Production of gamma rays by pulsed laser beam Compton scattering off GeV-electrons using a non-planar four-mirror optical cavity

1.6 MB·English

by  T. Akagi

PreprinttypesetinJINSTstyle-HYPERVERSION
Production of gamma rays by pulsed laser beam
Compton scattering off GeV-electrons using a
non-planar four-mirror optical cavity
2
1
T. Akagie, S. Arakid, J. Bonisa, I. Chaikovskaa∗, R. Chichea, R. Cizerona, M. Cohena,
0
2 E. Cormierb, P. Cornebisea, N. Deleruea, R. Flaminioc, S. Funahashid, D. Jehannoa,

n Y. Hondad, F. Labayea, M. Lacroixa, R. Mariea, C. Michelc, S. Miyoshie, S. Nagatae,
a
T. Omorid, Y. Peinauda, L. Pinardc, H. Shimizud, V. Soskova, T. Takahashie,
J
4 R. Tanakae, T. Terunumad, J. Urakawad, A. Variolaa, F. Zomera.

aLAL,CNRS-IN2P3,UniversitéParis-SudXI
]
h
CentreScientifiqued’Orsay,Bâtiment200,BP34,91898Orsaycedex,France
p
bCELIA,CNRS,UniversitédeBordeaux
-
c DomaineduHautCarré,43,RuePierreNoailles,33405Talence,France
c
a cLMA,CNRS-IN2P3,UniversitéClaudeBernardLyonI
s. BâtimentVirgo7,AvenuePierredeCoubertin,69622Villeurbannecedex,France
c dATF,HighEnergyAcceleratorResearchOrganization(KEK)
i
s 1-1Oho,Tsukuba,305-0801Ibaraki,Japan
y
eHiroshimaUniversity
h
p 1-3-2Kagamiyama,Higashi-Hiroshima739-8511,Japan
[ E-mail: [email protected]

2
v
4 ABSTRACT: As part of the positron source R&D for future e+−e− colliders and Compton based
3
8 compact light sources, a high finesse non-planar four-mirror Fabry-Perot cavity has recently been
5 installed at the ATF (KEK, Tsukuba, Japan) [1]. The first measurements of the gamma ray flux
.
1 producedwithasuchcavityusingapulsedlaserispresentedhere. Wedemonstratetheproduction
1
1 of a flux of 2.7 ± 0.2 gamma rays per bunch crossing (∼3×106 gammas per second) during the
1 commissioning.
:
v
i
X KEYWORDS: Positronsproduction;Comptonscattering;Fabry-Perotcavity;Gammarays;ATF.
r
a
∗Correspondingauthor.
Contents
1. Introduction 1
2. Experimentalsetup 2
2.1 AcceleratorTestFacility 2
2.2 Fabry-Perotcavity 2
2.3 Calorimeter 3
3. Expectedgammarayflux 4
4. Measurements 5
4.1 Scanningprocedureandfirstdata 5
4.2 Exampleofdata 6
4.3 Dataanalysis 7
4.4 Calibration 10
5. Results 11
6. Conclusionandoutlook 15
1. Introduction
HighEnergyphysics[2,3,4]aswellasappliedphysics[5,6,7,8]areshowingagreatinterestfor
intensefluxofhighenergyX-raysandgammarays.
ThesegammaoralternativelyX-rayscanbegeneratedbyComptonscattering[9]. Toachieve
ahighfluxofhighenergyphotonsdespitethelowcrosssectionofComptonscatteringonerequires
ahighaveragepowerlasersystembasedonaFabry-Perotcavity[10]togetherwithahighcurrent
electronbeam[11]. Inthiscontext,atwo-mirrorFabry-Perotcavityhasalreadybeensuccessfully
operatedattheAcceleratorTestFacility[12](ATF)ofKEK[13,14].
A solution to produce a high flux of circularly polarized gamma rays is to use a four-mirror
Fabry-Perot cavity (FPC) where laser pulses can be stacked to reach a high average power at the
interaction point (IP). A prototype of non-planar high finesse four-mirror Fabry-Perot cavity has
beeninstalledattheKEKATFandisdescribedindetailsin[1]. Theopticalsystemhasbeencom-
missionedduringsummer2010andelectron-photonComptoncollisionswereobservedonthefirst
attempt in October 2010. In this paper we present measurements of the gamma ray flux recorded
duringourcommissioning. Thedataanalyzedhereweretakenbeforethetragicearthquakewhich
struckJapaninMarch2011. Recoveryandimprovementworkisbeingcarriedoutsincebutisnot
coveredbythispaper.
–1–
Thispaperisorganisedasfollows. Section2describestheexperimentalsetupusedtoproduce
and measure the gamma rays. Section 3 discusses the gamma ray flux expected with our current
setup. Section 4 explains how the data were analyzed. Section 5 presents the gamma ray spectra
measured. Finally,themainresultsandfurtherstepsarediscussedinsection6.
2. Experimentalsetup
2.1 AcceleratorTestFacility
A detailed description of the ATF at KEK can be found in [12, 15, 16]. The 1.28 GeV damping
ring (DR) has a revolution period of 462 ns and operates at the radio frequency (RF) of 357 MHz
(165 RF buckets spaced by 2.8ns). Although a total of up to 3 trains of 10 bunches separated by
5.6nscanbeinjectedinthering,mostoperationsrunwithasinglebunchinthetrainorwithatrain
of up to 10 bunches [17]. Our FPC is installed in one of the straight sections of the damping ring
as shown on figure 1. After the collisions the gamma rays, propagating along the electron beam,
areextractedthroughawindowbeforepassingthroughseveralcollimators. Theyaremeasuredby
agammadetectorabout20mdownstreamtheIP.
ATF2 beam line
Detector
~20 m Gamma rays
Electron beam 4-mirror Fabry-Perot
cavity
m
6 53.4 m
7.
2
Damping Ring
Photo-cathode RF gun
1.3 GeV S-band LINAC
Figure1: LayoutoftheAcceleratorTestFacilityatKEK.Theredstarindicatestheapproximatelo-
cationofthe4-mirrorFabry-Perotcavitydescribedin[1]andusedforthemeasurementspresented
in this paper. The red arrow indicates the direction of propagation of the gamma rays produced.
Thegreyboxindicatesthegammaraydetector(adaptedfrom[14]).
ThemainparametersoftheelectronbeamattheATFDRaregiveninthetable1.
2.2 Fabry-Perotcavity
TheFPCiscomposedof2concavemirrorswitharadiusofcurvatureof0.5mandtwoflatmirrors
with a non-planar tetrahedron geometry. The mirrors have a very high reflectivity (1 - 1060 ppm
for one of them and 1 - 330 ppm for the others) leading to a cavity finesse of the order of 3000
–2–
Table1: ATFDRparameters.
Description Value
Electronenergy,E 1.28GeV
e
Electronbunchpopulation,N ∼0.5×1010
e
Electronbunchlength,σ /c=τ ∼20 ps
z e
Electronbeamsize,σ /σ ∼110/10 µm
x y
Revolutionperiod,T 462 ns
0
Emittance,γε 5×10−6 /3×10−8 m·rad
x/y
(correspondingtoapassivepowerenhancementofabout1000). Aroundtriptimeof5.6nsinthe
FPCissettomatchhalftheRFfrequencyoftheATFDRat178.5MHz.
A passively mode-locked oscillator is amplified in a microstructured active fiber and further
injected in the FPC. An all-digital double feedback system is used to lock the cavity on the ATF
clockandtoensurethatthelaserpulsesareproperlystackedwithinterferometricaccuracy.
TheFPCismountedonanopticaltablewhichisitselfmountedonmoversthatallowtocontrol
preciselytheverticalpositionoftheFPCwithrespecttothatofthelaserbeam.
Table 2 summarises the main parameters of the system and more details on the experimental
apparatuscanbefoundin[1].
Table2: Parametersofthelasersystemusedduringthedatatakingpresentedhere.
Description Value
Laserphotonenergy,E 1.2 eV(λ =1032nm)
ph ph
Laserspotsize,σ /σ 26/38 µm
x y
Laserrepetitionrate, f 178.5 MHz±4KHz
L
Finesse,F ∼3000
Laserpulselength,τ 68 ps
L
AveragepowerstoredinFPC,P ∼160 W
L
Crossingangle,φ 8 deg.
AsthereisanoddnumberofRFbucketsintheDRcollisionsonagivenelectronbunchoccur
onlyeveryotherturn,atafrequencyof1.08MHz.
Dataacquisition Thedataacquisitionreliesonseveraloscilloscopestoacquireinformationfrom
the FPC and from the calorimeter. The waveforms acquired are timestamped in a database and
saved on disk for later analysis. Data from the accelerator such as beam position and charge are
alsoacquiredandsavedinthedatabase.
2.3 Calorimeter
To detect the gamma rays produced by the collisions between the laser photons and the electrons,
weuseafastscintillationdetectormadeofbariumfluoride(BaF )coupledwithaPhotomultiplier
2
–3–
Tube(PMT).TwopolishedrectangularcrystalsofBaF withdimensions100mm×70mm×70mm
2
are glued to each other and wrapped into an absorbing tape. The 200 mm depth of the BaF
2
calorimeterisnearly10radiationlengths,ensuringhighdetectionefficiency[18].
Bariumfluorideiscommonlyusedasithastwoemissionspectrapeaksat220nmand310nm
with decay time constants of about 0.8 ns and 630 ns respectively. The decay time of the fast
component of BaF allows to resolve two successive pulses of gamma rays generated from the
2
electronbunchesspacedby5.6ns. Toeliminatetheslowcomponentofthescintillationanoptical
filterhasbeeninstalledinfrontofthePMT.AfastPMT(HamamatsuPhotonicsR3377)witharise
timeof0.7nsisused. DataacquisitionisperformedusingaLeCroyWS454oscilloscope(1GS/s,
500MHzbandwidth).
GEANT4simulations UsingGeant4[19]wesimulatedthedetectortostudythedevelopmentof
electromagneticshowersinsidetheBaF calorimeter.
2
Thisallowstoperformrealisticsimulationsofhighenergygammaspassingthroughthecalorime-
ter. Theinitialdistributionofthegammaraysisobtainedbyusingthebeam-beaminteractioncode
CAIN2.40 [20]. Once a gamma ray hits the calorimeter, scintillation light is uniformly emitted
along the path of the charged particles produced by the electromagnetic shower. The number of
opticalphotonsgeneratedisproportionaltotheenergylossoftheinitialgammarays. Thescintil-
lationlightthenpropagatesandfinallyreachesthePMTlocatedattheendofthescintillator.
3. Expectedgammarayflux
UsingtheComptonscatteringphysicalpropertiesitispossibletoestimatetheexpectednumberof
scatteredgammaraysintheexperiment. Neglectingthedivergenceoftheelectronandlaserbeam
attheIP,theluminosityfortheComptoncollisionsisgivenbyequation3.1in[21]:
cos(φ/2) 1
L =N N f , (3.1)
e ph (cid:113) (cid:113)
2π σ2 +σ2 (σ2 +σ2)cos2(φ/2)+(σ2 +σ2 )sin2(φ/2)
ye yph xph xe ze zph
dN
F = γ =σ ·L, (3.2)
Compton
dt
where N , N are the number of the electrons in the bunch and number of the photons in the
e ph
laserpulserespectively, f istherepetitionfrequency,φ istheangleofthecollisions,σ andσ
(cid:126)re (cid:126)rph
indicaterespectivelytheRMSsizesoftheelectronbunchandthelaserpulse. Once,theluminosity
oftheprocessisdefined,thefluxofgammaraysisgivenbyformula3.2,whereσ isthetotal
Compton
Comptonscatteringcrosssectionwhichisdeterminedbythemomentaoftheincidentelectronand
laserphoton[22].
Formulas 3.1 and 3.2 show that the emitted rate is inversely proportional to the transverse
electronandlaserbeamsizes. Theseformulasshowsalsothatacollisionanglereducesthegamma
rayfluxespeciallyinthecaseoflongelectronbunchesandlaserpulses.
The simulation code CAIN has been used to simulate the Compton scattering process. This
codesimulatestheinteractionbetweenoneelectronbunchandonelaserpulsewheretheCompton
cross section is calculated for all the possible polarization states [22] and takes into account the
–4–
beam geometry. Using the parameters listed in table 1 and 2 the energy spectrum and expected
numberofthescatteredgammaraysperbunchcrossinghavebeenobtained.
Asitwasmentionedbefore,intheexperimentthescatteredgammarayspassthroughseveral
collimatorsbeforeenteringthedetector. Thiscarriesoutagammaenergyspectrumselectionowing
totheenergy-anglecorrelationinComptonscattering. Figure2showsthesimulatedenergyspec-
trumofthegammarayswheretheredcolorreferstothegammaraystransmittedbythecollimators
andbluecolorreferstothegammaraysfilteredoutduetotheirenergy/angle.
The average energy of the gamma rays produced is ∼15 MeV. However, due to the limited
geometricalaperture,thegammaraysbelow15MeVdonotreachthedetector. Theaverageenergy
of the gamma rays reaching the detector is therefore 24 MeV. Later, these results are used for the
Geant4simulationandcalibrationofthecalorimeter.

Gamma rays blocked by the collimators
300 Gamma rays reaching the detector
250
200
s
e
Entri150
100
50
0
0 5 10 15 20 25 30
Energy [MeV]
Figure 2: Energy spectrum of the gamma rays. The energies of the gamma rays blocked by the
collimators are shown as well as the energies of the gamma rays entering the detector. Only the
gammarayswithenergiesabove15MeVareacceptedbythecollimators.
4. Measurements
4.1 Scanningprocedureandfirstdata
TheFPCwascommissionedinOctober2010andComptoncollisionswererecordedonthefirstat-
tempt(26October). Beforedatatakingweinvestigatedhowtheintensityofthesignalisaffectedby
thepositionofthelaserbeamwaistwithrespecttotheelectronbeampositionandsynchronization
betweenthelaserandelectronbeambydoingthepositionandthephasescans.
The search for the collisions area between the electrons and the laser photons has several
degrees of freedom and is not straightforward. The two most sensitive degrees of freedom are
the vertical position of the FPC (dimension orthogonal to the plane of the beams) and the relative
phasebetweenthelaserandtheelectrons(timedimension). Thesearchforcollisionsisperformed
inseveralsteps:
–5–
• TheFPCfrequencyisoffsetbyatleast100HzwithrespecttotheATFfrequency. Bydoing
this the relative phase between the electrons and the laser is automatically scanned at a rate
ofatleast100Hz.
• TheverticalpositionoftheFPCissweptslowly.
• Whileperformingthisverticalscanwelookonthedataacquisitionoscilloscopefortheap-
paritionofshortburstsintheoutputofthePMT(seefigure3). Thelengthandthefrequency
of these bursts are related to the difference between the frequency of the cavity and the fre-
quencyoftheATFclock.
• Oncetheseburstshavebeenfoundtheverticalpositionofthelaserisadjustedtomaximize
theirintensity.
• AtthisstagethelaserfrequencycanbelockedontheATFfrequency.
• The phase of the laser with respect to the electron beam must be adjusted to optimize the
intensityofthesignalobservedonthePMToutput1.
Figure3: Exampleofsignalobservedonouroscilloscopewhilescanningtheparametersspaceto
findthecollision area. The yellowlineisthe signalfromthePMT.The durationofthewaveform
is10msandtheverticalscaleis200mV(onthisdaythesignalwasamplified).
Theburstscorrespondtothetimeatwhichthephaseofthelasermatchesthatoftheelectrons.
ThedistancesbetweentheburstscorrespondstothedifferencebetweentheATFfrequencyandthe
laser frequency. Each burst contains several peaks spaced by the duration of two DR revolution.
The red line records the injection trigger (the trigger itself is off the screen). It was checked that
the bursts on the yellow lines disappear when the cavity is moved vertically by a distance greater
thanthebeamsize.
4.2 Exampleofdata
During data taking we record the signal from the PMT as well as the 357 MHz ATF clock and
the laser power transmitted by the FPC measured by a photodiode. To avoid the mistaking noise
from the injection or extraction kickers a typical data acquisition starts at least 200 ms after the
injectiontriggerisreceived. Atthistimeweexpectthebeamtobealmostfullydamped. Atypical
signal waveform from the calorimeter can be observed on figure 4. A full waveform contains
approximately200000samplesspacedby1ns.
1Ourexperienceshowsthatthevariationofthisphasefromdaytodayissmallhoweverwedoobservesmalldrifts
whichrequireanewphasescanoneachrun.
–6–
−3
x 10
2
0
−2
V] −4
e [ −6
g
a
olt −8
V
−10
−12
−14
0 1 2
Time [s] −4
x 10
Figure4: ThetypicalsignalshapeforthehighenergygammaraysproducedbyComptonscattering
observed at the anode output of the PMT. Each spike on the picture corresponds to the gamma
productionaftersuccessivebunchcrossingsover0.2ms. Datashowncorrespondtoasinglebunch
storedintheATFDR.
4.3 Dataanalysis
TherawdatarecordedcontaintheintensityofthePMToutputasafunctionoftime(seefigure4).
DataanalysiswascarriedouttoextracttheintensityoftheComptonsignalandremoveunwanted
backgrounds.
AsthedatawereacquiredduringthedifferentATFrunswithdifferentfillingmodeoftheDR
(1 train, 2 trains, 3 trains stored in the ATF DR) the correct number of Compton peaks and their
timingmustbefound. The357MHzATFclockisusedtodefinethebeginningofa924nsperiods
corresponding to the occurrence of the Compton signal (two ATF DR revolutions). All 924 ns
periods belonging to one data file (usually 0.2 ms long) are superimposed on top of each other to
find the number of the peaks per revolution and their time of arrival with respect to the beginning
of the period. Such technique is used to enhance the signal over the background reducing at the
sametimetheelectronicnoise. Thisresultsinamoreprecisewayfortheestimationofthenumber
ofthepeaksandtheirtiming. Anexampleofstackingcanbeseenonfigure5a.
Once the number of peaks and their positions within the period is found we define a gate
aroundthepositionwheretheComptonsignalisexpected. Weusethisgatetocalculatetheheight
and integral of the peak. The Compton peak height and its integral are our measure of the energy
depositedbythegammaraysinthecalorimeter. Thelengthofthegateissetto12nsinordertoen-
tirelycontainthesignalthusensuringacorrectevaluationoftheenergydeposited. Thedistribution
ofsuchgatesaroundthepeakispresentedonfigure5b.
ThebackgroundlevelanditsRMSarecalculatedwithineachperiodandaresubtractedfrom
thecorrespondingpeakheightandpeakintegral. Thebackgroundlevelisdefinedasthemeanover
thetimebasecorrespondingtoagivenperiodexcludingthegatecontainingthesignal. Theaverage
background level and its RMS for our data sample is estimated to be around 7 µV and 0.2 mV
respectively. Finally, the peak height and peak integral are calculated for every peak within the
–7–
x 10−3
0.1 2
0 0
g [V]−0.1 −2
stackin−0.2 age [V] −4
Voltage −−00..43 Volt −−86
−0.5 −10
−0.6 −12
0 0.2 0.4 0.6 0.8 1 0 5 10 15
Time [s] x 10−6 Time [ns]
(a) (b)
Figure 5: Stacking of all the 924 ns periods from the data file presented on figure 4 (figure 5a).
This stacking is used to find the number of the peaks and their timing. Time profile of the signals
foralltheperiodsfromthesamefileisshownonfigure5b.
period(seefigure6a).
Theshapeofthedetector’sresponsecreatesalinearrelationbetweenthetotalchargeandthe
maximum charge measured. This can be seen as a correlation between the calculated peak height
andpeakintegralasshownonfigure6b
x 10−11
0
0
−0.002
s]
V
V]−0.004 k [
ht [ pea−2
Peak heig−−00..000086 gral of the
e
−0.01 Int−4
−0.012
0 50 100 150 200 −0.012 −0.01 −0.008 −0.006 −0.004 −0.002 0
Period number Peak height [V]
(a) (b)
Figure6: Figure6agivesanexampleofthepeakheightdistributionfromthedatafilepresentedon
figure4. Peakheightvs. peakintegralforthesamedatafileisshownonfigure6b.
Quality of the data. Different quality cuts are applied to restrict the analysis to a high purity
sample. By putting limitations on them we can reject noise and obtain a set of good quality data
(seefigure7).
We want all the Compton peaks within one file to have the same phase with respect to the
ATF clock. It22 is achieved, in this case, by requiring the Compton signal to arrive 6 ns after the
–8–
beginningofthegate(seefigure7a).
Sometimes, the noise can dominate the signal (see figure 7b). Quite often, these signals can
evenleadtoanintegralvaluewiththewrongsign. Thisisillustratedonfigure6bbythedotswith
positive values of the integral of the peak. Such events have to be considered as picked up noise
andmustberejected. Forthis,weintroduceavariable:
150 1x 10−3 5x 10−11Data kept after signal arrival time cut
0 Data rejected by cut on pulse shape
0
Entries10500 Voltage [V]−−−321 eak integral [V s] −−105
P
−4 −15
00 2 4 6 8 10 12 −50 5 10 15 −0.03 −0.02 −0.01 0
Signal arrival time within a gate [ns] Time [ns] Peak height [V]
(a) (b) (c)
Figure7: Qualitycutillustration(figures7aand7b)showntogetherwiththeireffectonallthedata
sample (figure 7c). Figure 7a: the distribution of the signal arrival time within the gate for all the
signalsfoundinthedatafilepresentedonfigure4. Figure7bgivesanexampleofacorrectsignal
(magenta line) as opposed to the noise signal (blue, red line). Figure 7c shows the data sample
remainingafterthecutonthesignalarrivaltimeapplied(magentapoints)togetherwiththeevents
whicharerejectedduetothecutontheshapeofthepulse(datapointstaggedinblack).
2I (t )
max 0
V = , (4.1)
I(t −1)+I(t +1)
0 0
where I (t ) is the intensity of the signal taken at the time when it reaches maximum, I(t −1)
max 0 0
andI(t +1)aretheintensitiesofthesignalinthetwonearestdatapoints. Byimposing1≤V ≤10
0
thenoiseisfilteredoutasV reacheseithernegativeorveryhighpositivevaluesfornoisesignal. On
figure7bthemagentalinecorrespondstotherealComptonsignalforwhichourestimatorvariable
V isabout1whiletheredandbluelinesareforthenoiseeventsandinthiscaseV equalsto-2and
17respectively.
The effect of the cuts mentioned above is shown on figure 7c. Cut on the signal arrival time
preserve good linear relation between the peak height and peak integral apart from a few high
intensity data points for which our data acquisition system saturated. The cut on the shape of the
pulsehelpstoremovenoise. However,somelowintensityeventsstillremainwhichcorrespondto
thelowenergydepositedbythegammaraysinthecalorimeter.
Approximately half of the data taken and presented in this study have been rejected after the
differentqualitycutshavebeenapplied. Alltheresultscommentedbelowarebasedonhighquality
datasample.
Background While analysing the data we noticed that some data files contain peaks spaced by
about462nscorrespondingtooneATFDRrevolutionperiod. Withtheexperimentalsetupweuse
–9–