacuum, and the fundamental coupling of electromagnetism, inertia,
gravity, and spacetime.

Acknowledgments
The author is grateful to Ed Zampino, William Meyer, Geoff Landis, and Mike
LaPointe for discussions that helped distill the ambition of space drives into its
key scientific issues. Such individuals, those able to simultaneously entertain
imaginative departures and apply analytical rigor, are rare to find and a joy to
work with. Extra thanks are also due to William Meyer for his assistance with
the three-dimensional plots.

References
[1] Farmer, M., Science Fiction Citations for the Oxford English Dictionary, URL:
<http://www.jessesword.com/sf/list> [cited 4 May 2007].
[2] Millis, M. G., “Challenge to Create the Space Drive”, Journal of Propulsion and
Power, Vol. 13, No. 5, 1997, pp. 577 582.
[3] Bertolami, O., and Tajmar, M., Gravity Control and Possible Influence on Space
Propulsion: A Scientific Study. ESA CR (P) 4365, on Contract ESTEC 15464/01/
NL/Sfe, 2002.
[4] Hartle, J. B., Gravity; An Introduction to Einstein’s General Relativity, Addison
Wesley, San Francisco, CA, 2003.
 PREREQUISITES FOR SPACE DRIVE SCIENCE 171

[5] Carroll, B. W., and Ostlie, D. A., An Introduction to Modern Astrophysics, 2nd ed.,
Pearson Education, Saddle River, NJ, 2007.
[6] Robitaille, P. M., “WMAP: A Radiological Analysis,” Progress in Physics, Vol. 1,
2007, pp. 3 18.
[7] Spergel, D. N., Bean, R., Doré, O., Nolta, M. R., Bennett, C. L., Dunkley, J.,
Hinshaw, G., Jarosik, N., Komatsu, E., Page, L., Pereis, H. V., Verde, L., Halpern,
M., Hill, R. S., Kogut, A., Limon, M., Meyer, S. S., Odegard, N., Tucker, G. S.,
Weiland, J. L., Wollack, E., and Wright, E. L., “Three Year Wilkinson Microwave
Anisotropy Probe (WMAP) Observations: Implications for Cosmology,” The Astro
physical Journal Supplement Series, Vol. 170, No. 2, 2007, pp. 377 408.
[8] Riess, A. G., Kirshner, R. P., Schmidt, B. P., Jha, S., Challis, P., Garnavich, P. M.,
Esin, A. A., Carpenter, C., Grashius, R., Schild, R. E., Berlind, P. E., Huchra, J. P.,
Prosser, C. F., Falco, E. E., Benson, P. J., Briceño, C., Brown, W. R., Caldwell,
N., Dell’Antonio, I. P., Filippenko, A. V., Goodman, A. A., Grogin, N. A.,
Groner, T., Hughes, J. P., Green, P. J., Jansen, R. A., Kleyna, J. T., Luu, J. X.,
Macri, L. M., McLeod, B. A., McLeod, K. K., McNamara, B. R., McLean, B.,
Milone, A. A. E., Mohr, J. J., Moraru, D., Peng, C., Peters, J., Prestwich, A. H.,
Stanek, K. Z., Szentgyorgyi, A., and Zhao, P., “BVRI Light Curves for 22 Type Ia
Supernovae,” Astronomical Journal, Vol. 117, 1999, pp. 707 724.
[9] Goldhaber, G., Groom, D. E., Kim, A., Aldering, G., Astier, P., Conley, A., Deustua,
S. E., Ellis, R., Fabbro, S., Fruchter, A. S., Goobar, A., Hook, I., Irwin, M., Kim, M.,
Knop, R. A., Lidman, C., McMahon, R., Nugent, P. E., Pain, R., Panagia, N.,
Pennypacker, C. R., Perlmutter, S., Ruiz Lapuente, P., Schaefer, B., Walton, N. A.,
and York, T., “Timescale Stretch Parameterization of Type Ia Supernova B Band
Light Curves,” The Astrophysical Journal, Vol. 558, part 1, 2001, pp. 359 368.
[10] Huterer, D., and Turner, M. S., “Prospects for Probing the Dark Energy via
Supernova Distance Measurements,” Physical Review D, Vol. 60, No. 8, 1999,
pp. 081301 081306.
[11] Rubin, V. C., and Ford, W. K., Jr., “Rotation of the Andromeda Nebula from a Spec
troscopic Survey of Emission Regions,” Astrophysical Journal, Vol. 159, 1970, pp.
379 403.
[12] Kneib, J. P., Ellis, R. S., Smail, I., Couch, W. J., and Sharples, R. M., “Hubble Space
Telescope Observations of the Lensing Cluster Abell 2218,” Astrophysical Journal,
Vol. 471, 1996, pp. 643 656.
[13] Milgrom, M., “Dynamics with a Non Standard Inertia Acceleration Relation: An
Alternative to Dark Matter in Galactic Systems,” Annals of Physics, Vol. 229,
1994, pp. 384 415.
[14] Milgrom, M., “MOND Theoretical Aspects,” New Astronomy Reviews, Vol. 46,
2002, pp. 741 753.
[15] Rabounski, D., “The Relativistic Effect of the Deviation Between the CMB Tempera
tures Obtained by the COBE Satellite,” Progress in Physics, Vol. 1, 2007, pp. 24 26.
[16] Misner, C. W., Thorne, K. S., and Wheeler, J. A., Gravitation, W. H. Freeman & Co.,
New York, 1973.
[17] Rider, T. H., “Fundamental Constraints on Large Scale Antimatter Rocket Propul
sion,” Journal of Propulsion and Power, Vol. 13, No. 3, 1997, pp. 435 443.
[18] Taylor, E. F., and Wheeler, J. A., Spacetime Physics, W.H. Freeman and Co., San
Francisco, CA, 1966.
172 M. G. MILLIS

[19] Flynn, C., “On the Mass to Light Ratio of the Local Galactic Disc and the Optical
Luminosity of the Galaxy,” Monthly Notices of the Royal Astronomical Society,
Vol. 372, 2006, pp. 1149 1160.
[20] Cutnell, J. D., and Johnson, K. W. Physics, 3rd ed. Wiley, New York, 1995, p. 441.
[21] Eddington, A. S., Space, Time and Gravitation, Cambridge University Press,
Cambridge, U.K., 1920.
[22] de Felice, F., “On the Gravitational Field Acting as an Optical Medium,” General
Relativity and Gravitation, Vol. 2, 1971, pp. 347 357.
[23] Evans, J., Nandi, K. K., and Islam, A., “The Optical Mechanical Analogy in General
Relativity: Exact Newtonian Forms for the Equations of Motion of Particles and
Photons,” General Relativity and Gravitation, Vol. 28, 1996, pp. 413 439.
[24] Puthoff, H. E., “Polarizable Vacuum (PV) Approach to General Relativity,” Foun
dations of Physics, Vol. 32, 2002, pp. 927 943.
[25] Mach, E., The Science of Mechanics, 5th English ed., Open Court Publishing Co.,
London, 1942.
[26] Barbour, J. B., and Pfister, H. (eds), Mach’s Principle: From Newton’s Bucket to
Quantom Gravity (Einstein Studies), Birkhäuser, Boston, 1995.
[27] Weinberg, S., Gravitation and Cosmology: Principles and Applications of the
General Theory of Relativity, Wiley, New York, 1972.
[28] Bondi, H., and Samuel, J., “The Lense Thirring Effect and Mach’s Principle,”
Physics Letters A, Vol. 228, pp. 121 126.
[29] Ciufolini, I., and Wheeler, J. A., Gravitation and Inertia, Princeton Univ. Press,
Princeton, NJ, 1995.
[30] Symon, K. R., Mechanics, 3rd ed., Addison Wesley, Reading, MA, 1971.
[31] Foster, R. N., Innovation; The Attacker’s Advantage, Summit Books, New York,
1986.
[32] Seifert, H. (ed.), Space Technology, Wiley, New York, 1959.
[33] Berman, A. I., The Physical Principles of Astronautics: Fundamentals of Dynamical
Astronomy and Space Flight, Wiley, New York, 1961.
[34] Sternbach, R., and Okuda, M., Star Trek: The Next Generation Technical Manual,
Pocket Books, New York, 1991.
[35] Woodward, J. F., “Flux Capacitors and the Origin of Inertia,” Foundations of
Physics, Vol. 34, 2004, pp. 1475 1514.
[36] Millis, M. G., “Assessing Potential Propulsion Breakthroughs,” New Trends in Astro
dynamics and Applications, Belbruno, E., (ed.), Annals of the New York Academy of
Sciences, New York, Vol. 1065, 2005, pp. 441 461.
[37] Tajmar, M., and Bertolami, O., “Hypothetical Gravity Control and Possible Influence
on Space Propulsion,” Journal of Propulsion and Power, Vol. 21, No. 4, July Aug
2005, pp. 692 696.
[38] Podkletnov, E., and Nieminen, R. “A Possibility of Gravitational Force Shielding by
Bulk YBCO Superconductor,” Physica C, Vol. 203, 1992, pp. 441 444.
[39] Hathaway, G., Cleveland, B., and Bao, Y. “Gravity Modification Experiment Using a
Rotating Superconducting Disk and Radio Frequency Fields,” Physica C, Vol. 385,
2003, pp. 488 500.
[40] Resnick, R., and Halliday, D., Physics, 3rd ed., Wiley, New York, 1977.
[41] Boston, M., “Simplistic Propulsion Analysis of a Breakthrough Space Drive for
Voyager”, AIP Conference Proceedings No. 504, El Genk, M. S., (ed.), American
Institute of Physics, New York, 2000, pp. 1075 1078.
 PREREQUISITES FOR SPACE DRIVE SCIENCE 173

[42] Byers, D. C., “An Experimental Investigation of a High Voltage Electron Bombard
ment Ion Thruster,” Journal of the Electrochemical Society, Vol. 116, No. 1, 1969,
pp. 9 17.
[43] Forward, R. L., “Guidelines to Antigravity,” American Journal of Physics, Vol. 31,
1963, pp. 166 170.
[44] Alcubierre, M., “The Warp Drive: Hyper fast Travel Within General Relativity,”
Classical and Quantum Gravity, Vol. 11, 1994, p. L73.
[45] Morris, M. S., and Thorne, K. S., “Wormholes in Spacetime and Their Use for
Interstellar Travel: A Tool for Teaching General Relativity,” American Journal of
Physics, Vol. 56, 1988, pp. 395 412.
[46] Visser, M., Lorentzian Wormholes: From Einstein to Hawking, AIP Press,
New York, 1995.
[47] Krasnikov, S. V., “Hyperfast Interstellar Travel in General Relativity,” Physical
Review D, Vol. 57, 1998, p. 4760.
[48] Bussard, R. W, “Galactic Matter and Interstellar Flight,” Astronautica Acta, Vol. 6,
1960, pp. 179 194.
[49] Bond, A., “An Analysis of the Potential Performance of the Ram Augmented
Interstellar Rocket.” Journal of the British Interplanetary Society, Vol. 27, 1974,
pp. 674 688.
[50] Jackson, A., “Some Considerations on the Antimatter and Fusion Ram Augmented
Interstellar Rocket,” Journal of the British Interplanetary Society, Vol. 33, 1980,
pp. 117 120.
[51] Woodward, J. F., “Method for Transiently Altering the Mass of an Object to Facili
tate Their Transport or Change Their Stationary Apparent Weights,” US Patent #
5,280,864, 1994.
[52] Woodward, J. F., “A New Experimental Approach to Mach’s Principle and
Relativistic Gravitation,” Foundations of Physics Letters, Vol. 3, No. 5, 1990, pp.
497 506.
[53] Bondi, H., “Negative Mass in General Relativity,” Reviews of Modern Physics,
Vol. 29, No. 3, July 1957, pp. 423 428.
[54] Forward, R. L., “Negative Matter Propulsion,” Journal of Propulsion and Power,
Vol. 6, No. 1, Jan. Feb., 1990, pp. 28 37.
[55] Landis, G., “Comment on ‘Negative Mass Propulsion’,” Journal of Propulsion and
Power, Vol. 7, No. 2, 1991, p. 304.
[56] Winterberg, F., “On Negative Mass Propulsion,” 40th Congress of the International
Astronautical Federation, International Astronautical Federation Paper 89 668,
Malaga, Spain, Oct. 1989.
[57] Chew, G., Dolyle, M., and Stancati, M., Interstellar Spaceflight Primer, Final Report
on NASA Contract NASW 5067, Order 167, Science Applications, International
Corp., Schaumburg, IL, 2001.
[58] McInnes, C. R., and Brown, J. C., “Terminal Velocity of a Laser Driven Light Sail,”
Journal of Spacecraft, Vol. 27, No. 1, 1989, pp. 48 52.
[59] Carmouche, W. J., “Spaceship to Harness Radiations in Interstellar Flights,” US
Patent 5,058,833, 22 Oct 1991.
[60] Landis, G. A., “Reactionless Orbital Propulsion Using a Tether,” Acta Astronautica,
Vol. 26, No. 5, 1992, pp. 307 312; also as National Aeronautics and Space Admin
istration TM 101992, April 1989.
174 M. G. MILLIS

[61] Gueron, E., Maia, C. A. S., and Matsas, G. E. A., “ ‘Swimming’ Versus
‘Swinging’ Effects in Spacetime,” Physical Review D, Vol. 73, No. 2, id. 024020,
January 2006.
[62] Jefimenko, O. D., Gravitation and Cogravitation; Developing Newton’s Theory of
Gravitation to Its Physical and Mathematical Conclusions, Electret Scientific Co.,
West Virginia, 2006.
[63] Barceló, C., Liberati, S., and Visser, M., “Analogue Gravity,” Living Reviews Rela
tivity, Vol. 8, No. 12, [online article], URL: http://www.livingreviews.org/
lrr 2005 12 cited, 12 Sept. 2007.
[64] Winterberg, F. M., The Planck Aether Hypothesis: An Attempt For a Finitistic
Theory of Elementary Particles, Verlag Relativistichen Interpretationen, Karlsbad,
Germany, 2000.
 Chapter 4

Review of Gravity Control Within Newtonian and

General Relativistic Physics

Eric W. Davis
Institute for Advanced Studies at Austin, Austin, Texas

I. Introduction
RAVITY is a pervasive force that we are unable to affect. Unlike the
G electromagnetic force, for which an engineering prowess has been achieved,
gravity remains an immutable force. The scientific knowledge of gravity has
matured to where its behavior is well modeled, both from the Newtonian and
the more detailed general relativistic perspectives, but why it functions as
modeled is still not understood. With the accumulated knowledge, propulsion
devices can be engineered that overcome the force of gravity through indirect
means, but gravity itself cannot yet be modified. Aerodynamic machines fly in
air and rockets fly in the vacuum of space. We can even take advantage of
gravity, for example, with “gravity assists” for space maneuvers and for convert-
ing the Earth’s water cycle into electrical energy via hydroelectric dams, but here
again without any ability to change gravity itself.
Gravity is weak compared to the other fundamental forces in nature.
Considering the forces between an electron and a proton, the gravitational
force is 40 orders of magnitude weaker than the electric force. But near the
surface of the Earth, where 1024 kg of gravitational mass lies underneath, the
accumulated strength is significant. On the surface of the Earth, gravity is
strong enough to accelerate objects downward at 9.81 m/s2. Considering the
relative weakness of gravity compared to more controllable phenomena, it
seems reasonable to consider that gravity might one day become a phenomenon
that we can engineer.

Copyright # 2008 by the American Institute of Aeronautics and Astronautics, Inc. All rights
reserved.

Senior Research Physicist.

175
176 E. W. DAVIS

A colloquial expression for the goal of affecting gravity is “antigravity.” This term
specifically means the negation or repulsion of the force of gravity. A more general
term that encompasses this notion and other possibilities is “gravity control.”
This chapter examines a variety of approaches toward the goal of controlling
gravity. Both Newtonian and general relativistic perspectives will be explored.
The reader should bear in mind that most of these concepts are nowhere near
having any form of practicable engineering implementation. The intent is to fam-
iliarize the audience with the approaches that already exist, explain their limits
and unknowns, and then to identify lingering research objectives that are relevant
to the quest of controlling gravity.
The approaches covered include using Newton’s Law of Gravity to simply
nullify the gravity field of one body acting on another body by using a clever
arrangement of masses. Also, the theoretical possibility of antigravity appears
in quantum gravity theories, cosmological vacuum (or dark) energy theory,
and quantum field theory. We will review additional topics of gravity control
that include the production of artificial gravity via ultrahigh-intensity electric
or magnetic fields, the production and use of gravitational waves for rocket pro-
pulsion, antigravity (self-lifting) forces induced by the Casimir effect or by non-
retarded quantum interatomic dispersion forces in a curved spacetime (i.e., in a
background gravitational field), and speculative quantum unified field theories
that predict unusual new antigravity forces.

II. Gravity Control Within Newtonian Physics

The basic form of Newton’s Law of Gravity is given by the standard expression for
the gravitational force (Fgrav) that mutually acts between two masses [1]:

Gm1 m2
Fgrav ¼  (1)
r2

where the negative sign indicates that Fgrav is a mutual force of attraction, G is
Newton’s universal gravitation constant (6.673  10 11 Nm2/kg2), m1 and m2 are
two interacting masses, and r is the radial distance between the two masses (note:
we adopt MKS units throughout, unless otherwise indicated). We observe in
Eq. (1) that the force of gravity acting on a small test mass becomes stronger when
the other (gravitating) mass is larger in magnitude or when the distance between
them is very small, or both. Also recall that we can use Eq. (1) and Newton’s
Second Law of Motion (F ma) to define the magnitude of the gravitational accele-
ration ag that acts on a small test mass m due to a larger (gravitating) mass M [1]:

GM
ag ¼ (2)
r2

If we choose Earth to be the larger gravitating mass so that M M (5.98  1024 kg),
then according to Eq. (2), a small test mass m placed near the Earth’s surface,
whereby r  R (6.378  106 m), will experience a downward gravitational accel-
eration of ag ; g 9.81 m/s2.
 REVIEW OF GRAVITY CONTROL 177

A. Negating Newtonian Gravity the Hard Way

It is possible to design an antigravity machine that can nullify Earth’s gravity
field using Newton’s Law of Gravity. One way to use Eq. (1) to nullify the Earth’s
gravitational pull at a particular location would be to locate another planet of
equal mass above that location [2,3]. The forces from the two Earth masses
will cancel each other out over a broad region between them. Everything
within this broad region will be in free fall. This is not a practical solution for
aerospace flight, however, because we have no way to manipulate and control
another planetary-sized body.
Along similar lines, Forward [2,3] considered using a ball of ultradense
compact matter, corresponding to dwarf star or neutron star matter (1011 to
1018 kg/m3), having a diameter of 32 cm and a mass of 4  106 tons. This ultra-
dense ball will have a surface gravitational (attractive) force of 1-g. We could
then place this small, ultradense ball near the surface of the Earth and its 1-g
gravity field will cancel the Earth’s 1-g gravity field. All test objects placed in
the broad region between the small ultradense ball and the Earth will thus be
in free fall. Another option Forward [2 5] suggested would be to shape the
compact ultradense matter into a disk that is 45 cm in diameter and 10 cm
thick, with the same mass and density as the small ultradense ball. Its gravita-
tional acceleration is ag 4Grt, where r is the mass density of the disk and t
is its thickness. In this case, the disk will have a force of gravitational attraction
that is the same on both sides, and it will be uniform near the center of the disk
where the strength of the gravitational force will be 1-g. If this disk were to be
placed very close to and above the Earth’s surface, then there will be a gravita-
tional force of 2-g above the disk (equal to 1-g due to the Earth’s gravity field plus
1-g due to the top-side gravity field of the disk) while underneath the disk near its
center, there will be a gravity-free (or free fall) region because the Earth’s gravity
field underneath is canceled by the gravity field of the disk’s bottom side. While
these are interesting antigravity machines, they are unfortunately not feasible
from an engineering standpoint because we do not yet have the technology or
means to create and handle ultradense compact matter.

B. Approaches for Orbital Stability

As another approach, Fig. 1 shows a Six-Mass Compensator designed by
Forward [6]. This device approximately nullifies the Earth’s residual gravita-
tional tidal forces that act upon all orbiting bodies such as satellites and space
stations. In a free-fall orbit around the Earth, the overall downward gravitational
acceleration is canceled to first-order by the orbital motion alone. However, there
are residual tidal accelerations (or forces) that are very weak (micro-g). These
tidal forces are not exactly zero because the free-fall motion of orbiting bodies
does not entirely cancel the Earth’s gravity field.† Such weak tidal forces will

†
The only part of a body that is under absolutely zero net gravity force in a free fall orbit is its
center of mass. Those parts of a body located some distance away from the center of mass will experi
ence tidal forces due to gravity gradients (or differential acceleration). Residual tides arise when the
distant parts of an orbiting body follow their own orbits that are either higher (lower velocity), lower
(higher velocity), or in a plane tangent to the Earth with respect to the orbit of the center of mass. The
result is outward vertical and inward horizontal tidal forces.
178 E. W. DAVIS

Fig. 1 Orbital tidal patterns of Earth and the Six-Mass Compensator [6].

have a measurable impact upon sensitive lab experiments or manufacturing pro-

cesses that require an absolute 0-g environment. And the residual tidal forces will
grow larger in magnitude as an orbiting body grows larger in size and/or as the
size of 0-g sensitive experimental or manufacturing devices grow larger.
Forward’s Six-Mass Compensator is designed to reduce the weak horizontal
and vertical tidal forces by a factor 106. The magnitude and sign of the horizon-
tal and vertical tidal accelerations (a GM‘/R 3) in orbit are illustrated at the top
of Fig. 1, where R is the orbital distance ( R þ h, where h is the orbital altitude)
from the center of Earth (mass M M) and ‘ is the distance from the center of
mass of an arbitrary point on the orbiting body. We observe that the outward tidal
acceleration in the vertical direction is twice as strong as the uniform inward tidal
acceleration (compression) in the horizontal direction. Therefore, the compensa-
tor illustrated at the bottom of Fig. 1 generates forces that approximately cancel
 REVIEW OF GRAVITY CONTROL 179

the residual tidal forces. The compensator is comprised of six equal mass (m)
spheres arranged in a ring of radius r. The ring encloses the region to be compen-
sated and the magnitude of the compensating acceleration is a 3Gm‘/r 3. A
solid ring or any number of spherical masses greater than three can be used
instead, but Forward estimated that six spherical masses is optimum. Because
the compensator is nullifying only the weak residual orbital tidal forces, the com-
pensator spheres can be made of ordinary density matter (e.g., tungsten or lead)
having a mass of 100 kg and size of 20 cm. The plane of the ring compensator
must always be oriented tangent to the surface of the Earth below. The gravita-
tional tidal pattern of the ring compensator can be fine tuned to match the tidal
pattern of the Earth at any altitude by simply adjusting the radius and tilt of
the ring compensator. Because the Earth’s gravitational tidal forces grow stron-
ger at lower altitudes, the region to be compensated will grow smaller, while the
compensation region will grow larger at higher altitudes because the tidal forces
will grow weaker.
Unfortunately, this particular antigravity machine offers no propulsion benefit;
therefore, we will not consider it further.

C. Newtonian Levitation Energy Estimate

An ideal propulsion breakthrough could take the form of the antigravity-based
levitation of an aerospace vehicle within the Earth’s atmosphere. Rockets like the
Air Force DC-XA can hover above the ground for a time limited by the amount of
available rocket fuel [7]. An ideal antigravity propulsion device would allow for
the indefinite levitation of a vehicle above the Earth’s surface. It is illustrative to
estimate the energy required to levitate a 1-kg test mass above the Earth’s
surface. This will help quantify a potentially key engineering parameter for
such a levitation system. Millis [8] provides a generic estimate by considering
the amount of energy (E lev) required to nullify the magnitude of the Earth’s
gravitational potential energy for a test mass m hovering at height h above the
Earth’s surface:

GM m
E lev ¼ (3)
h
Equation (3) can also be derived by calculating how much energy is required to
completely remove a test mass from the Earth’s surface to infinity. Millis points
out that this calculation is more in line with the analogy to nullify the effect of
gravitational energy. Equation (3) also represents the energy required to stop a
test mass at the levitation distance h if it were falling in from infinity with zero
initial velocity.
Setting h  R and m 1 kg in Eq. (3), we get E lev 62.5 MJ. This is 2.05
times the kinetic energy required to put the test mass into low Earth orbit (LEO).
However, this estimate for E lev will require some adjustment depending on the
type of theory and its technological implementation because the operational ener-
getics of a putative antigravity propulsion system must be considered in conjunc-
tion with E lev.
180 E. W. DAVIS

D. Negative Matter Propulsion

Negative matter is a hypothetical form of matter whose active and passive
gravitational, inertial, and rest masses are opposite in sign to ordinary (positive)
matter. Thus, negative matter has negative mass. Negative matter is not to be con-
fused with antimatter, because antimatter is a form of positive matter having posi-
tive mass and quantum properties (i.e., spin and electric charge) that are opposite
to its positive matter counterparts. Forward [9,10] and others [11 15] studied the
physics of negative matter within the context of both Einstein’s Special and
General Theories of Relativity, and discovered many unusual properties that
could provide a form of breakthrough propulsion. These properties are given
below [9,10,14].
1) An object with negative mass gravitationally repels all other types of
matter, both positive and negative.‡
2) The gravitational coupling between a negative mass and a positive mass
does not violate the conservation of momentum, because negative matter in
motion has negative momentum.
3) The gravitational coupling between a negative mass and a positive mass
does not violate the conservation of energy, because negative matter in motion
has negative kinetic energy.
4) The total energy required to create negative matter, along with an
equal amount of positive matter, is zero, because negative matter has negative
rest-energy.
The consequences of these four properties upon a hypothetical negative matter
antigravity propulsion system are elucidated in the following sections.

1. Concepts for Negative Matter Propulsion

We can establish the fundamental basis for negative mass propulsion via a
thought experiment that demonstrates the negative matter property in item 1
above. If we take a ball of negative matter of mass ( M) that is initially at
rest, and place it near a rocket of positive equal mass (þM) that is also initially
at rest, then the gravitational field of the ball will repel the rocket, while the grav-
itational field of the rocket will attract the ball. The result of this unique gravita-
tional coupling is that the rocket and the ball will move off in the same direction
with an acceleration that is proportional to the force of gravity between them.§
Figure 2 illustrates this effect. This mechanism appears to provide an unlimited
amount of unidirectional acceleration without requiring either a reaction mass
or an energy source (see also, Ref. 15).
Forward [9,10] also provided a Newtonian analysis for the case when the
magnitude of M and þM are unequal. In this case, a consideration of the
motion of the center of mass of the combined system becomes important. So,

‡
Positive (ordinary) mass attracts all types of matter, including negative matter.
§
Otherwise the ball would violate the principle of equivalence which demands that all objects,
without exception, acquire the same acceleration in the same gravitational field.
 REVIEW OF GRAVITY CONTROL 181

Fig. 2 Negative matter propulsion [9,10].

in addition to the unidirectional acceleration of the combined system, a decrease

or increase in the separation of the two objects is superimposed. In a general rela-
tivistic analysis, Bonnor and Swaminarayan [12] showed that in the case of a uni-
formly accelerating pair of gravitationally interacting bodies with opposite
masses, the two masses are not quite equal and opposite because they are
being measured in an accelerating reference frame. For the two masses to keep
a constant separation in the accelerated reference frame, the negative mass
should be very slightly larger in magnitude than the positive mass, depending
on the initial separation.
The case of physically coupling a particle of negative matter and a (positive
matter) rocket together using a stiff spring was analyzed by Forward [9,10].
He designed a propulsion system that has the negative matter particle, of a
mass equal in magnitude to the entire rocket, located in an “engine room.” In
order to move the rocket forward, one must pull a spring attached to the back
wall of the engine room and hook it to the negative matter particle. The sub-
sequent inertial reaction of the negative matter particle will cause it to accelerate
in the forward direction, thus pulling the rocket forward with an acceleration that
is proportional to the strength of the spring. One would unhook the spring in order
to stop accelerating and, in order to decelerate the rocket to a stop, one would
replace the spring attached to the back wall of the engine room with a spring
attached to the front (or forward) wall. Forward also proved that this propulsion
system does not violate the laws of conservation of momentum and energy even
when the magnitudes of the two masses are not equal.}
Finally, Forward [9,10] also considered the case of physically coupling a par-
ticle of negative matter and a (positive matter) rocket together using electrostatic
forces. It is assumed in this case that negative matter comes in an electrically
charged variety, whereby the electric charge acts like a “handle” that can be
used to push or pull the negative matter particle at a distance using electrostatic
forces. It was shown that in both cases when the charges are both equal and oppo-
site, the same results of unidirectional motion of the combined system occurred at
an acceleration that is proportional to the strength of the electrostatic force.
Again, there was no violation of the laws of conservation of momentum and
energy even when the magnitudes of the two masses are not equal.

}
It is the elasticity of the spring that keeps the conservation laws from being violated, because the
spring also applies forces to the two objects.
182 E. W. DAVIS

2. Consideration of the Conservation Laws

Items 2 through 4 in the previous list appear to be controversial. In this section
we shall examine each of them in turn within the context of Newtonian mech-
anics. It would appear that negative matter propulsion automatically violates
both of the conservation laws of momentum and energy; however, it can be
proved that this is not so. In the following it is assumed that the magnitudes of
the positive mass and the negative mass are equal, but the results remain
unchanged even in the case when they are unequal. When the ball of negative
matter and the (positive matter) rocket were both initially at rest (i.e., zero vel-
ocity), their individual momentum is zero, and so their total initial momentum
is also zero. But Pwhen the two objects reach some final velocity n, their total
final momentum P is still zero, because:
X
P ¼ Pþ þ P

¼ ( þ M)n þ (  M)n (4)

¼0

where Pþ is the final momentum of the rocket and P is the final momentum of
the ball. Equation (4) shows that the negative (final) momentum of the ball
cancels out the positive (final) momentum of the rocket, and so momentum is
conserved. In addition, when the ball and the rocket are both initially at rest,
their individual kinetic energy is zero, and so their total initial kinetic energy
P After the two objects reach final velocity n, their total final kinetic
is also zero.
energy KE is still zero, because:
X
KE ¼ KE þ þ KE
1 1
¼ (þ M)n2 þ ( M)v2 (5)
2 2
¼0

where KE þ is the final kinetic energy of the rocket and KE is the final kinetic
energy of the ball. Equation (5) shows that the negative (final) kinetic energy of
the ball cancels out the positive (final) kinetic energy of the rocket, and so kinetic
energy is also conserved.
Finally, if we could somehow produce negative matter in order to exploit its
antigravity propulsion properties, then we must produce it along with equal
amounts of positive matter in order to conserve energy. This is exactly what is
required for the production of antimatter even though these two types of
matter are completely dissimilar. Antimatter production always occurs via the
production of particles and antiparticles in pairs. And so, negative matter pro-
duction must also invoke the production of negative matter particle positive
matter particle pairs. However, unlike the production of antimatter that requires
2mc 2 (c is the speed of light, 3.0  108 m/s) of combined rest-energy to produce
each particle antiparticle pair (each particle of mass, m), the energy cost for
 REVIEW OF GRAVITY CONTROL 183

producing a negative matter particle positive matter particle pair is (þM)c 2 þ

( M)c 2 0. This surprising counterintuitive result is possible because the nega-
tive rest-energy of the negative matter cancels the positive rest-energy of the
equal quantity of positive matter.
Forward [9,10] pointed out that negative rest-energy, negative momentum,
and negative kinetic energy are not standard concepts within Newtonian mech-
anics, but they do not appear to lead to any logical contradictions because they
are not (logically) forbidden by Special or General Relativity Theory. If negative
matter is forbidden, then the proof must come from some other theory of physics
(e.g., quantum field theory, quantum superstring theory, etc.).

3. Make or Break Issues: Negative Matter Propulsion

Can we produce negative matter in the lab and/or does it naturally exist some-
where in the universe? The answer to the first part of this question is no or not yet.
The answer to the second part is we do not know. One property that negative
matter has in common with antimatter is that, if it does exist naturally somewhere
in the universe, then it is just as scarce as antimatter. The cosmological abun-
dance of antimatter is negligible. This conclusion is partly based on the observed
minuscule amount of antimatter in cosmic rays. All of the antimatter present in
cosmic rays can be accounted for by radioactive decay processes or by nuclear
reactions involving ordinary matter. We also do not observe the signatures of
electron positron annihilation, or proton antiproton annihilation coming from
the edges of galaxies, or from places where two galaxies are near each other.
As a result, we believe that essentially all of the objects we see in the universe
are made of matter and not antimatter. Yet we can produce copious amounts
of antimatter in particle accelerator devices. Nearly the entire abundance of anti-
matter that was created during the Big Bang had been annihilated away prior to or
during cosmological nucleosynthesis. Likewise, negative matter could have been
produced along with equal amounts of matter during the Big Bang, but we can
speculate that some underlying universal quantum symmetry property was
broken as the universe cooled, thus causing the negative matter particles to be
destroyed by some unknown particle interaction processes. This may account
for the absence of naturally occurring negative matter.
Forward [9,10] speculated that the large voids existing between galaxies and
superclusters of galaxies are regions of negative matter. Because negative matter
repels itself, it cannot gravitationally bind into larger clumps of negative matter.
However, because Coulomb electrostatic forces are enormously stronger than
gravitational forces, electrically charged bulk negative matter attracts itself and
does so much more strongly than ordinary matter. Thus, negative matter can
become electrostatically bound into larger clumps [15]. This is because for nega-
tive matter, gravity and electric charge exchange roles: The “gravitational
charge” of negative matter gravitationally repels like-gravitationally charged
particles, whereas electrostatic forces will attract electrically charged negative
matter. So we could then imagine that the large extragalactic voids throughout the


The gravitational charge eg of a mass M is customarily defined as [14]: eg ; (2G)1/2M.
184 E. W. DAVIS

universe are places where negative matter has collected into huge electrically
bound clumps. This could explain the mystery of dark matter [16 18].
However, dark matter is not likely to be composed of negative matter because
dark matter gravitationally attracts all other matter (including other dark
matter), and the total cosmological energy budget requires that dark matter has
a positive rest-energy density (and positive pressure).
However, there is nothing to forbid the existence or artificial creation of nega-
tive matter. No logical contradiction proofs have succeeded in forbidding it. The
most serious known objection to negative matter is the one based on the Principle
of Causality or, equivalently, on the basis of the Second Law of Thermodyn-
amics. Terletskii [14] correctly points out that all other objections can be
reduced to the Principle of Causality or they can be shown to be linked with it
to some degree. Further, because it is recognized that the Principle of Causality
is a naturally apparent expression of the Second Law of Thermodynamics, then
the objections to negative matter lose all physical justification. Negative matter
should exist on an equal par with antimatter and all other exotic elementary
quantum matter (artificially created in accelerators) because Murray Gell-
Mann’s Totalitarian Principle states that in physics “anything which is not
prohibited is compulsory” [19]. Furthermore, exotic matter (e.g., antimatter,
negative matter, superluminal matter, etc.) is in no way precluded by either rela-
tivistic quantum field theory or Einstein’s two relativity theories because it is
these very theories that suggest their possibility. For example, C. D. Anderson’s
1932 discovery of anti-electrons (or positrons) in cosmic rays confirmed Dirac’s
1928 relativistic quantum mechanics prediction of antimatter.

III. Gravity Control Within General Relativity

In the sections that follow we describe and summarize the known types of anti-
gravity that can be derived from Einstein’s General Theory of Relativity.

A. Antigravity via Gravitomagnetic Forces

1. Gravitomagnetic Foundations
Heaviside [20] (in 1883), Einstein (prior to the 1915 publication of his General
Theory of Relativity), Thirring [21,22], and Thirring and Lense [23] showed that
General Relativity Theory provides a number of ways to generate non-Newtonian
gravitational forces via the splitting of gravitation into electric and magnetic field
type components [24]. These forces can be used to counteract the Earth’s grav-
itational field, thus acting as a form of antigravity. General Relativity Theory pre-
dicts that a moving source of mass-energy can create forces on a test body that are
similar to the usual centrifugal and Coriolis forces, although much smaller in
magnitude. These forces create accelerations on a test body that are independent
of the mass of the test body, and the forces are indistinguishable from the usual
Newtonian gravitational force. We can counteract the Earth’s gravitational field
by generating these forces in an upward direction at some spot on the Earth.
Forward [25] linearized Einstein’s general relativistic field equation and
developed a set of dynamic gravitational field relations similar to Maxwell’s
electromagnetic field relations. The resulting linearized gravitational field
 REVIEW OF GRAVITY CONTROL 185

relations are a version of Newton’s Law of Gravitation that obeys Special Rela-
tivity. The linearized gravitational field relations show that there is a unique cor-
respondence between the gravitational field and the electric field. For example,
the Newtonian gravitational field of an isolated mass is the gravitational
analog to the electric field of an isolated electric charge.
Likewise, there is an analogy to a magnetic field contained within the linear-
ized gravitational field relations. In Maxwellian electrodynamics, a magnetic
field is due to the flow of an electric charge or an electric current. In other
words, the electric field surrounding an electric charge in motion will appear
as a magnetic field to stationary observers. If the observers move along with
the charge, they see no relative motion, and so they will only observe the
charge’s electric field. Thus, the magnetic field is simply an electric field that
is looked at in a moving frame of reference. In an analogous fashion, the linear-
ized gravitational field relations show that if a (gravitational) mass is set into
motion and forms a mass current, then a new type of gravitational field is
created that has no source and no sink. This is called the Lense Thirring
effect, or rotational frame dragging effect, in which rotating bodies literally
drag spacetime around themselves.

2. Forward’s Dipole Gravitational Field Generator

Forward [26,27] used the linearized gravitational field relations plus aspects of
the Lense-Thirring effect to develop models for generating antigravity forces.
One example of an antigravity generator is based on a system of accelerated
masses whose mass flow can be approximated by the electrical current flow in
a wire-wound torus. According to Maxwellian electrodynamics, an electric
current flowing through a wire that is wrapped around a torus (or ring) causes
a magnetic field to form inside the torus. If the current I in the wire increases
with time, then the magnetic field, B, inside the torus also increases with time.
This time-varying magnetic field in turn creates a dipole electric field, E, as
shown in Fig. 3. The magnitude of the electric field at the center of the torus is
given by E (m0Nr 2İ)(4pR2t ) 1, where m0 is the vacuum permeability con-
7
stant (4p  10 H/m), N is the total number of turns of wire wound around

Fig. 3 Dipole electric field generator [27].

186 E. W. DAVIS

Fig. 4 Dipole gravitational field generator [27].

the torus, İ is the time rate-of-change of the electric current flowing through the
wire, r is the radius of one of the loops of wire, and Rt is the radius of the torus.
In a similar fashion, Forward’s antigravity device is a dipole gravitational field
generator. As shown in Fig. 4, a mass flow, T, through a pipe wound around a
torus induces a Lense Thirring field, P, to form inside the torus. If the mass
flow is accelerated, then the P-field increases with time, and thus a dipole grav-
itational field, G, is created. The magnitude of the antigravitational field at
the center of the torus is given by G (h0 Nr 2Ṫ)(4pR2t ) 1, where h0 is the
vacuum “gravitational permeability” constant (;16pG/c 2 3.73  10 26 m/
kg),†† N is the total number of turns of pipe wound around the torus, Ṫ is the
time rate-of-change of the mass current flowing through the pipe, r is the
radius of one of the loops of pipe, and Rt is the radius of the torus [27]. One
should note the striking similarity between the two equations for the dipole elec-
tric and dipole gravitational fields.
Using the above equation for G, Forward [26,27] showed that we would need
to accelerate matter with the density of a dwarf star through pipes as wide as a
football field wound around a torus with kilometer dimensions in order to
produce an antigravity field (at the center of the torus) of G  10 10 aacc,
where aacc is the acceleration of the dwarf-star-density matter through the
pipes. The tiny factor 10 10 is composed of the even smaller h0, which is the
reason why very large devices are required to obtain even a measurable
amount of acceleration. To counteract the Earth’s gravitational field of 1-g
requires an antigravity field of 1-g (vectored upward), and thus the dwarf-star-
density material within the pipes must achieve aacc 1011 m/s2 in order to
accomplish this effect.
Forward [5] also identified a configuration comprised of a rotating torus
of dense matter that turns inside out like a smoke ring as another type of dipole
gravitational field generator. As shown in Fig. 5, an inside-out turning ring of

†† 1
The vacuum “gravitational permittivity” constant is [25]: g0 ; (4pG) ¼ 1.19  109 kg . s2/m3.
 REVIEW OF GRAVITY CONTROL 187

Fig. 5 Inside-out whirling dense matter torus [5].

very dense mass (M) will create an upward force (of acceleration a) in the direc-
tion of the constant mass motion (Mv, where v is the mass velocity). This is also a
feature of the Lense Thirring effect. Forward’s linearization analysis generalizes
all of these effects into the following two key ingredients that are required to
produce antigravity forces:
1) Any mass with a velocity and an acceleration exerts many different
general relativistic forces on a test mass.
2) These forces act in the direction of the velocity and in the direction of the
acceleration of the originating mass.
In summary, these forces are equivalent to gravitational forces, which can be used
to cancel the Earth’s gravitational field.
One can also view this genre of devices as a gravity catapult machine in which
the machine pushes a body away using its general relativistic antigravity forces to
impart a change in velocity. A space launch operator on the ground wanting to
send a payload up into orbit would just ratchet up the strength of the (upward
directed) antigravity field to some value above 1-g, and after pressing the
release button the payload accelerates up and away into orbit. These devices
could also be placed in Earth orbit, stationed anywhere within the solar
system, or even distributed throughout the galaxy in order to establish a
network of gravity catapults. Space travelers could begin their trip by being
launched from the catapult on the Earth’s surface, and when they reach space
they would jump through various catapults as needed to reach their destination.

3. Make or Break Issues: Gravitomagnetic Antigravity

We presently do not have the technology to achieve the astronomical mass
densities, extreme velocities or accelerations of mass motion, and the large
device dimensions required to produce large enough antigravity forces for
useful propulsion. The issues we are faced with are: 1) dense materials, and 2)
gravitational properties of matter. Forward [27] suggested investigating
neutron neutron interactions. One could cool a gas of thermal neutrons from a
nuclear reactor to extremely low temperatures using magnetic confinement or
magneto-gravitational traps, and concentrate them into a small region through
188 E. W. DAVIS

the interaction of the trap’s magnetic field with the magnetic moment of the
neutrons. The Fermi energy‡‡ of the bound neutrons limits the neutron density
to 10 3 kg/m3. However, the formation of putative tetraneutrons§§ or the exist-
ence of a superconductive-type phase space condensation will create bosons that
do not have this limitation. It turns out that exotic quantum states of matter such
as Bose Einstein (BE) and Fermionic condensates}} transcend the Fermi energy
limit and thus possess highly unusual material properties. BE condensates were
first created in 1995 and Fermionic condensates in 2003, but both are still under-
going laboratory exploration.
As for the gravitational properties of matter, we know from electromagnet-
ism that the permeability (m) of magnetic materials such as iron is anomalously
large and nonlinear, which allows for the construction of highly efficient elec-
tromagnetic field generators. The gravitational equivalent to the magnetic per-
meability is a property of matter that is still largely unexplored. A material
possessing an anomalously large, very nonlinear gravitational permeability
(h) would be useful in the construction of highly efficient, very small scale
gravitational field generators. One would expect all materials to have an h
that is different from h0 because the atoms comprising any material have
quantum spin. Forward [27] reported that a rough estimate indicates there is
a very small difference between h and h0. It is thus necessary to implement a
coordinated theoretical program to determine the value of h for all known
forms of matter and an experimental program to find materials that might
possess anomalously large or nonlinear properties that can be used to intensify
time-varying gravitational fields.
Forward [27] also described an unsuccessful experimental attempt to find
materials that have the property of converting time-varying electromagnetic
fields into time-varying gravitational fields. This speculative property exploits
the fact that the magnetic and inertial moments are combined in an atom via
the usual quantum angular and spin momentum coupling. Other theoretical and
experimental concepts incorporating the use of rotating magnets to generate
gravitomagnetic/antigravity forces will be reviewed in Sect. IV.C.3, while
similar concepts using rotating superconductors are reviewed in Chapter 5.
Note particularly that Chapter 5 reviews the emerging experimental observations
of Martin Tajmar in which an apparent frame-dragging effect is observed near
super-cooled rotating rings as measured by ring laser gyros and accelerometers.
At the time of this writing these effects were being reported but not yet indepen-
dently confirmed.

‡‡
In condensed matter physics, this is the energy of the highest occupied quantum state in a system
of fermions at zero absolute temperature.
§§
These are a hypothetical stable cluster of four neutrons, but recent empirical evidence suggests
they exist. Readers should consult the technical literature for more information by using
“tetraneutrons” as a search term.
}}
BE or Fermionic condensates are a macroscopic collection of bosons or fermions (electrons,
nucleons, or atoms) that collapse into the same quantum state when they form at near zero absolute
temperature.
 REVIEW OF GRAVITY CONTROL 189

B. Exact Relativistic Antigravity Propulsion

Felber [28] used the Schwarzschild solution of Einstein’s general relativistic
field equation to find the exact relativistic motion of a payload in the gravitational
field of a mass moving with constant velocity. His analysis gives a relativistically
exact (strong gravitational field condition) calculation showing that a mass,
which radially approaches or recedes from a payload at a relative velocity of
ncrit . c/31/2 (ncrit ; critical velocity), will gravitationally repel the payload
as seen by distant inertial observers. In other words, any source mass, no
matter how large or small it is or how far away it is from a test body
(payload), will produce an antigravity field when moving at any constant velocity
above ncrit.
The exact relativistic strong-field condition that establishes the lower limit cri-
terion for ncrit to induce antigravity repulsion of a payload (as measured by distant
inertial observers in the rest frame of the source or in the initial rest frame of the
payload) is given by [28]:
  
3 L2 c GM
g2 . c 1  2 (6)
2 GMr 3 rc

In this expression, g ; (1 b2) 1/2 is the standard relativistic Lorentz trans-

formation factor which is a function of the normalized relativistic velocity
parameter b n/c, c ; c(r) 1 (2GM/rc 2) is the g00 (or time time)
component of the static Schwarzschild spacetime metric of a source (or
central) body of mass M, L is the constant specific angular momentum of a bal-
listic payload of mass m, and r is the radial distance of the approaching/receding
payload from M. One can solve the inequality in Eq. (6) for b (or n) under the
condition that a payload far from M, such that r  b (b is the periapsis distance
of the payload from M) and r  GM/c 2, and find that the payload will become
gravitationally repelled by M whenever g2 . 3/2 or b . 3 1/2. In order to
derive an exact solution, Felber considered the case for which M  m so that
the energy and momentum delivered to the payload has a negligible back-
reaction on the source body’s motion. He found that a strong gravitational field
is not required for antigravity propulsion because a weak-field solution achieves
the same results.
Felber discovered another interesting facet about this new relativistic antigrav-
ity effect. He found that there is also an antigravity field that repels bodies in the
backward direction with a strength that is one-half the strength of the antigravity
field in the forward direction. Thus, a stationary body will repel a test body that is
radially receding from it at any n . ncrit. To delineate the propulsion benefit from


A spacetime metric (ds 2) is a Lorentz invariant distance function between any two points in
spacetime that is defined by ds 2 ¼ gmndx mdx n where gmn is the metric tensor, which is a 44
matrix that encodes the geometry of spacetime and dx m is the infinitesimal coordinate separation
between two points. The Greek indices (m, n ¼ 0 . . . 3) denote spacetime coordinates, x 0. . . x 3,
such that x 1. . . x 3 ; space coordinates and x 0 ; time coordinate. The Schwarzschild metric is
ds2 ¼ 2(1 2 2GM/c 2r)c 2dt 2 þ (1 2 2GM/c 2r) 1dr 2 þ r2(du2 þ sin2u dw2). The corresponding metric
tensor is a diagonal matrix: gmn ¼ diag[2(1 2 2GM/c 2r), (1 2 2GM/c 2r) 1, r 2, r 2sin2u].
190 E. W. DAVIS

this technique, Felber determined that the maximum velocity (npmax-wf) that can
be imparted to a payload initially at rest by the weak (gravitational) field of
a larger source mass moving toward the payload at constant n . ncrit is npmax-wf
 c[b (3b) 1]. For the strong-field case, the maximum velocity (npmax-sf)
that can be imparted to the payload (initially at rest) by the larger source mass
moving toward the payload at any constant n is npmax-sf bc. Felber’s analysis
includes examples using black holes for the large source mass.
This form of antigravity propulsion is not too surprising because Misner et al.
[29], Ohanian and Ruffini [30], and Ciufolini and Wheeler [31] report that general
relativistic calculations show that the time-independent Kerr (spinning black
hole) gravitational field exhibits an inertial frame-dragging effect similar to grav-
itational repulsive forces in the direction of a moving mass at relativistic
velocities. This and Felber’s exact solution are among the genre of Lense
Thirring type effects that produce antigravity forces. It is interesting to note
that even though General Relativity Theory admits the generation of antigravity
forces at relativistic velocities [32], they have not been seen in laboratory exper-
iments because repulsive force terms are second and higher-order in the source
mass velocity. To invent a relativistic driver for a captured astronomical body
in order to use it to launch payloads into relativistic motion presents a large tech-
nical challenge for future experimenters.

C. Negative Energy Induced Antigravity

Negative energy density and negative pressure are acceptable results both
mathematically and physically in General Relativity and quantum field theories,
and negative energy/pressure manifests as gravitational repulsion (antigravity).
Chapter 15 provides a detailed review of the various forms of quantum field
theoretic sources of negative energy that are found in nature. For the purpose
of the present discussion, we will confine ourselves to a discussion of how
negative energy can be used to produce antigravity for the simplest case of coun-
teracting the Earth’s gravitational field. To counteract or otherwise reduce gravity
merely requires the deployment of a thin spherical shell (bubble) of negative
energy around an aerospace vehicle. This particular case study will serve as a
useful illustrative comparison with the Newtonian antigravity case discussed in
Section II.A.
We are interested only in the slow (nonrelativistic) motion, weak (gravity)
field regime that characterizes the physics of the Earth, Sun, other forms of
solar system matter, most interstellar matter (excluding compact dense stars
and black holes), and small test masses. In this case the time time component
of the Ricci curvature tensor (Rmn) is given by R00  Gr/c 2  (7.41 
10 28)r m 2. This is the primary quantity inside the general relativistic field
equation††† that encodes and measures the curvature of spacetime around a
source of mass-energy and characterizes the weak or strong gravity field

†††
The Einstein field equation is: Gmn ; Rmn 2 (1/2)gmnR ¼ 2(8pG/c 4)Tmn, where Gmn is the
Einstein curvature tensor and R ; R mm (the trace of Rmn) is the Ricci scalar curvature. In simplest
terms, this relation states that gravity is a manifestation of the spacetime curvature induced by a
source of matter (Tmn).
 REVIEW OF GRAVITY CONTROL 191

regime for all forms of astronomical mass density (r). For example, the Earth’s
mass density is 5500 kg/m3 so R00  4.08  10 24 m 2, which indicates that an
extremely flat space surrounds the Earth and thus the system is within the weak
field regime. Gravitational physics in the weak field regime is completely
described by the standard Schwarzschild spacetime metric, which leads to the
usual Newtonian and post-Newtonian gravitational physics.
Two simple approaches can be used to determine the negative energy density
required to counteract the Earth’s gravitational field: case a) integrate Einstein’s
general relativistic field equation, or case b) use an already derived result from
General Relativity that gives the repulsive force acceleration in terms of the
spacetime metric components. For case a, we have the generalized gravitational
Poisson equation from the Einstein field equation:
p 4p G p
R00 g00 ¼ 4 Tr(Tmn ) g00
c
4p G p
¼ 4 Tmm g00 (7)
c
4p G 
¼ 4 rE
c
where the definition
p
rE ; Tmm g00

is used, rE ; rest-energy density þ compressional potential energy (a.k.a.

pressure), g00 ; g00(r) is the time time component of the metric tensor gmn,
and Tr(Tmn) ; T mm is the trace (sum of diagonal matrix elements) of the stress
energy momentum tensor Tmn (a matrix quantity that encodes the density and
flux of a matter source’s energy and momentum). We use the identity
p p
R00 g00 ; r2 g00

to rewrite Eq. (7) as

p 4pG
r2 g00 ¼ 4 rE , (8)
c
where r2 is the standard Laplace differential operator. The left-hand side of Eq.
(8) is the gravitational potential. Now we integrate Eq. (8) once over a region of
space exterior to a ball (or thin spherical shell) of rest-energy density to obtain
p GM
jr g00 (r)j ¼ 2 ; g (acceleration, m=s2 ) (9)
r

where we use the standard spherically symmetric spacetime (or Schwarzschild)

coordinate system (t, r, u, w) in which time t, radial space coordinate r, and
angular space coordinates (u, w) have their usual meaning.
192 E. W. DAVIS

The second approach (case b) can be derived by recalling that in the exterior
Schwarzschild spacetime around a central mass M (a ball or thin spherical shell),
we have that
p GM
g00 (r) ¼ 1 
r
Because the definition is given that
p
g ; jr g00 (r)j

we then perform the radial derivative of (1 GM/r) and again arrive at Eq. (9).
Because from Special Relativity we have that M E/c 2 (for a given rest-
energy, E), a negative energy state is identical to a negative mass state (see
Sec. II.D) [14]. Thus we can replace the mass M in Eq. (9) with the negative
energy density rE rc 2 Mc 2/V by using the volume (V 4p r 2dr) of
a thin spherical shell of radius r and thickness dr, and rearrange quantities to
solve for rE to get the final result that we seek:

gc2
rE ¼
4pGdr
  (10)
 1:05  1027 3
¼ (J=m )
dr
where g is now the acceleration due to gravity near the Earth’s surface. Equation
(10) gives the negative energy density required to generate a repulsive gravita-
tional force that counteracts the Earth’s gravity field from the surface all the
way up to LEO (as g in LEO is only a few percent smaller than on the
surface). Any realistic value that one chooses for the bubble wall thickness dr
will give a negative energy density that will always be on the order of the equiv-
alent negative energy density of a dwarf star or neutron star. The technical chal-
lenge to implement this kind of antigravity, however, is daunting.
In the next section we discuss the case of a cosmological antigravity that is
generated by a form of matter having a positive energy density and negative
pressure. Another case of exotic matter having both negative energy density
and negative pressure is treated in Chapter 15 in which a special form of antigrav-
ity is generated that leads to the creation of faster-than-light spacetimes such as
traversable wormholes and warp drives.

D. Cosmological Antigravity
It turns out that there is already a naturally occurring antigravity force that acts
throughout the universe. Actually, this force acts upon the entire spacetime struc-
ture of the universe, and it is called cosmological inflation. Cosmological
inflation causes the universe to expand at an ever accelerating rate. In what
follows, we examine the nature of this cosmological antigravity force and its
potential breakthrough propulsion application.
 REVIEW OF GRAVITY CONTROL 193

1. Pressure as a Source of Gravity

Newtonian gravitation is modified in the case of a relativistic perfect-fluid
(where p  rE cannot be assumed). The stress-energy tensor T mn for this case
is [29]:
T mn ¼ (rE þ p)U m U n  pgmn (11)
where r is the fluid mass density, rE ; rc 2 is the fluid rest-energy density (or
just energy density), p is the fluid pressure, U m is the 4-velocity vector of the
fluid, and g mn is the metric tensor. We can contract the Einstein general relati-
vistic field equation using the identity g mm 4 to obtain R (8pG/c 4)T, which
is the Ricci curvature scalar. And so Eq. (11) becomes T rE 3p, which is
just the trace of T mn. Because T rE 3p, we get a modified gravitational
Poisson equation:

r2 f ¼ 4pG(rE þ 3p) (12)

where f is the gravitational potential. It should be noted that the energy density
and pressure are kept as separate terms as opposed to Eqs. (7) and (8) in the
previous section. Equation (12) means that a gas of particles all moving at
the same speed u has an effective gravitational mass density of r(1 þ u 2/c 2).
Thus, for example, a radiation-dominated fluid generates a gravitational
attraction twice as strong as one predicted by Newtonian gravity theory accord-
ing to Eq. (12).

2. Vacuum Energy of Einstein’s Cosmological Constant

A major consequence of the Einstein field equation is that pressure p becomes
a source of gravitational effects on an equal footing with the energy density rE.
One consequence of the gravitational effects of pressure is that a negative-
pressure equation of state that achieves rE þ 3p , 0 in Eq. (12) will produce
gravitational repulsion (i.e., antigravity). The Einstein field equation that includes
a cosmological constant L is:
8pG
Gmv þ Lgmn ¼  4 T mn (13)
c
where G mn is the Einstein curvature tensor. The L term, as it appears in Eq. (13),
represents the curvature of empty space. Now if one moves this term over to the
right-hand side of Eq. (13), which has become widespread practice in recent
times, then
 
8pG mn
G mn ¼  T þ Lg mn
(14)
c4
whereby this term now behaves like the stress-energy tensor of the vacuum,
mn
Tvac , which acts as a gravitational source:

mn Lc4 mn
Tvac ¼ g (15)
8pG
194 E. W. DAVIS

One should note that the absence of a preferred frame in Special Relativity
mn
means that Tvac must be the same (i.e., isotropic or invariant) for all observers.
There is only one isotropic tensor of rank 2 that meets this requirement: hmn
(the Minkowski flat spacetime metric tensor in locally inertial frames). Therefore,
mn
in order for Tvac to remain invariant under Lorentz transformations, the only
requirement we can have is that it must be proportional to hmn. This generalizes
in a straightforward way from inertial coordinates to arbitrary coordinates by
replacing h mn with g mn, thus justifying the curved spacetime metric tensor in
Eq. (15). By comparing Eq. (15) with the perfect-fluid stress-energy tensor in
Eq. (11), we find that the vacuum looks like a perfect fluid with an isotropic
pressure pvac, opposite in sign to the energy density rvac. Therefore, the
vacuum must possess a negative-pressure equation of state (according to the
First Law of Thermodynamics):

pvac ¼ rvac (16)

The vacuum energy density should be constant throughout spacetime, because a

gradient would not be Lorentz invariant. So by substituting Eq. (16) into rE þ 3p,
we see that

rvac þ 3pvac ¼ rvac þ 3(rvac )

¼ 2rvac (17)
,0

The vacuum equation of state is therefore manifestly negative. Last, when incor-
porating rvac into the Einstein field equation as a gravitational source term, and
comparing its corresponding (Lorentz invariant) stress-energy tensor rvac g mn
with Eq. (15), then the usual identification (or definition) is made that:

Lc4
rvac ; (18)
8pG

Thus the terms “cosmological constant” and “vacuum energy” are essentially
interchangeable in this perspective and mean the same thing (whereupon
rvac ; rL), which is seen in the present-day cosmological literature.
By substituting Eq. (18) into Eq. (16), one observes that a positive L will act to
cause a large-scale repulsion of space (because this gives a negative vacuum
pressure), whereas a negative L (giving a positive vacuum pressure) will cause
a large-scale contraction of space. Because L is a constant, the vacuum energy
is a constant (i.e., time-independent). This then implies a problem with energy
conservation in an expanding universe since we expect that energy density
decreases as a given volume of space increases, which is the case for the ordinary
matter and cosmic microwave background that is observed in extragalactic space.
In other words, the matter and radiation energy densities decay away as the uni-
verse expands while the vacuum energy density remains constant.
 REVIEW OF GRAVITY CONTROL 195

The cure for this apparent energy conservation problem is the vacuum
equation of state given by Eq. (16). A negative pressure is something like
tension in a rubber band. It takes work to expand the volume rather than work
to compress it. The proof of this is as follows [33]: the energy created in the
vacuum by increasing (expanding) space by a volume element dV is rvacdV,
which must be supplied by the work done by the vacuum pressure pvacdV
during the expansion of space, therefore we must have pvac rvac. In other
words, the work done by the vacuum pressure maintains the constant vacuum
energy density as space expands. Therefore, the vacuum acts as a reservoir of
unlimited energy that provides as much energy as needed to inflate any region
of space to any given size at constant energy density.

3. Dark Energy
Dark energy is an easily misunderstood form of energy in cosmology. There
are two sets of evidence pointing toward the existence of something else beyond
the radiation and (ordinary and dark) matter itemized in the overall cosmic
energy budget.‡‡‡ The first comes from a simple budgetary shortfall. The total
energy density of the universe is very close to critical. This is expected theoreti-
cally and it is observed in the anisotropy pattern of the cosmic microwave back-
ground (CMB). Yet, the total matter density inferred from observations is 26% of
critical. [Note: 26% total matter density 4% ordinary (baryonic) matter þ 22%
dark matter.] The remaining 74% of the energy density in the universe must be in
some smooth, unclustered form that is dubbed “dark energy.” The second set of
evidence is more direct. Given the energy composition of the universe, one can
compute a theoretical distance versus redshift diagram. This relation can then be
tested observationally.
Riess et al. [34] and Perlmutter et al. [35] reported direct evidence for dark
energy from their supernovae observations. Their evidence is based on the differ-
ence between the luminosity distance in a universe dominated by dark matter and
one dominated by dark energy. They showed that the luminosity distance is larger
for objects at high redshifts in a dark energy-dominated universe. Therefore,
objects of fixed intrinsic brightness will appear fainter if the universe is composed
of dark energy. The two groups measured the apparent magnitudes of a few dozen
type Ia supernovae at redshifts z 0.9, which are known to be standard distance
candles (meaning they have nearly identical absolute magnitudes at any cosmo-
logical redshift-distance).§§§ The supernovae data strongly disfavored (with high
confidence) the flat, matter-dominated (Vm 1, VL 0) universe and the
pure open universe (Vm 0.3, VL 0) models.}}} After this discovery, a lot

‡‡‡
Dark matter and dark energy are not to be confused. Dark matter is a nonluminous, nonabsorb
ing, nonbaryonic form of matter that only interacts with all other forms of matter via gravitational and
weak nuclear forces. Dark matter has a positive rest energy density and a nearly negligible positive
pressure. Thus, it has no beneficial application for breakthrough propulsion physics.
§§§
In cosmology, the redshift z serves as a surrogate for distance (in light years) or look back time.
}}}
Vm ¼ ratio of energy density contained in matter (as measured today) to the critical energy
density; VL ; Vvac ¼ ratio of energy density in a cosmological constant to the critical energy
density; rcr ; 3H20/8pG is the critical energy density, where H0 is the present day Hubble rate.
196 E. W. DAVIS

of attention was paid to choosing an appropriate name for this new energy.
“Quintessence” was one good choice because it expresses the fact that, after cos-
mological photons, baryons, neutrinos, and dark matter, there is a fifth essence in
the universe. More recently, the term dark energy is used more often, with quin-
tessence referring to the subset of models in which the energy density can be
associated with a time-dependent scalar field or a time-dependent cosmological
vacuum energy.
In analyzing the cosmological modeling results suggested by the type Ia super-
novae data, it becomes apparent that the only form of dark energy budgeted for in
the models is the cosmological constant. To consider other possibilities, we
evaluate the time evolution of the general relativistic conservation law for
energy, rmT mn rmT m0 0, where n 0 to signify time evolution and rm is
the covariant derivative (or spacetime curvature gradient), in an expanding uni-
verse as applied to the cosmological constant [29]:

@rE a_
þ (3rE þ 3p) ¼ 0 (19)
@t a

where a is the scale factor of the universe and ȧ is the time derivative of a.
Equation (19) is derived using Eq. (11) in the case of a perfect isotropic
fluid where there is no gravity and velocities are negligible such that U m
(1, 0, 0, 0), and the energy density and pressure evolve according to the continuity
and Euler equations. The only way Eq. (19) can be satisfied with constant
energy density is if the pressure is defined by Eq. (16). One might imagine
energy with a slightly different pressure and therefore energy evolution. Define
the equation of state w:
p
w¼ (20)
rE

A cosmological constant corresponds to wL ; wvac 1, matter (ordinary and

dark) to wmatter  0, and radiation to wrad 1/3. The earlier Riess and Perl-
mutter supernovae data (fixing the universe to be flat) showed that values of
wde . 0.52 for dark energy are strongly disfavored. In fact, Riess and a team
of collaborators (a.k.a. the “Higher-Z Team”) recently published new observa-
tional data and analysis that include a much larger survey of type Ia supernovae
that are at much higher cosmological redshift [36]. The measured spectra of
ancient (z
1, or up to 10 billion light-years distance or a look-back time of
up to 10 billion years ago) and recent (z 0.1, or 1 billion light-years distance
or a look-back time of 1 billion years ago) were compared and showed that
there was no evolutionary change in the physics that drives type Ia supernovae
explosions and their subsequent spectral luminosity output. This establishes the
efficacy of using type Ia supernovae as a standard distance candle for cosmolo-
gical dark energy surveys. The Higher-Z Team’s results also concluded, with


Nonrelativistic (ordinary and dark) matter has a very tiny positive p / Temp/m (Temp is absol
ute temperature, m is mass), while a relativistic gas (of radiation) has p ¼ rE /3 . 0.
 REVIEW OF GRAVITY CONTROL 197

98% confidence, that wde 1.0, and that this is a perpetual constant (over at
least 10 billion years time) [36]. This result falsifies all quintessence models
for cosmology. Therefore, a cosmological constant is consistent with the dark
energy data to a high degree of precision and statistical confidence whereby
we can now state that dark energy is the vacuum energy of Einstein’s cosmolo-
gical constant because wde wL 1 [37,38].
Equation (19) can be integrated to find the evolution of the dark energy density
rde rL as a function of the cosmological scale factor a:
8 9
< ð da0 =
0
rde / exp 3 ½1 þ w de (a ) (21)
: a0 ;
a

where a0 is the dummy integration variable for the scale factor. Because
wde 1 ( wL) is a constant in Eq. (21), we then obtain rde / a exp[ 3(1 þ
wde)] or rde rL / a 0. This is exactly what we expected on the basis of our pre-
vious analysis in Sec. III.D.2. For a comparison with this result, one should note
that rc 2 / a 3 for (ordinary and dark) matter and rrad / a 4 for radiation such
that rc 2 ! 0 and rrad ! 0 as a ! 1 while rde rL remains constant.

4. Cosmological Inflation as a Form of Antigravity

It has already been shown in the literature that an inflationary solution (i.e., an
accelerating universe) can solve the cosmological horizon problem. We know
that General Relativity Theory ties the expansion of the universe to the energy
within it, so we now ask what type of energy can produce acceleration. The
answer is found by examining the time time and space space components of
the Einstein field equation [29]:
 2
a_ 8p G
¼ r (22)
a 3 E
 
a€ 1 a_ 2
þ ¼ 4pGp (23)
a 2 a

where ä is the second time derivative of the scale factor a. One should note that
the left-hand side of Eq. (22) is also the square of the Hubble parameter H, and
that Eq. (22) is also known as the first Friedmann equation. Multiplying Eq. (22)
by 1/2 and then subtracting the result from Eq. (23) gives

a€ 4pG
¼ (rE þ 3p) (24)
a 3
Equation (24) is known as the second Friedmann equation or the cosmological
acceleration equation. Acceleration is defined by ä .0, so for this condition to be
met we require that rE þ 3p , 0 in Eq. (24). So inflation requires that p , rE/3
in order to meet this constraint. Because the energy density is always positive, the
198 E. W. DAVIS

pressure must be negative. We saw in the previous section that the accelerated
expansion (or cosmological inflation), which causes supernovae to appear very
faint, can be caused only by a dark (or vacuum) energy with p pvac , 0.

5. Breakthrough Propulsion Application of Dark/Vacuum Energy

If we could somehow harness a local amount of dark/vacuum energy, then
could we use its negative pressure property to produce an antigravity propulsion
effect? To answer this question we can use the experimentally measured value for
rde rL  2.4r0c 2  10 9 J/m3, where r0 is the present-day value of the total
cosmological mass density of (ordinary and dark) matter [36,39]. Using this
number we can work through the math and estimate that the total amount of
dark/vacuum energy contained within the volume of our solar system amounts
to the mass equivalent of a small asteroid. This means that its repulsive gravita-
tional influence upon planetary orbital dynamics inside the solar system is com-
pletely inconsequential. Only on the extragalactic-to-cosmological scale will its
repulsive gravitational property achieve strong enough influence over matter and
spacetime. On this basis, we conclude that it is highly unlikely, if not impossible,
that one will be able to invent a technology in the near future that can acquire and
exploit a near-cosmological amount of dark/vacuum energy to implement a
useful antigravity propulsion system. There are a number of unpublished
proposals on the Internet that claim to achieve this goal for a faster-than-light
“warp drive” propulsion concept using questionable theoretical approaches.
However, White and Davis [40] proposed a first-order experiment that is
designed to generate artificial dark/vacuum energy in the lab for the purpose
of exploring a warp drive propulsion concept that is rooted in D-Brane
quantum gravity theory (see Chapter 15).

IV. Miscellaneous Gravity Control Concepts

A. Artificial Gravity via the Levi-Civita Effect
Levi-Civita [41] considered the possibility of generating a static uniform
(cylindrically symmetric or solenoidal) magnetic or electric field to create an arti-
ficial gravity field in accordance with General Relativity Theory. Levi-Civita’s
spacetime metric for a static uniform magnetic field was originally conceived
by Pauli [42]:
( )
2 1 2 2 2 (x1 dx1 þ x2 dx2 )2
3 2
ds ¼ (dx ) þ (dx ) þ (dx ) þ 2
amag  ½(x1 )2 þ (x2 )2 
(25)
  3   2
x x3
 c1 exp þ c2 exp  (dx4 )2
amag amag

where c1 and c2 are integration constants that are determined by appropriate

boundary conditions, and x 1 . . . x 4 are Cartesian coordinates (x 1. . . x 3 ; space,
 REVIEW OF GRAVITY CONTROL 199

x 4 ; ct for time) with orthographic projection. The important parameter in

Eq. (25) is [42]:
  1=2
2 4pG 1
amag ¼ c B
m0 (26)
18 1
 (3:48  10 )B (m)

where amag is the constant radius of curvature of the spacetime geometry given by
Eq. (25), m0 is the vacuum permeability constant, and B is the magnetic field
intensity. The constant radius of curvature for the case of an electric field in
Eq. (25) is found by replacing B in Eq. (26) with the electric field intensity E
along with an appropriate change in the related vacuum electromagnetic
constants (i.e., Bm0 1/2 ! E11/20 ; where 10 is the vacuum permittivity constant,
8.8542  10 12 F/m). Thus, we have aelec  (1.04  1027)E 1 for the case of
a static uniform (cylindrically symmetric) electric field that induces artificial
gravity.
The physical meaning of the spacetime geometry described by Eq. (25) is
difficult to interpret because its mathematical form is arcane. Puthoff et al. [43]
recast Eq. (25) into a more transparent form using cylindrical spacetime coordi-
nates (t, r afield, u, w, z):
    2
z z
ds2 ¼  c1 exp þ c2 exp  c2 dt2
afield afield
 
þ a2field du2 þ sin2 u dw2 þ dz2 (27)

where afield (; amag or aelec) is the constant radius of curvature; time t, z-axis in
space, and angular space coordinates (u, w) have their usual meaning. We can
study Eq. (27) to better understand the spacetime geometry that the magnetic or
electric field induces. The spatial part of Eq. (27), a2field(du2 þ sin2u dw2) þ dz 2,
is recognized as the three-dimensional metric of a “hypercylinder” (denoted by
the topological product space S2  <, where S2 is the ordinary sphere and < is
the real line that defines the linear space dimension corresponding to the
z-axis). Equation (27) shows that Levi-Civita’s spacetime metric is simply a
spatial hypercylinder with a position dependent gravitational potential.
This spacetime geometry is interesting from the standpoint that it describes a
unique cylindrically shaped “trapped” space having an artificial gravity field that
depends on the magnitude of the applied electric or magnetic field intensity.
Puthoff et al. [43] showed that the gravitational potential inside this trapped
space has the effect of slowing down propagating light beams (i.e., reducing
the speed of light) to a minimum value at the center of the trapped space equidi-
stant from the ends of the solenoidal electric or magnetic field. The magnitude of
light-speed reduction is a function of the strength of the magnetic or electric field
intensity that is applied by a field generator. For example, we require B  4.93 
1018 Tesla to generate a gravitational field that slows a light beam to c/2 at the
200 E. W. DAVIS

center of the trapped space, assuming that the magnetic field region generated
by the solenoid is 1-m in length. This greatly exceeds the B-field intensity
of magnetars (1011 Tesla). The B-field energy density for our example is
B 2/2m0  1043 J/m3, which greatly exceeds the rest-energy density of a neutron
star (1035 J/m3). The equivalent electric field case requires E  1.47  1027
V/m with a corresponding energy density 10E 2/2  1043 J/m3. For either
electric or magnetic field case, the solenoidal field creates a spacetime
curvature with afield  0.71 m, which corresponds to an equivalent artificial gravi-
tational acceleration (at the center of the solenoid) of 3.19  1016 m/s2 (or
3.25  1015 g).
It is an interesting exercise to compare these magnetic and electric field
intensities with the critical quantum electrodynamic (QED) vacuum break-
down field intensities where nonlinear quantum effects begin to appear.
This is where intense fields pack enough energy to excite elementary par-
ticles out of the vacuum. The critical QED vacuum breakdown electric
field intensity is Ec 2m2e c 3/h e  1018 V/m while the critical breakdown
magnetic field intensity is Bc Ec/c  1010 Tesla, where Ec is defined by
the total rest-energy of an electron positron pair created from the vacuum
divided by the electron’s Compton wavelength and its charge, me is the
electron mass (9.11  10 31 kg), e is the electron charge (1.602  10 19 C),
and h is Planck’s reduced constant (1.055  10 34 J . s). In comparing these
field intensities with those from the previous example, we observe that the
1-m solenoid field intensities exceed the QED vacuum breakdown field inten-
sities by more than eight orders of magnitude, which implies that nonlinear
QED effects will appear and potentially influence, for better or for worse,
one’s attempt to create the trapped space geometry. The nonlinear QED
effects will likely cause some kind of quantum back-reaction upon the
trapped space geometry, but we are unable to predict the exact effects or
their magnitude.
On the other hand, note that the artificial gravitational acceleration of 3.25 
1015 g from the 1-m solenoid example is very extreme, and certainly not compa-
tible with space travelers who should experience accelerations 1-g over an
extended duration. For the purpose of a gravity control model that requires accel-
erations 1-g, we repeat the example of the 1-m solenoid Levi-Civita effect to
create a spacetime curvature equivalent to an artificial gravitational acceleration
of 1-g and ascertain the required electric or magnetic field intensities. In a 1-g
gravitational (acceleration) field, the speed of light is reduced by 0.42 m/s,
which corresponds to amag aelec  R. Using this result and working the
math backwards, we find that the magnetic field intensity required to
achieve this effect is B  5.46  1011 Tesla (energy density 1029 J/m3)
while the corresponding electric field intensity is E  1.63  1020 V/m
(energy density 1029 J/m3). These field intensities are still larger than Bc and
Ec by more than one to two orders of magnitude. Present-day tabletop
petaWatt laser technology is beginning to reach Ec- and Bc-field intensities in
the laboratory. Advanced pulsed-power technology is not too far behind. It
appears reasonable to stay alert to the possibility that laser technology will
evolve in the near future to achieve electric or magnetic field intensities on the
order of that required to generate a 1-g artificial gravity field.
 REVIEW OF GRAVITY CONTROL 201

B. Gravitational Wave Rockets

Can we exploit an intense beam of gravitational wave radiation to propel a
spacecraft? It turns out that Einstein’s General Theory of Relativity does allow
for a beam of gravitational waves to be used as a rocket propellant. Because grav-
itational waves are ripples in the shape (i.e., curvature) of spacetime, we can use
this propellant to attain acceleration simply by ejecting one “hard vacuum” (i.e.,
the beam of gravitational waves) into another (i.e., the background spacetime).
A common example of this propulsion effect is that of a star undergoing asym-
metric octupole collapse, which achieves a net velocity change of 100 to
300 km/s via the anisotropic emission of gravitational waves [44].
We elaborate on this further by pointing out an important result from General
Relativity: Because a gravitational wave has a definite energy, it therefore is a
source that induces its own gravitational field. This induced field is a second-
order effect in the hmn, which is tremendously amplified in the case of high-
frequency gravitational waves by the very large factor l 2 (note that n / 1/l
from wave optics, where n is frequency and l is wavelength) introduced by
the terms quadratic in @g hmn (i.e., terms second-order in 1/l) that comprise the
gravitational wave stress-energy pseudotensor [45].†††† Therefore, the gravita-
tional wave itself produces the background field upon which it propagates.

1. Gravitational Wave Rocket Based on Photon Rocket Spacetime Metric

Bonnor and Piper [46] performed a very lengthy and rigorous analysis for their
study of gravitational wave rocket motion. They obtained the gravitational wave
rocket equations of motion directly by solving the Einstein general relativistic
field equation using the spacetime metric of a photon rocket [46]:

ds2 ¼ 1  ( f 2 r2 c 4 ) sin2 u  (2frc 2 ) cos u  (2GMc 2 r 1 ) c2 dtret

2

þ 2cdtret dr  r 2 dV2  (2fr 2 c 2

sinu)cdtret du (28)

where the spherical coordinates (r, u, w) have their usual meaning, tret is retarded
time, dV2 ; du2 þ sin2u dw2, M ; M(tret) is the mass of a particle at the origin,
and f ; f(tret) is its acceleration along the negative direction of the polar axis.
Also note that Bonnor and Piper use a metric signature for ds 2 that is opposite
of the one used for ds 2 in the previous sections.‡‡‡‡ Equation (28) is the
solution of the Einstein field equation having a stress-energy tensor for the
photon fluid that propels the rocket: Tmn r 2(2cṀ 6M f cosu)jmjn, where
an overdot signifies differentiation with respect to tret, and jm is a null vector

††††
In the linearized general relativistic field equation for gravitational waves, hmn is the small first
order (weak field) perturbation metric tensor quantity, xg are general spacetime coordinates, and @g hmn
is the ordinary partial derivative of hmn with respect to xg.
‡‡‡‡
A metric signature counts how many dimensions of spacetime have a timelike or spacelike
character. This is strictly a matter of choice that is established by assigning a negative sign to the time
like component(s) and a positive sign to the spacelike component(s) in ds 2, or vice versa.
202 E. W. DAVIS

(i.e., a zero-length vector in four-dimensional spacetime that is orthogonal to

itself) [46]. The first term in parentheses quantifies the loss of mass and the
second term quantifies the change of momentum by the rocket because of the ani-
sotropic emission of photons.
However, the photon fluid stress-energy tensor must somehow be canceled out
so that one actually solves the vacuum Einstein field equation Rmn 0, because
the gravitational waves that propel the rocket are not a physical fluid; instead they
are ripples in the shape of spacetime that move through the surrounding back-
ground spacetime. Bonner and Piper therefore added terms to Eq. (28) that
produce new terms within the resulting vacuum field equation, which cancel
out the photon fluid stress-energy tensor in order to arrive at the equations of
motion. In order to carry out their program, they found that a gravitational
source loses mass by the emission of quadrupole waves and it gains momentum
from recoil when it emits quadrupole and octupole waves. Thus, the terms that
they added to Eq. (28) are those representing quadrupole and octupole gravita-
tional waves.
As the rocket emits gravitational waves, it will lose mass DM in terms of the
quadrupole oscillations induced by its internal motions because the conservation
of energy guarantees that radiation reaction forces will pull down the internal
energy of the rocket at the same rate as gravitational waves carry energy away [46]:

ðt
G ~j2 dtret
DM ¼ 
30c7
1
(29)
ðt
 (1:02  10 71
) ~j2 dtret (kg)
1

where j̃ is the third derivative with respect to retarded time of the time-dependent
function j(tret) in the mass quadrupole moment Q m‘2j(tret); m is the initial
mass of the rocket; ‘ is a linear dimension associated with the rocket; the quantity
inside the integral represents the power output in gravitational waves as being
roughly the square of the internal motion power flow and has the dimension of
[(energy)2/time]; and the negative sign indicates that mass is being lost from the
rocket. The rocket will acquire an acceleration (i.e., thrust) due to the combined
rates of change of the quadrupole and octupole moments induced by its internal
motions [46]:

mGl5
f ¼ (2_pq_  3p€q þ 3q€p)
630c7 (30)
73 5 2
 (4:84  10 ) m‘ (2_pq_  3p€q þ 3q€p) (m=s )

where ṗ and p̈ are the first and second retarded time derivatives of p(x0 ) ; d2 j(x0 )/
dx0 2, q̇ and q̈ are the first and second retarded time derivatives of q(x0 ) ; d3k(x0 )/
dx0 3, x0 is a dummy variable, and k(tret) is the time-dependent function in the
 REVIEW OF GRAVITY CONTROL 203

mass octupole moment O m‘3k(tret). If the rocket starts from rest at time tret 0,
then at a later time tret t1, the rocket will acquire a final velocity nfinal, which is
found by integrating Eq. (30) between these time limits [46]:

ðt1
mG‘5
nfinal ¼ p_ q_ dtret
315c7
0
(31)
ðt1
73
 (9:69  10 )m‘5 p_ q_ dtret (m/s)
0

To put things into a quantitative perspective, we use Eqs. (30) and (31) to make
estimates for f and nfinal in the simplest case of a gravitational wave rocket having a
linear dimension (radius or length) of 1-m and an initial mass of 1-kg. In this case,
Eq. (30) shows that the magnitude of the combined rates of change of the quadru-
pole and octupole moments (i.e., the combined quadrupole and octupole oscillations
due to the internal motions of the rocket) will need to be 1072 in order to generate
a propulsive acceleration of 1 m/s2 and attain a final velocity of 1 m/s (after inte-
grating the acceleration over an appropriate duration of time). If we scale our
example rocket up to a realistic linear dimension of 50 m and a mass of 105 kg,
the magnitude of the combined quadrupole octupole oscillations will need to
be 1058 in order to generate the same acceleration and final velocity. These two
examples demonstrate how very important the mass and linear dimension scaling
in m‘5 is for determining f and nfinal given some physically reasonable magnitude
for the combined quadrupole octupole oscillations. We conclude from this that
stellar-sized objects having ‘
109 m (
solar radius) and m
1030 kg (
solar
mass) will achieve superior propulsive f
104 m/s2 and nfinal
105 m/s for com-
bined quadrupole octupole oscillations of magnitude 102. This is physically con-
sistent with the gravitational wave propulsion effect produced by a star undergoing
asymmetric octupole collapse. In all cases, jDMj 10 71 times the internal motion
power flow via Eq. (29), and so will be minute.
Baker [47] reviewed a series of patent designs for devices that would produce
high-frequency gravitational waves in the lab for an exploratory study of their use
in rocket propulsion. The various devices involve active elements comprised of a
small mass or a system of small masses (e.g., coils and/or piezoelectric crystals
on the order of sub-millimeter and less than gravitational wave wavelength in
size) that undergo a rapid change in acceleration (of mass motion or angular
momentum), which he defines as the “jerk” (i.e., third-time derivative motion).
By applying a series of “rapid jerks” over a picosecond or less to the active
elements, using strong electric, magnetic, and electromechanical driving
forces, the devices will generate a significant quadrupole moment and thus
emit
1012 Hz gravitational waves. The upshot is that Baker’s devices rely on
very fast moving, high-frequency events (109 to 1012 Hz) and very strong elec-
tric, magnetic, or electromechanical driving forces in order to generate very large
magnitude quadrupole oscillations that give rise to large-magnitude gravitational
wave radiation power output (see also [48], [49]).
204 E. W. DAVIS

2. Producing Gravitons via Quantization of the Coupled

Maxwell Einstein Fields
In the previous section we saw that it will be necessary to violently “shake
things up” inside a rocket in order to produce gravitational waves for propulsion.
However, it turns out that there are a few alternative methods for generating
gravitational waves. These entail the production of quantized gravitational
waves, called gravitons, in the laboratory via specialized electromagnetic field
or elementary particle interactions. In this and the sections that follow, we
review these alternative concepts.
One candidate process uses electrostatic or magnetostatic fields to annihilate
an incident photon with production of a graviton. For this case, electrodynamics
and gravitation may be written in Hamiltonian form and a quantization carried
out in an approximation scheme. For weak gravitational fields we assume that
spacetime is almost flat, and that an almost-Minkowski metric tensor is appropri-
ate. Then we can write the spacetime metric tensor gmn as a Minkowski flat space-
time metric tensor hmn plus a small quantity of the first order:

gmn ¼ hmn þ hmn (32)

where hmn ; diag( 1, 1, 1, 1) and hmn is a small, first-order metric tensor quan-
tity representing a small perturbation in otherwise flat spacetime. We choose
spacetime coordinates such that gm0 hm0, which define a time-orthogonal coor-
dinate system in which the time axis is everywhere orthogonal to the spatial coor-
dinate “grid” (or curves) that form a given spatial hypersurface wherein we
perform our calculations. Using these assumptions we can write the Hamiltonian
H for the coupled Maxwell Einstein fields in the approximate form [50]:
ð

H ¼ HG þ HM þ (hij xi xj =8)  2hlk Frl Fmk drm d3 x (33)

which is written in quantum field theory natural units (i.e., G c h 1). The
first term in Eq. (33) is the gravitational field Hamiltonian HG, which contains the
gravitational field variables (hij) and momenta (pij); the second term is the
Maxwellian electromagnetic field Hamiltonian HM, which contains the electro-
magnetic field variables (vector potential Ai, which defines the electromagnetic
field-strength tensor Fij) and canonical momenta xi; and the third term is the
coupled field interaction term expressed as a three-dimensional (spatial)
volume integral. The Latin indices appearing on quantities inside the coupled
field interaction term represent spatial coordinates (x 1 . . . x 3), and drm ; diag(1,
1, 1) is the three-dimensional (spatial) unit tensor. Equation (33) describes sets of
gravitational and electromagnetic field oscillators. In considering the interaction
term, when the theory is quantized, we see from Eq. (33) that it is made up of
sums of products, each containing one gravitational field operator and two elec-
tromagnetic field operators. This interaction then implies that a photon can decay
into another photon and a graviton (see Fig. 6). A careful study of this process
(the math is very dense) shows that the interaction matrix elements for it do
not vanish unless all three particles propagate in the same direction. However,
all three particles are bosons possessing zero rest-mass (note: this is a class of
 REVIEW OF GRAVITY CONTROL 205

Fig. 6 Decay of photon (left) into a photon and graviton (right).

elementary particles having integer quantum spin of which photons and gravitons
are representative members). Energy and momentum can be strictly conserved
only if all particles propagate in the same direction. Therefore, this process
cannot occur except, possibly, at energies 1028 eV corresponding to decay
photon wavelengths 10 34 m [note: 1 eV 1.602  10 19 J and the Planck
 c 5/G)1/2  1.96  109 J  1028 eV].
energy is (h
Further study of the interaction term in Eq. (33) shows that graviton pro-
duction becomes possible if photons are incident on a Coulomb electrostatic or
magnetostatic field. The cross section will be very small. For a Coulomb scatterer
containing uniform electric or magnetic fields, with linear dimensions that are
large compared with the incident photon wavelength, the graviton production
cross section (s) is [50]:
8p 2 G‘E Coul
s¼
c4 (34)
 
 6:50  10 43 ‘E Coul (m )2

where ECoul is the Coulomb scatterer energy and ‘ is the scatterer’s linear dimen-
sion in the direction of propagation of the photon. Note the absence of Planck’s
constant in Eq. (34). This is not a totally unexpected result for the very simple
reason that the interaction of two boson fields has a classical limit.
An example theoretical estimate for the graviton production cross section is
that of a cubic meter of space containing 108 J of electric energy, which then
gives s  6.50  10 35 m2 according to Eq. (34). This is an exceedingly low
cross section. In contrast to this, our Galaxy has a total magnetic field energy
density  2  10 13 J/m3 (for a galactic magnetic field  7  10 10 Tesla) con-
tained within a galactic volume  1.80  1061 m3 possessing a linear dimension
(galactic diameter) ‘Galaxy  7.57  1020 m (or 80,000 light-years) [51,52]. There-
fore, the total galactic magnetic field energy is  3.61  1048 J giving the result
that s  1.78  1027 m2 from Eq. (34), and thus converting approximately 1
part in 1013 incident photons into gravitons by this scattering process. These grav-
itons should be detectable near the Earth with wavelengths (lgrav  ‘Galaxy)
ranging from extremely low frequencies (lgrav  0.1 parsec to 20 AU) to very
high frequencies (lgrav  3 m to 0.3 cm) [53].§§§§ Thus, an astronomical exper-
iment for detecting gravitons is possible to do with present technology.

§§§§
1 AU (mean Earth Sun distance) ¼ 1.50  1011 m, 1 parsec ¼ 3.26 light years, 1 light
year ¼ 9.46  1015 m.
206 E. W. DAVIS

To produce very high-to-ultrahigh frequency gravitons in the lab will require

the scattering of ultrahigh-intensity petaWatt laser photons (beam inten-
sity  1019 to 1034 W/m2) from a Coulomb-electrostatic or magnetostatic field
scatterer possessing a total energy .109 J. Such ultrahigh scatterer field energies
can potentially be produced in the laboratory using pulsed power systems. At
present the Z-Machine at Sandia National Laboratory achieves a peak magneto-
static field energy output of  several  106 J from the implosion of target hohl-
raums, which is only 0.1% of what is required to produce a significant number of
laboratory gravitons. Gravitons produced by pulsed power systems will necess-
arily be very high-to-ultrahigh frequency because their wavelengths will be
much smaller than the linear dimension of the Coulomb scatterer.
The X-1 machine, a follow-on device to the Z-Machine, is conceptually
designed to produce more than one order of magnitude higher field energy
( several  107 J), which is still only 1% of what is required. However, the
history of the Z-Machine program has shown that the energy and power output
achieved in practice have always been a factor of two or more larger than
design estimates, and that further design improvements to the implosion hohl-
raums led to energy and power outputs that increased by one order of magnitude
within 24 months. Therefore, it is expected that the X-1 machine, if it should
become operational, could be improved to the point of achieving field
energies  108 to 109 J, and that a third generation device will routinely
achieve field energies .109 J within a few years after that.
The small linear dimension of the scattering field (‘ 5 m) in a pulsed power
system remains a major obstacle to achieving a sizable cross section via Eq. (34)
and efficient graviton production rates. However, the role of the ultrahigh-
intensity laser is to compensate for this by saturating the scattering field with
an ultrahigh-intensity flux of incident laser photons, thus boosting graviton
production to significant measurable levels. Because the Coulomb scattering
field also polarizes the quantum vacuum zero-point fluctuations (a.k.a. virtual
particle pairs), there will then be additional (nonlinear QED) scattering of the
incident laser photons by the polarized vacuum resulting in the generation of
second-harmonic photons [54]. It is possible that second-harmonic photons
will contribute to or otherwise further boost graviton production in this
scheme. Additional research will be required to quantify this contribution as
well as provide an estimate for the overall graviton production rate and total grav-
iton radiation power achieved.
We now speculate on what a graviton rocket propulsion system might look
like based on the previous discussion. But first we point out that ultrahigh-
intensity lasers are tabletop devices, whereas pulsed power systems have been
miniaturized to fit inside commercial trucks. On this basis, we assume that an
ultrahigh-pulsed power (total output energy .109 J) and ultrahigh-intensity
laser system comprises a graviton rocket propulsion system that efficiently gen-
erates a high number of gravitons possessing a high radiation power output. The
gravitons are high-frequency, highly collimated and possess a well-defined
energy spectrum. We imagine the propulsion system to be comprised of many
long (‘ 500 m) linear arrays composed of individual ultrahigh-pulsed power
implosion hohlraum segments, and each of the linear arrays are further arranged
to form several cylindrically concentric super-arrays. Affixed at one end of the
 REVIEW OF GRAVITY CONTROL 207

concentric super-array is a cylindrically concentric array of ultrahigh-intensity

lasers. The lasers fire at each of the linear arrays the instant the hohlraums
implode, thereby generating an intense avalanche of collimated high-frequency
gravitons, which are emitted from the other end of the arrays and ejected out
the rear of the rocket. Perhaps it might be better to have the linear arrays arranged
to form concentric cones instead of concentric cylinders so that gravitons emitted
from each of the linear arrays can merge into a single beam. There is also the
benefit of an additional thrust component produced by the outgoing beam of
photons generated via the photon conversion process (per Fig. 6). Further work
will be required to estimate the necessary physical parameters and photon con-
version efficiency in order to define the propulsive performance for this graviton
rocket concept.

3. Graviton Production via Particle Accelerators

Novaes and Spehler [55] describe an alternative scheme for producing gravi-
tons in the lab using a laser and an e þe linear collider (e þ positively charged
positron/antielectron and e negatively charged electron). They evaluated the
cross section for the electron þ photon ! electron þ graviton reaction in the
framework of linearized gravitation. They analyzed this scattering reaction con-
sidering an incident photon coming either from a laser beam (assumed flash
energy of 2.5 J, same repetition rate as the electron beam pulse frequency, and
root-mean-square radius of 20 mm), or from a Compton back-scattering
process. They propose that the collision of laser photons of a few eV’s, at
small scattering angle, with an energetic electron beam is able to produce a
highly collimated graviton beam possessing a defined energy spectrum.
Four different e þe linear collider designs were examined by Novaes and
Spehler in their study: 1) SLAC (beam energy E beam 50 GeV and collider
flux-luminosity L 5  1029 cm 2s 1); 2) Palmer-G (E beam 250 GeV,
L 5.85  1033 cm 2s 1); 3) Palmer-K (E beam 500 GeV, L 11.1  1033
cm 2s 1); and 4) VLEPP (E beam 103 GeV, L 1033 cm 2s 1).}}}} The
laser photon energy was assumed to be 1 eV for each of these cases. For the
direct laser photon scattering process, the estimated total graviton radiation
power emitted (for each of the four colliders) ranges from 1.17  10 20 eV/s
(or 1.87  10 39 W, for SLAC) to 1.53  10 18 eV/s (or 2.45  10 37 W, for
VLEPP) [55]. For the back-scattered photon process, the estimated total graviton
radiation power emitted (for each of the four colliders) ranges from 1.69  10 28
eV/s (or 2.71  10 47 W, for SLAC) to 1.32  10 23 eV/s (or 2.11  10 42 W,
for VLEPP) [55]. We expect that the graviton production process outlined in the
previous and following sections will yield rates and emitted radiation power
many dozens of orders of magnitude higher than those estimated for the laser-
e þe collider scheme.
Chen [56] and Chen and Noble [57] reviewed the prospect for high-yield grav-
iton production in the lab using new advanced particle accelerator technology,
and confirmed that the most effective way to produce gravitons is through resonant

}}}}
1 GeV ¼ 109 eV.
208 E. W. DAVIS

photon-graviton conversion in a strong external electromagnetic field à la the Gert-

senshteı̆n effect, which is reviewed in the following section (see also Ref. 58). They
propose to achieve the excitation of 1011 to 1013 V/m plasma waves in an advanced
semiconductor or metal crystal channel collider by using either laser wakefield or
side-injected laser techniques. In crystal channels, the electrostatic fields are known
to be as large as 1016 V/m. Ultrahigh center-of-mass energies possibly approaching
the Planck energy scale will be made possible by the high-field-gradient accelera-
tion of charged particles along crystal channels that collide within these strong-
focusing crystal (atomic scale) channels.
Gravitational synchrotron radiation and gravitational bremsstrahlung radiation
are the two resonant excitation mechanisms that were identified by Chen [56]. In
the former mechanism, the external electromagnetic field interaction with a
charged particle serves as a means to bend its trajectory inside the collider,
thus causing the mass of the particle to behave like a gravitational charge.
Hence, a gravitational charge undergoing acceleration along a curved path will
emit gravitons in the form of gravitational synchrotron radiation. For this case,
the estimated total rate of graviton emission is 10 6 to 103 gravitons per
second with a frequency of 70 to 600 kHz using the design parameters of a
few conventional high energy storage ring facilities [56]. In the latter mechanism,
the collision of e þe beams in a high energy linear collider causes a substantial
fraction of beam energy to be lost through bremsstrahlung when particles are bent
by the strong collective macroscopic electromagnetic field of the oncoming
beam. In this case, the very intense collective field intrinsic to the colliding
beams (104 Tesla) and the very intense bremsstrahlung that penetrates
through such a field gives rise to resonant excitation of gravitational bremsstrah-
lung radiation. For this case, the estimated total rate of graviton emission is
10 25 to 10 22 incoherent gravitons per second and 10 15 to 10 3 coherent
gravitons per second using the design parameters of a few conventional high
energy storage ring facilities [56]. This result should not be too surprising
because high energy accelerators are not designed for optimized gravitational
wave production. Unfortunately, Chen does not provide any gravitational syn-
chrotron and bremsstrahlung radiation emission estimates using advanced
crystal channel collider technology, but only states that their length scale will
have to be enormous (100 km for a single crystal) to achieve graviton yields
that are many dozens of orders of magnitude larger than what could be achieved
using conventional accelerator technology.
As a final note, it is important to point out that a mobile (1 GeV) electron
accelerator less than 3-m in diameter and weighing a few tons was designed
by R. R. Wilson in 1952 [59]. Wilson’s mobile miniaturized accelerator was
designed to consume electrical power of a few kilowatts and achieve a continuous
g-ray beam flux-luminosity 108 cm 2s 1 for a beam projected out to 1-mile
distance with an intensity 1014 photons per second. A mobile miniaturized
accelerator could achieve beam parameters that are many orders of magnitude
larger than Wilson’s original design using present-day accelerator technology.
Therefore it is not out of the question to imagine that a modern-day version of
Wilson’s mobile miniaturized electron accelerator could be properly scaled
and optimized for use as a gravitational wave (graviton) rocket engine. The
question still remains whether the emitted graviton beam power will be of
 REVIEW OF GRAVITY CONTROL 209

large enough magnitude to produce any meaningful rocket propulsion. This is

addressed in Sec. IV.B.5.

4. Gertsenshteı̆n Effect
In this section we review the genre of schemes for producing gravitational
waves that are based upon the Gertsenshteı̆n effect. Gertsenshteı̆n [60] described
a new resonance phenomenon in which a propagating electromagnetic wave can
couple its electromagnetic field-strength tensor to that of a transverse background
electromagnetic field to give rise to a nontrivial stress-energy tensor, which
serves as a source for the linearized Einstein field equation to excite a gravita-
tional wave. In quantum language, this corresponds to a mixing between the pro-
pagating photon and a graviton via a Yukawa-type coupling mediated by a virtual
photon from the background field. Gertsenshteı̆n’s embodiment of this was that a
monochromatic light wave, of angular frequency v, propagating through a
stationary uniform transverse magnetic field produces an outgoing gravitational
wave, also of angular frequency v, with an amplitude that is proportional to
the distance traveled in the background magnetic field. The source of this grav-
itational wave is the mixed stationary-radiative term of the total electromagnetic
field-strength tensor, which is comprised of static and radiative magnetic fields.
The Gertsenshteı̆n effect is considered to be the primary method for the labora-
tory generation of gravitational waves.
Gravitational waves produced in this manner are often called Gertsenshteı̆n
waves (GWs). The production of GWs could not be tested in the lab because
the light and magnetic field intensities required to observe the effect were not
available in the 1960s. Gertsenshteı̆n estimated that only astronomical processes
could be capable of producing GWs. The very high light and magnetic field inten-
sities required to produce GWs became available with the advent of tabletop
ultrahigh-intensity lasers and ultrahigh-pulsed power technologies in the
mid-1990s.
There are several variations of the Gertsenshteı̆n effect published in the litera-
ture. Pustovoı̆t and Gertsenshteı̆n [61] proposed generating synchrotron radiation
via ultrarelativistic particles moving in a constant magnetic field to produce
GWs. A variation of this was the proposal by Sushkov and Khriplovich [62] to
generate synchrotron radiation via ultrarelativistic particles moving in the
Coulomb electric field of a huge electrical charge. Grishchuk and Sazhin [63]
devised a modified Gertsenshteı̆n effect in the form of an electromagnetic reso-
nator with a toroidal shape and rectangular cross section in which a periodic elec-
tromagnetic field is produced. The periodic electromagnetic field is the time-
dependent part of the electromagnetic field stress-energy tensor in the general
relativistic field equation, and thus serves as the source of GWs. The result is a
standing cylindrical gravitational wave that is produced in the focal region of
the resonator. The interference focusing of the gravitational radiation within
the resonator occurs via the interference of coherent gravitational waves freely
propagating from different parts of the resonator, because gravitational waves
cannot be reflected by the material walls of the resonator cavity. Portilla and
Lapiedra [64] developed a concept that produces both electromagnetic and grav-
itational waves by shaking an electric charge in a homogeneous stationary
210 E. W. DAVIS

magnetic field while the system is inside either a homogeneous or inhomo-

geneous dielectric medium. Their work dramatically improved upon the Gert-
senshteı̆n effect and its ability to produce GWs. And last, Navarro et al. [65]
take the previous scheme in another direction by proposing that high-frequency
GWs can be produced by irradiating a dielectric film in a resonant cavity.
The early Russian schemes all suffer from extremely low electromagnetic-to-
GW conversion efficiency, on the order of or lower than that estimated for the
photon decay process described in Sec. IV.B.2, and therefore, extremely low
GW luminosity (or power) output. The generic GW luminosity (LGW) produced
by the schemes described previously is given by [64]:

84pGB2 FEM D
LGW ¼ (35)
5ðn2  1Þ2 v4
where B is the static magnetic field intensity inside the resonator cavity, FEM is
the incident electromagnetic (light) flux with angular frequency v, n is the index
of refraction of the diffracting dielectric medium inside the resonator, and D is a
dimensionless parameter that depends on the combined refractive index of the
dielectric medium and vacuum pressure inside the resonator. Equation (35)
shows how important it is to use ultrahigh field intensities to produce a measur-
able GW luminosity. A better quantitative measure for the strength of the
Gertsenshteı̆n effect is given by the ratio of LGW to incident electromagnetic
luminosity, LEM [65]:
 2  2
LGW 9 B 10 9 Torr
¼ (5  10 )
LEM 1 Tesla p
    (36)
2
2p  1010 Hz Qf

v 5,000

where p is the vacuum pressure inside the resonator and Qf is the resonator’s quality
factor (defined as 2p times the ratio of the energy stored in a resonator to the energy
lost per cycle). As an illustrative example, we use Eq. (36) to estimate the LGW
produced if the National Institute of Standards and Technology’s Synchrotron
Ultraviolet Radiation Facility (SURF) were used to provide the incident light
flux. The SURF UV beam luminosity is LEM  106 W with v  2p  1012 Hz,
and we assume that the resonator has a magnetic field intensity B  10 Tesla
with p  10 9 Torr and Qf 5000 (typical for many resonator designs). Using
these parameters, we can expect to generate GWs with LGW  50 mW, which
should be within reach of carefully designed high-frequency gravitational wave
detectors. It is beyond the scope of the present discussion to review the large
number of proposed gravitational wave detectors and their underlying physics.

5. Make or Break Issues: Gravitational Wave Rockets

According to General Relativity, gravitational waves (or gravitons) must
propagate at the speed of light. This means that the exhaust velocity (ne) of a
 REVIEW OF GRAVITY CONTROL 211

gravitational wave rocket will also be the speed of light, as it is for photon
rockets. And the specific impulse (Isp) of a gravitational wave rocket is
Isp ne/g c/g  3.06  107 s, as it also is for photon rockets. This means
that gravitational wave rockets will operate at the maximum efficiency possible.
Furthermore, the propulsive performance (b) of a rocket propulsion system is
measured by its ability to generate thrust (Fthrust) for a given unit of exhaust jet
power (Pjet): b Fthrust/Pjet 21ff/ne, where 1ff ( 1) is the efficiency involved
in the conversion of stored energy into the kinetic energy of the exhaust stream.
For conventional chemical rockets (e.g., the space shuttle main engines),
ne  4.45  103 m/s and 1ff  0.5, hence b  2.25  10 4 N/W. For gravita-
tional wave rockets, b 21ff/c  (6.67  10 9)1ff N/W, where 1ff will be deter-
mined by the particular mechanism chosen to produce gravitational waves. By
comparing these two results, we see that gravitational wave rockets require
.104 times the jet power of chemical rockets in order to deliver the same 1 N
of thrust.
In the previous sections we learned that there are numerous proposals for pro-
ducing gravitational waves in the lab. The make or break issues for gravitational
wave rockets reside within the ability of any one of the proposals to achieve the
production of high-intensity, high-frequency gravitational waves with reasonable
efficiency. The electromechanical quadrupole octupole technique will require a
tradeoff between the mass and size scaling of the rocket and the magnitude of the
internal quadrupole-octupole oscillations that can be achieved. Our analysis
showed that only astronomical-sized objects having physically reasonable
internal quadrupole octupole oscillations can achieve high thrust and high
final velocities. If gravitational waves, or gravitons, are produced via electromag-
netic-gravitational resonance or resonant photon decay processes, then the make
or break issues reside within the ability to achieve the required ultrahigh (exter-
nal) electromagnetic field intensities. Tabletop ultrahigh-intensity laser technol-
ogy has reached the threshold where this becomes possible to do. However,
producing gravitational waves via conventional and advanced particle accelerator
techniques does not appear to be feasible at present because the gravitational
beam luminosity and power output is estimated to be too minute for effective
rocket propulsion. More theoretical and nascent experimental work needs to be
done in all of these areas to identify the best approach for producing gravitational
waves at a level that will support a possible rocket propulsion application.

C. Quantum Antigravity Propulsion

Quantum antigravity can be found within the very large genre of quantum gravity
theories in which repulsive gravity terms appear as quantum corrections to the clas-
sical Newtonian gravitational force law. Generally, one can derive such correction
terms by quantizing the Einstein general relativistic field equation or by starting
with a particular type of quantum field theory (e.g., supersymmetric field theory,
quantized five-dimensional Kaluza Klein unified field theories, quantum super-
strings/D-Brane theory, quantum loops or knots, Yang Mills theories, etc.) and
work backward to find the corresponding gravity theory. The particular mathemat-
ical form and quantitative magnitude that quantum correction terms can have
depends totally upon the quantization procedure and order of approximation used
212 E. W. DAVIS

in a given quantum gravity theory. However, the linearized semi-classical quantum

gravity theory is related to Einstein’s classical nonlinear General Relativity Theory
whereby the former uniquely implies the latter, provided that the graviton, which
exchanges the gravitational force between two massive particles or photons, is a
pure spin 2 particle. In this theory, the stress-energy tensor of the matter fields is
quantized while gravitation (via the Einstein curvature tensor) is still treated classi-
cally. Semi-classical quantum gravity is a quantum field theory in curved spacetime
that has been successful in reproducing a few of the predictions and many of the
foundational precepts of General Relativity Theory.
A particular example of what a quantum antigravity correction term looks like
was derived in 1984 by R. L. Forward and this author, with instruction provided
by R. P. Feynman and M. Scadron, during a summer quantum gravity seminar
sponsored by the Hughes Research Laboratories in Malibu, California. We
began by studying the Feynman quantization procedure for the case of single-
photon exchange between two charged particles, which tells us about the under-
lying nature and quantum corrections to the static Coulomb force. From this
study we discovered that the same is also true for the case of single-graviton
exchange between two massive spin 0 particles in connection with the static
Newtonian force. By applying Feynman’s quantization procedure [66 68] to
the linearized Einstein field equation in the nonrelativistic limit, we derived
the following static graviton-exchange potential [Vgrav(r)] for two spin 0 particles
undergoing a gravitational interaction:

2
Gm1 m2 4pGh 3
Vgrav (r) ¼  þ d (r) (37)
r c2

where m1 and m2 are the masses of the interacting particles, r is their radial sep-
aration, and d3(r) is the three-dimensional Dirac d-function with r the position
vector of some reference point in space. The first term in Vgrav(r) is immediately
recognized as the attractive Newtonian gravitational potential, while the second
quantum correction term is repulsive. Also, the second term is independent of the
interacting particle masses and can only be measured for bound quantum s-states
because the product of the coefficient 4p (Gh 2/c 2)  10 94 with the d-function
gives only a minute physical effect at the atomic scale. The second term
happens to be analogous to the usual quantum correction to the Coulomb or
nuclear force. If the two particles were to have non-zero quantum spin, then
Vgrav(r) will be modified by additional spin orbit and spin spin correction
terms. Furthermore, there are additional velocity-dependent corrections to
Vgrav(r) that generate the general relativistic post-Newtonian modifications of
the classical equation of motion of a particle in a gravitational field.
But the most important characteristic to observe about the quantum antigravity
correction term in Vgrav(r) is that its magnitude is incredibly minute, only affect-
ing bound quantum s-states. In general, quantum gravity correction terms at any
level of approximation, whether gravitationally repulsive or attractive, will have
coefficients G(h d/c k) (for d, k . 1), and therefore will not have a measurable
impact on any macroscopic system that embodies any form of breakthrough pro-
pulsion. Because these quantum corrections are so minute, and because there is
 REVIEW OF GRAVITY CONTROL 213

no single universally accepted quantum gravity theory to work with, investigators

have had little reason to look into the potential application of quantum gravity
correction terms to breakthrough propulsion physics.
However, this is not the entire story because there are many interesting
quantum field theoretic phenomena that exist outside of that which arise in
quantum corrections to Newtonian gravity. In what follows, we review the
recent discovery of antigravity forces that arise within both QED vacuum fluctu-
ation and nonretarded quantum interatomic dispersion force theories in curved
spacetime, as well as two highly speculative concepts that invoke some form
of quantum unified field theory.

1. Antigravity via Quantum Vacuum Zero-Point Fluctuation Force

Calloni et al. [69,70] explored the possibility of verifying the equivalence
principle for the zero-point energy of QED. They used semi-classical quantum
gravity theory to evaluate the net force produced by quantum vacuum zero-
point fluctuations (ZPF) acting on a rigid Casimir cavity in a weak gravitational
field. Their analysis assumed the rigid Casimir cavity to be a nonisolated system
at rest in the Earth’s gravitational field, which is modeled using the standard
Schwarzschild spacetime metric geometry, so that they could evaluate the regu-
mv
larized (or renormalized) stress-energy tensor, kTvac lren, of the quantized vacuum
electromagnetic field between two plane-parallel ideal metallic plates lying in a
mv
horizontal plane. kTvac lren encodes the Casimir effect, which has a negative
energy density and a negative pressure along the vertical (acceleration) axis
between the plates. (See Secs. II and V.B.1 in Chapter 18 for more information
about the Casimir effect.) Their results agreed with the equivalence principle
because they showed that quantum vacuum ZPF (i.e., virtual quanta) do gravitate
because the energy of each ZPF mode is redshifted by the factor ( g00)1/2
[1 (2GM/c 2r)]1/2 even though the modes remain unchanged. In other
words, the electromagnetic vacuum state in a weak gravitational field is
redshifted. This effect remains true for strong gravitational fields.
The resulting antigravity force (FCasGrav) derived by Calloni et al. is [70]:

p 2 Ah g
FCasGrav ¼
180cd 3
(38)
 43
A
 1:89  10 (N)
d3
where A is the area of the plates and d is their separation. Equation (38) states that
a Casimir device in a weak gravitational field will experience a tiny push in the
upward direction (i.e., the opposite direction with respect to the Earth’s gravita-
tional acceleration). This is consistent with the interpretation that the negative
Casimir energy in a gravitational field will behave like a negative mass (see,
e.g., Sec. II.D). FCasGrav is actually the sum of two separate force terms: the
00
first term arises from the Casimir energy encoded in kTvac lren, which is interpreted
as the Newtonian repulsive force on an object with negative energy, and the
second term arises from the pressure along the vertical (acceleration) axis,
214 E. W. DAVIS

which is interpreted as the mass contribution of the spatial part of the stress-
energy tensor. To evaluate FCasGrav for the case of any gravitating body of inter-
est, one must replace g in Eq. (38) with Eq. (2).
Calloni et al. further point out that a real Casimir cavity is an isolated system
in which the actual (total) resulting force is the Newtonian force on the sum of the
rest-Casimir energy and rest-mechanical mass whereby the contribution of the
vacuum ZPF leads to a gravitational repulsion (FCGexp) on the Casimir device
that is given by [70]:

1
FCGexp ¼ FCasGrav
4
(39)
 44
A
 4:73  10 (N)
d3

which is the force that should be experimentally tested. Equation (39) takes into
consideration that the contribution to the total force on a real cavity resulting
from the spatial part of the stress-energy tensor is balanced by the contribution
from the mechanical stress-energy tensor. Given that the typical dimensions of
a Casimir device are very small, it appears that FCGexp will be very difficult, if
not impossible, to measure using present-day lab technology.
However, Calloni et al. propose an experimental device that could signifi-
cantly magnify the repulsive force up to a measurable scale. Their proposed
device is a multilayered series of rigid Casimir cavities with each cavity consist-
ing of two thin metallic disks that are separated by a dielectric material which is
inserted to maintain rigidity. They suggest SiO2 for the dielectric material
because it is an efficient dielectric with low absorption over a wide range of
frequencies, and it is an inexpensive material that is easy to fabricate into
layers. The introduction of the dielectric material is equivalent to enlarging the
optical path length by the refractive index n so that the cavity plate separation
d 7! nd. The Casimir effect has been tested down to plate separations
60 nm, while separations 10 nm is possible with present technology. At
10 nm distances, dielectric absorption and finite conductivity are expected to
decrease the effective Casimir pressure compared to a cavity comprised of
perfect mirrors. For example, a plate separation of 6.5 nm corresponds to a
decreasing factor z of 0.07 for plates made of aluminum. Finite temperature
and plate surface roughness could also introduce additional corrections to the
Casimir pressure. Calloni et al. propose to magnify the total force by using
N‘ 106 layers of rigid cavities with each cavity having a diameter of 35 cm
and thickness of 100 nm, for a total device thickness of 10 cm.
All these engineering factors taken together led Calloni et al. to recast FCGexp
into the following new form [70]:

p 2 Ah g
FCGexp  zN‘
720c(nd)3
(40)
 44
 zN ‘ A
 4:73  10 (N)
(nd)3
 REVIEW OF GRAVITY CONTROL 215

Calloni et al. also suggest that a feasible experiment will require modulating
FCGexp in order to obtain a measurable force. They are investigating the possi-
bility of modulating z by varying the temperature in order to induce a periodic
transition from conducting state to superconducting state. They estimate that
doing this could achieve zmax  0.5, and thus produce a force FCGexp  10 14
N at a modulation frequency on the order of 10s of mHz for d 5 nm and
n 1.46 (for SiO2 dielectric). This result is more than two orders of magnitude
larger than the force that the VIRGO gravitational wave antenna is expected to
detect at several 10s of Hz. If one could fabricate a device consisting of 109
layers, then FCGexp  10 11 N. This suggests that cavities made from thin-film
deposited surfaces or photonic band-gap materials would be the best approach
for fabricating a multilayer Casimir device.
Bimonte et al. [71,72] also derived Eq. (38) for this very same problem by
using Green-function techniques in the Schwinger DeWitt quantum ether pre-
mv
scription for kTvac lren in a curved spacetime. They also computed the weak grav-
itational field-induced correction terms for the Casimir pressure on the plates
00
kTvac lren and the total energy ECasGrav stored in the Casimir device, which is
given by [71]:
 
p 2 Ah c 5 gd
E CasGrav ¼ 1þ 2 (J) (41)
720d 3 2c

The correction terms for the different (measurable) physical quantities of interest
are generally g/c 2.
Finally, Calloni et al. point out that the overriding concern with performing an
experiment to test FCGexp is whether cavities can be made sufficiently rigid, if the
effect of surface roughness and defects can be quantified to improve the force
estimate, and if the necessary signal modulation can be achieved in the lab.
However, micro- and nano-manufacturing is maturing to the point where rigidity,
surface roughness, and close plate separations are becoming routinely controlla-
ble. While the numerical estimate for FCGexp is quite feeble, it is still significant
because it is at the very low end of the macroscopic scale, and it might be possible
to devise advanced methods to magnify the force to a magnitude that benefits a
propulsion application. However, the upward force will have to be larger than the
weight of the propulsion system in order to achieve levitation. This could be very
difficult to do, but this is a concept ripe for further exploration.

2. Antigravity via Nonretarded Quantum Interatomic Dispersion Force

Pinto [73] evaluated the net lifting force produced by nonretarded electrostatic
dipole dipole interactions (i.e., nonretarded van der Waals dispersion forces)
acting on a quantum system of polarizable particles in a curved spacetime. The
foundation of Pinto’s study was the original discovery made by Fermi [74] that
classical electrostatic theory must be reformulated in a curved spacetime in
order to properly evaluate the effects of gravitation upon the Coulomb electric
field of a single charged particle. In this case, the Laplace equation of electro-
statics for a single charged particle can be generalized in the presence of a grav-
itational field and then extended to show that a system of classical charged
216 E. W. DAVIS

particles undergoes a gravity-induced self-lifting force. Fermi and other investi-

gators arrived at this counterintuitive result by computing the gravity-induced
self-force acting on an isolated electric dipole in a weak gravitational field and
showing that the self-force (times dipole size) is exactly equal to the gravitational
equivalent of the electrostatic internal energy of the dipole.
The net gravity-induced (electrostatic levitation) self-force (FDipGrav) is given
by [73]:
  
q2e g
FDipGrav ¼ (N) (42)
4 p 1 0 r c2

where qe is the electric charge on a particle and r is the radial distance between
two charged particles in the dipole. There is an additional term of order g 2/c 4 in
FDipGrav that is neglected because it is negligible in magnitude. Equation (42)
states that an electric dipole will experience a push in the upward direction (oppo-
site direction with respect to the Earth’s gravitational acceleration), that is, the
dipole undergoes self-acceleration in which one charged particle in the dipole
appears to be chasing the other charged particle. As an example, for a dipole com-
prised of two charges (e.g., an electron-proton system) held at fixed r to levitate in
the Earth’s gravitational field, r would have to be  10 15 m (the size of an
atomic nucleus). An experiment to test this prediction on such a small scale is
too difficult to control or measure.
An energy analysis done by Pinto showed that there is a distance r between
two charges (each of rest-mass m0) in a dipole (of mass Mdip 2m0) such that
their electrostatic potential energy, Udip q2e /4p10r, becomes equal to the
unrenormalized mass of the system as r ! 1. At this distance, the effective
total gravitational mass Mdip þ Udip/c 2 0 and the self-force alone can
support the dipole at rest against its own weight. The self-acceleration of the
dipole is such that the acceleration process can continue indefinitely, which
poses a problem for energy conservation because the dipole can be left to self-
accelerate for an arbitrary period of time and then stopped to harness the resulting
kinetic energy. This process could be used to extract unlimited energy from the
system. Pinto claims that there is no conflict with energy conservation because
the renormalized inertial mass of the accelerating system is Mdip-ren Mdip þ
Udip/c 2 0 and the total energy of the system is zero at all times regardless
of speed. This claim requires re-evaluation because there are subtle boundary
conditions involved that might have been overlooked in the analysis.
Fermi’s discovery led to a new subfield of research devoted to the study of
electrodynamics and dipole and interatomic dispersion forces in a curved space-
time. Pinto’s theoretical program extended the result of these studies by consid-
ering a system of polarizable atoms and adopting an approach in which the effect
of a gravitational field in General Relativity is modeled as an effective optical
medium. In other words, the spacetime vacuum is treated as a nonuniform
optical medium with a varying index of refraction that defines the components
of a flat spacetime metric geometry [43,75]. There is no spacetime curvature
due to sources of matter in this model, instead its equivalent general relativistic
effects (i.e., gravitation) are produced by varying the vacuum index of refraction,
 REVIEW OF GRAVITY CONTROL 217

comprised of the vacuum electromagnetic permittivity and permeability para-

meters, in response to the presence of matter sources. Pinto’s lengthy analysis
gives the van der Waals dispersion self-force for two polarizable atoms in a
curved spacetime (i.e., a weak gravitational field) as [73]:
g
(0)
FvdWGrav ¼ UvdW
c2
 
6q2e a50  g 
¼  (43)
4p10 r 6 c2
 57
 q2e
 2:44  10 (N)
r6

where a0 is the Bohr radius (5.292  10 11 m), r is the radial distance between
two atoms, and U(0) vdW is the flat spacetime van der Waals (interatomic potential)
interaction energy to second-order in quantum perturbation theory. Pinto used
Eq. (43) to estimate the gravity-induced self-acceleration (alift) for the case of
two hydrogen atoms in their ground state at r 20a0, and found that alift,H
FvdWGrav/2mH  4  10 15 m/s2 (mH mass of hydrogen atom). For the case
of two positronium (Ps) atoms, he found that alift,Ps  8  10 12 m/s2.
Pinto’s strategy is to dramatically magnify FvdWGrav to a large enough magni-
tude that it becomes viable for space propulsion applications. He claims that this
(0)
can be done by manipulating UvdW , which is dependent on the atomic polarizability
and strongly affected by the quantum state in which the atoms are prepared. Inter-
atomic forces can also be manipulated by means of external electromagnetic fields
that can transform van der Waals forces into a first-order interaction. He evaluated
a number of schemes and settled on the following techniques for manipulating dis-
persion forces: 1) excitation of polarizable atoms to Rydberg states in external
time-dependent electric fields, 2) polarizability resonant enhancement by laser
radiation, and 3) laser-induced near-zone orientational average of the dispersion
force. Also, in order to generate a macroscopic self-lifting force, it will be necess-
ary to apply these techniques to a cluster of trapped atoms because the total self-
lifting force acting on the center-of-mass of a trapped gas composed of Na identical
polarizable atoms is N2a times the self-lifting force acting on a single pair of inter-
acting atomic dipoles. Item 1 has a two-part contribution to the magnification of the
self-lifting force: a) one part from a2(v)E 2 due to the effect of external time-
dependent electric fields on atomic polarization, where a(v) is the atomic polariz-
ability as a function of the electric field frequency v and E is the electric field
intensity; and b) another part from using highly-excited Rydberg atoms (with prin-
cipal quantum number np 1 and Bohr radius an n2pa0) whose polarizability
scales as n7p. Item 2 leads to a magnification by factors of a(v)/a0  103 to 105
(a0 is the static value of the polarizability) via detuning of the laser excitation radi-
ation frequency from the nearest atomic transition resonance of the atoms in the
trapped cluster. Item 3 leads to a further magnification due to the effect of the
incident laser radiation on the dispersion force being averaged over all
directions, which changes the interatomic potential (/1/r 6 ) into a gravity-like
1/r potential.
218 E. W. DAVIS

Pinto’s study suggests that the combined effect of items 1 to 3 will magnify the
self-lifting force to the point where a cluster of trapped atoms will not only hover
unsupported in the Earth’s gravitational field, but will also generate an additional
upward thrust. On the basis of extensive theoretical and empirical studies, along
with the typical parameters for laboratory laser and optical atomic matter trap
technologies, he estimates that alift
1.5-g (in the upward direction). Trapped
atom gravimeters can be used to observe this effect in the lab. Pinto also
points out that other polarizable systems such as nanoparticles, microspheres,
and quantum dots can be used in place of atoms. The trapping of latex spheres
into a form of optical matter by means of intense laser radiation has already
been demonstrated in the lab. In addition, an analogy to the item 1 to 3 manipula-
tions that produce dramatically enlarged polarizabilities in trapped interacting
nanoparticles and microspheres have also been demonstrated in the lab.
Pinto proposes a levitation propulsion thruster in which the combined system of
trapped interacting polarizable particles and external confining fields forms a single
thruster element comprising a fraction of the mass of the entire vehicle. The reaction
of the self-lifting force exerted by this element against the external confining fields
results in the transfer of force (thrust) to the entire vehicle. In order to achieve levita-
tion, this requires that the upward thrust per polarizable particle be larger than its
own weight if the fraction of the thrusting mass is smaller than the mass of the
rest of the vehicle. The propulsive levitation condition is expressed as [73]:
Fthrust (Mveh þ mAN)g or Fthrust/mANg
1, where Fthrust is the total gravity-
induced thrust, Mveh is the vehicle mass, mA is the mass of individual polarizable
particles, and N is the total number of trapped polarizable particles.
Pinto identified numerous technical challenges that will have to be overcome
before this concept can be put to practice. One challenge is that polarizability res-
onant enhancement also leads to atomic transitions and decay that result in the
recoil and evaporation of atoms from inside the trap. Another is the difficulty
of maintaining continued confinement of a trapped cluster of polarizable particles
in a specific three-dimensional array while the cluster is simultaneously opposing
the amplified interatomic forces and producing thrust. The confinement lifetime
of trapped polarizable particles is finite and there is the possibility that these par-
ticles might be evaporated away or destroyed in a time that is too short to deliver
the required thrust to the vehicle. Therefore, a scheme for active repopulation of
the trapped cluster will have to be developed. The design of particle cluster traps
and associated external confinement fields are of primary importance to deter-
mine the effective thrusting time of every polarizable particle. In addition,
Rydberg atoms suffer from finite radiative lifetimes and are sensitive to external
perturbations, so dispersion force manipulation might lead to the ionization of
atoms. Tradeoffs will have to be made between all of the relevant system
parameters in order to discover the “sweet spot” that achieves levitation and
upward acceleration. These and other yet-to-be-identified technical challenges
need to be addressed via further empirical and theoretical studies.

3. Heim’s Quantum Theory for Space Propulsion

Heim [76 79], Heim et al. [80], and Dröscher and Heim [81] (a.k.a. the Heim
theory group; see also Ref. 82), propose a unification of the quantum field theory
 REVIEW OF GRAVITY CONTROL 219

of elementary particles with General Relativity via a six- or eight-dimensional

spacetime geometry, called Heim space. Heim space is a quantized space com-
prised of elemental surfaces with orientation (spin) and a size  (Planck
length)2 (h G/c 3)  10 70 m2. Heim space may also comprise several sub-
spaces, each of which are equipped with their own space metric. Combining
these subspaces by employing certain selection rules, a set of partial metric
tensors is obtained which form a polymetric that represents all of the known fun-
damental interaction forces. The partial metrics are interpreted as a physical
interaction field or elementary particle. However, Heim theory actually predicts
two more fundamental interaction fields in addition to the four experimentally
known ones.
The six fundamental interactions emerge in our four-dimensional spacetime
and represent real physical fields carrying energy. The two additional interaction
fields are identified as a gravitophoton interaction (i.e., the conversion of photons
into a gravitational-like field) and a quintessence or vacuum interaction (also a
conversion of photons into another type of gravitational-like field).
The gravitophoton interaction is mediated by two hypothetical massless gravito-
photon particles, one that is gravitationally attractive and the other that is grav-
itationally repulsive. The massless quintessence interaction particle mediates a
very weak repulsive gravitational-like force that is much smaller in magnitude
than the two gravitophoton interactions. Therefore, according to Heim theory,
the gravitational force that we experience is comprised of three different funda-
mental interactions having four different interaction quanta, namely the usual
gravitons of Heim-quantized General Relativity Theory, both repulsive and
attractive gravitophotons, and the repulsive quintessence (or vacuum) particle.
All of the related interaction coupling constants were derived and numerically
calculated by the Heim theory group. The gravitophoton and quintessence inter-
actions are Lorentz-invariant, and the laws of momentum and energy conserva-
tion are strictly obeyed. However, the speed of light is not the limiting speed for
an inertial transformation in Heim theory. The gravitophoton interaction also has
the peculiar property of reducing the inertia of a body.
The physical interpretation for the gravitophoton interaction field led the Heim
theory group to conclude that this field could be used to both accelerate a body
and to cause its transition into a parallel space that possibly admits superluminal
motion. Dröscher and Häuser [83 88] propose that these effects act as two sep-
arate modes of space propulsion that could lead to gravity control and reduction
of inertia at low energy densities. The first propulsion mode is based on the
prediction that a Heim Lorentz force Fgp arises as a byproduct of the gravitopho-
ton interaction. This force is structurally similar to the magnetic Lorentz force
[83 88]:
 
Fgp ¼ Lp em0 vT  B (N) (44)


The use of the word “quintessence” by the Heim theory group is spurious because it is
defined by the cosmological research community as a time dependent vacuum scalar field. The use
of this term is not justified because Heim theory indicates that the gravitophoton vacuum interaction
field has no time dependence and its coupling constant is fixed.
220 E. W. DAVIS

where Lp is a dimensionless (highly-nonlinear) function of the probability ampli-

tude of the gravitophoton particle, e is the electron charge, m0 is the vacuum per-
meability constant, nT is the circumferential velocity of a rotating body of mass
m, and B is the magnetic field vector. The Heim Lorentz force is a repulsive
gravitational force that is generated by the pair production of gravitophotons
(pair production energy is zero). Equation (44) states that an accelerating force
is generated by a freely rotating ring-magnet located above a current-loop that
produces an external inhomogeneous magnetic field surrounding the ring-
magnet, all of which gives rise to gravitophoton pair production via the conver-
sion of photons (that couple the electrons inside the current-loop) to the virtual
electrons inside the ring-magnet. Because of the relative strength of their coup-
ling constants and corresponding interaction cross sections, the repulsive gravi-
tophoton gives rise to a propulsive force that acts upon the rotating ring, while
the attractive gravitphoton is absorbed by the atomic nuclei comprising the
ring material. Furthermore, the virtual electrons are quantum vacuum zero-
point fluctuations that surround the atomic nuclei comprising the ring, which
are polarized by the externally applied magnetic field. The role of the rotating
ring’s circumferential speed, the virtual electrons inside it, and the external mag-
netic field is to catalyze the photon-to-gravitophoton conversion process inside
the ring so that the kinetic energy of the ring, under acceleration by the
Heim Lorentz force, is provided by the quantum vacuum and not by the external
magnetic field.
This configuration automatically suggested an experiment that Dröscher and
Häuser have detailed in their publications whereby an apparatus to test the
Heim Lorentz force prediction will also serve as a prototype “gravitophoton
field propulsion” device. Dröscher and Häuser calculated the magnitude of the
Heim Lorentz force for a range of ring rotation rates, external magnetic field
strengths, ring size and mass, etc. They predict Heim Lorentz forces ranging
from 7.14  10 43 N to 1.45  109 N for the case of a 100 kg ring (diameters
ranging from 1 m to 6 m) that is embedded in an external magnetic field
(ranging from 2 Tesla to 50 Tesla), and rotating with a circumferential speed
ranging from 103 m/s to 700 m/s [83 88].
The second mode of gravitophoton field propulsion is highly speculative.
Dröscher and Häuser suggest that the gravitophoton interaction can also be
exploited to reduce the inertial mass of a spacecraft by a factor of 104, which
will result in the increased speed of the spacecraft by a corresponding factor of
104 without increasing its kinetic energy. (This appears to violate the conserva-
tion of energy.) The inertia reduction is what causes the spacecraft to slip into a
parallel space where it undergoes superluminal motion. The physical process that
allegedly makes this possible is the emission of repulsive gravitophotons by the
gravitophoton field propulsion device (using a 60 Tesla magnetic field), which
convert the spacecraft’s own gravitons into repulsive quintessence particles
[83 88]. This process is claimed to reduce the gravitational potential of the
spacecraft, thus reducing its inertia. The explanation given by Dröscher and
Häuser for the parallel space transition process is somewhat obscure.
The reader should be cautioned that the edifice of Heim’s theory is very soph-
isticated and mathematically very dense. The efficacy of its theoretical foun-
dation has not been proven and its theoretical predictions have not been tested
 REVIEW OF GRAVITY CONTROL 221

in the lab nor published in the peer-reviewed literature. However, at the time
of this writing, the Heim theory group is working to establish their first lab
experiment.

4. Alzofon’s Antigravity Propulsion

Alzofon [89] began exploring the fundamental nature of the gravitational
force during the 1950s 1960s, and began to question whether it could be con-
trolled by artificial means. Many of his contemporaries in the aerospace engineer-
ing and sciences community were also exploring this question at a time when
several large aerospace industry firms were establishing their own gravity
physics research institutes in the United States. The goal of these institutes was
to find a physical principle that would lead to the development of a technology
for the artificial control of gravity in order to reduce, or otherwise, eliminate
the influence of the Earth’s gravitational field upon aerospace vehicles. The
hope was to revolutionize aerospace transportation by eliminating the need to
overcome the Earth’s gravity.
Alzofon later began a phenomenological study in which he proposed that a
specimen composed of a very pure isotope of aluminum (e.g., Al27) with small
spherical iron inclusions (or magnesium plus chromium inclusions), embedded
in a static magnetic field, and undergoing excitation by pulsed microwave radi-
ation, would generate a net decrease of the alloy’s weight to the point of levitation
[90]. This embodiment is said to alter the force of gravity by means of altering
nuclear entropy using dynamic nuclear orientation via pulsed polarization of
the magnetic moments of nucleons, and its interaction with polarized electron
spins. In this mechanism, the paramagnetic electron spin resonance and align-
ment disalignment cycle resonance of pulsed dynamic nuclear magnetic
moment orientation inside the specimen modifies the interaction between
nucleons and spacetime metric fluctuations, thus reducing the specimen’s inertial
mass. This is hypothesized to result in a corresponding reduction of the Earth’s
gravitational force acting upon the specimen [90,91].
At present, efforts are underway in private laboratories to expand the empirical
envelope of Alzofon’s original embodiment to see whether a measurable altera-
tion can be produced in the force of gravity acting on specimens. However, a
complete theoretical model for this concept is required in order to provide
self-consistent predictions that can be used to guide any empirical studies. No
complete, self-consistent theoretical model yet exists for this genre of gravity
control devices, and Alzofon’s model is a hypothesis at best. However, empirical
studies have been known throughout history to guide the way toward the devel-
opment of mature theoretical models when such is largely absent at the begin-
ning. This is how the laws of gravitation and motion, electrodynamics, optics,
quantum theory and many others got their start.

V. Conclusions
In this chapter we reviewed and analyzed a broad number of gravity control
concepts that are found within Newtonian gravity theory, General Relativity
Theory, semi-classical quantum gravity theory, quantum field theory, nonretarded
222 E. W. DAVIS

quantum interatomic dispersion force theory, and speculative quantum unified

field theories. We found that plausible mechanisms exist within Newtonian and
general relativistic theories whereby one could embody a realistic device that
produces gravity control or an antigravity force. However, we discovered that
there are daunting technical challenges that arise in each of the proposed embo-
diments. Mechanical embodiments that produce antigravity forces require
kilometer-sized apparatus, astronomical-sized masses and densities, or extreme
mass velocities and accelerations. There are other subtleties involved, such as
the possibility of different forms of matter having a highly nonlinear gravitational
permeability, which could dramatically mitigate such large-scale requirements.
Embodiments of antigravity devices that use negative matter are speculative
because negative matter has not been observed in nature or in the laboratory.
Dark energy (i.e., the vacuum energy of Einstein’s cosmological constant) and
dark matter, however, have both been observed. Our analysis showed that dark
energy has no useful breakthrough propulsion application even though it has
an antigravity property that drives the inflationary expansion of the universe.
And it turns out that dark matter has no special antigravity property, so we can
ignore it. Negative energy has been produced in the lab in very small quantities.
The technologies used for producing negative energy are nascent, and so it will be
some time before it can be ascertained whether they are capable of producing the
astronomical amounts of negative energy required to generate significant anti-
gravity forces.
Also, embodiments that produce artificial gravity or gravitational waves for
propulsion require electromagnetic field intensities approaching the breakdown
field intensities of the quantum vacuum, or they require particle accelerator ener-
gies that approach or exceed Planck energy. Modern laser technology has reached
field intensities approaching vacuum breakdown field intensities while pulsed
power technology is beginning to catch up. There is ongoing theoretical and
empirical testing of new alternative accelerator technologies designed to
achieve Planck energy on a tabletop. It is expected that particle beam luminosity
and power output from tabletop accelerators could be more than a dozen orders of
magnitude higher than what is achieved by conventional large-scale accelerator
technology. This is an ongoing development in experimental particle physics.
Antigravity forces generated by the Casimir effect or by nonretarded quantum
interatomic dispersion forces in a curved spacetime (i.e., gravitational field) are
very feeble, but there are proposals based on other theoretical and empirical
studies which suggest that these forces can be amplified to macroscopic level.
However, there are a number of difficult technical challenges to overcome in
order to achieve success. There are other antigravity concepts based on speculat-
ive quantum unified field theories, but it is beyond the scope of this chapter to
review all of them. However, two of the more well-known concepts were
reviewed, and our evaluation showed that they are highly speculative and
remain subject to theoretical and/or empirical verification.
The field of gravity control is ripe for focused, dedicated research. Laboratory
technology continues to evolve and push the envelope on what is possible.
Extreme field strengths, power, luminosity, and even mass motion are being
achieved by tabletop devices. So the technological means for exploring extreme
physical conditions in the lab exist and should be used to explore gravity control.
 REVIEW OF GRAVITY CONTROL 223

Acknowledgments
The author thanks the Institute for Advanced Studies at Austin and H. E.
Puthoff for supporting this work. Portions of this work previously originated
under Air Force Research Laboratory (AFMC) contract F04611-99-C-0025.
The author acknowledges the entire body of technical contributions made to
gravitational and breakthrough propulsion physics research over four decades
by R. L. Forward. Bob Forward was both a personal and professional mentor
to this author as well as to many others, and he will not be forgotten. The
author also extends many thanks to F. B. Mead, Jr. (AFRL/PRSP, Edwards
AFB, CA) and Brig. Gen. S. P. Worden (USAF ret., NASA-Ames) for encoura-
ging and supporting our exploration into this topic. Finally, I thank M. G. Millis
(NASA-Glenn), E. H. Allen (Lockheed Martin), J. Häuser (Univ. of Applied
Sciences), G. Hathaway (Hathaway Consulting), M. Tajmar (Austrian Res.
Centers), R. P. Feynman (CalTech), and M. Scadron (Univ. of Arizona) for
many useful discussions.

References
[1] Symon, K. R., Mechanics, 3rd ed., Addison Wesley, Reading, MA, 1971, pp. 10.
[2] Forward, R. L., Future Magic: How Today’s Science Fiction Will Become Tomor
row’s Reality, Avon Books, New York, 1988, pp. 103 132.
[3] Forward, R. L., Indistinguishable from Magic, Baen Books, New York, 1995,
pp. 148 178.
[4] Forward, R. L., “A New Gravitational Field,” Science Digest, Vol. 52, 1962, pp. 73 78.
[5] Forward, R. L., “Far Out Physics,” Analog Science Fiction/Science Fact, Vol. 95,
1975, pp. 147 166.
[6] Forward, R. L., “Flattening Spacetime Near the Earth,” Physical Review D, Vol. 26,
1982, pp. 735 744.
[7] Davis, E. W., “Advanced Propulsion Study,” Air Force Research Lab., Final Report
AFRL PR ED TR 2004 0024, Edwards AFB, CA, 2004, pp. 2 5.
[8] Millis, M. G., “Assessing Potential Propulsion Breakthroughs,” New Trends in Astro
dynamics and Applications, Belbruno, E. (ed.), Annals of the New York Academy of
Sciences, Vol. 1065, 2005, pp. 441 461.
[9] Forward, R. L., “Negative Matter Propulsion,” Journal of Propulsion and Power,
Vol. 6, No. 1, Jan. Feb. 1990, pp. 28 37.
[10] Forward, R. L., “Negative Matter Propulsion,” 24th AIAA/ASME/SAE/ASEE Joint
Propulsion Conference, AIAA Paper 1988 3168, Boston, MA, July 1988.
[11] Bondi, H., “Negative Mass in General Relativity,” Reviews of Modern Physics,
Vol. 29, No. 3, July 1957, pp. 423 428.
[12] Bonnor, W. B., and Swaminarayan, N. S., “An Exact Solution for Uniformly Accel
erated Particles in General Relativity,” Zeitschrift fur Physik A Hadrons and Nuclei,
Vol. 177, 1964, pp. 240 256.
[13] Bicak, J., Hoenselaers, C., and Schmidt, B. G., “Solutions of the Einstein Equations
for Uniformly Accelerated Particles Without Nodal Singularities,” Proceedings of
the Royal Society of London, Series A, Mathematical and Physical Sciences,
Vol. 390, 1983, “I. Freely Falling Particles in External Fields,” pp. 397 409; “II.
Self Accelerating Particles,” pp. 411 419.
224 E. W. DAVIS

[14] Terletskii, Y. P., Paradoxes in the Theory of Relativity, Plenum Press, New York,
1968, Chap. VI.
[15] Landis, G. A., “Comment on ‘Negative Matter Propulsion,’ ” Journal of Propulsion
and Power, Vol. 7, No. 2, 1991, p. 304.
[16] Clowe, D., Bradac, M., Gonzalez, A., Markevitch, M., Randall, S. W., Jones, C., and
Zaritsky, D., “A Direct Empirical Proof of the Existence of Dark Matter,” Astrophy
sical Journal Letters, Vol. 648, 2006, pp. L109 L113.
[17] Linder, E. V., “The Universe’s Skeleton Sketched,” Nature, Vol. 445, 2007, p. 273.
[18] Massey, R., et al., “Dark Matter Maps Reveal Cosmic Scaffolding,” Nature,
Vol. 445, 2007, pp. 286 290.
[19] Bilaniuk, O. M., and Sudarshan, E. C. G., “Particles Beyond the Light Barrier,”
Physics Today, Vol. 5, 1969, pp. 43 51.
[20] Heaviside, O., “A Gravitational and Electromagnetic Analogy,” The Electrician,
Vol. 31, 1893, pp. 281 282, 359.
[21] Thirring, H., “Über die Wirkung Rotierender Ferner Massen in der Einsteinschen
Gravitationstheorie,” Physikalische Zeitschrift, Vol. 19, 1918, pp. 33 39.
[22] Thirring, H., “Berichtigung zu Meiner Arbeit: ‘Über die Wirkung rotierender ferner
Massen in der Einsteinschen Gravitationstheorie,’ ” Physikalische Zeitschrift,
Vol. 22, 1921, pp. 29 30.
[23] Thirring, H., and Lense, J., “Über den Einfluss der Eigenrotation der Zentralkörper
auf die Bewegung der Planeten und Monde nach der Einsteinschen Gravitationsthe
orie,” Physikalische Zeitschrift, Vol. 19, 1918, pp. 156 163.
[24] Mashhoon, B., Hehl, F. W., and Theiss, D. S., “On the Gravitational Effects of Rotat
ing Masses: The Lense Thirring Papers Translated,” General Relativity and Gravita
tion, Vol. 16, 1984, pp. 711 750.
[25] Forward, R. L., “General Relativity for the Experimentalist,” Proceedings of the
Institute of Radio Engineers, Vol. 49, 1961, pp. 892 904.
[26] Forward, R. L., “Antigravity,” Proceedings of the Institute of Radio Engineers,
Vol. 49, 1961, p. 1442.
[27] Forward, R. L., “Guidelines to Antigravity,” American Journal of Physics, Vol. 31,
1963, pp. 166 170.
[28] Felber, F. S., “Exact Relativistic ‘Antigravity’ Propulsion,” Proceedings of the STAIF
2006: 3rd Symposium on New Frontiers and Future Concepts, El Genk, M. S. (ed.),
AIP Conference Proceedings Vol. 813, AIP Press, New York, 2006, pp. 1374 1381.
[29] Misner, C. W., Thorne, K. S., and Wheeler, J. A., Gravitation, W. H. Freeman & Co.,
New York, 1973.
[30] Ohanian, H., and Ruffini, R., Gravitation and Spacetime, 2nd ed., W. W. Norton &
Co., New York, 1994.
[31] Ciufolini, I., and Wheeler, J. A., Gravitation and Inertia, Princeton Univ. Press,
Princeton, NJ, 1995.
[32] Sachs, M., “The Mach Principle and the Origin of Inertia from General Relativity,”
Mach’s Principle and the Origin of Inertia, Sachs M., and Roy, A. R. (eds.), C. Roy
Keys, Inc., 2003.
[33] Peacock, J. A., Cosmological Physics, Cambridge Univ. Press, Cambridge, UK,
1999, pp. 25 26.
[34] Riess, A. G., et al., “Observational Evidence from Supernovae for an Accelerating
Universe and a Cosmological Constant,” Astronomical Journal, Vol. 116, 1998,
pp. 1009 1038.
 REVIEW OF GRAVITY CONTROL 225

[35] Perlmutter, S., et al., “Measurements of V and L from 42 High Redshift Superno
vae,” Astrophysical Journal, Vol. 517, 1999, pp. 565 586.
[36] Schwarzschild, B., “High Redshift Supernovae Indicate that Dark Energy Has Been
Around for 10 Billion Years,” Physics Today, Vol. 60, 2007, pp. 21 25.
[37] Riess, A. G., et al., “New Hubble Space Telescope Discoveries of Type Ia Superno
vae at z
1: Narrowing Constraints on the Early Behavior of Dark Energy,” Astro
physical Journal, Vol. 659, 2007, pp. 98 121.
[38] Astier, P., et al., “The Supernova Legacy Survey: Measurement of VM, VL and w
from the First Year Data Set,” Astronomy and Astrophysics, Vol. 447, 2006, pp.
31 48.
[39] Particle Data Group, “Review of Particle Physics,” Journal of Physics, Vol. 33, 2006,
pp. 210 232.
[40] White, H. G., and Davis, E. W., “The Alcubierre Warp Drive in Higher Dimensional
Spacetime,” Proceedings of the STAIF 2006: 3rd Symposium on New Frontiers and
Future Concepts, El Genk, M. S. (ed.), AIP Conference Proceedings, Vol. 813, AIP
Press, New York, 2006, pp. 1382 1389.
[41] Levi Civita, T., Rendiconti della Reale Accademia dei Lincei, Series 5, Vol. 26,
1917, p. 519.
[42] Pauli, W., Theory of Relativity, Dover, New York, 1981, pp. 171 172.
[43] Puthoff, H. E., Davis, E. W., and Maccone, C., “Levi Civita Effect in the Polarizable
Vacuum (PV) Representation of General Relativity,” General Relativity and Gravi
tation, Vol. 37, 2005, pp. 483 489.
[44] Bekenstein, J. D., “Gravitational Radiation Recoil and Runaway Black Holes,”
Astrophysical Journal, Vol. 183, 1973, pp. 657 664.
[45] Landau, L. D., and Lifshitz, E. M., The Classical Theory of Fields: Course of Theor
etical Physics Vol. 2, 4th rev. English ed., Butterworth and Heinemann, Oxford,
1998, pp. 347 350.
[46] Bonnor, W. B., and Piper, M. S., “The Gravitational Wave Rocket,” Classical and
Quantum Gravity, Vol. 14, 1997, pp. 2895 2904.
[47] Baker, R. M. L., Jr., “Preliminary Tests of Fundamental Concepts Associated with
Gravitational Wave Spacecraft Propulsion,” AIAA Space 2000 Conference & Expo
sition, AIAA Paper 2000 5250, Long Beach, CA, 2000.
[48] Baker, R. M. L., Jr., and Li, F. Y., “High Frequency Gravitational Wave (HFGW)
Generation by Means of X ray Lasers and Detection by Coupling Linearized GW
to EM Fields,” Proceedings of the STAIF 2005: 2nd Symposium on New Frontiers
and Future Concepts, El Genk, M. S. (ed.), AIP Conference Proceedings, Vol.
746, AIP Press, New York, 2005, pp. 1271 1281.
[49] Baker, R. M. L., Jr., Li, F. Y., and Li, R., “Ultra High Intensity Lasers for Gravita
tional Wave Generation and Detection,” Proceedings of the STAIF 2006: 3rd Sym
posium on New Frontiers and Future Concepts, El Genk, M. S. (ed.), AIP
Conference Proceedings, Vol. 813, AIP Press, New York, 2006, pp. 1352 1361.
[50] Weber, J., and Hinds, G., “Interaction of Photons and Gravitons,” Physical Review,
Vol. 128, 1962, pp. 2414 2421.
[51] Zombeck, M. V., Handbook of Space Astronomy & Astrophysics, 2nd ed., Cambridge
Univ. Press, Cambridge, UK, 1990, p. 82.
[52] Cox, A. N. (ed.), Allen’s Astrophysical Quantities, 4th ed., AIP Press, New York,
1999, p. 571.
226 E. W. DAVIS

[53] Douglass, D. H., and Braginsky, V. B., “Gravitational Radiation Experiments,” General
Relativity: An Einstein Centenary Survey, Hawking, S. W., and Israel, W. (eds.),
Cambridge Univ. Press, Cambridge, UK, 1979, p. 98.
[54] Kaplan, A. E., and Ding, Y. J., “Field gradient induced Second harmonic Generation
in Magnetized Vacuum,” Physical Review A, Vol. 62, 2000, p. 043805.
[55] Novaes, S. F., and Spehler, D., “Gravitational Laser Backscattering,” Physical
Review D, Vol. 47, 1993, pp. 2432 2434.
[56] Chen, P., “Resonant Photon Graviton Conversion in EM Fields: From Earth to
Heaven,” SLAC PUB 6666, Stanford Univ., Stanford, CA, 1994.
[57] Chen, P., and Noble, R. J., “Crystal Channel Collider: Ultra High Energy and
Luminosity in the Next Century,” Proceedings of the 7th Workshop on
Advanced Accelerator Concepts, Chattopadhyay, S., McCullough, J., and Dahl,
P. (eds.), AIP Conference Proceedings, Vol. 398, AIP Press, New York, 1997, pp.
273 285.
[58] De Logi, W. K., and Mickelson, A. R., “Electrogravitational Conversion Cross
Sections in Static Electromagnetic Fields,” Physical Review D, Vol. 16, 1977,
pp. 2915 2927.
[59] Schweber, S. S., “Defending Against Nuclear Weapons: A 1950s Proposal,” Physics
Today, Vol. 60, 2007, pp. 36 41.
[60] Gertsenshteı̆n, M. E., “Wave Resonance of Light and Gravitational Waves,” Soviet
Physics JETP, Vol. 14, 1962, pp. 84 85.
[61] Pustovoı̆t, V. I., and Gertsenshteı̆n, M. E., “Gravitational Radiation by a Relativistic
Particle,” Soviet Physics JETP, Vol. 15, 1962, pp. 116 123.
[62] Sushkov, O. P., and Khriplovich, I. B., “The Gravitational Radiation Emitted by an
Ultrarelativistic Charged Particle in an External Electromagnetic Field,” Soviet
Physics JETP, Vol. 39, 1974, p. 1.
[63] Grishchuk, L. P., and Sazhin, M. V., “Excitation and Detection of Standing Gravita
tional Waves,” Soviet Physics JETP, Vol. 41, 1976, pp. 787 793.
[64] Portilla, M., and Lapiedra, R., “Generation of High Frequency Gravitational Waves,”
Physical Review D, Vol. 63, 2001, p. 044014.
[65] Navarro, E. A., Portilla, M., and Valdes, J. L., “Test of the Generation of High Fre
quency Gravitational Waves by Irradiating a Dielectric Film in a Resonant Cavity,”
Proceedings of the STAIF 2004: 1st Symposium on New Frontiers and Future Con
cepts, El Genk, M. S. (ed.), AIP Conference Proceedings, Vol. 699, AIP Press,
New York, 2004, pp. 1122 1126.
[66] Feynman, R. P., and Hibbs, A. R., Quantum Mechanics and Path Integrals, McGraw
Hill, New York, 1965.
[67] Feynman, R. P., Morinigo, F. B., and Wagner, W. G., Feynman Lectures on Gravita
tion, Hatfield, B. (ed.), Addison Wesley, Reading, MA, 1995.
[68] Zee, A., Quantum Field Theory in a Nutshell, Princeton Univ. Press, Princeton, NJ,
2003, Chap. VIII.1.
[69] Calloni, E., Di Fiore, L., Esposito, G., Milano, L., and Rosa, L., “Gravitational
Effects on a Rigid Casimir Cavity,” International Journal of Modern Physics A,
Vol. 17, 2002, pp. 804 807.
[70] Calloni, E., DiFiore, L., Esposito, G., Milano, L., and Rosa, L., “Vacuum Fluctuation
Force on a Rigid Casimir Cavity in a Gravitational Field,” Physics Letters A,
Vol. 297, 2002, pp. 328 333.
 REVIEW OF GRAVITY CONTROL 227

[71] Bimonte, G., Calloni, E., Esposito, G., and Rosa, L. “Energy Momentum Tensor for
a Casimir Apparatus in a Weak Gravitational Field,” Physical Review D, Vol. 74,
2006, p. 085011.
[72] Bimonte, G., Calloni, E., Esposito, G., and Rosa, L. “Erratum: Energy Momentum
Tensor for a Casimir Apparatus in a Weak Gravitational Field,” Physical Review D,
Vol. 75, 2007, p. 089901(E).
[73] Pinto, F., “Progress in Quantum Vacuum Engineering Propulsion,” Journal of the
British Interplanetary Society, Vol. 59, 2006, pp. 247 256.
[74] Fermi, E., “Sull’elettrostatica di un Campo Gravitazionale Uniforme e sul Peso delle
Masse Elettromagnetiche,” Nuovo Cimento, Vol. 22, 1921, pp. 176 188.
[75] Puthoff, H. E., “Polarizable Vacuum (PV) Approach to General Relativity,” Foun
dations of Physics, Vol. 32, 2002, pp. 927 943.
[76] Heim, B., “Vorschlag eines Weges einer Einheitlichen Beschreibung der Elementar
teilchen,” Zeitschrift fur Naturforschung, Vol. A32, 1977, pp. 233 243.
[77] Heim, B., Elementarstrukturen der Materie: Einheitliche Strukturelle Quanten
feldtheorie der Materie und Gravitation, Band 1, 3rd ed., Resch Verlag, Innsbruck,
1998.
[78] Heim, B., Elementarstrukturen der Materie: Einheitliche Strukturelle Quanten
feldtheorie der Materie und Gravitation, Band 2, 2nd ed., Resch Verlag, Innsbruck,
1996.
[79] Heim, B., “Ein Bild vom Hintergrund der Welt,” Welt der Weltbilder, Imago Mundi,
Band 14, Resch, A. (ed.), Resch Verlag, Innsbruck, 1994.
[80] Heim, B., Dröscher, W., and Resch, A., Einfuhrung in Burkhard Heim Einheitliche
Beschreibung der Welt, Resch Verlag, Innsbruck, 1998.
[81] Dröscher, W., and Heim, B., Strukturen der Physikalischen Welt und ihrer Nichtma
teriellen Seite, Resch Verlag, Innsbruck, 1996.
[82] Willigmann, H., Grundriss der Heimschen Theorie, Resch Verlag, Innsbruck, 2002.
[83] Dröscher, W., and Häuser, J., “Physical Principles of Advanced Space Propulsion
Based on Heim’s Field Theory,” 38th AIAA/ASME/SAE/ASEE Joint Propulsion
Conference, AIAA Paper 2002 4094, Indianapolis, IN, July 2002.
[84] Dröscher, W., and Häuser, J., “Future Space Propulsion Based on Heim’s Field
Theory,” 39th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, AIAA
Paper 2003 4990, Huntsville, AL, July 2003.
[85] Dröscher, W., and Häuser, J., “Guidelines for a Space Propulsion Device Based on
Heim’s Quantum Theory,” 40th AIAA/ASME/SAE/ASEE Joint Propulsion Confer
ence, AIAA Paper 2004 3700, Ft. Lauderdale, FL, July 2004.
[86] Dröscher, W., and Häuser, J., “Heim Quantum Theory for Space Propulsion
Physics,” Proceedings of the STAIF 2005: 2nd Symposium on New Frontiers and
Future Concepts, El Genk, M. S. (ed.), AIP Conference Proceedings, Vol. 746,
AIP Press, New York, 2005, pp. 1430 1440.
[87] Dröscher, W., and Häuser, J., “Magnet Experiment to Measuring Space Propulsion
Heim Lorentz Force,” 41st AIAA/ASME/SAE/ASEE Joint Propulsion Conference,
AIAA Paper 2005 4321, Tucson, AZ, July 2005.
[88] Dröscher, W., and Häuser, J., “Spacetime Physics and Advanced Propulsion Con
cepts,” 42nd AIAA/ASME/SAE/ASEE Joint Propulsion Conference, AIAA Paper
2006 4608, Sacramento, CA, July 2006.
[89] Alzofon, F. E., “The Origin of the Gravitational Field,” Advances in the Astronauti
cal Sciences, Jacobs, H. (ed.), Vol. 5, Plenum Press, New York, 1960, pp. 309 319.
228 E. W. DAVIS

[90] Alzofon, F. E., “Anti Gravity with Present Technology: Implementation and Theor
etical Foundation,” 17th AIAA/SAE/ASME Joint Propulsion Conference, AIAA
Paper 81 1608, Colorado Springs, CO, July 1981.
[91] Alzofon, F. E., “The Unity of Nature and the Search for a Unified Field Theory,”
Physics Essays, Vol. 6, 1993, pp. 599 608.
 Chapter 5

Gravitational Experiments with Superconductors:

History and Lessons

George D. Hathaway
Hathaway Consulting Services, Toronto, Ontario, Canada

I. Introduction
HIS CHAPTER outlines some of the problems inherent in trying to discover
T and verify new forces, and it provides a historical survey of gravitational
experiments using superconductors. The development of superconductors,
and in particular high-temperature ceramic superconductors, has facilitated
the pursuit of a potential connection between gravity, electromagnetism, and
matter in the solid state. The discovery of yttrium barium copper oxide
(YBCO) ceramics that are superconducting at liquid nitrogen (LN2) temperatures
has allowed laboratories around the world to fabricate experimentally useful
superconductors cooled by relatively inexpensive LN2 rather than liquid
helium (LHe). However, experimenting at cryogenic temperatures necessitates
an increased awareness of a host of technical issues that can cause an expected
effect to be masked or otherwise interfered with. Hence, this chapter also deals
with several of the problems encountered by the experimentalist in the design
and execution of gravitational experiments with superconductors. For the theore-
tician, the possibility of superconductors representing a macroscopic quantum
object has provided several avenues for the development of theories connecting
gravity and gravitylike forces to matter. The exploration of these concepts is also
highlighted in this chapter.
The possible production of laboratory-scale gravitomagnetic fields using
high-temperature superconductors has recently been investigated by Li and
Torr [1 3] who expanded on earlier work by DeWitt [4] and Ross [5] to
include gravitomagnetic fields in the London equations for superconductors.

Copyright # 2008 by the American Institute of Aeronautics and Astronautics, Inc. All rights
reserved.

Director.

229
230 G. D. HATHAWAY

Podkletnov and Nieminen [6] published what is arguably the first possible
demonstration of an experimental link between high-temperature (LN2) super-
conductor effects and gravity, allegedly in the form of a “gravity shield.”
Although Podkletnov’s findings were later found not to be reproducible [7],
they suggested that the experimental search for gravity-related forces could be
undertaken in the laboratory. In addition, such work prompted theoreticians to
consider possible approaches for using superconductors as a special form of con-
densed matter capable of modifying and/or producing such forces.
Einstein maintained that gravity is a consequence of curved space-time
geometry, implying that we need to manipulate space itself to affect gravity.
General Relativity introduces a metric tensor theory of gravity, which allows
the prediction of gravitational interactions between bodies but does not explain
the fundamental physical basis of gravity. Similarly, Maxwell’s vector equations
allow us to predict the outcomes of electromagnetic interactions but do not
explain the fundamental basis for such interactions. It is possible to reformulate
the tensor format of General Relativity into a simpler vector format, valid for
weak field approximations and nonrelativistic velocities. Using perturbation
theory to compute the equations of motion in the simplified General Relativity
equations results in terms that have direct analogues in Maxwell’s equations,
where for example the flow of electric current is replaced by mass flow.
Forward [8,9] was one of the first to investigate this analogue. Of particular
interest is a term analogous to the Biot Savart magnetic field, which is generally
referred to as the gravitomagnetic field (also sometimes referred to as gravita-
tional frame dragging or the Lense Thirring effect). A second term is analogous
to the electrostatic Coulomb field, and is called the gravitoelectric field.
Essentially, the gravitomagnetic field produces a force between currents of
flowing matter while the gravitoelectric field produces a force between masses
themselves (the conservative Newtonian gravitational field). The term
gravitoelectromagnetic field is often used to refer to both the gravitoelectric
and gravitomagnetic fields.
In this analogue, gravity is thus comprised of a conservative velocity-
independent field and a velocity-dependent field analogous to the electric and
magnetic fields in electromagnetic theory. The simplified General Relativity/
Maxwell’s equations also lead to a Faraday-like law of induction, which can gen-
erate Newtonian gravitational fields from time-varying gravitomagnetic fields.
Modern experimental attempts to confirm the existence of the gravitomagnetic
field include highly accurate laser ranging of the Earth-moon distance, as well
as the gravity probe B satellite [10].
In addition to these fundamental considerations, a variety of other approaches
have appeared in the literature. Puthoff [11] extended an idea originally
articulated by Sakharov [12], which posits that gravity is related to quantum
fluctuations of the vacuum, sometimes called zero-point fluctuations. Alzofon
[13] presented an engineering approach to interacting with gravity by means of
altering nuclear entropy using dynamic nuclear orientation, that is, pulsed
polarization of the magnetic moments of nucleons by interaction with polarized
electron spins. Hughes analyzed the Kopernicky conjecture, which holds that
gravity is nothing other than the slight difference between forces of Coulomb
attraction and repulsion [14].
 GRAVITATIONAL EXPERIMENTS WITH SUPERCONDUCTORS 231

A distinction should be made between gravitational waves, gravitational

“force,” and anomalous forces. The existence of gravitational waves is a
prediction of General Relativity. Research on gravitational waves is generally
divided into two more-or-less distinct realms: low frequency (less than a few
hundred Hz) and high frequency. The search for low-frequency gravitational
waves currently utilizes large, heavy, long metal detectors, interferometers,
strain gauges, and accelerometers in an attempt to detect quadrupole gravitational
waves from cosmological sources, such as rapidly rotating binary star systems.
Several researchers postulate that high-frequency gravitational waves (HFGW)
could be produced under certain laboratory conditions in the near future [15].
It is envisioned that such HFGW could be used for communications, astronomy,
microscopy, and possibly even propulsion. The production and detection of these
waves may be mediated or influenced by superconductors. Some initial debate on
such considerations has been articulated by Woods [16,17]. If gravitational wave
interaction with laboratory-scale matter results in detectable forces, such forces
should be amenable to detection by available laboratory instruments.
Modifying gravity for propulsive purposes resolves itself into roughly two
categories: 1) neutralizing or negating the gravitational attraction of a nearby
body, and 2) providing a propulsive force or impulse to a spacecraft by manipu-
lating the underlying physical phenomenon which forms the basis of gravity. For
additional details, the reader is referred to Chapter 4.
In any experiment designed to produce a gravitylike force or to interact
directly with a local gravity field, the researcher typically must look for extremely
small deviations from a null result. Most observations to date demonstrate that
interactions between gravity and electromagnetic fields, given the densities and
strengths available to even the most well-equipped laboratory, are many orders
of magnitude smaller than contemporary equipment can detect. Braginski et al.
[18] determined that ordinary matter cannot be used to generate measurable
gravitational fields in the laboratory. Demonstrating that a new force has been
discovered in the laboratory will require the ability to distinguish between true
electromagnetic gravitational interactions and a host of other effects potentially
masquerading as these forces.

II. Experimental Traps and Pitfalls

In the course of trying to measure signals from potentially miniscule (if
existent) forces, constant vigilance for more prosaic explanations must be main-
tained. This is true when measuring not only the type of force (e.g., magnetic,
electrostatic, mechanical, gravitational), but also the origin of the force. In
even the most carefully designed experiments, it is easy to be fooled by a
litany of potential effects resulting from experimental procedures that may
have been poorly conceived or carried out. This section provides a brief overview
of some of those considerations.
For example, the direct determination of the mass of a test sample cooled to
cryogenic temperatures typically requires that the test sample be attached to a
measurement device (a balance, for example). If the balance is not cooled, the
connection between test sample and instrument must accommodate a very
great temperature difference, and problems may arise due to condensation and
232 G. D. HATHAWAY

buoyancy. In order to shield the test sample from the effects of condensation, the
sample can be enveloped either in a vacuum or a noncondensable (at the operat-
ing temperature) gas. Placing the sample in a vacuum mitigates both buoyancy
and condensation concerns, but may preclude effective cooling of the sample.
Hence a compromise must be struck between the need to isolate the sample
from the environment and the speed at which cooling is expected. Balances
that operate in vacuum are useful in this regard. It may be possible to weigh
the entire apparatus, cryogen, and all. However, this method introduces its own
host of problems, mostly resulting from boil-off of the cryogen.
Determination of the mass of a test sample that is near a superconductor
requires that the test sample be isolated from the environment as much as poss-
ible, to avoid being shaken, buoyed by air or gas currents, or otherwise perturbed.
However, if the sample needs to be cooled, the requirement for strict temperature
isolation can be somewhat relaxed.
Exacerbating the problems inherent in cryogenic experiments are those intro-
duced by nearby electric and magnetic fields. As the experimenter endeavors to
increase the mechanical isolation between the test sample and the local laboratory
environment, the sample’s susceptibility to electromagnetic and electrostatic
influences becomes greater.
The following summary of the some of the more important experimental
concerns is extracted from a compilation published in 2006 [19], which in
turn expanded on the joint paper by Reiss and Hathaway [20], which discussed
laboratory practices for experiments designed to detect gravitylike forces. The
following list is organized into general categories, and there will inevitably be
some overlap; it is by no means exhaustive.

A. Mechanical Effects
1. Thermal Effects
This includes thermally driven air, gas, or liquid currents, vapor condensation,
and thermal expansion issues when a difference in temperature is encountered.
Thermally driven currents act to artificially alter the weight of test samples
they encounter. It is often exceedingly difficult to calculate the resulting
forces, which tend to randomly fluctuate both temporally and spatially.
Condensates can alter the mass of the test article, with the most problematic
condensate in cryogenic experiments being water vapor from the air and, to a
lesser extent, nitrogen condensate when working at LHe temperatures. Exper-
iments must be designed to avoid any part of a cryogenically cooled apparatus
being in contact with condensable gasses. When dealing with severe temperature
differences, structural apparatus and other test articles must not be subjected to
thermal expansion or contraction, unless this effect can be accurately analyzed
and factored into the resulting measurements.

2. Buoyancy Effects
Materials display different buoyancy in different media. If a test sample is
weighed in air, then weighed in LN2, its apparent weight will clearly differ and be
shape dependent. More subtly, if the sample is cooled to cryogenic temperatures,
its buoyancy cross section changes as a function of its temperature, producing a
 GRAVITATIONAL EXPERIMENTS WITH SUPERCONDUCTORS 233

time-varying weight change that could be misinterpreted as a gravitational

interaction. Buoyancy differences may also arise in a single medium (such as air)
due to stratification and temperature differences in the medium.

3. Seismic/Vibration Effects
Scales, balances, and force sensors generally operate by detecting or
displaying the relative distances between fixed and movable masses, and it is
necessary to avoid introducing artificial mechanical motion/noise into the
measuring system. Vibrations can occur from laboratory and other sources,
such as compressors, fans, and local traffic (both pedestrian and vehicular), and
are usually transmitted into the measuring apparatus by structural elements. Miti-
gation is typically achieved either by loosening the structural design through the
installation of dampers, or tightening the design by anchoring the support struc-
tures to larger, relatively immobile masses. If these are not easily or completely
achievable, altering the frequency of the chief modes of vibration by adding or
subtracting masses is often helpful. Wires and cables carrying signals and
power may appear negligible as structural vibration carriers, but for experiments
at cryogenic temperatures the wires may freeze and become rigid. In addition,
the pulsing of current through any portion of a wire with curvature will
produce a flexure in the wire, leading to displacement or, for alternating currents,
oscillatory vibrations.

4. Vacuum Effects
All materials will to more or less extent outgas (release vapor) in vacuum,
hence the use of low outgassing materials for any experiment performed in a
vacuum or low ambient pressure environment is desirable. Even a low outgassing
material may have minute amounts of contaminant (oils, fingerprints, etc.) on its
surface, the release of which can be detected by sensitive balances. Other
contaminant material on the inside walls of the vacuum chamber housing the
sample may be released and coat or otherwise impinge upon the sample. Of
more subtle nature is the slight movement of the vacuum chamber and
encompassed test apparatus as the vacuum chamber is pumped down. Such
small movements can offset previously aligned balances, producing an apparent
weight change in a test sample simply due to a slight shifting in the support
structures. It is also worth noting the high degree of variation in outgassing
among materials that may be used in test apparatus or support structures involved
in such experiments. For instance, brass has a high outgassing rate due to the
presence of high vapor pressure zinc; if a brass test mass is used in a
vacuum environment, its actual weight loss could change over the course of
the experiment and mask other experimentally induced effects [21].

5. Liquid Effects
The evaporation or boiling of liquid (usually the cryogen) around a test sample
or the suspension system can clearly contribute noise into the measurement
system. This is particularly true when the test sample is suspended in, or is in
contact with, a liquid cryogen used to cool it below its transition temperature.
Boiling of the cryogen when inserting a sample will cause spurious readings
234 G. D. HATHAWAY

and potentially mask effects near the critical temperature until sample and
cryogen are in thermal equilibrium. Similarly, lifting the sample from a
cryogen produces high rates of evaporation (often in addition to dripping
cryogenic fluid), and will cause an apparent weight change as the temperature
of the test sample rises back to ambient thermal conditions.

B. Temporal Effects
Certain experiments require a test sample to be cooled in cryogenic gas or
liquid while being suspended from the movable arm of a balance. The precise
determination of sample temperature as a function of time is critical to ascribing
an observed weight change to a superconductive attribute. However, the attach-
ment of a thermocouple to the superconducting sample necessitates feeding wire
leads from the movable sample to the fixed laboratory frame on which the
temperature readout sits. Movable slip rings are too stiff and unpredictable to
be used as rotating/movable joints in most delicate experiments. Testing for
the attainment of critical transition temperature via Meissner flux expulsion is
often too cumbersome, and would of necessity disturb the balance. If direct
temperature determination proves exceedingly difficult, the best path is for
the experimenter to simply allow sufficient time for the test sample and any
supporting mass to reach the required equilibrium temperature.

C. Electromagnetic Effects
1. Magnetic Coupling
The incorporation of superconductors in gravity experiments necessarily
involves magnetic fields. Potential interactions with the magnetic fields arising
from current-carrying and signal wires must be taken into account in the vicinity
of the experimental apparatus. Residual magnetism can be found in certain
“nonmagnetic” metals and alloys, which may interact with the measurement
system. The assumption of perfect magnetic shielding by high-mu magnetic
shielding alloys is usually not valid, and care must be taken that the proper
level of electromagnetic shielding required by the experiment is obtained.
Time-varying electromagnetic fields generated by test equipment, power
sources, leads, and contacts can interact with nearby conducting bodies and
cause often subtle but nevertheless spurious measurements. More dramatically,
the sudden release of trapped magnetic fields in superconductors raised
above transition temperature can affect, and be affected by, nearby magnetic or
conductive structures.

2. Electric Coupling
As in the use of high-mu metals to shield magnetic fields, an over-reliance on a
Faraday cage for complete exclusion of DC or quasi-static electric fields is often
not realistic, and extreme care should be taken to account for and mitigate any
spurious electric fields that may interfere with the test specimen or measurement
apparatus.
 GRAVITATIONAL EXPERIMENTS WITH SUPERCONDUCTORS 235

3. Electromagnetic Coupling
Improper shielding or conductor pairs that are not tightly twisted can cause
unwanted cross talk on nearby signal cables, even shielded ones. Switching
transients in high-power circuits can easily couple into signal sensing circuits,
causing artificial or spurious readings; this is particularly important for circuits
incorporated into electronic balances. Using radio frequency (RF) power with
improperly matched loads or inappropriately sized matching networks can
cause intense electromagnetic interference in nearby circuits.

4. Grounding Issues
The physical layout of signal grounds should be carefully considered to avoid
nearby magnetic interference. Shield grounds should be checked for potential
ground loop problems by incorporating a single-point grounding connection.
Contacts should be checked for loose, corroded, or bi-metal connections which
can introduce unwanted contact potentials.

D. Electrostatic and Related Effects

1. Gradient Effects
The operation of an apparatus at high DC voltages relative to a nearby ground
will usually produce spatial gradients in the electric field, unless the experiment
is specifically designed to produce uniform fields in the vicinity of the test
apparatus. Such gradients may induce subtle motion in nearby free bodies
whether conductive or not, such as samples suspended from a balance.

2. Charge Pooling and Induced Charges

Even under high vacuum, without sufficient surface preparation and thermal
baking, the interior surfaces of vacuum chambers are covered with invisible
pools of surface charges, usually in the form of films of polar water molecules.
In very sensitive measurement systems, or where the sample mass being
weighed is in close proximity to a wall, the electrostatic forces between such
“patch charges” on the wall and test mass may interfere with proper weight
measurements. This applies whether or not the surfaces are conductive.

3. Ion and Molecular Effects

In high-voltage DC or AC experiments, even in vacuum, strong electric
fields can accelerate outgassed or residual ions in the vacuum chamber, which
in turn can impart momentum to a suspended test mass. This may occur either
because the ions are streaming off of or onto a sample held at high voltage, or
because the sample may be in the “ion wind” produced by a nearby source of
ions held at high voltage. Such ion wind effects are difficult to calculate, and
so generally need to be avoided by adhering to strict chamber cleanliness pro-
cedures, proper choice of vacuum compatible materials (including wire insula-
tion), and operation at chamber pressures below those needed to produce
significant ion wind effects. High voltages can cause ablation or sputtering of
even the best insulation.
236 G. D. HATHAWAY

4. Charge Leakage
Charges can leak from insulated conductors, either as a quasi-DC corona or
(more likely) in bursts called Trichel pulses in high-voltage experiments [22].
These may be negligible to the force-measuring apparatus at relatively low vol-
tages, but at some threshold value they can suddenly become problematic, and
often under generally unpredictable conditions. These can cause direct effects
by impinging on or interfering with balances, for instance, or indirect effects
through, for example, interference with nearby signal wires. Most commonly,
they cause weak conduction paths to nearby grounds, which can interfere with
high-gain amplifiers, and so on.

E. Experimental Expectations
The preceding is but a small sampling of the experimental artifacts that must be
considered when designing and performing any sensitive laboratory experiment,
particularly when searching for miniscule effects in support of controversial
theories. In addition, there are psychological pitfalls of which investigators who
design experiments are often unaware. For instance, confirmation bias is the ten-
dency to trust data that support one’s hypothesis, and to distrust (or ignore) data
that do not support it. After all, if my data are exactly as my theory predicts,
then the experiment is a good one and I made no errors! Another is errors in
logic. An example of one such incorrect syllogism might be “If A is true, then
we will observe B. We do observe B, therefore we have proven A is true.” This
false logic overlooks other explanations for observing B. Put more correctly the
logical statement is: “We do observe B, therefore one possible explanation is
that A is true, and there may be other explanations.” The all too common
neglect of prior probability flaunts Bayes’ theorem, which shows that if the
probability of a theory being true is low before the experiment, then a stronger
standard of proof is required to support its correctness. These and other con-
siderations must be recognized and added to the usual repertoire of experimental
design and protocol that makes up the bulk of the scientific method.

III. Historical Outline

It is instructive to follow the general historical development of the modern
search for a link between electromagnetism, matter, and gravity so that
future researchers can understand the scope of the investigation so far. This
outline focuses on those works related to superconductor experiments and
theory. For more general historical events and context, please refer to
Chapter 1. The reader is encouraged to pursue the references cited herein to
obtain a fuller appreciation of the effort that has been expended in this area
of physics.

A. Possibility of Gravitational Effects in the Laboratory

In a 1966 paper, DeWitt [4] proposed a magnetic analogue to the prediction by
Shiff and Barnhill [23] who showed that the electric field inside a regular
conductor is not zero in the presence of a Newtonian gravitational field.
DeWitt calculated the surface current resulting from nonzero magnetic and
 GRAVITATIONAL EXPERIMENTS WITH SUPERCONDUCTORS 237

Lense Thirring fields inside a superconductor in the presence of a nearby

gravitomagnetic/Lense Thirring field. The flux of the combination of magnetic
and gravitomagnetic fields was found to be quantized. He suggested that the
current could be measured in an apparatus that ensured that the current measure-
ment was far from the gravitomagnetic field. In 1983, Dewitt’s work was
expanded upon by Ross [5], who produced a modified set of London equations
as well as weak-field General Relativity equations in a form very similar to Max-
well’s equations. The latter were particularly useful to later researchers as they no
longer needed to deal with the complete set of tensor General Relativity
equations when dealing with practical calculations. These papers laid the theor-
etical foundations for the later work of Li, Torr, Tajmar, and others, to be dis-
cussed below.
Li and Torr combined and extended this prior work in a 1991 publication on
the effects of a superconductor on external gravitomagnetic and magnetic fields
[1]. Ordinarily, all magnetic fields are excluded from the interior of a type I
superconductor due to the Meissner effect. However, by solving the coupled
Maxwell, General Relativity, and London equations for the internal magnetic
and gravitomagnetic fields of superconductors exposed to external gravitomag-
netic and magnetic fields, Li and Torr predicted a small, residual, internal mag-
netic field in contrast to normal Meissner field expulsion. This in turn produces an
internal gravitomagnetic field. These internal fields are related to one another by
the Cooper pair mass-to-charge ratio. One year later, the same authors presented a
paper in which they used coupled Ginzburg Landau equations to calculate the
relative strengths of the electric and gravitational fields in superconductors in
the presence of magnetic and gravitomagnetic fields. They concluded that
under certain circumstances, a secondary gravitational field could be induced
inside the superconductor and “. . . provide a basis for the electrical generation
of gravitational fields in the laboratory” [2].
Around the same time Eugene Podkletnov, a Russian materials scientist on
staff at the Institute of Materials Science at the Tampere University of Technol-
ogy in Finland, published a paper with R. Nieminen on an apparent gravity
shielding experiment using spinning superconductors [6] (Fig. 1). At the heart
of his experimental design was the high-speed rotation of a relatively large
(14.5 cm diameter  6 mm thick) YBCO sintered ceramic superconducting
disk in the vapors of LHe. The disk was levitated by Meissner flux expulsion
over a single, large, toroidal support electromagnet, which was immersed in
LHe and powered by a variable-frequency supply between 50 Hz to 106 Hz. At
the periphery of the disk diameter, two additional but smaller electromagnets,
also powered by variable frequency supplies, were positioned. These two
“rotational” electromagnets were used to spin the disk in some unspecified
manner. A small nonconducting, nonmagnetic test mass was suspended from an
analytical balance about 15 mm from the top of the disk. The test mass weight
loss was observed to be about 0.3% when stable conditions were achieved.
Not surprisingly, these results were met with much skepticism. Apart from the
difficulty in explaining such a relatively large weight loss on purely theoretical
grounds, there was considerable doubt about the validity of the experimental
observations. The only information on the physical configuration of the
experiment was provided in the sketch reproduced in Fig. 1. Assuming this is
238 G. D. HATHAWAY

Plastic
Suspension
Thread
Electronic Balance

Cryostat
Two Magnets
with
Rotating
Field
Sample Plastic Film

Superconducting Disk

Toroidal
Support
Magnet

Liquid
Helium

Fig. 1 Podkletnov’s spinning disk experiment. (Adapted from Ref. 6.)

representative of the actual experimental setup, one notes that the sample, sample
suspension thread, and main balance arm are unprotected from air currents.
Concerning the cryostat, for instance, the only thing separating the vapors of
LHe from the laboratory atmosphere was a thin plastic film. Ordinarily, so
much water vapor and other gases would have condensed on the outer surface
of the film as to render it completely opaque, thus making the observation of
the disk extremely difficult. If the cryostat was actually designed roughly per
the sketch in Fig. 1, the LHe would be boiling so vigorously that it would
rupture any film unless adequate He gas escape was provided. As pointed out
by de Podesta and Bull [24], thermal currents and buoyancy changes above
such a cryostat would be so severe as to render the determination of the weight
of a test mass suspended only 15 mm above the disk (and, therefore, only a
few mm above a separating film covered with ice) virtually impossible. As
such, the design represented in the article appeared to be entirely unsuited for
the detailed and delicate measurements reported by Podkletnov. Other important
issues, such as how the disk was balanced, how it kept together at high speeds,
how much power was used to operate the coils, and what means were employed
 GRAVITATIONAL EXPERIMENTS WITH SUPERCONDUCTORS 239

to prevent the balance from being affected by the magnetic fields from the coils
were not addressed in the article.
Nevertheless, Podkletnov and Nieminen’s paper provoked some experimen-
talists to attempt replications of the essence of the experiment. The experimental
replications started out using less costly LN2 with either fixed or rotating
permanent magnets, and small (approximately 2 to 3 cm diameter) commercially
purchased superconducting disks. Several researchers, including Gonnelli at the
Politecnico of Turin (private communication), Woods et al. at the University of
Sheffield [25], and this author witnessed very slight apparent weight changes
while the disk was passing through its critical temperature, Tc. However, in
most cases, the apparent effect was generally no larger than the noise inherent
in the system, and further experimentation was not performed. At a subsequent
conference in Turin (discussed later in the Sec. III.B), Podkletnov emphasized
that unless the exact disk formulation was followed, high-frequency magnetic
fields employed (not permanent magnets), and a larger disk spun at lower temp-
eratures, the shielding effect would be extremely small.
After the Podkletnov and Nieminen publication, Li and Torr published another
article [3] expanding on their earlier investigations in an attempt to outline the
physical mechanism underlying the production of a gravitomagnetic field
inside a superconductor. Basic to their evaluation was an assumption of near-
zero magnetic permeability in type I superconductors, and the requirement of
coherent alignment of lattice ion spins in conjunction with a time-varying
applied magnetic vector potential field. With this framework, they determined
values for laboratory-scale induced internal gravitomagnetic fields and external
gravitoelectric fields, and predicted how these fields could be maximized.
After noting an additional assumption about Cooper pair to lattice ion mass
ratio, they state “. . . it appears that the electrically induced gravitoelectric field
should be readily detectable in the laboratory.”
In 1994, Kowitt [26] showed that the assumption of near-zero permeability
inside a superconductor used by Li and Torr in their 1992 and 1993 papers
was not valid. Another criticism of the Li and Torr results was from Harris
[27]. He argued that Li and Torr’s previous results were erroneous because
they assumed arbitrary (and extremely small) distances from the lattice ion to
the observer, thus producing unreasonably large effects. In fact, Harris pointed
out that the correct estimate of the induced gravitoelectric field outside a super-
conductor is some 20 orders of magnitude smaller than that predicted by Li and
Torr. These issues are further discussed in Sec. III.D.

B. Tests of Podkletnov Gravity Shielding Claims

In 1995 Li approached NASA Marshall Space Flight Center to support devel-
opment of her theory that an artificial gravity field could be measured outside a
superconductor. Funding was also secured for replicating the Podkletnov exper-
iment at NASA. After several preliminary experiments at NASA, Koczor et al.
failed to see the expected shielding effect in a Podkletnov-like experiment
[28]. However, they were using a small commercially available disk levitated
above permanent magnets at LN2 temperatures. The following year, after discus-
sions with Podkletnov, Noever and Koczor [29] published the results of their
240 G. D. HATHAWAY

investigations into nonrotating superconductor disks irradiated by radio

frequencies from 1 to 15 Mhz and detected a very weak gravity increase. This
finding was later shown by the same authors to likely be the result of an instru-
mentation artifact [30]. Notwithstanding, Koczor commissioned the fabrication
of a 27-cm bi-layer disk conforming to Podkletnov’s 1997 specifications
that had just been published [31], in the hopes of replicating their experiment.
Unfortunately, budgetary constraints forced NASA to stop further research in
this area few years later, with no further publications forthcoming.
In 1995 this author began preliminary experiments in Toronto after contacting
Podkletnov. The approach was to reproduce the original 1992 spinning disk
experiment with additional data from Podkletnov and to use a better cryogenic
design that allowed the possibility of mechanically spinning the disk. At our
Toronto laboratory, manufacturing also began on the large YBCO disks required
for the experiment.
Meanwhile, Podkletnov and Vuorinen attempted to publish the results of their
own updated spinning disk experiments in the British Journal of Physics D, but
the paper was withdrawn and later published on the Web [31]. This new
experiment involved a large AC-levitated, 27-cm diameter bi-layer sintered
YBCO disk spinning at 5000 rpm in two-phase high-frequency RF “rotation”
fields, and claimed to show gravitational shielding in the few percent range.
The cryostat design used in the updated experiments was somewhat better than
the original design in that there was considerably more shielding of the test
mass from buoyancy and thermal current effects.
In 1999, a private, unpublished symposium on Podkletnov’s work was hosted
by R. Gonnelli at the Politecnico in Turin. The state of the Toronto experimental
replication was presented, and Gonnelli and others presented their initial findings of
a possible tiny weight change in test samples suspended above a disk as it
passed through Tc. Podkletnov described how the “gravity shielding” effect was
initially discovered. Apparently his group had made large, sputtering target disks
of YBCO for single-crystal processing and, to ensure the correct uniformity and
porosity, the disks were set into rotation (presumably mechanically) while being
levitated over a “supporting solenoid.” This allowed quick and complete scanning
of the target’s surface by means of a small movable test magnet suspended above
the rotating disk and connected to an analytical chemical balance. When the
smoke from a technician’s pipe appeared to rise inexplicably above the apparatus,
they considered the possibility of gravitational shielding and substituted a
nonmagnetic, nonconducting test mass for the small suspended magnet. It should
be noted that the normal rotational speeds for magnetron sputtering targets are in
the 10s of rpm, whereas Podkletnov required speeds several orders of magnitude
greater than this. Subsequent information from Podkletnov indicated that, to
obtain the maximum stable test sample weight loss of about 0.3%, the optimum con-
ditions required operation of the three coils at frequencies of 105 Hz and disk
rotational speeds of several thousand rpm (E. Podkletnov, private communication).
Although the general consensus at the conference was that the Podkletnov
claims were not spurious, only the Toronto group elected to pursue a complete
replication.
Our own version of the Podkletnov spinning disk experiment was completed
in late 2001, with results published in 2003 [7] (Fig. 2). Here, an external motor
 GRAVITATIONAL EXPERIMENTS WITH SUPERCONDUCTORS 241

Top of Liquid
Helium
Dewar Insert

Motor to Spin Disk

Four Thermal Baffles

Supporting Stainless
Steel and Fiberglass
Driveshaft

Upper Bearing Plate

Three-Phase RF Coil Assembly

Threading Spinning
Superconductor Disk

Levitation Coils

Fig. 2 Apparatus for LHe test of Podkletnov’s “gravity shielding” claims [7].

spins the superconducting disk that is at the bottom of a deep LHe dewar. The
toroidal disk is threaded by three high-frequency coils and is levitated by three
solenoidal coils. The experiment showed a null result. Of potential interest to
the reader, the paper contains an overview of the experimental difficulties
arising from the nature of the experiment itself, and discusses the inability of
Podkletnov to supply critical data on the previous experiments. Unfortunately,
neither Podkletnov’s 1992 publication nor subsequent discussions with Podklet-
nov himself allowed a complete understanding of how the original experiment
was carried out.
Also of importance, in 1999 Reiss published [32] the results of precise
measurements of the weights of superconducting and non-superconducting
samples cooled below Tc. He found a slight (0.5%) weight increase of a
high-temperature superconductor for which he was not able to offer a prosaic
explanation. He weighed small disk-shaped samples held in a specially-made
capsule while dipping it into LN2. His analysis of possible artifacts is thorough
and very useful to other researchers investigating this area. In 2003, he published
242 G. D. HATHAWAY

[33] an update of his on-going LN2 experiments with increased precision and arti-
fact reduction. He was still observing weight changes during Tc transition to a
repeatable degree not achieved elsewhere.
However, during this same period, Tajmar et al. [34] attempted to replicate
several earlier weight-alteration experiments involving high-temperature
superconductors but with no success.

C. Podkletnov Force Beam Claims

At the Turin symposium, Podkletnov described further experiments with the
so-called high-voltage “gravity beam” (rather than gravity shielding) apparatus.
On a small (few cm2 in area) substrate was grown an array of single crystal
whiskers of YBCO. This plate was placed upright in a small LN2 dewar and
electrically attached to a small (200 KV) van de Graaff machine. A grounded
metal annular disk was placed a few centimeters laterally away. The whole
assembly was placed in a large bell jar, which was evacuated and backfilled
with argon to prevent YBCO degradation by water vapor. When the static
machine was operated, a light blue planar “discharge” was seen to pass from
the superconductor array to the annulus. At this instant, a pencil standing
upright on a table in an adjoining room fell over. The pencil was separated
from the experiment by a thick concrete wall.
In 2001, Podkletnov and his collaborator Modanese published a paper on the
Web concerning an enhanced version of the gravity beam experiment [35]. This
was the first general publication of the high-voltage impulse force experiment.
Enough technical description was available to allow assessment of the validity
of experimental setup. Unfortunately, many unresolved technical questions cast
considerable doubt on whether the experiment actually had been undertaken
and, once again, no confirmatory evidence was provided by Podkletnov or
Modanese. However, it presented a general summary of the theoretical work
by Modanese and had an extensive bibliography.

D. Gravitational Wave Transducers

In 2002, Chiao proposed using superconductors as gravitational wave transdu-
cers for RF radiation and attempted to demonstrate this experimentally, appar-
ently without success [36]. Harris provided an explanation for the apparent
failure by stating that neither gravitoelectric nor gravitomagnetic fields accom-
pany gravitational waves [37].
In 2005 Woods presented the results of his investigations into the concept of
using superconductors to mediate the detection of high-frequency gravitational
waves [16,17]. Using Li and Torr’s result that the phase velocity of gravitational
waves is smaller in superconductors than in ordinary matter by a factor of roughly
300, he reasoned that the refractive index for gravitational waves inside a
superconductor might be used to focus or manipulate such waves to increase
the possibility of their detection in the laboratory. However, his results are
highly dependent on the gravitational wave phase velocity being as predicted
by Li and Torr. He also examined the criticisms of Kowitt and Harris and
concluded that while Kowitt’s concerns may not be tenable, Harris’s
 GRAVITATIONAL EXPERIMENTS WITH SUPERCONDUCTORS 243

arguments were harder to refute and the issue might best be resolved through
experimental investigation.

E. Tajmar Experiments
In 2001 Tajmar and De Matos published a paper [38] which reviewed previous
work and showed that every electromagnetic field is coupled to a gravitoelectric
and gravitomagnetic field. According to Tajmar and De Matos, the coupling
“. . . is generally valid and does not require special properties like superconduc-
tivity.” The authors acknowledged the criticisms of Li and Torr by Kowitt and
Harris, and noted that the simple coupling coefficient they derive is exceedingly
small. However, they note that it can be increased by using massive ion currents
(e.g., a moving/rotating mass or a dense plasma) and alignment of electron and
nuclear spins. In a roughly concurrent publication [39], De Matos and Tajmar
extended these ideas and used a Barnett effect analogue to show “. . . any
substance (i.e., not necessarily superconductive) set into rotation becomes the
seat of a uniform intrinsic gravitomagnetic field . . .”
In the classic 1950 text on superfluids, London derived an expression for the
magnetic field produced by a rotating superconductor or superfluid, which was
proportional to the Cooper pair mass-to-charge ratio and the angular velocity
[40]. The value of this so-called “London moment” has been measured in the lab-
oratory by Tate et al. [41]. A general expression of the London moment can be
used to determine the Cooper pair mass. Surprisingly, the measured Cooper
pair mass, which had been predicted to be slightly smaller than twice that of
the electron, was actually slightly larger.
In a 2003 paper, Tajmar and De Matos [42] asked if a gravitational effect
might explain this mass discrepancy. By applying their previous work to this
“Cooper pair mass anomaly,” Tajmar and De Matos found that a relatively
huge internal gravitomagnetic field would be required to explain the mass
anomaly, of a magnitude that could be measured in the laboratory. The gravito-
magnetic field was predicted to have the form Bg 2vr /r, where r is the
Cooper pair mass density, r is the classical bulk mass density of the supercon-
ducting material, and v is the superconductor’s rotational speed. To test this
hypothesis, Tajmar proposed an experiment “. . . measuring the torque on a spin-
ning gyroscope produced by the gravitomagnetic field possibly generated by
rotating superconductors . . .” In 2006 Tajmar et al. [43] described the results
of just such an experiment (Fig. 3), which involved mechanical spinning of
niobium and high-temperature ceramic superconductor rings at LHe tempera-
tures. No external magnetic fields were applied; a sudden angular deceleration
of the superconducting rings was used to produce the acceleration required to
see the anticipated effect. Tajmar et al. claimed to have found the expected
large gravitomagnetic field as detected by nearby sensors such as accelerometers
and laser gyroscopes, which matched their theoretical predictions to within a
reasonable factor.
By 2007, Tajmar et al. [44] recognized that new data from improved exper-
iments did not match their prior predictions. Nevertheless, an unexplained
residual signal persisted that exhibited several unexpected features, including a
relatively large coupling constant of 10 8 between the observed acceleration
244 G. D. HATHAWAY

Support Rods Fixed to Building

Air Motor to Spin

Superconductor Ring

Shaft
Annular
Sensor Vacuum
Chamber

Position of
Sensors
(Typical)
Superconductor Ring

Cryostat
Liquid Helium

Fig. 3 Tajmar apparatus to search for possible frame-dragging effects. (Adapted

from Ref. 44.)

effect and the applied angular velocity. The effect appears to be proportional to
angular momentum and inversely proportional to temperature, after passing a
critical temperature (which is dependent on the material of the spinning ring
and not coincident with the superconducting critical temperature). In addition,
the effect is more pronounced in the clockwise rotation direction (as viewed
from above), and the effect does not decay as would a dipole field. While
Tajmar et al. endeavor to address all possible systematic errors or prosaic expla-
nations, they conclude: “The . . . measurements rule out our previous theoretical
model which predicted a coupling proportional to the material’s Cooper pair and
lattice mass density.” The residual signal observed in the most recent experiments
remains to be explained. At the time of this writing, this most recent reference
[44] was still undergoing review prior to its planned publication in the Time
and Matter Conference Proceedings, published by World Scientific Press.
In an experiment designed to test Tajmar’s results, a team from the
University of Canterbury in New Zealand [45] rotated a massive superconducting
lead ring near a very large ring laser gyroscope, which allowed measurement of
the Sagnac frequency difference between each rotational direction of the lead
mass. They measured a variation of Sagnac frequency deviation as a function
 GRAVITATIONAL EXPERIMENTS WITH SUPERCONDUCTORS 245

of superconductor rotational speed approximately 21 times less than that

predicted by Tajmar et al. Their data, however, do show a possible indication
of an effect similar to the Tajmar observations.
An explicit assumption in the work of Graham et al. [45] was that the putative
field was dipolar in nature. They noted that the magnitude of the effect claimed by
Tajmar is not dependent on the mass of the superconductor, but that the observed
effect would scale roughly as the volume of rotating material, assuming an
inverse-cube decay of a dipole field. Thus Graham et al. used a massive lead
ring, which was about three times the volume of that used by Tajmar.
However, the ring was positioned approximately 20 cm from one leg of the
gyroscope compared to the 5-cm distance used in the Tajmar experiment. In
addition, Tajmar used fiber optic gyroscopes as detectors, versus the New
Zealand’s team use of ring laser gyroscopes. Another difference was Tajmar’s
positioning of the gyroscopes slightly above the spinning ring, whereas the
spinning ring used by Graham et al. was in the plane of the gyroscope. Finally,
Graham et al. used angular velocities of approximately 100 rad/s, whereas the
Tajmar team used angular velocities around 400 rad/s.

F. Other Theoretical Directions

Theoretical work is still being produced on the interactions between the
gravitomagnetic field and accelerated matter. A recent paper by Ahmedov [46]
investigated the possibility of measuring a voltage due to the action of a gravito-
magnetic field on current flowing in orbiting semiconductors as an analogue of
the Hall effect resulting from a magnetic field. Rabounski and Borissova [47]
developed a nonholonomic metric in which a vertically accelerating mass (not
necessarily superconductive) perturbs the “background space” to produce a
zone of altered gravity near the mass. They invoke the vibrating balance
experiment of Kozyrev [48] to justify their claims.

IV. Summary and Future Directions

If real, a superconductor-mediated interaction between matter and gravity
would provide a method for manipulating gravitational fields, with potentially
tremendous benefits for future technology. Only a handful of researchers are
currently pursuing this avenue, and an increased understanding of the nature of
superconductivity is necessary to establish a firm basis from which to proceed.
One lingering question is whether there really is a Cooper pair mass anomaly.
A second area requiring further investigation is the precise determination
of the gravitational wave index of refraction inside a superconductor. Progress
in each of these areas will further benefit our understanding of superconductors
in general, and may result in new advances that can lead to future flight
systems. And lastly, further independent tests of the Tajmar observations
would help distinguish if a new effect is being observed, and if so, then better
characterize such effects.
In the spirit of the scientific enterprise, experimentalists must ensure that their
results, both positive and negative, are readily accessible for other researchers to
test. Theoreticians will then have more empirical evidence upon which to further
246 G. D. HATHAWAY

advance models of how Nature functions. And all must ensure that they are cognizant
of experimental pitfalls associated with the phenomena they seek to uncover.

References
[1] Li, N., and Torr, D. G., “Effects of a Gravitomagnetic Field on Pure
Superconductors,” Physics Review D, Vol. 43, No. 2, 1991, pp. 457 459.
[2] Li, N., and Torr, D. G., “The Gravitoelectrodynamics of Superconductors: A
Theoretical Basis for a Principle of Electrically Induced Gravitation,” Bulletin of
the American Physics Society, Vol. 37, No. 2, 1992, Session G8 paper 10.
[3] Li, N., and Torr, D. G., “Gravitoelectric Electric Coupling via Superconductivity,”
Foundations of Physics Letters, Vol. 6, No. 4, 1993, pp. 371 383.
[4] DeWitt, B. S., “Superconductors and Gravitational Drag,” Physics Review Letters,
Vol. 16, No. 24, 1966, pp. 1092 1093.
[5] Ross, D. K., “The London Equation for Superconductors in a Gravitational Field,”
Journal of Physics A, Vol. 16, 1983, pp. 1331 1335.
[6] Podkletnov, E., and Nieminen, R., “A Possibility of Gravitational Force Shielding by
Bulk YBa2Cu3O7-x Superconductor,” Physica C, Vol. 203, 1992, pp. 441 444.
[7] Hathaway, G., Cleveland, B., and Bao, Y., “Gravity Modification Experiment Using
a Rotating Superconducting Disk and Radio Frequency Fields,” Physica C, Vol. 385,
2003, pp. 488 500.
[8] Forward, R. L., “General Relativity for the Experimentalist,” Proceedings of the
IRE, Vol. 49, 1961, pp. 892 904.
[9] Forward, R. L., “Guidelines to Antigravity,” American Journal of Physics, Vol. 31,
1963, pp. 166 170.
[10] Fairbank, J. D., Deaver, Jr., B. S., Everitt, C. W. F., and Michelson, P. F. (eds.), Near
Zero: New Frontiers of Physics, W. H. Freeman and Co., New York, 1988, Chap. VI.
[11] Puthoff, H. E., “Gravity as a Zero Point Fluctuation Force,” Physics Review A, Vol.
39, 1989, pp. 2333 2342.
[12] Sakharov, A. D., “Vacuum Quantum Fluctuations in Curved Space and the Theory of
Gravitation,” Soviet Physics Doklady, Vol. 12, 1968, pp. 1040 1041.
[13] Alzofon F. E., “Anti Gravity with Present Technology: Implementation and
Theoretical Foundation,” 17th AIAA/SAE/ASME Joint Propulsion Conference,
AIAA Paper 81 1608, Colorado Springs, CO, July 1981.
[14] Hughes, W. L., “On Kopernicky’s Conjecture: Gravity Is a Difference Between
Electrostatic Attraction and Repulsion,” Galilean Electrodynamics, Vol. 15, No. 5,
2004, pp. 97 100.
[15] Gravitational Wave Conference: International High Frequency Gravitational Waves
(HFGW) Working Group, The Mitre Corporation, McLean, VA, May 2003.
[16] Woods, R. C., “Gravitational Waves and Superconductivity,” Space Technologies
and Applications International Forum 2005, El Genk, M. S. (ed.), American Institute
of Physics Conference Proceedings, Melville, NY, 2005.
[17] Woods, R. C., “A Novel Variable Focus Lens for HFGW,” Space Technologies and
Applications International Forum 2006, El Genk, M. S. (ed.), American Institute of
Physics Conference Proceedings, Melville, NY, 2006.
[18] Braginski, V. B., Caves, C. M., and Thorne, K. S., “Laboratory Experiments to Test
Relativity Gravity,” Physics Review D, Vol. 15, No. 5, 1977, pp. 2047 2068.
 GRAVITATIONAL EXPERIMENTS WITH SUPERCONDUCTORS 247

[19] Earthtech International. URL: http://earthtech.org/experiments/Hathaway Night

mare List.pdf [Accessed 3 March 2008].
[20] Reiss, H. D., and Hathaway, G. D., “Minimum Experimental Standards in the
Laboratory Search for Gravity Effects,” Space Technologies and Applications
International Forum 2006, El Genk, M. S. (ed.), American Institute of Physics Con
ference Proceedings, Melville, NY, 2006.
[21] O’Hanlon, J. F., A User’s Guide to Vacuum Technology, 2nd ed., Wiley, New York,
1989, p. 445.
[22] Gallagher, T. J., and Pearmain, A. J., High Voltage Measurement, Testing and
Design, Wiley, New York, 1983, p. 50.
[23] Schiff, L. I., and Barnhill, M. V., “Gravitation Induced Electric Field near a Metal,”
Physics Review, Vol. 151, No. 4, 1966, p. 1067.
[24] de Podesta, M., and Bull, M., “Alternative Explanation of ‘Gravitational Screening’
Experiments,” Physica C, Vol. 253, 1995, pp. 199 200.
[25] Woods, R. C., Cooke, S. G., Helme, J., and Caldwell, C. H., “Gravity Modification by
High Temperature Superconductors,” 37th AIAA/ASME/SAE/ASEE Joint Propul
sion Conference, AIAA Paper 2001 3363, Salt Lake City, UT, July 2001.
[26] Kowitt, M., “Gravitomagnetism and Magnetic Permeability in Superconductors,”
Physics Review B, Vol. 49, 1994, pp. 704 708.
[27] Harris, E. G., “Comments on ‘Gravitoelectric electric Coupling via Superconduc
tivity,’” Foundations of Physics Letters, Vol. 12, 1999, p. 201.
[28] Li, N., Noever, D., Robertson, T., Koczor, R., and Brantley, W., “Static Test for a
Gravitational Force Coupled to Type II YBCO Superconductors,” Physica C, Vol.
281, 1997, pp. 260 267.
[29] Noever, D., and Koczor, R., “Radio Frequency Illuminated Superconductive Disks:
Reverse Josephson Effects and Implications for Precise Measuring of Proposed
Gravity Effects,” NASA JPL 9th Advanced Space Propulsion Research Workshop
& Conference, Pasadena, CA, 1998.
[30] Noever, D., Koczor, R., Roberson, R., “Superconductor Mediated Modification of
Gravity? AC Motor Experiments with Bulk YBCO Disks in Rotating Magnetic
Fields,” 34th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, AIAA Paper
98 3139, Cleveland, OH, July 1998.
[31] Podkletnov, E., “Weak Gravitational Shielding Properties of Composite Bulk YBCO
Superconductor below 70K under EM Field,” LANL cond mat/9701074 v. 3, 16
Sept. 1997.
[32] Reiss, H. D., “A Possible Interaction Between Gravity and High Temperature Super
conductivity by a Materials Property?,” 15th European Conference on Thermophy
sical Properties, Wuerzburg, Germany, Sept. 1999.
[33] Reiss, H. D., “Weight Anomalies Observed During Cool Down of High Temperature
Superconductors,” Physics Essays, Vol. 16, 2003, pp. 236 253.
[34] Tajmar, M., Hense, K., Marhold, K., and De Matos, C., “Weight Measurements of
High Temperature Superconductors During Phase Transition in Stationary, Nonsta
tionary Condition and Under ELF Radiation,” Space Technologies and Applications
International Forum 2005, El Genk, M. S. (ed.), American Institute of Physics Con
ference Proceedings, Melville, NY, 2005.
[35] Podkletnov, E., and Modanese, G., “Impulse Gravity Generator Based on Charged
YBa2Cu3O7 y Superconductor with Composite Crystal Structure,” arXiv:
physics/0108005v2, Aug. 2001.
248 G. D. HATHAWAY

[36] Chiao, R., “Superconductors as Quantum Transducers and Antennas for Gravita
tional and Electromagnetic Radiation,” arXiv: gr qc/0204012, Apr. 2002.
[37] Harris, E., “Superconductors as Gravitational Wave Detectors,” 69th Annual
Meeting Southeastern Section, American Physics Society, No. NC.001, 2002.
[38] Tajmar, M., and De Matos, C., “Coupling of Electromagnetism and Gravitation in the
Weak Field Approximation,” Journal of Theoretics, Vol. 3, No. 1, 2001.
[39] De Matos, C., and Tajmar, M., “Gravitomagnetic Barnett Effect,” Indian Journal of
Physics, Vol. 75B, No. 5, 2001, pp. 459 461.
[40] London, F., Superfluids, Vol. 1, Wiley, New York, 1950, p. 78.
[41] Tate, J., Cabrera, B., Felch, S., and Anderson, J., “Precise Determination of the
Cooper Pair Mass,” Physics Review of Letters, Vol. 62, No. 8, 1989 pp. 845 848.
[42] Tajmar, M., and De Matos, C., “Gravitomagnetic Field of a Rotating Superconductor
and of a Rotating Superfluid,” Physica C, Vol. 385, 2003, pp. 551 554.
[43] Tajmar, M., Plesescu, F., Marhold, K., and De Matos, C., “Experimental Detection of
the Gravitomagnetic London Moment,” arXiv: gr qc/0603033, Mar. 2006.
[44] Tajmar, M., Plesescu, F., Seifert, B., Schnitzer, R., and Vasiljevich, I., “Search for
Frame Dragging Like Signals Close to Spinning Superconductors,” Proceedings of
the Time and Matter 2007 Conference, Bled, Slovenia, World Scientific Press, 2007.
[45] Graham, R. D., Hurst, R. B., Thirkettle, R. J., Rowe, C. H., and Butler, P. H.,
“Experiment to Detect Frame Dragging in a Lead Superconductor,” Physica C, 6
July 2007, submitted for publication.
[46] Ahmedov, B. J, “General Relativistic Galvano Gravitomagnetic Effect in Current
Carrying Conductors,” arXiv: gr qc/0701045v1, Jan. 2007.
[47] Rabounski, D., and Borissova, L., “A Theory of the Podkletnov Effect Based on
General Relativity: Anti Gravity Force Due to the Purturbed Nonholonomic Back
ground of Space,” Progress in Physics, Vol. 3, 2007, pp. 57 80.
[48] Kozyrev, N. A., “The Vibration Balance and Analysis of Its Operation,” Problems
of Research of the Universe, Vol. 7 in “Astrometry and Celestial Mechanics,”
Moscow Leningrad, 1978, pp. 582 584.
 Chapter 6

Nonviable Mechanical “Antigravity” Devices

Marc G. Millis
NASA Glenn Research Center, Cleveland, Ohio

I. Introduction
QUINTESSENTIAL example of nonviable approaches to create antigravity
A are the commonly proposed devices that attempt to convert mechanical
oscillations or gyroscopic motion into thrusting or gravity-defying effects.
When claimed as a breakthrough, such ideas are often summarily dismissed as
they are known to violate conservation of momentum. To offer a more effective
response, explanations are offered on why these devices appear to be break-
throughs, how they operate from the perspective of established physics, and
what tests would be required to provide more convincing evidence of how the
devices really operate. This also provides a starting point for critically assessing
future proposals of this type.

II. Oscillation Thrusters

The oscillation thruster, also describable as a sticktion drive, internal drive, or
slip-stick drive, is a commonly suggested device that uses the motion of internal
masses to create net thrust. One of the most famous oscillation thrusters is
the 1959 “Dean drive” described in Patent 2,886,976 [1]. A more recent and
simple example is shown in Fig. 1 [2] and further still, Fig. 2 displays an
example that uses rotating masses [3]. Although there are many versions, all
oscillation thrusters have the following common components: chassis to
support a system of masses; conveyor that moves the masses through an
asymmetric cycle; and power source for the conveyor.
A crucial feature is that the internal masses go through a cyclic motion where
the motion in one direction is quicker than in the other. The result is that the

This material is a work of the U.S. Government and is not subject to copyright protection in the
United States.

Propulsion Physicist, Propellant System Branch, Research and Technology Directorate.

249
250 M. G. MILLIS

Fig. 1 Linear oscillation thruster. A typical example of an oscillation thruster,

specifically from Patent 5,685,196 [2]. As the cam (60) rotates, a mass (50) moves
slowly in one direction and is allowed to return quickly in the other. The reaction
force from one part of this cycle is sufficient to overcome static friction, while the
reaction force is insufficient in the other part of the cycle. This leads to one-
directional motion, giving the illusion of net thrust.

whole device moves in surges across the ground, giving the appearance that a net
thrust is being produced without the expulsion of a reaction mass.
Because it would constitute a breakthrough to be able to move a vehicle
without expelling a reaction mass [4], these devices appear to be breakthroughs.
Regrettably, such devices are not breakthroughs because they still require a
connection to the ground to create net motion. The ground is the reaction mass,
and the frictional connection to the ground is a necessary feature to its operation.
More specifically, it is the difference between the static friction (sometimes
called sticktion) and the dynamic friction between the device and the ground

Fig. 2 Rotational oscillation thruster. A typical example of an oscillation thruster

that uses rotating masses with convolutions of cycles. The version shown is from
Patent 4,631,971 [3].
 NONVIABLE MECHANICAL “ANTIGRAVITY” DEVICES 251

that is required for their operation. Static friction, the amount of friction
encountered when contacting surfaces are not moving relative to one another,
is typically greater than the dynamic friction when the contacting surfaces are
moving relative to one another.
Recall that the device’s internal masses move faster in one direction and
slower in the other. When the masses move quickly, the device has enough reac-
tion force to overcome the static friction between itself and the ground, and the
device slides. When the internal masses return slowly in the other direction, the
reaction forces are not enough to overcome the static friction and the device stays
in its place. The net effect is that such slip-stick motion causes the device to scoot
across the floor.
If the device could be placed into orbit to follow a freefall trajectory absent of
any connection with external masses, then the center of mass of the entire system
would follow the freefall trajectory without deviation, while the device’s external
frame and its internal masses would just oscillate with respect to one another.
Similarly, if the device were dropped with its thrusting direction pointing
either up or down, the rate of fall of the center of mass of the system would be
identical regardless if the device were on or off.
To illustrate stick-slip operation, Fig. 3 offers an analogy. The right side of the
figure represents half of a cycle where the device does not move, while the left
side represents half of the cycle where the device does move. Even though the
total impulse (I F  t), as represented by the area of the rectangle, is equal
in both phases of the cycle, the force, F, represented by the width of the rectangle,
and its duration, t, represented by height of the rectangle, can vary. When they do,
they vary reciprocally so that their product remains constant. (Note that this is

Left-hand Half-Cycle Right-hand Half-Cycle

FL × tL = I = FR × tR
Area = Impulse, I

Duration
Duration Right, tR
Left, tL

Reaction Force Left, FL Reaction Force Right, FR

Critical Static Critical Static

Friction Force, Fc Friction Force, Fc

|FL| > |Fc| |FR| < |Fc|

Motion No Motion

Fig. 3 Equal impulses, different forces. This is a conceptual representation of the

F 3 t impulse of each half of a generic mechanical oscillator. Although the impulse
(area of rectangle) is equal in both halves of the cycle, the force on the right part
of the cycle, FR, is insufficient to overcome the critical friction, Fc, while the force
on the left, FL, is sufficient. The device would move during the left half-cycle and
remain stuck to the floor on the right half-cycle.
252 M. G. MILLIS

only a conceptual illustration; a more rigorous analysis would require integrating

a variable force over time.) With the impulse the same in both half-cycles, the
reaction force is less on the right-hand side than on the left. When compared
to the critical static friction force, Fc, which is the amount of force that must
be overcome to set the device in motion, it is clear that the reaction force for
the right half-cycle is less than this critical force, while on the left it is greater.
This means that the device would move during the left-hand half of the cycle,
but not during the right-hand half.
More rigorous analyses of a given device can take on a variety of forms,
depending Ð on the device in question. Options Ðinclude using the impulse represen-
tation ( F . dt) or the work representation ( F . dl) over all the phases of the
device’s cycle. Also, depending on the device, the number of phases could
vary. In Fig. 3, there are only two phases, but for the Foster device in Fig. 1,
for example, at least three phases would need to be assessed: 1) when the cam
(part 60) is displacing the mass (part 50); 2) when the mass returns under the
restoring force of spring (part 80); and 3) when the mass comes to a stop back
to its initial position.
Because such analyses are time-consuming and are not likely to be understood
by many amateurs, it is more effective to suggest rigorous experimental proof. A
fitting test is to place the device on a level pendulum stand, as illustrated in Fig. 4,
and compare the deflection between the on and off conditions of the device. A
sustained net deflection of the pendulum is indicative of genuine thrust. Conver-
sely, if the pendulum oscillates around its null position, which is the expected
finding, then the device is not creating net thrust.
To avoid possible spurious effects, it is advised to have the device and
its power supply on the pendulum, and to have the tallest pendulum possible
(i.e., l in Fig. 4). The reason for containing the power supply with the device is
to avoid having the power cords interfere with the free motion of the pendulum.

Center of Mass
Oscillation

q
l Thrusting
Direction

F
h

Fig. 4 Basics of a level pendulum test. A pendulum is a simple tool for measuring
lateral force and a level pendulum keeps the test platform from tilting during
operation. If a thrusting device can sustain a deflection of a level pendulum, then
there is strong evidence toward the claim of net thrust. It is anticipated, however,
that mechanical devices will instead oscillate the pendulum back and forth, with
the average position being zero deflection.
 NONVIABLE MECHANICAL “ANTIGRAVITY” DEVICES 253

The reasons for having a tall pendulum is to make its lateral motion (d in Fig. 4)
more pronounced for a given lateral force, F, and to reduce its natural oscillation
frequency to be much less than the oscillation frequency of the device. Spurious
results are possible if the oscillation frequency of the device and the pendulum
are similar. For reference, Eqs. (1) and (2) present common equations for
using pendulums to measure lateral thrusting (valid for small deflections,
approximating sine u  u ) [5].
Lateral force as a function of deflection:

F ¼ mg tan u (1)

Natural frequency of a pendulum:

r
1 g
f ¼ (2)
2p l
where:
F lateral force acting to deflect the pendulum (N)
f pendulum frequency (Hz)
g gravitational field (9.8 m/s2)
l length of the pendulum (m) (Fig. 4)
m mass at the lower end of the pendulum (other masses taken as
negligible) (kg)
u deflection angle (radians) (Fig. 4)
A level pendulum is recommended instead of a simple pendulum to avoid the
misleading effects from tilting the base, as illustrated in Fig. 5. Similarly,
the reason that an air track is not recommended is the misleading effects from

Fig. 5 Pitfall of simple pendulum. With a simple pendulum, a shift in the center of
mass of the device (from the motion of its internal masses), gives the false appearance
of a deflection when in fact the center of mass of the system is still directly underneath
the support.
254 M. G. MILLIS

Fig. 6 Pitfalls of air track test. In addition to the fact that air tracks posses some
friction, they are not suitable as a testing platform due to the potential for induced
tilting. If the center of mass of the device shifts forward, for example, the leading
edge of the air platform can tilt downward, creating a preferred channel for the
airflow, which in turn creates a propulsive force on the platform.

the tilting of its base, as illustrated in Fig. 6. When the internal masses of the
device shift off center, the base can tilt. In the case of the pendulum, this has
the effect of inducing an apparent deflection of the pendulum. In the case of
an air track this has the effect of deflecting more air away from the dipped
end, thereby thrusting the device from the reaction to the asymmetric airflow.
Another difficulty when using air tracks is that it is easy for amateur research-
ers to mistake steady-state motion as evidence of thrust. When activating the
device, or when setting an active device on an air track, it is easy to inadvertently
impart an initial velocity that will continue to move the device along the track.
This motion is not evidence of thrust, but rather of that initial velocity. If the
device is actually producing thrust (and the tilting issue described previously is
avoided), then the device will accelerate, moving faster and faster. Making the
distinction between steady-state motion (zero force) and a slightly accelerated
motion (thrusting) can be difficult with the naked eye. Instead, reliable measures
of velocity versus time are required.
To keep an open yet rigorous mind to the possibility that some physical
phenomenon has been overlooked, it should be necessary that future proposals
on these types of devices pass a pendulum test and that rigorous supporting
data addressing all possible false-positive conclusions be provided. Although a
“jerk” effect (time rate change of acceleration) [6] has sometimes been men-
tioned as a theoretical approach to understand such devices, no physical evidence
has been reported to substantiate that such a jerk effect exists, and theoretical dif-
ficulties arise because of momentum conservation violations. If successful net-
thrust tests are ever produced and if a genuine new effect is found, then
science will have to be revised because it would then appear as if such devices
were violating conservation of momentum.

III. Gyroscopic Antigravity

Another category of commonly purported mechanical breakthroughs consists
of a system of gyroscopes. A famous example is from the 1973 demonstration by
Eric Laithwaite, where a spinning gyro is shown to rise upward while it is forced
 NONVIABLE MECHANICAL “ANTIGRAVITY” DEVICES 255

Fig. 7 Laithwaite “propulsion system” patent. Quoting from US Patent 5,860,317:

“. . . the mass of the first gyroscope moves with an associated second movement of
the mass of the remainder of the system in substantially the opposite direction,
wherein the movement owing to the translation-dominated portion and is larger
than the movement owing to the precession-dominated portion of the motion,
hence moving the system” [8].

to presses.† Although such upward motion is a consequence of conservation of

angular momenta, it is easily misinterpreted as an “antigravity” effect.
Laithwaite, a professor of applied electricity at the Royal Institution of Great
Britain from 1967 to 1975 [7], and Dawson went on to patent a device (Fig. 7)
that claims to produce linear force from such torques [8].
Variations on the claim that gyroscopes can create linear thrust are common and
typically consist of forcing the axis of a spinning gyroscope to change its orien-
tation in a manner that causes the entire gyroscope to shift upward, thereby creating
the impression that an upward “antigravity” force exists [9]. Because a rigorous
analysis of this dramatic motion can be difficult, the ambiguities regarding its
real physics linger.
Such concepts can be viewed as trying to reverse the effect of a spinning top.
Rather than falling immediately over, a spinning top precesses. It appears as if an
invisible force is holding it up (Fig. 8). Although the gravitational field, g, which
is tilting the top, is rectilinear, the manner in which the top reacts to it produces
angular torques. These torques are working to conserve angular momentum in
response to gravity. The notion of reversing this process, of forcing the gyro-
scopic precessions to induce a linear force to oppose gravity (“antigravity”),
seems like reasonable symmetry to expect, but this is not the case.
To illustrate the basic operation of a typical gyroscopic antigravity device, a
simple version from Ref. 9 that is based on the Laithwaite demonstration
(Fig. 9) will be used at the working example. Laithwaite’s demonstration con-
sisted of just a singe 50-lb gyroscope rather than the dual version shown, but

†
Eigenbrot, I. V., email exchange between Dr. Ilya V. Eigenbrot, International and Science
Communication Officer, Imperial College London, and Marc Millis, NASA, Glenn Research
Center, Cleveland, OH, 15 Sept. 2004.
256 M. G. MILLIS

Before After
(1) The projection

Top view
of angular momentum
of top is reduced
relative to vertical
axis when tilted (2) The top begins
to precess around
vertical axis to
make up the loss
Side view

Fig. 8 Dynamics of a spinning top. As gravity, g, pulls downward on a spinning top,

the tilting changes the proportion of the top’s angular momentum that is projected
onto the vertical axis. To conserve angular momentum along this axis, the entire
top will begin to precess around. Note that this figure is intended only as a device
to help understand the direction of such precessions rather than as a rigorous
analysis of the system’s dynamics.

the principle of operation is the same. With the gyroscope up to speed, Laithwaite
(himself acting as the “main spindle” and “pivot” analogous to Fig. 9) was able to
lift the gyroscope by its stem (nonrotating beam that is coincident with the gyro-
scope’s axis), by torquing the stem around horizontally. In Fig. 9, a dual version
of the same thing occurs with the stems of two gyroscopes mounted to a “main
spindle.” When this spindle is rotated, the gyroscope’s stems will pivot upward
giving the impression of a lifting force. A mechanical stop is included to halt
the gyroscope’s tilting while it is still aligned to produce this “upward” force.
To understand this in terms of known physics, Fig. 10 presents a visual means
to help understand how conservation of angular momenta can lead to such effects.
This diagram should be viewed only as a device for understanding the direction
of torques from such devices, rather than as a rigorous means to fully convey the
system’s dynamics. Figure 10 shows the device before and after the main spindle
has been torqued. A helpful simplification to comprehend how conservation of
angular momentum functions is to consider only one axis at a time, so this illus-
tration concerns itself only with the angular momentum in the axis of the main
spindle. To set the initial angular momentum to zero, both the angular momentum
of the main spindle and the gyroscope will be set to zero in this view (left or
“before” side of Fig. 10). To achieve this zero-momentum initial condition, the
position of the gyroscope is set perpendicular to the main spindle so that it
projects none of its angular momentum along the axis of the main spindle.
Said another way, when viewed from the top (upper left in Fig. 10), there are
no apparent rotations. Along the view of this axis, there is zero angular
momentum.
In the “after” or right side or Fig. 10, the main spindle is rotating clockwise.
At this point it is not necessary to consider what caused this change, as this
 NONVIABLE MECHANICAL “ANTIGRAVITY” DEVICES 257

Fig. 9 Classical gyroscopic antigravity claim “Essential Design of the Laithwaite

Engine.” A typical version of a gyroscopic device that purports to exhibit an
antigravity effect [9]. With the gyroscopes spinning, the main spindle is torqued.
This causes the gyroscopes to flip upward at the pivot points, giving the impression
that some antigravity effect is lifting them.

illustration is intended only to help visualize the torque directions from angular
momenta conservation. Because the rotation of the main spindle has introduced
an angular momentum in this view, the stem of the gyroscope has shifted its
alignment so that the gyroscope now presents a countering angular momentum
so that their sum conserves the initial zero-valued angular momentum. When
tilted, the projection of the gyroscope in this view now provides a rotational con-
tribution. Even though it is only an ellipse in this view (i.e., a circle viewed at an
angle), this is enough to project a portion of its angular momentum into the plane
of rotation of the spindle. When the gyroscopes’ stems reach the end of their
allowed tilting angle, the torque that induced the tilt will still exist, but it will
be acting among the pivot, stem, and the stem’s stop. It is not an “upward”
force relative to the gravitational field, but rather a torque.
Experimental testing of these gyroscopic devices is not so easy as with the
oscillation thrusters. The key difference is that the gyroscopic thrusters require
alignment with the Earth’s gravitational field to operate. They do not produce
lateral thrust that can be tested with the pendulum methods previously discussed.
Instead, their weights must be measured. This is difficult because torques are
introduced between the device and the platform on which it is mounted that
is, reaction forces from its internal mechanisms. Furthermore, the frequency of
258 M. G. MILLIS

Before After
(1) The spindle
now has angular
1
Top View
momentum
2

The gyro is spinning, but none (2) To counter, the

of its angular momentum is gyroscope tilts to
projected in this view project a portion of
its momentum along
the spindle's axis
Side View

The spindle
2
is not
rotating

Fig. 10 Picturing conservation of angular momenta. The angular momenta of both

situations is identical but achieved differently. On the left, the gyroscope’s rotation is
aligned perpendicular to the axis of the spindle (zero contribution along the viewed
axis) and the spindle is not rotating, hence the total angular momentum is zero. On
the right, the spindle has been set into motion, introducing its own angular
momentum. To conserve angular momentum, the gyroscope tilts to project an
opposing portion of its angular momentum in the axis of the spindle, so that their
sum remains zero. Note that this is not a strict representation of the dynamics of
this system, but rather a device to help understand the directions that such
coupled wheels will move.

vibrations from the gyroscopes might interfere with the operation of any weight-
measuring device.
Drop tests, where fall times are measured to detect any changes in gravita-
tional acceleration, are not likely to be a viable testing option for two reasons.
First, typical gyroscopic thrusting devices require the presence of the gravita-
tional field, which vanishes in free-fall. To illustrate this with a simple experi-
ment, drop a spinning top that is precessing. The moment that it begins its free
fall, the precessing will stop. Second, the device is not likely to survive the
rapid deceleration at the end of its fall.
One testing option, as illustrated in Fig. 11, is to place two identical devices on
each end of a balance beam and look for any tilting of the balance beam when one
device is free to operate normally and the other has its stems locked to prevent the
upward motion of the gyroscope. Even this simple test can be difficult because it
requires two devices and can be vulnerable to spurious effects that could tilt a
sensitive balance. Also, the transient impulses from the starting and stopping
of the gyroscopes’ tilting will induce oscillations in the balance beam.
Again, to keep an open yet rigorous mind to the possibility that there has
been some overlooked physical phenomenon, it would be necessary that any
experiments explicitly address all the conventional objections and provide
convincing evidence to back up the claims. Any test results would have to be
 NONVIABLE MECHANICAL “ANTIGRAVITY” DEVICES 259

Inactive Antigravity Active Antigravity

Gyro Device Gyro Device

Does the Scale Tilt?

Fig. 11 Testing antigravity gyroscopes. One method of testing the antigravity claims
of gyroscopic devices is to place two at either end of a balance beam and to see if there
is any difference between an inactive and an operating device. Even though an
impulse is expected as the gyroscopes begin their upward tilt, a countering impulse
is equally expected when the gyroscopes reach the end of their allowed travel and
stop tilting upward. This sequence of impulses is likely to set the balance into
oscillation, but a sustained tilting of the balance beam is not expected.

rigorous, impartial, and address all possible causes that might lead to a false-
positive conclusion.
One approach to assess such concepts is the “control volume” analysis [10].
To start, a control volume is mathematically constructed around the device.
Next the vectors of all the forces and any phenomena that carry momenta that
cross the surface of that volume are summed and applied to the center of mass
of the system. This includes external masses, fluids, magnetic fields, gravitational
fields, electromagnetic radiation, and so on. Motions within the control volume
need not be considered, including mechanical devices, fluid flow, internally gen-
erated fields, and so on. If nothing crosses the surface that can carry momentum,
then there is no net acceleration.
The orientation of this control volume, however, is not fixed. The internal
motion of gyroscope can alter which direction a device will point without affect-
ing its rectilinear motion. Such is the case of “reaction wheels” discussed next.

IV. Unrelated Devices

It is necessary to make clear distinctions between these gyroscopic antigravity
devices and similar-sounding devices, specifically “reaction wheels,” spinning
superconductors, rotating masses in general relativity, and gyroscopic variants
of the previously described oscillation thrusters.
A reaction/momentum wheel or torque wheel is an established device used in
satellites to change the satellite’s pointing direction [11]. Even though the
angular momentum of the entire system remains constant, the angular orientation
and momentum of the external structure and its internal masses can be changed
with respect to one another. For example, by rotating an internal wheel
clockwise, the external cage (i.e., satellite body) will rotate counterclockwise.
This is a simple and effective way to change the pointing direction of satellites,
260 M. G. MILLIS

but such devices cannot be used to change the position of the center of mass of
the system.
Another similar-sounding claim is the gravity shielding claim involving spin-
ning superconductors [12]. These superconductor claims were later shown not to
be reproducible [13]. Further claims of gravitomagnetic effects using spinning
superconductors [14] are still under review at the time of this writing. These
are separate effects that should not be confused with claims of gyroscopic
antigravity.
The concept of using rotating masses to induce forces does have a treatment
within General Relativity that can lead to acceleration fields, but it is through
the very feeble effect of frame-dragging. In 1963, Robert Forward calculated
the induced acceleration field from a rapidly rotating ultra-dense toroid [15].
The magnitude of the induced effect is impractically trivial compared to the con-
figurations needed to produce the effect. This example does serve, however, as a
theoretical treatise on the subject and is discussed in more detail in Chapter 4.
When gyroscopic devices constrain their motions to a single plane instead of
the multi-direction axes versions that are the focus of this section, then they are
just a variation of the oscillation thrusters discussed previously. The version
shown in Fig. 2 is a classic example.

V. Conclusions
Claimed breakthroughs that are based on errant interpretations of mechanical
forces are common. Two devices in particular involve oscillating masses that
claim net thrust and gyroscopic devices that claim antigravity effects. The
oscillation thrusters are misinterpretations of differential friction, while the gyro-
scopic devices misinterpret torques as linear thrust. To help reduce the burden on
reviewers and to give would-be submitters the tools to assess these ideas on their
own, examples of these devices, their operating principles, and testing criteria
have been discussed.

References
[1] Dean, N. L., “System for Converting Rotary Motion into Unidirectional Motion,”
U.S. Patent 2,886,976, 19 May 1959.
[2] Foster, Sr., R. E., “Inertial Propulsion Plus/Device and Engine,” U.S. Patent
5,685,196, 11 Nov. 1997.
[3] Thornson, B. R., “Apparatus for Developing a Propulsion Force,” U.S. Patent
4,631,971, 30 Dec. 1986.
[4] Millis, M. G., “Assessing Potential Propulsion Breakthroughs,” New Trends in
Astrodynamics and Applications, Belbruno, E. (ed.), Annals of the New York
Academy of Sciences, New York, Vol. 1065, pp. 441 461.
[5] Resnick, R., and Halliday, D., Physics, 3rd ed, Wiley, New York, 1977, pp. 123, 314.
[6] Davis, “The Fourth Law of Motion,” Analog Science Fiction, Science Fact, 1962,
May, pp. 83 105.
[7] The Royal Institution Heritage: Ri People, Royal Institution of Great Britain,
London, UK, 2006; URL: http://www.rigb.org/rimain/heritage/ripeople.jsp [cited
1 June 2006].
 NONVIABLE MECHANICAL “ANTIGRAVITY” DEVICES 261

[8] Laithwaite, E., and Dawson, W., “Propulsion System,” U.S. Patent 5,860,317, 19 Jan.
1999.
[9] Childress, D. H., The Anti Gravity Handbook, Adventures Unlimited Press, Stelle,
IL, 1985, pp. 18, 20.
[10] Hill, P. G., and Peterson, C. R., Mechanics and Thermodynamics of Propulsion,
Addison Wesley, Reading, MA, 1965.
[11] Bayard, D. S. “An Optimization Result With Application to Optimal Spacecraft
Reaction Wheel Orientation Design,” Proceedings of the American Control Confer
ence 2001, Vol. 2, Arlington, VA, June 25 27, 2001, pp. 1473 1478.
[12] Podkletnov E., and Nieminen, R., “A Possibility of Gravitational Force Shielding
by Bulk YBCO Superconductor,” Physica C, Vol. 203, 1992, pp. 441 444.
[13] Hathaway, G., Cleveland, B., and Bao, Y., “Gravity Modification Experiment Using
a Rotating Superconducting Disk and Radio Frequency Fields,” Physica C, Vol. 385,
2003, pp. 488 500.
[14] Tajmar, M., Plesescu, F., Seifert, B., Schnitzer, R., and Vasiljevich, I., “Search for
Frame Dragging Like Signals Close to Spinning Superconductors,” Proceedings of
the Time and Matter 2007 Conference, Bled, Slovenia, World Scientific Press, 2007.
[15] Forward, R. L., “Guidelines to Antigravity.” American Journal of Physics, Vol. 31,
1963, pp. 166 170.
 Chapter 7

Null Findings of Yamishita

Electrogravitational Patent

Kenneth E. Siegenthaler and Timothy J. Lawrence†

U.S. Air Force Academy, Colorado Springs, Colorado

Nomenclature
a significance level (0.001)
D0 test value (zero)
m mean, general case
mNeg Unc difference in means for negative/uncharged scenario
mPos Unc difference in means for positive/uncharged scenario
d̄ difference in means, general case
Ha alternate hypothesis
H0 null hypothesis
n number of observations
n 1 degrees of freedom
s standard deviation, general case
sd difference in standard deviations
t test statistic
ta,n 1 rejection region

I. Introduction
A. Theoretical Background
LECTROGRAVITATIONAL theory holds that moving charges are
E responsible for the mysterious phenomenon we know as gravity. According

This material is a work of the U.S. Government and is not subject to copyright protection in the
United States.

Professor of Astronautics, Department of Astronautics.
†Director, Space Systems Research Center.

263
264 K. E. SIEGENTHALER AND T. J. LAWRENCE

to Nils Rognerud, author of the paper “Free Fall of Elementary Particles: On

Moving Bodies and Their Electromagnetic Forces,” gravity as we know it is
“. . . simply a pseudo-force, produced by the special non-shieldable dielectric
effect which is produced by the relativistic motions of orbital electrons of ordin-
ary matter” [1]. Much like a moving charge will induce a magnetic field, it will
also induce a gravitational field of a much smaller magnitude [1].
Essentially, according to electrogravitational theory, all gravity is produced by
the motion of orbital electrons in the atoms that comprise all matter. This sup-
posed dielectric effect is also additive, unlike the magnetic forces of randomly
oriented atoms that cancel each other out [1]. The obvious implication of this
statement is that more massive objects, hence possessing more orbital electrons
than less massive objects, will produce a greater electrogravitational force. This
is a simple explanation of why the Earth possesses a stronger gravitational field
than the moon, why the sun possesses a stronger gravitational field than the Earth,
and so on.
Figure 1 is a simple illustration depicting the manner in which a charged par-
ticle induces an electrogravitational field. Notice that the produced field is normal
to the direction of motion.
Figure 2 depicts how an atom produces an electrogravitational field around
itself. Again, this field is additive; it becomes stronger with the presence of
additional atoms [2].
Electrogravitational theory is much more involved than the brief description
provided above. However, this elementary description could be applied to
artificially inducing an electrogravitational field.

B. Experimental Background
By virtue of the assertion that moving particles will induce an electro-
gravitational field, a charged, rotating cylinder should produce an electrogravity

Fig. 1 Particle inducing an electrogravitational field.

 NULL FINDINGS OF YAMISHITA 265

Fig. 2 Atom inducing an electrogravitational field.

field normal to the plane of rotation. If the cylinder is sufficiently charged and
rotating rapidly enough, it should alter its weight equivalent in measurable
ways. Yamashita attempted this in his 1991 experiment [2]. The result he
obtained, an 11-g reduction in weight equivalent of a 1300-g device, is significant
[2]. Unfortunately, no one has been able to reproduce his experiment to date.
Figure 3 is an illustration of the device that Yamashita had envisioned for the
artificial induction of an electrogravitational field.
In order to test the validity of his hypothesis, Yamashita constructed a
device similar to Fig. 3 that rotated at approximately 50 revolutions per
second (3000 revolutions per minute) [2]. It was comprised of four main
parts: a base plate, an electrode, a rotor, and an electric motor [2]. All
parts, except the motor, were machined from aluminum. The electrode was
130 mm in diameter and 5-mm thick. It was coated on the interior with a
dielectric coating, although Yamashita makes no mention as to what
chemicals he used to achieve that purpose. The rotor was 127 mm in diam-
eter, 5-mm thick, and 60-mm high. It also was coated presumably with the
same dielectric coating, except on the outside. The machine, at rest, weight
equivalent was 1300 g.
In order to test his machine, Yamashita placed it on a scale with a resolution of
1 g. He tested the machine, uncharged, and rotated it to its maximum speed. The
difference in weight equivalent between the machine at rest and the machine at
this speed was less than 1 g. Yamashita concluded that there was, in fact, a
difference in weight equivalent that his scale could not detect, and he attributed
this to the rotor’s interaction with the surrounding air.
Yamashita applied a charge to the device by bringing into contact for one
minute the charged Van de Graaf generator’s spherical electrode and machine’s
266 K. E. SIEGENTHALER AND T. J. LAWRENCE

Fig. 3 Yamashita’s device.

electrode. He applied a current of 0.5 A to the motor. As the rotor accelerated, the
scale read increasingly lower weight equivalent until at top speed, it read a weight
equivalent of 1289 g. This represents a decrease of 11 g, or a 1% decrease in the
machine’s weight equivalent.
Yamashita then reversed the polarity of the rotor. To do this, he attached a
spherical electrode to the positive terminal of the generator. He brought the
sphere and the machine’s electrode into contact, although his patent does not
indicate for how long. With the polarity reversed, the machine increased its

Fig. 4 Yamashita’s experimental setup.

 NULL FINDINGS OF YAMISHITA 267

weight equivalent by 4 g at top speed. From this experiment, Yamashita con-

cluded that “horizontal rotation of a charged body generates a vertical force,”
“when the polarity of the charges supplied to the rotating body is reversed,
the direction of the generated vertical force is also reversed,” “the faster the
body is rotated, the stronger is the generated vertical force,” and “the direction
and strength of the generated vertical force does not depend on the direction of
the body” [2]. Yamashita arrived at the last conclusion after his machine
produced a force when oriented at an angle. Figure 4 depicts Yamashita’s exper-
imental setup.
No one has been able to reproduce Yamashita’s experiments to date. A 1%
decrease in the weight equivalent of an object is very significant, and it was
the aim of the experiments conducted at the United States Air Force Academy
to validate Yamashita’s claims and to determine whether or not electrogra-
vitational theory is valid.

II. Experiment 1
A. Methods
Efforts were made to follow Yamashita’s experiment as closely as possible.
His patent, although somewhat vague, gave enough information to conduct an
experiment reasonably true to his original.
The first step in attempting to replicate his experiment was to construct a
machine reasonably similar to the one he used. If electrogravitation is indeed a
real phenomenon, minor differences between the two machines theoretically
should not matter in terms of producing results.
In choosing an electric motor to power the rotor, a Global Super Cobalt
400 27T motor was selected. This motor was originally intended to power
radio-controlled aircraft, and it was for this reason that it was chosen to power
the replica device. The motor boasts a stall current of 64 amps and is capable
of reaching 19,500 RPM without a load. Compared to most other electric
motors, the Global motor is especially powerful and should have no trouble spin-
ning the rotor at speeds higher than those attained by Yamashita’s device. Higher
speeds, in theory, should induce a larger electrogravitational field that is easier to
measure.
A preliminary design was made using Autodesk Inventor Version 7.0. The
replica machine was designed specifically for the Global electric motor and,
therefore, it was not exactly the same as Yamashita’s device. Figure 5 depicts
the replica’s electrode.
Figures 6, 7, and 8 depict the rotor, base plate, and motor mounts, respectively.
Yamashita’s machine used an unspecified dielectric material to insulate the
electrode from the rotor. Dielectric materials come in many different forms,
ranging from gels to baked-on coatings to solid sheets. For the purposes of the
replica machine, the surfaces that required a dielectric coating were prepared
using one coat of gray automotive primer. To act as the dielectric, four coats
of blue enamel paint were applied over the primer. The particular enamel used
contained the chemical Xylene, which is known to have a dielectric constant
of 2.5 at 258C [3]. With the dielectric applied, the replica machine was
assembled. A specially made nylon washer was used to further insulate the
268 K. E. SIEGENTHALER AND T. J. LAWRENCE

Fig. 5 Replica machine electrode.

rotor from the motor shaft. Finally, to dampen vibrations, foam padding was
applied to the base of the machine.
Figure 9 depicts the appearance of the fully assembled machine. With the
machine fully assembled, a mathematical relationship could be developed

Fig. 6 Replica machine rotor.

 NULL FINDINGS OF YAMISHITA 269

Fig. 7 Replica machine base plate.

between input current and rotor speed. To power the rotor, an Energy Concepts,
Inc. Model 20600B high current power supply was used. Due to the high
power requirements of the Global electric motor, a regular power supply could
not provide sufficient power to turn the rotor, let alone attain the required rotation
rate. On the Model 20600B power supply, the 0-24 volt setting was selected. The
amps setting was then selected to display the supplied current. A vertical line was
drawn on the rotor for purposes of speed calibration. A Power Instruments Digist-
robe Model M64 strobe light (calibrated up to 10,000 RPM) was used in conjunc-
tion with the vertical calibration line to obtain the rotor speed. Input current was
varied, and the rotor speed associated with that current was recorded. Table 1
shows input currents and the resulting rotor speed.
With Table 1, it is relatively easy to develop a function that relates input
current to rotor speed using Microsoft Excel. Figure 10 is the Excel graph of
Table 1, with rotor speed function shown.
The data points fit well on the third-order polynomial trendline developed
by Excel. Equation (1) is the relationship between input current and rotational
speed.

rpm ¼ 75:341i3  686:26i2 þ 2229:2i  2047:1 (1)

270 K. E. SIEGENTHALER AND T. J. LAWRENCE

Fig. 8 Replica machine motor mount.

Fig. 9 Replica machine.

 NULL FINDINGS OF YAMISHITA 271

Table 1 Input current and rotor speed

Input current (amps) Rotor speed (RPM)

3 500
4 700
4.7 1,130
5 1,330
5.8 2,500

This equation is also shown on the graph. Knowledge of this equation was necess-
ary to the experiment. It would not be possible to use a strobe light to monitor the
rotor speeds during the test, because the electrode would effectively cover the
rotor. It should be noted that the rotor was always spun in a counterclockwise
direction; the Global motor is not designed to rotate in the other direction.
To test the replica machine, a thorough test plan was followed. This plan
required the use of the Model 20600B power supply to spin the rotor. It also
required the use of a Van de Graaf generator to charge the machine. For this
purpose, a Wabash Instrument Corp. Winsco Model N-100V generator was
used. This particular generator is capable of developing charges up to 250 kv
[4]. Finally, to record the possible weight equivalent to change, an Ohaus I-10
FE-7000 precision scale was used, which is capable of supporting up to 25 kg,
with a resolution of 1 g. Figure 11 is a photograph of the entire test apparatus;
Fig. 12 is a photograph of the machine, the scale, and the power supply.
First, it was necessary to determine whether or not the high levels of static
electricity from the Van de Graaf generator had any adverse effects on the
scale. To accomplish this, the device was charged for one minute and rotated
at full speed in close proximity to the scale. If the reading on the scale remained
constant throughout this process, then it would be assumed that there was no
interference.

Rotor Speed vs. Supply Current

RPM = 75.341i3 – 686.26i2 + 2229.2i – 2047.1
6000

5000
Rotor Speed (RMP)

4000

3000

2000

1000

0
0 1 2 3 4 5 6 7 8
Input Current (Amps)

Fig. 10 Rotor speed graph and function.

272 K. E. SIEGENTHALER AND T. J. LAWRENCE

Fig. 11 Test apparatus.

After establishing that the scale was not affected by the presence of static
electricity, the generator was switched off and the machine discharged. The
uncharged machine was then placed on the scale. Double-sided tape was used
to hold the underside of the machine to the scale. This ensured that the
machine did not vibrate across the smooth surface of the scale. The wires were
placed in such a way that they did not affect the readings on the scale
(Fig. 12). Throughout the course of the experiment, the machine would not
leave the scale.
Uncharged, the weight equivalent to of the machine was taken at 500 RPM
intervals up to its top speed of 5000 RPM. The purpose of this was to ensure

Fig. 12 Detail of machine, scale, and power supply.

 NULL FINDINGS OF YAMISHITA 273

that the machine’s vibrations did not cause erratic readings, as well as provide a
basis for comparison for charged operation. After this test, the machine was
allowed to come to a complete stop.
The next test involved charged operation. The machine was charged for one
minute from the Van de Graaf generator. A wire was used to connect the spheri-
cal electrode of the generator to the electrode on the machine. After the machine
was fully charged, the Van de Graaf generator was switched off and grounded to
ensure it carried no residual charge that might have caused interference. As was
the procedure of the previous test, the weight equivalent to of the machine was
recorded at 500 RPM intervals, from 0 to 5000 RPM. At these intervals, the
machine’s weight equivalent was recorded.
Afterwards, the machine was allowed to stop and was discharged. The Van de
Graaf generator was again used to charge the machine. This time, however,
contact was made through a wire from the generator’s positive electrode on its
base to the machine’s electrode for one minute. The previous test was again
conducted, this time with the opposite charge, and the weight equivalent
values were recorded.
The next series of tests involved testing, at full speed, various levels of charge
and their effects on the machine’s weight equivalent. Using the same procedures
previously described for charging, the machine was charged in 10-s intervals,
ranging from 10 to 60 s, from the negative spherical electrode of the generator.
At each of these charge intervals, the machine operated at full speed, and its
weight equivalents were recorded. Between each charge interval, the machine
was allowed to come to a complete stop and be fully discharged. For example,
the first step of this test involved charging the machine for 10 seconds.
It was then tested, stopped, and discharged. The second step involved
charging the machine for 20 s and repeating the same process. The third step
involved charging the machine for 30 s, and so on, until the machine received
one full minute of charging. Finally, this same test was conducted with the
machine receiving a positive charge from the electrode on the base of
the generator.
It was hoped that by following these procedures that Yamashita’s claim would
be validated. Furthermore, it was hoped that by using more thorough procedures,
electrogravitational theory could either be validated or rejected.

B. Results and Discussion

Although not conclusive by any means, the results of this experiment were
nonetheless interesting. By operating the fully charged machine in close proxi-
mity to the scale, it was determined that the static electricity had no adverse
effect on the other equipment, especially the scale.
The first test involved the uncharged machine at various speeds. Table 2
contains the input currents, associated speeds, and weight equivalents of
the machine.
As the rotor was accelerating, the scale fluctuated at +2 g above and below
the central value. When the rotor reached a steady speed, the fluctuation in
some cases disappeared; in others was +1 g about the central value. In these
cases, it was the central value that was recorded as the machine’s weight
274 K. E. SIEGENTHALER AND T. J. LAWRENCE

Table 2 Uncharged rotation, speed varied

Machine weight
Supply current (A) Rotor speed (RPM) equivalent to (g)

0.0 0 1315
3.0 500 1315
4.6 1000 1315
5.1 1500 1315
5.5 2000 1315
5.8 2500 1316
6.0 3000 1316
6.3 3500 1316
6.5 4000 1316
6.6 4500 1316
6.8 5000 1316

equivalent. These fluctuations for the most part seemed to be caused by the
natural vibrations in the machine, although the rotor’s interaction with the sur-
rounding air plays a part as well.
The next test involved varying rotor speed with a full negative charge. Table 3
contains the test data from this scenario, as well as the differences in weight
equivalent between the present and the uncharged test cases.
The next test involved varying rotor speed with a full positive charge. The data
for this scenario are listed in Table 4, again showing the change in weight equiv-
alent from the uncharged test case.
Figure 13 compares the percent of weight equivalent to deviation during the
rotation experiments of an uncharged, negatively charged and positively
charged cylinder of Experiment 1.

Table 3 Negative charge, speed varied

Machine weight Weight equivalent to

Rotor speed (RPM) equivalent to (g) change (g)

0 1315 0
500 1315 0
1000 1315 0
1500 1315 0
2000 1315 0
2500 1315 1
3000 1315 1
3500 1314 2
4000 1314 2
4500 1314 2
5000 1314 2
 NULL FINDINGS OF YAMISHITA 275

Table 4 Positive charge, speed varied

Machine weight Weight equivalent to

Rotor speed (RPM) equivalent to (g) change (g)

0 1315 0
500 1315 0
1000 1315 0
1500 1316 þ1
2000 1316 þ1
2500 1316 0
3000 1316 0
3500 1316 0
4000 1316 0
4500 1317 þ1
5000 1317 þ1

The next test involved varying negative charge at full rotational speed. Table 5
lists the resulting data.
After obtaining the data for the 60 s charge time at full speed, the machine
broke. An inaccessible screw holding the motor to the motor bracket came
loose, and with no way to adequately tighten it, the machine would vibrate vio-
lently above 500 RPM. At this point, enough data had been taken; the constant
speed, varying positive charge test was scrubbed.

EXPERIMENT 1 RESULTS
0.1

0.05
PECRENT WEIGHT DEVITAION

0
0 1000 2000 3000 4000 5000 6000

–0.05

–0.1
UNCHARGED ROTATION

NEGATIVE CHARGE
–0.15
ROTATION
POSITIVE CHARGE
ROTAION
–0.2
ROTOR SPEED (RPM)

Fig. 13 Comparison of the percent weight equivalent to deviation results for

Experiment 1.
276 K. E. SIEGENTHALER AND T. J. LAWRENCE

Table 5 Constant speed, varied negative charge

Machine weight Weight equivalent to

Charge time (s) equivalent to (g) difference (g)

0 1315 0
10 1315 1
20 1315 1
30 1315 1
40 1314 2
50 1314 2
60 1314 2

Enough data had been acquired to statistically test whether or not this
experiment offered proof that electrogravity was indeed a real phenomenon.
Even though there was an apparent decrease in weight equivalent through nega-
tively charged rotation and an increase through positively charged rotation (as
Yamashita had predicted), it may not have been a measurable enough of a
change to reach a definite conclusion.
The varied rotation speed with one minute charge will be of statistical interest.
The uncharged rotation will be used as a basis for separate comparison between
the negative- and the positive-charged cases. These data will be considered paired
data, because they consists of two observations on the same unit, that unit being
speed of rotation. Consequently, a t-test will be used. The null hypothesis, in this
case, is that there is no difference between the mean weight equivalent of the
uncharged scenario and that of whatever it is being compared to, either the posi-
tive or negative. On the contrary, the alternative hypothesis holds that there is a
difference in the means. The t-test will be used to either reject or accept the null
hypothesis. Rejection of the null means that it is statistically sound to accept elec-
trogravitational theory based on the experimental data. Accepting the null sig-
nifies the opposite, or the experimental data are statistically insufficient to
prove the existence of an electrogravitational force.
Table 6 shows all pertinent data for statistical analysis of the negative-
charged case.
The two hypotheses in this case are as follows:

H0 : mNeg Unc ¼0
Ha : mNeg Unc ,0 (2)

Table 6 Negative charge statistical data

Uncharged Negative Difference

m 1315.5455 1314.636364 0.90909

s 0.522233 0.504524979 0.94388
 NULL FINDINGS OF YAMISHITA 277

For this analysis, a significance level of a 0.001 will be used. Furthermore,

there are a total of n 11 observations. For this test, the test statistic will be as
follows:

d  D0
t¼ p (3)
Sd = n

This test statistic will be compared to t0.001,11 1 4.144. If the following

relationship is found to be true, then H0 will be rejected.

t , ta, n 1 (4)

Stepping through all of the math for the comparison of the uncharged and
negative-charged data, the following is found to be true:

0:90909  0
t¼ p ¼ 3:194 (5)
0:94388= 11

This particular test value, when compared to t0.001,11 1 4.144, makes

Eq. (4) false. Therefore, it is necessary to fail to reject the null hypothesis. The
experimental data for the negative case are insufficient to prove the existence
of an electrogravitational force.
The same test must be performed between the uncharged and the positive-
charged scenarios. Table 7 contains all the necessary data for this comparison.
Unlike the previous case, this case requires a different set of hypotheses. Ha
must be the opposite for this case, because positive rotation tended to cause
the machine to gain weight equivalent rather than lose weight equivalent. The
set of null hypotheses required for this test are as follows:

H0 : mPos Unc ¼0
Ha : mPos Unc .0 (6)

Likewise, the test relationship shown in Eq. (4) must be changed. The test
relationship now becomes:

t . ta,n 1 (7)

Table 7 Positive charge statistical data

Uncharged Positive Difference

m 1315.5455 1315.909091 0.363636

s 0.522233 0.70064905 0.8202
278 K. E. SIEGENTHALER AND T. J. LAWRENCE

Using Eq. (5) to obtain the test statistic,

p
t ¼ (0:363636  0)=(0:8202= 11) ¼ 1:470 (8)

Using t0.001,11 1 4.144 as in the previous case, Eq. (7) is not satisfied.
Therefore H0 must be accepted in this case also. There is not enough evidence
from the experimental data from the positive case that electrogravity is a real
force. Statistical analysis of the varied charge test is not necessary. The analysis
performed on the initial data proves that the experimental data are not sufficient to
confirm the existence of an electrogravitational force.

III. Experiment 2
A. Methods
After the results obtained in Experiment 1, another attempt was made to repli-
cate the results and make improvements. However, funding issues precluded the
implementation of a key improvement, which was conducting the experiment in a
vacuum environment. After several inquiries to different electric motor manufac-
turers, it was found that a vacuum rated motor capable of spinning the rotor at the
desired RPM would cost roughly $1000. Additionally, the manufacturers would
not be able to supply a motor in a fashion that would meet the timetable require-
ments. Many of the motors were out of stock or would have to be custom-made
for the purposes of this experiment. This postponed the arrival of a motor by
several months, precluding their use even if funding were available.
Another objective of the second experiment was to investigate the reason why
the previous experiment’s machine failed at high speeds. The rotor used pre-
viously was plagued with balancing issues. These were first believed to have
come from painting the rotor; however, further investigation concluded that
the rotor used in Experiment 1 was in fact out of balance without the paint and
that the hole drilled for the motor shaft interface was out of center. Additionally,
the motor mount previously designed was determined to be insufficient to keep
the machine stable at high RPM, especially given the imbalanced rotor. This con-
clusion was reached through inspection of the motor mount when fully assembled
and attached to the base plate and motor. The motor mount was comprised of only
two brackets, attached to the motor and base plate with a total of four screws as
illustrated in Fig. 8. The motor mount was also attached to the motor with hot
glue. The motor used was also questioned, as the slender shaft of the motor
may have contributed additional instabilities of the experiment at high speeds.
Another discrepancy found was that the electrode did not fully enclose the
rotor there was a hole at the top that provided access to a screw that held the
rotor to the motor interface. Whether or not this hole affected experimental
results has not been determined, but it can be hypothesized that such a hole
could introduce a larger possibility for aerodynamic effects as opposed to a
completely enclosed and sealed device. The electrode issued in Experiment 1
and Yamashita’s device can be seen in Figs. 3 to 5. From Yamashita’s drawings,
it is seen that neither iteration included a hole in the electrode component.
 NULL FINDINGS OF YAMISHITA 279

The experiment was conducted as closely as possible to Yamashita’s

experiment; however, the ambiguity of the European patent application led to
educated guesses on a few aspects during the design of the machine. In addition,
accuracy limitations of the manufacturing tools at the Air Force Academy intro-
duced errors to the components, more specifically the rotor. These ambiguities
and limitations led to many iterations of the machine while in the construction
phase of the experiment. The initial design for the machine can be seen in
Fig. 14, but the final iteration of the machine differed greatly from this
drawing. (The final version of the machine is shown later.)
An enlarged view of the new motor mount design can be seen in Fig. 15. It is
easily seen that this design would be much more stable than the two-bracket
design from the previous machine; however, implementation of this motor
mount was not possible due to issues with the motors, detailed later.
Given the criticality of accuracy in manufacturing the components, the tools
used were integral in allowing the components to meet exacting standards;
however, it was determined that the facilities at the Air Force Academy were
not capable of machining within such tolerances. The limitations present were
a great hindrance, especially in manufacturing the rotor. The rotor that was
manufactured for Experiment 2 was out of balance in addition to being out of
round. The out of roundness of the rotor was mainly due to the thinness of the
aluminum. When compared to Experiment 1 rotor (mounted on Experiment
2’s motor), the vibrations experienced by the new rotor were much greater; there-
fore Experiment 1’s rotor was reused in this new experiment. In addition it was
shortened to reduce the mass moment of inertia values. The same thickness was
kept to maintain the durability of the rotor.
After analysis of the previous experiment’s motor, it was hypothesized that the
motor shaft was bent, contributing to the vibrations experienced. Three different
motors were tested with the old rotor and each experienced similar vibration.
Although specific dimensions of the motor and motor shaft were not provided
in Yamashita’s patent application, further investigation of his drawings led to

Fig. 14 Initial design of the machine.

280 K. E. SIEGENTHALER AND T. J. LAWRENCE

Fig. 15 Motor mount.

the conclusion that a motor with a very thick shaft was used in his experiment;
therefore, a shaft diameter of about 0.25 in. was decided upon. Furthermore a
thicker shaft would bolster the stability of the spinning rotor given how imbal-
anced it was. Hobby shops did not provide motors with a shaft thickness in the
range of 0.25 in. therefore more creative means were employed to find a desirable
motor. A 1.30 VAC brushless electric fan motor with a shaft thickness of 6 mm
(slightly under 0.25 in.) was found at a Goodwill store. To power the motor, an
Energy Concepts, Inc. Model 20600B high current power supply was used.
The fan cover and blades were removed leaving only half of the fan casing, the
motor, speed control, and stand. The integrated speed control of the fan was set to
maximum for the entire experiment. A picture of that assembly can be seen in
Fig. 16.
Because of the peculiar way the motor was mounted to the fan assembly, it
would have been difficult and time consuming to design and build a new
mount for the fan motor. Therefore, it was decided that the base plate would
be mounted on top of the fan motor with the shaft protruding from the bottom
of the base plate. This is illustrated in Fig. 17.
Because the design had gone through several iterations at this point, different
sets of holes had been placed on the base plate. After initial construction it was
determined that these should covered with aluminum tape to mitigate airflow
through this section.
Additionally, the motor mount interface posed a problem as it presented more
opportunities for inaccuracies to be introduced. Experiments is rotor-to-motor
interface was held to the motor shaft by friction fit, but after many uses this inter-
face became loose. Another method of interfacing the motor to the rotor was
improvised by using a small hand drill chuck to clamp onto the motor shaft.
Running the motor at maximum speed with the chuck attached produced
minimal vibrations. Further detail as to how these components were interfaced
is illustrated in Fig. 18.
 NULL FINDINGS OF YAMISHITA 281

Fig. 16 Modified fan.

A reference line was drawn on the chuck. Then, using a strobe light rated to
over 18,000 RPM and the Model 20600B placed on the 130 VAC setting, the
maximum rotational speed of the motor was determined to be roughly
3390 rpm. Though this was much slower than the previous maximum speed, it
was sufficient to meet the specifications detailed in Yamashita’s experiment.
The method of determining the rotational speed of the rotor in relation to the
current supplied was identical to the previous experiment and is further discussed
later in the paper. The rotor-to-motor interface was attached to the chuck via
screw. Because the previous experiment’s interface produced minimal vibrations
when attached to the motor, it was also reused for this experiment; however it was
redrilled and tapped to hold a screw that fit into the chuck. The interface was
attached to the rotor via two countersunk screws that completed the assembly
for attaching the rotor to the motor shaft. Although much more complex, when
fully assembled the rotor spun with less vibration than the previous machine.

Fig. 17 Base plate mounted on fan.

282 K. E. SIEGENTHALER AND T. J. LAWRENCE

Fig. 18 Drill chuck on motor shaft.

Unfortunately, at high speeds these vibrations were still apparent. The final
assembly is illustrated in Fig. 19.
The vibrations were mitigated through manually balancing the rotor. Manual
balancing was necessary because machine shops offering professional balancing
services were not able to fit the rotor on their balancing machines as the shaft size
for the motor was too thin. A shop was found which was able to balance rotors
with smaller shaft sizes (specifically turbo chargers for cars), but this develop-
ment occurred too late in the manufacturing process to meet deadline
requirements.

Fig. 19 Rotor mounted on shaft.

 NULL FINDINGS OF YAMISHITA 283

Given that the hole drilled for the interface was off center, a micrometer was
used to determine the point on the rotor with the smallest radius. Additional mass
was added to this point by attaching solder to the inside of the rotor with alumi-
num tape. This addition greatly minimized the vibrations of the rotor, especially
at high speeds. Fine tuning of the rotor’s balance was achieved through an itera-
tive process of adding a mass (section of 18-gauge wire attached with duct tape),
spinning up the motor to its maximum velocity, judging whether the addition
decreased the total vibration of the machine, and moving the mass to a new
location. Locations with the least vibration were marked, compared, and the
optimum location of the additional mass was determined through observation.
After several masses were added to the rotor (all on the inside surface), it
became increasingly difficult to judge the differences in severity of vibration.
At this point it was considered that the rotor was balanced to the maximum
extent possible given the method implemented.
Once the rotor was balanced, it was painted and retested to see if the paint had
any noticeable affects. Because Yamashita’s patent application did not specify
the exact dielectric layer used, Vanguard Class F Red VSP-E-208 Insulating
Enamel was utilized for this experiment to insulate the inner surface of the elec-
trode and the outer surface of the rotor. This insulating enamel is specifically
designed to insulate electrical components. To ensure an even coating on the
outside of the rotor, the paint was applied as the rotor was spinning. When
applied the paint did not have any adverse affects on the balance of the rotor.
Once the paint was applied, a calibration curve that related rotor speed to
applied current was developed using the strobe light and by making a reference
line on the rotor. The results can be seen in Table 8.
From the data in Table 8 a linear regression between each point was derived
using the TREND function in Microsoft Excel. This function determined the
RPM associated with intermediate levels of current. The TREND function was
used between each point because the regression lines that Excel produced did

Table 8 Relation of amps to RPM

Amps RPM

0.20 621
0.21 1983
0.22 2860
0.23 3015
0.25 3165
0.30 3277
0.35 3330
0.40 3350
0.45 3368
0.50 3377
0.60 3385
0.74 3390
284 K. E. SIEGENTHALER AND T. J. LAWRENCE

Amps to RPM conversion

4000
3500
3000
RPM 2500
2000
1500
1000
500
0

45
35
22

25

6
2

0.
0.

0.
0.

0.

0.
Amps

Fig. 20 Graph of amp to RPM conversion.

not match well with the data. Figure 20 is a graph of the data points from the
Table 8.
A relation between current and RPM was required because the rotor would not
be visible once the electrode was attached.
When connecting the electrode to the assembly, styrofoam was used as an
insulative layer to separate the electrode from the base plate. Nylon screws
were used to hold the two components together. The final assembly of the
device can be seen in Fig. 21. Though the internal components of the machine
were exposed, the aerodynamic effects generated by the spinning rotor were
minimal, as detailed in the results section of this paper.
To measure the weight equivalent to change, a Mettler PM6100 scale was
used. Provided by the chemistry department, this scale had a resolution and
range of 0.01 g and 6100 g, respectively. The scale was grounded with a 28-
gauge wire in order to protect the equipment from static discharge. This wire

Fig. 21 Machine fully assembled.

 NULL FINDINGS OF YAMISHITA 285

Fig. 22 Wire used to ground the scale.

was oriented in such a way that it would not affect the weight equivalent of the
machine and is illustrated in Fig. 22.
A Wabash Instrument Corp. Winsco Model N-100V Van de Graf generator
was provided by the physics department and used to charge the machine. The
Van de Graf generator’s spherical electrode was attached to the machine’s elec-
trode via wire. Although Yamashita’s patent application depicted charging the
machine’s electrode by directly touching it with the Van de Graf generator’s elec-
trode, the sensitivity of the scale utilized precluded the implementation of that
procedure. Instead a wire was used to connect the generator’s electrode to the
machine’s. The procedures for the experiment are as follows.
With the power supply and the charge supply off, the weight equivalent of the
machine was measured. The wire for the motor was oriented in such a way that
they would not affect the weight equivalent of the machine. This was also true for
the wire used to charge the machine. These are illustrated in Figs. 23 and 24. The

Fig. 23 Clamp holding motor cord.

286 K. E. SIEGENTHALER AND T. J. LAWRENCE

Fig. 24 Clamp holding the charging wire.

mass measurement was made several times while shaking the wires connected to
the motor and power supply, and mass differences less than 0.1 g were observed.
Given the results of shaking the wires, they would need to be kept still during
experimentation.
The power supply was then turned on, and the rotor was accelerated rapidly.
From the acceleration test it was seen that the machine’s mass would fluctuate
by less than +0.1 g if maximum current were applied at rest. Further testing indi-
cated that a current increase of 0 to 0.74 amps in 60 seconds produced (relatively
slow acceleration) smaller fluctuations in mass. It was then decided that the rotor
would be allowed to run at determined amp levels for 10 seconds before proceed-
ing to further accelerate the rotor. Increasing the current by 0.05 amps and allowing
the rotor to spin at the amp level for 10 seconds, the mass readings were recorded.
To check whether the electrode held charge, the machine was connected to the
Van de Graf generator via wire, which can be seen in Fig. 25. The generator was
then turned on and allowed to charge the electrode for one minute. After one
minute the generator was disconnected from the electrode and turned off. The
device illustrated in Fig. 26 was then connected to the ground socket in a wall
outlet and brought within close proximity of the electrode. A spark was observed,
verifying that the electrode had been charged by the generator.
 NULL FINDINGS OF YAMISHITA 287

Fig. 25 Wire connecting the Van de Graf generator to the electrode.

Then, with the rotor at rest, the Van de Graf generator was turned on and
reconnected to the machine’s electrode. After one minute of charging the elec-
trode, the wire was disconnected from the machine’s electrode and the Van de
Graf generator was taken out of proximity and turned off.
The rotor was then accelerated to its top speed, taking readings of the scale at
intervals of 0.05 amps, remaining at each level for 10 s. These procedures were
then repeated with the positive terminal of the Van de Graf generator charging the
electrode. A third test was also performed with the generator attached and con-
tinually charging the machine at a low rate while the rotor accelerated.

B. Results and Discussion

When uncharged, the difference in mass between 0 RPM and 3390 RPM was
less than 0.02 g. This difference can be considered to be caused by an interaction

Fig. 26 Device used to check charge.

288 K. E. SIEGENTHALER AND T. J. LAWRENCE

Table 9 Uncharged operation

Amps RPM Mass (g) Change (g)

0 0 2585.01 0
0.21 1983 2585.02 0.01
0.24 3087.764 2585.01 0
0.29 3254.6 2585 0.01
0.34 3319.4 2585.01 0
0.39 3346 2585.02 0.01
0.44 3364.4 2585.01 0
0.49 3375.2 2585.01 0
0.54 3380.2 2585.01 0
0.59 3384.643 2585.00 0.01
0.64 3386.429 2585.01 0
0.69 3388.215 2585.02 0.01
0.74 3390 2585.01 0
0 0 2584.99 0.02

between the rotor rotation and the surrounding air as the machine was not fully
enclosed. Additionally, the slight vibrations that resulted from the machine’s
operation may have caused these fluctuations given the scale’s sensitivity.
Table 9 depicts the mass fluctuations when accelerating the uncharged electrode.
It is seen from the data that the machine’s mass stays fairly constant while
accelerating, indicating very smooth operation; the small fluctuations that do
occur may have been caused by slight interference from either air or small
vibrations from the machine itself.
When charging the electrode using the negative terminal of the Van de Graf
generator, the scale produced the mass changes illustrated in Table 10.

Table 10 Negatively charged operation

Amps RPM Mass Change

0 0 2584.99 0
0.21 1983 2584.99 0
0.24 3087.764 2584.99 0
0.29 3254.6 2584.98 0.01
0.34 3319.4 2584.98 0.01
0.39 3346 2584.99 0
0.44 3364.4 2584.97 0.02
0.49 3375.2 2584.97 0.02
0.54 3380.2 2584.97 0.02
0.59 3384.643 2584.99 0
0.64 3386.429 2584.99 0
0.69 3388.215 2585.01 0.02
0.74 3390 2584.99 0
0 0 2585.01 0.02
 NULL FINDINGS OF YAMISHITA 289

From Table 10 it is seen that no significant mass changes were registered

by the scale. Additionally, the very slight mass fluctuations that did occur
were not indicative of a weight equivalent loss pattern that the machine
should have been experiencing with negative charge. Though these results
are indicative that the theory behind Yamashita’s device does not hold true,
an inaccuracy of the machine’s dimensions may have been the cause of
such results.
A protruding screw, which attached the rotor to the drill chuck, caused a
gap of about 2 cm between the top of the rotor to the ceiling of the electrode.
This is not in accordance with the smaller gap evident from Yamashita’s draw-
ings. In addition, the mass of the machine was much greater than the mass of
either Berrettini’s [5] or Yamashita’s [2]. This was mainly due to the
additional mass incurred by integrating the fan into the entire assembly of
the machine.
When connected to the positive terminal, the machine was accidentally
nudged. The scale was allowed to settle and it settled on a new value of
2585.45 g. While connected to the positive terminal, the machine produced the
results presented in Table 11.
Figure 27 compares the percent of weight equivalent deviation during the
rotation experiments of an uncharged, negative, charged, and positive charged
cylinder of Experiment 2.
It is also seen in this case that the machine neither provided a significant
mass change or a general increase in mass difference. When continually
charged by the Van de Graf generator, the machine yielded similar results for
both the positive-charged and negative-charged case. To put this in perspective
for the application of this concept as a form of space travel, based upon the
results we obtained for the most precise mass displacement measurements con-
ducted in Experiment 2, it would take 830 years for a 500 kg spacecraft to
travel from Earth to Mars [6 9].

Table 11 Positively charged operation

Amps RPM Mass Change

0 0 2585.45 0
0.21 1983 2585.45 0
0.24 3087.764 2585.46 0.01
0.29 3254.6 2585.46 0.01
0.34 3319.4 2585.45 0
0.39 3346 2585.44 0.01
0.44 3364.4 2585.43 0.02
0.49 3375.2 2585.42 0.03
0.54 3380.2 2585.45 0
0.59 3384.643 2585.43 0.02
0.64 3386.429 2585.45 0
0.69 3388.215 2585.44 0.01
0.74 3390 2585.45 0
0 0 2584.43 0.02
290 K. E. SIEGENTHALER AND T. J. LAWRENCE

EXPERIMENT 2 RESULTS
0.001
PECRENT WEIGHT DEVITAION

0.0005

0
3050 3100 3150 3200 3250 3300 3350 3400 3450

-0.0005
UNCHARGED ROTATION
NEGATIVE CHARGE ROTATION
POSITIVE CHARGE ROTATION
-0.001

–0.0015
ROTOR SPEED (RPM)

Fig. 27 Comparison of the percent weight equivalent to deviation results for

Experiment 2.

IV. Conclusions
The experiments conducted at the United States Air Force Academy in no
way confirmed the existence of an electrogravitational force. Because of the
equipment and techniques used, the differences in weight equivalent actually
seen were too statistically insignificant to prove anything. The replica device,
however, did not show any evidence contrary to Yamashita’s claims. The first
claim, that a horizontal rotating body produces a vertical force, could not be dis-
proved. The spinning rotor and its effect on the replica device’s weight equivalent
did not disprove the first claim by virtue of the fact that the data needed to prove
or disprove this claim were statistically insignificant. Another one of Yamashita’s
claims, that reversing the polarity should reverse the direction of the force,
could not be disproved for the same reason. The device seemed to generally
decrease its weight equivalent when given a negative charge, and increase its
weight equivalent with a positive charge; again, however, this weight equivalent
change was not statistically conclusive. The final claim that could not be rebuked
is that the magnitude of the force generated increases with the speed of rotation of
the charged body. One can see from the tables that this did tend to happen,
although again the results were statistically insignificant. Yamashita’s fourth
claim, that a force could be produced in any direction, was not observed in
this experiment.
In order to be able to prove or disprove Yamashita’s claims, it would be absol-
utely necessary to know exactly what he did in his experiment. Unfortunately, the
ambiguity of his patent left a lot of room for guesswork. A number of assump-
tions were made in the absence of information from Yamashita’s patent and
these assumptions may be partly responsible for the fact that this experiment
did not produce any conclusive results.
 NULL FINDINGS OF YAMISHITA 291

These experiments neither denied nor confirmed the existence of an electro-

gravitational force. Conclusive proof or disproof would require better data than
there collected from these experiments.
However, we feel these experiments were successful because the finding that
these concepts were not significant and worthy of further research in the field of
breakthrough propulsion physics for NASA allowed resources to be allocated to
investigate other concepts.

Acknowledgments
This work would not be possible without all of the hard work and dedication of
the cadets. Cadets helped in the design, fabrication, testing, and analysis of all of
these experimental campaigns. We acknowledge John Bulmer, Christopher
Schlagheck, and Vinny Berrettini for their outstanding work. Without them,
this chapter would not have been possible.

References
[1] Rognerud, N., “Free Fall of Elementary Particles: On Moving Bodies and Their
Electromagnetic Forces” Technical Paper, Rognerud Research and Development,
Concord, CA, Physics Review D, 1994, submitted for publication.
[2] Yamashita, H., and Takayuki, Y., Machine for Acceleration in a Gravitational Field,
European Patent 0486243A2, 1991.
[3] Xylene: Material Safety Data Sheet, CHEM SUPPLY Pty Ltd, Gillman, South Africa,
Sept. 1999.
[4] Schlagheck, C., “Weight Equivalent to Loss Effects Associated with a Rotating
Charged Cylinder,” AIAA Student Paper Conference, Parks College, St Louis, Mis
souri, 17 19 April 2002.
[5] Berrettini, V., “The Coupling of Mass and Charge as a Propulsive Mechanism,” AIAA
Student Paper Conference, University of Minnesota, Minneapolis, Minnesota, 28 30
April 2004.
[6] Tajmar, M. and Jacinto de Matos, C., “Induction and Amplification of Non Newtonian
Gravitational Fields,” AIAA Paper 2001 3911, AIAA Joint Propulsion Conference,
Salt Lake City, Utah, 9 11 July 2001.
[7] Larson, W. J., and Wertz, J. R. (eds), Space Mission Analysis and Design, Microcosm
Press, El Segundo, CA, 1999.
[8] Humble, R. W., Henry, G. N., and Larson, W. J., Space Propulsion Analysis and
Design, McGraw Hill Company, New York, 1995.
[9] Angelo, J. A. Jr., and Buden, D., Space Nuclear Power, Orbit Book Company,
Malabar, FL, 1985.
 Chapter 8

Force Characterization of Asymmetrical Capacitor

Thrusters in Air

William M. Miller
Sandia National Laboratories, Albuquerque, New Mexico
Paul B. Miller†
East Mountain Charter High School, Sandia Park, New Mexico.
and
Timothy J. Drummond‡
Sandia National Laboratories, Albuquerque, New Mexico

I. Introduction
OR some time there have been claims of significant forces generated by
F voltage-excited asymmetric capacitors (colloquially known as “lifters”),
along with theories explaining them. In fact, reports concerning anomalous
force production by asymmetric capacitors have existed for more than 80
years. However, few of these were published in any readily obtainable form,
fewer still contained significant amounts of quantitative data, and many involved
extensive speculation as to the origin of this force. The authors believe that this
lack of data, combined with fanciful speculation, may have led most researchers
to ignore what might be an important phenomenon. Our purpose was to remedy
this situation by making careful measurements and letting the weight of the
evidence speak for itself as to the physics underlying the effect.
This chapter reports the results and conclusions of this experimental program.
While applying both uniform and sinusoidal voltages to asymmetric capacitors of

This material is a work of the U.S. Government and is not subject to copyright protection in the
United States.

Distinguished Member of Technical Staff, System Integration and Technology Department.
†
Student; previously at East Mountain Christian Academy, Tijeras, New Mexico.
‡
Principal Member of Technical Staff, Technology and Integration Department.

293
294 W. M. MILLER ET AL.

varying geometry, forces were measured to within +100 nN. In most cases,
voltage, current, and force were measured and absorbed power was calculated.
The force (direction and magnitude) was found to be polarity independent for
a uniform applied voltage. Sinusoidal excitation led to constant forces of similar
functional form (force as a function of voltage) of slightly lower magnitude.
Forces were observed well below the breakdown threshold in air. Frequency
dependencies were insignificant. A uniform magnetic field caused no quantitative
effect on the force. Geometrical variations in the asymmetry produced only small
variation in the measured force. These results are shown to be consistent with a
force produced by a corona wind originating at the capacitor electrode associated
with the higher electric field.
Early work on forces generated by high-voltage-excited asymmetric
capacitors was carried out from 1923 to 1926 by Dr. Paul Alfred Biefeld and
his student, Thomas Townsend Brown, both of Dennison University. Contempor-
ary publications [1] by Brown are sketchy, but refer to the effect he observed as
being related to “controlling gravitation” and “influencing the gravitational field”
with an “apparatus requiring electrical energy.” Many of the details involving this
can be found in a number of British and U.S. patents [2 6] dating from the late
1920s through the early 1960s. The British patent clearly refers to “controlling
gravitation” and “influencing the gravitational field” with an “apparatus
requiring electrical energy.” However, the later U.S. patents for different
devices utilizing an asymmetric electrical field make no such claims beyond
the production of mechanical forces. Brown did, however, claim in one patent
[6] the production of forces in a vacuum far from potentially interacting walls.
He was also issued one French patent on his devices [7].
Some work was done under the auspices of the U.S. government from the mid
1980s through 1990 to better understand this phenomenon. A report describing
the forces on an asymmetric capacitor held under vacuum in a chamber that
was large compared to the size of the device under test, did report the observation
of small forces [8]. Another report detailed possible relationships between elec-
tromagnetic fields and gravity [9].
More recent work consists primarily of that by amateur experimenters [10].§
Carefully controlled and published experiments continue to be rare. Bahder and
Fazi reported on theoretical and experimental work they performed at the Army
Research Laboratory [11]. Canning et al. reported on tests conducted under high
vacuum conditions, in a NASA report [12]; a more recent summary by Canning
appears in the next chapter of this book.

II. Summary of Theories

A number of theories have been put forth to explain the motive force of asym-
metric capacitors. Not all devices conform to the triangular wire and plate
arrangement widely described on the Internet. The common feature is that all
lifters are asymmetric capacitors. Brown patented both the wire and rectangular
plate geometry as well as the geometry of a sphere above, on the axis of, and

§
Naudin’s Web site is arguably the most thorough compilation.
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 295

normal to a circular plate. An assessment by the Office of Naval Research (ONR)

of Brown’s work in 1952 came to the conclusion that, where Brown had docu-
mented results, those results could be readily explained in the context of a
corona wind [13]. The ONR study dismissed Brown’s claim that results were
linked to lunar tides. More recently, studies have inconclusively tried to link
asymmetric capacitor phenomena to the Earth’s global electric or magnetic cir-
cuits [14,15]. Without invoking a “fifth force” in nature such interactions are
orders of magnitude too small to have a significant impact on lifter phenomena.
It has been hypothesized by T. Musha that a strong localized electric field gener-
ates a local gravitational field and that the mass associated with the electric field
energy is canceled by a negative mass associated with the induced gravitational
field. Musha asserts from his experiments that the effect of solar and lunar tides is
real. Musha’s results do not explain the observed directionality of the nominal
Biefeld Brown effect nor its independence of the polarity of the applied
field [16]. More credible assertions that the inertial mass of an object can be
manipulated electromagnetically have been put forth by Woodward and Brito
in separate approaches [17,18]. These approaches do not, however, claim to
explain the Biefeld Brown effect. An Air Force study postulated that by assum-
ing a five-dimensional continuum, an electrogravitic coupling could be derived to
explain the Biefeld Brown effect. This is a purely speculative approach that may
appear to predict the desired result. Such modeling is fraught with danger as any
such model will be severely limited by constraints deriving from both cosmology
and particle physics [19].

III. Overall Experiment Setup

In reviewing the literature of the previous 80 years, most reports described the
(uncalibrated) high voltage needed to lift a several-gram object against the force
of gravity. While this indicated the presence of an impressively large force, there
were few reports of accurate measurements of both the force and the voltage
required to generate it. In some cases, the force measurements were of a very
indirect nature, relating to the velocity of a spinning object against the retarding
forces of friction. We set out to remedy the continued lack of direct, detailed, and
precise measurements of this phenomenon with the hope of discerning among the
many theories that have been proposed as to its physical origin.
We built several asymmetric capacitors each consisting of a thin conduct-
ing wire (0.0042 in. uninsulated beryllium-copper wire; Alloy 25, Little Falls
Alloys, Inc. Paterson) separated from and parallel to a thin conducting plate
(0.0016 in. uninsulated aluminum foil; Reynolds 657). The particular shape
and size of the device depended on the particular experiment. Direct force
measurements (uniform and sinusoidal voltage excitation) were made on tra-
ditional triangular-shaped devices (which enhanced their stability), while geo-
metrical variations were made on linear devices to be described later.
Descriptions of the measurements made on each device are provided in Sec. IV
“Specific Setup” and the results of these measurements are provided in Sec. IV
“Results.”
Force measurements were made by suspending the various asymmetric
capacitors directly from the pan hook of a Mettler M3 microbalance on a thin
296 W. M. MILLER ET AL.

Fig. 1 Complete experimental setup. Note that the device under test has been
enhanced to clearly show the wires (black) and the supporting lines (white).

insulating polypropylene line (thickness of 0.0044 in.). This self-calibrating,

zeroing balance can be read to +1 mg (approximately +10 nN). Even though
the entire experimental setup (Fig. 1) was placed on a vibration isolation
mount supported by a poured concrete basement foundation, vibration and
other motions induced by operating the capacitors allowed accurate reading to
be made to no better than +10 mg (+100 nN). As the balance had a
maximum capacity of 3 g (with the balance pan in place) and somewhat more
than this with the pan removed (as it was for these experiments), the mass of
the capacitor that we used was limited. The force of the effect was measured
as the difference between the forces on the unexcited capacitor (the force due
to gravity on its mass) and that of the excited capacitor.
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 297

Exciting voltages were provided by high voltage uniform and sinusoidal power
supplies (described under the “Specific Setup” subsections). The voltages applied
to and the currents flowing into the capacitors were measured across a noninductive
resistive voltage divider and a noninductive current viewing resistor, respectively.
The resulting voltages were measured by a pair of calibrated zeroing digital
multimeters (HP3456A, Hewlett Packard Co.) to +1 mV. The relationship
between the voltage divider readings and the actual voltage applied was determined
by calibration using a high voltage probe (Fluke 80K-40). The current viewing
resistor was calibrated with the use of an impedance bridge (Model 1656, General
Radio). All measurements were performed at an altitude of 2304 m with an approxi-
mate air pressure of 76.5 kPa at a temperature between 48C and 168C.
Force, voltage, and current measurements were made as follows:
1) A measurement of the weight of the unexcited device was made with the
microbalance. Even though the device was in a cage shielded from external air
currents and on a vibration-damped table, some small variation in the measure-
ment always occurred. The maximum and minimum of this range was recorded.
These variations tended to be small when the force measured was small, and
larger at greater forces.
2) With all high voltage controls set to zero, measurements were also made of
the applied voltage and the unexcited current, providing a zero reference for later
comparison.
3) The high voltage excitation was then applied.
4) The device was allowed to stabilize and a reading of the maximum and
minimum force on the microbalance was made.
5) The net force due to the effect was taken as the difference between the
force on the excited device under test and that of the gravitational force on the
unexcited device.
6) Thirty readings each of the excitation voltage and the current flowing into
the device were made. The average and variance were then recorded.
7) This process was repeated from step 3 through step 6 for different exci-
tation voltages until a complete set of data was obtained. Typically, readings
were made starting at the lowest and working to the highest voltage.

IV. Results
A. DC Measurements
1. Specific Setup and Measurement Technique
The high voltage DC exciting voltage was provided by a Del Electronics Corp.
power supply capable of generating up to 30,000 volts at 10 mA (Fig. 2).
Therefore, for safety’s sake, the entire experiment was carried out inside a 12 in.
thick Plexiglas cage supported on top of this power supply that itself was sup-
ported on the vibration isolation table mentioned earlier. The device under test
was suspended from the Mettler M3 microbalance in such a way that when
measuring from any conducting extremity on the device to the insulating cage
walls there was 11 cm to the bottom, 25 cm to the top, 14 cm to either side,
10 cm to the front, and 12 cm to the back of the cage. As the device under test
298 W. M. MILLER ET AL.

Fig. 2 DC experiment setup. The resistor bridge was calibrated using the Fluke HV
probe using the pair of voltmeters. During data acquisition, the second voltmeter was
used to measure the voltage across the current viewing resistor.

had approximately 3 cm between the wire and the plate, the cage walls were at
least several times this characteristic distance away from the test device to mini-
mize wall interaction concerns.
This initial set of measurements was performed on a single triangular test
device built out of 2 mm  2.5 mm rectangular cross-section balsa wood, each
side of which was 200 mm long. The aluminum plate was 40 mm tall and the
wire was held 30 mm above this plate and in its plane by balsa wood supports.
To minimize corona discharge between the plate and the wire, the aluminum
was rolled over the upper balsa wood supporting stick (Fig. 3). A representative
picture of a later, but similar device is shown in Fig. 4. The device was supported
on three fine strands of polypropylene line attached to each of its three corners.
These lines were attached together above the center of the device and a single
strand of line then continued upward to the attachment on the balance. The
device was leveled horizontally so that the force generated up along the rotational
symmetry axis of the triangular lifter was directed vertically. Actual force,
voltage, and current measurements were carried out as described in Section III.

2. Results
Some initial qualitative experiments were performed with the symmetry axis
of the triangular device held horizontally. In this case the device generated a force
along the axis of the device directed from the plate toward the wire. Likewise,
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 299

Fig. 3 Original design of asymmetric capacitor.

when the device was suspended with its axis vertical, the force was again directed
along the axis from the plate to the wire. Quantitative measurements were made
of both the force developed by the device and the current flowing between
the wire and the plate as a function of the applied voltage. Figure 5 shows that
the force follows a power law dependence on voltage of approximately:

F ¼ k0 V 4 (1)

Fig. 4 Completed asymmetric capacitor.

300 W. M. MILLER ET AL.

Fig. 5 Measured force as a function of applied DC voltage. Note that the force is
identical for both polarities.

where F is the measured force, k0 is a proportionality constant, and V is the

applied voltage. It is exceptionally important to note that the force data for
both the positive and negative applied wire polarities were found to be identical
within the resolution of our measurements. We observed no polarity dependence
for any of the devices tested in this study. Another important point is that we were
able to make measurements at higher negative than positive voltages. This is con-
sistent with the fact that negative corona discharges are more stable than positive
corona discharges.
The current dependence of the force was also discovered to follow a power
law as shown in Fig. 6. This is clearly shown by plotting the force against the
current raised to the 2/3 power. This functional dependence is given as:

F ¼ k1 I (2=3) (2)
where again F is the measured force, k1 is a proportionality constant, and I is the
measured current. This relationship describes the data quite well and yields a
correlation coefficient between the data points and the power law of 0.998.
Again, both positive and negative polarities yield the same forces. There is no
polarity dependence observed. Extension of the best fit line shows an important
point; it passes through zero. This clearly implies that current must flow for a
force to be generated.
In order to understand the possible nature of any electrical conduction between
the wire and the plate, the applied voltage and the current delivered to the device
was plotted on a log-log graph. Figure 7 shows that the current follows a power
law with a slope of approximately six:

I ¼ k2 V 6 (3)
where k2 is a proportionality constant.
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 301

Fig. 6 Measured force as a function of the measured current raised to the 2/3
power.

Note that the functional dependence of the current-voltage curve was approxi-
mately independent of polarity and of small changes in the device geometry. The
“large 45-mm device” had a side length of 200 mm, a plate height of approxi-
mately 46 mm with a wire height of approximately 31 mm. The “large 40-mm
device” had a side length of 200 mm, a plate height of 39 mm and a wire
height of 30 mm, while the “small 45-mm device” had a side length of
135 mm, a plate height of 45 mm, and a wire height of 30 mm.
In order to better understand the role of current flow in the production of the
force, we took the “small” device and coated its wire with glyptol, a paintable
conformal insulator, and allowed it to dry. Such a coating would, we presumed,
suppress the formation of corona discharge in the immediate vicinity of the wire.
Thus, we might expect a decrease in the overall force developed.

Fig. 7 Plot of the current flowing into the device as a function of the applied voltage.
302 W. M. MILLER ET AL.

Fig. 8 Comparison of the force developed on identical devices, one with the wire
bare and the other with the wire covered in a thin layer of glyptol insulation. Both
measurements were taken with a positive polarity on the wire.

This effect can be seen in Fig. 8, which shows the measured force as a function
of the applied voltage for capacitors with bare and glyptol-coated wires. Both
measurements were made with the wire at positive polarity. Note that at the
highest voltage, the force-voltage curve of the insulated wire is approaching
that of the bare wire. We interpret this as an increase in the number of points
at which the glyptol has broken down along the length of the wire allowing a
relatively more uniform corona to form.

Fig. 9 Comparison of the force developed with bare and glyptol-insulated wires as a
function of the current to the 2/3 power. Note that again the extrapolated best linear
fit passes through zero for both wire types.
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 303

Fig. 10 Comparison of the current voltage relationship of bare versus glyptol-

insulated wires. The glyptol-insulated wire shows currents that are significantly
below those of the bare wire for intermediate voltages, but approaches the bare
wire results at both lower and higher voltages.

We expected that the force-current curve would be related to the nature of the
atmosphere only and that the force developed would be proportional to the
current that flowed, even if higher voltages were required to bring this about.
As can be seen in Fig. 9, this is not the case. The glyptol-insulated wire produced
a force much lower than that of the bare wire for a given measured current.
However, in both cases the extrapolated best linear fit passes through zero,
again confirming that current must flow for the force to be generated.
Finally, the current-voltage plot of Fig. 10 compares the various positive bare
wire results with that of the positive glyptol insulated wire. Unlike the bare wire
results that follow a single power law dependence, the insulated wire results show
a marked degree of curvature. At the lower voltages the bare and insulated wires
have similar values and slopes (although the insulated wire data are slightly
below that of the bare wire). At intermediate values of voltage, the insulated
wire data are significantly below that of the bare wire. Finally, at the highest
voltages used, the insulated wire data turn upward to approach the values for
the bare wire (much as was seen in the force-voltage curve of Fig. 8).

3. Discussion
The DC force data provide several indications of the nature of the force
generated by an asymmetric capacitor.
1) The magnitude and direction of the force are independent of the polarity of
the excitation voltage.
2) This force was also not related to the orientation of the device with respect
to the gravitational field of the Earth. It appeared to be solely related to geometry
of the asymmetric capacitor. Specifically, the force was always directed along the
symmetry axis of that capacitor in the direction from the plate to the wire.
3) Within the excitation regime studied (relatively high voltages and high
currents), the force exhibits a power law dependence with respect to voltage
304 W. M. MILLER ET AL.

(the force is proportional to the square of the voltage) and current (the force is
proportional to the current raised to the 2/3 power). We will later show that
these data were collected in what we will identify as the Fowler Nordheim
regime.
4) Current must flow for there to be a force. This was shown to be true
no matter the condition of the wire (bare or insulated). One can conclude,
therefore, that the effect, while requiring a voltage to be applied to generate
a current flow, is not so much an electric field phenomenon as a current
flow phenomenon. Even though these DC data do not show a force at very
low voltage, the presence of a current at lower voltages indicates that such a
force should be detected. Measurements at low voltages (described later) will
certainly show this to be true in the case of uniform excitation down to a few
hundred volts.
5) Devices with minor changes in the geometry behave in similar ways. This
will be explored further in the section on geometrical variations.

B. Sinusoidal Excitation Measurements

The fact that the force developed in the device was shown to be independent of
the polarity of the applied voltage led us to speculate that the device would also
develop a force under sinusoidal voltage excitation. In fact, Canning et al. [12]
and others have speculated that even with uniform voltages applied, the
leakage of current does not occur in a steady manner, but rather occurs as
pulses, implying that neither DC nor AC voltage excitation is truly uniform.

1. Specific Setup
We tested for the presence of electrical pulses under the application of uniform
negative DC voltage through the use of a 10 high voltage probe (Tektronix
PG122). This was placed on a PVC stand allowing the probe to be placed securely
and at a fixed distance between the wire and the plate of the device under test. The
output of the probe was fed into an oscilloscope (Tektronix R7844 dual beam
oscilloscope) and photographed with a digital camera. The uniform voltage
had to be raised above 8.3 kV before any effects were seen. (The results of this
will be described in detail.) However, the presence of these pulses (described
as Trichel pulses) encouraged us to proceed with the planned measurements of
the force on the device under sinusoidal high voltage excitation.
In order to make these measurements, an entirely different excitation setup
was needed. As we were interested in how any measured force changed with
voltage and with frequency, and as the voltage required might be in excess of
20 kV, we required a high voltage, wide frequency range amplifier. We were
unable to find any off-the-shelf equipment meeting our needs. Therefore, we
chose to adopt a technique utilizing a low voltage, high precision frequency
source (Hewlett Packard 3330B frequency synthesizer) feeding a high voltage
amplifier (Fluke 5205 high voltage wideband amplifier with a maximum
output of 1200 V at its 100 fixed gain), which in turn fed a custom designed
and built high voltage final transformer (modified Energy Systems pulse trans-
former). As this transformer had an original input to output ratio of 1:7, we
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 305

could not hope to achieve the final maximum voltage that we required, and so
rewrapped both its primary and secondary windings to match our input device
and our required output voltage. We made measurements on the original pulse
transformer core (14 turn primary bifilar winding, 102  2 turn balanced second-
ary winding) to help in the selection of the final design. Oscilloscope measure-
ments of the original transformer showed that frequencies from 100 Hz to
50 kHz were handled with no distortion or diminution of the transformer ratio.
For our final design we chose a primary winding of 93 turns of 28-gauge
copper wire per side (bifilar wound), and a secondary winding of 2809 turns
(per side of the balanced secondary). While the 93 turns of the primary
winding would not provide a sufficient load on the high voltage amplifier at
lower frequencies, selection of a greater number of turns would have made wrap-
ping the secondary impractical.
The four coils needed were wound onto PVC mandrels and then insulated with
high voltage Formvar varnish. Given our use of a bifilar primary and balanced
secondary, the designed output ratio was 1:15.1 and the measured ratio was
1:15.5. Insulation was provided by 4.5 gallons of Shell Aeroshell 30-weight
oil, dewatered and degassed at 1088C under a modest vacuum to achieve a
suitable breakdown voltage.
Figure 11 shows the calibration curve for this transformer measured as a func-
tion of frequency at low input voltage. One DMM (mentioned earlier) was used to
measure the ratio of input to output voltage. For frequencies below approximately
5 kHz, a flat voltage multiplication of 15.5 was measured. For use at higher
frequencies, we prepared a detailed calibration curve.
As the high voltage amplifier would not properly load into the transformer, we
were forced to place a 360 V resistor in series between the amplifier output and
the transformer input. While this decreased the input voltage to the transformer
and hence the output voltage (and was not ideal), it did allow data to be acquired.
The complete experimental setup for the AC measurements using the modified
transformer is given in Fig. 12. The Tektronix oscilloscope was used to measure

Fig. 11 Transformer voltage input-to-output ratio as a function of frequency.

306 W. M. MILLER ET AL.

Fig. 12 AC experiment setup. Oscilloscope was used to measure Trichel pulses.

the Trichel pulses while the device was under DC excitation only. We arranged to
apply either AC or DC excitation by reconnecting the DC signal source at the test
points to the right hand of the modified transformer.
Voltage measurements were made utilizing the root-mean-square (RMS)
feature of the digital multimeter (DMM). Current measurements proved to be
impossible due to ground loops. Due to a lack of time, the automation of the
measurements using the HP9845B was not completed; the equipment was run
manually. Each combination of voltage and frequency was measured 30 times
with the average reported in this chapter. In addition, we recorded the variance
of the observations. The error bars for these measurements were typically less
than 50 nN.

2. Results
In this case the wire was kept at a high negative voltage and the plate was
grounded. A single pulse is shown in Fig. 13. This clearly shows the double
peak characteristic of Trichel pulses. Figure 14 shows a series of oscilloscope pic-
tures obtained as the voltage between the wire and the plate was increased. In this
case the probe was kept approximately halfway between the wire and the plate of
the device under test. Additional data (not shown here) were taken with the probe
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 307

Fig. 13 Example of Trichel pulse.

in various other positions. Similar results were found, although the magnitude of
the DC offset of these pulses (at higher voltages) did change.
Note that when the voltage reached a certain threshold value (just below
8336 V for this geometry), infrequent negative-going pulses began to form. As
the voltage was increased, the pulse rate also increased. When the voltage
reached 8690 V the pulse rate was higher still and the DC voltage level increased.

Fig. 14 Onset of Trichel pulses. Voltage was raised to a) 8336 V, b) 8690 V,

c) 9323 V, d) 10,054 V, and e) 11,022 V, respectively.
308 W. M. MILLER ET AL.

Fig. 15 Number of Trichel pulses per millisecond is linearly correlated with the
measured current.

Above this voltage the DC component continued to increase, while the pulse rate
became uncountable as individual pulses overlapped.
If the current is solely related to the formation of the Trichel pulses, then the
pulse rate should be proportional to the measured current. In fact, a plot of these
two variables (Fig. 15) shows this to be the case. Had it increased more rapidly,
one might argue that generic leakage currents were superimposed onto the
current flow from the Trichel pulses. This is consistent with the physics of
Trichel pulses wherein the charge per pulse decreases with frequency due to
space charge effects.
These previous data indicate that even in the case of uniform excitation, a
regular time-periodic response can be obtained. These data caused us to pursue
direct AC measurements. The equipment (described in the previous subsection)
was then constructed to take these additional data.
Force and voltage measurements were made at 100 Hz, 200 Hz, 500 Hz,
1.0 kHz, 1.2 kHz, 2 kHz, 4 kHz, 5 kHz, 10 kHz, 20 kHz, 38 kHz, 50 kHz, and
100 kHz with the majority of these measurements taken at 1.2 kHz, 2 kHz,
4 kHz, and 38 kHz. Initially there appeared to be some form of resonance at
38 kHz, so data were first taken at this frequency. The thought was that if a
maximum of power would be absorbed by the device, a maximum force
should be observed. This did not prove to be true.
The data at 38 kHz (Fig. 16) show the same functional dependence that we
observed in the case of DC excitation. However, rather than being of large mag-
nitude as we had expected, the force was quite low. Compare the magnitude of
the force at 38 kHz with that measured for DC excitation and for AC excitation
at lower frequencies. The force is proportional to the square of the voltage. Note
that the scatter of the value of the force is large at the lower applied voltages, but
tightens as 1 kV is approached. We believe this to be due to the difficulty of
measuring forces in the region below 100 nN.
The force was measured at a series of voltages as the frequency was varied in
logarithmic steps from 100 Hz to 100 kHz. Only a relatively small change in
force was observed as a function of the frequency over this range.
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 309

Fig. 16 Force measured at 38 kHz excitation. Note that the force curve follows the
same power law as it did for DC excitation.

The data in Fig. 17 indicate that we can expect very little change in force as a
function of frequency. Our equipment was such that voltages above 1 kV could
only be obtained at frequencies in the 1 to 10 kHz range. Therefore, additional
data were obtained in this range and are presented below.
Figure 17 shows data taken at 1.2 kHz, 2 kHz, and 4 kHz. In this case the force
again behaves like a power law as before, following a voltage cubed relationship
at the higher voltages. However, at the lower voltages it follows a voltage squared
relationship.
These data clearly show several things: 1) the lifter is still capable of produ-
cing its force, even with an AC voltage applied to it, and 2) it appears that the
force of the lifter is the same no matter if the excitation by a sinusoidal (AC)
voltage or a DC voltage.

3. Discussion
These data elucidate several additional important facts about the force arising
from an asymmetric capacitor.
1) As expected from the polarity independence of force under DC excitation,
AC excitation also results in the production of significant forces. In fact, when
compared, the DC and AC forces are nearly identical (for the higher voltages
where such a comparison is possible).
2) The presence of Trichel pulses and their relationship to the force produced
under DC excitation shows that even for such an excitation, nonconstant effects
are taking place.
3) It is clear that measurable but small forces continue to be produced even at
extremely small voltages (down into the range of hundreds of volts). There is no
reason to believe that there is any limit in how low this can go, provided that
current is flowing. Practically, devices based on the phenomena of force produced
from asymmetric capacitors should concentrate on methods to increase current
flow at lower applied voltages.
310 W. M. MILLER ET AL.

Fig. 17 Force on device as a function of frequency. Note that the AC and DC forces
are nearly identical at the higher voltages.

C. Geometric Variations
Understanding the physics behind the force generation in an asymmetrical
capacitor should be facilitated by measurements of the force with different
asymmetries and hence in different geometrical configurations. However, such
variations are not easily accomplished using the standard triangular device that
had been used in the pervious experiments. A linear arrangement (to be described
below) was selected to allow easy variation of the wire-to-plate separation and
the plate depth, thus varying the magnitude of the asymmetry.

1. Specific Setup
The linear arrangement was formed from two vertical hollow, cylindrical
plastic supports, each 130-mm tall. In the center of these, another smaller
hollow plastic support, slightly longer than 60 mm, was placed such that this
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 311

support could rotate. On this was placed a 60-mm wide thin aluminum foil plate
such that this foil could be rolled around the horizontal support. The lower edge
of the plate was supported by a stiff copper wire of 65 mm in length, which was
centered and held rigid by slits in the edge of the vertical supports. A thin 60-mm
long wire was supported above the horizontal support via a series of holes placed
in the vertical supports and spaced every 5 mm above the horizontal pieces,
beginning at a 10 mm separation from the top of the aluminum foil plate.
These arrangements allowed the wire to be adjusted to any of 10 positions
above the plate (10 mm through 55 mm). Rigidity in the structure was ensured
by two thin plastic lines at the top and bottom of the arrangement that kept
tension on the vertical supports. The entire arrangement was hung from a triangu-
lar plastic line support and then from the center of the Mettler M3 balance, much
like the triangular device. The plate height could be adjusted continuously from
10 mm through 55 mm. Thus the depth of the plate and the height of the wire
could be varied without changing the underlying shape of the device under
test. For practical reasons, extra polyethylene weights were symmetrically
hung on the bottom of the device to provide a downward force on which the
device could exert force. (Recall that the balance can measure weights up to
3 g.) Weighting the device so that the total device weighed nearly this amount
guaranteed the maximum total measurement range. This also served to stabilize
the arrangement.
The data were taken using positive DC excitation on the wire via the setup
previously described (Fig. 2). The maximum voltage that could be used was
related to the minimum separation between the wire and the plate. For
example, with a 10-mm separation, a maximum of 13 kV could be applied,
while at greater than 20 mm, the full 30 kV capability of the high voltage DC
power supply could be used. The measurement procedure was identical to that
described in the DC measurements section (Section IV.A). However, as this
linear design was somewhat unstable against horizontal motion, higher voltages
tended to move the device back and forth more readily, thus leading to greater
variations in the vertical forces measured.

2. Results
As shown in Fig. 18, the force developed on the device with the plate depth set
to 55 mm varies with the wire height and voltage. As is expected from the pre-
vious DC force measurements, the force increases with applied voltage. These
data also show that for a given voltage, the force generally decreases as
the wire height increases. At very small separations between the wire and the
plate, with the highest voltages applied, the force decreased. There are a
number of modest maxima shown. For 10 to 18 kV, the maximum force is
generated at the minimum wire height measured, with a second, lower, local
maxima between 45 mm and 50 mm. At higher voltages, these second maxima
move toward lower wire heights, until at 30 kV, it is below 20 mm. At such
high voltages data could not be taken at the smallest wire heights as breakdown
would occur.
Most data taken at 15-mm wire height had to be discarded as these data
were taken with a poor connection to the aluminum plate. Additional data
312 W. M. MILLER ET AL.

Fig. 18 Force developed on the linear device as a function of wire height and
voltage. Note that there are gentle local maxima.

taken at 6 kV were also not shown, as they were not visible on a graph of this
scale. Nevertheless they behaved in a way similar to the other data between
10 and 18 kV. There were no unexpected results discovered, save the
gentle force maxima. While these may be useful in an engineering design
utilizing this effect, they were not of further use in understanding the physics
of the phenomena.
Likewise, we studied the effect of increasing the plate depth while using a
fixed wire height. The wire height was set at 25 mm to allow a full range of
voltages to be used. Our thought was that the greater plate depth might increase
the overall asymmetry of the system, thereby increasing the force. This
experiment was intended to answer a fundamental question about the force,
specifically, if increasing asymmetry was indeed related to the force production.
Figure 19 clearly shows that such an effect was not observed. Even when the data
were looked at very closely, there appeared to be no correlation between plate
depth and force. Plate depths from 17 mm to 55 mm were studied. These were
selected to be of sufficient range (greater than a factor of three) and an appropriate
magnitude (slightly less than to much greater than the characteristic wire height)
that some effect, if it were present, should have been manifested. Please note that
while the lowest force was recorded with the 17-mm plate depth, the second
lowest was found with the 55-mm plate depth. At best there might be a very
slight maxima at a plate depth of 45 mm; however, given the scatter in the
data, this experiment appears to have produced a null result.
As was the case in previous DC measurements, the force was found to vary as
a power law of both the voltage and the current. Again force varied as the voltage
cubed (Fig. 19) and as the current to the 2/3 power (Fig. 20). Furthermore, this
did not change appreciably as the plate depth was varied.
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 313

Fig. 19 Force as a function of voltage for a variety of plate depths (in mm). The wire
is fixed 25 mm from the plate. The force increases as the cube of the applied voltage.
Note that the variation in plate depth over the range studied does not result in
significant changes in the force developed.

The force power law as a function of voltage did appear to change as the wire
height was varied, as seen in Fig. 21. While the overall data at higher voltages
appear to follow the voltage cubed relationship described earlier, the very smal-
lest wire separations may have a slightly steeper slope. This also appeared to
change with the wire separation (being steeper with decreasing wire separations).

Fig. 20 Force as a function of current for a variety of plate depths (in mm). The
wire is fixed 25 mm from the plate The force increases as the current raised to
the 2/3 power.
314 W. M. MILLER ET AL.

Fig. 21 Force measured as a function of voltage for various wire separations (in
mm) and with a fixed plate depth of 55 mm. Note that the data do not all follow a
single power law dependence. A line showing a cubic dependence is indicated. For
low voltages the force is proportional to the voltage and at higher voltages it is
proportional to the cube of the voltage.

At the lowest voltages, the force is approximately proportional to the voltage,

with not all of the data behaving in this way.
The relationship of the force with current again behaved approximately as the
2/3 power as previously shown in Fig. 20. There might be very slight variations
with wire height (slightly higher slope) but this is difficult to confirm. Note the
wide scatter in the data.
One of the advantages of measuring the force, voltage, and current relation-
ships with the linear asymmetric capacitor was the complete absence of
corners found in the triangular device. This simpler geometry also allowed us
to investigate the different current regimes accessed as the voltage was increased.
As we saw in the earlier section on DC behavior, it is the current flow that is most
well-correlated with the production of the force. Thus, understanding the origin
of the current is most likely to help in understanding the origin of the force,
especially as the geometry was changed. The current and voltage data, collected
as described above, were replotted with 1000 divided by the voltage plotted on
the y-axis and with the natural logarithm of the ratio of the current to the
voltage squared plotted on the x-axis. Such a plot should be linear if Fowler
Nordheim currents are flowing. Fowler Nordheim tunneling, or field emission,
is a form of quantum tunneling in which current passes through an insulating
barrier (in this case, air) in the presence of a high electric field.
In this experiment, only data taken at the largest combined wire separation
(55 mm) and plate depth (55 mm) were plotted so that the form of graph might
clearly be seen. This representation of the data (Fig. 22) allowed three regimes
to be identified. At the lowest voltages 1000/V had its highest value. This
regime is seen to the far right-hand side of the graph. These data exhibit a straight
line relationship from the lowest voltages to approximately 6000 V (1000/V 
0.17 1/V). This first region we identify as the “non-self-sustaining current
regime.” Note that 6000 V is approximately the voltage at which we first saw
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 315

Fig. 22 Fowler Nordheim plot of linear device (55-mm wire separation and 55-mm
plate height). Note the three distinct conduction regimes highlighted by the three
lines.

the onset of the largest DC forces and where Trichel pulses were first observed,
although smaller forces within this regime could and were measured in both
the AC excitation and this experiment. The non-self-sustaining regime is domi-
nated by leakage currents that are not energetic enough to result in current
multiplication.
At somewhat higher voltages (from approximately 6000 V to 14,000 V with
corresponding values of 1000/V of from 0.17 to 0.07 1/V), another linear
region was found. We termed this the “corona discharge regime.” This is best
described as resulting from ion multiplicative effects. Even though a slight posi-
tive slope can be seen for this regime, we believe that the actual slope here is
approximately zero due to the additive nature of all of the current and the inability
to distinguish one particular current type within each regime. The greatest effect
tends to dominate these regimes. In this case the third regime tends to add to and
cause this slight slope.
The third and final regime at the highest voltages (above 14,000 V from
1000/V less than 0.07 1/V) we have identified as the Fowler Nordheim
regime. These very large currents dominate all others. This region is composed
of both Fowler Nordheim current and corona current although, being much
larger, the Fowler Nordheim current dominates. Interpretation of the data relies
heavily on these observations and will be discussed at length in Section V.
The three regimes found may help to explain the variation in the power law
dependence seen in Fig. 21. Note that the behavior changed at approximately
the voltages where the Fowler Nordheim and the corona discharge regimes meet.
All data for the varying wire height (with the fixed plate depth of 55 mm) are
shown in Fig. 23. The same behavior exhibited with the 55-mm 55-mm data can
be seen at all other wire heights, although less clearly as there is much overlap.
Finally, several sets of data are replotted in Fig. 24 to show a fact of potential
engineering significance, that of the force-to-power relationship for this linear
316 W. M. MILLER ET AL.

Fig. 23 Fowler Nordheim plot for 55-mm plate linear device as a function of wire
height. Note how the curves move upward with decreasing wire height.

system. Note that the force increases as approximately the square root of the
power. While we cannot claim to have an optimized system it is interesting to
note (empirically) that
p
F¼C P (4)

Again we define the force as F (in mN),pthe power as P (in mW) and the
constant as C, which is 1 (in units of mN= mW). We have every reason to
expect that the force will be proportional to the length of the asymmetric capaci-
tor. Given the length of this device (0.06 m) and for this geometry (wire height of
25 mm) and plate depths from 17 mm through 55 mm, the force per unit length
will follow a formula similar to Eq. (4):
p
f ¼c P (5)

Fig. 24 Force as a function of power for a variety of plate depths (in mm). The wire is
fixed 25 mm from the plate. The force increases as the square root of the power.
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 317

Where f is the force per unit length (in mN/m), again P is the applied power
in mW and p c is the specific force per unit length per unit power of
16:7 mN=(m  mW).

3. Discussion
These studies of the effect of geometry on the force developed in asymmetric
capacitors were surprising in that no large changes were noted with relatively
large changes in wire spacing and plate height, and hence in asymmetry.
Minor maxima were found as a function of wire height that became more
pronounced at the higher voltages. Again, interpretation of these results should
be made in terms of the flow of the current (given that no force was measured
unless current was observed to have flowed, as in Fig. 9) and not just on the
basis of the voltage applied. However, as Fig. 22 has shown, the values of the
voltage applied do help delineate the various regimes of current production.
At very small wire-plate separations, the total force generated was always
modest. This was almost certainly due to the current leakage between the wire
and the plate and the very limited voltages that could be applied without
leading to arcing of the system. When such breakdowns did occur, they did
not damage the device under test.
From a purely practical point of view, reasonably large forces can be
generated by the simple plate-and-wire arrangement. Clearly, the more efficient
generation of large currents with more modest voltages would raise the efficiency
of this system.

D. Magnetic Field Measurements

There has been speculation that the force generated by an asymmetric
capacitor is related to its interaction with the Earth’s magnetic field, which has
a magnitude on the order of 5  10 – 5 T. Both the magnitude and direction of
this field vary depending on the location on the Earth. In our location, just to
the east of Albuquerque, New Mexico, at approximately 358000 5000 north latitude,
1068180 5000 west longitude, and an altitude of approximately 2304 m above
mean sea level, the magnetic field had an approximate total magnitude of
5.0676  10 – 5 T and a direction of approximately 98570 east of north and
a dip from the horizontal of 628240 down into the Earth on the date of the
calculation (30 Nov. 2003) [20].

1. Specific Setup
Previously mentioned measurements of the lifter’s force were always made in
the Earth’s magnetic field. In order to discern if this (or any) magnetic field gave
rise to the observed force, a much larger artificial magnetic field was created and
the measurements made in it. To create a uniform magnetic field we designed,
built, and operated the experiment in a Helmholtz coil, which has a property of
producing a large, relatively uniform magnetic field of known magnitude
and direction near its center (Fig. 25). Each of the 2 coils had four windings
8-layers deep for a total of 32 turns of 6-gauge copper wire. The diameter of
the inner coil was 75.6 cm and that of the outer coil was 80.0 cm. We separated
318 W. M. MILLER ET AL.

Fig. 25 Setup to measure magnetic field effects.

the 2 coils by 38.1 cm. The coils were energized with a Sorensen DCR40-60A
power supply that could produce 40 V at 60 A, although they only needed
about 13 V and 60 A to produce a magnetic field approximately 100 times that
of the Earth (specifically 5.0  10 – 3 T). A current shunt was used to measure
the current flowing through the Helmholtz coil and hence the magnetic field pro-
duced. The calibration constant was calculated to be 0.2063 T/V.
It was possible to arrange the coils to produce both a horizontal and a
vertical magnetic field. Specifically, the horizontal component was directed
approximately upward, and the vertical component was directed downward.
However, a failure of the power supply precluded our taking data in the
horizontal direction.
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 319

2. Results
Force measurements were first made with an unpowered triangular
asymmetric capacitor (as described in Sec. IV.A on DC measurements) with
the Helmholtz coils de-energized. With the device unpowered and the
Helmholtz coils energized, no effect was observed within the experimental
resolution of +0.00025 mN. This indicates that the artificial magnetic field did
not affect the force measurement of the Mettler microbalance to within these
limits. Measurements were then made with the device powered to produce a
force, but without the Helmholtz coils energized. This provided a background
measurement against which to compare the force once the coils were energized.
Finally a measurement was made with the device powered and the coils ener-
gized. This force measurement was then compared to the one made with no
applied external magnetic field.
Table 1 summarizes the results with the external magnetic field pointing
vertically downward. Two experiments were performed, one with an insulated
wire (coated with glyptol to decrease the formation of any leakage or corona cur-
rents) and one left bare. The first number under the column titled “Net Force
Change” for each experiment represents the downward force on the unpowered
device. The other values in this column represent the net upward force that
was measured both with and without the external magnetic field. The final
column in the table (Range of Effect) represents the net effect when normalized
to the field that was applied compared to the Earth’s magnetic field. Note that
these are very small numbers. If there was a significant magnetic field effect
one would have expected the net force change to have increased proportionally
to the applied magnetic field.

3. Discussion
Note that in both cases, the range of the measured effect was very small,
equivalent to the measurement errors experienced. The applied magnetic fields
(5.034  10 – 3 T and 5.044  10 – 3 T) were nearly 113 times that of the
Earth’s vertical magnetic field component of 4.4577  10 – 5 T. If the majority
of the force were to arise from interactions with the Earth’s magnetic field,
then one would expect that a force of 2.03 mN or 7.07 mN would have increased
to 100 times that in the artificial field provided. They did not. We conclude that
within experimental error, magnetic fields have little or nothing to do with the
generation of the forces observed.

V. Discussion of Data as it Relates to Theories

The objective of a theory of asymmetric capacitors (or lifters) is to elucidate
the mechanism by which a motional force is generated by the application of a
sufficiently high voltage. In all lifter experiments to date for which precision
measurements of current were made, it has been found that there exists a mono-
tonic dependence of force upon current flowing through the lifter. This is true for
both static and sinusoidally time-varying applied potentials. The language here is
carefully chosen to avoid the confusion inherent in describing potentials as DC or
AC, terms which refer explicitly to the time variation of current. As discussed
 320

Table 1 Changes in measured force with magnetic fielda

Applied external
Measurement magnetic field (T) Applied voltage (V) Net force change (mN) Range of effect (%)

No B, no E 21.15515
No B with E 14,121 2.03218
B and E 0.005034 14,204 2.03218
Glyptol B and E maximum 14,204 2.02238 0.0043
B and E minimum 14,204 2.04199 20.0043
No B, no E 19.23957
No B with E 14,149 7.06790
B and E 0.005044 14,251 7.26893
W. M. MILLER ET AL.

Bare B and E maximum 15,915 7.22971 20.0197

B and E minimum 15,915 7.30816 20.0293
a
Measurements were made with both insulated and bare wires
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 321

below, the lifter will be considered to be a leaky capacitor. There are two possible
current paths: surface leakage over the dielectric supports separating the wire
from the plate, and free space currents consisting of ions and electrons. Herein
we will assume that only the free space currents generate a force. Under this
assumption we also observe that, even under static applied potentials, the
current is never truly a DC current as would be characteristic of a linear resistive
circuit (ignoring thermal noise). The current is time varying at high frequencies,
driven by inhomogeneity in the lifter components and the complexities of free
space transport of charged particles [21]. At best, the current may be characterized
by a root-mean-square (rms) average current. Even in the case of a sinusoidally
varying applied potential, the current should not be assumed to be a simple sinu-
soidal response. It is also important to emphasize that force must be dependent on
a current flow. Under a static applied potential, with no current flow, no work is
done by the system that could be transduced into a force capable of moving the
lifter a finite distance. Whatever mechanism is at work, it is assumed to obey the
symmetry of conservation of energy.
The classical application, which most closely relates to the phenomena
observed in lifters, is found in the electrical description of electrostatic precipi-
tation [22]. In a lifter the common configuration is that of a wire parallel to the
edge of a rectangular plate and occupying the same plane. A common geometry
used in the study of precipitator phenomena is that of a wire in the plane of a per-
pendicular bisector of a plate (foil) along its major axis. The wire is held parallel
to the plane of the plate. A precipitator is designed to have a single high electric
field electrode. This is accomplished by designing the large electrode to be free of
sharp edges. In discussions of lifter phenomena, the wire is often assumed to be a
unique high electric field anode although by nature of the design, often compli-
cated by crude assembly methods, this assumption is of questionable validity.
Precipitators are designed to operate with a self-sustaining corona discharge
plasma localized in the near neighborhood of the wire electrode. The polarity
of a corona discharge is always referenced to the polarity of the high field elec-
trode, in this case the wire. In air, both positive and negative corona discharges
are possible. In a positive corona discharge a current of positive ions is ejected
from the plasma, to be collected by the plate, while both electrons and negative
ions are attracted to the wire where they are collected or neutralized [23]. The
situation for a negative corona discharge is more complex [24]. Both electrons
and negative ions are ejected from the plasma surrounding the wire. At low cur-
rents electrons leaving the plasma form more negative ions by attachment to elec-
tronegative molecules, oxygen in particular. This forms a negative ion sheath
around the plasma sheath that, at higher currents, causes instability in the
plasma causing it to periodically collapse. The resulting pulsed current mode is
described as a Trichel pulse mode. The frequency of the Trichel pulses may
increase with current from a few Hz to hundreds of MHz. While Trichel pulses
have been implicated in the force generation mechanism, they occur only for
negative corona discharges, although comparable lifter force has been observed
under conditions of both positive and negative discharge.
A corona discharge, either positive or negative, is sustained by an avalanche
multiplication process in which electrons are accelerated by the high electric
fields near the wire to produce ions via impact and frequently generating more
322 W. M. MILLER ET AL.

electrons to participate in the ionization process. The avalanche process is self-

sustaining only above a certain threshold field. The electron source seeding the
avalanche derives from natural ionization events driven by UV light or other
background radiation. Below the self-sustaining threshold the seed source is con-
stant, although some current amplification does occur from non self-sustaining
avalanches. The authors are not aware of any critical studies of the current
characteristics in this subcorona regime. In the corona regime the current is
quadratic in voltage according to the law

I ¼ AV(V  Vo ) (6)

where I is the current, V is the applied voltage, and A and Vo are constants depen-
dent of the geometry of the system. Vo is the corona onset voltage. This equation
is known as the Townsend equation [25] and relates directly to Paschen’s law and
its critical parameter pd (p being the pressure and d the gap distance). For pd
values less than about 1000 torr-cm the Townsend equation holds true. The
current is carried entirely by ions. At sufficiently high voltages the corona is typi-
cally observed to collapse into a spark discharge in which a direct electron current
flows between electrodes limited only by the power supply. This is generally a
destructive event. In the early literature on precipitators, an intermediate
regime was often reported in which currents were observed to follow a power
law, I / BV n, with 4 , n , 6. This behavior was not interpreted at the time.
In the present work, it is found that this is a Fowler Nordheim electron emission
current flowing parallel with a reduced ion current. In lifters, the geometry allows
for substantial Fowler Nordheim currents in conjunction with either positive or
negative discharges. In the case of a positive corona, the electrons must be
sourced from the base of the lifter and not the wire. Because the net force is essen-
tially the same for positive and negative corona discharges for which a substantial
Fowler Nordheim current flows, it must be substantially an electron current
thereby contributing negligibly to the force.
Fowler Nordheim emission is a field-assisted tunneling current that is quali-
tatively described by

j ¼ A( y)(E2 =F) exp [B( y)F3=2 =E ] (7)

where j is the current density, E is the electric field, F is the work function of the
electron emitter,p and A( y) and B( y) are slowly varying functions of
y ¼ 3:79  10 5 E=F. Because F is so strongly dependent on the geometry,
smoothness, and cleanliness of the emitter, it is found to have an effective
value typically 10 to 100 times greater than its nominal value. With the
approximation that at the emitter surface the electric field is proportional to the
applied voltage and the substitution I j  (effective emission area), A( y)/F
and B( y)F 3/2 are subsumed, with these proportionality constants, into
unknown constants A  and B  . Equation (7) is then rearranged as

‘n(I=V 2 ) ¼ B=V þ ‘n(A ) (8)

 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 323

which yields linear graphs of Fowler Nordheim I V data. On the surface this
seems like a large number of approximations. However, as in a parallel diode
array, the bulk of the current flows through surfaces having comparable effective
F and surface fields. Locally the conditions required for Eq. (6) are satisfied and
dominate the total current. By emphasizing that a Fowler Nordheim current
emitted from an extended source is spatially inhomogeneous it is possible to
understand that a Fowler Nordheim emission current can coexist with a
corona ion current as spatially distinct phenomena.
The assumption that lifters “fly” by generation of a unipolar ion current at the
wire electrode is realistic. That the force is independent of the polarity of the
applied bias and massive ions may account for virtually all of the current.
Lifter forces resulting from a constant ion current have been analyzed by
Bahder and Fazi under two assumptions: ballistic transport of ions from wire
to skirt (ionic wind) and nonballistic transport were the ions assume a finite mobi-
lity determined by collisions with neutral air molecules (ionic drift) [11]. In many
references Bahder and Fazi’s “ionic drift” is referred to as ionic wind, ion-neutral
wind, or corona wind. The ion-neutral wind explicitly calls out the assumed trans-
fer of momentum from ions to neutral air molecules. Bahder and Fazi found that
ballistic transport resulted in forces orders of magnitude smaller than those
observed in lifters. The magnitude of the forces estimated to result from a
corona wind (ionic drift) matched nicely with observed lift forces. The limitation
of this analysis, however, was that they did not recognize that the lifter was oper-
ating in a corona regime and could not identify a source of ions sufficient to drive
the lifter. In contrast, Christenson and Moller, deliberately investigating ionic
drift propulsions in the 1960s, designed a circular lifter-like device that produced
air velocities of up to 7 ft/s (2.1 m/s) with the device fixed on a stand [26]. The
only significant difference from modern circular lifter designs is that a dense
array of point emitters, directed toward the skirt of the device, was used
in place of a wire. The I V response of their device was verified to obey the
Townsend equation. The air velocity was proportional to the square root of the
current flowing through the device.
Published results for corona plasmas generated by sinusoidal high voltage
excitation are far less common than for constant excitation. They are of interest
primarily as a loss mechanism pertinent to high voltage power transmission
lines [27]. The salient features are that a full cycle of applied potential results
in a positive corona for V . Vo, a nonplasma interphase for Vo . V . Vo, a
negative corona for V , Vo, followed by a final nonplasma interphase. It is
important to note that for either positive or negative coronas, there is a plasma
sheath around the wire containing both positive and negative ions. Some of the
extant literature seem to imply that space charge produced by a sinusoidally
excited plasma simply oscillates in space outside the plasma region in the
nonplasma interphase in favor of a net current [28,29]. However, the field
strength decreases monotonically with distance from the wire. There should be
a net current of alternating positive and negative ions. Ions leaving the wire
during their generative corona and drifting outward will see a weaker
return field as the potential swings toward the opposite polarity. Evidence for
such a net current is evident in the comparison of the AC and DC lifter data
already discussed.
324 W. M. MILLER ET AL.

A second-order effect that has been considered in precipitators derives from

the induced polarization of the ambient gasses. Such forces are variously referred
to as electrohydrodynamic (EHD), dielectrophoretic, thermodynamic, or Kelvin
forces. Any electric field will induce a dipole in a polarizable atom or molecule.
In a spatially uniform electric field dipoles will simply align with the field. It is
generally assumed that air subject to a high electric field constitutes a dilute
incompressible fluid of linearly polarizable dipoles. In a spatially nonuniform
electric field, dipoles will migrate to the regions of the highest electric fields
driven by the gradient in the electric field intensity, irrespective of the polarity
of the field. The magnitude of the force is proportional to the polarizability
(corresponding macroscopically to the permittivity of the working fluid) of the
dipolar molecules and r (E  E), the gradient of the dot product of the electric
field [30]. Common atmospheric gases have low polarizabilities. Ions produced
in the corona plasma and plasma reaction products such as ozone (O3) can
have polarizabilities orders of magnitude larger than neutral N2 or O2. In wire-
plate test configurations Kelvin forces are found to be responsible for circulating
flows with momentum transfer to the air approximately an order of magnitude
smaller than that transferred by the Coulomb force [31]. Because r (E  E ) is
independent of the polarity of the field, we have considered that these Kelvin
forces may account for the motional force acting on a lifter. The counter argument
has been that the Kelvin force ultimately results in a static situation where the
increasing gas pressure in the high field region balances the Kelvin force.
Coupled with an ionic flow, a recirculating flow is obtained, which reinforces
the corona wind in the plate perpendicular bisector plane containing the wire.
All of the atmospheric pressure results accumulated in this work, as well as
those reported in the open literature can be interpreted in the context of an ion
drift thrust mechanism using Eqs. (6) and (7). Claims for esoteric thrust mechan-
isms will not be considered here and it is asserted that all published atmospheric
results can be interpreted solely in the context of a corona ion drift wind as
asserted by Tajmar [32]. Tajmar was also able to exclude the existence of any
significant exotic force by enclosing a lifter in a sealed grounded enclosure
and demonstrating that the lifter was not able to impose a net force on the sur-
rounding air in the enclosure or on the enclosure itself. A unique contribution
of the present work has been to demonstrate that there are two distinct plasma
regimes that can provide significant lift: the pure corona regime and the
regime where an ion current flows in parallel with a much larger electron
current sourced by Fowler Nordheim emission. These regimes with be
distinguished as the corona wind and corona Fowler Nordheim regimes.
If we consider a pure corona wind as providing the motive force for lifters, one
can make the following positive assertions as being true of both lifters and expli-
cit corona wind studies. A corona wind will always flow from the high field
(plasma source) to the low field region of the device. In traditional lifters, includ-
ing ours, the high field region is typically a wire. For a corona wind the current
will be proportional to V 2. Negative coronas will be stable to higher voltages than
positive coronas. Only negative coronas generate Trichel pulses and substantially
more ozone than positive coronas. Wire vibrations may occur that are driven by
simple electromechanical coupling of the wire and the plate in the presence of
a plasma sheath [33]. No new physics are required to explain these phenomena.
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 325

A new observation for lifters is that currents in excess of the corona current are
apparently due to Fowler Nordheim emission. These currents appear on a
log-log plot as being proportional to V n, 4 , n , 6. Over a larger range of
validity they are described by the Fowler Nordheim Eq. (7). The Fowler
Nordheim currents are assumed to be pure electron currents that do not contribute
significantly to the total force. They do, however, reduce the rate of ion pro-
duction comparably in both positive and negative discharges. In the triangular
lifters studied here the propensity for Fowler Nordheim emission was strong
enough to dominate at all voltages. The onset of Fowler Nordheim current
flow apparently seeded the corona and no pure corona wind regime was observed.
In the linear geometry lifters the upper edge of the foil had a much larger
radius and a corona wind regime was easily observed prior to the onset of
Fowler Nordheim emission.

VI. Conclusions
Taken together, the experimental data on asymmetric capacitors data exhibit
the following:
1) The magnitude of the force is independent of the polarity of the applied
electrical excitation (AC and DC data).
2) The direction of the force is independent of the polarity of the applied elec-
trical excitation (AC and DC data).
3) The force can be generated both in line with and perpendicular to the
Earth’s gravitational field (DC data).
4) Current must flow for the force to be generated (DC data).
5) Electrical insulation of the wire leads to both a decrease in the current flow
and a decrease in the force generated (glyptol data).
6) The magnitude of the force is unaffected by magnetic fields up to 5  10 – 3
T (magnetic field data).
7) Even with a constant applied (DC) excitation, non-constant current flow
can be observed (DC data).
8) Forces of nearly identical magnitude were observed for both DC and AC
excitation.
9) Forces have been measured with small excitation voltages (DC data) and
very small excitation voltages (down to 100s of volts) (AC data).
10) Minor changes in the geometry of the asymmetric capacitors lead to
minor changes in the forces generated (geometry data).
11) Large changes in geometry lead to only modest changes in the forces
generated (geometry data).
12) All parameters that correlate with force production obey power laws
(voltage, current, electrical power) (DC, AC, geometry data).
13) The variation of the force with voltage follows a fourth power depen-
dence for DC excitation.
14) The variation of the force with voltage follows a third power dependence at
high voltages and a second power dependence for low voltages for AC excitation.
15) The force generated is proportional to the square root of the power
delivered to the device.
326 W. M. MILLER ET AL.

All of these data, when taken together, can be explained by the ion drift theory
as presented earlier. The complete description of the force is somewhat more
complicated and requires the consideration of second-order effects including
that of the dielectrophoretic (Kelvin) effect.

Acknowledgments
Sandia is a multiprogram laboratory operated by Sandia Corporation, a
Lockheed Martin Company, for the United States Department of Energy’s
National Nuclear Security Administration under Contract DE-AC04-94AL85000.

References
[1] Brown, T. T., “How I Control Gravitation,” Science and Invention, Aug. 1929.
[2] Brown, T. T., “A Method of and an Apparatus or Machine for Producing Force or
Motion,” British Patent 300,311, 15 Nov. 1928.
[3] Brown, T. T., “Electrokinetic Apparatus,” U.S. Patent 2,949,550, 16 Aug. 1960.
[4] Brown, T. T., “Electrokinetic Transducer,” U.S. Patent 3,018,394, 23 Jan. 1962.
[5] Brown, T. T., “Electrokinetic Generator,” U.S. Patent 3,022,430, 20 Feb. 1962.
[6] Brown, T. T., “Electrokinetic Apparatus,” U.S. Patent 3,187,206, 1 June 1965.
[7] Brown, T. T., “Appareils et Procédés Électrocinétiques,” (English translation “Elec
trokinetic Apparatuses and Processes”), French Patent 1,207,519, 21 Aug. 1961.
[8] Talley, R. L., “21st Century Propulsion Concept,” Air Force Astronautics Lab. Rept.
AFAL TR 88 031, AD A197 537, Edwards Air Force Base, CA, April 1988.
[9] Cravens, D. L., “Electric Propulsion Study,” Air Force Astronautics Lab. Rept.
AL TR 89 040, AD A227 121, Edwards AFB, CA, Aug. 1990.
[10] URL: http://jnaudin.free.fr/lifters/main.htm [cited 2 March 2007].
[11] Bahder, T. B., and Fazi, C., “Force on an Asymmetric Capacitor,” Army Research
Lab. Rept., ARL TR 3005, Adelphi, MD, June 2003.
[12] Canning, F. X., Melcher, C., and Winet, E., “Asymmetrical Capacitors for Propul
sion,” NASA Glenn Research Center, NASA/CR 2004 213312, Oct. 2004.
[13] Cady, W. M., “Thomas Townsend Brown: Electro Gravity Device,” Office of Naval
Research File 24 185, Pasadena, CA, Sept. 1952.
[14] Buehler, D. R., “Exploratory Research on the Phenomenon of the Movement of High
Voltage Capacitors,” Journal of Space Mixing, Vol. 2, 2004, pp. 1 22.
[15] Stephenson, G. V., “The Biefeld Brown Effect and the Global Electric Circuit,”
Space Technology and Applications International Forum STAIF 2005, El Genk,
M. S. (ed.), American Institute of Physics, 2005, pp. 1249 1255.
[16] Musha, T., “The Possibility of Strong Coupling Between Electricity and Gravita
tion,” Infinite Energy Magazine, Issue 53, Jan. Feb. 2004, pp. 61 64.
[17] Woodward, J. F., “Flux Capacitors and the Origin of Inertia,” Foundations of
Physics, Vol. 34, No. 10, 2004, pp. 1475 1514.
[18] Brito, H. H., “Experimental Status of Thrusting by Electromagnetic Inertia Manipu
lation,” Acta Astronautica, Vol. 54, 2004, pp. 547 558.
[19] Randall, L., Warped Passages: Unraveling the Mysteries of the Universe’s Hidden
Dimensions, HarperCollins Publishers, New York, 2005.
 ASYMMETRICAL CAPACITOR THRUSTERS IN AIR 327

[20] National Geophysical Data Center, National Oceanic and Atmospheric

Administration, “Magnetic Field Calculator”, URL: http://www.ngdc.noaa.gov/
seg/geomag/jsp/Downstruts/calcIGRFWMMM [cited 10 Feb. 2007].
[21] Carreno, F., and Bernabeu, E., “On Wire to Plane Positive Corona Discharge,”
Journal of Physics D, Vol. 27, 1994, pp. 2135 2144.
[22] White, H. J., Industrial Electrostatic Precipitation, Addison Wesley, Reading, MA,
1963, Ch. 4.
[23] Chen, J., and Davidson, J. H., “Electron Density and Energy Distributions in the
Positive DC Corona: Interpretation for Corona Enhanced Chemical Reactions,”
Plasma Chemistry and Plasma Processing, Vol. 22, No. 2, 2002, pp. 199 224.
[24] Chen, J., and Davidson, J. H., “Model of the Negative DC Corona Plasma: Compari
son to the Positive DC Corona Plasma,” Plasma Chemistry and Plasma Processing,
Vol. 23, No. 1, 2003, pp. 83 102.
[25] White, H. J., Industrial Electrostatic Precipitation, Addison Wesley, Reading, MA,
1963, Ch. 4.
[26] Christenson, E. A., and Moller, P. S., “Ion Neutral Propulsion in Atmospheric
Media,” AIAA Journal, Vol. 5, No. 10, 1967, pp. 1768 1773.
[27] Abdel Salam, M., and Shamloul, D., “Computation of Ion flow Fields of AC
Coronating Wires by Charge Simulation Techniques,” IEEE Transactions on
Electrical Insulation, Vol. 27, No. 2, 1992, pp. 352 361.
[28] Zhang, C. H., and MacAlpine, J. M. K., “A Phase Related Investigation of AC
Corona in Air,” IEEE Transactions on Dielectrics and Electrical Insulation, Vol.
10, No. 2, 2003, pp. 312 319.
[29] MacAlpine, J. M. K., and Zhang, C. H., “The Effect of Humidity on the
Charge/Phase Angle Patterns of AC Corona Pulses in Air,” IEEE Transactions on
Dielectrics and Electrical Insulation, Vol. 10, No. 2, 2003, pp. 506 513.
[30] Melcher, J. R., Continuum Electromechanics, MIT Press, Cambridge, MA, 1981, Ch. 3.
[31] Yabe, A., Mori, Y., and Hijikata, K., “EHD Study of the Corona Wind Between
Wire and Plate Electrodes,” AIAA Journal, Vol. 16, No. 4, 1978, pp. 340 345.
[32] Tajmar, M., “Biefeld Brown Effect: Misinterpretation of Corona Wind Phenomena,”
AIAA Journal, Vol. 42, No. 2, 2004, pp. 315 318.
[33] Kawasaki, M., and Adachi, T., “Mechanism and Preventive Method of Self Excited
Vibration of Corona Wire in an Electrostatic Precipitator,” Journal of Electrostatics,
Vol. 36, 1996, pp. 235 252.
 Chapter 9

Experimental Findings of Asymmetrical Capacitor

Thrusters for Various Gasses and Pressures

Francis X. Canning
Simply Sparsew Technologies, Morgantown, West Virginia

I. Introduction
SYMMETRICAL capacitor thrusters have been proposed as a source
A of propulsion. For over 80 years it has been known that a thrust results when
a high voltage is placed across an asymmetrical capacitor, when that voltage
causes a leakage current to flow. Chapter 8 provides experimental results using
the classical “lifter shape” in air, while this chapter provides results for several
geometries that are more “capacitor like” and that have greatly varying
amounts of asymmetry. Measurements are made in air, nitrogen, and argon at
atmospheric pressure and at various partial vacuums. The thrust these devices
produce has been measured for various voltages, polarities, and ground configur-
ations and their radiation in the very high frequency (VHF) range has been
recorded. A number of possible explanations for the source of the thrust are con-
sidered. Several of these are different from those considered in the previous
chapter. However, we also consider a model that assumes the thrust is due to elec-
trostatic forces interacting with the leakage current flowing across the capacitor.
It further assumes that this current involves charged ions which undergo multiple
collisions with air. These collisions transfer momentum. All of the measured data
were found to be consistent with this model.
The force produced by asymmetrical capacitor thrusters (ACT) was first
observed in 1922. A graduate student, T. T. Brown, working under his advisor,
Dr. Paul Biefeld, noticed a force on a device when a high voltage was applied, an
effect sometimes called the Biefeld Brown effect. T. T. Brown received a patent
in Great Britain for the use of this effect in 1928 [1]. More recently, this effect
has been used to produce devices commonly called “lifters.” Lifters are generally

Copyright # 2008 by the American Institute of Aeronautics and Astronautics, Inc. All rights
reserved.

Independent consultant.

329
330 F. X. CANNING

light-weight devices that have a high voltage supplied by attached wires. They have
generated much interest in hobbyists as they lift off of the ground in a way that
appears magical to the casual observer. One such device was patented in 1964 [2].
A common feature of these devices is that they apply a high voltage to an
asymmetrical capacitor. Some of these devices are called asymmetrical capacitor
thrusters. Not only are they asymmetrical, but they generally also have sharp
edges and/or sharp corners. One normally does not think of a capacitor as con-
suming power in its charged state. However, the combination of sharp features
and high voltage tends to produce a small leakage current causing power con-
sumption. Potentials in the range of 50,000 to 100,000 volts are commonly used.
This report describes some recent experiments [3,4] that were designed to
explain some of the confusing lore about how these devices function. For
example, some previous reports suggest that they always created a force
toward the side of the capacitor with the sharper physical features, as does
Chapter 8. Other reports state that the direction of the force changes when the
polarity of the excitation is changed. That conclusion is hard to reconcile with
other reports that observe that these devices function with both a DC and an
AC voltage applied. We control an additional factor in our experiments and as
a result provide a reasonable explanation for all of these observations.
The specific designs that we tested were chosen because they both generated
a relatively strong force and had features that would help in determining
the mechanism that produced the thrust. There have been several reports on
tests of such devices [3 6]. However, this chapter concentrates on reviewing
the interpretation of the tests reported in Refs. 3 and 4, in which this author
participated. New features of our experimental data are 1) the use of devices
with both weak and strong asymmetry, 2) simultaneous control of both the
polarity and ground location, and 3) simultaneous measurements of both the
current into the capacitor and the current out of the capacitor.

A. Lifter Geometries
A typical lifter is made from materials such as aluminum foil and wire. The
wire is near the aluminum foil (Fig. 1) and is on the upper side. The wire may

Fig. 1 Typical lifter.

 FINDINGS OF ASYMMETRICAL CAPACITOR THRUSTERS 331

be considered a sharper surface than the edge of the aluminum foil. The two
wires are charged at different potentials. For illustration, the top wire is shown
as plus, while a negative polarity is just as common.
Lifters seem to most often have only metallic surfaces, although dielectric
material may be used. Even lifters that are several feet in size weigh very
little, often less than an ounce. The previous chapter presented results for lifter
geometries, while we will only present results for asymmetrical capacitor
thruster (ACT) geometries.

B. Asymmetrical Capacitor Thruster Geometries

ACTs are similar in design to lifters, but tend to be more recognizable as
a capacitor with a significant asymmetry. Nevertheless, one side is more dis-
continuous than the other. For example, one design uses a disk and a cylinder,
where both bodies of revolution share the same axis of rotation.
For a disk and a (hollow) cylinder, the disk is considered to have the sharper
features. While each has a surface with an edge, we believe the presence of the
other side of the cylinder softens the discontinuity (more accurately, an abrupt
change) in the resulting electric fields. A numerical calculation for a two-
dimensional version of a disk and cylinder was performed in Ref. 3 to verify
this (see Section III). It was found that the electric field strength at the edge of
the disk when the cylinder was grounded was approximately twice the electric
field strength at the cylinder when the disk was grounded. This numerical
calculation for the two-dimensional case supports our arguments about sharper
features for the three-dimensional case.
One interesting feature that was used on some designs consisted of adding
short individual wires. These wires were obtained from a window screen. The
wires that were parallel to an edge were removed so that the wires that remained
all pointed in the same direction. These created a stronger “discontinuity”
(Section III clarifies what this means). Thus, we were able to compare the
performance of designs that did and did not have these sharper features.
Four devices were examined in detail. Device 1 is shown in Fig. 2. This is the
capacitor shape that others have tested in the past. It is expected that the disk
functions as having a sharper discontinuity than the cylinder. Others have
discussed the possible effect of dielectric materials. Thus, we created device 2
which is shown in Fig. 3.
It is important to understand the effects of greater and of less asymmetry.
Devices 1 and 2 may be considered to be less asymmetric than a standard

Fig. 2 Test device 1.

332 F. X. CANNING

Fig. 3 Test device 2.

lifter design. This is because a wire (Fig. 1) is nearly a one-dimensional object

while a disk or cylinder (Figs. 2 and 3) has surface area. Figure 3 one can see
dielectric material on the left half and a hollow copper cylinder on the right
half of the device.
For comparison we created devices 3 and 4, which have a much larger asym-
metry than both devices 1 and 2 and standard lifters (Fig. 1). As compared to
device 1, devices 3 and 4 make the disk end even sharper and the cylinder less
sharp. As shown in Fig. 4, device 3 has very fine wires on the disk. These fine
wires add to the discontinuity there. Also, the end of the cylinder nearest to
the disk has an added rounded collar, making it a smoother surface.
The final device tested was device 4 that was similar to device 3, with one
added feature. Wires pointing away from the disk were added to the rear end
of the cylinder (the end most distant from the disk). When sharp features (such
as these wires) are placed in a region with a varying electric potential, they
generally cause that potential to vary rapidly near that sharp feature. This
“near discontinuity” generally causes a large local electric field. (Because of
its similarity to device 3, device 4 is not illustrated.)

II. Experimental Setup

Each of the four devices was tested both in a vacuum chamber and in a con-
ducting box having the same dimensions as the vacuum chamber. Two polarities
are possible; one has the disk positive and the cylinder negative and the other
reversing that polarity. Fortunately, in building the apparatus, it was realized
that the ground for the box might be kept at the same potential as either the
disk or the cylinder, for either polarity.

Fig. 4 Device 3.
 FINDINGS OF ASYMMETRICAL CAPACITOR THRUSTERS 333

Fig. 5 Four wiring circuits.

The four possible combinations of polarity and ground are shown in Fig. 5 as
circuits A through D. Observing all four cases proved very illuminating for
understanding the physics of these devices. The resulting data provided an expla-
nation why previous anecdotal information from a variety of sources appeared to
be contradictory. That is, the location of the ground affected the results whereas
other researchers typically did not specify whether the ground was floating or
fixed, or where it was attached.
We also measured two different currents by measuring the current through
each lead. We expected that these two currents could be significantly different,
because there are three current paths. The first two travel from one side of the
capacitor to the other, either through air or through the supporting structure.
The third path goes through the “air” (or other gas or vacuum) to surfaces at
some distance from the ACT.

III. Qualitative Experimental Results

The results of the experiments in Ref. 3 are summarized in Sections III.A and
III.B. Section III.C gives interpretations of those results. These tests were
performed using a direct (nonalternating) current.

A. Tests Performed in Air at Atmospheric Pressure

1) Devices 1 and 2 always produced a force on the ACT toward the non-
grounded (charged) surface.
2) Devices 3 and 4 always produced a force on the ACT directed from the
cylinder toward the disk.
334 F. X. CANNING

3) Devices 1 and 2 produced a larger force when the disk was the non-
grounded surface.
4) The polarity (þ Vs ) had only a small effect on the magnitude of the force
produced (for ground constant).
5) When the cylinder was grounded, devices 3 and 4 produced a larger force
than devices 1 and 2.
6) When the cylinder was grounded, device 4 produced more thrust than
device 3.
7) The current to the live (nongrounded) side was always larger than the
current from the grounded side to ground.
8) When the box containing the apparatus was opened, the hair on one’s arm
stood on end.
Notice that when we say the polarity had only a small effect on the magnitude
of the force produced, we are assuming that when the polarity is changed, the
position of the ground does not change. That is, if one changed the polarity in
circuit A, one would then have circuit D. Similarly, if one changed the polarity
in circuit B, one would get circuit C. However, previous reports generally do
not mention the ground and we do not even know if the ground was fixed or
floating. Thus, if they change the polarity of circuit A, we do not know if their
configuration was circuit B or D, or something else. Thus, it is not surprising
that they reported varying results. For example, we observed that for the less
asymmetric devices (devices 1 and 2), changing both polarity and the ground
(e.g., going from circuit A to B) changed the direction of the force.

B. Tests Performed in a Partial Vacuum in Air and in other Gasses

1) At pressures such as 300 torr in air, the results were similar to atmospheric
pressure but forces were weaker.
2) Also, similar results as for air were found in argon and nitrogen, but with
somewhat smaller forces.
3) In air, the current flowed in bursts, and VHF radiation was observed.
4) In argon and nitrogen, the current did not flow in bursts and the VHF radi-
ation was absent.
5) These bursts in the current were observed on an oscilloscope.
6) In a significant vacuum, with one exception, no force was observed
although experimental sensitivity was low. The exception was a momentary
force that occurred at below 1/10,000 of a torr when a significant spark
(arcing) was observed. This occurred the first time a voltage was applied after
the vacuum chamber had been closed and the pressure reduced.

C. Interpretations of Results
For the tests performed in air, the air became significantly ionized. This
allowed a second current path from the live side to ground through air bypassing
the grounded side of the ACT. It is possible that the wires on the cylinder of
device 4 reduced the ionization as compared to device 3, and resulted in the
larger force observed. However, that is not clear because tests were not done
that controlled the ionization, such as by flushing the air.
 FINDINGS OF ASYMMETRICAL CAPACITOR THRUSTERS 335

The two devices that were only somewhat asymmetrical (devices 1 and 2)
produced a force in a direction that was determined by the location of the
ground. This is due to the live side having a large voltage gradient at its
surface because the voltage changes abruptly to the ambient value (approxi-
mately ground). A large voltage gradient gives a large electric field, which
causes ionization. The ions have the same charge (whether plus or minus)
as the nearby ACT surface, so they are repelled by it. This produces a force
on the ACT which in this case is directed from the grounded side toward
the live side.
The two devices that were highly asymmetric devices (devices 3 and 4) always
produced a force in a direction that was determined by the asymmetry. That is,
the force was always directed from the cylinder toward the disk, regardless of
the polarity or the ground location. A reasonable explanation for this is possible
because the air was clearly ionized in the box containing the ACT when it was in
use. The ionization was displayed in two ways. First, the current into the live side
of the ACT was larger than the current from the grounded side as measured
through the grounding wire. This significant current difference showed that
current was flowing through the air, indicating that it must be ionized. Second,
the air was found to make arm hairs stand up straight, again showing the air
was charged. This suggests that near the ACT, the ambient voltage could be
significantly different from ground due to the charges in the air. Thus, there
could be a significant voltage gradient between the grounded side of the ACT
and the air around it.
The voltage gradient near the grounded side is expected to be significant,
even though it likely is smaller than the voltage gradient near the surface of
the live (non-ground) side of the ACT. Thus, for a large enough asymmetry,
the asymmetry would be the determining factor in the net force on the ACT.
That is, charged particles may be created on both sides of the ACT, and they
would be repelled from their respective nearby side of the ACT. The charged
particles near each side produce forces in opposite directions. Thus, it is quite
reasonable that for a strong enough asymmetry, the net force is always directed
toward the sharper side (e.g., the one with the wires).
The current was found to flow continuously when an ACT was operated in
argon and in nitrogen, but to flow in bursts when operated in air. This was
measured both by a time resolved measurement of each current and indirectly
by observing the VHF radiation that resulted from the bursts of current when
the ACT was used in air. These bursts are a known phenomenon called
“Trichel pulses.” Because a force was produced in argon and in nitrogen
where these bursts do not occur, it appears that the mechanism of the force is
not inherently linked to these Trichel pulses.
The only time a noticeable force was created in a strong vacuum occurred
the first time a voltage was applied after the chamber had been closed.
A significant arcing was associated with this momentary force. Thus, it is
possible that some material was removed from one part of the ACT when that
arcing occurred. It is reasonable that moisture due to humid air may have
deposited on the ACT while the vacuum chamber was open. Thus, this event
may have been due to some material (moisture or otherwise) being ejected
from the ACT.
336 F. X. CANNING

IV. Numerical Calculations of Electric Fields

It is well known that high voltages combined with sharp surfaces cause high
electric fields. Those experienced in working with high voltages are very familiar
with this. The theoretical reason is very simple. An electric field is the rate of
change of the voltage as one measures the voltage at different spatial locations,
and the voltage changes rapidly near discontinuities in conducting objects.
Another way to see this is to consider a piece of wire charged to a potential of
one volt. The charges on the wire tend to repel each other, and are concentrated
(denser) at the ends of the wire. Thus, as one moves from an end of the wire off
the wire, the voltage changes rapidly (a large electric field) because of the large
number of nearby charges. If one moves away from the center of the wire, the
voltage changes slowly (a smaller electric field). A numerical simulation of
this was given for a lifter type of geometry in Ref. 3. A simulation was also
given for an ACT type of geometry, and we describe its results below.
The numerical results presented in Fig. 6 show that in parts a and b, the largest
electric field occurs on the charged (not grounded) plate. In part c, where the
plates are charged to equal strength (but different signs), the strongest field
occurs on the upper (more discontinuous) plate. This supports our arguments
(Section III.C) that the strength of the difference in the “discontinuity” versus
the strengths of the difference of the voltage on a plate from the ambient
voltage determines which electric field will be stronger, and thus the direction
of the net force produced.

Fig. 6 Equipotential lines for a two-dimensional conducting object. In a), the upper
plate is at one volt and the lower plates are at ground. In b), the upper plate is at
ground and the lower plates are at one volt. In c), the upper plate is at plus one
volt and the lower plates are at minus one volt.
 FINDINGS OF ASYMMETRICAL CAPACITOR THRUSTERS 337

V. Theories Versus Quantitative Experimental Results

It has been shown that all of the qualitative observations can be explained by a
model using a flow of ions. It remains to consider other theories and see if they
are plausible, and whether an ion flow model can predict the magnitude of the
force created. Several possible theories for the mechanism that creates the
force (or thrust) are now considered. Except for Section V.C, these are different
from the mechanisms considered in Chapter 8.

A. Ablative Material
Due to the high voltages and high electric fields that are present, material
might be removed from the disk or cylinder during continuous operation. The
possibility that continuously ejected material could provide the force was exam-
ined in Ref. 3. However, because these devices were operated for a large number
of hours with no visible physical change, an upper limit on the amount of mass
that might have been removed was easy to calculate. Using ejection velocities
due to thermal effects it was found that any force created by this would be sig-
nificantly less than that observed, so this effect is not a significant possible
mechanism.

B. Electrostatic Forces Involving Image Charges

This concept is simpler to analyze for a lifter than for our test geometry.
Consider the possibility that a lifter’s charges might interact with image
charges due to a ground plane. For example, a concrete floor might have metal
reinforcement that produces a current that can be described by image charges.
One might assume a perfectly conducting plane under the lifter. (A perfect
conductor would produce the strongest forces. A more realistic surface would
produce significantly weaker forces.) The lifter creates approximately a
charged dipole, which interacts with its image dipole. A simple calculation
shows the resulting force is many orders of magnitude too weak. Thus, this
cannot be the mechanism that produces the force. Adding dielectric material
(as is sometimes done) would increase the force significantly. However, there
is always an attractive force between a charge and its image, so this effect
would pull the lifter down rather than make it rise. Thus, with or without dielec-
tric material, such an electrostatic force would be in the wrong direction to
explain why a lifter lifts.

C. Ion Drift Causing Momentum Transfer to Air

A high electric field near a charged plate will cause charged particles to be
ejected. They will be charged with the same sign as that nearby plate (i.e.,
both plus or both minus), and will be repelled by it. If charged particles at one
side of the ACT are accelerated, move to the other side, and decelerate to
move with the ACT again, no net force is produced. The net change in momen-
tum of such a particle is zero. For a system consisting of these particles and the
ACT, the average force on these particles would be zero so the average force on
the ACT would also be zero. The only way a continuous force could be created on
an ACT due to these particles would be if these particles transferred momentum
338 F. X. CANNING

to something else. Of course, particles moving through air will have collisions
and thus transfer momentum to that air. This is analogous to a propeller
moving through air and transferring momentum.
Charged particles would have a large number of collisions traveling from
one side of an ACT to the other, and as a result would always be moving
much slower than the thermal velocity at sea level and room temperature.
Thus, their collision rate would be approximately unchanged due to their
motion. With this assumption, it is simple to compute the force produced for a
given voltage, distance across the ACT, and current. Further, it may be
assumed that all of the charged particles flow in one direction. With these
assumptions, the force that would be expected on an ACT was computed in
Ref. 3. All of the forces measured in Ref. 3 were found to be smaller than
the result of a computed force using an equation that we reproduce below
[Eq. (3)]. However, for device 4 with the cylinder grounded and the disk posi-
tively charged the result was close at 77% of the computed value. This is in
good agreement as Eq. (3) gives an upper limit to the possible force. If not all
charges flow in the same direction, there will be a partial cancellation reducing
the net force to a value smaller than that given by Eq. (3).
The energy efficiencies, forces produced, and other properties of an ACT or a
lifter may be understood from this ion drift model. The calculation of the force
produced seems straightforward. One first finds how long it takes a charge to
move across the gap between the charged surfaces. The current, multiplied by
this time, gives the charge in the gap. The applied voltage divided by the size
of the gap gives an average electric field. The electric field times the charge
gives a force. This seems simple, but there is one pitfall. As a charged particle
moves across the gap, it has multiple collisions with air (or whatever gas is
present). With these collisions, it takes orders of magnitude longer for a charge
to cross the gap than it would in a vacuum. If this effect is missed, then one
would calculate a force that is orders of magnitude too small.
The simple model used in Ref. 3 assumed that in air at standard conditions,
there are 1010 collisions per second. It was assumed that charged particles also
had this number of collisions with air, and that on average each collision
caused them to lose all of their momentum across the gap. With these assump-
tions, for an ion of mass m, the distance traveled between collisions would be

d0 ¼ (1=2) a t0 2 ¼ (1=2) a ½10 10

sec2 where a ¼ F=m ¼ ½eV=d=m (1)

In this equation, e is the charge on the ion, V is the voltage applied, and a is
the acceleration. The total time, t, to travel a distance, d, across the gap and
the resulting force are:

10
t ¼ 10 sec (d=d0 ) ¼ 2d (1010 = sec)=a ¼ 2d 2 (1010 = sec) m=(eV) (2)
10
F ¼ I  t  V=d ¼ 2d (10 = sec) mI=e (3)

When a 100 kilovolts produces a current of less than one milliamp, the computed
force is significantly smaller than a Newton.
 FINDINGS OF ASYMMETRICAL CAPACITOR THRUSTERS 339

This model is very approximate, and the resulting force is often a few times
smaller than it predicts. Equation (3) shows that if the current is fixed, then the
force produced is proportional to the size of the gap, which is the distance the
charges must flow. We observed that device 3 produced significantly less force
than device 4. There is a likely explanation for this. The air may have been
less ionized for device 4 and therefore the ions were nearly (but not completely)
all produced only on the disk side. For our highest-thrust device, device 4 with
circuit A, the measured force was 77% of the force computed from Eq. (3).
This strongly suggests that this ion drift model explains the origin of the force.

VI. Vacuum Results

Measurements were also made in a vacuum chamber containing dry air, argon
or nitrogen at various pressures. All of those results suggested that the force
produced by an ACT goes to zero as the pressure is reduced. Some measurements
in dry air are shown in Fig. 7.
Figure 7 shows that the forces produced by each ACT decrease as the pressure
was decreased. For each device/circuit combination shown, the same voltage
was used at each pressure. The devices were attached to an axle that rotated,
and the torque they produced is shown on the vertical axis. This has the advantage
that they are moving through fresh (less ionized) air and the disadvantage that the
force measurements are less accurate than for stationary devices. In all of our
measurements, no force was measured for very low pressures.
As the pressure is reduced, the optimal design parameters change. That is, one
might (possibly) design an ACT that is optimized for maximum thrust at a given
pressure. Regardless, however, the trend shown in Fig. 7 would still apply.

VII. Conclusions
Data from some recent tests performed on ACTs were reviewed and a
number of mechanisms were considered for how their thrust is produced.

Fig. 7 Operation in dry air at reduced pressures.

340 F. X. CANNING

These mechanisms were considered theoretically and in light of test results.

Only one mechanism seems plausible, and it relies on standard elementary
physics. This model consists of ions drifting from one electrode to the other
under electrostatic forces. They collide with air as they move, slowing them
down and increasing the time that each contributes to the force. Each collision
transfers momentum to the surrounding air, much as a propeller does. This
model was found to be consistent with all of the observations that were made.
This included how, for certain designs, the direction of the force changed depend-
ing on which side of the ACT was grounded. It also predicted how, for other
designs, the direction of the force did not change depending on which side of
the ACT was grounded. Furthermore, it approximately predicted the magnitude
of the force (thrust) that was measured. This model also predicted that the
direction of the thrust was independent of the polarity of the applied voltage.
Furthermore, it predicted which designs were most efficient at producing a
force. The force we measured always decreased to zero as the pressure was
reduced. In spite of previous speculation about possible new physical principles
being responsible for the thrust produced by ACTs and lifters, we find no
evidence to support such a conclusion. On the contrary, a multitude of details
about their operation is fully explained by a very simple theory that uses only
electrostatic forces and the transfer of momentum by multiple collisions.

Acknowledgments
The author thanks Jonathan Campbell at NASA Marshall and Marc G. Millis
at NASA Glenn for introducing him to this subject. Also, he thanks his coauthors
of Refs. 3 and 4 for performing much of the work that is reviewed here.

References
[1] Brown, T. T., “A Method of and an Apparatus or Machine for Producing Force or
Motion,” British Patent 300,311, Nov. 1928.
[2] deSeveresky, A. P., “Ioncraft,” U.S. Patent 3,130,945, 1964.
[3] Canning, F. X., Melcher, C., and Winet, E., “Asymmetrical Capacitors for Propul
sion,” NASA Glenn Research Center, NASA TM CR 2004 213312, Oct. 2004.
[4] Canning, F. X., Campbell, J., Melcher, C., Winet, E., and Knudsen, S. R., “Asymme
trical Capacitors for Propulsion,” Proceedings of the 53rd JANNAF Propulsion
Conference, Monterey, CA, Dec. 2005.
[5] Talley, R. L., “Twenty First Century Propulsion Concept,” NASA TM CR 2004
213312, 2004.
[6] Bahder, T. B., and Fazi, C., “Force on an Asymmetrical Capacitor Final Report,
Aug. Dec. 2002,” ARL TR 3005, NTIS Order Number ADA416740, 2002.
 Chapter 10

Propulsive Implications of Photon Momentum

in Media

Michael R. LaPointe
NASA Marshall Space Flight Center, Huntsville, Alabama

I. Introduction
N 1908, Hermann Minkowski published derivations of the electromagnetic
I momentum density and Maxwell stress tensor, the latter having an asymmetric
form in macroscopic media [1]. Dissatisfied with this inherent asymmetry, Max
Abraham soon thereafter derived a symmetric form for the stress tensor in a
material medium [2]. From these differing derivations was born the Abraham
Minkowski controversy, the resolution of which continues to be argued nearly
a century after their publications. Put succinctly, Abraham’s formulation predicts
that the momentum density of an electromagnetic field (photon) will decrease
when traversing media with an index of refraction greater than unity; Minkowski’s
derivation predicts that the photon momentum density will increase in such
media. The results from a limited number of experiments have been used to
argue in support of both derivations. The potential implications for breakthrough
propulsion arise from the asymmetric form of the stress tensor, which suggests
that net linear momentum may be imparted to a material medium via photon
drag effects or electromagnetic interactions with the vacuum. A sustained
motion of this self-contained system would allow propulsion without the expen-
diture of propellant or the application of external forces. Concepts proposing to
take advantage of this effect have appeared in the literature, and experimental
efforts have been undertaken to determine whether such ideas may be feasible.
This chapter reviews the origins and evolution of the Abraham Minkowski con-
troversy, and discusses the theories, interpretations, and experimental efforts
designed to resolve the controversy. Propulsion concepts and related experiments

This material is a work of the U.S. Government and is not subject to copyright protection in the
United States.

Project Manager, Science Research and Technologies Project Office.

341
342 M. R. LAPOINTE

seeking to exploit net material momentum are described, and potential theoretical
and experimental issues are identified. The chapter concludes with a summary
of recent developments and prospects for this class of potential breakthrough
propulsion devices.

II. Background
Maxwell’s derivation of the basic equations used to describe the propagation
of electromagnetic fields in vacuum was a crowning achievement of 19th-century
physics. This theoretical framework, built upon the experimental efforts of
Faraday, Ampere, Henry, and others, has stood the test of time unchanged and
serves as the underpinning of all large-scale electromagnetic devices in operation
today. In their differential form, Maxwell’s equations are written:

@B
rB¼0 rE¼
@t (1)
r 1 @E
rE¼ r  B ¼ m0 J þ
10 c2 @t

where B is the magnetic field vector, E is the electric field vector, 10 is the
permittivity of free space, m0 is the permeability
 1=2 of free space, and c is
the speed of light in vacuum, equal to m0 10 . The terms r and J represent
the charge and current density source terms, respectively, which give rise to
the free space electromagnetic fields.
Following standard practice [3,4], this set of equations is often recast in terms
of the potential functions V and A,

@A
B¼rA E ¼ rV  (2)
@t
which satisfy the Lorentz condition,

1 @V
rAþ ¼0 (3)
c2 @t
Defining the electromagnetic field tensor, F, with components
0 1
i Ex
B 0 Bz By 
B c CC
B i Ey C
B Bz 0 Bx  C
B c C
F¼B
i Ez C
(4)
B C
B By Bx 0 C
B c C
@ iE i Ey i Ez A
x
0
c c c
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 343

allows Maxwell’s equations to be written in a more compact form:

r
rE¼
10
) @m Fmn ¼ m0 Jn
1 @E
r  B ¼ m0 J þ
c2 @t
(5)

rB¼0
@B ) @m Fna þ @n Fsm þ @s Fmn ¼ 0
rE¼
@t

where the current and charge densities are related through the continuity
equation:

@r
rJþ ¼0 ) @m Jm ¼ 0 (6)
@t

Equation (2) expressing the potentials V and A can then be written as:
 
2 1 @2
r  2 2 A ¼ m0 J  
c @t 1 @2
  ) r2  Am ¼ m0 J m (7)
2 1 @2 r c2 @t2
r  2 2 V¼
c @t 10

with the Lorentz condition, Eq. (3), taking the form @m Am ¼ 0.

Of particular interest to this chapter, electromagnetic fields carry both energy
and momentum. Using the simplified notation above, we can define the quan-
tity, Fmn J n , whose three spatial components relate to the force density,
(rE þ J  B), and whose fourth component describes the power input per
unit volume, (ci E . J), to the particles that make up the current and charge
density source terms. This quantity can be written in terms of the divergence
of the symmetric, second-rank electromagnetic stress energy momentum
tensor, Tma,

Fmn J n ¼ @a Tma (8)

where

1  
Tma ¼ Fmn Fna þ 14dma Fst Fst (9)
m0

and dma is the Kronecker delta function (dma 1 for m a, zero otherwise).
344 M. R. LAPOINTE

The elements of Tma are given by:

0 1
T11 T12 T13 icgx
B T21 T22 T23 icgy C
T ¼B
@ T31
C (10)
T32 T33 icgz A
ci Sx ci Sy ci Sz uEM

where
 
1 1 1 2
Tij ¼ 10 Ei E j þ Bi B j  dij 10 E2 þ B (i, j  3) (11)
m0 2 m0

are the three-dimensional elements of the electromagnetic Maxwell stress tensor;

1
S¼ EB (12)
m0

is the Poynting vector, which describes the time rate of energy flow per unit area;

1
g ¼ 10 E  B ¼ S (13)
c2

is the momentum density carried by the electromagnetic field; and

1 1
uEM ¼ 10 E2 þ B2 (14)
2 2m 0

is the electromagnetic energy density, the total electromagnetic field energy per
unit volume. The conservation laws of energy and momentum are contained
within Eq. (8). For example, setting m 4 yields

F4n Jn ¼ @a T4a
i
E  J ¼ @1 T41 þ @2 T42 þ @3 T43 þ @4 T44
c
  (15)
i 1 @uEM
¼r  S þ
c ic @t
@uEM
E  J ¼ r  S 
@t

which is a statement for the conservation of energy. Similarly, summing Eq. (8)
over spatial indices 1 to 3 yields an expression for the conservation of
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 345

momentum for the electromagnetic field:

@g j X 3
 ri Tij ¼  f Lj (16)
@t i ¼1

where f L represents the Lorentz force density, f L ¼ rE þ J  B, acting on the

charge distribution.
Maxwell’s equations have been amply verified for electromagnetic wave
propagation in free space, and are commonly accepted pillars of contemporary
physics. In the early part of the 20th century, two well-known figures in theoreti-
cal physics, Hermann Minkowksi and Max Abraham, separately sought to extend
Maxwell’s equations to encompass electromagnetic wave propagation in regions
where matter is present. Their results, and the controversies that have followed,
are described in the following sections.

III. Electromagnetic Fields in Dielectric Media

Within a dielectric material, Maxwell’s equations take the form:

@B
rB¼0 rE¼
@t (17)
@D
rD¼r rH¼Jþ
@t
where D is the displacement field, H is the magnetic field intensity, and the
remaining symbols retain their usual meanings. For linear dielectrics, the consti-
tutive relations between electric and magnetic fields in vacuum and within the
dielectric material are given by:

D ¼ 1E B ¼ mH (18)

Here 1 is the static permittivity of the dielectric, equal to (10 þ x0) where x0 is the
static dielectric susceptibility, and m is the static magnetic permeability, equal to
m0(1 þ xm), where xm is the static magnetic susceptibility. The static dielectric
and magnetic susceptibilities may depend on the electric and magnetic fields,
respectively. The continuity equation takes the modified form:

@
rJþ ðr  DÞ ¼ 0 (19)
@t
Although Maxwell’s equations in matter are generally accepted in the above
form, the corresponding value of the electromagnetic stress energy momentum
tensor has been a continuing source of contention. While there have been many
competing derivations over the intervening decades [5], the two primary protago-
nists at the heart of the debate are Minkowski and Abraham, whose apparently
competing formulations are briefly described as follows.
346 M. R. LAPOINTE

A. Minkowski’s Formulation
In 1908, Minkowski published his derivation of the stress energy momentum
tensor for electromagnetic wave propagation in linearly responsive but spatially
stationary matter [1]. In Minkowski’s formulation, the spatial components of the
electromagnetic stress tensor become:
1
TijM ¼ Ei D j þ Bi H j  dij ðE  D þ H  BÞ (i, j  3) (20)
2

where the superscript M denotes the Minkowski form of the stress tensor. The elec-
tromagnetic momentum density chosen by Minkowski is:

gM ¼ D  B (21)

and the conservation of linear momentum for the electromagnetic field takes the
form:

@gMj X3
 ri TijM ¼ 0 (22)
@t i¼1

Using the Minkowski form of the momentum density, it can be shown that the
momentum flux density for an electromagnetic wave in a stationary material
medium increases proportionally to the index of refraction, n, of the material.
Assuming for illustration a plane electromagnetic wave traveling in the
z-direction, and ignoring losses and dispersion within the material, Maxwell’s
equations provide the simple
p solution, B n(E/c), where n is the index of refrac-
tion of the material, n ¼ m1c. Using Eqs. (14), (18), and (21), and substituting the
solution for B, yields:

1nE2 m
gzM ¼ 1EB ¼ ¼ n EM (23)
c c
Compared with an electromagnetic wave propagating in vacuum, the Minkowski
momentum density of an electromagnetic wave propagating within a material
medium increases by a factor proportional to the index of refraction of
the material.

B. Abraham’s Formulation
Dissatisfied with the asymmetric form of Minkowski’s energy momentum
tensor, Abraham developed, and in 1910 published, a symmetric form of the
electromagnetic stress tensor in spatially stationary matter [2]:

1 1
TijA ¼ (Ei D j þ E j Di þ Bi H j þ B j Hi )  dij ðE  D þ H  BÞ (i, j  3) (24)
2 2
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 347

with the superscript A denoting Abraham’s form for the stress tensor. The
electromagnetic momentum density in this formulation is given by:

1
gA ¼ ðE  H Þ (25)
c2
The conservation of linear momentum then takes the form:

@gAj X 3
 ri T ija ¼  f Aj (26)
@t i¼1

with the so-called Abraham force density, f A , given by:

@
f A ¼ 10 x ðE  BÞ (27)
@t
Using the Abraham formulation for momentum density, and again assuming a
simple plane wave solution to Maxwell’s equations, yields:

EH EB nE2 nE2 1 1 mEM

gA
z ¼ 2
¼ 2 ¼ 2 ¼ 2 ¼ (28)
c c m (c m)c n c n c

In Abraham’s formulation, the momentum flux density for an electromagnetic

wave traversing a stationary medium is inversely proportional to the index of
refraction of the material medium, and hence decreases from its value in
free space. This is clearly at odds with the results derived using Minkowksi’s
formulation, corresponding to a factor of n 2 difference in momentum flux
density between the two approaches. These apparently disparate conclusions
by two pre-eminent physicists ignited a smoldering debate, the resolution of
which remains in contention nearly a century later. Why these results
are important, and in particular why they are important for those seeking
to develop breakthrough propulsion capabilities, are discussed in the
following section.

IV. A Century of Controversy: Theory, Experiment,

and Attempts at Resolution
The momentum densities formulated by Minkowski and Abraham agree for
electromagnetic waves propagating through free space, where the index of refrac-
tion is unity. The dispute arises for electromagnetic waves passing through
material media. In Minkowski’s derivation, the electromagnetic wave (photon)
momentum increases by a factor of n compared to its vacuum value, while
in the Abraham derivation the photon momentum decreases by same factor.
In each case, a corresponding change in the momentum of the material balances
the change in photon momentum, such that total momentum is conserved.
Theoretical and experimental attempts to resolve this apparent inconsistency
between the two approaches have yielded considerable insight into the nature
348 M. R. LAPOINTE

of the electromagnetic stress energy momentum tensor, together with a better if

not yet complete understanding of the nature of photon propagation in dielectric
media. The following paragraphs highlight, in roughly chronological order, some
of the analytic and experimental results published over the past century. For a
more complete history, the reader is referred to Penfield and Haus [5], Brevik
[6], Bowyer [7], and the references therein.

A. Early Years: 1930 to 1960

1. Theoretical Developments
Early discussions on the correct form of the energy momentum tensor in
matter typically centered on the conservation of energy and of linear and
angular momentum. In 1935, Halpern demonstrated that electromagnetic
waves described by a symmetric energy momentum tensor will (as required)
propagate with the velocity of light, provided the four-dimensional divergence
and the diagonal sum of the energy momentum tensor vanish [8]. Applying
this theorem, he concluded that angular momentum is not conserved within the
Minkowski formulation, and the asymmetric tensor form is therefore unsatisfac-
tory. Angular momentum can be conserved in Abraham’s symmetric tensor deri-
vation, but requires that wave propagation within the medium will exert a force
on the medium. Halpern argues that conservation of angular momentum
takes precedence, and that Abraham’s form for the electromagnetic energy
momentum tensor is preferred.
This generally accepted interpretation was reinforced in a 1953 paper by
Balasz [9], who provided a simple thought experiment in support of this
stance. The gedanken consists of two enclosures in uniform motion with no
external forces acting on them. In one enclosure an electromagnetic
wave packet passes through a perfect, nondispersive dielectric rod in which
the wave velocity is lower than that in vacuum. An identical dielectric rod
and wave packet are in the other enclosure, but in this case the wave does
not pass through the rod. Started simultaneously, the electromagnetic wave
traveling through vacuum will be at a different point than an otherwise iden-
tical wave passing more slowly through the dielectric medium; this means
that the equivalent mass associated with the wave energy will also be at a
different location with respect to time. Because no external forces act on
the enclosures, the dielectric rod in which the slower wave packet is traveling
must move to keep the positions of the centers of mass of each enclosure dis-
placed by the same amount. This in turn implies that the dielectric rod
received momentum from the wave packet, and because the total system
momentum is conserved, the momentum of the electromagnetic wave inside
the dielectric must be different than the momentum of an otherwise identical
electromagnetic wave traveling through vacuum. Balasz then asks which of
the two tensor formulations can simultaneously satisfy the conservation of
momentum and center of mass requirements, and after an extended bit of
algebra shows that only the symmetric form of the energy momentum
tensor satisfies both conditions.
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 349

2. Jones and Richards Experiment

In 1954, the results of several carefully designed experiments with sufficient
accuracy to address the Abraham Minkowski controversy were reported by
Jones and Richards [10], who immersed various metallic reflectors in air and
in dielectric media and measured the radiation pressure exerted on each. The
dielectric media ranged from water (index of refraction, n 1.33) to carbon
disulphide (n 1.61); a 30-W tungsten lamp was used as the radiation source,
with short time constants employed to minimize heating and other spurious
forces. Their analysis of the experimental results demonstrated that the ratio of
radiation pressure on the reflector in a dielectric medium to that in air was
proportional to the refractive index of the dielectric material an experimental
confirmation that the Minkowski form of the electromagnetic stress energy
momentum tensor was correct. However, despite its flagrant challenge to the
prevailing theories supporting Abraham’s tensor formulation, the work was not
well known and received little attention at the time.

B. Intermezzo: 1960 to 1990

1. Simplest Cases Theory
In 1968, Shockley [11] employed a “simplest case” technique to once more
argue in favor of Abraham’s electromagnetic energy momentum tensor.
Employing center of mass considerations, Shockley’s thought experiment
consisted of a closed coaxial system, with an inner conducting rod and outer
conducting cylinder electrically connected through a resistor. Power is provided
by a battery. Shockley concludes, upon analysis of equivalent mass flows and
force densities within the system, that the volume integrated form of Abraham’s
momentum density yields the proper value for total system momentum. In an
author’s note added in proof, he credits the above cited work of Balasz with
using a similar simplest method approach for discerning the correct form of
the electromagnetic tensor, but points out that Balasz did not employ the same
force density arguments and as such, his arguments were less well-founded.
Nevertheless, and in spite of Jones and Richards experimental results, the
prevailing theoretical arguments continued to favor Abraham’s symmetric
form for the electromagnetic stress tensor.

2. Ashkin and Dziedzic Experiment

In 1973, Ashkin and Dziedzic [12] published the results of their experimental
efforts to measure the radiation pressure on an air water interface due to the
passage of focused laser light. Their underlying assumption is that the momentum
of light will change upon entering a dielectric medium from free space, due to
Fresnel reflection and via interactions with the medium. Assuming p0 is the
momentum of light in free space and p is the momentum in the dielectric
medium, the net change in momentum is given by fp0(1 þ R)g fp(1 R)g,
where n is the index of refraction of the medium and R is the Fresnel reflection
coefficient. The difference in momentum must be balanced by a mechanical force
on the medium, and they show that if the momentum in the medium is given by
p (n/c) . uEM, then light entering the dielectric will exert a net outward force at
350 M. R. LAPOINTE

the surface. If, instead, the electromagnetic field momentum within the dielectric
medium is given by p uEM/(nc), then the light will exert a net inward force
on the interface surface. These momentum expressions are equal to the
Minkowski (Eq. 23) and Abraham (Eq. 28) terms, respectively, and hence the
experiment could provide a discriminator between the competing concepts.
The experimental setup is discussed in detail in Ashkin and Dziedzic; the
results of their experiments demonstrated a net outward force on the free
surface of the liquid, in agreement with the predictions of the Minkowski form
for the momentum. These observations, which agree with the previously reported
results of Jones and Richards, caught the attention of theorists and spurred an
effort to better understand the nature of electromagnetic fields and forces
within dielectric media.

3. Gordon’s Analysis
In the same year that Ashkin and Dziedzic published the results of their
experiments, Gordon [13] published an analysis in which the concept of pseudo-
momentum was used to delineate between the Minkowski and Abraham forms for
momentum. Like Ashkin and Dziedzic, Gordon worked for Bell Telephone
Laboratories at the time and was keenly aware of the experimental efforts of
his colleagues. Gordon’s analysis showed that for nondispersive dielectric
media, Abraham’s tensor form correctly represents the momentum density of
electromagnetic fields, but the Minkowski form is not precluded if one correctly
takes into account the total radiation pressure acting on an object embedded in the
dielectric. For a broad range of conditions, Minkowski’s momentum density
(here termed a pseudomomentum) can be expressed as the relative dielectric
constant of the medium multiplied by Abraham’s form for the momentum
density. Gordon argues that the pseudomomentum form can be used to evaluate
the radiation pressure on objects embedded in dielectric media, if it is recognized
that the radiation pressure is actually a combination of the mechanical force
exerted directly by the field on the object, and the force exerted on the object
by mechanical pressures induced in the dielectric medium by the presence of
the electromagnetic field. Within this framework, Gordon analyzes the results
of the Ashkin and Dziedzic experiment, as well as that of Jones and Richards,
and concludes that theory and experiment are in complete accord: Abraham’s
momentum density is correct for nondispersive dielectric media, and the pseudo-
momentum formulation embodied by Minkowski’s momentum density correctly
determines the radiation pressure on objects embedded within such a medium. He
concludes with a cautionary note that laboratory experiments designed to
measure the true nature of electromagnetic momentum in dielectric media may
not be feasible.

4. Walker, Lahoz, and Walker Experiment

Undeterred, in 1975 Walker et al. [14] published a set of experimental results
which, unlike previous experiments, supported the Abraham form for the force
density. The premise of the experiment was that, by measuring the motion of a
material body placed in a time-varying electromagnetic field, the total
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 351

momentum can be separated into ordinary mechanical momentum and electro-

magnetic field momentum. The electromagnetic momentum can then be com-
pared to the Abraham and Minkowski predictions.† In the Walker et al.
experiment, a barium titanate disk with a thin aluminum coating was suspended
by a fine wire between the poles of a strong electromagnet, forming a torsion pen-
dulum. A time varying voltage was applied across the disk, providing a slowly
varying (quasi-stationary) orthogonal electric field. The resulting torsion pendu-
lum oscillations were readily measured, and were shown to be consistent with
predictions made using the Abraham form of the force density, to within an
experimental accuracy of +10%. The apparently unambiguous measurement of
a repeatable macroscopic effect explicitly due to the Abraham force term
appeared to rule out the Minkowski formulation by several standard deviations,
and lent strong experimental support to the Abraham side of the long simmering
theoretical debate.

5. Brevik’s Analysis
However, as the reader has probably anticipated, this result did not close the
door on the controversy but instead opened a window of opportunity to gain
deeper physical insight into the varying electromagnetic tensor formulations.
In 1979, Brevik [6] published a detailed review article that outlined and inter-
preted the experimental results of the preceding decades. Regarding the
Walker et al., experiment, Brevik notes that despite some shortcomings (low fre-
quencies, nonmagnetic test body, and constant magnetic field), the experiment
does provide direct experimental evidence for the existence of the Abraham
force term and the Minkowski tensor appears inadequate as a description for
quasi-static fields. However, Brevik goes on to note that Minkowski’s tensor
still has use at optical (and higher) frequencies, where the Abraham term is unob-
servable; in these instances, the relative simplicity of the Minkowski tensor is
often more convenient. Brevik also discusses the 1954 experimental results of
Jones and Richards, which he terms one of the most important experiments in
phenomenological electrodynamics. Following alternative derivations for the
radiation pressure, Brevik shows that either the Abraham or the Minkowski for-
mulation can account equally well for the experimental results; that the Abraham
form in this case would agree with the Minkowski results quoted by Jones and
Richards is due to the vanishing of the Abraham force term at high optical fre-
quencies. Rather than eliminating one form of the electromagnetic field tensor,
the experiment can instead be used to show the close relationship between the
two tensor forms at optical frequencies. Regarding the 1973 results of Ashkin
and Dziedzic, which again appear to support the Minkowski interpretation,

†
An unpublished but apparently similar experiment to detect the Abraham force density was
reported by R. P. James in his Stanford University dissertation. The diligent reader is referred to
James, R. P., Force on Permeable Matter in Time Varying Fields, Ph.D. dissertation, Dept. of Elec
trical Engineering, Stanford University, 1968. See also James, R. P., “A ‘Simplest Case’ Experiment
Resolving the Abraham Minkowski Controversy on Electromagnetic Momentum in Matter,”
[abstract] Proceedings Natural Academy Science, Vol. 61, Nov. 1968, pp. 1149 1150. The results
of James’s experiment are briefly discussed in Brevik [6].
352 M. R. LAPOINTE

Brevik provides a lengthy analysis of the experiment and demonstrates that

the Abraham force term under these reported conditions would not contribute
to the radiation pressure acting on the liquid interface. As such, the Abraham
and Minkowski surface force expressions would be identical, and the experiment
does not actually discriminate between the two. Brevik discusses and dissects a
number of other experiments as well, and concludes his review by noting that, in
most cases, it is sufficiently accurate and often simplest to use the divergence-free
form of the Minkowski tensor.

6. Photon Drag Experiment

Soon after Brevik published his review, Gibson et al. [15] published the results
of an experiment in which radiation pressure was measured by means of the
photon drag effect. This experiment brought to bear a new type of potential
discriminator through the direct measurement of photon momentum density. A
photon traversing a semiconductor material can transfer momentum to the elec-
trons in the valence or conduction bands of the material, generating an electric
field. The high refractive index of semiconductor materials provides significantly
different predictions between the Minkowski and Abraham momentum densities
[Eqs. (23) and (28)]; by measuring the electric field due to photon drag, the
correct tensor description can be established. Experimental results were obtained
using silicon and n-type and p-type germanium semiconductors and photon
wavelengths up to 1.2 mm; the results were shown to be correctly described by
the Minkowski formulation. In 1986, Brevik [16] analyzed the results of the
experiment and, while confirming that the Minkowski tensor provided the most
straightforward explanation of the results, showed that other tensor formulations
were equally able to predict the experimental findings. In concluding, Brevik
notes that the photon drag experiment shows the usefulness of Minkowski’s
tensor in the high-frequency range, from optical to at least millimeter frequen-
cies, but at much lower frequency limits, the results of Walker et al. show that
Minkowski’s tensor cannot be used and the Abraham form provides the correct
description of experimental results. While intermittent discussions continued to
appear in print, this resolution appears to have been generally accepted.

C. Recent Developments: 1990 to Today

1. Theoretical Advancements
The concept of electromagnetic pseudomomentum previously developed by
Gordon was significantly expanded upon in a 1991 paper by Nelson [17], who
used a general Lagrangian approach to develop momentum and pseudomomen-
tum conservation laws for electromagnetic fields interacting with a dispersive,
deforming dielectric medium. Upon quantizing the electromagnetic wave
energy, Nelson sums the momentum and pseudomomentum terms to find a
new conserved quantity, dubbed the wave-momentum, which enters into wave-
vector conservation and phase-matching relationships. In 2004, Garrison and
Chiao [18] published a quantization scheme for electromagnetic fields in a
weakly dispersive, transparent dielectric, and used it to develop canonical and
kinetic forms for the electromagnetic wave momentum. The canonical form
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 353

defines a unique operator that generates spatial transformations in a uniform

media; its physical significance has been established via energy and momentum
conservation during spontaneous emission, Cerenkov and Doppler effects, and
phase matching in nonlinear optical processes. Two choices arise for the
kinetic momentum operator, corresponding to the classical Minkowski and
Abraham momentum terms. Based on a review of prior experiments (outlined
previously), they conclude that the Abraham form is required to correctly
describe closed system experiments in which acceleration of the dielectric is
allowed, at least for classical, low-frequency electromagnetic fields.
In 2004 and 2005, Mansuripur published a series of papers dealing with
radiation pressure and linear momentum related to the once again percolating
Abraham Minkowski discussion. In his first paper [19], Mansuripur applies
the classical Lorentz force to evaluate the interaction of electromagnetic fields
with the charges and currents in conducting and dielectric materials. Following
a cursory review of plane wave reflection from perfect conductors, he next devel-
ops a detailed analysis of the momentum density of an electromagnetic plane
wave in an isotropic and homogeneous dielectric medium. Mansuripur finds
that the correct expression is actually given by an average of the Minkowski
and Abraham momentum densities; however, he notes that this result is limited
due to the effects of multiple plane wave interference that arises in most practical
applications. In such instances, the resulting force on the dielectric medium must
be calculated by a direct evaluation of the charge and current interactions with the
electromagnetic field components. He shows that the electric component of the
radiation field acts on the bound charges within the medium, while the magnetic
field component exerts a force on the bound currents. Sample calculations for
various configurations are provided, and are shown to be in good agreement
with prior Minkowski Abraham related experiments.
The following year Mansuripur extended his analysis to a detailed study of
radiation forces on a solid dielectric wedge immersed in a transparent medium
[20]. Again using the Lorentz law, he shows that the linear momentum of the
electromagnetic field within the dielectric is given by an equal contribution of
both the Minkowski and Abraham momentum terms. For a dielectric surrounded
by free space, he finds that the Lorentz force acting on the bound charges and cur-
rents exactly equals the time rate of change of the incident electromagnetic
momentum. For a dielectric immersed in the transparent liquid medium, the
time rate of change of the incident momentum is equal to the force exerted on
the wedge plus that experienced by the surrounding fluid. In his third paper pub-
lished that same year, Mansuripur derives an expression for the radiation pressure
of a quasimonochromatic optical plane wave incident on the flat surface of a
semi-infinite dielectric medium [21]. He shows that the combined mechanical
momentum acquired by the dielectric, plus the rate of flow of electromagnetic
momentum within the dielectric (i.e., the Abraham momentum), can together
account for the total optical momentum transferred to the dielectric. In this
same paper, he then addresses the results of prior photon drag experiments,
which showed that the charge carriers within the semiconductor acquired a
momentum equal to the Minkowski value. Again using the Lorentz force law,
he constructs a model of a thin absorbing dielectric layer (similar to a semicon-
ductor band gap) and finds that an incident captured photon of initial energy hn
354 M. R. LAPOINTE

transfers an amount of momentum nhn/c to the medium. The latter value is

equivalent to the Minkowski momentum of the photon in the dielectric
medium, which is greater than the total incident momentum of the photon.
This, in turn, means the dielectric must recoil an amount equal to the difference
between the incident photon’s initial momentum and the Minkowski momentum
value picked up by the excited charge carrier within the dielectric. Thus, by
using the Lorentz force law and paying careful attention to boundary conditions,
Mansuripur accounts for a broad range of previously published experi-
mental results.
Nearly concurrent with Mansuripur’s third paper, Loudon et al. [22] published
their own detailed analysis of momentum transfer via the photon drag effect.
They constructed a two-component optical system model, with the semiconduc-
tor material described by real phase and group refractive indices, and the charge
carriers modeled by an extinction coefficient. Using the Lorentz force approach,
they calculated the force on the charge carriers and semiconductor host material
and found that the momentum transfer to the charge carriers per photon was equal
to the experimentally determined Minkowski value, nhn/c, where n is the (phase)
refractive index, n is the optical frequency, and h is Planck’s constant. The
transfer of momentum to the host material is fixed by conservation of momentum
considerations, and requires the total momentum transfer per photon to have
the value (hn=c)[(n2 þ 1)=2n]. Loudon et al. then take their analysis one step
further, and calculate the time dependent Lorentz force when the incident light
is a narrow-band, single photon pulse. For pulse widths significantly shorter
than the attenuation length in the material, the momentum transfer to the
semiconductor can be split into surface and bulk contributions, and they find
that the total bulk momentum transfer (to charge carriers plus host) is equal
to the Abraham momentum value of hn/nc. As did Mansuripur, they argue
there is no conflict in these results, as they apply to different field and boundary
conditions. They further note that there is really no uniquely correct expression
for the transfer of momentum from light to matter, as both the Minkowski and
Abraham values can, in principle, be observed by appropriate measurements.

2. Bose Einstein Condensate Experiment

As theorists continued to hone in on a consensus interpretation, a new
experimental approach to measure electromagnetic momentum transfer made
its debut appearance. In 2005, Campbell et al. [23] published their measurements
of photon recoil momentum from a Bose Einstein condensate of rubidium
atoms. While photon recoil experiments in dilute gas are not new, the use of a
Bose Einstein condensate allowed Campbell et al. to significantly increase the
density of the medium, and to observe how altering the index of refraction
changes the atomic recoil frequency. The intent of the experiment was to
distinguish between two competing theories for the correct amount of atomic
recoil momentum expected for photon propagation through a dispersive
medium. If an atom absorbs a photon, and no motion is left within the system,
the recoil momentum should be h k, where k is the wave number of the photon.
If instead the atom interacts via dipole moment oscillations with the other
atoms in the medium, it will recoil with a momentum of nh k, where n is the
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 355

refractive index of the dilute medium. The results of the carefully conceived and
executed experiment demonstrated that the atoms do indeed recoil with a
momentum of nh  k. While not explicitly developed as a test of the Abraham or
Minkowski formulations, the experiment demonstrated that, at least under
these conditions, the Minkowski formulation predicted the correct momentum
transfer. The results sent a new ripple through the theoretical community,
which quickly rose to the challenge.

3. Leonhardt’s Analysis
Soon after the experimental results of Campbell et al. were published,
Leonhardt [24] calculated the energy and momentum balance in quantum dielec-
trics such as Bose Einstein condensates. He found that, in the nonrelativistic
limit, the total momentum density, g, can be expressed as:

g ¼ rh rw þ D  B (29)

where r is the number density and w is the phase of the condensate. This result
shows that Minkowski momentum is imprinted onto the phase of the quantum
dielectric, which agrees with the experimental results of Campbell et al.
However, Leonhardt goes on to show that the total momentum density can also
be written:

EH
g ¼ rmu þ (30)
c2
where m is the atomic mass and u is the flow velocity of the dielectric media,
defined as
 

1  1 EB
u¼ h rw þ 1  (31)
m m r

Leonhardt notes that the mechanical momentum, mu, differs from the canonical
momentum, h rw, by the Rontgen term ð1  1=mÞE  B=r arising from inter-
actions of the electromagnetic field with the moving electric and magnetic
dipoles of the dielectric medium. Leonhardt’s interpretation is that variations of
the Minkowski momentum are imprinted onto the phase, and the Abraham
momentum drives the flow, of the quantum dielectric. Leonhardt derives rela-
tivistic covariant formulations of Eqs. (29) through (31), and concludes that the
Abraham Minkowski controversy actually has its roots in the Rontgen
interaction.

4. Photonic Band Gaps as Future Testbeds

In 2006 Scalora et al. [25] published the results of detailed derivations for the
Minkowski momentum density and the Lorentz force density assuming a
356 M. R. LAPOINTE

dispersive medium. They evaluated the interactions between short optical pulses
incident on 1) dielectric substrates of finite length, 2) on micron-size multilayer
structures in free space and embedded within a dielectric medium, and 3) on a
negative index of refraction material (with negative values for 1 and m). They
used a numerical approach to solve the vector Maxwell equations and found
that for all the considered cases, the conservation of linear momentum could
be satisfied solely using the Poynting vector and the Abraham form for the
momentum density. They further surmise that neither the Minkowski momentum
density nor the average momentum density advocated by Mansuripur could
reproduce the Lorentz force in any of the circumstances considered. Of additional
interest, they calculate that a narrow band optical pulse of 600-fs duration and
1-MW/cm2 peak power, incident upon a multilayer photonic bandgap structure
with a mass of 10 5 g, may produce accelerations up to 108 m/s2. Although the
very short interaction times will limit sample displacement and velocity, they
suggest that such experiments could be used to test basic electromagnetic
phenomena and momentum transfer to macroscopic media.

D. Summary
Thus stands the famous Minkowski Abraham debate as of this writing. As
Gordon [13] notes, in a statement attributed to Blount, “The argument has not,
it is true, been carried on at high volume, but the list of disputants is very
distinguished.” That the debate shows no signs of abating is evidenced by the
growth of recent popular articles appearing in print [26 28], and it appears
that a final resolution, if indeed one exists, may await further experiment and
analysis. Other electromagnetic tensor formulations, of which there are several,
have not been discussed in the abbreviated and incomplete history given here;
the reader is referred to the excellent reviews of Penfield and Haus [5] and
Brevik [6] for a sampling of these alternative formulations. As we draw this
section to a close, perhaps the most succinct and useful summary of the
Abraham Minkowski controversy is provided by Pfeifer et al. [29], who state
that “Any choice of the electromagnetic energy momentum tensor is
equally valid provided the corresponding material counterpart is also taken
into consideration, as it is only the total energy momentum tensor which is
uniquely defined.” It is with these comforting words of conciliation that we
move from our review of this century-old debate, and step through the looking
glass into proposed propulsive applications.

V. Propulsion Concepts Based on Electromagnetic

Momentum Exchange
The subtle interplay between electromagnetic and mechanical momentum
has over the years inspired several creative and contentious attempts to
produce unidirectional forces and propulsive motion. Unlike typical low thrust
photon propulsion schemes in which internally generated electromagnetic
waves are beamed away from a spacecraft, or externally generated photons or
microwaves are reflected by a spacecraft, the concepts described in this section
attempt to exchange mechanical and electromagnetic momentum within a
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 357

closed system. Time varying electric and magnetic fields are generated within a
fixed volume, and an interchange between electromagnetic field momentum and
spacecraft momentum is postulated to generate mechanical motion within the
prescribed field boundaries. However, the subtleties inherent in the generation
and arrangement of the fields, together with the time-varying interactions
between fields and sources, can quickly obfuscate the underlying physics, and
great care must be exercised to conserve total system energy and momentum.
A cautionary note along these lines appeared in print over half a century
ago, and serves as a useful introduction to this class of propellantless
propulsion concepts.

A. Slepian’s Electromagnetic Space Ship

One of the most widely referenced publications on the possible generation of
unbalanced electromagnetic forces for propulsion is a short 1949 essay by
Joseph Slepian, a distinguished engineer who penned a number of brief
“electrical essays” designed to amuse and educate the readership of the Electrical
Engineering journal. In this particular essay [30], Slepian presented a “means of
propulsion which does not require any material medium upon which the propel-
ling thrust is exerted,” a prescient description of modern breakthrough propulsion
physics efforts. With the typical good humor displayed in his other essays,
Slepian proceeds to walk the reader through the various components that consti-
tute his hypothetical spacecraft, and ends the succinct tutorial with two questions:
1) is there an unbalanced force acting on the material system of the spaceship, and
2) can this unbalanced force be used to propel the craft? His answer to the first
question was yes; his answer to the second, no. Per the instructive nature of
his essays, both answers were explained in a short follow-up publication entitled
“Answer to Previous Essay,” which appeared the following month [31]. Because
the first essay is often cited in support of electromagnetic tensor propulsion
concepts, the salient points of Slepian’s enjoyable discourse and self-rebuttal
are both outlined below.
Slepian begins the construction of his hypothetical space drive using
elementary principles from classical electromagnetic theory. The reader is first
introduced to the standard Lorentz force: a current-conducting wire situated per-
pendicular to a magnetic field will feel a force in a direction perpendicular to both
the current and magnetic field. The second step of the argument reasserts
Newton’s classical Laws of Motion, and states that there is an equal but opposite
force acting on the poles of the (permanent) magnet that corresponds to the force
experienced by the current distribution. The third step in the construction of the
electromagnetic space ship replaces the permanent magnet with an electromagnet
in the form of separated solenoid coils that provide an equivalent magnetic field.
The force on the current distribution running perpendicular to the area between
the separated coils is countered by an equal but opposite force acting on the
solenoid coils, just as occurred for the poles of the permanent magnet. Slepian
then notes that the same forces will arise if an alternating current is used
in place of a direct current in the solenoid windings, if the current distribution
in the space between the coils also changes direction. In this case the force
358 M. R. LAPOINTE

Fig. 1 Slepian’s conceptual electromagnetic momentum generator [30].

on the current distribution remains unidirectional, with an equal and opposite

unidirectional counterforce acting on the electromagnet.
At this point Slepian breaks the continuous current-carrying wire with two
parallel plates (Fig. 1), and introduces the concept of displacement current.
Due to the gap between the plates, the conduction current flowing along the
wire is no longer continuous; however, the current flowing along the wire is
maintained by the displacement current between the plates, which is exactly
equal to the conduction current. The displacement current between the plates is
given by 10 (@E=@t), which appears in the set of Maxwell’s equations [Eq. (1)].
Slepian then shows that there will be an unbalanced force due to the displacement
current appearing in the space between the plates and the magnetic field, equal to:
ð 
@E
F1 ¼ 10  B dV (32)
@t

where V is the volume between the plates. By attaching the device to the side of
a spacecraft, the unbalanced force generated on the solenoid coils by the
interaction of the oscillating magnetic field with the displacement current
would move the craft in a direction perpendicular to both, providing propulsion
without propellant. Thus ends Slepian’s first essay, with an admonishment to the
reader that while in principle this unbalanced force exists, it cannot be used to
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 359

propel the ship. As with his other essays, Slepian provides his reasoning in a sep-
arate article published the following month, allowing the reader time to think
about the puzzle prior to giving the answer.
In his second essay, Slepian observes that the above calculations, while
correct, do not proceed quite far enough. According to Maxwell’s equations,
the time-changing magnetic field generates an electric field between the plates;
the interaction between the magnetic field and induced electric field provides a
force that acts on the parallel plates:
ð 
@B
F2 ¼ 10 E  dV (33)
@t

The time average of this force acting on the charged plates is just equal and oppo-
site to the time average of the force acting on the solenoid magnets, resulting in
no net (time-averaged) motion of the device. How then was Slepian able to assert
a positive answer to the question posed in his first essay regarding net unbalanced
forces acting on the system? A careful consideration of the oscillating electric
and magnetic fields in the space between the plates shows that they are in time
quadrature, hence a non-zero instantaneous resultant force does indeed exist.
However, this instantaneous force alternates in direction, hence the net, time-
averaged force acting on the system will be zero. As Slepian remarks, “Yes,
there will be an unbalanced net force on the space ship [as calculated], but it
will be an alternating force.” Hence, no net motion will result.
In his closing remarks, Slepian discusses the conservation of momentum and
energy for his hypothetical space drive. Combining the expressions for the forces
above, the (instantaneous) unbalanced force on the material system is given by:
ð  ð 
@E @B
F ¼ F1 þ F2 ¼ 10  B dV þ 10 E  dV
@t @t
ð
@
¼ ð10 E  BÞ dV (34)
@t

where the quantity in brackets under the last integral is recognized as the momentum
density of electromagnetic fields in free space [Eq. (13)]. The electromagnetic
energy density in the free space between the plates is then given by Eq. (14).
In the hypothetical space drive, the electric and magnetic fields are caused to
oscillate at high frequency, resulting in a rapidly changing momentum density
between the plates. By conservation of momentum, there will be an equal but
opposite rate of change in the momentum of the material system. This again
corresponds to an instantaneous unbalanced force acting on the system, whose
magnitude is given by Eq. (34). However, as noted above, this force alternates in
direction and its time average over an oscillation period still sums to zero.
While “Slepian drives” still appear in the literature, it is clear that Slepian
meant his electrical essay to be an informative and entertaining mental diversion
for his readers, with no pretext that such a drive would actually work. The
problem he posed and the solution he discussed serve to illuminate the difficulties
360 M. R. LAPOINTE

inherent in untangling and understanding the various field interactions that

underlie this class of breakthrough propulsion concepts. Far from advocating
that such a system would work, his essays instead provide a clarion call that
all such systems must remain consistent with the well-established laws of
momentum and energy conservation.

B. Electromagnetic Stress-Tensors and Space Drive Propulsion

More recently, Corum et al. [32,33] have sought to extend Slepian’s
hypothetical electromagnetic space drive to include the effects of nonstationary
boundary conditions. Using the definition of the displacement field, D ¼ 10 E,
Corum writes the electromagnetic momentum density force appearing in
Eq. (34) as:

@ðD  BÞ
fH ¼ (35)
@t
and renames it the Heaviside force density, f H , due to its appearance in the
derivation by Heaviside of Maxwell’s stress tensor [34]. In agreement
with Slepian, Corum notes that the time average of this force density, integrated
over a stationary bounding volume, will provide zero net force. Corum et al. then
speculate on the possibility of rectifying this force using a temporally discontinu-
ous action, for example by some appropriate modulation of the boundary con-
ditions. Corum et al. argue that if this is possible, then the average of
the integrated force will no longer vanish and unidirectional motion may
indeed take place.
Among the experiments Corum et al. cite in evidence of a possible
unidirectional force component is the early 20th-century work of R. Hartley,
inventor of the Hartley oscillator. Hartley observed a sustained, residual mech-
anical offset on the plates of a capacitor charged with alternating current [35]. In
the 1950s, Manley and Rowe expanded on these results and concluded that
time-varying and nonlinear elements may both contribute to the observed mech-
anical offset [36,37]. Corum et al. propose that a similar effect could occur in
the Slepian electromagnetic device, perhaps resulting in a small but sustained
unidirectional force on a spacecraft so equipped. In fiscal year 2000 and for a
few years following, Corum et al. received funding to conduct a further inves-
tigation of this effect. The intent of the research program was to experimentally
investigate whether the Heaviside force was real and observable, and Corum
et al. set up a program to try and rectify the force on a capacitor to cause uni-
directional movement of a wheeled device. The device chosen was a modifi-
cation of what is historically known as the ampere rail motor, a concept
similar to a modern rail gun in which a direct current is passed along conducting
rails and across a conductive load; the current through the load interacts with
the magnetic fields produced by the current in the rails to provide Lorentz accel-
eration of the device along the rails. The experiment sought to replicate this
using high frequency alternating currents through a capacitor placed in series
with the load, in which the hypothesized Heaviside force would produce
an offset force, resulting in small but perceptible movement of the device.
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 361

A concurrent experimental effort was set up to repeat the 1981 experiments of

Lahoz and Graham [38], who used a torsion pendulum to try and observe the
momentum density produced by static electric and magnetic field distributions.
As reported in 2001, both experimental efforts were proceeding to hardware
fabrication in anticipation of near-term testing; however, to date there has
been no further publications regarding their status or results, and it appears
that this activity is no longer being funded.
In 2002, the U.S. Air Force (USAF) Academy undertook an independent
investigation of the Heaviside force conjecture [39]. The effort employed
both theory and experiment to investigate the possibility of producing a uni-
directional Heaviside force as proposed by Corum et al. The theoretical analysis
showed that using the electromagnetic force density per se would not lead to
any discernable net thrust, but that the time derivative of the electromagnetic
momentum density, as represented by Eq. (35), might, in the words of the
report, cause an acceleration of space and hence a discernable force. The
USAF Academy research team used a laser interferometer arranged with one
beam passing through an apparatus held under vacuum, in which perpendicular
electric and magnetic fields would be generated; the other beam traversed a
similar space devoid of electromagnetic fields. The beams are recombined at
a photodetector, and the signal searched for any effect due to beam traversal
of the crossed electric and magnetic fields versus traversal when the fields are
absent. The thought was that the crossed and oscillating fields might
somehow produce an acceleration of space, which would then be picked up
as a change in the interference pattern of the recombined beams. As reported,
the laser interferometer did not find any acceleration of space. The team rede-
signed its experiment to look instead for possible oscillations on a parallel plate
capacitor using parallel applied and induced electric fields, corresponding to
possible linear electromagnetic momentum transfer. However, it does not
appear that sufficient funding remained to carry out the revised set of exper-
iments, and no results of the modified USAF Academy experiment have to
date been published. An independent analysis of the original USAF Academy
approach was undertaken by the Instituto Universitario Aeronautico in Argen-
tina, which determined that no net forces would be produced under this arrange-
ment, and that no acceleration of space or any matter within that space would
occur, confirming the null results of the experiment [40].
Further critiques of the conceptual use of the Heaviside force for propulsion
were published by Brito and Elaskar [41], who developed their own concept
for propellantless propulsion via the manipulation of electromagnetic fields
(see Sec. IV.C). Brito and Elaskar argue that the volume integrated force in
the vacuum space between the parallel plates of the system is identically equal
to zero, hence any rectification procedure applied to the instantaneous force
to generate unidirectional acceleration is meaningless. They argue that this
renders moot the possibility of temporally modulating the boundary conditions
to provide a non-zero time average force, unless polarized matter is present.
However, even modulating the boundary conditions on the dielectric will not
circumvent the problem because the momentum of the complete system must
still be conserved, resulting in a zero net force. Brito and Elaskar thus conclude
that Slepian’s instructional construct, and Corum et al.’s extension to include
362 M. R. LAPOINTE

a conjectured unbalanced Heaviside force, are not capable of producing

unidirectional acceleration of the system.

C. Electromagnetic Inertia Manipulation (EMIM) Propulsion

In the late 1990s, Brito began publishing his work on theoretical and
experimental efforts to use Minkowski’s form of the stress tensor as a basis
for propellantless breakthrough propulsion [41 46]. Dubbed electromagnetic
inertia manipulation, or EMIM, the fundamental concept is that the total space-
time momentum of a material spacecraft, plus any electromagnetic field it gener-
ates, must be a conserved quantity. By considering four-dimensional spacetime
instead of three-dimensional space, Brito argues that the four-velocity of the
spacecraft can be changed by a proper manipulation of the electromagnetic
portion of the total mass tensor (ship plus fields), analogous to an ice skater chan-
ging (angular) velocity by changing the moment of inertia. The EMIM system is
comprised of the spaceship plus the extended electromagnetic fields it generates;
the conjecture is that a net displacement force on the spacecraft would result from
momentum exchange with the electromagnetic field. A conceptual schematic of
the electromagnetic momentum generating device is shown in Fig. 2.
Brito proposed a set of related hypotheses to describe how such a system
might operate, with the formulations based on electromagnetic field momentum,
mass tensors, and electromagnetic force densities. While the mass tensor
formulation appears most often in his later publications, each approach is
briefly described.

Fig. 2 Brito’s conceptual EMIM generator [46]. (Reproduced with permission of

the American Institute of Aeronautics and Astronautics.)
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 363

1. Electromagnetic Field Momentum

The basis for this derivation directly relates to the original Abraham
Minkowski stress tensor controversy, and Brito re-examines the genesis of
Eqs. (21) and (24) to determine whether the non-zero momentum densities appar-
ently arising in Minkowski’s formulation can lead to a non-zero total electromag-
netic momentum for the coupled ship-field system. Brito invokes the results of
Furry [47] to show that the total electromagnetic momentum, G, for a static dis-
tribution of electric and magnetic fields in a material medium, whose magnitudes
rapidly decay to zero at infinity, can in general be expressed as:

ð "  
ð  2 #
1 2 1
G ¼ gdV ¼  x r 0  S þ 0 r  S dV (36)
c c

where x is a position vector, S is the Poynting vector, uEM is the total energy
density, and c0 is the speed of light in the medium, (m1) 1/2. Brito notes the
quantity in brackets is simply the divergence of the electromagnetic momentum
density, g; if this quantity is not divergence free everywhere, then it may be poss-
ible to produce a non-zero total electromagnetic momentum. This leads him to
adopt Minkowski’s formulation for the electromagnetic momentum density
(Eq. 21); Abraham’s formulation would result in zero total electromagnetic
momentum. Referring to Fig. 3, he shows that the quantity r . S vanishes,
which leaves the gradient in the velocity of light within the integral region as
the only remaining possibility for generating non-zero total electromagnetic
momentum. Such gradients might arise across the free surfaces of a finite

Fig. 3 Electromagnetic energy flow and momentum density in EMIM device [46].
(Reproduced with permission of the American Institute of Aeronautics and
Astronautics.)
364 M. R. LAPOINTE

dielectric material, and Brito argues that the device depicted in Fig. 3 provides
just such an arrangement.
Invoking the conservation of total mechanical and electromagnetic
momentum, Brito conjectures that the time rate of change of the volume inte-
grated electromagnetic momentum produced by the EMIM device will be
balanced by the time rate of change of the mechanical momentum of the physical
device, producing a force on the device and providing net thrust. Noting that the
electromagnetic fields fall rapidly to zero outside of the dielectric boundaries, he
shows that the net thrust, T, acting on the device is approximately equivalent to:
ð ð
@g @
Tffi dV ¼  ðD  BÞdV (37)
@t @t

where the integration is taken over the material volume of the dielectric medium. In
this approach, the possibility of generating net thrust explicitly hinges upon the
correctness of the Minkowski interpretation of the electromagnetic momentum
density. While similar to Corum’s development using the Heaviside force
density (Eq. 35), Brito’s analysis is restricted to integration volumes bounded by
the dielectric medium, with the discontinuity in the speed of light at the material
boundaries responsible for the generation of electromagnetic momentum and
the offsetting mechanical momentum required for total momentum conservation.

2. Mass-Inertia Tensor
In this formulation, Brito considers the spacecraft mass and the equivalent
mass of the electromagnetic field generated by the EMIM device aboard the
craft as a single point mass located at the center of mass of the total system.
He then derives a mass-inertia tensor, M, given by:
   
M ¼ m0 þ mEM I þ pEM ^ v =c2 (38)

where m0 represents the material rest mass, mEM is the equivalent mass of
the electromagnetic field in the spacecraft frame of reference, pEM is the
four-momentum of the electromagnetic field, v is the four-velocity of the
point center of mass, ^ is the wedge product, and I is the identity tensor.
For a closed system completely encompassing both material spacecraft and
electromagnetic field, the conservation of momentum is written:

dM  v þ M  dv ¼ 0 (39)

Combining Eqs. (38) and (39), Brito concludes that any change in the
electromagnetic field momentum will be exactly balanced by a change in the
mechanical (spacecraft) momentum. Restated, the formulation argues that
momentum can be exchanged between matter and field within the closed
system bounded by the electromagnetic field distribution, with the amount of
momentum that can be transferred to the spacecraft limited by the amount
of electromagnetic momentum contained within the closed system.
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 365

3. Electromagnetic Momentum Density

The third approach considered by Brito is again based on Minkowski’s
formulation for the electromagnetic momentum density, with both the displace-
ment and magnetic fields undergoing harmonic excitation. Within this frame-
work, Brito develops an expression for the time-averaged electromagnetic
thrust on a closed system. Related to the EMIM generator shown in Fig. 2, the
expression becomes:

1r vNIVd
kTl ¼ sinw (40)
2c2
where the applied voltage is V . sin(vt), the applied current is I . sin(vt þ w), 1r is
the relative permittivity of the medium, N is the number of turns in the magnet
coil, and d is the distance between capacitor plates. This equation provides a
set of tunable physical parameters that can be used to experimentally evaluate
the time averaged force acting on the proposed EMIM device.
Armed with the above hypotheses, Brito performed a series of experiments to
validate the electromagnetic inertia manipulation concept represented in Figs. 2
and 3. These experiments, and the current status of the EMIM effort, are dis-
cussed below.

4. EMIM Force Experiments

Beginning in the mid-1990s, an experimental embodiment of the conceptual
EMIM thruster shown in Fig. 2 was built, tested, and refined in an effort to
demonstrate sustained thrust via electromagnetic inertia manipulation. Barium
titanate (BaTiO3) ceramic with a high relative permittivity was used as the
dielectric medium. Periodic current and voltage supplied to the magnet coils
and capacitor plates, respectively, provided the magnetic and electric
displacement fields required to generate electromagnetic momentum within the
dielectric, with the fields falling rapidly to zero outside the device. Acting as a
resonant mass, the device was placed atop a thin vertical cantilever beam (res-
onant blade), which in turn rested on a vibration-free table. Piezoelectric strain
transducers were used to measure the displacement of the cantilever beam and
infer the net force acting on the system. A common power supply providing
100 VAC at 30 kHz energized three, 900-turn, parallel-mounted toroidal coils
connected in series with three parallel mounted, 10 nF, 8-mm-wide annular
capacitors. With this experimental configuration, Brito calculated that a peak
electromagnetic momentum of 5  10 11 N should be generated by the system,
capable of providing a time averaged thrust of around 5  10 6 N. A later
upgrade of the power supply to 200 VAC at 30 kHz doubled these estimates to
around 10 10 N and 10 5 N, respectively. Additional details on the experimental
configurations are included in Brito [43 45] and Brito and Elaskar [46].
Operating first with the 100-VAC supply, Brito measured the displacement of
the resonant mass under conditions in which either the coils or the capacitors were
energized, in which case no thrust is expected to occur, and when both the coils
and capacitors were energized, in which case a displacement was expected to
occur via electromagnetic momentum interchange with the mechanical system.
366 M. R. LAPOINTE

They note that uncertainties arising from mechanical and electrical interference
prevent a clean interpretation of the results; nevertheless, the claim is made that
the predicted force appears at a level sufficiently above the experimental back-
ground noise to demonstrate that a net propulsive effect does indeed occur. The
reader is referred to the experimental results presented by Brito and Elaskar
[46] to draw their own conclusion as to the validity of this claim.
As might be expected, there are a number of extraneous factors that must be
accounted for in the experimental setup and interpretation of results. Among
these are electrostatic and magnetic field interactions with laboratory surround-
ings, interactions with self-induced magnetic fields, differential heating of thruster
or test components, ground and air vibrations, geomagnetic field interactions,
coronal discharges, ion wind, and myriad other nuanced effects that must be
tracked and eliminated as potential sources for any observed net motion. Brito dis-
cusses and dismisses a number of these potential sources of uncertainty as negli-
gible [46], but the possibility of unaccounted interactions clearly remains. Such
effects are exacerbated by the rapidly pulsed nature of the device, as well as the
extremely low levels of anticipated net thrust. While it will be difficult to
account for all such interactions, extraordinary claims require extraordinary
proof, and a diligent effort is needed to better understand and eliminate any remain-
ing sources of uncertainty before claims of experimental proof are fully justified.

D. Summary
The work of Slepian, Corum, Brito, and others outlined in this section serves
to illuminate the challenges inherent in this approach to breakthrough propulsion.
The analysis of potential interactions between electromagnetic field and material
medium is complex, and sources of experimental noise are often exceedingly
difficult to discern. It is an enormous and perhaps unwarranted extrapolation to
move from still-debated forms of the electromagnetic stress tensor to conceptual
systems designed to generate net motion. Nevertheless, such endeavors do
provide additional understanding of electromagnetic field matter interactions,
and help clarify areas for potential future research. One such area with
possible implications for breakthrough propulsion is the Feigel hypothesis,
briefly discussed below.

VI. Feigel Hypothesis

In 2004, a new and controversial physical interpretation of the momentum
transfer between matter and field was published by Feigel [48], who applies a rela-
tivistic Lagrangian formulation to the interaction of light and matter in a dielectric
medium. He derives the equations of motion and conservation laws, and
finds that in the case of a liquid dielectric, the electromagnetic field interactions
will cause the medium to move. Returning to the Minkowski Abraham
controversy, he concludes that Abraham’s expression represents the momentum
of the field, but because the measured momentum also includes a matter contri-
bution, its value corresponds more closely to the Minkowski formulation. In a
novel piece of analysis, Feigel predicts that zero-point vacuum fluctuations
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 367

contribute to the motion of a dielectric media in crossed electric and magnetic

fields. This astonishing result raises the possibility that momentum can be
transferred from the zero-point fluctuations to matter, with the direction and
magnitude of the force controlled by externally applied electric and magnetic
fields. Feigel estimates that for a relative dielectric constant of 1.5, material
density of 103 kg/m3, and externally applied orthogonal electric and magnetic
fields of 105 V/m and 17 T, respectively, one could achieve a small but measurable
velocity of around 50 nm/s due to the contribution of high frequency vacuum modes.
Following this publication, there appeared a number of comments and replies
[49,50], culminating with a 2006 paper by van Tiggelen et al. [51] in which they
reanalyze Feigel’s results using a Lorentz-invariant description with field
regularization techniques to cope with vacuum divergence. They find that for
an infinite medium, there is no transfer of momentum via Feigel’s proposed
mechanism. However, they note that a vanishingly small momentum transfer
may occur in the squeezed vacuum states provided by inwardly moving parallel
plates in a Casimir geometry. For this exercise they assumed the Casimir plates
start at a large separation distance, with a magneto-electric material slab at rest in
the space between them. The crossed electric and magnetic fields are switched on,
and the plates are slowly brought closer together to a finite separation distance
(this would occur naturally due to the Casimir force acting to push the plates
together). The inward motion of the plates thus converts vacuum energy into
kinetic energy, but there is no transfer of momentum to the system because the
plates are moving with opposite momentum. However, the electromagnetic
momentum density in the space between the plates is changing with the plate
separation distance; this change must be compensated by a change in the
momentum density of the magneto-electric material in order for the total
system momentum to remain zero. Using similar material properties, van Tigge-
len et al. calculate that the medium would acquire a momentum density nearly 20
orders of magnitude smaller than that originally predicted by Feigel, which is
immeasurable.
However, in a companion report to the European Space Agency (ESA), van
Tiggelen et al. [52] calculate that the Feigel effect for a small magneto-electric
object immersed in an isotropic, monochromatic radiation field could be set
into motion if the external fields are suddenly switched on. They note that exper-
imental verification in this instance would be facilitated by the ease with which
monochromatic radiation fields can be generated and modulated. For a field
intensity of 10 kW/cm2, they predict a velocity approaching 10 5 cm/s could
be achieved for an object made of FeGaO3. Experimental verification could be
achieved using a 10-mg crystal of FeGaO3 mounted at the end of a piezo-resistive
cantilever. Immersed in a 10-kW/cm2 radiation field, the crystal would acquire a
momentum of approximately 10 10 g-cm/s. Using a DC magnetic field and per-
iodic switching of the radiation field could provide an oscillatory force on the
mass sample, of sufficient magnitude to be measured experimentally. Recently,
van Tiggelen, Rikken and Krzic proposed to perform these experimental
measurements, which could help determine whether the controversial effect
predicted by Feigel actually exists. As of this writing, no additional information
is available on the status of their proposed experimental effort.
368 M. R. LAPOINTE

VII. Conclusions
The still-debated physical mechanism responsible for the interchange of
electromagnetic and mechanical momentum in a dielectric medium retains an
allure for breakthrough propulsion research, offering a potentially self-contained
method of propulsion without the expenditure of propellant. However, purported
demonstrations of unidirectional motion from a rectified harmonic force or by
exchanging mechanical momentum with the momentum of an extended electro-
magnetic field remain to date elusive and unsubstantiated. Experimental efforts
are often susceptible to the extraneous noise and unwanted motion inherent in
most laboratory environments, while theoretical arguments may easily get lost
in the subtle accounting of electric and magnetic field interactions. We are
reminded once again of Slepian’s decades-old admonishment that all such pro-
posed systems must remain consistent with the conservation of energy and
momentum, indispensable principles in the continuing search for new break-
through propulsion concepts.

Acknowledgments
The author thanks C. Mercer, M. Millis, T. LaPointe, and J. Wilson for
several useful comments and corrections. Portions of this research were conducted
under the auspices of the Science Research and Technology Projects Office at the
NASA Marshall Space Flight Center, whose support is gratefully acknowledged.

References
[1] Minkowski, H., “Die Grundgleichungen fur die Elecktromagnetischen Vorgange in
Bewegten Korpen,” Nachrichten von der Gesellschaft der Wissenschaften zu
Gottingen, Vol. 53, 1908, pp. 53 111.
[2] Abraham, M., “Sull’Elettrodinamica di Minkowski,” Rendiconti Circolo Matematico
di Palermo, Vol. 30, 1910, pp. 33 46.
[3] Jackson, J. D., Classical Electrodynamics, 2nd ed., Wiley, New York, 1975, Ch. 6.
[4] Cook, D. M., The Theory of the Electromagnetic Field, Prentice Hall, Saddle River,
NJ, 1975, Chs. 6, 12, and 15.
[5] Penfield, P., Jr., and Haus, H. A., Electrodynamics of Moving Media, MIT Press,
Cambridge, MA, 1967.
[6] Brevik, I., “Experiments in Phenomenological Electrodynamics and the
Electromagnetic Energy Momentum Tensor,” Physics Reports, Vol. 53, No. 3,
1979, pp. 133 201.
[7] Bowyer, P., “The Momentum of Light in Media: the Abraham Minkowksi
Controversy,” dissertation, School of Physics and Astronomy, Southampton, UK,
Jan. 2005. URL: http://peter.mapledesign.co.uk/writings/physics/2005 dissertation
The Abraham Minkowski Controversy.pdf.
[8] Halpern, O., “A Theorem Connecting the Energy Momentum Tensor with the
Velocity of Propagation of Waves,” Physical Review, Vol. 48, Sept. 1935, pp.
431 433.
[9] Balasz, N. L., “The Energy Momentum Tensor of the Electromagnetic Field Inside
Matter,” Physical Review, Vol. 91, No. 2, July 1953, pp. 408 411.
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 369

[10] Jones, R. V., and Richards, J. C., “The Pressure of Radiation in a Refracting
Medium,” Proceedings of the Royal Society London, Series A, Vol. 221, 1954,
pp. 480 498.
[11] Shockley, W., “A ‘Try Simplest Cases’ Resolution of the Abraham Minkowski
Controversy on Electromagnetic Momentum in Matter,” Proceedings of the National
Academy Science, Vol. 60, 1968, pp. 807 813.
[12] Ashkin, A., and Dziedzic, J. M., “Radiation Pressure on a Free Liquid Surface,”
Physical Review Letters, Vol. 30, No. 4, 1973, pp. 139 142.
[13] Gordon, J. P., “Radiation Forces and Momenta in Dielectric Media,” Physics Review
A, Vol. 6, No. 1, 1973, pp. 14 21.
[14] Walker, G. B., Lahoz, D. G., and Walker, G., “Measurement of the Abraham Force
in a Barium Titanate Specimen,” Canadian Journal of Physics, Vol. 53, 1975,
pp. 2577 2568.
[15] Gibson, A. F., Kimmitt, M. F., Koohan, A. O., Evans, D. E., and Levy, G. F., “A
Study of Radiation Pressure in a Refractive Medium by the Photon Drag Effect,”
Proceedings of the Royal Society of London A, Vol. 370, 1980, pp. 303 311.
[16] Brevik, I., “Photon Drag Experiment and the Electromagnetic Momentum in
Matter,” Physics Review B, Vol. 33, No. 2, 1986, pp. 1058 1062.
[17] Nelson, D. F., “Momentum, Pseudomomentum, and Wave Momentum: Toward
Resolving the Minkowski Abraham Controversy,” Physics Review A, Vol. 44,
No. 6, 1991, pp. 3985 3996.
[18] Garrison, J. C., and Chiao, R. Y., “Canonical and Kinetic Forms of the Electromag
netic Momentum in an Ad Hoc Quantization Scheme for a Dispersive Dielectric,”
Physics Review A, Vol. 70, No. 5, 2004, p. 053862(8).
[19] Mansuripur, M., “Radiation Pressure and the Linear Momentum of the
Electromagnetic Field,” Optics Express, Vol. 12, No. 22, pp. 5375 5401.
[20] Mansuripur, M., “Radiation Pressure on a Dielectric Wedge,” Optics Express, Vol.
13, No. 6, 2005, pp. 2064 2074.
[21] Mansuripur, M., “Radiation Pressure and the Linear Momentum of Light in
Dispersive Dielectric Media,” Optics Express, Vol. 13, No. 6, 2005, pp. 2245 2250.
[22] Loudon, R., Barnett, S. M., and Baxter, C., “Radiation Pressure and Momentum
Transfer in Dielectrics: The Photon Drag Effect,” Physics Review A, Vol. 71,
No. 6, 2005, p. 063802(11).
[23] Campbell, G. K., Leanhardt, A. E., Mun, J., Boyd, M., Streed, E. W., Ketterle, W.,
and Pritchard, D. E., “Photon Recoil Momentum in Dispersive Media,” Physical
Review Letters, Vol. 94, No.17, 2005, 170403(4).
[24] Leonhardt, U., “Energy Momentum Balance in Quantum Dielectrics,” Physics
Review A, Vol. 73, No. 3, 2005, p. 032108(12).
[25] Scalora, M., D’Aguanno, G., Mattiucci, N., Bloemer, M. J., Centini, M., Sibilia, C.,
and Haus, J. W., “Radiation Pressure of Light Pulses and Conservation of Linear
Momentum in Dispersive Media,” Physics Review E, Vol. 73, No. 5, 2006,
p. 056604(12).
[26] Cho, A., “Momentum from Nothing,” Physics Review Focus, 2004, URL: http://
focus.aps.org/story/v13/st3.
[27] Leonhardt, U., “Momentum in an Uncertain Light,” Nature, Vol. 44, No. 14, 2006,
pp. 823 824.
[28] Buchanan, M., “Minkowski, Abraham and the Photon Momentum,” Nature Physics,
Vol. 3, Feb. 2007, p. 73.
370 M. R. LAPOINTE

[29] Pfiefer, R. N., Nieminen, T. A., Heckenberg, N. R., and Rubinsztein Dunlop, H.,
“Two Controversies in Classical Electromagnetism,” Proceedings of the SPIE:
Optical Trapping and Optical Micromanipulation III, Dholakia K., and Spalding,
G. (eds.), Vol. 6326, 2006, 62260H(10).
[30] Slepian, J., “Electrical Essay: Electromagnetic Space Ship,” Electrical Engineering,
Feb. 1949, pp. 145 146.
[31] Slepian, J., “Answer to Previous Essay,” Electrical Engineering, Mar. 1949,
p. 245.
[32] Corum, J. F., Dering, J. P., Pesavento, P., and Donne, A., “EM Stress Tensor Space
Drive,” Space Technologies and Applications International Forum 1999, El Genk,
M. S. (ed.), American Institute of Physics Conference Proceedings CP458,
Springer Verlag, New York, 1999, pp. 1027 1033.
[33] Corum, J. F., Keech, T. D., Kapin, S. A., Gray, D. A., Pesavento, P. V., Duncan,
M. S., and Spadaro, J. F., “The Electromagnetic Stress Tensor as a Possible Space
Drive Propulsion Concept,” 37th AIAA/ASME/SAE/ASEE Joint Propulsion
Conference, AIAA Paper 2001 3654, Salt Lake City, UT, July 2001.
[34] Heaviside, O., Electromagnetic Theory: Volumes 1, 2, and 3, Dover Pub., New York,
1950.
[35] Hartley, R., “Oscillations in Systems with Nonlinear Reactance,” Bell Systems
Telephone Journal, Vol. 15, No. 3, 1936, pp. 424 440.
[36] Manley, J. M., and Rowe, H. E., “Some General Properties of Nonlinear Elements
Part I: General Energy Relations,” Proceedings of the Institute Radio Engineering,
Vol. 44, No. 7, 1956, pp. 904 913.
[37] Rowe, H. E., “Some General Properties of Nonlinear Elements Part II:
Small Signal Theory,” Proceedings of the Institute Radio Engineering, Vol. 46,
No. 6, 1958, pp. 850 856.
[38] Lahoz, D. G., and Graham, G. M., “Experimental Decision on the Electromagnetic
Momentum Expression for Magnetic Media,” Journal of Physics A, Vol., 15,
1982, pp. 303 318.
[39] Seigenthaler, K. E., and Lawrence, T. J., “Breakthrough Propulsion Physics Research
at the United States Air Force Academy,” 42nd AIAA/ASME/SAE/ASEE Joint Pro
pulsion Conference, AIAA Paper 2006 4912, Sacramento, CA, July 2006.
[40] Eguinlian, A., and Gallo, B., “Critical Analysis of the Results Presented by
the United States Air Force Academy in Report USAFA TR 2003 03,” Instituto
Universitario Aeronautico Technical Report DMA 008/04, Cordoba, Argentina,
Mar. 2004.
[41] Brito, H. H., and Elaskar, S. A., “Overview of Theories and Experiments on Electro
magnetic Inertia Manipulation Propulsion,” Space Technologies and Applications
International Forum 2005, El Genk, M. S. (ed.), American Institute of Physics
Conference Proceedings CP746, Springer Verlag, New York, 2005, pp. 1395 1402.
[42] Brito, H. H., “A Propulsion Mass Tensor Coupling in Relativistic Rocket Motion,”
Space Technologies and Applications International Forum 1998, El Genk, M. S.
(ed.), American Institute of Physics Conference Proceedings CP420, Springer
Verlag, New York, 1998, pp. 1509 1515.
[43] Brito, H. H., “Propellantless Propulsion by Electromagnetic Inertia Manipulation:
Theory and Experiment,” Space Technologies and Applications International
Forum 1999, El Genk, M. S. (ed.), American Institute of Physics Conference
Proceedings CP458, Springer Verlag, New York, 1999, pp. 994 1004.
 PROPULSIVE IMPLICATIONS OF PHOTON MOMENTUM 371

[44] Brito, H. H., “Candidate In Orbit Experiment to Test the Electromagnetic Inertia
Manipulation Concept,” Space Technologies and Applications International
Forum 2000, edited by El Genk, M. S. (ed.), American Institute of Physics
Conference Proceedings CP458, Springer Verlag, New York, 2000, pp. 1032 1038.
[45] Brito, H. H., “Research on Achieving Thrust by EM Inertia Manipulation,” 37th
AIAA/ASME/SAE/ASEE Joint Propulsion Conference, AIAA Paper 2001 3656,
Salt Lake City, UT, July 2001.
[46] Brito, H. H., and Elaskar, S. A., “Direct Experimental Evidence of Electromagnetic
Inertia Manipulation Thrusting,” Journal of Propulsion and Power, Vol. 23, No. 2,
Mar. Apr. 2007, pp. 487 494; also published as 39th AIAA/ASME/SAE/ASEE
Joint Propulsion Conference, AIAA Paper 2003 4989, Huntsville, AL, Jul. 2003.
[47] Furry, W. H., “Examples of Momentum Distributions in the Electromagnetic Field
and in Matter,” American Journal Physics, Vol. 37, No. 6, 1969, pp. 621 636.
[48] Feigel, A., “Quantum Vacuum Contribution to the Momentum of Dielectric Media,”
Physical Review Letters, Vol. 92, No. 2, 2004, p. 020404(4).
[49] Schutzhold, R., and Plunien, G., “Comment on ‘Quantum Vacuum Contribution to
the Momentum of Dielectric Media,’” Physical Review Letters, Vol. 93, No. 26,
2004, 268901(1); Feigel replies in Physical Review Letters, Vol. 93, No. 26, 2004,
p. 268902(1).
[50] van Tiggelen, B. A., and Rikken, G. L., “Comment on ‘Quantum Vacuum
Contribution to the Momentum of Dielectric Media,’” Physical Review Letters,
Vol. 93, No. 26, 2004, p. 268903(1); Feigel replies in Physical Review Letters,
Vol. 93, No. 26, 2004, p. 268904(1).
[51] van Tiggelen, B. A., Rikken, G. L., and Krstic, V., “Momentum Transfer from
Quantum Vacuum to Magetoelectric Matter,” Physical Review Letters, Vol. 96,
No. 13, 2006, p. 130402(4).
[52] van Tiggelen, B. A., Rikken, G. L., and Krstic, V., “The Feigel Process: Lorentz
Invariance, Regularization, and Experimental Feasibility. Final Report of a Two
Month Study Funded by the ESA Advanced Concepts Team,” European Space
Agency Internal Report, Ariadna Grant No. 18806/04/NL/MV, 2005.
 Chapter 11

Experimental Results of the Woodward Effect

on a Micro-Newton Thrust Balance

Nembo Buldrini and Martin Tajmar†

Austrian Research Centers GmbH ARC, Seibersdorf, Austria

I. Introduction
HE need for efficient space propulsion systems has stimulated
T researchers to investigate new pathways in modern physics for a theory
or a phenomenon that could be exploited for this purpose. Nevertheless,
even if some intriguing developments exist in this field, there is lack of experi-
mental evidence: either the experiment gives inconclusive results, or the
required test is not affordable because of our limited current technological capa-
bility. In any case, there are no wholly accepted experimental proofs that demon-
strate such advanced propulsion systems are feasible. In 1990, James
F. Woodward [1] started to get interesting results in his experiments to demon-
strate that the inertia of a body is the result of Mach’s principle; that is, it
arises from the gravitational interaction with distant matter in the universe.
Since then, he and other investigators have been obtaining remarkable results
that are giving, step-by-step, a more complete view of the phenomenon,
suggesting that what is detected is a genuine effect. In this paper, we tested
some of these devices on a mN thrust balance in a large vacuum chamber. Our
results are in general about an order of magnitude below Woodward’s past
claims. In one configuration, a net thrust effect was observed but still needs
further investigation.

Copyright # 2008 by Austrian Research Centers GmbH ARC. Published by the American
Institute of Aeronautics and Astronautics, Inc., with permission.

Research Scientist, Space Propulsion and Advanced Concepts.
†
Head, Space Propulsion and Advanced Concepts.

373
374 N. BULDRINI AND M. TAJMAR

II. Theoretical Considerations

The focus of this paper is on experimental evaluation; therefore, we will
present just a brief description of the theoretical background, encouraging the
reader to consider the full theoretical aspects that have already been extensively
and rigorously treated [1 10]. In 1953, Dennis Sciama showed that inertial
reaction forces can be viewed as the reaction to the radiative action of chiefly
distant cosmic matter on accelerating local objects [11]. From the case of electro-
dynamics we know that radiation reaction includes third, as well as second, time-
derivatives in the equations of motion, and therefore some peculiar phenomena
not encountered in normal mechanics can take place. Starting from this assump-
tion, Woodward derived a series of equations that show what happens to the
mass-energy of an object “immersed” in the gravitational (or inertial) field of
cosmic matter when it is accelerated by some external force. Interestingly, it
turns out that the expression for the local source of the inertial (or gravitational)
field has time-dependent terms that can be engineered to be surprisingly large.
They are, however, only non-zero for sources that change their state of internal
energy as they are accelerated.
Following the formalism adopted in Ref. 5, proper matter density fluctuation
arising from Machian “gravinertial” interactions may be written as:
"     #
1 1 @2 E0 1 2 @E0 2
dr0 (t)   (1)
4pG r0 c2 @t2 r0 c 2 @t

where r0 is body density (kg . m 3), E0 is the internal energy of the body (J), and
G is Newton’s gravitational constant ( 6.67 . 10 11 m3 . kg 1 . s 2). Capacitors
excited by an alternating voltage are one possible system where large, rapid
fluctuations in Eo can easily be affected and should produce periodic mass fluctu-
ations dm0 that might be used to generate stationary forces. Mass fluctuation is just
the integral of dr0(t) over the volume of the capacitor, and the corresponding
integral of the time derivatives of E0, because @E0/@t is the power density will be:
"   #
1 1 @P 1 2 P2
dm0   (2)
4pG r0 c2 @ t r0 c 2 V

where P is the instantaneous power delivered to the capacitor and V its volume. The
predicted mass fluctuation can be computed using Eq. (2) which, after
differentiation of P P0 sin(2vt) and ignoring the second term on the right-
hand side, reads:
vP0
dm0 (t)  cos (2vt) (3)
2pGr0 c2
A simple way to get unidirectional force is to make the capacitor oscillate at a rate
that it is accelerated in one direction when the mass undergoes a positive fluctu-
ation, and in the other direction when the fluctuation is negative. Several devices
have been built that used piezoelectric materials to exert an oscillating force on a
capacitor subject to mass fluctuations [2,6,7]. A more reliable and convenient
 WOODWARD EFFECT ON MICRO-NEWTON THRUST BALANCE 375

method to apply this force is the use of a magnetic field perpendicular to the dis-
placement current in the capacitor [3 5,8,10,11]. If the phase between the
current in the coil and the voltage applied to the capacitor is 90 degrees, then a
Lorentz force from the displacement current and the field of the coil will act on
the dielectric with the correct timing to give rise to a unidirectional force. This
can be described with the following formalism:

FB ¼ i d  B  L (4)

where id is the displacement current in the capacitor (A), B the magnetic flux (T),
and L the distance between the capacitor plates (m). Considering that id and B are
orthogonal by construction and have the same frequency, the force may be
expressed as follows:

FB  Bid L cos w (5)

The total force acting on the supports of the device is the sum of the Lorentz
force and the counterbalancing lattice forces:

Ftot ¼ (FB þ Flat ) (6)

which, in the absence of mass fluctuations in the dielectric, turns out to be zero.
However, when mass fluctuations are taken into account, the time-average of Ftot
no longer vanishes in stationary circumstances if the phase relationship among
FB, id, and dmo is such that FB acts in phase with the mass fluctuation; thus we
can write:
 
dm 0
kFtot l  2 FB sin w (7)
m0

where the phase angle w is the one between the voltage applied to the capacitor and
the current in the inductor. The factor of two arises because the mass fluctuation
peaks with reversed sign when the lattice restoring forces act during each cycle.
Although several experiments have claimed to have detected the existence of
this mass fluctuation effect, the agreement of the results with the theory is not yet
completely fulfilled. Usually the predictions underestimate observations by a
factor of two to several hundreds, depending on the theoretical model adopted.
This mismatching can arise from several factors, such as the interactions at
atomic level involved in the production of such mass fluctuations that could
reveal themselves to be more complex than expected. Another theoretical
approach has been proposed by H. Brito and S. A. Elaskar based on purely elec-
tromagnetic considerations [12,13]. Electromagnetic fields possess momentum
carried by the crossed E and B fields. Although it could be shown that a torque
can be generated by such fields [14], the generation of linear momentum is
highly controversial and called “hidden momentum” throughout the literature
[15] (see also Chapter 10). Brito and Elaskar followed the “hidden momentum”
hypothesis and built a thruster very similar to Woodward’s device. Their theory
predicted higher thrust values compared to Woodward’s Machian approach.
376 N. BULDRINI AND M. TAJMAR

III. Thrust Balance

The balance used in this work [16] has been especially developed to measure
the thrust produced by In-FEEP (indium field emission electric propulsion)
thrusters, designed and manufactured at the Austrian Research Centers.
The force produced by a single thruster of this sort falls in the micro-Newton
range, reaching milli-Newtons for clustered systems.
The balance consists of a symmetric arm (a hollow rectangular aluminum
profile with a side length of 20 mm and a total length of about 60 cm), which
is free to rotate by means of two flexural pivots. The pivots have been selected
because of their low friction, high linearity, and negligible hysteresis. In addition
they are able to support large loads (in the order of a few kilograms). The selected
flexural pivot is a G-10 pivot from C-Flex, Inc. When the thruster, mounted to one
end of the arm, is switched on, it will exert a force on the arm, causing it to shift
its position. The deflection of the balance is a linear function of the applied force.
Additionally, the measurable thrust range is adjustable, changing either the pos-
ition of the sensor or the thruster (i.e., the distance to the center of rotation).
Moreover, the flexural pivot can be substituted by a pivot with a different tor-
sional spring rate.
Different sensors have been evaluated to measure the deflection of the thrust
balance. The sensor selected is a D64 fiber optic displacement sensor (Philtec,
Inc.), which works by measuring the reflection of light from the target surface.
A big advantage of this sensor is that the part inside the vacuum chamber consists
only of optical fibers, thus the signal cannot be influenced by electromagnetic
interference (Fig. 1).
The typical response curve of the sensor allows the operation of the sensor in
either the so-called “near side” or “far side.” The near side offers a higher sensi-
tivity at the cost of a decreased linear measurement range (this limits the measur-
able thrust range and makes the initial setup of the balance more complicated).
The far side has a lesser sensitivity but a much larger measurement range.

Fig. 1 The vacuum chamber and thrust balance mounted inside the chamber for an
In-FEEP test.
 WOODWARD EFFECT ON MICRO-NEWTON THRUST BALANCE 377

In order to simplify the initial setup of the balance and to enable use of both oper-
ating modes of the sensor, a movable sensor mount was designed. A vacuum-
compatible stepper motor driving a high precision stage allows the movement
of the sensor head. By following the response curve of the sensor one can set
the working point either in the near or far side of the sensor.
Because the flexural pivot used has low friction, the motion of the balance is
almost without damping. To reduce oscillations of the arm, a damping system had
to be included in the design. Possibilities for this included oil dampers,
electromagnetic dampers, and damping by electrostatic forces. The damping
system selected for the balance was the latter. The signal from the displacement
sensor is recorded into the computer and is differentiated, amplified, and fed into
the voltage command input of a high-voltage power supply connected to a special
comb assembly. This works as a velocity proportional damping system. The level
of damping can be conveniently set in the control software.
A typical response of the balance is shown in Fig. 2. When a calibration force
of 25 mN is applied by the electrostatic comb actuator, the arm moves to a new
position. The oscillations are gradually reduced by the damping system. After
switching off the voltage supply for the electrodes, the balance returns to its
original position. Note that when the calibration force is switched on again, the
balance is damped more strongly because the level of damping has been
changed in the control software. This is not the highest possible level,
however; further tests are needed to show the optimum damping. In general
it is preferred to reach a steady state as quickly as possible to maximize
measuring time.
Figure 3 displays the response of the balance to the force produced by a real
In-FEEP thruster. The black trace represents the thrust calculated using the
electrical values of the thruster; in this case the force was about 10 mN. In this

Fig. 2 Response of the balance to a calibration force.

378 N. BULDRINI AND M. TAJMAR

Fig. 3 Balance response to In-FEEP firing.

run the displacement sensor has been used only in the range with less sensitivity
to simplify the setup of the balance in the vacuum chamber. In the high sensitivity
range, resolutions of less than 1 mN are achievable.

IV. Setup
Figure 4 shows the typical balance assembly used to test mass fluctuation
devices. This is a preliminary layout where all the electrical connections are

Fig. 4 Mach-5C mass fluctuation device (right) mounted on the balance.

 WOODWARD EFFECT ON MICRO-NEWTON THRUST BALANCE 379

Fig. 5 Radial outgassing slit.

not yet implemented. The counterweight is located on the left-hand side of the
beam. The white Teflon blocks next to it constitute part of the damping system
actuator. The device to be tested is mounted on the other side of the balance
arm. The article is enclosed in an aluminum housing (122-mm external diameter,
54-mm height) that provides EM radiation shielding and mechanical coupling
with the thrust balance. Between the top cover and the main housing, a space
of several millimeters is retained (Fig. 5) to allow for proper outgassing of
the device (the tests are performed at a pressure of 10 6 mbar). This outgassing
slit solution was selected to avoid other problems during the test of the article
as well: an abrupt outgassing episode of some extent is expected every time
the device is connected to the power supply (as result of the increase of temp-
erature), and spurious thrust signals resulting from gas emission are probable
and must be avoided. The slit provides a radial outlet, so that even if the outgas-
sing is not uniformly distributed over the rim, the disturbance will be minimized
as the thrust vector will be perpendicular to the sensitive thrust axis of the
balance.
The devices are driven by two Carvin DCM2500 audio amplifiers, a model
similar to those used by Woodward [3] and March [8], in order to provide a
true replication of the original experiment. One amplifier drives the coil and
the other one the capacitors. Stepup transformers are needed to elevate the
voltage level from the output of the amplifiers. For the tests with Mach-6C
and its modified version Mach-6CP (see experimental results), an additional
380 N. BULDRINI AND M. TAJMAR

Fig. 6 Experiment setup with independent capacitor/coil driving.

pre-tuning circuit has been inserted in the capacitor line, between the amplifier
and the stepup transformers. This circuit permits higher capacitor voltages,
which should translate in higher thrust outputs. The signal generator used to
pilot the amplifiers is a two-channel wave synthesis generator connected via
USB to the computer. It is able to generate various waveforms with a
maximum frequency of 200 kHz. The phase shift of each signal can be adjusted
via software from 08 to 1808 continuously. Having the two signals supplied inde-
pendently makes it possible to check the dependence of the thrust produced on
the relative phase shift, thus eliminating some possible spurious results. Much
attention has to be paid in feeding the audio frequency power to the device.
The cables have to be robust enough to carry high power levels. Consequently,
mechanical interference with the balance cannot be excluded. The experiment
setup is shown in Fig. 6.
Dedicated Labview software is used to manage the operation of the device
and the acquisition of the data. For each shot, the start and duration of both
signals can be set independently.

V. Results
Three different devices, two of them (named Mach-5C and Mach-6C)
were provided and already tested by Woodward himself and have been character-
ized. All these devices are based on Eq. (7), where the unidirectional force is sup-
posed to be achieved rectifying the mass fluctuation of the capacitor dielectric by
means of the Lorentz force. The general structure of these devices consists of two
or more capacitors surrounded by a coil, which generates a magnetic field perpen-
dicular to the electric field established in the capacitors’ dielectric.
 WOODWARD EFFECT ON MICRO-NEWTON THRUST BALANCE 381

A. Mach-5C
The device named Mach-5C is comprised of two 2.2 nF 15 kV Vishay Cera-
Mite capacitors (Z5U dielectric, barium titanate, 1r  8,500) glued between the
halves of a toroidal inductor core and connected in parallel (measured total
capacitance, 4.95 nF). The magnetic flux through the capacitors needed to
produce thrust is generated by two coils wound over the capacitors. The measured
inductance of the two coils connected in series is 680 mH. In previous measure-
ment sessions with this device, Woodward and Vandeventer [10] recorded a
thrust of about 50 mN.
Since the first runs with the Mach-5C device, it was obvious that a pronounced
thermal drift signature was present in the arm displacement plot (Fig. 7), recog-
nizable from the fact that the trace does not return promptly back toward zero,
as should occur when the device is not energized. Further investigation identified
that the cause of the drift was due to a thermomechanical bending of the power
supplying wires that go through the balance to the device. Nevertheless, even if
such a spurious effect is present, it would not be difficult to see a real thrust signal
superimposed on it.
Figure 8 shows the comparison between two signals: one obtained with a 908
phase shift between the capacitor voltage and the coil current, which should
correspond to the maximum thrust production, and the other obtained with a
1808 phase shift, which should result in a zero thrust trace, per Eq. (7). Mach-
5C was energized with 1.6 kV to the capacitors and 2.2 A to the coil; the
frequency of the driving signal was 43 kHz. The coil was energized for 4
seconds starting at 115 sec in the plot. After the initial second, the capacitor
was energized for 4 sec. The total sequence lasted 5 seconds. The predicted
thrust at 908 was 5 mN and not 50 mN, because the voltage going to the capacitors
was 1.6 kV and not 3.5 kV, and the frequency was 43 kHz instead of 60 kHz as

Fig. 7 Typical trace obtained with the Mach-5C device setup.

382 N. BULDRINI AND M. TAJMAR

Fig. 8 908 and 1808 phase shift superimposed plots.

used by Woodward in his 50-mN test. From the graph it is possible to see a
small difference of several mN between the curves, but it is difficult to say if it
is due to a genuine effect, as the initial thermal conditions were not monitored.
That could lead to different thermal behavior of the wires, confusing the
actual results.
Using the lower frequency configuration we can safely say that the thrust
produced from the Mach-5C device must be less than 2 mN compared to the
5 mN extrapolated from Woodward’s results. However, we have to note that,
due to the heating of the capacitors, the phase between coil and capacitor is
changing so that if a Machian thrust is present, the maximum time that it acts
on the balance is only tenths of a second. Still, a trace on the balance should
be possible to see, as we will see in a later section where we tested the balance
response with short calibration pulses. As the 2-mN thrust limit is, however,
close to the 5 mN estimate, at best we can say that the results are inconclusive
from this experiment.

B. Mach-6C
More controllable conditions were achieved with the Mach-6C device (Fig. 9),
also built by Woodward, who equipped it with an embedded thermistor. Tests by
Woodward claim a thrust generated with this device in the 150-mN thrust range
[17]. Monitoring the temperature allows operation of the device far away from
the temperature range where the capacitor dielectric can get damaged; moreover,
it permits to begin each test with the same starting conditions. The Mach-6C
assembly comprises eight capacitors (Y5U dielectric, barium titanate,
1r  5000), connected in parallel (measured total capacitance, 4.20 nF) and a
coil wound in toroidal configuration over them (measured inductance,
 WOODWARD EFFECT ON MICRO-NEWTON THRUST BALANCE 383

Fig. 9 Mach-6C.

1.17 mH). A test from Woodward assessed that with Mach-6C, it was possible to
reach higher voltages (and thus higher thrusts) without incurring dielectric aging,
a phenomenon that was noted to reduce the performance of the “5” series. We
upgraded the driver setup so that it was possible to reach the required higher vol-
tages values at the capacitor leads (in excess of 3.2 kVp). The operation fre-
quency ranged between 50 kHz and 55 kHz. Figure 10a depicts the typical
trace of one Mach-6C run. The thermal drift has been virtually eliminated by
rearranging the wiring and reducing the firing time to about 2 sec. It appears
evident that no appreciable thrust signal is present in the balance arm

Fig. 10 Mach-6C measurements.

384 N. BULDRINI AND M. TAJMAR

displacement trace within the balance resolution of 1 mN. Compared with Wood-
ward’s results of 150 mN, we should have definitely seen a trace.
In order to check if the balance was working under the same conditions
as described above, Fig. 10b shows a trace achieved with similar working
parameters but with a 50 mN superimposed calibration pulse generated indepen-
dently by the electrostatic actuator (the duration of the pulse coincides with the
running time of the device). The idea behind this test was to show that, if there
were actually a propulsive effect coming from the device, it would not be
hidden by something occurring during the firing time (counter effects from the
wires, EMI, etc.). One problem encountered during experimentation with Mach
devices was the difficulty to keep the phase relationship stable between the
capacitor voltage and the coil current. Even if the firing time is very short, in
fact, the power levels supplied to the device are relatively high and this causes
the heating up of the several parts. As mentioned in the previous section,
heating directly translates into a phase shift of the applied voltage to the capaci-
tors relative to the phase of the current flowing into the coil. Furthermore, the
voltage applied to the capacitor varies as we are receding from the condition
of resonance in the circuit comprising the capacitor itself and the driving circui-
try. That said, we can assume that ideal thrust conditions only occur during the
first tenths of a second. Therefore, in order to evaluate the behavior of our
balance at different thrust duration values, a series of short pulses was generated
using the calibration actuator, as shown in Fig. 11. The first pulse is the usual cali-
bration and is performed before every device run to provide the actual scaling
factor. It consist of a single pulse (usually corresponding to a thrust of 50 mN)
generated while keeping the actuator switched on for the time needed by the
damping system to bring the balance arm to a static position. The pulses in
Fig. 11 were generated with a software-controlled timing procedure. From this

Fig. 11 Balance response to short pulses.

 WOODWARD EFFECT ON MICRO-NEWTON THRUST BALANCE 385

graph we can observe that it is possible to easily discriminate 50 mN pulses as

short as 0.2 sec.
In summary, even short pulses of 50 mN and 100 mN should have been clearly
identifiable with thrust peaks of up to 30 mN. In our experimental setup we can
say that if the Machian thrust effect is present, it must be at least one order of
magnitude less than in previous claims.

C. Mach-6CP
The Mach-6CP was tested in a different configuration with the power cables
fed through the outgassing slit instead of the bottom of the housing as for the
Mach-6C tests (Fig. 12). As we will see, that could have caused an interference
with the balance. The graphs in Fig. 13 show the results obtained with a new
version of Mach-6C, also provided by Woodward. The same device has been
potted (thus the “P” in the model number) in epoxy paste and enveloped in a
mild steel housing. This encapsulation provides an additional shield against
electromagnetic emission and eliminates the occurring of a possible corona
effect on the capacitor leads. It also decreases the cooling time between one
run and the next. A thrust of 100 mN to 200 mN was recorded in Woodward’s
facilities (in vacuum).
One standard procedure to discriminate an authentic effect from a false posi-
tive is to run the coil and the capacitors separately. If the effect is genuine, in fact,
it has to be present only when the coil and capacitors are working together, so that
the magnetic field generated by the coil can act on the displacing ions in the

Fig. 12 Mach-6CP.
386 N. BULDRINI AND M. TAJMAR

Fig. 13 Mach-6CP measurements.

capacitor dielectric lattice, producing the Lorentz force necessary to rectify the
mass fluctuations and produce net thrust. Figure 13 shows the comparison
between a shot with the capacitors and coil together, and a shot with only the
capacitors in operation. A thrust signature continues to be recorded when the
capacitors are working alone. The real cause of this pulse still must be assessed.
One possibility could be the different arrangement of the feeding cables (in the
“P” version, they have to exit the housing from the outgassing slit), together
with an inadequacy of their electromagnetic shielding. In fact, if some electric
fields are leaking out, they can interact with the fixed portions of the balance
or with the vacuum chamber wall, giving rise to a net force on the device
during operation time.
Actually, some electromagnetic field measurements were pursued (with the
device operating in air) and a leak in the cable shielding near to the aluminum
housing was found: an electric field strength of 2.5 kV/m was detected several
centimeters away from the housing, that can perhaps account for the observed
effect. The wires were then shielded properly so that a field strength of just 59
V/m was measured. Unfortunately, it has not been possible to date to
perform experiments with this last modified version, so we cannot yet confirm
the real nature of the effect. There is another explanation that can account for
the thrust effect recorded: because the capacitor leads are asymmetric, they
generate a magnetic field that can interact with the dielectric of the
capacitor itself and generate a thrust according to the mass fluctuation effect.
However, the fact that the thrust magnitude is not varying between the two
circumstances (in both cases the maximum swing of the balance arm is
indicating about 22 mN) suggests that the origin of the recorded effect is not of
Machian nature. Otherwise the magnetic field generated by the coil would
have added to the produced thrust and a difference between the plots should
have been seen.

D. 2-MHz Breadboard Device

A breadboard device has been built and tested here in Austrian Research
Centers to explore force production at higher frequencies and using different
 WOODWARD EFFECT ON MICRO-NEWTON THRUST BALANCE 387

dielectric materials. It consists of high-power ceramic capacitors, designed for

high frequency duty (dielectric: titanium dioxide, 1r  120) and a coil connected
in series, driven at resonance (resonance frequency: 2 MHz) by a compact class-E
type amplifier. Although the dielectric relative permittivity of titanium dioxide is
quite low compared to the materials employed in the capacitors of the Mach-5C/
6C devices, the dissipation factor is one order of magnitude lower, thus allowing
their use at higher frequency without incurring severe overheating. Driving the
coil and the capacitor in a series (or parallel) configuration has two desirable
advantages:
1) The voltage on the capacitor and the current in the coil always have a
relative phase shift of 908, thus, via Eq. (7), maximum thrust is obtainable.
2) At resonance, high voltage (or current) values are built up inside the
device leading to high-reactive (not dissipative) power levels on the order of
several kVA with a relatively low driving power, related only to the dissipation
factor of the circuit.
The device, as shown in Fig. 14, was operated in air on a commercial elec-
tronic balance (Sartorius MC-1). Voltages in excess of 2 kVp were obtained
with currents on the order of 4 to 5 A. An expected thrust varying from 1 mN
to 6 mN was calculated, depending on the theoretical approach used [9], but
no real thrust was detected within the sensibility of the used electronic balance
(0.1 mN). Allowing for the Machian effects to be a real phenomenon, a possible
explanation of the failure of this device could be that the dielectric used and its
geometry (very thin layer) is not suitable for producing mass fluctuations.
Accordingly, a further investigation using different type of capacitors seems to
be reasonable.

Fig. 14 2-MHz breadboard device.

388 N. BULDRINI AND M. TAJMAR

The kind of device here described appears to be an optimal breadboard to

continue Machian mass fluctuation testing because of its insensitiveness to
capacitor/coil phase shifts and its very compact layout. This last feature makes
this design suitable for self-contained setup testing, where device, amplifier,
and power source (battery) are placed together on the thrust-measuring apparatus,
thus eliminating spurious signals coming from the feeding cables.

VI. Conclusions
The aim of the reported experimental activity was to evaluate the propulsive
capabilities of devices based on the Machian mass fluctuation effect. Two test
devices, built and already tested by Woodward, have been characterized using
a highly sensitive mN thrust balance usually employed for electric ion propulsion.
Our results are, for the most part, not in agreement with the data collected by
Woodward, even if some interesting thrust plots were recorded during the last
test with the Mach-6CP device that suggest further investigation be pursued. In
general, we find that any Machian effect eventually produced by such devices
is likely more than an order of magnitude below Woodward’s past claims.
A device operating at higher frequencies and with different capacitors has
been designed and built at the Austrian Research Centers. No thrust has been
detected within the sensibility of the electronic balance used, although the
minimum thrust measurable with this instrument was well below the magnitude
predicted by the theoretical models. However, it must be taken into account that
the geometry and the dielectric material of the capacitors may influence the
expression of the effect in a way that is still not clear.
Recently, Woodward built a thrust balance similar to the one at the Austrian
Research Centers and initial tests seem to be in general agreement with our analy-
sis of the magnitude of the effect [17], though Woodward claims to see a small,
real effect. The thrust values of Mach-6CP-like devices are presently under inves-
tigation on this thrust balance.

Acknowledgments
The main credit goes to J. Woodward. Without his contributions and the con-
tinuing inspiration of his work during the last 10 years, this paper wouldn’t have
been realized. Many thanks go out to P. March, A. Palfreyman, P. Vandeventer,
T. Mahood, and other colleagues for providing useful discussions. We also
acknowledge the help of K. Marhold and B. Seifert for designing the original
mN thrust balance.

References
[1] Woodward, J. F., “A New Experimental Approach to Mach’s Principle and Relativis
tic Gravitation,” Foundation of Physics Letters, Vol. 3, No. 5, 1990, pp. 497 506.
[2] Woodward, J. F., “A Laboratory Test of Mach’s Principle and Strong Field Relati
vistic Gravity,” Foundations of Physics Letters, Vol. 9, No. 3, 1996, pp. 247 293.
[3] Woodward, J. F., “Flux Capacitors and the Origin of Inertia,” Foundations of
Physics, Vol. 34, No. 10, 2004, pp. 1475 1514.
 WOODWARD EFFECT ON MICRO-NEWTON THRUST BALANCE 389

[4] Woodward, J. F., “Life Imitating ‘Art’: Flux Capacitors and Our Future in Space
time,” Proceedings of Space Technology and Applications International Forum
(STAIF 2004), El Genk, M. S. (ed.), AIP Conference Proceedings 699, Melville,
NY, 2004, pp. 1127 1137.
[5] Woodward, J. F., “Tweaking Flux Capacitors,” Proceedings of Space Technology
and Applications International Forum (STAIF 2005), El Genk, M. S. (ed.), AIP
Conference 746, Melville, NY, 2005, pp. 1345 1352.
[6] Mahood, T. L., “A Torsion Pendulum Investigation of Transient Machian Effects,”
M.S. Thesis, California State University Fullerton (CSUF), 1999.
[7] Mahood, T. L., March, P., and Woodward, J. F., “Rapid Spacetime Transport and
Machian Mass Fluctuations: Theory and Experiments,” Proceedings of 37th
AIAA/ASME/SAE Joint Propulsion Conference, AIAA Paper 2001 3907, 9 July
2001.
[8] March, P., “Woodward Effect Experimental Verifications,” Space Technology and
Applications International Forum (STAIF 2004), El Genk, M. S. (ed.), AIP
Conference Proceedings 699, Melville, NY, 2004, pp. 1138 1145.
[9] March, P., and Palfreyman, A., “The Woodward Effect: Math Modeling and Contin
ued Experimental Verifications at 2 to 4 MHz,” Proceedings of Space Technology
and Applications International Forum (STAIF 2006), El Genk, M. S. (ed.), Ameri
can Institute of Physics, Melville, NY, 2006.
[10] Woodward, J. F., and Vandeventer, P., “Mach’s Principle, Flux Capacitors, and
Propulsion,” Proceedings of Space Technology and Applications International
Forum (STAIF 2006), El Genk, M. S. (ed.), American Institute of Physics, Melville,
NY, 2006.
[11] Sciama, D., “On the Origin of Inertia,” Monthly Notices of the Royal Astronomical
Society, 113, 1953, pp. 34 42.
[12] Brito, H. H., and Elaskar, S. A., “Direct Experimental Evidence of Electromagnetic
Inertia Manipulation,” 39th AIAA/ASME/SAE/ASEE Joint Propulsion Conference
and Exhibit, AIAA Paper 2003 4989, Huntsville, AB, July 2003.
[13] Brito, H. H., and Elaskar, S. A., “Overview of Theories and Experiments on Electro
magnetic Inertia Manipulation Propulsion,” Proceedings of Space Technology and
Applications International Forum (STAIF 2005), El Genk, M. S. (ed.), AIP Confer
ence Proceedings 746, Melville, NY, 2005, pp. 1395 1402.
[14] Graham, G. H., and Lahoz, D. G., “Observation of Static Electromagnetic Angular
Momentum in Vacuo,” Nature, Vol. 285, 1980, pp. 154 155.
[15] Hnizdo, V., “Hidden Momentum and the Electromagnetic Mass of a Charge and
Current Carrying Body,” American Journal of Physics, Vol. 65, No. 1, 1997,
pp. 55 65.
[16] Marhold, K., and Tajmar, M., “Micronewton Thrust Balance for Indium FEEP
Thrusters,” 41st AIAA/ASME/SAE Joint Propulsion Conference, AIAA Paper
2005 4387, Tucson, AZ, July 2005.
[17] Woodward, J. F., “Mach’s Principle and Propulsion: Experimental Results,”
Proceedings of Space Technology and Applications International Forum
(STAIF 2007), El Genk, M. S. (ed.), American Institute of Physics, Melville, NY,
2007.
 Chapter 12

Thrusting Against the Quantum Vacuum

G. Jordan Maclay
Quantum Fields LLC, Richland Center, Wisconsin

I. Introduction
HIS chapter addresses the question of how the properties of the quantum
T vacuum might be exploited to propel a spacecraft. Quantum electrodynamics
(QED), the theory of the interaction of light and matter that has made predictions
verified to 1 in 10 billion [1], predicts that the quantum vacuum, which is the
lowest state of the electromagnetic field, contains a fluctuating virtual photon
field. This fluctuating field is predicted to produce vacuum forces between
nearby surfaces [1 3]. Recently these Casimir forces have been measured and
found to agree with predictions [4 9]. If this virtual radiation pressure could
be utilized for propulsion, the goal of propellantless propulsion would be
achieved. Restrictions due to the conservation of energy and momentum
are discussed. A propulsion system based on an uncharged, conducting mirror
that vibrates asymmetrically in the vacuum is described. By the dynamic
Casimir effect, the mirror produces real photons that impart momentum and
result in a net acceleration. The acceleration is very small, but demonstrates
that the vacuum can be utilized in propulsion. Technological improvements,
some of which are proposed, may be used to increase the accelerating
force. Many questions remain about the supporting theory, and experiments
are needed to probe questions about the quantum vacuum that are far beyond
current theory.
Rockets employing chemical or ionic propellants require the transport of pro-
hibitively large quantities of propellant. If the properties of the quantum vacuum
could somehow be utilized in the production of thrust, that would provide a
decided advantage because the vacuum is everywhere. At this embryonic
stage, in the exceedingly brief history of interstellar spacecraft, we are trying

Copyright # 2008 by G. Jordon Maclay. Published by the American Institute of Aeronautics and
Astronautics, Inc., with permission.

Professor Emeritus, University of Illinois.

391
392 G. J. MACLAY

to distinguish between what appears possible and what appears impossible within
the context of our current understanding of quantum physics and the fundamental
laws of physics, particularly conservation of momentum and energy. Science
fiction writers have written about the use of the quantum vacuum to power space-
craft for decades but no research has validated this suggestion. Arthur C. Clark,
who proposed geosynchronous communications satellites in 1945, described a
“quantum ramjet drive” in 1985 in Songs of Distant Earth, and observed “If
vacuum fluctuations can be harnessed for propulsion by anyone besides
science-fiction writers, the purely engineering problems of interstellar flight
would be solved.Ӡ Australian science fiction writer Ken Ingle described, with
my fanciful suggestions, the Casimir vacuum drive in his soon to be published
book First Contact.
In the last 10 years great progress has been made experimentally in measuring
Casimir forces, which arise between closely spaced surfaces due to the quantum
fluctuations of the electromagnetic field, the quantum vacuum. Although the
forces tend to be small, practical applications of vacuum forces have recently
appeared in microelectromechancial systems (MEMS) devices [10 12].
QED predicts the behavior of the quantum vacuum, including vacuum forces
and the presence of a vast energy in empty space due to a fluctuating electro-
magnetic field. Unfortunately we do not yet have a proven method to propel
a spacecraft by harnessing the vast energy of vacuum fluctuations that QED
predicts, and therefore this chapter focuses on general considerations about
momentum transfer between the quantum vacuum and a spacecraft. The space-
craft proposed in this paper is described as a “gedanken spacecraft” because its
design is intended not as an engineering guide but just to illustrate possibilities.
Indeed, based on our current understanding of quantum vacuum physics, one
could reasonably argue that the gedanken spacecraft could be propelled more
effectively by simply oscillating a charged mirror that would emit electromag-
netic radiation or simply using a flashlight or laser to generate photons.
Although the performance of the vacuum-powered gedanken spacecraft as
presented is disappointing and is no more practical than a spacewarp [13],
the discussion illustrates many important ideas about the quantum vacuum,
and it suggests the potential role of quantum vacuum phenomena in a macro-
scopic system like space travel. In fact, with a breakthrough in materials,
methods, or fundamental understanding, this approach could become practical,
and we might be able to realize the dream of space travel as presented in science
fiction. Physicists have explored various means of locomotion depending on the
density of the medium and the size of the moving object. It would be interesting
to find an optimum method for moving in the quantum vacuum. Unfortunately
we currently have no simple way to mathematically explore these various
simple possibilities.

†
Clarke, A. C., Personal Communication. See the Acknowledgments in The Songs of Distant
Earth. Numerous science fiction writers, including Clarke, Asimov, and Sheffield have based space
craft on the quantum vacuum.
 THRUSTING AGAINST THE QUANTUM VACUUM 393

II. Physics of the Quantum Vacuum

A. Historical Background
Quantum mechanics is one of the great scientific achievements of the twen-
tieth century. It provides models that describe many properties of atoms and mol-
ecules, such as the optical spectra and transition probabilities. In its original form,
as developed by Schrodinger, Heisenberg, Bohr, and others, quantum mechanics
is a nonrelativistic theory that makes the ad hoc assumption that light is emitted
and absorbed by atoms in bundles, called photons. The electromagnetic field,
however, is treated as an ordinary classical field that obeys Maxwell’s relativistic
equations, not as a quantized field. Dirac, Heisenberg, Jordan, Dyson, and others
began formulation of a relativistic form of quantum mechanics, and made efforts
to quantize the electromagnetic field. This quantized field theory of particles and
light theory developed over the next few decades with numerous successful pre-
dictions. In 1948, Willis Lamb tested a crucial prediction of the field theory, that
the 2s and 2p levels of a hydrogen atom would have precisely the same energy.
Lamb sent a beam of hydrogen atoms through a cavity exposed to radio fre-
quency radiation, and determined that the 2s and 2p energy levels were in fact
split by an energy equivalent to 100 MHz [14]. Within days, Hans Bethe of
Cornell realized the problem and published the solution: The theoretical calcu-
lation did not consider the effects of the quantum vacuum on the energy levels
of the hydrogen atoms [15]. This ushered in the modern formulation of QED
of Feynmann, Schwinger, and Tomonaga [1].
In QED, particles and light are both treated as quantized fields that are fully
relativistic. Because the electromagnetic field is quantized, there may be 0, 1,
2, 3 or any number of photons present and, because the fields are relativistic,
they can be readily transformed to coordinate systems that are translated or
moving uniformly (Lorentz transformations). The entire formalism of QED
can be written in tensors that ensure the proper transformation properties of all
observables under a Lorentz transformation.
The pervasive and dynamic role of the vacuum state in QED was unexpected
to many physicists. The lowest state of the quantized electromagnetic field, which
is referred to as the quantum vacuum, was predicted to be filled with photons and
electron positron pairs that appear and disappear continuously and so rapidly
that no direct measurement of their presence is possible. Yet these so called
“virtual particles” affect measurable properties of atoms, such as the energy
levels, magnetic moments, and transition probabilities.
In retrospect, when arguing from nonrelativistic quantum mechanics and the
uncertainly principle, it was clear that the vacuum would contain a fluctuating
electromagnetic field once the field was treated as a quantized field. The field
variables, Ev and Bv, representing the electric and magnetic field at a frequency
v, are directly analogous to Pv, and Qv, the position and momentum of a harmo-
nic oscillator of frequencyv. The ground state of the harmonic oscillator has to
obey Heisenberg’s Uncertainty Principle: DPvDQv  h , where DPv is the uncer-
tainty in the momentum and DQv is the uncertainty in the position. In the lowest
state, the oscillator is still vibrating, with an energy 12h v. If it were not vibrating
but was motionless, then the uncertainty in its momentum would be zero. If we
knew the approximate position of the oscillator, then this state would violate
394 G. J. MACLAY

the Uncertainly
 Principle. The energy of the nth excited state of the oscillator
is n þ 12 h v.
Similarly quantized electric and magnetic fields cannot vanish, but must,
in their lowest state, fluctuate. This isotropic residual fluctuating electromag-
netic field, which is present everywhere at zero Kelvin temperature with all
electromagnetic sources removed, is often called the zero-point electromag-
netic field.
Quantum fluctuations occur in the particle fields as well as the electromagnetic
field, so the quantum vacuum is filled with virtual electron positron pairs, as
well as virtual photons. Before Lamb’s Nobel Prize-winning measurement,
most physicists felt comfortable ignoring the effects of the quantum fluctuations,
assuming that they just shifted the energy but did not have measurable conse-
quences. It turns out that quantum fluctuations affect virtually all physical pro-
cesses, including the mass, charge, and magnetic moment of all particles; the
lifetimes of excited atoms or particles; scattering cross sections; and the
energy levels of atoms. QED, which accounts for all the vacuum processes,
has made experimental predictions of magnetic moments and energy levels
that have been verified by experiment to 1 in 10 billion, the most accurate predic-
tions of any scientific theory [1,16].

B. Energy in the Quantum Vacuum

Zero-point field energy density is a simple and inexorable consequence of
quantum theory and the uncertainty principle, but it brings puzzling inconsisten-
cies with another well-verified theory, General Relativity. The energy in the
quantum vacuum at absolute zero, which is the lowest energy state of the electro-
magnetic field, is due to the presence of virtual photons of energy 12 h vn of all
possible frequencies:

1Xnmax
hv n
E0 ¼ (1)
2 n¼0

Usually a cutoff is used for the high frequencies, such as the frequency corre-
sponding to the Planck length of 10 34m which gives an enormous energy
density (about 10114 J/m3 or, in terms of mass, 1095 g/cm3). From the perspective
of General Relativity, this enormous energy density seems to make no physical
sense, and that is why the effects of the quantum fluctuations were neglected
for decades. Indeed such a large energy would, according to the General
Theory of Relativity, have a disastrous effect on the metric of spacetime. For
an infinite flat universe, this vacuum energy density would imply an outward
zero-point pressure that would rip the universe apart [17]. Astronomical
data, on the other hand, indicate that any such cosmological constant must
be  4 eV/mm3, or 10 29 g/cm3 when expressed as mass [18]. The discrepancy
here between theory and observation is about 120 orders of magnitude, and is
arguably the greatest quantitative discrepancy between theory and observation
in the history of science [19,20]! There are numerous approaches to solve this
 THRUSTING AGAINST THE QUANTUM VACUUM 395

“cosmological constant problem,” such as renormalization, supersymmetry,

string theory, and quintessence, but as yet this remains an unsolved problem.
Gradually, the belief has developed that only changes in the energy density
give observable effects [21].
Each virtual photon of frequency v and wave vector k ~ ðk ¼ 2p=lÞ has
associated with it a momentum h k.~ Because photons are in random directions,
the mean momentum of the vacuum fluctuations vanishes, but just as there are
fluctuations in the electric and magnetic fields consistent with the uncertainty
principle, there are fluctuations in the root mean square momentum. At
finite temperatures, real photons begin to appear in the quantum vacuum,
but their contribution to the total energy is much smaller than that of the
virtual photons.

C. Casimir Forces Predicted in 1948

About the same time as Lamb’s experiment, Heindrick Casimir, director of
research at Phillips Laboratories in the Netherlands, found some disagreements
between the experiment and his model for precipitation of phosphors used in
the manufacture of fluorescent light bulbs. Better agreement between theory
and experiment could be obtained if the van der Waal’s force between two
neutral, polarizable atoms somehow fell off more rapidly at larger distances
than had been supposed. A co-worker suggested that this might be related to
the finite speed of light, which prompted Casimir and Polder to reanalyze the
van der Waals interaction. They found that including the retardation effects
caused the interaction to vary as r 7 rather than r 6 at large intermolecular sep-
arations r, which gave agreement with the experiment.
Intrigued by the simplicity of the result, Casimir sought a deeper understand-
ing. A conversation with Bohr led him to an interpretation in terms of zero-point
energy, and the realization that, by simply considering the changes in vacuum
energy arising from the presence of surfaces in the vacuum, forces due to the
vacuum fluctuation would appear [22]. To understand this result, consider
how inserting two parallel surfaces into the vacuum causes the allowed
modes of the electromagnetic field to change. This change in the modes that
are present occurs since the electromagnetic field must meet the appropriate
boundary conditions at each surface. Thus surfaces alter the modes of oscil-
lation, and therefore the surfaces alter the energy density corresponding to
the lowest state of the electromagnetic field. In actual practice, the modes
with frequencies above the plasma frequency do not appear to be significantly
affected by the metal surfaces because the metal becomes transparent to radi-
ation above this frequency. In order to avoid dealing with infinite quantities,
the usual approach is to compute the finite change DE0 in the energy of the
vacuum due to the presence of the surfaces:

DE0 ¼ E½energy in empty space (2)

 ES ½energy in space with surfaces present
396 G. J. MACLAY

where the definition of each term is given in brackets. This equation can be
expressed as a sum over the corresponding modes:
1Xnmax
hvn  1
X
surfaces
hv0
DE0 ¼ m (3)
2 n¼0 2 m¼0

The quantity DE0 can be computed for various geometries. The forces F due
to the quantum vacuum are obtained by computing the change in the vacuum
energy for a small change in the geometry and differentiating. For example,
consider a hollow conducting rectangular cavity with sides a1, a2, a3. Let
en(a1, a2, a3) be the change in the vacuum energy due to the cavity, then the
force F1 on the side perpendicular to a1 is:

den
F1 ¼  (4)
da1

where den represents the infinitesimal change in energy corresponding to an

infinitesimal change in the dimension da1. This equation also represents the con-
servation of energy when the wall perpendicular to a1 is moved infinitesimally:

den ¼ F1da1 (5)

Thus, if we can calculate the vacuum energy density as a function of the

dimensions of the cavity, we can compute derivatives that give the forces on the
surfaces. For uncharged, perfectly conducting, parallel plates with a very large
area A, very close to each other (separation of d), the tangential component of
the electric field must vanish at the surface, and wavelengths longer than twice
the plate separation d are excluded. With the appropriate boundary conditions,
we can compute the change in vacuum energy and use Eq. (5) to predict an attrac-
tive (or negative) force between the plates:

K
F¼ (6)
d4
where

p 2 hc
K¼ (7)
240
This force F, commonly called the Casimir force, arises from the change in
vacuum energy density Epp from the free field vacuum density that occurs
between the parallel plates [21,23]:

K
E pp ðd Þ ¼ (8)
3d3
Two decades after Casimir’s initial predictions, a method was developed to
compute the Casimir force in terms of the local stress-energy tensor using
 THRUSTING AGAINST THE QUANTUM VACUUM 397

quantum electrodynamics [24]. Many innovations have followed. Vacuum forces

have been computed for other geometries besides the classic parallel plate geome-
try, such as a rectangular cavity, cube, sphere, cylinder, and wedge. For a cube or
sphere, the Casimir forces are outward or repulsive. For a rectangular cavity, the
Casimir forces on the different faces may be inward, outward, or zero depending
on the ratio of the sides. Situations arise in which there are inward forces on some
faces and outward forces on other forces [25]. It is difficult to understand these
unusual results intuitively.
The application of Casimir forces in space propulsion is motivated more
clearly by the interpretation of the parallel plate Casimir force as arising from
radiation pressure, the transfer of momentum from the virtual photons in the
vacuum to the surfaces [3]. It is this virtual radiation pressure that we propose
to explore as a possible driving force to generate net forces on an object, ulti-
mately to propel a spacecraft. There are very significant advantages if it is poss-
ible to use virtual radiation pressure: no propellant may be required, and there is
always something to push.

D. Dynamic Casimir Effect

In the dynamic Casimir effect the parallel plates are imagined to move rapidly,
which can lead to an excited state of the vacuum between the plates, meaning the
creation of real photons [26]. To understand this process from a physical perspec-
tive, imagine that in a real moving conductor, the surface charges must constantly
rearrange themselves to cancel out the transverse electric field at all positions.
This rapid acceleration of charge can lead to radiation. This effect, generally
referred to as the dynamic or adiabatic Casimir effect, has been reviewed but
not yet observed experimentally [21,23,27]. The vacuum field exerts a force on
the moving mirror that tends to damp the motion. Energy conservation requires
the existence of a radiation reaction force working against the motion of the
mirror [28]. This dissipative force may be understood as the mechanical effect
of the emission of radiation induced by the motion of the mirror. This force of
radiation reaction can be used to accelerate the mirror, or a spacecraft attached
to the vibrating mirror, as discussed in Section V.
The Hamiltonian is quadratic in the field operators, and formally analogous to
the Hamiltonian describing photon pair creation by parametric interaction of a
classical pump wave of frequency v0, with a nonlinear medium [29]. Pairs of
photons with frequencies v1 þ v2 v0, are created out of the vacuum state.
Furthermore, the photons have the same polarization, and the components of
the corresponding wave vectors k~1 and k~2 taken along the mirror surface must
add to zero because of the translational symmetry:

k~1  x^ þ k~2  x^ ¼ v1 sin u1 þ v2 sin u2 ¼ 0 (9)

This last equation relates the angles of emission of the photon pairs with
respect to the unit vector x^ , which is normal to the surface. It is interesting that
the photons emitted by the dynamic Casimir effect are entangled photons. This
analysis in terms of the analogous effective Hamiltonian is illuminating but
not complete for perfect mirrors, because no consistent effective Hamiltonian
398 G. J. MACLAY

can be constructed in this case with the idealized and pathological boundary
conditions. More realistic results are obtained assuming that the mirrors are
transparent above a plasma frequency.
The dynamic Casimir effect was studied for a single, perfectly reflecting
mirror with arbitrary nonrelativistic motion and a scalar field in three dimensions
in 1982 by Ford and Vilenkin [30]. They obtained expressions for the vacuum
radiation pressure on the mirror. In 2001, Barton extended the analysis using a
one-dimensional scalar field to a moving body with a finite refractive index
[31]. The vacuum radiation pressure and the radiated spectrum for a nonrelativis-
tic, perfectly reflecting, infinite, plane mirror was computed by Neto and
Machado for the electromagnetic field in three dimensions, and shown to obey
the fluctuation-dissipation theorem from linear response theory [28,32]. This
theorem shows the fluctuations for stationary body yield information about the
mean force experienced by the body in nonuniform motion. Jaekel and
Reynaud computed shifts in the mass of the mirror for a scalar field in two dimen-
sions [33]. The mirror mass is not constant, but rightfully a quantum variable
because of the coupling of the mirror to the fields by the radiation pressure. A
detailed analysis was done by Barton and Calogeracos in 1995 for a dispersive
mirror in one dimension that includes radiative shift in the mass of the mirror
and the radiative reaction force [34]. This model can be generalized to an infini-
tesimally thin mirror with finite surface conductivity and a normally incident
electromagnetic field.

E. Alternative Theories of Casimir Forces

The experimental verification of Casimir’s prediction is often cited as proof of
the reality of the vacuum energy density of quantum field theory. Yet, as Casimir
himself observed, other interpretations are possible:

The action of this force [between parallel plates] has been shown by clever
experiments and I think we can claim the existence of the electromagnetic
zero point energy without a doubt. But one can also take a more modest
point of view. Inside a metal there are forces of cohesion and if you take
two metal plates and press them together these forces of cohesion begin to
act. On the other hand you can start with one piece and split it. Then you
have first to break chemical bonds and next to overcome van der Waals
forces of classical type and if you separate the two pieces even further
there remains a curious little tail. The Casimir force, sit venia verbo, is the
last but also the most elegant trace of cohesion energy [35].

Several approaches to computing electromagnetic Casimir forces have been

developed that are not based on the zero-point vacuum fluctuations directly. In
the special case of the vacuum electromagnetic field with dielectric or conductive
boundaries, various approaches suggest that Casimir forces can be regarded as
macroscopic manifestations of many-body retarded van der Waals forces, at
least in simple geometries with isolated atoms [1,36]. Casimir effects have also
been derived and interpreted in terms of source fields in both conventional [1]
and unconventional [37] quantum electrodynamics, in which the fluctuations
appear within materials instead of outside of the materials. Lifshitz provided a
 THRUSTING AGAINST THE QUANTUM VACUUM 399

detailed computation of the Casimir force between planar surfaces by assuming

that stochastic fluctuations occur in the tails of the wavefunctions of atoms that
leak into the regions outside the surface These fluctuating tails can induce
dipole moments in atoms in a nearby surface, which leads to a net retarded
dipole-induced dipole force between the planar surfaces [38]. These various
approaches that are alternatives to conventional QED always postulate the exist-
ence of fluctuations in potentials, wave functions, or electromagnetic fields, and
give results consistent with QED formulations in the few cases of simple geome-
tries that have been computed [16].
It should be pointed out that all QED calculations must routinely include the
effects of the vacuum fluctuations in order to obtain the correct results. For
example, the spontaneous emission from excited atoms depends on transitions
induced by the vacuum field.

F. Limitations of Current Theoretical Calculations of Vacuum Forces

The parallel plate geometry (and the approximately equivalent sphere flat
plate geometry or sphere almost flat plate geometry) is essentially the only geo-
metry for which experimental measurements have been conducted and the only
geometry for which the vacuum forces between two separate surfaces
(assumed to be infinite) have been computed. In the calculations with spheres,
the radius of curvature of the sphere is very large compared to the separation;
therefore locally, the geometry is a parallel plate geometry. Vacuum forces are
known to exist in other experimental configurations between separate surfaces,
but rigorous calculations based on QED are very difficult and have yet to be com-
pleted [39]. Because it is experimentally possible to measure forces between
various separate surfaces with the improvement in experimental techniques, the-
oreticians may soon see the need for such computations.
Calculations of vacuum stresses for a variety of geometric shapes, such as
spheres, cylinders, rectangular parallelepipeds, and wedges are reviewed in
Refs. 2, 21, and 23. In general, calculations of vacuum forces become very
complex when the surfaces are curved, particularly with right angles. Diver-
gences in energy appear, and there are disagreements about the proper way to
deal with these divergences [40]. The material properties, such as the dielectric
constant and plasma frequency of the metal and the surface roughness also
affect the vacuum forces, and are often not treated realistically in theoretical cal-
culations. Indeed, in the Lizshitz formulation, the Casimir forces depend on the
permittivity and permeability as a function of the frequency over the entire fre-
quency range. Because this information is not generally available, approxi-
mations have to be made. In addition, usually only a spatial average of the
force for a given area for the ground state of the quantum vacuum field is com-
puted, and material properties such as binding energies are ignored, a procedure
which Barton has recently questioned [25,31,41].

III. Measurements of Casimir Forces

It was not until about 1998, that the parallel plate Casimir force was measured
accurately [4,9]. Corrections for finite conductivity and surface roughness have
been developed for the parallel plate geometry, and the agreement between
theory and experiment is now at about the 1% level for separations of about
400 G. J. MACLAY

0.1 to 0.7 mm [5]. In actual practice, the measurements are most commonly made
with one surface curved and the other surface flat, using the proximity force
theorem to account for the curvature. This experimental approach eliminates the
difficulties of trying to maintain parallelism at submicron separations. Mohideen
and collaborators have made the most accurate measurements to date in this
manner, using an atomic force microscope (AFM) that has a metallized sphere
about 250 mm in diameter attached to the end of a cantilever about 200-mm
long, capable of measuring picoNewton forces. The deflection of the sphere is
measured optically as it is moved close to a flat metallized surface [4]. The
more difficult measurement between two parallel plates has been made by Bressi
et al. who obtained results that are consistent with theory [6]. Measurements of
the force between two parallel surfaces, each with a small (1 nm) sinusoidal modu-
lation in surface height, have shown that there is a lateral force as well as the usual
normal force when the modulations of the opposing surfaces are not in phase [7].
Recent measurements have confirmed the predictions, including effects of
finite conductivity, surface roughness, and temperature, uncertainty in dielectric
functions, to the 1 to 2% level for separations from 65 to 300 nm [8].

A. Forces on Conducting Surfaces

Parallel plate Casimir forces go inversely as the fourth power of the separation
between the plates. The Casimir force per unit area between perfectly conducting
plates is equivalent to about 1 atm pressure at a separation of 10 nm, and so is a
candidate for actuation of MEMS. The relative strength of Casimir, gravitational,
and electrostatic forces for parallel, conducting surfaces is shown in Fig. 1 [42].
In 1995 the first analysis of a dynamic MEMS structure that used vacuum
forces was presented by Serry et al. [42]. They consider an idealized MEMS com-
ponent resembling the original Casimir model of two parallel plates, except that
one of the plates is connected to a stationary surface by a linear restoring force
(spring) and can move along the direction normal to the plate surfaces. The
model demonstrates that the Casimir effect could be used to actuate a switch,
and might be responsible in part for the “stiction” phenomenon in which micro-
machined membranes are found to latch onto nearby surfaces during the fabrica-
tion process. If the movable surface is vibrating, then an “anharmonic Casimir
oscillator” (ACO) results. Other MEMs structures using Casimir forces, such
as pistons and interdigitated combs, have been proposed [43].
In MEMS, surfaces may come into close proximity with each other, particularly
during processes of etching sacrificial layers in the fabrication process. To explore
stiction in common MEMS configurations, Serry et al. computed the deflection of
membrane strips and the conditions under which they would collapse into nearby
surfaces [44]. Measurements were done by Buks and Roukes on cantilever beams
to investigate the role of Casimir forces in stiction [45]. An experimental realiz-
ation of the ACO in a nanometer-scale MEMS system was recently reported by
Chan et al. [10]. In this experiment the Casimir attraction between a 500-mm
square plate suspended by torsional rods and a gold-coated sphere of radius 100
mm was observed as a sharp increase in the tilt angle of the plate as the sphere
plate separation was reduced from 300 nm to 75.7 nm. This “quantum mechanical
actuation” of the plate suggests “new possibilities for novel actuation schemes in
MEMS based on the Casimir force” [10]. In a refinement of this experiment, a
 THRUSTING AGAINST THE QUANTUM VACUUM 401

Fig. 1 Casimir, electrostatic, and gravitational pressures. A comparison of the

attractive pressures due to the Casimir effect and applied electrostatic voltage (V )
between two flat parallel plates of conductors in vacuum. Also shown are the
gravitational pressure on 2-mm thick (dashed line A) and 10-mm thick (line B)
silicon membranes [42].

novel proximity sensor was demonstrated in which the plate was slightly oscillated
with an AC signal, and the deflection amplitude observed with its rapid inverse
fourth power behavior gave an indication of the precise location of the nearby
sphere [11]. A measurement using a similar torsion oscillator was recently
reported using gold on the sphere and chromium on the plate [46].

B. Forces on Semiconductor Surfaces

One of the potentially most important configurations from the technological
viewpoint involves vacuum forces on semiconductor surfaces. The Casimir
force for a conducting material depends approximately on the plasma frequency,
beyond which the material tends to act like a transparent medium. For parallel
plates separated by a distance, d, the usual Casimir force is reduced by a factor
of approximately C(a) [1 þ (8lp/3pd)] 1, where lp is the wavelength corre-
sponding to the plasma frequency of the material [47]. Because the plasma fre-
quency is proportional to the carrier density, it is possible to tune the plasma
frequency in a semiconductor, for example, by illumination, by temperature, or
by the application of a voltage bias. In principle it should be possible to build
a Casimir switch that is activated by light, a device that would be useful in
optical switching systems. A very interesting measurement of the Casimir
force between a flat surface of borosilicate glass and a surface covered with a
film of amorphous silicon was done in 1979 by Arnold et al. [48]. They observed
an increase in the Casimir force when the semiconductor was exposed to light.
This experiment has yet to be repeated with modern methods and materials. As
402 G. J. MACLAY

a first step, Chen et al. have used an AFM to measure the force between a single
Si crystal and a 200-mm diameter gold coated sphere, and found good agreement
with theory using the Lizshitz formalism [49].

IV. Space Propulsion Implications

A. General Considerations
Conservation of energy and momentum place severe restrictions on what
mechanisms may be utilized to propel spacecraft. For example, if a spacecraft
is accelerating due to an interaction with the quantum vacuum, then it has to
be removing energy from the quantum vacuum. Further the increase in kinetic
energy must be equal or less than the decrease in energy in the quantum
vacuum. Some general constraints on using the vacuum for space travel, as
well as methods of altering the metric of spacetime for space travel, are outlined
in the paper by Puthoff et al. [50]. In the analysis of any proposed approaches, we
need to consider the momentum and energy of the field plus any objects in the
field. Consider, for example, a spacecraft mechanism such as a sail, that alters
the normally isotropic quantum vacuum energy density in a local region sur-
rounding the spacecraft. Let E v, ~r, ~rS be the change in the vacuum energy
density as a function of the frequency v, the position ~r measured with respect
to the center of the sail given by ~rS , which is measured  with respect to some
fixed location. If, in actual fact, this function E v, ~r , ~rS does not depend on ~rS
but has the same shape no matter where the sail is located, then the change in
vacuum energy due to the presence of the sail is constant. By the conservation
of energy, the sail is moving at a constant velocity, and cannot experience a
force due to its interaction with the quantum vacuum. In conclusion, if the
change in vacuum energy does not depend on the position of the spacecraft,
then the energy and momentum are constant.

B. Sails in the Vacuum

A variety of sail concepts have been proposed [51]. As we mentioned earlier,
we can view the vacuum as a source of radiation pressure from virtual photons.
The challenge is to design surfaces that alter the symmetry of the free vacuum
and produce a net force. Consider, for example, a sail made of two different
materials on opposite sides that absorb electromagnetic radiation differently.
Can we expect a net force on the sail? A simple classical analysis as shown in
Fig. 2 suggests the answer to this question.
For a given frequency, assume the radiation energy density is proportional to
cf (v, T ), the net momentum transfer DPv to the top surface is

DPv ¼ Av f ðv, T Þ þ Ev f ðv, T Þ þ 2Rv f ðv, T Þ (10)

where Av is the absorptivity, Ev is the emissivity, Rv the reflectivity, and T the

temperature. For a body in thermodynamic equilibrium, Av Ev, and by defi-
nition, 1 Av þ Rv. Using these restriction, it follows that DPv 2f (v, T ),
which is independent of the material properties. Therefore, even if the individual
Av, Ev, and Rv are different for the other side, the same relations hold and the
 THRUSTING AGAINST THE QUANTUM VACUUM 403

f (T ) is the photon density at frequency w and

temperature T
R is the surface reflectivity at frequency w
A in the absorptivity at w
E is the emissivity at w

Absorbed Emitted
photons photons
Reflected A × f (T ) E × f (T )
photons
2R f (T )

Fig. 2 Schematic of the momentum transfer from zero-point electromagnetic

radiation to a sail made from different materials at temperature T on the top and
bottom.

force on the opposite side of the sail just cancels this force, and there is no net
acceleration. This conclusion holds at every frequency. We assumed the tempera-
ture of the sail is the same on both sides because of the intimate contact. If the
radiation spectrum corresponds to that at zero temperature, then f (v, 0) describes
the zero-point field, and both sides of the sail would be at zero Kelvin. On the
other hand, if one made a sail in which a temperature gradient was maintained
across the sail, a net force might occur, and it would be a function of the
energy required to maintain the temperature difference.
There is a complication to this analysis: What happens if the sail is moving? If
the radiation density is due solely to the quantum vacuum at zero temperature
[cf (v, 0) h v3/2p2c 3], then the spectral energy density the sail sees does not
change with motion. The invariance of the spectrum of the zero-point fluctuations
with uniform motion is a special property of the zero-point quantum vacuum.
Without this property, one could distinguish a unique rest frame for the universe,
violating the intent of special relativity. On the other hand, the thermal fields of
real photons do not have this unique invariance. Hence uniform motion in a
thermal field results in a Doppler shifted spectrum. For a sail, this means that
the spectral energy density is different on the opposite sides of the sail and, pro-
vided the integral of the forces over all frequencies were different for the two
sides of the sail, it would be possible to obtain a net, thermally generated,
force. When one considers the restrictions on the frequency dependence of
404 G. J. MACLAY

dielectric coefficients due to causality, it is uncertain if one can generate a net

force with this method. The possibility remains unresolved.
Einstein considered this situation for an atom moving in an isotropic thermal
field, and showed that the increase in the atom’s kinetic energy upon absorption
and emission of radiation is balanced by the drag force if the thermal field follows
the usual Planck spectrum. Similarly if the atom is moving in the zero-point
vacuum field, there is no net force on the atom. But if the spectrum does not
have this form, net forces are possible.
By inserting surfaces into the vacuum, we can alter the spectrum of the
vacuum fluctuations, which results in net forces. Indeed, wherever there is an
inhomogeneous vacuum energy density, there will be a net force on a polarizable
~ ð xÞ2 l [16]. From a propulsion viewpoint, this suggests the
particle given by 12arkE
possibility of using vacuum forces to eject particles to generate a propulsive
force, an approach that may not offer any distinct advantage over more conven-
tional methods.
Friction due to the quantum vacuum has been predicted to exist between
two parallel infinite plates that have finite conductivity. The friction arises
because of the motion of charges in the surface of the metal moving to maintain
the boundary conditions.
In conclusion, for a sail to accelerate due to the quantum vacuum, the sail must
be removing energy from the vacuum. This prompts the question, by what pro-
cesses can one remove energy from the vacuum?

C. Inertia Control by Altering Vacuum Energy Density

One can make use of the negative vacuum energy density that arises in a par-
allel plate structure in an alternative approach to propulsion. Based on the prin-
ciples of General Relativity, one would expect the changes in vacuum energy DE
to correspond to a change in mass DE/c 2. Thus the negative energy density of the
vacuum in a parallel plate structure should result in the equivalent of a negative
mass object. Proposals have been made to test this hypothesis, stating that a nega-
tive vacuum energy leads to a reduction in mass by constructing stacks of capaci-
tors; however, the predicted effect is just beyond current measurement capability
From the theoretical perspective, one can estimate the positive and negative mass
contributions for a parallel plate capacitor made from plates that are only one
hydrogen atom thick and one atom apart, and still the total energy is positive.
Thus it appears that a parallel-plate Casimir cavity will always have a net
positive energy density and cannot be used to create a zero or negative mass
spacecraft, or initiate a wormhole [17]. Nevertheless, the negative vacuum
energy density may, with more effective approaches, be of use in reducing
inertia, which reduces kinetic energy and the amount of work required to accel-
erate a spacecraft to a given velocity [50].
There is another variant of a negative mass drive that deserves mention.
A system of charges that has a negative electrostatic potential energy DE
would also, by the principles of General Relativity, be expected to have a
negative associated mass DE/c 2. Thus, in a gravitational field, there would be
a levitating force. Pinto has explored this possibility, and suggests the effect
may be amplified and made measurable if highly polarizable hydrogen atoms
 THRUSTING AGAINST THE QUANTUM VACUUM 405

in a magnetic trap are exposed to isotropic laser radiation [52]. Although these
technically challenging enhancements may amplify the basic levitating force, it
still appears that the effective reduction in mass will be quite small compared
to the total mass.
Negative energy drives and the hypothesis linking inertia to vacuum fluctu-
ations [53] are discussed more fully in Chapters 3, 4, 13, and 15.

D. Dynamic Systems
Dynamic systems, in which something moves and interacts with the quantum
vacuum, may have the possibility of extracting energy from the vacuum.
Hence they may be able to accelerate a spacecraft. The movement might be a
macroscopic physical motion, a piezoelectrically driven surface, or the motion
of electrons within a semiconductor, possibly altering the plasma frequency or
the dielectric constant. We discuss one possible dynamic system next.

V. Vibrating Mirror Casimir Drive

It is possible to conceive of a vacuum spacecraft that operates by pushing on
the quantum vacuum with a vibrating mirror [54]. With a suitable trajectory, the
motion of a mirror in vacuum can excite the quantized vacuum electromagnetic
field with the creation of real photons.

A. Simple Model Based on Energy and Momentum Conservation

The important physical features of using the dynamic Casimir effect to accel-
erate a spacecraft can be seen in a simplified, heuristic model. Assume that the
spacecraft has an energy source, such as a battery, that powers a motor that
vibrates a mirror or a system of mirrors in a suitable manner to generate radiation.
We will assume that there are no internal losses in the motor or energy source.
We assume that at the initial time ti, the mirrors are at rest. Then the mirrors
are accelerated by the motor in a suitable manner to generate a net radiative
reaction on the mirror, and at the final time tf, the mirrors are no longer vibrating,
and the spacecraft has attained a non-zero momentum. We can apply the First
Law of Thermodynamics to the system of the energy source, motor, and mirror
at times ti, and tf:

DQ ¼ DU þ DW (11)

where DU represents the change in the internal energy in the energy source,
DW represents the work done on the mirrors moving against the vacuum,
and DQ represents any heat transferred between the system and the environment.
We will assume that we have a thermally isolated system and DQ 0 so

0 ¼ DU þ DW (12)

By the conservation of energy, the energy DU extracted from the battery goes
into work done on the moving mirror DW. Because the mirror has zero
vibrational and kinetic energy and zero potential energy at the beginning and
406 G. J. MACLAY

the end of the acceleration period, and is assumed to operate with no mechanical
friction, all work done on the mirror goes into the energy of the emitted radiation
DR, and the kinetic energy of the spacecraft of mass M

M ðDV Þ2
DW ¼ DR þ (13)
2

Thus the energy of the radiation emitted due to the dynamic Casimir effect equals

M ðDV Þ2
DR ¼ DU  .0 (14)
2

The frequency of the emitted photons depends on the Fourier components

of the motion of the mirror. We assume that the radiant energy can be expressed
as a sum of energies of ni photons each with frequency vi:
X
DR ¼ ni h vi (15)
i

The number of photons emitted depends on the cosine of the angle the photon
momentum makes with the normal to the surface, as Neto and Machado [28]
show. In this simplified calculation, we will assume that all photons are
emitted normally from one side of the accelerating surface. This assumption is
not valid, but it allows us to obtain a best-case scenario and illustrates the
main physical features. If all photons are emitted normally from one surface,
then the photon momentum transfer DP is
X hvi DR
DP ¼ ni ¼ (16)
i
c c

where c is the speed of light. Using Eq. (14), we obtain the result

DU M ðDV Þ2
DP ¼  (17)
c 2c

In a nonrelativistic approximation DP MDV and the change in velocity DV of

the spacecraft is to second order in DU/Mc 2:
 
DV DU DU 2
¼ þ (18)
c Mc2 Mc2

This represents a maximum change in velocity attainable by use of the dynamic

Casimir effect (or by the emission of electromagnetic radiation generated by
more conventional means) when the energy DU is expended. The ratio DU/
Mc 2 is expected to be a small number, and we can neglect the second term in
Eq. (18). As a point of reference, for a chemical fuel, the ratio of the heat of
 THRUSTING AGAINST THE QUANTUM VACUUM 407

formation to the mass energy is approximately 10 10. With this approximation,

we find the maximum value of DV/c equals DU/Mc 2, the energy obtained from
the energy source divided by the rest mass energy of the spacecraft. It follows that
the kinetic energy of the motion of the spacecraft Eke can be expressed as:

M ðDV Þ2 DU
Eke ¼ ¼ DU (19)
2 2Mc2
This result for the upper limit on the spacecraft kinetic energy shows that the
conversion of potential energy DU from the battery into kinetic energy of the
spacecraft is an inefficient process because DU/Mc 2 is a small factor. Almost
all of the energy DU has gone into photon energy. This inefficiency follows
because the ratio of momentum to energy for the photon is 1/c.
In our derivation, the internal energy of the system is used to create and emit
photons from some unspecified process; no massive particles are ejected from the
spacecraft (propellantless propulsion). We have neglected: 1) the change in the
mass of the spacecraft as the stored energy is converted into radiation, 2) radiative
mass shifts, 3) complexities related to high energy vacuum fluctuations and
divergences, and 4) all dissipative forces in the system used to make the
mirror vibrate. These assumptions are consistent with a heuristic nonrelativistic
approximation.
In this simplified model, we have not made any estimates about the
rate of photon emission and how long it would take to reach the maximum
velocity. For configurations considered in the literature, rates of photon emission
from the dynamic Casimir effect are estimated to be very low, typically 10 5
photons/sec or about 300 photons/yr [28]. Also we will have to vibrate the
mirror asymmetrically so that more photons are emitted from one side than
the other. In the derivation, however, we never made any assumptions about
the mechanism by which photons were generated, so the derivation holds quite
generally, whether we simply use a battery and a perfect lightbulb or a vibrating
charged surface.

1. Static Casimir Effect as Energy Source

The general analysis in the preceding section can be taken one step further to
suggest a spacecraft whose operation is totally based on quantum vacuum
properties. The vibrating motor in the spacecraft could be powered by energy
removed from the quantum vacuum using an arrangement of perfectly con-
ducting, uncharged, parallel plates. Detailed considerations about extracting
energy from the quantum vacuum are discussed in Refs. 55 58, and presented
in Chapter 18. The Casimir energy UC(x) at zero degrees Kelvin between
plates of area A, separated by a distance x is:

p 2 hcA
U C ð xÞ ¼  (20)
720 x3
If we allow the plates to move from a large initial separation a to a very small
final separation b then the change in the vacuum energy between the plates is
408 G. J. MACLAY

approximately:

U C ð x Þ ¼ U C ð bÞ  U C ð aÞ (21)

Substituting Eq. (20) gives the result

p 2 h cA
DUC   (22)
720 b3

The attractive Casimir force has done work on the plates, and, in principle, we
can build a device to extract this energy with a suitable, reversible, isothermal
process, and use it to accelerate the mirrors. We neglect any dissipative forces
in this device, and assume all of the energy DUC can be utilized. Thus the
maximum value of DVC/c obtainable using the energy from the Casimir force
“battery” is:

DVC p 2 1 h cA
¼ (23)
c 720 Mc2 b3

We can make an upper bound for this velocity by making further assumptions
about the composition of the plates. Assume that the plate of thickness L is made
of a material with a rectangular lattice that has a mean spacing of d, and that the
mass associated with each lattice site is m. Then the mass of one plate is:

m
MP ¼ AL (24)
d3

The density approximation is good for materials with a cubic lattice, and
within an order of magnitude of the correct density for other materials.
In principle it is possible to make one of the plates in the battery the same as
the plate accelerated to produce radiation by the dynamic Casimir effect. As the
average distance between the plates is decreased, the extracted energy is used to
accelerate the plates over very small amplitudes. If we assume we need to employ
two plates in our spacecraft, and that the assembly to vibrate the plates has neg-
ligible mass, then the total mass of the spacecraft is M 2MP and we obtain an
upper limit on the increase in velocity:

DVC p 2 h d 3
¼ (25)
c 1400 Lmc b3

The final velocity is proportional to the Compton wavelength (h /mc) of the
lattice mass m divided by the plate thickness L. Assume that the final spacing
between the plates is one lattice constant (d b), that the lattice mass m
equals the mass of a proton mp, and that the plate thickness F is one Bohr
radius a0 h2/mee 2, then we obtain (a is the fine structure constant with
 THRUSTING AGAINST THE QUANTUM VACUUM 409

approximate value of 1/137):

DVC p 2 ame
¼ (26)
c 1400 mp

(A real plate constructed with current technology might easily be three orders of
magnitude thicker.) Substituting numerical values we find:

DVC p2 1 1 8
¼ ¼ 2:78  10 (27)
c 1400 137 1800
This best-case scenario corresponds to a disappointing final velocity of about
8 m/s, about 103 times smaller than for a large chemical rocket. As anticipated,
the spacecraft is very slow despite the unrealistically favorable assumptions made
in the calculation, yet this simple gedanken experiment does demonstrate that it
may be possible to base the operation of a spacecraft entirely on the properties of
the quantum vacuum. Using an additional energy source can result in higher
terminal velocity.

B. Model for Propulsion Using Vibrating Mirrors

Assume we have a flat, perfectly reflecting, mirror whose equilibrium
position is x 0. At a time t where ti , t , tf the location of the mirror is
given by x(t). Neto has given an expression for the force per unit area F(t) on
such a mirror [32]:
hc  1 d4 xðtÞ d5 xðtÞ
F ðtÞ ¼ lim  5 5 (28)
dx!0 30p2 dx c4 dt4 c dt
where dx represents the distance above the mirror at which the stress-energy
tensor is evaluated. The second term represents the dissipative force that is
related to the creation of traveling wave photons, in agreement with its interpret-
ation as a radiative reaction. In computing the force due to the radiation from the
mirror’s motion, the effect of the radiative reaction on x(t) is neglected in the non-
relativistic approximation. The divergent first term can be understood in several
ways. Physically, it is a dispersive force that arises from the scattering of low fre-
quency evanescent waves. The divergence can be related to the unphysical nature
of the perfect conductor boundary conditions. Forcing the field to vanish on the
surface requires its conjugate momentum to be unbounded. Thus the average of
the stress-energy tensor kTmv l is singular at the surface for the same reason that
single-particle quantum mechanics would require a position eigenstate to have
infinite energy [30]. This divergent term can be lumped into a mass renormaliza-
tion, and therefore disappears from the dynamical equations when they are
expressed in terms of the observed mass of the body [31,32]. We will not
discuss this term further in this calculation, although we will return to the
general idea of radiative mass shift in our discussion. We will assume that diffrac-
tion effects are small for our finite plates.
410 G. J. MACLAY

The total energy radiated per unit plate area E can be expressed

ðt2
dxðtÞ
E ¼  dtF ðtÞ (29)
dt
t1

Substituting Eq. (28) for F(t) we find

h 1 ð d3 xðtÞ2
t2

E¼ dt (30)
30p2 c4 dt3
t1

The total impulse I per unit plate area can also be computed as the integral of the
force per unit area over time:
ðt2 h 1 d4 xðtÞ  !
d 4
xð t Þ 
I ¼ dt F ðtÞ ¼     (31)
30p 2 c4 dt4 t2 dt4 t1
t1

The total impulse I equals the mass of the system M per unit area times the change
in velocity DV in a nonrelativistic approximation:
I ¼ MDV (32)
We want to specify a trajectory for the mirror that will give a net impulse. One
of the trajectories that has been analyzed is that of the harmonic oscillator
[30,59]. In this case, the mirror motion is in a cycle and we can compute the
energy radiated per cycle per unit area and the impulse per cycle per unit area.
For a harmonic oscillator of frequency V and period T 2p/V, there is only
one Fourier component of the motion, so the total energy of each pair of
photons emitted is h V ¼ h ðv1 þ v2 Þ. For a harmonically oscillating mirror the
displacement is
xho ðtÞ ¼ X0 sin Vt (33)
A computation based on Eqs. (30) and (31) shows there will be a net power
radiated in a cycle, however, the dissipative force for the harmonic oscillator
Fho will average to zero over the entire cycle as shown in Fig. 3, so there will
be no net impulse.
In order to secure a net impulse, we need a modified mirror cycle. One such
model cycle can be readily constructed by using the harmonic function xho(t)
over the first and last quadrants of the cycle, where the force Fho is positive,
and a cubic function xc(t) over the middle two quadrants where Fho is negative:

X0 ðVt  pÞ3 3X0 ðVt  pÞ

xc ðtÞ ¼  (34)
2 ðp=2Þ3 2 p=2

The coefficients for the cubic polynomial are chosen so that at Vt p/2, 3p/2
the displacement and the first derivatives of xc(t) and xho(t) are equal. As can be
 THRUSTING AGAINST THE QUANTUM VACUUM 411

Fh o Xh o
1

0.5

t/T
0.2 0.4 0.6 0.8 1

–0.5

–1

Fig. 3 The displacement xho and the radiative reaction Fho (bold line) for a
harmonically oscillating mirror plotted as a function of the normalized time. For
convenience Fho and xho are normalized to one.

seen from Fig. 4, the cubic function xc(t) matches xho(t) quite closely in the interval
0.25 , t/T , 0.75. Of course the higher order derivatives do not match, and that is
precisely why the force differs.
The similarity in displacement and the difference in the resulting force is strik-
ing. For the mirror displacement xm(t) in our model we choose:

xm ðtÞ ¼ xho ðtÞ for 0 t=T 0:25; 0:75 t=T 1 (35)

xm ðtÞ ¼ xc ðtÞ for 0:25 , t=T , 0:75 (36)

Figure 5 shows xm(t) plotted with the corresponding force per unit area Fm(t)
obtained from Eq. (28). The force Fm(t) is positive in the first and last quarter
of the cycle, and vanishes in the middle, where the trajectory is described by

Xh Xc
1

0.5

t/T
0.2 0.4 0.6 0.8 1

0.5

Fig. 4 The normalized displacement xho(t) for a harmonically oscillating mirror

(solid line) and the cubic function xc(t), (bold dotted line).
412 G. J. MACLAY

Fm Xm
1

0.5

t/T
0.2 0.4 0.6 0.8 1

–0.5

-1

Fig. 5 The normalized displacement xm(t) and the corresponding normalized

radiative force Fm(t) (bold line), are shown as functions of the time. The force is
positive in the first and last quarters, and zero in the middle quarters of the cycle.

the cubic. The energy radiated per area per cycle for our model trajectory can be
obtained from Eq. (30):
hc  5
2 V
Em ¼  X (37)
60p 0 c

The total impulse per area per cycle for our model Im trajectory is

  4
h V
Im ¼  2
X0 (38)
15p c

The impulse is first order in h and is therefore typically a small quantum effect.
Thus for our model cycle, the change in velocity per second is DV/dt:

Im V
DVm =dt ¼ (39)
M

where M is the mass per unit plate area of the spacecraft, and we assume the
plate is the only significant mass in the gedanken spacecraft. In order to estimate
DVm, we can make some further assumptions regarding the mass of the plate
per unit area. As before, we can make a very favorable assumption regarding
the mass per unit area of the plates M mp/a0 2, which yields the change in
velocity per second:

  4
h V a0
DVm =dt ¼  2
X0 V (40)
15p c mp

If we substitute reasonable numerical values [59,60], a frequency of

V 3  1010 s 1 and an oscillation amplitude of X0 10 9 m, we find that
 THRUSTING AGAINST THE QUANTUM VACUUM 413

DVm/dt is approximately 3  10 – 20 m/s2 per unit area, not a very impressive

acceleration. Physically, one would imagine the surface of the mirror
vibrating with an amplitude of just one nanometer. This conservative limitation
in the amplitude arises because the maximum velocity of the boundary
is proportional to the elastic deformation, which cannot exceed about 10 2 for
typical materials. The energy radiated per area EmV is about 10 25 W/m2.
There are a number of methods to increase these values by many orders of
magnitude, as discussed below.
The efficiency of the conversion of energy expended per cycle in our model Em
into kinetic energy of the spacecraft Eke 12M(DVm)2 Im 2/2M is given in the
nonrelativistic approximation by the ratio:

Eke h 1 V3
¼ (41)
Em Mc p c
With our assumptions, the approximate value of this ratio is 10 – 26, making
this conversion an incredibly inefficient process.

C. Methods to Increase Acceleration

The dynamic Casimir effect has yet to be verified experimentally, hence there
have been a number of interesting proposals describing methods designed to
maximize the effect so it can be measured. In 1994, Law predicted a resonant
response of the vacuum to an oscillating mirror in a one-dimensional cavity
[61]. The behavior of cavities formed from two parallel mirrors that can move
relative to each other is qualitatively different from that of single plates. For
example, it is possible to create particles in a cavity with plates separating at con-
stant velocity [62]. The very interesting proposal by Lambrecht et al. concludes
that if the mechanical oscillation frequency is equal to an odd integer multiple of
the fundamental optical resonance frequency, then the rate of photon emission
from a vibrating cavity formed with walls that are partially transmitting with
reflectivity, r, is enhanced by a factor equal to the finesse f 1/ln(1/r) of the
cavity [59,60]. For semiconducting cavities with frequencies in the GHz range,
the finesse can be 109, giving our gedanken spacecraft an acceleration of 3 
10 11 m/s2 based on Eq. (40). Plunien et al. have shown that the resonant
photon emission from a vibrating cavity is further increased if the temperature
is raised [63]. For a 1-cm cavity, the enhancement is about 103 for a temperature
of 290K. Assume that one has a gedanken spacecraft with a vibrating cavity oper-
ating at about 290K providing a 1012 total increase in the emission rate. This
would result in an acceleration per unit area of the plates of 3  10 – 8 m/s2, a
radiated power of about 10 – 13 W/m2 and an efficiency Eke/Em of about 10 – 14.
After 10 years of operation, the gedanken spacecraft velocity would be approxi-
mately 10 m/s, which is about three orders of magnitude less than the current
speed of Voyager (17 km/s), obtained after a gravity assist maneuver around
Jupiter to increase the velocity. The burn-out velocity for Voyager at launch in
1977 was 7.1 km/s [64].
The numerical results for the model obviously depend very strongly on the
assumptions made. For the plate mass/area and system to vibrate the plate, we
414 G. J. MACLAY

have made unrealistically favorable estimates; for the oscillation frequency and
amplitude, we have made conservative estimates. It is possible that new
materials, with the ability to sustain larger strains, could make possible an ampli-
tude of oscillation orders of magnitude larger than 1 nm. Perhaps the use of nano-
materials, such as carbon nanotubes that support 5% strain [65], or “super” alloys
[66], would allow a much larger effective deformation. If the amplitude was
1 mm instead of 1 nm, the gedanken spacecraft, or some modification with
improved coupling to the vacuum, might warrant practical consideration.
Eberlein has shown that density fluctuations in a dielectric medium would also
result in the emission of photons by the dynamic Casimir effect [67]. This
approach may ultimately be more practical with large area dielectric surfaces
driven electrically at high frequencies. More theoretical development is needed
to evaluate the utility of this method. Other solid state approaches may also be
of value with further technological developments. For example, one can envision
making sheets of charge that are accelerated in MOS type structures. Yablono-
vitch has pointed out that the zero-point electromagnetic field transmitted
through a window, whose index of refraction is falling with time, shows
the same phase shift as if it were reflected from an accelerating mirror [68]. To
simulate an accelerating mirror, he suggested utilizing the sudden change in
refractive index that occurs when a gas is photoionized or the sudden creation
of electron hole pairs in a semiconductor, which can reduce the index of refrac-
tion from 3.5 to 0 in a very short time. Using subpicosecond optical pulses,
the phase modulation can suddenly sweep up low-frequency waves by many
octaves. By lateral synchronization, the moving plasma front with its large
change in the index of refraction can, in effect, act as a moving mirror exceed-
ing the speed of light. Therefore, one can regard such a gas or semiconductor
slab as an observational window on accelerating fields, with accelerations as
high as 1020 m/s [68]. Accelerations of this magnitude will have very high
frequency Fourier components. Equation (40) shows that the impulse goes as
the fourth power and the efficiency as the third power of the Fourier component
for an harmonic oscillator, which suggests that with superhigh accelerations,
an optimum time dependence of the field and an optimum shape of the wave-
fronts, one might be able to secure much higher fluxes of photons and a
much higher impulse per second, with a higher conversion efficiency. On the
other hand, a preliminary calculation by Lozovik et al. to investigate this
approach suggests that the accelerated plasma method may have limitations
in producing photons [69].
Ford and Svaiter have shown that it may be possible to focus the fluctuating
vacuum electromagnetic field [70]. This capability might be utilized to create
regions of higher energy density. This might be of use in a cavity in order to
increase the flux of radiated photons. There may be also enhancements due to
nature of the index of refraction for real materials. For example, Ford has com-
puted the force between a dielectric sphere, whose dielectric function is described
by the Drude model (based on a simple approximate model for free electrons in a
metal), and a perfectly reflecting wall, with the conclusion that certain large com-
ponents of the Casimir force no longer cancel. He predicts a dominant oscillatory
contribution to the force, in effect developing a model for the amplification of
vacuum fluctuations. Barton has shown that for materials with a fixed index of
 THRUSTING AGAINST THE QUANTUM VACUUM 415

refraction, the force for a one-dimensional scalar field goes as [(n 1)/n]2,
which suggests the possibility that one might be able to enhance the force by
selection of a material with a small index [31].
Another, albeit improbable, approach to a vacuum-facilitated gedanken space-
craft is to consider the possibility of adjusting the radiative mass shift, which we
have neglected until this point. There is a very small radiative shift of the mirror
due to its interaction with the vacuum, akin to the Lamb shift for an atom [34,59].
To measure a vacuum mass shift, a proposal was made recently to measure the
inertial mass shift in a multilayer Casimir cavity, which consists of 106 layers
of metal 100-nm thick and 35 cm in diameter, alternating with films of silicon
dioxide 5-nm thick [71]. The mass shift is anticipated to arise from the decrease
in the vacuum energy between the parallel plates. A calculation shows that
the mass shift for the proposed cavity is at or just beyond the current limit of
detectability.
It appears that if quantum vacuum engineering of spacecraft is to become
practical, and the dreams of science fiction writers are to be realized, we may
need to develop new methods to be able to manipulate changes in vacuum
energy densities that are near to the same order of magnitude as mass energy den-
sities. Then we would anticipate being able to shift inertial masses by a significant
amount. Because mass shifts in computations are often formally infinite, perhaps
such developments are not forbidden, and, with more understanding (and
serendipity!), may be controllable. With large mass shifts one might be able to
build a structure that had a small or zero inertial mass, which could be
readily accelerated.

VI. Unresolved Physics

Casimir effects are typically small and difficult to measure. In fact, measure-
ments have only been made for the simplest geometries, such as the parallel plate
geometry or the sphere plate geometry. No measurements have ever been made
that validate the dynamic Casimir effect. Assuming the effect is verified exper-
imentally, then the question is: Is it possible to amplify these effects and bring
them into the useful range? This is certainly one of the challenges of vacuum
engineering. Experiments are needed that explore some of the issues that are
beyond the present calculational ability of QED; for example, the effect of
complex geometries on vacuum forces, or the effect of massive fields or dense,
moving nuclear matter on the quantum vacuum. Is it possible to make a stable
vacuum field that has a large variation in energy density? What is the effect of
changes in vacuum energy on the index of refraction? Can energy density gradi-
ents be found on a length scale that is useful to humans? We need to greatly
increase our knowledge of the quantum vacuum. The development of a very sen-
sitive small probe that provided a frequency decomposition of the local vacuum
energy density would certainly be very helpful.
From the status of current research in Casimir forces, it is clear that we are at
the tip of the iceberg describing the properties of the quantum vacuum for real
systems with real material properties. For example, there is no general agreement
regarding the calculations of static vacuum forces for geometries other than
infinite parallel plates of ideal or real metals at a temperature of absolute zero.
416 G. J. MACLAY

Non-zero temperature corrections for flat, real metals are uncertain [72,73].
There are fundamental disagreements about the computation of vacuum forces
for spheres or rectangular cavities, and about how to handle real material
properties and curvature in these and other geometries [31]. Indeed, it is very
difficult to calculate Casimir forces for these simple geometries and to relate
the calculations to an experiment. Calculations have yet to be done for more
complex and potentially interesting geometries. The usual problems in QED
such as divergences due to unrealistic boundary conditions, to curvature, to
interfaces with different dielectric coefficients abound [74].

A. Possible Discriminating Tests

As discussed, to make practical spacecraft based on engineering the quantum
vacuum, it seems we need to find new boundary conditions for the vacuum that
can alter the vacuum energy density by orders of magnitude more than with our
current boundary conditions, which are primarily metal or dielectric surfaces.
Perhaps the use of new materials (e.g., those with a negative index of refraction,
or an ultra-high electrical carrier density, either steady state or transient), or novel
superconducting materials may open the door to new Casimir phenomena.
Recently the use of negative index materials was proposed to make a repulsive
Casimir force [75]. With significantly increased funding of research, some break-
throughs might be possible.
There are some important experiments that can help our understanding of
vacuum energy and Casimir forces that may lead to significant improvements
in our engineering capability. Experiments to verify the adiabatic Casimir
effect have been suggested in the literature. This is an important theoretical
issue that has ramifications in different fields, including astrophysics and elemen-
tary particle physics. Within the next five years, some clever experimental
approaches will probably be developed to explore the adiabatic Casimir effect.
Experiments measuring the Casimir forces for semiconductor surfaces would
be helpful in the development of new applications of vacuum forces and to
demonstrate that it is possible to alter the Casimir force by altering the carrier
density. The measurement of Casimir forces and energies for different geometry
and composition objects, such as rectangular cavities or spheres, would provide
some badly needed answers for theoretical modeling. Measurements of Casimir
forces between separate, nonplanar surfaces are also needed. There may be sur-
faces that have larger forces than the classic parallel plates, but we have yet to
measure anything but parallel plates.
Tests to determine if negative vacuum energy yields a negative force in a grav-
itation field have been proposed. This is a fundamental question. Order-of-
magnitude estimates suggest that a negative mass object is not possible, but
that the total mass may be reduced a measurable amount by contributions from
negative vacuum energy.

B. Application Implications
Because Casimir forces require small separations, they have been utilized in
micro- and nano-electromechanical systems (MEMS/NEMS). As microfabrica-
tion technology develops, nanodevices with shorter working distances and more
 THRUSTING AGAINST THE QUANTUM VACUUM 417

varied materials will be developed, resulting in more applications for Casimir

forces. An excellent recent survey by Capasso et al. discusses a variety of
devices, including the anharmonic Casimir oscillator and a surface with a Pd
film whose electrical properties change in the presence of hydrogen, which
could lead to a measurable change in the Casimir force [12]. Also reported
were repulsive Casimir forces for certain planar material combinations as pre-
dicted by the Lifshitz formulation, for example, a gold-covered planar surface
and a silica-covered surface when ethanol is the medium in between. An ultrasen-
sitive magnetometer was proposed using the repulsive Casimir force to levitate a
disk with a magnetic dipole in alcohol. Also proposed was a demonstration
of quantum torque, analogous to the Casimir force in origin [76]. In this
system, the birefringent material is floated by the repulsive Casimir force
above a fixed birefringent plate. Steady progress in the application of Casimir
forces in MEMS is to be expected.

VII. Conclusions
One objective was to illustrate some of the unique properties of the quantum
vacuum and how they might be utilized in propulsion of a spacecraft. We have
outlined some of the considerations for the use of vacuum energy to propel a
spacecraft and pointed out some directions in which some helpful discoveries
may lie.
We have demonstrated that it is possible in principle to propel a spacecraft
using the dissipative force an accelerated mirror experiences when photons are
generated from the quantum vacuum. Further, we have shown that one could,
in principle, utilize energy from the vacuum fluctuations to operate the vibrating
mirror assembly required. The application of the dynamic Casimir effect and the
static Casimir effect may be regarded as a proof of principle, with the hope that
the proven feasibility will stimulate more practical approaches exploiting known
or as yet unknown features of the quantum vacuum. A model gedanken spacecraft
with a single vibrating mirror was proposed that showed a very unimpressive
acceleration due to the dynamic Casimir effect of about 3  10 – 20 m/s2 with a
very inefficient conversion of total energy expended into spacecraft kinetic
energy. Employing a set of vibrating mirrors to form a parallel plate cavity
and raising the cavity temperature to about 290K increased the output by a
factor of the finesse of the cavity times approximately 103, yielding an accelera-
tion of about 3  10 8 m/s2 and a conversion efficiency of about 10 14. To
put this into perspective, after 10 years at this acceleration, the spacecraft
would have only attained a 10 m/s velocity. At least we have computationally
suggested (pending experimental verification of the phenomena predicted by
QED), that it is possible to use the quantum vacuum to propel a spacecraft.
This represents progress.
The vacuum effects that we computed scale with Planck’s constant and are
therefore very small. In order to have a practical spacecraft based on quantum
vacuum properties, it would be preferable that the vacuum effects scale as h0,
meaning that the effects are essentially independent of Plank’s constant and con-
sequently may be much larger. By itself this requirement does not guarantee a
large enough magnitude, but it certainly helps [59]. New methods of modifying
418 G. J. MACLAY

the quantum vacuum boundary conditions may be needed to generate the large
changes in free-field vacuum energy or momentum required if “vacuum
engineering” as proposed in this paper is ever to be practical. For example, the
vacuum energy density difference between parallel plates and the region
outside them in free space is simply not large enough in magnitude for our engin-
eering purposes. Energy densities, positive or negative, that are orders of magni-
tude greater are required. Such energy density regions may be possible, at least in
some cases. For example, a region appeared in the one-dimensional dynamic
system in which the energy density was below that of the Casimir parallel
plate region [61]. Similarly, more effective ways of transferring momentum to
the quantum vacuum than using photons generated with the adiabatic Casimir
effect are probably necessary if a spacecraft is to be propelled using the vacuum.
Although the results of our calculations using the vibrating mirrors to propel a
gedanken rocket are very unimpressive, it is important to not take our conclusions
regarding the final velocity in our simplified models too seriously. The choice of
numerical parameters can easily affect the final result by four orders of magni-
tude. The real significance is that a method has been described that illustrates
the possibility of propelling a rocket by coupling to the vacuum. It is possible
that there may be vastly improved methods of coupling. There are numerous
potential ways in which the ground state of the vacuum electromagnetic field
might be engineered for use in technological applications, a few of which we
have mentioned here. As the technology to fabricate small devices improves,
and as the theoretical capability of calculating quantum vacuum effects increases,
and as experiments help us understand the issues, it will be interesting to see
which possibilities prove to be useful and which remain curiosities.
Optical applications of vacuum engineering, such as fabricating lasers in cav-
ities to control spontaneous emission, have not been mentioned. Likewise, the
current astrophysical conundrums about dark energy and the cosmological con-
stant, which may relate to vacuum energy, were not discussed because we are
not very good at predicting the development of technology. In the 1980s we
thought artificial intelligence was going to revolutionize the world, but it did
not. In the 1960s, manufacturers were hard put to think of any reason why an indi-
vidual would want a home computer and today we wonder how we ever survived
without them. Circa 1900 an article was published in Scientific American proving
that it was impossible to send a rocket, using a conventional propellant, to the
moon. The result was based on the seemingly innocuous assumption of a
single-stage rocket. It is hoped that a paper on vacuum propulsion written 100
years from now, will also find amusing our failure to perceive the key issues
and see clearly how quantum propulsion should be done.

Acknowledgments
I gratefully acknowledge helpful comments and suggestions from Dan Cole, and
I thank Gabriel Barton, Carlos Villarreal Lujan, Claudia Eberlein, Peter Milonni,
and Paulo Neto for discussions about their work. I would like to express my
appreciation to the NASA Breakthrough Propulsion Physics Program and Marc
G. Millis for the support of earlier versions of this work. I am sad to report that
Robert L. Forward died while we were working on the Casimir drive.
 THRUSTING AGAINST THE QUANTUM VACUUM 419

References
[1] Milonni, P. W., The Quantum Vacuum: An Introduction to Quantum Electro
dynamics, Academic Press, San Diego, CA, 1994.
[2] Milton, K. A., The Casimir Effect, USA World Scientific, NJ, 2001.
[3] Milonni, P., Cook, R., and Groggin, M., “Radiation Pressure from the Vacuum:
Physical Interpretation of the Casimir Force,” Physical Review, Vol. A38, 1988,
p. 1621.
[4] Mohideen, U., and Anushree, R., “Precision Measurement of the Casimir Force
from 0.1 to 0.9 Micron,” Physical Review Letters, Vol. 81, 1998, p. 4549.
[5] Klimchitskaya, G., Roy, A., Mohideen, U., and Mostepanenko, V., “Complete
Roughness and Conductivity Corrections for Casimir Force Measurement,” Physi
cal Review A, Vol. 60, 1999, pp. 3487 3495.
[6] Bressi, G., Carugno, G., Onofrio, R., and Ruoso, G., “Measurement of the Casimir
Force between Parallel Metallic Plates,” Physical Review Letters, Vol. 88, 2002,
p. 041804.
[7] Chen F., Mohideen, U., Klimchitskaya, G. L., and Mostepanenko, V. M., “Demon
stration of the Lateral Casimir Force,” Physical Review Letters, Vol. 88, 2002,
p. 101801.
[8] Chen, F., Mohideen, U., Klimchitskaya, G. L., and Mostepanenko, V. M., “Theory
Confronts Experiment in the Casimir Force Measurements: Quantification of Errors
and Precision,” Physical Review A, Vol. 69, 2004, pp. 022117 1 11.
[9] Lamoreaux, S., “Measurement of the Casimir Force Between Conducting Plates,”
Physical Review Letters, Vol. 78, 1997, p. 5.
[10] Chan, H. B., Aksyuk, V., Kleiman, R., Bishop, D., and Capasso, F., “Quantum
Mechanical Actuation of Microelectromechanical Systems by the Casimir
Force,” Science, Vol. 291, 2001, p. 1941.
[11] Chan, H. B., Aksyuk, V. A., Kleiman, R. N., Bishop, D. J., and Capasso, F.,
“Nonlinear Micromechanical Casimir Oscillator,” Physical Review Letters, Vol.
87, 2001, p. 211801.
[12] Capasso, F., Munday, J., Iannuzzi, D., and Chan, H., “Casimir Forces and Quantum
Electrodynamical Torques: Physics and Nanomechanics,” IEEE Journal of Selected
Topics in Quantum Electronics, Vol. 13, 2007, pp. 400 414.
[13] Morris M. S., and Thorne, K. S., “Wormholes in Spacetime and Their Use for
Interstellar Travel: A Tool for Teaching General Relativity,” American Journal
of Physics, Vol. 56, 1988, pp. 395 412.
[14] Lamb, W., and Retherford, R., “Fine Structure of the Hydrogen Atom by a
Microwave Method,” Physics Review, Vol. 72, 1947, p. 241.
[15] Bethe, H., “The Electromagnetic Shift of Energy Levels,” Physics Review, Vol. 72,
1948, p. 339.
[16] Maclay, J., Fearn, H., and Milloni, P., “Of Some Theoretical Significance: Impli
cations of Casimir Effects,” European Journal of Physics, Vol. 22, 2001, p. 463.
[17] Visser, M., Lorentzian Wormholes: From Einstein to Hawking, AIP Press,
New York, 1996, pp. 81 87.
[18] Ostriker, J., and Steinhardt, P., “The Quintessential Universe,” Scientific American,
Vol. 284, 2001, pp. 46 53.
[19] Weinberg, S., “The Cosmological Constant Problem,” Reviews of Modern Physics,
Vol. 61, 1989, p. 1.
420 G. J. MACLAY

[20] Adler, R. J., Casey, B., and Jacob., O. C., “Vacuum Catastrophe: An Elementary
Exposition of the Cosmological Constant Problem,” American Journal of
Physics, Vol. 63, 1995, pp. 620 626.
[21] Plunien, G., Müller, B., and Greiner, W., “The Casimir Effect,” Physics Reports,
Vol. 134, 1986, p. 87.
[22] Casimir, H. B. G., “On the Attraction Between Two Perfectly Conducting Plates,”
Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen, Vol.
51, 1948, pp. 793 795.
[23] Bordag, M., Mohideen, U., and Mostepanenko, V., “New Developments in the
Casimir Effect,” Physics Reports, Vol. 353, 2001, p. 1.
[24] Brown L. S., and Maclay, G. J., “Vacuum Stress Between Conducting Plates: An
Image Solution,” Physical Review, Vol. 184, 1969, pp. 1272 1279.
[25] Maclay, J., “Analysis of Zero Point Energy and Casimir Forces in Conducting
Rectangular Cavities,” Physical Review A, Vol. 61, 2000, p. 052110.
[26] Moore, G. T., “Quantum Theory of the Electromagnetic Field in a Variable Length
One Dimensional Cavity,” Journal of Mathematical Physics, Vol. 11, 1970,
p. 2679.
[27] Birrell, N., and Davies, P. C. W., Quantum Fields in Curved Space, Cambridge Uni
versity Press, Cambridge, U.K., 1984, p. 48, 102.
[28] Neto P., Maia, A., and Machado, L. A. S., “Quantum Radiation Generated by a
Moving Mirror in Free Space,” Physics Review A, Vol. 54, 1996, p. 3420.
[29] Jaekel M., and Reynaud, S., “Motional Casimir Force,” Journal of Physics (France)
I, Vol. 2, 1992, p. 149.
[30] Ford, L., and Vilenkin, A., “Quantum Radiation by Moving Mirrors,” Physics
Review D, Vol. 25, 1982, p. 2569; 36.
[31] Barton, G., “Perturbative Casimir Energies of Dispersive Spheres, Cubes and
Cylinders,” Journal of Physics A, Vol. 34, 2001, p. 4083.
[32] Neto, P., “Vacuum Radiation Pressure on Moving Mirrors,” Journal of Physics A,
Vol. 27, 1994, p. 2167.
[33] Jaekel, M., and Reynaud, S., “Quantum Fluctuations of Mass for a Mirror in
Vacuum,” Physical Letters A, Vol. 180, 1993, p. 9.
[34] Barton, G., and Calogeracos, A., “On the Quantum Dynamics of a Dispersive
Mirror. I. Radiative Shifts, Radiation, and Radiative Reaction,” Annual of
Physics, Vol. 238, 1995, p. 227.
[35] Casimir, H. B. G., “Some Remarks on the History of the So called Casimir Effect,”
The Casimir Effect, 50 Years Later, Bordag, M. (ed)., World Scientific, Singapore,
1999, pp. 3 9.
[36] Power E. A., and Thirunamachandran, T., “Zero Point Energy Differences and
Many Body Dispersion Forces,” Physics Review A, Vol. 50, 1994, pp. 3929 3939.
[37] Schwinger, J., DeRaad, Jr., L. L., and Milton, K. A., “Casimir Effect in Dielectrics,”
Annuals of Physics (NY), Vol. 115, 1978, pp. 1 23.
[38] Lifshitz, E. M., “The Theory of Molecular Attractive Forces Between Solids,”
Soviet Physics JETP, Vol. 2, 1956, pp. 73 83.
[39] Maclay, J., and Hammer, J., “Vacuum Forces in Microcavities,” Seventh Inter
national Conference on Squeezed States and Uncertainty Relations, Boston, MA,
4 6 June 2001. URL: http://www.physics.umd.edu/robot 32.
[40] Deutsch D., and Candelas, P., “Boundary Effects in Quantum Field Theory,” Phy
siscal Reviews D, Vol. 20, 1979, pp. 20, 3063 3080.
 THRUSTING AGAINST THE QUANTUM VACUUM 421

[41] Barton, G., “Perturbative Casimir Energies of Spheres: Re orienting an Agenda,”

International Conference of Squeezed States and Uncertainty Relations, Boston,
May 2000. URL: http://www.physics.umd.edu/robot.
[42] Serry, M., Walliser, D., and Maclay, J., “The Anharmonic Casimir Oscillator,”
Journal of Microelectromechanical Systems, Vol. 4, 1995, p. 193.
[43] Maclay, J., “A Design Manual for Micromachines Using Casimir Forces: Prelimi
nary Considerations,” Proceedings of STAIF 00 (Space Technology and Appli
cations International Forum), El Genk, M. S. (ed.): American Institute of
Physics, New York, 2000.
[44] Serry, M., Walliser, D., and Maclay, J., “The Role of the Casimir Effect in the Static
Defection and Stiction of Membrane Strips in Microelectromechanical Systems
(MEMS),” Journal of Applied Physics, Vol. 84, 1998, p. 2501.
[45] Buks, E., and Roukes, M.L., “Stiction, Adhesion, and the Casimir Effect in
Micromechanical Systems,” Physical Review B, Vol. 63, 2001, p. 033402.
[46] Decca, R., Lopez, D., Fishbach, E., and Krause, D., “Measurement of the Casimir Force
Between Dissimilar Metals,” Physical Reviews, Vol. 91, 2003, p. 050402.
[47] Lambrecht A., and Reynaud, S., Comment on “Demonstration of the Casimir Force
in the 0.6 to 6 mm Range,” Physical Review Letters, Vol. 84, 2000, p. 5672.
[48] Arnold, W., Hunklinger, S., and Dransfeld, K., “Influence of Optical Absorption on the
van der Waals Interaction Between Solids,” Physical Review B, Vol. 19, 1979, p. 6049.
[49] Chen, F., Mohideen, U., Klimchitskaya, G. L., and Mostepanenko, V. M., “Inves
tigation of the Casimir Force Between Metal and Semiconductor Test Bodies,”
Physical Review A, Rapid Communication, Vol. 72, 2005, pp. 020101 020104.
[50] Puthoff, H. E., Little, S. R., and Ibison, M., “Engineering the Zero Point Field and
PolarizableVacuum for Interstellar Flight,” Journal of the British International
Society, Vol. 55, 2002, pp. 137 144.
[51] Millis, M. G., “NASA Breakthrough Propulsion Physics Program,” Acta Astro
nautica, Vol. 44, No. 2 4, 1999, pp. 175 182; “Challenge to Create the Space
Drive,” Journal of Propulsion and Power, Vol. 13, No. 5, 1997, pp. 577 582; and
“Prospects for Breakthrough Propulsion from Physics,” NASA TM 2004 213082,
May 2004.
[52] Pinto, F., “Progress in Quantum Vacuum Engineering Propulsion,” Journal of the
British International Society, Vol. 39, 2006, pp. 247 256.
[53] Haisch, B., Rueda, A., and Puthoff, H. E., “Inertia as a Zero Point Field Lorentz
Force,” Physical Review A, Vol. 49, 1994, pp. 678 694.
[54] Maclay, J., and Forward, R. W., “A Gedanken Spacecraft that Operates Using the
Quantum Vacuum (Dynamic Casimir Effect),” LANL Arxiv Physics/0303108,
Foundation of Physics, Vol. 34, 2004, p. 477.
[55] Forward, R. L., “Extracting Electrical Energy from the Vacuum by Cohesion of
Charged Foliated Conductors,” Physics Review B, Vol. 30, 1984, p. 1700.
[56] Cole, D. C., and Puthoff, H. E., “Extracting Energy and Heat from the Vacuum,”
Physics Review E, Vol. 48, 1993, p. 1562.
[57] Rueda, A., “Survey and Examination of an Electromagnetic Vacuum Accelerating
Effect and its Astrophysical Consequences,” Space Science Reviews, Vol. 53, 1990,
pp. 223 345.
[58] Cole, D. C., “Possible Thermodynamic Violations and Astrophysical Issues for
Secular Acceleration of Electrodynamic Particles in the Vacuum,” Physics
Review E, Vol. 51, 1995, p. 1663.
422 G. J. MACLAY

[59] Lambrecht, A., Jaekel, M., and Reynaud, S., “Motion Induces Radiation from a
Vibrating Cavity,” Physics Review Letters, Vol. 77, 1996, p. 615.
[60] Dodonov, V., and Klimov, A., “Generation and Detection of Photons in a
Cavity with a Resonantly Oscillating Boundary,” Physics Review A, Vol. 53,
1996, p. 2664.
[61] Law, C. K., “Resonance Response of the Vacuum to an Oscillating Boundary,”
Physics Review Letters, Vol. 73, 1994, p. 1931.
[62] Villarreal, C., Hacyan, S., Jauregui, R., “Creation of Particles and Squeezed States
Between Moving Conductors,” Physics Review A, Vol. 52, 1995, p. 594.
[63] Plunien, G., Schützhold, R., and Soff, G., “Dynamical Casimir Effect at Finite
Temperature,” Physics Review Letters, Vol. 84, 2000, p. 1882.
[64] Boston, M., “Simplistic Propulsion Analysis of a Breakthrough Space Drive for
Voyager,” Proc. STAIF 00 (Space Technology and Applications International
Forum), January 2000, Albuquerque, NM, El Genk, M.S. (ed.) American Institute
of Physics, New York, 2000.
[65] Bernholc, J., Brenner, C., Nardelli, M., Meunier, V., Roland, C., “Mechanical and
Electrical Properties of Nanotubes,” Annual Review of Material Research, Vol. 32,
2002, pp. 347 375.
[66] Saito, T., Furuta, T., Hurang, J. H., Kuramoto, S., Nishino, K., Suzuki, N., Chen, R.,
Yamada, A., Ito, K., Seno, Y., Nonaka, T., Ikehata, H., Nagasako, N., Iwamoto, C.,
Ikuhara, Y., and Sakuma, T., “Multifunctional Alloys Obtained via a Dislocation
Free Plastic Deformation Mechanism,” Science, Vol. 300, 2003, p. 464. My
thanks to Steven Hansën for this reference.
[67] Eberlein, C., “Quantum Radiation from Density Variations in Dielectrics,” Journal
of Physics A: Mathematics and General, Vol. 32, 1999, p. 2583.
[68] Yablonovitch, E., “Accelerating Reference Frame for Electromagnetic Waves in a
Rapidly Growing Plasma: Unruh Davies Fulling DeWitt Radiation and the Non
adiabatic Casimir Effect,” Physics Review Letters, Vol. 62, 1989, p. 1742.
[69] Lozovik, Y., Tsvetus, V., and Vinogradovc, E., “Femtosecond Parametric Exci
tation of Electromagnetic Field in a Cavity,” JEPT Letters, Vol. 61, 1995, p. 723.
[70] Ford, L. H., and Svaiter, N. F., “Focusing Vacuum Fluctuations,” Physics Review
A, Vol. 62, 2000, p. 062105.
[71] Calonni, E., DiFiore, L., Esposito, G., Milano, L., and Rosa, L., “Vacuum Fluctu
ation Force on a Rigid 38 Casimir Cavity in a Gravitational Field,” Physics Letters
A, Vol. 297, 2002, p. 328.
[72] Bezerra, V., Klimchitskaya, G., and Romero, C., “Surface Impedance and the
Casimir Force,” Physics Review A, Vol. 65, 2001, p. 012111.
[73] Genet, C., Lambrecht, A., and Reynaud, S., “Temperature Dependence of the
Casimir Force Between Metallic Mirrors,” Physics Review A, Vol. 62, 2000,
p. 012110.
[74] DeWitt, B., Physics in the Making, Sarlemijn A., and Sparnaay M. (eds.), Elsevier,
Amsterdam, 1989, pp. 247 272.
[75] Leonhardt, V., and Philkin, T., “Quantum Levitation by Left Handed Materials,”
New Journal of Physics, Vol. 9, 2007, p. 254.
[76] Enk, S., “Casimir Torque Between Dielectrics,” Physical Review A, Vol. 52, 1995,
p. 2569.
 Chapter 13

Inertial Mass from Stochastic Electrodynamics

Jean-Luc Cambier
U.S. Air Force Research Laboratory, Edwards Air Force Base,
Edwards, California

I. Introduction
HE origin of inertia, the resistance of an object to a change in its velocity
T (i.e., acceleration), has been the subject of various hypotheses throughout
the development of modern physics. This resistance implies that a force must
be applied to provide acceleration to an object (Newton’s law) and the constant
of proportionality is the so-called (inertial) “mass.” One of the most famous con-
jectures regarding inertia is the Mach principle, which postulates that interaction
with the entire universe is responsible for this property of matter. Such arguments
of a quasi-philosophical nature played an important role in the development and
refinement of the important physical concepts of causality, inertial frames, invar-
iant measures, and eventually the equivalence principle between inertial mass
and gravitational mass. While this led to formidable advances in our understand-
ing of the fundamental properties of the universe, Einstein’s work did nothing to
explain why mass seemed to be an intrinsic property of matter in the first place.
The advent of quantum mechanics and quantum field theory led to further insight,
but also more unexplained phenomena, such as the existence of fields in a non-
vanishing energy ground-state, thus potentially filling the universe with so-
called “zero-point energy” (Chapters 1 and 18), while interaction with this
field on all scales could lead to effective particle masses that are seemingly infi-
nite. The origin of mass is, therefore, still an open field of research. If mass is
strictly the result of the interaction with background fields, then one may hope
to be able someday to modify these fields and change the mass. This could
have obvious implications for propulsion provided this manipulation can be
done in a practical and efficient way. Mass reduction would then allow a

This material is a work of the U.S. Government and is not subject to copyright protection in the
United States.

Senior Research Scientist, Propulsion Directorate Aerophysics Branch.

423
424 J.-L. CAMBIER

higher acceleration for a given applied force, and near-vanishing of the mass
would allow travel at near the speed of light.
This type of interaction can also proceed when the background field is a
random, purely classical field. This is the basis for the stochastic electrodynamics
(SED) model, and in particular the hypothesis by Haisch, Rueda, and Puthoff [1]
that inertial mass can be the result of such interaction with a fluctuating field. In
this chapter, this hypothesis is examined and compared with the modern relativis-
tic, quantum electro-dynamics (QED) theory. The intent is to give the reader an
explanation of the premise of this theory and to identify the most critical and
unresolved issues for further research. It should be pointed out that any strengths
or weaknesses identified here pertain only to this particular hypothesis rather than
the general theory of stochastic electrodynamics, or the general notion that inertia
might be fundamentally tied to quantum fluctuations. It is hoped that the analysis
offered here will alert future researchers to the details that need to be clearly
addressed to proceed along these lines.

II. Background
SED is a theory of the interaction of point-like charged particles with fluctu-
ating electromagnetic fields of the vacuum, also called zero-point fields (ZPF),
which Haisch, Rueda, and initially Puthoff, among others, have been proposing
for a number of years [1 7] as an explanation for the origin of inertial mass.
In the first version of the theory, Haisch, et al. [1] (hereafter called HRP) postu-
late that the coupling of the charged particle with such fields leads to a oscillating
motion, that is, an example of “zitterbewegung” (literally, “jittering motion”) [8].
SED is intimately tied to the decades-old problem of the electromagnetic vacuum
and radiation spectrum. In 1910, Einstein and Hopf [9] used a classical oscillator
model to demonstrate (in the nonrelativistic approximation) that the oscillator
coupled to a background radiation field, was subject to a retarding force pro-
portional to the velocity, F Rv, where†
 
4p 2 e 2 v0 d r
R¼ r (v 0 )  (1)
5mc2 3 d v0

and r(v0) is the energy density of the field at the resonant frequency v0 of the
oscillator. When the background field consists only of the vacuum field (the
thermal component being absent), the energy density is proportional to v3, and
the bracketed term on the right side of Eq. (1) vanishes; thus, in the absence of
thermal radiation (T ; 0) there is no frictional force. It has been shown indepen-
dently by Marshall [10] and Boyer [11] that the v3 scaling law is the required
form for the Lorentz invariance, and, therefore, the vacuum field does not
exert any force in any inertial frame; this is an important and necessary result,
because observers in inertial frames cannot be, by definition, subject to forces
and accelerations.

†
Units throughout are ESU (see Ref. 12); to convert to SI units, replace e 2 ! e 2/4p10.
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 425

The main objective of the HRP thesis was to generalize the Einstein and Hopf
results to consider noninertial frames, and attempt to obtain another retarding
force proportional to the acceleration, F mia. By doing so, the inertial
mass (the coefficient mi) would be explained in terms of a coupling to the fluctu-
ating background field of the vacuum. The HRP analysis is almost entirely based
on the work by Boyer [11,13 15], where the classical oscillator is also modeled
by the linear motion of a point charged particle (“parton,” in the terminology of
high-energy particle physics). After averaging over the high-frequency
oscillations and in the limit of low acceleration and nonrelativistic motion,
the Lorentz force is identified as a retarding force (after selecting the
appropriate sign) proportional to the acceleration. The inertial mass is given in
Ref. 1 as:
 
h vL
mi ¼ (GvL ) (2)
c2
where vL is the upper limit‡ to the integration of all electromagnetic modes of the
ZPF and G is a damping constant. Equation (2) appears to indicate that the mass is
completely obtained as a result of the interaction with the ZPF, and is the basis of
the claims of inertia as the result of a drag force from the vacuum. There are,
however, a number of issues related to the actual computations performed to
obtain the result in Eq. (2) above, and to its interpretation. This chapter is orga-
nized as follows; first the HRP theory, as mostly described in the seminal paper
[1], is summarized, with some key aspects and assumptions highlighted; second,
the theory is compared with the conventional approach of quantum field theory in
order to bring to light the similarities and differences; third, we discuss more
recent results, as well as a way forward, that is, what should be done to
achieve further progress; finally, we provide a brief conclusion.

III. Stochastic Electrodynamics

We principally discuss here the original model used by HRP, which is an
extension of the classical oscillator model in the presence of an oscillating elec-
tromagnetic, with radiation reaction§:
2e2
m0 _€r ¼ m0 v20 r þ _€r þ eE0 (3)
3c3
Here, r is the amplitude of oscillation along the direction of the applied field,
which we can choose as the ŷ direction without loss of generality. This equation
of motion leads to a damping oscillating motion, with

2e2
G¼ (4)
3 m0 c 3

‡
The cutoff can be expressed in terms of energy, frequency or mass: L ¼ h vL ¼ mLc 2.
§
See Sec. 17.8 of Ref. 12. This model describes the motion of a point... like, charged mass subject to
an applied electric field and a restoring force (harmonic potential). The r term is the radiation reaction
force.
426 J.-L. CAMBIER

being the well-known classical damping coefficient (the so-called Abraham

Lorentz constant), which yields the natural line width of spectral lines. The elec-
tric field is also along the ŷ direction, so that the vector notation can be ignored.
When the field frequency is v  v0, the system is, of course, driven to resonance.
However, if the field is randomly fluctuating in time, its Fourier transform
contains many different frequencies. In this case, the equation of motion
becomes a Langevin equation, where the driving term is described by a
random variable Ĕ, with a probability distribution function (PDF) such that
kĔl; 0, but such that its variance is nontrivial, that is, kĔ2l=0. The averaging
procedure (brackets k l) is over many cycles of the fluctuating quantity, which
is considered equivalent to the PDF average (ergodic hypothesis). HRP consider
this fluctuating field to be the ZPF, which contains a wide range of
frequencies extending, presumably, up to the Planck frequency:
 5 1=2
c
vP ¼   1:8  1043 rad=s (5)
hG
Therefore, unless the spectrum of the fluctuating field decays exponentially
fast in the high-frequency region, extremely high driving frequencies will be
present. These can easily be averaged over time scales of interest. An exponen-
tially decreasing function of frequency is the spectral density of a thermal
(Planck) distribution. However, the complete energy density function of the
electromagnetic field
  
v2 d v hv h v
r(v)d v ¼ 2 3  hv=kT þ (6)
p c e 1 2
contains the nonthermal, vacuum contribution hv/2. Integration over all
frequencies for this nonthermal contribution leads to a divergent integral (“UV
catastrophe”); the Planck frequency [Eq. (5)] acts as an upper limit to regularize
the integral, although for any practical purposes, this cutoff frequency is so large
that it can be considered as virtually infinite.
The reference frame S of the oscillator (characterized by the equilibrium
position R)~ is assumed to have a uniform acceleration a with respect to the
laboratory frame,} K. We choose the direction of motion of the frame to be
the x^ -direction. At any given time, one can also define an inertial frame K0 ,
which coincides with S but is related to the laboratory frame, by a Lorentz trans-
formation, the parameters of which are time-dependent (see Fig. 1).
Accounting for the uniform acceleration, the oscillator dynamics are given by:
 
2e2 a2
m0 r€ ¼ m0 v20 r þ 3 _€r  2 r_ þ eE0 (7)
3c c

where the time derivative is with respect to the proper time in the S
frame, r_ ¼ dr=dt. HRP determine first a solution of Eq. (7) for the velocity,
then separately compute the ensemble average of the net Lorentz force

}
In Ref. 1, the authors almost exactly replicate the analysis and notation of Boyer [14].
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 427

ŷ
K
S

–
R

x̂

Fig. 1 Schematic of relative frame transformation, where S is the frame attached to

the center of motion of the oscillator, and K is the laboratory frame. The oscillating
motion ( y-direction) is transverse to the relative velocity of the two frames
(x-direction). The S frame is uniformly accelerated (“hyperbolic motion”) with
respect to the laboratory frame.

acting on the system from this initial solution. From the harmonic oscillator
approximation, the magnitude of the velocity is v  eE0/m0v, which can be
recast in the form:
  2  2
v2 hv lc
ffi a   a  (8)
c2 m0 c2 l
where we have used the energy density E2v  hv/l3 for a single mode of the
ZPF, and the following definitions:
e2
Fine-structure constant: a ¼  (9a)
hc
h
Compton wavelength: lc ¼ (9b)
m0 c
Let us first assume that m0 is the observed rest mass of the electron, me. Because
the vacuum modes extend to very short wavelengths, presumably down to the
Planck length almost 20 orders of magnitude smaller than the characteristic
length in Eq. (9b) for the electron the term in parentheses on the right side of
Eq. (8) can be very large. Thus, the charged particle motion would be ultrarela-
tivistic and the model oscillator described in Eq. (7) would not be accurate, unless
the cutoff is arbitrarily limited by the physically observed energy scale, that is,
l  lc, or the particle mass m0 is very large. We should also point
out that Eq. (7) is taken directly from Boyer [14], and is valid in the limit of
small oscillating amplitude. A derivation of this equation (notably the reaction
force) from a covariant formulation is given in Appendix A to this chapter.
Equation (7) has two fundamental parameters: m0 and v0. The nature of these
parameters is not clearly specified in the HRP model, except that m0 is considered
the “bare” mass scale of the oscillator,†† while v0 is considered essentially


In HRP [1], p. 49, it is explicitly mentioned that only very high frequencies contribute to the
inertial mass.
††
This connotation of “bare” mass comes from quantum field theory, as explained in Section IV.
428 J.-L. CAMBIER

irrelevant by the authors. For a harmonic oscillator, v0 describes the properties of

the restoring force (stiffness), when combined with the oscillator mass m0.
Because there are no details given regarding the model and the nature of this
restoring force, these remain as free, adjustable parameters.
Equation (7) is solved by applying Fourier transforms. Again following
Boyer [14]:
ð1
1=2
r(t) ¼ (2p) dV  h(V)  e iVt (10a)
1
ð1
1=2 iVt
E0 (t) ¼ (2p) dV  S(V)  e (10b)
1

which, when applied to Eq. (7), yields the solution:

e S(V)
h(V) ¼ (11)
m0 v20  V2  iGV(V2 þ a2 =c2 )
where G is given by Eq. (4). The oscillating velocity is given by its time-Fourier
component:
e VS
v~ ¼ i (12)
m0 v0  V  iGV(V2 þ a2 =c2 )
2 2

Note that Eq. (10b) is a Fourier transform in the time/frequency domain only;
it describes the electric field at the center of the oscillator, that is, E0 (r ; 0,t) in
the S frame. There is no dependence in the position of the charge itself, that is,
dependence on r, whether in the initial dynamical Eq. (7), where the electric
field is evaluated at E0 (r ; 0,t), or in the fluctuating velocity. This is valid as
long as the spatial dependence of the electromagnetic fields is negligible
within the amplitude of motion of the oscillator (small oscillator approximation).
Again, because the model considers wavelengths of the fluctuating vacuum field
as small as the Planck length, this approximation is debatable.
The Lorentz force is obtained by taking the product of the velocity with
the magnetic field. For the latter, HRP use the following Fourier
transformation‡‡:
ð
E0 (0,t)  g(1  b) d3 k  H~ ZP  ei(kR vt) (13)

with H~ ZP ¼ h v=2p2 . The electric field in Eq. (13) is a simplified version of

Eq. (10) of HRP, assuming a single polarization in the ŷ-direction.
The magnetic field for this mode is pointing in the ẑ-direction (+) and is
related to Eq. (13) by a factor 1/c, which can be ignored when using the
natural system. Note that Eq.p(13) includes a Lorentz transformation (of par-
ameters b V/c and g ¼ 1= 1  b2 ) from an inertial frame (K) where the
vacuum mode energy density is defined (H̃ZP) to the inertial frame (K0 ) that is
instantaneously at rest with the uniformly accelerated frame S that contains the

‡‡
To avoid lengthy descriptions, we have simplified Eq. (8) from HRP by explicitly considering
the z component only and a single field polarization.
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 429

oscillator. This explains why the phase contains a term proportional to R

(distance to the center of the oscillator) and the time t (and not the proper
time). At this point, we should point out that the velocity Eq. (12) depends
only on the square of the acceleration; any effect of acceleration (expanding in
powers of a 2) from the solution of the oscillator equation of motion would there-
fore be independent of its sign, and could never contribute to a force proportional
to the acceleration, F mi . a. In fact, there is no reason to keep the term pro-
portional to a 2/c 2 in Eq. (7) if the only intent is to derive an expression up to the
first order in the acceleration. Close examination of the derivation presented in
HRP shows that the linear term in a is obtained only from the phase of the
Fourier transform in Eq. (13), that is, from the expansion of the distance:§§
R  12a . t 2. Thus, the force computed in the HRP model arises from a dephasing
between the oscillating velocity and the oscillating magnetic field. After some
further algebraic manipulation and introducing a cutoff frequency (an adjustable
parameter), HRP arrive at an expression for the force:

hvL
F ¼ (GvL ) a (14)
2pc2

and the term in front of the acceleration is thereby identified as the inertial
mass. Radiative damping actually plays no role in the HRP model, and it is some-
what misleading to express the solution in Eq. (14) in terms of the damping con-
stant. In fact, it is easy to see from HRP that the G term in Eq. (14) is obtained
only from the leading prefactor e/m0 of Eq. (12) and the charge e of the
expression for the Lorentz force: ev  B. The inertial mass in Eq. (14) can be
expressed as:

2 m2
mi ffi a L (15)
3 m0

In the absence of any other natural cutoff scale, one should have
mL ffi mP ¼ (h c=G)1=2 , the Planck mass. Note that in order to obtain the
known mass of the electron from Eq. (15), that is, mi ; me, the bare mass m0 must
be of the order of 1012 kg (1056 GeV/c2) more than 20 orders of magnitude
greater than the Planck mass. One could argue that this is of no significance,
because the bare mass is a free parameter, but this is well above any fundamental
mass scale imaginable.}}
In the HRP model, there is also no provision for computing the mass in an
inertial frame; the so-called inertial mass can be computed only when the oscil-
lator is accelerated. One must therefore assume that the “rest” mass, obtained in

§§
There are some discrepancies in Eqs. (13) and (1) regarding the expression for the distance R
see Appendix A although these are of no significance for the final results.
}}
One could make a reverse argument and consider m0 to be the Planck mass and determine the
cutoff energy scale from Eq. (15). This would lead to mL  1018 eV/c2, an intermediate scale well
above the projected scale of super symmetry breaking and well below that of grand unified theory
(GUT), see, for example, Ref. [17]. However, because Eq. (15) is based on a classical model of
electrodynamics, there is very little relevance to the current state of the art in quantum field theory.
430 J.-L. CAMBIER

an inertial frame when a ; 0, is also the same. However, it is physically counter-

intuitive, within the context of this classical model, to imagine how the “parton”
with an initial bare rest mass of m0c 2 can lose energy when subjected to the high-
frequency chaotic motion induced by the ZPF in order to reach the observed
value of mec 2  0.512 MeV. In fact, simple arguments regarding the coupling
of a classical oscillator with a fluctuating field yield a resulting rest mass with
a similar scaling as the HRP result, albeit with a slightly different coefficient
(see Appendix B of this chapter). Therefore, the mass “correction” due to the
coupling of the oscillator with the vacuum field is always positive, and can be
extremely large if the cutoff energy is also large.
If the cutoff frequency is set instead to a value such that, with a proper
combination of the bare mass the result matches the observed electron mass,
 vc )2
2 (h
me c2 ffi a (16)
3 m0 c2
a new adjustable parameter is being introduced, which is equivalent to the “size”
of the oscillator. Indeed, a cutoff frequency vc indicates that all ZPF fluctuations
with wavelengths much smaller than the oscillator size c/vc do not contribute to
a net force due to the averaging process over random phases. However, the oscil-
lator size is now introduced as an ad hoc parameter in the model, making its use-
fulness questionable. The charged parton being coupled to the ZPF modes, one
can argue that any wavelength shorter than its fundamental length scale (i.e.,
the “extent” of the parton), the Compton length lc,0 h /m0c would not effec-
tively contribute to the jittering motion. Thus, the cutoff scale becomes the
same as the original, that is, mL  m0; any attempt to “explain” the electron
mass scale from a more fundamental scale (such as Planck mass) is, within the
context of this model, highly problematic. Note also that any high-mass parton
constituting an electron would, when annihilating with an “anti-parton” of a posi-
tron, result in much higher energy gamma rays than observed.††† Therefore,
the model is faced with a conundrum: either there exists a rest mass that is
excessively high and unphysical, or a particle size is introduced ad hoc to
match observations and the model essentially predicts and explains nothing.‡‡‡
One should point out that HRP did consider the case of m0 , 0 (their Appen-
dix B); in that case, one could still start with a large (negative) bare mass and
obtain a small (observed) finite mass after accounting for the kinetic energy of
zitterbewegung, while a positive inertial mass can be obtained by choosing the


As mentioned earlier, this scaling relies on a non relativistic approximation, which breaks
down when the cutoff energy is much larger than the physical rest mass energy and a different,
linear scaling is obtained when taking into account relativistic effects. Although HRP have not con
sidered the relativistic case, one can reasonable assume that a similar change in scaling could be
obtained in that model.
†††
This discrepancy could not be explained away by hand waving arguments that the jitter
motion would be reabsorbed by the ZPF, because the problem is with the rest mass energy, m0c 2,
and not the kinetic energy of zitterbewegung.
‡‡‡
It is somewhat ironic that in a response [6] to comments by Woodward and Mahood, Haisch and
Rueda make a critique of gravitational theory of inertia (Mach’s principle) as containing a circular
argument, while the authors appear unable to recognize the circularity of their own model.
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 431

opposite sign in the solution for the Lorentz force. A negative (unobservable)
bare mass is not a priori impossible in quantum field theory (QFT) because the
mass parameter is related to the shape of a field potential and a cosmological
phase transition. For example, one could consider in a classical model a negative
bare mass as arising from a strong (nonperturbative) gravitational solution
[16,18], albeit of doubtful relevance to physical space. However, this only
raises additional and even more fundamental issues to the problem, with little
perspective of resolution within the scope of classical theory.
Haisch and Rueda and others have attempted to develop the SED theory
further [2 7], but some of the problems mentioned above have persisted.
Abandoning an explicit model such as the damped oscillator, Haisch and
Rueda considered the interaction between the ZPF and a particle of a given
size (volume V0) and with a given dimensionless factor, h(v). The inertial
mass is now:
ð hv3
V0
mi ¼ 2 h(v) 2 3 d v (17)
c 2p c

Therefore, the model difficulties regarding ultraviolet divergences are swept

under the rug by assuming a form factor h(v). This approach would be viable
if the particle had a structure,§§§ which could be determined experimentally;
for example, the structure of the proton in terms of partons (later to be identified
as the quarks of the standard model) was a primary object of experimental study
during the 1970s. However, as far as we know [19] the electron has no structure
up to a scale of approximately 50 GeV; this makes the upper limit of the integral
in Eq. (17) much larger than the observed rest-mass of the electron. Thus once
again, we are faced with a disparity of scales, that is, one must artificially set
the free parameters of the theory to match observed values.
Among recent work, one must mention the use of a covariant approach to
compute the inertial mass [3] and a “quantized” version [4]. However, the con-
nection between rest mass and inertial mass is still not being made. For
example, Sunahata [4] claims that the particle, seen from the laboratory frame,
contains both a mechanical momentum p~ 0 ¼ g m0 ~v and a ZPF momentum, the
latter becoming a function of time as the particle is being accelerated. This
separation of variables is not consistent. The total momentum after acceleration
can be written as:
ð
~
P(t þ Dt) ¼ p~0 þ p~zp (t þ Dt) ¼ p~0 þ p~zp (t) þ mi a~dt (18)

However, the velocity has now changed, and obviously the “mechanical”
momentum should now be p~0 ¼ g m0 (~v þ a~ Dt). Therefore there cannot be a
separation between a “mechanical” momentum (with an unknown mass, m0)

§§§
Replacing a form factor (i.e., distribution of internal constituents) by a frequency dependent
permittivity makes no difference; the latter also implies the existence of a substructure with charac
teristic time and length scales.
432 J.-L. CAMBIER

and a ZPF contribution. There are other problems as well. In the HRP model, the
linear term in the acceleration a~ arises from the expansion of a phase difference
between the electric and magnetic field Fourier components. In the more recent
work by Rueda and Haisch [3] and Sunahata [4], the effect of the acceleration
arises from the Lorentz transformation of the field amplitudes. Indeed, by switch-
ing to a semi-quantized version of the SED, the vacuum expectation values of the
ZPF involve the product of one field with the Hermitian of another (in order to have
a nonvanishing contribution ,0ja . aþj0. of the creation and destruction oper-
ators). This results in a phase factor exp[iQ(k) iQ(k0 )] which, when combined
with a delta-function d(k~  k~0 ), can be eliminated. The mechanism by which the
acceleration modifies the ZPF Poynting vector (of which the only relevant com-
ponent is in the x-direction, i.e., EyBz) is very suspect. To summarize, Rueda
and Haisch obtain field amplitudes of the form:
0 1 0 1
0 0
~ vl
E 00 ~ 00vl ¼ @
¼ @ g (1  bk^ x ) A  H~ ZP (v), B 0 A  H~ ZP (v) (19)
0 g (k^ x  b)

Integration over the wave vectors keeps only terms of even order in kx, yielding
a different expression for the Poynting vector in the new frame of reference, which
is interpreted as a resistance to the acceleration. However, Eq. (19) can be obtained
for any Lorentz transformation. Thus, the model would predict a reaction force for
a particle in inertial motion, which is of course unphysical. In fact, the value of the
Poynting vector of the vacuum fields was already computed by Boyer [13] as a two-
point correlation function, and was found to be identically zero. Thus, the results of
Rueda and Haisch [3] are in contradiction with Boyer’s. Our own calculations
(see Appendix C of this chapter) verify Boyer’s results and the identically null
Poynting vector.
The recent work by Sunahata [4] has also been hailed as a major step forward by
using a quantized version of the ZPF. However, this is somewhat misleading,
because the nonvanishing expectation value of the ZPF is itself a result of
quantum theory, used in a classical model. Sunahata compared classical evaluations
of the ZPF energy density and ZPF Lorentz force with a quantized version of the
two-point correlation functions; this is not a full quantum treatment, because the
matter fields would also need to be quantized, and an ad hoc symmetrization of
the operator ordering was introduced (instead of time ordering). In order to better
appreciate the similarities and differences between classical and quantized versions,
it is worth examining the state-of-the-art in conventional theory, that is, a fully
quantized, relativistic theory: quantum electrodynamics or QED.

Fig. 2 External lines in QED.

 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 433

Fig. 3 Propagator lines in QED.

IV. Quantum Electrodynamics

In quantum field theory (QFT), where matter and fields are both described at
the quantum level (“second quantization”), interactions with fluctuations of the
vacuum field are a fundamental aspect of the theory. While a complete and rig-
orous description of QFT can be found in various monographs [20,21], it is worth
presenting here a simplistic summary of the essential elements of the theory. We
consider only the theory for quantum electrodynamics (QED), which describes
the coupling of charges to electromagnetic fields at this fundamental level.
There are some very simple rules to observe, which are related to the so-called
Feynman diagrams (Figs. 2 8):
1) Any observable state (whether field or matter) is always physical and can
be described as an “external” line, as shown below. The fact that the state is phys-
ical implies that momentum and energy conservation are applied, and that all
masses are the physical masses observed; these states are said to be “on-shell”
(Fig. 2).
2) Virtual states can be generated and propagated from one point to another;
these are called “propagator” and their momentum and energy can take a range of
values (the arrows shown on propagator lines in Fig. 3 are actually immaterial).
3) The coupling between a charge and a field is described by a three-leg
vertex (in QED), and is associated with one unit of charge, or coupling constant:
conservation of momentum-energy is applied at each vertex (Fig. 4).
4) Only topologically different diagrams need to be considered.
For example, Fig. 5 describes to the lowest order (Born approximation) the
scattering of two particles (A and B) by the exchange of a virtual photon, result-
ing in two outgoing particles A0 and B0 . The combination of the delta functions at
each vertex ensures that energy and momentum are conserved between the initial
and final products.
Given the QFT rules described above, one can also construct the diagram
in Fig. 6. In this case, a charged particle (electron, top solid line) is created at

Fig. 4 Vertex diagram (coupling) in QED.

434 J.-L. CAMBIER

A' B'

A B
Fig. 5 Lowest-order fermion scattering in QED.

point O along with an anti-particle (positron, bottom solid line), while the inter-
action between the two is mediated by a photon. All particles in this diagram are
virtual, that is, are not required to have positive energy or mass. In fact, one can
consider this process as the temporary excitation of an electron from below the
Fermi level of a condensed state, thus leaving a positive “hole” inside the con-
densed state. This process is therefore a vacuum fluctuation, and is a QFT gener-
alization of the Heisenberg principle. Note that because there are no external lines
in Fig. 6, that is, there are no observables; although these fluctuations can occur at
all times, they are inconsequential.}}} For such vacuum processes to have an
observable effect there must be external lines associated with the diagram.
Consider now the diagram obtained by splitting the bottom solid line in Fig. 6
into two external lines (Fig. 7).
This time, the process describes the interaction of an electron (incoming line at
left) with the vacuum field, during which a virtual electron (e ) is created, and
after coupling to the vacuum field again, a real (“on-shell”) electron is
produced again. Although Fig. 7 appears benign, it does have important conse-
quences, leading to a mass renormalization of the electron. Note that the
process is of order e 2, the square of the electron charge (there are two vertices),
or e 2/h c, the fine-structure constant. More complicated diagrams can be

e–

O P

e+
Fig. 6 Vacuum fluctuation diagram in QED.

}}}
This distinction is critical to the prospects of “extracting” energy from the vacuum: In QED,
while the fluctuations of the vacuum (i.e., interaction with temporary fields) can have a profound
and measurable effect, there is no known mechanism (i.e., diagram) that leads to net energy extraction.
Even the impact on gravity requires, at the fundamental level, external lines corresponding to gravity
field (gravitons), and the need for a quantum theory of gravity.
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 435

e*

Fig. 7 Electron self-energy diagram.

constructed as well; for example there can be repeated emission/absorption of the

virtual photons, leading to an expansion in powers of a, or there can be an inter-
action with another particle while the electron is in a virtual (“off-shell”) state, as
shown in Fig. 8. When the particle at the top of Fig. 8 is an atomic nucleus, the
diagram describes the Lamb shift of spectral lines. Because a ,, 1, one can con-
struct a formal expansion in powers of a, that is, an expansion in the number of
vertices. Using this perturbative approach, quantum electromagnetic properties
have been calculated to an exquisite degree of precision, making QED one of
the most accurate and successful theories known to date.
However, there are some subtle conceptual difficulties with QFT. Note that
the self-energy diagram (Fig. 7) contains a loop, and because the virtual particles
can have any value of momentum and energy, there is an integration over the
momentum-energy values corresponding to each loop. The problem is that diver-
gences may occur during this integration. In particular, the diagram (Fig. 7) is
UV-divergent and a cutoff must be introduced. This is basically the same
problem as in the classical model of ZPF interaction ; i.e., all vacuum
modes can interact with the particle, up to the Planck frequency.

e*

Fig. 8 Renormalized coupling diagram (Lamb shift).


An important difference is that in QFT the divergences are only logarithmic.
436 J.-L. CAMBIER

The solution to this problem, as distressing it may appear, is a balance between

two mass parameters. In the initial QFT formulation, the “mass” of particles
appears as a specific term in the field Lagrangian, for example, mcc  , where
c is a Dirac field describing an electron. The self-energy diagram contributes a
mass correction dm (as well as a rescaling of the wave function or field amplitude,
which does not need to be considered here), proportional to ln L, where L is the
cutoff energy used to regularize the loop integral. Thus, there is a so-called
counter-term dm  cc in the Lagrangian, and the overall contribution to the
Lagrangian should be a combination of a “bare” mass m0 (which is not observa-
ble) and the counter-term, such that the observable mass is me m0 þ dm. Two
important remarks are in order:
1) There is an infinite number of possible self-energy diagrams of the type
shown in Fig. 7, where additional vertices and loops can be added, yielding an
expansion in powers of a (according to the number of vertices V) as well as an
expansion in the number of loops (L), with each loop corresponding to an inte-
gration over momentum and energy. It would seem that one would need a
similar infinite number of counter-terms in the Lagrangian; however, this is
not the case for well-behaved theories. This is an important criterion in dis-
tinguishing candidate theories, and a QFT is said to be renormalizable if there
is a finite number of such counter-terms.
2) Although the bare parameters seem arbitrary and a new circular argument
may be at hand (that is, the bare mass being adjusted such that the complete result
matches observations), the bare mass can actually be a product of a higher-level
theory. For example, “masses” can be generated†††† from a QFT with a higher
degree of symmetry (Higgs mechanism). The bare mass would then have a scale
comparable to the energy scale at which this breakdown of symmetry occurs
(similar to a phase transition in condensed matter), but the so-called “radiative cor-
rections” (Fig. 7) would bring the observable mass to a different value. In this case,
the cutoff energy scale L would presumably be of the same order as this symmetry
breakdown scale. Furthermore, it is important to point out that the QED results are
based only on the method of perturbation, the only approach currently available to
perform this type of calculation. In addition to the mass, the other parameters of
the theory (in particular a, the coupling strength) also are being renormalized.‡‡‡‡
The overall behavior of the parameters follows a set of nonlinear transform-
ations the “renormalization group” [23] as function of energy scale. This beha-
vior plays a critical role in allowing the unification of all forces of nature.
The result from QED is a counter-term [20, Sec. 7.42]:
 
3a L2 1 3a L
dm ffi m ln 2 þ  m  ln (20)
4p m 2 p
2|fflfflfflfflfflffl{zfflfflfflfflfflffl}
m
L!1

where L is the energy cutoff and m is the physical, observable particle mass.

††††
Even negative masses.
‡‡‡‡
See also the discussion in [32] Sec. 11.10.
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 437

Thus, although there are similarities of concept between QED and SED, there
are also important differences:
1) QED is a more complete theory, which is fully quantized, relativistic, well
developed, and highly accurate. SED was until recently a classical model based
on some questionable assumptions and mathematical solutions.
2) Both SED and QED rely on the concept of “bare” parameters, which
become renormalized by interaction with the vacuum field (virtual particles).
The radiative correction in QED leads to an additive term (me m0 þ dm),
while SED predicts a multiplicative correction (for example, me  m2P/m0).
3) QED predicts logarithmic divergences, while SED predicts much more
severe quadratic divergences (m2P) in the nonrelativistic case and possibly
linear in the relativistic case.
4) The radiative corrections in QED yield renormalization of the rest mass,
but SED predicts the mass only in the case of acceleration (that is, inertial
mass), because the calculations in HRP depend on the existence of a noninertial
frame.
These points are debated further in the next section. Item 4 listed above raises
another question: does QED also predict a mass shift for accelerated particles?
The answer is yes, although the effect is exceedingly small. QFT can be extended
to fields at finite temperature; formally, this is done by extending the theory from
Minkowski space to Euclidean space with an imaginary time variable b now
associated with 1/kT. The Feynman diagram rules are modified accordingly
and include the known quantum distribution functions at finite temperature,
that is, Bose and Fermi statistics, for the propagators [24]. An important aspect
of finite-temperature QFT is the fact that the vacuum now contributes external
lines in the Feynman diagrams. These lines are only present because the
vacuum now has a thermal distribution of photons, that is, the number operator
(nv a þ v av) operating on the vacuum state j0. is non-zero. The presence of
real (thermal) vacuum photons plays a key role in removing infrared divergences
when computing scattering cross-sections (see [25]). The self-energy diagram at
finite-temperature also has a temperature-dependent contribution [25 27] of the
order dm  (T 2/m).
Davies and Unruh [28,29] showed the equivalence between acceleration
and finite-temperature by examining the correlation functions (i.e. propagators
in QFT) in an accelerated frame. The result is that the quantum vacuum in an
accelerated frame obtains a thermal distribution with a temperature:

kTU ¼ h a=2p c (21)

Therefore, the complete formalism and results of finite-temperature QFT are

directly applicable to the case of the accelerated frame, by simply using the
Unruh-Davies temperature in Eq. (21) in the results. Note the scale of the temp-
erature in Eq. (21): It would take an acceleration of 2.5  1020 m/s2 to yield a
temperature of 1 K; for this reason, the principal area where this effect has
been studied is black-hole physics. Obviously, the corrections due to the
Unruh Davies effect are exceeding small, but could potentially be observed in
future laboratory experiments [30 32]. Finite-temperature QFT effects have
438 J.-L. CAMBIER

significant impact mostly in the early times after the big bang, when the universe
was still at very high temperature.
It should also be pointed out that the effect of a uniform acceleration on the clas-
sical vacuum electromagnetic field was also computed by Boyer [13]. This leads to
a simple correction to the rest mass that is proportional to the Unruh Davies temp-
erature, that is, to the acceleration (see Appendix B). The finite-temperature QED
renormalization yields, again, to a logarithmic correction, and one could simply
use the Unruh Davies temperature in the final result to obtain the effect of a
uniform acceleration. Again, the impact is completely negligible except for
exceedingly large acceleration (i.e., near back-hole horizon events).

V. The Way Forward

The SED model of vacuum-induced inertia can be summarized by the
following observations:
1) The HRP model of SED is based on a classical oscillator model with radi-
ation reaction force and with uniform acceleration. The principal result of the
model is a resistance force that is claimed to be proportional to the acceleration,
with the constant being identified as the inertial mass. Higher-order terms (i.e.,
a 2t 2/c 2) are not computed, but would likely remain in a more complete solution,
thus potentially leading to deviations from Newton’s Law.
2) It is claimed in HRP that the SED is essentially a relativistic effect and
HRP carried a complex and cumbersome relativistic treatment throughout most
of the calculations, yet a nonrelativistic approximation is essential to the validity
of the model and is used throughout. In fact, the analysis could have been carried
completely with a nonrelativistic approximation, because the inertia effect is
obviously observed at low accelerations. Similarly, the examination of the math-
ematical treatment shows that the radiation dampening of the oscillator, the term
proportional to G in the denominator of Eq. (12) is also essentially irrelevant, and
that the result is obtained as a result of phase difference between the Fourier
components of the velocity and magnetic fields expressed in different frames
of reference.
3) Another argument [3] for the generation of inertial mass based on a
non-zero expectation value of the Poynting vector of the vacuum field is at
odds with prior calculations by Boyer [13], which show a vanishing contribution
(see Appendix C of this chapter).
4) The inertial mass depends on a bare mass m0 and a cutoff frequency vL,
which are both free parameters in the theory. Depending on the choice of this
cutoff frequency, the model either appears to yield unphysical results or has no
real predictive capability.
5) The model used a small oscillator approximation (no field variation in the
transverse direction) while the integration over ZPF modes includes very short
wavelengths. The model also decouples the oscillating motion from the accelera-
tion in the longitudinal direction due to the Lorentz force, essentially a nonrela-
tivistic approximation that is not valid for the high frequency modes.
Further extensions of the model “removed” some of the problems of the
original classical oscillator model of HRP by keeping the form factor adjustable;
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 439

however, disparity of mass scales remains a problem that prevents SED from
making any mass predictions from any more fundamental scale.
The SED theory, whether the original classical oscillator model or more recent
versions, lacks rigor and contains some errors as well as unnecessary compli-
cations that obscure much of the physics and the analysis. Nevertheless, there
remains the potential for some advances in the theory, which can be fruitful.
The original intent of the theory, i.e., determination of the inertial mass as a
result of coupling to vacuum energy fluctuations, may have been of interest,
but this objective appears to be lost in circular and philosophical (or meta-phys-
ical) arguments. Classical models of electron self-energy have been developed
for many decades,§§§§ even after the discovery of quantum mechanics and
quantum field theory. Although classical models can still bring some interesting
insight into the physics of a problem, as shown for example in the work by Boyer,
one must be aware of the fundamental limitations of the approach. In particular,
classical theory breaks down at small length and time scales, that is, when their
combination becomes of the order of the Planck constant.}}}} By contrast, QFT
and in particular QED is a fully relativistic, quantum theory that is extremely well
validated. One can argue that, conceptually, there are still some remaining diffi-
culties with QFT, and that it is prudent to re-examine the differences between
SED and QED (listed at the end of Section IV) before drawing conclusions on
ways to improve SED.

1) QED is a more complete and accurate theory: This is certainly the case,
because SED relies mostly on classical and nonrelativistic approximations that
are not quite valid. Thus, one can reasonably assume that with continued progress
in calculating SED effects, such as quantization and inclusion of relativistic
effects, SED would become better developed and more accurate. However, it
begs the question as to the fundamental difference between the two theories:
QED already includes full quantization and relativistic effects, and already
accounts for the coupling to vacuum fluctuations, doing so with high precision
(e.g., Lamb shift, magnetic moment). What then would a quantized, relativistic
version of SED offer? If such a theory yields the same result as QED, what
would be the difference, conceptually or computationally? If it predicts different
results, either SED is invalidated (e.g., does not reproduce experimental data), or
it predicts additional results that QED cannot. For example, the treatment by
HRP is obtained as an expansion in terms of the acceleration; higher-order
terms (a 2) are also present but not yet computed. Would such deviations from
Newton’s Law be measurable?
2) Treatment of bare parameters: Pending a grand unification theory or
validated string theory, the bare mass of the particles in QFT have a certain
degree of arbitrariness ; however, there are a number of important effects

§§§§
The reader is urged to consult the book by P. Milonni [33] for a review on past and current
models.
}}}}
Classical physics can be formally recovered from quantum mechanics in the limit h ! 0.

One can argue that some masses can be explained by a Higgs mechanism, but it mostly defers
the question to another level (i.e., why the symmetry breaking of a higher level QFT would occur at a
particular scale) until the ultimate theory can be obtained.
440 J.-L. CAMBIER

in QFT that make the problem conceptually less difficult; for example, the fact that
mass parameters can be associated with phase transitions; divergences are only
logarithmic (see items); fundamental symmetries (e.g., SUSY) can be at hand to
cancel divergences. As a classical model, it is much more difficult to interpret
the incongruity of bare parameters; for example, there is no mechanism for a clas-
sical particle to start with a very high rest-mass and end up with a lower total
energy when adding high-frequency oscillating motions, and the concept of nega-
tive mass in classical theory presents a number of difficulties. Therefore, it would
appear that either one needs to have a complete revolution in physics by reverting
back to a reinterpreted classical view of nature, or SED needs to graduate and
become a quantum theory. There appears to be some minor progress in that direc-
tion [4], but in that case, one is then led back to the issues mentioned in item 1.
3) Nature of divergences: One could argue that the exact nature of diver-
gences is not important, because these divergences are removed by setting appro-
priate values of the bare parameters of the theory. However, a logarithmic
divergence is much more appealing, especially while attempting to construct a
more fundamental theory. The renormalization of the bare parameters (mass,
charge, etc.) of a theory has become an essential aspect of modern QFT through
the use of renormalization group (RNG) theory [23]. Here, RNG is used explicitly
to predict at what energy scale the coupling parameters become similar and when
symmetry breaking of a higher-level theory may occur. By contrast, SED cannot
make such predictions, does not offer any explanations for the cutoff values or
form factors, and, generally speaking, yields much more severe divergences.
Again, one can argue that SED is still in its infancy, and that further developments
along the lines of quantization may yield more information. Although this is poss-
ible, one is led to wonder at what point SED would become relevant, and what
would then be the difference from QED, leading us back to items 1 and 2.
4) True effect of acceleration: It is puzzling that although the effect of
acceleration on the vacuum has already been examined by Unruh and Davies,
it is not being used by Rueda and Haisch in the development of SED.
One could use simple (even simplistic) arguments combining the classical
models and the basic effect of a transformed vacuum field spectrum as a result
of acceleration (Appendix D of this chapter) to examine the impact of accelerated
motion on mass and self-energy. Because the acceleration essentially brings a
thermal component††††† to the vacuum field, one could simply use the complete
(thermal þ zero-point) energy spectrum in the interaction with a classical oscil-
lator. The effect can also be computed easily in QED, using well-known
results from finite-temperature QFT, as described earlier.
We suggest that further development of SED theory would require the
following:
1) Computing with SED some effects of the vacuum fluctuations that can
be compared with multi-loop QED and known data (e.g., Lamb shift, electron
magnetic moment).

†††††
P. Milonni ([33], p. 64) describes it as a “promotion” of the quantum modes to the level of
thermal fluctuations. This effect has also been shown within the context of SED [15].
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 441

2) Computing higher-order effects (a 2) that are not predicted by conventional

theory and determine if experimental validation of those effects is possible.
3) Examining fundamental issues regarding the connection with the Unruh
Davies effect and similarities and differences with QED, in a clear and convin-
cing manner.
There is, of course, the problem of experimental validation. Any theory must
eventually be able to satisfactorily reproduce experimental results, explain obser-
vations that have otherwise remained beyond the capabilities of competing
models, and/or predict new observable effects. In this aspect, QED has a signifi-
cant advantage over SED. It may be that, to distinguish between QED and SED,
one may be led to a new physical regime, notably in the levels of acceleration.
The recent development of ultra-intense, ultra-short lasers may provide an
avenue toward high acceleration levels in a laboratory space, and could be
used if SED can be made to predict effects to higher orders in acceleration, for
example. However, the problem may not be so simple because the laser field
itself creates a zitterbewegung and an effective mass that can drown measurable
effects from the ZPF. Another approach is to seek the high accelerations near
astrophysical objects, but data interpretation is also subject to a lot of uncertainty.
Therefore, there does not appear to be an easy solution to the problem of exper-
imental validation,‡‡‡‡‡ especially if one must be able to make the clear distinc-
tion between expected QED effects and SED effects. Nevertheless, it is
absolutely critical for SED to be computable (i.e., does not rely on ad hoc par-
ameters that are adjustable), and predictable (i.e., able to make quantifiable pre-
dictions for feasible experiments).

VI. Conclusions
Of course the manipulation of inertial mass opens up many interesting
possibilities for space propulsion and access to space. As always, the benefits
have to be weighed against the costs; for example, one must determine
whether it is advantageous to expend a certain amount of energy in reducing
the inertial mass of a spacecraft, versus the energy needed to provide the
kinetic energy to that same spacecraft. Assuming the SED theory to be
correct, such control of the inertial mass must inevitably come from manipulation
of the vacuum field. To this day, there is no definite mechanism that allows us to
perform this on macroscopic scales. Even if one were to induce changes in the
vacuum field at precise locations and times, the Heisenberg principle
implies that such changes are bound to be microscopic, either on the energy or
time scale. Thus, manipulation of the inertial mass would require related,
major advances in our understanding of the quantum vacuum in general, and
extreme cooperation on the part of the laws of physics; that is, the discovery
of methods for field manipulation on macroscopic scales. Although we are

‡‡‡‡‡
We must emphasize that commonly cited experiments that are based on the interaction with
vacuum fluctuations, such as Casimir force and related cavities, do not necessarily provide a validation
of SED, since current state of the art quantum theories are readily used to predict such effects. There
must be a clear and measurable difference between QED and SED predictions.
442 J.-L. CAMBIER

now within the realm of fiction, one may suspect that this is likely to be either
absolutely impossible (i.e., being limited to microscales) or extremely costly
(requiring, for example, to generate copious amounts of anti-Higgs or other par-
ticles with similar effects). Nevertheless, it is much too early to have a definite
opinion on the subject, and it is best to wait until further evidence can be
obtained, from current efforts at unified theories or other, less conventional
approaches. One may be able to learn from surprising effects in condensed-
matter, which has often been an important application of QFT (see, for
example, Ref. 34).
The potential of SED for explaining fundamental particle properties is some-
what doubtful. The current state of the theory is not especially promising, being
afflicted by questionable mathematical approaches and physical models, and
caught in circular arguments regarding particle size, cutoff, form factors, etc.
Significant advances are still possible, and guidelines have been provided.
Until a number of significant steps have been accomplished (i.e., comparison
with QED, computation of known effects, prediction of new ones), SED will
remain at the fringes of scientific research with little prospect for major
impact. On the other hand, SED could potentially play a role in other areas,
such as plasma physics (e.g., noise generation, turbulence, chaotic heating),
astrophysics, and condensed matter.

Appendix A: Relativistic Transformation to a

Uniformly Accelerated Frame
Let us start with covariant formulation of the radiation reaction force [22]:

2e2 d2 um um
Fm ¼  R (A1)
3c ds2 c2

with

2e2 dun dun

R¼ (A2)
3c ds ds

and
 
gc 1
um ¼ with g ¼ q (A3)
g~v
1  V~ 2 =c2

Here, V~ ¼ dX=dt
~ is the velocity of the accelerated frame S with respect to the
laboratory frame, ds2 ¼ c2 d t2 ¼ c2 dt2  d~x2 is the invariant measure in a
Minkowski metric [i.e., hmv (1, 1, 1, 1), where t is the proper time], and
(t, ~x) are the coordinates of the particle in the laboratory frame.
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 443

The motion of the accelerated frame S, as seen from the laboratory frame, is:

" #
dX d V at
V¼ ! p ¼a!V ¼p !g
dt dt 1  V 2 =c2 1 þ (at=c)2 (A4)
q
¼ 1 þ (at=c)2

where a is the constant acceleration. From the definition of the invariant measure,
we have:

ð  at
dt c
t¼ p ! t ¼ sinh (A5)
1 þ (at=c)2 a c

which leads to:

at V at
g ¼ cosh and b ¼ ¼ tanh (A6)
c c c

The motion of the accelerated frame, after transformation to the proper time
variable, is:

c2 h  at i
X(t) ¼ cosh 1 (A7)
a c

which in the nonrelativistic limit (at , , c) yields X  at 2/2 as expected.

Note that HRP used a different expression [1, Eq. (9c)] for the distance X(t)
(c 2/a) cosh (at/c). This leads to a distance (and a phase in the Fourier transform
of the vacuum fields) that becomes infinite as the acceleration becomes very small,
and does not make the two frames coincide at t 0. By contrast, Boyer obtains the
expression in Eq. (A7) but incorrectly applies it when computing the phase shift in
the Fourier transforms. In both cases, this error does not change the results because
it is equivalent to a constant at finite acceleration.
Let us now compute the reaction force using the following definition of the
four-velocity, for an oscillator motion in the ŷ direction only:

0 1 0 1
gc gc
B gV C B gV C
um ¼ B C B C
@ g  dy=dt A ¼ @ r_ A (A8)
0 0
444 J.-L. CAMBIER

where we have now identified r_ ¼ dy=d t. The proper time derivative is:
0 1
a sinh (at=c)
dum B a cosh (at=c) C
¼B
@
C
A (A9)
dt r€
0

We have then

dun duv a2 þ r€ 2
¼ (A10)
ds ds c2

and the transverse reaction force is

 
2e2 ... a2 þ r€ 2
Fy ¼ r  _
r (A11)
3c3 c2

There is a significant difference between the formulation ...

of Boyer [14] and
HRP. Boyer dismisses the r€ 2 r_ term when compared with r because the former is
of the first order of the oscillator amplitude while the latter is of second order. In
the limit
...
of small oscillator, the r€ 2 r_ term can therefore be neglected when compared
with r . This would be correct, but not entirely prudent, because Boyer also keeps the
term proportional to a 2 to yield an equation of motion [see Eq. (7) in text]:
 
2e2 ... a2
m0 r€ ¼ m0 v20 r þ r  _
r þ eE0 (A12)
3c3 c2

This is valid as long as r€ ,, a2 . However, the assumption of small oscillator

size can easily be compensated by the high frequencies of oscillations. Boyer [14]
finds a solution for the average amplitude of oscillation (without considering the
r€ 2 r_ term) to be:

1
, r 2 .  (h =mv0 ) cothðp cv0 =aÞ (A13a)
2

from which one can easily derive

1
, r€ 2.  (h v30 =m) cothðp cv0 =aÞ (A13b)
2

Approximation of small oscillating amplitude made by Boyer corresponds to:

, r2 . ,, c2 =v20 (A14)
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 445

However, to neglect r€ against a 2 implies that one must also satisfy:

v20 c2 ,, a2 (A15)

In the limit of small acceleration, this implies very low frequencies. Because in
the limit in Eq. (A15) one can approximate coth z  1/z, the average oscillating vel-
ocity is given by:
kTU
, r_ 2 . ,, c2 (A16)
4 m0
where kTU h a/(2pc) is the Unruh Davies temperature (here a is the
magnitude of the acceleration, i.e., does not depend on the sign). Because
the Boyer oscillator model is for a specific direction, the total energy is given
by: E mc 2 m0c 2 þ 3kTU/2.

Appendix B: Estimate of Rest Mass from Classical Interaction with ZPF

One can estimate the impact of vacuum field fluctuations on the
average energy of a dipole by very simple arguments. We consider a simple
oscillator:§§§§§

ð
1
e e2 dv
m0 r€ ¼ eE ! m0 jvv j  jEv j ! , v2 . ¼ 2 ,Ev2 . (B1)
v m0 v2
0

The average energy density in the vacuum fluctuations can be obtained as:
ð   ð hv3
1 h vk
, 10 E2 þ B2 =m0 . ¼ (2)  d 3 k ¼ dv 2 3 (B2)
2 2 2p c

Using equipartition of energy, this leads to ,E2v . hv3/(2p210c 3), and

therefore the energy density of a single mode of the vacuum field scales as v3,
as previously explained. The kinetic energy of oscillation, in the nonrelativistic
approximation, is:
  2 ð L
2 2 h 1 a m2
,v . ¼ a  v d v ! m0 ,v2. ¼ c2 L (B3)
p mc 0 2 p m0

and we have recovered the same mass scaling as in the HRP

result [Eq. (15) in the text]. On time scales that are large compared to the

§§§§§
This model is essentially the same as the one used by Welton [35] for the Lamb shift, where
the average oscillator amplitude, with a logarithmic divergence, is used as a perturbation of the elec
trostatic potential of the electron nucleus system; see also Ref. 30, p. 90.
446 J.-L. CAMBIER

fluctuations,}}}}} this average kinetic energy is part of the total energy of the
oscillator, that is E mc 2  m0c 2 (1 þ a . m2L/m20). The nonrelativistic treat-
ment is valid as long as ,v 2. ,, c 2, which implies that the upper cutoff is
at most mL  m0. One can generalize to the relativistic case by taking into
account the Lorentz dilation factor, suchqthat g2b2  am2L/m20, from which
the rest mass becomes: m ¼ g m0  m0 1 þ a m2L =m20  a1=2 mL . Thus, in
the relativistic case, the quadratic divergence L2 is replaced by a linear diver-
gence (L), a common occurrence in classical models of self-energy.

Appendix C: Boyer’s Correlation Function

and Poynting Vector
The trajectory of the center of mass of the oscillator under uniform accelera-
tion with respect to the proper time (so-called “hyperbolic motion”) is described
by Eqs. (A6) and (A7). Thus, the Lorentz transformation that relates the inertial
frame L0 that is instantaneously at rest with S is determined by the time-
dependent parameters b(t), g(t) given by Eq. (A6):
   0  
g þbg ct ct
Lt ¼ such that: ¼ Lt (only t, x coordinates) (C1)
bg g x0 x

A plane wave “seen” by the moving observer will have its wave vector and
frequency shifted, while the phase of the wave remains invariant, that is,
f kxx vt kx0 x0 v0 t0 . Therefore,
      
v0 1 v g bg v
¼ Lt ¼ (C2)
ckx0 ckx bg g ckx

The electric and magnetic fields of the vacuum can be written by a Fourier
transform:
Xð 
~
E(~x, t) ¼ ~x  vt  ukl
d 3 k f (v) 1^ kl cos k~ (C3a)
l
X ð 
~ x, t) ¼
B(~ l
d 3 k f (v)(k^  1^ kl ) cos k~
~ x  v t  uk l (C3b)

}}}}}
An analogous situation occurs for electrons oscillating in a classical, externally imposed
electromagnetic field, such as a microwave or laser beam; although the electron oscillates in the
field, there is no net energy transfer between the field and electron. However, during a collision,
this kinetic energy becomes available and participates in the collision process, leading to a net absorp
tion of electromagnetic energy. The “jitter” motion of the classical electron is also unobservable until
interaction with another field, which occurs on longer time scales than the high frequency jitter
motion.
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 447

Here, ukl is a random phase, 1^ is the unit vector along the electric field of a given
mode, and l is the polarization (l +1). Averaging over the random phases
yields the following relations:

,cos ukl cos uk0 l0 . ¼ ,sin ukl sin uk0 l0 . ¼ 12dll0 d(k~  k~0 ) (C4a)
,cos ukl sin uk0 l0 . ¼ ,sin ukl. ¼ ,cos ukl. ¼ 0 (C4b)

The factor f (v) in Eqs. (C3a) and (C3b) is the spectral density which, for
Lorentz invariance of the vacuum field, must be proportional to v1/2. From
the Lorentz transformation laws of arbitrary electric and magnetic fields [12],
we can write the fields observed in the L0 frame (ignoring multiplicative
constants) as:
0 1
1^ x
X ð p B C
~ 0 (0, t) 
E l
d 3 k  v  @ g1^ y  bg(k^  1^ )z A cosðkx x  vt  ukl Þ (C5a)
g1^ z þ bg(k^  1^ )y
0 1
ð (k^  1^ )x
X p B C
~ 0 (0, t) 
B d 3 k  v  @ g(k^  1^ )y þ bg1^ z A cosðkx x  vt  ukl Þ (C5b)
l
g(k^  1^ )z  bg1^ y

This description follows exactly Boyer’s [13] and Haisch and Rueda [3]. We can
now compute the product of such fields and average over the random phases,
using Eqs. (C4a) and (C4b). Let us compute the correlation function between
electric and magnetic fields at two proper times t+ t + dt:

X ð
, Ey0 (t )B0z (tþ ) . ¼ l
d 3 k  v  g b^1y  (k^  1^ )z b c
h i

 gþ 1^ y  (k^  1^ )z bþ  cos kx (x  xþ )  v(t  tþ )
(C6)

where g+ cosh(at+/c) and b+ tanh(at+/c). Using also the notation

z+ at+/c, this leads to:

X ð

þ
, Ey (t )Bz (t ) . ¼ l
d 3 kv 1y cosh z  (k  1)z sinh z



 1y cosh zþ  (k  1)z sinh zþ cos (kx c2 =a)


(cosh z  cosh zþ )  (vc=a)(sinh z  sinh zþ ) (C7)
448 J.-L. CAMBIER

Let us examine the argument of the cosine function first which, using the
hyperbolic identities, can be written as:

kx c2
  
½. . .
¼ coshðzÞ cosh 12 dz  sinhðzÞ sinh 12 dz  coshðzÞ cosh 12 dz
a
 
 sinh (z) sinh 12 dz
vc
  
 coshðzÞ sinh 12 dz þ sinh (z) cosh 12 dz  coshðzÞ sinh 12 dz
a
 
 sinh (z) cosh 12 dz (C8)

where we have also defined z at/c and dz adt/c. Thus,

vc  kx c2 
½. . .
¼ 2 coshðzÞ sinh 12dz  2 sinh (z ) sinh 12dz
a a
  (C9)
c adt
¼ 2 sinh  ½v cosh (at=c)  ckx sinh (at=c)

a 2c

The last term in brackets can be identified from Eq. (C2) as the transformed
frequency v0 . This finally leads to a compact expression for the trigonometric
function:
  
0c adt
cos 2 v sinh (C10)
a 2c

Let us now turn our attention to the product of amplitudes. Expanding again
the hyperbolic functions, we have:


½. . .
 ½. . .
¼ 1^ y (k^  1^ )z cosh2 ðzÞ þ sinh2 (z)

  
 1^ 2y sinh (z) coshðzÞ þ sinh 12dz cosh 12dz

  
 (k^  1^ )2z sinh (z) coshðzÞ  sinh 12dz cosh 12dz (C11)

The sum over polarizations yields the following relations:

X X
1^ 2 ¼ 1  ky2 =k2 ,
l y l
(k^  1^ )2z ¼ 1  kz2 =k2

X
and l
1^ y (k^  1^ )z ¼ kx =k (C12)
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 449

while the hyperbolic functions can be replaced by the Lorentz transformation

parameters, that is, g cosh z and bg sinh z. The final result is:
" !  #
k 2 2
k x y k z
½. . .
 ½. . .
¼ g2 (1 þ b2 )  bg2 1 2 þ 1 2
k
|fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl} k k
(1) |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
(2)
" !  #  
ky2 kz2 1 adt
 1  1  2 2 sinh (C13)
k2 k c
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
(3)

To transform the complete expression in Eq. (C7) into an integral on the

transformed modes (those seen by the accelerated observer), we need to make
use of the transformation properties:

v0 ¼ g ðv  bckx Þ
kx0 ¼ g ðkx  bv=cÞ (C14)
ky0 ¼ ky , kz0 ¼ kz

from which we derive:

v0 ckx0 ¼ g2 (1 þ b2 )vckx  bg2 (v2 þ c2 kx2 ) (C15a)

and the Lorentz-invariant measure:

ð ð 3 0
d3 k d k
; (C15b)
v v0

Consider the first term: using v2 ; c 2k 2 and Eq. (C15a),

ð ð 3 0
2 2 kx 32 2 d k
g (1 þ b ) d k  v ¼ g (1 þ b )  vckx
k v0
ð 3 0
d k
0 0 
¼ 0
 v ckx þ bg2 (v2 þ c2 kx2 ) (C16)
v

The second term of Eq. (C13) yields:

ð ð 3
d3 k d k 2 2 2
(2bg 2 v2 ) þ (c bg )(ky þ kz2 ) (C17)
v v
450 J.-L. CAMBIER

Combining Eqs. (C16) and (C17), we obtain:

ð ð
d 3 k0
0 0 2 2 2 2
 k0
0
v ck x þ bg (c k  v ) ¼ d 3 k0v0 x0 (C18)
v k

Consider now the last term (3) in Eq. (C13). Using v2 ; c 2k 2,

ð ! ð ð 3 0
3
ky2  kz2 d3 k 2 d k 02
d kv ¼ c2 (ky  kz2 ) ¼ c2 (k  kz02 )
k2 v v0 y
ð " !  #
3 0 0
ky02 kz02
¼ d k v   1 2 þ 1 2 (C19)
k k

The complete result reads:

ð (" !  #   )
þ 3 0 0
ky02 kz02 1 adt kx0
, Ey (t )Bz (t ) . ¼ d k  v   1  2 þ 1  2 2 sinh þ 0
k k c k
 0  
2v c adt
 cos sinh (C20)
a 2c

This is exactly the result obtained by Boyer. The integration over the wave
vectors identically yields zero by symmetry arguments. Furthermore, if we evalu-
ate Eq. (C20) at the same proper time t (i.e., dt ; 0), the result is identically null
irrespective of the integration over the wave vectors. Therefore, it is clear that the
x-component of the Poynting vector should be zero. This result is very general; by
symmetry arguments (time reversal), one could also evaluate the phase-averaged
Poynting vector in the coincident K0 frame and transform back into the laboratory
frame and obtain similar results. The fact that the vacuum fluctuation spectrum is
Lorentz-invariant leads to that conclusion.
Haisch and Rueda compute a similar integral function,†††††† which is essen-
tially the sum of contributions (1) and (2) of eq. (C13), leading again to Eq.
(C18) which, after integration over angular variables, is identically null. This
result is identified Haisch and Rueda [3]‡‡‡‡‡‡ then attempt to obtain an
expression in the laboratory frame simply by replacing the variables of inte-
gration (wave vector). This is incorrect; the Poynting vector (or equivalently,
the radiation momentum density) must be transformed back into the laboratory
frame through a Lorentz transformation, thus acting on a quadrivector of
energy-momentum density. The only net momentum density that can be obtained
in the laboratory frame would then be the transform of the time-like component


In Eq. (67) of Ref. 13, the cosine argument should be v0 instead of v; this is a typographical
error, because the computation of the argument is the same for all two point correlations.
††††††
See Ref. 3, eq. (A30).
‡‡‡‡‡‡
See Ref. 3, eq. (C19).
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 451

in the accelerated frame, that is, the energy density. Because the acceleration
leads to a thermalization of the vacuum fluctuations, the net momentum should
then be equivalent to the pressure of thermal photons promoted from the
vacuum by the acceleration; in other words, the Unruh Davies effect should
be recovered and the effect would be excessively small.

Appendix D: Unruh –Davies and Classical Oscillator

Here we offer a very simple (trivial) alternative approach to the coupling
between a classical oscillator and ZPF in an accelerated frame. Consider
first the complete radiation spectrum, including both thermal and zero-point
components:
 
v3 1 1
r(v) ¼ 2 2 hv=kT þ (D1)
2p c e 1 2
which can be rewritten as:
 v3  
h hv
r(v) ¼ 2 2
coth (D2)
4p c 2kT
Consider now the oscillator model of Einstein and Hopf, who showed that
the reaction force exerted by a background field onto an oscillator moving at a
velocity v with respect to the field was:
 
4p2 e2 v dr
dF ¼  r dv  v (D3)
5mc2 3 dv
Of course the zero-point component does not contribute to the reaction force
because it is Lorentz-invariant, but the thermal component does (a finite tempera-
ture breaks down Lorentz invariance). Using Eq. (D2), the derivative of the
energy density is:

dr 3 h r
¼ rþ (D4)
dv v 2kT sinh 6  cosh 6

with 6 h v/2kT. Therefore the force element is (barring the physical constants)
of the form:

hv3 d v
dF /  v (D5)
2kT sinh2 (6)

For a uniformly accelerated frame, we have, of course:

v ¼ c tanh (at=c) (D6)

and because the frame is accelerated, the effective temperature of the

vacuum electromagnetic field is given by the Unruh Davies temperature:
452 J.-L. CAMBIER

kTU h a/2pc. In the limit of moderate acceleration (i.e., small temperature),

we have approximately:

hv3 2kT 2
dF /  d v  tanh (at=c)  jaj  (v dv)  (at=c) (D7)
2kT h v

This leads to an inertial mass that is:

dmi (t) / jajt  L2 (D8)

Note that this is a correction to the inertial (or rest) mass of the particle, that is,
this term is added to the existing rest mass. The term is quadratically divergent
with respect to the cutoff energy, as expected from classical model in the nonre-
lativistic approximation. The correction grows in time until saturation occurs
(when at/c  1), at which time the reaction force becomes constant and the
mass correction changes behavior. Of course, we do not expect this result to
be correct; instead, a QED calculation using finite-temperature formulation
leads to a mass correction that is proportional to a 2/m, still very small for
almost all cases of acceleration. The point of this exercise is to show that
simple classical arguments can be used to yield acceleration effects, but
leading to small corrections to the mass, rather than attempting to explain
away the complete mass as an interaction with the ZPF.

References
[1] Haisch, B., Rueda, A., and Puthoff, H. E., “Inertia as a Zero Point Field Lorentz
Force,” Physical Review A, Vol. 49, 1994, pp. 678 694.
[2] Haisch, B., and Rueda, A. “Reply to Michel’s ‘Comment on Zero Point Fluctuations
and the Cosmological Constant’,” The Astrophysical Journal, Vol. 488, 1997, pp.
563 565.
[3] Rueda, A., and Haisch, B., “Contribution to Inertial Mass by Reaction of the Vacuum
to Accelerated Motion,” Foundations of Physics, Vol. 28, 1998, pp. 1057 1108.
[4] Sunahata, H., “Interaction of the Quantum Vaccum with an Accelerated Object and
its Contribution to Inertia Reaction Force,” PhD Thesis, California State University,
2006.
[5] Haisch, B., and Rueda, A., “Reply to Michel’s ‘Comment on Zero Point Fluctuations
and the Cosmological Constant’,” The Astrophysical Journal, Vol. 488, 1997, pp.
563 565.
[6] Dobyns, Y., Haisch, B., and Rueda, A., Foundations of Physics, Vol. 30, arXiv:gr gc/
002069, 2000, pp. 59 80.
[7] Rueda, A., and Haisch, B., “Gravity and the Quantum Vacuum Inertia Hypothesis,”
Annalen der Physik, Vol. 14, 2005, pp. 479 498.
[8] Thaller, B., The Dirac Equation (Text and Monographs in Physics), Springer Verlag,
Berlin, 1992.
[9] Einstein, A., and Hopf, L., “Statistische Untersuchung der Bewegung eines Resona
tors in einem Strahlungsfeld,” Annalen der Physik, Vol. 338, 1910, pp. 1105 1115.
 INERTIAL MASS FROM STOCHASTIC ELECTRODYNAMICS 453

[10] Marshall, T. W., “Statistical Electrodynamics,” Proceedings of the Cambridge

Philosophical Society, Vol. 61, 1965, pp. 537 546.
[11] Boyer, T. H., “Derivation of the Blackbody Radiation Spectrum without Quantum
Assumptions,” Physical Review, Vol. 182, 1969, pp. 1374 1383.
[12] Jackson, J. D., Classical Electrodynamics, 2nd ed., Wiley, New York, 1982.
[13] Boyer, T. H., “Thermal Effects of Acceleration Through Random Classical Radi
ation,” Physical Review D, Vol. 21, 1980, pp. 2137 2148.
[14] Boyer, T. H., “Thermal Effects of Acceleration for a Classical Dipole Oscillator in
Classical Electromagnetic Zero Point Radiation,” Physical Review D, Vol. 29,
1984, pp. 1089 1095.
[15] Boyer, T. H., “Derivation of the Blackbody Radiation Spectrum from the
Equivalence Principle in Classical Physics with Classical Electromagnetic
Zero Point Radiation,” Physical Review D, Vol. 29, 1984, pp. 1096 1098.
[16] Miyamoto, U., and Maeda, H., “On the Resolution of Negative Mass Naked
Singularities,” Journal of Physics: Conference Series, Vol. 31, 2006, pp. 157 158.
[17] Lawrie, I. D., A Unified Grand Tour of Theoretical Physics, Institute of Physics Pub
lishing, Bristol, U.K., 2002.
[18] Cebeci, H., Sarioglu, Ö., and Tekin, B., “Negative Mass Solitons in Gravity,” Phys
ical Review D, Vol. 73, 2006, 064020.
[19] Veltman, M., Facts and Mysteries in Elementary Particle Physics, World Scientific,
New York, 2003.
[20] Itzykson, C., and Zuber, J. B., Quantum Field Theory, McGraw Hill, New York,
1980.
[21] Kaku, M., Quantum Field Theory: A Modern Introduction, Oxford University Press,
New York, 1993.
[22] Coleman, S., in Electromagnetism: Paths to Research, Tiplitz, D., ed., Plenum,
New York, 1982, pp. 183 210.
[23] Amit, D. J., Field Theory, The Renormalization Group and Critical Phenomena,
McGraw Hill, New York, 1978.
[24] Dolan, L., and Jakiw, R., “Symmetry Behavior at Finite Temperature,” Physical
Review D, Vol. 9, 1974, pp. 3320 3341.
[25] Cambier, J. L., Primack, J. R., and Sher, M., “Finite Temperature Radiative
Corrections to Neutron Decay and Related Processes,” Nuclear Physics B,
Vol. B209, 1982, pp. 372 388.
[26] Peresutti, G., and Skagerstam, B. S., “Finite Temperature Effects in Quantum Field
Theory,” Physics Letters B, Vol. 110, 1982, pp. 406 410.
[27] Qader, M., Masood, S., and Ahmed, K., “Second Order Electron Mass Dispersion
Relation at Finite Temperature,” Physical Review D, Vol. 44, 1991, 3322 3327.
[28] Davies, P. C. W., “Scalar Production in Schwarzschild and Rindler Metrics,” Journal
of Physics A, Vol. 8, 1975, pp. 609 616.
[29] Unruh, W. G., “Notes on Black hole Evaporation,” Physical Review D, Vol. 14,
1976, pp. 870 892.
[30] Chen, P., and Tajima, T., “Testing Unruh Radiation with Ultraintense Lasers,”
Physical Review Letters, Vol. 83, 1999, pp. 256 259.
[31] Leinass, J. M., “Unruh Effect in Storage Rings,” arXiv:hep th/0101054.
[32] Chen, P., “Press Release: Violent Acceleration and the Event Horizon,” June 2006,
http://home.slac.stanford.edu/pressre leases/2000/20000606.htm. [Cited 19 Nov.
2008].
454 J.-L. CAMBIER

[33] Milonni, P., The Quantum Vacuum, Academic Press, New York, 1994.
[34] Novoselov, K. S., Geim, A. K., Morozov, S. V., Jiang, D., Katsnelson, M. I., Grigor
ieva, I. V., Dubonos, S. V., and Firsov, A. A., “Two Dimensional Gas of Massless
Dirac Fermions in Graphene,” Nature, Vol. 438, 2005, 04233.
[35] Welton, T. A., “Some Observable Effects of the Quantum Mechanical Fluctuations
of the Electromagnetic Field,” Physical Review, Vol. 74, 1948, 1157 1167.
 Chapter 14

Relativistic Limits of Spaceflight

Brice N. Cassenti
Rensselaer Polytechnic Institute, Hartford, Connecticut

I. Introduction
HE Special Theory of Relativity is one of the foundations of modern physics.
T It provides a conceptual starting point for the General Theory of Relativity
and is a key ingredient of Quantum Field Theory. Yet many of its predictions
are counterintuitive. These counterintuitive predictions produce paradoxes that
are not readily explained; yet experimental results have verified the predictions
of Special Relativity to an extraordinary degree. Although the Special Theory of
Relativity is formulated for inertial systems, it can also be applied to accelerating
interstellar spacecraft and is useful for sizing interstellar spacecraft.
The Special Theory of Relativity was proposed by Einstein in 1905 [1] to
reconcile experimental results that did not agree with classical Newtonian mech-
anics. Special Relativity assumes that Maxwell’s equations for electromagnetism
are exactly correct by hypothesizing that the speed of light in a vacuum takes on
the same value in all inertial (constant velocity) reference frames. As a conse-
quence of this assumption, time cannot flow at the same rate in all inertial refe-
rence systems, and this is usually at the heart of the paradoxes. An excellent
introduction to both Special and General Relativity can be found in Einstein’s
own writings. A detailed mathematical treatment is presented in Ref. 2, while
Ref. 3 gives a more popular presentation. Taylor and Wheeler have written an
excellent introduction to Special Relativity [4].
Paradoxes are a common feature of the Special Theory of Relativity and are
very useful at contrasting our common notions of space and time and the

Copyright # 2008 by the American Institute of Aeronautics and Astronautics, Inc. All rights
reserved.
Based on “Faster than Light Paradoxes in Special Relativity,” AIAA 2006 4607, presented at the 41st
AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, Sacramento, CA, 10 12
July 2006.

Associate Professor, Department of Engineering and Science.

455
456 B. N. CASSENTI

results predicted from the foundations of Special Relativity. A detailed

discussion of these paradoxes can be found in Ref. 5. The paradoxes are even
more profound if speeds greater than the speed of light are considered. For
example, in cases where objects travel faster than the speed of light, there are
reference frames where an effect can precede its cause. Of course, this implies
time travel. Modern physics sometimes removes these effects by postulating
that nothing can move faster than light. But the General Theory of Relativity
does not exclude the possibility, nor does the Special Theory, and quantum
mechanics may actually require it. Of course, it may be that it is just our poor
understanding of space and time that causes the confusion and that new concepts
will help clarify.
Paradoxes, though, do not mean that the theory is inaccurate or in error.
Empirical evidence strongly favors the Special Theory of Relativity, and has
verified it to an extraordinary accuracy and, hence, we can use Special Relativity
as a basis for sizing interstellar vehicles.
The sections that follow will introduce the Special Theory of Relativity
and then cover some of the well-known paradoxes. Closing sections give some
empirical results and a basis for the description of the motion of relativistic
rockets.

II. Principle of Special Relativity

Special Relativity is based on the experimental fact that the speed of light in a
vacuum remains the same in all inertial frames regardless of the observer’s speed
with respect to the light source. Many experiments were performed in the late
1800s and early 1900s that verified that the speed of light did not vary with
the relative speed of the source, but a simple Galilean transformation of
Maxwell’s equations indicated that the speed of light should vary. Ad hoc the-
ories were proposed to satisfy the observations. The most complete was that of
Lorentz [1], who proposed a transformation where both space and time were
connected to the relative speed and, although he modified Maxwell’s equations
to account for motion with respect to the stationary ether, the effects were
forced to cancel exactly in order to agree with the experimental observations.
Einstein, though, assumed that Maxwell’s equations were exactly correct, and
that the fault was in the Galilean transformation.

A. Postulates and Lorentz Transformation

The two postulates of Einstein’s Special Theory of Relativity are that: 1) phys-
ical laws hold in all inertial reference systems, and 2) the speed of light (in a
vacuum) is the same in all inertial reference frames. These two postulates are all
that is required to find the transformations that Lorentz found in his ad hoc manner.
Consider two inertial space-time reference systems (primed and unprimed) as
shown in Fig. 1. The coordinates (t, x) are for the stationary system, and the coor-
dinates (t 0 , x0 ) are in the moving system. The origin of the primed system is
moving at the constant speed v along the x-axis. Because Maxwell’s equations
must remain linear in both systems, the coordinates (t 0 , x0 ) should be a linear com-
bination of the coordinates (t, x). Taking the origins of both systems to coincide,
 RELATIVISTIC LIMITS OF SPACEFLIGHT 457

Fig. 1 Inertial reference frames.

the coordinates must be related by

ct0 ¼ Act þ Bx
(1)
x0 ¼ Cct þ Dx

where c is the speed of light and A, B, C, D are constants that depend only on the
relative speed v. The quantities ct and ct0 are used because they are distances and
hence have the same units as x and x0 .
The speed of light, c, must be exactly the same in both coordinate systems.
Hence, light moving through dx in time dt in the unprimed coordinate system
at speed c must also move through dx0 and dt0 in the primed system at
speed c. Then

c2 dt 2  dx2 ¼ c2 dt02  dx02 ¼ c2 dt2 (2)

The quantity dt is referred to as the proper time and represents time intervals for
an object that is stationary in its own frame.
Substituting Eq. (1) into Eq. (2) and equating the coefficients of dx 2, dt 2, and
dx dt on both sides of the resulting equation gives

A ¼ D ¼ cosh u
(3)
B ¼ C ¼  sinh u

where u u(v). The function u(v) can be found by noting that a particle stationary
in the primed system is moving at v in the unprimed system. Using Eqs. (1) and (3)
dx0 Cc þ Dv
¼0¼ (4)
cdt0 Ac þ Bv
458 B. N. CASSENTI

Then
v
¼ tanh u (5)
c
Equation (1) can now be written as
ct0 ¼ g ðct  bxÞ
(6)
x0 ¼ g ðx  bctÞ
p
where b tanh u and g ¼ cosh u ¼ 1= 1  b2 . Taylor and Wheeler [4] refer to u
as the velocity parameter, but in the physics community, u is usually referred to as
the rapidity. The transformation in Eq. (6) is commonly known as the Lorentz
transformation [1].
We can now compare the coordinate systems by considering lines of constant
ct0 and x 0 (and constant ct and x) as shown in Fig. 2. Note that light emitted from
the origin will travel along the diagonal in both coordinate systems. Also note that
the primed coordinate system is collapsing around the diagonal of the unprimed
system and, at a relative speed equal to the speed of light, the primed coordinates
will have completely collapsed. Equation (5) indicates that v/c can only approach
the speed of light if u is finite and real. For b . 1, u is imaginary, as well as g.
The proper time also becomes imaginary as indicated by Eq. (2). Another way to
look at faster-than-light relative motion is that the spatial and temporal coordi-
nates switch. This clearly means that objects can travel both ways in time for
an observation of faster-than-light particles. Yet another way to look at this is
to consider what Fig. 2 would look like for the primed system moving at
greater than the speed of light. The ct0 axis would be on the x-coordinate side,

Fig. 2 Lorentz transformed coordinate systems.

 RELATIVISTIC LIMITS OF SPACEFLIGHT 459

and the x0 axis would be on the ct-coordinate side. Time in the primed system
would be a space-like coordinate to the observer in the primed system, and the
space coordinate in the direction would be time-like in the unprimed system.
In Special Relativity, physicists sometimes assume that nothing can go faster
than light and this completely removes those cases from consideration.

B. Lorentz Contraction and Time Dilation

Two significant consequences of Special Relativity are the apparent contrac-
tion and the apparent slowing of time for moving objects. The contraction of
moving objects is easiest to predict once a method of measuring length is
defined. We will define length as the Euclidean difference in coordinates at the
same value for the time coordinate. In our case this means the difference in the
x-coordinate locations for a specific value of ct. For a measurement of a rod
stationary along the x-axis in the unprimed system, we can set ct0 0 in Eq.
(6). Then in the primed system ct bx, and
q
x0 ¼ 1  b2 x (7)

and stationary lengths in the unprimed system appear to be contracted in the

primed system. Equation (7) is the Lorentz contraction. Of course a similar analy-
sis will show that stationary lengths in the primed system appear contracted in the
unprimed system.
Time dilation is the analogous situation for relative time measurements. Con-
sider a stationary particle at the origin of the unprimed system (i.e., x 0). Then
Eq. (6) yields ct bx and
q
ct ¼ 1  b2 ct0 (8)

Hence, time in the moving coordinate system appears to move slower. However,
for an observer in the primed system watching a clock stationary at the origin of
the unprimed system stationary (i.e., x0 0), we find from Eq. (6) that
q
ct0 ¼ 1  b2 ct (9)

Hence observers in both frames will predict the other’s clocks are moving
more slowly. If they continue to move at a constant relative velocity (i.e., both
systems are inertial), then they can only meet once and there will be no
contradictions.

III. Paradoxes in Special Relativity

A. Pole Vaulter Paradox
A popular paradox in Special Relativity is the “pole vaulter paradox.” In this
scenario a pole vaulter carries his pole horizontally at high speed. In our version
of the paradox, the pole is 15 feet long according to the pole vaulter, but he is
460 B. N. CASSENTI

moving so fast that an observer standing at a stationary barn that is 10 feet long
sees the pole contracted to half its length, which is now less than the length of the
barn. The front door of the barn is open as the pole enters. When the back end of
the pole enters, the front end has not reached the back door, and the front door
closes. Hence to the observer the pole is entirely enclosed by the barn. When
the front of the pole meets the back door, the back door is opened and the pole
leaves the barn. Of course to the pole vaulter, the barn appears half as long
(i.e., 5 feet) and the pole vaulter will note that the pole was never completely
in the barn. The pole vaulter will also note that the back door opened before
the front door closed. Obviously, the pole cannot both be longer and shorter
than the barn. The paradox is solved by tracking the front and back end of the
pole and barn in both the pole vaulter’s and the observer’s systems. The trans-
formations will show that the order in which the doors open and close is reversed
in each system, agreeing with the observations. The results will also show that the
time interval when the doors open and close is shorter than the time it would take
light to traverse the distance between the doors of the barn in the observer’s coor-
dinate system, and is also shorter than the time it would take light to traverse the
length of the pole in the pole vaulter’s coordinate system. Hence, closing the front
door cannot influence opening the back door if the speed of light is the maximum
speed in the universe.
This leads to the primary contradiction of Special Relativity with our
common low-speed experience. Space and especially time are not what we
picture. Newtonian physics exists in a universe where time is the same for all
observers, and Special Relativity says it cannot be if the speed of light is the
same in all inertial (nonaccelerating) reference systems.

B. Twin Paradox
The best-known paradox in Special Relativity is the twin paradox [4,6]. The
problem presented by this paradox involves time directly. Consider two fraternal
twins, Alice and Bob, preparing for a long space voyage. Bob is in mission
control while Alice will do the traveling. Alice departs at a very high speed so
that Bob sees Alice’s clocks running at half speed, but, of course, Alice sees
Bob’s clock running at half speed also. At the half way point Alice reverses
speed, so that on the return each sees the other’s clock as running at half
speed. When Alice stops to meet Bob, they both cannot think that each has
aged half as much, thus the paradox. Certainly the Lorentz transformations
apply to Bob, because at no time did he experience any acceleration, and
hence, Alice will have aged half as much as Bob. Obviously the problem is
with the acceleration Alice experiences. It can be shown that while Alice
changes her speed she will see Bob’s clock run faster. All three of her accelera-
tions will exactly compensate for the slowing during the constant velocity
portions of the trip, and both Bob and Alice will agree that each of their clocks
is correct when they meet. Actually, there are many ways to resolve the
paradox. Weiss [6] gives an excellent discussion. The section to follow on
constant acceleration rockets can be used to show that each will agree with the
clock readings on Alice’s arrival.
 RELATIVISTIC LIMITS OF SPACEFLIGHT 461

C. Faster-than-Light Travel
Objects that move faster than the speed of light can reverse the order of events
in some inertial systems. Figure 3 illustrates the appearance of two objects in two
inertial systems. The object moving from A to B is moving less than the speed of
light in the unprimed system. The projection of the two points A and B onto the
ct-axis clearly shows that A precedes B. We can also project the points A and B
onto the ct0 -axis, which again clearly shows that A precedes B.
If we now consider the path from C to D, the slope of the line shows that in the
unprimed system the speed of an object going from C to D is greater than the
speed of light. Again projecting onto the ct-axis shows that C precedes D in
the unprimed system, but in the primed system a projection onto the ct0 -axis
shows that D now precedes C. Hence, events that are observed for an object
moving faster than the speed of light can have effects that precede their cause.
For example, a dish that breaks going from C to D in the unprimed system
will magically reassemble itself when observed in the primed system. The sim-
plest resolution of the paradox is to again assume that physical objects cannot
move faster-than-light but, of course, this may not be the actual explanation.

D. Instant Messaging
Now consider the case illustrated in Fig. 4. Two observers, Art (A) and Don
(D), are stationary in the unprimed coordinate system. There are two other obser-
vers, Brenda (B) and Cathy (C), in the primed system, moving at a constant velo-
city v̄ with respect to the unprimed system. Art hands off a message to Brenda as
she goes by. Brenda then instantaneously transmits the message to Cathy. Cathy
hands the message to Don, who instantaneously sends the message to Art. The

Fig. 3 Lorentz transformation for faster-than-light-travel.

462 B. N. CASSENTI

Fig. 4 Instant messaging.

loop in Fig. 4 shows that Art receives the message before he hands it off to
Brenda. Art “now” has two messages. “After” completing the loop again Art
has four, and the process continues to double infinitely many times, creating
quite a puzzling paradox. The problem again lies with speeds greater than the
speed of light. The objects that are moving faster than light are messages,
which need not consist of a physical mass. Nevertheless, the information in the
messages exceeds the speed of light.
Again, the simplest way to remove the paradox is to assume that nature
does not allow any cause to propagate faster then the speed of light, but there
is another explanation. The infinite number of messages arriving at A is clearly
a nonlinear process and will not satisfy the linear relationships assumed in
Special Relativity. This nonlinear process may be related to the collapse of the
wave function in quantum mechanics, but a nonlinear process would also
violate the postulates of quantum mechanics. The nonlinear theory could result
in one of the infinite number of messages being chosen and resolve the
paradox. That is, the collapse of the wave function in quantum mechanics
could be the mechanism that resolves the paradox and may clear up the concep-
tual problems with entanglement [7]. Chapter 16 explores quantum entanglement
in more detail.
 RELATIVISTIC LIMITS OF SPACEFLIGHT 463

A better illustration of the collapse of the wave function is the case where Art
can randomly choose between two messages, one labeled Up and the other Down.
Art randomly chooses one of the two. For example, Art chooses Up and hands it
off to Brenda. Instantaneously Art will have three messages. Two of these are Up
and the third is Down. A random choice now is more likely to be Up. Of course
instantly there will be an infinite number of messages consisting of the actual
choice for the collapse of the wave function, and the other choice will effectively
not appear at all. Hence, the collapse of the wave function could be explained as a
closed self-referential spacetime loop [6].

IV. Empirical Foundations

Paradoxes in the Special Theory of Relativity seem to imply that the theory
may not be physically consistent, but the Special Theory of Relativity has been
demonstrated to be accurate to an extraordinary precision. The text by Bergmann
[8] contains an entire chapter on the experimental verification of the Special
Theory of Relativity. Born [9] has a more popular description of the Special
Theory of Relativity but it does contain references to experimental verification
scattered throughout his chapter on Special Relativity. A readily accessible
summary by Roberts [10] contains numerous references to experiments and
their results.
All of the experiments so far performed indicate that the Special Theory of
Relativity is an extremely accurate representation of the nature of the universe.
Of course, with over a century of experimental verification it will be difficult
to publish any experimental violations, nevertheless searches for violation do
continue (e.g., see the article by Kostelecky [11], which contains a popular
description for the continuing search for experimental violations). Several of
these are summarized below.

A. Michelson – Morley Experiment

The Michelson Morley experiment is the classic experiment referenced for the
verification of the fundamental tenets of the Special Theory of Relativity. It con-
sists of a Michelson interferometer where a single light beam is split into two
beams moving at right angles. When the beams are brought back together, inter-
ference fringes appear according to the path lengths of the split beams and the
speed of light in each direction. Variations in the travel times can then be measured
along different directions. Because the Earth has a preferred direction for its orbital
velocity about the sun, any change in the speed of light due to the motion of the
Earth will be recorded. The most recent experiment shows that there is no
change with direction for the velocity of light [10] to within 2 parts in 10,000.

B. Half-Lives of Relativistic Particles

One of the predictions of the Special Theory of Relativity are changes in the
time observed by individuals moving at relativistic velocities. Cosmic rays are a
natural source of high energy, (relativistic) particles and, in fact, the muon, which
is produced when high-energy cosmic rays collide with atoms in the upper
atmosphere, has been observed at ground level indicating that its life has been
464 B. N. CASSENTI

significantly extended beyond its nominal 2ms lifetime. Experiments done in par-
ticle accelerators are much more accurate. Experiments have shown accuracies in
time dilation predictions to better than 0.1% [12].

C. Doppler Shifts
The Doppler shift on the spectra of quasars, which are moving away at rela-
tivistic velocities, is a very accurate test of the theory of Special Relativity for
radial motion but the more important tests are those that involve transverse
motion. Here laboratory tests have verified the theory with accuracies of better
than 3  10 6 relative to the predicted value [9]. These tests are particularly
important because the predictions of Special Relativity are at odds with all non-
relativistic (Galilean) theories.

D. Cerenkov Radiation
Cerenkov radiation is produced when charged particles move through a
material at speeds faster than the speed of light in the material. This is analogous
to the motion of an aircraft through the air at faster than the local speed of sound
in the air. When these particles exceed the local speed of light in a material a
“light boom” instead of a sonic boom is created from radiation emitted by
the particle as it slows down. The emitted light propagates perpendicular to the
wave front of the light boom. Also, the intensity of the light increases with the
emitted frequency and therefore is preferentially blue. In a material the local
speed of light is reduced by a factor equal to the index of refraction. For
example, if particles move through water, which has an index of refraction of
4/3, then Cerenkov radiation will be observed with more than 3/4 of the
speed of light. Such speeds are easily exceeded in cosmic ray and elementary par-
ticle physics experiments. In fact, Cerenkov radiation is widely used in elemen-
tary particle physics experiments to differentiate relativistic particles of the same
momentum but different masses [11] and the measurements are completely con-
sistent with the Special Theory of Relativity.

E. Tachyon Searches
Tachyons are hypothetical particles that move faster than the speed of light
and, if they existed, could produce causality paradoxes. A brief popular descrip-
tion on tachyons can be found in Ref. 13. For a particle with a velocity exceeding
the speed of light, Eq. (11) implies thatpboth the energy and the momentum
become imaginary (i.e., a multiple of 1). This is consistent with a rest
mass that has an imaginary value. There is nothing in the Special Theory of Rela-
tivity that precludes such particles from existing, but reported searches have been
consistent with the observation that tachyons do not exist. The experimental
limits on the imaginary mass-energy of the tachyon particles [14] have been
reported as less than 104 eV.
Even if tachyons do exist, the Feinberg reinterpretation principle [15] could be
used to show that tachyons will not violate causality. Recall that the order for
cause and effect is dependent on the relative velocity of the observers. Causality
would be violated if a tachyon could send information into its own past. The
 RELATIVISTIC LIMITS OF SPACEFLIGHT 465

Feinberg reinterpretation principle states that sending a negative energy tachyon

backward in time (which could violate causality) would appear as positive energy
tachyons moving forward in time. According to the Feinberg reinterpretation
principle, the creation and absorption of tachyons cannot be differentiated by
observers undergoing relative motion. Hence, if the Feinberg reinterpretation
principle holds, it would be possible to use tachyons to send energy forward in
time, but it would not be possible to communicate backward in time [13].

V. Relativistic Rockets
Although Special Relativity applies to constant velocity processes, it can be
extended to accelerating objects for applications to interstellar spaceflight. The
extension only requires that we constantly update the reference velocity to an
accelerating system. Before proceeding we must find the relativistic form of
the momentum and energy so that we may apply the conservation of momentum
to relativistic rockets. If we multiply Eq. (2) by m20 c2 , where m0 is the mass in the
frame at rest, then
 2  2
dt dx
m20 c4 ¼ m20 c4  m20 c2 (10)
dt dt

The first quantity on the right-hand side is the relativistic energy, E, and the
second is the relativistic momentum, p [1,4]. Using the coefficients in Eq. (6),
the relativistic energy and the three-dimensional momentum form a four-
dimensional vector with components

E ¼ gm 0 c 2 (11a)
and a total momentum

pc ¼ gbm0 c2 (11b)

For small speeds (b  1 and cosh u  1 þ u2 =2), Eq. (5) yields

 
1 2 1
E  m0 c 1 þ u  m0 c2 þ m0 v2
2
(12)
2 2
The second term is the correct form for the kinetic energy and the first is the rest
mass energy.

A. Constant Acceleration Rocket

Many authors have considered relativistic rockets (e.g., see Refs. 16 and 17),
and all are based on the conservation of momentum and energy. Using the vel-
ocity parameter, the momentum is given by

p ¼ m0 c sinh u (13)
466 B. N. CASSENTI

Then

dp du
¼ m0 c cosh u (14)
dt dt

In the accelerating frame, u 0 and cosh u 1. Using Newton’s Second Law

dp du
¼ m0 c ¼ m0 a (15)
dt dt

where a is the acceleration felt in the accelerating (rocket) frame of reference.

Then the acceleration is simply given by

du
a¼c (16)
dt

Using Eq. (2), it follows that for all inertial reference frames

dt
¼ cosh u
dt
(17)
dx
¼ c sinh u
dt

For constant acceleration a, Eq. (16) can be integrated to yield

at
u¼ (18)
c

where u 0 at t 0. Substituting Eq. (18) into Eq. (17) and integrating,

c at 
t ¼ sinh
a c
2h at  i (19)
c
x¼ cosh 1
a c

where t 0, and x 0 at t 0. Dividing the last of Eq. (17) by the first yields

dx
¼ c tanh u (20)
dt
 RELATIVISTIC LIMITS OF SPACEFLIGHT 467

which reproduces Eq. (5). For small proper times, t, Eq. (19) and (20) become

tt
at 2
x (21)
2
dx
 at
dt
Equation (21) in the correct nonrelativistic expressions for constant acceleration.
Equations (18) through (20) can now be used to obtain results for a spacecraft
accelerating to relativistic speeds (see Chapter 2 for more applications.) Note that
Eq. (17) indicates that the rocket will cover distance dx in onboard time dt.
Hence, when the hyperbolic sine of the velocity parameter is greater than one,
the crew will experience travel at speeds greater than the speed of light.
Although, the stationary observer will note that the rocket is always moving
less than the speed of light according to Eq. (20), there is no contradiction.
Time intervals in the stationary observer’s frame, from Eq. (17), are always
larger than the crew’s changes in time. The speeds can become so large in the
crew’s frame of reference that they could circumnavigate our expanding universe
in less than their working lifetime if they accelerate continuously at one Earth
gravity. The conclusion is that relativity does not limit the speeds that can be
achieved. Of course the observer on the ground will be long gone, along with
the reference stellar system, before the circumnavigating crew returns.
We now have enough information to resolve the twin paradox. Time in the
accelerating twin’s reference frame can now be taken as the proper time, t,
and the readings of the clocks on return can be predicted for each twin. The analy-
sis shows no paradox.
There is more in the consideration of accelerations than just a resolution of the
twin paradox. The case of constant acceleration also points to an approach that
will allow the inclusion of gravitational accelerations. Because, in both cases,
the accelerations experienced do not depend on the mass, this fact can be used
to lead directly to the fundamental principle of the General Theory of Relativity.

B. Photon Rocket
We can estimate the amount of mass, or energy, required by considering a
photon rocket. Because a photon is a particle of light it has no rest mass and,
hence, m0 0 in Eq. (11). Then Eq. (10) in terms of energy and momentum
becomes

m20 c4 ¼ E2  p2 c2 ¼ 0 (22)

and, taking the positive root of Eq. (22),

E ¼ pc (23)
468 B. N. CASSENTI

for a photon. Consider the conservation of momentum in the rocket’s frame of

reference. If dEg is the energy of the photons released in proper time dt, then

dp 1 dEg du
¼ ¼ mc (24)
d t c dt dt
where m is the current mass of the rocket in the reference frame moving with the
rocket, but the conservation of energy yields, in the accelerating rocket frame
 
d Eg þ mc2 dEg dm
¼ þ c2 ¼0 (25)
dt dt dt
Substituting for the energy in Eq. (25) using Eq. (24) and integrating, yields
u
m ¼ mi e (26)

where mi, is the mass of the rocket at u 0. If the final mass occurs at u uf, then
the mass ratio is
mi
MR ¼ ¼ eu f (27)
mf

Although, for reasonable final masses the propellant mass required to circum-
navigate the universe would be less than the mass of the universe, it would never-
theless be enormous.
It should be noted that the above could be readily extended to mass annihil-
ation rockets [17,18] (i.e., antimatter rockets) and gives reasonable estimates
for the performance requirements of interstellar antimatter rockets. Antimatter
rockets require not only the application of energy and momentum conservation,
but also the conservation of the baryon (nucleon) number must be applied. A
baryon number of plus one is assigned to protons and neutrons, while antiprotons
and antineutrons have a baryon number of minus one. The conservation laws can
also be extended to the Bussard interstellar ramjet [19], where interstellar matter
is collected by a moving spacecraft and used as propellant. In Refs. 20 and 21 all
three conservation laws are applied to the interstellar ramjet.

VI. Conclusions
The Special Theory of Relativity is based on sound empirical evidence and
demonstrates to an extraordinary degree of accuracy that the speed of light is con-
stant in inertial reference frames. It is our concept of time that has had the most
dramatic change as a result of the Special Theory of Relativity and, again, it is our
concept of time that produces the most paradoxical results. This is especially true
when objects are postulated to travel faster than the speed of light. But these
faster-than-light paradoxes in the Special Theory of Relativity are more than
curiosities based on our concepts of time and space; they can point the way to
a more complete understanding of space and time. They can even add insight
into other paradoxes such as the collapse of the wave function in quantum
 RELATIVISTIC LIMITS OF SPACEFLIGHT 469

mechanics, and may even lead to new physical theories that may allow unlimited
access to the universe. If our concept of time can be clearly defined, then it may
be possible to resolve all of these paradoxes.
Although the Special Theory of Relativity is applicable only in inertial refer-
ence frames, it can be extended to accelerating spacecraft for sizing interstellar
missions. Only the conservation of momentum, energy, and baryon (nucleon)
number are required. Sizing, using a photon rocket as an example, will show
that although practical interstellar flight will be difficult and expensive, it is
not impossible.
Within the constraints of the Special Theory of Relativity it is clearly imposs-
ible to exceed the speed of light, and interstellar flight will always be handi-
capped by long travel times to outside observers. Our only hope of reducing
the long travel times may be through the General Theory of Relativity (e.g.,
warp drives) and/or quantum mechanics.

References
[1] Loventz, H. A., Einstein, A., Minkowski, H., Weyl, H., Perret, W., Sommerfeld, A.,
and Jeffery, G. B., The Principle of Relativity: A Collection of Original Memoirs
on the Special and General Theory of Relativity, Courier Dover Publications,
New York, 1952.
[2] Einstein, A., Relativity The Special and General Theory, Crown Publishers,
New York, 1961.
[3] Einstein, A., The Meaning of Relativity, Princeton University Press, Princeton, NJ,
1956.
[4] Taylor, E. F., and Wheeler, J. A., Spacetime Physics, W. H. Freeman and Co., San
Francisco, 1966.
[5] Yakov P., and Terletskii, Y., Paradoxes in the Theory of Relativity (translated from
Russian), Plenum Press, NY, 1968.
[6] Weiss, M., “The Twin Paradox.” URL: http://math.ucr.edu/home/baez/physics/
Relativity/SR/TwinParadox/twin vase.html.
[7] Goff, A., “Quantum Tic Tac Toe: A Teaching Metaphor for Superposition in
Quantum Mechanics,” American Journal of Physics, Vol. 74, No. 11, 2006,
pp. 962 973.
[8] Bergmann, P. G., Introduction to the Theory of Relativity, Dover Publications,
New York, 1976.
[9] Born, M., Einstein’s Theory of Relativity, Dover Publications, New York, 1965.
[10] Roberts, T. “What Is the Experimental Basis of Special Relativity?” URL: http://
math.ucr.edu/home/baez/physics/Relativity/SR/experiments.html#5.%20Twin%
20paradox.
[11] Kostelecky, A., “The Search for Relativity Violations,” Scientific American, Vol.
291, No. 3, Sept. 2004, pp. 93 101.
[12] Perkins, D. H., Introduction to High Energy Physics, 2nd ed., Addison Wesley,
Reading, MA, 1982.
[13] Weisstein, E., World of Physics. URL: http://en.wikipedia.org/wiki/Tachyon#
note feinberg67.
[14] Herbert, N., Faster Than Light Superluminal Loopholes in Physics, New American
Library, Markham, Canada, 1988.
470 B. N. CASSENTI

[15] Ramana Murthy, P. V., “Stability of Protons and Electrons and Limits on Tachyon
Masses,” Physion Review D, Vol. 7, No. 7, 1973, pp. 2252 2253.
[16] Forward, R. L., “A Transparent Derivation of the Relativistic Rocket Equation,”
AIAA/ASME/SAE/ASEE 31st Joint Propulsion Conference and Exhibit, AIAA
Paper 95 3060, San Diego, CA, 10 12 July 1995.
[17] Cassenti, B. N., “Antimatter Rockets and Interstellar Propulsion,” AIAA/ASME/
SAE/ASEE 29th Joint Propulsion Conference and Exhibit, AIAA Paper 93 2007,
San Diego, CA, 28 30 June 1993.
[18] Cassenti, B. N., “High Specific Impulse Antimatter Rockets,” AIAA/ASME/SAE/
ASEE 27th Joint Propulsion Conference and Exhibit, AIAA Paper 91 2548,
Sacramento, CA, 24 26 June 1991.
[19] Bussard, R. W., “Galactic Matter and Interstellar Flight,” Astronatica Acta, Vol. VI,
1960, pp. 179 195.
[20] Cassenti, B. N., and Coreano, L., “The Interstellar Ramjet,” AIAA/ASME/SAE/
ASEE 40th Joint Propulsion Conference and Exhibit, AIAA Paper 2004 3568,
Fort Lauderdale, FL. 11 14 July 2004.
[21] Cassenti, B. N., “Design Concepts for the Interstellar Ramjet,” Journal of the British
Interplanetary Society, Vol. 46, 1993, pp. 151 160.
 Chapter 15

Faster-than-Light Approaches in General Relativity

Eric W. Davis
Institute for Advanced Studies at Austin, Austin, Texas

I. Introduction
T was nearly two decades ago when science fiction media (TV, film, and
I novels) began to adopt traversable wormholes, and more recently “star-
gates,” for interstellar travel schemes that allowed their heroes to travel
throughout our galaxy. In 1985 physicists M. Morris and K. Thorne at
CalTech discovered the principle of traversable wormholes based on
Einstein’s General Theory of Relativity published in 1915. Morris and
Thorne [1] and Morris et al. [2] did this as an academic exercise and in
the form of problems for a physics final exam, at the request of Carl
Sagan who had then completed the draft of his novel Contact. This little
exercise led to a continuous line of insights and publications in general
relativity research, i.e., the study of traversable wormholes and time
machines. Wormholes are hyperspace tunnels through spacetime connecting
either remote regions within our universe or two different universes; they
even connect different dimensions and different times. Space travelers
would enter one side of the tunnel and exit out the other, passing through
the throat along the way. The travelers would move through the wormhole
at c, where c is the speed of light (3  108 m/s) and therefore not violate
Special Relativity, but external observers would view the travelers as
having traversed multi-light-year distances through space at faster-than-
light (FTL) speed. A “stargate” was shown to be a very simple special
class of traversable wormhole solutions to Einstein’s general relativistic
field equation [3,4].

Copyright # 2008 by the American Institute of Aeronautics and Astronautics, Inc. All rights
reserved.

Senior Research Physicist.

471
472 E. W. DAVIS

This development was later followed by M. Alcubierre’s formulation in 1994

of the “warp drive” spacetime metric,† which was another solution to the general
relativistic field equation [5]. Alcubierre derived a metric motivated by cosmolo-
gical inflation that would allow arbitrarily short travel times between two distant
points in space. The behavior of the warp drive metric provides for the simul-
taneous expansion of space behind the spacecraft and a corresponding contrac-
tion of space in front of the spacecraft. The warp drive spacecraft would
appear to be “surfing on a wave” of spacetime geometry. A spacecraft can be
made to exhibit an arbitrarily large apparent FTL speed ( c) as viewed by exter-
nal observers, but its moving local rest frame never travels outside of its local
comoving light cone and thus does not violate Special Relativity. A variety of
warp drive research ensued and papers were published during the next 14 years.
The implementation of FTL interstellar travel via traversable wormholes,
warp drives, or other FTL spacetime modification schemes generally requires
the engineering of spacetime into very specialized local geometries. The analysis
of these via the general relativistic field equation plus the resultant source matter
equations of state demonstrates that such geometries require the use of “exotic”
matter in order to produce the requisite FTL spacetime modification. Exotic
matter is generally defined by general relativity physics to be matter that
possesses (renormalized) negative energy density (sometimes negative
stress-tension positive outward pressure, a.k.a. gravitational repulsion or anti-
gravity), and this is a very misunderstood and misapplied term by the non-general
relativity community. We clear up this misconception by defining what negative
energy is and where it can be found in nature, as well as reviewing the experi-
mental concepts that have been proposed to generate negative energy in the
laboratory. Also, it has been claimed that FTL spacetimes are not plausible
because exotic matter violates the general relativistic energy conditions.
However, it has been shown that this is a spurious issue. The identification, mag-
nitude, and production of exotic matter is seen to be a key technical challenge,
however. FTL spacetimes also possess features that challenge the notions of
causality and there are alleged constraints placed upon them by quantum
effects. These issues are reviewed and summarized, and an assessment on the
present state of their resolution is provided.

II. General Relativistic Definition of Exotic Matter and the

Energy Conditions
In classical physics the energy density of all observed forms of matter (fields)
is non-negative. What is exotic about the matter that must be used to generate
FTL spacetimes is that it must have negative energy density and/or negative
flux [6]. The energy density is “negative” in the sense that the configuration of
matter fields that must be deployed to generate and thread a traversable wormhole

†
A spacetime metric (ds 2) is a Lorentz invariant distance function between any two points in
spacetime that is defined by ds 2 ¼ gmn dx m dx n, where gmn is the metric tensor, which is a 4  4
matrix that encodes the geometry of spacetime and dxm is the infinitesimal coordinate separation
between two points. The Greek indices (m, n ¼ 0 . . . 3) denote spacetime coordinates, x 0 . . . x 3, such
that x 1. . . x 3 ; space coordinates and x 0 ; time coordinate.
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 473

throat or a warp drive bubble must have an energy density, rE ( rc 2, where r is

the rest-mass density), that is less than or equal to its pressures, pi [1,3].‡ In many
cases, these equations of state are also known to possess an energy density that is
algebraically negative, i.e., the energy density and flux are less than zero. It is on
the basis of these conditions that we call this material property “exotic.” The con-
dition for ordinary, classical (nonexotic) forms of matter that we are all familiar
with in nature is that rE . pi and/or rE  0. These conditions represent two
examples of what are variously called the “standard” energy conditions: weak
energy condition (WEC: rE  0, rE þ pi  0), null energy condition (NEC:
rE þ pi  0), dominant energy condition (DEC), and strong energy condition
(SEC). These energy conditions forbid negative energy density between material
objects to occur in nature, but they are mere hypotheses. Hawking and Ellis [7]
formulated the energy conditions in order to establish a series of mathematical
hypotheses governing the behavior of collapsed-matter singularities in their
study of cosmology and black hole physics. More specifically, classical
general relativity allows one to prove lots of general theorems about the behavior
of matter in gravitational fields. Some of the most significant of these general
theorems are:

1) Focusing theorems (for gravitational lensing)

2) Singularity theorems (for gravity-dominated collapse leading to black
holes possessing a singularity)
3) Positive mass theorem (for positive gravitational mass)
4) Topological censorship (i.e., you cannot build a wormhole)

Each of these requires the existence of “reasonable” types of matter for their for-
mulation, i.e., matter that satisfies the energy conditions. In what follows, we will
not consider the impact or implications of the DEC or SEC because they add no
new information beyond the WEC and NEC.
The bad news is that real physical matter is not “reasonable” because the
energy conditions are, in general, violated by semiclassical quantum effects
(occurring at order h , which is Planck’s reduced constant, 1.055  10 34 J . s)
[3]. More specifically, quantum effects generically violate the averaged NEC
(ANEC). Furthermore, it was discovered in 1965 that quantum field theory has
the remarkable property of allowing states of matter containing local regions
of negative energy density or negative fluxes [8]. This violates the WEC,
which postulates that the local energy density is non-negative for all observers.
And there are also general theorems of differential geometry that guarantee
that there must be a violation of one, some, or all of the energy conditions
(meaning exotic matter is present) for all FTL spacetimes. With respect to creat-
ing FTL spacetimes, “negative energy” has the unfortunate reputation of alarm-
ing physicists. This is unfounded because all the energy condition hypotheses
have been tested in the laboratory and experimentally shown to be false
25 years before their formulation [9].

‡
Latin indices (e.g., i, j, k ¼ 1 . . . 3) affixed to physical quantities denote the usual three
dimensional space coordinates, x1. . . x3, indicating the spatial components of vector or tensor quantities.
474 E. W. DAVIS

Further investigation into this technical issue showed that violations of the
energy conditions are widespread for all forms of both “reasonable” classical
and quantum matter [10 14]. Furthermore, Visser [3] showed that all (generic)
spacetime geometries violate all the energy conditions. So the condition that
rE . pi and/or rE  0 must be obeyed by all forms of matter in nature is
spurious. Violating the energy conditions commits no offense against nature.
Negative energy has been produced in the laboratory and this will be discussed
in the following sections.

A. Examples of Exotic or “Negative” Energy Found in Nature

The exotic (energy condition-violating) fields that are known to occur in
nature are:
1) Static, radially dependent electric or magnetic fields. These are borderline
exotic, if their tension were infinitesimally larger, for a given energy density [7,15].
2) Squeezed quantum vacuum states: electromagnetic and other (non-
Maxwellian) quantum fields [1,16].
3) Gravitationally squeezed vacuum electromagnetic zero-point fluctuations [17].
4) Casimir effect, i.e., the Casimir vacuum in flat, curved, and topological
spaces [18 24].
5) Other quantum fields, states, or effects. In general, the local energy density
in quantum field theory can be negative due to quantum coherence effects [8].
Other examples that have been studied are Dirac field states: the superposition
of two single particle electron states and the superposition of two
multi-electron positron states [25,26]. In the former/latter, the energy densities
can be negative when two single/multi-particle states have the same number
of electrons/electrons and positrons or when one state has one more electron/
electron positron pair than the other.

Cosmological inflation [3], cosmological particle production [3], classical

scalar fields [3], the conformal anomaly [3], and gravitational vacuum polariz-
ation [10 13] are among many other examples that also violate the energy con-
ditions. Because the laws of quantum field theory place no strong restrictions on
negative energies and fluxes, then it might be possible to produce exotic phenom-
ena such as FTL travel [5,27,28], traversable wormholes [1 3], violations of the
second law of thermodynamics [29,30], and time machines [2,3,31]. There are
several other exotic phenomena made possible by the effects of negative
energy, but they lie outside the scope of the present study. In the following
section, we review the previously listed items 1 through 4 and examine their
applicability and technical maturity. Dirac field states are currently under study
by investigators. Also, we do not consider the issue of capturing and storing nega-
tive energy in what follows because free-space negative energy sources appear to
be a more desirable option for inducing FTL spacetimes than stored negative
energy, and because there is very little technical literature that addresses how
to capture and store negative energy (see Ref. 6). The issue of capturing and
storing negative energy will be left for future investigations.
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 475

B. Generating Negative Energy in the Lab

1. Static Radial Electric and Magnetic Fields
It is beyond the scope of this study to include all the technical configurations
by which one can generate static, radially dependent electric or magnetic fields.
Suffice it to say that ultrahigh-intensity tabletop lasers have been used to generate
extreme electric and magnetic field strengths in the lab. Ultrahigh-intensity lasers
use the chirped-pulse amplification (CPA) technique to boost the total output
beam power. All laser systems simply repackage energy as a coherent package
of optical power, but CPA lasers repackage the laser pulse itself during the ampli-
fication process. In typical high-power short-pulse laser systems, it is the peak
intensity, not the energy or the fluence, that causes pulse distortion or laser
damage. However, the CPA laser dissects a laser pulse according to its frequency
components and reorders it into a time-stretched lower-peak-intensity pulse of
the same energy [32 34]. This benign pulse can be then amplified safely to
high energy, and then only afterwards reconstituted as a very short pulse of enor-
mous peak power, a pulse that could never itself have passed safely through the
laser system. Made more tractable in this way, the pulse can be amplified to
substantial energies (with orders of magnitude greater peak power) without
encountering intensity-related problems.
The extreme output beam power, fields, and physical conditions that have
been achieved by ultrahigh-intensity tabletop lasers are [34]:
1) Power intensity  1019 to 1030 W/m2 (1034 W/m2 using SLAC as
a booster).
2) Peak power pulse 103 fs.
3) Electric field, E  1014 to 1018 V/m. [Note: Compare this with the
critical quantum electrodynamic (QED) vacuum breakdown E-field intensity,
Ec 2m2e c 3/h  e  1018 V/m, defined by the total rest-energy of an electron posi-
tron pair created from the vacuum divided by the electron’s Compton wavelength,
where me is the electron mass (9.11  10 31 kg) and e is the electron charge
(1.602  10 19 C)].
4) Magnetic field, B  several  106 Tesla (note: The critical QED vacuum
breakdown B-field intensity is Bc Ec/c  1010 Tesla).
5) Ponderomotive acceleration of electrons  1017 to 1030g (g is the
acceleration of gravity near the Earth’s surface, 9.81 m/s2).
6) Light pressure 109 to 1015 bars.
7) Plasma temperatures .1010 K.
The vigilant reader might assert that the electric and magnetic fields generated
by ultrahigh-intensity lasers are not static. But in fact, these fields are static over
the duration of the pulsewidth while at peak intensity. We find from the above
data that ultrahigh-intensity lasers can generate an electric field energy
density 1016 to 1028 J/m3 and a magnetic field energy density 1019 J/m3.
However, there remains the problem of engineering this type of experiment
because classical electromagnetic theory states that every observer associated
with the experiment will see a non-negative energy density that is / E 2 þ B 2,
where E and B are measured in an observer’s reference frame. It is not known
how to increase the tension in these fields using current physics, but some new
476 E. W. DAVIS

physics may provide an answer. This technical problem must be left for future
investigation.

2. Squeezed Quantum Vacuum

Substantial theoretical and experimental work has shown that in many
quantum systems the limits to measurement precision imposed by the quantum
vacuum zero-point fluctuations (ZPF) can be breached by decreasing the noise
in one observable (or measurable quantity) at the expense of increasing the
noise in the conjugate observable; at the same time the variations in the first
observable, say the energy, are reduced below the ZPF such that the energy
becomes “negative.” “Squeezing” is thus the control of quantum fluctuations
and corresponding uncertainties, whereby one can squeeze the variance of one
(physically important) observable quantity provided the variance in the (phys-
ically unimportant) conjugate observable is stretched/increased. The squeezed
quantity possesses an unusually low variance, meaning less variance than
would be expected on the basis of the equipartition theorem. One can in principle
exploit quantum squeezing to extract energy from one place in the ordinary
vacuum at the expense of accumulating excess energy elsewhere [1].
The squeezed state of the electromagnetic field is a primary example of a
quantum field that has negative energy density and negative energy flux. Such
a state became a physical reality in the laboratory as a result of the nonlinear-
optics technique of “squeezing,” i.e., of moving some of the quantum fluctuations
of laser light out of the cos[v(t z/c)] part of the beam and into the sin[v(t z/
c)] part (v is the angular frequency of light, t is time, and z denotes the z-axis
direction of beam propagation) [16,35 39]. The observable that gets squeezed
will have its fluctuations reduced below the vacuum ZPF. The act of squeezing
transforms the phase space circular noise profile characteristic of the vacuum
into an ellipse, whose semimajor and semiminor axes are given by unequal quad-
rature uncertainties (of the quantized electromagnetic field harmonic oscillator
operators). This applies to coherent states in general, and the usual vacuum is
also a coherent state with eigenvalue zero. As this ellipse rotates about the
origin with angular frequency v, these unequal quadrature uncertainties manifest
themselves in the electromagnetic field oscillator energy by periodic occurrences,
which are separated by one quarter cycle of both smaller and larger fluctuations
compared to the unsqueezed vacuum.
Morris and Thorne [1] and Caves [40] point out that if one squeezes the
vacuum (i.e., if one puts vacuum rather than laser light into the input port of a
squeezing device), then one gets at the output an electromagnetic field with
weaker fluctuations and thus less energy density than the vacuum at locations
where cos2[v(t z/c)] ffi 1 and sin2[v(t z/c)]  1; but with greater fluctu-
ations and thus greater energy density than the vacuum at locations where
cos2[v(t z/c)]  1 and sin2[v(t z/c)] ffi 1. Because the vacuum is defined
to have vanishing energy density, any region with less energy density than the
vacuum actually has a negative (renormalized) expectation value for the
energy density. Therefore, a squeezed vacuum state consists of a traveling elec-
tromagnetic wave that oscillates back and forth between negative energy density
and positive energy density, but has positive time-averaged energy density.
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 477

For the squeezed electromagnetic vacuum state, the energy density rE-sqvac is
given by [41]:
  
2hv
rE-sqvac ¼ sinh j ½sinh j þ cosh j cosð2v (t  z=c) þ dÞ (J=m3 ) (1)
L3

where L 3 is the volume of a large box with sides of length L (i.e., we put the
quantum field inside a box with periodic boundary conditions), j is the squeezed
state amplitude (giving a measure of the mean photon number in a squeezed
state), and d is the phase of squeezing. Equation (1) shows that rE-sqvac falls
below zero once every cycle when the condition cosh j . sinh j is met. It turns
out that this is always true for every nonzero value of j, so rE-sqvac becomes nega-
tive at some point in the cycle for a general squeezed vacuum state. On another
note, when a quantum state is close to a squeezed vacuum state, there will almost
always be some negative energy densities present.
Negative energy can be generated by an array of ultrahigh-intensity lasers
using an ultrafast rotating mirror system [42]. In this scheme a laser beam is
passed through an optical cavity resonator made of a lithium niobate (LiNbO3)
crystal that is shaped like a cylinder with rounded silvered ends to reflect light.
The resonator will act to produce a secondary lower frequency light beam in
which the pattern of photons is rearranged into pairs. The squeezed light beam
emerging from the resonator will contain pulses of negative energy interspersed
with pulses of positive energy.
In this concept both the negative and positive energy pulses are 10 15
second in duration. We could, in principle, arrange a set of rapidly rotating
mirrors to separate the positive and negative energy pulses from each other.
The light beam would be set to strike each mirror surface at a very shallow
angle while the rotation would ensure that the negative energy pulses would be
reflected at a slightly different angle from the positive energy pulses. A small
spatial separation of the two different energy pulses would occur at some distance
from the rotating mirror. Another system of mirrors would be needed to redirect
the negative energy pulses to an isolated location and concentrate them there.
Figure 1 illustrates this concept.
The rotating mirror system can actually be implemented via nonmechanical
means. A chamber of sodium gas is placed within the squeezing cavity and a
laser beam is directed through the gas. The beam is reflected back on itself by
a mirror to form a standing wave within the sodium chamber. This wave
causes rapid variations in the optical properties of the sodium, thus causing
rapid variations in the squeezed light so that we can induce rapid reflections of
pulses by careful design [36]. An illustration of this is shown in Fig. 2.
Another way to generate negative energy via squeezed light would be to man-
ufacture extremely reliable light pulses containing precisely one, two, three, etc.,
photons each and combine them together to create squeezed states to order [42].
Superimposing many such states could theoretically produce bursts of intense
negative energy. Figure 3 provides a conceptual diagram of this concept. Photo-
nic crystal research has already demonstrated the feasibility of using photonic
crystal waveguides (mixing together the classical and quantum properties of
478 E. W. DAVIS

Fig. 1 Proposed squeezed light negative energy generator.

Fig. 2 Sodium chamber negative energy separator [36].

Fig. 3 Alternative squeezed light negative energy generator.

 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 479

optical materials) to engineer light sources that produce beams containing

precisely one, two, three, etc., photons. For example, researchers at Melbourne
University used a microwave oven to fuse a tiny diamond, just 1/1000 of a
millimeter long, onto an optical fiber, which could be used to create a single
photon beam of light [43,44]. The combining of different beams containing
different (finite integer) numbers of photons is already state-of-the-art practice
via numerous optical beam combining methods that can readily be extended to
our application.
Finally, Ries et al. [45] experimentally demonstrated the first simple, scalable
squeezed vacuum source in the laboratory that consisted of a continuous-wave
diode laser and an atomic rubidium vapor cell. The experimental tools we
need to begin exploring the generation of negative energy for the purpose of
creating FTL spacetimes are just now becoming available.

3. Gravitationally Squeezed Electromagnetic Zero-Point Fluctuations

A natural source of negative energy comes from the effect product by gravita-
tional fields (of astronomical bodies) in space on the surrounding vacuum. For
example, the gravitational field of the Earth produces a zone of negative energy
around it by dragging some of the virtual quanta (a.k.a. vacuum ZPF) downward.
This concept was initially developed in the 1970s as a byproduct of studies on
quantum field theory in curved space [21]. However, Hochberg and Kephart
[17] derived an important application of this concept to the problem of creating
and stabilizing traversable wormholes. They showed that one can utilize the nega-
tive energy densities, which arise from distortion of the vacuum ZPF due to the
interaction with a prescribed gravitational background, to provide a violation of
the energy conditions. The squeezed quantum states of quantum optics provide
a natural form of matter having negative energy density.
The analysis, via quantum optics, showed that gravitation itself provides the
mechanism for generating the squeezed vacuum states needed to support stable
traversable wormholes. The production of negative energy densities via a
squeezed vacuum is a necessary and unavoidable consequence of the interaction
or coupling between ordinary matter and gravity, and this defines what is meant
by gravitationally squeezed vacuum states. The magnitude of the gravitational
squeezing of the vacuum can be estimated from the quantum optics squeezing
condition for given transverse momentum and (equivalent) energy eigenvalues,
j, of two electromagnetic ZPF field modes, such that this condition is subject
to j ! 0, and it is defined as [17]:
 
4 p c 2 r 2 M
j;
lg R M
(2)
8p r S
¼
l
where l is the ZPF mode wavelength, r is the radial distance from the center of
the astronomical body in question, R is the radius of the Earth (6.378  106 m),
M is the mass of the Earth (5.972  1024 kg), M is the mass of the astronomical
480 E. W. DAVIS

Table 1 Substantial gravitational squeezing occurs for vacuum ZPF

when l  8prS

Mass of body (kg) rS (m) l (m) rE-gsvac (J/m3)

Sun ¼ 2.00  1030 2.95  103 78.0  103 1.69  10 44

Jupiter ¼ 1.90  1027 2.82 74 2.08  10 32
Earth ¼ 5.98  1024 8.87  10 3 0.23 2.23  10 22
Typical mountain  1011 10 16 10 15 6.25  1035
8
Planck mass ¼ 2.18  10 3.23  10 35 8.50  10 34 1.20  10108
Proton ¼ 1.67  10 27 2.48  10 54 6.50  10 53 3.50  10184

body, and rS is the Schwarzschild radius of the astronomical body.§ Note that rS is
only a convenient radial distance parameter for any object under examination and
so there is no black hole collapse involved in this analysis. We can actually
choose any radial distance from the body in question to perform this analysis,
but using rS makes the equation simpler in form. Also note that Eq. (2) contains
an extra factor of two (compared to the j derived in Ref. 17) in order to account
for the photon spin. The squeezing condition plus Eq. (2) simply states that sub-
stantial gravitational squeezing of the vacuum occurs for those ZPF field modes
with l  8prS of the mass in question (whose gravitational field is squeezing
the vacuum). The corresponding local vacuum state energy density will be
rE-gsvac 2p 2hc/l4.
The general result of the gravitational squeezing effect is that as the gravita-
tional field strength increases, the negative energy zone (surrounding the mass)
also increases in strength. Table 1 shows when gravitational squeezing be-
comes important for sample masses and their corresponding rE-gsvac. It shows
that in the case of the Earth, Jupiter, and the Sun, the squeezing effect is extre-
mely feeble because only ZPF mode wavelengths above 0.2 m to 78 km are
affected, each having very minute rE-gsvac. For a solar mass black hole
(2.95 km radius), the effect is still feeble because only ZPF mode wavelengths
above 78 km are affected. But note that Planck mass objects will have an enor-
mously strong negative energy zone surrounding them because all ZPF mode
wavelengths above 8.50  10 34 m will be squeezed, in other words, all wave-
lengths of interest for vacuum fluctuations. Protons will have the strongest nega-
tive energy zone in comparison because the squeezing effect includes all ZPF
mode wavelengths above 6.50  10 53 m. Furthermore, a body smaller than a
nuclear diameter (10 16 m) and containing the mass of a mountain (1011
kg) has a fairly strong negative energy zone because all ZPF mode wavelengths
above 10 15 m will be squeezed. In each of these cases, the magnitude of the cor-
responding rE-gsvac is very large.
However, the estimates for the wavelengths in Table 1 might be too small.
L. H. Ford (private communication, 2007) argues that the analysis in Ref. 17 is

§
rS ¼ 2GM/c 2, where G is Newton’s universal gravitation constant (6.673  10 11 Nm2/kg2).
According to General Relativity Theory, this is the critical radius at which a spherically symmetric
massive body becomes a black hole, that is, at which light is unable to escape from the body’s surface.
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 481

in error because spacetime is flat on scales smaller than the local radius of curva-
ture, which is defined by the inverse square root of the typical Riemann curvature
tensor component in a local orthonormal frame, or ‘C  (r 3c 2/GM)1/2. Accord-
ing to Ford, only ZPF modes with l  ‘C will be squeezed by the gravitational
field. This leads to a different local vacuum state energy density (for r  rS) [11]:

2p 2 hc
rE-gsvac ¼  4
l
2p 2 h c
 (3)
‘4C
2p 2 hG 2 M 2
 (J=m3 )
c3 r 6
For example, near the surface of the Earth (r  R , M M ), ‘C  2.42  1011 m
and hence, Eq. (3) gives rE-gsvac  1.82  10 70 J/m3. Compare these values
with l  0.23 m and rE-gsvac  2.23  10 22 J/m3 in Table 1. The resolution
of this disagreement remains an open question.
We are presently unaware of any way to artificially generate gravitational
squeezing of the vacuum in the laboratory. This will be left for future investi-
gation. However, it is predicted to occur in the vicinity of astronomical matter.
Naturally occurring traversable wormholes in the vicinity of astronomical
matter would therefore become possible.

4. Vacuum Field Stress: Negative Energy from the Casimir Effect

The Casimir effect is by far the easiest and most well-known way to generate
negative energy in the lab. The Casimir effect that is familiar to most people is the
force associated with the electromagnetic quantum vacuum [46]. This is an
attractive force that must exist between any two neutral (uncharged), parallel,
flat, conducting surfaces (e.g., metallic plates) in a vacuum. This force has
been well measured and it can be attributed to a minute imbalance in the electro-
magnetic zero-point energy density inside the cavity between the conducting
surfaces versus the region outside of the cavity [47 49].
It turns out that there are many different types of Casimir effects found in
quantum field theory [18 20,24,50]. For example, if one introduces a single infi-
nite plane conductor into the Minkowski (flat spacetime) vacuum by bringing it
adiabatically from infinity so that whatever quantum fields are present suffer no
excitation but remain in their ground states, then the vacuum (electromagnetic)
stresses induced by the presence of the infinite plane conductor produces a
Casimir effect. This result holds equally well when two parallel plane conductors
(with separation distance d) are present, which gives rise to the familiar Casimir
effect inside a cavity. Note that in both cases, the spacetime manifold is made
incomplete by the introduction of the plane conductor boundary condition(s).
The vacuum region put under stress by the presence of the plane conductor(s)
is called the Casimir vacuum. The generic expression for the energy density of
the Casimir effect is rCE A(h c)d 4, where A z(D)/8p 2 in spacetimes
482 E. W. DAVIS

of arbitrary dimension D [18 20]. The appearance of the zeta-function z(D) is

characteristic of expressions for vacuum stress-energy tensors, kT mn lvac .} In our fam-
iliar four-dimensional spacetime (D 4) we have that A p 2/720. To calculate
kT mn lvac for a given quantum field is to calculate its associated Casimir effect.
Analogs of the Casimir effect also exist for fields other than the electromag-
netic field. When considering the vacuum state of other fields, one must consider
boundary conditions that are analogous to the perfect-conductor boundary
conditions for the electromagnetic field at the surfaces of the plates [18 20,
24]. Other fields are not electromagnetic in nature, that is to say they are non-
Maxwellian, and so the perfect-conductor boundary conditions do not apply to
them. It turns out that complete manifolds exhibit what is called the topological
Casimir effect for any non-Maxwellian fields. In order to define boundary
conditions for other fields we replace the conductor boundary conditions and
Minkowski spacetime by a manifold of the form <  S (i.e., a product space),
where < is the real line defining the time dimension for this particular product
space and S is a flat three-dimensional manifold (three-manifold) having any
one of the following topologies: <2  S1, <  T2, T3, <  K2, etc., < being
the real line that defines any linear space dimension (e.g., < line, <2 two-
dimensional plane, etc.), Tn being the n-torus, K2 the two-dimensional Klein
bottle, S1 the circle, etc.
The case S <2  S1 has the closest resemblance to the electromagnetic
Casimir effect, the difference being that instead of imposing conductor boundary
conditions, one imposes periodic boundary conditions on some of the space coor-
dinates in the three-manifold. When imposing this topological constraint on the
field theoretic calculation of the topological Casimir effect (for linear massless
fields), one finds that the generic expression for the energy density is also
rCE A(hc)d 4, where A +df (p 2/90), df is the number of degrees of
freedom (e.g., helicity states) per spatial point, the plus sign holds for boson
fields (giving a negative energy density), and the negative sign for fermion
fields (giving a positive energy density).
If one were to admit spin structure in the manifolds described above and the
field is spinorial, then there is another important subtlety that must be taken into
account when evaluating kT mn lvac . However, this introduces an additional com-
plexity involving the relationship between the spin structure and the global struc-
ture (i.e., the configuration space or fibre bundle) of the field in question whereby
the topology not only of the base manifold but of the fibre bundle itself has an
effect on kT mn lvac . In addition to this, there are (compactified) extra-space dimen-
sional quantum field (i.e., D-Brane or “brane world”) analogs of the Casimir
effect yet to be explored. But a detailed consideration of these for producing
FTL spacetimes is beyond the scope of this chapter and will be left for future
investigation. One proposal that exploits a brane world analog of the Casimir
effect to create a warp drive is described in Sec. III.B.
As a final note, we point out that the methods used to obtain the electromag-
netic kT mn lvac between parallel plane conductors can be used also when the

}
The angular brackets denote the quantum (vacuum state) expectation value of the stress energy
tensor T mn. Also note that stress energy is synonymous with energy momentum.
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 483

conductors are not parallel but are joined together along a line of intersection. If
the conductors have curved surfaces instead, then one obtains results that are
similar to the case of intersecting conductors. These geometries have also been
evaluated for the case of dielectric media. These particular cases will not be
considered further because there are technical subtleties involved that complicate
the calculations and application of the different approaches. This topic will also
be left for future investigation.

5. Casimir Effect: Moving Mirrors

Negative energy can be created by a single moving reflecting (conducting)
surface (a.k.a. a moving mirror). A mirror moving with increasing acceleration
generates a flux of negative energy that emanates from its surface and flows
out into the space ahead of the mirror [21,51]. This is essentially the simple
case of an infinite plane conductor undergoing acceleration perpendicular to its
surface. If the acceleration varies with time, the conductor will generally emit
or absorb photons (i.e., exchange energy with the vacuum), even though it is
neutral. This is an example of the well-known quantum phenomenon of para-
metric excitation. The parameters of the electromagnetic field oscillators (e.g.,
their frequency distribution function) change with time owing to the acceleration
of the mirror [52]. However, this effect is known to be exceedingly small, and it is
not the most effective way to produce negative energy for our purposes. We will
not consider this scheme any further.

6. Casimir Effect: Negative Energy for Traversable Wormholes

The electromagnetic Casimir effect can, in principle, be used to create a tra-
versable wormhole. The energy density rCE (p 2h c/720)d 4 within a
Casimir cavity is negative and manifests itself by producing a force of attraction
between the cavity walls. But cavity dimensions must be made exceedingly small
in order to generate a significant amount of negative energy for our purposes. In
order to use the Casimir effect to generate a spherically symmetric traversable
wormhole throat of radius rthroat, we will need to design a cavity made of
perfectly conducting spherically concentric thin plates with a plate separation
d of [2]:

 1 r !12
p3 4 hG
d¼ rthroat
30 c3 (4)
 p
¼ 4:05  10 18 rthroat (m)

To counteract the collapse of the cavity due to the Casimir force acting
between the plates, the plates will have equal electric charges placed upon
them to establish adequate Coulomb repulsion. (In a detailed analysis, the
electrostatic energy required to support the Coulomb repulsion between the
plates would be considered separately.) Equation (4) shows that a 1-km
radius throat will require a cavity plate separation of 1.28  10 16 m
(smaller than a nuclear diameter), which gives rEC 1.62  1036 J/m3
484 E. W. DAVIS

for this configuration. In contrast, a wormhole with a throat radius of 1 AU

(mean Earth Sun distance, 1.50  1011 m) will require a plate separation of
1.57  10 12 m (or 35% smaller than the electron’s Compton wavelength),
which results in an energy density of 7.14  1019 J/m3. There is no tech-
nology known today that can engineer a cavity with such minuscule plate
separations. In addition, such minuscule plate separations are unrealistic
because the Casimir effect switches over to the nonretarded field behavior
(d 3) of van der Waals forces when plate separations go below the wave-
length (10 nm) where they are no longer perfectly conducting [53]. We will
not consider this scheme further. However, future work will be necessary to
elucidate whether the various quantum field analogs of the Casimir effect can
provide a more reasonable technical solution.

III. Brief Review of Faster-than-Light Spacetimes

In Section I we briefly described the two primary classes of FTL spacetimes
found in General Relativity Theory: traversable wormholes and warp drives.
There are as yet no other (credible) classes of FTL spacetimes having to do
with General Relativity Theory, higher dimensional D-Brane theory, or (cred-
ible) alternative theories of gravity (e.g., Puthoff’s Polarizable-Vacuum Rep-
resentation of General Relativity, Yilmaz theory, etc.). Investigators have
discovered that General Relativity Theory and all of the (credible) alternative the-
ories of gravity are infested with traversable wormholes, time machines, and
some form of a warp drive. The small number of proposed alternative FTL space-
times is largely speculative and not yet on the same level of rigor as the warp
drive and traversable wormhole spacetimes.
How does one study the physics of FTL spacetimes within the framework
of General Relativity Theory? When studying spacetime physics, the normal
philosophy is to take the general relativistic field equation, add some form of
matter, make simplifying assumptions, and then solve to deduce what the
geometry of spacetime will be. This is very difficult to do because there
are 10 nonlinear, second-order partial differential equations with four redun-
dancies (arbitrary choice of spacetime coordinates) and four constraints
(energy-momentum conservation). There is a tremendous body of research
that takes exactly this approach, either analytically or numerically. However,
this is not the best strategy for understanding FTL spacetimes. The appropr-
iate strategy is to decide beforehand on the definition of the desired travers-
able wormhole or warp drive and decide what the spacetime geometry should
look like. Given the desired geometry, use the general relativistic field
equation to calculate the distribution of matter required to set up this geome-
try. Then one needs to assess whether the required distribution of matter is
physically reasonable and whether it violates any basic rules of physics, etc.
In the following sections we briefly outline the key results for traversable
wormhole and warp drive spacetimes.

A. Traversable Wormholes
Traversable wormholes represent a class of exact metric solutions of the
general relativistic field equation. The solutions are “exact” in the sense that
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 485

no approximations requiring a plethora of physical assumptions have to be made

to derive the appropriate spacetime geometry. To define a stable traversable
wormhole one needs to define the desirable physical requirements it is to have
in order to achieve the desired FTL travel benefit. The desired requirements
are the following [1,3]:
1) Travel time through the wormhole tunnel or throat should be  1 year as
seen by both the travelers and outside static observers.
2) Proper time as measured by travelers should not be dilated by relativistic
effects.
3) The gravitational acceleration and tidal-gravity accelerations between
different parts of the travelers’ body should be  1g when going through the
wormhole.
4) Travel speed through the tunnel/throat should be ,c.
5) Travelers (made of ordinary matter) must not couple strongly to the
material that generates the wormhole curvature; the wormhole must be threaded
by a vacuum tube through which the travelers can move.
6) There is no event horizon at the wormhole throat.
7) There is no singularity of infinitely collapsed matter residing at the worm-
hole throat.
See item 5b in Box 3 of Ref. 1 for the technical description of a trip through a
wormhole. These requirements then lead us to define a spherically symmetric
Lorentzian spacetime metric, ds 2, that prescribes the required traversable worm-
hole geometry [1,3]:

ds2 ¼ e2f(r) c2 dt2 þ ½1  b(r)=r 1 dr 2 þ r 2 dV2 (5)

where standard spherical coordinates are used (r: 2pr circumference;

0  u  p; 0  w  2p), t is time ( 1 , t , 1), dV2 du2 þ sin2udw2,
f(r) is the freely specifiable redshift function that defines the proper time lapse
through the wormhole throat, and b(r) is the freely specifiable shape function
that defines the wormhole throat’s spatial (hypersurface) geometry. The throat
is spherically shaped. There are several variations of Eq. (5), which define traver-
sable wormholes having different properties. The reader should consult Ref. 3 for
further details.
Figure 4 shows two diagrams representing the embedded space (Flamm) rep-
resentation of Eq. (5), which depicts the geometry of an equatorial (u p/2)
slice through space at a specific moment of time (t constant). The top of
Fig. 4 shows the embedding diagram for a traversable wormhole that connects
two different universes (i.e., an inter-universe wormhole). The bottom diagram
is an intra-universe wormhole with a throat that connects two distant regions
of our own universe. These diagrams aid in visualizing traversable wormhole
geometry and are merely a geometrical exaggeration.
There was originally one other criterion for defining a traversable wormhole,
which was that it must be embedded within the surrounding (asymptotically) flat
spacetime. However, Hochberg and Visser [54] proved that it is only the behavior
near the wormhole throat that is critical to understanding the physics, and that
486 E. W. DAVIS

Fig. 4 Inter-universe wormhole (top) and intra-universe wormhole (bottom).

(Courtesy of Ron Koczor, NASA-Marshall.)

a generic throat can be defined without having to make all the symmetry
assumptions and without assuming the existence of an asymptotically flat
spacetime in which to embed the wormhole. Therefore, one only needs to
know the generic features of the geometry near the throat in order to guarantee
violations of the NEC for certain open regions near the throat. So we are free
to place our wormhole anywhere in spacetime we want because it is only the
geometry and physics near the throat that matters for any analysis. This fact
led to the development of a number of different traversable wormhole throat
designs that are cubic shaped, polyhedral shaped, flat-face shaped, generic
shaped, etc. The reader should consult Ref. 3 for a complete technical review
of the various types (and shapes) of traversable wormhole solutions found in
General Relativity Theory.
We know that we need exotic or negative energy to create and thread open a
traversable wormhole. In this regard, we ask what kind of wormhole can one
make with less effort. To answer this question we can relate the local wormhole
geometry to the global topological invariant of the spacetime via the Gauss
Bonnet theorem [55]. In the Gauss Bonnet theorem the local wormhole geo-
metry is quantified by the energy density, U (in geometrodynamic units,
h G c 1), threading the wormhole throat plus a spatial curvature constant
(for the throat). The global topological invariant of spacetime is quantified by the
Euler number, xe, which is itself defined in terms of the genus, g, representing the
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 487

number of handles (or throats or tunnels) a wormhole can be assigned. These two
topological quantities are related via xe 2(1 g). Therefore, the (static)
wormhole Gauss Bonnet relation is given by U  xe/4 or U  (1 g)/2
[55]. (The case for dynamic traversable wormholes has results that are
similar to the static case.) This relation will help us decide if we want to build
a traversable wormhole having one throat, or two or more throats, and what
energy cost this will incur.
The following is the result of our analysis for traversable wormholes having:
1) One-handle/throat (i.e., flat torus or spherical wormhole topology) giving
g 1, thus xe 0, and so U  0.
2) Two-handles/throats giving g 2, thus xe 2, and so U  1/2.
3) Three-handles/throats giving g 3, thus xe 4, and so U  1;
and so on.
It is clear from this that, as the number of wormhole handles/throats increases,
the amount of negative energy required to create the wormhole will grow
larger in magnitude. This is an undesirable demand on any putative negative
energy generator. It is clear then that item 1 defines the most desirable engineer-
ing solution we can hope for: a one-handle/throat traversable wormhole that will
require zero or (arbitrarily) little negative energy to create. The magnitude of
energy condition violations and the amount of negative energy required to
build a traversable wormhole will be addressed in Sec. IV.

B. Warp Drives
Alcubierre [5] derived a spacetime metric motivated by cosmological inflation
that would allow arbitrarily short travel times between two distant points in
space. The “warp drive” metric uses coordinates (t, x, y, z) and curve (or world-
line) x xsh(t), y 0, z 0, lying in the t x plane passing through the origin.
Note that xsh is the x-axis coordinate position of the moving spaceship (or warp
bubble) frame. The metric specifying this spacetime is (also an exact solution to
the general relativistic field equation) [5]:

ds2 ¼ c2 dt2 þ ½dx  vsh (t) f (rsh (t))dt2 þdy2 þ dz2 (6)

where vsh(t) ; dxsh(t)/dt is the speed associated with the curve (or warp bubble
speed) and rsh(t) ; [(x xsh(t))2 þ y 2 þ z 2]1/2 is the Euclidean distance from the
curve. The warp bubble shape function f (rsh) is any smooth positive function that
satisfies f (0) 1 and decreases away from the origin to vanish when rsh . R for
some distance R. The geometry of each spatial slice is flat, and spacetime is flat
where f (rsh) vanishes but is curved where it does not vanish.
The driving mechanism of Eq. (6) is the York extrinsic time,q. This quantity is
defined as [5]:

vsh xsh df
q¼ (7)
c rsh drsh

The q behavior of the warp drive bubble provides for the simultaneous expansion
of space behind the spacecraft and a corresponding contraction of space in front
488 E. W. DAVIS

Fig. 5 York extrinsic time (q) plot.

of the spacecraft. Figure 5 illustrates the q behavior of the warp drive bubble
geometry. Thus the spacecraft is enveloped within a warp bubble and can be
made to exhibit an arbitrarily large FTL speed (vsh  c) as viewed by external
coordinate observers. Even though the worldlines inside the warp bubble region
are spacelike for all external observers, the moving spaceship (warp bubble)
frame itself never travels outside of its local comoving light cone and thus
does not violate Special Relativity. However, Alcubierre’s warp drive suffers
from the problem that at superluminal speed, the interior of the warp bubble is
causally disconnected from its surface and the exterior region, which means
that photons cannot pass from the inside to the outside of the bubble and so
there is no way of controlling the space warp in order to start, stop, or steer it.
And the starship passengers will have no way of seeing where they are going.
Other investigators have designed warp drive metrics similar to Eq. (6) but
with some modifications to the space warp geometry and corresponding
changes to the negative energy requirement [56 63]. A well-known example
is the proposal by Krasnikov [56] to fix the warp bubble causal disconnection
problem by specifying a “warp tube” spacetime in which a starship traveling
one way at ultra-relativistic speed creates a tube-shaped space warp behind
itself, and then it would return by traveling back through the warp tube at FTL
speed. Thus a starship can return from its interstellar journey shortly after it
left no matter how far away it traveled. White and Davis [64] give another
example in which Alcubierre’s warp drive can be reinterpreted in extra-space
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 489

dimensional brane-world theory as a spacetime expansion boost (i.e., like a scalar

multiplier acting) on the initial spacecraft speed. This mechanism recasts the
warp drive energy requirement into the equation of state for dark energy (a.k.a.
the vacuum energy of Einstein’s cosmological constant; see Section III.D in
Chapter 4 for more details), whereby there is no negative energy density but
only a negative pressure that is required to construct the warp drive effect. An
alternative model is the proposal by Obousy [65] and Obousy and Cleaver [66]
to engineer a version of Alcubierre’s warp drive via exploiting the Casimir
energy of the compactified extra-space dimensions in brane-world theories.
They propose that the vacuum energy of Einstein’s cosmological constant is a
function of the size of the extra-space dimensions and that if one could locally
manipulate the size of an extra-space dimension and thus locally manipulate
the cosmological constant, then one could facilitate the expansion or contraction
of the spacetime surrounding a spacecraft at will. Their model utilizes theoretical
results from brane-world theories, which show that the Casimir energy of extra-
space dimensions is related to the cosmological constant. But their warp drive
model differs entirely in that the approach they use is based on the physics of
higher-dimensional quantum field theory and not General Relativity Theory.
Finally, Puthoff [67] showed that Alcubierre’s warp drive is a particular case
of a broad, general approach that is called “metric engineering,” the details of
which provide support for the concept that reduced-time interstellar travel is
not fundamentally constrained by physical principles. The magnitude of energy
condition violations and the amount of negative energy required to build an Alcu-
bierre-type warp drive will be addressed in the next section.

IV. Make or Break Issues

A. Negative Energy Requirements and Energy Condition Violations
We know how to make small quantities of negative energy in the lab, but we
do not know if it is possible to make large quantities of negative energy. It was
pointed out in Section II that one, some, or all of the energy conditions must be
violated in order to build a FTL spacetime. And we also cautioned that this was
not a showstopper because the energy conditions have all been violated by nature
or by lab experiment prior to their formulation. However, the reader should be
forewarned that there are a number of published claims that the energy condition
violations can be avoided. M. Visser (unpublished lecture, 1998) points out that
these claims are just semantic games whereby investigators universally invoke
the following scenario: Divide the total stress-energy into weird matter plus
normal matter, push all the energy condition violations into the weird matter so
that the normal matter does not violate the energy conditions. Given that the
energy conditions are not absolute, such rearranging approaches are not necessary.

1. Traversable Wormhole Requirements

Traversable wormhole throats violate the NEC (or ANEC). So we ask how big
a violation is required. The answer is that we only need to calculate the amount of
negative energy that will be needed to generate and hold open a wormhole throat.
A simple formula for short-throat wormholes using the thin shell formalism gives
490 E. W. DAVIS

this quantity in terms of the equivalent mass (note: the energy density derived
from the general relativistic field equation is too complex to use for this mass
comparison) [3]:

‘throat c2
Mwh ¼ 
G
‘throat
¼ (1:35  1027 kg) (8)
1m
‘throat
¼ (0:71 MJ )
1m
where Mwh is the equivalent mass required to build the wormhole, ‘throat is a suit-
able measure of the linear dimension (width or diameter) of the throat, and MJ is
the mass of the planet Jupiter. One can also obtain the required energy, Ewh, by
multiplying both sides of Eq. (8) by c 2. Equation (8) shows that a mass of
0.71 MJ will be required to build a wormhole 1 m in size. As the wormhole
size increases, the mass requirement grows negative-large. Table 2 presents a
tabulation of the required negative (equivalent) mass as a function of sample
wormhole throat sizes. After being alarmed by the magnitude of the results,
one should note that Mwh is not the total mass of the wormhole as seen by
remote observers. The nonlinearity of the general relativistic field equation dic-
tates that the total mass is zero (actually, the total net mass being positive, nega-
tive, or zero in the Newtonian approximation depending on the details of the
negative energy configuration constituting the wormhole system). Finally,
Visser et al. [68] demonstrated the existence of spacetime geometries containing
traversable wormholes that are supported by arbitrarily small quantities of nega-
tive energy, and this was proved to be a general result. We will expand on this
further in Sec. IV.B.

2. Warp Drive Requirements

Warp drive spacetimes violate the WEC and NEC (also the DEC and SEC).
The amount of negative energy required to create a warp drive gives a

Table 2 Negative equivalent mass required for traversable

wormholea

‘throat (m) Mwh

1000 709.9 MJ
100 71 MJ
10 7.1 MJ
1 0.71 MJ
0.1 22.6 M
0.01 2.3 M
a
MJ ¼ 1.90  1027 kg, M ¼ 5.98  1024 kg.
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 491

measure of the WEC/NEC violations. Because the energy density for the
Alcubierre [5] and Natário [69] warp drive that is derived from the general
relativistic field equation is too complex to use for these comparative purposes,
we instead use a more simple formula to express the net energy required,
Ewarp, to build a warp bubble around a spaceship [70]:

v2warp c4 R2 s
E warp ¼ 
G (9)
 
¼  1:21  1044 v2warp R2 s

where vwarp [vwarp: (0, 1)] is the dimensionless speed of the warp bubble, R (. 0)
is the radius of the warp bubble, and s (. 0) is proportional to the inverse of the
warp bubble wall thickness D (i.e., s  1/D). One can also obtain the equivalent
mass, Mwarp, by dividing both sides of Eq. (9) by c 2. Equation (9) characterizes
the amount of negative energy that one needs to localize in the walls of the warp
bubble. Table 3 presents a tabulation of the required negative energy as a function
of the “warp factor,” vwarp. One can compare the values of Ewarp in the table with
the positive rest-energy contained in the Sun (1.79  1047 J). The consequence of
Eq. (9) and Table 3 is that if one wants to travel at hyperlight speeds, then the
warp bubble energy requirement will be an enormous negative number. And
this remains true even if one engineers an arbitrarily low sublight-speed warp
bubble. Engineering a warp drive bubble is quite daunting given these results.
One further complication arises from the serious flaw that was discovered in
the Alcubierre and Natário warp drives by Lobo and Visser [70]. They point
out that in the original version of the warp drive, the point at the center of the
warp bubble moves on a geodesic and is massless. The spaceship is always
treated as a test particle in Alcubierre’s warp drive and in all other incarnations
of it. Lobo and Visser corrected this flaw by constructing a more realistic model
of the warp drive spacetime by applying linearized gravity to the weak-field warp
drive case and testing the energy conditions to first and second orders of vwarp.
The fundamental basis of their model is that it specifically includes a finite
mass spaceship that interacts with the warp bubble. Their results verified that

Table 3 Negative energy required for warp bubblea

vwarp Ewarp (J)

5
10 (¼ 3 km/s) 3.03  1040
10 4 (¼ 30 km/s) 3.03  1042
0.01 (¼ 3000 km/s) 3.03  1046
0.5 (¼ 150,000 km/s) 7.59  1049
1 (¼ light speed) 3.03  1050
2 (¼ 600,000 km/s) 1.21  1051
10 (¼ 3.0  106 km/s) 3.03  1052
100 (¼ 3.0  107 km/s) 3.03  1054
a
Assume: R ¼ 50 m, s ¼ 103 m 1
.
492 E. W. DAVIS

all warp drive spacetimes violate the energy conditions and will continue to do so
for arbitrarily low warp bubble speed. They also found that the energy condition
violations in this class of spacetimes is generic to the form of the geometry under
consideration and is not a side effect of the superluminal properties. Based on
these facts plus Eq. (9) and Table 3, it appears that for all conceivable laboratory
experiments in which negative energy can be created in very small amounts,
the warp bubble speed will be absurdly low. It therefore appears unlikely that
warp drives will ever prove to be technologically practical unless new warp
bubble geometries or ways to generate astronomical amounts of negative
energy are found.

B. Quantum Inequalities and Spatial Distributions of Negative Energy

The quantum inequalities (QI) conjecture is an extension of the Heisenberg
uncertainty principle to curved spacetimes. Much research has been conducted
around this one topic alone. The literature is too numerous to cite here but the
reader should consult Refs. 6 and 41 for detailed information. The QI conjecture
relates (via model-dependent time integrals of the energy density along geode-
sics) the energy density of a free quantum field and the time during which this
energy density is observed. This conjecture was devised as an attempt to quantify
the amount of negative energy or energy condition violations required to build a
FTL spacetime. Investigators have invoked the QI to rule out many of the macro-
scopic wormhole and warp drive spacetimes. When generating negative energy
the QI postulate that: 1) the longer the pulse of negative energy lasts, the
weaker it must be; 2) a pulse of positive energy must follow and the magnitude
of the positive pulse must exceed that of the initial negative pulse; and 3) the
longer the time interval between the two pulses, the larger the positive pulse
must be. This actually sounds quite reasonable on energy conservation grounds
until one discovers that the Casimir effect and its non-Maxwellian quantum
field analogs violate all three conditions. There are also a number of squeezed
vacuum sources and Dirac field states that manifestly violate all three conditions.
Cosmological inflation, cosmological particle production, classical scalar fields,
the conformal anomaly, and gravitational vacuum polarization are among the
many other examples that also violate the QI. Visser [71] also points out that
observational data indicate that large amounts of “exotic matter” are required
to exist in the universe in order to account for the observed cosmological evol-
ution parameters. The QI have also not been verified by laboratory experiments.
The assumptions used to derive the QI and the efficacy of their derivation for
various cases has been called into question by numerous investigators. Krasnikov
[72] constructed an explicit counterexample for generalized FTL spacetimes
showing that the relevant QI breaks down even in the simplest FTL cases. And
he also addressed Fewster’s [73] technical arguments on this issue. It is important
to point out that the QIs have been mainly proven for free massless scalar fields in
flat two-dimensional Minkowski spacetime, so there remains the unanswered
questions of extending the QI into a four-dimensional curved spacetime model
(with or without boundaries) and how much negative energy density can arise
for interacting fields.
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 493

It turns out that Visser and coworkers [60,68,70,74] developed a superior way
to properly quantify the amount of negative energy or energy condition violations
required to build a FTL spacetime. They propose a quantifier in terms of a spatial
volume integral, which amounts to calculating the following definite integrals
[68,70,74]:

ð ð
 
rE dV  0 ; rE þ pi dV  0 (10)

with an appropriate choice of the integration measure dV [ 4p r 2dr or g 1/2dr du dw,

where g ; det(gmn) is the matrix determinant of gmn]. The amount of energy con-
dition violation is defined as the extent to which Eq. (10) can become negative.
The value of Eq. (10) provides information about the total amount of energy con-
dition violating matter that must exist for any given FTL spacetime under study.
It was further shown that Eq. (10) can be adjusted to become vanishingly small by
appropriate choice of parameters; therefore, examples can be constructed
whereby the energy condition violation can be made arbitrarily small. But the
violation cannot be made to vanish entirely.
Equation (9) is also a consequence of Eq. (10). Coupling of the finite spaceship
mass with the warp bubble leads to the quite reasonable condition that the net
total energy stored in the warp bubble be less than the total rest-energy of the
spaceship itself, which places a strong constraint upon the dimensionless speed
of the warp bubble [70]:

  1
G Mship Rship D 2
vwarp  2
c Rship R2
   1 (11)
 28
 Mship D 2
 7:41  10
R2

where Mship and Rship are the mass and size of the spaceship, respectively, and R
is the radius of the warp bubble. [One can multiply both sides of Eq. (11) by c to
convert to the MKS unit of speed.] Equation (11) indicates that for any reasonable
values of the engineering parameters inside the brackets, vwarp will be absurdly
low. This result is due to the intrinsic nonlinearity of the general relativistic
field equation. To illustrate this point, we insert the example starship parameters
from Table 3 (R 50 m, D  1/s 10 3 m) into Eq. (11) and assume
Mship 106 kg to find that vwarp  1.72  10 14 (or 5.16  10 6 m/s). Garden
snails can crawl faster than this. And if R and Mship are kept constant, then
D 3.37  1024 m (or 3.57  108 light-years) in order for vwarp  1, which is
an unrealistic requirement on the warp bubble design.
Equation (10) also gives the result that traversable wormholes require arbitra-
rily small amounts of negative energy to build [whereby Eq. (8) serves only as a
494 E. W. DAVIS

gross upper limit] such that within a wormhole spacetime we must have that [68]:
ð
rE ¼ 0; pr dV ! 0 (12)
C

where pr is the outward radial pressure required to hold the wormhole throat open.
We should point out that the Gauss Bonnet theorem (discussed in Sec. III.A)
predicted this result beforehand. Equation (12) is a result that is also due to the
intrinsic nonlinearity of the general relativistic field equation. This nonlinearity
also impacts the coupling of a finite spaceship mass with each side of a worm-
hole’s throat (or the mouth on each side of the throat) leading to a specialized
mass conservation law for the combined system of spacecraft and wormhole:
When finite mass spaceships traverse a wormhole they alter the equivalent
mass of the wormhole mouths they pass through [3]. The entrance mouth absorb-
ing the spacecraft gains equivalent mass while the exit mouth emitting it loses
equivalent mass. (This mass coupling and conservation law takes into
account the possibility that spaceships traversing the wormhole may lose or
gain some momentum and kinetic energy in the process, and it is assumed that
the two mouths are sufficiently far apart that their mutual gravitational interaction
is negligible.) This unusual result suggests, but does not prove, the possibility of a
fundamental limit on the total mass that can traverse a wormhole. The coupled
mass conservation law shows that for a sufficiently large net transfer of mass
the final equivalent mass of the exit mouth becomes negative. This is actually
a beneficial result because ANEC violations are required just to hold the worm-
hole throat open in the first place. If it appears that a runaway reaction might
occur, then it would be prudent for wormhole engineers to simply “turn off”
the wormhole for a brief moment and then “turn it back on” (i.e., “reset” the
wormhole) to restart space transportation operations.
It is on the basis of the foregoing discussion that we can say with confidence
that traversable wormholes appear to be the most viable form of FTL transport.
However, we still do not know how to construct a traversable wormhole because
General Relativity Theory only provides a recipe for the essential geometric and
material ingredients required to open and maintain one, but not the required
assembly instructions. Will we need to pull a traversable wormhole out of the
quantum spacetime foam and enlarge it to macroscopic scale, or will we need
to use extremely large spacetime curvatures to “punch a hole” through space?
Or are there construction techniques yet to be identified? The author is convinced
that the answer can only be found through empirical studies designed to decide
whether the present general relativistic recipe is enough to work with or an
additional construction mechanism is required.
On physical grounds Eq. (10) appears to be the correct negative energy/
energy condition violation quantifier. However, further work is needed to estab-
lish whether Eq. (10) is the correct quantifier to use overall and whether all (aver-
aged) energy condition theorems can be extended to include it.


Similar coupling and conservation law holds for the case of electrically charged matter that
traverse a (charged or uncharged) wormhole.
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 495

On another note, Borde et al. [75] have recast the QI conjecture into a new
program that seeks to study the allowed spatial distributions of negative energy
density in quantum field theory. Their study models free massless scalar fields
in flat two-dimensional Minkowski spacetime. Several explicit examples of
spacetime averaged QI were studied to allow or rule out some particular model
spatial distributions of negative energy. Their analysis showed that some geo-
metric configurations of negative energy can either be ruled out or else con-
strained by the QI restrictions placed upon the allowable spatial distributions
of negative energy. And there were found to be allowable negative energy distri-
butions in which observers would never encounter the accompanying positive
energy distribution so long as the QI restrictions and corresponding energy con-
ditions are violated. The extent to which the results of Borde et al.’s analysis can
be generalized to a four-dimensional curved spacetime (with or without bound-
aries) and interacting fields remains unsolved.

C. Observing Negative Energy in the Lab

Negative energy should be observable in lab experiments. The presence of
naturally occurring negative energy regions in space is predicted to produce a
unique signature corresponding to lensing, chromaticity, and intensity effects
in micro- and macro-lensing events on galactic and extragalactic/cosmological
scales [76 81]. It has been shown that these effects provide a specific signature
that allows for discrimination between ordinary (positive energy) and negative
energy lenses via the spectral analysis of astronomical lensing events. Theoretical
modeling of negative energy lensing effects has led to intense astronomical
searches for naturally occurring traversable wormholes in the universe. Computer
model simulations and comparison of their results with recent satellite obser-
vations of gamma ray bursts (GRBs) have shown that putative negative energy
(i.e., traversable wormhole) lensing events very closely resemble the main fea-
tures of some GRBs. Other research has found that current observational data
suggest that large amounts of naturally occurring “exotic matter” must have
existed sometime between the epoch of galaxy formation and the present in
order to properly quantitatively account for the “age-of-the-oldest-stars-in-the-
galactic halo” problem and the cosmological evolution parameters [71].
When background light rays strike a negative energy lensing region, they are
swept out of the central region thus creating an umbra region of zero intensity. At
the edges of the umbra the rays accumulate and create a rainbow-like caustic with
enhanced light intensity. The lensing of a negative energy region is not analogous
to a diverging lens because in certain circumstances it can produce more light
enhancement than does the lensing of an equivalent positive energy region.
Real background sources in lensing events can have non-uniform brightness dis-
tributions on their surfaces and a dependency of their emission with the observing
frequency. These complications can result in chromaticity effects, that is, in spec-
tral changes induced by differential lensing during the event. The quantification
of such effects is quite lengthy, somewhat model dependent, and with recent
application only to astronomical lensing events. Suffice it to say that future
work is necessary to scale down the predicted lensing parameters and character-
ize their effects for lab experiments in which the negative energy will not be of
496 E. W. DAVIS

astronomical magnitude. Present ultrahigh-speed optics and optical cavities,

lasers, photonic crystal (and related switching) technology, sensitive nano-
sensor technology, and other techniques are very likely capable of detecting
the very small magnitude lensing effects expected in lab experiments.
A nonoptical scheme for detecting negative energy in experiments was
recently reported by Davies and Ottewill [82] who studied the response of
switched particle detectors to static negative energy densities and negative
energy fluxes. Their model is based on a free massless scalar field in flat four-
dimensional Minkowski spacetime and utilized a simple generalization of the
standard monopole detector, which is switched on and off to concentrate the
measurements on periods of isolated negative energy density or negative
energy flux. The detector model includes an explicit switching factor whereby
five different switching functions (based on data windowing theory) are
defined and evaluated. In order to isolate the effects of negative energy a com-
parison is made for the response of a detector switched on and off during a
period of negative energy density (or negative energy flux) and that switched
on and off in the vacuum. The results shed light on the response of matter detec-
tors to pulses of negative energy of finite duration, and they showed that negative
energy should have the effect of enhancing de-excitation (i.e., induce cooling) of
the detector. This is the opposite of our experience with detectors that undergo
excitation when encountering “normal” matter or energy, and isolated detectors
placed in a vacuum naturally cool due to the usual thermodynamic reasons. But
Davies and Ottewill point out that the enhanced cooling effect they discovered
cannot be used to draw a thermodynamic conclusion because their modeling
was restricted to first order in perturbation theory. It is not possible at first
order to determine whether the enhanced cooling effects are due to the small vio-
lation of energy conservation expected in any process in which a general quantum
state collapses to an energy eigenstate, or whether they predict a systematic
reduction in the energy of the detector which has serious thermodynamic impli-
cations. However, Davies and Ottewill point out that their results are model
dependent and they found for their standard monopole detector model that
there is not always a simple relationship between the strength of the negative
energy density/flux and the behavior of the detector. Further research will be
necessary to resolve these issues.

D. Time Machines and Faster-than-Light Causality Paradox Issues

Simply put, any closed timelike curve (CTC), not necessarily a geodesic, in
spacetime defines a time machine. Traversable wormholes imply time machines,
and this discovery spawned numerous follow-on research on time machines
[2,3,42,83 91]. Given a traversable wormhole, it appears to be very easy to
build a time machine, although it will require a Herculean effort to induce a
time shift between the two wormhole mouths (via special relativistic or
general relativistic time dilation techniques). Comprehensive theoretical
studies show that the creation of a time machine might be the generic fate of a
traversable wormhole [3]. One should not become alarmed by this revelation
because it is a well-known and widely accepted fact that classical (and semi-
classical) General Relativity is infested with time machines whereby there are
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 497

numerous spacetime geometry solutions that exhibit time travel and/or have the
properties of time machines [3]. It should be pointed out that various alternative
theories of gravity, and quantum gravity models, are also infested with time
machines, traversable wormholes, and warp drives. Because we are only inter-
ested in exploiting (static or dynamic) traversable wormholes to facilitate FTL
travel between planets or stars, we will not consider their time machine properties
any further. It might be worth revisiting this topic when a traversable wormhole
can be successfully created in the lab (or in space).
Not surprisingly, warp drives also engender the appearance of CTCs [31,92].
Technical fixes have been proposed to mitigate against the formation of CTCs in
and around the warp bubble by performing some minor “surgery(ies)” to the warp
drive geometry. However, because we previously showed that warp drives may
not be technologically practical to implement, we will not consider this issue
any further.
There is ongoing, vibrant debate among some physicists over the effects of
FTL (i.e., superluminal) motion on causality. Because a comprehensive discus-
sion of this topic is beyond the scope of this chapter, we make the following
points in this regard:

1) Einstein’s general relativistic field equation admits solutions that violate

causality (e.g., the Kerr solution, Gödel’s universe, etc.).
2) There are no grounds for microcausality violations in accordance
with Ref. 93.
3) A new definition of chronology and/or causality is in order for FTL
motion.
4) Investigators have found that CTCs/time machines do not affect Gauss’s
theorem, and thus do not affect the derivation of global conservation laws from
differential ones [94]. The standard conservation laws remain globally valid
while retaining a natural quasi-local interpretation for spacetimes possessing
time machines (e.g., asymptotically flat wormhole spacetimes).

Thorne [95] states that it may turn out that causality is violated at the macroscopic
scale. Even if causality is obeyed macroscopically, then quantum gravity might
offer finite probability amplitudes for microscopic spacetime histories possessing
time machines. Li and Gott [96] found a self-consistent vacuum for quantum
fields in Misner space (a simple flat space with CTCs) for which the renormalized
stress-energy tensor is regular (in fact zero) everywhere. This implies that CTCs
could exist at least at the level of semi-classical quantum gravity theory.
As for items 1 and 3 in the list above, we point out that the fact that the
spacetime metric defines a global chronology in General Relativity cannot be
proven. That is because local physics does not determine the topology of space-
time, which could prevent the existence of a global chronology. In fact, Bruneton
[97] emphasizes the fact that relativistic field theories (including Einstein’s
General Relativity) do not involve a pre-existing notion of time and chronology.
The basis for this being that any relativistic field defines its own chronology on
the manifold by means of the Lorentzian metric along which it propagates. Thus
any Lorentzian metric induces a local chronology in the tangent space through
the usual special relativistic notions of absolute Lorentz-invariant future and
498 E. W. DAVIS

past. Because spacetime may be curved and has a nontrivial topology, the exist-
ence of a local chronology in the tangent space (of Special Relativity) does not
imply that a chronology over the whole manifold exists. As a result of this con-
sideration, Bruneton showed that if a signal made up of waves of some field pro-
pagates between two spacetime points, then these points can be time ordered with
the help of the field’s metric, and causality may be preserved even if the field pro-
pagates superluminally. It is standard practice for physicists to use the gravita-
tional metric field to define a chronology on the manifold when establishing
equations of motion for problems of interest. However, it is just one particular
field that exists on the manifold, and so there are no clear reasons why it
should be favored over the other fields in order to define a particular chronology.
Because the global properties of spacetime break causality in General Relativity,
this cannot be related to some intrinsic disease of superluminal motion.
Bruneton [97] further addresses item 3 by showing that causality requires the
existence of a global mixed chronology in spacetime in which classical fields can
propagate superluminally without threatening causality provided that we do not
refer to any preferred chronology in spacetime. To prove this Bruneton defined a
minimal formulation of causality in which no prior chronology is assumed. This
was done by considering a collection I of relativistic fields CI (gravity included)
that propagate along some metrics gI (tensor indices suppressed), all of which
together establishes a finite set of causal cones at each point of spacetime. The
cones defined by gI may be in any relative position with respect to one
another and they may even tip each other over depending on the location in
spacetime. The fields may also interact with each other, which will allow a phys-
ical signal to propagate in spacetime along different metrics. One, therefore, does
not want to prefer one metric with respect to the others because there is no reason
why some sets of local coordinates, or some rods and clocks, should be preferred
because General Relativity is covariant (i.e., coordinates are meaningless). This
framework then leads to the definition of an “extended” notion of future-directed
and timelike curves that requires that at each point on the curve, the tangent
vector is future-directed and timelike with respect to at least one of the gI .
From this construction, all of the extended notions of future, past, initial
conditions (for equations of motion), and other causal properties of spacetime
then follow.
This new causal framework then leads to an immediate resolution of the so-
called “grandfather paradox.” This paradox traditionally assumes that causality
(or determinism) holds although it cannot be defined because there is no available
notion of time ordering along a CTC. Because of this important key fact, the
claim that CTCs have bad causal behavior (in the form of grandfather-type para-
doxes) is unjustified. Bruneton [97] emphasizes that the paradox only arises if one
asks “the wrong question,” and so temporal paradoxes do not exist because they
are not related to causality but only to logic. He proved that there are CTCs that
cannot exist because an event along their curve will influence itself (i.e., the event
is not freely specifiable), while other types of CTCs can exist whereby an event
along their curve is freely specifiable (up to possible gauge invariance con-
straints, if any) [97]. It is important to stress that the latter type of CTCs
already occur in QED. When nontrivial QED vacua involving electromagnetic
or gravitational backgrounds, or finite temperature effects, and boundaries
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 499

(e.g., the Casimir effect) are taken into account, local Lorentz invariance is
broken at the loop level through vacuum polarization. Some of these cases
lead to superluminal light propagation [93,98 103].

E. Conservation of Momentum in Faster-than-Light Spacetimes

Are FTL spacetimes in the form of traversable wormholes or warp drives a
representative form of propellantless propulsion? If so, then what about the con-
servation of momentum in such spacetimes? When studying FTL spacetimes, one
finds a remarkable absence of any published technical papers that address the
topic of momentum conservation. What is not remarkable is that the conservation
laws are already incorporated into General Relativity Theory. Conservation laws
are based on uniform time and isotropic space. Once there is a condition in the
universe where time is not uniform and space is not isotropic, then those conser-
vation laws break down. For example, conservation of energy and momentum in
flat spacetime is rigorously given by the following divergence condition on a
source of matter: @nT mn 0, where T mn (in combination with the coefficient
8pGc 4) is the stress-energy tensor that encodes the density and flux of
the source’s energy and momentum (i.e., 8pGc 4T m0 is the density and
8pGc 4T mi is the flux), and @n denotes the usual flat-spacetime partial derivative
with respect to the spacetime coordinates. However, in curved spacetime this
condition becomes r nT mn 0, where rn is the covariant derivative or spacetime
curvature gradient, and the conservation of energy and momentum is now only
approximate such that the error is attributed to the gravitational field acting on
the matter and itself having some energy and momentum.
It appears that many physicists assert that the conservation laws are already
incorporated into General Relativity and so a discussion of momentum conserva-
tion in their published papers is unnecessary. We disagree because momentum
conservation for any FTL spacetime model must be clearly demonstrated in
order to gain confidence that a given model is correct, plus exploring such
issues is another venue for further discoveries.
The problem that needs to be addressed can be established via the following
scenarios:
1) Warp drives: Imagine there is a spacecraft that is surrounded by a warp
bubble. Place the coupled spacecraft warp bubble system inside an imaginary
box. The specific warp drive spacetime engineering and energy used to
implement the warp bubble is not important. Because space for all warp drive
geometries becomes asymptotically flat at large distances away from the
coupled system inside the box, we can place observers in remote, asymptotically
flat spacetime. These observers see the center-of-mass (COM) of the box
translate at FTL speeds across space.
2) Traversable wormholes: Imagine there is a spacecraft that passes through
the throat of a traversable wormhole. Place the coupled spacecraft traversable
wormhole system inside an imaginary box. The specific traversable wormhole
spacetime engineering and energy used to implement the traversable wormhole
is not important. Because space for all traversable wormhole geometries
becomes asymptotically flat at large distances away from the coupled system
500 E. W. DAVIS

inside the box, we can place observers in remote, asymptotically flat spacetime.
These observers see the spacecraft disappear as it enters one side of the wormhole
throat and reappear as it exits the other side such that the spacecraft appears to
translate at FTL speeds across space. In this case, the COM of this coupled
system moves as well.
3) Given these two scenarios, the problem we are interested in is the follow-
ing: Begin with a spacecraft of mass Mship in flat spacetime (using a flat space-
time coordinate system) at fixed location A before the warp bubble or
traversable wormhole is switched on. Switch on the warp bubble or traversable
wormhole. Assume that the spacecraft moves on a geodesic. Switch off the
warp bubble or traversable wormhole. The spacecraft is now in flat spacetime
at fixed location B in the original flat spacetime.
Note that both scenarios appear to be classifiable as a form of propellantless pro-
pulsion as seen within the reference frame of the spacecraft: in both cases the space-
craft does not require rocket engines (to exert thrust) in order to move FTL across
space because it is the FTL spacetime effect that does all the work. But there is a
slight difference for the case of a traversable wormhole. In that case, the spacecraft
must be self-propelled (under the action of a rocket engine) in order to move
through the throat. But it can move through the throat at any speed  c because a
traversable wormhole, which is a hyperspace shortcut, allows the spacecraft to
undergo apparent FTL motion as seen by remote observers.
In both FTL scenarios remote observers in flat spacetime view the coupled
spacecraft warp bubble/wormhole system inside the imaginary box as a “particle”
that is embedded in flat spacetime. And this “particle” appears to them as
spontaneously moving from one place to another. But such a “particle” cannot
spontaneously move from one place to another without emitting a propellant
along with a corresponding loss of mass from the coupled system. The particle’s
acceleration must be accompanied by a reciprocal momentum (“propellant”). In
other words, propellant here means something the remote observers see as compen-
sating for the momentum of the coupled spacecraft warp bubble/wormhole system
in the observers’ flat spacetime coordinate system.
We conjecture that some of the mass that must have been expended as propellant
is some form of reaction mass, which we provisionally assume to be gravitational
radiation. Now the passengers inside the spacecraft would be reluctant to call this
gravitational radiation “propellant” because they do not feel themselves being
accelerated because they move along a geodesic. By contrast, sufficiently remote
observers can only follow the COM of the spontaneously accelerated ponderable
matter and spacetime is flat elsewhere. The spacecraft’s spontaneous acceleration
is explained to them as being compensated for by the gravitational radiation
emitted in the other direction. That is how they see momentum to be conserved.
This issue needs to be studied in detail with quantitative results stated. This is
an excellent subject for future research.

V. Conclusions
FTL spacetimes hold the promise of allowing mankind to reach the far planets
and stars in our galaxy within a human lifetime to fulfill “the compelling urge of
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 501

man to explore and to discover,” which is guided by “the thrust of curiosity that
leads men to try to go where no one has gone before” [104]. In this assessment of
FTL spacetimes, we delineated a substantial number of crucial technical details
along with their accompanying “make or break” issues having either tentative or
firm results. The tentative results suggest a path for future research that is needed
to put certain crucial technical issues on a firmer footing.
We identified the two primary forms of FTL spacetimes found in General
Relativity Theory that can be created in principle: traversable wormholes and
warp drives. These specialized spacetimes require the introduction of negative
energy densities or fluxes in order to implement their geometries and FTL
effects. Our assessment concludes that we already make small amounts of nega-
tive energy in the lab, but we do not yet know if we can access larger amounts for
extended periods of time over extended spatial distributions for the purpose of
engineering a particular FTL spacetime. We found that there are proposals for
observing negative energy in outer space and in the lab, but further work is
needed to downscale astronomical techniques for use at the lab scale, and we
need to firm up our understanding of how lab detectors will respond to negative
energy in situ.
A detailed energy analysis showed that warp drives are not technologically
practical to implement due to their requirement for extremely large amounts of
negative energy (in order to achieve absurdly low “warp speeds”), while traver-
sable wormholes appear to be the most practical to implement because of their
very minimal negative energy requirement. The appearance of time machines
via warp drives or traversable wormholes, which can induce grandfather-type
causality paradoxes, might not be an issue because investigators have shown
that causality can be preserved in relativistic field theories even when there is
superluminal motion. A resolution to the causality paradoxes is found by using
an extended definition of causality which accounts for the fact that all relativistic
field theories (including Einstein’s General Relativity) do not involve a pre-
existing notion of time and chronology.
The issue of momentum conservation in FTL spacetimes has not been
addressed in the published literature. Here we introduce the nature of the
problem and propose a solution. We suggest that the coupled spacecraft warp
drive/wormhole system must eject some form of reaction mass and subsequently
lose mass in order to conserve momentum as seen by remote observers in asymp-
totically flat spacetime, whereas the passengers on board the spacecraft would not
see the loss of mass because they do not feel themselves being accelerated as they
move FTL across space. This proposed solution has a direct consequence on
whether one can consider FTL spacetimes to be a true form of propellantless pro-
pulsion. This appears to depend upon the frame of reference one is situated.
Limitations on length prevented us from giving a detailed discussion and
analysis of FTL spacetimes within the framework of alternative theories of
gravity or quantum gravity theories. Such theories remain speculative at
present due to a lack of experimental validation, but we encourage the interested
investigator to pursue this line of research to find any comparisons or contrasts
with the results obtained from Einstein’s General Relativity Theory. We hope
that this work will serve as a useful guide for the interested investigator who
wishes to pursue this topic on their own.
502 E. W. DAVIS

Acknowledgments
The author would like to thank the Institute for Advanced Studies at Austin
and H. E. Puthoff for supporting this work. Parts of this work previously origi-
nated under Air Force Research Laboratory (AFMC) contract F04611-99-C-
0025. The author extends thanks and appreciation to F. B. Mead, Jr. (AFRL/
PRSP, Edwards AFB, CA) and members of the Air Staff (Air Force HQ, Penta-
gon) for encouraging our exploration into this topic. He also thanks V. Teofilo
(Lockheed Martin), Lt. Col. P. A. Garretson (Air Force HQ, Pentagon), M. G.
Millis (NASA-Glenn), M. Ibison (Inst. for Advanced Studies at Austin),
S. Little (EarthTech Int’l), R. M. Wald (Univ. of Chicago), T. W. Kephart (Van-
derbilt Univ.), P. C. W. Davies (Arizona State Univ.), L. H. Ford (Tufts Univ.),
and H. G. White (NASA-Johnson) for many useful discussions or their contri-
butions to this chapter.

References
[1] Morris, M. S., and Thorne, K. S., “Wormholes in Spacetime and Their Use for Inter
stellar Travel: A Tool for Teaching General Relativity,” American Journal of
Physics, Vol. 56, 1988, pp. 395 412.
[2] Morris, M. S., Thorne, K. S., and Yurtsever, U., “Wormholes, Time Machines, and the
Weak Energy Conditions,” Physical Review Letters, Vol. 61, 1988, pp. 1446 1449.
[3] Visser, M., Lorentzian Wormholes: From Einstein to Hawking, AIP Press,
New York, 1995.
[4] Davis, E. W., “Teleportation Physics Study,” Air Force Research Laboratory, Final
Report AFRL PR ED TR 2003 0034, Air Force Materiel Command, Edwards
AFB, CA, 2004, pp. 3 11.
[5] Alcubierre, M., “The Warp Drive: Hyper fast Travel Within General Relativity,”
Classical and Quantum Gravity, Vol. 11, 1994, pp. L73 L77.
[6] Ford, L. H., and Roman, T. A., “Negative Energy, Wormholes and Warp Drive,”
Scientific American, Vol. 13, 2003, pp. 84 91.
[7] Hawking, S. W., and Ellis, G. F. R., The Large Scale Structure of Space Time,
Cambridge Univ. Press, Cambridge, U.K., 1973, pp. 88 91, 95 96.
[8] Epstein, H., Glaser, V., and Jaffe, A., “Nonpositivity of the Energy Density in Quan
tized Field Theories,” Nuovo Cimento, Vol. 36, 1965, pp. 1016 1022.
[9] Visser, M., “Wormholes, Baby Universes, and Causality,” Physical Review D,
Vol. 41, 1990, pp. 1116 1124.
[10] Visser, M., “Gravitational Vacuum Polarization. I. Energy Conditions in the
Hartle Hawking Vacuum,” Physical Review D, Vol. 54, 1996, pp. 5103 5115.
[11] Visser, M., “Gravitational Vacuum Polarization. II. Energy Conditions in the Boul
ware Vacuum,” Physical Review D, Vol. 54, 1996, pp. 5116 5122.
[12] Visser, M., “Gravitational Vacuum Polarization. III. Energy Conditions in the
(1 þ 1) Dimensional Schwarzschild Spacetime,” Physical Review D, Vol. 54,
1996, pp. 5123 5128.
[13] Visser, M., “Gravitational Vacuum Polarization. IV. Energy Conditions in the
Unruh Vacuum,” Physical Review D, Vol. 56, 1997, pp. 936 952.
[14] Barcelo, C., and Visser, M., “Twilight for the Energy Conditions?,” International
Journal of Modern Physics D, Vol. 11, 2002, pp. 1553 1560.
[15] Herrmann, F., “Energy Density and Stress: A New Approach to Teaching Electro
magnetism,” American Journal of Physics, Vol. 57, 1989, pp. 707 714.
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 503

[16] Drummond, P. D., and Ficek, Z. (eds.), Quantum Squeezing, Springer Verlag,
Berlin, 2004.
[17] Hochberg, D., and Kephart, T. W., “Lorentzian Wormholes from the Gravitation
ally Squeezed Vacuum,” Physics Letters B, Vol. 268, 1991, pp. 377 383.
[18] DeWitt, B. S., “Quantum Field Theory in Curved Spacetime,” Physics Reports,
Vol. 19C, 1975, pp. 295 357.
[19] DeWitt, B. S., “Quantum Gravity: The New Synthesis,” General Relativity: An
Einstein Centenary Survey, Hawking, S. W., and Israel W. (eds.), Cambridge
Univ. Press, Cambridge, MA, 1979, pp. 680 745.
[20] DeWitt, B. S., “The Casimir Effect in Field Theory,” Physics in the Making, Essays
on Developments in 20th Century Physics, In Honour of H. B. G. Casimir,
Sarlemijn, A., and Sparnaay, J. (eds.), North Holland Elsevier Science Publ.,
New York, 1989, pp. 247 272.
[21] Birrell, N. D., and Davies, P. C. W., Quantum Fields in Curved Space, Cambridge
Univ. Press, Cambridge, U.K., 1984.
[22] Saunders, S., and Brown, H. R. (eds.), The Philosophy of Vacuum, Clarendon Press,
Oxford, 1991.
[23] Milonni, P. W., The Quantum Vacuum: An Introduction to Quantum Electro
dynamics, Academic Press, San Diego, CA, 1994.
[24] Milton, K. A., The Casimir Effect: Physical Manifestations of Zero Point Energy,
World Scientific, NJ, 2001.
[25] Vollick, D. N., “Negative Energy Density States for the Dirac Field in Flat Space
time,” Physical Review D, Vol. 57, 1998, pp. 3484 3488.
[26] Yu, H., and Shu, W., “Quantum States with Negative Energy Density in the Dirac
Field and Quantum Inequalities,” Physics Letters B, Vol. 570, 2003, pp. 123 128.
[27] Olum, K. D., “Superluminal Travel Requires Negative Energies,” Physical Review
Letters, Vol. 81, 1998, pp. 3567 3570.
[28] Gao, S., and Wald, R. M., “Theorems on Gravitational Time Delay and Related
Issues,” Classical and Quantum Gravity, Vol. 17, 2000, pp. 4999 5008.
[29] Ford, L. H., “Quantum Coherence Effects and the Second Law of Thermo
dynamics,” Proceedings of the Royal Society of London, Series A, Mathematical
and Physical Sciences, Vol. 364, 1978, pp. 227 236.
[30] Davies, P. C. W., “Can Moving Mirrors Violate the Second Law of Thermo
dynamics?,” Physics Letters B, Vol. 11, 1982, p. 215.
[31] Everett, A. E., “Warp Drive and Causality,” Physical Review D, Vol. 53, 1996,
pp. 7365 7368.
[32] Perry, M. D., “Crossing the Petawatt Threshold,” Science & Technology
Review, Lawrence Livermore National Laboratory, Livermore, CA, Dec. 1996,
pp. 4 11.
[33] Perry, M. D., “The Amazing Power of the Petawatt,” Science & Technology Review,
Lawrence Livermore National Laboratory, Livermore, CA, March 2000, pp. 4 12.
[34] Mourou, G. A., Barty, C. P. J., and Perry, M. D., “Ultrahigh Intensity Lasers:
Physics of the Extreme on a Tabletop,” Physics Today, Vol. 51, 1998, pp. 22 28.
[35] Slusher, R. E., et al., “Observation of Squeezed States Generated by Four Wave
Mixing in an Optical Cavity,” Physical Review Letters, Vol. 55, 1985, pp. 2409 2412.
[36] Slusher, R. E., and Yurke, B., “Squeezed Light,” Scientific American, Vol. 254,
1986, pp. 50 56.
[37] Robinson, A. L., “Bell Labs Generates Squeezed Light,” Science, Vol. 230, 1985,
pp. 927 929.
504 E. W. DAVIS

[38] Robinson, A. L., “Now Four Laboratories Have Squeezed Light,” Science, Vol.
233, 1986, pp. 280 281.
[39] Saleh, B. E. A., and Teich, M. C., Fundamentals of Photonics, Wiley, New York,
1991, pp. 414 416.
[40] Caves, C. M., “Quantum Mechanical Noise in an Interferometer,” Physical Review
D, Vol. 23, 1981, pp. 1693 1708.
[41] Pfenning, M. J., “Quantum Inequality Restrictions on Negative Energy Densities in
Curved Spacetimes,” Ph.D. Dissertation, Dept. of Physics and Astronomy, Tufts
Univ., Medford, MA, 1998.
[42] Davies, P. C. W., How to Build a Time Machine, Penguin Books, New York, 2001.
[43] Rabeau, J. R., Chin, Y. L., Prawer, S., Jelezko, F., Gaebel, T., and Wrachtrup, J.,
“Fabrication of Single Nickel Nitrogen Defects in Diamond by Chemical Vapor
Deposition,” Cornell Univ. Library arXiv.org e Print Archive. URL: http://
arxiv.org/cond mat/0411245.pdf [cited 11 April 2004].
[44] Rabeau, J. R., Huntington, S. T., Greentree, A. D., and Prawer, S., “Diamond Chemi
cal Vapor Deposition on Optical Fibers for Fluorescence Waveguiding,” Cornell
Univ. Library arXiv.org e Print Archive. URL: http://arxiv.org/cond mat/
0411249.pdf [cited 11 April 2004].
[45] Ries, J., Brezger, B., and Lvovsky, A. I., “Experimental Vacuum Squeezing in
Rubidium Vapor via Self rotation,” Physical Review A, Vol. 68, 2003, 025801.
[46] Casimir, H. B. G., “On the Attraction Between Two Perfectly Conducting Plates,”
Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen, Vol. 51,
1948, pp. 793 796.
[47] Lamoreaux, S. K., “Demonstration of the Casimir Force in the 0.6 to 6 mm Range,”
Physical Review Letters, Vol. 78, 1997, pp. 5 8.
[48] Mohideen, U., “Precision Measurement of the Casimir Force from 0.1 to 0.9 mm,”
Physical Review Letters, Vol. 81, 1998, pp. 4549 4552.
[49] Chen, F., Mohideen, U., Klimchitskaya, G. L., and Mostepaneko, V. M., “Theory
Confronts Experiment in the Casimir Force Measurements: Quantification of
Errors and Precision,” Physical Review A, Vol. 69, 2004, 022117.
[50] Brown, L. S., and Maclay, G. J., “Vacuum Stress Between Conducting Plates: An
Image Solution,” Physical Review, Vol. 184, 1969, pp. 1272 1279.
[51] Walker, W. R., “Negative Energy Fluxes and Moving Mirrors in Curved Space,”
Classical and Quantum Gravity, Vol. 2, 1985, pp. L37 L40.
[52] Moore, G. T., “Quantum Theory of the Electromagnetic Field in a Variable Length
One Dimensional Cavity,” Journal of Mathematical Physics, Vol. 11, 1970,
pp. 2679 2691.
[53] Forward, R. L., “Alternate Propulsion Energy Sources,” Air Force Rocket Propulsion
Laboratory, Air Force Space Tech. Ctr. Space Div., Air Force Systems Command,
Final Report AFRPL TR 83 067, Edwards AFB, CA, 1983, pp. A1 A14.
[54] Hochberg, D., and Visser, M., “Geometric Structure of the Generic Static Traversa
ble Wormhole Throat,” Physical Review D, Vol. 56, 1997, pp. 4745 4755.
[55] Ida, D., and Hayward, S. A., “How Much Negative Energy Does a Wormhole
Need?,” Physics Letters A, Vol. 260, 1999, pp. 175 181.
[56] Krasnikov, S. V., “Hyperfast Travel in General Relativity,” Physical Review D,
Vol. 57, 1998, pp. 4760 4766.
[57] Everett, A. E., and Roman, T. A., “Superluminal Subway: The Krasnikov Tube,”
Physical Review D, Vol. 56, 1997, pp. 2100 2108.
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 505

[58] Broeck, C. V. D., “A ‘Warp Drive’ with More Reasonable Total Energy Require
ments,” Classical and Quantum Gravity, Vol. 16, 1999, pp. 3973 3979.
[59] Natário, J., “Newtonian Limits of Warp Drive Spacetimes,” General Relativity and
Gravitation, Vol. 38, 2006, pp. 475 484.
[60] Lobo, F., and Crawford, P., “Weak Energy Condition Violation and Superluminal
Travel,” Lecture Notes in Physics, Vol. 617, Springer, Berlin, 2003, pp. 277 291.
[61] Pfenning, M. J., and Ford, L. H., “The Unphysical Nature of Warp Drive,” Classical
and Quantum Gravity, Vol. 14, 1997, pp. 1743 1751.
[62] Coule, D. H., “No Warp Drive,” Classical and Quantum Gravity, Vol. 15, 1998,
pp. 2523 2527.
[63] Hiscock, W. A., “Quantum Effects in the Alcubierre Warp Drive Spacetime,”
Classical and Quantum Gravity, Vol. 14, 1997, pp. L183 L188.
[64] White, H. G., and Davis, E. W., “The Alcubierre Warp Drive in Higher Dimensional
Spacetime,” Proceedings of the STAIF 2006: 3rd Symposium on New Frontiers and
Future Concepts, El Genk, M. S. (ed.), AIP Conference Proceedings, Vol. 813, AIP
Press, New York, 2006, pp. 1382 1389.
[65] Obousy, R., “Supersymmetry Breaking Casimir Warp Drive,” Cornell Univ.
Library arXiv.org e Print Archive. URL: http://arxiv.org/gr qc/0512152.pdf
[cited 27 Dec. 2005].
[66] Obousy, R., and Cleaver, G., “Supersymmetry Breaking Casimir Warp Drive,”
Proceedings of the STAIF 2007: 4th Symposium on New Frontiers and Future
Concepts, El Genk, M. S. (ed.), AIP Conference Proceedings, Vol. 880, AIP
Press, New York, 2007, pp. 1163 1169.
[67] Puthoff, H. E., “SETI, the Velocity of Light Limitation, and the Alcubierre Warp
Drive: An Integrating Overview,” Physics Essays, Vol. 9, 1996, pp. 156 158.
[68] Visser, M., Kar, S., and Dadhich, N., “Traversable Wormholes with
Arbitrarily Small Energy Condition Violations,” Physical Review Letters, Vol. 90,
2003, 201102.
[69] Natário, J., “Warp Drive with Zero Expansion,” Classical and Quantum Gravity,
Vol. 19, 2002, pp. 1157 1165.
[70] Lobo, F. S. N., and Visser, M., “Fundamental Limitations on ‘Warp Drive’ Space
times,” Classical and Quantum Gravity, Vol. 21, 2004, pp. 5871 5892.
[71] Visser, M., “Energy Conditions in the Epoch of Galaxy Formation,” Science,
Vol. 276, 1997, pp. 88 90.
[72] Krasnikov, S., “Counter Example to a Quantum Inequality,” Cornell Univ. Library
arXiv.org e Print Archive. URL: http://arxiv.org/gr qc/0409007.pdf [cited 25
May 2005].
[73] Fewster, C. J., “Comments on ‘Counter Example to the Quantum Inequality,’ ”
Cornell Univ. Library arXiv.org e Print Archive. URL: http://arxiv.org/gr qc/
0409043.pdf [cited 10 Sept. 2004].
[74] Kar, S., Dadhich, N., and Visser, M., “Quantifying Energy Condition Violations in
Traversable Wormholes,” Pramana, Vol. 63, 2004, pp. 859 864.
[75] Borde, A., Ford, L. H., and Roman, T. A., “Constraints on Spatial Distributions of
Negative Energy,” Physical Review D, Vol. 65, 2002, 084002.
[76] Cramer, J. G., Foward, R. L., Morris, M., Visser, M., Benford, G., and Landis, G.L.,
“Natural Wormholes as Gravitational Lenses,” Physical Review D, Vol. 51, 1995,
pp. 3117 3120.
506 E. W. DAVIS

[77] Torres, D. F., Anchordoqui, L. A., and Romero, G. E., “Wormholes, Gamma Ray
Bursts and the Amount of Negative Mass in the Universe,” Modern Physics
Letters A, Vol. 13, 1998, pp. 1575 1581.
[78] Torres, D. F., Romero, G. E., and Anchordoqui, L. A., “Might Some Gamma Ray
Bursts be an Observable Signature of Natural Wormholes?,” Physical Review D,
Vol. 58, 1998, 123001.
[79] Anchordoqui, L. A., Romero, G., Torres, D. F., and Andruchow, I., “In Search for
Natural Wormholes,” Modern Physics Letters A, Vol. 14, 1999, pp. 791 797.
[80] Safonova, M., Torres, D. F., and Romero, G. E., “Macrolensing Signatures of
Large Scale Violations of the Weak Energy Condition,” Modern Physics Letters
A, Vol. 16, 2001, pp. 153 162.
[81] Eiroa, E., Romero, G. E., and Torres, D. F., “Chromaticity Effects in Microlensing
by Wormholes,” Modern Physics Letters A, Vol. 16, 2001, pp. 973 983.
[82] Davies, P. C. W., and Ottewill, A. C., “Detection of Negative Energy: 4 Dimen
sional Examples,” Physical Review D, Vol. 65, 2002, 104014.
[83] Nahin, P. J., Time Machines: Time Travel in Physics, Metaphysics, and Science
Fiction, 2nd ed., AIP Press, New York, 1999.
[84] Davies, P. C. W., The Physics of Time Asymmetry, Univ. of California Press,
Berkeley and Los Angeles, 1974.
[85] Davies, P. C. W., About Time: Einstein’s Unfinished Revolution, Simon and
Schuster, New York, 1995.
[86] Gott, J. R., Time Travel in Einstein’s Universe: The Physical Possibilities of Travel
Through Time, Houghton Mifflin Co., Boston, 2001.
[87] Thorne, K. S., Black Holes and Time Warps: Einstein’s Outrageous Legacy,
W. W. Norton & Co., New York, 1994.
[88] Pickover, C., Time: A Traveler’s Guide, Oxford Univ. Press, Oxford, 1999.
[89] Novikov, I. D., The River of Time, Cambridge Univ. Press, Cambridge, U.K., 1998.
[90] Deutsch, D., The Fabric of Reality, Penguin Press, London, 1997.
[91] Al Khalili, J., Black Holes, Wormholes and Time Machines, Inst. of Physics
Publishing, Bristol, UK, 1999.
[92] González Dı́az, P. F., “Warp Drive Space Time,” Physical Review D, Vol. 62,
2000, 044005.
[93] Drummond, I. T., and Hathrell, S. J., “QED Vacuum Polarization in a Background
Gravitational Field and Its Effect on the Velocity of Photons,” Physical Review D,
Vol. 22, 1980, pp. 343 355.
[94] Friedman, J., et al., “Cauchy Problem in Spacetimes with Closed Timelike Curves,”
Physical Review D, Vol. 42, 1990, pp. 1915 1930.
[95] Thorne, K. S., “Closed Timelike Curves,” GRP 340, CalTech, Pasadena, CA, 1983.
[96] Li, L. X., and Gott, J. R., “Self Consistent Vacuum for Misner Space and the
Chronology Protection Conjecture,” Physical Review Letters, Vol. 80, 1998,
pp. 2980 2983.
[97] Bruneton, J. P., “Causality and Superluminal Behavior in Classical Field Theories:
Applications to k Essence Theories and Modified Newtonian Dynamics like
Theories of Gravity,” Physical Review D, Vol. 75, 2007, 085013.
[98] Liberati, S., Sonego, S., and Visser, M., “Faster Than c Signals, Special Relativity,
and Causality,” Annals of Physics, Vol. 298, 2002, pp. 167 185.
[99] Scharnhorst, K., “On Propagation of Light in the Vacuum Between Plates,” Physics
Letters B, Vol. 236, 1990, pp. 354 359.
 FASTER-THAN-LIGHT APPROACHES IN GENERAL RELATIVITY 507

[100] Scharnhorst, K., “The Velocities of Light in Modified QED Vacua,” Annalen der
Physik (Leipzig), 8, Vol. 7, 1998, pp. 700 709.
[101] Latorre, J. I., Pascual, P., and Tarrach, R., “Speed of Light in Non trivial Vacua,”
Nuclear Physics B, Vol. 437, 1995, pp. 60 82.
[102] Dittrich, W., and Gies, H., “Light Propagation in Nontrivial QED Vacua,” Physical
Review D, Vol. 58, 1998, 025004.
[103] Shore, G. M., “Causality and Superluminal Light,” Cornell Univ. Library arXiv.org
e Print Archive. URL: http://arxiv.org/gr qc/0302116.pdf [cited 27 Feb. 2003].
[104] Killian, J. R., Jr., et al., “Introduction to Outer Space: An Explanatory Statement
Prepared by the President’s Science Advisory Committee,” The White House,
U.S. Government Printing Office, Washington, DC, March 1958.
 Chapter 16

Faster-than-Light Implications of Quantum

Entanglement and Nonlocality

John G. Cramer
University of Washington, Seattle, Washington

I. Introduction
HEN two photons emerge from a single quantum event, the state of one
W may be subtly connected to that of the other. The classical view is that,
once separated, such states are fixed according to mechanics and conservation
relations that act at the point of their origin, so that modifying one state later
will not affect the other. In quantum physics, however, as borne out by exper-
iment [1,2], the outcome of a measurement of the state of one of the photons,
even well after the point of joint creation, does affect the state of the other
photon. This connection is referred to as quantum entanglement, a phrase first
coined by Erwin Schrödinger [3]. The questions at issue in such experiments
are: 1) what is the causal connection between states acting in such phenomena,
and 2) can it possibly be used for sending state-to-state signals? The chapter
takes a close look at quantum entanglement, quantum nonlocality, the exper-
iments to explore them, and proposed experiments to test the causal and faster-
than-light (FTL) communication issues evoked by such physics.
Quantum entanglement describes the condition of separated parts of the same
quantum system in which each of the parts can only be described by referencing
the state of other part. This is one of the most counterintuitive aspects of quantum
mechanics, because classically one would expect system parts out of “local”
contact to be completely independent. Thus, entanglement represents a kind of
quantum “connectedness” in which measurements on one isolated part of an
entangled quantum system have nonclassical consequences for the outcome of
measurements performed on the other (possibly very distant) part of the same
system. This quantum connectedness that enforces the measurement correlation
and state-matching in entangled quantum systems is called quantum nonlocality.

Copyright # 2008 by John G. Cramer. Published by the American Institute of Aeronautics and
Astronautics, Inc., with permission.

Professor, Department of Physics.

509
510 J. G. CRAMER

Nonlocality was first highlighted by Albert Einstein and his coworkers Boris
Podolsky and Nathan Rosen in their famous EPR paper [4]. They argued that the
nonlocal connectedness of quantum systems requires a FTL connection that
appears to be in conflict with Special Relativity. Despite this objection,
quantum nonlocality has been demonstrated (see Sec. II) in many quantum
systems [1,2]. In the physics community it is now generally acknowledged to
be implicit in the quantum formalism as applied to entangled systems, although
there are a few “holdouts” who would require an explicit demonstration of non-
local signaling before admitting that nonlocality can be considered a real
quantum phenomenon.
The question that will be investigated here is whether quantum nonlocality is
the private domain of nature, or whether it can be used in experimental situations
to send signals from one observer to another. As we will see, there is presently no
compelling answer to this question. However, it is clear that if such nonlocal
communication were possible, it would have far reaching implications. In par-
ticular, it would represent an enabling technology for superluminal and retrocau-
sal signaling and communications.

II. Quantum Entanglement, Nonlocality, and EPR Experiments

In a quantum mechanical description of elementary entities like photons there is
a duality between the description as a particle and as a wave. Such objects can be
thought of as traveling through space as waves but delivering energy (and other con-
served quantities) at detection as particles. By choosing the kinds of measurements
made on such objects, one can force wave-like or particle-like behavior to be
exhibited in the measurements results. Between the entangled parts of a quantum
system (e.g., the emission of a pair of entangled photons), this wave-like or
particle-like behavior in a measurement on one part of the system may force
similar behavior in the other part. This will be considered further in Sec. V.
The quantum entanglement condition is usually a consequence of some con-
servation law acting within the system, so that the subsystems are connected by
the conservation law. For example, if two photons are emitted back to back in a
joint state that has zero angular momentum and positive parity, then whatever
linear or circular polarization state one photon is measured to have, the other
photon must have an identical polarization if measured in the same basis
(linear or circular). This condition must exist to ensure that the net angular
momentum of the two photon state is zero. In this situation, if the photons are
measured for circular polarization they must both be in states of right circular
polarization or of left circular polarization. Because linear polarization is a coher-
ent superposition of circular polarization states, if measured in the vertical/hori-
zontal linear polarization basis, they must be in the same vertical or horizontal
polarization state; if measured in the 458 left/right linear polarization basis,
they must be in the same 458 left or right polarization state.
Classically, such a polarization correlation condition could, in principle, exist
in some particular polarization basis but not in all of the several polarization
bases simultaneously. This is the underlying physics of the Bell inequalities
[5], which deal with the falloff rate of the correlations as the polarization basis
of one of the measurements is rotated in angle. The Bell inequalities demonstrate
mathematically that the predictions of semiclassical, local, hidden-variable
 QUANTUM ENTANGLEMENT AND NONLOCALITY 511

theories are inconsistent with those of standard quantum mechanics. Tests of such
polarization correlations have been the basis for a number of Bell inequality tests
(or so-called EPR experiments), in which the validity of the predictions of
quantum mechanics and the inadequacies of semi-classical local hidden-variable
theories have been demonstrated to high statistical precision [1,2].
It was later demonstrated [6,7] that the issues surrounding a violation of the
Bell inequalities could be separated into violations of either parameter indepen-
dence (i.e., the outcome probability of a measurement on one of a pair of
entangled particles is independent of the choice of parameters of a measurement
performed on the other member of the entangled pair) and violations of outcome
independence (i.e., the outcome probability of a measurement on one of a pair of
entangled particles is independent of the outcome of a measurement performed
on the other member of the entangled pair). The observation of a violation of
the Bell inequalities indicates a violation of either parameter independence or
outcome independence (or both). Outcome independence is fairly evident in
the quantum formalism, while parameter independence is more elusive and
depends on specific assumptions. We will consider the implications of this
dichotomy in the context of the “no-signal” theorems.
We note that there is some misinformation in the literature concerning the
chronology of successful EPR polarization correlation experiments, and here
we wish to set the record at least somewhat straighter. The experimental measure-
ment that first demonstrated a polarization correlation related to EPR nonlocality
was performed by C. S. Wu and I. Shanknov in 1949 [8], well before Bell’s work
and the subsequent interest in testing Bell’s inequality. Wu and Shanknov
showed that the linear polarizations of back-to-back entangled gamma rays
from electron positron annihilation (an L 0 negative parity state) were
anticorrelated, for example, if one photon was polarized vertically then the
other was polarized horizontally. They did not, however, investigate the falloff
of the correlation with angle, which is the basis of Bell inequality tests, nor
did they depict their results as the action of quantum nonlocality.
It required two more decades for the publication of John Bell’s pivotal work in
1966 [5]. In 1972 Freedman and Clauser [1] performed the first definitive Bell
inequality test by measuring the polarization correlation of entangled photons
from a positive parity L 0 atomic cascade in calcium (Fig. 1). Their results
were in agreement with the predictions of quantum mechanics and were incon-
sistent with local hidden-variable theories by 6.7 standard deviations. A decade
later in 1982, EPR measurements of the Aspect group [2] eliminated several
“loophole” scenarios that might constitute unlikely ways of preserving classical
locality and again demonstrated agreement with quantum mechanics and incon-
sistency with local hidden-variable theories, this time by 46 standard deviations.
A more recent example of an EPR experiment is the work of the Gisin group [9].
They used the fiber-optics cables owned by the Swiss Telephone System to
demonstrate the nonlocal connection between EPR measurements made at
locations in Geneva and Bern, Swiss cities with a line-of-sight separation of
156 km, a direct demonstration, if one was required, that quantum nonlocality
can operate over quite large distances.
Do these EPR experiments constitute a demonstration of the existence of
quantum nonlocality? There is more than one way of interpreting the implications
of the experimental results [1,2]. One can find much discussion in the literature as
512 J. G. CRAMER

Fig. 1 Schematic of the 1972 Freedman Clauser experiment [5].

to whether it is locality or “realism” (the objective observer independent reality

of external events) that has been refuted by these EPR measurements. We here
adopt the view that reality should be a given, and so we regard these experiments
as direct demonstrations of the intrinsic nonlocality of standard quantum
mechanics.
We note that the several polarization bases used in these kinds of polarization
EPR experiments make it straightforward to demonstrate the quantum nonlocal
connections but also make it effectively impossible to use those connections
for observer-to-observer signaling [10] because one would need to deduce
from the arriving photons the polarization basis that was being used in the
distant measurements. This is an aspect of the parameter independence men-
tioned earlier. Although each observer is free to choose the polarization basis
(e.g., circular right/left, linear horizontal/vertical, linear 458 left/right) for the
measurement, the observer is not free to force the photon into a particular state
of that basis, as would be required for nonlocal communication. However,
measuring polarization correlations in a system with angular momentum con-
straints is not the only way of demonstrating the nonlocal connection between
the entangled separated parts of a quantum system. We will discuss EPR exper-
iments that use momentum entanglement and will explore the question of whether
such quantum systems might provide a better vehicle for observer-to-observer
nonlocal communication, because by using momentum entanglement an observer
is able to force the photon into particle-like or wave-like behavior.

III. Quantum No-Signal Theorems

As Einstein asserted with his well-known “spooky actions at a distance”
comment, the enforcement of quantum correlations across space-like and nega-
tive time-like intervals by nonlocality is very counterintuitive. It appears to
imply the twin possibilities of superluminal communication between observers
and of reverse causation through back-in-time communication between
 QUANTUM ENTANGLEMENT AND NONLOCALITY 513

observers. However, over the years, a number of authors [11] have presented
“proofs” that such nonlocal observer-to-observer communication is impossible
within the formalism of standard quantum mechanics. These theorems assert
that, in separated measurements involving entangled quantum systems, the
quantum correlations will be preserved but there will be no effect apparent to
an observer in one sub-system if the character of the measurement is changed
in the other sub-system. Thus, it is asserted, nonlocal signaling is impossible.
As mentioned above, EPR experiments can be viewed [5,6] as demonstrating
violations of outcome independence or parameter independence or both.
Outcome independence cannot be used for nonlocal signaling, while parameter
independence could be used for such signaling. Thus, any test of nonlocal signal-
ing is, in effect, a test of the parameter independence of quantum phenomena and
the no-signal theorems are “proofs” of parameter independence.
Do these no-signal “proofs” really have the status of mathematical theorems?
Perhaps not. Recently it has been pointed out [12] that at least some of these
“proofs” ruling out nonlocal signaling are tautological, assuming that the
measurement process and its associated Hamiltonian are local, thereby building
the final conclusion of no signaling into their starting assumptions. Standard
quantum mechanical Bose Einstein symmetrization has been raised as a
counter-example, shown to be inconsistent with the initial assumptions of
some of these “proofs.” Therefore, at least from some perspectives, the possi-
bility of nonlocal communication in the context of standard quantum mechanics
remains open and appropriate for experimental testing.

IV. Nonlocality vs Special Relativity?

If nonlocal communication is possible, would it be in conflict with Special
Relativity, with its well-known prohibition against FTL signals? The answer to
this question is no.
The prohibition of signals with superluminal speeds by Einstein’s Theory of
Special Relativity is related to the fact that the definite simultaneity of two sep-
arated spacetime points is not Lorentz invariant. Because some hypothetical
superluminal signal could be used to establish a fixed simultaneity relation
between two such points, this would imply a preferred inertial frame and
would be inconsistent with Lorentz invariance and Special Relativity. In other
words, it would be inconsistent with the even-handed treatment of all inertial
reference frames in Special Relativity.
However, if a nonlocal signal could be transmitted through measurements at
separated locations performed on two entangled photons, the signal would be
“sent” at the time of the arrival of the photon in one location and “received” at
the time of arrival of the other photon. By varying path lengths to the two
locations, these events could be made to occur in any order and time separation
in any reference frame. Therefore, nonlocal signals (even superluminal and retro-
causal ones) could not be used to establish a fixed simultaneity relation between
two separated spacetime points, because the sending and receiving of such
signals do not have fixed time relations. The transmission and arrival instants
of a nonlocal signal cannot be used for synchronization because the transmission
and reception instants are path and delay-dependent variables.
514 J. G. CRAMER

To put it another way, the nonlocal connections of entangled photons lie along
segmented, light-like world lines that transform properly under Lorentz trans-
formations. Therefore, there is no conflict between nonlocal signaling and the
Lorentz invariance of special relativity. On the other hand, the principle of caus-
ality (i.e., cause must precede effect in all reference frames) appears very likely to
be violated (or at least violate-able) if nonlocal signaling is possible.
Is it possible that the universe does have some preferred reference frame,
perhaps that laid down by the cosmic microwave background or implied by
Mach’s principle? Perhaps, but if such a preferred frame existed, its existence
could not be established by nonlocal communication.

V. Momentum Domain Entanglement and EPR Experiments

Einstein’s original objection [4] that quantum mechanics appeared to be non-
local was made with arguments based on a gedanker experiment in the momen-
tum domain. However, almost all of the modern ERP experiments testing the Bell
inequality and demonstrating quantum nonlocality have been performed in the
polarization (i.e., angular momentum) domain, usually with linearly polarized
photons. Interestingly, it appears that if nonlocal quantum communication is
possible at all, it may be more easily achieved in the momentum domain of
Einstein’s original focus.
The optical process of spontaneous parametric down-conversion [11] turns out
to be a very useful way of generating photon pairs entangled in either the polar-
ization or the momentum domains. In this process a photon from a “pump laser”
interacts with a nonlinear crystal and is transformed into two photons with ener-
gies and vector momenta that add up to those of the original pump photon.
Depending on the type of down-conversion process, there are well-defined polar-
ization correlations between the entangled photons. The down-converted photons
may also be easily prepared in momentum-entangled states, because within the
nonlinear medium the vector momenta of the down-converted pair of photons
must add to give that of the pump photon.
The first measurement using momentum-entangled down-conversion photons
that might be related to nonlocal communication is the “ghost interference”
experiment reported in 1995 by Shih group [13]. It is shown schematically in
Fig. 2. They used degenerate collinear type-II down-conversion of 351 nm UV
pump radiation from an argon-ion laser passed through a 3.0 mm long BBO
(b-BaB2O4) crystal that had been cut with the optic axis at a phase-matching
angle of 42.28 to the pump beam to produce a pair of collinear momentum-
entangled 702-nm photons with opposite polarizations. The entangled photons
emerge from the crystal very nearly parallel with the pump beam. The pump
beam is then split off from the pair using refraction in a quartz prism (UV
prism), and the entangled photons are separated with a polarization-selecting
beam splitter (BS) that reflects the “extraordinary” vertically polarized photon
(e), and transmits the “ordinary” horizontally polarized photon (o). Both
photons are passed through 702 + 10 nm wavelength selective filters (f1,2) and
then detected (D1,2).
The experimenters demonstrated that passing the vertically polarized photon
(e) through a double or a single slit system before detection at D1 produced a
 QUANTUM ENTANGLEMENT AND NONLOCALITY 515

Fig. 2 Schematic of the 1995 ghost interference experiment of the Shih group.

“comb” interference distribution or a “bump” diffraction distribution, respect-

ively in the position X2 of the horizontally polarized photon (o) detected at D2,
when the pair of photons is examined in coincidence. In other words, the position
distribution of the straight-through photon shows patterns characteristic of the
single or double slit system through which its twin entangled photon passed.
Figure 3 shows the observed position distributions for the two cases.
From the viewpoint of nonlocal communication, we note that modifying the
slit system before D1 through which the reflected photon passes, which can be
thought of as the action of a “sending” observer, nonlocally causes an observable
change in the X2-position distribution of the undeflected photon, as detected by a
“receiving” observer at D2. This is a nonclassical effect that demonstrates the
nonlocal connection between the entangled pair that might form the basis for
transmission of a nonlocal signal between the two observers. However, the
ghost interference experiment does not, in the form reported, demonstrate non-
local communication because of its use of a classical communication link in
imposing the coincidence requirement between the detected photons.
In their paper, the authors comment that with the two-slit system in place, in
the absence of coincidences there is no observable two-slit interference pattern
distributions at either D1 or D2. They attribute this lack of an interference
“signal” to the horizontal variation in the creation position of the down-converted
photons. The variation is enough to cause the e photons to arrive at the two slits
with relative path lengths that may differ by more than a wavelength, thereby
randomly shifting and washing out any interference pattern.
The authors point out that there is a simple way of thinking about momentum-
entanglement measurements involving entangled photons. It can be shown from
Snell’s Law and conservation of momentum in the crystal that if one photon has a
516 J. G. CRAMER

Fig. 3 “Ghost interference” position distributions at X2.

small momentum that causes it to be slightly deflected to the right of the pump
beam by an angle u, then the twin entangled photon will be deflected to the
left by the same angle u, a situation reminiscent of reflection from a mirror.
This allows the experiment to be “unfolded” by replacing the effective reflection
by a straight-through path, as shown in Fig. 4. The point of the unfolding is that
the entangled photons behave exactly as they would if the direction of the
deflected photon was reversed, so that it originated at the detection point D1,
passed through one or two slits at C, D, and produced a one- or two-slit interfer-
ence pattern at D2.
Why is the coincidence needed? First, it should be clear from Fig. 2 that D2
detects not only the entangled twins of the photons that pass through the slit
openings, but also the entangled twins of the much larger number of photons
that are stopped by the opaque parts of the slits. Therefore, without coincidences
no interference pattern could be observed. Moreover, one can see from Fig. 4b
that detector D1 behind the slits receives light in a very localized region, and if
it were moved vertically in the diagram, the interference pattern at D2 would
 QUANTUM ENTANGLEMENT AND NONLOCALITY 517

Fig. 4 “Unfolding” the ghost interference experiment.

be shifted, with maxima becoming minima, and vice versa. Without coincidences
requiring a particular location for the detection at D1, the D2 distribution would
have to average over all possible D1 positions, washing out the two-slit interfer-
ence pattern. Therefore, because of the geometry used, the ghost interference
experiment required a coincidence to observe a two-slit interference pattern
like the one shown in Fig. 3a.
Another momentum-entangled EPR experiment was performed in 1998 at the
University of Innsbruck by Birgit Dopfer [14] and shown schematically in Fig. 5.
In the Dopfer experiment, moving a detector in one arm nonlocally changes the
observed interference pattern in the other arm. Dopfer used 351-nm UV pump
radiation from an argon-ion laser with type I down-conversion in a nonlinear
LiIO3 crystal cut with the optic axis at 908 to the pump beam. This produced a
pair of 702-nm momentum-entangled photons that emerged from the crystal at
angles of 28.28 to the right and left of the pump axis, as shown in Fig. 5.

Fig. 5 Schematic of the 1998 Dopfer experiment [5].

518 J. G. CRAMER

The lower entangled photon passed through a pair of slits and into a detector,
while the upper photon passed through a lens that could image the two slits to
perform a “which-way” measurement if detector D2 was placed two focal
lengths behind the lens (2f ). However, if detector D2 was placed in a position
one focal length behind the lens (f ), the slits were not imaged and light on the
reflected line passing through either slit could reach the detector at the same
points. This produced a result similar to that of the ghost interference experiment.
A structured two-slit interference pattern could be switched on and off by moving
a detector in the other arm of the experiment between the f and 2f positions.
Again, from the viewpoint of nonlocal communication, we note that moving
detector D2, which can be thought of as the action of a “sending” observer, non-
locally causes an observable change in the position distribution of the second
photon, as detected at “receiver” position D1. However, the Dopfer experiment
does not demonstrate nonlocal communication because, like the ghost interfer-
ence experiment, it requires a classical communication link to impose the
coincidence requirement between the detected photons because of the geometry
of the experiment.
Examination of these two experiments raises a very interesting question: Can
the coincidence requirement be removed? The answer to this question is not
clear. In principle, the two entangled photons are connected by nonlocality
whether or not they are detected in coincidence. The coincidence should there-
fore be removable. However, in both experiments the authors report that no
two-slit interference distribution is observed when the coincidence requirement
is relaxed. This may be explained by the action of coherence-entanglement com-
plementarity, as discussed in the next section.

VI. Coherence-Entanglement Complementarity

As discussed, the finite extent of the source is expected to limit the possibility
of observing a 2-slit interference pattern, which would be the “signal” if nonlocal
communication was possible. Figure 6 shows schematically this “thick source”
effect. The source volume on the left is the region of the nonlinear crystal that
is illuminated by the UV pump-laser beam directed along the u axis. The
source volume is a cylinder a few mm thick and a mm or so in radius with a
center point C. The source cylinder is assumed to be tilted at an angle u with
respect to the horizontal z axis on which the slit system and detector plane are
symmetrically centered. We note that u 08 in the ghost interference experiment
and u 28.28 in the Dopfer experiment. A horizontal distance Lxs away from the
source is a two-slit system, a pair of apertures a with center-to-center separation
d. Light passing through the slit system travels a horizontal distance Lsd and is
detected at detector plane at position x1.
If the point of photon production is off the z-axis, there will be a path length
difference between waves as they arrive at the two slits. In Fig. 6, waves from
points A and B could have path length differences greater that half a wavelength
and phase differences of greater than 1808. Roughly speaking, this shifts the inter-
ference pattern relative to waves created at central point C, so that maxima
become minima and vice versa. The net effect of averaging over all points in
the source volume therefore would be to wash out the two-slit interference
 QUANTUM ENTANGLEMENT AND NONLOCALITY 519

x
v
A z X
1
u
θ a
C
d
Lxs Lsd
a
Slit Detector
B Source P System Plane
Volume

Fig. 6 Thick-source effect (not to scale). Waves arriving at the two slits from points
A and B at the extrema of the source volume may have significant path length and
phase differences, while waves from the central point C are in phase at the slits.

pattern. That two-slit interference pattern must be observed unambiguously,

because it is the “signal” that would be used in any nonlocal communication.
The constancy of the relative phase at the two slits for photons arriving from
various parts of the source is called “coherence.” It should be clear that a point-
like source has perfect coherence and a source with a large solid angle as viewed
from the slits will have reduced coherence. The path length difference at the slits
is inversely proportional to Lxs to a good approximation, so making Lxs large (i.e.,
placing the slits some distance from the source volume), can reduce the path
length differences to a value that increases coherence and allows observation
of a sharp two-slit interference pattern signal at detector D1. Alternatively,
placing a thin double-concave diverging lens at point P of Fig. 6 can have the
same effect by causing the shorter path lengths to pass through a greater thickness
of lens glass. Such a lens would also demagnify the source, producing the equiv-
alent of a longer path length and smaller source solid angle.
However, increasing source coherence has another consequence. The momen-
tum entanglement of photons from the source arises from momentum conserva-
tion. Restricting the solid angle of the source as viewed from the slits means that
fewer photon pairs can be entangled and still satisfy the geometrical constraints
of the experimental configuration. The Saleh group at Boston University has
studied the complementarity between source coherence and two photon entangle-
ment [15] and has shown that there is a complementary relation between these
two characteristics. As the source-slit distance Lxs is increased, there is a
smooth transition from one-slit to two-slit interference patterns and a smooth
transition from a highly entangled source to a highly coherent source.
Nonlocal communication using momentum entanglement requires source
coherence. Source coherence is needed in order to observe the “signal” of a
two-slit interference pattern and two photon entanglement so that a measurement
of one of the photons “connects” with the interference pattern produced by
the other photon. Where there is coherence without entanglement or entangle-
ment without coherence, nonlocal communication with momentum-entangled
520 J. G. CRAMER

photons is not possible. The open question is whether there is a “sweet spot” in
the experimental design that embraces both partial coherence and partial
entanglement and that permits the transmission of nonlocal signals. This is
an unresolved issue that requires further theoretical consideration and
experimental testing.

VII. Nonlocal Communication vs Signaling

As we have pointed out above, the possibility of nonlocal communication is an
unresolved issue. It is perhaps likely that the coherence/entanglement trade-off is
nature’s way of preventing nonlocal signaling, but that has not been demon-
strated. In this section, we will assume than nonlocal signaling is possible and
will examine its implications. As will be seen, they are so far-reaching that
they could be taken as a syllogism that nature would not allow such things and
therefore nonlocal signaling must be impossible.
Figure 7 shows a variation of the ghost interference experiment [13] in which
the slit imaging procedure of the Dopfer experiment [14] is used to ensure that
entangled photon pairs passing through slits reach both detectors, and those inter-
cepted by the opaque regions of the slits reach neither detector. In particular, a
lens of focal length f is placed in the path after the BBO crystal, and before
the polarization splitter, so that both entangled photons pass through the lens.
A pair of slits S1 is placed at a path distance f beyond the lens in the path of
the “o” photons, which are linearly polarized horizontally (HLP) and are trans-
mitted by the splitter. As Dopfer has shown, because of momentum entangle-
ment, an image of slit system S1 will be formed by the “e” photons linearly
polarized vertically (VLP) at a path length f beyond the lens on the deflected
path at position S2, where a pair of “cleanup” slits are located that pass only
those photons whose entangled twins passed through S1. We note that because
of the optical geometry, this imaging occurs even for waves that pass through
the image points and ultimately interfere.
At the image position of each slit at S2 we place an optical fiber, as shown. The
fibers conduct the light to an optical switch, at which the light is either sent
directly to two avalanche photodiode detectors D (providing which-way infor-
mation about which S1 slit the photon entered), or alternatively is routed to an
optical combiner C and then detected, so that waves passing through both slits
can contribute constructively to the detection event. We note that this fiber
switching system is the fiber-optics equivalent of a Mach Zehnder interfero-
meter [16] in which one can activate and deactivate the last half-silvered
mirror by switching, so that which-way information can be switched on and off.
A quantum sensitive cooled CCD camera is substituted for detector D1 of the
ghost interference experiment [14] and is set to measure distributions like those
shown in Fig. 3. In the arrangement in Fig. 7, switching the optical fiber routing
can be considered to be an act of transmitting a binary 0 or 1 signal. If the switch
is in the position leading to the outer detectors, then which-way information is
available and the pattern detected by the camera should be a single-slit diffraction
pattern labeled “1” in Fig. 7. If the switch is in the position leading to the com-
biner and middle detector, waves from both slits contribute to the detection, no
 QUANTUM ENTANGLEMENT AND NONLOCALITY 521

Fig. 7 Slit-imaging coincidence-free version of the ghost interference experiment to

demonstrate nonlocal communication.

which-way information is available, and the pattern detected by the camera

should be the two-slit interference pattern labeled “0” in Fig. 7.
If the pattern observed by the camera can indeed be changed by switching the
optical fiber routing, then this would constitute a direct demonstration on nonlo-
cal communication. Such an observation would falsify the no-signal theorems
mentioned above, which require that in a non-coincidence scenario, no action
on one entangled photon can produce a “signal-capable” observable result at
the detection of the other entangled photon of the pair.
We should emphasize that demonstrating nonlocal communication with
momentum-entangled photons, as described above, is not the same as actually
sending a signal. It should be clear that no real signal can be communicated
with a single photon pair. It is only when multiple photons are detected that
the underlying distribution function can become apparent. We estimate that if
the distribution functions to be distinguished are a “pure” two-slit interference
pattern modulated by a diffraction envelope and a “pure” one-slit diffraction
pattern, then about 10 photon detections would be required for a 3s decision
between these two possibilities.
However, as mentioned, it is likely that if nonlocal communication is possible
at all, it would have to be accomplished in a situation where some compromise
522 J. G. CRAMER

between entanglement and coherence has been achieved, and such a compromise
would inevitably cause the two patterns to be distinguished to be more similar
and more difficult to separate. Therefore, the 10 photon detections cited above
must be taken as a rather optimistic lower limit, and it is likely that a significantly
larger number of detections (perhaps 100) would be required. The time
required to send a single bit of information would then be the product of the
photon detection rate in the two arms of the experiment times the number of
photons that must be detected to receive the signal. In principle, such a trans-
mission rate might be improved by pulsing the pump laser, so that “clusters”
of entangled photons would be received with each such pulse.

VIII. Superluminal and Retrocausal Nonlocal Communication

As mentioned, we will assume, for the sake of discussion, that nonlocal signal-
ing is possible and will consider its implications for the speed of transmission of
signals. For definiteness, schemes for doing this are based on the slit-imaging,
coincidence-free version of the ghost interference experiment already described
and shown in Fig. 7. In that system, the instant at which a nonlocal signal is sent is
the arrival of the VLP photon at the fiber-optics system on the left, and the instant
at which the signal is received is the arrival of the HLP photon at the camera at
the bottom of the diagram. Assuming the workability of this scheme, both the
instants of sending and receiving can be delayed, in principle, by the introduction
of delay paths (e.g., runs of fiberoptic cables) in the system. In particular, the send
and receive instants, occurring at widely separated locations connected by runs of
fiber-optics, could be tuned to occur simultaneously in any desired reference
frame. This would constitute a direct demonstration of superluminal signaling.
Additionally, the “send” instant could be made to occur well after the
“receive” instant in the system, constituting a direct demonstration of retrocausal
signaling. This is shown in Fig. 8. Here the cleanup two-slit system S2 becomes
the entrance for two 10-km long runs of fiber-optics that are carefully matched to
have identical exit phases at S3, the end of the fiber runs where the light enters the
optical switching arrangement described above. If the index of refraction of the
fiber is 1.5, light transiting the 10-km path requires about 50 ms. In the presence
of detection noise or the degradation of pattern visibility because of compromises
between entanglement and coherence, considerably more photon detection
events, say 100, might be required.
Assume that the fiber coils of Fig. 8 remain rolled up and are stored in a corner
of the laboratory and that the source can be made strong enough so that the
average rate at which the entangled photon pairs are detected is 10 MHz
(which would correspond to the efficient detection of about 3.0 nW of 702 nm
photons). Then the “sending” detector system and the “receiving” camera are
in the same room and separated by a distance of 1 m or less. If the switch is
set on the 0 or 1 position, the “message” that it is in that position begins to
arrive at the camera 50 ms before the switch position is moved. If 100 photon
counts constitute a signal, then the message could be received 40 ms before it
was sent. This would be a direct demonstration of retrocausal signaling using
nonlocal communication, and would constitute a direct violation of the principle
of causality.
 QUANTUM ENTANGLEMENT AND NONLOCALITY 523

BBO

VLP
10 km Double Convex Lens 351 n
Focal Length = f
S3 S2 IR Pass

Argon Ion Laser

D S Filter
IR Pass VLP Polarizing
Filter 702 nm Splitter
D C Switch

D S
Destination Image Slits IR Pass f
Apertures 10 km Path to Lens = f Filter

HLP
702 nm

Object Slits
Path to Lens = f S1

Camera

Fig. 8 Slit-imaging coincidence-free version of the ghost interference experiment to

demonstrate superluminal and retrocausal signaling.

IX. Paradoxes and Nonlocal Communication

The setup described above, with its retrocausal communication link, raises
some time-communication paradoxes. First, let us consider the issue of
“bilking.” Suppose that we construct one million linked systems of the type
shown in Fig. 8. Then the transmitted message would be received 40 sec
before it was sent. Now suppose that a tricky observer receives a message
from himself 40 sec in the future, and then decides not to send it. This produces
an inconsistent timelike loop, which has come to be known as a “bilking
paradox.” Could this happen? If not, what would prevent it?
There are discussions of such bilking paradoxes in the physics literature by
Wheeler and Feynman [17] who were considering the retrocausal aspects of the
advanced waves of absorber theory, and by Thorne and colleagues [18]
who were considering the paradoxes that might arise from timelike wormholes.
The general consensus of this work is that nature will forbid it and require a con-
sistent set of conditions. Thorne and co-workers showed that “nearby” to any
inconsistent paradoxical situation involving timelike wormholes there is always
a self-consistent situation that does not involve a paradox. As Sherlock Holmes
said, “When the impossible is eliminated, whatever remains, however improbable,
must be the truth.” These speculations assert that equipment failure producing a
consistent sequence of events is more likely than producing an inconsistency
between the send and receive events. The implications of this are that bilking
itself is impossible, but very improbable events could be produced in avoiding it.
The other issue raised by retrocausal signaling might be called the “immacu-
late conception” paradox. Suppose that you are using the setup described above,
and you receive from yourself in the future the manuscript of a best-selling novel
524 J. G. CRAMER

with your name listed as the author. You sell it to a publisher and become rich and
famous. And when the time subsequently comes for transmission, you duly send
the manuscript back to yourself, thereby closing the timelike loop and producing
a completely consistent set of events. But the question is, just who wrote the
novel? Clearly, you did not; you merely passed it along to yourself. Yet highly
structured information (the novel) has been created out of nothing. And in this
case, nature should not object, because there was no bilking and you produced
no inconsistent timelike loops.
It is not known how to resolve either of these paradoxes. All that can be said is
the following:
1) If nonlocal signaling is impossible, then it needs no resolution, but better,
more “air-tight” proofs of the impossibility of nonlocal signaling would be needed.
2) If nonlocal signaling is possible and can be used to form timelike loops,
then paradoxes become important subjects for further experimental testing,
study, and theoretical treatment.

X. Nonlinear Quantum Mechanics and Nonlocal Communication

So far, we have focused on the possibility that nonlocal communication may
be possible within the framework of standard quantum mechanics. However,
even if nonlocal communication is impossible in standard quantum mechanics,
there could also be another path to nonlocal communication.
The no-signal theorems described in Section III are based on the formalism
of standard quantum mechanics. Such “proofs” become invalid if quantum
mechanics is allowed to be slightly “nonlinear,” a technical term meaning
that when quantum waves are superimposed they may generate a small cross-
term not present in the standard formalism. Steven Weinberg, Nobel laureate
for his theoretical work in unifying the electromagnetic and weak interactions,
investigated a theory that introduces small nonlinear corrections to standard
quantum mechanics [19]. The onset of nonlinear behavior is seen in other
areas of physics (e.g., laser light in certain media) and, he suggested, might
also be present but unnoticed in quantum mechanics itself. Weinberg’s
nonlinear quantum mechanics subtly alters certain properties of the standard
theory, producing new physical effects that can be detected through precise
measurements.
Two years after Weinberg’s nonlinear quantum mechanics theory was pub-
lished, Joseph Polchinski published a paper demonstrating that Weinberg’s non-
linear corrections upset the balance in quantum mechanics that prevents
superluminal communication using EPR experiments [20]. Through the new non-
linear effects, separated measurements on the same quantum system begin to
“talk” to each other and FTL and/or backward-in-time signaling becomes poss-
ible. Polchinski describes such an arrangement as an “EPR telephone.”
The work by both Weinberg and Polchinski had implications that are
devastating for the Copenhagen representation of the wave function as “observer
knowledge.” Polchinski has shown that a tiny nonlinear modification trans-
forms the “hidden” nonlocality of the standard quantum mechanics formalism
into a manifest property that can be used for nonlocal observer-to-observer
 QUANTUM ENTANGLEMENT AND NONLOCALITY 525

communication. This is completely inconsistent with the Copenhagen “know-

ledge” interpretation.
Finally there is also the possibility of a nonlinear form of quantum mechanics
that reduces to linear quantum mechanics in the limiting case of weak gravity or
no gravity. Using the wave picture, it is possible to formulate a Laplace Beltrami
wave equation for curved space. The Laplace Beltrami operator on the left-hand
side contains information about the spacetime geometry (the metric tensor) and
operates on the wave function. On the right-hand side, is the same term found
in the flat space Klein Gordon wave equation. In flat space (no gravity), this
“curved-space” wave equation reduces to the Klein Gordon wave equation. In
addition, some theoretical physicists have studied Dirac-like wave operators on
curved space to handle spin.
There are many possibilities for this type of nonlinear curved space quantum
mechanics that can be explored. The Laplace Beltrami operator can be
expanded for any type of space-time geometry. Schwartzschild, Yilmaz,
Friedmann Walker, or even Morris Thorne traversable wormhole geometries
can all be explored. Nonlinear differential equations will result. But do physically
meaningful solutions exist? Most likely, they will not obey the principle of
superposition. However, there could be some intriguing possibilities. Analogous
to tunneling through a potential barrier, elementary particles might be able to
tunnel through spacetime geometries at superluminal speed.
But in the everyday world of weak gravity and fairly flat space, is quantum
mechanics indeed nonlinear? Atomic physics experiments have been used by a
number of experimental groups to test Weinberg’s nonlinear theory. So far,
these tests have all been negative, indicating that any nonlinearities in the
quantum formalism must be extremely small, if they exist at all. These negative
results are not surprising, however, because the atomic transitions used involve
only a few electron-volts of energy. If quantum mechanics does have nonlinear
properties, they would be expected to depend on energy and to appear only at
a very high energy scale and particularly at the highest energy densities. Nonlocal
communication tests should be made, if possible, with the highest energy particle
accelerators.

XI. Issues and Summary

A summary of key unresolved issues follows:
1) Can the intrinsic nonlocality of quantum mechanics be used for observer-
to-observer communication?
2) If nonlocal communication is possible, can it be used to send messages
faster than the speed of light (i.e., across spacelike intervals)?
3) If nonlocal communication is possible, can it be used to send messages
backward in time (i.e., across negative timelike intervals)?
4) If nonlocal back-in-time communication is possible, how can the para-
doxes that result from this capability be resolved?
5) Is quantum mechanics perfectly linear, or are there small nonlinearities,
perhaps consequences of quantum gravity, that could be exploited for faster-
than-light or backward-in-time communication?
526 J. G. CRAMER

XII. Conclusions
Ultimately, the question of whether nonlocal communication is possible is an
experimental one. The issue should be resolvable by testing for nonlocal com-
munication and observing what experimental limits appear. In particular, are
the limits of coherence/entanglement complementarity so severe that signaling
is precluded? Currently there is at least one experiment in progress that aims at
producing a coincidence free version of the ghost interference experiment. We
eagerly await the outcome of such tests.

Appendix: Glossary and Description of Key Concepts

Basis: In quantum mechanics, a choice of an observable quantity that may be
complimentary to another variable, so that both cannot be measured at the same
time. An example is the choice of measuring position, which prevents the simul-
taneous measurement of momentum. In EPR experiments, one must choose a
polarization basis; for example, linear polarization that may be either vertical
or horizontal. Because both circular polarization and 458 left/right polarization
are linear superpositions of vertical/horizontal polarization, they may not
be measured simultaneously. In quantum mechanics, the measurement causes
the wave function to collapse to a particular basis value, excluding other
possible values.
Bell’s Theorem: A mathematical proof by John S. Bell [5] demonstrating that in
a polarization-based EPR experiment, the falloff of correlations as the basis angle
of a polarization measurement is changed is qualitatively different, as predicted
by local hidden-variable theories and by standard quantum mechanics. In particu-
lar, local hidden-variable theories predict a linear falloff while quantum mech-
anics predicts a quadratic falloff. This difference in predictions is represented
as an inequality in measurement intensity ratios that all local hidden variable the-
ories must satisfy while quantum mechanics does not. Tests of these predictions
have been found to agree with quantum mechanics and to falsify local hidden-
variable theories.
Bilking Paradox: A type of back-in-time communication paradox in which an
inconsistent causal loop is created. A well-known example is the grandmother
paradox, a time travel scenario from science fiction in which a time traveler
travels to the past and kills his grandmother before she had children. The question
then arises, how he could have been born if his grandmother had no children?
Several works in the physics literature [17,18] have concluded that such trans-
temporal bilking is impossible, that nature will not permit inconsistent timelike
loops, and that it is more likely that some apparatus will fail than that a “bilk”
of nature could be achieved.
Causality: The observation, which is regarded as a law of physics, that a cause
must precede its effects as viewed in any and all reference frames. Sometimes
referred to as “cause and effect” or “the law of cause and effect.”
Correlations: The mathematical connection between two variables or two
measured quantities. As an example, in an EPR measurement, the basis polariz-
ation of one photon is selected, the basis polarization of the twin entangled
photon is varied, and the coincidence counting rate versus varied angle is
measured to establish the correlation between the two polarizations.
 QUANTUM ENTANGLEMENT AND NONLOCALITY 527

Coherence: Describes whether two waves (e.g., those arriving at a pair of slits or
at a detector) have a definite phase relation (in which case they are completely
coherent) or a random phase relation (in which case they are completely incoher-
ent) or something in between.
Coherence-Entanglement Complementarity: The theoretical expectation and
experimental observation [15] that perfect coherence and perfect entanglement
cannot be achieved for an entangled pair of photons at the same time.
Coherent Superposition: The formation of a quantum mechanical state (e.g.,
right circular polarization) by adding components of other states (e.g., left and
right polarization) with a definite complex phase between the added states.
Collapse: A quantum mechanical wave function is said to collapse to a particular
basis value when a measurement is made in that basis. For example, if a photon is
emitted isotropically (with equal probability in all directions), its wave function
is distributed uniformly over a sphere with a radius that grows at the speed of
light until it is detected. On detection, the photon’s wave function is localized
at the detection point and disappears everywhere else.
Entangled: The separated parts of the same quantum system are said to
be entangled when each of the parts can only be described by referencing
the state of other part. This is one of the most counterintuitive aspects of
quantum mechanics, because classically system parts out of “local” contact
should be completely independent. Thus, entanglement represents a kind of
quantum “connectedness” in which measurements on one isolated part of an
entangled quantum system have nonclassical consequences for the outcome
of measurements performed on the other (possibly very distant) part of the
same system.
EPR Experiment: A class of experiments with entangled particles,
usually photons, that demonstrate quantum nonlocality. A gedanken experiment
of this kind was first suggested in the famous 1935 paper by Einstein,
Podolsky, and Rosen [4], in which a set of criticisms of quantum mechanics
were presented.
Hidden-Variable Theories: A set of alternatives to quantum mechanics intended
to satisfy the objections of the EPR paper, in which the uncertainty principle does
not apply and a quantum system can simultaneously have definite values of
complementary variables like position and momentum, provided one of these
values is somehow “hidden.” Hidden-variable theories are usually also “local”
(see Locality) to deal with Einstein’s objection to the nonlocality of quantum
mechanics.
Immaculate Conception Paradox: A type of back-in-time communication
paradox in which a completely consistent causal loop produces information
with no known origin. An example is the book paradox, in which an author
receives a book in a message from the future. He publishes it, and when the
time comes, he transmits the manuscript to himself in the past. The question
then arises, who wrote the book? In this case, no inconsistent timelike loops
are involved, the arguments against bilking do not apply in this case.
Locality: The assumption that the correlations between parts of a system can only
be established while the subsystems are in contact (or speed-of-light communi-
cation) and that, once out of such contact, no changes in such correlations
are possible.
528 J. G. CRAMER

Nonlocality: The situation, apparently present in quantum mechanics, that

correlations between parts of a system can be established independent of the
separation of the parts in time and space.
Retro-Causal: Situations in theory or in the real world in which the effect
precedes the cause, in violation of the principle of causality.

References
[1] Freedman, S. J., and Clauser, J. F., “Experimental Test of Nonlocal Hidden Variable
Theories,” Physical Review Letters, Vol. 28, 1972, pp. 938 942.
[2] Aspect, A., Dalibard, J., and Roger, G., “Experimental Realization of Einstein
Podolsky Rosen Bohm Gedankenexperiment: A New Violation of Bell’s Inequal
ities,” Physical Review Letters, Vol. 49, 1982, pp. 91 95; Aspect, A., Dalibard, J.,
and Roger, G., “Experimental Test of Bell’s Inequalities Using Time Varying
Analyzers,” Physical Review Letters, Vol. 49, 1982, p.1804.
[3] Schrödinger, E., “Discussions of Probability Relations between Separated Systems,”
Proceedings of Cambridge Philosophical Society, Vol. 31, 1935, pp. 555 563; Vol.
32, 1936, pp. 446 451.
[4] Einstein, A., Podolsky, B., and Rosen, N., “Can Quantum Mechanical Description of
Physical Reality Be Considered Complete?,” Physical Review, Vol. 47, 1935, pp.
777 785.
[5] Bell, J. S., Physics, Vol. 1, 1964, p. 195; “On the Problem of Hidden Variables in
Quantum Mechanics,” Review of Modern Physics, Vol. 38, 1966, p. 447.
[6] Jarrett, J. P., “On the Physical Significance of the Locality Condition in the Bell
Argument,” Noûs, Vol. 18, 1984, p. 569.
[7] Shimony, A., “Events and Processes in the Quantum World,” Quantum Concepts in
Space and Time, Penrose, R., and Isham, C. J. (eds.), Clarendon Press, Oxford, 1986,
pp. 182 203.
[8] Wu, C. S., and Shanknov, I., “The Angular Correlation of Scattered Annihilation
Radiation,” Physical Review, Vol. 77, 1950, p. 136.
[9] Tittel, W., Brendel, J., Zbinden, H., and Gisin, N., “Violation of Bell Inequalities by
Photons More Than 10 km Apart,” Physical Review Letters, Vol. 81, 1998, pp.
3563 3566.
[10] Pagels, H., The Cosmic Code, Simon & Schuster, New York, 1982.
[11] Eberhard, P. H., “Bell’s Theorem Without Hidden Variables,” Nuovo Cimento,
Vol. B38, 1977, p. 75; “Bell’s Theorem and the Different Concepts of
Locality,” Nuovo Cimento, Vol. B46, 1978, p. 392; Ghirardi, G. C., Rimini, A.,
and Weber, T., “A General Argument Against Superluminal Transmission
Through the Quantum Mechanical Measurement Process,” Lett. Nuovo Cimento,
Vol. 27, 1980, pp. 293 298; Yurtsever, U., and Hockney, G., “Signaling, Entangle
ment, and Quantum Evolution Beyond Cauchy Horizons,” Classical and Quantum
Gravity, Vol. 22, 2005, pp. 295 312.
[12] Peacock, K. A., and Hepburn, B., “Begging the Signaling Question: Quantum Signal
ing and the Dynamics of Multiparticle Systems,” Proceedings of the Meeting of the
Society of Exact Philosophy, 1999, quant ph/9906036.
[13] Strekalov, D. V., Sergienko, A. V., Klyshko, D. N., and Shih, Y. H., “Observation of
Two Photon ‘Ghost’ Interference and Diffraction,” Physical Review Letters, Vol. 74,
No. 17, 1995, pp. 3600 3603.
 QUANTUM ENTANGLEMENT AND NONLOCALITY 529

[14] Dopfer, B., “Zwei Experimente zur Interferenz von Zwei Photonen Zuständen, Ein
Heisenbergmikroskop und Pendellösung,” PhD Thesis, University of Innsbruck,
1998; Zeilinger, A., “Experiment and the Foundations of Quantum Physics,”
Review Modern Physics, Vol. 71, 1999, S288 S297.
[15] Abouraddy, A. F., Nasr, M. B., Saleh, B. E. A., Sergienko, A. V., and Teich, M. C.,
“Demonstration of the Complementarity of One and Two Photon Interference,”
Physical Review A, Vol. 63, 2001, 063803.
[16] Zehnder, L., “Ein neuer Interferenzrefraktor,” Z. Instrumentenkunde, Vol. 11, 1891,
p. 275; L. Mach, “Über einen Interferenzrefraktor,” Z. Instrumentenkunde, Vol. 12,
1892, p. 89.
[17] Wheeler, J. A., and Feynman, R. P., “Classical Electrodynamics in Terms of Direct
Interparticle Action,” Review of Modern Physics, Vol. 21, 1949, pp. 425 433.
[18] Echeverria, F., Klinkhammer, G., and Thorne, K. S., “Billiard Balls in Wormhole
Spacetimes with Closed Timelike Curves: Classical Theory,” Physical Review D,
Vol. 44, 1991, pp. 1077 1099.
[19] Weinberg, S., “Precision Tests of Quantum Mechanics,” Physical Review Letters,
Vol. 62, 1989, pp. 485 490.
[20] Polchinski, J., “Weinberg’s Nonlinear Quantum Mechanics and the Einstein
Podolsky Rosen Paradox,” Physical Review Letters, Vol. 66, 1991, pp. 397 401.
 Chapter 17

Comparative Space Power Baselines

Gary L. Bennett†
Metaspace Enterprises, Emmett, Idaho

I. Introduction
PACE power is an integral part of any space mission without a source of
S electrical power nothing else works, including propulsion. Moreover, in a
number of advanced space propulsion concepts (e.g., electric propulsion,
fusion, etc.) power is essential for the advanced space propulsion concept to func-
tion. This chapter examines space power technology to convey its existing and
projected performance limits that are based on established science. This infor-
mation is provided both as a baseline for contemplating powering new propulsion
science techniques with known power methods, as well as to define the limits that
any new power methods would have to exceed to become competitive. Also, to
help convey a sense of progress, much of this information is presented in its
historical context. Because missions to the outer solar system and beyond will
require power sources independent of the sun, the focus of this chapter is on
nuclear power sources.
Figure 1 illustrates the basic subsystems of a typical Earth-orbiting spacecraft.
Within the electric power system (EPS) of a typical spacecraft, one can further
subdivide into the basic configuration shown in Fig. 2. In very simple terms,
an EPS can be thought of as consisting of three basic elements: 1) power
source; 2) power management and distribution (PMAD); and 3) energy storage

Copyright # 2008 by Gary L. Bennett. Published by the American Institute of Aeronautics and
Astronautics, Inc., with permission.
Unless otherwise indicated, mission information in this chapter was taken from Web sites main
tained by the mission organizations or by the National Space Science Data Center ,http://
nssdc.gsfc.nasa.gov/..

First presented as G. L. Bennett, “Space Nuclear Power: Opening the Final Frontier,” AIAA
Paper 2006 4191, 4th International Energy Conversion Engineering Conference, 26 29 June 2006,
San Diego, California.
†
Director.

531
532 G. L. BENNETT

Fig. 1 Typical subsystems on an Earth-orbiting spacecraft. (Artist’s illustration of

the U.S. Navy ultra-high-frequency follow-on satellites, courtesy of Hughes Space
and Communications.)

[1]. This chapter will focus on power sources and energy storage; however, the
spacecraft designer must not forget the ever-important PMAD element.
In this chapter we shall review the basic space power concepts and list their
current attributes. Historically, space power sources have been divided into
nuclear and non-nuclear where the former has included radioisotope power
sources and fission power sources, and the latter has generally included solar
power (e.g., photovoltaic systems) and energy storage (e.g., batteries or fuel
cells). Looking to advanced “classical” systems one can include fusion and anti-
matter. For the purposes of this chapter, both fusion and antimatter power sources
will be included under the umbrella term “nuclear.”
Figure 3, which is based on studies conducted at the Jet Propulsion Laboratory,
shows the attributes of the major classical power sources in terms of the theoretical

Fig. 2 Basic configuration of a spacecraft electric power system. (Adapted from

Ref. 1.)
 COMPARATIVE SPACE POWER BASELINES 533

Fig. 3 Theoretical performances of advanced non-nuclear and nuclear power

sources. (From Ref. 2.)

highest specific impulse they can produce. (Here specific impulse is expressed
in its natural unit of velocity, i.e., meters per second.) Free radicals are neutral
monatomic or polyatomic fragments produced by the dissociation of molecules.
An electronically excited state of an atom or molecule is said to be metastable if
its radiative lifetime is greater than a microsecond.
In terms of performance, whether energy release or specific impulse, the clas-
sical nuclear systems (fission, fusion, antimatter) have the clear advantage as
shown in Fig. 4 where candidate near-term missions such as a Comet Nuclear
Sample Return (CNSR) and Mars Rover Sample Return (MRSR) are compared
with more challenging missions such as the proposed Thousand Astronomical
Unit (TAU) mission beyond the solar system and a mission to Alpha Centauri.
However, there are occasions when non-nuclear power sources can help in
advancing human exploration of the solar system and beyond, so it is important
to consider them as well. For example, a mission planner might want to use a
solar electric propulsion (SEP) stage to send a spacecraft beyond Earth, dropping
that SEP stage when the sunlight becomes too dim, and continuing with gravity
assists and some type of nuclear power. Some have proposed using the Sun for a
gravity assist to send a spacecraft beyond the solar system.‡

‡
The use of another planetary body’s gravity to accomplish a mission has been successfully
employed on a range of NASA missions, e.g., Pioneers 10 and 11; Voyagers 1 and 2; Mariner 10;
Galileo; Ulysses; Cassini; MESSENGER; and New Horizons. Because energy is conserved in such
encounters the increase in velocity and/or change in direction for the spacecraft comes at the
expense of a minuscule decrease in the velocity of the planetary body.
534 G. L. BENNETT

Fig. 4 Mission velocity changes (Dv) as a function of specific energy. (Courtesy of

the Jet Propulsion Laboratory.)

The performance characteristics are also fundamentally clear when one

considers that the energies of chemical reactions are typically measured in electron
volts (eV) while nuclear reactions are orders of magnitude greater. For
example, the alpha particle (4He) decay of plutonium-238 (238Pu or Pu-238)
carries almost 5.5 MeV of energy. The fission of a uranium-235 nucleus (235U
or U-235) is about 200 MeV. The fusion reaction of a deuteron (2H) and a
tritium nucleus (“triton” or 3H) the so-called “d-t reaction” yields 17.6 MeV.
While this may not seem like much compared to fission reactions, the energy
per nucleon is much greater (3.5 MeV/nucleon for d-t fusion versus about
0.8 MeV/nucleon for 235U fission). In contrast, the annihilation of a nucleon via
a matter antimatter reaction would release over 938 MeV/nucleon. Radioactive
decay rates are fixed; however, in principle, the reaction rates for fission, fusion,
and antimatter can be increased as needed for the application.
Mission planners and space power designers have several top-level figures of
merit that they use. For power sources, the interest is in having a high specific
power (We/kg) and usually a high power density (We/m3). In the electric pro-
pulsion world, there is often a focus on the inverse, having a low specific mass
(kg/kWe). In the solar power community, there is also a desire to reduce the
area of the solar arrays; hence, the occasional requirement on areal power
density (We/m2). For energy storage systems, the focus is usually on having a
high specific energy (We-h/kg) and usually a high energy density (We-h/m3).
In addition, the mission planner needs to know the values at beginning-of-life
(BOL), beginning-of-mission (BOM), and end-of-mission (EOM).
 COMPARATIVE SPACE POWER BASELINES 535

II. Non-nuclear Space Power Sources

In the space power community, non-nuclear power sources typically include
solar and chemical. Solar can be divided into photovoltaic power sources and
solar dynamic power sources. In the former, sunlight is directly converted into
electrical power by means of a solar cell. Solar dynamic power sources utilize
solar heat to drive a working fluid in a “dynamic” conversion system such as
Brayton (gas-operated turbine-alternator-compressor), Rankine (two-phase
fluid-operated turbine-generator), and Stirling (gas-operated linear alternator).
Because dynamic conversion systems will be covered in Section III, this
section will focus on photovoltaic power sources.
Chemically produced power can come from batteries and fuel cells. Such
devices are usually referred to as energy storage systems (along with capacitors
and flywheels). Energy storage systems tend to be power rich but energy limited
(e.g., once the battery is discharged that is the end of the power until it can be
recharged if rechargeable). Power sources tend to be energy rich but power
limited (e.g., a radioisotope power source can operate for decades, which trans-
lates into a large energy release, but it is fixed in power).
Figure 3 shows the potential benefits of two other non-nuclear power sources:
free radicals and metastable propellants. In principle, these concepts offer the
potential of high specific impulses and high thrusts without the nuclear radiation
associated with “classical” nuclear propulsion concepts. However, storage and
control are issues yet to be satisfactorily addressed with some studies indicating
that when these factors are included, the performance of free radicals and meta-
stable propellants drops to that of nuclear thermal propulsion systems which do
not have these same storage and control issues [2]. With the ongoing research on
Bose Einstein condensates (BECs) and related laser and magnetic field studies,
there may yet be a breakthrough in the use of free radicals and metastable
propellants.

A. Photovoltaic Space Power Sources

In the early days of the space program, solar arrays produced about 15 We/kg
using solar cells that were about 10% efficient. (Unless otherwise specified solar
array and solar cell performance values are presented for 1 AU from the sun with
no intervening atmosphere, i.e., “air mass zero” or AM0.) The early space solar
cells were usually based on a single-crystal silicon material. The early Earth-
orbiting spacecraft received a few tens of watts from those early photovoltaic
power sources. Gradually the power production increased into the hundreds of
watts to today’s tens of kilowatts. The solar arrays on the International Space
Station (ISS) produce 110 kWe [3].
Since those early missions, the U.S. space community has focused on produ-
cing more power in lighter structures. One approach is to develop more efficient
solar cells. Efficiencies have been raised from the early 10% to over 27% in
production by using several layers of different materials (e.g., GaAs, GaInP,
Ge), which are typically referred to as dual-junction and triple-junction solar
cells. Efficiencies in laboratory cells beyond 35% are being measured [3].
It is possible to increase the specific power of solar arrays without using
advanced solar cells by employing low-mass structures. This approach was
536 G. L. BENNETT

used on NASA’s Advanced Photovoltaic Solar Array (APSA) program where

specific powers in the range of 130 We/kg for 12-kWe applications were demon-
strated with thin silicon solar cells for geosynchronous Earth orbit (GEO)
conditions [4,5].
Beginning in the late 1980s, NASA had two long-range goals for space solar
power: 1) to develop solar arrays that can provide 300 We/kg for deployable
planar arrays; and 2) to develop solar arrays that can provide 300 We/m2 at
100 We/kg in concentrator arrays (both at 1 AU from the sun) [5]. Because
the dual-junction and triple-junction solar cells are heavier than thin silicon
cells and thin-film cells, it may not be possible to have both attributes in the
same array. Already there are reports of arrays with specific powers of
300 We/kg with stowed packing densities on the order of 50 kWe/m3 [3].
Occasionally one will find studies in the literature of using extremely large
solar arrays to capture starlight for “generation ships” or “world ships” moving
between the stars. A quick calculation based on the inverse-square reduction in
insolation with distance from a star will easily show how unrealistically large
(with today’s technologies) such arrays would have to be. (A solar array that
would produce 1000 We at Jupiter would have to be over 25 times the size of
a 1000-We array operating in Earth orbit. At Pluto the array would have to be
over 1500 times as large!)
In choosing a solar array design, the mission planner and the spacecraft power
engineer need to consider such factors as [3]:
1) The range of insolation (solar intensity) of the mission (e.g., a solar probe
mission might travel between Jupiter and to within a few solar radii of the sun)
2) High intensity/high temperatures (e.g., Mercury and solar missions)
3) Low intensity/low temperatures (e.g., missions beyond Mars)
4) Very high power for solar electric propulsion (SEP)
5) High radiation fields (e.g., Europa, Jupiter)
6) Electrostatically clean arrays for fine magnetic measurements
7) Solar power in dusty environments (e.g., surface of Mars)
Of particular concern for mission planners considering sending a spacecraft to
the outer planets and beyond is the “LILT” (low-intensity, low-temperature)
effect. Solar arrays operating in Earth’s orbit typically have steady-state tempera-
tures on the order of 313 K to 343 K. Some benefit can be obtained by reducing
the temperature to about 223 K (at a solar distance of about 3 AU). Beyond about
3 AU, the combined effects of the low temperature and the low intensity degrade
the conversion efficiency of the solar cell [3].
Large solar arrays obviously have impacts on the attitude control of a space-
craft, which in turn affects the amount of attitude control propellant that must be
carried. In addition, concentrator arrays can have an even larger attitude control
effect than planar arrays because of their requirement to be more precisely
pointed at the sun.

B. Chemical Space Energy Storage Systems

Energy storage systems can provide backup power in case the primary power
source shuts down. Energy storage systems can also provide extra power for short
 COMPARATIVE SPACE POWER BASELINES 537

periods of time (e.g., “burst power”) when the power source is limited. The two
principal types of chemical energy storage systems are batteries and fuel cells.
Capacitors and flywheels are also candidates for energy storage in certain
applications.
Batteries have been the energy storage system of choice for most space mis-
sions. Batteries come in two types: 1) primary (single discharge) and 2) second-
ary (rechargeable). Primary batteries are typically used on missions that require
a one-time use of electrical power for a few minutes to several hours. The
state-of-practice on primary batteries is 250 We-h/kg with the goal of achieving
600 We-h/kg in 10 years [6].
Rechargeable batteries have been used principally for load leveling and to
provide power during eclipse periods. Historically, nickel-cadmium batteries
have been used; however, more recent space missions have used nickel-hydrogen
and lithium-based batteries. The state-of-practice on rechargeable, low-
temperature batteries is about 100 We-h/kg with the goal of reaching
200 We-h/kg in 10 years [6].
Fuel cells have been used on human missions (e.g., Gemini, Apollo, Space
Shuttle) because they can provide kilowatts of power. Fuel cells that can be
recharged are referred to as regenerative fuel cells (RFCs). NASA has had a
goal of developing the technology for a 1000 We-h/kg RFC with high efficiency
and greater than 5000-h reliable operation [5].

III. Nuclear Power Sources

Four types of classical nuclear power sources will be considered: 1) radioiso-
tope power; 2) fission power; 3) fusion power; and 4) antimatter power.

A. Radioisotope Power Sources

The development and use of nuclear power in space has enabled the human
race to extend its vision into regions that would not have been possible with
non-nuclear power sources. For example, in the bitterly cold, radiation-rich,
and poorly lit environments of the outer planets, only a rugged, solar-independent
power source has the capability to survive and function for long periods of time.
Even closer to the sun, environments can be too harsh or otherwise inhospitable
for more conventional power sources. For example, the long lunar nights (14
Earth days) create a major penalty for solar-powered systems. The frigid,
dimly lit polar regions of Mars are another potentially difficult location in
which to operate non-nuclear power sources for long periods of time.
Serious studies (such as Project Feedback) of using the immense potential
energy of the nucleus for power and propulsion in space began in the United
States shortly after the end of World War II. At a time when the full potential
of more conventional power and propulsion systems was not yet fully realized,
nuclear systems promised advantages in opening up the final frontier of space.
These early studies identified two types of nuclear power sources of potential
interest: radioisotope power sources and nuclear reactors [7].
Figure 5 illustrates the basic features of a radioisotope thermoelectric genera-
tor (RTG). The heart of the RTG is the radioisotope heat source that contains the
538 G. L. BENNETT

Fig. 5 Diagram of how a radioisotope thermoelectric generator works. (From [7,8].)

heat-producing radioisotope (plutonium-238 is used for U.S. space RTGs). The

thermal power from the heat source is transmitted to the thermoelectric elements
housed in the converter. The thermoelectric elements convert the thermal power,
into electrical power, which is sent to the spacecraft.
With advances in the more conventional technologies such as solar cells
and batteries it was more advantageous to use those technologies for the first
space flights, particularly because those early missions were in Earth orbit and
of short duration.
However, NASA and its predecessor, the National Advisory Committee on
Aeronautics (NACA), realized early on the need for nuclear systems in space
if the solar system was to be explored. During the Cold War, the U.S. Department
of Defense also realized the inherent advantages of nuclear systems, particularly
their hardness against certain types of countermeasures. (“Hardness” refers to the
ability to resist certain threats such as radiation or electromagnetic fields.
Because, by their nature, nuclear power sources operate in their own self-
generated radiation environment, their designs incorporate features that make
them resistant or “hard” to radiation effects. Reference 9 discusses some
aspects of survivability requirements for DoD spacecraft.)
Following on these early mission studies, the U.S. Atomic Energy Commis-
sion (USAEC), beginning in 1951, requested several studies on using radioiso-
topes and nuclear reactors to power spacecraft. As these studies were being
completed, K. C. Jordan and J. H. Birden, two researchers working at what
was then the USAEC’s Mound Laboratory (then operated by Monsanto Research
 COMPARATIVE SPACE POWER BASELINES 539

Corporation), built the first RTG in 1954. Even though this first RTG produced
only 1.8 mWe of power, it demonstrated the feasibility of coupling radioisotopes
with thermocouple-type (thermoelectric) conversion systems [7].
Table 1 summarizes the U.S. space missions that have utilized or are utiliz-
ing radioisotope and fission nuclear power (no fusion or antimatter missions
have been flown). The following sections discuss the various missions shown
in Table 1. The interested reader is referred to Refs. 10, 11, and 12 for more
details. Table 1 does not include those missions that used only radioisotope
heater units (RHUs).
All of the U.S. space nuclear power systems used thermoelectric technology to
convert the heat (thermal power) from the radioisotope heat source or nuclear
reactor into electrical power. The United States did, however, support research
into more advanced conversion technologies, including dynamic conversion
(Brayton, Rankine, and Stirling). The early RTGs used telluride-based materials
in their thermoelectric elements. The Systems for Space Nuclear Power (SNAP)
10A space nuclear reactor power system and the RTGs on all missions beginning
with Lincoln Experimental Satellites (LES) 8 and 9 (1976) used silicon-
germanium alloys.

Table 1 Uses of space nuclear power by the United Statesa

Transit Navy Navigational Satellites

† Transits 4A and 4B (1961) SNAP 3B (2.7 We)
† Transits 5BN 1 and 5BN 2 (1963) SNAP 9A (25 We)
† Transit TRIAD (1972) Transit RTG (35 We)
SNAPSHOT Space Reactor Experiment
† SNAP 10A nuclear reactor (1965) (500 We)
Nimbus 3 Meteorological Satellite
† SNAP 19B RTGs (1969) (2 at 28 We each)
Apollo Lunar Surface Experiments Packages
† Apollos 12 (1969), 14 (1971), 15 (1971), 16 (1972), 17 (1972) SNAP 27 (70 We
each)
Lincoln Experimental Satellites (Communications)
† LES 8 and LES 9 (1976) MHW RTG (2 per spacecraft at 154 We each)
Interplanetary Missions
† Pioneer 10 (1972) and Pioneer 11 (1973) SNAP 19 (4 per spacecraft at 40 We each)
† Viking Mars Landers 1 and 2 (1975) SNAP 19 (2 per Lander at 42 We each)
† Voyager 1 and Voyager 2 (1977) MHW RTG (3 per spacecraft at 158 We each)
† Galileo (1989) GPHS RTG (2 at  288 We each) (.3 year delay)
† Ulysses (1990) GPHS RTG (289 We) (.4 year delay)
† Cassini (1997) GPHS RTG (3 at  295 We each)
† New Horizons (2006) GPHS RTG (1 at 245.7 We) (most of fuel .21 years old)
a
Data are presented in the following reference: spacecraft/year launched, type of nuclear power
source, beginning of mission power. The odd numbered SNAP power sources used radioisotope
heating; the even numbered SNAP power sources used nuclear reactors to produce heat.
Abbreviations: GPHS RTG ¼ General purpose heat source radioisotope thermoelectric generator;
MHW RTG ¼ multi hundred watt radioisotope thermoelectric generator; SNAP ¼ systems for
nuclear auxiliary power.
540 G. L. BENNETT

1. Transit Navy Navigational Satellites

In 1958, researchers at the Johns Hopkins University Applied Physics Labora-
tory (JHU/APL) conceived the idea of a navigational satellite based on Doppler
technology as part of the APL-proposed Navy Navigation Satellite System. The
navigation satellite technologies developed under the Transit program are now in
use in the Global Positioning System (GPS). While solar cells and batteries
powered these early “Transit” satellites, JHU/APL accepted an offer from the
USAEC to have an auxiliary radioisotope power source, denoted by the
acronym SNAP-3B, for its Transit 4A and Transit 4B satellites [13]. And with
that act the era of space nuclear power was literally launched. The following sec-
tions discuss the Transit satellites in the order of launching.

a. Transits 4A and 4B. Transit 4A was launched on 29 June 1961 and was
quickly followed by the launch of Transit 4B on 15 November 1961. The objec-
tives of both missions were 1) to conduct navigation trials and demonstrations; 2)
to improve the understanding of the effects of ionospheric refraction on radio
waves; and 3) to increase knowledge of Earth’s shape and gravitational field.
Both spacecraft met all launch objectives [14].
Each 2.1-kg SNAP-3B RTG (one per spacecraft) produced about 2.7 We at
BOM from a radioisotope heat source that provided a thermal power of about
52.5 Wt. While these power levels seem low, both RTGs served to fill a critical
niche by powering the crystal oscillator that was the heart of the electronic system
used for Doppler-shift tracking. In addition, the RTGs powered the buffer-
divider-multiplier, phase modulators, and 54- and 324-MHz power amplifiers.
While various component failures interfered with a full analysis of the RTG per-
formance, there were sufficient data from other measurements to show that both
RTGs operated well beyond their design life of five years [14,15].
Figure 6 is an artist’s illustration of Transit 4A in space and Fig. 7 is a cutaway
of the SNAP-3B RTG.

b. Transits 5BN-1 and 5BN-2. Based on the successful performance of the

SNAP-3B RTGs on the Transit 4A and Transit 4B satellites, JHU/APL had the
confidence to design and fly two totally nuclear-powered navigational satellites
(Transit 5BN-1 and Transit 5BN-2). Each satellite used a new, higher power
RTG designated SNAP-9A. Each 12.3-kg SNAP-9A was designed to provide
25 We at a nominal 6 V for five years in space after one year of storage on
Earth [14,16].
Transit 5BN-1 was launched on 28 September 1963 from Vandenberg Air
Force Base (VAFB) in California. Transit 5BN-2 followed a little over two
months later on 6 December 1963 [14].
One of the objectives of the Transit 5BN series was to provide “a means by
which U.S. Navy ships may navigate anywhere in the world” [14]. Because
of some electronic problems, Transit 5BN-1 partially achieved this objective;
nevertheless, JHU/APL reported that: “All APL satellite navigational concepts
were validated using Satellite 5BN-1” [14].
In a summary report, JHU/APL stated that: “Satellite 5BN-1 was the first
artificial earth satellite to employ nuclear energy as its primary power source. . . .
In her role as a pioneer nuclear satellite, 5BN-1 demonstrated the extreme
 COMPARATIVE SPACE POWER BASELINES 541

Fig. 6 Artist’s concept of the Transit-4A satellite in orbit. The SNAP-3B RTG is
shown on “top” of the satellite. (Courtesy of Martin Nuclear.)

simplicity with which thermoelectric generators may be integrated into the design,
not only to provide the electrical power but also to aid in thermal control” [14].
Transit 5BN-2 reportedly met all launch objectives and was described by
JHU/APL as “the first truly operational navigation satellite” [14].
Figure 8 is an artist’s conception of the Transit 5BN-1 satellite showing it in
the gravity gradient stabilization mode with the SNAP-9A RTG mounted at the
aft end. Figure 9 is a photograph of a SNAP-9A RTG.

Fig. 7 Cutaway of the SNAP-3B RTG. SNAP-3B was 12.1 cm in diameter and 14-cm
high. The power was 2.7 We and the mass was 2.1 kg. (Courtesy of Martin Nuclear.)
542 G. L. BENNETT

Fig. 8 Artist’s illustration of the Transit 5BN-1 navigational satellite. (From

Ref. 14.)

c. Transit TRIAD. The TRIAD satellite (Fig. 10), launched on 2 September

1972 from VAFB, “was the first in a series of three experimental/operational
spacecraft designed to flight test improvements to the Navy Navigation Satellite
System” [14]. The principal power source was the Transit RTG (Fig. 11), which
was to provide a minimum EOM power of 30 We after five years at a minimum of
3 V. Four solar-cell panels and a 6-Ah nickel-cadmium battery provided auxiliary
power [17].
The 13.6-kg Transit RTG used a SNAP-19 radioisotope heat source (Section
III.A.6) coupled with light-weight thermoelectric panels using lead-telluride

Fig. 9 SNAP-9A RTG. The height was 26.7 cm and the fin span was 50.8 cm. The
power was 26.8 We and the mass was 12.3 kg. (Courtesy of Martin Nuclear.)
 COMPARATIVE SPACE POWER BASELINES 543

Fig. 10 Transit TRIAD satellite in orbit (artist’s conception). The TRANSIT RTG is
at the lower left.

technology. In effect, the heat source radiated to the panels so the RTG did not
have to be sealed [17].
JHU/APL reported that: “All TRIAD satellite and space technology
experiments were exercised and the TRIAD short-term objectives were

Fig. 11 Cutaway of the TRANSIT RTG. The RTG was approximately 36.3-cm high
by 61-cm across the flats. The TRANSIT RTG produced 35.6 We BOM with a mass of
13.6 kg. (Courtesy of TRW/GA/AEC.)
544 G. L. BENNETT

demonstrated” [14]. Despite a telemetry loss about one month into the mission that
precluded measuring the Transit RTG power, the functioning of various TRIAD
experiments showed that the Transit RTG more than met its objective [14].

2. Nimbus-3 Meteorological Satellite

The SNAP-9A development program moved on to the SNAP-19 technology
improvement program, which became the basis for NASA’s first outerplanetary
missions. First, though, came the checkout on NASA’s Nimbus-3 meteorological
satellite launched on 14 April 1969 from VAFB. Nimbus-3 was the first U.S.
weather satellite to make day and night global measurements from space of temp-
eratures at varying levels in the atmosphere. Figure 12 is an artist’s conception of
the Nimbus-3 meteorological satellite showing two SNAP-19 RTGs mounted
to the base platform. In addition to the two SNAP-19 RTGs, Nimbus-3 carried
solar arrays.
The requirement on the two SNAP-19 RTGs was to deliver 50 We to the
regulated-power bus after one year in orbit. The two 13.4-kg RTGs produced a
combined 56.4 We (49.4 We usable) at launch and 47 We one year later. This
nuclear power comprised about 20% of the total power delivered to the
regulated-power bus during that time, allowing a number of extremely important
atmospheric-sounder experiments to operate in a full-time duty cycle. Without
the RTGs, the total delivered power would have fallen below the load line
about two weeks into the mission [18,19].

Fig. 12 Nimbus-3 in orbit (artist’s conception). The two SNAP-19 RTGs are shown
mounted on the left. The RTGs had a height of 26.7 cm and a fin span of 53.8 cm. The
average power per RTG was 28.2 We with a mass of 13.4 kg. (Courtesy of NASA.)
 COMPARATIVE SPACE POWER BASELINES 545

3. Apollo Lunar Surface Experiments Packages

Nuclear power on the moon provides a number of important advantages,
chiefly the ability to deliver full power throughout the long (14 Earth day)
lunar night. This was one of the factors that led NASA to select RTG power
for its Apollo Lunar Surface Experiments Packages (ALSEPs). The program
objectives included acquiring scientific data to aid in determining the internal
structure and composition of the moon and the composition of the lunar atmos-
phere [20].
Five ALSEPs were placed on the moon, beginning with the Apollo 12 mission
(Fig. 13) whose ALSEP downlink acquisition started on 19 November 1969. The
Apollo 11 crew deployed a forerunner of ALSEP, known as the Early Apollo
Scientific Experiment Package (EASEP). The EASEP used solar cells for
power and two 15-Wt radioisotope heater units (RHUs) for warmth [20].
Power for each ALSEP was provided by a new RTG designated SNAP-27
(Fig. 14). The power requirement for the 19.6-kg RTG was to provide at least
63.5 We at 16-V DC for one year after lunar emplacement. (For Apollo 17, the
requirement was 69 We two years after emplacement [21].) All five ALSEP
SNAP-27 RTGs (Apollo 12, 14, 15, 16, and 17) exceeded their mission require-
ments in both power and lifetime, which enabled the ALSEP stations to gather
long-term scientific data on the moon [20].

4. Lincoln Experimental Satellites 8 and 9

Lincoln Experimental Satellites 8 and 9 (LES-8/9), which were built for the
U.S. Air Force, “were developed with the goal of demonstrating, in full-scale
operation (terminals as well as satellites), advanced technologies for strategic
communications links” (Fig. 15) [22]. Each spacecraft was powered by two
RTGs of a completely new design (Fig. 16). Termed “Multi-Hundred Watt

Fig. 13 Alan Bean removing the SNAP-27 on the Apollo 12 mission in November
1969. (Courtesy of NASA.)
546 G. L. BENNETT

Fig. 14 Cutaway of the SNAP-27 RTG. The converter was 46-cm high and 40.0-cm
across the fins. The RTG power was 63.5 We with a mass of 19.6 kg. (Courtesy of
GE.)

Radioisotope Thermoelectric Generator” (MHW-RTG), these RTGs were being

developed for multiple missions; however, the initial focus was on NASA’s
Grand Tour (later Voyager) mission. With minimal modifications, the MHW-
RTGs were used on LES-8/9 because they “offer impressive advantages of phys-
ical survivability by comparison with solar-cell arrays” [22].
For LES-8/9, each 39.69-kg MHW-RTG was to produce 125 We at 26 V at
the end of mission an operational life of at least five years after launch
(which occurred on 14 March 1976) [23]. BOM powers averaged about
154 We per RTG. The MHW-RTGs more than met this goal. LES-8 was
turned off on 2 June 2004 because of control difficulties (its two MHW-RTGs
were still providing usable electrical power). LES-9 continues to operate over
30 years after launch.
Writing 13 years after the launch, two of the LES-8/9 managers concluded,
“The RTGs were well worth the effort . . . they have performed superbly. They
provide continuous electrical power through the 70-min eclipses of the sun by the
Earth that LES-8 and LES-9 experience every day. The compatibility of these
rugged power sources with complex signal-processing circuitry has been well
established . . . the measures that were taken to assure the success of the RTGs
in LES-8 and 9 contributed directly to the success of the Voyager missions” [24].

5. Interplanetary Missions
From a public standpoint, the most spectacular uses of nuclear power sources
in space have come from the interplanetary missions, beginning with the launches
of the Pioneer 10 and Pioneer 11 spacecraft in 1972 and 1973, respectively. When
a spacecraft is sent where the sunlight is low (at Jupiter it is 25 times less than at
Earth; at Pluto, .900 times less than at Earth), where the temperatures are quite
 COMPARATIVE SPACE POWER BASELINES 547

Fig. 15 LES-8/9 in orbit. The two MHW-RTGs are shown mounted on top of one of
the satellites.

low (130 K at Jupiter), and where the radiation belts are very severe, the only
option is nuclear power. This section summarizes the RTG performance on U.S.
interplanetary missions from the Pioneers (Jupiter, Saturn) to the Vikings
(Mars), Voyagers (Jupiter, Saturn, Uranus, Neptune), Galileo (Jupiter), Ulysses
(solar polar), and Cassini (Saturn) to the most recent, the New Horizons
mission to Pluto.

a. Pioneers 10 and 11. Pioneers 10 and 11 were built to explore the

environment of Jupiter and the interplanetary medium beyond the orbit of
Mars and the asteroid belt. Pioneer 10 was launched on 2 March 1972 on a
fast trajectory to Jupiter (closest approach on 3 December 1973). The twin space-
craft Pioneer 11 was launched on 5 April 1973 to follow Pioneer 10 to Jupiter
(closest approach on 4 December 1974).
Each Pioneer spacecraft carried four 13.6-kg SNAP-19 RTGs, producing an
average of 40.3 We per RTG at BOM. The mission requirement was that the
four SNAP-19 RTGs on each Pioneer spacecraft had to produce 120 We total
at the Jupiter flyby. The RTGs more than met this requirement, which enabled
NASA to direct Pioneer 11 on for a first encounter with Saturn (closest approach
on 1 September 1979). The Pioneer 11 SNAP-19 RTGs exceeded the 90 We
required at Saturn by producing 119.3 We. By this time the RTGs on both space-
craft had operated well beyond their original 21-month requirement [25]. Both
Pioneer spacecraft also carried a number of small radioisotope heater units
548 G. L. BENNETT

Fig. 16 Cutaway of the MHW-RTG. The length was 58.31 cm and the overall
diameter was 39.73 cm. For LES-8/9 the average BOM power was 154 We per
RTG with an average mass of 39.69 kg. For Voyager 1/2, the average BOM power
was 158 We/RTG with an average mass of 37.69 kg. (Courtesy of GE.)

(RHUs) to keep equipment warm without the potential electromagnetic

interference of electrical heaters.
Both spacecraft continued to operate for decades after their original mission
objectives had been met and so gave scientists a wealth of information on the
interplanetary medium in the outer solar system. The last signal from Pioneer
10 was received on 22 January 2003, more than 30 years after launch. The last
signal from Pioneer 11 was received on 30 September 1995. Figure 17 is an
artist’s illustration of Pioneer 10, and Fig. 18 is a cutaway of the Pioneer
SNAP-19 RTG.

b. Viking Landers 1 and 2. In 1975, the United States embarked on the first
surface exploration of Mars with the launches of the two Viking missions. Each
Viking spacecraft consisted of an orbiter and a lander. In view of the hostile
environment of Mars (including dust storms and Antarctic-style temperatures),
NASA selected a modified SNAP-19 RTG to power the Viking landers (two
RTGs per lander). The modifications included the addition of a dome for gas
exchange with the main body of the RTG. The average BOM power of these
15.2-kg RTGs was 42.7 We per RTG. The requirement was to produce a
minimum of 35 We during the 90-day primary mission [25,26]. Figure 19
shows how the SNAP-19 RTGs were placed on a Viking lander. Viking 1 was
 COMPARATIVE SPACE POWER BASELINES 549

Fig. 17 Pioneer 10 leaving the solar system (artist’s conception). (Courtesy of

NASA.)

launched on 20 August 1975 and its lander reached the Martian surface on 20 July
1976. Viking 2 was launched on 9 September 1975 and its lander reached the
Martian surface on 3 September 1976.
All four SNAP-19 RTGs easily met the 90-day requirement allowing the
Viking landers to operate for years until other system failures led to a loss of

Fig. 18 Cutaway of the Pioneer SNAP-19 RTG. The height was 28.2 cm and the fin
span 50.8 cm. The average BOM power was 40.3 We per RTG. (Courtesy of TES.)
550 G. L. BENNETT

Fig. 19 Model of the Viking Mars lander showing the locations of the two Viking
SNAP-19 RTGs. The RTGs had a height of 40.4 cm and a fin span of 58.7 cm. The
average BOM power per RTG was 42.7 We and the average mass was 15.2 kg per
RTG. (Courtesy of NASA/TES.)

data. When the last data were received from Viking Lander 1 in November 1982,
it had been estimated that the RTGs were capable of providing sufficient power
for operation until 1994 18 years beyond the original mission requirement. Had
an accidental shutdown not occurred, it is fascinating to speculate if Viking
Lander 1 might have been ready to greet the Mars Sojourner when it landed on
4 July 1997.
It is worth noting that all three U.S. Mars Rovers (Sojourner, Opportunity, and
Spirit) carried 1-Wt light-weight radioisotope heater units (LWRHUs) to keep
them warm during the cold Martian nights.

c. Voyager 1 and Voyager 2. Following on the paths of Pioneers 10 and 11,

the more advanced Voyager 1 and Voyager 2 spacecraft were launched in 1977 to
conduct more detailed studies of Jupiter and Saturn, their satellites and mag-
netospheres as well as studies of the interplanetary medium. The successful com-
pletion of the Saturn flybys allowed NASA to send Voyager 2 to the first flybys of
Uranus and Neptune, completing most of the objectives for the originally planned
Grand Tour mission.
To power these two spacecraft so far from the sun, the Voyagers used multi-
hundred watt radioisotope thermoelectric generators (MHW-RTGs) similar in
design to those flown the year before on the U.S. Air Force communications
satellites LES-8/9 (Fig. 16 for LES-8/9 version). With some modifications,
the MHW-RTG mass was reduced to an average of 37.69 kg. The Voyager
mission required an EOM minimum power of 128 We per RTG four
years after launch. BOM powers averaged about 158 We per RTG [10,11].
(Both Voyager spacecraft also carried a number of small RHUs to keep
equipment warm.)
 COMPARATIVE SPACE POWER BASELINES 551

Fig. 20 Artist’s concept of the path of a Voyager spacecraft past Jupiter and Saturn.
The three MHW-RTGs are shown mounted below the spacecraft on a boom.

The two Voyager spacecraft (Fig. 20) have probably explored more territory
than any other spacecraft in human history. Discoveries ranged from finding 22
new satellites (3 at Jupiter; 3 at Saturn; 10 at Uranus; and 6 at Neptune) to wit-
nessing the first volcanic eruption on another solar system body (Io) and geysers
on another satellite (Triton). The two Voyager spacecraft are now operating in an
interstellar mode where they continue to provide new information beyond the
orbit of Pluto.

d. Galileo. With the completion of the successful Pioneer and Voyager

flybys scientists turned to an in-depth study of the largest planet in the solar
system. Conceived at the time of the Voyager launches, the Galileo spacecraft
was launched on 18 October 1989 on a series of gravity assist maneuvers to
take this technologically sophisticated orbiter and atmospheric probe to Jupiter.
There Galileo investigated the Jovian atmosphere; the Jovian satellites; the
Jovian magnetosphere; and energetic particles and plasma.
Powering Galileo in this hostile environment were two new RTGs designated
by the acronym GPHS-RTG (general-purpose heat source radioisotope thermo-
electric generator). At the time of fueling (BOL), a GPHS-RTG is capable of pro-
ducing about 300 We in a mass envelope of about 55.9 kg making this the highest
specific power nuclear power source ever flown by the United States. Because of
launch delays caused by the Challenger accident, the power requirement for the
Galileo GPHS-RTGs was reduced to 470 We (235 We per RTG) at EOM
552 G. L. BENNETT

Fig. 21 Galileo orbiter communicating with the Galileo probe. One of the two
GPHS-RTGs is shown on a boom “above” Galileo. (Courtesy of NASA/JPL.)

(71,000 h after BOM) [27]. The combined BOM power of the two GPHS-RTGs
was 577.2 We [28]. Galileo went into orbit on 7 December 1995 as the probe
entered the atmosphere of Jupiter. Figure 21 is an artist’s conception of the
Galileo spacecraft at Jupiter and Fig. 22 illustrates the features of the GPHS-
RTG. Both the Galileo orbiter and probe carried new, lower-mass 1-Wt RHUs,
which were termed the light-weight radioisotope heater unit or LWRHU.
Both GPHS-RTGs met their EOM power requirements allowing NASA and
JPL to extend the Galileo mission three times. Finally, on 21 September 2003,
after 35 orbits of Jupiter and with its propellant running low, Galileo was
sent into the atmosphere of Jupiter so that it would not collide with the

Fig. 22 Cutaway of the GPHS-RTG. The length is 114 cm and the fin span is
42.2 cm. At time of fueling, the GPHS-RTG can produce 300 We. For Galileo and
Ulysses, the average RTG mass was 55.9 kg. (Courtesy of LMA/DOE.)
 COMPARATIVE SPACE POWER BASELINES 553

oceanic satellite Europa. In its 14-year life, Galileo sent back data on Earth,
Venus, two asteroids (Gaspra and Ida and its moon Dactyl) and the Jovian
system plus views of comet Shoemaker-Levy 9 crashing into Jupiter. Galileo
also sent the first atmospheric probe into the atmosphere of a giant planet, a
probe kept warm with LWRHUs.

e. Ulysses. Originally planned as part of a two-spacecraft mission, the

Ulysses spacecraft was launched on 6 October 1990 “to investigate for the first
time as a function of heliographic latitude, the properties of the solar wind, the
structure of the sun/wind interface, the heliospheric magnetic field, solar radio
bursts and plasma waves, solar x-rays, solar and galactic cosmic rays and both
interstellar and interplanetary neutral gas and dust” [29]. Even though Ulysses
is essentially a solar polar mission, it needed the immense gravity of Jupiter to
bend its orbit out of the plane of the ecliptic. These regular trips out to 5 AU
(along with the Jovian radiation belts) dictated a nuclear power source (GPHS-
RTG) for Ulysses.
The Ulysses GPHS-RTG power requirements were modified because of the
more than four-year delay following the Challenger accident: provide a BOM
power of 277 We and an EOM power (42,000 h) of 245 We [27]. Telemetry
measurements indicated a BOM power of 284 We at the bus and 289 We at
the RTG connector, both in excess of the minimum required. The August 1995
‘EOM’ was reported to be 248 We, 3 watts above the minimum [28]. Because
of the excellent performance of the Ulysses GPHS-RTG, the mission has been
extended several times. In November 2006, Ulysses started its third polar orbit
of the sun and, beginning in November 2007, the spacecraft moved over the
sun’s north polar latitudes. The Ulysses mission ended in June 2008, having
operated more than three times longer than planned. In this time Ulysses has
returned a wealth of information on the latitudinal distribution of particles and
fields from our closest star.
Figure 23 is an artist’s illustration of the Ulysses mission showing its GPHS-
RTG mounted on the side of the main body of the spacecraft.

f. Cassini. Following on the successful Pioneer and Voyager flybys of

Saturn, scientists turned next to an orbiter/probe mission to provide a more in-
depth study of the ringed world. On 15 October 1997, the Cassini spacecraft
was launched to do, in a sense, what Galileo had done at Jupiter: carry out a
multiyear scientific study of the second largest gas giant in the solar system.
Mission objectives included studying the rings, the satellites, the magnetosphere,
and, in particular, Titan, the cloud-covered largest satellite of Saturn. Like
Galileo, Cassini carried a probe (Huygens), this one built by the European
Space Agency (ESA) and designed to land on Titan.
Cassini successfully went into orbit around Saturn in July 2004 after traveling
past Venus, Earth, and Jupiter. Huygens, which carried Galileo-style lightweight
radioisotope heater units (LWRHUs) to keep it warm, successfully landed on
Titan and provided dramatic views of that shrouded world. Cassini is now well
into its 4-year main mission of 74 orbits around Saturn with 44 flybys of Titan.
Because of mission complexity, Cassini needed more power than used on pre-
vious missions of this type. Three modified GPHS-RTGs provide that power. The
554 G. L. BENNETT

Fig. 23 Ulysses spacecraft shown in its second Jupiter encounter (November 2003 to
April 2004). The GPHS-RTG is shown mounted on “top” of the spacecraft. (Courtesy
of NASA/ESA/JPL.)

specification requirement for BOM was 826 We. The three GPHS-RTGs provided
887 We total at BOM. The average mass per RTG is 56.4 kg. The projected power
at 16 years after BOM is 640 We, which will exceed the specification requirement
of 596 We [28]. The prognosis is excellent that the Cassini RTGs will provide suf-
ficient power for an extended mission beyond the main mission. Figure 24 is an
artist’s illustration of the Cassini spacecraft showing two of the three GPHS-
RTGs. Cassini also carried LWRHUs to keep equipment warm.

g. New Horizons Pluto Kuiper Belt Flyby. One planet remained unex-
plored after the Pioneer and Voyager flyby missions: Pluto, once thought to be
the “last planet.” On 19 January 2006, the New Horizons spacecraft was launched
to fly by Pluto and its moon Charon. New Horizons will then continue on into the
Kuiper Belt where it will fly by one or more Kuiper belt objects (KBOs). On its
way to the Pluto/Charon system, New Horizons flew by Jupiter in late February
2007 for a gravity assist. During that four-month encounter period, New Horizons
studied Jupiter, adding to our knowledge of the solar system’s largest planet.
New Horizons is expected to fly by Pluto on 14 July 2015 (Fig. 25). From
about 2016 to 2020, following the Pluto/Charon flyby, New Horizons will
enter the KBO study phase of the mission.
In a region of space where the sunlight is less than 0.001 of what it is at Earth,
nuclear power is the only viable option. Thus, New Horizons carries a single
GPHS-RTG to provide power in the far reaches of the solar system. With
various modifications (including changes to the heat source modules), the
 COMPARATIVE SPACE POWER BASELINES 555

Fig. 24 Cassini shown over Titan as the Huygens probe descends toward Titan
(artist’s conception). Two of the three GPHS-RTGs are shown mounted at the
“upper end” in this figure. (Courtesy of NASA/JPL.)

flight mass is 57.8 kg. The BOM power from this GPHS-RTG was 245.7 We
versus a specification requirement of 237 We. This power was lower than what
has been produced in the GPHS-RTGs flown on earlier missions, primarily
because the New Horizons GPHS-RTG used some older fuel that had decayed
for over 21 years [30,31]. Still, it is expected that the GPHS-RTG will produce
about 200 We (versus the specification minimum of 191 We) at the time of
Pluto/Charon flyby, enough for New Horizons to conduct exciting scientific
studies of what was once thought to be our farthest planet [31].

6. Future Uses of Radioisotope Power Sources

In September 2009, NASA plans to launch the Mars Science Laboratory
(MSL), which will use a modified SNAP-19 RTG known as the Multi-Mission
Radioisotope Thermoelectric Generator (MMRTG). The MMRTG, like the
Viking SNAP-19, is designed to operate on the surface of Mars. The projected
power of the MMRTG at BOM is 121 We at a thermal power of 1982 Wt for
Mars noon and 123 We in deep space. After 14 years, these powers are projected
to be 96 We and 97 We, respectively. With a mass of about 45 kg, the specific
power works out to be about 2.7 We/kg.
Separately, NASA and the U.S. Department of Energy are sponsoring the
development of an advanced Stirling radioisotope generator (ASRG) to
produce 140 We BOM. With a mass of 20.2 kg, the specific power will be
556 G. L. BENNETT

Fig. 25 New Horizons flying over Pluto with Charon in the background. The GPHS-RTG
is shown mounted on the left side in this illustration. (Courtesy of JHU/APL; SwRI.)

almost 7 We/kg. The additional advantage of the ASRG is that the higher effi-
ciency (29%) of the Stirling conversion system will translate into much less
238
Pu being required; a definite advantage at a time when the United States is
no longer producing 238Pu.
Radioisotope power sources have been considered to power electric propulsion
thrusters, a concept designated radioisotope electric propulsion (REP). After being
sent on an Earth escape trajectory by a chemical upper stage, REP used in conjunc-
tion with gravity assists can enable some challenging science missions such as orbit-
ing the satellites of the outer planets. One can envision a combination of SEP, REP,
and gravity assists to send an interstellar explorer on its way.

B. Fission Space Power

Both the United States and the former Soviet Union have had active technol-
ogy and flight programs in fission space power. In addition to the fission space
power programs, both nations have had technology programs in fission propul-
sion (both nuclear thermal propulsion, NTP, and nuclear electric propulsion,
NEP).

1. U.S. Fission Space Power

The first nuclear reactor to be employed in space was launched by the U.S. Air
Force from VAFB on 3 April 1965. The reactor, which was designated SNAP-10A,
 COMPARATIVE SPACE POWER BASELINES 557

Fig. 26 SNAP-10 in orbit attached to the Agena bus (artist’s conception). The
overall length was 3.48 m and the mounting base diameter was 1.27 m. The total
system mass of the final flight unit was 435 kg. SNAP-10A was designed to produce
at least 500 We for 1 year. (Courtesy of Atomics International/AEC.)

built upon the heritage of the earlier SNAP reactor programs, in particular, SNAP-2
and SNAP-10. SNAPSHOT, as the experiment was named, was a test of the
operability of an automated space reactor. The power requirement for the 435-kg
SNAP-10A was to produce at least 500 We for 1 year [32]. Figure 26 is an
artist’s conception of the SNAP-10A reactor coupled to the Agena “spacecraft.”
Once in its 1288-km by 1307-km nuclear safe orbit, the reactor was started
up and operated. All went well for 43 days until a failure of a voltage regulator
in the Agena “spacecraft” caused a termination of the power operation. Neverthe-
less, SNAP-10A showed that it was feasible to remotely operate a liquid
metal-cooled nuclear reactor in space. The capability of SNAP-10A to operate
unattended for one year was demonstrated in a ground-test twin to the flight
reactor [32].
In addition to the SNAP-10A program, during the late 1950s to the early 1970s
the United States sponsored a number of other space reactor programs (e.g., SNAP-
10, SNAP-50/SPUR) that would use advanced power conversion systems
(e.g., Brayton and Rankine) to reach even higher powers (e.g., up to 1 MWe for
SNAP-50). In particular, the SNAP-50/SPUR (Space Power Unit Reactor)
program had nuclear electric propulsion (NEP) as one of its proposed uses.
From 1983 to 1994, the United States sponsored the SP-100 space nuclear
reactor power system technology program (Fig. 27). The goal was to develop a
space nuclear reactor technology that could support a range of projected future
missions including planetary surface operations and NEP for science missions.
The generic flight system (GFS) was established to support operational missions
requiring relatively high power (100-kWe class) for 10-year mission durations,
but scalable from about 10 kWe to 1000 kWe and with high specific power.
558 G. L. BENNETT

Fig. 27 SP-100 space nuclear reactor power system module showing the key
subsystems. The diameter and length of the main body (less the radiator panels)
are 3.5 m and 6 m, respectively, and the mass is 4575 kg for the 100-kWe GFS.
A deployable boom is used to maintain a separation distance of 22.5 m between
the reactor and the payload plane to keep the neutron and gamma doses within
specified values. (Courtesy of GE.)

The 100-kWe version with mature thermoelectric conversion was estimated to

have a specific power of 26 We/kg [33,34].
One study of coupling the SP-100 nuclear reactor with either Brayton or
Stirling dynamic power conversion systems at various temperatures for a lunar
base showed that for a 550-kWe power output, all of them were within 30% of
12,000 kg, corresponding to a specific mass of 21.8 kg/kWe or a specific
power of 45.8 We/kg [35].
Nuclear reactors have the advantage that they scale up well in terms of
reduced specific mass as the power increases. A scaling study completed in
1991 showed that coupling a nuclear reactor based on SP-100 technology to a
potassium Rankine power conversion system could produce 1000 kWe with a
mass of about 7600 kg, yielding a specific power of over 131 We/kg or a specific
mass of 7.6 kg/kWe. The corresponding masses for the Brayton and Stirling
systems were about 13,600 kg (Brayton) and over 15,000 kg (Stirling) [36].

2. U.S. Fission Propulsion

The classic nuclear propulsion system involves using a nuclear reactor as a
kind of heat exchanger to heat the propellant (hydrogen) and to expel it
through a nozzle as shown in Fig. 28. From about 1955 to about 1972, the
 COMPARATIVE SPACE POWER BASELINES 559

Fig. 28 Schematic of a generic nuclear rocket. (Courtesy of AEC/NASA.)

United States conducted research on a series of such reactors under the Rover and
NERVA (Nuclear Engine for Rocket Vehicle Applications) programs. Under the
NERVA program, the goal was to develop a nuclear rocket capable of producing
a thrust of 334 kN at a specific impulse of 8.1 km/s, which is almost double that
of the best classical chemical propulsion system [37,38].
Seventeen reactors, one nuclear safety reactor, and two ground experimental
engines were tested with the summary record performances listed in Table 2.

Table 2 Summary of U.S. nuclear rocket performancea

Power (Phoebus 2A) 4100 MWt

Thrust (Phoebus 2A) 930 kN
Hydrogen flow rate (Phoebus 2A) 120 kg/s
Equivalent specific impulse (Pewee) 8.3 km/s
Minimum reactor specific mass (Phoebus 2A) 2.3 kg/MW
Average coolant exit temperature (Pewee) 2550 K
Peak fuel temperature (Pewee) 2750 K
Core average power density (Pewee) 2340 MW/m3
Peak fuel power density (Pewee) 5200 MW/m3
Accumulated time at full power (NF 1) 109 minutes
Greatest number of restarts 28
a
From Refs. 37 and 38.
560 G. L. BENNETT

In theory, a nuclear rocket should have a specific impulse about three times
that of the best state-of-practice chemical rocket (liquid hydrogen liquid
oxygen, LH2 LO2). In practice, materials have limited the nuclear rocket’s
performance to about twice that of the chemical rocket. In an effort to overcome
the material limits, researchers have investigated nuclear rockets that are not so
dependent on materials, that is, the particle bed reactor (PBF), the liquid-core
reactor, the gas-core nuclear rocket (GCNR), and the closed-cycle gas-core
nuclear rocket or “nuclear light bulb” reactor (essentially a reacting 235U
plasma radiating through a transparent medium to the hydrogen propellant). In
theory, the GCNR offers the highest specific impulse (105 m/s) although the
engine mass would be on the order of 200 MT [39,40].
In the early days of the space program, a proposal was made to use fission
bombs as a propulsion system (Project Orion). In this concept, small fission
bombs would be released at the rate of 1 bomb every 1 to 10 s and exploded at
a distance on the order of 30 m to 300 m from the vehicle. The blast would inter-
act with a pusher plate that in turn would transmit the impulse to the vehicle
through a shock attenuation system. The specific impulse was estimated to be
in the range of 18 km/s to 25 km/s with a thrust-to-weight (T/W) of about
4. Approximately 2000 fission bombs were estimated to be required for a 250-
day roundtrip mission to Mars. The program was ended in 1965 after some sub-
scale tests with chemical explosives were conducted [40].
Another approach to realize the benefits of nuclear fission is the fission frag-
ment rocket in which the fission fragments themselves become the propellant
through the use of magnetic fields to guide them. Specific impulses on the
order of 106 m/s would be possible if this concept could be developed [41]. A
number of technology hurdles would have to be overcome including the
control of the fission fragments.
The high power densities achieved in the Rover/NERVA program indicate
what is possible with multi-megawatt nuclear reactors. If these high power den-
sities were coupled to advanced electric propulsion thrusters the solar system
would be open to human exploration. Comparisons of chemical, SEP, NEP,
and NTP for human missions to Mars have shown that NEP is a competitor in
terms of mass that can be delivered [42]. With reactors providing 200 MWe
and specific masses on the order of 1 kg/kWe to 2 kg/kWe, fast trip times
could be achieved. An artist’s illustration of one such concept, Odyssey, is
shown in Fig. 29. The reactor concept would produce 200 MWe and deliver
100 MT of payload. The length of the NEP spacecraft would be over 100 m [43].
A JPL top-level study indicated that the Odyssey spacecraft could make
230-day roundtrip times to Mars and 2-to-3-year roundtrip times to Jupiter.

3. Former Soviet Union Fission Program

The former Soviet Union also had an active space nuclear program, launching
about 31 (and perhaps 33) BUK space nuclear reactors to power spacecraft used
for marine radar observations. The BUK liquid metal-cooled fast reactor pro-
duced about 3 kWe from 100 kWt of thermal power using two-stage thermoelec-
tric elements. Launches of BUK may have begun as early as 1967 with the last
publicly known launch occurring in 1988 [44,45].
 COMPARATIVE SPACE POWER BASELINES 561

Fig. 29 Project Odyssey: a 200 MWe NEP concept for facilitating human explo-
ration of the solar system. (Courtesy of JPL.)

In 1987, the former Soviet Union launched two experimental thermal reactors
designated TOPAZ. These experimental reactors, which used an in-core thermio-
nic conversion system, reportedly produced about 5 kWe from a reactor thermal
power of 150 kWt [44,45]. Both the BUK and TOPAZ space reactor
power systems had masses on the order of 1 MT. In the late 1980s and early
1990s, certain elements of the U.S. Department of Defense showed some interest
in testing and flying another Russian thermionic reactor (ENISEY) that had
not been flown by the Russians [44]. In the United States, this reactor was erro-
neously dubbed “TOPAZ 2” (but it was not built by the Russian group that had
built TOPAZ).
In addition to nuclear reactors, the former Soviet Union has used RHUs and
small RTGs on some missions [45], and has also conducted tests of nuclear
fuel elements for a nuclear rocket.

C. Fusion Power and Propulsion

Nuclear fusion powers the stars and it offers the potential of powering inter-
planetary and, perhaps, interstellar spacecraft. Very light nuclides such as deuter-
ons and tritons have a relatively small binding energy per nucleon compared to
heavy nuclei such as uranium-235 (235U) and plutonium-239 (239Pu). Therefore,
the fusion of light nuclei generally results in the release of a relatively large
amount of energy per nucleon. Table 3 lists some of the more important fusion
reactions together with uranium (235U) fission for comparison.
562 G. L. BENNETT

Table 3 Fusion reactions

Energy release

Reaction Total (MeV) E/m (MeV/amu)

1
H þ 1H ! 2H þ ßþ 1.4 0.7
1
H þ 2H ! 3He þ g 5.5 1.8
1
H þ 7Li ! 4He þ 4He 17.4 2.2
3
He þ n (50%) 3.3 0.8
2
H þ 2H ! 3H þ 1H (50%) 4.0 1.0
2
H þ 3H ! 4He þ n 17.6 3.5
2
H þ 3He ! 4He þ 1H 18.4 3.8
2
H þ 6Li ! 4He þ 4He 22.4 2.8
3
H þ 3H ! 4He þ n þ n 11.3 1.9
3
He þ 3He ! 4He þ 1H þ 1H 12.9 2.1
235
U þ n ! (fission) 200 0.8

Note: 1H ¼ proton; 2H ¼ deuteron; 3H ¼ triton; He ¼ Helium; Li ¼ Lithium; U ¼ Uranium ßþ ¼

positron; g ¼ gamma ray; n ¼ neutron.
From Ref. 46.

The principal issues in making fusion work include using enough energy
(“temperature”) to overcome the electrostatic repulsion of the positive charges
in order to make the reaction happen and to confine the plasma long enough
for the reactions to occur. However, if fusion propulsion is achieved, the benefits
of fusion propulsion would greatly outweigh those of chemical and fission
propulsion.
Many nations have actively pursued fusion power for terrestrial applications.
For space propulsion, there is a very close coupling of the fusion power source
design and the propulsion technology. Classically, two different approaches
have been proposed to confine the plasma long enough to sustain a fusion
reaction: 1) inertial confinement fusion (ICF) and 2) magnetic confinement
fusion (MCF).
ICF utilizes a high-power source (e.g., lasers) to compress and heat a pellet of
fusion fuel to fusion ignition conditions. Figure 30 illustrates the ICF propulsion
concept. In the 1970s, the British Interplanetary Society investigated an interstel-
lar ICF concept under their Project Daedalus study. In the early 1990s, the JPL,
the Lawrence Livermore National Laboratory (LLNL), the Energy Technology
Engineering Center (ETEC), and NASA’s Johnson Space Center (JSC) studied
an ICF propulsion concept termed VISTA (vehicle for interplanetary space trans-
port applications). The VISTA vehicle was sized for a fast (100-day roundtrip)
human mission to Mars. The payload was to be 100 MT for a total mass of
5800 MT, of which 4100 MT would be hydrogen expellant and 40 MT would
be deuterium-tritium (d-t) fuel. A jet power of 20,000 MW was calculated for
30 Hz operation (30 d-t pellets ignited per second) with a specific impulse of
170 km/s [40].
MCF (shown diagrammatically in Fig. 31) uses strong magnetic fields to
confine the fusion plasma. In principle, MCF should provide similar performance
 COMPARATIVE SPACE POWER BASELINES 563

Fig. 30 Basic features of a fusion propulsion system using inertial confinement

fusion. (Courtesy of JPL.)

Fig. 31 Basic features of a fusion propulsion system using magnetic confinement

fusion. (Courtesy of JPL.)
564 G. L. BENNETT

to the ICF propulsion concepts, although the use of large magnets may make
MCF devices heavier than ICF devices.

D. Antimatter Power and Propulsion

As noted in Section I and Figs. 3 and 4, matter antimatter annihilation offers
the highest possible classical physical specific energy of any known reaction sub-
stance. The matter antimatter annihilation specific energy of 9  1016 J/kg is
many orders of magnitude greater than chemical (1  107 J/kg), fission
(8  1013 J/kg), or even fusion (3  1014 J/kg) reaction energies. Moreover,
matter antimatter annihilation can be induced spontaneously without the use
of complex reactor systems.
While no antimatter power sources have been built, studies have shown that
the design of the antimatter “power source” is strongly coupled to how it
would be used in propulsion. Historically, four basic antimatter thruster concepts
have been considered (Fig. 32).
Most antimatter power and propulsion studies have assumed the use of anti-
protons or neutral anti-hydrogen that would be produced in a terrestrial accelera-
tor. Because antimatter annihilates spontaneously when it comes in contact with
matter, the antimatter would have to be contained in an electromagnetic field and
stored separately in a high vacuum.
One Monte Carlo analysis of the four concepts shown in Fig. 32 found the effi-
ciencies for conversion of annihilation energy to energy available for propulsion
“to be approximately 75% for the solid core concept, 13% for the gas core
concept, 3% or less for the plasma core concept and 20% or less for the beam
core concept.” This study also observed that “the gas core, plasma core and

Fig. 32 Four basic antimatter propulsion concepts. (Courtesy of JPL.)

 COMPARATIVE SPACE POWER BASELINES 565

beam core concepts all suffered from additional plasma radiation losses which
are severe. The solid core concept, because of its higher density core, experiences
a substantial neutron production problem from nuclear charge-exchange” [47].
A separate analysis of a pulsed, antiproton-powered magnetically confined
hydrogen plasma engine generally found energy transfer efficiencies below
1% [48].
As a result of such studies, advocates have focused on alternative concepts to
raise the efficiency of conversion.

IV. Conclusions
In this chapter we briefly covered a number of non-nuclear and nuclear space
power sources that either are available now (solar, batteries, fuel cells, radioiso-
topes) or could be available in the future (fission, fusion, antimatter). Particular
attention was paid to radioisotope and fission power sources because these
have been used in a number of space missions and are most directly applicable
to missions to the outer solar system and beyond. The development of fusion
and antimatter power and propulsion systems would greatly facilitate these
explorations. From a “classical” standpoint, antimatter represents the “ultimate”
source of power for space exploration.
In terms of classical power sources for interplanetary and early interstellar
travel, one can summarize the foregoing sections with the following projected
or “ultimate” performance metrics as described in the parenthetical comments):
Photovoltaic: 1000 We/kg (future achievable at 1 AU with thin films)
Energy storage: 1000 We-h/kg (future goal)
Radioisotope:  10 We/kg (future)
Fission:  1000 We/kg (future at 100 s of MWe)
Fusion:  1000 We/kg (future)
Antimatter: 9  1016 J/kg (ultimate)

References
[1] Brandhorst, H. W., Chetty, P. R. K., Doherty, M. J., and Bennett, G. L., “Technol
ogies for Spacecraft Electric Power Systems,” Journal of Propulsion and Power,
Vol. 12, No. 5, Sept. Oct. 1996, pp. 819 827.
[2] Bennett, G. L., and Stone, J. R., “The NASA Advanced Propulsion Concepts
Program,” 1989 JANNAF (Joint Army Navy NASA Air Force) Propulsion Meeting,
Cleveland, OH, 23 25 May 1989.
[3] Surampudi, R., Hamilton, T., Rapp, D., Stella, P., Mardesich, N., Mondt, J., Nesmith,
B., Bunker, R. L., Cutts, J., Bailey, S. G., Curtis, H. B., Piszczor, M., Gaddy, E.,
Marvin, D., and Kazmerzki, L., Solar Cell and Array Technology for Future
Space Science Missions, JPL D 24454, Jet Propulsion Laboratory, Pasadena, CA,
Aug. 2002.
[4] Kurland, R. M., and Stella, P. M., “The Advanced Photovoltaic Solar Array Program
Update,” Proceedings of the 3rd European Space Power Conference (Graz, Austria),
European Space Agency Publication ESA WPP 054, Paris, France, Aug. 1993.
566 G. L. BENNETT

[5] Bennett, G. L., “The OAST Space Power Program,” Space Photovoltaic Research
and Technology 1989, NASA Conf. Publication 3107, NASA Lewis Research
Center, Cleveland, OH, 7 9 Nov. 1989, published 1991.
[6] Mondt, J., Halpert, G., Rapp, D., Surampudi, S., Peterson, C., Frank, H., Burke, K.,
Manzo, M., Wolff, F., Bragg, B., Rao, G., Vukson, S., Marsh, R., Plichta, E., Sutton,
R., Browning, V., Methlie, G., Hwang, W., and Savinall, R., Energy Storage Tech
nology for Future Space Science Missions, JPL D 30268, Jet Propulsion Laboratory,
Pasadena CA, Nov. 2004.
[7] Corliss, W. R., and Harvey, D. G., Radioisotopic Power Generation, Prentice Hall,
Englewood Cliffs, NJ, 1964.
[8] Corliss, W. R., and Mead, R. L., Power from Radioisotopes, U.S. Atomic Energy
Commission, No. 74 169081, 1971 (rev.).
[9] Bennett, G. L., “Survivability Considerations in the Design of Space Power
Systems,” Proceedings of the 23rd Intersociety Energy Conversion Engineering
Conference, Paper No. 889159, Denver, CO, 31 July 5 Aug. 1988.
[10] Bennett, G. L., Lombardo, J. J., and Rock, B. J., “U.S. Radioisotope Thermoelectric
Generator Space Operating Experience” (June 1961 Dec. 1982), Proceedings of the
18th Intersociety Energy Conversion Engineering Conference, Orlando, FL, 21 26
Aug. 1983; reprinted in The Nuclear Engineer, Vol. 25, No. 2, March/April 1984,
pp. 49 58.
[11] Bennett, G. L., and Skrabek, E. A., “Power Performance of U.S. Radioisotope
Thermoelectric Generators,” Proceedings ICT ’96, 15th International Conference
on Thermoelectrics, Pasadena, CA, 26 29 March 1996, IEEE No. 96TH8169,
Library of Congress No. 96 75531, pp. 357 372.
[12] Bennett, G. L., “Space Nuclear Power,” Encyclopedia of Physical Science and Tech
nology, 3rd ed., Vol. 15, Academic Press, New York, 2002, pp. 537 553.
[13] Wyatt, T., “The Gestation of Transit as Perceived by One Participant,” Johns
Hopkins APL Technical Digest, Vol. 2, No. 1, Jan. March 1981, pp. 32 38.
[14] Johns Hopkins University Applied Physics Laboratory, Artificial Earth Satellite
Designed and Fabricated by the Johns Hopkins University Applied Physics Labora
tory, JHU/APL Report SDO 1600 (rev.), Aug. 1980.
[15] Dick, P. J., and Davis, R. E., “Radioisotope Power System Operation in the Transit
Satellite,” AIEE Summer General Meeting, No. CP 62 1173, Denver, CO, 1962,
pp. 17 22.
[16] Dick, P. J., SNAP 9A Quarterly Progress Report No. 2, Radioisotope Fueled Ther
moelectric Power Conversion System Development, 1 Dec. 1961 Feb. 28 1962,
MND P 2700 2.
[17] Bradshaw, G. B., and Postula, F. D., “Beginning of Mission Flight Data on the
TRANSIT RTG,” Proceedings of the 8th Intersociety Energy Conversion Engineer
ing Conference, Paper 739091, Philadelphia, PA, 13 16 Aug. 1973.
[18] Fihelly, A. W., and Baxter, C. F., “Orbital Performance of the SNAP 19 Radioisotopic
Thermoelectric Generator Experiment,” Proceedings of the 6th Intersociety Energy
Conversion Engineering Conference, Paper 719152, Boston, MA, 3 5 Aug. 1971.
[19] Jaffe, H., and O’Riordan, P., “Isotope Power Systems for Unmanned Spacecraft
Applications,” Proceedings of the 7th Intersociety Energy Conversion Engineering
Conference, Paper 729088, San Diego, CA, 25 29 Sept. 1972.
[20] Bates, J. R., Lauderdale, W. W., and Kernaghan, H., ALSEP Termination Report,
NASA Ref. Publication 1036, 1979.
 COMPARATIVE SPACE POWER BASELINES 567

[21] Pitrolo, A. A., Rock, B. J., Remini, W. C., and Leonard, J. A., “SNAP 27 Program
Review,” Proceedings of the 4th Intersociety Energy Conversion Engineering Con
ference, Washington, DC, 22 26 Sept. 1969.
[22] Ward, W. W., “Developing, Testing, and Operating Lincoln Experimental Satellites
8 and 9 (LES 8/9),” M.I.T. Lincoln Laboratory Tech. Note 1979 3, 16 Jan. 1979.
[23] Kelly, C. E., “The MHW Converter (RTG),” Record of the 10th Intersociety Energy
Conversion Engineering Conference, Paper 759132, Newark, DE, 18 22 Aug. 1975.
[24] Ward, W. W., and Floyd, F. W., “Thirty Years of Research and Development in
Space Communications at Lincoln Laboratory,” The Lincoln Laboratory Journal,
Vol. 2, No. 1, Spring 1989.
[25] Brittain, W. M., and Skrabek, E. A., “SNAP 19 RTG Performance Update for the
Pioneer and Viking Missions,” Proceedings of the 18th Intersociety Energy Conver
sion Engineering Conference, Orlando, FL, 21 26 Aug. 1983.
[26] Brittain, W. M., “SNAP 19 Viking RTG Mission Performance,” Proceedings of the
11th Intersociety Energy Conversion Engineering Conference, Paper 769255, State
Line, NV, 12 17 Sept. 1976.
[27] Bennett, G. L., Hemler, R. J., and Schock, A., “Development and Use of the Galileo
and Ulysses Power Sources,” Space Technology, Vol. 15, No. 3, 1995, pp. 157 174;
originally presented as Paper IAF 94 R 1 362, 45th Congress of the International
Astronautical Federation, Jerusalem, Israel, 9 14 Oct. 1994.
[28] Lockheed Martin, GPHS RTGs in Support of the Cassini RTG Program, Final Tech
nical Report, 11 Jan. 1991 30 April 1998, Lockheed Martin Astronautics, Doc. No.
RR18, Valley Forge Operations, Aug. 1998.
[29] Ulysses objectives. URL: http://sci.esa.int/science e/www/object/index.cfm?
fobjectid ¼ 30990 [cited on 20 Jan. 2008].
[30] Ottman, G. K., and Hersman, C. B., “The Pluto New Horizons RTG and
Power System Early Mission Performance,” 4th International Energy Conversion
Engineering Conference, AIAA Paper 2006 4029, San Diego, CA, 26 29 June
2006.
[31] Cockfield, R. D., “Preparation of RTG F8 for the Pluto New Horizons Mission,” 4th
International Energy Conversion Engineering Conference, AIAA Paper 2006 4031,
San Diego, CA, 26 29 June 2006.
[32] Staub, D. W., SNAP 10A Summary Report, NAA SR 12073, Atomics International,
Canoga Park, CA, 25 March 1967.
[33] Josloff, A. T., Shepard, N. F., Chan, T. S., Greenwood, F. C., Deane, N. A., Stephen,
J. D., and Murata, R. E., “SP 100 Generic Flight System Design and Early Flight
Options,” Proceedings of the 11th Symposium on Space Nuclear Power and
Propulsion, American Institute of Physics, CP 301, Pt. 2, Woodbury, NY, 1994,
pp. 533 538.
[34] Josloff, A. T., and Mondt, J. F., “SP 100 Space Reactor Power System Readiness to
Support Emerging Missions,” Proceedings of the 11th Symposium on Space Nuclear
Power and Propulsion, American Institute of Physics, CP 301, Pt. 2, Woodbury, NY,
1994, pp. 539 545.
[35] Nainiger, J. J., and Mason, L. S., “Nuclear Reactor Power Systems for Lunar Base
Applications,” NTSE 92, Nuclear Technologies for Space Exploration, Proceedings
of the American Nuclear Society Topical Meeting on Nuclear Technologies for
Space Exploration, 16 19 Aug. 1991, Jackson, WY, American Nuclear Society
Order No. 700177, La Grange, IL, pp. 247 256.
568 G. L. BENNETT

[36] Cropp, L. O., Gallup, D. R., and Marshall, A. C., “Mass and Performance Estimates
for 5 to 1000 kW(e) Nuclear Reactor Power Systems for Space Applications,” Space
Power, Vol. 10, No. 1, 1991, pp. 43 78.
[37] Black, D. L., and Gunn, S. V., “Space Nuclear Propulsion,” Encyclopedia of Physical
Science and Technology, 3rd ed., Vol. 15, Academic Press, New York, 2002,
pp. 555 575.
[38] Bennett, G. L., Finger, H. B., Miller, T. J., Robbins, W. H., and Klein, M., “Prelude to
the Future: A Brief History of Nuclear Thermal Propulsion in the United States,”
A Critical Review of Space Nuclear Power and Propulsion, 1984 1993, El Genk,
M. S. (ed.), American Institute of Physics, New York, 1994, pp. 221 267.
[39] Ragsdale, R., “Open Cycle Gas Core Nuclear Rockets,” Nuclear Thermal
Propulsion, A Joint NASA/DOE/DOE Workshop, Proceedings of the Nuclear
Thermal Propulsion Workshop, Strongsville, OH, NASA Lewis Research
Center, Cleveland, OH, NASA Conference Publication 10079, 10 12 July 1990,
pp. 343 357.
[40] Burke, K. A. (ed.), Advanced Propulsion Concepts, Jet Propulsion Laboratory,
Jan. 1989.
[41] Schnitzler, B. G., Jones, J. L., and Chapline, G. F., Fission Fragment Rocket Scien
tific Feasibility Assessment, Idaho National Engineering Laboratory, EGG NERD
8585, June 1989.
[42] Braun, R. D., and Blersch, D. J., “Propulsive Options for a Manned Mars Transpor
tation System,” Journal of Spacecraft, Vol. 28, No. 1, Jan. Feb. 1991, pp. 85 92.
[43] Sercel, J., and Sargent, M., “Odyssey: A Vision of Human Exploration,” NASA
Headquarters presentation, Jet Propulsion Laboratory, Pasadena, CA, 2 Feb. 1989.
[44] Ponomarev Stepnoi, N. N., Talyzin, V. M., and Usov, V. A., “Russian Space Nuclear
Power and Nuclear Thermal Propulsion Systems,” Nuclear News, Vol. 43, No. 13,
pp. 33 46, Dec. 2000.
[45] Bennett, G. L., “A Look at the Soviet Space Nuclear Power Program,” Proceedings
of the 24th Intersociety Energy Conversion Engineering Conference, Paper 899009,
Crystal City, VA, 7 11 Aug. 1989.
[46] Hoisington, D. B., Nucleonics Fundamentals, McGraw Hill, New York, 1959.
[47] Callas, J. L., The Application of Monte Carlo Modeling to Matter Antimatter
Annihilation Propulsion Concepts, Jet Propulsion Laboratory, JPL D 6830, CA,
1 Oct., 1989.
[48] LaPointe, M. R., “Antiproton Powered Propulsion with Magnetically Confined
Plasma Engines,” Journal of Power and Propulsion, Vol. 7, No. 5, 1991, pp.
749 759; based on LaPointe, M. R., “Antiproton Annihilation Propulsion with Mag
netically Confined Plasma Engines,” Ph.D. Dissertation, Department of Chemical
and Nuclear Engineering, The University of New Mexico, Albuquerque, NM,
May 1989.
 Chapter 18

On Extracting Energy from the

Quantum Vacuum

Eric W. Davis and H. E. Puthoff†

Institute for Advanced Studies at Austin, Austin, Texas

I. Introduction
UANTUM theory predicts that the vacuum of space throughout the universe
Q is filled with electromagnetic waves, random in phase and amplitude, propa-
gating in all possible directions, and with a cubic frequency distribution. This
differs from the cosmic microwave background radiation, and is referred to as
the electromagnetic quantum vacuum, which is the lowest energy state of other-
wise empty space. When integrated over all frequency modes up to the Planck
frequency, nP (1043 Hz), it represents an energy density of as much as
10113 J/m3, which is far in excess of any other known energy source, although
only an infinitesimal fraction of it is accessible. Even if we are constrained to
integrate over all frequency modes only up to the nucleon Compton frequency
(1023 Hz, the characteristic frequency associated with the size of nucleons),
this energy density is still enormous (1035 J/m3). In addition, the electromag-
netic quantum vacuum is not alone; it intimately couples to the charged particles
in the Dirac sea of virtual particle antiparticle pairs, and thereby couples to the
other interactions inherent in the Standard Model (weak and strong force vacua).
Therefore, all the numbers just mentioned are subject to further refinement.
However, it should be noted that we can safely ignore any coupling of the
quantum electromagnetic vacuum to the quantum chromodynamic (QCD)
vacuum in the context of this chapter because the latter coexists in two phases:
1) the ordinary vacuum exterior to the hadron, which is impenetrable to quark
color, and 2) the vacuum interior of the hadron in which the Yang Mills fields
that carry color (gluons) propagate freely. Both vacuum phases are separated

Copyright # 2008 by the American Institute of Aeronautics and Astronautics, Inc. All rights
reserved.

Senior Research Physicist.
†
Director.

569
570 E. W. DAVIS AND H. E. PUTHOFF

by a boundary at the surface of the hadron on which the Yang Mills and quark
fields satisfy boundary conditions.
Even though this zero-point field (ZPF) energy seems to be an inescapable
consequence of quantum field theory, its energy density is so enormous as to
make it difficult to reconcile. Instead, many quantum calculations subtract
away the ZPF energy by ad hoc means (e.g., renormalization). However, we
observe the effects of the quantum vacuum ZPF that are responsible for a
variety of well-known physical effects, such as:
1) Lamb shift
2) Spontaneous atomic emission
3) Low-temperature van der Waals forces
4) Casimir effect
5) Source of photon shot and fluctuating radiation-pressure noise in lasers
6) Astronomically observed cosmological constant (a.k.a. dark energy, a
form of Casimir energy according to the Schwinger DeWitt quantum ether
prescription [1 4])

Rather than eliminate the ZPF energy (a.k.a. ZPE) from the equations, there is
much left to be learned by exploring the possibility that it is a real energy. From
this perspective, the ordinary world of matter and energy is like foam atop the
quantum vacuum sea. If the ZPF is real, then there is the possibility that it can
be tapped as a source of power or harnessed to generate a propulsive force for
space travel. This notion, of exchanging energy with the quantum vacuum, is
the focus of this chapter.
The propeller or the jet engine of an aircraft can push air backward to propel
the aircraft forward. A ship or boat propeller does the same thing in water. On
Earth there is air or water to push against. But a rocket in space has no material
medium to push against, and so it needs to carry propellant to eject in order to
provide momentum. A deep space rocket must start out with all the propellant
it will ever require, and this quickly results in the need to carry additional propel-
lant just to propel the propellant. The breakthrough one wishes to achieve in deep
space travel is to eliminate the need to carry propellant at all. How can one
generate a propulsive force without carrying and ejecting propellant?

II. Early Concepts for Extracting Energy and

Thermodynamic Considerations
The Casimir force is a force associated with the electromagnetic quantum
vacuum [5]. This force is an attraction between parallel uncharged metallic
plates that has now been well measured and can be attributed to a minute imbal-
ance in the ZPE density inside the cavity between the plates versus the region
outside the plates (Fig. 1) [6 8]. As shown in the figure, the vacuum is full of
virtual photons, but photons with wavelengths, l, more than twice the plate sep-
aration, d, are excluded from the space between them, which causes the imbal-
ance that pushes the plates together. However, this is not useful for propulsion
because it symmetrically pulls on the plates. If some asymmetric variation of
the Casimir force could be identified, then one could in effect sail through
space as if propelled by a kind of quantum fluctuation wind. This specific
notion is explored in Chapter 12.
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 571

Fig. 1 Schematic of Casimir effect cavity.

The other requirement for space travel is energy. It is sometimes assumed that
attempting to extract energy from the vacuum ZPF would somehow violate
the laws of thermodynamics. Fortunately, it turns out that this is not the case.
A thought experiment published by Forward [9,10] demonstrated how the
Casimir force could, in principle, be used to extract energy from the vacuum
ZPF. Forward showed that any pair of conducting plates at close distance experi-
ences an attractive Casimir force that is due to the electromagnetic ZPF of the
vacuum. A “vacuum-fluctuation battery” can be constructed by using the
Casimir force to do work on a stack of charged conducting plates, as shown in
Fig. 2. By applying a charge of the same polarity to each conducting plate, a
repulsive electrostatic force will be produced that opposes the Casimir force. If
the applied electrostatic force is adjusted to be always slightly less than the

Fig. 2 Vacuum-fluctuation battery. (From Ref. 9.)

572 E. W. DAVIS AND H. E. PUTHOFF

Casimir force, the plates will move toward each other and the Casimir force will
add energy to the electric field between the plates. The battery can be recharged
by making the electrical force slightly stronger than the Casimir force to re-
expand the foliated conductor.
Cole and Puthoff [11] verified that (generic) energy extraction schemes are not
contradictory to the laws of thermodynamics. For thermodynamically reversible
processes, no heat will flow at temperature T 0. However, for thermodynami-
cally irreversible processes, heat can be produced and made to flow, either at
T 0 or at any other T . 0 situation, such as by taking a system out of mechan-
ical equilibrium. Moreover, work can be done by or on physical systems, either at
T 0 or T . 0 situations, whether for a reversible or irreversible process.
However, if one is considering a net cyclical process on the basis of, say, the
Casimir effect, then energy would not be able to be continually extracted
without a violation of the Second Law of Thermodynamics. Thus, Forward’s
process cannot be cycled to yield a continuous extraction of energy. Here, the
recharging of the battery would, owing to frictional and other losses, require
more energy than is gained from the ZPF. There is no useful engine cycle in
this process; nonetheless, the plate-contraction phase of the cycle does demon-
strate the ability to cause “extraction” of energy from the ZPF. It does reflect
work done by the ZPF on matter.
Another illustrative example of an early scheme for extracting energy from the
ZPF is described in a patent by Mead and Nachamkin [12]. They propose that a
set of resonant dielectric spheres be used to extract energy from the ZPF and
convert it into electrical power. They consider the use of resonant dielectric
spheres, slightly detuned from each other, to provide a beat-frequency downshift
of the more energetic high-frequency components of the ZPF to a more easily
captured form. Figure 3 shows two embodiments of the invention. The device
includes a pair of dielectric structures (items 12, 14, 112, and 114 in the
figure) that are positioned proximal to each other and which intercept incident
ZPE radiation (items 16 and 116). The volumetric sizes of the structures are
selected so that they resonate at a particular frequency of the incident radiation.
But the volumetric sizes of the structures are chosen to be slightly different, so
that the secondary radiations emitted from them (items 18, 20, 24, 118, 120,
and 124) at resonance interfere with each other, thus producing a beat-frequency
radiation that is at a much lower frequency than that of the incident radiation, and
can be converted into electrical energy. A conventional metallic antenna (loop or
dipole type, or a RF cavity structure; items 22 and 122 in the figure) can then be
used to collect the beat-frequency radiation. This radiation is next transmitted
from the antenna to a converter via an electrical conductor or waveguide
(items 26 and 126) and converted to electrical energy. The converter must
include: 1) a tuning circuit or comparable device so that it can effectively
receive the beat frequency radiation, 2) a transformer to convert the energy to
electrical current having a desired voltage, and 3) a rectifier to convert the
energy to electrical current having a desired waveform (items 28, 30, 32, 34,
128, 130, and 132 in the figure).
The receiving structures are composed of dielectric material in order to diffract
and scatter the incident ZPE radiation. The volumetric sizing requirements for the
receiving structures are selected to enable them to resonate at a high frequency
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 573

Fig. 3 ZPE resonant dielectric spheres electrical power generator. (From Ref. 12.)

corresponding to the incident ZPE radiation, based on the parameters of frequency

of the incident ZPE radiation, and the propagation characteristics of the medium
(vacuum or otherwise) and the receiving structures. Because the ZPE radiation
energy density increases with increasing frequency, greater amounts of electro-
magnetic energy are potentially available at higher frequencies. Consequently,
the size of the receiving structures must be miniaturized in order to produce
greater amounts of energy from a system located within a space or volume of a
given size. Therefore, the smaller the size of the receiving structures, the greater
the amount of energy that can in principle be produced by the system.
Although a computer model study performed at the Air Force Research
Laboratory (Edwards Air Force Base, California) indicates that the invention
could work, no experimental study has been performed to validate this in the
lab (F. B. Mead, private communication, 2002). Regarding critiques, it is not
clear how the beat frequency can be picked up by the receiving loop antenna.
There is no nonlinear method in the invention showing that an electromagnetic
beat frequency can be generated and coupled to the loop. Without a nonlinear
coupling method there will be no sidebands, one of which would be frequency
downshifted and called the beat frequency. The coupling method requires the
generation of sidebands in the mixing of two different frequencies via a nonlinear
technique. However, an easy resolution to this potential deficiency is that the
resonant dielectric spheres could be constructed of a nonlinear dielectric material.
Although several novel ZPF energy extraction mechanisms have been pro-
posed in the popular and technical literature, no practicable technique has been
successfully demonstrated in the laboratory. To better understand how ZPE
extraction methods might work, it is necessary to characterize the physics of
574 E. W. DAVIS AND H. E. PUTHOFF

the ZPF and proposed energy extraction techniques, and to evaluate their
feasibility for application to space power and propulsion systems. In what
follows, we summarize the physics of the ZPF and the experimental investi-
gations being pursued to address the question of extracting energy from the
quantum vacuum.

III. Origin of Zero-Point Field Energy

A. Quantum Electrodynamics Theory
The basis of the ZPF is typically attributed to the Heisenberg uncertainty
principle. According to this principle, A and B are any two conjugate observables
that we are interested in measuring, and they obey the commutation relation
[A, B] ih  (i is the unit complex number and h is Planck’s reduced constant,
1.055  10 34 J . s). Their corresponding uncertainty relation is DADB  h/2,
where DA is the variance (a.k.a. uncertainty) of observable A and DB is that of
the conjugate observable B. This relation states that if one measures observable
A with very high precision (i.e., its uncertainty DA is very small), then a simul-
taneous measurement of observable B will be less precise (i.e., its uncertainty DB
is very large), and vice versa. In other words, it is not possible to simultaneously
measure two conjugate observable quantities with infinite precision. This
minimum uncertainty is not due to any correctable flaws in measurement, but
rather reflects the intrinsic fuzziness in the quantum nature of energy and
matter. Substantial theoretical and experimental work has shown that in many
quantum systems the limits to measurement precision is imposed by the
quantum vacuum ZPF embodied within the uncertainty principle. Nowadays
we rather see the Heisenberg uncertainty principle as a necessary consequence,
and therefore a derived result of the wave nature of quantum phenomena. The
uncertainties are just a consequence of the Fourier nature of conjugate pairs of
quantities (observables). For example, the two Fourier wave conjugates, time
and frequency, become the pair of quantum-particle conjugates, time and
energy; the two Fourier wave conjugates, displacement and wavenumber,
become the pair of quantum-particle conjugates, position and momentum. (For
more on this see, e.g., Ref. 13.)
Radio and microwaves, infrared light, visible light, ultraviolet light, X-rays,
and gamma rays are all forms of electromagnetic radiation. Classically, electro-
magnetic radiation can be pictured as waves flowing through space at the speed of
light. The waves are not waves of anything substantive, but are in fact ripples in
the state of a field. These waves carry energy, and each wave has a specific direc-
tion, frequency, and polarization state. This is called a “propagating mode of the
electromagnetic field.” A useful tool for modeling the propagating mode of the
electromagnetic field in quantum mechanics is the ideal quantum mechanical har-
monic oscillator, a hypothetical charged mass on a perfect spring oscillating back
and forth under the action of the spring’s restoring force. The Heisenberg uncer-
tainty principle dictates that a quantized harmonic oscillator (a.k.a. a photon
state) can never come entirely to rest, because that would be a state of exactly
zero energy, which is forbidden by the commutation relation outlined above.
Instead, every mode of the field has h v/2 (v is the mode or photon frequency
and hv is the energy of a single mode or photon) as its average minimum
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 575

energy in the vacuum. (This is a small amount of energy, but the number of
modes is enormous, and indeed increases as the square of the frequency. The
product of this minuscule energy per mode, multiplied by the huge spatial
density of modes, yields a very high theoretical energy density per unit
volume.) This ZPE term is added to the classical blackbody spectral radiation
energy density r(v)dv (i.e., the energy per unit volume of radiation in the
frequency interval (v, v þ dv)) [14]:

 hv hv
v2
r(v)d v ¼ þ dv
p2 c3 expðh v=kTÞ  1 2
hv3  
hv
¼ 2 3 coth dv (1)
2p c 2kT

where c is the speed of light (3.0 108 m/s), k is Boltzmann’s constant (1.3807 
10 23 J/K), T is the absolute temperature, and v 2pn is the angular frequency.
The factor outside the square brackets in the first line of Eq. (1) is the density of
mode (or photon) states (i.e., the number of states per unit frequency interval per
unit volume); the first term inside the square brackets is the standard Planck
blackbody radiation energy per mode; and the second term inside the square
brackets is the quantum zero-point energy per mode. Equation (1) is called the
Zero Point Planck (ZPP) spectral radiation energy density. Planck first added
the ZPE term to the classical blackbody spectral radiation energy density in
1912, although it was Einstein, Hopf, and Stern who actually recognized the
physical significance of this term in 1913 [14]. Direct spectroscopic evidence
for the reality of ZPE was provided by Mulliken’s boron monoxide spectral
band experiments in 1924, several months before Heisenberg first derived
the ZPE for a harmonic oscillator from his new quantum matrix mechanics
theory [15].
Following this line of reasoning, quantum physics predicts that all of space
must be filled with electromagnetic zero-point fluctuations (a.k.a. the
zero-point field or ZPF) creating a universal sea of ZPE. The density of this
energy depends critically on where the frequency of the zero-point fluctuations
ceases. Because space itself is currently thought to break up into a kind of
“quantum foam” at the Planck length, ‘P (10 35 m), it is argued that the ZPF
must cease at the corresponding nP. If true, then the ZPE density would be
10113 J/m3, 108 orders of magnitude greater than the radiant energy at the
center of the sun! Formally, in quantum electrodynamics (QED) theory, the
ZPE energy density is taken as infinite; however, arguments based on quantum
gravity considerations yield a finite cutoff at nP. Therefore, the spectral energy
density is given by r(v)dv (h v3/2p 2c 3)dv, which integrates to an energy
density, rE h  nP4/8p 2c 3  10113 J/m3. As large as the ZPE is, interactions with
it are typically cut off at lower frequencies depending on the particle coupling
constants or their structure. Nevertheless, the potential ZPF energy density predicted
by quantum physics is enormous.
576 E. W. DAVIS AND H. E. PUTHOFF

B. Stochastic Electrodynamics Theory

An alternative to QED, stochastic electrodynamics (SED) identifies the origin
of the ZPF as a direct consequence of a classical ZPF background. SED begins
with the ordinary classical electrodynamics of Maxwell and Lorentz. However,
instead of assuming the traditional homogeneous solution of the source-free
differential wave equations for the electromagnetic potentials, it instead con-
siders that due to multiple charged particles moving throughout the universe,
there is always a random electromagnetic radiation background present that
affects the particle(s) in any experiment. This new boundary condition
(random radiation background) replaces the prior null background of traditional
classical electrodynamics. Moreover, the principle of relativity dictates that iden-
tical experiments performed in different inertial frames must yield the same
result, and that this random classical electromagnetic radiation must be isotropic
in all inertial frames; it is invariant under scattering by a dipole oscillator, invar-
iant under redshift (Doppler, cosmological, gravitational, no Einstein Hopf drag
force), and must therefore have a Lorentz-invariant energy density spectrum. The
only energy density spectrum that obeys such conditions is one that is pro-
portional to the cubic power of the frequency. Interestingly, this is exactly the
same frequency dependence as that of the QED spectral ZPF energy density
described above, when the temperature T is set to zero in Eq. (1). Thus in
SED, the random radiation assumes the role of the ZPE of QED, and is termed
the classical electromagnetic ZPE. Planck’s constant appears then in SED as
an adjustable parameter that sets the scale of the ZPE spectral density.
The formulation of the SED model has evolved over time, beginning with the
work of Nernst in 1916 and the later foundational work of Marshall and Boyer in
the 1960s [14]. The original standard SED model was based on random phases
with fixed electric-field mode amplitudes. The more recent modified SED
model employs random phases with random electric-field mode amplitudes
and a full probability distribution for the ground state amplitude, in agreement
with quantum theory [16]. A comparison of SED with quantum theory shows
that the first and second moments of the spectral energy distribution are identical,
but beyond that, the distributions diverge widely. Nevertheless, several quantum
theory results have been reproduced by means of the SED approach, such
as [14,17]:
1) Quantum mechanical harmonic oscillator
2) Lamb shift
3) Blackbody radiation
4) van der Waals forces
5) Casimir forces
6) Diamagnetism
7) Davies Unruh effect

The strength of the SED model is that it is heuristically appealing, with trans-
parent derivations, and it is applicable to linear systems. SED calculations have
also been shown to be in one-to-one correspondence with the expectation values
of the Heisenberg quantum equations of motion for linear systems. Both SED and
QED will play a role in the discussions to follow.
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 577

IV. Review of Selected Experiments

In what follows, we outline each of the proposed experimental concepts that
were selected for theoretical and laboratory investigation. A subset of our pro-
posed concepts has undergone preliminary evaluation by Lockheed Martin
review panels involving both internal R&D personnel and outside experts on
theory and experimentation (V. Teofilo, private communication, 2005).

A. Voltage Fluctuations in Coils Induced by ZPF at High Frequency

In a series of experiments, Koch et al. [18 20] measured voltage fluctuations in
resistive wire circuits that are induced by the ZPF. Their result is striking
corroboration of the reality of the ZPF and proves that the ZPF can do real work
(cause measurable currents). Although the Koch et al. experiment detected minu-
scule amounts of ZPF energy, it shows the principle of ZPF energy circuitry to
detect vacuum fluctuations and opens the door to consideration of means
to extract useful amounts of energy. The secondary consequences on other phenom-
ena, if energy can be successfully extracted, have not yet been investigated.
Blanco et al. [21] have proposed a method for enhancing the ZPF-induced
voltage fluctuations in circuits. Theoretically treating a coil of wire as an
antenna, they argue that the antenna-like radiation resistance of the coil should
be included in the total resistance of the circuit, and suggest that this total
resistance should be used in the theoretical computation of ZPF-induced
voltage fluctuations. Because of the strong dependence of the radiation resistance
on the number of coil turns (quadratic scaling), coil radius (quartic scaling), and
frequency (quartic scaling), any enhanced ZPF-induced voltage fluctuations
should be measurable in the laboratory at readily accessible frequencies
(100 MHz compared to the 100 GHz range necessary in the Koch et al.
experiments).
In the theory of Blanco et al., random voltage fluctuations are conveniently
described by their frequency spectrum. That is, given a sufficient time interval
of measured voltages, the measurements are Fourier transformed to the frequency
domain to determine how the voltage fluctuations are distributed (e.g., quantity of
low-frequency, long-duration fluctuations relative to high-frequency, short-
duration fluctuations). Theoretically, the spectrum of voltage fluctuations,
S(v, T ), of a resistive circuit is given by [21]:
 
Rðv; TÞ hv hv
Sðv; TÞ ¼ coth (2)
p 2 2kT

where R(v, T) is the total resistance (ohmic plus radiative), v is the (angular)
frequency, and T is the absolute temperature. The resistance R(v, T) is tempera-
ture dependent through its ohmic contribution (the radiation resistance depends
only on frequency). Note the similar hyperbolic cotangent functions appearing
in Eq. (2) and in the second line of Eq. (1). The postulate of Blanco et al. is
that the total resistance must include the radiation resistance of the circuit [21]:

R(v; T) ¼ Rohmic (v; T) þ Rrad (v) (3)

578 E. W. DAVIS AND H. E. PUTHOFF

Fig. 4 Theoretical voltage spectral density of a tungsten coil.

Under the assumption that the wavelengths of the ZPF modes of interest are
larger than the dimensions of the circuit, the radiation resistance of a coil is
given by [21]:

2 p 2 N 2 av4
Rrad (v) ¼ (4)
3 c c

where N is the number of coil turns, and a is the radius of the coil winding.
According to Blanco et al., large enhancements in ZPF-induced voltage
fluctuations are possible. By reducing the temperature to minimize ohmic resist-
ance, making the coil of many turns and a large radius, and performing measure-
ments at high frequency, it should be possible to investigate this amplification
effect. The predicted coil-enhanced voltage spectrum can readily be computed.
The result is shown in Fig. 4 for a 1-cm diameter coil of 2000 turns, made of
38 AWG tungsten wire, and kept at a temperature of 3K. In Fig. 4, the upper
(dotted) curve represents the predicted voltage spectral density for the combined
ohmic plus radiation resistance. The lower (solid) curve is the predicted result
when radiation resistance is ignored. If the postulate of Blanco et al. is correct,
the enhancement in voltage fluctuations due to the antenna-like nature of the
coil should be easily measured at frequencies as low as 100 MHz (where the
coil enhancement effect is 100-fold for tungsten).
To successfully measure the ZPF-induced voltage fluctuations, the require-
ments of low temperature, large coil, and high frequency must be met. The
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 579

low-temperature requirement is met by performing the experiment in a cooled

dewar. Existing high-quality cryogenic dewars (pumped down to 3K) and sensi-
tive laboratory instruments are suitable for the measurements. The cold spot in
one particular dewar under consideration is cylindrical, 2.5 cm in both diameter
and height. The largest coil that can be installed will thus have a coil radius of
approximately a 1 cm. To keep the linear dimension of the coil small will
require a small wire thicknesses, perhaps b 0.01 cm (gauge 38 AWG). By
winding the coil in a number of layers (10 or 12 layers), a large number of
turns can be accommodated, perhaps N 2000 turns. To minimize ohmic
resistance, wire made of tungsten (W) is preferred; however, copper (Cu) is a
suitable alternative.
Voltage fluctuations in the 100 MHz range are easily detected using commer-
cially available laboratory equipment; hence this experiment could be performed
using tungsten without resorting to the more sophisticated Josephson junction
techniques required by Koch et al. for their higher frequency measurements.
For a copper wire coil, the magnitude of the enhancement effect is reduced some-
what compared to the tungsten results shown in Fig. 4. But for frequencies
approaching the GHz regime, the radiation resistance enhancement effect in
copper wire is still predicted to be over four orders of magnitude larger. Commer-
cial equipment readily allows measurements of the voltage spectrum in the GHz
regime. Therefore, given a cost tradeoff of copper versus tungsten coil fabrica-
tion, the use of copper coils may be preferred. A second coil can be used in a
control experiment constructed with the same parameters as the first coil, but
with half of its turns wound in the reverse direction. This will make the coil
non-inductive so that its voltage spectral density should correspond to the
lower solid curve in Fig. 4.

B. Zero-Point Field Energy Extraction by Ground

State Energy Reduction
As first analyzed by Boyer [22], and later refined by Puthoff [23], the follow-
ing paradox was addressed: although atomic ground states involve electrons in
accelerated motion, such states are nonetheless radiationless in nature, even
though it is well known from classical electrodynamics that charged particles
undergoing acceleration must always emit radiation. For the standard Bohr
ground state orbit of the hydrogen atom, this was interpreted as an equilibrium
process in which radiation by the electron in its ground state orbit was compen-
sated by absorption of radiation from the background vacuum electromagnetic
ZPE. This interpretation has recently been strengthened by the analyses of
Cole and Zou [24,25] using a SED model for the vacuum ZPE. Because the
balance between emitted orbital-acceleration radiation and absorbed ZPE radi-
ation is modeled as taking place primarily at the ground state orbital frequency,
one can consider the possibility of using this feature in some type of mechanism
to extract energy from the ZPF. One fundamental difference between the SED
interpretation and that of quantum mechanics is that in quantum mechanics,
the 1s state of the electron is regarded as having zero angular momentum,
whereas in the SED interpretation, the electron has an angular momentum
of mecre/137 (me electron mass, re electron radius, atomic fine structure
580 E. W. DAVIS AND H. E. PUTHOFF

constant a 1/137, and c/137 is the classical orbital velocity of the ground
state electron).
The Bohr radius of the hydrogen atom in the SED view is 0.529 Å. This
implies that the wavelength (l) of zero-point radiation responsible for sustaining
the orbit is 2p . 0.529 . 137 455 Å (or 0.0455 mm). It has been conjectured by
Puthoff and Haisch (private communication, 2004) that suppression of zero-point
radiation at this wavelength (and at shorter wavelengths) inside a Casimir micro-
cavity could result in the decay of the electron to a lower energy state determined
by a new balance between classical emission of an accelerated charge and
absorption of zero-point radiation at l , 455 Å, where l depends on the micro-
cavity plate separation (d ). Because the frequency of this orbit is 6.6  1015 Hz,
no matter how quickly the atom were to be injected into a Casimir microcavity,
one would assume that the decay process would be a slow one as experienced by
the orbiting electron. Figure 5 shows a schematic representation of a hydrogenic
atom in free space and inside a microcavity.
Consider the possibility that the decay to a new sub-Bohr ground state would
involve gradual release of energy in the form of heat, rather than a sudden optical
radiation signature. Because the binding energy of the electron is 13.6 eV, it is
estimated that the amount of energy released in this process could be on the
order of 1 to 10 eV for injection of the hydrogen atom into a Casimir cavity of
d 250 Å. Furthermore, consider the possibility that when the electron exits
the cavity it would reabsorb energy from the ZPF and be re-excited to its
normal state. If these conjectures were to be verified by experiment, then the
energy extracted in the process comes at the expense of the ZPF, which in the
SED interpretation propagates at the speed of light throughout the universe. In
effect, the energy would be extracted locally and replenished globally. The sec-
ondary consequences on other phenomena, if this energy conversion were to
succeed, have not yet been investigated. However, on a cautionary note, the con-
flicts between SED and QED theories (discussed in Sec. V) raise questions as to

Fig. 5 Energy released from ground state suppression of hydrogenic atom in a

microcavity. (rb 5 free-space Bohr orbit radius, rb0 5 suppressed Bohr orbit
radius, l 5 resonant wavelength of Bohr orbit, and Eout 5 released energy.)
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 581

whether the conjectured approach discussed here is viable. This issue is perhaps
best addressed by experiment for its resolution.
In terms of an experimental test, consider using monatomic gases or liquids
flowing in a block with Casimir tunnels, which has the following attributes: 1)
no dissociation process is required for monatomic gases or liquids, 2) heavier
element atoms are approximately two to four times larger than hydrogen and
thus can utilize and be affected by a larger Casimir cavity, and 3) heavier
elements have numerous outer shell electrons, several of which may be simul-
taneously affected by the reduction of zero-point radiation in a Casimir cavity.
All of the noble gas elements contain ns electrons. He (Z 2, r 1.2 Å) has
two 1s electrons. Ne (Z 10, r 1.3 Å) has two each of 1s and 2s electrons. Ar
(Z 18, r 1.6 Å) has two each of 1s, 2s, and 3s electrons. Kr (Z 36,
r 1.8 Å) has two of each of 1s, 2s, 3s, and 4s electrons. Xe (Z 54,
r 2.05 Å) has two of each of 1s, 2s, 3s, 4s and 5s electrons. Larger Casimir
cavities would also be expected to have an effect on the energetics of the outer
electron shells (at larger radii). One could therefore expect that a Casimir
cavity having d 0.1 mm could have an effect on reducing the energy levels
of the outermost pair of s electrons, and possibly also p electrons and intermediate
shell s electrons as well.
Continuing with this model, it is reasonable to expect that a 0.1 mm Casimir
cavity could result in a release of 1 to 10 eV for each injection of a He, Ne,
Ar, Kr, or Xe atom into such a cavity. According to Maclay [26], a long cylind-
rical Casimir cavity results in an inward force on the cavity walls due to the
exclusion of interior ZPF modes. In the “exclusion of modes” interpretation of
the Casimir force, this implies that a cylindrical cavity of diameter 0.1 mm
could yield the desired decay of outer shell electrons and subsequent release of
energy. If we let the length of the cylinder be 100 times the width, this results
in ‘ 10 mm for the length of the Casimir tunnel. Taking advantage of this
effect, Puthoff and Haisch (private communication, 2004) propose a segmented
tunnel consisting of alternating conducting and nonconducting materials, each
10 mm in length. In a length of 1 cm, there could be 500 such pairs in segments,
resulting in 500 energy releases (each yielding 1 to 10 eV) for each transit of an
atom through the entire 1-cm long Casimir tunnel.
Now consider a 1-cm3 block that is built up of 10 mm thick alternating layers
as described above. Assume that tunnels of 0.1-mm diameter could be drilled
through the cube perpendicular to the layers (this is not physically possible, of
course; tunnel manufacture must be done differently). If 10% of the cross
section comprises entrance to some 1.3 billion tunnels, then the amount of
energy released would be proportional to the flow rate of the gas through the
tunnels (for the number of entrances and exits through Casimir segments). A
flow rate of 10 cm/s through a total cross-sectional area of 0.1 cm2 yields
1 cm3 of gas per second flowing through the tunnels, which at STP would
be 2.7  1019 atoms. A very simple sealed, closed-loop pumping system could
maintain such a continuous gas flow. Because each atom interacts 500 times
during its passage, there would be 1.3  1022 transitions per second in the
entire cube of 1 cm3. An energy release of 1 to 10 eV per transition corresponds
to 2150 to 21,500 W of power released for the entire Casimir cube of tunnels.
However, again, all of this assumes that the chain of conjectures detailed
above is correct. Fortunately, this can be experimentally tested.
582 E. W. DAVIS AND H. E. PUTHOFF

Microcavity fabrication to match the atomic ground states is daunting because

there will potentially be fabrication irregularities that cause edge and surface
effects that act upon the particles as they enter or exit the Casimir region. And it
is not possible to drill 1.3 billion tunnels having diameters of 0.1 mm. However,
it should be feasible to use microchip technology to etch holes into the individual
layers first and then assemble the stack. Extremely fine coregistration and align-
ment of stacks would be an issue, but a surmountable one. A much smaller
number of layer pairs and tunnels would suffice for a measurable demonstration
of release of ZPE by this process. If such a small-scale demonstration succeeds,
larger versions that convert more energy could be built that also take advantage
of more efficient thermal-to-electrical energy conversion methods. If successful,
such apparatus could also be used to explore for secondary effects of converting
quantum vacuum energy into thermal then electrical energy.
Further investigation by Puthoff et al. [27] was based on the premise that the
above principle is broadly applicable to other than just atomic ground states. In
their experiment, H2 gas was passed through a 1-mm Casimir cavity to suppress
the ZPE radiation at the vibrational ground state of the H2 molecule. The antici-
pated signature for such a process would be an increase in the dissociation energy
of the molecule. Initial experiments, shown in Fig. 6, were carried out at the
Synchrotron Radiation Center at the University of Wisconsin at Madison,
where an intense UV beam is available to disassociate gas molecules. Further
experimentation to investigate this hypothesis has yet to be completed. (See
Chapter 20 in this text for additional details.)

Fig. 6 Experimental apparatus for ground state energy reduction tests.

 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 583

C. Tunable Casimir Effect

As previously discussed, the Casimir effect is a unique ZPF-driven quantum
force that occurs between closely spaced conductive cavity walls (or plates).
If left unfettered, the plates will collapse together and energy is converted from
the ZPF into heat (or other forms of energy) in accordance with the expression
E/A p 2h c/720d 3, where E/A is the energy per unit area of the plates and d
is the plate separation. Investigation of this mechanism by Cole and Puthoff [11]
showed that this process fully obeys energy conservation and thermodynamic laws.
Although the Casimir force is conservative (thus the Casimir device might
appear to be a one-shot device), the fact that the attractive Casimir force is
weaker for dielectric plates compared to conductive plates raises the possibility
of the use of thin-film switchable mirrors to obtain a recycling engine
[28 30]. Figure 7 shows a comparison of the strength of the Casimir force in
a conductive cavity with that in a dielectric cavity. In such an application the
plates are drawn together by the stronger force associated with the conducting
state and withdrawn after switching to the dielectric state. The engine cycle for
this concept is shown in Fig. 8. Assuming optimistic conditions for practical
devices (negligible energy required for switching; plate separation oscillations
between 30 nm and 15 nm for 1 cm2 plates; driving circuit 10 times the
weight of the Casimir plates, etc.), an estimate of the achievable power might
be obtained. Based on the described parameters, and assuming a switching
from a purely conductive state to a dielectric constant of K 4, yields a figure
of merit of 35  f(MHz) W/kg ( f switching rate) for the power density
[28]. This can be compared to the power density of 5 W/kg achieved by
current radioisotope thermoelectric generators. The predicted output power
per unit area for this experimental device is 10 6f (MHz)/4[d(mm)]3 W/cm2.

Fig. 7 Tunable Casimir effect: conductor versus dielectric.

584 E. W. DAVIS AND H. E. PUTHOFF

Fig. 8 Tunable Casimir effect: engine cycle.

Another “tunable” conductive-type plate experiment under consideration

involves the use of plates consisting of three-dimensional photonic crystals,
with the bandgap of the photons that can transmit through the structure being a
“tunable” value. Using microelectro-mechanical processing methods, Sandia
National Laboratories has produced such crystals and are researching methods
of actively modifying the structures while in use [31]. The technology require-
ments for this concept are the nano-fabrication of microcavities with thin-film
deposited surfaces, RF-driven piezoelectric mounts for cavity oscillation,
mirror-switching modality (e.g., hydrogen pressure modulation), and calori-
metric measurement of energy/heat production.
An initial experiment to explore this concept was recently performed by
Iannuzzi et al. [32]. They investigated the effect of hydrogen switchable
mirrors (HSMs) on the Casimir force. HSMs are shiny metals in their “as depos-
ited” state. However, when they are exposed to a hydrogen-rich atmosphere, they
become optically transparent. Because the electromagnetic ZPF depends on the
optical properties of the surfaces, the Casimir force of attraction between two
HSMs in air should be different than the attraction between the same HSMs
immersed in a hydrogen-rich atmosphere. That is because one expects that the
Casimir force will be much weaker when the HSM is in the transparent state
rather than in the reflective state. The experiment tested this for plate separations
of 70 to 400 nm.
Iannuzzi et al.’s experimental results showed that the Casimir force did not
noticeably decrease after filling the experimental apparatus with hydrogen.
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 585

This may have occurred for two reasons. First, the dielectric properties of the
HSMs used in the experiment are known only in a limited range of wavelengths
spanning 0.3 to 2.5 mm, while the experiment measured the transparency of the
HSMs over a wavelength range of 0.5 to 3 mm. This narrower wavelength span
excludes the rest of the electromagnetic ZPF modes having wavelengths shorter
than 0.5 mm and longer than 3 mm. The ZPF modes lying outside this narrow
wavelength span were not affected by the hydrogenation-induced transparency
of the HSMs, hence their contribution to the total Casimir force acting
between the HSMs was not included. One would expect to see a significant
decrease of the Casimir force if the hydrogenation-induced transparency of the
HSMs had affected all of the ZPF mode wavelengths ranging from IR to UV
(ZPF modes with l  2.5 mm will not give rise to large contributions to the
force). Second, the experiment demonstrated a property of the Lifshitz theory
(see Ref. 32 for more detail), that in order to significantly change the Casimir
force between surfaces at separations on the order of 100 nm it is not sufficient
just to change their optical (IR and visible) reflectivity, but it is necessary
to modify their dielectric functions over a much wider spectral range. This com-
ports with the first reason, and indicates that more theoretical and experimental
work is needed to overcome the shortcomings of this experiment, and to allow
for the design and testing of new experiments that can achieve Casimir plate
transparency over a wider spectral range.
A notion similar to the tunable Casimir effect involves changing the dimen-
sions of a rectangular “Casimir box.” Forward [33] proposed a paradox in
which energy could be extracted by altering the aspect ratio of a conductive rec-
tangular Casimir cavity over a specific cycle of dimension changes (e.g., varying
width while holding length constant). It was subsequently shown by Maclay
[26,34], that the Casimir energy inside the box is not isotropic, varying in such
a way that more work is expended in cycling the box dimensions than can be
extracted. It appears that no net gain of energy is theoretically possible in this
scheme. Whether such considerations apply to the tunable Casimir cavity
concept remains to be assessed.

D. Electromagnetic Vortex Phenomenon

Shoulders [35] developed an experimental program to explore the physics of
microscopic plasma vortices (a.k.a. force-free plasmoids), which are thought to
be a form of ball lightning [36]. This study was motivated by the earlier exper-
imental work of Wells at the Princeton University Plasma Physics Laboratory,
Bostick and Nardi at the Stevens Institute of Technology, and their collaborators
[37 44]. Shoulders became interested in the possibility of stable, quantized
force-free structures that could be taken apart by some process to yield a net
energy gain for power generation. The foundation for this speculation was
Nardi et al.’s [44] observation of strange electron concentrations they called
vortex filaments that formed in an electron beam made by plasma focus or rela-
tivistic electron beam machines, which exhibited electron concentrations that
appeared to violate the space charge law. Furthermore, Nardi et al. observed
that the vortex filaments were striking exposed materials (metals, dielectrics, cer-
amics, glass, etc.), boring smooth channels straight through them and sometimes
586 E. W. DAVIS AND H. E. PUTHOFF

exploding with such a large force that they created impact craters or holes in the
materials. Piestrup et al. [45] performed more recent experiments to investigate
this unusual phenomenon. This discovery inspired Shoulders to consider vortex
filaments as a potential new source of energy, and hence he named them electro-
magnetic vortices or “EVs.” However, given that he could not experimentally
verify the vortex nature of the phenomenon, he later redefined EV to mean elec-
trum validum (roughly translated as strong electron).
Bostick and Shoulders began collaborating and realized that EVs were much
easier to generate and observe using micro-arc discharge devices because they
are usually obscured in large high-power plasma machines by surrounding
plasma. This led Shoulders to design a series of low-voltage, low-power
micro-arc discharge (or condensed-charge emission) devices to produce EVs in
the lab. Figure 9 shows a schematic diagram for one embodiment of an EV
(pulse discharge source) device. The EVs are generated at the cathode tip and
then follow the path (dashed line above the dielectric) to the impact site on the
ground plane. The EVs generated by such devices were able to reproduce the
material damage observed in Nardi et al.’s earlier experiments. Figure 10
shows a scanning electron microscope (SEM) photograph of the damage inflicted
by a single EV burst fired along an aluminum-oxide ceramic plate. The EV bored
through the ceramic, forming a smooth symmetrical channel along its path.
Shoulders’ experimental studies claim that EVs have physical characteristics
corresponding to the phenomenon observed by Nardi et al. His conclusions were
that EVs are compact, spherically shaped balls (diameter 1 to 20 mm) of con-
densed high-density charge (1030 electrons/m3) with an internal electric field
.108 V/m, a charge-to-mass ratio of 1.7588  1011 Coulomb/kg ( electron’s
charge-to-mass ratio), and a surface current density of 6  1015 Amps/m2
[35]. Shoulders also reported that EVs are a source of (copious) X-rays; a
single EV discharge gun can produce multiple EVs in which the coupling
between adjacent EVs produces quasi-stable structures (chains); and EVs
respond like an electron under deflection by external fields of known polarity.
Because electrons would not be expected to bind together due to their mutual
Coulomb repulsion, a speculative model based on the vacuum electromagnetic
ZPF was formed to explain the existence of EVs. The emerging laboratory

Fig. 9 Schematic of EV (pulse discharge source) device (C 5 capacitor;

V 5 voltage). (From Ref. 46.)
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 587

Fig. 10 SEM of EV damage to ceramic plate (20-mm scale). (From Ref. 35.)

evidence led them to consider the hypothesis that the Casimir effect may be a
major contributing mechanism to the formation of EVs in micro-arc discharges.
This conjecture is based on models by Casimir [47] and Puthoff and Piestrup [48]
suggesting that the generation of a relatively cold, dense, non-neutral (charged)
plasma results in charge-condensation effects that may be attributable to a
Casimir-type pinch effect (i.e., ZPF-induced pressure forces) in which the
inverse square-law Coulomb repulsion is overcome by an attractive inverse
fourth-law Casimir force to yield a stable configuration of bound charges at
small dimensions. This is a derivative of Casimir’s semi-classical model of
the electron in which a dense shell-like distribution of charge might suppress
vacuum fields in the interior of the shell [47]. However, initial application
of Casimir’s model found that the vacuum field inside the modeled electron
was found to augment rather than offset the divergent Coulomb field thus
rendering the electron’s self-energy divergent. Puthoff [49] later resolved
this problem by developing a self-consistent vacuum fluctuation-based model
in which the net contribution to the point-like electron’s self-energy by its
Coulomb and vacuum fields vanishes, thus rendering a stable finite-mass
electron.
Shoulders and collaborators subsequently investigated different approaches to
extracting useful energy from the vacuum ZPF by way of exploiting the EV
phenomenon. Even though EVs can be easily produced in the lab, efforts to
test this hypothesis have not met with success (see Chapter 20 for more
details). However, this topic is suitable for future research.
588 E. W. DAVIS AND H. E. PUTHOFF

V. Additional Considerations and Issues

A. Quantum Electrodynamic Vacuum Revisited
1. Quantum Electrodynamic Vacuum as a Plenum
Continued theoretical and experimental research has revealed that the vacuum
constitutes an active agent that contributes to a host of phenomena ranging from
microscopic level shifts of atomic states to possible connections to the cause of
cosmological expansion [14,50]. As more of its attributes are explored, the
vacuum has been found to exhibit phenomena characteristic of an optical
medium, such as induced birefringence in the presence of an applied magnetic
field [51], and breakdown (decay) in the presence of external electric fields
[52 54]. The current view is that the vacuum has structure, and can be con-
sidered much like a medium of classical physics. However, the vacuum differs
significantly from that of a classical medium due to the existence of quantum
fluctuations. A primary attribute of quantum theory is the concept of matter
and field fluctuations, rooted in Heisenberg’s uncertainty principle.
In second-quantized QED theory, the theory that applies to the electromag-
netic vacuum, the canonical approach to representing fluctuations of the free
vacuum electromagnetic field is to express the field distribution in terms of
standing- or traveling-wave normal modes. Section I suggested that the large
value of the integrated ZPE density fuels the concept of potentially useful
vacuum energy conversion to other forms, should even some small part of the
spectral distribution be accessible for conversion by technological means.

2. Quantum Electrodynamic Vacuum as a Mathematical

“Placeholder” for Fluctuating Matter Fields
The treatment of the QED vacuum as a fluctuating plenum with (formally)
infinite energy density has caused some physicists to question the viability of
the second-quantized QED formalism. Jaynes, for example, in considering the
consequences for calculation of the Lamb shift of the 2s level of the hydrogen
atom under the assumption of a much more modest electron Compton frequency
cutoff (1021 Hz), calculates a fluctuating power flow for the Poynting vector of
6  1020 MW/cm2 comparable in every square centimeter to the total power
output of the sun and states that “real radiation of that intensity would do a
little more than just shift the 2s level by 4 microvolts [55].” However, despite
alternatives to the formalism of QED that have been suggested (more on this
later), second-quantized QED cannot be lightly dismissed. This is so even
though the infinities that must be dealt with by such procedures as renormaliza-
tion caused even one of its founders, Paul Dirac, to remark: “This is just not
sensible mathematics. Sensible mathematics involves neglecting a quantity
when it turns out to be small not neglecting it because it is infinitely great
and you do not want it! [56].”
A second argument that can be raised against using the QED formalism to
further explore vacuum fluctuation physics is that, despite the magnitude of the
energy density potentially associated with vacuum electromagnetic fluctuations,
observation of the cosmological constant a measure of net vacuum energy
density has a value that is only on the order of the average energy density of
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 589

matter in the universe 10 9 J/m3 [57, 58]. This leads to what is often referred to
as the 120 orders-of-magnitude problem, or “cosmological coincidence.” Rather
than the QED value being discounted, the resolution of this problem is thought to
lie in the domain of infinity (or divergent integral) cancellations, requiring
instead an accounting for the fine-tuning requirements of such cancellations
[59,60].
Again, the root cause of the difficulties that accompany second quantization of
the vacuum field is that an unbounded plenum possesses an infinite number of
degrees of freedom, each with its assigned ground-state fluctuation energy. In
an attempt to circumvent the difficulties associated with an unbounded,
second-quantized plenum, alternative approaches to QED have been explored
in the literature in some detail, a few of which are discussed in Sec. V.E. A
number of these alternative viewpoints interpret the second-quantized QED
vacuum with its infinite degrees of freedom as simply an overidealized mathe-
matical placeholder for “real” fields that originate in matter fluctuations whose
number of degrees of freedom is necessarily always limited. Nevertheless,
though the alternative formalisms and associated interpretations differ signifi-
cantly from the canonical approach, detailed calculations yield results identical
to those generated by the second-quantized field formalism. As a result, even
treated as a mathematical placeholder for matter fluctuation fields, at this point
in our discussion the QED value must be taken seriously. The proposed corollary
concerning the potentially significant conversion of QED vacuum energy to other
forms is further evaluated in the sections that follow.

B. Casimir Effect Revisited

The most-quoted quintessential configuration for the conversion of vacuum
energy to other forms of energy is the Casimir effect. As previously discussed,
when parallel conducting plates are placed in a vacuum, they attract one
another by a very weak force that varies inversely as the fourth power of the dis-
tance between them. First computed by Casimir in terms of van der Waals forces
(a matter-fields approach; see below), he soon realized that, because the force
turns out to be independent of the molecular details of the conductors, it could
be computed as a problem in vacuum energy, and that is the way it is now
generally presented in the literature (the “plenum approach”) [5,61].

1. Casimir Effect in the Plenum Picture

One begins with the free quantum vacuum electromagnetic field fluctuations,
and then determines their modification due to the insertion of two parallel plane
conductors (i.e., plates) as additional boundary conditions, which constrain a dis-
crete set of intra-cavity modes of integer half-wavelengths. Aside from an unob-
servable, high-frequency-cutoff-dependent, free-field term that remains from the
mathematical regularization procedure, the resulting (renormalized) vacuum
stress-energy tensor is given by kT mnlvac (p 2h c/720d 4)diag( 1, 1, 1, 3),
where the angular brackets denote the quantum (vacuum state) expectation
value of the tensor T mn, d is the plate separation, and diag( 1, 1, 1, 3)
denotes the diagonal elements of a 4  4 matrix [1 3,61]. (For this derivation,
590 E. W. DAVIS AND H. E. PUTHOFF

the vacuum fluctuations of other quantum fields are essentially undisturbed by the
presence of the conductors or are affected only in the immediate vicinity of the
atomic nuclei that they contain.) kT mnlvac represents the real physical stress
carried by the vacuum field fluctuations in the presence of the parallel plane con-
ductors, and it encodes the Casimir effect in terms of 1) an interaction energy per
unit area, E/A p 2h c/720d 3, and 2) a corresponding force per unit
2
area, F/A p h c/240d 4. If free to move in response to the attractive
Casimir force, the motion of the plates toward each other is understood in the
plenum approach to progressively eliminate intra-cavity modes, converting their
associated ground-state energies first into kinetic energy, and then, upon collision
of the plates, into heat. In Section II we described the Casimir force-driven collapse
of Forward’s charged slinky as a Casimir-type configuration for building up an
electric field to charge a battery, and how such processes were shown not to
violate either conservation of energy or thermodynamic constraints.

2. Casimir Effect in the Fluctuating Matter Fields Picture

Complementary to the vacuum mode description (plenum approach), the
Casimir effect can be described, like van der Waals attraction, as arising from
correlations in the state of electrons in the two plates through the intermediary
of their coupled fields. From this standpoint (matter-fields approach), there is
no requirement for the high energy density vacuum field of the plenum approach
to reside throughout all space.
Unfortunately with regard to energy generation, though the Casimir forces
involved can be of significance for microelectromechanical systems (MEMS)
applications [62], the associated Casimir energies involved are too small to be
considered of significance for energy applications, so if the possibility for
vacuum energy conversion exists, one must look elsewhere to other types of
matter-vacuum interactions.

C. Type I (Transient) and Type II (Continuous) Machines

A key feature of the Casimir process just described, regardless of viewpoint
(plenum or matter-fields), is that it is a “one-shot,” transient, energy-producing
machine. That is, after delivering its energy, E, the matter that comprises the
machine is in a “used” state (we commonly refer to this used matter as “ash”)
and cannot be restored to the original state without an input of energy that is
greater than or equal to E. This “one-shot” feature can be generalized to define
a category of machine we call a type I transient machine, with the Casimir
machine constituting the prototypical representative. Should gravitation even-
tually be traced to a vacuum ZPF origin as proposed by Sakharov [63], then
the fall of an object of mass m through a height h in a gravitational field (g
acceleration of gravity at Earth’s surface), delivering its gravitational energy
mgh upon impact with the ground, would constitute another example.
In contrast, one can envision a type II (continuous) machine in which vacuum
ZPF energy is converted to a useful form on a recycling basis without net alteration
to its own matter state. A hypothetical example is the tunable Casimir device that
we reviewed in Sec. IV.C. The cycle of energy generation would consist of the
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 591

collapse of conducting plates with delivery of energy, followed by separation of

plates switched to insulating mode for which the attractive force is considerably
weaker, only to be switched back to conducting mode for the next cycle, etc.
Provided the input switching energy required per cycle is less than the output
energy delivered per cycle, a continuous generation of energy without a net
change in matter configuration would result. A second example would be a non-
linear oscillator that continuously, on a steady-state basis, downshifted high-
frequency components of the vacuum ZPF spectrum to lower frequencies for
convenient collection and application, without a net change in its own operation.
Clearly, a type II machine would be far more useful than a type I machine for
energy extraction. Type II machines would constitute a fuelless energy source,
with the ambient vacuum ZPF providing essentially unlimited energy. For this
to be the case, however, another requirement needs to be satisfied, which we
address in the next section.

D. Degradability of the Vacuum

The possibility of continuous conversion of vacuum ZPE to other forms (i.e.,
by a type II machine) requires that, in principle, vacuum energy must be degrad-
able (i.e., continuously consumable), not just that there be a surfeit of energy in
place to harvest. This perspective leads to a remarkable question for deeper
explorations of QED. It turns out that the mathematical structure of QED is
based on a formalism in which the vacuum mode structure and vacuum fluctu-
ation energy per mode are quantized in what could be called a “hardwired”
fashion, that is, they possess fixed immutable values. Therefore, at the end of a
cycle of a hypothetical type II machine, in which both matter and vacuum
mode structure have been returned to their original states, the vacuum modes
must of necessity contain at a minimum the same “hardwired” energetic
content as before the cycle. Therefore, assuming local detailed-balance energy
conservation, continuous conversion of vacuum ZPE to other forms via a type
II machine is, from the QED viewpoint, forbidden in principle because the
vacuum as described by the QED formalism is nondegradable. (Globally,
vacuum energy is not conserved during cosmological expansion, with work
being done by the negative vacuum pressure to maintain positive constant
vacuum energy density and therefore increasing the vacuum energy [64].) This
outcome of second-quantized QED theory permits but two interpretations with
regard to continuous vacuum energy conversion: 1) QED theory, despite criti-
cisms that can be leveled against it, is correct in its description of vacuum fluc-
tuation dynamics, and even though vacuum ZPE exists, it cannot be continuously
converted to other forms, or 2) the axiomatic inconvertibility is an artifact of an
over-idealized mathematical structure, and therefore the possibility of conversion
remains an open question.‡ What is not in question, however, is that QED,
as an axiomatic, quantum formalism based on the concept of an immutable,

‡
A number of publications by E. T. Jaynes, A. O. Barut, and their collaborators are based on the
premise that second quantization is an unnecessary artifact of an overidealized formalism.
592 E. W. DAVIS AND H. E. PUTHOFF

nondegradable vacuum, does not support the concept of continuous vacuum

energy conversion.

E. Alternatives to Quantum Electrodynamics

As noted in Section V.A.1, despite its successes the second-quantized QED
formalism with its infinite vacuum degrees of freedom and associated infinite
energy density has been the subject of criticism and, as a result, alternatives
have been proposed and investigated in the literature. The alternatives run the
gamut from neoclassical theories in which matter is quantized but the fields
are not (e.g., the voluminous work of E. T. Jaynes), through classical theories
where both matter and fields are treated classically, with vacuum fluctuation
fields taken to be real but of a classical nature (e.g., SED), to formalisms
which eliminate the concept of vacuum fields altogether (e.g., direct-action
approaches investigated by A. O. Barut and others; see the references cited
below). We examine each of these briefly with regard to the possibility of
useful “vacuum energy conversion.”

1. Neoclassical Theories of Quantum Electrodynamic

Vacuum Fluctuation Effects
A major proponent of the neoclassical approach has been E. T. Jaynes, who
has questioned whether the quantized vacuum field is physically real or merely
an artifice of the second-quantized QED formalism. Based on the fact that the
QED formalism permits expression of effects in terms of quantized “self” or
“source” fields as an alternative to expression in terms of quantized vacuum fluc-
tuation fields, Jaynes advanced the hypothesis that QED effects can be attributed
to the self-fields of quantized matter without considering independent quantiza-
tion of the vacuum fields, expressions in terms of the latter just being a place-
holder for the former. Pointedly, with regard to QED being “the jewel of
physics because of its extremely accurate predictions,” Jaynes’s position is that
“those accurate experimental confirmations of QED come from the local
source fields, which are coherent with the local state of matter,” and that “the
quantized free field only tags along” [65]. Jaynes nonetheless arrived at a con-
clusion that we might call Jaynes’ axiom, namely, “This complete interchange-
ability of source-field effects and vacuum-fluctuation effects . . . shows that
source-field effects are the same as if vacuum fluctuations were present.”
Applied to the case of a radiating atom, Jaynes provides a specific example of
his conclusion with the statement “The radiating atom is indeed interacting
with an electromagnetic field of the intensity predicted by the zero-point
energy, but this is just the atom’s own radiation reaction field” [66]. As a
result, with the axiomatic second-quantized field formalism set aside, in the neo-
classical approach any consideration of the conversion of vacuum ZPE for use
must be displaced to consideration of the conversion and degradability of
source or matter-fields fluctuation energy for use, issues yet to be addressed in
the literature.
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 593

2. Stochastic Electrodynamics Model Revisited

SED is a classical (i.e., nonquantized) theory of particle-field interactions that
assumes the existence of classical particles and a classical random background
electromagnetic field distribution whose Lorentz-invariant spectral energy
density is chosen to match that originally appearing in second-quantized QED.
Given SED’s heuristic value of classical-like modeling and ease of calculation,
and its seeming ability to address many quantum mechanical problems with
success (as outlined in Sec. III.B), the SED approach has been employed in the
literature to explore vacuum energy conversion. In the absence of a formalism
for vacuum field quantization, there are no fundamental immutability constraints
that would mitigate against vacuum energy degradability, so that issue is not tes-
table under this formalism.
Investigations to date have included the use of cavity-QED techniques to
suppress atomic or molecular ground states [27], and evaluation of the use of a
nonlinear oscillator to continuously downshift high-frequency components of
the vacuum fluctuation spectrum to lower frequencies for convenient collection
and use. With regard to the latter, the result of a nonrelativistic SED analysis
is that the downshifting process acts to convert an initial hypothetical cubic-
frequency vacuum fluctuation spectrum toward a Rayleigh Jeans rather than a
Planck heat spectrum (the former being a low-energy approximation of the
latter) [67,68]. Extension of the analysis to the relativistic regime does not
alter this conclusion [69,70]. Though further work remains, these considerations
lead us to conclude that SED in its present form is incomplete and may not be
useful for the assessment of the potential conversion of vacuum energy to
other forms; its predictions concerning such must be treated with caution.
Additional shortcomings of the SED model include convoluted attempts to
derive interference effects or Schrödinger’s equation, and the difficulty in
explaining sharply defined stationary states (i.e., sharp atomic spectra), though
there have been many attempts [17]. QED and SED do not, in general, yield
the same results for nonlinear systems, although they are in agreement for the
range of linear systems examined. The apparent disagreements between SED
and QED are quite serious, and occur in areas in which QED is highly successful.
Perhaps the source of these difficulties lies in accurately dealing with the non-
linear stochastic differential equations in SED for these problems. Even still, it
is likely that differences will remain, which should clearly be testable by exper-
imental means [71]. For a very thorough, detailed and scholarly review of SED,
see Ref. 17 and the corresponding review by Cole and Rueda [72].
Given the heuristic value of certain aspects of SED modeling, but with the
shortcomings outlined above noted, SED theorists de la Peña and Cetto have pro-
posed a modification to SED they call LSED (linear SED) [73]. The modification
consists of the addition of three new constraining principles that result in a form
of convergence with nonrelativistic quantum mechanics while retaining some of
the appealing attributes of standard SED (e.g., quantum states being stable on the
basis of a dynamic balance between absorption and emission of background
vacuum fluctuation fields). The added constraints (e.g., an added constraint
that invokes detailed energy balance for separate frequencies) result in correcting
594 E. W. DAVIS AND H. E. PUTHOFF

several known problems with standard SED. For example, now the equilibrium
spectrum is Planck’s, not Rayleigh Jeans, and wavelike behavior of matter
and nonlocality issues can be addressed, etc. The issue of continuous vacuum
energy conversion has yet to be addressed in this new formalism, however, so
that remains for the future.

3. Quantum Electrodynamics Without Second-Quantized Fields

As yet another alternative to canonical second-quantized QED, Barut [74] has
proposed that effects attributable to vacuum ZPF can be derived with a theory in
which there are source (matter) fluctuation fields but no vacuum fluctuation
fields, and that even the former can be eliminated. Barut’s approach is developed
in considerable detail as an independent, self-consistent, formulation of QED in
its own right. Barut argues that effects normally attributed to vacuum fluctuations
in the second-quantized, linear theory of the radiation field can be equally well
computed within the framework of a non-second-quantized, nonlinear theory
based entirely on matter wave functions alone. His program is to assess how
far one can go in understanding radiative processes without second quantization
or vacua that fluctuate. Barut and his collaborators have successfully applied the
theory to the Lamb shift and spontaneous emission [75,76], problems of cavity
QED [77], Casimir Polder and van der Waals forces [78], calculations of
the ge 2 (gyromagnetic ratio, ge) factor of an electron [79 81], and the
Davies Unruh effect [82], among others.
Given that the formalism of second-quantized field operators are not used at
all in the Barut approach, the seemingly quantized properties of fields are
taken to simply reflect first quantization of the sources. Therefore, in the
absence of the independent existence of second-quantized field fluctuations,
the QED arguments concerning immutability and nondegradability of quantized
vacuum fluctuation fields, and the corollary proscription against potential conver-
sion of energy from such fields, do not apply. As in the neoclassical approach, the
question of the conversion of quantum ZPE to other forms must be diverted to
consideration of the global properties of matter fluctuation interactions in the
as-yet incomplete development of the Barut approach.

F. Examples of a Degradable Vacuum

In closing, we review three examples of a degradable vacuum that are pre-
dicted by the quantum field theory of curved spacetime and the Standard
Model of elementary particle physics.

1. Gravitational Squeezing of the Vacuum

In their study of traversable wormholes, Hochberg and Kephart [83] discov-
ered that the gravitational field of any astronomical body produces a zone of
negative energy around it by “dragging” some of the virtual quanta (a.k.a.
vacuum ZPF) downward. They applied their discovery to the problem of creating
and stabilizing traversable wormholes. Their quantum optics analysis showed
that there is a distortion of the vacuum electromagnetic ZPF due to the interaction
with a prescribed gravitational background, which results in “squeezed” vacuum
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 595

states that possess a negative energy density. Squeezing of the vacuum is a

quantum process that is roughly analogous to the compression of an ordinary
fluid. (See Sec. II.B.2 in Chapter 15 for further details.) This means that, as the
vacuum field is continuously being squeezed by the gravitational field of a
body, its energy is continuously being degraded with respect to the undisturbed
remote vacuum field.
The magnitude of the gravitational squeezing of the vacuum can be estimated
from the quantum optics squeezing condition for given transverse (to direction of
gravitational acceleration) momentum and (equivalent) energy eigenvalues,
j 8prS/l, of two electromagnetic ZPF field modes, subject to j ! 0, where
l is the ZPF mode wavelength and rS is the Schwarzschild radius of any astro-
nomical body under study [83]. (Note that rS 2GM/c 2 is the critical radius
at which a body of mass M collapses into a black hole, used here as a convenient
radial distance parameter to simplify the inequality, but there is no actual black
hole collapse involved in this mechanism; and G is Newton’s universal gravita-
tion constant, 6.673  10 11 Nm2/kg2.) This condition simply states that
substantial gravitational squeezing of the vacuum occurs for ZPF field modes
with l  8prS of the mass under study. (See Sec. II.B.3 in Chapter 15 for
further details.)
It is not clear whether this mechanism can be exploited to extract energy from
the vacuum. Conservation of energy suggests one of two possible outcomes: 1)
the lost energy is injected into the gravitational energy of the body, or 2) the
lost energy reappears as an accumulation of positive energy density ZPF
modes elsewhere in the universe. Further research will be needed to address
this question.

2. Redshifting the Vacuum

Calloni et al. [84,85] explored the possibility of verifying the equivalence
principle for the ZPE of QED. They used semiclassical quantum gravity
theory to evaluate the net force produced by the quantum vacuum ZPF acting
on a rigid Casimir cavity in a weak gravitational field that is modeled using
the standard Schwarzschild spacetime metric geometry.§ They evaluated the reg-
ularized (or renormalized) stress-energy tensor kT mnlvac of the quantized vacuum
electromagnetic field between two plane-parallel ideal metallic plates lying in a
horizontal plane. kT mnlvac encodes the Casimir effect, which has a negative
energy density and a negative pressure along the vertical (gravitational accelera-
tion) axis between the plates. Bimonte et al. [86,87] also studied this problem
using Green-function techniques in the Schwinger DeWitt quantum ether
prescription for kT mnlvac in a curved spacetime. The results from these studies
agreed with the equivalence principle and showed that quantum vacuum ZPF
does gravitate because the energy of each ZPF mode is redshifted by the factor

§
A spacetime metric is a Lorentz invariant distance function between any two points in spacetime,
which is defined in terms of a metric tensor, gmn, that encodes the geometry of spacetime (Greek
indices m,n ¼ 0 . . . 3 denote spacetime coordinates, x 0 . . . x 3, such that x 1 . . . x 3 ; space coordinates
and x 0 ; time coordinate).
596 E. W. DAVIS AND H. E. PUTHOFF

( g00)1/2 [1 (2GM/c 2r)]1/2 even though the modes remain unchanged

(here, M is the mass of a gravitating body, r is the radial distance from the
body, and g00 is the time-time component of the Schwarzschild metric tensor).
These studies suggest that cavity electromagnetic vacuum states are continu-
ously degrading inside a background gravitational field. But can we extract
energy from this mechanism? The answer to this question is not known at
present, but consideration of the conservation of energy suggests that the same
two possible outcomes given in the previous section would seem to apply: 1)
the lost energy is injected into the gravitational energy of the body, or 2)
the lost energy reappears as positive energy density ZPF modes elsewhere in
the universe. Further research will be needed to address this question as well.

3. Melting the Vacuum

In their study of the structured vacuum, Rafelski and Müller [53,54] analyze
the nature of the strongly interacting (QCD) vacuum and elucidate its character
from the Standard Model of particle physics and high energy particle accelerator
data. They concluded that in addition to the electroweak vacuum (i.e., the unified
electromagnetic and weak force vacua) there exists a dual QCD vacuum struc-
ture: one vacuum structure that is everywhere in space and consists of a compli-
cated soup of interacting gluons which confine the quarks (called the ordinary or
“frozen” vacuum); and another vacuum structure that is found inside elementary
particles (e.g., hadrons) and behaves like the dielectric vacuum of electro-
dynamics. In this second vacuum structure, particles that have a strong charge
(such as quarks or gluons) can move freely, but are confined by the frozen
vacuum that is everywhere else. This is called the perturbative or gluon or
“melted” vacuum, which can also be pictured as a quark gluon plasma. They
estimate that there is a “latent heat” of 1 GeV/fm3 (or 1035 J/m3)} associated
with the phase change of transforming from one vacuum structure to another
when the gluonic structures of the perturbative vacuum are melted. It is important
to point out here that this is a degradable vacuum structure.
This unusual dual vacuum structure led Rafelski and Müller to speculate on a
mechanism for the “burning of matter” as the ultimate source of energy in which
it might be possible that the energy contained within baryons could be converted
into useful energy. Their idea is to remove or destroy the three quarks residing
inside a baryon in order to gain energy, the latent heat, from the melted vacuum
inside the baryon. This process also entails the decay of the quarks via lepton-
quark interactions, which is a topic that is beyond the scope of this chapter. They
suggest that it might be possible that producing a quark gluon plasma in high
energy nuclear collisions could be a very efficient source of energy. In this
process atomic nuclei would be collided at high energy in order to form a compressed
high density zone in the region where the two nuclei overlap. This would lead to the
melting of the vacuum and the subsequent direct conversion of matter into radiation,
thus releasing 1035 J/m3 of energy density. This magnitude of energy density
would be very useful as a source of energy for space propulsion applications.

} 19
1 eV ¼ 1.602  10 J; 1 GeV ¼ 109 eV; 1 fm ¼ 10 15
m.
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 597

Rafelski and Müller point out that the commonly held view that the centers of
neutron stars are dead and cold, due to their nuclear fuel having burnt out and the
energy of gravitational collapse having been expended for the conversion of the
collapsed star into a gigantic atomic nucleus, is not the complete story. They hold
open the possibility that the entire rest-mass of all the baryons inside neutron stars
might become available and converted into heat. In their scenario, the core of a
neutron star is actually composed of condensed quark matter, and the rest-mass of
baryons is burnt up into radiation inside the quark core. They also point out that
supernovae explosions, gamma ray bursts, positron emission from the center of
our galaxy, quasars, and galactic nuclei have been observed to emit extreme
amounts of thermal energy, the mechanisms of which are still not understood
today.
Theoretical and laboratory studies of the dual QCD vacuum have been under-
way for over 20 years. The progress in experimental particle physics is such that
one gains an order of magnitude in the resolution (i.e., energy) of elementary par-
ticle structures roughly every decade. It is hoped that the commissioning of the
Large Hadron Collider in 2008 will lead to higher resolution probing of the
dual QCD vacuum structure, and help to determine whether there are deeper
grand unified and/or Higgs vacuum structures residing within quarks. This
topic is a rich area for future research, in which investigators can explore ways
to exploit the energy residing within new vacuum structures for space power
and propulsion applications.

4. Zero-Point Fluctuation Modes and Vacuum Field Energy

The above examples illustrate how the vacuum becomes degradable when per-
turbed under certain conditions. In each of the examples, the vacuum ZPF modes
were perturbed in such a way as to drive the QED field’s vacuum state energy
below zero, or the vacuum undergoes a phase change as in the QCD case. The
Casimir effect is another example in which certain ZPF modes are excluded by
boundary conditions that perturb the free-space vacuum ZPF modes, thus
driving the QED field energy below zero inside a Casimir cavity. In accordance
with the discussion in Secs. III.A and V.D, the ZPF modes serve only as
placeholders for a quantum field’s vacuum state calculations. Therefore, the
“hardwired” ZPF modes cannot be driven below the ground state. It is only a
quantum field’s overall vacuum state energy that can be driven below zero. So
we conjecture that the key to exploring the possibility of extracting energy
from the vacuum is to discover or invent additional mechanisms that perturb
the ZPF modes of a quantum field, which could then be technologically
implemented and tested in laboratory experiments.

VI. Conclusions
What are the conclusions that can be drawn from the considerations presented
here regarding the concept of continuous conversion of energy from the vacuum
electromagnetic ZPF?
First, we see that although the original inspiration for the concept of continu-
ous vacuum energy extraction came from second-quantized QED theory, it must
598 E. W. DAVIS AND H. E. PUTHOFF

be acknowledged that QED, as an axiomatic, quantized formalism based on

the concept of an immutable, non-degradable vacuum, does not support the
concept of continuous vacuum energy conversion. Given that second-quantized
QED is our most comprehensive quantum theory to date, its lack of support
for continuous vacuum energy conversion must be given serious consideration.
Second, SED as an alternative theory, whose formalism has been taken to
support the concept of continuous vacuum energy conversion, has enough short-
comings in its current state of development that one must conclude that it is not at
present an adequate tool for the assessment of potential vacuum energy conver-
sion. SED predictions must thus remain suspect in the absence of experimental
confirmation. The purpose of the experimental program outlined in Sec. IV is
meant to address this need.
Third, the concept of the conversion of energy from vacuum fluctuations is in
principle not falsifiable, given the unknowns that present theory has yet to resolve
(e.g., cosmological dark energy), and the numerous approaches currently being
brought to bear in the development of quantum theory.
Finally, even though experimental efforts at energy extraction from the
vacuum have been proposed or are already underway at various laboratories,
definitive theoretical support underpinning the concept of useful extraction of
energy from quantum fluctuations is not yet in place. Such support awaits theore-
tical developments that either posit a plenum that (unlike second-quantized QED)
can be shown to be degradable, or posit conversion of energy associated with
matter fluctuations, also in a degradable fashion. Because the quantum fluctu-
ations of interest are associated with quantum ground states, what is minimally
required are particle vacuum or particle particle interactions that result in the
formation of alternate lower-energy, ground states of matter/field configurations.
Suggested approaches to be explored are those which are known to yield results
consistent with the existence of vacuum fluctuation fields, but without the formal-
ism of independently postulated second-quantized vacuum fields. Whether useful
conversion of energy from quantum fluctuations can be accomplished, and iden-
tifying the unequivocal conditions under which this can be achieved, are yet to be
determined.
It has been argued that the QED vacuum is degradable under the action of
gravitation-induced quantum squeezing or redshifting. However, it is not
known whether these effects can be exploited for the extraction of energy from
the vacuum. The concept of a dual, degradable vacuum structure in QCD can
possibly lead to the generation of useful energy via the release of latent heat
from melting the QCD vacuum. The new generation of high energy particle col-
liders coming online in the very near future may yield new information about the
complex vacuum structure of the universe, allowing us to find ways to exploit its
energy content for revolutionary space propulsion applications.

Acknowledgments
The authors wish to thank our colleagues, V. Teofilo (Lockheed Martin),
B. Haisch (ManyOne Networks), L. J. Nickisch (Northwest Research Assoc.),
A. Rueda (California State University Long Beach), D. C. Cole (Boston
University), M. Ibison (Institute for Advanced Studies at Austin), S. Little
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 599

(EarthTech International), and M. Little (EarthTech International), for their

extensive technical contributions to this chapter. We also thank J. Newmeyer
(Lockheed Martin), E. H. Allen (Lockheed Martin), T. W. Kephart (Vanderbilt
University), P. C. W. Davies (Arizona State University), M. G. Millis (NASA-
Glenn), and M. R. LaPointe (NASA-Marshall) for their very useful comments.

References
[1] DeWitt, B. S., “Quantum Field Theory in Curved Spacetime,” Physics Reports,
Vol. 19C, 1975, pp. 295 357.
[2] DeWitt, B. S., “Quantum Gravity: The New Synthesis,” General Relativity: An
Einstein Centenary Survey, Hawking, S. W., and Israel, W. (eds.), Cambridge
University Press, Cambridge, MA, 1979, pp. 680 745.
[3] DeWitt, B. S., “The Casimir Effect in Field Theory,” Physics in the Making, Essays
on Developments in 20th Century Physics, In Honour of H. B. G. Casimir, Sarlemijn,
A., and Sparnaay, J. (eds.), North Holland Elsevier Science Publ., New York, 1989,
pp. 247 272.
[4] Birrell, N. D., and Davies, P. C. W., Quantum Fields in Curved Space, Cambridge
University Press, Cambridge, U.K., 1984.
[5] Casimir, H. B. G., “On the Attraction Between Two Perfectly Conducting Plates,”
Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen, Vol.
51, 1948, pp. 793 796.
[6] Lamoreaux, S. K., “Demonstration of the Casimir Force in the 0.6 to 6 mm Range,”
Physical Review Letters, Vol. 78, 1997, pp. 5 8.
[7] Mohideen, U., “Precision Measurement of the Casimir Force from 0.1 to 0.9 mm,”
Physical Review Letters, Vol. 81, 1998, pp. 4549 4552.
[8] Chen, F., Mohideen, U., Klimchitskaya, G. L., Mostepanenko, V. M., “Theory
Confronts Experiment in the Casimir Force Measurements: Quantification of
Errors and Precision,” Physical Review A, Vol. 69, 2004, 022117.
[9] Forward, R. L., “Alternate Propulsion Energy Sources,” Air Force Rocket Propulsion
Laboratory, Air Force Space Tech. Ctr. Space Div., Air Force Systems Command,
Final Report AFRPL TR 83 067, Edwards AFB, CA, 1983, pp. A1 A14.
[10] Forward, R. L., “Extracting Electrical Energy from the Vacuum by Cohesion of
Charged Foliated Conductors,” Physical Review B, Vol. 30, 1984, pp. 1700 1702.
[11] Cole, D. C., and Puthoff, H. E., “Extracting Energy and Heat from the Vacuum,”
Physical Review E, Vol. 48, 1993, pp. 1562 1565.
[12] Mead, Jr., F. B., and Nachamkin, J., “System Converting Electromagnetic Radiation
Energy to Electrical Energy,” U.S. Patent No. 5,590,031, 1996.
[13] Peres, A., Quantum Theory: Concepts and Methods, Kluwer, Dordrecht, The
Netherlands, 1993.
[14] Milonni, P. W., The Quantum Vacuum: An Introduction to Quantum
Electrodynamics, Academic Press, San Diego, CA, 1994, Chaps. 1, 2, 8.
[15] Mulliken, R. S., “The Band Spectrum of Boron Monoxide,” Nature, Vol. 114, 1924,
p. 349.
[16] Ibison, M., and Haisch, B., “Quantum and Classical Statistics of the Electromagnetic
Zero Point Field,” Physical Review A, Vol. 54, 1996, pp. 2737 2744.
[17] de la Peña, L., and Cetto, A. M., The Quantum Dice: An Introduction to Stochastic
Electrodynamics, Kluwer, Dordrecht, The Netherlands, 1996.
600 E. W. DAVIS AND H. E. PUTHOFF

[18] Koch, R. H., Van Harlingen, D. J., and Clarke, J., “Quantum Noise Theory for the
Resistively Shunted Josephson Junction,” Physical Review Letters, Vol. 45, 1980,
pp. 2132 2135.
[19] Koch, R. H., Van Harlingen, D. J., and Clarke, J., “Observation of Zero Point
Fluctuations in a Resistively Shunted Josephson Tunnel Junction,” Physical
Review Letters, Vol. 47, 1981, pp. 1216 1219.
[20] Koch, R. H., Van Harlingen, D. J., and Clarke, J., “Measurements of Quantum Noise
in Resistively Shunted Josephson Junctions,” Physical Review B, Vol. 26, 1982,
pp. 74 87.
[21] Blanco, R., França, H. M., Santos, E., and Sponchiado, R. C., “Radiative Noise in
Circuits with Inductance,” Physics Letters A, Vol. 282, 2001, pp. 349 356.
[22] Boyer, T. H., “Random Electrodynamics: The Theory of Classical Electrodynamics
with Classical Electromagnetic Zero Point Radiation,” Physical Review D, Vol. 11,
1975, pp. 790 808.
[23] Puthoff, H. E., “Ground State of Hydrogen as a Zero Point Fluctuation Determined
State,” Physical Review D, Vol. 35, 1987, pp. 3266 3269.
[24] Cole, D. C., and Zou, Y., “Quantum Mechanical Ground State of Hydrogen Obtained
from Classical Electrodynamics,” Physics Letters A, Vol. 317, 2003, pp. 14 20.
[25] Cole, D. C., and Zou, Y., “Analysis of Orbital Decay Time for the Classical
Hydrogen Atom Interacting with Circularly Polarized Electromagnetic Radiation,”
Physical Review E, Vol. 69, 2004, 016601.
[26] Maclay, G. J., “Analysis of Zero Point Electromagnetic Energy and Casimir
Forces in Conducting Rectangular Cavities,” Physical Review A, Vol. 61, 2000,
052110.
[27] Puthoff, H. E., Little, S. R., and Ibison, M., “Discussion of Experiment Involving
Lowering of Ground State Energy for Oscillation in a Casimir Cavity,” International
Conference on Squeezed States and Uncertainty Relations (ICSSUR 2001): Special
Session on Topics of Physics Related to Stochastic Electrodynamics, Boston
University, 2001.
[28] Puthoff, H. E., “Vacuum Energy Extraction by Conductivity Switching of Casimir
Force,” Technical Memo, Inst. for Advanced Studies at Austin, Austin, TX, 1985,
unpublished.
[29] Lipkin, R., “Thin Film Mirror Changes into a Window,” Science News, Vol. 149,
1996, p. 182.
[30] Pinto, F., “Engine Cycle of an Optically Controlled Vacuum Energy Transducer,”
Physical Review B, Vol. 60, 1999, pp. 14740 14755.
[31] Lin, S. Y., Moreno, J., and Fleming, J. G., “Three Dimensional Photonic Crystal
Emitter for Thermal Photovoltaic Power Generation,” Applied Physics Letters,
Vol. 83, 2003, pp. 380 382.
[32] Iannuzzi, D., Lisanti, M., Munday, J. N., and Capasso, F., “The Design of Long
Range Quantum Electrodynamical Forces and Torques Between Macroscopic
Bodies,” Solid State Communications, Vol. 135, 2005, pp. 618 626.
[33] Forward, R. L., “Apparent Endless Extraction of Energy from the Vacuum by Cyclic
Manipulation of Casimir Cavity Dimensions,” NASA Breakthrough Propulsion
Physics Workshop Proceedings, NASA/CP 1999 208694, NASA Lewis Research
Center, Cleveland, OH, 1999, pp. 51 54.
[34] Maclay, G. J., “Unusual Properties of Conductive Rectangular Cavities in the Zero
Point Electromagnetic Field: Resolving Forward’s Casimir Energy Extraction
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 601

Cycle Paradox,” Proceedings of the STAIF 1999: Conference on Applications of

Thermophysics in Microgravity and Breakthrough Propulsion Physics, El Genk,
M. S. (ed.), AIP Conf. Proc. Vol. 458, AIP Press, New York, 1999, pp. 968 973.
[35] Shoulders, K. R., EV: A Tale of Discovery, Jupiter Technologies, Austin, TX, 1987.
[36] Stenhoff, M., Ball Lightning: An Unsolved Problem in Atmospheric Physics, Kluwer
Academic/Plenum Publ., New York, 1999.
[37] Bostick, W. H., “Experimental Study of Ionized Matter Projected Across Magnetic
Field,” Physical Review, Vol. 104, 1956, p. 292.
[38] Bostick, W. H., “Experimental Study of Plasmoids,” Physical Review, Vol. 106,
1957, p. 404.
[39] Bostick, W. H., Prior, W., and Farber, E., “Plasma Vortices in the Coaxial Plasma
Accelerator,” Physics of Fluids, Vol. 8, 1965, p. 745.
[40] Wells, D. R., “Observation of Plasma Vortex Rings,” Physics of Fluids, Vol. 5, 1962,
p. 1016.
[41] Wells, D. R., “Axially Symmetric Force Free Plasmoids,” Physics of Fluids, Vol. 7,
1964, p. 826.
[42] Wells, D. R., “Injection and Trapping of Plasma Vortex Rings,” Physics of Fluids,
Vol. 9, 1966, p. 1010.
[43] Wells, D. R., and Schmidt, G., “Observation of Plasma Rotation Produced by an
Electrodeless Plasma Gun,” Physics of Fluids, Vol. 6, 1963, p. 418.
[44] Nardi, V., Bostick, W. H., Feugeas, J., and Prior, W., “Internal Structure of Electron
Beam Filaments,” Physical Review A, Vol. 22, 1980, pp. 2211 2217.
[45] Piestrup, M. A., Puthoff, H. E., and Ebert, P. J., “Measurement of Multiple Electron
Emission in Single Field Emission Events,” Journal of Applied Physics, Vol. 82,
1997, pp. 5862 5864.
[46] Shoulders, K. R., “Method and Apparatus for Production and Manipulation of High
Density Charge,” U.S. Patent No. 5,054,046, 1991.
[47] Casimir, H. B. G., “Introductory Remarks on Quantum Electrodynamics,” Physica,
Vol. 19, 1953, pp. 846 849.
[48] Puthoff, H. E., and Piestrup, M. A., “On the Possibility of Charge Confinement by
van der Waals/Casimir type Forces,” Cornell University Library E Print Physics
Abstracts Archive arXiv.org Database [online database]. URL: http://arxiv.org/
physics/papers/0408/0408114.pdf [cited Aug. 2004, revised 2005].
[49] Puthoff, H. E., “Casimir Vacuum Energy and the Semiclassical Electron,” Inter
national Journal of Theoretical Physics, Vol. 46, 2007, pp. 3005 3008.
[50] Volovik, G. E., The Universe in a Droplet of Helium, Clarendon Press, Oxford, 2003.
[51] Zavattini, E., Zavatti, G., Ruoso, G., Palacco, E., Milotti, E., Karuza, M.,
Di Domenico, G., Della Valle, F., Ciminio, R., Carusotto, S., Cantatore, G., and
Bregant, M., “Experimental Observation of Optical Rotation Generated in Vacuum
by a Magnetic Field,” Physical Review Letters, Vol. 96, 2006, 110406.
[52] Fulcher, L. P., Rafelski, J., and Klein, A., “The Decay of the Vacuum,” Scientific
American, Vol. 241, 1979, pp. 150 159.
[53] Rafelski, J., and Müller, B., The Structured Vacuum: Thinking About Nothing, Verlag
Harri Deutsch Publ., Frankfurt, Germany, 1985.
[54] Rafelski, J., “Vacuum Structure An Essay,” Vacuum Structure in Intense Fields,
Fried, H. M., and Müller, B. (eds.), NATO ASI Series, Series B: Physics Vol. 255,
Plenum Press, New York, 1991, pp. 1 28.
602 E. W. DAVIS AND H. E. PUTHOFF

[55] Jaynes, E. T., “Survey of the Present Status of Neoclassical Radiation Theory,”
Coherence and Quantum Optics, Mandel, L., and Wolf, E. (eds.), Plenum Press,
New York, 1973, pp. 35 81.
[56] Dirac, P. A. M., Directions in Physics, H. Hora and J. R. Shepanski (eds.), Wiley,
New York, 1978, pp. 76 77.
[57] Schwarzschild, B., “High Redshift Supernovae Indicate that Dark Energy Has Been
Around for 10 Billion Years,” Physics Today, Vol. 60, 2007, pp. 21 25.
[58] Particle Data Group, “Review of Particle Physics,” Journal of Physics G, Vol. 33,
2006, pp. 210 232.
[59] Weinberg, S., “The Cosmological Constant Problem,” Reviews of Modern Physics,
Vol. 61, 1989, pp. 1 23; see also, “The Cosmological Constant Problems,” Cornell
University Library E Print Physics Abstracts Archive arXiv.org Database [online
database]. URL: http://arxiv.org/PS cache/astro ph/pdf/0005/0005265.pdf
[cited 12 May 2000].
[60] Volovik, G. E., “Vacuum Energy: Myths and Reality,” International Journal of
Modern Physics D, Vol. 15, 2006, pp. 1987 2010.
[61] Milton, K. A., The Casimir Effect: Physical Manifestations of Zero Point Energy,
USA World Scientific, NJ, 2001, p. 13.
[62] Chan, H. B., Aksyuk, V., Kleiman, R., Bishop, D., and Capasso, F., “Quantum Mech
anical Actuation of Microelectromechanical Systems by the Casimir Force,” Science,
Vol. 291, 2001, pp. 1941 1944.
[63] Sakharov, A. D., “Vacuum Quantum Fluctuations in Curved Space and the Theory of
Gravitation,” Soviet Physics Doklady, Vol. 12, 1968, pp. 1040 1041.
[64] Peacock, J. A., Cosmological Physics, Cambridge University Press, Cambridge,
U.K., 1999, pp. 25 26.
[65] Jaynes, E. T., “Probability in Quantum Theory,” Complexity, Entropy, and the Physics
of Information, Zurek, W. (ed.), Addison Wesley, Reading, MA, 1990, pp. 381 403.
[66] Jaynes, E. T., “Electrodynamics Today,” Coherence and Quantum Optics, Mandel,
L., and Wolf, E. (eds.), Plenum Press, New York, 1978, pp. 495 509.
[67] Boyer, T. H., “Equilibrium of Random Classical Electromagnetic Radiation in the
Presence of a Nonrelativistic Nonlinear Electric Dipole Oscillator,” Physical
Review D, Vol. 13, 1976, pp. 2832 2845.
[68] Pesquera, L., and Claverie, P., “The Quartic Anharmonic Oscillator in Stochastic
Electrodynamics,” Journal of Mathematical Physics, Vol. 23, 1982, pp. 1315 1322.
[69] Blanco, R., Pesquera, L., and Santos, E., “Equilibrium between Radiation and Matter
for Classical Relativistic Multiperiodic Systems. I. Derivation of Maxwell
Boltzmann Distribution from Rayleigh Jeans Spectrum,” Physical Review D, Vol.
27, 1983, pp. 1254 1287.
[70] Blanco, R., Pesquera, L., and Santos, E., “Equilibrium between Radiation and Matter
for Classical Relativistic Multiperiodic Systems. II. Study of Radiative Equilibrium
with Rayleigh Jeans Radiation,” Physical Review D, Vol. 29, 1984, pp. 2240 2254.
[71] Cole, D. C., “Simulation Results Related to Stochastic Electrodynamics,” Proceed
ings of the International Conference on Quantum Theory: Reconsideration of Foun
dations 3, Adenier, G., Khrennikov, A., and Nieuwenhuizen, T. M. (eds.), AIP
Conference Proceedings Vol. 810, No. 1, AIP Press, New York, 2005, pp. 99 113.
[72] Cole, D. C., and Rueda, A., “The Quantum Dice: An Introduction to Stochastic
Electrodynamics,” Foundations of Physics, Vol. 26, 1996, pp. 1559 1562.
 ON EXTRACTING ENERGY FROM THE QUANTUM VACUUM 603

[73] de la Peña, L., and Cetto, A. M., “Contribution from Stochastic Electrodynamics
to the Understanding of Quantum Mechanics,” Cornell University Library
E Print Physics Abstracts Archive arXiv.org Database [online database]. URL:
http://arxiv.org/PS cache/quant ph/pdf/0501/0501011.pdf [cited 4 Jan. 2005].
[74] Barut, A. O., “Quantum Electrodynamics Based on Self Energy versus Quantization
of Fields: Illustration by a Simple Model,” Physical Review A, Vol. 34, 1986,
pp. 3502 3503.
[75] Barut, A. O., and van Huele, J. F., “Quantum Electrodynamics Based on Self Energy:
Lamb Shift and Spontaneous Emission without Field Quantization,” Physical Review
A, Vol. 32, 1985, pp. 3187 3195.
[76] Barut, A. O., Kraus, J., Salamin, Y., and Unal, N., “Relativistic Theory of the Lamb
Shift in Self Field Quantum Electrodynamics,” Physical Review A, Vol. 45, 1992,
pp. 7740 7745.
[77] Barut, A. O., and Dowling, J. P., “Quantum Electrodynamics Based on Self Energy:
Spontaneous Emission in Cavities,” Physical Review A, Vol. 36, 1987, pp. 649 654.
[78] Barut, A. O., and Dowling, J. P., “Quantum Electrodynamics Based on Self Energy,
without Second Quantization: The Lamb Shift and Long Range Casimir Polder
van der Waals Forces near Boundaries,” Physical Review A, Vol. 36, 1987,
pp. 2550 2556.
[79] Barut, A. O., Dowling, J. P., and van Huele, J. F., “Quantum Electrodynamics Based
on Self Fields, without Second Quantization: A Nonrelativistic Calculation of
g 2,” Physical Review A, Vol. 38, 1988, pp. 4405 4412.
[80] Barut, A. O., and Dowling, J. P., “Quantum Electrodynamics Based on Self Fields,
without Second Quantization: Apparatus Dependent Contributions to g 2,”
Physical Review A, Vol. 39, 1989, pp. 2796 2805.
[81] Barut, A. O., and Dowling, J. P., “QED Based on Self Fields: A Relativistic
Calculation of g 2,” Zeitschrift fur Naturforschung, Vol. A44, 1989, p. 1051.
[82] Barut, A. O., and Dowling, J. P., “Quantum Electrodynamics Based on Self Fields:
On the Origin of Thermal Radiation Detected by an Accelerated Observer,” Physical
Review A, Vol. 41, 1990, pp. 2277 2283.
[83] Hochberg, D., and Kephart, T. W., “Lorentzian Wormholes from the Gravitationally
Squeezed Vacuum,” Physics Letters B, Vol. 268, 1991, pp. 377 383.
[84] Calloni, E., Di Fiore, L., Esposito, G., Milano, L., and Rosa, L., “Gravitational
Effects on a Rigid Casimir Cavity,” International Journal of Modern Physics A,
Vol. 17, 2002, pp. 804 807.
[85] Calloni, E., DiFiore, L., Esposito, G., Milano, L., and Rosa, L., “Vacuum Fluctuation
Force on a Rigid Casimir Cavity in a Gravitational Field,” Physics Letters A, Vol.
297, 2002, pp. 328 333.
[86] Bimonte, G., Calloni, E., Esposito, G., and Rosa, L., “Energy Momentum Tensor for
a Casimir Apparatus in a Weak Gravitational Field,” Physical Review D, Vol. 74,
2006, 085011.
[87] Bimonte, G., Calloni, E., Esposito, G., and Rosa, L., “Erratum: Energy Momentum
Tensor for a Casimir Apparatus in a Weak Gravitational Field,” Physical Review D,
Vol. 75, 2007, 089901(E).
 Chapter 19

Investigating Sonoluminescence as a Means of

Energy Harvesting

John D. Wrbanek, Gustave C. Fralick,† Susan Y. Wrbanek,‡

and Nancy R. Hall§
NASA Glenn Research Center, Cleveland, Ohio

I. Introduction
ONOLUMINESCENCE has risen to a source of interest to those outside of
S the ultrasonic community over the last decade [1]. The processes of under-
standing the effect lead to the challenge of utilizing some of its more interesting
properties in practical applications.
The sonoluminescence phenomenon is defined as the generation of light from
sound waves, first discovered in the 1930s as a by-product of early work on sonar
[2]. The report in 1992 of the ultrasonic trapping of a single glowing bubble in a
flask of water generated a cascade of research [3]. The glow from the bubble was
found to be generated in bubbles compressed to at least 150 kPa in an extremely
short duration flash (,12 picoseconds), and had temperatures of at least
25,000 K for the single bubble [4,5]. Bubbles of noble gases were seen to flash
brighter, but the nature of the liquid was seen as playing a large role in the
flashes as well [6 8].
Shortly after experimental results on trapped single bubbles were published,
models were developed to explain these measurements. Simple shock calcu-
lations showed that peak temperatures inside the sonoluminescent bubbles

This material is a work of the U.S. Government and is not subject to copyright protection in the
United States.
Trade names and trademarks used in this report are for identification only. The usage does not
constitute an official endorsement, either express or implied, by NASA.

Sensor Electronics Engineer, Research and Technology Directorate.
†
Senior Sensor Electronics Engineer, Research and Technology Directorate.
‡
Optical Instrumentation Electronics Engineer, Research and Technology Directorate.
§
Fluid Physics Aerospace Engineer, Research and Technology Directorate.

605
606 J. D. WRBANEK ET AL.

could reach 3  108 K based on the collapse of an ideal spherical gas bubble
[9]. Assuming a nonspherical collapse, the high-speed jet striking the
opposite side of the bubble gave rise to the possibility of the water being frac-
tured on the molecular scale and generating light as fractoluminescence
[10,11]. The extremely rapid collapse of the bubble led to the theoretical
examination of sonoluminescence as an effect of quantum vacuum radiation
[12]. The lack of an afterglow suggested a cooperative optical emission, like
that of an optical laser or superradiance [13,14]. A model of the flash resulting
from a two-component plasma in the bubble (containing a low-density halo
and high-density core) was developed to explain the measured properties
[15]. The actual process may be a combination of any of the above
[16 18]. A simplified schematic of the current model of the process is
shown in Fig. 1.
Even as these theories are being explored, applications for the effect are taking
shape, from fusion containment [9,19 22] to thin film deposition systems [23]
and other useful applications [24 27]. Recently, claims put forward for the
generation of fusion reactions have caused controversy in the scholarly and
popular media [28 37]. However, if realized, harnessing the high-energy
release in safe, emission-free ultrasonic processes would lead to the development
of revolutionary power systems for in-flight use for both aircraft and spacecraft.
The benefits of an onboard fusion power system will result in reduced
fuel consumption, lower emissions, and reduced noise for many types of aircraft.

Fig. 1 Schematic of the sonoluminescence process. (After Ref. 1.)

 INVESTIGATING SONOLUMINESCENCE 607

A practical fusion power source would replace the conventional gas turbine
auxiliary power units and electrical generators in aircraft and fuel cells and
batteries in spacecraft, improving flight and mission capability. Longer-term
use of the power source will have both an environmental benefit and act as a
positive contributor to the country’s energy diversification, as well as enable
new missions for both air and space.

II. Approaches
A common approach for the generation of power from ultrasonic acoustic
waves is through the attempted initiation of fusion reactions using the
high energy of cavitation. Various methods have been referred to early on as
“cavitation fusion” and more technically as “acoustic inertial confinement
fusion” (AICF), but popularly it is “sonofusion” or “bubble fusion.” Confusion
with “cold fusion” is understandable, as the respective test cells in “sonofusion”
and “cold fusion” appear at first glance to be similar bubbling glass beakers. The
mechanisms are very different; “sonofusion” utilizes the ultra-high temperature
and pressure conditions of the sonoluminescence to initiate fusion, whereas
“cold fusion” supposedly utilizes the boson-like properties of deuterium gas in
a metal lattice at room temperature [38]. “Sonofusion” is hot fusion on the
pico-scale.
One of the early concepts of a fusion generator was by Hugh Flynn of the
University of Rochester, who patented the cavitation fusion reactor (CFR)
in 1982 [19]. The CFR design had six acoustic horns to cavitate liquid
lithium metal with hydrogen, deuterium, or helium gas added. The fusion
reactions initiated by the intense cavitation of the liquid metal fuses the gas
with the liquid metal. Liquid metal was used due to the high speed of
sound, and thus higher energy cavitations. The case would heat up from the
fusion reactions, allowing energy transfer to a heat exchanger. The CFR
was never built.
After the isolation of single bubble sonoluminescence, Seth Putterman’s
research group at the University of California Los Angeles (UCLA) in 1994
[9] and William Moss’ research group at Lawrence Livermore National Lab-
oratory in 1996 [20] addressed the possibility that conditions in sonolumines-
cence in water allows the fusion of two deuterium nuclei (D-D) and a
deuterium and tritium nuclei (D-T). In 1997, UCLA patented Putterman’s
apparatus for converting acoustic power to other useful forms of energy,
including D-T fusion reactions by introducing D-T gas into a flask filled
with water [21]. The apparatus is a flask with two piezoelectric transducers
attached on opposite sides to produce resonances to cavitate the fluid.
Unlike Flynn’s apparatus, Putterman’s apparatus was in use in their lab.
Also unlike Flynn’s apparatus, no method of extracting the energy from
fusion was outlined, but the flask is enclosed in a temperature-controlled
box for calorimetric calculations.
As part of research sponsored by the Electric Power Research Institute, Russ
George and Roger Stringham of E-Quest Sciences (now D2Fusion, Inc.) reported
the generation of molten “ejecta craters” and vents along with anomalous heating
608 J. D. WRBANEK ET AL.

of titanium, nickel, copper and palladium metal foils under cavitation in heavy
water in 1997 and 1998 [29,30]. The craters and vents were compared to
similar nanoparticles formed by hydrogen and deuterium implantation in palla-
dium targets reported by Okuyama et al. of the Nagoya Institute of Technology
[39]. The apparatus used two “sonicators” with disks for acoustic input to an
aluminum container. George and the Stringham’s calorimetric calculations and
the trace amounts of helium-4 that were detected suggested to the researchers
that the formation of the nanoparticles were the result of fusion reactions in
the cavitation. No report was given of any results using cavitations in light
water. As the research was purely to determine if cavitation in heavy water
could result in fusion reactions in metals, no method of extracting the resulting
energy was outlined.
The most publicized claim of sonofusion was in 2002 by Rusi Taleyarkhan
et al. of Purdue University [31,32]. Taleyarkhan’s group reported the generation
of tritium and neutron flux from cavitation of deuterated acetone. Regular “light”
acetone was used in a control run. The selection of deuterated acetone as the
liquid medium was due to the high accommodation coefficient of dissociated
acetone compared to water; that is, more of the vapor in the dissociated
acetone is stuck to the bubble walls than water vapor bubbles. Their models indi-
cate that the temperatures in the resulting implosion of cavitating bubbles will
result in temperatures of over 1 million Kelvin for acetone, but under one-half
million Kelvin for water. The results imply that higher cavitation temperatures
are found in liquids with higher vapor pressure rather than liquids with higher
speed of sound [9,19], surface tension, or viscosity [40]. The apparatus was a
cylinder with a piezoelectric transducer ring producing resonances to cavitate
the acetone. The cavitation sites were initiated with a burst of neutrons from a
14-MeV pulsed neutron source. They note that the sonoluminescence was not
single bubble, but formed clusters of a thousand bubbles at the cavitation site.
The 2002 Taleyarkhan experiment was repeated by Oak Ridge National Labora-
tory (ORNL), but no coincidence of neutrons with sonoluminescence flash was
distinguishable from the neutron background [33].
In 2006, the experiment was modified by Taleyarkhan’s group using alpha
particle production from uranyl nitrate in a mix of benzene, tetrachloroethene,
and deuterated acetone [34]. Experimental runs with light acetone, as well as
cells using light and heavy water mixed with uranyl nitrate, were performed as
controls. The group reported an increase in neutron and gamma ray flux, using
the deuterated acetone mixture that was not seen in the other mixtures, including
heavy water. Edward Forringer and his group at LeTourneau University were
able to reproduce the experiment [35]. Critical analysis of the published data
has led some to conclude that the neutron flux is too weak to provide definitive
evidence of nuclear reactions [36], or that the recorded neutron energies are
consistent with 252Cf emission [37]. It is not clear if the experiment ruled out
fission reactions from uranium.
Though not having any claims of nuclear fusion, the work of Ken Suslick’s
group at the University of Illinois at Urbana-Champaign is of note since they
concentrate on the chemistry at work in ultrasonic cavitation (sonochemistry)
[41 43]. Suslick’s appratus consists of a transducer horn in a test cell immersed
in a temperature-controlled bath [41]. Suslick’s group found localized point
 INVESTIGATING SONOLUMINESCENCE 609

heating due to cavitation in liquids and the resulting metal clustering and liquid
decomposition can be attributed to processes with temperatures between 2900 K
and 5200 K that do not include fusion processes [42,43]. Because the high temp-
eratures generated that make sonochemistry possible may also be used for power
generation, an effort for harvesting power from sonoluminescence should not be
focused solely on sonofusion results.

III. Challenges for Application

The application of sonoluminescence as a means of energy harvesting requires
a clear understanding of the heat generation process involved. In order for sono-
fusion claims to be verifiable, detection of their nuclear products is a necessity.
The reactions cited are those that fuse two deuterium atoms to form: 1)
helium-3 and a neutron, 2) tritium and a proton [D(d,p)T], or 3) helium-4 and
a gamma ray. These reactions are reported to occur either in a sonoluminescent
bubble alone or interacting with a palladium catalyst. Detection of these products
will verify whether or not fusion is involved in the heating process and
discrimination of the products will verify the specific process involved.
In order for a system to be applicable in flight hardware, the system needs to be
scaled down to become as small as possible. Auxiliary power units in aircraft
have a current design point of 440 kW with a mass of 1396 kg, the equivalent
of 315 W/kg specific power [44]. The International Space Station (ISS) electrical
power system contains 76 Ni-H battery cells to supply the 8-kW peak power
required when the station cannot use the power from its solar cell arrays [45].
The size of a typical cell is 350 ml in volume, approximately 1/3 kg in mass, pro-
viding the equivalent of 316 W/kg specific power [46]. For a substantial
improvement over state-of-the-art, a factor of 10 improvement in overall per-
formance (power, mass, and volume) is desired. The ultimate goal then is to
provide a power source capable of better than 1 kW/kg, with the volume less
than 100 ml and mass less than 0.1 kg. For a sonoluminescence-based power
source, most of the mass is in the transducers and support equipment. Minimizing
the test cell to 20 ml or less is a realistic criterion for achieving these goals.
As an example of the scaling challenges, a first order estimate on the minimum
size of a sonoluminescence test cell can be derived from the role of harmonic
resonances in bubble oscillations [47]. A gas bubble in a liquid has a free
oscillation frequency given by Eq. (1):
s
1 3 gP
v¼ (1)
a r

Here v is the angular frequency at resonance (v 2pf ), a is the bubble

radius, P is the initial liquid pressure, g is the heat capacity ratio for the air in
the bubble (g 5/3 for monatomic gasses, g 7/5 for diatomic gasses), and
r is the liquid density at initial pressure. The forced oscillations are driven by
the resonance frequency of the test cell given in Eq. (2):
k pc
v¼ (2)
r
610 J. D. WRBANEK ET AL.

Again, v is the angular frequency at resonance (v 2pf ), k is the harmonic, r is

the radius of the cell, and c is the speed of sound in the liquid. This can be set
equal to the free oscillation frequency to determine the maximum response.
Sonoluminescence takes place in bubbles with initial radius less than 10 mm,
so, theoretically, the smallest test cell (with k 1) for water ( r 1000 kg/m3,
c 1500 m/s) at atmospheric pressure (P 105 N/m2) can be calculated by
combining Eqs. (1) and (2):
s
r k 2 p 2 c2 r
¼ ¼ 230 (3)
a 3g P

With the result of Eq. (3), a test cell can theoretically be sized to 4.6 mm in
diameter with a resonance frequency of f 326 kHz, making the goal of
scaling the test cell to under 20 ml appear realistic.

IV. Sonoluminescence at NASA

Initial examination of the phenomenon has been performed by several
researchers sponsored by NASA via the NASA Research Announcement
(NRA) process. The research efforts involved measurements of single and/or
multibubble sonoluminescence in microgravity (0 g vertical acceleration) and
hypergravity (2 g vertical acceleration) performed on a KC-135 parabolic
research flight. One such experiment was performed by researchers from the Uni-
versity of Washington. The researchers discovered that light emission brightened
promptly by 20% and increased continually under microgravity, suggesting
buoyancy-driven instabilities are a critical limitation to the effect [48]. This
research was selected under the NRA process for a flight experiment. Flight hard-
ware for an ensuing experiment on ISS studying buoyancy-driven instabilities in
single bubble sonoluminescence was under development in 2003 (Fig. 2), but was
canceled following a redirection in NASA’s space exploration efforts.
As part of our mission to conduct basic and applied research on advanced
instrumentation technologies, NASA Glenn Research Center began an examin-
ation of sonoluminescence for instrumentation and measurement technique
development [49]. The objective of the effort was to investigate claims and the-
ories of power generation based on sonoluminescence. The approach would
initially compare the emission from bubbles in light water, heavy water, and
other solvents. Conclusions on sonoluminescence-based power generation con-
cepts would be formulated utilizing the data from our experiments.

A. Apparatus
The basic equipment for sonoluminescence consists of a flask containing
the liquid, ultrasonic transducers, a piezoceramic amplifier, and a function
generator [4]. Schematics of our test apparatus are shown in Figs. 3 and 4. The
transducers are driven by an amplified signal from the function generator to
saturate the fluid with ultrasonic waves. The voltage and current applied to
the transducers can be monitored and instrumentation (such as a microphone,
 INVESTIGATING SONOLUMINESCENCE 611

Fig. 2 Single bubble sonoluminescence flight hardware testing in 2003.

spectrometer, and photomultiplier tube) can be positioned to monitor the result-

ing sonoluminescence.
If the ultrasonic energy from the transducers can be concentrated into a small
enough uniform spot, and the liquid does not contain a large amount of dissolved
gas, the liquid can be ruptured in cavitation that is controlled. The cavitation can
cause a single glowing bubble to appear that is reasonably stable, known as
single-bubble sonoluminescence (SBSL) [4]. If the liquid is nearly saturated
with dissolved gas, the cavitation forms filamentary patterns that are chaotic as
the input energy is distributed throughout the flask. A cloud of glowing
bubbles can result, which is referred to as multi-bubble sonoluminescence
(MBSL) [50]. Either way, the sonoluminescence is located at or near regions
of maximum pressure [50].

B. Imaging
The initial apparatus for our tests was a resonating test apparatus as shown in
Fig. 3, consisting of a 250-ml round borosilicate glass flask filled with refriger-
ated (178C) distilled water with a pair of piezoceramics attached with epoxy
on opposite sides. Patterns of MBSL were produced at various frequencies.
Although most of the patterns were chaotic, two stable geometric patterns
were reproducible. The patterns of interest were rings having four and eight
nodules at 68.5 kHz and 93.0 kHz, respectively. The MBSL patterns were
612 J. D. WRBANEK ET AL.

Fig. 3 Resonating test apparatus and associated equipment.

faint, requiring 20 minutes of dark adaptation in the lab for observers to see
them. To determine the placement of fiber-optic instrumentation for future
investigations, the position of the MBSL in relation to the sides of the flask
needed to be recorded.

Fig. 4 Sonicator test apparatus and associated equipment.

 INVESTIGATING SONOLUMINESCENCE 613

Fig. 5 Enhanced image from video of a ring of MBSL. The arrows indicate the three
visible nodules.

A low lux astronomical video camera was purchased to image the patterns.
The camera was set to maximum sensitivity and images were recorded using a
2-s frame integration setting. Images were recorded with and without the fre-
quency generator on, and with and without room lights on. The resulting compi-
lation of these is seen in Fig. 5. Only three of the nodules of the four-nodule ring
are seen in the image. The eight-nodule ring was too faint to be resolved by this
method.
To help increase visibility of the MBSL, a sonoluminescence setup using a
high power Sonicator test apparatus as shown in Fig. 4 was assembled. The
test cell used initially was a 100-ml borosilicate glass beaker filled with refriger-
ated distilled water, but later a 50-ml quartz flask was used. Each used a titanium
high-intensity ultrasonic transducer horn probe in line with the signal generator
and amplifier instead of piezoceramics. As seen in the figure, the transducer
horn was inserted into the open top of the beaker and flask. Besides being
capable of delivering high acoustic power, the Sonicator test apparatus has the
added benefit that the transducer is not physically attached with epoxy to the
test cell as it is in the resonating test apparatus.
With the resonating test apparatus, the previously reported effect [10] of
brighter sonoluminescence glow in cooler water at the same driving voltage
was observed. In order to keep the water temperature cool and the cells stable,
a Peltier cooler plate was modified from a commercial kit and aluminum
support rings were machined for use with the Sonicator test apparatus. The alumi-
num support rings allowed thermal contact of the round flasks with the Peltier
cooler plate. The Peltier cooler was not used in the resonating apparatus, as it
would interfere with the vibrations of the flask. The MBSL patterns were gener-
ated in both the beaker and quartz flask cells, with and without support rings, by
614 J. D. WRBANEK ET AL.

this method. However, the patterns were not as defined as in the 250-ml flask, and
the video camera was not able to detect the patterns as well as the ring patterns
generated by the resonating test apparatus.
A 16-bit grayscale charged coupled device (CCD) imaging system for
terrestrial still deep-space astrophotography was purchased to image the
MBSL. Unlike the compiled image (Fig. 5) from three separate images, only
brightness, contrast level enhancement, and the application of a noise filter
were needed to achieve the images shown in Figs. 6 to 10. The MBSL filament
feature on the bottom of the flask in Fig. 6b is approximately 25-mm wide by 1-
mm thick, making the resolution approximately 100 mm/pixel. The filament
structure is revealed with a gradient map of sonoluminescence intensity as
false-color, rendered in black-and-white, in Figs. 7b to 10b.
The sonoluminescence filaments seen in Figs. 7 and 8 follow a conical pattern
from the horn to the base of the beaker and flask. This filament pattern has been
reported elsewhere [46], but in our images, the container is clearly visible, and no
brightening agent was introduced into the water. The 50-ml flask produced the
brightest MBSL (Fig. 8), possibly due to the small volume and thus increased
concentration of the applied ultrasonic power.
The CCD imaging system was then used to look at the rings formed by
the resonating test apparatus using the 250-ml flask, which had been at room
temperature for several months. The ring patterns were reproduced (Figs. 9
and 10), but dimmer and at higher frequencies than in Fig. 5 due to the room
temperature water. The rings in the images are smeared due to the variation of
the nodules positions about the pressure maximums. The images reveal the
rings as “wavy” rings, with nodules near the piezoceramics.
The nodules of the eight-nodule ring are revealed as pairs of nodules shifting
about the four maximums of the four-nodule ring (only six of the eight nodules
are visible in Fig. 10). In addition, when comparing Figs. 9 and 10, the four-
nodule ring forms in the center of the flask, but the eight-nodule ring
forms slightly lower. The shift in position is not apparent to the dark-adapted
observer because the flask is not clearly visible in the darkened lab. Thus, this

Fig. 6 Bottom of a 100-ml beaker with the Sonicator transducer horn a) off and b)
on. The MBSL filament appears centered at the bottom of the beaker (arrow). Both
pictures are 1-min exposures at f/1.2.
 INVESTIGATING SONOLUMINESCENCE 615

Fig. 7 Enhanced image of MBSL at 108.8 kHz in the 100-ml beaker with Sonicator.
Exposure time was 2 min at f/2.8. The field of view is 6.6 3 5.1 cm. Image shown in
a) grayscale and b) false-color (shown here in black-and-white), highlighting filament
structure of the MBSL. Dashed lines outline the conical area beneath the Sonicator
where most of the sonoluminescence occurred.

ability to image the flask and the sonoluminescence effect together is demon-
strated as necessary if any precision is desired for the placement of in situ
instrumentation.
A 16-bit three-color CCD imaging system for terrestrial still deep-space
astrophotography was purchased for color imaging the MSBL. The camera
was calibrated using a color rendition chart. The resulting true-color image of
the MBSL in a 50-ml flask is shown in black-and-white in Fig. 11 with contrast
enhanced. Red light from LEDs on the support equipment scattered by the flask
and transducer horn help reveal their presence.

Fig. 8 Enhanced images of MBSL at 106.5 kHz in the 50-ml flask with Sonicator.
Exposure time was 3 min at f/2.8. The field of view is 7.1 3 5.4 cm. Image shown
in a) grayscale and b) false-color (shown here in black-and-white), highlighting
filament structure of the MBSL. Dashed lines outline the conical area beneath the
Sonicator where most of the sonoluminescence occurred.
616 J. D. WRBANEK ET AL.

Fig. 9 Enhanced images of MBSL at 68.76 kHz in the resonating 250-ml flask.
Exposure time was 3 min at f/2.8. The field of view is 11.4 3 8.8 cm. Image shown
in a) grayscale and b) false-color (shown here in black-and-white), highlighting
filament structure of the MBSL. Dashed line outlines the position of the ring of
MBSL within the flask.

C. Sonoluminescence in Solvents
Previous studies of sonoluminescence brightness in various solvents give
empirical relationships of the brightness varying with the liquid’s viscosity,
surface tension, inverse of the vapor pressure, or a combination of these properties
[40]. Properties of several solvents that have been identified as producing some
form of sonoluminescence are reported in Table 1 [40,51]. The stability of the
sonoluminescence light output with temperature is also reported to be dependent
on the molar heat of vaporization of the liquid divided by the boiling point.
Table 1 shows that the brighter sonoluminescence should be seen in solvents
with higher boiling points (.1008C); however, a higher boiling point also

Fig. 10 Enhanced images of MBSL at 93.28 kHz in the resonating 250-ml flask.
Exposure time was 5 min at f/2.8. The field of view is 11.4 3 8.8 cm. Image shown
in a) grayscale and b) false-color (shown here in black-and-white), highlighting
filament structure of the MBSL. Dashed line outlines the position of the ring of
MBSL within the flask.
 INVESTIGATING SONOLUMINESCENCE 617

Fig. 11 True-color picture of sonoluminescence in light water (contrast-enhanced)

using the 50-ml flask with Sonicator, shown here in black-and-white. Exposure
time was 2 min at f/2.8.

indicates a greater sensitivity to temperature. Note that this is consistent with

Flynn’s use of higher density liquids for more violent cavitation, but contradictory
to Taleyarkhan’s use of higher vapor pressure liquids (see Sec. II).
From the information in Table 1, pure glycerin appears to be the best solvent
for use in sonoluminescence studies. Like the other, higher boiling point solvents,
glycerin is notoriously hydroscopic, eventually absorbing water from the ambient
air to stabilize as the 80% glycerin to 20% water mixture listed in Table 1. As
glycerin is relatively safe and more readily available than the other solvents,
NASA Glenn began sonoluminescence studies using glycerin. Note that
acetone and other low boiling point solvents, which are being explored elsewhere
[34,35], were rejected based on low viscosity and high vapor pressure properties
that are considered detrimental to sonoluminescence intensity.
A cloud of cavitation was generated with a Sonicator setup (Fig. 4) that
corresponded to sonoluminescence in the liquid as seen under dark conditions,
as shown in Fig. 12. The cavitation was particularly localized, allowing for a
promising target for the spectroscopy and radiation studies that are planned.

V. Indications of High Temperature Generation

Cavitation bubbles are formed by the vaporization of the liquid due to
changes in pressure. The liquid transitions to its vapor phase around nucleation
points that arise from a breakdown in surface tension of the liquid. This break-
down can be initiated by the random mechanical motion of the liquid molecules,
container geometry, impurities (liquid or gaseous), and cosmic radiation. The
vaporization can translate into high temperatures. The highly localized nature
 618

Table 1 Properties important to sonoluminescence for different solvents at 2088C [40,51]

Molar heat Calculated Intensity

of MBSL stability with
Density Viscosity Surface Vapor vaporization intensity temperature
Boiling (r) Mol. wt. (m) tension (g) pressure (Pv) (DHv) (g2/Pv relative dI/(IdT)
Solvent point (8C) (g/cm3) (g/mole) (cP) (dynes/cm) (Torr) (kJ/mole) to water) (8C21)

Methanol 64.7 0.415 32.04 0.597 22.6 93.3 35.2 0.0180 20.072
Ethanol 78.5 0.789 46.07 1.2 22.8 44 39.3 0.0386 20.078
Acetone 56.5 0.792 58.08 0.326 23.7 186 30.3 0.0099 20.064
Cyclohexane 81.4 0.779 84.16 1.02 25.5 84.8 29.9 0.0251 20.059
Carbon 76.8 1.595 153.84 0.969 27.0 91.3 29.9 0.0261 20.059
tetrachloride
Benzene 80.1 0.879 78.11 0.652 28.9 74.3 30.8 0.0369 20.060
Light water 100 1.00 18.02 1.00 73.1 17.5 40.7 1 20.076
J. D. WRBANEK ET AL.

Heavy water 101.4 1.11 20.03 1.25 73.1 16.6 42.1 1.13 20.078
80% Glycerin þ 121 1.209 77.28 60.1 66.6 3.5 85.7 4.16 20.151
20% water
Ethylene glycol 197 1.116 62.07 19.9 47.7 0.08 49.6 93.3 20.073
Sulfuric acid 290 1.788 184.15 25.4 55.1 0.00006 94.1 166,000 20.116
Glycerin 290 1.261 92.09 1490 63.4 0.001 87.9 13,200 20.108
 INVESTIGATING SONOLUMINESCENCE 619

Fig. 12 Cavitation in glycerin with room lights a) on and b) off (contrast enhanced)
showing localized sonoluminescence in a 40-ml container. The Sonicator transducer
was set to 56 kHz.

of sonoluminescence makes a direct measurement of the bubble temperature extre-

mely difficult with thermocouples or RTD probe. One method of comparing the
temperatures of two different environments is to compare the surface modification
of materials exposed to the different environments. Leveraging our expertise in thin
films for high temperature sensing applications, the effect of sonoluminescent
bubbles on thin films was investigated to determine differences in the temperature
of sonoluminescence in light water (H2O) and heavy water (D2O).

A. Platinum Films
The effect of exposure of thin films to sonoluminescence in light water and
heavy water was compared using several 1-mm thick, 6.1  6.1 mm substrates
of alumina (Al2O3) coated with thin films. The first two samples had 3 mm of
platinum deposited on the alumina substrates and were not annealed. One
sample was exposed to MBSL in light water, and another was exposed to
MBSL in heavy water. Images taken by a scanning electron microscope
(SEM) are shown in Fig. 13, and both samples have similar features. Each
sample had the general appearance of as-deposited platinum film. However,
larger and more frequent grain clusters were seen in the sample exposed to
sonoluminescence in heavy water. These features are typically observed after
annealing platinum films at high temperatures. If the larger grains were created
from exposure to sonoluminescence, their appearance would be an indication
that the sonoluminescence in heavy water generated higher temperatures than
the sonoluminescence in light water.

B. Palladium – Chromium Films

The three new thin film samples were fabricated by first patterning a 1-mm
thick sputter-deposited platinum (Pt) resistance temperature detector (RTD) on
1-mm thick, 6.1  6.1 mm substrates of alumina and annealing at 10008C for
8 hours. The RTD pattern was to be used for applications of miniature instrumen-
tation to characterize sonoluminescence. The samples were then over-coated with
620 J. D. WRBANEK ET AL.

Fig. 13 SEM image comparison of as-deposited, 3-mm thick films of pure platinum
coating alumina substrates exposed to sonoluminescence with a) H2O and b) D2O.

a 3-mm layer of palladium alloy with 13% chromium (PdCr). The PdCr alloy film
is typically used as a strain gauge in high temperature environments. For this
instance, the PdCr alloy was used because of the affinity of hydrogen to palladium
to allow the localization of sonoluminescence, which may contain hydrogen ions
[40,52]. Though the Pt itself has some limited affinity to hydrogen, the film does
not adhere well to alumina at high temperatures, and some delamination of the Pt
film was seen after annealing prior to PdCr deposition.
One of the new samples was exposed to MBSL in light water and another to
MBSL in heavy water. Viewed under SEM, the sample exposed to MBSL in light
water showed no significant modification of the film. The sample exposed to
MBSL in heavy water showed 4- to 5-mm diameter craters in the PdCr film over-
coating the Pt film when viewed under SEM. As the Pt film was already delaminat-
ing, the third sample was exposed to MBSL in both light water and heavy
water. Images of the exposures (Fig. 14) show that the sample was exposed to
sonoluminescence directly at the tip of the Sonicator horn. Again, the sample

Fig. 14 Film sample on Sonicator horn tip (arrows) exposed to SL in a) H2O and
b) D2O at 56 kHz in 100-ml beakers.
 INVESTIGATING SONOLUMINESCENCE 621

exposed to light water showed no modification of the film, but after the heavy water
MBSL exposure, the 4- to 5-mm diameter craters in the film were apparent when
viewed under SEM. No craters in the PdCr film directly deposited on the alumina
substrate were seen. Figure 15 gives a side-by-side comparison of the film at different

Fig. 15 Surface of the 3-mm thick PdCr films on 1-mm thick Pt patterns seen by
SEM under increasing magnification. The black surface at the bottom of the
craters is nonconducting alumina. The films were exposed to MBSL in a) H2O,
2603 magnification (mag), b) D2O, 2753 mag, c) H2O, 20003 mag, d) D2O,
20003 mag, e) H2O, 50003 mag, and f) D2O, 50003 mag.
622 J. D. WRBANEK ET AL.

Table 2 Summary of film modification by MBSL in light and heavy water

Film on alumina MBSL in H2O MBSL in D2O

Pt (as deposited) No damage No damage

PdCr (as deposited) No damage No damage
PdCr (as deposited) over Pt No further Craters through film;
(annealed) (some Pt damage isolated PdCr damage
delamination after annealing)

magnifications after the light water exposure and after the heavy water exposure. A
summary of the damages seen in the films is given in Table 2.
Table 3 gives a summary of some properties of the materials used in the
samples [53,54]. Delaminating of films is considered a zero-order effect of
mismatches between the two materials’ coefficient of thermal expansion
(CTE). The fact that grain failures are seen in the film with PdCr on Pt and not
the PdCr suggests that the expansion of the PdCr on Pt is greater than that of
the PdCr compared to alumina. However, Table 3 indicates that the adhering
PdCr has a greater CTE mismatch to the alumina substrate than the Pt film
that was delaminating.
A first-order estimate of the relative adhesion strength of films on oxides is the
energy involved in forming an oxide of the film (or heat of oxide formation) [55].
Table 3 shows that the PdCr alloy film has a more negative heat of oxide for-
mation, and is thus more favorable to adhesion on the alumina than platinum.
The energy imparted to the film at the sonoluminescence point locations could
be enough to vaporize or melt the film when not adhering to the substrate, but
not enough to vaporize or melt the film when it is adhering to the substrate.
The adhesion of the film to the substrate may be allowing the substrate to
absorb the energy via phonon vibrations, and the delaminating film does not
allow the energy to be transferred directly into the bulk substrate material. Alter-
natively, the loose islands may have heated up and “popped” off due to thermal
expansion. This effect was seen on two separate samples in the heavy water runs
but were not seen in the light water runs. The craters did not occur in the as-
deposited Pt films in either light or heavy water.
Some failures of the PdCr film that did not involve failure of the Pt film were
observed after exposure to heavy water. The SEM image of the largest volume
(14  8  3 mm) is shown in Fig. 16. Analysis by EDX of the failure area indi-
cates that remnants of the PdCr film remain at the exposed Pt surface, presumably
in the form of several 0.375-mm radius globules observed in the SEM image.
From Table 3, at least 6.9 kJ/cm3 are required to melt PdCr from 208C,
and 7.1 kJ/cm3 for platinum. The craters and the PdCr film damage may be
due to the same heating process of the Pt film. The lack of damage induced in
the as-deposited Pt film suggests the heating is not directly from excessively
high energy density of the MBSL in heavy water as compared to light water.
The primary result of these tests revealed modification of PdCr films by MBSL
generated in heavy water but not light water. The film modification indicates high
energy densities are generated, though not high enough to indicate net energy
generation by the MBSL itself. No modification was observed of platinum
films by MBSL in heavy or light water.
 Table 3 Sample substrate and film properties [53,54]

Coefficient of
thermal
Thermal Thermal expansion Heat of oxide
Density Melting point Heat of fusion Specific heat conductivity diffusivity (CTE) formation (kJ/
Material (g/cm3) (K) (J/g) (J/gK) (W/mK) (mm2/s) (1026/8C) mol)

Al2O3 3.92 2300 16,435 0.880 30.0 8.70 8 —

Pt 21.5 2141 113.6 0.113 71.6 29.5 10 2173
Pd – 13%Cr 11.39 .1828 189.4 0.271 71.8 23.3 15 2468
INVESTIGATING SONOLUMINESCENCE
623
624 J. D. WRBANEK ET AL.

Fig. 16 A large 20 3 10 mm failure of the PdCr film over Pt film. Globules of PdCr
are seen (arrows) on the 1-mm thick Pt film that is still adhering.

C. Comparison to Other Claims

The formation of submicron globules and film failures are consistent with the
results of George and Stringham [29,30], as well as Suslick [42,43]. The film
modifications suggest that sonochemical processes in the sonoluminescence in
heavy water reached temperatures of at least 1830 K. The lack of similar
modification in adherent platinum suggests that the temperatures of the cavitation
did not reach over 2140 K. Suslick reported the temperatures of cavitations in
hydrocarbon slurries can be as high as 5200 K. The globules as a remnant of
larger PdCr failure suggest two possibilities: 1) The film failed due to high temp-
erature melting and the globules are “residue,” or 2) the film was not fully adher-
ing when exposed and the globules are bits of the failing film forced into the
platinum undercoat. The former interpretation is consistent with George and
Stringham, and the latter consistent with Suslick. The point film failures are
similar to the void formation seen by George and Stringham, but without the
vent cones associated with ion implantation. The large grain failures seen in
our films after high temperature exposure suggests the point failures are due to
thermal expansion mismatch of the alumina and the film. The lack of similar
failures in the light water runs suggests that the heavy water cavitations did
get warmer.
In comparing our results with those of Suslick, George, and Stringham, we
cannot claim fusion processes are at work in our sonoluminescence runs based
on the film modification. However, the increased temperature in the heavy
water runs is not easily explainable. As noted earlier, the development of an
 INVESTIGATING SONOLUMINESCENCE 625

energy harvesting device is not dependent on the detection of fusion in sonolumi-

nescence. The determination of the nature of the film heating, and understanding
the role of light and heavy water in temperature generation, will influence
the design and the estimation of the performance characteristics of an energy
harvesting device.

VI. Energy Harvesting

The application of sonoluminescence as a means of energy harvesting requires
a clear understanding of the heat generation process involved. There have been
several claims of fusion reactions occurring in sonoluminescence [24 33].
A direct method to verify the reaction processes is by sampling in situ with radi-
ation detectors. The ultimate application of sonoluminescence for this effort is the
energy harvesting of extremely hot bubbles at point locations. New instrumen-
tation is under development at NASA Glenn for the energy harvesting
application.

A. Fusion Claims
The reactions cited in sonofusion claims are those that fuse two deuterium
atoms to form: 1) helium-3 and a neutron [D(d,n)3He], 2) tritium and a proton
[D(d,p)T], or 3) helium-4 and a gamma ray [D(d,g)4He]. These reactions
either can occur in a sonoluminescent bubble alone or interacting with a palla-
dium catalyst [Pd:D(d,b)Y reactions]. See Table 4 for the calculated energies
of the products of these reactions. In order for sonofusion claims to be verifiable,
detection of their nuclear products is a necessity. Detection of the reaction pro-
ducts in Table 4 will verify whether or not fusion is involved in the heating
process and discrimination of the products will verify the specific process
involved.

B. Radiation Detection
The detection of the fusion reactions in the sonoluminescence test cells is
made difficult by the nature of the liquids used to generate the reactions. The
low level ionizing radiation levels are difficult to detect through water and
heavy water due to their nuclear properties. The liquids can be analyzed for
tritium production or the detection of helium-3 and helium-4 out-gassing

Table 4 Calculated energies of reactions of interest

Rest mass (MeV/c2) Kinetic energy

Reaction Product (D ¼ 1875.6128 MeV/c2) (keV)

D(d,n)3He 3
He 2808.3914 822
n 939.5536 2458
D(d,p)T T 2807.9027 1265
P 938.2720 3786
D(d,g)4He 4
He 3727.3792 75.7
g 23,771
626 J. D. WRBANEK ET AL.

Fig. 17 Schematic of thin film coated scintillating detector.

during the tests, but background impurities and the diffusive nature of the gases
makes such analysis slow and difficult. However, some success has been reported
using liquid scintillators and track detectors [30,31].
A direct method to verify the reaction processes is by sampling the ionizing
radiation in situ with scintillation detectors. A thin film coated scintillating
detector is under development to identify fusion reactions occurring in sonolumi-
nescence. The detector consists of a coated scintillating cube or fiber, a wave-
shifting optical fiber, a fiber-optic connector, and a photomultiplier (PMT)
module. A schematic is shown in Fig. 17.
The polyvinyltoluene (PVT)-based scintillator} generates a pulse of light if an
ionizing particle interacts inside it, deposting energy. The intensity of the light is
directly proportional to the energy deposited by the interacting particle with the
ionization potential based on the particle’s atomic number. The scintillator used
is reported to be insensitive to photons with energies over 100 keV as well as
neutral particles. A thin film coating of rhodium (Rh), copper (Cu), or palladium
(Pd) on the scintillator functions as either an attenuator or a convertor to allow
possible fusion products to react with the scintillator, and also prevents the sono-
luminescence light from generating false readings. The nuclear properties of the
scintillator and the coatings are given in Table 5 [56,57].

}
Bicron BC 408 scintillator was used in the prototype detector design. “Bicron” is a registered tra
demark of Saint Gobain Ceramics & Plastics, Inc. for their inorganic and organic scintillators. This
usage is for identification only, and does not constitute an official endorsement, expressed or
implied, by NASA or AIAA.
 Table 5 Nuclear properties of scintillator and coatings [56,57]

Thermal neutron 76 keV alpha 3.75 MeV 2.5 MeV

Density Atomic mass absorption cross particle range proton range electron range
Material (g/cm3) (amu) section (barns) (mm) (mm) (mm)

Scintillator 1.032 6.25 0.35 0.7914 208.6 12.34

Rh 12.41 102.91 144.8 0.2724 42.34 1.480
Cu 8.96 63.546 3.78 0.2076 48.68 1.953
Pd 12.02 106.4 6.9 0.1797 44.83 1.552
INVESTIGATING SONOLUMINESCENCE
627
628 J. D. WRBANEK ET AL.

The relative response of the scintillator with coatings for each reaction was
calculated using the Monte Carlo program SRIM [58]. For each coating (Rh,
Cu, Pd) and ion (p, T, 3He, 4He), 10,000 input particles were used, and the
balance of the energy of the particles transmitted through the coatings was
assumed deposited into the scintillator. The sum of the deposited energies was
scaled based on the reported scintilator response to determine the light output.
For each fusion reaction scenario, the ouput of the detectors is normalized to
the detector with the largest signal. The results are shown in Table 6. Thus, com-
paring the response of each detector will allow a reasonable identification of the
reaction occuring in the cell. In this way, the fusion reactions that appear (or not)
in sonoluminescent conditions can be determined, limited by the statistics of the
actual counts and background radiation.
The output for the D(d,n)3He reaction was determined from the helium-3 ion
transport in SRIM, as well as utilizing a 3-mm thick rhodium film as a neutron
convertor. Rhodium emits 2.44 MeV electrons in the beta decay of caputured
neutrons, and has an excellent thermal neutron absorption cross section of
145 barns/atom, which will capture about 0.1% of the neutron flux per micron
thickness of film. The emitted electron will be easily detectable as the scintillator
sensitivity to electrons is reported to be five times that for protons of equivalent
energy.
The use of palladium in ultrasonic systems as well as our observations in
Figs. 13 and 14 suggest that the metal should be considered in radiation detection.
All three of the deuterium-deuterium reactions will be tested with and without a
palladium film included on the detectors. If palladium has a catalytic role in
fusion reactions, then the detectors with the palladium will be more sensitive
as the reactions are occurring inside the palladium film. The sensitivity was
modeled in SRIM assuming the reactions occur at the bottom of the 0.3 mm Pd
layer, based on the 0.375 mm radius globules seen in Fig. 15, with all the particles
deposited into the scintillator or rhodium.
The current assembly of the detectors is shown in Fig. 18. This assembly is
too sensitive to background counts to provide useful data at this time and
improvements in the assembly are proceeding. With modifications, the concept
could potentially be used in other applications, such as radiation monitoring at
high altitudes or in space environments. This effort was successfully leveraged
in radiation detector development for the application of dosimeters on surface
suits for lunar extravehicular activities [59].

Table 6 Relative sensitivities of detectors to reactions of interest

Detector film
coating D(d,p)T D(d,n)3He D(d,g)4He Pd:D(d,p)T Pd:D(d,n)3He Pd:D(d,g)4He

3.0 mm Rh 0.89 0.04

0.3 mm Cu 0.86 1.00 1.00
0.3 mm Pd 1.00 0.94 1.00 1.00 1.00
0.3 mm Pd/ 0.84 0.04 0.87 0.03
3.0 mm Rh
 INVESTIGATING SONOLUMINESCENCE 629

Fig. 18 Prototype thin film coated scintillating detector a) assembly and b) close-up
of coated 1/4-inch scintillator cube with a thin film of Pd.

C. Energy Harvesting Concept

The common goal of energy harvesting devices is to generate electricity from
the environment. In the case of energy harvesting from sonoluminescence, the
environment has point locations of extremely hot bubbles. The most direct
method of energy harvesting of such localized heating is by thermoelectric
conversion to electricity.
Thermoelectric voltage generation uses the Seebeck effect, which is the
generation of a potential difference between two ends of a conductor or semi-
conductor that are at different temperatures. The effect involves the movement
of charge carriers from the hot end of the conductor, where the carrier density
is forced lower, to the cold end, where the carrier density is allowed to be
greater. The electric field due to the new distribution of charge carriers leads
to a thermoelectric voltage. The voltage generation per degree difference is
referred to as the Seebeck coefficient. To complete the electrical circuit a differ-
ent conductor is used with its own Seebeck coefficient, making the net thermo-
electric voltage generated the difference of the two with a corresponding
relative Seebeck coefficient. The simplest example of this effect is a thermo-
couple used for temperature measurement [60].
If the thermoelectric voltage generated is high and the losses due to the
thermal conduction and electrical resistance of the conductors are low, then the
voltage generated by such a device can be used to generate power. Thermo-
electric power generation is used on deep space probes (such as the Viking,
Voyager, and Galileo missions) where solar power, fuel cells, and batteries are
not practical. As their name indicates, the heat source for radioisotope thermo-
electric generators (RTGs) on spacecraft is a radioactive pellet heating a
ceramic block; the cold side is the darkness of space.
To take advantage of the high temperature robustness of ceramics and the
miniature nonintrusiveness of thin films, NASA Glenn is investigating thin
film ceramic thermocouples for high-temperature environments. Initial results
630 J. D. WRBANEK ET AL.

revealed a chromium silicide film with a Seebeck coefficient to be high enough to

be considered for use in thermoelectric generators [60]. These results suggest that
ceramic thin films can be tailored for thermoelectric energy harvesting
applications.
A schematic of the energy harvesting concept is shown in Fig. 19. A 6-mm
diameter thermopile, originally used for fabrication of heat flux sensors
(Fig. 20), will be used in fabricating devices for the initial test of a thermoelectric
generator for sonoluminescence. The thermoelectric elements will be made of
high-temperature ceramics, covered with electrical insulation for use in the
liquid. The inner junction will be covered with a high-temperature insulator
embedded with a palladium catalyst based on the indications of high temperature
generation seen in Figs. 14 and 15.
The output of the generator can be estimated using the properties of the
chromium silicide film [60]. A thermoelectric voltage of 200 mV per junction
results, assuming sonoluminescent bubbles react on the inner junctions of the
thermopile with a 20008C temperature difference from the surrounding liquid.
The thermopile resistance should be about 200 V per junction, giving 0.2 mW
of power per junction. About half of the generated power is expected be lost to
the resistance of the thermopile. Thus, a 40-pair thermopile as shown in
Fig. 20 should generate 4 mW of power under sonoluminescence. Because this
28 mm2 generator is a fraction of the MBSL area seen in Fig. 13, an array of gen-
erators can conceivably output enough electrical power to match the acoustic
power input to the system. Clearly, the improvement of the Seebeck coefficient,
thermal resistance, or electrical conductivity is needed for the high-performance
thermoelectric generators needed to allow practical applications for energy
harvesting of sonoluminescence.

Fig. 19 Cross-section concept of a high-temperature thermopile for energy

harvesting sonoluminescence.
 INVESTIGATING SONOLUMINESCENCE 631

Fig. 20 A 6-mm diameter thermopile for initial test of ceramic thin film
thermoelectric device.

VII. Summary and Future Directions

Instrumentation techniques were explored at NASA Glenn to measure optical,
radiation, and thermal properties of the phenomenon of sonoluminescence, the
light generated using acoustic cavitation. The objective of the effort was to
investigate claims and theories of power generation based on sonoluminescence,
particularly from fusion reactions in the glowing bubbles. The approach was to
determine whether there is any difference in the emission from bubbles in light
water and heavy water and then from bubbles in other solvents to formulate
conclusions of sonoluminescence-based power generation concepts utilizing
the data from the experiments.
More generally, there remains much to learn about sonoluminescence. Efforts
to characterize sonoluminescence and thereby form a complete theoretical
description of the phenomenon have produced such a wide variety of information
that much more research needs to be done. Characterization efforts such as
searching for byproducts of chemical and nuclear reactions, completely
mapping the temperature distribution in a sonoluminescence cell, and spec-
troscopy will all contribute to a more complete understanding of the nature
of sonoluminescence. Because the results of any type of characterization
measurement seem to change with reported changes in cell size and shape,
temperature, pressure, liquid, dissolved gasses, and oscillator frequency and
power, there is ample room for careful detailed characterization to be made.
Optical emission spectra need to be recorded for many different sonolumines-
cence setups, for SBSL and MBSL, as near and far from the bubble(s) as is
practically achievable. For different liquids, differences in spectra, power, and
whether there are single or multiple bubbles should be carefully noted. Tempera-
ture maps should be made of as much of the liquid region as possible, with the
632 J. D. WRBANEK ET AL.

smallest probes that are practical to use. Evidence for chemical or nuclear
reactions and the specifics of the condition that produced the evidence must be
carefully noted.
High-speed timing of bubble flashes, and high-speed photography should be
employed to understand better the mechanics of the bubble behavior. Addition-
ally, high-speed photography or light scattering studies may help map bubble
motion in the fluid. Currently, widely disparate claims about the nature and beha-
vior of sonoluminescence are made by people who are not considering the same
conditions under which the phenomenon of sonoluminescence is achieved. Until
many detailed characterizations are complete, a detailed theoretical explanation
of the phenomenon remains an elusive hope. The listing of measurement con-
ditions and sonoluminescence characterizations to be varied in experiments on
both SBSL and MBSL is summarized in Table 7.
A more detailed understanding of the nature of sonoluminescence will help
define its practical uses, although even with our current limited understanding
of the phenomenon some practical uses have arisen already. The field of sono-
chemistry is already using the localized heating provided by ultrasound to
drive chemical reactions [24], modify surfaces [25], produce interparticle
melting in liquid solid chemical reactions [26], and produce quantum dots [27].
These techniques have already produced useful applications even without a
complete understanding of the phenomenon.
After all this, two issues must yet be resolved. The first is whether fusion
actually occurs in the bubbles. This has been discussed above and is the subject
of active research in laboratories around the world. One possibility for increasing
the likelihood of fusion is to take advantage of the fact that D-T fusion is roughly
two orders of magnitude more likely than D-D fusion. This could be done using
mixtures of suitable deuterated and tritiated solvents. D-D fusion, if it can be
made to work, does have the advantage that the supply of fuel is almost unlimited,
whereas the tritium required for D-T fusion does not occur in nature, but must be
manufactured from lithium in a reactor according to the reaction:

n þ 6 Li ! 4He þ T þ 4:8 MeV (4)

The second is whether, even if fusion does occur, enough power is prod-
uced for sonoluminescence to be a net energy source. After all, production of

Table 7 A list of testing variables for SBSL and MBSL

sonoluminescence experiments

Measurement conditions Sonoluminescence characterizations

Test cell geometry Flash intensity

Ambient conditions Three dimension temperature map
Test liquid and dissolved gasses Emission spectrum
Applied power and frequency High speed and low light photographs
Bubble location and motion Light scattering of bubble
Bubble creation method Precision timing of flash duration
Chemical and nuclear reaction products
 INVESTIGATING SONOLUMINESCENCE 633

sonoluminescence is a power intensive process; only a fraction of the electrical

power is converted into vibrational power in the piezo crystals, and only a fraction
of the vibrational power is transmitted into the liquid. Intuitively, the power required
for sonoluminescence should be less in small vessels; more power would be required
to initiate sonoluminescence in Lake Erie than in a 100-ml flask. One would think
that if the wave amplitude diminished with distance, perhaps because of dissipation
within the liquid, that the power required would scale as some characteristic dimen-
sion r. The frequency is inversely proportional to r, so if the dissipation is frequency
dependant the overall question of how the power required varies with size needs to be
investigated. From the result in Eq. (3), the minimum container diameter is of the
order of a few millimeters. In this case, one could envision a piezo-walled honey-
comb arrangement, with each small cell holding a bubble; some arrangement such
as this is probably the most efficient.
To date, a complete understanding of the nature of sonoluminescence remains
elusive. Sonoluminescence is often difficult to create, difficult to see (in many
cases), difficult to sustain, and difficult to characterize. Reports of days spent
acquiring one data set from a single event abound. However, difficulty should
not deter us from pursuing and attempting to understand the unknown. Facing
the difficulties of characterizing sonoluminescence and attempting to overcome
them may lead to more practical applications of this phenomenon with the
acquisition of new data. For “It is likely that those doing absurd experiments
will be the ones to make discoveries” [61].

VIII. Conclusions
Sonoluminescence, the generation of light from cavitation in fluids, has been
associated with claims of energy production. Instrumentation techniques were
explored at NASA Glenn Research Center to measure properties of the
phenomenon of sonoluminescence.
A resonating test cell and a Sonicator test cell were built to generate
sonoluminescence. Multi-bubble sonoluminescence in the tests cells was imaged
with low-light cameras in a variety of containers. Indications of high temperature
generation were observed in palladium chromium alloy films when exposed to
sonoluminescence in heavy water but not in light water. No indication of high-temp-
erature generation was observed in platinum films exposed to sonoluminescence in
heavy or light water, though the platinum film was estimated to require about the
same amount of energy to be modified as the palladium chromium film.
Localized bright sonoluminescence was generated in glycerin saturated with
water, allowing future spectroscopic and other optical investigations. A design
was presented for in situ radiation monitoring of the sonoluminescence, with
plans for future improvements. A concept was presented for harvesting the
effects seen in this study as electricity, however there is clear need for improving
the thermoelectric properties of the thin films for practical energy harvesting.
Characterization efforts, such as searching for byproducts of chemical and
nuclear reactions, completely mapping the temperature distribution in a
sonoluminescence cell, and spectroscopy, all contribute to a more complete
understanding of the nature of sonoluminescence. From these efforts, a clear
understanding of the heat generation process involved may lead to the successful
application of sonoluminescence as a means of energy harvesting.
634 J. D. WRBANEK ET AL.

Acknowledgments
This work was sponsored over the past several years through the Green
Propulsion and Cryogenic Advanced Development, the Low Emissions Alterna-
tive Power, and the Breakthrough Propulsion Physics projects at the NASA
Glenn Research Center. The authors thank Kenneth Weiland (retired) and
James Williams of the R&D Labs Technical Branch for their optical hardware
and electronics support in this effort. We also thank Drago Androjna of Sierra
Lobo, Inc. (retired) and José Gonzalez of Gilcrest Electric as members of the
NASA Glenn Research Center Test Facilities Operation, Maintenance, and
Engineering (TFOME) organization for laboratory support. Finally, we are
thankful to Jonathan Wright of University of Florida for his input in this work,
and Carl Chang of ASRC Aerospace Corporation at the NASA Glenn Research
Center for his technical review.

References
[1] Putterman, S. J., “Sonoluminescence: Sound into Light,” Scientific American,
February 1995, pp. 32 37.
[2] Frenzel, H., and Schultes, H., “Luminescenz im ultraschallbeschickten Wasser,”
Zeitschrift fur Physikalische Chemie, Vol. B27, 1934, p. 421.
[3] Gaitan, D. F., Crum, L. A., Church, C. C., and Roy, R. A., “Sonoluminescence and
Bubble Dynamics for a Single, Stable, Cavitation Bubble,” Journal Acoustics of
Society of America, Vol. 91, 1992, pp. 3166 3183.
[4] Moran, M. J., Haigh, R. E., Lowry, M. E., Sweider, D. R., Abel, G. R., Carlson, J. T.,
Lewia, S. D., Atchley, A. A., Gaitan, D. F., and Maruyama, X. K., “Direct
Observation of Single Sonoluminescence Pulses,” Nuclear Instrumentation and
Methods in Physical Reseach B, Vol. 96, No. 3 4, May 1995, pp. 651 656.
[5] Hiller, R., Putterman, S. J., and Barber, B. P., “Spectrum of Synchronous Picosecond
Sonoluminescence,” Physical Review Letters, Vol. 69, No. 8, August 1992,
pp. 1182 1184.
[6] Hiller, R., Weninger, K., Putterman, S. J., and Barber, B. P., “Effect of Noble
Gas Doping in Single Bubble Sonoluminescence,” Science, Vol. 266, 1994,
pp. 248 250.
[7] Hiller, R. A., and Putterman, S. J., “Observation of Isotope Effects in
Sonoluminescence,” Physical Review Letters, Vol. 75, No. 19, Nov. 1995,
pp. 3549 3551.
[8] Didenko, Y. T., McNamara, W. B., and Suslick, K. S., “Temperature of Multibubble
Sonoluminescence in Water,” Journal of Physical Chemistry, Vol. 103, No. 50, Dec.
1999, pp. 10783 10788.
[9] Barber, B. P., Wu, C. C., Lofstedt, R., Roberts, P. H., and Putterman, S. J.,
“Sensitivity of Sonoluminescence to Experimental Parameters,” Physical Review
Letters, Vol. 72, No. 9, Feb. 1994, pp. 1380 1383.
[10] Prosperetti, A., “A New Mechanism for Sonoluminescence,” Journal of the
Acoustical Society of America, Vol. 101, 1997, pp. 2003 2007.
[11] Prosperetti, A., “Old Fashioned Bubble Dynamics,” Sonochemistry and
Sonoluminescence, Crum, L. A., Mason, T. J., Reisse, J. L., and Suslick, K. S.
(eds.), Kluwer, 1999, pp. 159 164.
 INVESTIGATING SONOLUMINESCENCE 635

[12] Eberlein, C., “Sonoluminescence as Quantum Vacuum Radiation,” Physical Review

Letters, Vol. 76, No. 20, May 1996, pp. 3842 3845.
[13] Brodsky, A. M., Burgess, L. W., and Robinson, A. L., “Cooperative Effects in
Sonoluminescence,” Physics Letters A, Vol. 287, Sep. 2001, pp. 409 414.
[14] Schiffer, M., “Sonoluminescence: The Superradiance Paradigm,” 14 Oct. 1997.
URL: http://arxiv.org/pdf/quant ph/9710039 [cited October 2002].
[15] Moss, W. C., Clarke, D. B., and Young, D. A., “Star in a Jar,” Sonochemistry and
Sonoluminescence, Crum, L. A., Mason, T. J., Reisse, J. L., and Suslick, K. S.
(eds.), Kluwer, 1999, pp. 159 164.
[16] Yasui, K., “Effect of Liquid Temperature on Sonoluminescence,” Physics Review E,
Vol. 64, June 2001.
[17] Yasui, K., “Temperature in Multibubble Sonoluminescence,” Journal of Chemical
Physics, Vol. 115, No. 7, Aug. 2001, pp. 2893 2896.
[18] Margulis, M. A., and Margulis, I. M., “Contemporary Review on Nature of Sono
luminescence and Sonochemical Reactions,” Ultrasonics Sonochemistry, Vol. 9,
2002, pp. 1 10.
[19] Flynn, H. G., “Method of Generating Energy by Acoustically Induced Cavitation
Fusion and Reactor Therefor,” U.S. Patent 4,333,796, June 1982.
[20] Moss, W. C., Clarke, D. B., White, J. W., and Young, D. A., “Sonoluminescence and
the Prospects for Table Top Micro Thermonuclear Fusion,” Physics Letters A,
Vol. 211, No. 2, Feb. 1996, pp. 69 74.
[21] Putterman, S. J., Barber, B. P., Hiller R. A., and Lofstedt, R. M. J., “Converting Acous
tic Energy into Useful Other Energy Forms,” U.S. Patent 5,659,173, Aug. 1997.
[22] Crum, L. A., “Sonofusion: Star in a Jar?” Science Forum Colloquium Series, Univer
sity of Washington, 21 Nov. 2003. URL: http://www.washington.edu/research/
scienceforum/pdfs/Crum.pdf [cited December 2003].
[23] Nomura, S., and Toyota, H., “Sonoplasma Generated by a Combination of Ultrasonic
Waves and Microwave Irradiation,” Applied Physics Letters, Vol. 83, No. 22,
Dec. 2003, pp. 4503 4505.
[24] Didenko, Y. T., and Suslick, K. S., “The Energy Efficiency of Formation of Photons,
Radicals and Ions During Single Bubble Cavitation,” Nature, Vol. 418, July 2002,
pp. 394 397.
[25] Schewe, P., and Stein, B., “Detecting Megasonic Bubbles on Computer Chips,”
Physics News Update, No. 710, American Institute of Physics, Nov. 24, 2004.
[26] Doktycz, S. J., and Suslick, K. S., “Interparticle Collisions Driven by Ultrasound,”
Science, Vol. 247, March 1990, pp. 1067 1069.
[27] Dhas, N. A., Zaban, A., and Gedanken, A., “Surface Synthesis of Zinc Sulfide
Nanoparticles on Silica Microspheres: Sonochemical Preparation, Characterization,
and Optical Properties,” Chemistry of Materials, Vol. 11, No. 3, Feb. 1999, pp. 806 813.
[28] Jorné, J., “Ultrasonic Irradiation of Deuterium Loaded Palladium Particles
Suspended in Heavy Water,” Fusion Technology, Vol. 29, Jan. 1996, pp. 83 90.
[29] George, R., “Photographic Evidence for Micronuclear Explosions in Thin Metal Foils
Initiated by Intense Ultrasonic Cavitation,” Scanning, Vol. 19, No. 3, 1997, p. 196.
[30] Stringham, R., “Anomalous Heat Production by Cavitation,” Proceedings of the
1998 IEEE Ultrasonics Symposium, IEEE, 1998, pp. 1107 1110.
[31] Taleyarkhan, R. P., West, C. D., Cho, J. S., Lahey, Jr., R.T., Nigmatulin, R. I., and
Block, R. C., “Evidence for Nuclear Emissions during Acoustic Cavitation,”
Science, Vol. 295, March 2002, pp. 1868 1873.
636 J. D. WRBANEK ET AL.

[32] Taleyarkhan, R. P., Cho, J. S., West, C. D., Lahey, Jr., R. T., Nigmatulin, R. I., and
Block, R. C., “Additional Evidence of Nuclear Emission during Acoustic Cavita
tion,” Physical Reviews E, Vol. 69, No. 3, March 2004, 036109.
[33] Shapira, D., and Saltmarsh, M., “Nuclear Fusion in Collapsing Bubbles Is It There?
An Attempt to Repeat the Observation of Nuclear Emissions from Sonolumines
cence,” Physical Review Letters, Vol. 89, No. 10, Sept. 2002, 104302.
[34] Taleyarkhan, R. P., West, C. D., Lahey, Jr., R. T., Nigmatulin, R. I., Block, R. C., and
Xu, Y., “Nuclear Emissions during Self Nucleated Acoustic Cavitation,” Physical
Review Letters, Vol. 96, No. 3, Jan. 2006, 034301.
[35] Forringer, E. R., Robbins, D., and Martin, J., “Confirmation of Neutron Production
during Self Nucleated Acoustic Cavitation,” Transactions of the American Nuclear
Society, Nov. 2006, pp. 736 737.
[36] Lipson, A. G., “Comment on ‘Nuclear Emissions during Self Nucleated Acoustic
Cavitation,’ ” Physical Review Letters, Vol. 97, Oct. 2006, 149401.
[37] Naranjo, B., “Comment on ‘Nuclear Emissions during Self Nucleated Acoustic
Cavitation,’ ” Physical Review Letters, Vol. 97, Oct. 2006, 149403.
[38] Chubb, S., “An Overview of Cold Fusion Theory,” Thermal and Nuclear Aspects of
the Pd/D2O System, SSC TR 1862, Szpak, S., and Mossier Boss, P. A. (eds.), Spawar
Systems Center, San Diego, CA, Feb. 2002, Vol. 1, pp. 91 111.
[39] Okuyama, F., Muto, H., Tsujimaki, H., and Fujimoto Y., “Palladium Nanoparticles
Grown by Hydrogen and Deuterium Ion Bombardment,” Surface Science,
Vol. 355, March 1996, pp. L341 L344.
[40] Young, F. R., “Multibubble Sonoluminescence,” Sonoluminescence, CRC Press,
Boca Raton, FL, 2005, pp. 27 66.
[41] Flint, E. B., and Suslick, K. S., “Sonoluminescence from Nonaqueous Liquids:
Emission from Small Molecules,” Journal of the American Chemistry Society,
Vol. 111, No. 18, 1989, pp. 6987 6992.
[42] Suslick, K. S., and Doktycz, S. J., “The Sonochemistry of Zn Powder,” Journal of the
American Chemistry Society, Vol. 111, No. 61989, pp. 2342 2344.
[43] Suslick, K. S., Didenko, Y., Fang, M. M., Hyeon, T., Kolbeck, K. J., McNamara, III,
W. B., Mdleleni, M. M., and Wong, M., “Acoustic Cavitation and its Chemical Con
sequences,” Philosophical Transactions of the Royal Society of London, Series A,
Vol. 357, No. 1751, Feb. 1999, pp. 335 353.
[44] Freeh, J. E., Steffen, Jr., C. J., and Larosiliere, L. M., “Off Design Performance
Analysis of a Solid Oxide Fuel Cell/Gas Turbine Hybrid for Auxiliary Aerospace
Power,” NASA TM 2005 213805, Dec. 2005.
[45] Thaller, L. H., and Zimmerman, A. H., “Overview of the Design, Development, and
Application of Nickel Hydrogen Batteries,” NASA TP 2003 211905, June 2003.
[46] Cohen, F., and Dalton, P., “Update on International Space Station Nickel Hydrogen
Battery On Orbit Performance,” NASA TM 2003 212542, Aug. 2003; AIAA paper
2003 6065, Aug. 2003.
[47] Nigmatulin, R. I., Akhatov, I. Sh., Vakhitova, N. K., and Lahey, Jr., R.T.,
“Hydrodynamics, Acoustics and Transport in Sonoluminescence Phenomena,”
Sonochemistry and Sonoluminescence, Crum, L. A., Mason, T. J., Reisse, J. L.,
and Suslick, K. S. (eds.), Kluwer, 1999, pp. 127 138.
[48] Matula, T. J., “Single Bubble Sonoluminescence in Microgravity,” Ultrasonics,
Vol. 38, No. 1 8, March 2000, pp. 559 565.
 INVESTIGATING SONOLUMINESCENCE 637

[49] Wrbanek, J. D., Fralick, G. C., and Wrbanek, S. Y., “Development of Techniques to
Investigate Sonoluminescence as a Source of Energy Harvesting,” NASA TM 2007
214982, Sep. 2007; AIAA 2007 5596, July 2007.
[50] Lauterborn, W., and Ohl, C. D., “Acoustic Cavitation and Multi Bubble
Sonoluminescence,” Sonochemistry and Sonoluminescence, Crum, L. A., Mason,
T. J., Reisse, J. L., and Suslick, K. S. (eds.), Kluwer, 1999, pp. 97 104.
[51] Hodgman, C. D., Weast, R. C., and Selby, S. M. (eds.), Handbook of Chemistry and
Physics, The Chemical Rubber Publishing Co., Cleveland, OH, 1960.
[52] Beckett, M. A., and Hua, I., “Impact of Ultrasonic Frequency on Aqueous
Sonoluminescence and Sonochemistry,” Journal of Physical Chemistry A,
Vol. 105, No. 15, April 2001, pp. 3796 3802.
[53] Hulse, C. O., Bailey, R. S., Grant, H. P., Anderson, W. L., and Przybyszewski, J. S.,
“High Temperature Static Strain Gauge Development,” NASA CR 189044, Aug. 1991.
[54] Kubaschewski, O., and Hopkins, B. E., Oxidation of Metals and Alloys,
Butterworths, London, 1967, pp. 4 18.
[55] Campbell, D. S., “Mechanical Properties of Thin Films,” Handbook of Thin
Film Technology, Maissel, L. I., and Glang, R. (eds.), McGraw Hill, New York,
1970, pp. 12 3 12 50.
[56] Berger, M. J., Coursey, J. S., and Zucker, M. A., “ESTAR, PSTAR, and ASTAR:
Computer Programs for Calculating Stopping Power and Range Tables for
Electrons, Protons, and Helium Ions (version 1.2.2),” 2000; URL: http://physics.
nist.gov/Star [cited 15 September 2004]; NIST, Gaithersburg, MD.
[57] Sears, V. F., “Neutron Scattering Lengths and Cross Sections,” Neutron News, Vol.
3, No. 3, 1992, pp. 26 37; URL: http://www.ncnr.nist.gov/resources/n lengths/
list.html, NIST, Gaithersburg, MD [cited 02 January 2004].
[58] Ziegler, J. F., and Beirsack, J. P., SRIM, The Stopping and Range of Ions in Matter,
Software Package, Ver. 2003.26, SRIM.org, Annapolis, MD, 2004.
[59] Wrbanek, J. D., Wrbanek, S. Y., Fralick, G. C., and Chen, L. Y., “Micro Fabricated
Solid State Radiation Detectors for Active Personal Dosimetry,” NASA TM 2007
214674, Feb. 2007.
[60] Wrbanek, J. D., Fralick, G. C., Farmer, S. E., Sayir, A., Blaha, C. A., and Gonzalez,
J. M., “Development of Thin Film Ceramic Thermocouples for High Temperature
Environments,” NASA TM 2004 213211, Aug. 2004, also AIAA 2004 3549, July
2004.
[61] Lane, J. W., “Our Children Will Not Be Us,” Letter to APS News, Aug./Sept. 2007,
Vol. 16, No. 8, p. 4.
 Chapter 20

Null Tests of “Free Energy” Claims

Scott R. Little
EarthTech International, Austin, Texas

I. Introduction
HERE IS no question that our hydrocarbon-based energy system is slowly
T breaking down. The supply of fossil fuel is finite and combustion products
are steadily polluting our environment. Conventional nuclear energy also faces
challenges: undesirable waste products are produced, there is significant
public concern about the safety of nuclear power plants, the supply of fuel is
finite, and the technology is not suitable for most transportation applications.
But these terrestrial problems pale in comparison to the likely energy require-
ments for interstellar travel. For example, consider a hypothetical mission to
Alpha Centauri in a ship the size of a Boeing 757 (i.e., 105 kg). Acceleration
is 1 g to the midpoint and deceleration is 1 g until arrival at the destination 4.3
light-years from Earth. At the midpoint, the ship is moving at 0.95 c. For the pas-
sengers the trip takes only 3.6 years whereas 5.9 years go by for those who stay on
Earth. A yet-to-be-developed engine that converts its fuel entirely to energy and
beams it out the tailpipe drives the ship with the maximum possible fuel effi-
ciency. Despite this efficiency, the trip requires a staggering 3.8  106 kg of
fuel: 38 times the weight of the ship. The energy required for this ideal one-
way trip to Alpha Centauri is about 700 times the present (2008) annual
energy consumption of the world. (See Chapters 2 and 3 for additional infor-
mation on spaceflight energy requirements.)
Faced with such monumental problems humans naturally devise schemes to
solve them. Often, the scheme involves a device that is purported to produce
more energy than it takes to run it. Sometimes the inventor naively thinks the
device is simply creating the extra energy. More often the inventor believes
that the device is tapping a new source of energy.

Copyright # 2008 by the American Institute of Aeronautics and Astronautics, Inc. All rights
reserved.

Experimentalist.

639
640 S. R. LITTLE

Breakthrough energy claims are nothing new. Recorded history of perpetual

motion claims begin in the 13th century with a simple mechanical device attrib-
uted to Wilars de Honecort [1]. Leonardo da Vinci toyed with the idea of a
hydraulic perpetual motion machine [2]. In the 1700s Johann Bessler (a.k.a.
Orffyreus) developed a mechanical perpetual motion machine [3] that was
widely witnessed yet remains shrouded in mystery to this day. There are
numerous other devices, primarily invented by charismatic individuals with no
particular scientific training. As science progressed toward formal recognition
of the laws of thermodynamics in the mid 1800s, claims to perpetual motion
became considerably less acceptable but hardly less frequent. In 1870, Henry
Dircks eloquently described claimants to perpetual motion as follows, “A more
self-willed, self-satisfied, or self-deluded class of the community, making at
the same time pretension to superior knowledge, it would be impossible
to imagine. They hope against hope, scorning all opposition with ridiculous
vehemence, although centuries have not advanced them one step in the way of
progress” [4].
Fortunately, things have changed. In our present age of science and
technology almost everyone accepts the laws of thermodynamics. The majority
of new energy claims are therefore based upon the idea of tapping a new
source of energy. It is these claims that are the primary focus of this chapter.

II. Testing of Energy Claims

Testing of energy devices is conceptually straightforward. The output energy
of the device is measured, the apparent input energy is measured, and the two
quantities are compared. If the device is working as claimed, that is, extracting
energy from some unexpected source and delivering it to the output, the
measured output energy will exceed the apparent input energy. Otherwise,
the First Law of Thermodynamics requires that the output energy, including
any losses such as heat, must be precisely equal to the input energy.
In the case of an electrical input, measurement of the input energy is relatively
straightforward, particularly if the input is steady DC. If the input has temporal
variations, a sophisticated power analyzer may be required to achieve satisfactory
measurement accuracy. These instruments sample the voltage and current being
delivered to a device rapidly and simultaneously. Each pair of samples is multi-
plied together to obtain a measure of the instantaneous power flowing into the
device. These values are then integrated over time to obtain the electrical
energy consumed by the device. In the case of mechanical input like a rotating
shaft, a dynamometer, which measures torque and angular velocity, is required
to directly measure the mechanical input energy.
Measurement of the output energy is often more challenging. Devices that
produce an electrical output can be handled by the methods outlined above for
input energy measurement. Devices that output heat energy require some form
of calorimetry. Calorimetry is conceptually simple but, in practice, a great
effort is required to reduce errors, primarily systematic, to acceptable levels. It
is especially difficult to achieve accuracy levels better than 1% relative.
Compared to a calorimeter with 1% relative accuracy, at least an order of
magnitude more effort is required to achieve 0.1% accuracy [5].
 NULL TESTS OF “FREE ENERGY” CLAIMS 641

A frustrating situation arises when investigating energy claims. To the clai-

mant, a single null test often proves almost nothing. Regardless of the circum-
stances one can always say that the test was not conducted properly that the
right materials were not used, the apparatus was not assembled correctly, or
the planets were not properly aligned. Strictly speaking, from the claimant’s
viewpoint, an infinite number of null tests are required to disprove a claim. A
relatively small number of positive tests suffice to prove a claim.
With this in mind, the ideal condition under which to test an energy claim is
with the full cooperation of the claimant using the original apparatus. The testing
proceeds rapidly and, in the event of null results, the claimant can ensure as far as
possible that the tests were conducted properly. If the original apparatus is not
available, a new apparatus must be constructed. In this case, even with the full
cooperation of the claimant, there is more room for excuses should the tests
yield null results. If the claimant is noncooperative and the apparatus must be
constructed from various documents and records, very little can be proved if
the tests are null.

III. Some Tests of Breakthrough Energy Claims

This section contains accounts of some tests of breakthrough energy claims
conducted at Earthtech International (http://www.earthtech.org). Although the
results were negative, we present these cases primarily to promote better under-
standing of the problems involved in such testing. They are not to be taken as
indisputable refutation of the claims.

A. Zero-Point Energy Devices

Quantum physics predicts the existence of an electromagnetic zero-point field
whose energy density is so large that many physicists designate it as virtual.
Others treat the zero-point field as real energy that surrounds and pervades every-
thing leaving ordinary matter as an almost negligible foam riding on this vast sea
of energy. According to John A. Wheeler et al. [6], “. . . elementary particles
represent a percentage-wise almost completely negligible change in the locally
violent conditions that characterize the vacuum.” This viewpoint has led to
intense speculation about the possibility of utilizing zero-point energy. (See
Chapter 18 for broader discussions of this possibility.)
There is a real force that can be attributed to the zero-point field: the Casimir
force [7]. Experimentally confirmed [8], this force arises when conductive sur-
faces are placed in very close proximity, thus creating a cavity that eliminates
certain electromagnetic modes between the plates. The result is an imbalance
in the radiation pressure on the two sides of each plate [9], which produces a
net force that pushes the plates together. Some physicists, including Robert
Forward [10], have proposed that the Casimir force provides a means of extract-
ing energy from the zero-point field. Julian Schwinger [11] provided further
stimulus by suggesting that the energy released in sonoluminescence was due
to Casimir forces acting on the collapsing bubbles. (Chapter 19 discusses
additional sonoluminescence tests and testing methods.)
642 S. R. LITTLE

Our first and most extensive campaign to extract energy from the zero-point
field was an effort to replicate the energy claims associated with Ken Shoulders’
charge clusters [12], or EVs (electrum validum; i.e., strong electron) as they are
popularly known. Shoulders believes that at least one EV is formed in every
spark discharge. EVs are supposed to contain at least 109 electrons and exist
only during the transit from cathode to anode. Shoulders claims that sharply
pointed cathodes and a fast rise time for the applied high voltage pulse
promote EV formation. The connection with zero-point energy comes from
the hypothesis that the compression of electrons into a charge cluster is due
to attractive Casimir forces overpowering repulsive Coulomb forces at very
short range. Shoulders did a great deal of experimentation with EVs and
claimed in US Patent 5,018,180 (21 May 1991) to have observed 96 times
more energy released by an EV than required to produce the EV. We
pursued this claim for years and were never able to reproduce Shoulders’
results. We experimented with numerous configurations in an effort to
observe the direct electrical energy output that Shoulders had claimed in his
patent. Failing that, we attempted calorimetric measurements that were of com-
pounded difficulty because of the low energy levels involved, the difficulty of
accurately measuring the input energy delivered to the spark discharge, and
the overall difficulty of making sensitive measurements in such an electrically
noisy environment. Despite these problems we eventually managed to obtain
reasonable accuracy and reliability in our calorimetric measurements of
EVs and the results were uniformly negative. (See Chapter 18 for additional
information on EVs.)
Schwinger’s hypothesis led us to give serious consideration to several
cavitation-based energy claims. The Potapov device, invented by Yuri Potapov
in Moldavia, is an example. The Potapov device consisted simply of a swirl
chamber through which water was pumped vigorously to create a vortex and
cavitation. Potapov claimed that his device imparted up to 3 times more heat
energy to the water than the mechanical energy required to pump the water
through it. In this case we were able to obtain a genuine Potapov device for
testing and we had limited cooperation from Potapov himself. We constructed
a batch calorimeter system in which the device would be operated for a certain
period of time to heat up the water contained in a large insulated reservoir.
Water was pumped from the reservoir, through the device, and returned to the
reservoir. The pump was driven by an electric motor. For the input energy we
simply measured the electrical energy required to drive the motor using a
3-phase watt-hour meter. For the output energy we measured the increase in
water temperature and used the total weight of water in the tank to compute
the heat energy delivered to the water. Instead of the 300% efficiency
claimed for the Potapov device we observed only 80% at best. As a control,
we also measured the heating efficiency of a simple gate valve inserted in the
flow path in place of the Potapov device and adjusted to provide about
the same flow restriction. The valve heated the water just as efficiently as the
Potapov device. The testing went on for months as we struggled to communicate
with Potapov. He did not think we were operating his device properly and we
made numerous modifications at his request. The test results remained uniformly
negative (see http://www.earthtech.org/experiments).
 NULL TESTS OF “FREE ENERGY” CLAIMS 643

Another device (whose specific identity is protected by a nondisclosure

agreement with the developer) involved a motor-driven rotor in a close-fitting
housing. Water was forced through the gaps around the rotor where intense cavi-
tation was supposed to occur. This device was claimed to impart up to 50% more
heat energy to the water flowing through it than the mechanical energy required
to drive it. We tested this device using a larger version of the same batch calori-
meter described above. In this case we constructed a cradle dynamometer to
directly measure the mechanical input energy. With a 30-hp electric motor and
copious generation of steam by the device, this was a very exciting and some-
times dangerous experiment. However, our measurements never showed any
sign of excess energy. Furthermore, by comparing mechanical input energy to
heat output energy we were able to obtain a near-perfect energy balance in our
measurements, typically 99 + 1%. We were fortunate to have significant
cooperation from the developer of this device and, during a visit to our
facility, we accidentally discovered the source of most of his anomalous readings:
improper usage of his electrical power analyzer.
A related claim is that of sonofusion made primarily by Roger Stringham [13].
We first investigated this claim by constructing our own apparatus without
requesting any cooperation from Stringham. The apparatus consists of an ultra-
sonic transducer immersed in heavy water with a palladium target in close proxi-
mity. According to Stringham, more heat energy is produced in the apparatus
than the acoustic energy delivered by the transducer. We employed a water-
flow calorimeter to measure the heat output energy and we spent a great deal
of effort to learn how to accurately measure the high-voltage, 20-kHz, low
power-factor electrical input energy being delivered to the piezoelectric transdu-
cer. We evaluated several power analyzers and selected the Clarke-Hess 2330,
which we found to be significantly superior to other instruments with similar
accuracy specifications. We conducted a total of 48 runs with our apparatus,
12 of which used palladium targets. We never saw any sign of excess heat.
After communicating our results to Stringham we arranged to visit his laboratory
with a portable version of our calorimeter and our Clarke-Hess 2330. In other
words, we were given a chance to test his claim using his apparatus with his
full cooperation. We succeeded only in demonstrating that his input power
measurements were understating the actual input power. That was the cause of
his apparent excess heat on that day in 1999. More recently, Stringham has
explored this phenomenon with considerably reworked apparatus and claims
even higher excess energy production [14].
H. E. Puthoff [15] has calculated that the ground state of the hydrogen atom
can be explained as a dynamic balance between energy lost by the electron
due to acceleration radiation and energy absorbed from the zero-point field.
The fact that the space between Casimir plates is a region where the zero-point
field is reduced in energy density led to speculation that hydrogen might lose
some of its ground state energy when placed in such a cavity. If that were the
case, then that energy release would constitute zero-point energy conversion
and a circulation of hydrogen into and out of a Casimir cavity might produce a
continuous extraction of energy from the zero-point field. We have designed
and constructed several experiments to explore this hypothesis but without
success so far. Most of these experiments were attempts to detect heat energy
644 S. R. LITTLE

being released by hydrogen flowing through some form of Casimir cavity. We

first tried constructing cavities out of precision optical flats. For the ground
state of molecular hydrogen, the optimum cavity spacing is about 1 mm. We
monitored the gas temperature at the entrance and exit of the cavity looking
for signs of heating due to the release of ground state energy. Although a
small temperature increase in this experiment was observed, it turned out to be
due only to the Joule Thompson effect that, for hydrogen, results in a
warming of the gas as it flows through a restriction. We also tried using finely
powdered metals such as platinum and palladium to create a dense labyrinth of
Casimir scale passages. These experiments tended to produce an exciting
initial burst of heat when the flow of H2 was started. But the burst always
faded away after a minute or two and it could not be readily repeated. We
finally tracked this down to H2 þ O2 combustion catalyzed by the finely pow-
dered metal. It was not readily repeatable because the apparatus was nearly
sealed. Only after the apparatus had been sitting overnight or had been disas-
sembled was there sufficient O2 present for another heat burst.
In a different approach to testing this hypothesis we put hydrogen molecules in
a Casimir cavity and used absorption spectroscopy to look for a shift in their
ground state energy, which follows from the assumption that a depression of
the ground state energy would produce a corresponding increase in the dis-
sociation energy. Early attempts at this experiment were conducted at the
Synchrotron Radiation Center of the University of Wisconsin-Madison. Follow-
ing in the footsteps of molecular spectroscopy pioneer Gerhard Herzberg [16],
extreme ultraviolet radiation was used to probe the dissociation energy of H2
molecules in an appropriate Casimir cavity. Unfortunately we did not find any
evidence of a ground state shift in this work. (See http://www.earthtech.org/
experiments/src/srcreport.htm for full details of our synchrotron experiment.)
Further effort along this line by a consortium of researchers is planned [17].
The theory and operating principles behind this approach are discussed in
more detail in Chapter 18.

B. Electromagnetic Devices
A number of electromagnetic energy claims have been made over the past
150 years. Some of them are no more than a continuation of the quest for perpe-
tual motion but with magnets and coils replacing the weights and levers of the
earlier devices. Others, particularly the more recent claims, are not so easy to
dismiss and deserve to be investigated.
A relatively simple device called the Motionless Electromagnetic Generator
(MEG) (US Patent 6,362,718 26 March 2002) has been widely publicized on
the Internet. We constructed our own MEG using detailed construction plans
from an independent lab that reportedly had successfully replicated the excess
power results. The initial results with our MEG also apparently showed excess
power but we soon learned why. The MEG operates at high audio frequencies
and delivers several hundred volts and a few milliamperes to a load resistor.
This current was being measured using a 10-ohm current-viewing resistor in
series with the load resistor. With only a few milliamperes of current
through this resistor, the voltage developed is only a few 10s of millivolts.
 NULL TESTS OF “FREE ENERGY” CLAIMS 645

At these frequencies it is almost impossible to accurately measure this small

voltage while in intimate proximity to the output voltage, which is four orders
of magnitude higher. Capacitive coupling between the voltmeter or scope leads
significantly elevated the observed voltage across the current viewing resistor
resulting in overestimation of the output power. Supporting this conclusion
was the lack of heating in the load resistor that should have occurred had the
apparent output power been correct. When we changed the current viewing
resistor to 1000 ohms, which did not significantly affect the load impedance,
the apparent excess power disappeared.
The essence of another case, again covered by nondisclosure agreement, can
be presented as a good example of an axiom that we have come to embrace: the
fact that your instruments cost a lot of money does not guarantee that their results
will be accurate. A simple device was being driven by 60 Hz AC power. The
output of this device was also 60 Hz AC but at a different voltage. We employed
a Clarke-Hess 2330 power analyzer for the input and output power measurements
and obtained a mundane efficiency of 90% for the device. However, the
claimant was using a state-of-the-art high-bandwidth digital oscilloscope with
extensive waveform math capabilities (including power calculation) and a soph-
isticated clamp-on AC/DC current probe. Surprisingly, this $30,000 collection of
modern equipment, while capable of accurate power analysis on a wide variety of
waveforms over an impressive range of frequencies, made significant errors in
the measurement of power in these 60 Hz signals.

C. Cold Fusion
In March of 1989 Martin Fleischmann and Stanley Pons of the University of
Utah announced that they had succeeded in making D-D fusion occur in an
electrochemical cell near room temperature [18]. Compared to the ordinary
conditions required for this reaction, this claim was aptly named “cold fusion.”
The announcement of cold fusion generated intense interest as it promised to
solve most if not all of the energy problems on Earth. However, it also received
intense scrutiny as it appeared to violate known principles of nuclear
physics. With widespread failure to replicate the experiment, support for
continued experimentation rapidly waned.
Despite diminished support, a number of scientists continue to investigate cold
fusion. Hundreds of papers reporting positive results have been published (see the
cold fusion library at http://www.lenr-canr.org) and international conferences
are held every couple of years. However, to this day (2008), there exists no
cold fusion demonstration experiment. In other words, the cold fusion phenom-
enon is not sufficiently reproducible that it can be demonstrated on demand. This
situation greatly hampers cold fusion research because it makes application of the
scientific method almost impossible. It is extremely difficult to test hypotheses
when the experimental results are nearly random.
The primary signature of cold fusion is excess heat, which means that the
electrochemical cell produces more heat power than the electrical power used
to stimulate it. Thus calorimetry is often involved in testing cold fusion
experiments. In our laboratory we have expended a great deal of effort on the
development of calorimeters suitable for cold fusion experiments. Over the
646 S. R. LITTLE

years we have had the opportunity to test a relatively small number of cold fusion
cells, some that we constructed ourselves and some that were brought to our lab-
oratory by other investigators who had seen positive signs of excess heat in their
own labs. None of these cold fusion experiments have shown any convincing evi-
dence of excess heat in our calorimeters. We cannot say that we have never seen
any signs of excess heat in our laboratory because all calorimeters drift somewhat
and, inevitably, that drift sometimes goes in a positive direction and looks just
like a low level of genuine excess heat. When that occurs we strive to check
the calorimeter’s calibration as quickly and thoroughly as possible. Usually the
drift in calibration is evident and its magnitude matches, and thus explains, the
apparent excess heat signal. In a few cases the calibration check did not
explain the apparent excess heat signal, but when we returned the cell to the
calorimeter after the calibration check, the excess heat signal did not reappear.
This tantalizing behavior either means that the cell did produce low levels of
excess heat for a while or the calorimeter was simply drifting up and down in
unfortunate synchrony with our observations.
In our laboratory we have about 70 years of combined experience designing,
building, and operating various measuring systems. From this perspective, we
find calorimetry to be unusually susceptible to subtle systematic errors. Further-
more we have found it nearly impossible to anticipate the causes of these errors.
Their elucidation usually occurs only after the instruments have been constructed
and tested extensively.
The culmination of our efforts to build an accurate and reliable calorimeter for
cold fusion experimentation is an instrument we call MOAC (Mother Of All
Calorimeters). This instrument operates on a simple and fundamental principle.
Flowing water is used to extract the heat from the cell. The flow rate is measured
and the temperature rise of the water is measured. The product of the temperature
rise, the flow rate, and the specific heat of water yields the heat power being
extracted from the cell. Despite its simple concept, MOAC is not a simple instru-
ment. Two independent computer-based data acquisition systems monitor a total
of 45 parameters, including 22 separate temperatures. Fourteen analog outputs,
driven by proportional-derivative feedback algorithms, control various critical
parameters. Figure 1 shows a simplified block diagram of the system.
The cold fusion cell (marked CELL in Fig. 1) and heat exchanger are located in
a chamber whose walls are made almost perfectly insulating by a system that heats
the outer surface of each of the six wall panels so that its temperature matches that
of the corresponding inner surface. The inner and outer surface of each wall panel
is composed of a thick aluminum plate for thermal uniformity. Temperatures are
sensed by thermistors embedded in the center of each plate and the outer plates
are heated by electrical heating elements distributed over the surface (shown sche-
matically as long resistors in Fig. 1). Each wall panel is independently controlled
by a servo algorithm in the software. This active insulation ensures that virtually all
of the heat dissipated by the cell leaves the chamber via the flowing water.
A three-stage Peltier temperature regulator (which can add or remove heat
as needed) controls the temperature of the water entering the heat exchanger to
+ 0.00038C. A positive-displacement pump driven by a synchronous motor
powered by a crystal-based oscillator produces an exceedingly stable flow of
about 2.5 gm/s. A flowmeter consisting of an automated batch weighing
 NULL TESTS OF “FREE ENERGY” CLAIMS 647

Fig. 1 Block diagram of MOAC.

system measures the flow rate periodically and typically reports a standard
deviation of only + 0.0005 gm/s (i.e., 0.02% relative). A large well-insulated
enclosure houses the entire system. Air circulates over the calorimetry apparatus
and then is ducted to a two-stage Peltier air-conditioner where its temperature
is regulated to + 0.0018C before it re-enters the enclosure. Figure 2 is a
photograph of the entire system.
MOAC was designed to achieve + 0.1% relative accuracy at the typical input
power level of 10 watts (i.e., equivalent to + 0.01 watts). Accuracy is tested
using a standard electrolysis cell constructed of “inactive” materials (i.e.,
without palladium, lithium, or deuterium). The measured electrical input
power is compared to the measured heat output power. When freshly calibrated
and operating normally, the design accuracy of 0.1% relative is actually
achieved. However, performance can degrade over time, typically drifting up
to 0.03 watts a month after calibration. We believe this drift originates primarily
in the thermistors used to measure inlet and outlet water temperatures.
648 S. R. LITTLE

Fig. 2 Exterior of MOAC and controlling computers.

The space available for the cell is relatively large (about 10 cm  25 cm 

25 cm). The cell sits in a stirred air environment where it is not thermally
clamped to a specific temperature. MOAC exhibits excellent specimen versatility
by producing precisely the same reading regardless of the size, shape, or location
of the heat source.

IV. Conclusions
Testing of breakthrough energy claims is simple in concept but often difficult
in practice. It is most effectively done with the full cooperation of the claimant.
Systematic errors are common. Considerable diligence is required to ensure that
the measurement techniques employed are acceptably free from such problems.
The specific examples discussed here can guide other researchers in avoiding
similar experimental errors.

References
[1] Dircks, H., Perpetuum Mobile, E. & F.N. Spon, London, 1870, p. 1.
[2] Heaton, Mrs. C. W., Leonardo da Vinci and His Works, Kessinger Publishing, 2004,
pp. 154 155.
[3] “Perpetual Motion,” Encyclopedia Britannica, 1971 ed.
[4] Dircks, H., Perpetuum Mobile, E. & F.N. Spon, London, 1870, p. 354.
[5] McCullough, J. P., and Scott, D. W., Experimental Thermodynamics, Calorimetry of
Non reacting Systems, Butterworth, London, Vol. 1, 1968, p. 9.
[6] Misner, C. W., Thorne, K. S., and Wheeler, J. A., Gravitation, W.H. Freeman & Co.,
New York, 1973, p. 1202.
 NULL TESTS OF “FREE ENERGY” CLAIMS 649

[7] Casimir, H. B. G., “On the Attraction Between Two Perfectly Conducting Plates,”
Proceedings of the Koninklijke Nederlandse Akademie ran Wetenschappen, Vol.
51, 1948, pp. 793 796.
[8] Harris, B. W., Chen, F., Mohideen, U., “Precision Measurement of the Casimir Force
Using Gold Surfaces,” Physical Review A, Vol. 62, 2000, 052109/1 5.
[9] Milonni, P. W., Cook, R. J., and Goggin, M. E., “Radiation Pressure from the
Vacuum: Physical Interpretation of the Casimir Force,” Physical Review A, Vol.
A38, 1988, p. 1621.
[10] Forward, R. L., “Extracting Electrical Energy from the Vacuum by Cohesion of
Charged Foliated Conductors,” Physical Review B, Vol. 30, No. 4, 1984, p. 1700.
[11] Schwinger, J., “Casimir Light: The Souce,” Proceedings National Academy Science
USA, Vol. 90, March 1993, pp. 2105 2106.
[12] Shoulders, K. R., EV: A Tale of Discovery, Jupiter Technologies, Austin TX, 1987.
[13] Stringham, R. S., George, D. R., Tanzella, F. L., Williams, M., “Cavitation Induced
Heat in Deuterated Metals,” EPRI Report TR 108474, March 1998.
[14] Stringham, R., “1.6 MHz Sonofusion Measurement and Model,” Proceedings of the
American Physical Society, Denver, CO, 2007.
[15] Puthoff, H. E., “Ground State of Hydrogen as a Zero Point Fluctuation Determined
State,” Physical Review D, Vol. 35, 1987, p. 3266.
[16] Herzberg, G., “The Dissociation Energy of the Hydrogen Molecule,” Journal of Mol
ecular Spectroscopy, Vol. 33, 1970, pp. 147 168.
[17] Davis, E. W., Teofilo, V. L., Haisch, B., Puthoff, H. E., Nickisch, L. J., Rueda, A.,
and Cole, D. C. “Review of Experimental Concepts for Studying the Quantum
Vacuum Field,” Proceedings of the STAIF 2006: 3rd Symposium on New Frontiers
and Future Concepts, AIP Conference Proceedings, Vol. 813, El Genk, M. S. (ed.),
AIP Press, New York, pp. 1390 1401.
[18] Browne, M. W., “Fusion in a Jar: Announcement by Two Chemists Ignites Uproar,”
New York Times, March 28, 1989.
 Chapter 21

General Relativity Computational Tools and

Conventions for Propulsion

Claudio Maccone

I. Introduction
O ADDRESS the technical challenges of breakthrough spaceflight within
T the sciences of General Relativity (GR) and Quantum Field Theory
(QFT), computer computational tools are required. Calculations in GR and
QFT are extensive and involve complex notational conventions. The time it
takes to do these calculations by hand is prohibitive as is the risk of inducing
transcription errors. In this chapter we make a comparative review of the main
tensor calculus capabilities of the three most advanced and commer-
cially available “symbolic manipulator” tools. We also address the challenge
of the different conventions in tensor calculus that make it difficult or imposs-
ible to compare results obtained by different scholars in GR and QFT. To
proceed, conventions that would be useful for space propulsion research are
suggested, and then reviews of the software that take these options into
account follow.

A. Computational Conventions
Mathematical physicists, experimental physicists, and engineers have each
their own way of customizing tensors, especially by using the different metric
signatures, different metric determinant signs, different definitions of the basic
Riemann and Ricci tensors, and by adopting different systems of physical
units. This inconsistency hampers progress when trying to apply the advances
from one area to another.
To understand this point a little better, a comparison can be made with the
situation in Europe before the French revolutionaries adopted (and imposed)
the metric system; that is, each country had its own system of units, and a

Copyright # 2008 by Claudio Maccone. Published by the American Institute of Aeronautics and
Astronautics, Inc., with permission.

Retired scientist.

651
652 C. MACCONE

considerable amount of time had to be spent just to make the various results
numerically comparable.
In this chapter we examine the major convention choices and suggest which
choices are preferred for further exploring propulsion physics.

B. History of Computational Tool Development

NASA inaugurated the use of symbolic manipulators (called computer algebra
codes, or symbolic mathematics codes) in the early 1960s. NASA’s goal at that
time was to check the validity of a number of analytical results in celestial
mechanics that had been found by hand calculation in the previous 300 years.
NASA decided to solve the problem by creating a new code from scratch
called Macsyma, a lisp-written symbolic manipulator developed by the Artificial
Intelligence Laboratory of MIT from 1965 through 1982, and later taken over by
private corporations for further developments.
In the 1980s Macsyma was endowed with a tensor calculus package, the first
software in history capable of handling long analytical expressions for the
Riemann, Ricci, and stress-energy tensors required to compute analytical
solutions of both the Einstein and combined Einstein Maxwell equations.
Other products followed and these options are examined here in the context of
how well they match the recommended propulsion conventions.
When “C” became the standard programming language, two new research
companies produced new codes, which have since been commercially available:
“Mathematica,” developed by Wolfram Research, Inc., and “Maple,” developed
by a team of the University of Waterloo (Canada).
Initially, both these codes did not have a tensor package, but several tensor
add-on packages were created in the 1990s:
1) “MathTensor,” written by Leonard Parker and Steven M. Christensen,
which runs on Mathematica
2) “GRTensorII,” written by Kayll Lake and coworkers Peter Musgrave and
Denis Pollney of Queens University in Kingston, Ontario, which runs on Maple,
currently under development (a smaller version of it runs on Mathematica also)
3) “Ricci,” developed by John M. Lee of Washington University for Mathe-
matica 3.0 (more details are included in Ref. 1)
This chapter examines the above-mentioned codes in the context of their
utility to propulsion science research. The comparison is based on the
experience of explicit calculations in the field of GR and QFT that the author
has performed over the last 15 years when the latest releases of the above
codes became available.

II. Recommended Propulsion Computational Conventions

In both GR and QFT, there are different conventions on the metric signature,
the definition of which indices are contracted in the Riemann tensor to yield the
Ricci tensor, and the use of different systems of units. The options are so complex
that a number of excellent results in GR and QFT obtained by either pure
 GENERAL RELATIVITY COMPUTATIONAL TOOLS 653

mathematicians, theoretical physicists, applied technologists, or engineers are

hardly comparable to each other, thus hampering their applications to research
on spaceflight propulsion.
A good example of these difficulties is offered by the two pages printed
on red paper that opened the famous book Gravitation by Misner, Thorne,
and Wheeler in its first 1973 edition [2]. In these two pages the authors list 34
authoritative textbooks of GR that were published between 1922 and 1973 and,
for each book, they neatly specify: 1) the metric signature; 2) the definition of
the Riemann tensor; 3) the sign found by contracting Riemann to Ricci; and
4) whether the spacetime four-dimensional indices are denoted by Greek or
Latin letters.
These two pages give a good idea about the “mess” that freedom of arbitrary
conventions created in a century of GR.
Because propulsion science is in the midst of changes, now is a good time to
implement a set of standard conventions. To that end, here are suggested choices
for conventions and why each is recommened. The key convention choices are as
follows: sytem of units, sign convention for the metric, sign convention for the
determinant of the metric, definition of Riemann tensor in terms of Christoffel
symbols and their derivatives, and definition of which indices are used when con-
tracting the Riemann tensor into the Ricci tensor.

A. System of Units
While different branches of physics have their preferred units (e.g., esu, emu,
Gaussian, Heaviside Lorentz, geometrical, and Planck systems), using a mix of
these would impede comparisons. When exploring science to seek operating
principles for new space propulsion technology, it would be useful to adopt the
system of units most prevalent in engineering professions and ensure that the
equations explicitly list the pertinent natural constants.
Although it is common practice in mathematical physics to treat the fundamental
constants of the speed of light, c, Newton’s gravitational constant, G, and Plank’s
constant, h, as unity (c G h 1), this is not desired for propulsion research.
Although such “geometrized” units (c G h 1) help draw attention to the
power of the mathematical tools, it has the disadvantage of masking the role
played by these fundamental properties of nature. For spaceflight research it is rec-
ommended to explicitly include the natural constants in the equations.
Regarding the system of units, it is recommended to use the MKS system
(meter, kilogram, and second), also called SI for Système International.

B. Sign Convention for Metric

Another convention choice in Special and General Relativity is where to
assign the negative sign to the spatial terms or to the temporal term:

(ds)2 ¼ (c dt)2  x2  y2  z2 vs (ds)2 ¼ (cdt)2 þ x2 þ y2 þ z2 (1)

The version on the left is the convention recommended for spaceflight

considerations, while the convention on the right is often used in General
Relativity textbooks (e.g., Ref. 3). The reason that the version on the left is
654 C. MACCONE

preferred is because all time-like displacements, (ds)2, will be positive. “Time-

like” is the terminology for motion in spacetime that behaves according to the
normal causal relations. Conversely, “space-like” displacements are those that
cover more space in a given time than light-speed propagation will allow. It is
convenient for causal displacements to have a positive value while causality-
violating displacements have a negative value.

C. Sign Convention for Metric Determinant

When proceeding to investigate a particular metric it is necessary to calculate
the inverse of that metric. A step in that process is to find the metric determinant
for which there are two sign conventions, shown below.
   
 g00
 g01 g02 g03   g00
 g01 g02 g03 
 g10
 g11 g12 g13  g g11 g12 g13 
, 0 vs  10 .0 (2)
 g20
 g21 g22 g23   g20 g21 g22 g23 
 g30 g31 g32 g33   g30 g31 g32 g33 

The version on the left is the convention recommended for spaceflight

considerations. The convention on the right is used in some advanced QFTs.
The reason that the version on the left is preferred for spaceflight calculations
is simply because this is more common in ordinary GR and its immediate
generalizations.

D. Definition of Riemann Tensor in Terms of Christoffel Symbols

The Riemann curvature tensor is the mathematical tool for determining the cur-
vature of a spacetime metric by combining terms (Christoffel sysmbols) obtained
from moving a vector (using parallel transport) around a small closed path in
spacetime. There are two conventions to represent the indices in these treatments:

Rlnrs ; Glsn;r  Glrn;s þ Gasn Glar  Garn Glas

vs

Rlnrs ; Gsl n;r þ Grln;s  Gsan Gar

l
þ Gran Gas
l
(3)

The version below is the convention recommended for spaceflight consider-

ations, quite simply because it is the most widely adopted convention to define
the Riemann tensor in the various tensor packages (such as those running on
Macsyma and Mathematica). The convention above is simply the version
below with all the signs reversed.

E. Index Convention when Contracting Riemann to Ricci

Other options in GR deal with the indices on the Ricci tensor, which result
from contracting the Riemann tensor. There are two ways to represent this,
 GENERAL RELATIVITY COMPUTATIONAL TOOLS 655

with the preferred choice for propulsion research on the left:

Rnr ¼ Rlnlr vs Rnr ¼ Rlnrl (4)

Again, the reason that the version on the left is preferred for spaceflight
calculations is because it is the most widely adopted convention in the various
symbolic manipulators.

III. Representative Problems in Propulsion Science

Another factor to address when considering existing and future computational
tools is the nature of the calculations that need to be performed. Some represen-
tative examples include calculating the energy conditions needed to provide warp
drive metrics and calculating the energy conditions needed to provide traversable
wormhole metrics.

IV. Review of Existing Computational Tools

Before discussing the three major available computational tools, it must be
stressed that these tools continue to undergo changes. While the comparisons
here, compiled on 9 September 2007, might change later, the basis of the com-
parisons for their applicablity to space propulsion will remain valid for some
time to come and as new products become available. For example, some of the
newer codes such as Cadabra, Tensorial, and Cartan, and codes related to
Casimir effects, have not yet been assessed for their propulsion research
utility. The following website presented a comparison of various computer
algebra codes in 2007, and might continue to offer up-to-date information:
,http://en.wikipedia.org/wiki/Comparison of computer algebra systems..
To compare the three major software packages with the features desired for
space propulsion science research, Table 1 is offered. Following the table are
more detailed comments on each code. Note that within the major codes,
Maxima, Mathematica, and Maple, there are additional codes that are also
addressed.

A. Maxima
Maxima is a free (i.e., noncommercial) version of Macsyma that was recov-
ered by the late Professor William Shelter (1947 2001). Maxima is based on a
1982 version of Macsyma and is capable of a wide range of mathematical appli-
cations including basic tensors for GR. Volunteers are now further developing its
basic tensor package.
Macsyma, the initial embodiment that lead to Maxima, was an “elementary
math” package written around 1965 at the Artificial Intelligence Laboratory of
the MIT. The codes were designed to provide NASA and the then-ongoing
Apollo program with software capable of checking the analytical results in celes-
tial mechanics that had been piling up in the previous 300 years with no practical
possibility of checking them manually again. For instance, the French astronomer
Charles E. Delaunay (1816 1872) had spent 10 years of his life calculating new
and more accurate analytical results for the motion of the moon. He then spent
 656

Table 1 Summary of computational tools compared to propulsion research preferences (valid as of Sept. 2007)

Metric sign Sign of metric Riemann Ricci Cost

Propulsion preferences convention: time-like determinant: convention: Lower convention: estimate
Product Units: MKS (SI) is positive negative version of Eq. (3) Rnr ¼ Rlnlr (circa 2007)

Maxima (formerly Yes (load “units” No, metric is -þþþ Yes Yes Yes Freea
Macsyma) package)
Mathematica $1800
MathTensor Yes (user set) Yes (user set) Yes (user set) Yes (user set) Yes (user set) $536
Ricci No (user set) (user set) (user set) (user set) Freeb
Maple $1995
GRtensorII No No, metric is -þþþ Yes Yes Yes Freec
C. MACCONE

(for Maple)
Cadabra Not assessed Not assessed Not assessed Not assessed Not assessed Freed
a
Available from: ,http://sourceforge net/project/showfiles php?group_id¼4933.
b
Available from: ,http://www math washington edu/ lee/Ricci/.
c
Available from: ,http://grtensor phy queensu ca/.
d
Available from: http://www aei mpg de/ peekas/cadabra/; does not run on Windows
 GENERAL RELATIVITY COMPUTATIONAL TOOLS 657

10 additional years checking them, and finally published them in 1867. His
results were taken for granted for over a century, basically because no one was
willing to devote 20 years to check the results. But, in 1970, the advent of
Macsyma allowed this daunting task to be performed: in just 20 hours of
computer time André Deprit, Jacques Henrard, and Arnold Rom, at the Boeing
Scientific Research Labs in Seattle, checked all Delaunay’s results. They found
only one analytical mistake; on page 234 of the second volume of Delaunay’s
book titled Théorie de la Lune he incorrectly wrote one fraction as 13/16,
whereas the correct fraction is 33/16. Two further errors were just a consequence
of this one. Happily, this error in Delaunay’s calculations turned out not to be
vital for the Apollo flights to the moon. This success made Macsyma the pre-
ferred worldwide symbolic computational tool from 1965 onward.
Though the Macsyma source files are written in lisp, the user does not need to
be proficient in lisp. Instead, the user must learn a special “Macsyma language” to
perform the requested calculations. This is described in both the paper manual
and online help form. This programming language may not be immediately
helpful to the user because the names for the commands are invented a bit
“randomly,” without strict logical rules that the programmer’s mind may memor-
ize immediately (see the Macsyma Mathematics Reference Manual and
Macsyma System Reference Manual [4]). This situation is somewhat similar in
Maple but not in Mathematica, where strict name rules are enforced.

B. Mathematica
When “C” became the stardard programming language, Wolfram Research,
Inc. developed and commercially sold Mathematica. This remains a widely
used computational tool for various applications. Specific notebooks and even
more specialized codes are available to use with Mathematica, some may be
downloaded without cost and others are add-on commercial products.
In Mathematica, variables are defined using the convention of long, compound
input words, where uppercase letters denote the beginning of each new word
within the compound word. This convention is not used in Maxima or Maple.
This convention has pros and cons:
Pro: The meaning of the very long words is self-explanatory.
Con: It forces the user to input very long, case-sensitive words.

1. MathTensor
Leonard Parker of the University of Wisconsin at Milwaukee and Steven
M. Christensen of the University of North Carolina at Chapel Hill created
MathTensor in 1994, the first tensor symbolic manipulator running on Mathematica.
They also published a book [5] as the user manual (http://smc.vnet.net/
mathtensor.html). Like Mathematica, MathTensor also uses the same
case-sensitive, compound word convention.
A main advantage of MathTensor is that the user can set all of the following
conventions independently, by just assigning four parameters, respectively:
MetricSign (signature of the metric) 1, DetgSign (sign of the determinant of
the metric) 1, RmSign (definition of the Riemann tensor in terms of the
658 C. MACCONE

Christoffel symbols and their derivatives) 1, and RcSign (definition of the

Ricci tensor contraction from the Riemann tensor) 1.
Another unique advantage of MathTensor is that the user can freely adopt the
preferred unit system: one just has to set the variable “units” equal to either of the
following names in order to have all equations computed in the relevant units
system, respectively: emuUnits, esuUnits, GaussianUnits, HeavisideLorentz
Units, RationalizedMKSUnits or SIUnits. This is a nontrivial feature of Math-
Tensor, because the definition itself of some tensors, like the stress-energy
tensor of the electromagnetic field, depends on conventions adopted with the
units system. The amount of time saved by the user changing results from one
unit system to another one is remarkable.
When it comes to research utility, MathTensor is able to derive the set of
tensor equations (like the Einstein equations) as the Euler Lagrange equations
from a suitable variational principle (like the Einstein Hilbert action). This is
accomplished by MathTensor first by partial integration, and then computation
of the Euler-Lagrange equations. The result is that, even for Lagrangians that
are quadratic in the Riemann tensor, obtaining the relevant differential equations
is virtually immediate. Based on experience, however, it is not yet clear if these
operations are bug-free.
Regarding disadvantages, MathTensor is not currently capable of handling
Greek letters, so symbols like the electric permittivity, 10, and magnetic
permeability, m0, have to be written “eps0” and “mu0.” Worse still, the user
interface of MathTensor is very primitive. For example, one has to edit the
“Conventions.m” file of MathTensor (i.e., not a file of Mathematica itself) in
order to set up all the above conventions.
2. Ricci
Ricci is free software created by John M. Lee of the Mathematics Department
of the University of Washington. It performs general differential geometry calcu-
lations and tensor calculus. The relevant Web site is ,http://www.math.
washington.edu/ lee/Ricci/. .
Just one example of application of Ricci: the Beltrami (or Laplace-Beltrami)
operator is the wave propagation operator in Riemannian manifolds. Ricci
computes the Beltrami operator immediately, a feature not often found in other
codes. (For more features see Ref. 1.)

3. GRtensorII (for Mathematica)

Kyall Lake of Queen’s University in Kingston, Ontario, Canada, and
coworkers Peter Musgrave and Denis Pollney, created a tensor package specifi-
cally designed to tackle research problems in GR [6]. This package is called
“GRTensor II” and was originally developed to run with Maple. A more recent
version of it also runs on Mathematica. Because it was first developed for
Maple, its features are discussed more in the Maple section. It can be
downloaded for free from ,http://grtensor.phy.queensu.ca/. .

4. Others
Other recently developed packages that have not yet been evaluated for their
propulsion utility are Tensorial (version 3 and higher) and Cartan. Tensorial is a
 GENERAL RELATIVITY COMPUTATIONAL TOOLS 659

new tensor package of Mathematica. Cartan is a package for conducting

differential geometry and Lie groups and runs under Mathematica.

C. Maple
The other product that was introduced after “C” became the standard program-
ming language is Maple, developed by a team at the University of Waterloo
(Canada). It is a main competititor to Mathematica and is also widely used as
a computational tool for a variety of scientific and engineering applications.

1. Maple’s “Standard” Tensor Package

Raymond McLenaghan of the University of Waterloo in Canada created the
“standard” tensor package of Maple. McLenaghan is known among GR experts
as co-discoverer (with the Australian, John Carminati) of the Carminati
McLenaghan invariants (i.e., all possible invariants that one can construct by
contracting the Riemann and Weyl tensors in all possible modalities; Web site
,http://en.wikipedia.org/wiki/Carminati-McLenaghan invariants. .
The tensor package of Maple is endowed with all the facilities necessary in
GR. It includes a symbolic computing code to immediately list the Carminati
McLenaghan invariants for a given four-dimensional Riemannian manifold
(i.e., when the metric tensor is assigned). A similar capability also is included
in the GRTensorII package, described next.

2. GRTensorII (For Maple)

GRTensor II was specifically designed to tackle research problems in GR by
Kyall Lake of Queen’s University in Kingston, Ontario, Canada, and coworkers
Peter Musgrave and Denis Pollney. It was originally developed to run with Maple
(Lake had originally contributed to the tensor package of Macsyma).
To operate GRTensorII one has to learn Maple’s operating language and a
dedicated “GR” language to handle tensors, both old and “new” (e.g., tensors
at research level like the Bel-Robinson tensor).
Many advanced features of GRTensorII are quite remarkable, for example:
1) Automatic Petrov classification of Einstein spaces
2) Immediate solutions of the full set of Einstein Maxwell equations
3) Numerous and detailed exact solutions (Bondi metric, Stephani solution
for perfect fluids, Tomimatsu-Sato solution, Cosmological metrics, etc.). Five-
dimensional metrics of the Kaluza-Klein type are also starting to show up
among the worked examples, and this paves the way to future generalizations
to N 11 dimensions as typical of supergravity, string theories, etc.
4) “GRJunction” is an innovative tool within GRTensorII that allows the sym-
bolic computations of exact solutions made up by “cutting, pasting, and joining
together” already known, simpler exact solutions. This is a unique feature of
GRTensorII that no other tensor package has, including MathTensor.
The disadvantages of GRTensorII are that the user cannot select:
1) Signature of the metric (assumed to be space-like)
2) Sign of the determinant of the metric (assumed to be negative)
660 C. MACCONE

3) Definition of the Riemann tensor in terms of the Christoffel symbols and

their derivatives
4) Definition of the Ricci tensor from the Riemann tensor
5) MKS (or SI Système International) system of units
The inability to adjust these settings is serious as one may get a confusing factor
of 1 when comparing the same exact solutions of the Einstein or Einstein
Maxwell equations by using different tensor packages, like GRTensorII and
Macsyma, or GRTensorII and MathTensor. Furthermore, the inability to use
MKS units is disappointing and limits its utility for space propulsion
considerations.
In conclusion, GRTensorII is another available code that has advantages and
disadvantages relative to its utility for space propulsion science research.

D. Other Codes
In addition to the codes examined, there are a number of other codes that
might be useful but whose features for space propulsion research have not yet
been assessed. A couple of examples are described next.

1. Reduce (Web site: http://en.wikipedia.org/wiki/

Reduce computer algebra system)
Reduce is an existing code commonly used in both quantum physics (e.g., for
Dirac equations) and GR. Recently a spin-off of this code was created devoted to
GR called “GRG,” Web site ,http://sal.linet.gr.jp/A/1/GRG.html..

2. Casimir-Effect Codes
Jordan Maclay developed some codes within Mathematica for studying the
Casimir force for numerous geometries, not just the traditional parallel plate
configuration. The codes address estimating the forces on the faces of three-
dimensional cavities, whose results vary depending on the aspect ratio of the
cavity and can sometime become repulsive.
Similar codes were probably also written in Europe when the space mission to
measure the Casimir effect in space was proposed to ESA back in 2000. These
have recently become available through Dr. Martin Tajmar of ARC Seibersdorf
Research, in Seibersdorf, Austria [7].
It would be interesting to “match” these codes with GR codes to find possible
breakthrough propulsion physics links between GR and the Casimir effect.

V. Conclusions
As computer codes advance and as more research is conducted in propulsion
physics, it would be advantageous for future propulsion researchers to begin to
follow common conventions in GR equations. To that end, standardization of
sign and notational conventions are recommended as follows:
1) MKS (SI) sytem of units, with all natural constants (c, G, h) explicitly
shown in the equations.
 GENERAL RELATIVITY COMPUTATIONAL TOOLS 661

2) Sign convention for the metric where causal conditions result in a positive
spacetime displacement (i.e., a signature þ - - -).
3) Sign convention where the determinant of the metric is negative.
4) Definition of Riemann tensor as Rlnrs ; Glsn;r þ Glrn;s  Gasn Glar þ Garn Glas .
5) Definition of contracting the Riemann tensor into the Ricci tensor as
Rnr Rlnlr.

Next, the ability of existing codes to support these conventions and their ability to
support propulsion-relevant calculations is assessed. Presently there is no single,
ideal product. It is the intent of this chapter that the examination of software
options and the use of standardized conventions will assist future researchers
in making their work easier to conduct and easier to share with their colleagues.

Acknowledgments
The late John Anderson of NASA was the first to recognize the key role that
these symbolic codes might play in the development of breakthrough space pro-
pulsion. Subsequently, Marc G. Millis of NASA Glenn Research Center at Lewis
Field adapted John Anderson’s work into the Breakthrough Propulsion Physics
project. The author of this paper acknowledges Marc G. Millis for his interest
in a question as vital as “how will breakthrough propulsion research benefit
from symbolic tensor manipulators?” The support and friendship of the co-editor
of this book, Eric W. Davis of the Institute for Advanced Studies at Austin, is also
gratefully acknowledged. Finally, this author is personally indebted to one of the
most farsighted researchers in visionary fields, H. E. Puthoff, Director of the
Institute for Advanced Studies at Austin, for sharing support and enthusiasm!
Relativistic interstellar flight is difficult to model and investigate mathematically,
but someone has to start the process. In the centuries to come we will show that
we have not been working in vain.

References
[1] Lee, J. M., Ricci A Mathematica Package for Doing Tensor Calculations in Differential
Geometry, User Manual. URL: http://www.math.washington.edu/ lee/Ricci/.
[2] Misner, C. W., Thorne, K. S., Wheeler, J. A., Gravitation, W. H. Freeman & Co.,
New York, 1973.
[3] Hartle, J. B., Gravity: An Introduction to Einstein’s General Relativity, Addison
Wesley, San Francisco, 2003.
[4] Macsyma Mathematics Reference Manual and Macsyma System Reference Manual,
Macsyma Inc., Arlington, MA, 1993.
[5] Parker, L., and Christensen, S. M., MathTensor A System for Doing Tensor Analysis
by Computer, Addison Wesley, Reading, MA, 1994.
[6] Lake, K., Musgrave, P., and Pollney, D., GRTensor II User Manual. URL: http://
130.15.26.62/NewDemo/frame.html.
[7] Sedmik, R., and Tajmar, M., “CasimirSim A Tool to Compute Casimir Polder
Forces for Nontrivial 3D Geometries,” Proceedings of the STAIF 2007: 4th Sym
posium on New Frontiers and Future Concepts, El Genk, M. S. (ed.), AIP Conference
Proceedings Vol. 880, AIP Press, New York, 2007, pp. 1148 1155.
 Chapter 22

Prioritizing Pioneering Research

Marc G. Millis
NASA Glenn Research Center, Cleveland, Ohio

I. Introduction
TYPICAL challenge of any research project is to decide how best to disburse
A limited resources to the variety of competing options. When it comes to
seeking propulsion-specific advances from science, the situation is even
more challenging. First, research aimed at such revolutionary advancements is
different from the more common work of improving technology. Balancing the
vision required to extend beyond established knowledge, along with the rigor
required to make genuine progress, presents additional challenges beyond just
the science itself. Next, there is the unfamiliarity of the topic. This includes intro-
ducing the aerospace community to emerging science as well as introducing
scientists to the specific queries behind revolutionary spaceflight. Additionally,
existing research proposals span multiple disciplines of science and span different
levels of progress and applicability toward the goals. To use a colloquial
expression, this presents the challenge of comparing apples to oranges. This
difficulty is compounded by the fact that not all of the important questions are
yet represented by proposed approaches. There are likely to be unaddressed
issues that are more important than the approaches proposed so far. And lastly,
on topics that appear so far from fruition, the available resources are minimal.
This compounds the challenge of partitioning resources in portions sufficient to
ensure progress.
This chapter addresses these challenges primarily by examining the methods
and lessons learned from the NASA Breakthrough Propulsion Physics (BPP)
Project. This project, founded by this author, was designed to manage
research on gravity control and faster-than-light travel. It assessed 10 different
research approaches, produced 16 peer-reviewed journal articles [1 16], an

This material is a work of the U.S. Government and is not subject to copyright protection in the
United States.

Propulsion Physicist, Propellant System Branch, Research and Technology Directorate.

663
664 M. G. MILLIS

award-winning website [17], and covered this plus discretionary efforts thereafter
for a total cost of $1.6 M spread over 1996 to 2002 [18]. The findings of the spon-
sored research are described in Chapter 1. Regarding other evidence of the
project’s effectiveness, here is a quote from an independent review panel about
the BPP Project [19]:

[The Breakthrough Propulsion Physics Project] approach was unanimously

judged to be well thought out, logically structured and carefully designed
to steer clear of the undesirable fringe claims that are widespread on the
Internet. The claim that the timing is ripe for a modest program of this sort
was agreed to be justified: Clues do appear to be emerging within mainstream
science of promising new areas of investigations. The team concurred that
the 1997 BPP kickoff workshop did identify affordable candidate research
tasks which can be respectably pursued, and that the proposed research prior
itization criteria were a valid way to select from amongst these (and future)
proposals. The program approach was deemed to be sound: emphasizing
near term progress toward long term goals; supporting a diversity of
approaches having a credible scientific basis; holding workshops to exchange
ideas; solicit constructive criticism and assess progress; aiming toward test
able concepts [19].

The project’s operating principles, research solicitation process, evaluation

criteria, and other lessons learned are detailed. Introductory comparisons to the
larger NASA Institute of Advanced Concepts (NIAC) [20,21] and the still
larger Defense Advanced Research Projects Agency (DARPA) [22] are also
provided.
The first section of this chapter examines a variety of historic lessons about
revolutionary advances and then sets them in the context of seeking revolutionary
propulsion advances from science. These details are included to convey the
rationale behind the NASA project’s methods and also pertain to both NIAC
and DARPA.
Another section of this chapter is devoted to the challenge of combining vision
with rigor. Because it is common that novel ideas are reflexively dismissed and
that noncredible concepts are abundant on provocative topics, this specific chal-
lenge warrants special attention. Methods are suggested to avoid reflexive dismis-
sals, to efficiently filter out noncredible concepts, and to respond to nonrigorous
research submissions in a constructive manner.
All of these lessons are presented largely in the context of management, but
some lessons apply to individual research as well. The intent is to provide man-
agers and researchers insights on how to conduct research that is both visionary
and productive.
For readers who want to jump ahead to the recommendations without needing
their rationale, the specific recommendations can be found in the following
sections. Most of these are in the form of lists:
II.H. Recommendations Deduced from Historical Perspectives
III.C. Summary of Recommendations on Combining Vision and Rigor
IV. Research Project Operating Principles
 PRIORITIZING PIONEERING RESEARCH 665

V. Devising Prioritization Criteria (refer to the entire discussion, not

just the bullet list)
VII.C Project Metrics of Performance
VII.D Mitigating Revolutionary Research Risks (see Table 3)

II. Historical Perspectives

To better assess the conditions necessary to elicit technological revolutions, a
variety of historical perspectives are examined and then set in the context of
seeking the scientific discoveries that could lead to revolutionary space pro-
pulsion. This includes Fosters’s “S-Curve” of technological advancements,
Henderson’s “architectural innovations,” Dyson’s “tool-driven” perspectives,
Kuhn’s “paradigm shifts,” Hamming’s lecture on great researchers and important
problems, Sir Clarke’s Three Laws, the role played by science fiction, and John
Anderson’s “Horizon Methodology.”
The literature on the history of technological advancements is abundant, and
the coverage here is only a small sample of the depth of available information that
reflects similar patterns. Interested readers are encouraged to examine the cited
references for further insights. Following the explanations of these perspectives,
a list of recommendations is offered to help researchers, managers, and advocates
of revolutionary research improve the prospects for success. These recommen-
dations are based on the recurring patterns observed in history.

A. Institutional Patterns of Technological Revolutions

History has shown that simply improving existing technology is not sufficient
to sustain competitive advantage [23]. As illustrated by the recurring S-curve
pattern of technological advancement (Fig. 1), it is time to look for revolutionary
approaches when the existing methods have reached the point of diminishing
returns. For example, jet aircraft did not result from mastering piston-propeller
aircraft. Transistors were not invented by mastering vacuum tubes. Photocopiers
did not result from mastering carbon paper. The recurring theme is that entirely
different operating principles were pursued to sustain competitive advantage
rather than refinements of known methods.
The S-curve evolution shown in Fig. 1 is typical of any successful technology.
The pattern begins with the initial efforts where little advancement results until a
breakthrough occurs. The breakthrough, at the lower knee of the curves, is where
the technology has finally demonstrated its viability. After this point significant
progress is made as several embodiments are produced and the technology
becomes widely established. Eventually, however, the physical limits of the tech-
nology are reached, and continued efforts result in little additional advancement.
This upper plateau is the “point of diminishing returns.” To go beyond these
limits, a new alternative (with its own S-curve) must be created. Shifting to the
new alternative is what is meant by pursuing “revolutionary” advancements.
It has been found that it is most difficult for incumbent organizations to con-
sider such alternatives when their familiar approaches are at the point of dimin-
ishing returns [23,24]. By then, the institutions have become too uniquely adept
666 M. G. MILLIS

New Methods

(Jet Engines)
Performance

Physics Limit of Prior Method

Prior Methods

(Propellers)

Investment
Fig. 1 S-curve pattern of technology advancement.

at their accrued technology to consider alternatives. They are also tied so closely
with their existing customers that it is difficult to explore new opportunities.
Because new approaches emerge in a still-developing state and have unfamiliar
principles, it is also difficult for the incumbent to properly assess their merit. This
difficulty is compounded because the incumbents use their prior values to judge
the new approach, values that are rooted in the evaluation criteria for the differ-
ent, prior technology. When at the point of diminishing returns, established insti-
tutions prefer to modify or add new features to their technology rather than to
search for ways to go beyond their technology or to find new applications for
their technology. Sometimes this takes the form of changing the emphasis
away from performance to improving reliability and cost effectiveness.
The term for reconfiguring existing technology to address a new opportunity is
“architectural innovation” [25]. Even here, the incumbent organizations will typi-
cally dismiss such innovations because the new opportunity is seemingly irrele-
vant when viewed per their prior values. Additionally for architectural
innovations, their value is even harder to appreciate because the technical
aspects of the innovation do not appear like noteworthy advancements.
In the case of spaceflight, the space tourism entrepreneurs are examples of
such architectural innovations. They are taking existing technology and applying
it in new configurations and to reach new markets [26,27]. Although this example
is not directly relevant to frontiers of propulsion science, it is relevant in the
context of identifying obsolescing values. As evidenced by the emergence of
such firms outside the incumbent aerospace organizations, it is clear that the
original values that drove the emergence of spaceflight are no longer complete.
In other words, the criteria against which early spaceflight was judged are no
longer the only drivers of future progress. Significant changes have occurred
 PRIORITIZING PIONEERING RESEARCH 667

in societal values, technological options, and emerging science. Following

historical patterns, it is likely that the incumbent spaceflight organizations will
have difficulty recognizing and adapting to the contemporary opportunities and
constraints [23 25].
Considering these patterns, it is not surprising that the emergence of revolu-
tionary advances often come from outside the established organizations [28,
p. 30]. A classic aerospace example is how the Wright brothers (bicycle mech-
anics) succeeded in heavier-than-air manned flight well in advance of the govern-
ment funded (Smithsonian Institution) aerospace research of Samuel P. Langley.

B. Tool-Driven Revolutions
In Freeman Dyson’s Imagined Worlds (1997) scientific revolutions are cast as
the byproducts of new tools [29]. As examples, Dyson cites how the telescope led
to Galileo’s insights and how X-ray diffraction led Crick and Watson to under-
stand DNA structure. The recurring theme is that when new observational
tools become available, new phenomena are observed, which then leads to revo-
lutionary science.
Considering our contemporary situation, this list shows tools and observations
that are relevant to seeking revolutionary spaceflight along with citations for
where these are discussed in this book:
1) Hubble space telescope and other advanced ground telescopes are reveal-
ing further details about the constituents of our universe, such as the anomalous
redshifts from the most distant supernovae, and the apparent existence of “dark
matter” (discussed in Chapters 3, 4, and 15).
2) COBE and WMAP satellites are revealing new details of the cosmic micro-
wave background radiation [30] (discussed in Chapter 3).
3) Improved fidelity of laboratory instrumentation makes it possible to
conduct ever-more accurate experiments and correspondingly requiring extra
vigilance in their proper use (Chapters 5 and 20).
4) Symbolic mathematical computational tools make it easier to apply
complex mathematical techniques (Chapter 21).
5) Ultrahigh-intensity tabletop lasers can momentarily produce extreme elec-
tric and magnetic fields suitable for testing space-warping theory (Chapter 15).
6) Superconductivity provides the means to explore the intriguing phenom-
enon of absolute-zero electrical resistance and its secondary phenomena (see
Chapter 5).
7) Micro- and nano-structure engineering make it easier to tangibly explore the
physics of the very small, such as Casimir forces (discussed in Chapters 12 and 18).
8) Internet communication makes it easier to access a wider network of data
and collaborators. This includes those that were previously separated by geo-
graphy or disciplines.

C. Paradigm Shifts
The next perspective, articulated in 1962 by Thomas Kuhn in The Structure of
Scientific Revolutions [31], posits that scientific revolutions are sudden paradigm
668 M. G. MILLIS

shifts that occur after the accumulation of physical data that do not coherently
blend with the prior paradigms. Examples include the Copernicus model of pla-
netary orbits (heliocentric model), Newton’s laws of motion, Darwin’s evolution,
Einstein’s special relativity, and quantum mechanics.
In a 1970 edition to his book [31, see Postscript], Kuhn distinguishes between
two meanings of paradigm that were too closely woven together in earlier edi-
tions, leading to misinterpretations. One meaning is related to the scientific com-
munity as a whole, whereas the other is specific to sets of interpretations held for
natural phenomena. In the context of this discussion, the latter meaning is used,
specifically paradigms as sets of interpretations held for natural phenomena. In
this sense, it is time to more critically challenge old paradigms and suggest
new ones when accumulating evidence does not clearly fit the established
paradigms.
The organizational challenge when dealing with paradigms is the implicit
value system used to judge emerging possibilities. This is similar to the ten-
dencies discussed earlier regarding S-curves, where old values are used to
judge new situations. With paradigms there are implicit commitments within
incumbent organizations for setting work priorities. This results in a tendency
to reflexively dismiss novel approaches that are inconsistent with the established
paradigm. This issue is discussed in more detail in Sec. III: Combining Vision
and Rigor.
For now, to offer starting points for considering how the perspective of para-
digm shifts pertain to space propulsion, a brief list of physical observations or
lingering unknowns that have not yet been resolved follows. Many of these are
revealed by the new tools discussed previously and their relevance to space
propulsion is discussed in the cited chapters.
1) “Dark matter” as the contemporary paradigm to explain gravitational
lensing and anomalous binding of rotating galaxies (Chapters 3, 4, and 15).
2) “Dark energy” as the contemporary paradigm to explain the anomalous
redshifts of distant supernovae (Chapters 3, 4, and 15).
3) Incompatibility of General Relativity (large scale) and Quantum Mech-
anics (small scale). As one example, there is discrepancy of about 120 orders
of magnitude between the General Relativistic energy estimate for the
quantum vacuum and that from astronomical observations with Quantum
interpretations (Chapters 3, 12, 13, and 18).
4) Causality and the nature of time (Chapters 14 16).
It is curious to contrast these lingering unknowns to the notion presented in the
1996 book, The End of Science [32], where its premise is that all the important
questions have already been raised and are well on their way to being answered.

D. Great Researchers and Important Problems

In a 1986 lecture, Richard Hamming (a pioneer in error-correcting codes)
reflected on the distinctions between good and great researchers [33]. According
to Hamming, good researchers are those who make competent, incremental
advancements, while great researchers are those who achieve pioneering
advancements beyond their peers. The two major distinguishing characteristics
 PRIORITIZING PIONEERING RESEARCH 669

are the problems those researchers choose to tackle, and the self-critical drive
with which those problems are pursued.
Regarding the choice of problem, it was noted that the great researchers have
the courage to tackle the “important problems.” Important problems are defined
as the grand challenges that will make a significant difference if solved, and
where enough progress has been made to enable these problems to finally be
pursued. These are the problems that their peers will not attempt. Instead, the
good researchers opt to pursue objectives that are already well established in
their field and where there is little chance of failure. In terms of the S-curve
analogy discussed previously, the good researchers work to improve existing
methods, while the great researchers seek the new S-curves, the alternatives to
surpass the existing methods.
The choice of problem is not the only distinction. Hamming discussed numer-
ous characteristics that great researchers tend to possess. The recurring theme
is that the great researchers have both the confidence to approach the problem,
with enough self-doubt and awareness of their shortcomings to sustain their
objectivity. Here is an abbreviated list of the major characteristics identified by
Hamming:
1) Have the courage to tackle the important problems:
a) These are grand challenges that will make a significant difference, not
just the “safe” research.
b) Attackable, meaning that there is a way to begin solving the problems.
2) Start with independent thoughts and then collaborate with others.
3) Make steady progress; be driven and focused.
4) Redirect what is difficult into something easier.
5) Are open to learn things beyond their own field “Knowledge is like
compound interest.”
6) Tolerate ambiguity:
a) Believe enough in self to proceed.
b) Doubt self enough to honestly see flaws.
7) Sell themselves well.
a) Write well.
b) Present well.
c) Able to translate work into executive-level communications.
8) Honest with personal flaws and work toward overcoming them
(converting liabilities into assets).

E. Sir Arthur C. Clarke’s Three Laws

Evolving over the course of revisions to the book, Profiles of the Future,
1962 1973, Arthur C. Clarke posited the following three “laws” regarding the
reaction to revolutionary advancements [34]:
1) When a distinguished but elderly scientist states that something is possi-
ble, he is almost certainly right. When he states that something is impossible,
he is very probably wrong.
2) The only way of discovering the limits of the possible is to venture a little
way past them into the impossible.
670 M. G. MILLIS

3) Any sufficiently advanced technology is indistinguishable from magic.

While these oft-cited laws are not laws in the strictest sense, they do serve to
summarize key recurring themes and are consistent with many of the perspectives
already introduced.
Regarding the first law, which is in reference to such infamous quotes as:
“Space travel is utter bilge” uttered by Dr. Richard van der Riet Wooley, one
year before 1957’s Sputnik; and “The secrets of flight will not be mastered
within our lifetime, not within a thousand years” credited to Wilbur Wright,
two years before their 1903 Kitty Hawk flight, it is equally important to remem-
ber that history includes many errant ideas that were indeed critically flawed. Dis-
tinguishing these is easy in retrospect. The viable ideas survive (along with the
infamous dismissive quotes), while the errant ideas tend to be forgotten.
To assess revolutionary ideas as they emerge is difficult. Revolutionary ideas,
by their very nature, break from the familiar and can look as nonsensical at first as
the more common errant ideas. Here again the notions of paradigm shifts and
incumbent values pertain. Because this situation is complex and is a major chal-
lenge of revolutionary research, Sec. III: Combining Vision and Rigor is devoted
to addressing this issue.
When it comes to deliberately seeking revolutionary advances in spaceflight,
Clarke’s second law is more opportune because it suggests where to look for the
next revolutions. An oft-cited source of such inspirational impossibilities is
science fiction, which is examined in the next section.
Regarding Clarke’s Third Law, Robert Forward applied this perspective in his
book Indistinguishable from Magic, which included projections of interstellar
flight technology [35]. While many of the specific topics in Forward’s book
are addressed in more up-to-date terms here in this book, Forward’s book still
serves as an illustration of Clarke’s third law.

F. Science Fiction
It is generally recognized that science fiction is inspirational. Just to cite one
study, the American Astronautical Society (AAS) found that the rocketry pio-
neers, Tsiolkovsky, Goddard, von Braun, etc., were all inspired by the science
fiction of their day [36]. The captivating idea of rocket travel and its implications
for a better world spurred these pioneers to investigate how to make rocketry real.
There is also a common misperception that science fiction reliably predicts
future technologies. Although some fiction became real (nuclear submarines,
men on the moon, portable palm-sized phones, etc.) there is much fiction that
has not happen or did not happen as envisioned. An excellent example is when
men first walked on the moon [37, p. 142]. Although the Apollo crew size and
the launch location matched Jules Verne’s fiction, rockets were used instead of
a giant cannon and, most importantly, when it happened for real, people
around the whole world were watching on television. This unforeseen event of
the entire world being able to witness this historic event is profoundly significant
unto itself.
Regarding the scientific discoveries that could revolutionize spaceflight, many
science fiction stories have already broadcast what interstellar flight might look
 PRIORITIZING PIONEERING RESEARCH 671

like. Famous examples include Star Trek [38] and Star Wars [39,40]. This fiction,
however, is not sufficient for conducting research. Although entertaining, inspir-
ing, and thought provoking, the science behind such fiction was cast to provide a
setting for drama. Accordingly, it is often incomplete or compromised to create
dramatic tensions and expedient plots. As much as it might inspire, to follow its
fictional images too closely would limit possibilities, as it would constrain one to
only consider the approaches suggested in the fiction.
When examined in the context of its thought-provoking nature and without
taking its visions too literally, science fiction is akin to “brainstorming.” Brain-
storming is a step in explicit processes for enhancing the problem-solving
ability of teams [41]. In such processes, the focal problem is defined in the
first step. Next, through brainstorming, numerous ideas are collected to address
the problem. At this stage the ideas do not need to be correct, but merely provo-
cative. This is completely consistent with science fiction visions. Once a suite of
provocative ideas is collected, analytical rigor is applied to filter through the ideas
and refine a workable set of approaches. For the details of such techniques, the
reader is referred to Miller’s The Creative Edge [41] or any of the other numerous
books that document the human creative process and how to apply that process
explicitly in organizational settings.
Following along these lines, the next section summarizes a technique devel-
oped and applied to deliberately seek revolutionary research within NASA.

G. Horizon Mission Methodology

In 1996, John L. Anderson published his ‘Horizon Mission Methodology’ in
the Journal of the British Interplanetary Society and set it in the context of inter-
stellar flight [42]. This method is a systematic approach for provoking revolution-
ary research within organizations the kind of approaches indicative of the
alternative S-curves described earlier. The method employs lessons from prior
revolutions and deliberately uses science fiction for its inspirational value.
The process can be used by individuals or teams. Its first step is to impose a
general goal that is impossible to achieve with projected technology. The goal
of timely interstellar flight is an example. The use of impossible goals is deliber-
ately intended to counteract the habit of researchers to extrapolate their familiar
technologies to address problems. To use a colloquial expression, it forces the
researchers to think “out of the box.” Along with the seemingly impossible
goals is the requirement that the team considers that the goals are achievable
by some undefined, far-future technology. This is an application of Clarke’s
Second Law. Next, through brainstorming, the notions of science fiction are
used as placeholders for the embodiments of the ultimate successes. From
those provisional “solutions” the team is then asked to “look back from this
future” to identify the limiting assumptions. In other words, the team is asked
to determine the specific make-or-break issues that would have to be solved to
make such a future plausible. In short, this means defining the “grand challenges”
around which to aim research objectives. With those goals set, the next step is to
identify the knowledge gaps. By comparing the grand challenges in detail with
the accumulated knowledge to date, the key unknowns and critical issues are
identified. From there, the researchers are asked what steps could be taken to
672 M. G. MILLIS

begin addressing those unknowns and issues. In terms of Hamming’s lecture, this
means articulating the “important problems.”
The NASA BPP Project employed this Horizon method. It was used to devise
the Project’s grand challenges, specifically its three goals: 1) nonpropellant pro-
pulsion, 2) hyper-fast travel, and 3) the energy breakthroughs related to those two
goals. In a subsequent exercise with geographically dispersed participants, the
foundational knowledge related to these challenges was collected [43]. And
finally, still following the spirit of this Horizon method, a workshop was con-
vened to deliberately identify relevant research tasks [2,44].
The next section lists recommendations based on the lessons just described.
These recommendations are intended to help researchers, managers, and advo-
cates of revolutionary research improve their prospects for success.

H. Recommendations Deduced from Historic Lessons

From historically based perspectives, it is clear that pioneers who seek support
from established organizations face both technical and organizational challenges.
The Horizon Mission methodology [42] is one approach to begin soliciting revo-
lutionary approaches within established organizations, but this by itself appears
insufficient when considering organizational impediments. To help overcome
organizational impediments, it is recommended that advocates of pioneering
research also perform the following programmatic tasks:
1) Compare the goals of the organization to the ultimate performance limits
of the organization’s technology. Make it clear which revolutionary advance-
ments are required to fully satisfy those goals. In the case of viable interstellar
flight, for reaching other habitable worlds, this goal cannot be achieved by
merely improving technology. Alternatives from undiscovered science must be
sought (see Chapters 2, 3, 14, and 17).
2) Familiarize the organization about the emerging possibilities, building on
scholarly publications and impartially identifying both their strengths and weak-
nesses. Offer foundational information that will help the organization compre-
hend the opportunities and challenges. In the case of the frontiers of
propulsion science, this book is a step to address this need.
3) To enable the organization to properly assess the emerging alternatives,
develop the criteria against which the emerging alternatives can be compared
to one another and to the organization’s broader goals. This avoids the
pitfall of assessing fledgling alternatives in terms of criteria that are specific
to prior, more-developed methods. In the case of propulsion science, this
chapter offers examples of criteria and the selection processes used by the
NASA BPP project.
4) Using the foundational information in comparison to the revolutionary
goals, identify the “important problems”; those problems that are both highly rel-
evant and approachable. John Anderson’s Horizon Mission methodology is one
approach to identify such problems within organizations [42].
5) Closely related to identifying important problems is to place emphasis
on physical observations rather than on contemporary paradigms of those
observations. Paradigms are interpretations to explain observations of nature.
 PRIORITIZING PIONEERING RESEARCH 673

Empirical data are nature (partially revealed within the constraints of the given
observation or experiment). For example, “dark energy” is a paradigm a
working hypothesis to explain the anomalously large redshifts of light from
distant supernovae. The redshifts and luminosity data are the empirical obser-
vations. Conversely, the liability of emphasizing empiricism is that data are
more limited than a reliable theory. A reliable theory (which is more substantial
than a paradigm) can predict observations that have not yet been made. At the
current stage of seeking revolutionary spaceflight, where the theories still have
not been fully formulated, it is advantageous to place more emphasis first on
empirical observations, and then derive theories to fit those data, independent
of such paradigms as dark matter and dark energy.
6) Tailor the earliest research proposals to fit within the relatively minor
resources available to the far-future options, and then build on the progress
from those steps to demonstrate the value of continuing research toward revolu-
tionary advancements.
7) And lastly, realizing that many revolutions come from outside the
expected organizations, look to other organizations and disciplines for opportu-
nities. In terms of revolutionary propulsion science, it is prudent to approach
the undiscovered physics from the aerospace perspective and seek the visions
for the next aerospace advances in terms of emerging physics.
In subsequent sections of this chapter, the specific programmatic methods of
the NASA BPP Project are described. Before proceeding, it is necessary to
examine yet another set of lessons from dealing with revolutionary ideas the
challenge of balancing vision and rigor.

III. Combining Vision and Rigor

Exploring the edge of knowledge for profound discoveries evokes special
challenges. In addition to the normal challenges of scientific research discover-
ing how nature truly works the provocative nature of the edge of knowledge can
encumber research. First, by pursuing truly profound improvements in the human
condition, the stakes are higher and accordingly emotions run higher. Second, by
operating on the genuine edge of knowledge instead of exploring refinements of
established knowledge, controversial ideas are encountered. This combination of
heightened emotions and controversy can encumber the normal, productive dis-
course of scientific study. Both skeptics and optimists can prematurely reach con-
flicting conclusions and, in their zeal, fail to communicate with the impartiality
needed to rigorously identify, test, and resolve the real issues.
Considering that most historic breakthroughs originally sounded like fringe
ideas, it is not surprising that many of the proposals for revolutionary spaceflight
might sound too visionary at first, or at least unfamiliar. It is therefore difficult to
sort out the fringe ideas that may one day evolve into tomorrow’s breakthroughs
from the more numerous, erroneous ideas.
This section examines this challenge with the in-going premise that a combi-
nation of vision and rigor is needed to make genuine progress on such topics.
Vision, where one extends beyond the known using imagination and opportunis-
tic optimism, is necessary to break from mere extrapolations of existing
674 M. G. MILLIS

solutions. It is a necessary characteristic of venturing in search of new S-curves

and to break free from legacy paradigms, and to seek out the important problems.
It is a necessary characteristic to apply Clarke’s Second Law. Perhaps this is what
Einstein meant when he said “Imagination is more important than knowledge.”
Conversely, vision by itself is insufficient, because it lacks the critical analysis
to move forward. By itself, vision is just akin to science fiction. To make
genuine progress, analytical rigor is also needed. Just as vision by itself is
inadequate, rigor alone is insufficient to make revolutionary progress. Rigor
by itself does not take one beyond the known. By combining rigor with
vision, reliable contrasts between the known and unknown become possible.
Make-or-break issues can be identified. Progress is made with the resolution of
these issues.
This section examines several facets of how to enable this combination of
vision and rigor. This includes avoiding the tendency to reflexively dismiss unfa-
miliar ideas and the challenge of dealing with nonrigorous concepts that are
endemic to this topic and that taint serious research. Lessons from psychological
studies and from the NASA BPP project are offered.

A. Absence of Vision: Reflexive Dismissals

As reflected by the investigations of Foster [23], Utterback [28], Kuhn [31],
and Clarke [34], it is common to have established experts summarily dismiss
emerging possibilities. Statistically speaking, one is likely to be correct by dis-
missing all unfamiliar assertions. Viable revolutionary ideas are rare while
errant ideas are easily generated. Such reflective dismissals are not the same as
rigorously applying skepticism to identify critical issues.
To reliably determine feasibility of new ideas requires openness to consider
the possibility and to rigorously apply skepticism to check for weak points and
to judge how well these weaknesses are addressed. To apply skepticism in the
assessment of a new idea is time consuming, because the unfamiliarity inherent
in revolutionary ideas requires more time to assess. With familiar approaches
there are already precedents with which to compare. With unfamiliar approaches,
it takes time to check the citations and ensure that the assertions are logically con-
structed. It is analogous to assessing a topic outside of one’s normal discipline.
This is compounded because emerging ideas are inherently less refined, which
can make them appear nonrigorous when compared to well-established knowl-
edge. Conducting a full assessment of an unfamiliar approach is comparable to
a full research task unto itself.
In the context of managing research solicitations for pioneering ideas, this pre-
sents a challenge; if a rigorous proposal review is potentially as intensive as the
proposed research tasks themselves, how does one efficiently filter out errant
approaches while still allowing the potentially viable ideas that might otherwise
be hastily dismissed? In short, one technique is to scrutinize the rigor of the
approach rather than trying to judge its feasibility. This technique is described
in more detail in the section “NASA Breakthrough Propulsion Physics Project
Operating Principles.” Before detailing that technique, it is necessary to
examine more closely the characteristics of nonrigorous work.
 PRIORITIZING PIONEERING RESEARCH 675

B. Vision Without Rigor

Individuals who dabble in vision without rigor are common. Regrettably, such
nonrigorous enthusiasts can taint the serious pursuit of revolutionary research;
hence it needs to be addressed here. The grand claims typical of nonrigorous
work can attract undue media attention [45]. When nonrigorous approaches are
reported in the press they exacerbate the difficulty of focusing attention on more
promising approaches. Some works specifically related to the pursuit of spaceflight
breakthroughs have appeared in such books as Voodoo Science [46]. One of the first
lectures on this situation is the 1953 colloquium by Dr. Irving Langmuir who coined
the term “pathological science” [47]. Other examples appear in a short paper “Some
Observations on Avoiding Pitfalls in Developing Future Flight Systems” [48]. To
address this issue, this section examines the extent of this situation, corresponding
psychological considerations, techniques to identify noncredible approaches, and
then techniques for responding to noncredible submissions.

1. Statistics on the Extent of Nonrigorous Submissions

In the course of the NASA BPP Project, two years of unsolicited correspon-
dence (2000 and 2001) were analyzed to determine the relative proportion of non-
rigorous enthusiasts [49]. The intent at the time was to provide management an
estimate of the labor required to reply to these submissions. In the context of
this chapter, these statistics convey the magnitude of nonrigorous activities. No
statistics are available to reflect the proportion of researchers who possess
rigor without vision.
The findings are presented in Table 1. These statistics counted the number of
messages rather than the number of submitters. Multiple messages from a single
sender, for example, were all counted. Roughly one-third of all the submissions
to the NASA project were from nonrigorous individuals requesting reviews or
other support for their ideas. Given that an objective review and response can
take roughly three days to prepare, and considering the rate of incoming sub-
missions, it is estimated that a fulltime staff of three or four researchers would
be needed just to respond to the nonrigorous individual requests.
Within that group of nonrigorous individuals, there is another subset that war-
rants mention, the “lunatic fringe.” The term lunatic fringe is a colloquial
expression for those people whose ideas are accompanied by delusions of gran-
deur and/or paranoia. Please be advised that this discussion is not meant to be
glib, trivial, or taken lightly. Roughly one-third of the nonrigorous correspon-
dences (specifically 9% of total unsolicited correspondence) displayed such
characteristics. Delusions of grandeur are evidenced when the submitter states
that their device or theory is the answer to solve the world’s problems, or
other statements to that effect. Paranoia is easier to recognize, where the submit-
ter expresses fear about their safety or about the theft of their ideas. Often these
characteristics coexist, such as when the submitters express concern that their
submission is so valuable that it will evoke suppression by evil conspirators.

2. Techniques to Identify Nonrigorous Submissions

Research managers who are scouting for potentially revolutionary ideas need
to find an efficient way to screen submissions. As already shown, the proportions
676 M. G. MILLIS

Table 1 Unsolicited correspondence statistics from the NASA Breakthrough

Propulsion Physics Project

1000 unsolicited submissions per year (2000 2001) 100%

Professional researchers: 32%

. News of recent peer reviewed publications
. Inquiries about future research solicitations
. Employment inquires
Amateur researchers requesting feedback 31%
. “Here is my breakthrough . . .”
. “Please evaluate my idea . . .”
. “Please help me advance my idea . . .”
(About one third of amateur requests, or 9% of all
submissions, display paranoia or delusions of grandeur)
Public inquiries: 30%
. “Please tell me more about . . .”
. “What is your assessment of . . .?”
. “Please help me with my homework.”
Invitations to conferences/workshops 4%
Press interview requests 2%
Public speaking requests 1%

of nonrigorous enthusiasts are high and there is a tendency in reviews to prema-

turely dismiss unfamiliar concepts, concepts that might actually be viable. To
conduct thorough, rigorous reviews has already been shown to be as intensive
as conducting the research task itself.
The recommended tactic is to focus on the rigor with which the concepts are
articulated rather than on the nature of the proposed concept. A lack of rigor
is easier to detect. Classic symptoms of nonrigorous work are reflected in
Langmuir’s lecture on “pathological science” [47], Carl Sagan’s “baloney detec-
tor” [50], John Baez’s “Crackpot Index” [51], and lessons from the NASA BPP
Project.
One of the earliest examinations on the phenomenon of noncredible science
occurred in a 1953 colloquium by Dr. Irving Langmuir who coined the term
“pathological science” [47]. In this lecture, Langmuir identified the following
six symptoms of unproductive and misleading work (quoted directly from
R. N. Hall’s transcription of the lecture [47]):

1) The maximum effect that is observed is produced by a causative agent

of barely detectable intensity, and the magnitude of the effect is substantially
independent of the intensity of the cause.
2) The effect is of a magnitude that remains close to the limit of detectability;
or, many measurements are necessary because of the very low statistical sig
nificance of the results.
 PRIORITIZING PIONEERING RESEARCH 677

3) Claims of great accuracy.

4) Fantastic theories contrary to experience.
5) Criticisms are met by ad hoc excuses thought up on the spur of the
moment.
6) Ratio of supporters to critics rises up to somewhere near 50% and then
falls gradually to oblivion.

Of these largely self-explanatory symptoms, the first bears further expla-

nation. In the cases that Langmuir studied (N-rays, Davis Barnes effect, mito-
genic rays, etc.), the proponents avoided tests to increase the magnitude of
their effect by increasing the magnitude of its hypothesized cause. Instead, the
phenomena were treated as if they were independent of controllable causes.
Another list comes from Carl Sagan’s The Demon-Haunted World [50]. Here
Sagan offers a list of roughly 30 characteristics for identifying fallacious or frau-
dulent arguments, sometimes referred to as “baloney detectors.” Although the list
is too long to quote here in its entirety, the following seven items reflect many
of its key points. (The full list has been posted on the Internet at ,http://
www.xenu.net/archive/baloney detection.html . .)

1) Wherever possible there must be independent confirmation of the facts.

2) If there is a chain of argument every link in the chain must work.
3) Is the hypothesis testable? Can it, at least in principle, be falsified (shown
to be false by some unambiguous test)? Can others duplicate the experiment
and get the same result?
4) Was there observational selection (counting the hits and forgetting the
misses)?
5) Are the statistics of small numbers (such as drawing conclusions from
inadequate sample sizes)?
6) Is there a misunderstanding of the nature of statistics (e.g., President
Eisenhower expressing astonishment and alarm on discovering that fully
half of all Americans have below average intelligence!)?
7) Is there confusion of correlation and causation?

A similar list is the the Crackpot Index from John Baez. It provides a scoring
system for rating potentially revolutionary contributions to physics, but where the
higher scores refer to less-rigorous assertions. The list is based on common
symptoms of nonrigorous submission, such as offers for prize money to
anyone who proves their theory wrong (item 13) or that the “scientific establish-
ment” is engaged in a “conspiracy” to prevent their work from gaining its well-
deserved fame (item 34). The full list of 37 items can be found at ,http://
math.ucr.edu/home/baez/crackpot.html. [51].
From the experience of the NASA BPP Project, these behaviors were similarly
observed. It is common for the nonrigorous proponents to blame their lack of
success on the phenomenon of reflexive dismissals, without considering the possi-
bility that their work has flaws. This lack of self-criticality was observed as a
recurring theme among unsuccessful researchers. Another consistent behavior
of nonrigorous submissions was the lack of relevant reference citations. This is
a particularly easy characteristic to use as a checklist item in research solicitations.
678 M. G. MILLIS

3. Additional Psychological Considerations

It has been found in psychological studies that those who are most incompe-
tent also lack the competence to realize their incompetence [52]. This is not
meant to be a glib comment; this is a real psychological characteristic that com-
pounds the difficulty of responding to nonrigorous research.
The cited study covered many different tests, including humor, grammar, and
logic; the trends were similar throughout. The study divided the participants into
quartiles based on their test scores and then compared how the participants per-
ceived their abilities without knowing the scores. The trend is that the poorest
performers are the least aware of their limitations. There was, however, a cross-
over point typically with the third quartile who tended to accurately judge their
ranking. The most competent quartile had the opposite perspective. They
tended to underestimate how well they fared compared to their peers. The self-
assessments of their test scores tended to be accurate, but the top quartile
tended to overestimate the abilities of others. In other words, they tend to think
others were similarly competent to themselves.
The study also found that raising the skill level of the less competent helped
them better realize their limits. By teaching the less competent how to improve
their skills, they become more aware of their deficiencies. In the context of
dealing with nonrigorous submissions, this means that there is a benefit from
teaching the submitters how to take the next step with their work rather than to
just dismiss their work.

4. Responding to Nonrigorous Submissions

Because of the possibility that researchers and managers of provocative
research will occasionally be asked to respond to nonrigorous submissions, the
following recommendations are offered. These recommendations are offered
with the observation (from the NASA BPP project [49]) that poorly crafted feed-
back can result in additional time-consuming and unproductive consequences.
Recall that the poorest performers are the least aware of their limitations. If
not dealt with constructively, they can counter with emotionally charged corre-
spondence that also brings in others into their dispute. This short list of rec-
ommendations is followed with explanations and examples.
1) Do not reply to submissions that display delusions of grandeur or paranoia.
2) Give the submitters a task to perform that is aimed at improving the rigor
of their work and at a level commensurate with their abilities.
3) Make the successful completion of that task a condition of continued
correspondence.
4) When stating that the submitter’s claims run contrary to known physics,
use terms such as “appears to violate a well-established law of nature.”
Based on the advice of a psychologist, it is recommended to not send any
response at all to the lunatic fringe, those whose submissions display delusions
of grandeur and/or paranoia (author’s discussion with Dr. Joseph R. Wasdovich,
Employee Assistance Program Manger and psychologist, NASA Glenn Research
Center, 11 December 2002). The reason is that a technical response will not
provide the submitter with the kind of help they need and will only encourage
 PRIORITIZING PIONEERING RESEARCH 679

more unproductive correspondence. Delusions of grandeur are evidenced when

the submitter states that their device or theory is the answer to solve the
world’s problems, or other statements to that effect. Paranoia is easier to recog-
nize, where the submitter expresses fear about their safety or about the theft of
their ideas. Often these characteristics coexist, such as when the submitters
express concern that their submission is so valuable that it will evoke suppression
by competing conspirators.
The option to not reply to any submissions is a viable option. Even though
reviews of unsolicited submissions were occasionally conducted early in the
NASA BPP Project, eventually the project adopted the policy of not reviewing
any unsolicited submissions due to the volume of correspondence. For formal
solicitations, however, where the submitter has followed the rules of submission,
debriefings are often required.
Most of the nonrigorous submissions are curiosity-driven and the individuals
simply do not know where to turn for guidance. Normal venues for assessments
(journal peer-reviews or analytical assessments) are excessively difficult for non-
rigorous individuals; it has been found to be more effective to give the submitters
a task that is within their ability to perform and that will educate them to the criti-
cal details. Sometimes this can be as simple as showing them an example of a
similar, previously submitted idea that was proven not to work as claimed.
This is also consistent with the psychological study from Kruger and Dunning
[52] where it was found that poor performers are more apt to realize their
actual level of capabilities if taught to improve. A quintessential example is
where mechanical oscillators are claimed to produce net thrust (see Chapter 6).
In this case, the experimenter can be asked to perform further experiments
using a level pendulum and instructed on how to distinguish vibrations from
net thrust. This gives the submitters a task that is both within their ability and
that will illustrate that their device does not operate as they originally interpreted.
It is an exercise they can learn from.
When drafting the response letter, it is not effective to simply dismiss the ideas
as violating known physics. This does not give the submitters a path to improving
themselves or for learning the real operation of their device, plus it tends to evoke
inflammatory, emotionally charged reactions. Remember, according to the study
by Kruger and Dunning [52], they are likely to be unaware of the extent of their
limitations. It is necessary in the response to raise the issue of a violation of
physics, but in a way that leaves them an option to prove otherwise. Also,
rather than using the phrase “laws of physics,” it is more constructive to use
phrases such as “laws of nature.” The distinction is that nature is inarguably
impartial and the basis of the physics.
The following is an example of a more effective response letter:

Your device appears to violate conservation of momentum (or insert which

ever natural law applies to their case), a well evidenced law of nature. To
convince us that something other than this is occurring with your device,
we require that you perform additional tests to a higher standard of proof.
More detailed suggestions are included for such tests as well as explanations
for why other devices, that appear similar to yours, were found not to be
viable. If, after following our suggestions and ensuring that all false positive
680 M. G. MILLIS

effects have been dismissed, you will then have more convincing evidence
that your device is operating in a novel manner. At that point we will
gladly reconsider your submission. Until such conditions are met, we
regret that we do not have the resources to maintain correspondence or
provide further assistance in your investigations. We wish you the best in
your endeavors.

This response puts the burden of proof back onto the submitter, setting clear
and reasonable steps to move ahead to the next level of legitimacy. In cases
where the submitters have built devices, it is recommended that the response
dictate further experimental tests as a condition of deeper inquiry. By giving
the submitter a reasonable next step and advice on what pitfalls to avoid, it
gives them the means to learn from their own experiences. By making successful
tests a condition for engaging in future correspondence, it relieves reviewers from
further time-consuming correspondence, yet leaves open the option of learning
about possible positive results.

C. Summary of Lessons on Combining Vision and Rigor

Extracting the key points from the previous discussions, below is a short list of
recommendations for successfully combining vision and rigor. These address
both the tendency to reflexively dismiss novel ideas (an absence of vision), as
well as to efficiently filter out nonrigorous ideas by taking advantage of their
recurring critical flaws.
1) In evaluations, focus on the rigor with which the work is constructed
rather than on trying to assess the viability of its approach.
2) Checklist of characteristics that reflect required rigor:
a) The submitter is aware of the focal make break issues related to their
approach.
b) The submitter is cognizant of the reliable relevant literature. Note,
however, that it is common that credible researchers are not aware
of all of the relevant literature. Some omissions are reasonable
to expect.
c) Any alternative and unconventional interpretations of known phenom-
ena are accompanied by correct citations of those phenomena.
d) The submitter cites examples of their prior publications or work to
reflect their competence to conduct the proposed work.
3) Checklist of the characteristics that reflect noncredible work. The appear-
ance of any of these is sufficient to reject the submission:
a) The submitter does not provide any reference citations.
b) The submitter is unaware or does not mention the critical issues
underlying their concept.
c) The submitter displays delusions of grandeur or paranoia.
d) The submitter blames others for their lack of prior success.
4) The bottom line evaluation question is: “Has the researcher demonstrated
that the results will be rigorous enough to be a reliable conclusion upon which to
base future decisions, regardless if their idea works or not?”
 PRIORITIZING PIONEERING RESEARCH 681

IV. Research Project Operating Principles

Considering the historical perspectives and the need to combine vision and
rigor, managing a research project devoted to revolutionary advances presents
challenges beyond managing nearer-term research. As shown in Fig. 2, the
manager faces the challenge of balancing a number of opposites. In addition to
balancing vision and rigor, the manager must demonstrate progress in the near-
term on goals whose solutions are not immediately conceivable. Further, the
research approaches are likely to be numerous and diverse, while the manager
can only focus on a few. And lastly, the supported research has to be within
budgets, yet in portions sufficient to make genuine progress.
The NASA BPP Project devised the operating strategies described below to
deal with these challenges [18]. These strategies are offered as an example for
others who are responsible for managing revolutionary research. Later in this
chapter, the specific research selection criteria are also detailed. The items in
this introductory list are subsequently discussed in detail.
1) Reliability: Define success as acquiring reliable knowledge, rather than as
achieving a breakthrough.
2) Immediacy: Focus research on the immediate unknowns, make-or-break
issues, or curious effects, rather than extrapolating beyond reliable foundations.
3) Measured: Measure the relative maturity of the research using recognized
steps of scientific and technological progress.
4) Iteration: Achieve overall progress by repeating cycles of solicitation,
research, and reassessment, where the cycle duration is roughly two to four years.
5) Diversified: Explore multiple, divergent research approaches simul-
taneously.
6) Impartial: Judge selections based on credibility and relevance, rather than
trying to predict feasibility.
7) Empirical: Preference is given to experiments and empirical observations
over purely analytical studies.
8) Published: Results are published, regardless of outcome.

A. Reliable Knowledge Gained as the Success Criteria

Although it is a common practice when advocating research to emphasize the
ultimate technical benefits, this practice is not constructive on topics as visionary
and provocative as seeking revolutionary advances in spaceflight. Instead, it is
more constructive to emphasize the reliability of the information to be gained.
Compared to other propulsion research, the current frontiers of propulsion
science are at their infancy. It is expected, therefore, that any practical embodi-
ment is years, perhaps decades, away, if not impossible. Although breakthroughs,
by their very definition, happen sooner than expected, no breakthrough is genuine
until it has been proven to be genuine. Hence, the reliability of the information is
a paramount prerequisite to the validity of any conclusions.
To place the emphasis where it is needed, no research approach should be con-
sidered unless it is sufficiently rigorous, regardless of the magnitude of claimed
benefit. Success is defined as acquiring reliable knowledge, rather than as achiev-
ing a breakthrough.
682 M. G. MILLIS

Visionary Rigorous

Long-Term Goals Near-Term Progress

Divergent Options Need to Focus

Sufficient Investment Affordable

Fig. 2 Management challenges for revolutionary research.

This success criterion even means that a failed concept (test, device, etc.) is
still a success if the information gleaned from that failure provides a reliable
foundation for future decisions. This is a departure from the more common
notion of judging success by how closely outcomes match expectations. In the
more common approach, a task fails if the device does not work as desired,
regardless of the lessons gained from the attempt. While such expectation-
specific success criteria are appropriate for manufactured goods, they can be det-
rimental to fundamental research. By placing the emphasis on the fidelity of the
findings, it encourages researchers to apply rigor to their work and to take the
risks necessary to discover what others have overlooked. It also makes it easier
to accept the results as they are, rather than to be tempted to skew the findings
to match the expectations.

B. Immediate Research Steps

Another technique to shift the emphasis away from provocative situations and
toward constructive practices is to focus the research on the immediate questions
at hand. These immediate unknowns, issues, and curious effects can be identified
by comparing established and emerging science to the ultimate propulsion goals.
A subsequent section in this chapter, “Identifying Important Problems,” deals
with methods to assist with this step. The scope of any research task should
ideally be set to the minimum level of effort needed to resolve an immediate
“go/no-go” decision on a particular issue. This near-term focus for long-range
research also makes the tasks more manageable and more affordable. Specifi-
cally, it is recommended that any proposed research be configured to reach a
reliable conclusion in one to three years. Should the results be promising, a
sequel can be proposed in the next solicitation cycle.
This is a departure from the Phase I and Phase II practice of both the Small
Business Innovative Research (SBIR) [53] and the NIAC [20,21] programs. In
these systems, phase-I awards are 6 to 12 month feasibility studies, whereas
Phase II awards are larger and longer-term awards to move from feasibility
toward an application embodiment. While this may be a prudent strategy for
 PRIORITIZING PIONEERING RESEARCH 683

technology that is approaching fruition, such a two-stage approach is premature

for basic research, even for applied basic research. For example, the Phase I and
Phase II approach is based on a system where success is based on the feasibility of
the concept as opposed to revolutionary research where success is based on the
lessons learned.

C. Measured Progress
To help identify a suitable research increment and to provide managers a
means to measure progress, the Scientific Method has been adapted as a readiness
scale in a manner similar to how the technology readiness levels are used to
measure technological progress [54]. The readiness scale developed for the
BPP project consists of three stages that gauge the applicability of the work
(reflecting how research can evolve from the more general, to the more specific
application), and within each of these three stages, the five steps of the scientific
method are repeated (from recognizing the problem, through testing the hypoth-
esis). This equates to 15 levels of relative maturity, with the most advanced level
being equivalent to Technology Readiness Level 1 (i.e., basic principles observed
and reported).

1. Three Stages of Applicability

The scope of scientific research can span from the very general, broad-sweep-
ing considerations all the way down to specific details of a given application. The
more focused that a research increment is toward a desired application, the
greater its applicability, as reflected in the following three stages:
Stage 1: General Physics. The research topic deals with general underlying
physics related to the desired application.
Stage 2: Critical Issues. The research topic deals with an immediate
unknown, critical make break issue, or curious effect relevant to the desired
application.
Stage 3: Desired Effects. The research topic deals with a specific effect or
device for answering the goals of the application, such as force-inducing
effects or the conversion of energy from one form to another.

2. Five Levels of the Scientific Method

Within each of these ranges of applicability, the Scientific Method can gauge
interim progress. The following definitions for the steps of the Scientific Method
have been tailored to reflect applied research progress. Because applied research
implies a tangible product, these steps distinguish between empirical approaches
(those based on the emergence of empirical effects) and theoretical approaches
(those based on theory). The most significant distinction is with the final stage,
where the hypotheses are tested. For applied science, again emphasizing a tangi-
ble product, this final stage can only be satisfied with an empirical test. Another
noteworthy distinction from the common definitions of the Scientific Method is
the inclusion of the zeroth step. This gives a placeholder for emerging opportu-
nities that have not yet been addressed.
684 M. G. MILLIS

Step 0: Pre-Science
Empirical: Observations of an unconfirmed anomalous effect have been
reported (includes observations of natural phenomena or claims of unverified
devices), or
Theory: A correlation between a desired goal (or unsolved problem) and the
existing knowledge base has been articulated.
Step 1: Problem Formulated
Empirical: An experiment has been defined that can collect the data required
to isolate and characterize the anomalous effect, or
Theory: A goal (or problem) has been defined specifically enough to identify
the remaining knowledge gaps toward achieving the goal (or solving the
problem).
Step 2: Data Collected
Empirical: Data have been collected and analyzed from experiment to isolate
and characterize the anomalous effect, or
Theory: The relevant data to fill the critical knowledge gaps, identified in the
previous step, have been collected through experiment, observation, or math-
ematical proof (this level includes assessments of theory using mathematical
analysis).
Step 3: Hypothesis Proposed
Empirical: A mathematical representation of the physical principles under-
lying an effect has been offered to explain the effect and predict additional (tes-
table) effects, or
Theory: A mathematical representation of the relation between physical
phenomena has been offered that addresses the goal (or problem) formulated
previously.
Step 4: Hypothesis Tested and Results Reported
The hypothesis has been tested by comparison to observable phenomena or by
experiment sufficiently to determine if it appears viable, and the results reported.
Note: In the context of applied research, testing of a hypothesis must be empiri-
cal; that means it must be done by comparison to observable phenomena or by
experiment, rather than just by mathematical proof. Although mathematical
proof can be used to test the consistency of a theory against known science,
such a mathematical test alone is not sufficient to warrant achievement of
Step 4. Instead, a mathematical test of a theory reflects achieving Step 2.

After a given research objective has been ranked relative to this 15-level scale,
the next logical increment of research would be to advance that topic to the next
readiness level. This is consistent with the incremental research strategy.
A “Research Summary Form” is presented in the appendix.

D. Iterated Research
To accumulate progress over the long term, it is recommended to solicit a suite
of proposals every two to three years, and to let the findings of the prior suite
influence the next round of selections. This provides an opportunity for new
approaches, sequels to the positive results, and redirections around null results.
At any point, if a research task leads to the discovery of a new propulsion or
 PRIORITIZING PIONEERING RESEARCH 685

energy effect, it can be pulled out of this process into its own advancement plan.
This strategic approach is recommended for high-gain/high-risk research, where
cycles of peer-reviewed solicitations can examine a diverse portfolio of options,
and where the decisions build on the lessons learned from prior cycles of
research.
Again, for basic research aimed at revolutionary advances, this iterative strat-
egy is recommended over the Phase I and Phase II strategy of SBIRs and NIAC
projects. The distinction is that revolutionary research requires taking risks
beyond testing seemingly feasible approaches. This is also tied to shifting the
definition of success from that of feasibility to that of lessons learned.

E. Diversified Portfolio
It is far too soon, in the course of seeking spaceflight breakthroughs, to down-
select to just one or two hot topics. Instead, a variety of research approaches
should be investigated in each review cycle. In simple terms, this is to diversify
the research portfolio. This is different than the more common practice with
advanced propulsion research where further advancements are primarily sought
on the technical approaches already under study. Although this more common
strategy can produce advances on the chosen topics, it faces the risk of overlook-
ing potentially superior emerging alternatives and the risk that support will wane
unless the chosen topics produce unambiguous positive results.

F. Impartial Reviews
When inviting research on the edge of knowledge, controversial ideas are
encountered. Considering that most historic breakthroughs originally sounded
like fringe ideas, it is not surprising that many of the proposals for breakthrough
spaceflight might sound too visionary at first, or at least unfamiliar. It is therefore
difficult to sort out the fringe ideas that may one day evolve into tomorrow’s
breakthroughs from the more numerous, erroneous ideas. During proposal
reviews, it is common to have some reviewers reflexively assume that unfamiliar
ideas will not work. To reliably determine technical feasibility, however, is
beyond the scope of a proposal review, constituting a full research task unto
itself. Instead of expecting proposal reviewers to judge technical feasibility, it
is recommended to have reviewers judge if the task is leading to a result that
other researchers will consider as a reliable conclusion upon which to base
future investigations. This includes the possibility of learning from null results.
This posture of judging credibility rather than prejudging feasibility is one of
the ways of being open to visionary concepts while still sustaining rigor.
This posture also eases the burden on the reviewers. By asking them to focus
on rigor rather than feasibility, it is easier for them to review work beyond their
immediate area of expertise. The hallmarks of rigor, and the absence thereof, are
relatively easy to spot and are similar across a wide range of disciplines.

G. Empirical Emphasis
When seeking advancements that can eventually lead to new technology, there
is a decided preference toward tangible observations over purely analytical
studies with all other factors being equal (cost, technical maturity, etc.).
686 M. G. MILLIS

Experiments, being hardware, are considered closer than theory to becoming

technology. Also, experiments are considered a more direct indicator of how
nature works. Theories are interpretations to explain observations of nature,
while the empirical data are nature, partially revealed within the constraints of
the given experiment. This tactic also helps to break from the limitations implicit
in dominant paradigms.

H. Publishing Results
The final recommendation is to ensure that the research findings are published,
regardless of outcome. Results, pro or con, set the foundations for guiding the next
research directions. Although there can be a reluctance to publish null results when
a given approach is found not to work, such dissemination will prevent other
researchers from repeatedly following deadends. Again, by defining success as
gaining reliable knowledge, such dissemination of lessons-learned becomes easier.

V. Devising Prioritization Criteria

Before detailing the selection criteria developed by the NASA BPP
Project, this section describes the process by which those criteria were developed,
including the scoring equation. These methods are offered for other managers
who are tasked to develop scoring criteria tailored to their own projects.
The BPP evaluation process was adapted from a procedure developed by
Bruce Banks during his tenure at the NASA Glenn Research Center and
evolved over several years and among a variety of teams. A documented
example of this process is the selection of the replacement thermal control
materials for the Hubble Space Telescope [55]. This decision-making process
is designed for prioritizing options where there are many issues of varying
importance and influence that make such selections complex. A multiplicative
scoring method is employed, based on principles of probability and statistics,
which is significantly more sensitive than additive scoring methods.
The process has the following advantages and characteristics:
1) It is ideal for making decisions when many complex and often conflicting
issues must be considered.
2) The process is applicable to prioritizing choices and making decisions on
almost any topic such as:
a) Strategic planning (determining which of several potential plans is
best)
b) Personnel selections or promotions
c) Career options
d) Contractor selections
e) Major purchases
3) Every issue and every opinion is considered.
4) One individual or team can perform the process, but the team approach is
recommended on those decisions where a team consensus on the final decision
is paramount.
5) The decision-making process results in excellent “buy-in” by those using
the process, because all issues and everyone’s opinions can be taken into account.
 PRIORITIZING PIONEERING RESEARCH 687

6) The process employs multiplicative scoring that is significantly more

sensitive to critical issues than additive scoring methods.
7) The decisions are prioritized in a quantified manner.
8) The resulting decisions are highly defendable.
9) The process minimizes skewing from overly assertive people.
10) The process can be performed via e-mail.
11) The mathematical methods of the process can be automated in software.

A. Participants
The process starts by assembling a team of representative experts and custo-
mers of the desired technology. “Customers” are those who are sponsoring the
research solicitation, and “experts” are representative practitioners who are
capable of conducting the research. Through brainstorming and voting, the
team defines the relevant evaluation criteria, and then narrows these criteria
down to a minimal list with weighting factors for each. The group also must dis-
tinguish among those criteria that are mandatory (criteria that must be met), and
those criteria that are just enhancing. It is essential that the customers for the
research concur with the criteria, and it is crucial that the other participants
concur before applying the criteria to actually evaluate the options.

B. Required Characteristics of Evaluation Criteria

To determine the criteria to be used in the selection, it is helpful to have a suite
of illustrative options (i.e., a sample of items whose prioritization is sought). With
these up for discussion, it is easier to start collecting a list of evaluation factors.
When the committee discusses the evaluation criteria, every proposed criterion
should be listed, regardless of how many people feel it is meritorious. After all
the criteria are listed, individual or group voting will be done to determine the
relative importance of each.
Evaluation criteria should be:
1) Phrased in a positive sense (express as desired characteristics or freedom
from undesirable characteristics)
2) Phrased such that there is majority acceptance of the wording
3) Independent of each other (no duplication of issues)
4) Address single issues (issues should not contain the word “and”)
5) Able to be numerically scored (or graded)
6) Include all relevant issues

C. Relative Weighting of Criteria

Once the criteria have been selected, the committee decides, by consensus or
majority vote, which criterion is the most important. The “relative importance” of
this criterion is assigned a value of 1. Voters decide their values for relative
importance of each evaluation criterion knowing that:
1) The most important criterion is rated 1, and is not changeable.
2) Because the most important criterion has been rated 1, the relative import-
ance of the remaining criteria will have proportional importance values between
688 M. G. MILLIS

0 and 1. For example, a criterion half as important as the most important criterion
should be rated 0.5.
3) Values can be entered that have up to three decimal places to the right of 0.

D. Multiplicative Scoring Principles

To quickly filter out substandard submissions, it is desired to have a feature
whereby any failure to meet a mandatory criterion will eliminate the entire sub-
mission from competition. To provide this feature as an integral part of an evalu-
ation system, the total score is determined by multiplying together, rather than by
adding, the individual criterion scores. In this manner, any zero score (failing
grade) on any mandatory criterion will result in a total score of zero.
To implement such a system, there are three details to take into account:
1) how to handle nonmandatory criteria, 2) how to handle weighting func-
tions, and 3) how to normalize scores. The sample equation below illustrates
a multiplicative system for two mandatory criteria and one nonmandatory
criterion:
 a  b  
A B C þ Cmin c
Total Score ¼ (1)
NA NB NC

where:
A, B, C criteria scores
a, b, c weighting factors, where 1 is the maximum value, and lower
priorities are fractions of 1
NA, NB, NC normalizing values or functions
Cmin a preset, non-zero value to prevent the parenthetical term from
equaling zero, in the event that C 0, thereby making criterion
C nonmandatory

To allow nonmandatory criteria into a multiplicative system, two different

approaches can be employed. The easiest, and the way employed with the
NASA BPP process, is to just assign a score range for that criteria where the
lowest possible score is not zero. The alternate approach, shown in the equation
above, is to include a non-zero value in the criterion’s equation. This second
approach, however, complicates how normalizing functions are included.
To accommodate weighting factors, exponents are used analogous to the way
coefficients are used for additive systems. It is recommended to use positive
values equal to or less-than 1 for these exponents, where an exponent of 1 rep-
resents the highest priority. An exponent of 0.5 represents a criterion that is
half as important.
In practice, the effect of the weighting functions also is tied to the maximum-
point-value that each criterion can attain. Therefore, it is necessary that each
criterion be normalized to the same maximum-point-value (the terms within
the parentheses). For normalization, which means equalizing each criterion
prior to applying its weighting exponent, a simple fractional coefficient is
applied, so that the maximum possible values of all the criteria are equal.
 PRIORITIZING PIONEERING RESEARCH 689

Although a generic set of equations can be derived for how to implement a

multiplicative system that accommodates all possibilities of mandatory and non-
mandatory criteria, and accommodates criteria with differing scoring ranges, it is
far simpler to implement the system with constraints on the scoring ranges. If all
criteria have the same maximum point value, no normalization is required. If all
nonmandatory criteria have a non-zero value as their minimum possible score,
then no additional constants or associated normalization functions are required.

E. Scholastic Grading Standard

Experience has shown that an evaluation depends not only on the perceived
merit of the idea, but also on the evaluators’ interpretations of how to score
the idea. For example, if the scoring range is 0 to 25 on a given criterion, such
as with the Small Business Innovative Research evaluations [53], two different
evaluators may use significantly different point values to mean the same grade.
To avoid this problem, it is recommended that a familiar and limited grading
system be used, such as the scholastic four-point scale:
A (4 points) Excellent or outstanding, meeting the criteria to the maximum
amount
B (3 points) Good, or well above average
C (2 points) Average, or the score to use if there is no reason to score high
or low
D (1 point) Poor or well below average
F (0 points) Fails to meet the criteria
In those cases where these discriminators do not fit, it is still recommended to
have the scoring range limited to about 5 gradations where possible, and to
have clear text explanations to accompany each gradation. Because the final
scores combine several criteria, it is possible to get sufficient distinctions with
the total final scores even with such limited gradations.

VI. NASA Breakthrough Propulsion Physics Research Solicitation

This section is based on the 1999 research solicitation conducted by the
NASA BPP Project [56]. It covers the selection criteria, the composite scoring
equation, the two-stage review process, and the lessons learned. A “Research
Summary Form,” with instructions, is included in the appendix to help convey
pertinent details.

A. NASA Propulsion Physics Criteria and Scoring Equation

Using methods very similar to those just described, the NASA BPP Project
created the following prioritization criteria. In the early stages of the project, a
Product Definition Team helped devise the first set of criteria and their weighting
factors [43]. This team included a mix of both the customers of the research
(NASA, DOD, DOE) and practitioners of research (physicists from a variety of
organizations). A revised set of criteria were tested during a workshop in 1997
[44], and a further revised set [18] was finally adapted for use in the formal
solicitation that spanned 1999 through 2000 [56]. The criteria listed below
incorporate revisions from the lessons learned through the solicitation process.
690 M. G. MILLIS

1. Criteria
There are three categories of selection criteria. Technical Relevance relates
directly to the project’s Grand Challenges, the Credibility criteria judge the
rigor of the research, and the Resource criteria address affordability and timeli-
ness. The total composite score is achieved by multiplying the individual criteria
scores as illustrated in Eq. (2) in the next section.

Technical Relevance:
1) Gain: Magnitude of performance improvement relative to all three of the
Grand Challenges, assuming the approach under consideration ultimately reaches
fruition.
2) Empiricism: Does the topic deal with tangible physical effects or just
theory?
3) Readiness: The present maturity of the topic/concept under study as
measured using the Applied Science Readiness Levels.
4) Progress: Magnitude of progress to be achieved by the research task, as
measured by the difference in the readiness now (criteria 3), and the anticipated
readiness level to be reached upon completion of the task, as measured using the
Applied Science Readiness Levels.
Credibility:
5) Foundations: Based on credible references.
6) Contrasts: Compared to current credible interpretations.
7) Tests: Leading toward a discriminating test.
8) Results: Probability that the task will result in knowledge that will be a
reliable foundation for future decisions.
Resources:
9) Triage: Will it be done anyway or is it unique to this project?
10) Cost: Funding required (reciprocal scoring factor).
11) Time: Time required to complete task (reciprocal scoring factor).

Some revisions were needed based on the lessons learned during the actual
solicitation. The version presented above is the revised version. In the original
solicitation, the Technical Relevance group of criteria (called Directness in the
NRA solicitation) and the Scientific Method Readiness Scale were separate.
When submitters and reviewers were scoring readiness, they often overlooked
how readiness and applicability were linked, sometimes leading to contradictory
assessments of readiness and progress. This is why the revised readiness scale has
these two criteria explicitly interwoven. The other problem was that there was a
“lineage” criteria that was found to be redundant to the “Probability of Successful
Completion” criteria. Lineage has been deleted and is now integral to “Credible
Results” (criteria 8).

2. Composite Equation
The total composite score is achieved by multiplying the individual scores, as
illustrated in Eq. (2). This has the feature whereby a failure to meet any mandatory
 PRIORITIZING PIONEERING RESEARCH 691

Table 2 Evaluation criteria correlations to composite scoring equation

Variable Score Normalizing Weighting

Criteria name range Equation variable variable
a
1: Gain G 0 4 G ¼ (GM þ WG
GS þ GE)/3
Gain, Goal 1 Mass GM 0 4
Gain, Goal 2 Speed GS 0 4
Gain, Goal 3 Energy GE 0 4
2: Empiricism E 1 4 WE
3: Readiness (now) RN 0 15 RNV ¼ 15/4b WRN
4: Progress P 0 15 P ¼ RA RN PNV ¼ 1/2c WP
Readiness (after)
RA 0 15
5: Credible Foundations CF 0 4 WCF
6: Credible Contrasts CC 0 4 WCC
7: Credible Tests CT 0 4 WCT
8: Credible Results CR 0 4 WCR
9: Triage TR 1 4 WTR
d
10: Cost (discrete bands) C 0 4 WC
d
11: Time (discrete bands) TI 1 4 WTI
a
The 1/3, embedded in the Total Gain equation, is to normalize the total gain to a maximum value of 4.
b
The Readiness Normalization Value (RNV) is set to 15/4 so that a ranking of Technology Readiness
Level 1 (score ¼ 15) equates to a score of 4.
c
The Progress Normalization Value (PNV) is set to 1/2, so that the typical progress of incrementing up
one level of the Applied Science Readiness Levels equates to a score of 2 (an average condition).
d
For Cost and Time, the scoring gradations are set to discrete bands rather that directly entering time or
cost, and where the higher scores refelct lower cost and time. Specific scoring gradations examples are
in the appendix.

criterion (zero score on criteria) will result in a total score of zero. Table 2 lists the
characteristics of each criteria for traceability to the equation.
Total Score ¼ GWG EWE (RN=RNV)WRN (P=PNV)WP
 CF WCF CC WCC CT WCT CRWCR TRWTR C WC TI WTI (2)

3. Research Summary Form

To streamline the review process, the proposal submitters were asked to
encapsulate the key points of their proposal onto a Research Summary Form.
This form is configured to pull together the information that pertained most
directly to the evaluation criteria, and to present this information in a standard
format to make it easier for the reviewers to find the key information. This
form, along with its instructions, is provided in the appendix. Separate instruc-
tions are included for both the submitters and the reviewers to illustrate
the difference between the collection of information and how the reviewer
692 M. G. MILLIS

examines that information. Note how the inputs on the form trace directly to the
evaluation criteria.
On a real proposal, the summary form would be accompanied by a 10- to 20-
page proposal document. To better convey the meaning and application of
selection criterion, the instructions for the reviewers that are presented in the
appendix deal with each criteria in detail. For example, one area of expanded
detail instructs the reviewers how to apply the four credibility criteria to accom-
modate the following five possible proposal situations:
1) New unreported effect
2) Known, unconfirmed effect (such as the Tajmar observations of frame
dragging in the vicinity of very low temperature rotating rings [57])
3) Known, confirmed effect
4) New theory/hypothesis
5) Known theory
From the lessons of the actual 1999 solicitation, the use of such a proposal
summary sheet was found to be effective. This made it very easy to quickly
filter out noncompliant proposals and to focus on the key points needing
further scrutiny. This might also become a useful tool for pre-proposal screen-
ing, and submissions of just the summary sheet could be used as a first screen-
ing, and only those submissions with acceptable summary sheets are invited to
submit a full proposal. With the 1999 solicitation, a similar form was required
from the reviewers, which made it easy to compile the scores. The version pre-
sented in the appendix combines both the submitter and reviewer inputs onto
one sheet.
Another possibility that was under development when the project funding was
deferred, was to adapt such a form to allow Internet submissions where appropri-
ate information fields would be configured to automatically reject fringe sub-
missions. This format was also to serve as the outline for archiving BPP
research findings into an electronic database.

B. Inviting Relevant Research Proposals

The NASA BPP solicitation used a NASA Research Announcement
(NRA) format that gave submitters the flexibility to suggest how their work
could contribute to the solicitation’s general goals. A lesson learned from the
solicitation is that not all the important issues were covered by a viable proposal.
If the project does not have another venue through which to direct work, such as
an in-house lab, then the broader success of the project is entirely dependent on
the proposals.
Before research proposals should be solicited, it is prudent to identify focal
issues to help guide submissions. Workshops are one way to address this issue,
provided that the workshop is designed for that purpose [44]. The Horizon
Mission methodology, already discussed, is yet another approach. In addition,
the NASA BPP Project used a “traceability map” to help contrast the founda-
tional science to the propulsion goals. This map and its methods are discussed
in Sec. VII.A.
 PRIORITIZING PIONEERING RESEARCH 693

C. Review and Selection Process

The research prioritization and selection process employed by the NASA BPP
project followed a two-stage peer-review process. In the first stage, proposals
were numerically scored relative to the criteria. In the second stage, a diverse
suite of proposals were selected from the top-ranking candidates.
Having a two-stage review process was found to be effective. The first stage
allowed each proposal to receive a thorough review, and the second stage allowed
a smaller team to effectively compare the proposals. The first stage produced a
large amount of data that was, by itself, not sufficient to identify the best propo-
sals. Even after these data were analyzed to obtain an average score for each
proposal and the value for the standard deviation of the reviewers’ scores (Fig.
3), further discussion was required to effectively review and distill these results.
In the first stage of the review process, proposals were subject to scientific
review by discipline specialists consistent with the topic matter of the proposal.
On topics as visionary as breakthrough propulsion physics, it is prudent to have
numerous reviewers to provide well-rounded assessments. For the 1999 2000
solicitation, at least four reviewers per proposal were secured for all 60 proposals,
with virtually all proposals being scored by five reviewers. Proposals were
reviewed by a combination of in-house and selected external reviewers, with
due regard for conflict-of-interest and protection of proposal information.
These external reviewers included individuals from government labs, univer-
sities, or industry.
Although it was challenging to secure the volunteer services of the more than
50 reviewers required for this coverage, having this many reviewers allowed

Fig. 3 Scores of proposals from the 1999 BPP Project solicitation.

694 M. G. MILLIS

several important functions to be employed. First, it was easy to avoid conflicts of

interest by not having any proposals scored by people from their own or compet-
ing institutions. Next, it was possible to match the areas of expertise of the pro-
posals and reviewers, which varied greatly because there is a wide span of
physics underlying the BPP Project’s ambitions. As already noted under the oper-
ating principles section, having the reviewers focus on rigor rather than feasibility
made it easier for them to review topics that were not identical to their area of
expertise.
With multiple reviewers per proposal, it was possible to calculate the standard
deviations of scores to flag any disparate reviews for further scrutiny. This
feature, of assessing the relative agreement among the reviewers of a given pro-
posal, was found to be an effective tool for screening out questionable proposals
and to identify problems with any reviews. Figure 3 reflects the scoring results
using these techniques. In particular, note how the standard deviations are
lower for the highest-scoring proposals whereas the mid-range proposals have
larger deviations, and many of the failing proposals have zero deviation. This
reflects how the best and worst proposals are more obvious (less disagreement
among the reviewers) and the medium-range proposals are more ambiguous.
The second stage of the review process involved the compilation and review of
these scores by a smaller team of reviewers, again with due regard for conflict-of-
interest and protection of proposal information. This team consisted of 10
government employees from multiple government labs, including NASA, the
Department of Energy, and the Department of Defense. With a smaller but still
diverse group of reviewers, it was relatively easy to sort through the total scores,
evaluate the disparate reviews, and then select a diverse suite of research from top-
ranking proposals. This second-stage review required about 2.5 days to complete.
A suite of different approaches is desired to “diversify the portfolio” because it
is not possible to tell which research paths will lead most directly to revolutionary
advancements. This means that the proposals selected for award were not
necessarily selected contiguously from the highest-ranking set. For example, if
the top two ranking proposals are both to perform an experimental test of
theory A, and the third-ranking proposal is to test theory B, then it is the prero-
gative of this selecting team to award the best proposal on A and the third-ranking
proposal on B, while skipping the second-ranking proposal on A if this supports
diversification of research.
Five proposals were selected for award and their final results were presented at
the 2001 Joint Propulsion Conference [58 62] along with one in-house
supported task [63]. Some of these went on to have publications in peer-reviewed
journals as well [2 14]. The intent was to use the conference sessions as a step
toward preparing for the next solicitation around which additional work could be
proposed toward promising areas and the null results dispensed.

VII. From Individual to Overall Progress

This last group of recommendations address building from the individual task
progress into overall progress. This covers three aspects: 1) plotting the task
progress relative to the goals, 2) providing metrics for upper level managers,
and 3) mitigating the risks associated with revolutionary research.
 PRIORITIZING PIONEERING RESEARCH 695

A. Task Progress in the Big Picture

As stated in the Operating Principles, overall progress results from the
accumulation of findings from individual research tasks. When managing a
project, however, it is helpful to more specifically convey how such individual
items compare to the general state of art. One method discussed is to measure
progress with the Applied Science Readiness Scale (Sec. IV.C “Measured
Progress”). Another tool toward this objective is the Traceability Map, which
helps show how the individual tasks are connected to both the founding
science and the project’s goals. Figure 4 presents an abbreviated version to
introduce the concept. In the left-most column, the various disciplines of
physics are listed, which then branch out into specific items of interest relative
to propulsion science. In the right-most column, the desired goals are listed,
which then branch out into a variety of hypothetical concepts for achieving
the desired effects. These concepts help define the critical questions.
Between these two extremes reside the individual tasks that address the rel-
evant effects, unknowns, and issues.
Figure 5 presents a more substantive example showing work supported by
the NASA BPP project. In the figure, the shaded blocks represent work sup-
ported by the project, some of which encompass further details. Numbers in
brackets refer to the resulting publications (listed with this chapter’s references).
Of these, the items having Refs. 3 to 13 and 58 to 62 were from the 1999 2000
competitive research selection; Ref. 63 is a small in-house task; Ref. 14 is a
small grant; and Refs. 1, 16, and 49 represent work accomplished on dis-
cretionary time. For a version of a traceability map that shows all the items
known to the BPP Project (144 blocks total), the reader is referred to page 16
of the project’s management report [18]. As further illustration, a traceability

Fig. 4 Introduction to the concept of the traceability map.

696 M. G. MILLIS

"Important Problems"
Rigorous Foundations Goal-Driven Visions
Disciplines Hypotheses & Issues & Concepts &
Curiosities Categories
& Tools Tests Unknowns Devices

Defining the problem for space drive research, Millis [1]

Electric potentials affect
clocks? Gravity effects via
Kinematics Ringermacher [60] electrodynamics? Space drive sails

Superconductors Net external thrust

&gravity, w/ Cavendish
balance Quantum vacuum Field space drives
Electromagnetics
Robertson [59] as reaction mass?

MASS: Thrust without propellant or beamed energy

Modify gravity
Superconductors

Tests of Woodward transient inertia, Cramer [61] Modify inertia

General relativity Mechanical

Nonviable Mechanical Antigravity, Millis [49]
device claims

Tests of Schlicher thruster, Fralick [63] Electromagnetic

momentum

Testing tangibility of quantum vacuum, Maclay Quantum vacuum propulsion

Maclay [13]
Casimir cavity forces, Maclay [5]

Repulsive Casimir forces, Maclay [58] Casimir implications

Quantum theory Maclay [8]
Casimir cavity superlattice - tuning
Esquivel-Sirvent [7]
Casimir forces in nanostructures
Micro-electro- Esquivel-Sirvent [9]
mechanical sys Casimir forces in arbitrary dielectrics
(MEMS) Mochan [10]
Casimir forces modified by band gaps,
Villarreal [11]
Quantum of negative refraction materials
Milonni [12]

Warp metrics and Yilmaz theory, Modify

SPEED: faster-than-light
Tests with amplified tunnel, Malloy Zampino [16] spacetime
Time domain group velocity, Mohahedi [3]

Freq domain group velocity, Mohahedi [4]

Quantum noise, Segev [6]

Superluminal but Is quantum

Quantum
causal tunneling tunneling really
nonlocality
Mojahedi [62] faster-than-light?
ENERGY

Deep Dirac energy Quantum

Thermodynamics
Thermodynamics Ground state energy reduction
levels, Deck [14] vacuum energy

Fig. 5 NASA BPP task traceability map.

map for this book is provided in the preface. Unfortunately, printing limitations
prevented showing the numerous interconnections between the items on the
book’s traceability map.
It is important to stress that items shown in Fig. 5 do not represent the full
range of work or even the most critical issues, but rather the opportunities that
were available to the project at the time. Many critical questions and unknowns
still remain unaddressed. Likewise, the traceability map for this book presented
in the Preface only covers the items discussed in this book. Neither is a complete
set of possibilities.
 PRIORITIZING PIONEERING RESEARCH 697

To help identify required research, John Anderson’s Horizon Methodology

[42] can be employed. This method was introduced earlier in the context of
historic lessons, and now it will be illustrated here in the context of populating
a traceability map with representative research task options. Also, this is
identical to identifying the “important problems” as in the context of Hamming’s
lecture [33].

B. Horizon Mission Methodology Applied to Traceability Map

1) Define impossible goals. These appear in the right-most column.
2) Apply Clarke’s second law and brainstorming to expand those goals into
notional concepts. These concepts appear just to the left of the goals, and connec-
tors indicate which goals are addressed by which concepts.
3) Contrast these notional concepts to our established knowledge base, by:
a) First listing the relevant foundational knowledge in the left-most
column, and then
b) Branch out from those disciplines to identify active areas of research
that pertain to the notions emerging from the right-hand side of the
chart. This includes curious effects and open hypotheses and tests.
4) “Look back from the future” of the notional concepts to identify the criti-
cal issues and unknowns that prevent their creation. These issues and unknowns
connect the notional concepts with items on the left of the diagram.
5) Complete drawing the linkages between the individual items on the
diagram, linking established disciplines to the desired revolutionary advances.
Additionally, this map could assist in prioritizing those options. In principle,
because an individual task is linked to the concepts it addresses and the foundational
science upon which it is based, the number of those connections is a mea-
sure of its relative importance. The research options that are linked to the most
concepts are more important, and the ones linked to more foundational physics
are likelier to be easier to study. The notion of applying the numerical techniques
of fault tree analysis for such prioritizations has been introduced by Zampino
[64], but has not been advanced further. This is a potential area for improvement.

C. Project Metrics of Performance

In addition to conveying how individual tasks progress maps to the
general state of art, it is also useful to address how to measure progress and
risks of the project as a whole. To that end, the following project metrics are
provided:
1) Number of incremental unknowns resolved. When numerically counting
these, devise a system whereby further advanced approaches are weighted with
higher point values to reflect their greater importance (as gauged by the readiness
levels).
2) Number of findings published in peer-reviewed literature with the goal
being 100% of submissions accepted for publication.
3) Number of citations of the published works.
698 M. G. MILLIS

4) Number of visionary notions converted into new research approaches.

5) Number of students inspired (can only count those that send comments).
6) Number of spin-offs or educational materials produced.
To measure the overall performance of such a pioneering research project, all
of the productivity measures above can be tallied, and then divided by the amount
of resources (funding and time) consumed to achieve them:
P
Project metrics
Relative performance ¼ (3)
($)  (Duration)

By including resources, it provides a means to gauge efficiency over time and

to compare the effectiveness of this project to others in an organization (provided
that the other projects are judged according to similar metrics).

D. Managing Revolutionary Project Risks

Risk management takes on a different meaning than that typically associated
with technology development. In contrast to the issues of hardware reliability,
research projects deal with knowledge. Therefore, the risks are associated with
the reliability and relevance of the knowledge. Table 3 describes those risks
along with the mitigation strategy for each.

Table 3 Mitigating revolutionary research risks

Risks Mitigations

Credibility damaged by nonrigorous Emphasize building rigorous, incremental

research reporting advancements in knowledge, rather than
requiring breakthroughs, and by
collaborating with academia and other
institutions, including peer reviews.
Leadership stature damaged by missing Pursue visionary research, extending well
relevant breakthrough or by being too beyond the known and into areas that
conservative other organizations choose not to address.
Sustain active scouting for ongoing
research, forging widespread
collaborations to find the most reliable
advanced research. Support as much
research as possible.
Competitive advantage weakened from Threshold of attention is when devices can
premature disclosures (for individuals or be engineered. Disclosure strategy is to
commercially sponsored research) publish only enough for others to
independently confirm the validity of the
operating principles, followed by the
marketing of improved versions once
independent confirmation is reliably
published.
 PRIORITIZING PIONEERING RESEARCH 699

VIII. Final Lessons and Comparisons

The BPP Project was found to be effective, although areas of improvement
were identified and discussed in the 2004 project management report [18].
Some improvements, particularly refinements to the research criteria themselves,
had already been incorporated. Other planned improvements included shifting
the operation of the project to a nonprofit consortium of government, university,
and industry. This shift was intended to improve the ability of the project to
engage with both academia and industry, respond to the large number of inter-
ested researchers and research approaches, create and maintain a database of
research findings and future options, and improve continuity to better manage
progress. These revisions were underway when funding for the project was
terminated in 2003 [65, p. 325].
The BPP Project is not the only government-funded effort to seek revolution-
ary advances. Others include the larger NASA Institute of Advanced Concepts
(NIAC) and the still larger Defense Advanced Research Projects Agency
(DARPA). For an introduction to the accomplishments produced by NAIC and
DARPA, the reader is referred Ref. 20 for NIAC and Ref. 22 for DARPA.
Because both of these other efforts have a different scope and operating
budget, direct one-to-one comparisons are inappropriate, but some features
merit mention. First, regarding the differences in scope: The BPP Project
only addressed emerging science relevant to space propulsion and had an
effective budget of approximately $0.2M/yr (total budget divided by years
of operation) [18]. NIAC’s scope was broader, seeking revolutionary concepts
on “architectures or systems” of space exploration in general, and had an
effective budget of more than 10 times the BPP effort, specifically about
$3.4M/yr [20]. NIAC began in 1998 and was terminated in 2007. DARPA
has been in existence since 1958 and has an even larger scope and budget,
seeking revolutionary advances on any aspects related to defense. In 2007,
DARPA’s portion of the budget that was devoted to space activities was
almost $500M [22], which is roughly two orders of magnitude higher than
NIAC funding. With these differences, the mechanisms of how work is
selected and supported are quite different and would be difficult to objectively
compare.
What can be examined are some of the operating principles followed by both
DARPA and NIAC. For example, DARPA rotates its staff so they are in pos-
itions no longer than six years to avoid the “we tried that, it didn’t work” syn-
drome. This tactic is one way to avoid institutional paradigms or the hesitation
to venture beyond current approaches. Another DARPA practice is that they
have “the freedom to fail.” This tactic allows researchers to take the risks
necessary to extend beyond the known. For NIAC, one of its founding charac-
teristics addressed the issue of incumbent limitations. NIAC was directed to
seek innovations from outside NASA. NASA researchers were not allowed to
compete for NIAC funding. Also, one of the criteria was to judge if the work
was revolutionary instead of evolutionary. And finally, to reflect its operating
attitude, the following quote is prominently displayed on their 2005 brochure
[21]: “Don’t let your preoccupation with reality stifle your imagination . . .”
700 M. G. MILLIS

These practices are indicative of the same historical lessons discussed at the
beginning of this chapter.

IX. Conclusions
Although pioneering research is difficult, enough lessons have accumulated
from history to guide the management of projects devoted to revolutionary
research. These historic lessons plus the precedents set with the NASA BPP
Project are offered as guidance to future managers and researchers of revolution-
ary concepts.
A key recommendation is to combine vision with rigor. Vision is needed to
extend beyond existing knowledge, whereas rigor is needed to impartially
compare those visions to accrued knowledge. The intent from that contrast is
to identify the critical issues, make/break questions, and curious effects related
to the desired goals. Once articulated, these become the important problems
for future, pioneering research.
Another key recommendation is to not attempt to judge technical feasibility
during proposal review, because that would constitute a research task unto itself.
Instead, focus attention on judging if the proposed work will reach a reliable
conclusion upon which other researchers and managers can make sound
decisions for the future. To make such judgments, focus on the following
four credibility criteria: foundations, contrasts, testability, and results. Foun-
dation refers to the source material from which the proposal was based. Con-
trast is how the proposed concepts are compared to the accrued knowledge:
Does the work possess the self-criticality that is characteristics of great
researchers? Does the work reflect a working knowledge of accrued science
and technology? Testability asks if the research is advancing toward a discrimi-
nating test. And finally, results refers to how likely the results, pro or con, will
be reliable enough to become a basis for future management and research
decisions. This, in large part, can be reflected by the track record of the research
team.
From the experiences of the NASA project over its seven years, some
additional useful strategies have been:
1) Breaking down the long-range goals into near-term immediate “go/no-
go” research objectives that can each be assessed within one to three years.
2) Devising a numerical means to impartially compare research options and
inherently reject nonrigorous submissions.
3) Addressing a diversified portfolio of research approaches.
Other insights into revolutionary research have been offered that demonstrate
that revolutionary research faces implementation challenges in addition to the
challenges of the research itself, and that these challenges can be met. It is
hoped that by articulating these leading-edge project challenges and mitigating
strategies, that other leading-edge research projects can improve their prospects
for success.
PRIORITIZING PIONEERING RESEARCH 701

Appendix

I. Research Summary Form

702 M. G. MILLIS

II. Instructions for the Research Summary Form

A. Instructions for the Submitter
The submitter only fills in blocks that are labeled with a capital letter.
The reviewers will fill in the blocks in the “Review Column.” The entries onto
this summary form should only be an introduction to the proposed work as
opposed to a thorough explanation of the work. Reference citations are
required parts of this form that are intended to give interested readers leads to
further study.
BLOCK A: Submitted by
This block contains three separate fields: Submitter’s first name with middle
initial (as one field), submitter’s last name, and the date of submission (mm/
dd/yy format).
BLOCK B: Title of Submission
For published results, enter the full reference citation in Block B. For proposed
tasks or suggested inquiries, enter an appropriate title. The use of familiar key-
words is encouraged to help convey context.
BLOCK C: Submission Type
Check one box to reflect the type of submission (research findings, proposed
tasks, or suggested areas of inquiry), and note the other associated data blocks to
skip or include.
Note: Adding this block allows the form to also be used for cataloging results
or provisional research suggestions into a database.
BLOCK D: Central Issue, Unknown, or Observation Under Study
Provide a short paragraph to reflect the key focus of the completed (or pro-
posed) work. What is the central, single question that this work aims to
resolve? Use Block N to list the primary references that gave rise to this incre-
ment of work.
BLOCK E: Findings
This block should only be filled out if results of published research are
being submitted. In that case, provide a short paragraph reflecting the conclusions
of the research.
BLOCK F: Sequels Expected
This block should only be filled out if results of published research are being
submitted. In that case, indicate whether continued investigations are expected.
Note that null-findings, where no sequels are expected, are still considered valu-
able results worth disseminating.
BLOCK G: Relevance to Project
Provide a brief paragraph explaining how the research is relevant to the ultimate
goals of the project. For general physics research, some traceability to the final
goals needs to be articulated.
BLOCK H: Ultimate Improvement
This is a continuation of the relevance question, specifically addressing the
degree to which the research is relevant to the ultimate three goals of the
 PRIORITIZING PIONEERING RESEARCH 703

project. For these questions, assume that the work reaches a successful con-
clusion and that all subsequent works building on its result ultimately lead to a
new propulsion or power device. For each of the three project goals (mass,
speed, energy), check the box pertaining to the statement that best describes
the ultimate envisioned performance of this new propulsion or power device.
Granted, it may be difficult to predict this because the concept may be far from
fruition, but these questions provide a way to distinguish among different
levels of ambition.
MASS: Discover new propulsion physics that eliminates or dramatically reduces
the need for propellant.
0 Not applicable to this goal.
1 Applicability doubtful.
2 Applicable, but potential impact unknown.
3 Intended to significantly reduce propellant requirement.
4 Intended to eliminate the need for propellant and the need for directed
energy.
(The term “directed energy” means any form of energy sent from a central
location such as from the Earth or sun.)
SPEED: Discover how to circumvent existing speed limits to dramatically
reduce transit times.
0 Not applicable to this goal.
1 Applicability doubtful.
2 Applicable, but potential impact unknown.
3 Intended to eliminate speed constraints caused by limits of propellant
or energy supply.
4 Intended to circumvent the light speed limit.
ENERGY: Discover new energy physics sufficient for interstellar flight.
0 Not applicable to this goal.
1 Applicability doubtful.
2 Applicable, but potential impact unknown.
3 Better energy conversion physics, but still limited to a consumable
onboard supply.
4 Intended to provide energy sources and conversion methods accessible
in flight.
BLOCK I: Increment of Work Proposed or Reported
Provide a brief paragraph describing the work performed (or proposed) to
resolve the critical issue, unknown, or observation under study that was specified
in Block D.
BLOCK J: Type of Research
Use this block to specify whether this work is theoretical or experimental, by
checking the one box that best describes the type of research, using the options
offered below:
Study Comparative study, data collection, or literature search.
Theory Theoretical work only, without empirical investigations.
704 M. G. MILLIS

Expmt Experimental tests or empirical observations only, not aimed at veri-

fying or extending a specific theory.
Ex&Th Experiment or empirical observations coupled with theory. (This
implies that the theory is sufficiently advanced to numerically predict effects or
assess the utility and scalability of the effects beyond just a single demonstration
experiment.)
BLOCKS K and L: Readiness and Progress †
Using the readiness levels described below, specify the level of progress of the
work by checking the corresponding boxes. Note that two columns are required to
be checked: one specifying the readiness before starting the task, and the other
for the readiness after completion of the task (actual completion level for
reported results, and predicted completion level for proposed research).
Answer this question within the limits of the specific increment of research
being addressed. Note that “successful completion” is defined as completing the
proposed work and learning more about reaching the breakthrough, rather than
actually achieving the breakthrough.
BLOCK M: Contrasting or Skeptical Challenges to the Proposed (or
Reported) Approach
Provide a brief paragraph summarizing the key skeptical challenges against
the work. For proposed work, mention what is being done to answer the
challenges. For the results of completed work, the response to the challenges
can be addressed in the Findings section (Block E), and/or the Increment of
Work section (Block I). Use Block Q to list the primary references for
the skeptical challenges. In those cases where specific challenges have not
been published, at least describe the weak points of the work and cite textbooks
where the basic foundations behind such issues are found. For the BPP Project,
researchers are required to explicitly cite and respond to the skeptical challenges
to the work.
BLOCK N: Founding References
List the primary references upon which the work is based. These are the refer-
ences from which the central issue, described previously in Block D, can be
found.
BLOCK O: Representative Graphic (Optional)
This block is optional. The intent of this block is to provide a quick visual icon
representing the essence of the work. In the final compiled document where many

†
The Applied Science Readiness Levels were devised to provide a measure for the progress of
applied science in an analogous manner to how the Technology Readiness Levels are used to rank rela
tive maturity of engineering developments. Specifically, these Applied Science Readiness Levels
consist of three stages for applicability (reflecting how applied research evolves from the more
general to the more specific), and within each of these three levels, five steps of the Scientific
Method are repeated. This equates to 15 levels of relative maturity, with the most advanced level
being equivalent to Technology Readiness Level (TRL) 1: Basic Principles Observed and Reported.
The form also has a placeholder for progress beyond this level (TRL2). A more thorough description
of these levels in provided in Sec. IV.C.
 PRIORITIZING PIONEERING RESEARCH 705

summary sheets are presented, such visual icons are useful to quickly find items
of interest. Provide a graphic that represents some key feature of the work such as
graphs of important data, photographs of hardware, significant equations, dia-
grams, etc.

BLOCK P: Related Disciplines

Check all boxes that apply. This block is used for identifying those disciplines
that would be needed in a review and identifying which disciplines might benefit
from the findings of the described research.
BLOCK Q: Contrasting or Skeptical References
List the primary references that describe the skeptical challenges to this work,
as previously discussed in Block M. In those cases where specific challenges have
not been published, at least cite textbooks where the basic foundations behind
such issues are found.
BLOCK R: Prior Publications to Reflect Qualifications of Proposed
Researchers
For proposed work, list examples of prior publications from the principal
Investigator that demonstrate the ability to conduct and complete the proposed
research.
BLOCK S: Resources
In addition to specifying the cost (in 1000s of U.S. dollars), and the
anticipated duration of the proposed research in their appropriate boxes,
also specify the “triage” situation. Triage is used here to predict whether the
proposed work will be conducted anyway, or whether it is uniquely suited to the
BPP project. The explanations accompany the checkbox options are as follows:
Elsewhere Certain to be credibly done without the support of the project.
? Unknown. May or may not be done without the support of the project.
Only-BPP Exclusively suited to the BPP Project.
BLOCK T: Performing Organization
Specify the affiliation of the Principal Investigator under which the work
has been conducted or is being proposed. Only the name and general location
of the organization is required, not necessarily the full address. It is assumed
for proposed work that the Submitter is the Principal Investigator or a represen-
tative of the Principal Investigator. In the case of published findings, it is assumed
the first author cited in Block C is the Principal Investigator.
BLOCK U: Other Sponsors (Optional)
If the proposed (or completed) work is (or was) cosponsored by another affilia-
tion, please list this here. Only the name and general location of the organization
is required.

B. Instructions for Evaluators

The submitter only fills in blocks that are labeled with a capital letter.
Reviewers will fill in the blocks in the “Review Column.” The entries on this
summary form are only an introduction to the proposed work as opposed to a
706 M. G. MILLIS

thorough explanation of the work. See the full text of the proposal before evalu-
ating. The instructions that follow explain how the entry blocks correspond to the
criteria. More information is provided to help interpret these criteria.
Reviewer ID#
Self-explanatory
SCORES
The first and last rows of this block are part of the automated scoring equations
that will register data once all the information has been entered. The middle row
is where you enter your subjective assessment of the proposal, using the follow-
ing familiar scholastic gradations. Unless otherwise specified, these gradations
will be used for other criteria scores:
A (4 points) Excellent or outstanding, meeting the criteria to the maximum
amount.
B (3 points) Good, or well above average.
C (2 points) Average, or the score to use if there is no reason to score high
or low.
D (1 point) Poor or well below average.
F (0 points) Fails to meet the criteria.
The bottom row, “Automated Pre-Score,” is an automatic score generated
from the inputs of the submitter as they select gradations from their own input
blocks. The equation that supports this function is set up such that failure to
meet any mandatory criteria by virtue of the information on the form will
render a zero score. This includes looking for null entries for required inputs
(especially Blocks M, N, Q, and R).

SEQUELS JUSTIFIED
In the case where results are reported instead of being a proposal (see Block
C), judge if the issues (Block D), findings (Block E), the submitter’s assessment
(Block F), and relevance (Block G) merit further inquiry. Select one of the three
stoplight options provided.

RELEVANCE: 1. Gains Toward Goals

This criterion grades how the proposal relates to all three of the BPP Grand
Challenges, assuming that the concept behind the proposed research ultimately
reaches fruition. Each Grand Challenge is graded separately and the final Gain
criteria involves the sum of these three subset scores.
It is mandatory that the proposed work seek advances in science that are in
some way relevant to the Project’s three propulsion challenges or any critical
issues or unknowns related to these goals. The scope is limited to further
advances in science from which genuinely new technology can eventually
emerge, technology to surpass the limits of existing methods as opposed to
further developments of known technology. This means that if the proposed
work only builds on known technology, then it fails this criterion.
For each of the three Grand Challenges, specify which of the statements best
describes the ultimate achievable performance of the concept being addressed by
the proposal, while entertaining the assumption that a final embodiment ultimately
functions as desired. Granted, it may be difficult to predict this ultimate impact
because the concepts may be far from fruition or the concept may appear not to
 PRIORITIZING PIONEERING RESEARCH 707

be viable. For grading this criterion, assume for the moment, that the concept is
viable and will reach fruition. Other criteria will grade readiness and credibility.
Note that this is a mandatory criterion, which means that a failure to meet
this criterion (a zero score on all three subcriteria) will result in a total score of
zero. Because the scores for all three Grand Challenges will be added, it is
only mandatory that one of the three goals be addressed.
Grand Challenge 1—MASS: Discover new propulsion physics that eliminates
or dramatically reduces the need for propellant. Scoring Gradations:
0 Not applicable to this goal (default answer if no answer specified).
2 Applicable, but potential impact unknown.
3 Intended to significantly reduce propellant requirement.
4 Intended to eliminate the need for propellant and the need for directed energy.
(The term “directed energy” means any form of energy sent from a central
location such as from the Earth or sun.)
Grand Challenge 2—SPEED: Discover how to circumvent existing speed
limits to dramatically reduce transit times. Scoring Gradations:
0 Not applicable to this goal (default answer if no answer specified).
2 Applicable, but potential impact unknown.
3 Intended to eliminate speed constraints caused by limits of propellant or
energy supply.
4 Intended to circumvent the light speed limit.
Grand Challenge 3—ENERGY: Discover new energy physics to power
these propulsion devices at levels sufficient for interstellar flight. Scoring Grada-
tions:
0 Not applicable to this goal (default answer if no answer specified).
2 Applicable, but potential impact unknown.
3 Better energy conversion physics, but still limited to a consumable
onboard supply.
4 Intended to provide energy sources and conversion methods accessible
in flight.

RELEVANCE: 2. Empiricism
Does the topic deal with tangible physical effects or just theory? Because this
project is interested in advancements that can eventually lead to new technology,
and because empiricism is necessary to validate theories, there is a decided prefer-
ence toward empirical observations over purely analytical studies, all other factors
being equal. Experiments, being hardware, are considered closer than theory to
becoming technology. Also, experiments are considered a more direct indicator
of how nature works. Theories are interpretations to explain observations of
nature, while the empirical data are nature. The most desired research task is an
experiment that is coupled with theory. Experiments that are backed by a sound
theoretical foundation provide a means to numerically assess the utility and
scalability of the effects beyond just a single demonstration experiment. The
next preference is experimental work by itself; for example, to independently
708 M. G. MILLIS

test a claimed anomalous effect. The next preference is theoretical work by itself.
Lowest on this priority scale is work that only involves comparative studies of
existing approaches or literature searches. Scoring Gradations:

1 Comparative study, data collection, or literature search.

2 Theoretical work only, without empirical investigations.
3 Experimental tests or empirical observations only.
4 Experiment or empirical observations coupled with theory.

RELEVANCE: 3 –4. Readiness and Progress

Specify the level of readiness and progress of the work by checking the
corresponding boxes. Note that two columns are required to be checked: one spe-
cifying the readiness at the before starting the task, and the other for the readiness
after completion of the task (actual completion level for reported results, and
predicted completion level for proposed research). Note that item 4: Calculated
Progress will be automatically filled in by the difference of the two columns,
including application of its normalization function.
Answer this question within the limits of the specific increment of research
being addressed. Note that “successful completion” is defined as completing the
proposed work and learning more about reaching the breakthrough, rather than
actually achieving the breakthrough.
CREDIBILITY: 5. Foundations
This criterion grades how well the proposed work is grounded in credible
foundations. The proposed work must be based in some way on data or theories
that are in the peer-reviewed literature. Note: Requiring reference citations is
one of the techniques to filter out “fringe” submissions. This is a mandatory cri-
terion, which means that failure to meet this criterion (zero score) will result in
a total score of zero. Grade this criterion on how well the author identifies the
most relevant references for the topic of investigation. A variety of specific
guidelines are provided below on how this criterion maps to different proposed
situations. The scoring gradations follow the previously described scholastic
grading scale.
New, Unreported Effect: In cases where an unconfirmed anomalous effect is
being investigated (where the effect has not yet been independently
reported or confirmed in the peer-reviewed literature), the author must
cite peer-reviewed references to indicate why the newly observed phenom-
enon would be considered anomalous. For example, in the case of anom-
alous thrust observations, where no reaction mass is readily apparent, it
would be appropriate to cite references on momentum conservation.
Note: Requiring an admission that the effect does not match the physics
known to date is one of the techniques to filter out “fringe”
submissions. If the author assumes that the effect is genuine, despite not
having been independently confirmed, or despite the appearance of contra-
dicting known physics, then the author fails this criterion. For consistency
in this case, it is expected that the readiness level specified under criteria
3 should match Scientific Method Step 0 (Pre-Science) at whatever
Applicability Stage fits the proposed work.
 PRIORITIZING PIONEERING RESEARCH 709

Known, Unconfirmed Effect: In cases where an unconfirmed anomalous

effect is being investigated that has been reported in the literature, then
these references must be cited. It is expected, in this case, that the existing
literature would already list suspect causes for a “false positive” to help
guide independent confirmation or refutation. For consistency in this case,
it is expected that the readiness level specified under criteria 3 should
match Scientific Method Step 1 (Problem Formulated), at whatever Applica-
bility Stage fits the proposed work. If the suspect causes have not yet been
articulated in the literature, then the readiness level specified under criteria
3 should match Scientific Method Step 0 (Pre-Science). If the author
assumes that the effect is genuine, despite not having been independently
confirmed, or despite the appearance of contradicting known physics, then
the author fails this criterion.
Known, Confirmed Effect: In the case where the proposed work builds on an
effect that has already been confirmed in the peer-reviewed literature, then the
author must cite those references. For consistency in this case, it is expected that
the readiness level specified under criteria 3 should match Scientific Method
Step 2 (Data Collected) at whatever Applicability Stage fits the proposed work.
New Theory: In the case where work involves a theory that is not yet in the
peer-reviewed literature, then the author must cite peer-reviewed references
of the data or phenomena with which they are claiming consistency. It is
not necessary that the author agree with current interpretations of these
data, but it is mandatory that the theories are consistent with credible empirical
evidence. For consistency in this case, it is expected that the readiness level
specified under criteria 3 should match Scientific Method Step 2 (Data
Collected) at whatever “Applicability Stage” fits the proposed work.
Known Theory: In the case where the proposed work builds on a theory that is
already in the peer-reviewed literature, then the author must cite those refer-
ences. For consistency in this case, it is expected that the readiness level speci-
fied under criteria 3 should match Scientific Method Step 3 (Hypothesis
Proposed) at whatever Applicability Stage fits the proposed work.
CREDIBILITY: 6. Contrasts
This criterion grades how well the author articulates how the proposed work
compares to existing credible interpretations, relative to the BPP Grand Chal-
lenges. This is to ensure that an idea is oriented toward the goals of the
Project, and to ensure that the author has done the requisite homework on the
existing literature. This not only checks for relevance to BPP, but also positions
the work to address the next criterion of a discriminating test. Note: Requiring
reference citations is one of the techniques to filter out “fringe” submissions.
This is a mandatory criterion which means that a failure to meet this criterion
(zero score) will result in a total score of zero. Grade this criterion by how
well the author identifies the most relevant literature and on their understanding
of this literature. Again, the usual scholastic scoring gradations apply.
Also, recall from the objectives of the BPP project, that the perceived correct-
ness of the author’s alternative interpretations is not being judged with this cri-
terion. Unless there is some obvious error, it is considered too difficult to reliably
710 M. G. MILLIS

determine such feasibility during a proposal review. Such an assessment would

constitute a full research task unto itself. The burden of addressing feasibility,
via a discriminating test, is addressed by criteria 7. Instead, judge how well the
author demonstrates an understanding of the current, credible interpretations
that are cited, and the author’s ability to contrast this prior knowledge to the
approach they are offering. Recall that the BPP Project examines emerging
physics in the context of propulsion and power, and as such, there is latitude to
consider alternative perspectives beyond that from general physics. Even
though the current credible interpretations have already passed their own rigorous
tests, this does not imply that such interpretations are a complete or best represen-
tation of the actual underlying physics, especially in the context of propulsion and
power. Conversely, however, if the proposed interpretation has already been
raised and dismissed in the open literature, then the author must cite and
address these references.
A variety of specific guidelines are provided below on how this criteria maps
to different proposed situations.
New, Unreported Effect: In cases where an unconfirmed anomalous effect is
being investigated, where the effect has not yet been reported or confirmed in
the peer-reviewed literature, then the author must focus on comparing the
effect with other, credibly known effects that might lead to a false-positive
conclusion. In the prior criteria (“foundations”), the author had to acknowl-
edge that the effect was anomalous. In this criterion, the author must demon-
strate that they are astute to the conventional interpretations that must be
tested to determine if the effect is genuinely new. References that cover
these conventional interpretations must be cited. Also, the author must
explain why the effect (if genuine) might be advantageous to the BPP chal-
lenges. If the author does not address the issue of ruling out the suspect
causes, then the author fails this criterion.
Known, Unconfirmed Effect: In cases where an unconfirmed anomalous effect
is being investigated, that has already been reported in the literature, two
different scenarios can apply. If the existing published report (that the
author had to cite under “foundations”) did not list a well-rounded set of
suspect causes for a “false positive,” then judge adherence to this criterion
in the same manner stated for the case of new unreported effects. On the
other hand, if the existing published report did sufficiently list suspect
causes for a “false positive,” then the original report will suffice for the
required “contrast” citation, but the author must still demonstrate that they
are astute to the conventional interpretations necessary to determine if the
effect is genuinely new. Also, the author must explain why their proposed
investigation is more applicable than the existing or past investigations into
the effect. In the case where null results were previously published, the
author must cite these and explain why these prior tests were incomplete, in
error, or why a reinvestigation is warranted. Reference citations for these
other investigations are required. Also, the proposal must explain why the
effect (if genuine) might be advantageous to the BPP challenges. If the
author does not address the issue of ruling out the known, suspect causes,
then the author fails this criterion.
 PRIORITIZING PIONEERING RESEARCH 711

Known, Confirmed Effect: In the case where the proposed work builds on an
effect that has already been confirmed in the peer-reviewed literature, then the
author must explain why the effect might be relevant or advantageous to the
propulsion challenges and why the investigation is more applicable to BPP
than the prior or ongoing investigations into the effect. Reference citations
for the confirmation publication, and for the prior or ongoing investigations,
are required. If the author is challenging the current interpretations of the
effect, then also judge this criterion by the guidance offered under “new
theory,” below.
New Theory: In the case where a theory is proposed that is not yet in the
peer-reviewed literature, then it is mandatory that the new theories be com-
pared to the contemporary theories that address the same phenomenon. Refer-
ence citations for the contemporary theories are required. The comparison
must explain why the new theory would be more advantageous to the propul-
sion challenges than the contemporary theories. Judge this criterion by how
well the author demonstrates an understanding of the existing theories, and
on the author’s ability to identify the unresolved issues of both theories
with respect to the goals of breakthrough propulsion or power. The author
must also demonstrate a willingness to consider that the theory might be in
error, by identifying its weak points. If the author assumes that the theory is
correct, without it having been confirmed with rigorous empirical tests, then
the author fails this criterion.
Known Theory: In the case where the proposed work builds on a theory that is
already in the peer-reviewed literature, then the author must describe how the
work will be more applicable to BPP than the prior or ongoing work on the
same theory. Reference citations for the contemporary theories are required.
If the theory is still under debate in the open literature, then the author must
acknowledge its potential weaknesses, and cite references that highlight
these issues. Judge this criterion by how well the author demonstrates an
understanding of the known theory, its debated issues, and on the author’s
ability to identify how the theory applies to the goals of breakthrough propul-
sion or power.

CREDIBILITY: 7. Credible Tests

This criterion grades how well the research advances the topic toward a dis-
criminating test. It is required that the proposed work be leading toward a discrimi-
nating test or actually be a discriminating test. If a discriminating test can be
completed within the budget and time guidelines requested of proposals, it is necess-
ary that the test actually be proposed. Otherwise, it is sufficient to propose the design
of an experiment for a make-or-break test, or to further advance a theory toward
testable predictions. Again, the usual scholastic scoring gradations apply.
This requires that the author must identify the critical make-or-break issues for
their immediate area of investigation. Also, the proposed next-step must be con-
sistent with the scientific method, with due consideration for the current status of
the topic as specified by the author. Further note that, depending on the status of
the proposed task, independent verification may be warranted. In such a case, the
vested interests of the Principle Investigator must be taken into account. This is a
712 M. G. MILLIS

mandatory criterion, which means that a failure to meet this criterion (zero score)
will result in a total score of zero. A variety of specific guidelines are provided
below on how this criterion maps to different proposed situations.
Unconfirmed Effect (Reported or Not): In cases where an unconfirmed anom-
alous effect is being investigated, a discriminating test must be suggested that
could distinguish between possible conventional explanations or whether this
is a genuine new effect. The task should propose to at least design a discrimi-
nating experiment, or to actually conduct an experimental test. The work will
be considered more credible if the proposal concentrates only on the exper-
imental methods rather than on speculating on a new cause for the effect.
For consistency in this case, it is expected that the completion readiness
level (the level anticipated after the task is completed), will be at least at
Scientific Method Step 1 (Problem Formulated) for an experiment design,
or Scientific Method Step 2 (Data Collected) if an experimental test is actually
planned.
Known, Confirmed Effect: In the case where the proposed work builds on an
effect that has already been confirmed in the peer-reviewed literature, a
logical next step would be to develop a theory to describe the anomaly. It
would also be appropriate to propose a reconfiguration of the effect so that
its propulsive or energy implications could be assessed. For consistency in
this case, it is expected that the completion readiness level could be
anywhere between Scientific Method Step 2 (Data Collected) through Scien-
tific Method Step 4 (Hypothesis Tested), depending on the breadth of the
proposed work.
Theory: In the case where the proposed work deals with theory, it is mandatory
that the new theories are at least matured to the point where mathematical
models are offered (this is one of the “fringe” filters). Then, either further
mathematical analysis to predict testable effects, comparison to credible
empirical observations, or experimental tests must be proposed that can
bring the theory closer to a correctness resolution. An actual empirical test
is preferred. For consistency in this case, it is expected that the completion
readiness level could be anywhere between Scientific Method Step 1
(Problem Formulated) through Scientific Method Step 4 (Hypothesis
Tested), depending on the breadth of the proposed work.
CREDIBILITY: 8. Results
This criterion grades the expected fidelity of the conclusions to be reached at
the end of the proposed task. Will the task result in knowledge that will be a
reliable foundation for future research decisions? Again, the usual scholastic
scoring gradations apply.
Successful completion of the research task is defined as learning more about
reaching the breakthrough, rather than actually achieving the breakthrough.
Negative test results are considered progress. What is required, for successful
completion, is that the work reaches a credible resolution that is clearly commu-
nicated. If it is likely that the work can be completed within the funding and time
allocations specified, and that the results will be accepted by other researchers as
a credible foundation for future work, then a high score is warranted. Base this
 PRIORITIZING PIONEERING RESEARCH 713

assessment on a combination of the realism of the proposed work, its cost and
schedule, and on the credentials of the proposed research team and their facilities.
If cost-sharing is mentioned in the proposal, judge this criterion on the total
resources to be devoted, not just the amount to be charged to NASA. Consider
the clarity and quality of the proposal and any prior publications from the
authors as a good reflection of the clarity and quality of the final product. Note
too that, depending on the status of the proposed task, independent verification
may be warranted. In such cases the vested interests of the Principle Investigator
must be taken into account to ensure that there is no conflict of interest in the
outcome of the device, phenomenon, or theory under test. This is a mandatory
criterion, which means that a failure to meet this criterion (zero score) will
result in a total score of zero.

RESOURCES: 9. Triage
Will this research be done anyway or must this project support it? This cri-
terion addresses the possibility that the BPP project can save its resources if
the topic is likely to be explored without support of the BPP project. Specify
which statement best describes the situation. Note that this is not a mandatory cri-
terion. A minimum score here will only result in demoting an overall “A” grade
to a “C” grade. Scoring gradations:
1 (D) Certain to be credibly done without the support of the BPP project.
2 (C) Unknown.
3 (A) Exclusively suited to the BPP project.

RESOURCES: 10. Cost

This is a reciprocal scoring factor that addresses practical resource concerns.
The more costly the work, the lower the overall score, all other factors being
equal. Scoring gradations follows. (Note that the threshold values are based on
1999 costs.)
1 0 (F) If the cost is outrageous, then assign a failing grade.
2 (D) Cost  $400K: Below average.
3 (C) Cost $200K: Average.
4 (B) Cost $100K: Good or well above average.
5 (A) Cost  $50K: Excellent (but verify that this is realistic for the work
offered).

RESOURCES: 11. Time

This is a reciprocal scoring factor that addresses practical resource concerns.
The longer to reach a reliable conclusion, the lower the overall score, all other
factors being equal. Scoring Gradations:
1 (D) Duration  3 years: Below average.
2 (C) Duration 2 years: Average.
3 (B) Duration 1.5 years: Good or well above average.
4 (A) Duration  1 year: Excellent (but verify that this is realistic for the
work offered).
714 M. G. MILLIS

References
[1] Millis, M. G., “Challenge to Create the Space Drive,” Journal of Propulsion and
Power, Vol. 13, No. 5, 1997, pp. 577 582.
[2] Millis, M. G., “NASA Breakthrough Propulsion Physics Program,” Acta Astronautica,
Vol. 44, No. 2 4, 1999a, pp. 175 182.
[3] Mojahedi, M., Schamiloglu, E., Hegeler, F., and Malloy, K. J., “Time Domain Detec
tion of Superluminal Group Velocity for Single Microwave Pulses,” Physical Review
E, Vol. 62, 2000, pp. 5758 5766.
[4] Mojahedi, M., Schamiloglu, E., Agi, K., and Malloy, K. J., “Frequency Domain
Detection of Superluminal Group Velocities in a Distributed Bragg Reflector,”
IEEE Journal of Quantum Electronics, Vol. 36, 2000, pp. 418 424.
[5] Maclay, G. J., “Analysis of Zero Point Electromagnetic Energy and Casimir Forces in
Conducting Rectangular Cavities,” Physical Review A, Vol. 61, 2000, 052110 1
052110 18.
[6] Segev, B., Milonni, P. W., Babb, J. F., and Chiao, R. Y., “Quantum Noise and Super
luminal Propagation,” Physical Review A, Vol. 62, 2000, 0022114 1 0022114 15.
[7] Esquivel Sirvent, R., Villarreal, C., and Cocoletzi, G. H., “Superlattice Mediated
Tuning of Casimir Forces,” Physical Review A, Vol. 64, 2001, 052108 1 052108 4.
[8] Maclay, G. J., Fearn, H., and Milonni, P. W., “Of Some Theoretical Significance: Impli
cations of Casimir Effects,” European Journal of Physics, Vol. 22, 2001, pp. 463 469.
[9] Esquivel Sirvent, R., Villarreal, C., Mochan, W. L., and Cocoletzi, G. H., “Casimir
Forces in Nanostructures,” Physica Status Solidi (b), 230, 2002, pp. 409 413.
[10] Mochan, W. L., Villarreal, C., and Esquivel Sirvent, R., “On Casimir Forces in
Media with Arbitrary Dielectric Properties,” Revista Mexicana de Fisica, Vol. 48,
2002, p. 339.
[11] Villarreal, C., Esquivel Sirvent, R., and Cocoletzi, G. H., “Modification of Casimir
Forces due to Band Gaps in Periodic Structures,” International Journal of Modern
Physics A, Vol. 17, 2002, pp. 798 803.
[12] Milonni, P. W., and Maclay, J., “Quantized Field Description of Light in Negative
Index Media,” Optics Communications, Vol. 228, 2003, pp. 161 165.
[13] Maclay, J., Forward, R., “A Gedanken Spacecraft that Operates Using the Quantum
Vacuum (Dynamic Casimir Effect),” Foundations of Physics, Vol. 34, 2004, pp. 477 500.
[14] Deck, R., Amar, J., and Fralick, G., “Nuclear Size Correction to the Energy Levels
of Single Electron and Muon Atoms,” Journal of Physics G, Vol. 38, 2004,
pp. 2173 2186.
[15] Millis, M. G., “Assessing Potential Propulsion Breakthroughs,” New Trends in Astro
dynamics and Applications, Belbruno, E. (ed.), Annals of the New York Academy of
Sciences, New York, Vol. 1065, 2005, pp. 441 461.
[16] Zampino, E. J. “Warp Drive Metrics and the Yilmaz Theory,” Journal of the British
Interplanetary Society, Vol. 59, 2006, pp. 226 229.
[17] Millis, M. G., “Warp Drive, When?,” 2006. URL: ,http://www.nasa.gov/centers/
glenn/research/warp/warp.html. [cited 27 Feb. 2008].
[18] Millis, M. G., “Breakthrough Propulsion Physics Project: Project Management
Methods,” NASA TM 2004 213406, 2004.
[19] Merkle, C. L. (ed.), “Ad Astra per Aspera, Reaching for the Stars,” Report of the
Independent Review Panel of the NASA Space Transportation Research Program,
Jan. 1999.
 PRIORITIZING PIONEERING RESEARCH 715

[20] Cassanova, R., Jennings, D., Turner, R., Little, D., Mitchell, R., and Reilly, K.,
NASA Institute of Advanced Concepts, 9th Annual & Final Report, 2006 2007, 12
July 2006 31 August 2007.
[21] Bradley, A. (ed.), The First Five Years and Beyond, NASA Institute of Advanced
Concepts. URL: ,http://www.niac.usra.edu. [cited 29 Feb. 2008].
[22] Wislon, J. R., “Fifty Years of Inventing the Future, 1958 2008 (DARPA),”
Aerospace America, Vol. 46, No. 2, 2008, pp. 30 43.
[23] Foster, R. N., Innovation: The Attacker’s Advantage, Summit Books, New York,
1986.
[24] Shepherd, D. A., and Shanley, M., “Common Wisdom on the Timing of the Entry,”
New Venture Strategy, Sage, Thousand Oaks, CA, 1998, pp. 1 14.
[25] Henderson, R. M., and Clark, K. B., “Architectural Innovation: The Reconfiguration
of Existing Product Technology and the Failure of Established Firms,” Administra
tive Science Quarterly, Vol. 35, 1990, pp. 9 30.
[26] “Bezos in Space,” Newsweek on MSNBC News (5 May). URL: ,http://
www.msnbc.com/news/9o4842.asp. [cited 29 May 2003].
[27] “Race to Blast Tourists into Space,” CNN.com (21 March). URL,http://
www.cnn.com/2006/TECH/space/03/20/space.tourism.ap/index.html. [cited
23 March 2006].
[28] Utterback, J. M., “Dominant Designs and the Survival of Firms,” Mastering the
Dynamics of Innovation, Harvard Business School Press, Cambridge, MA, 1994.
[29] Dyson, F., Imagined Worlds. Harvard Univ. Press, Cambridge, MA, 1997.
[30] Legacy Archive for Microwave Background Data Analysis (LAMBDA), NASA
Goddard Space Flight Center (7 Oct.). URL: ,http://lambda.gsfc.nasa.gov/.
[cited 13 March 2006].
[31] Kuhn, T. S., The Structure of Scientific Revolutions, 2nd ed., Univ. Chicago Press,
Chicago, 1970.
[32] Horgan, J., The End of Science, Addison Wesley, Reading, MA, 1996.
[33] Hamming, R., “You and Your Research,” Morris Research & Engineering
Center Lecture, 7 May 1986. URL: , http://www.cs.virginia.edu/robins/
YouAndYourResearch.pdf. [cited 17 Feb. 2006].
[34] Clarke, A. C., Profiles of the Future: An Inquiry into the Limits of the Possible,
Bantam Books, New York, 1972.
[35] Forward, R. L., Indistinguishable from Magic, Baen Books, New York, 1995.
[36] Emme, E. M. (ed.), Science Fiction and Space Futures Past and Present, American
Astronautical Society, San Diego, CA, 1982.
[37] von Puttkamer, J., “Reflections on a Crystal Ball: Science Fact vs. Science Fiction,”
Science Fiction and Space Futures Past and Present, Emme, E. M. (ed.), American
Astronautical Society, San Diego, CA, pp. 137 150.
[38] Star Trek, Gene Roddenberry, NBC, 78 episodes, 8 Sept. 1966 3 June 1969.
[39] Star Wars movies, Lucasfilm Ltd, 1997, 1980, 1983, 1999, 2002 and 2005.
[40] Rodley, E. (ed.), Star Wars Where Science Meets Imagination. National Geo
graphic, Washington, DC, 2005.
[41] Miller, W. C., The Creative Edge: Fostering Innovation Where You Work, Addison
Wesley, Reading, MA, 1987.
[42] Anderson, J. L., “Leaps of the Imagination: Interstellar Flight and the Horizon
Mission Methodology,” Journal of the British Interplanetary Society, Vol. 49,
1996, pp. 15 20.
716 M. G. MILLIS

[43] Millis, M. G., Breakthrough Propulsion Physics Research Program, NASA TM

107381, 1997.
[44] Millis, M. G., and Williamson, G. S., NASA Breakthrough Propulsion Physics Work
shop Proceedings, NASA CP 1999 208694, Cleveland, OH, 12 14 Aug. 1997.
[45] Hartz, J. and Chappell, R., Worlds Apart: How the Distance Between Science
and Journalism Threatens America’s Future, 2nd ed., First Amendment Center,
Nashville TN 1998. URL: ,http://www.firstamendmentcenter.org/about.
aspx?id¼6270. [Accessed May 2005].
[46] Park, R. L., Voodoo Science: The Road from Foolishness to Fraud. Oxford Univer
sity Press, New York, 2000.
[47] Langmuir, I., “Pathological Science, Colloquium at The Knolls Research Labora
tory,” 18 Dec. 1953, transcribed and edited by Hall, R. N; URL: ,http://
www.cs.princeton.edu/ken/Langmuir/langmuir.htm. [cited 27 Feb. 2008].
[48] Bennett, G. L. “Some Observations on Avoiding Pitfalls in Developing Future
Flight Systems,” 33rd AIAA/ASME/SAE/ASEE Joint Propulsion Conference,
AIAA 97 3209, Seattle, WA, July 1997.
[49] Millis, M. G., “Responding to Mechanical Antigravity,” NASA/TM 2006 214390,
2006.
[50] Sagan, C., and Druyan, A., The Demon Haunted World: Science as a Candle in the
Dark, Ballantine Books, New York, 1997.
[51] Baez, J., Crackpot Index, 1998. URL: ,http://math.ucr.edu/home/baez/crackpot.
html. [cited 27 Feb. 2008].
[52] Kruger, J., and David, D., “Unskilled and Unaware of It: How Difficulties in Recog
nizing One’s Own Incompetence Lead to Inflated Self Assessments.” Personality
and Social Psychology, Vol. 77, No. 6, 1999, pp. 121 1134.
[53] (SBIR) evaluations. URL: http://sbir.nasa.gov [cited May 2004].
[54] Hord, R. M., CRC Handbook of Space Technology: Status and Projections, CRC
Press, Boca Raton, FL, 1985.
[55] Townsend, J. A., Hansen, P.A., McClendon, M. W., de Groh, K. K., and Banks,
B. A., “Ground Based Testing of Replacement Thermal Control Materials for the
Hubble Space Telescope,” High Performance Polymers, Vol. 11, 1999, pp. 63 79.
[56] “NASA Research Announcement: Research and Development Regarding ‘Break
through’ Propulsion,” NRA 99 LeRC 1, NASA Lewis Research Center, 9 Nov.
1999.
[57] Tajmar, M., Plesescu, F., Seifert, B., Schnitzer, R., and Vasiljevich, I., “Search for
Frame Dragging Like Signals Close to Spinning Superconductors,” Proceedings of
the Time and Matter 2007 Conference, Bled, Slovenia, World Scientific Press, 2007.
[58] Maclay, J., Hammer, J., George, M., Sanderson, L., and Clark, R., “First Measure
ment of Repulsive Quantum Vacuum Forces,” Joint Propulsion Conference,
AIAA Paper 2001 3359, July 2001.
[59] Roberson, T., “Exploration of Anomalous Gravity Effects by rf Pumped Magnetized
High T Superconducting Oxides,” Joint Propulsion Conference, AIAA Paper 2001
3364, July 2001.
[60] Ringermacher, H., Conradi, M., Browning, C., and Cassenti, B., “Search for Effects
of Electric Potentials on Charged Particle Clocks,” Joint Propulsion Conference,
AIAA Paper 2001 3906, July 2001.
[61] Cramer, J. “Tests of Mach’s Principle with a Mechanical Oscillator,” Joint Propul
sion Conference, AIAA Paper 2001 3908, July 2001.
 PRIORITIZING PIONEERING RESEARCH 717

[62] Mojahedi, M., and Malloy, K., “Superluminal but Causal Wave Propagation,” Joint
Propulsion Conference, AIAA Paper 2001 3909, July 2001.
[63] Fralick, G., and Niedra, J., “Experimental Results of Schlicher’s Thrusting Antenna,”
Joint Propulsion Conference, AIAA Paper 2001 3357, July 2001.
[64] Zampino, E. J., and Millis, M. G., “The Potential Application of Risk Assessment to
the Breakthrough Propulsion Physics Project,” Annual Reliability and Maintainabil
ity Symposium 2003 Proceedings, 2003, pp. 164 169.
[65] Budget of the United States Government, Fiscal Year 2003; Executive Office, Office
of Management & Budget. GPO, Washington, DC, 2002; URL:,http://
www.gpoaccess.gov/usbudget/fy03/pdf/bud27.pdf. [cited 18 April 2006].
 Subject Index

Abraham Lorentz constant, 425 Anderson, John J., 671 672

Abraham Minkowski controversy, Angular momenta, 256
22, 341 342 Anharmonic Casimir oscillator
Rontgen interaction, 355 (ACO), 401
theoretical history, 348 352 Anomalous forces, 231
Abraham’s formulation, 346 347 Anomalous redshifts, 131
Ashkin and Dziedzic, 349 350 Antigravity
Balasz, 348 cosmological, 192 198
Brevik’s analysis, 351 352 exact relativistic antigravity
Gordon’s analysis, 350 propulsion, 189 190
Halpern, 348 gravitomagnetic foundations, 184
Jones and Richards experiment, 349 negative energy induced, 190 192
photon drag experiment, 352 unrelated devices
Shockley, 349 frame dragging, 260
Walker experiment, 350 351 gyroscopic antigravity, 254 259
Absence of vision, characteristics of, 674 oscillation thrusters, 249 254
Absolute reference frame, 135 reaction/momentum wheel,
Accelerated matter, gravitomagnetic 259 260
field and, 245 spinning superconductors, 260
Acceleration increase, Casimir drive and, torque wheel, 259 260
413 415 Antimatter
Acceleration, effects on mission trip time, fusion propulsion and, 77 78
61 63 production of, 75
ACO. See anharmonic Casimir oscillator. Antimatter annihilation
Acoustic inertial confinement fusion propulsion, 42
(AICF), 607 Antimatter catalyzed nuclear
ACT. See Advanced Concepts Team. fission/fusion, 78
ACT. See asymmetrical capacitor Antimatter propulsion, 106, 107
thrusters. Antimatter rocket issues, 102 103
Active gravitational mass, 161, 162 Antimatter rockets
Additive scoring method, 686 light sails vs, 93 94
Advanced Concepts Team (ACT), propulsion system, 85
19 20 Antimatter space power,
nonviable concepts, 19 20 564 565
Advanced nuclear electric propulsion Apollo Lunar Surface Experiments
(NEP), 66 Packages (ALSEPs), 545
Advanced Stirling radioisotope generator Areal density, 55
(ASRG), 555 Ashkin and Dziedzic, 349 350
AFM. See atomic force microscope. ASRG. See advanced Stirling
AICF. See acoustic inertial confinement radioisotope generator.
fusion. Asymmetric capacitors lifters
Alcubierre warp drive, 8 9, 487 489 (Biefeld Brown effect),
Alpha Centauri mission, 86 88 nonviability of, 16
ALSEPs. See Apollo Lunar Surface Asymmetrical capacitor thrusters
Experiments Packages. ablative material, 337
Alzofon’s antigravity propulsion, 221 corona discharge regime, 315

719
720 SUBJECT INDEX

Asymmetrical capacitor thrusters Batteries, 537

(Continued) Beamed energy propulsion, 79
corona wind, 324 325 laser electric, 79
current flow, 335 solar thermal, 79
DC measurements, forces generated, Beamed momentum light sails, 107
303 304 Beamed momentum propulsion, 79 84
dielectrophoretic forces, 324 electromagnetic catapults, 83 84
early development of, 329 330 sails
early history, 293 294 electromagnetic, 82 83
electrogravitic coupling, 295 light, 80 82
electrohydrodynamic forces, 324 solar, 80
electromagnetic manipulation, 295 Beginning of life (BOL) values, 534
experiments, 333 335 Beginning of mission (BOM) values, 534
atmospheric pressure, 333 334 Bell inequality tests, 511
DC measurements, 297 304 Bias drive, 165 167
partial vacuum, 334 Biefeld Brown effect. See asymmetric
results, 297 304 capacitor lifters.
set ups, 295 297 Bonnor and Piper, 201
vacuum chamber, 339 Bonnor and Swaminarayan, 181
force power law, 312, 313 314 Bose Einstein condensate experiment,
Fowler Nordheim, 314, 315 354 355
geometric variations, 310 317 Boyer’s correlation function, 446 451
linear arrangement, 310 BPP. See Breakthrough Propulsion
geometries of, 331 332 Physics (BPP) Project.
grounding issues, 333 Breadboard device Micro Newton thrust
high voltage, 295, 330 balance, 386 388
image charges, 337 Breakthrough Propulsion Physics (BPP)
induced gravitational field, 295 Project, 11 17, 663 692
ion drift, 337 339 goals/grand challenges, 12
thrust mechanism, 324 kickoff workshop, 13
Kelvin forces, 324 research proposal invitations, 692
lifter geometries of, 330 331 review and selection process,
magnetic field measurements, 693 694
317 319 scoring equation, 689 692
corona discharge, 321 322 tasks, 13 14
Fowler Nordheim emission, Breakthrough propulsion studies, 1 2
322 323, 325 Abraham Minkowski debate, 22
Townsend equation, 322 advance concepts, 3, 7
numerical calculations, 336 Sanger Study, 7
Sinusoidal excitation measurements, Breakthrough Propulsion Physics
304 309 Project, 11 17
theories, 294 295 coupling of electromagnetism and
corona wind, 295 gravity, 1
thermodynamic forces, 324 early efforts, 2
voltage gradients, 535 Einstein’s Special Relativity, 1
Atmospheric pressure, asymmetrical ESA General Studies Programme,
capacitor thruster experiments 19 20
and, 333 334 Institute for Advanced Studies at
Atomic force microscope (AFM), 400 Austin, 11
Podkletnov disk, 9 10
Balance response, Micro Newton thrust Project Greenglow, 17 19
balance and, 378 quantum tunneling, 1, 9
Balasz, 348 quantum vacuum, 10 11
 SUBJECT INDEX 721

reality based, 2 3 Cavitation fusion reactor (CFR), 607

rotating superconductors, 21 22 Cerenkov radiation, 464
timeline of, 3 6 CFR. See cavitation fusion reactor.
transient inertia phenomena, 22 Charge clusters, Shoulders, Ken, 642
vacuum fluctuation energy, 1 Charge leakage, superconductor
Vision 21, 7 8 experiments and, 236
warp drive, 1 Charge pooling, superconductor
physics, 8 9 experiments and, 235
wormholes, 8 9 Chemical rocket propulsion, 65 66
Brevik’s analysis, 351 352 high energy density matter, 65
Brito, 362 366 space shuttle main engine, 65
Buoyancy effects, superconductor Chemical space power, 536 537
experiments and, 232 233 batteries, 537
Bussard interstellar ramjet, 59 61, fuel cells, 537
72 73, 151 Christoffel symbols, Riemann
cylindrical geometry magnetic Tensor, 654
confinement fusion reactor, Clarke, Arthur C, three laws, 669 670
60 61 Classical interaction, ZPF and, 445
extraterrestrial resource utilization, Classical oscillator, 451 452
72 73 Classical rocket equations, 43 45,
momentum drag, 60 61 108 110
ram scoop design, 60 61 propellant loss, 46 48, 114 116
Closed cycle gas core nuclear rocket, 560
Calorimetry, 640 Closed timelike curve (CTC), 496
design, 645 646 CMB. See cosmic microwave
Cartan, 658 659 background.
Casimir cavity, 213 214 Coherence entanglement
Casimir drive complementarity, 518 520
applications of, 416 417 Cold fusion, 645 648
testing, 416 calorimetry design, 645 646
unresolved physics issues, 415 417 Conservation Laws, negative matter
vibrating mirror, 405 415 propulsion and, 182 183
Casimir dynamic effect, 397 398 Conservation of momentum and energy,
Casimir effect, 391, 395 397, 481 484, 128 130, 167, 359
570 572, 583 585, inertia, 129
589 590, 641 traversable wormholes, 501
alternative theories, 398 399 warp drives, 501
analytical codes, 660 Constant acceleration rocket, 465 467
fluctuating matter fields, 590 Continuous machine, 590 591
measuring of, 399 402 Control volume analysis, 259
anharmonic oscillator, 401 Cooper pair mass anomaly, 243
atomic force microscope, 400 Corona discharge, 321 322
microelectromechanical systems, regime, 315
400 401 Corona plasmas, 323
moving mirrors, 483 Corona wind, 296, 324 325
plenum, 589 Corum, 360
semiconductor surfaces, 401 402 Cosmic microwave background
traversable wormholes, 483 484 (CMB), 195
Cassini project, 553 554 radiation, 132
Catalyzed nuclear fission, 77 78 Cosmological antigravity, 192 198
Catapults, electromagnetic, 83 84 dark energy, 195 197
Cavitation energy claims, 642 dark/vacuum energy, 198
Potapov device, 642 vacuum energy, 193 195
722 SUBJECT INDEX

Cosmological inflation, 197 198 Dielectric media

Coupled Maxwell Einstein fields, Abraham’s formulation, 346 347
204 207 electromagnetic fields, 345 347
gravitons, 204 207 Minkowski’s formulation, 346
Coupled phenomena, 20 Dielectrophoretic forces, 324
Cruise velocity, effects on mission trip Differential sail, 154 155
time, 61 63 Diode sail, 155 156
CTC. See closed timelike curve. Dipole gravitational field generator,
Current flow, 335 185 187
Curvature of spacetime, 137 Disjunction drives, 162 164
Cyclindrical geometry magnetic active gravitational mass, 162
confinement fusion (MCF) Displacement current, 358
reactor, 60 61 Distances, interstellar flight technology
and, 32
Daedalus fusion rocket, 69, 70 71 Doppler shifts, 464
Damping oscillation motion, 425 426 Drop tests, 258
Abrahm Lorentz constant, 426 Dynamic Casimir effect, 397 398
Damping system, 377 Dynamic friction, 251
Dark energy, 20, 131 132, 195 197, 668 Dynamic space propulsion, 405
anomalous redshifts, 131
cosmic microwave background, 195 Earth to orbit energy, 146 147
Dark matter, 3, 20, 132, 668 Rocket Equation, 147
galaxy rotation, 132 Space Shuttle, 147, 148
gravitational lensing, 132 Einstein’s General Theory of Relativity,
Dark/vacuum energy, 198 184 196
warp drive, 198 antigravity, 184 186
DC measurements, asymmetrical cosmological, 192 198
capacitor thrusters and, exact relativistic antigravity
297 304 propulsion, 189 190
De Matos, 243 computational tools, 651 661
Deep Dirac energy assessment, 14 Field equation, 136
Deep space travel energy, 143 149 negative energy induced,
Earth to orbit energy, 146 147 190 192
General Relativity, 146 Electric coupling, superconductor
gravitational dipole toroid, 146 experiments and, 234
Kinetic Energy Equation, 144 Electric power system (EPS), 531
Krasnikov Tubes, 146 components, 531
levitation energy, 147 149 power management and
Newtonian equations, 146 distribution, 531
Rocket Equation, 144 Electric propulsion, 66
type comparison, 145, 146 advanced nuclear, 66
warp drives, 146 low specific mass, 534
wormholes, 146 radioisotope thermoelectric
Degradable quantum vacuum, 594 597 generator, 66
gravitational squeezing, 594 595 solar, 66
melting, 596 597 Electrogravitational theory, 263 264
redshifting, 595 596 Electrogravitic coupling, 295
vacuum field energy, 597 Electrohydrodynamic (EHD)
zero point fluctuation modes, 597 forces, 324
DeWitt, 236 237 Electromagnetic catapults, 83 84
Diametric drive, 160 162 Electromagnetic coupling,
gravitational mass, 161 superconductor experiments
inertial mass, 161 and, 235
 SUBJECT INDEX 723

Electromagnetic device energy claims, Energy momentum stress tensor, 137

644 645 EPS. See electric power system.
Motionless Electromagnetic Equivalence Principle, 140
Generator, 644 ESA General Studies Programme,
Electromagnetic effects, superconductor Advanced Concepts Team,
experiments and, 234 235 19 20
Electromagnetic field momentum, Escape velocity, 148
363 364 ETRU. See extraterrestrial resource
Electromagnetic fields, dielectric media utilization.
and, 345 347 EV. See electromagnetic vortex.
Electromagnetic inertia manipulation Exact relativistic antigravity propulsion,
(EMIM) propulsion, 189 190
362 366 Exotic matter, 472 484
Brito, 362 366 Extraterrestrial resource utilization
electromagnetic field momentum, (ETRU), 72 73
363 364
electromagnetic momentum Faster than light (FTL) spacetimes
density, 365 closed timelike curve, 496
experiments, 365 366 conservation of momentum, 499 500
mass inertia tensor, 364 General Theory of Relativity and,
Electromagnetic manipulation, 471 501
asymmetrical capacitor motion on causality, 497 499
thrusters and, 295 time machines, 496 499
Electromagnetic momentum traversable wormholes, 484 487
density, 365 warp drives, 487 489
Electromagnetic momentum exchange, Faster than light travel paradoxes, 461
356 366 Feigel hypothesis, 366 367
Slepian’s space ship, 357 360 Fermi energy, 188
Electromagnetic sails, 82 83 Feynman diagrams, 433, 437
MagOrion, 83 external lines, 437
particle beam driven, 83 Field drives, 159 168
Electromagnetic stress tensors, bias, 165 167
360 362 conservation of momentum, 167
Corum, 360 diametric, 160 162
Hartley oscillator, 360 disjunction, 162 164
Heaviside force density, 360 362 gradient potential, 164 165
Electromagnetic vortex (EV) net external force requirements,
phenomenon, 585 587 164, 167
Electrostatic effects, superconductor Fission fragment propulsion, 42, 67 69
experiments and, 235 236 rocket, 6
Electrostatic forces, asymmetrical sail, 68 69
capacitor thrusters and, 337 Fission space power, 556 561
EM torsion theory testing, 15 nuclear electric propulsion, 556
EMIM. See electromagnetic inertia nuclear thermal propulsion, 556
manipulation propulsion. Soviet Union, 560 561
Empirical observations, 685 686 United States, 556 558
End of mission (EOM) values, 534 Fission thermal propulsion, 66 67
Energy conditions, 472 484 nuclear engine for rocket vehicle
violating field, 474 applications, 66
Energy density function, 426 Fluctuating plenum, 588 589
Energy storage systems, 534 Fluctuation matter fields, Casmir effect
high energy density, 534 and, 590
high specific energy, 534 Force beams, Podkletnov, 242
724 SUBJECT INDEX

Force power law, 312, 313 314 Gamma ray bursts (GRB), 495
Forward, dipole gravitational field Gas core nuclear rocket (GCNR), 560
generator, 185 187 GCNR. See gas core nuclear rocket.
Forward, Robert, 2, 7 General purpose heat source radioisotope
Forward’s negative matter propulsion thermoelectric generator
analysis, 181 (GPHS RTG), 551
Forward’s Six Mass Compensator, General Relativity, 146, 184 196, 219,
177 179 230, 394
Fowler Nordheim, 314, 315 Quantum Mechanics vs, 668
Fowler Nordheim emission, General Relativity computational tools,
322 323, 325 651 661
Frame dependent considerations, existing types, 655 660
135 136 Casimir Effect codes, 660
Frame dependent effects, 21 history, 652
Mach’s principle, 21 Macsyma, 652, 655, 657
Frame dragging, 185 187, 243 245, 260 Maple, 659 660
Fraudulent submissions, 675 677 Mathematica, 657 659
Free energy, testing claims of, 639 648 Maxima, 656, 657
cold fusion, 645 648 Reduce, 660
electromagnetic devices, 644 645 propulsion computational conventions,
output energy measurement, 640 652 656
zero point energy devices, 641 644 summary, 656
Free radicals, 533 General Theory of Relativity, 184 186
Friction, 250 251 energy condition, 472 484
dynamic, 251 violating field, 474
static, 251 exotic matter, 472 484
Fuel cells, 537 faster than light spacetimes, 471 501
Fusion, 77 78 wormholes, 471 472
Fusion propulsions, 69 73 negative energy, 474
all onboard types, 69 theorems, 473
antimatter catalyzed fission fragment Generational ships, 94 95
sail, 78 Geometric variations, asymmetrical
Bussard interstellar ramjet, 72 73 capacitor thrusters and,
combinations, 77 78 310 317
antimatter catalyzed nuclear Gertsenshtein effect, 209 210
fission/fusion, 78 Gertsenshtein waves (GW), 209
Daedalus, 69, 70 71 Gliese mission, 88 90
inertial confinement, 69 Gordon’s analysis, 350
magnetic confinement, 69 pseudo momentum, 350
matter antimatter annihilation, 73 76 GPHS RTG. See general purpose heat
muon catalyzed fusion, 78 source radioisotope
Fusion ramjet issues, 105 thermoelectric generator.
Fusion reactions, 562 Gradient effects, superconductor
inertial confinement, 562 experiments and, 235
magnetic confinement, 562 Gradient potential drive, 164 165
Fusion space power, 561 564 Gravitational bremsstrahlung
fusion reactions, 562 radiation, 208
Fusion, sonoluminescence energy Gravitational dipole toroid, 147
harvesting and, 625 Gravitational lensing, 132
Gravitational mass, 161
Galactic hydrogen, 133 active, 161
Galaxy rotation, 132 passive, 161
Galileo spacecraft, 551 552 Gravitational scalar potentials, 167
 SUBJECT INDEX 725

Gravitational squeezing, 594 595 quantum antigravity propulsion,

Gravitational synchrotron radiation, 208 211 221
Gravitational wave rockets, Gravity shielding, Podkletnov, 239 242
coupled Maxwell Einstein fields, Greenglow, Project, 17 19
204 207 Ground state energy, 644
Gertsenshtein effect, 209 210 reduction, 579 582
photon rocket spacetime metric, Grounding issues
201 203 asymmetrical capacitor thrusters, 333
realistic use of, 210 211 superconductor experiments, 235
Gravitational wave transducers, GRtensorII, 658, 659 660
242 243 Gyroscopic antigravity, 254 259
Gravitational waves, 231 angular momenta, 256
high frequency, 231 testing, 257 259
Gravitationally altered launch pad, control volume, 259
140 143 drop tests, 258
analyses of, 143
Gravitationally squeezed electromagnetic Haisch model, 429 431
zero point fluctuations, Haisch, Rueda and Puthoff (HRP), 424
479 481 Halpern, 348
Gravitomagnetic field Hamming, Richard, 668 669
accelerated matter and, 245 Harmonic oscillator, 427
rotating superconductors and, 21 22 Hartley oscillator, 360
Gravitomagnetic foundations, 184 Heaviside force density, 360 362
Fermi energy, 188 HEDM. See high energy density matter.
Forward’s dipole gravitational field Heim’s quantum theory, 218 221
generator, 185 187 General Relativity, 219
Lense Thirring effect, 185 gravitophoton interaction, 219 220
Maxwellian electrodynamics, 185 Heisenberg uncertainty principle, 574
reality of, 187 188 Helicopter analogy, 148
Graviton production, 204 207 HFGW. See high frequency gravitational
particle accelerators, 207 209 waves.
Gravitophoton interaction Hidden momentum, 375
field, 219 220 High energy density, 534
Gravity High energy density matter (HEDM), 65
electromagnetism and, 1 propellant, 97
modification of, 231 High specific power, 534
Gravity control High frequency gravitational wave
Alzofon’s antigravity propulsion, 221 producer, 203
Einstein’s General Theory of High frequency gravitational waves
Relativity, 184 198 (HFGW), 231
field drives, 159 168 Hooper antigravity coils, nonviability
gravitational wave rockets, 201 211 of, 16
Heim’s quantum theory, 218 221 Horizon Mission Methodology, 697
Levi Civita effect, 198 200 Anderson, John L., 671 672
negating, 177 HSMs. See hydrogen switchable mirrors.
negative matter propulsion, 180 184 Human generational issues, 94 95
Newtonian Levitation Energy Hydrogen switchable mirrors
Estimate, 179 (HSMs), 584
Newtonian physics, 176 184
nonretarded quantum interatomic ICF. See inertial confinement fusion.
dispersion force, 215 218 ICP. See inertial confinement propulsion.
orbital stability, 177 179 Image charges, 337
Six mass Compensator, 177 179 electrostatic forces and, 337
726 SUBJECT INDEX

Imaging, multi bubble sonoluminescence critical nonpropulsion technologies,

(MBSL) and, 611 615 99 100
Impossible missions, concept of, 36 40 dust grains, 106
Impulse representation, 252 engineering vs physical feasibility of,
Impulse treatment, 148 101 105
Indexing convention, contracting evaluation, 40 53
Riemann to Ricci, 654 655 infrastructure, 42
Indigenous reaction mass, 130 137 technology, 42
cosmic microwave background velocity change capability, 40 41
radiation, 132 fundamentals, 43 65
dark energy, 131 132 Bussard interstellar fusion ramjet
dark matter, 132 mission performance, 59 61
galactic hydrogen, 133 mission trip time, 61 63
quantum vacuum fluctuations, 132 133 relativistic light sail equations, 53 59
spacetime, 134 137 rocket equation, 43 53
virtual particle pairs, 133 fusion ramjet issues, 105
Indigenous space phenomena, 131 human and cultural effects, 94
Induced charges, superconductor impossible missions, 36 40
experiments and, 235 interstellar dust, 63 65
Induced gravitational field, 295. See also laser light sails, 106
gravity control. issues, 103 104
Induction sail, 155 light sails vs antimatter
Inertia, 129 rockets, 93 94
Inertia control, 156 159, 404 405 minimum communications system
Inertia modified rocket, 156 power, 100
Inertial confinement fusion (ICF), 562 mission requirements, 33 35
Inertial confinement propulsion (ICP), 69 robotics, 34
Inertial frame, 134 Vision Mission, 33
Inertial mass, 161 multiple generations, 94 95
Inertial mass, stochastic electrodynamics possible missions, 84 95
and, 432 452 Alpha Centauri, 86 88
Inertial space drive modifications, 156 159 antimatter rocket propulsion, 85
inertia modified rocket, 156 Gliese, 876, 88 90
oscillatory inertia thruster, 157 159 light sail relativistic effects, 92
In FEEP (indium field emission electric single stage light sail, 85
propulsion) thrusters, 376 Vision Mission, 90 92
Infinite specific impulse perspective, 138 reliability vs reparability, 99
Input current/rotational speed equation, rocket based propulsion, 65 78
269, 271 societal investment in, 100
Instant messaging paradoxes, 461 463 specific energy density, 96 98
Institute for Advanced Studies specific impulse, 36 40, 95 96
at Austin, 11 specific power, 96 98
Internal drive, 249 stage dry mass, 36 40
Interstellar dust grains, 106 wet mass, 36 40
Interstellar dust particles, 63 65 Interstellar fusion ramjet mission
Interstellar flight technology performance, 59 61
antimatter, 106, 107 Interstellar jet propulsions, 151 152
antimatter rocket issues, 102 103 Bussard ramjet, 151
beamed energy, 79 Interstellar travel, barriers to, 12
beamed momentum light sails, 107 Ion drift, 324, 337 339
beamed momentum propulsion, 79 84 Ion effects, superconductor experiments
challenge of, 32 42 and, 235
distances, 32 Iterated research, 684 685
 SUBJECT INDEX 727

Jones and Richards experiment, 349 Lorentz force law, 353 354, 428 429
Journal of the British Interplanetary Lorentz Transformation, 456 459
Society, 2 Loudon, 354
Low specific mass, 534
Kelvin forces, 324 LWRHU. See lightweight radioisotope
Kinetic Energy Equation, 144 heater units.
Kowitt, 239
Krasnikov Tubes, 146 Mach’s principle, 21, 135, 167, 373
absolute reference frame, 135
Lamb, Willis, 393, 394 frame dependent considerations,
Langrangian, 352, 436 135 136, 515 516
Laser beam power, 54 55 Mach 5C Micro Newton thrust balance,
Laser electric propulsion (LEP), 79 381 382
Laser light sails, 106 Mach 6C Micro Newton thrust balance,
issues, 103 104 382 385
Lense Thirring effect (rotational frame Mach 6CP Micro Newton thrust balance,
dragging effect), 185 187, 385 386
243 245, 260 Macsyma, 652, 655, 657
Leonhardt’s analysis, 355 lisp, 657
Abraham Minkowski tensor calculus package, 652
controversy Rontgen Magnetic confinement fusion
interaction, 355 (MCF), 562
LEP. See laser electric propulsion. reactor, 60 61
Levi Civita effect, 198 200 Magnetic confinement propulsion
quantum electrodynamic vacuum (MCP), 69
breakdown field, 200 Magnetic coupling, superconductor
Levitation energy, 147 149 experiments and, 234
damped oscillation, 148 Magnetic fields, 475 576
escape velocity, 148 measurements, asymmetrical capacitor
helicopter analogy, 148 thrusters and, 317 319
impulse, 148 changes in, 320
normal accelerated motion, 148 Magnetic Orion (MagOrion), 83
thermodynamics, 148 Mallove, Eugene, 2
Li, 237, 239 Mansuripur, 353 354
Lifter geometries, 330 331 Lorentz force law, 353 354
Lifters. See asymmetrical capacitor Minkowski Abraham controversy,
thrusters. 353 354
Light sails, 80 82 Maple, 659 660
antimatter rockets vs, 93 94 GRTensorII, 659 660
diameter, 54 tensor package, 659
hypothetical sail drives, 152 156 Mars Science Laboratory, 555 556
mass, 55 Multi Mission Radioisotope
relativistic effects, 92 Thermoelectric Generator, 555
Lightspeed limit, 134 Mass inertia tensor, 364
Lightweight radioisotope heater units Mass renormalization, 434
(LWRHUs), 553 Mass, characteristics of, 161
Lincoln satellites, 545 546 Mathematica, 657 659
Linear arrangement, asymmetrical Cartan, 658 659
capacitor thrusters and, 310 GRtensorII, 658
Liquid core reactor, 560 MathTensor, 657 658
Liquid effects, superconductor Ricci, 658
experiments and, 233 234 Tensorial, 658 659
Lisp, 657 MathTensor, 657 658
728 SUBJECT INDEX

Matter antimatter annihilation Momentum, 128 130

propulsion, 73 76 conservation of, 167, 499 500
antimatter, 75 Momentum domain entanglement,
vehicle sizing, 75 76 514 518
Matter antimatter annihilation reaction, Motionless Electromagnetic Generator
48, 116, 117, 119 (MEG), 644
Maxima, 656, 657 Moving mirrors, 483
Maxwell’s equation, 342 345 Multi bubble sonoluminescence
Maxwellian electrodynamics, 185 (MBSL), 611
MBSL. See multi bubble imaging, 611 615
sonoluminescence. testing variables, 632
MCF. See magnetic confinement fusion. Multi Mission Radioisotope
MCP. See magnetic confinement Thermoelectric Generator
propulsion. (MMRTG), 555
MEG. See motionless electromagnetic Multiplicative scoring method, 686,
generator. 688 689
Melting the vacuum, 596 597 Muon catalyzed fusion, 78
MEMS. See microelectromechanical
systems. NASA, sonoluminescence and, 610 617
Metric determinant sign Negative energy, 474
convention, 654 drives, 404 405
Metric sign convention, 653 654 generation of, 475 484
Michelson Morely experiment, 463 Casimir effect, 481 484
Microelectromechanical gravitationally squeezed
systems (MEMS), 21, 392, electromagnetic zero point
400 401 fluctuations, 479 481
Micro Newton thrust balance magnetic fields, 475 476
balance response, 378 squeezed quantum vacuum,
damping system, 377 476 479
experiment setup, 378 380 static radial electric, 475 476
2 MHz breadboard device, induced antigravity, 190 192
386 388 observation of, 495 496
Mach 5C, 381 382 gamma ray bursts, 495
Mach 6C, 382 385 quantum inequalities, 492 495
Mach 6CP, 385 386 spatial distributions, 492 495
hidden momentum, 375 traversable wormholes, requirements
In FEEP thrusters, 376 of, 489 490
sensors, 376 377 warp drive, requirements of, 490 492
theoretical considerations, 374 376 Negative matter propulsion, 180 184
Woodward effect, 373 388 actual existence of, 183 184
Minkowski’s formulation, 346 analysis of, 180
Minkowski Abraham controversy. Bonnor and Swaminarayan, 181
See Abraham Minkowski Forward, 181
controversy. Conservation Laws, 182 183
Mission trip time, 61 63 diametric drive, 160 162
cruise velocity, 61 63 Principle of Causality, 184
effects of acceleration, 61 63 Second Law of Thermodynamics, 184
MKS system, 653 Totalitarian Principle, 184
MMRTG. See Multi Mission Nelson, 352
Radioisotope Thermoelectric Lagrangian approach, 352
Generator. NEP. See nuclear electric propulsion.
Molecular effects, superconductor NERVA. See nuclear engine for rocket
experiments and, 235 vehicle applications.
 SUBJECT INDEX 729

Net external force requirements, gradient Particle accelerators, 207 209

potential drive and, 164, 167 Particle beam driven electromagnetic
New Horizons project, 554 sails, 83
Newton’s gravitational constant, Particle bed reactor (PBR), 560
166, 167 Passive gravitational mass, 161
Newtonian Levitation Energy PBR. See particle bed reactor.
Estimate, 179 PDF. See probability distribution
Newtonian physics, 128, 146, 176 184 function.
Nieminen, R., 237 239 Photon band gaps, 355 356
Nimbus 3 meterological satellite, 544 Photon drag experiment, 352
Nonlinear quantum mechanics, Photon implications, 345 347
524 525 dielectric media, 345 347
Nonlocal communication, 520 522 Photon momentum
Nonlocal signaling, 520 522 Abraham Minkowski controversy,
Non nuclear space power, 535 537 341 342
chemical, 536 537 theoretical history, 348 352
photovoltaic, 535 536 Abraham’s formulation, 346 347
Nonretarded quantum interatomic electromagnetic inertia manipulation
dispersion force, 215 218 propulsion, 362 366
challenge to, 218 electromagnetic momentum exchange,
Pinto, 215 218 356 366
Nonrigorous submissions, 675 Slepian’s space ship, 357 360
responding to, 678 680 electromagnetic stress tensors,
Normal accelerated motion, 148 360 362
NTP. See nuclear thermal propulsion. Feigel hypothesis, 366 367
Nuclear electric propulsion (NEP), Maxwell’s equation, 342 345
66, 556 Minkowski’s formulation, 346
Nuclear engine for rocket vehicle Nelson, 352
applications (NERVA), 66 Lagrangian approach, 352
Nuclear space power, 537 565 photon band gaps, 355 356
radioisotopes, 537 556 theoretical history
Nuclear thermal propulsion (NTP), 556 Bose Einstein condensate
experiment, 354 355
Onboard fusion propellants, 69 Leonhardt’s analysis, 355
Orbital stability, gravity control and, Loudon, 354
177 179 Mansuripur, 353 354
Oscillation thrusters, 249 254 Photon rocket, 467 468
friction, 250 251 spacetime metric, 201 203
impulse representation, 252 Bonnor and Piper, 201
internal drive, 249 high frequency gravitational wave
nonviability of, 16 producer, 203
single plane motion, 260 Photovoltaic space power, 535 536
slip stick drive, 249 solar arrays, 536
sticktion drive, 249, 250 Pinto, 215 218
Oscillatory inertia thruster, 157 159 Pioneer projects, 547 548
Output energy measurement, 650 Piper and Bonner, 201
calorimetry, 640 Planck mass, 429
Platinum films, 619
Palladium chromium films, 619 623 Plenum, 588
resistance temperature detector, 619 Casimir effect and, 589
Paradigm shifts, research and, fluctuating, 588 589
667 668 PMAD. See power management and
Paradoxes, 523 524 distribution.
730 SUBJECT INDEX

Podkeltnov gravity shield, 239 242 Feynman diagrams, 433

nonviability of, 16 less second quantized fields, 594
Podkletnov disk, 9 10 quantum field theory, 433
testing of, 14 15 theory, 574 575
Podkletnov, Eugene, 237 239 Heisenberg uncertainty
Podkletnov, force beams and, 242 principle, 574
Polarization correlation condition, vacuum, 588
510 511 breakdown field, 200
Pole vaulter paradox, 459 460 Casimir effect, 589 590
Polyvinyltoluene based fluctuation effects, neoclassical
scintillator, 626 theories, 592
Potapov device, 642 plenum, 588
Power management and distribution fluctuating, 588 589
(PMAD), 531 Quantum entanglement, 509 525
Poynting vector, 432, 446 551 Bell inequality tests, 511
Principle of Causality, 184, 496, 522 coherence entanglement
Probability distribution function complementarity, 518 520
(PDF), 426 definition of, 509
Project Greenglow, 17 19 early experiments, 510 511
Project metrics of performance, 697 698 glossary, 526 528
Project risk management, 698 momentum domain entanglement,
Propellant loss, 46 51 514 518
classical rocket equations and, nonlinear quantum mechanics, 524 525
114 116 nonlocal communication, 520 522
relativistic rocket equations and, nonlocal signaling, 520 522
116 122 paradoxes, 523 524
Propulsion computational conventions, polarization correlation condition,
652 656 510 511
Christoffel symbols, Riemann Tensor, quantum nonlocality, 509 510
654 quantum no signal theorems,
indexing, 654 655 512 513
system of units, 653 retrocausal nonlocal communication,
metric determinant sign convention, 522
654 superluminal communication, 522
metric sign convention, 653 654 Quantum field theory (QFT), 433, 651
MKS system, 653 Lagrangian, 436
Pseudo momentum, 350 looping of, 435
Publishing, research findings and, 686 mass renormalization, 434
Pulsed propulsion, 66 67 Unruh Davies temperature, 437
vacuum fluctuation, 434
QED. See quantum electrodynamics. Quantum inequalities (QI), negative
QFT. See quantum field theory. energy and, 492 495
QI. See quantum inequalities. Quantum mechanics, 393
Quantum antigravity propulsion, General Relativity vs, 668
211 221 Lamb, Willis, 393, 394
quantum vacuum zero point Quantum nonlocality
fluctuation force, 213 215 definition of, 509 510
Quantum chromodynamic, 569 early experiments, 510 511
Quantum electrodynamics 391, 392, Special Relativity vs, 513 514
433 438 Quantum no signal theorems, 512 513
alternatives to, 592 594 Quantum tunneling, 1, 9
stochastic electrodynamics, 437, testing, 15
438 441, 593 594 nonviability of, 16
 SUBJECT INDEX 731

Quantum vacuum used by United States, 539

Casimir forces, 391, 392, 395 397 advanced Stirling radioisotope
alternative theories, 398 399 generator, 555
dynamic effect, 397 398 Apollo Lunar Surface Experiments
measuring, 399 402 Packages, 545
degradable, 594 Cassini project, 553 554
effects, 21 Galileo spacecraft, 551 552
microelectromechanical general purpose heat source
structures, 21 radioisotope thermoelectric
extracting energy from, 569 598 generator, 551
Casimir force, 570 572, 641 lightweight radioisotope heater
degradability, 591 592 units, 553
early concepts, 570 574 Lincoln satellites, 545 546
experiments, 577 587, 642 Mars Science Laboratory,
electromagnetic vortex 555 556
phenomenon, 585 587, 642 New Horizons project, 554
ground state energy reduction, Nimbus 3 meterological satellite,
579 582, 644 544
tunable Casimir effect, 583 585 Pioneer projects, 547 548
ZPF voltage fluctuations, transit satellites, 540 544
577 579 Ulysses project, 553
quantum chromodynamic, 569 Viking space probes, 548 550
quantum electrodynamics Radioisotope thermoelectric generator
vacuum, 588 (RTG), 66, 537 538, 629
type I transient machine, 590 591 Ram scoop design, momentum drag,
type II continuous machine, 60 61
590 591 Reaction/momentum wheel, 259 260
zero point field, 570 Redshifting vacuum, 595 596
zero point field energy, 574 576 Reduce, 660
fluctuations, 132 133 Reflexive demissals, 674
General Relativity, 394 Relativistic effects on sail acceleration,
history, 393 394 57 59
quantum mechanics, 393 Relativistic light sail equations, 53 59
microelectromechanical systems, 392 areal density, 55
quantum electrodynamics, 391, 392 laser beam power, 54 55
space propulsion, 402 405 light sail diameter, 54
dynamic systems, 405 light sail mass, 55
negative energy drives, 404 405 relativistic effects, 57 59
sails, 402 404 thermally limited light sail
vacuum energy density, 404 405 acceleration, 55 57
studies, 10 11 Relativistic particle half lives, 463 464
zero point energy, 10 11 Relativistic rockets, 465 468
theoretical calculation limitations, 399 constant acceleration, 465 467
vibrating mirror Casimir drive, equation, 45 46, 110 113
405 415 propellant loss, 48 51, 116 122
zero point field energy, 394 395 matter antimatter annihilation
zero point fluctuation force, 213 215 reaction, 48
Casimir cavity, 213 214 Relativistic transformation, uniformly
accelerated frame, 442 445
Radiation detection, 625 628 Research, prioritizing of, 663 713
Radioisotope power absence of vision, 674
sources, 537 556 Breakthrough Propulsion Physics
thermoelectric generator, 537 538 project, 663, 689 694
732 SUBJECT INDEX

Research, prioritizing of (Continued) fission thermal, 66 67

criteria development, 686 689 fusion, 69 73
additive scoring method, 686 pulsed, 66 67
multiplicative scoring method, 686, upper performance limits, 65
688 689 Rocket equations, 43 53, 108 122,
participants, 687 144, 147,
scholastic grading standards, 689 classical, 43 45, 108 110
weighing of, 687 688 comparison of, 51 53
fraudulent work, 675 677 relativistic, 45 46, 110 113
historical perspectives, 665 673 Rocket inertia, modification of,
Arthur C. Clarke, 669 670 138 143
Horizon Mission Methodology, Rotating superconductors, gravito
671 672 magnetic fields in, 21 22
lessons learned, 672 673 Rotational frame dragging effect, 185,
paradigm shifts, 667 668 243 245
researchers, 668 669 Rotational speed/input current equation,
science fiction, 670 671 269, 271
technology revolutions, 665 667 RTD. See resistance temperature detector.
tool driven revolutions, 667 RTG. See radioisotope thermoelectric
Horizon Mission Methodology, 697 generator.
operating principles, 681 694
diversified portfolio, 685 Sails
empirical emphasis, 685 686 acceleration, relativistic effects on,
immediate research steps, 682 683 57 59
impartial reviews, 685 beamed momentum light, 107
iterated, 684 685 drives, 152 156
measured progress, 683 differential, 154 155
publishing, 686 diode, 155 156
reliable knowledge, 681 682 induction, 155
Scientific Method, 683 684 electromagnetic, 82 83
performance project metrics, 697 698 fission fragment propulsion for, 68 69
project risk management, 698 issues in use of, 103 104
Richard Hamming, 668 669 laser light, 106
sample procedure, 701 714 light, 80
task progress, 695 697 propulsion, quantum vacuum and,
vision without rigor, 675 680 402 404
Resistance temperature detector (RTD), solar, 80
619 Sanger Study, 7
Rest mass, ZPF and, 445 SBSL. See single bubble
Retrocausal nonlocal communication, 522 sonoluminescence.
Reviews, impartially of, 685 Schlicher Thruster testing, 14
Ricci, 658 Scholastic grading standards, 689
Ricci, Riemann from, 654 655 Science fiction, 670 671
Riemann Tensor, 654 Scientific Methods, 683 684
Riemann, Ricci to, 654 655 Scintillator, 626
Rigor S curve evolution, 665 666
importance of, 675 680 Second Law of Thermodynamics, 184
nonrigorous submissions, 675 67 Second quantized electrondynamics, 594
Robotic interstellar missions, 34 SED. See stochastic electrodynamics.
Rocket based propulsion, 65 78 Seismic effects, superconductor
chemical, 65 66 experiments and, 233
electric, 66 Semiconductor surfaces, Casimir forces
fission fragment, 67 69 and, 401 402
 SUBJECT INDEX 733

Sensors, 376 377 palladium chromium films, 619 623

SEP. See solar electric propulsion. platinum films, 619
Shockley, 349 multi bubble, 611
simplest case theory, 349 NASA, 610 617
Shoulders, Ken, 642 apparatus, 610 611
SI, 653 single bubble, 611
Simplest case theory, 349 solvents, 616 617, 618
Single plane motion, 260 Soviet Union fission space power,
Single bubble sonoluminescence 560 561
(SBSL), 611 Space drives
testing variables, 632 conservation of momentum, 128 130
Single stage light sail mission, 85 deep space travel energy, 143 149
Single stage to geosynchronous Earth Equivalence Principle, 140
orbit (SSTO GEO), 97 gravitationally altered launch pad,
Sinusoidal excitation measurements, 140 143
304 309 hypothetical mechanisms, 149 168
Sinusoidal high voltage excitation, field drives, 159 168
corona plasmas, 323 inertial modifications, 156 159
Six mass Compensator, Forward, 177 179 interstellar jet propulsions,
Slepian’s electromagnetic space ship, 151 152
357 360 listing of, 150
conservation of momentum and sail drives, 152 156
energy, 359 indigenous reaction mass, 130 137
displacement current, 358 infinite specific impulse perspective,
Slip stick drive, 249 138
Society’s investment, interstellar flight net external force requirements,
technology and, 100 164, 167
Solar arrays, 536 Newtonian physics, 128
Solar electric propulsion (SEP), 66, 533 physics of, 168
Solar sails. See light sails. practicality of, 168
Solar thermal propulsion (STP), 79 rocket inertia, 138 143
Solvents, sonoluminescence and, strategy, 14
616 617, 618 sustainability of, 168
Sonofusion, 606 610, 617 633, 643 Space power, 531 565
Roger Stringham, 643 antimatter, 564 565
Sonoluminescence, 605 633 electric power system, 531
approaches, 607 609 electric propulsion, 534
acoustic inertial confinement energy storage systems, 534
fusion, 607 fission, 556 561
applications, 609 610 fusion, 561 564
cavitation fusion reactor, 607 high power density, 534
Suslick, Ken, 608 609 high specific power, 534
Yaleyarkhan, Rusi, 608 non nuclear, 535 537
energy harvesting, 625 630 nuclear, 537 565
concepts, 629 639 performance characteristics, 534
fusion, 625 solar electric propulsion, 533
polyvinyltoluene based scintilator, specific impulse, 532 533
626 values at
radiation detection, 625 628 beginning of life, 534
radioisotope thermoelectric beginning of mission, 534
generators, 629 end of mission, 534
high temperature generation, 617, Space shuttle main engine (SSME), 65
619 625 Space Shuttle, energy used, 147, 148
734 SUBJECT INDEX

Spaceflight, relativistic limits of, 455 469 harmonic oscillator, 427

photon, 467 468 inertial mass, 432 452
rockets, 465 468 Lorentz force, 428 429
Special Theory of Relativity, 455 459 modeling, 593 594
Spacetime, 134 137 Planck mass, 429
inertial frame, 134 probability distribution function, 426
lightspeed limit, 134 quantum electrodynamics (QED) vs,
Mach’s principle, 135 433 438, 437, 438 441
Young’s modulus, 136 semi quantized, 432
Spatial distributions, negative energy theory, 424, 574
and, 492 495 zero point fields, 424
Spatial terms, 653 654 ZPF Poynting vector, 432
Special Relativity, quantum nonlocality STP. See solar thermal propulsion.
vs, 513 514 Stringham, Roger, 643
Special Theory of Relativity Superconductors
experiments re, 463 465 anomalous forces, 231
Cerenkov radiation, 464 experiment considerations, 231 236
Doppler shifts, 464 buoyancy effects, 232 233
Michelson Morley, 463 charge leakage, 236
relativistic particle half lives, charge pooling, 235
463 464 electric coupling, 234
Tachyon searches, 464 465 electromagnetic coupling, 235
paradoxes, 459 463 electromagnetic effects, 234 236
faster than light travel, 461 gradient effects, 235
instant messaging, 461 463 grounding issues, 235
pole vaulter, 459 460 induced charges, 235
twin, 460 ion effects, 235
principles, 456 liquid effects, 233 234
Lorentz Transformation, 456 459 magnetic coupling, 234
postulates, 456 459 molecular effects, 235
Specific energy density, 96 98 seismic/vibration effects, 233
Specific impulse, 36 40, 532 533 temporal effects, 234
free radicals, 533 thermal effects, 232
metastable, 533 vacuum effects, 233
technology, 95 96 General Relativity, 230
Specific power, 96 98 gravitational
Spinning superconductors, 260 force, 231
Squeezed quantum vacuum 476 479 waves, 231
SSME. See space shuttle main engine. gravitomagnetic field and accelerated
SSTO GEO. See single stage to matter, 245
geosynchronous Earth orbit. gravity shielding, Podkletnov,
Stage dry mass, 36 40 239 242
Stargate, 471 history, 236 245
Static Casimir effect, 407 409 DeWitt, 236 238
Static friction, 251 force beams, Podkletnov, 242
Static radial electric, 475 476 gravitational wave transducers,
Sticktion drive, 249, 250 242 243
Stochastic electrodynamics (SED) Kowitt, 239
background, 425 432 Li and Torr, 237, 239
damping oscillation motion, 425 426 Podkletnov and Nieminen,
disadvantages of, 442 237 239
energy density function, 426 Tajmar experiments, 243 245
Haisch model, 429 431 modifying of gravity, 231
 SUBJECT INDEX 735

spinning, 260 United States fission space power,

yttrium barium copper oxide 556 558
ceramics, 229 rockets, 559
Superluminal communication, 522 closed cycle gas core nuclear
Suslick, Ken, 608 609 rocket, 560
Swaminarayan and Bannor, 181 gas core nuclear rocket, 560
liquid core reactor, 560
Tachyon searches, 464 465 particle bed reactor, 560
Tajmar experiments, 185, 243 245 Unruh Davies, 451 452
Cooper pair mass anomaly, 243 temperature, 437
De Matos, 243
Task progress, 695 697 Vacuum chamber, 339
Technology revolutions Vacuum effects, superconductor
institutional patterns, 665 667 experiments and, 233
S curve evolution, 665 666 Vacuum energy degradability, 591 592
Temporal effects, superconductor Vacuum energy density propulsion,
experiments and, 234 404 405
Temporal terms, 653 654 inertia control, 404 405
Tensor calculus package, 652 Vacuum energy, 193 195
Tensorial, 658 659 Vacuum field energy, 597
Thermal effects, superconductor Vacuum fluctuation, 434
experiments and, 232 energy, 1
Thermally limited light sail Vacuums, asymmetrical capacitor
acceleration, 55 57 thruster experiments and, 334
Thermodynamic forces, 324 Van de Graaf generator, 271, 273
Thermodynamics, 148 Velocity change capability, 40 41
Thrust balance, 376 378 Vibrating mirror Casimir
Time machines, 496 499 drive, 405 415
Tool driven revolutions, 667 acceleration increase, 413 415
Torque wheel, 259 260 modeling, 409 413
Torr, 237, 239 static, 407 409
Totalitarian Principle, 184 unresolved physics issues, 415 417
Townsend equation, 322 Vibration effects, superconductor
Transient inertia phenomena. See experiments and, 233
Woodward effect. Viking space probes, 548 550
Transit satellites, 540 544 Virtual particle pairs, 133
TRIAD, 542 544 Vision Mission, 33, 90 92
Traversable wormholes, 472, 483 487 Vision 21, 7 8
conservation of momentum and, 499 500 Voltage gradients, 535
construction of, 494
physical requirements, 485 486 Walker experiment, 350 351
requirements of, 489 490 Warp drives, 1, 22 23, 146, 198, 472,
TRIAD satellites, 542 544 487 489
Tunable Casimir effect, 583 585 Alcubierre, 487 489
hydrogen switchable mirrors, 584 conservation of momentum and, 499
Twin paradoxes, 460 physics, 8 9
Type I transient machine, 590 591 Alcubierre, 8 9
Type II continuous machine, 590 591 requirements of, 490 492
York extrinsic time, 487
Ulysses project, 553 Wet mass, 36 40
Uncertainty Principle, 132 Woodward effect, 22, 373 388
Uniformly accelerated frame, relativistic Mach’s principle, 373
transformation, 442 445 oscillatory inertia thruster, 157 158
736 SUBJECT INDEX

Woodward’s transient inertia, Zero point energy (ZPE) devices,

testing of, 15, 378 388 a641 644
Wormholes, 8 9, 146, 471 472 Casimir force, 570 572, 641
stargate, 471 cavitation, 642
traversable, 472 charge clusters, 585 587, 642
warp drive, 472 ground state energy, 579 582, 644
sonofusion, 643
Yaleyarkhan, Rusi, 608 Zero point energy (ZPE) history, 10 11
Yamishita electrogravitational patent, Zero point fields (ZPF), 424, 570
263 290 classical interaction, 445
experiment background, 264 267, Haisch, Rueda and Puthoff, 424,
278 287 425 444
input current/rotational speed Poynting vector, 432
equation, 269, 271 rest mass, 445
Van de Graaf generator, 271 voltage fluctuations, 577 579
experiment conclusions, 290 291 Zero point fields energy, 394 395,
experiment results, 273 278, 574 576
287 290 quantum electrodynamics theory,
theoretical background, 263 264 574 575
YBCO. See yttrium barium copper stochastic electrodynamics theory, 574
oxide ceramics. Zero point fluctuation
York extrinsic time, 487 modes, 597
Young’s modulus analogy to, ZPE (ZPF energy), 570
136 137 ZPE. See zero point energy.
Yttrium barium copper oxide ceramics ZPF. See zero point field.
(YBCO), 229 ZPF energy. See ZPE.
 Author Index

Bennett, G. L. . . . . . . . . . . . . 531 Lawrence, T. J. . . . .. . . . . . . . 263

Buldrini, N. . . . . . . . . . . . . . . 373 Little, S. R. . . . . . .. . . . . . . . 639
Cambier, J.-L. . . . . . . . . . . . . . 425 Maccone, C. . . . . . .. . . . . . . . 651
Canning, F. X. . . . . . . . . 329, 341 Maclay, G. J. . . . . .. . . . . . . . 391
Cramer, J. G. . . . . . . . . . . . . . 511 Miller, P. B. . . . . . .. . . . . . . . 293
Cassenti, B. N. . . . . . . . . . . . . 457 Miller, W. M. . . . . .. . . . . . . . 293
Davis, E. W. . . . . . . 175, 473, 569 Millis, M. G. . . . . .. 127, 249, 663
Drummond, T. J. . . . . . . . . . . . 293 Puthoff, H. E. . . . . .. . . . . . . . 569
Fralick, G. C. . . . . . . . . . . . . . 605 Siegenthaler, K. E. .. . . . . . . . 263
Frisbee, R. H. . . . . . . . . . . . . . . 31 Tajmar, M. . . . . . . .. . . . . . . . 373
Gilster, P. A. . . . . . . . . . . . . . . . . 1 Wrbanek, J. D. . . . .. . . . . . . . 605
Hall, N. R. . . . . . . . . . . . . . . . 605 Wrbanek, S. Y. . . . .. . . . . . . . 605
Hathaway, G. D. . . . . . . . . . . . 229

737
 Supporting Materials

Many of the topics introduced in this book are discussed in more detail in other
AIAA publications. For a complete listing of titles in the Progress in Astronautics
and Aeronautics Series, as well as other AIAA publications, please visit
http://www.aiaa.org.

739