19.12.2012 Views

A compendium on beam transport and beam ... - IRUVX-PP – the...

A compendium on beam transport and beam ... - IRUVX-PP – the...

A compendium on beam transport and beam ... - IRUVX-PP – the...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

A <str<strong>on</strong>g>compendium</str<strong>on</strong>g> <strong>on</strong> <strong>beam</strong> <strong>transport</strong><br />

<strong>and</strong> <strong>beam</strong> diagnostic methods for<br />

Free Electr<strong>on</strong> Lasers<br />

<strong>IRUVX</strong>-<strong>PP</strong> Experts’ Report<br />

A. Lindblad, S. Svenss<strong>on</strong>, K. Tiedtke<br />

Partners of <strong>IRUVX</strong>-<strong>PP</strong> <strong>–</strong> <strong>the</strong> preparatory phase of EuroFEL


Imprint<br />

Publishing <strong>and</strong> c<strong>on</strong>tact:<br />

Deutsches Elektr<strong>on</strong>en-Synchrotr<strong>on</strong> DESY<br />

<strong>IRUVX</strong>-<strong>PP</strong> Project Coordinator<br />

Notkestr. 85, 22607 Hamburg, Germany<br />

Tel.: +49 40 8998-4130<br />

www.eurofel.eu<br />

ISBN 978-3-935702-45-4<br />

This <str<strong>on</strong>g>compendium</str<strong>on</strong>g> is nei<strong>the</strong>r for sale nor may be resold.<br />

Editors:<br />

Dr. Andreas Lindblad , Prof. Dr. Svante Svenss<strong>on</strong>, Dr. Kai Tiedtke<br />

Copy deadline: March 2011<br />

Cover layout: M<strong>on</strong>ika Illenseer<br />

Printing: K<strong>on</strong>rad Triltsch GmbH, Ochsenfurt-Hohestadt<br />

Editorial note:<br />

The authors of <strong>the</strong> individual scientific c<strong>on</strong>tributi<strong>on</strong>s published<br />

in this <str<strong>on</strong>g>compendium</str<strong>on</strong>g> are fully resp<strong>on</strong>sible for <strong>the</strong> c<strong>on</strong>tents.<br />

Cover:<br />

FLASH experimental hall at DESY in Hamburg <strong>and</strong> diffracti<strong>on</strong> image.<br />

(Photos: © Heiner Mueller-Elsner / Agentur-Focus.de; DESY)


A <str<strong>on</strong>g>compendium</str<strong>on</strong>g> <strong>on</strong> <strong>beam</strong> <strong>transport</strong><br />

<strong>and</strong> <strong>beam</strong> diagnostic methods for<br />

Free Electr<strong>on</strong> Lasers<br />

<strong>IRUVX</strong>-<strong>PP</strong> Experts’ Report<br />

A. Lindblad, S. Svenss<strong>on</strong>, K. Tiedtke<br />

Partners of <strong>IRUVX</strong>-<strong>PP</strong> <strong>–</strong> <strong>the</strong> preparatory phase of EuroFEL


This book was set with L ATEX using <strong>the</strong> memoir class


Foreword<br />

The development <strong>and</strong> use of X-ray free electr<strong>on</strong> lasers have underg<strong>on</strong>e a remarkable<br />

development during <strong>the</strong> last decade: <strong>the</strong> first lasing in <strong>the</strong> VUVat 100 nm wavelength<br />

was achieved in <strong>the</strong> year 2000; in Europe <strong>the</strong> soft X-ray free electr<strong>on</strong> laser Flash<br />

started operati<strong>on</strong>s in 2005 followed by <strong>the</strong> hard X-ray facility Lcls in <strong>the</strong> USA;<br />

during <strong>the</strong> time of writing of this book <strong>the</strong> first seeded X-ray free electr<strong>on</strong> laser has<br />

been commissi<strong>on</strong>ed, <strong>the</strong> Fermi@Elettra. During <strong>the</strong> following decade we will see<br />

upgrades to <strong>the</strong> currently operating facilities, as well as <strong>the</strong> emergence of new sources.<br />

Within <strong>the</strong> coming years <strong>the</strong>re will be hard X-ray free electr<strong>on</strong> laser operating<br />

<strong>on</strong> three c<strong>on</strong>tinents: The European X-Fel, Lcls in North America <strong>and</strong> Scss in<br />

Japan. The progress in accelerator technology <strong>and</strong> in free electr<strong>on</strong> laser science has<br />

resulted in emerging c<strong>on</strong>cepts for more compact <strong>and</strong> more efficient sources, which<br />

can make it feasible for nati<strong>on</strong>al laboratories to build such facilities; In more or<br />

less advanced planning <strong>and</strong> design stages, this is under c<strong>on</strong>siderati<strong>on</strong> in Sweden,<br />

Switzerl<strong>and</strong> (SwissFEL), South Korea (PAL X-Fel) <strong>and</strong> o<strong>the</strong>r countries.<br />

Freeelectr<strong>on</strong>lasers haveunprecedented<strong>beam</strong> properties<strong>and</strong>arecurrently<strong>the</strong>most<br />

intense <strong>and</strong> well collimated man-made phot<strong>on</strong> source in <strong>the</strong> UV to <strong>the</strong> hard X-ray<br />

range. A pulsed X-ray source like this allows for investigati<strong>on</strong> of matter <strong>on</strong> <strong>the</strong> scales<br />

natural to <strong>the</strong> nano-world: femtosec<strong>on</strong>ds <strong>and</strong> nanometers <strong>–</strong> current development aims<br />

for X-rays <strong>on</strong> <strong>the</strong> natural atomic scales: attosec<strong>on</strong>ds <strong>and</strong> ˚Angströms. As access to<br />

this type of source is becoming easier to gain <strong>the</strong> enthusiastic user community is ever<br />

exp<strong>and</strong>ing <strong>–</strong> a significant step essential for lowering <strong>the</strong> threshold for scientists, often<br />

coming from <strong>the</strong> synchrotr<strong>on</strong> or laser communities but not experts in <strong>the</strong> free electr<strong>on</strong><br />

laser field, to c<strong>on</strong>duct experimental work utilizing <strong>the</strong> sources’ unique capabilities. As<br />

<strong>the</strong> field is relatively young (at least for <strong>the</strong> X-ray regime) <strong>the</strong> forthcoming decade(s)<br />

will see <strong>the</strong> field mature bey<strong>on</strong>d pi<strong>on</strong>eering experiments into a broad plethora of<br />

experimental applicati<strong>on</strong>s with an equally broad <strong>and</strong> diverse user community from<br />

many different fields.<br />

Although obviously attractive with regards to phot<strong>on</strong> <strong>beam</strong> quality, <strong>the</strong> c<strong>on</strong>structi<strong>on</strong>,<br />

commissi<strong>on</strong>ing <strong>and</strong> operati<strong>on</strong> of a free electr<strong>on</strong> laser facility is technically very<br />

challenging. The accelerated electr<strong>on</strong> <strong>beam</strong> needs to have very high quality from <strong>the</strong><br />

very start, <strong>and</strong> throughout <strong>the</strong> accelerator <strong>and</strong> <strong>the</strong> magnet array of <strong>the</strong> undulator(s)<br />

requiring specific knowledge in, l<strong>and</strong> development, <strong>beam</strong> dynamics <strong>and</strong> simulati<strong>on</strong>s,<br />

electr<strong>on</strong><strong>beam</strong>diagnostics, synchr<strong>on</strong>izati<strong>on</strong>technology, undulatortechnology<strong>and</strong>laser<br />

seeding. The <strong>beam</strong> of X-rays with extremely high fluence needs to be <strong>transport</strong>ed to<br />

i


ii Foreword<br />

<strong>the</strong> user experiment with <strong>the</strong> resulting dem<strong>and</strong>s <strong>on</strong> X-ray optics with regard to surface<br />

characteristics, damage etc.In additi<strong>on</strong>, each phot<strong>on</strong>pulse needs to be diagnosed<br />

<strong>and</strong> its characteristics stored for later inclusi<strong>on</strong> in data-analysis, for poly-color experiments<br />

(involving o<strong>the</strong>r laser pulses) informati<strong>on</strong> for pulse-synchr<strong>on</strong>izati<strong>on</strong> needs to<br />

be available <strong>and</strong> so forth.<br />

In this phase, <strong>the</strong> funding programmes of <strong>the</strong> European Commissi<strong>on</strong> (EC) for <strong>the</strong><br />

development of new research infrastructures helped interested European instituti<strong>on</strong>s<br />

enormously to join <strong>the</strong>ir forces for developing key technologies required for <strong>the</strong> design<br />

<strong>and</strong> c<strong>on</strong>structi<strong>on</strong> of <strong>the</strong> next generati<strong>on</strong> free electr<strong>on</strong> laser sources in Europe:<br />

• The European Fel design study project EUROFEL funded by <strong>the</strong> 6 th Framework<br />

Programme for a period of three years, 2005<strong>–</strong>2007, focused <strong>on</strong> comp<strong>on</strong>ents<br />

<strong>the</strong> electr<strong>on</strong> accelerator <strong>and</strong> <strong>the</strong> free electr<strong>on</strong> laser itself.<br />

• This project was followed by Iruvx-<strong>PP</strong>, <strong>the</strong> preparatory project for a sustainable<br />

future European FEL c<strong>on</strong>sortium called EuroFEL. Iruvx-<strong>PP</strong> includes<br />

significant funding for fur<strong>the</strong>r technical developments regarding both <strong>the</strong> electr<strong>on</strong><br />

<strong>and</strong> phot<strong>on</strong> <strong>beam</strong>s.<br />

This <str<strong>on</strong>g>compendium</str<strong>on</strong>g> is an amalgamati<strong>on</strong> of <strong>the</strong> collaborative efforts of two of Iruvx-<br />

<strong>PP</strong>’s workpackes, <strong>the</strong> third <strong>and</strong> seventh <strong>–</strong> whose efforts were focused <strong>on</strong> X-ray phot<strong>on</strong><br />

<strong>beam</strong> optics <strong>and</strong> diagnostics. C<strong>on</strong>trasting <strong>the</strong> mainly political <strong>and</strong> structural objectives<br />

of Iruvx-<strong>PP</strong>, it provides an immediate practical result <strong>and</strong> delivers timely <strong>–</strong><br />

toge<strong>the</strong>r with an introducti<strong>on</strong> to free electr<strong>on</strong> laser technology <strong>and</strong> science <strong>–</strong> an urgently<br />

needed text which details <strong>the</strong> latest developments in <strong>the</strong> field in a manner also<br />

accessible for n<strong>on</strong>-experts.<br />

Ithasnotbeenintendedtoprovideacomplete indepthaccountof<strong>the</strong>whole fieldof<br />

free electr<strong>on</strong> laser science but ra<strong>the</strong>r to provide: (i: a presentati<strong>on</strong> of key technologies<br />

<strong>and</strong> c<strong>on</strong>cepts needed to underst<strong>and</strong> <strong>the</strong> various requirements that are imposed <strong>on</strong> <strong>the</strong><br />

electr<strong>on</strong> <strong>and</strong> phot<strong>on</strong> <strong>beam</strong>s, which limitati<strong>on</strong>s different technology choices impose <strong>on</strong><br />

<strong>the</strong> final output of <strong>the</strong> free electr<strong>on</strong> laser; (ii, detail, sometimes exp<strong>and</strong> <strong>and</strong> put into<br />

c<strong>on</strong>text <strong>the</strong> various internal reports generated in <strong>the</strong> workpackages. Taken toge<strong>the</strong>r<br />

both ambiti<strong>on</strong>s have generated a nice recollecti<strong>on</strong> of <strong>the</strong> collaborative efforts of <strong>the</strong><br />

members c<strong>on</strong>stituting <strong>the</strong> workpackages in an accessible form that hopefully will help<br />

future students <strong>and</strong> colleagues to get acquainted with <strong>the</strong> exciting field of free electr<strong>on</strong><br />

laser science.<br />

Josef Feldhaus, Hamburg, 25 th February, 2011.<br />

Scientific coordinator of <strong>the</strong> Iruvx-<strong>PP</strong> project


Preface<br />

Caveat Lector<br />

This book’s humble beginning was as a summarizing report of <strong>the</strong> collective efforts<br />

within <strong>the</strong> 3 rd & 7 th workpackage of <strong>the</strong> Iruvx-<strong>PP</strong> programme. During <strong>the</strong> time of<br />

writing (<strong>the</strong> last twelve m<strong>on</strong>ths of <strong>the</strong> programme) <strong>the</strong> members of <strong>the</strong> workpackages<br />

produced reports <strong>on</strong> <strong>the</strong>ir findings <strong>–</strong> often c<strong>on</strong>taining reviews of <strong>the</strong> various fields<br />

where <strong>the</strong> undertakings took place.<br />

Each such report have been, more or less, basis for secti<strong>on</strong>s <strong>and</strong> indeed whole<br />

chapters in this book, notably in <strong>the</strong> sec<strong>on</strong>d part. The reports’ titles <strong>and</strong> <strong>the</strong>ir authors<br />

are menti<strong>on</strong>ed in <strong>the</strong> very beginning of <strong>the</strong> chapters where material from <strong>the</strong>m have<br />

been used.<br />

A summarizing report may be interesting in itself, but to heighten <strong>the</strong> appeal to<br />

a broader audience <strong>and</strong> to future students <strong>and</strong> colleagues it was thought that when<br />

adding c<strong>on</strong>text to <strong>the</strong> reports (who by <strong>the</strong>mselves deal mainly with X-ray optics<br />

<strong>and</strong> phot<strong>on</strong> diagnostic methods) a good introductory textbook describing <strong>the</strong> various<br />

challenges pertainingtowork<strong>and</strong>developmentoffree electr<strong>on</strong> lasers couldberealized.<br />

Hence <strong>the</strong> first part of <strong>the</strong> book describes <strong>the</strong> physics <strong>and</strong> c<strong>on</strong>cepts that govern<br />

free electr<strong>on</strong> lasers. The various comp<strong>on</strong>ents of a free electr<strong>on</strong> laser are also described<br />

<strong>–</strong> from <strong>the</strong> electr<strong>on</strong> gun to <strong>the</strong> undulators, <strong>and</strong> some X-ray optics. The sec<strong>on</strong>d part<br />

of <strong>the</strong> book focuses heavily <strong>on</strong> diagnostic methods that can be used to quantify <strong>the</strong><br />

properties of <strong>the</strong> free electr<strong>on</strong> laser phot<strong>on</strong> <strong>beam</strong>. In <strong>the</strong> last chapters of part two<br />

<strong>the</strong> current running <strong>and</strong> planned facilities are described in <strong>the</strong> light of what we have<br />

learned in <strong>the</strong> firstpart of <strong>the</strong> book, planned <strong>and</strong>performed experiments are described<br />

as to give an orientati<strong>on</strong> of what scientists are trying to achieve with <strong>the</strong> facilities.<br />

The bibliography of this book c<strong>on</strong>tains both references to books <strong>and</strong> articles that<br />

reflect <strong>the</strong> current state of <strong>the</strong> art, at least for part two of <strong>the</strong> book, at <strong>the</strong> time of<br />

writing (February 2011). No such list is ever complete <strong>and</strong> will so<strong>on</strong> become dated<br />

<strong>–</strong> however through citati<strong>on</strong>s to those papers <strong>the</strong> reader will so<strong>on</strong> find where <strong>the</strong> field<br />

have developed. Without any doubt many of <strong>the</strong> references will remain key references<br />

for quite a foreseeable time.<br />

iii


iv Preface<br />

Acknowledgements<br />

I would like to extend my gratitude to my co-editors Prof. Df. Svante Svenss<strong>on</strong><br />

<strong>and</strong> Dr. Kai Tiedkte for giving me <strong>the</strong> opportunity to work with this book. I also<br />

thank <strong>the</strong>m for <strong>the</strong> many nice discussi<strong>on</strong>s <strong>and</strong> <strong>the</strong> good criticism during <strong>the</strong> authoring<br />

process. Naturally I also like to thank <strong>the</strong> authors of <strong>the</strong> reports which c<strong>on</strong>stitute<br />

parts of this book without which I would literally have started with a blank page.<br />

Thanksto<strong>the</strong>manynicepresentati<strong>on</strong>s<strong>on</strong><strong>the</strong>workshops<strong>and</strong>freeelectr<strong>on</strong>laser c<strong>on</strong>ference<br />

last year I have been kindly introduced to <strong>the</strong> fantastic subject of free electr<strong>on</strong><br />

laser science. Again, for <strong>the</strong> opportunity to work with it <strong>and</strong> writing a book about it<br />

I will be forever grateful.<br />

Andréas Lindblad, Uppsala, 28 th February, 2011.


C<strong>on</strong>tents<br />

Foreword i<br />

Preface iii<br />

C<strong>on</strong>tents v<br />

I Free electr<strong>on</strong> lasers <strong>–</strong> a primer<br />

1 Introducti<strong>on</strong> 3<br />

1.1 Historical exposé & scientific background . . . . . . . . . . . . . . . . 3<br />

1.2 Laboratory X-ray <strong>and</strong> UV/Vis phot<strong>on</strong> sources . . . . . . . . . . . . . . 7<br />

X-ray tube <strong>and</strong> anode sources . . . . . . . . . . . . . . . . . . . . . . . 8<br />

Synchrotr<strong>on</strong> light sources . . . . . . . . . . . . . . . . . . . . . . . . . 9<br />

Lasers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10<br />

High Harm<strong>on</strong>ic Generati<strong>on</strong> Lasers . . . . . . . . . . . . . . . . . . . . 11<br />

1.3 Free electr<strong>on</strong> Lasers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12<br />

1.4 Development of X-ray free electr<strong>on</strong> lasers . . . . . . . . . . . . . . . . 15<br />

1.5 Seeding schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15<br />

eSase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17<br />

HGHG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17<br />

EEHG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18<br />

Harm<strong>on</strong>ic afterburners . . . . . . . . . . . . . . . . . . . . . . . . . . . 18<br />

1.6 Definiti<strong>on</strong>s used throughout <strong>the</strong> book . . . . . . . . . . . . . . . . . . . 19<br />

Brilliance <strong>and</strong> Brightness . . . . . . . . . . . . . . . . . . . . . . . . . 19<br />

Emittance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19<br />

2 Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties 23<br />

2.1 Radiati<strong>on</strong> from a moving charge . . . . . . . . . . . . . . . . . . . . . 23<br />

Maxwell’s laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23<br />

Charged particle at rest or moving with c<strong>on</strong>stant velocity . . . . . . . 24<br />

The fields from a charge in arbitrary moti<strong>on</strong> . . . . . . . . . . . . . . . 25<br />

Frequency <strong>and</strong> coherence of synchrotr<strong>on</strong> radiati<strong>on</strong> . . . . . . . . . . . 27<br />

v


vi C<strong>on</strong>tents<br />

2.2 Radiati<strong>on</strong> from a bending magnet . . . . . . . . . . . . . . . . . . . . . 29<br />

2.3 Undulator radiati<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30<br />

The undulator equati<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . 30<br />

2.4 Microbunching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37<br />

Interacti<strong>on</strong> between <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> radiati<strong>on</strong> field . . . . . 37<br />

Exp<strong>on</strong>ential gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39<br />

Scaled free electr<strong>on</strong> laser equati<strong>on</strong>s . . . . . . . . . . . . . . . . . . . . 41<br />

2.5 Sase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44<br />

Coherence properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45<br />

3 Free electr<strong>on</strong> laser ”hardware” 47<br />

3.1 A prototypical FEL amplifier . . . . . . . . . . . . . . . . . . . . . . . 47<br />

3.2 Electr<strong>on</strong> guns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47<br />

General requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48<br />

Thermi<strong>on</strong>ic emitters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49<br />

Photocathode emitters . . . . . . . . . . . . . . . . . . . . . . . . . . . 50<br />

Normally c<strong>on</strong>ducting guns . . . . . . . . . . . . . . . . . . . . . . . . . 50<br />

Superc<strong>on</strong>ducting guns . . . . . . . . . . . . . . . . . . . . . . . . . . . 51<br />

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52<br />

3.3 Radio-frequency driven accelerators . . . . . . . . . . . . . . . . . . . . 52<br />

The accelerating RF-field . . . . . . . . . . . . . . . . . . . . . . . . . 54<br />

Energy gain in a radiofrequency driven accelerating cavity . . . . . . . 57<br />

Warm technology: Copper . . . . . . . . . . . . . . . . . . . . . . . . . 58<br />

Superc<strong>on</strong>ducting technology . . . . . . . . . . . . . . . . . . . . . . . . 58<br />

3.4 Undulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59<br />

Undulator tolerances example . . . . . . . . . . . . . . . . . . . . . . . 60<br />

4 X-ray optics 63<br />

4.1 Dem<strong>and</strong>s <strong>on</strong> optics precisi<strong>on</strong> at free electr<strong>on</strong> lasers . . . . . . . . . . . 63<br />

4.2 Focussing mirrors <strong>–</strong> back-reflecting geometry example . . . . . . . . . 65<br />

4.3 Damage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65<br />

4.4 Diffracti<strong>on</strong> gratings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66<br />

4.5 M<strong>on</strong>ochromators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67<br />

4.6 Beam attenuators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69<br />

5 Beam-splitting methods 73<br />

5.1 Introducti<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73<br />

5.2 Beam-splitter specificati<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . 74<br />

5.3 Amplitude divisi<strong>on</strong> <strong>beam</strong> splitters . . . . . . . . . . . . . . . . . . . . 76<br />

Partially transmitting materials . . . . . . . . . . . . . . . . . . . . . . 76<br />

Crystal diffracti<strong>on</strong> <strong>beam</strong> splitters . . . . . . . . . . . . . . . . . . . . . 78<br />

Gratings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80<br />

Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83<br />

5.4 Wavefr<strong>on</strong>t divisi<strong>on</strong> <strong>beam</strong>splitters . . . . . . . . . . . . . . . . . . . . . 83<br />

Beamline apertures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83<br />

Knife-edge mirrors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84<br />

Knife-edge crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84<br />

Fresnel bi-mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85<br />

Slotted or perforated mirrors . . . . . . . . . . . . . . . . . . . . . . . 85


C<strong>on</strong>tents vii<br />

Structured arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87<br />

5.5 Time-based splitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88<br />

5.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89<br />

Techniques requiring <strong>the</strong> least development . . . . . . . . . . . . . . . 89<br />

Techniques requiring more development . . . . . . . . . . . . . . . . . 90<br />

II Beam diagnostics<br />

6 Introducti<strong>on</strong> 95<br />

6.1 Diagnostics categorizati<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . 95<br />

Subcategorizati<strong>on</strong>s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96<br />

6.2 C<strong>on</strong>clusi<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97<br />

7 Spectral diagnostics: Intensity & Energy 99<br />

7.1 X-ray spectrometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99<br />

7.2 Intensity/Beam energy . . . . . . . . . . . . . . . . . . . . . . . . . . . 100<br />

Gas m<strong>on</strong>itor detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . 101<br />

Calorimeters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102<br />

Solid state devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103<br />

7.3 Phot<strong>on</strong>-energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104<br />

I<strong>on</strong> time-of-flight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104<br />

Electr<strong>on</strong> time-of-flight . . . . . . . . . . . . . . . . . . . . . . . . . . . 106<br />

8 Beam cross-secti<strong>on</strong> diagnostics 109<br />

8.1 Introducti<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109<br />

The ideal cross-secti<strong>on</strong> diagnostic . . . . . . . . . . . . . . . . . . . . . 109<br />

Distributi<strong>on</strong> of diagnostics al<strong>on</strong>g <strong>the</strong> <strong>beam</strong> . . . . . . . . . . . . . . . . 110<br />

C<strong>on</strong>tent of this chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . 111<br />

8.2 Definiti<strong>on</strong>s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112<br />

8.3 Direct imaging of <strong>the</strong> <strong>beam</strong> . . . . . . . . . . . . . . . . . . . . . . . . 113<br />

Imaging a replica of <strong>the</strong> <strong>beam</strong> . . . . . . . . . . . . . . . . . . . . . . . 114<br />

Summary of imaging techniques . . . . . . . . . . . . . . . . . . . . . . 117<br />

8.4 Scanning techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117<br />

Scanning wire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117<br />

Scanning crossed wires . . . . . . . . . . . . . . . . . . . . . . . . . . . 118<br />

Scanning slit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118<br />

Scanning pinhole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119<br />

Scanning knife-edge . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119<br />

Summary of scanning techniques . . . . . . . . . . . . . . . . . . . . . 119<br />

8.5 I<strong>on</strong>izati<strong>on</strong> <strong>beam</strong>profile detectors . . . . . . . . . . . . . . . . . . . . . . 120<br />

8.6 Imaging i<strong>on</strong> chambers . . . . . . . . . . . . . . . . . . . . . . . . . . . 120<br />

Fluorescence detecti<strong>on</strong> in residual gas m<strong>on</strong>itors . . . . . . . . . . . . . 124<br />

8.7 Sampling techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126<br />

8.8 Spot size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129<br />

Techniques useable with unattenuated <strong>beam</strong>s . . . . . . . . . . . . . . 129<br />

Photoi<strong>on</strong>isati<strong>on</strong> saturati<strong>on</strong> of rare gases . . . . . . . . . . . . . . . . . 130<br />

8.9 Techniques requiring attenuated <strong>beam</strong>s . . . . . . . . . . . . . . . . . . 131<br />

Wire, knife-edge <strong>and</strong> slit scans . . . . . . . . . . . . . . . . . . . . . . 131


viii C<strong>on</strong>tents<br />

Photographic film . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131<br />

Gas-filled detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132<br />

Charge coupled device (CCD) . . . . . . . . . . . . . . . . . . . . . . . 132<br />

Multichannel plate (MCP) . . . . . . . . . . . . . . . . . . . . . . . . . 133<br />

Solid State Detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133<br />

8.10 Positi<strong>on</strong> <strong>and</strong> centroiding . . . . . . . . . . . . . . . . . . . . . . . . . . 133<br />

Sampling techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133<br />

8.11 Wavefr<strong>on</strong>t measurements . . . . . . . . . . . . . . . . . . . . . . . . . . 141<br />

8.12 THz/IR techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142<br />

Thermal detecti<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143<br />

Phot<strong>on</strong>ic detecti<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144<br />

New detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145<br />

IR <strong>and</strong> THz <strong>beam</strong> profiling . . . . . . . . . . . . . . . . . . . . . . . . 145<br />

8.13 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148<br />

9 Pulse length, profile <strong>and</strong> jitter 151<br />

9.1 Introducti<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151<br />

9.2 Cross-correlati<strong>on</strong> techniques . . . . . . . . . . . . . . . . . . . . . . . . 153<br />

9.3 Electro-optic techniques . . . . . . . . . . . . . . . . . . . . . . . . . . 155<br />

9.4 Autocorrelati<strong>on</strong> techniques . . . . . . . . . . . . . . . . . . . . . . . . 155<br />

Intensity autocorrelati<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . 156<br />

Autocorrelati<strong>on</strong> techniques for complete pulse characterizati<strong>on</strong> . . . . 159<br />

Frequency Resolved Optical Gating (FROG) . . . . . . . . . . . . . . 160<br />

Polarizati<strong>on</strong>-gate FROG (PG FROG) . . . . . . . . . . . . . . . . . . 162<br />

Self-diffracti<strong>on</strong> FROG (SD FROG) . . . . . . . . . . . . . . . . . . . . 163<br />

Transient-grating FROG (TG FROG) . . . . . . . . . . . . . . . . . . 163<br />

Sec<strong>on</strong>d-harm<strong>on</strong>ic-generati<strong>on</strong> FROG (SHG FROG) . . . . . . . . . . . 163<br />

Third-harm<strong>on</strong>ic-generati<strong>on</strong> FROG (THG FROG) . . . . . . . . . . . . 164<br />

9.5 Spectral Phase Interferometry for Direct Electric-field Rec<strong>on</strong>structi<strong>on</strong> 165<br />

9.6 Reflectivity modulati<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . 166<br />

9.7 Streak cameras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167<br />

9.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168<br />

10 Free electr<strong>on</strong> laser experiments 173<br />

10.1 The ”holy grails” of free electr<strong>on</strong> laser experiments . . . . . . . . . . . 173<br />

”Molecular movies” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173<br />

”Single-molecule/nanostructure imaging” . . . . . . . . . . . . . . . . 174<br />

”Single-shot spectroscopy/imaging” . . . . . . . . . . . . . . . . . . . . 174<br />

10.2 Time-resolved spectroscopies . . . . . . . . . . . . . . . . . . . . . . . 174<br />

UV/Vis pump-X-ray probe spectroscopy . . . . . . . . . . . . . . . . . 174<br />

Nexafs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175<br />

10.3 Imaging <strong>and</strong> Crystallography . . . . . . . . . . . . . . . . . . . . . . . 175<br />

10.4 N<strong>on</strong>-linear X-ray science . . . . . . . . . . . . . . . . . . . . . . . . . . 176<br />

Photoi<strong>on</strong>izati<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176<br />

11 Free Electr<strong>on</strong> Laser facilities 179<br />

11.1 Operating facilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180<br />

11.2 Flash . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180<br />

Injector <strong>and</strong> accelerator . . . . . . . . . . . . . . . . . . . . . . . . . . 181


C<strong>on</strong>tents ix<br />

Undulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181<br />

sFlash . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181<br />

Flash-II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181<br />

Experimental stati<strong>on</strong>s: . . . . . . . . . . . . . . . . . . . . . . . . . . . 182<br />

11.3 Scss <strong>–</strong> X-fel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183<br />

Scss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183<br />

Injector & accelerator . . . . . . . . . . . . . . . . . . . . . . . . . . . 183<br />

Undulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183<br />

X-ray free electr<strong>on</strong> laser/ Spring-8 . . . . . . . . . . . . . . . . . . . . 183<br />

Injector & accelerator . . . . . . . . . . . . . . . . . . . . . . . . . . . 183<br />

Undulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184<br />

11.4 Fermi@Elettra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184<br />

Injector & accelerator . . . . . . . . . . . . . . . . . . . . . . . . . . . 184<br />

Undulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184<br />

Experiments: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185<br />

11.5 Lcls <strong>–</strong> Linac Coherent Light Source . . . . . . . . . . . . . . . . 186<br />

Injector <strong>and</strong> accelerator . . . . . . . . . . . . . . . . . . . . . . . . . . 186<br />

Undulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186<br />

Lcls-II proposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186<br />

Experiments: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187<br />

11.6 Facilities under c<strong>on</strong>structi<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . 188<br />

11.7 The European Xfel . . . . . . . . . . . . . . . . . . . . . . . . . . . 188<br />

Injector <strong>and</strong> accelerator . . . . . . . . . . . . . . . . . . . . . . . . . . 188<br />

Undulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188<br />

Experiments: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189<br />

11.8 SwissFEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190<br />

SwissFEL injector test facility . . . . . . . . . . . . . . . . . . . . . . 190<br />

<strong>the</strong> SwissFEL proposal . . . . . . . . . . . . . . . . . . . . . . . . . . 190<br />

Injector & accelerator . . . . . . . . . . . . . . . . . . . . . . . . . . . 191<br />

Undulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191<br />

Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191<br />

11.9 Proposed facilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192<br />

11.10PAL-X-fel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192<br />

Accelerator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192<br />

Undulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192<br />

12 Outlook & C<strong>on</strong>clusi<strong>on</strong>s 195<br />

12.1 Current trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195<br />

More compact sources <strong>and</strong> alternative approaches . . . . . . . . . . . . 195<br />

Higher repetiti<strong>on</strong> rates . . . . . . . . . . . . . . . . . . . . . . . . . . . 196<br />

Polarizati<strong>on</strong> c<strong>on</strong>trol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196<br />

Coherence & Seeding . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197<br />

Harder X-rays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198<br />

12.2 C<strong>on</strong>clusi<strong>on</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199<br />

Bibliography 201


Part I<br />

Free electr<strong>on</strong> lasers <strong>–</strong> a primer


1. Introducti<strong>on</strong><br />

Written by: A. Lindblad<br />

1.1 Historical exposé & scientific background<br />

Ever since <strong>the</strong> advent of electricity it has been possible for humankind to produce<br />

artificial light sources not emanating from chemical processes, i.e. different from c<strong>and</strong>les,<br />

b<strong>on</strong>-fires, <strong>and</strong> <strong>the</strong> like. With <strong>the</strong> discovery of X-rays by Wilhelm Röntgen in<br />

1895 1 for which he subsequently got <strong>the</strong> first Nobel prize in physics 1901.<br />

Hisdiscovery was made possible with <strong>the</strong>adventof vacuum<br />

tube discharge sources in <strong>the</strong> end of <strong>the</strong> 19 th century <strong>–</strong> in<br />

a discharge tube (invented by William Crookes <strong>and</strong> o<strong>the</strong>rs,<br />

Figure 1.1) electr<strong>on</strong>s travel between two electrodes in a gas<br />

tube between which a high electric voltage have been applied,<br />

Röntgen discovered that, even though he covered <strong>the</strong> cathode,<br />

using cardboard, wood, books (<strong>and</strong> seemingly whatever came<br />

in h<strong>and</strong>y) a phosphorous screen placed in <strong>the</strong> o<strong>the</strong>r end of<br />

his laboratory room still glowed. In part II of this book we<br />

will see how such screens are still in use to characterize X-ray<br />

sources 2 .<br />

Figure 1.1: A Crookes tube<br />

from 1910’s.<br />

The interacti<strong>on</strong> of X-ray phot<strong>on</strong>s with matter can generate many answers in <strong>the</strong><br />

scientific inquiry <strong>–</strong> spurred by <strong>the</strong> rapid development of quantum <strong>the</strong>ory <strong>–</strong> <strong>and</strong> thus<br />

our underst<strong>and</strong>ing of atoms <strong>and</strong> matter, combined with <strong>the</strong> development of X-ray<br />

lightsources with ever-increasing quality (Figure 1.3). As evident in <strong>the</strong> short list<br />

below both scientific topics ranging from biology, chemistry <strong>and</strong> physics <strong>and</strong> industrial<br />

research <strong>and</strong> development (notably <strong>the</strong> pharmaceutical industry) have greatly<br />

benefitted from <strong>the</strong> development of this area of science.<br />

• X-ray imaging <strong>–</strong> medical imaging, materials’ science, safety<br />

1 Reported in Über eine neue Art v<strong>on</strong> Strahlen (http://de.wikisource.org/wiki/Ueber eine<br />

neue Art v<strong>on</strong> Strahlen) <strong>–</strong> published <strong>the</strong> December 28, 1895, where Röntgen refers his discovery as<br />

”X-rays” (a noti<strong>on</strong> which <strong>the</strong> modest Röntgen preferred, in many languages this type of radiati<strong>on</strong> is<br />

still known as Röntgen-rays). An english translati<strong>on</strong> can be found in Nature 53, 274<strong>–</strong>276 (January<br />

23, 1886). The discovery was deemed important enough to be translated <strong>and</strong> communicated within<br />

a m<strong>on</strong>th of <strong>the</strong> original publicati<strong>on</strong>.<br />

2 For instance, Secti<strong>on</strong> 8.3, (see page 113)<br />

3


4 1. Introducti<strong>on</strong><br />

• X-ray diffracti<strong>on</strong> <strong>–</strong> crystallography, materials’ science<br />

• X-ray absorpti<strong>on</strong> <strong>and</strong> emissi<strong>on</strong> spectroscopies <strong>–</strong> materials’ science, l<strong>on</strong>g range<br />

order.<br />

• X-ray photoelectr<strong>on</strong> spectroscopies <strong>–</strong> materials’ science, chemical b<strong>on</strong>ds, spinresolved<br />

Ofcourse <strong>the</strong>technological developmentoftechniquesfor particle accelerators, sample<br />

h<strong>and</strong>ling, radiati<strong>on</strong> detecti<strong>on</strong>, vacuum have both received <strong>and</strong> given synergetic effects<br />

in society at large.<br />

High energy phot<strong>on</strong>s or particles from radioactive sources 3 is an alternative as<br />

particle sources, though <strong>the</strong>y lack tunability <strong>and</strong> <strong>the</strong> energy is often too high to be<br />

used for many of <strong>the</strong> important applicati<strong>on</strong>s of X-rays. Applicati<strong>on</strong>s of radioactive<br />

radiati<strong>on</strong> have been found elsewhere, notably in radiati<strong>on</strong> <strong>the</strong>rapy, as gene-markers,<br />

<strong>the</strong> carb<strong>on</strong>-14 age-determinati<strong>on</strong> method etc.<br />

In Ernest Ru<strong>the</strong>rford’s laboratory (1909), Hans Geiger <strong>and</strong> Ernest Marsded used<br />

a <strong>beam</strong> of alpha particles generated from a radioactive decay of <strong>the</strong> element radium<br />

impinging <strong>on</strong> a gold foil to find out how charge was distributed within atoms. They<br />

intended to investigate <strong>the</strong> prevailing plum pudding model 4 where <strong>the</strong> positive charge<br />

was delocalized (<strong>the</strong> pudding) <strong>and</strong> <strong>the</strong> electr<strong>on</strong>s submerged (as <strong>the</strong> plums). The gold<br />

foil was surrounded by a sheet of zinc sulfide which would fluoresce when hit by alpha<br />

particles. From <strong>the</strong> plum model it was expected that <strong>the</strong> alpha particles would not<br />

scatter at all, however it was observed that a significant fracti<strong>on</strong> of <strong>the</strong> alphaparticles<br />

was backscattered by very small c<strong>on</strong>centrated positive charged objects in <strong>the</strong> gold<br />

film <strong>–</strong> this was taken as a str<strong>on</strong>g indicati<strong>on</strong> of <strong>the</strong> existence of an atomic nucleus, a<br />

result which was not expected. In 1911 Ru<strong>the</strong>rford explained <strong>the</strong> experiment in terms<br />

of scattering which resulted in <strong>the</strong> Ru<strong>the</strong>rford ”planetary” model of <strong>the</strong> atom.<br />

In Ru<strong>the</strong>rford’s laboratory significant developments followed this, especially <strong>the</strong><br />

development of particle accelerators with <strong>the</strong> purpose of splitting <strong>the</strong> atom. A note<br />

<strong>on</strong> <strong>the</strong> development of some particle accelerator techniques <strong>and</strong> c<strong>on</strong>cepts emanating<br />

from Ru<strong>the</strong>rford <strong>and</strong> his students, relevant for this book, can be found <strong>on</strong> page 52.<br />

For a lot of medical purposes (<strong>the</strong>rapeutic <strong>and</strong> o<strong>the</strong>rwise) radioactive materials<br />

are increasingly phased out to <strong>the</strong> benefit of particle accelerators where c<strong>on</strong>trol of <strong>the</strong><br />

particle energy is possible <strong>and</strong> thus <strong>the</strong> interacti<strong>on</strong> length <strong>and</strong> dose can be c<strong>on</strong>trolled.<br />

Theories <strong>on</strong> <strong>the</strong> nature of light <strong>–</strong> a small historical primer<br />

In <strong>the</strong> 17 th century investigati<strong>on</strong>s into <strong>the</strong> nature of light were pursued by means<br />

of <strong>the</strong> modern scientific method. Light was thought of as waves or particles <strong>–</strong> both<br />

with <strong>the</strong>ir shortcomings. The discourse between <strong>the</strong> two, seemingly incommensurate,<br />

st<strong>and</strong>points were ultimately resolved by <strong>the</strong> advent of quantum mechanics in <strong>the</strong><br />

beginning of <strong>the</strong> 20 th century <strong>–</strong> which argue that light (<strong>and</strong> matter) can behave both<br />

as waves <strong>and</strong> particles.<br />

3 That is: phot<strong>on</strong>s, electr<strong>on</strong>s or <strong>the</strong>ir antiparticles: positr<strong>on</strong>s, or 4 He nuclei <strong>–</strong> gamma, γ, beta,<br />

β ± <strong>and</strong> alpha, α radiati<strong>on</strong>, respectively. The greek names for radiati<strong>on</strong> were all coined by <strong>the</strong><br />

english physicist Ernest Ru<strong>the</strong>rford. For his work <strong>on</strong> radioactivity <strong>–</strong> transmutati<strong>on</strong>, <strong>the</strong> noti<strong>on</strong><br />

of half-life, <strong>and</strong> <strong>the</strong> differentiati<strong>on</strong> between alpha <strong>and</strong> beta rays <strong>–</strong> he received <strong>the</strong> Nobel prize in<br />

chemistry 1908.<br />

4 The plum pudding model of <strong>the</strong> atom was proposed by J. J. Thoms<strong>on</strong> in 1904.


1.1. Historical exposé & scientific background 5<br />

Robert Hooke <strong>and</strong> Christiaan Huygens both published <strong>the</strong>ories (In <strong>the</strong> 1660s <strong>and</strong><br />

late 1670s respectively) <strong>on</strong> light that built <strong>on</strong> light rays being waves. The main<br />

propositi<strong>on</strong> was that light, being waves, would not be affected by gravity <strong>–</strong> thus<br />

slowing down up<strong>on</strong> entering a denser medium. The wave <strong>the</strong>ory though assumed <strong>the</strong><br />

existence of a medium in which <strong>the</strong> waves could propagate through: <strong>the</strong> luminiferous<br />

æ<strong>the</strong>r 5 .<br />

Isaac Newt<strong>on</strong> described light as particles (corpuscles) to explain reflecti<strong>on</strong> of light<br />

(Opticks, 1704). Perhaps owing to his work <strong>on</strong> gravity, he postulated that <strong>the</strong> corpuscles<br />

were accelerated when <strong>the</strong>y entered a denser medium, because of <strong>the</strong> larger<br />

gravitati<strong>on</strong>al influence <strong>on</strong> <strong>the</strong> particles from <strong>the</strong> medium, which helped explain refracti<strong>on</strong>.<br />

Nowadays we know that this was a step in <strong>the</strong> wr<strong>on</strong>g directi<strong>on</strong> since <strong>the</strong><br />

speed of light is slower in a denser medium than in a dilute medium.<br />

Investigati<strong>on</strong>s into <strong>the</strong> various properties of light such as refracti<strong>on</strong>, reflecti<strong>on</strong>,<br />

polarizati<strong>on</strong> <strong>and</strong> diffracti<strong>on</strong> undertaken in <strong>the</strong> beginning of <strong>the</strong> 19 th century tilted<br />

<strong>the</strong> balance in favor of <strong>the</strong> wave <strong>the</strong>ory. However <strong>–</strong> as both <strong>the</strong>ories made different<br />

predicti<strong>on</strong>s <strong>on</strong> how <strong>the</strong> speed of light changed up<strong>on</strong> entering a denser medium <strong>–</strong> <strong>the</strong><br />

most c<strong>on</strong>vincing test to be made, i.e. measuring <strong>the</strong> actual speed of light had to wait<br />

until 1850 <strong>and</strong> Lé<strong>on</strong> Focault for a precise enough experiment to be performed 6 . His<br />

results favored <strong>the</strong> wave <strong>the</strong>ory of light <strong>and</strong> thus put <strong>the</strong> particle <strong>the</strong>ory of light out<br />

of <strong>the</strong> scientific limelight<br />

In <strong>the</strong> later part of <strong>the</strong> 19 th century James Clerk Maxwell formulated <strong>the</strong> governing<br />

equati<strong>on</strong>s of electromagnetism. His <strong>the</strong>ory built <strong>on</strong> that electromagnetic waves<br />

travelled at a c<strong>on</strong>stant speed <strong>–</strong> equal to <strong>the</strong> speed of light. 1862 he published <strong>the</strong><br />

noti<strong>on</strong> that light was a form of electromagnetic wave in On <strong>the</strong> physical lines of force.<br />

A full <strong>the</strong>ory, describing ma<strong>the</strong>matically <strong>the</strong> behavior of electric <strong>and</strong> magnetic fields <strong>–</strong><br />

<strong>the</strong> celebrated Maxwell’s equati<strong>on</strong>s 7 <strong>–</strong> was published in 1873 by Maxwell in A treatise<br />

<strong>on</strong> electricity <strong>and</strong> magnetism 8 . Heinrich Hertz c<strong>on</strong>firmed Maxwell’s <strong>the</strong>ory by generating<br />

<strong>and</strong> detecting radiowaves in a laboratory setting <strong>and</strong> proving that radiowaves<br />

5<br />

The existence of <strong>the</strong> æ<strong>the</strong>r medium was cast into str<strong>on</strong>g doubt by <strong>the</strong> famous Michels<strong>on</strong>-<br />

Morley experiment (this is <strong>the</strong> Michels<strong>on</strong> with <strong>the</strong> interferometer setup). They reas<strong>on</strong>ed al<strong>on</strong>g <strong>the</strong><br />

following lines: when <strong>the</strong> earth revolves around <strong>the</strong> sun it should produce a substantial wind in <strong>the</strong><br />

æ<strong>the</strong>r medium <strong>–</strong> thus, <strong>the</strong> speed of light would be slightly different depending <strong>on</strong> <strong>the</strong> experiment<br />

(at <strong>the</strong> earth surface) was facing <strong>the</strong> wind (foul wind) or facing from <strong>the</strong> wind (fair wind). The<br />

changes in speed of light, both daily <strong>and</strong> seas<strong>on</strong>al, was expected to be very small <strong>–</strong> hence <strong>the</strong> need<br />

for an interferometer which would split up a <strong>beam</strong> <strong>and</strong> propagate <strong>the</strong> <strong>beam</strong>s al<strong>on</strong>g l<strong>on</strong>g arms <strong>and</strong><br />

recombine <strong>the</strong>m, which would produce a interference pattern if <strong>the</strong> <strong>beam</strong>s did not propagate in<br />

<strong>the</strong> same manner, i.e. a <strong>beam</strong> propagating parallel to <strong>the</strong> æ<strong>the</strong>r wind would propagate slower than<br />

<strong>on</strong>e propagating perpendicular. Michels<strong>on</strong> <strong>and</strong> Morley combined <strong>the</strong>ir efforts in 1887 <strong>and</strong> <strong>the</strong>ir<br />

experiments cast doubt <strong>on</strong>, but did not disprove, <strong>the</strong> existence of <strong>the</strong> æ<strong>the</strong>r medium; <strong>the</strong>ir most<br />

important c<strong>on</strong>tributi<strong>on</strong> remains <strong>the</strong> interferometer that bears <strong>the</strong>ir names.<br />

6<br />

The Fizeau-Focault apparatus c<strong>on</strong>sisted of a light source <strong>and</strong> a two mirrors spaced 35 kilometers<br />

apart. The mirror closet to <strong>the</strong> lightsource was rotating at a c<strong>on</strong>stant angular rate. The elapsed<br />

time for <strong>the</strong> light to pass <strong>the</strong> distance ℓ between <strong>the</strong> mirrors is 2ℓ/c (c being <strong>the</strong> speed of light).<br />

During <strong>the</strong> flight of <strong>the</strong> light <strong>the</strong> moving mirror will have rotated away from its original positi<strong>on</strong>.<br />

The angle at which <strong>the</strong> returning light is observed is <strong>the</strong>n α = dα 2h<br />

dt c . A drawing of <strong>the</strong> original<br />

experiment can be found at: http://imgbase-scd-ulp.u-strasbg.fr/displayimage.php?pos=-212905<br />

(From his collected works Volume Two - Recueil des travaux scientifiques de Lé<strong>on</strong> Foucault,<br />

1878).<br />

7<br />

Stated in detail below (see page 23).<br />

8<br />

The basic equati<strong>on</strong>s was in fact published already in 1865 in a paper entitled A dynamical<br />

<strong>the</strong>ory of <strong>the</strong> electromagnetic field[1].


6 1. Introducti<strong>on</strong><br />

exhibited <strong>the</strong> same properties as light waves, e.g. reflecti<strong>on</strong>, refracti<strong>on</strong>, interference<br />

<strong>and</strong> diffracti<strong>on</strong>.<br />

Electromagnetic radiati<strong>on</strong> can i<strong>on</strong>ize atoms. This was discoveredbyHeinrich Hertz<br />

in 1886 in <strong>the</strong> course of <strong>the</strong> inquires described above.<br />

The apparatus Hertz used was a high voltage inducti<strong>on</strong> coil to create a discharge<br />

between two pieces of brass <strong>and</strong> a piece of copper wire with a brass sphere <strong>on</strong> <strong>on</strong>e end<br />

<strong>and</strong> <strong>on</strong> <strong>the</strong> o<strong>the</strong>r a sharp point directed towards <strong>the</strong> sphere. The basic idea is that<br />

<strong>the</strong> charges in <strong>the</strong> discharge oscillate back <strong>and</strong> forth thus emitting electromagnetic<br />

radiati<strong>on</strong>; if <strong>the</strong> emitted light would create ano<strong>the</strong>r spark between <strong>the</strong> tip of <strong>the</strong> wire<br />

<strong>and</strong> <strong>the</strong> brass sphere light would <strong>the</strong>n be proven to be electromagnetic waves.<br />

The photoelectric effect<br />

During 1886, Hertz carried out a series of experiments with his apparatus showing<br />

that electromagnetic waves were reflected through prisms, that it was polarized etc.,<br />

just <strong>the</strong> same properties as light waves. The <strong>on</strong>ly snag was that it sometimes was very<br />

hard to see <strong>the</strong> tiny spark created at <strong>the</strong> wire tip 9 , to improve this Hertz enclosed<br />

<strong>the</strong> wire in a dark casing <strong>–</strong> which reduced <strong>the</strong> intensity of <strong>the</strong> spark. He so<strong>on</strong> found<br />

out that if <strong>the</strong> part of <strong>the</strong> casing shielding <strong>the</strong> discharge was removed <strong>the</strong> intensity<br />

was not reduced <strong>and</strong> that glass, but not quartz, reduced <strong>the</strong> intensity <strong>–</strong> quartz being<br />

transparent to ultra-violet light.<br />

By using a quartz prism to disperse <strong>the</strong> electromagneticwaves<br />

it was also foundthat<strong>the</strong>greatest intensity<br />

of <strong>the</strong> detected spark was obtained in parts of <strong>the</strong> dispersed<br />

light which were above <strong>the</strong> visible range. Hertz<br />

reported his observati<strong>on</strong>s in Annalen der Physik 10 but<br />

offered noexplanati<strong>on</strong>s of <strong>the</strong>phenomena 11 <strong>–</strong> hera<strong>the</strong>r<br />

c<strong>on</strong>cluded that this phenomen<strong>on</strong> was probably of no<br />

practical use whatsoever <strong>–</strong> as we will see below this<br />

was a ra<strong>the</strong>r pessimistic c<strong>on</strong>clusi<strong>on</strong>.<br />

In 1899, J.J. Thomps<strong>on</strong> observed that negative particles<br />

were emitted when a metal surface was exposed<br />

to ultra-violet light. Later, in 1902, P. v<strong>on</strong> Lenard observed<br />

that <strong>the</strong> emitted particles’ kinetic energy 12 did<br />

E kin<br />

f0<br />

frequency f<br />

Figure 1.2: A minimum frequency of<br />

<strong>the</strong> phot<strong>on</strong>s are required to i<strong>on</strong>ize a<br />

material.<br />

depend <strong>on</strong> <strong>the</strong> color (frequency) of <strong>the</strong> light <strong>and</strong> not <strong>on</strong> <strong>the</strong> intensity of <strong>the</strong> light.<br />

In <strong>on</strong>e of his annus mirabilis (1905) papers Einstein gave a ma<strong>the</strong>matical descripti<strong>on</strong><br />

of <strong>the</strong> photoelectric effect[3]. The i<strong>on</strong>izati<strong>on</strong> was described to be caused by <strong>the</strong><br />

absorpti<strong>on</strong> of a ‘light quantum’ <strong>and</strong> that different materials had different <strong>on</strong>set frequencies<br />

f0 for electr<strong>on</strong> emissi<strong>on</strong> was explained by that <strong>the</strong> size of <strong>the</strong> energy packet<br />

needed to be large enough to overcome <strong>the</strong> first i<strong>on</strong>izati<strong>on</strong> potential of <strong>the</strong> material.<br />

9<br />

As <strong>on</strong>e remedy it was suggested that a suitably prepared frog’s leg would serve equally well<br />

as a detector.<br />

10<br />

In, Über einen Einfluss des ultravioletten Lichtes auf die electrische Entladung[2].<br />

11<br />

The spark in <strong>the</strong> detector was enhanced by charges knocked out from <strong>the</strong> material in <strong>the</strong><br />

detector by phot<strong>on</strong>s from <strong>the</strong> ultra-violet parts of <strong>the</strong> spectrum emitted from <strong>the</strong> discharge.<br />

12<br />

More exactly he measured <strong>the</strong> stopping potential of <strong>the</strong> emitted electr<strong>on</strong>s. He performed<br />

<strong>the</strong> experiment by shining light <strong>on</strong> <strong>the</strong> positively charged plate of a parallel plate capacitor <strong>and</strong><br />

observing <strong>the</strong> potential that causes <strong>the</strong> induced current to become zero.


1.2. Laboratory X-ray <strong>and</strong> UV/Vis phot<strong>on</strong> sources 7<br />

For this work he was awarded <strong>the</strong> Nobel prize of physics 1921 “for his services<br />

to Theoretical Physics, <strong>and</strong> especially for his discovery of <strong>the</strong> law of <strong>the</strong> photoelectric<br />

effect”.<br />

The use of quantized energy-levels was used in 1900 by Max Planck to explain <strong>the</strong><br />

distributi<strong>on</strong> of radiati<strong>on</strong> from a black body. To avoid <strong>the</strong> ultra-violet catastrophe of<br />

classical electrodynamics, i.e. that <strong>the</strong> radiati<strong>on</strong> energy distributi<strong>on</strong> tends to infinity<br />

for short wavelengths, he assumed that <strong>the</strong> energy of <strong>the</strong> emitting oscillators was<br />

quantized. By stating that light c<strong>on</strong>sists of discrete energy packets, Einstein could<br />

formulate an equati<strong>on</strong> that explained <strong>the</strong> photoelectric effect:<br />

εkin = �ω −φ<br />

where <strong>the</strong> kinetic energy of <strong>the</strong> photoelectr<strong>on</strong> is related to <strong>the</strong> frequency of <strong>the</strong> light<br />

<strong>and</strong> <strong>the</strong> work needed to escape <strong>the</strong> material. � is Planck’s c<strong>on</strong>stant (divided by 2π)<br />

<strong>and</strong> ω = 2πf is <strong>the</strong> angular frequency of <strong>the</strong> light. The slope of <strong>the</strong> line in Figure 1.2<br />

is thus Planck’s c<strong>on</strong>stant.<br />

In secti<strong>on</strong> 3.2, we will see how <strong>the</strong> photoelectric effect is utilized as <strong>on</strong>e way to<br />

provide a high quality source of electr<strong>on</strong>s to be used in accelerators. The effect also<br />

forms <strong>the</strong> basis for electr<strong>on</strong> spectroscopy which is an important experimental field as<br />

well as a diagnostic possibility for free electr<strong>on</strong> laser light sources (as elaborated up<strong>on</strong><br />

in Secti<strong>on</strong> 7.3).<br />

1.2 Laboratory X-ray <strong>and</strong> UV/Vis phot<strong>on</strong> sources<br />

Peak brilliance [Phot./(s⋅ mrad 2 ⋅ mm 2 ⋅ 0.1% b<strong>and</strong>w.)]<br />

10 35<br />

10 30<br />

10 25<br />

10 20<br />

10 15<br />

10 10<br />

X-ray tubes<br />

Peak Brilliance of X−ray sources<br />

3rd gen. synchrotr<strong>on</strong>s<br />

2nd gen. synchrotr<strong>on</strong>s<br />

X-FEL<br />

10<br />

1850 1900 1950 2000 2050<br />

5<br />

Time<br />

Average brilliance [Phot./(s ⋅ mrad 2 ⋅ mm 2 ⋅ 0.1% b<strong>and</strong>w.)]<br />

10 30<br />

10 25<br />

10 20<br />

10 15<br />

10 10<br />

X-ray tubes<br />

Average brilliance of X−ray sources<br />

3rd gen. synchrotr<strong>on</strong>s<br />

2nd gen. synchrotr<strong>on</strong>s<br />

1st gen. synchrotr<strong>on</strong>s 1st gen. synchrotr<strong>on</strong>s<br />

X-FEL<br />

10<br />

1850 1900 1950 2000 2050<br />

5<br />

Time<br />

Figure 1.3: Development of <strong>the</strong> brilliance of man-made X-ray sources. The brilliance is a measurement of <strong>beam</strong><br />

quality that measures how directi<strong>on</strong>al <strong>and</strong> pointy a source in combinati<strong>on</strong> by how str<strong>on</strong>gly it emits light within<br />

a certain b<strong>and</strong>width, (see page 19).


8 1. Introducti<strong>on</strong><br />

Vanode<br />

V<br />

- +<br />

X-rays<br />

Figure 1.5: A schematic of an improved Crooks tube Figure 1.1.<br />

Water out<br />

Water in<br />

ThedevelopmentofX-raysources withrespectintensity<strong>and</strong>source qualityhave, as<br />

shown in Figure 1.3 developed superexp<strong>on</strong>entially 13 since <strong>the</strong> end of <strong>the</strong> 19 th century.<br />

A technological development like this is logical, when <strong>on</strong>e c<strong>on</strong>siders <strong>the</strong> accumulated<br />

knowledge of relevant physics <strong>and</strong> related areas 14 .<br />

When <strong>the</strong> X-ray tubes reached <strong>the</strong>ir optimum potential,<br />

that is when <strong>the</strong> effort in terms of time <strong>and</strong> m<strong>on</strong>ey did not<br />

yield enough payback to develop <strong>the</strong> technology fur<strong>the</strong>r <strong>–</strong> accelerator<br />

based synchrotr<strong>on</strong> light sources took over as <strong>the</strong> developingtechnology.<br />

Currentlybothsynchrotr<strong>on</strong>storage rings<br />

<strong>and</strong> free electr<strong>on</strong> laser develop in parallel since <strong>the</strong>y satisfies<br />

different dem<strong>and</strong>s from <strong>the</strong> user community.<br />

1 10<br />

�ω [keV]<br />

Figure 1.4: Generic spec-<br />

X-ray tube <strong>and</strong> anode sources<br />

trum of X-rays from a tube<br />

source.<br />

The X-ray tubes used by Wilhelm Röntgen <strong>and</strong> o<strong>the</strong>rs in <strong>the</strong><br />

years leading up to <strong>and</strong> <strong>the</strong> first years into <strong>the</strong> 1900’s was<br />

limited in intensity (<strong>and</strong> brightness) largely by <strong>the</strong> power that<br />

<strong>the</strong> anode could dissipate without melting.<br />

In Figure 1.5 a schematic of an X-ray tube of a model dating from 1913 is shown.<br />

The electr<strong>on</strong>s are produced by <strong>the</strong>rmi<strong>on</strong>ic emissi<strong>on</strong> 15 from a heated tungsten filament<br />

serving as <strong>the</strong> cathode; <strong>the</strong> electr<strong>on</strong>s are accelerated towards an anode target which<br />

is watercooled. This type of tube can produce powers up to 18 kW.<br />

ThegenericX-rayspectrumfrom asourcewhichgeneratesX-raysbybombardment<br />

of a target with electr<strong>on</strong>s is presented in Figure 1.4. On top of a broad feature ranging<br />

over several keV of phot<strong>on</strong>-energies sharp features emanating from discrete atomic<br />

13 That is, for instance, faster than Moore’s law for <strong>the</strong> transistor density of integrated circuits.<br />

14 The rate of publicati<strong>on</strong> in physics <strong>and</strong> chemistry was also growing faster than an exp<strong>on</strong>ential<br />

curve during <strong>the</strong> first part of <strong>the</strong> 20 th century[4].<br />

15 Discussed fur<strong>the</strong>r in <strong>the</strong> c<strong>on</strong>text of electr<strong>on</strong> guns for free electr<strong>on</strong> laser below in Secti<strong>on</strong> 3.2<br />

(see page 47).


1.2. Laboratory X-ray <strong>and</strong> UV/Vis phot<strong>on</strong> sources 9<br />

transiti<strong>on</strong>s in <strong>the</strong> target which generates X-rays with precisely determined energies.<br />

The broad smooth feature arises from bremsstrahlung from electr<strong>on</strong>s that decelerate<br />

in <strong>the</strong> target <strong>and</strong> <strong>the</strong>reby emits X-rays.<br />

An improvement still up<strong>on</strong> this is <strong>the</strong> rotating anode source where <strong>the</strong> area of<br />

illuminati<strong>on</strong> is increased to allow more efficient cooling of <strong>the</strong> anode by letting it<br />

rotate in vacuum. It is desirable to keep <strong>the</strong> illuminated spot as small as possible<br />

to increase <strong>the</strong> brilliance of <strong>the</strong> source. Since <strong>the</strong> surface temperature of <strong>the</strong> anode<br />

can easily reach above 2000 ◦ C <strong>the</strong> cooling needs to be very efficient <strong>and</strong> <strong>the</strong> rotati<strong>on</strong><br />

speed kept high. Modern rotating anode sources can produce up to 100 kW of X-ray<br />

power thanks to this development.<br />

Rotatinganode sources are amature technology which <strong>the</strong>refore is used for medical<br />

purposes (imaging <strong>and</strong> <strong>the</strong>rapy) <strong>and</strong> as laboratory sources both for scientific research<br />

<strong>and</strong> in industry for material diagnostic purposes.<br />

Synchrotr<strong>on</strong> light sources<br />

First generati<strong>on</strong> synchrotr<strong>on</strong> light sources were particle physics electr<strong>on</strong> storage rings<br />

where <strong>the</strong> synchrotr<strong>on</strong>light producedin <strong>the</strong>dipole magnets (usedtobend<strong>the</strong> electr<strong>on</strong><br />

<strong>beam</strong> in a quasicircular orbit) was used by o<strong>the</strong>rs in a parasitic fashi<strong>on</strong>. For particle<br />

physics experiments, <strong>the</strong> producti<strong>on</strong> of light in <strong>the</strong> accelerating structures is a, more<br />

or less significant, problem. The first observati<strong>on</strong> of man-made synchrotr<strong>on</strong> light was<br />

published in 1947 [5].<br />

Sec<strong>on</strong>d generati<strong>on</strong> synchrotr<strong>on</strong>s were built in <strong>the</strong> early 1970s to dedicatedly supply<br />

synchrotr<strong>on</strong> radiati<strong>on</strong> for a growing number of researchers from various fields; <strong>the</strong><br />

multitude of scientific disciplines attracted by synchrotr<strong>on</strong> radiati<strong>on</strong> can in part be<br />

explained by that <strong>the</strong> emitted spectrum is from <strong>the</strong> infrared to <strong>the</strong> hard x-rays with<br />

an well defined polarisati<strong>on</strong>. The light was still produced in <strong>the</strong> bending magnets<br />

that accelerates <strong>the</strong> electr<strong>on</strong>s so that <strong>the</strong> electr<strong>on</strong>s can be stored in a quasi-circular<br />

orbit.<br />

Third generati<strong>on</strong> synchrotr<strong>on</strong>s are used today<br />

<strong>and</strong> use inserti<strong>on</strong> devices, such as wigglers<br />

<strong>and</strong> undulators[6] to produce radiati<strong>on</strong> with<br />

even higher brilliance <strong>and</strong> power. The inserti<strong>on</strong><br />

devices are periodic magnetic structures (Figure<br />

1.8) which, stated bluntly, makes <strong>the</strong> electr<strong>on</strong>s<br />

turn more often which produces a higher<br />

radiati<strong>on</strong> power than just <strong>on</strong>e bending magnet.<br />

However, since a wiggler do not need to bend<br />

<strong>the</strong> orbit <strong>–</strong> <strong>and</strong> thus ideally should not influence<br />

it <strong>–</strong> <strong>the</strong> magnetic field can be much higher<br />

which not <strong>on</strong>ly increases <strong>the</strong> radiati<strong>on</strong> power<br />

but also shifts <strong>the</strong> wavelength of <strong>the</strong> phot<strong>on</strong>s<br />

upward (which is why wigglers are sometimes<br />

called wavelength shifters). An undulator has<br />

generally a lot more magnetic periods than a<br />

Intensity<br />

Phot<strong>on</strong> energy<br />

Figure 1.6: Typical intensity distributi<strong>on</strong>s from<br />

dipole (dashed), wiggler (red, solid) <strong>and</strong> undulator<br />

(filled) radiati<strong>on</strong>.<br />

wiggler with weaker magnetic fields which keeps <strong>the</strong> deflecti<strong>on</strong>s from <strong>the</strong> central orbit<br />

small such that phot<strong>on</strong>s emitted at an earlier instant can interfere with phot<strong>on</strong>s<br />

emitted at future times which produces a build up of emitted power at certain frequencies.<br />

In chapter 2, we will investigate with some detail <strong>and</strong> rigor <strong>the</strong> properties


10 1. Introducti<strong>on</strong><br />

of undulator radiati<strong>on</strong> <strong>–</strong> as this is a prerequisite for <strong>the</strong> underst<strong>and</strong>ing of <strong>the</strong> free<br />

electr<strong>on</strong> laser formalism.<br />

Figure 1.6 presents how <strong>the</strong> general appearance of <strong>the</strong> spectra from different inserti<strong>on</strong><br />

devices can look like. In <strong>the</strong> case of <strong>the</strong> undulator <strong>the</strong>re is sharp peaks where <strong>the</strong><br />

c<strong>on</strong>diti<strong>on</strong> for positive interference is fulfilled. The peaks are distributed at multiples<br />

of <strong>the</strong> first res<strong>on</strong>ance (i.e. harm<strong>on</strong>ics of <strong>the</strong> fundamental energy where <strong>the</strong> c<strong>on</strong>diti<strong>on</strong><br />

is first fulfilled).<br />

Synchrotr<strong>on</strong> storage rings provide almost a c<strong>on</strong>tinuous source of radiati<strong>on</strong> with<br />

repetiti<strong>on</strong> rates in <strong>the</strong> 100’s of MHz something which gives rise to high average power.<br />

The high repetiti<strong>on</strong> rate is often too much for time-resolved measurements which<br />

is why synchrotr<strong>on</strong>s are sometimes run in ”low filling modes” where <strong>on</strong>ly <strong>on</strong>e or a few<br />

bunches of electr<strong>on</strong>s travel around <strong>the</strong> orbit which brings down <strong>the</strong> repetiti<strong>on</strong> rate to<br />

<strong>the</strong> order of 1 MHz. The bunches are still relatively l<strong>on</strong>g (in <strong>the</strong> order of picosec<strong>on</strong>ds);<br />

experiments dem<strong>and</strong>ing short pulses can use femtosec<strong>on</strong>d laser slicing of <strong>the</strong> electr<strong>on</strong><br />

bunches before <strong>the</strong>y pass through a bending magnet, or undulator, this modulates<br />

part of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>and</strong> causes that part of <strong>the</strong> <strong>beam</strong> to emit a short pulse<br />

which is separated from <strong>the</strong> radiati<strong>on</strong> fro <strong>the</strong> majority part of <strong>the</strong> bunch.<br />

Onecan c<strong>on</strong>cludethat synchrotr<strong>on</strong>radiati<strong>on</strong> hasmany attractive properties which,<br />

as menti<strong>on</strong>ed, have made its use widespread over a broad range of scientific communities,<br />

i.e. tunable light with an (extremely) well defined polarizati<strong>on</strong> (which can be<br />

linear, circular or elliptical) available over a very large range of energies. It also have<br />

an inherent timestructure that can be manipulated to provide relatively l<strong>on</strong>g x-ray<br />

pulses in <strong>the</strong> MHz repetiti<strong>on</strong> rate domain; in combinati<strong>on</strong> with lasers o<strong>the</strong>r timestrutures<br />

o<strong>the</strong>r types of time-resolved experiments are enables. A (significant) fracti<strong>on</strong> of<br />

<strong>the</strong> light is also transversely coherent, which enable imaging experiments making use<br />

of <strong>–</strong> for instance <strong>–</strong> phase c<strong>on</strong>trast at X-ray wavelengths.<br />

Lasers<br />

The <strong>the</strong>oretical framework for <strong>the</strong> Laser was laid down by Albert Einstein in his<br />

Zur Quanten<strong>the</strong>orie der Strahlung [7] from 1917. In this paper he detailed how light<br />

quanta are absorbed <strong>and</strong> emitted by atoms 16 . By introducing probabilities for <strong>the</strong><br />

processes of absorpti<strong>on</strong>, sp<strong>on</strong>taneous emissi<strong>on</strong> <strong>and</strong> stimulated emissi<strong>on</strong> he was able<br />

to quantify how atomic spectral lines was formed.<br />

The processes of absorpti<strong>on</strong> <strong>and</strong> emissi<strong>on</strong> of light works in <strong>the</strong> intuitive way; Stimulated<br />

emissi<strong>on</strong> describes how <strong>the</strong> presence of electromagnetic radiati<strong>on</strong> causes atoms<br />

in a higher energy state to decay into a lower <strong>on</strong>e. One can <strong>the</strong>n imagine a process<br />

where, if we were to pump atoms to a higher energy state, decays from this state will<br />

<strong>the</strong>n in turn stimulate o<strong>the</strong>r atoms in <strong>the</strong> ensemble to decay to.<br />

If, in a medium, <strong>the</strong> number of atoms in <strong>the</strong> higher energy state is larger than<br />

<strong>the</strong> number in <strong>the</strong> lower energy state <strong>the</strong> amount of stimulated emissi<strong>on</strong> is larger<br />

than that absorbed in <strong>the</strong> ensable <strong>–</strong> <strong>the</strong> amount of light in <strong>the</strong> medium is amplified.<br />

By placing this medium between two mirrors, an optical res<strong>on</strong>ator, <strong>the</strong> light passes<br />

through <strong>the</strong> gain medium many times before it is extracted somehow, this gives rise to<br />

an even more efficient amplficati<strong>on</strong>: we have a proper laser. The word (or acr<strong>on</strong>ym)<br />

Laser can be spelled out to Light Amplificati<strong>on</strong> by Stimulated Emissi<strong>on</strong> of Radiati<strong>on</strong>.<br />

16 Building <strong>on</strong> Max Planck’s seminal paper from 1901 Ueber das Gesetz der Energieverteilung<br />

im Normalspectrum[8] c<strong>on</strong>sidered to be <strong>on</strong>e of <strong>the</strong> birthplaces for quantum <strong>the</strong>ory.


1.2. Laboratory X-ray <strong>and</strong> UV/Vis phot<strong>on</strong> sources 11<br />

The first laser was built in 1960 by T. H. Maiman <strong>and</strong> c<strong>on</strong>sisted of an solid stateflashlamp<br />

pumped artificially grown ruby crystal (emitting red laser light at 694<br />

nanometer wavelength)[9]. Shortly <strong>the</strong>reafter a laser with a gasmixture as a gain<br />

medium was dem<strong>on</strong>strated.<br />

The number of available laser media is large 17 <strong>and</strong> currently covers a wavelength<br />

range between millimeters down to below 200 nm.<br />

There is a multitude of n<strong>on</strong>-linear optical processes that can be used to manipulate<br />

laser light. Of particular interest for <strong>the</strong> lasers to use toge<strong>the</strong>r with free electr<strong>on</strong> lasers<br />

is harm<strong>on</strong>ic generati<strong>on</strong> of shorter wavelengths. This phenomen<strong>on</strong> was first discovered<br />

in a quartz crystal in 1961[10], where integer multiples of <strong>the</strong> driving laser’s frequency<br />

was observed.<br />

Third harm<strong>on</strong>ic generati<strong>on</strong> using a gas as a medium was observed a few years<br />

later[11]. The intensity of <strong>the</strong> generated harm<strong>on</strong>ics drops very fast <strong>–</strong> <strong>the</strong> process can<br />

be understood (in <strong>the</strong> regime of weak fields) as an atom absorbs several phot<strong>on</strong>s which<br />

in turn are emitted as <strong>on</strong>e; <strong>the</strong> probability of absorbing n phot<strong>on</strong>s drops with n.<br />

High Harm<strong>on</strong>ic Generati<strong>on</strong> Lasers<br />

If <strong>on</strong>e ventures out of <strong>the</strong> weak field regime <strong>the</strong>re is a possibility for ano<strong>the</strong>r process<br />

to occur: high harm<strong>on</strong>ic generati<strong>on</strong>. Laser light is sh<strong>on</strong>e into a gas sustaining high<br />

enough energy density (typically in <strong>the</strong> order of 10 14 W/cm 2 a fracti<strong>on</strong> of <strong>the</strong> laser<br />

power can be c<strong>on</strong>verted into (odd) higher harm<strong>on</strong>ics of <strong>the</strong> original laser pulse. This<br />

allows for <strong>the</strong> creati<strong>on</strong> of UV <strong>and</strong> even soft X-ray pulses. Typically <strong>the</strong> repetiti<strong>on</strong><br />

rates for such systems range from a few Hz to KHz (<strong>the</strong> same as <strong>the</strong> driving laser)<br />

<strong>and</strong> even attosec<strong>on</strong>d pulses can be created.<br />

High harm<strong>on</strong>ic generati<strong>on</strong> can be understood via a semi-classical picture[12]: A<br />

sufficiently str<strong>on</strong>g laser field can perturb atomic potentials enough to allow <strong>the</strong> outermost<br />

electr<strong>on</strong>s (illustrated as a red wave-packet in Figure 1.7).<br />

This allows <strong>the</strong> electr<strong>on</strong> wave-functi<strong>on</strong> to tunnel out of <strong>the</strong> atomic potential into<br />

<strong>the</strong> c<strong>on</strong>tinuum (2 in <strong>the</strong> figure) during <strong>the</strong> first half cycle of <strong>the</strong> laser pulse; during<br />

<strong>the</strong> sec<strong>on</strong>d part of <strong>the</strong> laser cycle <strong>the</strong> electr<strong>on</strong> wavefuncti<strong>on</strong> finds itself <strong>on</strong> a str<strong>on</strong>gly<br />

attractive potential leading back to <strong>the</strong> atom where it came from <strong>–</strong> up<strong>on</strong> recombinati<strong>on</strong><br />

<strong>the</strong> system will emit <strong>the</strong> excess energy as a high energy phot<strong>on</strong> with significantly<br />

higher energy than that of <strong>the</strong> driving laser. There is also <strong>the</strong> possibility of higher<br />

harm<strong>on</strong>ics of this moti<strong>on</strong> to occur <strong>and</strong> <strong>the</strong>nce higher phot<strong>on</strong> energies.<br />

Unlike in <strong>the</strong> weak field regime <strong>the</strong> intensity of <strong>the</strong> higher harm<strong>on</strong>ics do not drop<br />

with <strong>the</strong> harm<strong>on</strong>ic number in a simple decreasing manner. Indeed, <strong>the</strong> higher harm<strong>on</strong>ics<br />

have roughly <strong>the</strong> same intensities up until a cut-off energy.<br />

This cut-off energy can be understood from <strong>the</strong> recombinati<strong>on</strong> model menti<strong>on</strong>ed<br />

above, with <strong>the</strong> i<strong>on</strong>izati<strong>on</strong> potential of <strong>the</strong> medium being Ip:<br />

Ecut-off = Ip +3.17·Up<br />

<strong>and</strong> <strong>the</strong> p<strong>on</strong>dermotive energy being <strong>the</strong> average energy of a free electr<strong>on</strong> in <strong>the</strong> linearly<br />

polarized laser field E (with angular frequency ω.<br />

Up = e2 E 2<br />

4meω 2<br />

17 Diagram of laser lines from commercially available sources:<br />

http://en.wikipedia.org/wiki/File:Commercial laser lines.svg


12 1. Introducti<strong>on</strong><br />

Laser field<br />

1<br />

3<br />

X-ray phot<strong>on</strong><br />

Figure 1.7: Illusrati<strong>on</strong> of <strong>the</strong> high harm<strong>on</strong>ic generati<strong>on</strong> process.<br />

This process is effective (that is, it works) for linearly polarized light <strong>–</strong> elliptically<br />

polarized light accelerates <strong>the</strong> electr<strong>on</strong>s in such a way that it misses <strong>the</strong> i<strong>on</strong>ized atom<br />

<strong>on</strong> <strong>on</strong> <strong>the</strong> returning path so no recombinati<strong>on</strong> occur. At very high energy densities<br />

(10 16 W/cm 2 ) <strong>the</strong> ”magnetic” term in <strong>the</strong> Lorentz force equati<strong>on</strong> (Equati<strong>on</strong> 2.5)<br />

becomes significant, which causes <strong>the</strong> accelerati<strong>on</strong> to deviate from <strong>the</strong> intended return<br />

path.<br />

In <strong>the</strong> c<strong>on</strong>text of free electr<strong>on</strong> lasers, ordinary lasers <strong>and</strong> high harm<strong>on</strong>ic generati<strong>on</strong><br />

lasers are interesting both for experiments in combinati<strong>on</strong> with free electr<strong>on</strong> laser radiati<strong>on</strong>,i.e.<br />

two-color experiments as discussed in chapter 10, <strong>and</strong> as a way to increase<br />

<strong>the</strong> quality of <strong>the</strong> free electr<strong>on</strong> laser light in ways that will be elaborated up<strong>on</strong> below<br />

(see page 15).<br />

1.3 Free electr<strong>on</strong> Lasers<br />

In a c<strong>on</strong>venti<strong>on</strong>al laser <strong>the</strong> average output power is limited by how much of <strong>the</strong> unused<br />

power (which is significantly larger than <strong>the</strong> output power) that can be dissipated by<br />

<strong>the</strong> active medium. Moreover <strong>the</strong> light from a laser is seldom diffracti<strong>on</strong> limited<br />

owing to heat effects in <strong>the</strong> lasing medium <strong>and</strong> n<strong>on</strong>-linear processes taking place in<br />

<strong>the</strong> medium.<br />

C<strong>on</strong>trasting this is <strong>the</strong> free electr<strong>on</strong> laser process which can be close to unity in<br />

efficiency. In a free electr<strong>on</strong> laser <strong>the</strong> amplificati<strong>on</strong> of <strong>the</strong> electromagnetic field occurs<br />

by <strong>the</strong> interacti<strong>on</strong> between an electr<strong>on</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> radiati<strong>on</strong> field it creates when<br />

moving through a periodic magnetic structure. Hence <strong>the</strong> operating wavelength is<br />

tunable via machine parameters such as electr<strong>on</strong> <strong>beam</strong> energy, <strong>and</strong> magnetic field<br />

strength.<br />

Figure 1.8 depicts three different ways of producing free electr<strong>on</strong> laser radiati<strong>on</strong>:<br />

bunches in a storage ring passes through a l<strong>on</strong>g undulator; an oscillator where <strong>the</strong><br />

interacti<strong>on</strong> where <strong>the</strong> electromagnetic radiati<strong>on</strong> interacts with <strong>the</strong> electr<strong>on</strong> bunches<br />

many times; an amplifier, where <strong>the</strong> electr<strong>on</strong>s pass <strong>on</strong>ce through an l<strong>on</strong>g undulator<br />

structure <strong>–</strong> <strong>the</strong> interacti<strong>on</strong> between <strong>the</strong> electromagnetic field <strong>and</strong> <strong>the</strong> electr<strong>on</strong> <strong>beam</strong><br />

is str<strong>on</strong>g enough for <strong>on</strong>e pass to be sufficient.<br />

2


1.3. Free electr<strong>on</strong> Lasers 13<br />

Figure 1.8: Different ways of generating coherent laser radiati<strong>on</strong>. From left to right, electr<strong>on</strong>s are fed into a<br />

l<strong>on</strong>g undulator ei<strong>the</strong>r from a storage ring or in a storage ring but with mirrors that reflect part of <strong>the</strong> pulse to<br />

modulate <strong>the</strong> electr<strong>on</strong> bunch even fur<strong>the</strong>r, or from a linear accelerator where <strong>the</strong> electr<strong>on</strong> bunch gets modulated<br />

by <strong>the</strong> light field it generates.<br />

There is a number of free electr<strong>on</strong> lasers operating in <strong>the</strong> world today (chapter 11),<br />

covering light wavelengths from <strong>the</strong> infrared to <strong>the</strong> x-ray regi<strong>on</strong>s. This book will cover<br />

free electr<strong>on</strong> lasers that are providing light in <strong>the</strong> X-ray range, which are amplifiers<br />

operating in <strong>the</strong> high-gain regime.<br />

Sase<br />

The physical process that governs <strong>the</strong> functi<strong>on</strong> of a free electr<strong>on</strong> laser is abbreviates<br />

Sase <strong>–</strong> Self-Amplified Sp<strong>on</strong>taneous Emissi<strong>on</strong>. In <strong>the</strong> next chapter we will discuss this<br />

in more detail, here follows a short introducti<strong>on</strong>.<br />

The spectrum from <strong>the</strong> phot<strong>on</strong>s emitted from an electr<strong>on</strong> bunch traveling through<br />

an undulator will c<strong>on</strong>tain a large degree of incoherent radiati<strong>on</strong> <strong>and</strong> a small part of<br />

coherent radiati<strong>on</strong>. The latter occurs since a small number of electr<strong>on</strong>s, by chance,<br />

happentoberadiatingcoherently. In<strong>the</strong>undulatorspectrumthus, anumberofspikes<br />

will be seen <strong>on</strong> top of <strong>the</strong> broad sp<strong>on</strong>taneous emissi<strong>on</strong> spectrum. As <strong>the</strong> occurrence<br />

Log radiati<strong>on</strong> power<br />

Distance<br />

Figure 1.9: Growth of <strong>the</strong> radiated power in a Sase-mode as a functi<strong>on</strong> of travelled distance in <strong>the</strong> undulator.<br />

The process saturates when <strong>the</strong> microbunching is maximal.


14 1. Introducti<strong>on</strong><br />

Phot<strong>on</strong>s/s 0.03% BW<br />

10 25<br />

10 23<br />

10 21<br />

10 19<br />

1 10 100 1000<br />

Phot<strong>on</strong> energy [eV]<br />

Figure 1.10: Computed Phot<strong>on</strong> flux into 0.03% b<strong>and</strong>width at 3.63 kA at <strong>the</strong> Lcls at 1.5 ˚A wavelength. The<br />

two spikes are <strong>the</strong> 1 st <strong>and</strong> 3 rd harm<strong>on</strong>ics of <strong>the</strong> fundamental Sase mode, which sits <strong>on</strong> a broad sp<strong>on</strong>taneous<br />

radiati<strong>on</strong> background.<br />

of <strong>the</strong> coherent radiati<strong>on</strong> is r<strong>and</strong>om <strong>the</strong> number of coherently radiating modes per<br />

electr<strong>on</strong> bunch will follow a Poiss<strong>on</strong> distributi<strong>on</strong>.<br />

The radiati<strong>on</strong> field that is created from <strong>the</strong> accelerati<strong>on</strong> of <strong>the</strong> electr<strong>on</strong> bunch<br />

through an undulator in a storage ring is often c<strong>on</strong>sidered to be weak enough as to<br />

not have an effect <strong>on</strong> <strong>the</strong> electr<strong>on</strong> bunch, i.e. <strong>the</strong> electr<strong>on</strong> bunches do not interact<br />

with <strong>the</strong> radiati<strong>on</strong> field. This is true if <strong>the</strong> electr<strong>on</strong> density is low enough <strong>and</strong> if <strong>the</strong><br />

overlap between <strong>the</strong> radiati<strong>on</strong> field <strong>and</strong> <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> is small.<br />

However, if <strong>the</strong> electr<strong>on</strong> density is high enough <strong>and</strong> if <strong>the</strong> quality, (called emittance<br />

18 ), of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> is high <strong>the</strong> interacti<strong>on</strong> between <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>and</strong><br />

<strong>the</strong> radiated field can become substantial.<br />

If<strong>the</strong>interacti<strong>on</strong>between<strong>the</strong>electr<strong>on</strong>s in<strong>the</strong>bunch<strong>and</strong><strong>the</strong>radiati<strong>on</strong>fieldisstr<strong>on</strong>g<br />

enough amicrobunchingof <strong>the</strong> electr<strong>on</strong> bunchoccurs. This means that as <strong>the</strong> electr<strong>on</strong><br />

travelsal<strong>on</strong>g <strong>the</strong>undulatorstructure<strong>the</strong>electr<strong>on</strong> densitybecomes modulatedwith <strong>the</strong><br />

wavelength of <strong>the</strong> radiati<strong>on</strong> field. This enhances <strong>the</strong> coherent emissi<strong>on</strong> fur<strong>the</strong>r, which<br />

in turn enhances <strong>the</strong> micro-bunching <strong>and</strong> amplifies <strong>the</strong> radiati<strong>on</strong> field. The radiati<strong>on</strong><br />

mode thus gets amplified <strong>and</strong> <strong>the</strong> degree of coherence increases. The growth of <strong>the</strong><br />

radiati<strong>on</strong> mode’s strength is exp<strong>on</strong>ential until <strong>the</strong> process saturates.<br />

Free electr<strong>on</strong> laser radiati<strong>on</strong> has a number of unique features owing to <strong>the</strong> amplificati<strong>on</strong><br />

process outlined above. In Figure 1.10 an example of <strong>the</strong> brilliance from <strong>the</strong><br />

Lcls free electr<strong>on</strong> laser source is shown. The peak intensity <strong>and</strong> brilliance is many<br />

orders of magnitude larger than can be produced by o<strong>the</strong>r sources (Figure 1.3). The<br />

18 See <strong>the</strong> discussi<strong>on</strong> <strong>on</strong> emittance below.


1.4. Development of X-ray free electr<strong>on</strong> lasers 15<br />

average brilliance, which is limited by<strong>the</strong> (generally) low repetiti<strong>on</strong> rate of <strong>the</strong> driving<br />

linear accelerators are still orders of magnitudes above that of synchrotr<strong>on</strong> storage<br />

rings. In <strong>the</strong> figure <strong>the</strong> two sharp spikes mark <strong>the</strong> fundamental <strong>and</strong> third harm<strong>on</strong>ics<br />

of <strong>the</strong> Sase radiati<strong>on</strong> modes. Those spikes sit <strong>on</strong> top of a significant sp<strong>on</strong>taneous radiati<strong>on</strong><br />

background that encompasses several orders of magnitude of phot<strong>on</strong> energies.<br />

• Pulse lengths down to tens of femtosec<strong>on</strong>ds are also orders of magnitude shorter<br />

than those of a synchrotr<strong>on</strong>. Ordinary lasers <strong>and</strong> high harm<strong>on</strong>ic generati<strong>on</strong><br />

lasers can surpass this figure but can not deliver <strong>the</strong> brilliance at X-ray wavelengths<br />

c<strong>on</strong>sidered here.<br />

• Free electr<strong>on</strong> laser radiati<strong>on</strong> has full transverse coherence (if <strong>the</strong> Sase process<br />

has reached saturati<strong>on</strong>), i.e. it is diffracti<strong>on</strong> limited.<br />

The Sase process <strong>on</strong>ly amplifies certain modes at <strong>the</strong> time (<strong>and</strong> possibly <strong>the</strong>ir harm<strong>on</strong>ics)<br />

The sp<strong>on</strong>taneous radiati<strong>on</strong> spectrum extends very high up in phot<strong>on</strong> energy,<br />

towards 1 MeV.<br />

1.4 Development of X-ray free electr<strong>on</strong> lasers<br />

The c<strong>on</strong>cept of free electr<strong>on</strong> lasing was developed in <strong>the</strong> early 1970s[13]. The predicti<strong>on</strong><br />

that sp<strong>on</strong>taneous emissi<strong>on</strong> from electr<strong>on</strong> bunches traveling through a periodic<br />

magnetic fieldcould experience exp<strong>on</strong>ential gain was dem<strong>on</strong>strated 1977[14]. The first<br />

free electr<strong>on</strong> laser to reach saturati<strong>on</strong> was <strong>the</strong> Leutl free electr<strong>on</strong> laser located at <strong>the</strong><br />

Advanced Phot<strong>on</strong> Source at <strong>the</strong> Arg<strong>on</strong>ne Nati<strong>on</strong>al Laboratory, Illinois, USA[15].<br />

During<strong>the</strong>1980s <strong>and</strong>1990s asignificantresearcheffortwas c<strong>on</strong>ductedastodevelop<br />

<strong>the</strong> <strong>the</strong>ory for free electr<strong>on</strong> laser as to investigate <strong>the</strong> feasibility of increasing <strong>the</strong><br />

phot<strong>on</strong> energies into <strong>the</strong> UV <strong>and</strong> X-ray ranges[16<strong>–</strong>18]. The first lasing of a hard<br />

X-ray free electr<strong>on</strong> laser occurred in 2009[19].<br />

Around <strong>the</strong> world <strong>the</strong>re is a number of laboratories harboring free electr<strong>on</strong> lasers<br />

operating with phot<strong>on</strong> energies across <strong>the</strong> electromagnetic spectrum, from <strong>the</strong> microwave<br />

regi<strong>on</strong> to <strong>the</strong> hard X-ray regi<strong>on</strong>s.<br />

Facilities for <strong>the</strong> VUV/X-ray regi<strong>on</strong> dem<strong>and</strong> ra<strong>the</strong>r high electr<strong>on</strong> <strong>beam</strong> energies,<br />

thus linear electr<strong>on</strong> accelerators (retired from use for particle physics experiments)<br />

have been fitted with undulators <strong>and</strong> hence c<strong>on</strong>verted to free electr<strong>on</strong> laser; Flash<br />

in Hamburg <strong>and</strong> Lcls in Stanford are from this category. Dedicated facilites for <strong>the</strong><br />

X-ray range are also being deployed, for instance <strong>the</strong> Fermi@Elettra in Italy <strong>and</strong> <strong>the</strong><br />

Scss in Japan. In chapter 11 currently operating facilites <strong>and</strong> some of those under<br />

various stages of commissi<strong>on</strong>ing <strong>and</strong> planning are described in more detail.<br />

1.5 Seeding schemes<br />

As discussed previously, <strong>the</strong> Sase process starts up from shot-noise in <strong>the</strong> electr<strong>on</strong><br />

<strong>beam</strong> <strong>–</strong> thus, even though <strong>the</strong> transverse coherence is very good (optimal) <strong>the</strong> l<strong>on</strong>gitudinal<br />

coherence is poor. This means that <strong>the</strong> phot<strong>on</strong> spectrum is different for<br />

each pulse, both regarding <strong>the</strong> number of modes that are radiating, which frequencies<br />

dominate <strong>the</strong> spectrum <strong>and</strong> <strong>the</strong> <strong>beam</strong> energy.<br />

This can be seen in Figure 1.12: <strong>the</strong> frequency <strong>and</strong> intensity distributi<strong>on</strong> vary <strong>on</strong> a<br />

shot-to-shot basis. Intuitively this phenomen<strong>on</strong> can be understood by c<strong>on</strong>sidering <strong>the</strong>


16 1. Introducti<strong>on</strong><br />

10 3<br />

10 −2<br />

10 −5<br />

0.5×10 −6<br />

10 −8<br />

10 −10<br />

10 −12<br />

Name Radio Microwave Infrared Visible UV X-ray Gamma<br />

Wavelength [m]<br />

Frequency [Hz]<br />

Corresp<strong>on</strong>ding<br />

temperature of<br />

radiating blackbody<br />

Buildings Humans Butterflies Needle Point Protozoans Molecules Atoms Atomic Nuclei<br />

10 4<br />

10 8<br />

10 12<br />

10 15<br />

10 16<br />

10 18<br />

1 K 100 K 10,000 K 10,000,000 K<br />

−272 °C −173 °C 9,727 °C ~10,000,000 °C<br />

Figure 1.11: Wavelengths <strong>and</strong> frequencies in <strong>the</strong> electromagnetic spectrum.<br />

Figure 1.12: The Sase phot<strong>on</strong> spectrum from <strong>the</strong> Flash free electr<strong>on</strong> laser.<br />

stochastic start of <strong>the</strong> amplificati<strong>on</strong> process; in <strong>the</strong> beginning of <strong>the</strong> undulator many<br />

modes radiate energy, those serve as seeding radiati<strong>on</strong> for <strong>the</strong> durati<strong>on</strong> of <strong>the</strong> traversing<br />

of <strong>the</strong> (l<strong>on</strong>g) remainder of <strong>the</strong> undulator. The modes that ”fit” <strong>the</strong> res<strong>on</strong>ance<br />

c<strong>on</strong>diti<strong>on</strong> for <strong>the</strong> undulator spectrum will get progressively more amplified whereas<br />

those modes that are not radiating res<strong>on</strong>antly do not gain in energy. The res<strong>on</strong>ance<br />

c<strong>on</strong>diti<strong>on</strong> may be fulfilled by several modes simultaneously, more or less well which<br />

give rise to many spikes in <strong>the</strong> final spectrum. For each new bunch <strong>the</strong> process start<br />

10 20


1.5. Seeding schemes 17<br />

over again <strong>and</strong> thus a new set of modes arise from <strong>the</strong> stochastic start-up.<br />

Naturally <strong>the</strong>re is a desire to have a more stable free electr<strong>on</strong> laser phot<strong>on</strong> spectrum.<br />

Both <strong>the</strong> jitter in time between pulses, <strong>the</strong> pulse energy <strong>and</strong> <strong>the</strong> spectral<br />

c<strong>on</strong>tent prohibit <strong>the</strong> maximum performance of both <strong>the</strong> free electr<strong>on</strong> laser itself <strong>and</strong><br />

<strong>the</strong> possible experiments that can be performed.<br />

To circumvent <strong>the</strong> stochastic startup of <strong>the</strong> radiati<strong>on</strong>-field amplificati<strong>on</strong> <strong>the</strong> most<br />

straightforward way <strong>–</strong> at least c<strong>on</strong>ceptually <strong>–</strong> is to pre-modulate <strong>the</strong> electr<strong>on</strong> <strong>beam</strong>’s<br />

energy with a str<strong>on</strong>g laser field, <strong>the</strong>n c<strong>on</strong>vert this energy modulati<strong>on</strong> into a density<br />

modulati<strong>on</strong>. The c<strong>on</strong>versi<strong>on</strong> between energy <strong>and</strong> density modulati<strong>on</strong> can be d<strong>on</strong>e in<br />

a magnetic structure (chicane or wiggler/undulator) since <strong>the</strong> different parts of <strong>the</strong><br />

bunch take different paths through such a structure.<br />

eSase<br />

Figure 1.13: Energy modulati<strong>on</strong> of an electr<strong>on</strong> bunch.<br />

If <strong>the</strong> electr<strong>on</strong> bunch are modulated with a laser in a wiggler <strong>and</strong> subsequently pass<br />

through an undulator structure significantly shorter gain length (<strong>and</strong> thus shorter<br />

undulators) can be achieved as compared to normal Sase operati<strong>on</strong> (hence Enhanced-<br />

Sase)[20].<br />

HGHG<br />

ESASE<br />

Modulator Radiator<br />

High-Gain Harm<strong>on</strong>ic-Generati<strong>on</strong> is a frequency upc<strong>on</strong>versi<strong>on</strong> scheme, designed to upc<strong>on</strong>vert<br />

<strong>the</strong> fundamental frequency of <strong>the</strong> laser to a much higher frequency[21].<br />

This scheme has been dem<strong>on</strong>strated[22] <strong>and</strong> it includes a short modulator where<br />

<strong>the</strong> energy-densityc<strong>on</strong>versi<strong>on</strong> starts, followed by a chicane (two bendingmagnets that<br />

bends away <strong>and</strong> returns <strong>the</strong> <strong>beam</strong> al<strong>on</strong>g <strong>the</strong> orignal path). The chicane compresses<br />

<strong>the</strong> bunch fur<strong>the</strong>r which enhances <strong>the</strong> density modulati<strong>on</strong> even more before <strong>the</strong> <strong>beam</strong><br />

enters <strong>the</strong> sec<strong>on</strong>d undulator (<strong>the</strong> radiator) where <strong>the</strong> free electr<strong>on</strong> laser process takes<br />

place. Since <strong>the</strong> electr<strong>on</strong> bunch is pre-modulated when it enters <strong>the</strong> radiator <strong>the</strong><br />

spectrum from this type of free electr<strong>on</strong> laser is significantly more intense in <strong>the</strong><br />

fundamental mode (up to 10 6 times) <strong>and</strong> narrow (since, ideally, all <strong>the</strong> spectral intensity<br />

is put into <strong>on</strong>e mode <strong>and</strong> its harm<strong>on</strong>ics). The shot to shot repeatability is also


18 1. Introducti<strong>on</strong><br />

much better as <strong>the</strong> free electr<strong>on</strong> laser pulse is a up-c<strong>on</strong>verted versi<strong>on</strong> of <strong>the</strong> original<br />

laser pulse (at least <strong>the</strong> part of <strong>the</strong> spectrum arising from <strong>the</strong> part of <strong>the</strong> electr<strong>on</strong><br />

bunch that became modulated)[22]. The Hghg scheme can of course be cascaded,<br />

HGHG<br />

Dispersive secti<strong>on</strong><br />

Modulator Radiator<br />

that is putting several stages in fr<strong>on</strong>t of each o<strong>the</strong>r, to achieve increasingly shorter<br />

wavelengths.<br />

EEHG<br />

The Echo-Enabled Harm<strong>on</strong>ic Generati<strong>on</strong> free electr<strong>on</strong> laser scheme[23] have recently<br />

been dem<strong>on</strong>strated experimentally at <strong>the</strong> Next Linear Collider Test Accelerator at<br />

<strong>the</strong> SLAC Nati<strong>on</strong>al Accelerator Laboratory, Stanford, USA[24].<br />

The principle is similar to that of <strong>the</strong> Hghg scheme above with <strong>the</strong> important<br />

difference that two modulators with two different laser seedings is used before <strong>the</strong> undulator.<br />

This enables more intricate c<strong>on</strong>trol of <strong>the</strong> path differences <strong>the</strong> particles with<br />

different energies takes in <strong>the</strong> two different chicanes <strong>–</strong> allowing <strong>the</strong> density modulati<strong>on</strong><br />

of <strong>the</strong> electr<strong>on</strong>s to occur at a shorter frequency than <strong>the</strong> original laser pulses; at <strong>the</strong><br />

entrance of <strong>the</strong> radiator <strong>on</strong>e can get a pre-bunched electr<strong>on</strong> <strong>beam</strong> for a significantly<br />

shorter wavelength than any of those of <strong>the</strong> seed-lasers.<br />

Harm<strong>on</strong>ic afterburners<br />

Harm<strong>on</strong>ic afterburners c<strong>on</strong>sist of <strong>on</strong>e or more undulators placed after <strong>the</strong> main radiator.<br />

They are ei<strong>the</strong>r used to enhance <strong>the</strong> radiated power in a harm<strong>on</strong>ic of <strong>the</strong><br />

fundamental <strong>–</strong> thus reaching shorter wavelengths[25]. Extra undulators can also be<br />

used to c<strong>on</strong>trol <strong>the</strong> polarizati<strong>on</strong> of <strong>the</strong> emitted light to some degree[26, 27].


1.6. Definiti<strong>on</strong>s used throughout <strong>the</strong> book 19<br />

1.6 Definiti<strong>on</strong>s used throughout <strong>the</strong> book<br />

Brilliance <strong>and</strong> Brightness<br />

The intensity of a source can be regarded as <strong>the</strong> flow of energy per unit time per unit<br />

source area<br />

I = dE<br />

dtdxdy<br />

The flux of a source Φ<br />

Φ =<br />

�<br />

1 dI<br />

ω dω dω<br />

is defined as <strong>the</strong> number of phot<strong>on</strong>s per unit time, per unit surface area of <strong>the</strong> source.<br />

As figures of merit for a light source <strong>the</strong> intensity <strong>and</strong> flux are ra<strong>the</strong>r blunt tools<br />

since <strong>the</strong>y say nothing about <strong>the</strong> directi<strong>on</strong>ality of <strong>the</strong> source. If we differentiate <strong>the</strong><br />

flux with respect to <strong>the</strong> solid angle dΩ we obtain <strong>the</strong> brightness:<br />

B = dΦ<br />

dΩ<br />

with units [phot<strong>on</strong>s/s/mm 2 /mrad].<br />

Brilliance, or spectral brightness[28] is defined as <strong>the</strong> number of phot<strong>on</strong>s emitted<br />

per unit time, per unit solid angle, per unit source area inside a b<strong>and</strong>width chosen to<br />

be 0.1%<br />

Br = d2 Φ<br />

dωdΩ<br />

thus <strong>the</strong> unit [phot<strong>on</strong>s/s/mm 2 /mrad/0.1%BW]. This defines <strong>the</strong> brightness of a<br />

source inside a certain frequency envelope centered around a certain frequency ω.<br />

Spectral brightness is closely related to <strong>the</strong> emittance of <strong>the</strong> of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong><br />

source. The emittance is <strong>the</strong> product of <strong>the</strong> <strong>beam</strong> divergence <strong>and</strong> <strong>the</strong> transverse size<br />

of <strong>the</strong> <strong>beam</strong> al<strong>on</strong>g each directi<strong>on</strong> perpendicular to <strong>the</strong> propagati<strong>on</strong> of <strong>the</strong> electr<strong>on</strong>s.<br />

Emittance<br />

When an ensemble of charged particles propagates through an accelerator <strong>the</strong>y move<br />

al<strong>on</strong>g an orbit through <strong>the</strong> accelerator structure (composed of guiding magnets <strong>and</strong><br />

accelerating cavities etc.). Each particle in <strong>the</strong> ensemble move al<strong>on</strong>g a trajectory,<br />

i.e. <strong>the</strong> orbit is described by <strong>the</strong> ensemble moti<strong>on</strong> is an average of <strong>the</strong> individual<br />

trajectories.<br />

The instantaneous positi<strong>on</strong> of an particle can be described by <strong>the</strong> tripple [x,y,s]<br />

(as defined in Figure 1.15). In a linear accelerator <strong>the</strong> ˆs directi<strong>on</strong> coincides with <strong>the</strong><br />

ˆz coordinate. For a complete descripti<strong>on</strong> of <strong>the</strong> particle’s state we also need to define<br />

coordinates that are proporti<strong>on</strong>al to <strong>the</strong> <strong>the</strong> momentum of <strong>the</strong> particles: [x ′ ,y ′ ,E],<br />

with x ′ = px/p, y ′ = py/p which describe <strong>the</strong> angular deviati<strong>on</strong>, perpendicular to<br />

<strong>the</strong> directi<strong>on</strong> of moti<strong>on</strong>, from <strong>the</strong> ideal orbit. For relativistic particles <strong>the</strong> energy is<br />

approximately equal to <strong>the</strong> particle momentum E ≈ cp. In some instances it is more<br />

c<strong>on</strong>venient to define <strong>the</strong> energy in terms of its deviati<strong>on</strong> from <strong>the</strong> ensemble average,<br />

as this gives a measure which is independent of <strong>the</strong> total energy.


20 1. Introducti<strong>on</strong><br />

10<br />

8<br />

6<br />

4<br />

2<br />

0<br />

2<br />

1<br />

Potential energy surface <strong>and</strong> phase plot<br />

0<br />

X’<br />

−1<br />

Figure 1.14: Part of a quadratic potential energy surface with a phase portrait. The circles describe orbits with<br />

c<strong>on</strong>stant energy, i.e. a particle would move al<strong>on</strong>g those circles.<br />

−2<br />

A <strong>beam</strong> can thus be defined as occupying a certain volume in a 6-dimensi<strong>on</strong>al<br />

phase space. This volume stays c<strong>on</strong>stant if <strong>the</strong> ensemble’s moti<strong>on</strong> is such that <strong>the</strong> total<br />

energy of <strong>the</strong> system do not change (i.e. evolves according to Hamilt<strong>on</strong>’s equati<strong>on</strong>s<br />

of moti<strong>on</strong>) 19 . For c<strong>on</strong>venience <strong>the</strong> phase space of <strong>the</strong> ensemble is usually represented<br />

by two-dimensi<strong>on</strong>al projecti<strong>on</strong>s x,x ′ ; Figure 1.14 shows a phaseportrait <strong>and</strong> a potential<br />

energy surface for a particle moving in a quadratic potential (for instance a<br />

gravitati<strong>on</strong>al force), <strong>the</strong> circles represent orbits with c<strong>on</strong>stant energy.<br />

Horiz<strong>on</strong>tal emittance (for instance al<strong>on</strong>g <strong>the</strong> x directi<strong>on</strong>) εx is <strong>the</strong> area of an ellipse<br />

encompassing <strong>the</strong> majority of <strong>the</strong> particles (usually this is taken to be <strong>the</strong> area of a<br />

root-mean-square (rms) sense). A measure of <strong>the</strong> average phase space area covered by<br />

<strong>the</strong> particles in <strong>the</strong> x,x ′ -plane can <strong>the</strong>n be computed, assuming a distributi<strong>on</strong> al<strong>on</strong>g<br />

<strong>the</strong> ideal orbit such that 〈x〉 = 〈x ′ 〉 = 0<br />

−2<br />

−1<br />

εx = � 〈x 2 〉〈x ′2 〉−〈xx ′ 〉 2<br />

This quantity is c<strong>on</strong>served as l<strong>on</strong>g as <strong>the</strong> moti<strong>on</strong> in this directi<strong>on</strong> is independent from<br />

<strong>the</strong> moti<strong>on</strong> in <strong>the</strong> o<strong>the</strong>r directi<strong>on</strong>s (something which is comm<strong>on</strong> in accelerators)[29].<br />

19 As a c<strong>on</strong>sequence of Liouville’s <strong>the</strong>orem, i.e. that <strong>the</strong> volume in phase space stays c<strong>on</strong>stant if<br />

<strong>the</strong> system evolves under <strong>the</strong> influence of c<strong>on</strong>servative forces <strong>on</strong>ly. Forces that depend <strong>on</strong> positi<strong>on</strong><br />

<strong>on</strong>ly are c<strong>on</strong>servative, e.g. gravity <strong>and</strong> Coloumb forces. It is less comm<strong>on</strong> that forces that depend<br />

<strong>on</strong> <strong>the</strong> momentum are c<strong>on</strong>servative, an important excepti<strong>on</strong> to <strong>the</strong> latter is magnetic forces which<br />

do not change <strong>the</strong> momentum magnitude, <strong>on</strong>ly its directi<strong>on</strong>; thus magnetic forces are c<strong>on</strong>servative.<br />

0<br />

X<br />

1<br />

2


1.6. Definiti<strong>on</strong>s used throughout <strong>the</strong> book 21<br />

ˆx<br />

ˆy<br />

ˆs<br />

� 〈x〉 2<br />

� 〈x ′ 〉 2<br />

Figure 1.15: Definiti<strong>on</strong> of <strong>the</strong> rms width <strong>and</strong> angular spread of particles in a bunch.<br />

In Figure 1.15 <strong>the</strong> quantities inside <strong>the</strong> square-root are depicted.<br />

The emittance, as defined above, is also preserved as l<strong>on</strong>g as <strong>the</strong> particles are not<br />

accelerated. It is <strong>the</strong>refore customary to use <strong>the</strong> normalized emittance which makes<br />

<strong>the</strong> emittance comparable even if <strong>the</strong> <strong>beam</strong> has underg<strong>on</strong>e accelerati<strong>on</strong>[30].<br />

The normalized emittance is related to <strong>the</strong> <strong>beam</strong> energy via <strong>the</strong> relativistic factors<br />

β = v<br />

<strong>and</strong> γ = √ 1<br />

c<br />

1−β2 :<br />

εn = βγε<br />

which is invariant al<strong>on</strong>g <strong>the</strong> accelerator structure in <strong>the</strong> absence of radiati<strong>on</strong>. C<strong>on</strong>sequently,<br />

<strong>the</strong> transverse size of a charged particle <strong>beam</strong> shrinks as it gets accelerated<br />

as √ βγ. As will be seen below, <strong>the</strong> emittance needs to be very low for a free electr<strong>on</strong><br />

laser as <strong>the</strong> generated radiati<strong>on</strong> needs to overlap substantially with <strong>the</strong> electr<strong>on</strong><br />

<strong>beam</strong> for <strong>the</strong> lasing process to occur. Toge<strong>the</strong>r with <strong>the</strong> current <strong>and</strong> <strong>the</strong> <strong>beam</strong> energy<br />

<strong>the</strong> transverse <strong>and</strong> l<strong>on</strong>gitudinal emittances are very important figures of merit for<br />

accelerators.


22 1. Introducti<strong>on</strong><br />

Summary<br />

• X-rays can be used to investigate <strong>the</strong> electr<strong>on</strong>ic <strong>and</strong> geometrical<br />

structure of matter <strong>–</strong> by spectroscopic or scattering experiments<br />

respectively. Thus <strong>the</strong>y are used by a broad scientific<br />

community for experiments, i.e. for both fundamental <strong>and</strong><br />

applied investigati<strong>on</strong>s in medicince, biology, chemistry <strong>and</strong><br />

physics.<br />

• The brilliance (quality) of X-ray sources have developed exp<strong>on</strong>entially<br />

since <strong>the</strong> discovery of X-ray radiati<strong>on</strong>.<br />

• Today <strong>the</strong> most brilliant, man-made, X-ray source is <strong>the</strong> free<br />

electr<strong>on</strong> laser. O<strong>the</strong>r X-ray sources are X-ray tubes, anodes,<br />

synchrotr<strong>on</strong> storage rings <strong>and</strong> high-harm<strong>on</strong>ic generati<strong>on</strong><br />

(HHG) lasers.<br />

• Compared to solid state lasers <strong>and</strong> HHG lasers a free electr<strong>on</strong><br />

laser do not have any limit <strong>on</strong> <strong>the</strong> output power imposed by<br />

<strong>the</strong> laser medium, it being an electr<strong>on</strong> <strong>beam</strong> in vacuum <strong>and</strong><br />

not a gas or a solid.<br />

• Thebasic process thatgovernsfree electr<strong>on</strong>laser amplificati<strong>on</strong><br />

is abbreviated Sase <strong>–</strong> Self Amplified Sp<strong>on</strong>taneous Emissi<strong>on</strong>.<br />

This process describes how <strong>the</strong> radiati<strong>on</strong> field is amplified bya<br />

relativistic electr<strong>on</strong> <strong>beam</strong> moving through aperiodic magnetic<br />

field (i.e. in an undulator) when <strong>the</strong> radiati<strong>on</strong> field modulates<br />

<strong>the</strong> electr<strong>on</strong> <strong>beam</strong> density (microbunching). A nice introductory<br />

descripti<strong>on</strong> of <strong>the</strong> free electr<strong>on</strong> laser process can be found<br />

in Ref. [31].<br />

• Sase is a positive feedback process <strong>and</strong> <strong>the</strong> amplificati<strong>on</strong> of<br />

<strong>the</strong> radiati<strong>on</strong> field can be exp<strong>on</strong>ential. The process reach saturati<strong>on</strong><br />

when <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> is microbunched with <strong>the</strong><br />

periodicity of emitted radiati<strong>on</strong>. The process starts up from<br />

electr<strong>on</strong> shot-noise in <strong>the</strong> <strong>beam</strong>, thus <strong>the</strong> number of radiati<strong>on</strong><br />

modes follow a Poiss<strong>on</strong> distributi<strong>on</strong>.<br />

• The spectrum of <strong>the</strong> emitted phot<strong>on</strong>s is broad (from incoherent<br />

sp<strong>on</strong>taneous undulator radiati<strong>on</strong>) with a number of<br />

sharp spikes corresp<strong>on</strong>ding to <strong>the</strong> coherent radiati<strong>on</strong> from<br />

Sase-modes. The radiati<strong>on</strong> in <strong>the</strong> modes have full transverse<br />

coherence <strong>and</strong> each mode is diffracti<strong>on</strong> limited.<br />

• The energy <strong>and</strong> intensity distributi<strong>on</strong> from each pulse is<br />

unique since <strong>the</strong> process starts up from noise. Methods to manipulate<br />

<strong>the</strong> electr<strong>on</strong> <strong>beam</strong> serving to enhance <strong>the</strong> shot toshot<br />

repeatability of <strong>the</strong> spectrum utilize lasers to seed <strong>the</strong> <strong>beam</strong><br />

before entering <strong>the</strong> periodic magnetic structure (undulator).<br />

In fortuitous cases this can lock <strong>the</strong> radiati<strong>on</strong> into a single<br />

Sase-mode which <strong>the</strong>n becomes <strong>on</strong>e milli<strong>on</strong> times more intense<br />

<strong>the</strong>n <strong>the</strong> corresp<strong>on</strong>ding unmanipulated Sase-spectrum.


2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

Written by: A. Lindblad<br />

2.1 Radiati<strong>on</strong> from a moving charge<br />

This chapter c<strong>on</strong>tains certain elements of classical electrodynamics, which is needed<br />

in <strong>the</strong> following chapters. For <strong>the</strong> underlying framework <strong>and</strong> definiti<strong>on</strong>s see <strong>the</strong> book<br />

of Jacks<strong>on</strong> <strong>and</strong> Schwinger’s article from 1949[32, 33].<br />

Maxwell’s laws<br />

Gauss’ law for <strong>the</strong> electric field states that <strong>the</strong> flux of <strong>the</strong> electric field through a<br />

closed surface S is proporti<strong>on</strong>al to <strong>the</strong> enclosed total charge:<br />

�<br />

E·dA =<br />

S<br />

Q<br />

ǫ0<br />

(2.1)<br />

Analogously <strong>the</strong>re is a Gauss’ law for <strong>the</strong> magnetic field B; <strong>the</strong> field lines of <strong>the</strong><br />

magnetic field must be closed <strong>–</strong> thus <strong>the</strong> net flux through a closed surface must be<br />

zero, i.e. �<br />

B·dA = 0 (2.2)<br />

S<br />

This implies that <strong>the</strong>re are no magnetic m<strong>on</strong>opoles, if <strong>the</strong>re were this equati<strong>on</strong> would<br />

also have a source term as <strong>the</strong> equati<strong>on</strong> for <strong>the</strong> electric field.<br />

For <strong>the</strong> electric <strong>and</strong> magnetic fields to be coupled<br />

<strong>the</strong> flux of ei<strong>the</strong>r <strong>the</strong> electric or <strong>the</strong> magnetic fields<br />

needs to change; if <strong>the</strong> flux of <strong>the</strong> magnetic field<br />

changes over time, <strong>the</strong>n <strong>the</strong> electromotive force al<strong>on</strong>g<br />

a closed loop <strong>on</strong> <strong>the</strong> surface S is proporti<strong>on</strong>al to <strong>the</strong><br />

flux:<br />

S1<br />

I<br />

∂S S2<br />

E<br />

I<br />

�<br />

E·dℓ = −<br />

∂S<br />

B<br />

Figure2.1: Surfacessharing<strong>the</strong> same<br />

bounding c<strong>on</strong>tour ∂S.<br />

∂ΦB<br />

∂t<br />

(2.3)<br />

The sum of <strong>the</strong> magnetic fields through a closed<br />

loop <strong>on</strong> a surface enclosing a current is proporti<strong>on</strong>al<br />

23


24 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

to that current (Ampere’s law): �<br />

B · dℓ = µ0I. However, as seen in Figure 2.1 it<br />

∂S<br />

is easy to c<strong>on</strong>struct a situati<strong>on</strong> where <strong>the</strong> same bounding c<strong>on</strong>tour is shared by two<br />

surfaces where <strong>on</strong>ly <strong>on</strong>e of <strong>the</strong> surfaces c<strong>on</strong>tain <strong>the</strong> current in Ampère’s law, whereas<br />

<strong>the</strong> o<strong>the</strong>r <strong>on</strong>e c<strong>on</strong>tains a changing electric field <strong>–</strong> here we use a discharging parallelplate<br />

capacitor for this purpose; even though no charge flows between <strong>the</strong> capacitor<br />

plates <strong>the</strong>re is still a current ID flowing inside <strong>the</strong> capacitor (although <strong>the</strong>re is a<br />

vacuum between <strong>the</strong> plates here), by Ampère’s law we have just stated <strong>the</strong>re should<br />

be an induced magnetic field.<br />

Using Gauss’ law for <strong>the</strong> electric field, assuming a static surface, we can get:<br />

�<br />

dQ<br />

= I = ǫ0 dS ·<br />

dt ∂E ∂E<br />

≈ −Sǫ0<br />

∂t ∂t<br />

S<br />

This current <strong>and</strong> <strong>the</strong> c<strong>on</strong>ducti<strong>on</strong> current must be equal since <strong>the</strong>y must sum to zero.<br />

We divide by <strong>the</strong> area of <strong>the</strong> surface element S <strong>and</strong> use current densities (i.e. divide<br />

I by <strong>the</strong> area as well, I/S = J) <strong>and</strong> we get <strong>the</strong> result: <strong>the</strong> Ampère-Maxwell equati<strong>on</strong>:<br />

�<br />

∂S<br />

B ·dℓ = µ0<br />

�<br />

S<br />

�<br />

J+ǫ0<br />

∂E<br />

∂t<br />

�<br />

·dS (2.4)<br />

Equati<strong>on</strong>s 2.1, 2.2, 2.3, <strong>and</strong> 2.4 are <strong>the</strong> Maxwell equati<strong>on</strong>s that c<strong>on</strong>situte <strong>the</strong> basis<br />

for classical electrodynamics; via Gauss’ <strong>and</strong> Stoke’s <strong>the</strong>orems we can formulate <strong>the</strong>m<br />

also in differential form as<br />

Gauss’ law ∇·E = ρ<br />

ǫ0<br />

Gauss’ law for magnetic field ∇·B = 0<br />

Maxwell-Faraday equati<strong>on</strong> ∇×E = − ∂B<br />

∂t<br />

Ampère-Maxwell equati<strong>on</strong> ∇×B = µ0J+µ0ǫ0 ∂E<br />

∂t<br />

The electric <strong>and</strong> magnetic fields also couple via <strong>the</strong> force <strong>the</strong> fields exert <strong>on</strong> a<br />

charge moving in <strong>the</strong>m in <strong>the</strong> expressi<strong>on</strong> for <strong>the</strong> Lorentz force equati<strong>on</strong>:<br />

F = −q[E+v×B] (2.5)<br />

The energy flux of <strong>the</strong> fields can be found via <strong>the</strong> expressi<strong>on</strong> for Poynting’s vector:<br />

S = 1<br />

Charged particle at rest or moving with c<strong>on</strong>stant velocity<br />

µ0<br />

(E×B) = 1<br />

|E|<br />

cµ0<br />

2 ˆz (2.6)<br />

A charged particle at rest surrounds itself with an electric field that can be written<br />

(Coulomb’s law):<br />

E = 1 q<br />

4πǫ0 r2ˆr This is a special case of Gauss’ law for <strong>the</strong> electric field, Equati<strong>on</strong> 2.1.<br />

Since <strong>the</strong> electric field is static in <strong>the</strong> situati<strong>on</strong> when <strong>the</strong> particle is at rest no<br />

magnetic field is induced (Ampére-Maxwell equati<strong>on</strong> 2.4. In <strong>the</strong> case of a uniformly


2.1. Radiati<strong>on</strong> from a moving charge 25<br />

moving charge we have a c<strong>on</strong>stant current which creates a static magnetic field (Equati<strong>on</strong><br />

2.4 again) <strong>–</strong> no electric field is induced (Equati<strong>on</strong> 2.3).<br />

In both cases c<strong>on</strong>sidered here nochange in <strong>the</strong> kinetic energy of <strong>the</strong> particle occurs,<br />

thus no energy exists that can be transferred to <strong>the</strong> electromagnetic radiati<strong>on</strong> field.<br />

The fields from a charge in arbitrary moti<strong>on</strong><br />

Following Feynman[34, 35] we write <strong>the</strong>electric field from acharge in arbitrary moti<strong>on</strong><br />

as<br />

E = q<br />

� � ′ ′<br />

ˆr r′ d ˆr<br />

+<br />

4πǫ0 r ′2 c dt r ′2<br />

�<br />

+ 1<br />

c2 d 2<br />

dt2ˆr′ �<br />

(2.7)<br />

<strong>and</strong> <strong>the</strong> magnetic field cB = ˆr ′ × E. The primed quantites is to remember that we<br />

have to evaluate <strong>the</strong>se quantites at retarded time t ′ = t− r′<br />

<strong>–</strong> this is a c<strong>on</strong>sequence<br />

c<br />

of <strong>the</strong> finite speed of light, at <strong>the</strong> observati<strong>on</strong> point p in Figure 2.2 signals observed<br />

at time t was created when <strong>the</strong> charge was at t ′ . The sec<strong>on</strong>d term corresp<strong>on</strong>ds to a<br />

linear extrapolati<strong>on</strong> of <strong>the</strong> Coulomb field (velocity times <strong>the</strong> time-delay r ′ /c), such<br />

that when <strong>the</strong> velocity tends to zero we retain <strong>the</strong> normal Coulomb field.<br />

The first two terms are both proporti<strong>on</strong>al to <strong>the</strong> inverse squared distance <strong>and</strong> thus<br />

decays fast with respect to <strong>the</strong> distance, at least compared to <strong>the</strong> third term which is<br />

proporti<strong>on</strong>al to <strong>the</strong> inverse distance. Therefore, <strong>the</strong> last term in <strong>the</strong> equati<strong>on</strong> above<br />

is called <strong>the</strong> radiati<strong>on</strong> field since it survives even when r → ∞.<br />

Electric <strong>and</strong> magnetic field lines must be to be c<strong>on</strong>tinuous; looking at <strong>the</strong> right-side<br />

of Figure 2.2 we can c<strong>on</strong>sider a charge that get accelerated a very short time towards<br />

a n<strong>on</strong>-relativistic velocity (v ≪ c). The signal at X that was emitted at time t = 0<br />

have its fr<strong>on</strong>t at a distance c∆t from <strong>the</strong> signal that was emitted at t = ∆t; for <strong>the</strong><br />

electric field lines to be c<strong>on</strong>tinuous <strong>the</strong>re must exist a perpendicular comp<strong>on</strong>ent of<br />

<strong>the</strong> electric field which is proporti<strong>on</strong>al to <strong>the</strong> velocity in that directi<strong>on</strong> (<strong>and</strong> thus <strong>the</strong><br />

acellerati<strong>on</strong>). The parallel comp<strong>on</strong>ent is given by <strong>the</strong> first two terms in <strong>the</strong> equati<strong>on</strong><br />

for <strong>the</strong> electric field above.<br />

1<br />

˙x<br />

r ′<br />

r ′<br />

c 2<br />

˙x ˙ xt<br />

Figure 2.2: Signalsreceived at <strong>the</strong> observati<strong>on</strong>pointp attime twascreated at<strong>the</strong> retarded time t’. Equati<strong>on</strong>2.7<br />

attempts to account for <strong>the</strong> particles moti<strong>on</strong> by linearly extrapolating <strong>the</strong> Coulomb-field to guess <strong>the</strong> particles<br />

current positi<strong>on</strong>.<br />

r<br />

p<br />

2<br />

1<br />

ϑ<br />

˙x⊥<br />

˙x�<br />

c∆t<br />

E⊥<br />

E�


26 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

|E⊥/E�| = ˙x⊥t<br />

c∆t = ¨x⊥ ·∆t·t<br />

=<br />

c∆t<br />

¨x⊥ ·t<br />

c = {t = r/c} = ¨x⊥ ·r<br />

c2 The magnitude of <strong>the</strong> electric field outside <strong>the</strong> sphere is given by Gauss’ law, which<br />

gives <strong>the</strong> parallel comp<strong>on</strong>ent: E� = q<br />

4πǫ0<br />

1<br />

r 2, yielding <strong>the</strong> perpendicular comp<strong>on</strong>ent<br />

we seek:<br />

E⊥ = q ¨x⊥<br />

4πǫ0 c2 q 1<br />

sinϑˆr =<br />

r 4πǫ0 c2 d 2<br />

dt2ˆr (2.8)<br />

in <strong>the</strong> relativistic case we must take care to evaluate <strong>the</strong> derivative at retarded time.<br />

Radiated power from a charged particle in n<strong>on</strong>-relativistic moti<strong>on</strong><br />

Equipped with <strong>the</strong> expressi<strong>on</strong> for <strong>the</strong> Poynting vector <strong>and</strong> <strong>the</strong> expressi<strong>on</strong> for <strong>the</strong><br />

electric <strong>and</strong> magnetic fields for a charge in accelerated moti<strong>on</strong> we are now equipped<br />

to find <strong>the</strong> expressi<strong>on</strong> for <strong>the</strong> radiated power.<br />

In Figure 2.2 we can see that we can write <strong>the</strong> radiati<strong>on</strong> part of <strong>the</strong> electric field<br />

(<strong>the</strong> o<strong>the</strong>r terms will fall off as 1/r 4 in <strong>the</strong> expressi<strong>on</strong> for <strong>the</strong> Poynting vector <strong>and</strong><br />

thus carry an insignificant energy flux compared to <strong>the</strong> radiati<strong>on</strong> field)<br />

E⊥ =<br />

q<br />

4πǫ0c2 1<br />

¨xsinϑ (2.9)<br />

r<br />

with directi<strong>on</strong> ˆr×(ˆr× ˆx). <strong>the</strong> magnetic field we get straightforwardly cB = ˆr×E⊥.<br />

Inserting this into Equati<strong>on</strong> 2.6 we get:<br />

|S| = 1<br />

c4 q 2<br />

(4πǫ0) 2<br />

¨x 2<br />

r2 sin2ϑ (2.10)<br />

To get <strong>the</strong> total radiated power we integrate over all angles<br />

��<br />

P = |S|dS =<br />

� π<br />

0<br />

|S|2πr 2 sinϑdϑ = q2<br />

6πǫ0c 3¨x<br />

This is <strong>the</strong> Larmor formula[36] for <strong>the</strong> total radiati<strong>on</strong> power 1 .<br />

Radiated power from a relativistic charged particle<br />

(2.11)<br />

Equati<strong>on</strong> 2.10 describes <strong>the</strong> radiati<strong>on</strong> field from a charge in n<strong>on</strong>-relativistic moti<strong>on</strong>,<br />

with <strong>the</strong> typical sin 2 ϑ look of <strong>the</strong> power distributi<strong>on</strong> (which is similar to that of<br />

a dipole-antenna); <strong>the</strong> left-side of Figure 2.3 shows <strong>the</strong> radiati<strong>on</strong> lobes from a n<strong>on</strong>relativistic<br />

charge moving al<strong>on</strong>g <strong>the</strong> z-axis that gets accelerated in <strong>the</strong> directi<strong>on</strong> of<br />

1<br />

The Larmor formula is <strong>the</strong> result <strong>on</strong>e obtains within <strong>the</strong> framework of classical electrodynamics.<br />

If <strong>on</strong>e accounts for <strong>the</strong> quantum nature of <strong>the</strong> emitted phot<strong>on</strong>, <strong>and</strong> thus <strong>the</strong> resulting recoil<br />

exerted <strong>on</strong> <strong>the</strong> electr<strong>on</strong>, <strong>the</strong> radiated power is slightly less[33]:<br />

�<br />

P = PLarmor 1 − 55<br />

16 √ �<br />

ε<br />

3 E<br />

with ε as <strong>the</strong> classical phot<strong>on</strong> energy <strong>and</strong> E <strong>the</strong> total energy.


2.1. Radiati<strong>on</strong> from a moving charge 27<br />

<strong>the</strong> x-axis. We will now try to get a look-<strong>and</strong>-feel for how <strong>the</strong> radiati<strong>on</strong> power is<br />

distributed up<strong>on</strong> accelerati<strong>on</strong> at relativistic velocities.<br />

In <strong>the</strong> rest-frame of <strong>the</strong> particle (i.e. a reference frame moving with <strong>the</strong> same speed<br />

as <strong>the</strong> particle) S ⋆ <strong>the</strong> radiati<strong>on</strong> flux will look as in <strong>the</strong> n<strong>on</strong>-relativistic case described<br />

above, that is a dipole-field; this is shown in <strong>the</strong> left side of Figure 2.3. A radiated<br />

electromagnetic wave with wavelength λ in a certain directi<strong>on</strong> in <strong>the</strong> rest-frame will,<br />

will have a wavevector k ⋆ in that frame. To see how this wavevector looks in ”our”<br />

laboratory frame we must do a Lorentz-transformati<strong>on</strong>.<br />

If we c<strong>on</strong>sider <strong>the</strong> velocity to be solely in <strong>the</strong> ˆz directi<strong>on</strong> we can write <strong>the</strong> Lorentztransformati<strong>on</strong><br />

from <strong>the</strong> starred to <strong>the</strong> unstarred system as kz = 2γk ⋆ z, where γ is<br />

<strong>the</strong> relativistic factor<br />

1<br />

�<br />

1− v2<br />

c 2<br />

1<br />

= √ . As we approach more <strong>and</strong> more relativistic<br />

1−β2 velocities more <strong>and</strong> more of <strong>the</strong> radiati<strong>on</strong> is focussed in <strong>the</strong> forward directi<strong>on</strong>; <strong>the</strong><br />

opening angle of <strong>the</strong> radiati<strong>on</strong> c<strong>on</strong>e shrinks:<br />

θ ≈ kx<br />

kz<br />

≈ k⋆ x<br />

2γk ⋆ z<br />

= tanθ⋆<br />

2γ<br />

≈ 1<br />

2γ<br />

This is a very important property from <strong>the</strong> point of view of light producing accelerators,<br />

<strong>the</strong> more <strong>the</strong> particles get accelerated, <strong>the</strong> more focussed <strong>the</strong> radiati<strong>on</strong> gets.<br />

It can also be seen from <strong>the</strong> bottom in <strong>the</strong> figure that smaller opening angles get<br />

shifted to higher frequencies than those with larger opening angles <strong>–</strong> thus <strong>the</strong>re will<br />

be a spatial frequency distributi<strong>on</strong> of <strong>the</strong> electromagnetic radiati<strong>on</strong> with <strong>the</strong> highest<br />

frequencies in <strong>the</strong> middle; this is a angle dependent Doppler shift of <strong>the</strong> radiati<strong>on</strong>.<br />

To find <strong>the</strong> magnitude of <strong>the</strong> radiated power <strong>on</strong>e must generalize Larmor’s formula<br />

to take into account relativistic effects (see e.g.[32]); for <strong>the</strong> treatment here it suffices<br />

to know that for <strong>the</strong> same accelerating force leads to a factor γ 2 higher radiati<strong>on</strong><br />

power <strong>–</strong> it is thus more ec<strong>on</strong>omical to have light producing structures that accelerate<br />

<strong>the</strong> particles transversely ra<strong>the</strong>r than accelerating <strong>the</strong>m l<strong>on</strong>gitudinally (at least from<br />

<strong>the</strong> light producti<strong>on</strong> point-of-view). This is also <strong>the</strong> reas<strong>on</strong> as to why linear colliders<br />

are historically <strong>the</strong> weap<strong>on</strong> of choice for particle physics experiments where, notably<br />

for light particles, synchrotr<strong>on</strong> radiati<strong>on</strong> losses is undesirable since <strong>the</strong> particle energy<br />

is <strong>the</strong> critical parameter.<br />

If we let rcmc 2 = e 2 define <strong>the</strong> classical particle radius rc <strong>and</strong> ρ <strong>the</strong> bending radius<br />

of <strong>the</strong> orbit caused by <strong>the</strong> deflecting magnetic field, <strong>the</strong> total radiated power becomes:<br />

Pγ = 2<br />

3 rcmc2cβ 4 γ 4<br />

ρ2 (2.12)<br />

InFigure 2.3 we see thatmost of<strong>the</strong> radiati<strong>on</strong> will beradiated in <strong>the</strong>forward directi<strong>on</strong><br />

with a polarisati<strong>on</strong> perpendicular to <strong>the</strong> bending magnetic field (σ mode) <strong>–</strong> however,<br />

some of <strong>the</strong> power will be emitted in a mode off axis with circular polarizati<strong>on</strong> (π<br />

mode). It can be shown [37] that <strong>the</strong> relative power between <strong>the</strong> modes are:<br />

Pσ = 7 1<br />

Pγ, Pπ =<br />

8 8 Pγ<br />

Frequency <strong>and</strong> coherence of synchrotr<strong>on</strong> radiati<strong>on</strong><br />

Synchrotr<strong>on</strong> radiati<strong>on</strong> from a charged particle in a circular orbit at relativistic speeds<br />

= φ, this angular<br />

is characterized by a searchlightlike lobe of radiati<strong>on</strong> of width 1<br />

γ


28 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

a ⋆ a<br />

Θ ⋆<br />

v ≪ c v � c<br />

k ⋆<br />

θ ⋆<br />

k ⋆ z<br />

k ⋆ x<br />

ˆx<br />

L<br />

ˆz<br />

k<br />

θ<br />

kz<br />

θ<br />

kx ≈ k ⋆ x<br />

Figure 2.3: N<strong>on</strong>relativistic (left, starred quantities) <strong>and</strong> relativistic (right) dipole radiati<strong>on</strong> fields. With k =<br />

2π/λ we get from <strong>the</strong> starred reference frame to <strong>the</strong> laboratory (observers’) frame via a Lorentz transformati<strong>on</strong><br />

L.<br />

interval is swept during <strong>the</strong> time ∆t ⋆ during which <strong>the</strong> particle moves <strong>the</strong> length<br />

∆ℓ ⋆ = v ⋆ ∆t ⋆ ⋆ φ<br />

= v <strong>–</strong> all in <strong>the</strong> particles frame of reference.<br />

ω0<br />

Due to <strong>the</strong> relativistic effects in play (length c<strong>on</strong>tracti<strong>on</strong> <strong>and</strong> time dilati<strong>on</strong>) an<br />

observer will measure a compressed pulse width of length:<br />

≈<br />

∆t = ∆t ⋆ − ∆ℓ⋆<br />

c = ∆t⋆ − v⋆∆t ⋆<br />

c =<br />

�<br />

1− v⋆<br />

�<br />

∆t<br />

c<br />

⋆ �<br />

= 1− v⋆<br />

�<br />

φ<br />

≈<br />

c ω0<br />

�<br />

1− v⋆<br />

�<br />

�<br />

1<br />

=<br />

c γω0<br />

1− v⋆<br />

c<br />

��<br />

1+ v⋆<br />

c<br />

1+ v⋆<br />

c<br />

�<br />

� � � �<br />

⋆ 2<br />

1 v 1<br />

≈ 1−<br />

γω0 c 2γω0<br />

= 1<br />

2γ 3 ω0<br />

In <strong>the</strong> last steps we have, since v ⋆ ≈ c in an accelerator, made <strong>the</strong> approximati<strong>on</strong><br />

1+v ⋆ /c = 2. The spectral width of a pulse of durati<strong>on</strong> ∆t is ∆ω � 1/∆t. Thus, in<br />

<strong>the</strong> case of synchrotr<strong>on</strong> radiati<strong>on</strong> we can expect to observe radiati<strong>on</strong> with frequencies<br />

up to about<br />

ωMax ≈ 2γ 3 ω0<br />

The spectrum will c<strong>on</strong>sist of Fourier comp<strong>on</strong>ents nω0 from n = 1 to n ≈ 2γ 3 . A<br />

spectrum from a bending magnet in a storage ring will thus be a broad spectrum<br />

peaked at <strong>the</strong> critical frequency (see e.g. [32, 38]). The critical frequency is usually<br />

defined as <strong>the</strong> frequency where half <strong>the</strong> integral power lies below <strong>and</strong> half lies above<br />

this frequency.<br />

In a storage ring, or linear accelerator <strong>the</strong>re also exist collective effects since <strong>the</strong>re<br />

is a number of particles in <strong>the</strong> <strong>beam</strong>, say N; how <strong>the</strong> particles are distributed have a<br />

large impact <strong>on</strong> <strong>the</strong> emitted radiati<strong>on</strong>. If we c<strong>on</strong>sider <strong>the</strong> case of a storage ring three<br />

cases can be discerned.


2.2. Radiati<strong>on</strong> from a bending magnet 29<br />

e −<br />

1<br />

γ<br />

1<br />

γ<br />

∆t<br />

Figure 2.4: The time during which an observer is illuminated...<br />

• The particles are evenly distributed in a c<strong>on</strong>stant current al<strong>on</strong>g <strong>the</strong> ring <strong>–</strong> this<br />

leads to perfect cancellati<strong>on</strong> of <strong>the</strong> radiati<strong>on</strong> fields since <strong>the</strong>ir phases are evenly<br />

distributed.<br />

• If every particle is closer than a typical wavelength (in a short ”bunch”) <strong>the</strong>n<br />

<strong>the</strong>ir individual phase differences will be small <strong>and</strong> <strong>the</strong> field from each particle<br />

will be amplified N times <strong>–</strong> leading to a N 2 fold increase in <strong>the</strong> radiated power.<br />

This is coherent radiati<strong>on</strong>.<br />

• If <strong>the</strong> particles are unevenly distributed in <strong>the</strong> ring <strong>the</strong>n √ N of <strong>the</strong> particles<br />

will have phases that are not perfectly r<strong>and</strong>om 2 . Then <strong>the</strong> fields from those √ N<br />

particles will amplify that of <strong>the</strong> √ N o<strong>the</strong>rs which leads to that <strong>the</strong> incoherent<br />

radiati<strong>on</strong> is proporti<strong>on</strong>al to N.<br />

In a synchrotr<strong>on</strong> <strong>beam</strong> from a storage ring with bunches of N particles both incoherent<br />

<strong>and</strong> coherent parts of <strong>the</strong> radiati<strong>on</strong> field are normally present (Equati<strong>on</strong> 2.32).<br />

2.2 Radiati<strong>on</strong> from a bending magnet<br />

Asseen abovein Equati<strong>on</strong>2.12, <strong>the</strong>power is inverselyproporti<strong>on</strong>al to<strong>the</strong>radiusof <strong>the</strong><br />

orbit squared. The upper limit for a single magnet is thus set by <strong>the</strong> strength of <strong>the</strong><br />

deflecting magnetic field <strong>and</strong> <strong>the</strong> <strong>beam</strong> energy. To maximize <strong>the</strong> radiated power <strong>on</strong>e<br />

thus needs to have smaller magnet gaps <strong>and</strong> higher magnetic field strengths combined<br />

with a high <strong>beam</strong> energy. Thus <strong>the</strong>re exist a limit as of how intense radiati<strong>on</strong> <strong>on</strong>e can<br />

create with a bending magnet.<br />

For a single bending magnet <strong>the</strong>re exist ano<strong>the</strong>r side-c<strong>on</strong>diti<strong>on</strong> <strong>–</strong> that is that <strong>the</strong><br />

magnet should not bend <strong>the</strong> electr<strong>on</strong>s away from <strong>the</strong> orbit in <strong>the</strong> storage ring, thus a<br />

2 Imagine that <strong>the</strong> emitted radiati<strong>on</strong> is shot-noise <strong>–</strong> thus characterizedby a Poiss<strong>on</strong> distributi<strong>on</strong>,<br />

<strong>the</strong>n for a sufficiently large number of particles <strong>the</strong> signal to noise ratio will be <strong>the</strong> square root of <strong>the</strong><br />

number of emitters, since <strong>the</strong> variance is √ N for a normal distributi<strong>on</strong> <strong>–</strong> which can be c<strong>on</strong>sidered<br />

applicable since for large N <strong>the</strong> Poiss<strong>on</strong>-distributi<strong>on</strong> tends towards <strong>the</strong> normal distributi<strong>on</strong>.


30 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

too str<strong>on</strong>g magnetic field can not be utilized. To circumvent this we can imagine an<br />

array of str<strong>on</strong>g magnets that eventually returns <strong>the</strong> electr<strong>on</strong> to <strong>the</strong> intended orbit.<br />

If <strong>the</strong> array of magnets are str<strong>on</strong>g, such that <strong>the</strong> deflecti<strong>on</strong> from <strong>the</strong> central orbit<br />

is large so that <strong>the</strong> light pulses from different bunches do not overlap, we will get a<br />

spectrum similar to that of a single bending magnet but more intense <strong>and</strong> <strong>–</strong> with <strong>the</strong><br />

now str<strong>on</strong>ger magnetic field <strong>–</strong> shorter critical wavelength. Such a device is called a<br />

wiggler or wavelength shifter. The output power exceeds that of a bending magnet<br />

by twice <strong>the</strong> number of periods, i.e. 2·N.<br />

We can also imagine a device with a smaller magnetic field strength where <strong>the</strong><br />

deflecti<strong>on</strong>s from <strong>the</strong> central orbit is not too large, instead we employ many more<br />

poles to get <strong>the</strong> intensity str<strong>on</strong>ger. In this type of scheme <strong>the</strong> emitted radiati<strong>on</strong> field<br />

can positively interfere for certain wavelengths (this c<strong>on</strong>diti<strong>on</strong> will be specified below)<br />

resulting in a spectrum with a fundamental radiati<strong>on</strong> mode <strong>and</strong> its harm<strong>on</strong>ics whose<br />

wavelength corresp<strong>on</strong>ds to a given combinati<strong>on</strong> of magnetic field strengths <strong>and</strong> <strong>beam</strong><br />

energy. This type of device is called an undulator. In a free electr<strong>on</strong> laser this device<br />

is a critical comp<strong>on</strong>ent which ultimately sets <strong>the</strong> properties of <strong>the</strong> emitted radiati<strong>on</strong>.<br />

The output power exceeds that of a bending magnet by <strong>the</strong> square of <strong>the</strong> number of<br />

periods: N 2 .<br />

2.3 Undulator radiati<strong>on</strong><br />

The undulator was described already in 1951 as a source for synchrotr<strong>on</strong> X-ray radiati<strong>on</strong><br />

[6]. In Figure 2.5 a schematic of <strong>the</strong> alternating-magnetic pole scheme of a<br />

permanent magnet undulator is shown <strong>–</strong> usually <strong>the</strong> number of poles is much larger<br />

since <strong>the</strong> number of deflecti<strong>on</strong>s for <strong>the</strong> electr<strong>on</strong> bunch dictates <strong>the</strong> intensity <strong>and</strong> spectral<br />

quality of <strong>the</strong> radiati<strong>on</strong> emitted (as we will derive below).<br />

λu<br />

Figure 2.5: A schematic of <strong>the</strong> periodic magnet structure of an undulator. In yellow <strong>the</strong> electr<strong>on</strong> path is shown.<br />

The undulator equati<strong>on</strong><br />

The undulator strength parameter K<br />

The magnetic field in an periodic magnet structure can be written, if we c<strong>on</strong>sider that<br />

we look at <strong>the</strong> structure from above as in Figure 2.5:<br />

� �<br />

2π<br />

B(z) = B0cos ˆy = B0cos(kuz) ˆy (2.13)<br />

z<br />

λu<br />

θ<br />

ˆz


2.3. Undulator radiati<strong>on</strong> 31<br />

Remembering <strong>the</strong> expressi<strong>on</strong> for <strong>the</strong> Lorentz force (Equati<strong>on</strong> 2.5) <strong>and</strong> Newt<strong>on</strong>’s<br />

sec<strong>on</strong>d law we can write<br />

dp<br />

dt<br />

= [˙p] = e(v×B) ⇒ ˙px = −evzBy<br />

where in <strong>the</strong> last step we have used v = vzˆz; we have also used <strong>the</strong> approximati<strong>on</strong><br />

that <strong>the</strong> electric field created is weak enough so that <strong>the</strong> interacti<strong>on</strong> is negligible <strong>–</strong><br />

later we will see what c<strong>on</strong>sequences occur when this is not <strong>the</strong> case (see secti<strong>on</strong> 2.5<br />

at (see page 44)).<br />

this is <strong>the</strong> equati<strong>on</strong> of moti<strong>on</strong> for <strong>the</strong> electr<strong>on</strong> in <strong>the</strong> periodic magnetic structure 3 .<br />

If we insert Equati<strong>on</strong> 2.13 into <strong>the</strong> equati<strong>on</strong> of moti<strong>on</strong> we get<br />

˙px = −e dz<br />

dt B0cos[kuz]<br />

which after integrati<strong>on</strong> with respect to time gives:<br />

mγvx = − eB0<br />

sin(kuz)<br />

ku<br />

(<strong>the</strong> relativistic factor γ enters from <strong>the</strong> expressi<strong>on</strong> for relativistic momentum) which<br />

defines <strong>the</strong> undulator strength parameter K:<br />

vx = − eB0<br />

sin(kuz) = −Kc sin(kuz) (2.14)<br />

mekuγ γ<br />

If <strong>the</strong> deflecti<strong>on</strong> from <strong>the</strong> ˆz directi<strong>on</strong> can be c<strong>on</strong>sidered small we can use tanθ ≈ θ<br />

toge<strong>the</strong>r with <strong>the</strong> approximati<strong>on</strong> that vz = c to obtain:<br />

�<br />

tanθ = vx<br />

vz<br />

≈ vx<br />

�<br />

≈ θ = −<br />

c<br />

K<br />

γ sin(kuz)<br />

Usually <strong>the</strong> n<strong>on</strong>-dimensi<strong>on</strong>al parameter K is written as:<br />

K = λueB0<br />

2πmec<br />

≈ 0.9337B0λu<br />

(2.15)<br />

where <strong>the</strong> latter holds true if <strong>the</strong> magnetic field strength is measured in Tesla <strong>and</strong> <strong>the</strong><br />

undulator period in centimeters.<br />

The c<strong>on</strong>diti<strong>on</strong> of coherent emissi<strong>on</strong><br />

In an undulator <strong>the</strong> transverse moti<strong>on</strong> is small, thus significant parts of <strong>the</strong> radiati<strong>on</strong><br />

field emitted at different times will overlap as <strong>the</strong> electr<strong>on</strong> transverse <strong>the</strong> undulator.<br />

This will result in positive <strong>and</strong> negative interference of certain wavelengths which<br />

are <strong>on</strong> or off res<strong>on</strong>ance with <strong>the</strong> period of <strong>the</strong> magnetic field. The c<strong>on</strong>diti<strong>on</strong> for<br />

a wavelength to experience positive interference can be derived without too much<br />

effort in terms of <strong>the</strong> undulator strength parameter, <strong>the</strong> relativistic factor γ <strong>and</strong> <strong>the</strong><br />

deflecti<strong>on</strong> angle.


32 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

A<br />

λu<br />

λucosθ<br />

Figure 2.6: Light traveling <strong>the</strong> path AB interferes with light emitted from A at an angle θ (path AB’).<br />

C<strong>on</strong>siderthat<strong>the</strong>electr<strong>on</strong> emitradiati<strong>on</strong>at<strong>the</strong>wave-crest 4 labelled AinFigure2.6<br />

<strong>and</strong> at some later instance τ emit radiati<strong>on</strong> at crest B.<br />

In Figure 2.6 <strong>the</strong> path through an undulator for an electr<strong>on</strong> moving in <strong>the</strong> field<br />

described by Equati<strong>on</strong> 2.13. The relati<strong>on</strong>ship between <strong>the</strong> emitted wavelength λs<br />

<strong>and</strong> <strong>the</strong> period of <strong>the</strong> undulator λu can be found by c<strong>on</strong>sidering <strong>the</strong> c<strong>on</strong>diti<strong>on</strong> for<br />

c<strong>on</strong>structive interference of <strong>the</strong> wavefr<strong>on</strong>ts emitted at A <strong>and</strong> B respectively.<br />

If we denote <strong>the</strong> average l<strong>on</strong>gitudinal velocity ˜vz <strong>the</strong>n <strong>the</strong> time difference between<br />

<strong>the</strong>topoints of emissi<strong>on</strong> is τ = λu/˜vz. The pathdifference between <strong>the</strong>twopoints will<br />

be a multiple of some wavelength λs of <strong>the</strong> emitted light; multiples of this wavelength<br />

will experience positive interference <strong>and</strong> will dominate <strong>the</strong> spectrum.<br />

c λu<br />

˜vz<br />

����<br />

AB<br />

θ<br />

B<br />

−λucosθ = nλs<br />

� �� �<br />

AB’<br />

n ∈ Z+ (2.16)<br />

We now need to seek <strong>the</strong> average l<strong>on</strong>gitudinal velocity. Remembering Equati<strong>on</strong><br />

2.14, we have an expressi<strong>on</strong> for <strong>the</strong> velocity in <strong>the</strong> transverse directi<strong>on</strong> <strong>–</strong> thus we<br />

can write:<br />

�<br />

vz = � v 2 −v 2 x =<br />

v2 c2<br />

−K 2<br />

γ2 sin2kuz where, by using β = v/c <strong>and</strong> γ = 1/ � 1−β 2 , <strong>on</strong>e can by breaking c out of <strong>the</strong><br />

square-root find:<br />

�<br />

vz = c 1− 1 1<br />

−<br />

γ2 γ2K2 sin2 �<br />

kuz = c 1− 1<br />

γ2 � �<br />

K2 2 sin kuz<br />

Knowing that γ >> 1 <strong>and</strong> <strong>the</strong> McLaurin series for √ 1+x = 1+ 1<br />

2<br />

�<br />

vz ≈ c 1− 1<br />

2γ2 � � 2 2<br />

1+K sin kuz �<br />

θ<br />

ˆz<br />

x− 1<br />

8 x2 +...<br />

3 In an helical undulator <strong>the</strong> magnetic field can be written B = Bu [ˆxcos(kuz) + ûsin(kuz)].<br />

4 The strength of <strong>the</strong> radiati<strong>on</strong> is proporti<strong>on</strong>al to <strong>the</strong> accelerati<strong>on</strong>, Equati<strong>on</strong> 2.11.


2.3. Undulator radiati<strong>on</strong> 33<br />

which averaged over <strong>on</strong>e half oscillati<strong>on</strong> gives <strong>the</strong> sought average velocity 5 :<br />

�<br />

˜vz ≈ c 1− 1<br />

2γ2 �<br />

1+ K2<br />

��<br />

2<br />

which, inserted into Equati<strong>on</strong> 2.16, gives:<br />

⎡<br />

nλs = λu⎣<br />

1<br />

1− 1<br />

2γ2 �<br />

1+ K2<br />

�<br />

� − 1−<br />

2<br />

θ2<br />

�<br />

2<br />

⎤<br />

⎦<br />

(2.17)<br />

where we have exp<strong>and</strong>ed <strong>the</strong> cosine for small θ, for <strong>the</strong> first term we again use <strong>the</strong><br />

1<br />

fact that 1/γ is very small: 1−x = 1+x+x2 +..., which gives <strong>the</strong> final result, for<br />

n = 1:<br />

λs = λu<br />

2γ2 �<br />

1+ 1<br />

2 K2 +(γθ) 2<br />

�<br />

(2.18)<br />

In a helical undulator <strong>the</strong> 1<br />

2 K2 term becomes K 2 . This equati<strong>on</strong> is usually referred to<br />

as <strong>the</strong> undulator equati<strong>on</strong> which defines <strong>the</strong> wavelength which satisfies <strong>the</strong> res<strong>on</strong>ance<br />

c<strong>on</strong>diti<strong>on</strong> up<strong>on</strong> which <strong>the</strong> radiati<strong>on</strong> is experiencing positive interference.<br />

On axis (θ = 0) <strong>the</strong> total time of travel through <strong>the</strong> magnetic structure will be<br />

∆t = Nu∆T, with Nu being <strong>the</strong> number of poles <strong>–</strong> this yields a linewidth of <strong>the</strong><br />

radiati<strong>on</strong> given by: ∆ω<br />

ω<br />

= 1<br />

Nu .<br />

For future reference we note that owing to <strong>the</strong> properties of <strong>the</strong> Fourier transform:<br />

The pulse length will be increased if it is larger than λNu <strong>and</strong> <strong>the</strong> line width will be<br />

increased if <strong>the</strong> energy spread of <strong>the</strong> <strong>beam</strong> superseeds<br />

A detailed look at <strong>the</strong> equati<strong>on</strong>s of moti<strong>on</strong> in an undulator<br />

1<br />

2Nu .<br />

To derive <strong>the</strong> frequency spectrum of undulator radiati<strong>on</strong> it is c<strong>on</strong>venient to know a bit<br />

more about <strong>the</strong> trajectories of <strong>the</strong> particles in <strong>the</strong> undulator, so far we <strong>on</strong>ly c<strong>on</strong>cerned<br />

ourselves with <strong>the</strong> transverse comp<strong>on</strong>ent of <strong>the</strong> moti<strong>on</strong> as this gives us expressi<strong>on</strong>s<br />

for <strong>the</strong> wiggler/undulator strength parameter K <strong>and</strong> <strong>the</strong> res<strong>on</strong>ance c<strong>on</strong>diti<strong>on</strong> for <strong>the</strong><br />

emitted radiati<strong>on</strong>.<br />

It is here c<strong>on</strong>venient to use Hamilt<strong>on</strong>ian dynamics, as this will simplify our discussi<strong>on</strong><br />

later when we c<strong>on</strong>sider what happens when we can no l<strong>on</strong>ger make <strong>the</strong> approximati<strong>on</strong><br />

that <strong>the</strong> radiated electric field is de-coupled from <strong>the</strong> moti<strong>on</strong> of <strong>the</strong> electr<strong>on</strong>s.<br />

The Hamilt<strong>on</strong>ian for an electr<strong>on</strong> in an electromagnetic field with vector potential A<br />

can be written: �<br />

H = (p−eA) 2 c2 +(mc2 ) 4 (2.19)<br />

From Maxwell’s equati<strong>on</strong>s it can be inferred that, since we know <strong>the</strong> rotati<strong>on</strong> <strong>and</strong><br />

curl of <strong>the</strong> electric <strong>and</strong> magnetic fields <strong>on</strong>e can formulate <strong>the</strong>m in terms of a vector<br />

potential A <strong>and</strong> a scalar potential φ. This is Helmholtz <strong>the</strong>orem of vector calculus;<br />

for <strong>the</strong> B field this relati<strong>on</strong> is particularly simple B = ∇×A <strong>–</strong> thus (with <strong>the</strong> aid of<br />

Equati<strong>on</strong> 2.13)<br />

π� 5 1<br />

π<br />

0<br />

sin 2 xdx = 1<br />

2


34 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

A = B<br />

sin(kuz)ˆx<br />

ku<br />

is <strong>the</strong> vector potential defining <strong>the</strong> undulator field. The Hamilt<strong>on</strong>ian described<br />

above do not have any explicit time dependence in <strong>the</strong> coordinates; for such a system<br />

<strong>the</strong> Hamilt<strong>on</strong>ian describes <strong>the</strong> total energy of <strong>the</strong> system, i.e. H = γmc 2 . Moreover,<br />

as <strong>the</strong> magnetic field d<strong>on</strong>ot perform any work <strong>on</strong> <strong>the</strong> particles, as it <strong>on</strong>ly changes <strong>the</strong>ir<br />

trajectory <strong>–</strong> we get an added b<strong>on</strong>us in that <strong>the</strong> can<strong>on</strong>ical momenta in <strong>the</strong> transverse<br />

directi<strong>on</strong>s of <strong>the</strong> system is c<strong>on</strong>served, hence[39]:<br />

˙px = − ∂H<br />

∂x<br />

˙py = − ∂H<br />

∂y<br />

Both momenta are thus c<strong>on</strong>stant <strong>and</strong> we can choose this c<strong>on</strong>stant to be zero if<br />

we c<strong>on</strong>sider <strong>the</strong> electr<strong>on</strong>’s velocity up<strong>on</strong> its entrance in <strong>the</strong> magnetic structure to be<br />

completely axial. The trajectories are given by<br />

˙x = ∂H<br />

∂px<br />

˙y = ∂H<br />

∂py<br />

= px −eAx<br />

= py −eAy<br />

= 0,<br />

= 0<br />

γm ,<br />

γm<br />

We use <strong>the</strong> definiti<strong>on</strong> of <strong>the</strong> undulator strength parameter K <strong>and</strong> divide by c to<br />

obtain expressi<strong>on</strong>s for <strong>the</strong> velocities<br />

βx = − K<br />

γ<br />

sin(kuz), βy = 0<br />

from <strong>the</strong> definiti<strong>on</strong> of <strong>the</strong> relativistic parameter γ = 1/ � 1−β 2 we can get an<br />

expressi<strong>on</strong> for <strong>the</strong> axial velocity βz as well:<br />

βz = K2<br />

4γ2β0 cos(2kuβ0ct)+β0<br />

(2.20)<br />

where β0 is <strong>the</strong> average l<strong>on</strong>gitudinal velocity (Equati<strong>on</strong> 2.17). The l<strong>on</strong>gitudinal velocity<br />

is thus slowed down since a part of <strong>the</strong> kinetic energy is transferred to <strong>the</strong><br />

transverse moti<strong>on</strong>.<br />

It can be expected that <strong>the</strong> velocity in <strong>the</strong> forward directi<strong>on</strong> will be much greater<br />

than <strong>the</strong> transverse deviati<strong>on</strong>s, thus we can c<strong>on</strong>sider <strong>the</strong> l<strong>on</strong>gitudinal trajectory as<br />

approximately given by βoct+z0 in <strong>the</strong> integrati<strong>on</strong> of Equati<strong>on</strong> 2.20<br />

K 2<br />

z(t) = 1<br />

8β2 sin(2kuβ0ct)+β0ct+z0 (2.21)<br />

0 kuγ2 The transverse comp<strong>on</strong>ent can be obtained similarily as was d<strong>on</strong>e above:<br />

x(t) = 1<br />

β0<br />

K<br />

kuγ coskuβ0ct+x0<br />

In <strong>the</strong> rest frame of <strong>the</strong> electr<strong>on</strong> it will <strong>the</strong>refore describe a figure eight moti<strong>on</strong>.


2.3. Undulator radiati<strong>on</strong> 35<br />

Frequency distributi<strong>on</strong> of undulator radiati<strong>on</strong><br />

Using <strong>the</strong> vector potential introduced above we are now equipped to derive <strong>the</strong> frequency<br />

spectrum from a relativistic electr<strong>on</strong> in an undulator field.<br />

Ano<strong>the</strong>r way to express <strong>the</strong> radiated power per unit solid angle is[32]:<br />

dP(t)<br />

dΩ<br />

= |A(t)|2<br />

<strong>the</strong>nce <strong>the</strong> total radiated energy becomes <strong>the</strong> time-integral (assuming that <strong>the</strong> radiati<strong>on</strong><br />

field drops sufficiently fast for large times, past <strong>and</strong> future, so that <strong>the</strong> radiated<br />

energy is finite)<br />

dW<br />

dΩ =<br />

�<br />

|A(t)| 2 dt<br />

The relativistic fields can be derived from <strong>the</strong> vector potential A <strong>and</strong> <strong>the</strong> scalar<br />

potential ψ, <strong>the</strong> Liénard-Wiechert potentials[32, 40] which takes care of <strong>the</strong> fact that<br />

<strong>the</strong> radiati<strong>on</strong> perceived by an observer was generated at an earlier instant (i.e. at<br />

retarded time t ′ = t−R/c).<br />

�<br />

c<br />

A(t) =<br />

4π [RErad]<br />

⎡<br />

�<br />

c<br />

ret = ⎣<br />

4π<br />

ˆn×<br />

�<br />

(ˆn−β)× ˙ �<br />

β<br />

(1−β · ˆn) 3<br />

⎤<br />

⎦<br />

<strong>the</strong> electric field here have some resemblance to <strong>the</strong> <strong>on</strong>e described by Equati<strong>on</strong> 2.8,<br />

however in <strong>the</strong> present case all <strong>the</strong> c<strong>on</strong>sequences of <strong>the</strong> relativistic velocity of <strong>the</strong><br />

moti<strong>on</strong> have been manifested.<br />

To analyze <strong>the</strong> frequency spectrum of <strong>the</strong> radiati<strong>on</strong> it is c<strong>on</strong>venient to express <strong>the</strong><br />

radiated energy in terms of <strong>the</strong> Fourier transform, where we can make use of <strong>the</strong><br />

Parseval <strong>the</strong>orem 6<br />

dW<br />

dΩ =<br />

∞�<br />

−∞<br />

|A(ω)| 2 dω<br />

If integrati<strong>on</strong> is taken for positive values of <strong>the</strong> frequencies, since negative <strong>on</strong>es do<br />

not have any physical meaning, we can write, as <strong>the</strong> integr<strong>and</strong>:<br />

|A(ω)| 2 +|A(−ω)| 2 = 2|A(ω)| 2<br />

where <strong>the</strong>last equality hold if A(t)is real so thatA(−ω) = A ∗ (ω). Wemay formulate<br />

<strong>the</strong> argument of <strong>the</strong> integral as <strong>the</strong> intensity as <strong>the</strong> double derivative of <strong>the</strong> intensity,<br />

as perceived <strong>on</strong> <strong>the</strong> directi<strong>on</strong> ˆn from <strong>the</strong> source<br />

dW<br />

dΩ =<br />

∞�<br />

0<br />

d 2 I(ω,ˆn)<br />

dΩdω dω<br />

6 Loosely stated: <strong>the</strong> integral of <strong>the</strong> square of a functi<strong>on</strong> is identical to <strong>the</strong> integral of <strong>the</strong> square<br />

of its Fourier series, i.e. with appropriate units: <strong>the</strong> energy c<strong>on</strong>tained in a waveform is identical<br />

to <strong>the</strong> energy c<strong>on</strong>tained in <strong>the</strong> sum of its various frequency comp<strong>on</strong>ents.<br />

ret


36 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

We need to find <strong>the</strong> Fourier transform of <strong>the</strong> vector potential to proceed<br />

�<br />

e2 A(ω) =<br />

8π2 ∞�<br />

e<br />

c<br />

iωt<br />

⎡<br />

⎣ ˆn×<br />

�<br />

(ˆn−β)× ˙ �<br />

β<br />

(1−β · ˆn) 3<br />

⎤<br />

⎦ dt<br />

−∞<br />

we should now change variables as to take explicit care of <strong>the</strong> retarded time, t ′ +<br />

R(t ′ )/c = t, we also assume that <strong>the</strong> unit vector ˆn stays c<strong>on</strong>stant in time <strong>–</strong> that<br />

is, our observati<strong>on</strong> point is sufficiently far away from <strong>the</strong> regi<strong>on</strong> in space where <strong>the</strong><br />

accelerati<strong>on</strong> takes place. The distance R(t ′ ) ≈ x− ˆn· ˆr(t ′ ).<br />

A(ω) =<br />

� e 2<br />

8π 2 c<br />

∞�<br />

−∞<br />

e iω(t′ −ˆn·ˆr(t ′ )/c) ˆn×<br />

ret<br />

�<br />

(ˆn−β)× ˙ �<br />

β<br />

(1−β · ˆn) 2<br />

dt ′<br />

it can be shown that <strong>on</strong>e can rewrite <strong>the</strong> integr<strong>and</strong> in terms of time derivative<br />

�<br />

ˆn× (ˆn−β)× ˙ �<br />

β<br />

(1−β · ˆn) 2 = d<br />

dt ′<br />

� �<br />

ˆn×(ˆn×β)<br />

1−β · ˆn<br />

with this in mind <strong>on</strong>e can integrate Equati<strong>on</strong><br />

2.22 by parts to obtain a significantly simpler<br />

expressi<strong>on</strong> for <strong>the</strong> frequency distributi<strong>on</strong><br />

e 2 ω 2<br />

4π 2 c<br />

�<br />

� ∞�<br />

�<br />

�<br />

�<br />

�<br />

−∞<br />

ˆn×(ˆn×β)e iω(t′ −ˆn·ˆr(t ′ )/c) dt ′<br />

(2.23)<br />

On axis, <strong>the</strong> fundamental harm<strong>on</strong>ic of <strong>the</strong><br />

undulator will thus have a frequency distributi<strong>on</strong><br />

given by:<br />

d 2 I<br />

dωdΩ = e2N 2 uγ 2 K 2<br />

�<br />

c 1+ K2<br />

� �2 sinx1<br />

� F1(K) 2<br />

x1<br />

2<br />

(2.24)<br />

where F1 is a difference between Bessel functi<strong>on</strong>s<br />

such that<br />

F1(K) = [J0(κ)−J1(κ)] 2<br />

where <strong>the</strong> arguments are written as κ =<br />

Fur<strong>the</strong>rmore<br />

�<br />

�<br />

�<br />

�<br />

�<br />

�<br />

2<br />

K 2<br />

�<br />

4 1+ K2<br />

2<br />

1<br />

0.9<br />

0.8<br />

0.7<br />

0.6<br />

0.5<br />

0.4<br />

0.3<br />

0.2<br />

0.1<br />

0<br />

−30 −20 −10 0 10 20 30<br />

(2.22)<br />

Figure 2.7: Undulator-radiati<strong>on</strong> in frequencyspace<br />

(amplitude scaled to unity) shows an oscillatory<br />

behaviour dominated by <strong>the</strong> sinx/x<br />

term.<br />

ω −ωr<br />

x1 = πNu<br />

ωr<br />

where ωr is <strong>the</strong> res<strong>on</strong>ant frequency obtained from <strong>the</strong> undulator equati<strong>on</strong> above:<br />

ωr = 2ckuγ2<br />

1+ K2<br />

2<br />

�.


2.4. Microbunching 37<br />

2.4 Microbunching<br />

Interacti<strong>on</strong> between <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> radiati<strong>on</strong> field<br />

Up until now we have c<strong>on</strong>sidered <strong>the</strong> coupling between <strong>the</strong> electr<strong>on</strong>s’ moti<strong>on</strong> <strong>and</strong><br />

<strong>the</strong> radiati<strong>on</strong> field to be negligible. If we c<strong>on</strong>sider <strong>the</strong> possibility for energy to be<br />

transferredback<strong>and</strong>forthbetween<strong>the</strong>electr<strong>on</strong><strong>beam</strong>inanundulator<strong>and</strong><strong>the</strong>radiated<br />

electromagnetic field we will find that free electr<strong>on</strong> laser amplificati<strong>on</strong> can occur. In<br />

<strong>the</strong> following we will entertain this possibility <strong>and</strong> find out <strong>the</strong> c<strong>on</strong>diti<strong>on</strong>s that enables<br />

this amplificati<strong>on</strong> to happen.<br />

An energy-modulati<strong>on</strong> to occur al<strong>on</strong>g an electr<strong>on</strong> bunch can be accounted for via<br />

<strong>the</strong> acti<strong>on</strong> of <strong>the</strong> part of <strong>the</strong> Lorentz force (Equati<strong>on</strong> 2.5) c<strong>on</strong>taining <strong>the</strong> electric field:<br />

dW = F·ds = −eE·ds = −E ·v ·ds (2.25)<br />

In an undulator <strong>the</strong> electr<strong>on</strong> velocity have comp<strong>on</strong>ents parallel to <strong>the</strong> electric field.<br />

Ex(z,t) = Ecos(kz −ωt+φ0) (2.26)<br />

To find <strong>the</strong> flow of energy per time we need an expressi<strong>on</strong> for <strong>the</strong> electr<strong>on</strong> velocity<br />

in an undulator <strong>–</strong> this is given by Equati<strong>on</strong> 2.14, hence:<br />

= ecE0K<br />

2γ<br />

dW<br />

dt<br />

cK<br />

= −eE0cos(ks−ωt+φ0)<br />

γ sin(kus) =<br />

� �� �<br />

Eq. 2.14<br />

{sin([k +ku]s−ωt+φ0)−sin([k −ku]s−ωt+φ0)} =<br />

= − ecE0K<br />

2γ<br />

[sinΨ+ −sinΨ−] (2.27)<br />

For <strong>the</strong> energy transfer between <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> radiati<strong>on</strong> field to be<br />

efficient over <strong>the</strong> whole undulator structure, <strong>the</strong> phase between <strong>the</strong> sinosoidal terms<br />

within <strong>the</strong> brackets in Equati<strong>on</strong> 2.27 needs to be c<strong>on</strong>stant, i.e.:<br />

0 = dΨ±<br />

dt<br />

= −[k ±ku] ds<br />

dt<br />

The resulting wavelength can thus be calculated:<br />

λ = 2π<br />

k<br />

= ± 2π<br />

2kuγ 2<br />

�<br />

1+ K2<br />

2<br />

K2<br />

2<br />

−ω ≈ −kc1+ ±kuc (2.28)<br />

2γ2 �<br />

= λu<br />

2γ 2<br />

�<br />

1+ K2<br />

�<br />

2<br />

(2.29)<br />

in <strong>the</strong> last step, <strong>the</strong> negative branch of soluti<strong>on</strong>s have been dropped since <strong>on</strong>ly wavelengths<br />

larger than zero make physical sense. Our hope is to make Ψ+ c<strong>on</strong>stant to<br />

fulfill <strong>the</strong> res<strong>on</strong>ance c<strong>on</strong>diti<strong>on</strong> defined by Equati<strong>on</strong> 2.28. This is fulfilled if <strong>the</strong> electr<strong>on</strong>s<br />

lag behind <strong>the</strong> radiati<strong>on</strong> field <strong>on</strong>e λu per undulator period. The energy between<br />

<strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> radiati<strong>on</strong> field is exchanged with <strong>the</strong> same wavelength as<br />

sp<strong>on</strong>taneous undulator radiati<strong>on</strong>.


38 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

ˆy<br />

ˆy<br />

ˆy<br />

ˆy<br />

ˆx<br />

ˆx<br />

ˆx<br />

ˆx<br />

v<br />

E<br />

0<br />

0<br />

λu<br />

Figure 2.8: The interacti<strong>on</strong> between <strong>the</strong> electr<strong>on</strong>s in <strong>the</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> radiati<strong>on</strong> field give rise to a density<br />

modulati<strong>on</strong> of <strong>the</strong> electr<strong>on</strong> bunches. This occurs most efficiently if <strong>the</strong> <strong>the</strong> electr<strong>on</strong> bunch lags behind <strong>the</strong><br />

radiati<strong>on</strong> field with <strong>on</strong>e λu per period.<br />

dW<br />

dt<br />

v<br />

The electr<strong>on</strong> velocity is parallel with <strong>the</strong> electric field twice per period <strong>–</strong> <strong>the</strong>re<br />

= 0. The p<strong>on</strong>dermotive phase Ψ+, related to <strong>the</strong> negative branch as Ψ− =<br />

Ψ+ − 2kus when dΨ ±<br />

dt<br />

E<br />

= 0, oscillates twice per period <strong>and</strong> <strong>on</strong> average cancels out,<br />

i.e. for a homogeneous e - -distributi<strong>on</strong> half <strong>the</strong> electr<strong>on</strong>s gain energy while <strong>the</strong> o<strong>the</strong>r<br />

half looses it.<br />

As a result <strong>the</strong> electr<strong>on</strong>s, if bunched from <strong>the</strong> start, tend to become density modulated<br />

with <strong>the</strong> periodicity of <strong>the</strong> radiati<strong>on</strong> field <strong>–</strong> this process is called microbunching.<br />

The wavelength of <strong>the</strong> density modulati<strong>on</strong> is thus given also by Equati<strong>on</strong> 2.29.<br />

With <strong>the</strong> res<strong>on</strong>ant c<strong>on</strong>diti<strong>on</strong> fulfilled <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> radiati<strong>on</strong> field can<br />

exchange energy over several (many) undulator periods which, taken toge<strong>the</strong>r with<br />

<strong>the</strong> microbunching, can lead to a net gain of energy in <strong>the</strong> radiati<strong>on</strong> field.<br />

So far we have c<strong>on</strong>sidered <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> to be m<strong>on</strong>oenergetic <strong>–</strong> it is instructive<br />

to c<strong>on</strong>sider also a <strong>beam</strong> with an energy spread (which will later be seen to relate to<br />

v<br />

E<br />

λr<br />

ˆz<br />

ˆz<br />

ˆz<br />

ˆz


2.4. Microbunching 39<br />

o<strong>the</strong>r figures of merit for free electr<strong>on</strong> laser <strong>beam</strong>). We denote <strong>the</strong> energy spread by<br />

∆γ, <strong>the</strong>n we can write <strong>the</strong> p<strong>on</strong>dermotive phase change as:<br />

dΨ+<br />

dt<br />

K2<br />

2<br />

= −kc1+<br />

2<br />

�<br />

1 1<br />

−<br />

(γ +∆γ) 2 γ2 �<br />

≈ 2kuc ∆γ<br />

γ<br />

����<br />

=η<br />

(2.30)<br />

where in <strong>the</strong> last step <strong>the</strong> relative energy spread has been defined. Knowing this we<br />

can formulate <strong>the</strong> energy transfer rate:<br />

dW<br />

dt<br />

= dη<br />

dt γmec2<br />

(2.31)<br />

Using <strong>the</strong> two relati<strong>on</strong>ships found above we can write an equati<strong>on</strong> system in <strong>the</strong> two<br />

variables Ψ <strong>and</strong> η: ⎧<br />

⎪⎨ dΨ+<br />

dt<br />

⎪⎩<br />

dη<br />

dt<br />

= 2kucη<br />

= − eE0K<br />

2mecγ2 sinΨ<br />

with Ψ = Ψ+ as <strong>the</strong> phase of <strong>the</strong> radiati<strong>on</strong> field compared to <strong>the</strong> electr<strong>on</strong>s with<br />

∆γ = 0 somewhere where dW<br />

= 0. The system of differential equati<strong>on</strong>s can be<br />

dt<br />

combined to a single sec<strong>on</strong>d order differential equati<strong>on</strong>:<br />

¨Ψ+Ω 2 sinΨ = 0, Ω 2 = eE0kuK<br />

meγ 2<br />

For small oscillati<strong>on</strong>s (i.e. sinx ≈ x) this is nothing but an harm<strong>on</strong>ic oscillator:<br />

¨x+ω 2 x = 0. Large oscillati<strong>on</strong>s, where <strong>on</strong>e needs to keep <strong>the</strong> sine-term intact, cause<br />

<strong>the</strong> frequency to decrease; analogous to, for instance, a swing that can make a full<br />

loop.<br />

For <strong>the</strong> electr<strong>on</strong>s that deviate from <strong>the</strong> ”central” energy more energy is lost than<br />

gained. This fur<strong>the</strong>r drives <strong>the</strong> bunching since ”fast” electr<strong>on</strong>s slow down <strong>and</strong> ”slow”<br />

electr<strong>on</strong>s get accelerated. This l<strong>on</strong>gitudinal moti<strong>on</strong> of <strong>the</strong> electr<strong>on</strong>s reduce <strong>the</strong> coupling(c<strong>on</strong>tained<br />

in K)between <strong>the</strong> electr<strong>on</strong> bunch<strong>and</strong> <strong>the</strong>radiati<strong>on</strong> field, acorrecti<strong>on</strong><br />

need thus to be made, i.e.<br />

� � 2<br />

K<br />

K −→ K J0<br />

4+2K 2<br />

� � 2<br />

K<br />

−J1<br />

4+2K 2<br />

��<br />

For K close to unity <strong>the</strong> reducti<strong>on</strong> caused by <strong>the</strong> Bessel functi<strong>on</strong>s within <strong>the</strong> brackets<br />

is 0.9 <strong>and</strong> tends toward 0.7 for large K. This correcti<strong>on</strong> is comm<strong>on</strong>ly written as [JJ]<br />

in free electr<strong>on</strong> laser litterature.<br />

Exp<strong>on</strong>ential gain<br />

The radiated power in a spectrum emanating from sp<strong>on</strong>taneous radiati<strong>on</strong> is proporti<strong>on</strong>al<br />

to <strong>the</strong> number of emitters, i.e. no phase correlati<strong>on</strong> exist. In a free electr<strong>on</strong><br />

laser <strong>beam</strong> ideally all electr<strong>on</strong>s in <strong>the</strong> <strong>beam</strong> emit in phase, as stated previously such<br />

an ensamble’s radiated power is proporti<strong>on</strong>al to <strong>the</strong> square of <strong>the</strong> number of emitters.


40 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

In a storage ring, part of <strong>the</strong> electr<strong>on</strong>s in a bunch emit in phase <strong>–</strong> <strong>the</strong> number of<br />

such emitters can be enhanced by various laser slicing schemes, all striving to increase<br />

<strong>the</strong> number of coherent emitters. � In general � <strong>the</strong> power from such a mixed ensamble<br />

2<br />

can be described by P = P0 Ne +NeFe ;<br />

��<br />

Fe =<br />

�2 cos(2πz/λ)S(z)dz<br />

(2.32)<br />

where S(z) is <strong>the</strong> l<strong>on</strong>gitudinal density distributi<strong>on</strong> of <strong>the</strong> electr<strong>on</strong>s, The number of<br />

electr<strong>on</strong>s in <strong>the</strong> storage ring case is Ne ∼ 10 10 .<br />

In a free electr<strong>on</strong> laser amplifier three collective phenomena c<strong>on</strong>tribute towards <strong>the</strong><br />

quadratic dependence <strong>on</strong> <strong>the</strong> number of emitters:<br />

1. Modulati<strong>on</strong> of <strong>the</strong> electr<strong>on</strong>s’ energies due to <strong>the</strong> interacti<strong>on</strong> with <strong>the</strong> radiati<strong>on</strong><br />

field.<br />

2. Change in <strong>the</strong> electr<strong>on</strong>s’ l<strong>on</strong>gitudinal positi<strong>on</strong>s from path length differences in<br />

<strong>the</strong> combined potential created by <strong>the</strong> radiati<strong>on</strong> field <strong>and</strong> <strong>the</strong> undulator field.<br />

3. A, so far, ignored growth of <strong>the</strong> radiati<strong>on</strong> field which enhance <strong>the</strong> two aforementi<strong>on</strong>ed<br />

effects.<br />

If we let e i2πz/λ−iωt describe <strong>the</strong> radiati<strong>on</strong> field (which develops according to <strong>the</strong><br />

wave equati<strong>on</strong>) <strong>and</strong> E = E0e iΦ describe <strong>the</strong> transverse oscillati<strong>on</strong>s of <strong>the</strong> electr<strong>on</strong>s,<br />

<strong>on</strong>e may formulate <strong>the</strong> three cooperating processes in a more specific language:<br />

� �<br />

∂ ∂<br />

+ E = i<br />

c∂t ∂z<br />

µ0<br />

2 aw<br />

�<br />

j<br />

e iΦ j<br />

γj<br />

(2.33)<br />

The right h<strong>and</strong> side of this expressi<strong>on</strong> clearly have a maximum when <strong>the</strong> phases Φj<br />

are <strong>the</strong> same, i.e. when <strong>the</strong> electr<strong>on</strong>s emit in phase. Fur<strong>the</strong>rmore, <strong>the</strong> strength of this<br />

maximum is clearly larger with increasing number of electr<strong>on</strong>s j.<br />

mc 2dγ<br />

dt<br />

aw<br />

= eE0 sin(Φ+Ψ) (2.34)<br />

γ<br />

This equati<strong>on</strong> describes how <strong>the</strong> energy transfer occurs between <strong>the</strong> electr<strong>on</strong>s <strong>and</strong> <strong>the</strong><br />

radiati<strong>on</strong> field. Here aw describes <strong>the</strong> coupling <strong>on</strong>-axis in <strong>the</strong> undulator so that for<br />

a planar undulator λu = λ0<br />

2γ2 � 2<br />

1+a w +γ 2 θ 2� , for a planar undulator a 2 w = K 2 /2.<br />

Lastly, <strong>the</strong> equati<strong>on</strong> below describes <strong>the</strong> energy modulati<strong>on</strong> of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong><br />

which give rise to <strong>the</strong> microbunching of <strong>the</strong> electr<strong>on</strong>s. If <strong>the</strong> l<strong>on</strong>gitudinal velocity<br />

βz differs from <strong>the</strong> res<strong>on</strong>ant velocity ˜ βz <strong>the</strong> electr<strong>on</strong> slips in p<strong>on</strong>dermotive phase,<br />

electr<strong>on</strong>s with higher velocity move forward while slower <strong>on</strong>es are retarded.<br />

dΦ<br />

dt<br />

= 2πc<br />

λ<br />

� βz<br />

˜βz<br />

�<br />

−1<br />

(2.35)


2.4. Microbunching 41<br />

Scaled free electr<strong>on</strong> laser equati<strong>on</strong>s<br />

The equati<strong>on</strong>s above can be formulated compactly by utilizing <strong>the</strong> Pierce parameter:<br />

ρ = λu 4π<br />

4π<br />

2 ·j0 ·K 2 rms ·A 2 jj<br />

IA ·λu ·γ 3<br />

(2.36)<br />

where j0 is <strong>the</strong> <strong>beam</strong>’s current density, IA = mc 3 /e = 17 kA (<strong>the</strong> Alfvén current). Ajj<br />

is <strong>the</strong> coupling coefficient between <strong>the</strong> electr<strong>on</strong> <strong>and</strong> phot<strong>on</strong> <strong>beam</strong>s <strong>–</strong> for a helical field<br />

it is equal to unity, whereas for a planar sinusoidal magnetic field it is a combinati<strong>on</strong><br />

of Bessel functi<strong>on</strong>s of <strong>the</strong> first kind: Ajj = [J0(κ) − J1(κ)] with argument κ =<br />

K 2 rms/[2·(1+K 2 rms)]. To c<strong>on</strong>nect it str<strong>on</strong>ger with <strong>the</strong> results above we may write it<br />

as functi<strong>on</strong> of <strong>the</strong> undulator parameters <strong>and</strong> properties of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong>:<br />

� �2/3 F1Kγ0Ωf<br />

ρ =<br />

4cγ 2 rku<br />

with this we can write a detuning parameter describing how <strong>the</strong> electr<strong>on</strong> <strong>beam</strong>’s<br />

energy relates to <strong>the</strong> res<strong>on</strong>ant energy of <strong>the</strong> radiated field:<br />

δ = γ2 0 −γ 2 r<br />

2γ 2 rρ<br />

≈ ∆γ<br />

γρ<br />

= η<br />

ρ<br />

where <strong>the</strong> approximati<strong>on</strong> holds for small deviati<strong>on</strong>s from <strong>the</strong> res<strong>on</strong>ant energy. The<br />

res<strong>on</strong>ant energy is given by <strong>the</strong> undulator equati<strong>on</strong> as:<br />

γr = kr<br />

2ku<br />

� µ0Nee 2 c 2<br />

1+K 2 /2<br />

2<br />

The Ωf is <strong>the</strong> plasma frequency which enters as a parameter in <strong>the</strong> space<br />

γme<br />

charge coefficient which describes <strong>the</strong> repulsi<strong>on</strong> forces between <strong>the</strong> electr<strong>on</strong>s in <strong>the</strong><br />

<strong>beam</strong> <strong>–</strong> something which counteracts <strong>the</strong> micro-bunching. The balance between <strong>the</strong><br />

space charge effects <strong>and</strong> micro-bunching is <strong>on</strong>e of <strong>the</strong> reas<strong>on</strong>s why <strong>the</strong> amplificati<strong>on</strong><br />

eventually saturates. The space charge coefficient is defined as:<br />

σ = Ωfγ0<br />

�<br />

1<br />

ckrρ 1+K 2 /2<br />

We write our new set of variables, describing <strong>the</strong> energy-spread, field amplitude <strong>and</strong><br />

positi<strong>on</strong>-time, following[41]:<br />

η ′ = η<br />

ρ<br />

A = F1Kkr<br />

4γrkuρ2 �<br />

−iKre iΨ�<br />

˜z = 2ckrργ2 r<br />

γ2 t<br />

0<br />

where in <strong>the</strong> sec<strong>on</strong>d equati<strong>on</strong> we have introduced <strong>the</strong> complex amplitude of <strong>the</strong> vector<br />

potential. For <strong>the</strong> details <strong>on</strong> <strong>the</strong> derivati<strong>on</strong> <strong>–</strong> not important for <strong>the</strong> following


42 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

<strong>–</strong> <strong>the</strong> reader is referred to B<strong>on</strong>ifacio et al. [41]. With <strong>the</strong> new variables above our<br />

equati<strong>on</strong>s 2.35, 2.35 <strong>and</strong> 2.35 can be normalized to <strong>the</strong> system of equati<strong>on</strong>s:<br />

˙θ = δ +η ′<br />

˙η ′ = − � A+iσ 2� e −θ�� e iθ −c.c.<br />

˙<br />

A = � e −θ�<br />

This 1-D system of equati<strong>on</strong>s can be solved analytically for a few idealized cases,<br />

o<strong>the</strong>rwise we have to utilize numerical methods to solve <strong>the</strong>m. Without energyspread<br />

<strong>and</strong> using <strong>the</strong> reas<strong>on</strong>able ansatz A ∝ e iΛ˜z we find that <strong>the</strong> cubic dispersi<strong>on</strong> relati<strong>on</strong>:<br />

� (Λ+δ) 2 −σ 2 � Λ = −1<br />

reduces to Λ 3 = −1.<br />

This equati<strong>on</strong> have three soluti<strong>on</strong>s describing <strong>the</strong> free electr<strong>on</strong> laser collective instability:<br />

<strong>the</strong> real soluti<strong>on</strong>, which is oscillatory, <strong>and</strong> two complex soluti<strong>on</strong>s which is<br />

decaying <strong>and</strong> growing respectively. In <strong>the</strong> start-up phase <strong>the</strong> three soluti<strong>on</strong>s have<br />

comparable magnitudes, after a certain time <strong>the</strong> growing soluti<strong>on</strong> will dominate, <strong>and</strong><br />

<strong>the</strong> field amplitude grows exp<strong>on</strong>entially. Within <strong>the</strong> linear <strong>on</strong>e dimensi<strong>on</strong>al framework<br />

we can not explain where this growth process ends (for instance via <strong>beam</strong> blow-up<br />

due to space-charge effects) as it is a n<strong>on</strong>-linear process.<br />

Inserting ˜z = 2kuρz (using z = ct <strong>and</strong> no energy spread) into our ansatz gives <strong>the</strong><br />

scales soluti<strong>on</strong> A ∝ e iΛz/Lg , defining <strong>the</strong> <strong>on</strong>e-dimensi<strong>on</strong>al gain-length 7 :<br />

Lg = λu<br />

2 √ 3πρ<br />

which is <strong>the</strong> undulator-length needed for <strong>the</strong> field to grow a factor e. As will discussed<br />

below, differentfactors combinetolimit<strong>the</strong>gainafteracertainlengthin<strong>the</strong>undulator<br />

<strong>–</strong> it can be shown that this length is[31]:<br />

Ls ≈ 22Lg<br />

Significant effort is put to keep <strong>the</strong> gain-length (<strong>and</strong> thus saturati<strong>on</strong>-length) as short<br />

as possible for ec<strong>on</strong>omical reas<strong>on</strong>s.<br />

The results here is valid for a m<strong>on</strong>o-energetic <strong>beam</strong>, introducing energy-spread<br />

effects <strong>the</strong> gain negatively as it prevents <strong>the</strong> bunching of <strong>the</strong> electr<strong>on</strong>s at <strong>the</strong> proper<br />

phase.<br />

The output power at saturati<strong>on</strong> of <strong>the</strong> free electr<strong>on</strong> laser can also be expressed in<br />

terms of ρ:<br />

�<br />

〈γ〉Ipeakmc<br />

Psat = ρP<strong>beam</strong> = ρ<br />

2�<br />

e<br />

Besides maximizing ρ to shorten <strong>the</strong> undulator, a large value of <strong>the</strong> parameter also<br />

give greater radiati<strong>on</strong> output power.<br />

7 By solving −ℜe (iΛ) = √ 3/2


2.4. Microbunching 43<br />

Saturati<strong>on</strong> <strong>–</strong> limiting factors for <strong>the</strong> gain<br />

The Pierce parameter defined above shows up in several relati<strong>on</strong>s highlighting <strong>the</strong><br />

dem<strong>and</strong>s <strong>on</strong> <strong>the</strong> quality of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> for <strong>the</strong> free electr<strong>on</strong> laser amplificati<strong>on</strong><br />

to be efficient. Electr<strong>on</strong>s transfer energy to <strong>the</strong> radiati<strong>on</strong> field until <strong>the</strong>y fall out of<br />

<strong>the</strong> b<strong>and</strong>width of <strong>the</strong> free electr<strong>on</strong> laser <strong>and</strong> <strong>the</strong> synchr<strong>on</strong>ism c<strong>on</strong>diti<strong>on</strong> is no l<strong>on</strong>ger<br />

fulfilled. We may thus formulate <strong>the</strong> efficiency of a free electr<strong>on</strong> laser simply by<br />

restating <strong>the</strong> last equati<strong>on</strong> as ρ = Psat/P<strong>beam</strong> <strong>–</strong> <strong>the</strong> Pierce parameter is thus a measure<br />

<strong>on</strong> how good a machine is when it comes to c<strong>on</strong>verting electr<strong>on</strong> <strong>beam</strong> energy to<br />

radiati<strong>on</strong> energy.<br />

Am<strong>on</strong>g <strong>the</strong> limiting factors for <strong>the</strong> gain process (<strong>and</strong> indeed <strong>the</strong> free electr<strong>on</strong><br />

laser process as a whole) is:<br />

• Energy spread around <strong>the</strong> res<strong>on</strong>ant energy<br />

• Deviati<strong>on</strong> of <strong>the</strong> mean energy from <strong>the</strong> res<strong>on</strong>ant energy<br />

• The size <strong>and</strong> divergence of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong>, that is normalized emittance.<br />

• Diffracti<strong>on</strong> effects in <strong>the</strong> <strong>beam</strong>.<br />

In Equati<strong>on</strong> 2.24 we found that <strong>the</strong> frequency spectrum of undulator radiati<strong>on</strong><br />

is proporti<strong>on</strong>al to � �<br />

sinw 2<br />

with w ∝ (γ − γr)/γr. The gain curve, showing how a<br />

w<br />

mode with arbitrary energy will get amplified in <strong>the</strong> undulator, for <strong>the</strong> free electr<strong>on</strong><br />

laser process can be shown to be proporti<strong>on</strong>al to <strong>the</strong> derivative of that functi<strong>on</strong>, i.e.<br />

G(ω) ∝ − d<br />

� �2 sinw<br />

dω w<br />

The proporti<strong>on</strong>ality c<strong>on</strong>stant c<strong>on</strong>tains parameters that define <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>and</strong><br />

radiati<strong>on</strong> field properties as dictated by <strong>the</strong> list above <strong>–</strong> a detailed treatment requires<br />

assumpti<strong>on</strong>s <strong>on</strong> <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> details would vary 8 .<br />

8 See for instance <strong>the</strong> books by Saldin <strong>and</strong> co-workers, or Wiedemann[42, 43], or for a more<br />

compact account <strong>the</strong> review in Ref. [44]. Said references also serve as good general references for<br />

this chapter for readers who want a more detailed <strong>and</strong> perhaps more stringent account of matters<br />

than presented in this introductory text to <strong>the</strong> subject at h<strong>and</strong>.<br />

Gain<br />

1<br />

0.5<br />

−10 −5 5 10<br />

−0.5<br />

−1<br />

Figure 2.9: The small-signal gain functi<strong>on</strong> for a free electr<strong>on</strong> laser amplifier.<br />

w


44 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

In Figure 2.9 <strong>the</strong> derivative of <strong>the</strong> negative cardinal sine functi<strong>on</strong> is plotted, this<br />

is <strong>the</strong> functi<strong>on</strong>al dependence <strong>on</strong> <strong>the</strong> <strong>beam</strong>’s energy for <strong>the</strong> gain curve as given by <strong>the</strong><br />

equati<strong>on</strong> above. It can be seen that a <strong>beam</strong> must have a certain, not too large, energy<br />

spread <strong>and</strong> ideally a small shift towards higher energies than <strong>the</strong> res<strong>on</strong>ance energy<br />

to have a positive gain (for a nearly m<strong>on</strong>ochromatic <strong>beam</strong> <strong>the</strong> optimal shift is about<br />

+1.2 <strong>on</strong> <strong>the</strong> gain curve).<br />

An energy spread prevents efficient microbunching of all electr<strong>on</strong>s with <strong>the</strong> same<br />

p<strong>on</strong>dermotivephase, this smears out <strong>the</strong>electr<strong>on</strong> bunchwhich preventsefficient transfer<br />

of energy into <strong>the</strong> res<strong>on</strong>ant mode with (fundamental) wavelength λr. The res<strong>on</strong>ant<br />

energy being γr = � (λu/2λ)·(1+a 2 w) <strong>the</strong> initial energy spread σ ′ of <strong>the</strong> electr<strong>on</strong><br />

<strong>beam</strong> should be kept<br />

σ ′<br />

≪ ρ<br />

γr<br />

The undulator res<strong>on</strong>ant energy needs, of course, to be tuned to <strong>the</strong> <strong>beam</strong> energy,<br />

i.e. <strong>the</strong> energies needs to be matched:<br />

� �<br />

�〈γ〉−γr<br />

�<br />

� �<br />

� � ≪ ρ<br />

γr<br />

Greater efficiency is obtained if <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> phot<strong>on</strong> <strong>beam</strong> overlap<br />

exactly. An electr<strong>on</strong> oscillates around <strong>the</strong> central path through <strong>and</strong> undulator with a<br />

period that is muchl<strong>on</strong>ger than<strong>the</strong> undulator’s period; this moti<strong>on</strong> is called abetatr<strong>on</strong><br />

moti<strong>on</strong>. Part of <strong>the</strong> electr<strong>on</strong>’s kinetic energy is <strong>the</strong>nce partiti<strong>on</strong>ed into <strong>the</strong> executi<strong>on</strong><br />

of this betatr<strong>on</strong> moti<strong>on</strong> which slows down <strong>the</strong> electr<strong>on</strong>, effectively acting as an energy<br />

spread. The betatr<strong>on</strong> amplitude functi<strong>on</strong> β ′ have a direct effect <strong>on</strong> <strong>the</strong> normalized<br />

transverse emittance of <strong>the</strong> phot<strong>on</strong> <strong>beam</strong> which dictates <strong>the</strong> following dem<strong>and</strong>:<br />

ǫn ≪ 4γβ′ 〈γ〉<br />

ρ<br />

The beta oscillati<strong>on</strong> needs to be optimized to strike a balance between a high electr<strong>on</strong><br />

density <strong>and</strong> space charge effects which degrades <strong>the</strong> emittance <strong>–</strong> a good starting point<br />

is to set it equal to <strong>the</strong> gain length, β ′ ≈ Lg ≈ λ<br />

ρ. This gives us a c<strong>on</strong>diti<strong>on</strong> <strong>on</strong> how<br />

4π<br />

<strong>the</strong> emittances of <strong>the</strong> electr<strong>on</strong> (ǫ) <strong>and</strong> <strong>the</strong> phot<strong>on</strong> <strong>beam</strong> (diffracti<strong>on</strong> limited λ/4π)<br />

should be matched:<br />

ǫ = ǫn λ<br />

<<br />

γ 4π<br />

If this c<strong>on</strong>diti<strong>on</strong> is fulfilled nei<strong>the</strong>r <strong>beam</strong> diverges faster than <strong>the</strong> o<strong>the</strong>r.<br />

Diffracti<strong>on</strong> effects in <strong>the</strong> <strong>beam</strong> softens <strong>the</strong> coupling between <strong>the</strong> radiati<strong>on</strong> field <strong>and</strong><br />

<strong>the</strong> electr<strong>on</strong> <strong>beam</strong>, to account for those properly a more intricate three-dimensi<strong>on</strong>al<br />

model needs to be c<strong>on</strong>structed[17]. A 3-d analogue to our <strong>on</strong>e dimensi<strong>on</strong>al Pierce<br />

parameter can be found where <strong>the</strong> relati<strong>on</strong>s states are still fulfilled.<br />

2.5 Sase<br />

So far our treatment of <strong>the</strong> free electr<strong>on</strong> laser problem have been d<strong>on</strong>e without any<br />

assumpti<strong>on</strong> <strong>on</strong> <strong>the</strong> nature of <strong>the</strong> radiati<strong>on</strong> mode(s) that are amplified. In <strong>the</strong> previous<br />

chapter, schemes aimed to create c<strong>on</strong>diti<strong>on</strong>s for amplificati<strong>on</strong> of already coherently<br />

λu


2.5. Sase 45<br />

radiati<strong>on</strong> modes from <strong>the</strong> start were discussed. Such schemes were not employed in<br />

<strong>the</strong> X-ray range until recently <strong>–</strong> at <strong>the</strong> Fermi@Elettra free electr<strong>on</strong> laser in Italy,<br />

which successfully dem<strong>on</strong>strated Hghg seeding in december 2010.<br />

The process of Sase utilizes <strong>the</strong> broadb<strong>and</strong> sp<strong>on</strong>taneous undulator radiati<strong>on</strong> from<br />

<strong>the</strong> first few gain-lengths of <strong>the</strong> undulator secti<strong>on</strong> as a seed for <strong>the</strong> remainder of <strong>the</strong><br />

amplificati<strong>on</strong> process. Owing to this <strong>the</strong> res<strong>on</strong>ance c<strong>on</strong>diti<strong>on</strong> is always fulfilled for<br />

some modes of <strong>the</strong> radiati<strong>on</strong>, thus <strong>the</strong> number of radiati<strong>on</strong> modes <strong>and</strong> <strong>the</strong>ir energy is<br />

sampled from a r<strong>and</strong>om distributi<strong>on</strong> (within <strong>the</strong> b<strong>and</strong>width of <strong>the</strong> amplifier which is<br />

also related to <strong>the</strong> Pierce parameter: ∆ω<br />

= ρ) <strong>on</strong> a shot-to-shot basis.<br />

ω<br />

The shot-to-shot fluctuati<strong>on</strong>s of <strong>the</strong> radiati<strong>on</strong> pulse energy follows a Gamma distributi<strong>on</strong><br />

[45] whose free parameter M can be interpreted as <strong>the</strong> number of spikes<br />

in <strong>the</strong> final frequency spectrum. The length of <strong>the</strong> individual spikes is proporti<strong>on</strong>al<br />

to <strong>the</strong> gain-length <strong>and</strong> <strong>the</strong> wavelength of <strong>the</strong> radiati<strong>on</strong> as λ/λuLg. Shorter pulses<br />

increases <strong>the</strong> number of spikes as <strong>the</strong> energy c<strong>on</strong>tent of such pulse is broader. The<br />

width of <strong>the</strong> Gamma distributi<strong>on</strong> is inversly proporti<strong>on</strong>al to √ M; <strong>the</strong> fluctuati<strong>on</strong> of<br />

<strong>the</strong> power is distributed as a negative exp<strong>on</strong>ential[42].<br />

Coherence properties<br />

Coherence means that <strong>the</strong> relative phase of waves is fixed. Spatial coherence between<br />

two radiati<strong>on</strong> sources means that <strong>the</strong> phot<strong>on</strong>s originating from <strong>the</strong>m occupy<br />

<strong>the</strong> same volume in phase space. In practice this means that emitters within <strong>the</strong><br />

coherence-length/area can amplify each o<strong>the</strong>r by c<strong>on</strong>structive interference if <strong>the</strong> phase<br />

relati<strong>on</strong>ship means that <strong>the</strong>y are equal.<br />

Temporal coherence can be thought of in <strong>the</strong> same manner: waves emitted at<br />

different times have a phase-correlati<strong>on</strong> that is predictable. The time during which<br />

<strong>the</strong> phase-relati<strong>on</strong>ship remains locked is referred to as <strong>the</strong> coherence-time. Waves<br />

emitted during this time interval (e.g. from electr<strong>on</strong>s al<strong>on</strong>g <strong>the</strong> electr<strong>on</strong> bunch) can<br />

c<strong>on</strong>structively interfere with each o<strong>the</strong>r if <strong>the</strong> phases are equal. The coherence time<br />

τ is intimately related to <strong>the</strong> spectral width ∆λ of <strong>the</strong> source via:<br />

∆λ = λ2<br />

cτ<br />

A l<strong>on</strong>g coherence time thus ensures a narrow spectral b<strong>and</strong>width of <strong>the</strong> source. This<br />

is <strong>the</strong> case for a normal Laser. In a Sase free electr<strong>on</strong> laser many modes are excited<br />

with various coherence-times for each meaning that each mode give rise to a narrow<br />

spike in <strong>the</strong> spectrum, whereas <strong>the</strong> compound spectrum is broad[45].<br />

Sase ensures very high transverse (spatial) coherence at <strong>the</strong> saturati<strong>on</strong> point[45<strong>–</strong><br />

47]; towards <strong>the</strong> end of <strong>the</strong> linear gain regime ideally all electr<strong>on</strong>s in <strong>the</strong> <strong>beam</strong> radiate<br />

in c<strong>on</strong>cert. This makes <strong>the</strong> free electr<strong>on</strong> laser extremely attractive for diffractive<br />

imaging[48] <strong>and</strong> o<strong>the</strong>r experiments relying heavily <strong>on</strong> this property of <strong>the</strong> X-ray<br />

source[49]. As we have seen earlier <strong>the</strong> ratio between <strong>the</strong> coherent radiati<strong>on</strong> part of<br />

<strong>the</strong> radiati<strong>on</strong> spectrum <strong>and</strong> <strong>the</strong> in-coherent (sp<strong>on</strong>taneous) radiati<strong>on</strong> can be extremely<br />

high in a free electr<strong>on</strong> laser whereas in a storage ring <strong>the</strong> relati<strong>on</strong>ship <strong>the</strong> coherent<br />

part of <strong>the</strong> undulator spectrum is significantly lower[50].


46 2. Synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong> its properties<br />

Summary<br />

• Synchrotr<strong>on</strong> radiati<strong>on</strong> is radiati<strong>on</strong> emitted from accelerated<br />

relativistic charged particles.<br />

• This type of radiati<strong>on</strong> can be produced in bending magnets in<br />

storage rings. The number of phot<strong>on</strong>s emitted <strong>and</strong> <strong>the</strong> quality<br />

of <strong>the</strong> phot<strong>on</strong> <strong>beam</strong> can be optimized with magnetic arrays.<br />

<strong>–</strong> Bending magnet <strong>–</strong> magnetic field strength limited by <strong>the</strong><br />

c<strong>on</strong>diti<strong>on</strong> that <strong>the</strong> electr<strong>on</strong>s should stay in orbit in <strong>the</strong><br />

storage ring. The number of emitted phot<strong>on</strong>s is proporti<strong>on</strong>al<br />

to <strong>the</strong> <strong>beam</strong> energy <strong>and</strong> <strong>the</strong> curvature of <strong>the</strong> bend<br />

(denote this flux Φ).<br />

<strong>–</strong> A wiggler can have a str<strong>on</strong>ger magnetic field since it returns<br />

<strong>the</strong> electr<strong>on</strong> <strong>beam</strong> to its original path <strong>–</strong> it is a sequence<br />

of bending magnets. The phot<strong>on</strong> flux is proporti<strong>on</strong>al<br />

to <strong>the</strong> numberof magnetic periods, i.e. ∝ Nw×Φ.<br />

<strong>–</strong> An undulator have many magnetic periods that have<br />

smaller field strengths than <strong>the</strong> wigglers <strong>–</strong> however <strong>the</strong><br />

sp<strong>on</strong>taneous emissi<strong>on</strong> can c<strong>on</strong>structively interfere for certain<br />

wavelengths yielding a dependence of <strong>the</strong> flux that<br />

is proporti<strong>on</strong>al to <strong>the</strong> square of <strong>the</strong> number of magnetic<br />

periods, i.e. ∝ N 2 u ×Φ.<br />

• In a free electr<strong>on</strong> laser <strong>the</strong> electr<strong>on</strong> bunches become density<br />

modulated with <strong>the</strong> period of <strong>the</strong> radiati<strong>on</strong> field. The phot<strong>on</strong><br />

flux is <strong>the</strong>refore proporti<strong>on</strong>al to <strong>the</strong> number of cooperating<br />

electr<strong>on</strong>s in <strong>the</strong> <strong>beam</strong> in additi<strong>on</strong> to <strong>the</strong> undulator flux: ∝<br />

N 2 u ×Ne ×Φ.<br />

• Anundulatorhasaspectrumwithharm<strong>on</strong>icsofafundamental<br />

frequency. The frequency spectrum is dominated by <strong>the</strong> sinc<br />

functi<strong>on</strong> ∼ sinx<br />

. The b<strong>and</strong>width of an undulator is inversely<br />

x<br />

proporti<strong>on</strong>al to <strong>the</strong> number of periods.<br />

• The gain-functi<strong>on</strong> for a free electr<strong>on</strong> laser amplifier can be described<br />

by<strong>the</strong> derivative of <strong>the</strong> sinc functi<strong>on</strong>: we must have an<br />

electr<strong>on</strong> <strong>beam</strong> with slightly higher energy than <strong>the</strong> res<strong>on</strong>ance<br />

energy with a small energy spread for optimal gain.<br />

• Sase starts up from noise in <strong>the</strong> <strong>beam</strong> resulting in a r<strong>and</strong>om<br />

selecti<strong>on</strong> of radiating modes <strong>–</strong> giving rise to a broad<br />

spectrum with many spikes with poor l<strong>on</strong>gitudinal coherence.<br />

Each spike is diffracti<strong>on</strong> limited.<br />

• Short-pulses <strong>and</strong> low charge gives fewer radiating modes[51].<br />

Am<strong>on</strong>ochromator betweentwoundulatorsecti<strong>on</strong>s canbeused<br />

to select a desired mode[52].<br />

• Seeding schemes serves to increase <strong>the</strong> micro-bunching before<br />

<strong>the</strong>freeelectr<strong>on</strong>laseramplificati<strong>on</strong> takesplacewhichincreases<br />

<strong>the</strong> degree of l<strong>on</strong>gitudinal coherence.<br />

• Simulati<strong>on</strong>s Genesis[53], Fast[54]. See for instance <strong>the</strong> start<br />

to end simulati<strong>on</strong>s of <strong>the</strong> Lcls facility described in Ref. [55].


3. Free electr<strong>on</strong> laser ”hardware”<br />

Written by: A. Lindblad<br />

3.1 A prototypical FEL amplifier<br />

In this chapter we will have a closer look at what is actually inside <strong>the</strong> tunnel of a<br />

free electr<strong>on</strong> laser facility. Key technologies will be presented <strong>and</strong> at which facilities<br />

<strong>the</strong>y are utilized.<br />

A free electr<strong>on</strong> laser c<strong>on</strong>sists, in principle, of four parts before <strong>the</strong> user experiments<br />

(<strong>and</strong> <strong>the</strong>ir optics): (1) an electr<strong>on</strong> gun optimized for low emittance which injects a<br />

short intense electr<strong>on</strong> bunch into (2) a linear accelerator structure followed by (3) <strong>on</strong>e<br />

or more undulator secti<strong>on</strong>(s) where <strong>the</strong> free electr<strong>on</strong> laser process takes place; in some<br />

cases <strong>the</strong> three first principal secti<strong>on</strong>s are followed in turn by an (4) gas-attenuator<br />

secti<strong>on</strong>.<br />

e - -gun<br />

3.2 Electr<strong>on</strong> guns<br />

Accelerating structure<br />

Figure 3.1: Schematic of a free electr<strong>on</strong> laser facility.<br />

Undulator(s) Gas attenuator<br />

The electr<strong>on</strong> emitter source at <strong>the</strong> start of <strong>the</strong> accelerator is comm<strong>on</strong>ly referred to<br />

as an electr<strong>on</strong> gun. In <strong>the</strong> ideal case it produce a m<strong>on</strong>oenergetic current spike with<br />

minimal spatial extent. This current spike (see Figure 3.2), produced with electr<strong>on</strong><br />

emissi<strong>on</strong>, is <strong>transport</strong>ed <strong>and</strong> focussed with an electric field (sometimes in combinati<strong>on</strong><br />

with a magnetic field) to <strong>the</strong> exit of <strong>the</strong> gun.<br />

Since electr<strong>on</strong>s carry charge <strong>the</strong>y can not be compressed into an arbitrarily small<br />

<strong>beam</strong>. Space charge effects effectively limits <strong>the</strong> minimal size of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong>.<br />

As will be seen below this is <strong>on</strong>e of <strong>the</strong> principal limitati<strong>on</strong>s when c<strong>on</strong>structing guns<br />

for free electr<strong>on</strong> lasers.<br />

47


48 3. Free electr<strong>on</strong> laser ”hardware”<br />

I [A]<br />

Time [t]<br />

y<br />

x<br />

I [A]<br />

Time [t]<br />

Figure 3.2: Ideal properties of an electr<strong>on</strong> gun (left) compared to real life properties.<br />

Electr<strong>on</strong> guns can be divided into categories by <strong>the</strong> method of electric field generati<strong>on</strong><br />

(direct current (DC) radiofrequency (RF)), by <strong>the</strong> method of electr<strong>on</strong> emissi<strong>on</strong>,<br />

i.e. <strong>the</strong>rmi<strong>on</strong>ic, photocathode, cold emissi<strong>on</strong> or plasma source. A general divisi<strong>on</strong> also<br />

exist for accelerators between hot <strong>and</strong> cold technology, i.e. normally c<strong>on</strong>ducting vs.<br />

superc<strong>on</strong>ducting <strong>–</strong> both of which have <strong>the</strong>ir pros <strong>and</strong> c<strong>on</strong>s.<br />

A direct current electrostatic <strong>the</strong>rmi<strong>on</strong>ic gun is arguably <strong>the</strong> simplest: a hot cathode<br />

emits electr<strong>on</strong>s through <strong>the</strong>rmi<strong>on</strong>ic emissi<strong>on</strong> (i.e. <strong>the</strong> cathode is hot enough so<br />

that electr<strong>on</strong>s can escape from <strong>the</strong> surface) cased by <strong>the</strong> heating from <strong>the</strong> direct current<br />

going through <strong>the</strong> cathode <strong>–</strong> <strong>the</strong> electr<strong>on</strong>s from <strong>the</strong> cathode are <strong>the</strong>n accelerated<br />

away via an electrostatic field.<br />

Of course <strong>the</strong> accelerating field need not be static. Ei<strong>the</strong>r an RF pulse provide<br />

<strong>the</strong> accelerating potential or a pulsed DC field can provide <strong>the</strong> same. For instance,<br />

<strong>the</strong> Scss free electr<strong>on</strong> laser at <strong>the</strong> Spring-8 site in Japan a <strong>the</strong>rmi<strong>on</strong>ic gun is used<br />

toge<strong>the</strong>r with a pulsed DC field of 500 kV with a CeB6 cathode[56, 57].<br />

Nei<strong>the</strong>r cold emissi<strong>on</strong> (also called field emissi<strong>on</strong>) nor plasma source electrodes are<br />

used for free electr<strong>on</strong> laser electr<strong>on</strong> guns. Field emissi<strong>on</strong> emitters are though an area<br />

of currentresearch[58]. Photocathode guns <strong>on</strong> <strong>the</strong>o<strong>the</strong>r h<strong>and</strong> are used bothat Flash,<br />

Fermi <strong>and</strong> at <strong>the</strong> Lcls. A photocathode gun is c<strong>on</strong>structed from a laser that can<br />

photoi<strong>on</strong>ize a cathode.<br />

General requirements<br />

Generally, <strong>the</strong> <strong>beam</strong> dynamics throughout <strong>the</strong> accelerator downstream from <strong>the</strong> electr<strong>on</strong><br />

gun can be described by independent l<strong>on</strong>gitudinal <strong>and</strong> transverse parts. In this<br />

approximati<strong>on</strong> a few parameters are critical at <strong>the</strong> start of <strong>the</strong> undulator: charge per<br />

bunch, <strong>the</strong> geometrical transverse emittance <strong>and</strong> <strong>the</strong> l<strong>on</strong>gitudinal emittance.<br />

For lasing at X-ray wavelengts, <strong>the</strong> geometric emittance must be smaller than<br />

λ/4π (where λ is <strong>the</strong> phot<strong>on</strong> wavelength), i.e.that <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> must overlap<br />

sufficiently/totally with <strong>the</strong> generated phot<strong>on</strong>-<strong>beam</strong>.<br />

The geometric emittance is proporti<strong>on</strong>al to <strong>the</strong> normalized emittance divided by<br />

<strong>the</strong> <strong>beam</strong> energy[59]<br />

εg = εn<br />

E<br />

thus, a small normalized emittance allow for a lowering of <strong>the</strong> <strong>beam</strong> energy of <strong>the</strong><br />

accelerator <strong>–</strong> which in turn lowers <strong>the</strong> total cost for <strong>the</strong> facility.<br />

The emittance of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> is defined by <strong>the</strong> electr<strong>on</strong> gun where <strong>the</strong> charge<br />

cloud to be accelerated is created. The ultimate performance of a free electr<strong>on</strong> laser is<br />

ultimately determined by quality of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> at <strong>the</strong> start of <strong>the</strong> accelerating<br />

y<br />

x


3.2. Electr<strong>on</strong> guns 49<br />

structure. More precisely <strong>the</strong> emittance of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> needs to be low, as this<br />

quantity can <strong>–</strong> in <strong>the</strong> best case scenario <strong>–</strong> be c<strong>on</strong>served throughout <strong>the</strong> accelerator.<br />

Since performance (<strong>and</strong> cost efficiency) can be gained by c<strong>on</strong>structing an low emittance<br />

electr<strong>on</strong> gun a lot of effort have been (<strong>and</strong> still is) put into research <strong>and</strong> development<br />

in this area (for a recent overview see W. A. Ferrario[60]).<br />

The l<strong>on</strong>gitudinal emittance <strong>and</strong> <strong>the</strong> bunch charge define two important parameters<br />

for <strong>the</strong> lasing process: <strong>the</strong> energy spread <strong>and</strong> <strong>the</strong> peak current. The path difference<br />

of <strong>the</strong> electr<strong>on</strong>s (caused by, for instance energy spread) in <strong>the</strong> undulator over <strong>on</strong>e<br />

gain-length must be very small (very much smaller than <strong>the</strong> radiati<strong>on</strong> wavelength)<br />

allow microbunching to occur.<br />

Table 3.1, c<strong>on</strong>stitutes a wish-list for electr<strong>on</strong> guns suitable for X-ray free electr<strong>on</strong><br />

lasers. As will become evident, it is hard to find guns that simultaneously fulfill all of<br />

<strong>the</strong> menti<strong>on</strong>ed points. The accelerator structure (normal c<strong>on</strong>ducting, superc<strong>on</strong>ducting)<br />

<strong>and</strong> user dem<strong>and</strong>s <strong>on</strong> <strong>the</strong> facility will guide <strong>the</strong> choices.<br />

Parameter Value & Comments<br />

Repetiti<strong>on</strong> rate Hz to 100’s of MHz<br />

Charge per bunch Tens of pC to ∼ nC<br />

Normalized emittance ∼ 0.1 to ∼ 1 µm<br />

Energy at gun exit ≥∼ 0.5 MeV<br />

E-field at <strong>the</strong> cathode ≥∼ 10 MV/m<br />

B-field compatibility emittance compensati<strong>on</strong><br />

Spatial distributi<strong>on</strong> c<strong>on</strong>trollable<br />

Bunch length (rms) fs to 10’s of ps<br />

Vacuum 10 −9 −10 −11 mbar<br />

Load-lock compatibility Facilitate cathode replacement<br />

High reliability User facility operati<strong>on</strong><br />

Table 3.1: Some requirements for X-ray free electr<strong>on</strong> laser electr<strong>on</strong> guns.<br />

The design parameters of a facility vis-à-vis average brightness <strong>and</strong> flux sets <strong>the</strong><br />

repetiti<strong>on</strong> rate <strong>and</strong> peak current; <strong>the</strong> desired radiati<strong>on</strong> wavelength sets <strong>the</strong> electr<strong>on</strong><br />

energy <strong>and</strong> undulator energy <strong>and</strong> field; phot<strong>on</strong> pulse length <strong>and</strong> radiati<strong>on</strong> field intensity<br />

c<strong>on</strong>strain choices of seeding schemes, peak current, total charge, etc.<br />

Choice of cathode materials range from pure metals to various semic<strong>on</strong>ductor compounds<br />

(e.g. CeB6[56], ZrC[61]). They are chosen with respect to <strong>the</strong>ir stability <strong>and</strong><br />

quantum efficiency. With low repetiti<strong>on</strong> rates (up to about 1 kHz) presently available<br />

lasers can in combinati<strong>on</strong> with a low quantum efficiency material (i.e. QE ≈<br />

10 −5 −10 −4 ) achieve a high enough photocurrent to be employed. Megahertz repetiti<strong>on</strong><br />

rates require materials with higher efficiency in <strong>the</strong> order of percents.<br />

Thermi<strong>on</strong>ic emitters<br />

The normalized rms emittance of electr<strong>on</strong>s emitted from a hot cathode of radius rc<br />

can be described by[62]:<br />

εn = βγ � 〈x2 〉〈x ′2 〉 = γ rc<br />

�<br />

kBT<br />

2 mc2


50 3. Free electr<strong>on</strong> laser ”hardware”<br />

-<br />

-<br />

-<br />

-<br />

- - - -<br />

-<br />

- - - - -<br />

+<br />

+<br />

+<br />

+<br />

-<br />

-<br />

-<br />

-<br />

-<br />

- - - -<br />

- - - - -<br />

Figure 3.3: Thermi<strong>on</strong>ic emitter (left) <strong>and</strong> photocathode emitter (right) with a static accelerating gradient.<br />

with T being <strong>the</strong> cathode temperature <strong>and</strong> kB Boltzman’s c<strong>on</strong>stant. Clearly <strong>the</strong> key<br />

to a low emittance is to have a small cathode to begin with. At <strong>the</strong> SCSS (Spring-8)<br />

a <strong>the</strong>rmi<strong>on</strong>ic gun using a CeB6 cathode operating at 1450 ◦ C produces a 3 A peak<br />

current with a emittance as low as 0.4π mm mrad[56].<br />

Photocathode emitters<br />

The emittance from a photo-cathode depends str<strong>on</strong>gly <strong>on</strong> material properties, thus<br />

materials incombinati<strong>on</strong>withlasertechnologyis<strong>the</strong>reforeanactiveareaofresearch[63,<br />

64].<br />

Photocathode guns are used at <strong>the</strong> majority of free electr<strong>on</strong> laser facilities around<br />

<strong>the</strong> world.<br />

• Lcls: currently polycrystaline Cu, CsBr coated Cu being c<strong>on</strong>sidered as a future<br />

alternative[65].<br />

• Flash: Cs2Te.<br />

Semi-c<strong>on</strong>ductor cathode materials are more sensitive to degrading processes such<br />

as i<strong>on</strong> backscattering <strong>and</strong> surface degradati<strong>on</strong> than <strong>the</strong>ir metal low efficiency counterparts.<br />

Never<strong>the</strong>less, by asserting proper vacuum c<strong>on</strong>diti<strong>on</strong>s Cs : GaAs <strong>and</strong> Cs2Te<br />

are operating at user facilities .<br />

For an comprehensive investigati<strong>on</strong> <strong>on</strong> different cathode materials <strong>and</strong> a perspective<br />

of <strong>the</strong> current research efforts see[66]. The emittance can be reduced by cooling<br />

of <strong>the</strong> cathode[62].<br />

Normally c<strong>on</strong>ducting guns<br />

Static (DC) accelerati<strong>on</strong><br />

In this type of electr<strong>on</strong> gun <strong>the</strong> particles are accelerated by a static 1 (or pulsed field<br />

that is static during <strong>the</strong> duty cycle, i.e. when an electr<strong>on</strong> bunch is to be accelerated<br />

into <strong>the</strong> accelerator structure), as depicted in Figure 3.3.<br />

1 Cockcroft & Walt<strong>on</strong> used an electrostatic linear acceleratorwhere an alternatingcurrentsource<br />

is rectified by diodes <strong>and</strong> capacitors to achieve a voltage multiplicati<strong>on</strong> over several stages. They<br />

used <strong>the</strong> machine to split lithium atoms with 400 keV prot<strong>on</strong>s <strong>–</strong> <strong>the</strong> results were published 1932[67];<br />

for this achievement <strong>the</strong>y were rewarded <strong>the</strong> Nobel prize in physics 1951 for ”Transmutati<strong>on</strong> of<br />

atomic nuclei by artificially accelerated atomic particles”.<br />

+<br />

+<br />

+<br />

+


3.2. Electr<strong>on</strong> guns 51<br />

The charges are accelerated byaforce proporti<strong>on</strong>al to <strong>the</strong> gradient of <strong>the</strong> potential,<br />

i.e. <strong>the</strong> voltage difference F = −q∇φ. The energy gained is ∆E = qU with <strong>the</strong> unit<br />

often givenin electr<strong>on</strong> volts (eV).Withthis typeof voltage multiplicati<strong>on</strong> itis possible<br />

to reach voltages of 1-2 MV.<br />

Radio frequency (RF) accelerati<strong>on</strong><br />

An oscillating electromagnetic field can be used to accelerate charged particles. If <strong>the</strong><br />

particles moti<strong>on</strong> is matched so that <strong>the</strong>y interact res<strong>on</strong>antly with <strong>the</strong> rf-field a very<br />

high amount of accelerati<strong>on</strong> can be achieved (see page 52).<br />

Guns operating in <strong>the</strong> L- <strong>and</strong> S-b<strong>and</strong>s 2 (∼ 1−2 GHz <strong>and</strong> 2−4 GHz) have already<br />

been c<strong>on</strong>structed <strong>and</strong> successfully employed in photoinjector schemes, notably <strong>the</strong><br />

Lcls gun at SLAC[68]. Normally c<strong>on</strong>ducting RF-guns can be c<strong>on</strong>sidered a mature<br />

technology that exhibit several important performance parameters in-line with what<br />

is dem<strong>and</strong>ed from a free electr<strong>on</strong> laser point-of-view: <strong>the</strong>y are capable of producing a<br />

high field gradient (up to150 MV/m) which allow <strong>the</strong>extracti<strong>on</strong> of high peak currents<br />

in short bunches; <strong>the</strong>y permit <strong>the</strong> use of emittance compensati<strong>on</strong> through <strong>the</strong> use of<br />

solenoidal magnetic fields; <strong>the</strong>y are compatible with a large numberof various cathode<br />

materials.<br />

The limiting factor is <strong>the</strong> power density exerted <strong>on</strong> <strong>the</strong> cavity walls when <strong>the</strong>y are<br />

submitted to a high accelerating field gradient. A high radio frequency implies that<br />

<strong>the</strong> cavities are comparatively small which makes efficient dissipati<strong>on</strong> of <strong>the</strong> generated<br />

heat through a cooling system technically challenging. Hence <strong>the</strong> repetiti<strong>on</strong> rate for<br />

is limited to a maximum somewhere between 100 Hz <strong>and</strong> about 10 kHz (depending <strong>on</strong><br />

<strong>the</strong> RF-frequency). The small cavities also imply that <strong>the</strong> apertures are small which<br />

can also impair pumping which may generate vacuum quality c<strong>on</strong>cerns.<br />

Below a certain frequency <strong>the</strong> heat load <strong>on</strong> <strong>the</strong> cavity walls becomes manageable<br />

in such a way (with lower RF frequency <strong>the</strong> cavities become larger which decreases<br />

<strong>the</strong> power density) that a c<strong>on</strong>tinuous wave (CW) operati<strong>on</strong> mode can be allowed[69].<br />

Lower frequency implies a lower accelerating gradient which are still higher than <strong>the</strong><br />

alternative ”varm technology” direct current counter parts.<br />

The interest for this type of operati<strong>on</strong> is large from <strong>the</strong> user community since it<br />

allows a higher repetiti<strong>on</strong> rate than stated above, even for a n<strong>on</strong>-superc<strong>on</strong>ducting<br />

apparatus. Owing to <strong>the</strong> corresp<strong>on</strong>dence to RF technology employed at storage ring<br />

this technology is well matured which ensures reliability <strong>and</strong> simplicity hard to find<br />

in o<strong>the</strong>r schemes.<br />

Superc<strong>on</strong>ducting guns<br />

A scheme where superc<strong>on</strong>ducting radiofrequency (SRF) accelerator cavities are combined<br />

with photocathode laser electr<strong>on</strong> guns can potentially allow for <strong>the</strong> producti<strong>on</strong><br />

of electr<strong>on</strong> <strong>beam</strong>s of sufficient quality for usage in free electr<strong>on</strong> lasers at very high<br />

repetiti<strong>on</strong> rates[70].<br />

An overview of <strong>the</strong> current research <strong>and</strong> developments have been published by A.<br />

Arnold <strong>and</strong> co-workers[71].<br />

2 RF sources are classified into VHF, UHF, microwave <strong>and</strong> millimetre waveb<strong>and</strong>s. The microwave<br />

b<strong>and</strong>s are divided into <strong>the</strong> following categories: <strong>the</strong> L b<strong>and</strong>, 1.12-1.7 GHz; S b<strong>and</strong>, 2.6-3.95<br />

GHz; C b<strong>and</strong>, 3.95-5.85 GHz; X b<strong>and</strong>, 8.2-12.4 GHz; K b<strong>and</strong>, 18.0-26.5 GHz. The millimetre wave<br />

b<strong>and</strong> is between 30 <strong>and</strong> 300 GHz.


52 3. Free electr<strong>on</strong> laser ”hardware”<br />

The Meissner effect (exclusi<strong>on</strong> of B-field from superc<strong>on</strong>ducting cavity walls) makes<br />

<strong>the</strong> inclusi<strong>on</strong> of emittance reducing B-fields in <strong>the</strong> source regi<strong>on</strong> problematic. The<br />

use of higher order cavity modes that generate a magnetic comp<strong>on</strong>ent achieving <strong>the</strong><br />

emittance reducti<strong>on</strong> have been proposed <strong>and</strong> are under investigati<strong>on</strong>[72, 73].<br />

Summary<br />

Table 3.2 presents <strong>the</strong> current best <strong>beam</strong> performance as obtained from different<br />

gun technologies. The low emittance of <strong>the</strong> Lcls gun is achieved thanks to <strong>the</strong> low<br />

peak current operati<strong>on</strong>. The PITZ gun has a 10 Hz structure with 1 MHz pulse<br />

substructure. The Rossendorf setup apparently suffers from a damaged cavity <strong>–</strong><br />

impairing <strong>the</strong> strength of <strong>the</strong>ir accelerating field.<br />

Gun Technology Rate Acc. Field E εn C<br />

[Hz] [MV/m] [MeV] [µm] [pC]<br />

Lcls NC RF 120 140 6 0.5 250<br />

3 GHz 0.14 20<br />

PITZ NC RF 10 60 > 5 1.3 1000<br />

(Flash) 1.3 GHz<br />

JLab DC 75·10 6<br />

6 0.35 3 140<br />

Scss Pulsed DC 60 ∼ 60 0.5 0.6 300<br />

Rossendorf SRF 125·10 6<br />

5 ˜1 3 80<br />

Table 3.2: Performances of existing guns employing different technologies. Table obtained from W. A. Barletta<br />

et al. [74].<br />

3.3 Radio-frequency driven accelerators<br />

RF RF RF<br />

RF RF RF RF RF<br />

Figure3.4: Different rfaccelerati<strong>on</strong>schemes. From lefttoright: in<strong>the</strong> betatr<strong>on</strong>chargedparticlesare accelerated<br />

in a spiral path in a static magnetic field; in <strong>the</strong> microtr<strong>on</strong> <strong>the</strong> magnetic field is static but <strong>the</strong> orbit is stretched<br />

l<strong>on</strong>ger for each pass to adapt to <strong>the</strong> particles’ higher kinetic energy; in <strong>the</strong> synchrotr<strong>on</strong> <strong>the</strong> magnetic field<br />

strength is risen per turn to compensate for <strong>the</strong> higher kinetic energy (in a storage ring <strong>the</strong> rf power is matched<br />

to <strong>the</strong> synchrotr<strong>on</strong> radiati<strong>on</strong> losses, hence <strong>the</strong> orbit is kept stable with a c<strong>on</strong>stant magnetic field strength). A<br />

linear accelerator successively accelerate <strong>the</strong> particles without bending <strong>the</strong>ir orbit.<br />

A radio-frequency (RF) accelerator use power from a single RF generator to create<br />

an alternating electric field gradient over <strong>the</strong> gaps of <strong>the</strong> accelerating secti<strong>on</strong>s. Figure<br />

3.4 different particle accelerati<strong>on</strong> schemes are presented, all can be understood<br />

from <strong>the</strong> Lorentz force equati<strong>on</strong> (Equati<strong>on</strong> 2.5): <strong>on</strong>ly an electric field can change


3.3. Radio-frequency driven accelerators 53<br />

<strong>the</strong> kinetic energy of a particle, whereas an magnetic field can change <strong>the</strong> orbit of a<br />

particle (since it exerts a force perpendicular to <strong>the</strong> particle’s velocity) 3 .<br />

In 1924 G. Ising suggested that time-varying electric fields could be used for <strong>the</strong><br />

accelerati<strong>on</strong> of charged particles through a periodic structure of drift-tubes[75]. The<br />

first successful operati<strong>on</strong> of a radio frequency driven linear accelerator was dem<strong>on</strong>strated<br />

in 1928 by R. Wiederöe[76].<br />

In Figure 3.5 an accelerator of this type is outlined. The lengths of <strong>the</strong> drift tubes<br />

needs to be progressively l<strong>on</strong>ger as <strong>the</strong> particles’ velocity increases al<strong>on</strong>g <strong>the</strong>structure.<br />

However, when <strong>the</strong>ir velocity is sufficiently close to <strong>the</strong> speed of light <strong>the</strong>y mainly pick<br />

up energy, thus after a while <strong>the</strong> length of <strong>the</strong> drift tubes need not to be increased.<br />

Since <strong>the</strong> length of <strong>the</strong> structures in a Wiederöe linac is βλ/2 it makes ec<strong>on</strong>omical<br />

sense to choose higher radio frequency <strong>–</strong> since <strong>the</strong> overall accelerator would become<br />

shorter. However, <strong>the</strong> structure radiates energy as P = ωrfCV 2<br />

rf <strong>–</strong> <strong>the</strong> losses thus<br />

increase with <strong>the</strong> radio-frequency ωrf, <strong>the</strong> gap capacitance C <strong>and</strong> <strong>the</strong> voltage squared.<br />

If <strong>the</strong> drift tube is placed in a cavity <strong>the</strong> electromagnetic energy is also stored within<br />

<strong>the</strong> structure in a magnetic field (owing to <strong>the</strong> cavities inductive properties). The<br />

res<strong>on</strong>ant frequency of <strong>the</strong> cavity can of course be tuned (via <strong>the</strong> cavity radius) to<br />

match that of <strong>the</strong> accelerating field.<br />

The mass of an electr<strong>on</strong> is 1832 times smaller than a prot<strong>on</strong>, an electr<strong>on</strong> thus<br />

achieves relativistic velocities much faster with <strong>the</strong> same accelerating force. Therefore<br />

linear accelerators for electr<strong>on</strong>s generally have structures that have equal length since<br />

<strong>the</strong> velocity factor is β ≈ 1 already after <strong>the</strong> electr<strong>on</strong> gun accelerating structure.<br />

The governing principles for linear accelerators using res<strong>on</strong>ant accelerati<strong>on</strong> are<br />

slightlymorec<strong>on</strong>volutedthanthoseof<strong>the</strong>electrostatic accelerators menti<strong>on</strong>edabove[77].<br />

We will c<strong>on</strong>sider here <strong>on</strong>ly a few key features of <strong>the</strong>ir comp<strong>on</strong>ents necessary for underst<strong>and</strong>ing<br />

of accelerator parameters necessary for free electr<strong>on</strong> laser.<br />

3 At relativistic speeds v ≈ c <strong>the</strong> sec<strong>on</strong>d term in F = −q[E + v × B] may be about 300 times<br />

larger than <strong>the</strong> first term already at 1 T magnetic strength (which is readily achievable technologically).<br />

+ - + -<br />

+<br />

βλ/2<br />

Figure 3.5: Schematic of <strong>the</strong> cavity structure of a Wiederöe accelerator. To <strong>the</strong> right <strong>the</strong>re is a source for <strong>the</strong><br />

particles to be accelerated.


54 3. Free electr<strong>on</strong> laser ”hardware”<br />

The accelerating RF-field<br />

A good starting point to get a feeling for how <strong>the</strong> accelerating electric field within<br />

a accelerator cavity looks is to c<strong>on</strong>sider <strong>the</strong> field within a capacitor. If we, initially,<br />

assume <strong>the</strong> field to vary very slowly with <strong>the</strong> angular frequency ω we can write <strong>the</strong><br />

field as:<br />

E = E0e iωt<br />

(3.1)<br />

i.e. with <strong>the</strong> alternating field <strong>the</strong> charges <strong>on</strong> <strong>the</strong> plates gets depleted <strong>and</strong> accumulated<br />

sequentially.<br />

Γ<br />

E<br />

B<br />

Figure 3.6: A capacitor c<strong>on</strong>nected to an alternating current source stores both an electric <strong>and</strong> a magnetic field.<br />

The Γ c<strong>on</strong>tour (dashed) is an integrati<strong>on</strong> path.<br />

We know that a varying electric field induces a magnetic field (from <strong>the</strong> Ampère-<br />

Maxwell equati<strong>on</strong>, eq. 2.4). Inside <strong>the</strong> capacitor we have no stored current <strong>and</strong> thus<br />

(Figure 3.6):<br />

c 2<br />

�<br />

Γ<br />

B·dℓ = ∂<br />

�<br />

E·dS (3.2)<br />

∂t S<br />

where S is <strong>the</strong> area enclosed within Γ. The c<strong>on</strong>tour is a circle with radius r.<br />

c 2 B2πr = ∂<br />

∂t Eπr2 , thus B = iωr<br />

2c 2E0eiωt<br />

(3.3)<br />

Where we have made use of our definiti<strong>on</strong> of <strong>the</strong> oscillating electric field.<br />

If <strong>the</strong> time derivative of <strong>the</strong> electric field is identically zero (i.e. a static electric<br />

field) all energy in <strong>the</strong> capacitor was stored in <strong>the</strong> electric field. Now with a<br />

time-varying field <strong>the</strong>re is <strong>the</strong> additi<strong>on</strong>al possibility of storing energy in <strong>the</strong> induced<br />

magnetic field.<br />

In <strong>the</strong> center of <strong>the</strong> capacitor (r = 0) <strong>the</strong>re is no magnetic field, elsewhere <strong>the</strong>re is<br />

an induced field that varies with <strong>the</strong> distance from <strong>the</strong> center. Since such a magnetic<br />

field is present <strong>the</strong>re is a increasing perturbati<strong>on</strong> of <strong>the</strong> electric field with increasing


3.3. Radio-frequency driven accelerators 55<br />

r. It is now possible to c<strong>on</strong>struct a correcti<strong>on</strong> to <strong>the</strong> electric field that takes this into<br />

account using Faraday’s law eq. 2.3:<br />

E = E1 +E2<br />

Γ2<br />

S0<br />

(3.4)<br />

Figure 3.7: The capacitor viewed from <strong>the</strong> side, depicting <strong>the</strong> surface <strong>and</strong> integrati<strong>on</strong> path used for Faraday’s<br />

law.<br />

The sec<strong>on</strong>d term in <strong>the</strong> superpositi<strong>on</strong> is required to be zero in <strong>the</strong> center, it is also<br />

<strong>the</strong> <strong>on</strong>ly term c<strong>on</strong>tributing to <strong>the</strong> line integral in (Figure 3.7):<br />

�<br />

Γ2<br />

E·dℓ = − ∂ΦB<br />

∂t<br />

The flux of <strong>the</strong> magnetic field in a vertical strip of width dr is B(r)hdr (imagine that<br />

we split <strong>the</strong> surface S0 into strips). The right h<strong>and</strong> side boils down to −E2(r)h:<br />

�<br />

−hE2(r) = −h B(r)dr<br />

Here it can be seen that <strong>the</strong> correcti<strong>on</strong> field does not depend <strong>on</strong> <strong>the</strong> separati<strong>on</strong> of<br />

<strong>the</strong> fields, <strong>on</strong>ly <strong>on</strong> <strong>the</strong> distance from <strong>the</strong> center. The equati<strong>on</strong> above gives E2(r) =<br />

− ω2 r 2<br />

4c 2 E0e iωt , this gives us <strong>the</strong> corrected electric field in <strong>the</strong> capacitor<br />

E = E1 +E2 =<br />

�<br />

1− ω2 r 2<br />

4c 2<br />

�<br />

E0e iωt<br />

(3.5)<br />

Our obtained electric field now differs significantly from <strong>the</strong> <strong>on</strong>e we started out<br />

with to obtain <strong>the</strong> magnetic field above. Since we are dealing with fields we can<br />

repeat <strong>the</strong> same procedure for <strong>the</strong> magnetic field, i.e. use <strong>the</strong> ansatz that<br />

B = B1 +B2<br />

With B1 = iωr<br />

2c 2E0e iωt . To find <strong>the</strong> correcti<strong>on</strong> term we may use <strong>the</strong> Ampère-Maxwell<br />

equati<strong>on</strong> again (as we did above in eq. 3.2):<br />

c 2 B22πr = ∂<br />

∂t<br />

�<br />

S<br />

E·dS


56 3. Free electr<strong>on</strong> laser ”hardware”<br />

The flux of <strong>the</strong> electric field is taken through <strong>the</strong> circle enclosed by Γ in Figure 3.6<br />

<strong>and</strong> we get:<br />

B2(r) = − iω3r 3<br />

16c4 This gives us a new correcti<strong>on</strong> to <strong>the</strong> electric field via<br />

E3(r) = ∂<br />

∂t<br />

�<br />

B2(r)dr<br />

We obtain a new term in <strong>the</strong> expansi<strong>on</strong> of <strong>the</strong> electric field as:<br />

�<br />

E = 1− 1<br />

22 �<br />

ωr<br />

�2 +<br />

c<br />

1<br />

22 ·42 �<br />

ωr<br />

� �<br />

4<br />

E0e<br />

c<br />

iωt<br />

if we were to c<strong>on</strong>tinue we would get an increasingly large expansi<strong>on</strong> within <strong>the</strong> paren<strong>the</strong>sis<br />

which c<strong>on</strong>tinues as (slightly rewritten):<br />

�<br />

1− 1<br />

(1!) 2<br />

�<br />

ωr<br />

�2 +<br />

2c<br />

1<br />

(2!) 2<br />

�<br />

ωr<br />

�4 −<br />

2c<br />

1<br />

(3!) 2<br />

�<br />

ωr<br />

� �<br />

6<br />

+...<br />

2c<br />

This series is, bydefiniti<strong>on</strong>, <strong>the</strong>Bessel functi<strong>on</strong> of <strong>the</strong> first kind(J0) with argument<br />

ωr/c. The oscillating electric field in <strong>the</strong> capacitor can thus be written in <strong>the</strong> very<br />

compact form:<br />

E = J0(ωr/c)E0e iωt<br />

Bessel’s functi<strong>on</strong>s are soluti<strong>on</strong>s to <strong>the</strong> wave equati<strong>on</strong> in a cylindrical geometry. The<br />

subscript zero denotes that this soluti<strong>on</strong> is independent of <strong>the</strong> polar coordinate. In<br />

Figure 3.8 we can see that <strong>the</strong> functi<strong>on</strong> have a zero around 2.4 (it is actually 2.405).<br />

This implies that a pair of plates have a res<strong>on</strong>ant<br />

frequency2.405c/r <strong>–</strong>ofcourse <strong>the</strong>reisalso <strong>the</strong>possibility<br />

for harm<strong>on</strong>ics of this frequency, functi<strong>on</strong>s that will<br />

haveadditi<strong>on</strong>al nodes inside <strong>the</strong>cavity <strong>–</strong>meaningthat,<br />

<strong>the</strong> electric field <strong>and</strong> magnetic fields will exchange en-<br />

ergy in an oscillating manner indefinitely (as a capac-<br />

itor <strong>and</strong> an inductance coupled toge<strong>the</strong>r). If <strong>the</strong> fields<br />

were enclosed in a cylinder with c<strong>on</strong>ducting walls this<br />

would hold if <strong>the</strong> walls were perfect c<strong>on</strong>ductors.<br />

Imperfect c<strong>on</strong>ducting walls implies that <strong>the</strong> oscillating<br />

fields gradually gets drained of <strong>the</strong>ir energy due<br />

to resistive losses.<br />

To describe a res<strong>on</strong>ating cavity, we may imagine<br />

that <strong>the</strong> oscillating field occur in a hollow cylinder<br />

c<strong>on</strong>taining <strong>the</strong> same oscillating field as above (<strong>the</strong> so-<br />

J0(x)<br />

1<br />

0. 8<br />

0. 6<br />

0. 4<br />

0. 2<br />

0<br />

0 1 2 3<br />

r[c/ω]<br />

Figure 3.8: The Bessel functi<strong>on</strong> of<br />

<strong>the</strong> first kind.<br />

luti<strong>on</strong>s are basically <strong>the</strong> same). The lowest mode in such a cylinder is usually denoted<br />

TM010 which can be written:<br />

Ez = E0J0(kcr)cos(ωt−kzz)<br />

The wave numberkz describes <strong>the</strong> dependenceof <strong>the</strong> field up<strong>on</strong> <strong>the</strong> cut-offwavelength<br />

λc of <strong>the</strong> cavity <strong>–</strong> <strong>on</strong>ly electromagnetic waves with wavelengths shorter than <strong>the</strong>


3.3. Radio-frequency driven accelerators 57<br />

cylinder diameter can propagate through <strong>the</strong> structure: kz = √ k2 −k2 c, where kc =<br />

2π/λc <strong>and</strong> <strong>the</strong> free space wavenumber k = ω/c. We require that <strong>the</strong> field vanish at<br />

<strong>the</strong> surface of <strong>the</strong> cylinder, letting <strong>the</strong> radius of <strong>the</strong> cylinder be ρ, thus kc = 2.405/ρ<br />

must hold (2.405 being <strong>the</strong> first zero of <strong>the</strong> Bessel functi<strong>on</strong> J0).<br />

Ifanelectr<strong>on</strong> wouldhave<strong>the</strong>same speedas <strong>the</strong>phasevelocityof<strong>the</strong>wavedescribed<br />

above it would be accelerated, however <strong>the</strong> phase velocity of <strong>the</strong> wave is:<br />

vphase = ω c<br />

= �<br />

kz 1− λ2<br />

λ2 > c<br />

c<br />

The c<strong>on</strong>diti<strong>on</strong> of accelerati<strong>on</strong> is thus possible to fulfill for a very short distance <strong>on</strong>ly<br />

after which <strong>the</strong> wave would reverse <strong>the</strong> field orientati<strong>on</strong> <strong>and</strong> decelerate <strong>the</strong> electr<strong>on</strong><br />

again <strong>–</strong> <strong>on</strong> average no net gain in energy would be possible.<br />

To overcome this, i.e. to make a working accelerating structure, <strong>the</strong> phase velocity<br />

of <strong>the</strong> wave can be slowed down for a certain frequency to match <strong>the</strong> speed of light<br />

by introducing irises al<strong>on</strong>g <strong>the</strong> cavity. Such a structure will naturally perturb <strong>the</strong><br />

fields of <strong>the</strong> perfect cylinder, a working accelerating structure is obtained when <strong>the</strong><br />

wavelength is chosen such that it is a multiple of <strong>the</strong> distances between <strong>the</strong> iris-discs,<br />

λ = nd (see Figure 3.9). The modes inside <strong>the</strong> disc-loaded cavities are usually named<br />

after <strong>the</strong> phase advance kzd = 2π/n of <strong>the</strong> wave per sub-cavity.<br />

Beam<br />

2π/3 mode π mode<br />

Figure 3.9: Two comm<strong>on</strong> modes used in disc-loaded cavities.<br />

Energy gain in a radiofrequency driven accelerating cavity<br />

If U0 ≫ mec 2 <strong>the</strong> electr<strong>on</strong> speed is close to <strong>the</strong> speed of light. Let z = −d/2 be <strong>the</strong><br />

entrance of <strong>the</strong> cavity (corresp<strong>on</strong>ding to a time ωt = −π/2), a particle entering <strong>the</strong><br />

cavity may experience <strong>the</strong> electric field at a phase φ relative to <strong>the</strong> peak field (defining<br />

zero phase).<br />

Then <strong>the</strong> field <strong>on</strong> axis (in <strong>the</strong> ˆz directi<strong>on</strong> with r = 0 in cylindrical coordinates)<br />

can be written:<br />

Ez = E0cos(ωt(z)−φ)<br />

since, <strong>on</strong> axis <strong>the</strong> Bessel functi<strong>on</strong> J0 is equal to unity. If <strong>the</strong> particle is at z <strong>the</strong> time<br />

is given by t(z) = � z<br />

dzv(z). The energy gain is <strong>the</strong>n:<br />

0<br />

�<br />

∆U = −qE·z = q<br />

d/2<br />

−d/2<br />

E0cos(ωt(z)−φ)dz<br />

We know that during <strong>the</strong> passage <strong>the</strong> change in velocity is small <strong>–</strong> we may set t(z) ≈<br />

z/v. With <strong>the</strong> use of a trig<strong>on</strong>ometric identity we can have <strong>the</strong> phase angle outside<br />

<strong>the</strong> integrati<strong>on</strong>. Defining <strong>the</strong> accelerating gradient V = E0d <strong>and</strong> transit time T<br />

∆U = qE0dT cosφ = qVT cosφ<br />

d


58 3. Free electr<strong>on</strong> laser ”hardware”<br />

<strong>the</strong> transit time T = sin(ωd/(2v))<br />

ωd/(2v) is a correcti<strong>on</strong> to <strong>the</strong> particle accelerati<strong>on</strong> owing to<br />

<strong>the</strong> time variati<strong>on</strong> of <strong>the</strong> field while <strong>the</strong> particles transverse <strong>the</strong> cavity.<br />

is <strong>the</strong> energy gain of an electr<strong>on</strong> passing an accelerator of length L. The phase φ<br />

is measured relative to that of <strong>the</strong> traveling wave.<br />

Often <strong>on</strong>e does not operate <strong>the</strong> accelerators at <strong>the</strong> point of maximum accelerati<strong>on</strong><br />

since <strong>on</strong>e wants to introduce a energy chirp (gradient) al<strong>on</strong>g <strong>the</strong> particle bunch to<br />

achieve a shorter bunch.<br />

Warm technology: Copper<br />

Accelerating structures built in Copper have been used for decades in both storage<br />

rings <strong>and</strong> linear accelerators, thusthis technology can be c<strong>on</strong>sidered to be verymature<br />

<strong>and</strong> soluti<strong>on</strong>s are often available commercially. Copper is chosen for its good electric<br />

c<strong>on</strong>ductivity coupled with a high <strong>the</strong>rmal c<strong>on</strong>ductivity.<br />

The cheap cost per GeV (about 15 Milli<strong>on</strong> USD) must though be put against <strong>the</strong><br />

limited average brightness that can be delivered to <strong>the</strong> users.<br />

Cavity structures for <strong>the</strong> L, S, C <strong>and</strong> X b<strong>and</strong>s are currently being used for different<br />

purposes. Thanks to <strong>the</strong> cheap cost <strong>the</strong> technique is being developed for 100-150<br />

MV/m for <strong>the</strong> X-b<strong>and</strong>, mainly for <strong>the</strong> TeV lept<strong>on</strong> linear collider but with obvious<br />

synergetic effects for linear accelerators for X-ray producti<strong>on</strong>[78].<br />

C-b<strong>and</strong> normally c<strong>on</strong>ducting copper structures operating at accelerating gradients<br />

of 35 MV/m are used at <strong>the</strong> Scss free electr<strong>on</strong> laser.<br />

Superc<strong>on</strong>ducting technology<br />

Superc<strong>on</strong>ducting technology is vastly more expensive than <strong>the</strong> warm counterpart;<br />

however, <strong>the</strong> high average brightness required for larger facilities (especially those<br />

where it is envisi<strong>on</strong>ed that many experimental stati<strong>on</strong>s are served simultaneously,<br />

e.g. <strong>the</strong> European X-Fel) this might be <strong>the</strong> <strong>on</strong>ly way forward.<br />

Since <strong>the</strong> short durati<strong>on</strong> <strong>and</strong> low emittance of <strong>the</strong> electr<strong>on</strong> bunches would not be<br />

preserved in a synchrotr<strong>on</strong>, Sase free electr<strong>on</strong> lasers are based <strong>on</strong> linear accelerators,<br />

ei<strong>the</strong>r employing normally c<strong>on</strong>ducting or superc<strong>on</strong>ducting rf structures as sketched<br />

in Figure 3.9. In c<strong>on</strong>trast to <strong>the</strong> case of a c<strong>on</strong>stant current, <strong>the</strong> resistance of a<br />

superc<strong>on</strong>ductor does not completely vanish in <strong>the</strong> presence of an radiofrequency field,<br />

see e.g.[79]. It is never<strong>the</strong>less much smaller in superc<strong>on</strong>ducting niobium at e.g. 2 K<br />

than for copper at room temperature.<br />

Therefore, <strong>the</strong> quality factor Q of a superc<strong>on</strong>ducting cavity (where Q/2π is <strong>the</strong><br />

ratio of<strong>the</strong> energy stored in <strong>the</strong>system <strong>and</strong> <strong>the</strong>energy dissipated per oscillati<strong>on</strong> cycle)<br />

is of <strong>the</strong> order of 10 1 0, compared to below 10 5 for copper cavities. On <strong>the</strong> o<strong>the</strong>r h<strong>and</strong>,<br />

for 1Wof power lost in asuperc<strong>on</strong>ductingcavity at 2K, <strong>the</strong>cryogenic system requires<br />

almost 1 kW to keep <strong>the</strong> temperature c<strong>on</strong>stant. This toge<strong>the</strong>r with <strong>the</strong> technological<br />

complexity <strong>and</strong> a fundamental limitati<strong>on</strong> given by <strong>the</strong> critical magnetic field makes<br />

<strong>the</strong> choice not so obvious.<br />

It required a committee of internati<strong>on</strong>al experts in 2004 to identify <strong>the</strong> superc<strong>on</strong>ducting<br />

TESLA technology as <strong>the</strong> best soluti<strong>on</strong> for <strong>the</strong> Internati<strong>on</strong>al Linear Collider<br />

[80].<br />

At Flash, six accelerati<strong>on</strong> stages, each accommodating eight nine-cell<br />

TESLA cavities used to accelerate <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> to 1 GeV[81]. The average


3.4. Undulators 59<br />

h<br />

D<br />

Figure 3.10: Schematic of an Apple-I variable polarizati<strong>on</strong> undulator. The short magnet in <strong>the</strong> start have λu/8.<br />

electric field is about 20 MV/m with <strong>on</strong>e meter cavity lengths. By optimizing cavity<br />

design <strong>and</strong> with <strong>the</strong> advent of new techniques to clean <strong>the</strong> cavity surfaces, fields larger<br />

than 50 MV/m have been dem<strong>on</strong>strated[82].<br />

The European X-ray free electr<strong>on</strong> laser will employ TESLA cavities with a field of<br />

21 MV/m. Lcls at Stanford, <strong>on</strong> <strong>the</strong> o<strong>the</strong>r h<strong>and</strong>, is based <strong>on</strong> <strong>the</strong> normal-c<strong>on</strong>ducting<br />

SLAC linac, <strong>and</strong> some o<strong>the</strong>r projects, like <strong>the</strong> Scss XFEL in Japan for example, have<br />

opted for normal-c<strong>on</strong>ducting rf structures as well.<br />

C<strong>on</strong>tinuoswaveoperati<strong>on</strong> withTESLAcavitieshaverecentlybeendem<strong>on</strong>strated[83]<br />

3.4 Undulators<br />

Most free electr<strong>on</strong> lasers use planar undulators with a fixed undulator period, which<br />

<strong>the</strong>n provide linearly polarized light c<strong>on</strong>taining a fundamental wavelength <strong>and</strong> its<br />

harm<strong>on</strong>ics. If <strong>on</strong>e desires to suppress <strong>the</strong> higher harm<strong>on</strong>ics, for some reas<strong>on</strong>, a<br />

quasiperiodic scheme can be employed[84]. A quasiperiodic undulator scheme generally<br />

degrade <strong>the</strong> performance of <strong>the</strong> undulator for <strong>the</strong> fundamental which makes<br />

such devices unattractive for <strong>the</strong> use in, at least, Sase free electr<strong>on</strong> laser, since <strong>the</strong><br />

cost is highly dependent <strong>on</strong> <strong>the</strong> undulator length <strong>and</strong> an decrease in phot<strong>on</strong> density<br />

necessitates a l<strong>on</strong>ger gain length.<br />

However, oftenpolarizati<strong>on</strong>s o<strong>the</strong>rthan<strong>the</strong>linearisdesiredby<strong>the</strong>usercommunity.<br />

By introducing a magnetic field parallel to <strong>the</strong> central path through <strong>the</strong> undulator<br />

(superimposing <strong>the</strong> vertical field) it is, in principle, possible to generate arbitrary<br />

polarizati<strong>on</strong> directi<strong>on</strong>s[85].<br />

InFigure3.10(adaptedfromRef.[86])aschematicofanApple-1(AdvancedPlanar<br />

Polarized Light Emitter) undulator. Two magnetic fields (sinusoidal <strong>and</strong> helical) are<br />

superimposed <strong>on</strong> each o<strong>the</strong>r with variable strength <strong>and</strong> phase. The type II <strong>and</strong> III<br />

varieties looks similar but have different pole geometries[87, 88].<br />

By placing a variable polarizati<strong>on</strong> device after <strong>the</strong> Sase undulator or using it<br />

directly as <strong>the</strong> primary undulator in <strong>the</strong> free electr<strong>on</strong> laser elliptically polarized light<br />

can be delivered to <strong>the</strong> user experiments.<br />

The res<strong>on</strong>ance wavelength in this type (Apple-1) of undulator looks slightly differ-<br />

y<br />

x<br />

z


60 3. Free electr<strong>on</strong> laser ”hardware”<br />

ent than our expressi<strong>on</strong> found earlier as we now have two fields in <strong>the</strong> undulator:<br />

λr = λu<br />

E2 �<br />

1+ K2 x +K 2� y<br />

2<br />

Here we obtain <strong>the</strong> res<strong>on</strong>ance wavlength in ˚Angström if <strong>the</strong> undulator period is given<br />

in millimeters <strong>and</strong> <strong>the</strong> <strong>beam</strong>’s energy in GeV. The coupling parameter between <strong>the</strong><br />

undulator field <strong>and</strong> <strong>the</strong> radiati<strong>on</strong> fields is also different with <strong>the</strong> [JJ] Bessel-functi<strong>on</strong><br />

factor reducing to unity[17].<br />

Undulator tolerances example<br />

Using <strong>the</strong> Pierce parameter ρ (Equati<strong>on</strong> 2.36) <strong>on</strong>e can estimate <strong>the</strong> accuracy needed<br />

to be achieved, with errors from different sources: temperature, alignment, undulator<br />

gap accuracy <strong>and</strong> flatness change up<strong>on</strong> undulator gap changes.<br />

The fundamental wavelength have to be tuned within<br />

ρ ≥ ∆λ<br />

λ<br />

(3.6)<br />

For an error to have a large impact <strong>on</strong> <strong>the</strong> amplificati<strong>on</strong> process it has to be in<br />

effect for at least <strong>on</strong>e gain length. As an example we can use <strong>the</strong> European Xfel<br />

which has a gain length of 10 meters <strong>and</strong> undulator secti<strong>on</strong>s that are 5 meters l<strong>on</strong>g <strong>–</strong><br />

<strong>the</strong> ρ = 3·10 −3 at 0.1 nm for this facility. If all error sources are equally severe, <strong>the</strong>n<br />

<strong>on</strong>e can obtain an idea of <strong>the</strong> tolerances required for an free electr<strong>on</strong> laser undulator<br />

system[89]. We can express <strong>the</strong> b<strong>and</strong>width as a functi<strong>on</strong> of different errors impacting<br />

<strong>the</strong> magnetic field in <strong>the</strong> undulators:<br />

�<br />

�<br />

∆λ = �<br />

� ∂λ<br />

�<br />

��<br />

�<br />

∂B�<br />

∆B2 temp. +∆B 2 gap +∆B 2 flat (3.7)<br />

from this <strong>on</strong>e obtains that<strong>–</strong>for Equati<strong>on</strong> 3.6 tobe fulfilled <strong>–</strong><strong>the</strong> alignment between<br />

undulator secti<strong>on</strong>s needs to be within ±100 µm; <strong>the</strong> temperature stability of <strong>the</strong><br />

whole undulator system within ±0.08 K <strong>–</strong> <strong>and</strong> that <strong>the</strong> gap adjustment accuracy <strong>and</strong><br />

<strong>the</strong> flatness preservati<strong>on</strong> up<strong>on</strong> gap changes needs to be better than ±1 µm.


3.4. Undulators 61<br />

Summary<br />

• The basic comp<strong>on</strong>ents of a free electr<strong>on</strong> laser before <strong>the</strong> user’s<br />

experiments are:<br />

<strong>–</strong> An electr<strong>on</strong>-gun. Which can be of <strong>the</strong>rmi<strong>on</strong>ic, photocathode<br />

or field emissi<strong>on</strong> type. The electr<strong>on</strong> gun defines<br />

<strong>the</strong> emittance <strong>and</strong> <strong>the</strong> number of electr<strong>on</strong>s in <strong>the</strong><br />

bunches.<br />

<strong>–</strong> A linear accelerator. Defines <strong>the</strong> repetiti<strong>on</strong> rate of <strong>the</strong><br />

system. Superc<strong>on</strong>ducting accelerators have higher repetiti<strong>on</strong><br />

rates (kHz-MHz) than normally c<strong>on</strong>ducting (100’s<br />

of Hz).<br />

<strong>–</strong> Undulator(s) sets <strong>the</strong> wavelength <strong>and</strong> polarizati<strong>on</strong> of <strong>the</strong><br />

X-rays.<br />

<strong>–</strong> (Gas attenuator) - can be used to limit <strong>the</strong> intensity at<br />

<strong>the</strong> experiments.


4. X-ray optics<br />

The material presented here in this chapter is partly adapted from ”Survey of<br />

in situ metrology for <strong>the</strong> measurement of damage to FEL phot<strong>on</strong> <strong>transport</strong><br />

optics” by A. J. Glees<strong>on</strong>, Iruvx WP7, 2010 by A. Lindblad.<br />

It is a challenge to c<strong>on</strong>struct any optical system for X-rays. The requirements <strong>on</strong><br />

optical elements at free electr<strong>on</strong> laser is often different than those imposed at synchrotr<strong>on</strong><br />

X-ray sources because of <strong>the</strong> high peak-power ra<strong>the</strong>r than <strong>the</strong> high average<br />

power. The dem<strong>and</strong>s <strong>on</strong> precisi<strong>on</strong> is also higher since <strong>the</strong> phot<strong>on</strong> <strong>beam</strong> from <strong>the</strong> free<br />

electr<strong>on</strong> laser undulator(s) have a very high brilliance that <strong>on</strong>e wishes to preserve to<br />

as high degree as possible. In <strong>the</strong> following secti<strong>on</strong>s we will investigate how stringent<br />

those dem<strong>and</strong>s are <strong>and</strong> <strong>the</strong> techniques that can be used to achieve this.<br />

Also c<strong>on</strong>nected to <strong>the</strong> high brilliance <strong>and</strong> fluence is damage to optical elements <strong>and</strong><br />

methods to minimize/prevent it. This will be discussed below before <strong>the</strong> chapter end,<br />

c<strong>on</strong>taining descripti<strong>on</strong>s of dispersive optics, m<strong>on</strong>ochromators <strong>and</strong> <strong>beam</strong> attenuators.<br />

If we write <strong>the</strong> index of refracti<strong>on</strong> for X-rays in matter as:<br />

n = 1−δ −iβ<br />

with δ <strong>and</strong> β as <strong>the</strong> refracti<strong>on</strong> <strong>and</strong> absorpti<strong>on</strong> coefficients respectively a number of<br />

c<strong>on</strong>clusi<strong>on</strong>s can be drawn. In <strong>the</strong> X-ray regime <strong>the</strong> reflecti<strong>on</strong> coefficient δ is in <strong>the</strong><br />

order of 10 -5 to 10 -7 , generally thus a refractive optic would not be very effective. The<br />

absorpti<strong>on</strong> coefficient is very high in <strong>the</strong> UV-soft X-ray regi<strong>on</strong> for all elements which<br />

present yet ano<strong>the</strong>r tamper, however for materials c<strong>on</strong>stituted of elements with low<br />

mass (<strong>and</strong> thus small nuclear charge) this coefficient is small above 1.5 keV <strong>–</strong> since<br />

<strong>the</strong> photoi<strong>on</strong>izati<strong>on</strong> cross-secti<strong>on</strong>s become very low <strong>the</strong>re. For instance, <strong>the</strong> deepest<br />

laying single-i<strong>on</strong>izati<strong>on</strong> threshold for carb<strong>on</strong> is around 300 eV.<br />

Absorpti<strong>on</strong> can be fur<strong>the</strong>r traded into reflecti<strong>on</strong> by having a grazing incidence of<br />

<strong>the</strong> X-rays <strong>on</strong>to <strong>the</strong> optical elements[90]. In <strong>the</strong> soft X-ray regi<strong>on</strong> this is <strong>the</strong> general<br />

approach utilized. For hard X-rays reflecti<strong>on</strong> planes in crystals of low Z materials can<br />

be used.<br />

4.1 Dem<strong>and</strong>s <strong>on</strong> optics precisi<strong>on</strong> at free electr<strong>on</strong> lasers<br />

A real-life mirror is not perfect <strong>and</strong> imperfecti<strong>on</strong>s of various kinds distort <strong>and</strong> degrades<br />

<strong>the</strong> mirror’s performance. To analyze <strong>the</strong> impact of various errors <strong>on</strong>e usually<br />

63


64 4. X-ray optics<br />

compares with an idealized surface which perfectly transfer <strong>the</strong> source point to <strong>the</strong><br />

image point.<br />

It can be shown that low frequency surface errors, between millimeters <strong>and</strong> λ,<br />

distort <strong>the</strong> wavefr<strong>on</strong>t but still image <strong>the</strong> incident wave into <strong>the</strong> image plane (often<br />

taken to be within <strong>the</strong> 1/e intensity points in a Gaussian spot). Higher frequency<br />

errors scatter <strong>the</strong> incident waves outside <strong>the</strong> image plane[91].<br />

The authors of Ref. [91] have developed a model classifying an optical systems<br />

performance in terms of <strong>the</strong> slope error (deviati<strong>on</strong> from ideal shape) <strong>and</strong> surface<br />

roughness <strong>–</strong> <strong>the</strong> system coherence length describes <strong>the</strong> limit between ’high’ <strong>and</strong> ’low’<br />

frequencies in a natural way:<br />

λ<br />

W = √ 2<br />

Θcosϑi<br />

treating Θ as <strong>the</strong> given angular radius for <strong>the</strong> image <strong>and</strong> ϑi <strong>the</strong> incident angle relative<br />

to <strong>the</strong> surface parallel we have, as a functi<strong>on</strong> of <strong>the</strong> operating wavelength λ <strong>the</strong><br />

wavelength W which is <strong>the</strong> surface spatial wavelength that diffracts intensity into <strong>the</strong><br />

1/e radius of a Gaussian spot.<br />

For a diffracti<strong>on</strong> limited source (as is free electr<strong>on</strong> laser <strong>and</strong> modern synchrotr<strong>on</strong><br />

storage rings) <strong>the</strong> W is roughly equal to <strong>the</strong> illuminated length (ℓ) of <strong>the</strong> optical<br />

element. For <strong>the</strong> X-ray optics c<strong>on</strong>sidered here this wavelength thus lies between λ<br />

<strong>and</strong> ℓ.<br />

The degradati<strong>on</strong> of transmissi<strong>on</strong> <strong>and</strong> image quality in <strong>the</strong> optical system due to<br />

mirror errors may be described by <strong>the</strong> deviati<strong>on</strong> from <strong>the</strong> ideal case, <strong>the</strong> so-called<br />

Strehl-factor:<br />

I(0) 8<br />

≈ 1−<br />

I0(0) Θ2δ2 �<br />

4π<br />

−<br />

λ cosϑi<br />

�2 σ 2<br />

with δ <strong>and</strong> σ are <strong>the</strong> rms values of <strong>the</strong> slope error <strong>and</strong> surface roughness within <strong>the</strong><br />

b<strong>and</strong>-pass of <strong>the</strong> optical comp<strong>on</strong>ent[91].<br />

Both<strong>the</strong>slopeerror<strong>and</strong><strong>the</strong>surface roughnessareintegral propertiesof<strong>the</strong>surface,<br />

i.e. <strong>the</strong> accumulated error is what affects <strong>the</strong> <strong>beam</strong>:<br />

δ 2 = 4π 2<br />

1/W �<br />

1/L<br />

dfxS1(fx)f 2 x; σ 2 =<br />

�<br />

1/λ<br />

1/W<br />

dfxS1(fx)<br />

Owing to this property of <strong>the</strong> imaging system <strong>on</strong>e is faced with an array of different<br />

problems when it comes to manufacturing <strong>and</strong> commissi<strong>on</strong>ing a device with such<br />

stringent dem<strong>and</strong>s <strong>on</strong> surface quality:<br />

• Manufacturing<br />

<strong>–</strong> Polishing a surface down to a few nanometers peak-to-peak roughness over<br />

a large area (typically <strong>the</strong> mirrors are above 25 cm l<strong>on</strong>g)<br />

<strong>–</strong> Process such a large substrate to a determined shape.<br />

• Metrology of <strong>the</strong> manufactured surface with high (sub-nm) accuracy[92, 93]. A<br />

detailed account of techniques <strong>and</strong> metrology methods for <strong>the</strong> manufacturing<br />

of X-ray optical elements is given in <strong>the</strong> book ”Modern developments in X-ray<br />

<strong>and</strong> Neutr<strong>on</strong> Optics” edited by A. Erko et al.[94].


4.2. Focussing mirrors <strong>–</strong> back-reflecting geometry example 65<br />

• Moving <strong>and</strong> mounting <strong>the</strong> finished element into its positi<strong>on</strong> in <strong>the</strong> optical system<br />

of <strong>the</strong> facility without degrading <strong>the</strong> surface. One way to circumvent this<br />

problem is to make mirrors that have an mount where <strong>the</strong> surface can be mechanically<br />

distorted to compensate for <strong>the</strong> errors <strong>–</strong> so called adaptive optics (see<br />

e.g. Ref. [95]).<br />

As an example we here take <strong>the</strong> offset mirrors at <strong>the</strong> Lcls, <strong>the</strong>y have figure errors<br />

<strong>on</strong> <strong>the</strong> order of 2 nm (rms)[96] <strong>and</strong> because of <strong>the</strong> reflectivity c<strong>on</strong>diti<strong>on</strong> <strong>on</strong> <strong>the</strong> grazing<br />

incidence <strong>the</strong>y need to be of different lengths for <strong>the</strong> soft <strong>and</strong> hard X-ray regi<strong>on</strong>s:<br />

• soft X-rays, 25 cm l<strong>on</strong>g, bor<strong>on</strong>-carbide-coated a total of four.<br />

• hard X-rays, pair of 45 cm l<strong>on</strong>g mirrors with SiC-coating <strong>and</strong> 25 keV cut-off.<br />

Allows <strong>the</strong> throughput of <strong>the</strong> 3 rd harm<strong>on</strong>ic of <strong>the</strong> 8.3 keV fundamental.<br />

4.2 Focussing mirrors <strong>–</strong> back-reflecting geometry example<br />

So far we have <strong>on</strong>ly discussed plane mirrors, <strong>the</strong>re is naturally a dem<strong>and</strong> for focussing<br />

mirrors for experiments that utilize high phot<strong>on</strong>-density. Figure 4.1 shows a princi-<br />

Figure 4.1: Spherical/Parabolic mirror used to focus <strong>the</strong> free electr<strong>on</strong> laser <strong>beam</strong>. <strong>the</strong> focuspoint do not overlap<br />

with <strong>the</strong> incoming <strong>beam</strong> because of <strong>the</strong> <strong>beam</strong>stopper (black).<br />

ple for an experiment that uses a backscattering geometry with a focussing mirror<br />

that focus <strong>the</strong> phot<strong>on</strong><strong>beam</strong> to a point. Such set-ups are utilized, for instance, when<br />

investigating multiphot<strong>on</strong> i<strong>on</strong>izati<strong>on</strong> of gases. One such experiment utilized a Mo-Si<br />

multilayer spherical mirror with 68% reflectance (this number is made possible by<br />

progress in EUV-litography[97]) <strong>–</strong> which could <strong>the</strong>n be focussed down to 3 to 5 µm<br />

focus diameters[98].<br />

4.3 Damage<br />

Because of <strong>the</strong> high peak powers <strong>and</strong> short pulses routinely achieved at free electr<strong>on</strong><br />

laser it is entirely possible that new kinds of damage mechanisms that degrade optical<br />

comp<strong>on</strong>ents in X-ray <strong>beam</strong>lines have to be c<strong>on</strong>sidered. Often our knowledge for such<br />

mechanisms emanate from <strong>the</strong> world of storage ring synchrotr<strong>on</strong>s where ra<strong>the</strong>r a high<br />

average power load provides <strong>the</strong> source for potential damage to optical elements <strong>and</strong><br />

coatings.<br />

At a free electr<strong>on</strong> laser <strong>the</strong> high peak power could potentially render an (costly)<br />

optical comp<strong>on</strong>ent useless within fracti<strong>on</strong>s of a sec<strong>on</strong>d[99]. The research of various<br />

damagemechanismscausedbythiskindofsources is<strong>the</strong>reforeanactivefield[100,101].<br />

The empirical data for damage thresholds that exist are often very specific to a certain<br />

(synchrotr<strong>on</strong>) <strong>beam</strong>line <strong>and</strong> are thus unique to <strong>the</strong> flux <strong>and</strong> operating wavelength


66 4. X-ray optics<br />

under c<strong>on</strong>siderati<strong>on</strong>. Recently materials for optics <strong>and</strong> <strong>the</strong>ir coatings employed at Xray<br />

free electr<strong>on</strong> laser around <strong>the</strong> world have begun to be studied in a more systematic<br />

fashi<strong>on</strong>[101, 102], notably <strong>the</strong> low Z materials B4C <strong>and</strong> SiC[103].<br />

Damage to opical elements at a free electr<strong>on</strong> laser can be thought of as direct <strong>and</strong><br />

in-direct: direct damage could be ablati<strong>on</strong> cratering of <strong>the</strong> surface, distorti<strong>on</strong> due to<br />

<strong>the</strong> heat-load caused by <strong>the</strong> pulse, etc. In-direct damage mechanisms are more subtle<br />

<strong>and</strong> include damage to multilayers by diffusi<strong>on</strong> or chemical modificati<strong>on</strong> of surfaces<br />

<strong>and</strong> layers, changes to <strong>the</strong> refractive index.<br />

Careful m<strong>on</strong>itoringof<strong>the</strong><strong>beam</strong>line performance is thusimportant<strong>and</strong><strong>the</strong>methods<br />

described in <strong>the</strong> sec<strong>on</strong>d part of this book inherently give informati<strong>on</strong> that can be used<br />

to diagnose <strong>the</strong> optics.<br />

4.4 Diffracti<strong>on</strong> gratings<br />

A diffracti<strong>on</strong> grating (henceforth taken to be syn<strong>on</strong>ymous to ’grating’ <strong>on</strong>ly) is a periodically<br />

structuredsurface thatdivides <strong>and</strong>diffracts alight<strong>beam</strong> intoseveral <strong>beam</strong>lets<br />

propagating in different directi<strong>on</strong>s depending <strong>on</strong> <strong>the</strong>ir wavelength. This can readily<br />

be understood from <strong>the</strong> Huygens-Fresnel principle <strong>–</strong> stating that each point <strong>on</strong> a<br />

wavefr<strong>on</strong>t act as an independent source <strong>and</strong> that by adding up <strong>the</strong> c<strong>on</strong>tributi<strong>on</strong> from<br />

such pointsources <strong>the</strong> properties of any subsequent point can be found.<br />

Light reflected from an an ideal grating can be c<strong>on</strong>sideredt to equal to that from<br />

emitted from a set of infinitely l<strong>on</strong>g narrow slits spaced with distance d from each<br />

o<strong>the</strong>r. If we add up <strong>the</strong> <strong>beam</strong>s from each slit at some point (far) away from <strong>the</strong><br />

grating <strong>the</strong> optical path difference between <strong>the</strong> <strong>beam</strong>s will casue positive <strong>and</strong> negative<br />

interference owing to <strong>the</strong> phase difference between <strong>the</strong> waves. If <strong>the</strong> path difference<br />

between <strong>the</strong> slits are some multiple of <strong>the</strong> wavelength d = n·λ positive interference<br />

occur <strong>–</strong> for a given wavelength this c<strong>on</strong>diti<strong>on</strong> will be fulfilled at some angle ϑ away<br />

from <strong>the</strong> normal (taken to be perpendicular to <strong>the</strong> surface)<br />

dsinϑ = n·λ<br />

Ifweallow for <strong>the</strong>incominglight’s (c<strong>on</strong>sidered tobeaplanewave)angle tobedifferent<br />

from that of <strong>the</strong> outgoing we get a generalizati<strong>on</strong> of <strong>the</strong> equati<strong>on</strong> above:<br />

d(sinϑout −sinϑin) = n·λ (4.1)<br />

This equati<strong>on</strong> is comm<strong>on</strong>ly referred to as <strong>the</strong> grating equati<strong>on</strong>. Since <strong>the</strong> result was<br />

obtained using <strong>the</strong> phase differences this holds for regular structures, i.e. wavefr<strong>on</strong>t<br />

distorti<strong>on</strong>s occur through <strong>the</strong> irregularities of <strong>the</strong> grating.<br />

The soluti<strong>on</strong> corresp<strong>on</strong>ding to n = 0 is <strong>the</strong> zeroth order comp<strong>on</strong>ent menti<strong>on</strong>ed<br />

before (thisisakintospecular reflecti<strong>on</strong>atamirror); <strong>the</strong>n<strong>on</strong>zerosoluti<strong>on</strong> corresp<strong>on</strong>ds<br />

to different diffracti<strong>on</strong> orders.<br />

A grating is thus a dispersive optical comp<strong>on</strong>ent that can be used in at normal or<br />

grazing incidence todeflect a certain wavelength part of <strong>the</strong> <strong>beam</strong> intoadefinedangle.<br />

At short wavelengths (in <strong>the</strong> EUV/soft X-ray regi<strong>on</strong>s) it is necessary to operate with<br />

grazing incidences since <strong>the</strong> reflectivity materials increase with decreasing angles.<br />

The zeroth order diffracti<strong>on</strong> have <strong>the</strong> same diffracti<strong>on</strong> angle as <strong>the</strong> incoming angle<br />

whereas <strong>the</strong> diffracted orders have a different angle as compared to <strong>the</strong> incoming light<br />

(angles α <strong>and</strong> β in Figure 4.3). Since <strong>the</strong> grating splits <strong>the</strong> <strong>beam</strong>, ei<strong>the</strong>r <strong>the</strong> diffracted


4.5. M<strong>on</strong>ochromators 67<br />

ordersor<strong>the</strong>zerothordercanbeusedfor<strong>beam</strong>diagnostic purposeswithoutdisturbing<br />

<strong>the</strong> user’s experimental stati<strong>on</strong>s downstream <strong>–</strong> this aspect is fur<strong>the</strong>r discussed below<br />

in chapter 5 (see page 73),<br />

Because of <strong>the</strong>ir dispersive nature gratings act as b<strong>and</strong>pass filters <strong>on</strong> <strong>the</strong> radiati<strong>on</strong><br />

<strong>and</strong> can thus be used to define <strong>the</strong> wavelength at <strong>the</strong> experimental stati<strong>on</strong>s (see<br />

below).<br />

Gratings that are blazed produce a maximum efficiency vis-à-vis reflectivity into<br />

a certain order (o<strong>the</strong>r than <strong>the</strong> zeroth) for <strong>the</strong> light that hit <strong>the</strong> grating. The higher<br />

reflectivity decreases <strong>the</strong> potential for radiati<strong>on</strong> damage to <strong>the</strong> grating.<br />

4.5 M<strong>on</strong>ochromators<br />

Although <strong>the</strong> b<strong>and</strong>width of <strong>the</strong> radiati<strong>on</strong> from free electr<strong>on</strong> laser can be very small,<br />

e.g. at Flash it is about 1%[104], many experiments <strong>–</strong> especially spectroscopic dittos<br />

<strong>–</strong> have a more stringent dem<strong>and</strong> <strong>on</strong> <strong>the</strong> spectral purity of <strong>the</strong> light. This is <strong>on</strong>e reas<strong>on</strong><br />

why m<strong>on</strong>ochromators are present at some free electr<strong>on</strong> laser<strong>–</strong> o<strong>the</strong>r reas<strong>on</strong>s include<br />

<strong>the</strong> increase of stability of <strong>the</strong> central wavelength as discussed below.<br />

Some experimental cevats c<strong>on</strong>cerning free electr<strong>on</strong> laser radiati<strong>on</strong><br />

A Sase pulse from a free electr<strong>on</strong> laser can be c<strong>on</strong>sidered as being built up from a<br />

series of spikes arising from <strong>the</strong> radiating l<strong>on</strong>gitudinal modes that happended to be<br />

amplified as that pulse passed through <strong>the</strong> undulator structure. Each spike in <strong>the</strong><br />

spectrum is transform limited 1 whereas <strong>the</strong> spectral distributi<strong>on</strong> in<strong>the</strong> pulseis chaotic<br />

in <strong>the</strong> sense that each pulse have a unique spectral compositi<strong>on</strong>. Thus a Sase pulse<br />

c<strong>on</strong>sists of many independently radiating modes whose intensity varies from pulse to<br />

pulse <strong>–</strong> hence <strong>the</strong> median wavelength of <strong>the</strong> pulses change, as well as <strong>the</strong> spectral<br />

intensity distributi<strong>on</strong>, <strong>on</strong> a pulse to pulse basis. Additi<strong>on</strong>ally, as seen in Figure 1.10,<br />

<strong>the</strong> radiati<strong>on</strong> emitted in <strong>the</strong> Sase part of <strong>the</strong> spectrum sits <strong>on</strong> top of a large broad<br />

background of sp<strong>on</strong>taneous undulator radiati<strong>on</strong>.<br />

If <strong>the</strong> undulators are not perfectly helical (that is <strong>the</strong> electr<strong>on</strong>s emits n<strong>on</strong>-circularly<br />

polarized radiati<strong>on</strong>) <strong>the</strong> spectrum will c<strong>on</strong>tain harm<strong>on</strong>ics of <strong>the</strong> fundamental wavelength<br />

emitted. In certain instances this can be viewed as fortuitous since <strong>the</strong> harm<strong>on</strong>ics<br />

will have shorter wavelengths (but <strong>the</strong>n <strong>the</strong> intense fundamental becomes a<br />

problem at <strong>the</strong> experiment) or as a serious drawback if it is <strong>the</strong> fundamental wavelength<br />

that is to be used <strong>–</strong> <strong>and</strong> <strong>the</strong> harm<strong>on</strong>ics becomes a problem at <strong>the</strong> experiment<br />

side.<br />

Seeded free electr<strong>on</strong> laser schemes (secti<strong>on</strong> 1.5, see page 15) improves up<strong>on</strong> <strong>the</strong><br />

basic Sase scheme in various ways <strong>–</strong> all striving to improve <strong>the</strong> quality of <strong>the</strong> emitted<br />

radiati<strong>on</strong> by enhancing <strong>the</strong> microbunching of <strong>the</strong> electr<strong>on</strong>s in various ways. To avoid<br />

timing errors between <strong>the</strong> seed pulse(s) <strong>and</strong> <strong>the</strong> electr<strong>on</strong> bunch (timing jitter) <strong>the</strong><br />

seed pulse is often significantly shorter than <strong>the</strong> electr<strong>on</strong> bunch. Hence, <strong>on</strong>ly a part<br />

of <strong>the</strong> electr<strong>on</strong>s experiences <strong>the</strong> additi<strong>on</strong>al density modulati<strong>on</strong> owing to <strong>the</strong> seeding<br />

process, in additi<strong>on</strong> to <strong>the</strong> seeded free electr<strong>on</strong> laser radiati<strong>on</strong> <strong>on</strong>e thus obtains a<br />

Sase background from <strong>the</strong> rest of <strong>the</strong> electr<strong>on</strong> bunch (as well as <strong>the</strong> background from<br />

sp<strong>on</strong>taneous undulator radiati<strong>on</strong> menti<strong>on</strong>ed before). Eventhough <strong>the</strong> Sase part of<br />

1 i.e. <strong>the</strong> spectral width is minimal for a given pulse-length. For a Gaussian pulse this implies<br />

that <strong>the</strong> time-b<strong>and</strong>width product is 0.44.


68 4. X-ray optics<br />

<strong>the</strong> <strong>beam</strong> do not reach saturati<strong>on</strong> it can still carry significant power. The unseeded<br />

part of <strong>the</strong> <strong>beam</strong> can thus disturb downstream experiments both with <strong>the</strong> background<br />

itself <strong>and</strong> that it arrives before, <strong>and</strong> extends after, <strong>the</strong> main pulse.<br />

Benefits <strong>and</strong> drawbacks of from a m<strong>on</strong>ochromator<br />

A m<strong>on</strong>ochromator is a device that filters a phot<strong>on</strong> <strong>beam</strong> with regard to wavelength<br />

<strong>–</strong> acting essentially as a wavelength b<strong>and</strong>-pass filter. C<strong>on</strong>sidering <strong>on</strong>ly <strong>the</strong> wavelength<br />

variati<strong>on</strong> issue first, it is obviously beneficial to filter <strong>the</strong> phot<strong>on</strong> <strong>beam</strong> with a<br />

m<strong>on</strong>ochromator since it resolves many of <strong>the</strong> issues menti<strong>on</strong>ed above. A m<strong>on</strong>ochromator<br />

provides:<br />

• Suppressi<strong>on</strong> of <strong>the</strong> sp<strong>on</strong>taneous emissi<strong>on</strong> background.<br />

• Selecti<strong>on</strong>ofanarrow(er)wavelengthrange <strong>–</strong>essentially translatingcentralwavelength<br />

jitter to intensity jitter. The latter is significantly easier to measure <strong>on</strong><br />

a pulse by pulse basis 2 to be used in <strong>the</strong> analysis of o<strong>the</strong>r experimental data.<br />

• Selecti<strong>on</strong> of a single harm<strong>on</strong>ic. Ei<strong>the</strong>r filtering away <strong>the</strong> fundamental if higher<br />

harm<strong>on</strong>ics <strong>and</strong> thus shorter wavelengths are desired, or suppressing <strong>the</strong> c<strong>on</strong>tributi<strong>on</strong><br />

from higher orders if <strong>the</strong> fundamental wavelength is to be used.<br />

• Selecting <strong>the</strong>seeded part of <strong>the</strong>spectrum <strong>–</strong>suppressing <strong>the</strong>sp<strong>on</strong>taneous undulator<br />

emissi<strong>on</strong> <strong>and</strong> <strong>the</strong> Sase background. The sp<strong>on</strong>taneous <strong>and</strong> Sase background<br />

which lies directly underneath <strong>the</strong> seeded part will not be filtered away <strong>and</strong> thus<br />

c<strong>on</strong>tribute to a intensity jitter of <strong>the</strong> m<strong>on</strong>ochromatized pulse.<br />

From a wavelength stability aspect it is thus very advantageous to insert a m<strong>on</strong>ochromator<br />

before <strong>the</strong>experiments. This is providedthat<strong>the</strong>intensityjitter betweenpulses<br />

can be measured <strong>on</strong> a pulse by pulse basis which can be included in <strong>the</strong> data analysis<br />

of <strong>the</strong> experiments. Ways of measuring <strong>the</strong> intensity are discussed in chapter 7 (see<br />

page 99).<br />

The price to pay for <strong>the</strong> wavelength stability is that <strong>the</strong> transmitted power through<br />

<strong>the</strong> optical system gets smaller (per each new optical element) <strong>and</strong> that <strong>the</strong> pulses<br />

get temporally stretched. The time-stretch can be c<strong>on</strong>trolled or compensated at <strong>the</strong><br />

cost of transmissi<strong>on</strong> as will be seen below.<br />

For high energy phot<strong>on</strong>s (larger than about 2 keV) it is not possible to use gratings<br />

because <strong>the</strong>ir cut-off energy prohibits transmissi<strong>on</strong>. Instead crystal m<strong>on</strong>ochromators<br />

can be employed which is c<strong>on</strong>structed from channel cut crystals (often Si) optimized<br />

to transmit <strong>on</strong>e phot<strong>on</strong> energy optimally. At Lcls <strong>on</strong>e such m<strong>on</strong>ochromator is built<br />

into <strong>the</strong> <strong>beam</strong> <strong>transport</strong> system that is optimized for 8.3 keV[96].<br />

Time-stretching<br />

In <strong>the</strong> soft X-ray ranges <strong>the</strong> incidence angle <strong>on</strong> any optical element needs to be small<br />

(i.e. <strong>the</strong> <strong>beam</strong> impinges at <strong>the</strong> optic at grazing incidence). Thus <strong>the</strong>re will be a time<br />

difference between <strong>the</strong> parts of <strong>the</strong> <strong>beam</strong> that hits <strong>the</strong> surface first <strong>and</strong> <strong>the</strong> parts<br />

2 O<strong>the</strong>rwise <strong>the</strong> full spectrum of <strong>the</strong> pulse (i.e. both wavelength <strong>and</strong> intensity distributi<strong>on</strong>)<br />

needs to be measured <strong>on</strong> a pulse by pulse basis. This is slightly more involved than measuring <strong>the</strong><br />

intensity in this manner, as discussed in chapter 7 (see page 99).


4.6. Beam attenuators 69<br />

that hit at a later time, as illustrated in Figure 4.2. The time-stretching can be<br />

reversed by letting <strong>the</strong> <strong>beam</strong> bounce off a sec<strong>on</strong>d surface <strong>–</strong> at <strong>the</strong> cost of reduced<br />

transmissi<strong>on</strong>, it can also be c<strong>on</strong>trolled as it is proporti<strong>on</strong>al to <strong>the</strong> number of grooves<br />

that are illuminated by <strong>the</strong> <strong>beam</strong>[105]. By illuminating <strong>the</strong> grating parallel to <strong>the</strong><br />

grooves (off-plane mounting) <strong>the</strong> time-stretching can be reduced also[106, 107].<br />

Figure 4.2: Time stretching of a pulse arising from <strong>the</strong> grazing incidence.<br />

α<br />

β<br />

α<br />

β 0 th order<br />

Figure 4.3: A time-stretch compensating double grating setup. The <strong>beam</strong> moves from left to right in a Z-shaped<br />

path <strong>and</strong> <strong>the</strong> outgoing <strong>beam</strong> is parallel to <strong>the</strong> incoming.<br />

4.6 Beam attenuators<br />

Although not a proper optical element per se, gas attenuators are a integral comp<strong>on</strong>ent<br />

of <strong>the</strong> <strong>beam</strong> <strong>transport</strong> system of free electr<strong>on</strong> laser providing a robust mean of<br />

c<strong>on</strong>trolling <strong>the</strong> intensity of <strong>the</strong> radiati<strong>on</strong> at <strong>the</strong> experiments. The intensity can be<br />

c<strong>on</strong>trolled in a c<strong>on</strong>tinuous manner over several orders of magnitude with <strong>the</strong> limit set<br />

essentially by vacuum c<strong>on</strong>siderati<strong>on</strong>s, i.e. how good <strong>the</strong> differential pumping is.<br />

Beam attenuators <strong>and</strong> filters based <strong>on</strong> metal foils <strong>and</strong> metal/ceramic windows free<br />

electr<strong>on</strong> laser <strong>beam</strong> can be used at higher energies provided that <strong>the</strong>y can withst<strong>and</strong><br />

<strong>the</strong> fluence of <strong>the</strong> X.rays.<br />

There is str<strong>on</strong>g indicati<strong>on</strong>s that, at least, gas attenuators do not distort <strong>the</strong> wavefr<strong>on</strong>t<br />

[104] thus c<strong>on</strong>serving <strong>the</strong> coherence properties of <strong>the</strong> x-ray <strong>beam</strong>.<br />

Figure 4.4 describes a generic attenuator, at <strong>the</strong> Lcls <strong>the</strong>re are solid attenuators<br />

included whereas at Flash <strong>and</strong> Scss <strong>on</strong>ly <strong>the</strong> gas attenuator is needed (owing to


70 4. X-ray optics<br />

Intensity detector<br />

Pumping<br />

Solid attenuators<br />

Gas input<br />

Intensity detector<br />

Pumping<br />

Figure 4.4: Schematic of a gas attenuator with intensity m<strong>on</strong>itors before <strong>and</strong> after.<br />

<strong>the</strong>ir wavelength ranges). Intensity m<strong>on</strong>itors are usually put before <strong>and</strong> after <strong>the</strong><br />

attenuators 3 .<br />

The gas attenuator at Flash is 15 m l<strong>on</strong>g <strong>and</strong> s<strong>and</strong>wiched between intensity m<strong>on</strong>itors.<br />

It is operated ei<strong>the</strong>r with nitrogen for <strong>the</strong> wavelength ranges 60-19 nm, <strong>and</strong><br />

for smaller wavelengths xen<strong>on</strong> or krypt<strong>on</strong> can be employed. The differential pumping<br />

system is fast <strong>and</strong> changes of transmissi<strong>on</strong> of up to four orders of magnitude can be<br />

provided by <strong>the</strong> attenuator within minutes[104, 108].<br />

At <strong>the</strong> Scss a smaller gas-cell is used with arg<strong>on</strong> as attenuator gas[109, 110].<br />

Lclsemploysa4.5metersl<strong>on</strong>ggas attenuatorusingnitrogenfor operati<strong>on</strong>between<br />

800 <strong>and</strong> 2000 eV phot<strong>on</strong> energies. The pressure range is up to about 10 mbar <strong>and</strong><br />

thus provide an attenuati<strong>on</strong> span of four orders of magnitude. A set of solid beryllium<br />

attenuators (0.1 to 32 mm thickness) provide <strong>the</strong> attenuati<strong>on</strong> for <strong>the</strong> harder X-rays<br />

up to 8 keV.<br />

3 Intensity diagnostic devices will be discussed fur<strong>the</strong>r in secti<strong>on</strong> 7.2 (see page 100).


4.6. Beam attenuators 71<br />

Summary<br />

• The dem<strong>and</strong>s <strong>on</strong> X-rayoptical elements for free electr<strong>on</strong> lasers<br />

are more stringent than those utilized for e.g. storage ring<br />

lightsources <strong>–</strong> owing to <strong>the</strong>, in most instances, higher peak<br />

power of free electr<strong>on</strong> lasers compared to <strong>the</strong> high average<br />

power exerted from storage ring synchrotr<strong>on</strong>s.<br />

• Damage<strong>on</strong>opticalelementschanges<strong>the</strong>propertiesof<strong>the</strong>radiati<strong>on</strong><br />

at <strong>the</strong> user stati<strong>on</strong>s vis-á-vis spectral c<strong>on</strong>tent, intensity,<br />

coherence <strong>and</strong> temporal structure.<br />

• Optical elements used at free electr<strong>on</strong> lasers generally suffers<br />

damage primarily from ablati<strong>on</strong> caused by <strong>the</strong> high peak brilliance<br />

of <strong>the</strong> source.<br />

• In<strong>the</strong>VUV<strong>and</strong> soft X-rayregi<strong>on</strong>s diffracti<strong>on</strong> gratings areuniversally<br />

adopted as <strong>the</strong> dispersive optic necessary to provide<br />

wavelength selecti<strong>on</strong> in m<strong>on</strong>ochromators.<br />

• Simulati<strong>on</strong> codes: Shadow[111], Xtrace[112], Spectra.


5. Beam-splitting methods<br />

The material presented here in this chapter is partly adapted from ”Survey of<br />

<strong>beam</strong> splitters” by M. A. Bowler, A. J. Glees<strong>on</strong>, D Laundy, <strong>and</strong> M. D. Roper.<br />

Iruvx WP7, 2010 by A. Lindblad.<br />

5.1 Introducti<strong>on</strong><br />

The number of <strong>beam</strong>lines, where experiments can be carried out, at Free Electr<strong>on</strong><br />

Laser (FEL) facilities is much less than <strong>on</strong> synchrotr<strong>on</strong>s, <strong>and</strong> may not be enough to<br />

meet user dem<strong>and</strong>. One possibility of increasing <strong>the</strong> number of users being able to<br />

carry out experiments <strong>on</strong> free electr<strong>on</strong> lasers is to split <strong>the</strong> radiati<strong>on</strong> in a <strong>beam</strong>line<br />

into two or more separate paths allowing more than <strong>on</strong>e experiment to be carried out<br />

simultaneously. This could be achieved by splitting <strong>the</strong> phot<strong>on</strong> <strong>beam</strong>, or if <strong>the</strong> pulse<br />

repetiti<strong>on</strong> rate is slow enough, by kicking <strong>the</strong> electr<strong>on</strong> bunches into different paths.<br />

This chapter is mainly c<strong>on</strong>cerned with <strong>the</strong> former technique, <strong>and</strong> gives a survey of<br />

techniques for splitting a phot<strong>on</strong> <strong>beam</strong>.<br />

The type of <strong>beam</strong> splitter used will depend <strong>on</strong> <strong>the</strong> wavelength range over which it<br />

is required to operate. This chapter c<strong>on</strong>centrates <strong>on</strong> <strong>the</strong> wavelengths from <strong>the</strong> vacuum<br />

ultra-violet (VUV) to <strong>the</strong> soft X-ray (SXR) regi<strong>on</strong>s. However, menti<strong>on</strong> will be made<br />

of techniques employed at o<strong>the</strong>r wavelengths, at harder X-ray energies of relevance for<br />

example to XFEL, <strong>and</strong> at much l<strong>on</strong>ger wavelengths in <strong>the</strong> infrared regime of relevance<br />

to existing free electr<strong>on</strong> lasers <strong>and</strong> possible future far infrared (FIR) sources.<br />

Beam splitters are also usedtopartiti<strong>on</strong> <strong>the</strong>radiati<strong>on</strong> for pump-probeexperiments,<br />

for sending a small fracti<strong>on</strong> of <strong>the</strong> radiati<strong>on</strong> for diagnostic purposes, <strong>and</strong> for input to<br />

interferometers. Beam splitters designed for <strong>the</strong>se <strong>and</strong> o<strong>the</strong>r uses have been included<br />

as <strong>the</strong>y may be able to be adapted to fullfil <strong>the</strong> requirements for allowing multiple<br />

end-stati<strong>on</strong>s to be used simultaneously.<br />

Beam splitters can be divided into two main categories. Firstly, <strong>the</strong>re are those<br />

that divide <strong>the</strong> amplitude of <strong>the</strong> radiati<strong>on</strong>, such as:<br />

• partiallytransmittingmirrors -metallic foils, multilayers, pellicles, <strong>beam</strong>splitter<br />

plates<br />

• <strong>beam</strong> splitter cubes<br />

• thin crystals<br />

73


74 5. Beam-splitting methods<br />

• diffracti<strong>on</strong> gratings<br />

• wire grids<br />

Sec<strong>on</strong>dly, <strong>the</strong>re are those that divide <strong>the</strong> wavefr<strong>on</strong>t, such as<br />

• knife edge mirrors or crystals<br />

• Fresnel bi-mirror<br />

• slotted or perforated mirrors<br />

• structured arrays<br />

In <strong>the</strong> case of a pulsed source, it is also possible to divide <strong>the</strong> <strong>beam</strong> in time by<br />

using moving mirrors (translating, rotating, vibrating) to deflect pulses in different<br />

directi<strong>on</strong>s.<br />

Each type of <strong>beam</strong> splitter is discussed in turn, <strong>and</strong> in secti<strong>on</strong> 5.6 a list is drawn up<br />

of those which may be suitable c<strong>and</strong>idates for splitting a main free electr<strong>on</strong> laser <strong>beam</strong><br />

for multiple experiments for different wavelength regimes. General points for <strong>the</strong><br />

specificati<strong>on</strong> of a <strong>beam</strong> splitter are given in secti<strong>on</strong> 2 before <strong>the</strong> survey of <strong>the</strong> different<br />

types.<br />

5.2 Beam-splitter specificati<strong>on</strong><br />

A sample specificati<strong>on</strong> is given in Table 5.1 for a <strong>beam</strong> splitter <strong>on</strong> an XUV-FEL.<br />

The full specificati<strong>on</strong> of a <strong>beam</strong> splitter will include <strong>the</strong> maximum power loading,<br />

overall efficiency, relative intensity of <strong>the</strong> split <strong>beam</strong>s, minimum angular separati<strong>on</strong>,<br />

polarisati<strong>on</strong> <strong>and</strong> wavefr<strong>on</strong>t preservati<strong>on</strong>. The required power h<strong>and</strong>ling <strong>and</strong> efficiency<br />

to a large extent will determine <strong>the</strong> incident angle <strong>and</strong> <strong>the</strong> material(s) used. The<br />

incidence angle will have a large impact <strong>on</strong> <strong>the</strong> size of <strong>the</strong> optic. Cost <strong>and</strong> ease of<br />

manufacture of <strong>the</strong> comp<strong>on</strong>ent are also important c<strong>on</strong>siderati<strong>on</strong>s.<br />

FEL radiati<strong>on</strong> has a high degree of polarisati<strong>on</strong> <strong>and</strong> coherence. In c<strong>on</strong>sidering<br />

<strong>the</strong> suitability of different types of <strong>beam</strong> splitter, <strong>the</strong> starting assumpti<strong>on</strong> is that <strong>the</strong><br />

quality of <strong>the</strong> incident <strong>beam</strong> is to be preserved in <strong>the</strong> split <strong>beam</strong>s, i.e. polarisati<strong>on</strong>,<br />

wavefr<strong>on</strong>t shape <strong>and</strong> pulse durati<strong>on</strong> are preserved as far as possible. However, it may<br />

be that in order to increase <strong>the</strong> efficiency <strong>the</strong>se assumpti<strong>on</strong>s have to be relaxed, at<br />

least in <strong>on</strong>e if not both of <strong>the</strong> <strong>beam</strong>s. Obviously <strong>the</strong> exact <strong>beam</strong> qualities required<br />

will depend <strong>on</strong> <strong>the</strong> experiment <strong>and</strong> preserving all properties may not be important<br />

in all cases <strong>and</strong> such a relaxati<strong>on</strong> may not matter. Different splitting techniques will<br />

compromise different aspects of <strong>the</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> correct choice will depend <strong>on</strong> <strong>the</strong><br />

experiment; <strong>the</strong>re is unlikely to be a universal <strong>beam</strong> splitter.<br />

The minimum required efficiency will be set by <strong>the</strong> ability of <strong>the</strong> optics to h<strong>and</strong>le<br />

<strong>the</strong> power load or <strong>beam</strong> fluence. The efficiency requirement as determined by <strong>the</strong> flux<br />

needs of <strong>the</strong> experiments falls outside <strong>the</strong> scope of this survey. It is however assumed<br />

that <strong>the</strong> experiments <strong>on</strong> each branch will be equally dem<strong>and</strong>ing of phot<strong>on</strong> flux <strong>and</strong><br />

so <strong>the</strong> <strong>beam</strong> should be split roughly into equal parts. It is however noted that for<br />

some applicati<strong>on</strong>s <strong>the</strong> free electr<strong>on</strong> laser be may be too powerful, <strong>and</strong> it may be that<br />

a <strong>beam</strong> splitter could be used as an attenuator <strong>on</strong> <strong>on</strong>e <strong>beam</strong> path.<br />

A fur<strong>the</strong>r point to note is that he space available for <strong>the</strong> <strong>beam</strong>lines may well also<br />

limit <strong>the</strong> maximum angular separati<strong>on</strong> <strong>and</strong> in general it may not be practical to have<br />

very large angular separati<strong>on</strong>s between <strong>the</strong> <strong>beam</strong>s.


5.2. Beam-splitter specificati<strong>on</strong> 75<br />

Parameter Value Comments<br />

Phot<strong>on</strong> energy 8-100 eV<br />

Average power 10-100 W Depends <strong>on</strong> rep. rate<br />

Beam fluence ≤ 50 mJ/cm 2<br />

Value at 100 eV, normal incidence<br />

Grazing angle < 5 ◦ , preferable ∼ 3 ◦<br />

Determined by reflectivity<br />

requirements <strong>and</strong> ablati<strong>on</strong><br />

threshold<br />

Size 100-200 mm by 50 mm Depends <strong>on</strong> grazing angle<br />

Overall efficeny > 50% This may increase if power<br />

loading becomes problematic<br />

Intensity split Roughly equal in each<br />

branch<br />

Polarisati<strong>on</strong> Ideally small/negligable This will be difficult for<br />

lower energies<br />

Wavefr<strong>on</strong>t distorti<strong>on</strong> Ideally <strong>the</strong> modal structure<br />

of <strong>the</strong> <strong>beam</strong> should be preserved<br />

Minimum angular 5 mrad (0.3<br />

separati<strong>on</strong> of <strong>beam</strong>s<br />

◦ )<br />

Vacuum compatibil- UVH bakable hydrocarb<strong>on</strong> free for operaityti<strong>on</strong><br />

near <strong>the</strong> carb<strong>on</strong> edge<br />

Table 5.1: Sample specificati<strong>on</strong> for a <strong>beam</strong>-splitter for XUV radiati<strong>on</strong>.


76 5. Beam-splitting methods<br />

5.3 Amplitude divisi<strong>on</strong> <strong>beam</strong> splitters<br />

Partially transmitting materials<br />

Partially transmitting mirrors can be made from metallic foils or multilayers. When<br />

<strong>the</strong>se are deposited <strong>on</strong> glass <strong>the</strong>y are termed plate <strong>beam</strong> splitters, or <strong>on</strong> a thin (of<br />

<strong>the</strong> order of µm thickness) stretched polymer membranes, <strong>the</strong>y are termed pellicles.<br />

Thin crystals are also used as <strong>beam</strong> splitters in <strong>the</strong> harder X-ray regime.<br />

Plate <strong>beam</strong> splitters, by <strong>the</strong>ir nature of being deposited <strong>on</strong> glass, operate mainly<br />

<strong>the</strong> visible <strong>and</strong> near infrared (NIR) but <strong>the</strong>ir range can be extended to 200 nm using<br />

UV grade fused silica as <strong>the</strong> substrate. They will normally have <strong>the</strong> <strong>beam</strong> splitter<br />

coating <strong>on</strong> <strong>the</strong> fr<strong>on</strong>t face of <strong>the</strong> plate <strong>and</strong> an anti-reflecti<strong>on</strong> coating <strong>on</strong> <strong>the</strong> back to<br />

prevent ghosting <strong>and</strong> <strong>the</strong> substrate will be slightly wedged to eliminate interference<br />

fringes. They can be ei<strong>the</strong>r polarising or n<strong>on</strong>-polarising. Polarising <strong>beam</strong> splitters<br />

use bi-refringent materials <strong>and</strong>/or are operated near <strong>the</strong> Brewster Angle. Plate <strong>beam</strong><br />

splitters can be made for high power applicati<strong>on</strong>s, but those that are readily available<br />

are mainly single wavelength, single angle devices, made for specific laser lines. As an<br />

example, ESCO[113] make a range of <strong>beam</strong> splitters with a 10% b<strong>and</strong>pass, damage<br />

threshold 300 - 500 W/cm 2 , designed to work at 45 ◦ incidence for a range of laser<br />

lines from <strong>the</strong> UV to IR.<br />

Broadb<strong>and</strong> dielectric plate <strong>beam</strong> splitters are available in <strong>the</strong> visible <strong>and</strong> NIR, e.g.<br />

Newport[114] provide <strong>beam</strong> splitters covering <strong>the</strong> wavelength ranges 480 - 700 nm,<br />

700 - 950 nm, 1290 - 1500 nm. These operate at 45 ◦ angle of incidence <strong>and</strong> all are<br />

slightly polarising. The damage threshold is quoted as typically 500 W/cm 2 CW, 0.5<br />

J/cm 2 with 10 ns pulses.<br />

C<strong>on</strong>venti<strong>on</strong>al plate <strong>beam</strong> splitters may not be applicable for ultra-fast use due to<br />

dispersi<strong>on</strong> in <strong>the</strong> transmitted <strong>beam</strong>s. Newport provide thin (3mm) ”ultra-fast” plate<br />

splitters made from UV quality fused-silica operating over 700 - 950 nm for s-polarised<br />

light <strong>and</strong> 680 - 1060 nm for p-polarised light. Again <strong>the</strong>se operate at 45 ◦ angle of<br />

incidence <strong>and</strong> split <strong>the</strong> radiati<strong>on</strong> nearly 50:50 <strong>and</strong> are usable for pulses in <strong>the</strong> 100’s<br />

of fs regime.<br />

Pellicles<br />

Pellicles operate in <strong>the</strong> wavelength range 400 - 2500 nm. They have <strong>the</strong> advantage<br />

over plates in that <strong>the</strong>re is no ghost image due to <strong>the</strong>ir extreme thinness (except for<br />

extremely short pulses which would <strong>on</strong>ly be a few cycles l<strong>on</strong>g at <strong>the</strong>se wavelengths).<br />

For uncoated pellicles <strong>the</strong> reflectivity is about 10%, transmissi<strong>on</strong> about 90%, but<br />

coatings have been made which increase <strong>the</strong> reflectance up to <strong>the</strong> order of 50 %.<br />

Typical damage thresholds are 2 W/cm 2 CW <strong>and</strong> 1 J/cm 2 with 10 ns pulses for<br />

uncoated membranes, which makes <strong>the</strong>m unsuitable for free electr<strong>on</strong> laser <strong>beam</strong>lines.<br />

The thin membranes are also susceptible to vibrati<strong>on</strong> <strong>and</strong> interference fringes. For<br />

fur<strong>the</strong>r details see for example: [114, 115].<br />

As an extensi<strong>on</strong> to what is normally c<strong>on</strong>sidered to be a pellicle, a <strong>beam</strong> splitter for<br />

an FTIR spectrometer has been made for <strong>the</strong> THz free electr<strong>on</strong> laser at KAERI by<br />

coating a polyester film with silver [116]. At 3 THz, <strong>the</strong> absorpti<strong>on</strong> of <strong>the</strong> polyester<br />

film is about 2%, <strong>and</strong><strong>the</strong> balance between reflecti<strong>on</strong> <strong>and</strong>transmissi<strong>on</strong> was obtained by<br />

coating <strong>the</strong> film with several tens of nanometers of silver to create <strong>the</strong> <strong>beam</strong> splitter.


5.3. Amplitude divisi<strong>on</strong> <strong>beam</strong> splitters 77<br />

In an analogous manner, thin foils of a low-Z material might be used at soft x-ray<br />

<strong>and</strong> shorter wavelengths.<br />

For example, Lcls propose to use 30 µm thick polished beryllium foils at a grazing<br />

angle (∼ 1 ◦ for energies around 8 keV) above <strong>the</strong> critical angle to reflect a small<br />

( 0.01%) part of <strong>the</strong> <strong>beam</strong> to a diagnostic instrument, while transmitting 99% of <strong>the</strong><br />

radiati<strong>on</strong>[117]. Because low-Z materials give a sharp reflecti<strong>on</strong> cut-off, achieving a<br />

more balanced split will require <strong>the</strong> grazing angle to be adjusted to suit <strong>the</strong> phot<strong>on</strong><br />

energy <strong>and</strong> hence <strong>the</strong> angle of <strong>beam</strong> separati<strong>on</strong> would be a functi<strong>on</strong> of phot<strong>on</strong> energy.<br />

At lower phot<strong>on</strong> energies high transmissi<strong>on</strong> will not be possible e.g. it is effectively<br />

zero below 1 keV for a 30 µm Be foil.<br />

Extensi<strong>on</strong> of this technique into <strong>the</strong> XUV would require excepti<strong>on</strong>ally thin foils<br />

where <strong>the</strong> challenge of manufacturing <strong>and</strong> supporting a surface of sufficient quality<br />

for good reflecti<strong>on</strong> would be very severe indeed. Where absorpti<strong>on</strong> is high, <strong>the</strong>re is<br />

also <strong>the</strong> c<strong>on</strong>cern of radiati<strong>on</strong> damage.<br />

Multi-layer <strong>beam</strong> splitters<br />

Multi-layer <strong>beam</strong>splitters canbemade for <strong>the</strong>XUV<strong>and</strong>shorter wavelengths, however<br />

in general <strong>the</strong>y <strong>on</strong>ly operate for a given wavelength <strong>and</strong> incident angle. Multi-layers<br />

are usually formed <strong>on</strong>SiC, SiNor Si3N4 membranes butobviously for <strong>the</strong>transmissi<strong>on</strong><br />

multi-layers required for <strong>beam</strong> splitters, absorpti<strong>on</strong> in <strong>the</strong> membrane is a problem at<br />

XUV wavelengths. Much work has been d<strong>on</strong>e in <strong>the</strong> 13<strong>–</strong>16 nm wavelength range<br />

for applicati<strong>on</strong>s in lithography using multi-layers <strong>on</strong> Si3N4 membranes, however in<br />

general <strong>the</strong> overall efficiencies are not very high. Some examples found <strong>on</strong> <strong>the</strong> web<br />

with overall efficiency greater than 20% include:<br />

• Near-normal incidence (10 ◦ ) Mo/Si multilayer - reflectivity 28%, transmissi<strong>on</strong><br />

32% [118]<br />

• Ru/SiNbilayeroperatingat17.5 ◦ incidence -reflectivity<strong>and</strong>transmissi<strong>on</strong> about<br />

11% [118]<br />

• Mo/Si multilayer <strong>on</strong> silic<strong>on</strong> nitride at 7.2 ◦ incidence - reflectivity 20%, transmissi<strong>on</strong><br />

22% [119]<br />

Some work has been d<strong>on</strong>e <strong>on</strong> manufacturing free-st<strong>and</strong>ing multi-layers. For example,<br />

Haga, et al. [120] report <strong>the</strong> manufacture of a 10 by 10 mm 2 Mo/Si multilayer<br />

where <strong>the</strong> silic<strong>on</strong> nitride membrane is etched away. The outer Mo layers were replaced<br />

by Ru which was much more resistant to <strong>the</strong> etching process. They also investigated<br />

<strong>the</strong> effect of <strong>the</strong> substrate smoothness <strong>on</strong> <strong>the</strong> efficiency of <strong>the</strong> multi-layer. Measurements<br />

using synchrotr<strong>on</strong> radiati<strong>on</strong> revealed that <strong>the</strong> multilayer worked as a <strong>on</strong>e-to-<strong>on</strong>e<br />

<strong>beam</strong> splitter whose reflectivity<strong>and</strong> transmittance for s-polarised radiati<strong>on</strong> at 13.4 nm<br />

are both 27% at an angle of incidence of 45 ◦ , i.e. <strong>the</strong> efficiency is similar to supported<br />

multilayers. More recent work <strong>on</strong> free-st<strong>and</strong>ing multilayers has not been identified.<br />

By varying <strong>the</strong> thickness of <strong>the</strong> multi-layers with depth, <strong>the</strong> wavelength <strong>and</strong>/or<br />

angle range can be increased, e.g. for reflecting multi-layers a reflectivity of larger<br />

than 40% is expected over <strong>the</strong> energy range 4 - 9 keV using 20 Mo/Si bi-layers with<br />

five different spacings <strong>on</strong> an SiO2 substrate[94]. If <strong>the</strong>se multi-layers could be made<br />

<strong>on</strong> transparent substrates, <strong>the</strong>ir use as <strong>beam</strong>-splitters could be investigated. Note<br />

again <strong>the</strong>re may be a problem of dispersi<strong>on</strong> which may preclude <strong>the</strong>ir use for very<br />

high time resoluti<strong>on</strong> applicati<strong>on</strong>[121].


78 5. Beam-splitting methods<br />

Radiati<strong>on</strong> damage is an area for c<strong>on</strong>cern for <strong>the</strong> use of multilayers in free electr<strong>on</strong><br />

laser <strong>beam</strong>lines. Recently Bajt, et al. [122] reported that a normal incidence parabolic<br />

Mo/Si multilayer mirror with a reflectivity of about 60%, used to focus 13.5 nm<br />

radiati<strong>on</strong> at Flash did not show signs of damage. This efficiency is similar to <strong>the</strong><br />

overall efficiency of some of <strong>the</strong> <strong>beam</strong> splitters menti<strong>on</strong>ed above. Obviously more<br />

work needs to be d<strong>on</strong>e in this area. However, it is unlikely that a multilayer <strong>beam</strong><br />

splitter would be used for <strong>the</strong> current applicati<strong>on</strong> due to <strong>the</strong> wavelength specificity as<br />

well as possible power h<strong>and</strong>ling issues.<br />

Beam splitter cubes<br />

Beam splitter cubes <strong>on</strong>sist of 2 right-angled prisms, made of fused silica for UV<br />

applicati<strong>on</strong>s, normally cemented toge<strong>the</strong>r with an epoxy. Broadb<strong>and</strong> cubes, with<br />

b<strong>and</strong>widths of 300 - 400 nm, are available over <strong>the</strong> wavelength range 400 - 1600 nm,<br />

<strong>and</strong> may alter <strong>the</strong> polarizati<strong>on</strong> of <strong>the</strong> light by about 10%. There are n<strong>on</strong>polarizing<br />

cubes designed for specific laser lines in <strong>the</strong> UV, for moderate power laser operati<strong>on</strong>.<br />

The epoxy absorbs radiati<strong>on</strong> <strong>and</strong> hence cannot be used in high power applicati<strong>on</strong>s.<br />

However, cubes which do not use adhesives to cement <strong>the</strong> prisms toge<strong>the</strong>r have been<br />

recently designed for high power laser applicati<strong>on</strong>s, e.g. Showa Optr<strong>on</strong>ics[123] market<br />

a cube for <strong>the</strong> UV over <strong>the</strong> range 193nm - 355 nm with a damage threshold at 248<br />

nm of 1 J/cm 2 <strong>and</strong> Precisi<strong>on</strong> Phot<strong>on</strong>ics Corporati<strong>on</strong>[124] make cubes for high power<br />

applicati<strong>on</strong>s (damage threshold of 10 J/cm 2 at 1100 nm) in <strong>the</strong> visible <strong>and</strong> IR.<br />

Crystal diffracti<strong>on</strong> <strong>beam</strong> splitters<br />

Perfect crystals are comm<strong>on</strong>ly used as m<strong>on</strong>ochromating elements for hard X-rays<br />

(working at X-ray energies above about 2 keV). An X-ray entering a perfect crystal<br />

scatters from <strong>the</strong> regularly space atomic planes - a process known as dynamical<br />

diffracti<strong>on</strong>[125]. This leads to a very sharp reflectivity curve with a b<strong>and</strong>pass (∆E/E)<br />

for a given angle of incidence (Bragg angle) that is typically less than 10 −4 . The b<strong>and</strong>pass<br />

depends <strong>on</strong> <strong>the</strong> crystal material (diam<strong>on</strong>d is narrower than silic<strong>on</strong> which in turn<br />

is narrower than germanium), <strong>on</strong> <strong>the</strong> crystal planes used (planes with lower order<br />

indices generally give narrower reflecti<strong>on</strong>s) <strong>and</strong> <strong>on</strong> whe<strong>the</strong>r Bragg or Laue geometry<br />

is employed. In Bragg reflecting geometry, <strong>the</strong> X-rays enter <strong>and</strong> exit from <strong>the</strong> same<br />

surface of <strong>the</strong> crystal while in Laue geometry, <strong>the</strong> X-rays enter <strong>and</strong> exit from different<br />

surfaces. In Bragg geometry, <strong>the</strong> X-rays penetrate a short distance into <strong>the</strong> crystal<br />

<strong>and</strong> <strong>the</strong> reflectivity curve is given by a Darwin curve, which approximates to <strong>the</strong><br />

ideal top hat shape. A general c<strong>on</strong>sequence of <strong>the</strong> narrow b<strong>and</strong>pass associated with<br />

dynamical diffracti<strong>on</strong> is that <strong>the</strong> temporal profile of a short X-ray pulse is affected.<br />

When <strong>the</strong> radiati<strong>on</strong> source b<strong>and</strong>pass is larger than <strong>the</strong> crystal reflecti<strong>on</strong> b<strong>and</strong>pass,<br />

as is <strong>the</strong> case with X-ray FEL radiati<strong>on</strong>, a crystal m<strong>on</strong>ochromator can be used as a<br />

<strong>beam</strong> splitter. This method uses a crystal to deflect a small energy b<strong>and</strong> from <strong>the</strong><br />

broader b<strong>and</strong> source to <strong>on</strong>e experiment while <strong>the</strong> remaining transmitted radiati<strong>on</strong><br />

is passed <strong>on</strong>to a sec<strong>on</strong>d experiment. Thin crystals are normally used to minimise<br />

absorpti<strong>on</strong>. Crystal reflecti<strong>on</strong>s are sensitive to heating from <strong>the</strong> radiati<strong>on</strong> which can<br />

result in changes in lattice parameter <strong>and</strong> lattice orientati<strong>on</strong> which may degrade <strong>the</strong><br />

deflected <strong>beam</strong>. The ideal <strong>beam</strong> splitter crystal should have low X-ray absorpti<strong>on</strong> to<br />

minimise losses to <strong>the</strong> X-ray <strong>beam</strong> <strong>and</strong> to reduce X-ray <strong>beam</strong> power absorbed inside<br />

<strong>the</strong> crystal.


5.3. Amplitude divisi<strong>on</strong> <strong>beam</strong> splitters 79<br />

This method of splitting <strong>beam</strong>s for multiple experiments is used <strong>on</strong> some undulator<br />

<strong>and</strong> high-energy wiggler X-ray <strong>beam</strong>s at synchrotr<strong>on</strong> radiati<strong>on</strong> sources. In <strong>the</strong> lower<br />

X-ray energy range, diam<strong>on</strong>d is often used as <strong>the</strong> <strong>beam</strong> splitter as it has low Xray<br />

absorpti<strong>on</strong> <strong>and</strong> also good <strong>the</strong>rmal c<strong>on</strong>ductivity that allows <strong>the</strong> heat to dissipate<br />

efficiently, reducing <strong>the</strong>rmal distorti<strong>on</strong> of <strong>the</strong> crystal. The difficulty of obtaining<br />

diam<strong>on</strong>d crystals with high enough perfecti<strong>on</strong> for use as m<strong>on</strong>ochromators is a problem<br />

that has been recognised[126]. A sec<strong>on</strong>d m<strong>on</strong>ochromator crystal may be used to<br />

deflect <strong>the</strong> <strong>beam</strong> back into <strong>the</strong> incident <strong>beam</strong> directi<strong>on</strong> <strong>and</strong> if <strong>the</strong> first reflecti<strong>on</strong> is<br />

a diam<strong>on</strong>d (111) reflecti<strong>on</strong> it is possible to use a germanium (220) reflecti<strong>on</strong> as <strong>the</strong><br />

sec<strong>on</strong>d reflecti<strong>on</strong> as <strong>the</strong> d-spacings of <strong>the</strong> two reflecti<strong>on</strong>s are very similar. Using a<br />

germanium reflecti<strong>on</strong> for <strong>the</strong> sec<strong>on</strong>d crystal has <strong>the</strong> advantage that <strong>the</strong> reflecti<strong>on</strong><br />

has a broad width making alignment easier <strong>and</strong> reducing loss of X-ray flux caused<br />

by strain within <strong>the</strong> diam<strong>on</strong>d crystal. The additi<strong>on</strong> of a sec<strong>on</strong>d crystal may also<br />

allow <strong>the</strong> X-ray energy to be changed while maintaining almost c<strong>on</strong>stant X-ray <strong>beam</strong><br />

directi<strong>on</strong>. This method of X-ray <strong>beam</strong> splitting is used at <strong>the</strong> ESRF <strong>on</strong> <strong>the</strong> Troika<br />

<strong>beam</strong>line ID10, which has three undulator segments <strong>and</strong> has available a selecti<strong>on</strong> of<br />

m<strong>on</strong>ochromator crystals - diam<strong>on</strong>d (111), diam<strong>on</strong>d (220), <strong>and</strong> silic<strong>on</strong> (111) - allowing<br />

a variati<strong>on</strong> in b<strong>and</strong>width <strong>on</strong> two side stati<strong>on</strong>s. Ano<strong>the</strong>r example at <strong>the</strong> ESRF is <strong>the</strong><br />

macromolecular crystallography <strong>beam</strong>line ID14 which has three diam<strong>on</strong>d (111) <strong>and</strong><br />

germanium (220) m<strong>on</strong>ochromator pairs, in line, to give three 13.3 keV fixed energy<br />

stati<strong>on</strong>s <strong>and</strong> a single variable energy stati<strong>on</strong> using <strong>the</strong> straight through transmitted<br />

<strong>beam</strong>.<br />

In <strong>the</strong> higher X-ray energy range, X-ray absorpti<strong>on</strong> is lower <strong>and</strong> silic<strong>on</strong> or germanium<br />

may be used as <strong>the</strong> <strong>beam</strong> splitting crystal. This allows <strong>the</strong> b<strong>and</strong>pass of <strong>the</strong><br />

reflecti<strong>on</strong> to be varied by applying a bend to <strong>the</strong> crystal. An example of this applicati<strong>on</strong><br />

at <strong>the</strong> ESRF is <strong>the</strong> high-energy <strong>beam</strong>line ID15 which can use two bent crystals<br />

to deflect <strong>beam</strong> from <strong>the</strong> wiggler source to two side stati<strong>on</strong>s at X-ray energies from<br />

30 keV up to 300 keV.<br />

Where <strong>beam</strong> splitting is required for X-ray interferometry, perfect crystals are<br />

used. Laue interferometers[127] use <strong>the</strong> forward transmitted <strong>and</strong> reflected <strong>beam</strong>s at<br />

<strong>the</strong>crystalexitsurface tosplit<strong>the</strong><strong>beam</strong>. Interferometric <strong>beam</strong>splitters usingmultiple<br />

reflecti<strong>on</strong>s[128] or <strong>the</strong> <strong>beam</strong>s reflected from <strong>and</strong> transmitted through a crystal whose<br />

thickness is less than <strong>the</strong> extincti<strong>on</strong> depth can also be used. These techniques might<br />

be useful where multiple low-b<strong>and</strong>pass (∆E/E ∼ 10 −5 ) <strong>beam</strong>s are required, though<br />

<strong>the</strong>re could be <strong>the</strong>rmal problems since <strong>the</strong> radiati<strong>on</strong> out of this b<strong>and</strong>pass will be<br />

absorbed in <strong>the</strong> crystal.<br />

A similar applicati<strong>on</strong> to that of X-ray interferometry is <strong>the</strong> use of crystal <strong>beam</strong><br />

splitters in optical delay lines. A splitter/delay line at 8 keV has been designed<br />

for <strong>the</strong> Lcls using Si (400) crystals[129]. A 10 µm thick Si (400) crystal has 80%<br />

reflectivity in a b<strong>and</strong>pass of ∆E/E ∼ 1.4·10 −5 with a transmissi<strong>on</strong> outside this regi<strong>on</strong><br />

of 75%. The reflected <strong>and</strong> transmitted <strong>beam</strong>s are sent via different sets of crystals<br />

to be ultimately recombined <strong>on</strong> a comm<strong>on</strong> path but with a variable delay created<br />

by translating <strong>on</strong>e set of crystals. Two pulses are created by tuning <strong>the</strong> crystal sets<br />

to wavelengths that differ by more than <strong>the</strong> b<strong>and</strong>pass of <strong>the</strong> crystal reflecti<strong>on</strong> but<br />

with both c<strong>on</strong>tained within <strong>the</strong> b<strong>and</strong>width of <strong>the</strong> free electr<strong>on</strong> laser pulse. A similar<br />

device working at 8.39 keV with Si (511) crystals <strong>and</strong> 12.4 keV with Si (553) has been<br />

c<strong>on</strong>structed at HASYLAB<strong>and</strong> tested <strong>on</strong> <strong>beam</strong>lines at DORIS-III, PETRA-II <strong>and</strong> <strong>the</strong><br />

ESRF[130, 131].


80 5. Beam-splitting methods<br />

Gratings<br />

Gratings provide a very simple way of splitting a <strong>beam</strong>, though <strong>the</strong>re are a number<br />

of practical limitati<strong>on</strong>s <strong>and</strong> c<strong>on</strong>sequences. A grating divides an incoming <strong>beam</strong> of a<br />

given wavelength into a series of orders spaced around <strong>the</strong> ”zeroth” order <strong>beam</strong>, which<br />

has a diffracti<strong>on</strong> angle equal to <strong>the</strong> incidence angle. Inside orders have a diffracti<strong>on</strong><br />

angle (measured from <strong>the</strong> grating normal) smaller in magnitude than <strong>the</strong> incidence<br />

angle <strong>and</strong> outside orders have a diffracti<strong>on</strong> angle that is larger in magnitude than <strong>the</strong><br />

incidence angle. By c<strong>on</strong>venti<strong>on</strong>, diffracti<strong>on</strong> angles that are <strong>on</strong> <strong>the</strong> opposite side of <strong>the</strong><br />

normal are negative.<br />

For a <strong>beam</strong> splitter, <strong>on</strong>e could choose to use ei<strong>the</strong>r <strong>on</strong>e of <strong>the</strong> first order <strong>and</strong> <strong>the</strong><br />

zeroth order <strong>beam</strong>s, or <strong>the</strong> first inside <strong>and</strong> <strong>the</strong> first outside order <strong>beam</strong>s. Higher<br />

diffracti<strong>on</strong> orders are unlikely to be used as <strong>the</strong>y have lower efficiency.<br />

A possible advantage of using <strong>the</strong> first inside <strong>and</strong> outside orders is that <strong>the</strong> grating<br />

acts as a m<strong>on</strong>ochromator for both. C<strong>on</strong>versely, <strong>the</strong> zeroth order c<strong>on</strong>tains all <strong>the</strong><br />

spectral c<strong>on</strong>tentof<strong>the</strong>incident<strong>beam</strong>upto<strong>the</strong>cut-offdeterminedby<strong>the</strong>reflectivityof<br />

<strong>the</strong> grating coating <strong>and</strong> so a sec<strong>on</strong>d grating would be needed to give a m<strong>on</strong>ochromatic<br />

<strong>beam</strong>. However, this would allow independent choice of wavelengths for <strong>the</strong> two<br />

<strong>beam</strong>s. The relative intensity of <strong>the</strong> first diffracted to <strong>the</strong> zeroth order depends <strong>on</strong><br />

<strong>the</strong> shape of <strong>the</strong> diffracti<strong>on</strong> grating grooves. A careful choice of <strong>the</strong> groove profile<br />

based <strong>on</strong> modelling <strong>the</strong> efficiency with a code such as Gradif 1 can give a certain<br />

degree of tuning of <strong>the</strong> relative intensity, but <strong>on</strong>e cannot expect perfect c<strong>on</strong>trol if <strong>the</strong><br />

grating is to operate over a wide range of wavelengths.<br />

One should also remember that <strong>the</strong> required zero order efficiency could be, <strong>and</strong><br />

probably will be, at a wavelength that is unrelated to <strong>the</strong> first order wavelength.<br />

Therefore, a comprehensive set of calculati<strong>on</strong>s would have to be performed to give<br />

not <strong>on</strong>ly <strong>the</strong> first order efficiency as a functi<strong>on</strong> of wavelength, but also <strong>the</strong> zeroth<br />

order efficiency as a functi<strong>on</strong> of both wavelength <strong>and</strong> incidence angle. There is thus<br />

no uniquevalue to <strong>the</strong> efficiency at a particular wavelength into <strong>the</strong> zeroth order <strong>beam</strong><br />

as it depends <strong>on</strong> wavelength of <strong>the</strong> first order <strong>beam</strong> <strong>and</strong> hence angle of <strong>the</strong> grating.<br />

Anyrequirement totune <strong>the</strong>grating leads to <strong>the</strong> most significant practical problem<br />

with using a grating as a <strong>beam</strong> splitter. As <strong>the</strong> grating is rotated to select a different<br />

wavelength, <strong>the</strong> angle between <strong>the</strong> first <strong>and</strong> zeroth orders will change. Similarly, <strong>the</strong><br />

angle between <strong>the</strong> first inside <strong>and</strong> first outside orders that have <strong>the</strong> same wavelength<br />

will also change. (C<strong>on</strong>versely, <strong>the</strong> wavelength into a fixed outside order directi<strong>on</strong> will<br />

change in <strong>the</strong> opposite directi<strong>on</strong> <strong>and</strong> at a different rate to <strong>the</strong> wavelength in a fixed<br />

inside order directi<strong>on</strong>). This is very inc<strong>on</strong>venient for a <strong>beam</strong> splitter that is to feed<br />

fixed end stati<strong>on</strong>s from a fixed source. One soluti<strong>on</strong> is to place a plane mirror, which<br />

can both translate <strong>and</strong> rotate, after <strong>the</strong> grating. This mirror intercepts <strong>the</strong> zeroth<br />

order <strong>beam</strong> <strong>and</strong> steers it always to a fixed output path. The disadvantages of this are<br />

that <strong>the</strong> path length will change (so changing <strong>the</strong> overall <strong>beam</strong> timing) <strong>and</strong> that <strong>the</strong><br />

mechanism is complicated <strong>and</strong> so pr<strong>on</strong>e to introducing small errors giving temporal<br />

<strong>and</strong> spatial jitter to <strong>the</strong> <strong>beam</strong>.<br />

An alternative approach is to use <strong>the</strong> SX700 type of variable included angle grating<br />

mount. In this, a plane mirror rotates about an axis not in <strong>the</strong> mirror surface <strong>and</strong><br />

allows <strong>the</strong> included angle at <strong>the</strong> grating to be varied by just a single rotati<strong>on</strong>. The<br />

1 Gradif is an update of <strong>the</strong> LUMNAB code of M. Nevière, et al. [132]


5.3. Amplitude divisi<strong>on</strong> <strong>beam</strong> splitters 81<br />

mechanically simpler mount will work with less mechanical error, but <strong>the</strong>re is still a<br />

path length change, <strong>and</strong> so <strong>the</strong> overall <strong>beam</strong> timing is a functi<strong>on</strong> of wavelength.<br />

Providing <strong>the</strong> grating is working in collimated light, <strong>the</strong>re can a free choice of<br />

included angle without changing <strong>the</strong> focusing properties of <strong>the</strong> m<strong>on</strong>ochromator. This<br />

free choice of included angle can be used to ei<strong>the</strong>r keep a fixed angle between <strong>the</strong><br />

first diffracted order <strong>and</strong> <strong>the</strong> zeroth order, or to keep <strong>the</strong> first inside <strong>and</strong> first outside<br />

orders at a <strong>the</strong> same wavelength with a fixed angle between <strong>the</strong>m.<br />

In <strong>the</strong> case where <strong>the</strong> angle between <strong>the</strong> first <strong>and</strong> zeroth order is to be fixed, <strong>on</strong>e<br />

also finds that this is <strong>the</strong> same c<strong>on</strong>straint for scanning a blazed grating ’<strong>on</strong>-blaze’ <strong>and</strong><br />

thus has a potential advantage in optimising <strong>the</strong> first order efficiency. This would<br />

naturally be at <strong>the</strong> expense of <strong>the</strong> zeroth order efficiency at that wavelength, but this<br />

will not be a problem if <strong>the</strong> two branches are to operate at different wavelengths.<br />

If <strong>the</strong> angle between <strong>the</strong> first inside <strong>and</strong> zeroth order <strong>beam</strong> is ψ, <strong>the</strong>n <strong>the</strong> grating<br />

equati<strong>on</strong> under <strong>the</strong> c<strong>on</strong>straint that ψ is fixed becomes<br />

Nmλ = 2sin<br />

� ψ<br />

2<br />

�<br />

cos<br />

� ψ<br />

2 −β<br />

�<br />

(5.1)<br />

where β is <strong>the</strong> diffracti<strong>on</strong> angle. For a blazed grating operating <strong>on</strong>-blaze, <strong>the</strong> blaze<br />

angle should be ψ/2.<br />

A similar expressi<strong>on</strong> is obtained if <strong>the</strong> angle between <strong>the</strong> first inside <strong>and</strong> outside<br />

orders is to be kept c<strong>on</strong>stant <strong>and</strong> both orders are to pass <strong>the</strong> same wavelength. If<br />

β + is <strong>the</strong> diffracti<strong>on</strong> angle of <strong>the</strong> first inside order <strong>and</strong> β − <strong>the</strong> diffracti<strong>on</strong> angle of <strong>the</strong><br />

first outside order, <strong>the</strong>n <strong>the</strong> angle between <strong>the</strong>m is<br />

ξ = β + −β −<br />

<strong>and</strong> <strong>the</strong> grating equati<strong>on</strong> for c<strong>on</strong>stant ξ becomes<br />

Nλ = 2sin<br />

� ξ<br />

2<br />

�<br />

cos<br />

� ξ<br />

2 −β+<br />

�<br />

(5.2)<br />

(5.3)<br />

which gives <strong>the</strong> diffracti<strong>on</strong> angle of <strong>the</strong> inside order for agiven wavelength, from which<br />

<strong>the</strong> incidence angle <strong>and</strong> diffracti<strong>on</strong> angle of <strong>the</strong> outside order can be calculated.<br />

In both <strong>the</strong>se modes of operati<strong>on</strong>, <strong>the</strong> tuning range will be limited if <strong>the</strong> grazing<br />

angles at <strong>the</strong> grating are not to become large (<strong>and</strong> hence <strong>the</strong> efficiencies low). In<br />

general, <strong>the</strong> angles between <strong>the</strong> orders (ψ or ξ) will have to be


82 5. Beam-splitting methods<br />

for <strong>the</strong> sec<strong>on</strong>d grating with c<strong>on</strong>sequent impact <strong>on</strong> <strong>the</strong> cost <strong>and</strong> fur<strong>the</strong>r path length<br />

variati<strong>on</strong>s. The system with a translating <strong>and</strong> rotating plane mirror that intercepts<br />

with just <strong>the</strong> zeroth order <strong>beam</strong> after <strong>the</strong> grating would be simpler in this respect as<br />

both <strong>the</strong> grating mounts could simply be fixed included angle type.<br />

Alternatively, c<strong>on</strong>ical diffracti<strong>on</strong> mounting can be used to reduce <strong>the</strong> pulse stretch<br />

(since <strong>the</strong> effective number of illuminated groves is reduced when compared with <strong>the</strong><br />

classical mount). The fur<strong>the</strong>r advantage of <strong>the</strong> c<strong>on</strong>ical mount is that <strong>the</strong> diffracti<strong>on</strong><br />

efficiency should be much higher. However, in <strong>the</strong> c<strong>on</strong>text of a <strong>beam</strong> splitter, <strong>the</strong><br />

c<strong>on</strong>ical mounting leads to difficulties. It is normal to operate <strong>the</strong> grating in fixed<br />

altitude mode as this allows <strong>the</strong> grating to be tuned with just a simple rotati<strong>on</strong><br />

(about an axis parallel to <strong>the</strong> grooves). Unfortunately, in this mode, <strong>the</strong> zeroth order<br />

(for example) changes directi<strong>on</strong> in both <strong>the</strong> horiz<strong>on</strong>tal <strong>and</strong> vertical planes. Capturing<br />

<strong>and</strong> steering <strong>the</strong> zeroth order <strong>beam</strong> into a fixed directi<strong>on</strong> will be very difficult. Fixed<br />

azimuth mode would make capturing <strong>the</strong> zeroth order <strong>beam</strong> easier, but scanning <strong>the</strong><br />

grating requires translating <strong>and</strong> rotating mirrors.<br />

A scheme for using diffracti<strong>on</strong> gratings as a <strong>beam</strong> splitter has been proposed for<br />

FERMI@Elettra for wavelengths around 40 nm. Light from <strong>the</strong> 0 <strong>and</strong> +1 orders<br />

is used. For <strong>the</strong> 1st order radiati<strong>on</strong>, a sec<strong>on</strong>d grating followed by a translatable<br />

plane mirror allows <strong>the</strong> pulse stretch to be compensated <strong>and</strong> gives a fixed positi<strong>on</strong><br />

for <strong>the</strong> output <strong>beam</strong>[133]. An efficiency of about 30% has been estimated for both<br />

<strong>the</strong> zero <strong>and</strong> first orders using at 100 lines/mm grating, 84 ◦ incidence angle, over <strong>the</strong><br />

wavelength range of 34 - 46 nm. This design requires <strong>the</strong> manufacture of a l<strong>on</strong>g (60<br />

mm) grating to a very high specificati<strong>on</strong>, as well as a translating mirror hence it is<br />

technically challenging.<br />

The above applicati<strong>on</strong>s are all for reflecti<strong>on</strong> gratings. Transmissi<strong>on</strong> gratings can<br />

also be used to split <strong>the</strong> <strong>beam</strong> with <strong>the</strong> advantage that, since <strong>the</strong> gratings are used<br />

at (or near) normal incidence, <strong>the</strong> geometry of <strong>the</strong> split <strong>beam</strong>s is much simpler.<br />

Manufacturing transmissi<strong>on</strong> gratings that can work in <strong>the</strong> XUV to x-ray regi<strong>on</strong> is of<br />

course a major technical challenge. Achieving adequate dispersi<strong>on</strong> (<strong>and</strong> hence <strong>beam</strong><br />

separati<strong>on</strong>) at short wavelengths requires <strong>the</strong> grating period to very small. The aspect<br />

ratio (depth/width) of <strong>the</strong> grating bars is inevitably quite high if <strong>the</strong> gratings are to<br />

be robust <strong>and</strong> <strong>the</strong>y are <strong>the</strong>refore not suitable for str<strong>on</strong>gly divergent light. The risk of<br />

ablati<strong>on</strong> of <strong>the</strong> grating in a free electr<strong>on</strong> laser pulse is also a major c<strong>on</strong>cern.<br />

In <strong>the</strong> XUV, a transmissi<strong>on</strong> grating will almost certainly behave as an amplitude<br />

grating, since it will be difficult to make <strong>the</strong> grating bars anything o<strong>the</strong>r than 100%<br />

absorbing. Str<strong>on</strong>g absorpti<strong>on</strong> means <strong>the</strong> grating must also be made unsupported,<br />

which is very difficult. Assuming a perfect amplitude grating can be made, <strong>the</strong> maximum<br />

first order efficiency occurs when <strong>the</strong> groove width is half <strong>the</strong> grating period<br />

<strong>and</strong> is just 10% <strong>and</strong> <strong>the</strong> zeroth order efficiency is 25%.<br />

C<strong>on</strong>versely, for hard x-rays, it is difficult to make <strong>the</strong> aspect ratio of <strong>the</strong> grating<br />

bars high enough to be perfectly absorbing <strong>and</strong> <strong>the</strong> grating will functi<strong>on</strong> as a phase<br />

grating. The efficiency depends <strong>on</strong> <strong>the</strong> grating material <strong>and</strong> bar profile, <strong>and</strong> this offers<br />

scope for tuning <strong>the</strong> grating performance.<br />

Weitkamp, et al. [134] used a R<strong>on</strong>chi phase grating to make an x-ray shearing<br />

interferometer. The grating is made by electr<strong>on</strong>-<strong>beam</strong> lithography from silic<strong>on</strong> <strong>and</strong><br />

has a groove period of 2 µm <strong>and</strong> depth of 9 µm. They operated <strong>the</strong> grating at 12.4<br />

keV (1 ˚A), but were <strong>on</strong>ly producing a small shear in <strong>the</strong> <strong>beam</strong>. The phase shift of <strong>the</strong><br />

grating was tuned by tilting <strong>the</strong> grating about an axis perpendicular to <strong>the</strong> grooves<br />

<strong>and</strong> <strong>the</strong> <strong>beam</strong> axis so as to vary <strong>the</strong> effective thickness of <strong>the</strong> grating. This allowed


5.4. Wavefr<strong>on</strong>t divisi<strong>on</strong> <strong>beam</strong>splitters 83<br />

<strong>the</strong> phase shift to be increased to π <strong>and</strong> hence <strong>the</strong> zeroth diffracti<strong>on</strong> order eliminated<br />

<strong>and</strong> first order efficiencies maximised. Naturally, this technique can <strong>on</strong>ly be employed<br />

when <strong>the</strong> grating bars have sufficient transmissi<strong>on</strong>, <strong>and</strong> is thus <strong>on</strong>ly applicable for<br />

harder x-rays due to <strong>the</strong> difficulty in manufacturing very thin gratings with such a<br />

fine structure.<br />

Gratings can also be used to send a small part of <strong>the</strong> <strong>beam</strong> (in first order) to a<br />

spectrometer for measuring <strong>the</strong> spectral c<strong>on</strong>tent of <strong>the</strong> <strong>beam</strong>, while most of <strong>the</strong> <strong>beam</strong><br />

is in <strong>the</strong> zeroth order <strong>and</strong> is passed down <strong>the</strong> <strong>beam</strong>line to <strong>the</strong> experiments. The groove<br />

profile of <strong>the</strong> grating is tuned to give just sufficient intensity in <strong>the</strong> diffracted order<br />

for <strong>the</strong> spectrometer to work (thus maximising <strong>the</strong> <strong>beam</strong> intensity to <strong>the</strong> experiment)<br />

<strong>and</strong> <strong>the</strong> line spacing is varied to give a flat field for <strong>the</strong> dispersed spectrum <strong>on</strong> <strong>the</strong><br />

detector. Such an instrument is being installed at Flash [135].<br />

A point of c<strong>on</strong>cern for operating gratings in free electr<strong>on</strong> laser <strong>beam</strong>s is <strong>the</strong> effect<br />

of <strong>the</strong> structured surface <strong>on</strong> <strong>the</strong> likelihood of ablati<strong>on</strong> since parts of <strong>the</strong> surface may<br />

make very steep angles to <strong>the</strong> incoming <strong>beam</strong>. The effect of <strong>the</strong> groove shape <strong>on</strong><br />

<strong>the</strong> ablati<strong>on</strong> threshold is not understood. If this were to be a problem it would be<br />

exacerbated with classical grating mounts where <strong>the</strong> grating is likely to operate at<br />

quite a steep angle to <strong>the</strong> <strong>beam</strong> at l<strong>on</strong>g wavelength part of <strong>the</strong> spectrum. In this<br />

regard, c<strong>on</strong>ical diffracti<strong>on</strong> in c<strong>on</strong>stant altitude mode has an advantage as <strong>the</strong> <strong>beam</strong><br />

strikes <strong>the</strong> grating at effectively <strong>the</strong> same grazing angle even as <strong>the</strong> grating is tuned.<br />

Grids<br />

Wire grids can be used as polarising <strong>beam</strong> splitters for FIR radiati<strong>on</strong> <strong>and</strong> are used<br />

in Fourier Transform Infra-Red (FTIR) spectrometers. Radiati<strong>on</strong> with polarisati<strong>on</strong><br />

parallel to <strong>the</strong> wires is reflected, whereas <strong>the</strong> comp<strong>on</strong>ent polarised perpendicular to<br />

<strong>the</strong> wires is transmitted. As an example, <strong>the</strong> Millimetre-Wave Technology group[136]<br />

in <strong>the</strong> Space Science <strong>and</strong> Technology Department of <strong>the</strong> UK Science <strong>and</strong> Technology<br />

Facilities Council (STFC) has manufactured wire grids for a FTIR spectrometer to be<br />

used <strong>on</strong> <strong>the</strong> THz <strong>beam</strong>line <strong>on</strong> ALICE, <strong>the</strong> energy-recovery-linac-based developmental<br />

light source at Daresbury Laboratory.<br />

5.4 Wavefr<strong>on</strong>t divisi<strong>on</strong> <strong>beam</strong>splitters<br />

Beamline apertures<br />

The simplest method of splitting a <strong>beam</strong> by wavefr<strong>on</strong>t divisi<strong>on</strong> is to use apertures to<br />

divide <strong>the</strong> <strong>beam</strong> into two spatially separated parts, as is often d<strong>on</strong>e <strong>on</strong> synchrotr<strong>on</strong><br />

sourceswhere<strong>the</strong>reisalarge fanofradiati<strong>on</strong>(e.g.fromdipolesorwavelengthshifters).<br />

The disadvantage ofthis is that <strong>the</strong> angular separati<strong>on</strong> of<strong>the</strong> two <strong>beam</strong>s is determined<br />

by <strong>the</strong> natural divergence of <strong>the</strong> source, which will be ra<strong>the</strong>r small for a free electr<strong>on</strong><br />

laser. Never<strong>the</strong>less, <strong>the</strong> technique might be suitable for separating off a small part<br />

from <strong>the</strong> extremity of <strong>the</strong> <strong>beam</strong> to pass to a <strong>beam</strong> m<strong>on</strong>itor or some <strong>beam</strong> diagnostics,<br />

or for coherent scattering experiments from small samples. If greater separati<strong>on</strong> is<br />

required, a mirror could be used to deflect <strong>on</strong>e <strong>beam</strong> downstream of <strong>the</strong> aperture,<br />

though this is effectively <strong>the</strong> same situati<strong>on</strong> as <strong>the</strong> knife-edge mirror described below.<br />

As with o<strong>the</strong>r methods that divide <strong>the</strong> wavefr<strong>on</strong>t, edge diffracti<strong>on</strong> effects are likely to


84 5. Beam-splitting methods<br />

be significant. The design of <strong>the</strong> aperture would have to eliminate <strong>the</strong> risk of radiati<strong>on</strong><br />

damage.<br />

Knife-edge mirrors<br />

The radiati<strong>on</strong> can be split by inserting a mirror part way into <strong>the</strong> <strong>beam</strong> to deflect a<br />

porti<strong>on</strong> of it. As an example, a grazing incidence (3 ◦ ) knife-edge mirror is used as<br />

part of an autocorrelator system designed at BESSY for use in soft X-ray pump-probe<br />

experiments at Flash[137]. The autocorrelator was designed to work up to 200 eV,<br />

<strong>and</strong> <strong>the</strong> slope error <strong>on</strong> <strong>the</strong> mirror is required to be less than 0.5 mikro-rad. This slope<br />

error must be maintained up to <strong>the</strong> cutting edge of <strong>the</strong> mirror <strong>and</strong> this is very difficult<br />

to achieve with c<strong>on</strong>venti<strong>on</strong>al polishing techniques as <strong>the</strong>re is always some ”roll-off”<br />

at <strong>the</strong> mirror edges. The required quality was achieved by cutting away <strong>the</strong> end of<br />

<strong>the</strong> mirror after polishing to remove <strong>the</strong> roll-off regi<strong>on</strong>. Alternatively, <strong>the</strong> edge of <strong>the</strong><br />

mirror could be masked with an aperture (see above).<br />

Beam splitter 3°<br />

Figure 5.1: Schematic of knife-edge grazing incidence mirror <strong>beam</strong> splitter for soft X-rays.<br />

For harder X-rays, <strong>the</strong> grazing angle needs to be smaller to maintain a high reflectivity.<br />

The required minimum <strong>beam</strong> separati<strong>on</strong> of 0.3 ◦ in <strong>the</strong> sample specificati<strong>on</strong><br />

above means that <strong>the</strong> grazing angle must be greater than 0.15 ◦ . For carb<strong>on</strong>, <strong>the</strong><br />

reflectivity is greater than 98% for energies from 1 keV to 10 keV at this incidence<br />

angle, but <strong>the</strong> mirror could become very l<strong>on</strong>g. For slightly larger angles <strong>the</strong> cut-off in<br />

reflectivity with increasing energy will be <strong>the</strong> main limiting factor in <strong>the</strong> usable wavelength<br />

range, e.g. <strong>the</strong> reflectivity of carb<strong>on</strong> decreases rapidly above about 3 keV for a<br />

grazing angle of 0.5 ◦ . Using a coating with higher atomic number (nickel, rhodium)<br />

will give a reflectivity cut-off at higher energy for a given angle, but such materials<br />

do not have as high a damage threshold as carb<strong>on</strong>. The main c<strong>on</strong>cern about using<br />

a mirror as a <strong>beam</strong> splitter is that edge diffracti<strong>on</strong> could have a significant effect <strong>on</strong><br />

<strong>the</strong> wavefr<strong>on</strong>t, especially as <strong>the</strong> <strong>beam</strong> is being cut in <strong>the</strong> middle where <strong>the</strong> amplitude<br />

is likely to be highest. This type of <strong>beam</strong> splitter cannot <strong>the</strong>refore be expected to<br />

produce two <strong>beam</strong>s that are lower intensity replicas of <strong>the</strong> original <strong>beam</strong>.<br />

Knife-edge crystals<br />

To achieve higher angular separati<strong>on</strong> at shorter wavelengths, <strong>the</strong> knife-edge mirror<br />

could be replaced by a knife-edge crystal. For example, <strong>the</strong> Bragg angle of <strong>the</strong> silic<strong>on</strong><br />

(111) reflecti<strong>on</strong>at8keVis14.3 ◦ , whichgivesa<strong>beam</strong>deflecti<strong>on</strong>angle of28.6 ◦ . Doublecrystal<br />

arrangements could be used to give a deflected <strong>beam</strong> with a fixedexit directi<strong>on</strong><br />

as <strong>the</strong> crystal is rotated to tune <strong>the</strong> phot<strong>on</strong> energy. The splitting crystal would have<br />


5.4. Wavefr<strong>on</strong>t divisi<strong>on</strong> <strong>beam</strong>splitters 85<br />

to be translated orthog<strong>on</strong>ally to <strong>the</strong> <strong>beam</strong> directi<strong>on</strong> or rotated about its cutting edge<br />

if <strong>the</strong> fracti<strong>on</strong> of <strong>the</strong> <strong>beam</strong> it intercepts is to remain c<strong>on</strong>stant. Edge diffracti<strong>on</strong> <strong>and</strong><br />

<strong>the</strong> impact <strong>on</strong> <strong>the</strong> wavefr<strong>on</strong>t would still be an issue, especially as it may be difficult<br />

to maintain a perfect crystalline structure to its edge. There would also be a change<br />

in <strong>the</strong> pulse length of <strong>the</strong> <strong>beam</strong> deflected by <strong>the</strong> crystal due to <strong>the</strong> m<strong>on</strong>ochromating<br />

acti<strong>on</strong> of <strong>the</strong> crystal, <strong>and</strong> <strong>the</strong> reduced b<strong>and</strong>width will also limit <strong>the</strong> overall efficiency<br />

for <strong>the</strong> reflected <strong>beam</strong>.<br />

Fresnel bi-mirror<br />

Fresnel bi-mirrors c<strong>on</strong>sist of two flat mirrors, normally joined al<strong>on</strong>g <strong>on</strong>e edge <strong>and</strong><br />

inclined at an angle to each o<strong>the</strong>r <strong>–</strong> see Figure 5.2. In general <strong>the</strong>y are used in<br />

interferometers where <strong>the</strong> radiati<strong>on</strong> reflected from <strong>the</strong> two parts of <strong>the</strong> mirror is<br />

inclined towards each o<strong>the</strong>r, resulting in interference. An example is <strong>the</strong>ir use in an<br />

interferometer <strong>on</strong> SU7 at SUPERACO[138], where silica mirrors at a grazing angle of<br />

3−6 ◦ were used for 4.4 nm. While bi-mirrors could be used to cross <strong>the</strong> <strong>beam</strong>s over<br />

each o<strong>the</strong>r <strong>and</strong> separate <strong>the</strong>m, it is hard to see any advantage over <strong>the</strong> use of a single<br />

grazing incidence knife-edge mirror<br />

a<br />

a’<br />

Fresnel’s bi-mirror<br />

Figure 5.2: Basic design of <strong>the</strong> Fresnel mirror interferometer. Rays a <strong>and</strong> a ′ create an interference pattern<br />

when <strong>the</strong>y add up.<br />

Slotted or perforated mirrors<br />

In this method, <strong>the</strong> incident wavefr<strong>on</strong>t is coherently split <strong>on</strong> a microscopic scale with<br />

<strong>the</strong> use of holes or slots cut in a mirror. This method is easier for l<strong>on</strong>ger wavelengths,<br />

but has been realised in <strong>the</strong> XUV regime, e.g. a slotted mirror has been used in an<br />

interferometer <strong>on</strong> <strong>beam</strong>line 9.3.2 <strong>on</strong> <strong>the</strong> ALS[139]. This is similar to <strong>the</strong> proposal<br />

for a prototype <strong>beam</strong> splitter to be made for <strong>IRUVX</strong> by AZM at BESSY[140], in<br />

which arrays of small holes will be machined in a grazing incidence mirror. An area<br />

of c<strong>on</strong>cern is <strong>the</strong> ability to manufacture high aspect ratio holes to a great enough<br />

precisi<strong>on</strong> <strong>and</strong> surface quality. The effect <strong>on</strong> <strong>the</strong> coherence of <strong>the</strong> radiati<strong>on</strong> must also<br />

be c<strong>on</strong>sidered.<br />

p


86 5. Beam-splitting methods<br />

Losses in reflecti<strong>on</strong> due to diffracti<strong>on</strong> as well as filling factor have to be taken<br />

into account for mirrors with holes in <strong>the</strong>m. One such example in <strong>the</strong> field of microelectromechanical<br />

systems (MEMS), an area of c<strong>on</strong>siderable research effort, has been<br />

found. A key comp<strong>on</strong>ent for future nanoscale optical devises is <strong>the</strong> freest<strong>and</strong>ing<br />

micro-machined mirror <strong>–</strong> as shown in Figure 5.3.<br />

Figure 5.3: A MEMS free-space optical reflector. Taken from Zou, et al. [141] .Note <strong>the</strong> scale at <strong>the</strong> top of<br />

<strong>the</strong> figure <strong>–</strong> <strong>the</strong> line is 200 µm l<strong>on</strong>g.<br />

The holes in <strong>the</strong> mirror surface serve no useful purpose regarding <strong>the</strong> functi<strong>on</strong><br />

of <strong>the</strong> reflective surface. They are comm<strong>on</strong>ly referred to as release holes <strong>and</strong> allow<br />

release etchant to flow behind <strong>the</strong> mirror <strong>on</strong>ce <strong>the</strong> micromachining process is finished,<br />

releasing <strong>the</strong> comp<strong>on</strong>ent from <strong>the</strong> bulk material. In[141], <strong>the</strong> losses in <strong>the</strong> reflectivity<br />

due to <strong>the</strong> filling factor <strong>and</strong> diffracti<strong>on</strong> are calculated for light of wavelength 632.8<br />

nm as a functi<strong>on</strong> of hole size from 5 - 23 µm <strong>and</strong> spacings of 10 - 30 µm. For <strong>the</strong><br />

case of 21 µm square holes spaced at 30 µm, <strong>the</strong> filling factor is 50%, <strong>and</strong> <strong>the</strong> loss in<br />

reflected light due to diffracti<strong>on</strong> was estimated to be 14%. For <strong>the</strong> same spacing, as


5.4. Wavefr<strong>on</strong>t divisi<strong>on</strong> <strong>beam</strong>splitters 87<br />

<strong>the</strong> hole size decreases, <strong>the</strong> diffracti<strong>on</strong> loss increases, so that for a 10 µm hole size <strong>the</strong><br />

filling factor loss is 11%, <strong>and</strong> <strong>the</strong> diffracti<strong>on</strong> loss 15%.<br />

Much bigger holes relative to <strong>the</strong> wavelength were used in <strong>the</strong> <strong>beam</strong> splitter for<br />

<strong>the</strong> ALS interferometer, designed to operate between 60 eV <strong>and</strong> 100 eV. Slots were<br />

cut in a highly polished (rms roughness ∼ 3 ˚A) single crystal silic<strong>on</strong> wafer. Based <strong>on</strong><br />

<strong>the</strong> coherence properties of <strong>the</strong> incident X-ray <strong>beam</strong>, <strong>the</strong> width <strong>and</strong> spacing of <strong>the</strong><br />

slots were set to 50 µm <strong>and</strong> 100 µm respectively. The slots were 15 mm in length<br />

<strong>and</strong> created by chemical etching. The completed assembly was <strong>the</strong>n coated with<br />

molybdenum. A major c<strong>on</strong>cern in <strong>the</strong> manufacturing was to keep <strong>the</strong> mirror flat to<br />

within 1 µrad, <strong>and</strong> to maintain <strong>the</strong> surface smoothness.<br />

Structured arrays<br />

A ’1-D capillary <strong>beam</strong> splitter’ has been designed to work around 13.9 nm (89 eV) <strong>on</strong><br />

a plasma-based XUV source[142]. It c<strong>on</strong>sists of a stack of 20 µm thick plates, 7.9 mm<br />

l<strong>on</strong>g <strong>and</strong> separated by 130 µm. The plates are tilted at a small angle (grazing angle<br />

of 0−0.8 ◦ ) with respect to <strong>the</strong> incoming radiati<strong>on</strong>. Light can ei<strong>the</strong>r pass unhindered<br />

between <strong>the</strong> plates, or else undergo a single reflecti<strong>on</strong>, causing two <strong>beam</strong>s to emerge<br />

at different angles. At 0.8 ◦ angle of incidence, all <strong>the</strong> light has to be reflected to get<br />

through <strong>the</strong> device. The transmittance (efficiency of <strong>the</strong> direct path) varies from 85%<br />

to 0% depending <strong>on</strong> <strong>the</strong> angle of incidence. The efficiency of <strong>the</strong> reflected light varies<br />

from 0% to 75% <strong>and</strong> 15% of <strong>the</strong> light is lost by absorpti<strong>on</strong> <strong>on</strong> <strong>the</strong> fr<strong>on</strong>t edges of <strong>the</strong><br />

plates. If a similar design were to be used for splitting free electr<strong>on</strong> laser <strong>beam</strong>s, <strong>the</strong><br />

stack would need to be designed to absorb a smaller fracti<strong>on</strong> of <strong>the</strong> radiati<strong>on</strong>. The<br />

effect <strong>on</strong> <strong>the</strong> wavefr<strong>on</strong>t would also need to be investigated.<br />

Capillary arrays have been used in <strong>the</strong> keV regime to suppress higher order harm<strong>on</strong>ics<br />

passed by a m<strong>on</strong>ochromator <strong>on</strong> a bending magnet <strong>beam</strong>line at BESSY[143].<br />

This device uses a double reflecti<strong>on</strong> in <strong>the</strong> array, <strong>the</strong> grazing angle of 3.6 mrad giving<br />

an 89% efficiency at each reflecti<strong>on</strong>, so <strong>the</strong> reflected light is in <strong>the</strong> same directi<strong>on</strong> as<br />

<strong>the</strong> incident; however this example shows that single reflecti<strong>on</strong> capillary plates could<br />

be used at this wavelength.<br />

An alternative c<strong>on</strong>structi<strong>on</strong> scheme could be to adopt <strong>the</strong> same technology used to<br />

form Multilayer Laue Lenses (MLLs)[144]. The x-ray lens technology developed by<br />

Arg<strong>on</strong>ne Nati<strong>on</strong>al Laboratory c<strong>on</strong>sists of many individual layers precisely sputtered<br />

<strong>on</strong>to a silic<strong>on</strong> wafer. The multilayer stack is <strong>the</strong>n ”sliced” to form a thin transmissi<strong>on</strong><br />

element in which <strong>the</strong> stacked diffracting surfaces are oriented almost parallel with <strong>the</strong><br />

optical axis. Although developed as a Fresnel lens structure for micro-focus applicati<strong>on</strong>s,<br />

it is foreseeable that <strong>the</strong> same technology could be used to create a <strong>beam</strong><br />

splitter, ei<strong>the</strong>r by employing a similar geometry to that used in <strong>the</strong> capillary <strong>beam</strong><br />

splitter stack or as a knife edge diffractive element redirecting a proporti<strong>on</strong> of <strong>the</strong><br />

<strong>beam</strong>.<br />

Possible drawbacks for this unproven technology include radiati<strong>on</strong> damage to <strong>the</strong><br />

multilayer structure <strong>and</strong> achieving <strong>the</strong> necessary manufacturing tolerances for both<br />

<strong>the</strong> microstructure fabricati<strong>on</strong> <strong>and</strong> material slicing.


88 5. Beam-splitting methods<br />

5.5 Time-based splitting<br />

An alternative to splitting <strong>the</strong> wavefr<strong>on</strong>t in space is to send alternate pulses or trains<br />

of pulses to different <strong>beam</strong>lines. One scenario is to send hours/minutes worth of<br />

pulses to <strong>on</strong>e experiment while <strong>the</strong> o<strong>the</strong>r endstati<strong>on</strong> is changing samples, checking<br />

data, etc. This could be d<strong>on</strong>e simply using a switching mirror. At <strong>the</strong> o<strong>the</strong>r end of<br />

<strong>the</strong> time scale, is may be possible to use vibrating mirrors or alternating mirrors <strong>and</strong><br />

slots <strong>on</strong> a rotating disc to send alternate pulses to two end-stati<strong>on</strong>s. The feasibility<br />

of this will depend <strong>on</strong> <strong>the</strong> repetiti<strong>on</strong> rate of <strong>the</strong> free electr<strong>on</strong> laser.<br />

An early example utilising a disc rotating at 25 Hz was designed <strong>and</strong> used at<br />

<strong>on</strong> a synchrotr<strong>on</strong> radiati<strong>on</strong> source at DESY. A grazing incidence mirror occupied a<br />

segment of roughly a quarter of <strong>the</strong> disc area, with an equally sized slot occupying<br />

about ano<strong>the</strong>r quarter. The system operated for wavelengths 20 - 280 eV, <strong>the</strong> grazing<br />

angle of <strong>the</strong> mirror being 4 ◦ [145].<br />

In order to increase <strong>the</strong> repetiti<strong>on</strong> rate, <strong>on</strong>e could use <strong>the</strong> fact that <strong>the</strong> free electr<strong>on</strong><br />

laser <strong>beam</strong> is much smaller than those from a synchrotr<strong>on</strong> radiati<strong>on</strong> source <strong>and</strong><br />

c<strong>on</strong>sider a system of several mirrors <strong>and</strong> slots <strong>on</strong> a disc, <strong>and</strong> also increase <strong>the</strong> rotati<strong>on</strong>al<br />

speed of <strong>the</strong> disc. High-speed slotted discs have been designed for use as <strong>beam</strong><br />

choppers. For example, a <strong>beam</strong> chopper for operati<strong>on</strong> at phot<strong>on</strong> energies below 50<br />

eV has been built by Forschungszentrum Jülich GmbH for use <strong>on</strong> <strong>the</strong> synchrotr<strong>on</strong><br />

radiati<strong>on</strong> source at BESSY. The disc is aluminium alloy with a diameter of 338 mm<br />

<strong>and</strong> has 1252 slots cut in <strong>the</strong> edge. It uses magnetic bearings <strong>and</strong> is designed to spin<br />

at about 1 kHz. It was reported at <strong>the</strong> SRI meeting in 2008 that <strong>the</strong> chopper had<br />

been successfully tested at 1 kHz <strong>and</strong> will be commissi<strong>on</strong>ed with phot<strong>on</strong> pulses in <strong>the</strong><br />

autumn of 2008[146].<br />

Asec<strong>on</strong>dexample, inthiscaseusingairbearings, is<strong>the</strong>chopperwhichwasdesigned<br />

to chop <strong>the</strong> 4.3 MHz pulsed <strong>beam</strong> from <strong>the</strong> VUV-FEL <strong>on</strong> proposed 4GLS facility to<br />

100 kHz. The disc would have to have withstood a high heat load from <strong>the</strong> 400<br />

W of <strong>beam</strong> power as well as high mechanical stresses from <strong>the</strong> high speed rotati<strong>on</strong>.<br />

C<strong>on</strong>sequently, martensitic stainless steel was chosen for <strong>the</strong> disk material. The disc<br />

had a polished chamfered edge to reflect <strong>the</strong> unwanted light at 2.5 ◦ grazing angle. A<br />

prototype chopper with just 2 slots has been built <strong>and</strong> tested by Fluid Film Devices<br />

Ltd[147]. The disc has a diameter of 134 mm <strong>and</strong> was tested up to 500 Hz, half <strong>the</strong><br />

design frequency, before <strong>the</strong> project was disc<strong>on</strong>tinued due to 4GLS being replaced by<br />

<strong>the</strong> NLS project.<br />

If <strong>on</strong>e wanted to use <strong>the</strong> chamfered edge to reflect <strong>the</strong> light into a <strong>beam</strong>line for<br />

use by an experiment, <strong>the</strong> edge would need to be polished to an optical quality<br />

surface. An alternative would be to attach mirrors to <strong>the</strong> disc. Both opti<strong>on</strong>s require<br />

fur<strong>the</strong>r feasibility studies. For <strong>the</strong> martensitic steel prototype described above, <strong>the</strong><br />

manufacturer was not able to polish to <strong>the</strong> edge to optical quality, though <strong>the</strong> main<br />

aim was simply to reflect as much of <strong>the</strong> incident power as possible to prevent it<br />

from being absorbed in <strong>the</strong> disc. Distorti<strong>on</strong>s at <strong>the</strong> edge of <strong>the</strong> disc could also be<br />

a problem. For <strong>the</strong> sec<strong>on</strong>d opti<strong>on</strong>, <strong>the</strong> feasibility of having mirrors b<strong>on</strong>ded <strong>on</strong>to a<br />

high-speed rotating disc would need to be investigated. Fur<strong>the</strong>rmore, this technique<br />

would be difficult to apply to X-rays since <strong>the</strong> very grazing angle required for good<br />

reflectivity would lead to a thick disc <strong>and</strong> very high mechanical loads.<br />

At Flash <strong>the</strong> X-ray <strong>beam</strong> can be switched between two user experiments by a mechanically<br />

switching mirror. This reduces <strong>the</strong> repetiti<strong>on</strong> rate at <strong>the</strong> user experiments


5.6. Summary 89<br />

to 2.5 Hz (with maximal switching frequency). The stated accuracy of <strong>the</strong> movement<br />

back <strong>and</strong> forth is a few µ m in <strong>the</strong> positi<strong>on</strong> <strong>and</strong> about 1 arcsec<strong>on</strong>d in <strong>the</strong> angle[148].<br />

5.6 Summary<br />

As <strong>the</strong> preceding secti<strong>on</strong>s have shown, <strong>the</strong>re is a wide variety of techniques that can<br />

be used divide a phot<strong>on</strong> <strong>beam</strong>. Some of <strong>the</strong>se techniques are very well established,<br />

whilst some are more developmental. A key issue for free electr<strong>on</strong> laser sources is that<br />

<strong>the</strong> properties of <strong>the</strong> radiati<strong>on</strong> produced extend into new ground when compared<br />

with c<strong>on</strong>venti<strong>on</strong>al laboratory or synchrotr<strong>on</strong> sources. The short wavelengths, high<br />

transverse coherence, short pulse durati<strong>on</strong> <strong>and</strong> high pulse energy all mean that <strong>the</strong>re<br />

is some element of development required for any splitting technique.<br />

Two lists of techniques are presented for fur<strong>the</strong>r c<strong>on</strong>siderati<strong>on</strong> for use mainly in <strong>the</strong><br />

soft x-ray regime. The first list gives those techniques that would seem amenable to<br />

rapid development into practical <strong>beam</strong> splitters. The sec<strong>on</strong>d list gives <strong>the</strong> techniques<br />

that show potential but will require more extensive development to over come <strong>the</strong><br />

technical challenges. In all cases, <strong>the</strong> properties of a <strong>beam</strong> splitter derived from a<br />

particular technique will be somewhat specialised <strong>and</strong> so <strong>the</strong>re will be no ”universal”<br />

<strong>beam</strong> splitter. The technique chosen will depend <strong>on</strong> <strong>the</strong> nature of <strong>the</strong> phot<strong>on</strong> <strong>beam</strong><br />

being split <strong>and</strong> <strong>the</strong> particular needs of <strong>the</strong> experiment.<br />

Techniques requiring <strong>the</strong> least development<br />

Diffracti<strong>on</strong> gratings<br />

Reflecti<strong>on</strong> gratings give amplitude divisi<strong>on</strong> <strong>and</strong> since <strong>the</strong>y can be made to very high<br />

tolerances, splitting with good wavefr<strong>on</strong>t c<strong>on</strong>trol should be possible. Polarizati<strong>on</strong><br />

effects will be present but should <strong>on</strong>ly be significant at VUV <strong>and</strong> XUV wavelengths.<br />

At least <strong>on</strong>e of <strong>the</strong> <strong>beam</strong>s will be m<strong>on</strong>ochromatic. Additi<strong>on</strong>al optics are required to<br />

keep <strong>the</strong> output <strong>beam</strong> directi<strong>on</strong>s c<strong>on</strong>stant if tuning of <strong>the</strong> phot<strong>on</strong> energy is required.<br />

Pulse stretch is inevitable for ultra-short pulses, but can be corrected with a sec<strong>on</strong>d<br />

grating. In general, flexibility is high, but <strong>the</strong> systems are likely to be complex <strong>and</strong><br />

optimizati<strong>on</strong>s will give a more restricted operating envelope. The likely operating<br />

range is from <strong>the</strong> VUV to <strong>the</strong> soft X-rays. Infrared gratings are also feasible, though<br />

<strong>the</strong>re are arguably easier splitting techniques for this part of <strong>the</strong> spectrum.<br />

Knife-edge mirrors<br />

A knife-edge mirror is <strong>on</strong>e of <strong>the</strong> simplest techniques <strong>and</strong> also offers a very flexible<br />

splitting with minimal c<strong>on</strong>straints. Both <strong>beam</strong>s are essentially spectrally unmodified,<br />

though <strong>the</strong> reflected <strong>beam</strong> will be filtered at wavelengths shorter than <strong>the</strong> mirror<br />

reflecti<strong>on</strong> cut-off. Very low grazing angles will thus be required for hard X-ray operati<strong>on</strong>,<br />

<strong>and</strong> so <strong>the</strong> splitting angle will become very small. (Bragg reflecti<strong>on</strong> from a<br />

crystal will give larger angles but with <strong>the</strong> disadvantages of reduced b<strong>and</strong>width <strong>and</strong><br />

<strong>the</strong> need to rotate <strong>the</strong> crystal for wavelength tuning). The splitting is achieved by<br />

wavefr<strong>on</strong>t divisi<strong>on</strong> <strong>and</strong> so <strong>the</strong> main disadvantage is that nei<strong>the</strong>r <strong>beam</strong> is a replica at<br />

reduced intensity <strong>the</strong> incident <strong>beam</strong> since diffracti<strong>on</strong> effects at <strong>the</strong> edge will distort<br />

<strong>the</strong> wavefr<strong>on</strong>t. However, pulse lengths should be preserved. A key technical challenge


90 5. Beam-splitting methods<br />

is achieving a mirror with an edge that does not badly degrade <strong>the</strong> <strong>beam</strong> at <strong>the</strong> divisi<strong>on</strong><br />

regi<strong>on</strong>. Knife-edged mirrors could be designed to operate over <strong>the</strong> entire spectral<br />

range from infrared to X-ray wavelengths (though not necessarily in <strong>on</strong>e device).<br />

Slotted mirrors<br />

A slotted (or perforated) mirror may overcome <strong>the</strong> key disadvantage of <strong>the</strong> knifeedge<br />

mirror by dividing <strong>the</strong> wavefr<strong>on</strong>t periodically in space (thus making each <strong>beam</strong><br />

closer to aless intense replica of <strong>the</strong> incident <strong>beam</strong>) whilst maintaining <strong>the</strong> advantages<br />

of transparency (to <strong>the</strong> pulse properties) <strong>and</strong> flexibility. The technical challenge of<br />

making <strong>the</strong> mirror is however much more severe. There is no inherent restricti<strong>on</strong> <strong>on</strong><br />

spectral range, though <strong>the</strong> technical challenges will become more severe at shorter<br />

wavelengths <strong>and</strong> <strong>the</strong> devices may be limited to <strong>the</strong> soft X-ray regime <strong>and</strong> below.<br />

Crystal diffracti<strong>on</strong> <strong>beam</strong> splitters<br />

These are mainly applicable to use at shorter wavelengths (< 6 ˚A), but have <strong>the</strong><br />

advantage in this range over mirror type splitters of allowing much greater angular<br />

separati<strong>on</strong>s. Wavelengthtuningrequires arotati<strong>on</strong> of <strong>the</strong>crystal <strong>and</strong> henceadditi<strong>on</strong>al<br />

optics to maintain a c<strong>on</strong>stant output <strong>beam</strong> positi<strong>on</strong>. Laue <strong>and</strong> Bragg reflecti<strong>on</strong>s<br />

are always highly m<strong>on</strong>ochromatic (unless <strong>on</strong>e distorts <strong>the</strong> crystal lattice) <strong>and</strong> so <strong>the</strong><br />

full b<strong>and</strong>width of <strong>the</strong> free electr<strong>on</strong> laser is not preserved. Accompanying this is<br />

a stretching in <strong>the</strong> pulse length <strong>and</strong> whilst this stretching should be small (a few<br />

femtosec<strong>on</strong>ds), it cannot be reversed as is <strong>the</strong> case with a grating. Only in <strong>the</strong> case<br />

of very thin crystals can a broadb<strong>and</strong> transmissi<strong>on</strong> be achieved, though <strong>the</strong>re will be<br />

a missing comp<strong>on</strong>ent matching <strong>the</strong> b<strong>and</strong>width of Laue/Bragg reflecti<strong>on</strong> of <strong>the</strong> split<br />

<strong>beam</strong>. The most important area for technical development is in achieving higher<br />

quality diam<strong>on</strong>d crystals, which are attractive due to <strong>the</strong>ir high <strong>the</strong>rmal c<strong>on</strong>ductivity<br />

<strong>and</strong> resistance to radiati<strong>on</strong> damage.<br />

Techniques requiring more development<br />

Multilayers<br />

A transmissi<strong>on</strong> multilayer is <strong>the</strong> most likely way that a plate <strong>beam</strong> splitter can be<br />

realized at short wavelengths. In fact, <strong>the</strong> multilayer <strong>beam</strong> splitter will be more akin<br />

to a pellicle. The technical challenges are thus making a membrane that is robust<br />

<strong>and</strong> flat but has adequate transmissi<strong>on</strong>. If <strong>the</strong> membrane is small, this will be easier<br />

but <strong>the</strong>n <strong>the</strong> <strong>beam</strong> fluence will be high if <strong>the</strong> <strong>beam</strong> must be focussed <strong>on</strong>to it in order<br />

to fit through it. A larger membrane would allow operati<strong>on</strong> in <strong>the</strong> diverged light but<br />

may be challenging to keep flat <strong>and</strong> vibrati<strong>on</strong> free.<br />

Structured arrays<br />

These devices have similar properties to slotted mirrors but differ in <strong>the</strong> approach<br />

to fabricati<strong>on</strong>. Operati<strong>on</strong>al range will depend <strong>on</strong> <strong>the</strong> fabricati<strong>on</strong> technique (e.g. <strong>the</strong><br />

overall thickness of <strong>the</strong> structure is linked to <strong>the</strong> angular acceptance). Wavefr<strong>on</strong>t<br />

qualitywill beverydependent<strong>on</strong><strong>the</strong>manufacturing<strong>and</strong>thusdevelopmentisrequired.


5.6. Summary 91<br />

Time-based splitters<br />

Mechanical switching to direct alternate pulses (or trains of pulses) to different <strong>beam</strong>lines<br />

is a logical approach for an inherently pulsed source. The potential advantage to<br />

<strong>the</strong> experimentis that all received pulses c<strong>on</strong>tain <strong>the</strong> full free electr<strong>on</strong> laser output<strong>and</strong><br />

can (in principle) be unchanged in all o<strong>the</strong>r properties. The difference is a lowering of<br />

<strong>the</strong> repetiti<strong>on</strong> rate or pulse structure (e.g. <strong>the</strong> change from a uniform pulse train to a<br />

macropulsed train). For some experiments this may offset <strong>the</strong> advantage of <strong>the</strong> intact<br />

pulse single pulse energy. In any case, <strong>the</strong>re is a c<strong>on</strong>siderable engineering challenge<br />

of achieving a mechanical based system that can deflect alternating pulses or pulse<br />

trains at rates of <strong>the</strong> order of a kHz or higher. High speed operati<strong>on</strong> will probably<br />

be restricted to <strong>the</strong> VUV/XUV to ease <strong>the</strong> challenge of making a fast moving mirror<br />

that can deflect a complete pulse.


92 5. Beam-splitting methods<br />

Summary<br />

• Splitting <strong>the</strong> phot<strong>on</strong> <strong>beam</strong> can be d<strong>on</strong>e to serve several experiments<br />

in parallel, it also enables <strong>the</strong> use of part of <strong>the</strong> <strong>beam</strong><br />

for diagnostic purposes. The trade off is that less (in some<br />

sense) of <strong>the</strong> <strong>beam</strong> gets h<strong>and</strong>ed out for <strong>the</strong> various purposes.<br />

• A <strong>beam</strong> can be split in <strong>the</strong> amplitude (using e.g. a semitransparent<br />

mirror), frequency (a dispersive optic is needed)<br />

<strong>and</strong> temporal (using for instance a moving mirror) domains.<br />

• Of <strong>the</strong> techniques listed in this chapter not all are suitable<br />

for <strong>the</strong> use at free electr<strong>on</strong> laser. Often <strong>the</strong> limiting factor is<br />

<strong>the</strong> intended splitting techniques robustness when it comes to<br />

withst<strong>and</strong>ing <strong>the</strong> high peak power at X-ray wavelengths.<br />

• In some instances <strong>the</strong> following techniques are already in use<br />

at free electr<strong>on</strong> laser whereas some need some kind of adaptati<strong>on</strong><br />

for <strong>the</strong> purpouse:<br />

<strong>–</strong> Diffracti<strong>on</strong> gratings (dispersive). Induces a pulsestretching<br />

whose length depends <strong>on</strong> how many grooves<br />

are illuminated. Pulse-stretch gets reversed if two gratings<br />

are used, at <strong>the</strong> expense of lower transmissi<strong>on</strong>.<br />

<strong>–</strong> Crystal diffracti<strong>on</strong> splitter <strong>–</strong> effective below ∼ 6˚A wavelengths.<br />

An attractive material is diam<strong>on</strong>d crystals, owing<br />

to <strong>the</strong>ir high <strong>the</strong>rmal c<strong>on</strong>ductivity combined with<br />

good radiati<strong>on</strong> hardness[149]. Thediffracti<strong>on</strong> in<strong>the</strong>crystalcausesan<strong>on</strong>-reversiblepulsestretchofafewfemtosec<strong>on</strong>ds.<br />

<strong>–</strong> Knife-edge mirrors<br />

<strong>–</strong> Slotted mirrors<br />

• O<strong>the</strong>r techniques have been employed for specific purposes<br />

<strong>and</strong> proven to be applicable in <strong>the</strong> free electr<strong>on</strong> laser regime<br />

of experimental c<strong>on</strong>diti<strong>on</strong>s. Introducing <strong>the</strong>m for general utilizati<strong>on</strong><br />

require fur<strong>the</strong>r development <strong>and</strong> research:<br />

<strong>–</strong> Multilayers<br />

<strong>–</strong> Structured arrays<br />

<strong>–</strong> Time-based splitters


Part II<br />

Beam diagnostics


6. Introducti<strong>on</strong><br />

Written by: A. Lindblad<br />

Since <strong>the</strong> Sase-process starts up from white noise <strong>–</strong> i.e. <strong>the</strong> initial shot-noise in <strong>the</strong><br />

<strong>beam</strong> is uniformly distributed, each light-pulse will have different temporal, spatial<br />

<strong>and</strong>spectral properties. Over<strong>the</strong>course ofmanyshots saidproperties average out<strong>and</strong><br />

<strong>the</strong> machine attains its ”average” properties in terms of intensity, pulse-length <strong>and</strong><br />

spectral purity; it is <strong>the</strong>refore natural that <strong>beam</strong> diagnostic methods are an integral<br />

part of any free electr<strong>on</strong> laser-project (see, for instance Ref. [89]).<br />

Diagnostics of <strong>the</strong> phot<strong>on</strong>s (<strong>and</strong> electr<strong>on</strong>s) also provide valuable feedback to <strong>the</strong><br />

accelerator part of <strong>the</strong> machine. This feedback is essential to ensure stable operati<strong>on</strong><br />

<strong>and</strong> l<strong>on</strong>g up-time for <strong>the</strong> users.<br />

The various schemes that exist to enhance <strong>the</strong> overall shot-to-shot repeatability,<br />

e.g. HGHG, EEHG 1 , may seem to loosen <strong>the</strong> dem<strong>and</strong> <strong>on</strong> diagnostics <strong>–</strong> however <strong>the</strong><br />

delicate problem of overlapping a laser pulse (which can be of HHG type <strong>–</strong> also requiring<br />

its own diagnostics) with <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> still requires a strict characterizati<strong>on</strong><br />

of <strong>the</strong> light as in <strong>the</strong> Sase case.<br />

In this chapter <strong>the</strong> various categories of diagnostics are presented, as well as <strong>the</strong><br />

different subcategories that we can sort <strong>the</strong>m into.<br />

6.1 Diagnostics categorizati<strong>on</strong><br />

Figure 6.1 gives an indicati<strong>on</strong> of how we can categorize <strong>the</strong> different diagnostic methods<br />

that needs to be employed to characterize <strong>the</strong> free electr<strong>on</strong> laser phot<strong>on</strong> <strong>beam</strong>.<br />

• Beam cross-secti<strong>on</strong><br />

<strong>–</strong> Transverse (also al<strong>on</strong>g <strong>the</strong> <strong>beam</strong> path to discern <strong>the</strong> focus-size)<br />

<strong>–</strong> L<strong>on</strong>gitudinal<br />

• Pulse arrival time<br />

• The jitter between pulses<br />

• Intensity / Pulse energy<br />

1 Secti<strong>on</strong>s 1.5, see page 17 <strong>and</strong> 1.5, see page 18.<br />

95


96 6. Introducti<strong>on</strong><br />

y<br />

x<br />

z<br />

I<br />

Figure 6.1: Phot<strong>on</strong> pulses needs to be diagnosed vis-à-vis spatial <strong>and</strong> temporal extents. Spectral properties<br />

can be inferred from <strong>the</strong> temporal distributi<strong>on</strong>.<br />

• Spectral c<strong>on</strong>tent<br />

• Median energy<br />

The <strong>beam</strong>’s size in space, <strong>and</strong> how it varies al<strong>on</strong>g <strong>the</strong> optical system is important<br />

to characterize since those properties needs to be known to know <strong>the</strong> focus points<br />

al<strong>on</strong>g <strong>the</strong> <strong>beam</strong> <strong>–</strong> both for simulati<strong>on</strong> purposes <strong>and</strong> experiments.<br />

The l<strong>on</strong>gitudinal cross-secti<strong>on</strong> of <strong>the</strong> <strong>beam</strong> gives <strong>the</strong> pulse length <strong>and</strong> shape. This<br />

is c<strong>on</strong>nected both to <strong>the</strong> spectral c<strong>on</strong>tent <strong>and</strong> <strong>the</strong> pulse to pulse arrival time jitter.<br />

Forexperimentsthatare c<strong>on</strong>sideringtemporal propertiesofmatter (pump-<strong>and</strong>-probe)<br />

this is crucial informati<strong>on</strong>.<br />

The pulse energy (or intensity) follows a gamma-distributi<strong>on</strong> depending <strong>on</strong> how<br />

many radiating modes that are present[104] <strong>–</strong> hence a shot-to-shot measurement of<br />

<strong>the</strong> intensity is necessary to have at a free electr<strong>on</strong> laser facility.<br />

The centroid of <strong>the</strong> phot<strong>on</strong> energy fluctuates about 0.5% at Flash[150], obviously<br />

a shot-to-shot measurement of this needs to be provided, both to <strong>the</strong> users <strong>and</strong> to <strong>the</strong><br />

staff h<strong>and</strong>ling <strong>the</strong> accelerator itself.<br />

Spectral diagnostics (Intensity <strong>and</strong> energy) are discussed in chapter 7 below.<br />

Various methods c<strong>on</strong>cerning <strong>the</strong> measurements of <strong>the</strong> transverse spatial extent of<br />

<strong>the</strong> phot<strong>on</strong> <strong>beam</strong> are described in chapter 8 (see page 109).<br />

A survey of pulse length, profile <strong>and</strong> jitter diagnostics can be found in chapter 9<br />

(see page 151).<br />

Subcategorizati<strong>on</strong>s<br />

Shot-to-shot / Average<br />

Many properties of <strong>the</strong> free electr<strong>on</strong> laser phot<strong>on</strong> <strong>beam</strong> needs to be known at a shotto-shot<br />

basis. A spectroscopic experiment, for instance, needs to know <strong>the</strong> incoming<br />

phot<strong>on</strong>s’ energy <strong>and</strong> intensity per pulse as to allow sorting of <strong>the</strong> experimental<br />

data. Knowing just <strong>the</strong> average intensity <strong>and</strong> phot<strong>on</strong> energy does not allow this<br />

post-experiment analysis of <strong>the</strong> data.<br />

Average properties, <strong>on</strong> <strong>the</strong> o<strong>the</strong>r h<strong>and</strong> says a lot about <strong>the</strong> l<strong>on</strong>g term stability<br />

of <strong>the</strong> free electr<strong>on</strong> laser <strong>and</strong> may also serve as a measurement of <strong>the</strong> c<strong>on</strong>diti<strong>on</strong> of<br />

<strong>transport</strong> optics. Averaging measurements are usually more stable than <strong>the</strong>ir shotto-shot<br />

counterparts <strong>–</strong> <strong>and</strong> can thus be used to calibrate o<strong>the</strong>r diagnostics.<br />

t


6.2. C<strong>on</strong>clusi<strong>on</strong> 97<br />

Transparent/Opaque/Blocking<br />

A diagnostic can be, more or less, invasive, i.e. how much it affects <strong>the</strong> phot<strong>on</strong> <strong>beam</strong>’s<br />

properties for experiments <strong>and</strong> o<strong>the</strong>r diagnostics downstream. An positi<strong>on</strong> measurement<br />

based <strong>on</strong> <strong>the</strong> photo-current generated <strong>on</strong> a slit will cut away parts of <strong>the</strong> <strong>beam</strong>,<br />

thus reducing <strong>the</strong> available intensity downstream from <strong>the</strong> diagnostic.<br />

As can be inferred from <strong>the</strong> title, we can divide <strong>the</strong> degree of invasiveness of a<br />

diagnostic tool into transparent, opaque <strong>and</strong> blocking. As <strong>on</strong>-line diagnostics it is<br />

preferable to have transparent diagnostics since <strong>the</strong>y have least effect <strong>on</strong> <strong>the</strong> <strong>beam</strong>.<br />

For commissi<strong>on</strong>ing of diagnostics, <strong>beam</strong>-<strong>transport</strong> elements, calibrating of diagnostics<br />

<strong>and</strong> accelerator c<strong>on</strong>diti<strong>on</strong>ing opaque <strong>and</strong> blocking diagnostics can be used.<br />

As discussed previously (chapter 5) it is possible to split off part of <strong>the</strong> <strong>beam</strong><br />

to be diagnosed in parallel to <strong>the</strong> downstream experiments/diagnostics. Hence even<br />

a blocking diagnostic that o<strong>the</strong>rwise have <strong>the</strong> needed specificati<strong>on</strong>s can be used to<br />

provide informati<strong>on</strong> to <strong>the</strong> downstream activities.<br />

Phot<strong>on</strong>-energy range<br />

Physical processes that may be used as basis for a diagnostic can be challenging to<br />

find. For infrared <strong>and</strong> THz <strong>beam</strong>s it is not possible to photo-i<strong>on</strong>ize a gas, whereas for<br />

UV <strong>and</strong> soft X-rays <strong>the</strong> cross-secti<strong>on</strong> for this processes is very large <strong>–</strong> <strong>and</strong> in turn, for<br />

hard X-rays <strong>the</strong> cross-secti<strong>on</strong>s for photo-i<strong>on</strong>izati<strong>on</strong> become very small.<br />

Hence, <strong>the</strong> phot<strong>on</strong>-energy range(s) of interest needs to be taken into account in<br />

designing <strong>the</strong> diagnostic array for a facility.<br />

6.2 C<strong>on</strong>clusi<strong>on</strong><br />

There is no such thing as an perfect diagnostic, i.e. a device that measure a property<br />

<strong>on</strong> a shot-to-shot basis in a transparent manner for any phot<strong>on</strong>-energy with negligible<br />

error. Thus we have torely <strong>on</strong> acombinati<strong>on</strong> of diagnostics tomeasure <strong>the</strong>desired unknowns<br />

of <strong>the</strong> phot<strong>on</strong> <strong>beam</strong>. To ensure <strong>the</strong> integrity of <strong>the</strong> resulting diagnostics array<br />

<strong>on</strong>e needs to calibrate <strong>the</strong>m against each o<strong>the</strong>r (<strong>and</strong> possibly additi<strong>on</strong>al diagnostics).


98 6. Introducti<strong>on</strong><br />

Summary<br />

• A diagnostic needs to be more robust <strong>and</strong> more user friendly<br />

than an experiment, as <strong>the</strong> informati<strong>on</strong> provided is used to<br />

analyze experimental data <strong>–</strong> for users, as feedback to <strong>the</strong> running<br />

of <strong>the</strong> accelerator <strong>and</strong> for commissi<strong>on</strong>ing of <strong>the</strong> <strong>transport</strong><br />

optics.<br />

• Diagnostics can be divided into:<br />

<strong>–</strong> Beam cross-secti<strong>on</strong> <strong>–</strong> transverse <strong>and</strong> l<strong>on</strong>gitudinal<br />

<strong>–</strong> Arrival time, jitter<br />

<strong>–</strong> Intensity <strong>and</strong> Pulse energy<br />

<strong>–</strong> Spectral c<strong>on</strong>tent<br />

<strong>–</strong> Median energy<br />

• Any diagnostic method can also be fur<strong>the</strong>r characterized by:<br />

<strong>–</strong> its ability to measure <strong>on</strong> a shot-to-shot basis or if it provides<br />

an average measurement of <strong>the</strong> property.<br />

<strong>–</strong> if it is transparent, opaque or <strong>beam</strong> stopping/blocking for<br />

a downstream experiment.<br />

<strong>–</strong> what phot<strong>on</strong>-energy range it can be used in, i.e. infrared/THz,<br />

UV, soft X-ray <strong>and</strong> hard X-rays.<br />

• No perfect diagnostic exists (nei<strong>the</strong>r an universal that measures<br />

everything, nor optimal for all phot<strong>on</strong> ranges). Thus<br />

a judicious combinati<strong>on</strong> of diagnostics needs to be composed<br />

into a diagnostics array distributed al<strong>on</strong>g <strong>the</strong> <strong>beam</strong>-path that<br />

can be used to<br />

<strong>–</strong> Measure <strong>–</strong> during <strong>the</strong> course of o<strong>the</strong>r experiments<br />

<strong>–</strong> Calibrate <strong>–</strong> o<strong>the</strong>r diagnostics during commissi<strong>on</strong>ing<br />

<strong>–</strong> Cross-check <strong>–</strong> <strong>the</strong> integrity of <strong>the</strong> diagnostics array<br />

• Beam-splitters allows <strong>the</strong> use of opaque/blocking diagnostics<br />

in parallel to o<strong>the</strong>r experiments <strong>and</strong> diagnostics.


7. Spectral diagnostics: Intensity & Energy<br />

Written by: A. Lindblad<br />

In this chapter some ways to infer <strong>the</strong> spectral properties (or some spectral property)<br />

of <strong>the</strong> free electr<strong>on</strong> laser light. During <strong>the</strong> course of Part I of <strong>the</strong> book we have<br />

gotten <strong>the</strong> indicati<strong>on</strong> that both seeded <strong>and</strong> Sase free electr<strong>on</strong> laser provide light that<br />

need to be diagnosed <strong>on</strong> a shot to shot basis.<br />

The Sase process start up from current shot noise in <strong>the</strong> <strong>beam</strong> that is subsequently<br />

amplified with a few radiati<strong>on</strong> modes c<strong>on</strong>tributing to <strong>the</strong> final spectrum at/after<br />

saturati<strong>on</strong>. Each pulse is <strong>the</strong>refore unique <strong>and</strong> for an experiment to be meaningful<br />

<strong>the</strong> mean intensity <strong>and</strong> mean-energy of <strong>the</strong> pulses can be c<strong>on</strong>sidered tobe <strong>the</strong> minimal<br />

informati<strong>on</strong> to be provided to <strong>the</strong> user. In various seeding schemes <strong>the</strong> quality of <strong>the</strong><br />

pulses needs to be m<strong>on</strong>itored to ensure stability. The harm<strong>on</strong>ic c<strong>on</strong>tent of <strong>the</strong> pulses<br />

<strong>and</strong> <strong>the</strong> sp<strong>on</strong>taneous emissi<strong>on</strong> background levels needs to be diagnosed as well.<br />

Ideally <strong>the</strong> full spectrum of <strong>the</strong> pulse should be measured <strong>on</strong> a pulse to pulse basis<br />

in a manner that is transparent to <strong>the</strong> users of <strong>the</strong> <strong>beam</strong>. This is often not practically<br />

possible, as will be seen in <strong>the</strong> following, but this can be overcome by dividing <strong>the</strong><br />

<strong>beam</strong>between<strong>the</strong>experiment<strong>and</strong><strong>the</strong>diagnostics utilizing, e.g.a<strong>beam</strong>splittingdevice<br />

as discussed above in chapter 5.<br />

7.1 X-ray spectrometry<br />

The obvious spectral diagnostic of <strong>the</strong> fel <strong>beam</strong> is to measure it, or parts of it directly<br />

with an X-ray spectrometer. With <strong>on</strong>e or more gratings it is possible to disperse <strong>the</strong><br />

phot<strong>on</strong> <strong>beam</strong> c<strong>on</strong>verting energy spread into a spatial distributi<strong>on</strong>. In most instances<br />

variable line spacing gratings is used to focus a higher order diffracted <strong>beam</strong> <strong>on</strong>to a<br />

focal plane, passing <strong>the</strong> 0 th order reflected <strong>beam</strong> for <strong>the</strong> experiment (Figure 7.1).<br />

Rewriting <strong>the</strong> grating equati<strong>on</strong> (Equati<strong>on</strong> 4.1) in terms of line density<br />

sinα−sinβ = n·λσ0<br />

<strong>on</strong>e can vary <strong>the</strong> groove depth according to a third degree polynomial (with x al<strong>on</strong>g<br />

<strong>the</strong> grating) as:<br />

σ(x) = σ0 +σ1x+σ2x 2 +σ3x 3<br />

99


100 7. Spectral diagnostics: Intensity & Energy<br />

Incoming FEL pulse<br />

α<br />

β<br />

focal plane<br />

O th order<br />

Figure 7.1: A variable line space grating disperses different wavelengths al<strong>on</strong>g <strong>the</strong> same focal plane.<br />

λ range 6-40 nm 20-60 nm<br />

central line spacing 900 l/mm 300 l/mm<br />

fracti<strong>on</strong> in 1 st order 8-0.5% 11-1.5%<br />

diffracti<strong>on</strong> angle 83.8-74.5 ◦<br />

83.5-79 ◦<br />

Res. power (CCD) > 7000 > 4000<br />

incident angle 88 ◦<br />

88 ◦<br />

coating C, Ni C<br />

Table 7.1: Design parameters of <strong>the</strong> flash VLS grating spectrometer <strong>–</strong> as given in Ref. [104].<br />

Then <strong>on</strong>e can choose σ(x) so that <strong>the</strong> abberati<strong>on</strong> <strong>and</strong> spectral defocusing effects are<br />

minimized. With proper materials chosen between <strong>on</strong>e <strong>and</strong> ten percent of <strong>the</strong> radiati<strong>on</strong><br />

is dispersed <strong>on</strong>to <strong>the</strong> focal plane (this percentage also depend <strong>on</strong> <strong>the</strong> wavelength).<br />

This kind of spectrometers provide <strong>the</strong> full spectrum around a certain phot<strong>on</strong><br />

energy <strong>–</strong> harm<strong>on</strong>ics are usually filtered away by <strong>the</strong> grating. The detecti<strong>on</strong> can be<br />

d<strong>on</strong>e ei<strong>the</strong>r by CCD camera or strip detectors. At Flash <strong>the</strong> spectrum is recorded<br />

by a Ce:YAG screen imaged by a CCD camera. The camera can record images at a<br />

rate of 5 Hz[104].<br />

Similar set-ups are used at Lcls[151] <strong>and</strong> at Fermi@Elettra[152].<br />

7.2 Intensity/Beam energy<br />

Earlier we discussed <strong>the</strong> use of m<strong>on</strong>ochromators in combinati<strong>on</strong> with <strong>the</strong> free electr<strong>on</strong><br />

laser<strong>beam</strong> <strong>–</strong> in secti<strong>on</strong> 4.5 (see page 67) <strong>–</strong> as a means to c<strong>on</strong>vert <strong>the</strong> spectral jitter<br />

(wavelength <strong>and</strong> intensity fluctuati<strong>on</strong>s per pulse) into intensity jitter.<br />

Intensitycan be measured <strong>on</strong> a per pulse basis or as an average property over many<br />

pulses, of course a fast enough detecti<strong>on</strong> scheme can be made to integrate over many<br />

pulses if <strong>on</strong>e needs to measure <strong>the</strong> average intensity. As free electr<strong>on</strong> laser pulses are<br />

often in <strong>the</strong> order of tenths of femtosec<strong>on</strong>ds (or shorter) a measuring <strong>on</strong> a pulse by<br />

pulse basis poses quite a challenge as will be seen below.<br />

As for most diagnostics we can roughly divide intensity measurements into opaque<br />

<strong>and</strong>transparent <strong>–</strong>as seenfrom animagined experimentalstati<strong>on</strong> downstream from <strong>the</strong><br />

diagnostic. Diagnostics which are, more or less, transparent are possible to distribute<br />

al<strong>on</strong>g a <strong>beam</strong> path leading up to an experimental stati<strong>on</strong>, whereas opaque diagnostics


7.2. Intensity/Beam energy 101<br />

needs to use ei<strong>the</strong>r a part of <strong>the</strong> free electr<strong>on</strong> laser <strong>beam</strong> using a <strong>beam</strong>splitter, or be<br />

placed after <strong>the</strong> experiment (provided that <strong>the</strong> experiment is somewhat transparent).<br />

Besides <strong>the</strong> gas m<strong>on</strong>itor detectors <strong>and</strong> solid state devices described below a way of<br />

discerning <strong>the</strong> average energy of <strong>the</strong> free electr<strong>on</strong> laser <strong>beam</strong> have been invented at<br />

<strong>the</strong> Lcls[96]. By determining <strong>the</strong> energy loss for different trajectories of <strong>the</strong> electr<strong>on</strong><br />

<strong>beam</strong> through<strong>the</strong>undulatorsameasure of<strong>the</strong>average pulseenergycan bedetermined<br />

within 1 <strong>and</strong> 5%.<br />

Gas m<strong>on</strong>itor detectors<br />

I<strong>on</strong>/electr<strong>on</strong> detecti<strong>on</strong><br />

To provide a transparent diagnostic of <strong>the</strong> intensity at <strong>the</strong> Flash free electr<strong>on</strong> laser a<br />

an intensity m<strong>on</strong>itor based <strong>on</strong> photoi<strong>on</strong>izati<strong>on</strong> of gases <strong>and</strong> detecti<strong>on</strong> of i<strong>on</strong>s <strong>and</strong><br />

electr<strong>on</strong>s have been developed [153] at Flash. These detectors are placed before <strong>and</strong><br />

after <strong>the</strong> gas attenuator[104]. The latter is places immediately before entering <strong>the</strong><br />

experimental hall.<br />

By measuring <strong>the</strong> yield (electr<strong>on</strong> or i<strong>on</strong>) from a known density of a gas it is possible<br />

to c<strong>on</strong>clude <strong>the</strong> intensity of <strong>the</strong> phot<strong>on</strong> <strong>beam</strong> via:<br />

N = Nγ ·ρ·σ(E)·ℓ (7.1)<br />

relates <strong>the</strong> number of particles i<strong>on</strong>ized (N) with <strong>the</strong> number of phot<strong>on</strong>s Nγ; <strong>the</strong> target<br />

density (ρ); <strong>the</strong> photoi<strong>on</strong>izati<strong>on</strong> cross-secti<strong>on</strong> at <strong>the</strong> phot<strong>on</strong> energy in questi<strong>on</strong> σ(E)<br />

<strong>and</strong> <strong>the</strong> length of <strong>the</strong> interacti<strong>on</strong> volume ℓ.<br />

-V<br />

+V<br />

Figure 7.2: A Faraday cup counts <strong>the</strong> i<strong>on</strong>s <strong>and</strong> electr<strong>on</strong>s produced by photoi<strong>on</strong>izati<strong>on</strong> of a target gas to produce<br />

a measure of <strong>the</strong> intensity of <strong>the</strong> FEL phot<strong>on</strong> <strong>beam</strong> via Equati<strong>on</strong> 7.1.<br />

The gas m<strong>on</strong>itor detector scheme is shown in Figure 7.2. The charged particles<br />

(electr<strong>on</strong>s <strong>and</strong> i<strong>on</strong>s) created by <strong>the</strong> i<strong>on</strong>izati<strong>on</strong> of <strong>the</strong> gas is separated <strong>and</strong> accelerated<br />

byahomogeneous electric field, <strong>and</strong> can thus be detected separately. The gas pressure<br />

is typically held at 10 -6 mbar, as to disturb <strong>the</strong> downstream experiments minimally.<br />

The charged particles are detected by Faraday cups.<br />

With this type of detector it is possible to measure <strong>the</strong> pulse intensity with an<br />

error less than 10% with a jitter between pulses of 1%. The latter is dominated by<br />

<strong>the</strong> signal statistics <strong>and</strong> holds for more than 10 10 phot<strong>on</strong>s per pulse.<br />

+


102 7. Spectral diagnostics: Intensity & Energy<br />

It is important to note that, since <strong>the</strong> detected intensity depends <strong>on</strong> <strong>the</strong> crosssecti<strong>on</strong><br />

1 for photo-i<strong>on</strong>izati<strong>on</strong> <strong>the</strong> sensitivity of <strong>the</strong> detecti<strong>on</strong> scheme varies str<strong>on</strong>gly<br />

with phot<strong>on</strong> energy. This is an issue to c<strong>on</strong>sider when this type of detector is to<br />

be used for hard X-rays, where cross-secti<strong>on</strong>s are several orders of magnitude lower<br />

than in <strong>the</strong> VUV/soft X-ray regi<strong>on</strong>. The sensitivity of <strong>the</strong> scheme when it comes to<br />

<strong>the</strong> detecti<strong>on</strong> of intensities of higher harm<strong>on</strong>ics in <strong>the</strong> free electr<strong>on</strong> laser pulses are<br />

<strong>the</strong>refore also impaired.<br />

Photoluminescence detecti<strong>on</strong><br />

At <strong>the</strong> Lcls a gas m<strong>on</strong>itor detector have been developed that detects <strong>the</strong> fluorescence<br />

of nitrogen with two photomultiplier tubes[96, 154].<br />

Magnetic coils around a 30 cm l<strong>on</strong>g <strong>and</strong> 8 cm wide pressure chamber (with 0.015-<br />

0.9 mbar of gas pressure) c<strong>on</strong>fine <strong>the</strong> photoelectr<strong>on</strong>s produced by <strong>the</strong> free electr<strong>on</strong><br />

laser <strong>beam</strong>, <strong>the</strong>se electr<strong>on</strong>s excite <strong>the</strong> surrounding nitrogen gas. The de-excitati<strong>on</strong> of<br />

<strong>the</strong> molecules occur by <strong>the</strong> emissi<strong>on</strong> of phot<strong>on</strong>s in <strong>the</strong> UV range between 300 to 400<br />

nm. During <strong>the</strong> course of detecti<strong>on</strong> of cosmic rays it have been c<strong>on</strong>cluded that <strong>the</strong><br />

yield in this spectrum depends weakly <strong>on</strong> <strong>the</strong> exciting electr<strong>on</strong> energy.<br />

Calorimeters<br />

A calorimeter (also called radiometer in <strong>the</strong> present c<strong>on</strong>text <strong>and</strong> will henceforth be<br />

denoted thusly) measure, in principle, <strong>the</strong> integral spectral power impinging <strong>on</strong> it.<br />

Hence, <strong>the</strong> sensitivity is equal for <strong>the</strong> harm<strong>on</strong>ic c<strong>on</strong>tent in <strong>the</strong> pulses. St<strong>and</strong>ard<br />

for measuring radiated power from UV/X-ray lightsources around <strong>the</strong> world (i.e. at<br />

PTB[155] in Germany, NIST[156] in <strong>the</strong> USA, <strong>and</strong> NMIJ[157, 158] in Japan).<br />

The device c<strong>on</strong>sists of a cryogenically cooled temperature sensor in fr<strong>on</strong>t of a target<br />

up<strong>on</strong> which <strong>the</strong> radiati<strong>on</strong> impinges <strong>on</strong> (Figure 7.3). The pulse energy is given by:<br />

E = ∆T<br />

s·f<br />

(7.2)<br />

where <strong>the</strong> temperature difference (∆T) is <strong>the</strong> temperature raise in <strong>the</strong> absorbing<br />

cavity, f is <strong>the</strong> repetiti<strong>on</strong> rate of <strong>the</strong> source <strong>and</strong> s is <strong>the</strong> <strong>the</strong>rmal resp<strong>on</strong>se of <strong>the</strong><br />

system. The radiated power is <strong>the</strong>n simply P = E ·f.<br />

The <strong>the</strong>rmal resp<strong>on</strong>se of <strong>the</strong> system can have more or less inertia, but <strong>the</strong> time<br />

c<strong>on</strong>stant for transiti<strong>on</strong> from <strong>on</strong>e equilibrium of <strong>the</strong> system to ano<strong>the</strong>r up<strong>on</strong> changing<br />

absorbed power is measured in minutes. Hence, average energies per pulse can be obtained,<br />

however with small systematic errors (dominated by <strong>the</strong> intensity fluctuati<strong>on</strong>s<br />

of <strong>the</strong> radiati<strong>on</strong>).<br />

A radiometer of this kind is also effectively a <strong>beam</strong> stopper <strong>and</strong> needs to be placed<br />

at <strong>the</strong> very end of <strong>the</strong> <strong>beam</strong> <strong>transport</strong> system, i.e. after <strong>the</strong> experiment(s) or operate<br />

at a split of branch of <strong>the</strong> <strong>beam</strong>. This type of diagnostic can <strong>the</strong>refore be used as a<br />

st<strong>and</strong>ard which to calibrate o<strong>the</strong>r intensity diagnostics against, for instance <strong>the</strong> gas<br />

m<strong>on</strong>itor detectors menti<strong>on</strong>ed above, or during commissi<strong>on</strong>ing runs <strong>and</strong> as a machine<br />

diagnostic.<br />

1 Cross-secti<strong>on</strong>s for <strong>the</strong> noble gases have been tabulated by K. Tiedtke <strong>and</strong> co-workers for <strong>the</strong><br />

noble gases up to 300 eV phot<strong>on</strong> energy[153].


7.2. Intensity/Beam energy 103<br />

Solid state devices<br />

Photodiodes<br />

Temp. sensor<br />

Liquid He<br />

Liquid N2<br />

Heater<br />

Figure 7.3: Illustrati<strong>on</strong> of a cryogenic calorimeter setup.<br />

Phot<strong>on</strong> <strong>beam</strong><br />

Photodiodes, comm<strong>on</strong>ly employed at synchrotr<strong>on</strong> laboratories for intensity diagnostic<br />

purposes have shown to exhibit quite large errors when employed at free electr<strong>on</strong><br />

laser, at Scss in Japan <strong>on</strong>e such measurements yielded an error of about 30% of <strong>the</strong><br />

intensity.<br />

Bolometers<br />

A solid state bolometric sensor that can withst<strong>and</strong> <strong>the</strong> power from a free electr<strong>on</strong><br />

laser laser <strong>beam</strong> can be realized from a device where <strong>the</strong> resistance change of a colossal<br />

magneto-resistance <strong>the</strong>rmistor film up<strong>on</strong> absorpti<strong>on</strong> of electromagnetic radiati<strong>on</strong>.<br />

With a sufficiently efficient coupling to a cooling substrate <strong>the</strong> operati<strong>on</strong> can be fast<br />

enough to allow very precise measurement of <strong>the</strong> <strong>beam</strong> energy <strong>on</strong> a pulse to pulse<br />

basis <strong>–</strong> since <strong>the</strong>y can be operated close to <strong>the</strong>ir metal-insulator transiti<strong>on</strong> where <strong>the</strong><br />

resistance changes are verylarge. At low temperatures <strong>the</strong>c<strong>on</strong>tributi<strong>on</strong>s from <strong>the</strong>rmal<br />

noise in <strong>the</strong> circuit is also lower.<br />

The resistivity <strong>and</strong> <strong>the</strong> temperature of such <strong>the</strong>rmistor films can be varied depending<br />

<strong>on</strong> <strong>the</strong>ir compositi<strong>on</strong>[159]. The sensitivity of <strong>the</strong> resistance up<strong>on</strong> temperature<br />

changes (understood as 1<br />

∂R<br />

R ∂T<br />

) can be as large as 10%/K.<br />

Assuming that <strong>the</strong> sensor is efficiently coupled to <strong>the</strong> cooling substrate <strong>the</strong>n <strong>the</strong><br />

current signal S from a voltage biased sensor is proporti<strong>on</strong>al to <strong>the</strong> total X-ray energy<br />

<strong>and</strong> inversely proporti<strong>on</strong>al to <strong>the</strong> heat capacity of <strong>the</strong> substrate C:<br />

∆T ∝ Efel<br />

C<br />

(7.3)


104 7. Spectral diagnostics: Intensity & Energy<br />

Assuming that <strong>the</strong> noise in <strong>the</strong> circuit mainly arises from <strong>the</strong>rmal noise (i.e. Johns<strong>on</strong><br />

noise) of <strong>the</strong> sensor <strong>and</strong> <strong>the</strong> readout (in,en) we can write <strong>the</strong> signal to noise ratio<br />

(S/N) during an integrati<strong>on</strong> time τ[159]:<br />

S √ Vbias<br />

τ =<br />

N R2 ∂R<br />

∂T �<br />

4kBT/R+i 2<br />

n +(en/R) 2∆(t)√ τ (7.4)<br />

For low resistances <strong>and</strong> low temperatures this expressi<strong>on</strong> is maximized. Low noise<br />

operati<strong>on</strong>al amplifers have en as low as 1 nV/ √ Hz <strong>and</strong> in in <strong>the</strong> orders of pA/ √ Hz.<br />

Signal to noise ratios of 100 000 can be obtained for integrati<strong>on</strong> times of millisec<strong>on</strong>ds.<br />

Neodynium str<strong>on</strong>tium manganese oxide (NSMO) sensors grown <strong>on</strong> a silic<strong>on</strong>e substrate<br />

can be used to measure <strong>the</strong> total energy <strong>on</strong> a pulse to pulse basis with very<br />

small errors[96, 160].<br />

Even-though this diagnostic interrupts <strong>the</strong> <strong>beam</strong> (which can be worked around as<br />

seen earlier) <strong>the</strong> devices can be absolutely calibrated with an ordinary pulsed laser.<br />

Lastly this kind of detectors have a sensitivity that is linear over three orders of<br />

magnitude of incoming <strong>beam</strong> energy. This kind of detectors are currently in use (<strong>and</strong><br />

being fur<strong>the</strong>r developed) at <strong>the</strong> Lcls free electr<strong>on</strong> laserwhere <strong>the</strong>y also serve as to<br />

calibrate <strong>the</strong> photoluminescence detectors described above.<br />

At 8.3 keV <strong>the</strong> average energy is determined with a pulse to pulse jitter of about<br />

8%[96].<br />

7.3 Phot<strong>on</strong>-energy<br />

Measuring <strong>the</strong> phot<strong>on</strong>-pulse directly, as outlined above in secti<strong>on</strong> 7.1, needs <strong>the</strong> <strong>beam</strong><br />

to be split (in amplitude or in frequency) or to be placed at <strong>the</strong> very end after an<br />

experiment <strong>–</strong> that, in turn, needs to be transparent to <strong>the</strong> diagnostic. In various<br />

circumstances this is not desirable or even possible <strong>–</strong> ei<strong>the</strong>r because <strong>the</strong> experiment<br />

is phot<strong>on</strong>-hungry <strong>and</strong> can not afford to use <strong>the</strong> loss of transmissi<strong>on</strong> from a <strong>beam</strong><br />

splitting device, <strong>and</strong>/or <strong>the</strong> experiment is opaque (i.e. effectively a <strong>beam</strong>-stopper).<br />

One way to circumvent this is to measure <strong>the</strong> time-of-flight of <strong>the</strong> i<strong>on</strong>s <strong>and</strong>/or electr<strong>on</strong>s<br />

in amannersimilar to<strong>the</strong>intensitymeasurement with <strong>the</strong> gas-m<strong>on</strong>itor detectors<br />

described above (secti<strong>on</strong> 7.2). Since <strong>the</strong> arrival time of <strong>the</strong> pulses at <strong>the</strong> diagnostic is<br />

known (or can be inferred from o<strong>the</strong>r diagnostics) we can measure different aspects<br />

of <strong>the</strong> i<strong>on</strong>izati<strong>on</strong> processes occurring in, for instance, a rare gas via <strong>the</strong> time-of-flight<br />

of <strong>the</strong> produced charged particles in a homogeneous electric (<strong>and</strong>/or magnetic) field.<br />

I<strong>on</strong> time-of-flight<br />

The time of flight of i<strong>on</strong>s through an electric field is proporti<strong>on</strong>al to <strong>the</strong>ir mass <strong>and</strong><br />

charge <strong>–</strong> for a small source regi<strong>on</strong> <strong>the</strong> flight times for different charge to mass ratios<br />

are deterministic <strong>and</strong> depends <strong>on</strong>ly <strong>on</strong> electric field strengths <strong>and</strong> <strong>the</strong> geometry of <strong>the</strong><br />

apparatus, usually referred to as a Wiley-MacLaren spectrometer[161]. I<strong>on</strong> time-offlight<br />

is thus a mass-spectroscopic measurement, in essence.<br />

The resoluti<strong>on</strong> of a Wiley-Maclaren setup (spatial <strong>and</strong> temporal) can be analyzed,<br />

let us spell it out in detail: define<br />

qsEs +dEd<br />

Ut = , k =<br />

sEs<br />

Ut<br />

qsEs


7.3. Phot<strong>on</strong>-energy 105<br />

s<br />

d0,E1 d1,E2 D<br />

Figure 7.4: The field in <strong>the</strong> drift tube is zero. The ratio between <strong>the</strong> fields is uniquely determined by <strong>the</strong><br />

geometry given by s,d1 <strong>and</strong> D.<br />

With T as <strong>the</strong> total flight time, zero initial kinetic energy <strong>and</strong> starting positi<strong>on</strong> s,<br />

<strong>the</strong> positi<strong>on</strong> where i<strong>on</strong>s having s± 1<br />

δs pass each o<strong>the</strong>r can be found where<br />

2<br />

�<br />

dT �<br />

� = 0 ⇒ D = 2sk<br />

ds<br />

3/2<br />

�<br />

1<br />

1−<br />

k + √ �<br />

d<br />

k s<br />

� 0,s<br />

This focus c<strong>on</strong>diti<strong>on</strong> is <strong>the</strong> same for all i<strong>on</strong>s <strong>and</strong> is independent of <strong>the</strong> systems total<br />

energy. Hence, if s, d, <strong>and</strong> D is given, <strong>the</strong> ratio Ed/Es since k can <strong>on</strong>ly have <strong>on</strong>e<br />

physically reas<strong>on</strong>able value.<br />

T(U,s) has ei<strong>the</strong>r a maximum, minimum, or an inflecti<strong>on</strong> point where dT<br />

�<br />

� = 0, ds 0,s<br />

<strong>the</strong> latter can be found with:<br />

d 2 T<br />

ds2 �<br />

�<br />

� = 0,⇒ d<br />

s =<br />

� �<br />

k −3 D<br />

k 2s<br />

� 0,s<br />

For <strong>the</strong> parameter values chosen for best resoluti<strong>on</strong>, two-field systems are operated<br />

smaller than <strong>the</strong> righth<strong>and</strong>-side.<br />

at <strong>the</strong> maxiumum, i.e. with <strong>the</strong> d<br />

s<br />

Spatial resoluti<strong>on</strong><br />

Let Ms be <strong>the</strong> largest mass for which <strong>the</strong> flight times are discernibly different,<br />

Tm+1 −Tm = τ ≈ Tm<br />

�<br />

s<br />

�2 ⇒ Ms ≈ 16k<br />

2m ∆s<br />

The latter is true whenever k ≫ 1 <strong>and</strong> k ≫ d/s.<br />

This shows that <strong>the</strong> distance to <strong>the</strong> sourcepoint s must be larger than <strong>the</strong> spread<br />

∆s for space resoluti<strong>on</strong> to be made adequate <strong>–</strong> this, in turn, is dependent <strong>on</strong> D<br />

being large since D/s determines k. Increasing d (thus increasing k) improves space<br />

resoluti<strong>on</strong>.


106 7. Spectral diagnostics: Intensity & Energy<br />

Energy resoluti<strong>on</strong><br />

C<strong>on</strong>sider two i<strong>on</strong>s at <strong>the</strong> same initial positi<strong>on</strong> s but with opposite velocities al<strong>on</strong>g <strong>the</strong><br />

spectrometer axis, equal in magnitude, <strong>the</strong>n <strong>the</strong> turn-around time is<br />

∆T = 1.02 2√ 2mU0<br />

qEs<br />

Then <strong>the</strong> maximum resolvable mass becomes, with D/s given by <strong>the</strong> focusing c<strong>on</strong>diti<strong>on</strong>:<br />

ME = 1<br />

� � √<br />

Ut k +1 k +1<br />

√ −<br />

4 U0 k k + √ �<br />

d<br />

k s<br />

To find <strong>the</strong> overall resoluti<strong>on</strong> a compromise between Ms <strong>and</strong> ME has to be found.<br />

Higherorderfocussingcanbeachievedbyhavingadditi<strong>on</strong>aldrift-secti<strong>on</strong>s<strong>and</strong>fields[162]<br />

<strong>–</strong> for those <strong>the</strong> flight times, etc. can still be calculated analytically[163].<br />

I<strong>on</strong> time-of-flight as a diagnostic<br />

A precise way of determining <strong>the</strong> charge to mass ratio of photo-i<strong>on</strong>s allows us to<br />

measure <strong>the</strong> phot<strong>on</strong>-energy. The photoi<strong>on</strong>izati<strong>on</strong> cross-secti<strong>on</strong> depends <strong>on</strong> phot<strong>on</strong><br />

energy, i.e. <strong>on</strong>e can measure it directly via <strong>the</strong> intensity of <strong>on</strong>e charged state; a much<br />

more stable measurement of<strong>the</strong>phot<strong>on</strong>energyis obtained via<strong>the</strong>ratiobetween singly<br />

<strong>and</strong> doubly charged i<strong>on</strong>s (or higher charge states). All this assumes that <strong>the</strong> crosssecti<strong>on</strong>s<br />

as functi<strong>on</strong>s of energy <strong>and</strong> charge state are known beforeh<strong>and</strong>. Moreover, <strong>the</strong><br />

phot<strong>on</strong>-density must be low enough to ensure that <strong>on</strong>ly single-phot<strong>on</strong> events occur <strong>–</strong><br />

something which is usually fulfilled for an unfocussed free electr<strong>on</strong> laser <strong>beam</strong>[153].<br />

Juranić et al. have used a Wiley-MacLaren spectrometer to measure <strong>the</strong> phot<strong>on</strong><br />

energyatFlashwithuncertaintiesstayingbelow1%upto150eVphot<strong>on</strong>energy[164].<br />

I<strong>on</strong> time-of-flight measurements can be shielded from perturbing magnetic fields<br />

easier, <strong>and</strong> provides higher countrates than <strong>the</strong> electr<strong>on</strong> measurements described below.<br />

The time-of-flight times are also l<strong>on</strong>ger since <strong>the</strong> particles have higher mass<br />

which makes <strong>the</strong> measurement less dem<strong>and</strong>ing with respect to <strong>the</strong> data-acquisiti<strong>on</strong>.<br />

Electr<strong>on</strong> time-of-flight<br />

The time-of-flight for <strong>the</strong> electr<strong>on</strong>s can be c<strong>on</strong>verted into kinetic energy with <strong>the</strong> aid<br />

of known i<strong>on</strong>izati<strong>on</strong> energies of atomic states <strong>–</strong> hence this is a spectroscopic measurement.<br />

If performed at high enough resoluti<strong>on</strong> electr<strong>on</strong> time-of-flight can thus give<br />

informati<strong>on</strong> <strong>on</strong> <strong>the</strong> spectral c<strong>on</strong>tent of <strong>the</strong> pulse as well as <strong>the</strong> mean energy. Potentially<br />

this spectrum also give access to <strong>the</strong> pulse length[165] as discussed in chapter 9.<br />

This allows for measurement of <strong>the</strong> phot<strong>on</strong>-energy <strong>and</strong> <strong>the</strong> fluctuati<strong>on</strong>s of it, as<br />

have been d<strong>on</strong>e near 93 eV at Flash[150]. The resoluti<strong>on</strong> in <strong>the</strong> spectra for <strong>the</strong> He<br />

1s photoelectr<strong>on</strong> line was about 100 meV.<br />

The limitati<strong>on</strong> of this photo-electr<strong>on</strong> spectroscopic measurement is <strong>the</strong> relatively<br />

low countrate (in <strong>the</strong> order of thous<strong>and</strong>s of electr<strong>on</strong>s per pulse). The kinetic energy<br />

of <strong>the</strong> electr<strong>on</strong>s is also not invariant with changing intensity of <strong>the</strong> free electr<strong>on</strong><br />

laser <strong>beam</strong>, where <strong>the</strong> attractive forced from <strong>the</strong> positively charged i<strong>on</strong>s cases <strong>the</strong><br />

kinetic energy to be lowered for <strong>the</strong> electr<strong>on</strong>s <strong>–</strong> which, if not accounted for, induces<br />

an apparent decrease in phot<strong>on</strong>-energy with increasing phot<strong>on</strong>-densities.


7.3. Phot<strong>on</strong>-energy 107<br />

However, <strong>the</strong> lower cross-secti<strong>on</strong> at higher energies lowers <strong>the</strong> Coulomb attracti<strong>on</strong><br />

betweeni<strong>on</strong>s<strong>and</strong>electr<strong>on</strong>s, since<strong>the</strong>numberofelectr<strong>on</strong>-i<strong>on</strong>pairs(forsingle-i<strong>on</strong>ziati<strong>on</strong><br />

events) is:<br />

Nparis = Nphot<strong>on</strong>s ·ngas ·σgas ·ℓ (7.5)<br />

where ℓ is <strong>the</strong> length of <strong>the</strong> interacti<strong>on</strong> regi<strong>on</strong> <strong>and</strong> n = p/kBT for an ideal gas. The<br />

n<strong>on</strong>-ideality of <strong>the</strong> gas can be accounted for via a correcti<strong>on</strong> using <strong>the</strong> mean-free<br />

path[150].<br />

For higher phot<strong>on</strong>-energies thus, electr<strong>on</strong> time-of-flight measurements can be a<br />

viable opti<strong>on</strong> for determining <strong>the</strong> phot<strong>on</strong>-energy <strong>and</strong> its fluctuati<strong>on</strong>s.


108 7. Spectral diagnostics: Intensity & Energy<br />

Summary<br />

• The <strong>beam</strong> energy (or intensity) is a critical parameter for user<br />

experiments <strong>and</strong> o<strong>the</strong>r diagnostics.<br />

• Gas m<strong>on</strong>itor detectors analyze <strong>the</strong> total yield occurring from<br />

photoi<strong>on</strong>izati<strong>on</strong> (which is proporti<strong>on</strong>al to <strong>the</strong> <strong>beam</strong> energy) of<br />

a target gas by accelerating <strong>the</strong>m in an electric (<strong>and</strong> possibly<br />

an magnetic field). This is a diagnostic that is transparent.<br />

• Ano<strong>the</strong>r transparent diagnostic is photoluminescence measurements<br />

of, for instance, nitrogen gas.<br />

• Calorimeters (Radiometers) is a measurement that gives <strong>the</strong><br />

average intensity of <strong>the</strong> pulses in a very exact manner. This<br />

is an opaque diagnostic which can also be used to calibrate<br />

o<strong>the</strong>r intensity diagnostics.<br />

• Bolometers provides a measurement of <strong>the</strong> <strong>beam</strong> energy over<br />

a large phot<strong>on</strong> range <strong>–</strong> for instance with 8% error at 8.3 keV<br />

phot<strong>on</strong> energy at <strong>the</strong> Lcls.<br />

• Diodes have shown to have ra<strong>the</strong>r large errors when used at<br />

free electr<strong>on</strong> laser, e.g. 30% at <strong>the</strong> Scss.<br />

• Bolometers <strong>and</strong> diodes are opaque diagnostics <strong>and</strong> thus require<br />

splitting or a transparent user experiment.<br />

• X-ray spectrometry measures <strong>the</strong> dispersed fel <strong>beam</strong> from a<br />

(VLS) diffractive grating. Thus obtaining <strong>the</strong> full spectrum<br />

of <strong>the</strong> free electr<strong>on</strong> laser <strong>beam</strong> <strong>on</strong> a shot to shot basis.<br />

• I<strong>on</strong> time-of-flight measurements diagnoses <strong>the</strong> phot<strong>on</strong>-energy<br />

of <strong>the</strong> <strong>beam</strong> with relatively small errors <strong>–</strong> assuming that <strong>the</strong><br />

cross-secti<strong>on</strong>s for <strong>the</strong> i<strong>on</strong>ized gas is known as a functi<strong>on</strong> of<br />

energy <strong>and</strong> <strong>the</strong> various charged states.<br />

• Electr<strong>on</strong> time-of-flight can provide a measurement of <strong>the</strong> spectrum<br />

of <strong>the</strong> pulse <strong>and</strong> provides a way of discerning <strong>the</strong> fluctuati<strong>on</strong>s<br />

in phot<strong>on</strong>-energy.<br />

• Electr<strong>on</strong> <strong>and</strong> i<strong>on</strong> time-of-flight measurements that are transparent<br />

to <strong>the</strong> user’s experiments.<br />

• Time-of-flight measurements utilize photoi<strong>on</strong>izati<strong>on</strong> for which<br />

<strong>the</strong> cross-secti<strong>on</strong>s drops fast after 100-200 eV <strong>–</strong> i.e. for higher<br />

phot<strong>on</strong>-energies <strong>the</strong>y may be impractical due to low countrates.


8. Beam cross-secti<strong>on</strong> diagnostics<br />

The material presented here in this chapter is adapted from ”Survey of diagnostic<br />

techniques for measuring <strong>the</strong> <strong>beam</strong> cross-secti<strong>on</strong> of ultra-short phot<strong>on</strong><br />

pulses” by M. A. Bowler, A. J. Glees<strong>on</strong> <strong>and</strong> M. D. Roper. Iruvx WP7, 2010<br />

by A. Lindblad.<br />

8.1 Introducti<strong>on</strong><br />

By measuring <strong>the</strong> profile of <strong>the</strong> generated phot<strong>on</strong>-<strong>beam</strong> many important parameters<br />

important for <strong>the</strong> <strong>beam</strong> <strong>transport</strong> towards <strong>the</strong> experiments can be learned. The<br />

<strong>beam</strong>-profile is also an important diagnostic for <strong>the</strong> machine <strong>and</strong> as an parameter for<br />

experiments (e.g. <strong>the</strong> spot-size determines <strong>the</strong> energy density at <strong>the</strong> experiment).<br />

The specific informati<strong>on</strong> that is required is:<br />

• The transverse intensity distributi<strong>on</strong> of <strong>the</strong> pulse <strong>–</strong> allowing <strong>the</strong> source size,<br />

positi<strong>on</strong> <strong>and</strong> quality factor (M 2 ) to be calculated from <strong>the</strong> sec<strong>on</strong>d moment 1 .<br />

• The centroid (first moment) of <strong>the</strong> transverse pulse profile <strong>–</strong> giving <strong>beam</strong> positi<strong>on</strong><br />

<strong>and</strong> (in combinati<strong>on</strong> with a sec<strong>on</strong>d measurement of <strong>the</strong> same pulse at<br />

different l<strong>on</strong>gitudinal distance) <strong>the</strong> <strong>beam</strong> angle.<br />

• The pulse wavefr<strong>on</strong>t <strong>–</strong> giving a complete descripti<strong>on</strong> of <strong>the</strong> spatial properties of<br />

<strong>the</strong> <strong>beam</strong>.<br />

• The focused spot size of <strong>the</strong> <strong>beam</strong> <strong>–</strong> critical for optimizing adaptive mirrors to<br />

give <strong>the</strong> best focus<strong>and</strong> for achieving high fluences for n<strong>on</strong>-linear studies.<br />

The ideal cross-secti<strong>on</strong> diagnostic<br />

Any measurement that gives a full <strong>beam</strong> profile can be used to calculate <strong>the</strong> <strong>beam</strong><br />

centroid but a measurement that gives <strong>on</strong>ly centroid informati<strong>on</strong> gives no profile<br />

informati<strong>on</strong>. Centroiding measurementsare never<strong>the</strong>less useful where <strong>the</strong>y are less<br />

1 The quality factor for a diffracti<strong>on</strong> limited gaussian phot<strong>on</strong> <strong>beam</strong> at wavelength lambda is<br />

λ/π. For a real <strong>beam</strong> <strong>the</strong> product of <strong>the</strong> minimum waist-size of <strong>the</strong> <strong>beam</strong> with <strong>the</strong> divergence of<br />

<strong>the</strong> <strong>beam</strong> in <strong>the</strong> far field is called <strong>the</strong> <strong>beam</strong> parameter product (B<strong>PP</strong>). The ratio between <strong>the</strong> B<strong>PP</strong><br />

<strong>and</strong> λ/π is <strong>the</strong> quality measure M 2 . For a diffracti<strong>on</strong> limited <strong>beam</strong> this ratio is unity; for real<br />

<strong>beam</strong>s this measure is larger than unity.<br />

109


110 8. Beam cross-secti<strong>on</strong> diagnostics<br />

invasive or can give pulse by pulse informati<strong>on</strong>. For a free electr<strong>on</strong> laser source, <strong>the</strong><br />

ideal diagnostic would have <strong>the</strong> following properties:<br />

• A sub-µm spatial resoluti<strong>on</strong> to allow <strong>the</strong> focused <strong>beam</strong> to be measured.<br />

• A field of view of ∼ 1 cm to allow <strong>the</strong> unfocused <strong>beam</strong> to be measured.<br />

• N<strong>on</strong>-invasive <strong>–</strong> give negligible disrupti<strong>on</strong> to <strong>the</strong> <strong>beam</strong>.<br />

• Sensitive <strong>–</strong> to allow a single pulse to be measured.<br />

• Wide dynamic range with linear resp<strong>on</strong>se <strong>–</strong> to measure attenuated <strong>and</strong> fullpower<br />

<strong>beam</strong>s, focused <strong>and</strong> unfocused, fundamental <strong>and</strong> harm<strong>on</strong>ics without damage,<br />

or excessive noise.<br />

• Large b<strong>and</strong>width <strong>–</strong> to measure every pulse up to MHz repetiti<strong>on</strong> rates in real<br />

time.<br />

• Broadb<strong>and</strong> <strong>–</strong> work at THz/IR or from <strong>the</strong> VUV to hard X-rays.<br />

• Spectrally discriminating <strong>–</strong> to separate signals from e.g. fundamental <strong>and</strong> harm<strong>on</strong>ics.<br />

The numberof <strong>the</strong>seproperties thatagiven practical m<strong>on</strong>itor will needwill depend<br />

<strong>on</strong> <strong>the</strong>particular applicati<strong>on</strong> or locati<strong>on</strong> in <strong>the</strong>phot<strong>on</strong> <strong>transport</strong> system. For example,<br />

sub-µm resoluti<strong>on</strong> is <strong>on</strong>ly required to measure <strong>the</strong> size of <strong>the</strong> <strong>beam</strong> at a focus, it will<br />

not be necessary for a m<strong>on</strong>itor to be able to wavelength-discriminate if it is situated<br />

after a m<strong>on</strong>ochromator, <strong>and</strong> diagnostics situated after <strong>the</strong> experiment do not need to<br />

be n<strong>on</strong>-invasive.<br />

Distributi<strong>on</strong> of diagnostics al<strong>on</strong>g <strong>the</strong> <strong>beam</strong><br />

feedback<br />

to source<br />

Source<br />

On-line centroid m<strong>on</strong>itors for<br />

<strong>beam</strong> positi<strong>on</strong> <strong>and</strong> angle<br />

Pro�lin m<strong>on</strong>itors for<br />

quality factor analysis<br />

Centroid m<strong>on</strong>itors for<br />

mirror alignment<br />

Beamline optics<br />

Pro�ling m<strong>on</strong>itor for<br />

commisi<strong>on</strong>ing<br />

Focus spot-size<br />

m<strong>on</strong>itor<br />

Wavefr<strong>on</strong>t<br />

sensor<br />

Figure 8.1: A schematic <strong>on</strong> how to distribute various diagnostics to determin transverse <strong>beam</strong> properties.<br />

Figure 8.1 shows schematically how <strong>the</strong> various diagnostics might be distributed<br />

al<strong>on</strong>g a <strong>beam</strong>line. Before any optics it will be necessary to have n<strong>on</strong>-invasive (or<br />

minimally invasive) ”<strong>on</strong>-line” diagnostics that give at least <strong>beam</strong> positi<strong>on</strong> <strong>and</strong> angle<br />

informati<strong>on</strong> c<strong>on</strong>tinuously <strong>and</strong> individually for every pulse. These m<strong>on</strong>itors could<br />

be used in feedback c<strong>on</strong>trol of <strong>the</strong> machine to ensure <strong>the</strong> phot<strong>on</strong> <strong>beam</strong> is stable in<br />

positi<strong>on</strong> <strong>and</strong> angle. There should also be diagnostics that can give a full pulse profile<br />

at three or more positi<strong>on</strong>s such that <strong>the</strong> source positi<strong>on</strong> <strong>and</strong> size <strong>and</strong> <strong>beam</strong> quality<br />

factor M 2 <strong>and</strong> can be determined[166]. The main applicati<strong>on</strong> here would be during<br />

commissi<strong>on</strong>ing <strong>and</strong> so <strong>the</strong> m<strong>on</strong>itors could be invasive, though if that were case it


8.1. Introducti<strong>on</strong> 111<br />

would not be possible to calculate <strong>the</strong> quality factor for a single pulse. Therefore a<br />

n<strong>on</strong>-invasive measurement is preferred, but it need not work at <strong>the</strong> full rep rate of<br />

<strong>the</strong> free electr<strong>on</strong> laser since statistical methods can be used to infer <strong>the</strong> overall <strong>beam</strong><br />

quality.<br />

After each optical element a <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor is required to ensure <strong>the</strong> <strong>beam</strong><br />

alignment is correct. These might just give centroid informati<strong>on</strong> <strong>and</strong> may average<br />

over a number of pulses. Ideally, <strong>the</strong>y would be <strong>on</strong>-line so <strong>the</strong> <strong>beam</strong> stability can be<br />

c<strong>on</strong>stantly m<strong>on</strong>itored, though this may be unnecessary. Additi<strong>on</strong>al profiling m<strong>on</strong>itors<br />

will also be available specifically for commissi<strong>on</strong>ing <strong>and</strong> diagnosing problems; <strong>the</strong>se<br />

can be invasive.<br />

At <strong>the</strong> experiment it is necessary to determine <strong>the</strong> positi<strong>on</strong> <strong>and</strong> quality of <strong>the</strong> <strong>beam</strong><br />

focus to:<br />

• Ensure multiple <strong>beam</strong>s can be overlapped spatially<br />

• Optimise adaptive mirrors<br />

• Achieve <strong>the</strong> highest <strong>beam</strong> fluence<br />

• Positi<strong>on</strong> <strong>the</strong> sample at <strong>the</strong> focus<br />

It would generally be acceptable for <strong>the</strong>se measurements to be invasive since, with<br />

a stable source, <strong>the</strong>y would normally <strong>on</strong>ly be needed during commissi<strong>on</strong>ing <strong>and</strong> experimental<br />

set-up. Depending <strong>on</strong> need, <strong>the</strong> measurements could ei<strong>the</strong>r be averaged<br />

over many pulses (e.g. for positi<strong>on</strong>ing) or measure just a single pulse (e.g. focus quality).<br />

If a measure of every pulse is needed outside <strong>the</strong> realm of commissi<strong>on</strong>ing <strong>the</strong>n<br />

a n<strong>on</strong>invasive detector or a detector remote from <strong>the</strong> sample positi<strong>on</strong> is needed. The<br />

latter is quite easy with gas-phase experiments (which are almost transparent to <strong>the</strong><br />

<strong>beam</strong>) since <strong>the</strong> detector can be placed after <strong>the</strong> experiment. It would not be possible<br />

to measure <strong>the</strong> focus directly, so <strong>the</strong> wavefr<strong>on</strong>t would have to be measured <strong>and</strong><br />

reverse-propagated to rec<strong>on</strong>struct <strong>the</strong> <strong>beam</strong> focus.<br />

Direct measurement of <strong>the</strong> wavefr<strong>on</strong>t would be <strong>the</strong> optimal way to measure <strong>the</strong><br />

<strong>beam</strong> as it allows for simulated propagati<strong>on</strong> to arbitrary positi<strong>on</strong>s al<strong>on</strong>g <strong>the</strong> <strong>transport</strong><br />

path, e.g. back to <strong>the</strong> source. However, wavefr<strong>on</strong>t measurement is invasive <strong>and</strong> so<br />

cannot be a general ”<strong>on</strong>-line” diagnostic. Fur<strong>the</strong>rmore, accurate propagati<strong>on</strong> across<br />

an optical element requires detailed informati<strong>on</strong> <strong>on</strong> <strong>the</strong> element surface shape at a<br />

large range of spatial frequencies <strong>and</strong> thus <strong>the</strong> accuracy of any predicti<strong>on</strong> will fall as<br />

<strong>the</strong> simulati<strong>on</strong> traverses more optical elements.<br />

C<strong>on</strong>tent of this chapter<br />

In this chapter, a broad survey of <strong>the</strong> range of techniques that have been used to<br />

profile n<strong>on</strong>-visible phot<strong>on</strong> <strong>beam</strong>s is presented. In <strong>the</strong> VUV to X-ray range, most<br />

existing diagnostics have been developed for use <strong>on</strong> synchrotr<strong>on</strong> radiati<strong>on</strong> sources.<br />

The challenges <strong>the</strong>re tend to be ra<strong>the</strong>r different, <strong>and</strong> <strong>the</strong> techniques may not be easy<br />

to modify for free electr<strong>on</strong> laser use. Specific problems for FEL <strong>beam</strong>s include:<br />

• Damage from ablati<strong>on</strong> ra<strong>the</strong>r than high average power.<br />

• The need for pulse resolved ra<strong>the</strong>r than time averaged measurements.<br />

• The need to avoid <strong>beam</strong> disrupti<strong>on</strong> through coherent diffracti<strong>on</strong> effects.


112 8. Beam cross-secti<strong>on</strong> diagnostics<br />

It is worthwhile to describe briefly several comm<strong>on</strong> techniques that are used in <strong>the</strong><br />

diagnostics as follows:<br />

• Scanning <strong>–</strong> in which a scan through <strong>the</strong> <strong>beam</strong> secti<strong>on</strong> is made, recording <strong>the</strong><br />

intensity in a step-wise manner to build up <strong>the</strong> profile<br />

• Imaging <strong>–</strong> in which <strong>the</strong> entire intensity profile, in <strong>on</strong>e or two dimensi<strong>on</strong>s, is<br />

measured in <strong>on</strong>e shot<br />

• Sampling <strong>–</strong> in which a small part of <strong>the</strong> <strong>beam</strong> is extracted <strong>and</strong> <strong>the</strong> required<br />

informati<strong>on</strong> deduced from this whilst <strong>the</strong> bulk of <strong>the</strong> <strong>beam</strong> passes <strong>on</strong> to <strong>the</strong><br />

experiment<br />

• Replicating <strong>–</strong> in which <strong>the</strong> <strong>beam</strong> profile informati<strong>on</strong> is transferred to ano<strong>the</strong>r<br />

medium <strong>and</strong> that measured to give <strong>the</strong> actual profile.<br />

In some cases a mixture of <strong>the</strong>se techniques is used, for example in imaging a<br />

replica of <strong>the</strong> <strong>beam</strong>.<br />

In <strong>the</strong> THz <strong>and</strong> IR ranges, diagnostics at existing free electr<strong>on</strong> laser sources tend to<br />

belimitedtocharacterizing<strong>the</strong><strong>beam</strong>at<strong>the</strong>experimentra<strong>the</strong>rthanhavingdistributed<br />

diagnostics al<strong>on</strong>g <strong>the</strong> <strong>transport</strong> system. The available means of detecting IR <strong>and</strong> THz<br />

radiati<strong>on</strong> restrict <strong>the</strong> range of diagnostic techniques that can be applied, in particular<br />

because <strong>the</strong> l<strong>on</strong>g wavelength radiati<strong>on</strong> cannot directly i<strong>on</strong>ize materials. For <strong>the</strong> near<br />

<strong>and</strong> mid-IR, techniques can be adapted from <strong>the</strong> visible regime.<br />

In secti<strong>on</strong> 8.4 will be described those techniques that can give a profile of <strong>the</strong><br />

<strong>beam</strong> intensity. In some cases this will be al<strong>on</strong>g a single axis (or two axes with two<br />

instruments situated orthog<strong>on</strong>ally), <strong>and</strong> in o<strong>the</strong>rs a complete 2-dimensi<strong>on</strong>al map of<br />

<strong>the</strong> <strong>beam</strong> intensity will be recorded. Secti<strong>on</strong> 8.10 will describe those techniques that<br />

give <strong>on</strong>ly informati<strong>on</strong> about <strong>the</strong> positi<strong>on</strong> of <strong>the</strong> <strong>beam</strong> centroid (”centre of gravity”,<br />

or first moment, of <strong>the</strong> intensity). How <strong>the</strong> complete wavefr<strong>on</strong>t can be measured is<br />

described in secti<strong>on</strong> 8.11 <strong>and</strong> specific applicati<strong>on</strong>s of <strong>beam</strong> profiling to determine <strong>the</strong><br />

size <strong>and</strong> quality of <strong>the</strong> focused <strong>beam</strong> are described in secti<strong>on</strong> 8.8. Diagnostics for IR<br />

<strong>and</strong> THz wavelengths will be described collectively in secti<strong>on</strong> 8.12.<br />

8.2 Definiti<strong>on</strong>s<br />

In this chapter we will look into various schemes as how to determine <strong>the</strong> transverse<br />

(x,y) distributi<strong>on</strong> of phot<strong>on</strong>s in <strong>the</strong> <strong>beam</strong> 2 . Most of <strong>the</strong> methods also carry over to<br />

more applicati<strong>on</strong>s of particle <strong>beam</strong>s in general.<br />

Let us define <strong>the</strong> problem more specifically, c<strong>on</strong>sider <strong>the</strong> particle <strong>beam</strong> to have a<br />

Gaussian distributi<strong>on</strong> in <strong>the</strong> x,y plane orthog<strong>on</strong>al to <strong>the</strong> directi<strong>on</strong> of <strong>the</strong> <strong>beam</strong> (<strong>the</strong><br />

s directi<strong>on</strong>)<br />

N(x,y) ∝ 1<br />

e<br />

σxσy<br />

−1 2(x 2 /σx+y 2 /σy)<br />

Al<strong>on</strong>g <strong>the</strong> s-axis <strong>the</strong> distributi<strong>on</strong> is ideally a step functi<strong>on</strong>. The measurement of this<br />

profile will be covered more specifically in chapter 9.<br />

Various schemes can be envisi<strong>on</strong>ed <strong>on</strong> how to measure <strong>the</strong> profiles c<strong>on</strong>cerned, each<br />

with its own merits <strong>–</strong> as a first rough divisi<strong>on</strong> we can divide <strong>the</strong>m into invasive <strong>and</strong><br />

n<strong>on</strong>-invasive methods:<br />

2 Diagnostics c<strong>on</strong>cerning <strong>the</strong> l<strong>on</strong>gitudinal (s) extent of <strong>the</strong> pulses will be covered in chapter 9.


8.3. Direct imaging of <strong>the</strong> <strong>beam</strong> 113<br />

y<br />

x<br />

Figure 8.2: Projecti<strong>on</strong> of <strong>the</strong> spatial density distributi<strong>on</strong> in <strong>the</strong> x.y plane (left) <strong>and</strong> <strong>on</strong>to individual coordinateaxes<br />

(right).<br />

invasive <strong>–</strong> or direct measurements<br />

Direct imaging of <strong>the</strong> <strong>beam</strong><br />

Wire grids<br />

Scanning wires, slits, knife-edges, pin-holes<br />

n<strong>on</strong>-invasive <strong>–</strong> or in-direct measurements<br />

rest gas i<strong>on</strong>izati<strong>on</strong><br />

photo dissociati<strong>on</strong><br />

synchrotr<strong>on</strong> light<br />

Compt<strong>on</strong> scattering<br />

8.3 Direct imaging of <strong>the</strong> <strong>beam</strong><br />

The simplest way to image <strong>the</strong> transverse footprint of a phot<strong>on</strong> <strong>beam</strong> is to directly<br />

illuminate an array detector such as a CCD with <strong>the</strong> <strong>beam</strong>[167, 168]. Providing <strong>the</strong><br />

detector is sensitive to <strong>the</strong> phot<strong>on</strong> wavelength, a direct readout of <strong>the</strong> <strong>beam</strong> profile is<br />

achieved.<br />

If <strong>the</strong> array is two-dimensi<strong>on</strong>al, <strong>the</strong>n a complete transverse map of <strong>the</strong> <strong>beam</strong> can<br />

be recorded. The spatial resoluti<strong>on</strong> of a CCD detector is determined, in principle,<br />

by <strong>the</strong> photo-site size <strong>and</strong> spacing in <strong>the</strong> detector array. In practice this limit cannot<br />

be achieved due to diffusi<strong>on</strong>, where electr<strong>on</strong>s created in <strong>on</strong>e pixel are collected in an<br />

adjacent pixel.<br />

To record an image of every free electr<strong>on</strong> laser pulse individually, <strong>the</strong> frame rate<br />

of <strong>the</strong> camera must of course match <strong>the</strong> pulse repetiti<strong>on</strong> rate. Extremely high-speed<br />

cameras are available for optical imaging (e.g. 600,000 fps with <strong>the</strong> NAC Memrecam<br />

GX-8 3 ) <strong>and</strong> commercial high-speed X-ray imaging services are available (e.g. Speed<br />

Visi<strong>on</strong> Technologies Inc. offer frame rates of 1000 fps 4 ). Speeds for scientific applicati<strong>on</strong>s<br />

are typically much lower.<br />

3 www.nacinc.com<br />

4 www.speedvisi<strong>on</strong>tech.com/speedvisi<strong>on</strong>-x-ray.php<br />

s<br />

I<br />

I<br />

x<br />

y


114 8. Beam cross-secti<strong>on</strong> diagnostics<br />

For example, Princet<strong>on</strong> Instruments 5 manufacture<br />

CCD cameras for both direct <strong>and</strong> indirect (q.v.) X-ray<br />

imaging that have readout rates of 2 MHz or 100 kHz<br />

depending <strong>on</strong> <strong>the</strong> sensitivity <strong>and</strong> signal to noise ratio.<br />

This rate equates to roughly <strong>the</strong> rate at which a single<br />

pixel can be read out since all <strong>the</strong> data passes through<br />

a single serial register. Thus <strong>the</strong> frame rate depends<br />

<strong>on</strong> <strong>the</strong> number of pixels. They offer a 512 x 512 pixel<br />

camera (with 13 µm pixel spacing) which could thus<br />

be read out at <strong>on</strong>ly ∼ 7.5 fps.<br />

Never<strong>the</strong>less, it is clear that higher speeds are<br />

achievable, though <strong>the</strong> balance of sensitivity <strong>and</strong> noise<br />

needs to be c<strong>on</strong>sidered. We might <strong>the</strong>refore reas<strong>on</strong>ably<br />

expect to be able to record every pulse at repetiti<strong>on</strong><br />

rates of about 1 kHz. For single pulse imaging at repetiti<strong>on</strong><br />

rates above <strong>the</strong> detector frame rate, <strong>the</strong> camera<br />

would need to be gated to record just <strong>on</strong>e pulse in <strong>the</strong><br />

Figure 8.3: A typical scintillator<br />

screen setup.<br />

frame cycle. Gating at <strong>the</strong> nanosec<strong>on</strong>d level is available with optical cameras, but is<br />

presumably achieved using electro-optical shutters which are not available for X-rays.<br />

A mechanical shutter would be needed to pick just <strong>on</strong>e pulse <strong>and</strong> this would be also<br />

be challenging to design <strong>and</strong> synchr<strong>on</strong>ize.<br />

The high fluence of a free electr<strong>on</strong> laser pulse also raises significant c<strong>on</strong>cerns over<br />

ablati<strong>on</strong>, radiati<strong>on</strong> damage, <strong>and</strong> linearity / saturati<strong>on</strong> effects. The saturati<strong>on</strong> level<br />

is a compromise with spatial resoluti<strong>on</strong> since if <strong>the</strong> pixels are made smaller <strong>the</strong>n<br />

<strong>the</strong>y can hold less charge <strong>and</strong> will saturate earlier. In fact, CCD detectors are not<br />

often used to measure <strong>the</strong> direct <strong>beam</strong> <strong>on</strong> synchrotr<strong>on</strong> sources because <strong>the</strong>y saturate<br />

too easily. Fedotov in 2000[169] reported that an ”old” 1200LC1 type CCD would<br />

saturate with 500 phot<strong>on</strong>s at 10 keV compared with 5-50 phot<strong>on</strong>s for ”more modern”<br />

CCDs. Therefore, a c<strong>on</strong>venti<strong>on</strong>al CCD is limited to <strong>the</strong> direct measurement of <strong>on</strong>ly<br />

attenuatedorspatiallydilute<strong>beam</strong>s. Directdetecti<strong>on</strong>withsolid-state arraysoffers<strong>the</strong><br />

potential of energy discriminati<strong>on</strong>, which might allow <strong>the</strong> fundamental <strong>and</strong> harm<strong>on</strong>ics<br />

to be distinguished. However, this may <strong>on</strong>ly be possible if each pixel collects no more<br />

than <strong>on</strong>e phot<strong>on</strong>[167].<br />

Ano<strong>the</strong>r factor to c<strong>on</strong>sider is <strong>the</strong> wavelength sensitivity of <strong>the</strong> CCD. A st<strong>and</strong>ard<br />

fr<strong>on</strong>t illuminated CCD detector will be insensitive to VUV phot<strong>on</strong>s as <strong>the</strong>y cannot<br />

penetrate through <strong>the</strong> inactive layer <strong>on</strong> <strong>the</strong> detector surface. C<strong>on</strong>versely, <strong>the</strong> sensitivity<br />

can drop with harder x-rays as <strong>the</strong>y can pass straight through <strong>the</strong> depleti<strong>on</strong><br />

layer of <strong>the</strong> detector. Different types of CCD are thus needed for different parts of <strong>the</strong><br />

spectrum,for example ’back thinned’ for soft X-rays less than 2 keV, ordinary ’fr<strong>on</strong>t<br />

illuminated’ from 2 to 10 keV, <strong>and</strong> ’deep depleti<strong>on</strong>’ type for X-rays above 10 keV.<br />

Imaging a replica of <strong>the</strong> <strong>beam</strong><br />

A comm<strong>on</strong> approach for imaging techniques is to measure indirectly via <strong>the</strong> producti<strong>on</strong><br />

of a replica of <strong>the</strong> <strong>beam</strong>. One way of achieving this is by illuminating a luminescent<br />

screen with <strong>the</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong>n imaging <strong>the</strong> luminescence through a viewport<br />

<strong>on</strong> <strong>the</strong> vacuum vessel with an optical camera[170]. This allows for higher ultimate<br />

5 www.princet<strong>on</strong>instruments.com/products/xraycam/


8.3. Direct imaging of <strong>the</strong> <strong>beam</strong> 115<br />

Source<br />

pinhole array<br />

YAG<br />

Viewport<br />

Mirror<br />

Camera<br />

Beam profile<br />

Figure 8.4: The use of a YAG screen toge<strong>the</strong>r with a pinhole array for X-ray <strong>beam</strong> profiling as described by<br />

Bol<strong>and</strong> <strong>and</strong> co-workers.<br />

spatial resoluti<strong>on</strong> than with direct CCD imaging. The resoluti<strong>on</strong> is determined by <strong>the</strong><br />

quality <strong>and</strong> thickness of <strong>the</strong> screen, <strong>the</strong> magnificati<strong>on</strong> <strong>and</strong> aberrati<strong>on</strong>s of <strong>the</strong> camera<br />

optics, <strong>and</strong> <strong>the</strong> camera sensor resoluti<strong>on</strong>. The ultimate limit is set by diffracti<strong>on</strong> of<br />

<strong>the</strong> visible light through <strong>the</strong> camera optics, <strong>and</strong> is thus ∼ 0.5 µm. The viewport must<br />

be of high optical quality if it is not to degrade <strong>the</strong> image. On <strong>the</strong> Australian Light<br />

Source diagnostic <strong>beam</strong>line, Bol<strong>and</strong> et al. [170] placed a pinhole array before a YAG<br />

screen (Figure 8.4) so that several fully resolved images are recorded <strong>and</strong> this allows<br />

<strong>the</strong> <strong>beam</strong> divergence as well as <strong>the</strong> profile to be measured. This approach could be<br />

useful with a free electr<strong>on</strong> laser since it would simplify single-pulse measurement of<br />

<strong>the</strong> <strong>beam</strong> divergence.<br />

With traditi<strong>on</strong>al powder phosphors a grain size of c. 1 µm is available <strong>and</strong> <strong>the</strong> resoluti<strong>on</strong><br />

as defined by <strong>the</strong> line-spread functi<strong>on</strong> is approximately equal to <strong>the</strong> thickness<br />

of <strong>the</strong> phosphor layer[171]. But if <strong>the</strong> phosphor layer is made too thin (less than a<br />

few µm) <strong>the</strong>n <strong>the</strong> light yield decreases dramatically <strong>and</strong> performance is compromised.<br />

Optically transparent luminescent screens, or scintillators, are a better choice for<br />

achieving high spatial resoluti<strong>on</strong>. Since <strong>the</strong> camera is now focussed <strong>on</strong> a transparent<br />

screen, <strong>the</strong> screen thickness <strong>and</strong> depth of field of <strong>the</strong> camera optics play an important<br />

part in <strong>the</strong> resoluti<strong>on</strong>. The scintillators must <strong>the</strong>refore be thin, have a surface of high<br />

optical quality, <strong>and</strong> c<strong>on</strong>tain high Z elements to increase absorpti<strong>on</strong>. Cerium doped<br />

aluminum garnets are most often used. Koch et al. [171] used 5 µm YAG:Ce <strong>on</strong> 170<br />

µm undoped YAG crystal in combinati<strong>on</strong> with diffracti<strong>on</strong>-limited microscope objectives<br />

<strong>and</strong> achieved spatial resoluti<strong>on</strong>s of less than 1 µm in micro-imaging experiments<br />

at <strong>the</strong> ESRF. Tous et al. [172] achieved resoluti<strong>on</strong> of about 1 µm with an anode X-ray<br />

source using YAG:Ce <strong>and</strong> LuAG:Ce, both of 20 µm thickness.<br />

The b<strong>and</strong>width of such a system will be limited by <strong>the</strong> camera read-out rate <strong>and</strong><br />

ultimately by <strong>the</strong> decay time of <strong>the</strong> luminescent material. The decay time of some<br />

luminescent materials can be roughly 1 µs, but <strong>the</strong> visible luminescence from YAG:Ce<br />

is very fast at ∼ 70 ns[173] <strong>and</strong> is thus fast enough to resolve single pulses at even 1<br />

MHz. As discussed above, cameras with frame rates up to 600,000 Hz are available,<br />

though not with c<strong>on</strong>tinuous output at that rate. It would need to be checked that<br />

<strong>the</strong>y would have <strong>the</strong> sensitivity to record <strong>the</strong> luminescence from a single free electr<strong>on</strong><br />

laser pulse (<strong>the</strong> m<strong>on</strong>ochrome sensitivity of <strong>the</strong> Memrecam GX-8 referred to is 20,000<br />

ISO). With slower frame rates, it should be possible to select just <strong>on</strong>e pulse in <strong>the</strong><br />

frame cycle by gating <strong>the</strong> luminescence with a high-speed electro-optical shutter such<br />

as a Pockels cell. Clearly, <strong>the</strong> total number of phot<strong>on</strong>s collected from <strong>the</strong> screen in<br />

this mode will be low <strong>and</strong> so an image intensified camera may be required. These are


116 8. Beam cross-secti<strong>on</strong> diagnostics<br />

available commercially, for example <strong>the</strong> Lambert Instruments LI2CAM 6 ,which has a<br />

frame rate of just 15 fps but can be gated down to 2 ns.<br />

Itis also possible tocreate anelectr<strong>on</strong> replicaof<strong>the</strong>phot<strong>on</strong><strong>beam</strong> profilebydirectly<br />

illuminating a multichannel plate with <strong>the</strong> free electr<strong>on</strong> laser <strong>beam</strong>. A phosphor<br />

screen placed near <strong>the</strong> plate c<strong>on</strong>verts <strong>the</strong> electr<strong>on</strong>s to visible light whilst preserving<br />

<strong>the</strong> spatial origin of <strong>the</strong> electr<strong>on</strong>s, <strong>and</strong> <strong>the</strong> luminescence from <strong>the</strong> screen is imaged<br />

by a CCD camera. This approach was used by Yang et al. to image <strong>the</strong> profile of<br />

<strong>the</strong> X-ray pulses from a Compt<strong>on</strong> back-scattering source[174] (though <strong>the</strong> image was<br />

integrated over many pulses due to <strong>the</strong> very low intensity). The spatial resoluti<strong>on</strong><br />

will be determined by <strong>the</strong> channel pore size, which at 10 µm would give a similar<br />

resoluti<strong>on</strong> to direct CCD imaging. The sensitivity would be determined by <strong>the</strong> photoemissivity<br />

of <strong>the</strong> channel plate, which is dependent <strong>on</strong> wavelength. Coatings can be<br />

added to enhance <strong>the</strong> photoemissi<strong>on</strong> at l<strong>on</strong>g wavelengths. At shorter wavelengths,<br />

<strong>the</strong> efficiency drops as <strong>the</strong> X-rays are absorbed more in <strong>the</strong> bulk of <strong>the</strong> channel plate<br />

material. At very short wavelengths, <strong>the</strong> X-rays may penetrate through <strong>the</strong> pore walls<br />

<strong>and</strong> this will reduce <strong>the</strong> spatial resoluti<strong>on</strong>.<br />

The b<strong>and</strong>width will still be limited by <strong>the</strong> camera frame rate, but <strong>the</strong> ultimate<br />

limit will also be influenced by <strong>the</strong> time of <strong>the</strong> avalanching process in <strong>the</strong> pores,<br />

as well as <strong>the</strong> luminescence decay. The b<strong>and</strong>width could be greatly increased by<br />

replacing <strong>the</strong> phosphor screen <strong>and</strong> camera with an anode array. Assuming parallel<br />

readout of each anode, <strong>the</strong> detector resp<strong>on</strong>se could be reduced to <strong>the</strong> multichannel<br />

plate resp<strong>on</strong>se time, which could be in <strong>the</strong> nanosec<strong>on</strong>d regi<strong>on</strong>. This would allow every<br />

pulse to be measured, but <strong>the</strong> spatial resoluti<strong>on</strong> would be limited by <strong>the</strong> anode array<br />

size. Alternatively, <strong>the</strong> charge pulse from <strong>the</strong> multichannel plate as <strong>the</strong> phot<strong>on</strong> pulse<br />

strikes it could be used to trigger <strong>the</strong> gating of <strong>the</strong> camera to select a single pulse.<br />

This would obviate <strong>the</strong> need for synchr<strong>on</strong>izati<strong>on</strong> to <strong>the</strong> X-ray <strong>beam</strong> itself. If this is<br />

not possible, <strong>the</strong> timing trigger from <strong>the</strong> free electr<strong>on</strong> laser pulse could be used to<br />

gate <strong>the</strong> multichannel plate.<br />

A refinement of this approach would be to illuminate a photo-emissive material<br />

(e.g. a metal foil) <strong>and</strong> image <strong>the</strong> emitted photoelectr<strong>on</strong>s. A key requirement for this<br />

to work is preserving <strong>the</strong> spatial informati<strong>on</strong> encoded into <strong>the</strong> photoelectr<strong>on</strong>s whilst<br />

<strong>the</strong>y <strong>transport</strong>ed to <strong>the</strong> multichannel plate. Similar problems face <strong>the</strong> techniques<br />

described in secti<strong>on</strong>s 8.6 <strong>and</strong> 8.7.<br />

An imaging detector based <strong>on</strong> <strong>the</strong> photoc<strong>on</strong>ductive effect in Type IIa CVD diam<strong>on</strong>d<br />

has been developed at <strong>the</strong> APS byShu et al. [175]. This detector is an extensi<strong>on</strong><br />

of work d<strong>on</strong>e <strong>on</strong> quadrant detectors (q.v.). A 127 µm thick diam<strong>on</strong>d disc was patterned<br />

<strong>on</strong> both sides with sixteen 0.2 µm thick <strong>and</strong> 175 µm wide aluminum strips.<br />

The patterns <strong>on</strong> <strong>the</strong> two sides are orthog<strong>on</strong>al such that a 16 by 16 pixel array is<br />

created with 175 µm by 175 µm pixel size. The strips <strong>on</strong> <strong>on</strong>e side are c<strong>on</strong>nected to a<br />

bias supply via a 16-channel switch, whilst those <strong>on</strong> <strong>the</strong> o<strong>the</strong>r side are c<strong>on</strong>nected to<br />

sixteen discrete current amplifiers.<br />

The disc has a high transmissi<strong>on</strong> for hard X-rays (around 10 keV) <strong>and</strong> thus <strong>the</strong><br />

m<strong>on</strong>itor can be used <strong>on</strong>-line for c<strong>on</strong>stant <strong>beam</strong> m<strong>on</strong>itoring. When <strong>the</strong> X-ray <strong>beam</strong><br />

passes through <strong>the</strong> diam<strong>on</strong>d, <strong>the</strong> localized c<strong>on</strong>ductivity rises in proporti<strong>on</strong> to <strong>the</strong><br />

absorbed X-ray power. Thus, <strong>the</strong> current passing through <strong>the</strong> diam<strong>on</strong>d when <strong>the</strong> bias<br />

is applied is highest where <strong>the</strong> incident X-ray <strong>beam</strong> is most intense. In order to build<br />

up a two dimensi<strong>on</strong>al picture of <strong>the</strong> <strong>beam</strong> intensity profile, <strong>the</strong> bias is applied to each<br />

6 www.lambert-instruments.com


8.4. Scanning techniques 117<br />

strip <strong>on</strong> <strong>on</strong>e side in turn <strong>and</strong> <strong>the</strong> current from <strong>the</strong> sixteen strips <strong>on</strong> <strong>the</strong> o<strong>the</strong>r side<br />

recorded.<br />

The spatial resoluti<strong>on</strong> is naturally fairly low, but sufficient to give a true <strong>beam</strong><br />

profile assuming <strong>the</strong> <strong>beam</strong> is not highly structured <strong>and</strong> overlaps with enough of <strong>the</strong><br />

electrode strips. The b<strong>and</strong>width is limited by <strong>the</strong> need to apply <strong>the</strong> bias voltage to<br />

each strip in turn. Never<strong>the</strong>less, <strong>the</strong> data acquisiti<strong>on</strong> system was able to scan at 300<br />

to 3000 columns per sec<strong>on</strong>d (from 19 to 190 Hz). Of course, this still means that<br />

<strong>the</strong> system will give <strong>on</strong>ly aline profile <strong>and</strong> not a full image of a single free electr<strong>on</strong><br />

laser pulse; <strong>the</strong> electrode structure would have to be modified to give simultaneous<br />

readout of all 256 channels for single pulse image.<br />

Summary of imaging techniques<br />

All <strong>the</strong> imaging techniques described above are invasive. Direct imaging with a CCD<br />

<strong>and</strong> direct illuminati<strong>on</strong> of a micro-channel plate will stop <strong>the</strong> entire <strong>beam</strong> <strong>and</strong> so<br />

could <strong>on</strong>ly be used during commissi<strong>on</strong>ing or if <strong>the</strong> experiment is transparent (e.g. gas<br />

phase) so that <strong>the</strong> detector can be placed after <strong>the</strong> experiment. The o<strong>the</strong>r techniques<br />

could be made partially transparent. For example, hard X-rays could pass through a<br />

sufficiently thin luminescent screen without excessive loss. The luminescent intensity<br />

is proporti<strong>on</strong>al to <strong>the</strong> number of phot<strong>on</strong>s absorbed <strong>and</strong> so some losses are required for<br />

<strong>the</strong> detector to work. On <strong>the</strong> o<strong>the</strong>r h<strong>and</strong>, since photoemissi<strong>on</strong> is essentially a surface<br />

phenomen<strong>on</strong>, a very thin <strong>and</strong> thus highly transmitting foil would give <strong>the</strong> same signal<br />

as a thick <strong>on</strong>e. In any case, <strong>the</strong> disrupti<strong>on</strong> to a coherent free electr<strong>on</strong> laser <strong>beam</strong><br />

as it passes through a screen or foil would probably be unacceptable. There are also<br />

c<strong>on</strong>cerns over radiati<strong>on</strong> damage, ablati<strong>on</strong>,linearity <strong>and</strong> saturati<strong>on</strong> with high-fluence,<br />

ultra-short free electr<strong>on</strong> laser pulses. The risk of ablati<strong>on</strong> would need to be c<strong>on</strong>trolled<br />

by using <strong>the</strong>se techniques <strong>on</strong>ly where <strong>the</strong> <strong>beam</strong> is spatially dilute, <strong>and</strong> this might also<br />

be sufficient to prevent saturati<strong>on</strong> <strong>and</strong> n<strong>on</strong>-linear effects.<br />

The wavelength resp<strong>on</strong>se of <strong>the</strong> detectors needs to be c<strong>on</strong>sidered. It is obviously<br />

desirable for <strong>the</strong> detector to work over as wide a range of phot<strong>on</strong> energies as possible.<br />

However, a broad-b<strong>and</strong> resp<strong>on</strong>se can cause problems if it is very n<strong>on</strong>-linear. For<br />

example, photo-emissi<strong>on</strong> yield is much greater at VUV than X-ray wavelengths <strong>and</strong><br />

this can distort an X-ray <strong>beam</strong> profile if <strong>the</strong>re is low level VUV light with a different<br />

spatial pattern to <strong>the</strong> X-ray profile.<br />

8.4 Scanning techniques<br />

Scanning wire<br />

This is perhaps, at a glance, <strong>the</strong> simplest soluti<strong>on</strong> to <strong>the</strong> present problem: scan a<br />

wire through <strong>the</strong> <strong>beam</strong>. A thin metal wire (such as tungsten) is step-scanned through<br />

<strong>the</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> <strong>beam</strong> intensity deduced from, for example, <strong>the</strong> drain current in<br />

<strong>the</strong> wire[176], <strong>the</strong> intensity of electr<strong>on</strong>s emitted from <strong>the</strong> wire, or <strong>the</strong> intensity of<br />

scattered or fluorescent light from <strong>the</strong> wire, all of which should be proporti<strong>on</strong>al to <strong>the</strong><br />

<strong>beam</strong> intensity hitting <strong>the</strong> wire. The signal at each positi<strong>on</strong> in <strong>the</strong> scan is <strong>the</strong> result<br />

of an integrati<strong>on</strong> of <strong>the</strong> <strong>beam</strong> intensity al<strong>on</strong>g <strong>the</strong> illuminated length of <strong>the</strong> wire, i.e.<br />

in <strong>the</strong> directi<strong>on</strong> orthog<strong>on</strong>al to <strong>the</strong> scan. The measured profile is thus a c<strong>on</strong>voluti<strong>on</strong><br />

of <strong>the</strong> actual <strong>beam</strong> profile <strong>and</strong> <strong>the</strong> illuminated length of <strong>the</strong> wire. This means <strong>the</strong>


118 8. Beam cross-secti<strong>on</strong> diagnostics<br />

scanningxwires.eps<br />

Figure 8.5: Working principle of a scanning crossed-wire m<strong>on</strong>itor, after Ref. [178].<br />

measured profile will be distorted when <strong>the</strong> orthog<strong>on</strong>al <strong>beam</strong> secti<strong>on</strong> is not uniform<br />

such that <strong>the</strong> illuminated length varies through <strong>the</strong> scan[177]. This will occur, for<br />

example when <strong>the</strong> wire is tilted relative to an elliptical <strong>beam</strong>.<br />

The resoluti<strong>on</strong> of <strong>the</strong> wire type m<strong>on</strong>itor depends <strong>on</strong> <strong>the</strong> step resoluti<strong>on</strong> of <strong>the</strong><br />

scanning mechanism, <strong>the</strong> diameter of <strong>the</strong> wire relative to <strong>the</strong> <strong>beam</strong> width, <strong>and</strong> <strong>the</strong><br />

straightness <strong>and</strong> uniformity of <strong>the</strong> wire. Radiati<strong>on</strong> scattering from <strong>the</strong> edges of <strong>the</strong><br />

wire will degrade <strong>the</strong> resoluti<strong>on</strong>. Thin wires give higher resoluti<strong>on</strong>, but are harder to<br />

cool. If a wire gets too hot, <strong>the</strong>rmi<strong>on</strong>ic emissi<strong>on</strong> can occur even before melting <strong>and</strong><br />

this will distort <strong>the</strong> profile when electr<strong>on</strong> detecti<strong>on</strong> is used to infer <strong>the</strong> <strong>beam</strong> intensity.<br />

Scanning crossed wires<br />

Two wires are crossed at 90 ◦ <strong>and</strong> scanned through <strong>the</strong> <strong>beam</strong> al<strong>on</strong>g a directi<strong>on</strong> at 45 ◦<br />

to each wire <strong>–</strong> see Figure 8.5. In this way, two orthog<strong>on</strong>al profiles can be derived in<br />

<strong>on</strong>e scan, but detecti<strong>on</strong> is limited to drain current to allow <strong>the</strong> signals to be distinguished.<br />

Each profile is still an integrati<strong>on</strong> in <strong>the</strong> orthog<strong>on</strong>al directi<strong>on</strong>.The instrument<br />

developed by Fajardo <strong>and</strong> Ferrer[178] has a reported resoluti<strong>on</strong> of better than 5 µm,<br />

limited by mechanical reproducibility <strong>and</strong> vibrati<strong>on</strong> in <strong>the</strong> scanning mechanism. The<br />

finite resp<strong>on</strong>se time of <strong>the</strong> amplifiers results in a difference in apparent positi<strong>on</strong> <strong>on</strong><br />

forward <strong>and</strong> backward scans if <strong>the</strong> scan speed is too high; this effect was not observed<br />

at a speed below ∼ 1 mm/sec.<br />

Scanning slit<br />

A fine slit is scanned through <strong>the</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> transmitted intensity measured by<br />

a detector (e.g. a photodiode, a micro channel plate, or a luminescent screen <strong>and</strong><br />

video camera combinati<strong>on</strong>)[179].Again, <strong>the</strong> intensity in <strong>the</strong> orthog<strong>on</strong>al directi<strong>on</strong> is<br />

integrated by <strong>the</strong> length of <strong>the</strong> slit. The result is analogous to <strong>the</strong> scanning wire<br />

but <strong>the</strong> technique is more invasive, though easier to cool. Scanning slits are however


8.4. Scanning techniques 119<br />

useful for probing aperture related aberrati<strong>on</strong>s in optical systems, though <strong>the</strong> <strong>beam</strong><br />

spread after <strong>the</strong> slit caused by diffracti<strong>on</strong> can c<strong>on</strong>fuse <strong>the</strong> results.<br />

Scanning pinhole<br />

A refinement of <strong>the</strong> scanning slit in which <strong>the</strong> <strong>beam</strong> profile is not integrated al<strong>on</strong>g <strong>the</strong><br />

directi<strong>on</strong> orthog<strong>on</strong>al to <strong>the</strong> scan directi<strong>on</strong> <strong>and</strong> so true line profiles through <strong>the</strong> <strong>beam</strong><br />

can be made. With two-axis c<strong>on</strong>trol of <strong>the</strong> pin-hole moti<strong>on</strong>, line profiles in arbitrary<br />

directi<strong>on</strong>s through <strong>the</strong> <strong>beam</strong> can be made. The pin-hole is comm<strong>on</strong>ly achieved by<br />

using ei<strong>the</strong>r two slits crossed at 90 ◦ to make a rectangular pinhole or with four jaws<br />

making <strong>the</strong> two orthog<strong>on</strong>al slits. The advantage of <strong>the</strong> latter is that <strong>the</strong> size of <strong>the</strong><br />

pin-hole can be easily changed, though a4-axis driveis needed<strong>and</strong> verysmall pinholes<br />

are not realizable because of diffracti<strong>on</strong> at each jaw <strong>and</strong> <strong>the</strong> necessary l<strong>on</strong>gitudinal<br />

separati<strong>on</strong> between <strong>the</strong>m.<br />

Scanning knife-edge<br />

A blade is scanned through <strong>the</strong> <strong>beam</strong> such that it successively obscures (or reveals)<br />

more <strong>and</strong> more of <strong>the</strong> <strong>beam</strong>. The intensity passing <strong>the</strong> blade is recorded by a detector<br />

such as a photodiode or mictro-channel plate[180]. The recorded signal is <strong>the</strong> integral<br />

of <strong>the</strong> profile in <strong>the</strong> scan directi<strong>on</strong> <strong>and</strong> <strong>the</strong> deduced profile is integrated in <strong>the</strong> orthog<strong>on</strong>al<br />

directi<strong>on</strong>. If <strong>the</strong> profile is being measured at a focus, a wire of diameter greater<br />

than <strong>the</strong> <strong>beam</strong> size can be used instead of an edge[181]. This has <strong>the</strong> advantage that<br />

it is easy to achieve a smooth edge with a wire than with a blade.<br />

Summary of scanning techniques<br />

The most important drawback when applying <strong>the</strong>se techniques to a free electr<strong>on</strong><br />

lasersource is that <strong>the</strong>y cannot be used to measure a single pulse due to <strong>the</strong> step-wise<br />

nature of <strong>the</strong> measurement. In additi<strong>on</strong>, <strong>the</strong> techniques are invasive (i.e. disruptive<br />

to <strong>the</strong> <strong>beam</strong>), especially as <strong>the</strong> coherent nature of <strong>the</strong> phot<strong>on</strong> <strong>beam</strong> will mean <strong>the</strong><br />

disrupti<strong>on</strong> will be enhanced due to diffracti<strong>on</strong> effects. Therefore, in <strong>the</strong> c<strong>on</strong>text of a<br />

free electr<strong>on</strong> laser source, <strong>the</strong>se techniques are useful nei<strong>the</strong>r during commissi<strong>on</strong>ing<br />

(when <strong>the</strong> invasive nature would not be an issue but <strong>the</strong> inability to resolve single<br />

pulses is), nor during operati<strong>on</strong> (as <strong>the</strong> invasive nature makes <strong>the</strong>m unsuitable as <strong>on</strong>line<br />

m<strong>on</strong>itors). In additi<strong>on</strong>, <strong>the</strong> high pulse intensity of an unattenuated free electr<strong>on</strong><br />

laser fundamental gives serious c<strong>on</strong>cern over <strong>the</strong> damage to <strong>the</strong> scanning elements.<br />

With a free electr<strong>on</strong> laser source, <strong>the</strong>re are two modes of damage, viz. melting <strong>and</strong><br />

ablati<strong>on</strong>. At low repetiti<strong>on</strong> rates (less than 1 kHz) melting is unlikely as <strong>the</strong> average<br />

power is less than in a synchrotr<strong>on</strong> <strong>beam</strong>, but ablati<strong>on</strong> is likely unless <strong>the</strong> pulses<br />

are attenuated (for example with a gas absorber, as discussed above in chapter 4 see<br />

page 69). Without attenuati<strong>on</strong>, <strong>the</strong>re is also <strong>the</strong> possibility of reaching n<strong>on</strong>-linear<br />

regimes where <strong>the</strong> measured signal is no l<strong>on</strong>ger proporti<strong>on</strong>al to <strong>the</strong> incident intensity,<br />

e.g. through space charge effects.


120 8. Beam cross-secti<strong>on</strong> diagnostics<br />

8.5 I<strong>on</strong>izati<strong>on</strong> <strong>beam</strong>profile detectors<br />

The <strong>beam</strong>-profile from a high-energy <strong>beam</strong> of particles (phot<strong>on</strong>s, electr<strong>on</strong>s, neutr<strong>on</strong>s...)<br />

can be measured by detecting <strong>the</strong> i<strong>on</strong>s (or electr<strong>on</strong>s) resultant from i<strong>on</strong>izati<strong>on</strong><br />

events occurring in a residual gas intersecting <strong>the</strong> <strong>beam</strong>. By accelerating <strong>the</strong><br />

i<strong>on</strong>s in a homogeneous electric field towards a multichannel plate detector (see Figure<br />

8.6) combined with a positi<strong>on</strong> sensitive read-out anode an image of <strong>the</strong> <strong>beam</strong> can<br />

be recorded.<br />

From Equati<strong>on</strong> 2.5 <strong>on</strong>e would expect <strong>the</strong> i<strong>on</strong>s to get accelerated in straight lines<br />

towards <strong>the</strong> detector since <strong>the</strong> force <strong>on</strong> <strong>the</strong> charged particle is directly proporti<strong>on</strong>al<br />

to <strong>the</strong> electric field strength; <strong>the</strong> i<strong>on</strong>ized particles, however, will have velocity comp<strong>on</strong>ents<br />

in o<strong>the</strong>r directi<strong>on</strong>s besides normal to <strong>the</strong> detector <strong>–</strong> this will give rise to a<br />

broadening of <strong>the</strong> profile which can be minimized by applying a str<strong>on</strong>ger electric field.<br />

Amicrochannel plate detector also haveafiniteresoluti<strong>on</strong> since <strong>the</strong>actual channels<br />

are in <strong>the</strong> order of 10 µm in diameter. There is also a possibility that several channels<br />

detect a single event which will blur <strong>the</strong> image fur<strong>the</strong>r.<br />

Anode<br />

I<strong>on</strong><br />

MCP 1<br />

MCP 2<br />

Output signal<br />

Figure 8.6: Microchannel plates mounted in a chevr<strong>on</strong> geometry with voltages coupled to detect i<strong>on</strong>s. When an<br />

i<strong>on</strong> hit a microchannel plate it gives rise to several sec<strong>on</strong>dary i<strong>on</strong>izati<strong>on</strong>s <strong>–</strong> <strong>the</strong> electr<strong>on</strong>s from those i<strong>on</strong>izati<strong>on</strong>s,<br />

in turn, gives rise to o<strong>the</strong>r sec<strong>on</strong>dary electr<strong>on</strong>s, thus an amplificati<strong>on</strong> occurs. The resulting charge cloud hits<br />

<strong>the</strong> anode <strong>and</strong> gives rise to an electric current which can be read out. An alternative to this scheme exist where<br />

a phosphor plate is mounted behind <strong>the</strong> anode which can be read out optically with, for instance, an CCD<br />

camera.<br />

At Flash this type of <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor, using a phosphor screen toge<strong>the</strong>r<br />

with a camera, gives a reported spatial resoluti<strong>on</strong> of c. 50 µm[182]. Obviously this<br />

type of detector also gives <strong>the</strong> positi<strong>on</strong> of <strong>the</strong> <strong>beam</strong>.<br />

8.6 Imaging i<strong>on</strong> chambers<br />

An established technique for profiling particle <strong>beam</strong>s in high-energy accelerators is to<br />

image <strong>the</strong> <strong>beam</strong> path through <strong>the</strong> residual gas (or gas added at a very low pressure<br />

gas, i.e. less than 10 -5 mbar) in <strong>the</strong> <strong>beam</strong> <strong>transport</strong> system. As <strong>the</strong> <strong>beam</strong> passes<br />

through <strong>the</strong> gas, it will i<strong>on</strong>ize it <strong>and</strong> <strong>the</strong> i<strong>on</strong> density is proporti<strong>on</strong>al to <strong>the</strong> particle<br />

density in <strong>the</strong> <strong>beam</strong>. Thus, if <strong>the</strong> i<strong>on</strong>s can be channelled linearly to a luminescent<br />

screen with a uniform electric field, <strong>the</strong> intensity of luminescence is also proporti<strong>on</strong>al<br />

to particle density <strong>and</strong> <strong>the</strong> <strong>beam</strong> profile can be recorded with a video camera. The i<strong>on</strong>


8.6. Imaging i<strong>on</strong> chambers 121<br />

signal is usually amplified by generating multiple electr<strong>on</strong>s for each incident i<strong>on</strong> with<br />

a micro-channel plate placed in fr<strong>on</strong>t of <strong>the</strong> screen. The benefit of this system is that,<br />

when using just <strong>the</strong> residual gas, <strong>the</strong> m<strong>on</strong>itor is completely n<strong>on</strong>-invasive. However,<br />

<strong>the</strong> achievable resoluti<strong>on</strong> has tended to be ra<strong>the</strong>r poor, of <strong>the</strong> 1 mm. This is mainly<br />

due toperturbati<strong>on</strong> of <strong>the</strong> drifting i<strong>on</strong>s by<strong>the</strong> electric field of <strong>the</strong> particle <strong>beam</strong>, which<br />

causes <strong>the</strong>mtospread awayfrom <strong>the</strong>regi<strong>on</strong>s ofhighi<strong>on</strong> density,<strong>and</strong>sobroadening<strong>the</strong><br />

recorded image. I<strong>on</strong>s, ra<strong>the</strong>r than electr<strong>on</strong>s, are usually detected since <strong>the</strong>ir greater<br />

mass means <strong>the</strong>y are less susceptible to <strong>the</strong>se perturbati<strong>on</strong>s. Never<strong>the</strong>less, Fischer<br />

<strong>and</strong> Koopman were able to improve <strong>the</strong> resoluti<strong>on</strong> by measuring <strong>the</strong> electr<strong>on</strong>s[183].<br />

They achieved this by placing <strong>the</strong> i<strong>on</strong> chamber in a uniform magnetic field parallel to<br />

<strong>the</strong> electr<strong>on</strong> trajectories. The emitted electr<strong>on</strong>s will precess about <strong>the</strong> field <strong>and</strong> so can<br />

be channelled more linearly. The resoluti<strong>on</strong> was improved such that an actual <strong>beam</strong><br />

width of 1 mm RMS could be measured without instrumental broadening (implying<br />

a resoluti<strong>on</strong> not worse than a few hundred µm).<br />

The same measurement principle can be applied to an X-ray <strong>beam</strong>, with which<br />

<strong>the</strong>re is <strong>the</strong> advantage that <strong>the</strong> <strong>beam</strong> will not perturb <strong>the</strong> electr<strong>on</strong>s <strong>and</strong> i<strong>on</strong>s directly<br />

<strong>on</strong>ce created. However, preserving accurately <strong>the</strong>ir spatial point of origin is still<br />

difficult. The approach used by Ioudin et al. [184] is to encode <strong>the</strong> point of <strong>the</strong> spatial<br />

origin of <strong>the</strong> i<strong>on</strong>s <strong>on</strong>to <strong>the</strong>ir kinetic energy by accelerating <strong>the</strong>m with an extracti<strong>on</strong><br />

field <strong>–</strong> see Figure 8.7.<br />

The fur<strong>the</strong>r <strong>the</strong> i<strong>on</strong>s are from <strong>the</strong> cathode when created, <strong>the</strong> greater <strong>the</strong> kinetic energy<br />

<strong>the</strong>y gain during <strong>the</strong> accelerati<strong>on</strong>. The i<strong>on</strong>s are accelerated towards <strong>the</strong> entrance<br />

slit of an electrostatic energy analyzer, which disperses <strong>the</strong>m in kinetic energy <strong>on</strong>to a<br />

micro-channel plate. Thus positi<strong>on</strong> al<strong>on</strong>g <strong>the</strong> plate corresp<strong>on</strong>ds to spatial coordinate<br />

of i<strong>on</strong> generati<strong>on</strong>. The multi-channel plate amplifies <strong>the</strong> i<strong>on</strong> signal, which is imaged<br />

using a phosphor screen <strong>and</strong> video camera.<br />

This instrument, called <strong>the</strong> Beam Cross-secti<strong>on</strong> Image Detector (BCID), gives a<br />

profile of <strong>the</strong> <strong>beam</strong> in two axes <strong>–</strong> <strong>the</strong> secti<strong>on</strong> al<strong>on</strong>g x is mapped al<strong>on</strong>g <strong>the</strong> microchannel<br />

plate as shown in <strong>the</strong> figure, X1 = 2XEe/Ea, whilst <strong>the</strong> secti<strong>on</strong> al<strong>on</strong>g y is<br />

mapped in <strong>the</strong> orthog<strong>on</strong>al directi<strong>on</strong> <strong>on</strong> <strong>the</strong> plate. The spatial resoluti<strong>on</strong> is determined<br />

by a number of factors.I<strong>on</strong>s generated at l<strong>on</strong>gitudinally adjacent positi<strong>on</strong>s a <strong>and</strong> b to<br />

q2 in Figure 8.7 are mapped to different positi<strong>on</strong>s <strong>on</strong> <strong>the</strong> channel plate despite having<br />

<strong>the</strong> same kinetic energy. The slit width determines <strong>the</strong> l<strong>on</strong>gitudinal acceptance <strong>and</strong><br />

hence <strong>the</strong> positi<strong>on</strong>al spread <strong>on</strong> <strong>the</strong> plate <strong>and</strong> must thus be kept small to c<strong>on</strong>trol<br />

this blurring. The analyzer resoluti<strong>on</strong> is determined by <strong>the</strong> energy dispersi<strong>on</strong> <strong>and</strong><br />

<strong>the</strong> camera spatial resoluti<strong>on</strong>. Aberrati<strong>on</strong>s in <strong>the</strong> analyzer will distort <strong>the</strong> measured<br />

profile. The accuracy in <strong>the</strong> y-directi<strong>on</strong> is determined by how linearly <strong>the</strong> i<strong>on</strong> chamber<br />

/ encoder can preserve <strong>the</strong> transverse positi<strong>on</strong> al<strong>on</strong>g <strong>the</strong> length of <strong>the</strong> slit <strong>and</strong> <strong>the</strong><br />

linearity of <strong>the</strong> analyzer resp<strong>on</strong>se in <strong>the</strong> plane orthog<strong>on</strong>al to <strong>the</strong> dispersi<strong>on</strong> plane. The<br />

ultimate resoluti<strong>on</strong> is limited by<strong>the</strong> extent towhich <strong>the</strong> i<strong>on</strong>s acquire extra momentum<br />

in <strong>the</strong> y- <strong>and</strong> z-directi<strong>on</strong>s as <strong>the</strong> result of inter-i<strong>on</strong> scattering <strong>and</strong> <strong>the</strong> influence of stray<br />

external (especially magnetic) fields. These effects will be reduced by increasing <strong>the</strong><br />

gradient of <strong>the</strong> extracting field to increase <strong>the</strong> kinetic energy of <strong>the</strong> i<strong>on</strong>s. But, if <strong>the</strong><br />

kinetic energy is too high, <strong>the</strong>n <strong>the</strong> analyzer performance will be degraded.<br />

The sensitivity of <strong>the</strong> technique is limited by <strong>the</strong> number of i<strong>on</strong>s generated in <strong>the</strong><br />

residual gas <strong>and</strong> <strong>the</strong> detecti<strong>on</strong> efficiency. I<strong>on</strong> generati<strong>on</strong> is more likely at VUVthan Xray<br />

wavelengths <strong>and</strong> thus <strong>the</strong> spectral c<strong>on</strong>tent of <strong>the</strong> <strong>beam</strong> will influence <strong>the</strong> recorded<br />

profile. To measure <strong>the</strong> X-ray profile in, for example, <strong>the</strong> <strong>beam</strong> from a synchrotr<strong>on</strong><br />

dipole source would require <strong>the</strong> l<strong>on</strong>g wavelength comp<strong>on</strong>ents to be removed by a filter


122 8. Beam cross-secti<strong>on</strong> diagnostics<br />

X<br />

Extracti<strong>on</strong> field Ee<br />

+<br />

-<br />

X1<br />

q1<br />

q2<br />

X-ray <strong>beam</strong><br />

+<br />

Phosphor screen<br />

MCP<br />

-<br />

y<br />

x<br />

z<br />

Analyzer field Ea<br />

Figure 8.7: The <strong>beam</strong> cross-secti<strong>on</strong> image detector (BCID) as described by Ioudin et al. [184].<br />

such as a beryllium window[185]. This would however facilitate <strong>the</strong> additi<strong>on</strong> of a gas<br />

such as arg<strong>on</strong> or xen<strong>on</strong> at a low pressure to improve <strong>the</strong> i<strong>on</strong> count. In all cases,<br />

<strong>the</strong> necessity for a small analyzer slit limits <strong>the</strong> total count rate <strong>and</strong> all reported<br />

measurements of X-ray <strong>beam</strong>s have been d<strong>on</strong>e with a gas pressure of 10 -5 to 10 -3<br />

mbar <strong>and</strong> by integrating up to 256 frames with frame rate of 12.5 Hz [185, 186]. It<br />

is thus debatable whe<strong>the</strong>r <strong>the</strong> technique could have sufficient efficiency to record a<br />

single shot of a free electr<strong>on</strong> laser <strong>beam</strong>. There is little informati<strong>on</strong> in <strong>the</strong> references<br />

<strong>on</strong> <strong>the</strong> spatial resoluti<strong>on</strong> achieved with <strong>the</strong> BCID.<br />

A residual gas <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor is also being developed for use as an X-ray<br />

<strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor at PETRA III at HASYLAB. The RGBPM as described by<br />

Ilinski et al. [187] uses a layout similar to <strong>the</strong> original particle <strong>beam</strong> m<strong>on</strong>itors, i.e. <strong>the</strong><br />

kinetic energy encoding approach of Ioudin is not used <strong>and</strong> <strong>the</strong> spatial coordinate of<br />

<strong>the</strong> <strong>beam</strong> is mapped directly <strong>on</strong>to <strong>the</strong> detector. The generated i<strong>on</strong>s or electr<strong>on</strong>s are<br />

drifted in an applied electric field to a micro-channel plate (MCP) which produces<br />

an intensified image <strong>on</strong> a phosphor screen <strong>–</strong> Figure 8.8. The <strong>beam</strong> profile is recorded<br />

in <strong>on</strong>e axis <strong>on</strong>ly, this being in <strong>the</strong> directi<strong>on</strong> perpendicular to both <strong>the</strong> drift field <strong>and</strong><br />

<strong>beam</strong> directi<strong>on</strong>. The profile is recorded for a finite length in <strong>the</strong> propagati<strong>on</strong> directi<strong>on</strong>


8.6. Imaging i<strong>on</strong> chambers 123<br />

X-ray <strong>beam</strong><br />

Phosphor<br />

MCP<br />

Guide electrode<br />

Guide electrode<br />

Repeller electrode<br />

Field axis<br />

Profile axis<br />

Propagati<strong>on</strong> axis<br />

Figure 8.8: Schematic of PETRA-III RGBPM; a vertical profile of <strong>the</strong> <strong>beam</strong> is measured.<br />

of <strong>the</strong> <strong>beam</strong> determined by <strong>the</strong> l<strong>on</strong>gitudinal aperture of <strong>the</strong> detecti<strong>on</strong> system.<br />

Improvements in <strong>the</strong> resoluti<strong>on</strong> come from close attenti<strong>on</strong> to <strong>the</strong> quality of <strong>the</strong><br />

electric field, initial kinetic energy of <strong>the</strong> i<strong>on</strong>s or electr<strong>on</strong>s, resoluti<strong>on</strong> of <strong>the</strong> detector<br />

system, <strong>and</strong> data processing. To quote: ”The electrical field has to be uniform in<br />

order to provide aberrati<strong>on</strong> free <strong>beam</strong> profile. Broadening of <strong>the</strong> <strong>beam</strong> profile occurs<br />

due to electrical field n<strong>on</strong>-uniformity <strong>and</strong> presence of <strong>the</strong> transverse comp<strong>on</strong>ent of <strong>the</strong><br />

electrical field. This broadening should not exceed <strong>the</strong> broadening of <strong>the</strong> <strong>beam</strong> profile<br />

which is caused by <strong>the</strong> initial transverse kinetic energy of <strong>the</strong> i<strong>on</strong>s or electr<strong>on</strong>s. The<br />

resoluti<strong>on</strong> of detecti<strong>on</strong> system is defined by <strong>the</strong> MCP, <strong>the</strong> phosphor screen, optical<br />

coupling <strong>and</strong> signal background ratio. A proper data processing allows for sub-pixel<br />

resoluti<strong>on</strong>.” The guide electrodes shown in Figure 8.8 are designed to improve <strong>the</strong><br />

repeller field uniformity.<br />

A prototype RGXBPM was tested <strong>on</strong> ID6 at <strong>the</strong> ESRF at 29 m from <strong>the</strong> centre<br />

of a straight c<strong>on</strong>taining three undulators. The m<strong>on</strong>itor was located after a diam<strong>on</strong>d<br />

window of 300 µm thickness. Two readout systems were tried, viz. optical by imaging<br />

<strong>the</strong> phosphor screen with a CCD camera, <strong>and</strong> current by using a multi-channel plate<br />

with a split saw-tooth electrode. The channel plate had a channel diameter of 12 µm<br />

<strong>and</strong> <strong>the</strong> CCD camera had a resoluti<strong>on</strong> of 15 µm/pixel. With <strong>the</strong> optical readout, each<br />

CCD frame had an integrati<strong>on</strong> period of 300 ms <strong>and</strong> <strong>the</strong> <strong>beam</strong> centre of gravity could<br />

be calculated for each frame with a ring current of 68 mA <strong>and</strong> a residual gas pressure<br />

of ∼10 -7 mbar. A 5 µm step in <strong>the</strong> RGBPM positi<strong>on</strong> could clearly be resolved. The<br />

noise was not quoted, but looked to be 2 to 3 µm. The resoluti<strong>on</strong> with <strong>the</strong> splitelectrode<br />

detecti<strong>on</strong> was about twice that with <strong>the</strong> optical detecti<strong>on</strong>, with a 10 µm


124 8. Beam cross-secti<strong>on</strong> diagnostics<br />

step being resolved.<br />

An important limitati<strong>on</strong> of <strong>the</strong>se devices for measuring X-ray <strong>beam</strong>s is that <strong>the</strong><br />

cross-secti<strong>on</strong> for creating i<strong>on</strong>s is much greater at VUV than at X-ray wavelengths <strong>and</strong><br />

thus <strong>the</strong> sensitivity of m<strong>on</strong>itor is str<strong>on</strong>gly dependent <strong>on</strong> <strong>the</strong> <strong>beam</strong> spectral c<strong>on</strong>tent.<br />

This is a real issue <strong>on</strong> synchrotr<strong>on</strong> sources where <strong>the</strong> l<strong>on</strong>g wavelength halo around an<br />

undulator <strong>beam</strong> will give a false broadening of <strong>the</strong> measured profile even thoughit is<br />

at a low intensity compared to <strong>the</strong> <strong>on</strong>-axis X-rays[187]. In both <strong>the</strong> measurements<br />

of Ioudin <strong>and</strong> Ilinski, windows in <strong>the</strong> X-ray <strong>beam</strong> acted as high-pass filters. Such<br />

filtering is unlikely to be possible <strong>on</strong> a free electr<strong>on</strong> laser source (because of <strong>the</strong><br />

ablati<strong>on</strong> risk <strong>and</strong> <strong>beam</strong> disrupti<strong>on</strong>) but <strong>the</strong> spectral c<strong>on</strong>tent in such a source should<br />

not c<strong>on</strong>tain <strong>the</strong> l<strong>on</strong>g wavelength comp<strong>on</strong>ents anyway. Never<strong>the</strong>less, it is important to<br />

c<strong>on</strong>sider any possible sources of stray light (from dipoles, steering magnets etc <strong>and</strong><br />

<strong>the</strong> sp<strong>on</strong>taneous radiati<strong>on</strong> from <strong>the</strong> undulators) that might propagate with <strong>the</strong> free<br />

electr<strong>on</strong> laser <strong>beam</strong>.<br />

It should also be noted that <strong>the</strong> RGBPM is principally being designed to measure<br />

<strong>the</strong> centroid of <strong>the</strong> <strong>beam</strong>, this being achieved through measuring <strong>and</strong> analyzing <strong>the</strong><br />

<strong>beam</strong> profile. Thus, a small amount of profile broadening can be accounted for. Of<br />

course, bi-axial profiling requires a sec<strong>on</strong>d m<strong>on</strong>itor. There is also <strong>the</strong> need to fur<strong>the</strong>r<br />

improve <strong>the</strong> resoluti<strong>on</strong> <strong>and</strong> speed of <strong>the</strong> m<strong>on</strong>itor if it is to meet <strong>the</strong> requirements of<br />

a free electr<strong>on</strong> laser <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor.<br />

Fluorescence detecti<strong>on</strong> in residual gas m<strong>on</strong>itors<br />

When atoms in a gas are excited or i<strong>on</strong>ized by VUV to soft X-ray phot<strong>on</strong>s, <strong>the</strong><br />

dominant decay process is Auger emissi<strong>on</strong>. However, at hard X-ray wavelengths,<br />

decay by fluorescence becomes a significant process.There is <strong>the</strong>refore <strong>the</strong> opti<strong>on</strong> of<br />

recording <strong>the</strong> <strong>beam</strong> profile by imaging <strong>the</strong> fluorescent light. The principle is <strong>the</strong> same<br />

as with imaging <strong>the</strong> i<strong>on</strong> or electr<strong>on</strong> emissi<strong>on</strong>. The density of fluorescent phot<strong>on</strong>s is<br />

proporti<strong>on</strong>al to <strong>the</strong> density in X-rays in <strong>the</strong> <strong>beam</strong>. With fluorescence, <strong>the</strong>re is <strong>the</strong><br />

advantage that <strong>the</strong> phot<strong>on</strong>s will not be perturbed <strong>on</strong>ce emitted since <strong>the</strong> probability<br />

of scatter in a low-pressure gas will be negligible. However, <strong>the</strong> phot<strong>on</strong> emissi<strong>on</strong> is<br />

not instantaneous with <strong>the</strong> atom being excited by <strong>the</strong> x-ray. Thus, by <strong>the</strong> time <strong>the</strong><br />

atom emits, it will have moved under <strong>the</strong> influence of <strong>the</strong> space charge of all <strong>the</strong> i<strong>on</strong>s.<br />

The average moti<strong>on</strong> will <strong>the</strong>refore be away from <strong>the</strong> <strong>beam</strong> centre <strong>and</strong> this will result<br />

in a broadening of <strong>the</strong> fluorescence profile.<br />

This broadening effect is much more significant when measuring charged particle<br />

<strong>beam</strong> profiles since <strong>the</strong> i<strong>on</strong>s are also under <strong>the</strong> influence of <strong>the</strong> <strong>beam</strong> space charge.<br />

This effect has been observed when measuring <strong>the</strong> profile of prot<strong>on</strong> <strong>beam</strong>s[188]. When<br />

measuring an X-ray <strong>beam</strong> in a low pressure gas, <strong>the</strong> effect of inter-i<strong>on</strong> repulsi<strong>on</strong> is<br />

probably negligible. The i<strong>on</strong>s will still move r<strong>and</strong>omly before emitting <strong>the</strong> phot<strong>on</strong>s,<br />

<strong>and</strong> statistically more will move away from <strong>the</strong> <strong>beam</strong> centre than towards it. Thus,<br />

<strong>the</strong>re can still be some broadening effect that will depend <strong>on</strong> <strong>the</strong> lifetime of <strong>the</strong> excited<br />

state, which will in turn depend <strong>on</strong> <strong>the</strong> gas being i<strong>on</strong>ized.<br />

This type of m<strong>on</strong>itor is in use at <strong>the</strong> Cornell synchrotr<strong>on</strong> CHESS, where <strong>the</strong> m<strong>on</strong>itors<br />

are called video <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itors (VBPM). The layout is shown schematically<br />

in Figure 8.9. The <strong>beam</strong> tube is filled with helium at atmospheric pressure.<br />

Whilst this undoubtedly increases <strong>the</strong> fluorescence intensity, it would seem <strong>the</strong> main<br />

reas<strong>on</strong> for this is that <strong>the</strong> entire <strong>beam</strong>line is gas filled after a window of unspecified<br />

material which isolates it from <strong>the</strong> machine vacuum. (These are of course hard X-ray


8.6. Imaging i<strong>on</strong> chambers 125<br />

Vert. profile camera<br />

He gas<br />

Horiz. profile camera<br />

Figure 8.9: Schema of <strong>the</strong> CHESS VBPM system.<br />

X-ray <strong>beam</strong><br />

<strong>beam</strong>lines). The video cameras image <strong>the</strong> fluorescence via plane mirrors to prevent<br />

<strong>the</strong>m being damaged by scattered radiati<strong>on</strong> coming through <strong>the</strong> viewports.<br />

One issue that affects <strong>the</strong> resoluti<strong>on</strong> of <strong>the</strong> VBPM is <strong>the</strong> depth of field of <strong>the</strong><br />

video camera lens. The depth of field of <strong>the</strong> vertical profile camera is smaller than<br />

<strong>the</strong> horiz<strong>on</strong>tal <strong>beam</strong> size <strong>and</strong> so edges of <strong>the</strong> <strong>beam</strong> are not in focus. The blurring of<br />

<strong>the</strong> image leads to a broadening of <strong>the</strong> recorded <strong>beam</strong> profile. This is not an issue<br />

when <strong>the</strong> main purpose of <strong>the</strong> m<strong>on</strong>itor is to determine <strong>the</strong> <strong>beam</strong> centroid, as in this<br />

applicati<strong>on</strong>, but would be if accurate profiling were required. The spatial resoluti<strong>on</strong><br />

is also dependent of <strong>the</strong> lens magnificati<strong>on</strong> <strong>and</strong> camera CCD pixel size, which need<br />

to be chosen with <strong>the</strong> size of <strong>the</strong> <strong>beam</strong> to be measured in mind, <strong>and</strong> <strong>the</strong> number of<br />

digitizati<strong>on</strong> levels <strong>and</strong> overall signal to noise ratio. The camera system was tested to<br />

have an accuracy of 0.4 µm fwhm.<br />

The imaging system of <strong>the</strong> VBPM runs at 15 frames per sec<strong>on</strong>d <strong>and</strong> <strong>the</strong> intensity<br />

map is usually averaged over 10 frames. This low data rate is not an issue with an<br />

essentially c<strong>on</strong>tinuous source like a synchrotr<strong>on</strong>, but would be unacceptable for a free<br />

electr<strong>on</strong> laser source. A faster camera could of course be used; <strong>the</strong> main c<strong>on</strong>cern is<br />

whe<strong>the</strong>r enough fluorescence can be generated with a single free electr<strong>on</strong> laser pulse to<br />

be measurable. Enhancement of <strong>the</strong> signal byadding agas at a low pressure should be<br />

feasible; most X-ray free electr<strong>on</strong> lasers have a windowless gas attenuator anyway (c.f.<br />

chapter 4, p. 69). But <strong>the</strong> pressure limit will be determined by absorpti<strong>on</strong>, vacuum


126 8. Beam cross-secti<strong>on</strong> diagnostics<br />

c<strong>on</strong>taminati<strong>on</strong> (<strong>the</strong> gas cell cannot be isolated with windows) <strong>and</strong> self-modulati<strong>on</strong> of<br />

<strong>the</strong> intense phot<strong>on</strong> pulse as it passes through <strong>the</strong> gas.<br />

8.7 Sampling techniques<br />

The ideal sampling process would be to extract a time slice from <strong>the</strong> <strong>beam</strong> so that<br />

<strong>the</strong> full spatial profile can be measured of <strong>the</strong> slice whilst allowing <strong>the</strong> majority of<br />

<strong>the</strong> <strong>beam</strong> to pass <strong>on</strong>to <strong>the</strong> experiment. Unfortunately, extracting a time slice of a<br />

femtosec<strong>on</strong>d durati<strong>on</strong> X-ray pulse is not practical. Therefore,sampling in <strong>the</strong> spatial<br />

domain has to be used, for which <strong>the</strong>re are two approaches. One can remove just a<br />

small part of <strong>the</strong> spatial extent of <strong>the</strong> <strong>beam</strong>, which leads to <strong>on</strong>ly centroid informati<strong>on</strong><br />

(see secti<strong>on</strong> 8.10), or <strong>the</strong> <strong>beam</strong> can be sampled by uniformly removing part of <strong>the</strong><br />

overall intensity, which allows a profile to be measured. Clearly, <strong>the</strong> main objective is<br />

to remove as little of <strong>the</strong> overall <strong>beam</strong> intensity as possible to maximize <strong>the</strong> intensity<br />

reaching <strong>the</strong> experiment. Profiling by sampling requires an intensity <strong>beam</strong> splitter,<br />

whichisdifficulttoachievein<strong>the</strong>VUVtosoftX-rayrange. Thusmostdem<strong>on</strong>strati<strong>on</strong>s<br />

of this approach have been at shorter wavelengths.<br />

In an effort to make an effectively n<strong>on</strong>-invasive profiling measurement, van Silfhout<br />

developed a replicating technique in which <strong>the</strong> bulk of <strong>the</strong> X-ray <strong>beam</strong> passes through<br />

a thin ”featureless” foil angled in <strong>on</strong>e plane to <strong>the</strong> <strong>beam</strong> whilst a small fracti<strong>on</strong> of<br />

<strong>the</strong> X-rays are scattered from <strong>the</strong> foil <strong>and</strong> are imaged by a linear photodiode array<br />

to give line profile of <strong>the</strong> <strong>beam</strong>[189]. Because <strong>the</strong> x-rays are scattered ra<strong>the</strong>r than<br />

reflected, a Soller slit is used to give accurate mapping of each scattering point <strong>on</strong><br />

<strong>the</strong> foil to a single diode in <strong>the</strong> array -see Figure 8.10. Never<strong>the</strong>less, <strong>the</strong> measured<br />

profile must be dec<strong>on</strong>voluted from an instrumental functi<strong>on</strong> determined from <strong>the</strong><br />

profile measured with a very small <strong>beam</strong> footprint <strong>on</strong> <strong>the</strong> foil. The limiting spatial<br />

resoluti<strong>on</strong> is determined by <strong>the</strong> spatial sampling of <strong>the</strong> projected <strong>beam</strong> footprint <strong>on</strong><br />

<strong>the</strong> foil <strong>and</strong> can thus be improved by angling <strong>the</strong> foil at a more grazing angle to<br />

<strong>the</strong> <strong>beam</strong>. A <strong>beam</strong> positi<strong>on</strong> accuracy of 1 µm was claimed. The technique can be<br />

extended to 3-D imaging by tilting <strong>the</strong> foil in two planes <strong>and</strong> using a crossed-Soller slit<br />

<strong>and</strong> diode array. A more recent paper[190] shows how <strong>the</strong> technique can give a fast<br />

output of <strong>the</strong> <strong>beam</strong> profile (still at in <strong>on</strong>e dimensi<strong>on</strong>). The ultimate limit was 10 kHz,<br />

determined by <strong>the</strong> readout speed of <strong>the</strong> electr<strong>on</strong>ics, though <strong>the</strong> actual measurements<br />

were performed at 2400 Hz. The obvious c<strong>on</strong>cerns for free electr<strong>on</strong> laser use are: 1)<br />

ablati<strong>on</strong> damage to <strong>the</strong> foil, 2) disrupti<strong>on</strong> to <strong>the</strong> <strong>beam</strong> through coherent diffracti<strong>on</strong><br />

if <strong>the</strong> foil is not perfectly featureless <strong>and</strong> 3) <strong>the</strong> limitati<strong>on</strong> to hard x-rays to get high<br />

transmissi<strong>on</strong> through <strong>the</strong> foil.<br />

Ano<strong>the</strong>r approach, tested <strong>on</strong> an undulator <strong>beam</strong>line at SPring-8, is described in<br />

Kudo et al. [191]. A 30 µm thick CVD diam<strong>on</strong>d film was grown <strong>on</strong> a silic<strong>on</strong> substrate<br />

<strong>and</strong><strong>the</strong>silic<strong>on</strong> was etchedaway from acentralregi<strong>on</strong> of10mmdiameter. Theexposed<br />

diam<strong>on</strong>d has silic<strong>on</strong> doping from <strong>the</strong> substrate that causes it to photo-luminesce<br />

at 739 nm when exposed to an X-ray <strong>beam</strong>. The luminescence was imaged by a<br />

CCD camera giving a 2-D picture of <strong>the</strong> <strong>beam</strong> profile. The diam<strong>on</strong>d film is almost<br />

transparent to hard X-rays (50% transmissi<strong>on</strong> at 3300 eV, 90% at 6200 eV, for a<br />

density of 3.5 g/cm 3 ) <strong>and</strong> so most of <strong>the</strong> <strong>beam</strong> passes through it to <strong>the</strong> experiment.<br />

The luminescence resp<strong>on</strong>se to <strong>beam</strong> intensity was shown to be linear over at least<br />

four orders of magnitude. However, <strong>the</strong> luminescence is stimulated by a wide range<br />

of X-ray wavelengths <strong>and</strong> so <strong>the</strong> method is unable to distinguish between <strong>the</strong> <strong>on</strong> axis


8.7. Sampling techniques 127<br />

X-ray <strong>beam</strong><br />

Diode array<br />

Scattered X-rays<br />

Thin foil<br />

Soller slit<br />

Figure 8.10: The profiling system of van Silfhout, based <strong>on</strong> a Soller slit collimating <strong>the</strong> scattered X-rays from<br />

a thin foil.<br />

emissi<strong>on</strong> at <strong>the</strong>res<strong>on</strong>ant wavelength <strong>and</strong><strong>the</strong> lower energy radiati<strong>on</strong> outside <strong>the</strong>central<br />

emissi<strong>on</strong> c<strong>on</strong>e of <strong>the</strong> undulator, leading to an artificially broadened <strong>beam</strong> profile.<br />

However, this should not be an issue with a free electr<strong>on</strong> laser <strong>beam</strong>. No specific<br />

data for spatial resoluti<strong>on</strong> is given, but this will be determined by <strong>the</strong> combinati<strong>on</strong><br />

of <strong>the</strong> camera resoluti<strong>on</strong> <strong>and</strong> <strong>the</strong> <strong>beam</strong> footprint <strong>on</strong> <strong>the</strong> diam<strong>on</strong>d. B<strong>and</strong>width will be<br />

determined by <strong>the</strong> readout rate of <strong>the</strong> camera <strong>and</strong> <strong>the</strong> persistence of <strong>the</strong> luminescence<br />

in <strong>the</strong> diam<strong>on</strong>d. Diam<strong>on</strong>d has high <strong>the</strong>rmal c<strong>on</strong>ductivity <strong>and</strong> <strong>the</strong> film, mounted in<br />

a water-cooled assembly, was able to withst<strong>and</strong> <strong>the</strong> full pink <strong>beam</strong> <strong>on</strong> BL46XU at<br />

Spring-8. Power loading is less of an issue with a free electr<strong>on</strong> laser <strong>beam</strong> but <strong>the</strong>re<br />

is a significant risk of ablati<strong>on</strong> if <strong>the</strong> diam<strong>on</strong>d is exposed to <strong>the</strong> full intensity of <strong>the</strong><br />

fundamental. A fur<strong>the</strong>r c<strong>on</strong>cern is that <strong>the</strong> CVD diam<strong>on</strong>d is polycrystalline <strong>and</strong> thus<br />

<strong>the</strong>re may be significant disrupti<strong>on</strong> to a coherent X-ray <strong>beam</strong> <strong>on</strong> passing through <strong>the</strong><br />

film.<br />

An alternative approach to intensity sampling is to Bragg reflect part of <strong>the</strong> <strong>beam</strong><br />

from a thin crystal <strong>on</strong>to a 2-D CCD detector whilst most of <strong>the</strong> <strong>beam</strong> intensity passes<br />

through <strong>the</strong> crystal. This also allows <strong>the</strong> <strong>beam</strong> to be fully imaged <strong>and</strong> so profile<br />

informati<strong>on</strong> to be extracted, Fajardo <strong>and</strong> Ferrer used a 500 µm thickberyllium crystal<br />

in a white <strong>beam</strong> from an undulator at <strong>the</strong> ESRF[192]. The crystal was set a 45 ◦ <strong>and</strong><br />

so measured X-rays at 4.45 keV. The quality of <strong>the</strong> image is degraded by <strong>the</strong> effects<br />

of mosaic spread <strong>and</strong> <strong>the</strong> finite extincti<strong>on</strong> depth of <strong>the</strong> crystal. This does not affect


128 8. Beam cross-secti<strong>on</strong> diagnostics<br />

centroiding resoluti<strong>on</strong> but does affect any profile measurement.<br />

Themajor disadvantage is <strong>the</strong>spectral dependenceof <strong>the</strong>reflected<strong>and</strong>transmitted<br />

<strong>beam</strong>s.Varying <strong>the</strong> Bragg angle to work at different wavelengths would make a very<br />

complicated system. Fur<strong>the</strong>rmore, in <strong>the</strong> c<strong>on</strong>text of a narrow-b<strong>and</strong>width free electr<strong>on</strong><br />

laser source, <strong>the</strong> impact <strong>on</strong> <strong>the</strong> transmitted <strong>beam</strong> will be significant. The crystal acts<br />

as a b<strong>and</strong>-cut filter <strong>and</strong>, because a significant fracti<strong>on</strong> of <strong>the</strong> radiati<strong>on</strong> could lie inside<br />

<strong>the</strong>crystal rocking curvewidth, <strong>the</strong>notchin <strong>the</strong>transmitted spectrumwould be large.<br />

The mosaic spread of <strong>the</strong> crystal is also likely to cause str<strong>on</strong>g diffractive disrupti<strong>on</strong> to<br />

<strong>the</strong> transmitted <strong>beam</strong>. Finally, <strong>the</strong> technique is <strong>on</strong>ly applicable to hard X-rays where<br />

adequate transmissi<strong>on</strong> through <strong>the</strong> crystal is possible.<br />

At lower phot<strong>on</strong> energies, a diffracti<strong>on</strong> grating can be used to split <strong>the</strong> <strong>beam</strong> in<br />

intensity (as discussed previously in chapter 5). This can be utilized in <strong>on</strong>e of two<br />

ways; ei<strong>the</strong>r <strong>the</strong> first-order diffracted <strong>beam</strong> is sent to <strong>the</strong> diagnostic <strong>and</strong> <strong>the</strong> zeroth<br />

order to <strong>the</strong> experiment or vice versa.<br />

For a <strong>beam</strong>line that requires a m<strong>on</strong>ochromator, it makes sense to use <strong>the</strong> zeroth<br />

order <strong>beam</strong> for <strong>the</strong> diagnostic. In such a case, it is necessary to ensure <strong>the</strong> zeroth<br />

order<strong>beam</strong> directi<strong>on</strong> isfixedas <strong>the</strong>gratingis turnedtotune<strong>the</strong>wavelength, o<strong>the</strong>rwise<br />

<strong>the</strong> diagnostic system will have to move c<strong>on</strong>siderably to follow it.In a normal fixed<br />

included angle m<strong>on</strong>ochromator, this can <strong>on</strong>ly be achieved by using additi<strong>on</strong>al steering<br />

mirror(s) to catch <strong>the</strong> zeroth order <strong>beam</strong> <strong>and</strong> redirect it to <strong>the</strong> diagnostic. Such a<br />

mechanism could be both large <strong>and</strong>complicated since translati<strong>on</strong> as well as rotati<strong>on</strong> of<br />

<strong>the</strong> steering mirrors is likely to be necessary. An alternative approach is possible with<br />

a variable included angle m<strong>on</strong>ochromator such as <strong>the</strong> SX700 mount when operated in<br />

collimated light. In this m<strong>on</strong>ochromator it is possible to operate in <strong>the</strong> so-called ’<strong>on</strong>blaze’<br />

mode <strong>and</strong> this maintains a fixed angle between zeroth <strong>and</strong> first order <strong>beam</strong>s.<br />

In this mode, when <strong>the</strong> angle between <strong>the</strong> first <strong>and</strong> zeroth order <strong>beam</strong>s is ψ, <strong>the</strong><br />

wavelength λ is related to <strong>the</strong> diffracti<strong>on</strong> angle β by<br />

N ·n·λ = 2·sin<br />

� ψ<br />

2<br />

�<br />

cos<br />

� ψ<br />

2 −β<br />

�<br />

(8.1)<br />

Whilst this mode also c<strong>on</strong>veniently maximizes <strong>the</strong> first order efficiency when a blazed<br />

grating with a blaze angle of ψ/2 is used, <strong>the</strong>re is a significant reducti<strong>on</strong> in tuning<br />

range for a grating of a given line density <strong>and</strong> thus more gratings are required to cover<br />

a wide phot<strong>on</strong> energy range.<br />

In <strong>the</strong> alternative approach when <strong>the</strong> zeroth order is sent to <strong>the</strong> experiment, <strong>the</strong><br />

grating should ideally be kept at a fixed angle to minimize <strong>the</strong> reflecti<strong>on</strong>s in <strong>the</strong><br />

main <strong>beam</strong> path. The diffracted order will thus be reflected at a different angle<br />

as <strong>the</strong> free electr<strong>on</strong> laser wavelength is tuned <strong>and</strong> <strong>the</strong> profile m<strong>on</strong>itor will need to<br />

follow it. This will be easier than in <strong>the</strong> case where <strong>the</strong> grating is turned since<br />

<strong>the</strong> angular range over which <strong>the</strong> diffracted order moves will be smaller. Thus it<br />

should be possible to arrange <strong>the</strong> imaging detector to move <strong>on</strong> a rail to follow <strong>the</strong><br />

<strong>beam</strong>. It must be remembered that <strong>the</strong> diffracted order will need to be imaged<br />

<strong>on</strong>to <strong>the</strong> detector. This is best achieved by using a varied-line-spacing grating that<br />

also allows aberrati<strong>on</strong>s to be corrected. This is basically <strong>the</strong> approach used in <strong>the</strong><br />

diagnostic spectrometer at Flash[193], discussed elsewhere. When applied to spatial<br />

imaging,<strong>the</strong>re are two important c<strong>on</strong>siderati<strong>on</strong>s when interpreting <strong>the</strong> measurement.<br />

Firstly, residual aberrati<strong>on</strong>s could still be present <strong>and</strong> sec<strong>on</strong>dly, <strong>the</strong> image will be<br />

blurred due to <strong>the</strong> dispersi<strong>on</strong> of <strong>the</strong> pulse spectrum. A detailed design study would


8.8. Spot size 129<br />

need to be performed to investigate whe<strong>the</strong>r this technique could give a useful pulse<br />

profile diagnostic.<br />

Ano<strong>the</strong>r factor that should be remembered is that <strong>the</strong> positi<strong>on</strong> of <strong>the</strong> <strong>beam</strong> at <strong>the</strong><br />

diagnostic will be dependent <strong>on</strong> <strong>the</strong> grating angle <strong>and</strong> thus <strong>the</strong> technique is not ideal<br />

for getting informati<strong>on</strong> <strong>on</strong> absolute positi<strong>on</strong>s <strong>and</strong> how <strong>the</strong> <strong>beam</strong> is moving.<br />

8.8 Spot size<br />

From <strong>the</strong> users’ perspective, <strong>on</strong>e of <strong>the</strong> critical factors determining <strong>the</strong> performance<br />

of a free electr<strong>on</strong> laser is <strong>the</strong> size <strong>and</strong> quality of <strong>the</strong> focused <strong>beam</strong> spot. For many<br />

experiments, <strong>the</strong> requirement is for as small a focused spot as possible to maximize<br />

<strong>the</strong> flux density or fluence <strong>and</strong> to restrict interacti<strong>on</strong> to a specific, targeted sample<br />

area.Transverse intensity profiling is also critical when aligning <strong>the</strong> focusing optics<br />

to optimize <strong>the</strong> spot shape by minimizing any asymmetry <strong>and</strong> eliminating tails <strong>and</strong><br />

flares. A precise measurement of <strong>the</strong> spot size is crucial to determining <strong>the</strong> absolute<br />

flux density, which must be known for some experiments.<br />

The most desirable soluti<strong>on</strong> for spot size determinati<strong>on</strong> would have <strong>the</strong> ability<br />

to directly image <strong>the</strong> spot with a suitable two-dimensi<strong>on</strong>al, high resoluti<strong>on</strong>, high<br />

repetiti<strong>on</strong> rate detector which is suitably robust to accept <strong>the</strong> unattenuated <strong>beam</strong>.<br />

However, extensive experience <strong>on</strong> synchrotr<strong>on</strong> sources has dem<strong>on</strong>strated that <strong>the</strong><br />

current generati<strong>on</strong> of CCD <strong>and</strong> proporti<strong>on</strong>al gas-filled detectors can be permanently<br />

damaged bymomentaryexposureto<strong>the</strong>unattenuatedX-ray<strong>beam</strong>ofarelatively large<br />

(several millimeters) diameter. Whilst such devices should not be excluded from a<br />

survey of possible techniques it is reas<strong>on</strong>able to suggest that <strong>the</strong> majority of 2-D<br />

imaging systems will require significant attenuati<strong>on</strong> <strong>and</strong> shielding to provide a usable<br />

service life <strong>on</strong> a free electr<strong>on</strong> laser source. It should be noted that attenuating <strong>the</strong><br />

<strong>beam</strong> using solid filters or foils has been dem<strong>on</strong>strated to affect <strong>the</strong> <strong>beam</strong> wavefr<strong>on</strong>t<br />

at Flash although <strong>the</strong> effect of using a gas attenuators is negligible 7 .<br />

The remainder of this secti<strong>on</strong> is <strong>the</strong>refore subdivided into two secti<strong>on</strong>s covering<br />

those techniques that can be used with <strong>the</strong> unattenuated <strong>beam</strong>, <strong>and</strong> those for which<br />

significant attenuati<strong>on</strong> will be required.<br />

Techniques useable with unattenuated <strong>beam</strong>s<br />

Ablati<strong>on</strong> crater analysis<br />

One of <strong>the</strong> most established methods of determining <strong>the</strong> <strong>beam</strong> cross secti<strong>on</strong> in c<strong>on</strong>venti<strong>on</strong>al<br />

high power UV-Vis-IRlasers is byanalysis of <strong>the</strong> laser ablati<strong>on</strong> crater imprinted<br />

into a well characterized sample, predominantly PMMA (polymethyl methacrylate).<br />

PMMA is used extensively because of its well characterized ablati<strong>on</strong> characteristics<br />

across a wide wavelength spectrum, its short heat diffusi<strong>on</strong> length <strong>and</strong> <strong>the</strong> predominance<br />

of n<strong>on</strong>-<strong>the</strong>rmal processes when exposed to ultra-short laser pulses. The low<br />

ablati<strong>on</strong> threshold of PMMA also allows characterizati<strong>on</strong> to be performed <strong>on</strong> less<br />

efficient or highly attenuated <strong>beam</strong>lines.<br />

Techniques exist to comprehensively rec<strong>on</strong>struct <strong>the</strong> <strong>beam</strong> profile from <strong>the</strong> ablative<br />

imprint[194]. The <strong>beam</strong> can be characterized at any point al<strong>on</strong>g <strong>the</strong> <strong>beam</strong>line or<br />

7 P. Juranić et al. Desy internal presentati<strong>on</strong>.


130 8. Beam cross-secti<strong>on</strong> diagnostics<br />

Open multiplier<br />

Diff. pumping<br />

TOF<br />

I<strong>on</strong>s I<strong>on</strong>s<br />

±2 cm<br />

-<br />

+<br />

-<br />

+<br />

GMD<br />

e -<br />

Faraday cup<br />

Faraday cup<br />

FEL <strong>beam</strong><br />

Figure 8.11: Setup for <strong>the</strong> i<strong>on</strong> yield saturati<strong>on</strong> measurement at a <strong>beam</strong> focus.<br />

end stati<strong>on</strong> where it is possible to place a sample of PMMA. The material is readily<br />

available, inexpensive <strong>and</strong> can be easily shaped. It is particularly suitable for<br />

characterizing <strong>the</strong> <strong>beam</strong> profile in user-supplied sample holders <strong>and</strong> envir<strong>on</strong>mental<br />

chambers.<br />

Despite all <strong>the</strong>se advantages, <strong>the</strong>re are very significant practical barriers to using<br />

crater analysis bey<strong>on</strong>d <strong>the</strong> initial commissi<strong>on</strong>ing stage. Measurements can <strong>on</strong>ly be<br />

taken of a single pulse (although some form of carousel or sample changer could<br />

<strong>the</strong>oretically be used for verylow repetiti<strong>on</strong> rates). The technique is clearly disruptive<br />

to any experimental data collecti<strong>on</strong> <strong>and</strong> requires significant time tomount <strong>and</strong> remove<br />

<strong>the</strong> ablati<strong>on</strong> sample. Although it may be possible to make a limited determinati<strong>on</strong><br />

of <strong>the</strong> <strong>beam</strong> cross-secti<strong>on</strong> using in-situ optical microscopy, detailed analysis of <strong>the</strong><br />

<strong>beam</strong> quality will be c<strong>on</strong>ducted offline. The measurements to determine <strong>the</strong> crater<br />

profile typically incorporate both optical <strong>and</strong> atomic force microscopy,so results are<br />

far from instantaneous. It is possible to envisage a certain degree of automati<strong>on</strong> being<br />

employed if <strong>the</strong> number of routine measurements justified <strong>the</strong> investment. Even so,<br />

it is difficult to imagine <strong>the</strong> time between sample exposure <strong>and</strong> accurate <strong>beam</strong> cross<br />

secti<strong>on</strong> determinati<strong>on</strong> being reduced below <strong>the</strong> level of ”several hours”.<br />

Photoi<strong>on</strong>isati<strong>on</strong> saturati<strong>on</strong> of rare gases<br />

An advanced, n<strong>on</strong>-disruptive technique for spot size minimizati<strong>on</strong> has been developed<br />

at Flash by A.A. Sorokin et al. [195]. This is based <strong>on</strong> <strong>the</strong> saturati<strong>on</strong> effect of <strong>the</strong><br />

photoi<strong>on</strong>isati<strong>on</strong> of rare gases. At lower irradiance levels <strong>the</strong>re exists a linear relati<strong>on</strong>ship<br />

between <strong>the</strong> number of incident phot<strong>on</strong>s <strong>and</strong> <strong>the</strong> number of i<strong>on</strong>s generated.<br />

However as <strong>the</strong> irradiance increases (e.g. due to <strong>the</strong> reducti<strong>on</strong> of <strong>the</strong> spot size at <strong>the</strong><br />

ideal focus positi<strong>on</strong>) <strong>the</strong> fracti<strong>on</strong> of <strong>the</strong> atoms in <strong>the</strong> interacti<strong>on</strong> regi<strong>on</strong> that are i<strong>on</strong>ized<br />

approaches unity. The i<strong>on</strong> yield thus diminishes with respect to <strong>the</strong> number of<br />

incident phot<strong>on</strong>s as <strong>the</strong> interacti<strong>on</strong> regi<strong>on</strong> <strong>and</strong> focus coincide. This saturati<strong>on</strong> effect<br />

can be measured with an apertured time-of-flight (TOF) spectrometer mounted perpendicular<br />

to <strong>the</strong> <strong>beam</strong> directi<strong>on</strong> (see secti<strong>on</strong> 7.3, see page 104), whilst measurement<br />

of <strong>the</strong> absolute number of phot<strong>on</strong>s per pulse is made using a gas m<strong>on</strong>itor detector<br />

(<strong>the</strong> GMD presented in secti<strong>on</strong> 7.2, see page 100) <strong>–</strong> Figure 8.11. Careful selecti<strong>on</strong> of<br />

<strong>the</strong> target gas type <strong>and</strong> pressure relative to <strong>the</strong> energy of <strong>the</strong> phot<strong>on</strong> <strong>beam</strong> is required<br />

to restrict <strong>the</strong> photoi<strong>on</strong>isati<strong>on</strong> process to <strong>on</strong>e-phot<strong>on</strong> single i<strong>on</strong>izati<strong>on</strong>[196].


8.9. Techniques requiring attenuated <strong>beam</strong>s 131<br />

The technique is not quantitative; ra<strong>the</strong>r <strong>the</strong> maximum level of saturati<strong>on</strong> of <strong>the</strong><br />

photoi<strong>on</strong>isati<strong>on</strong> signal indicates <strong>the</strong> optimum focus positi<strong>on</strong>. Fur<strong>the</strong>r diagnostics are<br />

requiredtoquantify<strong>the</strong>resultant<strong>beam</strong>size,although photoi<strong>on</strong>isati<strong>on</strong> saturati<strong>on</strong>could<br />

be calibrated using a quantitative process (e.g. ablati<strong>on</strong> crater analysis) for a given<br />

lateral positi<strong>on</strong> <strong>and</strong> gas pressure. One benefit of <strong>the</strong> technique is that it is n<strong>on</strong>disruptive,<br />

since both <strong>the</strong> TOF spectrometer <strong>and</strong> <strong>the</strong> GMD do not impinge directly<br />

<strong>on</strong> <strong>the</strong> incident phot<strong>on</strong>s.The system has currently been tested using macropulses of<br />

Flash free electr<strong>on</strong> laser at a phot<strong>on</strong> energy of 38 eV.<br />

8.9 Techniques requiring attenuated <strong>beam</strong>s<br />

Wire, knife-edge <strong>and</strong> slit scans<br />

For <strong>the</strong> majority of existing synchrotr<strong>on</strong>-based micro-focus experiments, slit scanning<br />

methods are usedtocharacterize <strong>the</strong> <strong>beam</strong> size <strong>and</strong> profile[197]. Ingeneral, diffracti<strong>on</strong><br />

or scattering effects from <strong>the</strong> scanning object are not taken into c<strong>on</strong>siderati<strong>on</strong> when<br />

calculating <strong>the</strong> <strong>beam</strong> size. These techniques are covered in more detail in relati<strong>on</strong> to<br />

general <strong>beam</strong> profiling in secti<strong>on</strong> 8.4. For synchrotr<strong>on</strong> radiati<strong>on</strong> sources, any <strong>the</strong>rmal<br />

loading issues are overcome by implementing water cooling to <strong>the</strong> slit jaws or apertures.<br />

However,<strong>the</strong> fluence levels at <strong>the</strong> focus of a pulsed free electr<strong>on</strong> laser source will<br />

accentuate <strong>the</strong> difficulties associated with ablati<strong>on</strong>, for which significant attenuati<strong>on</strong><br />

will be <strong>the</strong> <strong>on</strong>ly soluti<strong>on</strong>.<br />

Photographic film<br />

This is included for <strong>the</strong> sake of completeness, <strong>and</strong> as a dem<strong>on</strong>strati<strong>on</strong> that a relatively<br />

low technology soluti<strong>on</strong> can prove extremely useful. Film was used extensively <strong>on</strong><br />

sec<strong>on</strong>d-generati<strong>on</strong> synchrotr<strong>on</strong>s e.g. SRS, Daresbury UK, to produce a permanent<br />

record of <strong>the</strong> size <strong>and</strong> shape of <strong>the</strong> focused X-ray <strong>beam</strong>. In this example <strong>the</strong> film used<br />

was Polaroid 55 large-format self-developing film which was chosen for its ease of use,<br />

speed, limited cost (at <strong>the</strong> time) <strong>and</strong> small grain size which led to comparatively high<br />

resoluti<strong>on</strong> images.Its thinness also made it practical in an experimental set-up as it<br />

could be placed practically anywhere in <strong>the</strong> hard X-ray <strong>beam</strong> path without disrupti<strong>on</strong><br />

to user equipment.<br />

One could suggest that photographic film is still viable for free electr<strong>on</strong> laser applicati<strong>on</strong>s<br />

since <strong>the</strong> grain size of high quality film stock can be as small as 0.5 µm,<br />

which is far smaller than <strong>the</strong> photo site spacing of even <strong>the</strong> best CCD sensors. Some<br />

form of pulse selecti<strong>on</strong> would be required to give single-pulse imaging <strong>and</strong> high levels<br />

of attenuati<strong>on</strong> would be required to avoid ablati<strong>on</strong> damage <strong>and</strong> over-exposure of<br />

<strong>the</strong> film. Since <strong>the</strong> film is by its nature disposable, damage would at least not be a<br />

catastrophe. In practical terms though, this is of little relevance since suitable film<br />

is now difficult to obtain in appropriate quantities <strong>and</strong> requires lengthy processing,<br />

digitizati<strong>on</strong> <strong>and</strong> subsequent image analysis. Polaroid self-developing film is no l<strong>on</strong>ger<br />

in producti<strong>on</strong>.<br />

For informati<strong>on</strong>, <strong>the</strong>re is a closely related material which can also be used for <strong>beam</strong><br />

size determinati<strong>on</strong>; namely self-developing X-ray dosimetry film 8 . Similar products<br />

8 E. g. http://www.gafchromic.com.


132 8. Beam cross-secti<strong>on</strong> diagnostics<br />

have been used <strong>on</strong> synchrotr<strong>on</strong>s for <strong>beam</strong> profiling <strong>and</strong> have broadly similar advantages<br />

<strong>and</strong> drawbacks as self-developing photographic film. However, <strong>the</strong> resoluti<strong>on</strong><br />

of dosimetry film (ca. 100 µm) is significantly worse than that for high quality photographic<br />

film <strong>and</strong> will be too low for many focused <strong>beam</strong> measurements <strong>on</strong> a free<br />

electr<strong>on</strong> laser source.<br />

Gas-filled detectors<br />

Gas proporti<strong>on</strong>al detecti<strong>on</strong> systems for X-rays are comm<strong>on</strong>ly based <strong>on</strong> ei<strong>the</strong>r wire<br />

grids[198] or, more recently, <strong>on</strong> metal strips lithographically printed <strong>on</strong>to ei<strong>the</strong>r traditi<strong>on</strong>al<br />

circuitboards orglass substrates[199]. Existinglarge-field commercial detectors<br />

typically have pixel sizes of <strong>the</strong> order of tens of micr<strong>on</strong>s, although it should be possible<br />

to lithographically produce dedicated <strong>beam</strong> imaging detectors with smaller pixel<br />

sizes,<strong>and</strong> resoluti<strong>on</strong> could be fur<strong>the</strong>r improved using pixel interpolati<strong>on</strong> algorithms.<br />

However, although <strong>the</strong>se detectors are sought after for <strong>the</strong>ir potential for measuring<br />

high global count rates <strong>and</strong> energy discriminati<strong>on</strong>,<strong>the</strong>y suffer from severe n<strong>on</strong>-linearity<br />

due to space charge effects. Thus high count-rates that are very localized, as would<br />

be experienced with free electr<strong>on</strong> laser <strong>beam</strong> imaging, would cause problems. Additi<strong>on</strong>ally,<br />

at high count rates <strong>the</strong>re are also problems with rapid c<strong>on</strong>taminati<strong>on</strong> of<br />

<strong>the</strong> gas <strong>and</strong> detector wires/elements,requiring very high gas flow rates <strong>and</strong> <strong>the</strong>refore<br />

high maintenance costs. These issues make <strong>the</strong>m highly unsuitable for direct <strong>beam</strong><br />

imaging purposes.<br />

Charge coupled device (CCD)<br />

Direct-detecti<strong>on</strong> CCD cameras for X-ray energies are commercially available with<br />

pixel sizes down to 8 x 8 µm 9 . This is not sufficient for <strong>the</strong> str<strong>on</strong>gest focusing that<br />

will be used <strong>on</strong> free electr<strong>on</strong> laser, where spot sizes around 1 µm or less will achieved.<br />

Depending <strong>on</strong> <strong>the</strong> energy range to be covered, a selecti<strong>on</strong> can be made between back<br />

thinned, fr<strong>on</strong>t illuminated <strong>and</strong> deep depleti<strong>on</strong> CCD types. However, <strong>the</strong>se sensors are<br />

specifically not designed for high flux density applicati<strong>on</strong>s; saturati<strong>on</strong> <strong>and</strong> damage<br />

occur at low irradiance.<br />

Significant effort has been channelled intodevelopingradiati<strong>on</strong> hard CCD cameras,<br />

<strong>the</strong> initial driver coming from improving <strong>the</strong> l<strong>on</strong>gevity of space-borne astr<strong>on</strong>omical<br />

instruments <strong>and</strong> remote observati<strong>on</strong> systems in <strong>the</strong> nuclear industry. Defects can be<br />

generated in <strong>the</strong> bulksilic<strong>on</strong> through exposure toradiati<strong>on</strong>, <strong>and</strong><strong>the</strong>se defects can <strong>the</strong>n<br />

become electrically active, leading to fur<strong>the</strong>r space-charge, charge leakage <strong>and</strong> charge<br />

trapping issues in CCDs[200]. Three key approaches have been taken to minimize <strong>the</strong><br />

degradati<strong>on</strong> effects inCCDs; i) engineering soluti<strong>on</strong>s such as guard structures, voltage<br />

biasing <strong>and</strong> a reducti<strong>on</strong> in <strong>the</strong> thickness of charge channels, ii) material developments<br />

such as reducing <strong>the</strong> c<strong>on</strong>taminati<strong>on</strong> defects in silic<strong>on</strong> or replacement of silic<strong>on</strong> with<br />

alternative insulator materials <strong>and</strong> iii) alternative structures such as p-channel CCDs<br />

or charge injecti<strong>on</strong> devices (CIDs).<br />

CIDs individually address each pixel in <strong>the</strong> detector <strong>and</strong> do not suffer from leakage<br />

of stored charge from <strong>on</strong>e pixel to ano<strong>the</strong>r. Commercial nuclear inspecti<strong>on</strong> CID<br />

systems are available for X-rays, infrared <strong>and</strong> <strong>the</strong> ultra-violett with a stated pixel<br />

9 See, for example, www.hamamatsu.com.


8.10. Positi<strong>on</strong> <strong>and</strong> centroiding 133<br />

size of 11.5 x 11.5 µm <strong>and</strong> a frame rate of up to 25 Hz 10 . A Thermo Scientific CID<br />

camera has been employed as part of <strong>the</strong> transverse <strong>beam</strong> profiler for <strong>the</strong> IFMIF-<br />

EVEDA prototype deuter<strong>on</strong> accelerator[201].<br />

As discussed in secti<strong>on</strong> 8.3, luminescent screens in c<strong>on</strong>juncti<strong>on</strong> with CCD cameras<br />

can be used for indirect detecti<strong>on</strong>. Using a scintillati<strong>on</strong> screen, imaging of <strong>the</strong> Xray<br />

<strong>beam</strong> from SRS dipole <strong>beam</strong>line 8.2 has been dem<strong>on</strong>strated using an attenuated<br />

Phot<strong>on</strong>ic Science CCD system. At <strong>the</strong> intensity levels experienced <strong>the</strong>re,<strong>the</strong> main<br />

risk of damage was assumed to be <strong>the</strong>rmal damage to <strong>the</strong> screen. For free electr<strong>on</strong><br />

laser <strong>the</strong> highest risk of damage will be ablati<strong>on</strong> of <strong>the</strong> screen, <strong>and</strong> hence attenuati<strong>on</strong><br />

of <strong>the</strong> <strong>beam</strong> will be essential.<br />

Multichannel plate (MCP)<br />

A number of commercially available <strong>beam</strong> imaging systems exist based <strong>on</strong> fibrecoupled<br />

MCP technology 11 <strong>and</strong> a large range of c<strong>on</strong>versi<strong>on</strong> media exist to enable<br />

<strong>the</strong> imaging of neutr<strong>on</strong> <strong>and</strong> electr<strong>on</strong> <strong>beam</strong>s in additi<strong>on</strong> to IR, UV <strong>and</strong> X-ray radiati<strong>on</strong>.<br />

However, <strong>the</strong>se systems are of limited applicati<strong>on</strong> to free electr<strong>on</strong> laser <strong>beam</strong>s<br />

since <strong>the</strong> inherent amplificati<strong>on</strong> due to <strong>the</strong> MCP is at odds with <strong>the</strong> requirement to<br />

str<strong>on</strong>gly attenuate <strong>the</strong> <strong>beam</strong>.<br />

Solid State Detectors<br />

For many applicati<strong>on</strong>s <strong>on</strong> existing synchrotr<strong>on</strong> sources, silic<strong>on</strong> pixel detectors (often<br />

referred to as M<strong>on</strong>olithic Active Pixel Sensors, MAPS) or hybrid pixel detectors,<br />

typified by <strong>the</strong> PILATUS system[202], are seen as a significant emergent technology.<br />

They combine volume manufacture with small pixel sizes,large dynamic range, high<br />

count-rates <strong>and</strong> fast readout. However, in comm<strong>on</strong> with CCD-based systems <strong>the</strong><br />

irreliance <strong>on</strong> silic<strong>on</strong> based lithographic plates makes <strong>the</strong>m susceptible to radiati<strong>on</strong><br />

damage at modest <strong>beam</strong> intensities.<br />

8.10 Positi<strong>on</strong> <strong>and</strong> centroiding<br />

Sampling techniques<br />

Beam centroidingis typically performed bysamplingpart of<strong>the</strong> <strong>beam</strong>, ei<strong>the</strong>r spatially<br />

or in intensity. The basic objective in measuring <strong>the</strong> centroid is to measure <strong>the</strong> <strong>beam</strong><br />

positi<strong>on</strong> <strong>and</strong> angle mainly for <strong>the</strong> purposes of <strong>beam</strong> stabilizati<strong>on</strong> through feedback<br />

c<strong>on</strong>trol.<br />

Solid state photo-emissi<strong>on</strong> m<strong>on</strong>itors<br />

Blade m<strong>on</strong>itors are an example of spatial sampling <strong>and</strong> are widely used <strong>on</strong> synchrotr<strong>on</strong>s.<br />

Mortazavi et al. [203] describe an early implementati<strong>on</strong> at <strong>the</strong> NSLS<br />

(Brookhaven) giving a sensitivity of a few micr<strong>on</strong>s. Thin tungsten blades positi<strong>on</strong>ed<br />

edge-<strong>on</strong> to <strong>the</strong> <strong>beam</strong> intercept <strong>the</strong> periphery of <strong>the</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> induced photocurrent<br />

is proporti<strong>on</strong>al to <strong>the</strong> amount of <strong>beam</strong> <strong>the</strong>y intercept. If two identical blades are<br />

10 www.<strong>the</strong>rmo.com/com/cda/product/.<br />

11 www.sciner.com/MCP/index.htm <strong>and</strong> www.<strong>beam</strong>imaging.com.


134 8. Beam cross-secti<strong>on</strong> diagnostics<br />

positi<strong>on</strong>ed ei<strong>the</strong>r side of <strong>the</strong> <strong>beam</strong>, <strong>the</strong>n <strong>the</strong> relative intensity <strong>the</strong>y see is a measure<br />

of <strong>the</strong> <strong>beam</strong> positi<strong>on</strong> relative to <strong>the</strong> centre of <strong>the</strong> gap between <strong>the</strong> blades. These<br />

devices are thus centroid m<strong>on</strong>itors giving <strong>beam</strong> positi<strong>on</strong> relative to <strong>the</strong> blade gap <strong>–</strong><br />

but <strong>the</strong>re is an underlying assumpti<strong>on</strong> that <strong>the</strong> <strong>beam</strong> intensity profile is symmetric.<br />

Johns<strong>on</strong> <strong>and</strong> Oversluizen[204] also assert that ”<strong>the</strong> apparent deviati<strong>on</strong> of <strong>the</strong> phot<strong>on</strong><br />

<strong>beam</strong> from <strong>the</strong> m<strong>on</strong>itor centroid depends <strong>on</strong> <strong>the</strong> <strong>on</strong> <strong>the</strong> size of <strong>the</strong> [...] <strong>beam</strong> relative<br />

to <strong>the</strong> blade gap”, though it is not clear why this should be if <strong>the</strong> detector <strong>and</strong><br />

phot<strong>on</strong> <strong>beam</strong> are symmetric (<strong>the</strong>re are however problems created by stray light from<br />

adjacent magnets, q.v.). It is however clearly important that <strong>the</strong> blades be identical<br />

both physically <strong>and</strong> also in terms of photoelectric yield. The latter is a c<strong>on</strong>diti<strong>on</strong><br />

that is hard to ensure given that photoemissi<strong>on</strong> is a predominantly surface effect <strong>and</strong><br />

so susceptible to c<strong>on</strong>taminati<strong>on</strong> from <strong>the</strong> residual vacuum. Aging of <strong>the</strong> blades in a<br />

white synchrotr<strong>on</strong> <strong>beam</strong> is likely <strong>and</strong> if <strong>the</strong> blades age differently <strong>the</strong>n this would lead<br />

to l<strong>on</strong>g-term change in <strong>the</strong> null positi<strong>on</strong> of <strong>the</strong> <strong>beam</strong> relative to <strong>the</strong> gap.<br />

A similar approach uses two parallel horiz<strong>on</strong>tal wires or rods with a fixed gap[205].<br />

The advantage of <strong>the</strong> blade approach is that blades are easier to cool than wires <strong>and</strong><br />

less invasive than rods. Alkire et al. [206] report a relative accuracy of ±5 µm for<br />

<strong>the</strong>ir m<strong>on</strong>itor which uses 1.5 mm diameter tungsten rods of 12 cm length <strong>and</strong> with a<br />

centre to centre spacing of 7.9 mm.<br />

Tungsten or molybdenum are often used as <strong>the</strong> blade material since <strong>the</strong>ir high<br />

melting point <strong>and</strong> hardness makes <strong>the</strong>m resistant to <strong>the</strong> high powers produced by<br />

3 rd generati<strong>on</strong> synchrotr<strong>on</strong> sources. However, when high-K undulators are used <strong>on</strong><br />

high-energy synchrotr<strong>on</strong>s <strong>the</strong> total power can be such that accidental exposure of<br />

such materials to <strong>the</strong> central part of <strong>the</strong> <strong>beam</strong> could be damaging. At <strong>the</strong> APS,<br />

CVD diam<strong>on</strong>d was chosen since it has 10 times higher <strong>the</strong>rmal c<strong>on</strong>ductivity than<br />

molybdenum, <strong>and</strong> lower <strong>the</strong>rmal expansi<strong>on</strong> allied to high strength <strong>and</strong> stiffness[207].<br />

These m<strong>on</strong>itors have sub-micr<strong>on</strong> sensitivity. For initial tests, <strong>the</strong> diam<strong>on</strong>d was coated<br />

with tungsten to give high photo-emissivity <strong>and</strong> electrical c<strong>on</strong>ductivity. Later, a 1<br />

µm gold coating <strong>on</strong> 150 µm diam<strong>on</strong>d blades was used[208]. This approach raises <strong>the</strong><br />

prospect that it may be possible to coat <strong>the</strong> blades with a low-Z material (e.g. carb<strong>on</strong><br />

or beryllium) that would resist ablati<strong>on</strong> from a free electr<strong>on</strong> laser <strong>beam</strong>, which would<br />

o<strong>the</strong>rwise be a major c<strong>on</strong>cern when using this type of m<strong>on</strong>itor <strong>on</strong> a such a <strong>beam</strong>.<br />

An advantage of <strong>the</strong> blade type m<strong>on</strong>itor is that extra blades can be added in<br />

various c<strong>on</strong>figurati<strong>on</strong>s such as a vertical cross[209], diag<strong>on</strong>al cross [210], etc. <strong>–</strong> see<br />

Figure 8.12. Thus, a single m<strong>on</strong>itor allows bi-axial positi<strong>on</strong>al informati<strong>on</strong> to be<br />

calculated. Two such m<strong>on</strong>itors l<strong>on</strong>gitudinally separated give <strong>beam</strong> angle informati<strong>on</strong>,though<br />

each should use a different blade arrangement to prevent <strong>the</strong> blades of<br />

<strong>the</strong> sec<strong>on</strong>d m<strong>on</strong>itor from lying in <strong>the</strong> shadow of <strong>the</strong> blades of <strong>the</strong> first m<strong>on</strong>itor[207].<br />

For <strong>the</strong> centroid measurement to give an absolute positi<strong>on</strong> <strong>and</strong> angle, <strong>the</strong> m<strong>on</strong>itors<br />

must be accurately mapped to an external reference frame. However, if <strong>the</strong>re is an<br />

asymmetry in <strong>the</strong> resp<strong>on</strong>se of <strong>the</strong> blades (e.g. due to surface c<strong>on</strong>taminati<strong>on</strong>) <strong>the</strong>n,<br />

when <strong>the</strong> m<strong>on</strong>itor is nulled, <strong>the</strong> <strong>beam</strong> centroid will not be centered <strong>on</strong> <strong>the</strong> gap. There<br />

is thus an error in <strong>the</strong> inferred absolute positi<strong>on</strong> relative to <strong>the</strong> external reference<br />

frame. Thus <strong>the</strong> absolute positi<strong>on</strong>al accuracy of a blade type m<strong>on</strong>itor is not always<br />

easy to define.<br />

In<strong>the</strong>caseofsynchrotr<strong>on</strong>lightfromdipoles, <strong>the</strong>sedetectorscanbemadeeffectively<br />

transparent to <strong>the</strong> X-ray <strong>beam</strong> by being designed to intercept <strong>on</strong>ly <strong>the</strong> UV light that<br />

hasamuchlarger openingangle than<strong>the</strong>X-rays, whichthuspassundisturbedthrough<br />

<strong>the</strong> blade gap (Figure 8.13). The same approach can also be used with undulator


8.10. Positi<strong>on</strong> <strong>and</strong> centroiding 135<br />

(a) (b) (c) (d)<br />

Figure 8.12: Possible arrangements for bi-axial blade m<strong>on</strong>itors. (a) simple vertical cross; (b) a diag<strong>on</strong>al cross<br />

allows better horiz<strong>on</strong>tal sensitivity when <strong>the</strong> horiz<strong>on</strong>tal opening angle changes c<strong>on</strong>siderably if an undulator is<br />

tuned from low to high K; (c) <strong>and</strong> (d) upstream <strong>and</strong> downstream arrangements used at <strong>the</strong> APS to prevent<br />

shadowing effects [40] (<strong>the</strong> tilted horiz<strong>on</strong>tal blades reduce <strong>the</strong> sensitivity to dipole radiati<strong>on</strong>).<br />

UV<br />

M<strong>on</strong>itor blades<br />

X-rays<br />

Figure 8.13: Figure 9 -A blade m<strong>on</strong>itor can be transparent at short wavelengths <strong>on</strong> a synchrotr<strong>on</strong> source due<br />

to <strong>the</strong> reducti<strong>on</strong> in opening angle as <strong>the</strong> wavelength decreases.<br />

sources since <strong>the</strong> emissi<strong>on</strong> outside <strong>the</strong> central c<strong>on</strong>e is at a l<strong>on</strong>ger wavelength than <strong>the</strong><br />

fundamental.However, free electr<strong>on</strong> laser radiati<strong>on</strong> does not have <strong>the</strong> same properties<br />

<strong>and</strong> so this approach cannot be used. There will be sp<strong>on</strong>taneous radiati<strong>on</strong> from <strong>the</strong><br />

undulator, but this will be at a low intensity due to <strong>the</strong> low average current <strong>and</strong> may<br />

notbeofsufficientintensitytomeasure. Fur<strong>the</strong>rmore, thisapproachhasitsdrawbacks<br />

since <strong>the</strong> phot<strong>on</strong>s being detected are not <strong>the</strong> <strong>on</strong>es being used in <strong>the</strong> experiment <strong>and</strong><br />

<strong>on</strong>e must assume <strong>the</strong>re is a unique <strong>and</strong> c<strong>on</strong>sistent spatial correlati<strong>on</strong> between <strong>the</strong><br />

spectral comp<strong>on</strong>ents in <strong>the</strong> <strong>beam</strong>.<br />

Whilst <strong>the</strong> range of wavelengths present in a synchrotr<strong>on</strong> <strong>beam</strong> can be used to advantage<br />

as described above, it can lead to problems when using photo-emissive type<br />

m<strong>on</strong>itors with inserti<strong>on</strong> devices <strong>on</strong> synchrotr<strong>on</strong> sources. The electr<strong>on</strong> yield is much<br />

greater at VUV wavelengths <strong>and</strong> thus <strong>the</strong>se m<strong>on</strong>itors are disproporti<strong>on</strong>ately sensitive<br />

to <strong>the</strong> low-level l<strong>on</strong>g-wavelength radiati<strong>on</strong> coming from <strong>the</strong> dipoles <strong>and</strong> steering magnets<br />

that surround <strong>the</strong> inserti<strong>on</strong> device. Extreme lengths have been used to overcome<br />

this problem at <strong>the</strong> APS where <strong>the</strong> machine lattice has been modified to separate <strong>the</strong><br />

inserti<strong>on</strong> device light from <strong>the</strong> stray light in order to facilitate sub-micr<strong>on</strong> level orbit


136 8. Beam cross-secti<strong>on</strong> diagnostics<br />

correcti<strong>on</strong>[211]. Galimberti et al. [212] describe a different approach that makes <strong>the</strong><br />

<strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor <strong>on</strong>ly sensitive to <strong>the</strong> X-rays that are being used in <strong>the</strong> experiment.<br />

They have improved <strong>the</strong> blade m<strong>on</strong>itor by adding electr<strong>on</strong> energy analyzers<br />

to measure <strong>the</strong> blade signals. Thus <strong>the</strong>y can select <strong>on</strong>ly <strong>the</strong> electr<strong>on</strong>s emitted by <strong>the</strong><br />

fundamental radiati<strong>on</strong> <strong>and</strong> so isolate it from any low energy background <strong>and</strong> even<br />

electr<strong>on</strong>s emitted by higher harm<strong>on</strong>ics from <strong>the</strong> undulator. This is a much more sophisticated<br />

approach to that suggested by Warwick et al. [210] in which <strong>the</strong> blades are<br />

reverse-biased to prevent <strong>the</strong> low energy electr<strong>on</strong>s leaving <strong>the</strong> surface. Whe<strong>the</strong>r such<br />

precauti<strong>on</strong>s would be needed <strong>on</strong> a free electr<strong>on</strong> laser source would clearly depend <strong>on</strong><br />

<strong>the</strong> nature of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>transport</strong>, but it would be natural to assume <strong>the</strong> effect<br />

would be much reduced since <strong>the</strong> electr<strong>on</strong> path is straight <strong>and</strong> <strong>the</strong> l<strong>on</strong>g undulator<br />

length should limit <strong>the</strong> acceptance aperture of radiati<strong>on</strong> from o<strong>the</strong>r sources.<br />

In general, blade type m<strong>on</strong>itors when used <strong>on</strong> undulator sources are sensitive to<br />

<strong>the</strong> undulator tuning due to <strong>the</strong> changing radiati<strong>on</strong> pattern as <strong>the</strong> undulator K-value<br />

is changed. This makes positi<strong>on</strong> c<strong>on</strong>trol to sub-micr<strong>on</strong> levels difficult. A smart XBPM<br />

system (SBPM) has been developed at <strong>the</strong> APS in which <strong>the</strong> resp<strong>on</strong>se of <strong>the</strong> XPBM is<br />

automatically characterized under all possible operating c<strong>on</strong>diti<strong>on</strong>s of <strong>the</strong> undulator<br />

so that <strong>the</strong>se effects can be automatically corrected for[208].<br />

The effect of any change in <strong>beam</strong> footprint with wavelength must be c<strong>on</strong>sidered<br />

carefully for free electr<strong>on</strong> laser <strong>beam</strong>s.This source approximates to a coherent Gaussian<br />

source <strong>and</strong> thus <strong>the</strong> divergence will be roughly proporti<strong>on</strong>al to wavelength. We<br />

can <strong>the</strong>refore expect a c<strong>on</strong>siderable change in <strong>beam</strong> footprint at <strong>the</strong> m<strong>on</strong>itor as <strong>the</strong><br />

wavelength of it tuned. This will certainly change <strong>the</strong> sensitivity <strong>and</strong> resoluti<strong>on</strong> of <strong>the</strong><br />

m<strong>on</strong>itor at different wavelengths <strong>and</strong> <strong>the</strong>re is also <strong>the</strong> issue of <strong>the</strong> inferred positi<strong>on</strong><br />

being dependent <strong>on</strong> <strong>beam</strong>size. Possibly more important are <strong>the</strong> risk of ablati<strong>on</strong> <strong>and</strong><br />

<strong>the</strong> disrupti<strong>on</strong> to <strong>the</strong> <strong>beam</strong>. Providing <strong>the</strong> blades sit <strong>on</strong>ly in <strong>the</strong> wings of <strong>the</strong> <strong>beam</strong>,<br />

ablati<strong>on</strong> <strong>and</strong> diffractive disrupti<strong>on</strong> to <strong>the</strong> downstream <strong>beam</strong> may not be an issue.<br />

But if <strong>the</strong> exp<strong>and</strong>s significantly, <strong>the</strong> blades will cut <strong>the</strong> <strong>beam</strong> at a positi<strong>on</strong> of greater<br />

intensity <strong>and</strong> so both <strong>the</strong> risk of ablati<strong>on</strong> <strong>and</strong> diffractive disrupti<strong>on</strong> will increase. It<br />

may thus be necessary for a blade type m<strong>on</strong>itor to have blades that move depending<br />

<strong>on</strong> <strong>the</strong> source’s wavelength.<br />

In order to eliminate <strong>the</strong> sensitivity of a blade m<strong>on</strong>itor to <strong>beam</strong> size, schemes using<br />

two triangular wedge swith a gap between <strong>the</strong>m running at 45 ◦ to <strong>the</strong> horiz<strong>on</strong>tal have<br />

been developed[204] (see Figure 8.14(a)). The gap must be designed to intercept<br />

enough of <strong>the</strong> <strong>beam</strong> to get a decent signal with <strong>the</strong> smallest expected <strong>beam</strong>size (i.e. at<br />

<strong>the</strong> shortest wavelength). This means that any significant increase in <strong>beam</strong> size will<br />

resultinasignificantloss inthroughput. Avariablegapwill <strong>the</strong>refore beessential with<br />

a free electr<strong>on</strong> laser source. Even <strong>the</strong>n, this type of m<strong>on</strong>itor will be more disruptive to<br />

<strong>the</strong> <strong>beam</strong> <strong>and</strong> be at greater risk of ablati<strong>on</strong> <strong>the</strong>n blade m<strong>on</strong>itors <strong>and</strong> so is unlikely to<br />

be useful with a free electr<strong>on</strong> laser source. A refinement of this scheme was reported<br />

by Mitsuhashi <strong>and</strong> tested at Spring-8[213]. Here, two wedge m<strong>on</strong>itors are placed<br />

to just intercept <strong>the</strong> periphery of <strong>the</strong> <strong>beam</strong>, thus reducing <strong>the</strong> fracti<strong>on</strong> of <strong>the</strong> <strong>beam</strong><br />

intercepted <strong>and</strong> allowing bi-axial m<strong>on</strong>itoring. Initially, <strong>the</strong> m<strong>on</strong>itors were composed<br />

of triangular wedges but this design was found to be sensitive undulator gap. Thus a<br />

revised symmetrical scheme was developed <strong>and</strong> tested <strong>–</strong> see Figure 8.14(b) <strong>–</strong> <strong>and</strong> this<br />

reduced <strong>the</strong> sensitivity to <strong>the</strong> gap around ten times. Note <strong>the</strong> wedges are angled to<br />

<strong>the</strong> <strong>beam</strong> to reduce <strong>the</strong> power loading.<br />

The wedge type m<strong>on</strong>itor can be extended to give bi-axial positi<strong>on</strong> detecti<strong>on</strong> by<br />

using a classic quadrant detector. These can be made using metal-foil photodiodes


8.10. Positi<strong>on</strong> <strong>and</strong> centroiding 137<br />

X-ray <strong>beam</strong><br />

Wedge plates<br />

Signal out<br />

(a) (b)<br />

Upper electrodes<br />

Lower electrodes<br />

Figure 8.14: (a) The Wedge positi<strong>on</strong> m<strong>on</strong>itor eliminates <strong>the</strong> sensitivity to <strong>beam</strong> footprint of <strong>the</strong> blade counterpart<br />

but is more invasive; (b) <strong>the</strong> symmetric design of Mitsuhashi et al. reduces <strong>the</strong> <strong>beam</strong> loss <strong>and</strong> has low<br />

sensitivity to undulator tuning.<br />

but are also available commercially using semic<strong>on</strong>ductor juncti<strong>on</strong> photodiodes[214].<br />

One problem with using semi-c<strong>on</strong>ductor devices is that <strong>the</strong>y tend to be insensitive<br />

at <strong>the</strong> edges <strong>and</strong> thus <strong>the</strong> amount of <strong>beam</strong> overlap needed to give a signal is increased<br />

to <strong>the</strong> detriment of <strong>the</strong> transmitted <strong>beam</strong>. Kenney et al. [215] describe how activeedge<br />

silic<strong>on</strong> detectors, which are active to within a few micr<strong>on</strong>s of <strong>the</strong> detector edge,<br />

can be used in various geometries for positi<strong>on</strong>, profile <strong>and</strong> intensity m<strong>on</strong>itoring. For<br />

example, a quadrant detector with a small hole at <strong>the</strong> centre can be used to m<strong>on</strong>itor<br />

<strong>the</strong> stability of a tightly focused <strong>beam</strong>. The <strong>beam</strong> is focused through <strong>the</strong> hole <strong>and</strong><br />

<strong>on</strong>ly <strong>the</strong> periphery of <strong>the</strong> <strong>beam</strong> is stopped by <strong>the</strong> detector. The focus positi<strong>on</strong> is<br />

stabilized using feedback c<strong>on</strong>trol based <strong>on</strong> <strong>the</strong> quadrant signals. An example device<br />

with 100 µm diameter hole is pictured whilst test were made <strong>on</strong> a simpler single<br />

element torus with 200 µm hole. This device showed excellent resp<strong>on</strong>se uniformity<br />

to 12.5 keV X-rays. The effect of diffracti<strong>on</strong> at <strong>the</strong> hole with a coherent free electr<strong>on</strong><br />

laser <strong>beam</strong> is a potential problem with this approach, as is <strong>the</strong> high risk of damaging<br />

<strong>the</strong> detector if <strong>the</strong> central part of <strong>the</strong> <strong>beam</strong> is inadvertently steered <strong>on</strong>to it.<br />

A significant disadvantage of <strong>the</strong> classic quadrant detector is that <strong>the</strong> fracti<strong>on</strong> of<br />

<strong>the</strong> radiati<strong>on</strong> transmitted, i.e. that which can pass through <strong>the</strong> hole in <strong>the</strong> middle,<br />

may not be large enough. Shu et al. at <strong>the</strong> APS[216, 217] have developed a quadrant<br />

detector that has a high transmissi<strong>on</strong> to hard X-rays. The detector is based <strong>on</strong> a<br />

25 mm diameter CVD diam<strong>on</strong>d disc of 150 µm thickness. This has a transmissi<strong>on</strong><br />

of 78% at 10 keV. The quadrant pattern is formed with a 0.2 µm aluminum coating<br />

<strong>on</strong> <strong>the</strong> disc. In its simplest form, <strong>the</strong> photocurrent from <strong>the</strong> aluminum sectors is<br />

independently m<strong>on</strong>itored <strong>and</strong> gives <strong>the</strong> positi<strong>on</strong>al informati<strong>on</strong>.The diam<strong>on</strong>d in this<br />

scheme simply acts as a support for <strong>the</strong> aluminum electrodes that can withst<strong>and</strong> <strong>the</strong><br />

intense synchrotr<strong>on</strong> <strong>beam</strong>.<br />

A more sophisticated approach uses <strong>the</strong> photoc<strong>on</strong>ductive properties of insulatingtype<br />

(IIa) CVD diam<strong>on</strong>d in which <strong>the</strong> diam<strong>on</strong>d becomes c<strong>on</strong>ductive when exposed to<br />

<strong>the</strong>X-rays. Thealuminium patternis replicated <strong>on</strong>bothsidesof<strong>the</strong>diam<strong>on</strong>ddisc <strong>and</strong><br />

a bias applied across <strong>the</strong> fr<strong>on</strong>t <strong>and</strong> back electrodes <strong>–</strong> Figure 8.15. On exposure to Xrays,<br />

<strong>the</strong> diam<strong>on</strong>d becomes c<strong>on</strong>ductive <strong>and</strong> a current flows. Since <strong>the</strong> c<strong>on</strong>ductivity is<br />

dependent<strong>on</strong> <strong>the</strong> absorbed X-raypower, <strong>the</strong> current is proporti<strong>on</strong>al to<strong>the</strong> intercepted<br />

X-ray intensity. Also,<strong>the</strong> sensitivity increases with phot<strong>on</strong> energy <strong>and</strong> <strong>the</strong> detector


138 8. Beam cross-secti<strong>on</strong> diagnostics<br />

Bias<br />

Diam<strong>on</strong>d disc<br />

Al quadrant electrodes<br />

X-ray <strong>beam</strong><br />

Figure 8.15: Quadrant detector based <strong>on</strong> <strong>the</strong> photoc<strong>on</strong>ductivity of diam<strong>on</strong>d.<br />

Output signals<br />

less sensitive to stray light from bending magnets etc.<br />

There are o<strong>the</strong>r ways of spatially sampling <strong>the</strong> <strong>beam</strong> such as using a pin-hole<br />

array[218] but <strong>the</strong>se are significantly more invasive <strong>and</strong> ablati<strong>on</strong> damage is highly<br />

likely.<br />

Solid state fluorescence m<strong>on</strong>itors<br />

Measurement of <strong>the</strong> incident phot<strong>on</strong> intensity <strong>on</strong> <strong>the</strong> m<strong>on</strong>itor by recording <strong>the</strong> photoemissi<strong>on</strong><br />

in some way is widely used because <strong>the</strong> electr<strong>on</strong> yields are high <strong>and</strong> photoemissi<strong>on</strong><br />

is <strong>the</strong> dominant de-excitati<strong>on</strong> process until <strong>the</strong> Ga K-edge at ∼ 10 keV.<br />

Never<strong>the</strong>less, <strong>the</strong> radiative yield is appreciable above c. 5 keV <strong>and</strong> thus fluorescence<br />

presents an alternative detecti<strong>on</strong> route. Alkire et al. [219] describe a simple positi<strong>on</strong><br />

m<strong>on</strong>itor c<strong>on</strong>sisting of a 0.5 µm of Cr or Ti through which <strong>the</strong> <strong>beam</strong> (5 to 25 keV)<br />

passes with little attenuati<strong>on</strong>. An array of four PIN photodiodes surround <strong>the</strong> <strong>beam</strong><br />

axis in a vertical cross arrangement just upstream of <strong>the</strong> foil <strong>–</strong> Figure 8.16. The diodes<br />

record <strong>the</strong> fluorescence signal <strong>and</strong> give <strong>the</strong> same effect as <strong>the</strong> four blades of a normal<br />

bi-axial <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor. The advantage is that <strong>the</strong> <strong>beam</strong> is measured over its<br />

full cross-secti<strong>on</strong> <strong>and</strong> so <strong>the</strong> true centre of intensity of <strong>the</strong> <strong>beam</strong> is measured. The<br />

measured positi<strong>on</strong> sensitivity was 1 <strong>–</strong> 2 µm. As with o<strong>the</strong>r techniques that involve<br />

passing <strong>the</strong> <strong>beam</strong> through a foil, <strong>the</strong> c<strong>on</strong>cerns with a free electr<strong>on</strong> laser <strong>beam</strong> are<br />

ablati<strong>on</strong> of <strong>the</strong> foil a diffractive disrupti<strong>on</strong> to <strong>the</strong> transmitted <strong>beam</strong>.


8.10. Positi<strong>on</strong> <strong>and</strong> centroiding 139<br />

X-ray <strong>beam</strong><br />

Foil<br />

Photo-diodes<br />

Figure 8.16: Photodiode array BPM used to collect fluorescence from a thin foil.<br />

Gas phase photo-i<strong>on</strong>isati<strong>on</strong> m<strong>on</strong>itors<br />

Split-plate i<strong>on</strong> chambers are examples of intensity sampling. This approach is more<br />

sophisticated than blade m<strong>on</strong>itors as it samples <strong>the</strong> whole spatial extent of <strong>the</strong> <strong>beam</strong><br />

<strong>and</strong> should thus give a more reliable measure of <strong>the</strong> <strong>beam</strong> centre of intensity if <strong>the</strong><br />

<strong>beam</strong> is asymmetric or inhomogeneous. The <strong>beam</strong> passes through a gas at a low<br />

pressure between electrode plates with a high voltage bias between <strong>the</strong>m, <strong>and</strong> <strong>the</strong> i<strong>on</strong><br />

yield is recorded as a current from <strong>the</strong> plates. By splitting <strong>on</strong>e of <strong>the</strong> plates diag<strong>on</strong>ally<br />

(Figure 8.17), each half of <strong>the</strong> split plate receives a different signal depending <strong>on</strong> <strong>the</strong><br />

<strong>beam</strong> positi<strong>on</strong> relative to <strong>the</strong> middle of <strong>the</strong> slit <strong>and</strong> <strong>the</strong> i<strong>on</strong> chamber records positi<strong>on</strong><br />

parallel to <strong>the</strong> plane of <strong>the</strong> plates. The diag<strong>on</strong>al split improves <strong>the</strong> linearity of <strong>the</strong><br />

m<strong>on</strong>itor at <strong>the</strong> expense of sensitivity near <strong>the</strong> null positi<strong>on</strong>[220].<br />

A split i<strong>on</strong>-chamber that can measure <strong>the</strong> vertical positi<strong>on</strong> of two partially overlapping<br />

<strong>beam</strong>s simultaneously has been developed at <strong>the</strong> Cornell High Energy Synchrotr<strong>on</strong><br />

Source CHESS with a reported accuracy better than 10 µm <strong>and</strong> a b<strong>and</strong>width<br />

greater than100Hzoveralinear rangeof5mm[221]. Theaccuracywas limitedmainly<br />

by drift in <strong>the</strong> analogue signal electr<strong>on</strong>ics <strong>and</strong> <strong>the</strong> i<strong>on</strong> chamber was actually able to<br />

resolve movements at <strong>the</strong> micr<strong>on</strong> level. Differentiati<strong>on</strong> of <strong>the</strong> two <strong>beam</strong>s was achieved<br />

by designing <strong>the</strong> collecting field produced by <strong>the</strong> bias electrodes such that <strong>the</strong> col-


140 8. Beam cross-secti<strong>on</strong> diagnostics<br />

X-ray <strong>beam</strong><br />

+<br />

Bias plate<br />

Wedge plates<br />

Figure 8.17: Arrangement of <strong>beam</strong> <strong>and</strong> plates for a split plate i<strong>on</strong> chamber.<br />

Signal out<br />

lecting electrodes collect <strong>on</strong>ly i<strong>on</strong>s from <strong>the</strong> n<strong>on</strong>-overlapping edges of <strong>the</strong> two <strong>beam</strong>s.<br />

Thus, <strong>the</strong> i<strong>on</strong> path is short <strong>and</strong> this also improves linearity <strong>and</strong> b<strong>and</strong>width.<br />

Two make a bi-axial measurement, two i<strong>on</strong>-chambers are placed sequentially with<br />

<strong>the</strong>ir plate-pairs orthog<strong>on</strong>al. Schildkamp <strong>and</strong> Praderv<strong>and</strong>[222] describe a system<br />

tested at CHESS which achieved a resoluti<strong>on</strong> below 1 µm with a b<strong>and</strong>width of 1000<br />

Hz.<br />

Important c<strong>on</strong>siderati<strong>on</strong>s when using i<strong>on</strong> chambers are:<br />

• different gases may be needed to cover different phot<strong>on</strong> energy ranges;<br />

• <strong>the</strong> effect of saturati<strong>on</strong> <strong>and</strong> recombinati<strong>on</strong> <strong>on</strong> <strong>the</strong> measured current;<br />

• <strong>the</strong> transit time of <strong>the</strong> electr<strong>on</strong>s <strong>and</strong> i<strong>on</strong>s.<br />

A noble gas is preferred as it eliminates asymmetries that result when using polar<br />

molecules <strong>and</strong> lighter gases reduce <strong>the</strong> transit time of <strong>the</strong> i<strong>on</strong>s to <strong>the</strong> plates <strong>and</strong> so<br />

help reduce <strong>the</strong> build up of space charge in <strong>the</strong> chamber. Whilst <strong>the</strong> i<strong>on</strong> chamber<br />

cannot be damaged by high intensities, <strong>the</strong> probability of recombinati<strong>on</strong> before <strong>the</strong><br />

i<strong>on</strong>s reach <strong>the</strong> plates increases in proporti<strong>on</strong> to <strong>the</strong> <strong>beam</strong> intensity <strong>and</strong> <strong>the</strong> square of<br />

<strong>the</strong> distance travelled. If care is not taken, false positi<strong>on</strong> changes can be deduced as<br />

<strong>the</strong> <strong>beam</strong> intensity is changed. The upper limit <strong>on</strong> <strong>the</strong> detecti<strong>on</strong> b<strong>and</strong>width is set be<br />

<strong>the</strong> transit time of <strong>the</strong> i<strong>on</strong>s to <strong>the</strong> plates; increasing <strong>the</strong> b<strong>and</strong>width requires reducing<br />

<strong>the</strong> plate gap <strong>and</strong> increasing <strong>the</strong> bias potential[223].<br />

The o<strong>the</strong>r main limitati<strong>on</strong> is <strong>the</strong> need for a gas. With a transversely coherent<br />

<strong>beam</strong>, it is not desirable to use windows to isolate <strong>the</strong> gas <strong>and</strong> so <strong>the</strong> chamber must be<br />

isolated by means of differential pumping. In <strong>the</strong> c<strong>on</strong>text of a free electr<strong>on</strong> laser this is<br />

not such a major issue as gas attenuators will be a comm<strong>on</strong> feature <strong>and</strong> <strong>the</strong> significant


8.11. Wavefr<strong>on</strong>t measurements 141<br />

advantage of <strong>the</strong> i<strong>on</strong>izati<strong>on</strong> chamber is that it can be c<strong>on</strong>sidered n<strong>on</strong>-invasive if <strong>the</strong><br />

gas pressure is low enough. The i<strong>on</strong> chamber can thus be used as an <strong>on</strong>-line diagnostic<br />

giving pulse-by-pulse <strong>beam</strong> centroid positi<strong>on</strong>.<br />

8.11 Wavefr<strong>on</strong>t measurements<br />

A wavefr<strong>on</strong>t is defined by <strong>the</strong> surface <strong>on</strong> which <strong>the</strong> radiati<strong>on</strong> field has <strong>the</strong> same phase<br />

(assuming a m<strong>on</strong>ochromatic source). The directi<strong>on</strong> of propagati<strong>on</strong> of <strong>the</strong> radiati<strong>on</strong> is<br />

perpendicular to this surface, hence measurements of <strong>the</strong> propagati<strong>on</strong> directi<strong>on</strong> can<br />

be used to find <strong>the</strong> wavefr<strong>on</strong>t <strong>–</strong> this is <strong>the</strong> principle of <strong>the</strong> Hartmann sensor described<br />

below.<br />

A complete characterizati<strong>on</strong> of <strong>the</strong> radiati<strong>on</strong> field at a particular wavelength requires<br />

a measurement of <strong>the</strong> magnitude <strong>and</strong> relative phase of <strong>the</strong> field. These results<br />

could <strong>the</strong>n be input into simulati<strong>on</strong>s to deduce <strong>the</strong> radiati<strong>on</strong> field at o<strong>the</strong>r locati<strong>on</strong>s in<br />

<strong>the</strong> <strong>beam</strong>line. Such measurements could in principle be made using a Hartmann-type<br />

wavefr<strong>on</strong>t sensor for a m<strong>on</strong>ochromatic source. For narrow-b<strong>and</strong> sources such as free<br />

electr<strong>on</strong> lasers, <strong>the</strong> wavefr<strong>on</strong>t sensor can be used to measure <strong>the</strong> tilt of <strong>the</strong> wavefr<strong>on</strong>t<br />

<strong>and</strong> ray tracing used to find <strong>the</strong> shape of <strong>the</strong> <strong>beam</strong> at o<strong>the</strong>r locati<strong>on</strong>s, assuming that<br />

diffracti<strong>on</strong> effects are not important.<br />

The Hartmann sensor works by splitting <strong>the</strong> <strong>beam</strong> to be diagnosed into an array<br />

of ”mini-<strong>beam</strong>s” ei<strong>the</strong>r by passing <strong>the</strong> <strong>beam</strong> through grid of holes (Hartmann plate)<br />

or an array of lenslets (Shack-Hartmann array) <strong>and</strong> comparing <strong>the</strong> resultant pinhole<br />

images or microlens focal positi<strong>on</strong>s <strong>on</strong> a 2D detector with those from a reference<br />

wavefr<strong>on</strong>t. Variati<strong>on</strong>s in <strong>the</strong> positi<strong>on</strong> of each resultant spot can <strong>the</strong>n be used to<br />

determine <strong>the</strong> local slope in <strong>the</strong> wavefr<strong>on</strong>t. Let Sx ij denote <strong>the</strong> measured slope in <strong>the</strong><br />

x-directi<strong>on</strong> at <strong>the</strong> spot i,j <strong>on</strong> <strong>the</strong> detector, <strong>the</strong>n<br />

Sx ij = dWij<br />

dx<br />

λ dφ<br />

=<br />

2π dx<br />

(8.2)<br />

where W is <strong>the</strong> optical path difference <strong>and</strong> φ is <strong>the</strong> phase. A similar equati<strong>on</strong> can<br />

be written for <strong>the</strong> y-directi<strong>on</strong>. The field strength can be derived from <strong>the</strong> intensity<br />

of <strong>the</strong> spots. The actual wavefr<strong>on</strong>t W has to be rec<strong>on</strong>structed from <strong>the</strong> measured<br />

slopes <strong>–</strong> <strong>the</strong>re are two main methods of doing this, ei<strong>the</strong>r by assuming <strong>the</strong> wavefr<strong>on</strong>t<br />

can be written as a low (normally sec<strong>on</strong>d) order polynomial in <strong>the</strong> local co-ordinates<br />

(z<strong>on</strong>al method) or by exp<strong>and</strong>ing <strong>the</strong> wavefr<strong>on</strong>t in terms orthog<strong>on</strong>al functi<strong>on</strong>s, e.g. 2D<br />

Legendre polynomials (modal method)[224]. The modal rec<strong>on</strong>structi<strong>on</strong> is useful in<br />

being able to identify <strong>the</strong> c<strong>on</strong>tributi<strong>on</strong> of different aberrati<strong>on</strong>s to <strong>the</strong> wavefr<strong>on</strong>t shape.<br />

For X-rayapplicati<strong>on</strong>s a Hartmann plate is used due to <strong>the</strong> lack of microlens arrays<br />

capable of focussing <strong>the</strong> <strong>beam</strong>. For IR <strong>and</strong> UV energies suitable microlens arrays<br />

are available, <strong>the</strong>ir commercial design <strong>and</strong> manufacture having been stimulated by<br />

comm<strong>on</strong> use in ophthalmology <strong>and</strong> corrective laser eye surgery.<br />

The use of wavefr<strong>on</strong>t measurements in commissi<strong>on</strong>ing afree electr<strong>on</strong> laser <strong>beam</strong>line<br />

was dem<strong>on</strong>strated at Flash using a Hartmann sensor from Imagine Optics[225]. The<br />

sensor c<strong>on</strong>tained a Hartmann plate with a 51x51 pinhole array, each pinhole being a<br />

110 µm square tilted 25 ◦ to aid close packing <strong>and</strong> prevent interference with adjacent<br />

holes. A direct CCD camera was used as <strong>the</strong> detector. The reference spherical<br />

wavefr<strong>on</strong>t was generated by inserting a pinhole into <strong>the</strong> <strong>beam</strong>line.


142 8. Beam cross-secti<strong>on</strong> diagnostics<br />

From <strong>the</strong> wavefr<strong>on</strong>t measurements <strong>on</strong> <strong>beam</strong>line 2 (BL2) 12 , <strong>the</strong> depth of focus of<br />

<strong>the</strong> ellipsoidal mirror was calculated using ray tracing <strong>and</strong> a small astigmatism was<br />

found which was cured by remounting a switching mirror. The design focal spot size<br />

was still not obtained, but analysis of <strong>the</strong> wavefr<strong>on</strong>t measurements <strong>on</strong> BL2 <strong>and</strong> <strong>on</strong><br />

<strong>the</strong> unfocussed BL3 showed that this was not due to any remaining aberrati<strong>on</strong>s but<br />

to <strong>the</strong> source being larger than expected during that phase of operati<strong>on</strong>. The critical<br />

yaw angle of <strong>the</strong> toroidal mirror <strong>on</strong> BL1 was also set using <strong>the</strong> wavefr<strong>on</strong>t sensor.<br />

Based <strong>on</strong> this experience, an optimized wavefr<strong>on</strong>t measurement system, using a<br />

320 µm hole pitch, has subsequently been designed in c<strong>on</strong>juncti<strong>on</strong> with <strong>the</strong> Laser<br />

Laboratorium Göttingen for use as an end-stati<strong>on</strong> diagnostic <strong>on</strong> Flash. At <strong>beam</strong>line<br />

2 this wavefr<strong>on</strong>t sensor have been used to align a grazing incidence ellipsoidal mirror<br />

<strong>–</strong> decreasing <strong>the</strong> wavefr<strong>on</strong>t distorti<strong>on</strong> from 52.6 nm (peak-to-valley) <strong>and</strong> 9.2 nm (rms)<br />

to 12 nm <strong>and</strong> 2.6 nm respectively. This by correcting <strong>the</strong> mirror’s pitch -0.29 mrad<br />

<strong>and</strong> its yaw by -0.06 mrad[226]. At BL1 <strong>the</strong> focal spot size <strong>and</strong> <strong>the</strong> positi<strong>on</strong> of <strong>the</strong><br />

<strong>beam</strong>-waist (<strong>and</strong> its fluctuati<strong>on</strong> <strong>on</strong> a shot to shot basis) have been found with <strong>the</strong><br />

same sensor system[227].<br />

At <strong>the</strong> Scss a Hartman wavefr<strong>on</strong>t sensor have been used to do single-shot measurements<br />

to characterize <strong>the</strong> spatial propertis of <strong>the</strong> Sase radiati<strong>on</strong>[228].<br />

8.12 THz/IR techniques<br />

A major difficulty in IR <strong>and</strong> THz diagnostics is that of a suitable means of detecti<strong>on</strong><br />

of <strong>the</strong> l<strong>on</strong>g wavelength radiati<strong>on</strong>. The low quantum energy of IR <strong>and</strong> THz radiati<strong>on</strong><br />

means it is <strong>on</strong>ly able to stimulate vibrati<strong>on</strong>al <strong>and</strong> rotati<strong>on</strong>al modes in molecules <strong>and</strong><br />

cannot directly cause i<strong>on</strong>isati<strong>on</strong>. Therefore, measurement techniques that depend <strong>on</strong><br />

<strong>the</strong> measurement of emitted electr<strong>on</strong>s cannot be used in <strong>the</strong> IR <strong>and</strong> THz. There are<br />

two basic techniques for detecting IR <strong>and</strong> THz radiati<strong>on</strong> that are well established <strong>and</strong><br />

commercially available, namely:<br />

• Thermal<br />

• Phot<strong>on</strong>ic (quantum)<br />

The detectors can ei<strong>the</strong>r be photovoltaic (an induced voltage drives a current in <strong>the</strong><br />

detector circuit) or photoc<strong>on</strong>ductive (an induced resistivity change results in a voltage<br />

change in <strong>the</strong> detector circuit).<br />

In all cases, background radiati<strong>on</strong> is <strong>the</strong> ultimate limiting factor <strong>and</strong> detectors may<br />

require (cryogenic) cooling <strong>and</strong> / or <strong>the</strong>rmal shielding depending <strong>on</strong> <strong>the</strong> radiati<strong>on</strong><br />

levels <strong>and</strong> wavelengths to be measured <strong>and</strong> <strong>the</strong> required signal to noise ratio.<br />

Important factors to c<strong>on</strong>sider when selecting <strong>and</strong> IR detector are:<br />

• Photo-sensitivity (Resp<strong>on</strong>sivity) <strong>–</strong> The output voltage (or current) per watt<br />

of incident radiati<strong>on</strong> power. Units: A / W or V / W.<br />

• Noise equivalent power (NEP) <strong>–</strong> The amount of incident radiati<strong>on</strong> that gives<br />

a signal equal to <strong>the</strong> inherent noise level, i.e. that gives a signal to noise ratio<br />

of 1. Units: W / √ Hz.<br />

12 For a layout of <strong>the</strong> Flash <strong>beam</strong>lines, see Ref. [104]


8.12. THz/IR techniques 143<br />

• Detectivity D ∗ <strong>–</strong> The photo sensitivity per unit active area of <strong>the</strong> detector.<br />

The specific c<strong>on</strong>diti<strong>on</strong>s under which <strong>the</strong> detectivity was measured are usually<br />

given as a functi<strong>on</strong> of: temperature (K) or radiati<strong>on</strong> wavelength (µm), chopping<br />

frequency <strong>and</strong> b<strong>and</strong>width. Units: cm. √ Hz / W<br />

• Spectral resp<strong>on</strong>se <strong>–</strong> How <strong>the</strong> output varies with incident wavelength.<br />

• Resp<strong>on</strong>se time <strong>–</strong> How quickly <strong>the</strong> output rises <strong>and</strong> falls in resp<strong>on</strong>se to an<br />

input pulse. Gating may be needed to measure individual pulses if <strong>the</strong> resp<strong>on</strong>se<br />

is slow.<br />

• Background Limited Infrared Photodetecti<strong>on</strong> (BLIP) <strong>–</strong> The ultimate<br />

detecti<strong>on</strong> limit determined by fluctuati<strong>on</strong>s in <strong>the</strong> background radiati<strong>on</strong> flux in<br />

<strong>the</strong> ideal case of zero noise generated in <strong>the</strong> detector <strong>and</strong> processing circuits.<br />

This is inversely proporti<strong>on</strong>al to <strong>the</strong> square root of <strong>the</strong> background radiati<strong>on</strong><br />

flux.<br />

Thermal detecti<strong>on</strong><br />

Thermal detectors involve <strong>the</strong> measurement of a temperature dependent phenomen<strong>on</strong><br />

such as:<br />

• Temperature dependent resistance (bolometers),<br />

• Thermoelectric effect (<strong>the</strong>rmocouples <strong>and</strong> <strong>the</strong>rmopiles),<br />

• Thermal expansi<strong>on</strong> (Golay cells), <strong>and</strong><br />

• Pyroelectric effect (<strong>the</strong>rmally induced change in <strong>the</strong> surface charge of polarised<br />

crystals)<br />

Thermal detectors use <strong>the</strong> radiati<strong>on</strong> as heat <strong>and</strong> thus <strong>the</strong> photo-sensitivity is independent<br />

of wavelength. Of course, as <strong>the</strong> wavelength increases, more phot<strong>on</strong>s are<br />

needed to give <strong>the</strong> same incident power. Therefore <strong>the</strong> quantum yield is inversely<br />

proporti<strong>on</strong>al to wavelength. Since <strong>the</strong> envir<strong>on</strong>ment is a str<strong>on</strong>g source of l<strong>on</strong>g wavelength<br />

radiati<strong>on</strong>, achieving <strong>the</strong> required signal to noise at l<strong>on</strong>ger wavelengths becomes<br />

increasingly difficult.<br />

The spectral resp<strong>on</strong>se of <strong>the</strong>rmal detectors can be tailored by placing a window<br />

with a suitable transmissi<strong>on</strong> b<strong>and</strong> over <strong>the</strong> detector. It is also highly desirable to use<br />

windows that, as far as possible, block <strong>the</strong> parts of <strong>the</strong> black body spectrum from <strong>the</strong><br />

envir<strong>on</strong>ment that are outside <strong>the</strong> output spectrum of <strong>the</strong> free electr<strong>on</strong> laser source.<br />

Thermal detectors tend to have a slow resp<strong>on</strong>se <strong>and</strong> are more suitable for timeaveraged<br />

measurements or measurements <strong>on</strong> c<strong>on</strong>tinuous wave sources. For example,<br />

<strong>the</strong> resp<strong>on</strong>se of pyroelectric detectors is at <strong>the</strong> millisec<strong>on</strong>d level, though cryogenically<br />

cooled bolometers can resp<strong>on</strong>d much faster than this.<br />

Thepyroelectriceffectdoesnotproduceapermanentvoltage <strong>on</strong><strong>the</strong>crystalbecause<br />

<strong>the</strong> induced charge <strong>on</strong> <strong>the</strong> crystal surface dissipates through internal leakage <strong>and</strong> i<strong>on</strong>s<br />

in <strong>the</strong> air. Thus, pyroelectric detectors produce a signal <strong>on</strong>ly when <strong>the</strong> temperature<br />

of <strong>the</strong> crystal changes <strong>and</strong> <strong>the</strong>y <strong>on</strong>ly resp<strong>on</strong>d to pulsed sources. Use with a c<strong>on</strong>tinuous<br />

wave source requires <strong>the</strong> input light to be chopped, giving a signal of opposite sign at<br />

each opening <strong>and</strong> closing of <strong>the</strong> chopper.


144 8. Beam cross-secti<strong>on</strong> diagnostics<br />

Phot<strong>on</strong>ic detecti<strong>on</strong><br />

Phot<strong>on</strong>ic detectors are semic<strong>on</strong>ductors with b<strong>and</strong>-gaps that are narrow enough for<br />

<strong>the</strong> small energy quanta of IR phot<strong>on</strong>s to excite electr<strong>on</strong>s across it. They are more<br />

sensitive <strong>and</strong> have a higher resp<strong>on</strong>se b<strong>and</strong>width (i.e. are faster) than <strong>the</strong>rmal detectors.<br />

Background <strong>the</strong>rmal excitati<strong>on</strong> will result in a significant dark output <strong>and</strong> some<br />

sort of cooling is normally required.<br />

Phot<strong>on</strong>ic detectors operate over specific wavelength ranges determined by <strong>the</strong><br />

b<strong>and</strong>-gap. The b<strong>and</strong>-gap <strong>and</strong> resp<strong>on</strong>se time also tend to vary with temperature,<br />

so cooling a detector to a level at which <strong>the</strong> intrinsic noise is low enough may shift<br />

its spectral resp<strong>on</strong>se out of <strong>the</strong> required wavelength range.<br />

Most commercial phot<strong>on</strong>ic detectors work in <strong>the</strong> near to far-IR wavelength range<br />

from 0.75 to 15 µm (1.65 eV to 80 meV). Table 8.1 below 13 lists various phot<strong>on</strong>ic IR<br />

detectors.<br />

Detector Spectral<br />

resp<strong>on</strong>se [µm]<br />

Temp<br />

[K]<br />

D ∗ [cm, √ Hz/W] Type<br />

PbS 1 to 3.6 300 D ∗ (500,600,1) = 10 9<br />

PbSe 1.5 to 5.8 300 D ∗ (500,600,1) = 10 8<br />

InAs 2 to 5 213 D ∗ (500,1200,1) = 2·10 9<br />

InSb 2 to 16 77 D ∗ (500,1000,1) = 2·10 10<br />

Ge 0.8 to 1.8 300 D ∗ (λp) = 10 11<br />

InGaAs 0.7 to 1.7 300 D ∗ (λp) = 5·10 15<br />

InAs 1 to 3.1 77 D ∗ (500,1200,1) = 10 10<br />

InSb 1 to 5.5 77 D ∗ (500,1200,1) = 2·10 10<br />

HgCdTe 2 to 16 77 D ∗ (500,1000,1) = 10 10<br />

Ge:Au 1 to 10 77 D ∗ (500,900,1) = 10 11<br />

Ge:Hg 2 to 14 4.2 D ∗ (500,900,1) = 8·10 9<br />

Ge:Cu 2 to 30 4.2 D ∗ (500,900,1) = 5·10 9<br />

Ge:Zn 2 to 40 4.2 D ∗ (500,900,1) = 5·10 9<br />

Si:Ga 1 to 17 4.2 D ∗ (500,900,1) = 5·10 9<br />

Si:As 1 to 23 4.2 D ∗ (500,900,1) = 5·10 9<br />

Table 8.1: List of phot<strong>on</strong>ic IR detectors <strong>and</strong> operating ranges<br />

Intrinsic,<br />

Photoc<strong>on</strong>ductive<br />

Intrinsic,<br />

Photovoltaic<br />

Extrinsic<br />

A widely used phot<strong>on</strong>ic detector in existing IR-FEL facilities is <strong>the</strong> mercurycadmium-telluride<br />

or MCT detector. This is because <strong>the</strong> b<strong>and</strong>-gap <strong>and</strong> hence l<strong>on</strong>gwavelength<br />

cut-off can be tailored by adjusting <strong>the</strong> relative proporti<strong>on</strong>s of CdTe <strong>and</strong><br />

HgTe. The l<strong>on</strong>g wavelength limit of commercially available detectors is ∼ 24 µm 14 .<br />

13<br />

taken from Hamamatsu Technical Informati<strong>on</strong> SD-12 ”Characteristics an use of infrared detectors”.<br />

www.hamamatsu.com<br />

14<br />

InfraRed Associates Inc., www.irassociates.com


8.12. THz/IR techniques 145<br />

The website of <strong>the</strong> Felix IR Laser facility records that a Ge:Ga detector works<br />

from 10 to 200 µm 15 . Liquid helium cooling is of course essential at such a l<strong>on</strong>g<br />

wavelength.<br />

New detectors<br />

The increasing exploitati<strong>on</strong> of IR radiati<strong>on</strong> <strong>on</strong> synchrotr<strong>on</strong> radiati<strong>on</strong> sources has stimulated<br />

development into new types of IR <strong>and</strong> THz detectors that aim to give<br />

• Faster resp<strong>on</strong>se<br />

• Wider spectral resp<strong>on</strong>se<br />

• Higher quantum yield<br />

• Lower noise<br />

• Larger arrays<br />

• Faster array read-out<br />

• Higher spatial resoluti<strong>on</strong><br />

Developing <strong>and</strong> future technologies include:<br />

• Niobium nitride (NbN) superc<strong>on</strong>ducting bolometers (50 ps resp<strong>on</strong>se time, size<br />

0.1 x 1 µm)<br />

• Transiti<strong>on</strong> Edge Superc<strong>on</strong>ducting (TES) detectors (a superc<strong>on</strong>ductor is held<br />

near <strong>the</strong> transiti<strong>on</strong> temperature <strong>and</strong> thus a small amount of added heat gives<br />

an exaggerated c<strong>on</strong>ductivity change, improving sensitivity <strong>and</strong> resp<strong>on</strong>se time).<br />

• Quantum Well Infrared Photodetectors (QWIPs)<br />

• Zero-bias Schottky diodes<br />

Schottky diodes made at <strong>the</strong> Space Science Centre at Ru<strong>the</strong>rford Applet<strong>on</strong> Laboratory<br />

have been tested <strong>on</strong> <strong>the</strong> ALICE accelerator at Daresbury. The fastest has<br />

a measured resp<strong>on</strong>se time of 20 ns (1/e). They are able to distinguish (though not<br />

completely resolve) <strong>the</strong> individual THz pulses from <strong>the</strong> electr<strong>on</strong> bunch train with 81<br />

MHz repleti<strong>on</strong> rate (12 ns period).<br />

IR <strong>and</strong> THz <strong>beam</strong> profiling<br />

Scanning<br />

The simplest way to generate a profile in <strong>on</strong>e or two dimensi<strong>on</strong>s is to raster scan a<br />

detector element (<strong>the</strong>rmal or phot<strong>on</strong>ic) through <strong>the</strong> <strong>beam</strong>. The spatial resoluti<strong>on</strong> is<br />

limited by <strong>the</strong> size of <strong>the</strong> detector or <strong>the</strong> defining slit in fr<strong>on</strong>t of it. Diffracti<strong>on</strong> at a<br />

defining slit will limit <strong>the</strong> resoluti<strong>on</strong> if <strong>the</strong> slit size is comparable to <strong>the</strong> wavelength.<br />

Because IR <strong>beam</strong>s are relatively easy to manipulate, it is also possible to optically<br />

raster <strong>the</strong> <strong>beam</strong> over <strong>the</strong> fixed detector. This is likely to be faster than moving <strong>the</strong><br />

detector since <strong>the</strong> optical raster will use angular shifts of a lens or mirror whilst <strong>the</strong><br />

detector raster will use linear moti<strong>on</strong>s of <strong>the</strong> detector.<br />

15 Felix IR Laser facility: www.rijnhuizen.nl/en/felix/facilities


146 8. Beam cross-secti<strong>on</strong> diagnostics<br />

Scanning a linear array detector is more efficient, though <strong>the</strong> spatial sampling of<br />

<strong>the</strong> scan in <strong>the</strong> directi<strong>on</strong> al<strong>on</strong>g <strong>the</strong> array is fixed by <strong>the</strong> array spacing. The array<br />

must also be matched to <strong>the</strong> <strong>beam</strong> size, whilst a single element scan can used <strong>on</strong> an<br />

arbitrary <strong>beam</strong> size.<br />

imaging<br />

Infrared imaging is of course widespread. Even a st<strong>and</strong>ard CCD based digital camera<br />

can image in <strong>the</strong> near infrared, whilst security applicati<strong>on</strong>s have driven a huge development<br />

of sub-optical imaging systems. Here we will <strong>the</strong>refore c<strong>on</strong>sider <strong>the</strong> more<br />

challenging area of l<strong>on</strong>ger wavelength imaging <strong>and</strong> imaging which is tailored to <strong>the</strong><br />

nature of free electr<strong>on</strong> laser sources.<br />

It is c<strong>on</strong>ceivable that any of <strong>the</strong> basic types of single element IR detector can<br />

be built into an imaging array, ei<strong>the</strong>r 1-dimensi<strong>on</strong>al or 2-dimensi<strong>on</strong>al. The specific<br />

technical questi<strong>on</strong>s that need to be addressed are:<br />

• The size of <strong>the</strong> detector elements<br />

• The spacing of <strong>the</strong> detector elements<br />

• The read-out time<br />

• Sensitivity <strong>and</strong> spectral resp<strong>on</strong>se<br />

Achieving high spatial resoluti<strong>on</strong> requires small <strong>and</strong> closely spaced detector elements,<br />

but this can lead to problems with sensitivity at l<strong>on</strong>g wavelengths since <strong>the</strong><br />

low quantum efficiency means <strong>the</strong> signal generated in each element is too small to<br />

detect. For example, a commercial pyroelectric array detector was found to be unable<br />

to detect <strong>the</strong> THz output of <strong>the</strong> quantum cascade laser at Leeds University, UK.<br />

C<strong>on</strong>siderati<strong>on</strong> must also be given to diffractive effects at <strong>the</strong>se l<strong>on</strong>g wavelengths.<br />

Diffracti<strong>on</strong> at defining apertures could lead to a decrease <strong>the</strong> spatial resoluti<strong>on</strong>. Does<br />

a sensor element that is smaller than <strong>the</strong> wavelength of <strong>the</strong> radiati<strong>on</strong> accurately record<br />

<strong>the</strong> incident intensity?<br />

Beam splitters<br />

The detectors described in here are opaque to <strong>the</strong> <strong>beam</strong> <strong>and</strong> thus cannot be used as<br />

part of a n<strong>on</strong>invasive detector. However, it is relatively easy to split an IR <strong>and</strong> THz<br />

<strong>beam</strong> in amplitude such that <strong>on</strong>ly a small part of it passes to <strong>the</strong> detector <strong>and</strong> <strong>the</strong><br />

rest passes to <strong>the</strong> experiment 16 . The most suitable techniques for IR <strong>and</strong> THz are:<br />

• Plates <strong>and</strong> pellicles <strong>–</strong> can be made of various materials depending <strong>on</strong> <strong>the</strong><br />

transmissi<strong>on</strong> b<strong>and</strong> required. Pellicles can be coated to enhance <strong>the</strong> reflectivity<br />

at <strong>the</strong> expense of transmissi<strong>on</strong>.<br />

• Wire grids <strong>–</strong> <strong>the</strong>se are polarising <strong>beam</strong> splitters, suitable for <strong>the</strong> l<strong>on</strong>gest wavelengths<br />

<strong>on</strong>ly (determined by <strong>the</strong> wire spacing).<br />

16 See chapter 5 (see page 73).


8.12. THz/IR techniques 147<br />

Wavefr<strong>on</strong>t sensing<br />

The Shack-Hartmann sensor is a st<strong>and</strong>ard instrument for measuring wavefr<strong>on</strong>ts in<br />

<strong>the</strong> optical <strong>and</strong> IR regi<strong>on</strong>. Commercial instruments tend to be limited to operati<strong>on</strong><br />

in <strong>the</strong> near-IR (∼ 2 µm wavelength), probably because of <strong>the</strong> types of IR sensitive<br />

array detectors that are available with sufficient pixel count, density <strong>and</strong> sensitivity.<br />

CCD cameras would seem to be <strong>the</strong> st<strong>and</strong>ard detector, giving megapixel array size<br />

with spacing in <strong>the</strong> 20 µm range. High pixel count <strong>and</strong> small spatial separati<strong>on</strong> is<br />

required to allow accurate centroid determinati<strong>on</strong> of a large number of micro-<strong>beam</strong>s,<br />

both of which are necessary for accurate wavefr<strong>on</strong>t rec<strong>on</strong>structi<strong>on</strong>.<br />

Extending operati<strong>on</strong> to <strong>the</strong> far-infrared would require, for example, an MCT array<br />

detector. It is not clear that an MCT array with sufficient pixel count <strong>and</strong> density<br />

is technically feasible, let al<strong>on</strong>e cost effective. Alternatively, <strong>and</strong> for operati<strong>on</strong> in <strong>the</strong><br />

THz regi<strong>on</strong>, <strong>the</strong> array would have to be made of <strong>the</strong>rmal detectors. As menti<strong>on</strong>ed<br />

before, pyroelectric array detectors are commercially available. These currently have<br />

ra<strong>the</strong>r low pixel counts compared with CCD cameras <strong>and</strong> a detailed technical assessment<br />

would have to be made to decide <strong>on</strong> <strong>the</strong> minimum pixel count that would be<br />

required. This assessment would also need to c<strong>on</strong>sider <strong>the</strong> effects of <strong>the</strong> much str<strong>on</strong>ger<br />

diffracti<strong>on</strong> <strong>and</strong> <strong>the</strong> Hartmann plate of micro-lens array.<br />

Electro-Optical Imaging<br />

Electro-optical sampling of THz radiati<strong>on</strong> uses <strong>the</strong> Pockels effect in electro-optic crystals.<br />

A THz pulse acts like a transient bias that induces a transient polarisati<strong>on</strong> in<br />

<strong>the</strong> crystal. The polarisati<strong>on</strong> induces a birefringence in <strong>the</strong> crystal that is probed by<br />

a synchr<strong>on</strong>ous optical laser <strong>beam</strong>, <strong>the</strong> probe <strong>beam</strong> undergoing a polarisati<strong>on</strong> change<br />

as it passes through <strong>the</strong> crystal.<br />

Wu et al.[229] describe how to exploit <strong>the</strong> EO effect to produce a 2-dimensi<strong>on</strong>al<br />

image of <strong>the</strong> THz <strong>beam</strong> cross-secti<strong>on</strong>, see Figure 8.18. The THz <strong>beam</strong> was focused<br />

at a ZnTe EO crystal <strong>and</strong> was overlapped with a co-propagating optical laser <strong>beam</strong>.<br />

The optical field probes <strong>the</strong> spatial distributi<strong>on</strong> of <strong>the</strong> electric field in <strong>the</strong> crystal that<br />

<strong>the</strong> THz radiati<strong>on</strong> induces. Crossed-polarisers ei<strong>the</strong>r side of <strong>the</strong> EO crystal c<strong>on</strong>vert<br />

<strong>the</strong> resulting polarisati<strong>on</strong> modulati<strong>on</strong> of <strong>the</strong> optical <strong>beam</strong> as it passes through <strong>the</strong><br />

crystal into an intensity modulati<strong>on</strong> that is recorded by a CCD camera, which thus<br />

gives a 2-dimensi<strong>on</strong>al intensity image of <strong>the</strong> THz <strong>beam</strong>.<br />

This technique can be directly applied to determining <strong>the</strong> focus size of THz <strong>beam</strong>s.<br />

Obviously, a very fast optical laser that is synchr<strong>on</strong>ised to <strong>the</strong> THz pulse is also<br />

required. The Pockels effect is very fast <strong>and</strong> imaging rate is limited by <strong>the</strong> camera.<br />

Gating of <strong>the</strong> camera should allow single pulse measurement provided <strong>the</strong>re is enough<br />

modulati<strong>on</strong> in <strong>the</strong> optical <strong>beam</strong> to produce a measurable signal at <strong>the</strong> camera.<br />

Measurement of unfocussed <strong>beam</strong>s is likely to be limited by <strong>the</strong> phot<strong>on</strong> density of<br />

<strong>the</strong> THz <strong>and</strong> optical probe <strong>beam</strong>s, since <strong>the</strong> optical <strong>beam</strong> must spatially overlap <strong>the</strong><br />

THz <strong>beam</strong> for a complete 2-D image to be recorded in <strong>on</strong>e shot. The available size<br />

of EO crystals will be ano<strong>the</strong>r factor limiting <strong>the</strong> largest <strong>beam</strong> footprint that can be<br />

measured. These limitati<strong>on</strong>s could be overcome by using an optical system such as a<br />

telescope to compress <strong>the</strong> <strong>beam</strong> diameter without modifying <strong>the</strong> wavefr<strong>on</strong>t.


148 8. Beam cross-secti<strong>on</strong> diagnostics<br />

THz <strong>beam</strong><br />

8.13 Summary<br />

Pellicle<br />

Probe <strong>beam</strong><br />

Polariser<br />

ZnTe crystal<br />

Analyser<br />

Figure 8.18: Layout of <strong>the</strong> EO imaging system of Wu.<br />

CCD<br />

Camera<br />

The ideal or ”universal” diagnostic for a free electr<strong>on</strong> laser source would be able to:<br />

• Give a full spatial image of <strong>the</strong> phot<strong>on</strong> <strong>beam</strong>.<br />

• Do this for every pulse produced.<br />

• Not change <strong>the</strong> phot<strong>on</strong> pulse in any significant manner.<br />

• Operate over a wide spectral range (at least sufficient to cover <strong>the</strong> full output<br />

range of <strong>the</strong> source)<br />

At <strong>the</strong> moment, such a universal diagnostic is not possible.<br />

• Techniques that give full spatial profiles tend to be slow <strong>and</strong> invasive<br />

• Faster techniques often <strong>on</strong>ly give informati<strong>on</strong> such as <strong>beam</strong> centroid<br />

• N<strong>on</strong>-invasive techniques are too insensitive to measure a single pulse<br />

• Operating wavelength range is str<strong>on</strong>gly limited by <strong>the</strong> detecti<strong>on</strong> technique employed<br />

For radiati<strong>on</strong> with wavelengths from <strong>the</strong> VUV to X-rays, <strong>the</strong>re is a c<strong>on</strong>siderable<br />

range of diagnostics employed <strong>on</strong> synchrotr<strong>on</strong> radiati<strong>on</strong> sources that could be developed<br />

for use <strong>on</strong> free electr<strong>on</strong> laser sources. Many of <strong>the</strong>se techniques involve transferring<br />

<strong>the</strong> spatial informati<strong>on</strong> to electr<strong>on</strong>s that are detected <strong>and</strong> analysed to extract <strong>the</strong><br />

spatial informati<strong>on</strong>. Imaging detecti<strong>on</strong> is often achieved by c<strong>on</strong>verting <strong>the</strong> electr<strong>on</strong>s<br />

to visible phot<strong>on</strong>s with <strong>the</strong> aid of luminescent screens. For harder X-rays, fluorescence<br />

is <strong>the</strong> dominant process <strong>and</strong> detecti<strong>on</strong> of <strong>the</strong> fluorescent phot<strong>on</strong>s, probably indirectly<br />

by c<strong>on</strong>versi<strong>on</strong> to visible phot<strong>on</strong>s, is a better approach.<br />

For <strong>the</strong> IR <strong>and</strong> THz, <strong>the</strong>se approaches cannot be used <strong>and</strong> <strong>the</strong> choice of detecti<strong>on</strong><br />

method is much more limited. Phot<strong>on</strong>ic detectors are fast <strong>and</strong> efficient but are mainly<br />

limited to detecting in <strong>the</strong> near to far infrared, up to ∼ 24 µm with HgCdTe (though<br />

∼200 µm is possible with Ge:Ga). Thermal detectors must be used at frequencies<br />

below a few THz <strong>and</strong> <strong>the</strong>se tend to be slow. However, output at <strong>the</strong>se extremely l<strong>on</strong>g<br />

wavelengths is not generally in <strong>the</strong> realm of free electr<strong>on</strong> laser sources. Electro-optical<br />

techniques are probably a better soluti<strong>on</strong> for <strong>the</strong> very l<strong>on</strong>g wavelengths.<br />

It is thus inevitable that a range of diagnostics <strong>and</strong> detectors will be employed<br />

depending <strong>on</strong> <strong>the</strong> informati<strong>on</strong> required <strong>and</strong> <strong>the</strong> use to which it is to be put, for<br />

example:


8.13. Summary 149<br />

• Centroiding techniques for pulse by pulse <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itoring <strong>and</strong> feedback<br />

• Invasive imaging for optimising <strong>the</strong> source <strong>and</strong> phot<strong>on</strong> <strong>transport</strong> during commissi<strong>on</strong>ing<br />

• Fast imaging arrays or wavefr<strong>on</strong>t sensors situated behind gas-phase experiments<br />

The use of <strong>beam</strong> splitters to separate a part of <strong>the</strong> <strong>beam</strong> for <strong>the</strong> more sophisticated<br />

diagnostics whilst <strong>the</strong> bulk of <strong>the</strong> <strong>beam</strong> is passed to <strong>the</strong> experiment are also likely to<br />

be a comm<strong>on</strong> feature of <strong>the</strong> phot<strong>on</strong> <strong>transport</strong> systems due to <strong>the</strong> lack of truly n<strong>on</strong>invasivediagnostics.<br />

These splitterscanberemovedfrom <strong>the</strong><strong>beam</strong>pathfor maximum<br />

throughput<strong>and</strong>/orif<strong>the</strong>diagnostic isnotrequired. Ideally, <strong>the</strong><strong>beam</strong>splittersshould<br />

divide <strong>the</strong> <strong>beam</strong> in amplitude so <strong>the</strong> diagnostic <strong>beam</strong> has <strong>the</strong> same <strong>beam</strong> profile as<br />

<strong>the</strong> measured <strong>beam</strong> (but at lower intensity). This should be straightforward in <strong>the</strong><br />

IR <strong>and</strong> THz, but more challenging in <strong>the</strong> VUV <strong>and</strong> Soft Xray, where knife-edged<br />

mirrors that divide <strong>the</strong> wavefr<strong>on</strong>t are <strong>the</strong> easiest splitter to implement. Multilayer<br />

<strong>and</strong> slotted-mirror type <strong>beam</strong> splitters would be required to give amplitude <strong>beam</strong><br />

divisi<strong>on</strong>.<br />

Despite <strong>the</strong> c<strong>on</strong>siderable challenges that must be faced when making good spatial<br />

diagnostics for free electr<strong>on</strong> laser <strong>beam</strong>s, <strong>the</strong>re are a wide range of techniques<br />

that can be employed. A judicious combinati<strong>on</strong> of techniques will allow <strong>the</strong> required<br />

informati<strong>on</strong> to be measured.


150 8. Beam cross-secti<strong>on</strong> diagnostics<br />

Summary<br />

• Cross-secti<strong>on</strong> diagnostics measures <strong>the</strong> transverse intensity<br />

distributi<strong>on</strong> of <strong>the</strong> <strong>beam</strong>. For both optimizati<strong>on</strong>, commissi<strong>on</strong>ing<br />

of experiments <strong>and</strong> instruments it is important to know<br />

where <strong>the</strong> <strong>beam</strong> is <strong>and</strong> how large it is.<br />

• Measuring <strong>the</strong> focus size can be d<strong>on</strong>e ei<strong>the</strong>r via ablati<strong>on</strong> crater<br />

analysis (<strong>the</strong> size of a hole in a thin film), by <strong>the</strong> saturati<strong>on</strong> of<br />

an i<strong>on</strong>izati<strong>on</strong> process in a gas or by using a wave-fr<strong>on</strong>t sensor.<br />

• Techniques developed for synchrotr<strong>on</strong> sources may not be immediately<br />

used at free electr<strong>on</strong> lasers or not at all.<br />

• Examples of invasive techniques are:<br />

<strong>–</strong> Direct imaging<br />

<strong>–</strong> Wire grids<br />

<strong>–</strong> Scanning wires, slits, knife-edges, pin-holes. All have <strong>the</strong><br />

drawback that <strong>the</strong>y are not single shot.<br />

• N<strong>on</strong>-invasive techniques can be used while o<strong>the</strong>r experiments<br />

are running.<br />

<strong>–</strong> Rest gas i<strong>on</strong>izati<strong>on</strong><br />

<strong>–</strong> Photo dissociati<strong>on</strong><br />

<strong>–</strong> Synchrotr<strong>on</strong> light<br />

<strong>–</strong> Compt<strong>on</strong> scattering<br />

• Acombinati<strong>on</strong>oftechniquesisneeded, inpractice, todiagnose<br />

<strong>the</strong> <strong>beam</strong>:<br />

<strong>–</strong> Centroiding techniques for pulse by pulse <strong>beam</strong> positi<strong>on</strong><br />

m<strong>on</strong>itoring<br />

<strong>–</strong> Invasive imaging for optimizing <strong>and</strong> commisi<strong>on</strong>ing<br />

<strong>–</strong> Fast imaging devices or wavefr<strong>on</strong>t sensors situated behind<br />

gas-phase experiments.<br />

• In <strong>the</strong> THz range photodiodes in <strong>the</strong> infrared or <strong>the</strong>rmal detectors<br />

for l<strong>on</strong>ger wavelengths can be used. The latter <strong>on</strong>ly<br />

for averaging measurements.<br />

• Beamsplitters offers <strong>the</strong> possibility to use invasive diagnostics<br />

in parallel to o<strong>the</strong>r experiments.


9. Pulse length, profile <strong>and</strong> jitter<br />

The material presented here in this chapter is partly adapted from ”Survey<br />

of diagnostics techniques for measuring <strong>the</strong> temporal properties of ultra-short<br />

phot<strong>on</strong> pulses” by M. A. Bowler, A. J. Glees<strong>on</strong> <strong>and</strong> M. D. Roper. Iruvx<br />

WP7, 2009 by A. Lindblad.<br />

9.1 Introducti<strong>on</strong><br />

In chapter 8 methods to discern <strong>the</strong> transverse extent of a phot<strong>on</strong>-<strong>beam</strong> were discussed.<br />

Here we will review methods as to determine <strong>the</strong> length <strong>and</strong> profile of a<br />

pulse; <strong>the</strong> jitter between <strong>the</strong> pulses is also an important temporal parameter which is<br />

necessary to measure.<br />

A key measure of <strong>the</strong> performance of a freeelectr<strong>on</strong><br />

laser is <strong>the</strong> temporal properties of <strong>the</strong><br />

phot<strong>on</strong>-<strong>beam</strong>. The pulse length is required for<br />

<strong>the</strong> integral pulse power at <strong>the</strong> experiment; <strong>the</strong><br />

pulse profile determines <strong>the</strong> ”quality” of <strong>the</strong><br />

pulse in terms of length <strong>and</strong> height of, <strong>and</strong> deviati<strong>on</strong><br />

from, <strong>the</strong> ideal pedestal shape in Figure<br />

9.1 (i.e. deviati<strong>on</strong> from <strong>the</strong> transform limit<br />

for <strong>the</strong> spectral c<strong>on</strong>tent of <strong>the</strong> pulse) <strong>–</strong> this is<br />

also related to <strong>the</strong> jitter, since if <strong>the</strong> pulse shape<br />

lacks repeating structure <strong>the</strong> centroid will be<br />

shifted <strong>on</strong> a pulse to pulse basis which is equivalent<br />

to a shift in relative occurrence times; <strong>the</strong><br />

pulse jitter is a r<strong>and</strong>om fluctuati<strong>on</strong> in <strong>the</strong> arrival<br />

time of a pulse, by necessity this needs<br />

to be correlated to ano<strong>the</strong>r timing event. For<br />

δt<br />

∆t<br />

Figure 9.1: Ideal pulses have square profiles of<br />

length δt, occurring with frequency ∆t −1 ; <strong>the</strong><br />

pulse shape can be significantly deviated from<br />

<strong>the</strong> ideal square <strong>and</strong> occur within a frequency<br />

envelope defined by a time-jitter.<br />

pump-probe experiments this is obviously a crucial parameter.<br />

The profile <strong>and</strong> timing of a free electr<strong>on</strong> laser pulse is certainly going to change <strong>on</strong><br />

a pulse by pulse basis by an amount that will cause difficulty to at least some experiments.<br />

This is especially true for Sase operati<strong>on</strong> because <strong>the</strong> pulses are generated<br />

from r<strong>and</strong>om noise. Therefore, <strong>the</strong>re is a general need for <strong>the</strong> temporal diagnostics to<br />

be permanently ”<strong>on</strong>-line” so that <strong>the</strong> profile <strong>and</strong> timing of every pulse can be measured.<br />

Such a diagnostic should obviously impose a negligible change <strong>on</strong> <strong>the</strong> pulse<br />

151


152 9. Pulse length, profile <strong>and</strong> jitter<br />

being measured. The diagnostic thus needs to be ei<strong>the</strong>r effectively transparent to <strong>the</strong><br />

pulse, or at <strong>the</strong> very least interrupt <strong>on</strong>ly a small part <strong>and</strong> pass <strong>the</strong> major part of <strong>the</strong><br />

pulse undisturbed to <strong>the</strong> experiment.<br />

Hence, <strong>the</strong>re is a very dem<strong>and</strong>ing specificati<strong>on</strong> <strong>on</strong> an ideal pulse timing diagnostic<br />

apparatus:<br />

• Measure all pulses at <strong>the</strong> repetiti<strong>on</strong> rate of <strong>the</strong> machine, i.e. in <strong>the</strong> kHz <strong>and</strong><br />

MHz regimes.<br />

• Meaure <strong>the</strong> intensity profiles of <strong>the</strong> pulses with a temporal resoluti<strong>on</strong> of about<br />

1 femtosec<strong>on</strong>d <strong>–</strong> carrying in mind that this resoluti<strong>on</strong> needs to be maintained<br />

over pulse lengths that can have a durati<strong>on</strong> of hundreds of femtosec<strong>on</strong>ds.<br />

• Measure <strong>the</strong> arrival time of <strong>the</strong> pulse with a resoluti<strong>on</strong> in <strong>the</strong> femtosec<strong>on</strong>d<br />

domain.<br />

• Be transparent to <strong>the</strong> pulse, or sample a minuscule part of <strong>the</strong> pulse.<br />

• Work in a spectral range from <strong>the</strong> VUV to <strong>the</strong> hard X-rays.<br />

A diagnostic tool that fulfils all of <strong>the</strong> above dem<strong>and</strong>s can not be found. Below<br />

we will survey <strong>the</strong> techniques available in <strong>the</strong> VUV <strong>and</strong> X-ray regime toge<strong>the</strong>r with<br />

<strong>the</strong>ir limitati<strong>on</strong>s vis-à-vis <strong>the</strong> ideal outlined above.<br />

Many of <strong>the</strong> techniques that are currently being used <strong>and</strong> developed are based<br />

around cross-correlati<strong>on</strong> with an external optical laser (a high-power IR Ti:Sapphire)<br />

<strong>and</strong> can give both pulse length <strong>and</strong> pulse jitter (relative to <strong>the</strong> optical laser). These<br />

techniques have started out as multi-shot measurements since <strong>the</strong>y initially required<br />

scanning <strong>the</strong> IR pulse delay. But <strong>the</strong>re is a lot of activity in developing <strong>the</strong> techniques<br />

for single-shot use <strong>and</strong> in improving<strong>the</strong> temporal resoluti<strong>on</strong> to<strong>the</strong>femtosec<strong>on</strong>d<br />

level (see page 153).<br />

Electro-optical sampling also uses cross-correlati<strong>on</strong> with an optical laser <strong>and</strong> is<br />

used mainly for electr<strong>on</strong> bunch measurements but can also be used for measuring <strong>the</strong><br />

lengths of THz pulses directly (see page 155). An alternative approach to pulse length<br />

measurement is auto-correlati<strong>on</strong>, (see page 155).<br />

Simple intensity autocorrelati<strong>on</strong> <strong>on</strong>ly gives a pulse length <strong>and</strong> not <strong>the</strong> profile. However,<br />

more sophisticated autocorrelati<strong>on</strong> techniques have been developed in <strong>the</strong> visible<br />

<strong>and</strong> UV that can give <strong>the</strong> full pulse profile of pulses as short as a few femtosec<strong>on</strong>ds.<br />

There is some possibility to extend <strong>the</strong>se techniques to shorter wavelengths, though<br />

<strong>the</strong> limit here is not clear.<br />

Reflectivity modulati<strong>on</strong> of a semic<strong>on</strong>ductor by a free electr<strong>on</strong> laser pulse has been<br />

used to give single shot measurements with a temporal resoluti<strong>on</strong> of 40 fs. Streak<br />

cameras areawell establishedtechnologyformeasuringpulseswithpicosec<strong>on</strong>dlengths<br />

<strong>and</strong> <strong>the</strong>y are being developed to achieve resoluti<strong>on</strong>s of a few hundred femtosec<strong>on</strong>ds.<br />

Streak cameras can give single pulse measurements but <strong>on</strong>ly at limited repetiti<strong>on</strong><br />

rates.<br />

Recently an elegant way of achieving few femtosec<strong>on</strong>ds resoluti<strong>on</strong> in <strong>the</strong> timedomain<br />

was dem<strong>on</strong>strated at Flash by Tavella <strong>and</strong> co-workers. Their technique<br />

utilized <strong>the</strong> terahertz radiati<strong>on</strong> generated in <strong>the</strong> undulator, which is <strong>the</strong>n phasecorrelated<br />

to <strong>the</strong> X-ray pulse. The optical laser system of <strong>the</strong> facility can <strong>the</strong>n be diagnosed<br />

toge<strong>the</strong>r with he terahertz radiati<strong>on</strong> without disturbing <strong>the</strong> X-ray pulse[230].


9.2. Cross-correlati<strong>on</strong> techniques 153<br />

9.2 Cross-correlati<strong>on</strong> techniques<br />

Cross-correlati<strong>on</strong> techniques are currently <strong>the</strong> most widely used techniques at XUV<br />

wavelengths as <strong>the</strong>y give pulse length <strong>and</strong> jitter informati<strong>on</strong>. A wide range of techniques<br />

based <strong>on</strong> <strong>the</strong> photo-i<strong>on</strong>isati<strong>on</strong> of gases are being developed as <strong>the</strong>y have <strong>the</strong><br />

potential to be transparent to <strong>the</strong> phot<strong>on</strong> <strong>beam</strong>.<br />

With a known reference laser pulse, cross-correlati<strong>on</strong> between this <strong>and</strong> a X-ray<br />

pulse can be used to measure <strong>the</strong> relative jitter of <strong>the</strong> free electr<strong>on</strong> laser pulse with<br />

respect to <strong>the</strong> laser pulse; in some circumstances said measurement can be used to<br />

estimate <strong>the</strong> length of <strong>the</strong> X-ray pulse.<br />

The p<strong>on</strong>dermotive energy is given by<br />

Up(t) = e 2 E 2 a(t)cos(ωℓt+φ)<br />

where Ea is <strong>the</strong> amplitude of <strong>the</strong> laser field, ωℓ <strong>the</strong> laser frequency <strong>and</strong> φ <strong>the</strong> phase<br />

respectively[231].<br />

In most experimental c<strong>on</strong>figurati<strong>on</strong>s, <strong>the</strong> presence of side-b<strong>and</strong>s would be <strong>the</strong> dominant<br />

effect, but this can be suppressed by measuring <strong>the</strong> photo-electr<strong>on</strong>s ejected in<br />

a directi<strong>on</strong> perpendicular to <strong>the</strong> polarisati<strong>on</strong> axis of <strong>the</strong> laser radiati<strong>on</strong>, allowing <strong>the</strong><br />

shift due to <strong>the</strong> p<strong>on</strong>deromotive energy to be observed.<br />

The intensity of <strong>the</strong> side-b<strong>and</strong>s is proporti<strong>on</strong>al to magnitude of <strong>the</strong> photoelectr<strong>on</strong><br />

wave-vector[232]<br />

k = 1<br />

�<br />

� 2m(�ωfel −Ip)<br />

Dec<strong>on</strong>voluti<strong>on</strong> of <strong>the</strong> side-b<strong>and</strong> intensity as a functi<strong>on</strong> of delay is obviously not a<br />

single shot measurement. In order to determine <strong>the</strong> pulse length, <strong>the</strong> jitter between<br />

<strong>the</strong> two pulses <strong>and</strong> <strong>the</strong> X-ray pulse length must be small enough not to dominate<br />

<strong>the</strong> cross-correlati<strong>on</strong> curve. Such a technique can be successfully applied to measure<br />

<strong>the</strong> pulse length of soft X-ray radiati<strong>on</strong> generated by HHG where <strong>the</strong> jitter between<br />

<strong>the</strong> generating infrared pulse <strong>and</strong> <strong>the</strong> resulting HHG radiati<strong>on</strong> is very small. In <strong>the</strong><br />

case of radiati<strong>on</strong> from a Sase free electr<strong>on</strong> laser, <strong>the</strong> cross-correlati<strong>on</strong> curve will most<br />

likely to be dominated by <strong>the</strong> jitter between <strong>the</strong> pulses, <strong>and</strong> can in fact be used to<br />

measure <strong>the</strong> distributi<strong>on</strong> of <strong>the</strong> jitter. It may be possible that this technique could<br />

be used for seeded free electr<strong>on</strong> lasers where <strong>the</strong> pulses will be more stable <strong>and</strong> <strong>the</strong><br />

jitter smaller.<br />

An example of being able to estimate <strong>the</strong> time delay from a single shot measurement<br />

is given in Radcliffe et al. [232], where <strong>the</strong>y used 13.8 nm pulses from Flash<br />

of about 20 fs l<strong>on</strong>g to i<strong>on</strong>ise Xe <strong>and</strong> measured <strong>the</strong> sideb<strong>and</strong> intensity in <strong>the</strong> presence<br />

of 120 fs pulses from a Ti:Sapphire laser. This case is favourable for <strong>the</strong> formati<strong>on</strong><br />

of side-b<strong>and</strong>s due to <strong>the</strong> relatively large photo-electr<strong>on</strong> wave-vector <strong>and</strong> up to four<br />

high energy side-b<strong>and</strong>s were obtained. From <strong>the</strong>oretical simulati<strong>on</strong>s of <strong>the</strong> side-b<strong>and</strong><br />

intensity, <strong>the</strong> number of side-b<strong>and</strong>s gives a measure of <strong>the</strong> laser field intensity during<br />

<strong>the</strong> FEL pulse, <strong>and</strong> hence <strong>the</strong> overlap of <strong>the</strong> pulses. In this experiment, a precisi<strong>on</strong><br />

of better than 50 fs was achieved for <strong>the</strong> relative delay, but note that <strong>the</strong> sign of<br />

<strong>the</strong> relative delay cannot be obtained. The cross-correlati<strong>on</strong> curve has a FWHM of<br />

about 600±50 fs, which is dominated by <strong>the</strong> jitter, <strong>and</strong> gives <strong>the</strong> FWHM of <strong>the</strong> jitter<br />

distributi<strong>on</strong> of 590 fs. It should be possible to determine <strong>the</strong> sign of <strong>the</strong> jitter by<br />

using a chirped laser pulse (as also suggested in <strong>the</strong> XFEL TDR [89] for <strong>the</strong> Auger<br />

electr<strong>on</strong> measurements menti<strong>on</strong>ed below) but this idea has yet to be tested.


154 9. Pulse length, profile <strong>and</strong> jitter<br />

Ifthis work were tobe extendedtoveryshort X-rayspulses, <strong>the</strong>nit would be found<br />

that<strong>the</strong>spectralwidthof<strong>the</strong>X-raypulsewouldsmear out<strong>the</strong>side-b<strong>and</strong>s. This canbe<br />

overcome by observing <strong>the</strong> photo-electr<strong>on</strong> spectra for electr<strong>on</strong>s ejected at right angles<br />

to <strong>the</strong> polarisati<strong>on</strong> of <strong>the</strong> laser. In this case, as noted above, sideb<strong>and</strong> formati<strong>on</strong> is<br />

suppressed, <strong>and</strong> <strong>on</strong>e can measure <strong>the</strong> red-shift in <strong>the</strong> energy of <strong>the</strong> electr<strong>on</strong>s due to <strong>the</strong><br />

p<strong>on</strong>deromotive force exerted by <strong>the</strong> laser field. This has been d<strong>on</strong>e successfully using<br />

an HHG source of SXR (90 eV) phot<strong>on</strong>s co-focussed with <strong>the</strong> generating 770 nm laser<br />

pulses [231]. A delay scan yielded an X-ray pulse width of <strong>the</strong> order of 2 fs FWHM;<br />

again, extending this to free electr<strong>on</strong> lasers would require <strong>the</strong> relative jitter to be of a<br />

similar size to <strong>the</strong> pulse width. For a seeded free electr<strong>on</strong> laser, this may be possible<br />

if <strong>the</strong> X-ray pulse timing is dictated by <strong>the</strong> laser seed. A significant disadvantage of<br />

<strong>the</strong>se techniques is <strong>the</strong> need to co-propagate <strong>the</strong> FEL <strong>and</strong> IR pulses to a comm<strong>on</strong><br />

focus. For <strong>on</strong>-line use, this requires a permanent extra focus to be included in <strong>the</strong><br />

<strong>beam</strong> <strong>transport</strong> system, which is not always c<strong>on</strong>venient or desirable. A perpendicular<br />

geometry would be more flexible - this has been used in <strong>the</strong> c<strong>on</strong>figurati<strong>on</strong> adapted by<br />

Cunovic et al. [233] where <strong>the</strong>y have used <strong>the</strong> sec<strong>on</strong>d method of obtaining <strong>the</strong> pulse<br />

lengthbymapping<strong>the</strong>temporalco-ordinateof<strong>the</strong>pulsetospatial co-ordinates. In<strong>the</strong><br />

proof of principle experiment, outlined below, 32 nm radiati<strong>on</strong> from Flash was used<br />

to photo-i<strong>on</strong>ize Kr in <strong>the</strong> presence of a pulse from a Ti:Sapphire laser. In principle<br />

this is a single shot technique, but for this first experiment <strong>the</strong> data were too noisy<br />

<strong>and</strong> had to be summed over several shots.<br />

The unfocussed free electr<strong>on</strong> laser <strong>beam</strong>, of FWHM 7 mm, entered <strong>the</strong> experimental<br />

chamber c<strong>on</strong>taining low pressure Kr through an aperture 500 µm wide. The FEL<br />

<strong>beam</strong> was crossed at right angles with a focussed <strong>beam</strong> from a short pulse Ti:Sapphire<br />

laser. The widthof <strong>the</strong>X-ray<strong>beam</strong> in<strong>the</strong>chamber is equivalentto1.7 ps, muchl<strong>on</strong>ger<br />

than <strong>the</strong> laser pulse length of 150 fs, <strong>and</strong> <strong>on</strong>e has to assume that <strong>the</strong> intensity variati<strong>on</strong><br />

over <strong>the</strong> width of <strong>the</strong> central part of <strong>the</strong> X-ray <strong>beam</strong> accepted by <strong>the</strong> aperture<br />

is not great. Hence as <strong>the</strong> laser <strong>beam</strong> traverses <strong>the</strong> pulse, <strong>the</strong> intensity of <strong>the</strong> side<br />

b<strong>and</strong>s in <strong>the</strong> photo-electr<strong>on</strong> spectra will map out <strong>the</strong> intensity of <strong>the</strong> free electr<strong>on</strong><br />

laser <strong>beam</strong> as a functi<strong>on</strong> of time. The interacti<strong>on</strong> regi<strong>on</strong> is imaged by an electr<strong>on</strong><br />

lens system c<strong>on</strong>taining a retarding grid which <strong>on</strong>ly allows <strong>the</strong> passage of electr<strong>on</strong>s<br />

whose energy has been upshifted by <strong>the</strong> laser <strong>beam</strong>. A line will be formed in <strong>the</strong> 2D<br />

electr<strong>on</strong> detector, corresp<strong>on</strong>ding to <strong>the</strong> different emissi<strong>on</strong> time of electr<strong>on</strong>s, which in<br />

turn yields informati<strong>on</strong> about <strong>the</strong> intensity variati<strong>on</strong> of <strong>the</strong> X-ray pulse.<br />

In order to be able to use this technique for single shot measurement, <strong>the</strong> signal<br />

would need to be increased by about two orders of magnitude from that obtained in<br />

[233]. This could be achieved by increasing <strong>the</strong> target gas density in c<strong>on</strong>juncti<strong>on</strong> with<br />

<strong>the</strong> free electr<strong>on</strong> laser pulse energy, but <strong>the</strong>re is a limit to <strong>the</strong> increase before space<br />

charge effects become important. In <strong>the</strong> measurement reported in [233] <strong>the</strong> total<br />

pulse energy was 2 - 15 µJ before <strong>the</strong> entrance aperture to <strong>the</strong> chamber. If photoelectr<strong>on</strong>s<br />

with higher energies were produced, <strong>the</strong> signal would also be increased as<br />

<strong>the</strong> side-b<strong>and</strong> producti<strong>on</strong> increases with increasing energy as noted above.<br />

For use as an <strong>on</strong>line diagnostic, <strong>the</strong> experimental set-up would need to be redesigned<br />

without <strong>the</strong> entrance aperture as this will disrupt <strong>the</strong> <strong>beam</strong> through edge<br />

diffracti<strong>on</strong> effects. There does not seem to be an intrinsic need for aperturing <strong>the</strong><br />

<strong>beam</strong> <strong>on</strong> entry to <strong>the</strong> experimental chamber.<br />

An alternative to detecting electr<strong>on</strong>s from direct i<strong>on</strong>isati<strong>on</strong> is to use electr<strong>on</strong>s<br />

generated by Auger decay of an inner shell hole. The energy of <strong>the</strong> electr<strong>on</strong>s can also<br />

be modified by <strong>the</strong> presence of a str<strong>on</strong>g laser field, creating side-b<strong>and</strong>s. An advantage


9.3. Electro-optic techniques 155<br />

of using Auger electr<strong>on</strong>s is that <strong>the</strong>ir energy spectrum is fixed <strong>and</strong> does not depend<br />

<strong>on</strong> <strong>the</strong> energy width of <strong>the</strong> exciting X-ray pulse, <strong>and</strong> <strong>the</strong> Auger decay times are short<br />

compared with <strong>the</strong> typical durati<strong>on</strong> of free electr<strong>on</strong> laser pulses. If <strong>the</strong> laser pulse were<br />

chirped so that <strong>the</strong> side b<strong>and</strong>s are broadened, <strong>the</strong>n some measure of <strong>the</strong> X-ray pulse<br />

length can be obtained as well as <strong>the</strong> jitter <strong>–</strong> e.g. in <strong>the</strong> European X-Fel technical<br />

design report it is said that a linear chirp of 1 eV in a laser pulse of width of 300 fs<br />

would lead to a broadening of 0.25 eV of <strong>the</strong> sideb<strong>and</strong> for an 80 fs X-ray pulse[89].<br />

9.3 Electro-optic techniques<br />

The electro-optic technique uses birefringence induced in an electro-optic crystal by<br />

<strong>the</strong>electricfieldofTHzpulsesorelectr<strong>on</strong>pulses. Theamountofbirefringencedepends<br />

<strong>on</strong> <strong>the</strong> electric field <strong>and</strong> is probed by m<strong>on</strong>itoring <strong>the</strong> change of polarizati<strong>on</strong> of a<br />

short optical laser pulse. The limitati<strong>on</strong> to <strong>the</strong> THz regime for direct phot<strong>on</strong> pulse<br />

measurements is due <strong>the</strong> availability of crystals with a suitable electro-optic resp<strong>on</strong>se<br />

functi<strong>on</strong>.<br />

For electr<strong>on</strong> bunches, <strong>the</strong> electro-optic detecti<strong>on</strong> method makes use of <strong>the</strong> fact<br />

that <strong>the</strong> local electric field of a highly relativistic electr<strong>on</strong> bunch moving in a straight<br />

line is almost entirely c<strong>on</strong>centrated perpendicular to its directi<strong>on</strong> of moti<strong>on</strong>. The<br />

limitati<strong>on</strong>s to temporal resoluti<strong>on</strong> are discussed in Berden et al. [234] <strong>and</strong> include <strong>the</strong><br />

EO material selecti<strong>on</strong>, crystal thickness, wavelength dependence <strong>and</strong> <strong>beam</strong> <strong>–</strong> probe<br />

displacement. Philips et al. [235] have dem<strong>on</strong>strated measurement of electr<strong>on</strong> bunch<br />

lengths of 118 fs FWHM using a 35 fs probe laser at <strong>the</strong> Flash facility.<br />

The method can be used to m<strong>on</strong>itor <strong>the</strong> relative timing of an external laser with<br />

<strong>the</strong> electr<strong>on</strong> bunch, <strong>and</strong> using this informati<strong>on</strong> to improve <strong>the</strong> resoluti<strong>on</strong> of crosscorrelati<strong>on</strong><br />

data. A proof-of-principle experiment has been carried out at Flash by<br />

Azima et al. [236] where <strong>the</strong>y have measured <strong>the</strong> relative arrival time of <strong>the</strong> electr<strong>on</strong><br />

bunch<strong>and</strong> <strong>the</strong>laser pulseat <strong>the</strong>entrance to<strong>the</strong>free electr<strong>on</strong> laser as well as measuring<br />

<strong>the</strong> side-b<strong>and</strong> formati<strong>on</strong> in <strong>the</strong> photo-electr<strong>on</strong> spectra from Xe in <strong>the</strong> presence of <strong>the</strong><br />

laser field. A plot of side-b<strong>and</strong> intensity against <strong>the</strong> delay between <strong>the</strong> laser <strong>and</strong> <strong>the</strong><br />

free electr<strong>on</strong> laser has a rms width of 410 fs which is mainly due to jitter in <strong>the</strong> free<br />

electr<strong>on</strong> laser pulse. When <strong>the</strong> electro-optical data are used to correct for <strong>the</strong> jitter,<br />

<strong>the</strong> rms width of <strong>the</strong> curve is reduced to 100 fs. They call this technique TEO <strong>–</strong><br />

timing by electro-optic sampling.<br />

9.4 Autocorrelati<strong>on</strong> techniques<br />

The advent of ultra-short (< 100 fs) optical pulses from lasers has driven a significant<br />

amount of development into measuring <strong>the</strong> temporal profile of such pulses. Since<br />

<strong>the</strong>re are no detectors with a fast enough temporal resp<strong>on</strong>se to directly measure <strong>the</strong><br />

intensity profile of pulses with lengths shorter than a few hundred femtosec<strong>on</strong>ds,<br />

<strong>the</strong> approach taken has been to probe <strong>the</strong> pulse with ano<strong>the</strong>r short phot<strong>on</strong> pulse. Of<br />

course, this leads to <strong>the</strong> immediate problem of producing a suitably short probe pulse.<br />

In autocorrelati<strong>on</strong> measurements, <strong>the</strong> probe is created by splitting <strong>the</strong> original pulse<br />

into two (or in some cases more) parts that are each a replica of <strong>the</strong> original <strong>and</strong> <strong>on</strong>e<br />

part is used to probe <strong>the</strong> o<strong>the</strong>r. Since <strong>the</strong> pulse being measured is its own reference,<br />

autocorrelati<strong>on</strong> is thus used to study shape of pulses ra<strong>the</strong>r than timing jitter (though


156 9. Pulse length, profile <strong>and</strong> jitter<br />

a technique such as Spider (q.v.) can be used to derive <strong>the</strong> time correlati<strong>on</strong> functi<strong>on</strong><br />

of a train of pulses).<br />

Intensity autocorrelati<strong>on</strong><br />

When optical lasers producing ultra-short pulses were first developed, <strong>the</strong> instrument<br />

that was most likely to be employed to measure <strong>the</strong> pulses was a simple intensity<br />

autocorrelator. This was not because <strong>the</strong> autocorrelator was especially good at this<br />

job, but ra<strong>the</strong>r because autocorrelati<strong>on</strong> was about <strong>the</strong> <strong>on</strong>ly technique that could be<br />

used to measure pulses with durati<strong>on</strong>s below 100 fs since <strong>the</strong> fastest alternative<br />

instruments, such as streak cameras, were limited to temporal resoluti<strong>on</strong>s of several<br />

hundred femtosec<strong>on</strong>ds at best <strong>and</strong> more generally picosec<strong>on</strong>ds.<br />

A sec<strong>on</strong>d-order intensity autocorrelator is relatively easy to produce. The key<br />

comp<strong>on</strong>ents are a <strong>beam</strong> splitter to divide <strong>the</strong> <strong>beam</strong> into two (in principle identical)<br />

replicas of <strong>the</strong> input pulse, a means of altering <strong>the</strong> path length travelled by <strong>on</strong>e of <strong>the</strong><br />

replicas so as to vary <strong>the</strong> relative temporal delay between <strong>the</strong>m in precise increments,<br />

<strong>and</strong> a means of recombining <strong>the</strong>m so that <strong>the</strong>y overlap spatially. There must also be<br />

some means of measuring <strong>the</strong> temporal overlap as a functi<strong>on</strong> of <strong>the</strong> delay. Typically,<br />

this is achieved by spatially overlapping <strong>the</strong> two <strong>beam</strong>s in some instantaneouslyresp<strong>on</strong>ding,<br />

n<strong>on</strong>-linear medium so that a signal is produced that is proporti<strong>on</strong>al to<br />

<strong>the</strong> overlap intensity I(t)I(t−τ).<br />

A detector records <strong>the</strong> intensity of this signal, after separating it by some means<br />

from <strong>the</strong> transmitted replica pulses. Since <strong>the</strong> detector will inevitably have a resp<strong>on</strong>se<br />

thatisveryslowcomparedwith<strong>the</strong>ultra-shortpulselength, itautomatically performs<br />

a time integrati<strong>on</strong> of <strong>the</strong> signal. The signal as afuncti<strong>on</strong> of <strong>the</strong> relative temporal delay<br />

τ is thus <strong>the</strong> autocorrelati<strong>on</strong> functi<strong>on</strong> of <strong>the</strong> temporal intensity profile of <strong>the</strong> original<br />

pulse.<br />

�∞<br />

A(τ) = dtI(t)I(t−τ) (9.1)<br />

−∞<br />

Because <strong>the</strong> phase informati<strong>on</strong> is lost in this measurement, <strong>the</strong> Fourier transform of<br />

<strong>the</strong> autocorrelati<strong>on</strong> functi<strong>on</strong> is just <strong>the</strong> power spectrum of <strong>the</strong> original pulse. Autocorrelati<strong>on</strong><br />

thus gives <strong>the</strong> pulse length but not <strong>the</strong> pulse profile. Indeed, <strong>on</strong>e has to<br />

assume a pulse profile even to extract a pulse length. The autocorrelati<strong>on</strong> also tends<br />

to smear out any structure (such as satellite pulses) <strong>and</strong> in general any number of<br />

pulse shapes can give <strong>the</strong> same autocorrelati<strong>on</strong> functi<strong>on</strong>. Thus, changing <strong>the</strong> assumed<br />

pulse profile can give significant changes to <strong>the</strong> extracted pulse length. In general,<br />

autocorrelati<strong>on</strong> measurements are pr<strong>on</strong>e to systematic errors (for example caused by<br />

misalignment) since <strong>the</strong> <strong>on</strong>ly c<strong>on</strong>straints <strong>on</strong> <strong>the</strong> trace are that it should have a maximum<br />

at zero relative delay, be zero at delays much larger than <strong>the</strong> pulse width, <strong>and</strong><br />

be symmetric with respect to zero delay. In fact, <strong>the</strong> autocorrelati<strong>on</strong> of complicated<br />

pulses always tends to a sharp spike sitting <strong>on</strong> a smooth pedestal 1 .<br />

In <strong>the</strong> visible <strong>and</strong> UV regi<strong>on</strong>s <strong>the</strong> n<strong>on</strong>-linear medium is usually a sec<strong>on</strong>d harm<strong>on</strong>ic<br />

generati<strong>on</strong> (SHG) crystal which has a sec<strong>on</strong>d-order n<strong>on</strong>-linearity in its electric susceptibility<br />

<strong>and</strong> produces light at twice <strong>the</strong> frequency of <strong>the</strong> input light <strong>and</strong> with <strong>and</strong><br />

intensity that is proporti<strong>on</strong>al to <strong>the</strong> product of <strong>the</strong> intensities of <strong>the</strong> two input pulses.<br />

1 See, for instance, http://www.swampoptics.com/tutorials autocorrelati<strong>on</strong>.htm


9.4. Autocorrelati<strong>on</strong> techniques 157<br />

The SHG signal thus rises quadratically with intensity of <strong>the</strong> original pulse. The SHG<br />

signal is generally detected through a spectrometer to isolate it from <strong>the</strong> fundamental<br />

light.<br />

The short wavelength limit for SHG producti<strong>on</strong> is determined by <strong>the</strong> transmissi<strong>on</strong><br />

of <strong>the</strong> n<strong>on</strong>-linear crystals. The crystal must be transparent at half <strong>the</strong> wavelength<br />

of <strong>the</strong> incident light for <strong>the</strong> SHG signal to be detectable. Petrov et al. [237] have<br />

dem<strong>on</strong>strated autocorrelati<strong>on</strong> measurements to 250 nm (5 eV) using SrB4O7 (SBO),<br />

which is transparent to 125 nm. A fur<strong>the</strong>r limitati<strong>on</strong> of SHG is <strong>the</strong> finite b<strong>and</strong>width<br />

over which <strong>the</strong> crystal will functi<strong>on</strong>. It is important that <strong>the</strong> SHG signal be phasematched<br />

to <strong>the</strong> original pulses since o<strong>the</strong>rwise <strong>the</strong>re can be a spectral filtering effect<br />

that can significantly distort <strong>the</strong> autocorrelati<strong>on</strong> functi<strong>on</strong>[238]. For very short pulses,<br />

this means <strong>the</strong> crystals have to be very thin. The crystals also need to be thin to<br />

limit dispersi<strong>on</strong> that will change <strong>the</strong> pulse being measured. To overcome some of <strong>the</strong><br />

limitati<strong>on</strong>s of SHG from crystals, Dai et al. [239] have dem<strong>on</strong>strated using SHG from<br />

metal surfaces. This will work from <strong>the</strong> far-infrared to <strong>the</strong> plasma frequency of <strong>the</strong><br />

metal being used (e.g. about 10 eV for gold).<br />

Autocorrelati<strong>on</strong> measurements have also been dem<strong>on</strong>strated using two-phot<strong>on</strong> absorpti<strong>on</strong><br />

inside a semic<strong>on</strong>ductor-based phot<strong>on</strong> detector [240]. The advantage here is<br />

that <strong>the</strong> n<strong>on</strong>-linear medium <strong>and</strong>detector are combined <strong>and</strong> so <strong>the</strong> experimental set-up<br />

is simplified. However, a detector with a b<strong>and</strong>gap that is between <strong>the</strong> phot<strong>on</strong> energy<br />

<strong>and</strong> twice <strong>the</strong> phot<strong>on</strong> energy is required <strong>and</strong> so different materials will be required for<br />

different wavelengths. Extending <strong>the</strong> technique to <strong>the</strong> XUV is also unlikely to be possible.<br />

Worth noting is that Liu et al. [241, 242] have shown that autocorrelati<strong>on</strong> can<br />

give <strong>the</strong> full pulse profile if <strong>the</strong> pulse is split into three <strong>beam</strong>s. In this triple-optical<br />

autocorrelati<strong>on</strong> for direct pulse shape measurement (Toad), <strong>the</strong> delay between all <strong>the</strong><br />

pulses must be varied <strong>and</strong> <strong>the</strong> <strong>beam</strong>s are focused into a third-harm<strong>on</strong>ic generati<strong>on</strong><br />

(THG) crystal. Thus, although <strong>the</strong> technique requires <strong>on</strong>ly time-domain measurements,<br />

<strong>the</strong> extensi<strong>on</strong> to wavelengths shorter than <strong>the</strong> optical will be significantly<br />

more challenging than extending two-<strong>beam</strong> autocorrelati<strong>on</strong>.<br />

Making a two-<strong>beam</strong> autocorrelator that works at VUV <strong>and</strong> shorter wavelengths is<br />

challenging but not impossible. A successful instrument has been built for Flash by<br />

BESSY[137] <strong>and</strong> a simpler system has been made by <strong>the</strong> University of Hamburg[243]<br />

(cross-correlators between X-ray <strong>and</strong> optical pulses have also been achieved[244]).<br />

A significant challenge is achieving <strong>the</strong> initial <strong>beam</strong> splitting. Amplitude divisi<strong>on</strong><br />

whilst preserving <strong>the</strong> quality of <strong>the</strong> original pulse-fr<strong>on</strong>t is probably impossible at<br />

VUV to SXR wavelengths. In <strong>the</strong> BESSY instrument, <strong>the</strong> <strong>beam</strong> is divided spatially<br />

(i.e. wavefr<strong>on</strong>t divisi<strong>on</strong>) using a knife-edged mirror <strong>and</strong> <strong>on</strong>e must thus assume that<br />

<strong>the</strong> temporal profile of <strong>the</strong> <strong>beam</strong> is <strong>the</strong> same over <strong>the</strong> <strong>beam</strong> cross-secti<strong>on</strong>. Diffracti<strong>on</strong><br />

effects from <strong>the</strong> wavefr<strong>on</strong>t divisi<strong>on</strong> are a c<strong>on</strong>cern but are said to be minor in <strong>the</strong> focal<br />

plane[137]. A sec<strong>on</strong>d knife-edged mirror after <strong>the</strong> delay line is used to overlap <strong>the</strong><br />

two <strong>beam</strong>s by angling <strong>on</strong>e of <strong>the</strong>m slightly with respect to <strong>the</strong> o<strong>the</strong>r. The delay of<br />

<strong>on</strong>e <strong>beam</strong> with respect to <strong>the</strong> o<strong>the</strong>r is achieved through an optical delay line using<br />

translating mirrors. The relative delay range available is -3/+25 ps <strong>and</strong> is limited<br />

by <strong>the</strong> overall mechanical design <strong>and</strong> <strong>the</strong> size of <strong>the</strong> delay line mirrors, since <strong>the</strong><br />

<strong>beam</strong> moves al<strong>on</strong>g <strong>the</strong>se as <strong>the</strong> delay is varied. The mechanical specificati<strong>on</strong>s (e.g.<br />

mirror quality, angular precisi<strong>on</strong>, translati<strong>on</strong> accuracy <strong>and</strong> overall stability) are very<br />

challenging for short wavelength operati<strong>on</strong>.<br />

The shortest wavelength that can be measured is determined by <strong>the</strong> reflectivity<br />

of <strong>the</strong> mirrors since <strong>the</strong>re are four mirrors per <strong>beam</strong> path. Making <strong>the</strong> instrument


158 9. Pulse length, profile <strong>and</strong> jitter<br />

relatively compact prevents <strong>the</strong> use of extremely grazing angles of incidence; <strong>the</strong> fixed<br />

path arm uses 3 ◦ grazing whilst <strong>the</strong> variable path arm uses 6 ◦ grazing <strong>and</strong> <strong>the</strong>refore<br />

<strong>the</strong> intensity of <strong>the</strong>two pulses is not <strong>the</strong> same. With carb<strong>on</strong> coatings, good reflectivity<br />

is possible from 30 to 200 eV. Different coatings (e.g. nickel) could be used to extend<br />

this range to higher phot<strong>on</strong> energies.<br />

A mechanically much simpler instrument has been developed at Synchrotr<strong>on</strong> Soleil<br />

for use as a VUV Fourier Transform interferometer[245]. This also uses wavefr<strong>on</strong>t<br />

divisi<strong>on</strong> to split <strong>the</strong> <strong>beam</strong> but in this design two roof-mirrors are used so that <strong>the</strong><br />

optical assembly for each <strong>beam</strong> path is m<strong>on</strong>olithic. However, this means <strong>the</strong>re are<br />

two 90 ◦ deflecti<strong>on</strong>s in each path <strong>and</strong> so <strong>the</strong> l<strong>on</strong>gest operating wavelength is in <strong>the</strong><br />

VUV.<br />

Autocorrelati<strong>on</strong> measurements <strong>on</strong> pulses from an HHG source using ano<strong>the</strong>r splitmirror<br />

approach have been reported by Tzallas et al. [246]. In <strong>the</strong>ir work, a focusing<br />

wavefr<strong>on</strong>t divideris madebysplittingaspherical mirror intotwohalves. Atranslati<strong>on</strong><br />

of <strong>on</strong>e half-mirror al<strong>on</strong>g <strong>the</strong> surface normal gives a relative delay between <strong>the</strong> parts of<br />

<strong>the</strong> <strong>beam</strong> reflected from each half whilst <strong>the</strong> mirror also focuses <strong>the</strong> split <strong>beam</strong>s to a<br />

comm<strong>on</strong> positi<strong>on</strong> in <strong>the</strong> detecti<strong>on</strong> medium. Two-phot<strong>on</strong> i<strong>on</strong>ized He gas is used as <strong>the</strong><br />

n<strong>on</strong>-linear medium <strong>and</strong> <strong>the</strong> He i<strong>on</strong> yield measured by a time-of-flight spectrometer.<br />

Stigmatic imaging with a spherical mirror requires <strong>the</strong> system to operate at nearnormal<br />

incidence. The upperphot<strong>on</strong> energy is thus limited ifasimple metallic coating<br />

is used. Multilayer coatings could be used to allow operati<strong>on</strong> at shorter wavelengths,<br />

though <strong>the</strong> operating b<strong>and</strong>width would be quite narrow for a given multilayer.<br />

The final challenge in <strong>the</strong> XUV autocorrelator is a means of measuring <strong>the</strong> autocorrelati<strong>on</strong><br />

signal. In <strong>the</strong> measurements reported in [247] using <strong>the</strong> BESSY/Flash<br />

instrument, <strong>the</strong> temporal coherence was deduced from <strong>the</strong> fringe visibility of <strong>the</strong> spatial<br />

interference pattern as a functi<strong>on</strong> of relative delay. In <strong>the</strong>se measurements, <strong>the</strong><br />

mutual coherence functi<strong>on</strong>, which is closely related to <strong>the</strong> auto-correlati<strong>on</strong> functi<strong>on</strong>,<br />

is measured. The mutual coherence functi<strong>on</strong> is defined by[248]:<br />

�<br />

Γ12(τ) =<br />

u1(t+τ)u ∗<br />

2(t)dt<br />

where ui:s are <strong>the</strong> field values at <strong>the</strong> two slits.<br />

An alternative detecti<strong>on</strong> technique that gives a sec<strong>on</strong>d-order signal like SHG is<br />

two-phot<strong>on</strong> i<strong>on</strong>isati<strong>on</strong> [249<strong>–</strong>251]. The strength of <strong>the</strong> two-phot<strong>on</strong> i<strong>on</strong>isati<strong>on</strong> signal<br />

produced when <strong>on</strong>e phot<strong>on</strong> is c<strong>on</strong>tributed from each <strong>beam</strong> is clearly proporti<strong>on</strong>al <strong>the</strong><br />

temporal intensity overlap of <strong>the</strong> two <strong>beam</strong>s. The wavelength range over which such<br />

schemes will work depends <strong>on</strong> <strong>the</strong> i<strong>on</strong>isati<strong>on</strong> potential of <strong>the</strong> gas. For first i<strong>on</strong>isati<strong>on</strong><br />

potentials, this ranges from 12-24 eV for helium to 4.5 to 9 eV for toluene.<br />

Two-phot<strong>on</strong> single-i<strong>on</strong>isati<strong>on</strong> above 24 eV is unlikely to be practical due to <strong>the</strong><br />

need to discriminate against single-phot<strong>on</strong> i<strong>on</strong>isati<strong>on</strong> events which will give an increasingly<br />

str<strong>on</strong>g background of first order signal as <strong>the</strong> phot<strong>on</strong> energy increases.<br />

Nakajima <strong>and</strong> Nikolopoulos [252, 253] have made a <strong>the</strong>oretical study of using twophot<strong>on</strong><br />

doublei<strong>on</strong>isati<strong>on</strong> of helium, which would cover <strong>the</strong> range 40 to 54 eV. This<br />

scheme could be extended to shorter wavelengths by using heavier elements (Li, Be,<br />

B etc) <strong>and</strong> <strong>the</strong>ir higher i<strong>on</strong>isati<strong>on</strong> stages [254]. Never<strong>the</strong>less, c<strong>on</strong>tinuous coverage of<br />

a wide phot<strong>on</strong> energy range will not be possible with i<strong>on</strong>isati<strong>on</strong> based techniques. In<br />

summary, autocorrelati<strong>on</strong> is a feasible pulse length measuring technique in <strong>the</strong> VUV<br />

<strong>and</strong> XUV, though <strong>on</strong>ly <strong>the</strong> pulse length is measured <strong>and</strong> <strong>the</strong> technique is usually


9.4. Autocorrelati<strong>on</strong> techniques 159<br />

multi-shot, though adapti<strong>on</strong> to single shot should be possible (see for example <strong>the</strong><br />

descripti<strong>on</strong> of single shot Frog in secti<strong>on</strong> 3.2.1). The pulse length extracti<strong>on</strong> is also<br />

not particularly robust due to <strong>the</strong> need to assume a pulse profile <strong>and</strong> <strong>the</strong> susceptibility<br />

to systematic errors. Autocorrelators do not give informati<strong>on</strong> of pulse timing (jitter).<br />

The mechanical dem<strong>and</strong>s of <strong>the</strong> instrument are high but can be surmounted. Finding<br />

a suitable n<strong>on</strong>-linear detecti<strong>on</strong> technique to measure <strong>the</strong> auto-correlati<strong>on</strong> signal is<br />

even more dem<strong>and</strong>ing <strong>and</strong> unlikely to give c<strong>on</strong>tinuous wavelength coverage. Autocorrelators<br />

are invasive in that <strong>the</strong>y modify <strong>the</strong> <strong>beam</strong> passing through <strong>the</strong>m, but <strong>the</strong>y<br />

can be designed so that <strong>the</strong>y can be inserted into or removed from <strong>the</strong> <strong>beam</strong> path<br />

without affecting <strong>the</strong> operati<strong>on</strong> of downstream optical elements.<br />

Autocorrelati<strong>on</strong> techniques for complete pulse characterizati<strong>on</strong><br />

Complete characterisati<strong>on</strong> of <strong>the</strong> pulse intensity profile requires <strong>the</strong> measurement of<br />

both <strong>the</strong> spectral <strong>and</strong> phase informati<strong>on</strong> so that <strong>the</strong> electric field can be computed<br />

as a functi<strong>on</strong> of time. There are three distinct approaches to this that have been<br />

used in <strong>the</strong> visible <strong>and</strong> near-visible spectral regi<strong>on</strong>s, viz. spectrographic, tomographic<br />

<strong>and</strong> interferometric. In all cases, filtering of <strong>the</strong> pulses is required. It is <strong>the</strong> practical<br />

realizati<strong>on</strong> of some of <strong>the</strong>se filters that make extending <strong>the</strong> techniques to shorter<br />

wavelengths so difficult.<br />

The two classes of filter that are important in current pulse length analysis techniques<br />

are time-stati<strong>on</strong>ary filters (in which <strong>the</strong> time of incidence of <strong>the</strong> input pulse<br />

does not affect <strong>the</strong> output) <strong>and</strong> frequency stati<strong>on</strong>ary filters (where <strong>the</strong> output is not<br />

changed by arbitrary shifts in frequency of <strong>the</strong> input). Frequency-stati<strong>on</strong>ary filters<br />

are time n<strong>on</strong>-stati<strong>on</strong>ary. These filters can be fur<strong>the</strong>r classified as amplitude-<strong>on</strong>ly or<br />

phase-<strong>on</strong>ly i.e. <strong>the</strong>y modulate <strong>on</strong>ly <strong>the</strong> amplitude or <strong>on</strong>ly <strong>the</strong> phase of <strong>the</strong> input.<br />

Finally, <strong>the</strong> filter may have a linear or n<strong>on</strong>-linear resp<strong>on</strong>se (with frequency or time as<br />

appropriate).<br />

A n<strong>on</strong>-dispersive delay line is a simple filter that adds <strong>the</strong> same time delay to all<br />

<strong>the</strong> frequency comp<strong>on</strong>ents in <strong>the</strong> pulse. A delay line is thus a linear spectral phase<br />

modulator, since a linear (with frequency) shift in <strong>the</strong> spectral phase is equivalent to<br />

a time shift. A spectrometer is an example of spectral amplitude filter.<br />

These are both time-stati<strong>on</strong>ary filters. Both <strong>the</strong>se types of filter are relatively easy<br />

to implement across a wide range of wavelengths, though some thought must be given<br />

to <strong>the</strong> frequency resp<strong>on</strong>se functi<strong>on</strong> (e.g. b<strong>and</strong>width, resoluti<strong>on</strong>) of practical devices.<br />

The simplest time-n<strong>on</strong>-stati<strong>on</strong>ary filters are a time-gate <strong>and</strong> a frequency shifter.<br />

A time gate is a time-n<strong>on</strong>stati<strong>on</strong>ary amplitude filter <strong>and</strong> is used to take time slices<br />

of a pulse. A linear temporal phase modulator gives a linear variati<strong>on</strong> of phase with<br />

time, which is <strong>the</strong> same as a translati<strong>on</strong> or shift of <strong>the</strong> frequency axis, <strong>and</strong> thus gives<br />

a spectral shift or shear to <strong>the</strong> pulse. Time-n<strong>on</strong>-stati<strong>on</strong>ary filters are more difficult to<br />

implement, especially for ultra-short pulses.<br />

Spectrographic techniques are probably <strong>the</strong> most widely used pulse profiling techniques.<br />

They work by measuring a two-dimensi<strong>on</strong>al representati<strong>on</strong> of <strong>the</strong> <strong>on</strong>e- dimensi<strong>on</strong>al<br />

field, i.e. <strong>the</strong>y are phase-space measurements. This is <strong>the</strong> critical step since it<br />

allows a generally unambiguous retrieval of <strong>the</strong> phase informati<strong>on</strong> in a way that is not<br />

possible with a <strong>on</strong>e-dimensi<strong>on</strong>al measurement. The <strong>on</strong>ly ambiguities are <strong>the</strong> absolute<br />

phase <strong>and</strong> absolute arrival time.


160 9. Pulse length, profile <strong>and</strong> jitter<br />

Frequency Resolved Optical Gating (FROG)<br />

An example of a spectrographic technique is Frog or Frequency-Resolved Optical<br />

Gating. This uses a sequential spectral filter <strong>and</strong> a time-gate, in ei<strong>the</strong>r order, followed<br />

by an intensity detector. Depending <strong>on</strong> <strong>the</strong> order of <strong>the</strong> filters, <strong>the</strong> recorded signal<br />

is a measure ei<strong>the</strong>r of <strong>the</strong> spectrum of a series of time slices or a measure of <strong>the</strong><br />

time of arrival of a series of spectral slices. The technique is thus operating in <strong>the</strong><br />

timefrequency domain (phase-space) <strong>and</strong> is both temporally <strong>and</strong> spectrally resolved.<br />

The technique is not limited to just autocorrelati<strong>on</strong> measurements (i.e. it will work<br />

with external gate pulses), but in all practical applicati<strong>on</strong>s <strong>the</strong> pulse is used to analyse<br />

itself in manner that is an extensi<strong>on</strong> of intensity autocorrelati<strong>on</strong>.<br />

τ<br />

input-pulse<br />

Beam-splitter<br />

Probe, E(t)<br />

Gate, E(t − τ)<br />

Spectrometer<br />

N<strong>on</strong>-linear medium<br />

Figure 9.2: The basic layout of a Frog experiment; τ marks <strong>the</strong> variable time-delay between <strong>the</strong> probe <strong>and</strong><br />

gate pulses.<br />

In <strong>the</strong> first descripti<strong>on</strong> of Frog by Kane <strong>and</strong> Trebino [255], a replica of <strong>the</strong> pulse<br />

to be measured is used as <strong>the</strong> time gate. Thus, <strong>the</strong> initial pulse is split into two pulses<br />

(gate g(t) <strong>and</strong> probe E(t)) <strong>and</strong> <strong>the</strong> gate pulse has a variable temporal delay applied<br />

to it <strong>–</strong> see Figure 9.2. The two pulses are focused <strong>and</strong> overlapped spatially in an<br />

instantaneous n<strong>on</strong>-linear medium (in this case, self-diffracti<strong>on</strong> due to <strong>the</strong> electr<strong>on</strong>ic<br />

Kerr effect in glass is <strong>the</strong> n<strong>on</strong>-linear process). The diffracted light is <strong>the</strong>n passed to a<br />

spectrometer <strong>and</strong> a complete spectrum recorded, i.e. <strong>the</strong> spectrogram:<br />

�<br />

� �∞<br />

�<br />

SE(ω,τ) = �<br />

� E(t)g(t−τ)e<br />

�<br />

−iωt �<br />

�2<br />

�<br />

dt�<br />

�<br />

�<br />

−∞<br />

The spectrogram is recorded for a range of relative delays of <strong>the</strong> probe <strong>and</strong> gate that<br />

is sufficiently wide to give zero temporal overlap of <strong>the</strong> gate as it is shifted from before<br />

to after <strong>the</strong> probe pulse. Frog is thus essentially a spectrally resolved autocorrelati<strong>on</strong><br />

(as seen from <strong>the</strong> similarity of <strong>the</strong> equati<strong>on</strong> above <strong>and</strong> Equati<strong>on</strong> 9.1).


9.4. Autocorrelati<strong>on</strong> techniques 161<br />

If self-diffracti<strong>on</strong> is used as <strong>the</strong> n<strong>on</strong>-linear effect <strong>the</strong> signal pulse is given by<br />

ES(t,τ) ∝ [E(t)] 2 E ∗ (t−τ), which yields<br />

�<br />

� �<br />

�<br />

Ifrog(ω,τ) = �<br />

�<br />

�<br />

∞<br />

−∞<br />

[E(t)] 2 E ∗ (t−τ)e −iωt dt<br />

It does not matter (i.e. it does not degrade <strong>the</strong> temporal resoluti<strong>on</strong>) that <strong>the</strong> gate<br />

has <strong>the</strong> same time-width as <strong>the</strong>pulse beingmeasured, though ashorter pulse would be<br />

preferable. However, <strong>the</strong> gate pulse should not be too short as an infinitely short gate<br />

yields <strong>on</strong>ly intensity informati<strong>on</strong>, (whilst a CW gate would yield <strong>on</strong>ly <strong>the</strong> spectrum).<br />

The use of <strong>the</strong> pulse to gate itself does however complicate <strong>the</strong> inversi<strong>on</strong> process since<br />

<strong>on</strong>e cannot input any knowledge of <strong>the</strong> gate pulse into <strong>the</strong> analysis.<br />

Ano<strong>the</strong>rpointt<strong>on</strong>oteis that, if<strong>the</strong>Frog measurement recordsan equalnumberN<br />

temporal slices <strong>and</strong> spectral slices, <strong>the</strong>n N 2 measurements are made in total. But <strong>the</strong><br />

analysis will yield <strong>on</strong>ly N intensity <strong>and</strong> N phase values, i.e. 2N derived values. There<br />

is thus a lot of data redundancy in <strong>the</strong> measurement, though this does c<strong>on</strong>tribute<br />

to making <strong>the</strong> data inversi<strong>on</strong> give unambiguous results. In fact, <strong>the</strong> unambiguous<br />

nature of Frog is a <strong>on</strong>e of its most important features (<strong>and</strong> is quite c<strong>on</strong>trary to<br />

simple autocorrelati<strong>on</strong>). Although <strong>the</strong> technique as described above is multi-shot<br />

due to <strong>the</strong> need to scan <strong>the</strong> time delay, <strong>the</strong> technique can be adapted for single shot<br />

measurements [255, 256]. This is achieved by focusing <strong>the</strong> pulses to lines in a comm<strong>on</strong><br />

plane <strong>and</strong> crossing <strong>the</strong> lines at an angle. The positi<strong>on</strong> al<strong>on</strong>g ei<strong>the</strong>r of <strong>the</strong> line foci<br />

is now a linear functi<strong>on</strong> of relative delay between <strong>the</strong> two pulses. If <strong>the</strong> line foci are<br />

orthog<strong>on</strong>al to <strong>the</strong> dispersi<strong>on</strong> plane of <strong>the</strong> spectrometer, <strong>the</strong>n an imaging spectrometer<br />

with 2-D detector will be able to record spectrum in <strong>on</strong>e plane simultaneously with<br />

delay in <strong>the</strong> orthog<strong>on</strong>al plane. Thus <strong>the</strong> spectrum can be recorded as a functi<strong>on</strong> of<br />

delay <strong>and</strong> frequency in <strong>on</strong>e shot. The length of <strong>the</strong> line foci <strong>and</strong> relative angle will<br />

determine <strong>the</strong> range of delays that is recorded, which must be sufficient to give zero<br />

overlap of <strong>the</strong> pulses at each end for <strong>the</strong> Frog measurement to be successful. The size<br />

of <strong>the</strong> delay step is determined by <strong>the</strong> spatial resoluti<strong>on</strong> of <strong>the</strong> spectrometer detector.<br />

The experimental c<strong>on</strong>figurati<strong>on</strong> as described in [255] is quite simple since both<br />

<strong>the</strong> delay line <strong>and</strong> spectrometer are straightforward. Though <strong>the</strong> technique has <strong>the</strong><br />

disadvantage of being invasive, it would seem to be extensible to wavelengths shorter<br />

than <strong>the</strong> visible <strong>and</strong> UV. The key limitati<strong>on</strong>s are <strong>the</strong> need to find a suitable <strong>beam</strong><br />

splitter, <strong>and</strong> in particular <strong>the</strong> need for a n<strong>on</strong>-linear process to mix <strong>the</strong> two <strong>beam</strong>s<br />

<strong>and</strong> give a signal proporti<strong>on</strong>al to <strong>the</strong> combined intensity. As with autocorrelati<strong>on</strong>,<br />

two-phot<strong>on</strong> i<strong>on</strong>isati<strong>on</strong> would be possible n<strong>on</strong>-linear process, at least at lower phot<strong>on</strong><br />

energies.<br />

Norin et al. [257] used two-photo i<strong>on</strong>isati<strong>on</strong> from Xen<strong>on</strong> to study <strong>the</strong> chirp of <strong>the</strong><br />

5th harm<strong>on</strong>ic (15.5 eV) radiati<strong>on</strong> produced by HHG in a method that is described<br />

as similar to Frog. The HHG is produced from xen<strong>on</strong> from frequency doubled IR<br />

radiati<strong>on</strong> (Ti:Sapph)whilst<strong>the</strong>probepulseissplitofffrom <strong>the</strong>mainIR<strong>beam</strong>. Amagnetic<br />

bottle spectrometer is used to collect <strong>the</strong> photoelectr<strong>on</strong> spectrum as a functi<strong>on</strong><br />

of probe <strong>beam</strong> delay. The intensity of <strong>the</strong> sideb<strong>and</strong> corresp<strong>on</strong>ding to <strong>the</strong> absorpti<strong>on</strong><br />

of <strong>on</strong>e 15.5 eV <strong>and</strong> <strong>on</strong>e IR phot<strong>on</strong> is measured as a functi<strong>on</strong> of relative delay. This<br />

allowed <strong>the</strong> extracti<strong>on</strong> of <strong>the</strong> linear chirp in <strong>the</strong> HHG pulse <strong>and</strong> its length, but not<br />

<strong>the</strong> actual pulse shape.<br />

�<br />

�<br />

�<br />

�<br />

�<br />

�<br />

2


162 9. Pulse length, profile <strong>and</strong> jitter<br />

Sekikawa et al. [258] were able to fully characterize <strong>the</strong> 5th harm<strong>on</strong>ic pulse of a<br />

Ti:Sapphire laser using twophot<strong>on</strong> i<strong>on</strong>isati<strong>on</strong> in Frog (TPI Frog). This is because<br />

<strong>the</strong>y show that <strong>the</strong> TPI spectrum is equivalent to <strong>the</strong> spectrally resolved SHG used<br />

in ”c<strong>on</strong>venti<strong>on</strong>al” Frog. They also claim <strong>the</strong> technique is scaleable to XUV <strong>and</strong> SXR<br />

pulses through <strong>the</strong> detecti<strong>on</strong> of two-phot<strong>on</strong> absorpti<strong>on</strong> from <strong>the</strong> K-shell to a free state<br />

in bor<strong>on</strong>.<br />

Frog is probably <strong>the</strong> most widely used technique for measuring <strong>the</strong> pulse profile of<br />

ultra-short optical pulses. There are a number of different geometries for measuring<br />

Frog as described by Trebino et al. [256]. They are summarized briefly below.<br />

Polarizati<strong>on</strong>-gate FROG (PG FROG)<br />

The basic layout of PG Frog is shown in Figure 9.3. The input pulse is split into two<br />

equal replicas. One replica (<strong>the</strong>probe) is sent throughcrossed polarisers <strong>and</strong> <strong>the</strong>o<strong>the</strong>r<br />

(<strong>the</strong> gate) through a half-wave plate to give a ±45 ◦ linear polarisati<strong>on</strong> with respect<br />

to <strong>the</strong> first. The two replicas are <strong>the</strong>n spatially overlapped in a material with a fast<br />

third-order susceptibility such as fused silica. The gate pulse induces birefringence<br />

in <strong>the</strong> silica through <strong>the</strong> electr<strong>on</strong>ic Kerr effect <strong>and</strong> so <strong>the</strong> silica acts as a wave plate<br />

<strong>and</strong> rotates <strong>the</strong> polarisati<strong>on</strong> of <strong>the</strong> probe <strong>beam</strong> slightly which allows some light to be<br />

transmitted through a polarisati<strong>on</strong> analyser. Because this occurs <strong>on</strong>ly when <strong>the</strong> gate<br />

pulse is present in <strong>the</strong> silica, <strong>the</strong> transmitted intensity as a functi<strong>on</strong> of relative delay<br />

is an autocorrelati<strong>on</strong> measurement of <strong>the</strong> pulse. Spectrally resolving <strong>the</strong> transmitted<br />

light thus gives <strong>the</strong> Frog measurement.<br />

λ 2 -plate<br />

τ<br />

Polariser<br />

input-pulse<br />

Beam-splitter<br />

Probe, E(t)<br />

Gate, E(t − τ)<br />

Photodetector<br />

Figure 9.3: The basic layout of PG Frog.<br />

Fused Silica<br />

Filter<br />

Polariser<br />

The biggest advantage of PG Frog is that <strong>the</strong>re are no ambiguities <strong>on</strong> inversi<strong>on</strong><br />

so that <strong>the</strong> pulse characterisati<strong>on</strong> is complete in all cases. Ano<strong>the</strong>r advantage is that<br />

<strong>the</strong> n<strong>on</strong>-linear process is automatically phase matched so alignment is easy. The main<br />

disadvantage is that <strong>the</strong> polarisers must be of high quality; an extincti<strong>on</strong> coefficient


9.4. Autocorrelati<strong>on</strong> techniques 163<br />

of better than 10 −5 is recommended. This makes <strong>the</strong> polarisers expensive <strong>and</strong> more<br />

importantly fairly thick, which can introducedispersi<strong>on</strong> to<strong>the</strong> pulse<strong>and</strong> sochange <strong>the</strong><br />

pulse being measured, a particular problem for <strong>the</strong> shortest of pulses. Also, because<br />

a third-order n<strong>on</strong>-linearity is used, <strong>the</strong> sensitivity of <strong>the</strong> technique is reduced.<br />

The fact that PG Frog polarises <strong>the</strong> <strong>beam</strong> is <strong>the</strong> biggest impediment to extending<br />

<strong>the</strong> technique to wavelengths shorter than <strong>the</strong> UV. Polarisers <strong>and</strong> half-wave plates<br />

with <strong>the</strong>requiredperformance are impossible tomakefor <strong>the</strong>XUV<strong>and</strong>X-rayregimes.<br />

Achieving a large phase shift in this spectral range is <strong>on</strong>ly possible using anomalous<br />

dispersi<strong>on</strong> near an absorpti<strong>on</strong> edge <strong>and</strong> so polarisers would be limited to narrow<br />

spectral ranges. Even <strong>the</strong>n, absorpti<strong>on</strong> is str<strong>on</strong>g <strong>and</strong> <strong>the</strong> performance would not meet<br />

<strong>the</strong> exacting requirements for PG Frog. Finally, a third-order n<strong>on</strong>-linear process<br />

wouldbeneededtoextract<strong>the</strong>Frog signal. Thereseemslittle possibilityofextending<br />

PG FROG to wavelengths shorter than 250 nm (5 eV).<br />

Self-diffracti<strong>on</strong> FROG (SD FROG)<br />

Self-diffracti<strong>on</strong>Frogis<strong>the</strong>techniqueasoriginally describedbyKane<strong>and</strong>Trebino[255]<br />

<strong>–</strong> see Figure 9.2. As with PG Frog, <strong>the</strong> electr<strong>on</strong>ic Kerr effect is used as a third-order<br />

n<strong>on</strong>-linear process. In SD Frog however, <strong>the</strong> intensity oscillati<strong>on</strong>s of <strong>the</strong> interfering<br />

<strong>beam</strong>s induce a refractive-index grating in <strong>the</strong> silica <strong>and</strong> this diffracts each of<br />

<strong>the</strong> <strong>beam</strong>s in a different directi<strong>on</strong>. The first order diffracti<strong>on</strong> of <strong>on</strong>e of <strong>the</strong> <strong>beam</strong>s is<br />

spectrally resolved to make <strong>the</strong> Frog measurement.<br />

The advantage of SD Frog <strong>the</strong>refore, is that <strong>the</strong> <strong>beam</strong>s can have <strong>the</strong> same polarisati<strong>on</strong><br />

<strong>and</strong> polarisers are not required. Applicati<strong>on</strong> to bey<strong>on</strong>d <strong>the</strong> UV is <strong>the</strong>refore<br />

potentially more straightforward than with PG Frog.<br />

However, a third-order n<strong>on</strong>-linear process is still required. Self-diffracti<strong>on</strong> is not actually<br />

ideal even in <strong>the</strong> visible since it is not a phase-matched process. The n<strong>on</strong>-linear<br />

medium must <strong>the</strong>refore be kept thin (


164 9. Pulse length, profile <strong>and</strong> jitter<br />

as sec<strong>on</strong>d-order processes are inherently more efficient than third-order <strong>on</strong>es. The<br />

probe <strong>and</strong> gate <strong>beam</strong>s are overlapped spatially in an SHG crystal <strong>and</strong> a signal at<br />

twice <strong>the</strong>ir frequency is produced with intensity that is proporti<strong>on</strong>al to <strong>the</strong> product<br />

of <strong>the</strong> individual intensities. The SHG signal is spectrally resolved as functi<strong>on</strong> of <strong>the</strong><br />

relative pulse delay. Since each frequency in <strong>the</strong> original pulse is up-c<strong>on</strong>verted by <strong>the</strong><br />

crystal, <strong>the</strong> SHG spectrum is directly correlated to <strong>the</strong> original pulse spectrum. It<br />

is thus important to ensure that <strong>the</strong> crystal works over a wide enough b<strong>and</strong>width to<br />

up-c<strong>on</strong>vert <strong>the</strong> entire b<strong>and</strong>width of <strong>the</strong> original pulse.<br />

Failure to record <strong>the</strong> entire pulse b<strong>and</strong>width will prevent <strong>the</strong> Frog inversi<strong>on</strong> from<br />

working. Since <strong>the</strong> crystal b<strong>and</strong>width is inversely proporti<strong>on</strong>al to crystal thickness,<br />

short pulses require very thin crystals, e.g. for measuring 100 fs pulse at 800 nm,<br />

<strong>the</strong> maximum crystal thickness should be ∼ 300 µm for KDP (potassium dihydrogen<br />

phosphate) <strong>and</strong> ∽ 100 µm for BBO (β-barium borate).<br />

Of course, operati<strong>on</strong> at VUV to X-ray wavelengths will require <strong>the</strong> crystal to be<br />

replaced by ano<strong>the</strong>r n<strong>on</strong>-linear medium in any case. The opti<strong>on</strong>s for this are discussed<br />

in Secti<strong>on</strong> 9.4. SHG Frog has <strong>the</strong> most potential to be extended to operati<strong>on</strong> at<br />

wavelengths shorter than <strong>the</strong> visible/UV part of <strong>the</strong> spectrum. A final point that<br />

should be noted with SHG Frog is that it is ambiguous with respect to <strong>the</strong> directi<strong>on</strong><br />

of time since SHG Frog traces are always symmetric with delay. Thus it is not<br />

possible to tell if a pulse with a satellite has <strong>the</strong> satellite before or after <strong>the</strong> main<br />

pulse.<br />

Third-harm<strong>on</strong>ic-generati<strong>on</strong> FROG (THG FROG)<br />

THG Frog is very similar to SHG Frog except that third-harm<strong>on</strong>ic generati<strong>on</strong> is<br />

used as <strong>the</strong> n<strong>on</strong>-linear process. The advantage of this is that <strong>the</strong> third-order n<strong>on</strong>linear<br />

process removes <strong>the</strong> time-directi<strong>on</strong> ambiguity inherent in SHG Frog (except<br />

in <strong>the</strong> particular case of a Gaussian pulse profile with a purely linear chirp).<br />

A potentially important variati<strong>on</strong> of THG Frog is where surface third-harm<strong>on</strong>ic<br />

generati<strong>on</strong> is used (STHG)[259]. Not <strong>on</strong>ly is STHG a relatively efficient process<br />

(compared to o<strong>the</strong>r third-order processes) but by interacting <strong>on</strong>ly at <strong>the</strong> surface <strong>the</strong><br />

phase-matching b<strong>and</strong>width is very large. This is of particular value when measuring<br />

pulses of <strong>on</strong>ly a few fs in durati<strong>on</strong> which require SHG crystals of <strong>on</strong>ly a few tens of<br />

µm in thickness.<br />

Table 9.1: Different FROG Schemes<br />

Type Esig(t,τ)<br />

SHG E(t)E(t−τ)<br />

Pol. Gate E(t)|E(t−τ)| 2<br />

Self. diff. E(t) 2 E(t−τ) ∗<br />

THG E(t) 2 E(t−τ)


9.5. SPIDER 165<br />

9.5 Spectral Phase Interferometry for Direct Electric-field Rec<strong>on</strong>structi<strong>on</strong><br />

An example of an interferometric pulse profiling technique is Spectral Phase Interferometry<br />

for Direct Electric-field Rec<strong>on</strong>structi<strong>on</strong> or Spider[260]. The c<strong>on</strong>cept of <strong>the</strong><br />

technique is to have a linear spectral phase filter in parallel with a linear temporal<br />

phase filter. These are followed by a spectral amplitude filter before an intensity<br />

detector. The input pulse is divided into two replicas <strong>and</strong> <strong>on</strong>e replica passes through<br />

each of <strong>the</strong> phase filters before being recombined <strong>and</strong> passing through <strong>the</strong> amplitude<br />

filter before reaching <strong>the</strong> detector. If <strong>the</strong> linear temporal phase filter is adjusted from<br />

null, <strong>the</strong>n <strong>on</strong>e of <strong>the</strong> replicas is given a spectral shift relative to <strong>the</strong> o<strong>the</strong>r. On recombining<br />

<strong>the</strong> pulses <strong>the</strong>y interfere spectrally (beat) <strong>and</strong> <strong>the</strong> apparatus becomes a<br />

spectral shearing interferometer. The spectral interferogram is recorded <strong>on</strong> <strong>the</strong> detector<br />

as <strong>the</strong> spectral amplitude filter is tuned over <strong>the</strong> b<strong>and</strong>width of <strong>the</strong> pulse, i.e. <strong>the</strong><br />

spectrometer resolves <strong>the</strong> frequency mixed signal. This informati<strong>on</strong> toge<strong>the</strong>r with<br />

knowledge of <strong>the</strong> applied spectral shear is sufficient to rec<strong>on</strong>struct <strong>the</strong> electric field of<br />

<strong>the</strong> pulse through direct inversi<strong>on</strong>.<br />

The linear spectral phase filter can be used to introduce a temporal delay between<br />

<strong>the</strong> two replica pulses. This additi<strong>on</strong>al degree of freedom allows <strong>the</strong> correlati<strong>on</strong> of<br />

pulses in a train of n<strong>on</strong>-identical pulses to be calculated[260], but o<strong>the</strong>rwise can, in<br />

principle, besettoanarbitraryvalue. Inpracticehowever, <strong>the</strong>timedelayisimportant<br />

as it adds an extra phase to <strong>the</strong> frequency spectrum of <strong>on</strong>e pulse replica relative<br />

to <strong>the</strong> o<strong>the</strong>r <strong>and</strong> this is used to ensure several fringes per independent frequency<br />

comp<strong>on</strong>ent. This makes an unambiguous reading of <strong>the</strong> spectral phases from <strong>the</strong><br />

spectral interferogram possible (because <strong>the</strong> ac <strong>and</strong> dc terms of <strong>the</strong> Fourier transform<br />

are well separated) <strong>and</strong> thus ensures that <strong>the</strong> inversi<strong>on</strong> can successfully recover <strong>the</strong><br />

electric field.<br />

In terms of <strong>the</strong> practical implementati<strong>on</strong> of Spider, <strong>the</strong> all-important temporal<br />

phase filter is <strong>the</strong> hardest to implement. The spectral phase filter is just a n<strong>on</strong>dispersive<br />

delay line <strong>and</strong> <strong>the</strong> spectral amplitude filter is a spectrometer with sufficient<br />

b<strong>and</strong>width <strong>and</strong> resoluti<strong>on</strong> to record <strong>the</strong> pulse b<strong>and</strong>width <strong>and</strong> resolve <strong>the</strong> spectral<br />

interference fringes. Because <strong>the</strong> final measurement is spectral, Spidercan be applied<br />

to single pulses (assuming <strong>the</strong>re is enough signal to noise to extract <strong>the</strong> interferogram<br />

accurately).<br />

Returning to <strong>the</strong> temporal phase filter, it is necessary to achieve an appropriate<br />

amountofspectralshear. In[260], <strong>the</strong>useofelectro-opticphasemodulator(EOPM)is<br />

c<strong>on</strong>sidered togive insufficientspectral shear. Therefore, amethodusingup-c<strong>on</strong>versi<strong>on</strong><br />

of <strong>the</strong>two replica pulses is described. A broadb<strong>and</strong>n<strong>on</strong>-linear material is required <strong>and</strong><br />

eachof<strong>the</strong>replicapulses isup-c<strong>on</strong>vertedthroughmixingwithaquasi-c<strong>on</strong>tinuouswave<br />

of different centre frequency. The up-c<strong>on</strong>verted pulses are thus centred at a different<br />

frequency <strong>and</strong> <strong>the</strong> required spectral shift is achieved.<br />

In [261], up-c<strong>on</strong>versi<strong>on</strong> is again used, but <strong>the</strong> two replica pulses are temporally<br />

displaced <strong>and</strong> mixed with a str<strong>on</strong>gly chirped pulse in <strong>the</strong> n<strong>on</strong>-linear material. The<br />

temporal delay results in mixing with a different frequency range of <strong>the</strong> chirped pulse<br />

<strong>and</strong> again <strong>the</strong> spectral shift is achieved. The fringe separati<strong>on</strong> in <strong>the</strong> interferogram is<br />

now inversely proporti<strong>on</strong>al to <strong>the</strong> temporal separati<strong>on</strong>. Note that <strong>the</strong> spectral phase<br />

filter (delay line) of <strong>the</strong> original c<strong>on</strong>cept is now linked to<strong>the</strong> role of <strong>the</strong> temporal phase<br />

filter <strong>and</strong> cannot thus be adjusted independently. Indeed, <strong>the</strong> temporal delay <strong>and</strong><br />

spectral shear are linked through <strong>the</strong> chirp of <strong>the</strong> pulse used to drive<strong>the</strong> upc<strong>on</strong>versi<strong>on</strong>.


166 9. Pulse length, profile <strong>and</strong> jitter<br />

The spectral <strong>and</strong> temporal shifts must be chosen to suit <strong>the</strong> length of <strong>the</strong> input<br />

pulse <strong>and</strong> <strong>the</strong> spectrometer resoluti<strong>on</strong>, <strong>and</strong> this places c<strong>on</strong>straints <strong>on</strong> <strong>the</strong> length of<br />

pulses that can be measured. The most important factor is to ensure that <strong>the</strong> delay<br />

is small enough to ensure <strong>the</strong> fringes can be resolved whilst not so small that <strong>the</strong> inversi<strong>on</strong><br />

process cannot unambiguously determine <strong>the</strong> phases. It is thus apparent that<br />

extending <strong>the</strong> Spider technique to shorter (XUV, SXR) wavelengths is dependent <strong>on</strong><br />

a satisfactory method of achieving <strong>the</strong> required spectral shear. (We will assume that<br />

splitting <strong>the</strong> <strong>beam</strong> by means of a knife-edged mirror will prove to be satisfactory since<br />

true amplitude divisi<strong>on</strong> is likely to be impossible). The use of n<strong>on</strong>-linear materials<br />

is prevented by str<strong>on</strong>g absorpti<strong>on</strong>. Ano<strong>the</strong>r approach that has been suggested for<br />

<strong>the</strong> XUV is to use two-colour, two-phot<strong>on</strong> atomic i<strong>on</strong>isati<strong>on</strong> to transfer <strong>the</strong> frequency<br />

spectrum of <strong>the</strong> phot<strong>on</strong> to <strong>the</strong> photoelectr<strong>on</strong> energy spectrum[262]. This however<br />

brings additi<strong>on</strong>al problems since an electr<strong>on</strong> spectrometer is now needed to resolve<br />

<strong>the</strong> photoelectr<strong>on</strong> spectrum, which c<strong>on</strong>tains <strong>the</strong> interferogram. Matching <strong>the</strong> fringe<br />

spacing <strong>and</strong> interferogram b<strong>and</strong>width to <strong>the</strong> performance of available electr<strong>on</strong> spectrometers<br />

will place limits <strong>on</strong> <strong>the</strong> range of pulses that can be analysed. For example,<br />

an electr<strong>on</strong> spectrometer with a challenging high resoluti<strong>on</strong> of 1 meV can <strong>on</strong>ly probe<br />

a temporal range of 660 fs from <strong>the</strong> uncertainty relati<strong>on</strong>.<br />

In [263], Smirnova et al. describe an approach for measuring fast processes with<br />

phot<strong>on</strong> pulses that are l<strong>on</strong>g relative to <strong>the</strong> process. The approach is to spectrally<br />

shear, by dressing with a weak IR field, a correlated two-electr<strong>on</strong> spectrum of photoi<strong>on</strong>isati<strong>on</strong><br />

<strong>and</strong> Auger electr<strong>on</strong>s created when an XUV pulse interacts with <strong>the</strong> sample.<br />

The spectrally shifted electr<strong>on</strong> spectrum interferes with <strong>the</strong> original spectrum <strong>and</strong> <strong>the</strong><br />

phase is mapped to an amplitude modulati<strong>on</strong> in <strong>the</strong> spectral intensity <strong>and</strong> thus can be<br />

measured. The technique is thus Spider with electr<strong>on</strong>s. The relevance here is that<br />

<strong>the</strong>re is no reas<strong>on</strong> in principal why <strong>the</strong> same approach cannot be used to measure<br />

<strong>the</strong> field of <strong>the</strong> X-ray pulse since <strong>the</strong> spectral c<strong>on</strong>tent of <strong>the</strong> pulse will be encoded<br />

into <strong>the</strong> electr<strong>on</strong> spectrum. The key requirement for <strong>the</strong> technique is that very tight<br />

synchr<strong>on</strong>isati<strong>on</strong> (sub-fs) between <strong>the</strong> XUV pulse profile <strong>and</strong> <strong>the</strong> phase of <strong>the</strong> IR pulse<br />

is required. The technique will also be flux intensive since <strong>the</strong> correlated two-electr<strong>on</strong><br />

spectrum must be measured, but given sufficient phot<strong>on</strong>s per pulse a single shot<br />

measurement can be made.<br />

In summary, <strong>the</strong> extensi<strong>on</strong> of optical techniques like Frog <strong>and</strong> Spiderto <strong>the</strong> XUV<br />

<strong>and</strong> shorter wavelengths looks challenging. Progress in this area has no doubt been<br />

limited by <strong>the</strong> lack of sources giving ultra-short pulses in this spectral range, <strong>and</strong> <strong>on</strong>e<br />

might <strong>the</strong>refore reas<strong>on</strong>ably expectmore progress in<strong>the</strong>future. The advantage of<strong>the</strong>se<br />

techniques is that <strong>the</strong>y give <strong>the</strong> exact pulse profile by calculating <strong>the</strong> electric field <strong>and</strong><br />

thus provide complete informati<strong>on</strong>. But <strong>the</strong> techniques are also invasive. The input<br />

pulse must be divided <strong>and</strong> manipulated <strong>and</strong> as such <strong>the</strong>y are not <strong>the</strong> obvious choice<br />

for a pulse-by pulse diagnostic even when single pulse characterisati<strong>on</strong> is possible.<br />

9.6 Reflectivity modulati<strong>on</strong><br />

Maltezopoulos et al. [264] describe how <strong>the</strong> free electr<strong>on</strong> laser <strong>beam</strong> incident at a<br />

glancing angle <strong>on</strong> <strong>the</strong> surface of a GaAs substrate modifies <strong>the</strong> reflectivity of <strong>the</strong><br />

GaAs to visible light in proporti<strong>on</strong> to <strong>the</strong> intensity of <strong>the</strong> X-rays. Thus, a visible laser<br />

(frequency doubled from near-infrared 800 nm) is used to simultaneously illuminate<br />

<strong>the</strong> area of <strong>the</strong> GaAs illuminated by <strong>the</strong> X-ray pulse <strong>and</strong> <strong>the</strong> reflected visible light


9.7. Streak cameras 167<br />

imaged <strong>on</strong>to a CCD. The intensity distributi<strong>on</strong> <strong>and</strong> positi<strong>on</strong> of <strong>the</strong> visible image give<br />

<strong>the</strong> free electr<strong>on</strong> laser pulse profile <strong>and</strong> timing.<br />

The technique is inherently pulse-by-pulse, <strong>and</strong> does not use co-propagati<strong>on</strong>, but<br />

is not transparent. Also, <strong>the</strong> visible light must overlap <strong>the</strong> entire area of illuminati<strong>on</strong><br />

from <strong>the</strong> free electr<strong>on</strong> laser <strong>and</strong> so some sort of focus of <strong>the</strong> X-rays may be required.<br />

The technique is thus not an <strong>on</strong>-line diagnostic but could be used for checking <strong>the</strong><br />

<strong>beam</strong> at <strong>the</strong> experimental end-stati<strong>on</strong> (where <strong>the</strong>re is likely to be a focus anyway).<br />

Alternatively, since this technique does not require such high <strong>beam</strong> intensities as <strong>the</strong><br />

cross-correlati<strong>on</strong> methods described in Secti<strong>on</strong> 9.3, it may be possible to split off a<br />

small part of <strong>the</strong> FEL <strong>beam</strong> <strong>and</strong> use that to m<strong>on</strong>itor <strong>the</strong> arrival time of <strong>the</strong> pulse. In<br />

any case, <strong>the</strong> temporal resoluti<strong>on</strong> is limited by <strong>the</strong> length of <strong>the</strong> visible pulse <strong>and</strong> <strong>the</strong><br />

space-to-time correlati<strong>on</strong> of <strong>the</strong> visible imaging system. The work presented gives a<br />

resoluti<strong>on</strong> of about 40 fs rms.<br />

Gahl et al. [244] suggest that <strong>the</strong> key limitati<strong>on</strong> to <strong>the</strong> temporal resoluti<strong>on</strong> is<br />

<strong>the</strong> visible probe pulse length, of <strong>the</strong> order 120 <strong>–</strong> 150 fs in this example. They<br />

also suggest that <strong>the</strong> time required for <strong>the</strong> GaAs surface to recover to its original<br />

level of reflectivity could be of <strong>the</strong> order of hundreds of picosec<strong>on</strong>ds, but this would<br />

<strong>on</strong>ly be a limit in multi-shot measurements at repetiti<strong>on</strong> rates in <strong>the</strong> GHz regi<strong>on</strong>.<br />

A fur<strong>the</strong>r c<strong>on</strong>siderati<strong>on</strong> is <strong>the</strong> potential for radiati<strong>on</strong> damage of <strong>the</strong> GaAs surface.<br />

Maltezopoulos et al. [264] used an optimised <strong>beam</strong> fluence of 13 ± 5 mJ/cm 2 rms<br />

for <strong>the</strong>ir measurements, commenting that permanent damage was observed at an<br />

unstated higher fluence, thus requiring <strong>the</strong> GaAs surface to be renewed. C<strong>on</strong>versely,<br />

measurements at a lower average fluence led to very weak c<strong>on</strong>trast in <strong>the</strong> images,<br />

making temporal determinati<strong>on</strong> unreliable.<br />

A related technique has been recently proposed by Krejcik [265]. In his scheme,<br />

<strong>the</strong> X-rays strike a magnetised film <strong>and</strong> cause a change in <strong>the</strong> magnetisati<strong>on</strong> which<br />

is probed using MOKE (Magneto- Optical Kerr Effect). An IR laser illuminates<br />

<strong>the</strong> magnetic film <strong>and</strong> undergoes a small relative polarisati<strong>on</strong> rotati<strong>on</strong> (∼ 1 ◦ ) in <strong>the</strong><br />

part that is reflected from <strong>the</strong> area <strong>the</strong> X-rays illuminated. The polarisati<strong>on</strong> of <strong>the</strong><br />

reflected IR <strong>beam</strong> is analysed <strong>and</strong> mapped over <strong>the</strong> <strong>beam</strong> profile. Thus, a time to<br />

space mapping of <strong>the</strong> arrival time of <strong>the</strong> X-ray pulse is achieved. The advantage of<br />

<strong>the</strong> MOKE analysis is that <strong>the</strong> magnetisati<strong>on</strong> change is extremely rapid <strong>and</strong> so <strong>the</strong>re<br />

is no issue with <strong>the</strong> resp<strong>on</strong>se time of <strong>the</strong> process reducing <strong>the</strong> temporal resoluti<strong>on</strong><br />

achieved. The disadvantage is that <strong>the</strong> magnetic material needs a res<strong>on</strong>ance at a<br />

magnetically active orbital at <strong>the</strong> X-ray wavelength of interest. Thus <strong>the</strong> technique<br />

is not applicable to arbitrary wavelengths.<br />

9.7 Streak cameras<br />

Streak cameras work <strong>on</strong> similar principles to oscilloscopes <strong>and</strong> cathode-ray tubes <strong>and</strong><br />

work by mapping time to a spatial coordinate in <strong>the</strong> detector. The incident phot<strong>on</strong><br />

<strong>beam</strong> is focused <strong>on</strong>to a slit <strong>and</strong> <strong>the</strong>n passes through a photocathode. The photoemitted<br />

electr<strong>on</strong>s are drawn between two parallel plates that are also parallel to <strong>the</strong><br />

slit. A high-speed sweep voltage, synchr<strong>on</strong>ised to <strong>the</strong> pulse arrival, is applied across<br />

<strong>the</strong> plates. This gives an angular deflecti<strong>on</strong> of <strong>the</strong> electr<strong>on</strong>s that is directly related to<br />

<strong>the</strong>ir time of arrival <strong>and</strong> thus to <strong>the</strong> durati<strong>on</strong> of <strong>the</strong> pulse. The deflected electr<strong>on</strong>s are<br />

<strong>the</strong>n multiplied with a micro-channel plate (MCP) before impacting <strong>on</strong> a phosphor<br />

screen. The streak pattern <strong>on</strong> <strong>the</strong> screen is imaged by a CCD or o<strong>the</strong>r suitable


168 9. Pulse length, profile <strong>and</strong> jitter<br />

detector, using a light intensifier tube if required. The advantages of streak cameras<br />

are <strong>the</strong> applicability across a wide wavelength range (X-rays to near-IR, although not<br />

with a single camera), single-shot time resoluti<strong>on</strong> in <strong>the</strong> picosec<strong>on</strong>d range for off-<strong>the</strong>shelf<br />

systems, <strong>and</strong> <strong>the</strong> ability to work at high repetiti<strong>on</strong> rate, albeit with reduced<br />

temporal resoluti<strong>on</strong>. They can also provide informati<strong>on</strong> <strong>on</strong> intensity <strong>and</strong> positi<strong>on</strong> in<br />

additi<strong>on</strong> to <strong>the</strong> temporal informati<strong>on</strong> in a single measurement.<br />

Historically, <strong>the</strong> resoluti<strong>on</strong> of streak cameras has been restricted to several picosec<strong>on</strong>ds<br />

in single-shot mode. In multiple-shot averaging systems, <strong>the</strong> use of a synchr<strong>on</strong>ised<br />

sine-wave sweep voltage adversely affects temporal resoluti<strong>on</strong> by making <strong>the</strong>se<br />

systems highly susceptible to source jitter. However, <strong>the</strong>re have been recent improvements<br />

in off <strong>the</strong> shelf systems <strong>and</strong> <strong>the</strong> fastest commercially available streak cameras<br />

claim a time resoluti<strong>on</strong> of


9.8. Summary 169<br />

ties, we ideally need to be able to measure each <strong>and</strong> every pulse that is used in <strong>the</strong><br />

experiment.<br />

The temporal diagnostics techniques thus need to be automatic, reliable, n<strong>on</strong>invasive<br />

<strong>and</strong> capable of h<strong>and</strong>ling large amounts of data. And yet a key point to note<br />

is that many of <strong>the</strong> techniques are currently at <strong>the</strong> level of experiments in <strong>the</strong>mselves.<br />

This is not unexpected since sources of ultra-short pulses at VUV <strong>and</strong> shorter<br />

wavelengths are a relatively new phenomen<strong>on</strong>, <strong>and</strong> it is <strong>on</strong>ly through availability of<br />

such sources that <strong>the</strong> diagnostic techniques can be developed. In <strong>the</strong> early days of<br />

ultra-short optical wavelength pulses, <strong>the</strong> principal diagnostic was <strong>the</strong> relatively uninformative<br />

autocorrelator, <strong>and</strong> yet sophisticated diagnostics that fully characterize<br />

<strong>the</strong> pulse profile can now be bought ”off <strong>the</strong> shelf”.<br />

We should not however be complacent that developments to shorter wavelengths<br />

will rapidly follow <strong>the</strong> availability of <strong>the</strong> sources. Certainly, basic sec<strong>on</strong>d-order autocorrelati<strong>on</strong><br />

as been successfully dem<strong>on</strong>strated into <strong>the</strong> soft X-ray, but this gives<br />

<strong>on</strong>ly informati<strong>on</strong> about <strong>the</strong> pulse length, <strong>and</strong> even <strong>the</strong>n assumpti<strong>on</strong>s have to be made<br />

about <strong>the</strong> pulse profile.<br />

Atoptical wavelengths, autocorrelati<strong>on</strong> was extendedthrough <strong>the</strong> additi<strong>on</strong> of spectral<br />

analysis (e.g. FROG) <strong>and</strong> this allowed complete pulse profile retrieval. But a key<br />

requirement in autocorrelati<strong>on</strong> techniques is a n<strong>on</strong>-linear process that can mix <strong>the</strong><br />

two pulses <strong>and</strong> produce a signal that is proporti<strong>on</strong>al to <strong>the</strong> autocorrelati<strong>on</strong> functi<strong>on</strong>.<br />

Here we are limited not <strong>on</strong>ly by <strong>the</strong> availability of suitable physical processes but<br />

also by detectors since <strong>the</strong> n<strong>on</strong>-linear signal may be accompanied by a str<strong>on</strong>g background<br />

that is hard to discriminate from <strong>the</strong> signal of interest. This is particularly<br />

true when <strong>the</strong> physical process produces electr<strong>on</strong>s ra<strong>the</strong>r than light, as will generally<br />

be <strong>the</strong> case at short wavelengths where i<strong>on</strong>izati<strong>on</strong> is a dominant process. Two-phot<strong>on</strong><br />

i<strong>on</strong>izati<strong>on</strong> is <strong>the</strong> most widely cited physical process that might fulfill <strong>the</strong> requirements<br />

<strong>and</strong> auto-correlati<strong>on</strong> measurements have been successfully performed into <strong>the</strong> XUV.<br />

The wavelength range over which two-phot<strong>on</strong> i<strong>on</strong>isati<strong>on</strong> will work is limited by<br />

<strong>the</strong> i<strong>on</strong>isati<strong>on</strong> potentials of available gases <strong>and</strong> so different gases are needed to cover<br />

different wavelength ranges. Fur<strong>the</strong>rmore, <strong>on</strong>ce <strong>the</strong> energy of a single phot<strong>on</strong> is above<br />

<strong>the</strong> gas IP, <strong>the</strong>n <strong>the</strong> large background of single-phot<strong>on</strong> i<strong>on</strong>isati<strong>on</strong> events is likely to<br />

swamp <strong>the</strong> two-phot<strong>on</strong> signal. Thus two-phot<strong>on</strong> i<strong>on</strong>isati<strong>on</strong> is <strong>on</strong>ly viable at phot<strong>on</strong><br />

energies between half of <strong>and</strong> <strong>the</strong> full IP of <strong>the</strong> gas. Two-phot<strong>on</strong> double i<strong>on</strong>isati<strong>on</strong><br />

may <strong>the</strong>n be <strong>the</strong> way forward into <strong>the</strong> soft X-ray, but at <strong>the</strong> moment such schemes<br />

are untested.<br />

A fundamental limitati<strong>on</strong> of autocorrelati<strong>on</strong> techniques is that <strong>the</strong>y cannot provide<br />

informati<strong>on</strong> about <strong>the</strong> timing of <strong>the</strong> pulse since it is measured against itself. Thus,<br />

<strong>the</strong>re has been a lot of development in crosscorrelati<strong>on</strong> techniques in which <strong>the</strong> free<br />

electr<strong>on</strong> laser pulse is measured against a known pulse from an IR laser. As well<br />

as pulse length informati<strong>on</strong>, we can gain informati<strong>on</strong> <strong>on</strong> timing jitter relative to <strong>the</strong><br />

laser, which is often very c<strong>on</strong>venient, for example when <strong>the</strong> laser is also used in pumpprobe<br />

experiments. A significant amount of work has been undertaken in this area<br />

<strong>and</strong> many possible approaches to cross-correlati<strong>on</strong> have been dem<strong>on</strong>strated or at least<br />

suggested.<br />

A well-established approach is to measure <strong>the</strong> intensity <strong>and</strong>/or number of sideb<strong>and</strong>s<br />

that appear <strong>on</strong> <strong>the</strong> photo-i<strong>on</strong>isati<strong>on</strong> spectrum of a gas as <strong>the</strong> infrared laser falls<br />

in <strong>and</strong> out of temporal overlap with <strong>the</strong> free electr<strong>on</strong> laser pulse. This requires <strong>the</strong><br />

infrared laser <strong>and</strong> X-ray pulse to be spatially overlapped in <strong>the</strong> i<strong>on</strong>isati<strong>on</strong> chamber,<br />

<strong>and</strong> thus has some implicati<strong>on</strong>s for <strong>the</strong> optical layout of <strong>the</strong> <strong>beam</strong>line. Never<strong>the</strong>less,


170 9. Pulse length, profile <strong>and</strong> jitter<br />

<strong>the</strong> gas absorbs so little that <strong>the</strong> X-ray pulse is unaltered <strong>and</strong> <strong>the</strong> diagnostic meets our<br />

requirement for being n<strong>on</strong>-invasive. In <strong>the</strong> first implementati<strong>on</strong>s of this sort of crosscorrelator,<br />

complete pulse characterisati<strong>on</strong> required measurement over a successi<strong>on</strong><br />

of free electr<strong>on</strong> laser pulses with differing infrared pulse delays. The pulse profile<br />

measure was thus an average profile <strong>and</strong> not shot-by-shot. This limitati<strong>on</strong> can be<br />

overcome byc<strong>on</strong>verting<strong>the</strong> temporal informati<strong>on</strong> tospatial informati<strong>on</strong> <strong>and</strong> recording<br />

<strong>the</strong> photo-electr<strong>on</strong> spectrum with an imaging detector. Proof of principle experiments<br />

have been d<strong>on</strong>e but more development is needed to improve sensitivity <strong>and</strong> temporal<br />

resoluti<strong>on</strong>.<br />

There are o<strong>the</strong>r approaches to cross-correlati<strong>on</strong> that have been proposed, using<br />

for example Auger electr<strong>on</strong>s <strong>and</strong> chirped infrared pulses. This is an active area of<br />

research at <strong>the</strong> moment. The comm<strong>on</strong> <strong>the</strong>me is detecti<strong>on</strong> through <strong>the</strong> i<strong>on</strong>isati<strong>on</strong> of<br />

gases, thus all <strong>the</strong>se schemes rely <strong>on</strong> electr<strong>on</strong> detectors, which adds significantly to<br />

<strong>the</strong> complexity of <strong>the</strong> measurement.<br />

A different approach to cross-correlati<strong>on</strong> is that of reflectivity modulati<strong>on</strong>. Although<br />

this is an invasive measurement, it is more sensitive than gas based measurements<br />

<strong>and</strong> so not all <strong>the</strong> <strong>beam</strong> is needed. Thus, a small part of <strong>the</strong> <strong>beam</strong> can be<br />

split off <strong>and</strong> sent to <strong>the</strong> cross-correlator whilst <strong>the</strong> rest of <strong>the</strong> <strong>beam</strong> is passed to <strong>the</strong><br />

experiment. Since <strong>the</strong> technique is relatively simple <strong>and</strong> is inherently pulse resolved<br />

(but limited to <strong>the</strong> external laser repetiti<strong>on</strong> rate), it is quite attractive. Some work is<br />

needed to improve <strong>the</strong> temporal resoluti<strong>on</strong> however.<br />

Streak cameras are a l<strong>on</strong>g established instrument for measuring short pulses. However,<br />

current instruments are too limited in terms of temporal resoluti<strong>on</strong> <strong>and</strong> repetiti<strong>on</strong><br />

rate. There are a number of active development programs aimed at addressing<br />

<strong>the</strong>se points <strong>and</strong> it seems likely that temporal resoluti<strong>on</strong> of 100 fs will be possible<br />

whilst <strong>the</strong> use of ’smart’ detectors will give operati<strong>on</strong> to several kHz.<br />

Electro-optical techniques are not useful for directly measuring <strong>the</strong> profile of <strong>the</strong><br />

free electr<strong>on</strong> laser pulses since <strong>the</strong>y <strong>on</strong>ly functi<strong>on</strong> at THz frequencies. But <strong>the</strong>y can<br />

be usefully employed for m<strong>on</strong>itoring <strong>the</strong> timing of <strong>the</strong> electr<strong>on</strong> bunch relative to an<br />

external laser. This is likely to give important informati<strong>on</strong> <strong>on</strong> <strong>the</strong> overall timing of<br />

<strong>the</strong> X-ray pulses.


9.8. Summary 171<br />

Summary<br />

• Pulse length <strong>and</strong>jitter can bediagnosed with cross-correlati<strong>on</strong><br />

techniques. They rely <strong>on</strong> encoding <strong>the</strong> electr<strong>on</strong>s emanating<br />

from a photoi<strong>on</strong>izati<strong>on</strong> of a gas spectrally with energy from<br />

a known laser pulse. The main photoi<strong>on</strong>izati<strong>on</strong> peaks will<br />

<strong>the</strong>n get side-b<strong>and</strong>s (also called satellites) whose intensity is<br />

proporti<strong>on</strong>al to <strong>the</strong> delay between <strong>the</strong> laser phot<strong>on</strong> pulse <strong>and</strong><br />

that of <strong>the</strong> free electr<strong>on</strong> laser. With a chirped laser pulse (<strong>the</strong><br />

phot<strong>on</strong>-energy varies over <strong>the</strong> pulse) <strong>the</strong> pulse length may also<br />

be extracted from such a measurement.<br />

• Autocorrelati<strong>on</strong> techniques uses <strong>the</strong> pulse itself, <strong>the</strong> pulse<br />

length is deduced from splitting <strong>the</strong> pulse <strong>and</strong> recombining<br />

it after <strong>the</strong> two parts of <strong>the</strong> pulse have traversed different<br />

optical paths. The correlati<strong>on</strong> between different parts of <strong>the</strong><br />

pulse can <strong>the</strong>n be measured. Although <strong>the</strong> technique is single<br />

shot a pulse shape has to be assumed to extract <strong>the</strong> length of<br />

<strong>the</strong> pulse. More advanced autocorrelati<strong>on</strong> techniques Frog<br />

<strong>and</strong> Spider can also deduce <strong>the</strong> pulse shape.<br />

• Reflectivity modulati<strong>on</strong> of a GaAs surface induced by an optical<br />

laser can be used to extract <strong>the</strong> pulse length since <strong>the</strong><br />

free electr<strong>on</strong> laser <strong>beam</strong> modifies <strong>the</strong> reflectivity of <strong>the</strong> surface.<br />

The positi<strong>on</strong> <strong>and</strong> intensity of <strong>the</strong> visible reflected light<br />

can <strong>the</strong>n be recorded <strong>on</strong> a pulse-by-pulse basis but not in a<br />

manner transparent to <strong>the</strong> experiments. The resoluti<strong>on</strong> limit<br />

is quoted to <strong>the</strong> 40 fs rms currently. Modulati<strong>on</strong> of o<strong>the</strong>r processes<br />

have also been suggested, such as <strong>the</strong> Magneto-optical<br />

Kerr effect.<br />

• Streak cameras work <strong>on</strong> similar principles to oscilloscopes <strong>and</strong><br />

cathode-ray tubes <strong>and</strong> work by mapping time to a spatial coordinate<br />

in <strong>the</strong> detector. The incident phot<strong>on</strong> <strong>beam</strong> is focused<br />

<strong>on</strong>to a slit <strong>and</strong> <strong>the</strong>n passes through a photocathode.<br />

The photo-emitted electr<strong>on</strong>s are drawn between two parallel<br />

plates that are also parallel to <strong>the</strong> slit. A high-speed sweep<br />

voltage, synchr<strong>on</strong>ised to <strong>the</strong> pulse arrival, is applied across<br />

<strong>the</strong> plates. This gives an angular deflecti<strong>on</strong> of <strong>the</strong> electr<strong>on</strong>s<br />

that is directly related to <strong>the</strong>ir time of arrival <strong>and</strong> thus to<br />

<strong>the</strong> durati<strong>on</strong> of <strong>the</strong> pulse. The deflected electr<strong>on</strong>s are <strong>the</strong>n<br />

multiplied with a micro-channel plate before impacting <strong>on</strong> a<br />

phosphor screen. Current time-resoluti<strong>on</strong> for streak cameras<br />

are about 300 fs.


10. Free electr<strong>on</strong> laser experiments<br />

Written by: A. Lindblad<br />

A free electr<strong>on</strong> laser provide a tunable pulsed (transversely) coherent phot<strong>on</strong> <strong>beam</strong><br />

with unprecedented brilliance. The <strong>beam</strong> can thus be focussed to small spots where<br />

a very high X-ray phot<strong>on</strong> density can be acquired.<br />

The properties outlined, enable a number of experiments that are, more or less,<br />

unique. The large number of phot<strong>on</strong>s per pulse (often tens of billi<strong>on</strong>s or more) allow<br />

time-resolved imaging <strong>and</strong> spectroscopies <strong>–</strong> where combinati<strong>on</strong>s of <strong>the</strong> free electr<strong>on</strong><br />

laser <strong>beam</strong> <strong>and</strong> laser <strong>beam</strong>s make pump <strong>and</strong> probe experiments more advanced. The<br />

high degree of coherence make single-shot imaging of nano-structures possible. The<br />

high X-ray phot<strong>on</strong> density enable <strong>the</strong> study of n<strong>on</strong>-linear processes in <strong>the</strong> X-ray<br />

regime, as well as <strong>the</strong> ability to study processes with low cross-secti<strong>on</strong>s or with low<br />

target density (e.g. imaging of single unsupported nanostructures/molecules).<br />

In this chapter some of <strong>the</strong> experiments carried out at free electr<strong>on</strong> laser will be<br />

highlighted <strong>–</strong> our treatment here will not by any means be all encompassing but is<br />

ra<strong>the</strong>r intended to give an introducti<strong>on</strong> to which kinds of experiments successfully<br />

exploits <strong>the</strong> unique properties of this kind of X-ray source.<br />

One may c<strong>on</strong>dense <strong>the</strong> experimental efforts into: imaging <strong>and</strong> measuring atoms<br />

<strong>and</strong> atomic processes <strong>on</strong> <strong>the</strong> nanometer <strong>and</strong> femtosec<strong>on</strong>d scales. Adding that <strong>the</strong><br />

development is towards ˚Angström <strong>and</strong> attosec<strong>on</strong>d time-scales.<br />

10.1 The ”holy grails” of free electr<strong>on</strong> laser experiments<br />

”Molecular movies”<br />

The short pulse length of free electr<strong>on</strong> laser sources makes it possible, ei<strong>the</strong>r via<br />

splitting or sub-pulse structure, to get a X-ray pump - X-ray probe experiment. With<br />

<strong>the</strong> additi<strong>on</strong> of external lasers X-ray pump, UV/Vis Laser-probe or UV/Vis-pump<br />

X-ray probe is made possible. All of this in <strong>the</strong> femtosec<strong>on</strong>d regime.<br />

This, potentially, makes it possible to follow a photo-excited process like a movie,<br />

but with a femtosec<strong>on</strong>d frame-rate.<br />

173


174 10. Free electr<strong>on</strong> laser experiments<br />

”Single-molecule/nanostructure imaging”<br />

Real-space imaging of nanostructures <strong>and</strong> ultimately single-molecules can be realized<br />

by <strong>the</strong> rec<strong>on</strong>structi<strong>on</strong> of <strong>the</strong> image via a recorded image in momentum space obtained<br />

from X-ray scattering from <strong>the</strong> object.<br />

”Single-shot spectroscopy/imaging”<br />

Single-shot experiments are made possible by <strong>the</strong> large number of phot<strong>on</strong>s available<br />

that can get focussed down to a very small point owing to <strong>the</strong> <strong>beam</strong> quality. Since <strong>the</strong><br />

pulses are short <strong>and</strong> intense <strong>the</strong> data for imaging <strong>and</strong> spectroscopy can be acquired<br />

before <strong>the</strong> sample explodes. This is also a cornerst<strong>on</strong>e for <strong>the</strong> success of time-resolved<br />

measurements as outlined above.<br />

In <strong>the</strong> following some examples from each field will be given. Both planned <strong>and</strong><br />

already performed experimentsare presentedas toshowwhat is d<strong>on</strong>etoday <strong>and</strong>where<br />

different groups <strong>and</strong> facilities intends to head in <strong>the</strong> near future.<br />

10.2 Time-resolved spectroscopies<br />

Core-level photo-electr<strong>on</strong> spectroscopies give informati<strong>on</strong> about <strong>the</strong> chemical state<br />

of <strong>the</strong> c<strong>on</strong>stituting atoms (being in free molecules or atoms, molecules or atoms <strong>on</strong><br />

surfaces or in solids). By using phot<strong>on</strong>-pulses time-resolved spectroscopies can be<br />

performed. Atsynchrotr<strong>on</strong>s<strong>and</strong>HHGlaser sources <strong>the</strong>numberofphot<strong>on</strong>sperpulse is<br />

verylowcomparedtothoseatafreeelectr<strong>on</strong>laser (atleast generally)whichiswhythis<br />

kind of experiments needs to be performed at X-ray free electr<strong>on</strong> laser facilites[271].<br />

Many fundamental properties of matter can be studied with this type of spectroscopy,<br />

e.g. magnetizati<strong>on</strong> dynamics, reflectivity changes <strong>and</strong> molecular dissociati<strong>on</strong><br />

dynamics.<br />

UV/Vis pump-X-ray probe spectroscopy<br />

With an ordinary laser synchr<strong>on</strong>ized (with a known time-delay) to <strong>the</strong> free electr<strong>on</strong><br />

laser X-rays it is possible to study <strong>the</strong> development of <strong>the</strong> core level photo-electr<strong>on</strong><br />

spectrum. At Flash, <strong>the</strong> Ge 3d photoemissi<strong>on</strong> from a n-doped Ge crystal have been<br />

studied in this manner[272].<br />

Time-resolved pump-probe experiments at <strong>the</strong> Lcls[273]: N2 molecule studied<br />

with 1.05 keV X-rays (10 11 phot<strong>on</strong>s/pulse) with a Wiley-McLaren i<strong>on</strong> time-of-flight<br />

spectrometer. The counting rate was about 3 Hz. Comparing to <strong>the</strong> 30 Hz repetiti<strong>on</strong><br />

rate of <strong>the</strong> X-ray source focussed <strong>on</strong> 3 µm 2 <strong>and</strong> 10 13 molecules/cm 3 .<br />

Timeresolvedcorelevelphoto-electr<strong>on</strong>spectroscopy(asdescribedalsoinRef.[274])<br />

studied of atoms <strong>and</strong> molecules (in gases, clusters, liquids <strong>and</strong> solids) combined with<br />

lasers will be an important investigative tool at free electr<strong>on</strong> laser to study how <strong>the</strong><br />

chemical state of an atom can change with time depending <strong>on</strong> how its neigbours are<br />

excited.<br />

Velocity map imaging have been used sucessfully for characterizati<strong>on</strong> of <strong>the</strong> overlap<br />

between free electr<strong>on</strong> laser X-rays <strong>and</strong> 800 nm optical laser <strong>–</strong> for instance using<br />

hydrogen[275] as <strong>the</strong> target.<br />

X-ray/X-ray Auto-correlati<strong>on</strong> spectroscopy at <strong>the</strong> X-Fel have been discussed by<br />

Grübel[276].


10.3. Imaging <strong>and</strong> Crystallography 175<br />

Nexafs<br />

Ce:YAG screen<br />

Sample<br />

Free electr<strong>on</strong> laser <strong>beam</strong><br />

Grating<br />

Figure 10.1: Possible set-up for a single shot Nexafs measurement (adapted from Ref. [278]).<br />

X-ray absorpti<strong>on</strong> near an i<strong>on</strong>izati<strong>on</strong> threshold (Nexafs) allow for <strong>the</strong> mapping of<br />

unoccupied states in a sample <strong>–</strong> <strong>the</strong> absorpti<strong>on</strong> intensity is recorded as a functi<strong>on</strong><br />

of phot<strong>on</strong>-energy. The absorpti<strong>on</strong> may be studied with any decay product from<br />

<strong>the</strong> excitati<strong>on</strong>, e.g. i<strong>on</strong>s, electr<strong>on</strong>s, phot<strong>on</strong>s. A possible way of doing Nexafs in a<br />

single-shot fashi<strong>on</strong> is depicted in Figure 10.1 <strong>–</strong> a grating acts as a <strong>beam</strong>splitter which<br />

c<strong>on</strong>tinuously disperse <strong>the</strong> X-ray pulse over a sample, thus <strong>the</strong> need for sweeping <strong>the</strong><br />

phot<strong>on</strong>-energy is eliminated <strong>and</strong> a single-shot measurement made possible[277, 278].<br />

10.3 Imaging <strong>and</strong> Crystallography<br />

The high fluence of free electr<strong>on</strong> laser <strong>beam</strong>s allow imaging of micro- <strong>and</strong> nanoparticles<br />

via X-ray scattering <strong>on</strong> a shot-to-shot basis. A problem that needs to be<br />

circumvented is <strong>the</strong> radiati<strong>on</strong> damage <strong>and</strong> subsequent explosi<strong>on</strong> of <strong>the</strong> particles up<strong>on</strong><br />

<strong>the</strong> multi-i<strong>on</strong>izati<strong>on</strong> <strong>–</strong> this problem becomes more <strong>and</strong> more substantial with decreasing<br />

particles size. For a 50 femtosec<strong>on</strong>d l<strong>on</strong>g pulse <strong>the</strong> resoluti<strong>on</strong> limit have been<br />

estimated to 0.2 nm because of <strong>the</strong> blurring caused by <strong>the</strong> Coulumb-explosi<strong>on</strong> of <strong>the</strong><br />

sample[279].<br />

A dem<strong>on</strong>strati<strong>on</strong> of single nano-sized particle X-ray diffractive imaging have been<br />

performedbyBogan<strong>and</strong>co-workers[280]whereanærodynamiclensprovidednanoparticles<br />

from an electrospray source. A part of <strong>the</strong>ir set-up is shown in Figure 10.2.<br />

There is clearly a strive towards imaging ever smaller objects[281] <strong>–</strong> <strong>and</strong> ultimately<br />

single molecules, both in free form <strong>and</strong> in <strong>the</strong>ir natural envir<strong>on</strong>ment, i.e. in water or


176 10. Free electr<strong>on</strong> laser experiments<br />

Figure 10.2: A prototypical single particle X-ray diffractive imaging (after Ref. [280]).<br />

in cells[282]. Recently a virus particle was imaged using overlays of recorded patterns<br />

from single particles <strong>–</strong> which shows that even if <strong>the</strong> object explodes because of <strong>the</strong><br />

deposited X-ray energy it survives l<strong>on</strong>g enough to be imaged[283].<br />

The imaging activities at Flash have recently been reviewed[284], as well as <strong>the</strong><br />

specific strives towards coherent diffractive imaging [50]. At <strong>the</strong> Lcls <strong>the</strong>re is an<br />

endstati<strong>on</strong> dedicated to coherent X-ray imaging[285]: <strong>the</strong> (CXI) instrument[286]. At<br />

Fermi@Elettra <strong>the</strong> first operati<strong>on</strong>al <strong>beam</strong>line is intended for coherent scattering experiments.<br />

As free electr<strong>on</strong> laser promise to deliver X-rays of sufficient quality to allow singleparticle/single-molecule<br />

imaging <strong>and</strong> that phase-retrieval algorithms become ever<br />

more sophisticated this area of research attracts a lot of effort. With phase-retrieval<br />

informati<strong>on</strong> (lost in <strong>the</strong> imaging process) this will ultimately allow for single-particle<br />

tomographic measurements of nano-particles, proteins <strong>and</strong> parts of cells.<br />

Figure 10.3 depicts X-ray scattering through a r<strong>and</strong>omly ordered sample (powder<br />

diffracti<strong>on</strong>). If <strong>the</strong> <strong>beam</strong> is incoherent <strong>and</strong> wide (Debeye-Scherrer scattering) <strong>the</strong><br />

scattering angle is proporti<strong>on</strong>al to <strong>the</strong> mean distance between <strong>the</strong> scattering centers in<br />

<strong>the</strong>samples. If<strong>the</strong><strong>beam</strong>issmall <strong>and</strong>sufficientlycoherent<strong>on</strong>estill obtainsinformati<strong>on</strong><br />

about <strong>the</strong> mean distances in <strong>the</strong> film via <strong>the</strong> scattering angle <strong>–</strong> but <strong>on</strong> <strong>the</strong> detector<br />

a speckle interference pattern occur instead of diffuse rings. The speckles angular<br />

extent is inversely proporti<strong>on</strong>al to <strong>the</strong> <strong>beam</strong> width.<br />

By overlaying <strong>the</strong> X-ray scattering images from two time-delayed pulses Gün<strong>the</strong>r<br />

<strong>and</strong> co-workers have recently dem<strong>on</strong>strated that sequential imaging with femtosec<strong>on</strong>d<br />

resoluti<strong>on</strong> is indeed possible to obtain for nanometer sized objects[287]. This is a<br />

significant step towards time-resolved imaging of objects at <strong>the</strong> atomic scale.<br />

10.4 N<strong>on</strong>-linear X-ray science<br />

Photoi<strong>on</strong>izati<strong>on</strong><br />

Withveryhigh irradiance levels (towards 10 16 W/cm 2 )at 93eV phot<strong>on</strong>energy, Xe 21+<br />

have been observed in i<strong>on</strong>-time of flight spectra <strong>–</strong> this corresp<strong>on</strong>ds to <strong>the</strong> absorpti<strong>on</strong>


10.4. N<strong>on</strong>-linear X-ray science 177<br />

a<br />

d<br />

∼ λ/a<br />

∼ λ/d<br />

Figure 10.3: X-ray diffracti<strong>on</strong> using diffuse <strong>and</strong> coherent <strong>beam</strong>s.<br />

of about 57 phot<strong>on</strong>s for that atom (or 5 keV absorbed X-ray energy)[288]. This<br />

amount of energy deposited in a single atom allow <strong>the</strong> study for many processes of<br />

fundamental nature. With shorter pulses in <strong>the</strong> attosec<strong>on</strong>d regime (<strong>the</strong> ”atomic unit<br />

of time” being 24 as) processes may even be possible to study <strong>on</strong> <strong>the</strong> same time-scale<br />

as <strong>the</strong> electr<strong>on</strong>’s travel-time around <strong>the</strong> nucleus.<br />

Sequential i<strong>on</strong>izati<strong>on</strong> of atomic arg<strong>on</strong> have been studied at <strong>the</strong> Scss free electr<strong>on</strong><br />

laser. They find evidence of a sequential electr<strong>on</strong> emissi<strong>on</strong> from <strong>the</strong> absorbing atoms<br />

during <strong>the</strong> pulse durati<strong>on</strong>[289], access to even shorter pulses would naturally benefit<br />

<strong>the</strong> study of this kind of physical phenomena.<br />

A summary of <strong>the</strong> findings <strong>on</strong> different rare gases is provided in Ref [98]. In this<br />

reference a survey of different possible experimental set-ups is also provided.<br />

Multiphot<strong>on</strong> i<strong>on</strong>izati<strong>on</strong> of atomic clusters have also been studied in this regard<br />

(see, e.g. Ref. [290]) since this allows for a detailed study of <strong>the</strong> Coulomb explosi<strong>on</strong><br />

from <strong>the</strong> interacti<strong>on</strong> of nano-particles with free electr<strong>on</strong> laser light.<br />

Recently Fang <strong>and</strong> colleagues have reported <strong>on</strong> an experiment where double corehole<br />

i<strong>on</strong>izati<strong>on</strong> in nitrogen molecules[291] is dem<strong>on</strong>strated. This type of experiment<br />

give insight in how fast phot<strong>on</strong>s are absorbed during <strong>the</strong> pulse since <strong>the</strong> single corei<strong>on</strong>izedspecies<br />

havealifetime in<strong>the</strong>lowfemtosec<strong>on</strong>dregime. Anexperimentinvolving<br />

X-ray emissi<strong>on</strong> following two-phot<strong>on</strong> absorpti<strong>on</strong> have also been suggested by Sun <strong>and</strong><br />

coworkers[292].


178 10. Free electr<strong>on</strong> laser experiments<br />

Diff. pumping<br />

Open multiplier<br />

TOF<br />

I<strong>on</strong>s<br />

±2 cm<br />

Figure 10.4: Focussed free electr<strong>on</strong> laser <strong>beam</strong> set-up for <strong>the</strong> study of multi-phot<strong>on</strong> i<strong>on</strong>izati<strong>on</strong> of gases (from<br />

Figure 8.11).<br />

Summary<br />

• Free electr<strong>on</strong> lasers enable <strong>the</strong> study of atoms <strong>and</strong> atomic<br />

processes, ei<strong>the</strong>r via imaging or spectroscopic measurements,<br />

<strong>on</strong> nanometer <strong>and</strong> femtosec<strong>on</strong>d scales.<br />

• Current source developments strive to refine <strong>the</strong> scales to<br />

atomic <strong>on</strong>es, i.e. ˚Angströms <strong>and</strong> attosec<strong>on</strong>ds.<br />

• Naturally an free electr<strong>on</strong> laser experimentshould exploit <strong>on</strong>e,<br />

or several of <strong>the</strong> source’s unique properties vis-à-vis:<br />

<strong>–</strong> Coherence<br />

<strong>–</strong> Brilliance<br />

<strong>–</strong> Fluence<br />

<strong>–</strong> Time structure<br />

• The holy grails of free electr<strong>on</strong> laser experiments are:<br />

<strong>–</strong> Single-shot imaging of single molecules<br />

<strong>–</strong> Single-shot spectroscopy of single molecules<br />

<strong>–</strong> Molecular-movies<br />

-<br />

+


11. Free Electr<strong>on</strong> Laser facilities<br />

Written by: A. Lindblad<br />

Inthischapter, free electr<strong>on</strong> laser facilites around<strong>the</strong>world will beinvestigated visà-vis<br />

<strong>the</strong> different technology choices made <strong>and</strong> how <strong>the</strong> performance of <strong>the</strong> facilites<br />

have been affected by those choices. Naturally such an overview will be somewhat<br />

limited <strong>and</strong> <strong>the</strong> focus lies <strong>on</strong> <strong>the</strong> currently operating facilites <strong>and</strong> <strong>the</strong> intended developments<br />

of <strong>the</strong>m <strong>–</strong> also, <strong>on</strong>ly free electr<strong>on</strong> lasers that lase in <strong>the</strong> VUV- <strong>and</strong> X-ray<br />

regimes are c<strong>on</strong>sidered here.<br />

Facility E εn λmin Rate Pol.<br />

Flash O SC 1.2 < 2 4.45 8·10 3<br />

No<br />

Lcls O NC 14 1 0.12 120 No<br />

XFEL C SC 17.5 1.4 0.1 27·10 3 Yes<br />

XFEL/SPring-8 C NC 8 0.8 0.1 60 No<br />

Fermi@Elettra C NC 1.7 1 4 50 Yes<br />

SwissFEL D NC 6 0.4 0.1 100 Yes<br />

Pal XFEL D NC 10 1 0.1 60 No<br />

Lcls-II D NC 14 1 0.6 120 Yes<br />

Flash-II D SC 1.2 1-1.5 4 10 No<br />

Table 11.1: The electr<strong>on</strong> energy is given in GeV <strong>and</strong> <strong>the</strong> repetiti<strong>on</strong> rate in Hz; <strong>the</strong> normalized emittance is<br />

given in µm; <strong>the</strong> minimum wavelength in nm. Informati<strong>on</strong> in <strong>the</strong> table is adapted from <strong>the</strong> review by McNiel<br />

<strong>and</strong> Thomps<strong>on</strong>[293].<br />

In Table 11.1 <strong>the</strong> facilites around <strong>the</strong> world currently operati<strong>on</strong>al (O), under c<strong>on</strong>structi<strong>on</strong><br />

(C) <strong>and</strong> in advanced technical planning <strong>and</strong> design phases (D) are listed<br />

toge<strong>the</strong>r with <strong>the</strong> facilities’ choice <strong>on</strong> normally c<strong>on</strong>ducting technology (NC) or superc<strong>on</strong>ducting<br />

technology (SC) <strong>and</strong> some o<strong>the</strong>r parameters.<br />

The locati<strong>on</strong>s of <strong>the</strong> facilites around <strong>the</strong> world are as follows: <strong>the</strong> Flash <strong>and</strong><br />

<strong>the</strong> European X-Fel[89] are located in Hamburg, Germany; <strong>the</strong> Lcls free electr<strong>on</strong><br />

laser uses <strong>the</strong> SLAC linear accelerator, which can be found in Stanford, USA[294];<br />

<strong>the</strong> XFEL/SPring-8 is located in <strong>the</strong> Harima area in Japan[295]; Fermi@Elettra<br />

is in Trieste, Italy; <strong>the</strong> SwissFEL[49] can be found in Switerl<strong>and</strong> <strong>and</strong> <strong>the</strong> Pal XFEL<br />

in Korea[296].<br />

179


180 11. Free Electr<strong>on</strong> Laser facilities<br />

Lcls-II <strong>and</strong> Flash-II[297] are extensi<strong>on</strong>s of those currently operating facilities.<br />

11.1 Operating facilities<br />

11.2 Flash<br />

Flash is situated in Hamburg, Germany (www.flash.de). The facility deliver ultrashort<br />

femtosec<strong>on</strong>d coherent radiati<strong>on</strong> in <strong>the</strong> wavelength range between 47 <strong>and</strong> 4.12<br />

nm. The shorter wavelength lies inside <strong>the</strong>, so called, water window where water is<br />

transparent to X-rays.<br />

Since 2005, Flash is a user facility serving a large variety of experiments. Typical<br />

user operati<strong>on</strong> parameters during <strong>the</strong> 2 nd user period from Nov 26, 2007 to Aug 16,<br />

2009:<br />

Wavelength range (fundamental) 4.12 <strong>–</strong> 47 nm<br />

Average single pulse energy 10 <strong>–</strong> 100 µJ<br />

Pulse durati<strong>on</strong> (FWHM) 10<strong>–</strong>400 fs<br />

Peak power (from av.) 1 <strong>–</strong> 5 GW<br />

Average power (at 500 pulses/s) ∼ 15 mW<br />

Spectral width (FWHM) ∼ 1 %<br />

Peak Brilliance 10 29 −10 30<br />

Flash is a high-gain Sase free electr<strong>on</strong> laser. The lasing process is initiated by<br />

<strong>the</strong> sp<strong>on</strong>taneous undulator radiati<strong>on</strong>, <strong>and</strong> <strong>the</strong> free electr<strong>on</strong> laser works <strong>the</strong>n in <strong>the</strong> socalled<br />

Self-Amplified Sp<strong>on</strong>taneous Emissi<strong>on</strong> (Sase)mode without needingan external<br />

input signal.The electr<strong>on</strong> bunches are produced in a laser-driven photoinjector <strong>and</strong><br />

accelerated by a superc<strong>on</strong>ducting linear accelerator. The RF-gun based photoinjector<br />

allows <strong>the</strong> generati<strong>on</strong> of electr<strong>on</strong> bunches with tiny emittances - m<strong>and</strong>atory for an<br />

efficient SASE process.The superc<strong>on</strong>ductingtechniques allows to accelerate thous<strong>and</strong>s<br />

of bunches per sec<strong>on</strong>d, which is not possible with o<strong>the</strong>r technologies.<br />

At intermediate energies of 130 <strong>and</strong> 470 MeV <strong>the</strong> electr<strong>on</strong> bunches are l<strong>on</strong>gitudinally<br />

compressed, <strong>the</strong>reby increasing <strong>the</strong> peak current from initially 50-80 A to 1-2<br />

kA - as required for <strong>the</strong> lasing process in <strong>the</strong> undulator. The 30 m l<strong>on</strong>g undulator<br />

c<strong>on</strong>sists of permanent NdFeB magnets with a fixed gap of 12 mm, a period length of<br />

27.3 mm <strong>and</strong> peak magnetic field of 0.47 T. The electr<strong>on</strong>s interact with <strong>the</strong> undulator<br />

field in such a way, that so called micro bunches are developed. These micro bunches<br />

radiate coherently <strong>and</strong> produce intense X-ray pulses. Finally, a dipole magnet deflects<br />

<strong>the</strong> electr<strong>on</strong> <strong>beam</strong> safely into a dump, while <strong>the</strong> X-ray radiati<strong>on</strong> propagates to <strong>the</strong><br />

experimental hall.<br />

e - -gun<br />

Accelerator<br />

Figure 11.1: Schematic of <strong>the</strong> Flash free electr<strong>on</strong> laser facility.<br />

Sase undulator


11.2. Flash 181<br />

Injector <strong>and</strong> accelerator<br />

Cs2Te photocathode inside a 1.3 GHz normally-c<strong>on</strong>ducting radiofrequency cavity. 0.5<br />

to 1 nC per pulse at 10 Hz. L-b<strong>and</strong> cavity (1.3 GHz). The pulse trains generated<br />

can be up to 800 microsec<strong>on</strong>ds l<strong>on</strong>g <strong>and</strong> <strong>the</strong> pulse spacing is usually 1 mikrosec<strong>on</strong>d<br />

(1 MHz). The peak current of <strong>the</strong> uncompresssed bunch is around 70 A. The bunch<br />

compressi<strong>on</strong> occurring during <strong>the</strong> acclerati<strong>on</strong>-stages in two steps up to 2 kA.<br />

The accelerator secti<strong>on</strong> c<strong>on</strong>sists of 5 TESLA linear accelerator secti<strong>on</strong>s (each c<strong>on</strong>taining<br />

eight superc<strong>on</strong>ducting radiofrequency cavities) that can accelerate <strong>the</strong> electr<strong>on</strong>s<br />

to 1.2 GeV.<br />

Undulators<br />

The Flash undulator system c<strong>on</strong>sists of six 4.5 meter l<strong>on</strong>g permanent magnet undulators<br />

having 27.3 mm period with a fixed 12 mm gap. The undulator parameter K<br />

is 1.23 (corresp<strong>on</strong>ding to a peak magnetic field of 0.48 T).<br />

Infrared/THz undulator<br />

An electromagnetic l<strong>on</strong>g period undulator (400 mm period) placed after <strong>the</strong> SASE-<br />

FELundulatorsystemprovidesfar-infrared pulsesthataresynchr<strong>on</strong>izedto<strong>the</strong>VUV/soft<br />

X-ray phot<strong>on</strong> pulses <strong>–</strong> <strong>the</strong>y are emitted from <strong>the</strong> same electr<strong>on</strong> bunch[298]. The<br />

wavelength can be tuned between 1 <strong>and</strong> 200 µm (300 THz-1.5 THz). For wavelengths<br />

shorter than 20 µm <strong>the</strong> pulse energies of 10 µJ can be obtained.<br />

sFlash<br />

Undulator placed before <strong>the</strong> Sase undulator secti<strong>on</strong> of Flash.<br />

e - -gun<br />

Accelerator<br />

Seed<br />

sFlash undulator<br />

sFlash out<br />

Figure 11.2: Schematic of <strong>the</strong> sFlash seeding extensi<strong>on</strong> to Flash.<br />

Sase undulator<br />

C<strong>on</strong>sists of an extra 10 m l<strong>on</strong>g variable gap undulator before <strong>the</strong> current Flash<br />

undulator system. The seed laser system is a HHG setup using an arg<strong>on</strong> gas jet,<br />

<strong>the</strong> 21st harm<strong>on</strong>ic of a 800 nm Ti:Sa laser is used with a pulse length of 40 fs. The<br />

seeded free electr<strong>on</strong> laser radiati<strong>on</strong> is <strong>the</strong>n <strong>transport</strong>ed to a different experimental hall<br />

situated <strong>on</strong> top of <strong>the</strong> Flash hall.<br />

Flash-II<br />

Is a proposed extensi<strong>on</strong> of <strong>the</strong> Flash facility in Hamburg, Germany (see page 180)<br />

(http://flash.desy.de/flash ii/).<br />

Besides <strong>the</strong> operating FEL undulator at Flash, a sec<strong>on</strong>d undulator line with a<br />

separate tunnel<strong>and</strong> experimental hall is envisaged toextend<strong>and</strong> enhance <strong>the</strong>capacity<br />

of <strong>the</strong> Flash facility.


182 11. Free Electr<strong>on</strong> Laser facilities<br />

The facility will use a two stage Hghg seeding scheme (see page 17) with variable<br />

gap undulators to ensure shot-to-shot pulse reproducibility with variable femtosec<strong>on</strong>d<br />

durati<strong>on</strong>, gigawatt peak power, <strong>and</strong> full transverse <strong>and</strong> l<strong>on</strong>gitudinal coherence. A<br />

seeding opti<strong>on</strong> using HHG (High Harm<strong>on</strong>ic Generati<strong>on</strong>) laser generati<strong>on</strong> in gases to<br />

reach short seeding wavelengths is also c<strong>on</strong>tained within this project[299].<br />

The new experimental hall will house about six new <strong>beam</strong>lines with <strong>on</strong>e user<br />

<strong>beam</strong>line housing phot<strong>on</strong> diagnostics <strong>and</strong> <strong>beam</strong> manipulati<strong>on</strong> facilities.<br />

The electr<strong>on</strong> <strong>beam</strong> from <strong>the</strong> linear accelerator will be shared between <strong>the</strong> two<br />

Flash undulator chains with a fast <strong>beam</strong> switch. The project is currently in <strong>the</strong><br />

technical design phase[297, 299].<br />

e - -gun<br />

Accelerator<br />

Experimental stati<strong>on</strong>s:<br />

seed<br />

Seed<br />

sFlash und.<br />

Flash-II undulator<br />

Figure 11.3: Schematic of <strong>the</strong> Flash-II free electr<strong>on</strong> laser.<br />

sFlash out<br />

Flash-1 Sase und.<br />

• PG1<strong>–</strong>am<strong>on</strong>ochromatizedsmall focus<strong>beam</strong>linewithapermanenthigh-resoluti<strong>on</strong><br />

two-stage phot<strong>on</strong> spectrometer for inelastic scattering experiments.<br />

• PG2 <strong>–</strong> 50 µm focus after a high-resoluti<strong>on</strong> plane-grating m<strong>on</strong>ochromator.<br />

• BL1 <strong>–</strong> 100 µm spot after a toroidal mirror.<br />

• BL2 <strong>–</strong> 25 µm spot after an ellipsoidal mirror.<br />

• BL3 <strong>–</strong> a <strong>beam</strong>line without focussing optics.<br />

• THz undulator <strong>beam</strong>line <strong>–</strong> provides pulsed, coherent THz pulses that can be<br />

usedincombinati<strong>on</strong>with<strong>the</strong>freeelectr<strong>on</strong>laserradiati<strong>on</strong>from<strong>the</strong>Sase-undulator.


11.3. Scss <strong>–</strong> X-fel 183<br />

11.3 Scss <strong>–</strong> X-fel<br />

Scss<br />

The Spring-8 compact SASE source, Japan[295] is a test-bed for technologies to be<br />

used for <strong>the</strong> japanese X-fel. It has also been used to do accelerator research, for<br />

instance <strong>the</strong> first seeding experiments using a HHG laser[300].<br />

e - -gun<br />

Injector & accelerator<br />

Accelerator<br />

Undulators<br />

Figure 11.4: Schematic of <strong>the</strong> Scss free electr<strong>on</strong> laser test facility.<br />

The electr<strong>on</strong> gun used in <strong>the</strong> injector is of <strong>the</strong> <strong>the</strong>rmi<strong>on</strong>ic type, based <strong>on</strong> a singlecrystal<br />

CeB6 cathode which provides a <strong>beam</strong> current of 1 A at <strong>the</strong> initial <strong>beam</strong><br />

energy of 500 keV. In <strong>the</strong> accelerating structure <strong>the</strong> peak current becomes 300 A after<br />

bunch-compressi<strong>on</strong> with bunch-charges of 0.3 nC. The accelerating structure after<br />

<strong>the</strong> injector (which c<strong>on</strong>tains two S-b<strong>and</strong> accelerating structures) is composed C-b<strong>and</strong><br />

accelerating structures operating at 35 MV/m <strong>–</strong> yielding a final <strong>beam</strong> energy of 250<br />

MeV.<br />

Undulators<br />

Two, in vacuum, 600 period permanent magnet undulators with a minimum gap of<br />

3 mm <strong>and</strong> a period length of 15 mm. The maximum K is 1.5 <strong>and</strong> <strong>the</strong> radiati<strong>on</strong><br />

wavelength is between 30 <strong>and</strong> 61 nm.<br />

X-ray free electr<strong>on</strong> laser/ Spring-8<br />

This project was founded in 2006. The c<strong>on</strong>structi<strong>on</strong> is scheduled between 2006-2010<br />

<strong>and</strong> <strong>the</strong> start of operati<strong>on</strong>s is planned to 2011. The aim is to produce coherent<br />

radiati<strong>on</strong> of 1 ˚Angström wavelength with an 8 GeV electr<strong>on</strong> <strong>beam</strong> from a normalc<strong>on</strong>ducting<br />

C-b<strong>and</strong> (5.712 GHz) accelerator.<br />

Injector & accelerator<br />

The electr<strong>on</strong> gun was developed at <strong>the</strong> SCSS (see above) <strong>and</strong> is of <strong>the</strong> <strong>the</strong>rmi<strong>on</strong>ic<br />

type. The repetiti<strong>on</strong> rate of <strong>the</strong> accelerator is 60 Hz for macropulses c<strong>on</strong>taining 50<br />

sub-pulses. The X-ray pulse frequency is thus 3000 Hz. The peak current is 4 kA<br />

after <strong>the</strong> accelerati<strong>on</strong> up to 8 GeV (with bunch charges of 0.2 nC). The 128 C-b<strong>and</strong><br />

accelerators operating around 35 MV/m.


184 11. Free Electr<strong>on</strong> Laser facilities<br />

Undulators<br />

The in vacuum undulators is of hybrid type using permanent magnets <strong>and</strong> ir<strong>on</strong> yokes.<br />

The gap can be varied between 2 <strong>and</strong> 40 mm with a nominal operati<strong>on</strong> point at 4 mm<br />

gap (giving K=1.9). The resulting X-ray wavelength range is between 3 <strong>and</strong> 0.1 nm<br />

in bunches that are 50 fs l<strong>on</strong>g. At 1 ˚Angström wavelength <strong>the</strong> output power is 0.4<br />

mJ per pulse, <strong>the</strong> phot<strong>on</strong> flux is estimated to be 2·10 11 phot<strong>on</strong>s per pulse. The peak<br />

brightness is calculated to be in <strong>the</strong> order of 10 33 phot<strong>on</strong>s/mm 2 /mrad 2 /0.1%bw.<br />

11.4 Fermi@Elettra<br />

The Italian X-ray free electr<strong>on</strong> laser 1 is located near Trieste in Italy[301].<br />

e - -gun<br />

Injector & accelerator<br />

”Cern”-type, ”Fermi”-type acc. struct.<br />

Fel-1<br />

Fel-2<br />

Figure 11.5: Schematic of <strong>the</strong> Fermi@Elettra free electr<strong>on</strong> laser.<br />

The injector is of <strong>the</strong> photocathode variety with a copper target photoi<strong>on</strong>ized by <strong>the</strong><br />

3 rd harm<strong>on</strong>ic ofaTi:Salaser. Thephotocathodeis placedinanS-b<strong>and</strong>radiofrequency<br />

accelerating structure (1.6 cells).<br />

Seven”Cern”-type accelerating structures followed by seven S-b<strong>and</strong>”Fermi” structures<br />

provide <strong>the</strong> accelerati<strong>on</strong> up to 1.5 GeV. The repetiti<strong>on</strong> rate is 50 Hz.<br />

After <strong>the</strong> accelerator <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> can be steered ei<strong>the</strong>r into <strong>the</strong> first or <strong>the</strong><br />

sec<strong>on</strong>d undulator arrays.<br />

Undulators<br />

Following <strong>the</strong> linear accelerator part of <strong>the</strong> facility, <strong>the</strong>re is a switch for <strong>the</strong> electr<strong>on</strong><br />

<strong>beam</strong> <strong>–</strong> allowing <strong>the</strong> alternate usage of two different free electr<strong>on</strong> laser undulator<br />

systems; <strong>the</strong> first of which have been installed <strong>and</strong> <strong>the</strong> sec<strong>on</strong>d <strong>on</strong>e is to follow.<br />

Fel-1<br />

The first undulator system is a seeded Hghg cascade utilizing a tunable seed laser in<br />

<strong>the</strong> wavelength range 240-360 nm.<br />

The first undulator, referred to as <strong>the</strong> modulator where <strong>the</strong> microbunching density<br />

modulati<strong>on</strong> occurs is a 160 mm period 3.04 m l<strong>on</strong>g planar permanent magnet<br />

undulator with an undulator strength between K=3.9 to 4.9. This provided <strong>the</strong> basic<br />

wavelength that is to be up-c<strong>on</strong>verted in <strong>the</strong> radiator undulator system.<br />

Between <strong>the</strong> modulator <strong>and</strong> <strong>the</strong> sec<strong>on</strong>d undulator system <strong>the</strong>re is a 1 meter l<strong>on</strong>g<br />

<strong>beam</strong> <strong>transport</strong> secti<strong>on</strong> (which is a small chicane). The radiator undulator system<br />

1 http://www.elettra.trieste.it/FERMI/.


11.4. Fermi@Elettra 185<br />

c<strong>on</strong>sists of four Apple-type variable polarizati<strong>on</strong> undulators <strong>–</strong> tuned to a harm<strong>on</strong>ic<br />

n of <strong>the</strong> modulator wavelength (thus optimized for λ/n). They have a undulator<br />

period of 65 mm installed in four 2.34 meter l<strong>on</strong>g segments separated by 1.06 meter<br />

l<strong>on</strong>g <strong>transport</strong> secti<strong>on</strong>s, yielding a total length of <strong>the</strong> radiator system of 12.48 meters.<br />

Their undulator strengths are between 2.4 <strong>and</strong> 4.<br />

Including <strong>transport</strong> secti<strong>on</strong>s, <strong>the</strong> total length of <strong>the</strong> Fel-1 system is thus 16.5<br />

meters. This part of <strong>the</strong> free electr<strong>on</strong> laser facility provides radiati<strong>on</strong> in <strong>the</strong> range<br />

between 20 <strong>and</strong> 100 nm (see Table 11.2).<br />

Fel-2<br />

The sec<strong>on</strong>d free electr<strong>on</strong> laser of <strong>the</strong> Fermi will be built after <strong>the</strong> first <strong>on</strong>e is completed.<br />

It is intended to be a cascaded double HGHG scheme (using two modulators<br />

<strong>and</strong> two radiators). This scheme is flexible enough to also allow HHG seeding <strong>and</strong><br />

Echo-enabled harm<strong>on</strong>ic generati<strong>on</strong> in <strong>the</strong> future. By using Apple-II undulators <strong>the</strong><br />

polarizati<strong>on</strong> can be c<strong>on</strong>trolled <strong>and</strong> tuned in a range around 10% of <strong>the</strong> energy defined<br />

by <strong>the</strong> <strong>beam</strong> energy[302].<br />

Parameter Fel-1 Fel-2<br />

Wavelength range (fundamental) 20 <strong>–</strong> 100 nm 3-20 nm<br />

Phot<strong>on</strong> energy 62 <strong>–</strong> 12 eV 413<strong>–</strong>62 eV<br />

Pulse durati<strong>on</strong> (FWHM) 20 <strong>–</strong> 40 fs < 100 fs<br />

Repetiti<strong>on</strong> rate 10-50 Hz 10-50 Hz<br />

Peak power (from av.) 1 <strong>–</strong> 5 GW 1 GW<br />

Spectral width (FWHM) ∼ 20−40 meV ∼ 20−40 meV<br />

Phot<strong>on</strong>s per pulse 2·10 14 at 100 nm 10 13 at 10 nm<br />

Peak Brilliance 10 29 −10 30<br />

Table 11.2: Phot<strong>on</strong> output parameters for Fermi@Elettra. Taken from <strong>the</strong> laboratory homepage.<br />

Experiments:<br />

• DIPROI <strong>–</strong> Diffracti<strong>on</strong> <strong>and</strong> projecti<strong>on</strong> imaging<br />

• LDM <strong>–</strong> Low density matter<br />

• EIS <strong>–</strong> Elastic <strong>and</strong> inelastic scattering.


186 11. Free Electr<strong>on</strong> Laser facilities<br />

11.5 Lcls <strong>–</strong> Linac Coherent Light Source<br />

The Lcls was designed to become <strong>the</strong> world’s first hard X-ray free electr<strong>on</strong> laser with<br />

<strong>the</strong> fundamental wavelength lying between 0.55 <strong>and</strong> 10 keV[96, 294, 303]. The facility<br />

is located at <strong>the</strong>SLACnati<strong>on</strong>al accelerator laboratory inMenloPark, California, USA<br />

<strong>and</strong> went into operati<strong>on</strong> during <strong>the</strong> year 2009 2 .<br />

e - -gun<br />

Injector <strong>and</strong> accelerator<br />

Accelerator<br />

Figure 11.6: Schematic of <strong>the</strong> Lcls free electr<strong>on</strong> laser.<br />

Undulator<br />

The Lcls free electr<strong>on</strong> laser is currently utilizing <strong>the</strong> last kilometre of <strong>the</strong> SLAClinear<br />

electr<strong>on</strong> accelerator (which is 3 km l<strong>on</strong>g in total). The SLAC accelerator use S-b<strong>and</strong><br />

radiofrequency-fields (2,856 MHz) in copper cavities <strong>and</strong> can accelerate electr<strong>on</strong>s up<br />

to 50 GeV - if <strong>the</strong> whole three kilometers available is used. As <strong>the</strong> Lcls uses <strong>on</strong>ly<br />

<strong>on</strong>e kilometer electr<strong>on</strong> energies of 3.5 to 15 GeV can be delivered to <strong>the</strong> undulators.<br />

A photocathode radiofrequency electr<strong>on</strong> gun serves as injector; <strong>the</strong> gun c<strong>on</strong>sists<br />

of a copper target that is photoi<strong>on</strong>ized by a frequency trippled Ti:Sapphire system.<br />

This can operate up to 120 Hz with bunch charges of 0.25 nC (with a peak current of<br />

35 A). With bunch compressi<strong>on</strong> during <strong>the</strong> accelerati<strong>on</strong> <strong>the</strong> peak current is amplified<br />

up to 3.5 kA.<br />

Undulators<br />

The undulator secti<strong>on</strong> of Lcls is 132 m l<strong>on</strong>g in total <strong>and</strong> c<strong>on</strong>sists of thirty-three 3.4<br />

m l<strong>on</strong>g planar permanent magnet undulators - separated by short <strong>and</strong> l<strong>on</strong>g <strong>transport</strong><br />

secti<strong>on</strong>s. The undulator period is 30 mm with a fixed gap of 6.8 mm. The undulator<br />

parameter K is 3.5 (corresp<strong>on</strong>ding to a magnetic field of 1.25 T at <strong>the</strong> menti<strong>on</strong>ed<br />

gap).<br />

Lcls-II proposal<br />

Proposed extensi<strong>on</strong> of <strong>the</strong> Linac Coherent Light Source in Stanford, California, USA<br />

[304] (see page 186). This proposal c<strong>on</strong>siders <strong>the</strong> use of <strong>the</strong> 2 nd km (of 3 km in total,<br />

<strong>the</strong> first kilometer remains untouched) of <strong>the</strong> SLAC linear accelerator. By adding<br />

a sec<strong>on</strong>d electr<strong>on</strong> gun <strong>and</strong> injector 1 km upstream from <strong>the</strong> Lcls injector <strong>the</strong> mid<br />

kilometer of <strong>the</strong> accelerator could be used to operate a soft X-ray free electr<strong>on</strong> laser in<br />

2 Lcls homepage: https://slacportal.slac.stanford.edu.


11.5. Lcls <strong>–</strong> Linac Coherent Light Source 187<br />

parallel to <strong>the</strong> existing Lcls if <strong>the</strong> electr<strong>on</strong> bunches are made to travel in a bypass<br />

line running side by side to <strong>the</strong> last kilometer (used for <strong>the</strong> accelerati<strong>on</strong> of <strong>the</strong> electr<strong>on</strong><br />

bunches for <strong>the</strong> o<strong>the</strong>r free electr<strong>on</strong> laser). Besides extending <strong>the</strong> phot<strong>on</strong>-energy range<br />

of <strong>the</strong> facility this would also allow simultaneous operati<strong>on</strong> of two free electr<strong>on</strong> laser in<br />

parallel which would increase <strong>the</strong> number of users significantly.<br />

e<br />

Bypass<br />

- -gun Undulators<br />

Accelerator<br />

Figure 11.7: Schematic of <strong>the</strong> Lcls-II proposal. The grey area marks <strong>the</strong> existing Lcls facility.<br />

Upgrade of <strong>the</strong> existing facility<br />

An upgrade of <strong>the</strong> current Lcls is c<strong>on</strong>sidered to allow a larger gap (<strong>and</strong> thus higher<br />

energies). A sec<strong>on</strong>d harm<strong>on</strong>ic afterburner optimizing <strong>the</strong> bunching at 0.62 ˚A wavelength<br />

will be added in 2010 to produce half <strong>the</strong> fundamental wavelength efficiently.<br />

This will be achieved by increasing <strong>the</strong> gap of <strong>the</strong> last 8-10 undulators to give <strong>the</strong>m<br />

a K of 2.26 (as opposed to 3.5 for <strong>the</strong> 23-25 undulators).<br />

The soft X-ray undulators<br />

Withtwoundulators, each36ml<strong>on</strong>ghaving6-60˚Awavelengthrange throughvariable<br />

gap, a number of possible opportunities are given: self-seeding by placing a grating<br />

between <strong>the</strong> undulators, Eehg-seeding, etc. It is also possible to c<strong>on</strong>trol <strong>the</strong> polarizati<strong>on</strong><br />

of <strong>the</strong> light by adding Apple-II type undulators at <strong>the</strong> end of each undulator<br />

that would offer variable polarizati<strong>on</strong>.<br />

Since <strong>the</strong> accelerator for <strong>the</strong> soft X-ray free electr<strong>on</strong> laser operate in <strong>the</strong> 3-7 GeV<br />

range, <strong>the</strong> repetiti<strong>on</strong> rate can be increased to 320 Hz.<br />

Experiments:<br />

• AMO <strong>–</strong> Atomic, molecular <strong>and</strong> optical science.<br />

• CXI <strong>–</strong> Coherent X-ray imaging.<br />

• MEC <strong>–</strong> Matter in extreme c<strong>on</strong>diti<strong>on</strong>s.<br />

• SXR <strong>–</strong> Soft X-ray materials science.<br />

• XCS <strong>–</strong> X-ray correlati<strong>on</strong> spectroscopy.<br />

• X<strong>PP</strong> <strong>–</strong> X-ray pump probe.


188 11. Free Electr<strong>on</strong> Laser facilities<br />

11.6 Facilities under c<strong>on</strong>structi<strong>on</strong><br />

11.7 The European Xfel<br />

The European Xfel will be 3.4 km l<strong>on</strong>g, starting at <strong>the</strong> DESY-Bahrenfeld site in<br />

Hamburg, Germany <strong>and</strong> <strong>the</strong> research campus will be situated south of <strong>the</strong> town of<br />

Schenefeld[305]. Thestartofoperati<strong>on</strong>s isplannedtobein2014(http://www.xfel.eu/).<br />

Injector <strong>and</strong> accelerator<br />

The injector is of <strong>the</strong> radio-frequency accelerated photocathode variety with a Cs2Te<br />

target (very similar to <strong>the</strong> <strong>on</strong>e used at Flash presented above). Both <strong>the</strong> injector <strong>and</strong><br />

<strong>the</strong> linear accelerator c<strong>on</strong>sists of superc<strong>on</strong>ducting L-b<strong>and</strong> (1.3 GHz) Tesla cavities<br />

that will operate at 23.6 MV/m. The repetiti<strong>on</strong> rate between macropulses is 10 Hz,<br />

each such pulse can house 3000 pulses.<br />

In total <strong>the</strong> accelerator will be 1.6 km l<strong>on</strong>g with 116 accelerator modules, each 12<br />

m l<strong>on</strong>g. The peak current is projected to be 5 kA with a bunch charge of 1 nC. The<br />

maximum <strong>beam</strong> energy attainable is 20 GeV, but normal operati<strong>on</strong> will be at 17.5<br />

Gev electr<strong>on</strong> energy.<br />

Undulators<br />

Sase 2<br />

Sase 1<br />

U 1<br />

Sase 3<br />

U 2<br />

e - dump<br />

Figure 11.8: The electr<strong>on</strong> <strong>beam</strong> distributi<strong>on</strong> between different undulators in <strong>the</strong> European Xfel.<br />

Toaccommodate <strong>the</strong>differentdem<strong>and</strong>sfrom <strong>the</strong>experimentalside<strong>–</strong><strong>and</strong>toprovide<br />

maximum flexibility for future upgrades <strong>–</strong> <strong>the</strong> 17.5 GeV electr<strong>on</strong> <strong>beam</strong> can be sent<br />

al<strong>on</strong>g two different paths (Figure 11.8) servicing five different undulators.<br />

Sase 1 is a planar permanent magnet undulator optimized to deliver 12.4 keV<br />

hard X-rays (λ = 0.1 nm) <strong>–</strong> <strong>the</strong> magnetic period is 35.6 <strong>and</strong> <strong>the</strong> gap is 10 mm (giving<br />

K=3.3). The total undulator length is 201.3 m.<br />

Sase 2 provides tunable hard X-rays between 3.1 <strong>and</strong> 12.4 keV. It is a planar permanent<br />

magnet undulator that thus provides linearly polarized light. This undulator<br />

have a magnet period of 47.9 mm with a gap adjustable between 19 <strong>and</strong> 10 mm (K


11.7. The European Xfel 189<br />

between 2.8 <strong>and</strong> 6.1). The total length is 256.2 m.<br />

Sase 3 is an Apple II device which can deliver circularly or linearly polarized light<br />

in <strong>the</strong> range between 0.8-3.1 keV. If <strong>the</strong> accelerator is operated at 10 GeV electr<strong>on</strong><br />

energy <strong>the</strong> range becomes 0.25-1.0 keV. The undulator’s period is 80 mm with gaps<br />

between 23 <strong>and</strong> 10 mm. The total length is 128.1 m.<br />

U 1 & U 2 provides sp<strong>on</strong>taneous synchrotr<strong>on</strong> radiati<strong>on</strong> in <strong>the</strong> range between<br />

20-100 keV by utilizing <strong>the</strong> ”spent” electr<strong>on</strong> <strong>beam</strong> after <strong>the</strong> Sase 2 undulator. The<br />

length is 61 m each.<br />

To attain <strong>the</strong> accuracy <strong>and</strong> maintainability of such a large undulator system <strong>the</strong><br />

undulators are built up from 5 m l<strong>on</strong>g magnet arrays intersected by 1.1 m l<strong>on</strong>g<br />

<strong>transport</strong> secti<strong>on</strong>s.<br />

Sase 1 Sase 2 Sase 3 Unit<br />

Ee− 17.5 17.5 17.5 10 GeV<br />

λ 0.1 0.1-0.4 0.4-1.6 4.9 nm<br />

EPhot. 12.4 12.4-3.1 3.1-0.8 0.25 keV<br />

Ppeak 20 20-80 80-130 150 GW<br />

˜P 65 65-260 260-420 490 W<br />

Phot./pulse 10 12<br />

0.1−1.6·10 13<br />

0.16−1.0·10 13<br />

3.7·10 14 #<br />

Flux 3.0·10 16<br />

0.3−4.8·10 17<br />

0.48−4.8·10 17<br />

1.1·10 19 #/s<br />

Peak brill. 5.0·10 33<br />

0.5−2.2·10 33<br />

0.22−2.3·10 33<br />

1.0·10 32<br />

a)<br />

Aver. brill. 1.6·10 25<br />

0.16−6.5·10 34<br />

0.65−5.9·10 24<br />

2.8·10 23<br />

a)<br />

Table 11.3: Radiati<strong>on</strong> parameters of <strong>the</strong> three free electr<strong>on</strong> laser undulators at <strong>the</strong> European Xfel <strong>–</strong> from<br />

simulati<strong>on</strong>s[89]. a)brilliance is given in <strong>the</strong> unit phot<strong>on</strong>/0.1% bw/s/mm 2 /mrad 2 . The pulse durati<strong>on</strong> is 100 fs<br />

overall.<br />

Experiments:<br />

• FXE Femtosec<strong>on</strong>d X-ray Experiments<br />

• HED High-Energy Density matter experiments<br />

• SPB Single Particles, Clusters & Biomolecules<br />

• MID Materials Imaging <strong>and</strong> Dynamics<br />

• SQS Small Quantum Systems<br />

• SCS Spectroscopy & Coherent Scattering


190 11. Free Electr<strong>on</strong> Laser facilities<br />

11.8 SwissFEL<br />

SwissFEL injector test facility<br />

Currently in operati<strong>on</strong> at <strong>the</strong> Paul Scherrer Institut (PSI) in Switerl<strong>and</strong> is <strong>the</strong> injector<br />

testbed[306]for<strong>the</strong>proposedSwissFEL. Theintentwiththisfacility istodevelop<strong>the</strong><br />

electr<strong>on</strong> gun <strong>and</strong> injector technology sufficiently to achieve low enough emittance for<br />

free electr<strong>on</strong> laser operati<strong>on</strong>. An important parameter for <strong>the</strong> intended free electr<strong>on</strong><br />

laser is that it will have two pulses per macrobunch with a defined delay between <strong>the</strong><br />

(i.e. minimal jitter) subpulses <strong>–</strong> this to allow very precise X-ray pump- X-ray probe<br />

experiments <strong>–</strong> something which also is being developed at <strong>the</strong> PSI.<br />

A photocathode electr<strong>on</strong> gun enclosed in a radiofrequency S-b<strong>and</strong> accelerating<br />

module provides <strong>the</strong> electr<strong>on</strong> bunches for <strong>the</strong> linear accelerator, at <strong>the</strong> cathode (depending<br />

<strong>on</strong> l<strong>on</strong>g or short pulse mode) it delivers 22 or 3 A at <strong>the</strong> cathode. The accelerator<br />

structure is based <strong>on</strong> four normally c<strong>on</strong>ducting S-b<strong>and</strong> accelerati<strong>on</strong> modules<br />

(operating at 20 MV/m) followed by a fourth harm<strong>on</strong>ic X-b<strong>and</strong> cavity (which decelerates<br />

<strong>the</strong> electr<strong>on</strong>s somewhat to attain a better bunch compressi<strong>on</strong>, giving shorter<br />

pulses). The repetiti<strong>on</strong> rate of <strong>the</strong> source is 100 Hz <strong>and</strong> <strong>the</strong> <strong>beam</strong> energy is 250 MeV.<br />

<strong>the</strong> SwissFEL proposal<br />

e - gun<br />

Accelerator secti<strong>on</strong>s<br />

Seed<br />

Figure 11.9: Schematic of <strong>the</strong> proposed SwissFEL.<br />

d’Artagnan Athos<br />

Aramis<br />

Undulators<br />

The time-schedule <strong>on</strong> <strong>the</strong> project’s homepage 3 indicates that <strong>the</strong> planned start of<br />

operati<strong>on</strong>s is targeted towards in 2016.<br />

This seeded free electr<strong>on</strong> laser is envisi<strong>on</strong>ed to deliver transform-limited pulses in<br />

<strong>the</strong> soft X-ray range. The highest electr<strong>on</strong> <strong>beam</strong> energy at <strong>the</strong> entrance of <strong>the</strong> hard<br />

X-ray undulator secti<strong>on</strong> is 5.8 GeV. To facilitate experiments examining temporal<br />

processes <strong>the</strong> macropulse durati<strong>on</strong> range between 6 <strong>and</strong> 30 femtosec<strong>on</strong>ds with 50 ns<br />

separati<strong>on</strong> between two subpulses. To achieve short pulses <strong>the</strong> bunch charge is kept<br />

between 10 <strong>and</strong> 200 pC. For <strong>the</strong> soft X-rays (less than 1.6 keV) circular polarizati<strong>on</strong><br />

can be employed (for instance for magnetism measurements).<br />

3 http://www.psi.ch/swissfel/swissfel


11.8. SwissFEL 191<br />

Wavelength range 7-0.1 nm<br />

Rep. Rate 100 Hz (of 2 sub-pulses, 50 ns separati<strong>on</strong>)<br />

Pulse durati<strong>on</strong> (FWHM) 30 or 6 fs (high/low charge mode)<br />

Peak power 10 GW<br />

Peak Brilliance 1-10·10 32<br />

Injector & accelerator<br />

Table 11.4: Main parameters of <strong>the</strong> SwissFEL.<br />

The photocathode injector uses a S-b<strong>and</strong> <strong>and</strong> X-b<strong>and</strong> accelerati<strong>on</strong> before <strong>the</strong> first<br />

bunch-compressi<strong>on</strong>, about 24 meters l<strong>on</strong>g. This is followed by three C-b<strong>and</strong> accelerating<br />

linear accelerati<strong>on</strong> secti<strong>on</strong>s, in total 208 meters l<strong>on</strong>g. The C-b<strong>and</strong> parts were<br />

chosen over <strong>the</strong> S-b<strong>and</strong> (used for <strong>the</strong> test facility) as <strong>the</strong>y preserve <strong>the</strong> emittance<br />

better (provided a good alignment)[307], <strong>the</strong> power c<strong>on</strong>sumpti<strong>on</strong> is lower <strong>and</strong> <strong>the</strong><br />

number of RF stati<strong>on</strong>s fewer[308]. For <strong>the</strong> soft-X-ray secti<strong>on</strong> <strong>on</strong>ly two are used before<br />

<strong>the</strong> electr<strong>on</strong> <strong>beam</strong> enter <strong>the</strong> undulator system, giving a final electr<strong>on</strong> <strong>beam</strong> energy<br />

of 2.1 or 3.4 GeV <strong>–</strong> <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> is accelerated to 5.8 GeV in <strong>the</strong> last secti<strong>on</strong>,<br />

exclusively serving <strong>the</strong> hard X-ray undulator system. The total length of <strong>the</strong> whole<br />

free electr<strong>on</strong> laser is 715 meters. Two pulse modes are envisi<strong>on</strong>ed with 13 or 2.1 fs.<br />

Undulators<br />

The details <strong>on</strong> <strong>the</strong> undulators is from a presentati<strong>on</strong> given by T. Garvey at <strong>the</strong> 32 nd<br />

free electr<strong>on</strong> laser c<strong>on</strong>ference[308].<br />

• Aramis <strong>–</strong> an in-vacuum planar undulator system with variable gap to produce<br />

Sase free electr<strong>on</strong> laser radiati<strong>on</strong> in <strong>the</strong> 0.1 to 70 nm range. The undulator<br />

period is 15 mm (with a gap larger than 4 mm <strong>and</strong> K=1.2) <strong>and</strong> <strong>the</strong> total length<br />

is 60 meters in 4 meter secti<strong>on</strong>s.<br />

• Athos <strong>–</strong> <strong>the</strong> undulator system is built up from 40 mm period Apple-II type<br />

undulators with a fixed 6.5 mm gap <strong>and</strong> full polarizati<strong>on</strong> c<strong>on</strong>trol. Intended<br />

for both Sase <strong>and</strong> seeded free electr<strong>on</strong> laser operati<strong>on</strong> in <strong>the</strong> wavelength range<br />

between 0.7 <strong>and</strong> 7 nm.<br />

• d’Artagnan <strong>–</strong> intended for use as <strong>the</strong> radiator in aHHG seeded HGHG cascade<br />

toge<strong>the</strong>r with <strong>the</strong> Athos undulator as radiator. This extends <strong>the</strong> energy range<br />

of this part of <strong>the</strong> SwissFEL to l<strong>on</strong>ger wavelengths.<br />

Experiments<br />

The scientific program at <strong>the</strong> SwissFEL is focussed towards <strong>the</strong> study of ultra-fast<br />

phenomena at <strong>the</strong> nanoscale. The double pulse structure <strong>and</strong> <strong>the</strong> strive to create very<br />

short pulses is a testament to this goal.<br />

According to <strong>the</strong> scinece case 4 for <strong>the</strong> Swissfel <strong>the</strong> scientific program will focus<br />

up<strong>on</strong> five areas:<br />

4 http://www.psi.ch/swissfel/CurrentSwissFELPublicati<strong>on</strong>sEN/SwissFEL Science Case<br />

small.pdf


192 11. Free Electr<strong>on</strong> Laser facilities<br />

1. Ultrafast magnetizati<strong>on</strong> dynamics at <strong>the</strong> nanoscale.<br />

2. Catalysis <strong>and</strong> soluti<strong>on</strong> chemistry, i.e. <strong>the</strong> lifetime <strong>and</strong> structure of short-lived<br />

intermediate states <strong>on</strong> surfaces or in soluti<strong>on</strong>.<br />

3. Coherent diffracti<strong>on</strong> of nanostructures <strong>–</strong> lens-less imaging can provide atomic<br />

resoluti<strong>on</strong> of biological <strong>and</strong> inorganic nanostructures.<br />

4. Ultrafast biochemistry<br />

5. Time-resolved spectroscopies of correlated electr<strong>on</strong> materials.<br />

11.9 Proposed facilities<br />

Besides <strong>the</strong> proposed extensi<strong>on</strong>s of currently existing facilites (as <strong>the</strong> Flash-II, Lcls-<br />

II, SwissFEL) <strong>and</strong> projects under c<strong>on</strong>structi<strong>on</strong> as <strong>the</strong> European <strong>and</strong> <strong>the</strong> Spring-8<br />

X-fels included here <strong>the</strong> Korean X-fel project.<br />

11.10 PAL-X-fel<br />

The X-fel at <strong>the</strong> Poohang accelerator laboratory in Korea is currently in <strong>the</strong> technical<br />

design phase[309]. Here we will c<strong>on</strong>sider <strong>the</strong> design assuming a 10 GeV linear<br />

accelerator. Shorter more compact designs using 3.7 GeV linear accelerators envisi<strong>on</strong>ing<br />

<strong>the</strong> use of <strong>on</strong>ly <strong>the</strong> third harm<strong>on</strong>ic for free electr<strong>on</strong> laser radiati<strong>on</strong> have been<br />

c<strong>on</strong>sidered.<br />

Accelerator<br />

The accelerator secti<strong>on</strong> c<strong>on</strong>sists of three S-b<strong>and</strong> secti<strong>on</strong>s (4, 28 <strong>and</strong> 96 three meter<br />

l<strong>on</strong>g radiofrequency accelerating structures) <strong>and</strong> <strong>on</strong>e short (0.6 m) X-b<strong>and</strong> secti<strong>on</strong> <strong>–</strong><br />

<strong>the</strong> total length of <strong>the</strong> S-b<strong>and</strong> secti<strong>on</strong> is 550 m. The accelerating gradients in <strong>the</strong><br />

S-b<strong>and</strong> structures is 27 MV/m. Simulati<strong>on</strong>s of both 1 nC <strong>and</strong> <strong>and</strong> 0.2 nC charges<br />

have been d<strong>on</strong>e. With 1 nC <strong>the</strong> radiated output power is 6 GW, 0.2 nC gives about<br />

2 GW.<br />

Undulators<br />

The undulator c<strong>on</strong>sidered is a 94 m l<strong>on</strong>g (or around 100 m) in vacuum undulator with<br />

a period of 2.23 cm <strong>and</strong> undulator parameter K=2.22.<br />

As this facility is in <strong>the</strong> technical design stage no informati<strong>on</strong> <strong>on</strong> <strong>the</strong> intended<br />

experimental program is available in <strong>the</strong> time of writing.


11.10. PAL-X-fel 193<br />

Summary<br />

• The currently two operating soft <strong>and</strong> hard X-ray free electr<strong>on</strong>laser<br />

are Flash, Hamburg, Germany<strong>and</strong>Lcls, Stanford,<br />

USA.<br />

• Facilities currently being built is <strong>the</strong> Fermi@Elettra, Trieste,<br />

Italy; <strong>the</strong> european <strong>and</strong> <strong>the</strong> japanese X-fels.<br />

• Am<strong>on</strong>g facilities in detailed technical design phase are <strong>the</strong><br />

SwissFel in Switzerl<strong>and</strong> <strong>and</strong> <strong>the</strong> PAL-X-fel in Korea.<br />

• In c<strong>on</strong>trast to <strong>the</strong> o<strong>the</strong>r facilities <strong>the</strong> Flash free electr<strong>on</strong><br />

laser have less permanent in house experiments (encouraging<br />

users to bring <strong>the</strong>ir own set-ups).<br />

• Both Flash <strong>and</strong> <strong>the</strong> Lcls are currently specifying upgrades<br />

to <strong>the</strong> facilites. Flash have a seeding experiment already.<br />

Both are c<strong>on</strong>sidering additi<strong>on</strong>s to <strong>the</strong> current accelerator layout<br />

that would increase <strong>the</strong> number of experimental secti<strong>on</strong>s.<br />

Notably both facilites, currently producing Sase free electr<strong>on</strong><br />

laser radiati<strong>on</strong> <strong>–</strong> aims to install seeded HGHG or EEHG cascades.<br />

• Most facilites are trying to (or at least have <strong>the</strong> opti<strong>on</strong> to) run<br />

with lower bunch charges as to shorten <strong>the</strong> pulses.<br />

• The overall length of <strong>the</strong> newer facilites is shorter, owing to<br />

advances made in technology to reduce <strong>the</strong> emittance of <strong>the</strong><br />

electr<strong>on</strong> <strong>beam</strong>.


12. Outlook & C<strong>on</strong>clusi<strong>on</strong>s<br />

Written by: A. Lindblad<br />

12.1 Current trends<br />

More compact sources <strong>and</strong> alternative approaches<br />

A significant effort is currently being undertaken to find technology that reduce <strong>the</strong><br />

normalized emittance of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> <strong>and</strong> <strong>the</strong> length of <strong>the</strong> undulators <strong>–</strong> both<br />

factors that severely impacts <strong>the</strong> overall length <strong>and</strong> cost of a free electr<strong>on</strong> laser.<br />

The radiofrequency systems driving <strong>the</strong> cavities of <strong>the</strong> linear accelerators are also<br />

important cost drivers.<br />

The European X-fel <strong>and</strong> <strong>the</strong> Lcls is 3.4 km <strong>and</strong> 3 km l<strong>on</strong>g respectively. They use<br />

L-b<strong>and</strong> <strong>and</strong> S-b<strong>and</strong> accelerator technology respectively toge<strong>the</strong>r with photocathode<br />

electr<strong>on</strong> guns. In c<strong>on</strong>trast <strong>the</strong> X-fel at Spring-8 will use C-b<strong>and</strong> accelerators (with<br />

frequencies four times that of L-b<strong>and</strong> <strong>and</strong> twice that of S-b<strong>and</strong>) <strong>and</strong> a <strong>the</strong>rmi<strong>on</strong>ic gun<br />

<strong>–</strong> both c<strong>on</strong>tributing to <strong>the</strong> significantly shorter accelerator part of <strong>the</strong> facility of 750<br />

m. Also c<strong>on</strong>tributing to this is <strong>the</strong> in vacuum undulators with small gaps that allow<br />

<strong>the</strong> undulator period to be shortened[310] <strong>–</strong> similar undulators are c<strong>on</strong>sidered for <strong>the</strong><br />

SwissFel.<br />

The incentives are thus many to make more compact X-ray sources, not <strong>on</strong>ly<br />

motivated by cost reducti<strong>on</strong> but also scientifically[311], <strong>on</strong>e such source would be <strong>the</strong><br />

plasma wake-field accelerators[312]. In such a device electr<strong>on</strong>s are accelerated in a<br />

bubble created in a plasma by a str<strong>on</strong>g laser propagating through a medium (often<br />

a gas). A plasma wake-field accelerator can be used toge<strong>the</strong>r with an undulator to<br />

create sp<strong>on</strong>taneous X-ray radiati<strong>on</strong>[313]. A questi<strong>on</strong> yet to be answered is if <strong>the</strong><br />

electr<strong>on</strong> <strong>beam</strong> in <strong>the</strong> plasma wake-field can be made with high enough quality to <strong>and</strong><br />

if collective effects like microbunching can occur to allow free electr<strong>on</strong> lasing. The<br />

repetiti<strong>on</strong> rate of <strong>the</strong> driving lasers is currently slow, in <strong>the</strong> order of 10’s of Hz.<br />

XFELO (X-ray free electr<strong>on</strong> laser oscillator) <strong>and</strong> <strong>the</strong> regenerative free electr<strong>on</strong><br />

laser amplifier are suggested low gain amplificati<strong>on</strong> systems, both use cavity feedback<br />

to achieve lasing with cavities made from highly reflecting diam<strong>on</strong>d mirrors that work<br />

in <strong>the</strong> X-ray range. Unlike <strong>the</strong> single-pass high-gain systems described in this book<br />

both <strong>the</strong> XFELO <strong>and</strong> <strong>the</strong> regenerative FEL amplifiers dem<strong>and</strong> high repetiti<strong>on</strong> rate<br />

195


196 12. Outlook & C<strong>on</strong>clusi<strong>on</strong>s<br />

electr<strong>on</strong> guns <strong>–</strong> o<strong>the</strong>rwise <strong>the</strong> electr<strong>on</strong>bunches do not arrive in time to meet <strong>the</strong> pulse<br />

in <strong>the</strong> cavity.<br />

Higher repetiti<strong>on</strong> rates<br />

Normally c<strong>on</strong>ducting accelerator technology limits <strong>the</strong> repetiti<strong>on</strong> rate to about 100<br />

Hz, e.g. at <strong>the</strong> Lcls. Currently Flash <strong>and</strong>, in <strong>the</strong> near future, <strong>the</strong> European X-Fel<br />

operate with phot<strong>on</strong> pulses that have MHz substructures.<br />

Achieving high average brilliance is an active area of pursuit <strong>–</strong> as it opens up<br />

new fields of research <strong>and</strong> allows for serving many user experiments in parallel or in<br />

series (at reduced intensity or repetiti<strong>on</strong> rate respectively). The technology (superc<strong>on</strong>ducting<br />

accelerator secti<strong>on</strong>s? electr<strong>on</strong> guns, etc.) for driving a X-ray free electr<strong>on</strong><br />

laser with MHz frequency exist in principle with <strong>the</strong> cost being <strong>the</strong> biggest obstacle.<br />

With o<strong>the</strong>r comp<strong>on</strong>ents of <strong>the</strong> free electr<strong>on</strong> laser becoming shorter <strong>and</strong> more efficient<br />

this deterring factor may be less str<strong>on</strong>g in <strong>the</strong> future. Generic c<strong>on</strong>cepts for what<br />

would be <strong>the</strong> necessary comp<strong>on</strong>ents in a free electr<strong>on</strong> laser facility achieving MHz<br />

repetiti<strong>on</strong> rates have been studied by J. Corlett <strong>and</strong> co-workers[314].<br />

Polarizati<strong>on</strong> c<strong>on</strong>trol<br />

We fleetingly touched up<strong>on</strong> <strong>the</strong> subject of polarizati<strong>on</strong>s different from <strong>the</strong> linear inplane<br />

variety produced by planar undulators with a vertical magnetic field in chapter<br />

3 when different types of undulators were discussed.<br />

Typically <strong>the</strong> first generati<strong>on</strong> free electr<strong>on</strong> laser in <strong>the</strong> VUV <strong>and</strong> X-ray ranges<br />

utilize <strong>the</strong> aforementi<strong>on</strong>ed planar undulators with vertical magnetic fields to generate<br />

<strong>the</strong> phot<strong>on</strong> <strong>beam</strong> generating linearly polarized light. Many experiments, for instance<br />

those c<strong>on</strong>cerning spin properties <strong>and</strong> dynamics of materials, e.g. magnetic properties<br />

<strong>and</strong>magnetizati<strong>on</strong>/de-magnetizati<strong>on</strong> processes, requirecircularly (helically) polarized<br />

light for <strong>the</strong>ir inquiry.<br />

On paper <strong>the</strong>re it is not harder to produce helical light than linearly polarized, in<br />

practice helical undulators involve more moving parts <strong>and</strong> are thus more intricate to<br />

build <strong>and</strong> utilize. This is probably <strong>the</strong> main reas<strong>on</strong> why <strong>the</strong> first generati<strong>on</strong> of X-ray<br />

free electr<strong>on</strong> laser have chosen to operate with planar undulators <strong>–</strong> as seen earlier <strong>the</strong><br />

error tolerances for <strong>the</strong> undulator systems are tight without <strong>the</strong> complexity of, e.g. an<br />

Apple type undulator.<br />

Above roughtly 2 keV phot<strong>on</strong> energies it is possible to produce befringement transparent<br />

materials that c<strong>on</strong>vert linearly polarized X-rays to helical polarizati<strong>on</strong>s given<br />

that <strong>the</strong> thickness <strong>and</strong> orientati<strong>on</strong> of <strong>the</strong> crystals c<strong>on</strong>spire to do so (e.g. diam<strong>on</strong>d<br />

crystals with <strong>the</strong> proper thickness will do this). In many cases this will be a true<br />

<strong>on</strong>e-shot experiment which destroys <strong>the</strong> crystal <strong>–</strong> <strong>the</strong> cost is though microscopical<br />

compared to installing a new undulator secti<strong>on</strong> in a free electr<strong>on</strong> laser.<br />

In <strong>the</strong> soft X-ray regime it is however necessary to use an undulator to create<br />

helical polarizati<strong>on</strong>, am<strong>on</strong>g o<strong>the</strong>r things <strong>the</strong> absorbance for soft X-rays in materials<br />

is too high for experiments to be efficiently executed.<br />

Besides <strong>the</strong> Apple type of undulators described earlier, two planar undulators with<br />

<strong>the</strong> undulator planes rotated with respect to each o<strong>the</strong>r can be used to produce tilted<br />

linear <strong>and</strong> helical polarizati<strong>on</strong>s. The c<strong>on</strong>figurati<strong>on</strong> drawn in <strong>the</strong> top of Figure 12.1<br />

will create tilted linearly polarized light if <strong>the</strong> undulators are placed at a distance


12.1. Current trends 197<br />

that cause <strong>the</strong>m to emit in phase. If, <strong>on</strong> <strong>the</strong> o<strong>the</strong>r h<strong>and</strong>, a magnetic chicane secti<strong>on</strong> is<br />

installed between <strong>the</strong>undulators <strong>the</strong>delay between <strong>the</strong>exit of<strong>the</strong> electr<strong>on</strong> bunchfrom<br />

<strong>the</strong> first <strong>and</strong> entrance of <strong>the</strong> sec<strong>on</strong>d undulator can be varied with <strong>the</strong> magnetic field<br />

strength (as <strong>the</strong> path becomes l<strong>on</strong>ger for str<strong>on</strong>ger magnetic fields since <strong>the</strong> deviati<strong>on</strong><br />

becomes larger). Theintroduceddelaycase aphase-shift of<strong>the</strong>emittedphot<strong>on</strong>swhich<br />

give rise to a (tunable) polarizati<strong>on</strong> change from linear tofully circular polarized light.<br />

Chicane<br />

Figure 12.1: Planar undulators with <strong>the</strong> plane shifted (here 90 ◦ ) can produce light with tilted linear polarizati<strong>on</strong><br />

when <strong>the</strong>y emit in phase <strong>–</strong> or circular polarizati<strong>on</strong> if <strong>the</strong> phase can be shifted with a delay introduced with a<br />

chicane secti<strong>on</strong> between <strong>the</strong> undulators.<br />

It is not necessary for <strong>the</strong> whole undulator array of <strong>the</strong> free electr<strong>on</strong> laser to be<br />

composed of helical undulators. An electr<strong>on</strong> bunch which is microbunched in a planar<br />

undulator emit coherently in a helical undulator as well if <strong>the</strong> undulators are tuned<br />

to <strong>the</strong> same wavelength. C<strong>on</strong>siderable effort can thus be saved in c<strong>on</strong>structing an<br />

undulator array that can produce all kinds of polarizati<strong>on</strong> of <strong>the</strong> light. Most upgrades<br />

to currently operating facilities <strong>and</strong> new facilities have this capacity included in <strong>on</strong>e<br />

way or ano<strong>the</strong>r since <strong>the</strong> experiments enabled are numerous <strong>and</strong> studies phenomena<br />

of general scientific interest.<br />

Coherence & Seeding<br />

Throughout this book we have noted that nei<strong>the</strong>r l<strong>on</strong>gitudinal, nor transverse coherence<br />

of a free electr<strong>on</strong> laser is perfect. This is reflected in <strong>the</strong> spectral b<strong>and</strong>width,<br />

which is inversely proporti<strong>on</strong>al to <strong>the</strong> number of undulator periods. We may improve<br />

up<strong>on</strong> this number with a m<strong>on</strong>ochromator before <strong>the</strong> experiment, trading intensity for<br />

higher m<strong>on</strong>ochromaticity.<br />

Differentup-shiftingschemesinvolvingtraditi<strong>on</strong>allasers <strong>and</strong>HHGlasers, e.g.Hghg,<br />

Eehg <strong>and</strong> o<strong>the</strong>r seeding schemes are an integral part of most new free electr<strong>on</strong><br />

laser projects <strong>–</strong> notably <strong>the</strong> Fermi@Elettra, <strong>the</strong> next free electr<strong>on</strong> laser to come<br />

<strong>on</strong>-line. A significant improvement up<strong>on</strong> <strong>the</strong> coherence properties of <strong>the</strong> free electr<strong>on</strong><br />

laser radiati<strong>on</strong> can be expected from this, especially if <strong>the</strong> seed power dominates <strong>the</strong><br />

initial Sase power. Currently, <strong>the</strong> limit of effective seeding using an HHG laser source<br />

is around 10 nm <strong>–</strong> mainly set by <strong>the</strong> power of <strong>the</strong> driving lasers. Up-shifting schemes<br />

<strong>the</strong>refore often use ordinary lasers because of <strong>the</strong>ir higher repetiti<strong>on</strong> rate, which will<br />

be needed for many newfree electr<strong>on</strong> laser facilites operating in <strong>the</strong> kHz/MHz regi<strong>on</strong>s.<br />

At<strong>the</strong>Lcls aself-seedingopti<strong>on</strong>is includedin<strong>the</strong>upgradeproposal, whichwill use<br />

<strong>the</strong> generated radiati<strong>on</strong> to seed <strong>the</strong> Sase process in <strong>the</strong> following undulator secti<strong>on</strong>s


198 12. Outlook & C<strong>on</strong>clusi<strong>on</strong>s<br />

<strong>–</strong> something which potentially overcomes <strong>the</strong> limits imposed <strong>on</strong> external laser seeding<br />

schemes vis-à-vis wavelength <strong>and</strong> repetiti<strong>on</strong> rate.<br />

The use of low bunch charges reduce <strong>the</strong> emittance of <strong>the</strong> <strong>beam</strong> <strong>and</strong> increases <strong>the</strong><br />

probability of <strong>on</strong>ly <strong>on</strong>e Sase mode radiating efficiently (which improves <strong>the</strong> pulse<br />

length).<br />

Harder X-rays<br />

It is problematic to reduce <strong>the</strong> phot<strong>on</strong>-wavelength fur<strong>the</strong>r than 1 ˚A while maintaining<br />

free electr<strong>on</strong> laser amplificati<strong>on</strong> efficiently <strong>–</strong> <strong>the</strong> main limits being electr<strong>on</strong> <strong>beam</strong><br />

emittance <strong>and</strong> for very hard X-rays <strong>the</strong> recoil up<strong>on</strong> phot<strong>on</strong> emissi<strong>on</strong> smears out <strong>the</strong><br />

microbunching (having an extremely short period as is at those wavelengths). At <strong>the</strong><br />

Lcls a significant porti<strong>on</strong> of <strong>the</strong> density modulati<strong>on</strong> of <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> is present<br />

at half of <strong>the</strong> fundamental wavelength, giving access to sub-˚A wavelengths when <strong>the</strong><br />

first undulator secti<strong>on</strong> is followed by undulators that have <strong>the</strong> sec<strong>on</strong>d harm<strong>on</strong>ic of<br />

<strong>the</strong> radiati<strong>on</strong> as fundamental wavelength. The expected power at 16 keV is about<br />

10% of that found in <strong>the</strong> 8 keV fundamental[303]. A proposal for a free electr<strong>on</strong><br />

laser producing 50 keV X-rays is actively worked up<strong>on</strong>[315].<br />

γ-rays <strong>and</strong> <strong>the</strong> quantum free electr<strong>on</strong> laser<br />

The c<strong>on</strong>cept of <strong>the</strong> quantum free electr<strong>on</strong> laser (QFEL) have been introduced by<br />

R. B<strong>on</strong>ifacio <strong>and</strong> o<strong>the</strong>rs (see [316] <strong>and</strong> references <strong>the</strong>rein). We may check when<br />

<strong>the</strong> phot<strong>on</strong>’s momentum becomes comparable to <strong>the</strong> electr<strong>on</strong> momentum spread in<br />

<strong>the</strong> <strong>beam</strong> using <strong>the</strong> Pierce parameter ρ (see chapter 2). Classically, <strong>the</strong> momentum<br />

spread in <strong>the</strong> electr<strong>on</strong> <strong>beam</strong> is mcγrρ. The phot<strong>on</strong>’s momentum is proporti<strong>on</strong>al to<br />

<strong>the</strong> wavenumber k with � as <strong>the</strong> proporti<strong>on</strong>ality c<strong>on</strong>stant.<br />

Letting <strong>the</strong> ratio between <strong>the</strong> two momenta define <strong>the</strong> ”quantum FEL parameter”<br />

as:<br />

¯ρ = ρ mcγr<br />

�k<br />

in <strong>the</strong> soft X-ray regime <strong>the</strong> wavenumber of <strong>the</strong> phot<strong>on</strong>s is small compared to <strong>the</strong><br />

momentum spread in <strong>the</strong> electr<strong>on</strong> <strong>beam</strong>, for a normal Sase-free electr<strong>on</strong> laser this<br />

means that ¯ρ ≫ 1. For very high momenta <strong>the</strong> phot<strong>on</strong> momentum will have comparable<br />

magnitude to <strong>the</strong> electr<strong>on</strong> momenta. In this limit we must take into account<br />

that <strong>the</strong> momentum recoil can <strong>on</strong>ly assume discrete values (i.e. multiples of �k). If<br />

nothing is d<strong>on</strong>e to synchr<strong>on</strong>ize <strong>the</strong> phot<strong>on</strong> emissi<strong>on</strong> <strong>the</strong> recoils will occur r<strong>and</strong>omly<br />

<strong>and</strong> widen <strong>the</strong> momentum distributi<strong>on</strong> in <strong>the</strong> <strong>beam</strong> with loss of <strong>the</strong> microbunching.<br />

For an electr<strong>on</strong> <strong>beam</strong> with small enough energy spread, <strong>the</strong> quantum free electr<strong>on</strong><br />

laser operate by colliding this electr<strong>on</strong> <strong>beam</strong> with a counterpropagating laser <strong>beam</strong><br />

having very high power 1 . The laser <strong>beam</strong> (which can be operated at µm wavlengths)<br />

<strong>the</strong>n effectively acts as an undulator with extremely short period. If <strong>on</strong>e is in <strong>the</strong><br />

regime where ¯ρ ≤ 1 all electr<strong>on</strong>s experience <strong>the</strong> same recoil losing �k. The radiati<strong>on</strong><br />

from this process is temporally coherent up into γ-ray energies.<br />

Experiments with this type of radiati<strong>on</strong> may extend <strong>the</strong> applicability of free electr<strong>on</strong><br />

laser (in a broad sense) into <strong>the</strong> sub-atomic regimes with <strong>the</strong> study of processes<br />

involving, e.g. nuclear transiti<strong>on</strong>s.<br />

1 For <strong>the</strong> so inclined, <strong>the</strong> process is that of collective Compt<strong>on</strong> backscattering.


12.2. C<strong>on</strong>clusi<strong>on</strong> 199<br />

A X-ray free-prot<strong>on</strong> laser have been suggested as a way to overcome <strong>the</strong> ”quantum<br />

momentumlimitati<strong>on</strong>”of<strong>the</strong>Saseprocess, thisbyusingprot<strong>on</strong>sra<strong>the</strong>rthanelectr<strong>on</strong>s<br />

in <strong>the</strong> particle <strong>beam</strong> <strong>–</strong> although feasible, it would require a huge scale facility such<br />

as <strong>the</strong> LHC as a driver[317]. As noted in <strong>the</strong> reference though, this may be tested<br />

as a parasitic undertaking at <strong>the</strong> storage ring, much as synchrotr<strong>on</strong> radiati<strong>on</strong> studies<br />

started out in <strong>the</strong> late 1940’s.<br />

12.2 C<strong>on</strong>clusi<strong>on</strong><br />

Free electr<strong>on</strong> laser operating in <strong>the</strong> VUV <strong>and</strong> X-ray ranges offer scientists a light<br />

source with unique spectral properties. The success of Flash <strong>and</strong> Lcls <strong>and</strong> <strong>the</strong><br />

pi<strong>on</strong>eering experiments carried out <strong>the</strong>re mark <strong>the</strong> potential for a broad <strong>and</strong> deep<br />

endeavor to investigate nature in ways hi<strong>the</strong>rto not possible with X-rays using that<br />

degree of precisi<strong>on</strong>: nanometers <strong>and</strong> femtosec<strong>on</strong>ds with developments pushing those<br />

limits towards ˚Angströms <strong>and</strong> attosec<strong>on</strong>ds <strong>–</strong> <strong>the</strong> true atomic scales of space <strong>and</strong> time.<br />

With <strong>the</strong> advent of successful improvements of <strong>the</strong> spectral characteristics of <strong>the</strong><br />

Sase experimental work <strong>and</strong> interpretati<strong>on</strong> of data will become easier <strong>and</strong> facilitate<br />

<strong>the</strong> utilizati<strong>on</strong> of this type of source for a broader user community.<br />

As <strong>the</strong> first generati<strong>on</strong> of X-ray free electr<strong>on</strong> laser is augmented, upgraded <strong>and</strong> <strong>the</strong><br />

sec<strong>on</strong>d generati<strong>on</strong> of facilities are brought <strong>on</strong>-line in <strong>the</strong> future <strong>the</strong> number of possible<br />

experiments will surely increase as well as <strong>the</strong> intricacy of <strong>the</strong> investigati<strong>on</strong>s.


Bibliography<br />

[1] J. C. Maxwell. A dynamical <strong>the</strong>ory of <strong>the</strong> electromagnetic field. Philos. Trans.<br />

R. Soc. L<strong>on</strong>d<strong>on</strong>, 155:459, 1865.<br />

[2] H. Hertz. Über einen einfluss des ultravioletten lichtes auf die electrische entladung.<br />

Ann. Phys., 267(8):983<strong>–</strong>1000, 1887.<br />

[3] A. Einstein. Über einen die erzeugung un verw<strong>and</strong>lung des lichtes betreffenden<br />

heuristischen gesichtspunkt. Annalen der Physik, IV(17):132, 1905.<br />

[4] P. Larsen <strong>and</strong> M. v<strong>on</strong> Ins. The rate of growth in scientific publicati<strong>on</strong> <strong>and</strong> <strong>the</strong><br />

decline in coverage provided by science citati<strong>on</strong> index. Scientometrics, 84:575<strong>–</strong><br />

603, 2010.<br />

[5] F.R. Elder, A. M. Gurewitsch, R. V. Langmuir, <strong>and</strong> H. C. Pollock. Radiati<strong>on</strong><br />

from electr<strong>on</strong>s in a synchrotr<strong>on</strong>. Phys. Rev., 71(11):829<strong>–</strong>830, Jun 1947.<br />

[6] H. Motz. Applicati<strong>on</strong>s of <strong>the</strong> radiati<strong>on</strong> from fast electr<strong>on</strong> <strong>beam</strong>s. J Appl. Phys.,<br />

22(5), 1951.<br />

[7] A.Einstein. Zurelektrodynamikbewegter körper. Physik. Zeitschr., XVIII:121<strong>–</strong><br />

128, 1917.<br />

[8] M. Planck. Ueber das gesetz der energieverteilung im normalspectrum. Annalen<br />

der Physik, 309:553<strong>–</strong>563, 1901.<br />

[9] T. H.Maiman. Stimulated optical radiati<strong>on</strong> inruby. Nature, 187(4736):493<strong>–</strong>494,<br />

1960.<br />

[10] P. A. Franken, A. E. Hill, C. W. Peters, <strong>and</strong> G. Weinreich. Generati<strong>on</strong> of optical<br />

harm<strong>on</strong>ics. Phys. Rev. Lett., 7(4):118<strong>–</strong>119, 1961.<br />

[11] G. H. C. New <strong>and</strong> J. F. Ward. Optical third-harm<strong>on</strong>ic generati<strong>on</strong> in gases.<br />

Phys. Rev. Lett., 19(10):556<strong>–</strong>559, 1967.<br />

[12] P. B. Corkum. Plasma perspective <strong>on</strong> str<strong>on</strong>g field multiphot<strong>on</strong> i<strong>on</strong>izati<strong>on</strong>. Phys.<br />

Rev. Lett., 71(13):1994<strong>–</strong>1997, 1993.<br />

[13] J. M. J. Madey. Stimulated emissi<strong>on</strong> of bremsstrahlung in a periodic magnetic<br />

field. J. Appl. Phys., 42:1906<strong>–</strong>1913, Apr 1971.<br />

201


202 Bibliography<br />

[14] D. A. G. Deac<strong>on</strong>, L. R. Elias, J. M. J. Madey, G. J. Ramian, H. A. Schwettman,<br />

<strong>and</strong> T. I. Smith. First operati<strong>on</strong> of a free-electr<strong>on</strong> laser. Phys. Rev. Lett.,<br />

38(16):892<strong>–</strong>894, Apr 1977.<br />

[15] P. G. O’Shea <strong>and</strong> H. P. Freund. Free-Electr<strong>on</strong> Lasers: Status <strong>and</strong> Applicati<strong>on</strong>s.<br />

Science, 292(5523):1853<strong>–</strong>1858, 2001.<br />

[16] C. Pellegrini <strong>and</strong> S. Reiche. The development of x-ray free-electr<strong>on</strong> lasers. IEEE<br />

J. Sel. Top. Quant., 10(6):1393 <strong>–</strong> 1404, 2004.<br />

[17] Z. Huang <strong>and</strong> K.-J. Kim. Review of x-ray free-electr<strong>on</strong> laser <strong>the</strong>ory. Phys. Rev.<br />

ST Accel. Beams, 10(3):034801, Mar 2007.<br />

[18] E L Saldin, E A Schneidmiller, <strong>and</strong> M V Yurkov. Statistical <strong>and</strong> coherence<br />

properties of radiati<strong>on</strong> from x-ray free-electr<strong>on</strong> lasers. New Journal of Physics,<br />

12(3):035010, 2010.<br />

[19] B. McNeil. Free electr<strong>on</strong> lasers: First light from hard x-ray laser. Nat. Phot<strong>on</strong>,<br />

3(7):375<strong>–</strong>377, 2009.<br />

[20] A. A. Zholents. Method of an enhanced self-amplified sp<strong>on</strong>taneous emissi<strong>on</strong> for<br />

x-ray free electr<strong>on</strong> lasers. Phys. Rev. ST Accel. Beams, 8(4):040701, 2005.<br />

[21] L.-H. Yu, M. Babzien, I. Ben-Zvi, L. F. DiMauro, A. Doyuran, W. Graves,<br />

E. Johns<strong>on</strong>, S. Krinsky, R. Mal<strong>on</strong>e, I. Pogorelsky, J. Skaritka, G. Rakowsky,<br />

L. Solom<strong>on</strong>, X. J. Wang, M. Woodle, V. Yakimenko, S. G. Biedr<strong>on</strong>, J. N.<br />

Galayda, E. Gluskin, J. Jagger, V. Sajaev, <strong>and</strong> I. Vasserman. High-Gain<br />

Harm<strong>on</strong>ic-Generati<strong>on</strong> Free-Electr<strong>on</strong> Laser. Science, 289(5481):932<strong>–</strong>934, 2000.<br />

[22] A. Doyuran, M. Babzien, T. Shaftan, L. H. Yu, L. F. DiMauro, I. Ben-Zvi, S. G.<br />

Biedr<strong>on</strong>, W. Graves, E. Johns<strong>on</strong>, S. Krinsky, R. Mal<strong>on</strong>e, I. Pogorelsky, J. Skaritka,<br />

G. Rakowsky, X. J. Wang, M. Woodle, V. Yakimenko, J. Jagger, V. Sajaev,<br />

<strong>and</strong> I. Vasserman. Characterizati<strong>on</strong> of a high-gain harm<strong>on</strong>ic-generati<strong>on</strong><br />

free-electr<strong>on</strong> laser at saturati<strong>on</strong>. Phys. Rev. Lett., 86(26):5902<strong>–</strong>5905, Jun 2001.<br />

[23] Dao Xiang <strong>and</strong> Gennady Stupakov. Echo-enabled harm<strong>on</strong>ic generati<strong>on</strong> free<br />

electr<strong>on</strong> laser. Phys. Rev. ST Accel. Beams, 12(3):030702, Mar 2009.<br />

[24] D. Xiang, E. Colby, M. Dunning, S. Gilevich, C. Hast, K. Jobe, D. McCormick,<br />

J. Nels<strong>on</strong>, T. O. Raubenheimer, K. So<strong>on</strong>g, G. Stupakov, Z. Szalata, D. Walz,<br />

S. Wea<strong>the</strong>rsby, <strong>and</strong> M. Woodley. Dem<strong>on</strong>strati<strong>on</strong> of <strong>the</strong> echo-enabled harm<strong>on</strong>ic<br />

generati<strong>on</strong> techniquefor short-wavelengthseeded free electr<strong>on</strong> lasers. Phys. Rev.<br />

Lett., 105(11):114801, 2010.<br />

[25] E.Schneidmiller<strong>and</strong>M. Yurkov. Anopti<strong>on</strong>offrequencydoublerat<strong>the</strong>european<br />

xfelfor generati<strong>on</strong> ofcircularly polarized radiati<strong>on</strong> in<strong>the</strong>wavelengthrangedown<br />

to 1 - 2.5 nm. In Proc. FEL2010, Aug. 2010. Malmö, Sweden.<br />

[26] K.-J. Kim. Circular polarizati<strong>on</strong> with crossed-planar undulators in high-gain<br />

fels. Nucl. Instrum. Meth. A, 445(1-3):329 <strong>–</strong> 332, 2000.<br />

[27] H. Geng, Y. Ding, <strong>and</strong> Z. Huang. Crossed undulator polarizati<strong>on</strong> c<strong>on</strong>trol for<br />

x-ray fels in <strong>the</strong> saturati<strong>on</strong> regime. Nucl. Instrum. Meth. A, 622(1):276 <strong>–</strong> 280,<br />

2010.


Bibliography 203<br />

[28] D. M. Mills, J. R. Helliwell, ˚A. Kvick, T. Ohta, I. A. Robins<strong>on</strong>, <strong>and</strong> A. Authier.<br />

Brightness, spectral brightness or brilliance <strong>–</strong> Report of <strong>the</strong> Working Group <strong>on</strong><br />

Synchrotr<strong>on</strong> Radiati<strong>on</strong> Nomenclature. J. Synchr. Rad., 12(3):385, 2005.<br />

[29] A. M. Sessler. Accelerator <strong>beam</strong> emittance. Am. J. Phys., 60(8):760<strong>–</strong>762, 1992.<br />

[30] P. G. O’Shea. Reversible <strong>and</strong> irreversible emittance growth. Phys. Rev. E,<br />

57(1):1081<strong>–</strong>1087, 1998.<br />

[31] G. Margarit<strong>on</strong>do <strong>and</strong> P. R. Rebernik. A simplified descripti<strong>on</strong> of X-ray freeelectr<strong>on</strong><br />

lasers. J. Synchr. Rad., 18(2), Mar 2011.<br />

[32] J. D. Jacks<strong>on</strong>. Classical Electrodynamics. Wiley, 3rd editi<strong>on</strong>, 1998.<br />

[33] J. Schwinger. On<strong>the</strong>classical radiati<strong>on</strong> ofaccelerated electr<strong>on</strong>s. Physical review,<br />

75(12):1912, 1949.<br />

[34] Richard Feynman, Robert Leight<strong>on</strong>, <strong>and</strong> Mat<strong>the</strong>w S<strong>and</strong>s. The Feynman Lectures<br />

<strong>on</strong> Physics: Volume 1, volume 1 of The Feynman Lectures <strong>on</strong> Phsyics.<br />

Addis<strong>on</strong>-Wesley, Bost<strong>on</strong>, 2nd editi<strong>on</strong> editi<strong>on</strong>, 1963.<br />

[35] Richard Feynman. The Feynman Lectures <strong>on</strong> Physics: Volume 2, volume 2 of<br />

The Feynman Lectures <strong>on</strong> Phsyics. Addis<strong>on</strong>-Wesley, Bost<strong>on</strong>, 1963.<br />

[36] Joseph Larmor. A dynamical <strong>the</strong>ory of <strong>the</strong> electric <strong>and</strong> luminiferous medium.<br />

part iii. relati<strong>on</strong>s with material media. Phil. Trans. R. Soc. L<strong>on</strong>d. A, 190:205<strong>–</strong><br />

493, 1897.<br />

[37] Helmut Wiedemann. Particle Accelerator Physics: Volume I <strong>and</strong> II. Springer,<br />

2004.<br />

[38] R. Aloisio <strong>and</strong> P. Blasi. Theory of synchrotr<strong>on</strong> radiati<strong>on</strong>: I. coherent emissi<strong>on</strong><br />

from ensembles of particles. Astropart. Phys., 18(2):183 <strong>–</strong> 193, 2002.<br />

[39] H. Goldstein. Classical Mechanics. Addis<strong>on</strong>-Wesley, 2nd editi<strong>on</strong>, 1980.<br />

[40] B. Thidé. Electromagnetic field <strong>the</strong>ory. 2nd editi<strong>on</strong>, 2010.<br />

[41] R. B<strong>on</strong>ifacio, C. Pellegrini, <strong>and</strong> L.M. Narducci. Collective instabilities <strong>and</strong><br />

high-gain regime in a free electr<strong>on</strong> laser. Opt. Comm., 50(6):373 <strong>–</strong> 378, 1984.<br />

[42] E.V. Schneidmiller E.L. Saldin <strong>and</strong> M.V. Yurkov. The Physics of Free Electr<strong>on</strong><br />

Lasers. Springer-Verlag, 1st editi<strong>on</strong> editi<strong>on</strong>, 2000.<br />

[43] Helmut Wiedemann. Particle Accelerator Physics. Springer, 3 rd ed. editi<strong>on</strong>,<br />

2007.<br />

[44] R. B<strong>on</strong>ifacio, F. Casagr<strong>and</strong>e, G. Cerchi<strong>on</strong>i, L. de Salvo Souza, P. Pierini, <strong>and</strong><br />

N. Piovella. Physics of <strong>the</strong> high-gain fel <strong>and</strong> superradiance. La Rivista del<br />

Nuovo Cimento, 13:1<strong>–</strong>69, 1990.<br />

[45] E. L. Saldin, E. A. Schneidmiller, <strong>and</strong> M. V. Yurkov. Statistical <strong>and</strong> coherence<br />

properties of radiati<strong>on</strong> from x-ray free-electr<strong>on</strong> lasers. New J. Phys.,<br />

12(3):035010, 2010.


204 Bibliography<br />

[46] G. Gel<strong>on</strong>i, E. Saldin, L.Samoylova, E.Schneidmiller, H.Sinn, Th.Tschentscher,<br />

<strong>and</strong> M. Yurkov. Coherence properties of <strong>the</strong> european xfel. New J. Phys.,<br />

12(3):035021, 2010.<br />

[47] A. Singer, I. A. Vartanyants, M. Kuhlmann, S. Duesterer, R. Treusch, <strong>and</strong><br />

J. Feldhaus. Transverse-coherence properties of <strong>the</strong> free-electr<strong>on</strong>-laser flash at<br />

desy. Phys. Rev. Lett., 101(25):254801, 2008.<br />

[48] I. A.Vartanyants<strong>and</strong>A. Singer. Coherence properties ofhardx-raysynchrotr<strong>on</strong><br />

sources <strong>and</strong> x-ray free-electr<strong>on</strong> lasers. New J. Phys., 12(3):035004, 2010.<br />

[49] B. D. Patters<strong>on</strong>, R. Abela, H.-H. Braun, U. Flechsig, R. Ganter, Y. Kim,<br />

E. Kirk, A. Oppelt, M. Pedrozzi, S. Reiche, L. Rivkin, Th. Schmidt, B. Schmitt,<br />

V. N. Strocov, S. Tsujino, <strong>and</strong> A. F. Wrulich. Coherent science at <strong>the</strong> swissfel<br />

x-ray laser. New J. Phys., 12(3):035012, 2010.<br />

[50] I. A. Vartanyants, A. P. Mancuso, A. Singer, O. M. Yefanov, <strong>and</strong> J. Gulden.<br />

Coherence measurements <strong>and</strong> coherent diffractive imaging at flash. J. Phys. B:<br />

At. Mol. Opt. Phys., 43(19):194016, 2010.<br />

[51] J.B. Rosenzweig, D. Alesini, G. And<strong>on</strong>ian, M. Boscolo, M. Dunning, L. Faillace,<br />

M. Ferrario, A. Fukusawa, L. Giannessi, E. Hemsing, G. Marcus, A. Marinelli,<br />

P. Musumeci, B. O’Shea, L. Palumbo, C. Pellegrini, V. Petrillo, S. Reiche,<br />

C. R<strong>on</strong>sivalle, B. Spataro, <strong>and</strong> C. Vaccarezza. Generati<strong>on</strong> of ultra-short, high<br />

brightness electr<strong>on</strong> <strong>beam</strong>s for single-spike sase fel operati<strong>on</strong>. Nucl. Instrum.<br />

Meth. A, 593(1-2):39 <strong>–</strong> 44, 2008.<br />

[52] J. Feldhaus, E. L. Saldin, J. R. Schneider, E. A. Schneidmiller, <strong>and</strong> M. V.<br />

Yurkov. Possible applicati<strong>on</strong> of x-ray optical elements for reducing <strong>the</strong> spectral<br />

b<strong>and</strong>width of an x-ray sase fel. Opt. Comm., 140(4-6):341 <strong>–</strong> 352, 1997.<br />

[53] S. Reiche. Genesis 1.3: a fully 3d time-dependent fel simulati<strong>on</strong> code. Nucl.<br />

Instrum. Meth. A, 429(1-3):243 <strong>–</strong> 248, 1999.<br />

[54] E. L. Saldin, E. A. Schneidmiller, <strong>and</strong> M. V. Yurkov. Fast: a three-dimensi<strong>on</strong>al<br />

time-dependent fel simulati<strong>on</strong> code. Nucl. Instrum. Meth. A, 429(1-3):233 <strong>–</strong><br />

237, 1999.<br />

[55] S. Reiche, C. Pellegrini, J. Rosenzweig, P. Emma, <strong>and</strong> P. Krejcik. Start-to-end<br />

simulati<strong>on</strong> for <strong>the</strong> lcls x-ray fel. Nucl. Instrum. Meth. A, 483(1-2):70 <strong>–</strong> 74, 2002.<br />

[56] K. Togawa, H. Baba, K. Onoe, T. Inagaki, T. Shintake, <strong>and</strong> H. Matsumoto.<br />

Ceb6 electr<strong>on</strong> gun for <strong>the</strong> soft x-ray fel project at spring-8. Nucl. Instrum.<br />

Meth. A, 528(1-2):312 <strong>–</strong> 315, 2004.<br />

[57] H. Hanaki, T. Asaka, H. Ego, H. Kimura, T. Kobayashi, S. Suzuki, T. Hara,<br />

A. Higashiya, T. Inagaki, N. Kumagai, H. Maesaka, Y. Otake, T. Shintake, <strong>and</strong><br />

H. Tanaka. Injector system for x-ray fel at spring-8. In Proc. of EPAC08, 2008.<br />

Genoa, Italy.<br />

[58] K. L. Jensen, P. G. O’Shea, D. W. Feldman, <strong>and</strong> J. L. Shay. Emittance of a<br />

field emissi<strong>on</strong> electr<strong>on</strong> source. J. Appl. Phys., 107:014903, 2010.


Bibliography 205<br />

[59] L. Serafini <strong>and</strong> J. B. Rosenzweig. Envelope analysis of intense relativistic quasilaminar<br />

<strong>beam</strong>s in rf photoinjectors:ma <strong>the</strong>ory of emittance compensati<strong>on</strong>. Phys.<br />

Rev. E, 55(6):7565<strong>–</strong>7590, 1997.<br />

[60] M. Ferrario. Overview of fel injectors. In Proc. EPAC 2006, June 2006. Edinburgh,<br />

Scotl<strong>and</strong>.<br />

[61] R. Ganter, R. Bakker, C. Gough, S. C. Leemann, M. Paraliev, M. Pedrozzi,<br />

F. Le Pimpec, V. Schlott, L. Rivkin, <strong>and</strong> A. Wrulich. Laser-photofield emissi<strong>on</strong><br />

from needle cathodes for low-emittance electr<strong>on</strong> <strong>beam</strong>s. Phys. Rev. Lett.,<br />

100(6):064801, Feb 2008.<br />

[62] J. E. Clendenin, T. Kotseroglou, G. A. Mulhollan, D. T. Palmer, <strong>and</strong> J. F.<br />

Schmerge. Reducti<strong>on</strong> of <strong>the</strong>rmal emittance of rf guns. Nucl. Instrum. Meth. A,<br />

455(1):198 <strong>–</strong> 201, 2000.<br />

[63] K. L. Jensen, P.G. O’Shea, <strong>and</strong> D. W. Feldman. Emittance of a photocathode:<br />

Effects of temperature <strong>and</strong> field. Phys. Rev. ST Accel. Beams, 13(8):080704,<br />

2010.<br />

[64] V.G. Tkachenko · A.I. K<strong>on</strong>drashev · I.N. Maksimchuk. Advanced metal alloy<br />

systems for massive high-current photocathodes. Appl. Phys. B, 98:839<strong>–</strong>849,<br />

2010,.<br />

[65] J. R. Mald<strong>on</strong>ado, Z. Liu, D. H. Dowell, R. E. Kirby, Y. Sun, P. Pianetta, <strong>and</strong><br />

F. Pease. Robust csbr/cu photocathodes for <strong>the</strong> linac coherent light source.<br />

Phys. Rev. ST Accel. Beams, 11(6):060702, 2008.<br />

[66] D.H. Dowell, I.Bazarov, B. Dunham, K. Harkay, C. Hern<strong>and</strong>ez-Garcia, R.Legg,<br />

H. Padmore, T. Rao, J. Smedley, <strong>and</strong> W. Wan. Cathode r&d for future light<br />

sources. Nucl. Instrum. Meth. A, 622(3):685 <strong>–</strong> 697, 2010.<br />

[67] J. D. Cockcroft <strong>and</strong> E. T. S. Walt<strong>on</strong>. Experiments with High Velocity Positive<br />

I<strong>on</strong>s. II. The Disintegrati<strong>on</strong> of Elements by High Velocity Prot<strong>on</strong>s. Proc. Roy.<br />

Soc. A, 137(831):229<strong>–</strong>242, 1932.<br />

[68] P. R. Bolt<strong>on</strong>, J. E. Clendenin, D. H. Dowell, M. Ferrario, A. S. Fisher, S. M.<br />

Gierman, R. E. Kirby, P. Krejcik, C. G. Limborg, G. A. Mulhollan, D. Nguyen,<br />

D. T. Palmer, J. B. Rosenzweig, J. F. Schmerge, L. Serafini, <strong>and</strong> X. J. Wang.<br />

Photoinjector design for <strong>the</strong> lcls. Nucl. Instrum. Meth. A, 483(1-2):296 <strong>–</strong> 300,<br />

2002.<br />

[69] K.Baptiste, J.Corlett, S.Kwiatkowski, S.Lidia, J.Qiang, F. Sannibale, K.S<strong>on</strong>nad,<br />

J. Staples, S. Virostek, <strong>and</strong> R. Wells. A cw normal-c<strong>on</strong>ductive rf gun for<br />

free electr<strong>on</strong> laser <strong>and</strong> energy recovery linac applicati<strong>on</strong>s. Nucl. Instrum. Meth.<br />

A, 599(1):9 <strong>–</strong> 14, 2009.<br />

[70] A. Arnold, H. Büttig, D. Janssen, T. Kamps, G. Klemz, W.D. Lehmann,<br />

U. Lehnert, D. Lipka, F. Marhauser, P. Michel, K. Möller, P. Murcek, Ch.<br />

Schneider, R. Schurig, F. Staufenbiel, J. Stephan, J. Teichert, V. Volkov, I. Will,<br />

<strong>and</strong> R. Xiang. Development of a superc<strong>on</strong>ducting radio frequency photoelectr<strong>on</strong><br />

injector. Nucl. Instrum. Meth. A, 577(3):440 <strong>–</strong> 454, 2007.


206 Bibliography<br />

[71] J. Teichert A. Arnold. Overview of superc<strong>on</strong>ducting photo injectors. In Proc.<br />

SRF 2009 C<strong>on</strong>ference, Sept. 2009. Berlin, Germany.<br />

[72] V. Volkov D. Janssen. Emittance compensati<strong>on</strong> in a superc<strong>on</strong>ducting rf photoelectr<strong>on</strong><br />

gun by a magnetic rf field. In Proc. EPAC 2004, July 2004.<br />

Lucerne,Switzerl<strong>and</strong>.<br />

[73] K. Flöttmann, D. Janssen, <strong>and</strong> V. Volkov. Emittance compensati<strong>on</strong> in a superc<strong>on</strong>ducting<br />

rf gun with a magnetic mode. Phys. Rev. ST Accel. Beams,<br />

7(9):090702, 2004.<br />

[74] W.A. Barletta, J. Bisognano, J.N. Corlett, P. Emma, Z. Huang, K.-J. Kim,<br />

R. Lindberg, J.B. Murphy, G.R. Neil, D.C. Nguyen, C. Pellegrini, R.A. Rimmer,<br />

F. Sannibale, G. Stupakov,R.P. Walker, <strong>and</strong>A.A.Zholents. Freeelectr<strong>on</strong> lasers:<br />

Present status <strong>and</strong> future challenges. Nucl. Instrum. Meth. A, 618(1-3):69 <strong>–</strong> 96,<br />

2010.<br />

[75] G. Ising. Arkiv för Matematik, Astr<strong>on</strong>omi och Fysik, 18:1, 1924.<br />

[76] R. Wiederöe. Arch. Elektrotech., 21:387, 1928.<br />

[77] S. Humphries. Principles of charged particle accelerati<strong>on</strong>, 1999.<br />

[78] V. A. Dolgashev, S.G. Tatawi, A.D. Yeremian, Y. Higashi, <strong>and</strong> B. Spataro.<br />

Status of high power tests of normal c<strong>on</strong>ducting single-cell st<strong>and</strong>ing wave structures.<br />

In Proc. IPAC’10, May 2010. Koyoto, Japan.<br />

[79] H. Padamsee. Rf Superc<strong>on</strong>ductivity::Science, Technology, <strong>and</strong> Applicati<strong>on</strong>s.<br />

Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2009.<br />

[80] H. Hayano. Progress <strong>and</strong> plans for r&d <strong>and</strong> <strong>the</strong> c<strong>on</strong>ceptual design of <strong>the</strong> ilc<br />

main linacs. In Proc of 2005 Part. Acc. C<strong>on</strong>f., 2005. Knoxville, Tenessee, USA.<br />

[81] B. Aune, R. B<strong>and</strong>elmann, D. Bloess, B. B<strong>on</strong>in, A. Bosotti, M. Champi<strong>on</strong>,<br />

C. Crawford, G. Deppe, <strong>and</strong> et. al. Superc<strong>on</strong>ducting tesla cavities. Phys. Rev.<br />

ST Accel. Beams, 3(9):092001, Sep 2000.<br />

[82] F. Furuta, K. Saito, T. Saeki, H. Inoue, Y. Morozumi, T. Higo, Y. Higashi,<br />

H. Matsumoto, S. Kazakov, H. Yamaoka, K. Ueno, Y. Kobayashi, R.S. Orr, <strong>and</strong><br />

J. Sekutowicz. Experimental comparis<strong>on</strong> at kek of high gradient performance of<br />

different single cell superc<strong>on</strong>ducting cavity designs. In Proc. EPAC 2006, 2006.<br />

Edinburgh, Scotl<strong>and</strong>.<br />

[83] O. Kugeler, A. Neumann, W. Anders, <strong>and</strong> J. Knobloch. Adapting tesla technology<br />

for future cw light sources using hobicat. Rev. Sci. Instrum., 81(7):074701,<br />

2010.<br />

[84] S. Hashimoto <strong>and</strong> S. Sasaki. C<strong>on</strong>cept of a new undulator that will suppress <strong>the</strong><br />

rati<strong>on</strong>al harm<strong>on</strong>ics. Nucl. Instrum. Meth. A, 361(3):611 <strong>–</strong> 622, 1995.<br />

[85] V. A. Papadichev. Polarizati<strong>on</strong> in free electr<strong>on</strong> lasers. Nucl. Instrum. Meth. A,<br />

375(1-3):483 <strong>–</strong> 486, 1996.


Bibliography 207<br />

[86] S. Sasaki, K. Miyata, <strong>and</strong> T. Takada. A new undulator for generating variably<br />

polarized radiati<strong>on</strong>. Jap. J. Appl. Phys., 31(Part 2, No. 12B):L1794<strong>–</strong>L1796,<br />

1992.<br />

[87] Shigemi Sasaki. Analyses for a planar variably-polarizing undulator. Nucl.<br />

Instr. <strong>and</strong> Meth. A, 347(1-3):83 <strong>–</strong> 86, 1994.<br />

[88] J. Bahrdt, W. Frentrup, A. Gaupp, M. Scheer, <strong>and</strong> U. Englisch. Magnetic field<br />

optimizati<strong>on</strong> of permanent magnet undulators for arbitrary polarizati<strong>on</strong>. Nucl.<br />

Instr. <strong>and</strong> Meth. A, 516(2-3):575 <strong>–</strong> 585, 2004.<br />

[89] M. Altarelli et al. XFEL: The European X-Ray Free-Electr<strong>on</strong> Laser. Technical<br />

Design Report. DESY, Hamburg, 2006. http://xfel.desy.de.<br />

[90] A. Gibaud <strong>and</strong> S. Hazra. X-ray reflectivity <strong>and</strong> diffuse scattering. Curr. Sci.,<br />

78(12):1467<strong>–</strong>1477, 2000.<br />

[91] E. L. Church <strong>and</strong> P. Z. Takacs. Specificati<strong>on</strong> of glancing- <strong>and</strong> normal-incidence<br />

x-ray mirrors [also erratum 34(11)3348(nov1995)]. Opt. Eng., 34(2):353<strong>–</strong>360,<br />

1995.<br />

[92] F. Siewert, J. Buchheim, T. Zeschke, G. Brenner, S. Kapitzki, <strong>and</strong> K. Tiedtke.<br />

Sub-nm accuracy metrology for ultra-precise reflective x-ray optics. Nucl. Instrum.<br />

Meth. A, In Press, Corrected Proof:<strong>–</strong>, 2010.<br />

[93] S.P. Hau-Riege<strong>and</strong>H.N.Chapman. Reflecti<strong>on</strong> ofattosec<strong>on</strong>d x-rayfree electr<strong>on</strong><br />

laser pulses. Rev. Sci. Instrum., 78(1), 2007.<br />

[94] J. I. Larruquert, A. G. Michette, Ch. Morawe, C. H. Borel, <strong>and</strong> B. Vidal.<br />

Specially Designed Multilayers. Springer, 2008.<br />

[95] M. Idir, P. Mercere, M. H. Modi, G. Dovillaire, X. Levecq, S. Bucourt, L. Escolano,<br />

<strong>and</strong> P. Sauvageot. X-ray active mirror coupled with a hartmann wavefr<strong>on</strong>t<br />

sensor. Nucl. Instrum. Meth. A, 616(2-3):162 <strong>–</strong> 171, 2010.<br />

[96] S. Moeller, J. Arthur, A. Brachmann, R. Coffee, F.-J. Decker, Y. Ding, D. Dowell,<br />

S. Edstrom, P. Emma, Y. Feng, A. Fisher, J. Frisch, J. Galayda, S. Gilevich,<br />

J. Hastings, G. Hays, P. Hering, Z. Huang, R. Ivers<strong>on</strong>, J. Krzywinski,<br />

S. Lewis, H. Loos, M. Messerschmidt, A. Miahnahri, H.-D. Nuhn, D. Ratner,<br />

J. Rzepiela, D. Schultz, T. Smith, P. Stefan, H. Tompkins, J. Turner, J. Welch,<br />

B. White, J. Wu, G. Yocky, R. Bi<strong>on</strong>ta, E. Ables, B. Abraham, C. Gardener,<br />

K. F<strong>on</strong>g, S. Friedrich, S. Hau-Riege, K. Kishiyama, T. McCarville, D. McMah<strong>on</strong>,<br />

M. McKernan, L. Ott, M. Pivovaroff, J. Robins<strong>on</strong>, D. Ryutov, S. Shen,<br />

R. Soufli, <strong>and</strong> G. Pile. Phot<strong>on</strong> <strong>beam</strong>lines <strong>and</strong> diagnostics at lcls. Nucl. Instrum.<br />

Meth. A, In Press, Uncorrected Proof, 2010.<br />

[97] T. Feigl, S. Yulin, N. Benoit, <strong>and</strong> N. Kaiser. Euv multilayer optics. Microelectr<strong>on</strong>.<br />

Eng., 83(4-9):703 <strong>–</strong> 706, 2006.<br />

[98] M. Richter, S. V. Bobashev, A. A. Sorokin, <strong>and</strong> K. Tiedtke. Multiphot<strong>on</strong><br />

i<strong>on</strong>izati<strong>on</strong> of atoms with soft x-ray pulses. J. Phys. B: At. Mol. Opt. Phys.,<br />

43(19):194005, 2010.


208 Bibliography<br />

[99] A. R. Khors<strong>and</strong>, R.Sobierajski, E. Louis, S. Bruijn, E. D. van Hattum, R. W. E.<br />

van de Kruijs, M. Jurek, D. Klinger, J. B. Pelka, L. Juha, T. Burian, J. Chalupsky,<br />

J. Cihelka, V. Hajkova, L. Vysin, U. Jastrow, N. Stojanovic, S. Toleikis,<br />

H. Wabnitz, K. Tiedtke, K. Sokolowski-Tinten, U. Shymanovich, J. Krzywinski,<br />

S. Hau-Riege, R. L<strong>on</strong>d<strong>on</strong>, A. Glees<strong>on</strong>, E. M. Gulliks<strong>on</strong>, <strong>and</strong> F. Bijkerk.<br />

Single shot damage mechanism of mo/si multilayer optics under intense pulsed<br />

xuv-exposure. Opt. Express, 18(2):700<strong>–</strong>712, 2010.<br />

[100] J. Kuba, A. Woott<strong>on</strong>, R. M. Bi<strong>on</strong>ta, R. Shepherd, E. E. Fill, T. Ditmire,<br />

G. Dyer, R. A. L<strong>on</strong>d<strong>on</strong>, V. N. Shlyaptsev, J. Dunn, R. Booth, S. Bajt, R. F.<br />

Smith, M. D. Feit, R. Levesque, <strong>and</strong> M. McKernan. X-ray optics research for<br />

free electr<strong>on</strong> lasers: study of material damage under extreme fluxes. Nucl.<br />

Instrum. Meth. A, 507(1-2):475 <strong>–</strong> 478, 2003.<br />

[101] S. P. Hau-Riege, R. A. L<strong>on</strong>d<strong>on</strong>, R. M. Bi<strong>on</strong>ta, R. Soufli, D. Ryutov, M. Shirk,<br />

S. L. Baker, P. M. Smith, <strong>and</strong> P. Nataraj. Multiple pulse <strong>the</strong>rmal damage<br />

thresholds of materials for x-ray free electr<strong>on</strong> laser optics investigated with an<br />

ultraviolet laser. Appl. Phys. Lett., 93(20):201105, 2008.<br />

[102] S. P. Hau-Riege, R. A. L<strong>on</strong>d<strong>on</strong>, H. N. Chapman, <strong>and</strong> M. Bergh. Soft-x-ray<br />

free-electr<strong>on</strong>-laser interacti<strong>on</strong> with materials. Phys. Rev. E, 76(4):046403, 2007.<br />

[103] S. P. Hau-Riege, R. A. L<strong>on</strong>d<strong>on</strong>, A. Graf, S. L. Baker, R. Soufli, R. Sobierajski,<br />

T. Burian, J. Chalupsky, L. Juha, J. Gaudin, J. Krzywinski, S. Moeller,<br />

M. Messerschmidt, J. Bozek, <strong>and</strong> C. Bostedt. Interacti<strong>on</strong> of short x-ray pulses<br />

with low-z x-ray optics materials at <strong>the</strong> lcls free-electr<strong>on</strong> laser. Opt. Express,<br />

18(23):23933—23938, Nov 2010.<br />

[104] K. Tiedtke, A. Azima, N. v<strong>on</strong> Bargen, L. Bittner, S. B<strong>on</strong>figt, S. Düsterer,<br />

B. Faatz, U. Frühling, M. Gensch, Ch. Gerth, N. Guerassimova, U. Hahn,<br />

T. Hans, M. Hesse, K. H<strong>on</strong>kavaar, U. Jastrow, P. Juranic, S. Kapitzki, B. Keitel,<br />

T. Kracht, M. Kuhlmann, W. B. Li, M. Martins, T. Nú nez, E. Plönjes,<br />

H. Redlin, E. L. Saldin, E. A. Schneidmiller, J. R. Schneider, S. Schreiber,<br />

N. Stojanovic, F. Tavella, S. Toleikis, R. Treusch, H. Weigelt, M. Wellhöfer,<br />

H. Wabnitz, M. V. Yurkov, <strong>and</strong> J. Feldhaus. The soft x-ray free-electr<strong>on</strong><br />

laser flash at desy: <strong>beam</strong>lines, diagnostics <strong>and</strong> end-stati<strong>on</strong>s. New J. Phys.,<br />

11(2):023029, 2009.<br />

[105] M. D. Roper. Matching a variable-included-angle grating m<strong>on</strong>ochromator to<br />

<strong>the</strong> properties of a soft x-ray fel source to achieve a c<strong>on</strong>trolled temporal stretch.<br />

Nucl. Instrum. Meth. A, In Press, Corrected Proof:<strong>–</strong>, 2010.<br />

[106] L. Poletto <strong>and</strong> P. Villoresi. Time-delay compensated m<strong>on</strong>ochromator in <strong>the</strong> offplane<br />

mount for extreme-ultraviolet ultrashort pulses. Appl. Opt., 45(34):8577<strong>–</strong><br />

8585, 2006.<br />

[107] F. Frassetto, L. Poletto, J. I. Larruquert, <strong>and</strong> J. A. Mendez. Efficiency measurements<br />

<strong>on</strong> gratings in <strong>the</strong> off-plane mount for a high-resoluti<strong>on</strong> grazing-incidence<br />

xuv m<strong>on</strong>ochromator. volume 7077, page 707712, 2008.<br />

[108] U. Hahn <strong>and</strong> K. Tiedtke. The gas attenuator of flash at desy. AIP C<strong>on</strong>f. Proc.,<br />

879:276<strong>–</strong>282, 2007.


Bibliography 209<br />

[109] M. Kato, N. Saito, T. Tanaka, Y. Morishita, H. Kimura, H. Ohashi, M. Nagas<strong>on</strong>o,<br />

M. Yabashi, K. T<strong>on</strong>o, T. Togashi, A. Higashiya, <strong>and</strong> T. Ishikawa. Pulse<br />

energyof<strong>the</strong>extreme-ultravioletfree-electr<strong>on</strong> laser atspring-8determinedusing<br />

a cryogenic radiometer. Nucl. Instrum. Meth. A, 612(1):209 <strong>–</strong> 211, 2009.<br />

[110] N. Saito, P. N. Juranić, M. Kato, M. Richter, A. A. Sorokin, K. Tiedtke, U. Jastrow,<br />

U. Kroth, H. Schöppe, M. Nagas<strong>on</strong>o, M. Yabashi, K. T<strong>on</strong>o, T. Togashi,<br />

H. Kimura, H. Ohashi, <strong>and</strong> T. Ishikawa. Radiometric comparis<strong>on</strong> for measuring<br />

<strong>the</strong> absolute radiant power of a free-electr<strong>on</strong> laser in <strong>the</strong> extreme ultraviolet.<br />

Metrologia, 47(1):21, 2010.<br />

[111] C. Welnak, G.J. Chen, <strong>and</strong> F. Cerrina. Shadow: A synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>and</strong><br />

x-ray optics simulati<strong>on</strong> tool. Nucl. Instrum. Meth. A, 347(1-3):344 <strong>–</strong> 347, 1994.<br />

[112] S. Trabelsi-Bauer, M. Bauer, R. Steininger, <strong>and</strong> T. Baumbach. Xtrace: a new<br />

ray tracing tool for simulati<strong>on</strong> of x-ray <strong>beam</strong>lines at anka. volume 6667, page<br />

66670Q, 2007.<br />

[113] Esco. http://www.escoproducts.com/.<br />

[114] Newport. http://www.newport.com.<br />

[115] Optarius. http://www.optarius.com.<br />

[116] Abstracts for FEL’05 c<strong>on</strong>ference, 2005. TU<strong>PP</strong>061.<br />

[117] LCLSCDR(slac-r-593), secti<strong>on</strong>9.4.2.2. http://www-ssrl.slac.stanford.edu/lcls/cdr/.<br />

[118] H. Takenaka, S. Ichimaru, T. Haga, T. Ohchi, H. Ito, Y. Muramatsu, E.M.<br />

Gulliks<strong>on</strong>, <strong>and</strong> R.C.C. Perera. Fabricati<strong>on</strong> of soft x-ray <strong>beam</strong> splitters for use<br />

in <strong>the</strong> wavelength regi<strong>on</strong> around 13 nm. J. Phys. IV France, 104:251<strong>–</strong>254, 2003.<br />

[119] 2005. Proc. Of SPIE 60340D.<br />

[120] T. Haga, M.C.K. Tin<strong>on</strong>e, M. Shimada, T. Ohkubo, <strong>and</strong> A. Ozawa. J. Synchrotr<strong>on</strong><br />

Rad., 5:690, 1998.<br />

[121] R. Tatchyn, 1999. SLAC Report SLAC-TN-05-039.<br />

[122] 2008. Proc. Of SPIE 65860J.<br />

[123] Showa optr<strong>on</strong>ics. http://www.soc-ltd.co.jp/en/seihin/seimitsu/laser.html.<br />

[124] Precisi<strong>on</strong> phot<strong>on</strong>ics corporati<strong>on</strong>. http://www.precisi<strong>on</strong>phot<strong>on</strong>ics.com/prod<br />

main.asp?id=101.<br />

[125] B. Batterman <strong>and</strong> H. Cole. Rev. Mod. Phys., 36:681, 1964.<br />

[126] A.K. Freund. Opt. Eng., 34:432<strong>–</strong>440, 1995.<br />

[127] U. B<strong>on</strong>se <strong>and</strong> M. Hart. Appl. Phys. Letts., 6:155, 1965.<br />

[128] W. Graeff <strong>and</strong> U.Z. B<strong>on</strong>se. Phys. B-C<strong>on</strong>dens. Mat., 27(1):19, 1977.<br />

[129] J. Arthur LCLS DOE Review (2002); also Secti<strong>on</strong> 9.2.2.7 of <strong>the</strong> LCLS CDR<br />

(SLAC-R-593).


210 Bibliography<br />

[130] W. Roseker, A. Ehnes, H. Franz, O. Leupold, H. Schulte-Schrepping,<br />

<strong>and</strong> G. Grübel. http://hasyweb.desy.de/science/annual reports/2006<br />

report/part1/c<strong>on</strong>trib/30/18974.pdf, 2006. DESY Annual Report.<br />

[131] W. Roseker, H. Franz, A. Ehnes, H. Schulte-Schrepping, <strong>and</strong><br />

G. Grübel. http://hasyweb.desy.de/science/annual reports/2007<br />

report/part1/c<strong>on</strong>trib/30/22202.pdf, 2007. DESY Annual Report.<br />

[132] M. Nevière, P. Vincent, <strong>and</strong> R. Petit. Nouv. Rev. Opt., 5:65, 1974.<br />

[133] L. Poletto, P. Azzolin, <strong>and</strong> G. T<strong>on</strong>dello. Beam-splitting <strong>and</strong> recombining of freeelectr<strong>on</strong>-laser<br />

extreme-ultraviolet radiati<strong>on</strong>. Appl. Phys. B, 78(7<strong>–</strong>8):1009<strong>–</strong>1011,<br />

2004.<br />

[134] T. Weitkamp, B. Nöhammer, A. Diaz, C. David, <strong>and</strong> E. Zigler. X-ray wavefr<strong>on</strong>t<br />

analysis <strong>and</strong> optics characterizati<strong>on</strong> with a grating interferometer. Appl. Phys.<br />

Lett., 86:054101, 2005.<br />

[135] R. Reininger, J. Feldhaus, E. Pl<strong>on</strong>jes, R. Treusch, M.D. Roper, F.M. Quinn,<br />

<strong>and</strong> M.A. Bowler. SRI-03: AIP C<strong>on</strong>f. Proc., 705(1):572, 2004.<br />

[136] http://www.sstd.rl.ac.uk/facilities/mmt-new/interfer.html.<br />

[137] An X-ray autocorrelator <strong>and</strong> delay line for <strong>the</strong> VUV- FEL at TTF/DESY, 2005.<br />

Proc. Of SPIE 59200D.<br />

[138] J. Svatos, D. Joyeux, D. Phalippou, <strong>and</strong> F. Pollack. Opt. Lett., 18:1367, 1993.<br />

[139] E. J. Moler, Z. Hussain, R. M. Duarte, <strong>and</strong> M. R. Howells. Design <strong>and</strong> performance<br />

of a soft x-ray interferometer for ultra-high resoluti<strong>on</strong> fourier transform<br />

spectroscopy. J. Electr. Spectrosc. Rel. Phen., 80:309 <strong>–</strong> 312, 1996. Proceedings<br />

of <strong>the</strong> 11th Internati<strong>on</strong>al C<strong>on</strong>ference <strong>on</strong> Vacuum Ultraviolet Radiati<strong>on</strong> Physics.<br />

[140] D. Sch<strong>on</strong>delmaier, talk at <strong>IRUVX</strong> ”kick-off” meeting, 2008.<br />

[141] J. Zou, M. Balberg, C. Byrne, C. Liu, <strong>and</strong> D.J. Brady. Optical properties<br />

of surface micromachined mirrors with etch holes. J Microelectromech. Syst.,<br />

8:506, 1999.<br />

[142] 2005. Proc. Of SPIE 59190F.<br />

[143] A. Erko, N. Langhoff, A.A. Bjeoumikhov, <strong>and</strong> V.I. Beloglasov. Nucl. Instrum.<br />

Meth., 467:832, 2001.<br />

[144] H.C. Kang, J. Maser, G.B. Stephens<strong>on</strong>, C. Liu, R. C<strong>on</strong>ley, A.T. Macr<strong>and</strong>er,<br />

<strong>and</strong> S. Vogt. Phys. Rev. Lett., 96:127401, 2006.<br />

[145] W. Gudat, K. Kunz, <strong>and</strong> J. Karlau. Appl. Opt., 3:1412, 1974.<br />

[146] B. Lindinau, F. Janssen, <strong>and</strong> M. Leyendecker, 2008. presentati<strong>on</strong> at SRI2008,<br />

Saskato<strong>on</strong>.<br />

[147] M.A. MacD<strong>on</strong>ald, private communicati<strong>on</strong>.


Bibliography 211<br />

[148] R. Sternberger, M. Sachwitz, A. D<strong>on</strong>at, U. Gensch, R. Heller, L.V. Vu, U. Hahn,<br />

H. Schulte-Schrepping, <strong>and</strong> K. Tiedtke. A fast switching mirror unit at flash.<br />

Nucl. Instrum. Meth. A, In Press, Corrected Proof, 2010.<br />

[149] Y. V. Shvyd’ko, S. Stoupin, A. Cunsolo, A. H. Said, <strong>and</strong> X. Huang. Highreflectivity<br />

high-resoluti<strong>on</strong> x-ray crystal optics with diam<strong>on</strong>ds. Nat. Phys.,<br />

6(3):196<strong>–</strong>199, 2010.<br />

[150] M. Wellhöfer, J. T. Hoeft, M. Martins, W. Wurth, M. Braune, J. Viefhaus,<br />

K. Tiedtke, <strong>and</strong> M. Richter. Photoelectr<strong>on</strong> spectroscopy as a n<strong>on</strong>-invasive<br />

method to m<strong>on</strong>itor sase-fel spectra. J. Instr., 3(02):P02003, 2008.<br />

[151] A. Woott<strong>on</strong>, J. R. Arthur, Jr. T. W. Barbee, R. M. Bi<strong>on</strong>ta, R. A. L<strong>on</strong>d<strong>on</strong>, H.-S.<br />

Park, D. Ryutov, E. A. Spiller, <strong>and</strong> R. O. Tatchyn. X-ray optics <strong>and</strong> diagnostics<br />

for first experiments <strong>on</strong> <strong>the</strong> linac coherent light source (lcls). volume 4500, pages<br />

113<strong>–</strong>122. SPIE, 2001.<br />

[152] F. Frassetto, D. Cocco, M. Zangr<strong>and</strong>o, <strong>and</strong> L. Poletto. On-line spectrometer<br />

for fel radiati<strong>on</strong> at fermi@elettra. Nucl. Instrum. Meth. A, 593(1-2):129 <strong>–</strong> 131,<br />

2008.<br />

[153] K. Tiedtke, J. Feldhaus, U. Hahn, U. Jastrow, T. Nunez, T. Tschentscher, S. V.<br />

Bobashev, A. A. Sorokin, J. B. Hastings, S. Möller, L. Cibik, A. Gottwald,<br />

A. Hoehl, U. Kroth, M. Krumrey, H. Schöppe, G. Ulm, <strong>and</strong> M. Richter. Gas<br />

detectors for x-ray lasers. J. Appl. Phys., 103(9):094511, 2008.<br />

[154] S. P. Hau-Riege, R. M. Bi<strong>on</strong>ta, D. D. Ryutov, <strong>and</strong> J. Krzywinski. Measurement<br />

of x-ray free-electr<strong>on</strong>-laser pulse energies by photoluminescence in nitrogen gas.<br />

J. Appl. Phys., 103(5):053306, 2008.<br />

[155] H. Rabus, V. Persch, <strong>and</strong> G. Ulm. Synchrotr<strong>on</strong>-radiati<strong>on</strong>-operated cryogenic<br />

electrical-substituti<strong>on</strong> radiometer as <strong>the</strong> high-accuracy primary detector st<strong>and</strong>ard<br />

in <strong>the</strong> ultraviolet, vacuum-ultraviolet, <strong>and</strong> soft-x-ray spectral ranges. Appl.<br />

Opt., 36:5421, 1997.<br />

[156] P.-S. Shaw, K.R.Lykke, R.Gupta, T.R. O?Brian, U.Arp, H.H.White, T.B. Lucatorto,<br />

J.L. Dehmer, <strong>and</strong> A.C. Parr. Ultraviolet radiometry with synchrotr<strong>on</strong><br />

radiati<strong>on</strong> <strong>and</strong> cryogenic radiometry. Appl. Opt., 38:18, 1999.<br />

[157] Y. Morishita, N. Saito, <strong>and</strong> I. H. Suzuki. Comparis<strong>on</strong> of <strong>the</strong> absolute soft xray<br />

intensity between a cryogenic radiometer <strong>and</strong> an i<strong>on</strong> chamber. J. Electr<strong>on</strong><br />

Spectrosc. Relat. Phenom., 144, 2005.<br />

[158] M. Kato, A. Nohtomi, Y. Morishita, T. Kurosawa, N. Arai, I.H. Suzuki, <strong>and</strong><br />

N. Saito. Development of <strong>the</strong> soft x-ray intensity measurement with a cryogenic<br />

radiometer. AIP C<strong>on</strong>f. Proc., 879, 2007.<br />

[159] S. Friedrich, L. Li, L.L. Ott, Rajeswari M. Kolgani, G.J. Y<strong>on</strong>g, Z.A. Ali, O.B.<br />

Drury, E. Ables, <strong>and</strong> R.M. Bi<strong>on</strong>ta. Design of a bolometer for total-energy measurement<br />

of <strong>the</strong> linear coherent light source pulsed x-ray laser. Nucl. Instrum.<br />

Meth. A, 559(2):772 <strong>–</strong> 774, 2006.


212 Bibliography<br />

[160] O.B. Drury, G.J. Y<strong>on</strong>g, R.M. Kolagani, Y<strong>on</strong>g Liang, C.S. Gardner, E. Ables,<br />

K.W. F<strong>on</strong>g, R.M. Bi<strong>on</strong>ta, <strong>and</strong> S. Friedrich. Fabricati<strong>on</strong> of cryogenic manganite<br />

bolometers to measure <strong>the</strong> total energy at <strong>the</strong>lcls free electr<strong>on</strong> x-raylaser. IEEE<br />

Trans. Nucl. Sci., 56(3):1114 <strong>–</strong> 1120, 2009.<br />

[161] W.C. Wiley<strong>and</strong>I.H.MacLaren. Time-of-flightmassspectrometer withimproved<br />

resoluti<strong>on</strong>. Rev. Sci. Instrum., 26:1150<strong>–</strong>1157, 1955.<br />

[162] J. H. D. El<strong>and</strong>. Meas. Sci. Technol., 4:1522<strong>–</strong>1524, 1993.<br />

[163] D. P. Seccombe <strong>and</strong> T. J. Reddish. Rev. Sci. Instrum., 72:1330<strong>–</strong>1338, 2001.<br />

[164] P. N. Juranić, M. Martins, J. Viefhaus, S. B<strong>on</strong>figt, L. Jahn, M. Ilchen,<br />

S. Klumpp, <strong>and</strong> K. Tiedtke. Using i-tof spectrometry to measure phot<strong>on</strong> energies<br />

at fels. J. Instr., 4(09):P09011, 2009.<br />

[165] V. Ayvazyan, N. Baboi, I. Bohnet, R. Brinkmann, M. Castellano, P. Castro,<br />

L. Catani, S. Choroba, A. Cianchi, M. Dohlus, H. T. Edwards, B. Faatz, A. A.<br />

Fateev, J. Feldhaus, K. Flöttmann, A. Gamp, T. Garvey, H. Genz, Ch. Gerth,<br />

V. Gretchko, B. Grigoryan, U. Hahn, C. Hessler, K. H<strong>on</strong>kavaara, M. Hüning,<br />

R. Ischebeck, M. Jabl<strong>on</strong>ka, T. Kamps, M. Körfer, M. Krassilnikov, J. Krzywinski,<br />

M. Liepe, A. Liero, T. Limberg, H. Loos, M. Lu<strong>on</strong>g, C. Magne, J. Menzel,<br />

P. Michelato, M. Minty, U.-C. Müller, D. Nölle, A. Novokhatski, C. Pagani,<br />

F. Peters, J. Pflüger, P. Piot, L. Plucinski, K. Rehlich, I. Reyzl, A. Richter,<br />

J. Rossbach, E. L. Saldin, W. S<strong>and</strong>ner, H. Schlarb, G. Schmidt, P. Schmüser,<br />

J. R. Schneider, E. A. Schneidmiller, H.-J. Schreiber, S. Schreiber, D. Sertore,<br />

S. Setzer, S. Simrock, R. Sobierajski, B. S<strong>on</strong>ntag, B. Steeg, F. Stephan, K. P.<br />

Sytchev, K. Tiedtke, M. T<strong>on</strong>utti, R. Treusch, D. Trines, D. Türke, V. Verzilov,<br />

R. Wanzenberg, T. Weil<strong>and</strong>, H. Weise, M. Wendt, I. Will, S. Wolff, K. Wittenburg,<br />

M. V. Yurkov, <strong>and</strong> K. Zapfe. Generati<strong>on</strong> of gw radiati<strong>on</strong> pulses from a<br />

vuv free-electr<strong>on</strong> laser operating in <strong>the</strong> femtosec<strong>on</strong>d regime. Phys. Rev. Lett.,<br />

88(10):104802, 2002.<br />

[166] Optical <strong>beam</strong> quality in free-electr<strong>on</strong> lasers, 2006. Proceedings of FEL 2006,<br />

Bessy, Berlin, Germany.<br />

[167] D. H. Lumb. Applicati<strong>on</strong>s of charge coupled devices to x-ray astrophysics missi<strong>on</strong>s.<br />

Nucl. Instrum. Meth. A, 288(1):219 <strong>–</strong> 226, 1990.<br />

[168] A. Attaelmanan, A. Rindby, P. Voglis, <strong>and</strong> A. Shermeat. X-ray <strong>beam</strong> profile<br />

measurements with ccd detectors. Nucl. Instrum. Meth. B, 82(3):481 <strong>–</strong> 488,<br />

1993.<br />

[169] M. G. Fedotov. Ccd detectors for x-ray synchrotr<strong>on</strong> radiati<strong>on</strong> applicati<strong>on</strong>. Nucl.<br />

Instrum. Meth. A, 448(1-2):192 <strong>–</strong> 195, 2000.<br />

[170] M.J. Bol<strong>and</strong>, R.T. Dowd, G. LeBlanc, M.J. Spencer, Y.E. Tan, <strong>and</strong> A. Walsh.<br />

X-ray <strong>and</strong> optical diagnostic <strong>beam</strong>lines at <strong>the</strong> australian synchrotr<strong>on</strong> storage<br />

ring. In Proc. EPAC 2006 C<strong>on</strong>ference, June 2006. Edinburgh, Scotl<strong>and</strong>.<br />

[171] A. Koch, C. Raven, P. Spanne, <strong>and</strong> A. Snigirev. X-ray imaging with submicrometer<br />

resoluti<strong>on</strong> employing transparent luminescent screens. J. Opt. Soc.<br />

Am. A, 15(7):1940<strong>–</strong>1951, Jul 1998.


Bibliography 213<br />

[172] J. Tous, M. Horváth, L. Pina, K. Blazek, <strong>and</strong> B. Sopko. High-resoluti<strong>on</strong> applicati<strong>on</strong><br />

of yag:ce <strong>and</strong> luag:ce imaging detectors with a ccd x-ray camera. Nucl.<br />

Instrum. Meth. A, 591(1):264 <strong>–</strong> 267, 2008.<br />

[173] X. Yang, H. Li, Q. Bi, L. Su, <strong>and</strong> J. Xu. Growth of large-sized ce:y3al5o12<br />

(ce:yag) scintillati<strong>on</strong> crystal by <strong>the</strong> temperature gradient technique (tgt). J.<br />

Cryst. Grow., 311(14):3692 <strong>–</strong> 3696, 2009.<br />

[174] J. Yang, T. Nakajyo, Y. Okada, F. Sakai, T. Yanagida, M. Yorozu, A. Endo,<br />

S. Ito, <strong>and</strong> K. Takasago. Spatial profile measurement of femtosec<strong>on</strong>d lasercompt<strong>on</strong><br />

x-rays. In Proc. EPAC 2002 C<strong>on</strong>ference, pages 784<strong>–</strong>786, June 2002.<br />

Paris, France.<br />

[175] D.Shu, P.K.Job, J.Barraza, T. Cundiff, <strong>and</strong>T. M.Kuzay. Cvd-diam<strong>on</strong>d-based<br />

positi<strong>on</strong> sensitive photoc<strong>on</strong>ductive detector for high-flux x-rays <strong>and</strong> gamma<br />

rays. In Proc. PAC’99 C<strong>on</strong>ference, pages 2090<strong>–</strong>2092, March-April 1999. New<br />

York, USA.<br />

[176] F.-P. Wolf <strong>and</strong> W. Peatman. Synchrotr<strong>on</strong> radiati<strong>on</strong> diagnostics at bessy. Nucl.<br />

Instrum. Meth. A, 246(1-3):408 <strong>–</strong> 410, 1986.<br />

[177] M. Lankosz <strong>and</strong> J. Sieber. A simple micro<strong>beam</strong> profiling technique for x-ray<br />

optics. Rev. Sci. Instrum., 71(7):2640<strong>–</strong>2643, 2000.<br />

[178] P. Fajardo <strong>and</strong> S. Ferrer. Ultrahigh-vacuum-compatible positi<strong>on</strong> <strong>and</strong> shape<br />

m<strong>on</strong>itor for high brilliance synchrotr<strong>on</strong> radiati<strong>on</strong> <strong>beam</strong>s. Rev. Sci. Instrum.,<br />

66(2):1882<strong>–</strong>1884, 1995.<br />

[179] R. W. Schoenlein, A. Cavalleri, H. H. W. Ch<strong>on</strong>g, T. E. Glover,<br />

P. A. Heimann, A. A. Zholents, <strong>and</strong> M. S. Zolotorev. Generati<strong>on</strong><br />

of femtosec<strong>on</strong>d synchrotr<strong>on</strong> pulses: Performance <strong>and</strong> characterizati<strong>on</strong>.<br />

http://escholarship.org/uc/item/34001226, 2003.<br />

[180] Y. Iketaki, Y. Horikawa, S. Mochimaru, K. Nagai, T. Kiyokura, M. Oshima,<br />

<strong>and</strong> A. Yagishita. Study of <strong>the</strong> x-ray micro<strong>beam</strong> for scanning microscopes. J.<br />

Elect. Spectrosc. Rel. Phenom., 80:353 <strong>–</strong> 356, 1996.<br />

[181] Y. Suzuki, M. Awaji, Y. Kohmura, A. Takeuchi, H. Takano, N. Kamijo,<br />

S. Tamura, M. Yasumoto, <strong>and</strong> K. H<strong>and</strong>a. X-ray micro<strong>beam</strong> with sputteredsliced<br />

fresnel z<strong>on</strong>e plate at spring-8 undulator <strong>beam</strong>line. Nucl. Instrum. Meth.<br />

A, 467-468(Part 2):951 <strong>–</strong> 953, 2001.<br />

[182] I<strong>on</strong>izati<strong>on</strong> profile m<strong>on</strong>itor to determine spatial <strong>and</strong> angular stability of FEL<br />

radiati<strong>on</strong> of FLASH, 2008. Proceedings of EPAC08, Genoa, Italy.<br />

[183] C. Fischer <strong>and</strong> J. Koopman. I<strong>on</strong>isati<strong>on</strong> profile m<strong>on</strong>itor tests in <strong>the</strong> sps. In Proc.<br />

DIPAC’99 C<strong>on</strong>ference, pages 128<strong>–</strong>130, May 1999. Chester, UK.<br />

[184] L. Ioudin, V. Mikhailov, V. Rezvov, V. Sklyarenko, A. Artemev, S. Peredkov,<br />

T. Rakhimbabev, M. Lem<strong>on</strong>nier, S. Megtert, <strong>and</strong> M. Roulliay. Synchrotr<strong>on</strong><br />

radiati<strong>on</strong> parameters registrati<strong>on</strong> using <strong>beam</strong> cross-secti<strong>on</strong> image detector: The<br />

first experiments. Nucl. Instrum. Meth. A, 405(2-3):265 <strong>–</strong> 268, 1998.


214 Bibliography<br />

[185] A.N. Artemiev, L.I. Ioudin, V.H. Lightenshtein, V.G. Mikhailov, S.S. Peredkov,<br />

T.Y. Rakhimbabaev, V.A. Rezvov, V.I. Sklyarenko, M. Lem<strong>on</strong>nier, S. Megtert,<br />

<strong>and</strong> M. Roulliay. The development of <strong>the</strong> i<strong>on</strong>isati<strong>on</strong> detectors for measuring<br />

<strong>the</strong> main parameters of <strong>the</strong> accelerated i<strong>on</strong>ising <strong>beam</strong>s. In Proc. EPAC’96<br />

C<strong>on</strong>ference, pages 1716<strong>–</strong>1718, June 1996. Barcel<strong>on</strong>a, Spain.<br />

[186] A.N. Artemiev, N.A. Artemiev, L.I. Ioudin, S.T. Latushkin, V.G. Mikhailov,<br />

V.P. Moryakov, D.G. Odintsov, V.A. Rezvov, A.G.Valentinov, Y. Cerenius, <strong>and</strong><br />

A. Svenss<strong>on</strong>. An i<strong>on</strong>izati<strong>on</strong> detector of synchrotr<strong>on</strong> <strong>beam</strong> spatial parameters:<br />

possible applicati<strong>on</strong>s. In Proc. EPAC’98 C<strong>on</strong>ference, pages 1535<strong>–</strong>1537, June<br />

1998. Stockholm, Sweden.<br />

[187] P. Ilinski, U. Hahn, H. Schulte-Schrepping, <strong>and</strong> M. Degenhardt. Residual<br />

gas x-ray <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor development for petra iii. AIP C<strong>on</strong>f. Proc.,<br />

879(1):782<strong>–</strong>785, 2007.<br />

[188] J. Camas, R. J. Colchester, G. Ferioli, R. Jung, <strong>and</strong> J. Koopman. Preliminary<br />

test of a luminescence profile m<strong>on</strong>itor in <strong>the</strong> cern sps. In Proc. DIPAC’99<br />

C<strong>on</strong>ference, pages 95<strong>–</strong>99, May 1999. Chester, UK.<br />

[189] R. G. van Silfhout. A high-precisi<strong>on</strong> x-ray <strong>beam</strong>-positi<strong>on</strong> <strong>and</strong> profile m<strong>on</strong>itor<br />

for synchrotr<strong>on</strong> <strong>beam</strong>lines. J. Synchrotr<strong>on</strong> Rad., 6:1071<strong>–</strong>1075, 1999.<br />

[190] R.G. van Silfhout, S. Manolopoulos, N.R. Kyele, <strong>and</strong> K. Decanniere. In situ<br />

high-speed synchrotr<strong>on</strong> x-ray <strong>beam</strong> profiling <strong>and</strong> positi<strong>on</strong> m<strong>on</strong>itoring. Sensor<br />

Actuat A-Phys, 133(1):82 <strong>–</strong> 87, 2007.<br />

[191] T. Kudo, S. Takahashi, N. Nariyama, T. Hir<strong>on</strong>o, T. Tachibana, <strong>and</strong> H. Kitamura.<br />

Synchrotr<strong>on</strong> radiati<strong>on</strong> x-ray <strong>beam</strong> profile m<strong>on</strong>itor using chemical vapor<br />

depositi<strong>on</strong> diam<strong>on</strong>d film. Rev. Sci. Instrum., 77(12):123105, 2006.<br />

[192] P. Fajardo <strong>and</strong> S. Ferrer. Beam m<strong>on</strong>itor for undulator white radiati<strong>on</strong> in <strong>the</strong><br />

hard-x-ray range. Rev. Sci. Instrum., 66(2):1879<strong>–</strong>1881, 1995.<br />

[193] R. Reininger, J. Feldhaus, E. Pl<strong>on</strong>jes, R. Treusch, M. D. Roper, F. M. Quinn,<br />

<strong>and</strong> M. A. Bowler. Spectrometer based <strong>on</strong> a vls grating for diagnostics of a<br />

vacuum-ultraviolet free electr<strong>on</strong> laser. AIP C<strong>on</strong>f. Proc., 705(1):572<strong>–</strong>575, 2004.<br />

[194] J. Chalupsk´y, L. Juha, J. Kuba, J. Cihelka, V. Hájková, S. Koptyaev, J. Krása,<br />

A. Velyhan, M. Bergh, C. Caleman, J. Hajdu, R. M. Bi<strong>on</strong>ta, H. Chapman,<br />

S. P. Hau-Riege, R. A. L<strong>on</strong>d<strong>on</strong>, M. Jurek, J. Krzywinski, R. Nietubyc, J. B.<br />

Pelka, R. Sobierajski, J. Meyer ter Vehn, A. Tr<strong>on</strong>nier, K. Sokolowski-Tinten,<br />

N. Stojanovic, K. Tiedtke, S. Toleikis, T. Tschentscher, H. Wabnitz, <strong>and</strong> U. Zastrau.<br />

Characteristics of focused soft x-ray free-electr<strong>on</strong> laser <strong>beam</strong> determined<br />

by ablati<strong>on</strong> of organic molecular solids. Opt. Express, 15(10):6036<strong>–</strong>6043, 2007.<br />

[195] A. A. Sorokin, A. Gottwald, A. Hoehl, U. Kroth, H. Schoppe, G. Ulm,<br />

M. Richter, S. V. Bobashev, I. V. Domracheva, D. N. Smirnov, K. Tiedtke,<br />

S. Dusterer, J. Feldhaus, U. Hahn, U. Jastrow, M. Kuhlmann, T. Nunez,<br />

E. Pl<strong>on</strong>jes, <strong>and</strong> R. Treusch. Method based <strong>on</strong> atomic photoi<strong>on</strong>izati<strong>on</strong> for spotsize<br />

measurement <strong>on</strong> focused soft x-ray free-electr<strong>on</strong> laser <strong>beam</strong>s. Appl. Phys.<br />

Lett., 89(22):221114, 2006.


Bibliography 215<br />

[196] M. Richter, S.Bobashev, A. Sorokin, <strong>and</strong> K.Tiedtke. N<strong>on</strong>linear photoi<strong>on</strong>izati<strong>on</strong><br />

in <strong>the</strong> soft x-ray regime. Appl. Phys. A, 92:473<strong>–</strong>478, 2008.<br />

[197] R.J. Davies, M. Burghammer, , <strong>and</strong> C. Riekel. An overview of <strong>the</strong> esrf’s id13<br />

microfocus <strong>beam</strong>line. Synchr. Rad. Nat. Sci., 5(1-2), 2006.<br />

[198] R. A. Lewis, A. Berry, C. J. Hall, W. I. Helsby, <strong>and</strong> B. T. Parker. The rapid<br />

detector system - first user data. Nucl. Instrum. Meth. A, 454(1):165 <strong>–</strong> 172,<br />

2000.<br />

[199] J. Zhang, J. Chen, J. Han, R. Li, Q. Liu, M. Su, Wang W., Y. Wang, Y. Xie,<br />

N. Yao, H. Zhao, J. Zhao, <strong>and</strong> J. Zhao. The design <strong>and</strong> mass producti<strong>on</strong> <strong>on</strong><br />

resistive plate chambers for <strong>the</strong> besiii experiment. Nucl. Instrum. Meth. A,<br />

580(3):1250 <strong>–</strong> 1256, 2007.<br />

[200] C. DaVia. Jan 2003.<br />

[201] J.M. Carm<strong>on</strong>a, B. Bra nas, A. Ibarra, <strong>and</strong> I. Podadera Alised. Transverse profile<br />

m<strong>on</strong>itors based <strong>on</strong> fluorescence for ifmif-eveda accelerator. In Proc. DIPAC’09<br />

C<strong>on</strong>f., pages 1223<strong>–</strong>1225, May 2009. Basel, Switzerl<strong>and</strong>.<br />

[202] B. Henrich, A. Bergamaschi, C. Broennimann, R. Dinapoli, E.F. Eikenberry,<br />

I. Johns<strong>on</strong>, M. Kobas, P. Kraft, A. Mozzanica, <strong>and</strong> B. Schmitt. Pilatus: A<br />

single phot<strong>on</strong> counting pixel detector for x-ray applicati<strong>on</strong>s. Nucl. Instrum.<br />

Meth. A, 607(1):247 <strong>–</strong> 249, 2009.<br />

[203] P. Mortazavi, M. Woodle, H. Rarback, D. Shu, <strong>and</strong> M. Howells. High flux<br />

phot<strong>on</strong> <strong>beam</strong> m<strong>on</strong>itor. Nucl. Instrum. Meth. A, 246(1-3):389 <strong>–</strong> 393, 1986.<br />

[204] E. D. Johns<strong>on</strong> <strong>and</strong> T. Oversluizen. Uhv photoelectr<strong>on</strong> x-ray <strong>beam</strong> positi<strong>on</strong><br />

m<strong>on</strong>itor. Nucl. Instrum. Meth. A, 291(1-2):427 <strong>–</strong> 430, 1990.<br />

[205] S. M. Heald. A simple photoelectr<strong>on</strong> x-ray <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor for synchrotr<strong>on</strong><br />

radiati<strong>on</strong>. Nucl. Instrum. Meth. A, 246(1-3):411 <strong>–</strong> 412, 1986.<br />

[206] R.W. Alkire, E.P. Sullivan, F.D. Michaud, W.J. Trela, <strong>and</strong> R.J. Bartlett. The<br />

x8c dual wire <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor. Nucl. Instrum. Meth. A, 350(1-2):13 <strong>–</strong><br />

16, 1994.<br />

[207] D. Shu, B. Rodricks, J. Barraza, T. Sanchez, <strong>and</strong> T. M. Kuzay. The aps x-ray<br />

undulator phot<strong>on</strong> <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor <strong>and</strong> tests at chess <strong>and</strong> nsls. Nucl.<br />

Instrum. Meth. A, 319(1-3):56 <strong>–</strong> 62, 1992.<br />

[208] D. Shu, J. Barraza, H. Ding, T. M. Kuzay, <strong>and</strong> M. Ramanathan. Progress of<br />

<strong>the</strong> aps high heat load x-ray <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor development. AIP C<strong>on</strong>f.<br />

Proc., 417(1):173<strong>–</strong>177, 1997.<br />

[209] E. D. Johns<strong>on</strong> <strong>and</strong> T. Oversluizen. Compact high flux phot<strong>on</strong> <strong>beam</strong> positi<strong>on</strong><br />

m<strong>on</strong>itor. Rev. Sci. Instrum., 60(7):1947<strong>–</strong>1950, 1989.<br />

[210] T. Warwick, D. Shu,B. Rodricks, <strong>and</strong>E. D.Johns<strong>on</strong>. Prototype phot<strong>on</strong>positi<strong>on</strong><br />

m<strong>on</strong>itors for undulator <strong>beam</strong>s at <strong>the</strong> advanced light source. Rev. Sci. Instrum.,<br />

63(1):550<strong>–</strong>553, 1992.


216 Bibliography<br />

[211] G. Decker <strong>and</strong> O. Singh. Method for reducing x-ray background signals from<br />

inserti<strong>on</strong> device x-ray <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itors. Phys. Rev. ST Accel. Beams,<br />

2(11):112801, 1999.<br />

[212] A. Galimberti, R. Borghes, G. Paolucci, R. Presacco, G. Paolicelli, <strong>and</strong> G. Stefani.<br />

First results of <strong>the</strong> novel phot<strong>on</strong> <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itor for undulator<br />

<strong>beam</strong>lines of elettra. Nucl. Instrum. Meth. A, 467-468:221 <strong>–</strong> 225, 2001.<br />

[213] T. Mitsuhashi, A. Ueda, <strong>and</strong> T. Katsura. High-flux phot<strong>on</strong> <strong>beam</strong> positi<strong>on</strong><br />

m<strong>on</strong>itor. Rev. Sci. Instrum., 63(1):534<strong>–</strong>537, 1992.<br />

[214] R. G. van Silfhout. A <strong>beam</strong> tracking optical table for synchrotr<strong>on</strong> x-ray <strong>beam</strong>lines.<br />

Nucl. Instrum. Meth. A, 403(1):153 <strong>–</strong> 160, 1998.<br />

[215] C.J. Kenney, J. Hasi, Sherwood Parker, A.C. Thomps<strong>on</strong>, <strong>and</strong> E. Westbrook.<br />

Use of active-edge silic<strong>on</strong> detectors as x-ray <strong>beam</strong> m<strong>on</strong>itors. Nucl. Instrum.<br />

Meth. A, 582(1):178 <strong>–</strong> 181, 2007.<br />

[216] D. Shu, J. Barraza, T. M. Kuzay, G. Naylor, <strong>and</strong> P. Elleaume. Tests of <strong>the</strong><br />

aps x-ray transmitting <strong>beam</strong> positi<strong>on</strong> m<strong>on</strong>itors at esrf. In Proc. PAC’97 C<strong>on</strong>f.,<br />

pages 2210<strong>–</strong>2212, May 1997. Vancouver, B. C., Canada.<br />

[217] D. Shu, T. M. Kuzay, Y. Fang, J. Barraza, <strong>and</strong> T. Cundiff. Syn<strong>the</strong>tic diam<strong>on</strong>dbasedpositi<strong>on</strong>-sensitive<br />

photoc<strong>on</strong>ductivedetectordevelopmentfor<strong>the</strong>advanced<br />

phot<strong>on</strong> source. J. Synchrotr<strong>on</strong> Rad., 5:636<strong>–</strong>638, 1998.<br />

[218] ˚A . Anderss<strong>on</strong>, V. Schlott, M. Rohrer, A. Streun, <strong>and</strong> O. Chubar. Electr<strong>on</strong><br />

<strong>beam</strong> profile measurements with visible <strong>and</strong> x-ray synchrotr<strong>on</strong> radiati<strong>on</strong> at <strong>the</strong><br />

swiss light source. In Proc. EPAC 2006 C<strong>on</strong>f., pages 1223<strong>–</strong>1225, June 2006.<br />

Edinburgh, Scotl<strong>and</strong>.<br />

[219] R. W. Alkire, G. Rosenbaum, <strong>and</strong> G. Evans. Design of a vacuum-compatible<br />

high-precisi<strong>on</strong> m<strong>on</strong>ochromatic <strong>beam</strong>-positi<strong>on</strong> m<strong>on</strong>itor for use with synchrotr<strong>on</strong><br />

radiati<strong>on</strong> from 5 to 25 kev. J. Synchrotr<strong>on</strong> Rad., 7:61<strong>–</strong>68, 2000.<br />

[220] J. Tischler <strong>and</strong> P.L. Hartman. Photo-electr<strong>on</strong> <strong>and</strong> i<strong>on</strong>izati<strong>on</strong> chamber positi<strong>on</strong><br />

m<strong>on</strong>itors for chess. Nucl. Instrum. Meth., 172(1-2):67 <strong>–</strong> 71, 1980.<br />

[221] W. Schildkamp. N<strong>on</strong>destructive positi<strong>on</strong> m<strong>on</strong>itor for synchrotr<strong>on</strong> radiati<strong>on</strong>.<br />

Nucl. Instrum. Meth. A, 258(2):275 <strong>–</strong> 280, 1987.<br />

[222] W. Schildkamp <strong>and</strong> C. Praderv<strong>and</strong>. Positi<strong>on</strong> m<strong>on</strong>itor <strong>and</strong> readout electr<strong>on</strong>ics<br />

for undulator <strong>and</strong> focused bending magnet <strong>beam</strong>lines. Rev. Sci. Instrum.,<br />

66(2):1956<strong>–</strong>1959, 1995.<br />

[223] M. G. Billing. Beam positi<strong>on</strong> m<strong>on</strong>itors for storage rings. Nucl. Instrum. Meth.<br />

A, 266(1-3):144 <strong>–</strong> 154, 1988.<br />

[224] W. H. Southwell. Wave-fr<strong>on</strong>t estimati<strong>on</strong> from wave-fr<strong>on</strong>t slope measurements.<br />

J. Opt. Soc. Am., 70(8):998<strong>–</strong>1006, 1980.<br />

[225] M. Kuhlmann, E. Plönjes, K. Tiedtke, S. Toleikis, P. Zeitoun, J. Gautier,<br />

T. Lefrou, D. Douillet, P. Mercère, S. Bucourt G. Dovillaire, X. Levecq, <strong>and</strong><br />

M. Ajardo. Wave-fr<strong>on</strong>t observati<strong>on</strong>s at flash. In Proc. FEL 2006, 2006. BESSY,<br />

Berlin, Germany.


Bibliography 217<br />

[226] B. Flöter, P. Juranic, P. Großmann, S. Kapitzki, B. Keitel, K. Mann, E. Plönjes,<br />

B. Schäfer, <strong>and</strong> K. Tiedtke. Euv hartmann sensor for wavefr<strong>on</strong>t measurements<br />

at <strong>the</strong> free-electr<strong>on</strong> laser in hamburg. New J. Phys., 12(8):083015, 2010.<br />

[227] B. Flöter, P. Juranic, P. Großmann, S. Kapitzki, B. Keitel, K. Mann, E. Plönjes,<br />

B. Schäfer, <strong>and</strong> K. Tiedtke. Beam parameters of flash <strong>beam</strong>line bl1 from hartmann<br />

wavefr<strong>on</strong>t measurements. Nucl. Instrum. Meth. A, In Press, Corrected<br />

Proof:<strong>–</strong>, 2010.<br />

[228] P. Mercère, R. Bachelard, M.-E. Couprie, M. Idir, O. Chubar, J. Gautier,<br />

G. Lambert, P. Zeitoun, S. Bucourt, G. Dovillaire, X. Levecq, H. Kimura,<br />

H. Ohashi, T. Hara, A. Higashiya, T. Ishikawa, M. Nagas<strong>on</strong>o, <strong>and</strong> M. Yabashi.<br />

Spatial characterizati<strong>on</strong> of sase-fel of scss test accelerator. In Proc. FEL 2009,<br />

2009. Liverpool, UK.<br />

[229] Q. Wu, T. D. Hewitt, <strong>and</strong> X.-C. Zhang. Two-dimensi<strong>on</strong>al electro-optic imaging<br />

of thz <strong>beam</strong>s. Appl. Phys. Lett., 69(8):1026<strong>–</strong>1028, 1996.<br />

[230] F. Tavella, N. Stojanovic, G. Gel<strong>on</strong>i, <strong>and</strong> M. Gensch. Few-femtosec<strong>on</strong>d timing<br />

at fourth-generati<strong>on</strong> x-ray light sources. Nat. Phot<strong>on</strong>., pages XX<strong>–</strong>XX, 2011.<br />

<strong>on</strong>line advance publicati<strong>on</strong>.<br />

[231] M. Drescher, M. Hentschel, R. Kienberger, G. Tempea, C. Spielmann, G. A.<br />

Reider, P.B. Corkum, <strong>and</strong>F.Krausz. X-rayPulses Approaching<strong>the</strong>Attosec<strong>on</strong>d<br />

Fr<strong>on</strong>tier. Science, 291(5510):1923<strong>–</strong>1927, 2001.<br />

[232] P. Radcliffe, S. Düsterer, A. Azima, H. Redlin, J. Feldhaus, J. Dardis, K. Kavanagh,<br />

H. Luna, J. Pedregosa Gutierrez, P. Yeates, E. T. Kennedy, J. T.<br />

Costello, A. Delserieys, C. L. S. Lewis, R. Taïeb, A. Maquet, D. Cubaynes, <strong>and</strong><br />

M. Meyer. Single-shot characterizati<strong>on</strong> of independent femtosec<strong>on</strong>d extreme ultraviolet<br />

freeelectr<strong>on</strong> <strong>and</strong>infrared laser pulses. Appl. Phys. Lett., 90(13):131108,<br />

2007.<br />

[233] S. Cunovic, N. Müller, R. Kalms, M. Krikunova, M. Wiel<strong>and</strong>, M. Drescher,<br />

Th. Maltezopoulos, U. Frühling, H. Redlin, E. Plönjes-Palm, <strong>and</strong> J. Feldhaus.<br />

Time-to-space mapping in a gas medium for <strong>the</strong> temporal characterizati<strong>on</strong> of<br />

vacuum-ultraviolet pulses. Appl. Phys. Lett., 90(12):121112, 2007.<br />

[234] G. Berden, W. A. Gillespie, S. P. Jamis<strong>on</strong>, E.-A. Knabbe, A. M. MacLeod,<br />

A. F. G. van der Meer, P. J. Phillips, H. Schlarb, B. Schmidt, P. Schmüser, <strong>and</strong><br />

B. Steffen. Benchmarking of electro-optic m<strong>on</strong>itors for femtosec<strong>on</strong>d electr<strong>on</strong><br />

bunches. Phys. Rev. Lett., 99(16):164801, 2007.<br />

[235] Single Shot L<strong>on</strong>gitudinal Bunch Profile Measurements by Temporally Resolved<br />

Electrooptical Detecti<strong>on</strong>, 2007. Proceedings of DIPAC 2007, Venice, Italy.<br />

[236] A. Azima, S. Düsterer, P. Radcliffe, H. Redlin, N. Stojanovic, W. Li, H. Schlarb,<br />

J. Feldhaus, D. Cubaynes, M. Meyer, J. Dardis, P. Hayden, P. Hough,<br />

V Richards<strong>on</strong>, E. T. Kennedy, <strong>and</strong> J. T. Costello. Time-resolved pumpprobe<br />

experiments bey<strong>on</strong>d <strong>the</strong> jitter limitati<strong>on</strong>s at flash. Appl. Phys. Lett.,<br />

94(14):144102, 2009.


218 Bibliography<br />

[237] V. Petrov, F. Noack, D. Shen, F. Pan, G. Shen, X. Wang, R. Komatsu, <strong>and</strong><br />

V. Alex. Applicati<strong>on</strong> of <strong>the</strong> n<strong>on</strong>linear crystal srb4o7 for ultrafast diagnostics<br />

c<strong>on</strong>verting to wavelengths as short as 125 nm. Opt. Lett., 29(4):373<strong>–</strong>375, 2004.<br />

[238] A. Weiner. Effect of group velocity mismatch <strong>on</strong> <strong>the</strong> measurement of ultrashort<br />

optical pulses via sec<strong>on</strong>d harm<strong>on</strong>ic generati<strong>on</strong>. IEEE J. Quantum Electr<strong>on</strong>.,<br />

19:1276<strong>–</strong>1283, 1983.<br />

[239] J. Dai, H. Teng, <strong>and</strong> C. Guo. Sec<strong>on</strong>d- <strong>and</strong> third-order interferometric autocorrelati<strong>on</strong>s<br />

based <strong>on</strong> harm<strong>on</strong>ic generati<strong>on</strong>s from metal surfaces. Opt. Comm.,<br />

252(1-3):173 <strong>–</strong> 178, 2005.<br />

[240] J.K.Ranka,A.L.Gaeta, A.Baltuska, M.S.Pshenichnikov,<strong>and</strong>D.A.Wiersma.<br />

Autocorrelati<strong>on</strong> measurement of 6-fs pulses based <strong>on</strong> <strong>the</strong> two-phot<strong>on</strong>-induced<br />

photocurrent in a gaasp photodiode. Opt. Lett., 22(17):1344<strong>–</strong>1346, 1997.<br />

[241] T.-M. Liu, Y.-C. Huang, G.-W. Chern, K.-H. Lin, C.-J. Lee, Y.-C. Hung, <strong>and</strong><br />

C.-K. Sun. Triple-optical autocorrelati<strong>on</strong> for direct optical pulse-shape measurement.<br />

Appl. Phys. Lett., 81(8):1402<strong>–</strong>1404, 2002.<br />

[242] T.-M. Liu, Y.-C. Huang, G.-W. Chern, K.-H. Lin, Y.-C. Hung, C.-J. Lee,<br />

<strong>and</strong> C.-K. Sun. Characterizati<strong>on</strong> of ultrashort optical pulses with thirdharm<strong>on</strong>ic-generati<strong>on</strong><br />

based triple autocorrelati<strong>on</strong>. IEEE J. Quantum Electr<strong>on</strong>,<br />

38(11):1529<strong>–</strong>1535, 2002.<br />

[243] W. F. Schlotter, F. Sorgenfrei, T. Beeck, M. Beye, S. Gieschen, H. Meyer,<br />

M. Nagas<strong>on</strong>o, A. Föhlisch, <strong>and</strong> W. Wurth. L<strong>on</strong>gitudinal coherence measurements<br />

of an extreme-ultraviolet free-electr<strong>on</strong> laser. Opt. Lett., 35(3):372<strong>–</strong>374,<br />

2010.<br />

[244] C. Gahl, A. Azima, M. Beye, M. Deppe, K. Döbrich, U. Hasslinger, F. Hennies,<br />

A. Melnikov, M. Nagas<strong>on</strong>o, A. Pietzsch, M. Wolf, W. Wurth, <strong>and</strong> A. Föhlisch.<br />

A femtosec<strong>on</strong>d x-ray/optical cross-correlator. Nat. Phot<strong>on</strong>., 2:165<strong>–</strong>169, 2008.<br />

[245] N. de Oliveira, D. Joyeux, D. Phalippou, J. C. Rodier, F. Polack, M. Vervloet,<br />

<strong>and</strong> L. Nah<strong>on</strong>. A fourier transform spectrometer without a <strong>beam</strong> splitter for<br />

<strong>the</strong> vacuum ultraviolet range: From <strong>the</strong> optical design to <strong>the</strong> first uv spectrum.<br />

Rev. Sci. Instrum., 80(4):043101, 2009.<br />

[246] P. Tzallas, D. Charalambidis, N. A. Papadogiannis, K. Witte, <strong>and</strong> G. D.<br />

Tsakiris. Sec<strong>on</strong>d-order autocorrelati<strong>on</strong> measurements of attosec<strong>on</strong>d xuv pulse<br />

trains. J. Mod. Opt., 52(2):321 <strong>–</strong> 338, 2005.<br />

[247] R. Mitzner, B. Siemer, M. Neeb, T. Noll, F. Siewert, S. Roling, M. Rutkowski,<br />

A. A. Sorokin, M. Richter, P. Juranic, K. Tiedtke, J. Feldhaus, W. Eberhardt,<br />

<strong>and</strong> H. Zacharias. Spatio-temporal coherence of free electr<strong>on</strong> laser pulses in <strong>the</strong><br />

soft x-ray regime. Opt. Expr., 16(24):19909<strong>–</strong>19919, 2008.<br />

[248] E. Hecht. Optics (4th Editi<strong>on</strong>). Addis<strong>on</strong> Wesley, 2001.<br />

[249] T. Sekikawa, T. Ohno, T. Yamazaki, Y. Nabekawa, <strong>and</strong> S. Watanabe. Pulse<br />

compressi<strong>on</strong> of a high-order harm<strong>on</strong>ic by compensating <strong>the</strong> atomic dipole phase.<br />

Phys. Rev. Lett., 83(13):2564<strong>–</strong>2567, 1999.


Bibliography 219<br />

[250] Y. Kobayashi, T. Ohno, T. Sekikawa, Y. Nabekawa, <strong>and</strong> S. Watanabe. Pulse<br />

width measurement of high-order harm<strong>on</strong>ics by autocorrelati<strong>on</strong>. Appl. Phys. B,<br />

70(3):389<strong>–</strong>394, 2000.<br />

[251] P. Tzallas, D. Charalambidis, N. A. Papadogiannis, K. Witte, <strong>and</strong> G. D.<br />

Tsakiris. Direct observati<strong>on</strong> of attosec<strong>on</strong>d light bunching. Nature,<br />

426(6964):267<strong>–</strong>271, 2003.<br />

[252] T. Nakajima <strong>and</strong> L. A. A. Nikolopoulos. Use of helium double i<strong>on</strong>izati<strong>on</strong> for<br />

autocorrelati<strong>on</strong> of an xuv pulse. Phys. Rev. A, 66(4):041402, Oct 2002.<br />

[253] P. Lambropoulos <strong>and</strong> L. A. A. Nikolopoulos. Angular distributi<strong>on</strong>s in double<br />

i<strong>on</strong>izati<strong>on</strong> of helium under xuv sub-femtosec<strong>on</strong>d radiati<strong>on</strong>. New J. Phys.,<br />

10(2):025012, 2008.<br />

[254] 4GLS C<strong>on</strong>ceptual Design Report Chapter 10, secti<strong>on</strong> 10.8.1.3. 2006.<br />

[255] D. J. Kane <strong>and</strong> R. Trebino. Characterisati<strong>on</strong> of arbitrary femtosec<strong>on</strong>d<br />

pulses using frequency-resolved optical characterisati<strong>on</strong> of arbitrary femtosec<strong>on</strong>d<br />

pulses using frequency-resolved optical characterisati<strong>on</strong> of arbitrary femtosec<strong>on</strong>d<br />

pulses using frequency-resolved optical grating. IEEE J. Quantum<br />

Electr., 29(2):571<strong>–</strong>579, 1993.<br />

[256] R. Trebino, K.W. DeL<strong>on</strong>g, D. N. Fittinghoff, J. N. Sweetser, M. A. Krumbügel,<br />

B. A. Richman, <strong>and</strong> D. J. Kane. Measuring ultrashort laser pulses in <strong>the</strong> timefrequency<br />

domain using frequency-resolved optical gating. Rev. Sci. Instrum.,<br />

68(9):3277<strong>–</strong>3295, 1997.<br />

[257] J. Norin, J. Mauritss<strong>on</strong>, A. Johanss<strong>on</strong>, M. K. Raarup, S. Buil, A. Perss<strong>on</strong>,<br />

O. Dühr, M. B. Gaarde, K. J. Schafer, U. Keller, C.-G. Wahlström, <strong>and</strong><br />

A. L’Huillier. Time-frequency characterizati<strong>on</strong> of femtosec<strong>on</strong>d extreme ultraviolet<br />

pulses. Phys. Rev. Lett., 88(19):193901, 2002.<br />

[258] T. Sekikawa, T. Katsura, S. Miura, <strong>and</strong> S. Watanabe. Measurement of <strong>the</strong><br />

intensity-dependent atomic dipole phase of a high harm<strong>on</strong>ic by frequencyresolved<br />

optical gating. Phys. Rev. Lett., 88(19):193902, 2002.<br />

[259] T. Tsang, M.A. Krumbügel, K.W. DeL<strong>on</strong>g, D.N. Fittinghoff, <strong>and</strong> R. Trebino.<br />

Frequency-resolved optical-gating measurement of ultrashort pulses using surface<br />

third-harm<strong>on</strong>ic generati<strong>on</strong>. Opt. Lett., 21(17):1381<strong>–</strong>1383, 1996.<br />

[260] I.A.WalmsleyC. Iac<strong>on</strong>is, V.W<strong>on</strong>g. Directinterferometric techniquesforcharacterising<br />

ultrashort optical pulses. IEEE J. Sel. Top. Quantum Electr<strong>on</strong>., 4:285<strong>–</strong><br />

294, 1998.<br />

[261] I.A. Walmsley C. Iac<strong>on</strong>is. Self-referencing spectral interferometry for measuring<br />

ultrashort optical pulses. IEEE J. Sel. Top. Quantum Electr<strong>on</strong>., 35:501<strong>–</strong>509,<br />

1999.<br />

[262] H. G. Muller. Rec<strong>on</strong>structi<strong>on</strong> of attosec<strong>on</strong>d harm<strong>on</strong>ic beating by interference<br />

of two-phot<strong>on</strong> transiti<strong>on</strong>s. Appl. Phys. B., 74(Suppl.):S17<strong>–</strong>S21, 2002.


220 Bibliography<br />

[263] O. Smirnova, V. S. Yakovlev, <strong>and</strong> M. Ivanov. Use of electr<strong>on</strong> correlati<strong>on</strong> to<br />

make attosec<strong>on</strong>d measurements without attosec<strong>on</strong>d pulses. Phys. Rev. Lett.,<br />

94(21):213001, 2005.<br />

[264] T. Maltezopoulos, S. Cunovic, M. Wiel<strong>and</strong>, M. Beye, A. Azima, H. Redlin,<br />

M. Krikunova, R. Kalms, U. Frühling, F. Budzyn, W. Wurth, A. Föhlisch,<br />

<strong>and</strong> M. Drescher. Single-shot timing measurement of extreme-ultraviolet freeelectr<strong>on</strong><br />

laser pulses. New J. Phys., 10(3):033026, 2008.<br />

[265] P. Krejcik. X-ray pulse length characterizati<strong>on</strong> using <strong>the</strong> surface magneto optic<br />

kerr effect. Technical report, SLAC-PUB-11946, 2006.<br />

[266] Fesca 200 / fesca 6860. www.hamamatsu.com.<br />

[267] C. B<strong>on</strong>té, M. Harm<strong>and</strong>, F. Dorchies, S. Magnan, V. Pitre, J.-C. Kieffer, P. Audebert,<br />

<strong>and</strong> J.-P. Geindre. High dynamic range streak camera for subpicosec<strong>on</strong>d<br />

time-resolved x-ray spectroscopy. Rev. Sci. Instrum., 78(4):043503, 2007.<br />

[268] www.fastlite.com.<br />

[269] J. Feng, H. J. Shin, J. R. Nasiatka, W. Wan, A. T. Young, G. Huang, A. Comin,<br />

J. Byrd, <strong>and</strong> H. A. Padmore. An x-ray streak camera with high spatio-temporal<br />

resoluti<strong>on</strong>. Appl. Phys. Lett., 91(13):134102, 2007.<br />

[270] J. Larss<strong>on</strong>. Ultrafast, jitter-free x-ray streak camera that uses single-phot<strong>on</strong><br />

counting. Opt. Lett., 26(5):295<strong>–</strong>297, 2001.<br />

[271] A. Pietzsch, A. Föhhlisch, M. Beye, M. Deppe, F. Hennies, M. Nagas<strong>on</strong>o,<br />

E. Suljoti, W. Wurth, C. Gahl, K. Döbrich, <strong>and</strong> A. Melnikov. Towards time<br />

resolved core level photoelectr<strong>on</strong> spectroscopy with femtosec<strong>on</strong>d x-ray freeelectr<strong>on</strong><br />

lasers. New J. Phys., 10(3):033004, 2008.<br />

[272] C. Bostedt, H. N. Chapman, J. T. Costello, J. R. Crespo López-Urrutia,<br />

S. Düsterer, S. W. Epp, J. Feldhaus, A. Föhlisch, M. Meyer, T. Möller,<br />

R. Moshammer, M. Richter, K. Sokolowski-Tinten, A. Sorokin, K. Tiedtke,<br />

J. Ullrich, <strong>and</strong> W. Wurth. Experiments at flash. Nucl. Instrum. Meth. Phys.<br />

A, 601(1-2):108 <strong>–</strong> 122, 2009.<br />

[273] J. M. Glownia, J. Cryan, J. Andreass<strong>on</strong>, A. Belkacem, N. Berrah, C. I. Blaga,<br />

C. Bostedt, J. Bozek, L. F. DiMauro, L. Fang, J. Frisch, O. Gessner, M. Gühr,<br />

J. Hajdu, M. P. Hertlein, M. Hoener, G. Huang, O. Kornilov, J. P. Marangos,<br />

A. M. March, B. K. McFarl<strong>and</strong>, H. Merdji, V. S. Petrovic, C. Raman, D. Ray,<br />

D. A. Reis, M. Trigo, J. L. White, W. White, R. Wilcox, L. Young, R. N. Coffee,<br />

<strong>and</strong> P. H. Bucksbaum. Time-resolved pump-probe experiments at <strong>the</strong> lcls. Opt.<br />

Express, 18(17):17620<strong>–</strong>17630, 2010.<br />

[274] A. Pietzsch, A. Föhlisch, M. Beye, M. Deppe, F. Hennies, M. Nagas<strong>on</strong>o,<br />

E. Suljoti, W. Wurth, C. Gahl, K. Döbrich, <strong>and</strong> A. Melnikov. Towards time<br />

resolved core level photoelectr<strong>on</strong> spectroscopy with femtosec<strong>on</strong>d x-ray freeelectr<strong>on</strong><br />

lasers. New J. Phys., 10(3):033004, 2008.


Bibliography 221<br />

[275] P. Johnss<strong>on</strong>, A. Rouzée, W. Siu, Y. Huismans, F. Lépine, T. Marchenko,<br />

S. Düsterer, F. Tavella, N. Stojanovic, H. Redlin, A. Azima, <strong>and</strong> M. J. J.<br />

Vrakking. Characterizati<strong>on</strong> of a two-color pump<strong>–</strong>probe setup at flash using a<br />

velocity map imaging spectrometer. Opt. Lett., 35(24):4163<strong>–</strong>4165, 2010.<br />

[276] G. Grübel. X-ray phot<strong>on</strong> correlati<strong>on</strong> spectroscopy at <strong>the</strong> european x-ray freeelectr<strong>on</strong><br />

laser (xfel) facility. Comptes Rendus Physique, 9(5-6):668 <strong>–</strong> 680, 2008.<br />

[277] B. D. Patters<strong>on</strong> <strong>and</strong> R. Abela. Novel opportunities for time-resolved absorpti<strong>on</strong><br />

spectroscopy at <strong>the</strong> x-ray free electr<strong>on</strong> laser. Phys. Chem. Chem. Phys.,<br />

12:5647<strong>–</strong>5652, 2010.<br />

[278] D. P. Bernstein, Y. Acremann, A. Scherz, M. Burkhardt, J. Stöhr, M. Beye,<br />

W. F. Schlotter, T. Beeck, F. Sorgenfrei, A. Pietzsch, W. Wurth, <strong>and</strong><br />

A. Föhlisch. Near edge x-ray absorpti<strong>on</strong> fine structure spectroscopy with xray<br />

free-electr<strong>on</strong> lasers. Appl. Phys. Lett., 95(13):134102, 2009.<br />

[279] S. P. Hau-Riege, R. A. L<strong>on</strong>d<strong>on</strong>, H. N. Chapman, A. Szoke, <strong>and</strong> N. Timneanu.<br />

Encapsulati<strong>on</strong> <strong>and</strong> diffracti<strong>on</strong>-pattern-correcti<strong>on</strong> methods to reduce <strong>the</strong> effect<br />

of damage in x-ray diffracti<strong>on</strong> imaging of single biological molecules. Phys. Rev.<br />

Lett., 98(19):198302, 2007.<br />

[280] M. J. Bogan, W. H. Benner, S. Boutet, U. Rohner, M. Frank, A. Barty, M. M.<br />

Seibert, F. Maia, S. Marchesini, <strong>and</strong> S. Bajt. Single particle x-ray diffractive<br />

imaging. Nano Lett., 8(1):310<strong>–</strong>316, 2008.<br />

[281] H. N. Chapman, P. Fromme, A. Barty, T. A. White, R. A. Kirian, A. Aquila,<br />

M. S. Hunter, J. Schulz, D. P. DeP<strong>on</strong>te, U. Weierstall, R. B. Doak, F. R.<br />

N. C. Maia, A. V. Martin, I. Schlichting, L. Lomb, N. Coppola, R. L. Shoeman,<br />

S. W. Epp, R. Hartmann, D. Rolles, A. Rudenko, L. Foucar, N. Kimmel,<br />

G. Weidenspointner, P. Holl, M. Liang, M. Bar<strong>the</strong>lmess, C. Caleman, S. Boutet,<br />

M. J. Bogan, J. Krzywinski, C. Bostedt, S. Bajt, L. Gumprecht, B. Rudek,<br />

B. Erk, C. Schmidt, A. Homke, C. Reich, D. Pietschner, L. Struder, G. Hauser,<br />

H. Gorke, J. Ullrich, S. Herrmann, G. Schaller, F. Schopper, H. Soltau, K.-U.<br />

Kuhnel, M. Messerschmidt, J. D. Bozek, S. P. Hau-Riege, M. Frank, C. Y.<br />

Hampt<strong>on</strong>, R. G. Sierra, D. Starodub, J. G. Williams, J. Hajdu, N. Timneanu,<br />

M. M. Seibert, J. Andreass<strong>on</strong>, A. Rocker, O. J<strong>on</strong>ss<strong>on</strong>, M. Svenda, S. Stern,<br />

K. Nass, R. Andritschke, C.-D. Schroter, F. Krasniqi, M. Bott, K. E. Schmidt,<br />

X. Wang, I. Grotjohann, J. M. Holt<strong>on</strong>, T. R. M. Barends, R. Neutze, S. Marchesini,<br />

R. Fromme, S. Schorb, D. Rupp, M. Adolph, T. Gorkhover, I. Anderss<strong>on</strong>,<br />

H. Hirsemann, G. Potdevin, H.Graafsma, B. Nilss<strong>on</strong>, <strong>and</strong> J. C. H. Spence. Femtosec<strong>on</strong>d<br />

x-ray protein nanocrystallography. Nature, 470(7332):73<strong>–</strong>77, 2011.<br />

[282] C. M. Kewish, P. Thibault, O. Bunk, <strong>and</strong> F. Pfeiffer. The potential for twodimensi<strong>on</strong>al<br />

crystallography of membrane proteins at future x-ray free-electr<strong>on</strong><br />

laser sources. New J. Phys., 12(3):035005, 2010.<br />

[283] M. M. Seibert, T. Ekeberg, F. R. N. C. Maia, M. Svenda, J. Andreass<strong>on</strong>,<br />

O. J<strong>on</strong>ss<strong>on</strong>, D. Odic, B. Iwan, A. Rocker, D. Westphal, M. Hantke, D. P.<br />

DeP<strong>on</strong>te, A. Barty, J. Schulz, L. Gumprecht, N. Coppola, A. Aquila, M. Liang,<br />

T. A. White, A. Martin, C. Caleman, S. Stern, C. Abergel, V. Seltzer, J.-M.


222 Bibliography<br />

Claverie, C. Bostedt, J. D. Bozek, S. Boutet, A. A. Miahnahri, M. Messerschmidt,<br />

J. Krzywinski, G. Williams, K. O. Hodgs<strong>on</strong>, M. J. Bogan, C. Y.<br />

Hampt<strong>on</strong>, R. G. Sierra, D. Starodub, I. Anderss<strong>on</strong>, S. Bajt, M. Bar<strong>the</strong>lmess,<br />

J. C. H. Spence, P. Fromme, U. Weierstall, R. Kirian, M. Hunter, R. B. Doak,<br />

S. Marchesini, S. P. Hau-Riege, M. Frank, R. L. Shoeman, L. Lomb, S. W.<br />

Epp, R. Hartmann, D. Rolles, A. Rudenko, C. Schmidt, L. Foucar, N. Kimmel,<br />

P. Holl, B. Rudek, B. Erk, A.Homke, C. Reich, D. Pietschner, G. Weidenspointner,<br />

L. Struder, G. Hauser, H. Gorke, J. Ullrich, I. Schlichting, S. Herrmann,<br />

G. Schaller, F. Schopper, H. Soltau, K.-U. Kuhnel, R. Andritschke, C.-D.<br />

Schroter, F. Krasniqi, M. Bott, S. Schorb, D. Rupp, M. Adolph, T. Gorkhover,<br />

H. Hirsemann, G. Potdevin, H. Graafsma, B. Nilss<strong>on</strong>, H. N. Chapman, <strong>and</strong><br />

J. Hajdu. Single mimivirus particles intercepted <strong>and</strong> imaged with an x-ray<br />

laser. Nature, 470(7332):78<strong>–</strong>81, 2011.<br />

[284] R. Treusch <strong>and</strong> J. Feldhaus. Flash: new opportunities for (time-resolved) coherent<br />

imaging of nanostructures. New J. Phys., 12(3):035015, 2010.<br />

[285] H. N. Chapman <strong>and</strong> K. A. Nugent. Coherent lensless x-ray imaging. Nat.<br />

Phot<strong>on</strong>, 4(12):833<strong>–</strong>839, 2010.<br />

[286] S. Boutet <strong>and</strong> G. J Williams. The coherent x-ray imaging (cxi) instrument at<br />

<strong>the</strong> linac coherent light source (lcls). New J. Phys., 12(3):035024, 2010.<br />

[287] C. M. Gün<strong>the</strong>r, B. Pfau, R. Mitzner, B. Siemer, S. Roling, H. Zacharias,<br />

O. Kutz, I. Rudolph, D. Sch<strong>on</strong>delmaier, R. Treusch, <strong>and</strong> S. Eisebitt. Sequential<br />

femtosec<strong>on</strong>d x-ray imaging. Nat. Phot<strong>on</strong>., 5(2):99<strong>–</strong>102, 2011.<br />

[288] A. A. Sorokin, S. V. Bobashev, T. Feigl, K. Tiedtke, H. Wabnitz, <strong>and</strong><br />

M. Richter. Photoelectric effect at ultrahigh intensities. Phys. Rev. Lett.,<br />

99(21):213002, 2007.<br />

[289] K. Motomura, H. Fukuzawa, L. Foucar, X.-J. Liu, G. Prümper, K. Ueda,<br />

N. Saito, H. Iway ama, K. Nagaya, H. Murakami, M. Yao, A. Belkacem, M. Nagas<strong>on</strong>o,<br />

A. Higashiya, M. Yabashi, T. Ishikawa, H. Ohashi, <strong>and</strong> H. Kimura.<br />

Multiple i<strong>on</strong>izati<strong>on</strong> of atomic arg<strong>on</strong> irradiated by euv free-electr<strong>on</strong> laser pulses<br />

at 62 nm: evidence of sequential electr<strong>on</strong> strip. J. Phys. B: At. Mol. Opt. Phys.,<br />

42(22):221003, 2009.<br />

[290] K. Hoffmann, B. Murphy, B. Erk, A. Helal, N. K<strong>and</strong>adai, J. Keto, <strong>and</strong> T. Ditmire.<br />

High intensity femtosec<strong>on</strong>d xuv pulse interacti<strong>on</strong>s with atomic clusters.<br />

High Ener. Dens. Phys., 6(2):185 <strong>–</strong> 189, 2010.<br />

[291] L.Fang, M.Hoener, O.Gessner, F.Tarantelli, S.T. Pratt, O.Kornilov, C.Buth,<br />

M. Gühr, E. P. Kanter, C. Bostedt, J. D. Bozek, P. H. Bucksbaum, M. Chen,<br />

R. Coffee, J. Cryan, M. Glownia, E. Kukk, S. R. Le<strong>on</strong>e, <strong>and</strong> N. Berrah. Double<br />

core-hole producti<strong>on</strong> in n2: Beating <strong>the</strong> auger clock. Phys. Rev. Lett.,<br />

105(8):083005, 2010.<br />

[292] Y.-P. Sun, Z. Rinkevicius, C.-K. Wang, S. Carniato, M. Sim<strong>on</strong>, R. Taïeb, <strong>and</strong><br />

F. Gel’mukhanov. Two-phot<strong>on</strong>-induced x-ray emissi<strong>on</strong> in ne<strong>on</strong> atoms. Phys.<br />

Rev. A, 82(4):043430, 2010.


Bibliography 223<br />

[293] B. W. J. McNeil <strong>and</strong> N. R. Thomps<strong>on</strong>. X-ray free-electr<strong>on</strong> lasers. Nature<br />

Phot<strong>on</strong>., 4:814<strong>–</strong>821, 2010.<br />

[294] P. Emma, R. Akre, J. Arthur, R. Bi<strong>on</strong>ta, C. Bostedt, J. Bozek, A. Brachmann,<br />

P. Bucksbaum, R. Coffee, F.-J. Decker, Y. Ding, D. Dowell, S. Edstrom,<br />

A. Fisher, J. Frisch, S. Gilevich, J. Hastings, G. Hays, Ph. Hering, Z. Huang,<br />

R. Ivers<strong>on</strong>, H. Loos, M. Messerschmidt, A. Miahnahri, S. Moeller, H.-D. Nuhn,<br />

G. Pile, D. Ratner, J. Rzepiela, D. Schultz, T. Smith, P. Stefan, H. Tompkins,<br />

J. Turner, J. Welch, W. White, J. Wu, G. Yocky, <strong>and</strong> J. Galayda. First lasing<br />

<strong>and</strong> operati<strong>on</strong> of an ˚angstrom-wavelength free-electr<strong>on</strong> laser. Nature Phot<strong>on</strong>,<br />

4(9):641<strong>–</strong>647, 2010.<br />

[295] T Shintake, H. Tanaka, T. Hara, T. Tanaka, K. Togawa, M. Yabashi, Y. Otake,<br />

Y. Asano, T. Bizen, T. Fukui, S. Goto, H. Atsushi Higashiya, T. Hir<strong>on</strong>o,<br />

N. Hosoda, T. Inagaki, S. Inoue, M. Ishii, Y. Kim, H. Kimura, M. Kitamura,<br />

T. Kobayashi, H. Maesaka, T. Masuda, S. Matsui, T. Matsushita, X. Marechal,<br />

M. Nagas<strong>on</strong>o, H. Ohashi, T. Ohata, T. Ohshima, K. Onoe, K. Shirasawa,<br />

T. Takagi, S. Takahashi, M. Takeuchi, K. Tamasaku, R. Tanaka, Y. Tanaka,<br />

T. Tanikawa, T. Togashi, S. Wu, A. Yamashita, K. Yanagida, C. Zhang, H. Kitamura,<br />

<strong>and</strong> T. Ishikawa. A compact free-electr<strong>on</strong> laser for generating coherent<br />

radiati<strong>on</strong> in <strong>the</strong> extreme ultraviolet regi<strong>on</strong>. Nat. Phot<strong>on</strong>., 2(9):555<strong>–</strong>559, 2008.<br />

[296] E.-S. Kim <strong>and</strong> M. Yo<strong>on</strong>. Beam dynamics in a 10-gev linear accelerator for <strong>the</strong><br />

x-ray free electr<strong>on</strong> laser at pal. IEEE T. Nucl. Sci., 56(6):3597<strong>–</strong>3606, 2009.<br />

[297] B. Faatz, N. Baboi, V. Ayvazyan, V. Bal<strong>and</strong>in, W. Decking, S. Duesterer,<br />

H.-J. Eckoldt, J. Feldhaus, N. Golubeva, K. H<strong>on</strong>kavaara, M. Koerfer, T. Laarmann,<br />

A. Leuschner, L. Lilje, T. Limberg, D. Noelle, F. Obier, A. Petrov,<br />

E. Ploenjes, K. Rehlich, H. Schlarb, B. Schmidt, M. Schmitz, S. Schreiber,<br />

H. Schulte-Schrepping, J. Spengler, M. Staack, F. Tavella, K. Tiedtke, M. Tischer,<br />

R. Treusch, M. Vogt, A. Willner, J. Bahrdt, R. Follath, M. Gensch,<br />

K. Holldack, A. Meseck, R. Mitzner, M. Drescher, V. Miltchev, J. Rönsch-<br />

Schulenburg, <strong>and</strong> J. Rossbach. Flash ii: Perspectives <strong>and</strong> challenges. Nucl.<br />

Instrum. Meth. A, In Press, Corrected Proof, 2010.<br />

[298] M. Gensch, L. Bittner, A. Chesnov, H. Delsim-Hashemi, M. Drescher, B. Faatz,<br />

J. Feldhaus, U. Fruehling, G.A. Gel<strong>on</strong>i, Ch. Gerth, O. Grimm, U. Hahn,<br />

M. Hesse, S. Kapitzki, V. Kocharyan, O. Kozlov, E. Matyushevsky, N. Morozov,<br />

D. Petrov, E. Ploenjes, M. Roehling, J. Rossbach, E.L. Saldin, B. Schmidt,<br />

P. Schmueser, E.A. Schneidmiller, E. Syresin, A. Willner, <strong>and</strong> M.V. Yurkov.<br />

New infrared undulator <strong>beam</strong>line at flash. Infrared Phys. Technol., 51(5):423 <strong>–</strong><br />

425, 2008.<br />

[299] B.Faatz, N. Baboi, V. Ayvazyan, V. Bal<strong>and</strong>in, W. Decking, S. Duesterer, H.-J.<br />

Eckoldt, J. Feldhaus, N. Golubeva, M. Koerfer, T. Laarmann, A. Leuschner,<br />

L. Lilje, T. Limberg, D. Noelle, F. Obier, A. Petrov, E. Ploenjes, K. Rehlich,<br />

H. Schlarb, B. Schmidt, M. Schmitz, S. Schreiber, H. Schulte-Schrepping,<br />

J. Spengler, M. Staack, F. Tavella, K. Tiedtke, M. Tischer, R. Treusch, A. Willner,<br />

J. Bahrdt, R. Follath, M. Gensch, K. Holldack, A. Meseck, R. Mitzner,<br />

M. Drescher, V. Miltchev, J. Rönsch-Schulenburg, <strong>and</strong> J. Rossbach. Flash ii: A<br />

seeded future at flash. In Proc. IPAC’10, May 2010. Koyoto, Japan.


224 Bibliography<br />

[300] G. Lambert, T. Hara, D. Garzella, T. Tanikawa, M. Labat, B. Carre, H. Kitamura,<br />

T. Shintake, M. Bougeard, S. Inoue, Y. Tanaka, P. Salieres, H. Merdji,<br />

O. Chubar, O. Gobert, K. Tahara, <strong>and</strong> M. E. Couprie. Injecti<strong>on</strong> of harm<strong>on</strong>ics<br />

generated in gas in a free-electr<strong>on</strong> laser providing intense <strong>and</strong> coherent extremeultraviolet<br />

light. Nat. Phys., 4(4):296<strong>–</strong>300, 2008.<br />

[301] E. Allaria, C. Callegari, D. Cocco, W. M. Fawley, M. Kiskinova, C. Masciovecchio,<br />

<strong>and</strong> F. Parmigiani. The fermi@elettra free-electr<strong>on</strong>-laser source for coherent<br />

x-ray physics: phot<strong>on</strong> properties, <strong>beam</strong> <strong>transport</strong> system <strong>and</strong> applicati<strong>on</strong>s.<br />

New J. Phys., 12(7), 2010.<br />

[302] E. Allaria. The sec<strong>on</strong>d stage of fermi@elettra: a seeded fel in <strong>the</strong> x-ray spectral<br />

range. In 31 st Internati<strong>on</strong>al FEL c<strong>on</strong>f., August 2009. Liverpool, UK.<br />

[303] J. N. Galayda, J. Arthur, D. F. Ratner, <strong>and</strong> W. E. White. X-ray free-electr<strong>on</strong><br />

lasers—present <strong>and</strong> future capabilities. J. Opt. Soc. Am. B, 27(11):B106<strong>–</strong>B118,<br />

2010.<br />

[304] https://slacportal.slac.stanford.edu/sites/lcls public/lcls<br />

ii/Pages/default.aspx.<br />

[305] W. Ackermann et al. Operati<strong>on</strong> of a free-electr<strong>on</strong> laser from <strong>the</strong> extreme ultraviolet<br />

to <strong>the</strong> water window. Nat. Phot<strong>on</strong>., 1, 2007.<br />

[306] http://www.psi.ch/swissfel/CurrentSwissFELPublicati<strong>on</strong>sEN/SwissFEL<br />

Injector CDR 310810.pdf.<br />

[307] http://www.c<strong>and</strong>le.am/seminars/PDF 2009 11 19/Grigoryan B.pdf.<br />

[308] T. Garvey. Status of <strong>the</strong> psi x-ray free electr<strong>on</strong> laser. Talk at <strong>the</strong> FEL2010<br />

c<strong>on</strong>ference, Malmö, Sweden, Aug., 2010.<br />

[309] J. S. Oh, D. E. Kim, E. S. Kim, S. J. Park, H. S. Kang, T. Y. Lee, T. Y. Koo,<br />

S. S. Chang, C. W. Chung, S. H. Nam, <strong>and</strong> W. Namkung. 0.3-nm sase-fel at<br />

pal. Nucl. Instrum. Meth. A, 528(1-2):582 <strong>–</strong> 585, 2004.<br />

[310] B. McNeil. Free-electr<strong>on</strong> lasers: A down-sized design. Nat. Phot<strong>on</strong>, 2(9):522<strong>–</strong><br />

524, 2008.<br />

[311] K. Nakajima. Compact x-ray sources: Towards a table-top free-electr<strong>on</strong> laser.<br />

Nat. Phys., 4(2):92<strong>–</strong>93, 2008.<br />

[312] V. Malka, J. Faure, Y. A. Gauduel, E. Lefebvre, A. Rousse, <strong>and</strong> K. T. Phuoc.<br />

Principles <strong>and</strong> applicati<strong>on</strong>s of compact laser-plasma accelerators. Nat. Phys.,<br />

4(6):447<strong>–</strong>453, 2008.<br />

[313] H.-P. Schlenvoigt, K. Haupt, A. Debus, F. Budde, O. Jackel, S. Pfotenhauer,<br />

H. Schwoerer, E. Rohwer, J. G. Gallacher, E. Brunetti, R. P. Shanks, S. M.<br />

Wiggins, <strong>and</strong> D. A. Jaroszynski. A compact synchrotr<strong>on</strong> radiati<strong>on</strong> source driven<br />

by a laser-plasma wakefield accelerator. Nat. Phys., 4(2):130<strong>–</strong>133, 2008.<br />

[314] J. Corlett, J. Byrd, W. M. Fawley, M. Gullans, D. Li, S. M. Lidia, H. Padmore,<br />

G. Penn, I. Pogorelov, J. Qiang, D. Robin, F. Sannibale, J. W. Staples,<br />

C. Steier, M. Venturini, S. Virostek W. Wan, R. Wells, R. Wilcox, J. Wurtele,<br />

<strong>and</strong> A. Zholents. A high repetiti<strong>on</strong> rate vuv-soft x-ray fel c<strong>on</strong>cept. IEEE, 2007.


Bibliography 225<br />

[315] S. J. Russell, K. A. Bishofberger, B. E. Carlsten, D. C. Nguyen, <strong>and</strong> E. I.<br />

Smirnova. C<strong>on</strong>ceptual design of a 20 gev electr<strong>on</strong> accelerator for a 50 kev<br />

x-ray free-electr<strong>on</strong> laser using emittance exchange optics <strong>and</strong> crystallographic<br />

mask. In Proc. PAC’09 C<strong>on</strong>f., page TH5PFP036, May 2009. Vancouver, B. C.,<br />

Canada.<br />

[316] Harm<strong>on</strong>ics in a quantum free electr<strong>on</strong> laser: Towards a compact, coherent<br />

[gamma]-ray source. Opt. Comm., 284(4):1004 <strong>–</strong> 1007, 2011.<br />

[317] E. G. Bess<strong>on</strong>ov. X-ray free-prot<strong>on</strong> lasers. Ultrashort Wavelength Lasers II,<br />

2012(1):287<strong>–</strong>294, 1994.


This <str<strong>on</strong>g>compendium</str<strong>on</strong>g> serves both as a summary of <strong>the</strong> reports generated<br />

by <strong>the</strong> <strong>IRUVX</strong> experts <strong>on</strong> optics <strong>and</strong> phot<strong>on</strong> diagnostics <strong>and</strong> as an<br />

introducti<strong>on</strong> to Free Electr<strong>on</strong> Lasers, <strong>the</strong>ir functi<strong>on</strong>, <strong>and</strong> some of <strong>the</strong><br />

necessary technologies to make <strong>the</strong>m work.<br />

The ambiti<strong>on</strong> has been to encompass <strong>the</strong> subject of Free Electr<strong>on</strong> Laser<br />

technology in an introductory manner (part I), which supports <strong>the</strong> ra<strong>the</strong>r<br />

deep excursi<strong>on</strong> into phot<strong>on</strong> diagnostic methods that follows (part II). The<br />

list of references is extensive <strong>and</strong> could be used for more in depth studies.<br />

Members of <strong>IRUVX</strong>-<strong>PP</strong> WP7 <strong>and</strong> WP3 Expert Groups Phot<strong>on</strong> Beam<br />

Transport <strong>and</strong> Diagnostics <strong>and</strong> Metrology for FEL Optics:<br />

Rafael Abela, Günter Brenner, Anna Bianco, Mari<strong>on</strong> Bowler, Roberto<br />

Cimino, Daniele Cocco, Henrik Enquist, Uwe Flechsig, Christopher Gerth,<br />

Anth<strong>on</strong>y Glees<strong>on</strong>, Fini Jastrow, Ulf Johanss<strong>on</strong>, Libor Juha, Pavle Juranić,<br />

Barbara Keitel, Jörgen Larss<strong>on</strong>, Andreas Lindblad, Eric Louis, Bernd<br />

Löchel, Rolf Mitzner, Paul Morin, Robert Nietubyć, Luca Poletto, Paul<br />

Radcliffe, Amparo Rommeveaux, Mark Roper, Frank Siewert, Ryszard<br />

Sobierajski, Andrey Sorokin, Giovanni Sostero, Sibylle Spielmann-Jaeggi,<br />

Svante Svenss<strong>on</strong>, Cristian Svetina, Muriel Thomasset, Kai Tiedtke, Peter<br />

van der Slot, Hubertus Wabnitz, Christian Weniger, Marco Zangr<strong>and</strong>o.<br />

This work is supported by <strong>IRUVX</strong>-<strong>PP</strong>, an EU co-funded project under FP7<br />

(Grant Agreement 211285).<br />

www.eurofel.eu

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!