ANDREI SERYI

Boca Raton London New York

CRC Press is an imprint of the

Taylor & Francis Group, an informa business
Open Access funded by SCOAP3

Copyright 2021 Andrei Seryi

This eBook was converted to open access in 2021 through the sponsorship of SCOAP3
licensed under the terms of the creative commons Attribution-NonCommercial 4.0
International License (https://creativecommons.org/licenses/by-nc/4.0/) which permits
use, sharing, adaptation distribution and reproduction in any medium or format, as
long as you give appropriate credit to the author(s) and the source, provide a link to
the creative commons license and indicate if changes were made, this license does not
permit the Contribution to be used commercially.

CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

'JSTU
JTTVFE
JO
QBQFSCBDL


© 2016 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

*4#/


QCL


*4#/


ICL

DOI: 10.1201/b18696

This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been
made to publish reliable data and information, but the author and publisher cannot assume responsibility for the valid­
ity of all materials or the consequences of their use. The authors and publishers have attempted to trace the copyright
holders of all material reproduced in this publication and apologize to copyright holders if permission to publish in this
form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may
rectify in any future reprint.

The Open Access version of this book, available at www.taylorfrancis.com, has been made available under a Creative
Commons Attribution-Non Commercial 4.0 International.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for
identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data

Seryi, Andrei, author.

Unifying physics of accelerators, lasers and plasma / Andrei Seryi.
pages cm
Includes bibliographical references and index.
ISBN 978-1-4822-4058-0 (alk. paper)
1. Particle accelerators. 2. Lasers. 3. Plasma (Ionized gases) I. Title.

QC787.P3S465 2016
539.7’3--dc23 2015012802

Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
To Elena

 Contents in Brief

1 Basics of Accelerators and of the Art of Inventiveness 1

2 Transverse Dynamics 21

3 Synchrotron Radiation 43

4 Synergies between Accelerators, Lasers and Plasma 57

5 Conventional Acceleration 75

6 Plasma Acceleration 105

7 Light Sources 127

8 Free Electron Lasers 143

9 Proton and Ion Laser Plasma Acceleration 165

10 Advanced Beam Manipulation, Cooling, Damping and Stability 185

11 Inventions and Innovations in Science 225

vii
 Contents

Foreword xxv

Preface xxvii

Author xxix

1 Basics of Accelerators and of the Art of Inventiveness 1

1.1 Accelerators and society 1

1.2 Acceleration of what and how 2

1.2.1 Uses, actions and the evolution of accelerators 3

1.2.2 Livingston plot and competition of technologies 4

1.3 Accelerators and inventions 6

1.4 How to invent 8

1.4.1 How to invent — evolution of the methods 8

1.5 TRIZ method 9

1.5.1 TRIZ in action — examples 10

1.6 TRIZ method for science 13

1.7 AS-TRIZ 14

1.8 Creativity 17

2 Transverse Dynamics 21

2.1 Maxwell equations and units 21

2.2 Simplest accelerator 22

2.3 Equations of motion 24

2.3.1 Motion of charged particles in EM fields 24

2.3.2 Drift in crossed E × B fields 25

2.3.3 Motion in quadrupole fields 25

2.3.4 Linear betatron equations of motion 26

2.4 Matrix formalism 27

2.4.1 Pseudo-harmonic oscillations 27

2.4.2 Principal trajectories 27

2.4.3 Examples of transfer matrices 30

2.4.4 Matrix formalism for transfer lines 30

2.4.5 Analogy with geometric optics 31

2.4.6 An example of a FODO lattice 32

2.4.7 Twiss functions and matrix formalism 33

2.4.8 Stability of betatron motion 33

2.4.9 Stability of a FODO lattice 34

2.4.10 Propagation of optics functions 34

2.5 Phase space 35

2.5.1 Phase space ellipse and Courant–Snyder invariant 35

2.6 Dispersion and tunes 36

ix

x Contents

2.6.1 Dispersion 36

2.6.2 Betatron tunes and resonances 38

2.7 Aberrations and coupling 38

2.7.1 Chromaticity 38

2.7.2 Coupling 39

2.7.3 Higher orders 39

3 Synchrotron Radiation 43

3.1 SR on the back of an envelope 43

3.1.1 SR power loss 43

3.1.2 Cooling time 45

3.1.3 Cooling time and partition 46

3.1.4 SR photon energy 46

3.1.5 SR — number of photons 47

3.2 SR effects on the beam 48

3.2.1 SR-induced energy spread 48

3.2.2 SR-induced emittance growth 49

3.2.3 Equilibrium emittance 50

3.3 SR features 51

3.3.1 Emittance of single radiated photon 51

3.3.2 SR spectrum 52

3.3.3 Brightness or brilliance 52

3.3.4 Ultimate brightness 53

3.3.5 Wiggler and undulator radiation 54

3.3.6 SR quantum regime 55

4 Synergies between Accelerators, Lasers and Plasma 57

4.1 Create 58

4.1.1 Beam sources 58

4.1.2 Lasers 60

4.1.3 Plasma generation 62

4.2 Energize 63

4.2.1 Beam acceleration 63

4.2.2 Laser amplifiers 63

4.2.3 Laser repetition rate and efficiency 64

4.2.4 Fiber lasers and slab lasers 64

4.2.5 CPA — chirped pulse amplification 65

4.2.6 OPCPA — optical parametric CPA 66

4.2.7 Plasma oscillations 67

4.2.8 Critical density and surface 68

4.3 Manipulate 68

4.3.1 Beam and laser focusing 68

4.3.2 Weak and strong focusing 69

4.3.3 Aberrations for light and beam 69

4.3.4 Compression of beam and laser pulses 71

4.3.5 Beam cooling 72

4.3.6 Optical stochastic cooling 73

4.4 Interact 73

 Contents xi

5 Conventional Acceleration 75

5.1 Historical introduction 75

5.1.1 Electrostatic accelerators 76

5.1.2 Synchrotrons and linacs 77

5.1.3 Wideröe linear accelerator 78

5.1.4 Alvarez drift tube linac 79

5.1.5 Phase focusing 80

5.1.6 Synchrotron oscillations 80

5.2 Waveguides 81

5.2.1 Waves in free space 81

5.2.2 Conducting surfaces 82

5.2.3 Group velocity 82

5.2.4 Dispersion diagram for a waveguide 83

5.2.5 Iris-loaded structures 84

5.3 Cavities 86

5.3.1 Waves in resonant cavities 86

5.3.2 Pill-box cavity 87

5.3.3 Quality factor of a resonator 87

5.3.4 Shunt impedance — Rs 88

5.3.5 Energy gain and transit-time factor 88

5.3.6 Kilpatrick limit 89

5.4 Power sources 90

5.4.1 IOT — inductive output tubes 90

5.4.2 Klystron 91

5.4.3 Magnetron 92

5.4.4 Powering the accelerating structure 93

5.5 Longitudinal dynamics 94

5.5.1 Acceleration in RF structures 94

5.5.2 Longitudinal dynamics in a travelling wave 95

5.5.3 Longitudinal dynamics in a synchrotron 95

5.5.4 RF potential — nonlinearity and adiabaticity 99

5.5.5 Synchrotron tune and betatron tune 99

5.5.6 Accelerator technologies and applications 101

6 Plasma Acceleration 105

6.1 Motivations 105

6.1.1 Maximum field in plasma 106

6.2 Early steps of plasma acceleration 107

6.3 Laser intensity and ionization 108

6.3.1 Laser pulse intensity 108

6.3.2 Atomic intensity 108

6.3.3 Progress in laser peak intensity 109

6.3.4 Types of ionization 110

6.3.5 Barrier suppression ionization 110

6.3.6 Normalized vector potential 111

6.3.7 Laser contrast ratio 112

6.3.8 Schwinger intensity limit 113

6.4 The concept of laser acceleration 114

6.4.1 Ponderomotive force 114

xii Contents

6.4.2 Laser plasma acceleration in nonlinear regime 116

6.4.3 Wave breaking 116

6.4.4 Importance of laser guidance 117

6.5 Betatron radiation sources 118

6.5.1 Transverse fields in the bubble 118

6.5.2 Estimations of betatron radiation parameters 119

6.6 Glimpse into the future 120

6.6.1 Laser plasma acceleration — rapid progress 120

6.6.2 Compact radiation sources 121

6.6.3 Evolution of computers and light sources 122

6.7 Plasma acceleration aiming at TeV 122

6.7.1 Multi-stage laser plasma acceleration 122

6.7.2 Beam-driven plasma acceleration 123

6.8 Laser-plasma and protons 124

7 Light Sources 127

7.1 SR properties and history 127

7.1.1 Electromagnetic spectrum 127

7.1.2 Brief history of synchrotron radiation 128

7.2 Evolution and parameters of SR sources 129

7.2.1 Generations of synchrotron radiation sources 129

7.2.2 Basic SR properties and parameters of SR sources 130

7.3 SR source layouts and experiments 131

7.3.1 Layout of a synchrotron radiation source 132

7.3.2 Experiments using SR 133

7.4 Compton and Thomson scattering of photons 135

7.4.1 Thomson scattering 136

7.4.2 Compton scattering 136

7.4.3 Compton scattering approximation 138

7.4.4 Compton scattering characteristics 138

7.5 Compton light sources 139

8 Free Electron Lasers 143

8.1 FEL history 143

8.2 SR from bends, wigglers and undulators 144

8.2.1 Radiation from sequence of bends 144

8.2.2 SR spectra from wiggler and undulator 145

8.2.3 Motion and radiation in sine-like field 146

8.3 Basics of FEL operation 147

8.3.1 Average longitudinal velocity in an undulator 147

8.3.2 Particle and field energy exchange 148

8.3.3 Resonance condition 149

8.3.4 Microbunching conceptually 149

8.4 FEL types 150

8.4.1 Multi-pass FEL 150

8.4.2 Single-pass FEL 151

8.5 Microbunching and gain 152

8.5.1 Details of microbunching 152

8.5.2 FEL low-gain curve 155

 Contents xiii

8.5.3 High-gain FELs 156

8.6 FEL designs and properties 157

8.6.1 FEL beam emittance requirements 157

8.6.2 FEL and laser comparison 158

8.6.3 FEL radiation properties 158

8.6.4 Typical FEL design and accelerator challenges 159

8.7 Beyond the fourth-generation light sources 160

9 Proton and Ion Laser Plasma Acceleration 165

9.1 Bragg peak 166

9.2 DNA response to radiation 169

9.3 Conventional proton therapy facilities 171

9.3.1 Beam generation and handling at proton facilities 172

9.3.2 Beam injectors in proton facilities 173

9.4 Plasma acceleration of protons and ions — motivation 176

9.5 Regimes of proton laser plasma acceleration 176

9.5.1 Sheath acceleration regime 177

9.5.2 Hole-boring radiation pressure acceleration regime 179

9.5.3 Light-sail radiation pressure acceleration regime 180

9.5.4 Emerging mechanisms of acceleration 181

9.6 Glimpse into the future 182

10 Advanced Beam Manipulation, Cooling, Damping and Stability 185

10.1 Short and narrow-band 185

10.1.1 Bunch compression 185

10.1.2 CSR — coherent synchrotron radiation 188

10.1.3 CSR effects on the beam longitudinal phase space 189

10.1.4 Short laser pulse and Q-switching techniques 192

10.1.5 Q-switching methods 193

10.1.6 Regenerative amplifiers 194

10.1.7 Mode locking 195

10.1.8 Self-seeded FEL 196

10.2 Laser–beam interaction 196

10.2.1 Beam laser heating 197

10.2.2 Beam laser slicing 198

10.2.3 Beam laser harmonic generation 199

10.3 Stability of beams 200

10.3.1 Stability of relativistic beams 200

10.3.2 Beam–beam effects 200

10.3.3 Beam break-up and BNS damping 203

10.3.4 Landau damping 205

10.3.5 Stability and spectral approach 207

10.4 Beam or pulse addition 209

10.4.1 Optical cavities 210

10.4.2 Accumulation of charged particle bunches 211

10.4.3 Coherent addition of laser pulses 212

10.4.4 Resonant plasma excitation 213

10.5 Cooling and phase transfer 214

10.5.1 Beam cooling methods 214

xiv Contents

10.5.2 Electron cooling, electron lens and Gabor lens 214

10.5.3 Laser cooling 216

10.6 Local correction 217

10.6.1 Final focus local corrections 217

10.6.2 Interaction region corrections 219

10.6.3 Travelling focus 220

10.6.4 Crabbed collisions 221

10.6.5 Round-to-flat beam transfer 222

11 Inventions and Innovations in Science 225

11.1 Accelerating Science TRIZ 226

11.2 Trends and principles 227

11.2.1 TRIZ laws of technical system evolution 227

11.2.2 From radar to high-power lasers 228

11.2.3 Modern laws of system evolution 229

11.3 Engineering, TRIZ and science 229

11.3.1 Weak, strong and cool 230

11.3.2 Higgs, superconductivity and TRIZ 230

11.3.3 Garin, matreshka and Nobel 231

11.4 Aiming for Pasteur quadrant 233

11.5 How to cross the Valley of Death 236

11.6 How to learn TRIZ in science 240

11.7 Let us be challenged 242

Final Words 245

Bibliography 247

Index 253

 List of Figures

1.1 Gear-like structure in jumping insects as an illustration of nature’s inven­

tiveness. Burrows and Sutton, 2013. Reproduced with permission. 2

1.2 Helical solenoid channel. 2

1.3 Basic principles of acceleration — electrostatic, betatron, in an EM wave

in an accelerating structure. 2

1.4 Uses of accelerated beams — sending to target, colliding with another

beam, characterization of the beam or separation into species, generation

of useful radiation. 3

1.5 Actions on accelerated beams — acceleration, focusing, generation of

radiation, colliding. 4

1.6 Livingston plot of evolution of accelerators. 5

1.7 Evolution of technologies — saturation and replacement by newer tech­

nologies. 5

1.8 Van der Graaf accelerator. 6

1.9 Cyclotron accelerator. 6

1.10 Synchrotron accelerator. 6

1.11 Strong focusing concept. 6

1.12 Collective acceleration. 7

1.13 Electron cooling concept. 7

1.14 Stochastic cooling concept. 7

1.15 Plasma acceleration concept. 7

1.16 Illustration of TRIZ in action — initial specific problem. 10

1.17 Illustration of the flow of the TRIZ algorithm. 11

1.18 Illustration of TRIZ in action — specific solution. 12

1.19 Valery Bryusov’s electron as an analogy to the TRIZ inventive principle of

nested dolls. 13

1.20 High energy physics detectors, which have a layered “nested” structure,

reflecting the TRIZ inventive principle of Russian dolls. 13

1.21 Particle interaction event observed in a cloud chamber invented by Wilson

in 1911 (left), and in a bubble chamber invented by Glaser in 1952 (right). 14

1.22 Carbon wire beam profile monitor. 15

1.23 Laser wire beam profile monitor. 16

1.24 Looking at the world through the prism of TRIZ. Illustration by Sasha

Seraia. 17

1.25 TRIZ vs. brainstorming. Illustration by Sasha Seraia. 18

2.1 Simple electron gun. 22

2.2 Electron gun with Pierce electrode and collector made in the form of a

Faraday cup. 23

2.3 Motion of charged particles in a uniform magnetic field. 24

2.4 Drift in crossed E×B fields. 25

2.5 Magnetic fields and forces acting on a particle in a quadrupole. 25

2.6 Frenet–Serret curvilinear coordinate system. 26

xv

xvi List of Figures

2.7 Shifted circles cross. 26

2.8 Illustration of the origin of weak focusing in dipoles. 26

2.9 Illustration of pseudo-harmonic oscillations and cosine-like and sine-like

principal trajectories. 28

2.10 Linear matrix approach for evaluation of the evolution of the particle

coordinates in a transfer line. 29

2.11 Mechanism of the edge focusing of bending magnet in a horizontal plane. 30

2.12 For illustration of transfer matrix of two quadrupoles separated by a drift. 31

2.13 A doublet can focus simultaneously in both planes. 31

2.14 Optical telescope with two lenses. 32

2.15 FODO lattice. 32

2.16 Betatron motion in phase space. 36

2.17 Evolution of phase-space ellipse. Locations: (a) in D and (c) in F

quadrupoles, and (b) in between. 36

2.18 Bending magnet creates dispersion. 36

2.19 Dispersion. 37

2.20 Chromaticity of a focusing quadrupole. 38

2.21 Skew quad fields and forces. 39

2.22 Sextupole fields and forces. 39

2.23 Octupole magnet forces. 39

3.1 Synchrotron radiation — conceptual explanation. 44

3.2 Illustrating the characteristic volume used in Eq. 3.3. 44

3.3 RF cavity restores only longitudinal momentum, thus other degrees of

freedom are cooled due to synchrotron radiation. 45

3.4 SR and remote observer. 47

3.5 SR can cause excitation of oscillation of particles and corresponding emit­

tance growth. 49

3.6 For illustration of emittance of a single photon. 51

3.7 SR spectrum and its approximations for low and high energies. 52

3.8 For illustration of brilliance or brightness. 53

3.9 Radiation from sequence of bends. 54

3.10 Wiggler and undulator radiation. 55

3.11 SR spectrum in classical and quantum regimes. 55

4.1 Synergies between accelerators, lasers and plasma. 57

4.2 Discussion of synergies will follow this sequence. 58

4.3 Thermal cathode e-gun. 58

4.4 One-and-a-half-cell RF photocathode electron gun. 59

4.5 Surface-plasma Penning H − ion source. 59

4.6 Conceptual diagram of the laser-driven ion source. 60

4.7 Laser diagram. 60

4.8 Three-level laser. 60

4.9 Conceptual diagram of a ruby laser. Quartz flash tube serves as the pump

source and ruby crystal as the gain medium. 61

4.10 Diode laser in application for pumping of Nd:YAG laser. 62

4.11 Paschen discharge curve for hydrogen gas. 62

4.12 Fields of a relativistic electron bunch can produce field ionization of gas. 62

4.13 Electrostatic and betatron acceleration. RF cavity and RF structure. 63

4.14 Laser amplifier. Flash lamp emits in broad spectrum. Gain medium am­
plifies selected wavelength. 63

 List of Figures xvii

4.15 The cat intuitively knows the inventive principle of changing her surface-

to-volume ratio depending on the external temperature. 64

4.16 Volume-to-surface ratio S/V in units of (2πV )1/3 vs L/R. 64

4.17 Schematic of a fiber laser and cross section of refractive index. 65

4.18 Schematics of CPA — chirped pulse amplification. 65

4.19 Optical parametric generation in nonlinear crystals. 66

4.20 Optical parametric chirped pulse amplification — OPCPA. 66

4.21 Plasma oscillations. 67

4.22 Laser penetration to plasma and critical density. 68

4.23 Beam or light in the focus. 68

4.24 Weak focusing. 69

4.25 Strong focusing. 69

4.26 Lenses made from different glasses compensate chromatic aberrations. 70

4.27 Compensation of chromatic aberration by inserting nonlinear sextupole

magnets in a dispersive region. 70

4.28 Laser pulse stretcher. 71

4.29 Laser pulse compressor. 71

4.30 Bunch compressor. 72

4.31 Electron cooling. 72

4.32 Stochastic cooling. 72

4.33 Standard stochastic cooling. 73

4.34 Optical stochastic cooling. 73

4.35 Layout of optical stochastic cooling system in an accelerator. 73

5.1 A cathode ray tube TV as an example of an accelerator. 75

5.2 Cockcroft–Walton generator. 76

5.3 Van der Graaf accelerator. 76

5.4 Tandem electrostatic accelerator. 76

5.5 Pelletron charging mechanism. 77

5.6 Synchrotron and linac. 77

5.7 Wideröe linear accelerator. 78

5.8 Voltage in Wideröe linac. 79

5.9 Alvarez drift tube linac. 79

5.10 RFQ structure. 80

5.11 Synchrotron oscillations. 81

5.12 TEM wave in free space. 82

5.13 Boundary conditions on perfectly conducting surfaces. 82

5.14 Two-wave interference. 82

5.15 Waves in a waveguide, two extreme cases. 83

5.16 Dispersion of a waveguide, two extreme cases. 83

5.17 Intermediate case. 84

5.18 Dispersion of a waveguide. 84

5.19 Iris-loaded accelerating structure. 85

5.20 Qualitative behavior of dispersion curve in iris-loaded structures. 85

5.21 Extended dispersion diagram of an iris-loaded structure. 86

5.22 Cylindrical pill-box cavity. 87

5.23 Examples of pill-box cylindrical cavity modes with electric field lines

shown. 87

5.24 The RF gap — space between entrance and exit irises of cavity resonator

in drift tube linac. 89

xviii List of Figures

5.25 Breakdown Kilpatrick limit (lower curve) and Wang–Loew limit (upper

curve). 90

5.26 Schematic of an inductive output tube. 91

5.27 Schematic of a klystron. 91

5.28 Schematic of a magnetron. 93

5.29 Feeding RF power into an accelerating structure. Field lines show electric

and magnetic fields of the corresponding cavity modes. 93

5.30 Acceleration in a travelling wave structure (left) and in a standing wave

structure (right). The wave and particles’ position in different moments of

time are shown. 94

5.31 Synchronous and lagging particles in a synchrotron ring. 96

5.32 Motion in RF potential. 96

5.33 RF bucket trajectories in a linearized case are ellipses. 98

5.34 RF voltage and phase space and RF potential for cases below and above

the transition energy. 98

5.35 Qualitative evolution of the longitudinal phase space (energy vs phase,

for vertical and horizontal axes, correspondingly) of the beam for an

increasing number of synchrotron periods. 99

5.36 RF bucket in the case of fast acceleration. 99

5.37 Betatron oscillations modulated by synchrotron motion (left) and a corre­
sponding spectrum (right) with betatron tune and synchrotron sidebands. 100

5.38 A generic linear collider. 101

5.39 A generic free electron laser. 102

6.1 For illustration of plasma beat wave and self-modulated laser wakefield

acceleration. 107

6.2 Plasma wakefield acceleration — PWFA. 107

6.3 Laser wakefield acceleration — LWFA. 107

6.4 Laser focused to a tight spot. 108

6.5 Qualitative overview of the progress in laser peak intensity. 109

6.6 Types of ionization: (a) direct, (b) multi-photon, (c) tunneling. 110

6.7 Barrier suppression ionization. 111

6.8 Qualitative temporal profile of a CPA-compressed laser pulse. 113

6.9 Laser acceleration — conceptually. Linear regime. 114

6.10 For illustration of the mechanism of the ponderomotive force. 115

6.11 Bubble formation. 116

6.12 Laser plasma acceleration in nonlinear regime — conceptually. 116

6.13 Wave breaking concept — the wave nonlinearity gradually rises from top

to bottom. 116

6.14 Capillary channel technique of laser plasma acceleration. 117

6.15 Cylindrical symmetry in the plasma bubble. 118

6.16 Laser plasma betatron source — conceptually. Wave breaking and self-

injection — (a). Oscillation of accelerating electron beams in the plasma

bubble — (b)-(d), sequential time moments. Betatron radiation produced

by oscillating beams — (e). 119

6.17 Laser plasma betatron radiation light source — conceptually. 121

6.18 Computers’ evolution. 122

6.19 Light sources’ evolution. 122

6.20 Beam-driven plasma acceleration — conceptually. 123

6.21 Sheath laser plasma acceleration of protons or ions. 124

 List of Figures xix

7.1 Electromagnetic spectrum covered by SR and Compton sources. 127

7.2 Photon attenuation in water in comparison with a typical protein. 128

7.3 Generations of SR sources. Brightness is expressed in the units of the

number of photons per s · mm2 · mrad 2 · 0.1%BW . 129

7.4 Generic SR light source with multiple X-ray beamlines and showing typical

allocation of beamlines to experiments. 131

7.5 Schematics of a generic third-generation SR light source. 132

7.6 Current in SR light source without (a) and with (b) top-up injection mode. 133

7.7 Crystal monochromator of X-rays. Symmetric case (a) and asymmetric

case (b). 133

7.8 Absorption (left) and phase contrast (right) X-ray imaging and comparison

of reconstructed image (middle). 134

7.9 Pump-probe experiment arrangement. Here T and n are revolution period

and number of bunches in the SR ring, Δt is time delay between the pump

laser pulse and SR probe pulse. 135

7.10 Thomson scattering. 136

7.11 Compton backscattering. Initial photon with wavelength λ1 and after scat­
tering with λ2 . 136

7.12 Compton scattering in the rest frame of an electron and relativistic invari­
ants. 137

7.13 Compton scattering — definition of frequencies and angles. 138

7.14 Generic Compton source of linac type. 139

7.15 Generic Compton light source based on electron storage ring. 140

8.1 Trajectory and radiation in a sequence of bending magnets. 144

8.2 Wiggler (top) and bending magnet (bottom) SR spectra. 144

8.3 Radiation from wiggler, regime of K » 1. 145

8.4 Time profile of radiation observed from wiggler. 145

8.5 Spectrum from wiggler (left) and undulator (right), qualitative comparison.

Dashed line on the left spectrum corresponds to the spectrum from bends

of the same strength. Horizontal axis is in units of λu /(2γ 2 ). 145

8.6 Radiation from undulator, with K « 1. 146

8.7 Time profile of radiation observed from undulator. 146

8.8 Trajectory and radiation in sine-like field. 146

8.9 EM wave and particle trajectory — straight (left) and wiggling (right) in an

undulator. 148

8.10 EM wave-particle resonance condition of energy transfer. 149

8.11 Microbunching. Density of the beam along the longitudinal coordinate

for the initial noise (left), intermediate regime of microbunching (middle)

and saturated microbunching (right). 150

8.12 Multi-pass FEL. 151

8.13 Single-pass FEL. 151

8.14 Radiation in an FEL undulator composed of permanent magnets. 152

8.15 Illustrating solutions of FEL-pendulum equation and microbunching for

different initial conditions. The initial beam (I) is on-energy and when

bunched (II) demonstrates symmetrical profile of beam density (B). 154

8.16 Microbunching in a case when the initial beam is slightly off energy. 154

8.17 FEL low-gain curve. 156

8.18 High-gain FELs, typical behavior of the emitted power — exponential

growth eventually turned into saturation. 156

xx List of Figures

8.19 Peak brilliance (left) and temporal resolution (right) of typical FEL in com­
parison with third-generation SR sources. 158

8.20 Generic layout of a compact light source driven by an LPWA. 160

8.21 For illustration of filamentation. An intact paper sheet (top) may have very

low volume; however, when crumpled (bottom) it will have its effective

volume increased by orders of magnitude. 161

9.1 Photon matter interaction, qualitatively. 166

9.2 Absorption of photons (dotted lines) in comparison with absorption of

protons in media. Overlaying multiple Bragg peaks creates a near uniform

dose distribution in a certain target volume. 167

9.3 Radiation effects on DNA. 169

9.4 Generic proton or heavy-ion therapy facility. 171

9.5 The elements of the proton therapy beamline. 172

9.6 Pencil beam scanning. 172

9.7 Schematic of a cyclotron. 173

9.8 Schematics of a synchrocyclotron. 174

9.9 Schematics of an isochronous cyclotron. 175

9.10 Example of a field profile in an isochronous cyclotron. 175

9.11 Sheath laser acceleration of protons. 177

9.12 TNSA spectum, qualitative behavior. 178

9.13 Radiation pressure acceleration concept. 179

9.14 Hole-boring radiation pressure laser acceleration of protons. 179

9.15 Light-sail radiation pressure laser acceleration of protons. 180

10.1 Velocity bunching. Initial beam (a) and compressed beam (b). 186

10.2 Four-magnet chicane. 186

10.3 Energy–time correlation and bunch compression. 186

10.4 Incoherent radiation (left) and coherent radiation (right). 189

10.5 Qualitative comparison of the spectrum of coherent synchrotron radiation

in comparison with the spectrum of incoherent SR. 189

10.6 Two-particle model of the beam and its field. 189

10.7 Illustration of the tail field overtaking the head of the bunch in the mech­
anism of coherent synchrotron radiation. 190

10.8 Shape function F0 (top plot) of coherent synchrotron radiation for a bunch

with Gaussian density profile (bottom plot). 191

10.9 Q-switching technique. In step one (a) the pump builds up large inversion

in the gain media. In step two (b) the laser cavity switches from low to

high-Q. 192

10.10 Examples of active Q-switching methods. Rotating mirror (A), Electro­

optic (B) and Acousto-optic (C). 193

10.11 Passive Q-switching — saturable absorber (A) and SESAM (B). 194

10.12 Schematics of a regenerative amplifier. 194

10.13 Mode-locked laser (left) and the laser output (right) in the normal (a) and

mode-locked (b) regimes. 195

10.14 Self-seeded FEL. 196

10.15 Laser heater. 197

10.16 Beam laser slicing. 198

10.17 Echo-enabled harmonic generation scheme — EEHG. 199

10.18 Phase space (top) and density profile (bottom) of an EEHG-modulated

beam. 199

 List of Figures xxi

10.19 Fields of the bunch and head–tail effects. 200

10.20 Flat beam collision in an IR of a typical linear collider. 201

10.21 Fields of the flat beam. 201

10.22 Beamstrahlung. 202

10.23 Consequent moments of high-disruption beam collision. 202

10.24 Beam break-up instability of a single beam. Fields left by the bunch are
shown qualitatively. Beam evolution from the initial unperturbed shape
(A) to the final BBU-distorted shape (B). 203

10.25 BNS damping method. 204

10.26 Detuned structure as a cure for multi-bunch BBU instability. 205

10.27 For illustration of Landau damping mechanism. 206

10.28 Velocity spread and Landau damping. 206

10.29 Power spectrum. 207

10.30 Examples of power spectrum P(ω, k) (left), spectral response function

G(k) and characteristic function of the feedback F(ω). 208

10.31 Examples of optical cavities. Plane-parallel (A), concentric/spherical (B)

and confocal (C) configurations. 210

10.32 Examples of four-mirror optical cavity suitable for electron beam–laser

interaction. 210

10.33 Phase-space stacking. 211

10.34 Transverse phase-space stacking. Consecutive moments. 211

10.35 Longitudinal phase-space stacking. 212

10.36 Charge-exchange injection. 212

10.37 Concept of fiber laser coherent combination of pulses. 213

10.38 Electron cooling, stochastic cooling and ionization cooling concepts. 214

10.39 Electron cooling or electron lens. 215

10.40 Conceptual schematic of a Gabor lens. 215

10.41 Relations of velocities of proton and electron beams in different configu­

rations: electron cooling, electron lens, Gabor lens. 216

10.42 Laser cooling steps. Absorption of a photon by an atom (a); excited state

of the atom (b); emission of a photon (c). 217

10.43 Relation between laser wavelength and Doppler shifted resonance ab­
sorption of an atom moving in different directions. 217

10.44 Final focus with local chromaticity correction. 218

10.45 Conceptual layout of experimental detector and beamlines in the interac­

tion region of a linear collider. 219

10.46 Standard solenoid (A) and interaction region dual solenoids (B). 220

10.47 Hourglass effect. 220

10.48 Travelling focus collisions. 221

10.49 Collisions of the beams with crossing angle at the IP. Normal (A) and

crabbed (B) collisions. 221

10.50 Crab cavity and its fields. 222

10.51 Beamline magnetic elements and phase-space portraits of the beam sub­
jected to flat-to-round beam transformation. Initial flat beam, vortex, par­
allel beam in the solenoid. 222

11.1 Stimulated emission depletion microscopy (STED) and TRIZ inventive

principle of matreshka and system–antisystem. Excitation laser pulse (a),

de-excitation pulse (b) and remaining fluorescence (c). Improvement of

resolution of a protein imaging due to STED is shown qualitatively on the

right. 232

xxii List of Figures

11.2 One-dimensional, linear model of research. 233

11.3 Dynamic linear model of technology transfer. 233

11.4 Revised dynamic model of research and technology transfer. 234

11.5 Pasteur quadrant and accelerator science. 235

11.6 The units of quantitative assessments of research in the Pasteur quadrant. 235

11.7 Academia–industry–investor puzzle caused by different motivations of the

three participating groups. 238

11.8 Working on a portfolio of compact X-ray light sources can help in cross­
ing the “Valley of Death” between accelerator science and technological

innovation. 238

11.9 A solution to the academia–industry–investor puzzle — work on three

designs of compact X-ray sources. 239

11.10 The FCC beam energy will be comparable to that of an airplane, while the
beam will need to be focused at the interaction point to micron-scale size
— an analogy with a plane passing through the needle’s eye. 242

 List of Tables

1.1 TRIZ contradiction matrix for speed–temperature parameters and indexes

of the corresponding inventive principles 12

1.2 Elements of TRIZ contradiction matrix 12

1.3 TRIZ inventive principles 13

1.4 Emerging AS-TRIZ parameters 15

1.5 Emerging AS-TRIZ principles 15

1.6 Emerging AS-TRIZ contradiction matrix and indexes of inventive principles 17

5.1 Operating frequencies and typical parameters for RF cavities 101

8.1 FEL and laser comparison 158

8.2 NC and SC FEL, typical parameters 159

11.1 Updated table of emerging AS-TRIZ parameters 226

11.2 Updated table of emerging AS-TRIZ inventive principles 226

xxiii
 Foreword

Accelerators were invented in the early 1930s and first devel­

oped in a significant way by E.O. Lawrence at the University
of California, at Berkeley. Initial ideas arose in response to
the needs of particle physics expressed by Ernest Rutherford
as early as 1924, but right from the outset, Lawrence used
his early cyclotron accelerators for both pure research in par­
ticle (then called nuclear) physics and more practical appli­
cations. Much of the development was funded by their use
in the field of medicine for isotope production and therapy.
After the World War II years 1939 to 1945, came the inven­
tion of the synchrotron — a clever extension of the cyclotron
principle to undercut the rising cost of cyclotrons and extend
their energy range to allow the production of more massive
fundamental particles. Readers who follow the description
of the TRIZ inventive process in this book may wonder if this
invention was an accidental example of what TRIZ calls the
“Russian Doll” technique.
Soon it was realized that electron synchrotrons were a
prolific and controllable source of synchrotron radiation, ini­
tially in the ultraviolet spectrum and later down to the wave­
lengths of X-rays. A large number of these SR sources were
built and their beams used for scattering experiments, to
fathom the structure of new materials for engineering and
elucidate the molecular structure of protein and the other
complex molecules that have come to dominate our under­
standing of today’s life sciences. To extend the scope of these
studies in order to allow single molecules to be reconstructed
from their scattering patterns, a completely new concept —
the free electron laser — was invented. This has already been
used in several countries as a basis for the construction of
research facilities that rival the large colliders of particle
physics in scale and budget. Finding a new approach to the
production of X-rays in this way has been another leap in hu­
man imagination, opening the door to new and more power­
ful applications in a field where those applications bear an
almost immediate return on investment. It is the purpose of
this book to explore how such leaps in imagination may be
stimulated.
To be significant, such inventive processes need to iden­
tify a symbiotic relationship between two apparently quite
different fields — for the free electron laser these were ac­
celerators and lasers — and then find a common factor that
can be used to improve both fields. This technique has been
recently developed and refined as a replacement for simple

xxv
xxvi unifying physics of accelerators, lasers and plasma

brainstorming. It is called TRIZ and has already been identi­

fied by industry as a means towards more rapid technological
progress. We discuss here how it may be used to cut the Gor­
dian Knot1 of rising costs and complexity, which threaten to
impede the development of more powerful instruments for
both pure and applied research.
One does not have to look far to find a candidate for
a symbiotic technique to pair with today’s accelerators. A
laser beam penetrating a plasma generates accelerating fields
many orders of magnitude greater than today’s accelerators.
This book for the first time puts these two techniques side by
side so that we may identify commonalities of solution and
stimulate today’s students of accelerator and laser/plasma
physics to work together and use TRIZ to solve the problem.
The book’s author heads the John Adams Institute, whose
participating universities include postgraduate students in
both the laser and accelerator fields. We are well placed to
kindle the fire of invention amongst them, and we urge those
responsible for young and inventive minds elsewhere to join
this mission. Our students are already applying our ideas to
the Future Circular Collider project of CERN.

Edmund Wilson — Geneva, March 7, 2015

1 The Gordian Knot is a legend of Phrygian Gordium associated with

Alexander the Great. It is often used as a metaphor for disentangling an in­
tractable problem.
 Preface

The aim of this book is to build bridges and connections be­

tween three areas of physics that are essential for developing
the next generation of accelerators: accelerators, lasers and
plasma. These three fields of physics will be introduced in
tandem with the industrial methodology of inventiveness, a
method which teaches that similar problems and solutions
appear again and again in seemingly dissimilar disciplines.
This methodology of inventiveness will, ultimately, further
enhance connections between the aforementioned fields, and
will grant the reader a novel perspective.
The text is suitable for students of various levels between
senior undergraduate and graduate physics who are inter­
ested in enhancing their ability to work successfully on the
development of the next generation of facilities, devices and
scientific instruments manufactured from the synergy of ac­
celerators, laser and plasma. I would also recommend this
book to anyone interested in scientific innovations.
The idea for the book Unifying Physics of Accelerators,
Lasers and Plasma came not by accident — it naturally re­
sulted from a search for the best method to suitably train
the undergraduate and graduate students at John Adams In­
stitute and its affiliate universities. The aim is to teach sev­
eral physics disciplines in a coherent way, simultaneously
ensuring that this training would develop and stimulate
innovativeness.
Materials for this book developed gradually, starting from
a single lecture presented at JINR Dubna in March 2014, to
several lectures given to Oxford undergraduates in May 2014,
and eventually to a full week-long US Particle Accelerator
School course given in June 2014. The text, to a large ex­
tent, follows the materials developed for this USPAS course.
Interaction with the students during this course proved to be
a significant inspiration for converting these materials into
book form, as well as the lectures, presentations of mini-
projects by student teams and tutorial sessions that were fo­
cused on analyzing key inventions that shaped the discussed
scientific areas.
The style of this book is perhaps different from a typical
textbook on accelerator physics, or books on lasers or plasma.
Here, we tend to use qualitative discussion and prefer to
avoid heavy math, using back-of-the-envelope type deriva­
tions and estimations whenever possible (even if they pro­
vide approximate formulas). We believe that this is the better

xxvii
xxviii unifying physics of accelerators, lasers and plasma

way to convey physics principles to the reader and is certainly

preferable when discussing such a broad area of physics.
I would like to express my gratitude to my many col­
leagues who directly or indirectly helped with this endeavor.
Particular thanks go to Professor Riccardo Bartolini, Profes­
sor Emmanuel Tsesmelis and Professor Ted Wilson for joint
work on the short-option course for Oxford undergraduates,
which was the first step toward the development of the US­
PAS course that gradually led to development of this book.
I am grateful to the students of my USPAS-2014 course for
inspiration and enthusiastic discussions and also to Profes­
sor Bill Barletta who supported the idea of the USPAS course,
discussed the very first time with Bill in early April 2013.
I am grateful to the Rector of Novosibirsk State Univer­
sity, Mikhail Fedoruk, for creating an inspiring opportunity
to present a lecture on Science and Inventiveness to a cohort
of 1,000 first-year students on September 1, 2014; the prepa­
rations for this lecture significantly stimulated the thought
process for writing the first chapter of this book.
I am grateful to Professor Zulfikar Najmudin for help
with the materials for ion plasma acceleration and thankful
to Professor Peter Norreys and Professor Bob Bingham for
discussions on the topics connecting chirped pulse amplifi­
cation and radar inventions.
I am in immense debt to Professor Ted Wilson, who made
tremendous efforts to be the first reader of the manuscript
and who gave me a lot of valuable comments.
I am grateful to my John Adams Institute colleagues
— Professors Riccardo Bartolini, Stewart Boogert, Simon
Hooker, Ivan Konoplev, Zulfikar Najmudin and Mike
Partridge for reading the manuscript or its individual chap­
ters and giving their valuable comments. I would like to
express gratitude to all of my colleagues and friends who
helped with the preparation of this book and whom I failed
to thank here directly.
I am also delighted to thank Francesca McGowan, CRC
Press editor, who on April 30, 2013 appeared out of the blue
in my office asking for directions to a meeting — this unex­
pected conversation helped to crystallize my thoughts toward
this book. I appreciate all the aid Francesca and her CRC
Press colleagues provided during the writing process.
I would like to express my deep thanks to my family, first
of all to my daughter Sasha, who designed the book cover,
made several illustrations and helped with grammar.
And most importantly, my eternal gratitude to my wife
Elena, who put a massive amount of effort into converting
lecture materials to LaTeX, created more than 200 illustra­
tions for this book and gave me support and motivation,
without which this book would never have been finished.
 Author

Andrei Seryi, currently a professor at Oxford University and

Director of the John Adams Institute, graduated from Novosi­
birsk State University in 1986 and received his Ph.D. from
the Budker Institute of Nuclear Physics in 1994. He worked
at Stanford Linear Accelerator Center until 2010, where he
led the design and first stages of implementation of the Fa­
cility for Advanced Accelerator Experimental Tests (FACET)
project and coordinated the Beam Delivery efforts for the lin­
ear collider. He served as a deputy spokesperson of the ATF
International Collaboration for ATF2 Project, is a chairper­
son or a member of a number of advisory committees and
a Fellow of American Physical Society. As well as teaching
at numerous accelerator schools, he has contributed to the
developments of beam–beam compensation, electron cool­
ing, beam-delivery systems, stability of colliders and novel
acceleration methods. His professional interests include: the
development of accelerator science applications to discovery
science, industry, healthcare and energy; organizing of scien­
tific research; project management; crisis management in sci­
entific and technological areas; inventions and innovations;
and developing novel training approaches and methods.

xxix
 1
Basics of Accelerators and of
the Art of Inventiveness

1.1 Accelerators and Can you imagine that electrons

society 1 Are planets circling their Suns?
1.2 Acceleration of Space exploration, wars, elections
what and how 2 And hundreds of computer tongues
1.3 Accelerators and
inventions 6 Author’s translation of 1920 poem of
Valery Bryusov “The World of
1.4 How to invent 8
Electron”
1.5 TRIZ method 9
1.6 TRIZ method for In this chapter we will discuss the basic terms related to
science 13 accelerators. We will also describe the framework and meth­
1.7 AS-TRIZ 14 ods of the theory of inventive problem solving.
1.8 Creativity 17

1.1 Accelerators and society

Accelerators are essential for science and society — they are
in use in high energy physics, nuclear physics, healthcare and
life science. They are important to industry, the development
of new materials, energy and security, and can be applied to
many other fields.
It is enough to state just three reasons why society needs
accelerators: tens of millions of patients receive accelerator-
based diagnoses and treatments each year in hospitals and
About 30% of Nobel prizes in clinics around the world; all products that are processed,
Physics are due to use of ac­ treated or inspected by particle beams have a collective an­
celerators. nual value of more than $500B; and a significant fraction of
the Nobel prizes in physics are directly connected to the use
of accelerators.
If you are a young person thinking about your future, you
might be interested to know that accelerator science can lead
to attractive career prospects, because whether you are more
inclined towards a theoretical approach, experiments or com­
puter modeling, there will be an array of tasks and oppor­
tunities ready for you. Moreover, the knowledge gained and
developed in accelerator science has connections and appli­
cations to surprisingly remote disciplines, from stock market
predictions to planetary motion dynamics.
The recent discovery of the Higgs boson by the Large
Hadron Collider — the grandest machine ever built — has
once again demonstrated accelerator science’s and associated
technologies’ potential to reveal some of the most fundamen­
tal constructs of nature.

DOI: 10.1201/b18696-1 1
2 unifying physics of accelerators, lasers and plasma

Accelerator science demonstrates a rich history of inven­

tions, often inspired by nature itself. We are very often moti­
vated by nature and try to compete with it, not always know­
ing who invented certain things first — nature or humans.
For example, it is perhaps a common belief that gears were in­
vented by humans. In fact, insects have been using gears1 for
millions of years! The mechanism allows the insect to jump
FIGURE 1.1 in a straight line, its left and right leg synchronized by a gear-
Gear-like structure in jump­ like connection — see Fig. 1.1 — which works better than a
ing insects as an illustra­ synchronization via nervous signals.
tion of nature’s inventive­ The shapes created by nature often inspire our creativity
ness. Burrows and Sutton, in accelerator physics as well — e.g., the spiral-shaped Muon
2013. Reproduced with per­ collider cooling channel (consisting of an integrated helical
mission. solenoid and accelerating cavities interleaved with absorbers;
see Fig. 1.2) was possibly inspired by the double-helical DNA.
We hope that examples like this, together with a rigorous in­
ventiveness methodology described in this book will arm the
reader with a new systematic approach that will enable effi­
cient inventiveness.
FIGURE 1.2
Helical solenoid channel.
1.2 Acceleration of what and how
Accelerators can be either giant or tiny, but all have similar­
ities and the same subsystems. A giant accelerator, such as
SLAC’s linear accelerator and a cathode ray tube TV (which is
also an accelerator, albeit smaller), both have all of the main
components of a modern linear accelerator or collider. This
includes a source of charged particles, an acceleration area,
a drift region with focusing and steering, and a target or de­
tector (represented in the case of a TV by the phosphorous
screen).
When discussing acceleration, we assume that we accel­
erate a bunch of particles — a compact cloud of, for exam­
ple, electrons or positrons, protons or antiprotons, ions or any
other charged particles.
The simplest accelerating mechanism is electrostatic di­
rect acceleration — caused by DC voltage and a correspond­
ing electric field. Another method is betatron acceleration,
which is caused by a magnetic field changing in time which,
according to Maxwell’s equation E · dt = −d/dt B · dS, cre­
ates a curl of electric field E suitable for acceleration. The
third method is acceleration in an electro-magnetic wave;
however, one should note that an EM wave in free space can­
not continuously accelerate particles along the direction of its
FIGURE 1.3 propagation, as its E and B components are transverse to the
Basic principles of acceler­ direction of the EM wave’s propagation. Therefore, in order
ation — electrostatic, beta­ to use an EM wave for acceleration, one needs to change the
tron, in an EM wave in an ac­
1 M. Burrows and G. Sutton, Interacting Gears Synchronize Propulsive Leg
celerating structure.
Movements in a Jumping Insect, Science, 341, 13 Sep 2013.
 basics of accelerators and of the art of inventiveness 3

structure of its fields, which can be achieved by propagating

the EM wave in an appropriately shaped accelerating struc­
ture. These three methods are illustrated in Fig. 1.3.
Assuming we know how to accelerate the beam, we can
ask the question of why would we want to do that? That is,
how are we planning to use the accelerated beam? One can
foresee at least four different uses of the beam, as illustrated
in Fig. 1.4. We can direct the accelerated beam onto a tar­
get, either for scientific experiments (e.g., in nuclear physics)
or for modifying or treating the target itself. We could direct
two accelerated beams onto each other, as is typically done
for high energy physics experiments. Acceleration could also
be used to characterize the beam or perhaps to separate it into
different species or isotopes. Finally, we can use the acceler­
ated beam to generate useful radiation.

FIGURE 1.4
1.2.1 Uses, actions and the evolution of accelerators Uses of accelerated beams
Having discussed in the previous section why we need to ac­ — sending to target, collid­
celerate the beam and how an accelerated beam can be used, ing with another beam, char­
we will now take this moment to define the basic actions acterization of the beam or
that can be applied to the beam: acceleration, focusing and separation into species, gen­
cooling, and the generation of radiation, as well as the corre­ eration of useful radiation.
sponding parameters and characteristics of these actions.
In cases of acceleration we aim to find out the final energy
of the particles and usually prefer to achieve as high a rate as
possible of the energy change (usually called the accelerating
gradient). If the electrostatic accelerating voltage is U0 then
the final energy is E = γmc2 equal to E = eU0 + mc2 , where γ
is relativistic factor, γ = 1 + eU0 /(mc 2 ), m is the rest mass of
the particle, e is its charge and c is speed of light.
Whether we plan to send the beam to a target or collide it
with another beam, we strive to achieve a certain flux of parti­
cles; we therefore may need to focus the beam to a small size
on the target or at the interaction point with the oncoming
beam. As it is with light, a sequence of focusing and defo­
cusing lenses (in this case electromagnetic lenses) focus the
beams, as is illustrated in Fig. 1.5.
Using lenses to focus the beam does not affect its so-called
phase-space volume, which is usually called emittance ε and
defined as an area of the ellipse occupied by the beam in
coordinate-angle phase space (for example, x and x' as illus­
trated in Fig. 1.5). If the two transverse planes are indepen­
dent (not coupled), then both εx and εy emittances are con­
served. Emittance ε is usually defined in units of m · rad or
mm · mrad. If the beam is accelerated, the so-called normal­
ized emittance εn = γε is conserved. If the beam emittance is
large (which can especially be true for positrons or antipro­
tons, which are created in “hot” collisions of the initial beam
with the target) it can be particularly difficult to focus such
4 unifying physics of accelerators, lasers and plasma

a beam to a small size. Therefore, the next important action

applied to the beam is cooling, which is intended to reduce
the emittance of the beam.
In the context of the generation of radiation, we concern
ourselves with reducing both the sizes of the beam in the
emitting region, and its angular spread (as emitted radiation
usually follows the direction of the particles). Therefore, low
emittance and beam cooling may again be a necessity. The
corresponding characteristic, which describes the radiation
generation, is called brightness and is defined as the number
of photons emitted per unit of time from a certain area into a
certain solid angle (see Fig. 1.5). One also typically adds “into
a certain bandwidth” (such as, for example, 1% of the cen­
tral wavelength) as one is usually interested only in a certain
spectral range of the emitted radiation. Brightness is there­
fore defined in the units of number of photons/(s · m2 rad 2 (% ·
bandwidth)).
Finally, when two beams are colliding, we aim at focusing
the beams into the smallest possible size at the interaction
point, in order to maximize the probability of interaction.
In this case, we are therefore interested in the characteris­
tic called luminosity L, which is defined in such a way that
the product of luminosity and the cross section of interac­
tion σ (which has the units of m2 ) gives the number of events
per unit of time. The luminosity is then defined in units of
FIGURE 1.5
1/(s · m2 ).
Actions on accelerated
These basic actions or manipulations that can be applied
beams — acceleration,
to the beam help to define the evolution of accelerators.
focusing, generation of
Scientific and technological advances in the area of accel­
radiation, colliding.
erators have focused on an increase of energy of accelerated
beams, mastering acceleration techniques (including the ac­
celeration of different species) and on an increase of acceler­
ating gradients. The need for smaller beam sizes facilitated
the development of beam focusing and cooling methods. De­
mands for higher brightness stimulated mastery of the meth­
ods of radiation generation. Desires to increase luminosity
of colliding beams led to improvements of a whole class of
techniques, from emittance preservation to stabilization of
nanometer beams. Lastly, one of the biggest motivations for
accelerator progression was their potential to be applied in
various scientific and technological areas.

1.2.2 Livingston plot and competition of technologies

The history of accelerators and various accelerator technolo­
gies can be summarized in a so-called “Livingston plot”
where the equivalent energy of an accelerated beam is plotted
against time — see Fig. 1.6. One can clearly see that, over the
course of many decades, the maximum achieved energy grew
exponentially. It was the development of different accelerator
technologies that enabled this exponential growth.
 basics of accelerators and of the art of inventiveness 5

H9

H9
/+&
7HYDWURQ
H9
Equivalent energy of fixed target accelerator

7H9

7H9 6\QFKURWURQV
FROOLGHUV
7H9

*H9 HHFROOLGHUV
%HWDWURQ
*H9

*H9 /LQDFV

0H9 &\FORWURQV

0H9 *HQHUDWRUV

0H9

.H9
         

FIGURE 1.6
Livingston plot of evolution of accelerators.

The Livingston plot also depicts that new accelerating

technologies replaced each other once the previous technol­
ogy had reached its full potential. This evolution and satura­
tion of particular acceleration technologies, and birth of the
new technologies, is a common phenomenon in any techno­
logical or scientific field — illustrated in Fig. 1.7.
The three most recent decades shown in Fig. 1.6, repre­
sented by Tevatron and LHC colliders, exhibit a much slower
exponential energy growth over time. This may be an indi­
cation that the existing technologies of acceleration came to
their maximum potential, and that further progress would
demand creation of new accelerating methods — one which FIGURE 1.7
was more compact and more economical. There are several Evolution of technologies —
emerging acceleration techniques, such as laser-driven and saturation and replacement
beam-driven plasma acceleration, which can bring the Liv­ by newer technologies.
ingston plot back to the exponential path — which we will
further discuss later on.
6 unifying physics of accelerators, lasers and plasma

1.3 Accelerators and inventions

Accelerator science exhibits a rich history of inventions. Let
us briefly skim through some of the most influential inven­
tions. Those interested in further reading are recommended
to explore the fascinating story of accelerators described in
Engines of Discovery.2
Between 1900 and 1925, radioactive source experiments
initiated by Rutherford created a demand for higher energy
beams.
From 1928 to 1932, Cockcroft and Walton developed elec­
trostatic acceleration using voltage multiplication created
FIGURE 1.8 out of diodes and oscillating voltage — ultimately reaching
Van der Graaf accelerator. around 700 kV of voltage. At about the same time, Van der
Graaf created a method of voltage charging wherein a rubber
mechanical belt would carry charges deposited onto the belt
by sharp needles via the ionization of gas molecules — which
achieved voltages at around 1.2 MV (see Fig. 1.8).
Resonant acceleration development commenced in 1928
with Ising establishing the concept and Wideroe building the
first linac.
In 1929, Livingston built his small prototype of the cy­
FIGURE 1.9 clotron as his PhD thesis, inspired by Lawrence, who studied
Cyclotron accelerator. Wideroe’s thesis (see Fig. 1.9). This concept was then realized
later in large scale by Lawrence.
In 1942, the principle of magnetic induction helped Kerst
build the first betatron.
In 1944, the synchrotron (see Fig. 1.10) was invented
by Oliphant, while MacMillan and Veksel independently in­
vented the principle of RF phase stability, which made longi­
tudinal focusing of beams possible.
In 1946, Alvarez built the proton linac by using an RF
FIGURE 1.10 structure with drift tubes (in progressive wave in 2π mode).
Synchrotron accelerator. In 1950, Christofilos developed strong focusing, which he
later patented as the alternate gradient concept (transverse
strong focusing). Fig. 1.11 illustrates the strong focusing as a
gutter with its edges bent up and down in a sine-like manner.
The story of the invention of strong focusing has an interest­
ing aspect — it is usually attributed to Courant and Snyder,
since the Christophilos patent was pointed out only after the
Cosmotron team had announced the idea. This example is
FIGURE 1.11 certainly relevant for anyone considering whether to publish
Strong focusing concept. or to patent their ideas.
In 1951, a tandem Van der Graaf accelerator was devel­
oped by Alvarez, thusly upgrading the electrostatic acceler­
ation concept. A charge-exchange stripping foil was placed
at the high voltage point and the source of negative ions was
placed at ground potential (which was also much more prac­
tical for its servicing). This resulted in a voltage twice as large.
2 Engines of Discovery, A Century of Particle Accelerators, A. Sessler and
E. Wilson, World Scientific, 2014.
 basics of accelerators and of the art of inventiveness 7

In 1953, Courant, Snyder, and Livingston built the weak

focusing 3.3 GeV Cosmotron in Brookhaven, and in 1957,
Veksler built a 10 GeV (which was the world record at that
time) synchrophasotron in Dubna, whose magnet weighed 

36,000 tons and was registered in the Guinness Book of 

World Records. This record in energy was overtaken by the 
CERN Proton Synchrotron in 1959, constructed under the  

  
leadership of Sir John Adams. The CERN PS was the first  
 
strong focusing accelerator, closely followed by the AGS at 
Brookhaven. 
 
In 1956, Veksler suggested the principle of collective ac­  

celeration (see Fig. 1.12), which became the predecessor to a
variety of collective methods based on plasma acceleration. FIGURE 1.12
In 1956, Kerst discussed the concept of colliding beams Collective acceleration.
and in 1961, the e+e- collider — the concept for a particle-
antiparticle collider — was invented by Touschek.
The collider concept created the need to develop the
methods used to decrease beam emittances, in particular for
antiparticles, and in 1967, Budker proposed electron cool­
ing as a way to increase the proton or antiproton beam den­
sity (see Fig. 1.13). Shortly thereafter, Van der Meer proposed
stochastic cooling (see Fig. 1.14) in 1968 as a way to compress
the beam’s phase space.
In 1970, RFQ was invented when Kapchinski and
FIGURE 1.13
Telyakov build the first radiofrequency quadrupole that al­
Electron cooling concept.
lowed a simultaneous focusing and acceleration of the beam.
This technology allowed a much more efficient and compact
acceleration of ions and protons from a very low energy to an
energy as high as a few MeV.
In 1971, Madey developed the principle of a Free Electron
Laser (FEL), a method that allowed the production of coher­
ent, hard X-rays of unprecedented brightness.
In 1979, Tajima and Dawson proposed acceleration of the
beam in plasma waves excited by a laser (see Fig. 1.15). It was
only much later that suitable lasers became available and this
method started to become competitive.
From around 1980, superconducting magnets were devel­
oped in various labs around the world, which allowed for a FIGURE 1.14
drastic increase of beam energy in circular acceleration. De­ Stochastic cooling concept.
velopment of this technology still continues today (8 Tesla
magnets are routinely used in LHC, and magnets exceeding
10 Tesla are being developed for LHC upgrade).
Likewise, the superconducting RF (radio-frequency) tech­
nology (created around 1980) is currently being used and is
still being developed today in many labs and industrial com­
panies, allowing an increase in the RF gradient and efficiency,
and thus the eventual energy and power of the beams.
The most recent decades have been rich with inventions as FIGURE 1.15
well. The years between 1990 and the present day have seen Plasma acceleration con­
a photon collider concept, an (experimentally verified) crab cept.
8 unifying physics of accelerators, lasers and plasma

waist collision and integrable optics for storage rings, just to

name a few.
The variety of the inventions (many of which will be dis­
cussed later on in further detail) demonstrates the past suc­
cesses of our field, but we are concerned with a different ques­
tion: what can we look forward to inventing in the future, and
how are we to invent it more efficiently?
In the next section we will introduce the methodical ap­
proaches to inventiveness used in industry, and will then ex­
plain how these methods can be applied to, in this case, ac­
celerator science.

1.4 How to invent

Methods described in this section are rarely known in sci­
ence, but they are widely used in industry.
In his March 7, 2013 contribution to Forbes,3 “What
Makes Samsung Such An Innovative Company?” the author,
Haydn Shaughnessy, wrote: “But it was ... that became the
Formal inventive approaches bedrock of innovation at Samsung. In 2003 ... led to 50 new
rarely known in science but patents for Samsung and in 2004 one project alone, a DVD
widely used in industry pick-up innovation, saved Samsung over $100 million. ... is
now an obligatory skill set if you want to advance within
Samsung.” The reader may speculate on what the dots sig­
nify — but read on.
What is this magic method, and what is the word inten­
tionally omitted in the above quote? The answer will be given
in just a few pages.

1.4.1 How to invent — evolution of the methods

Let’s start by recalling the techniques of problem solving,
starting from a well-known brute force or exhaustive search
method. In this case, any potential solutions and ideas are
considered and evaluated one by one. It is easy to imagine
how inefficient this method may be.
An improved variation of the exhaustive search method
is called brainstorming. Its author, Alex Osborn, introduced
this method in the 1950s. Brainstorming is a psychological
approach that helps to solve problems and create inventions
by separating the process of idea generation from the process
of critical analysis.
The method of brainstorming has its limitations — the ab­
sence of critical feedback, which is the main feature of the
method, is simultaneously its handicap, as feedback is re­
quired to develop and improve an idea.
The next method that emerged on the scene, the so-called
3 http://www.forbes.com/sites/haydnshaughnessy/2013/03/07/why-is­
samsung-such-an-innovative-company/
 basics of accelerators and of the art of inventiveness 9

synectics, was introduced by George Prince and William Gor­

don in an attempt to improve brainstorming. One of the main
feature of synectics is that it is assumed to be practiced by a
permanent group of experts, whose members, with time, be­
come less sensitive to critics from among their peers and thus
become more efficient in inventing. This method emphasized
the importance of seeing familiarity in an unknown and vice
versa, which helps to solve new and unfamiliar problems us­
ing known methods. The authors of synectics stressed the im­
portance of approaching a problem with a fresh gaze, and
also stressed the use of analogies to generate this attitude.
The analogies employed by synectics can be direct (any
analogy, e.g., from nature), emphatic (attempting to look at
the problem by identifying yourself with the object), sym­
bolic (finding a short symbolic description of the problem
and the object), or even metaphorical (describing the prob­
lem in terms of fairy-tales and legends).
Synectics is still a variation of the exhaustive search
method and is in fact the limit of what can be achieved, main­
taining the brute force method of an exhaustive search.
There must, then, be a better inventive approach free from
the irrationality of synectics. Indeed, why would one employ
analogies and metaphors and other irrational factors in order
to come to a formula — the action has to happen itself — a nat­
ural and universal inventive formula of an ideal invention?
Such a formula (which means that the action needed to solve
the problem should happen by itself, i.e., without introduc­
ing any additional systems) should indeed be programmed
into the process of any inventive solution, aimed at an ap­
propriately selected part of the object, and with precise iden­
tification of the physical contradiction and intended physi­
cal action. These features are the characteristics of TRIZ, the
powerful inventive method described in the next section.

1.5 TRIZ method

The abbreviation TRIZ (pronounced [treez]) can be translated
as the Theory of Inventive Problem Solving. TRIZ was devel­
oped by Genrikh Altshuller in the Soviet Union in the mid­
20th century. The author, while working in the patent office
since 1946, analyzed many thousands of patents, trying to
discover patterns to identify what makes a patent successful.
Between 1956 and 1985, following his work in the patent The creator of TRIZ G.S.
office, Altshuller formulated TRIZ and, together with his Altshuller (October 15,
1926 – September 24,
team of supporters, developed it further.
1998). Published with per­
Altshuller and his team devised four key constructs of the mission of the Official G.S.
TRIZ methodology: Altshuller Foundation.
• The same problems and solutions appear again and again

but in different industries.

10 unifying physics of accelerators, lasers and plasma

• There is a recognizable technological evolution path for

all industries.

• Innovative patents (about a quarter of the total) used sci­

ence and engineering theories outside their own area or
industry.

• An innovative patent uncovers and solves contradictions.

Together with these main clauses, the authors created a
detailed methodology, which we will review in the next sec­
tion before making the connection to physics.
And as the reader has likely already guessed, it was the
word “TRIZ” that was skipped in the quote at the beginning
of the previous section. It was TRIZ that became the bedrock
of innovation at Samsung, and which can do so much more if
applied to other existing fields such as accelerator science.

1.5.1 TRIZ in action — examples

Let us consider the TRIZ approach with an example that is
often cited in TRIZ textbooks. Imagine that we need to polish
an optical lens with an abrasive stick (see Fig. 1.16) and we
need to do it quickly. However, there is a problem: polishing
the lens generates heat, which degrades the optical proper­
ties of the produced lens. Existing cooling methods are inef­
fective, as one cannot supply adequate and uniform cooling
to each abrasive particle.

FIGURE 1.16
Illustration of TRIZ in action — initial specific problem.

The first step of the TRIZ approach consists of identifying

the pair of contradicting parameters: the one that needs to be
improved and the one that simultaneously gets worse. In the
considered case of the lens polishing, these parameters are
the speed (the one to be improved) and the temperature (the
one that gets worse).
The second step of the TRIZ algorithm starts by identi­
fying if anyone else has solved such a contradiction in the
past. And it is here that the power of TRIZ lies — hundreds
of thousands of analyzed patents allowed the TRIZ team to
put together a table, the contradiction matrix, which helps to
find relevant inventive principles corresponding to a particu­
lar pair of contradicting parameters.
 basics of accelerators and of the art of inventiveness 11

The beauty of the TRIZ method is also illustrated by the

fact that the entire field of engineering can be described by
the TRIZ contradiction matrix using only 39 parameters, and
that the number of TRIZ inventive principles is also remark­
ably small — only 40.
The second step of the TRIZ algorithm involves checking4
the TRIZ matrix for this pair of contradicting parameters and
then identifying the relevant, in this case, inventive princi­
ples.
Having identified the necessary inventive principles, we
are ready for the third and final step of the TRIZ algorithm
— translating a generic inventive principle into a specific so­
lution. Graphically, the flow of the TRIZ algorithm is shown
in Fig. 1.17.

1) Define the 2) Use contradiction 3) Translate

problem in terms of matrix to obtain generic inventive
generic contradiction relevant inventive principles into
parameters principles specific solution

TRIZ Generic TRIZ Generic

Problem Solution

Specific Specific
Problem Solution

FIGURE 1.17
Illustration of the flow of the TRIZ algorithm.

In our particular example of the lens that needs to be pol­

ished, the contradiction matrix zoomed into the crossing of
the speed parameter, which needs to improve without dam­
aging the temperature parameter, is pictured in Table 1.1.
The inventive principles listed (according to their numer­
ical index) in the speed–temperature cell of the contradic­
tion matrix are: 2-Taking out; 28-Mechanics substitution; 30­
Flexible shells and thin films; 36-Phase transitions.
The next step is to select the inventive principle that is
most suitable which in our example can be judged to be the
latter one — 36-Phase transition (use of phenomena occur­
ring during phase transitions, such as volume changes, loss
or absorption of heat, etc., according to the description of this
TRIZ principle).
The corresponding specific solution to the lens polish­
4 See, e.g., http://www.triz40.com/
12 unifying physics of accelerators, lasers and plasma

TABLE 1.1
TRIZ contradiction matrix for speed–temperature parameters and
indexes of the corresponding inventive principles
Improving parameter Parameter that deteriorates
... 9.Speed ... 17.Temperature ...
... ... ... ... ... ...
9.Speed ... ... ... 2,28,30,36 ...
... ... ... ... ... ...
17.Temperature ... ... ... ... ...
... ... ... ... ... ...

ing dilemma (which can be suggested according to the se­

lected inventive principle of phase transition) is to use ice
together with abrasive particles, which provides efficient uni­
form cooling, as illustrated in Fig. 1.18.

FIGURE 1.18
Illustration of TRIZ in action — specific solution.

Before we discuss the connection between TRIZ and sci­

ence, allow us to present a complete list of the standard TRIZ
contradicting parameters (shown in Table 1.2) as well as the
list of inventive principles (Table 1.3). A detailed descrip­
tion of the TRIZ principles is also available at the web page
mentioned in footnote 4 of this chapter.

TABLE 1.2
Elements of TRIZ contradiction matrix
No.Parameter No.Parameter No.Parameter
1. Weight of moving obj. 14. Strength 27. Reliability
2. Weight of stat. obj. 15. Durability of mov. obj. 28. Measurement accuracy
3. Length of moving obj. 16. Durability of stat. obj. 29. Manufacturing precision
4. Length of stat. obj. 17. Temperature 30. Object-affected harmful
5. Area of moving object 18. Illumination intensity 31. Object-generated harmful
6. Area of stationary obj. 19. Energy use by mov. obj. 32. Ease of manufacture
7. Volume of moving obj. 20. Energy use by stat. obj. 33. Ease of operation
8. Volume of stat. obj. 21. Power 34. Ease of repair
9. Speed 22. Loss of energy 35. Adaptability or versatility
10. Force (Intensity) 23. Loss of substance 36. Device complexity
11. Stress or pressure 24. Loss of information 37. Difficulty of detecting
12. Shape 25. Loss of time 38. Extent of automation
13. Stability of the object 26. Quantity of substance 39. Productivity
 basics of accelerators and of the art of inventiveness 13

TABLE 1.3 TRIZ inventive principles

No.Principle No.Principle No.Principle
1. Segmentation 15. Dynamics 29. Pneumatics, hydraulics
2. Taking out 16. Partial or excessive actions 30. Flexible shells, thin films
3. Local quality 17. Another dimension 31. Porous materials
4. Asymmetry 18. Mechanical vibration 32. Color changes
5. Merging 19. Periodic action 33. Homogeneity
6. Universality 20. Continuity of useful action 34. Discarding, recovering
7. Russian dolls 21. Skipping 35. Parameter changes
8. Anti-weight 22. Blessing in disguise 36. Phase transitions
9. Preliminary anti-action 23. Feedback 37. Thermal expansion
10. Preliminary action 24. Intermediary 38. Strong oxidants
11. Beforehand cushioning 25. Self-service 39. Inert atmosphere
12. Equipotentiality 26. Copying 40. Composite materials
13. “The other way round” 27. Cheap short-lived objects
14. Spheroidality — Curvature 28. Mechanics substitution

1.6 TRIZ method for science

We have finally arrived at the section that will elucidate the
meaning of the epigraph to this chapter.
The TRIZ inventive principle of Russian dolls (nested
dolls, or matreshka) can be applied not only to engineering,
but to many other areas, including science. A rather spectac-
FIGURE 1.19
ular example is the construction of a high energy physics de-
Valery Bryusov’s electron as
tector, where many different sub-detectors are inserted into
an analogy to the TRIZ inven­
one another, like a nested doll, in order to enhance the accu­
tive principle of nested dolls.
racy of detecting elusive particles (see Fig. 1.20).

FIGURE 1.20
High energy physics detectors, which have a layered “nested”
structure, reflecting the TRIZ inventive principle of Russian dolls.
14 unifying physics of accelerators, lasers and plasma

The reader can now see that the 1920 poem by Valery
Bryusov, which describes an electron as a planet in its own
world (see Fig. 1.19), can also be seen as reflection of the
nested doll inventive principle, this time in poetic science fic­
tion.

FIGURE 1.21
Particle interaction event observed in a cloud chamber invented
by Wilson in 1911 (left), and in a bubble chamber invented by
Glaser in 1952 (right).

TRIZ textbooks also often cite Wilson’s cloud chamber

(invented in 1911) and Glaser’s bubble chamber (invented
in 1952) as examples, in the terminology of TRIZ, of a sys­
tem and anti-system. Indeed, the cloud chamber works on
the principle of bubbles of liquid created in gas, whereas the
bubble chamber uses bubbles of gas created in liquid (see
Fig. 1.21).
It is thus reasonable to ask: would it have taken almost
half a century to invent the bubble chamber had the TRIZ
anti-system principle been applied?
The systematic application of TRIZ to science is indeed a
valid question, and moreover, it can give us new insights. We
will now discuss TRIZ in connection with accelerator science.

1.7 AS-TRIZ
The TRIZ method was originally created for engineers. How­
ever, the methodology is universal and, as the previous sec­
tion demonstrated, can largely be applied to science, and in
particular to accelerator science.
Many of the parameters from the TRIZ contradiction ma-
AS-TRIZ is currently be- trix, as well as TRIZ inventive principles, can be directly ap­
ing developed. More exam- plied to problems arising in accelerator science. Still, accel­
ples will appear in the fol- erator science is a distinct discipline, and it is only natural
lowing chapters. to add accelerator science-related parameters and inventive
 basics of accelerators and of the art of inventiveness 15

principles to TRIZ. We call this extension Accelerating Science

TRIZ or AS-TRIZ, highlighting via its name a wide applica­

bility of the method to various areas of science, even those

beyond the field of accelerators.

Below are just a few suggested additional parameters for

the AS-TRIZ matrix of contradictions (which are based on

manipulated beams and corresponding characteristics dis­

cussed in Section 1.2.1), and just a couple of inventive prin­
ciples to start with — see Table 1.4 and Table 1.5. We will

populate this table as we move along the chapters.

TABLE 1.4 TABLE 1.5

Emerging AS-TRIZ parameters Emerging AS-TRIZ principles
No. Parameter No. Principle
1. Energy 1. ...
2. Rate of energy change 2. ...
3. Emittance 3. Undamageable or already damaged
4. Luminosity 4. Volume-to-surface ratio
5. Brightness 5. ...
6. Intensity ... ...
7. Efficiency 21. ...
8. Power
9. Integrity of materials
10. Time duration or length
11. Spatial extent
12. Sensitivity to imperfections
13. Cooling rate
14. ...

... ...

21. ...

The first inventive principle shown in Table 1.5, num­

ber 3: “Undamageable or already damaged,” is illustrated by

the three examples below, while 4, “Volume-to-surface ra­

tio,” naturally arises, for example, from Maxwell’s equation,

where an integral on a surface is connected to the integral

over volume EdS ∝ ρdV , or from other similar equations,

e.g., from thermodynamics. We will discuss examples corre­

sponding to this principle later on in the book.

Let us illustrate an application of AS-TRIZ to a couple of

accelerator science examples.

The first example is the beam profile monitor with its

tungsten or carbon wire, shown in Fig. 1.22. In order to mea­

sure the beam’s profile, the wired frame needs to cross the

beam very quickly. The problem is that, as beam intensity in­

creases, the beam and energy losses in the wire also increase,

and the wire can get damaged after a single use. The physical

contradiction in this case is between the parameters intensity

FIGURE 1.22
(to be improved) and integrity (what degrades). Carbon wire beam profile

In order to solve the beam profile monitor problem, let monitor.

16 unifying physics of accelerators, lasers and plasma

us apply emerging AS-TRIZ principle 3: “Undamageable or

already damaged.” That is, we must replace the material that
can be damaged with another media which either cannot be
damaged (light) or is already “damaged” (e.g., plasma).
Selecting light as a medium that cannot be damaged nat­
urally arrives at an already well-known solution — a beam
profile monitor based on laser wire, shown in Fig. 1.23.

+LJKSRZHUODVHU

%HDPVSOLWWHU

'HIOHFWRU
\ 0LUURUV

[ ]

&RPSWRQVFDWWHUHG V

(OHFWURQEXQFK

/DVHUEHDP
UD\GHWHFWRU
%HQGLQJPDJQHW

(OHFWURQEHDPWUDMHFWRU\

In the laser wire the bunch

interacts with optical light FIGURE 1.23
that acts like the physical Laser wire beam profile monitor.
wire of the mechanical beam
profile monitor. The next example is a standard glass mirror. The prob­
lem here is that, as the intensity of the laser increases, the
mirror can get damaged. The contradiction is again between
intensity and integrity. Let us apply the same principle 3 of
AS-TRIZ, and replace the mirror with something that is al­
ready damaged — plasma. Indeed, such a solution is already
known in laser and plasma science — the so-called plasma
mirror. A plasma mirror is created by focusing a laser beam
onto a piece of glass or gas. It will then be transmitted un­
til the intensity reaches a certain value sufficient to produce
plasma on the surface. At that moment, the reflective index
of the surface changes, and the laser beam is then reflected
from the plasma’s free electrons.
The final example for this chapter, one that we will return
to for further details later on, is the dilemma of accelerat­
ing structures. Made from metal, the acceleration structures
(either normal conducting or super-conducting) are suscepti-
 basics of accelerators and of the art of inventiveness 17

ble to breakdowns that limit the useful accelerating gradient,

typically to about 100 MeV/m.
The problem of an accelerating cavity can therefore be for­
mulated as follows: as the rate of energy E change (accelerat­
ing gradient) increases, the surface of cavities gets damaged
with occasional breakdowns. The contradiction is therefore
between rate of E change (to be improved) and integrity (what
degrades).
Applying again principle 3 of AS-TRIZ, and replacing ac­
celerating structure with plasma, we come to the well-known
concept of plasma acceleration, when an “accelerating struc­
ture” is temporarily produced in plasma by a driving laser
pulse.
By taking note of these three examples, we already can
start to fill in the AS-TRIZ contradiction matrix with inven­
tive principles — see Table.1.6.

TABLE 1.6
Emerging AS-TRIZ contradiction matrix and indexes
of inventive principles
Improving parameter Parameter that deteriorates
... ... 9.Integrity ...
... ... ... ... ...
2.Rate of E change ... ... 3,... ...
... ... ... ... ...
6.Intensity ... ... 3,... ...
... ... ... ... ...

Many more examples related to TRIZ and AS-TRIZ will

be discussed in the following chapters. However, the style of
the book from this point on will change — the main text will
follow the standard style of scientific textbooks, while the oc­
casional notes (primarily in the margins) will point out rele­
vant connections to TRIZ. We will also use TRIZ in order to
highlight similarities between the three discussed subjects —
physics of accelerators, lasers and plasma.
These two parallel narratives, about science and about in­
ventions, will eventually merge again in the final chapter,
where we will summarize our discussions about the art of
inventiveness and accelerator science.

1.8 Creativity
Use of TRIZ does not take away your creativity! It instead
strives to encourage and inspire you. FIGURE 1.24
It helps to find potential solutions, narrowing down the Looking at the world through
choices, but does not find the final answer for you. By arming the prism of TRIZ. Illustration
you with a considerably deep breadth of knowledge, TRIZ by Sasha Seraia.
18 unifying physics of accelerators, lasers and plasma

FIGURE 1.25
TRIZ vs. brainstorming. Illustration by Sasha Seraia.

can help you to be successful if you choose to apply its meth­

ods (see Fig. 1.25 and Fig. 1.24).
Moreover, use of TRIZ will help us to build bridges be­
tween accelerators, lasers and plasma, and indeed even pos­
sibly other currently unrelated sciences, technology or engi­
neering. Remember, one of the most important principles of
TRIZ is that “the same problems and solutions appear again
and again but in different disciplines.”
 basics of accelerators and of the art of inventiveness 19

EXERCISES
1.1 Analyze the evolution of technical or scientific systems.
Discuss the evolution of any scientific or technical (or
accelerator-science-related) area, identifying successive tech­
nologies, which arise, saturate and get replaced by new ap­
proaches and solutions.
1.2 Analyze inventions or discoveries using TRIZ and AS-TRIZ.
A plasma mirror is often used in the situation when a stan­
dard metal mirror cannot withstand the power density of the
laser. Analyze this technology in terms of the TRIZ and AS­
TRIZ approach, identifying a contradiction and a general in­
ventive principle that were used (could have been used) in
this invention.
1.3 Analyze inventions or discoveries using TRIZ and AS-TRIZ.
Analyze and describe any scientific or technical (possibly, re­
lated to accelerator science) invention/discovery in terms of
the TRIZ and AS-TRIZ approaches, identifying a contradic­
tion and an inventive principle that were used (could have
been used) in this invention or discovery.
1.4 Developing AS-TRIZ parameters and inventive principles.
Based on what you already know about accelerator science,
discuss and suggest the possible additional parameters for
the AS-TRIZ contradiction matrix, as well as the possible ad­
ditional AS-TRIZ inventive principles.
 2
Transverse Dynamics

2.1 Maxwell equations Let’s begin our discourse about the basics of accelerator
and units 21 physics with the topic of the transverse dynamics of charged
2.2 Simplest particles. This will lead into a discussion of the basics of syn­
accelerator 22 chrotron radiation and of acceleration in the following chap­
2.3 Equations of ters, intermediated by a dialogue on the synergies between
motion 24 accelerators, lasers and plasma.
2.4 Matrix formalism
27
2.5 Phase space 35
2.6 Dispersion and 2.1 Maxwell equations and units
tunes 36 We start by recalling the Maxwell equations with an empha­
2.7 Aberrations and sis on their systems of units, focusing in particular on SI and
coupling 38 Gaussian-cgs systems. The microscopic Maxwell equations
(i.e., equations in vacuum) expressed in SI units, in both dif­
ferential and integral form, are
This Maxwell equation ex­
plains the universality of the
inventive principle of chang- ρ 1
∇·E = or E · dS = ρdV (2.1)
ing the volume-to-surface ra- ε0 ∂Ω ε0 Ω
tio.
∇ · B = 0 or B · dS = 0 (2.2)
∂Ω

Permittivity of free space

ε0 ≈ 8.85 F/m or A2 s 4 kg −1 m−3
∂B d
∇×E = − or E · dt = − B · dS (2.3)
∂t ∂Σ dt Σ
Vacuum permeability μ0 =
1/(c2 ε0 ) ≈ 1.26 × 10−6 N ·
A−2 ( )
∂E
∇ × B = μ0 J + ε0 or
∂t
d
B · dt = μ0 J · dS + μ0 ε0 E · dS (2.4)
∂Σ Σ dt Σ
Speed of light in vacuum c ≈
2.99 × 108 m/s The Lorentz force acting on a charged particle in an electric
and magnetic field in SI units is expressed as

F = q (E + v × B) (2.5)

While SI is the standard, the Gaussian system is more natu­

ral for electromagnetism. The differential Maxwell equations
expressed in Gaussian-cgs units are

DOI: 10.1201/b18696-2 21
 22 unifying physics of accelerators, lasers and plasma

∇ · E = 4πρ , ∇ · B = 0
( )
1 ∂B 1 ∂E
∇×E = − , ∇×B = 4πJ + (2.6)
c ∂t c ∂t
Charge of electron in SI:
≈ 1.6 × 10−19 C or A · s And the Lorentz force is
v
F = q E+ ×B (2.7)
c
Charge of electron in Gaus­
sian units: ≈ 4.8 × 10−10 cgs The above equation highlights why the Gaussian system is
units so useful: the electric and magnetic fields are expressed in
the same units, which stresses that these fields have a similar
nature.
Deriving a formula, instead Throughout this text we will use both SI and Gaussian
of using quantities such as e units. We will, however, plan to construct equations or at
or h, express the end result least the end results in such a way that allows for easy conver­
via more natural quantities sion between different units. As a result, we will avoid hav­
(me c 2 , re , λe , α, etc.). ing quantities such as electric charges or Planck’s constant in
our equations, and instead will use only quantities of length,
speed and energy.
Let’s write down some of the equations that will be useful
for such conversions into the natural and unit-independent
form.
The most useful one is for the classical radius of an elec­
tron:
1 e2 e2
SI : re = , Gauss : re = (2.8)
4πε0 me c2 me c 2
re ≈ 2.82 · 10 m −15

And the one for the fine structure constant:

e2 e2
SI : α = , Gauss : α = (2.9)
(4πε0 ) nc nc
α ≈ 1/137
The latter also gives us the reduced Compton wavelength for
the electron (λ̄e = re /α ≈ 3.86 · 10−13 m), which is very use­
ful whenever the Planck’s constant needs to be hidden in an
equation.

2.2 Simplest accelerator

An example of a simple accelerator is a thermionic gun, in
which thermionic emission from the cathode generates elec­
trons that are accelerated across a high voltage gap to the an­
ode. A variety of materials can be used for these cathodes.
FIGURE 2.1 Pure metals are the most robust choice but require extremely
Simple electron gun. high temperatures. On the other hand, the oxides cathodes
are covered with alkaline earth metal oxides (e.g., the BaO
impregnated tungsten cathode), and provide a higher emis­
sion at lower temperatures. However, they are more delicate

DOI: 10.1201/b18696-1
 transverse dynamics 23

as they require a better vacuum for their operation and are

vulnerable to poisoning by heavier elements present in the
residual gas.
In thermionic guns, a grid can be installed between the
anode and cathode. Applying pulsed voltage to the grid al­
lows for the generation of a train of pulses suitable for conse­
quent RF acceleration.
In an electron gun, the electrons generated by thermionic
emission tend to repel from each other as illustrated in
Fig. 2.1, resulting in reduced beam quality. The additional
focusing electrodes can help to maintain the quality of the
beam. In particular, it was shown by J.R. Pierce in 1954 that
in planar geometry, an electrode at the potential of the cath­
ode inclined at 67.5◦ (called the Pierce angle), as shown in
Fig. 2.2, will help to maintain a parallel flow of electrons.
Any intermediate accelerating electrodes or the anode would
need to be placed along the equipotential lines, as shown in
Fig. 2.2, to maintain the parallel flow.

FIGURE 2.2
Electron gun with Pierce electrode and collector made in the
form of a Faraday cup.

The same Fig. 2.2 also illustrates the concept of the beam
anode (collector) made in the shape of a Faraday cup — a use­
ful device for the accurate measurement of electron current.
Any secondary charged particles emitted from the walls are
eventually absorbed and do not affect the measurements of
the current.
Controlling the beam shape and beam quality in high
density electron guns often requires the use of an accompa­
nying solenoid magnetic field. Let us now consider equations
for the motion of charged particles in electromagnetic fields.
24 unifying physics of accelerators, lasers and plasma

2.3 Equations of motion

2.3.1 Motion of charged particles in EM fields
The motion of a particle with charge q in an electric E and
magnetic B fields is given by the following equations:

dp dE
= q(E + v × B) , = F·v (2.10)
dt dt
where the momentum p and energy E of the particle are

p = m0 γv , E = m0 γc2

and m0 is the particle mass, and γ is the relativistic factor.

Let’s consider the case of the uniform magnetic field when
the equation of motion simplifies to

dγv
m0 = qv × B or m0 γ v̇x = q vy B and m0 γ v̇y = −q vx B
dt
which then can be rewritten as
( )2
qB qB
v̈x = v̇ = − vx
m0 γ y m0 γ

which has a solution of

v0
vx = v0 cos(ωt) or x = sin(ωt)
ω
where
qB
ω= (2.11)
m0 γ
This solution describes motion (see Fig. 2.3) with a radius of
FIGURE 2.3 v0 v0 m0 γ
Motion of charged particles ρ= = (2.12)
ω qB
in a uniform magnetic field.
Let’s rewrite this radius (called the Larmor radius) in both
systems of units:
p pc
SI : ρ = Gaussian : ρ = (2.13)
qB qB

This kind of motion is observed, for example, in dipoles —

magnets intended primarily for bending the trajectories of
Magnetic rigidity Bρ [T esla · charged particles.
m] ≈ 3.3356 p [GeV /c] A quantity called magnetic rigidity Bρ is often used to de­
scribe motion in magnetic fields. It is defined as
p pc
SI : Bρ = Gaussian : Bρ = (2.14)
q q

and for a particle with the elementary charge and momentum

p given in GeV /c is equal to Bρ[T esla · m] ≈ 3.3356 p[GeV /c]
or Bρ[kGs · cm] ≈ 3335.6 p[GeV /c].
 transverse dynamics 25

2.3.2 Drift in crossed E × B fields

(
While we are on this topic, let’s consider the case of uniform
E and B fields that are perpendicular — a situation often en­
countered when dealing with plasma and beams.
%
Qualitatively, if a particle is initially at rest in these
crossed fields, it is initially pulled by the electric field, and
then, due to emerging velocity, the magnetic field turns it
around. When the direction of the particle’s motion reverses,
the electric field slows it down and eventually stops the par­
ticle some distance away from its initial position. After that,
the aforementioned motion begins anew. The resulting tra­ Y  Y!
jectory of the particle resembles Fig. 2.4. As a result, these
equations of motion predict a particle drift with constant ve­ FIGURE 2.4
locity, which is perpendicular to both E and B and with its Drift in crossed E×B fields.
value given by
E×B E×B
SI : vd = Gaussian : vd = c 2 (2.15)
B2 B
The efficiency of the Gaussian system of units continues to
astound us! Not only does it give us an intuitively clear and
beautiful formula, but it also immediately shows that the
above equation is valid only in the assumption that the elec­
tric field is much smaller than the magnetic field E « B.

2.3.3 Motion in quadrupole fields \

%\
Let’s consider motion in a quadrupole magnet (see Fig. 2.5) %[
1 6
where, ideally, the fields depend linearly on the distance from
the center of the magnet:
[
Bx = Gz and Bz = Gx
where G is the gradient of the quadrupole (and we use z ≡ y
6 1
in this section). We will rewrite the equation of motion as
\
dγ v )\
m0 = qv×B )[
dt 1 6
and using Cartesian coordinates:
[
q q
ẍ = − ˙ ,
G sx z̈ = ˙
Gsz
m0 γ m0 γ
6 1
q
and s̈ = ˙ − zz
G(xx ˙ )
m0 γ
FIGURE 2.5
Changing the independent variable from time to path length,
Magnetic fields and forces
and considering only the case of small deviations from the
acting on a particle in a
axis, reduces these equations to
quadrupole.
x'' − Kx = 0 and z'' + Kz = 0 (2.16)
q ∂Bz q
where K= = G (2.17)
p ∂x p
This brings us to the discussion of linear betatron motion.
26 unifying physics of accelerators, lasers and plasma

2.3.4 Linear betatron equations of motion

The particle trajectory in an accelerator is initially defined
by dipole magnets, therefore a curvilinear coordinate system is
y best used to describe the motion of the particles; see Fig. 2.6.
A reference orbit is usually selected, corresponding to an ideal
particle, which typically has a nominal energy and zero trans­
verse offsets and angles.
The focusing elements, quadrupoles and higher-order el­
x s ements are placed in space so that their centers correspond to
the reference orbit.
FIGURE 2.6 The motion of a charged particle with respect to the refer­
Frenet–Serret curvilinear co­ ence orbit and along the curvilinear abscissa (s in Fig. 2.6), and
ordinate system. influenced by the the magnetic fields of dipole magnets and
quadrupole magnets, is given by the linear Hill’s equations
d 2y
+ Ky (s)y = 0 (2.18)
ds 2
where transverse coordinate y stands for either a horizontal
or vertical axis (y = x, z).
Let’s write these equations down for the horizontal
1 1 ∂Bz (s)
Kx (s ) = − (2.19)
ρ 2 (s) Bρ ∂x
and the vertical
1 ∂Bz (s)
Kz (s) = (2.20)
Bρ ∂x
planes. We see that the equation almost exactly resembles
those derived in the previous section (see Eq. 2.16 and
Eq. 2.17) except for an additional term 1/ρ2 (s), which corre­
FIGURE 2.7
sponds to a weak focusing of a dipole.
Shifted circles cross.
The origin of the term corresponding to the weak focusing
in a dipole can be illustrated by the following example. Con­
sider the shifted circles in Fig. 2.7. They cross in two points
(we will ignore second-order effects). Translating this exam­
ple of shifted circles to a dipole magnet whose field fills a
half plane as shown in Fig. 2.8, we conclude by observation
% that the trajectories of the particles in this dipole exhibit an
equivalent “focusing” with the wavelength of motion (along
FIGURE 2.8 the curvilinear coordinate s) given by 2πρ, corresponding to
Illustration of the origin of x = x0 sin (s/ρ)
weak focusing in dipoles.
or to the following equation
d 2x x
+ =0 (2.21)
ds2 ρ 2
which has the same term 1/ρ 2 (s) as the Hill’s equations above.
The above derivations generally assumed no periodicity;
however, in a circular machine, Kx , Kz and ρ are periodic.
These are linear equations and can be integrated, which will
be discussed in the next section.
 transverse dynamics 27

2.4 Matrix formalism

2.4.1 Pseudo-harmonic oscillations
Let’s look for the solution of the Hill’s Eq. 2.18 in following
form ) [ ]
y(s) = εy βy (s) cos φ y (s) − φ (2.22)
where
s
ds '
φ y (s) = (2.23)
βy (s ' )
s0

which describes pseudo-harmonic oscillations. Here the beta

functions β (in x and z) are proportional to the square of the
envelope of the oscillations. The functions φ(s) (also in x and Beta functions β are propor­
z) describe the phase of the oscillations. tional to the square of the en-
We can find the differential equation for the beta func- velope of the oscillations
tions, by substituting the form Eq. 2.22 into the Hill’s equa­
tion. We use the following equation first
β ' (s) ε .

y ' (s) = cos(φ(s) − φ) − φ ' (s) εβ(s) sin(φ(s) − φ)

2 β(s)
and the second derivatives
⎡ ⎤
''
''
⎢⎢⎢ β (s) β ' 2 (s) . '2 ⎥ ⎥⎥ √
y (s) = ⎣⎢ . − − β(s)φ (s)⎥⎦ ε cos(φ(s) − φ)−
2 β(s) 4β 3/2 (s)
⎡ ⎤
⎢⎢ '' . β ' (s)φ ' (s) ⎥ √
⎥⎥
− ⎢⎢⎣φ (s) β(s) + . ⎥⎦ ε sin(φ(s) − φ)
β(s)
which we substitute to Hill’s equation and proceed by equat­
ing the coefficients to zero in front of sin and cos parts. We Alpha function is defined as
therefore obtain α = −β ' /2
1 '' 1 '2
ββ − β + k(s)β 2 = 1 (2.24)
2 4
and
1
φ 'y (s) = (2.25)
βy (s)
which represents the differential equations for beta function
and the betatron phase.

2.4.2 Principal trajectories

The solutions of Hill’s equation can be found in the form of
principal trajectories. These are two particular solutions of the
homogeneous Hill’s equation
y '' + k(s)y = 0
which satisfy the following initial conditions
C (s0 ) = 1; C ' (s0 ) = 0; cosine−like solution
28 unifying physics of accelerators, lasers and plasma

S (s0 ) = 1; S ' (s0 ) = 1; sine−like solution

The general solution can then be written as a linear combina­
tion of the principal trajectories

y(s) = y0 C(s) + y '0 S(s)

The pseudo-harmonic oscillations, and the cosine-like and

sine-like principal trajectories are illustrated in Fig. 2.9 for
a FODO chain of quadrupole magnets.

' ) ' ) ' ) ' ) '

V
DX

VDX
[ D
DX

FRVLQHOLNHWUDMHFWRU\
VDX
D
D
[
DX

VLQHOLNHWUDMHFWRU\ VDX
D

FIGURE 2.9
Illustration of pseudo-harmonic oscillations and cosine-like and
sine-like principal trajectories.

Furthermore, we can derive the connection between the

principal trajectories and pseudo harmonic oscillations. Let’s
express the amplitude and angle functions as
.
y(s) = εβ(s) cos (φ(s) − φ)

ε
y ' (s) = − [sin (φ(s) − φ) + α(s) cos (φ(s) − φ)]
β(s)
in terms of the principal trajectories

y(s) = y0 C(s ) + y '0 S(s)

Using simple algebraic manipulations, we find that

β(s) .
C (s) = (cos φ(s) + α0 sin φ(s)) , S(s) = β(s)β0 sin φ(s)
β0
 transverse dynamics 29

and correspondingly for the reverse formula

S(s)
φ(s) = arctg
β0 S(s) − α0 C(s)

1 S 2 (s) + [β0 S(s) − α0 C(s)]2

β (s ) =
β0 β0 S(s) − α0 C(s)
or, in a simpler shape
r 12
1 S(s)
β(s) =
β0 sin φ(s)
)
−S ' (s) β + cos φ(s)
β(s)
0
α(s) =
sin φ(s)
Having derived the above equations, we can now see that
we can describe the evolution of the particle trajectories in
a transfer line or in a circular accelerator by means of matrix
formalism (see Fig. 2.10). In other words, linear transforma­
tions that are enabled by the linearity of the Hill’s equations
express as
( ) ( )( )
y(s) C(s) S(s) y(s0 )
= (2.26)
y ' (s) C ' (s) S ' (s) y ' (s0 )

The matrix elements C(s) and S(s) depend only on the mag­
netic lattice and not on the initial conditions of the particle.

\ 7UDQVIHUPDWUL[
&6 \
&
6

\
\

\

\
[ V
V [
FIGURE 2.10
Linear matrix approach for evaluation of the evolution of the
particle coordinates in a transfer line.

The transfer matrix is therefore given by

( )
C(s) S(s)
M1→2 = (2.27)
C ' (s) S ' (s)

The described approach allows for the possibility of using the

matrix formalism to describe the evolution of the coordinates
of a charged particle in a magnetic lattice.
30 unifying physics of accelerators, lasers and plasma

2.4.3 Examples of transfer matrices

The most common elements in accelerators are drifts, bending
magnets, which steer the trajectory, quadrupoles, which pro­
vide transverse focusing, and the accelerating section, which
accelerates the beam. Each of these elements can be repre­
sented by a particular transfer matrix. Let’s consider here a
couple of the simplest examples. The transfer matrix of a drift
space is simply ( )
1 d
M= (2.28)
0 1
Bending magnets, especially in large accelerators where
each individual bend provides only a tiny bit of bending,
have transfer matrices very close to those of drift space except
when edge focusing (illustrated in Fig. 2.11) needs to be taken
into account. In the latter, the faces of the bending magnet
are not perpendicular to the reference trajectory. As clearly
seen from the figure, this can provide additional defocusing
on the horizontal plane1 due to edge effects.
The matrix of a focusing quadrupole is
( √ √ )
cos( |K|L) √1 sin( |K|L)
M= √ √ |K| √ (2.29)
− |K| sin( |K|L) cos( |K|L)

and the matrix of a defocusing quadrupole is

⎛ √ √ ⎞
⎜⎜ cosh( KL) √1 sinh( KL) ⎟⎟
M = ⎝⎜⎜ √ √ K √ ⎟⎠⎟ (2.30)
K sinh( KL) cosh( KL)

For a thin lens, L → 0 with KL staying finite, these matrices

correspondingly become
( ) ( )
1 0 1 0
FIGURE 2.11 MF = and MD = (2.31)
−|K|L 1 KL 1
Mechanism of the edge fo­
cusing of bending magnet in The higher-order elements such as sextupole and octupole
a horizontal plane. magnets are often used in accelerators for nonlinear cor­
rections. In linear approximation their transfer matrices are
equivalent to the corresponding matrix of a drift space. They
will have, however, higher-order terms and can be described
by higher-order matrices (as will be discussed later in this
chapter).

2.4.4 Matrix formalism for transfer lines

The matrix formalism is very practical for computing prop­
agation through transfer lines, especially since the transfer
matrix of each individual element of the beamline needs to
be calculated only once. In a purely linear transfer matrix, the
1 Vertical plane edge focusing can also occur but only in the case of a finite
vertical gap of the bending magnet.
 transverse dynamics 31

overall linear matrix of the beamline, computed as a step-by­

step matrix multiplication of all individual elements, would
satisfactorily describe the propagation of particles through
the beamline.
As a practical example, let’s consider a pair of thin
quadrupoles of different polarity separated by a drift space
as shown in Fig. 2.12. The overall horizontal transfer matrix
of such a system is given by
( )( )( ) ⎛ ⎞
1 0 1 L 1 0 ⎜⎜ 1 − fL L ⎟⎟
1→2
Mx = 1 ⎜
= ⎝⎜ 1 ⎟
⎟
f2 1 0 1 −1 1f1 −1 1+ L ⎠
f∗ f2

1 1 1 L
where = − + (2.32)
f ∗ f1 f2 f1f2
The quantity f ∗ can be considered as an effective focal dis­
tance of a system of two lenses — such a system is usually FIGURE 2.12
called a focusing doublet. The overall vertical transfer ma- For illustration of transfer
trix M1→2 y is obtained by reversing the signs of f 1 and f 2 . By matrix of two quadrupoles
referring to Eq. 2.32, we can see that there is a region of pa- separated by a drift.

rameters where the sign of f ∗ is the same and positive for

both the horizontal and vertical planes (for example, when

f 1 = f 2 ), which corresponds to focusing in both planes.

2.4.5 Analogy with geometric optics

As we can see, the particle trajectories can be described us­
ing matrix formalism in a very similar way to ray propaga­
tion in an optical system. The magnetic quadrupoles play the
role of focusing and defocusing lenses but, unlike an optical
lens, a magnetic quadrupole focuses in one plane and defo­
cuses in the other plane due to the nature of Maxwell’s equa­
tions. However, as we have just shown in the previous sec­
tion, a doublet of quadrupoles can focus simultaneously in
both planes, as illustrated in Fig. 2.13.

I I I I
KRUL]RQWDO YHUWLFDO

FIGURE 2.13
A doublet can focus simultaneously in both planes.

The similarity and difference with geometrical optics can

be highlighted by the following example. Let’s consider An optical telescope with two
an optical telescope consisting of two lenses as shown in lenses can provide arbitrary
Fig. 2.14. It is intuitively clear and in fact can be proven that demagnification.
32 unifying physics of accelerators, lasers and plasma

two lenses (in linear approximation), properly spaced and lo­

cated, can provide an arbitrary demagnification.
In a direct analogy to geometrical optics, two focusing
doublets are needed in order to create a telescope with arbi­
trary demagnification in the case of magnetic element optics
(i.e., four quadrupoles appropriately placed and spaced).

FIGURE 2.14
Optical telescope with two lenses.
The comparison of geometrical optics to magnetic element
optics is a powerful method that often helps for back-of-the-
Four quadrupoles are needed envelope evaluations of various optical systems.
to create a telescope with ar­
bitrary demagnification for
charged particle optics. 2.4.6 An example of a FODO lattice
Let’s consider one more practical example — an alternating
sequence of focusing (F) and defocusing (D) quadrupoles sep­
arated by a drift (O) — this is a so-called FODO lattice; see
Fig. 2.15.

) 2 ' 2 )

V

HQYHORSH

/

FIGURE 2.15
FODO lattice.

The transfer matrix of the FODO cell can be derived as

( )( )( )( ) ⎛ ⎞
⎜⎜⎜1 + 2f L 1 + 4Lf ⎟⎟⎟
L
1 0 1 L 1 0 1 L
M= 2 2 = ⎜⎝⎜ L 2 ⎟⎟
− f1 1 0 1 f1 1 0 1 − 2f 2 1 − 2fL − 4fL 2 ⎠

We will use this expression in the following section to evalu­

ate beam stability.
 transverse dynamics 33

2.4.7 Twiss functions and matrix formalism

The optical functions — beta, alpha and gamma (defined be­
low) are called Twiss functions. Using the formulas defined in
the previous sections, the matrix elements can be expressed
via the optics functions at the beginning and end of the beam-
line: ( )
C(s) S(s)
Ms0 →s = = (2.33)
−C ' (s) S ' (s)
⎛ ) . ⎞
⎜⎜ β(s)
Δφ Δφ) Δφ ⎟⎟
⎜⎜⎜ β0 (cos + α 0 sin
)
β(s)β 0 sin
⎟⎟⎟
= ⎜⎜ (α(s)−α ) cos Δφ+(1+α(s)α ) sin Δφ ⎟⎟
⎜⎝ − 0
√ 0 β0
[cos Δφ − α(s) sin Δφ] ⎟⎠
β(s)β0 β(s)

Here β0 , α0 and the phase φ 0 (in Δφ = φ − φ 0 ) correspond to

the beginning of the transfer line. The expressions above are
called Twiss parameterization of the transfer matrices.
So far, we haven’t yet assumed any periodicity in the
transfer line. However, if we now consider a periodic ma­
chine, then the transfer matrix over a single turn (single turn
map) would reduce to
( )
cos μ + α0 sin μ β0 sin μ
Ms0 →s0 = (2.34)
−γ0 sin μ cos μ − α0 sin μ

where the gamma function is defined as

1 + α02
γ0 = (2.35)
β0
and where we used μ = Δφ to define the phase advance for
one turn.

2.4.8 Stability of betatron motion

Having considered periodic transfer maps in the previous
section, we are now ready to discuss stability of the multi-
turn motion.
Consider a circular accelerator with a transfer matrix,
which for one turn equals to M. Let’s rewrite the Twiss pa­
rameterization for M given by Eg.2.34 as

M = cos μ · I + sin μ · J (2.36)

( ) ( )
1 0 α0 β0
where I = and J =
0 1 −γ0 −α0
After n turns, the particle coordinates will be given by the
successive application of the one-turn transformation matrix
n times, as follows:

x1 = M x0 ... x2 = M2 x0 ... x n = M n x0

The beauty of the parametrization given in Eq. 2.36 is that, as

34 unifying physics of accelerators, lasers and plasma

can be easily proven, J2 = −I and thus M2 = cos 2μ·I+sin 2μ·J.

Similarly, one can show that
( )
n cos nμ + α0 sin nμ β0 sin nμ
M = (2.37)
−γ0 sin nμ cos nμ − α0 sin nμ

Observing the expression for this periodic transverse

map, one can conclude that stability of the transverse motion
necessarily requires the phase advance μ to be a real number,
which ensures that the multi-turn motion represents stable
oscillations. The condition of μ being real can be re-written
as | cos μ| < 1 or as a more general expression involving the
trace or spur (sum of its diagonal elements) of the transfer
matrix
1
| cos μ| = |tr M| < 1 (2.38)
2
The criteria defined above is a necessary condition for a trans­
fer line to be suitable for multi-turn stable dynamics. We are
now ready to apply this criteria to a practical example.

2.4.9 Stability of a FODO lattice

Let’s apply the stability criteria expressed as Eq. 2.38 to the
Twiss parameterization of the matrix or the FODO cell de­
rived in Section 2.4.6
⎛ ⎞ ( )
⎜⎜⎜ 1 + 2f L 1 + 4fL ⎟⎟⎟
L
cos μ + α sin μ β sin μ
M = ⎝⎜⎜ 2 ⎟⎟ =
− L2 1 − L − L 2 ⎠
2f 2f 4f
−γ sin μ cos μ − α sin μ

The trace of this transfer matrix is given by

L2
tr M = 2 −
4f 2

And the stability criteria thus requires

L2 L
| cos μ| = 1 − <1 or f > (2.39)
8f 2 4

The resulting criteria f > L/4 is intuitively very clear and it

can also be understood from the analogy with geometric op­
tics and from considerations of the behavior of the beam en­
velope. Looking at Fig. 2.15, one can observe that f = L/4
would correspond to the situation when the size of the enve­
lope in the defocusing quadrupole approaches zero, and even
stronger quadrupoles (i.e. lower f ) would make it impossible
to sketch a repeatable finite envelope.

2.4.10 Propagation of optics functions

As we have discussed above, the coordinates of the particles
can be propagated via a transfer line using the matrices of the
 transverse dynamics 35

transfer line defined by the principal trajectories

( )
C(s) S(s)
M1→2 = (2.40)
C ' (s) S ' (s)

Similarly, we can write down the expression that computes

the propagation of the optics function along the transfer lines
using a matrix based on the principal trajectories
⎛ ⎞ ⎛ ⎞⎛ ⎞
⎜⎜⎜ β ⎟⎟⎟ ⎜⎜⎜ C '
2 −2CS S2 ⎟⎟⎟ ⎜⎜⎜ β0 ⎟⎟⎟
⎜⎜ α ⎟⎟⎟ = ⎜⎜⎜ −CC CS ' + SC ' −SS ' ⎟⎟ ⎜⎜ α ⎟⎟ (2.41)
⎝⎜ ⎠ ⎝ ⎠⎟ ⎝⎜ 0 ⎠⎟
γ C '2 −2C ' S ' S '2 γ0

The initial values of the optical functions in this equation are

either determined by the periodicity conditions as in the case
of a circular machine, or correspond to the initial values at
the entrance of the system as in the case of a transfer line.
Let’s consider two cases as examples. Drift space ex­
presses as
( ) ⎛ ⎞ ⎛ ⎞⎛ ⎞
⎜⎜⎜ β ⎟⎟⎟ ⎜⎜⎜ 1 −2s s2 ⎟⎟⎟ ⎜⎜⎜ β0 ⎟⎟
M=
1 s
→ ⎜⎜⎜ α ⎟⎟⎟ = ⎜⎜⎜ 0 1 −s ⎟⎟⎟ ⎜⎜⎜ α0 ⎟⎟⎟
0 1 ⎝ ⎠ ⎝ ⎠⎝ ⎟⎠
γ 0 0 1 γ0

We can see that the β function has a parabolic behavior in

correlation to the drift length.
A thin focusing quadrupole of focal length f = 1/KL
⎛ ⎞ ⎛ ⎞⎛ ⎞
( ) ⎜ β ⎟⎟ ⎜⎜ 1 0 0 ⎟⎟ ⎜⎜ β0 ⎟⎟
⎜
⎜ ⎟ ⎜ ⎟
→ ⎜⎜⎜ α ⎟⎟⎟ = ⎜⎜⎜⎜ KL 0 ⎟⎟⎟ ⎜⎜⎜⎜ α0 ⎟⎟⎟⎟
1 0
M= 1
KL 1 ⎝ ⎠ ⎝ 2 ⎟⎠ ⎝ ⎠
γ (KL) 2KL 1 γ0

In this case, it is the γ function that has a parabolic behavior

in correlation to the inverse focal length of the quadrupole.

2.5 Phase space

2.5.1 Phase space ellipse and Courant–Snyder invariant
To summarize our discussion about betatron motion, let’s dis­
cuss the evolution of a phase space ellipse.
The solutions of the Hill’s equation

d 2y
+ Ky (s)y = 0 (2.42)
ds 2
discussed in Section 2.4.1 is reproduced below
.
y(s) = εβ(s) cos (φ(s) − φ) (2.43)

ε
y ' (s) = − [sin (φ(s) − φ) + α(s) cos (φ(s) − φ)]
β(s)
36 unifying physics of accelerators, lasers and plasma

2
/ tan 2

/ /

Area

FIGURE 2.16
Betatron motion in phase space.

This solution describes an evolution of an ellipse in phase

space (y, y ' ). The parameters of the ellipse are described in
Fig. 2.16.
Hill’s equations have a remarkable property — they have
an invariant:
A(s) = β y '2 + 2αyy ' + γ 2 y 2 = const. = ε (2.44)
This can be proven by substituting the solutions of Hill’s
equations Eq. 2.43 into Eq. 2.44 for A(s).
The quantity A(s) is called the Courant–Snyder invariant
and is connected to the area of the ellipse in phase space with
FIGURE 2.17 a factor of π as Area = πε.
Evolution of phase-space el­ The Courant–Snyder invariant — and thus the area of
lipse. Locations: (a) in D and the ellipse — stay constant independent of the optics of the
(c) in F quadrupoles, and (b) beamline. As illustrated in Fig. 2.17, the ellipse rotates and
in between. its shape may change while its area remains invariant.

2.6 Dispersion and tunes

2.6.1 Dispersion
We have so far assumed that the particles of the beam have
a nominal energy equal to that of the energy of the reference
particle. In practice, however, there is always some energy
offset or energy spread within the beam. The function that
characterizes the orbit of an off-energy particle in an acceler­
ator is called the dispersion function.
The primary effect of the energy offset is the difference
of the trajectory in bending magnets — as illustrated in
FIGURE 2.18 Fig. 2.18. After the propagation of an off-energy particle in
Bending magnet creates dis­ a magnet, both an offset and angle of the orbit are created.
persion. In linear approximation, the radius of curvature of the trajec­
tory for an off-energy particle expresses as
( )
1 eB eB 1 Δp
= = Δp
≈ 1 − (2.45)
ρ(s) p p0 1 + p ρ0 (s) p0
0
 transverse dynamics 37

Substituting this to Eq. 2.19 gives us the Hill’s equation for

an off-energy particle:
( )
d 2x 1 1 Δp
2
− K(s) − 2 (s) x = (2.46)
ds ρ ρ(s) p0

The answer to the above equation can be found using the fol­
lowing form:
dp
x = x0 + D (2.47)
p
where D is the dispersion function and x0 describes betatron
oscillation around the dispersive orbit as shown in Fig. 2.19.

'

FIGURE 2.19
Dispersion.

Substituting this to Eq. 2.46 will yield the following equa­

tion, which governs the evolution of the dispersion function:
( )
d 2D 1 1
− K(s) − 2 D= (2.48)
ds2 ρ (s) ρ(s)

Dispersion function D can also be expressed in terms of

the principal trajectories as
s s
C(t) S(t)
D(s) = S(s) dt − C(s) dt (2.49)
ρ(t) ρ(t)
s0 s0

s s
' ' C(t) S(t)
D (s) = S (s) dt − C ' (s) dt (2.50)
ρ(t) ρ(t)
s0 s0

We have assumed in this section that all bending occurs in

a horizontal plane and therefore only a horizontal dispersion
function is nonzero. This may not be the case in instances of
vertically bending magnets or coupling, discussed below.
As dispersion affects the space location of the reference
orbit for off-energy particles, it thus also affects the orbit’s
path length. Defining C as circumference or orbit path length
in curvilinear coordinates, the deviation of the path length
can be shown to be given by

dp D(s)
ΔC = ds (2.51)
p ρ(s)
38 unifying physics of accelerators, lasers and plasma

We can also compute the so-called momentum compaction fac­

tor with
dC/C 1 D(s)
αc = = ds (2.52)
dp/p C ρ(s)
which will be discussed further in Chapter 5, which deals
with longitudinal dynamics.

2.6.2 Betatron tunes and resonances

Taking into account the definition of the betatron phase in
Eq. 2.23, we can write the phase advance over one turn of a
circular machine as
ds
Δφ C = 2π Q = (2.53)
β(s)

The quantity Q is called the betatron tune. The betatron tunes

are essential quantities used to analyze the stability of a cir­
cular accelerator. In particular, if Q is an integer number, res­
onance conditions occur, as a tiny disturbance at some place
along the orbit would repeat for many turns, accumulating
into a large disruption of the particle’s motion.
The general equations for resonance conditions of the be­
tatron tunes can be written as follows:

m Qx + n Q z = k (2.54)

where m, n and k are integer numbers and where |m| + |n|

is called the order of the resonance. Resonances of the lowest
orders are the most dangerous for the stability of the particle
motion and thus must be carefully avoided by proper ma­
chine optics design.

2.7 Aberrations and coupling

2.7.1 Chromaticity
 Offsets of energy in the particles cause not only dispersion

! but also result in different focusing strengths of the magnetic
elements (as illustrated in Fig. 2.20).
The quadrupole strength for off-energy particles in the
first order can be approximated as follows

e ∂By e ∂By k
k1 = = = 0 ≈ k0 (1 − δ) (2.55)
p ∂x p0 (1 + δ) ∂x 1+δ
FIGURE 2.20 Taking into account the equations that describe beta func­
Chromaticity of a focusing tion evolution, we can show that the betatron tunes shift in
quadrupole. correlation to changes in focusing strength:

1
ΔQ = β(s)ΔK(s)ds (2.56)
4π
 transverse dynamics 39
\ %\
Using the above approximation for the off-energy quadrupole
strength, we thus write the expression for chromaticity Q ' as
6 %[
describing the dependence of the betatron tune on the energy
offset of the particle and define it as the derivative of the be­ 1 1[
tatron tunes with respect to the relative energy change:
dQ 1
Q' = =− β(s)k0 (s)ds (2.57)
dδ 4π 6
\ )\
6 )[
2.7.2 Coupling
Throughout this chapter we have assumed that the motions 1 1[
of particles in horizontal and vertical planes are independent.
This could indeed be the case if machine optics consist of
bending magnets and quadrupoles that are perfectly placed
in space. However, any rotational misalignments of these el­ 6
ements can create coupling of the horizontal and vertical mo­ FIGURE 2.21
tions. Skew quad fields and forces.
Coupling can also be created by other magnetic elements
such as solenoids (especially strong coupling can occur when
a solenoid overlaps with quadrupole field, mixing different
types of symmetry), or by misaligned nonlinear magnets such
as sextupoles or octupoles, etc.
Some amount of coupling is unavoidable in a real ma­
chine and it usually needs to be corrected. A standard way
to correct coupling (or to create it on purpose if needed) is
to use skew quadrupoles — these are standard quadrupoles
rotated by 45◦ as shown in Fig. 2.21.

2.7.3 Higher orders

In storage rings, chromaticity is defined as a dependence of
the betatron tunes on energy.
In single-path beamlines, it is more convenient to use
other definitions. Let us first recall the linear matrix ap­
proach
xiout = Ri j xjin (2.58)
FIGURE 2.22
this time noting all six components of the vector of interest, Sextupole fields and forces.
adding to the two coordinates and their angles the longitudi­
nal offset Δl as well as the energy offset δ

xi = (x, x' , y, y ' , Δl, δ)' (2.59)

The second, third and other higher terms that can re­
sult from nonlinear elements such as a sextupole shown in
Fig. 2.22 or octupole (Fig. 2.23) can be included in the matrix
formalism in a similar manner:

xiout = Ri j xjin + Ti j k xjin xkin + Ui j k n xjin xkin xnin + ... (2.60)

FIGURE 2.23
where T and U are the second and third-order matrices. Octupole magnet forces.
40 unifying physics of accelerators, lasers and plasma

Unlike in the storage rings, in transfer line design, one

usually calls chromaticity the second-order elements T126
and T346 . All other high-order terms are just aberrations,
purely chromatic (like T166 , which is second-order disper­
sion), or chromo-geometric (like U32446 ).
 transverse dynamics 41

EXERCISES
2.1 Chapter materials review.

Define the region of parameters where a pair of thin

quadrupoles will focus the beam in both planes.

2.2 Chapter materials review.

Prove the Eq. 2.39, which defines the stability of a FODO

beamline, geometrically, using the analogy with a traditional

geometrical optics.

2.3 Chapter materials review.

A parallel proton beam of E=200 MeV enters a beamline. It is

necessary to focus this beam into a point at a 3 m distance

from the entrance. Estimate the necessary parameters of a

quadrupole system (gradients, lengths) that can perform this

task.

2.4 Mini-project.

Consider the same proton beam as in the previous exercise,

as well as the same focusing requirements. Assume that the

focusing is performed by a continuous, cylindrical electron

beam. Estimate the necessary electron density which can per­

form the focusing task. Select electron beam energy, deter­
mine the electron current and discuss and select an optimal

design of the electron beam system, as well as its feasibility.

2.5 Analyze inventions or discoveries using TRIZ and AS-TRIZ.

Analyze and describe scientific or technical inventions de­

scribed in this chapter in terms of the TRIZ and AS-TRIZ ap­
proaches, identifying a contradiction and an inventive prin­
ciple that were used (could have been used) for these inven­
tions.

2.6 Developing AS-TRIZ parameters and inventive principles.

Based on what you already know about accelerator science,

discuss and suggest the possible additional parameters for

the AS-TRIZ contradiction matrix, as well as the possible ad­

ditional AS-TRIZ inventive principles.

 3
Synchrotron Radiation

3.1 SR on the back of In this chapter we will consider one of the most important
an envelope 43 phenomena that governs the behavior of accelerators — syn­
3.2 SR effects on the chrotron radiation (SR).
beam 48 SR can be both helpful, as it yields the creation of high
3.3 SR features 51 brightness radiation sources, and harmful, as it can deterio­
rate the beam by creating additional energy spread and beam
emittance growth.
Traditional derivations of SR equations are rather mathe­
matically involved. However, in this chapter we will use sim­
plified back-of-the-envelope style derivations, which never­
theless obtain all of the important characteristics of SR with
high accuracy.

3.1 SR on the back of an envelope

In our simple picture, the SR is the result of the charged par­
ticle leaving part of its fields behind when it is moving on a
curved trajectory. The part of the field that is left behind (or
radiated) cannot catch up with the motion of the particle, as
it cannot move faster than the speed of light.
Armed with this concept, let’s estimate the power loss due
to SR, the typical energy of the emitted photons, and other
parameters of synchrotron radiation as well as the most im­
portant effects that SR inflicts on the beam.

3.1.1 SR power loss

The straightforward concept of SR described above is repre­
sented in Fig. 3.1. In this instance, the particle moving with
velocity v (which is close to the speed of light) on a radius R
has its field lines pointing mostly transversely, and the part
of field moving further away (on the radius R + r) would be
left behind, as it cannot move faster than c.
The radius r can be evaluated as
c R
r =R − 1 ≈ 2 (3.1)
v 2γ

where we assumed that γ » 1 or β ≈ 1 and thus (1 − v/c) =

(1 − β)(1 + β)/(1 + β) ≈ (1 − β 2 )/2 = 1/(2γ 2 ).

DOI: 10.1201/b18696-3 43
44 unifying physics of accelerators, lasers and plasma
)LHOGOHIW
EHKLQG

R R+r

)LHOGOLQHV

FIGURE 3.1
Synchrotron radiation — conceptual explanation.

The energy in the field that is left behind (radiated) can be

estimated (we use Gaussian units in this section) as a volume
integral of the field squared:

W≈ E 2 dV (3.2)

The field E can be estimated as the field on the radius r from

the particle
e
E≈ 2
r
and the characteristic volume can be estimated (see Fig. 3.2)
as
V ≈ r 2 ds (3.3)
where ds is an element of the path along the orbit, as we will
eventually wish to find the power lost per unit of length.
FIGURE 3.2
The energy loss per unit length can thus be written as:
Illustrating the characteristic
volume used in Eq. 3.3. dW e 2
≈ E2 r2 ≈ 2 r2 (3.4)
ds r
and after substituting
R
r≈
2γ 2
we get an estimate of:

dW e2 γ 4
≈ 2 (3.5)
ds R
which we can compare with the exact formula:
dW 2 e2 γ 4
= (3.6)
ds 3 R2
and conclude that our rough estimations give a very reason­
able result.
 synchrotron radiation 45

3.1.2 Cooling time

Knowledge of the SR-caused power losses immediately allow
us to estimate the SR cooling time of the beam in a storage
ring.
First of all, let’s — for convenience’s sake — rewrite the
exact formula for power lost per unit length

dW 2 e 2 γ 4
=
ds 3 R2
in the way that does not depend on the systems of units:

dW 2 re γ 4 2
= mc (3.7)
ds 3 R2
The SR energy loss of a particle per turn is therefore:

4π re γ 4 2
U0 = mc (3.8)
3 R
The effect of particle cooling due to SR is based on the fact
that when an electron radiates a photon, its momentum de­
creases. Taking into account that while the beam of particles
can have a range of angles within the beam, the accelerating
RF cavity would restore only the longitudinal part of momen­
tum, whereas the transverse degrees of freedom of the parti­
cles will be cooled down as illustrated in Fig. 3.3.

FIGURE 3.3
RF cavity restores only longitudinal momentum, thus other de­
grees of freedom are cooled due to synchrotron radiation.

Let’s estimate the cooling time τ of a particle with energy

E0 cycling with revolution period T0 in a circular machine as
τ = E0 T0 /U0 , which yields

2π R γmc 2
τ≈ (3.9)
c U0

After substitution, the inverse cooling time can be written as

2 c re γ 3
τ −1 ≈ (3.10)
3 R2
46 unifying physics of accelerators, lasers and plasma

3.1.3 Cooling time and partition

In the previous section we estimated the inverse cooling time
as τ −1 ≈ 2 c re γ 3 /(3R2 ). Traditionally, there is a factor of 2 in
the definition in the cooling time:

1 c re γ 3
τ = 2E0 T0 /U0 ⇒ τ −1 = (3.11)
3 R2
We will use this latter definition in this section.
We can express the evolution of the beam emittance under
the influence of an SR damping as

ε(t) = ε0 exp( − 2 t /τ ) (3.12)

Both transverse planes, as well as the longitudinal motion

in rings, are usually coupled. Thus we can expect that the
damping will be distributed between these degrees of free­
dom in some proportion depending on details of the optics.
Distribution of cooling between the degrees of freedom
is defined by the so-called partition numbers Jx , Jy and JE ,
which we mention here without derivations. The cooling time
of a degree of freedom is correspondingly
τ
τi = (3.13)
Ji

The total radiated power due to SR is fixed and constant,

therefore
τi−1 = const. (3.14)
which corresponds to the partition theorem

Ji = 4 (3.15)

for a typical accelerator

Jx ≈ 1 , Jy ≈ 1 , JE ≈ 2 (3.16)

and adjusting the optics of the machine changes the distribu­

tion of the partition numbers.

3.1.4 SR photon energy

For γ » 1 the emitted pho- In order to estimate the typical energy of the SR photons,
tons go into 1/γ cone. we need to make an assumption that is based on relativistic
kinematics: the radiation of relativistic particles is emitted
into a cone with angular spread of 1/γ.
Let’s take this assumption into account when examining
the radiation emitted during motion along the curved trajec­
tory shown in Fig. 3.4 and ask a question — during what time
interval Δt would the remote observer see the emitted fields?
 synchrotron radiation 47


Y  F
$ % 2EVHUYHU

5



FIGURE 3.4
SR and remote observer.

Looking at this figure we conclude that the emitted radi­

ation is observed during particle travel along the 2R/γ arc.
Let’s now keep in mind that the radiation travels at speed
c, while particles travel at v. At point B shown in Fig. 3.4 the
separation between radiation and the particles is given by
2R v
ds ≈ 1− (3.17)
γ c
Therefore, the observer will see radiation during the follow­
ing time interval:
ds 2R R
Δt ≈ ≈ (1 − β) ≈ (3.18)
c cγ c γ3
We can thus proceed to estimate the characteristic fre­
quency of emitted photons as the inverse of the time duration
of the flash, as seen by the observer:

1 c γ3
ωc ≈ ≈ (3.19)
Δt R
Comparing the above with the exact formula that we re­
produce here without derivation

3 c γ3
ωc = (3.20)
2 R
we can again see that our back-of-the-envelope estimations
give pretty accurate results while not hiding the physics of
the phenomena behind heavy math.

3.1.5 SR — number of photons

Having estimated the characteristic energy of the SR photons,
we can now estimate the number of photons emitted per unit
length by a single electron.
48 unifying physics of accelerators, lasers and plasma

Let’s use our estimation for the rate of energy loss

dW /ds ≈ e 2 γ 4 /R2 and the estimation of the characteristic fre­
quency of photons ωc ≈ cγ 3 /R, and rewrite the latter in terms
of the photon energy:

γ3 nc γ3
εc = nωc ≈ = λ mc2 (3.21)
R R e
where
e2 e2 r
re = α = λe = e
mc 2 nc α
The number of photons emitted per unit length can be
obtained by dividing the energy loss per unit length by the
energy of the photons

dN 1 dW α γ
≈ ≈ (3.22)
ds εc ds R

It is also practical to derive an expression for the number

of photons emitted per unit of the bending angle θ

N ≈ αγθ (3.23)

which is given by a remarkably simple and clear formula.

3.2 SR effects on the beam

The derived characteristics of SR allow evaluation of the ef­
fects of SR on the beam.

3.2.1 SR-induced energy spread

The energy
√ spread ΔE/E will grow due to statistical fluctua­
tions ( N ) of the number of emitted SR photons and there­
fore can be estimated as

d (ΔE/E)2 dN 1
≈ εc2 (3.24)
ds ds (γmc 2 )2

which gives the following estimation

d (ΔE/E)2 re λ e γ 5
≈ (3.25)
ds R3
Comparing this with the exact formula

d (ΔE/E)2 55 re λe γ 5
= √ (3.26)
ds 24 3 R3

confirms good √ accuracy of the estimation as the numerical

factor 55/(24 3) ≈ 1.32.
 synchrotron radiation 49

3.2.2 SR-induced emittance growth

Let’s estimate the beam emittance growth rate due to syn­
chrotron radiation. The qualitative picture of the phe­
nomenon is shown in Fig. 3.5.

(PLWSKRWRQ

FIGURE 3.5
SR can cause excitation of oscillation of particles and correspond­
ing emittance growth.

In this diagram, the dispersion function η shows how the

equilibrium orbit shifts when particle energy changes due to
photon emission. Correspondingly, when a photon is emitted
and the energy of the particle becomes equal to E +ΔE (where
ΔE is negative), the particle starts to oscillate around a new
equilibrium orbit. The amplitude of oscillation will be equal
to
Δx ≈ η ΔE/E
Let’s compare this with the betatron beam size given by

σx = (εx βx )1/2

and write an estimate for the emittance growth as

Δεx ≈ Δx2 /β

By expanding the equation, we obtain an estimation for

the emittance growth:
2
dεx η 2 d (ΔE/E) η 2 re λe γ 5
≈ ≈ (3.27)
ds βx ds βx R3
In the above estimation we ignored the dependence of β
and η on s; however, these dependences can alter the results.
The exact formula, which takes into account the derivatives
of the Twiss functions, is as follows:
' ' 2
η 2 + βx η − βx η /2
dεx 55 re λe γ 5
= √ (3.28)
ds βx 24 3 R3
� =H /
50 unifying physics of accelerators, lasers and plasma

where the parenthesis with Twiss functions in front of the nu­

merical coefficient is usually called H. As we see, the back-of­
the-envelope estimation correctly captures the most impor­
tant features of the phenomenon and also produces a useful
and simple expression.

3.2.3 Equilibrium emittance

The SR-induced cooling of the beam emittance and SR-
induced emittance growth would naturally balance, so that
the beam emittance would eventually reach equilibrium
value.
Let’s look at the estimated rate of emittance growth:
2
dεx η 2 d (ΔE/E) η 2 re λe γ 5
≈ ≈
ds βx ds βx R3

and the SR cooling rate

dε 2 1 c re γ 3
=− ε with τ −1 =
ds cτ 3 R2
and equate them to obtain an expression for the horizontal
equilibrium emittance:

c τ η 2 re λe γ 5
εx0 ≈ (3.29)
2 βx R3

or, after substitution:

3 η 2 λe γ 2
εx0 ≈ (3.30)
2 βx R

In the equations above, we ignored dependence of R on longi­

tudinal coordinate s. In order to obtain more accurate formu­
las from these equations, one needs to use the values (1/R2 )
and (1/R3 ), which are averaged over the orbit period.
In the vertical plane, SR’s contribution to emittance is
only due to 1/γ angles of emitted photons, but usually the
impact on highly relativistic beams is negligibly small.
The vertical equilibrium emittance is therefore usually
defined not by SR directly, but by the coupling coefficient k
(which is « 1) of x-y planes:

εy0 ≈ k εx0

In the above, we ignored partition numbers, but they can

be taken into account in accurate calculations. The equilib­
rium energy spread of the beam can also be calculated in a
similar manner.
 synchrotron radiation 51

3.3 SR features
Performance of SR-based light sources depends on spatial
and spectral characteristics of synchrotron radiation. Below
we will introduce basic relevant characteristics, leaving de­
tailed discussion for Chapter 7.

3.3.1 Emittance of single radiated photon

In order to discuss the ultimate brightness of the SR-based
light sources, we require knowledge of the emittance of a sin­
gle photon radiated from the curved beamline.
To answer this question we must first evaluate the size of
the emitting region seen by the remote observer in the case of
a single electron, as illustrated in Fig. 3.6.

R
σ≈
2γ 2

FIGURE 3.6
For illustration of emittance of a single photon.

We first note that the angles of photons coming from the

emitting region are spread over σ ' ≈ 1/γ. The size of the emit­
ting region is given by the height of the arc segment and is
thus equal to σ ≈ R/(2γ 2 ).
The estimate for the emittance of an SR photon emitted
by a single electron can therefore be written as

R
εph = σ σ ' ⇒ εph ≈ (3.31)
2γ 3

(In a similar way as above, one can estimate beta function of

photons as β = σ/σ ' .)
Let’s rewrite the photon emittance equation using the ex­
pression for photon wavelength

2π c c γ 3 λc
ωc = ≈ ⇒ εph ≈ (3.32)
λc R 4π

We can see here that the emittance of synchrotron radiation

is directly related to its wavelength. This is not a coincidence
52 unifying physics of accelerators, lasers and plasma

in a single example, but is actually an inherent property of

photon radiation:
λ
εph = (3.33)
4π
This, together with information about the SR spectrum in the
next section, will bring us to discuss the brightness of SR light
sources.

3.3.2 SR spectrum
In all the estimations above, we assumed that the photons
emitted are monoenergetic. It is not exactly the case, and in
reality the energy of the photons will be distributed around
the characteristic frequency of the SR photons ωc .
Accurate mathematics, which we do not show here, pre­
dicts that the SR spectrum looks like the one shown in
Fig. 3.7.

FIGURE 3.7
SR spectrum and its approximations for low (curve a — behaves
as 4/3 · x1/3 ) and high (curve b — behaves as 7/9 · x1/2 e−x )
energies.

We can indeed see that a large fraction of the photons will

have energies close to ωc . However, there is also a lower en­
ergy tail, as well as some fraction of higher energy photons.
It is also natural to expect that the photons’ angular distri­
bution will deviate from the 1/γ rule, and indeed, the lower
energy photons typically have larger angular spread.

3.3.3 Brightness or brilliance

Following discussion of the SR spectrum, we can introduce
the notion of bandwidth — the interval of interest in the spec­
trum of photon frequencies. This bandwidth is denoted here
BW and is expressed, typically, in %.
Let’s assume that our photon beam is emitted from the
 synchrotron radiation 53

area As and the emitted radiation has angular opening angles

of ΔΦ and Δψ, as shown in Fig. 3.8.

FIGURE 3.8
For illustration of brilliance or brightness. Here As is the emitting
area and ΔΨ and ΔΦ are opening angles of emitted photons.

The first concept to make note of is flux, which is ex­

pressed as the number of photons emitted per units of time
and per unit of bandwidth:

Flux = Photons/ (s · BW ) (3.34)

The brilliance (or brightness) is then defined as flux per

unit of emitting area and product of angles:

Brilliance = Flux/(As · ΔΦ · Δψ)

and is usually expressed in units of [Photons/(s · mm· mrad 2 ·

BW )].
In a typical case of Gaussian distributions, the definition
of brilliance is based on the total effective sizes and diver­
gences
flux
brilliance = (3.35)
4π 2 Σx Σx' Σy Σy '
where the total effective sizes include contributions from
electrons as well as photons:
) )
Σx = σx,e
2 2
+ σph,e σx = εx βx + (Dx σε )2 (3.36)

) )
' 2
Σx ' = σx2' ,e + σph,e σx' = εx βx + (D 'x σε )2

and similarly for the other plane.

3.3.4 Ultimate brightness

As we have seen, brilliance is defined by the overall effec­
tive emittance, which convolves electron and photon distri­
butions: ) )
εef f = σe2 + σph
2
σe2' + σph
2
' (3.37)
54 unifying physics of accelerators, lasers and plasma

Since the lowest photon emittance depends on the photon

wavelength (Eq. 3.33), the smallest overall emittance will be
obtained when:
εe = σe σe' ≤ εph (3.38)
which corresponds to a diffraction-limited source.
In modern SR sources, the typical radiation of interest
spans from 100 eV to 100 keV in terms of photon energy. Let’s
o take an example of 12.4 keV of photon energy which corre-
Wavelength λ ≈ 1 A corre­
sponds to 12.4 keV photons. sponds to εph ≈ 8 pm. For typical third-generation SR light
sources, the horizontal emittance εx is usually between 1 and
5 nm and the vertical emittance εy between 1 and 40 pm.
Thus, such rings are close, in terms of its emittance, to the ul­
timate performance in the vertical plane. However, they are
many orders of magnitude away from the diffraction-limited
emittance in the horizontal plane. We will discuss the ques­
tions of brilliance in closer detail in Chapter 7, which is ded­
icated to light sources.

3.3.5 Wiggler and undulator radiation

Let’s consider radiation from a sequence of bends, and in par­
ticular, let’s assume that the bends are arranged in a sequence
with + − + − +− polarity with a period of λu , so that the trajec­
tory of the particle wiggles as shown in Fig. 3.9.

FIGURE 3.9
Radiation from sequence of bends.

An external observer will see photons emitted by the par­

ticle during its travel along the arc 2R/γ. Let’s now define the
parameter K as the ratio of the wiggling period to the length
of this arc:
K ∼ γ λu /R (3.39)
We can qualitatively see that if 2R/γ « λu /2, then the
radiation emitted at each wiggle is independent. This corre­
sponds to K » 1 and is called the wiggler regime.
On the other hand, if 2R/γ » λu /2, then we are in a
regime where the entire wiggling trajectory contributes to ra­
diation (therefore interference leads to coherence of radiation,
 synchrotron radiation 55

which is explored in detail in further chapters). This corre­

sponds to K « 1 and is called the undulator regime.
As we see, the undulator parameter K ∼ γ λu /R defines dif­
ferent regimes of synchrotron radiation: K » 1 is the wiggler
regime, K « 1 is the undulator regime.

FIGURE 3.10
Wiggler and undulator radiation.

Fig. 3.10 shows the differences between radiation from a

single bend and from a sequence of bends in wiggler and un­
dulator regimes, respectively. We will consider this topic in
more detail in the Chapters dedicated to light sources and to
FELs, which are respectively 7 and 8.

3.3.6 SR quantum regime

Let’s define the parameter “Upsilon”as Υ = nωc /E. The mean­
ing of this parameter depends on the regimes of SR.
When the parameter Υ « 1, its physical meaning is the
ratio of the characteristic photon energy to the energy of a
single electron in the beam.
However, when Υ ∼ 1 and higher, the classical regime of
synchrotron radiation is not applicable, and the quantum SR
formulae of Sokolov–Ternov should be used. Such a situation
may happen in particular in collision or highly relativistic
focused beam, e.g., in linear colliders.
In a quantum regime, the shape of the SR spectrum
changes, as there should not be a photon emitted that has
energy larger than the energy of the initial particle.
The qualitative dependence of the SR spectrum in classi­ FIGURE 3.11
cal and quantum regimes is shown in Fig. 3.11. SR spectrum in classical and
Though the quantum SR is unlikely to occur in radiation quantum regimes.
from bends, it can happen in SR during beam collisions, as
beams focused to tiny spots can produce enormous fields that
cause the oncoming particles to radiate in a quantum regime.
56 unifying physics of accelerators, lasers and plasma

EXERCISES
3.1 Chapter materials review.

A proton beam of E=50 TeV circulates in a 100 km perimeter

ring. Estimate the synchrotron radiation energy loss per turn,

the characteristic energy of the emitted photons and the cool­

ing time.

3.2 Chapter materials review.

Describe how one needs to change the optics of third-

generation SR sources in order to approach the diffraction-

limited SR source, particularly in the horizontal plane.

3.3 Chapter materials review.

In a manner similar to how the equilibrium emittance was es­

timated in this chapter, derive the equilibrium energy spread

of the beam.

3.4 Mini-project.

Define the approximate parameters (energy, sizes, fields in

bending magnets) of a second-generation SR source aiming

to achieve 10 keV of X-rays.

3.5 Analyze inventions or discoveries using TRIZ and AS-TRIZ.

Analyze and describe scientific or technical inventions de­

scribed in this chapter in terms of the TRIZ and AS-TRIZ ap­
proaches, identifying a contradiction and an inventive prin­
ciple that were used (could have been used) for these inven­
tions.

3.6 Developing AS-TRIZ parameters and inventive principles.

Based on what you already know about accelerator science,

discuss and suggest the possible additional parameters for

the AS-TRIZ contradiction matrix, as well as the possible ad­

ditional AS-TRIZ inventive principles.

 4
Synergies between
Accelerators, Lasers and
Plasma
4.1 Create 58 In this chapter we will discuss the synergies between acceler­
4.2 Energize 63 ators, lasers and plasma.
4.3 Manipulate 68 Within the scope of our interests, the themes linking all
4.4 Interact 73 three areas (accelerators, lasers, plasma), include, among oth­
ers, beam sources, laser beam generation, wave propagation
(in structures and in plasma), laser propagation in plasma,
plasma acceleration, radiation (synchrotron, betatron) and
free electron lasers, as illustrated in Fig. 4.1.

%HDPVRXUFHVODVHUEHDP
JHQHUDWLRQ
:DYHSURSDJDWLRQLQ
VWUXFWXUHVLQSODVPD
/DVHUSURSDJDWLRQLQ
SODVPDFULWLFDOGHQVLW\
3ODVPDDFFHOHUDWLRQ
5DGLDWLRQV\QFKURWURQ
3
EHWDWURQ
)UHH(OHFWURQ/DVHUV

O
D
V
P
Synergy — cross-fertilizing
interaction of considered
D
areas of physics. FIGURE 4.1
Synergies between accelerators, lasers and plasma.

DOI: 10.1201/b18696-4 57
58 unifying physics of accelerators, lasers and plasma

Some of the themes that connect both accelerators and

lasers are: focusing (chromaticity, aberrations, beam quality),
cavities (RF and optical), laser-beam interaction (ponderomo­
tive force), laser imprints (on e-beam in wigglers), cooling (e-,
stochastic, optical stochastic, laser), Compton X-ray sources,
chirped pulse amplification and bunch or pulse compression.
Accelerators’ and plasma’s connecting themes include: in­
stabilities (plasma oscillation, beam instability, e-cloud, e-
ion), beam–beam effects, plasma-focusing lenses, plasma mir­
rors, collision-less Landau damping (in plasma, in beams)
and echo effects (in beams, in plasma).
Lastly, lasers’ and plasma’s themes include: gas lasers, op­


tical parametric chirped pulse amplification and harmonic




generation. In this chapter and in this entire text, we will

touch on only some of the topics mentioned above.


In the following sections we will discuss accelerators,
lasers and plasma in the order of how we interact with
them in the real world: First, we create the beam/light
FIGURE 4.2
pulse/plasma wave; then we prepare them for use, i.e., en-
Discussion of synergies will
ergize (accelerate, amplify, excite in plasma) them or manip­
follow this sequence.
ulate (focus, compress, stretch, etc.) them; and, lastly, use or
interact with them. This sequence is illustrated in Fig. 4.2.

4.1 Create
Let’s discuss the topic of beam, light pulse and plasma wave
creation in the Create — Energize — Manipulate — Interact
sequence.

4.1.1 Beam sources

We will begin our dialogue on particle sources by starting
with leptons (electrons, positrons, muons, etc.) and later mov­
ing on to hadrons (protons, antiprotons, etc.) and ions.
The simplest form of an electron source is the thermal
cathode gun (Fig. 4.3). According to the “three-halves power”
law (or the Child-Langmuir law), the space charge effects of
FIGURE 4.3 the non-relativistic accelerated electron beam often limit the
Thermal cathode e-gun. current in electron guns.

I = P · U 3/2 (4.1)

where I is the current, U is the cathode to anode voltage and

the coefficient P is called perveance.
A much more modern source of electrons is the laser-
driven photocathode gun (illustrated in Fig. 4.4). Photo elec­
tron guns contain electrons that are generated by a laser field
via the photoelectric effect. Such a gun usually consists of
one-and-a-half RF cavities and a photocathode made either
out of metal (e.g., copper) or from a special alloy. While a
pure metal photocathode is robust, its quantum efficiency (the
 synergies between accelerators, lasers and plasma 59

number of electrons per number of incident photons) is usu­

ally quite low, of the order of 10−4 . Photocathodes with alkali
metals such as cesium can have a quantum efficiency of 10%
or higher.
The photocathode guns driven by a laser pulse are also
very suitable for production of short electron beams. The
laser pulse is usually sent to the cathode at a small angle as
shown in Fig. 4.4 to avoid interference with the accelerating
electron beam.

FIGURE 4.4
One-and-a-half-cell RF photocathode electron gun.

As seen from Eq. 4.1, photoguns can provide a much

higher pulsed electron current — due to a higher acceler­
ating voltage — produced at the cathode by the RF struc­
ture. The typical accelerating voltage in guns ranges from
around 50 MV/m in an L-band, 100 MV/m in an S-band and
200 MV/m in an X-band. RF frequencies bands: L: 1­
Producing positrons usually requires creating e + e − pairs 2 GHz, S: 2-4 GHz, C: 4­
followed by separating the positrons. An electron or photon 8 GHz, X: 8-12 GHz.
beam with a sufficient amount of energy is sent onto a target
where the e+ e− pairs will be produced. Separated positrons
are then accelerated and sent to a damping ring, where their
emittance will decrease due to radiation damping.
Ion or proton beams are produced by plasma-based ion
sources — a large variety of source types exists. An example
shown in Fig. 4.5 depicts a Penning source in which a mag­
netic trap is arranged in the cathode-anode area of an elec­
tron beam, ionizing gas via discharge. Ions of certain charges
are extracted with the help of an electrode, and are then sepa­
rated and sent for further acceleration and miscellaneous use.
In particular for negative ion sources, cesiation (developed by
V. Dudnikov; when a small amount of Cs atoms is added into
the gas) is often used to significantly enhance the emission of
negative ions. FIGURE 4.5
Plasma in ion sources can be created by various means Surface-plasma Penning H −
such as ionization via laser (which was realized in the ion ion source.
source developed at CERN, shown in Fig. 4.6). In this exam­
60 unifying physics of accelerators, lasers and plasma

ple, the laser beam is focused by a metal mirror on a tar-

Consider whether evolution get, ionizing the target and producing plasma. The ions of
from discharge ion source plasma are then directed to the extraction electrodes via a
to laser-driven ion source hole in the mirror.
can suggest a new inventive This brings us to the discussion of lasers.
principle.
0LUURU

3ODVPD
%HDP
7DUJHW

0LUURU

,QVXODWRU
8 8
FIGURE 4.6
Conceptual diagram of the laser-driven ion source.

4.1.2 Lasers
Lasers are a source of coherent light, and a laser diagram
is shown in Fig. 4.7. The main laser components are the
gain medium (which amplifies the light), the resonator (which
gives optical feedback) and the pump source (which creates
population inversion).

( /
([FLWHG
SRSXODWLRQ
D E
/
6WLPXODWHG
3XPS 6WLPXODWLQJ
SKRWRQ F
HPLVVLRQ
/
*URXQGVWDWH
FIGURE 4.7 FIGURE 4.8
Laser diagram. Three-level laser.

The gain medium contains atoms with specific conditions

for the energy levels and for the lifetime of an excited state of
the atom, suitable for arranging a three-level laser as described
in the following paragraph. The pump gets the atom pop­
ulation from the ground state L1 to the higher energy level
L3 . The excited population gets from L3 to L2 through non-
radiative decay as the lifetime of the L3 state is very short and
all the population in state L3 decays to state L2 . The lifetime
of energy state L2 is, in contrast, very long and therefore, a
population inversion occurs with respect to the state L1 . Once
the population inversion is obtained, stimulated emission can
occur when an incident photon forces the population to drop
 synergies between accelerators, lasers and plasma 61

from level L2 to L1 (as shown in Fig. 4.8), resulting in an op­

tical gain.
As you can see from the above description, a minimum of An advantage of four-level
laser systems is that much
three levels are needed to arrange a laser, but it is possible to
less pumping power is
create four and higher-level laser systems as well. needed for the creation of
The first laser built was a three-level ruby laser emitting population inversion.
a 694 nm wavelength (shown conceptually in Fig. 4.9). The
first mass-produced laser was a four-level He-Ne gas laser
emitting a 694 nm wavelength. These first lasers were char­
acterized by low emission power and very low efficiency —
typically around 0.01-0.1%.
A CO2 gas laser emitting a 10.6 μm wavelength, on the
other hand, has an efficiency close to 30% and a high power
level measuring up to kW in CW. It is also interesting to note
that the quantization of the CO2 molecule’s vibrational and
rotational states enables this laser’s system levels.

FIGURE 4.9
Conceptual diagram of a ruby laser. Quartz flash tube serves as
the pump source and ruby crystal as the gain medium.

One more example of an efficient laser (at around 40%)

is the diode laser. In this laser, the levels of the system are
enabled by the quantization of energies of holes and elec­
trons in the semiconductor diode. Due to the compactness of
the diode laser, its output light has a very large divergence,
low coherence and usually low power. However, the versatil­
ity of wavelength output and high efficiency makes this type
of laser ideal for pumping — the excitation of gain medium
in high-power laser amplifiers. In the latter, the low power, Nd:YAG — Neodymium-doped
low coherence and large divergence output of the laser diode yttrium aluminum garnet:
is amplified in a gain medium such as Nd:YAG, resulting in Nd:Y3 Al5 O12 .
a high-power, high-coherence, high-efficiency laser beam as Yb:YAG — Ytterbium-doped
illustrated in Fig. 4.10. Note that pumping by the diode laser YAG.
occurs at a shorter wavelength than the output radiation, in
agreement with Fig. 4.8.
62 unifying physics of accelerators, lasers and plasma

When used for laser pump­

ing, many low-power diode
lasers are usually assem­
bled into arrays called “diode
bars.” FDYLW\

FIGURE 4.10
Diode laser in application for pumping of Nd:YAG laser.

The laser diode output can also be applied to fiber optics and
fiber amplifiers, as we will discuss in the next section.

89ROW 4.1.3 Plasma generation

 To conclude our discussion regarding creation, let’s briefly
consider the process of how plasma is made. The most well-
known method is discharge ionization in which plasma is
 produced by DC voltage or a pulse typically applied to a low
pressure gas. The voltage U that creates discharge depends
on the product of the gap g to gas pressure P, i.e. (P × g) — the

    behavior of this voltage is represented by the Paschen curve as
3[J7RUUFP shown in Fig. 4.11.
FIGURE 4.11 Several factors are important in sustaining conditions ap­
Paschen discharge curve for propriate for a breakdown. On one side, the energy that the
hydrogen gas. electron gains during its acceleration over the mean free path
(before the next collision) should be larger than the first ion­
ization energy of the gas molecules. A lower pressure (longer
Ionization energy of hydro- mean free path) would therefore create preferable conditions
gen is ∼ 13.6 eV. for a breakdown. Multiple collisions and multiple electron
ionizations are required in order to create suitable conditions
for an avalanche and thus a breakdown. On the other hand,
overly low P × g would mean that electrons would not have
enough collisions while traveling through the gap, making
an avalanche less probable. The balance of these effects is re­
flected in the behavior of the Paschen’s curve.
( Plasma can also be created by field ionization. Both a well-
% focused relativistic electron beam (see Fig. 4.12) and a laser
can have a sufficient field to ionize gas. A gas or target can
then be instantaneously ionized if the field level reaches an
atomic field scale:

FIGURE 4.12 EBeam » EAtomic or ELaser » EAtomic

Fields of a relativistic elec­
tron bunch can produce field As we will discuss in Chapter 6, due to the multi-photon
ionization of gas. ionization and tunneling effects, much lower-level fields than
atomic ones are often sufficient for ionization.
 synergies between accelerators, lasers and plasma 63

4.2 Energize
Next comes the process of energizing the beam, laser or
plasma. In the case of a charged particle beam, this involves
acceleration (electrostatic, betatron, acceleration in RF cavi­
ties and structures, or plasma acceleration). In the case of a
laser, it involves amplification (standard, chirped pulse am­
plification, optical parametric chirped pulse amplification,
etc.), and for plasma, excitation of waves by various means
such as short pulse of a laser or beam.

4.2.1 Beam acceleration

The simplest and exceedingly widespread acceleration
method is electrostatic acceleration (Fig. 4.13). Not so widely
used now is betatron acceleration principle, which is based
on Maxwell’s Eq. 2.4.
Perhaps the most versatile acceleration method involves
the use of resonators — also known as RF cavities. Cavities
can be arranged in structures and made to be suitable for the
acceleration of any types of particles from electrons (which
almost immediately become relativistic) to protons or ions.
Often, acceleration is combined with focusing, either via
magnetic quadrupoles inserted between acceleration sec­
tions, or via an EM wave with a quadrupole component as
in a RFQ accelerating structure.

4.2.2 Laser amplifiers

One of the possible principles of laser amplification is that
a pumped gain medium of the amplifier amplifies light at
the wavelength of the oscillator laser, which is made with the
same material as the pumped gain medium of the amplifier.
The diagram of such a laser amplifier is shown in Fig. 4.14. FIGURE 4.13
The technical challenges caused by laser light amplifi­ Electrostatic and betatron ac­
cation actually resulted in numerous inventions and break­ celeration. RF cavity and RF
throughs, as discussed in Chapter 1. structure.

FIGURE 4.14
Laser amplifier. Flash lamp emits in broad spectrum. Gain
medium amplifies selected wavelength.
64 unifying physics of accelerators, lasers and plasma

In particular, lasers with ultra-short and ultra-high power

pulse have had their share of challenges. Ultra-short intense
pulses can cause nonlinear effects in the medium, while the
high-power pulses cause heating up of the amplifier medium.
These issues limit the repetition rate, power and efficiency of
laser systems. Some of the most powerful lasers fire just once
every few hours!

4.2.3 Laser repetition rate and efficiency

Let’s take a look at the challenges of high average power
lasers from the point of view of AS-TRIZ. The TRIZ-
formulated problem is as follows: as the intensity of the
pulsed laser light increases, it takes much more time for the
medium to cool down and be ready for its next use. The iden­
tified contradiction is between the parameter that needs to be
improved, INTENSITY, and the parameter that gets worse,
REPETITION RATE.
A general principle that can solve this contradiction can
be looked up in nature (see Fig. 4.15) or taken from AS-TRIZ,
where Principle 4, Volume-to-surface ratio, suggests changing
this ratio in order to alter the object’s characteristics (such as
its cooling rate or its fields).

15o C 22o C 25o C 40o C

FIGURE 4.15
The cat intuitively knows the inventive principle of changing her
surface-to-volume ratio depending on the external temperature.

4.2.4 Fiber lasers and slab lasers

Fiber laser technology is a perfect illustration of the inventive
principle of changing the surface-to-volume ratio.
Indeed, if the main technical issue with high average
power laser amplifiers is cooling the active medium, then
one would try to increase its surface, without changing its
volume, to enhance thermal exchange with the environment.
Fig. 4.16 clearly demonstrates that the typical geometry of
the active medium (when its length L is similar to its radius
R) is the most disadvantageous, in terms of cooling.
On the other hand, either increasing L/R towards fiber
geometry of the active medium, or decreasing L/R toward
FIGURE 4.16 slab geometry, would both have advantages in terms of the
Volume-to-surface ratio S/V surface-to-volume ratio S/V , thus making it easier to manage
in units of (2πV )1/3 vs L/R. their temperature regimes.
 synergies between accelerators, lasers and plasma 65

FIGURE 4.17
Schematic of a fiber laser and cross section of refractive index.

Correspondingly, the fiber laser and slab lasers are the tech­
nologies most suitable for achieving higher average power Fiber laser technology uses
laser pulses, higher repetition rate and higher wall-plug ef­ the principle of larger surface­
ficiency. The main components of a fiber laser are shown in to-volume ratio.
Fig. 4.17.

4.2.5 CPA — chirped pulse amplification

Invented for lasers by G. Mourou and D. Strickland, chirped
pulse amplification (or CPA) is a process that stretches the
short pulse in time, amplifies the now-longer laser pulse and
then compresses the amplified pulse. This multi-step pro- CPA technique was origi­
cess relieves the optical component of laser amplifiers from nally developed for radar,
pulsed power, thus reducing nonlinear effects and avoid- while chirped pulses could
ing material damage. The stretching and compressing of the also be observed in nature,
pulses are based on the time–energy correlation property of e.g., in the voices of bats.
the laser pulse.

FIGURE 4.18
Schematics of CPA — chirped pulse amplification.
66 unifying physics of accelerators, lasers and plasma

The schematics of a laser stretcher and compressor are

presented in Section 4.3.4. The stretcher and compressor use
a pair of diffraction gratings and rely on the fact that the an­
gle of reflection from the grating depends on the wavelength,
which sends different colors along different paths. The spec­
tral width of the short laser pulse with duration τ is approxi­
mately given by Δf ∼ 1/τ. This finite spread of the spectrum
around the carrier frequency makes such laser pulses suitable
for spatial–spectral manipulation.
The schematics of CPA are shown in Fig. 4.18. The ini­
tial short pulse is provided by a short-pulse oscillator. The
first pair of gratings disperses the spectrum and stretches the
pulse by a factor of about a thousand (for visibility, the lon­
gitudinal extent of the pulses in Fig. 4.18 is shown qualita-
Amplification of chirped tively). After stretching, the pulse is long and has a low peak
pulses was used in radar and
power, which is thus safer for amplification. After passing the
is now used in lasers — this
trend from microwave to power amplifier, the pulse is sent to the second pair of grat­
optical range can be taken as ings, which reverses the dispersion of the first pair and com­
one of the generic principles presses the pulse, producing a high-energy ultra-short laser
of AS-TRIZ. pulse.
The invention of CPA was one of the factors that ulti­
mately pushed laser technology to such peak power levels
that these lasers became a possible competitor alongside par­
ticle accelerators.
4.2.6 OPCPA — optical parametric CPA
Another method of laser amplification is called OPCPA —
optical parametric CPA. Its principle is based on nonlinear
properties of crystals (typically barium borate or BBO crys­
V
tals) which, being subjected to radiation of wavelength ωs ,
generate radiation at two frequencies — ω1 and ω2 where, as
energy is conserved, ω1 + ω2 = ωs (as shown in Fig. 4.19).
In optical parametric amplification, the input consists of
FIGURE 4.19 two beams: the pump at ωs and the signal at ω1 . The OPA out-
Optical parametric genera- put is the amplified ω beam and weakened ω beam, plus an
1 s
tion in nonlinear crystals. additional idler beam at ω2 .

FIGURE 4.20
Optical parametric chirped pulse amplification — OPCPA.
 synergies between accelerators, lasers and plasma 67

The optical parametric amplifier system can be fed with a The main advantage of
frequency-stretched signal pulse, as illustrated in Fig. 4.20. OPCPA is that it works via
This makes the OPA system into a chirped pulse method a parametric process, i.e., no
known as OPCPA. energy is left in the nonlinear
The OPCPA method is versatile; it can work from CW to crystal and everything comes
femtosecond range in terms of pulse length, from UV to Ter­ out in the form of light. This
aHz in terms of light wavelength, and from mW to TW and is beneficial for high energy
or high mean power systems
PW in terms of the peak power.
since the thermal issues are
eliminated.
4.2.7 Plasma oscillations
Jumping from lasers back to plasma topics, let’s discuss the
process of energizing the plasma, i.e., creating oscillations in
plasma.
Imagine that there is a region in plasma where electrons
of density n shift with respect to the ions by a distance of x as
shown in Fig. 4.21. Applying Gauss’s law

1
E · dS = ρdV
∂Ω ε0 Ω

will yield the value of an electric field produced by displaced

charges
nex
E= (4.2)
ε0
Writing an equation for the electrons’ non-relativistic motion

d 2x ne 2 x
F=m = −eE = − (4.3)
dt 2 ε0

will then give us the expression for the oscillation frequency:

1HJDWLYH 3RVLWLYH
ne 2
ωp2 = (4.4)
ε0 m

Recalling the advice to express the end result in a form inde­

pendent of the systems of units, we use [

1 e2
re =
4πε0 me c 2

to rewrite the oscillating frequency or the plasma frequency as:

ωp2 = 4πnc 2 re (4.5)

,QWHJUDOFRQWRXU
We can also write a practical formula for f p = ωp /(2π): ,RQ
(OHFWURQ
f p ≈ 9000 n1/2 (Hz) (4.6)
FIGURE 4.21
where n is expressed in (cm−3 ). Plasma oscillations.
68 unifying physics of accelerators, lasers and plasma

4.2.8 Critical density and surface

Let’s take this moment to qualitatively consider the process
of laser plasma penetration and the corresponding factors of
critical density and critical surface.
When a laser hits a target or a dense gas, the target surface
or gas is heated and ionized, which forms plasma. Hot plasma
then starts expanding into the vacuum, creating a gradient of
plasma density as illustrated in Fig. 4.22, with a respective
gradient of plasma frequency.
Qualitatively, if the plasma frequency ωp in a particular
layer of plasma is larger than the laser frequency ω, then
the plasma electrons can move fast enough and can thus cre­
ate electric currents that will screen the fields of the laser
EM wave. Therefore, lasers can penetrate plasma only to the
point when
ωp < ω
The critical density is therefore
FIGURE 4.22
Laser penetration to plasma ω2
nc = (4.7)
and critical density. 4πc 2 re
In other words, the laser beam of frequency ω cannot pene­
trate areas with n > nc .

4.3 Manipulate
Let’s consider some of the ways we can manipulate beams,
laser pulses or plasma.
For particle beams the topics of interest include focus­
ing (weak, strong, chromaticity, aberrations); compressing;
cooling (electron, stochastic, optical stochastic, laser); phase
plane exchange; transverse stability, etc. For laser beams we
are interested in focusing; compression; phase locking; har­
monic generation, etc. Plasma-manipulation topics include
plasma focusing; Landau damping; and laser self-initiated fo­
cusing in the plasma channel.
We will touch on some of these in this section, and several
other topics will be discussed in Chapter 10.

4.3.1 Beam and laser focusing

Optics of light and optics of charged particles have a lot of
2a 0 a0 similarities. However, different names are sometimes used
for the same quantities in beam and light optics. To illustrate
this, consider a situation wherein the laser or particle beam
is focused into a point so that its minimum size at the waist is
equal to a0 (also called w0 in light optics). The distance from
0 the
√ waist point to the point where the beam size increases by
2 is called the Rayleigh length ZR in light optics and is equiv­
FIGURE 4.23 alent to the Twiss beta function β0 at the location of the waist
Beam or light in the focus. point in optics of charged particles (see Fig. 4.23).
 synergies between accelerators, lasers and plasma 69

4.3.2 Weak and strong focusing

Analogies between light and beam optics — as well as me­
chanical analogies (in the spirit of TRIZ or synectics) — can
help us build up an intuitive understanding of complex phe­
nomena. Let’s illustrate this in an example of weak versus
strong focusing.
Weak focusing can be brought about by bending dipoles
with the same gradient all along the perimeter of an accel­
erator (refer to Fig. 4.24). In this figure, the weak focusing
is compared to the motion of a ball in a gutter. Looking at
Eq. 2.19 and Eq. 2.20, one can conclude that there are condi­
tions when the beam is focused simultaneously in the x and y
planes (due to the presence of an additional focusing term in
bends in the x plane). However, as follows from these equa­
tions or as Fig. 2.8 suggests, such focusing is weak — the spa­
tial period of particle oscillation in this focusing field is of the
order of the orbital circumference.
Weak focusing accelerators were mainly built in the early
days. One of the disadvantages of weak focusing accelerators FIGURE 4.24
was the large transverse oscillations of particles, leading to Weak focusing.
wide apertures and correspondingly large and heavy mag­
nets. For example, the 10 GeV weak focusing Synchropha­
sotron built in Dubna in 1957 (the biggest and the most
powerful in its time), was registered in the Guinness Book
of World Records for housing the heaviest magnet system,
weighing 36,000 tons.
On the other hand, strong focusing can be generated via a
sequence of focusing–defocusing quadrupoles, with the over­
all effect being equivalent to focusing, if certain conditions
are satisfied. These conditions can be understood from the
gutter analogy shown in Fig. 4.25 where the gutter is now
bent, first of all, much more strongly, and second, it is bent
up and down (see also Fig. 1.11). As clearly seen from the
picture, stable motion is possible in this event only if the par­
ticle passes the areas of the downward-bent gutter near the
center, in a way similar to what it would do in a FODO lat­
tice. CERN’s Proton Synchrotron, the first operating strong
focusing proton accelerator, reached 24 GeV in 1959. It is
constructed as a 200-m diameter ring, with a magnet weight
of 3,800 tons — weighing ten times less for twice the energy
of Dubna’s Synchrophasotron — resulting in a clear demon­ FIGURE 4.25
stration of the advantages of the strong focusing approach. Strong focusing.

4.3.3 Aberrations for light and beam

Following Newton, we know that sunlight consists of a spec­
trum of colors and that each different color focuses differently
— these are called chromatic aberrations. In precise opti­ A curved mirror, in contrast
cal devices, such as photo cameras, the chromatic aberrations to a lens, does not produce
need to be compensated. chromatic aberrations.
70 unifying physics of accelerators, lasers and plasma

For light, one uses lenses made from different materi­

&URZQ
als to compensate for chromatic aberrations, as illustrated in
)OLQW
Fig. 4.26. In this example, using a strong focusing lens made
from a low-dispersion crown glass coupled with a weak de-
focusing high-dispersion flint glass can correct the chromatic
aberrations.
For particle beams, chromatic aberrations are caused by
the dependence of the focusing strength on the beam energy.
There are no different materials and no specific Maxwell’s
equations in particle optics, therefore, other means have to
FIGURE 4.26 be used for chromatic compensation in accelerators.
Lenses made from different
glasses compensate chro­ (
'LSROH
matic aberrations. %HDP 0DJQHWV
'LUHFWLRQ
(!

FIGURE 4.27
Compensation of chromatic aberration by inserting nonlinear
sextupole magnets in a dispersive region.

Let’s consider chicane arrangements of four bending

dipoles as shown in Fig. 4.27. Such an arrangement creates a
dispersive region, where orbit position depends on the energy.
Let’s insert a sextupole magnet with normalized strength S
(or K ' ) into the dispersive area

1 δ2 B
S= (4.8)
2!Bρ δx2

The sextupole magnet produces the following effect (kick)

on the angle of the beam trajectory

x' = x' + S x2 − y 2 and y ' = y ' − S 2xy (4.9)

In the dispersive region, one needs to replace x with x +ηδ

where η is the dispersion, thus the sextupole kick will contain
the energy-dependent focusing since the terms in the equa­
tion below correspond to a quadrupole with an effective gra­
dient equal to S · η:

x' ⇒ S(x + ηδ)2 ⇒ 2S η xδ + .. (4.10)

y ' ⇒ −S 2 (x + ηδ) y ⇒ −2S η yδ + ..

Therefore, such an arrangement of nonlinear magnets placed
in a dispersive region can be used for chromatic correction in
the event of beam optics. One should also note that the terms
that are not shown in Eq. 4.3.3, containing x2 and δ 2 , are un­
wanted additional aberrations — geometrical aberrations and
higher-order chromatic aberrations. They often need to be
 synergies between accelerators, lasers and plasma 71

corrected, either by special optic arrangements and appropri­

ate placement of sextupole pairs, or by higher-order magnets.

4.3.4 Compression of beam and laser pulses

Energy-z correlation is used
Compression of laser pulses relies on the dependence of the
to compress/stretch either
light reflection angle from the grating on the light wave- laser pulse or particle bunch
length, in the case when the grating spacing is comparable to — this is one more general
the wavelength. This produces an optical system where the principle of AS-TRIZ.
path length through the system depends on the light’s wave­
length.

FIGURE 4.28
Laser pulse stretcher.

The combination of a pair of grated plates, lenses and

mirrors can create a laser pulse stretcher — as realized in
Fig. 4.28.
Equally, a laser pulse compressor can be arranged with a
pair of gratings, as shown in Fig. 4.29.

The optical telescope placed

inside the pulse stretcher is
needed in order to provide a
FIGURE 4.29 “negative distance” between
Laser pulse compressor. the gratings.
72 unifying physics of accelerators, lasers and plasma

Laser pulse compressor gratings are technological mar­

vels. In order to stay below the damage threshold of the grat­
ing material, the size of the plates has to be around a meter in
diameter for some of the highest power lasers. Taking into ac­
count that the typical space between grooves is around a mi­
crometer, we can imagine the high levels of accuracy needed
to produce such plates.
Charged particle beam compression is based on the same
principle of the path length dependence on the particle’s en­
ergy. Dispersion is created using dipole bending magnets and
the initial beam is arranged to have an E − z correlation, via
its acceleration off-crest of RF voltage.
(
(

'LSROH W
W (
0DJQHWV
, ,
%HDP
(!

W W
FIGURE 4.30
Bunch compressor.

Fig. 4.30 shows a bunch compressor and the phase space of

the beam before and after compression.

4.3.5 Beam cooling

Beam cooling methods are usually necessary for antiparticles,
e.g., antiprotons p̄, as they are produced on a target very “hot”
— with large emittance and energy spread. The cooling meth­
ods are intended to decrease beam emittances.
Electron cooling, invented by G.I. Budker and realized at
Novosibirsk, consists of creating a region along the orbit of
the beam where a “cold” electron beam will co-propagate to­
gether with a “hot” antiproton beam (see Fig. 4.31) at the
FIGURE 4.31
same velocity. Energy exchange between the beams will even­
Electron cooling.
tually result in cooling of the p̄ beam.
Another method of cooling the antiproton beams was pro­
posed by S. van der Meer and realized at CERN. In this ap­
proach, a beam particle’s betatron oscillation is detected by a
pick-up electrode, and then the signal is amplified and sent
via a short path across the ring onto a kicker (see Fig. 4.32).
This will apply a kick to the same particle, resulting in a re­
duction of its oscillations. This method is called stochastic
cooling and was indispensable for ensuring the discovery of
W and Z bosons (Carlo Rubbia and Simon van der Meer, No­
bel Prize in Physics, 1984).
FIGURE 4.32 As it should be apparent from the description of stochas­
Stochastic cooling. tic cooling, this method can clearly have drawbacks in the
 synergies between accelerators, lasers and plasma 73

case of a large number of particles in the beam, as the pick­

up electrode would see signals from many particles, which
would then smear individual contributions.

4.3.6 Optical stochastic cooling

“Standard” stochastic cooling entails sampling of the beam
by electrostatic pick-up electrodes (see Fig. 4.33), and there­
fore its cooling rate is limited by the system bandwidth,
which is defined by pick-up length. An extension of the
method, optical stochastic cooling, uses optical pick-ups and
optical amplifiers as shown in Fig. 4.34, resulting in poten­
tial increases of bandwidth and of the cooling rate by several
orders of magnitude.

FIGURE 4.33
Standard stochastic cooling.

2SWLFDO
DPSOLILHU
4XDGUXSROH 'LSROH
ZLJJOHU ZLJJOHU

FIGURE 4.35
Layout of optical stochastic cooling system in an accelerator.

In optical stochastic cooling, quadrupole wigglers and dipole FIGURE 4.34

wigglers play the role of pick-up electrodes and kickers as Optical stochastic cooling.
shown in Fig. 4.35.
Beam cooling methods, and stochastic cooling in partic­
ular, are excellent examples that demonstrate the discussed
AS-TRIZ principle — the evolution of technical systems from
microwave frequencies into optical range.

4.4 Interact
We have reached the final step of the sequence of Create —
Energize — Manipulate — Interact.
Once the beam is accelerated, it can be used in a vari­
ety of ways — from creating a radiation (synchrotron, beta­
tron) source, a free electron laser, a collider, a spallation neu­
tron source, to using beams for particle therapy, industrial
or security applications, or energy applications in accelerator-
driven systems — ADS. Equally, amplified and compressed
laser pulses can be used in a variety of ways; in the context
of accelerator physics they can be used as a driver for parti­
cle acceleration, as the main component of a Compton X-ray
source or of a Photon collider, among other things.
We will touch on some of these further on in this text.
74 unifying physics of accelerators, lasers and plasma

EXERCISES
4.1 Chapter materials review.

The plasma density is 1017 cm−3 . A laser light of which wave­

length could still penetrate such a plasma? Also, estimate the

corresponding plasma frequency.

4.2 Chapter materials review.

Taking into account a) Fermat’s principle of the least amount

of time for light propagation through an optical system, and

b) the observation that the diffraction angle should approach

the law for reflection from a mirror as the wavelength be­

comes very short (and hence diffraction becomes less impor­
tant), explain qualitatively why an optical telescope is needed

inside the laser pulse stretcher, but not required in the laser

pulse compressor.

4.3 Analyze inventions or discoveries using TRIZ and AS-TRIZ.

Analyze and describe scientific or technical inventions de­

scribed in this chapter in terms of the TRIZ and AS-TRIZ ap­
proaches, identifying a contradiction and an inventive prin­
ciple that were used (could have been used) for these inven­
tions.

4.4 Developing AS-TRIZ parameters and inventive principles.

Based on what you already know about accelerator science,

discuss and suggest the possible additional parameters for

the AS-TRIZ contradiction matrix, as well as the possible ad­

ditional AS-TRIZ inventive principles.

 5
Conventional Acceleration

5.1 Historical In this chapter we will discuss conventional acceleration. Be­

introduction 75 ginning with a historical introduction, we will then touch on
5.2 Waveguides 81 waveguides, resonant cavities fed by RF power generators,
5.3 Cavities 86 and conclude with an overview of the basics of linacs and
5.4 Power sources 90 longitudinal dynamics.
5.5 Longitudinal
dynamics 94

5.1 Historical introduction

The “Livingston plot” presented in Fig.1.6 of Chapter 1 over­
sees the development of accelerators over the past several
decades. These days, we refer to them as conventional accel­
erators. These accelerators are also human-made, in contrast
with, for example, a cosmic accelerator, which has the capac­
ity to produce particles with energies exceeding 1020 eV , or
lightning, which produces fairly minimal acceleration.

FIGURE 5.1
A cathode ray tube TV as an example of an accelerator.

To contrast the composition of older accelerators with that

of a contemporary one, we will first look at an accelerator
familiar to everyone — a cathode ray tube TV (Fig.5.1), which
is a common example of an accelerator with all subsystems
present: a beam source, focusing and steering, an accelerating
region and an interaction or a target area (i.e., a fluorescent
screen).

DOI: 10.1201/b18696-5 75
76 unifying physics of accelerators, lasers and plasma

8
5.1.1 Electrostatic accelerators
& In an inspirational push in favor of accelerator development,
&
8
Rutherford spoke at The Royal Society in 1928, lamenting,
“I have long hoped for a source of positive particles more
&
&
8
energetic than those emitted from natural radioactive sub­
stances.”
& The first accelerator developed was the Cockcroft–Walton
& generator which is based on a system of multiple rectifiers
(Fig.5.2). In this generator, voltage generated by the cascade
a 8 circuit
2πI 2 3 1 2 1
Utot = 2Un − n + n + n (5.1)
ωC 3 4 12
FIGURE 5.2
Cockcroft–Walton genera­ depends on the number of cells n, as well as generated cur­
tor. rent I , capacity C and frequency ω.
The Cockcroft–Walton accelerator helped to make Ruther­
&ROOHFWLQJ
0HWDOVSKHUH EUXVK
ford’s dream a reality. In the first-ever transmutation exper­
iment, the accelerated 700 keV protons were sent onto a
lithium target, resulting in the production of helium.
Practical reasons (i.e., size of the device, performance of
capacitors and diodes) limited the generated voltages and
corresponding energies of the accelerated particles up to
about ∼4 MV . Cockcroft–Walton accelerators can typically
generate beam currents of several hundred mA with CW or
pulsed particle beams of few μs pulse lengths.
5XEEHU
Another example of an electrostatic accelerators is the
EHOW “Van der Graaf” (see Fig.5.3), where a metal brush deposits
0HWDO
EUXVK charges onto a rubber belt, carrying the charges into a metal
sphere where they are later collected by another brush. The
exchange of charges is performed with help from a discharge
between the sharp tips of the needles of the brushes and the
belt.
The ion source in the “Van der Graaf” accelerator is lo­
FIGURE 5.3 cated inside of the metal sphere, and is charged to a high
Van der Graaf accelerator. voltage. An electrical generator (mechanically connected to

&KDUJHH[FKDQJHWDUJHW 1HJDWLYHLRQVRXUFH

$FFHOHUDWLQJWXEH 6WHHOSUHVVXUHWDQN

  

6WHHULQJPDJQHW
&KDUJHGEHOW +LJKYROWDJH *UDGLHQWULQJV
FIGURE 5.4
Tandem electrostatic accelerator.
 conventional acceleration 77

the upper axis of the rubber belt pulley) produces electrical

power needed to operate the ion source or any other inter­
nal control electronics. The ions produced by the source are
accelerated on their way down from the sphere.
With any electrostatic accelerator, it is difficult to achieve
high energies due to limitations determined by the size of the
vessels. In particular, the highest recorded energy created via
the Van der Graaf accelerator was around ∼25 MeV.
A version of the Van der Graaf accelerator that can pro­
duce twice higher energy of the particles is called a tandem
accelerator (Fig.5.4). In this case, negative ions are accelerated
and, when they reach the charged sphere, they pass through
a foil or gas target to perform their charge exchange. Now
positive, ions continue to accelerate towards ground poten­
tial and thus reach twice the voltage of the charged sphere.
An additional advantage of the tandem accelerator is having
its ion source located at ground potential, which significantly
simplifies its operation and maintenance.
Another version of the charge-carrying mechanism is re-
alized in a pelletron, where instead of the rubber belt, metal FIGURE 5.5
pellets are connected by non-conductive links into a chain Pelletron charging mecha­
(see Fig.5.5). nism.

“Particles should be con­

5.1.2 Synchrotrons and linacs strained to move in a circle
of constant radius thus en-
Synchrotrons can accelerate particles to much higher energies
abling the use of an annular
than electrostatic devices can. Their name is derived from the
ring of magnetic field...which
process of synchronous change of the magnetic field of bend-
would be varied in such
ing magnets according to the growing energy of the acceler-
a way that the radius of
ated beam (see the quote at right). curvature remains constant
Many modern accelerators are synchrotrons (e.g., the as the particle gains en-
Large Hadron Collider at CERN is a synchrotron). These ergy through successive ac­
celerations by an alternat­
types of accelerators can reach very high energies, limited
ing electric field applied be-
only (especially for electrons) by synchrotron radiation and
the cost of their construction. tween coaxial hollow elec­
trodes.” Mark Oliphant, Oak
As we will momentarily explain, time-varying fields are
Ridge, TN, 1943.
the necessary conditions for acceleration of charged particles
to high energies, and in particular for overcoming the limi­
tations of the maximum achieved energy in the electrostatic
accelerators.
Both linear and circular accelerators use EM fields oscil­
lating in resonant cavities to achieve acceleration. In circular
accelerators, particles follow an orbit guided by a magnetic
field and return to the same accelerating cavity on every turn,
while in linac accelerators the particles follow a straight path
through a sequence of cavities, as shown in Fig.5.6.
Powerful radio-frequency (RF) systems produce the re­
quired powerful electric fields in the resonant cavities. Ac­
celerators of this kind progressed in large part due to the FIGURE 5.6
telecommunications industry, which drove the development Synchrotron and linac.
78 unifying physics of accelerators, lasers and plasma

of power systems with frequencies ranging from a few MHz

to several GHz.
Recalling Maxwell' s equation and its integral form
∂B d
∇×E = − or E · dt = − B · dS
∂t ∂Σ dt Σ

we see that if the particle moves on an enclosed orbit, like in

a synchrotron, finite acceleration (i.e., nonzero contour inte­
gral of E) would not be possible without a time-dependent
magnetic field B. On the other hand, time-dependent mag­
netic flux can provide acceleration when — either in a linac
or a synchrotron — the EM field oscillates in the resonant
cavity and the particles receive a finite energy increment at
each pass through the cavities.

5.1.3 Wideröe linear accelerator

In 1925, Ising inspired the new technological branch of ac­
celerators by realizing that limitations imposed by corona
formation and discharge in electrostatic accelerators can be
overcome with the use of alternating voltages. This was car­
ried further by Wideröe who, in 1928, performed the first
successful test of the linac based on this principle.
The Wideröe linac has a series of drift tubes arranged
along its beam axis and connected with an alternating high
frequency RF voltage: V (t) = Vmax sin(ωt) as shown in Fig.5.7.
During the first half of the RF period in the acceleration pro­
cess, voltage applied to the first drift tube accelerates charged
particles emerging from the ion source. Particles then enter
the first drift tube and pass through it. Meanwhile, the direc­
tion of the RF field reverses without affecting the particles —
seeing as the drift tube acts as a Faraday cage and shields the
particles from external fields. When the particles reach the
gap between the first and the second drift tubes, they accel­
erate and the process repeats in the following gaps as well.

,RQVRXUFH 'ULIWWXEHV
O O O O OL OL

    L L

%HDP

5)JHQHUDWRU

FIGURE 5.7
Wideröe linear accelerator.
 conventional acceleration 79

¨ linac after passing

The energy of the particle in Wideroe
the i-th drift tube is

Ei = iqVmax sin Ψ0 (5.2)

where q is the charge of the particle and Ψ0 is average phase

of the RF voltage that particles feel as they cross the gaps
(see Fig.5.8). As we can see, the energy is proportional to the
number of stages i passed by the particle. Furthermore, the
largest voltage to ground in the entire system never exceeds
Vmax . This lets us reach high energies without using voltage
levels, which can cause electrical breakdown.
The accelerating gaps between drift tubes in the Wideroe ¨
linac must increase in sync with the monotonically increas­
ing velocity of the particle. Taking into account that the half-
period of RF τRF /2 should correspond to a particle passing
with velocity vi through one drift section, we write the dis­
tance between i-th and (i + 1)-th gaps as

vi τRF 1 iqVmax sin Ψ0

ei = = FIGURE 5.8
2 f RF 2m Voltage in Wideröe linac.
which we expanded using Eq.5.2.

5.1.4 Alvarez drift tube linac

¨
The Alvarez linac is conceptually quite similar to the Wideroe
linac. It differs in that, in an Alvarez linac, the accelerating
voltage at individual drift tubes is created by an RF wave in a
container (a tank made of a good conductor such as copper),
in which the drift tubes are located (see Fig.5.9). The drift
tubes may have magnets installed inside to focus the beam
during acceleration.

,RQVRXUFH &RSSHUHQYHORSH

%HDP

5)RVFLOODWRU
FIGURE 5.9
Alvarez drift tube linac.

The drift tube linacs are still used (particularly in

hadrons), but they are being replaced by better-performing
RFQ-structures (Fig.5.10). Due to periodic transverse varia­
tions of their shape, such structures allow for the creation of
not only accelerating fields, but also focusing fields. Recent
 80 unifying physics of accelerators, lasers and plasma

progress via 3D computer simulations of RF fields and beam

dynamics resulted in a widespread use of RFQs.
Most of the complicated linac designs discussed above
are applicable to hadrons, which (in practically achievable
accelerating gradients) become relativistic rather slowly, af­
ter a hundred or so meters of acceleration. Correspondingly,
the size and shape of accelerating cavities need to vary along
the hadron linac, so as to match their increasing velocity. For
electrons, which have already become highly relativistic after
a few tens of cm of acceleration, relatively simpler linac de­
sign is possible, in which cavities of the same size and shape
are regularly placed along the acceleration path.
FIGURE 5.10
RFQ structure. 5.1.5 Phase focusing
As we have seen in the previous section, the energy trans­
ferred to particles in a drift tube linac depends on the voltage
amplitude Vmax and phase Ψ0 . We should note that a small
deviation of the nominal voltage Vmax would result in a par-
ticle velocity that no longer matches the design velocity fixed
by the length of drift sections. In this case, the particles would
undergo a phase shift relative to the RF voltage and the syn­
chronization of particle motion with respect to RF field will
be eventually lost.
The system can be made to self-adjust to such small de­
viations if we use Ψ0 < π/2 so that the effective accelerating
voltage is Veff < Vmax , as shown in Fig.5.8. In this instance, if
a particle gains too much energy in the preceding stage and
is travelling faster than the ideal particle and arrives at the
next acceleration stage earlier, it will then feel the average RF
phase Ψ = Ψ0 − ΔΨ and will be accelerated by the voltage
V 'eff = Vmax sin (Ψ0 − ΔΨ) < Vmax sin Ψ0 (5.3)
which is below the ideal voltage. The particle will thus gain
less energy and will slow down and return to the nominal
velocity. The process will then repeat and the particles will
continuously oscillate about the nominal phase Ψ0 , exhibit­
ing the phenomena of phase focusing. If dissipation is present
in the system (such as SR damping), these oscillations would
eventually be forced to decay and the particles would gather
around the synchronous phase. A similar effect of bunching
can happen without damping, due to acceleration.

5.1.6 Synchrotron oscillations

The phase focusing of particles in accelerators manifests it­
self as the periodic longitudinal particle motion about the
nominal phase, and is called synchrotron oscillation.
In circular accelerators, as the ideal particle encounters
the RF voltage at exactly the nominal phase on each revolu­
tion, the RF frequency ωRF must be an integer multiple of the
 conventional acceleration 81

revolution frequency ωrev

h = ωRF /ωrev (5.4)

where h is called the harmonic number of the ring: an inte­

ger chosen to allow matching the frequency of available RF
generators.

  








 


 



  


 




FIGURE 5.11
Synchrotron oscillations.

For relativistic particles, the phase focusing is enabled by

the dependence of the circular path on energy, as shown in
Fig.5.11. The stable point in this case is on the declining slope
of the sine wave, as particles with Δp/p > 0 will travel on the
longer path, come to the RF cavity in the next turn somewhat
later, and ultimately lower their acceleration. As the reader
can already guess, for a particle that is not yet relativistic and
whose velocity depends on energy, the situation is quite dif­
ferent — we will discuss this further later on in this chapter.

5.2 Waveguides
Prior to engaging in detailed discussion of RF accelerat­
ing cavities, let’s introduce — via simple considerations and
analogies — the basic properties of waveguides, as they pro­
vide important and intuitive understanding applicable for
accelerating structures.

5.2.1 Waves in free space

The velocity of an EM wave in a vacuum and in a medium is
1 1
vacuum : v = c = √ , medium : v = √ (5.5)
ε0 μ0 ε0 εr μ0 μr

where εr is the dielectric constant and μr is the magnetic per­

meability of the medium.
The amplitudes of electric and magnetic fields in an
EM wave in a vacuum are exactly the same if expressed in
Gaussian-cgs units (demonstrating the naturalness of this
system of units) and relates as E = cB in SI units.
82 unifying physics of accelerators, lasers and plasma
\
A plane EM wave with transverse electric and magnetic fields
(
(TEM wave) propagating in free space in x-direction is shown
] in Fig.5.12. The quantity
% %
P = (E × B) /μ0 (5.6)
(
[
is called the Poynting vector and equals to the local power flux.

FIGURE 5.12
TEM wave in free space. 5.2.2 Conducting surfaces
In the derivations presented in this section, we discuss the
behavior of EM waves bounded in metal boxes; therefore, we
need to recall the boundary conditions of a wave at a perfectly
conducting metallic surface.
On the surface of a perfect conductor, the tangential com­
ponent of an electric field E|| and the normal component of a
magnetic field B⊥ will vanish, as illustrated in Fig.5.13.
A non-ideal surface with conductivity σ is characterized
by skin depth; an EM wave entering a conductor is dampened
to 1/e of its initial amplitude at the depth
1
δS = . (5.7)
πf μ0 μr σ
This allows us to introduce the notion of surface resistance

FIGURE 5.13 1
Rsurf = (5.8)
Boundary conditions on per­ σ δS
fectly conducting surfaces. which plays an important role in determining performance
of accelerating cavities.

5.2.3 Group velocity

In preparation to discuss dispersion properties of waveguides
and RF structures, let us recall the derivation of group velocity
— the propagation velocity of energy (and information) in an
EM wave.
Consider the interference between two continuous waves
of slightly different frequencies ω ± dω and wavenumbers k ±
dk (see Fig.5.14):
E = E0 sin [(k + dk) x − (ω + dω) t]
FIGURE 5.14 +E0 sin [(k − dk) x − (ω − dω) t]
(5.9)
Two-wave interference. = 2E0 sin [kx − ωt] cos [dk x − dω t]
= 2E0 f 1 (x, t) f 2 (x, t)
The last line of Eq.5.9 contains two functions, f1 and f2 .
The first function corresponds to a continuous wave with the
mean wavenumber and frequency: f 1 (x, t) = sin [kx − ωt]. In
this wave, any given phase is propagated such that kx − ωt
remains constant, which gives us the equation for the phase
velocity of the wave:
ω
vp = (5.10)
k
 conventional acceleration 83

The same can be obtained by requesting the convective deriva­ Convective derivative — the
tive (∂/∂t + vp ∂/∂x) of f 1 to be equal to zero, which again re­ term originates from fluid
sults in mechanics — is the deriva­
∂f (x, t) /∂t ω tive taken with respect to a
vp = − 1 =
∂f 1 (x, t) /∂x k moving coordinate system.
The second function in Eq.5.9 describes the evolution of
the envelope of the pattern: f 2 (x, t) = cos [dk x − dω t]. Again,
any point in the envelope propagates such that the quantity
dk x − dω t remains constant and therefore its velocity, i.e.,
the group velocity, is given by

∂f 2 (x, t) /∂t dω
vg = − = (5.11)
∂f 2 (x, t) /∂x dk

This prepares us for discussion of the notions of dispersion

and group or phase velocity of a waveguide.

5.2.4 Dispersion diagram for a waveguide

We will start our conversation regarding wave propagation
down a waveguide using two extreme cases.
First of all, if the wavelength λ of an EM wave in free
space is much shorter than the transverse size a of the wave­
guide λ « a (as shown in Fig.5.15, case (i)), then the wave­
guide does not matter, and we expect the dispersion at large
ω to approach the equation for free space (i.e. ω/c = k). With
the goal of deriving the dependence of frequency ω against
wavenumber k = 2π/λ in a waveguide, let us place a corre­
sponding segment on a waveguide dispersion curve at a high
frequency (Fig.5.16, case (i)).

D 
L LL

FIGURE 5.15
Waves in a waveguide, two extreme cases.
F
Another extreme case is shown in Fig.5.15, (ii), when half L
of a wavelength in free space equals the waveguide trans­
verse size. As can be seen from this diagram, the longest
wavelength for which the boundary conditions at a perfectly LL Y F
conducting surface of the waveguide can still be satisfied, is
given by λ/2 ≤ a. This defines the cut-off parameters λc = 2a
or ωc = πc/a; that is, waves with wavelengths longer than λc
cannot propagate in the waveguide.
As Fig.5.15 (ii) suggests, the case of ω = ωc corresponds to FIGURE 5.16
an infinite wavelength in in the waveguide, or k = 0. We thus Dispersion of a waveguide,
plot a corresponding point in Fig.5.16. two extreme cases.
84 unifying physics of accelerators, lasers and plasma

In the intermediate region of frequencies, the dispersion

curve should connect the point (ii) and region (i) in Fig.5.16.
Looking at the wave in a waveguide from a simple geomet­
rical point of view (Fig.5.17) and considering corresponding

D similar triangles, one can write
 ω 2 ωc 2

= k2 + (5.12)

c c
FIGURE 5.17 thus describing dispersion of the wave in a rectangular
Intermediate case. waveguide, graphically shown in Fig.5.18.

F
N

Y F
F

N 
FIGURE 5.18
Dispersion of a waveguide.

Eq.5.12 suggests that, for any wavenumber k, the frequency

is always greater than the cut-off frequency. Looking at the
slope (i.e. derivative dω/dk = vg ) of the curve in Fig.5.18 we
can also observe that the longer the wavelength or lower the
frequency, the slower the group velocity, and at the cut-off
frequency no energy flows along the waveguide.
Eq.5.12 and Eq.5.10 also help find that

vp vg = c 2 (5.13)

which tells us that, in a waveguide, the phase velocity is al­

ways larger than the speed of light.

5.2.5 Iris-loaded structures

We can conclude from the previous section that acceleration
in a waveguide is not possible because the phase velocity of
the wave exceeds that of light. Particles that travel slower
than the wave would be periodically accelerating or decel­
erating, achieving zero net acceleration when averaged over
a long time interval.
In order to make the acceleration possible, one needs to
modify the waveguide to reduce the phase velocity to an ap­
propriate value below the speed of light, so as to match the
velocity of the particle.
Reduction of the phase velocity can be achieved by using
iris-shaped screens installed into the waveguide with a con­
stant step along the axes shown in Fig.5.19.
 conventional acceleration 85

K
E
G D

FIGURE 5.19
Iris-loaded accelerating structure.
)
The dispersion relation in a waveguide ω = c kz2 + (2π/λc )2
changes due to installation of the irises. Qualitatively, behav­
ior at low k resembles that of the waveguide curve, defined
by the smallest diameter of the irises. At higher k, with the
installation of irises, the curve flattens off and crosses the
boundary of vϕ = c in the region of kz = π/d as illustrated
in Fig.5.20.

8QGLVWXUEHG
VWUXFWXUH

!F 'LVFORDGHG
VWUXFWXUH

F

F N
G
FIGURE 5.20
Qualitative behavior of dispersion curve in iris-loaded structures.

With a properly selected iris separation d and other pa­

rameters of the irises, the phase velocity of the iris-loaded
structure can be set to any value, making it suitable for parti­
cle acceleration with arbitrary v/c.
An extended version of the dispersion diagram curve of
an iris-loaded structure can also be constructed while noting
that adding a multiple of 2π/d to the wavenumber k as

2nπ
kn = k0 +
d
would still satisfy the periodic boundary conditions at the
86 unifying physics of accelerators, lasers and plasma

irises. Therefore, the dispersion curve repeats itself at space

harmonics corresponding to different integers, n, as shown in
Fig.5.21.

F 
N

  

    N
FIGURE 5.21
Extended dispersion diagram of an iris-loaded structure.

The first rising slope of the dispersion curve shown in

Fig.5.21 is usually used for acceleration.

5.3 Cavities
In this section, we will consider general properties of reso­
nant cavities, their quality factors, shunt impedance, and will
introduce the definition of the resonance modes.

5.3.1 Waves in resonant cavities

In preparation for a discussion about the resonance modes
in the cavity, we first recall a general solution of the wave
equation, which can be written as

W (r, t) = Aei(ωt+k·r) + Bei(ωt−k·r) (5.14)

This describes the sum of two waves — one moving in one

direction and the other in the opposite direction.
In the case wherein the wave is totally reflected from a
conductive surface, both amplitudes need to be the same, i.e.,
A = B, and we can therefore rewrite Eq.5.14 as

W (r, t) = Ae iωt (e ik·r + e−ik·r ) = 2A cos(k · r)e iωt (5.15)

This equation describes the field configuration with a static

in time amplitude 2A cos (k · r), therefore corresponding to a
standing wave.
Consider now that the waveguide we discussed earlier has
a finite length e and its entrance and exit are closed by two
conducting surfaces, forming a resonance cavity. The reso­
nant wavelengths of this cavity can be determined noting that
a stable standing wave can form in this fully enclosed cavity
 conventional acceleration 87

if we assume that the following condition is satisfied

λz
e=q with q = 0, 1, 2, ... (5.16)
2
Therefore, only certain well-defined wavelengths λ are
present in the cavity.
Near the resonant wavelength, the resonant cavity be­
haves like an oscillator with a high quality factor Q, allowing it
to build up high voltages that can be used for particle acceler­
ation. The cavities are often modeled as electrical oscillators,
with their Q-value determined by losses of equivalent indi­
vidual coils, capacitors and resistances of the circuit model.

5.3.2 Pill-box cavity

An enclosed section of a waveguide (either rectangular or
cylindrical) forms the simplest RF cavity, called a pill-box cav­ K
ity. ]
The conventionally accepted classification of the modes
in pill-box cavities separates the cases of transverse electric or
TE modes (zero electric field along the axis) and transverse U 
magnetic or TM modes (zero magnetic field along the axis).
As Eq.5.16 suggests, many modes can exist in a cavity, as
defined by the corresponding dimension of the pill-box, and
the integer number of the mode. The corresponding integer
indexes are used to identify a particular mode. In a rectan­ FIGURE 5.22
gular pill-box, a mode can be called TEklm or TMklm where Cylindrical pill-box cavity.
the integer indexes indicate the number of half-wavelength
variations across the corresponding dimension (x, y, z) of the
cavity.
Cylindrical pill-box cavities are very common in acceler-
ators. An example of a cylindrical pill-box cavity with holes
for the beam is shown in Fig.5.22. The classification of the
modes in cylindrical cavities is very similar, TEklm or TMklm , 70
but in this case the indexes refer to the polar coordinates (ϕ,
r and z).
Examples of modes in cylindrical pill-box cavities, with ]
longitudinal electric fields and no variation over ϕ (TM0lm ),
which are therefore suitable for use as an accelerating cavity,

are shown in Fig.5.23. 70

5.3.3 Quality factor of a resonator ]

The quality factor of a resonator — Q — is defined as the ratio
of the energy stored in the cavity to the energy dissipated per
oscillating cycle, divided by 2π FIGURE 5.23
Examples of pill-box cylindri-
Ws Ws cal cavity modes with elec-
Q= =ω (5.17)
Wd Pd tric field lines shown.
88 unifying physics of accelerators, lasers and plasma

Here, Ws is the energy stored in the cavity, Wp is the energy

dissipated per cycle divided by 2π, Pd is the power dissipated
in the cavity walls and ω is the frequency of the cavity.
The stored energy over the cavity volume is
ε0 μ0
Ws = |E|2 dv + |H|2 dv (5.18)
2 2
where the first integral is the energy stored in the E-field and
In vacuum H = B/μ0 in SI the second integral corresponds to the energy in the H-field.
units. In an EM wave in space or a cavity, the energy oscillates back
and forth between these two contributions.
The losses in the cavity are calculated by taking into ac­
count the finite conductivity σ of the cavity walls. Since the
linear density of the current j along the walls of a perfect con­
ductor can be written as

j = n×H (5.19)

where n is the vector normal to the surface, we can therefore

equate the power dissipated in the cavity walls to
Rsurf
Pd = |H|2 ds (5.20)
2 s

where the integral is taken over the inner surface of the con­
ductor, and the surface resistance is given by Eq.5.8.

5.3.4 Shunt impedance — Rs

The so-called shunt impedance Rs relates the accelerating volt­
age V to the power that needs to be fed into the cavity to
compensate for the dissipation in the walls Pd .
The accelerating voltage along the path followed by the
beam in an electric field Ez is

V= Ez (x, y, z) de (5.21)
pass

and is taken as peak-to-peak value. The shunt impedance is

then defined as
V2
Rs = (5.22)
2Pd
and is another important characteristic of an accelerating
cavity.

5.3.5 Energy gain and transit-time factor

The energy gain of a particle as it travels a distance through
the accelerating structure depends only on potential differ­
ence crossed by particle:
.
U = K PRF lRs (5.23)
 conventional acceleration 89

where PRF is the RF power supplied to the cavity, l is

the length of the accelerating structure, Rs is the shunt
impedance and K is a correction factor (typically ≈ 0.8).
In accelerators such as drift tube linacs, a so-called
“transit-time factor” also plays a role and needs to be taken
into account in order to evaluate the average energy gain.
Consider an accelerating gap corresponding to the space  ]
between drift tubes in a linac structure (Fig.5.24). The accel­ 
erating field in this gap is uniform along the axis and depends (]
sinusoidally on time Ez = E0 cos (ωt + φ), where the phase φ
refers to the particle in the middle of gap z = 0 at t = 0. The
field varies as the particle traverses the gap, making the cav­
ity less efficient and the resultant energy gain only a fraction
of the peak voltage. FIGURE 5.24
The transit-time factor Γ is the ratio of the energy actually The RF gap — space between
given to a particle passing the cavity center at the peak field entrance and exit irises of
to the energy that would be received if the field were con­ cavity resonator in drift tube
stant with time at its peak value. Taking into account that the linac.
energy gained over the gap G is:
+G/2
sin (ωG/2βc)
V= E0 cos (ωt + φ) dz = E0 G cos φ (5.24)
ωG/2βc
−G/2

we write the following expression for the transit-time factor

Γ = sin (ωG/2βc)/ (ωG/2βc).

5.3.6 Kilpatrick limit

The performance of any normal conducting accelerating
structure depends on its susceptibility to RF breakdown
(which can occur at very high fields). Empirically derived
around 1950, the Kilpatrick limit expresses the relation be­
tween the accelerating frequency and maximum achievable
accelerating field:

f [MHz] = 1.64 Ek2 e−8.5/Ek (5.25)

where Ek is expressed in [MV/m] and is depicted by the lower

curve in Fig.5.25.
Significant efforts and technological developments in­
tended to improve surface quality and cleanness have re­
sulted in a considerable increase of achievable accelerating
gradients. In particular, Wang and Loew’s empirical formula,
devised in 1997, suggests the following behaviors:

E [MV /m] = 220f 1/3 (5.26)

where f is expressed in [GHz] — shown by the upper curve

in Fig.5.25.
The E ∼ f 1/3 dependence in Eq.5.26 was, for a long time,
90 unifying physics of accelerators, lasers and plasma

an inspiration and a driving force for developing higher gra­

dients acceleration at higher (multi-tens of GHz) frequencies.
This dependence, however, eventually was not confirmed for
the practical parameters of the accelerator RF pulses, where
it was observed that in these regimes the maximum gradients
appear to be rather independent of the frequency.

(09P







    

I*+]
FIGURE 5.25
Breakdown Kilpatrick limit (lower curve) and Wang–Loew limit
(upper curve).

5.4 Power sources

Sinusoidal power ranging from a few kW to a few MW is
needed to drive the accelerating structures. This is commonly
achieved by using RF power amplifiers such as triodes and
tetrodes (operating from a few MHz to a few hundred MHz),
inductive output tubes (suitable for CW applications, provid­
ing tens of kW at a high efficiency) and klystrons (which typ­
ically operate above a few hundred MHz) and have proven to
be the most effective power generators for accelerator appli­
cations at higher frequencies.

5.4.1 IOT — inductive output tubes

The inductive output tube, or IOT, (invented by Andrew Haeff
around 1939) is based on the principle that a toroidal cavity
surrounding an electron beam of oscillating intensity could
extract power from the beam without intercepting the beam
itself.
The oscillating EM fields carried with the beam excite the
modes found inside the toroidal cavity, which allows RF en­
ergy to be transferred from the beam to a waveguide or coax­
ial cable connected to the resonator via a coupling loop.
A schematic design of an IOT is shown in Fig.5.26.
 conventional acceleration 91

FIGURE 5.26
Schematic of an inductive output tube.

5.4.2 Klystron
Building on the success of the IOT, Sigurd and Russell Varian
added a cavity resonator at the beginning of the beamline in
order to provide a signal input for the inductive output tube.
This input resonator acts as a pair of control grid electrodes
performing velocity modulation in the electron beam. This re­
sults in bunching developed in the beam by the time it arrives
at the output resonator. In the latter the amplitude variation
will be converted to energy extracted from the beam. The Var­
ian brothers called their invention a klystron.

FIGURE 5.27
Schematic of a klystron.
92 unifying physics of accelerators, lasers and plasma

A schematic design of a klystron is shown in Fig.5.27.

Klystrons are similar to small linear accelerators, and op­
erate as follows. First, electrons are emitted from a round
cathode, which has a large surface area. Electrons are then
accelerated by a voltage of a few tens of kV. A round beam is
formed with a current from a few amperes to tens of amperes.
The beam is focused by near-cathode electrodes and is further
Permanent magnet focusing focused by the solenoid field (which is essential to ensuring
has been used in klystrons effective beam transportation). The particles leaving the cath­
with the intention to increase ode pass through an input cavity operating at TM011 mode,
their wall-plug efficiency. which is fed from an external pre-amplifier.
The klystron output power is given by

Pklystron = ηU0 Ibeam (5.27)

where U0 is the klystron supply voltage (e.g., 350 kV), Ibeam is

the electron beam current (e.g., 420 A) and η is the klystron
efficiency (e.g., 45%). The numbers given above as examples
correspond to the SLAC 5054 klystron running at 2.856 GHz.
The klystron efficiency (which typically ranges from 45% to
65%) is one of the most important parameters and is the sub­
ject of continuous innovations.
IOTs, klystrons and similar RF power generators — while
still extremely popular — are gradually being replaced by
solid-state devices, for frequencies lower than 1 GHz in par-
The trend of RF power sources ticular. The solid-state RF power systems have the advan­
technology development is a tage of compactness, higher efficiency and also reliability. The
good illustration of TRIZ modular design of solid-state RF power sources, when each
conclusions about the evolu- module contributes only a small fraction of the total power,
tion of technical systems. makes it possible to significantly increase the reliability of
the entire system.

5.4.3 Magnetron
Magnetron is an RF power source popular in particular for
CW applications. In a magnetron the cylindrical cathode is
located in the center and magnetic field is applied along
the axis of the device. When the electrons are moving from
the cathode to the anode, the magnetic field turns them and
makes them move on spiral trajectories, creating azimuthal
variations of the electron trajectories.
At certain parameters (of the voltage and magnetic field)
the azimuthal variations of the electron trajectories will
match the azimuthal spacing of the cavities arranged around
the perimeter on the anode body, exciting fields in the cavi­
ties. The exited fields in their turn will enhance the velocity
modulation and thus the spatial density modulation of the
electron beam. Increased modulations will further increase
the fields generated in the cavities. The magnetron amplitude
will grow until saturation (due to effects related to the elec­
tron beam space charge) and remain constant.
 conventional acceleration 93

FIGURE 5.28
Schematic of a magnetron.

A schematic of a magnetron is shown in Fig.5.28.

A magnetron, in contrast to klystrons and IOTs, is not
an amplifier, but a generator, which starts from initial noise.
Therefore, the phase of the RF power generated by a mag­
netron is arbitrary in principle.
Correspondingly, one of the difficulties of using mag­
netrons in accelerators when more than one magnetron is re­
quired is the difficulty of phase-locking those devices.

5.4.4 Powering the accelerating structure

An accelerating structure of a linac or other accelerator is typ­
ically fueled by pulsed power RF tubes — klystrons.
RF power is transported from a klystron in a different
mode than that used for acceleration. Therefore, the RF cou­

,ULVORDGHGVWUXFWXUH

7(ZDYH

70ZDYH

&RXSOLQJVORW

FIGURE 5.29
Feeding RF power into an accelerating structure. Field lines show
electric and magnetic fields of the corresponding cavity modes.
94 unifying physics of accelerators, lasers and plasma

pler of the acceleration cavity must ensure an appropriate

mode conversion.
It is typical to feed the RF power into the linac struc­
ture by a TE10 mode of EM wave in a rectangular waveguide.
The mode is then converted by a coupling slot into a cylin­
drical TM01 mode in the accelerating cavity, as illustrated in
Fig.5.29.

5.5 Longitudinal dynamics

In this section, we will discuss the basics of longitudinal dy­
namics in travelling and standing wave linacs, as well as in
synchrotrons.

5.5.1 Acceleration in RF structures

Particle acceleration in linacs is achieved with RF structures,
using EM modes with the electric field pointing in the longi­
tudinal direction (the direction of the charged particle’s mo­
tion). The RF electric field can be provided by either travelling
wave structures or standing wave structures.
The acceleration conditions demand that the phase veloc­
ity of the travelling wave and the particle velocity be equal,
so therefore disk-loaded structures are used to slow down the
phase velocity of the electric field vp < c to achieve synchro­
nism. In an appropriately synchronized travelling wave, the
bunch of charge particles experience a constant electric field

Ez = E0 cos (φ) (5.28)

as illustrated in Fig.5.30 (left plot).

FW 
(= F (= F F

 
 ]  ]

FIGURE 5.30
Acceleration in a travelling wave structure (left) and in a stand­
ing wave structure (right). The wave and particles’ position in
different moments of time are shown.

In a standing wave structure, the electromagnetic field is

the sum of two travelling waves running in opposite direc­
tions. Only the forward-travelling wave takes part in the ac­
celeration process.
The electric field that the particle bunch observes in a
 conventional acceleration 95

standing wave is a varying function of time

Ez = E0 cos (ωt + φ) sin (kz) (5.29)

as illustrated in Fig.5.30 (right plot).

The standing wave, despite its seeming disadvantage in
comparison with the travelling wave (in terms of the average
field seen by the bunch), is actually much more suitable in
certain cases, such as for superconducting cavities.

5.5.2 Longitudinal dynamics in a travelling wave

Consider a particle moving in the E field of a travelling wave

Ez = E0 cos(ωt − kz) (5.30)

with a phase velocity vp = ω/k. The equations that describe

the particle motion in the longitudinal plane in this field are

dpz dE
= eE0 cos(ωt − kz) and = eE0 ż cos(ωt − kz) (5.31)
dt dt
We will define the synchronous particle as
dEs
= eE0 vs cos ϕs (5.32)
dt
and for any other particle, we will use, as coordinates, the
deviations from the energy W and position u of the syn­
chronous particle

E = Es + W and z = zs + u (5.33)

Then, after changing variables to

ω
ϕ = kz − ωt = ϕs − u (5.34)
vs
we will obtain the system of equations for a particle motion
in a travelling wave:

dW dϕ ω W
= eE0 [cos ϕ − cos ϕs ] , =− 3 3 (5.35)
ds ds βs γs c mc 2

These describe the motion in the so-called “RF bucket” in a

longitudinal phase space (ϕ , W) and feature stable enclosed
trajectories as well as unstable trajectories. We will discuss
the phase space trajectories and motion in the “RF bucket”
in detail in the following section, after deriving similar equa­
tions for the case of acceleration in a synchrotron.

5.5.3 Longitudinal dynamics in a synchrotron

Acceleration in a synchrotron is provided by the longitudinal
electric fields generated in RF cavities placed on the orbit.
96 unifying physics of accelerators, lasers and plasma

FIGURE 5.31
Synchronous and lagging particles in a synchrotron ring.

A particle in an RF cavity changes its energy according to the

phase of the RF field in the cavity

ΔE = eV (t) = eVo sin(ωRF t + ϕs ) (5.36)

The synchronous particle (see Fig.5.31) is the particle that ar­

rives at the RF cavity when the voltage is such that it exactly
compensates the average energy losses U0

ΔE = U0 = eV0 sin(ϕs ) (5.37)

There are two points in time in the RF potential where the

particle will get the correct energy from the RF wave: point
P1 and point P2 (see Fig.5.32), and, as we can guess, one is
stable and the other is unstable.
In Section 5.1.6 we began our discussion of the dynamics
of the particle arriving earlier or later than the synchronous
particle, making a simplifying assumption that the particle is
FIGURE 5.32 ultrarelativistic.
Motion in RF potential. In the case of arbitrary energy, we need to take into ac­
count the dependence of the particle’s time of flight around
the ring of energy, which includes dependence of the circum­
ference and particle velocity on energy.
The synchronous particle with a nominal energy E and ve­
locity v travels around the nominal circumference C in time
T so that T = C/v. Taking a logarithm and differentiating this
expression will yield

dT dC dv
= − (5.38)
T C v
The first component of the above equation, dC/C, is ex­
pressed via the momentum compaction factor αc , which con­
nects the momentum deviation of the particle dp with the
orbit length difference dC:

dC dp
= αc (5.39)
C p
 conventional acceleration 97

The momentum compaction factor αc depends on the design

of the focusing lattice of the ring and can be either positive
(which is most common) or negative (particles with higher E
travel over a shorter orbit — which seems counterintuitive,
but is possible, although rare).
The second component of Eq.5.38 can be expressed as
dv dβ 1 dp
= = 2 (5.40)
v β γ p
Taking these two components together will yield
( ) ( )
dT 1 dp 1 dE
= αc − 2 ∼ αc − 2 (5.41)
T γ p γ E
which shows us that the time of flight depends on the energy
deviation, on the relativistic factor γ and on the momentum
compaction factor.
We now can conclude that if αc − 1/γ 2 > 0, the point
P2 in Fig.5.32 is stable (in contrast to phase stability as de­
scribed for a linac earlier). Indeed, according to this assump­
tion, a particle with a higher energy has a longer flight time
and therefore arrives later at the RF cavity, undergoes lower
RF voltage (point P2 is on a negative slope), thus gaining less
energy, and so tends to return to a nominal energy. Similar
deliberations can show that a particle with a lower energy
will, in this case, gain more energy in a manifestation of the
“principle of phase stability,” which enables the capture of
particles in the RF potential.
In the opposite case, αc − 1/γ 2 < 0, similar logic can lead
us to a conclusion of stability of the point P1 (as for a linac).
We have now arrived at the need to introduce the notion
of transition energy. In rings with positive αc passing, during
acceleration, the energy corresponding to the gamma factor
1
γt = (5.42)
αc1/2
corresponds to the moment of stability flipping from point
P1 to point P2 on the RF slope. Preserving the quality of the
accelerated beam requires a quick switch of the phase of the
RF voltage at the moment of passing the transition energy.
Now, let’s derive the longitudinal beam dynamics equa­
tions for particle acceleration in a synchrotron. We start by
considering a particle moving in an electric field of a travel­
ling wave Ez = E0 cos(ωt − kz) with a phase velocity vf = ω/k.
Equations describing the motion in the longitudinal plane are
dpz dE
= eE0 cos(ωt − kz) and = eE0 ż cos(ωt − kz) (5.43)
dt dt
We again define the synchronous particle as
dEs
= eE0 vs cos φ s (5.44)
dt
98 unifying physics of accelerators, lasers and plasma

and use deviations from its energy and time to describe an

arbitrary particle

E = Es + ε and t = ts + τ (5.45)

Using these definitions, we obtain the first equation

dε
= eE0 [cos(ωτ + φ s ) − cos φ s ] (5.46)
ds
and using Eq.5.41 to define the momentum compaction fac­
tor at high energy (γ » 1), we obtain the second equation
FIGURE 5.33 ( )
RF bucket trajectories in a dT 1 dE dτ αc dε
linearized case are ellipses. ∼ α c − → ∼ (5.47)
T γ2 E dt Es dt

which together describe an RF bucket in the longitudinal

phase space with coordinates (τ, ε).
Rewriting these equations for the RF bucket for γ » 1 in
3 3 terms of derivatives of time yields
qV0 αc
ε' = [sin(ϕs + ωτ) − sin ϕs ] and τ' = ε (5.48)
L Es
Linearizing these equations for the motion in the RF
6HSDUDWUL[ bucket gives us
e dV αc
ε' = τ and τ' = ε (5.49)
T0 dτ Es
which corresponds to the phase space motion with elliptical
8QVWDEOH trajectories as in Fig.5.33, with angular frequency defined as
6WDEOH
αc eV̇
ωs2 = (5.50)
T0 E0
which is called synchrotron frequency.
6HSDUDWUL[ In a realistic case of a practical accelerator design, we
often cannot limit ourselves to a linear approximation and
would need to consider the full nonlinear equations Eqs.5.48.
We will explore these equations now, looking at them from a
different perspective — via an analogy with classical mechan­
8QVWDEOH
ics. The equivalent equation can be rewritten as
6WDEOH
d 2ε
= U τ2 (5.51)
dt 2
where U is an analog of a potential energy in an oscillator.
The shape of the potential and corresponding phase space
trajectories are shown in Fig.5.34 for two cases, below and
FIGURE 5.34 above the transition energy. The potential energy analogy and
RF voltage and phase space its shape help to make sense of the behavior of the trajecto­
and RF potential for cases be­ ries. They are ellipse-like in the vicinity of the stable point
low and above the transition where the potential is parabolic, but become distorted and
energy. eventually unstable as they come closer to the saddle point
 conventional acceleration 99

of the potential where U flattens and passes through a local

maximum, creating conditions for certain trajectories to take
an arbitrarily long time for passing through the saddle point
region. The trajectory that originates from the vicinity of the
saddle point and separates the stable area from an unstable
area is called the separatrix.

5.5.4 RF potential — nonlinearity and adiabaticity

The RF potential, as we can see in Eqs.5.51, is intrinsically
nonlinear. Particles that are located close to the separatrix
will have longer periods of oscillations and will therefore lag
behind in their rotation in phase space with respect to parti­
cles in the center. This is illustrated in Fig.5.35, which shows
the qualitative behavior of a particle bunch in an RF poten­
tial (neglecting radiation damping) with an increasing num­
ber of synchrotron periods. We can see that while the beam
distortions are negligible after one synchrotron period, after
ten and especially after fifty synchrotron periods the longi­
tudinal phase space distribution is completely distorted. The
nonlinearity of the RF bucket results in filamentation of the
phase space, causing an effective increase of the longitudinal
emittance.

FIGURE 5.35
Qualitative evolution of the longitudinal phase space (energy vs
phase, for vertical and horizontal axes, correspondingly) of the
beam for an increasing number of synchrotron periods. Left to
right: initial distribution, after one synchrotron period, after ten
and after fifty periods.

Another assumption we made in the previous sections is

that the acceleration is adiabatic, i.e., dγs /ds ∼ 0. If this as­
sumption is violated, numerical integration of the equations
of motion need to be used, which will then show that the RF
bucket distorts further and takes the shape of a “golf club” as
FIGURE 5.36
illustrated qualitatively in Fig.5.36.
RF bucket in the case of fast
acceleration.
5.5.5 Synchrotron tune and betatron tune
Synchrotron oscillation in circular accelerators, as we saw in
the previous sections, is a multi-turn phenomenon (the syn­
chrotron frequency may correspond to many tens or hun­
100 unifying physics of accelerators, lasers and plasma

dreds of turns), i.e., such oscillations are very slow. The syn­
chrotron tune is connected to the synchrotron frequency as
follows
2πTrev
QS = (5.52)
ωS
where Trev is the period of revolution around the orbit. Typi­
cally, QS « 1.
In contrast, transverse betatron oscillations, in circular ac­
celerators with strong focusing, are necessarily fast (there are
usually many tens or hundreds of oscillations per revolution
period). The betatron tune is defined as the number of oscil­
lations around the ring or the ratio of the betatron frequency
to the revolution frequency

μ 1 ds'
Q= =
2π 2π β(s ' )

where the integral is taken along the accelerator circumfer­

ence. Typically, Q » 1.

[

W

7UHY 7UHY 7UHY 7UHY 4V 4 4V

FIGURE 5.37
Betatron oscillations modulated by synchrotron motion (left) and
a corresponding spectrum (right) with betatron tune and syn­
chrotron sidebands.

The betatron Q is momentum dependent and this can link

the two motions together. When synchrotron motion is thus
coupled to betatron motion, this can manifest itself in the sig­
nals of pick-up electrodes, which measure the beam transverse
oscillations as qualitatively shown in Fig.5.37. As the main
fast (betatron) signal is now modulated with a slow (syn­
chrotron) component

x ∝ sin (2πQ t /Trev ) (1 + Δ sin (2πQS t /Trev ))

the spectrum of transverse motion will now include syn­

chrotron sidebands at Q − QS and Q + QS in addition to the
main betatron frequency as is illustrated in Fig.5.37.
 conventional acceleration 101

5.5.6 Accelerator technologies and applications

Table 5.1 shows typical accelerating gradients for RF cavities
of various frequencies for linear accelerators.

TABLE 5.1
Operating frequencies and typical parameters for RF cavities
Warm cavities Gradient Repetition rate
S-band (3GHz) 15-25 MV/m 50-300 Hz
C-band (5-6 GHz) 30-40 MV/m <100 Hz
X-band (12 GHz) 100 MV/m <100 Hz
Superconducting cavities Gradient Repetition rate
L-band (1.3 GHz) < 35 MV/m up to CW

As we can see, the achievable gradient generally increases

with frequency, consistent with frequency dependences of
Eq.5.25 and Eq.5.26 (these relationships, however, haven’t yet
been demonstrated for frequencies higher than 12 GHz).
The practically achievable gradient is one of the main fac­
tors that define the size of accelerator-based facilities of vari­
ous kinds. Let’s consider a couple of examples.
In a high energy physics application, a linear collider
(Fig.5.38) aiming at 500 GeV energy in the center of mass
(CM), and built with L-band superconducting cavities, would
be around 30 km long. A collider aiming at 3 TeV CM, built
with an X-band normal conductive cavities, would be almost
50 km long.

FIGURE 5.38
A generic linear collider.

While the linac length constitutes a major fraction of the

length of a linear collider, in a free electron laser, the linac
length is a noticeable fraction of the overall length (between
around a quarter to a half). As an example, an FEL (see
Fig.5.39) with an electron beam energy of around 10 GeV
built with S-band or C-band technology would be about a
kilometer long.
102 unifying physics of accelerators, lasers and plasma

FIGURE 5.39
A generic free electron laser.

These examples demonstrate the motivations for develop­

ing new accelerator technologies that would make these and
other accelerator-based facilities and instruments more com­
pact. This brings us to the next chapter, where we will delve
into methods of plasma acceleration.
 conventional acceleration 103

EXERCISES
5.1 Chapter materials review.

Derive the Eq.5.35 for particle motion in the travelling wave.

5.2 Chapter materials review.

Discuss design approaches to the beam optics that would re­

sult in achieving a negative value of the momentum com­
paction factor in a synchrotron.

5.3 Mini-project.

Define very approximate parameters (sizes, magnetic fields,

parameters of RF system) of a 200 MeV rapid-cycling proton

synchrotron capable of operating at a 10 Hz repetition rate.

Assume injection at 1 MeV.

5.4 Analyze inventions or discoveries using TRIZ and AS-TRIZ.

Analyze and describe scientific or technical inventions de­

scribed in this chapter (e.g., tandem, RFQ, bunch and pulse

compressors) in terms of the TRIZ and AS-TRIZ approaches,

identifying a contradiction and an inventive principle that

were used (could have been used) for these inventions.

5.5 Developing AS-TRIZ parameters and inventive principles.

Based on what you already know about accelerator science,

discuss and suggest the possible additional parameters for

the AS-TRIZ contradiction matrix, as well as the possible ad­

ditional AS-TRIZ inventive principles.

 6
Plasma Acceleration

6.1 Motivations 105 Plasma acceleration is an emerging and promising field,

6.2 Early steps of whose rapid progress is enabled by developments in laser
plasma acceleration technology — particularly by the method of chirped pulse
107 amplification. Plasma accelerators of electrons — the pri­
6.3 Laser intensity and mary focus of this chapter — are the backbones of future
ionization 108 compact light sources. Proton and ion plasma acceleration,
6.4 The concept of laser briefly discussed here and in closer detail in Chapter 9, are a
acceleration 114 potential way to improve future medical accelerators.
6.5 Betatron radiation The aim of this chapter, after a brief introduction and dis­
sources 118 cussion of the motivations for pursuit of plasma acceleration,
6.6 Glimpse into the is to develop the framework that will help us to estimate the
future 120 parameters of laser plasma-based light sources, so as to be
6.7 Plasma acceleration prepared for Chapter 7.
aiming at TeV 122
6.8 Laser-plasma and
protons 124
6.1 Motivations
The “Livingston plot,” which depicts the energy of acceler­
ated beams versus time (Fig. 1.6), illustrates the great his­
tory of accelerators and related inventions. It also shows the
Recalling the use of Acceler- signs of saturation, highlighting the need for the next break­
ating Science TRIZ method: through in accelerator technology.
1) Define the problem in Traditionally, accelerating structures have been made
terms of generic contradic­ from metal (normal conductive or super-conductive) and
tion parameters (Table 1.4). are typically limited in their accelerating gradient to Ez <
To be improved: rate of E 100 MeV/m. This limitation is imposed by the properties of
change; what gets worse: in- the materials — since damage to the accelerating structure’s
tegrity. walls (deterioration of their integrity) limits the gradient.
2) Use the contradiction ma- The “accelerating structures” produced on the fly in
trix (Table 1.5) to obtain the plasma by a laser pulse are, however, made from a material
relevant inventive principle that is already “damaged” (plasma), and therefore do not ex­
— replace material that can hibit the same limitations due to the material’s properties.
be damaged with other me- Plasma acceleration was first proposed by T. Tajima and
dia, which either cannot be J. Dawson in 1979, which was, in fact, too early for laser and
damaged (light) or is already beam technologies to be ready to realize the proposed ap­
“damaged” (e.g., plasma). proach. Consequent parallel developments of laser and beam
3) Translate the generic in- technologies — specifically those aimed at creating short,
ventive principle into a spe­ powerful pulses — created the new reality making the laser
cific solution — plasma ac­ plasma acceleration the area with the highest degree of syn­
celeration. ergy between the physics of plasma, lasers and accelerators.
We should also recall (see margin notes) our discussion

DOI: 10.1201/b18696-6 105

106 unifying physics of accelerators, lasers and plasma

from Chapter 1 regarding the use of the AS-TRIZ approach

to post-facto analyze the invention of plasma acceleration.

6.1.1 Maximum field in plasma

By using plasma as an accelerating medium, we can remove
the limitation of the accelerating gradient relating to the ma­
terial’s damage threshold. The maximum field in plasma will
still be limited, but by other factors.
Let us look again at plasma oscillation, as illustrated in
Fig. 4.21, and briefly recall that this diagram allowed us to
estimate the plasma frequency ωp . Assuming that a fraction
of the charges is shifted by distance x, we selected an inte­
gral contour that enclosed the displaced charges, and we then
equated, according to Maxwell’s equations, the surface inte­
gral of electric field — E · dS to the volume integral of charge
density — ρdV /ε0 , to obtain the electric field created by the
shifted charges E = nex/ε0 , which creates the restoring force.
The equation of motion F = md 2 x/dt 2 = −eE = −ne 2 x/ε0 then
gave us the oscillation frequency ωp2 = ne 2 /(ε0 m), which, with
Plasma frequency in Hz is use of 4πε0 re = e 2 /(me c 2 ), we rewrote as: ωp2 = 4πnc 2 re — the
f p ∼ 9000 n1/2 where plasma angular plasma frequency.
density n is in cm−3 . Very similar calculations allow us to estimate the maxi­
mum accelerating field in plasma. Imagine that the plasma
oscillation in Fig. 4.21 is excited by a charged object moving
with velocity c. In the case where a total charge separation is
achieved in plasma, the maximal field is estimated assuming
c
x ∼ λp ∼ (6.1)
ωp

which results in the following for the maximum field

nec mcωp
Emax ∼ = (6.2)
ε0 ωp e

or equivalently
ωp
eEmax � mc 2 (6.3)
c
Maximum accelerating field We can use the practical formula f p ∼ 9000 n1/2 where n is
in plasma eEmax ≈ defined in cm−3 to obtain a formula for the maximum possi­
−3 1/2 ble accelerating field in plasma:
1 GeV/cm · n/10 cm
18 .
eV 1/2
eEmax ≈ 1 ·n cm−3 (6.4)
cm
This means that 1 GeV/cm accelerating gradient can be
achieved for plasma of 1018 cm−3 density.
Theoretical predictions, made back in 1979, of the prin­
cipal feasibility of such large accelerating gradients were an
essential driving force towards the development of plasma
acceleration technology.
 plasma acceleration 107

6.2 Early steps of plasma acceleration

We see that GeV/cm requires plasma with n=1018 cm−3 . The
plasma wavelength

c 1017 cm−3
λp = or λp ≈ 0.1mm (6.5)
fp n

corresponding to a plasma density of 1018 cm−3 is around

λp ≈ 30 μm (or around 100 fs). Thus, short sub-100-fs pulses
are needed to excite plasma towards GeV/cm accelerating
gradients.
In the absence of such short laser pulses, in the late 1970s
and early 1980s, other methods of plasma excitation were
suggested (by J.M. Dawson, 1979) such as the plasma beat
wave accelerator (PBWA) and the self-modulated laser wake-
field accelerator (SMLWFA); see Fig. 6.1.

S

D E F
FIGURE 6.1
For illustration of plasma beat wave and self-modulated laser
wakefield acceleration.

In the PBWA, two laser pulses with envelopes as in S

Fig. 6.1.a and frequencies differing by ωp overlap to create
a beating at the plasma’s frequency as shown in Fig. 6.1.b.
This combined laser pulse is sent into plasma where it cre­
ates plasma excitation as shown in Fig. 6.1.c.
In contrast to the previous method, in the SMLWFA,
only a single laser pulse is sent into the plasma (Fig. 6.1.a),
where an instability (which we will not discuss here in de­ FIGURE 6.2
tail) results in a self-modulation of the long laser pulse at λp Plasma wakefield accelera­
(Fig. 6.1.b), which again creates plasma excitation at wave­ tion — PWFA.
length λp (Fig. 6.1.c).
As a result of beam and laser technologies development, S
short sub-ps pulses of laser or beams became available and
thus prompted rapid progress of plasma acceleration.
The plasma wakefield acceleration (PWFA) method uses
a short, high energy particle bunch to excite the plasma
(Fig. 6.2). Similarly, a short laser pulse of high intensity can
be used in a laser wakefield acceleration (LWFA) method
(Fig. 6.3). In both of these cases, a high amplitude plasma FIGURE 6.3
wave is created, which can then be used for acceleration. We Laser wakefield acceleration
will discuss these methods in detail in the following sections. — LWFA.
108 unifying physics of accelerators, lasers and plasma

6.3 Laser intensity and ionization

Laser acceleration requires laser pulses of short duration and
high intensity. In order to prepare for a quantitative discus­
sion of the intensities required for plasma acceleration, let us
introduce the basic concepts related to this subject.

6.3.1 Laser pulse intensity

Laser intensity (in a vacuum) is defined (in SI and Gaussian
units respectively) as
1
I= ε E2 c (SI) (6.6)
2 0 max
Recall
ε0 ≈ 8.8 · 10−12 A2 s 4 /(kg m3 ) 1 2
I=
E c (Gaussian) (6.7)
8π max
The intensity I is usually measured in Watts per cm2 . The
It is useful to remember that corresponding relation between electric field and intensity
in an EM wave the field am- in practical units is:
plitudes of 300 V/cm and ≈1 [ ]
V I 1/2
Gauss are equivalent. Emax � 2.75 × 109 (6.8)
cm 1016 W /cm2
Similarly, for the magnetic field:

I 1/ 2
Bmax [Gauss] � 9.2 × 106 16
(6.9)
10 W /cm2
For example: a laser with 30 Joules energy in a 30-fs
(10 μm)-long pulse corresponds to (assume rectangular dis­
tribution in space and time) a peak power of 1015 watts and,
if focused to a spot with a diameter of 3 μm as illustrated in
Fig. 6.4, produces the intensity in the focus of 1022 W/cm2 .
FIGURE 6.4 According to the equations given above, the fields in the focus
Laser focused to a tight spot. of such a laser will approach 10,000 Mega Gauss.
6.3.2 Atomic intensity
In order to develop a quantitative understanding of laser in­
tensity values, it is best to compare the field of an intense
laser with atomic fields — particularly with the field in a hy­
drogen atom.
The Bohr radius is given by:

n2
aB = = 5.3 × 10−9 cm (6.10)
me 2
The corresponding field is then defined as:
e
Ea = (Gaussian units) (6.11)
aB2
e V
= ≈ 5.1 × 1011 (SI)
4πε0 aB2 m
 plasma acceleration 109

The corresponding atomic intensity is thus equal to

ε0 cEa2 W
Ia = � 3.51 × 1016 2 (6.12)
2 cm
A laser with intensity higher than the above will ionize gas
immediately. However, as we will show in the next sections,
ionization can occur well below this threshold due to multi-
photon effects or tunneling ionization.

6.3.3 Progress in laser peak intensity

Lasers able to produce peak intensity of atomic levels given
by Eq. 6.12 were not available until the mid-1980s. This is
illustrated in Fig. 6.5, which shows a qualitative overview of
the progress in laser peak intensity throughout history.
The invention of the chirped pulse techniques — CPA and
OPCPA — was a breakthrough in laser peak power, allowing
reaching and exceeding atomic intensities (indicated by the
line b in Fig. 6.5).

,QWHQVLW\:FP


 F
5HODWLYLVWLFUHJLPHD  

$WRPLFLQWHQVLW\ E

 &3$ D

)LHOGLRQL]DWLRQ
RI K\GURJHQ

 <HDU

     

FIGURE 6.5
Qualitative overview of the progress in laser peak intensity.

Fig. 6.5 also shows two other important intensity levels:

the first one corresponds to the field ionization of hydrogen
(line a in this figure), discussed in the next section, and the
second one corresponds to the relativistic optics case (line
c) when electrons become relativistic in the laser field (dis­
cussed in Section 6.3.6).
110 unifying physics of accelerators, lasers and plasma

One more vital intensity limit relevant to Fig. 6.5 —

but not shown as it would be significantly off-scale (around
2 · 1029 W/cm2 ) — is the Schwinger intensity limit, which cor­
responds to the case when the laser field can produce e+ e−
pairs from a vacuum (this will be discussed further in Sec­
tion 6.3.8).

6.3.4 Types of ionization

There are several types of ionization of interest, some of
which are shown in Fig. 6.6, depicting the potential well of
an electron in an atom.
In direct ionization (Fig. 6.6.a), the photon transmits
enough energy to an electron to overcome the potential bar­
rier in one interaction. In a multi-photon ionization (Fig. 6.6.b),
the electron obtains the energy needed to overcome the po­
tential barrier via the process of multi-photon absorption.
Tunneling ionization (Fig. 6.6.c) can occur when an electron
quantum tunnels through the potential barrier.

D E F
FIGURE 6.6
Types of ionization: (a) direct, (b) multi-photon, (c) tunneling.

The tunneling ionization, as presented in Fig. 6.6.c, and

with somewhat larger laser field intensity, will turn into an­
other ionization mechanism called barrier suppression ion­
ization.

6.3.5 Barrier suppression ionization

Barrier suppression ionization (BSI) occurs when the laser field
distorts the potential of an atom in such a way that the elec­
tron can freely escape the potential well.
The Coulomb potential of a hydrogen atom distorted by a
homogeneous field E can be written as (in Gaussian units):

e2
V (x) = − − eEx (6.13)
x
The distorted potential is shown in Fig. 6.7. The position of
the maximum of the potential on the right side of the plot is:

xmax = (e/E)1/2 (6.14)

 plasma acceleration 111

9[
H [

[PD[
 [
(LRQ

FIGURE 6.7
Barrier suppression ionization.

and the value of the potential at the maximum is

1/2
V (xmax ) = 2 e 3 E (6.15)

Equating the potential value at the maximum V (xmax ) to the

hydrogen atom ionization potential Eion
e2
Eion = ≈ 13.6eV (6.16)
2aB
gives us the critical field for the hydrogen atom
e E
εc = 2
= a (6.17)
16aB 16

As we can see, it is a small fraction of the atomic field Ea

given by Eq. 6.11. The laser intensity corresponding to the
BSI mechanism is then
Ia
Ic = ≈ 1.4 · 1014 W /cm2 (6.18)
256
which is more than two orders of magnitude lower than the
atomic intensity Ia given by Eq. 6.12. The intensity Ic is indi­
cated by line a in Fig. 6.5.

6.3.6 Normalized vector potential

The laser field can be written in terms of the vector potential
of the laser field A as
∂A
E=− , B = ∇×A (6.19)
c∂t
For a linearly polarized field:
A = A0 cos (kz − ωt) e⊥ (6.20)
112 unifying physics of accelerators, lasers and plasma

where e⊥ is transverse unit vector. We can see that the field

and vector potential amplitudes are connected via
A0 ω
E0 = (6.21)
c
Comparing momentum gained by an electron e− in one cycle
of laser field
eE
e E Δt ∼ (6.22)
ω
with its rest mass me c, we can see that it is better to define the
normalized vector potential as
eA
a= (6.23)
me c 2
with its amplitude given by
eE0
a0 = (6.24)
me ωc
The amplitude a0 will indicate if the electron motion in the
laser field is relativistic: a0 » 1, or nonrelativistic: a0 « 1.
The normalized vector potential amplitude in practical
units can be written as
⎛ [ ]⎞1
⎜⎜ I W /cm2 ⎟⎟ 2
⎜⎜ ⎟⎟
a0 ≈ ⎜⎜ ⎟ · λ [μm] (6.25)
⎝ 1.37 · 1018 ⎟⎠

where λ = 2πc/ω is the wavelength of the laser. For example,

for a red laser with λ = 0.65 μm, the value a0 = 1 reached
at intensity of I ≈ 3 · 1018 W /cm2 (as indicated by line c in
Fig. 6.5).

6.3.7 Laser contrast ratio

As we see in Fig. 6.5, different phenomena related to laser-
matter interaction and plasma acceleration occur at signifi­
cantly different intensities. This brings us to a dialogue re­
garding the temporal contrast ratio of a laser pulse.
The spatial contrast — the ratio of intensity at the laser
focus to the intensity outside of the focus — is a standard
concept intuitively known to everyone from everyday life.
For CPA-compressed pulses, which involve manipula­
tions and exchanges between energy and longitudinal phase
space coordinates, it is appropriate to introduce the notion of
the temporal contrast ratio — a function of time given by the
ratio of the peak laser intensity to the intensity in the front or
back of the pulse.
A qualitative spatial profile of a CPA-compressed laser
pulse is shown in Fig. 6.8 in terms of the contrast ratio, in
logarithmic scale. It is typical that a short sub-ps pulse is ac­
companied by many tens of ps low-intensity pulses, as well
 plasma acceleration 113



&RQWUDVWUDWLR
0DLQSXOVH


3UHSXOVHV 3RVWSXOVHV


   
WSV

FIGURE 6.8
Qualitative temporal profile of a CPA-compressed laser pulse.

as short pre-pulses or post-pulses, which are typically caused

by nonlinear properties of the elements of the CPA system
and non-ideal properties of the initial laser pulse.
The high contrast ratio is often the key parameter in
plasma acceleration, as even a relatively low intensity can ei­
ther ionize or destroy the target long before the arrival of the
main high-intensity short pulse.

6.3.8 Schwinger intensity limit

A laser of high enough intensity can generate e+ e − pairs from
a vacuum. According to the time-energy uncertainty princi­
ple
n
ΔE Δt ≥ (6.26)
2

a virtual e +
e − pair (thus ΔE ∼ me c2 ) can appear for a short
duration of time Δt ∼ n/(me c 2 ). If an electric field ES acting
on the pair during Δt increases the momentum of e + or e −
by about mc (i.e. Δt · e ES ∼ mc), the particles then become
real and thus materialize from the vacuum thanks to the laser
field. The corresponding laser field (ignoring factors of two in
the estimations) is hence given by

m2e c3
ES = ≈ 1.3 · 1018 V/m (6.27)
en
and is called a Schwinger limit field — the scale above which
the linear electrodynamics become invalid. The correspond­
ing laser intensity is

IS ≈ 2 · 1029 W/cm2 (6.28)

Reaching such laser intensity in practice would undoubtedly

create a new scientific and technological breakthrough.
114 unifying physics of accelerators, lasers and plasma

6.4 The concept of laser acceleration

We are now ready to discuss the concept of laser plasma ac­
celeration — see Fig. 6.9 — wherein a powerful laser pulse
enters gas (which can either be pre-ionized or not). We first
note that the contrast ratio of the laser is not infinite, and so
the ionization front starts in the gas at the front tail of the
laser pulse, much in advance of the arrival of the main laser
pulse. We then note that the main laser pulse needs to be of
a length similar to or shorter than the plasma wavelength in
order to excite the plasma efficiently.

FIGURE 6.9
Laser acceleration — conceptually. Linear regime.

The electrons of the plasma can be trapped in the wave

and then accelerated. Maximum acceleration can occur when
the laser pulse causes total separation of the electrons and
ion charges of the plasma; this regime is nonlinear, and the
cavity that is formed in the plasma and can trap and acceler­
ate electrons is called a bubble. Usually, electrons are trapped
and accelerated in the first bubble. The mechanism of bub­
ble formation in a strongly nonlinear approximation will be
discussed in detail in the following section.

6.4.1 Ponderomotive force

The formation of a bubble is the result of ponderomotive force
that a laser pulse confined in space exerts on the plasma elec­
trons.
We start from the assumption that the laser field E is ho­
mogeneous:
E = E0 cos (ωt) (6.29)
 plasma acceleration 115

The corresponding transverse motion of electrons is:

F eE eE
ÿ = = ⇒ y = − 02 cos (ωt) (6.30)
m m mω
We then assume that the field E has a gradient in transverse
direction y and thus can be expressed as:
∂E0
E = E0 (y) cos (ωt) ≈ E0 cos (ωt) + y cos (ωt) (6.31)
∂y
We then find the force acting on an electron e− averaged over
time is:
eE ∂E
(F)t = − 02 cos (ωt) · 0 cos (ωt) (6.32)
mω ∂y t
Replacing (cos2 ) with 1/2, we rewrite it as
e2 ∂E e 2 ∂E02
(F)t = − 2
E0 0 = − (6.33)
2mω ∂y 4mω 2 ∂y
We see from this equation that, since the intensity I ∝ E 2 , the
ponderomotive force is proportional to the gradient of the
laser intensity
∂I
(F)t ∝ −e2 (6.34)
∂y
and that the direction of the force is such that the pondero­
motive force pushes electrons out from the high intensity re­
gion.

FIGURE 6.10
For illustration of the mechanism of the ponderomotive force.
Laser with transverse intensity gradient (left), qualitative picture
of the electron motion (middle) and the intensity profile with
direction of the force shown (right).

We also note that the ponderomotive force is independent

on the sign of the charge — positrons will be pushed out of
the laser pulse just as well.
An alternative explanation of the ponderomotive force
can also be suggested (see Fig. 6.10): the electrons oscillating
in the time-varying laser field are pushed away more force­
fully when they are in a higher intensity region — which
on average repels the charged particles from the laser pulse’s
high intensity area.
116 unifying physics of accelerators, lasers and plasma

6.4.2 Laser plasma acceleration in nonlinear regime

Ponderomotive force plays a key role in the formation of
the accelerating bubble in the nonlinear (also called blow-out)
regime of laser plasma acceleration.
The ponderomotive force of a short (typically ∼50 fs) and
intense (typically ∼ 1018 W/cm2 ) laser pulse expels plasma
electrons while heavier ions stay at rest. The expelled elec­
trons are immediately attracted back to the ions, forming the
first bubble, as shown in Fig. 6.11 and Fig. 6.12, thus creating
a plasma wave that trails behind the laser pulse. The gradi­
ent of the density of electrons creates spatial oscillation of the
FIGURE 6.11
electric field within plasma (reaching ∼100 GV/m) which can
Bubble formation.
accelerate particles.

FIGURE 6.12
Laser plasma acceleration in nonlinear regime — conceptually.

Having formed the first bubble, the electrons continue

their oscillations around the ions, but their motions quickly
become incoherent and so the second and subsequent bub­
bles gradually become smaller. In a sense, only the first bub­
ble (and sometimes the second) is useful for acceleration.

6.4.3 Wave breaking

The high accelerating gradient in the plasma is useful only
if a particle beam can be injected into the bubble. Luckily,
self-injection of background plasma electrons into the plasma
bubble can occur through the wave breaking phenomenon.
Wave breaking transpires when, within the nonlinear
regime, certain particles outrun the wave, as is represented
in Fig. 6.13 in the analogy with ocean waves.
Other methods of getting particles into the bubble in­
clude injection of an external electron beam (challenging if
FIGURE 6.13
the bunches are short) and various other methods that create
Wave breaking concept —
an electron bunch inside of the bubble in the right place and
the wave nonlinearity gradu­
at the right time. These typically involve using multiple laser
ally rises from top to bottom.
pulses and mixes of gases with different ionization potentials.
 plasma acceleration 117

6.4.4 Importance of laser guidance

As a laser pulse travels through the gas or plasma, several
competing effects are taking place.
The most notable is diffraction; indeed, a laser beam fo­
cused to a size of several tens of μm will diffract very fast.
Other effects include dephasing — the gradual separation of
the accelerating beam (which quickly become relativistic)
from the laser (which propagates in gas or plasma slower
than the speed of light) — and also depletion — the gradual
decrease of the laser intensity.
Additional peculiar effects include longitudinal compres­
sion of the laser pulse by plasma waves; self-focusing, in par­
ticular due to the relativistic effect (the electrons of plasma
at the axis become relativistic and have higher masses, affect­
ing the plasma refraction coefficient); and ionization-caused
diffraction (gas on the axis where intensity is higher will be
ionized first, affecting diffraction).

FIGURE 6.14
Capillary channel technique of laser plasma acceleration.

A possible solution that could solve some of the issues

listed above involves creating a channel within the plasma
where a special density profile n(r) will be formed to assist
guiding the laser pulse for a significant distance.
This solution was realized in practice in the form of a cap­
illary discharge channel developed at Oxford University by
S. Hooker (ca. 2006). A schematic of the capillary channel is
shown in Fig. 6.14. In this example, a sub-mm hole is created
in a sapphire block, hydrogen gas is delivered to the capil­
118 unifying physics of accelerators, lasers and plasma

lary via side holes, and discharge electrodes act to pre-form

plasma with a density profile featuring minimum on the axis
(as gas near cold walls of the capillary has higher density ac­
cording to P = n k T = conts) — such a density profile is suit­
able for refraction-assisted laser guidance.
The capillary channel technique was essential in exceed­
ing the GeV barrier in laser plasma acceleration for the
first time ever (W. Leemans et al., 2006), creating a mono-
energetic 1 GeV beam after accelerating in just 3 cm of
plasma.

6.5 Betatron radiation sources

Beams accelerated in a laser-formed plasma bubble can os­
cillate, which generates synchrotron (betatron) radiation.
Strong radial electric fields within plasma bubbles are re­
sponsible for the electrons experiencing transverse oscilla­
tions. This can generate bright betatron radiation in an ex­
tensive range of photon energy (around 1 - 100 keV). In this
section we will estimate the expected parameters of radiation
produced by a laser plasma source.

6.5.1 Transverse fields in the bubble

Transverse oscillations of the accelerating electron beam in
the plasma bubble are caused by a transverse focusing force
that is produced by ions. We can assume that the ions are
heavy and are stationary within the bubble. The ions produce
a focusing force that can be determined using

E · dS = 4π ρdV (Gaussian units) (6.35)

and by assuming cylindrical symmetry as in Fig. 6.15. The

focusing force is therefore
eE = 2πne 2 r (6.36)
An electron with relativistic factor γ will oscillate in this field
FIGURE 6.15
as
Cylindrical symmetry in the d 2 r 2πne 2 r ωp2
plasma bubble. = = r (6.37)
ds2 γmc2 2γc2
The period of oscillation is thus given by
.
λ = 2γ λp (6.38)

(and is changing during electron acceleration).

 plasma acceleration 119

6.5.2 Estimations of betatron radiation parameters

In Chapter 3 we estimated the characteristics of synchrotron
radiation, in particular the energy loss per unit length:

dW 2 e 2 γ 4
= (6.39)
ds 3 R2
the characteristic frequency of photons:

3 c γ3
ωc = (6.40)
2 R
and the number of photons emitted per unit length:

dN α γ
= (6.41)
ds R
Fig. 6.16 shows a qualitative representation of the evo­
lution of the plasma bubble and oscillation of the accelerat­
ing beam. We assume that the beams are self-injected into
the bubble due to the wave breaking phenomena when two
beamlets overshoot and enter the bubble symmetrically from
the top and bottom (as shown in Fig. 6.16.a). Their initial
transverse velocity forces the beamlets to oscillate in the fo­
cusing field of the ions. The beamlets then continue to simul­
taneously accelerate while exhibiting transverse oscillations.
Let’s assume that the amplitude of beam oscillations in
the plasma bubble is equal to rb and the period of oscillations
is λ. The radius of the curvature of the beam trajectory in this
case is equal to:
λ2
R= (6.42)
4π2 rb
Substituting the period of oscillation given by Eq. 6.38, we
obtain for the radius of the curvature
γ λ2p
R= (6.43)
2π2 rb

Substituting this into Eq. 6.40, we get an estimation of the

radiation wavelength for the laser plasma betatron source:
2 FIGURE 6.16
1 λp 1
λc = (6.44) Laser plasma betatron
3π rβ γ 2
source — conceptually.
Using Eq. 6.41 together with Eq. 6.43 we can also estimate Wave breaking and self-
the number of photons Nγ emitted per λ: injection — (a). Oscillation
of accelerating electron
. rb
Nγ ≈ 2γ 2π α 2
(6.45) beams in the plasma bubble
λp — (b)-(d), sequential time
moments. Betatron radiation
Let us consider a practical example of a beam accelerated produced by oscillating
in the bubble characterized by λp = 0.03 mm, to up to 1 GeV beams — (e).
120 unifying physics of accelerators, lasers and plasma

(γ = 2 · 103 ). The synchrotron radiation is most notable at

higher energies, thus we will ignore lower energy radiation
occurring during acceleration. The oscillation amplitude rb
can only be guessed very approximately and usually needs
to be obtained via careful simulations. In a very rough ap­
proximation, rb is around 1%–10% of the bubble size (which
is ∼ λp ). We therefore assume that rb =0.001mm. Substituting
this into the above equations, we obtain λc = 0.025 nm (or
∼ 50 keV) and Nγ per λ is ∼ 0.3 per each accelerated elec­
tron. Considering that the accelerating bunch can carry tens
of pC to nC charge, we can conclude that such a light source
can generate many hard X-ray photons.

6.6 Glimpse into the future

Plasma acceleration is a technique that opens new opportu­
nities for creating scientific, technological and medical in­
struments. In this section we will give a brief review of the
progress made in laser acceleration to date, and, after com­
paring the evolution of plasma accelerators to the evolution
of computers in the latter half of the 20th century, we will
take a glimpse into the future.

6.6.1 Laser plasma acceleration — rapid progress

The last decade has yielded rapid progress in the field of laser
plasma acceleration. The pace of research and development
in this area received a significant boost in 2004 when the first
quasi-monoenergetic beam was generated.1
In 2006, Oxford and Berkeley teams2 broke the GeV bar­
rier in laser plasma acceleration and demonstrated quasi­
monoenergetic properties of this accelerated beam. Demon­
stration of these promising properties further increased the
research momentum, and applications of the accelerated
beam started to be developed, based on generation of radi­
ation in conventional as well as plasma wigglers (betatron ra­
diation).
The first use of laser plasma-produced betatron radia­
tion for biological imaging was reported3 in 2011. Multi-GeV
laser plasma acceleration4 was mastered to produce around
4 GeV beams and further progress is expected.
1 S. Mangles, Nature, 2004.

2 W. Leemans et al., Nature Physics 2006.

3 S. Kneip et al., Applied Physics Letters, 2011.

4 W. Leemans et al., Phys. Rev. Letters, 2014.

 plasma acceleration 121

6.6.2 Compact radiation sources

Laser-driven plasma accelerators can already generate elec­
tron beams with several GeV of energy, ∼10 fs bunch dura­
tion and ∼10-100 pC of charge per bunch.
These parameters make laser plasma technology poten­
tially suitable for creating compact radiation sources (see
Fig. 6.17). As the beams in these sources are created by a laser
in the first place, such sources would also have the advantage
of automatic synchronization of accelerated electron beams
and generated X-rays with the initial laser pulses. Such syn­
chronization is a powerful asset for time-resolved studies.
The far-from-desirable parameters of laser plasma light
sources so far include the repetition rate, wall-plug efficiency
and beam qualities (energy spread and emittance).
The repetition rate and efficiency are presently limited by
laser technology and are the subjects of active research by
many groups. Various promising ideas that are beyond the
scope of this book have been suggested and are being devel­
oped.
The beam quality, which is a focus of significant atten­
tion, is gradually improving, and eventually the laser plasma
accelerated beam might be suitable for the generation of co­
herent radiation in a free electron laser application.

FIGURE 6.17
Laser plasma betatron radiation light source — conceptually.

Modern synchrotron-based light sources are large ma­

chines with perimeters of several hundred meters. The linac­
based free electron lasers can be around a kilometer or more
in length. Both of these types of light sources operate with
electron beams of a few to about 10 GeV.
Despite the fact that similar electron energies can be
reached in a much more compact laser plasma accelerator,
it is unlikely that laser plasma-based light sources would en­
122 unifying physics of accelerators, lasers and plasma

tirely replace conventional light sources in the foreseeable fu­

ture. It is more probable that the types of light sources will
evolve in a similar manner to that of computers.

6.6.3 Evolution of computers and light sources

Early computers were large, bulky and slow. Development of
compact personal computers started in the early 1980s, but
was not accepted immediately: “IBM bringing out a personal
computer would be like teaching an elephant to tap dance,”
as newspapers mocked around 1981. Still, as we know today,
a plethora of large computers and super-computers co-exist
with a variety of personal and compact computers, from lap-
tops to mobile phones and smart watches (Fig. 6.18).
The expected future evolution of light sources may follow
a similar pattern. The synchrotron-based light sources and
FELs — (a) and (b) in Fig. 6.19 — will eventually be joined by
compact plasma-based light sources — (c) in Fig. 6.19 — as a
result of intense research, commercialization, and work with
users, the industry and economists — efforts of all of which
will result in a change of the paradigm.
All types of light sources will then continue to co-exist
and national-scale facilities will be complemented by a vari­
ety of compact plasma acceleration based light sources.
FIGURE 6.18
Computers’ evolution.
6.7 Plasma acceleration aiming at TeV
While application of plasma acceleration to compact light
sources is practically within reach, the application of plasma
acceleration technology to high energy physics discovery ma­
chines is significantly further away. In this section, we will
briefly review some of the primary challenges to plasma ac­
celeration on the way to TeV energy.

6.7.1 Multi-stage laser plasma acceleration

In laser plasma acceleration, the laser pulse propagating
through a medium (plasma) has v < c and the accelerat­
ing electrons that quickly become relativistic will soon de-
FIGURE 6.19 phase from the plasma wave. Consequently, if we are aiming
Light sources’ evolution. at multi-tens of GeV or TeV acceleration of electrons, many
stages of acceleration will be necessary.
The length of a single stage can be estimated by taking
into account that the group velocity of a laser pulse is given
by )
vg = 1 − ωp2 /ω2 (6.46)
and that the dephasing occurs when an electron outruns the
wave by a half of a period. For a relativistic electron the de­
 plasma acceleration 123

phasing time td is thus given by

(c−vg ) td = λp /2 (6.47)
We then substitute the expression for the group velocity and
get the estimate for the dephasing length:
Ld ≈ λp ω2 /ωp2 (6.48)
For example, for a laser with a wavelength of 1 μm and
λp = 30 μm, the dephasing length is Ld ≈ 30 mm. With an
accelerating gradient (at the corresponding plasma density)
of 1 GeV/cm, a single stage could yield around 3 GeV. Accel­
eration to a TeV would thus require several hundred stages.
Preservation of beam qualities during multi-stage accelera­
tion is the research area that promises significant advances in
the near future.

6.7.2 Beam-driven plasma acceleration

Plasma can be excited not with a laser pulse, but with a short
intense bunch of charged particles (e.g., electrons). In this
case, the bubble will be formed due to the bunch’s field, and
will have a very similar shape and properties to the laser ex­
ample.
This beam-driven acceleration approach (Fig. 6.20) has an
advantage in that the driver beam has v = c, and thus dephas­
ing of the witness beam from the driver is no longer an issue.
Another advantage of this method is that the driver beam can
carry much more energy than a laser pulse.

FIGURE 6.20
Beam-driven plasma acceleration — conceptually.

These advantages manifested themselves via a much

higher beam energy achieved in beam-driven acceleration —
the maximum final energy obtained so far is around 80 GeV
with an initial beam energy of 42 GeV (which acted in this
SLAC linac energy doubling experiment5 both as a witness as
well as a driver).
5 I. Blumenfeld et al., Nature, 2007.
124 unifying physics of accelerators, lasers and plasma

6.8 Laser-plasma and protons

So far, we have only discussed acceleration of electrons. How­
ever, plasma acceleration of protons and ions is possible and
is being actively developed. We will touch here on this subject
only briefly, and leave the details for Chapter 9.
The major reason why laser plasma acceleration of pro­
tons should be developed is its potential value to proton
therapy. Conventional proton therapy systems — which re­
quire around 250 MeV protons and include beam sources (cy­
clotron or synchrotron), beamlines and especially the beam
delivery gantries — are large and expensive. A desire to cre­
ate compact laser plasma acceleration proton therapy sys­
tems is one of the main motivations for development of
plasma acceleration of protons.
Proton and ion laser plasma acceleration is a rapidly de­
veloping area, but it is not quite yet ready for the textbooks.
Different models, with different assumptions and simplifica­
tions, are created to explain and predict beam properties in
different regimes of acceleration. One of the models, illus­
trated in Fig. 6.21, is sheath laser acceleration of protons. In
this case, the laser heats and ionizes a foil, creating a sheath
of hot electrons moving away from the foil, which in turn
pull and accelerate the ions from the plasma. This particular
FIGURE 6.21 mechanism, unfortunately, creates a very wide energy spec­
Sheath laser plasma acceler­ trum for the accelerated beam. Moreover, its scaling with the
ation of protons or ions. laser power is not favorable. We will discuss other more suit­
able mechanisms in Chapter 9.
 plasma acceleration 125

EXERCISES
6.1 Chapter materials review.

What laser intensity (in W/cm2 ) would correspond to a nor­

malized vector potential of a0 = 10, and what are the max­
imum values of the electric and magnetic fields in the laser

wave, for a ruby laser or for a CO2 laser?

6.2 Chapter materials review.

Assume that we get 1 GeV electron bunch from a laser

plasma accelerator, and would like to create an undulator

from plasma using the focusing force of the plasma’s ions.

Suggest the plasma parameters and the amplitude of the

beam oscillation that would correspond to the radiation at

the boundary of the undulator regime, i.e., with the undula­

tor parameter equaling K = 1.

6.3 Mini-project.

Assume that the LHC proton beam of E = 7 TeV is going to

be used as a driver for the plasma acceleration of electrons.

Select the parameters for the plasma and estimate to what

bunch length you would need to compress the LHC beam so

that it could be used for plasma acceleration. Define, roughly,

the parameters of the corresponding bunch compressor.

6.4 Analyze inventions or discoveries using TRIZ and AS-TRIZ.

Analyze and describe scientific or technical inventions de­

scribed in this chapter in terms of the TRIZ and AS-TRIZ ap­
proaches, identifying a contradiction and an inventive prin­
ciple that were used (could have been used) for these inven­
tions.

6.5 Developing AS-TRIZ parameters and inventive principles.

Based on what you already know about accelerator science,

discuss and suggest the possible additional parameters for

the AS-TRIZ contradiction matrix, as well as the possible ad­

ditional AS-TRIZ inventive principles.

 7
Light Sources

7.1 SR properties and In this chapter, we will focus on light sources — first syn­
history 127 chrotron radiation light sources and then Compton light
7.2 Evolution and sources. For the former, we will base our observations on re­
parameters of SR sults derived in Chapter 3, and will then introduce the nec­
sources 129 essary formalism for the latter.
7.3 SR source layouts
and experiments
131
7.4 Compton and 7.1 SR properties and history
Thomson scattering We know from courses on electrodynamics that electromag­
of photons 135 netic radiation is emitted by charged particles when they are
7.5 Compton light accelerated. In the particular case when the relativistic par­
sources 139 ticles move on a curved trajectory — i.e., when they are ac­
celerated radially (when the velocity is perpendicular to the
acceleration vector) — the emitted electromagnetic radiation
is called synchrotron radiation.
This SR was at first considered a nuisance, as it causes en­
ergy losses in the accelerated particles. Its unique properties,
however, were eventually shown to have the potential to pave
the path to a new and important type of scientific instrument:
the SR sources.

7.1.1 Electromagnetic spectrum

The usefulness of SR and Compton sources is due to the fact
that they can cover a large range of the electromagnetic spec­
trum (Fig. 7.1), from infrared (IR) and ultraviolet (UV) to
vacuum ultraviolet (VUV: the wavelength range strongly ab­
sorbed in air — thus the name), and on to hard X-rays and
γ-rays.

FIGURE 7.1
Electromagnetic spectrum covered by SR and Compton sources.

DOI: 10.1201/b18696-7 127

128 unifying physics of accelerators, lasers and plasma

The EM spectrum shown in Fig. 7.1 also indicates a partic­

ular range of wavelengths — the so-called water window is de­
fined as the range between the K-absorption edges of oxygen
(0.53 keV) and carbon (0.28 keV) — see Fig. 7.2. In this range
of soft X-ray energies, water is relatively transparent, which
simplifies SR experiments when biological samples need to
be used as water solutions.

FIGURE 7.2
Photon attenuation in water in comparison with a typical protein.

The success and widespread use of SR and Compton sources

is thanks to their flexible spectral parameters. This allows one
to select the photon energy required for an experiment by
adjusting the energy of the beam or laser, or the field strength
of wigglers or undulators (useful, for example, when trying to
fit within the range of the water window).

7.1.2 Brief history of synchrotron radiation

In 1944, D. Ivanenko and I. Pomeranchuk predicted that the
maximum energy of electrons in a betatron is limited due to
energy losses caused by radiation of relativistic electrons.
This radiation was first observed around 1947, by acci­
dent, in a General Electric 70 MeV synchrotron (this gave
the name synchrotron to the observed radiation), in the vis­
ible spectrum. It is interesting to note that earlier deliber­
ate attempts to find this radiation in a betatron had failed,
as researchers looked for the radiation in a microwave range
where the betatron walls were opaque.
The first physics experiments with SR were conducted
in 1956 at Cornell, on a 320 MeV synchrotron. In this run,
D. Tomboulian and P. Hartman studied the spectral and an­
gular properties of the radiation and also made the first
soft X-ray spectroscopy experiments, investigating the trans­
 light sources 129

parency of beryllium and aluminum foils near the K and L

edges.
The National Bureau of Standards (now National Institute
of Standards and Technology) was the next to use SR prop­
erties to their advantage, modifying a section of a vacuum
chamber of a 180 MeV electron synchrotron to enable access
to SR. Soon, it was apparent that the era of SR light sources
had begun.

7.2 Evolution and parameters of SR sources

A large demand for new scientific instruments stimulated
enormous advances in SR light source technology. In just a
few decades, several generations of SR sources technology
have been developed, each exhibiting an improvement with
every evolutionary step.

7.2.1 Generations of synchrotron radiation sources

It is now a tradition to distinguish the many generations of SR
sources according to the following classifications. The first-
generation light sources are the accelerators built for high en­
ergy physics, nuclear physics or other purposes, which used
for synchrotron radiation experiments parasitically.
A large demand for SR experiments resulted in the
construction of dedicated accelerators, creating the second-
generation of purpose-built synchrotron light sources. The
SRS at Daresbury, England, was the first dedicated machine
(operated between 1981 and 2008).
The second-generation light sources employed SR emitted
from bending magnets. Advances in accelerator science and
technology, inspired by a demand from SR users, quickly cre­
ated an opportunity for the next technological breakthrough:
the third-generation light sources — accelerators optimized
for high brilliance due to low electron beam emittance and
the use of insertion devices (wigglers and undulators). Exam­
ples of such SR sources include the European Synchrotron
Radiation Facility (ESRF) in France, the Diamond light source FIGURE 7.3
in the UK and many others. Generations of SR sources.
The third-generation of SR sources is presently the most Brightness is expressed in
widespread. There are several tens of such machines around the units of the number of
the world and the number is growing, following demand in photons per s · mm2 · mrad 2 ·
the field of science. The brightness of the third-generation 0.1%BW .
machines is several orders of magnitude higher than that of
the previous generation (see Fig. 7.3) and exceeds — by about
ten orders of magnitude — the brightness of the sources
available in the beginning of the 20th century.
The fourth-generation light source was brought to
fruition via the free electron laser idea, which was developed
130 unifying physics of accelerators, lasers and plasma

in the 1970s by John Madey. An essential part of the FEL,

the undulator (invented in 1947 by Ilya Ginzburg) provided
a significant cornerstone for the foundation of FEL technol­
ogy development.
The fourth-generation light sources enjoy all the latest de­
velopments of accelerator science: they are linac-based with
low-emittance photo-injectors, assisted with several bunch
compressors that help to obtain the ultra-short bunches.
Examples of such FELs include: FLASH (Germany), LCLS
(USA), and SACLA (Japan).
The next generation of SR light sources will inevitably ar­
rive, and very soon. The exact design and underlying tech­
nology are not yet determined, and in fact several ideas are
in competition, one being the ultimate storage rings discussed
in Chapter 3 and another being related to plasma accelera­
tion light sources discussed in Chapter 6, among others.

7.2.2 Basic SR properties and parameters of SR sources

Without repeating the formulae, which can be found in
Chapter 3, let us briefly recall our approach for deriving im­
portant parameters of SR sources and of the synchrotron ra­
diation itself.
Recall that a simple back-of-the-envelope treatment of SR
was possible when the amount of radiation left behind was
determined from simple geometrical consideration and the
volume integral over the field squared gave us the energy lost
per unit of length. This led us immediately to the estimation
of the cooling time and, after considering the quantum (sta­
tistical) character of radiation, we came up with an estimate
for the equilibrium horizontal emittance of the beam (with
vertical equilibrium emittance determined by the coupling
of the ring).
Knowledge of equilibrium emittances, as well as knowl­
edge of the single photon emittance that we estimated on the
way, firstly led us to an estimation of the SR flux and then of
the brightness of SR sources.
High-intensity photon flux (defined as the number of pho­
tons per second and per spectral bandwidth) allows for either
rapid experiments or the use of weakly scattering objects or
crystals. The brightness of a light source (also called, inter-
Brilliance = Photons/ changeably, brilliance or spectral brightness) is determined
s · mm2 · mrad 2 · BW as the number of photons emitted per second per surface area
and per solid angle and per fraction of a spectral bandwidth.
The high brilliance of SR sources is enabled by low emittance
(and thus a low emitting area) and a low divergence of radia­
tion emitted by an ultra-relativistic beam.
SR covers a broad light spectrum, from microwaves to
hard X-rays, hence allowing a variety of experiments. Experi­
ments that require precise photon energies can benefit either
from the use of a monochromator or from the ability to adjust
 light sources 131

the emission wavelength of the insertion devices. Polarization

of SR can be either linear or circular, as required by the ex­
periment, and depends on the design of insertion devices.
The coherence properties of SR sources vary — from par­
tial coherence in third-generation sources to full temporal co­
herence in early FELs and full temporal and spatial coherence
in the latest designs of pre-seeded FELs. We will discuss some
of the corresponding methods in Chapter 10.
The temporal structure of SR radiation in third- and
fourth-generations span many orders of magnitude, starting
from tens of ps SR flashes in ring-based sources to ultra-short
tens of fs flashes in FELs. The sources are designed to have
high spatial and temporal stability of the emitted radiation,
to submicron levels in space, and tens of fs levels in time.
Discussing the dependences of SR properties on parame­
ters in Chapter 3 led us to the concept of the ultimate bright­
ness of a diffraction-limited SR source — the ultimate storage
ring where the equilibrium horizontal emittance is at least as The tendencies to increase
small as the emittance of radiated photons. Reaching the ul- the ring perimeter and tighten
timate parameters, as Eq. 3.30 suggests, may require increas- the focusing (increased seg­
ing the perimeter of the ring and decreasing the average of mentation of the ring) are
η 2 /βx i.e., employing tighter focusing, which can be achieved demonstrated by the devel­
by shortening the distances between focusing elements. oping plans of upgrades of
many third-generation SR
sources as well as by the
emerging designs of ultimate
7.3 SR source layouts and experiments storage rings.
The modern third-generation light sources are state-of-the­
art facilities that provide multiple X-ray beamlines for a vari­
ety of experiments (Fig. 7.4).

0DFURPROHFXODU
6RIW&RQGHQVHG
&U\VWDOORJUDSK\
0DWWHU
6XUIDFHVDQG
(OHFWURQ ,QWHUIDFHV
*XQ
6SHFWURVFRS\ /LQDF
%RRVWHU
6\QFKURWURQ
6SHFWURVFRS\

0DWHULDOV

(QJLQHHULQJDQG
(QYLURQPHQW

FIGURE 7.4
Generic SR light source with multiple X-ray beamlines and show­
ing typical allocation of beamlines to experiments.
132 unifying physics of accelerators, lasers and plasma

In this section, we will look into a generic layout of such

sources, and will briefly touch on their experimental capa­
bilities.

7.3.1 Layout of a synchrotron radiation source

A schematic of a generic third-generation synchrotron radia­
tion source is shown in Fig. 7.5. Electrons are typically gener­
ated in an RF gun and accelerated in a linac (usually to a few
hundred MeV), further accelerated to the required energy (of
a few GeV) in a booster, and then injected into the storage
ring where the circulating electrons emit an intense beam of
synchrotron radiation.
The optics of the storage ring are arranged in such a way
so they have many empty drift sections where insertion devices
(ID) — wigglers and undulators — can be installed. Each of
these IDs will direct light into a corresponding X-ray beam-
line, which can then be tailored to a particular type of ex­
periment (life science, materials, etc.). The typical amount of
X-ray beamlines is a couple of dozen, as illustrated in Fig. 7.4.

FIGURE 7.5
Schematics of a generic third-generation SR light source.

The second-generation sources use radiation emitted in

For example, for B = 1.4 T bending magnets, which emit a continuous spectrum char-
and E = 3 GeV εc = 8.4 keV. acterized by critical energy εc , which can be estimated as
εc (keV) = 0.665 B(T)E 2 GeV.
The third-generation employs insertion devices (undula­
tors and wigglers). Either of these devices is a periodic array
of magnetic poles that provide a sinusoidal magnetic field B
on axis: B = (0, B0 sin(ku z), 0) where ku = 2π/λu . The max­
imal radius of the curvature of the orbit in the sinusoidal
 light sources 133

field R defines two distinct regimes (see Section 3.3.5). If

2R/γ « λu /2 (parameter K ∼ γ λu /R » 1), then the radia­
tion emitted at each period of sin-like field is independent;
this corresponds to a wiggler regime (similar to radiation
from a sequence of bends). The insertion devices working in
a wiggler regime are usually used to achieve high photon en­
ergies and flux.
In contrast, the undulator regime corresponds to the case
2R/γ » λu /2 (or K « 1), which means that the entire wig­
gling trajectory will contribute to radiation. Undulators are
typically used to generate high brilliance radiation in a quasi-
monochromatic spectrum — the bandwidth of undulator ra­
&XUUHQWDUEXQLWV
diation is inversely proportional to the number of undulator 
E
periods Nu and can be estimated as Δf /f ∼ 1/Nu . The precise 
definition and meaning of parameter K, as well as the wave­  D

length of undulator radiation, will be discussed in the next

chapter, in relation to FELs.
To a large extent, the scientific performance of third-    
7LPHKRXUV
generation SR sources depends on their stability. The cur­
rent in the storage ring, decaying between injection cycles
FIGURE 7.6
(due to the Touschek effect: intrabeam scattering resulting in
Current in SR light source
a change of particle momentum and its consequent loss on
without (a) and with (b) top-
the energy acceptance aperture) forces the power of emitted
up injection mode.
SR to change, affecting the temperature regime and stability
of the ring and of the X-rays’ beamlines. The top-off injection
(also called top-up) is the operation regime (see Fig. 7.6) that
keeps the beam current in the ring almost constant, improv­
ing the stability significantly. In this regime, a small amount
of current is injected into the ring much more frequently than
in the standard regime.

;UD\V
7.3.2 Experiments using SR
Synchrotron radiation allows for a wide array of experiments,
ranging from utilization of phenomena in X-ray scattering, D
X-ray absorption or X-ray fluorescence, to various advanced
methods that enhance the resolution of obtained images (e.g.,
relying on X-ray absorption near the atomic spectral edges &U\VWDOSODQHV
of particular elements contained in the studied samples). Al­
ready, a variety of imaging methods allow for the use of SR
in biological, chemical, medical and material studies, and in
many other areas of science and technology. ;UD\V
SR experiments often require an X-ray beam with a well-
defined wavelength. Monochromatization is typically per­
formed by the crystal monochromators. A variety of config­
urations of crystal monochromators are possible — Fig. 7.7 E
shows two particular arrangements. In both of these cases,
the geometry is selected in such a way that the desired X-ray FIGURE 7.7
wavelength λ corresponds to Bragg conditions Crystal monochromator of X-
rays. Symmetric case (a) and
n λ = 2 d sin θ (7.1) asymmetric case (b).
134 unifying physics of accelerators, lasers and plasma

which correspond to the maximal reflection of X-rays from

the crystal. In the above equation, n is an integer, d is the
distance between the crystal planes and θ is the angle be­
tween X-rays and scattering crystal planes. The monochro­
mator plates are usually made from crystals of Si or Ge. The
symmetrical configuration shown in Fig. 7.7.a is standard,
while the asymmetrical one (Fig. 7.7.b) allows for the increase
in angular resolution — thus narrowing the resulting energy
spread of the X-ray beam.

FIGURE 7.8
Absorption (left) and phase contrast (right) X-ray imaging and
comparison of reconstructed image (middle).

Due to the small size of the area that emits X-rays, the SR
light sources can utilize an advanced technique called phase
contrast imaging — shown in Fig. 7.8 in comparison with stan­
dard absorption imaging.
Phase contrast imaging is particularly appropriate for
studies of biological objects where the density difference, and
thus absorption difference between different tissues, is mini­
mal, which complicates the goal of achieving high-resolution
images relying on absorption (left part of Fig. 7.8). However,
benefiting from the point-like nature of the emitting source,
one can increase the distance between the object and the de­
tector plane, and rely instead on refraction of X-rays caused
by the density variations in the object. The consequent inter­
ference pattern on the detector plane will have much sharper
features, thus reconstructing images with better resolution
and contrast (right side of Fig. 7.8).
The phase contrast imaging technique is especially bene­
ficial for laser plasma betatron light sources (see Chapter 6),
as the emitting areas can have sizes below a micrometer. The
relatively low average brightness of such sources would then
be compensated by higher spatial resolutions, which are ad­
 light sources 135

ditionally enhanced by extremely short temporal durations

of X-ray flash.
To conclude this section, let us recapitulate the capabili­
ties of the modern SR sources by referring to a well-known
example. In 1952, DNA structure was studied in R. Franklin
and R. Gosling’s experiments using an X-ray tube that had
a brilliance of around 108 ph/ sec /mm2 /mrad2 /0.1 BW . At
that time, the duration of exposure needed to acquire the
necessary statistics was typically as long as one day (around
105 sec). The modern third-generation light sources with a
brilliance of the order of 1020 can provide the same exposure
in just 100 ns.
The much higher brightness is not the only advantage
of modern SR sources. Engaging lasers in combination with
SR sources creates a completely new type of experiment —
a pump-probe configuration, in which a laser pulse synchro­
nized with the beam revolution in an SR source excites the
object just before the main X-ray pulse arrives from the SR
source (as illustrated in Fig. 7.9).

FIGURE 7.9
Pump-probe experiment arrangement. Here T and n are revo­
lution period and number of bunches in the SR ring, Δt is time
delay between the pump laser pulse and SR probe pulse.

The specific feature of this kind of experiment is the abil­

ity to vary the time delay Δt between the pump and probe
pulses. The pump-probe experiments are therefore in partic­
ular useful for studies of ultra-fast phenomena such as spin
dynamics in metals, structural molecular dynamics of pro­
teins, photosynthesis, ultrafast photo-switching and many
others.

7.4 Compton and Thomson scattering of photons

Thomson and Compton processes describe the scattering of
an EM wave or photon on a charged particle. The Compton
scattering, in particular, can be very useful for creating com­
136 unifying physics of accelerators, lasers and plasma

pact X-ray sources — enabled by the development of electron

accelerators and laser technologies.
In this section — after reviewing the basic formalism of
the Thomson and Compton processes — we will discuss the
typical design and characteristics of Compton X-ray sources.

7.4.1 Thomson scattering

The elastic scattering of an electromagnetic plane wave by an
electron at rest (or low energy E) with mass me and charge q,
is a process known as Thomson scattering.
The total cross section of a classical Thomson scattering is
given by the following equation:
(0ZDYH 8π 2
σT h = r ≈ 0.665 · 10−28 [m2 ] (7.2)
3 e
And the differential cross section, illustrated in Fig. 7.10, is
equal to
dσ 1
= r 2 1 + cos2 θ (7.3)
dΩ 2 e
Thomson scattering is an approximation of an elastic pro­
FIGURE 7.10 cess — the energies of the particle and photon are the same
Thomson scattering. before and after the scattering (i.e., the recoil of the electron
can be neglected, in contrast to the Compton scattering).

7.4.2 Compton scattering

Compton scattering describes the inelastic process where we
can no longer neglect the transfer of energy between the par­
ticle and the photon.
We are, in particular, interested in the instance when a
collision between a high-energy electron and a low-energy
photon results in a substantial fraction of the electron energy
being transferred to the photon. In the laboratory reference
frame, this manifests as backscattering of the photon with a
significant energy boost; this process is known as Compton
backscattering (or inverse Compton scattering), as illustrated
in Fig. 7.11.


H
(  PF

FIGURE 7.11
Compton backscattering. Initial photon with wavelength λ1 and
after scattering with λ2 .
 light sources 137

In the approximation of small angles and relativistic elec­

trons (see Fig. 7.11), the wavelength of the photon after
Compton backscattering is described by the following equa­
tion:

λ 2 = λ 1 1 + θ 2 γ 2 / 4γ 2 (7.4)
As we can see from Eq. 7.4, in the case of relativistic
electrons, the photon gains considerable energy after inter­
action: its wavelength is shortened by the factor of 4γ 2 . Let’s
consider two examples in the case of green light with λ1 = Green laser (532 nm) scat­
532 nm (corresponding photon energy is 2.33 eV). If the to- tered from an 18.6 MeV elec­
tal electron energy is 5.11 MeV (γ = 10) then λ2 = 1.33 nm tron beam turns into X-rays
(equivalent to 0.93 keV energy of the photons). For a slightly with 0.1 nm wavelength.
larger energy of electrons of 18.6 MeV (γ = 36.5) the scat­
tered wavelength would reach an angstrom: λ2 = 0.1 nm (or
12.4 keV).
A derivation of the Compton process kinematics can be
obtained through considering the relativistic invariants of
electrons and photons before and after collision, as illustrated
in Fig. 7.12. Below, we will reproduce only the final results.

 I I VLQ  I  FRV 

%HIRUHFROOLVLRQ
$IWHUFROOLVLRQ

PH
 ]
(ISI

FIGURE 7.12
Compton scattering in the rest frame of an electron and relativis­
tic invariants.

The resulting expression for the total Compton scattering

cross section in the center of mass reference frame is as fol­
lows (we assumed c = 1 in this and the next section):
r 2 1
2π me2 re2 E
σtot = 2
ln cm (7.5)
Ecm m2e
where
Ecm = m2e + 2pf ωf (1 − cos θ) (7.6)
In the laboratory reference frame the total cross section is:
( )
2πre2 4 8 1 8 0.5
σtot = 1− − 2 ln (1 + x1 ) + + − (7.7)
x1 x1 x 1 2 x1 (1 + x1 )2
The differential cross section in the laboratory frame:
( ) ( )
dσ 2
ωf 2 x1 x2
= 2r e 4y (1 + y) − − (7.8)
dΩ m e x1 x2 x1
138 unifying physics of accelerators, lasers and plasma

where the parameters are defined as follows:

ωi ωf 1 1
x1 = 2γ (1−β cos φ 1 ) ; x2 = −2γ (1−β cos φ 2 ) ; y= +
me me x1 x2

Here β is the relativistic factor and angles ϕ1 and ϕ1 are de­

fined as shown in Fig. 7.13 in the next section.

7.4.3 Compton scattering approximation

We will now consider the approximation of the Compton
cross section given in the previous section in case x1 « 1
or γ ω1 « me . In this instance, the expression in the curved
parenthesis in Eq. 7.7 will simplify to
( )
4 8 1 8 1 4 x1
1− − ln (1 + x1 ) + + − →
x1 x12 2 x1 2(1 + x1 )2 3

Therefore the total cross section in this approximation is

given by the following equation:

8π re2
σtot = (7.9)
3
which shows that the total Compton scattering cross section
is very close to the Thomson one.

 I


H
 L

FIGURE 7.13
Compton scattering — definition of frequencies and angles.

7.4.4 Compton scattering characteristics

In the approximation described above, we can evaluate the
rate of emitted X-rays as the product of the Thomson cross
section and the luminosity L characterizing interaction be­
tween the electron and laser beams. Assuming head-on colli­
sion of the beams, we write for the luminosity:

Ne Nγ f
L= (7.10)
2π σx σy
 light sources 139

In the above, Ne and Nγ are the numbers of electrons and

photons in the colliding bunches and f is the repetition fre­
quency of collisions. The beam sizes in Eq. 7.10 are the con­
volution of electron and photon beam sizes:
)
σ = σe2 + σγ2 (7.11)

The above equations help us to estimate the rate of X-ray pro­

duction as
dNγ
= σtot L (7.12)
dt
In relativistic approximation γ » 1, the expression for the
final frequency of Compton scattered photons is given by the
following equation

2γ 2 ωi (1 − cosφ 1 )
ωf ≈ ωi
(7.13)
1 + (γφ 2 )2 + 2γ m (1 − cosφ 1 )
e

which identifies the following characteristics of Compton

scattering. There is a clear dependence between scattered
photon energy and its angle — this can be useful for the
selection of monochromatic beam with help of collimation.
The majority of the X-ray flux is emitted into a cone with
an opening angle of 4/γ. The photons of maximum energy
ωc = 4 ωi γ 2 come from a head-on collision (ϕ1 = π), while
the photons of half-maximum energy come from a ϕ1 = π/2
collision.
The equations and dependencies defined above aid us in
estimating basic parameters of Compton-based light sources.
Let us now consider the design of such light sources and their
typical characteristics.

7.5 Compton light sources

A generic Compton light source based on a linear accelerator
is shown in Fig. 7.14.
In this design, a train of electron bunches produced by

FIGURE 7.14
Generic Compton source of linac type.
140 unifying physics of accelerators, lasers and plasma

a photo RF gun is accelerated in a linac up to a few tens of

MeV (or several hundred MeV, depending on the applica­
tion). It is then compressed longitudinally and sent for col­
lision with laser bunches accumulated in the laser cavity. The
laser pumps the cavity through a semi-transparent mirror
and the laser intensity buildup in the cavity can exceed the
intensity in the single laser pulse by more than a hundred­
fold, increasing the brightness of the linac-based Compton
source.
Further enhancement of the brightness of the linac-based
Compton source can be achieved by employing the method
of energy recovery. The Compton cross section of electron-
photon interaction is rather low and therefore, in typical con­
figurations, the majority of electrons will pass through the
laser pulses without interacting. The major fraction of the
electron beam can therefore be decelerated, after Compton
interaction, and its energy recovered, reducing the required
RF power and increasing the current of the electron beam.
The energy recovery based Compton sources are mostly suit­
able for superconducting linac technology.
Another type of Compton source is one based on an
electron storage ring, as illustrated in Fig. 7.15. In this
case, a short linac injects an electron beam into a compact
ring, where — due to SR — the electron beam emittances
are cooled, thus helping to achieve a higher luminosity of
electron-laser interaction. The laser cavity is typically located
around or inside one of the straight sections of the ring with
the interaction region (IR) in the center.
Parameter ranges of existing or planned Compton sources
allow for their application in a variety of areas. In particular,
10-30 MeV accelerator produces (with a typical laser) X-rays
tuneable from a few keV to around 50 keV, which can be ap­
plied to high resolution clinical imaging systems or various
types of biomedical research.
Either the linac- or ring-based Compton source can be
rather compact, fitting in a room of a few meters by a few
meters, which enables the use of such sources in areas where
it would not be possible before. For example, a THOMX1
FIGURE 7.15 Compton source is being considered for use in cultural her­
Generic Compton light itage applications, and might be installed in a museum for
source based on electron nondestructive studies of precious paintings without the
storage ring. need to transport them.
Compton sources aimed at larger energies of X-ray pho­
tons, toward the 1–5 MeV range, have their particular niche.
The phenomenon of nuclear resonance fluorescence helps
to create imaging instruments with excellent isotopic sen­
sitivity; therefore such Compton sources can also assist in
nuclear-waste management.
1 A. Variola et al., THOMX Conceptual Design Report. LAL RT 09/28,
SOLEIL/SOU-RA-2678, 2010.
 light sources 141

The Compton sources based on a linac, mentioned above,

require superconducting RF (SRF) technology to achieve
maximal brightness. Compton sources are compact and
therefore may be suitable for small organizations, labs, hospi­
tals or universities. However, most of the modern SRF cavities
with frequencies around 1 GHz require operation at 2 K tem­
perature. There is, nonetheless, a contradiction here as the
cryogenics system, in particular for 2 K, can be bulky and ex­
pensive.
A possible solution for this issue is to revert to lower fre­
quency SRF cavities, which operate at 4 K temperatures, re­
sulting in a significant reduction in complexity, size and cost
of the cryogenic system. The practical frequency range of SRF
cavities that work at 4 K is around 200-500 MHz. The larger
transverse size and lower accelerating gradient of the lower
frequency cavities can sometimes be compensated by modi­
fying their design, e.g., use of spoke cavities.
142 unifying physics of accelerators, lasers and plasma

EXERCISES
7.1 Chapter materials review.

Describe the methods of the creation of a monochromatic X-

ray beam in Compton sources.

7.2 Chapter materials review.

In laser plasma acceleration, the final energy of an acceler­

ated electron beam is 1 GeV. The wavelength of the laser used

for laser plasma acceleration is 800 nm. Part of the same laser

pulse is redirected with mirrors to collide head-on with the

accelerated electron beam. Estimate the energy of photons

created in such a Compton source and the angular spread of

the photons.

7.3 Mini-project.

Select a desired photon energy of a laser plasma acceleration

betatron X-ray source (e.g., from 1 to 100 keV) and devise

a consistent set of basic parameters describing the source.

Discuss the justifications for selecting particular values of

certain parameters (for plasma or laser, etc.). Estimate the

brightness of the source.

7.4 Mini-project.

Select a desired photon energy of a Compton X-ray source

(e.g., from 1 keV to 10 MeV) and devise a consistent set of

basic parameters describing the source. Discuss the reasons

for selecting particular values of certain parameters (for elec­

tron beam, laser, etc.). Estimate the brightness of the source.

7.5 Analyze inventions or discoveries using TRIZ and AS-TRIZ.

Analyze and describe scientific or technical inventions de­

scribed in this chapter in terms of the TRIZ and AS-TRIZ ap­
proaches, identifying a contradiction and an inventive prin­
ciple that were used (could have been used) for these inven­
tions.

7.6 Developing AS-TRIZ parameters and inventive principles.

Based on what you already know about accelerator science,

discuss and suggest the possible additional parameters for

the AS-TRIZ contradiction matrix, as well as the possible ad­

ditional AS-TRIZ inventive principles.

 8
Free Electron Lasers

8.1 FEL history 143 In this chapter we will continue to build upon the results of
8.2 SR from bends, the previous chapters — particularly Chapter 3 (Synchrotron
wigglers and Radiation) and Chapter 7 (Light Sources) — and will discuss
undulators 144 the present reigning champion among the X-ray light sources:
8.3 Basics of FEL the free electron laser.
operation 147 We will begin with a brief, historical introduction. Then,
8.4 FEL types 150 we will recall the properties of radiation from a sequence of
8.5 Microbunching and bends, wigglers and undulators, and then discuss how their
gain 152 radiation spectra compare. Next, we will follow up on un­
8.6 FEL designs and dulator resonance conditions and microbunching. Finally, we
properties 157 will discuss the precise physical meaning and exact defini­
8.7 Beyond the tion of the undulator parameter K, which was introduced in
fourth-generation earlier chapters as a qualitative factor.
light sources 160 Following this introduction of basic FEL concepts, we will
discuss FEL designs and parameters, as well as possible fu­
ture advances in the evolution of FEL technology.

8.1 FEL history

The FEL concept, as well as the term itself, was suggested by
John Madey in the early 1970s during his work at Stanford
University. His research benefited from the earlier work by
Hanz Motz who, in 1953, built an undulator (which was pro­
posed in Vitaly Ginzburg’s 1947 theoretical paper, wherein
the undulator was described as a device generating electro­
magnetic radiation via relativistic electrons).
In 1971, John Madey wrote his first FEL-related published
journal entry on the subject of stimulated emissions of radi­
ation in a periodic magnetic field. This was shortly followed
by a patent on FEL filed in 1972 — but that’s a whole ‘nother
story. The FEL created by Madey used a 43 MeV electron
beam to create radiation with wavelengths of 3.4 μm in a 5 m
helical undulator, with a 3.2 cm period and a field of 0.24 T.
Following this pioneering work, many FELs have been
created all around the world — their wavelengths gradu­
ally shortening as the technology matured. The most mod­
ern FELs have recently reached the Angstrom range, creating
unsurpassed possibilities for discovery science, bio-medical
studies and technology-aimed research.

DOI: 10.1201/b18696-8 143

144 unifying physics of accelerators, lasers and plasma

8.2 SR from bends, wigglers and undulators

We started our discussion of FELs in Chapter 3, where we ap­
proached radiation from wigglers and undulators and com­
pared them to radiation from bending magnets. We recall
that, for relativistic electrons with γ » 1, the emitted pho-
Recall that γ 3 dependence of tons go into 1/γ cone and, if the radius of the curvature of
ωc is due to the length of the trajectory in the magnetic field is R, then the external ob­
the emitting arc (∝ 1/γ) and server will see the photons emitted during the particle travel
photon and particle velocity along the arc 2R/γ. This allowed us to estimate the character-
difference (v − c ∝ 1/γ 2 ). istic frequency of SR as ωc = 1.5c γ 3 /R.
We also recall that the extra factor of γ 2 appears in this
formula for ωc due to the difference between the speed of
photons c and the speed of particles v, estimated as (1−v/c) =
1/(2 γ 2 ). Knowledge of the characteristic frequency allows
us to determine the spectral characteristic of the SR emitted
from the bending magnets.

8.2.1 Radiation from sequence of bends

Assume that a set of bending magnets are arranged in a se­
quence with +–+– polarity with period λu , so that the particle
trajectory through this sequence of magnets wiggles as illus­
trated in Fig. 8.1.

% % % %

5 5
5 5
X


FIGURE 8.1
Trajectory and radiation in a sequence of bending magnets.

Precise definition of K fol- If the length of the emitting region (that a remote ob­
lows in just a couple of pages. server can see) is much less than the length of an individual
bend, i.e., 2R/γ « λu /2, then the radiation emitted in each
bend is independent. Such an arrangement of bends is called
ORJ6   a wiggler (and corresponds to K » 1 where K ∼ γ λu /R, and
where λu /R can be noted as being approximately equal to the
maximal angle of the trajectory).
In a wiggler configuration, the spectrum of emitted SR is
thus expected to resemble to the spectrum from the bends —
the spectrum shape will be similar to the spectrum from a
ORJ   bend while the amplitude will be multiplied by the number
FIGURE 8.2 of wiggles, as shown qualitatively in Fig. 8.2.
Wiggler (top) and bending The opposite regime 2R/γ » λu /2 is different — the en-
magnet (bottom) SR spectra. tire wiggling trajectory contributes to radiation. It is logical
 free electron lasers 145

to assume that there is a wavelength in this K « 1 undulator

regime that would be resonant with the emitting particle. In­
deed, it will emit radiation at this wavelength at each wiggle,
allowing coherent buildup of the amplitude. We will look at
the spectrum of undulators and also this resonant condition
in detail in the following sections.

8.2.2 SR spectra from wiggler and undulator

Despite the similarity of the overall shape of the SR spectra of
bend magnets and wigglers, the external observer will note
an important difference in the details of the spectra.
The fields emitted from wigglers and detected by the
external observer manifest themselves as periodic signals
— short flashes repeating with a period corresponding to
the time of flight between wiggler periods, as illustrated in
Fig. 8.3 and Fig. 8.4.

FIGURE 8.4
FIGURE 8.3 Time profile of radiation ob­
Radiation from wiggler, regime of K » 1. served from wiggler.
Given the periodic nature of radiation emitted by wigglers,
the spectrum of SR wiggler radiation should consist of har­
monics defined by the wiggler period corrected by the factor
(1 − v/c) = 1/(2γ 2 ), which takes into account the relative ve­
locity of particles and radiation (illustrated in Fig. 8.5 at left).
The relative width of each peak in the wiggler spectrum cor­
responds to the number of wiggles Nw , i.e., Δλ/λ ≈ 1/Nw .
As the the entire trajectory contributes to radiation emit­
ted from the undulator, the time structure of the observed ra­
diation is periodic and continuous, as shown in Fig. 8.6 and
Fig. 8.7.
: DX

 

 

 
          
DX DX
FIGURE 8.5
Spectrum from wiggler (left) and undulator (right), qualitative
comparison. Dashed line on the left spectrum corresponds to
the spectrum from bends of the same strength. Horizontal axis
is in units of λu /(2γ 2 ).
146 unifying physics of accelerators, lasers and plasma

FIGURE 8.6
Radiation from undulator, with K « 1.

FIGURE 8.7 The spectrum corresponding to the observed undulator ra-

Time profile of radiation ob- diation will thus contain just one harmonic at a wavelength
served from undulator. close to λu /(2γ 2 ). We will clarify this statement in the next
section.

8.2.3 Motion and radiation in sine-like field

In the example that we considered in Fig. 8.1, we assumed
that the segmented field of the wiggler is uniform and
sharply changes its sign during the transitions between seg­
ments. This may correspond to zero-aperture magnets but, in
practice, is not possible.
A much better way to estimate the field of a wiggler or
undulator is to assume that the field is sine-like, i.e.,

By (z) = B0 sin(ku z) (8.1)

where ku = 2π/λu .
Let us consider a trajectory through such a sine-like field
and let us parameterize it in such a way that the maximum
angle of the trajectory is equal to K/γ, as shown in Fig. 8.8.

FIGURE 8.8
Trajectory and radiation in sine-like field.

The trajectory parametrization can therefore be written as

( ) ( )
K λu 2π z ' K 2π z
x= sin and x = cos (8.2)
γ 2π λu γ λu

This shows us that if K < 1, then the trajectory angle is al­

ways less than 1/γ and the external observer will be able to
see the emitted fields without interruptions; ultimately, the
entire trajectory contributes to radiation.
Let us now connect the maximum field in the undulator
 free electron lasers 147

B0 with parameter K. The bending radius is connected to the

curvature of trajectory via

d 2x 1
=
dz2 R
which gives us
λu γ
K= (8.3)
2πR
and by substituting the expression for the bending radius in
the magnetic field R = p c/(e B0 ) we obtain

λu eB0
K= (8.4)
2πmc2
This finally give us the precise definition of the undulator
parameter.
We are now ready to discuss the basics of FEL operation.

8.3 Basics of FEL operation

In a third-generation light source, the phase relationship be­
tween the radiation emitted by each electron is random and
therefore the spatial and temporal coherence of the radiation
is poor. Even if an undulator were inserted into the ring, the
electrons would emit radiation incoherently.
In contrast to the third-generation light sources, operation
of a free electron laser relies on microbunching of the beam
caused by interaction of the radiation with the beam, i.e., the
beam interacts with itself via the radiation it emits.
Microbunching in FEL happens primarily at the resonant
wavelength determined by the undulator parameter.
Once the electron beam is microbunched, each mi­
crobunch emits radiation as a single particle of a large charge,
in phase with each other — i.e., coherently. Correspondingly,
the radiation power and brightness of FEL will scale as Ne2
and not as Ne as in third-generation sources, thus giving an
enormous boost in performance.
We will now consider the relevant phenomena — reso­
nance condition, energy exchange and microbunching — step
by step.

8.3.1 Average longitudinal velocity in an undulator

The average longitudinal velocity in an undulator is an im­
portant parameter that determines the resonant wavelength.
If the particle moves in a free space, its longitudinal velocity
is approximated as
( )
1
vz0 = β c ≈ c 1 − (8.5)
2γ 2
148 unifying physics of accelerators, lasers and plasma

For the sine-like trajectory that we parametrized above, the

transverse velocity is given by
( )
K 2π z
vx = β c sin (8.6)
γ λu
In the second-order approximation, the longitudinal velocity
can be written as follows
( )
1 vx2
vz ≈ β c 1− (8.7)
2 β 2 c2
and therefore the average longitudinal velocity can be ex­
pressed as ( ( ))
1 K2
(vz ) ≈ c 1− 2 1 + (8.8)
2γ 2
We can see that, in comparison with free space, there is an
additional longitudinal retardation, which is due to the trans­
verse velocity in an undulator. This retardation is equal to
K2
c (8.9)
4γ2
and is determined by the undulator parameter.

8.3.2 Particle and field energy exchange

Energy exchange between an EM wave and electron depends
on the electric field and velocity:
dW
= eE·v (8.10)
dt
If the electrons have only the longitudinal velocity (de­
picted in the left image of Fig. 8.9) no energy can be trans­
ferred between the electrons and the EM wave, as in this case
eE · v = 0.
On the other hand, an electron beam with sine-like trajec­
tory as in an undulator overlaid with an EM wave (right plot
in Fig. 8.9) can exhibit an energy exchange between the EM
wave and electrons as in this case
e E·v 0 (8.11)

FIGURE 8.9
EM wave and particle trajectory — straight (left) and wiggling
(right) in an undulator.
 free electron lasers 149

because v⊥ 0. Therefore, if electrons have a transverse ve­

locity, energy can be transferred between electrons and EM
wave — which is the principle FEL relies upon.

8.3.3 Resonance condition

For certain λ of an EM wave, a resonant energy transfer be­
tween electrons and the EM wave can occur — as illustrated
in Fig. 8.10.

(OHFWURQWUDMHFWRU\
9[
X 9[ V
([

([
FIGURE 8.10
EM wave-particle resonance condition of energy transfer.

The necessary condition for the resonant energy transfer

is that the EM wave slips forward with respect to an electron
by a λ/2 per half period of electron trajectory, i.e.,

λ = λu (1 − (vz ) /c) (8.12)

Taking into account the average velocity in an undulator

defined by Eq. 8.8, we therefore obtain, for the resonant EM
wavelength: ( )
λ K2
λ = u2 1 + (8.13)
2γ 2
We note that, in the undulator case when K « 1, the reso­
nance wavelength is very close to the relativistically trans­
formed undulator period λu /(2γ 2 ).
We should also note that slippage by 3(λ/2), 5(λ/2), 7(λ/2)
and so on would also be in resonance, which may result in
generation of odd, higher harmonics.

8.3.4 Microbunching conceptually

The interaction of particles with the resonant EM wave that
we defined in the previous section can create energy modula­
tion in the particle beam.
150 unifying physics of accelerators, lasers and plasma

FIGURE 8.11
Microbunching. Density of the beam along the longitudinal co­
ordinate for the initial noise (left), intermediate regime of mi­
crobunching (middle) and saturated microbunching (right).

As particles move along the curved sine-like trajectory,

this energy modulation can result in different routes taken
over different trajectories, depending on the particles’ energy.
Different path lengths can in turn create density modulations
along the beam.
An initial EM wave of resonant wavelength can be ex­
ternal (seeding) or can emerge from the noise that is always
present in the beam. The latter corresponds to the self am­
plified spontaneous emission (SASE) process, which is illus­
trated in Fig. 8.11. Here the corresponding harmonics from
the noise of the initial distribution evolve through the linear
regime into saturation regime where complete modulation of
density eventually occurs. The resulting microbunches emit
coherently at wavelength λ with radiated power P∼ N 2 .

8.4 FEL types

The two major kinds of FELs are single pass and multi pass.
While multi-pass FELs were built primarily in the earlier
days of FEL technology, single-pass devices are dominating
the arena today.

8.4.1 Multi-pass FEL

Multi-pass FEL is similar to a standard laser: mirrors help
to build up the optical amplitude on the resonance har­
monic while the electron beam trajectory is arranged to pass
through the undulator while avoiding mirrors, as shown in
Fig. 8.12.
Similar to a normal laser, a single-pass FEL stores the
radiation in a cavity. Such systems usually have low gain
and therefore many reflections of radiation off of mirrors are
needed to create sufficient amplitude. The multi-pass design
is suitable in particular for the light sources used in visible or
near-visible ranges of radiation.
 free electron lasers 151

FIGURE 8.12
Multi-pass FEL.

8.4.2 Single-pass FEL

In a single-pass FEL, the radiation has to grow within a single
passage of the beam through the undulator (Fig. 8.13). Such
FELs can be either seeded or the SASE type.
The need to use single-pass systems is actually a necessity
dictated by an absence of good mirrors in the X-ray spectral
region. The single-pass system has to be a high-gain system,
which puts extreme constraints on the quality of the electron
beam, as well as on the accuracy of the undulator. A typical
permanent magnet undulator is shown in Fig. 8.14.

FIGURE 8.13
Single-pass FEL.
152 unifying physics of accelerators, lasers and plasma

FIGURE 8.14
Radiation in an FEL undulator composed of permanent magnets.

8.5 Microbunching and gain

The term FEL contains the word laser; however, FEL can be
explained entirely using the approach of classical electrody­
namics. In this section, we will consider microbunching in
detail and then discuss the FEL gain factor.

8.5.1 Details of microbunching

Interaction of the radiation emitted in an undulator with the
electron bunch itself can, in certain conditions, be sufficiently
strong to generate a significant modulation of the electrons’
energy in the beam.
In this case, the energy change of the particles will occur
due to the coupling between the transverse (typically hori­
zontal) oscillation of the electron in the undulator and the
transverse (thus also horizontal) component of the electric
field of the emitted EM plane wave. The energy change in
this case can be written as
dE
= eE · v = eEx vx (8.14)
dt
One can also contrast this with the case of acceleration in the
RF cavities, where the energy change occurs because of the
coupling between the longitudinal velocity of the electron in
the RF cavity and the longitudinal component of the electric
field in the RF cavity:
dE
= eE · v = eEz vz (8.15)
dt
We will now consider the process that leads to mi­
 free electron lasers 153

crobunching in detail. The relevant equations of motion are

dp e dE
= eE + v × B and = eE · v (8.16)
dt c dt
where momentum and energy are

p = me γv and E = me c 2 γ (8.17)

and where B and E are the magnetic field of the undulator

and of the undulator radiation.
We will define the undulator magnetic field as

B = B0 0 , cos (ku z) , 0 (8.18)

where B0 is the amplitude of the undulator fields. And we

will define the undulator harmonic radiation as

E = E0 cos α , 0, 0

B = E0 0 , cos α, 0 (8.19)
where E0 is the amplitude of the radiation harmonic and

α = kz − ωt + ϕ and ω = kc (8.20)

As you can see, we have simplified the undulator radiation by

representing it as a plane wave.
The equations defined above can be integrated. Let’s first
change the independent variable from t to z and then inte­
grate the equation. We can find that the transverse velocity
will be expressed as
K eE0
βx = − sin ku z − sin α (8.21)
γ me ωcγ

where the second term corresponds to modulation of the ve­

locity due to the plane EM wave, which represents the radia­
tion.
The energy change expressed in terms of relativistic fac­
tors can be expressed as
r 1
eE0 eE0
γ̇ = − cos α · sin α + K sin ku z (8.22)
me cγ me ωc

These two equations, Eq. 8.21 and Eq. 8.22, constitute a

system of first-order differential equations for (z,γ). We will
solve these equations in the following approximations. We
will assume that the radiation amplitude is small and will
therefore keep only the terms with the first order in E0 . We
will also assume that the gain is small, i.e., the energy change
of the electrons is low (Δγ « γ). And finally, we will assume
that the radiation wavelength is very close to the fundamen­
tal undulator radiation wavelength and that we will average
154 unifying physics of accelerators, lasers and plasma

all quantities over one undulator period in order to remove

fast oscillations.
We begin by introducing the following variable:
ζ = ku z + α = (k + ku )z − ωt + ϕ (8.23)
The above-defined system of the first-order differential equa­
tions can be transformed into a single differential equation of
second order:
eE0 (ku + k)[J0 (ξ) − J1 (ξ)](1 + K 2 /2)K
ζ̈ = − sin ζ (8.24)
2me γ 4
where ξ = K 2 /(4 + 2K 2 ) and J0 and J1 are Bessel functions.
Eq. 8.24 (above) can be expressed as
ζ¨ + Ω 2 sin ζ = 0 (8.25)
which is the so-called FEL-pendulum equation, which de­
scribes the interaction of particles with radiation in an FEL.

6HSDUDWUL[
$

,


 ,,

  
%

  
FIGURE 8.15
Illustrating solutions of FEL-pendulum equation and mi­
crobunching for different initial conditions. The initial beam (I)
is on-energy and when bunched (II) demonstrates symmetrical
profile of beam density (B).

$

,


 ,,
6HSDUDWUL[
  
%

  
FIGURE 8.16
Microbunching in a case when the initial beam is slightly off
energy.
 free electron lasers 155

Take note that the equations Eq. 8.24 or Eq. 8.25 describe
the familiar motion in a separatrix where the period of os­
cillations depends on initial conditions, particularly whether
the particle is infinitely close to the separatrix, in which case
the period becomes infinitely long. Examples of solutions to
this equation for different initial conditions are shown in
Fig. 8.15 and Fig. 8.16.
In the EM wave-particle interaction process described by
the above equations, each electron will gain or lose energy
depending on the relative phase ζ(0) between the transverse
oscillation in the undulator and the phase of the radiation
plane wave. The energy change (ignoring the small terms of
the order of O(Ω 2 )) can be written as

eE0 K [J0 (ξ) − J1 (ξ)] L sin (ν/2)

Δγ = − sin (ζ(0) + ν/2) (8.26)
2me c 2 βz0 γ0 ν/2

where L is the length of the undulator and

( )
ω
ν = k + ku − L (8.27)
cβ0

By averaging the above equation over the initial phases

ζ(0) of the electrons, we obtain the average energy variation:
( )3
eE0 K [J0 (ξ) − J1 (ξ)] Ω 2 L d sin ν/2 2
(Δγ)ϕ = (8.28)
8me cγ0 cβz0 dν ν/2

As there is a balance and conservation of energy, the variation

of the electron energy is equivalent to a variation of the gen­
erated EM wave’s energy in the FEL. This allows us to define
the FEL gain.

8.5.2 FEL low-gain curve

The FEL gain G can be defined as a relative change of the
wave’s energy, equaling to the change of the energy of all elec­
trons involved in the interaction:
ΔEtot N
G= = −me c2 < Δγ>ϕ (8.29)
W0L W0L

where W0L is the initial energy of the wave over the entire
length of the undulator.
For a bunch with the peak current I and transverse area
Σb = FΣL the gain can be written as

πK 2 [J0 (ξ) − J1 (ξ)]2 ku L3 (1 + βz0 ) F I d sin ν/2 2

G=− 3
2γ 3 βz0 ΣL I0 dν ν/2
(8.30)
The corresponding low-signal and low-gain FEL curve is
shown in Fig. 8.17.
156 unifying physics of accelerators, lasers and plasma



*DLQ



    
FIGURE 8.17
FEL low-gain curve.

The positive gain in Fig. 8.17 corresponds to an amplifica­

tion of the EM wave (a standard FEL case) while the negative
gain corresponds to an acceleration of the beam and decrease
of the EM wave’s amplitude. The latter case relates to the so-
called inverse FEL.

8.5.3 High-gain FELs

Most modern FELs, especially those aimed at hard X-rays, are
high-gain systems. In these FELs, the gain is so large that the
EM wave amplitude changes within a single pass in the un­
dulator, and therefore our estimations determined in the pre­
vious section need to be revised.
A detailed analysis — which takes into account the wave
equation with driving terms determined by the oscillat­
ing current density of the beam — predicts an exponential
growth of the radiation power of

P (s) ∝ e s/Lg (8.31)

with the exponent given by the following
( )1/3
1 4γ 3 me
Lg = √ (8.32)
3 μ0 K 2 e 2 ku ne
The exponential growth continues until saturation is reached,
at which point the emitted power starts to oscillate as illus­
trated in Fig. 8.18.

FIGURE 8.18
High-gain FELs, typical behavior of the emitted power — expo­
nential growth eventually turned into saturation.
 free electron lasers 157

Analysis of a high-gain curve such as the one shown in

Fig. 8.18 helps us to determine an optimal length of the un­
dulator for a particular FEL design.

8.6 FEL designs and properties

We will now review typical accelerator parameters, require­
ments and radiation characteristics of modern FELs.

8.6.1 FEL beam emittance requirements

As we have determined in Chapter 3, the emittance of a syn­
chrotron radiation photon beam is given by Eq. 3.33 and is
equal to
λ
εph =
4π
It is intuitive to assume that efficient generation of FEL radi­
ation (the term, lasing, is often used) requires a fair amount
of overlap between the electron and the photon beam.
The geometrical emittance of an electron beam needed for
efficient lasing, therefore, needs to be smaller than the one for
the photon beam. Thus,

λ
ε≤ (8.33)
4π
or in terms of the normalized emittance:
λ
εN ≤ γ (8.34)
4π
As an example, for λ = 0.2 nm and γ = 3 · 104 (≈ 15 GeV),
the required normalized emittance is ≤ 0.5 mm·mrad, which
results in a necessity to use a very bright electron source. We
can also note that the requirement for geometrical emittance
can be eased for higher energy electron beams since, during
acceleration, the geometrical emittance decreases in inverse
proportion to the beam energy.
An important clarification of the above emittance require­
ment relates to the concept of slice emittance. As we discussed
above, radiation slips with respect to the electron beam by λ
for every λu . Even if the undulator is 100 m long, for any rea­
sonable undulator period (e.g., λu ≈ 1 cm) and wavelength
(assume λ ≈ 0.1 nm), the total slippage will be around 1 μm.
A typical electron bunch is usually much longer. Therefore,
only a small longitudinal fraction of the bunch contributes
to a particular spatial portion of generated radiation. The re­
quirement for the emittance defined above in Eq. 8.33 is thus
applicable to the slice that is generating this portion of the
radiation.
158 unifying physics of accelerators, lasers and plasma

8.6.2 FEL and laser comparison

It can be useful to compare the properties and features of
FELs and lasers side by side — see Table 8.1. Amongst many
similarities, the most notable difference is in their theoretical
foundations: while the laser is a purely quantum device, an
FEL is entirely classical.
TABLE 8.1 FEL and laser comparison
LASER FEL
Characteristics Source of narrow, monochromatic Source of narrow, monochromatic
and coherent light beams and coherent light beams
Configurations Oscillator or amplifier Oscillator or amplifier
First demonstration 1960 1977
Laser media Solids, liquids, gases Vacuum with electron beam
in periodic magnetic field
Energy storage Potential energy of electrons Kinetic energy of electrons
Energy pump Light or applied electric current Electron accelerator
Theoretical basis Quantum mechanics Relativistic mechanics
and electrodynamics
Wavelength Energy levels of laser medium Electron energy, magnetic
definition field strength and period

8.6.3
8.6.3 FEL
FELradiation
radiationproperties
properties
Due
Duetotothe
thecoherent
coherentnature
nature of
of radiation,
radiation, FELs
FELs can
can provide
provide
peak
peak brilliance of around 8–10 orders of magnitudes larger
brilliance of around 8–10 orders of magnitudes larger
than
thanthat
thatofofstorage
storagering
ringlight
lightsources
sources(see
(see Fig.
Fig. 8.19).
8.19).

 
UGJHQHUDWLRQ
 /LJKW6RXUFHV

7HPSRUDOUHVROXWLRQIVUPV

WKJHQHUDWLRQ
 WKJHQHUDWLRQ
/LJKW6RXUFHV
3HDNEULOOLDQFH

/LJKW6RXUFHV





WKJHQHUDWLRQ
 UGJHQHUDWLRQ /LJKW6RXUFHV
/LJKW6RXUFHV ORZFKDUJH



 
             

3KRWRQHQHUJ\H9 5HSHWLWLRQUDWHN+]

FIGURE 8.19
Peak brilliance (left) and temporal resolution (right) of typical FEL
in comparison with third-generation SR sources.
 free electron lasers 159

Since the repetition rates of the typical FELs is still below

that of storage rings, the average brilliance of FELs exceeds
that of third-generation light sources typically by about 2–4
orders of magnitude.
The FELs are also unsurpassed in temporal resolution,
exceeding the third-generation by 2–4 orders of magnitude,
able to reach femtosecond resolutions in special configura­
tions of low-charge sliced beams (the methods of ultra-short
bunch generation are discussed in more detail in Chapter 10).

8.6.4 Typical FEL design and accelerator challenges

Modern high-brightness FELs aimed at hard X-ray ranges are
sophisticated machines. A typical layout of such an FEL is
shown in Fig. 5.39. A high-brightness laser-driven photo RF
gun, operating from a kHz to potentially MHz range, can
power a linear accelerator, a normal conducting (NC) or a su­
per conducting (SC). Typically, two bunch compressors, at in­
termediate and final energies, are used to achieve the shortest
possible beam length. The accelerated and compressed elec­
tron beam is sent into an undulator, or switched between dif­
ferent undulators.
Practical constraints (size, cost) split the FEL designs into
two families — NC and SC FELs, which operate under two
distinct sets of parameters. Normal conducting FELs can have
the highest electron beam energy and therefore the shortest
X-ray wavelength; however, they are pulsed and usually have
rather low (sub-kHz) repetition rates.
Superconducting FELs, on the other hand, can be either
pulsed (but with long pulses and many bunches in a sin­
gle pulse) or are continuously operating (CW) with repeti­
tion rates of bunches up to MHz. They, however, are gener­
ally limited in electron energy and cannot reach the shortest
wavelength (see Table 8.2).
TABLE 8.2
NC and SC FEL, typical parameters
Characteristics NC-FEL SC-FEL
Linac energy 6–15 GeV 2–4 GeV
Linac frequency S, C, X band L band
Repetition frequency 50–200 Hz kHz–MHz
Operation mode pulsed long pulse to CW
Minimum wavelength 0.1 nm 0.3–1 nm

Accelerator physics challenges in FELs start from the

source — a very low emittance gun is needed, seeing as the
normalized emittance cannot be improved in the linac. Accel­
eration and compression through the linac needs to be per­
formed to keep the emittance low while managing the col­
lective effects of high-brightness beams (which is not trivial,
160 unifying physics of accelerators, lasers and plasma

particularly at a low energy). Compression of the bunch to a

tens of fs range is challenged by coherent synchrotron radiation
(CSR), which will be discussed in closer detail in Chapter 10.
Another group of challenges related to FELs is simultane­
ously connected to both the physics and technologies of accel­
erators and lasers. Firstly, this refers to the synchronization of
conventional lasers and FEL electron sources for pump-probe
experiments. And secondly, conventional lasers are essential
to applications in FELs for the seeding of radiation in order to
ensure improved temporal coherence of generated X-rays, as
well as for high harmonic generation (HHG) (discussed further
in Chapter 10).

8.7 Beyond the fourth-generation light sources

Let us summarise the properties of third- and fourth-
generation light sources and outline prospects for future gen­
erations.
The third generation (storage rings) and FEL have com­
plementary properties. The storage ring synchrotron light
sources are extremely stable, can serve several tens of user
beamlines, and their radiation can approach full transverse
coherence for ultimate diffraction-limited storage rings.

FIGURE 8.20
Generic layout of a compact light source driven by an LPWA.

The fourth-generation (FELs) sources have high bright­

ness, short pulses and full transverse coherence. However,
FELs can serve only a few beamlines at a time. Hard X-ray
 free electron lasers 161

high brightness FELs are machines that require considerable

resources for their construction and can usually be realized
only as national-scale facilities.
New technological solutions are required if we are to
build more economical and compact radiation sources.
Progress with laser plasma accelerators in the last several
years has opened up new possibilities to work towards com­
pact betatron radiation sources and has also created an aspi­
ration in the community to use this technology to drive com­
pact FELs.
The first generation of undulator radiation in the soft X-
ray range driven by an LWFA-produced beam has already
been obtained. In this first experiment, a laser plasma wake-
field accelerator produced 55–75 MeV electron bunches,
which were then sent to an undulator to generate visible to
IR-range synchrotron radiation.1
A generic layout of a compact light source based on laser
plasma acceleration and an undulator is shown in Fig. 8.20.
Such compact sources are often referred to as table-top al­
though at present time a more appropriate term would be
“room-sized.”
Laser plasma wakefield accelerators demonstrated the
possibility of generating a GeV beam with promising elec­
tron beam qualities, including a normalized emittance of the
order of 1 mm mrad and an energy spread of close to 1% for
the entire bunch.
Recall that lasing in FEL requires the beam slice to have
appropriately small emittances; these characteristics are al­
ready within reach for laser plasma acceleration and will
likely be achieved with relatively modest improvements on
what has presently been obtained. The beam energy spread
will, however, require more noticeable improvements from
the presently achieved values of a few percent for the entire
beam to around a few hundredth of a percent for a radiation-
generating slice.
Furthermore, an FEL based on laser plasma acceleration
will require significant improvement to the stability of LPWA
beams; at present, the repeatability of its beam parameters is
poor and important characteristics of beams can often exhibit
nearly 100% pulse-to-pulse fluctuations.
We should also note that an election beam generated in a
plasma bubble typically has inconvenient ratios between its
size and its angular spread. While the beam size may be of FIGURE 8.21
the order of a micron, the angular spread can reach several For illustration of filamenta­
milliradians. Even if the calculated emittance of such a beam tion. An intact paper sheet
is very small, its usefulness strongly depends on the ability to (top) may have very low vol­
quickly capture this beam into an appropriately designed fo­ ume; however, when crum­
cusing channel, in order to avoid effective emittance growth pled (bottom) it will have its
due to filamentation — nonlinear wrapping-around in the effective volume increased
by orders of magnitude.
1 Schlenvoigt et al., Nature Phys. 4, 130 (2008).
162 unifying physics of accelerators, lasers and plasma

beam phase space as illustrated (in a synectics approach of

using analogies) by a paper sheet that increases its effective
volume by orders of magnitude after being crumpled — see
Fig. 8.21.
Preservation of the emittance of an LPWA beam during its
transport from the plasma bubble into the undulator would
require properly catching the accelerated beam into a focus­
ing system. The focusing distances of the first lenses of the
catching optics need to be comparable to the focusing in the
plasma bubble, and the focusing distances of further lenses
should gradually increase to match the focusing strength of
the undulator beamline. An alternative approach would be to
create an undulator inside of the plasma — a concept which,
if proven feasible, may create light sources of ultimate com­
pactness.
 free electron lasers 163

EXERCISES
8.1 Chapter materials review.

Evaluate the slice emittance requirements for an FEL based

on 1 GeV electron beam. Discuss the factors affecting the re­

quirements for the slice energy spread of this beam in this

FEL.

8.2 Chapter materials review.

Discuss the phenomenon of filamentation in plasma acceler­

ation in connection to the TRIZ inventive principle of prelim­
inary action.

8.3 Mini-project.

Define, very approximately, the main parameters (energy,

length, undulator field and step size) of a linac-based FEL

aimed at 15 KeV X-ray energy.

8.4 Analyze inventions or discoveries using TRIZ and AS-TRIZ.

Analyze and describe scientific or technical inventions de­

scribed in this chapter in terms of the TRIZ and AS-TRIZ ap­
proaches, identifying a contradiction and an inventive prin­
ciple that were used (could have been used) for these inven­
tions.

8.5 Developing AS-TRIZ parameters and inventive principles.

Based on what you already know about accelerator science,

discuss and suggest the possible additional parameters for

the AS-TRIZ contradiction matrix, as well as the possible ad­

ditional AS-TRIZ inventive principles.

 9
Proton and Ion Laser Plasma
Acceleration

9.1 Bragg peak 166 In Chapter 6, we primarily discussed the plasma accelera­
9.2 DNA response to tion of electrons. There are strong motivations, however, for
radiation 169 development of plasma acceleration of protons and ions, as
9.3 Conventional further advances in this area could improve current ways of
proton therapy treating tumors.
facilities 171 The advantages of using protons and ions (in comparison
9.4 Plasma acceleration to using electrons and X-rays) are associated with the phe­
of protons and ions nomenon of Bragg peak — we will start our discussion by
— motivation 176 looking at this phenomenon in detail.
9.5 Regimes of proton Regarding the application of accelerators for therapy, it is
laser plasma important to know how either the electrons or protons can af­
acceleration 176 fect the living cells. We will therefore briefly overview, in this
9.6 Glimpse into the chapter, how the mechanisms of DNA respond to radiation.
future 182 The conventional accelerator systems used for particle
beam therapy for tumors typically require protons with
around 250 MeV of energy. The accelerator is typically a
synchrotron or a cyclotron (so far the most popular). A
beam therapy facility usually includes beam delivery gantries
(which send the beams in a selected target volume), as well
as collimators and degraders. We will review the design
and functionality of conventional beam-therapy facilities in
a later section.
A conventional accelerator system for particle beam ther­
apy, with a beam source and several gantries, may require
several thousands of square meters of space, making these
respective facilities large and expensive to maintain. On the
other hand, plasma acceleration of protons and ions is one of
the most promising means by which we can make beam ther­
apy more affordable and more accessible to patients. After we
discuss the motivations to apply plasma acceleration to this
area, we will briefly overview the present understanding of
the different regimes and mechanisms of proton/ion plasma
acceleration.
We would like to once again note that proton/ion laser
plasma acceleration is a rapidly developing area and, more­
over, it involves rather complicated physics. There are var­
ious regimes of acceleration identified and several mecha­
nisms that explain the experimental results. The different
mechanisms of acceleration compete, and often more than
one mechanism acts at the same time. The mechanisms of ac­
celeration are often just approximate models that represent
a gradual improvement of our understanding of proton/ion

DOI: 10.1201/b18696-9 165

166 unifying physics of accelerators, lasers and plasma

plasma acceleration. The state of this rapidly developing area

makes it hardly suitable for textbooks just yet. We will there­
fore grant only a cursory look at an overview of the presently
identified mechanisms and their scaling rules. We will con­
clude this chapter with a glimpse into the future.

9.1 Bragg peak

The treatment of tumors with the help of X-rays or pro-
ton/ion beams is based on the delivery of energy to malignant
cells of the target volume, intended to prevent and eradicate
the growth of these unwanted cells. We will review the bio­
logical effects of radiation in the next section, and will now
compare the effects of the photons with the effects of charged
particles.

FIGURE 9.1
Photon matter interaction, qualitatively.

Photons penetrating through a medium will lose their en­

ergy due to several factors. An incident photon can interact
with an atom and get absorbed, causing a photoelectron to be
ejected from the atom. The effect is dominant at lower ener­
gies (see Fig. 9.1). An incident photon can also lose part of
its energy via the Compton effect — the inelastic collision of
photons with the electrons of the atoms. Finally, if the energy
of the photon is sufficient, it can create an e+ e− pair. This ef­
fect has a threshold character and manifests itself for photons
with energies greater than about an MeV.
 proton and ion laser plasma acceleration 167

The photons that are used in radiation therapy are usually

produced from electron beams of several MeV, converted on
a solid target to photons via the process of bremsstrahlung. As
interaction between the photon beam and medium results in
a gradual decrease of intensity with the depth of the absorb­
ing medium, the so-called stopping power S(E) is gradually
decreasing too, as shown in Fig. 9.2. The stopping power is
defined as −dE/dx — the energy loss per unit length taken
with the minus sign.
In contrast to photons, charged particles lose their energy
in matter primarily through a Coulomb interaction with the
outer-shell electrons of the absorber’s atoms. Excitation and
ionization of atoms result in a gradual slowdown of the par­
ticle. A slower moving particle will interact with an atom for
longer time, resulting in a larger energy transfer. Therefore,
the charged particles will have an increased energy loss per
unit length at the end of their passage through the medium.
Qualitatively, this is the origin of the Bragg peak — the peak
in energy loss that occurs just before the particles come to a
complete stop.

FIGURE 9.2
Absorption of photons (dotted lines) in comparison with absorp­
tion of protons in media. Overlaying multiple Bragg peaks creates
a near uniform dose distribution in a certain target volume.

The above-mentioned phenomenon is named after William

Bragg, who discovered it in 1903. The Bragg peak is the rea­
son why protons are better for treatment of tumors than X-
rays, in some cases, as protons localize the deposited dose in
168 unifying physics of accelerators, lasers and plasma

the destined target volume (see Fig. 9.2), this minimizes the
impact on healthy tissues — especially in cases where the tar­
get volume is located close to critical organs.
Quantitatively, the Bragg peak can be explained by the
following formula for the mean energy loss of moderately rel­
ativistic heavy particles (Bethe equation):
( )
dE 2Z 1 1 2me c 2 β 2 γ 2 Wmax 2
(− ) ≈ Kz ln −β (9.1)
dx A β2 2 I2
Here, z is the charge number of an incident particle, Z
and A are the charge number and atomic mass of the ab­
sorber, respectively, and β and γ are the relativistic factors
of the incident particle. The parameters under the logarithm
are: I — the mean excitation energy of the atom’s electron,
and Wmax — the maximum energy transfer in a single col-
Penetration ranges for pro- lision. For a particle with mass M, the latter is defined
tons in water: as Wmax = 2me c 2 β 2 γ 2 /(1 + 2γme /M + (me /M)2 ). The coeffi­
250 MeV — 38 cm; cient K is defined as follows: K = 4πNA re2 me c 2 where NA is
200 MeV — 26 cm; Avogadro’s number. The coefficient K approximately equals
150 MeV — 15.6 cm; 0.3 MeV · cm2 /mol.
100 MeV — 7.6 cm; The usefulness of the Bragg peak for treating tumors was
50 MeV — 2.2 cm. first realized by Robert R. Wilson in 1946. Overlaying several
Bragg’s peaks described by Eq. 9.1 creates a uniform dose dis­
tribution in a given volume (as illustrated in Fig. 9.2) — this is
often called a spreadout Bragg peak. Such overlaying requires
an adjustment to the energy and intensity of each individual
proton beam.
The ideally sharp Bragg peak is, in practice, somewhat
spread — firstly due to the statistical character of interaction,
and secondly due to nuclear interactions between the protons
and absorber, which happen with some probability.
The sharpness of the Bragg peak is an enabling feature of
proton therapy, but simultaneously it is a factor that increases
the sensitivity of the method to errors, especially to the errors
in the predicted depth range. In a particular case when the
target volume is located near a critical organ, ideally, one can
completely eliminate irradiation of the critical organ while
filling the entire target volume uniformly. In practice, how­
ever, one cannot obtain a sharp irradiation boundary of the
irradiated volume, due to the necessity to allow for some un­
certainties of the depth range. Possible motion of the critical
organs during irradiation — as well as shrinkage of the tumor
(and possibly corresponding shift of the critical organs) as
the treatment progresses — are also important factors, which
need to be taken into account in proton therapy planning.
The above-mentioned sensitivity to errors places a partic­
ularly strong requirement on the energy of protons in the
cases when plasma acceleration is used. The beam needs to
have a well-defined energy. This can be ensured either via
predictable plasma acceleration or by an appropriate energy-
selection system.
 proton and ion laser plasma acceleration 169

9.2 DNA response to radiation

The underlying mechanism of X-ray and proton cancer ther­
apy is the ability to prevent the replication of malignant
cancer cells. This control is achieved by damaging the DNA
of cancerous cells. Let’s now consider the basic facts about
DNA’s response to radiation.
The effect of ionizing radiation on DNA can pro­
duce many outcomes, from neutral to negative (stimulating
growth of cancer cells), or positive (eliminating cancer cells)
depending on the initial state of the irradiated object, the ra­
diation dose and the use of concomitant agents — pharmaceu­
tical substances used prior or together with irradiation.
In some cases, radiation simply does not cause DNA dam­
age, but passes through the cell without any side effects. With
an increased dose, the radiation starts to physically affect the
DNA. The effects are distinguished as direct and indirect.
Direct radiation effects take place when X-rays create ions
and corresponding electrons (e-) that physically break the nu­
cleotide pairs of the DNA (see Fig. 9.3).

FIGURE 9.3
Radiation effects on DNA.

Nevertheless, in the majority of cases, X-rays act indirectly

when they induce water radiolysis and consequent produc­
tion of OH- hydroxide — highly active free radicals — that
170 unifying physics of accelerators, lasers and plasma

then form hydrogen peroxide and cause single-strand DNA

breaks.
We should also note that the effects described above are
also very similar to the case of irradiation by protons, tak­
ing into account that, in this instance, the electrons and
bremsstrahlung photons are produced when the protons lose
their energy to electrons of the absorbing medium.
Several minor DNA defects — such as one-strand dimer
formations (caused by UV light — see bottom part of Fig. 9.3)
or single-strand breaks could be corrected by the internal cell
mechanism, base excision repair, which involves endonucle­
ases and DNA polymerase. Nevertheless, some errors can also
appear at this step (for example, wrong nucleotide(s) inser­
Double-strand breaks repre- tion could cause a future coding mismatch).
sent the most detrimental Double-strand breaks are more difficult to repair and thus
damage produced in DNA by represent the most detrimental damage produced in DNA by
ionizing radiation. ionizing radiation. When a cluster of complex DNA damage
events occur together, for example from dense ionization at
the terminal end of an energetic electron track, there is the
highest probability of either mis-repair or failure to repair
damage. A loss of entire parts of the chromosomes contain­
ing tumor-suppressor genes — an amplification of oncogenic-
potential regions — could lead to carcinogenesis, the creation
of tumor cells. Unrepairable damage can trigger the apoptosis
mechanism, where defective cells undergo “cell suicide” and
are, thus, eliminated. Mis-repair of DNA with cross-linked
or mutated chromosomes may not be immediately lethal, but
will lead to further DNA damage or cell death when the cell
next attempts division — mitotic catastrophe.
Various examples of the consequences of DNA irradiation
are shown on the right side of Fig. 9.3. They include tem­
porary arrest at different stages of the cell cycle (mitosis (M),
both check-points (G1 and G2), or DNA replication/synthesis
(S)). The irradiation effects also may include inhibition of
transcription and a variety of cell division abnormalities, all
of which could result in a defective cell death.
In molecular biology, a small The positive effects of ionizing radiation are caused by the
molecule drug (or a small same mechanisms, since what could be detrimental for a nor-
compound) is a low molecu- mal cell could also be destructive for a malignant one.
lar weight organic compound In order to further enhance the destructive effect of X-rays
that may help to regulate bi- on malignant cells, some concomitant agents such as small
ological processes. molecule oncology drugs could be involved. The upper limit
for a small molecule’s weight is approximately 900 daltons,
which allows rapid diffusion across cell membranes so that
Dalton (Da) is the standard they can reach intracellular sites of action.
unit defined as one twelfth of
The effects of irradiation and the use of concomitant
the mass of an unbound neu­
tral atom of C12 in its nu­ agents can lead to an increased number of difficult-to-repair
clear and electronic ground inter-strand cross links, as shown in Fig. 9.3. The apoptosis
state. In other words, one Da mechanism, launched by the strong synergetic effect of these
is approximately equal to a two techniques, leads to a more efficient elimination of ma-
mass of a proton or neutron. lignant cells.
 proton and ion laser plasma acceleration 171

9.3 Conventional proton therapy facilities

The design of conventional proton therapy facilities is dic­
tated by several factors, such as the size of the proton source,
the number of treatment beamlines and the size of the gantry.
A generic layout of a proton therapy facility is shown in
Fig. 9.4. Typically, either a cyclotron or a synchrotron is used
as a source for a proton facility. In the event that ions are used
for therapy (e.g., carbon, which gives certain advantages), a
cyclotron can be used to inject the beam into a synchrotron.
The beamline distribution system then directs the accel­
erated beam into the treatment rooms. The beam can have a
fixed location in the room, or can be brought into the room
via a gantry, which allows flexibility in the direction of the
arriving beam.

FIGURE 9.4
Generic proton or heavy-ion therapy facility.

A typical facility would usually have several treatment

rooms with fixed beams and one or more rooms with gantries.
The gantry is a complicated mechanical device that needs to
rotate the entire section of a beamline while maintaining its
precise (usually sub-mm) alignment. The energy of the pro­
tons or ions and the achievable strength of the bending mag­
nets of the gantries would typically make the gantry on the
order of ten meters in size and a hundred of tons in weight.
The use of superconducting magnets in the gantries is pos­
sible and would typically reduce the weight of the gantry by
around a factor of four.
The overall size of the proton/ion treatment facilities is
defined by the size of the accelerator and beamlines, but,
most notably, by the size and the height of the gantries. The
172 unifying physics of accelerators, lasers and plasma

cost of the components and the size of the necessary building

are the prevailing limiting factors — there are presently sev­
eral tens of facilities operating worldwide, while the gener­
ally accepted estimation states that there should be one pro­
ton treatment facility for every 10 million people.

9.3.1 Beam generation and handling at proton facilities

The typical elements of the proton therapy system are illus­
trated in Fig. 9.5.
First, the proton beam coming from the source passes
through a scatterer, which increases the divergence of the
beam. The transverse beam shape is then adjusted by a multi-
leaf steel collimator to match the shape of the target volume.
The leaf of the collimator can be mechanically controlled on
a sufficiently short time scale. Instead of an adjustable multi-
leaf collimator, a fixed-shape brass collimator can be used,
tailored for a particular case.

FIGURE 9.5
The elements of the proton therapy beamline.

Another element of the system is called the bolus, or

range-shifter, and serves to compensate for the depth differ­
ence of different regions of the target volume. The compen­
sator is usually made from polyethylene and reduces the en­
ergy of the beam to fit the required penetration depth at each
point.
FIGURE 9.6 The elements as described above are, in particular, suit­
Pencil beam scanning. able for the beam coming from standard cyclotrons, the en­
 proton and ion laser plasma acceleration 173

ergy of which usually cannot be changed easily. In the case of

synchrotrons, where the beam energy can be finely scanned,
the pencil beam scanning approach can be employed, as shown
in Fig. 9.6. In this case, the beam should be small enough so
that one could “paint” the desired target volume transversely,
and simultaneously control the depth by appropriately ad­
justing the beam energy.

9.3.2 Beam injectors in proton facilities

So far, the most widely used accelerator for proton therapy
facilities has been a cyclotron.
A standard cyclotron is shown in Fig. 9.7. A constant mag­
netic field (often arranged to decrease with the radius to en­
sure transverse focusing) houses the electrodes (dees) where
oscillating voltage is applied.
The standard cyclotron is intended for CW operation, and
this is in fact one of its main advantages. The frequency of
accelerating voltage ω0 and the magnetic field B0 are con­
stant and are connected via the equation for the time of flight
around the orbit
v qB
ω0 = = 0 (9.2)
R mγ
where m is the mass at rest and γ is the relativistic factor that
can change during acceleration.
Since the condition v = ω0 R must be valid up to the mo­
ment of extraction, the final energy of the cyclotron is fixed
by its geometry and cannot be changed. Applications of the
standard cyclotron to proton therapy thus have to rely on the
use of range-shifters to adjust the penetration depth.
In addition to the fixed final energy, standard cyclotrons
also suffer from the relativistic effects. Once protons acceler­
ate to several tens of MeV, the perfect relation between the
revolution time and frequency of the field starts to break.

FIGURE 9.7
Schematic of a cyclotron.
174 unifying physics of accelerators, lasers and plasma

In contrast to a standard cyclotron, a synchrocyclotron can

have variable energies of the accelerated beam and it can also
achieve higher energies — several hundreds of MeV.
In the synchrocyclotron, the relativistic effects are com­
pensated by continuously decreasing the frequency of the ac­
celerating voltage during acceleration so that
ω0
ω= (9.3)
γ(t)

The time-varying accelerating frequency also means that only

a bunch of a certain length can be in sync with the field — the
synchrocyclotron therefore cannot accelerate CW current but
can only produce pulsed beams.

FIGURE 9.8
Schematics of a synchrocyclotron.

The electrode configuration in synchrocyclotrons is also

different — only one dee remains, while the other electrode
has a modified open shape as illustrated in Fig. 9.8. Adjust­
ments to the final energy can be achieved by modifying the
relations between the magnetic field and accelerating fre­
quency, making it possible to accelerate protons to GeV ener-
gies. The pulsed mode of operating synchrocyclotrons, how­
ever, limits their duty factor and thus limits their intensity.
The variability of the final energy of a synchrocyclotron
is an advantage that keeps attracting attention to their po­
tential use as proton therapy machines. A notable recent de-
Compare this with Section velopment1 includes a design of an iron-free superconduct­
10.6.2 and analyze this de­ ing synchrocyclotron, where dual nested solenoids are used
sign from the TRIZ point of to cancel the external fields of the device. The iron-free de-
view. sign makes the synchrocyclotron light and compact, and par-
ticularly suitable for mounting directly on a gantry.
Another type of cyclotron — the isochronous cyclotron —
compensates the relativistic effects by allowing the magnetic
1 A. Radovinsky et al., MIT report PSFC/RR-13-9, 2013.
 proton and ion laser plasma acceleration 175

field to increase with the radius of the orbit as

B = γ(R) B0 (9.4)
To increase the field with a radius in an isochronous cy­
clotron, shims attached to the poles can be used in such a
way that the azimuthal fraction of the shims increase with
the radius, as illustrated in Fig. 9.9.

FIGURE 9.9
Schematics of an isochronous cyclotron.

The field increasing with the radius contradicts the re­

quirements for weak focusing, and therefore strong focusing
is arranged in isochronous cyclotrons by employing a spiral
shape for the pole shims (often called flutter configuration),
so that the regions of the field transition shown in Fig. 9.10
would introduce edge focusing on the beam.
The isochronous cyclotrons, as their acceleration fre­
quency is constant, are suitable for CW operation. Proton
energies of around a GeV can be achieved by compensating 

for the relativistic effects. The final energy of isochronous cy­ 

clotrons is fixed, as its adjustments require modification of

the poles.
CW operation and the ability to reach higher energy made
isochronous cyclotrons very popular for proton therapy ap­
plications. Superconducting isochronous cyclotrons have also
been developed, with the goal to increase the compactness of
proton therapy devices.
Synchrotrons are naturally also a possible choice as an ac­
celerator for use in proton therapy. Their advantage is that FIGURE 9.10
they can provide variable energies; however, there are also a Example of a field profile in
number of disadvantages. First of all, the size: synchrotrons an isochronous cyclotron.
are larger than cyclotrons. Second, synchrotrons are pulsed
machines, while proton therapy benefits from a slow dose
delivery. Therefore, slow extraction methods, such as those
based on excitation of nonlinear resonances, may need to
176 unifying physics of accelerators, lasers and plasma

be used. Overall, synchrotrons are a viable choice and they

are being used in many proton therapy facilities around the
world.

9.4 Plasma acceleration of protons and ions —

motivation
The motivation for developing laser plasma acceleration
techniques for beam therapy can be defined by the following
factors. On the one hand, there are advantages to using pro­
tons instead of X-rays for certain tumors, particularly in pe­
diatric cases, where use of protons significantly reduces the
probability of tumor recurrence or side effects due to lower
dose for non-target tissues. On the other hand, widespread
use of proton therapy facilities is limited by their overall cost.
Proton therapy systems require 250 MeV beams (or above
330–350 MeV when protons are also used for diagnos­
tic imaging). Such systems, especially the beam delivery
gantries, are large and expensive (see Fig. 9.4). The cost and
size arguments are even more pronounced for heavy ion ther­
apy systems based on the use of carbon ions, which create
certain therapeutic advantages.
Meanwhile, laser plasma acceleration has demonstrated
rapid progress, delivering several GeV in energy of quasi mo­
noenergetic electron beams and more than 100 MeV proton
beams. The progress is due to advances in lasers, where a
CPA laser beam of a few hundred TW or around a PW (corre­
sponding, e.g., to 400 J energy in 400 ns or 30 J in 30 fs), when
focused to a 5 μm spot, can create an intensity on the order of
1025 W/m2 — which is suitable for proton plasma accelera­
tion. Such lasers, while still bulky and expensive, are rapidly
improving and will become more efficient and compact in the
future.
Ultimately, the desire to create compact laser plasma ac­
celeration proton therapy systems is one of the main motiva­
tions for developing plasma acceleration of protons.

9.5 Regimes of proton laser plasma acceleration

In this section, we will very briefly describe several different
regimes (mechanisms) of laser-driven proton acceleration.
We will start with a discussion of an already classical
mechanism called sheath acceleration, or TNSA — target nor­
mal sheath acceleration. As we will see, this classical method
provides protons with too large an energy spread and poor
scaling with laser power.
We will then follow up on more recent and promising
mechanisms that rely on radiation pressure acceleration, specif­
 proton and ion laser plasma acceleration 177

ically the mechanisms of hole-boring and light-sail accelera­

tion. The latter has favorable scaling with laser power and
promising perspectives for achieving near-monoenergetic
beams.
We will briefly touch upon other mechanisms that have
been identified, such as shock acceleration and the relativis­
tic transparency regime — also called the break-out afterburner.
Various academic review papers2,3,4 can be consulted to gain
deeper insights into the discussed areas.
Once more, we would like to stress that the men­
tioned mechanisms of plasma acceleration represent a grad­
ual improvement of our understanding of this very compli­
cated phenomenon. Different mechanisms compete, and of­
ten more than one mechanism is active in a particular case. As
we strive to sharpen our understanding of these mechanisms,
further significant progress in this area will be inevitable.

9.5.1 Sheath acceleration regime

Sheath acceleration (TNSA — target normal sheath accelera­
tion regime) is illustrated conceptually in Fig. 9.11. Here, the
laser pulse is focused on a thin metal foil that creates plasma.
The plasma electrons quickly become relativistically hot and
leave the foil, creating a sheath of charge, which then pulls
out the ions and protons from the plasma.

FIGURE 9.11
Sheath laser acceleration of protons.
2 A. Macchi et al., Rev. Mod. Physics, 85, 751 (2013).
3 H. Daido et al., Rep. Prog. Phys. 75, 056401 (2012).
4 M. Borghesi et al., Fusion Science and Technology, 49, 412 (2006).
178 unifying physics of accelerators, lasers and plasma

The surface of the foil is typically contaminated (which is,

in this case, useful) by a thin layer of hydrogen that becomes
the source of protons that are most readily accelerated. Ac­
celeration of protons and ions happens on both sides of the
foil. Still, the side opposite the laser irradiation side exhibits
sharper boundaries of the electron sheath, providing higher
energies and better beam quality of the accelerated protons.
Experimental investigations of the TNSA mechanism us­
ing lasers with intensities above 1019 W/cm2 demonstrated5
that multi-Mev ion acceleration from the rear surface of thin
foils is possible.
The scaling rules for the TNSA mechanism can be derived
by taking into account that a Debye sheath will form from hot
electrons on the rear of the foil, and the potential difference
U through this sheath will be of the order of the electron tem­
perature Te multiplied by the Boltzman constant: U ≈ kB Th .
Considering that the temperature of electrons √ is related to
laser intensity I and wavelength λ as Te ∝ I λ2 , one can ob­
tain a scaling for the maximal energy of accelerating protons
as √
Wmax ≈ kB Te ∝ I λ2 (9.5)
In the above equation, we have assumed in the first ap­
proximation that the maximum energy of the protons is equal
to the value of the potential difference U. In fact, energies sev­
eral times larger than that are expected, taking into account
that the sheath is expanding and protons are “surfing” on the
expanding potential.
In some of the first experiments,6 proton energies close
to 20 MeV were obtained. The spectrum of protons was,
however, very broad and there were not many protons at
the large energy side of the spectrum. Qualitative behavior
of the TNSA proton spectrum is shown in Fig. 9.12. Even
though later experiments demonstrated that the proton en­
ergies close to 100 MeV were possible (see review7 ), these
disadvantageous qualities of the spectrum remained.


Some of the characteristic properties of TNSA-produced
beams include a large divergence (of several degrees) com­
bined with micron-scale beam size. Such a beam may for­
  
 



mally have a low emittance; however, it will quickly filament
and the emittance will increase if the beam is not captured in
 an appropriate focusing system. The TNSA mechanism can
produce 1011 to 1013 protons per shot, but not many of those


 
protons will be at the high-energy edge of the spectrum.

 

Scaling rules predicted by TNSA models allow us to
make projections (with large uncertainties) towards reaching
FIGURE 9.12 a proton energy of around 200 MeV. These projections have
TNSA spectum, qualitative
behavior. 5 Maksimchuk et al., PRL, 84, 4108 (2000); Snavely et al., PRL, 85, 2945
(2000).
6 E.L. Clark et al., PRL, 84, p.670. (2000)
7 M. Borghesi et al., Plasma Phys. Control. Fus., 50, 124040 (2008).
 proton and ion laser plasma acceleration 179

shown that the required laser intensities are on the order of

1021 W/cm2 .
Various ways to improve TNSA have been considered,
starting from the brute force method to increase the laser in­
tensity, to more subtle methods involving the enhancement of
the laser energy transfer to electrons and increase of the elec­
tron density. Reduction of foil thickness and reduced mass
targets, enhanced coupling by a conically shaped target and
use of nano-particle structured targets, among others, have
all been used with some degree of success.
Despite the mentioned improvements, the TNSA mecha­
nism still has a major disadvantage due to the shape of the
spectrum of accelerated particles — the number of particles
at the high end of the spectrum remains very low.

9.5.2 Hole-boring radiation pressure acceleration regime

The radiation pressure acceleration mechanism is based on
the effects equivalent to radiation pressure, which light ex­
hibits on a mirror when it reflects from its surface, as illus­
trated in Fig. 9.13. The radiation pressure for perfect reflec­
tion is proportional to the laser intensity IL as

2 IL
PL = (9.6)
c
Similar pressure can be applied to a thin foil, upon whose
surface plasma quickly forms. The radiation pressure effect FIGURE 9.13
is transmitted into the plasma by electrons via the pondero- Radiation pressure accelera­
motive force. Displaced electrons produce space charge that tion concept.
creates a steady pressure, which in its turn transfers the effect
to the ions.
Two versions of radiation pressure mechanisms have re­
ceived distinct names — the hole-boring and the light-sail
mechanism.
In the case of hole-boring, the space charge due to elec­
trons acts on ions that are pushed into the overdense plasma,
initially compressing the foil and then pushing a region of
the foil forward, as illustrated conceptually in Fig. 9.14.
The derivation of the approximate scaling rules for the

FIGURE 9.14
Hole-boring radiation pressure laser acceleration of protons.
180 unifying physics of accelerators, lasers and plasma

hole-boring mechanism can be obtained by including the ra­

diation light pressure into the fluid equation of motion, and
transferring it to the reference frame where the shock is sta­
tionary (i.e., where the time derivatives are zero). This will
yield ρu 2 = PL and therefore the velocity of ions can be esti­
mated as
IL
u≈ (9.7)
ρc
Consider an example of a laser with IL = 1021 W/cm2
shining on an aluminum target with ρ = 2.7 103 kg/m3 . The
resulting velocity of ions in this case is u ∼ 0.01c or about
3 mm/ps. The energy of the protons associated with this
shock is 60 keV. By “bouncing” the stationary protons of the
shock front, the protons can gain twice the velocity or four
times the energy. Reducing the density of the foil could lead
to a further increase of the energy of accelerated ions — the
latter can be achieved by using gas targets.
The most attractive feature of the hole-boring radiation
pressure mechanism is that the resulting proton beam has
been demonstrated8 to have a nearly monochromatic peak at
the maximum energy.

9.5.3 Light-sail radiation pressure acceleration regime

In the light-sail9 regime, the radiation pressure mechanism is
taken to the extreme when the foil is so light that it starts to
accelerate immediately as a whole, as shown schematically in
Fig. 9.15.

FIGURE 9.15
Light-sail radiation pressure laser acceleration of protons.
8 C.A. Palmer et al., Phys. Rev. Lett., 106, 014801 (2011).

9 T. Esirkepov et al. Phys. Rev. Lett., 92, 175003 (2004).

 proton and ion laser plasma acceleration 181

In a simple model of light-sail acceleration, one can as­

sume that an area A is illuminated with an intensity IL and if
the light reflection is perfect, the force acting on this region
of the foil is given by
IL dv v
F = 2A =m = Adρ i (9.8)
ρc dt t
where m is the mass of this segment of the foil, m = A d ρ,
with d being the foil thickness and ρ the density. The velocity
of ions will therefore grow proportionally to the laser inten­
sity and laser pulse duration:
2IL τ
vi ≈ (9.9)
ηc
where η = ρ d — the areal density. As we can see, the energy
scaling in the light-sail regime is more favorable than in the
hole-boring case, as in the non-relativistic case of light-sail
acceleration the velocity of ions is proportional to IL and the
energy is thus proportional to IL2 .
The light-sail radiation pressure mechanism cannot be
completely decoupled from the competing TNSA mecha­
nism. The electron heating — which was ignored in the sim­
plified picture above — may cause foil deterioration. A pos­
sible improvement for the latter involves use of a circularly
polarized laser, which reduces the effects of TNSA and foil
heating.
The scaling of a light-sail mechanism for 10 PW pulses
with laser intensity of 1022 W/cm2 predicts10 that GeV pro­
ton beams with good near-monochromatic spectral character­
istics can be produced.
A Rayleigh–Taylor instability of the shape of the foil
can develop during acceleration, resulting in deterioration
of the resulted spectrum. However, this instability has been
shown11 to be stabilized by simultaneous acceleration of
multi-species ions.
Overall, the proton energy scaling as (IL τ/η)2 for the
light-sail radiation pressure regime is the most favorable and
promising.

9.5.4 Emerging mechanisms of acceleration

Various other mechanisms have been suggested that describe
the behavior of laser plasma acceleration of ions in certain
parameter ranges.
In particular, the break-out afterburner regime has been de­
scribed12 as the mechanism based on the appearance of the
10 B. Qiao et al., Phys. Rev. Lett., 102, 145002 (2009).
11 B. Qiao et al., PRL, 105, 1555002 (2010), T. Pu Yu et al., PRL, 105, 065002
(2010).
12 Yin et al., Phys. Rev. Lett. 107, 045003 (2011).
182 unifying physics of accelerators, lasers and plasma

relativistic transparency of an initially solid target. As a re­

sult, this leads to the enhancement of ion acceleration via the
TNSA mechanism.
Shock acceleration is a mechanism13 based on the ap­
pearance of a high Mach number electrostatic shock in the
overdense plasma created by the sharp front of the laser
pulse. The propagating electrostatic shock will then reflect
the plasma ions to the doubled velocity of the shock, resulting
in an appearance of monochromatic proton peaks observed in
experiments using overdense gas jet targets.
As we can see, while the number of forthcoming theoret­
ical and experimental tasks and questions is still very high
in this area, waiting for an inquisitive mind, there are mech­
anisms of proton plasma accelerations which, given the rate
of the laser technology progress, can already yield the prac­
tically useful techniques for creating beam therapy facilities
based on proton plasma acceleration.

9.6 Glimpse into the future

Creating a more compact and affordable design of a proton
therapy facility is a formidable task, one that is attracting the
attention of many research teams worldwide. The plasma ac­
celeration community, together with laser and conventional
acceleration communities, are joining forces to solve this task.
Techniques of beam control and energy selection, along
with methods of capturing divergent and chromatic beams,
all developed alongside conventional accelerators and com­
bined with novel opportunities enabled by modern and fu­
ture lasers, should help us to find a way to create a viable
design and a prototype system that can be used efficiently in
practice.
In Chapter 10, we will briefly review various methods of
beam control and manipulation. Some of these methods can
be particularly applicable to the design of a plasma proton
acceleration facility.

13 L. Silva et al., PRL 92, 015002 (2004); D. Haberberger et al., Nature Phys.,
8, 95 (2012).
 proton and ion laser plasma acceleration 183

EXERCISES
9.1 Chapter materials review.

Discuss the advantages and challenges of particle beam ther­

apy in comparison to X-ray therapy.

9.2 Chapter materials review.

Discuss the key requirements for the laser and the target that

may result in 200 MeV mono-energetic beams of protons in

laser plasma acceleration.

9.3 Mini-project.

Discuss and develop a plan to create a 250 MeV proton source

based on plasma acceleration, aiming for it to be applied in

the medical field. Select approximate laser parameters and

target parameters, and discuss their requirements. Discuss

and select a method for energy monochromatization or en­

ergy collimation/selection. Describe why you selected these

particular values of certain parameters (for target or laser,

collimation or monochromatization system, etc.).

9.4 Analyze inventions or discoveries using TRIZ and AS-TRIZ.

Analyze and describe scientific or technical inventions de­

scribed in this chapter (e.g., isochronous cyclotron) in terms

of the TRIZ and AS-TRIZ approaches, identifying a contra­

diction and an inventive principle that were used (could have

been used) for these inventions.

9.5 Developing AS-TRIZ parameters and inventive principles.

Based on what you already know about accelerator science,

discuss and suggest the possible additional parameters for

the AS-TRIZ contradiction matrix, as well as the possible ad­

ditional AS-TRIZ inventive principles.

 10
Advanced Beam
Manipulation, Cooling,
Damping and Stability
10.1 Short and We have arrived at the second-to-last chapter. The scope of
narrow-band 185 this chapter is a mixture of beam and pulse manipulation
10.2 Laser–beam topics, selected based on the degree of the synergy they ex­
interaction 196 hibit in four areas that we previously discussed. The first
10.3 Stability of beams three areas are accelerators, lasers and plasma (see Chapter
200 4), and the fourth area is the methodology of inventiveness.
10.4 Beam or pulse We will start this chapter with the methods and tech­
addition 209 niques for creating short bunches of charged particles and
10.5 Cooling and phase also short laser pulses. We will follow with a discussion of
transfer 214 various topics of beam stability and ways to enhance the in­
10.6 Local correction tensity of both the particle beams and the laser pulses. We
217 will conclude with an overview of cooling methods, as well
as local disturbance correction methods, aiming to connect
these and other techniques to Accelerating Science TRIZ.

10.1 Short and narrow-band

The topic of bunch and pulse compression is an area that
shows strong synergy between accelerators and lasers. This
is a broad area that includes techniques of bunch or pulse
compression as well as techniques for creating short bursts
of radiation. We will also pay attention to some effects that
may prevent the formation of ultra-short bunches — notably
coherent synchrotron radiation. Connected to the topic of short
pulse or bunch creation is the theme of narrow-band radia­
tion, where techniques developed in optics and lasers merge
with the accelerator-based techniques.

10.1.1 Bunch compression

Short electron bunches are typically produced by RF photo
guns where the cathode is illuminated by a short laser pulse.
Such photo injectors produce relatively long pulses (a few ps),
while for many applications, such bunches would be too long
and femtosecond bunches would be required instead.
Some compression of the beam may be performed already
in the photo injector, via the technique of velocity bunching. In
this case, by properly adjusting the moment when the laser
illuminates the cathode with the RF field in the gun, one
can create correlation between the energy of electrons and
the longitudinal position within the bunch. The difference of

185
DOI: 10.1201/b18696-10
186 unifying physics of accelerators, lasers and plasma

energy for beams that are still weakly relativistic can create
a difference in longitudinal velocities. If the faster particles
are arranged to be at the tail of the bunch (as illustrated in
Fig. 10.1), they can catch up, resulting in a shortening of the
bunch.
Because velocity bunching is based on the velocity’s de­
pendence on energy, it can work only for weakly relativis­
tic beams and this, in fact, is the method’s main shortcom-
D
ing. Weakly relativistic electron beams can, in particular, suf­
fer strongly from space charge effects, limiting the degree of
compression.
Therefore, achieving ultra-short electron bunches is usu-
E ally done at higher energies, when the beam is relativistic
and space charge effects are less severe. Shortening the bunch
FIGURE 10.1 length in this case is usually achieved with a magnetic com-
Velocity bunching. Initial pression system. When discussing bunch compressors, we
beam (a) and compressed will build on the foundation established in Chapter 4, further
beam (b). expanding it towards an analytical description of the process.
A typical arrangement that can compress the bunch is a
beamline made of four bending magnets of opposite polar­
ity arranged as shown in Fig. 10.2 — called a chicane. In this
chicane, the time of flight (or equivalently the path length) is
different for different energies.
In order to exploit the dependence of the time of flight (or
FIGURE 10.2
path length) on the particle’s energy, we need to introduce an
Four-magnet chicane.
energy–time correlation within the bunch. This correlation
can be created using the electric field of an RF cavity, properly
phased with the beam.

FIGURE 10.3
Energy–time correlation and bunch compression.

The RF cavity will create an energy chirp along the bunch

— the necessary condition for bunch compression to work;
see Fig10.3. The chirp is phased in such a way that the parti­
cles in the tail have a higher energy and will therefore travel
 advanced beam manipulation, cooling, damping and stability 187

on a straighter path through the chicane, catching up with the

synchronous particle, and resulting in bunch compression.
The bunch compression process can also be described an­
alytically. Let’s use the linear transfer matrices to evaluate the
evolution of the longitudinal position and relative energy off­
set (z, δ) as the particle propagates through the RF cavity and
then the bunch compressor.
The first step is to evaluate how the coordinates (z, δ)
change in the RF cavity. Passing through the RF cavity (as­
suming the cavity is thin), the longitudinal coordinate does
not change, while the energy change depends on the initial
position z0 as follows:

z1 = z0
(10.1)
δ1 = δ0 + eVERF cos π
2 − kRF z0
0

The above equations are equivalent to the following matrix

transformation expressed in linear approximation in (z, δ) as
follows (refer to Eq. 2.58 and Eq. 2.59 for definitions of the
indexes of the coordinate vector and of the transfer matrix):
( ) ( ) ( )
z1 1 0 z0
≈ · (10.2)
δ1 R65 1 δ0

where
eVRF
R65 = sin (ϕRF ) kRF (10.3)
E0
The next step is to take into account the bunch compres­
sor itself. In the chicane, the particle coordinate will change
according to the following general expression (the higher-
order terms are defined as in Eq. 2.60):

z2 = z1 + R56 δ1 + T566 δ12 + U5666 δ13 . . .

(10.4)
δ2 = δ1

which can be linearly approximated as

( ) ( ) ( )
z2 1 R56 z1
≈ · (10.5)
δ2 0 1 δ1

The full transformation is given by multiplying the matrices

of each element, which can be computed to be given by the
following:
( ) ( ) ( )
z2 z0 1 + R65 R56 R56
≈ M· M= (10.6)
δ2 δ0 R65 1

Dependence of the final bunch length on the initial beam

length is therefore given by the following equation

σz2 = |1 + R65 R56 |σz0 (10.7)

We can see that, in order to achieve maximal compression, we

188 unifying physics of accelerators, lasers and plasma

should adjust the value of R65 (energy chirp induced by the

RF cavity) in such a way that

R65 R56 ≈ −1 (10.8)

in which case the compression factor C will be the largest and

the final bunch length
σz0
σz2 = where C = 1/|1 + R65 R56 | (10.9)
C
will be minimal.
Let us now discuss the minimal achievable bunch length.
While in the linear approximation, the condition R65 R56 = −1
suggests that C is infinite and the resulting final bunch length
is zero, this cannot be achieved in practice.
First of all, taking second and higher-order terms into ac­
count will give us non-zero final bunch lengths for any initial
parameters.
Secondly, there is a limitation caused by the longitudi­
nal emittance of the beam. The linear transformation of the
(z, δ) coordinates preserves the longitudinal beam emittance
according to the Liouville’s theorem (it’s often said that the
corresponding transformation is symplectic). The longitudi­
nal emittance is given by the following:
)
ε = σz2 σδ2 − σzδ
2
(10.10)

which tells us that the minimum reachable bunch length is

limited to the product of the beam energy spread σδ times
the matrix element R56 .
Furthermore, additional limitations to the achievable
compression come from the effects associated with the beam’s
high peak current that we have neglected in the linear ap­
proximation denoted above. These effects include longitudi­
nal space charge, wakefields and coherent synchrotron radia­
tion (CSR). When taken into account, these effects can pro­
duce serious degradation to the quality of the beams and
limit the achievable minimum bunch length.

10.1.2 CSR — coherent synchrotron radiation

We will now evaluate the coherent synchrotron radiation effect
in a back-of-the-envelope fashion, following the example of
Chapter 3.
The characteristic frequency of SR

3 c γ3
ωc =
2 R
defines the higher edge of the SR spectrum while radiation is
emitted at any frequencies ω below ωc .
Let us assume that we have a bunch with N electrons
 advanced beam manipulation, cooling, damping and stability 189

and its bunch length is σ. For certain frequencies ω of syn­

chrotron radiation, its wavelength ≈ c/ω can be longer than
the bunch length σ; the beam at these frequencies will radiate
coherently, as illustrated in Fig. 10.4.

FIGURE 10.4
Incoherent radiation (left) and coherent radiation (right).

The two regimes can thus be defined as c/ω < σ — inco­

herent radiation — and c/ω > σ — coherent radiation.
For frequencies where c/ω is longer than σ, the beam ra­
diates coherently, increasing radiated power proportionally
to the square of the number of particles involved, i.e., as N 2 .
As the typical bunch population N is ∼ 109 to ∼ 1010 , this
effect increases the radiated power tremendously.
Assuming that the spectrum of incoherent SR is given by
I(ω), the CSR spectrum can be approximated as
r ( )1
ωσ 2
Ib (ω) = I (ω) N 1 + N exp − (10.11)
c
A typical spectrum described by the above equation is shown
in Fig. 10.5.


3RZHUDX

 a1

&RKHUHQW




a1 ,QFRKHUHQW



  
    
     

)UHTXHQF\7+]
FIGURE 10.5
Qualitative comparison of the spectrum of coherent synchrotron
radiation in comparison with the spectrum of incoherent SR.

10.1.3 CSR effects on the beam longitudinal phase space

We will now make a back-of-the-envelope estimate1 of the FIGURE 10.6
force that CSR exerts on the beam. We will use an approxi- Two-particle model of the
1 beam and its field.
Following derivations from Ya. Derbenev et al., Tesla-FEL Report 1995–
05.
190 unifying physics of accelerators, lasers and plasma

mation of the two-particle model. In this instance, the beam is

represented just by two particles — at the head and in the tail
of the bunch, as illustrated in Fig. 10.6, where the particles —
as well as its field — are shown qualitatively. As we will see,
the CSR effects are essentially caused by the possibility for
the tail field to overtake the head particle while the beam is
moving on a curved trajectory.
Let us thus define the bunch as a two-particle system with
the length between the head and the tail s equal to rms bunch
length of the initial beam σ. The EM fields of the relativistic
bunch are distributed primarily in transverse directions, as
indicated in Fig. 10.6.
Assuming that the beam is moving along the curved tra­
jectory shown in Fig. 10.7, we can determine the moment at
which the field of the tail, radiated at point A, and moving
on the straight line, will catch up with the head of the beam.
The condition for the catch-up (overtake) to happen in point
B can be expressed as follows:

Arc (AB) − s = |AB| (10.12)

which can be rewritten as

R θ − s = 2R sin (θ/2) , which gives s ≈ R θ 3 /24 (10.13)

Thus the overtaking distance L0 can be estimated as

1/3
L0 = |AB| = R θ ≈ 2 3sR2 (10.14)

Knowing the overtaking distance, we can determine the char­

acteristic transverse distance r, which is equal to the distance
between the head particle at the point of overtake and the
axis of propagation of the SR fields that travel along the ini­
tial trajectory (Fig. 10.7).

FIGURE 10.7
Illustration of the tail field overtaking the head of the bunch in
the mechanism of coherent synchrotron radiation.
 advanced beam manipulation, cooling, damping and stability 191

The transverse characteristic distance can be estimated as fol­

lows:
1/3
r = L0 θ/2 = 2 9s 2 R (10.15)
We can now estimate the field of the bunch tail (radiated at
point A), acting on the head at point B.
Assume that the beam is uniform and has the linear
change density eN λ where λ = 1/eb and eb is the length of
the bunch.
The values of the transverse field at the characteristic dis­
tance for the linear charge density beam are as follows:
2Ne λ
E⊥ = H⊥ ≈ (10.16)
r
In order to find the longitudinal field acting on the head,
we need to multiply the transverse field by the angle: E⊥ · θ.
The longitudinal force acting on the head can thus be esti­
mated as
2N e 2 λθ 2N e2 λ
F|| = eE⊥ · θ = = (10.17)
r (3sR2 )1/3

Let us now assume that s = eb = 31/2 σ (the latter assumes

that the bunch distribution is Gaussian). The estimate for the
longitudinal force thus becomes

2N e2 2N re mc2
F|| ≈ 2/3 4/3
= (10.18)
3R σ 3R2/3 σ 4/3
which is a rather accurate back-of-the-envelope estimate.
Accurate derivations of the CSR effects for a realistic
Gaussian bunch can show that the longitudinal force acting



) V 

WDLO KHDG

        




        

FIGURE 10.8
Shape function F0 (top plot) of coherent synchrotron radiation
for a bunch with Gaussian density profile (bottom plot).
192 unifying physics of accelerators, lasers and plasma

on the beam is very close to the estimate Eq. 10.18 and con­
tains an extra term: the shape function F0 , which depends on
the position of the particle within the bunch.

2N re mc2
F|| ≈ F0 (10.19)
3R2/3 σ 4/3
The shape of the CSR shape function is shown in Fig. 10.8.
As we can see, its amplitude is close to one, confirming the
back-of-the-envelope estimate. The shape function changes
its sign. While the head of the bunch slightly accelerates due
to the CSR effect, the major part of the bunch decelerates.
The effects of CSR are particularly important in bunch
compressors and can prevent achieving ultra-short bunches.
CSR can cause bunch instability and microbunching, and
can therefore deteriorate the longitudinal phase space of the
beam.
To conclude, one should note that the mechanism of CSR’s
creation — depicted in Fig. 10.7 — suggests that, in certain
parameters, there may be a “cure” for CSR, because the vac­
uum chamber where the beam and fields propagate can par­
tially shield the fields and reduce the CSR effects.

10.1.4 Short laser pulse and Q-switching techniques

The methods of generating short laser pulses often rely on
so-called Q-switching techniques. In this case, the laser gain
medium is placed in the optical cavity (the quality factor
Q of this cavity can be controlled). In the initial moment,
the Q factor is set to a low value, and pumping of the laser
medium will then build up a large population inversion in
the medium.
Once a sufficient inversion population is achieved, the Q
factor of the laser cavity is suddenly increased, which results
in quick buildup of the light in the cavity and an avalanche of
stimulated emission. A giant and short pulse is thus emitted
from the laser cavity. This technique is illustrated in Fig. 10.9.

FIGURE 10.9
Q-switching technique. In step one (a) the pump builds up large
inversion in the gain media. In step two (b) the laser cavity
switches from low to high-Q.
 advanced beam manipulation, cooling, damping and stability 193

10.1.5 Q-switching methods

The methods used to switch the Q factor in the laser cavity
can be divided into two categories: passive and active. Active
methods may involve mechanical effects, e.g., rotation of the
laser cavity mirrors or applying an EM field to a substance
inserted into the cavity (which changes its optical properties
under the influence of the field). Passive methods involve spe­
cial mirrors that can change the Q factor reflectivity when the
power level of radiation reaches a certain level.

FIGURE 10.10
Examples of active Q-switching methods. Rotating mirror (A),
Electro-optic (B) and Acousto-optic (C).

Some of the active methods are presented in Fig. 10.10.

The first active Q-switching method is mechanical and in­
volves the use of a rotating mirror. In this case, lasing will
happen only at the moment when the mirror is parallel to
the other mirror. A rotating cylinder — to which many mir­
rors are connected — is sometimes used because it generates
many pulses for each rotation of the cylinder.
The speed of the mechanical active Q-switching meth­
ods is obviously limited by the mechanical strength of the
movable objects, induced vibration and other factors. Active
methods that minimize mechanical motion can avoid these
limitations. The second active method is based on the use of
the acousto-optic effect — the Bragg diffraction of light from
the planes of a varied refractive index created in a crystal by
the applied sound wave. When the acoustic wave transducer
is switched on, a certain fraction of light diffracts away from
the main light path in the laser cavity, which can then be used
to change the cavity’s Q value. The third active method is
based on the electro-optic effect — the dependence of optical
properties on objects such as absorption or refraction (called
Pockels effect) on the applied electric field.
Similar to the active methods, various passive methods of
Q-switching exist. Two examples shown in Fig. 10.11 include
a saturable absorber and a SESAM. The saturable absorber
uses the fact that certain absorbers become transparent when
they reach saturation and cannot absorb radiation any fur­
ther. At the moment when they become transparent, increas­
ing the Q factor of the cavity can lead to lasing.
194 unifying physics of accelerators, lasers and plasma

FIGURE 10.11
Passive Q-switching — saturable absorber (A) and SESAM (B).

A saturable absorber can be built based on doped YAG

crystals or GaAs, on media with immersed quantum dots, or
semiconductors (called semiconductor saturable absorber mir­
rors — SESAM). Another possible design of a passive Q-
switch is based on a thin-film absorber where a quarter-wave
plate is combined with a thin-film polarizer (TFP), serving as
a hold-off polarizer in the Q-switched laser cell.

10.1.6 Regenerative amplifiers

The technology employed in Q-switching techniques is used
in regenerative laser amplifiers (often called “regens”), which
are designed to generate short, powerful laser pulses. In a “re­
gen,” a laser amplifier is placed inside of a Q-switched optical
cavity (see Fig. 10.12) and it operates as follows.

FIGURE 10.12
Schematics of a regenerative amplifier.
 advanced beam manipulation, cooling, damping and stability 195

First, the gain medium (amplifier crystal in Fig. 10.12) is

pumped by the laser diode. A short initial pulse from the
“master oscillator” is then injected into the input of the sys­
tem through the thin-film polarizer and the Faraday rotator,
both of which are switched on for this short moment in order
to let the input pulse pass.
The initial pulse then undergoes many round trips
through the system, its power level getting amplified each
time it passes the gain medium until it reaches a high level.
Finally, another switch is powered (Pockels cell in Fig. 10.12)
and the amplified pulse is released from the system.

10.1.7 Mode locking

Mode locking is the technique for achieving short laser pulses.
In this method a rapid modulator is installed in the laser cav­
ity, one that can open for short moments exactly in sync with
the pulse’s round-trip time around the cavity; see Fig. 10.13.
Therefore, all the generated and amplified photons will be
clustered only within those short moments when the modu­
lator is open.
The amount of time of an average round trip in a laser res­
onator can be on the order of a nanosecond, while the mod­
ulator can open for periods of tens of femtoseconds. Corre­
spondingly, the peak power in the mode-locked lasers can be
of many orders of magnitude higher than the average power.

FIGURE 10.13
Mode-locked laser (left) and the laser output (right) in the normal
(a) and mode-locked (b) regimes.

The name of the technique — mode locking — is a term

derived from the analysis of this problem in the frequency
domain. The laser cavity has a certain bandwidth that can
support many longitudinal modes — all of which can coex­
ist. In a normal laser, the phases between different modes are
random. The mode locking created by the modulator would,
in fact, create a certain fixed relation between the phases of
all modes. In this case, there will be only one point in the
laser cavity at each moment of time where the modes’ electric
196 unifying physics of accelerators, lasers and plasma

fields will add together constructively — which is equivalent

to the case of a single short pulse travelling inside the laser
cavity.

10.1.8 Self-seeded FEL

Various techniques used in optics and particularly in laser
systems are ideologically similar to techniques employed for
creating short pulses of FEL radiation. The self-seeding tech­
nique described in this section has certain similarities with
the mode-locking method described in the previous section.
Recall that, in the SASE regime, the FEL lasing starts from
noise, and also remember that the length of the slice that con­
tributes to radiation is much shorter than the length of the
electron bunch.
This means that each individual slice can generate radia­
tion at slightly different wavelengths near the resonant wave­
length, and, moreover, the amplitude of the radiation coming
from each slice can be slightly different, as illustrated in the
P(λ) spectrum on the left side of Fig. 10.14.

%HQGLQJPDJQHWV
6$6()(/ 6HOIVHHGHG)(/
HEHDP HEHDP

3 3
&U\VWDOPRQRFKURPDWRU

FIGURE 10.14
Self-seeded FEL.

The output structure of the FEL can be considerably im­

proved using the self-seeding approach. In this event, the FEL
is split into two parts. Before entering the second part, radi­
ation generated in the first part is passed through a crystal
monochromator. In order to have the timing maintained, the
beam is simultaneously passed through a four-bend chicane.
The second part of the FEL is thus seeded with narrow
spectrum radiation, which continues to be amplified along
the way in the undulator; the resulting output spectrum of
FEL then contains a narrow peak, as shown in Fig. 10.14 on
the right side.

10.2 Laser–beam interaction

Interaction of laser light with electron beams in wigglers of­
fers a wide range of techniques that can be used to manip­
 advanced beam manipulation, cooling, damping and stability 197

ulate the electron beam or to improve the properties of FEL

radiation.
We recall from Chapter 8 that the light of a resonant har­
monic will be exponentially amplified in an undulator, tak­
ing the energy from the electron beam. The inverse can also
be true — sending laser light at the resonant harmonic into
the undulator will affect the electron beam (inverse FEL), cre­
ating variations of its energy.
The inverse FEL principle is the basis for many techniques
used to manipulate electron beams, create short pulses of
radiation, generate higher harmonics, create two-color FEL
pulses and many others uses.

10.2.1 Beam laser heating

Short laser pulses can be used to improve the properties of
short pulses of FEL radiation. The first example we will con­
sider is the laser heater.
The generation of femtosecond-short X-ray pulses re­
quires fs-short electron bunches. However, as we saw earlier
in this chapter, the creation of short bunches in bunch com­
pressors can be complicated due to CSR effects that can cause
instability and microbunching.
Instabilities can often be suppressed (thanks to decoher­
ence — see discussion of Landau damping in Section 10.3.4
later in this chapter) if the beam has sufficient spread of its
relevant phase space coordinates (the energy spread, in the
case of CSR instability).
The issue that has been observed in FELs is that the beam
energy spread coming from a photocathode gun is extremely
small, and insufficient for suppressing CSR instability.
A method has been developed at SLAC to introduce addi­
tional uncorrelated energy spreads into the beam: laser heat­
ing (see Fig. 10.15).
In this case, the laser light of the resonant wavelength co­
propagates with an electron beam through a wiggler. This
system acts as an inverse FEL and introduces an additional

FIGURE 10.15
Laser heater.
198 unifying physics of accelerators, lasers and plasma

energy spread to the beam, sufficient to cure the CSR insta­

bility.

10.2.2 Beam laser slicing

Beam laser slicing is a technique that selects a femtosecond
short portion of radiation from a much longer initial electron
pulse.
In this method,2 a very short laser pulse overlaps with
the center of a longer bunch in the undulator or wiggler; see
Fig. 10.16. The laser wavelength λL matches the undulator
resonance condition
( )
λW K2
λL = 1+
2γ 2 2

and the interaction of the light with the beam in the undula­
tor will therefore produce modulation of energy in the short
beam slice overlapped with the laser pulse.

FIGURE 10.16
Beam laser slicing.

The electron beam coming out from the undulator will

therefore have a short fs region with an increased energy
spread. The energy spread of this short region can be con­
verted to spatial variation using a dispersive beamline sec­
tion. The resulting beam can then pass through a bend to
generate synchrotron radiation and, following that, the SR
corresponding to the short portion of the electron beam can
be transversely separated from the rest of the pulse using col­
limators.
The method described above can be especially suitable for
ring-based SR sources, where the natural length of the elec­
tron bunch is a picosecond long. The application of the laser
slicing technique can therefore help in generating femtosec­
2 A. A. Zholents and M. S. Zolotorev, Phys. Rev. Lett. 76, 912 (1996).
 advanced beam manipulation, cooling, damping and stability 199

ond synchrotron radiation pulses, even for third-generation

sources.

10.2.3 Beam laser harmonic generation

Harmonic generation is the technique that can help to produce
X-ray photons of much higher energies. We recall from Sec­
tion 8.3.3 that odd higher harmonics can also resonate and
can thus be generated in an FEL. An FEL would normally
generate primarily the first harmonic. In order to primar­
ily generate a higher-order harmonic, external seeding is re­
quired. There are, however, no conventional lasers of appro­
priately short wavelengths that can be used for such seeding.

8QGXODWRU 8QGXODWRU
,QLWLDOHEHDP /DVHU

/DVHU
%HQGLQJPDJQHWV
%HQGLQJPDJQHWV
( ( ( (
(

    

              

] ] ] ] ]

FIGURE 10.17
Echo-enabled harmonic generation scheme — EEHG.
(
A technique invented3 by G. Stupakov is currently solv­
ing the problem: echo-enabled harmonic generation (EEHG).
The term echo came from plasma physics and refers to the 
phenomenon of a spontaneous appearance of a wave with
wavenumber k3 in plasma, at a certain time after the initial
excitation waves with wavenumbers k1 and k2 would decay   
]
via the collisionless Landau damping mechanism. 'HQVLW\DX
It is useful, however, to discuss the mechanism of EEHG

without referring to the physics of plasma echo. The EEHG
technique is based on laser-beam interaction in wigglers and 
distortion of the z − E phase space in four-bend magnetic chi­
canes; see Fig. 10.17. Here, the first wiggler and laser with 
  
wavenumber k1 create sine-like modulation of the z −E phase ]
space.
The first chicane applies z = z + R56 · ΔE/E0 transforma­ FIGURE 10.18
tion that deforms the sine-like phase space into diagonally Phase space (top) and den­
distorted lines. The second wiggler and laser k2 modulate the sity profile (bottom) of an
beam again, and the second chicane creates another distor­ EEHG-modulated beam.
3 G. Stupakov, PRL 102, 074801 (2009).
200 unifying physics of accelerators, lasers and plasma

tion, which, if properly adjusted, produces repeating vertical

lines in the z − E plane (see Fig. 10.18), which correspond to
spatial density modulation of the beam at high harmonics of
the lasers.
Spatial harmonics that can be created with this method
are defined by k = nk1 + mk2 , where the integers n and m
can be large, granting the potential for seeding at very short
wavelengths.

10.3 Stability of beams

Stability of particle beams or laser pulses is usually one of
the most important requirements of any design. Any practical
realization of a beamline or laser system has imperfections:
static (caused by misalignments) or dynamic (caused, e.g., by
vibrations) positioning errors. In this section, we will look at
just a few selected examples from this wide topic.

10.3.1 Stability of relativistic beams

Head–tail effects are often the cause of instabilities and dete­
rioration of bunch properties, and occur when an initial off­
set of the bunch head creates much larger oscillations in the
bunch tail.
However, we need to take into account that the fields of
a relativistic bunch are mostly transverse, as illustrated in
Fig. 10.19. Therefore, in the free space, the motion of the tail
7DLO of the bunch would be independent of the motion of the head,
+HDG and in particular as to whether the head has offset and/or os­
cillations.
FIGURE 10.19 For the head–tail instability to develop, it is necessary to
Fields of the bunch and have an “agent” to carry the information about the offset from
head–tail effects. the bunch head to the bunch tail.
This so-called agent can be, for example, the fields in­
duced by the bunch in the surrounding accelerating struc­
tures. The field of the opposite colliding bunch can also play
the role of this agent. We will study our examples starting
with the latter.

10.3.2 Beam–beam effects

Modern electron–positron colliders aiming at the highest lu­
minosity4 require tiny beam sizes at their interaction region
(IR). An example of the beam sizes for a 500 GeV CM e+ e−
ILC collider project is shown in Fig. 10.20. Note that these
beams are extremely flat, to ensure that the energy losses of
the beams caused by synchrotron radiation during beam col­
lision are significantly suppressed.
4 International Linear Collider Technical Design Report, 2013.
 advanced beam manipulation, cooling, damping and stability 201

FIGURE 10.20
Flat beam collision in an IR of a typical linear collider.

For a beam with uniform density distribution and trans­

verse sizes σx,y , the maximum field at the boundary of the
beam can be estimated using the Gauss theorem Eds = 4πQ,
which thus gives for this maximum field:
eN
E≈ (10.20)
(σx σz )
where we take into account that, for the beam with uniform
distribution in the longitudinal direction, the surface integral
ds turns into the contour integral de with charge Q taken
per unit of length.
Taking the contour integral in the above derivation along
the contour inside of the beam, we conclude that the field
inside of the beam grows linearly. The fields outside of the
beam can be estimated in the same manner. It is important
to note that, if the beam is very flat, the integral would be
almost independent of the offset from the beam in the y di­
rection until the point when the vertical offset reaches values
comparable with σx .
\
a [

a \
(

a \
FRQVW

\
_(_aH1 [ ]
[ a [
FIGURE 10.21
Fields of the flat beam.

Therefore, for vertical offsets ranging between σy and σy , the U

field will be almost constant and after that will start to de­
crease as 1/r, as illustrated in Fig. 10.21.
202 unifying physics of accelerators, lasers and plasma

Fields of the beam cause synchrotron radiation in the par­

ticles of the opposite beam. As previously mentioned, this ra­
diation is called beamstrahlung (Fig. 10.22). It results in an
increase of the energy spread of the beams and smears out
the luminosity spectrum. The produced SR photons create
large numbers of e+ e− pairs that create background in the ex­
perimental detector. Beamstrahlung energy losses can be es­
timated using the formulas from Chapter 3. We will leave
these estimations for your exercises while we return to the
transverse dynamic effect.
The transverse fields of the flat beam estimated above pro­
FIGURE 10.22 duce an angular kick for a particle travelling through the op-
Beamstrahlung. posite colliding bunch. For Gaussian transverse beam distri-
bution, and for a particle near the axis, the total angular kick
of the particle after collision is estimated as:

dx 2N re
Δx' = =− ·x
dz γσx σx + σy

dy 2N re
Δy ' = =− ·y (10.21)
dz γσy σx + σy
We can now introduce the notion of the Disruption param­
eter.
2N re σz 2N re σz
Dx = and Dy = (10.22)
γσx σx + σy γσy σx + σy

As follows from Eq. 10.21, the disruption parameter is

inversely proportional to the focusing distance f beam corre­
sponding to the focusing field of the bunch: Dy ∼ σz /f beam .

FIGURE 10.23
Consequent moments of high-disruption beam collision.
 advanced beam manipulation, cooling, damping and stability 203

A low-disruption parameter D « 1 means that the bunch

acts as a thin lens. Its high values (D » 1) link to the instance
when particles oscillate in the field of the opposite colliding
bunch.
. The number of oscillations is approximately equal to
D/(2π) and if D is around or bigger than ∼ 20, there will be
a couple or more oscillations during collision.
Imagine now that the beams had small transverse offsets
before the collision. The beams would attract each other and
start to oscillate. As the collision developed further, the new
portions of the beam would already have larger initial dis­
placements. The oscillations would grow in an unstable man­
ner, limited only by the finite length of the bunch.
An example of beam–beam collisions with Dy ∼ 24 and
with an 0.1σy initial offset between the beams is shown in
Fig. 10.23. We can see that the second half of the collision is
indeed noticeably disrupted.
Collisions in high disruption regimes have both their
challenges and their advantages. As we can note in Fig. 10.23,
the middle part of the collision shows that the beams are
nicely focused on each other. The beam–beam focusing forces
the beams to squeeze tighter and thus give an additional en­
hancement to the luminosity, which is an advantage. How­
ever, beam–beam instability may develop, and therefore the
luminosity enhancement would be compromised by a higher
sensitivity to the initial offsets.
We will later discuss one more collision scenario where
the disruption regime may become useful — in the method
aimed to overcome the hourglass effect (see Section 10.6).

10.3.3 Beam break-up and BNS damping

The pursuit of high-charge and high-current beams leads to
challenges caused by various imperfections. One particular
issue is caused by wakefields.
The interaction of the charged beam with the accelerating
structure, or the vacuum chamber in general, can generate

FIGURE 10.24
Beam break-up instability of a single beam. Fields left by the
bunch are shown qualitatively. Beam evolution from the initial
unperturbed shape (A) to the final BBU-distorted shape (B).
204 unifying physics of accelerators, lasers and plasma

electromagnetic fields — wakefields — that can act back on

the bunch itself.
In the RF accelerating structure, these fields can build up
resonantly and disrupt the bunch itself. This is called the
beam break-up (BBU) instability and it can happen either for
a single bunch or in a multi-bunch case.
The single-beam break-up instability is illustrated in
Fig. 10.24. The cure for the single-bunch BBU is called BNS
damping, according to the names of its creators.5
The mechanism of how BNS damping mitigates single-
bunch BBU is explained in the following: assume that the
bunch has an offset with respect to the center of the accel­
erating cavity. The bunch head will excite a transverse dipole
wakefield W (proportional to the offset) that will cause trans­
verse deflection of the tail, which can consequently result in
BBU.
We can note that the wake W acting on the tail is an ad­
ditional defocusing. The BNS damping recipe is thus to intro­
duce additional focusing to the bunch tail, which would then
compensate the wakefield W acting on the tail.
In order to compensate the wake W acting on the tail, one
needs to decrease the energy of the tail in such a way that
the effectively increasing focusing via lenses in the acceler­
ator channel exactly cancels out the wakefield’s defocusing
effect. The necessary energy difference between the head and
the tail of the bunch is achieved for the BNS damping by plac­
ing a bunch off-crest in the RF pulse, which then creates cor­
responding optimal BNS energy spread over the bunch (E − z
correlation), as illustrated in Fig. 10.25.

5)
:

+HDG
+HDG
7DLO 7DLO :

$FFHOHUDWLQJVWUXFWXUH

FIGURE 10.25
BNS damping method.

BBU can also occur in trains of bunches, wherein accu­

mulated wakefields act on the next bunch and the following
bunches in the train, enhancing their oscillation. A possible
method to cure the multi-bunch BBS is to enhance decoher­
ence of transverse modes that the beam excites. This can be
5 V. Balakin, A. Novokhatsky and V. Smirnov, in Proc. of the 12th Int.
Conf. on High Energy Accelerators, Fermilab, 1983.
 advanced beam manipulation, cooling, damping and stability 205

achieved via gradual variation of the parameters (such as iris

radii) of the individual cells in the accelerating cavity.
In this case, as qualitatively shown in Fig. 10.26, the time
structure of excited modes will be slightly different, and the
total sum wakefield will quickly lose coherence, minimiz­
ing the kick on the following bunch. Often, in addition to
variation of accelerating cells, additional passive damping
elements are inserted into the accelerating structures to ab­
sorb particularly the higher-order modes. Such structures are
called damped detuned structures and can be suitable for ac­
celerating trains of high-charge bunches with separation be­
tween bunches as minimal as just a few RF cycles.
Before moving on to the next section, we would like to
make a few remarks. We have seen in this section that BNS
damping prevents BBU instability by maintaining the con­
ditions for coherent motion of the bunch head and tail (i.e.,
preventing relative oscillations of the head with respect to the
tail). It is helpful to look at the word damping, in “BNS damp­
ing” (or the word echo, in “EEHG”) broadly. The method to
avoid BBU instability in multi-bunch cases involves making
use of decoherence.
7RWDOZDNHDX








]

FIGURE 10.26
Detuned structure as a cure for multi-bunch BBU instability.

We will now look at one more mechanism — Landau

damping, which is used both in plasma and accelerators, but
has, in some cases, a different meaning within these two ar­
eas.

10.3.4 Landau damping

Landau damping6 is the mechanism first discovered in
plasma. It describes a collisionless damping of collective
plasma oscillations. This mechanism can be illustrated via
the following approximate analogy.
6 L. Landau, J. Phys. USSR 10 (1946).
206 unifying physics of accelerators, lasers and plasma





  
 

 
 







FIGURE 10.27
For illustration of Landau damping mechanism.

Let us consider an ocean wave that travels with phase ve­

locity vw . Assume that a duck floats along the wave with the
velocity v, which is very close to vw ; see Fig. 10.27.
The duck would soon be “trapped” in the ocean wave,
which means that if the duck was initially moving faster than
the wave (v > vw ), it would slow down and thus the wave
would gain energy from the duck. In the opposite case, if the
duck was initially slower (v < vw ), the wave energy would be
transmitted to the duck.
The overall damping of the wave can therefore occur if
the distribution of velocities of ducks (or particles in plasma)
decreases for larger values of velocity — which is indeed typi­
cally the case, as shown in Fig. 10.27 for Maxwellian distribu­
tion. As there are fewer faster particles than slower particles,
the collective wave in plasma will be damped by the Landau
mechanism.
It is important to note that, for the Landau damping to
IY be possible, the distribution function should have a nonzero
number of particles (ducks) at the wave velocity vw . There-
fore, if the initial spread of velocities of ducks or particles
was not sufficient, increasing it as shown in Fig. 10.28 would
help to enhance Landau damping.
Landau damping is a very important mechanism for ac­

YZ Y celerators too, as it helps to provide beam stability for either
transverse or longitudinal motion. Increasing the spread of
certain beam parameters, as was just mentioned above, is of­
FIGURE 10.28
ten also done in accelerators, as a possible way to provide
Velocity spread and Landau
improved stability.
damping.
In particular, increasing the energy spread in circular ac­
celerators would result in, via the nonzero momentum com­
paction factor, the spread of revolution periods, thus helping
the longitudinal stability of the beam. Increasing the energy
spread of the beam with a laser heater may help in coping
with CSR-caused microbunching.
Spread of betatron frequencies (caused by energy spread
and nonzero chromaticity) is often also helpful to the stability
of the beams, as well as an additional spread introduced by
the so-called Landau octupoles — the magnets often inserted
 advanced beam manipulation, cooling, damping and stability 207

into the beamline specifically for the purpose of enhancing

the tune spread and thus helping the beam stability.
We would like to mention here that increasing the spread
of tunes or energy often helps accelerators simply because
it enhances the decoherence, preventing particles from as­
sembling into collective motion. The term Landau damping
is sometimes used in cases where the term decoherence would
be more appropriate.7 Details of a particular physics setting
define which of those two mechanisms is acting in each situ­
ation.

10.3.5 Stability and spectral approach

Beam or laser pulse stability issues are often analyzed based
on a spectral approach. We will omit detailed and rigorous
definitions here and focus only on a few highlights.
The disturbance — e.g., the position of a particular mag­
netic lens or optical element of a laser (we will define it as
x(t)) — often has the same characteristics as a random pro­
cess. In this case, the notion of the power spectral density p(f )
(dubbed “power spectrum”) should be used instead of the
usual Fourier spectrum.
The key property of the power spectrum is that its integral
is equal to the variance σ of the signal x:
+∞
σ 2 = (x2 ) = p(f ) df (10.23)
−∞

where we assume that the average of x is zero: (x) = 0.

Disturbance to our beam or laser line often comes from 3I
certain frequencies. The power spectrum is useful here as
it can identify the contribution from a particular frequency
range according to the following (Fig. 10.29):
f 2

σ 2 ( f 1 < f < f 2 ) = (x2 ) = p(f ) df (10.24)

f 1

Quantitative analysis of the influence of disturbances of­

ten requires the knowledge of the expected relative displace­ I I I
ment of a beamline element during a specific time duration τ.
The associated variance (contribution to it from a frequency FIGURE 10.29
range [f 1 ; f 2 ]) can also be calculated from the power spec- Power spectrum.
trum:
f2
([x(t + τ) − x(t)]2 )t = p(f ) 2[1 − cos(ωt)]df (10.25)
f1

You may notice that the description given above is not suf­
ficient for analyzing beamlines that have many elements dis­
tributed in space. In addition to time, we also need to take
spatial information into account.
7 Werner Herr, CERN Accelerator School, 2013.
208 unifying physics of accelerators, lasers and plasma

Therefore, our analysis should involve a spectrum that

depends on frequency as well as on wavenumber — a two-
dimensional power spectrum8 P(ω, k).
The 2D spectrum (also called PWK spectrum) can eval­
uate expected relative displacements of two beamline ele­
ments, separated by a certain distance L and after a certain
time interval T . Assuming that at t = 0 the beamline was per­
fectly straight, the variance of the relative misalignment after
a time T of two points separated by the distance L is given by
+∞
dω dk
σ 2 (T , L) = P(ω, k) 4[1−cos(ωT )] [1−cos(kL)] (10.26)
−∞ 2π 2π

Since the formula Eq. 10.26 can predict the stability of any
two elements in our beamline, we can also evaluate the sta­
bility of the entire beamline by properly taking into account
all of the elements.

FIGURE 10.30
Examples of power spectrum P(ω, k) (left), spectral response
function G(k) and characteristic function of the feedback F(ω).

Combining the coefficients that determine how displace­

ment x of an individual beam element or laser line con­
tributes to, for example, displacement of the beam/light at
the focus xout , we can construct a so-called spectral response
function G(k). The variance of the relative misalignment of
the output beam/light after a time T depends on G(k) as
+∞
dω dk
σ 2 (T ) = P(ω, k) 2[1−cos(ωT )] G(k) (10.27)
−∞ 2π 2π

The variance of the misalignment defined above in

Eq. 10.27 depends on the observation time T and typically
8 A. Seryi and O. Napoly, Phys. Rev. E. v.53, 5323, 1996.
 advanced beam manipulation, cooling, damping and stability 209

grows with T . To stabilize the beam/light at the focus, a feed­

back would usually need to be applied, which can be de­
scribed by the characteristic function of the feedback F(ω). In
this case the variance of the relative misalignment of the out­
put beam/light is a constant, and is given by
+∞
dω dk
σ2 = P(ω, k) F(ω) G(k) (10.28)
−∞ 2π 2π

Examples of a 2D power spectrum as well as a spectral

response function and characteristic of feedback for a par­
ticular beamline (final focus system beamline) are shown in
Fig. 10.30.
A particularly useful and insightful approximation for a
2D spectrum is the one related to the ATL law, which was first
suggested9 to describe space–time motion of ground (earth),
on which the focusing elements of a beamline are installed.
For this “ATL law,” the variance of the misalignment for
elements separated by distance L and observed after time T
(described above by Eq. 10.26) is simply

σ 2 (T , L) = A · T · L (10.29)

where A is the parameter of the model that could be esti­

mated, for example, from ground motion measurements of
our site (where the elements of beamline are installed) or
perhaps of our optical table (where the laser elements are
mounted).
The 2D power spectrum corresponding to the “ATL law”
is given by
A
P(ω, k) = 2 2 (10.30)
ω k
The 2D spectrum shown in Fig. 10.30 includes the ATL com­
ponent; however, it is included in a corrected way, at higher
frequencies and wavenumbers, to ensure integrability of the
equations.
The term “law” in “ATL law” should be taken cautiously
here, as this is certainly a model, an estimate that is appli­
cable in certain ranges of parameters and certain situations.
It is, however, amazing how many different situations and
wide parameter ranges10 there are in which this approxima­
tion can be found valid.

10.4 Beam or pulse addition

The topic of beam or pulse addition is one where there is
an important difference between the area of accelerators —
9 B. A. Baklakov et al., INP Report 91–15, 1991.

10 V. Shiltsev, Phys. Rev. Lett. 104, 238501 (2010).

210 unifying physics of accelerators, lasers and plasma

where charged particles are used — and lasers, where we deal

with light.
In this section, we will briefly discuss the methods of laser
pulse addition in the optical cavities in comparison with in­
jection methods used in accelerators, return to lasers to touch
on a promising technique to coherently combine fiber laser
signals, and finish with resonant plasma excitation by a train
of laser pulses.

10.4.1 Optical cavities

Optical cavities consisting of two or more mirrors are a
widespread element used in practically any laser or optics
system.
Optical cavities, in the event when one of the mirrors is
semi-transparent, are efficient tools for adding laser pulses
together for their accumulation to higher intensities.
Various arrangements of optical cavities are possible, and
some examples are shown in Fig. 10.31.

$ % &

FIGURE 10.31
Examples of optical cavities. Plane-parallel (A), concen­
tric/spherical (B) and confocal (C) configurations.

An optical cavity suitable for hosting interaction between

an electron beam and a laser (for Compton sources in partic­
ular) can be composed of two or more mirrors.
One of the important concerns is the cavity’s stability and
the light pattern inside of it, and also how it reacts to small
deviations in its parameters.
It has been found that, while two-mirror systems are pos­

(OHFWURQV (OHFWURQV (OHFWURQV

FIGURE 10.32
Examples of four-mirror optical cavity suitable for electron beam–
laser interaction.
 advanced beam manipulation, cooling, damping and stability 211

sible, the four-mirror optical cavities are more practical,11

and are especially more impervious to disturbances.
Moreover, the four-mirror cavities can also be more suit­
ably arranged when one needs to pass the electron beam
around the cavities, as shown in the third case presented in
Fig. 10.32.
There is an essential difference between the accumulation
of pulses of light and the accumulation of bunches of parti­
cles, as we will discuss momentarily.

10.4.2 Accumulation of charged particle bunches Liouville' s theorem: In the

vicinity of a particle, the par-
The accumulation of bunches of charged particles (e.g., at in­
ticle density in phase space
jection) obeys Liouville’s theorem (see its full definition in the is constant if the particle
margin). moves in an external mag-
The consequence of the theorem is the conservation of netic field or in a general field
emittance ε, i.e., the area p dq of the phase space beam por- in which the forces do not de-
trait is constant (during accelerating, γε is constant). pend upon velocity.
Beam cooling methods (SR, electron, stochastic) do not
S
contradict Liouville’s theorem, as in these cases the cooling
forces acting on the particle depend on its velocity.
One important consequence of the theorem is that we can­
not stack bunches of charged particles on top of each other in
the phase space. If we would like to accumulate them to in­ T
crease the total intensity, we must inject the bunches into the
neighboring areas of the phase space (Fig. 10.33).
Phase-space stacking and accumulation of bunches dur­
ing injection can be done in various ways. Fig. 10.34 shows FIGURE 10.33
an example of transverse stacking. Phase-space stacking.

FIGURE 10.34
Transverse phase-space stacking. Consecutive moments.

Longitudinal stacking is also possible, as illustrated in

Fig. 10.35. Fast-switching bending magnets or electrostatic
deflectors — kickers — are used to perform bunch stacking
during injection.
11 A. Variola et al., ThomX — Conceptual Design Report, LAL, 2010.
212 unifying physics of accelerators, lasers and plasma

There is one particular injection mechanism that can

overcome limitations imposed by Liouville’s theorem — the
charge-exchange injection. In this case, H− ions are injected
into the beamline through a thin foil. Ion interactions at the
foil strip away two electrons and the ions are therefore con­
verted (at the foil) to protons.

,QMHFWHGEHDP

6
(PSW\SKDVHYROXPH

FIGURE 10.35
Longitudinal phase-space stacking.

This charge-exchange injection can be applied many times

into the same phase space. Since the foil is inserted into the
accelerator on the way to the main circulating beam (see
Fig. 10.36), the number of injections is limited by the scat­
tering of the circulating beam on this foil.

%HQGLQJPDJQHWV

S
,QMHFWHGEHDP

+

6WULSSLQJIRLO

S &LUFXODWLQJEHDP

FIGURE 10.36
Charge-exchange injection.

10.4.3 Coherent addition of laser pulses

Returning to the laser theme, we would like to briefly men­
tion a technique recently suggested12 to coherently add laser
pulses.
Refer to Sections 4.2.3 and The technique is aimed in particular to fiber lasers that
4.2.4 for discussion of fiber can produce pulses with high efficiency (tens of percent) and
laser repetition rate and TRIZ high repetition rates (tens of kHz and higher). The idea of
inventive principles.
12 Gerard Mourou et al., Nature Photonics 7, 258–261 (2013).
 advanced beam manipulation, cooling, damping and stability 213

the technique, which is presently under development, is to

coherently combine the output of many fiber lasers, with
the goal to produce laser systems with high repetition rates,
high energy and high efficiency. The concept is illustrated in
Fig. 10.37.
The technique of coherent fiber laser pulse combination
is striving to solve the deficiency of present high-peak power
lasers (such as Ti:Sapphire or Nd:glass lasers) both of which
have high peak powers but very low repetition rates (ranging
from one shot per hour to around a Hz) and low efficiency
(less than 0.01%).

FIGURE 10.37
Concept of fiber laser coherent combination of pulses.

This technique, once developed, can serve as an efficient

laser driver for laser plasma acceleration based light sources,
which can be applied to the field of medicine, among others.

10.4.4 Resonant plasma excitation

We will mention one more accelerator technique that is cur­
rently being developed — resonant plasma excitation.13
In this case, the amplitude of plasma oscillation accumu­
lates slowly, while plasma oscillations are excited by a train
of low-energy laser pulses distanced at plasma wavelengths.
Such a technique is particularly suitable for use in conjunc­
tion with fiber lasers because it can provide the driving laser
trains with high efficiency and high repetition rates.
This technique, once finalized, can also offer the path to
efficient laser-driven plasma acceleration based light sources
(which have applications beyond accelerator physics).
13 S.M. Hooker et al., J.Phys. B47 (2014) 234003.
214 unifying physics of accelerators, lasers and plasma

10.5 Cooling and phase transfer

We will now examine several topics related to beam cooling
and phase space manipulation.

10.5.1 Beam cooling methods

The three main methods of beam cooling (apart from SR) are
shown conceptually in Fig. 10.38.

FIGURE 10.38
Electron cooling, stochastic cooling and ionization cooling con­
cepts.

In addition to electron and stochastic cooling, which we

have already discussed, we add ionization cooling — which
is based on subjecting particles to ionization losses in an ab­
sorber, and then restoring the longitudinal component of the
particles’ momentum.
Stochastic and electron cooling are well-developed tech­
niques. In particular, stochastic cooling was essential for the
discovery of W and Z Bosons (C. Rubbia and S. van der Meer,
1984 Nobel Prize in Physics). Electron cooling was most re­
cently used for improving the operation of the Tevatron col­
lider.
Conversely, ionization cooling is a concept still under de­
velopment. Conceptually simple, it is quite challenging tech­
nologically. However, this technique might be the only way to
reduce the emittance of short-lived particles such as muons.

10.5.2 Electron cooling, electron lens and Gabor lens

In order to stimulate our TRIZ-inspired discussion, we will
now draw a parallel between the three different techniques,
which have completely different purposes but have, never­
theless, similarities.
Electron cooling, as we discussed, uses an electron beam
co-propagating with the proton beam with equal velocities
in order to enable their energy exchange. The electron beam
is guided by a magnetic field. It is essential for the e and p
velocities to be equal. The conceptual layout of an electron
cooler is shown in Fig10.39.
 advanced beam manipulation, cooling, damping and stability 215

An electron lens14 is geometrically a very similar device

wherein the proton (or antiproton) beam passes through the
electron beam and receives additional focusing due to the
fields of the electron beam; the lens aims to mitigate variation
of betatron tunes between different bunches of proton beams.
As the main effect of the lens comes from the electric field of
the electron bunch, the direction of the e-beam travel is not
essential (usually, counter-propagating beams are used). The
e-lens concept is shown in the same Fig. 10.39.

3URWRQEHDP

% %
3 % 3

HEHDP
8 3 HFRROLQJ
&DWKRGH 3 HOHQV &ROOHFWRU
FIGURE 10.39
Electron cooling or electron lens.

We will also mention one more arrangement involving an

electron beam and a magnetic field — the Gabor lens. Sug­
gested15 in 1947, this technique recently attracted attention16
in connection to its use for laser plasma ion acceleration, for
energy selection of accelerated protons.

FIGURE 10.40
Conceptual schematic of a Gabor lens.
14 V. Shiltsev et al., PR-ST-AB 2, 071001 (1999).

15 D. Gabor, Nature 160, 89–90 (19 July 1947).

16 J. Pozimski et al., Laser and Particle Beams, v. 31, 04, 2013, pp. 723–733.

216 unifying physics of accelerators, lasers and plasma

A conceptual schematic of a Gabor lens is shown in

Fig. 10.40. In this lens, the electron beam is formed by a cath­
ode with a hole in the center. Electrons trapped in the center
of the lens are attracted to the anode and are contained by the
magnetic field.
A Gabor lens can potentially accumulate a large electron
charge. In a steady state, the electrons rotate around the axis
and the electrostatic repulsion — together with the centrifu­
gal force — will balance the radial Lorentz force produced
by the magnetic field. Using this assumption, the maximum
density of electrons in the Gabor lens can be estimated as:

B2
n=− (10.31)
8π me c 2

which can be high enough to be considered for use as a lens

created by the radial field of the stored electrons.
The electrons stored in the Gabor lens move longitudi­
nally in both directions. In Fig. 10.41, we summarize the re­
lationships between the velocities of the proton (antiproton,
ion) beams and the velocities of electrons in these three de­
vices.

HFRROLQJ HOHQV *DERUOHQV

3 3 3
H H H

FIGURE 10.41
Relations of velocities of proton and electron beams in different
configurations: electron cooling, electron lens, Gabor lens.

In the above, we have considered three systems that in­

volve electron beams or stored electrons. These systems are
aimed at different applications and details of their functions
are very different. We assume, however, that it is useful to
look at these systems simultaneously, as discussion of analo­
gies between these systems is helpful for our TRIZ analy­
sis (which runs in parallel with the main accelerator-laser­
plasma story of this book).

10.5.3 Laser cooling

We conclude this section on cooling methods with laser cool­
ing of ions. The method relies on the Doppler frequency shift,
which is the key to laser cooling.
The cooling process involves the following steps (see
Fig. 10.42). First, the laser photon coming from a certain di­
rection hits the atom and is absorbed. The excited atom then
re-emits a photon into a random direction.
 advanced beam manipulation, cooling, damping and stability 217

D E F
W  W 

FIGURE 10.42
Laser cooling steps. Absorption of a photon by an atom (a);
excited state of the atom (b); emission of a photon (c).

The laser is in resonance with the atoms only when they

are moving towards the laser, but not if they are moving side­
ways or away as shown in Fig. 10.43. This condition ensures
the eventual cooling of the ions.

/DVHU
$EVRUSWLRQ

FIGURE 10.43
Relation between laser wavelength and Doppler shifted reso­
nance absorption of an atom moving in different directions.

10.6 Local correction

In accelerators — as well as in any other technical fields —
any unwanted disturbances are corrected either locally or
non-locally. The universality of the approaches can also be
observed with help from TRIZ, which includes relevant in­
ventive principles (principle of preliminary anti-action).
Non-local correction is often used when it is not possible
to correct the disturbance at its origin. The general issue with
this approach lies in its non-locality — a correction needs to
propagate and be properly preserved from the point of pre­
liminary correction to the point where it needs to act.
Local corrections, if they can be used, are often superior,
as they correct the disturbance at the origin. In this section,
we will discuss several examples of local corrections.

10.6.1 Final focus local corrections

A final focusing system (FF) of any collider is aimed to produce
a small beam size at the interaction point (IP). The final lenses
of the system (usually arranged in a final doublet — FD) are
the strongest and produce the largest chromaticity. The val­
ues of chromaticity are usually very large (if left uncorrected,
218 unifying physics of accelerators, lasers and plasma

they increase the beam size at IP by orders of magnitude) and

need to be compensated.
In a traditional final focus design, the chromaticity cor­
rection is performed in dedicated optical sections upstream
from IP. In these sections, sextuple magnets are installed
in dispersion sections and create some chromaticity there,
which will then propagate to the FD, where they will cancel
the FD chromaticity.
The performance of such an FF with non-local chromatic­
ity compensation is often limited, as various additional dis­
turbances (e.g., the SR-generated energy spread), which occur
between the chromatic correction section and the FD, can al­
ter the cancellation conditions.
An alternative design17 exists: a final focus with local
chromatic correction, shown conceptually in Fig. 10.44.

FIGURE 10.44
Final focus with local chromaticity correction.

In this design, chromaticity is canceled locally by two sex­

tupoles interleaved with FD, while a bend upstream gener­
ates the dispersion (which is necessary for the sextuples to
generate chromaticity) across the final doublet. The value of
dispersion in the FD is usually chosen so that it does not in­
crease the beam size in the FD by more than 20 to 30% for a
typical energy spread of the beam.
Local chromatic compensation by FD sextuples needs to
compensate the FD chromaticity without introducing other
unwanted aberrations. In a vertical plane, the chromaticity
compensation is straightforward, and therefore the vertical
FD sextupole should be tuned to completely cancel the verti­
cal FD chromaticity.
However, the FD horizontal sextupoles also introduce hor­
17 P. Raimondi and A. Seryi, Phys. Rev. Lett., 86 (17), 3779, (2001).
 advanced beam manipulation, cooling, damping and stability 219

izontal second-order dispersion. Second-order dispersion is

produced by FD quadrupoles as well as FD sextupoles, while
the latter produce just half as much of second-order disper­
sion as the FD quadrupole. Therefore, if we tune the FD hor­
izontal sextupole to cancel the FD chromaticity, half of the
uncompensated second-order dispersion will remain.
A solution to this issue consists of allowing the beta-
matching section (shown schematically in the beginning of
the beamline in Fig. 10.44) to produce as much horizontal
chromaticity as the final doublet, so that the horizontal FD
sextupoles will run twice as strong and simultaneously can­
cel the second-order dispersion and horizontal chromaticity.
Geometric aberrations to the final doubled sextupoles are
cancelled in this design by two more sextupoles placed in
phase with the FD sextupoles and upstream from the bend.
The final focus system can be reversed and the IP consid­
ered as an entrance point that captures a strongly diverging
beam, such as the one coming out of the laser plasma acceler­
ating bubble. This beam will also have an energy spread and
associated chromaticity effects in the capture optics. A very
small but divergent beam corresponds to a small beta func­
tion at the origin; therefore the chromatic effect for such a
beam should be significant. The local chromaticity correction
approach can therefore be applicable also to the laser-plasma
capture optics.

10.6.2 Interaction region corrections

The interaction region of linear colliders exhibits several
design inventions that can illustrate several TRIZ inventive
principles.

FIGURE 10.45
Conceptual layout of experimental detector and beamlines in the
interaction region of a linear collider.
220 unifying physics of accelerators, lasers and plasma

A typical layout of a detector and beamlines is shown in

Fig. 10.45. One can see here that an anti-solenoid is inserted
into the beamline. This anti-solenoid is needed to cancel the
beam coupling effects that the main solenoid produces on the
beam.
The problem with the anti-solenoid is that a huge force
will be exerted on it because of the main solenoid. A solution
implemented18 in IR design is to use a dual anti-solenoid,
where the outer coil with current I2 = −I1 (R1 /R2 )2 cancels
the external field; this makes it force-neutral (see Fig. 10.46).

FIGURE 10.46
Standard solenoid (A) and interaction region dual solenoids (B).

From the point of view of TRIZ, this dual solenoid is an

example of application of the nested doll and system–anti­
system inventive principles.

10.6.3 Travelling focus

The travelling focus regime19 of beam collisions employs
beam–beam focusing strengths in order to overcome the
hourglass effect.
The hourglass effect (see Fig. 10.47) prevents reduction of
the beta function at the IP to values lower than the length of
the bunch, as further increases in focusing distort the bunch
(into an hourglass shape) without increases to the luminosity.
FIGURE 10.47 In a travelling focus regime, the beam is focused to βy «
Hourglass effect. σz and the location of focus dynamically changes during the
collision. The focal point of the colliding bunches is made to
coincide with the location of the head of the opposite bunch.
This helps to optimally use additional focusing due to beam–
beam forces and keeps the beams properly focused on each
other during the entire collision.
18 B. Parker, ca. 2002. A similar solution is used in nuclear magnetic reso­
nance (NMR) scanners.
19 V. Balakin, ca. 1991.
 advanced beam manipulation, cooling, damping and stability 221

FIGURE 10.48
Travelling focus collisions.

Fig. 10.48 shows a simulation of traveling focus. The

arrows show the position of the focus point during col­
lision. This method hasn’t yet been experimentally tested.
One of the particular difficulties to this method consist of
an increased sensitivity to imperfections, i.e., to the initial
beam offset. This example gives us, however, additional in­
formation for a TRIZ-like analysis of the local correction ap­
proaches.

10.6.4 Crabbed collisions

The interaction region shown in Fig. 10.45 involves a collision
of beams with a certain nonzero crossing angle θc .
With crossing angle θc , the projected x-size is
)
σx2 + θc2 σz2 ≈ θc σz
[
Taking the IP beam parameters shown on Fig. 10.20 and tak­ ,3
ing θc ≈ 15 mrad, we can conclude that the projected hori­ $
zontal size is equal to several micrometers — which is several
times larger than the nominal size. This is illustrated in the
upper part of Fig. 10.49, where incomplete overlaps of the
beams are apparent.
The crossing angle collision will result, therefore, in a sub­
stantial (by several times) reduction in luminosity, unless this [
effect is locally compensated.
Compensation of the crossing angle effect can be achieved ,3
by giving the bunch a z-correlated kick in such a way that the %
beam starts to rotate in the horizontal plane and arrives at the 5)NLFN
IP properly overlapping with the opposite beam — as shown
at the bottom part of Fig. 10.49.
The z-dependent kick on the bunch can be produced by FIGURE 10.49
a special transverse mode cavity called a crab cavity, which Collisions of the beams with
does not disturb the central particle of the beam, but kicks crossing angle at the IP. Nor­
the head and the tail particles in opposite directions in the mal (A) and crabbed (B) col­
x-plane. The design of a crab cavity and its fields are shown lisions.
in Fig. 10.50.
222 unifying physics of accelerators, lasers and plasma

FIGURE 10.50
Crab cavity and its fields.

10.6.5 Round-to-flat beam transfer

In the final section of this chapter we will discuss round-
to-flat or flat-to-round beam transformation, suggested by
Y. Derbenev.20

FIGURE 10.51
Beamline magnetic elements and phase-space portraits of the
beam subjected to flat-to-round beam transformation. Initial flat
beam, vortex, parallel beam in the solenoid.
20 Ya. Derbenev, Adapting Optics for High Energy Electron Cooling, Uni­
versity of Michigan Report No. UM-HE-98–04, 1998.
 advanced beam manipulation, cooling, damping and stability 223

This method is primarily intended for the beams that have

significant asymmetries between phase planes. In particular,
flat beams are often generated in accelerators. For example,
SR rings naturally have y emittance much smaller than the x
emittance.
It has been shown by Derbenev that a triplet of skew
quadrupoles can transform such a flat beam to a vortex, as
illustrated in Fig. 10.51.
The beam with vortex-like phase space distribution can
be transformed further, noting that an edge of the solenoid
also creates a vortex-like phase space portrait for an initially
parallel beam.
Therefore, sending the vortex beam created by the skew
triplet into the solenoid with appropriate strength will result
in creating a round beam with zero angles — as shown in the
final step in Fig. 10.51.
The transformation described above can be used in many
accelerator systems. A particular example21 uses this trans­
formation to reduce space charge effects of flat beams in
damping rings with long straight sections by employing flat­
to-round and round-to-flat transformations at the beginnings
and exits of each section, respectively. The round beam in
straight sections will have lower space charge effects, improv­
ing the beam stability.

21 R. Brinkmann, Ya. Derbenev and K. Floettmann, Phys. Rev. ST Accel.

Beams 4, 053501 (2001).
224 unifying physics of accelerators, lasers and plasma

EXERCISES
10.1 Chapter materials review.

Describe the typical approaches for preventing beam break­

up instability in linacs. Discuss if these approaches can be

used in other situations considered in this book.

10.2 Chapter materials review.

Using the formulas in Chapter 3, derive an estimate for SR

energy losses (beamstrahlung) for colliding e+ e− bunches. Jus­

tify the need to use flat beams in a high energy e+ e− col­
lider. Estimate the energy losses for beam parameters shown
in Fig. 10.20.
10.3 Chapter materials review.
For the same beams as in the previous exercise, estimate the
number of emitted beamstrahlung photons per particle in the
case when the oncoming positron bunch has either a 3 nm or
30 nm vertical offset with respect to the electron bunch.
10.4 Mini-project.
Define approximate parameters of a 1 GeV compact SR
source ring aimed at 10 keV X-rays, based on top-off, on-orbit
and on-energy injection by a laser plasma acceleration sys­
tem. Discuss the research steps required for implementing
such a concept.
10.5 Analyze inventions or discoveries using TRIZ and AS-TRIZ.
Analyze and describe scientific or technical inventions de­
scribed in this chapter (e.g., coherent combination of laser
pulses) in terms of the TRIZ and AS-TRIZ approaches, iden­
tifying a contradiction and an inventive principle that were
used (could have been used) for these inventions.
10.6 Developing AS-TRIZ parameters and inventive principles.
Based on what you already know about accelerator science,
discuss and suggest the possible additional parameters for
the AS-TRIZ contradiction matrix, as well as the possible ad­
ditional AS-TRIZ inventive principles.
 11
Inventions and Innovations in
Science

11.1 Accelerating At last, we have reached the final chapter.

Science TRIZ 226 We have seen, throughout this book, that the three tech­
11.2 Trends and nical fields we discussed — accelerators, lasers and plasma
principles 227 — sometimes have a lot in common, things such as scientific
11.3 Engineering, TRIZ approaches and solutions, similar inventions and technolog­
and science 229 ical breakthroughs. We have seen several examples when the
11.4 Aiming for Pasteur TRIZ, or AS-TRIZ, inventive principles can be retrospectively
quadrant 233 identified in various scientific inventions.
11.5 How to cross the In this final chapter, we would like to return to the topic
Valley of Death we began to discuss in the very first chapter of this book —
236 the methodology of inventiveness.
11.6 How to learn TRIZ We will start by reviewing the updated Accelerating Sci­
in science 240 ence TRIZ principles.
11.7 Let us be Following that, we will then look critically at some of the
challenged 242 suggested AS-TRIZ inventive principles, and discuss whether
these are indeed principles or rather trends of the evolution
of scientific systems. In this discourse, we will recall the def­
inition of the laws of the evolution of technical systems as
described by the standard TRIZ principles. After observing
the parallels between the evolution of radar and lasers, we
will conclude that some of the suggested AS-TRIZ inventive
principles should be redefined as the laws of the evolution of
technical/scientific systems.
We will then continue to consider the inventive principles
and the laws of the evolution of systems by taking a very gen­
eral approach, and will look at various fields of science and
even beyond those.
The next topic we will discuss is the way to take these sci­
entific inventions and apply them in practical settings. We
will look at the linear and two-dimensional models of inno­
vation and discuss the notion of the Pasteur quadrant in ap­
plication to accelerator science in particular.
The well-known challenge of “crossing the Valley of
Death” in technological innovation will be our next topic, and
we will suggest a possible method for crossing this so-called
valley in regards to the field of sci-tech, with help from a se­
lection of compact light source projects.
In the next section we will offer a possible approach of
using TRIZ to teach future generations of science students. In
fact, the method we have used throughout this book has not
been the ready-to-use standard TRIZ; instead, we immersed

DOI: 10.1201/b18696-11 225

226 unifying physics of accelerators, lasers and plasma

ourselves in the proactive process of developing an extension

of TRIZ — Accelerating Science TRIZ.
And finally, we will aspire to apply these developed prin­
ciples to new, challenging projects and their unsolved prob­
lems.

11.1 Accelerating Science TRIZ

First of all, we hope that the reader has already guessed the
play of words in the term “Accelerating Science TRIZ.” The
word “accelerating” does not refer here to accelerators — the
devices that help particles reach higher energies. Instead, the
word “accelerating” in AS-TRIZ implies that science itself can
be accelerated via the application of these inventive princi­
ples.
Throughout this book we have seen several recurring in­
ventive principles that can be added into the tables of the AS­
TRIZ extension. Shown below are the updated tables where a
number of parameters for the AS-TRIZ matrix of contradic­
tion (Table 11.1), as well as several inventive principles (Ta­
ble 11.2) for the AS-TRIZ, are included.

TABLE 11.1 TABLE 11.2

Updated table of emerging AS-TRIZ Updated table of emerging AS-TRIZ
parameters inventive principles
No. Parameter No. Principle
1. Energy 1. ...
2. Rate of energy change 2. ...
3. Emittance 3. Undamageable or already damaged
4. Luminosity 4. Volume-to-surface ratio
5. Brightness 5. Local correction
6. Intensity 6. Transfer between phase planes
7. Efficiency 7. From microwave to optical
8. Power 8. Time–energy correlation
9. Integrity of materials ... ...
10. Time duration or length 21. ...
11. Spatial extent
12. Sensitivity to imperfections
13. Cooling rate
14. ...

... ...

21. ...

In addition to the principles of “undamageable or already

damaged” and “volume-to-surface ratio” mentioned already
in Chapter 1, we have included the inventive principle re­
lated to local correction (based on various examples from
the focusing systems in beams or lasers), transfer between
 inventions and innovations in science 227

phase planes (based, for example, on Derbenev’s transforma­

tion), transition from microwave to optical (recall radar and
CPA, or conventional and laser acceleration), and using time–
energy correlation (laser pulse or bunch compression).
The reader can perhaps already add other inventive prin­
ciples to this list — in fact, one of the purposes of keeping
the unfilled spaces in the tables above is to fill them in, dur­
ing the proactive process of learning.

11.2 Trends and principles

Standard TRIZ describes not only the inventive principles,
but also the laws (trends) of the evolution of technical sys­
tems. Knowledge of these evolution laws can help to create
innovative inventions.
There are just a few laws/trends of the evolution of tech­
nical systems and they can be described in broad terms. The
inventive principles, however, are more concrete, as they are
unique approaches that help to solve contradictions — there
are several dozens of these.
The lists of additional AS-TRIZ inventive principles de­
scribed in Table 11.2 should not be interpreted as cast in
stone. Compiling such tables should instead be considered
as a process of proactive learning parallel to the application
of the TRIZ method to science. A critical analysis is therefore
appropriate.
With this notion in mind, we would now like to look more
carefully at the inventive principle “from microwave to opti­
cal” that we introduced in Table 11.2, and discuss whether
this is indeed a principle or rather a general law (trend) of
the evolution of technical systems.

11.2.1 TRIZ laws of technical system evolution

Standard TRIZ defines three types of laws of technical system
evolution — static, kinematic and dynamic.
The three static laws are as follows. The TRIZ law of the
completeness of the parts of the system states that a system
should have the following four parts: an engine, a transmis­
sion, a working unit and a control element. The law of energy
conductivity of the system says that every technical system is a
transformer of energy and the energy should circulate freely
and efficiently through these four main parts of the system.
The law of harmonizing the rhythms of the parts of the system
suggests that the frequencies and periodic motions of these
parts should be in sync with each other.
The three kinematic laws are defined in standard TRIZ as
follows. The law of increasing the degree of ideality of the system
states that the “ideality,” which is a qualitative ratio between
228 unifying physics of accelerators, lasers and plasma

all desirable benefits of the system and its cost or other harm­
ful effects, should have the tendency to increase. The law of
uneven development of parts of a system suggests that the differ­
ent parts of a technical system will evolve differently, leading
to new technical and physical contradictions. And finally, the
law of transition to a super-system states that a system which
has exhausted the possibilities of further significant improve­
ment is included in a super-system as one of its parts.
Finally, the two dynamic laws are as follows. The law of
transition from macro to micro level suggests that the develop­
ment of working organs proceeds initially on a macro and
then more and more on a micro level. And the law of in­
creasing involvement of fields in the system states that the fields
evolve from mechanical fields to electro-magnetic fields.
We note that the phrase in the previous paragraph,
“the fields evolve from mechanical fields to electro-magnetic
fields,” can be extended and clarified as meaning that, within
the electro-magnetic spectrum, the evolution should follow
from RF frequencies to optical frequencies.
We can conclude from this comparison and discussion
that “from microwave to optical,” which we initially iden­
tified as an inventive principle, is in fact better suited to be
defined as a general trend of technical/scientific system evo­
lution.
Let’s highlight this conclusion by using an example of the
invention of CPA in connection to radar.

11.2.2 From radar to high-power lasers

In 1935, Robert Watson-Watt developed the RDF (Radio-
Direction Finding), which later become known as RADAR
(RAdio Detection And Ranging). Development of the radar
device took place at Ditton Park in England, which later be­
came known as the Appleton Laboratory and, after merging
with the Rutherford Laboratory, became the Rutherford Ap­
pleton Laboratory — currently the home of many accelerators
and lasers (and located just a half-hour drive from Oxford —
the place where this book was written).
The operation of radar involves chirped pulse amplifica­
tion. The positive influence of radars on the span of the 20th
century cannot be discounted. Indeed, we would not have
been party to the tremendous and revolutionary advances in
laser technology had it not been for the chirped pulse ampli­
fication — CPA, developed around 1985 by D. Strickland and
G. Mourou.1
The development of CPA stimulated an exponential
growth of available laser powers. From around 1990 to the
1 D. Strickland and G. Mourou, Compression of amplified chirped optical
pulses. Opt. Commun. 56, 219–221 (1985).
 inventions and innovations in science 229

early 2000s, the peak laser power increased by more than two
orders of magnitude, reaching PW levels.
High-power lasers, in their turn, enabled progress in
many technical and scientific areas. For instance, nuclear
physics, the production of Giga–Gauss magnetic fields, fu­
sion science, studies of material properties in extreme con­
ditions, laboratory astrophysics, studies of turbulence, non­
linear quantum electrodynamics and many other fields ben­
efitted from the development of CPA lasers.
The example of radar and CPA once more asserts TRIZ’s
founding proposal that the same problems and solutions ap­
pear again and again, but in different disciplines. It also il­
lustrates the general trend of technical/scientific system evo­
lution (i.e., the transition from the microwave part of the EM
spectrum to the optical range).

11.2.3 Modern laws of system evolution

The laws/trends of the evolution of systems were defined by
standard TRIZ several decades ago. The laws are very broad
and will likely remain valid for a long time. Still, it is tempt­
ing to look at these laws from a present-day perspective and,
initially, see whether the laws need to and/or can be updated,
and, secondly, attempt to apply these laws to other areas of
science, and also to seemingly unrelated areas.
In order to illustrate the tendencies that can be redefined
as the laws of the evolution of scientific systems, let us recall
the example of tumor therapy wherein the effects of a proton
beam can be combined and synergistically enhanced by using
small compounds. The latter can be created specifically for a
particular patient, based on a genetic analysis of their tumor.
This example shows, first of all, the trend of using more
than one technique for achieving synergy and a superior over­
all effect. And second, this illustrates the universal trend of
developing solutions and services tailored to specific persons.
The reader can certainly suggest other examples that il­
lustrate these or other trends. The fascinating question of
whether the noted trends are indeed modern general laws of
scientific system evolution is better left for further discussion
outside of the framework of this text.

11.3 Engineering, TRIZ and science

There are many well-founded connections between science
and engineering. We will take this time to discuss several par­
ticular aspects of these connections while highlighting their
link to TRIZ methodology.
230 unifying physics of accelerators, lasers and plasma

11.3.1 Weak, strong and cool

Recall our discussion relating to the focusing in accelerators
that is needed to keep particle trajectories near the center.
The first accelerators had weak focusing with spatial periods
greater than the perimeter of the accelerator. The trajectories
of particles in a weak focusing accelerator had large trans­
verse excursions, requiring extremely large vacuum chamber
apertures.
Strong focusing was suggested as a better way to focus
particles by involving the usage of a sequence of focusing–
defocusing elements. This example in itself may serve as an
illustration of the TRIZ inventive principle of system and an­
tisystem.
Strong FODO focusing may When strong focusing was suggested, it was not totally
illustrate TRIZ inventive
clear if it would work in practice. The realization of the
principle system–antisystem.
first strong focusing proton accelerator was led by Sir John
Adams, who had the courage to cancel (in October, 1952) the
already approved 10 GeV weak focusing accelerator for a to­
tally innovative 25 GeV Proton Synchrotron.
The risk was worth it — in 1959, Sir John Adams an­
nounced that CERN’s Proton Synchrotron had just reached
24 GeV and had surpassed Dubna’s Synchrophasotron world
record of 10 GeV.
Another example of scientific risk-taking relates to the de­
velopment of electron beam cooling (which was especially
necessary for antiparticles such as antiprotons). The method
was suggested by G.I. Budker, the founder and first direc­
tor of the Institute of Nuclear Physics in Novosibirsk. Budker
was the mind behind many inventions in the field of physics,
including the concept of electron cooling.
However, when electron cooling was first proposed by
Budker, the general consensus from the scientific community
was “brilliant idea, but unfortunately non-realistic.” Luckily,
Novosibirsk scientists did not listen to these predictions and
successfully constructed the first e-cooler around 1974 and
have built many more since then.
There is one more parallel between these two stories. Sir
John Adams had a unique combination of scientific and en­
gineering abilities, whereas Lev Landau once called Budker a
“relativistic engineer.” As it happens, the art of inventiveness
(TRIZ) originated from engineering.

11.3.2 Higgs, superconductivity and TRIZ

In the first chapter, we discussed the connection between the
TRIZ inventive principle of the nested doll (matreshka) and
Bryusov’s poem describing the world of an electron and par­
ticle detectors, which are arranged like a matreshka doll — a
system within another system.
The particle detectors are the devices that (in tandem
 inventions and innovations in science 231

with accelerators) helped to discover2 the Higgs boson —

the essential building block of the Standard Model of particle
physics.
A particular connection we would like to highlight is the
relationship between Higgs and superconductivity. In a re­
cent article,3 A. Pashkin and A. Leitenstorfer reminded us
that “... the theoretical proposal of the Higgs mechanism was
actually inspired by ideas from condensed matter physics ...
In 1958, Anderson discussed the appearance of a coherent
excited state in superconducting condensates with sponta­
neously broken symmetry... On page 1145 of this issue, Mat­
sunaga et al. report direct observation of the Higgs mode in
the conventional superconductor niobium nitride (NbN) ex­
cited by intense electric field transients.”
The above passage shows us once again that that the gen­
eral conclusion of TRIZ that “the same problems and solu­
tions appear again and again but in different disciplines” is
applicable to science, too.

11.3.3 Garin, matreshka and Nobel

One of the first inspirations and predictions of a device sim­
ilar to today’s laser may have appeared in a 1926 novel by
Aleksey Tolstoy, The Hyperboloid of Engineer Garin. In that
story, a device was described that was capable of producing
a ray of light of immense power. It is fascinating to note that
technical drawings were included in this novel — a signifi­
cant attraction for curious readers (even despite the fact that The focusing mirror in
Garin’s hyperboloid was
the drawings referred to non-existing materials).
made from “shamonite” —
The adventures described in Tolstoy’s novel were extraor­ an extremely durable mate­
dinary. This fictional and powerful ray was responsible for rial imagined by the writer.
many astounded remarks. One such was, “Can you imagine Fictitious and non-existing
what opportunities are opening now? Nothing in nature can in early 20th century, such
withstand the power of the beam of light — buildings, forts, a material can perhaps be
battleships, airships, rocks, mountains, the earth’s crust — created in the 21st century
everything could be penetrated, destroyed, cleaved with my thanks to the advent of new
beam.” These are the words Tolstoy put into the mouth of the engineered materials.
story’s protagonist — the engineer Garin.
The lasers developed later in the century4 did not adhere
to the design in The Hyperboloid of Engineer Garin, and were
luckily used for peaceful purposes.
In connection to lasers and TRIZ, we would like to men­
tion here one particular recent invention, stimulated emission
depletion microscopy (STED), which was developed by Stefan
W. Hell and his colleagues.5
The stimulated emission depletion microscopy allows for
2 Francois Englert and Peter W. Higgs, Nobel Prize in Physics, 2013.

3 A. Pashkin and A. Leitenstorfer, Science 345, 1121 (2014).

4 C. Townes, N. Basov and A. Prokhorov, Nobel Prize in Physics, 1964.

5 E. Betzig, S. W. Hell and W. E. Moerner, Nobel Prize in Chemistry, 2014.

232 unifying physics of accelerators, lasers and plasma

an increase of resolution in optical microscopes by a factor of

around five. It does this by applying two pulses of laser light
to an object under study — an excitation pulse and, shortly
afterwards, a de-excitation pulse.
A fluorescent dye introduced into the object is first excited
and then depleted with the second pulse of an appropriately
different wavelength. The key feature of this technique is the
usage of the depleting laser pulse with a special spatial pro­
file, i.e., its minimum intensity is located in the center, as il­
lustrated in Fig. 11.1.

FIGURE 11.1
Stimulated emission depletion microscopy (STED) and TRIZ in­
ventive principle of matreshka and system–antisystem. Excitation
laser pulse (a), de-excitation pulse (b) and remaining fluores­
cence (c). Improvement of resolution of a protein imaging due
to STED is shown qualitatively on the right.

The minimum intensity region of the second laser can be

several times smaller than the wavelength. Therefore, detect­
ing the remaining fluorescence spot will result in improving
the imaging resolution to values several times lower than the
laser wavelength.
Considering this invention within the framework we have
discussed in this book, we can also note a remarkable con­
nection to TRIZ. From the perspective of the theory of inven­
tive problem solving, the STED method is an illustration of
the use of the principle of system and antisystem (excitation
and de-excitation laser pulse), perhaps combined with the in­
ventive principle of the matreshka (one laser pulse is located
geometrically inside of the other pulse).
We have already observed examples where system­
antisystem and matreshka inventive principles combined to
create novel systems (e.g., dual force neutral solenoids). This
suggests that combinations of these or other inventive princi­
ples can be especially efficient in solving problems in various
scientific fields.
 inventions and innovations in science 233

11.4 Aiming for Pasteur quadrant

Science has always been a driver for the economy. This is a
commonly accepted statement — however, the mechanisms
of its impact are complicated. Still, their analysis is necessary,
and not only from a philosophical point of view, but also in
order to optimize the research priorities and define the strate­
gies for technological innovation.
One of the attempts to analyze the model for research and
technology transfer was done by Vannevar Bush, who, during
World War II, was instrumental in reorganizing the research
and science community according to the needs of that dif­
ficult time. Vannevar Bush’s 1945 report6 has since defined
post-war scientific policy in the United States and in many
other countries (supposedly, for decades to come).
In this report, Bush describes what would later be called a
one-dimensional or linear model for research and technology
transfer. Bush claims that research which is more basic is less
applied and vice versa (illustrated in Fig. 11.2). According
to Bush, applied research invariably drives out pure research
if the two are mixed, and therefore, basic research must be
completely isolated from considerations of use.

FIGURE 11.2
One-dimensional, linear model of research.

Correspondingly, the dynamic linear model of technology

transfer resembles a pipeline wherein government funding
stimulates basic research, which then in turn feeds applied
research, and ultimately results in technology and product
development, with society eventually benefiting from the
process (see Fig. 11.3).
These views of the relationship between basic science and

FIGURE 11.3
Dynamic linear model of technology transfer.
6 Science, the Endless Frontier: a Report to the President, by Vannevar Bush,
Director of the Office of Scientific Research and Development, July 1945.
234 unifying physics of accelerators, lasers and plasma

technological innovation have since then been analyzed, crit­

icized, and a new model has been developed.
The contradiction between these linear models and prac­
tice can be illustrated via the example of accelerator science
and technology. The invention of “strong focusing” in the
1950s was a revolutionary change in accelerator technology.
It enabled numerous applications. This invention may have
come about as a result of a pure fundamental interest — how­
ever, it was developed as a result of the pursuit of a certain
concrete goal, and was made possible due to certain technolo­
gies available at that time.

FIGURE 11.4
Revised dynamic model of research and technology transfer.

A new model of research and technology transfer was sug­

gested by Donald Stokes, who worked on the Advisory Com­
mittee on Research for the National Science Foundation.
In his report to the NSF, and in the book he subse­
“Research with consideration quently published,7 Donald Stokes argued against Bush’s
of use” is an effective way for linear model and introduced the notion of use-inspired re-
science to provide an impact search — “research with consideration of use” — which re-
on our economy and society. defined the paradigm of the relationship between basic sci­
ence and technological innovation. A revised dynamic model
suggested by Stokes is illustrated in Fig. 11.4.
7 Donald E. Stokes, Pasteur’s Quadrant — Basic Science and Technological
Innovation, Brookings Institution Press 1997.
 inventions and innovations in science 235

To portray his ideas, Donald Stokes suggested consider­

ing research on a two-dimensional plane, where the axes are
fundamental knowledge impact and consideration of use.
Donald Stokes’ plot is now called a Pasteur quadrant and
it is illustrated in Fig. 11.5.

FIGURE 11.5
Pasteur quadrant and accelerator science.

A characteristic example of a purely fundamental scien­

tific pursuit is the research works of Niels Bohr on the struc­
ture of nuclei, while another example of purely practical ac­
tivity is Thomas Edison’s developmental work on the fila­  


ments of light bulbs — as shown in Fig. 11.5.   

The quantitative assessments of these examples — or of 



other research projects placed on this graph — can be done  

  

by evaluating the number of either the academic papers or

patents, resulting from a particular research, as illustrated in
Fig. 11.6. 


Donald Stokes suggested, however, that an optimized ap­
proach should balance the fundamental pursuit of knowl­
edge with consideration of use, which is illustrated by the FIGURE 11.6
works of Louis Pasteur — indicated in Fig. 11.5 as a preferred The units of quantitative as­
direction. sessments of research in the
In the field of modern accelerator science and technology, Pasteur quadrant.
characteristic examples for two sections of the quadrant are:
colliders intended for the exploration of fundamental prop­
236 unifying physics of accelerators, lasers and plasma

erties of elementary particles (on one axis), and accelerator-

based devices for medical applications (on the other axis).
The criteria suggested by Donald Stokes are universal and
applicable to any scientific and technological area. By apply­
ing these criteria to accelerator science and technology, we
can conclude that the preferred direction — which best bal­
ances the fundamental pursuit of knowledge with consid­
eration of use — will be the direction to create novel light
sources. Efforts made in this direction could potentially pro­
duce scientific instruments applicable to the investigation of
protein structures or materials, which may also be almost di­
rectly applicable to the creation of new medicines or metals
with controllable properties.
This analysis also shows that — as in many other disci­
plines — accelerator science and technology, in synergy with
lasers and plasma physics, can truly span the entire range of
directions — from pure fundamental science to pure applied
development.
The research and technology innovation model is indeed
not just linear, but at least two-dimensional, as Stokes out­
lined in his revised dynamic model. In the 21st century, the
driving forces of technological innovation — as well as the
global pact between science and society — are different than
they were in the middle of the last century. The revised cri­
teria — as illustrated in this section via the example of ac­
celerator science — are universal, and can be applied to any
discipline, which can help us to optimize the impact of our
research investments on our economy and society.

11.5 How to cross the Valley of Death

Powerful beams of light, heat rays — now called lasers — are
features of H.G. Wells’ and Alexey Tolstoy’s science fiction
stories that have excited many generations of future inven­
tors and scientists. These stories thrilled crowds of kids (and
adults) as they rambled along back streets dreaming of hav­
ing lasers in their pockets.
The curiosity and imagination of the younger generation
is the fuel that enables the development of our civilization.
The challenge for governments, educational institutions and
societies is to understand how to nurture and later harness
these attributes.
The first visible light lasers (with a wavelength of light
of about half a micrometer) were typically big when they
were created half a century ago, even huge, and certainly
not pocket size. Now, miniature lasers are in CD players,
bar-scanners in shops, laser-pointers — practically every­
where. However, lasers with a much shorter wavelength, in
the Angstrom range (light in the X-ray spectrum), have only
just become available. Called free electron lasers, they are a
 inventions and innovations in science 237

kilometer long. Due to their short wavelength, X-rays are al­

ready indispensable for the analysis of protein structure, syn­
thetic molecules, new materials and many other objects. The
size and affordability of such X-ray lasers are major obstacles
that hinder the widespread use of compact X-ray lasers. Cre­
ating compact X-ray lasers is the challenge that accelerator
science now needs to confront.
Particle accelerators have already impacted many areas
of our lives with their medical and industrial uses, as well
as with their help in creating research instruments. Tens of
millions of patients receive accelerator-based diagnoses and
treatments each year, worldwide. The total annual market
value for all products that are treated or inspected by acceler­
ators is more than $500B.8 Approximately 30% of the Nobel
prizes in physics, as well as many in other areas, are directly
connected to the use of accelerators.9
The ideas that enabled the use of accelerators in everyday
life and industry were developed decades ago. New ideas will
be essential for ensuring the future impact of this field.
Conventional accelerators, no matter how advanced they
may be, are primarily based on the acceleration of particles
in cavities — metal vessels shaped to resonate and create ac­
celerating fields. The ability of metals to tolerate high elec­
tromagnetic fields is intrinsically limited. However, an accel­
erating wave can be created when gas is ionized and excited
by an intense beam of particles or by a laser pulse, becoming
plasma. Plasma is an indestructible medium and is able to
withstand a thousand-times-higher accelerating gradient.
Accelerator science and technology is on the edge of a
breakthrough brought on by synergy with laser and plasma
physics. The most immediate outcome that this synergy will
enable is the creation of novel, compact X-ray lasers and
light sources. The direct collision of beam and laser light
also opens up another opportunity for the creation of X-ray
sources via the use of the Compton effect (when visible light
photons are reflected from a relativistic electron beam and
thereby decrease their wavelength down to Angstrom levels).
Science is indeed the driver of our civilization’s progress.
However, the journey from initial ideas and experimental
demonstrations to widespread commercial applications is
long and difficult. Various studies performed in different
countries have all found a gap, a so-called “Valley of Death”
in technology transfer. It is difficult to bridge the middle
range of the technological readiness of ideas. On one end, the
research institutions are usually not positioned to develop
ideas into commercial applications, while on the other end,
the risk is often too high for industry to pick up ideas that
are too fresh and undeveloped.
8 Accelerators for America’s Future, Department of Energy, 2009.

9 E. Haussecker and A. Chao, Physics in Perspective, 2011.

238 unifying physics of accelerators, lasers and plasma

The challenge originates from the different motivations,

methods and timescales of three key players: academic in­
stitutions, industry and investors.10 Their corresponding
aims and motivations — the front-end fundamental scien­
tific results, the development of commercial devices in the
foreseeable future, and optimization of investments versus
risk/return factors — are often incompatible (Fig. 11.7).

FIGURE 11.7
Academia–industry–investor puzzle caused by different motiva­
tions of the three participating groups.

Accelerators in synergy with lasers and plasma may, in fact,

offer a solution for the academia–industry–investor puzzle
through simultaneous, parallel work on a portfolio of three
different types of compact X-ray light sources (Fig. 11.8).

FIGURE 11.8
Working on a portfolio of compact X-ray light sources can help
in crossing the “Valley of Death” between accelerator science
and technological innovation.
10 A government can be an investor.
 inventions and innovations in science 239

FIGURE 11.9
A solution to the academia–industry–investor puzzle — work on
three designs of compact X-ray sources.

A solution to the academia–industry–investor puzzle in­

volving work on three designs of compact X-ray sources is
illustrated in Fig. 11.9.
Compton X-ray sources — the first type of source in the
solved puzzle — are now actively being developed and are a
lower-risk investment for industrial use. Yet a more challeng­
ing, but promising, Compton source requires superconduct- How to solve the academia–
ing acceleration to allow for a much higher electron beam cur- industry–investor puzzle? Work-
rent and X-ray brightness. This second option resides in the ing on a portfolio of compact
middle of the range for both the projected availability and light sources may give an an-
risk/return. swer.
The most challenging, but also the most promising source
is an X-ray source based on laser plasma acceleration — ini­
tially a betatron source, and ultimately, perhaps in less than a
decade, a free electron laser. If we properly schedule the rel­
ative progression of the different stages of research and de­
velopment for these three types of X-ray sources, we could
balance the typical risks associated with the development of
innovative products, and the opportunities they offer.
Such work on compact X-ray sources fits into to the pre­
ferred direction of the Pasteur quadrant. Moreover, research
and development in this direction may result in the creation
240 unifying physics of accelerators, lasers and plasma

of instruments that every university will aspire to have. The

ability to obtain a compact X-ray laser would revolutionize
contemporary science and technology yet again — not unlike
how the spread of near-visible light lasers impacted science
and industry during the 20th century.

11.6 How to learn TRIZ in science

In this second-to-last section, I would like to convey to the
reader my personal impressions of the best ways to learn (and
also teach) the methodologies of inventiveness in the science
departments of universities. These impressions are based on
the development of training programs involving TRIZ at the
John Adams Institute for Accelerator Science — JAI — at Ox­
ford University, Imperial College and Royal Holloway Uni­
versity of London.
The JAI is a center of excellence in the UK for advanced
and novel accelerator technology. We focus on researching
and developing accelerator science and techniques, as well
as graduate training. Our cohort of PhD students numbers
around three dozen, and on average we educate about six
PhD experts per year. Like any other research organization
or graduate school, we are motivated to turn the results of
our research into having a positive societal impact, as well as
endeavoring to find ways to be more inventive and innova­
tive. For the latter, we recently (several years ago) started to
look at TRIZ and developed a specific, scientifically oriented
way to introduce this method to our PhD students.
Let us briefly review TRIZ and its connection to science
graduate schools. TRIZ is the methodology of inventiveness
that was specifically developed for engineering fields. It was
created during the second half of the 20th century and grad­
ually became one of the most powerful tools in the industrial
world. TRIZ has become, according to Forbes, the bedrock of
innovations of such companies as Samsung, as well as many
others.
While the TRIZ method of industrial inventiveness was
originally created for engineering, this methodology is uni­
versal and can also be applied to science. However, experi­
ence shows that knowledge of TRIZ is nearly nonexistent in
the scientific departments of western universities.
Moreover, it is not rare to hear about unsuccessful at­
tempts to introduce TRIZ into graduate courses of scientific
departments of universities. Indeed, in many or most of these
cases, the apparent reason for the failure was that the canon­
ical version of TRIZ was introduced to science PhD students
in the same way that TRIZ is taught to engineers in indus­
trial companies. This appears to be a mistake, seeing as sci­
ence students are rightfully more critically minded and justi­
fiably sceptical about overly prescriptive step-by-step meth­
 inventions and innovations in science 241

ods. (As we have seen above, TRIZ involves finding a pair

of contradicting parameters in a problem, which then — us­
ing the TRIZ inventive tables created by TRIZ teams, based
on the analysis of hundreds of thousands of past inventions
— immediately leads to selecting just a few suitable inven­
tive principles, narrowing down the choice and resulting in a
much faster solution to a problem.)
The approach of introducing TRIZ to JAI graduate stu­
dents is different, and takes into account lessons learned by
its predecessors. Instead of teaching our graduate students
the ready-to-use methodology, we are effectively taking them
through the process of recreating parts of the TRIZ method­
ology by analyzing various inventions and discoveries from
scientific disciplines, showing that these inventive principles
can be efficiently applied to science. Moreover, in the process
of this development, we often found that additional inven­
tive principles, more suitable for scientific disciplines, can be
introduced and added to standard TRIZ — we call this ex­
tension Accelerating Science TRIZ (the play on words is now
apparent — the word “accelerating” is not referring to accel­
erators any more, but highlights that TRIZ can help to boost
science).
The approach to teaching TRIZ described above has now
been successfully introduced to JAI graduate students and
was also successfully implemented at a course in the US­
PAS (US Particle Accelerator School), and has also been intro­
duced in the JUAS (European Joint Universities Accelerator
School) and APPEAL school (a one-day to one-week course
for high-school teachers that we conduct every year as part of
our outreach activity).
TRIZ methodology is another way to look at the world.
Combined with science it creates a powerful and eye-opening
amalgam of science and inventiveness. This methodology is
particularly helpful for building bridges of understanding
between completely different scientific disciplines, and so is
also naturally useful to educational and research organiza­
tions that endeavor to break barriers between disciplines.
Ultimately, my recommendations on learning and teach­
ing TRIZ in universities can be summarized as follows. In­
troducing TRIZ courses/lessons to university students is
only the first step. However, while teaching these TRIZ
courses/lessons, avoid the canonical, ready-to-use version
of TRIZ. Instead, take the students through the process of
proactively adapting TRIZ for science. Arrange a sequence of
relatively short TRIZ lessons throughout the entire duration
of education in the university, so that students can gradually
build up their ability to use TRIZ methods in order to build
bridges of understanding between different disciplines that
they will interact with during their education and research.
242 unifying physics of accelerators, lasers and plasma

11.7 Let us be challenged

Many of the examples of TRIZ-like inventions in science that
we considered in this book have already been made, and we
only analyzed the applicability of TRIZ post-factum.
It is natural to wonder whether TRIZ and AS-TRIZ can ac­
tually help to inspire and create new scientific inventions and
innovations, especially in regards to projects that continue to
manifest many unsolved obstacles.
One example of such a project is a circular collider cur­
rently being considered as a successor to the Large Hadron
Collider at CERN — the Future Circular Collider (FCC).11 The
FCC project is looking to create a 100 km tunnel infrastruc­
ture in order to house a 100 TeV CM proton–proton collider
(the e+ e− and proton-e options are also being considered).
This project has many scientific and technical tasks and
challenges that need to be solved. Notably, one issue is that
the synchrotron radiation at these high energies starts to in­
fluence the proton beams. The total energy in each circulating
proton beam is expected to exceed 8 GJ, which is equivalent
to a kinetic energy of an Airbus-380 flying at 720 km/h. Not
only does such a beam need to be handled safely in the bend­
FIGURE 11.10 ing magnets, but it also needs to be focused in the interac­
The FCC beam energy will be tion region to a micron spot size — which is practically the
comparable to that of an air­ equivalent of literally having to pass through a needle’s eye
plane, while the beam will It remains to be seen if TRIZ and AS-TRIZ methodology
need to be focused at the can be applied to such a large-scale project as the FCC, as it
interaction point to micron- brings a whole array of new, difficult and exciting challenges
scale size — an analogy with to the table. Nonetheless, it is certainly a project that can only
a plane passing through the flourish with the application of our knowledge and inventive­
needle’s eye. ness.

11 Future Circular Collider — http://cern.ch/fcc

 inventions and innovations in science 243

EXERCISES
11.1 Mini-project.
A proposal for a Higgs and top factory is based on the de­
sign of an electron-positron collider (175 GeV for each beam)
in a tunnel with a 50 km circumference. Assuming that the
maximum power delivered by the RF system to each beam
is limited to 50 MW, estimate the maximum current that can
be stored in such a collider. Make an assumption about the
focusing system at the interaction point, and evaluate beam
emittances and the collider’s luminosity.
11.2 Analyze inventions or discoveries using TRIZ and AS-TRIZ.
A liquid anode X-ray tube is a contemporary technology that
increases the photon flux. Analyze this technology in terms of
the TRIZ and AS-TRIZ approach, identifying a contradiction
and a general inventive principle that was used (could have
been used) in this invention.
11.3 Analyze inventions or discoveries using TRIZ and AS-TRIZ.
A liquid jet or liquid target is often used in high energy
physics experiments, particularly for the production of an­
tiprotons. A liquid jet allows withstanding the power of the
incident proton beam. Analyze this technology in terms of
the TRIZ and AS-TRIZ approach, identifying a contradiction
and a general inventive principle that were used (could have
been used) in this invention.
11.4 Analyze inventions or discoveries using TRIZ and AS-TRIZ.
Analyze and describe scientific or technical inventions de­
scribed in this book in terms of the TRIZ and AS-TRIZ ap­
proaches, identifying a contradiction and an inventive prin­
ciple that were used (could have been used) for these inven­
tions.
11.5 Developing AS-TRIZ parameters and inventive principles.
Based on what you have learned in this book, discuss and
suggest the possible additional parameters for the AS-TRIZ
contradiction matrix, as well as the possible additional AS­
TRIZ inventive principles.
 Final Words

Writing this book was an absolutely wonderful and enjoy­

able experience, worth every hour of long nights and week­
ends spent working on it. It has been about two years since I
first had the inkling to write it, although the actual writing,
from the first word of the first chapter to the last sentence,
has taken about nine months.
The book is different from many (if not all) textbooks
on accelerators, lasers and plasma because firstly, it looks at
these areas together, and secondly, it mixes them with the
methodology of inventiveness — the glue that pulls these ar­
eas close to each other just as gluons connect the quarks in a
Λ-Barion, as illustrated below.

There are still many scientific topics I would have wanted

to discuss and include in this book, but I had to resist this
temptation, primarily based on the expectation that feedback
from the readers will be invaluable for the work on possible
future editions.

Andrei Seryi — Oxford, March 5, 2015

245
 Bibliography

Altshuller, G. (1999). Innovation Algorithm: TRIZ, systematic innovation and technical cre­
ativity (first ed.). Worcester, MA: Technical Innovation Center, Inc.
Amaldi, U. et al. (2013). Elementary Particles, Volume 21: Subvolume C: Accelerators and
Colliders , Group I: Elementary Particles, Nuclei and Atoms. Berlin, Heidelberg, New
York: Springer-Verlag.
Badelek, B. et al. (2004). The photon collider at TESLA. International Journal of Modern
Physics A 19(30), pp.5097–5186. New Jersey, London, Singapore, Beijing, Shanghai,
Hong Kong, Taipei, Chennai: World Scientific Publishing Co. Pte. Ltd.
Baklakov, B. and others (1991). Investigation of seismic vibrations and relative displacements
of linear collider VLEPP elements. In Proc. of 1991 IEEE PAC. p.3273. San-Francisco, CA:
Proc. of 1991 IEEE PAC.
Balakin, V. (1991). Travelling Focus Regime for Linear Collider VLEPP. In Proc. 7th ICFA
Workshop on Beam Dynamics. Los Angeles, CA: Proc. 7th ICFA Workshop on Beam
Dynamics.
Balakin, V., Nokhatsky, A. and Smirnov, V. (1983). VLEPP: Transverse Beam Dynamics. In
Proceedings of the 12th Int. Conf. on HEACC. Batavia, IL: FERMILAB Print.
Bellan, P. (2006). Fundamentals of Plasma Physics. New York: Cambridge University Press
Inc.
Blumenfeld, I. et al. (2007). Energy doubling of 42 GeV electrons in a metre-scale plasma
wakefield accelerator. Nature 445, pp.741–744. London: Macmillan Publishers Limited.
Borghesi, M. et al. (2006). Fast Ion Generation by High-Intensity Laser Irradiation of Solid
Targets and Applications. Science and Technology 49(3), pp.412–439. La Grange Park,
IL: American Nuclear Society.
Borghesi, M. et al. (2008). Laser-driven proton acceleration: source optimization and radio­
graphic applications. Plasma Phys. Control. Fusion (50, 124040). Bristol, Philadelphia,
Beijing, Tokyo: IOP Publishing Ltd.
Bosco, A. et al. (2008). A Two-Dimensional Laser-Wire Scanner for Electron Accelerators.
Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spec­
trometers, Detectors and Associated Equipment 592(3), pp.162–170. Elsevier.
Bourgeois, N., Cowley, J. and Hooker, S. M. (2013). Two-Pulse Ionization Injection into
Quasilinear Laser Wakefields. Phys. Rev. Lett. 111(15, 155004). Ridge, NY: American
Physical Society.
Boyd, T. and Sanderson, J. (2003). The Physics of Plasmas. New York: Cambridge University
Press Inc.
Brinkmann, R. et al. (2001). A low emittance, flat-beam electron source for linear colliders.
Phys. Rev. ST Accel. Beams 4(5, 053501). Ridge, NY: American Physical Society.

247
248 unifying physics of accelerators, lasers and plasma

Budker, G.I. et al. (1976). Experimental studies of electron cooling. Particle Accelerators 7,
pp.197–211. United kingdom: Gordon and Breach, Science Publishers Ltd.

Burrows, M. and Sutton, G. (2013). Interacting Gears Synchronize Propulsive Leg Movements
in a Jumping. Science 341(6151), pp.1254–1256. Washington, DC: American Association
for the Advancement of Science.

Bush, V. (1945). Science, the Endless Frontier: a Report to the President. Science Educa­
tion 29(4), p.218. Washington, D. C.: U.S. Government Printing Office.

Chao, A. et al. (2013). Handbook of Accelerator Physics and Engineering (second ed.). New
Jersey, London, Singapore, Beijing, Shanghai, Hong Kong, Taipei, Chennai: World Sci­
entific Publishing Co. Pte. Ltd.
Chen, F. (1974). Introduction to Plasma Physics. New York and London: Plenum Press.

Clark, E. et al. (2000). Measurements of Energetic Proton Transport through Magnetized

Plasma from Intense Laser Interactions with Solids. Phys. Rev. Lett. 84(18), p.670. Ridge,
NY: American Physical Society.

Daido, H. et al. (2012). Review of laser-driven ion sources and their applications. Rep. Prog.
Phys. (75, 056401). Bristol, Philadelphia, Beijing, Tokyo: IOP Publishing Ltd.
Danilov, V. and Nagaitsev, S. (2010). Nonlinear accelerator lattices with one and two ana­
lytic invariants. Phys. Rev. ST Accel. Beams 13(084002). Ridge, NY: American Physical
Society.
Davidson, R. (2001). Physics of Nonneutral Plasmas. New Jersey, London, Singapore, Bei­
jing, Shanghai, Hong Kong, Taipei, Chennai: World Scientific Publishing Co. Pte. Ltd.
Delayen, J. R. (2012). Superconducting spoke cavities for electrons and high-velocity proton
linacs. In Proceedings of LINAC2012, 03 Technology, 3A Superconducting RF, TH1A03,
pp.758–762. Tel Aviv, Israel: Proceedings of LINAC2012.
Dendy, R. (1990). Plasma Dynamics. New York: Oxford University Press Inc.
Derbenev, Y. (1998). Adapting Optics for High Energy Electron Cooling. In University of
Michigan Report No. UM HE 98-04. Ann Arbor, MI: University of Michigan Report.
Derbenev, Y. et al. (1995). Microbunch Radiative Head-Tail Interaction. In Tesla-FEL Report.
Hamburg, Germany: DESY Print.
Diver, D. (2013). Plasma Formulary for Physics, Astronomy, and Technology (second ed.).
Weinheim, Germany: WILEY-VCH Verlag.
Edwards, D. et al. (2004). An Introduction to the Physics of High Energy Accelerators (second
ed.). Weinheim, Germany: WILEY-VCH Verlag.

Eliezer, S. (2002). The Interaction of High-Power Lasers with Plasma. Bristol, Philadelphia,
Beijing, Tokyo: IOP Publishing Ltd.

Esirkepov, T. et al. (2004). Highly Efficient Relativistic-Ion Generation in the Laser-Piston

Regime. Phys. Rev. Lett. 92(7, 175003). Ridge, NY: American Physical Society.

Gabor, D. (1947). A Space-Charge Lens for the Focusing of Ion Beams. Nature 160, pp.89–90.
London: Macmillan Publishers Limited.
 bibliography 249

Gadd, K. (2011). TRIZ For Ingeneers: Enabling Inventive Problem Solving. Chichester, West
Sussex: John Wiley & Sons, Ltd., Oxford Creativity.

Gibbon, P. (2005). Short pulse laser interactions with matter. An introduction. London: Im­
perial College Press.

Haberberger, D. et al. (2012). Collisionless shocks in laser-produced plasma generate monoen­

ergetic. Nature Physics 8(1), pp.95–99. London: Macmillan Publishers Limited.

Haseroth, H. et al. (1996). Laser Ion Source Development for Heavy Ions. In Proceedings
of the XVIII International linear Accelerator Conference Linac96, v.2, WE202. pp.570–
574. Geneva, Switzerland: CERN, European Organization for Nuclear Research.
Haussecker, E. and Chao, A.W. (2011). The Influence of Accelerator Science on Physics Re­
search. Physics in Perspective (PIP). 13(2), pp.146–160. Bristol, Philadelphia, Beijing,
Tokyo: IOP Publishing Ltd.

Herr, W. (2013). New Tools for Non Linear Dynamics I, II. In The CERN Accelerator School
2013, Trondheim, Norway. Geneva, Switzerland: CERN, European Organization for
Nuclear Research.

Hidding, B.B. et al. (2012). Beyond injection : Trojan horse underdense photocathode plasma
wakefield acceleration. In Advanced Accelerator Concepts. AIP Conference Proceedings.,
pp.570–575. Melville, NY: AIP.

Hooker, S. et al. (2014). Multi-Pulse Laser Wakefield Acceleration: A New Route to Efficient,
High-Repetition-Rate Plasma Accelerators and High Flux Radiation Sources. J. Phys. B 47,
(234003). Bristol, Philadelphia, Beijing, Tokyo: IOP Publishing Ltd.

Hooker, S. and Webb, C. (2010). Lasers physics (Oxford Master Series in Physics). New York:
Oxford University Press Inc.

Huang, Zh. and Ruth, R.D. (1998). Laser-electron storage ring. Phys. Rev. Lett. 80, pp.976–
979. Ridge, NY: American Physical Society.

Jackson, J. (1999). Classical Electrodynamics (third ed.). Hoboken, NJ: John Wiley & Sons,
Ltd.

Jaroszynski, D. et al. (2009). Laser-Plasma Interactions. Boca Raton, FL: CRC Press, Taylor
Francis Group.

Kneip, S. et al. (2011). X-ray phase contrast imaging of biological specimens with femtosecond.
Applied Physics Letters 99. Melville, NY: AIP Publishing LLC.
Landau, L. (1946). On the vibration of the electronic plasma. JETP (16), p.574. Russia:
International Academic Publishing Company (IAPC) “Nauka/Interperiodica”.

Lee, S. (2012). Accelerator Physics (third ed.). New Jersey, London, Singapore, Beijing,
Shanghai, Hong Kong, Taipei, Chennai: World Scientific Publishing Co. Pte. Ltd.
Leemans, W. et al. (2006). GeV electron beams from a centimetre-scale accelerator. Nature
Physics 2, pp.696–699. London: Macmillan Publishers Limited.
Leemans, W. et al. (2014). Multi-GeV Electron Beams from Capillary-Discharge-Guided Sub­
petawatt Laser Pulses in the Self-Trapping Regime. Phys. Rev. Lett. 113, (245002). Ridge,
NY: American Physical Society.
250 unifying physics of accelerators, lasers and plasma

Macchi, A. et al. (2013). Ion acceleration by superintense laser-plasma interaction. Rev. Mod.
Phys. 85, pp.751–793. Ridge, NY: American Physical Society.

Maksimchuk, A. et al. (2000). Forward Ion Acceleration in Thin Films Driven by a High-
Intensity Laser. Phys. Rev. Lett. 84(18), pp.4108–4111. Ridge, NY: American Physical
Society.

Mangles, S. et al. (2004). Monoenergetic beams of relativistic electrons from intense laser-
plasma interactions. Nature 431, pp.535–538. London: Macmillan Publishers Limited.

Mendonca, J. (2001). Theory of Photon Acceleration. Bristol, Philadelphia, Beijing, Tokyo:

IOP Publishing Ltd.
Mourou, G. et al. (2013). The future is fibre accelerators. Nature Photonics (7), pp.258–261.
London: Macmillan Publishers Limited.

Najmudin, Z. (2014). Laser Driven Systems. In The CERN Accelerator School 2014,
Geneva, Switzerland. Geneva, Switzerland: CERN, European Organization for Nuclear
Research.

Palmer, C.A. et al. (2011). Monoenergetic Proton Beams Accelerated by a Radiation Pressure
Driven . Phys. Rev. Lett. 106(1, 014801). Ridge, NY: American Physical Society.
Palmer, R.B. (1988). Energy Scaling, Crab Crossing and the Pair Problem. SLAC-PUB-4707.
Menlo Park, CA: SLAC–PUB.
Parker, B. et al. (2007). The Superconducting Magnets of the ILC Beam Delivery System.
SLAC-PUB-12832. Menlo Park, CA: SLAC–PUB.
Pashkin, A. and Leitenstorfer, A. (2014). Particle physics in a superconductor. Sci­
ence 345(6201), pp.1121–1122. Washington, DC: American Association for the Ad­
vancement of Science.
Pozimski, J. and Aslaninejad, M. (2013). Gabor lenses for capture and energy selection of laser
driven ion beams in cancer treatment. Laser and Particle Beams 31, pp.723–733. New
York: Cambridge University Press Inc.
Qiao, B. et al. (2009). Stable GeV Ion-Beam Acceleration from Thin Foils by Circularly Po­
larized Laser Pulses. Phys. Rev. Lett. 102(14, 145002). Ridge, NY: American Physical
Society.

Qiao, B. et al. (2010). Radiation-Pressure Acceleration of Ion Beams from Nanofoil Targets: The
Leaky Light-Sail Regime. Phys. Rev. Lett. 105(15, 155002). Ridge, NY: American Physical
Society.
Radovinsky, A. et al. (2013). Variable Energy Acceleration in a Single Iron-Free Synchrocy­
clotron. In Variable Energy Acceleration in a Single Iron-Free Synchrocyclotron. Cam­
bridge, MA: Plasma Science and Fusion Center Massachusetts Institute of Technology
PSFC/RR-13-9.

Raimondi, P. et al. (2008). Suppression of beam-beam resonances in crab waist collisions. In

Proceedings of EPAC08, Genoa, Italy, pp.2620–2622. Proceedings of EPAC08.

Raimondi, P. and Seryi, A. (2001). Novel Final Focus Design for Future Linear Colliders. Phys.
Rev. Lett. 86(17, 3779). Ridge, NY: American Physical Society.
 bibliography 251

Saldin, E. et al. (2010). The Physics of Free Electron Lasers (Advanced Texts in Physics). Berlin
and Heidelberg, Germany: Springer-Verlag.
Schlenvoigt, H.-P. et al. (2008). A compact synchrotron radiation source driven by a laser-
plasma wakefield accelerator. Nature Physics 4, pp.130–133. London: Macmillan Pub­
lishers Limited.
Seryi, A. and Napoly, O. (1996). Influence of ground motion on the time evolution of beams in
linear colliders. Phys. Rev. E 53(5, 5323). Ridge, NY: American Physical Society.
Sessler, A. and E. Wilson (2014). Engines of Discovery. A Century of Particle Accelerators
(Revised and Expanded Edition ed.). New Jersey, London, Singapore, Beijing, Shanghai,
Hong Kong, Taipei, Chennai: World Scientific Publishing Co. Pte. Ltd.
Shiltsev, V. et al. (1999). Considerations on compensation of beam-beam effects in the Tevatron.
Phys. Rev. ST Accel. Beams 2(071001). Ridge, NY: American Physical Society.
Shiltsev, V. (2010). Observations of Random Walk of the Ground in Space and Time. Phys.
Rev. Lett. 104(23, 238501). Ridge, NY: American Physical Society.
Silva, L. et al. (2004). Proton Shock Acceleration in Laser-Plasma Interactions. Phys. Rev.
Lett. 92(1, 015002). Ridge, NY: American Physical Society.
Snavely, R. et al. (2000). Intense High-Energy Proton Beams from Petawatt-Laser Irradiation
of Solids. Phys. Rev. Lett. 85(14), p.2945. Ridge, NY: American Physical Society.
Spence, D.J. and Hooker, S.M. (2001). Investigation of a hydrogen plasma waveguide. Phys.
Rev. E 63(1,015401(R)). Ridge, NY: American Physical Society.
Strickland, D. and Mourou, G. (1985). Compression of amplified chirped optical pulses. Optics
Communications 56(3), pp.219–221. Elsevier.
Stupakov, G. (2009). Using the Beam-Echo Effect for Generation of Short-Wavelength Radia­
tion. Phys. Rev. Lett. 102(7, 074801). Ridge, NY: American Physical Society.
Variola, A. et al. (2010). ThomX - Conceptual Design Report. In ThomX - Conceptual Design
Report, pp.1–136. LAL RT 09/28, SOLEIL/SOU-RA-2678.
Wang, J.W. and Loew, G.A. (1997). Field Emission and RF Breakdown in High-Gradient Room-
Temperature Linac Structures. SLAC-PUB-7684. Menlo Park, CA: SLAC–PUB.
Wiedemann, H. (2007). Particle Accelerators Physics (third ed.). Berlin and Heidelberg,
Germany: Springer-Verlag.
Wilson, E.J.N. (2001). An Introduction to Particle Accelerators. New York: Oxford University
Press Inc.
Wilson, E.J.N. (2011). Sir John Adams: his legacy to the world of particle accelerators. John
Adams Memorial Lecture, 2009. Geneva, Switzerland: CERN, European Organization for
Nuclear Research.
Yin, L. et al. (2011). Three-Dimensional Dynamics of Breakout Afterburner Ion Accelera­
tion Using High-Contrast Short-Pulse Laser and Nanoscale Targets. Phys. Rev. Lett. 107(4,
045003). Ridge, NY: American Physical Society.
Yu, T.-H. et al. (2010). Stable Laser-Driven Proton Beam Acceleration from a Two-Ion-Species.
Phys. Rev. Lett. 105(6, 065002). Ridge, NY: American Physical Society.
252 unifying physics of accelerators, lasers and plasma

Zholents, A. and Zolotorev, M. (1996). Femtosecond X-Ray Pulses of Synchrotron Radiation.

Phys. Rev. Lett. 76(6), p.912. Ridge, NY: American Physical Society.
 Index

Aberrations, 38, 40, 58, Beam break-up instability, 140–142, 147, 159,
68–70, 169, 218, 204 161, 226, 239
219 BBU instability, 204, 205 Brilliance, 52, 53, 54, 129,
Absorption imaging, 134 Beam collector, 23 130, 133, 135, 158,
Accelerating cavity, 17, 63, Beam laser slicing, 198 159
77, 80, 88, 94, 204, Beam-driven acceleration, Bryusov’s electron, 1, 13,
205 123 14, 230
Accelerating gradient, 3, 4, Beamstrahlung, 202, 224 Bubble chamber, 14
17, 80, 89, 101, Beta function, 27, 38, 51, Budker, 7, 72, 230
105–107, 116, 123, 68, 219, 220 Bunch, 2, 16, 62, 94, 95,
141, 237 Betatron, 5, 6, 128 99, 107, 116, 120,
Accelerating structures, Betatron acceleration, 2, 121, 123, 135, 139,
16, 81, 90, 105, 200, 63 140, 152, 155, 157,
205 Betatron motion, 25, 26, 159, 161, 174, 185,
Accelerator-driven 33, 35–37, 72, 100 186, 188, 190–192,
systems, 73 Betatron phase, 27, 38 196–198, 200,
Acousto-optic effect, 193 Betatron radiation, 57, 73, 202–205, 211, 215,
Alpha function, 27 118, 119–121, 161 220, 221
Altshuller, 9 Betatron radiation sources, Bunch compressor, 72,
Alvarez linac, 79 118, 134, 142, 161, 125, 130, 159, 186,
Analogies, 9, 34, 69, 81, 239 187, 192, 197
162, 216 Betatron tune, 38, 39, 99, Bunching, 80, 91

Apoptosis mechanism, 170 100, 206, 215

Art of inventiveness, 1, 9, Bethe equation, 168 C-band, 101

17, 230 BNS damping, 203, 204, Capillary channel, 117,

AS-TRIZ, 15, 16, 17, 19, 205 118

41, 56, 64, 66, 71, Bohr radius, 108 Carcinogenesis, 170
73, 74, 103, 106, Booster, 131, 132 Cathode poisoning, 23
125, 142, 163, 183, Boundary conditions, 82, Center of mass energy, 101
224–227, 242, 243 83, 85 Center of mass reference
Accelerating Science Bragg conditions, 133 frame, 137
TRIZ, 15, 226 Bragg diffraction, 193 CERN, 7, 59, 69, 72, 77,
ATL law, 209 Bragg peak, 165, 166, 167, 230, 242
Atomic fields, 108 168 Cesiation in ion source, 59
Atomic intensities, 109 Brainstorming, 8, 9, 17 Charge exchange, 6, 76, 77
Avalanche, 62, 192 Break-out afterburner Charge-exchange
acceleration, 181 injection, 212
Breakdown, 17, 62, 79, 89, Chicane, 70, 186, 187, 196,
Bandwidth, 4, 52, 53, 73, 90 199
130, 133, 195 Bremsstrahlung, 167, 170 Child-Langmuir law, 58
Barrier suppression Brightness, 4, 7, 15, 51–53, Chirped pulse
ionization, 110, 110 129, 130, 134, 135, amplification, 58,
Base excision repair, 170 63, 65, 105, 228

253
254 Index

CPA, 66, 109, 112, 113,

Cooling time, 45, 46, 56,
Drift tubes, 6, 79, 79, 89

176, 227–229
130
Dubna’s

Chromatic aberrations, 69,

Coulomb interaction, 167
synchrophasotron,

70
Coupling, 3, 37, 38, 39, 50,
7, 69, 230

Chromaticity, 38, 39, 40,

90, 93, 94, 130, 152,
Dudnikov ion source, 59

58, 68, 206, 207,

220

Echo-enabled harmonic

217–219
Courant–Snyder invariant,

generation, 199

Chromosomes, 170
35, 36

EEHG, 199

Cloud chamber, 14
Crab cavity, 221, 223

Edge focusing, 30, 175

Cockcroft–Walton
Crab waist collision, 7

Electrical breakdown, 79

generator, 76
Crabbed collisions, 221

Electro-optic effect, 193

Coherent laser pulse

Critical density, 57, 68

Electromagnetic spectrum,

combination, 213
Critical surface, 68

127

Coherent synchrotron
Cross section, 4, 136–138,

Electron cooling, 7, 72,

radiation, 160, 185,

140

214, 215, 216, 230

188, 189–191
Crown glass, 70

Electron lens, 214, 215,

CSR, 160, 188, 189–192,

CSR instability, 197, 198

216

197, 206
Curvilinear coordinate

Electrostatic acceleration,

Collective acceleration, 7
system, 26

2, 6, 63

Collective oscillations, 205

Cyclotron, 5, 6, 124, 165,

EM wave, 2, 63, 68, 81, 82,

Collider, 1, 2, 5, 7, 55, 73,

171, 172, 173

83, 88, 94, 108, 135,

77, 101, 200, 201,

136, 148–150, 152,

214, 217, 219, 224,

Dalton, 170
153, 155, 156

235, 242, 243

Debye sheath, 178
Emittance, 3, 4, 7, 15, 43,

Collimation, 139, 165,

Decoherence, 197, 204,
46, 49–51, 53, 54,

172, 183, 198

205, 207, 207
59, 72, 99, 121, 129,

Collisionless damping,
Demagnification, 31, 32
131, 140, 157, 159,

205
Dephasing, 117, 122, 123
161, 162, 178, 188,

Compact radiation
Dephasing length, 123
211, 214, 223, 226

sources, 121, 161

Depletion, 117
Endonucleases, 170

Compton backscattering,
Differential cross section,
Energy chirp, 186, 188

136, 137
136, 137
Energy recovery, 140

Compton light source,

Diffraction, 117
Engineer Garin, 231

139, 140
Diffraction-limited source,
Engineered materials, 231

Compton scattering, 135,

54, 56, 131, 160
Envelope, 27, 34, 83, 107

136, 137–139, 166

Diode laser, 61, 62
Equilibrium emittance, 50,

Compton wavelength, 22
Direct ionization, 110
130

Concomitant agents, 169,

Dispersion, 38, 66, 70, 72,
Equilibrium orbit, 49

170
82, 84
Evolution of computers,

Contradicting parameters,
Dispersion diagram, 83,
120, 122

10, 11, 12, 241

85, 86
Evolution of light sources,

Contradiction matrix, 10,

Dispersion function, 36,
122

11, 11, 13, 17, 19,

37, 40, 49, 218, 219
Evolution of technologies,

41, 56, 74, 103, 105,

Disruption parameter,
5

125, 142, 163, 183,

202, 203

224, 243
DNA, 2, 135, 165, 169, 170
Faraday cup, 23

Convective derivative, 83
DNA polymerase, 170
Fermat’s principle, 74

Cooling, 4, 7, 10, 12, 15,

Doppler shift, 216
Fiber lasers, 64, 212, 213

45, 46, 50, 58, 64,

Double-strand DNA
Field ionization, 62, 109

68, 72, 73, 185, 211,

break, 169, 170
Filamentation, 99, 161,

214–217, 226, 230

Drift tube linac, 79, 80, 89
163

 Index 255

Final focus, 209, 217, 218,

Higgs boson, 1, 231
K-absorption edge, 128

219
High harmonic generation,
Kilpatrick limit, 89, 90

Flat-to-round beam
160, 199
Klystron, 90, 91, 92, 93

transformation, 222
HHG, 160

Flint glass, 70
Higher harmonics, 149,

Flux, 3, 53, 130, 133, 139,

197, 199
L-band, 59, 101

243
Higher-order aberrations,
Landau damping, 58, 68,

Focusing, 2–4, 6, 7, 16, 23,

39
197, 199, 205, 206,

26, 30–32, 35, 38,

Hill’s equation, 26, 27, 29,
207

58, 63, 68–70, 75,

35–37
Landau octupoles, 206

79–81, 97, 100, 117,

Hole-boring acceleration,
Large Hadron Collider, 1,

118, 131, 162, 173,

177, 179
77, 242

175, 178, 202–204,

Hourglass effect, 203, 220
LHC, 1, 5, 7, 125

209, 215, 217, 220,

Hydrogen ionization
Larmor radius, 24

226, 230, 234

potential, 111
Laser, 7, 16–18, 21, 57–59,

Focusing doublet, 31, 32

Hyperboloid of engineer
60, 61–64, 66–68,

FODO lattice, 28, 32, 34,

Garin, 231
71, 72, 105,

69
107–116, 118,

Fourier spectrum, 207

Impregnated cathode, 22
121–124, 128,

Free electron laser, 7, 57,

Inductive output tubes,
135–138, 140, 150,

73, 101, 102, 121,

90, 91
152, 158, 160,

129, 143, 147, 236,

IOT, 90, 91, 92, 93
176–182, 185,

239
Injection, 103, 116, 119,
192–200, 206–210,

FEL, 7, 55, 101, 122,

132, 171, 210, 211,
212, 213, 216, 217,

129–131, 133, 143,

212, 224
225–228, 231, 232,

144, 147, 149–152,

Innovations in science,
236–238, 240, 245

154–161, 196, 197,

225
Laser amplifiers, 61, 63,

199
Insertion devices, 129,
64, 65, 194

Frenet–Serret coordinate
131–133
Laser contrast ratio, 112

system, 26
Integrable optics, 8
Laser cooling, 216, 217

Future Circular Collider,

Interaction point, 3, 4,
Laser guidance, 117, 118

242
217, 242, 243
Laser heater, 197, 206

FCC, 242
Interaction region, 140,
Laser intensity, 108,

200, 219–221, 242

111–113, 115, 117,

Gabor lens, 214, 215, 216

Inventive principles, 10,

125, 140, 178, 179,

Gain medium, 60, 61, 63,

11–15, 17, 212, 217,

181

192, 195

219, 220, 225–227,

Laser pulse compressor,

Gamma function, 33

232, 241
66, 71

Gantry, 171, 174

Inverse Compton
Laser pulse stretcher, 66,

Gaussian-cgs units, 21

scattering, 136
71

Generations of SR sources,

Inverse FEL, 156, 197

Laser pumping, 61

129

Ionization cooling, 214

Laser wakefield

Geometric optics, 31, 34

Iris-loaded structures, 84,

acceleration, 107

Grated plates, 71

85, 93
Laser wire, 16

Group velocity, 82, 83, 84,

Isochronous cyclotron,
Lasing, 157, 161, 193

122, 123

174, 175, 183

Laws of evolution of

Harmonic generation, 199

technical systems,

Harmonic number, 81
John Adams, 7, 230
227

Head–tail effects, 200

John Adams Institute,
Light-sail acceleration,

Head-on collision, 138,

xxix, 240
177, 179

139
JAI, 240, 241
Linac, 5, 6, 75, 77–80, 89,

256 Index

93, 94, 97, 101, 123, Nd:YAG, 61, 62 Phase-locking, 93

131, 132, 139, 159 Nested dolls inventive Phase-space stacking, 211
Linear collider, 55, 101, principle, 13, 220, Photocathode gun, 58, 59,
201, 219 230, 232 197
Liouville’s theorem, 188, Non-local correction, 217 Photoelectron, 166
211 Nonlinear Photon collider, 7, 73
Livingston plot, 4, 5, 75, electrodynamics, Physical contradiction, 9,
105 113 15, 228
Local chromaticity Nonlinear regime of Pierce electrode, 23
correction, 218, 219 plasma Pill-box cavity, 87
Local correction, 217, 221, acceleration, 116 Plasma acceleration, 5, 7,
226 Normalized emittance, 4, 17, 57, 63, 102,
Longitudinal compression, 157, 159, 161 105–108, 112–114,
117 Normalized vector 116–118, 120,
Lorentz force, 21, 22, 216 potential, 111, 112, 122–124, 130, 161,
Luminosity, 4, 15, 138, 125 165, 168, 176, 177,
140, 200, 202, 203, 181, 182, 213, 239
220, 221, 226, 243 Octupole magnets, 30 Plasma beat wave
Optical cavities, 210, 211 accelerator, 107
Magnetic rigidity, 24 Optical parametric CPA, Plasma bubble, 114, 116,
Magnetron, 92, 93 66 118, 119, 161, 162
Master oscillator, 195 OPCPA, 66, 67, 109 Plasma frequency, 67, 68,
Matreshka inventive Optical stochastic cooling, 74, 106
principle, 13, 220, 73 Plasma mirror, 16, 19, 58
230, 232 Optical telescope, 31, 32 Plasma wakefield
Matrix formalism, 27, Oxides cathodes, 22 acceleration, 107
29–31, 33, 39 Pockels cell, 195
Maxwell equations, 21 Parametric process, 67 Ponderomotive force, 58,
Maxwellian distribution, Partition numbers, 46, 50 114, 115, 116, 179
206 Partition theorem, 46 Population inversion, 60,
Microbunching, 143, 147, Paschen curve, 62 192
149, 150, 152, 154, Pasteur quadrant, 225, Power spectral density,
192, 197, 206 233, 235, 239 207
Mini-project, 41, 56, 103, Patent, 6, 8–10, 143, 235 Poynting vector, 82
125, 142, 163, 183, Pelletron, 77 Principal trajectories, 27,
224, 243 Pencil beam scanning, 28, 35, 37
Misalignment, 39, 200, 172, 173 Proton therapy, 124, 168,
208, 209 Penning ion source, 59 171, 172, 173–176,
Mitotic catastrophe, 170 Permittivity of free space, 182
Mode conversion, 94 21 Pseudo-harmonic
Mode locking, 195, 196 Perveance, 58 oscillations, 27, 28
Momentum compaction Phase contrast imaging, Pump-probe experiment,
factor, 38, 96–98, 134 135, 160
206 Phase focusing, 80, 81 PWK spectrum, 208, 209
Monochromatization, 133, Phase space, 7, 35, 36, 72,
139, 183 95, 98, 99, 112, 162, Q-switching techniques,
Multi-photon ionization, 189, 192, 197, 199, 192, 194
62, 110 211, 212, 214, 223 Quadrupole magnet, 25,
Multi-stage plasma Phase stability, 6, 97 26, 28
acceleration, 122 Phase velocity, 82, 83–85, Quality factor, 86, 87, 192
Muon collider, 2 94, 95, 97, 206 Quantum efficiency, 58, 59
 Index 257

Radar, 65, 66, 225, 227,

Self-seeding, 196
Standard Model, 231

228, 229
Semiconductor saturable
Standing wave, 86, 94, 95

Radiation pressure
absorber mirrors,
Stimulated emission, 60,

acceleration, 176,
194
143, 192

179, 180
SESAM, 193, 194
Stimulated emission

Random process, 207

Separatrix, 98, 99, 154,
depletion

Rayleigh length, 68
155
microscopy, 231,

Reference orbit, 26, 37

Sextupole magnets, 30, 70
232

Regenerative laser
Sextupole pairs, 71
Stochastic cooling, 7, 72,

amplifiers, 194
Shamonite, 231
73, 214

Research with
Sheath acceleration, 124,
Stopping power, 167

consideration of
176, 177
Strong focusing, 6, 7, 69,

use, 234
Shock acceleration, 177,
70, 100, 175, 230,

Resonances, 38, 86, 140,

182
234

143, 147, 149, 150,

Shunt impedance, 86, 88,
Superconducting cavities,

175, 198, 217

89
95, 101

Resonant acceleration, 6
SI units, 21, 81
Superconducting magnets,

Resonant cavities, 75, 77,

Single-strand DNA break,
7, 171

86
170
Superconducting RF, 7,

Resonant energy transfer,

Skew quadrupoles, 39,
141

149
222, 223
Surface resistance, 82, 88

Resonant plasma
Skin depth, 82
Symmetry, 39, 118, 231

excitation, 210, 213

Slab lasers, 64, 65
Symplectic

Retardation, 148
SLAC, 2, 92, 123, 197
transformation, 188

RF breakdown , 89
SLAC energy doubling
Synchrocyclotron, 174

RF bucket, 95, 98, 99

experiment, 123
Synchronous particle,

RF coupler, 93
Slice emittance, 157
95–97, 187

RF systems, 77
Small compound, 169,
Synchronous phase, 80

RFQ structures, 7, 79, 80,

170, 229
Synchrophasotron, 7, 69,

80
Small molecule drug, 170
230

Round-to-flat beam
Sokolov–Ternov SR
Synchrotron, 6, 77

transformation, 222
formulae, 55
Synchrotron frequency,

Russian dolls inventive

Solid-state RF, 92
98, 99, 100

principle, xxv, 13,

Space charge effects, 58,
Synchrotron oscillations,

220, 230, 232

186, 223
80, 81

Spallation neutron source,

Synchrotron radiation, 21,

S-band, 59, 101

73
43, 44, 45, 49, 51,

Saddle point, 98, 99

Spectral response
55, 56, 77, 119, 120,

Schwinger intensity limit,

function, 208, 209
127–130, 132, 133,

110, 113
Spreadout Bragg peak, 168
143, 157, 160, 161,

Seeding, 150, 160, 199,

Spur of matrix, 34
198–200, 202, 242

200
SR characteristic
Synchrotron sidebands,

Self amplified
frequency, 47
100

spontaneous
SR damping, 46, 80
Synchrotron tune, 99, 100

emission, 150
SR photon energy, 46
Synectics, 9, 69, 162

SASE, 150, 151, 196

SR quantum regime, 55
Synergies, 21, 57, 58

Self-focusing, 117
SR spectrum, 52, 55
System-antisystem

Self-injection, 116, 119

SR-induced emittance
inventive principle,

Self-modulated laser
growth, 49
13, 220, 232

wakefield
Stability of betatron

accelerator, 107
motion, 33

258 Index

Tandem accelerator, 6, 76,

TRIZ, 9, 11, 12, 14, 15,
Upsilon parameter, 55

77, 103
17–19, 41, 56, 64,

Target normal sheath

69, 74, 92, 103, 125,
Vacuum permeability, 21

acceleration, 176,
142, 163, 174, 183,
Vacuum ultraviolet, 127

177
212, 214, 216, 217,
Valley of Death in
TNSA, 176–179, 181,
219–221, 224–227,
technology transfer,
182
229–232, 240–243
225, 236–238
TE modes, 87
Theory of inventive
Van der Graaf accelerator,

Technological evolution,
problem solving, 1,
6, 76, 77

10
9, 232
Velocity bunching, 185,

Technology transfer
TRIZ algorithm, 10, 11
186

models, 233
Tunneling, 62
Velocity modulation, 91,

TEM wave, 82
Tunneling ionization, 109,
92

Tevatron collider, 5, 214

110
Visible spectrum, 128

Thermionic gun, 22, 23

Twiss functions, 33, 49, 50
Volume-to-surface ratio

Thin-film polarizer, 194,

inventive principle,

195
Ultimate storage rings,
15, 21, 64, 226

Thomson scattering, 135,

130, 131

136
Undamageable or already
Wakefields, 188, 203–205
Time-resolved studies, 121
damaged inventive
Wall-plug efficiency, 65,

TM modes, 87
principle, 15, 16,
121

226, 227
Water window, 128

Tolstoy’s hyperboloid, 231

Undulator, 54, 55, 101,

Wave breaking, 116, 119

Top-off injection, 133

125, 128–130, 132,

Waveguides, 81, 82

Touschek effect, 133

Trace of matrix, 34
133, 135, 143–155,
Weak focusing, 7, 26, 69,

Transfer matrices, 29, 30,

157, 159, 161–163,
175, 230

187
196–199
Wideroe linac, 6, 78, 79

Transit-time factor, 89
Undulator parameter, 55,
Wiggler, 54, 58, 73, 120,

Transition energy, 97, 98

125, 143, 147, 148
128, 129, 132,

Undulator radiation, 55,

143–146, 196–199

Travelling focus, 220, 221

Travelling wave, 94, 95,

133, 146, 153, 161
Wiggler regime, 55, 133

Undulator regime, 55,

97, 103

125, 133, 145

X-band, 59, 101