Undergraduate Texts in Physics

Ray LaPierre

Getting Started in
Quantum Optics
Undergraduate Texts in Physics

Series Editors
Kurt H. Becker, NYU Polytechnic School of Engineering, Brooklyn, NY, USA
Jean-Marc Di Meglio, Matière et Systèmes Complexes, Université Paris Diderot,
Bâtiment Condorcet, Paris, France
Sadri D. Hassani, University of Illinois at Urbana-Champaign, Urbana, IL, USA
Morten Hjorth-Jensen, University of Oslo, Oslo, Norway
Michael Inglis, Patchogue, NY, USA
Bill Munro, NTT Basic Research Laboratories, Optical Science Laboratories,
Atsugi, Kanagawa, Japan
Susan Scott, Australian National University, Acton, ACT, Australia
Martin Stutzmann, Walter Schottky Institute, Technical University of Munich,
Garching, Bayern, Germany
Undergraduate Texts in Physics (UTP) publishes authoritative texts covering topics
encountered in a physics undergraduate syllabus. Each title in the series is suitable as
an adopted text for undergraduate courses, typically containing practice problems,
worked examples, chapter summaries, and suggestions for further reading. UTP
titles should provide an exceptionally clear and concise treatment of a subject at
undergraduate level, usually based on a successful lecture course. Core and elective
subjects are considered for inclusion in UTP.
UTP books will be ideal candidates for course adoption, providing lecturers with
a firm basis for development of lecture series, and students with an essential
reference for their studies and beyond.
Ray LaPierre

Getting Started in Quantum

Optics
Ray LaPierre
Hamilton, ON, Canada

ISSN 2510-411X ISSN 2510-4128 (electronic)

Undergraduate Texts in Physics
ISBN 978-3-031-12434-1 ISBN 978-3-031-12432-7 (eBook)
https://doi.org/10.1007/978-3-031-12432-7

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland
AG 2022
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of
illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by
similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

Getting Started in Quantum Optics was born from lecture notes developed for an
introductory course in quantum optics offered to undergraduate students in the
Department of Engineering Physics at McMaster University (Hamilton, Ontario,
Canada). The book was written for students who have completed an introductory
course on quantum mechanics and electromagnetism, but otherwise have no back-
ground in quantum optics. Getting Started in Quantum Optics is mostly intended as a
self-contained introduction to the theory of quantum optics with the beginner
in mind.
This book covers canonical quantization, the quantum harmonic oscillator, vac-
uum fluctuations, Fock states, the single photon state, the quantum optics treatment
of the beam splitter and the interferometer, multimode quantized light, and coherent
and incoherent states. The book provides a treatment of squeezed light and its use in
the Laser Interferometer Gravitational-Wave Observatory (LIGO). The Heisenberg
limit is described, along with NOON states and their application in super-sensitivity,
super-resolution, and quantum lithography. Applications of entanglement and coin-
cidence measurements are described including ghost imaging, quantum illumina-
tion, absolute photodetector calibration, and interaction-free measurement. Light-
matter interaction, atomic clocks, and atom cooling and trapping are included. The
book does not cover quantum computing, which is a topic treated by another book by
the author [1]. Together with quantum computing, the topics of this book attempt to
form an almost complete introductory description of the “second quantum
revolution”.
Since the book is intended for the undergraduate beginner, I try not to say “it is
easy to show” too often. Rigorous derivations are given, although some calculations
are completed as exercises in the book and heuristic arguments are occasionally
employed to avoid getting bogged down too much in calculations. The emphasis is
on physical understanding. Thus, in some instances, a simplification of the topics is
presented for pedagogical reasons. For example, the canonical quantization of light

v
vi Preface

is presented using entirely electric and magnetic fields, rather than the usual
approach of using the vector potential. I do not use density matrices anywhere in
this book. Students are expected to be familiar with the Dirac bra-ket notation and
tensor product of states.

Reference

1. Ray LaPierre, Introduction to Quantum Computing (Springer, 2021). https://doi.

org/10.1007/978-3-030-69318-3
Acknowledgments

I am grateful to McMaster University, the Department of Engineering Physics, my

many colleagues, my students, and my family for inspiring me to write this book.
Any errors in the book are entirely my own.

vii
How to Use This Book

This book is intended for a single semester (~12 week) elective course on quantum
optics, comprised of approximately 36 1-hour lectures (3 hours per week). Chapters
are intended to be covered consecutively. Instructors may wish to begin with an
overview of the bra-ket notation, inner product, expectation values, tensor product,
and related topics. A suggested lecture schedule is:
Lecture 1: Chap. 1 – Canonical Quantization
Lecture 2–3: Chap. 2 – Quantum Harmonic Oscillator
Lecture 4–5: Chap. 3 – Canonical Quantization of Light
Lecture 6: Chap. 4 – Fock States and the Vacuum
Lecture 7: Chap. 5 – Single Photon State
Lecture 8–9: Chap. 6 – Single Photon on a Beam Splitter
Lecture 10–11: Chap. 7 – Single Photon in an Interferometer
Lecture 12: Chap. 8: Entanglement
Lecture 13–14: Chap. 9 – Multimode Quantized Radiation
Lecture 15–16: Chap. 10 – Coherent State
Lecture 17: Chap. 11 – Coherent State on a Beam Splitter
Lecture 18–19: Chap. 12 – Incoherent State
Lecture 20–21: Chap. 13 – Homodyne and Heterodyne Detection
Lecture 22–23: Chap. 14 – Coherent State in an Interferometer
Lecture 24–25: Chap. 15 – Squeezed Light
Lecture 26–27: Chap. 16 – Squeezed Light in an Interferometer
Lecture 28–29: Chap. 17 – Heisenberg Limit
Lecture 30–31: Chap. 18 – Quantum Imaging
Lecture 32–33: Chap. 19 – Light–Matter Interaction
Lecture 34: Chap. 20 – Atomic Clock
Lecture 35–36: Chap. 21 – Atom Cooling and Trapping
The book assumes that students have successfully completed an introductory
course in quantum mechanics, which is typically in the second year of a 4-year
undergraduate program in physics or related disciplines. Thus, this book is intended

ix
x How to Use This Book

as a course for the third or fourth year of an undergraduate program, or the entry level
of a graduate program.
Each chapter includes exercises which can be completed by the student as
homework assignments or used for tutorial instruction. A solutions manual is
available from the publisher for qualified instructors. Each chapter also includes
references for more advanced study, and further reading is listed at the end of
the book.
Contents

1 Canonical Quantization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Hamiltonian Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Canonical Quantization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Commutation Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Quantum Harmonic Oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1 Classical Harmonic Oscillator . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Quantum Harmonic Oscillator . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Dirac Formalism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.4 Number Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.5 Annihilation Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.6 Creation Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.7 Creating Excited States from the Ground State . . . . . . . . . . . . 18
2.8 Expectation Value and Uncertainty . . . . . . . . . . . . . . . . . . . . . 21
2.9 Heisenberg Uncertainty Relation . . . . . . . . . . . . . . . . . . . . . . 23
2.10 Some Important Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3 Canonical Quantization of Light . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.1 Single Mode of Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2 Quadrature Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3 Classical Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.4 Canonical Quantization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.5 Time-Dependence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.6 Quadrature Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.7 Physical Observables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.8 Photons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

xi
xii Contents

4 Fock States and the Vacuum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.1 Photon Number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2 Electric Field of the Fock State . . . . . . . . . . . . . . . . . . . . . . . 42
4.3 Vacuum Fluctuations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.4 Experimental Evidence of Vacuum Fluctuations . . . . . . . . . . . 44
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5 Single Photon State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.1 Single Photon State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.2 Photodetection Signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.3 Single Photon Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.4 Single Photon Detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
6 Single Photon on a Beam Splitter . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.1 Classical Beam Splitter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.2 Quantum Beam Splitter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.3 Input/Output Transformation . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.4 Single Photon on a Beam Splitter . . . . . . . . . . . . . . . . . . . . . . 62
6.5 Coincident Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6.6 Second-Order Correlation Function . . . . . . . . . . . . . . . . . . . . 65
6.7 Entangled State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6.8 Hanbury Brown–Twiss Experiment . . . . . . . . . . . . . . . . . . . . 70
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
7 Single Photon in an Interferometer . . . . . . . . . . . . . . . . . . . . . . . . . 71
7.1 Classical Light Interference . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7.2 Quantum Light Interference . . . . . . . . . . . . . . . . . . . . . . . . . . 74
7.3 Wave-Particle Duality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
8 Entanglement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
8.1 Entangled States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
8.2 EPR Paradox and Hidden Variables . . . . . . . . . . . . . . . . . . . . 79
8.3 CHSH Inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
8.4 Testing the CHSH Inequality . . . . . . . . . . . . . . . . . . . . . . . . . 81
8.5 Quantum Key Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
9 Multimode Quantized Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
9.1 Multimode Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
9.2 Quantized Multimode Radiation . . . . . . . . . . . . . . . . . . . . . . . 85
9.3 Vacuum Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
9.4 Single Photon Wavepacket . . . . . . . . . . . . . . . . . . . . . . . . . . 88
9.5 Spontaneous Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Contents xiii

10 Coherent State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
10.1 Coherent State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
10.2 Coherent State as a Superposition of Fock States . . . . . . . . . . . 94
10.3 Photon Number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
10.4 Poisson Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
10.5 Electric Field of Coherent State . . . . . . . . . . . . . . . . . . . . . . . 98
10.6 Phasor Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
10.7 Time-Dependence of Coherent State . . . . . . . . . . . . . . . . . . . . 101
10.8 Quadratures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
10.9 Displacement Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
10.10 Number-Phase Uncertainty Relation . . . . . . . . . . . . . . . . . . . . 107
10.11 Revisiting the Fock State . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
11 Coherent State on a Beam Splitter . . . . . . . . . . . . . . . . . . . . . . . . . 111
11.1 Photodetection Probability . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
11.2 Coincidence Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . 113
12 Incoherent State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
12.1 Incoherent State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
12.2 Electric Field of Incoherent State . . . . . . . . . . . . . . . . . . . . . . 115
12.3 Photodetector Signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
12.4 Photon Number Distribution . . . . . . . . . . . . . . . . . . . . . . . . . 118
12.5 Photon Number Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . 119
12.6 Comparison of Different Types of Light . . . . . . . . . . . . . . . . . 121
Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
13 Homodyne and Heterodyne Detection . . . . . . . . . . . . . . . . . . . . . . . 125
13.1 Homodyne Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
13.2 Heterodyne Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
14 Coherent State in an Interferometer . . . . . . . . . . . . . . . . . . . . . . . . 131
14.1 Coherent Light Interference . . . . . . . . . . . . . . . . . . . . . . . . . . 131
14.2 Coincident Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
14.3 Homodyne Signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
14.4 Uncertainty in the Homodyne Signal . . . . . . . . . . . . . . . . . . . 136
15 Squeezed Light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
15.1 Classical Description of Nonlinear Optics . . . . . . . . . . . . . . . . 139
15.2 Quantum Description of Squeezing . . . . . . . . . . . . . . . . . . . . 141
15.3 Squeezing Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
15.4 Electric Field of Squeezed Light . . . . . . . . . . . . . . . . . . . . . . . 144
15.5 Quadratures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
15.6 Squeezed Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
15.7 Fragility of Squeezing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
15.8 Squeezed Vacuum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
15.9 Photon Number Distribution of Squeezed Light . . . . . . . . . . . . 152
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
xiv Contents

16 Squeezed Light in an Interferometer . . . . . . . . . . . . . . . . . . . . . . . . 157

16.1 Homodyne Signal Using Squeezed Light . . . . . . . . . . . . . . . . 157
16.2 Laser Interferometer Gravitational-Wave
Observatory (LIGO) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
17 Heisenberg Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
17.1 Heisenberg Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
17.2 Phase Shifter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
17.3 NOON States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
17.4 Super-sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
17.5 Super-resolution and Quantum Lithography . . . . . . . . . . . . . . 169
17.6 Producing a NOON State . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
17.7 Hong–Ou–Mandel Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
17.8 High NOON State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
17.9 Quantum State Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . 174
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
18 Quantum Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
18.1 Nonlocal Interference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
18.2 Ghost Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
18.3 Quantum Illumination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
18.4 Absolute Photodetector Calibration . . . . . . . . . . . . . . . . . . . . 180
18.5 Interaction-Free Measurement (Elitzur–Vaidman
Bomb Problem) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
19 Light–Matter Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
19.1 Jaynes–Cummings Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . 183
19.2 Spontaneous Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
19.3 Cavity Quantum Electrodynamics . . . . . . . . . . . . . . . . . . . . . 187
19.4 Circuit QED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
19.5 Rydberg Atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
19.6 Rabi Oscillations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
19.7 Collapse and Revival of Rabi Oscillations . . . . . . . . . . . . . . . . 194
19.8 Example of a CQED Experiment . . . . . . . . . . . . . . . . . . . . . . 197
19.9 Dressed Atom–Cavity States and Vacuum Rabi Splitting . . . . . 199
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
20 Atomic Clock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
20.1 Quartz Oscillators—Before Atomic Time . . . . . . . . . . . . . . . . 205
20.2 Resonance Frequency in the Cs Atom . . . . . . . . . . . . . . . . . . . 205
20.3 Stern–Gerlach Apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
20.4 Thermal Atomic Clock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
20.5 Improvements to the Atomic Clock . . . . . . . . . . . . . . . . . . . . 209
20.6 Applications of the Atomic Clock . . . . . . . . . . . . . . . . . . . . . 210
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Contents xv

21 Atom Cooling and Trapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211

21.1 Paul Trap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
21.2 Laser Cooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
21.3 Magneto-optical Trap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
21.4 Sisyphus Cooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
21.5 Dipole Trap, Optical Tweezers, and Optical Lattice . . . . . . . . . 219
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222

Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

Appendix 1: Derivation of Lamb Shift . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Appendix 2: Derivation of Casimir Formula . . . . . . . . . . . . . . . . . . . . . . 231
Appendix 3: Derivation of Normalization Constant in Single Photon
Wavepacket . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Appendix 4: Derivation of Planck’s Distribution Law . . . . . . . . . . . . . . . 235
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
Chapter 1
Canonical Quantization

We begin with Hamiltonian mechanics and a method called “canonical quantiza-

tion”, developed by Paul Dirac, used for finding the Hamiltonian of a quantum
system from its classical counterpart. Once you know the Hamiltonian of the
quantum system, you can determine its quantum properties from the time-dependent
Schrodinger equation. The canonical quantization procedure gives us the canoni-
cally conjugate variables of the system that satisfy a commutation relation, such as
the Heisenberg uncertainty relation.

1.1 Hamiltonian Mechanics

Hamiltonian mechanics was formulated by William Rowan Hamilton in 1833.

Hamiltonian mechanics is equivalent to Newton’s laws of motion but provides a
simplification of the analysis for many dynamical systems. Another approach is
Lagrangian mechanics, which we leave to the reader as a topic for independent
study. In Hamiltonian mechanics, a system is described by canonically conjugate
variables denoted by qi and pi:

q 1 , q2 , . . . , q i , . . . ; p1 , p 2 , . . . , pi , . . . ð1:1Þ

qi and pi are also called the generalized position and momentum coordinates,
respectively. For example, q1, q2 and q3 may refer to the actual position coordinates
(x, y, z) of a particle and p1, p2 and p3 correspond to its linear momentum ( px, py and
pz). If there is more than one particle, then q4, q5, q6, p4, p5 and p6 are the
corresponding variables for the second particle, and so on. In general, qi and pi
may represent dynamic variables other than position and momentum, depending on
the system. For example, to describe a pendulum (Exercise 1.1), it is easier to assign
qi as the angle of the pendulum and pi as the angular momentum. The qi and pi
variables, if they are canonically conjugate variables, satisfy the Hamilton equations:

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 1

R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_1
2 1 Canonical Quantization

Fig. 1.1 A potential, U(x)

dqi ∂H
¼ ð1:2Þ
dt ∂pi
dpi ∂H
¼ ð1:3Þ
dt ∂qi

where H is the Hamiltonian and t is the time. The Hamiltonian is the total energy of
the system (kinetic energy plus potential energy) expressed in terms of the general-
ized coordinates.
To illustrate Hamilton’s approach, let us find the equations of motion for a
particle of mass m in a one-dimensional potential, U(x), shown in Fig. 1.1. Although
U(x) is actually the potential energy, physicists often abbreviate this simply as “the
potential”. In this example, suppose the generalized coordinates (qi, pi) are the
position (x) and momentum ( p) of the particle:

q!x ð1:4Þ
dx
p!m ð1:5Þ
dt

The Hamiltonian is the total energy (kinetic energy plus potential energy) expressed
in terms of the generalized coordinates from Eqs. (1.4) and (1.5):

p2
H¼ þ U ð xÞ ð1:6Þ
2m

Using Eq. (1.6), the Hamilton equations become

dx ∂H p
¼ ¼ ¼v ð1:7Þ
dt ∂p m
dp ∂H ∂U
¼ ¼ ¼F ð1:8Þ
dt ∂x ∂x

where v is the velocity and F is the force. Equation (1.7) is simply the definition of
momentum, while Eq. (1.8) reproduces the correct dynamical equation according to
Newton’s laws of motion. Thus, x and p satisfy the Hamilton equations, and we say
that x and p are canonically conjugate variables.
1.2 Canonical Quantization 3

Instead, suppose the generalized coordinates (qi, pi) are taken as the position (x)
and velocity (v) of the particle:

q!x ð1:9Þ
p!v ð1:10Þ

The Hamiltonian expressed in terms of the generalized coordinates, Eqs. (1.9) and
(1.10), is now

1
H ¼ mv2 þ U ðxÞ ð1:11Þ
2

The Hamilton equations become

dx ∂H
¼ ¼ mv ðincorrectÞ ð1:12Þ
dt ∂v
dv ∂H ∂U
¼ ¼ ¼ F ðincorrectÞ ð1:13Þ
dt ∂x ∂x

dt ¼ v is clearly not equal to mv, and

Equations (1.12) and (1.13) are incorrect, since dx
dv
dt is not equal to F according to Newton’s laws of motion. Thus, x and v are not
canonically conjugate variables.

Exercise 1.1 Solve the equation of motion for a simple pendulum using
Newton’s laws. Repeat using Hamiltonian mechanics.

1.2 Canonical Quantization

Canonical quantization is a prescribed method of finding the Hamiltonian of a

quantum system. The procedure was developed by Paul Dirac in 1925 (Fig. 1.2).
Dirac proposed that any system for which we have a classical description can be
quantized according to the procedure of canonical quantization. In canonical quan-
tization, the generalized coordinates of the classical description, found by
Hamilton’s approach (described in the previous section), are replaced by the
corresponding quantum operators (denoted by a “hat”,b):

canonical
quantization
H ðq1 , ..., qi , ...; p1 , .. ., pi , .. .Þ ! H b ðb
q1 , . .., b
qi , . ..; b
p1 , ... , b
pi , ...Þ
ð1:14Þ
4 1 Canonical Quantization

Fig. 1.2 Paul Dirac (Nobel

Prize in Physics in 1933).
(Credit: Wikimedia
Commons [1])

where the classical description is on the left and the quantum description is on the
right. The Hamiltonian of the quantum system, H b , is expressed in terms of the
generalized coordinates (now operators) on the right-hand side of Eq. (1.14). For
example, according to Sect. 1.1, the generalized coordinates for a particle of mass
m in a potential, U(x), are x and p. The Hamiltonian for the corresponding quantum
system becomes

canonical
p2
b¼ b
p 2 quantization
H¼ þ U ðxÞ ! H þ U ðbxÞ ð1:15Þ
2m 2m

Once you know H b of the quantum system, you can determine its quantum
properties from the time-dependent Schrodinger equation:

∂jψi b
iħ ¼ H jψi ð1:16Þ
∂t

where jψi is the state of the system and ħ is the reduced Planck constant (ħ ¼ h/2π).
You may remember from introductory quantum mechanics that Eq. (1.16) reduces to
the time-independent Schrodinger equation for stationary states:

b j ψn i ¼ E n j ψ n i
H ð1:17Þ

where En are the eigenenergies and jψni are the eigenstates (basis states) of the
system.
1.3 Commutation Relations 5

1.3 Commutation Relations

qi , b
Dirac showed that the canonically conjugate variables (b pj) of the quantum system
satisfy the commutation relation:
 
bqi , b
pj ¼ iħδij ð1:18Þ

where δij is the Kronecker function and, by definition,

 
qi , b
b pj ¼ b pj  b
qi b pj b
qi ð1:19Þ

Thus, when i ¼ j, we say that b qi and bpi “do not commute”; that is, ½b
qi , b
pi  ¼
b pi  b
qib pi b
qi ¼ iħ. Otherwise, the operators commute. For example, when the gener-
alized coordinates (bqi , b
pi ) are the position and momentum, we have

½bx, b
px  ¼ iħ ð1:20Þ
 
by, b
py ¼ iħ ð1:21Þ
 
bz, b
pz ¼ iħ ð1:22Þ

Thus, position and momentum along the same direction do not commute (e.g., bx and
b
px do not commute), while position and momentum along different directions do
commute (e.g., bx and b
py commute). Eq. (1.20) leads to the well-known Heisenberg
uncertainty relation:

ħ
ΔxΔpx  ð1:23Þ
2

with the same relation for the y and z directions arising from Eq. (1.21) and (1.22),
respectively. In Eq. (1.23), Δx is the uncertainty in position x and Δpx is the
uncertainty in momentum along x. Uncertainty is defined as the standard deviation
or root mean square (rms) error:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Δx ¼ hðx  hxiÞ2 i ð1:24Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ hx2 þ hxi2  2xhxii ð1:25Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ h x2 i  h x i 2 ð1:26Þ

where the brackets hi denote an average (in quantum mechanics, this is called the
“expectation value” of x). In general, Eq. (1.26) gives the uncertainty in a measurable
quantity x, and will be used frequently throughout this book. Note that (Δx)2 is
known as the variance of x.
6 1 Canonical Quantization

Heisenberg’s uncertainty relation (also known as the uncertainty principle) limits

the accuracy with which a pair of canonically conjugate variables can be measured at
the same time. Equation (1.23), for example, tells us that we cannot measure the
position and momentum simultaneously to arbitrary accuracy. This has nothing to do
with the accuracy of the instruments but is rather a fundamental limitation of nature.
Conjugate variables that do not commute cannot be measured simultaneously
without some minimum uncertainty. We will derive Eq. (1.23) in the next chapter
in the case of the quantum harmonic oscillator, which is the most important quantum
system in quantum optics.

Reference

1. https://commons.wikimedia.org/wiki/File:Paul_Dirac,_1933.jpg
Chapter 2
Quantum Harmonic Oscillator

The quantum harmonic oscillator (QHO) is introduced using the canonical quanti-
zation of the classical harmonic oscillator. An alternative formalism of the QHO due
to Dirac is introduced along with the creation and annihilation operators. The
expectation value and uncertainty of the position and momentum are derived,
resulting in the Heisenberg uncertainty relation.

2.1 Classical Harmonic Oscillator

Consider a classical system comprised of a particle of mass, m, and position, x,

moving in a one-dimensional parabolic potential:

1 1
U ðxÞ ¼ kx2 ¼ mω2 x2 ð2:1Þ
2 2
pffiffiffiffiffiffiffiffiffi
where k is a force constant and ω ¼ k=m is the angular frequency. The harmonic
oscillator arises in a wide variety of classical systems, but most often as a mass on a
spring described by Hooke’s law (F ¼  kx). The generalized coordinates for this
system are simply the position and momentum:

q!x ð2:2Þ
dx
p!m ð2:3Þ
dt

and the Hamiltonian becomes

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 7

R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_2
8 2 Quantum Harmonic Oscillator

p2 1
H¼ þ mω2 x2 ð2:4Þ
2m 2

The Hamilton equations become

dx ∂H p
¼ ¼ ¼v ð2:5Þ
dt ∂p m
dp ∂H ∂U
¼ ¼ mω2 x ¼  ¼F ð2:6Þ
dt ∂x ∂x

The first equation is the definition of momentum  ( p ¼ mv), while the second
equation reproduces Newton’s equation F ¼ dp dt . Thus, x and p satisfy the Hamilton
equations (they give the correct dynamical behavior) and are therefore canonically
conjugate variables.
Equations (2.5) and (2.6) are easily solved. Combining the two equations gives

d2 x
¼ ω2 x ð2:7Þ
dt 2

with the solution

x ¼ a cos ðωt þ φÞ ð2:8Þ

where the amplitude, a, and phase, φ, are determined by initial conditions. Equiv-
alently, the solution may be written as

x ¼ Aeiωt þ c:c: ð2:9Þ

where c.c. denotes the complex conjugate, and

a
A ¼ eiφ ð2:10Þ
2

Note that a is a real number, but A is a complex number. The negative sign in the
exponent of Eq. (2.9) is by convention, although the positive exponent is also a valid
solution.

2.2 Quantum Harmonic Oscillator

As described in Chap. 1, the Hamiltonian for the quantum harmonic oscillator

(QHO) is obtained by canonical quantization where the coordinates (x, p) are
replaced by their quantum operators:
2.2 Quantum Harmonic Oscillator 9

canonical
p2 1
b¼ b
p 2
1 quantization
H¼ þ mω2 x2 ! H þ mω2bx2 ð2:11Þ
2m 2 2m 2

where bx and b
p obey the commutation relation:

½bx, b
p ¼ iħ ð2:12Þ

Equation (2.12) can be used to find the momentum operator b p in terms of the
x coordinate. Starting from Eq. (2.12) and according to the definition of the com-
mutation relation:

ðbx b
pb
p bxÞjψi ¼ iħjψi ð2:13Þ

Expanding the left side of Eq. (2.13) gives

bx b
pjψi  b
p bxjψi ¼ iħjψi ð2:14Þ

If jψi is in the position representation (i.e., jψi represents the familiar wavefunction,
ψ(x)), then the operator bx is simply the position, x; that is, bxjψi ¼ xjψi . Thus,
Eq. (2.14) becomes

pjψi  b
xb pxjψi ¼ iħjψi ð2:15Þ

In the second term on the left, b

pxjψi, we apply the rules of partial differentiation, that
is, the operator b
p operates on x while keeping jψi constant, and then b p operates
on jψi while keeping x constant. This gives

pjψi  ðb
xb pxÞjψi  xðb
pjψiÞ ¼ iħjψi ð2:16Þ

In the second term on the left, b

p operates on x only. In the third term on the left, bp
operates on jψi only. The first and third terms on the left of Eq. (2.16) cancel, giving

ðb
pxÞjψi ¼ iħjψi ð2:17Þ

or

b
px ¼ iħ ð2:18Þ

The solution to Eq. (2.18) is

ħ ∂
b
p¼ ð2:19Þ
i ∂x
10 2 Quantum Harmonic Oscillator

Equation (2.19) is easily checked by substitution in Eq. (2.18). Thus, the left-hand
side of Eq. (2.18) is obtained by applying the differential operator b p to x, giving
ħ ∂ ħ
i ∂x ðx Þ ¼ i , which is identical to the right-hand side of Eq. (2.18). Eq. (2.19) is a
relationship that should be familiar from introductory quantum mechanics—it is the
momentum in the position representation.
Using Eq. (2.19), the Hamiltonian in Eq. (2.11) can be written entirely in terms of
the position, x:

b ¼  ħ d þ 1 mω2 x2
2 2
H ð2:20Þ
2m dx2 2

Finally, the stationary eigenstates and eigenenergies of the QHO can be found by
substituting the Hamiltonian of Eq. (2.20) into the time-independent Schrodinger
equation, Eq. (1.17), giving

ħ2 d 2 1
 jψ i þ mω2 x2 jψn i ¼ E n jψn i ð2:21Þ
2m dx2 n 2

The solution to Eq. (2.21) can be found in any introductory textbook on quantum
mechanics. The eigenstates are

 14 1 2
1 mω 2ξ
jψn i ¼ pffiffiffiffiffiffiffiffi
ffi e H n ð ξÞ ð2:22Þ
2n n! πħ
mω1=2
where Hn(ξ) are Hermite polynomials of degree n; n ¼ 0, 1, 2, . . .; and ξ ¼ ħ x.
The first few Hermite polynomials, Hn(ξ), are

H0 ¼ 1 ð2:23Þ
H 1 ¼ 2ξ ð2:24Þ
H 2 ¼ 4ξ2  2 ð2:25Þ
H 3 ¼ 8ξ3  12ξ ð2:26Þ
H 4 ¼ 16ξ  48ξ þ 12
4 2
ð2:27Þ
H 5 ¼ 32ξ  160ξ þ 120ξ
5 3
ð2:28Þ

The eigenenergies resulting from Eq. (2.21) are

 
1
En ¼ n þ ħω . . . , n ¼ 0,1,2, . . . ð2:29Þ
2
2.3 Dirac Formalism 11

Fig. 2.1 The first few

eigenenergies and
eigenstates (wavefunctions)
of the quantum harmonic
oscillator (QHO)

The first few eigenstates and eigenenergies of the QHO are shown in Fig. 2.1. An
important feature of the QHO is the existence of a minimum energy (ground state
energy), E 0 ¼ 12 ħω, corresponding to n ¼ 0 in Eq. (2.29).

2.3 Dirac Formalism

b can be
Paul Dirac formulated an alternative approach to solve the QHO. Suppose H
factorized as follows:

b {O
b ¼O
H b þ E0 ð2:30Þ

b { is the Hermitian conjugate. If jψni is an eigenstate

b is some operator and O
where O
b then the eigenenergies are
of H,

b ni
En ¼ hψn j Hjψ ð2:31Þ

b gives
Substituting Eq. (2.30) for H

b {O
En ¼ hψn jðO b þ E 0 Þjψn i
ð2:32Þ
b { Ojψ
¼ hψn jO b n i þ E0

This means:

En  E0 ð2:33Þ

b jψ0 i ¼ 0, then the minimum energy (ground state energy, E0) is found.
If O
b and P:
At this point, it is helpful to define dimensionless operators, Q b
rffiffiffiffiffiffiffi
b¼ mω
Q bx ð2:34Þ
ħ
12 2 Quantum Harmonic Oscillator

rffiffiffiffiffiffiffiffiffiffi
b¼ 1
P b
p ð2:35Þ
mħω

It is easily shown that Eqs. (2.11) and (2.12) become

 2 
b ¼ ħω Q
H b þP
b2 ð2:36Þ
2
h i
b P
Q, b ¼i ð2:37Þ

b like Eq. (2.30), we can rewrite H

To perform the factorization of H, b in terms of
b and P:
Q b
    
b ¼ ħω p1ffiffiffi Q
H b p1ffiffiffi Q
b  iP b þ1
b þ iP ð2:38Þ
2 2 2

b and P
Note that, according to Eq. (2.37), Q b do not commute, which leads to the factor
of ½ in Eq. (2.38).

Exercise 2.1 Derive Eq. (2.38).

Equation (2.38) inspires us to define two new operators:

 
1 b b
a ¼ pffiffiffi Q
b þ iP ð2:39Þ
2
 
1 b b
a{ ¼ pffiffiffi Q
b  iP ð2:40Þ
2

where b a{ is called the annihilation,

a is called the creation or raising operator and b
destruction or lowering operator. The reason for these names will become clear later.
With these definitions, the Hamiltonian in Eq. (2.38) becomes
 
b ¼ ħω b 1
H a{b
aþ ð2:41Þ
2

We have succeeded in factoring H b , similar to Eq. (2.30). We will see below that
b
ajψ0 i ¼ 0, so Eq. (2.41) gives the ground state energy of E 0 ¼ 12 ħω, identical to
that from Eq. (2.29).
Note that Eq. (2.41) does not contain bx or b p . One of the beautiful aspects of
Dirac’s formalism is that the expression for Hb in Eq. (2.41) is applicable to any type
of oscillator—for example, the quantization of atomic vibrations in a crystal
(phonons), the quantization of flux and charge in an LC circuit oscillator, and the
2.3 Dirac Formalism 13

quantization of the electromagnetic field (photons). The last example is the topic of
the next chapter.
Using Qb and P,
b we can also derive the commutation relation for b
a and ba{ :
h i
b a{ ¼ b
a, b a{  b
ab a{ b
a
        
1 b b p1ffiffiffi Q b  p1ffiffiffi Q
b  iP b p1ffiffiffi Q
b  iP b þ iP
b
¼ pffiffiffi Q þ iP
h2 2  22 2 i
1 b2 b b b b b 2
b b b b b þP b2
¼ Q  iQ P þ iP Q þ P  Q þ iQ P  iP Q
2  
¼ i QbPbP bQb
h i
b P
¼ i Q, b
¼ iðiÞ
¼1

where we have used Eq. (2.37). Thus, we arrive at the important result:
h i
b a{ ¼ 1
a, b ð2:42Þ

or, by definition:

b
aba{  b
a{ b
a¼1 ð2:43Þ

Thus, ba and ba{ do not commute. Eq. (2.43) will have far-reaching consequences
throughout this book.
Note that, by definition, Eqs. (2.39) and (2.40) indicate that b
a 6¼ ba{ . Thus, b
a and
{ {
a are not Hermitian. This means that b
b a and b
a do not correspond to any physical
observable (Exercise 2.2). However, we can form observables from combinations of
b a{. For example, Q
a and b b and P
b can be derived in terms of ba and ba{ . By rearranging
Eqs. (2.39) and (2.40), we obtain
 
b ¼ p1ffiffiffi b
Q a{
aþb ð2:44Þ
2
 
b¼p i
P ffiffiffi b a{
ab ð2:45Þ
2

b¼Q
Q b { and P b{ , so Q
b¼P b and P
b are Hermitian and may correspond to observables
(position and momentum, respectively).

Exercise 2.2 Show that operators corresponding to physical observables must

be Hermitian.
14 2 Quantum Harmonic Oscillator

2.4 Number Operator

The number operator is defined as

b ¼b
N a{ b
a ð2:46Þ

Thus, we can express the Hamiltonian for the QHO as

   
b ¼ ħω b 1 b þ1
H a{b
aþ ¼ ħω N ð2:47Þ
2 2

For simplicity and by convention, let us adopt a new notation for the eigenstate of a
QHO, replacing jψni with the ket, jni. jni is called the number state or Fock state,
named after the Russian theorist, Vladimir Fock. Thus,

b j ni ¼ E n j n i
H ð2:48Þ

b and Eq. (2.29) for En, we obtain

Substituting Eq. (2.47) for H
   
ħω Nb þ 1 jni ¼ ħω n þ 1 jni ð2:49Þ
2 2

Thus, canceling 12 ħωjni on both sides of Eq. (2.49), we get

b jni ¼ njni
N ð2:50Þ

The number operator, N b ¼b a{b

a, operating on jni gives the eigenvalue, n.
Any eigenstate b b
 of N with eigenvalue n is also an eigenstate of H with eigenvalue
E n ¼ ħω n þ 2 . The states jni form an orthonormal basis:
1

hnjmi ¼ δnm ð2:51Þ

where n and m ¼ 0, 1, 2, . . . . Also, the states jni form a complete basis, meaning any
state jψi of the QHO can be written as a superposition of the jni states.
As an illustration of the Dirac formalism, let us find the ground state
wavefunction for the QHO. Applying Eq. (2.50) with n ¼ 0, we get

b j0i ¼ 0
N

or

a{b
b aj0i ¼ 0 ð2:52Þ

Applying the bra, h0j, to both sides gives the inner product:
2.5 Annihilation Operator 15

a{b
h0jb aj0i ¼ 0 ð2:53Þ

or, equivalently, the modulus is zero:

kb
aj0ik ¼ 0 ð2:54Þ

Since the eigenstates are normalized (kj0ik ¼ 1), we must have

b
aj0i ¼ 0 ð2:55Þ
  pffiffiffiffiffi qffiffiffiffiffiffiffi
a ¼ p1ffiffi2 Q
Substituting b b þ iP b ¼ mω bx, P
b ,Q b¼ p ¼ ħi ∂
mħω b p, bx ¼ x, and b
1
ħ ∂x
into
Eq. (2.55) gives the differential equation:


ħ ∂
xþ j 0i ¼ 0 ð2:56Þ
mω ∂x

The solution to Eq. (2.56) is

 14
mω
e 2
1 mω 2
j0i ¼ ħ x ð2:57Þ
πħ

which can be verified by substitution into Eq. (2.56). Equation (2.57) is the correct
ground state wavefunction of the QHO, according to Eq. (2.22) with n ¼ 0. Also, as
mentioned previously, baj0i ¼ 0, which gives the ground state energy of E 0 ¼ 12 ħω
from Eq. (2.47).
The Dirac formalism presents an alternative approach to solve the QHO as
compared to Sect. 2.2. At the end of Sect. 2.7, we will see how to generate the
excited state wavefunctions (i.e., j1i, j2i, . . .). First, we need some more formalism.

2.5 Annihilation Operator

In the previous section, we found baj0i ¼ 0. What does the general case, b
ajni, give?
b jni ¼ njni, we have
Starting with the definition of the number operator, N

a{b
b a j ni ¼ n j n i ð2:58Þ

Applying b
a to both sides gives

b
aba{b
aj ni ¼ b
anjni ð2:59Þ

Using the commutation relation, Eq. (2.43) gives

16 2 Quantum Harmonic Oscillator

a{ b
ðb ajni ¼ b
a þ 1Þb anjni ð2:60Þ

Rearranging, we get

a{b
ðb a j n i ¼ ð n  1Þ b
aÞb ajni ð2:61Þ

or

bb
N ajni ¼ ðn  1Þb
ajni ð2:62Þ

Thus, b b with eigenvalue n  1. In other words, b

ajni is an eigenstate of N ajni gives the
state jn  1i:

b
ajni ! jn  1i ð2:63Þ

For this reason, b

a is called the lowering or annihilation operator. It lowers n to n  1.
In other words, it destroys one quantum of excitation of the QHO.
Is b
ajni normalized? Let us suppose that b ajni produces the state jn  1i with
coefficient (normalization factor) cn:

j ψi ¼ b
ajni ¼ cn jn  1i ð2:64Þ

Using jψi ¼ b
ajni, the inner product hψj ψi is

a{ b
hψjψi ¼ hnjb ajni ¼ n ð2:65Þ

Alternatively, using jψi ¼ cnjn  1i, the inner product hψj ψi is


hψjψi ¼ n  1jcn cn jn  1 ¼ jcn j2 ð2:66Þ

where cn is the complex conjugate of cn. Equating the two inner products in
Eqs. (2.65) and (2.66) gives the normalization factor:
pffiffiffi
cn ¼ n ð2:67Þ

Thus, from Eq. (2.64)

pffiffiffi
b
a j ni ¼ nj n  1 ð2:68Þ
2.6 Creation Operator 17

2.6 Creation Operator

In the previous section, we found that the annihilation operator, b

ajni, gives the state
pffiffiffi
jn  1i with normalization factor n. What about b a{ jni? Again, let us start with the
definition of the number operator, Nb jni ¼ njni, or

a{b
b a j ni ¼ n j n i ð2:69Þ

a{ to both sides gives

Applying b

a{ b
b a{ b a{ n j n i
a j ni ¼ b ð2:70Þ

Using the commutation relation, Eq. (2.43) gives

a{ ðb
b aba{  1Þjni ¼ b
a{ nj ni ð2:71Þ

Rearranging, we get

a{b
b aba{ jni ¼ ðn þ 1Þb
a{ jni ð2:72Þ

or

bb
N a{ jni ¼ ðn þ 1Þb
a { j ni ð2:73Þ

Thus, b b with eigenvalue n þ 1. In other words, b

a{ jni is an eigenstate of N a{ jni gives
the state jn þ 1i:

a{ jni ! jn þ 1i
b ð2:74Þ

a{ is called the raising or creation operator. It raises n to n þ 1. In other words, b

b a{
creates one quantum of excitation of the QHO.
a{ jni normalized? Let us suppose that b
Is b a{ jni produces the state jn þ 1i with
coefficient (normalization factor) cn:

a{ jni ¼ cn jn þ 1i
j ψi ¼ b ð2:75Þ

a{ jni, the inner product hψj ψi is

Using jψi ¼ b

hψjψi ¼ hnjba{ jni ¼ hnjðb

ab a{ b
a þ 1Þjni ¼ n þ 1 ð2:76Þ

Alternatively, using jψi ¼ cnjn þ 1i, the inner product hψjψi is

18 2 Quantum Harmonic Oscillator


hψjψi ¼ n þ 1jcn cn jn þ 1 ¼ jcn j2 ð2:77Þ

Equating the two inner products, Eqs. (2.76) and (2.77), gives the normalization
factor:
pffiffiffiffiffiffiffiffiffiffiffi
cn ¼ nþ1 ð2:78Þ

Thus, from Eq. (2.75),

pffiffiffiffiffiffiffiffiffiffiffi
a{ jni ¼
b n þ 1 jn þ 1 ð2:79Þ

2.7 Creating Excited States from the Ground State

Starting with the j0i state, we can create any excited state using multiple applications
of the creation operator:

b a{ Þn j0i ¼ n ðb
Nðb
n
a{ Þ j 0 i ð2:80Þ
n
Equation (2.80) states that ðba{ Þ j0i is an eigenstate of the number operator, N,b with
n 
eigenvalue, n. Also, ðb { b
a Þ j0i is an eigenstate of H with eigenvalue En¼ ħω n þ 12 .
Let us prove Eq. (2.80). To do so, we need to arrange all b a{ operators to the left of
all b
a operators. This arrangement is called the normal order. This is a commonly used
trick in quantum optics that will be found throughout this book (indeed, we already
used this trick in Eqs. (2.60) and (2.76)). The normal ordering is done by repeated
application of the commutation relation, b a{ ¼ b
ab a{ b
a þ 1. We start with the left side of
Eq. (2.80) and apply the definition of N b ¼b {
ab a:
 
b a{ Þn j0i ¼ b
Nðb a{ b
a b { {
b
a ⋯b a{ j 0 i
a|fflfflfflfflffl{zfflfflfflfflffl} ð2:81Þ
n times
 
a{ b
¼b a{ b
ab a{ b
a{ ⋯b a{ j 0 i
|fflfflfflfflffl{zfflfflfflfflffl} ð2:82Þ
n1 times

a{ is repeated n times, and similarly for

In Eq. (2.81), “n times” means that b
subsequent equations. Using the commutation relation, b
aba{ ¼ b
a{ b
a þ 1, gives
2.7 Creating Excited States from the Ground State 19

 n  
b b
N a{ j 0 i ¼ b
a{ b
a{ b
aþ1 ba{ b
a{ ⋯b a{ j 0 i ð2:83Þ
|fflfflfflfflffl{zfflfflfflfflffl}
n1 times

a{b
¼b a{ b
aba{ba{ ⋯b a{ j 0 i þ b
|fflfflfflfflffl{zfflfflfflfflffl} a{ b
a{ ⋯b a{ j0i
|fflfflfflfflffl{zfflfflfflfflffl} ð2:84Þ
n1 times n times

If we repeat the normal ordering on the first term of Eq. (2.84), we get
 n  
b b
N a{ j0i ¼ b
a{ b
a{ ba{ b
aþ1 ba{ba{ ⋯b a{ j 0 i þ b
a{ b
a{ ⋯b a { j 0i ð2:85Þ
|fflfflfflfflffl{zfflfflfflfflffl} |fflfflfflfflffl{zfflfflfflfflffl}
n2 times n times

{ { { { { { { { { { {
¼b
abab
abab
ab a ⋯b a j 0i þ b
|fflfflfflfflffl{zfflfflfflfflffl} ab a ⋯b a j0i þ b
|fflfflfflfflffl{zfflfflfflfflffl} b
a ⋯b a{ j0i
a|fflfflfflfflffl{zfflfflfflfflffl} ð2:86Þ
n2 times n times n times

We see that the last two terms in Eq. (2.86) are identical. Each application of the
normal ordering shifts the annihilation operator b a once toward the right and adds the
{ { {
term b
a b
a ⋯b a
|fflfflfflfflffl{zfflfflfflfflffl} j 0 i. Repeated application of the normal ordering procedure eventually
n times
results in

b a { Þ n j 0i ¼ b
Nðb a{ b
a{ ⋯b a{b
aj0i þ nb a{ ⋯b a { j 0i ð2:87Þ
|fflfflfflfflffl{zfflfflfflfflffl}
n times

Finally, the first term becomes 0 (since b

aj0i ¼ 0), giving

b a{ Þn j0i ¼ nb
Nðb a{ b
a{ ⋯b a{ j0i ð2:88Þ
|fflfflfflfflffl{zfflfflfflfflffl}
n times

{ n
¼ nðb
a Þ j0i ð2:89Þ

Thus, we have proven Eq. (2.80).

n
a{ Þ j0i normalized? That is, does hψj ψi ¼ 1?:
Is jψi ¼ ðb

hψjψi ¼ h0jb
a|fflfflfflb ⋯b a{ b
ffla} b
a{zfflffl a{ ⋯b a{ j 0 i
|fflfflfflfflffl{zfflfflfflfflffl} ð2:90Þ
n times n times
 
¼ h0j b
a|fflfflfflb ⋯bffl} b
a{zfflffl a a{ b
ab a{ b
a{ ⋯b a{ j 0 i
|fflfflfflfflffl{zfflfflfflfflffl} ð2:91Þ
n1 times n1 times

Applying the commutation relation to the middle b

aba{ pair:
20 2 Quantum Harmonic Oscillator

 
{
hψjψi ¼ h0j b
a|fflfflfflb ⋯b
a{zfflffl
ffl}a b
a b
a þ 1 a{b
b a{ ⋯b a{ j 0 i
|fflfflfflfflffl{zfflfflfflfflffl} ð2:92Þ
n1 times n1 times
 
¼ h0j b
a|fflfflfflb ⋯bffl} b
a{zfflffl a a{ b
a b { {
b
a ⋯b a{ j0i þ h0j b
a|fflfflfflfflffl{zfflfflfflfflffl} |fflfflfflb
a ⋯b a{ b
ffla} b
a{zfflffl a{ ⋯b a{ j 0 i
|fflfflfflfflffl{zfflfflfflfflffl} ð2:93Þ
n1 times n1 times n1 times n1 times

Repeated application (n times) of the normal ordering procedure results in

hψjψi ¼ n h0j b
a|fflfflfflb ⋯b a{ b
ffla} b
a{zfflffl a{ ⋯b a{ j0i
|fflfflfflfflffl{zfflfflfflfflffl} ð2:94Þ
n1 times n1 times

Now we repeat the normal ordering procedure again, resulting in

hψjψi ¼ nðn  1Þ h0j b

a|fflfflfflb ⋯b
a{zfflffl a{ b
ffla} b a{ ⋯b a{ j0i
|fflfflfflfflffl{zfflfflfflfflffl} ð2:95Þ
n2 times n2 times

Continuing in this manner, we eventually get

hψjψi ¼ nðn  1Þðn  2Þ . . . ð1Þ ¼ n! ð2:96Þ

where n! refers to the factorial of n. Hence, the normalized state is

n
a{ Þ
ðb
jni ¼ pffiffiffiffi j0i ð2:97Þ
n!

pffiffiffi
Exercise 2.3 Starting from Eq. (2.97), show that b
ajni ¼ n jn  1i, identical
to Eq. (2.68).
pffiffiffiffiffiffiffiffiffiffiffi
a{ jni ¼
Exercise 2.4 Starting from Eq. (2.97), show that b n þ 1 jn þ 1 ,
identical to Eq. (2.79).

Starting with Eq. (2.57) for the wavefunction of the j0i state, we can generate the
wavefunction for the excited states of the QHO using Eq. (2.97):
 n  14 1 mω 2
1 mω
jni ¼ pffiffiffiffi ba{ e 2 ħ x ð2:98Þ
n! πħ
  pffiffiffiffiffi qffiffiffiffiffiffiffi
a{ ¼ p1ffiffi2 Q
Substituting b b  iP
b ,Qb ¼ mω bx, P b¼ p ¼ ħi ∂x
∂
mħω b p, bx ¼ x, and b
1
ħ into
Eq. (2.98) gives the differential equation:
2.8 Expectation Value and Uncertainty 21

" rffiffiffiffiffiffiffi rffiffiffiffiffiffiffi ! #n

 14
1 mω mω ħ ∂
e 2 ħ x
1 mω 2
jni ¼ pffiffiffiffiffiffiffiffi
ffi x ð2:99Þ
2 n! πħ
n ħ mω ∂x

mω12
If we introduce our prior notation of ξ ¼ ħ x, we get

 14 
 n
1 mω ∂
e 2 ξ
1 2
jni ¼ pffiffiffiffiffiffiffiffi
ffi ξ ð2:100Þ
n
2 n! πħ ∂ξ

By induction, you can show that

n n
∂ ∂ ξ2
e2 ξ ¼ ð1Þn e2 ξ
1 2 1 2
ξ e ð2:101Þ
∂ξ ∂ξn

Using Eq. (2.101), we can write Eq. (2.100) as

 14 n
1 mω n 12 ξ2 ∂ ξ2
jni ¼ pffiffiffiffiffiffiffiffi
ffi ð 1 Þ e ne ð2:102Þ
n
2 n! πħ ∂ξ
 14 1 2
1 mω 2 ξ
¼ pffiffiffiffiffiffiffiffi ffi e H n ðξÞ ð2:103Þ
2 n! πħ
n

where
n
∂ ξ2
H n ðξÞ ¼ ð1Þn eξ
2
e ð2:104Þ
∂ξn

You can show that Eq. (2.104) generates Eqs. (2.23), (2.24), (2.25), (2.26), (2.27)
and (2.28). Eq. (2.103) is identical to Eq. (2.22). Thus, we have generated the excited
states for the QHO using the Dirac formalism.

2.8 Expectation Value and Uncertainty

The average or expectation value of Q (dimensionless position) for the QHO in the
ground state, j0i, is

b 1
hQ0 i ¼ h0j Qj0i ¼ pffiffiffi h0jðb a{ Þj0i ¼ 0
aþb ð2:105Þ
2

since b a{ ¼ ðb
aj0i ¼ 0 and h0jb aj0iÞ{ ¼ 0. hQ0i ¼ 0 makes sense because the average
position of a harmonic oscillator is indeed zero.
22 2 Quantum Harmonic Oscillator

Although the average position is zero, the mean squared position of a harmonic
oscillator is not zero. The average of Q2 for the QHO in the ground state is


b 2 j0i ¼ 1 h0jðb
Q0 2 ¼ h0j Q abaþb
aba{ þ b
a{ b a{ b
aþb a{ Þj0i ð2:106Þ
2

The first term h0jbabaj0i ¼ 0 since baj0i ¼ 0: The third term h0jb a{ b b
aj0i ¼ h0j Nj0i ¼
{ {
0: The fourth term h0jbaba j0i ¼ 0, since the creation operators generate the state j2i,
and j0i and j2i are orthogonal (h0j 2i ¼ 0). This leaves only one of the four terms in
Eq. (2.106), giving

 2 1
ab
Q0 ¼ h0jb a{ j0i ð2:107Þ
2

Using the commutation relation for the normal ordering gives

 1 1
a{ b
Q0 2 ¼ h0jðb a þ 1Þj0i ¼ ð2:108Þ
2 2

since h0jb a{ b b j0i ¼ 0.

a j 0i ¼ h 0 j N
Using Eqs. (2.105) and (2.108), the uncertainty of Q for the QHO in the ground
state is (recall Eq. (1.26))

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffi
 2 1
ΔQ0 ¼ Q0  hQ0 i ¼ 2
ð2:109Þ
2

The square of an uncertainty, such as (ΔQ0)2, is called the variance. Eq. (2.109) is
also called the standard deviation or root-mean-square deviation. It gives the spread
or dispersion in the possible values of Q0.
Similarly, the average or expectation value of P (dimensionless momentum) for
the QHO in the ground state is

hP0 i ¼ 0 ð2:110Þ

Again, this makes sense because the average momentum of a harmonic oscillator is
indeed zero. Although the average momentum is zero, the mean squared momentum
of a harmonic oscillator is not zero. The average of P2 for the QHO in the ground
state is

 1
P0 2 ¼ ð2:111Þ
2

Finally, the uncertainty in P for the QHO in the ground state is

2.9 Heisenberg Uncertainty Relation 23

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffi
1
ΔP0 ¼ P0  hP0 i ¼
2 2
ð2:112Þ
2

Exercise 2.5 Derive Eqs. (2.110) and (2.111).

2.9 Heisenberg Uncertainty Relation

Using Eqs. (2.109) and (2.112), we find

1
ΔQ0 ΔP0 ¼ ð2:113Þ
2

or, using Eqs. (2.34) and (2.35)

ħ
Δx0 Δp0 ¼ ð2:114Þ
2

In general, for the state jni

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffi
1
ΔQ ¼ Qn  hQn i ¼ n þ
2 2
ð2:115Þ
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffi
 2 1
ΔP ¼ Pn  hPn i2 ¼ n þ ð2:116Þ
2

and

1
ΔQΔP ¼ n þ ð2:117Þ
2

Equation (2.117) indicates that

1
ΔQΔP  ð2:118Þ
2

with the equality for the ground state (n ¼ 0). Hence, the Gaussian wavefunction
(Eq. (2.57)), corresponding to the ground state (n ¼ 0), is the state of the QHO with
minimum uncertainty.

Exercise 2.6 Derive Eqs. (2.115) and (2.116).

24 2 Quantum Harmonic Oscillator

Using Eqs. (2.34) and (2.35), we obtain

ħ
ΔxΔp  ð2:119Þ
2

Equation (2.119) is the Heisenberg uncertainty relation, familiar from introductory

quantum mechanics. Note that the ground state of the QHO (n ¼ 0) has the smallest
uncertainty permitted by the Heisenberg uncertainty relation. The uncertainty rela-
tion is consistent with the existence of a zero-point energy E0. We can never have a
harmonic oscillator with zero energy, because that would require its position and
momentum to both have the precise value of zero (according to Eq. (2.11)), in
contradiction to Eq. (2.119).

2.10 Some Important Relations

Below, we summarize the important results of this chapter for the QHO that will be
used frequently throughout this book:

b ¼b
Number operator : N a{ b
a ð2:120Þ
b jni ¼ njni, n ¼ 0,1,2, . . .
N ð2:121Þ
 
b ¼ ħω b 1
Hamiltonian, H a{baþ ð2:122Þ
2
b j ni ¼ E n j n i
H ð2:123Þ
 
1
E n ¼ ħω n þ ð2:124Þ
2
h i
Commutation relation : b a{ ¼ 1
a, b ð2:125Þ
 n
ba{
jni ¼ pffiffiffiffi j0i ð2:126Þ
n!
pffiffiffi
b
a j n i ¼ n j n  1i ð2:127Þ
pffiffiffiffiffiffiffiffiffiffiffi
a{ jni ¼ n þ 1 jn þ 1
b ð2:128Þ
b
aj0i ¼ 0 ð2:129Þ
Orthonormality : hnjmi ¼ δnm ð2:130Þ

Further information can be obtained from many introductory quantum mechanics

books [1, 2] or the Further Reading provided at the end of this book.
References 25

References

1. B.H. Bransden and C.J. Joachain, Introduction to quantum mechanics (John Wiley & Sons,
1989).
2. H.C. Ohanian, Principles of quantum mechanics (Prentice Hall, 1990).
Chapter 3
Canonical Quantization of Light

Albert Einstein (Fig. 3.1) proposed the existence of light quanta in a series of
publications beginning in 1905 [1–5], which he famously used to explain the
photoelectric effect. The “first quantization” was the quantum mechanics of particles
developed in the 1920s by Heisenberg, Schrodinger, Dirac, and others. The “second
quantization” refers to the canonical quantization of the electromagnetic field, that is,
the quantization of light. The second quantization is generally presented using the
b We leave this to the reader as a topic for independent study.
vector potential, A.
Instead, we adopt a simpler analysis here using only the electric and magnetic fields.
In this chapter, the canonical quantization of light is presented by analogy to a
quantum harmonic oscillator. The electric field operator is derived along with the
quadrature operators. The standard quantum limit is introduced.

3.1 Single Mode of Radiation

Maxwell’s equations provide a classical description of electric and magnetic fields.

Maxwell’s equations in free space (charge density ρ ¼ 0, current density J ¼ 0) are

—E¼0 ð3:1Þ
—B¼0 ð3:2Þ
∂B
—E¼ ð3:3Þ
∂t
1 ∂E
—B¼ ð3:4Þ
c2 ∂t

where c is the speed of light in vacuum.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 27

R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_3
28 3 Canonical Quantization of Light

Fig. 3.1 Albert Einstein

(Nobel Prize in Physics in
1921). (Credit: Wikimedia
Commons [6])

Fig. 3.2 Expression for the

classical electric field and
corresponding coordinate polarizaon
(unit vector)
system for a single mode
complex #

An oscillating electromagnetic field (E(r, t), B(r, t)) with angular frequency ω that
satisfies Maxwell’s equations is called a mode. The simplest possible mode is a
travelling plane monochromatic electromagnetic wave in vacuum. The latter is
called a “single mode” of radiation, which refers to an electromagnetic field of a
single angular frequency, ω, and polarization, ε. The classical electric field of a
single mode can be expressed as

Eðr, t Þ ¼ ε εðt Þ eikr þ c:c: ð3:5Þ

where E(r, t) is the field vector, ε is a unit vector describing the polarization of the
electric field, ε(t) is a complex number describing the field amplitude that includes
the time-dependence and initial phase of the electric field, k is the wavevector
(jkj ¼ k ¼ ω/c), and c.c. is the complex conjugate that makes E(r, t) real. Here,
ε is a real vector for simplicity, describing linear polarization. In general, ε is a
complex number, needed to describe circular or elliptical polarization. Figure 3.2
illustrates the single mode field and the corresponding coordinate system.
Similarly, the magnetic field of a single mode is given by
3.1 Single Mode of Radiation 29

kε
Bðr, t Þ ¼ εðt Þ eikr þ c:c: ð3:6Þ
ω

E, B and k are mutually orthogonal (transverse waves), and E and B are in phase.

Exercise 3.1 Check that Eqs. (3.5) and (3.6) satisfy Maxwell’s equations if
the time-dependence is given by ε(t) ¼ eiωt.

For simplicity, consider a single mode of radiation propagating in the x direction

with polarization ε ¼ y. Equations (3.5) and (3.6) simplify to

Eðr, t Þ ¼ y εðt Þ eikx þ c:c: ð3:7Þ

1
Bðr, t Þ ¼ z εðt Þ eikx þ c:c: ð3:8Þ
c

Here, y and z are unit vectors in the y and z directions, respectively. By definition,
—  B is given by
     
∂Bz ∂By ∂Bx ∂Bz ∂By ∂Bx
—  B=  xþ  yþ  z ð3:9Þ
∂y ∂z ∂z ∂x ∂x ∂y

Substituting Eq. (3.8) gives


ik  
—  B = ð0  0Þx þ 0  εðt Þe þ c:c: y þ ð0  0Þz
ikx
ð3:10Þ
c

or, using k ¼ ω/c:


ω
—  B ¼ i 2 E ð3:11Þ
c

Comparing Eq. (3.11) with Eq. (3.4) gives

∂E
¼ iωE ð3:12Þ
∂t

or, since the time-dependence is contained entirely in ε(t), Eq. (3.12) gives

dεðt Þ
¼ iωεðt Þ ð3:13Þ
dt

with the solution:

30 3 Canonical Quantization of Light

εðt Þ ¼ ε1 eiðωtþφÞ ð3:14Þ

where φ is a phase determined by the initial conditions, and ε1 is the field amplitude.
The superscript “1” reminds us that the field is a single mode (single frequency, ω).
Note that the positive exponential, ε(t)¼ ε1eþi(ωt þ φ), is also a solution to Eq. (3.13),
although Eq. (3.14) is usually adopted by convention and describes a wave travelling
in the þx direction in Eq. (3.7).

3.2 Quadrature Components

Rather than writing ε(t) with an amplitude and phase, as in Eq. (3.14), we may write
it as a complex number with a real and imaginary component:

1
εðt Þ ¼ iε1 pffiffiffi ðQ þ iPÞ ð3:15Þ
2

where Q and P are dimensionless real numbers. The i and p1ffiffi2 in Eq. (3.15) are present
by convention. ε1 is a constant with units of electric field. Q and P are real numbers
and are dynamical variables describing the time-dependence of the field. Equation
(3.15) is summarized in Fig. 3.3.
Substituting the complex field amplitude, Eq. (3.15), into Eq. (3.5) gives
 
1 1 ikr
EðrÞ ¼ iεε pffiffiffiðQ þ iPÞe  pffiffiffi ðQ  iPÞe
1 ikr

2 ikr  2 ikr 
pffiffiffi 1 e  eikr e þ eikr ð3:16Þ
¼ ε 2ε Q P
2i 2
pffiffiffi 1
¼ ε 2ε ½Q sin ðkrÞ þ P cos ðkrÞ

where we have used the Euler relation. Alternatively, we may write

pffiffiffi 1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
EðrÞ ¼ ε 2ε Q2 þ P2 sin ðk  r þ φÞ ð3:17Þ

where φ ¼ tan1(P/Q). Hence, according to Eq. (3.16), Q and P are two components
of the field that are out of phase by π/2. Q and P are known as the quadrature

Fig. 3.3 Expression for the

complex amplitude of a
classical single mode field

Real #, Complex #,
units of electric field dimensionless
3.3 Classical Hamiltonian 31

components of the field. The field associated with a single mode of radiation may
be described by two independent components—its magnitude and phase in
Eq. (3.5) or, alternatively, by its two quadrature components (Q and P) in Eq. (3.16).

Exercise 3.2 Show Eqs. (3.16) and (3.17) are equivalent.

3.3 Classical Hamiltonian

The classical expression for the energy of an electromagnetic field is

Z  
1 2 1 2
H¼ Eo jEj þ jBj dV ð3:18Þ
2 μo
V

where Eo is the permittivity of free space, μo is the permeability of free space, and
V is the volume of integration. Eo and μo satisfy the relation c ¼ (Eoμo)1/2 where
c is the speed of light in vacuum. Here, E2o jEj2 and 2μ1 jBj2 is the energy density
o
(energy per unit volume) of the electric and magnetic fields, respectively. Using
Eqs. (3.5) and (3.6), the electric field and magnetic field contributions to the
energy are equal, giving
Z
H ¼ Eo jEj2 dV ð3:19Þ
V

which is a well-known result for classical electromagnetic waves. Equation (3.17)

gives
Z
2 
H ¼ 2Eo ε1 Q2 þ P2 sin 2 ðk  r þ φÞdV ð3:20Þ
V

We need to choose a volume for the integration of Eq. (3.20). The volume V could be
a real volume (light confined in a cavity by two mirrors), the finite volume associated
with a wavepacket or pulse of light, or waves confined in a fictitious box with
periodic boundary conditions. The latter is a commonly used trick in quantum optics
(indeed, throughout physics) and is discussed below.
To evaluate the integral in Eq. (3.20), we imagine the light confined in a fictitious
box with side length L. The traveling wave satisfies periodic boundary conditions for
the field, E(x ¼ 0) ¼ E(x ¼ L ), as shown for a few modes in Fig. 3.4 (similarly for
the y and z directions). We imagine the field approaching the right side of the box and
“wrapping around” to the left side of the box. We can approach a continuous range
of wavelengths by choosing a very large box (large L ). With these boundary
32 3 Canonical Quantization of Light

Fig. 3.4 Periodic boundary

conditions for the field along the
x, y, or z direction, leading to
quantized values of the
wavevector kx ¼ 2π L nx , k y ¼
L ny and k z ¼ L nz where nx, y, z
2π 2π

is an integer

conditions, the integration of sin2(k  r þ φ) over the volume of the box becomes 12,
giving
2 
H ¼ Eo V ε1 Q2 þ P2 ð3:21Þ

where V ¼ L3 is the volume of the box.

By analogy to Eq. (2.36), we can think of a single mode field as a harmonic
oscillator where we found

ħω 2 
H¼ Q þ P2 ð3:22Þ
2

Equating Eqs. (3.21) and (3.22) gives

rffiffiffiffiffiffiffiffiffiffi
ħω
ε ¼
1
ð3:23Þ
2Eo V

ε1 is called the “one photon amplitude of the mode with frequency ω”. The concept
of the photon will be introduced later.

3.4 Canonical Quantization

Now that we have the classical Hamiltonian, Eq. (3.22), we can perform the
canonical quantization presented in Chap. 1. Let us define the canonical variables,
pffiffiffi
q¼ ħQ ð3:24Þ
pffiffiffi
p¼ ħP ð3:25Þ

The Hamiltonian in Eq. (3.22) becomes

ω 2 
H¼ q þ p2 ð3:26Þ
2

The Hamilton equations give

3.4 Canonical Quantization 33

dq ∂H
¼ ¼ ωp ð3:27Þ
dt ∂p
dp ∂H
¼ ¼ ωq ð3:28Þ
dt ∂q

Equations (3.27) and (3.28) are two coupled differential equations describing the
dynamics for light. Do they give the expected dynamics described by Eq. (3.13)? Let
us combine the two equations into a single differential equation. Multiplying
Eq. (3.28) by i and adding Eq. (3.27) gives

d
ðq þ ipÞ ¼ iωðq þ ipÞ ð3:29Þ
dt

or, from Eqs. (3.15), (3.24) and (3.25):

d
εðt Þ ¼ iω εðt Þ ð3:30Þ
dt

which is identical to Eq. (3.13). Thus, q and p give the correct dynamical equation.
According to the canonical quantization procedure, b q and b
p are quantum operators
and are canonically conjugate variables. As canonically conjugate variables, b
q and b
p
obey the commutation relation:

½b
q, b
p ¼ iħ ð3:31Þ

and, from Eq. (3.26), we can write the quantum Hamiltonian as


b ¼ω b
H q2 þ b
p2 ð3:32Þ
2

b in terms of the
Equivalently, using Eqs. (3.24) and (3.25), we can express H
b and P:
dimensionless operators, Q b

2 
b ¼ ħω Q
H b þP
b2 ð3:33Þ
2

and
h i
b P
Q, b ¼i ð3:34Þ

b and P
Here, the operators Q b are analogous to Eqs. (2.34) and (2.35) for the quantum
harmonic oscillator (QHO). However, Q b and Pb do not correspond to position and
b b
momentum of a photon! Here, Q and P correspond to two components of the electric
field that are π/2 out of phase (quadrature components of the field), analogous to
position and momentum (bx, bp) in a mechanical oscillator.
34 3 Canonical Quantization of Light

At this point, we can recall the earlier definition of b a{ and the commutation
a and b
relation from Chap. 2:


1 b b
a ¼ pffiffiffi Q
b þ iP ð3:35Þ
2


1 b b
a{ ¼ pffiffiffi Q
b  iP ð3:36Þ
2
h i
b a{ ¼ 1
a, b ð3:37Þ

Using Eqs. (3.35), (3.36) and (3.37) in Eq. (3.33), we obtain


b ¼ ħω b 1
H a{b
aþ ð3:38Þ
2

which is the familiar Hamiltonian for the QHO. A more rigorous derivation of
Eq. (3.38) using the vector potential and the normal modes of the field in a finite
volume gives the same result (for more information on the latter approach, refer to
the Further Reading provided at the end of this book).
Recalling the classical field, we have

Eðr, t Þ ¼ ε εðt Þ eikr þ c:c: ð3:39Þ

with

1
εðt Þ ¼ iε1 pffiffiffi ðQ þ iPÞ ð3:40Þ
2

Thus,
 
1 1
Eðr, t Þ ¼ iεε
1
pffiffiffi ðQ þ iPÞeikr  pffiffiffi ðQ  iPÞeikr ð3:41Þ
2 2

According to the canonical quantization, the quantum field is obtained by replacing

b and P:
the classical variables, Q and P, with the quadrature operators, Q b


 
b ðrÞ ¼ iεε1 p1ffiffiffi Q
E b eikr  p1ffiffiffi Q
b þ iP b  iP
b eikr ð3:42Þ
2 2

or, using Eqs. (3.35) and (3.36):

3.5 Time-Dependence 35


b ðrÞ ¼ iεε1 b
E a{ eikr
aeikr  b ð3:43Þ

Note that the time-dependence is absent from Eqs. (3.42) and (3.43), which is
discussed in the next section.

3.5 Time-Dependence

According to the “Schrodinger picture”, the states evolve in time, while the operators
are time-independent. In this viewpoint, the time-dependence of E b is absent, because
b is an operator:
E


b ðrÞ ¼ iεε1 b
E a{ eikr
aeikr  b ð3:44Þ

The time-dependence is contained in the state, not in the operator:

jψðt Þi ¼ jψð0Þi eiωt ð3:45Þ

The physical observable is

b ðrÞjψðt Þi
hEðr, t Þi ¼ hψðt ÞjE ð3:46Þ

Note that for stationary states jψ(t)i ¼ jψ(0)i eiωt while hψ(t)j¼ hψ(0)jeþiωt, so the
time-dependence in Eq. (3.46) cancels out.
Alternatively, according to the “Heisenberg picture”, the operators are time-
dependent while the states are time-independent and keep their initial value at
some time t0:

b ðr, t Þ ¼ iεε1 ðb
E a{ eiðkrωtÞ Þ
aeiðkrωtÞ  b ð3:47Þ

The physical observable is

b ðr, t Þjψðt 0 Þi
hEðr, t Þi ¼ hψðt 0 ÞjE ð3:48Þ

As we will see later, the Heisenberg viewpoint is useful in some circumstances, for
example, when describing photodetection events at two different times.
36 3 Canonical Quantization of Light

3.6 Quadrature Operators

Repeating the results obtained above, we have the quantum description of the field:


b ðrÞ ¼ iεε1 b
E aeikr  b a{ eikr


 
1 b b eikr  p1ffiffiffi Qb  iP
b eikr
¼ iεε1 pffiffiffi Q þ iP
2 ikr  2 ikr  ð3:49Þ
pffiffiffi e  eikr e þ eikr
¼ εε 2 Q
1 b P b
h 2i i 2
p ffiffi
ffi
¼ εε1 2 Q b sin ðkrÞ þ Pb cos ðkrÞ

b and P
Q b describe the two components of the field that are π out of phase.
2
b and P:
Recall the definitions of Q b


b ¼ p1ffiffiffi b
Q a{
aþb ð3:50Þ
2


b¼p i
P ffiffiffi b a{
ab ð3:51Þ
2

Using Eqs. (3.50) and (3.51), the expectation values (average) of Q b and P
b in the
QHO state, jni, are easily evaluated using the results from Chap. 2:

1
hQi ¼ pffiffiffi hnjðb a{ Þjni ¼ 0
aþb ð3:52Þ
2
i
hPi ¼ pffiffiffi hnjðb a{ Þjni ¼ 0
ab ð3:53Þ
2

Exercise 3.3 Derive Eqs. (3.52) and (3.53).

The expectation value of Q2 is

  1
aþb
Q2 ¼ hnjðb a{ Þðb
aþba{ Þjni
2 ð3:54Þ
1
ab
¼ hnjðbaþb a{ þ b
ab a{ b
aþb a{ b
a{ Þjni
2

Examining each term, we have hnjb

ab a{b
ajni ¼ 0 and hnjb a{ j ni ¼ 0, giving
3.6 Quadrature Operators 37

  1
ab
Q2 ¼ hnjðb a{ þ b
a{ b
aÞjni ð3:55Þ
2

Using the commutation relation, ½b a{  ¼ 1, gives

a, b

  1
a{ b
Q2 ¼ hnjð1 þ 2b aÞjni
2 ð3:56Þ
1
¼nþ
2

Similarly,

  1
P2 ¼ n þ ð3:57Þ
2

The uncertainties are

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffi
 2 1
ΔQ ¼ Q  hQ i ¼ n þ
2
ð3:58Þ
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffi
 1
ΔP ¼ P2  hPi2 ¼ n þ ð3:59Þ
2

which is the same as Eqs. (2.115) and (2.116).

The product of uncertainties is

1
ΔQΔP ¼ n þ ð3:60Þ
2

which is identical to Eq. (2.117). According to Eq. (3.60), the state j0i with n ¼ 0 is
the minimum uncertainty state with ΔQΔP ¼ 12. Thus,

1
ΔQΔP  ð3:61Þ
2

which is identical to Eq. (2.118). Equation (3.61) is known as the standard quantum
limit (SQL).
We can represent hQi and hPi in Fig. 3.5. According to Eqs. (3.52) and (3.53),
Q and P are centred at the origin in Fig. 3.5, because their average is zero. However,
according to Eqs. (3.58) and (3.59), there is a spread (uncertainty) in Q and P,
represented by the gray circle with area approximated by ΔQΔP, in accordance with
Eq. (3.61). Note, however, that the distribution represented by the gray circle is
actually a Gaussian distribution according to Eq. (2.57) for the state j0i. Q and P do
not commute and cannot be measured simultaneously, so a series of different
measurements are needed to determine Q and P. The circle represents what would
be obtained after plotting many measurements of Q and P.
38 3 Canonical Quantization of Light

Fig. 3.5 Quadrature

representation of the
quantum harmonic oscillator
in state j0i

3.7 Physical Observables

As mentioned in Chap. 2, b a and ba{ are not Hermitian, meaning they do not
correspond to any physical observable. However, the field E, b number operator N b,
b b b
Hamiltonian H, and quadratures Q and P are Hermitian, although b a and b{
a are not.
b N
Thus, E, b , H,
b Q b and P b correspond to physical observables. For example, at low
frequencies (e.g., radio), antennas can directly detect the electric field. Photodetec-
tors measure light intensity proportional to the number of photons, given by the
number operator N b ¼b a{ b
a (photon-resolving detectors also exist that can determine
the number of photons in a state). Bolometers measure field energy from a rise in
temperature associated with the Hamiltonian, H.b In Chap. 12, we will see that Qb and
b can be measured by a technique called homodyne detection.
P

b N,
Exercise 3.4 Show that E, b H,
b Qb and P
b are Hermitian.

Another Hermitian operator is the momentum, b

p. The momentum operator can be
written as

b a{ b
p ¼ ħkb a ð3:62Þ

b
p applied to the state, jni, gives

b a{b
pjni ¼ ħkb ajni ð3:63Þ
¼ ħknjni

Thus, jni is an eigenstate of b

p with eigenvalue nħk.
References 39

Fig. 3.6 Gilbert Lewis.

(Credit: AIP Emilio Segrè
Visual Archives,
photograph by Francis
Simon)

3.8 Photons

The concept of photon emerges from the Dirac formalism applied to the electro-
magnetic field. jni corresponds to a state with energy nħω above the vacuum energy
and momentum nħk. Thus, a photon is an elementary excitation of the quantized
electromagnetic field that resembles a particle with energy ħω and momentum ħk.
The state jni contains n photons. The annihilation operator, b
a, destroys a photon and
{
b
the creation operator, a , creates a photon. Gilbert Lewis (Fig. 3.6), an American
chemist, coined the term “photon” in 1926 [7]. In general, a quantized field allows
particle creation and annihilation, and is the foundation for quantum field theory.

References

1. A. Einstein, Ann. Phys., Lpz. 17 (1905) 132.

2. A. Einstein, Ann. Phys., Lpz. 20 (1906) 199.
3. A. Einstein, Ann. Phys., Lpz. 22 (1907) 180.
4. A. Einstein, Phys. Z. 10 (1909) 185.
5. A. Einstein, Phys. Z. 10 (1909) 817.
6. File: Einstein 1921 by F Schmutzer – restoration.jpg. (2020, September 26). Wikimedia Com-
mons, the free media repository. Retrieved 10:26, December 7, 2020 from https://commons.
wikimedia.org/w/index.php?title¼File:Einstein_1921_by_F_Schmutzer_-_restoration.jpg&
oldid¼471895083
7. G.N. Lewis, The conservation of photons, Nature 118 (1926) 874.
Chapter 4
Fock States and the Vacuum

In this chapter, we further examine the properties of the Fock or number state, jni.
The average photon number and average electric field of a Fock state are derived,
along with their uncertainty. We show that the Fock state has the unusual property of
having a precise number of photons, n, but with average electric field of zero.
Although the average field is zero, there is an uncertainty or fluctuation in the
field. The properties of the Fock state j0i, called the vacuum, are presented along
with the concept of vacuum field fluctuations and their experimental consequences.

4.1 Photon Number

The average number of photons (expectation value of n) in Fock state jni is

b
hni ¼ hnj Njni a{b
¼ hnjb ajni ¼ n ð4:1Þ

b operator twice:
The expectation value of n2 is obtained by applying the N
 2
b 2 jni ¼ nhnj Njni
n ¼ hnjN b ¼ n2 ð4:2Þ

The uncertainty in n is
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Δn ¼ hn2 i  hni2 ¼ n2  ðnÞ2 ¼ 0 ð4:3Þ

Thus, the Fock state jni has a definite number of photons, n, with zero uncertainty.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 41

R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_4
42 4 Fock States and the Vacuum

Exercise 4.1 We could write Nb 2 as b

a{b
aba{ b a{b
a. Show that if we put b a{b
ab a in the
{ {
ab
normal order, we get hnjb ab
abajni ¼ n , the same as Eq. (4.2).
2

4.2 Electric Field of the Fock State

Recall from Chap. 3 that the electric field operator is

b ðrÞ ¼ iεε1 ðb
E a{ eik ∙ r Þ
aeik ∙ r  b ð4:4Þ

b in the state jni is

The average (expectation value) of E

b
hEi ¼ hnjEjni
a{ eikr Þjni
aeikr  b
¼ iεε1 hnjðb ð4:5Þ
¼ 0
pffiffiffi
since b ajni ¼ n jn  1i, and hnj n  1i ¼ 0 because these are orthogonal states.
Similarly, hnjba{ jni ¼ 0. Thus, we have the unusual situation where the Fock state,
jni, has a precise number of photons, n, but the average electric field is zero. This
tells us that Fock states are a nonclassical form of light. Due to their nonclassical
nature, Fock states are very difficult to produce. In the next chapter, we will examine
the single photon state, that is, the Fock state with n ¼ 1, including methods to
produce this state.
Although the average electric field of a Fock state is zero, the field can still
fluctuate about the average. The average of the square of the field (expectation value
of Eb 2 ) is

 
E2 ¼ hnjE b 2 jni
a{ eikr Þðb a{ eikr Þjni ð4:6Þ
2
aeikr  b
¼ ði  iÞðε1 Þ hnjðb aeikr  b
1 2 2 2ikr { { { 2 2ikr
¼ ðε Þ hnjðb
a e b
aba b
ab a Þ e
a þ ðb Þjni

Keeping only the non-zero terms in Eq. (4.6), we get

   2
ab
E2 ¼ ε1 hnjðb a{ þ b
a{ b
aÞjni ð4:7Þ

Putting Eq. (4.7) into the normal order (or, alternatively, using Eqs. (2.127) and
(2.128)) gives
4.3 Vacuum Fluctuations 43

   2
E2 ¼ ε1 ð2n þ 1Þ ð4:8Þ

The field fluctuations are described by the uncertainty:

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
ΔE ¼ E2  hEi2
ð4:9Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ ε1 2n þ 1

Of course, there are corresponding fluctuations in the magnetic field too. Assuming
we could repeat a field measurement multiple times on identically prepared Fock
states, we would obtain a distribution in the measurement result according to
Eq. (4.9), and an average of zero according to Eq. (4.5). Equation (4.9) tells us
that the field fluctuations increase as the number of photons increases.

4.3 Vacuum Fluctuations

Classically, we think of the vacuum as an absence of everything. Quantum mechan-

ically, this is not true. For n ¼ 0, that is, the vacuum state j0i, we have ΔE ¼ ε1.
Thus, j0i is a minimum uncertainty state with the least uncertainty in the electric
field. The j0i state also corresponds to the minimum uncertainty of the quadratures
where ΔQΔP ¼ 12 with equal uncertainty in Q and P. We can summarize the
vacuum properties as follows:

j0i : vacuum state with n ¼ 0 photons ð4:10Þ

b j 0i ¼ 0
h ni ¼ h 0 j N ð4:11Þ
   
1 1 1
Vacuum energy : E 0 ¼ ħω n þ ¼ ħω 0 þ ¼ ħω ð4:12Þ
2 2 2
D E
Average electric field hEi ¼ 0jEj0 b ¼0 ð4:13Þ

Field fluctuations : ΔE ¼ ε1 ð4:14Þ

1
ΔQΔP ¼ , minimum uncertainty state ð4:15Þ
2

There are no photons in the vacuum and, on average, there is no electric field in the
vacuum either. However, there are fluctuations, ε1, in the field. These field fluctua-
tions have quadrature values at the minimum allowed by the uncertainty relation,
ΔQΔP ¼ 12. We can think of “empty” space as filled with these field fluctuations.
44 4 Fock States and the Vacuum

4.4 Experimental Evidence of Vacuum Fluctuations

Are vacuum fluctuations real or are they just a figment of the physicist’s imagina-
tion? The experimental consequence of hni ¼ 0 and hEi ¼ 0 for the vacuum state is
that we cannot directly photodetect the vacuum. However, we have an abundance of
indirect evidence for vacuum fluctuations, including spontaneous emission, the
Lamb shift, the magnetic moment of the electron, the Casimir effect, and the van
der Waals force. We consider each of these below.
In quantum mechanics, an energy eigenstate of the Hamiltonian is a stationary
state. This means that an electron in an excited state (an eigenstate of the Hamilto-
nian) should remain there forever, in the absence of any external perturbations.
Spontaneous emission, involving the transition of an electron from an excited to the
ground state, is explained by the perturbation on the electron from the field fluctu-
ations of the vacuum. Aspects of spontaneous emission are described in Chaps. 9
and 19.
The Lamb shift, named after Willis Lamb (Fig. 4.1a), is a correction to certain
energy levels of the hydrogen atom. This effect is due to the perturbation of the
electron by the field fluctuations of the vacuum. The electric field fluctuations of the
vacuum cause a rapid oscillation of the electron in the Coulomb potential of the
hydrogen atom, causing a slight shift in the energy levels of the s orbitals. A

(a) (c)

(b) 2 2
3/2 3/2
2 1/2
2 1/2 2 1/2 2 1/2

1 1/2 1 1/2

Bohr Dirac QED

Fig. 4.1 (a) Willis Lamb (Nobel Prize in Physics in 1955). (Credit: AIP Emilio Segrè Visual
Archives, W. F. Meggers Gallery of Nobel Laureates Collection). (b) Energy levels according to
Bohr, Dirac, and quantum electrodynamics (QED). The Lamb shift is the increase of the 2s1/2 level
according to QED. (c) Original experimental data of the Lamb shift by Lamb and Retherford. The
energy levels vary in a magnetic field due to the Zeeman effect. The experimental data (circles) are
shifted down by 1000 Megacycles for easier comparison with the theory (solid lines). (Reprinted
with permission from Lamb Jr. and Retherford [1], https://doi.org/10.1103/PhysRev.72.241.
Copyright 1947 by the American Physical Society)
4.4 Experimental Evidence of Vacuum Fluctuations 45

derivation of the Lamb shift is given in Appendix 1. The energy levels of the
hydrogen atom according to Bohr, Dirac, and quantum electrodynamics (QED) are
shown in Fig. 4.1b. According to the simple hydrogenic model by Bohr, the energy
levels of the hydrogen atom depend only on the principal quantum number, n, and
not on the orbital angular momentum quantum number l (l ¼ 0 for the s orbital and
l ¼ 1 for the p orbital). Dirac extended the Bohr theory to include the effects of
electron spin (spin–orbit coupling). In Dirac theory, the 2s1/2 (n ¼ 2, l ¼ 0, j ¼ 12 )
and 2p1/2 (n ¼ 2, l ¼ 1, j ¼ 12) energy levels are degenerate (the subscript indicates
the total angular momentum quantum number, j, which includes orbital angular
momentum and electron spin). Finally, QED includes the effect of the vacuum field
fluctuations. The increase of the 2s1/2 energy level predicted by QED is called the
Lamb shift, which removes the degeneracy between the 2s1/2 and 2p1/2 energy levels.
The energy splitting is about 1058 MHz, in the microwave frequency range. The
Lamb shift in the hydrogen spectrum was first measured in 1947 by Lamb and
Retherford (Fig. 4.1c) [1] using microwave resonance. In 1947, Hans Bethe
explained the Lamb shift in terms of the vacuum fluctuations [2]. These experimental
and theoretical efforts laid the foundation for QED.
The magnetic moment associated with the spin of the electron is given by
μ ¼  gμBms where μB is the Bohr magneton and ms is the spin quantum number
(ms ¼  12 for the electron). g, called the “spin g-factor”, is a correction factor
predicted to have a value of 2 from the relativistic Dirac equation. However, the
spin g-factor has been measured and has the actual value of 2.00231930436182. The
small deviation from 2 is due to the effect of vacuum fluctuations on the electron
magnetic moment, which can be calculated in QED. The theoretical prediction
agrees with the experimentally measured value to about 1 part in a billion, making
QED one of the most accurate physical theories that exists!
The vacuum energy can even have an influence on macroscopic systems. The
Casimir effect, introduced in 1948 by Hendrik Casimir (Fig. 4.2a) [3], is a physical
force associated with the vacuum field fluctuations. If two parallel neutral
conducting plates are held closely together in vacuum, a force (called the Casimir
force) pushes the two plates together. In classical electromagnetism, the force
between neutral plates is zero. The force arises quantum mechanically due to a
restriction in the allowed mode frequencies of the vacuum state between the plates
due to the electromagnetic boundary conditions imposed by the plates. The electric
field at the conducting plates must go to zero, according to Maxwell’s equations for a
perfect conductor. Thus, only certain discrete wavelengths can exist between the
plates, as shown in Fig. 4.2b. Specifically, the wavevector perpendicular to the plate
surface can take only certain discrete values of kz ¼ nπ/d where n ¼ 1, 2, 3, . . .,
corresponding to standing waves between the plates. Therefore, there is a reduction
in the number of allowed vacuum modes between the plates. As the separation
between the plates increases, the number of allowed modes between the plates also
increases, and hence, the vacuum energy between the plates increases. This tells us
that the energy is minimized when the plates have zero separation, meaning there is a
mutual attraction between the two plates. The difference in zero-point energy, ΔE,
for the continuum of modes versus those between the plates can be calculated, and
then the Casimir force per unit area of the plates as a function of plate separation
46 4 Fock States and the Vacuum

(a) (c)

(d)

(b)

Fig. 4.2 (a) Hendrik Casimir. (Credit: AIP Emilio Segrè Visual Archives, Physics Today Collec-
tion). (b) Discretization of the vacuum mode wavelengths between parallel conducting plates
leading to the Casimir effect. (c) Experimental method of measuring the Casimir force using a
plate–sphere geometry [8]. (d) Experimental data of Casimir force versus plate–sphere separation.
The solid line is the theoretical prediction [8]. ((c) and (d) are reprinted with permission from
Mohideen and Roy [8], https://doi.org/10.1103/PhysRevLett.81.4549. Copyright 1998 by the
American Physical Society)

d can be determined from F ¼  ∂ΔE ∂d

. An equivalent view is that the radiation
pressure from the vacuum is greater outside the plates than between the plates,
causing a force that pushes the plates together. An analogy is the attractive force
between two closely spaced ships at sea due to the restricted standing waves between
them [4]. The resulting attractive force per unit area of the plates falls off sharply
with separation according to [3]

π 2 ħc 1
F¼ ð4:16Þ
240 d4

A simple heuristic derivation of the 1/d4 force dependence is given in Appendix 2. If

d ¼ 1 μm, then the Casimir force is only 107 N/cm2. However, the force rises
rapidly with decreasing plate separation. Thus, if d ¼ 1 nm, then the force is 105
N/cm2. Casimir stated in his 1948 publication [3], “although the effect is small, an
4.4 Experimental Evidence of Vacuum Fluctuations 47

(a) (b)

Fig. 4.3 (a) Edward Mills Purcell (Nobel Prize in Physics in 1952). (Credit: Wikimedia Commons
[13]). (b) Enhancement in spontaneous emission from an atom in a cavity (right) compared to free
space (left). (Reprinted by permission from Springer Nature, Vahala [14]. Copyright 2003)

experimental confirmation seems not unfeasible and might be of a certain interest.”

A decade later, in 1958, M.J. Sparnaay experimentally measured the force between
two plates in qualitative agreement with Casimir’s prediction [5]. Nearly 50 years
after Casimir’s prediction, S.K. Lamoreaux [6, 7] and others [8], beginning in 1997,
used modern experimental techniques such as piezoelectric transducers to control the
plate positions very accurately. The plates were also replaced with more convenient
geometries, such as a plate and a sphere (Fig. 4.2c), since it is difficult to keep two
large plates parallel with small separation. These experiments confirmed the theory
to about 1% accuracy (Fig. 4.2d). Thus, the Casimir force is more than just a
theoretical curiosity. It is a real effect in microelectromechanical systems and
nanotechnology [9].
The Purcell effect, similar in some respects to the Casimir effect, is an enhance-
ment of the spontaneous emission from an atom placed inside a resonant cavity. In
the 1940s, Edward Mills Purcell (Fig. 4.3) discovered the effect named after him
[10]. The cavity modifies the density of modes at the transition frequency of the
atom, thereby modifying the spontaneous emission rate of the atom. An enhance-
ment of the spontaneous emission rate is observed if the standing electric field in the
cavity is resonant with the electronic transition energy of the atom. The enhancement
factor, called the Purcell factor, is the ratio of the mode density in the cavity to that in
vacuum and is given by [10]

3Qλo 3
F¼ ð4:17Þ
4π 2 V

where Q is the quality factor of the cavity, λo is the wavelength in the cavity, and V is
the mode volume inside the cavity. The quality factor is a measure of the “sharpness”
of the cavity mode and is given by Q ¼ λo=Δλ, where Δλ is the linewidth of the cavity
48 4 Fock States and the Vacuum

mode. If the cavity dimensions are on the order of the field wavelength, V~λo3, then
the Purcell factor is on the order of Q, which can be many orders of magnitude large
(e.g., Q values on the order of 10–100 are observed in semiconductor cavities [11]).
It is also possible to suppress the spontaneous emission of the atom if there are no
mode frequencies in the cavity that match the transition frequency of the atomic
emission. The atom cannot emit a photon because the cavity contains no mode with a
matching frequency to accept it. Thus, spontaneous emission is not a property of the
atom, but can be controlled by the atom’s environment. This is of both fundamental
and practical significance for the engineering of light sources [12].
Finally, the van der Waals force is an attractive force between two neutral atoms
in a vacuum. The electrostatic attraction between two neutral atoms can be explained
by the creation of induced electric dipole moments in the atoms by the vacuum field
fluctuations.

Exercise 4.2 The Casimir effect is a force produced from vacuum fluctua-
tions. Could free energy be harnessed from the vacuum?

References

1. W.E. Lamb, Jr. and R.C. Retherford, Fine structure of the hydrogen atom by a microwave
method, Phys. Rev. 72 (1947) 241.
2. H.A. Bethe, The electromagnetic shift of energy levels, Phys. Rev. 72 (1947) 339.
3. H.B.G. Casimir, On the attraction between two perfectly conducting plates, Proc. Kon. Ned.
Akad. Wet. 51 (1948) 793.
4. S. L. Boersma, A maritime analogy of the Casimir effect, Am. J. Phys. 64 (1996) 539.
5. M.J. Sparnaay, Measurements of attractive forces between flat plates, Physica 24 (1958) 751.
6. S.K. Lamoreaux, Demonstration of the Casimir force in the 0.6–6μm range, Phys. Rev. Lett.
78 (1997) 5.
7. S.K. Lamoreaux, The Casimir force: background, experiments, and applications, Rep. Prog.
Phys. 68 (2004) 201.
8. U. Mohideen and A. Roy, Precision measurement of the Casimir force from 0.1 to 0.9 μm,
Phys. Rev. Lett. 81 (1998) 4549.
9. F. Capasso, J.N. Munday, D. Iannuzzi and H.B. Chan, Casimir forces and quantum
electrodynamical torques: Physics and nanomechanics, IEEE J. Selected Topics in Quantum
Electronics 13 (2007) 400.
10. E.M. Purcell, Spontaneous emission probabilities at radio frequencies, Phys. Rev. 69 (1946)
681 (note B10).
11. J.-M. Gérard and B. Gayral, Strong Purcell effect for InAs quantum boxes in three-dimensional
solid-state microcavities, J. Lightwave Technol. 17 (1999) 2089.
12. T.F. Krauss and R.M. De La Rue, Photonic crystals in the optical regime – past, present and
future, Progress in Quantum Electronics 23 (1999) 51.
13. https://commons.wikimedia.org/wiki/File:Edward_Mills_Purcell.jpg
14. K.J. Vahala, Optical microcavities, Nature 424 (2003) 839.
Chapter 5
Single Photon State

In the previous chapter, we examined the properties of the special Fock state, j0i. In
this chapter, another special Fock state, j1i, called the single photon state, is
presented. The quantum optics treatment of photodetection is explained. Methods
of generating and detecting single photons are described.

5.1 Single Photon State

Single photons are required for quantum communications, quantum computing,

quantum metrology, and quantum information processing based on the optical
approach. Using the results of Chap. 3, the single photon state j1i (Fock state with
n ¼ 1) has the following properties:

h ni ¼ 1 ð5:1Þ
Δn ¼ 0 ð5:2Þ
3
E 1 ¼ ħω ð5:3Þ
2
b j 0i ¼ 0
hEi ¼ h0jE ð5:4Þ
pffiffiffi 1
ΔE ¼ 3ε ð5:5Þ
3
ΔQΔP ¼ ð5:6Þ
2

According to Eqs. (5.5) and (5.6), the single photon state is not a minimum
uncertainty state. Note that the single photon energy, ħω, is the photon energy
measured above the vacuum energy, E1  E0 ¼ ħω.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 49

R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_5
50 5 Single Photon State

5.2 Photodetection Signal

The direct detection of a single photon of a Fock state jni destroys a single photon,
represented by b
ajni. Intuitively, light intensity is expected to be proportional to the
number of photons:

a{ ajni ¼ n
I / hni ¼ hnjb ð5:7Þ

Alternatively, according to classical physics, intensity measured by a photodetector

is proportional to the square of the electric field (I / E2). Quantum mechanically, we
can write the electric field as

b ðrÞ ¼ iεε1b
E b þ ðrÞ þ E
aeik ∙ r þ h:c: ¼ E b  ðrÞ ð5:8Þ
þ
where h.c. is the Hermitian conjugate. E b ðrÞ ¼ iεε1b aeik ∙ r is called the positive

frequency component of the field, and E b ðrÞ ¼ iεε b1
a{ eik ∙ r is called the negative
frequency component of the field. The positive frequency component contains the
annihilation operator, and is thus responsible for photon absorption, while the
negative frequency component contains the creation operator and is responsible
for emission of a photon. Since we expect intensity measurements at a photodetector
to be related to absorption, we can write
 þ 2
b 
I / E ðrÞjni ð5:9Þ

b  ðrÞE
¼ hn j E b þ ðrÞjni ð5:10Þ
 2
¼ ε1 hnjb a{bj
a ni ð5:11Þ
 2
¼ ε1 n ð5:12Þ

Note that the intensity is proportional to hnjb a{ b

ajni, which is equal to the photon
number as we expect intuitively from Eq. (5.7). By setting n ¼ 1 in Eq. (5.12), (ε1)2
can be interpreted as the “intensity of a single photon”, and ε1 is the “field amplitude
of a single photon”. A word of caution is warranted, however, because the interpre-
tation of ε1 as the “field amplitude of a single photon” is not quite correct, since
hEi ¼ 0 for the single photon state.
Suppose we start with a single photon state, j1i. Direct photodetection destroys
the photon, and the state becomes

b
aj1i ¼ j0i ð5:13Þ
5.3 Single Photon Sources 51

consistent with Eq. (2.127). If we measure again, the state becomes

b
aj0i ¼ 0 ð5:14Þ

according to Eq. (2.129). By definition, you cannot detect a single photon twice,
since the first photodetection event destroys the single photon.
Note that we are speaking above about direct detection of a photon, which
destroys the photon by absorption. It is sometimes desirable to perform a measure-
ment of a photon without destroying the photon; that is, we want to make a
measurement of the photon without disturbing it and then let it continue on its
way. This measurement process is known as “quantum non-demolition” (QND)
measurement. QND usually involves forming an entangled state of the particle to be
measured with another particle called the “meter particle” (entanglement is discussed
in Chap. 8). A measurement of the meter particle is performed, which tells us the
state of the measurement particle via the entanglement. The interested reader may
obtain more information on QND in Refs. [1, 2].

5.3 Single Photon Sources

The ideal single photon source will produce single photons deterministically
(on demand) with high collection efficiency, at a high repetition rate (GHz), at
room temperature, at the desired wavelength (or spectrum), and with definite polar-
ization. Ideally, the emitter can be triggered to emit single photons on demand by
optical or electrical excitation. The photons should be produced with a high quantum
yield (ideally 100%), meaning that each trigger event results in a single photon
emission with certainty. Each photon wavepacket should be created in the identical
state, so that they are indistinguishable from each other (i.e., the photons should have
identical spectra, spatial distribution, and polarization); hence, they can interfere
perfectly as required for quantum information processing applications. The indistin-
guishability of photons can be tested through a Hong–Ou–Mandel experiment,
described in Chap. 17. The source should produce single photons of high purity,
meaning that two-photon correlations are negligible (see Chap. 6 for photon corre-
lations and their measurement). Thus far, there is no single photon source that meets
all the above criteria perfectly, and the development of improved single photon
sources remains an active research topic.
The two main types of single photon source available today are radiative decay in
two-level systems or spontaneous parametric down-conversion (SPDC) in nonlinear
crystals [3]. In two-level systems, spontaneous emission from an excited state to the
ground state creates a single photon wavepacket (wavepackets are discussed in
Chap. 9). Two-level systems include atoms, ions (Fig. 5.1a), nitrogen-vacancy
(NV) centers in diamond (Fig. 5.1b) or other color centers, semiconductor quantum
dot heterostructures (Fig. 5.1c, d), and semiconductor nanocrystals (Fig. 5.1e).
52 5 Single Photon State

Fig. 5.1 Single photon sources. (a) Atomic or ionic emitters (represented by the circles). Elec-
trodes are used to confine the ions, and lasers are used to control the state of the ions and read out
their state by fluorescence emission. (b) Illustration of the crystal structure of the N-V center in
diamond (green atoms are carbon), containing a substitutional nitrogen atom and a carbon vacancy.
(c) Transmission electron microscopy image of GaAs quantum dots (QDs) formed along the length
of a GaP nanowire. Nanowire diameter is about 50 nm. (d) Cross-sectional transmission electron
microscopy image of an InGaAs quantum dot in a GaAs matrix. (e) Colloidal quantum dots
showing size-dependent fluorescence under ultraviolet light. (b, d, e). (Credit: Wikimedia
Commons [4])

Quantum dots (QDs) are structures in which electron motion is strongly confined
in all three dimensions (with energy quantization in all directions), leading to
discrete energy levels. Today, QDs may be realized by a variety of different
techniques and in a range of different materials (Fig. 5.1c–e). For example, QDs
can be formed along the length of a nanowire by epitaxial growth methods
(Fig. 5.1c), as a two-dimensional array of QDs on a surface using the Stranski–
Krastanov process (Fig. 5.1d), or as nanocrystals in colloidal solutions (Fig. 5.1e).
Spontaneous parametric down-conversion (SPDC) is a nonlinear process that can
occur in certain crystals such as potassium dihydrogenphosphate (KDP), beta barium
borate (BBO), or lithium niobate (LiNbO3). In SPDC, a pump laser occasionally
undergoes a nonlinear interaction with the crystal. This interaction produces two
lower-frequency photons (ωs, ωi), called the signal and idler, from the higher-
frequency photon (ωp) of the incident pump laser (Fig. 5.2). The frequencies and
5.4 Single Photon Detectors 53

Fig. 5.2 (a) Spontaneous (a) (b)

parametric down-
conversion (SPDC) in a
nonlinear crystal. (b) Energy
SPDC
diagram of the SPDC
process

wavevectors of the three photons satisfy the conservation of energy and momentum
(the latter is called the phase matching condition):

ωp ¼ ωs þ ω i ð5:15Þ
kp ¼ ks þ ki ð5:16Þ

The SPDC process is called degenerate if the down-converted photons have the same
frequency (ωs ¼ ωi ¼ ωp/2) and is called nondegenerate otherwise. In general, the
photons leaving the crystal propagate in different directions as illustrated in
Fig. 5.2a. Thus, the idler photon can be used as a trigger at a detector to announce
(“herald”) the arrival of the signal photon at another detector. For this reason, SPDC
is called a heralded single photon source. In addition, the photon pairs along certain
directions can be polarization entangled (entanglement is discussed in Chap. 8). A
drawback of the SPDC process is that the photons are produced by a probabilistic
process, rather than being produced on demand.

5.4 Single Photon Detectors

Remarkably, the human retina can detect single photons [5]. More practical single
photon detectors include the single photon avalanche diode (SPAD), the
photomultiplier tube (PMT), the superconducting nanowire single photon detector
(SNSPD), the electron multiplying charge-coupled device (EMCCD) camera, and
the intensified CCD (ICCD) camera [3]. Single photon detectors should have a high
detection efficiency and ideally be able to count individual photons.
A photomultiplier tube (PMT) (Fig. 5.3a) uses a photocathode to convert incident
photons into electrons. The electrons are subsequently amplified by secondary
emission at a series of dynodes resulting in electron multiplication and eventual
collection at an anode. PMTs have largely been replaced by semiconductor
photodiodes.
A single photon avalanche diode (SPAD) (Fig. 5.3b) uses the avalanche multi-
plication process in a silicon or III-V semiconductor p-n junction to convert an
incident photon into an electrical pulse.
The electron multiplying charge-coupled device (EMCCD) camera (Fig. 5.3c)
uses the avalanche process in an electron multiplying register of a charge-coupled
device (CCD) camera to amplify the CCD image [6]. The intensified CCD (ICCD)
camera (Fig. 5.3e) uses an image intensifier mounted in front of a CCD. The
54 5 Single Photon State

photocathode dynode
(a) (b)

anode
(c) (d)

(e)

Fig. 5.3 Single photon detectors. (a) Photomultiplier tube (PMT). (b) Cross-sectional illustration
of the p-n junctions in a single photon avalanche diode (SPAD). (c) Electron multiplying charge-
coupled device (EMCCD) camera. (d) Superconducting nanowire single photon detector (SNSPD),
showing the meandering superconducting nanowire. (e) Intensified CCD (ICCD) camera, showing
photocathode (gray), microchannel plate (red), and phosphor (green). (b–e). (Credit: Wikimedia
Commons [7])

intensifier consists of a photocathode to convert incident photons into electrons

(similar to the PMT). The electrons are amplified using a microchannel plate,
basically a microscopic version of the dynodes in a PMT. The electrons are then
converted by a phosphor back to photons for detection by a CCD. EMCCD and
ICCD cameras are used in quantum imaging (Chap. 18).
The superconducting nanowire single photon detector (SNSPD) (Fig. 5.3d) uses
the photon energy to break superconducting Cooper pairs, creating a resistive state in
a current-biased superconducting nanowire. The photon is detected as a voltage
pulse due to the change in resistance.

Exercise 5.1 Investigate and explain the working principles in more detail for
single photon sources and detectors.
References 55

References

1. P. Grangier, J.A. Levenson and J.-P. Poizat, Quantum non-demolition measurements in optics,
Nature 396 (1998) 537.
2. G. Nogues et al., Seeing a single photon without destroying it, Nature 400 (1999) 239.
3. M.D. Eisaman, J. Fan, A. Migdall and S.V. Polyakov, Single-photon sources and detectors, Rev.
Sci. Instrum. 82 (2011) 071101.
4. https://commons.wikimedia.org/wiki/File:Nitrogen-vacancy_center.png; https://commons.
wikimedia.org/wiki/File:Gaas_inas_quantum_dot.jpg; https://commons.wikimedia.org/wiki/
File:Quantum_Dots_with_emission_maxima_in_a_10-nm_step_are_being_produced_at_
PlasmaChem_in_a_kg_scale.jpg;
5. J.N. Tinsley et al., Direct detection of a single photon by humans, Nat. Commun. 7 (2016) 12172.
6. L. Zhang et al., A characterization of the single-photon sensitivity of an electron multiplying
charge-coupled device, J. Phys. B: At. Mol. Opt. Phys. 42 (2009) 114011.
7. https://commons.wikimedia.org/wiki/File:SPAD_Cross-section.gif; https://commons.
wikimedia.org/wiki/File:EMCCD2_color_en.svg; https://en.wikipedia.org/wiki/File:NIST_
SEM_Image_of_Superconducting_Nanowire_Single_Photon_Detector.jpg; https://commons.
wikimedia.org/wiki/File:Image_intensifier_diagram.png
Chapter 6
Single Photon on a Beam Splitter

The beam splitter is an important optical element in quantum optics experiments.

The classical and quantum treatment of the beam splitter is presented. We derive the
photodetection probabilities for a single photon on a beam splitter, including single
photon detection and double photon detection (coincidence counting). The correla-
tion function is introduced for classical and quantum light. We show that the beam
splitter creates an entangled state from a single photon input. The Hanbury Brown–
Twiss experiment is introduced for characterizing light sources.

6.1 Classical Beam Splitter

The beam splitter is an important optical element in both classical and quantum
optics experiments. As shown in Fig. 6.1, the beam splitter contains two input ports
(labelled 1 and 2) and two output ports (labelled 3 and 4), also called modes. The
beam splitter contains an interface that splits the incident electric field into a reflected
and transmitted field. The reflection coefficient, r, is the fraction of incident field that
is reflected. The transmission coefficient, t, is the fraction of incident field that is
transmitted. r and t are complex numbers, which describe the magnitude and phase
change of the incident field upon reflection and transmission.
For example, consider an electric field incident from a medium of low refractive
index onto a medium of high refractive index (this case is called “external reflec-
tion”). According to the Fresnel equations familiar from classical optics [1], the
reflected field in this situation will undergo a π phase shift (corresponding to a
change in field amplitude of eiπ ¼  1) and the transmitted field will have zero phase
shift. Conversely, if the field is incident from the opposite direction (from high to
low refractive index, called “internal reflection”), then neither the reflected nor the
transmitted field has any phase change. Thus, the beam splitter gives a π phase shift
upon reflection from one direction only. This situation can arise in a beam splitter
composed of a semitransparent metallic film on a glass substrate, for example.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 57

R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_6
58 6 Single Photon on a Beam Splitter

Fig. 6.1 A beam splitter π phase

with input fields, E1 and E2, shi
and output fields, E3 and E4,
measured by detectors D3
and D4. The dot indicates
the side with a π phase shift
of the reflected field

Throughout this book, a “dot” as in Fig. 6.1, will indicate the side of the beam splitter
that results in a π phase shift for the reflected field. It should be noted, however, that
the phase convention can vary [2, 3] depending on the technical design of the beam
splitter, and nowadays, beam splitters are usually composed of dielectric (not
metallic) films. A common phase convention is to adopt either a π phase shift in
one path of the beam splitter, or an i phase shift in both paths of the beam splitter.
Different phase conventions do not change our conclusions. Here, we adopt the π
phase shift convention.
As shown in Fig. 6.1, the classical output field, E3, is

E 3 ¼ rE1 þ tE2 ð6:1Þ

and the output field, E4, is

E 4 ¼ tE1  rE2 ð6:2Þ

where r and t are now real numbers and we have included the π phase shift in
Eq. (6.2) due to reflection from the side of the beam splitter with the “dot” as seen in
Fig. 6.1. We assume the same polarization for all fields, so we only need to consider
each field amplitude and phase (we can treat the electric field as a scalar, rather than
as a vector).
For simplicity, suppose the input beam at the top of Fig. 6.1 is absent, so E2 ¼ 0.
The output fields are then

E3 ¼ rE 1 ð6:3Þ
E4 ¼ tE 1 ð6:4Þ

and the corresponding intensities measured at the detectors, D3 and D4, are propor-
tional to the square modulus of the fields, jE3j2 and jE4j2:

I 3 / jE3 j2 ¼ jr j2 jE1 j2 ¼ RjE 1 j2 ð6:5Þ

I 4 / jE 4 j2 ¼ jt j2 jE 1 j2 ¼ T jE 1 j2 ð6:6Þ
6.2 Quantum Beam Splitter 59

where R ¼ jrj2 and T ¼ jtj2 are called the reflectance and transmittance, respectively.
They refer to the fraction of light intensity, which is reflected or transmitted (rather
than r and t, which refer to the fields). Adding Eqs. (6.5) and (6.6) gives

jE3 j2 þ jE 4 j2 ¼ ðR þ T ÞjE 1 j2 ð6:7Þ

Due to the conservation of energy,

jE 3 j2 þ jE 4 j2 ¼ jE 1 j2 ð6:8Þ

That is, the input and output intensities must match. Comparing Eqs. (6.7) and (6.8),
we must have

RþT ¼1 ð6:9Þ

which is just another way of stating the conservation of energy. For the general case
with both inputs present (E1 6¼ 0, E2 6¼ 0), we must have

jE3 j2 þ jE 4 j2 ¼ jE 1 j2 þ jE 2 j2 ð6:10Þ

This results in the general condition:

rr  þ tt  ¼ 1 ð6:11Þ
 
rt  tr ¼ 0 ð6:12Þ

where r and t are the complex conjugate of r and t, respectively. Equation (6.11) is
equivalent to Eq. (6.9).

Exercise 6.1 Derive Eqs. (6.11) and (6.12), referring to Fig. 6.1.

6.2 Quantum Beam Splitter

The quantum approach to the beam splitter, first introduced in 1987 [4–6], replaces
the classical fields with the corresponding quantum operators, as shown in Fig. 6.2.
The fields are

E b1 þ tE
b 3 ¼ rE b2 ð6:13Þ
b 4 ¼ tE
E b 1  rE
b2 ð6:14Þ

analogous to Eqs. (6.1) and (6.2).

60 6 Single Photon on a Beam Splitter

Fig. 6.2 A beam splitter

with quantum operators for
the input and output fields

Fig. 6.3 A beam splitter

with annihilation operators
corresponding to the input
and output fields

Since the field operators are related to the annihilation operators (Eq. (3.43)), the
annihilation operators are also given by the reflection and transmission coefficients
in the same manner as the fields, as shown in Fig. 6.3:

b
a3 ¼ rb
a1 þ tb
a2 ð6:15Þ
b
a4 ¼ tb
a1  rb
a2 ð6:16Þ

Taking the Hermitian conjugate of Eqs. (6.15) and (6.16) gives

a{3 ¼ r b
b a{1 þ t b
a{2 ð6:17Þ

a{4 ¼ t b
b a{1  r b
a{2 ð6:18Þ

or, since r and t are real numbers in our case:

a{3 ¼ rb
b a{1 þ tb
a{2 ð6:19Þ

a{4 ¼ tb
b a{1  rb
a{2 ð6:20Þ

Note that Eqs. (6.15) and (6.16) can be summarized in matrix form:
    
b
a3 r t b
a1
¼ ð6:21Þ
b
a4 t r b
a2

The transformation matrix in Eq. (6.21) is unitary as required for any quantum
transformation.
6.3 Input/Output Transformation 61

Exercise 6.2 Using the commutator relations of the beam splitter input
operators, show that the correct commutator relations are obtained for the
output operators.

Let us find an expression for the input annihilation operators in terms of the
output operators. Starting with Eq. (6.15) and multiplying both sides by r gives

rb
a3 ¼ r  rb
a1 þ r  tb
a2 ð6:22Þ

Similarly, starting with Eq. (6.16) and multiplying both sides by t gives

tb
a4 ¼ t  tb
a1  t  rb
a2 ð6:23Þ

Adding Eqs. (6.22) and (6.23) gives

ðr r þ t  t Þb
a1 þ ðr  t  t  r Þb
a2 ¼ r  b
a3 þ t  b
a4 ð6:24Þ

From Eqs. (6.11) and (6.12), Eq. (6.24) simplifies to

a1 ¼ r  b
b a3 þ t  b
a4 ð6:25Þ

Similarly, we can prove that

a2 ¼ t  b
b a3  r  b
a4 ð6:26Þ

Taking the Hermitian conjugate of Eqs. (6.25) and (6.26) gives

a{1 ¼ rb
b a{3 þ tb
a{4 ð6:27Þ

a{2 ¼ tb
b a{3  rb
a{4 ð6:28Þ

These relations will be useful in our upcoming derivations.

Exercise 6.3 Derive Eq. (6.26).

6.3 Input/Output Transformation

Suppose we have some input state to the beam splitter, jψini. The output state can be
obtained by applying a transformation matrix, similar to Eq. (6.21):
62 6 Single Photon on a Beam Splitter

jψout i ¼ Ujψin i ð6:29Þ

The expectation value hOouti, associated with some output operator of the beam
b out , is
splitter, O

b out jψout i
hOout i ¼ hψout jO ð6:30Þ

Using Eq. (6.29):

b out Ujψin i
hOout i ¼ hψin jU { O ð6:31Þ

b out U can be interpreted as the operator O

The transformation U { O b out (expressed in
terms of the output modes 3 and 4) transformed into an input operator (expressed in
terms of the input modes 1 and 2):

O b out U
b in ¼ U { O ð6:32Þ

Thus,

b in jψin i
hOout i ¼ hψin jO ð6:33Þ

In summary, hOouti can be determined using either the input space or the output
space:

b out jψout i ¼ hψin jO

hOout i ¼ hψout jO b in jψin i ð6:34Þ

This idea will become more clear in the next section where it will be used to simplify
the analysis of the beam splitter.

6.4 Single Photon on a Beam Splitter

Consider the beam splitter in Fig. 6.4 with a single photon input state, j1i1, on port
1 and nothing on port 2. “Nothing” is represented by the vacuum state, j0i2,
on port 2. The combined input state is denoted as jψini ¼ j1i1j0i2. Single
photons are launched into the beam splitter and the detection events at D3 and
D4 are counted. After repeating the experiment many times, the probability of
detection at D3 and D4 is determined. For each photon launched into the beam
splitter, the probability of detection at D3 is
6.4 Single Photon on a Beam Splitter 63

Fig. 6.4 Single photon on

input port 1

hn3 i hn3 i b 3 jψout i

P3 ¼ ¼ ¼ hψoutj N ð6:35Þ
h n1 i 1

where Nb 3 is the number operator for D3 and jψouti is the output state at D3. hn1i
D   E
 { 
represents the single photon input to port 1, that is, hn1 i ¼ ψin b a1 b
a1 ψin ¼
a{1b
2 h0 j 1 h1 jb a1 j1i1 j0i2 ¼ 1. According to the definition of the number operator, we
have

a{3b
P3 ¼ hψout jb a3 jψout i ð6:36Þ

We can convert the output space of Eq. (6.36) into the input space using Eq. (6.34).
We replace jψouti with jψini and the output operators, ba{3 and b
a3 , are expressed in
terms of the input operators using Eqs. (6.15) and (6.17):

a{1 þ t b
P3 ¼ hψin jðr b a{2 Þðrb
a1 þ tb
a2 Þjψin i ð6:37Þ

Using jψini ¼ j1i1j0i2 gives

a{1b
P3 ¼ 2 h0 j 1 h1 jðr  rb a{1b
a1 þ r  tb a{2b
a2 þ t  rb a{2b
a1 þ t  tb a2 Þj1i1 j0i2 ð6:38Þ

Using the properties of the creation and annihilation operators on the states j0i and
j1i, only the first term in Eq. (6.38) survives, giving:

P3 ¼ 2 h0 j 1 h1 jr rj1i1 j0i2 ¼ r  r 2 h0j0i2 1 h1j1i1 ð6:39Þ

In Eq. (6.39), 2h0j0i2 ¼ 1 and 1h1j1i1 ¼ 1 since the states are normalized, giving

P3 ¼ r  r ¼ jr j2 ¼ R ð6:40Þ

Here, R is the probability of a single photon from port 1 being detected at D3. Note
that R is the reflectance from the beam splitter. Equation (6.40) is the same as the
classical result for the fraction R of light intensity reflected by the beam splitter from
port 1 to D3.
64 6 Single Photon on a Beam Splitter

Similarly, we can calculate the probability of detection at D4:

a{4b
P4 ¼ hψout j b a4 jψout i ð6:41Þ

Using Eqs. (6.16) and (6.18) gives

a{1  r b
P4 ¼ hψin jðt b a{2 Þðtb
a1  rb
a2 Þjψin i ð6:42Þ

a{1b
¼ 2 h0 j 1 h1 jðt  tb a{1b
a1  t  rb a{2b
a2  r  tb a{2b
a1 þ r  rb a2 Þj 1i1 j 0i2 ð6:43Þ
¼ 2 h0 j 1 h1 jðt  t Þj 1i1 j 0i2 ð6:44Þ

¼ t t ¼ jt j ¼ T
2
ð6:45Þ

Here, T is the probability of a single photon from port 1 being detected at D4. Note
that T is the transmittance through the beam splitter. Equation (6.45) is the same as
the classical result for the fraction T of light intensity transmitted by the beam splitter
from port 1 to D4. Note that the total probability for detection at either detector adds
to unity as required, since P3 þ P4 ¼ R þ T ¼ 1.

6.5 Coincident Measurements

Let us calculate the probability, P34, of simultaneous detection at D3 and D4.

Simultaneous detection events are called coincidence or correlation measurements.
Classically, we would expect a fraction R of a classical light intensity reflected to D3
and a fraction T transmitted to D4, allowing simultaneous detection. Classically, we
expect the intensity to be proportional to (E3)2 ¼ R(E1)2 at detector D3 and
(E4)2 ¼ T(E1)2 at detector D4, giving the probability of double detection:

RðE1 Þ2 T ðE 1 Þ2
Classical : P34 ¼ ¼ RT ð6:46Þ
ðE 1 Þ2 ðE 1 Þ2

Let us calculate the probability of a coincident detection, P34, for a single photon
input on the beam splitter, as illustrated in Fig. 6.5. The simultaneous measurement
is described by the operator ba4 b
a3 :

P34 ¼ hψout j ðb a3 Þ{ ðb
a4 b a{3b
a3 Þjψout i ¼ hψout j b
a4 b a{4b
a4 b
a3 jψout i ð6:47Þ

Using Eqs. (6.15) to (6.18) gives

6.6 Second-Order Correlation Function 65

Fig. 6.5 Correlation or

coincidence measurement
(simultaneous detection at
D3 and D4) with single
photon input on port 1

Coincidence
counts

a{1 þ tb
P34 ¼ 2 h0 j 1 h1 jðr b a{2 Þðt b
a{1  r b
a{2 Þðtb
a1  rb
a2 Þðrb
a1 þ tb
a2 Þj1i1 j0i2 ð6:48Þ

Evaluating all the terms of Eq. (6.48) gives

P34 ¼ 0 ð6:49Þ

Quantum mechanically, double detections are not possible for a single photon,
which is very different than the classical result of Eq. (6.46). The single photon is
detected at D3 with probability R, or at D4 with probability T, but never both
simultaneously. Here, we have a nonclassical correlation. The absence of double
detections must be the case if the concept of “single photon” is to make any sense at
all. You can only detect a single photon once, either at D3 or D4.

Exercise 6.4 Evaluate Eq. (6.48), verifying that P34 ¼ 0.

The single photon beam splitter could be used as a random number generator.
With a 50:50 beam splitter (R ¼ T ¼ 0.5), we have a probability P3 ¼ P4 ¼ 0.5 that a
single photon is detected at either D3 or D4. A single photon is launched into the
beam splitter, and a 0 bit is assigned for detection at D3, while a 1 bit is assigned for
detection at D4. After launching many single photons, one at a time, into the beam
splitter, a random sequence of bits is generated, 00110101110. . . The random
sequence of bits can be used to generate a random number.

6.6 Second-Order Correlation Function

The correlations described in the previous section are usually described by a second-
order correlation function, g(2)(τ), introduced in 1963 by Roy Glauber (Fig. 6.6), a
pioneer of quantum optics [7]. The 2005 Nobel Prize in Physics was divided, one
half awarded to Roy J. Glauber “for his contribution to the quantum theory of optical
coherence,” the other half jointly to John L. Hall and Theodor W. Hänsch “for their
contributions to the development of laser-based precision spectroscopy, including
the optical frequency comb technique.”
66 6 Single Photon on a Beam Splitter

Fig. 6.6 Roy J. Glauber.

(Credit: Photograph by Jane
Reed, Harvard News Office,
courtesy AIP Emilio Segrè
Visual Archives, Gift of Roy
Glauber, 2006)

First, we look at the classical definition of the second-order correlation function,

which is given by

ð2Þ hI ðt ÞI ðt þ τÞi hE  ðt ÞE ðt ÞE ðt þ τÞE ðt þ τÞi

gclassical ðτÞ ¼ ¼ ð6:50Þ
hI ðt Þi2 hE ðt ÞEðt Þi2

where I / jEj2 ¼ EE. The brackets, hi, indicate an average to account for intensity
ð2Þ
fluctuations during the measurement time. gclassical ðτÞ describes the correlation
between two temporally separated intensity signals with time difference τ from
ð2Þ
one source. If τ ¼ 0, gclassical ð0Þ is especially interesting, because it gives the
probability of simultaneous detection events at two detectors, normalized to the
probability of individual detection events at either detector. The “ 0 ” means no time
delay between the two simultaneous detections.
Suppose the input to the beam splitter is treated as a classical source of light. For
classical light, we have
 2 2
ð2Þ RE 1 TE 1
gclassical ð0Þ ¼    ð6:51Þ
RhE 1 i2 T hE1 i2

R and T cancel out, and since E 21 is proportional to the light intensity, we get
 2
ð2Þ I
gclassical ð0Þ ¼ ð6:52Þ
hI i2

Next, we can use the Cauchy–Schwarz inequality, which states

6.6 Second-Order Correlation Function 67

 2
I  hI i2 ð6:53Þ

for any positive random variable. Thus,

ð2Þ
gclassical ð0Þ  1 ð6:54Þ

We can rewrite Eq. (6.50) for quantum light by replacing the electric field with its
corresponding operator in the normal order:

  þ þ
hE b ðt þ τ ÞE
b ð t ÞE b ðt þ τ ÞE
b ðt Þi
gð2Þ ðτÞ ¼ ð6:55Þ
b  ð t ÞE
b þ ðt Þi
2
hE
 þ
where Eb and Eb are the negative and positive frequency components, respectively,
of the field operator introduced in Chap. 5. The correlation function, g(2)(0), for
quantum light becomes

a{ b
hb a{ b
ab
ai
g ð 2 Þ ð 0Þ ¼ 2
ð6:56Þ
a{ b
hb ai

a{ b
Using the commutation relation, b a¼ba{  1, gives
ab

a{ ðb
hb a{  1Þb
ab ai
gð2Þ ð0Þ ¼ 2
ð6:57Þ
a{ b
hb ai

a{ b
hb a{b
ab a{ b
ab ai
¼ 2
ð6:58Þ
a{b
hb ai

Recognizing b a{b
a as the number operator, we can write Eq. (6.58) in various
equivalent forms:
 2
ð2Þ hn2  ni hnðn  1Þi n  h ni
g ð0Þ ¼ 2
¼ 2
¼ ð6:59Þ
hni h ni h ni 2

Using the variance, (Δn)2 ¼ hn2i  hni2, gives

ðΔnÞ2 þ hni2  hni

gð2Þ ð0Þ ¼ ð6:60Þ
h ni 2
68 6 Single Photon on a Beam Splitter

ðΔnÞ2  hni
¼1þ ð6:61Þ
h ni 2

Let us examine the case for a Fock state jni on a beam splitter. The probability of
detection at D3 and D4 are:

a{ b
P3 ¼ Rhnjb ajni ¼ Rn ð6:62Þ
{
P4 ¼ T hnjb
abajni ¼ Tn ð6:63Þ

and the product is

P3 P4 ¼ RTn2 ð6:64Þ

The probability of double detection is

a{b
P34 ¼ RT hn j b a{ b
ab
ajni ð6:65Þ

a{ ðb
¼ RT hn j b a{  1Þb
ab aj ni ð6:66Þ

a{b
¼ RT hn jðb a{ b
ab a{b
ab aÞj ni ð6:67Þ
2

¼ RT n  n ð6:68Þ
¼ RTnðn  1Þ ð6:69Þ

Thus, for the Fock state, we get

P34 nðn  1Þ n  1
Fock state : gð2Þ ð0Þ ¼ ¼ ¼ <1 ð6:70Þ
P3 P4 n2 n

which we could have also obtained directly from Eq. (6.59). The result in Eq. (6.70)
ð2Þ
is very different from classical light where gclassical ð0Þ  1. In the quantum case, the
detection of a photon destroys it and changes the state, leaving one photon less (thus,
we get n(n  1) in the numerator of Eq. (6.70)). This reduces the probability of
double detection, which is not considered in the classical expression. This quantum
effect (g(2)(0) < 1) is called “anticorrelation” or “antibunching”. The first measure-
ment of antibunching in Ref. [8] showed the nonclassical or quantum nature of light.
If we set n ¼ 1 for a single photon state, we obtain g(2)(0) ¼ 0 from Eq. (6.70), as
expected. The detection of a single photon destroys it, leaving zero probability of
double detection.
6.7 Entangled State 69

6.7 Entangled State

If we have no input to the beam splitter, then we expect the action of the beam splitter
to give

beam
splitter
j 0i1 j 0i2 ! j 0i3 j 0i4 ð6:71Þ
|fflfflfflffl{zfflfflfflffl} |fflfflfflffl{zfflfflfflffl}
input output

The single photon input state in Fig. 6.4 can be expressed as

a{1 j0i1 j0i2

jψin i ¼ j1i1 j0i2 ¼ b ð6:72Þ

According to Eqs. (6.34) and (6.71), Eq. (6.72) can be expressed in terms of the
output space using Eq. (6.27):

beam
splitter  
a{1 j0i1 j0i2 ! rb
b a{3 þ tb
a{4 j0i3 j0i4 ð6:73Þ

¼ rj 1i3 j 0i4 þ t j 0i3 j 1i4 ð6:74Þ

The output in Eq. (6.74) is called an entangled state of a photon in the D3 path and
the D4 path. An entangled state is a state, which cannot be separated or factored into
individual product states; that is, jψi 6¼ jψ3i3jψ4i4. Equation (6.74) cannot be
factored into the product of two individual states (try it). If you are not familiar
with entanglement, do not worry. We will cover this topic in more detail in Chap. 8.
Equation (6.74) tells us that the single photon input on port 1 results in a
superposition of the single photon in mode 3 with zero photons in mode 4, and
vice versa. The probability amplitude for a single photon along the D3 path is given
by r, that is, the coefficient of j1i3j0i4 in Eq. (6.74). The corresponding probability is
the modulus squared of the probability amplitude, jrj2 ¼ R, the same as Eq. (6.40).
The probability amplitude for a single photon along the D4 path is given by t, that is,
the coefficient of j0i3j1i4 in Eq. (6.74). The corresponding probability is the
modulus squared of the probability amplitude, jtj2 ¼ T, the same as Eq. (6.45). As
seen in Eq. (6.74), the probability of joint detection at D3 and D4, represented by the
state j1i3j1i4, is zero.
70 6 Single Photon on a Beam Splitter

6.8 Hanbury Brown–Twiss Experiment

The concept of coincidence measurements, illustrated in Fig. 6.5, was first proposed
for radio astronomy by Robert Hanbury Brown and Richard Twiss in 1954 [9], and
later extended to optical signals [10]. Known as the Hanbury Brown–Twiss exper-
iment, it is now a standard method to characterize light sources, including whether an
emitter is a good source of single photons. For a single photon input to the beam
splitter, we expect g(2)(0) ¼ 0, which is very different than the classical result of
g(2)(0)  1 from Eq. (6.54). g(2)(0) below 103 has been measured for certain single
photon sources [11].

References

1. F.L. Pedrotti, L.M. Pedrotti and L.S. Pedrotti, Introduction to optics (3rd Edition, Pearson,
2006).
2. M. W. Hamilton, Phase shifts in multilayer dielectric beam splitters, Amer. J. Phys.
68 (2000) 186.
3. F. Hénault, Quantum physics and the beam splitter mystery, Proc. SPIE 9570, The Nature of
Light: What are Photons? VI, 95700Q (10 September 2015); https://doi.org/10.1117/12.
2186291
4. S. Prasad, M.O. Scully and W. Martienssen, A quantum description of the beam splitter, Opt.
Commun. 62 (1987) 139.
5. Z.Y. Ou C.K. Hong and L. Mandel, Relation between input and output states for a beam
splitter, Opt. Commun. 63 (1987) 118.
6. H. Fearn and R. Loudon, Quantum theory of the lossless beam splitter, Opt. Commun.
64 (1987) 485.
7. R. J. Glauber, Quantum theory of optical coherence, Phys. Rev. 130 (1963) 2529.
8. H. J. Kimble, M. Dagenais, L. Mandel, Photon antibunching in resonance fluorescence, Phys.
Rev. Lett. 39 (1977) 691.
9. R. Hanbury Brown and R.Q. Twiss, A new type of interferometer for use in radio astronomy,
Philosophical Magazine 45 (1954) 663.
10. R. Hanbury Brown and R.Q. Twiss, Correlation between photons in two coherent beams of
light, Nature 177 (1956) 27.
11. M.D. Eisaman, J. Fan, A. Migdall and S.V. Polyakov, Single-photon sources and detectors,
Rev. Sci. Instrum. 82 (2011) 071101.
Chapter 7
Single Photon in an Interferometer

Another important device in quantum optics is the interferometer. Building on the

results of the previous chapter for a beam splitter, the classical and quantum optics
treatment of the Mach-Zehnder interferometer is introduced. The case of a single
photon in an interferometer is treated, which introduces the concept of wave-particle
duality.

7.1 Classical Light Interference

The Mach-Zehnder (MZ) interferometer, illustrated for classical light in Fig. 7.1,
superposes the light field on a detector from two possible paths. Note that the two
beam splitters in Fig. 7.1 are reversed. The input field E1 along path 1 to detector D3
undergoes two transmissions (t2) and a phase shift of eikz1 associated with the
distance z1. The input field E1 along path 2 to detector D3 undergoes two reflections
(r and
r) and a phase shift of eikz2 associated with the distance z2. Therefore, the
field at detector D3 is a result of the superposition of the field along the two paths:


E3 ¼ t 2 eikz1
r 2 eikz2 E 1 ð7:1Þ

We assume r and t are real numbers, so r ¼ r and t ¼ t. The light

intensity (or detection probability) at D3 is proportional to the square of the field,
I3/ (E3)2 ¼ E3E3. From Eq. (7.1), we get



I 3 ¼ t 2 eikz1
r 2 eikz2 t 2 e
ikz1
r 2 e
ikz2 I 1 ð7:2Þ

where I1/ (E1)2 ¼ E1E1. Evaluating Eq. (7.2) gives

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 71

R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_7
72 7 Single Photon in an Interferometer

Fig. 7.1 Mach-Zehnder

interferometer with classical
light

h  i
I 3 ¼ R2 þ T 2
RT eikðz1
z2 Þ þ e
ikðz1
z2 Þ I 1 ð7:3Þ
 
¼ R2 þ T 2
2RT cos ðkΔzÞ I 1 ð7:4Þ

where Δz ¼ z1
z2 is the path length difference between path 1 and path 2, and
kΔz is the corresponding phase difference.
Similarly, the field at detector D4 is


E 4 ¼ rteikz1 þ treikz2 E 1 ð7:5Þ

and the intensity is



I 4 ¼ rteikz1 þ treikz2 rte
ikz1 þ tre
ikz2 I 1 ð7:6Þ
h  i
¼ 2RT þ RT eikðz1
z2 Þ þ e
ikðz1
z2 Þ I 1 ð7:7Þ

¼ ½2RT þ 2RT cos ðkΔzÞ I 1 ð7:8Þ

Adding Eqs. (7.4) and (7.8), we get


I 3 þ I 4 ¼ R2 þ T 2 þ 2RT I 1 ð7:9Þ

¼ ðR þ T Þ2 I 1 ð7:10Þ
¼ I1 ð7:11Þ

since R þ T ¼ 1. The total output light intensity at D3 and D4 is equal to the input
light intensity, as expected from the conservation of energy.
For 50:50 beam splitters (R ¼ T ¼ 0.5), we get

1 2 kΔz
I 3 ¼ ½1
cos ðkΔzÞ I 1 ¼ I 1 sin ð7:12Þ
2 2
7.1 Classical Light Interference 73

Fig. 7.2 Light output

intensity at D3 (red) and D4
,
(blue) versus phase
difference, kΔz


1 2 kΔz
I 4 ¼ ½1 þ cos ðkΔzÞ I 1 ¼ I 1 cos ð7:13Þ
2 2

Let us suppose that we change the path length difference, Δz, between path 1 and
2 (for example, by moving one of the mirrors) and we measure the intensities at D3
and D4 as a function of the phase difference, kΔz. The result from Eqs. (7.12) and
(7.13) is plotted in Fig. 7.2. The oscillating intensities are due to alternating
constructive and destructive interference of the two light fields along path 1 and
2 as the path length difference, Δz, changes. In particular, if kΔz ¼ 0 or any integer
multiple of 2π, then all of the input light appears at port 4 and none at port 3. We call
port 3 the “dark port”. A small change in the path length difference, Δz, will appear
as some light on the dark port. In this way, we can detect some path length
difference, Δz.
Let us calculate the probability of double detection, that is, simultaneous
detection at both D3 and D4. Classically, for 50:50 beam splitters in the interferom-
eter, we get

kΔz
kΔz
I 1 sin 2 I 1 cos 2
P34 ¼ 2
2
2
ð7:14Þ
ðI 1 Þ
 
2 kΔz 2 kΔz
¼ sin cos ð7:15Þ
2 2
1
¼ sin 2 ðkΔzÞ ð7:16Þ
4
74 7 Single Photon in an Interferometer

7.2 Quantum Light Interference

Let us now derive the output for the case of a single photon input, as shown in
Fig. 7.3. Here, we clearly need a quantum description. The annihilation operator
associated with photodetection at D3 is



b
a3 ¼ t 2 eikz1
r 2 eikz2 ba1 þ
treikz1
rteikz2 ba2 ð7:17Þ

The first term follows from Eq. (7.1). The second term derives from the input on
port 2, which is vacuum. Similar to Eq. (6.35), the probability of single photon
detection at D3 is

b 3 jψout i
P3 ¼ hψout jN ð7:18Þ

Using Eq. (7.17), we can express Eq. (7.18) in terms of the input space. Assuming
r and t are real, we get
h

 { i
P3 ¼ 2 h0j1 h1j t 2 e
ikz1
r 2 e
ikz2 t 2 eikz1
r 2 eikz2 b
a1 b
a1 j 1 i 1 j 0 i 2 ð7:19Þ

where all terms related to b

a2 are omitted, because they result in zero when applied to
the vacuum input, j0i2, on port 2. Evaluating Eq. (7.19) gives
 
P3 ¼ R2 þ T 2
2RT cos ðkΔzÞ ð7:20Þ

which is the same as the classical result.

Similarly, the annihilation operator associated with photodetection at D4 is



b
a4 ¼ rteikz1 þ treikz2 ba1 þ
r 2 eikz1 þ t 2 eikz2 ba2 ð7:21Þ

The probability of single photon detection at D4 is

Fig. 7.3 Mach-Zehnder

interferometer with single
photon input
7.2 Quantum Light Interference 75

b 4 jψout i
P4 ¼ hψout jN ð7:22Þ

or, in terms of the input space:

h

 { i
P4 ¼ 2 h0j1 h1j rte
ikz1 þ tre
ikz2 rteikz1 þ treikz2 b
a1 b
a1 j1i1 j0i2 ð7:23Þ

¼ 2RT þ 2RT cos ðkΔzÞ ð7:24Þ

which is the same as the classical result.

Exercise 7.1 Derive P3 and P4 if the two beam splitters in Fig. 7.3 are
identically oriented with the “dot” on top. How does it compare to the results
obtained above?

The probability of double (simultaneous) detection at D3 and D4 is

a{3b
P34 ¼ hψout jb a{4b
a4 b
a3 jψout i ð7:25Þ

Evaluating Eq. (7.25), using Eqs. (7.17) and (7.21), gives

P34 ¼ 0 ð7:26Þ

which is different from the classical case (but identical to the beam splitter in
Chap. 6). Simultaneous detection events at D3 and D4 are possible with classical
light, but not with single photons.

Exercise 7.2 Derive Eq. (7.26).

Suppose a single photon is input to the MZ interferometer for a given path length
difference Δz, resulting in a detection event at either D3 or D4. The experiment is
repeated many times, and the number of counts at D3 and D4 is tallied for a given Δz.
Next, the experiment is repeated for different path length differences, Δz. The
number of counts at D3 and D4 versus the phase, kΔz, can then be plotted. This
experiment has been done, using an atomic cascade as a heralded source of single
photons, with the results shown in Fig. 7.4 [1]. The quantum result in Fig. 7.4 is
identical to the classical result in Fig. 7.2. An interference pattern occurs after
counting many single photon detection events, although only a single photon
traverses the interferometer at a time and only a single photon is ever detected
at D3 or D4!
76 7 Single Photon in an Interferometer

Fig. 7.4 Number of counts in outputs D3 (labeled as MZ1) and D4 (MZ2) as a function of the path
difference Δz (one channel corresponds to a λ/50 variation of Δz). (a) 1 s counting time per channel.
(b) 15 s counting time per channel. (Reproduced with permission from Grangier et al. [1])

7.3 Wave-Particle Duality

The fact that a single photon is detected at either D3 or D4, and there are no double
detections in beam splitter or interferometer experiments, indicates that the photon
exhibits particle behavior. On the other hand, after counting many single photon
detection events, an interference pattern occurs that is indicative of wave behavior.
This happens because the photon probability amplitude is put into a superposition of
both paths, resulting in interference. Photodetection collapses the state to either D3 or
D4. This is a manifestation of “wave-particle duality”.

Exercise 7.3 In the 1970s, the physicist John Archibald Wheeler proposed a
thought experiment, called the delayed choice experiment, which further
demonstrates wave-particle duality. Describe the delayed choice experiment
and its implications for the description of light.

Reference

1. P. Grangier, G. Roger and A. Aspect, Experimental evidence for a photon anticorrelation effect
on a beam splitter: A new light on single-photon interferences, Europhys. Lett. 1 (1986) 173.
Chapter 8
Entanglement

There exist multiphoton states that cannot be expressed as a product of individual

photon states. These states are called entangled states. Einstein used hidden variables
in an attempt to explain the probabilities of quantum mechanics. John Bell proposed
a test, using entangled states, showing that quantum mechanics cannot be explained
by hidden variables.

8.1 Entangled States

In Chap. 6 (Sect. 6.7), we saw that the beam splitter output was an entangled state.
Let us examine a two-photon state, which can be written in general as a product of
two superpositions:



jψi ¼ jψi1 jψi2 ¼ α1 j0i1 þ β1 j1i1 α2 j0i2 þ β2 j1i2 ð8:1Þ

where the subscript on each ket indicates the photon number (photon 1 or photon 2),
and j0i and j1i represent the two possible orthogonal states of each photon—for
example, horizontal and vertical polarization, or two different paths in a beam
splitter or interferometer. Expanding this two-photon state gives

jψi ¼ α1 α2 j0i1 j0i2 þ α1 β2 j0i1 j1i2 þ β1 α2 j1i1 j0i2 þ β1 β2 j1i1 j1i2 ð8:2Þ

Suppose you are given a composite state of two photons (e.g., Eq. (8.2)) and
asked for the state of the individual photons. To answer this question, you
would work backward to factor the state, obtaining Eq. (8.1). This is known as a
separable state.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 77

R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_8
78 8 Entanglement

There are some states for which this factoring is impossible; that is, you cannot
write the composite state as a product of the individual states:

jψi 6¼ jψi1 jψi2 ð8:3Þ

These are known as entangled states. In entangled states, you cannot talk about the
state of the photons individually—they are somehow intertwined. Note that a general
two-photon state, jψi ¼ α00j0i1j0i2 þ α01j0i1j1i2 þ α10j1i1j0i2 þ α11j1i1j1i2, is
usually entangled rather than separable – entanglement is normal in quantum
mechanics!
There are four entangled two-photon states that are commonly encountered,
known as the Bell states (we have dropped the particle subscripts):

1
jΦþ i ¼ pffiffiffi ðj00i þ j11iÞ ð8:4Þ
2
1
jΨþ i ¼ pffiffiffi ðj01i þ j10iÞ ð8:5Þ
2
1
jΦ i ¼ pffiffiffi ðj00i  j11iÞ ð8:6Þ
2
1
jΨ i ¼ pffiffiffi ðj01i  j10iÞ ð8:7Þ
2

The states in Eqs. (8.4), (8.5), (8.6) and (8.7) cannot be factored into the product of
two individual photon states (try it) according to the definition of entanglement.
Suppose we prepare the entangled state jΨi (Eq. (8.7)) between two photons and
then separate them by large distances (note that entangled states can be prepared by
the various processes discussed in Chap. 5 such as atomic cascades, biexciton
recombination in QDs, or spontaneous parametric down-conversion). Let us suppose
that j0i represents vertical polarization and j1i represents horizontal polarization.
After separating the photons, an individual (let us call her Alice) could perform a
polarization measurement on her photon and another individual (let us call him Bob)
could perform a subsequent polarization measurement on his photon. If Alice
measures vertical polarization (j0i), Bob will measure horizontal polarization (j1i),
and vice versa. The measurements are correlated between Alice and Bob, and this
occurs no matter the distance between them. There appears to be instantaneous
action at a distance or “nonlocality” in entanglement, which Einstein called “spooky
action at a distance”. Could Alice and Bob use entanglement to communicate
instantaneously across vast distances? The theory of special relativity states that
information cannot travel faster than the speed of light. Does entanglement violate
special relativity?
In fact, entanglement does not violate special relativity, because no information
is being transmitted. Upon measurement, Alice will collapse the entangled state,
jΨ i ¼ p1ffiffi2 ðj01i  j10iÞ, to a separable state—either j01i with probability ½, or j10i
8.2 EPR Paradox and Hidden Variables 79

with probability ½; that is, Alice obtains either vertical or horizontal polarization
with 50% probability (i.e., random), and Bob obtains the opposite polarization. Alice
cannot control which of these two states she obtains. Subsequent measurement by
Bob will result in the opposite polarization state to Alice, but his measurement will
likewise appear to him to be completely random—no information is sent. Special
relativity remains intact.

8.2 EPR Paradox and Hidden Variables

In a famous 1935 paper [1], Albert Einstein, Boris Podolsky, and Nathan Rosen
(known as EPR) sought to demonstrate by the “EPR paradox” that quantum
mechanics was incomplete. EPR were concerned by the instantaneous action at a
distance, or “nonlocality”, implied by entanglement. Quantum mechanics also seems
to violate “realism”. “Realism” means that particles have definite properties that are
independent of any measurement.
Suppose we toss a coin. In principle, it is possible to know whether it will land
heads or tails if we keep track of a lot of information about the system (called
“degrees of freedom”), such as the forces applied during the toss, the air currents, the
height of the toss, etc. However, all these physical properties are impossible to
calculate in practice, so the most we can do is ascribe a probability distribution for
the toss outcome resulting in Pheads ¼ 12 and Ptails ¼ 12 . This outcome occurs from
averaging the many degrees of freedom that we do not have access to. This principle
also forms the basis for statistical thermodynamics.
Einstein and many others believed that quantum mechanics was like this; that is,
they proposed that the probabilities in quantum mechanics are deterministic (versus
probabilistic) and have some underlying causes that are “hidden”; that is, that we
cannot access (analogous to the unknown variables during the coin toss). These
underlying causes were called “hidden variables”. If we knew the hidden variables,
we would be able to calculate a definite measurement outcome, rather than just
probabilities.
Many quantum pioneers, exemplified by Einstein, believed in “local realism”
where the state of particles is defined when they are created. However, the “hidden
variables” only allow us to determine the probability of these states. Einstein
famously said: “God does not play dice with the universe”. Also, with regards to
realism, Einstein said “Do you believe the moon exists only when you look at it?”
Others, exemplified by Bohr, believed in the possibility of superpositions and
entanglement. They believed that no definitive statements about a physical system
may be made until a measurement is made. Particle properties do not exist until we
measure them. It turns out that Bohr was correct; but how do we prove it?
80 8 Entanglement

8.3 CHSH Inequality

In 1964, the physicist, John Bell, proposed a test for quantum mechanics by
a, b
measuring the spin states along three different directions (b b, bc) for many entangled
pairs of electron spins [2]. In 1969, Clauser, Horne, Shimony, and Holt (CHSH)
refined the Bell test to one that was more amenable to experiment [3]. Rather than
using entangled pairs of electron spins, CHSH proposed a test using entangled pairs
of photon polarizations. Suppose a polarization entangled state, jψi ¼ p1ffiffi2 
ðjVH i  jHV iÞ, is prepared where H represents horizontal polarization and V rep-
resents vertical polarization. Suppose a represents the result of a measurement along
two orthogonal polarization directions for Alice. The measurement results are
assigned the values þ1 and 1; that is, a ¼  1. For example, for the state jψi,
a would correspond to a polarization measurement. If the polarization measurement
yields horizontal polarization, then the value a ¼ þ 1 is assigned to the measure-
ment result. Alternatively, if the polarization measurement yields vertical polariza-
tion, then the value a ¼  1 is assigned to the measurement result. Similarly,
a0 ¼  1 represents the measurement results made by Alice along a different set of
orthogonal polarization directions (e.g., 45 and 135 ). Similarly, Bob can measure
along two sets of orthogonal polarizations with results b ¼  1 (corresponding to
polarization along 22.5 or 112.5 ) and b0 ¼  1 (corresponding to polarization
along 67.5 and 157.5 ). If a ¼  1 and a0 ¼  1, it follows that either a þ a0 ¼ 0, in
which case a  a0 ¼  2. Otherwise, a  a0 ¼ 0, in which case a þ a0 ¼  2.
Therefore, we define a quantity S:

S ¼ ða0 þ aÞb þ ða0  aÞb0 ¼ a0 b þ ab þ a0 b0  ab0 ¼ 2 ð8:8Þ

Quantities such as ab represent a coincidence or correlation measurement. For

example, a and b are obtained in measurements made by Alice and Bob, respec-
tively, for each entangled photon pair and the product ab is determined. After
repeated measurements on many photon pairs, the average value of S will depend
on the probability distribution of each coincidence measurement. For example, the
average of ab, habi, will be given by the probability distribution (Pþþ)(1) þ (P)
(1) þ (Pþ)(1) þ (Pþ)(1) where, for example, Pþþ is the probability of both
Alice and Bob obtaining a measurement value of þ1. After many trials, we must
obtain

jhSij ¼ jha0 bi þ habi þ ha0 b0 i  hab0 ij  2 ð8:9Þ

where the brackets hi represents an average over many coincidence measurements.

Eq. (8.9) is the CHSH inequality. Different versions of the CHSH inequality exist,
depending on how Eq. (8.9) is expressed. The signs in Eq. (8.9) are not so important
as long as one of them is different than the others.
8.4 Testing the CHSH Inequality 81

Fig. 8.1 Measurement

directions corresponding to
a, a0, b, and b0 equal to
+1 for the CHSH inequality
(1 corresponds to the
orthogonal directions)

Equation (8.9) is a classical prediction. What does quantum mechanics

predict? We can show for photons in the entangled state, jψi ¼ p1ffiffi2 ðjHV i  jVH iÞ,
that habi ¼  cos (2θ) where θ is the angle between the two measurement
directions (Exercise 8.1). For the angles mentioned earlier, and shown in Fig. 8.1,
habi ¼  cos ð2  22:5 Þ ¼  p1ffiffi2, ha0 b0 i ¼  cos ð2  22:5 Þ ¼  p1ffiffi2, ha0 bi ¼  cos
pffiffiffi
ð2  22:5 Þ ¼  p1ffiffi, and hab0 i ¼  cos ð2  67:5 Þ ¼ p1ffiffi, which gives jhSij ¼ 2 2,
2 2
that is, greater
pffiffiffi than the classical prediction of 2! It can be shown that the upper bound of
jhSij is 2 2 in the case when a, b, a0, and b0 measurement directions are separated by
successive 22.5  angles as shown in Fig. 8.1 [4].

Exercise 8.1 Show that for photons in the entangled state jψi ¼ p1ffiffi2 
ðjHV i  jVH iÞ, we get habi ¼  cos (2θ) where θ is the angle between the
measurement directions.

8.4 Testing the CHSH Inequality

The physicist, Alain Aspect (Fig. 8.2), famously performed an experiment in 1981
(Fig. 8.3), using polarization entangled photons produced by an atomic cascade,
showing violation of the CHSH inequality because of quantum mechanics
[5]. Polarizers were used to choose the polarization basis and split orthogonal
polarization states to separate detectors. The path length between the source and
the “Alice” detector is shorter than that for the “Bob” detector, so Alice performs her
measurement before Bob. The experiment is repeated many times with entangled
photon pairs, verifying a violation of the CHSH inequality. In the decades since
Aspect’s test, various “loopholes” in the test have been closed [7–9], confirming that
quantum mechanics is not consistent with a hidden variables theory. Quantum
mechanics violates local realism. Bohr was correct. However, the “spooky action
at a distance” still seems mysterious and remains an inspiration to both physicists
and philosophers.
82 8 Entanglement

Fig. 8.2 Alain Aspect

(Nobel Prize in Physics in
2022). (Credit: Wikimedia
Commons [6])

Fig. 8.3 Optical test of the Polarizer Polarizer

CHSH inequality

Source
Alice Bob

Correlaons

8.5 Quantum Key Distribution

Entangled states provide a means of secure communication called quantum key

distribution (QKD). In 1991, Arthur Ekert proposed the E91 QKD protocol using
entangled photons [10]. Alice and Bob receive one photon each of a polarization
entangled pair from a source located either at Alice’s or Bob’s position, or some-
where else. For each photon they receive, Alice and Bob choose a polarization
measurement basis and record the result of the measurement. Alice and Bob can
perform a Bell test using their measurement results. If the Bell inequality is not
violated, then an eavesdropper (called “Eve”) must have destroyed the entanglement
by eavesdropping. If the Bell inequality is violated, then there was no eavesdropper,
and Alice and Bob can use their measurement results to generate a secret key to
encrypt their messages. These implementations are referred to as entanglement-
based protocols, Ekert protocols, or EPR protocols.
Entanglement provides a new resource that does not exist in classical physics.
Further information on the applications of entanglement in quantum computing and
QKD are available in Ref. [11]. In Chap. 18, applications of entanglement in
metrology (quantum imaging) will be examined.
References 83

References

1. A. Einstein, B. Podolsky and N. Rosen, Can quantum-mechanical description of physical

reality be considered complete?, Phys. Rev. 47 (1935) 777.
2. J.S. Bell, On the Einstein Podolsky Rosen paradox, Physics Physique Fizika 1 (1964) 195.
3. J.F. Clauser, M.A. Horne, A. Shimony and R.A. Holt, Proposed experiment to test local hidden-
variable theories, Phys. Rev. Lett. 23 (1969) 880.
4. Lecture notes of John Preskill, Chapter 4, http://theory.caltech.edu/~preskill/ph219/index.
html#lecture
5. A. Aspect, P. Grangier and G. Roger, Experimental tests of realistic local theories via Bell’s
theorem, Phys. Rev. Lett. 47 (1981) 460.
6. Attribution: The Royal Society. This file is licensed under the Creative Commons Attribution-
Share Alike 3.0 Unported license (https://creativecommons.org/licenses/by-sa/3.0/
deed.en). File: Alain-Aspect-ForMemRS.jpg. (2020, September 17). Wikimedia Commons,
the free media repository. Retrieved 14:41, December 7, 2020 from https://commons.
wikimedia.org/w/index.php?title¼File:Alain-Aspect-ForMemRS.jpg&oldid¼462669099
7. G. Weihs et al., Violation of Bell’s inequality under strict Einstein locality conditions, Phys.
Rev. Lett. 81 (1998) 5039.
8. C. Abellán et al., Challenging local realism with human choices, Nature 557 (2018) 212.
9. D. Rauch et al., Cosmic Bell test using random measurement settings from high-redshift
quasars, Phys. Rev. Lett. 121 (2018) 080403.
10. A.K. Ekert, Quantum cryptography based on Bell’s theorem, Phys. Rev. Lett. 67 (1991) 661.
11. R.R. LaPierre, Introduction to quantum computing (Springer, 2021).
Chapter 9
Multimode Quantized Radiation

In the previous chapters, we only considered single mode radiation, that is, radiation
with a single frequency, ω, or wavevector, k. In fact, single mode radiation is a “toy
model” or approximation, since all real radiation is multimode. In reality, an infinite
wave train of a single frequency doesn’t exist. In this chapter, we introduce a more
realistic description of light, called multimode light. The quantum optics treatment
for the detection of a spontaneously emitted single photon wavepacket is presented
as an example of multimode radiation.

9.1 Multimode Radiation

The general solution to Maxwell’s equations in vacuum for classical electromagnetic

waves is a superposition of fields:
X
Eðr, t Þ ¼ εl εl ðt Þeikl r þ c:c: ð9:1Þ
l

where l refers to the mode with polarization εl, amplitude εl, wavevector kl, and
frequency ωl (ωl ¼ ckl). The amplitude of mode l is given by

εl ðt Þ ¼ εl ð0Þeiωl t ð9:2Þ

9.2 Quantized Multimode Radiation

The canonical quantization of multimode radiation leads to the Hamiltonian:

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 85

R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_9
86 9 Multimode Quantized Radiation

X   X  
b¼
H ħωl Nbl þ 1 ¼ al { b
ħωl b al þ
1
ð9:3Þ
l
2 l
2

with energy eigenvalues

X  1

E¼ ħωl nl þ ð9:4Þ
l
2

where nl ¼ 0, 1, 2, . . . is the number of photons for each of the modes l ¼ 1, 2, . . .

Like Eq. (9.1), the most general solution for the quantized radiation field is a
superposition of field operators with different frequencies (recall Eq. (3.43)):
X  
b ðrÞ ¼
E iεl ε1l b al { eikl r
al eikl r  b ð9:5Þ
l

where ε1l is the one photon amplitude of mode l with frequency ωl:
rffiffiffiffiffiffiffiffiffiffi
ħωl
ε1l ¼ ð9:6Þ
2Eo V

Each of the modes is decoupled and can be treated as independent quantum

harmonic oscillators.
The creation and annihilation operator for each mode l do not commute:
h i
b a{l ¼ 1
al , b ð9:7Þ

while the operators for different modes, l and m, do commute:

h i
b a{m ¼ 0
al , b ð9:8Þ

Equations (9.7) and (9.8) may be written succinctly as

h i
b a{m ¼ δlm
al , b ð9:9Þ

The annihilation operator destroys one quantum of excitation (photon) of mode l:

pffiffiffiffi
b
al j n l i ¼ nl jnl  1i ð9:10Þ

while the creation operator creates one photon of mode l:

pffiffiffiffiffiffiffiffiffiffiffiffi
a{l jnl i ¼
b nl þ 1 jnl þ 1i ð9:11Þ
9.3 Vacuum Energy 87

The Fock state of mode l can be generated from the vacuum state:
 nl
a{l
b
jnl i ¼ pffiffiffiffiffi j0l i ð9:12Þ
nl !

In general, a multimode Fock state can be written as

j ψ i ¼ j n 1 i  j n 2 i  . . . ¼ j n1 i j n 2 i . . . ¼ j n 1 , n 2 , . . . i ð9:13Þ

where  denotes the tensor product. Alternatively, the tensor product can be written
more succinctly as jn1ijn2i. . . or jn1, n2, . . .i as shown in Eq. (9.13). An example of a
multimode state generated from the vacuum is
2 { n1  { n2   nl 3
b
a1 b
a2 a{l
b
j n 1 , n 2 , . . . , nl , . . . i ¼ 4 p ffiffiffiffiffiffi p ffiffiffiffiffiffi . . . ffiffiffiffiffi . . .5
p
n1 ! n2 ! nl !

 j 0 1 , 0 2 , . . . , 0l , . . . i ð9:14Þ

Equation (9.14) denotes a multimode quantized state with n1 photons in mode 1 of

energy ħω1 and momentum ħk1 ¼ ħω1 =c , n2 photons in mode 2 of energy ħω2
and momentum ħk2 ¼ ħω2 =c, etc. Each mode is generated from the corresponding
vacuum state j01i, j02i, etc. Equation (9.13) assumes that the state is factorizable.
There are also entangled states, which cannot be factorized in this manner. An
example of a multimode entangled state will be given in Sect. 9.4.
The multimode number operator is
X {
b¼
N b
al b
al ð9:15Þ
l

where
X
b jn1 , n2 , . . .i ¼ ðn1 þ n2 þ . . .Þjn1 , n2 , . . .i ¼
N nl jn1 , n2 , . . .i ð9:16Þ
l

P
nl represents the total number of photons present in the state jn1, n2, . . .i.
l

9.3 Vacuum Energy

Let us apply the Hamiltonian, Eq. (9.3), to the vacuum state j01, . . ., 0l, . . .i:
88 9 Multimode Quantized Radiation

X
b j01 , . . . , 0l , . . .i ¼ 1 ħ
H ωl j 0 1 , . . . , 0 l , . . . i ð9:17Þ
2 l

which gives the total energy

1 X
E ¼ ħ ωl ð9:18Þ
2 l

where l ¼ 1, 2, ⋯ corresponding to each of the modes. Since l spans all integers, the
vacuum energy is infinite! As we saw in Chap. 4, the vacuum has real consequences
in the presence of matter or boundary conditions (such as the Casimir effect).
Richard Feynman, Julian Schwinger, and Sin-Itiro Tomonaga were awarded the
1965 Nobel Prize in Physics for developing a sophisticated method of
“renormalization” used to deal with the infinity of the vacuum energy in calculations.
However, we usually do not need to worry about the infinite vacuum energy, since
we measure the photon energy relative to the vacuum level; that is, we can only
measure energy differences.

9.4 Single Photon Wavepacket

b with eigenvalue of 1, which could be a

A single photon state is an eigenstate of N
multimode state. An example of a single photon multimode state is the superposition

1
jψi ¼ pffiffiffi ðj11 i þ j12 iÞ ð9:19Þ
2

which is a single photon in mode 1 with frequency ω1 and in mode 2 with frequency
ω2. Note that
X {
b j ψi ¼
N b
al b
al jψi
l
 
1
¼ pffiffiffi b a{1b a{2b
a1 j 1 1 i þ b a2 j12 i
2 ð9:20Þ
1
¼ pffiffiffi ð1j11 i þ 1j12 iÞ
2
¼ 1j ψ i

b with eigenvalue 1; that is, by definition, jψi is a single

Thus, jψi is an eigenstate of N
photon state.
9.5 Spontaneous Emission 89

Exercise 9.1 Show that the single photon state given by Eq. (9.19) is not an
eigenstate of the Hamiltonian.

In general, a one photon wavepacket is given by

X
j ψi ¼ c l j 1l i ð9:21Þ
l

b with eigenvalue 1. The coefficients cl must satisfy the

which is an eigenstate of N
normalization condition:
X 2
j cl j ¼ 1 ð9:22Þ
l

9.5 Spontaneous Emission

As an example of the application of multimode radiation, we evaluate the detection

of a single photon wavepacket emitted by spontaneous emission. A single photon
wavepacket can be produced by spontaneous emission of a photon due to transition
of an electron from an excited state jei to a ground state jgi of an atom, as illustrated
in Fig. 9.1a. We suppose the electron was prepared in the excited state at time t ¼ 0
and has a lifetime γ 1. After some time t0  γ 1, a transition of the electron occurs
from the excited state to the ground state accompanied by the emission of a single
photon wavepacket. Here, it is useful to use the Heisenberg picture of the time-
dependence (Sect. 3.5). Like Eq. (9.21), the single photon wavepacket emitted at
some time t0, using the Heisenberg picture, is given by
X
jψðt 0 Þi ¼ cl j1l i ð9:23Þ
l

Fig. 9.1 (a) Energy (a) (b)

diagram and (b) Lorentzian 2
distribution of frequencies lifeme, γ−1
in a multimode single
photon wavepacket
produced by spontaneous ħ 0 = −
emission ∆ =γ
90 9 Multimode Quantized Radiation

We suppose the emission from an atomic transition, like Fig. 9.1a, has a Lorentzian
distribution of frequencies, centered at ω0 ¼ (Ee  Eg)/ħ and with width γ, as shown
in Fig. 9.1b (Exercise 9.2). The Lorentzian distribution of frequencies is due to
the finite lifetime, γ 1, of the excited state. According to the energy-time
uncertainty relation, a finite lifetime γ 1 produces a spread γ in frequency (hence,
a multimode state). The probability amplitudes cl of the modes l with frequency ωl
are described by

K
cl ¼ ð9:24Þ
ðωl  ω0 Þ þ iγ=2

where Δω ¼ γ is the width of the frequency distribution. Eq. (9.24) gives a

Lorentzian function for the probabilities jclj2 of the modes l:

K2
jcl j2 ¼ ð9:25Þ
ðωl  ω0 Þ2 þ γ 2 =4

Here, K is a normalization constant used to satisfy the normalization condition in

Eq. (9.22). K is derived in Appendix 3. Eq. (9.25) gives the natural linewidth of the
spontaneously emitted radiation in the absence of other broadening mechanisms.

Exercise 9.2 Why is the distribution of frequencies from a lifetime-limited

spontaneous emission described by a Lorentzian function?

Using the Heisenberg picture, the photodetector signal (number of photon

counts), I, like Eq. (5.9), is proportional to
 þ 2
b 
I / E ðrÞjψi ð9:26Þ
 2
X 
 i½k l zωl ðtt 0 Þ 
/ b al e jψðt 0 Þi ð9:27Þ
 l 

where t0 þ z/c is the delay time due to emission of the wavepacket at time t0 and
propagation delay over distance z from the source to the detector. Substituting the
wavepacket, Eq. (9.23), gives
 2
X 
 i½k l zωl ðtt 0 Þ 
I /  cl e j 0i  ð9:28Þ
 l 

aj1i ¼ j0i. Substituting kl ¼ ωcl and defining τ ¼ t  t0  z/c

where we have used b
gives
9.5 Spontaneous Emission 91

 2
X 
 iωl τ 
I /  cl e j0i ð9:29Þ
 l 

Substituting Eq. (9.24) gives

 2
X 
 Keiωl τ 
I/ j 0i  ð9:30Þ
 l ðωl  ω0 Þ þ iγ=2 

For a continuous frequency distribution, Eq. (9.30) becomes

 Z  2
 iω τ eiΩτ 
I/
Ke 0
dΩ j 0i 
 ð9:31Þ
Ω þ iγ=2

where Ω ¼ ωl  ω0 . The integral in Eq. (9.31) is the Fourier transform of a

Lorentzian. The integration results in
   2
 
I / Keiω0 τ 2πiH ðτÞeγτ=2 j0i ð9:32Þ

where H(τ) is the Heaviside or step function, shown in Fig. 9.2. Finally, the
photodetection signal is proportional to

Fig. 9.2 Heaviside or step

function, H(τ)
Number of counts

Fig. 9.3 Photodetector signal, I (number of photon counts) versus time for a single photon
wavepacket. The dashed line is the envelope of the wavepacket, representing the number of counts
measured at a photodetector. The electric field oscillations around frequency ω0, shown within the
envelope, are too rapid in the optical domain to be directly detected (there would be many more
oscillations than depicted here)
92 9 Multimode Quantized Radiation

Fig. 9.4 Time-correlated single photon counting (TCSPC) from a single indium arsenide (InAs)
quantum dot. Blue circles correspond to emission due to exciton recombination and red squares to
biexciton recombination. Black squares are the instrument response. The single exponential fit, like
Eq. (9.33), is shown by the solid white lines giving γ 1 in nanoseconds. (Reprinted by permission
from Springer Nature, M. Birowosuto et al. [1])

I / H ðτÞeγτ ð9:33Þ

Thus, the expected photodetector signal is a sharp increase followed by an expo-

nential decay as shown by the dashed line in Fig. 9.3a. The time-dependence
described by Eq. (9.33) is observed experimentally as shown, for example, in
Fig. 9.4 for indium arsenide quantum dots measured using “time-correlated single
photon counting” (TCSPC). The photodetector signal is obtained as a histogram
from many single photon detection events as a function of time.

Reference

1. M. Birowosuto et al., Fast Purcell-enhanced single photon source in 1,550-nm telecom band
from a resonant quantum dot-cavity coupling, Sci. Rep. 2 (2012) 321.
Chapter 10
Coherent State

A coherent state, such as that from a laser, is not just a bunch of photons. How do we
describe a state that is a coherent wave like that from a laser? We will see that the
coherent state can be described as a superposition of Fock states. The properties of
the coherent state are derived, including the Poisson distribution of photon number.
The quadrature components of the coherent state are derived, showing that the
coherent state is a minimum uncertainty state, leading to the shot noise limit. The
phasor representation of the electric field is introduced, and the number-phase
uncertainty relation is derived.

10.1 Coherent State

The definition of a coherent state is

b
ajαi ¼ αjαi ð10:1Þ

where the eigenvalue, α, is a complex number (α 2 ℂ) given by

α ¼ jαjeiφ ð10:2Þ

We are using the standard notation where α labels both the eigenstate and the
eigenvalue. The Hermitian conjugate of Eq. (10.1) gives

a{ ¼ hαjα
hαjb ð10:3Þ

According to Eq. (10.1), a coherent state is an eigenstate of the annihilation

operator with eigenvalue α. This means that a coherent state is also an eigenstate of
b þ with eigenvalue proportional to α. Thus, unlike Fock states that
the field operator E

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 93

R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_10
94 10 Coherent State

have average electric field equal to zero (hEi ¼ 0), we will see that Eq. (10.1) means
that the expectation value of the electric field will not vanish.

10.2 Coherent State as a Superposition of Fock States

It is useful to express jαi in terms of the basis states jni (superposition of single mode
Fock states):

X1
αn
jαi ¼ ejαj =2
2
pffiffiffiffi jni ð10:4Þ
n¼0 n!

We can check that Eq. (10.4) is correct by verifying that it satisfies the eigenvalue
equation, Eq. (10.1):

2 X1
αn
ajαi ¼ ejαj =2
b pffiffiffiffi b
ajni ð10:5Þ
n¼0 n!
pffiffiffi
Using Eq. (2.127), b
aj ni ¼ njn  1i, gives

2 X1
αn pffiffiffi
ajαi ¼ ejαj =2
b pffiffiffiffi njn  1i ð10:6Þ
n¼1 n!

Note that the index in Eq. (10.5) starts at n ¼ 0, while

pffiffi
the index in (10.6) starts at
n ¼ 1, since the lowest possible state is j0i. Since pffiffiffi
n
¼ pffiffiffiffiffiffiffiffiffi
1 ffi
, we get
n! ðn1Þ!

2 X
1
αn
ajαi ¼ ejαj =2
b pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi jn  1i ð10:7Þ
n¼1 ðn  1Þ!
!
2 X1
α n1
¼ α ejαj =2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi jn  1i ð10:8Þ
n¼1 ðn  1Þ!

We can change the starting index in Eq. (10.8) back to 0, giving the term in brackets
2 P
1
αffiffiffi
as jαi ¼ ejαj =2
n
p
n!
jni , which is identical to Eq. (10.4). Thus, Eq. (10.8)
n¼0
becomes

b
ajαi ¼ αjαi ð10:9Þ

which is Eq. (10.1).

10.3 Photon Number 95

Is Eq. (10.4) normalized? We need to verify that hαjαi ¼ 1:

2 X 1
ðα Þn ðαÞn
0

hαjαi ¼ ejαj pffiffiffiffipffiffiffiffiffi hnjn0 i ð10:10Þ

n,n0 ¼ 0 n! n0 !

The states jni and jn0i are orthonormal; that is, hnjn0 i ¼ δn,n0 , meaning that each term
in the summation is nonzero only if n ¼ n0, which gives

2 X1
jαj2n
hαjαi ¼ ejαj ð10:11Þ
n¼0
n!

The summation in Eq. (10.11) contains the well-known expansion for the exponen-
tial function:

X1
jαj2n
ejαj ¼
2
ð10:12Þ
n¼0
n!

Therefore, from Eq. (10.11),

2 2
hαjαi ¼ ejαj ejαj ¼ 1 ð10:13Þ

Thus, we have proven that the state jαi, as written in Eq. (10.4), is normalized.

Exercise 10.1 Show that two coherent states satisfy jhαjβij2 ¼ ejαβj . We
2

say that two coherent states are “quasiorthogonal”; that is, the two states
become increasingly orthogonal with the separation of α and β in the complex
plane.

10.3 Photon Number

The average photon number of a coherent state is

b
hni ¼ hαjNjαi ð10:14Þ

a{b
¼ hα j b aj αi ð10:15Þ
¼ hαjα αjαi ð10:16Þ
¼ jαj2 ð10:17Þ

Note that we can interpret Eq. (10.17) as giving a connection between the particle
(photon) view and the wave view. hni is the average photon number (particle view),
while jαj2 is proportional to the square of a field amplitude (the wave intensity).
96 10 Coherent State

Next, let us calculate the uncertainty in n. To do so, we first calculate hn2i:

 2
a{ b
n ¼ hαjb aba{ b
ajαi ð10:18Þ

Using Eqs. (10.1) and (10.3), we get

 2
n ¼ jαj2 hαjb
aba{ jαi ð10:19Þ

Putting this in the normal order gives

 2
n ¼ jαj2 hαjð1 þ ba{baÞjαi ð10:20Þ
 
¼ jαj2 1 þ jαj2 ð10:21Þ

¼ jαj2 þ jαj4 ð10:22Þ

¼ h n i þ h n i2 ð10:23Þ

The uncertainty in n is
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Δn ¼ hn2 i  hni2 ð10:24Þ

Using Eq. (10.23), we get

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Δn ¼ hni þ hni2  hni2 ð10:25Þ
pffiffiffiffiffiffiffi
¼ h ni ð10:26Þ

Equation (10.26) is known as the shot noise limit. Note that, unlike the Fock state, a
coherent state is not an eigenstate of the number operator, since there is a dispersion
(uncertainty) in the photon number. Also note that the coherent state is not an
eigenstate of the Hamiltonian. However, it is easy to calculate the average energy,
b
hαjH jαi ¼ hαjħω (b {
b
a a þ 2Þjαi ¼ ħω hni þ 12 .
1

10.4 Poisson Distribution

In Eq. (10.4), we expressed the coherent state as a superposition of Fock states:

X
1
2 X1
αn
jαi ¼ cn jni ¼ ejαj =2 pffiffiffiffi jni ð10:27Þ
n¼0 n¼0 n!
10.4 Poisson Distribution 97

Fig. 10.1 Poisson 0.2

distribution Pn for average
photon number hni ¼ 5,
0.16
10 and 20

0.12

Pn
0.08

0.04

0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
n

We are interested in the probability distribution of the photon number, Pn ¼ jcnj2:

jαj2n
Pn ¼ jcn j2 ¼ ejαj
2
ð10:28Þ
n!

or, using Eq. (10.17):

h ni n
Pn ¼ ehni ð10:29Þ
n!

Equation (10.29) is the Poisson probability distribution! Pn is the probability of

detecting n photons in a fixed time interval if hni is the average number of photons in
that time interval. The Poisson distribution, Pn, is shown in Fig. 10.1 for various
average values hni. You can see that the spread (uncertainty) increases as
n increases. It can be shown that, as hni increases, the p Poisson
ffiffiffiffiffiffiffi distribution
approaches a Gaussian distribution with standard deviation of hni [1]:

hnin large n 1 
ðnhniÞ2
P n ¼ e h n i

! Pn ¼ pffiffiffiffiffiffiffiffiffiffiffiffi e 2hni ð10:30Þ

n!ffl}
|fflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflffl 2π hni
Poisson distribution
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
Gaussian distribution

Suppose we have a source with hni ¼ 1. The probability of observing 0, 1, 2 or

3 photons from Eq. (10.29) is (note that the factorial of 0 is equal to 1)

P0 ¼ 0:368 ð10:31Þ
P1 ¼ 0:368 ð10:32Þ
P2 ¼ 0:184 ð10:33Þ
98 10 Coherent State

P3 ¼ 0:061 ð10:34Þ

Unlike the single photon Fock state with hni ¼ 1, there is a probability of 0.184 of
measuring two photons simultaneously in a coherent state with hni ¼ 1, and an equal
probability of 0.368 for measuring 0 or 1 photon. Therefore, a coherent source with
hni ¼ 1 is not a good single photon source!
Suppose we attenuate a coherent source (e.g., attenuate a laser beam), such that
hni  1. Would this produce a good single photon source? Equation (10.29) gives

P0  1 ð10:35Þ
P1  hni ð10:36Þ
h ni 2
P2  ð10:37Þ
2

P0 is less than but close to 1. This means that a measurement would result in zero
photons most of the time. A single photon would be detected with probability hni,
2
two photons with probability hn2i , etc. Note that the sum of probabilities will yield 1 if
we continue with the higher-order terms. For hni  1, like in a strongly attenuated
laser source, there is still a finite (albeit low) probability for double photon detection,
and most of the time zero photons will be measured in a given time interval. You
cannot produce an efficient single photon source by attenuating a coherent source!

10.5 Electric Field of Coherent State

Recall that the electric field operator is

 
b ðrÞ ¼ iεε1 b
E a{ eikr
aeikr  b ð10:38Þ

The average electric field for a coherent state is

aeikr  b
hEðrÞi ¼ iεε1 hαjðb a{ eikr Þj αi ð10:39Þ
 
¼ iεε1 αeikr  α eikr ð10:40Þ

¼ iεε1 αeikr þ c:c: ð10:41Þ

If we include the time-dependence (see Sect. 10.7), we get

hEðr, t Þi ¼ iεε1 αeiðkrωtÞ þ c:c: ð10:42Þ

Using Eq. (10.2), we get

10.5 Electric Field of Coherent State 99

hEðr, t Þi ¼ 2εε1 jαj sin ðk  r  ωt þ φÞ ð10:43Þ

Equation (10.43) looks like a classical monochromatic traveling plane wave and is
different than the number (Fock) state where hEi ¼ 0. The coherent state is the
quantum description of light that most closely resembles classical coherent light,
such as that from a laser. However, unlike the classical description of light, the
electric field of the coherent state has an uncertainty that we calculate below. For this
reason, coherent states are often called “quasiclassical” states.
As usual, let us find the uncertainty in electric field, ΔE, by first calculating hE2i.
Using Eq. (10.38):

 
b 2 jαi
E2 ¼ hαjE ð10:44Þ
 2
¼  ε1 hαjðb aeikr  ba{ eikr Þðb
aeikr  ba{ eikr Þj αi ð10:45Þ
 2 2
¼  ε1 hαjðb a2 e2ikr  b
aba{  b a{ b
a þ ðba{ Þ e2ikr Þj αi ð10:46Þ
 2 2
a2 e2ikr  ð1 þ b
¼  ε1 hαjðb a{ b
aÞ  ba{ ba þ ðba{ Þ e2ikr Þj αi ð10:47Þ
 2 h i
¼  ε1 ðαÞ2 e2ikr  1  2α α þ ðα Þ2 e2ikr ð10:48Þ

From Eqs. (10.41) and (10.48), we get

   2
E2 ¼ hEi2 þ ε1 ð10:49Þ

Using Eq. (10.49), the uncertainty ΔE is

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
ΔE ¼ E2  hEi2 ð10:50Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ hEi2 þ ðε1 Þ2  hEi2 ð10:51Þ

¼ ε1 ð10:52Þ

Thus, according to our discussion in Chap. 4, a coherent state is a minimum

uncertainty state with field fluctuations ε1.
100 10 Coherent State

10.6 Phasor Representation

Choosing r ¼ 0 for simplicity, we can write Eq. (10.43) as

hEðt Þi ¼ 2εε1 jαj sin ðωt þ φÞ ð10:53Þ

Using Eq. (10.53), hE(t)i can be represented as a rotating phasor (rotating point on
a circle in the complex plane), as shown in Fig. 10.2a, with amplitude 2ε1jαj, initial
phase φ at t ¼ 0, and rotating clockwise at rate ω. Note the negative sign in the
amplitude of Eq. (10.53). Thus, hE(t)i is given by the projection onto the imag-
inary axis, as shown in Fig. 10.2b. According to Eq. (10.30), the gray circle in
Fig. 10.2a represents a Gaussian distribution (as depicted in Fig. 10.1) in the limit of
large hni ¼ jαj2; that is, large phasor amplitude. The standard deviation ΔE ¼ ε1 of
the Gaussian distribution gives an uncertainty in the field of ε1 around the average,
and a width of 2ε1 represented by the diameter of the gray circle in Fig. 10.2a. The
signal-to-noise ratio (SNR) may be given by

2jαjε1 pffiffiffiffiffiffiffi
SNR ¼ 1
¼ jαj ¼ hni ð10:54Þ
2ε

Classical optics (including laser light) can be explained as a macroscopic limit of

quantum optics. For example, suppose a laser contains hni~10 10
pffiffiffiffiffiffiffiphotons in its
cavity. The uncertainty in photon number from Eq. (10.26) is hni  105 , which
is much less than hni. As α (or hni) increases, the field becomes more classical. Note

(a) Rotang phasor in complex plane: (b) Projecon onto imaginary axis

Fig. 10.2 (a) Phasor representation of the field hE(t)i for the coherent state. (b) Projection of the
rotating phasor onto the imaginary axis gives hE(t)i with uncertainty ε1 represented by the dashed
lines and red arrows
10.7 Time-Dependence of Coherent State 101

that the electric field uncertainty ε1 remains constant, but becomes negligible in
comparison to the amplitude for large electric fields; that is, the gray circle in
Fig. 10.2a becomes a relatively small rotating point and approximates the phasor
for a classical wave.

10.7 Time-Dependence of Coherent State

Repeating Eq. (10.27), the coherent state is a superposition of Fock states:

X
1
2 X1
αn
jαi ¼ cn jni ¼ ejαj =2 pffiffiffiffi jni ð10:55Þ
n¼0 n¼0 n!

The Fock states are an eigenstate of the Hamiltonian with eigenvalue En ¼ nħω
(ignoring the vacuum energy). We know the time-dependence of the Fock states in
the Schrodinger picture is jnðt Þi ¼ eiEn t=ħ jni . Thus, the time-dependence of the
coherent state is given by

X
1
jαðt Þi ¼ cn eiEn t=ħ jni ð10:56Þ
n¼0

or, using Eq. (10.55), we get

2 X1
αn
jαðt Þi ¼ ejαj =2
pffiffiffiffi einωt jni ð10:57Þ
n¼0 n!

The time-dependence of the state from Eq. (10.57) can be written succinctly as
 
jαðt Þi ¼ αð0Þeiωt ð10:58Þ

or, in terms of the eigenvalue:

αðt Þ ¼ αð0Þ eiωt ð10:59Þ

We see that a coherent state remains a coherent state for all time. However, a
coherent state is not an eigenstate of the Hamiltonian. Therefore, a coherent state
evolves in time as shown in Fig. 10.2.
Let us find the time-dependent expectation values for the Q and P quadratures.
Recalling Eqs. (2.44) and (2.45), in the Heisenberg picture, we get

b 1 1  
Q ¼ hQi ¼ hαj Qjαi ¼ pffiffiffi hαjðb a{ Þjαi ¼ pffiffiffi αeiωt þ α eiωt
aþb ð10:60Þ
2 2
102 10 Coherent State

Using Eq. (10.2), we get

 
1
Q ¼ pffiffiffi jαjeiðωtþφÞ þ jαjeiðωtþφÞ ð10:61Þ
2
pffiffiffi
¼ 2jαj cos ðωt þ φÞ ð10:62Þ

Similarly,

b
P ¼ hPi ¼ hαjPjαi ð10:63Þ
i
¼ pffiffiffi hαjðb
ab a{ Þjαi
2
i  
¼ pffiffiffi αeiωt  α eiωt
2
 
i
¼ pffiffiffi jαjeiðωtþφÞ  jαjeiðωtþφÞ
2
pffiffiffi
¼ 2jαj sin ðωt þ φÞ ð10:64Þ

Recalling the classical harmonic oscillator from Chap. 2, Q represented the dimen-
sionless position and P represented the dimensionless momentum. We see that the
expectation values, Q and P for the electric field, oscillate with time and are 90 out
of phase with each other just like the position and momentum of a classical harmonic
oscillator. It is also in this sense that coherent states are the “most classical of states”,
since they are analogous to the dynamics of a classical harmonic oscillator. The
coherent state produces an oscillating Gaussian wavepacket (Fig. 10.3, red),
analogous to a classical particle oscillating in a parabolic potential. If the wavepacket
is to reproduce a classical particle, we better make sure that its width is not
changing with position or time as it oscillates. In the next section, we show that
the Gaussian wavepacket of the coherent state is a minimum uncertainty state whose
width (ΔQ, ΔP) remains the same for all time.

Fig. 10.3 A particle oscillating in a parabolic potential, U(x). The quantum analogue is a Gaussian
wavepacket (red) representing the particle position (Q quadrature), while the momentum is
represented by the P quadrature
10.8 Quadratures 103

10.8 Quadratures

In the Schrodinger picture (set t ¼ 0 in Eqs. (10.62) and (10.64)), we have

pffiffiffi pffiffiffi 1
Q¼ 2jαj cos φ ¼ 2 Re ðαÞ ¼ pffiffiffi ðα þ α Þ ð10:65Þ
2

and

pffiffiffi pffiffiffi i
P¼ 2 jαj sin φ ¼ 2 ImðαÞ ¼ pffiffiffi ðα  α Þ ð10:66Þ
2

Next, let us find the uncertainties, starting with hQ2i

 
b 2 jαi ¼ 1 hαjðb
Q2 ¼ hαjQ aþb a{ Þðb
aþb a{ Þjαi ð10:67Þ
2
1
¼ hαjðbabaþb a{ þ b
ab a{ b
aþba{ba{ Þjαi ð10:68Þ
2

Using the commutation relation, ba{ ¼ 1 þ b

ab a{ b
a, for normal ordering gives

 2 1
Q ¼ hαjðb ab a{b
a þ 1 þ 2b aþb a{ b
a{ Þjαi ð10:69Þ
2
 
1
¼ ðαÞ2 þ 1 þ 2jαj2 þ ðα Þ2 ð10:70Þ
2
h i
1
¼ ð α þ α Þ 2 þ 1 ð10:71Þ
2

Thus, the uncertainty is

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

ΔQ ¼ Q2  hQi2 ð10:72Þ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2ffi
h i 
1 1
¼ ðα þ α Þ2 þ 1  pffiffiffi ðα þ α Þ ð10:73Þ
2 2
1
¼ pffiffiffi ð10:74Þ
2

Similarly,
104 10 Coherent State

1
ΔP ¼ pffiffiffi ð10:75Þ
2

Thus,

1
ΔQΔP ¼ ð10:76Þ
2

Hence, there is an uncertainty relation between the electric field at a certain point in
time and the electric field at a quarter cycle later. The coherent state is a minimum
uncertainty state satisfying the standard quantum limit, like the ground state of the
quantum harmonic oscillator. A coherent state distributes its quantum mechanical
uncertainties equally between the Q and P quadratures.

Exercise 10.2 Derive Eq. (10.75).

The quadrature representation, shown in Fig. 10.4, represents α in the complex

plane,p according
ffiffiffiffiffiffiffi to Eq. (10.2), with magnitude (radius of the circle) given by
jαj ¼ hni and with initial phase φ. The projections give the real and imaginary
parts of α according to Eqs. (10.65) and (10.66). The gray circle represents what
would be obtained from many different measurements of P and Q with spread given
by Eqs. (10.74) and (10.75).
We remind the reader from Eq. (3.49) that the electric field can be written in terms
of the quadrature operators as

Fig. 10.4 Quadrature

representation of the
coherent state
10.9 Displacement Operator 105

pffiffiffiffi h i
b ðr, t Þ ¼ εε1 2 P
E b sin ðk  r  ωt Þ
b cos ðk  r  ωt Þ þ Q ð10:77Þ

giving

b ðr, t Þjαi
hEðr, t Þi ¼ hαjE ð10:78Þ
¼ 2εε1 jαj ½ sin φ cos ðk  r  ωt Þ þ cos φ sin ðk  r  ωt Þ ð10:79Þ

Is there a way to fix φ, so that we can detect the quadrature components? In fact,

there is! The method is known as homodyne detection, which is treated in Chap. 13.

10.9 Displacement Operator

Note that the vacuum state gave b aj0i ¼ 0: Thus, comparing with Eq. (10.1), the
vacuum state can be considered as a coherent state with α ¼ 0. It is useful to think of
the coherent state as a “displaced” vacuum, as depicted in Fig. 10.5. The displace-
ment can be performed by a displacement operator given by
 { 
b αb
a αb
DðαÞ ¼ e a
ð10:80Þ

Fig. 10.5 Coherent state

represented as a displaced
vacuum

coherent
state

vacuum
state
106 10 Coherent State

We can rewrite this operator using the following theorem for the exponential of
operators, often called Glauber’s formula:
 
b b bb 1bb
eAþB ¼ eA eB e2 A, B ð10:81Þ
h i
b B
where A, b 6¼ 0 and providing that

h h ii h h ii
b A,
A, b B
b ¼ B, b B
b A, b ¼0 ð10:82Þ

b ¼ αb
The latter conditions are met with A b ¼ αb
a{ and B a from Eq. (10.80). In this
case,
h i
b B
A, b ¼ jαj2 ð10:83Þ

The (tedious!) proof of Eq. (10.81) can be done by the Taylor expansion of the
exponentials.

b ¼ αb
Exercise 10.3 Prove Eqs. (10.82) and (10.83) for A b ¼ α ab:
a{ and B

Using the above results, we get

 {  {
b ðαÞ ¼ e αba αba ¼ e12jαj eαba eαba
2
D ð10:84Þ

b ðαÞ applied
which is just another form of the displacement operator. Let us evaluate D
to the vacuum state j0i. First, using the Taylor expansion of the exponential gives

X1
ðαbaÞ n
eα ba j0i ¼

j 0i ¼ j 0i ð10:85Þ
n¼0
n!

since b
an j0i ¼ 0, except for n ¼ 0. Next,
 n
{ X
1 a{
αb X1  
αn { n
eαba j0i ¼ j0i ¼ b
a j0i ð10:86Þ
n¼0
n! n¼0
n!

Using Eq. (2.97) gives

{ X1
αn
eαba j0i ¼ pffiffiffiffi jni ð10:87Þ
n¼0 n!
10.10 Number-Phase Uncertainty Relation 107

Thus,

{ X1
αn
b ðαÞj0i ¼ e12jαj eαba eαba j0i ¼ e12jαj
2 2
D pffiffiffiffi jni ð10:88Þ
n¼0 n!

which is just the number representation of the coherent state. Thus, the displacement
operator takes the Gaussian wavepacket of the j0i state (see Eq. (2.57) and Fig. 2.1)
and translates it, creating a coherent state with the same uncertainty. Similarly,
within a global phase factor, Db ðβÞjαi ¼ jα þ βi, which results in displacement of
the coherent state itself.

b ðβÞjαi ¼ jα þ βi within a phase factor.

Exercise 10.4 Show that D

10.10 Number-Phase Uncertainty Relation

As shown in Fig. 10.6, the quadrature spread also corresponds to a spread in

amplitude, Δα, and phase, Δφ. Using the small angle approximation, the spread in
phase from Fig. 10.6 is

Fig. 10.6 Number–phase

uncertainty relation in the
complex plane (phasor
representation of α)

Radius,
108 10 Coherent State

ΔP ΔP
Δφ ¼ ¼ pffiffiffiffiffiffiffi ð10:87Þ
R h ni

where we have used:

R2 ¼ jαj2 ¼ hni ð10:88Þ

Differentiating Eq. (10.88), we obtain the spread in photon number:

pffiffiffiffiffiffiffi
Δn ¼ 2RΔR ¼ 2 hniΔQ ð10:89Þ
pffiffiffiffiffiffiffi
where we have used R ¼ hni and ΔR ¼ ΔQ (see Fig. 10.6). Thus,
!
 pffiffiffiffiffiffiffi 
ΔP
ΔnΔφ ¼ pffiffiffiffiffiffiffi 2 hniΔQ ¼ 2ΔQΔP ð10:90Þ
h ni

Using Eq. (10.76), we obtain the number–phase uncertainty relation:

ΔnΔφ
1 ð10:91Þ

As the photon number becomes larger, the phase becomes less uncertain, that is, we
approach a classical state.

10.11 Revisiting the Fock State

Now that the phasor representation (Fig. 10.2a) has been introduced, we can revisit
the Fock state and intuitively understand it using the phasor representation. Recall
that the photon number is well defined for the Fock state with zero uncertainty. The
average electric field of the Fock state is zero, but the uncertainty in electric field is
nonzero. How can we reconcile these results in a phasor representation? A Fock state
can be pictured as a superposition of many phasors, as depicted in Fig. 10.7a. The
photon number, represented by the phasor amplitude, is well defined, but the phase
angle is completely undefined, in accordance with the number–phase uncertainty
relation. The resulting electric fields in Fig. 10.7b are given by the projection of the
many phasors on the imaginary axis in Fig. 10.7a. The superposition of all these
fields results in zero average electric field, hEi ¼ 0. However, the field fluctuations
will be nonzero, such that hE2i 6¼ 0 and thus ΔE 6¼ 0.
Reference 109

(a) (b)

Fig. 10.7 (a) Fock state represented by a superposition of phasors (only a few are shown in red).
(b) Resulting electric fields by projection of the phasors on the imaginary axis in (a), resulting in
zero average electric field, hEi ¼ 0, but nonzero field fluctuations, hE2i 6¼ 0 and ΔE 6¼ 0

Reference

1. John R. Taylor, An introduction to error analysis: The study of uncertainties in physical

measurements (2nd ed., Univ. Science Books, 1997).
Chapter 11
Coherent State on a Beam Splitter

In this short chapter, the photodetection probability of the coherent state on a beam
splitter and the probability of coincidence measurements are derived. Unlike the
single photon state, double photon counts (coincidences) occur for the coherent
state, as we would expect for a classical coherent source like a laser.

11.1 Photodetection Probability

Let us consider a coherent state on a beam splitter, as shown in Fig. 11.1. The input
state is jαi1j0i2 with a coherent state on port 1 and “nothing” (vacuum) on port
2. The probability of detection at D3 is

hnout i hn3 i
P3 ¼ ¼ 2 ð11:1Þ
hnin i jαj

where we have used hnini ¼ jαj2 for the coherent state input. Let us determine hn3i:

a{3b
hn3 i ¼ hψout jb a3 jψout i ð11:2Þ

Writing Eq. (11.2) in terms of the input space, we get

a{1 þ t b
hn3 i ¼ 2 h0j1 hαjðr b a{2 Þðrb
a1 þ tb
a2 Þjαi1 j0i2 ð11:3Þ

a{1b
¼ 2 h0j1 hαjðr  rb a{1b
a1 þ r  tb a{2b
a2 þ t  rb a{2b
a1 þ t  tb a2 Þjαi1 j0i2 ð11:4Þ

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 111
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_11
112 11 Coherent State on a Beam Splitter

Fig. 11.1 Coherent state on

a beam splitter

Remember that the operators act on the state with the same subscript. Thus,

a{1b
2 h0j1 hαjb a2 jαi1 j0i2 ¼0 ð11:5Þ

a{2b
2 h0j1 hαjb a1 jαi1 j0i2 ¼0 ð11:6Þ

a{2b
2 h0j1 hαjb a2 jαi1 j0i2 ¼0 ð11:7Þ

Using Eqs. (11.5), (11.6) and (11.7), Eq. (11.4) becomes

a{1b
hn3 i ¼ 2 h0j1 hαjr  r b a1 jαi1 j0i2 ð11:8Þ

¼ r r αα ¼ Rjαj 2
ð11:9Þ

Thus, from Eq. (11.1),

P3 ¼ R ð11:10Þ

which is the same as the classical result.

Exercise 11.1 Derive Eqs. (11.5), (11.6) and (11.7).

Similarly, the probability of detection at D4 is

hnout i hn4 i
P4 ¼ ¼ 2 ð11:11Þ
hnin i jαj

where hn4i is

a{4b
hn4 i ¼ hψout jb a4 jψout i
ð11:12Þ
a{1  r b
¼ 2 h0j 1 hαjðt b a{2 Þðtb
a1  rb
a2 Þjαi1 j0i2
11.2 Coincidence Measurements 113

 
a{1b
¼ 2 h0j1 hαj t  tb a{1b
a1  t  rb a{2b
a2  r  tb a{2b
a1 þ r  rb a2 jαi1 j0i2 ð11:13Þ

Keeping only the non-zero terms gives

a{1b
hn4 i ¼ 2 h0j1 hαjt  t b a1 jαi1 j0i2 ð11:14Þ

¼ t  t αα ¼ T jαj2 ð11:15Þ

Thus,

P4 ¼ T ð11:16Þ

which is the same as the classical result.

11.2 Coincidence Measurements

The probability of simultaneous detection at D3 and D4 (correlation or coincidence

measurement) is

a{3b
hn34 i ¼ hψout jb a{4b
a4 b
a3 jψout i ð11:17Þ

Expressed in terms of the input space, we get

a{1 þ t b
hn34 i ¼ 2 h0j1 hαjðr b a{2 Þðt b
a{1  r b
a{2 Þðtb
a1  rb
a2 Þ
 ðrb a1 þ tb a2 Þjαi1 j0i2 ð11:18Þ

Note that, if we expand the terms, Eq. (11.18) is in the normal order. Retaining only
the non-zero terms, we get

a{1 Þðt b
hn34 i ¼ 1 hαjðr b a{1 Þðtb
a1 Þðrb
a1 Þjαi1 ð11:19Þ

¼ r t  tr jαj4 ð11:20Þ
¼ RT jαj4 ð11:21Þ

Normalizing with hni2 ¼ jαj4, we get

P34 ¼ RT ð11:22Þ

which is the same as the classical result. This is what we would expect from a
coherent source such as a laser.
114 11 Coherent State on a Beam Splitter

The second-order correlation function, g(2)(0), is easily calculated. We have

P3 ¼ R, P4 ¼ T, and P34 ¼ RT. Thus,

P34
gð2Þ ð0Þ ¼ ¼1 ð11:23Þ
P3 P4

Coherent sources do not produce anticorrelation. This result could also be derived
from the result of Chap. 6 (Eq. (6.61)):

ðΔnÞ2  hni
gð2Þ ð0Þ ¼ 1 þ ð11:24Þ
h ni 2

For the coherent state, we found (Δn)2 ¼ hni, that is, the shot noise limit. Thus,
according to Eq. (11.24), g(2)(0) ¼ 1, identical to Eq. (11.23).

Exercise 11.2 What is the probability of detection at D3 and D4 if the

coherent source in Fig. 11.1 is replaced with a Fock state, jni1? What would
be the probability of double detection? How does this differ from the results
for a coherent state? What measurements could you make to distinguish the
Fock state from the coherent state?
Chapter 12
Incoherent State

In the previous chapter, we studied coherent states like that produced by a laser.
However, the light encountered in almost all situations of everyday life is incoherent
light (also called chaotic light or thermal light)—for example, from an incandescent
source (resistance filament), the Sun, blackbody radiation, etc. In this chapter, the
quantum optics treatment of incoherent or thermal light is introduced. We derive the
properties of incoherent light, including the photon number distribution and corre-
lation function, and compare with other types of light.

12.1 Incoherent State

Recall from Chap. 10 that coherent sources are described by a well-defined phase:

α ¼ jαjeiφ ð12:1Þ

In contrast, incoherent (thermal) sources are made of many independent emitters:

jψi ¼ jα1 i  jα2 i  . . . jαl i . . . ¼ Πl jαl i ð12:2Þ

αl ¼ jαl je iφl
ð12:3Þ

The incoherent state is described by a multimode state where the phase φl is

randomly distributed from 0 to 2π.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 115
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_12
116 12 Incoherent State

12.2 Electric Field of Incoherent State

Before describing the incoherent state, we summarize some key results for the
coherent state. The electric field operator for a mode l is
 
b ðr, t Þ ¼ iεl ε1 b
E a l e iðkl rωl t Þ
 b
a l
{ iðkl rωl t Þ
e ð12:4Þ
l

where we have included the time-dependence (Heisenberg picture). Equation (12.4)

can be written as a sum of positive and negative frequency components:

b þ ðr, t Þ þ E
b ðr, t Þ ¼ E
E b  ðr, t Þ ð12:5Þ

where

b þ ðr, t Þ ¼ iεl ε1b

E iðkl rωl t Þ
ð12:6Þ
l al e

h þ i{
b  ðr, t Þ is the Hermitian conjugate of E
and E b þ ðr, t Þ; that is, E
b  ðr, t Þ¼ E
b ðr, t Þ .
When applied to the coherent state, Eq. (12.5) gives the average electric field as

hEðr, t Þi ¼ iεl ε1l αl eiðkl rωl tÞ þ c:c: ð12:7Þ

which looks like a classical electric field.

The electric field for an incoherent state can be written as a sum of coherent fields
with random phases:
X
hEðr, t Þi ¼ iεl ε1l jαl jeiφl eiðkl rωl tÞ þ c:c: ð12:8Þ
l

which can be written as a sum of positive and negative frequency components:

hEðr, t Þi ¼ Eþ ðr, t Þ þ E ðr, t Þ ð12:9Þ

where
X
Eþ ðr, t Þ ¼ iεl ε1l jαl jeiφl eiðkl rωl tÞ ð12:10Þ
l

and E(r, t) is the complex conjugate of Eþ(r, t); that is, E(r, t) ¼ [Eþ(r, t)]. In the
incoherent state, the phases φl of each mode are randomly distributed, so the
summation involving eiφl is zero. The result is a summation of phasors with random
phase angles, resulting in zero average electric field. Equivalently, the average over
many modes l of eiφl , represented by an overbar, is zero:
12.3 Photodetector Signal 117

eiφl ¼ 0 ð12:11Þ

Thus, from Eqs. (12.9) and (12.10), the average electric field is zero:

hEðr, t Þi ¼ 0 ð12:12Þ

12.3 Photodetector Signal

Similar to Chap. 5, the photodetector signal (intensity or number of counts) associ-

ated with the incoherent state can be written as

b þ ðr, t Þjψijj
2
I / jjE ð12:13Þ
P
b þ ðr, t Þ ¼
where jψi is the incoherent state and E aeiðkl rωl tÞ is the positive
iεl ε1l b
l
frequency component of the field operator. Eþ(r, t) is the eigenvalue of the operator
b þ ðr, t Þ. Thus, Eq. (12.13) becomes
E

I / jjEþ ðr, tÞjψijj ¼ jEþ ðr, tÞj ¼ E ðr, tÞEþ ðr, tÞ

2 2
ð12:14Þ

where Eþ(r, t) is given by Eq. (12.10). This results in a double summation:

I / E ðr, t ÞEþ ðr, t Þ ¼ Σl Σm ε1l ε1m jαl jjαm jeiðφl φm Þ ei½ðkl km Þrðωl ωm Þt ð12:15Þ

The summation only contributes a finite value when φl ¼ φm; otherwise, the
summation over many modes is zero. Equivalently, the average of the phase term is

eiðφl φm Þ ¼ δlm ð12:16Þ

Hence, with l ¼ m in Eq. (12.15), we get the measured intensity as

X 2 2
I/ ε1l jαl j ð12:17Þ
l

which is equal to the sum of the intensities from the individual modes without any
cross terms from different modes.
118 12 Incoherent State

12.4 Photon Number Distribution

Max Planck (Fig. 12.1a) correctly described the blackbody radiation spectrum
(Fig. 12.1b) by treating the atomic vibrations of the material at finite temperature
as quantum harmonic oscillators (QHOs) [1]. Later, Einstein proposed the  quanti-

zation of light itself. The probability of a QHO having energy E n ¼ ħω n þ 12 is
given by a Boltzmann distribution:

eEn =kT
Pn ¼ ð12:18Þ
Z

where Z is the partition function familiar from statistical thermodynamics:

X
1
Z¼ eEn =kT ð12:19Þ
n¼0

P
1
Z is a normalization factor that ensures Pn ¼ 1: Thus, substituting
  n¼0
E n ¼ ħω n þ 12 into Eq. (12.19) gives

X
1
Z ¼ eħω=2kT enħω=kT ð12:20Þ
n¼0

Eq. (12.20) contains a geometric series:

Fig. 12.1 (a) Max Planck (Nobel Prize in Physics in 1918). (Credit: AIP Emilio Segrè Visual
Archives, Gift of Jost Lemmerich). (b) Blackbody radiation spectrum for different temperatures
12.5 Photon Number Uncertainty 119

X
1
1
enħω=kT ¼ ð12:21Þ
n¼0
1  eħω=kT

which gives

eħω=2kT
Z¼ ð12:22Þ
1  eħω=kT

Hence,
 
Pn ¼ enħω=kT 1  eħω=kT ð12:23Þ

Pn ultimately leads to the Planck distribution law for blackbody radiation derived in
Appendix 4 and shown in Fig. 12.1b.
From Eq. (12.23), the average photon number is (Exercise 12.1)

X
1
1
h ni ¼ nPn ¼ ð12:24Þ
n¼0
eħω=kT  1

Equation (12.24) is the famous Bose–Einstein distribution for photons. For example,
at room temperature (300 K) and λ¼500 nm, we get hni~1042! At 6000 K (surface
of the Sun) and λ¼500 nm, we still only get hni~102.

Exercise 12.1 Derive Eq. (12.24).

From Eqs. (12.23) and (12.24), we get (Exercise 12.2):

h ni n
Pn ¼ ð12:25Þ
ð1 þ hniÞnþ1

This probability distribution is shown in Fig. 12.2 for various values of hni. Note that
the most probable photon number is n ¼ 0, and the probability distribution decays
exponentially with n as expected from Eq. (12.18).

Exercise 12.2 Derive Eq. (12.25).

12.5 Photon Number Uncertainty

To calculate the photon number uncertainty, Δn, for incoherent light, we start with
the following relation:
120 12 Incoherent State

Fig. 12.2 Probability distribution Pn for thermal light with (a) hni ¼ 0.01, (b) hni ¼ 0.1, (c)
hni ¼ 1, and (d) hni ¼ 5

 2 X
n  hni ¼ hniðhni  1Þ ¼ nðn  1ÞPn ð12:26Þ
n

Evaluating the summation gives (Exercise 12.3)

X
nðn  1ÞPn ¼ 2hni2 ð12:27Þ
n

Thus, from Eq. (12.26),

 2
n ¼ 2hni2 þ hni ð12:28Þ

Finally, the uncertainty is

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Δn ¼ hn2 i  hni2 ð12:29Þ
12.6 Comparison of Different Types of Light 121

Substituting Eq. (12.28) gives

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Δn ¼ hni2 þ hni ð12:30Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ hniðhni þ 1Þ ð12:31Þ

The uncertainty or fluctuation described by Eq. (12.31) for incoherent light is super-
Poissonian;
pffiffiffiffiffiffiffi that is, the fluctuations are greater than Poissonian where Δn was equal
to hni (the shot noise limit).

Exercise 12.3 Derive Eq. (12.27).

12.6 Comparison of Different Types of Light

Now that we have described incoherent light, we can review the different types of
light covered thus far. A comparison of the photon number distribution for the
different types of light is shown in Fig. 12.3 for an average photon number hni ¼ 5.
The Fock state, j5i, has 5 photons with no uncertainty (Δn ¼ 0). The coherent state
follows a Poisson distribution, while the incoherent light (thermal light) has the
distribution given by Eq. (12.25). As might be expected, the thermal distribution is
broader than that from a coherent source (note the change in horizontal scale).

Exercise 12.4 Plot the probability distributions in Fig. 12.3 for hni ¼ 1.

Fig. 12.3 Comparison of photon number distribution for different types of light for hni ¼ 5
122 12 Incoherent State

Table 12.1 Correlation func- State g2(0) Photon statistics

tion for different types of light
Classical 1 Super-Poissonian,
Bunched
Fock state <1 Sub-Poissonian,
Antibunched
Coherent state 1 Poissonian
Incoherent state 2 Super-Poissonian,
Bunched

Recall that we can measure correlations or coincidences between two detectors in

a Hanbury Brown–Twiss experiment and obtain the correlation function g2(0),
which is an important method of characterizing light sources. The second-order
correlation function, g2(0), for thermal light can be derived from the result of
Chap. 6:

ðΔnÞ2  hni
gð2Þ ð0Þ ¼ 1 þ ð12:32Þ
h ni 2

For thermal light, we found (Δn)2 ¼ hni2 þ hni. Thus, according to Eq. (12.32),
g(2)(0) ¼ 2. Even if you take a thermal source and spectrally and spatially filter it to
look like a laser, it can be distinguished from a laser by a second-order correlation
function of 2 rather than 1 for a laser (coherent source). The thermal state has a
higher probability to emit more than one photon at the same time. This effect is
called photon bunching. Table 12.1 summarizes the correlation function for different
types of light.
Recall that the variance of the photon number for a Fock state is zero. Any light
with a sub-Poissonian distribution of photon number, like the Fock state, is called
antibunched. Antibunched photons are distributed more uniformly in time as com-
pared to the photons in a coherent (Poissonian) beam having the same average
number of photons per unit time. Thus, there will be less variance in the number
of photons counted by a photodetector per unit time for antibunched light compared
to coherent light. This would reduce any noise in a measurement based on counting
photons.
Until now, we have only considered the g2(0) correlation function with no time
delay between the measurement at one detector and the other. In general, we can
measure g2(τ) where τ is the time delay between measurements at the two detectors.
To determine g2(τ), we must consider many modes and how they interfere with each
other, which is described by a coherence time. The bunching or antibunching only
happens for time delays (τ) shorter than the coherence time, which is typically very
short for thermal light as assumed in Eq. (12.17). Thus, the g2(τ) function will appear
as shown qualitatively in Fig. 12.4.
Reference 123

Fig. 12.4 Correlation 2.

function, g2(τ), for different thermal
types of light: thermal or (bunched)
bunched light (red),
coherent light (green), and coherent
antibunched light from a 1.
Fock or number state (the
case of a single photon
SPS
source (SPS) is shown; blue) (an-bunched)

Interphoton delay,

Reference

1. M. Planck, Annalen der Physik 4 (1901) 553-563.

Chapter 13
Homodyne and Heterodyne Detection

In this chapter, homodyne detection and heterodyne detection on a beam splitter are
introduced. Homodyne detection is a powerful yet simple method for measuring the
quadrature components of electric field. Using heterodyne detection, we can down-
convert a high-frequency signal to lower frequency.

13.1 Homodyne Detection

In this section, we consider the method of homodyne detection (Fig. 13.1).

Homodyne detection means that the frequency of the state on each of the two
input ports of a beam splitter or interferometer is equal:

Homodyne detection: ω1 ¼ ω2 ð13:1Þ

Consider a beam splitter with reflection coefficient and transmission coefficient

equal to p1ffiffi2:

1
r ¼ t ¼ pffiffiffi ð13:2Þ
2

This gives a reflectance R ¼ 12 and transmittance T ¼ 12, that is, a 50:50 beam splitter.
The input state is

jψin i ¼ jψi1 jαLO i2 ð13:3Þ

where jαLOi2 is a coherent state called the “local oscillator” (LO), and jψi1 is also
typically a coherent state.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 125
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_13
126 13 Homodyne and Heterodyne Detection

Fig. 13.1 Homodyne

detection using a beam
splitter

The homodyne signal is the difference in photodetector signals measured between

detectors D3 and D4, which is proportional to the difference in photon number
counts, hn3i  hn4i:

Homodyne signal / hn3 i  hn4 i ð13:4Þ

Let us begin by calculating hn3i:

a{3b
hn3 i ¼ hψout jb a3 jψout i ð13:5Þ

Expressed in terms of the input state, we have

1 {
a þb
hn3 i ¼ hψin j ðb a{2 Þðba1 þ b
a2 Þjψin i ð13:6Þ
2 1
1 {
ab
¼ hψin j ðb a þb a{1b
a2 þ b a{2b a{2b
a1 þ b a2 Þjψin i ð13:7Þ
2 1 1

The factor of 12 comes from the reflection and transmission coefficients. Similarly,

a{4b
hn4 i ¼ hψout jb a4 jψout i ð13:8Þ
1 {
¼ hψin j ðb a b a{2 Þðb
a1  ba2 Þjψin i ð13:9Þ
2 1
 
1 {
¼ hψin j b a1  b
a1b a{1b
a2  ba{2b
a1 þ b a{2b
a2 jψin i ð13:10Þ
2

Thus, the homodyne signal is

 
hn3 i  hn4 i ¼ hψin j b a{1b
a2 þ ba{2b
a1 jψin i ð13:11Þ
 
a{1b
¼ 2 hαLO j1 hψ1 j b a2 þ ba{2b
a1 jψ1 i1 jαLO i2 ð13:12Þ
13.1 Homodyne Detection 127

a{1 þ αLOb
¼ hψ1 jðαLOb a1 Þjψ1 i ð13:13Þ

n3  b
Note that b n4 is Hermitian, and therefore can be an observable.
In general, αLO is a complex number and can be written as

αLO ¼ jαLO jeiφLO ð13:14Þ

Hence,
 
a{1 þ eiφLO b
hn3 i  hn4 i ¼ jαLO j hψ1 j eiφLO b a1 jψ1 i ð13:15Þ

Using the Euler relation for eiφLO and eiφLO , we get

hn3 i  hn4 i ¼ jαLO j

D h i E
a{1 þ ð cos φLO  i sin φLO Þb
 ψ1 j ð cos φLO þ i sin φLO Þb a1 jψ1
ð13:16Þ

Rearranging, we obtain

pffiffiffi  
1
h n3 i  h n4 i ¼ 2jαLO j½ cos φLO hψ1 jpffiffiffi b a{1 þ b
a1 jψ1 i ð13:17Þ
2
i
a{1  b
þ sin φLO hψ1 jpffiffiffi ðb a1 Þjψ1 i
2

b and P
Finally, from the definition of the quadrature operators Q b in Eqs. (2.44) and
(2.45), respectively, we obtain
pffiffiffi
hn3 i  hn4 i ¼ b 1 i þ sin φLO hψ1 jPjψ
2jαLO j½ cos φLO hψ1 jQjψ b 1 i ð13:18Þ

Thus, we can measure each quadrature of the state jψ1i by choosing the phase of the
local oscillator, φLO ¼ 0 or π/2. Also, note that jαLOj can be large, giving amplifi-
cation to the homodyne signal.
Suppose jψ1i is a coherent state, jψ1i ¼ jα1i, with eigenvalue given by

α1 ¼ jα1 jeiφ1 ð13:19Þ

Then,
128 13 Homodyne and Heterodyne Detection

Fig. 13.2 Quadrature

representation of a coherent
state. The Q quadrature is
obtained when φLO ¼ 0, and
the P quadrature is obtained
when φLO ¼ π/2

  pffiffiffi pffiffiffi
b 1 i ¼ p1ffiffiffi α þ α1 ¼ 2 Reðα1 Þ ¼ 2jα1 j cos φ1
Q ¼ hQi ¼ hα1 j Qjα ð13:20Þ
1
2
  pffiffiffi pffiffiffi
b 1 i ¼ pi ffiffiffi α1  α1 ¼ 2Imðα1 Þ ¼ 2jα1 j sin φ1
P ¼ hPi ¼ hα1 j Pjα ð13:21Þ
2

and

hn3 i  hn4 i ¼ 2jαLO j ½ cos φLO ðjα1 j cos φ1 Þ þ sin φLO ðjα1 j sin φ1 Þ ð13:22Þ

We see that Q and P are related to the real and imaginary parts of α1, and can be
represented in the complex plane as shown in Fig. 13.2. We already saw this in
Chap. 10, but now we have a way of measuring the quadrature components by
selection of the local oscillator phase φLO using homodyne detection. The
Q quadrature is obtained when φLO ¼ 0, and the P quadrature is obtained when
φLO ¼ π/2. We can think of the local oscillator as a “strobe light” that takes a
snapshot of the light field of jα1i at periodic times.

13.2 Heterodyne Detection

In the previous section, we considered homodyne detection where the frequency of

the two input signals is equal. In this section, we briefly consider heterodyne
detection, where the two input ports contain coherent sources, jα1i1 and jα2i2, at
13.2 Heterodyne Detection 129

Fig. 13.3 Heterodyne

detection using a beam
splitter

two different frequencies, ω1 6¼ ω2, as shown in Fig. 13.3. jα2i2 is the local oscillator,
while jα1i1 is the signal of interest.
Let us calculate the average photon number at D3:

a{3b
hn3 i ¼ hψout jb a3 jψout i ð13:23Þ
1 {
¼ hψin j ðb a þb a{2 Þðb
a1 þ b a2 Þjψin i ð13:24Þ
2 1
 
1 {
¼ hψin j b a1 þ b
a1 b a{1b
a2 þ b a{2b
a1 þ b a{2b
a2 jψin i ð13:25Þ
2
 
1
¼ 2 hα2 j1 hα1 j ba{1b
a1 þ b a{1b
a2 þ b a{2b
a1 þ b a{2b
a2 jα1 i1 jα2 i2 ð13:26Þ
2

If we include the time-dependence in the operators (Heisenberg picture), we get

 
1
hn3 ðt Þi ¼ jα1 j2 þ jα2 j2 þ jα1 jjα2 jei½ðω1 ω2 Þtþφ2 φ1  þ jα1 jjα2 jei½ðω1 ω2 Þtþφ2 φ1 
2
ð13:27Þ
 
1
¼ jα1 j2 þ jα2 j2 þ jα1 jjα2 j cos ½ðω1  ω2 Þt þ φ2  φ1  ð13:28Þ
2

We obtain the familiar form for the interference of two coherent signals. A hetero-
dyne signal with angular frequency, ω1  ω2, is observed. Using heterodyne
detection, small-frequency shifts of a signal from the local oscillator frequency can
be measured, which is important for many applications such as Doppler lidar.

Exercise 13.1 Describe some other applications of heterodyne detection.

Chapter 14
Coherent State in an Interferometer

We revisit the coherent state and determine the probability of single and double
photon detection in an interferometer and the expression for homodyne detection.
The uncertainty and signal-to-noise ratio (SNR) of the homodyne signal is analyzed,
leading to an important conclusion—the SNR for the coherent state arises from the
uncertainty in the field quadrature of the vacuum input to the interferometer.

14.1 Coherent Light Interference

Let us examine the output of a coherent state in an interferometer as shown in

Fig. 14.1. We have
   
a3 ¼ t 2 eikz1  r 2 eikz2 b
b a1 þ rteikz1  treikz2 ba2 ð14:1Þ

The probability of detection at D3 is

hn3 i hψout jb
n3 jψout i
P3 ¼ ¼ ð14:2Þ
hnin i jαj2

Converting to the input state, retaining only the non-zero terms, and assuming r and
t are real numbers gives
h   { i
2 h0j1 hαj t 2 eikz1  r 2 eikz2 t 2 eikz1  r 2 eikz2 b
a1 b
a1 jαi1 j0i2
P3 ¼ ð14:3Þ
jαj2
 
¼ R2 þ T 2  2RT cos ðkΔzÞ ð14:4Þ

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 131
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_14
132 14 Coherent State in an Interferometer

Fig. 14.1 Coherent state in

an interferometer

which is the same as the classical result. For the usual case of a 50:50 beam splitter,
Eq. (14.4) gives
 
kΔz
P3 ¼ sin 2
ð14:5Þ
2

Similarly,
   
b
a4 ¼ rteikz1 þ treikz2 ba1 þ r 2 eikz1 þ t 2 eikz2 ba2 ð14:6Þ

The probability of detection at D4 is

hn4 i hψout jb
n4 jψout i
P4 ¼ ¼ ð14:7Þ
hnin i jαj2

Converting to the input state, retaining only the non-zero terms, and assuming r and
t are real numbers gives
h   { i
2 h0j1 hαj rteikz1 þ treikz2 rteikz1 þ treikz2 b
a1 b
a1 jαi1 j0i2
P4 ¼ ð14:8Þ
jαj2
¼ 2RT þ 2RT cos ðkΔzÞ ð14:9Þ

which is the same as the classical result. For the usual case of a 50:50 beam splitter,
Eq. (14.9) gives
14.3 Homodyne Signal 133

 
kΔz
P4 ¼ cos 2
ð14:10Þ
2

Note that P3 þ P4 ¼ 1, as required for probabilities.

14.2 Coincident Detection

The probability of double detection at D3 and D4 is

a{3b
hψout jb a{4b
a4 b
a3 jψout i
P34 ¼ ð14:11Þ
jαj4

Retaining the non-zero terms gives

   
2 h0j1 hαjt 2 eikz1  r 2 eikz2 rteikz1 þ treikz2 rteikz1 þ treikz2
 2 ikz  { {
t e 1  r 2 eikz2 b a1 b
a1 b
a1 b
a1 jαi1 j0i2
P34 ¼ ð14:12Þ
jαj4
¼ P3 P4 ð14:13Þ

which is the same as the classical result.

Exercise 14.1 Derive Eq. (14.13).

14.3 Homodyne Signal

Let us determine the homodyne signal of a coherent state in an interferometer, as

shown in Fig. 14.2. Note that we moved the coherent state to port 2. The input state is

jψin i ¼ j0i1 jαi2 ð14:14Þ

with vacuum on port 1 and the coherent state on port 2. The operators are
   
b
a3 ¼ t 2 eikz1  r 2 eikz2 ba1 þ rteikz1  treikz2 b a2 ð14:15Þ
 ikz   
a4 ¼ rte 1 þ treikz2 b
b a1 þ r 2 eikz1 þ t 2 eikz2 ba2 ð14:16Þ

Let us consider the usual case of a 50:50 beam splitter. Equations (14.15) and (14.16)
simplify to
134 14 Coherent State in an Interferometer

Fig. 14.2 Homodyne

detection in an
interferometer

1  ikz1  1 
b
a3 ¼ e  eikz2 b a1  eikz1 þ eikz2 ba2 ð14:17Þ
2 2
1  1 
a4 ¼ eikz1 þ eikz2 b
b a1  eikz1  eikz2 ba2 ð14:18Þ
2 2

The annihilation operators are

1  ikz1  { 1  ikz  {
a{3 ¼
b e  eikz2 b
a1  e 1 þ eikz2 b
a2 ð14:19Þ
2 2
1  { 1  ikz  {
a{4 ¼ eikz1 þ eikz2 b
b a1  e 1  eikz2 b
a2 ð14:20Þ
2 2

From Eqs. (14.17), (14.18), (14.19) and (14.20), the number operators are

i
n3 ¼ b
b a{3b a{1b
a3 ¼ sin 2 ðkΔz=2Þb a1 þ a{1b
sin ðkΔzÞb a2
2 ð14:21Þ
i
 a{2b
sin ðkΔzÞb a{2b
a1 þ cos 2 ðkΔz=2Þb a2
2

and

i
n4 ¼ b
b a{4b a{1b
a4 ¼ cos 2 ðkΔz=2Þb a1  a{1b
sin ðkΔzÞb a2
2 ð14:22Þ
i
þ a{2b
sin ðkΔzÞb a{2b
a1 þ sin 2 ðkΔz=2Þb a2
2

where Δz ¼ z1  z2.

Exercise 14.2 Derive Eqs. (14.21) and (14.22).

14.3 Homodyne Signal 135

The average photon numbers at D3 and D4 are

a{3b
hn3 i ¼ 2 hαj1 h0jb a3 j0i1 jαi2 ¼ jαj2 cos 2 ðkΔz=2Þ ð14:23Þ

a{4b
hn4 i ¼ 2 hαj1 h0jb a4 j0i1 jαi2 ¼ jαj2 sin 2 ðkΔz=2Þ ð14:24Þ

Exercise 14.3 Derive Eqs. (14.23) and (14.24).

Let us take the homodyne signal as hn4  n3i ¼ hn4i  hn3i. From Eqs. (14.23)
and (14.24), we get
    
kΔz 2 kΔz
hn4 i  hn3 i ¼ jαj 2
sin 2
 cos ð14:25Þ
2 2

¼ jαj2 cos ðkΔzÞ ð14:26Þ

This signal is shown in Fig. 14.3.

Suppose we want to measure small displacements (e.g., movements of the
mirrors) corresponding to a path length difference or phase change, ε. The greatest
sensitivity to displacement occurs near φ ¼ π2 or 3π
2 where the slope of the homodyne
signal, hn4i  hn3i, in Fig. 14.3 is steepest. Let us consider small phase changes
ε from π2; that is,

π
kΔz ¼ þε ð14:27Þ
2

Equation (14.26) becomes


π
hn4 i  hn3 i ¼ jαj2 cos þε ð14:28Þ
2

Fig. 14.3 Homodyne signal

for the interferometer in
Fig. 14.2
136 14 Coherent State in an Interferometer

Using a trigonometric identity, Eq. (14.28) becomes

h

 i
π π
hn4 i  hn3 i ¼ jαj2 cos cos ðεÞ  sin sin ðεÞ ð14:29Þ
2 2
¼ jαj2 sin ε ð14:30Þ

Since jαj2 ¼ hn2i for the coherent state,

hn4 i  hn3 i ¼ hn2 i sin ε ð14:31Þ

According to the Taylor expansion, small ε gives:

h n4 i  h n3 i  h n 2 i ε ð14:32Þ

Equation (14.32) indicates that we have good sensitivity to displacements in the

interferometer if hn2i is large; that is, we want an intense laser.

14.4 Uncertainty in the Homodyne Signal

Let us determine the uncertainty in hn4i  hn3i. First, we need to determine

h(n4  n3)2i. From Eqs. (14.21) and (14.22), we have

n4  b
b a{4b
n3 ¼ b a4  ba{3b
a3 ð14:33Þ
    
kΔz kΔz
¼ cos 2  sin 2 a{1b
b a{1b
a1  i sin ðkΔzÞb a2
2 2
    
{ 2 kΔz 2 kΔz
a2 b
þ i sin ðkΔzÞb a1 þ sin  cos a{2b
b a2 ð14:34Þ
2 2

a{1b
¼ cos ðkΔzÞb a1  i sin ðkΔzÞba{1b a{2b
a2 þ i sin ðkΔzÞb a1  cos ðkΔzÞba{2b
a2 ð14:35Þ



¼ cos ðkΔzÞ b a{1b a{2b
a1  b a2  i sin ðkΔzÞ b a{1b a{2b
a2  b a1 ð14:36Þ

Using Eq. (14.27):





π π
b
n4  b
n3 ¼ cos þε a{1b
b a{2b
a1  b a2  i sin þε b a{1b
a2  ba{2b
a1 ð14:37Þ
2 2



¼  sin ε ba{1b a{2b
a1  b a2  i cos ε b a{1b a{2b
a2  b a1 ð14:38Þ
14.4 Uncertainty in the Homodyne Signal 137

Using Eq. (14.38) for small displacements (ε ! 0), ðb n3 Þ2 becomes:

n4  b



n3 Þ 2 ¼  b
n4  b
ðb a{1b a{2b
a2  b a1 ba{1b a{2b
a2  b a1 ð14:39Þ

a{1b
¼ b a{1b
a2 b a{1b
a2 þ b a{2b
a2 b a{2b
a1 þ b a{1b
a1 b a{2b
a2  b a{2b
a1 b a1 ð14:40Þ

Using the commutation relation to put the second term in the normal order, we get


ðb a{1b
n3 Þ2 ¼ b
n4  b a{1b
a2 b a{1 1 þ b
a2 þ b a{2b a{2b
a1 þ b
a2 b a{1b
a1 b a{2b
a2  b a{2b
a1 b a1 ð14:41Þ

Using Eq. (14.41), we can now evaluate the average, h(n4  n3)2i. For simplicity,
let us assume α is a real number (α 2 ℝ; equivalent to zero phase angle), giving:
h   { i
a{1b
hðn4  n3 Þ2 i ¼ 2 hαj1 h0j α2b a{1 þ 1 þ α2 b
a1 b
a1 þ α2b a{1  α2b
a1 b a1 b
a1 j0i1 jαi2
ð14:42Þ

We assume a strong coherent state, so 1 þ α2~α2, giving:


a{1b
hðn4  n3 Þ2 i ¼ α2 1 h0j b a{1  b
a{1b
a1  b a{1 þ b
a1 b a1 b
a1 j 0 i 1 ð14:43Þ

The first, second, and fourth terms are zero. Applying the normal ordering to the
third term gives

a{1b
hðn4  n3 Þ2 i ¼ α2 1 h0jð1 þ b a1 Þ j 0 i 1 ð14:44Þ
¼ α2 ¼ hn2 i ð14:45Þ

Finally, we can evaluate the uncertainty:

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Δðn4  n3 Þ ¼ h ð n 4  n 3 Þ 2 i  h n4  n 3 i ð14:46Þ

Using Eqs. (14.32) and (14.45) gives

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi
Δðn4  n3 Þ ¼ h n 2 i  ε h n 2 i  h n2 i ð14:47Þ

The uncertainty is equal to the shot noise limit, as shown in Fig. 14.4. The signal-to-
noise ratio (SNR) may be defined as
138 14 Coherent State in an Interferometer

Fig. 14.4 Homodyne signal

showing the uncertainty,
pffiffiffiffiffiffiffiffiffi
hn2 i, indicated by the
dashed curves and red arrow

hn2 iε pffiffiffiffiffiffiffiffiffi
SNR ¼ pffiffiffiffiffiffiffiffi
ffi ¼ hn2 i ε ð14:48Þ
h n2 i

The only way to increase the SNR is to increase the source power or the measure-
ment time, corresponding to an increase in hn2i. This approach has its limitations due
to damage to the optical
pffiffiffiffiffiffiffiffiffi system caused by exposure to high optical power.
How does the hn2 i uncertainty arise? From Eq. (14.43), we have

2
a{1  b
hðn4  n3 Þ2 i ¼ α2 1 h0j b a1 j0i1 ð14:49Þ

b operator is
Recall that the P


i
b¼p
P ffiffiffi b a{
ab ð14:50Þ
2

Thus,

b 2 j 0i 1
hðn4  n3 Þ2 i ¼ 2α2 1 h0jP ð14:51Þ

¼ 2hn2 iðΔPÞ2 ð14:52Þ

If we substitute ΔP ¼ p1ffiffi2 in Eq. (14.52) for the vacuum state, we reproduce

Eq. (14.45). We see that the SNR for the coherent state arises from the uncertainty
in the P quadrature of the vacuum on port 1! This is another of the experimental
consequences of the vacuum state, in addition to others presented in Chap. 4 (Lamb
shift, Casimir effect, etc.). In the next two chapters, we will see how we can use this
insight to improve the SNR in an interferometer below the shot noise limit.
Chapter 15
Squeezed Light

Thus far, we have described Fock states, coherent light, and incoherent light.
Quantum optics has discovered many other states of light. In this chapter, we
examine one of the most useful of these, called “squeezed light”. We show that
the uncertainty in the phase or amplitude quadrature of squeezed light can be reduced
as compared to coherent light, making squeezed light very useful in metrology. The
squeezed vacuum state is introduced, and the fragility of the squeezed state is
explained.

15.1 Classical Description of Nonlinear Optics

One of the new states of light discovered in quantum optics is “squeezed light”,
which allows us to surpass the shot noise limit. Although squeezed light had been
theoretically studied for a long time, the first experimental success for producing
squeezed light was by Slusher et al. in 1985 [1]. The most efficient means of
producing squeezed light has used parametric oscillation in nonlinear media. Exam-
ples of nonlinear materials include lithium niobate (LiNbO3) or potassium titanyl
phosphate (KTP).
The classical description of the parametric oscillator involves the nonlinear
dielectric polarization (dipole moment per unit volume). “Parametric” means that
some parameter of the oscillator varies periodically in time (e.g., periodically
varying the length of a swinging pendulum). In the optical parametric process, a
strong coherent source (a pump laser) modulates the polarization of a nonlinear
material due to a nonlinear dependence of the polarization P on the applied field. The
pump field is a strong coherent field (laser), which is our parametric drive. The
dependence of the polarization on the total electric field is given by (do not confuse
polarization P with the P quadrature)

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 139
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_15
140 15 Squeezed Light

Fig. 15.1 Production of Squeezed

squeezed light using a coherent
nonlinear material state

PðE Þ ¼ E0 χð1Þ E þ E0 χð2Þ E2 þ ⋯higherorder terms ð15:1Þ

|fflfflffl{zfflfflffl} |fflfflfflffl{zfflfflfflffl}
Pð1Þ P ð 2Þ

where χ(1) is the first-order linear electric susceptibility, χ(2) is the second-order
electric susceptibility (a nonlinear term), and so on for higher-order nonlinear terms.
Equation (15.1) can be understood as a power series, which arises physically from an
anharmonic crystal potential in the nonlinear medium. Typical values for χ(1) are on
the order of unity (χ(1) is dimensionless), χ(2) is typically on the order of 1012 m/V,
and successive higher-order terms usually decrease quickly (some terms may be zero
due to crystal symmetry). Thus, the nonlinear terms of Eq. (15.1) are much less than
the linear term and become evident only in the presence of a strong electric field
(e.g., from a laser).
Consider Fig. 15.1 where a coherent field at angular frequency ω and pump field
at 2ω are put into a nonlinear material such as KTP. The total applied electric field
(assuming the position r ¼ 0) can be expressed as

E ¼ A cos ðωt þ φÞ þ B cos ð2ωt Þ ð15:2Þ

The output field is proportional to the polarization P of the material. According to

Eq. (15.1), the linear term is

Pð1Þ ðE Þ ¼ E0 χð1Þ ½A cos ðωt þ φÞ þ B cos ð2ωt Þ ð15:3Þ

and the second-order term is

 
Pð2Þ ðEÞ ¼ E0 χð2Þ A2 cos 2 ðωt þ φÞ þ B2 cos 2 ð2ωt Þ þ 2AB cos ðωt þ φÞ cos ð2ωt Þ
ð15:4Þ

Using trigonometric identities, Eq. (15.4) can be expressed purely in cosine terms:
15.2 Quantum Description of Squeezing 141

(
ð2Þ ð2Þ 1 2 1
P ðE Þ ¼ E0 χ A ½1 þ cos ð2ωt þ 2φÞ þ B2 ½1 þ cos ð4ωt Þ ð15:5Þ
2 2
)
þ AB½ cos ðωt  φÞ þ cos ð3ωt þ φÞ

P(2) contains a constant component and oscillating components at frequency ω, 2ω,

3ω and 4ω. The ω component of the second-order nonlinear term, Pðω2Þ , is

Pðω2Þ ¼ E0 χð2Þ AB cos ðωt  φÞ ð15:6Þ

From Eq. (15.3), the linear term of the same frequency is

Pðω1Þ ¼ E0 χð1Þ A cos ðωt þ φÞ ð15:7Þ

The Pðω1Þ and Pðω2Þ terms can constructively interfere if φ ¼ 0. This constructive
interference process is called optical parametric amplification (OPA). On the other
hand, destructive interference occurs if φ ¼ π/2. This destructive interference
process can also work on the vacuum state, producing the “squeezed vacuum
state”. Thus, the optical nonlinearity creates correlations between photons that can
be exploited to reduce the noise below the shot noise limit.

15.2 Quantum Description of Squeezing

In Fig. 15.1, we refer to the input and output as mode b and mode a, respectively. In
the quantum description of squeezing, two photons at frequency ω (the down-
converted photons) are created in the output mode a, and one photon at frequency
2ω is destroyed in the input mode b (see Fig. 5.2b in Chap. 5). The reverse process is
b Hermitian). The interaction Hamiltonian
also possible (it is also required to make H
takes the form
 {

b ¼ ħg b
H ab
abb þb a{ b
a{ b b ð15:8Þ
 { 
a2b a{ b
2
¼ ħg b b þb b ð15:9Þ

where g is a coupling constant related to the nonlinear susceptibility terms in

Eq. (15.1). We assume the mode b input is a strong coherent state (e.g., a laser),
which is our parametric drive. Thus, for a coherent state,
142 15 Squeezed Light

b
bjβi ¼ βjβi ð15:10Þ

where β is given by

β ¼ jβjeiφ ð15:11Þ

Equation (15.10) states that the coherent state jβi is an eigenstate of the annihilation
operator with eigenvalue β, which is the definition of the coherent state. We also
assume that jβi is an eigenstate of the creation operator with eigenvalue β:

{
b
b jβi  β jβi ð15:12Þ

In other words, we assume that we can replace the annihilation and creation
operators with the complex numbers, β and β, respectively, which is valid for a
strong coherent classical source where adding or removing one photon from the field
makes a negligible difference. We can then rewrite the Hamiltonian in Eq. (15.9) as
 
b ¼ ħgjβj b
H a{2 eiφ
a2 eiφ þ b ð15:13Þ

15.3 Squeezing Operator

Using the unitary evolution operator, we define the squeezing operator, b

S:
^
b
S ¼ e ħ
iHt
ð15:14Þ

b is the Hamiltonian from Eq. (15.13). Thus, we get

where H
2 {2

S ¼ eigjβjt ba e þba e
b iφ iφ
ð15:15Þ

Choosing φ ¼  π/2 gives

 2 {2 
S ¼ e2 ba ba
b R
ð15:16Þ

where R ¼ 2gjβjt is called the squeeze parameter. The factor of 2 in R is present by

convention. We can tune R by selecting the phase φ. Note that b b is
S is unitary since H
Hermitian, resulting in
15.3 Squeezing Operator 143

1 {
b
S ðRÞ ¼ b
S ðRÞ ¼ b
SðRÞ ð15:17Þ

What does the squeezing operator bS do? The squeezing operator applied to a
coherent state produces a new state:

b
Sjαi ¼ jα, Ri ð15:18Þ

The new state is called the squeezed state, denoted by jα, Ri. To characterize this
state, let us find the average electric field, quadrature values (Q, P), and uncertainties
(ΔE, ΔQ, ΔP) for this new state.
{
First, we need to find hα, Rjb ajα, Ri ¼ hαjbSb ab
Sjαi. For this purpose, we use the
Baker–Hausdorff formula:
h i h h ii h h h iii
bb bB b 1 b b 1 b b b 1 b b b b
eB Ae ¼Aþ B, A þ B, B, A þ B, B, B, A þ . . . ð15:19Þ
1! 2! 3!

The Baker–Hausdorff formula can be proven using the Taylor expansion of the
exponential functions:
! !
bb bB b2
B b2
B
eB Ae ¼ b þ þ ... A
1þB b 1B
b þ þ ... ð15:20Þ
2! 2!
   2 
bþ B
¼A bA
bA bBb þ1 B b AbþA
bBb2  2B
bAbB
b þ ... ð15:21Þ
2!

Using the definition of the commutation relation, Eq. (15.21) is identical  to

b
Eq. (15.19). We consider the case for which A ¼ b b
a and B ¼  2 bR
a b
2 {2
a . It is
h i h i
b b
easy to show that B, a ¼ Rb {
a and B, b b{
a ¼ Rb a: Thus, after using the Baker–
Hausdorff formula, we get

b{ R2 R3
Sb ab
S¼b a{ R þ þ . . .
a 1 þ þ ... b ð15:22Þ
2! 3!

The terms in brackets are the Taylor expansion for the hyperbolic functions,
coshR and sinhR:

{
b
Sb ab
S¼b a{ sinh R
a cosh R  b ð15:23Þ

where

eR þ eR
cosh R ¼ ð15:24Þ
2
144 15 Squeezed Light

and

eR  eR
sinh R ¼ ð15:25Þ
2

Similarly,

{ {
b
Sbab a{ cosh R  b
S¼b a sinh R ð15:26Þ

With Eqs. (15.23) and (15.26) in our toolbox, we can find the expectation value and
uncertainty for the field and its quadratures, as shown in the following sections.

  h i
b¼b
Exercise 15.1 Show that for A b ¼ R b
a and B a 2
 b
a {2
, b b
B, a a{
¼ Rb
2
h i
b b
and B, a{ ¼ Rb a.

15.4 Electric Field of Squeezed Light

Recall that the quantum operator for the electric field (Heisenberg picture) is
 
b ðr, t Þ ¼ iεε1 b
E a{ eiðk ∙ rωtÞ
aeiðk ∙ rωtÞ  b ð15:27Þ

The average electric field for the squeezed state is

b ðr, t Þjα,Ri
hEi = hα,RjE ð15:28Þ

Using Eqs. (15.23) and (15.26) gives

hEi ¼ iεε1 ðα cosh R  α sinh RÞeiðk ∙ rωtÞ þ c:c: ð15:29Þ

¼ iεε1 α0 eiðk ∙ rωtÞ þ c:c: ð15:30Þ

where

α0 ¼ α cosh R  α sinh R ð15:31Þ

If α is a real number (α 2 ℝ), then

hEi ¼ 2εε1 α0 sin ðk ∙ r  ωt Þ ð15:32Þ

15.4 Electric Field of Squeezed Light 145

Equation (15.30) or (15.32) resembles a coherent state with amplitude α0, similar to
Eqs. (10.42) and (10.43), respectively. Thus, the squeezing of a coherent state jαi
produces another coherent state jα, Ri. However, as we will see below, the new
coherent state jα, Ri (the squeezed state) has some special properties.
If α 2 ℝ, then according to Eqs. (15.24), (15.25) and (15.31),

eR þ eR eR  eR
α0 ¼ α  ð15:33Þ
2 2

¼ α eR ð15:34Þ

For simplicity, let us choose the position r ¼ 0. The uncertainty in electric field
becomes (Exercise 15.2)
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ΔE ¼ ε1 e2R cos 2 ðωt Þ þ e2R sin 2 ðωt Þ ð15:35Þ

If the squeezing parameter, R < 0, then the minimum field uncertainty occurs when
ωt ¼ 0, π, . . .:

R < 0 : ΔE min ¼ ε1 eR when ωt ¼ 0,π, . . . ð15:36Þ

Conversely, the maximum field uncertainty occurs when t ¼ π2 , 3π

2 , . . .:

π 3π
R < 0 : ΔE max ¼ ε1 eR when ωt ¼ , , . . . ð15:37Þ
2 2

We see that the uncertainty depends on time. There are times (ωt ¼ 0, π, . . .) when
the uncertainty is less than ε1 and there are times (ωt ¼ π2 , 3π2 , . . . ) when the
uncertainty is greater than ε . We can think of the squeezed state as an oscillating
1

Gaussian wavepacket, like the coherent state (Fig. 10.3), but whose width (standard
deviation) varies with time.
Conversely, if R > 0:

π 3π
R > 0 : ΔE min ¼ ε1 eR when ωt ¼ , , . . . ð15:38Þ
2 2
R > 0 : ΔE max ¼ ε1 eR when ωt ¼ 0,π, . . . ð15:39Þ

Thus, the time (or phase) when squeezing occurs depends on the squeeze
parameter, R.
146 15 Squeezed Light

Exercise 15.2 Derive Eq. (15.35).

The average electric field, Eq. (15.32), can be represented in the complex plane by
a rotating phasor, as shown in Fig. 15.2a (with r ¼ 0), and with the uncertainty
represented by the gray oval. Figure 15.2 represents the case where R < 0. The
average electric field and its uncertainty, shown in Fig. 15.2b, is obtained by
projection of the phasor onto the imaginary axis. Note the negative sign in the
amplitude of Eq. (15.32). Thus, hE(t)i is given in Fig. 15.2b. As the phasor rotates,
the uncertainty in electric field is smallest when ωt ¼ 0, π, . . . and greatest when
ωt ¼ π2 , 3π
2 , . . ., consistent with Eqs. (15.36) and (15.37), respectively. This case is
called phase squeezing and corresponds to a reduction in the uncertainty of the
P quadrature below the standard quantum limit (SQL) at certain times.
Conversely, Fig. 15.3 shows the case for R > 0. Here, the uncertainty is smallest
when ωt ¼ π2 , 3π 2 , . . . and greatest when ωt ¼ 0, π, . . ., corresponding to Eqs. (15.38)

(a) (b)
Projecon onto imaginary axis

Fig. 15.2 (a) Phasor representation of squeezed field if R < 0. (b) Resulting average electric
field, hE(t)i, and its uncertainty (dashed lines and red arrows)

(a) (b)
Projecon onto imaginary axis

Fig. 15.3 (a) Phasor representation of squeezed field if R > 0. (b) Resulting average electric
field, hE(t)i, and its uncertainty (dashed lines and red arrows)
15.5 Quadratures 147

and (15.39), respectively. This case is called amplitude squeezing or Q quadrature

squeezing and corresponds to a reduction in the uncertainty of the Q quadrature
below the SQL at certain times.
Thus, we can measure either the phase or the amplitude below the SQL, but only
for a short period of time. As we will see in Chap. 16, we need homodyne detection
(our “strobe light”) to take advantage of the quadrature squeezing.

15.5 Quadratures

The Q quadrature of squeezed light is

b Ri
Q ¼ hQi ¼ hα, RjQjα, ð15:40Þ

where
 
b ¼ p1ffiffiffi b
Q a{
aþb ð15:41Þ
2

Using Eq. (15.18), we get

{
 
1
Q ¼ pffiffiffi hαjb
S b a{ b
aþb Sjαi ð15:42Þ
2

Using Eqs. (15.23) and (15.26), we get

1
Q ¼ pffiffiffi hαj½ðb a{ sinh RÞ þ ðb
a cosh R  b a{ cosh R  b
a sinh RÞjαi ð15:43Þ
2

Using the definition of a coherent state in Eq. (10.1) gives

1
Q ¼ pffiffiffi ½ðα cosh R  α sinh RÞ þ ðα cosh R  α sinh RÞ ð15:44Þ
2

Using Eqs. (15.24) and (15.25), we get

1
Q ¼ pffiffiffi eR ðα þ α Þ ð15:45Þ
2

If α 2 ℝ,
148 15 Squeezed Light

1
Q ¼ pffiffiffi eR ð2αÞ ð15:46Þ
2

It is straightforward to show that (Exercise 15.3)


b 2 jα, Ri ¼ 1 e2R ð2αÞ2 þ 1 e2R
Q2 ¼ hα, Rj Q ð15:47Þ
2 2

which gives the uncertainty

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
ΔQ ¼ Q2  hQi2 ¼ pffiffiffi eR ð15:48Þ
2

Similarly, the uncertainty in the P quadrature is

1
ΔP ¼ pffiffiffi eR ð15:49Þ
2

Recall that the quadratures for the coherent state were ΔQ ¼ ΔP ¼ p1ffiffi2, representing
a minimum uncertainty state or standard quantum limit (SQL). For squeezed light
with R < 0, we have an exponential increase (eR) in Q but an exponential decrease
(eR) in P compared to the SQL, and vice versa for R > 0. However, the product of the
uncertainties is the same as that of the minimum uncertainty state; that is, the
uncertainty relation is still satisfied:

1
ΔQΔP ¼ ð15:50Þ
2

The quadrature representation of the state is shown in Fig. 15.4 for R < 0, and in
Fig. 15.5 for R > 0. In the coherent state, the uncertainties are distributed equally
between the two quadratures. In the squeezed state, the uncertainty in one quadrature
is decreased at the expense of increasing the other. Recalling the number-phase

Fig. 15.4 Quadrature

representation for phase Phase squeezing
squeezing quadrature squeezing
15.6 Squeezed Power 149

Fig. 15.5 Quadrature

representation for amplitude Amplitude squeezing
squeezing quadrature squeezing

uncertainty relation, we see that the noise is redistributed between the amplitude and
phase of the field.

Exercise 15.3 Derive Eq. (15.47).

15.6 Squeezed Power

The uncertainty of certain quadratures for squeezed light is reduced compared to that
of coherent light. However, it is the signal-to-noise ratio (SNR) that really matters.
Let us check the power in the beam for squeezed light versus coherent light, and then
calculate the SNR. For a coherent state of amplitude α0 we have

b 0 i ¼ jα0 j2
hni ¼ hα0 jNjα ð15:51Þ

For a squeezed state, we have

b Ri
hni ¼ hα, RjNjα, ð15:52Þ

a{b
¼ hα, Rjb aj α, Ri ð15:53Þ
{ {
¼ hα j b
Sbab ab
S j αi ð15:54Þ
{
Since b
S is unitary (b
Sb
S ¼ 1), we can write

{ { {
 {  { 
hni ¼ hαjb
SbabSb
Sbab
Sjαi ¼ hαj b
Sb S b
a{ b Sbab
S jαi ð15:55Þ

Using Eqs. (15.23) and (15.26)

150 15 Squeezed Light

hni ¼ hαjðb a{ cosh R  ba sinh RÞðba cosh R  ba{ sinh RÞjαi ð15:56Þ
 
¼ hαj ba{ b
a cosh 2 R  ba{ b
a{ cosh R sinh R  bab
a sinh R cosh R þ b a{ sinh 2 R jαi
ab
ð15:57Þ

Using the commutation relation to produce normal ordering gives

hni¼ hαj½ba{b
a cosh 2 Rba{ b
a{ coshRsinhRb
ab
a sinhRcoshR ð15:58Þ
 
þ b a{ b
a þ1 sinh 2 Rjαi
  h i
¼ jαj2 cosh 2 R þ sinh 2 R  ðα Þ2 þ ðαÞ2 cosh R sinh R þ sinh 2 R ð15:59Þ

Using Eq. (15.31) and assuming α 2 ℝ, we get

hni ¼ ðα0 Þ þ sinh 2 R

2
ð15:60Þ

For typical values of R (~2  3), sinh2R  (α0)2, and we get

hni  ðα0 Þ
2
ð15:61Þ

which is the same as Eq. (15.51). Thus, a coherent state and squeezed state of the
same amplitude α0 have approximately the same power in the beam. However, the
uncertainty of the squeezed state is reduced compared to that of the coherent state.
Thus, the SNR of the squeezed state is improved compared to a coherent state of the
same power.

15.7 Fragility of Squeezing

The reason we do not use squeezing routinely is due to the fragility of the squeezed
state. The quadrature squeezing is easily lost due to absorption, scattering, or other
loss mechanisms. We can represent these losses by a beam splitter, as shown in
Fig. 15.6. A fraction of the squeezed light is lost (represented by the reflection),
while the remainder continues (represented by the transmission). The operator for
the transmitted light is

b
a4 ¼ tb
a1  rb
a2 ð15:62Þ

The Q quadrature at D4 is
15.8 Squeezed Vacuum 151

Fig. 15.6 Beam splitter

model of optical losses

 
1 1
hQ4 i ¼ 2 h0j1 hα, Rj pffiffiffi b a{4 jα, Ri1 j0i2 ¼ pffiffiffi eR ð2αÞt
a4 þ b ð15:63Þ
2 2

which is the same as Eq. (15.46) but multiplied by the transmission coefficient, t. The
uncertainty can be calculated, giving:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2 1  1=2
ΔQ4 ¼ Q4  hQ4 i2 ¼ pffiffiffi t 2 e2R þ r 2 ð15:64Þ
2

Note that we recover Eq. (15.48) when we set t ¼ 1 and r ¼ 0 in Eq. (15.64) for a
lossless beam splitter. We see that the squeezing is reduced due to the r2 term in
Eq. (15.64), representing the optical loss. The r2 term arises due to the Q quadrature
of the vacuum fluctuations in port 2! For this reason, the squeezing is easily
destroyed unless great care is taken to remove optical losses.

Exercise 15.4 Derive Eqs. (15.63) and (15.64).

15.8 Squeezed Vacuum

As mentioned earlier, the optical parametric amplification process also works for the
vacuum state, resulting in a reduction (squeezing) of the uncertainty of the electric
field of the vacuum state. The squeezing operator applied to the vacuum produces the
“squeezed vacuum”, that is, a coherent state with α ¼ 0:

b
SðRÞj0i ¼ j0, Ri ð15:60Þ

Based on the preceding results, the properties of squeezed vacuum include

b ðr, t Þj0, Ri ¼ 0
hEi ¼ h0, RjE ð15:61Þ
b j 0, Ri ¼ 0
hQi ¼ h0, Rj Q ð15:62Þ
b j 0, Ri ¼ 0
hPi ¼ h0, Rj P ð15:63Þ
152 15 Squeezed Light

Fig. 15.7 Quadrature

representation of the squeezed vacuum
vacuum state j0i (dashed
circle) and the squeezed
vacuum state j0, Ri (gray
oval)

1
ΔQ ¼ pffiffiffi eR , R < 0 ð15:64Þ
2
1
ΔP ¼ pffiffiffi eR , R < 0 ð15:65Þ
2
a{b
hni ¼ h0,Rjb aj0,Ri ¼ sinh 2 R ð15:66Þ

Notably, the average photon number of the squeezed vacuum is not zero. The
quadrature representation of the P squeezed vacuum (R < 0) is shown in Fig. 15.7.

15.9 Photon Number Distribution of Squeezed Light

We saw before that the photon number distribution is Poissonian for the coherent
state. Since the uncertainty of the phasor along the radius in Fig. 15.4 is larger than
that for a coherent state, we have a super-Poissonian distribution in the photon
number for the phase squeezed state. Conversely, the uncertainty in the phasor
along the radius in Fig. 15.5 for the amplitude squeezed state is reduced compared
to that for a coherent state, giving a sub-Poissonian photon number distribution.
Let us derive the photon number distribution of the squeezed vacuum state. We
start with the vacuum state, which satisfies

b
aj0i ¼ 0 ð15:67Þ

Applying the squeeze operator to Eq. (15.67), and using the fact that the squeeze
{
Sb
operator is unitary (b S ¼ 1), gives
15.9 Photon Number Distribution of Squeezed Light 153

{
bab
SbSb Sj0i ¼ 0 ð15:68Þ

or, using Eq. (15.60)

{
bab
SbS j0, Ri ¼ 0 ð15:69Þ

Similar to the derivation of Eq. (15.23), using the Baker–Hausdorff formula, we

have

{
bab
SbS ¼b a{ sinh R
a cosh R þ b ð15:70Þ

Thus, the squeezed vacuum satisfies the eigenvalue equation:

 
b a{ sinh R j0, Ri ¼ 0
a cosh R þ b ð15:71Þ

To determine the photon number distribution, we want to express the vacuum state
as a superposition of Fock states:

X
1
j0, Ri ¼ cn jni ð15:72Þ
n¼0

Then, the photon number distribution will be given by the probabilities, jcnj2. If we
substitute Eq. (15.72) into the eigenvalue equation in Eq. (15.71), we get

 X
1
b a{ sinh R
a cosh R þ b cn jni ¼ 0 ð15:73Þ
n¼0

Equation (15.73) gives the recursion relation

rffiffiffiffiffiffiffiffiffiffiffi
n
cnþ1 ¼  tanh R c ð15:74Þ
n þ 1 n1

The probability of the squeezed vacuum state having an odd number of photons is
zero. This is not surprising because photons are created or annihilated in pairs from
the vacuum according to the optical parametric process. We start from the vacuum
and the squeezing operator adds or subtracts two photons at a time. Only the even
numbered coefficients contain the vacuum state. The solution to Eq. (15.74) for even
solutions is
154 15 Squeezed Light

 1=2
ð2n  1Þ‼
c2n ¼ ð1Þn ð tanh RÞn c0 ð15:75Þ
ð2nÞ‼

where n is an integer, and !! denotes the double factorial. The double factorial, for
example n!!, is the product of all the integers from 1 up to n that have the same parity
(odd or even) as n. For example, if n is even, then n‼ ¼ n(n  2)(n  4). . .(4)(2).
c0 can now be determined from the normalization condition

X
1
jc2n j2 ¼ 1 ð15:76Þ
n¼0

resulting in
!
X1
ð tanh RÞ2n ð2n  1Þ‼
jc0 j2 1þ ¼1 ð15:77Þ
n¼1
ð2nÞ‼

Next, we use the following (not obvious!) identities:

X1
ðzÞn ð2n  1Þ‼
1þ ¼ ð1  zÞ1=2 ð15:78Þ
n¼1
ð 2n Þ‼

ð2nÞ‼ ¼ 2n n! ð15:79Þ
n
2 ð2nÞ!
ð2n  1Þ‼ ¼ ð15:80Þ
n!

which gives

1
c0 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð15:81Þ
cosh R

and
pffiffiffiffiffiffiffiffiffiffi
ð2nÞ! ð tanh RÞn
c2n ¼ ð1Þ n
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð15:82Þ
2n n! cosh R

From Eq. (15.72), we get

pffiffiffiffiffiffiffiffiffiffi
X
1
ð2nÞ! ð tanh RÞn
j0, Ri ¼ ð1Þ n
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi j2ni ð15:83Þ
n¼0
2n n! cosh R

Thus, the probability of detecting 2n photons from the vacuum squeezed state is
given by the square of the probability amplitude in Eq. (15.83):
References 155

Fig. 15.8 Photon number 0.4

distribution for squeezed
vacuum state from
Eq. (15.84) with R ¼ 2 0.3

Pn
0.2

0.1

0
0 2 4 6 8 10
n

ð2nÞ! ð tanh RÞ2n

P2n ¼ ð15:84Þ
22n ðn!Þ2 cosh R

Figure 15.8 shows an example of the photon number distribution from Eq. (15.84)
for a squeezed vacuum with R ¼ 2.
The squeezed coherent state can be obtained by applying the squeeze operator to
the vacuum, b SðRÞj0i ¼ j0, Ri, and then applying the displacement operator intro-
b ðαÞb
duced in Sect. 10.9, D SðRÞj0i ¼ jα, Ri. This results in a recursion relation that
can be solved for the photon number distribution, similar to the previous procedure
for the squeezed vacuum. The squeezed coherent state contains a mixture of even
and odd photon numbers with the possibility of sub-Poissonian statistics. A deriva-
tion can be found in Refs. [2–4].

References

1. R.E. Slusher et al., Observation of squeezed states generated by four-wave mixing in an optical
cavity, Phys. Rev. Lett. 55 (1985) 2409.
2. C.C. Gerry and P.L. Knight, Introductory quantum optics (2005, Cambridge University Press),
pp. 160-165.
3. R.W. Henry and S.C. Glotzer, A squeezed state primer, Amer. J. Phys. 56 (1988) 318.
4. J.J. Gong and P.K. Aravind, Expansion coefficients of a squeezed coherent state in the number
state basis, Amer. J. Phys. 58 (1990) 1003.
Chapter 16
Squeezed Light in an Interferometer

The uncertainty and signal-to-noise ratio (SNR) of a homodyne signal using

squeezed light in an interferometer is derived, showing SNR below the shot noise
limit. The use of squeezed light for improved detection of gravitational waves is
presented as a key application.

16.1 Homodyne Signal Using Squeezed Light

In Chap. 14, we learned that the SNR for the coherent state arises from the
uncertainty in the field quadrature of the vacuum input to the interferometer. To
improve the SNR, we can replace the vacuum, j0i, in the interferometer with
squeezed vacuum, j0, Ri! Using squeezed light, we will see that we can beat the
shot noise limit in interferometry. This approach was proposed by Carlton
M. Caves [1].
The input state to the interferometer in Fig. 16.1 is jψini ¼ j0, Ri1jαi2. The
local oscillator is the coherent state jαi2 on input port 2, while j0, Ri1 is the
vacuum squeezed state on input port 1. The operator for the homodyne signal
from Chap. 14 is
   
n4  b
b a{1b
n3 ¼  sin ε b a{2b
a1  b a2  i cos ε ba{1b a{2b
a2  b a1 ð16:1Þ

Retaining the non-zero terms gives

a{1b
hn4 i  hn3 i ¼ sin ε 2 hαj1 h0, Rjðjαj2  b a1 Þj0, Ri1 jαi2 ð16:2Þ

From Eq. (15.66), we get

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 157
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_16
158 16 Squeezed Light in an Interferometer

Fig. 16.1 Interferometer

with squeezed vacuum on
port 1 and coherent state on
port 2

 
hn4 i  hn3 i ¼ sin ε jαj2  sinh 2 R ð16:3Þ

For a strong coherent state, we have jαj2  sinh2R, which gives

hn4 i  hn3 i  jαj2 sin ε ð16:4Þ

¼ hn2 i sin ε ð16:5Þ

For a small phase change (small ε), we get

h n 4 i  h n3 i  h n2 i ε ð16:6Þ

which is the same result as that obtained in Chap. 14 for the coherent state.
Next, we evaluate h(n4  n3)2i. Assuming small displacements in the interfer-
ometer (ε ! 0) and α real, we get

2
a{1  b
hðn4  n3 Þ2 i ¼ α2 1 h0, Rjðb a1 Þ j0, Ri1 ð16:7Þ
¼ α2 e2R ð16:8Þ
¼ e h n2 i
2R
ð16:9Þ

Exercise 16.1 Derive Eqs. (16.7) and (16.8).

From Eqs. (16.6) and (16.9), the uncertainty is

16.2 Laser Interferometer Gravitational-Wave Observatory (LIGO) 159

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
D E pffiffiffiffiffiffiffiffiffi
Δ ð n4  n 3 Þ ¼ ðn4  n3 Þ2  hn4  n3 i  eR hn2 i ð16:10Þ

The shot noise is reduced by a factor eR for R < 0. The SNR is

ε hn2 i
SNR ¼ pffiffiffiffiffiffiffiffiffi
eR h n2 i
pffiffiffiffiffiffiffiffiffi R
¼ ε h n2 i e ð16:11Þ

Compared to Eq. (14.48), the SNR is improved by a factor eR for R < 0 due to the
P quadrature squeezing (phase squeezing) of the light on port 1.

Exercise 16.2 Derive Eq. (16.10).

Today, noise suppression in excess of 10 dB has been achieved, corresponding to

a factor of 10 reduction in noise, or R~2.3. Most squeezing experiments are based on
optical parametric amplification (OPA). Figure 16.2 illustrates the homodyne signal
measured for various states of light in one of the earlier squeezing experiments [2].

16.2 Laser Interferometer Gravitational-Wave

Observatory (LIGO)

The most spectacular application of squeezed light is the Laser Interferometer

Gravitational-wave Observatory (LIGO), located in Hanford, Washington,
(Fig. 16.3) and Livingston, Louisiana, in the USA [3]. These two facilities each
use a Michelson interferometer to measure displacements of space itself associated
with passing gravitational waves. On Sept 14, 2015, the LIGO detectors saw space
“vibrate” due to gravitational waves generated by the merger of two black holes as
predicted by Einstein’s general theory of relativity [4]. The 2017 Nobel Prize in
Physics was awarded to Rainer Weiss, Barry Barish, and Kip Thorne for their work
on gravitational wave detection.
LIGO is a Michelson interferometer with 4 km long arms. Gravitational waves
cause a displacement of the interferometer mirrors, resulting in a change in phase ε:

ε ¼ kΔx ð16:12Þ

or, after rearranging,

λε
Δx ¼ ð16:13Þ
2π
160 16 Squeezed Light in an Interferometer

Fig. 16.2 Experimentally

measured electric field
versus time (homodyne
signal) for (from the top) the
coherent state, phase-
squeezed state, squeezing at
a phase of φ ¼ 48 ,
amplitude-squeezed state
and squeezed vacuum state.
(Reprinted by permission
from Springer Nature,
G. Breitenbach, S. Schiller
and J. Mlynek, Nature 387
(1997) 471 [2])

Fig. 16.3 Aerial photograph of the LIGO Hanford facility, showing the 4 km long arms of the
interferometer. (Credit: Wikimedia Commons [3])
16.2 Laser Interferometer Gravitational-Wave Observatory (LIGO) 161

Thus, the minimum displacement Δx that can be measured is determined by the

minimum uncertainty in ε, assuming the wavelength is well known.
We can determine a phase change, ε, from a change in some measurement
parameter, M, and the sensitivity of the measurement parameter to the phase, dM
dε .
Thus,

M
ε ¼ dM ð16:14Þ
dε

The minimum phase we can measure corresponds to the minimum possible value of
M that can be measured:

M min
ε min ¼ dM
ð16:15Þ
dε

The minimum value of M that can be measured corresponds to its uncertainty, ΔM:

ΔM
ε min ¼ dM
ð16:16Þ
dε

Let us suppose that M corresponds to a homodyne measurement. For a coherent state

in an interferometer, we found the homodyne signal was given by Eq. (14.32):

M ¼ hn4 i  hn3 i ¼ hn2 iε ð16:17Þ

The sensitivity is therefore

dM
¼ h n2 i ð16:18Þ
dε

The uncertainty was given by the shot noise limit (Eq. (14.47)):
pffiffiffiffiffiffiffiffiffi
ΔM ¼ Δðn4  n3 Þ  hn2 i ð16:19Þ

Finally, according to Eq. (16.13):

pffiffiffiffiffiffiffiffiffi
ΔM h n2 i pffiffiffiffiffiffiffiffiffi
ε min ¼ ¼ ¼ 1= hn2 i ð16:20Þ
dM
dε
h n2 i

Note that Eq. (16.20) can also be derived from the number-phase uncertainty
relation, ΔnΔε ¼ 1, which gives Δε ¼ Δn
1
¼ pffiffiffiffiffiffi
1
for the shot noise limit.
hn2 i
LIGO uses a wavelength of λ ¼1 μm and optical power of 100 kW in the
interferometer arms, corresponding to a photon flux of ~1024 photons/s. Assuming
162 16 Squeezed Light in an Interferometer

a measurement time of 1 s, this corresponds to hn2i~1024. Hence, using Eq. (16.13),

the minimum displacement that can be measured is
 6 
λ 10 m
Δx min ¼ pffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffi  1019 m ð16:21Þ
2π hn2 i 2π 1024

or about 104 times smaller than the diameter of a proton! For the 4 km path length in
the LIGO interferometer, this corresponds to a relative displacement (strain) of

Δx 1019 m
  1022 ð16:22Þ
x 4000 m

Implementing squeezed light on the empty port of the interferometer can reduce the
uncertainty by a factor of eR (R < 0) according to Eq. (16.10) and improve the strain
sensitivity in Eq. (16.22) by a factor eR. The resulting improvement in strain
sensitivity is shown in Fig. 16.4 [5, 6]. The use of squeezed light in LIGO has
significantly increased the rate of observed gravitational-wave events.

Fig. 16.4 Improvements in strain sensitivity in LIGO achieved by using squeezed light. The sharp
lines are due to parasitic effects such as mechanical resonances. (Reprinted by permission from
Springer Nature, J. Aasi et al., Nature Photonics 7 (2013) 613 [6])
References 163

Exercise 16.3 Investigate and discuss other potential applications of

squeezed light.

References

1. C.M. Caves, Quantum-mechanical noise in an interferometer, Phys. Rev. D 23 (1981) 1693.

2. G. Breitenbach, S. Schiller and J. Mlynek, Measurement of the quantum states of squeezed light,
Nature 387 (1997) 471.
3. https://web.archive.org/web/20170614214509/http://www.ligo.org/multimedia/gallery/lho-
images/Aerial5.jpg
4. B.P. Abbott et al., Observation of gravitational waves from a binary black hole merger, Phys.
Rev. Lett. 116 (2016) 061102.
5. J. Gea-Banacloche and G. Leuchs, Squeezed states for interferometric gravitational-wave
detectors, J. Mod. Opt. 34 (1987) 793.
6. J. Aasi et al., Enhanced sensitivity of the LIGO gravitational wave detector by using squeezed
states of light, Nature Photonics 7 (2013) 613.
Chapter 17
Heisenberg Limit

The uncertainty in photon number of a coherent source leads to the shot noise limit.
We could eliminate this uncertainty by using a Fock state, which has a definite
photon number. This approach produces the “Heisenberg limit”, which is the
ultimate limit on measurement uncertainty. Multiphoton entangled states (called
NOON states) in an interferometer are introduced as a means of achieving the
Heisenberg limit. Super-sensitivity and super-resolution are introduced as applica-
tions. Methods of producing NOON states are described, and we introduce the
Hong–Ou–Mandel (HOM) effect and linear optical quantum state engineering.

17.1 Heisenberg Limit

As discussed in Chap. 16, we found the minimum phase that can be measured is
given by

ΔM
ε min ¼ dM
ð17:1Þ
dε

where ΔM is the uncertainty in a measurement parameter, M, and dMdε is the sensitivity

of ε with respect to M. For homodyne detection of a coherent
pffiffiffiffiffiffiffi state in an interfer-
ometer, we found in the previous chapter that ΔM ¼ hni and dM dε ¼ hni, giving
ε min ¼ p1ffiffiffiffiffi , which is the shot noise limit. This can be improved to ε min ¼
pffiffiffiffiffiffiffi hni
eR = hni by using squeezed light (with R < 0). Can we do any better than this?
Let us suppose that the smallest measurement we can make is ΔM ¼ 1,
corresponding to the measurement of a single photon. The minimum possible
phase measurement is then

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 165
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_17
166 17 Heisenberg Limit

ΔM 1
ε min ¼ ¼ ð17:4Þ
dM
dε
h ni
pffiffiffiffiffiffiffi
Equation (17.4) is called the Heisenberg limit, which is much better than the 1= hni
value of the shot noise limit! One can also obtain Eq. (17.4) from the number-phase
uncertainty relation, ΔεΔn ¼ 1. If we have hni photons, the maximum uncertainty is
Δn ¼ hni. Then, from the number-phase uncertainty relation, we get ε min ¼ Δε ¼
Δn ¼ hni, which is the same as Eq. (17.4).
1 1

For example, as discussed in the previous chapter, the LIGO interferometer with
hni ¼ 1024 photons and shot-noise limited phase uncertainty gave ε min ¼ p1ffiffiffiffiffi ¼
hni
1012 , corresponding to a minimum displacement measurement of Δx~1019 m.
In the Heisenberg limit, the phase uncertainty would be ε min ¼ h1ni ¼ 1024 ,
corresponding to a minimum displacement measurement of Δx~1031 m, a
12 order of magnitude improvement! This scale is approaching the Planck scale of
1035 m where quantum gravity and the graininess of space-time become relevant!
In the following sections, we examine approaches to achieving the Heisenberg limit.

17.2 Phase Shifter

A phase shifter is an optical device that imparts a phase shift, eiφ, to a classical field.
A phase shifter could simply be a piece of glass of certain thickness and refractive
index, or it could correspond to an optical path length difference (e.g., in an
interferometer). As shown in Fig. 17.1a, a phase shifter will impart a phase shift,
eiφ, to a coherent state, jαi, which resembles a classical field. However, something
very different happens for the number state. As shown in Fig. 17.1b, a number state
undergoes a phase shift of einφ in passing through a phase shifter. Let us explore how
we might use this effect.

Exercise 17.1 Investigate and explain why a number state undergoes a phase
shift of einφ in passing through a phase shifter, as compared to eiφ for a
coherent state.

Fig. 17.1 The effect of a

phase shifter on (a) a (a) Coherent state:
coherent state and (b) a
number state
(b) Number state:
17.4 Super-sensitivity 167

17.3 NOON States

Consider an interferometer where one path of the interferometer contains a phase

shifter. For this analysis, we adopt a simplified view of the interferometer, as shown
in Fig. 17.2, where the input and output optics (mirrors and beam splitters) are
consolidated into boxes on the left and right. We suppose we can create a state in the
two paths of the interferometer where we have n photons that are either all in the
upper mode (input path 1 before the phase shifter) and none in the lower mode 2, or
vice versa (i.e., we have a superposition of the two possibilities):

jn, 0i þ j0, ni
jψi ¼ pffiffiffi ð17:5Þ
2

The first number in each of the Dirac brackets in Eq. (17.5) denotes the number of
photons in the upper path, and the second number denotes the number of photons in
the lower path. This state is a superposition state of jn, 0i and j0, ni. This is called a
“NOON” state when n  2. Note that the NOON state is an entangled state; that is, it
is a nonseparable state. As n becomes large, the NOON state is called a “Schrodinger
cat state”. The NOON state was first discussed in 1989 by Barry Sanders in the
context of decoherence of Schrodinger cat states [1].
In the upper path of the interferometer, we insert a phase shifter. According to
Fig. 17.1b, the resulting state after the phase shifter is

einφ jn, 0i þ j0, ni

j ψi ¼ pffiffiffi ð17:6Þ
2

where φ ¼ kΔx represents an optical path length difference between the two modes.

17.4 Super-sensitivity

Suppose we measure a homodyne signal, hn4  n3i, at the output of the interferom-
eter in Fig. 17.2. For a coherent state in an interferometer, we previously found
(Eq. (14.26))

Fig. 17.2 Effect of phase

shifter on a NOON state in
an interferometer
168 17 Heisenberg Limit

Fig. 17.3 Homodyne signal

for a coherent state (red)
versus a NOON state with
n ¼ 3 (green) in an
interferometer

hn4 i  hn3 i ¼ hni cos ðφÞ ð17:7Þ

where hni is the average number of photons in the coherent state. Equation (17.7) is
represented by the red curve in Fig. 17.3. If we operate near the quadrature with
φ ¼ π2 þ ε, we previously obtained, for small ε,

hn4 i  hn3 i ¼ hni sin ðεÞ  hni ε ð17:8Þ

For a NOON state, Eq. (17.7) changes to

hn4 i  hn3 i ¼ n cos ðnφÞ ð17:9Þ

where n is the number of photons in the NOON state. Figure 17.3 illustrates
Eq. (17.9) for n ¼ 3. If we operate near φ ¼ π2 þ ε, we obtain for small ε:

hn4 i  hn3 i ¼ n sin ðnεÞ  n ðnεÞ ð17:10Þ

We see that the slope at the horizontal crossings (e.g., π2 ) for the number state
increases by a factor of n compared to the coherent state. Thus, comparing
Eq. (17.8) and (17.10), we have an improvement in phase sensitivity by a factor n.
This improvement is called “super-sensitivity”.
With the NOON state, we need to detect hn4i  hn3i ¼ n photons at a time:

h n 4 i  h n3 i ¼ n ð17:11Þ

or, from Eq. (17.10):

1
ε¼ ð17:12Þ
n

which is the Heisenberg limit, identical to Eq. (17.4).

Due to the number-phase uncertainty relation, we expect the minimum uncer-
tainty in phase when the uncertainty in photon number is greatest. In NOON states,
we are completely uncertain if all n photons are in mode 3 (and none in 4) or if all
17.5 Super-resolution and Quantum Lithography 169

n photons are in mode 4 (and none in 3). This gives the maximum uncertainty in the
number of photons of Δn ¼ n. Thus, from the number-phase uncertainty relation,

1 1
ΔnΔε ¼ 1 ! Δε ¼ ¼ ð17:13Þ
Δn n

which is again the Heisenberg limit.

17.5 Super-resolution and Quantum Lithography

Figure 17.3 shows a reduction in wavelength from λ for a coherent state to λ/n for a
NOON state in an interferometer. λ/n is referred to as the de Broglie wavelength of
the photon. The de Broglie wavelength of a single photon is

h hc
λ¼ ¼ ð17:14Þ
p E

The de Broglie wavelength of a biphoton is

h hc
λ¼ ¼ ð17:15Þ
p 2E

Continuing in this manner, the de Broglie wavelength of the NOON state is

h hc
λ¼ ¼ ð17:16Þ
p nE

which is called a “superphoton” [2].

A superphoton with reduced de Broglie wavelength would allow improvements
in the spatial resolution of imaging systems. In a conventional imaging system, the
Rayleigh criterion states that diffraction limits the smallest separation between two
object points that can be resolved in an optical image. Classical physics tells us that
the minimum angular separation of two object points is 1.22λ/D, where D is the
diameter of the collecting aperture. “Super-resolution” achieves a spatial resolution
in an optical imaging system using superphotons with reduced de Broglie wave-
length that can beat the Rayleigh criterion. The main idea of sub-Rayleigh imaging is
to replace intensity measurements with spatially resolved n photon detection using
photon-number-resolving detectors.
However, NOON states suffer a significant weakness. The probability of
detecting n photons arriving at the same place decreases exponentially with n due
to scattering; that is, NOON states are exponentially sensitive to absorption. Beer’s
law states that classical or coherent light will decay exponentially (eγ ) according to
170 17 Heisenberg Limit

an absorption coefficient, γ. However, the NOON state absorption becomes enγ .

This reduces the slope at the horizontal crossings, killing the super-sensitivity
advantage. Thus, super-resolution has only been demonstrated with a few
photons [3].
One of the proposed applications of super-resolution is quantum lithography,
where the superphoton with reduced de Broglie wavelength is used to expose
photoresists for improved spatial resolution in lithography [4]. Besides the difficulty
in creating NOON states, the problem with quantum lithography is that it has proven
very difficult to make efficient n-photon-sensitive absorbing resists.
Other applications of quantum metrology include the entanglement of ion states,
which could improve the accuracy of atomic clocks. We could measure a frequency
to an accuracy of Δω ¼ Δφ 1 ffiffi
t ¼ tn, instead of Δω ¼ t n. We could also improve transit
1 p

time measurements, distance measurements, and clock synchronization (used in

GPS) [5].

17.6 Producing a NOON State

A superposition of one photon in mode 3 and none in mode 4 (j1i3j0i4), and one
photon in mode 4 and none in mode 3 (j0i3j1i4), can be achieved using a beam
splitter (Fig. 17.4). As we saw previously in Chap. 6, for a 50:50 beam splitter and
single photon input on one of the ports, we get

1  
jψout i ¼ pffiffiffi j1i3 j0i4 þ j0i3 j1i4 ð17:17Þ
2

This is an entangled state of a photon in mode 3 and mode 4. Technically, this is not a
NOON state, since we need n  2. The following sections will describe how to
create a NOON state with n  2.

Fig. 17.4 Producing a

NOON state with n ¼ 1
17.7 Hong–Ou–Mandel Effect 171

17.7 Hong–Ou–Mandel Effect

A NOON state with n ¼ 2 can be produced by a single photon input on both port
1 and 2 of the beam splitter in Fig. 17.4:

jψin i ¼ j1i1 j1i2 ð17:18Þ

a{1b
¼b a{2 j0i1 j0i2 ð17:19Þ

The output state is

  
jψout i ¼ rb a{3 þ tba{4 tb a{3  rb
a{4 j0i3 j0i4 ð17:20Þ
 
a{3b
¼ rtb a{3  r 2b
a{3b
a{4 þ t 2b
a{4b
a{3  trb
a{4b
a{4 j0i3 j0i4 ð17:21Þ

Recalling Eq. (2.127), we get

pffiffiffi pffiffiffi
jψout i ¼ 2rt j2i3 j0i4  r 2 j1i3 j1i4 þ t 2 j1i3 j1i4  2tr j0i3 j2i4 ð17:22Þ

For a 50:50 beam splitter (r ¼ t ¼ p1ffiffi2), Eq. (17.22) becomes

1  
jψout i ¼ pffiffiffi j2i3 j0i4  j0i3 j2i4 ð17:23Þ
2

Equation (17.23) represents a NOON state with n ¼ 2. This arises due to

destructive interference of the probability amplitudes for a single photon in both
modes 3 and 4, destroying the j1i3j1i4 state. This interference is a quantum mechan-
ical effect called the Hong–Ou–Mandel (HOM) effect, named after C.K. Hong,
Z.Y. Ou, and L. Mandel [6]. The HOM effect results in only two-photon detection
occurring randomly at either D3 or D4 as depicted in Fig. 17.5.
Perfect destructive interference of the j1i3j1i4 mode occurs only when the two
photon inputs are indistinguishable in every respect, that is, when they have the same

(a) (b) (c) (d)

Fig. 17.5 The four possible photon paths leading to the Hong–Ou–Mandel (HOM) effect. The
negative signs arise due to one of the reflections, leading to destructive interference of the
amplitudes in (b) and (c). Only the outputs in (a) and (d) remain. The photon paths are labelled
with different colors, although the photons are indistinguishable
172 17 Heisenberg Limit

Table 17.1 Comparison of quantum versus classical expectation for HOM experiment with 50:50
beam splitter
Mode 3 Mode 4 Quantum probability Classical probability
j2i j0i 2jrj2jtj2 ¼ 1/2 jrj2jtj2 ¼ 1/4
j0i j2i 2jrj2jtj2 ¼ 1/2 jrj2jtj2 ¼ 1/4
j1i j1i (jtj2  jrj2)2 ¼ 0 jrj4 + jtj4 ¼ 1/2

frequency, same polarization, same coefficients (probability amplitudes) of the

multimode state, and when they arrive at the same time on the beam splitter.
Hence, the HOM effect can be used to measure the indistinguishability of two
photons.
The classical expectation versus the quantum results for the HOM experiment are
summarized in Table 17.1. Classically, we expect each of the fields to reflect from
the beam splitter with probability jrj2 and transmit with probability jtj2. Conversely,
the quantum probabilities are obtained from the square of the probability amplitudes
in Eq. (17.21). The quantum result is very different than the classical one. Note that
the probabilities sum to unity in either case. Classically, we expect coincidence
counts with 50% probability. Therefore, the HOM effect should give coincidence
counts with probability less than 50% to demonstrate quantum interference.
The experimental setup for the original HOM measurement is shown in
Fig. 17.6a. The two photons were generated by parametric down-conversion in a
KDP crystal. The two photons were redirected by mirrors (M1, M2) onto a beam
splitter (BS) and two detectors (D1, D2) for coincidence counting. A time delay
between the measurement at D1 and D2 is implemented by translation of the beam
splitter. The results shown in Fig. 17.6b reveal nearly zero coincidence counts (the
“HOM dip”) for simultaneous measurement (corresponding to BS position near
300 μm) where the two photon modes overlap temporally. The temporal overlap
of the two photons decreases by displacing the beam splitter. The coincidence counts
increase as the wavepacket of the two photons no longer overlaps perfectly. A
review of the HOM effect is available in Ref. [7].

17.8 High NOON State

A NOON state with large n is called a “high NOON state”. Figure 17.7 illustrates a
possible method for producing a high NOON state. The method, proposed in 2001
[8], uses two interferometers coupled by a Kerr material. The Kerr material uses the
nonlinear Kerr effect to produce a π phase shift when one photon is incident from the
upper interferometer but zero phase shift otherwise. The lower interferometer is set
up so that all n photons input on mode 1 will go to the output of mode 3 and none to
mode 4 when there is no phase shift from the Kerr material. However, when the Kerr
material gives a π phase shift, all n photons in the lower interferometer will go in
mode 4 and none in mode 3. The upper interferometer produces an entangled state
with a superposition of 0 and 1 photon incident on the Kerr phase shifter, producing
17.8 High NOON State 173

Fig. 17.6 (a) Experimental setup and (b) results for the original Hong–Ou–Mandel (HOM)
measurement [6]. (Reprinted with permission from C.K. Hong, Z.Y. Ou and L. Mandel, Phys.
Rev. Lett. 59 (1987) 2044, https://doi.org/10.1103/PhysRevLett.59.2044 [6]. Copyright 1987 by
the American Physical Society)

Fig. 17.7 Producing a high

NOON state using nonlinear
optics (Kerr effect)

Kerr
174 17 Heisenberg Limit

Fig. 17.8 Producing a high

NOON state from linear
optics

a superposition of 0 and π phase shift. The lower interferometer then produces a

superposition of n photons in mode 3 (jni3j0i4) for 0 phase shift and n photons in
mode 4 (j0i3jni4) for π phase shift, resulting in the NOON state. Unfortunately, the
Kerr effect is extremely weak. Attempts to boost the Kerr effect involve placing an
optical cavity around a Kerr material [9], or using an approach known as electro-
magnetically induced transparency (EIT) [10].

Exercise 17.2 Investigate and explain electromagnetically induced transpar-

ency (EIT) and describe its applications in quantum optics.

Another approach to producing high NOON states is to use entirely linear optics,
as shown in Fig. 17.8. Due to a generalization of the HOM effect, an input state of
j3ij3i results in an output from the first beam splitter with a superposition of j6ij0i,
j2ij4i, j4ij2i and j0ij6i. If we perform a coincidence measurement on this state and
obtain j1ij1i, then the measurement removes the j6ij0i and j0ij6i states from the
superposition. The state collapses to a superposition of j1ij3i and j3ij1i, since the
measurement removed one photon from each mode. Passage through the final beam
splitter produces the NOON state with n ¼ 4, again due to the HOM effect (Exercise
17.3). This is a heralded process. If we obtain the j1ij1i state upon measurement,
then we proceed; otherwise, we start over. Thus, this approach will succeed with
some probability.

Exercise 17.3 Show that the last beam splitter in Fig. 17.6 produces a NOON
state with n ¼ 4 from j1ij3i + j3ij1i.
References 175

17.9 Quantum State Engineering

The previous sections highlight the use of mirrors, phase shifters, and beam splitters
to produce a desired quantum state. These devices are the resources available to
realize a desired unitary transformation, together with photon counting measure-
ments. Putting these elements together efficiently to realize a desired quantum state,
which we call “quantum state engineering”, is an active research area for all-optical
quantum information processing. Knill et al. [11] have shown how linear optics
together with photon counting measurements can enable quantum information
processing without the need for the nonlinear Kerr effect. This approach is called
linear optical quantum computing (LOQC). Although this approach does away with
the inefficiency of nonlinear processes, it requires photon counting measurements
(called “post-selection”) and additional photons (called “ancillary photons”) to yield
the desired state with some probability.

References

1. B.C. Sanders, Quantum dynamics of the nonlinear rotator and the effects of continual spin
measurement, Phys. Rev. A 40 (1989) 2417.
2. J. Jacobson et al., Photonic de Broglie waves, Phys. Rev. Lett. 74 (1995) 4835.
3. M. D’Angelo, Two-Photon Diffraction and Quantum Lithography, Phys. Rev. Lett. 87 (2001)
013602.
4. R.W. Boyd and J.P. Dowling, Quantum lithography, status of the field, Quantum Information
Processing 11 (2012) 891.
5. V. Giovannetti et al., Quantum-enhanced measurements: Beating the standard quantum limit,
Science 306 (2004) 1330.
6. C.K. Hong, Z.Y. Ou and L. Mandel, Measurement of subpicosecond time intervals between two
photons by interference, Phys. Rev. Lett. 59 (1987) 2044.
7. Frédéric Bouchard et al., Two-photon interference: the Hong–Ou–Mandel effect, Rep. Prog.
Phys. 84 (2021) 012402.
8. C.C. Gerry and R.A. Campos, Generation of maximally entangled photonic states with a
quantum-optical Fredkin gate, Phys. Rev. A 64 (2001) 063814.
9. Q.A. Turchette et al., Measurement of conditional phase-shifts for quantum logic, Phys. Rev.
Lett. 75 (1995) 4710.
10. M.D. Lukin, Trapping and manipulating photon states in atomic ensembles, Rev. Mod. Phys.
75 (2003) 457.
11. E. Knill, R. Laflamme and G.J. Milburn, A scheme for efficient quantum computation with
linear optics, Nature 409 (2001) 46.
Chapter 18
Quantum Imaging

Quantum imaging, involving entangled light and coincidence measurements, is

providing a revolution in metrology. Nonlocal interference, ghost imaging, quantum
illumination, absolute detector calibration, and interaction-free measurement are
presented as leading applications. These applications demonstrate the spooky behav-
ior of quantum mechanics. For example, we will show you how to observe an object
without any photon ever interacting with it!

18.1 Nonlocal Interference

Figure 18.1 demonstrates the nonlocality of quantum mechanics in a double-slit

interference experiment. A barium borate crystal (Ba(BO2)2, called BBO) is used to
create entangled photon pairs by parametric down-conversion of a laser source. The
two photons pass through a slit (A1 and A2) and a light pattern is measured by
vertically scanning the detectors D1 and D2 (the scan direction is out of the page in
Fig. 18.1, which is perpendicular to the slits). Coincidence counts (double photon
detection at D1 and D2) are also measured for different detector positions. Single
photon detection at either D1 or D2 results in the corresponding slit pattern, A1 or A2,
respectively, in Fig. 18.2b. However, when coincidence counts are measured at both
D1 and D2, then an interference pattern (Fig. 18.2c, d) results from the combined slit
pattern, A1A2, corresponding to a double-slit. The interference pattern cannot be
measured by photon counting at the individual detectors but is contained nonlocally
in the entangled photon pair. The double-slit interference pattern can only be
observed in the coincidence counting, even though the individual detector measure-
ments have nothing to do with double-slit patterns. Neither beam alone contains the
information required to construct the double-slit image. This experiment demon-
strates “nonlocal interference”.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 177
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_18
178 18 Quantum Imaging

Fig. 18.1 (a) Experiment

demonstrating nonlocal
interference. (b) The slit
patterns, A1 and A2, and the
combined slit pattern, A1A2.
(Reprinted with permission
from Fonseca et al. [1].
Copyright 1992 by the
American Physical Society)

Fig. 18.2 (a) Slit patterns. (b) Single detector counts corresponding to A1 or A2. (c, d) D1 and D2
coincidence counts showing interference from the combined slit pattern, A1A2. (Reprinted with
permission from Fonseca et al. [1]. Copyright 1992 by the American Physical Society)

18.2 Ghost Imaging

Ghost imaging (GI) (also called coincidence imaging, two-photon imaging, or

correlated-photon imaging) is a technique that uses two correlated optical fields to
form an image of an object. A typical GI setup (Fig. 18.3) comprises an entangled
photon pair, in which one photon interacts with an object and is subsequently
detected by a nonspatially resolving (“bucket”) detector. The other photon of the
entangled photon pair is detected by a spatially resolving detector (e.g., a 2D
camera). Coincidence detection of photons at the bucket detector and photons at
the imaging detector is then performed to build an image of the object placed along
the bucket detector path. Neither beam alone contains the information required to
18.3 Quantum Illumination 179

Object
Bucket
detector
Laser PDC Coincidences
2D
detector

Fig. 18.3 Experiment for ghost imaging. A laser illuminates a parametric down-converter (PDC) to
create entangled photon pairs. One photon is directed along an object path with a bucket detector,
while the other photon is directed to a 2D detector. An image is formed from coincidence counting

reconstruct an image of the object. The bucket detector is spatially unresolved,

simply detecting all photons passed by the object. Similarly, the spatially resolving
detector measures the position of all photons incident on it without any information
about the object. However, the correlation between these two beams can extract an
image. It should be noted that GI can also be conducted using classical light (thermal
source) but with lower contrast or visibility [2–4]. Entanglement provides a conve-
nient and efficient source of correlated photons that produce greater visibility of the
correlated image.
One of the advantages of GI is that it allows imaging at wavelengths for which
efficient bucket detectors exist, but cameras do not. For example, one could generate
pairs of entangled photons with different wavelengths (355 nm converted to
1550 nm and 460 nm). The object is illuminated at 1550 nm, but the image is
formed on the 2D detector from 460 nm (in coincidence). This allows imaging in the
infrared using a visible CCD camera. This method is called two-color GI [5]. GI is
also of interest when imaging through diffusive media, such as free-space measure-
ments in the presence of fog or the diffusive medium of biological samples. GI is
further described in Ref. [6, 7].

18.3 Quantum Illumination

Quantum illumination achieves subshot noise imaging by coincidence counting, as

shown in Fig. 18.4 [8, 9]. An object of interest (bird) is obscured by noise (cage)
from thermal illumination, forming a “classical image”. To discern the object from
the noise, the object and a reference are illuminated with a quantum source (quantum
illumination, QI). One photon of an entangled pair is measured by a spatially
resolving detector forming the quantum reference, while the other photon illumi-
nates the object, which is added to the classical image. Coincidence counting of the
quantum reference and the classical image will remove the noise (cage), since the
noise is not contained in the coincidence counts. Using this technique in the radio or
microwave region is known as “quantum radar”.
180 18 Quantum Imaging

Fig. 18.4 (a) Experiment (a)

and (b) results
demonstrating quantum Quantum
reference
illumination (QI). (Credit:
Reprinted from Ref. [8] Laser PDC Coincidences
Object Classical
under the terms of the
(bird) image
Creative Commons
Thermal Noise
Attribution license)
source (cage) Classical
illuminaon

(b)

Fig. 18.5 Quantum method DUT

of absolute photodetector
calibration

Laser PDC Coincidences

Trigger

18.4 Absolute Photodetector Calibration

Quantum methods allow us to perform the calibration of photodetector efficiency

without the need for a calibrated light source [10, 11]. If we can produce a known
steady rate of single photons with known optical power on a photodetector, then we
have a means of calibrating the response of a photodetector. In Fig. 18.5, entangled
photon pairs are produced by parametric down-conversion (PDC). The total number
of trigger counts in the trigger arm, ntrig ¼ ηtrign, is measured where ηtrig is the
efficiency of photodetection for the trigger detector and n is the number of incident
photons. The total number of coincidence events, nC ¼ ηDUTηtrign, is measured
where ηDUT is the detection efficiency of the device-under-test (DUT) in the DUT
channel. Thus, the ratio of nC and ntrig gives us the unknown efficiency of the DUT:
18.5 Interaction-Free Measurement (Elitzur–Vaidman Bomb Problem) 181

ηDUT ¼ nC =ntrig ð18:1Þ

The difficulty of this approach is the slow rate of single photon generation from
sources available today, corresponding to only nW or less of optical power. Faster
rates of single photon generation would improve absolute methods of detector
calibration of use in radiometry.

18.5 Interaction-Free Measurement (Elitzur–Vaidman

Bomb Problem)

One of the weirdest consequences of quantum mechanics is “interaction-free mea-

surement” (IFM), also called the Elitzur–Vaidman (E–V) bomb problem, which was
conceived in 1993 by Avshalom Elitzur and Lev Vaidman [12]. IFM is a method of
detecting an object without any photon ever interacting with it [13–17] (e.g.,
detecting a photon-sensitive bomb without exploding it, as originally conceived by
Elitzur and Vaidman).
Suppose single photons enter an interferometer, one at a time. As shown in
Fig. 18.6a, the interferometer is initially adjusted so that the photon always triggers
detector D1 (with destructive interference at D2). Now if an object is placed in the
upper arm of the interferometer, as shown in Fig. 18.6b, then photons are absorbed
by the object 50% of the time (50% probability), and D1 or D2 is triggered each with
25% probability for each photon entering the interferometer. If D2 is triggered, then
we know an object must be present. In this case, the object is detected without a
photon ever having interacted with it! In fact, we can improve the detection prob-
ability to near unity by using a cavity [18]. The E–V experiment can ascertain the
existence of an object in a given region of space, although no light ever “touched”
the object. IFM allows imaging in the dark!

(a) (b)

Fig. 18.6 Demonstration of interaction-free measurement. (a) The interferometer is set up such
that, with the object absent, the probability of detection at D1 and D2 is 100% and 0%, respectively.
(b) With the object present, the probability of detection at D1 and D2 is both 25%. In particular,
detection at D2, which was previously absent in (a), now occurs due to the presence of the object,
although no photon interacted with the object
182 18 Quantum Imaging

Exercise 18.1 Describe some applications of interaction-free measurement.

References

1. E.J.S. Fonseca et al., Quantum interference by a nonlocal double slit, Phys. Rev. A 60 (1992)
1530. https://doi.org/10.1103/PhysRevA.60.1530.
2. Gatti et al., Correlated imaging, quantum and classical, Phys. Rev. A 70 (2004) 013802.
3. R.S. Bennink et al., Quantum and classical coincidence imaging, Phys. Rev. Lett. 92 (2004)
033601.
4. B.I. Erkmen and J.H. Shapiro, Signal-to-noise ratio of Gaussian-state ghost imaging, Phys.
Rev. A 79 (2009) 023833.
5. R.S. Aspden et al., Photon-sparse microscopy: visible light imaging using infrared illumina-
tion, Optica 2 (2015) 1049.
6. T.B. Pittman et al., Optical imaging by means of two-photon quantum entanglement, Phys. Rev.
A 52 (1995) R3429.
7. K.W.C. Chan, M.N. O’Sullivan and R.W. Boyd, Two-color ghost imaging, Phys. Rev. A
79 (2009) 033808.
8. Gregory et al., Imaging through noise with quantum illumination, Sci. Adv. 6 (2020) eaay2652.
9. G. Brida et al., Experimental realization of sub-shot-noise quantum imaging, Nature Photonics
4 (2010) 227.
10. S.V. Polyakov and A.L. Migdall, High accuracy verification of a correlated-photon-based
method for determining photon-counting detection efficiency, Opt. Express 15 (2007) 1390.
11. D.N. Klyshko, Use of two-photon light for absolute calibration of photoelectric detectors,
Quantum Electron. 7 (1980) 1932.
12. A. Elitzur and L. Vaidman, Quantum Mechanical Interaction-Free Measurements, Found.
Phys. 23 (1993) 987.
13. R.H. Dicke, Interaction-free quantum measurements: A paradox?, Am. J. Phys. 49 (1981) 925.
14. L. Vaidman, On the realization of interaction-free measurements, Quant. Opt. 6, 119 (1994).
15. P. Kwiat, H. Weinfurter, T. Herzog, A. Zeilinger, and M. A. Kasevich, Interaction-free
measurement, Phys. Rev. Lett. 74 (1995) 4763.
16. A.G. White, J. R. Mitchell, O. Nairz, and P. G. Kwiat, Interaction-free imaging, Phys. Rev. A
58, 605 (1998).
17. Y. Zhang et al., Interaction-free ghost-imaging of structured objects, Optics Express
27 (2019) 2212.
18. T. Tsegaye, E. Goobar, A. Karlsson, G. Björk, M.Y. Loh and K.H. Lim, Efficient interaction-
free measurements in a high-finesse interferometer, Phys. Rev. A 57 (1998) 3987.
Chapter 19
Light–Matter Interaction

The interaction between light and matter is essential for the generation, manipula-
tion, and detection of quantum states used in quantum information processing. In this
chapter, we describe the quantum treatment of light–matter interaction. The Jaynes–
Cummings Hamiltonian is derived and used to explain spontaneous emission and
Rabi oscillations. Cavity quantum electrodynamics and Rydberg atoms are intro-
duced as a means of enhancing light–matter interaction. Collapse and revival of Rabi
oscillations is examined for coherent light in a cavity. The dressed atom–cavity
states are derived and used to explain vacuum Rabi splitting.

19.1 Jaynes–Cummings Hamiltonian

When dealing with the interaction of light with matter, a semiclassical description is
often employed where the atom is quantized but light is treated as a classical wave
(Fig. 19.1a). Here, we introduce the fully quantum approach where both the atom
and light are quantized (Fig. 19.1b). We assume the atom has two states, a ground
state labelled j0i and an excited state labelled j1i, with energy separation ħω10.
These two states are orthonormal: hij ji ¼ δij with i, j ¼ 0 or 1. The light field is
quantized with orthonormal number states jni, and photon energy ħω separating the
number states.
The Hamiltonian of the combined atom–light system is

b ¼H
H bR þ H
bA þ H bI ð19:1Þ

where H b R is the light (radiation) Hamiltonian, and H

b A is the atomic Hamiltonian, H bI
is an interaction Hamiltonian describing the interaction between the light and
the atom.
The atomic Hamiltonian can be written as

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 183
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_19
184 19 Light–Matter Interaction

Fig. 19.1 Light–matter Light field Atom

interaction.
(a) Semiclassical approach
where light is treated as a (a) Semi-classical:
classical wave and the atom
is quantized. (b) Fully
quantum approach where
both the atom and light are
quantized (b) Fully quantum:

b A ¼ ħω10 j1ih1j
H ð19:2Þ

where the reference energy is the ground state (E0 ¼ 0). If we apply Hb A to the atomic
b
ground state, we get H A j0i ¼ ħω10 j1ih1j0i ¼ 0, since h1j 0i ¼ 0. If we apply H b A to
b
the excited state, we get H A j1i ¼ ħω10 j1ih1j1i ¼ ħω10 j1i, since h1j 1i ¼ 1. Thus,
b A has eigenstates j0i and j1i and energy eigenvalues 0 and ħω10 , respectively.
H
Equation (19.2) can also be written as

b A ¼ ħω10 b
H σ{b
σ ð19:3Þ

where

σ { ¼ j1ih0j
b ð19:4Þ

and

b
σ ¼ j0ih1j ð19:5Þ

σ{b
Note that b σ ¼ j1ih0j0ih1j ¼ j1ih1j since h0j0i ¼ 1, so we obtain Eq. (19.2)
from (19.3). bσ { is a raising operator because bσ { j0i ¼ j1ih0j0i ¼ j1i, thus raising the
atomic state from j0i to j1i. b σ is a lowering operator because b σ j1i ¼ j0ih1j1i ¼ j0i,
thus lowering the atomic state from j1i to j0i. Consequently, b σ { and bσ are called
{
raising and lowering operators, analogous to b a and b a for the radiation. Note that
σ { j1i ¼ j1ih0j1i ¼ 0 because there is no state above j1i, and b
b σ j0i ¼ j0ih1j0i ¼ 0
because there is no state below j0i; that is, in the two-level atomic system, there are
no states above the excited state nor below the ground state.
The radiation Hamiltonian is the familiar form for a quantum harmonic oscillator:

b R ¼ ħωb
H a{ b
a ð19:6Þ
19.1 Jaynes–Cummings Hamiltonian 185

where the vacuum energy (12 ħω) is set equal to zero by assuming we have a lot of
field quanta so that the vacuum energy is comparatively negligible.
Using the classical expression for the dipole energy, we can write the interaction
Hamiltonian as

H I ¼ b b ðrÞ
p∙E ð19:7Þ

where b b ðrÞ is the electric field operator

p is the atomic dipole moment operator and E
for the light. Quantum mechanically, the dipole moment operator can be written as

b
p ¼ p ðj1ih0j þ j0ih1jÞ ð19:8Þ

We see that b
p couples the j0i and j1i state. Using Eqs. (19.4) and (19.5), we can write
b
p in the form:
 
b
p¼p bσ{ þ b
σ ð19:9Þ

By choosing a position such that k ∙ r ¼ π2 in Eq. (3.44), the electric field operator can
be written in a convenient form:
 
b ¼ εε1 b
E a{
aþb ð19:10Þ

where ε1 is the one photon amplitude:

rffiffiffiffiffiffiffiffiffiffi
ħω
ε ¼
1
ð19:11Þ
2Eo V

and ε is the electric field polarization. Thus, the interaction Hamiltonian in Eq. (19.7)
can be written as
  
b I ¼ ħg b
H σ{ þ b
σ b a{
aþb ð19:12Þ

g describes the strength of the light–matter coupling and is given by

rffiffiffiffiffiffiffiffiffiffiffiffi
p ∙ ε ε1 ω
g¼ ¼ p∙ε ð19:13Þ
ħ 2ħEo V

Equation (19.12) contains four terms. The b σ {b

a term describes the absorption
(annihilation) of a photon and creation of an atomic excitation. The b a{ term
σ {b
describes the creation of an atomic excitation and creation (emission) of a photon,
which violates conservation of energy. The b σb
a term describes the destruction of an
atomic excitation and annihilation (absorption) of a photon, which also violates the
186 19 Light–Matter Interaction

a{ term describes the destruction of an atomic

σb
conservation of energy. Finally, the b
excitation and emission of a photon. The b a{ and b
σ {b σb
a terms are eliminated, since
they violate the conservation of energy. Thus, only the bσ {b a{ terms are kept,
σb
a and b
giving:
 
b I ¼ ħg b
H σ {b a{
σb
aþb ð19:14Þ

The operators have a time dependence given by

b að0Þeiωt
að t Þ ¼ b ð19:15Þ

a{ ðt Þ ¼ b
b a{ ð0Þeiωt ð19:16Þ
b σ ð0Þeiω10 t
σ ðt Þ ¼ b ð19:17Þ

σ { ðt Þ ¼ b
b σ { ð0Þeiω10 t ð19:18Þ

Thus, in the Heisenberg picture, the interaction Hamiltonian from Eq. (19.12) takes
the form:
 
b I ¼ ħg b
H σ {b
aeiðω10 ωÞt þ b a{ eiðω10 þωÞt þ b
σ {b aeiðω10 þωÞt 1 b
σb a{ eiðω10 ωÞt ð19:19Þ
σb

The b a{ term and the b

σ {b σb
a term vary much more rapidly compared to the other two
terms. Thus, the fast evolution of the b a{ and b
σ {b σb
a terms averages to zero during the
much slower time evolution of the other two terms. Ignoring the b a{ and b
σ {b σba terms in
this manner is called the “rotating wave approximation (RWA)”. This provides
another justification for ignoring these terms, besides being unphysical (violating
conservation of energy) as described above.
Combining Eqs. (19.2), (19.6) and (19.14) into Eq. (19.1) gives
 
b ¼ ħω10 b
H σ{b a{ b
σ þ ħωb a þ ħg bσ {b a{
σb
aþb ð19:20Þ

Equation (19.20) is known as the Jaynes–Cummings Hamiltonian, introduced by

E.T. Jaynes and F.W. Cummings in the 1960s [1]. Although relatively simple, the
Jaynes–Cummings Hamiltonian predicts a wide range of phenomenon.

19.2 Spontaneous Emission

The light–matter system in Fig. 19.1 can be described by a quantum state ji, ni where
i ¼ 0 or 1 indicates the atomic state, and n ¼ 0, 1, 2, . . . indicates the Fock (number)
state of the light. Suppose the initial state is j0, ni, meaning the atom occupies the
19.3 Cavity Quantum Electrodynamics 187

ground state, and the light field contains n photons in the Fock state. The absorption
of a photon results in the final state j1, n  1i; that is, the atom is promoted to the
excited state and the field loses one photon by annihilation (absorption). The dipole
coupling between the initial and final states is given by

b I j0, ni ¼ h1, n  1jħgðb

h1, n  1jH σ {b a{ Þj0, ni
σb
aþb ð19:21Þ
pffiffiffi
¼ ħg n ð19:22Þ

Conversely, the emission of a photon from an initial state j1, ni results in the final
state j0, n þ 1i. The dipole coupling between the initial and final states in this case is
 
b I j1, ni ¼ h0, n þ 1jħg b
h0, n þ 1jH σ {b a{ j1, ni
σb
aþb ð19:23Þ
pffiffiffiffiffiffiffiffiffiffiffi
¼ ħg n þ 1 ð19:24Þ

Exercise 19.1 Derive Eqs. (19.22) and (19.24).

Equation (19.24) indicates that we can have emission of a photon from the excited
atom even in the case of n ¼ 0; that is, the initial state is vacuum. This explains the
process of spontaneous emission from an atom in an excited state even in the absence
of any externally applied field. We can think of the field fluctuations in the vacuum
as causing spontaneous emission.

19.3 Cavity Quantum Electrodynamics

Equation (19.14) indicates that strong light–matter interactions can occur when the
light–atom coupling, g, is large. Equation (19.13) shows that one method of making
g large is to make the mode volume V small. This can be done by trapping the atom
in a cavity of small volume formed by two mirrors, as shown in Fig. 19.2 (methods
of atom trapping are discussed in Chap. 21). This approach is known as cavity
quantum electrodynamics (CQED). In this case, we can think of the atom-photon
interaction as being enhanced by the many round trips that a photon makes across the
atom due to reflection back and forth between the mirrors [2]. There are many other
ways of implementing CQED such as micropillars, microdisks, microspheres, and
photonic crystals, as shown in Fig. 19.3 [3]. We have already seen some other
consequences of CQED such as the Purcell effect in Chap. 4.
188 19 Light–Matter Interaction

Mirror Mirror

Fig. 19.2 Illustration of cavity quantum electrodynamics (CQED). γ, T1 and κ denote various loss
mechanisms (γ: rate of spontaneous emission into free space; κ: rate of photon transmission due to
finite reflectivity of the cavity mirrors; T1: rate of atom escape from the cavity)

Fig. 19.3 Illustration of various CQED implementations. The microcavities are organized by
column according to the confinement method used and by row according to quality factor Q of
the cavity. Upper row: micropost or pillar, microdisk, add/drop filter, photonic crystal cavity.
Lower row: Fabry-Perot bulk optical cavity, microsphere, and microtoroid. n is the material
refractive index, V is the mode volume, Q is the quality factor, and F is the finesse. Two
Q values are cited for the add/drop filter: one for a polymer design, QPoly, and the second for a
III–V semiconductor design, QIIIV. (Reprinted by permission from Springer Nature, Vahala
[3]. Copyright 2003)

Exercise 19.2 Explore the microcavities in Fig. 19.3 and explain how the
devices are fabricated.
19.5 Rydberg Atoms 189

19.4 Circuit QED

Another way to enhance atom–light interaction is to create “artificial atoms”, a

system that behaves like a two-level atom but with a stronger coupling. A common
“artificial atom” is a superconducting circuit containing Josephson junction circuit
elements that act as a two-level system. The superconducting circuit (Fig. 19.4c) is
coupled to a microwave resonator (Fig. 19.4a, c). Due to the small mode volumes
made possible by the microwave waveguide, the g coupling between microwave
photons and the superconducting circuit can be orders of magnitude larger than real
atomic systems. These superconducting circuits are called “circuit QED” (cQED)
and are presently among the leading technologies in quantum computing.

19.5 Rydberg Atoms

Another method to increase the atom–light coupling, g, according to Eq. (19.13), is

to increase the dipole moment p of the atom in the cavity of a CQED system.
Rydberg atoms are atoms with a valence electron excited to a high principal quantum
number (n~50) with a large orbital radius. In such an orbit, the balance between the

Fig. 19.4 (a) Optical micrograph showing circuit quantum electrodynamics (cQED). The trans-
mission line “wiggles” to increase its cavity length. (b) The cavity mirror is formed by a gap in the
transmission line. (c) The superconducting circuit (called a “transmon qubit”) with two Josephson
junctions. (Reprinted by permission from Springer Nature, Wallraff et al. [4]. Copyright 2004)
190 19 Light–Matter Interaction

Fig. 19.5 A Rydberg atom Hydrogen atom Rydberg atom

resembles a hydrogen atom
proton: electron:
(not to scale!) electron:
core:

charge of the inner electrons and the nucleus produces a hydrogen-like system with
charge þe surrounded by a single valence electron. Thus, the valence electron feels a
hydrogen-like Coulomb potential (Fig. 19.5). According to Bohr’s simple model of
the hydrogen atom, familiar from introductory quantum mechanics, the radius of the
electron orbit is given by

r ¼ n2 a0 ð19:25Þ

where a0~0.53 Å is the Bohr radius and n is the principal quantum number.
According to Eq. (19.25), the average radius of the electron orbit scales as n2, and
the resulting electron orbit is about 2500 atomic diameters for n~50; that is, on the
order of 0.1 micron! This results in a huge dipole moment ( p = er) due to the large
electron-nucleus separation (r).
Bohr’s model also accurately predicts the wavelengths of the energy transitions,
known as the Rydberg formula (hence the name, “Rydberg atoms”):
 
1 1 1
¼ RH  ð19:26Þ
λ n1 2 n 2 2

where RH~1.0974  107 m1 is the Rydberg constant for hydrogen, and n1 and n2
are the principal quantum numbers associated with the transition. For large n and for
transitions between adjacent energy levels, we can approximate Eq. (19.26) as
 
1 2
¼ RH 3 ð19:27Þ
λ n

For n~50, Eq. (19.27) gives λ~6 mm, that is, in the microwave range. This gives us
an idea of the required cavity dimensions to support a single mode field. Using
Rydberg states, interactions between atoms are more easily controlled [5, 6]. A
review of Rydberg atoms in the context of quantum computing is available in
Ref. [7].
19.6 Rabi Oscillations 191

19.6 Rabi Oscillations

Suppose the initial state of an atom–light system in a cavity is j1, ni, meaning we
start in the excited state of the atom (j1i) with n photons in the cavity (jni).
Following an atomic transition to the ground state accompanied by photon emission,
the state becomes j0, n þ 1i; that is, we add one photon to the cavity. The atom–light
system can also exist in a superposition of these two states:

jψðt Þi ¼ c1,n ðt Þj1, ni þ c0,nþ1 ðt Þj0, n þ 1i ð19:28Þ

In general, Eq. (19.28) represents an entangled state of the combined atom-field

system.
The dynamics of the system are given by the time-dependent Schrodinger
equation:

d j ψð t Þ i b
iħ ¼ H jψðt Þi ð19:29Þ
dt

The right-hand side of Eq. (19.29) is evaluated using the Jaynes–Cummings

Hamiltonian:
h  i
b jψðt Þi ¼ ħω10 b
H σ{b a{ b
σ þ ħωb σ {b
a þ ħg b a{
σb
aþb

 ðc1,n j1, ni þ c0,nþ1 j0, n þ 1iÞ ð19:30Þ

Using Eq. (2.127) and (2.128), we get

b jψðt Þi ¼ ħω10 c1,n j1, ni þ ħωnc1,n j1, ni þ ħωðn þ 1Þc0,nþ1 j0, n þ 1i

H
pffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffi
þ ħgc0,nþ1 n þ 1 j1, ni þ ħgc1,n n þ 1 j0, n þ 1i ð19:31Þ
pffiffiffiffiffiffiffiffiffiffiffi
¼ E i c1,n j1, ni þ Ef c0,nþ1 j0, n þ 1i þ ħgc0,nþ1 n þ 1 j1, ni
pffiffiffiffiffiffiffiffiffiffiffi
þ ħgc1,n n þ 1 j0, n þ 1i ð19:32Þ

where E i ¼ ħω10 þ ħωn is the initial energy with the atom in the excited state and
n photons in the field, and Ef ¼ ħωðn þ 1Þ is the final energy with the atom in the
ground state and n þ 1 photons in the field. Here, we assume ħω10 ¼ ħω; that is, the
cavity field is resonant with the atomic transition. The left-hand side of Eq. (19.29) is
192 19 Light–Matter Interaction

d j ψð t Þ i d j1, ni
iħ _ j1, ni þ iħc1,n
¼ iħc1,n þ iħ_c0,nþ1 j0, n þ 1i
dt dt
dj0, n þ 1i
þ iħc0,nþ1 ð19:33Þ
dt
¼ iħ_c1,n j1, ni þ E i c1,n j1, ni þ iħ_c0,nþ1 j0, n þ 1i þ E f c0,nþ1 j0, n þ 1i ð19:34Þ

Equating Eqs. (19.32) and (19.34) gives

pffiffiffiffiffiffiffiffiffiffiffi
c_ 1,n ¼ ig n þ 1 c0,nþ1 ð19:35Þ

and
pffiffiffiffiffiffiffiffiffiffiffi
c_ 0,nþ1 ¼ ig n þ 1 c1,n ð19:36Þ

Equations (19.35) and (19.36) are two coupled differential equations. Combining
these equations gives

€c1,n ¼ g2 ðn þ 1Þc1,n ð19:37Þ

The initial condition is j1, ni; that is, the atom is initially in the excited state. Thus, c1,
n(0) ¼ 1 and c0, n þ 1(0) ¼ 0. Solving Eq. (19.37) with this initial condition gives

 pffiffiffiffiffiffiffiffiffiffiffi
c1,n ðt Þ ¼ cos gt n þ 1 ð19:38Þ

and substituting Eq. (19.38) into (19.36) and solving gives

 pffiffiffiffiffiffiffiffiffiffiffi
c0,nþ1 ðt Þ ¼ i sin gt n þ 1 ð19:39Þ

Finally, substituting Eqs. (19.38) and (19.39) into Eq. (19.28) gives
   
Ωn t Ωn t
jψðt Þi ¼ cos j1, ni  i sin j0, n þ 1i ð19:40Þ
2 2

where
pffiffiffiffiffiffiffiffiffiffiffi
Ωn ¼ 2g n þ 1 ð19:41Þ

Ωn is called the Rabi frequency, named after Isidor I. Rabi (Fig. 19.6).
In the semiclassical model, where the atom is quantized but the light field is
treated as a classical field, the Rabi frequency is Ω ¼ p ∙ E=ħ where p ∙ E is the
classical expression for the dipole energy. In the fully quantum model, where both
the atom and the field are quantized, the Rabi frequency is quantized according to
Eq. (19.41).
19.6 Rabi Oscillations 193

Fig. 19.6 Isidor I. Rabi

(Nobel Prize in Physics in
1944). (Credit: Wikimedia
Commons [8])

Fig. 19.7 Probabilities

Probability

P1, n(t) (dashed line) and

P0, n þ 1(t) (solid line),
showing Rabi oscillations

According to Eq. (19.40), the probability of the atom-field system being in the
j1, ni state is
 
Ωn t
P1,n ðt Þ ¼ jc1,n j2 ¼ cos 2 ð19:42Þ
2

and the probability of being in the j0, n þ 1i state is

 
Ωn t
P0,nþ1 ðt Þ ¼ jc0,nþ1 j2 ¼ sin 2 ð19:43Þ
2

We see that the state oscillates between the j1, ni and j0, n þ 1i state with the Rabi
frequency Ωn. These are called Rabi oscillations as shown in Fig. 19.7.
Equation (19.41) shows that Rabi oscillations exist even in the initial absence of
light (n ¼ 0) with frequency Ω0 ¼ 2g. These are called vacuum Rabi oscillations,
caused by spontaneous emission of the atom by the vacuum. The emitted photon is
given up to the cavity where it can be repeatedly absorbed and remitted by the atom.
The observation of vacuum Rabi oscillations was demonstrated in Ref. [9].
Of course, no system is perfect, and the Rabi oscillations eventually decay due to
losses and decoherence associated with coupling to the environment (Fig. 19.2). By
making the light–atom coupling (g) large by using a cavity, many Rabi oscillations
are possible before photon loss due to spontaneous emission into free space (rate γ),
194 19 Light–Matter Interaction

photon transmission loss due to finite reflectivity of the cavity mirrors (rate κ), and
atom escape from the cavity (rate T1), as shown in Fig. 19.2. CQED aims to have
g  κ, γ, T1 so that many cycles of the Rabi oscillations can be observed before
decay. This is the so-called strong coupling regime of CQED where light and atomic
quanta play a dominant role in the system dynamics. Often, one characterizes the
system by combining the parameters into a single dimensionless quantity called the
cooperativity, defined as (assuming T1 is negligible):

g2
C¼ ð19:44Þ
κγ

Obviously, one wants C  1 for the strong coupling regime. In the strong coupling
regime, irreversible spontaneous emission changes to a reversible exchange of
energy between the atom and the cavity mode. With sufficiently strong coupling,
one can even make a maser (microwave laser), for example, by passing Rydberg
atoms one at a time through a cavity, where field buildup occurs by cumulative
atomic emission from each atom [10].

Exercise 19.3 Give typical values for g, κ, γ, and T1 for a few cavity systems
(e.g., from Fig. 9.3).

19.7 Collapse and Revival of Rabi Oscillations

Let us suppose that a coherent state exists inside a cavity with a two-level atom.
Recall that the coherent state in the number representation is

X
1
jα i ¼ c n j ni ð19:45Þ
n¼0

with Poisson probability distribution of the photon number n:

hnin
jcn j2 ¼ ehni ð19:46Þ
n!

In the coherent state, each of the Fock states n will undergo Rabi oscillation with
frequency given by Eq. (19.41). Let’s see what happens when we superimpose all
these Rabi oscillations.
We assume the atom is initially in a superposition of the ground and excited state:

jψð0Þiatom ¼ c0 j0i þ c1 j1i ð19:47Þ

Also, we assume the field is initially in a coherent state according to Eq. (19.45):
19.7 Collapse and Revival of Rabi Oscillations 195

X
1
jψð0Þifield ¼ c n j ni ð19:48Þ
n¼0

The initial atom-field state is then given by the tensor product:

jψð0Þi ¼ jψð0Þiatom  jψð0Þifield ð19:49Þ

If we take the case of c0 ¼ 0 and c1 ¼ 1 (the atom is initially in the excited state), then
the solution to Schrodinger’s equation (similar to Eqs. (19.38) and (19.39)) gives

X
1  pffiffiffiffiffiffiffiffiffiffiffi  pffiffiffiffiffiffiffiffiffiffiffi

j ψð t Þ i ¼ cn cos gt n þ 1 j1, ni  i sin gt n þ 1 j0, n þ 1i ð19:50Þ

n¼0

We define the atomic inversion, which is the difference in probabilities of the

atom in the excited and ground state:

W ðt Þ ¼ P1 ðt Þ  P0 ðt Þ ð19:51Þ
X
1  pffiffiffiffiffiffiffiffiffiffiffi
¼ jcn j2 cos 2gt n þ 1 ð19:52Þ
n¼0

X1
h ni n  pffiffiffiffiffiffiffiffiffiffiffi
¼ ehni cos 2gt n þ 1 ð19:53Þ
n¼0
n!

Equation (19.53) is plotted in Fig. 19.8, pshowing

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the Rabi oscillations. For large
coherent fields, the Rabi frequency is 2g phffiffiffiffiffiffi
niffi þ 1, associated with the mean photon
number hni, since the dispersion Δn ¼ hni is small. However, the Rabi oscilla-
tions eventually decay (collapse), followed by a revival, followed by another
collapse, etc. This is known as collapse and revival of the Rabi oscillations. This
occurs due to the interference of Rabi oscillations with different quantized Rabi
frequencies. The Rabi oscillations fall in and out of phase, leading to destructive and

(a) 1 (b) 1
W(t)

W(t)

0 0

-1 -1
0 10 20 30 40 50 60 0 10 20 30 40 50 60
tg tg

Fig. 19.8 Collapse and revival of Rabi oscillations for average photon number of (a) hni ¼ 10 and
(b) hni ¼ 30. Time t is in units of 1/g
196 19 Light–Matter Interaction

constructive interference. In other words, a beat note is formed between the different
Rabi frequencies.
Note that Rabi oscillations can be explained by a semiclassical theory where the
atomic system has quantized energy levels but the light field is classical. However,
the repeated collapse and revival of the Rabi oscillations can only be explained with
quantized fields via the Jaynes–Cummings model. This is the reason why collapse
and revival of the Rabi oscillations are of such great interest in quantum optics. The
first observation of Rabi collapse and revival is Ref. [11].
The collapse of the Rabi oscillations occurs when all the different Fock state
components become out of phase by π and destructively interfere. For a coherent
field with large average photon number hni, we can approximate
pffiffiffiffiffiffi
ffi the photon distri-
bution as a Gaussian with a standard deviation hni: Thus, we can estimate the
collapse time tC as occurring when

ΩhniþΔn t C  ΩhniΔn t C ¼ π ð19:54Þ

According to Eq. (19.54), we get

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2g hni þ Δn t C  2g hni  Δn t C ¼ π ð19:55Þ

or
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffi 1 pffiffiffiffiffiffiffi 1
2g hni 1 þ pffiffiffiffiffiffiffi t C  2g hni 1  pffiffiffiffiffiffiffi t C ¼ π ð19:56Þ
h ni h ni
pffiffiffiffiffiffiffi
Since 1= hni  1 for large hni, we can use a binomial expansion in Eq. (19.56),
which gives
! !
pffiffiffiffiffiffiffi 1 pffiffiffiffiffiffiffi 1
2g hni 1 þ pffiffiffiffiffiffiffi t C  2g hni 1  pffiffiffiffiffiffiffi t C ¼ π ð19:57Þ
2 h ni 2 h ni

or

π
tC ¼ ð19:58Þ
2g

Thus, the collapse time is independent of photon number as observed qualitatively in

Fig. 19.8.
We expect the Rabi oscillations to begin reviving at a time tR when neighboring
Fock states become in phase giving constructive interference. Thus,
19.8 Example of a CQED Experiment 197

Ωhniþ1 t R  Ωhni t R ¼ 2π ð19:59Þ

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffi
2g hni þ 1 tR  2g hni t R ¼ 2π ð19:60Þ
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffi 1 pffiffiffiffiffiffiffi
2g hni 1 þ t R  2g hni t R ¼ 2π ð19:61Þ
h ni

Using a binomial expansion, we get

 
pffiffiffiffiffiffiffi 1 pffiffiffiffiffiffiffi
2g hni 1 þ t R  2g hni t R ¼ 2π ð19:62Þ
2 h ni

or
pffiffiffiffiffiffiffi
2π h ni
tR ¼ ð19:63Þ
g

Thus, the revival time is related to the average photon number, as observed quali-
tatively in Fig. 19.8.

19.8 Example of a CQED Experiment

Figure 19.9 illustrates a cavity QED experiment [12]. An oven (O) produces a beam
of rubidium atoms which are then prepared into Rydberg states (B). The Rydberg
atoms enter a cavity (C) containing a coherent cavity mode produced by a micro-
wave generator (S). While in the cavity, the Rydberg atoms undergo Rabi oscilla-
tions between two principal quantum numbers (n ¼ 50 and 51). The Rydberg atoms
exit the cavity and their state is detected. The detection mechanism uses an electric
field to strip the electron in the excited state (n ¼ 51) off the atom where it is then
detected with a channeltron electron multiplier. If the atom is in the ground state
(n ¼ 50), the electric field is not sufficient to ionize the atom. Thus, the state of the
atom (ground or excited state) can be determined by charge detection. The

Fig. 19.9 Experimental

setup for the results of
Fig. 19.10. O: oven; B:
Rydberg atom state
preparation; S: coherent
source; C: cavity; D:
detector. (Reprinted with
permission from Brune et al.
[12]. Copyright 1996 by the
American Physical Society)
198 19 Light–Matter Interaction

Fig. 19.10 The probability versus time of finding the Rydberg atom in the ground state, showing
Rabi oscillations due to (A) vacuum field with hni ¼ 0.06 (due to a small thermal field) and (B–D)
coherent field with hni ¼ 0.40, 0.85 and 1.77, respectively. The points are experimental (with error
bars in (A) only for clarity); the solid lines correspond to theoretical fits. (a–d) Corresponding
pffiffiffi pffiffiffi pffiffiffi
Fourier transforms. Frequencies ν ¼ 47 kHz, 2ν, 3ν, and 4ν are indicated by vertical dotted
lines. (α – δ) Corresponding photon number distribution inferred from experimental signals
(points). Solid lines show the theoretical thermal (α) or coherent (β, γ, δ) distributions, which
best fit the data. (Reprinted with permission from Brune et al. [12]. Copyright 1996 by the American
Physical Society)

experiment is repeated with many different travel times through the cavity due to the
Maxwell velocity distribution of the atoms. The travel times are known according to
the difference in time between the state preparation and detection of the atom. Thus,
probabilities of ground and excited state populations can be obtained as a function of
time and plotted to observe the Rabi oscillations.
Figure 19.10 shows the results of an experiment performed using the setup of
Fig. 19.9 [12]. It is illustrative of the kind of results that are obtained in cavity QED
19.9 Dressed Atom–Cavity States and Vacuum Rabi Splitting 199

experiments. Figure 19.10(A) shows vacuum Rabi oscillations with a frequency

Ω0 ¼ 47 kHz obtained from the Rb atoms passing through the cavity. In Fig. 19.10
(B–D), different microwave field amplitudes were turned on, corresponding to a
coherent field with different average photon numbers in the cavity. Here, a weak
collapse and revival of the oscillations can be observed. Figure 19.10(a–d) are the
Fourier transforms of (A–D), showing the discrete frequency components of the
Rabi oscillations and the quantized nature of the field. Figure 19.10(a) shows the
case of vacuum Rabi oscillations with frequency Ω0 ¼ 47 kHz, while Fig. 19.10(b–
d) show additional frequency components when the microwave generator is turned
on, corresponding to the coherent field. The Rabi frequencies of each Fock state in
pffiffiffiffiffiffiffiffiffiffiffi
the coherent field are given by Ωn p ¼ffiffiffiΩ0 n þ 1 , according to Eq. (19.41). Fig-
ure 19.10(b–d) shows the expected n nonlinearity. Figure 19.10(α–δ) shows the
amplitudes of the frequency components. The amplitude components follow a
thermal radiation distribution in Fig. 19.10(α) in the absence of any microwave
photons, and a Poisson distribution in (β–δ) as expected for a coherent field (with the
microwave source turned on). The average photon numbers are obtained from these
distributions.

19.9 Dressed Atom–Cavity States and Vacuum Rabi

Splitting

In Fig. 19.1b, we treated the quantized levels of the atom and of the cavity field as
separate. The system is described by states of the form:

jψ1 i ¼ j1, ni ð19:64Þ

jψ2 i ¼ j0, n þ 1i ð19:65Þ

Using these as the basis states, we can write the Jaynes–Cummings Hamiltonian as a
matrix with elements given by
D E
b ij ¼ ψi jHjψ
H b j ð19:66Þ

where i and j are indices indicating the matrix element (i, j ¼ 0, 1), which gives
0 pffiffiffiffiffiffiffiffiffiffiffi 1
ħω
ħωn þ 10 ħg n þ 1
B
b ¼@ 2 C
H pffiffiffiffiffiffiffiffiffiffiffi A ð19:67Þ
ħω
ħg n þ 1 ħωðn þ 1Þ  10
2

where we have shifted the zero of energy to ħω210 between the atomic ground and
excited state energy, and we have ignored the zero-point energy. The diagonal
200 19 Light–Matter Interaction

elements correspond to the energy eigenvalues without the interaction term (ground
state energy with n þ 1 photons in the field, and excited state energy with n photons
in the field), while the off-diagonal elements correspond to the interaction term of the
Hamiltonian (dipole interaction).
At resonance (ω ¼ ω10), the new eigenstates of the coupled atom–cavity system
are

1
jn i ¼ pffiffiffi ðj1, ni þ j0, n þ 1iÞ ð19:68Þ
2

or, in the vector representation,

 
1 1
jn i ¼ pffiffiffi ð19:69Þ
2 1

with corresponding eigenvalues:

  pffiffiffiffiffiffiffiffiffiffiffi
1
E  ¼ ħω n þ  ħg n þ 1 ð19:70Þ
2

jni are called the dressed states (i.e., dressed by the photons), while the basis states,
j1, ni and j0, n þ 1i, are called the “bare” states. The dressed states are entangled;
that is, they cannot be expressed as a tensor product of states involving only the atom
and another part involving only the field.

Exercise 19.4
Check that Eqs. (19.68) and (19.70) satisfy the time-independent Schrodinger
b jn i ¼ E  jn i.
equation, H

For the off-resonance case (ω 6¼ ω10), the solution to Schrodinger’s equation

gives
 
1 ħ
E  ¼ ħω n þ  Ωn ðΔ Þ ð19:71Þ
2 2

where Ωn(Δ) is called the generalized Rabi frequency that is defined as

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Ωn ð Δ Þ ¼ Δ2 þ 4g2 ðn þ 1Þ ¼ Δ2 þ Ωn 2 ¼ Δ2 þ Ω0 2 ðn þ 1Þ ð19:72Þ

and Δ is called the detuning that is defined as

Δ ¼ ω10  ω ð19:73Þ

The eigenstates corresponding to Eq. (19.71) become

19.9 Dressed Atom–Cavity States and Vacuum Rabi Splitting 201

jnþ i ¼ cos ðθ=2Þj1, ni þ sin ðθ=2Þj0, n þ 1i ð19:74Þ

and

jn i ¼  sin ðθ=2Þj1, ni þ cos ðθ=2Þj0, n þ 1i ð19:75Þ

where θ is called the mixing angle with

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 Ωn ðΔÞ  Δ
sin ðθ=2Þ ¼ pffiffiffi ð19:76Þ
2 Ωn ðΔÞ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 Ωn ð Δ Þ þ Δ
cos ðθ=2Þ ¼ pffiffiffi ð19:77Þ
2 Ωn ðΔÞ

Exercise 19.5 Show that at resonance, Eqs. (19.74) and (19.75) reduce to
Eq. (19.68), and Eq. (19.71) reduces to Eq. (19.70). Show that far from
resonance, Eqs. (19.74) and (19.75) reduce to the bare states, j1, ni and j0,
n þ 1i.

We can represent the energy levels as shown in Fig. 19.11, which is called the
“Jaynes–Cummings ladder”. The uncoupled energy levels are shown on the left of
Fig. 19.11. At resonance, each pair of levels (j1, ni, j0, n þ 1i) would be degenerate.
Off resonance, there is a difference of energy (ħΔ) for the j1, ni and j0, n þ 1i state.
Figure 19.11 shows the case of red-detuning (Δ = ω10  ω > 0). The energy levels
are split by the atom-field coupling, as shown on the right of Fig. 19.11. This shift of
the energy levels is called the “AC Stark effect”. We will consider this effect further
in Chap. 21 for its use in atom cooling.
The splitting of the energy levels can be measured by a pump-probe experiment
where a strong pump laser drives the splitting of the dressed states, while a weaker
probe laser drives a transition with another energy level. The observed energy
doublet is known as the Autler-Townes effect and is a manifestation of the AC
Stark effect.
The energy splitting near resonance is known as an “avoided crossing”. This is a
common phenomenon in other coupled oscillators such as two coupled pendulums
or two coupled LC electronic oscillators. The similar resonance frequencies of two
oscillators become split when the two oscillators are coupled together.
resonance (Δ ¼ 0), the coupled energy levels are split by an amount
Atpffiffiffiffiffiffiffiffiffiffiffi
ħg n þ 1 . The splitting at n ¼ 0 becomes ħg ¼ ħΩ0 =2, which is known as
the vacuum Rabi splitting (VRS). VRS was first observed in Ref. [13] by measuring
transmission spectra through a cavity with a single mode and one atom present on
average. Since then, VRS has been observed in a wide variety of systems [14].
The observation of VRS and corresponding avoided crossings is an indication of the
strong coupling regime.
202 19 Light–Matter Interaction

Fig. 19.11 Jaynes–

Cummings ladder for
red-detuned light. The left-
hand side shows the energy
levels for the uncoupled
atom–light states, and the
right-hand side shows the
energy levels for the
coupled atom–light states

References

1. E.T. Jaynes and F.W. Cummings, Comparison of quantum and semiclassical radiation theories
with application to the beam maser, Proc. IEEE 51 (1963) 89.
2. G. Rempe, R. J. Thompson, H. J. Kimble and R. Lalezari, Measurement of ultralow losses in an
optical interferometer, Optics Letters 17 (1992) 363.
3. K.J. Vahala, Optical microcavities, Nature 424 (2003) 839.
4. A. Wallraff et al., Strong coupling of a single photon to a superconducting qubit using circuit
quantum electrodynamics, Nature 431 (2004) 162.
5. E. Urban et al., Observation of Rydberg blockade between two atoms, Nature Phys.
5 (2009) 110.
6. A. Browaeys, D. Barredo and T. Lahaye, Experimental investigations of dipole–dipole inter-
actions between a few Rydberg atoms, J. Phys. B: At. Mol. Opt. Phys. 49 (2016) 152001.
7. M. Saffman, T.G. Walker and K. Mølmer, Quantum information with Rydberg atoms, Rev.
Mod. Phys. 82 (2010) 2313.
8. File: II Rabi.jpg. (2020, October 21). Wikimedia Commons, the free media repository. Retrieved
15:45, December 7, 2020 from https://commons.wikimedia.org/w/index.php?title¼File:II_
Rabi.jpg&oldid¼496554939.
9. R. Bose, T. Cai, K.R. Choudhury, G.S. Solomon and E. Waks, All-optical coherent control of
vacuum Rabi oscillations, Nature Photon. 8 (2014) 858.
10. D. Meschede, H. Walther and G. Müller, One-atom maser, Phys. Rev. Lett. 54 (1985) 551.
References 203

11. G. Rempe, H. Walther and N. Klein, Observation of quantum collapse and revival in a
one-atom maser, Phys. Rev. Lett. 58 (1987) 353.
12. M. Brune et al., Quantum Rabi oscillation: A direct test of field quantization in a cavity, Phys.
Rev. Lett. 76 (1996) 1800.
13. R.J. Thompson, G. Rempe, and H.J. Kimble, Observation of normal-mode splitting for an atom
in an optical cavity, Phys. Rev. Lett. 68 (1992) 1132.
14. G. Khitrova, Vacuum Rabi splitting in semiconductors, Nature Physics 2 (2006) 81.
Chapter 20
Atomic Clock

In the late 1930s, Isidor Rabi introduced the idea of using atomic resonances as
frequency standards, that is, as a means of keeping time, which eventually became
the atomic clock. The atomic clock is a perfect example of light-matter interaction
being put to practical use. Due to the atomic clock, we can measure time with less
uncertainty than any other physical quantity.

20.1 Quartz Oscillators—Before Atomic Time

To put the atomic clock in context, let us first describe a ubiquitous time-keeping
technology that predates the atomic clock—the quartz crystal oscillator. Bell labs
built the first quartz crystal oscillator clock in 1927, which soon began to replace
pendulums or other mechanical clocks as the standard for time measurement. Quartz
oscillators are based on the mechanical resonance (vibration) of a quartz crystal. An
AC voltage applied at the resonance frequency of the quartz crystal causes it to
continuously vibrate by the piezoelectric effect. The quartz oscillation frequency can
be tuned from a few kHz to hundreds of MHz, depending on the size and shape of the
quartz crystal and how it is cut. Due to the regularity of the vibration frequency, the
quartz crystals can be used for time-keeping and are much more accurate than
pendulums or other mechanical oscillators. They have a typical accuracy on the
order of 1 second per month (i.e., they will gain or lose on the order of 1 second per
month).

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 205
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_20
206 20 Atomic Clock

20.2 Resonance Frequency in the Cs Atom

Many modern applications require clocks that are much more accurate than the
quartz clock, which spurred the development of the atomic clock. The atomic clock
is based on the electron transition between two energy levels in the 133Cs atom.
These two levels, labelled E0 and E1 in Fig. 20.1, are associated with the hyperfine
splitting of the valence electron energy of the 133Cs atom with principal quantum
number n ¼ 6 and orbital angular momentum quantum number l ¼ 0. The hyperfine
splitting is due to the Zeeman effect because of the magnetic field from the nuclear
magnetic moment of the atom. The valence electron can occupy the lower energy
level with spin down (ħ=2Þ and the upper energy level with spin up (þħ=2).
A Cs atom with an electron prepared in the E0 state will absorb a photon when it is
subjected to an incident electromagnetic wave tuned to the resonance frequency, f0,
of the two levels according to

E1  E0
f0 ¼ ð20:1Þ
h

Continued application of the incident field causes stimulated emission of a photon as

the electron transitions back to E0 from E1. The incident field causes alternating
absorption and emission of a photon.
According to the international system of units (Système International d’unités,
informally known as the metric system), the second is defined by taking “the fixed
numerical value of the cesium frequency, f0, the unperturbed ground-state hyperfine
transition frequency of the cesium-133 atom, to be 9,192,631,770 when expressed in
the unit Hz, which is equal to s1”. The frequency f0 is defined to be exactly
9,192,631,770 Hz (i.e., in the microwave range) to closely match older definitions
of the second.

Fig. 20.1 Resonant absorption of a photon (depicted as the field in red) causes an electron
transition from E0 (with spin down) to E1 (with spin up) as shown on the left. Continued application
of the field causes stimulated emission of a photon and electron transition from E1 to E0 as shown on
the right
20.3 Stern–Gerlach Apparatus 207

20.3 Stern–Gerlach Apparatus

The atomic clock relies on an accurate measurement of the Cs transition frequency,

f0. This relies on the quantum state preparation of Cs atoms (i.e., preparing Cs atoms
in the ground state, E0). One way of performing the state preparation is the Stern–
Gerlach (SG) apparatus, shown in Fig. 20.2. The SG experiment was first performed
in 1922 by Otto Stern (Nobel Prize in Physics in 1943) and Walther Gerlach. In the
SG experiment, a beam of atoms is produced by an oven and collimated by a slit
(Ag atoms were used in the original SG experiment, while Cs atoms are used today
for the atomic clock). The beam of atoms is passed through an inhomogeneous
magnetic field produced by the shaped poles of a permanent magnet. This field will
interact with the magnetic dipole moment of the atom, if any, and deflect it. This is
easy to understand by consideration of the classical expression for the potential
energy, U, of a magnetic dipole moment oriented along a magnetic field, B. If the
magnetic field is predominantly along the z direction, then

U ¼ μz Bz ð20:2Þ

The force on the magnetic moment is

∂U ∂B
Fz ¼  ¼ μz z ð20:3Þ
∂z ∂z

The dipole moment will be deflected in the direction that decreases its potential
energy.
According to Eq. (20.2), a magnetic moment parallel to the magnetic field can
decrease its energy by moving to regions of higher magnetic field, while a magnetic
moment pointing opposite the direction of the magnetic field can decrease its energy
by moving to regions of lower magnetic field. Thus, the Cs atoms passing through
the SG apparatus will be split into two beams, one having atoms with spin-up
valence electron and the other having atoms with spin-down valence electron. By

Fig. 20.2 Schematic of the Inhomogeneous

Stern–Gerlach apparatus
Beam of B-field
atoms

Oven
Slit

Magnet
208 20 Atomic Clock

blocking one of the two beams and allowing the other beam to pass, the SG
apparatus can be used as a state selection machine; that is, the SG apparatus is a
filter for electron spin. An alternative method for state selection, which will not be
discussed here, is optical pumping (optical pumping is described briefly in
Chap. 21).

Exercise 20.1 Explain the physics responsible for deflection of Cs atoms into
two separate beams in the Stern–Gerlach apparatus, as shown in Fig. 20.2.

20.4 Thermal Atomic Clock

The thermal atomic clock (Fig. 20.3) is based on thermally generated beams of Cs
atoms. Cs is chosen because it has a low melting point making it easy to form a
vapor, it has only one stable isotope (the hyperfine splitting will be identical for all
atoms), it has a low ionization energy (it is easily ionized for detection), and it has a
large hyperfine splitting due to a large nuclear spin.
The Cs beam passes through a SG apparatus allowing only Cs atoms with spin
down (energy E0) to pass. Then a microwave field is applied to flip the spin by
resonant absorption to the level E1. Another spin filter (SG apparatus) allows only
atoms with spin up to pass to a detector. The Cs atoms are then detected, for
example, by ionization followed by an ion detector or by state-dependent fluores-
cence emission. The detector signal is maximum when the applied microwave
frequency exactly matches (is resonant with) the Cs transition frequency. The
detector signal is used in a feedback loop to lock the microwave oscillator to the

SG SG

Detector

Microwave cavity
Cs oven

Frequency
= 9,192,631,770 Hz
control

Quartz
oscillator
clock

Fig. 20.3 The thermal atomic clock. An oven and collimating slit produce a beam of Cs atoms. A
Stern–Gerlach (SG) apparatus is used to prepare all atoms in the spin-down state with energy E0
before entering a microwave cavity. The microwave field of the cavity is resonantly absorbed by the
Cs atoms causing a spin flip to the energy E1. A second SG filter only allows atoms in the E1 (spin
up) state to pass to a detector. The detector signal is used as feedback for the frequency control
20.5 Improvements to the Atomic Clock 209

Cs transition frequency, which in turn can be used to provide a very stable AC

voltage for a quartz crystal oscillator clock. Note that this description of the atomic
clock is somewhat simplified—in practice, the atomic transition (spin flip) is
achieved using a Ramsey interferometer. The interested reader may obtain further
information from Refs. [1, 2].

Exercise 20.2 Investigate and explain the working principle of the Ramsey
interferometer.

20.5 Improvements to the Atomic Clock

The accuracy of the atomic clock (or any clock) can be quantified by the Q factor,
defined as follows:

f0
Q¼ ð20:4Þ
Δf

where Δf is the spread or uncertainty in the transition frequency. A perfect clock

would have a perfectly defined frequency f0 with zero uncertainty (Δf ¼ 0), giving an
ideally infinite Q. Since all atoms of Cs are identical, the atomic oscillator essentially
forms an almost perfect oscillator with a high Q. The Q for a thermal atomic clock is
~1010, which is much better than quartz oscillators that have Q~104  106. Δf (and
hence Q) is ultimately limited by the natural linewidth of the atomic transition. In
practice, another contribution to Δf in the thermal atomic clock is the Doppler effect
that shifts the atomic transition frequency due to Doppler broadening because of the
different speeds of the atoms exiting the oven. In addition, the Cs atoms pass quickly
through the microwave cavity, limiting the interaction time. According to the
frequency-time uncertainty relation, the limited interaction time leads to a spread
in frequency.
Better clocks are made by trapping and cooling the atoms to reduce the Doppler
broadening and increase the interaction time. Atom trapping and cooling are treated
in the next chapter. Cs fountain clocks are based on the magneto-optical trapping of
Cs atoms and can realize an uncertainty of 1 s in several hundred million years.

Exercise 20.3 Investigate and explain the working principle of the Cs foun-
tain clock.

The next evolution of the atomic clock is to use atomic transitions in the optical
rather than the microwave range, using atoms such as 199Hg, 27Al, 40Ca, 174Yb and
87
Sr. The microwave field would be replaced by an optical laser to induce the atomic
transition. Optical frequencies are ~105 times higher than microwave frequencies
210 20 Atomic Clock

(THz versus GHz). This will further increase the Q factor and allow the measurement
of smaller time intervals, making huge reductions in uncertainty possible. The
measured frequencies of today’s optical clocks are estimated to gain or lose no
more than 1 s over the age of the universe!

20.6 Applications of the Atomic Clock

Coordinated Universal Time (UTC) serves as the official time reference for most of
the world. UTC is computed from a weighted average of a network of nearly
450 atomic clocks located around the world. Computers, phones, and other devices
have internal clocks (quartz oscillators) that need periodic correction. The clocks are
synchronized to UTC using network time protocol.
Another application of the atomic clock is the Global Positioning System (GPS).
GPS is a global navigation satellite system based on triangulation and transit time
measurements. The satellites carry atomic clocks for precise measurements of the
transit times.
The frequency of atomic clocks can be altered slightly by gravity (due to general
relativity), magnetic fields (due to the Zeeman effect), electric fields (due to the Stark
effect), and other phenomena. This sensitivity enables atomic clocks as ultra-precise
measurement tools used, for example, in precision tests of special and general
relativity, probing the merger of quantum mechanics and relativity, or tools in
metrology [3, 4].

Exercise 20.4 Due to the accuracy of atomic clocks, it is desirable to express

other measurement units (e.g., the meter) in terms of time (the second).
Investigate the Système International d’unités (SI, informally known as the
metric system), and establish which other units depend on the definition of the
second.

References

1. E.F. Arias and G. Petit, The hyperfine transition for the definition of the second, Ann. Phys.
(Berlin) 531 (2019) 1900068.
2. E.O. Gobel and U. Siegner, The new international system of units (SI): Quantum metrology and
quantum standards (Wiley-VCH, 2019).
3. T. Bothwell et al., Resolving the gravitational redshift across a millimetre-scale atomic sample,
Nature 602 (2022) 420.
4. B. Stray et al., Quantum sensing for gravity cartography, Nature 602 (2022) 590.
Chapter 21
Atom Cooling and Trapping

In recent decades, it has become possible to trap individual atoms and ions. These
methods enable the interaction of light with a cloud of atoms or even with individual
atoms. There are three general methods that are widely used to trap atoms or ions:
radiation pressure, magnetic dipole forces, and electric dipole forces. In this chapter,
the major trapping and cooling techniques are reviewed with an emphasis on the
basic physical principles.

21.1 Paul Trap

Ions can be trapped in a “Paul trap”, invented in the 1950s by Wolfgang Paul (Nobel
Prize in Physics in 1945) [1]. According to Gauss’s law, it is impossible to trap a
single charge along all three directions in free space by static electric fields alone
(Earnshaw’s theorem), since there can be no net inward force (∇ ∙ E ¼ 0) to constrain
the motion of the ions. There will be at least one direction where ions can escape. We
can, however, use time varying electromagnetic fields to trap charge.
Before atomic trapping, the ions are first prepared as an atomic vapor, produced
simply by evaporating the material in a vacuum chamber. These atoms can be
ionized by an electron beam, laser beam, or high electric field to strip an electron
off the atom. Beþ, Mgþ, Caþ, Srþ, Baþ, Znþ, Cdþ, Hgþ and Ybþ are commonly
used ions, which have single valence electrons after ionization.
Figures 21.1 and 21.2 illustrate a Paul trap. An alternating potential is applied to
electrodes, resulting in a saddle-shaped potential that rotates at the AC frequency
(typically radio frequencies). The alternating forces create a trap for ions. Coulomb
repulsion of the ions distributes them in a linear chain (1D crystal) along the trap
(perpendicular to the page in Fig. 21.2). In 2016, more than 200 Be ions were
trapped [2]. In 2018, a quantum register of 20 trapped ions were entangled
[3]. Microfabricated 2D ion trap arrays are under development where RF and DC

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 211
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7_21
212 21 Atom Cooling and Trapping

Fig. 21.1 Schematic of a Paul trap. The ions are shown as black dots. The potential on the four
large electrodes alternates to create a “rotating saddle potential” that traps the ions. A laser is used to
excite Rabi oscillations. Fluorescence from the ions can be read by a CCD camera or photodetector
to determine the state of the ions

Fig. 21.2 A Paul trap illustrating the alternating potential at two different times (separated by a
half-period of the AC potential), and the resulting electric field lines (E) and forces (F) on a
positively charged ion. The resulting saddle potential rotates, resulting in an ion trap

fields can be applied to planar electrodes to move the ions around on the chip (ion
shuttling) like electrons in a CCD camera [4].

21.2 Laser Cooling

Although fields are used to trap the ions, an alternative method is needed to reduce
their energy (i.e., to cool them), so that the thermal energy does not cause unwanted
atomic transitions. For this purpose, the method of laser or Doppler cooling is used,
which works for both ions and neutral atoms with two electronic energy levels. As its
name suggests, Doppler cooling involves the Doppler effect.
A laser of frequency ωl is red-detuned from an electronic energy transition of the
atom, meaning its frequency is slightly lower than the transition frequency ωa of the
21.2 Laser Cooling 213

(a) speed, (b)

=0
Detuning
ħ (Doppler shi) ħ

(c) (d)
ħ

Fig. 21.3 Principle behind Doppler cooling. (a) Atomic motion toward the laser beam results in a
Doppler shift of the photon frequency and resonant absorption. (b) Other atomic motion, such as a
stationary atom, results in no photon absorption. (c) Each spontaneous emission occurs in a random
direction, with momentum averaging to zero. (d) Three pairs of orthogonal counterpropagating
bbb
laser beams will cool the atom along all three directions (x,y,z)

atom. An atom approaching the laser source with speed v will observe in its frame of
reference a higher photon frequency due to the Doppler effect:


v
ωa ¼ ωl 1 þ ð21:1Þ
c

where c is the speed of light; that is, the Doppler shift is ωl vc ¼ kv. The Doppler-
shifted frequency becomes resonant with the electronic transition of the approaching
atom, meaning the atom will absorb the photon (Fig. 21.3a). The momentum of the
photon ( p ¼ h/λ) is transferred to the atom, giving it a “kick” in a direction that is
opposite to the atomic motion. Atomic motion away from the laser beam, or a
stationary atom, will not absorb the photon because ωa 6¼ ωl (Fig. 21.3b). After
photon absorption, the atom will move back into its ground state by spontaneous
emission of the absorbed photon. The spontaneous emission occurs in a random
direction (Fig. 21.3c). Hence, after many repeated absorption and emission events,
the emission recoil momentum averages to zero, but the absorption recoil momen-
tum does not. The emitted photon has energy ħωa, while the absorbed photon has
energy ħωl < ħωa. As a result, the atom momentum and kinetic energy are reduced,
cooling the atom to lower temperatures (the velocity distribution gets compressed).
By using three pairs of laser beams oriented along orthogonal directions (Fig. 21.3d),
the laser cooling can be applied along all three dimensions. This method of laser
cooling is also called “optical molasses”, since the lasers act as a viscous force that
slows the atoms (it can be shown that F /  v, like a classical damping force,
because the Doppler shift is proportional to velocity).
214 21 Atom Cooling and Trapping

Fig. 21.4 Left to right: Steven Chu [5], Claude Cohen-Tannoudji [6], and William Phillips [7];
Nobel Prize in Physics in 1997. (Credit: Wikimedia Commons [5–7])

Note that Doppler cooling does not provide any restoring force; that is, it does not
trap atoms but only cools them. Other methods are needed (e.g., a Paul trap) in
combination with laser cooling to both trap and cool atoms. The limiting temperature
(called the Doppler limit, TD) is determined by the random walk of the atom caused
by spontaneous emission. The Doppler limit is given by TD ¼ ħγ/2kB where γ is the
rate of spontaneous emission (γ 1 is the natural lifetime of the excited state). The
factor of 1/2 arises from the equal occupancy of the ground and excited states at high
light intensities where stimulated emission balances absorption. For typical values of
the natural linewidth, the temperature TD is on the order of 100 μK, which is well
below the temperature that can be achieved by cryogenic cooling methods. Steven
Chu, Claude Cohen-Tannoudji, and William Phillips (Fig. 21.4) were awarded the
1997 Nobel Prize in Physics for their work on laser cooling and atom trapping.

21.3 Magneto-optical Trap

A Paul trap can trap ions, but this approach will not work for neutral atoms.
Alternatively, we can use magneto-optical forces to trap and cool neutral atoms.
Classically, the potential energy of a magnetic moment in a magnetic field is

U ¼ μ ∙ B ð21:2Þ

This results in a shift of the energy levels of the atom, known as the Zeeman effect,
due to the interaction of the magnetic dipole moment of the atom with the applied
magnetic field. Quantum mechanically, Eq. (21.2) is modified according to

U ¼ mgμB Bz ð21:3Þ
21.3 Magneto-optical Trap 215

where m is the quantum number for the z-component of the total (orbital þ spin)
angular momentum, g is the Landé g-factor (a correction due to quantum electrody-
namics; see Chap. 4), μB is the Bohr magneton, and Bz is the magnetic field along
z. The magneto-optical trap (MOT), discussed below, uses the combined effects of
an inhomogeneous magnetic field and an optical field to provide both cooling and
trapping.

Exercise 21.1 Calculate the potential well depth, U ¼ μΔB, for typical atomic
magnetic moment values of the order of the Bohr magneton, μ~μB, and typical
values of the variation in laboratory magnetic field, 0.01 T. Estimate the
temperature at which atoms could be feasibly trapped.

In the MOT, an anti-Helmholtz coil is used to create an inhomogeneous magnetic

field (a weak quadrupole magnetic field; Fig. 21.5a). The anti-Helmholtz coil
consists of two solenoids of opposite current direction with axis aligned along the
z-direction. These create a field-free region (B ¼ 0 at the trap center, between the two
coils) surrounded by regions of increasing magnetic field in all directions away from
the trap center. An atom moving away from the zero-field region in the center of the
trap toward, for example, the þz or z direction between the two solenoids, will

(a) (b)

(c)

Fig. 21.5 (a) Anti-Helmholtz coil, showing a cross-section of the magnetic field lines. (b) Energy
levels (ground state with m ¼ 0, and Zeeman split excited state with m ¼  1, 0, þ 1); light
polarization (σ , σ +) of the laser beams (blue arrows); energy transition for absorption of σ  photon
(green arrow); and restoring force on an atom (red arrow). (c) Three pairs of laser beams provide 3D
trapping
216 21 Atom Cooling and Trapping

experience an increasing magnitude of the magnetic field (B ¼ B0 zbz ) and a

corresponding Zeeman shift of its energy levels.
Consider an atom with m ¼ 0 for the ground state and three sublevels (m ¼  1,
0, þ 1) for the excited state. Thus, according to Eq. (21.3), there is no Zeeman
splitting of the ground state, while the excited state is Zeeman split into three
sublevels in the presence of a magnetic field. This energy splitting increases as the
field increases, according to Eq. (21.3). The resulting energy levels are depicted in
Fig. 21.5b.
A photon has spin angular momentum of ħ, and projection of this angular
momentum along the z axis (the propagation direction) gives Sz ¼ mħ where
m ¼  1, 0, or þ1, which is called left circularly polarized light (denoted σ ),
linearly polarized light (denoted Π), and right circularly polarized light (denoted σ þ),
respectively. The absorption of light by an atom can only change m of the atom by
1, 0, or þ1 due to conservation of angular momentum. σ þ photons have m ¼ 1 and
can only increase m of an atom by one unit upon absorption of a photon. Conversely,
σ  photons have m ¼  1 and can only decrease m of the atom by one unit upon
absorption of a photon. Finally, linearly polarized photons have m ¼ 0 (they can be
considered as a superposition of σ þ and σ  photons) and do not change m of the
atom upon absorption.
In the MOT, two counterpropagating laser beams, one with σ þ polarization and
the other with σ  polarization, are established along the z-axis for Doppler cooling of
an atomic gas (Fig. 21.5b). σ þ light propagates along the þz direction, while σ  light
propagates along the z direction. An atom moving along þz will have the Doppler-
shifted σ  light resonant with the m ¼ 0 to m ¼  1 transition of the atom
(Fig. 21.5b), while σ þ light is far out of resonance. Conversely, an atom moving
along z will be laser cooled by absorption of the σ þ photon (m ¼ 0 to m ¼ 1
transition), but with negligible effect from the σ  photon. Thus, the recoil of the
atom from photon absorption is toward the field-free (B ¼ 0) region located at z ¼ 0,
which provides a restoring force on the atom toward the trap center. The same
principles apply along the bx and by directions. By directing three counterpropagating
laser beams along the three orthogonal directions, a centralized cloud of about 109
cold atoms can be trapped and cooled near the field-free region at the trap center
(Fig. 21.5c).
The MOT is the most used cooling technique, allowing cooling of neutral atoms
down to temperatures of ~10  100 μK. Among the new phenomenon enabled by a
MOT is the Bose-Einstein condensate (BEC). The BEC is a group of bosonic atoms
cooled to near absolute zero. A gas of atoms, such as Rb, is first laser cooled in a
MOT followed by evaporative cooling. Evaporative cooling relies on the escape of
high-energy atoms by lowering the trap potential, like cooling of a cup of coffee by
blowing on it. Evaporative cooling can achieve nanoKelvin temperatures. When the
atoms are cooled and trapped to a sufficient density (n), their de Broglie wavelengths
(λ) become long enough to begin overlapping. When the condition nλ3~1 is met, the
atoms undergo a phase transition and enter a quantum degenerate state, which is
called a BEC as predicted by Einstein in 1924. All atoms of the BEC can be
described by a single wavefunction and act essentially as one single atom. Under
21.4 Sisyphus Cooling 217

these conditions, one has a coherent source of atoms enabling atomic interferometry,
atomic lasers, and other coherent phenomenon, providing new tools for ultraprecise
metrology. The first BEC was realized in 1995 [8] using Rb atoms and later Na
atoms [9], which earned Eric A. Cornell, Wolfgang Ketterle, and Carl E. Wieman the
Nobel Prize in Physics in 2001.

21.4 Sisyphus Cooling

Sisyphus cooling, also known as polarization gradient cooling, is a method of

obtaining sub-Doppler cooling. Sisyphus cooling involves a combination of
multilevel atoms, polarization gradients, light shifts, and optical pumping.
Sub-Doppler cooling was discovered when temperature below the Doppler limit
was realized in MOTs with atoms that have multiple levels in the ground state.
The method uses two counterpropagating laser beams with orthogonal linear
polarizations, which may be described by

E ¼ E 0bx cos ðωt  kzÞ þ E0by cos ðωt þ kzÞ

ð21:4Þ
¼ E0 ðbx þ byÞ cos ωt cos kz þ E 0 ðbx  byÞ sin ωt sin kz

Equation (21.4) describes a light field with a polarization gradient along bz . For
example, at z ¼ 0, Eq. (21.4) gives E ¼ E0 ðbx þ byÞ cos ðωt Þ, which describes linearly
polarized light (denoted Π) at an angle of π/4 to the x-axis. Similarly, at z ¼ λ/4, the
light is linearly polarized but at an angle of π/4 to the x-axis. At z ¼ λ/8, the field
becomes E ¼ pE0ffiffi2 ðbx þ byÞ cos ωt þ pE0ffiffi2 ðbx  byÞ sin ωt . In the latter case, the sine and
cosine terms give temporal dependences that are π2 out of phase, which describes left
circularly polarized light (denoted σ ). Similarly, at z ¼ 3λ/8, the light becomes right
circularly polarized (denoted σ þ). Between the linear and circularly polarized light,
the light is elliptically polarized. Thus, the light polarization oscillates between σ 
and σ þ and back again over a distance of λ/2.
Consider an atom with possible states given by jj, mi where j is the total (spin þ
orbital) angular momentum and m is the total angular momentum quantized along
the z-axis. For a given j, the allowed values of m are m ¼ 0,  1,  2, . . .  j,
according to the rules of angular momentum in quantum mechanics. We suppose the
atom has a ground state described by j ¼ 12 with two sublevels given by m ¼  12, and
an excited state given by j ¼ 32 with 4 sublevels given by m ¼  32 ,  12 , 12 , 32 :
Due to the conservation
  of angular momentum,3 3 the allowed transition with σ þ
light is from the 12 , 12 ground
 state to the  ,
2 2 excited state, or from the 1 ,  1
2 2
ground state to the 3 , 1 excited state. The allowed transition with σ  light is from
1 1 3  2 2 1  3 
 ,   
2 to 2 ,  2 , or from 2,  2 to 2 ,  
1 1 3
2 2 . Finally, with
 linearly
 (Π)
 polarized

light, only transitions from 12 , 12 to 32 , 12 , or from 12 ,  12 to 32 ,  12 are
allowed. The allowed transitions are illustrated in Fig. 21.6 along with the relative
218 21 Atom Cooling and Trapping

Fig. 21.6 Allowed

3 3 3 1 3 1 3 3
transitions from ground to ,−
2 2
,−
2 2
,
2 2
,
2 2
excited state, showing the
photon polarizations
involved in the transitions.
2 1 1 2
The numbers are the
Clebsch–Gordan
1 3 3 3 3
1
coefficients, whose square
gives the transition
probabilities − + − +

transition probabilities given by the square of the Clebsch–Gordan coefficients,

according to the quantum mechanical theory of angular momentum.
The light shift (AC Stark effect) considered in Chap. 19 (Fig. 19.11) is dependent
on the dipole coupling (g) between the ground and excited states. This coupling
depends on the transition probabilities between the states (i.e., g is proportional to
the square of the Clebsch–Gordan coefficients). Thus, the transition probabilities
depend on the light polarization, as shown in Fig. 21.6. For example, examination of
the Clebsch–Gordan coefficients in Fig. 21.6 shows that the transition from the
m ¼ 12 state is three times more probable than from the m ¼  12 state for σ þ
polarization (i.e., the m ¼ 12 state is more strongly coupled to the light field). Thus,
the light shift for σ þ is three times greater when starting from m ¼ 12 compared to
m ¼  12 . Conversely, for σ  polarization, the transition from the m ¼  12 state is
three times more probable than from the m ¼ 12 state, and the corresponding light
shift is three times greater. Finally, the transition probability and the light shift for
linearly polarized light is equal from m ¼ 12 and m ¼  12, and equal to 23 that of the
maximum for circularly polarized light. In summary, the light shift of the ground
state sublevels (m ¼  12Þ oscillates in space with the same periodicity as the
polarization, as shown in Fig. 21.7.
An atom starting in the m ¼ 12 ground state sublevel and moving along z will lose
kinetic energy (and gain potential energy) as it climbs the energy hill. Eventually,
 the
atom encounters σ polarization, allowing a transition from the  1
, 1
ground state
   2 2
to the 32 ,  12 excited state (with Δm ¼  1). The atom will then spontaneously
emit a photon and decay to the lower m ¼  12 ground state sublevel, losing energy in
the process. This transition process is known as “optical pumping”. The potential
energy gained by the atom in climbing the hill is radiated away by the spontaneous
emission, because the frequency of emission is greater than the frequency of
absorption (see Fig. 21.7). The amount of energy lost is equal to the difference in
light shift. As the atom continues moving along z, now in the m ¼  12 ground state
1 σ þ1polarization, where it
sublevel, it again climbs the energy hill and next encounters
can again undergo optical pumping but now from the  ,  ground state to the
3 1 2 2 
 , 1 1
2 2 excited state (Δm ¼ þ 1), and then back to the lower 2 , 2 ground state.
21.5 Dipole Trap, Optical Tweezers, and Optical Lattice 219

Energy

Polarizaon

Fig. 21.7 Principle of Sisyphus cooling. Ground state energy levels with m ¼ 12 (red) and m ¼  12
(blue) are affected by the polarization-dependent light shift (the light shift is shown relative to zero
energy). An atom starting at z ¼ 0 in the m ¼ 12 ground state sublevel and moving along the
z direction climbs uphill in energy (black dashed arrow). Optical pumping (green arrows) occurs
upon absorption of a σ  polarized photon. The process repeats in the m ¼  12 ground state sublevel
with absorption of a σ + polarized photon

The process repeats along the z axis, resulting in a continuous loss of energy and
sub-Doppler cooling to temperature on the order of μK. The name “Sisyphus
cooling” derives from an analogy of the above process to a figure in Greek mythol-
ogy who was doomed to forever roll a stone up a hill only to have it roll down again.

21.5 Dipole Trap, Optical Tweezers, and Optical Lattice

It is also possible to create a trapping potential using an induced electric dipole

moment in an inhomogeneous electric field. Classically, the potential energy of an
electric dipole moment in an electric field is given by

U ¼ p ∙ E ð21:5Þ

In an inhomogeneous field along z, the force is

∂U ∂ðpz Ez Þ
Fz ¼  ¼ ð21:6Þ
∂z ∂z

The induced dipole moment is proportional to electric field, according to pz ¼ αEz

where α is the atomic polarizability. Thus, the force is proportional to the gradient of
the light intensity, I (since I / E2). This produces a force toward regions of higher
220 21 Atom Cooling and Trapping

Fig. 21.8 (a) An atom (a) (b)

trapped in the focus of a
laser beam. (b) Atoms
trapped in a 2D optical
lattice. (Credit: Wikimedia
Commons [12])

light intensity (e.g., toward the waist of a tightly focused laser beam; Fig. 21.8a).
This approach is called dipole trapping or “optical tweezers”, developed by Arthur
Ashkin who was awarded the 2018 Nobel Prize in Physics. Optical tweezers are now
an important tool to hold and manipulate microscopic objects such as biological
molecules or living cells. In addition to trapping microscopic objects, the dipole
interaction with a light field can also hold and manipulate atoms. The first optical
trap was able to confine about 500 Na atoms [10].
In 1968, V.S. Letokhov proposed using the dipole force that arises from the light
shift in a standing wave to confine atoms in microscopic dimensions [11]. Using a
pattern of crossed laser beams, it is possible to set up a standing wave interference
pattern with maxima and minima in the light intensity with a period on the order of
the laser wavelength. The AC Stark effect or light shift varies with the light intensity,
resulting in a gradient force on an atom given by F z ¼  ∂ð∂z ΔE Þ
where ΔE is the
spatially varying light shift. Atoms immersed in a laser field, which is red-detuned
below atomic resonance (like in Fig. 19.11), have their ground state energy shifted
down with increasing light intensity (negative light shift), and atoms are therefore
attracted toward locations with maximum light intensity (toward the antinodes of the
standing wave). The opposite occurs if the optical field is blue-detuned where atoms
are attracted to the locations of minimum light intensity (nodes of the standing light
wave). Using orthogonal laser beams, a 2D trap potential called an “optical lattice”
can be created that enables dipole trapping of individual atoms at the nodes or
antinodes of the light field in a 2D array like an egg carton (Fig. 21.8b). The trap
potentials of an optical lattice are rather weak (a few mK), so it is necessary to first
precool the atoms (e.g., to μK temperature in a MOT) and then superimpose the
optical lattice to rearrange the atoms into the array.
The optical lattice is analogous to a conventional atomic crystal, but with a
periodic potential of a few microns, about 103 times larger than the period of an
atomic crystal. The de Broglie wavelength of the cooled atoms is on the order of the
size of the trap, so the atomic motion in the trap must be described quantum
mechanically. Atoms can tunnel between the potential wells of the optical lattice
just like electrons in a crystal, and the motion of atoms through the periodic potential
of an optical lattice can be described by a band structure derived from the Kronig–
Penney model familiar from introductory quantum mechanics or solid-state physics
[13]. Unlike an atomic crystal, however, we are free to vary the depth, period, or
geometry of the potential wells by adjusting the light intensity, wavelength, or angles
between the light beams. Thus, the array of atoms can be used to perform quantum
21.5 Dipole Trap, Optical Tweezers, and Optical Lattice 221

simulations; that is, using one quantum system (the optical lattice) to simulate
another one (like condensed matter systems such as superconductivity). One could
also use optical lattices to control interactions between neighboring atoms and
perform “microchemistry”. Using light fields, it is possible to manipulate and steer
atoms in a manner similar to mirrors, lenses, diffraction gratings, and other optical
elements, creating a new domain of “atom optics”. Such atom manipulation opens
the possibility for atom-by-atom nanofabrication. It is even possible to trap and
manipulate atoms into complex three-dimensional arrangements (Fig. 21.9) [14].

Fig. 21.9 Fluorescence from Rb-87 atoms arranged into 3D arrays. (Reprinted by permission from
Springer Nature, Barredo et al. [14]. Copyright 2018)
222 21 Atom Cooling and Trapping

Such arrangements could be used for quantum computing. A review of these and
other aspects of atom trapping and cooling is provided in the classic reference by
Metcalf and Straten [15].

References

1. W. Paul, Electromagnetic traps for charged and neutral particles, Rev. Mod. Phys.
62 (1990) 531.
2. J.G. Bohnet et al., Quantum spin dynamics and entanglement generation with hundreds of
trapped ions, Science 352 (2016) 1297.
3. N. Friis et al., Observation of entangled states of a fully controlled 20-qubit system, Phys. Rev.
X 8 (2018) 021012.
4. C.D. Bruzewicz et al., Trapped-ion quantum computing: Progress and challenges, Appl. Phys.
Rev. 6 (2019) 021314.
5. Attribution: The Royal Society. This file is licensed under the Creative Commons Attribution-
Share Alike 3.0 Unported license (https://creativecommons.org/licenses/by-sa/3.0/
deed.en). File: Professor Steven Chu ForMemRS headshot.jpg. (2020, October 30). Wikimedia
Commons, the free media repository. Retrieved 18:38, December 8, 2020 from https://
commons.wikimedia.org/w/index.php?title¼File:Professor_Steven_Chu_ForMemRS_
headshot.jpg&oldid¼507616885
6. Author: Amir Bernat. This file is licensed under the Creative Commons Attribution-Share Alike
4.0 International, 3.0 Unported, 2.5 Generic, 2.0 Generic and 1.0 Generic license (https://
creativecommons.org/licenses/by-sa/4.0/). File:Claude Cohen-Tannoudji.JPG. (2020, August
28). Wikimedia Commons, the free media repository. Retrieved 18:06, December 8, 2020 from
https://commons.wikimedia.org/w/index.php?title¼File:Claude_Cohen-Tannoudji.JPG&
oldid¼444509221
7. Author: Markus Pössel. This file is licensed under the Creative Commons Attribution-Share
Alike 3.0 Unported license (https://creativecommons.org/licenses/by-sa/3.0/deed.en). File: Wil-
liam D. Phillips.jpg. (2020, October 26). Wikimedia Commons, the free media repository.
Retrieved 18:08, December 8, 2020 from https://commons.wikimedia.org/w/index.php?
title¼File:William_D._Phillips.jpg&oldid¼502669574.
8. M.H. Anderson, J.R. Ensher, M.R. Matthews, C.E. Wieman and E.A. Cornell, Observation of
Bose-Einstein condensation in a dilute atomic vapor, Science 269 (1995) 198.
9. K.B. Davis, M.-O. Mewes, M.R. Andrews, M.J. Van Druten, D.S. Durfee, D.M. Kurn and
W. Ketterle, Bose-Einstein condensation in a gas of sodium atoms, Phys. Rev. Lett.
75 (1995) 3969.
10. S. Chu, J.E. Bjorkholm, A. Ashkin and A. Cable, Experimental observation of optically trapped
atoms, Phys. Rev. Lett. 57 (1986) 314.
11. V.S. Lethokov, Narrowing of the Doppler width in a standing light wave, JETP Lett.
7 (1968) 272.
12. https://commons.wikimedia.org/wiki/File:Lattice_mott.JPG
13. Y. Castin and J. Dalibard, Quantization of atomic motion in optical molasses, Europhys. Lett.
14 (1991) 761.
14. D. Barredo et al., Synthetic three-dimensional atomic structures assembled atom by atom,
Nature 561 (2018) 79.
15. H.J. Metcalf and P. van der Straten, Laser cooling and trapping (1999, Springer).
Further Reading

H.-A. Bachor and T.C. Ralph, A guide to experiments in quantum optics (2019, Wiley-VCH, 3rd
ed.).
C.C. Gerry and P.L. Knight, Introductory quantum optics (2005, Cambridge University Press).
G. Grynberg, A. Aspect and C. Fabre, Introduction to quantum optics (2010, Cambridge University
Press).
R.R. LaPierre, Introduction to quantum computing (2021, Springer).
P. Meystre, Quantum optics (2021, Springer).
H.J. Metcalf and P. van der Straten, Laser cooling and trapping (1999, Springer).

© The Editor(s) (if applicable) and The Author(s), under exclusive license to 223
Springer Nature Switzerland AG 2022
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7
Appendix 1: Derivation of Lamb Shift

A heuristic derivation of the Lamb shift was put forward by Theodore A. Welton in
1948 [A1.1]. We assume our atom sits inside a box containing the electric field of the
vacuum modes, like Fig. 3.4. These electric fields of the vacuum modes shake the
electron. The average energy density (energy per unit volume) inside the box
associated with each mode k of the electric field (Ek) is

rffiffiffiffiffiffiffiffiffiffi2
1 1
2 1 ħω 1 ħω
hU k i ¼ Eo hðE k Þ2 i ¼ Eo ε1 ¼ Eo ¼ ðA1:1Þ
2 2 2 2Eo V 4 V

where ε1 is the one-photon field amplitude of the mode k with frequency ω ¼ ck. The
energy associated with the electric field is equal to the energy density multiplied by
the volume of the box:

1 1
hU k iV ¼ hU k iL3 ¼ ħω ¼ ħck ðA1:2Þ
4 4

Equation (A1.2) simply tells us that the ground state energy of each vacuum mode,
2 ħω,is divided equally between the electric and magnetic fields. Thus,
1

ħck
hð E k Þ 2 i ¼ ðA1:3Þ
2Eo V

The acceleration of the electron due to each electric field of mode k is

F eE k
ak ¼ ¼ ðA1:4Þ
m m

The oscillatory motion of the electron is given by

© The Editor(s) (if applicable) and The Author(s), under exclusive license to 225
Springer Nature Switzerland AG 2022
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7
226 Appendix 1: Derivation of Lamb Shift

δr k ðt Þ ¼ δr k cos ðωt Þ ðA1:5Þ

The acceleration, ak, is given by

d2 ðδr k ðt ÞÞ
ak ð t Þ ¼ ¼ ω2 δr k ðt Þ ¼ k 2 c2 δr k ðt Þ ðA1:6Þ
dt 2

Thus, from Eqs. (A1.4) and (A1.6), the displacement amplitude is

e 1
δr k ¼ E ðA1:7Þ
mc2 k k2

The mean square displacement from Eqs. (A1.7) and (A1.3) is

 2    
e 1 e 2 ħc 1
hðδrk Þ2 i ¼ hðE k Þ2 i ¼ ðA1:8Þ
mc2 k4 mc2 2Eo V k3

The total mean square displacement h(δr)2i is a summation of h(δrk)2i over all modes
k:
X
hðδr Þ2 i ¼ hðδr k Þ2 i ðA1:9Þ
k
 2  
e ħc X 1
¼ ðA1:10Þ
mc2 2Eo V k k 3

Note that Eq. (A1.9) is valid because random variables add in quadrature. This is the
familiar rule for root-mean-square (rms) addition.
We assume the fictitious box is very large, so the modes are closely spaced in
wavevector (k is continuous) and we can replace the summation with an integral:
 2  Z 1
ρð k Þ 3
e ħc
hðδr Þ2 i ¼ d k ðA1:11Þ
mc2 2Eo V 0 k3

ρ(k) is the density of modes, that is, the number of modes between k and k þ dk in
k-space.
To determine ρ(k), note that the wavevectors for waves in the box are quantized
by periodic boundary conditions according to

2π 2π 2π
kx ¼ ¼ ¼ nx ðA1:12Þ
λ L=nx L

Similarly,
Appendix 1: Derivation of Lamb Shift 227

2π
n ky ¼ ðA1:13Þ
L y
2π
kz ¼ nz ðA1:14Þ
L

3
Each mode takes up a volume in k-space of 2π
L . The number of modes in the range
k to k þ dk is


2 4πk2 dk L3 k2 dk
ρðkÞdk ¼
2π 3 ¼ ðA1:15Þ
π2
L

where the first factor of 2 arises from the two allowed polarizations.
Thus,
 2  Z 1
2 e ħc 1
hðδr Þ i ¼ dk ðA1:16Þ
πmc2 2Eo 0 k

We see that the volume V ¼ L3 of the box cancels out as it should since the box was
fictitious.
The integral in Eq. (A1.16) is divergent, so we need to choose suitable limits in
the integration. The maximum wavelength is associated with the size of the atom,
2
which is twice the Bohr radius, 2a0 ¼ 2E 0h
πme2 . This maximum wavelength gives the
minimum wavevector:

2π π 2 me2
k min ¼ ¼ ðA1:17Þ
2a0 E 0 h2

The minimum wavelength is taken to be the Compton wavelength, λc ¼ mc h

. The
Compton wavelength gives a fundamental limitation on measuring the position of a
particle. This minimum wavelength gives the maximum wavevector:

2π 2πmc
k max ¼ ¼ ðA1:18Þ
λc h

Thus, the integral becomes

Z k max      
1 k max 4E0 ħc 1
dk ¼ ln ¼ ln ¼ ln ðA1:19Þ
k min k k min e 2 πα

e2
where α ¼ 4πE
1
0 ħc  137
1
is the fine structure constant. Thus,
228 Appendix 1: Derivation of Lamb Shift

 2    
2 e ħc 1
hðδr Þ i ¼ ln ðA1:20Þ
πmc2 2Eo πα

Because of the electron displacement, the electron experiences a change in

potential according to

1
ΔV ¼ V ðr þ δr Þ  V ðr Þ ¼ δr∇V þ ðδr Þ2 ∇2 V þ . . . ðA1:21Þ
2

where we have used a Taylor expansion. According to Eq. (A1.5), the average of δr
is zero, leaving:

1
ΔV ¼ ðδr Þ2 ∇2 V ðA1:22Þ
2

Here V is the Coulomb potential of the hydrogen atom:

1 e2
V ¼ ðA1:23Þ
4πE0 r

The Laplacian, ∇2V, for the Coulomb potential is

e2
∇2 V ¼ δ ðr Þ ðA1:24Þ
E0

where δ(r) is the Dirac delta function. The average becomes

∇2 V ¼ ψj∇2 Vjψ ðA1:25Þ

where ψ is the hydrogen wavefunction. Using Eq. (A1.24), we get

2
e2
∇ V ¼ j ψ ð 0Þ j 2 ðA1:26Þ
E0

Equation (A1.26) is zero at the origin for the p orbital and nonzero only for the
s orbital. For the 2s orbital, the wavefunction at the origin is

1
ψ2s ð0Þ ¼ ðA1:27Þ
ð8πa0 3 Þ1=2

Thus,
Appendix 1: Derivation of Lamb Shift 229

1
hΔV i ¼ hðδr Þ2 i h∇2 Vi ðA1:28Þ
2
 
1 5 2 1
¼ α mc ln ðA1:29Þ
2π πα

This shift is about 1540 MHz, close to the observed shift of 1058 MHz. A more exact
calculation using the full theory of quantum electrodynamics provides a measure-
ment of the fine-structure constant α to better than one part in a million.

References

A1.1. T.A. Welton, Some observable effects of the quantum-mechanical fluctuations of the elec-
tromagnetic field, Phys. Rev. 74 (1948) 1157.
Appendix 2: Derivation of Casimir Formula

Here, we present a simple heuristic argument for the 1/d4 dependence of the Casimir
force. The zero-point energy between the plates is

1 X
E ¼ ħ ωn ðA2:1Þ
2 n

1 X
¼ ħc k n ðA2:2Þ
2 n

where c is the speed of light.

The wavevectors perpendicular to the plates, describing standing waves, are
quantized according to

nπ
kn ¼ ðA2:3Þ
d

where n is an integer and d is the plate separation.

The density of states is ρ(k)dk / (Ad)k2dk where A is the plate area, and Ad is the
mode volume between the plates. Thus, we can replace the summation in Eq. (A2.2)
by an integration, using the density of states:
Z k max
E / Adħc k 3 dk ðA2:4Þ
k min

According to Eq. (A2.3), the minimum wavevector is k min ¼ πd . The maximum

wavevector corresponds to the maximum allowed frequency, which is taken to be
the plasma frequency of the metal where the plates become transparent. Thus,
evaluating the integral gives

© The Editor(s) (if applicable) and The Author(s), under exclusive license to 231
Springer Nature Switzerland AG 2022
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7
232 Appendix 2: Derivation of Casimir Formula

k max
E / Adħck 4 k min ðA2:5Þ
  4 
π
E / Adħc k max 
4
ðA2:6Þ
d
1
E / Aħc ðA2:7Þ
d3

The force per unit area F on the plates is

1 ∂E ħc
F¼ / 4 ðA2:8Þ
A ∂d d

Equation (A2.8) gives the correct 1/d4 dependence of the Casimir force. The
derivation is not quite correct because only modes with wavevectors perpendicular
to the plates are quantized, and one must use periodic boundary conditions for modes
with wavevectors parallel to the plates. A more detailed calculation [A2.1, A2.2]
gives the correct numerical factor:

π 2 ħc 1
F¼ ðA2:9Þ
240 d4

References

A2.1. S. Scheel and S.Y. Buhmann, Macroscopic quantum electrodynamics — concepts and
applications, Acta Phys. Slov. 58 (2008) 700.
A2.2. S.K. Lamoreaux, The Casimir force: background, experiments, and applications, Rep. Prog.
Phys. 68 (2004) 201.
Appendix 3: Derivation of Normalization
Constant in Single Photon Wavepacket

We start with the Lorentzian distribution of mode probabilities:

K2
jcl j2 ¼ ðA3:1Þ
ðωl  ω0 Þ2 þ γ 2 =4

The normalization condition becomes

X
1 X
1
K2
jcl j2 ¼ ¼1 ðA3:2Þ
l¼0 l¼0 ðωl  ω0 Þ2 þ γ 2 =4

Let us replace the discrete summation with an integration:

X
1 Z 1
K2 K2
¼ ρðωl Þdωl ðA3:3Þ
l¼0 ðωl  ω0 Þ2 þ γ 2 =4 0 ðωl  ω0 Þ2 þ γ2 =4

where ρ(ωl) is the density of modes, and ρ(ωl)dωl is the number of modes in the
frequency interval from ωl to ωlþ dωl.
Suppose we have a wavepacket of length L. The allowed wavevectors of the
standing modes become discrete:

2π 2π π
kl ¼ ¼ ¼ n ðA3:4Þ
λ 2L=nl L l
πc
ω l ¼ k l c ¼ nl ðA3:5Þ
L

© The Editor(s) (if applicable) and The Author(s), under exclusive license to 233
Springer Nature Switzerland AG 2022
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7
234 Appendix 3: Derivation of Normalization Constant in Single Photon Wavepacket

Rearranging,

dnl L
ρðωl Þdω ¼ dω ¼ dω ðA3:6Þ
dωl πc

Thus,
Z 1 Z
K2 LK 2 1 1
ρ ðω Þdω ¼ dωl ðA3:7Þ
0
2
ðωl  ω0 Þ þ γ =4
2
l l
πc 2
0 ðωl  ω0 Þ þ γ 2 =4
Z
4LK 2 1 1
¼ dω ðA3:8Þ
πcγ 2 0 4ðωl  ω0 Þ2 =γ 2 þ 1 l

Let u ¼ 2(ωl  ω0)/γ, du ¼ (2/γ) dωl, giving

Z 1 Z 1
4LK 2 1 2LK 2 1
dωl ¼ du ðA3:9Þ
πcγ 2 0
2 2
4ðωl  ω0 Þ =γ þ 1 πcγ 0 u2 þ 1
R1 1
Using 0
1
u2 þ1 du ¼ tan 1 ðuÞ0 ¼ π=2 gives
Z 1
2LK 2 1 LK 2
du ¼ ðA3:10Þ
πcγ 0 u2 þ1 cγ

Now we apply the normalization condition:

LK 2
¼1 ðA3:11Þ
cγ

which gives
rffiffiffiffiffi
cγ
K¼ ðA3:12Þ
L
Appendix 4: Derivation of Planck’s
Distribution Law

Starting with Eq. (12.23):

 
Pn ¼ enħω=kT 1  eħω=kT ðA4:1Þ

The energy is given by

X
1
hE i ¼ En Pn ðA4:2Þ
n¼0

Ignoring the zero-point energy, we get

X
1  
hE i ¼ nħωenħω=kT 1  eħω=kT ðA4:3Þ
n¼0
 X
1
¼ ħω 1  eħω=kT n enħω=kT ðA4:4Þ
n¼0

Let a ¼ eħω=kT , giving

X
1
hE i ¼ ħω ð1  aÞ nan ðA4:5Þ
n¼0

P1
Next, we use the geometric series n¼1 na
n
¼ ð1a
a
Þ2
, giving

a
hE i ¼ ħω ðA4:6Þ
1a

© The Editor(s) (if applicable) and The Author(s), under exclusive license to 235
Springer Nature Switzerland AG 2022
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7
236 Appendix 4: Derivation of Planck’s Distribution Law

eħω=kT
¼ ħω ðA4:7Þ
1  eħω=kT
ħω
¼ ħω=kT ðA4:8Þ
e 1

Next, we need to find the density of modes; that is, how many modes exist with
frequency between ω and ω þ dω. Expressing Eq. (A1.15) in terms of frequency
(k ¼ ωc ), we get

L3 2
ρðωÞdω ¼ ω dω ðA4:9Þ
π 2 c3

The number of states per unit volume between frequency ω and ω þ dω is

1 1
ρðωÞdω ¼ 2 3 ω2 dω ðA4:10Þ
L3 π c

Thus, the energy density between ω and ω þ dω is

1 2 ħω
U ðωÞdω ¼ ω dω ðA4:11Þ
π 2 c3 eħω=kT  1
h ω3
U ðωÞdω ¼ dω ðA4:12Þ
2π 3 c3 eħω=kT  1

Equation (A4.12) gives the Planck radiation law as

h ω3
U ð ωÞ ¼ ðA4:13Þ
2π 3 c3 eħω=kT  1

U(ω) is called the spectral energy density with units of Jm3/s1, that is, energy per
unit volume per unit frequency. Thus, if we integrate Eq. (A4.13) with respect to
frequency, ω, we get the energy density (energy per unit volume in units of Jm3).
Similarly, the spectral energy density could be expressed in terms of wavelength:

8πhc 1
U ð λÞ ¼ ðA4:14Þ
λ5 ehc=λkT  1

Classically, we expect kT of energy per mode of the radiation, giving

1 ω2
U ð ωÞ ¼ ρðωÞkT ¼ 2 3 kT ðA4:15Þ
V π c

which is known as the Rayleigh–Jeans result. This classical result diverges with
increasing ω, leading to the so-called uv catastrophe. Plank’s result solves this
problem by quantizing the radiation, leading to perfect agreement with experiment.
Index

A Clauser, Horne, Shimony and Holt (CHSH),

Absolute photodetector calibration, 180 80–82
Amplitude squeezing, 147, 149 Coherent state, 93–108, 111–116, 121, 122,
Annihilation operator, 7, 15–17, 19, 39, 50, 60, 125, 127, 128, 131–138, 141–143, 145,
61, 63, 74, 86, 93, 134, 142 147–152, 155, 157, 158, 160, 161,
Anti-bunching, 122 165–169, 194
Anti-correlation, 114 Coincidence, 57, 64, 65, 70, 80, 111, 113, 114,
Aspect, A., 81, 82 122, 172, 174, 177–180
Atomic clock, 170, 205–210 Collapse and revival, 183, 194–197, 199
Commutation relation, 1, 5–6, 9, 13, 15, 17–19,
22, 33, 34, 37, 67, 103, 137, 143, 150
B Complex plane, 95, 100, 104, 107, 128, 146
Baker–Hausdorff formula, 143, 153 Correlation, 51, 57, 64–67, 80, 113–115, 122,
Bare states, 201 123, 141, 179
Basis states, 4, 94, 199, 200 Creation operator, 17–18, 22, 39, 50, 86, 142
Beam splitter (BS), 57–71, 75–77, 111–114,
125, 126, 129, 132, 133, 150, 151,
170–172, 174 D
Bell inequality, 82 de Broglie wavelength, 169, 170, 216, 220
Boltzmann, 118 Dipole, 48, 139, 185, 187, 189, 190, 192, 200,
Bunching, 122 207, 211, 214, 218–220
Dipole trap, 219–222
Dirac, P., 1, 3, 4, 11
C Displaced vacuum, 105
Canonical quantization, 1–8, 27, 32–35, 85 Doppler broadening, 209
Canonically conjugate variables, 1–3, 5, 6, Doppler cooling, 212–214, 216
8, 33 Dressed states, 200, 201
Casimir effect, 44–48, 88, 138 Dynamic variables, 1
Cauchy–Schwarz inequality, 66
Cavity, 31, 47, 48, 100, 174, 181, 183,
187–191, 193, 194, 197–201, 208, 209 E
Chaotic light, 115 Eigenenergies, 4, 10, 11
Circuit QED (cQED), 187–189, 194, 197, 199 Eigenstate, 4, 10, 11, 14–18, 38, 44, 88, 89, 93,
Classical harmonic oscillator, 7–8, 102 96, 101, 142, 184, 200

© The Editor(s) (if applicable) and The Author(s), under exclusive license to 237
Springer Nature Switzerland AG 2022
R. LaPierre, Getting Started in Quantum Optics, Undergraduate Texts in Physics,
https://doi.org/10.1007/978-3-031-12432-7
238 Index

Electric field, 27, 28, 30, 31, 33, 38, 41–45, 47, Heralded, 53, 75, 174
50, 57, 58, 67, 91, 93, 94, 98, 99, 101, Hermite polynomial, 10
102, 104, 108, 109, 116, 117, 125, 139, Hermitian, 11, 13, 38, 50, 60, 61, 93, 116, 127,
140, 143–147, 151, 160, 185, 197, 141, 142
210–212, 219, 225 Heterodyne, 125–129
Electromagnetic wave, 28, 31, 85, 206 Homodyne, 38, 105, 125–128, 131, 133–138,
Electromagnetically induced transparency 147, 157–161, 165, 167, 168
(EIT), 174 Hong-Ou-Mandel (HOM), 165, 171–174
Electron multiplying charge-coupled device
(EMCCD), 53, 54
Elitzur, A., 181 I
Entangled, 51, 53, 57, 69, 77–82, 87, 165, 167, Idler, 52, 53
170, 172, 177–180, 191, 200, 211 Incoherent, 115–122, 139
Entanglement, 51, 53, 69, 77–82, 170, 179 Intensified CCD (ICCD), 53, 54
Excited state, 15, 18–21, 44, 51, 89, 90, 183, Interaction Hamiltonian, 141, 183, 185, 186
184, 187, 191, 192, 194, 195, 197–200, Interaction-free measurement (IFM), 177,
214–218 181, 182
Expectation value, 5, 7, 21–23, 36, 41, 42, 62, Interference, 71–76, 129, 131, 132, 141, 171,
94, 101, 102, 144 172, 177, 178, 181, 195, 196, 220

F J
Fock state, 14, 41–50, 68, 87, 93–96, 98, 101, Jaynes–Cummings Hamiltonian, 183–186, 199
108, 109, 114, 121, 122, 139, 153, 165, Jaynes–Cummings ladder, 201, 202
187, 194, 196, 199
Fountain clock, 209
K
Kerr effect, 172–175
G
Gaussian, 23, 37, 97, 100, 102, 107, 145, 196
Generalized coordinates, 2–5, 7 L
Ghost imaging (GI), 177–179 Lamb shift, 44, 45, 138, 225–229
Glauber, R.J., 65, 66, 106 Laser cooling, 212–214
Gravitational waves, 157, 159, 162 Laser interferometer gravitational-wave
Ground state, 14, 15, 18–24, 44, 51, 89, 104, observatory (LIGO), 159–162, 166
183, 184, 186, 191, 195, 197, 198, 206, Lewis, G., 39
207, 213, 215–219 Light-atom coupling, 187, 193
Ground state energy, 11, 12, 15, 200, 219, Linear optical quantum computing (LOQC), 175
220, 225 Lithium niobate (LiNbO3), 52
Local oscillator (LO), 125, 127–129, 157

H
Hamilton equations, 1–3, 8, 32 M
Hamiltonian, 1–4, 7, 8, 10, 12, 14, 31–34, 38, Mach-Zehnder interferometer, 71, 72, 74
44, 85, 87, 89, 96, 101, 142, 183, 184, Magnetic fields, 27, 28, 31, 43, 44, 206, 207,
186, 191, 199, 200, BNF–32 210, 214–216, 225
Hamilton, W.R., 1 Magnetic moment, 44, 45, 206, 207, 214, 215
Hanbury Brown–Twiss, 70, 122 Magneto-optical trap (MOT), 214–217, 220
Heaviside function, 91 Maxwell’s equations, 27–29, 45, 85
Heisenberg, 6, 27, 35, 89, 90, 101, 116, 129, Mode, 28, 31, 32, 34, 45–48, 57, 62, 69, 85–88,
144, 186 90, 116, 117, 122, 141, 167–172, 174,
Heisenberg limit, 165–175 187–189, 194, 197, 225–227, 231–233,
Heisenberg uncertainty relation, 1, 5, 7, 23–24 236
Index 239

Momentum operator, 9, 38 Potassium titanyl phosphate (KTP), 139, 140

Multimode, 85–92, 115, 172 Purcell effect, 47, 187

N Q
Non-linear, 139–141, 172, 173, 175 Quadrature, 27, 30–31, 33, 34, 36–38, 43,
Non-local interference, 177, 178 93, 101–105, 107, 125, 127, 128,
NOON states, 165, 167–174 131, 138, 139, 143, 144, 146–152,
Normal order, 18, 42, 67, 96, 113, 137 157, 159, 168, 226
Number operator, 14–15, 17, 18, 38, 63, Quantum dot (QD), 51, 52, 92
67, 87, 96, 134 Quantum electrodynamics (QED), 44, 45, 183,
Number state, 14, 41, 123, 166, 168, 183 187–189, 197, 198, 215, 229
Number-phase uncertainty relation, 148, 161, Quantum harmonic oscillator (QHO), 6–24, 27,
166, 168, 169 33, 34, 36, 38, 104, 118, 184
Quantum illumination (QI), 177, 179, 180
Quantum imaging, 54, 82, 177–181
O Quantum lithography, 169, 170
One photon amplitude, 32, 86, 185 Quantum radar, 179
Optical clock, 210 Quartz oscillators, 205–210
Optical lattice, 219–222
Optical parametric amplification (OPA), 141,
151, 159 R
Optical parametric process, 139, 153 Rabi frequency, 192, 193, 195, 196, 199, 200
Optical tweezers, 219–222 Rabi oscillations, 183, 191–199, 212
Orthonormal, 14, 95, 183 Reflectance, 59, 63, 125
Reflection coefficient, 57, 125
Rotating wave approximation (RWA), 186
P Rydberg atoms, 183, 189, 190, 194, 197, 198
Parametric down-conversion (PDC), 51–53,
78, 172, 177, 179, 180
Paul trap, 211, 212, 214 S
Phase, 8, 28–31, 33, 36, 53, 57, 58, 71–73, Schrodinger, 27, 35, 101, 103, 195, 200
75, 93, 100, 102, 104, 107, 108, Schrodinger cat state, 167
115–117, 127, 128, 135, 137, 139, Second order correlation function, 65–68
142, 145, 147, 149, 152, 158–161, Second quantization, 27
165, 166, 168, 172, 174, 195, 196, Shot noise, 93, 96, 114, 121, 137–139, 141,
216, 217 157, 159, 161, 165, 166
Phase shifters, 166, 167, 172, 175 Signal-to-noise ratio (SNR), 100, 131, 137,
Phase squeezing, 146, 148, 159 149, 150, 157
Phasor, 93, 100, 101, 107–109, 116, 146, 152 Single mode, 27–32, 85, 94, 190, 201
Photodetector signal, 90–92, 117, 126 Single photon detector, 53–54
Photomultiplier tube (PMT), 53, 54 Single photon source (SPS), 51–53, 98, 123
Photon, 13, 32, 33, 38, 39, 41–43, 48–54, Single photon state, 42, 49–54, 68, 88, 89, 111
57–78, 80–82, 86–93, 95, 97, 98, 100, Single photon wavepacket, 51, 85, 89, 91,
111, 119, 121, 122, 131, 141, 142, 153, 233–234
154, 161, 165–172, 174, 175, 177–181, Single-photon avalanche diode (SPAD), 53, 54
183, 185–189, 191, 193, 196, 199, 200, Sisyphus cooling, 217–219
206, 213, 215, 216, 218, 219 Spontaneous emission, 44, 47, 48, 51, 89–92,
Photon number, 41–42, 50, 77, 93, 95–97, 100, 183, 186–188, 193, 194, 213, 214, 218
108, 115, 118, 119, 121, 122, 126, 129, Squeeze parameter, 142, 145
135, 152–155, 165, 168, 194–199 Squeezed light, 139–155, 157–163, 165
Poisson, 93, 96, 97, 121, 194, 199 Squeezed vacuum, 139, 141, 151–153, 155,
Post-selection, 175 157, 158, 160
240 Index

Squeezing operator, 142–144, 151, 153 Unitary, 60, 142, 149, 152, 175
Standard quantum limit (SQL), 27, 37, 104,
146–148
Stern–Gerlach apparatus, 207–208 V
Superconducting nanowire single-photon Vacuum, 27, 28, 31, 41–48, 62, 74, 85,
detector (SNSPD), 53, 54 87, 88, 105, 106, 111, 131, 133,
Superphoton, 169 138, 141, 151–155, 157, 187, 193,
Super-Poissonian, 121, 122, 152 198, 211, 225
Super-resolution, v, 165, 169–170 Vacuum energy, 39, 45, 49, 87, 88, 101, 185
Super-sensitivity, v, 165, 167–169 Vacuum Rabi oscillations, 193, 199
Vacuum Rabi splitting (VRS), 183, 199–202
Vaidman, L., 181
T Van der waals force, 44, 48
Tensor, 87, 195, 200 Variance, 5, 22, 67, 122
Thermal light, 115, 120–122
Time-dependent Schrodinger equation, 1, 4, 191
Time-independent Schrodinger equation, 4, W
10, 200 Wavepacket, 31, 51, 89–91, 102, 107,
Transformation, 60–62, 175 145, 172, 233
Transmission coefficients, 57, 60, 125, 126, 151 Wave-particle duality, 71, 76
Transmittance, 59, 64, 125

Z
U Zero-point fluctuations, 45
Uncertainty, 5–7, 21–24, 37, 41, 43, 49, 90, 93,
96, 97, 99–104, 107, 108, 119–121, 131,
136–139, 143–152, 157, 158, 161, 162,
165, 166, 168, 169, 205, 209, 210