EPJWebofConferenceswillbesetbythepublisher DOI:willbesetbythepublisher (cid:13)c Ownedbytheauthors,publishedbyEDPSciences,2016 Fast Timing for High-Rate Environments with Micromegas ThomasPapaevangelou1,a, DanielDesforge1, EstherFerrer-Ribas1, IoannisGiomataris1, Cyprien Godinot1, Diego Gonzalez Diaz2, Thomas Gustavsson3, Mariam Kebbiri1, Eraldo Oliveri2, Filippo Resnati2, Leszek Ropelewski2, 6 1 GeorgiosTsiledakis1,RobVeenhof2,andSebastianWhite2,4 0 1CEA,IRFU,Centred’EtudesNucléairesdeSaclay,Gif-sur-Yvette91191,France 2 2EuropeanOrganizationforNuclearResearch(CERN),Genève1211,Switzerland n 3LaboratoireInteractions,DynamiquesetLasers(LIDyL)CEA,CNRS,UniversitéParis-Saclay,GifsurYvette91191,France a 4PrincetonUniversity,Princeton,NewJersey08544,USA J 2 1 Abstract.Thecurrentstateoftheartinfasttimingresolutionforexistingexperimentsisoftheorderof100ps onthetimeofarrivalofbothchargedparticlesandelectromagneticshowers.CurrentR&Donchargedparticle ] timingisapproachingthelevelof10psbutisnotprimarilydirectedatsustainedperformanceathighratesand t e underhighradiation(aswouldbeneededforHL-LHCpileupmitigation).WedemonstrateaMicromegasbased d solutiontoreachthislevelofperformance. TheMicromegasactsasaphotomultipliercoupledtoaCerenkov- - s radiatorfrontwindow,whichproducessufficientUVphotonstoconvertthe∼100pssingle-photoelectronjitter n intoa timingresponse of the order of 10-20 ps per incident charged particle. A prototype has been built in i ordertodemonstratethisperformance. Thefirstlaboratorytestswithapico-second laserhaveshownatime . s resolutionoftheorderof27psfor∼50primaryphotoelectrons,usingabulkMicromegasreadout. c i s y h 1 Introduction also the advantagesof being low-cost, simple to operate, p andarepotentiallyinsensitivetoirradiationeffects. [ From2025onward,itisplannedtooperatetheLHCwith Modernphoto-lithographictechnologyhasenabledthe 2 typically 140 collisions per proton bunch crossing (HL- development of a series of novel Micro Pattern Gaseous v LHC phase), which will greatly complicate the analysis, Detectors (MPGD): Micro-Strip Gas Chamber (MSGC), 3 particularly in the forward regions, where it will be very Gas Electron Multiplier (GEM), Micro-mesh Gaseous 2 difficulttolinkthetrackswiththeprimaryvertex(associ- Structure(Micromegas),andmanyothersrevolutionizing 1 0 atedwiththeonlyinterestingcollision)andthustoprevent cellsizelimitsformanygasdetectorapplications.TheMi- 0 theformationoffakejetscreatedbytherandomstacking cromegasdetectorfulfilstheneedsofhigh-luminositypar- 1. or pile-up of tracks from several secondary vertices. A ticle physics experiments, and compared to classical gas 0 solution proposedto fight against this pile-up effect is to counters, it offers intrinsic high rate capability, excellent 6 measure very accurately the time of arrival of the parti- spatialresolutionandexcellenttimeresolution [3,4]. 1 cles [1]. A time resolution of a few tens of picoseconds : v withaspatialgranularitybetterthan10mm(dependingon 2 Micromegas for fasttiming i pseudo-rapidity)willbeneededtoobtainafakejetrejec- X tionratethatisacceptableforphysicsanalyses. Suchper- 2.1 Detectionprinciple r a formances should be obtained with a detector capable of Theconceptof aMicromegasdetectorcoupledto a front withstandingtheveryimportantradiationlevelsexpected window that acts as Cerenkov-radiator, equipped with a intheHL-LHCera. photocathode, is examined in order to address its timing The ongoing R&D on timing for charged particles is performance. The radiator should produce sufficient UV approaching a level of 10 ps [2] but is not primarily di- photonsandshouldbeequippedwithanappropriatepho- rected at sustained performance at high rates and under tocathodeinordertocreatethenecessarynumberofelec- high radiation (as would be needed for HL-LHC pileup tronstoconverttheexpected∼100pssingle-photoelectron mitigation).Severaltechnologiesarepresentlyundereval- timejitterintoatimingresponseof10-20psperincident uation within the LHC collaborations: Silicon detectors particle. The photocathode has to be spark-resistant and with some built-in amplification, Silicon detectors with- musthavegoodagingproperties,appropriateforthehigh out amplification and gaseous detectors. Gaseous detec- particlefluxenvironmentofHL-LHC. tors have the potential of very good timing performance, Ingaseousdetectorsthetimeresolutionistypicallyof providedtheamplificationgapissmallenough.Theyhave the order of several ns because of the time jitter of pri- ae-mail:[email protected] mary ionization (for high energy MIPs) in the drift gap. EPJWebofConferences Ourstrategytoimprovethetimeresponseistominimize photons. IfListhethicknessincm,T(E)isthetransmis- thisjitterbycreatingsimultaneouslyallprimaryelectrons sion of the crystal and QE(E) the quantum efficiency of at about the same distance from the amplification region thephotocathode,thenumberofproducedphotoelectrons andbyminimizingthedriftspace. Theprimaryelectrons willbe: arecreatedbytheCerenkovlightwhichisemittedbythe chargedparticle when it crosses a 2-3 mm MgF crystal. 1 2 N ≈370L T(E)QE(E) 1− dE A thin CsI photocathode is deposited on the micromesh p.e. Z n2(E)! (reflective mode) or on the bottom of the crystal (semi- transparent mode) as shown in figure 1, and it is used to For a MgF2 crystal and a CsI photocathode with convertUV photonsto electrons. The time spread of the QE(E) takenfrom[7], T(E)from[8] andn(E) from[9], Cerenkovlight(∼10ps)istoosmalltolimittheprojected we expect 370 × 0.89 = 330 p.e./cm (figure 2). How- timeaccuracy. ever, this QE concerns operation in reflective mode. In thesemitransparentmodeitisexpectedtobesignificantly lower(factor2-3),so,fora3mmMgF crystalwewould 2 expect30-50photoelectrons. 1 MgF2 Transmission E CsI Q.E. (reflective) Q n or 0.8 1T -× 1 Q/nE2( E× ) 1(o -) 1/n2(E) sio ∑T×QE 1 - 1 ∆E = 0.89 s n2(E) mi 0.6 s n a Tr 0.4 0.2 Figure1.Schematicviewofthechargedparticledetectionprin- ciple, in the two modes; Reflective mode: a thin CsI layer is 5 6 7 8 9 10 11 12 depositedontopofthemicromesh;Semitransparentmode: CsI E [eV] isdepositedonthebottomsurfaceoftheUVcrystal,defininga driftgapwiththemicromesh. Figure2. Transmission,refractiveindexofMgF andquantum 2 efficiencyofCsIinreflectivemodewhichwereusedforthecal- culationofthenumberofphotoelectrons. In the semitransparent mode the photocathode is de- positedatthebottomfaceoftheMgF crystalatadistance 2 of 100-300μm from the micromesh to limit longitudinal diffusion. A strong field can be applied in the drift gap allowing some pre-amplification. The advantagesof this 2.3 Diffusioninthedriftregion configuration are the high single-electron sensitivity due tohighertotalgain,thepossibilitytousewovenmeshMi- Since the time jitter in photoelectron creation from the cromegas (bulk technology [5]) and the fabrication sim- Cerenkovlightissmall(<10ps),thedominanteffectlim- plicity. However,thepartialexposureofthephotocathode itingthetimeresolutionwillbethediffusionofthephoto- totheionbackflowmaycauseagingandrobustnessissues. electronsinthedriftregion. Inthereflectivemodethephotocathodeisdepositedonthe Inthecaseofthereflectivephotocathode,thejitter in Micromegasmeshandnostrongdriftfieldisneeded. The thedriftpathwillreflectthedifferenceinthepathdepend- main advantagesofthis methodis the veryeffectivepro- ing on the creation point on the mesh and its distance to tectionofthephotocathodefromtheionbackflow(better theamplificationgap,sofinepitchmeshesarepreferable. aging properties, robustness) and the complete suppres- Simulationswerecarriedouttoinvestigatethesemitrans- sion of gas ionization in the drift region. However, this parent mode. Magboltz simulation results are shown in mode is limited to flat mesh detectors (Microbulk tech- figure3wherethetimespreadduetoelectrondiffusionin nologyisthemoreappropriate[6]),whiletheinternalre- a200μmdriftgapisplottedasfunctionofthepreampli- flectionofthe Cerenkovlightonthecrystal-gasinterface ficationforseveralgasmixtures. Afirstassessmentbased mightlimittheefficiency. onMagboltzthusindicatesthatasinglephotoelectrontime resolutionof100-200pscouldbereachableforrelatively 2.2 Photoelectronproduction small(orevenintheabsenceof)pre-amplification. Compared to noble-gas admixtures, pure quenchers AMIPpassingthroughacrystalwillemit: show potentially a smaller time spread, but demand stronger electric fields in order to achieve the same gain. d2N α2z2 370 1 Theoptimumselectionwillbeatradeoffbetweenthemin- = sin2θ ≈ 1− dEdx r m c2 c eVcm n2(E)! imumtimejitterandthemaximumsignaltonoiseratio. e e MPGD2015 Magboltz simulations theUV laser, ismountedonthetopsurfaceofthe cham- 1400 ber. Thechamberandthecomponentswereproperlycho- Ne/ethane (95/5) 1300 ps] 1200 Npuer/ee tehtahnaen e (90/10) sen andspecialprecautionwastakenwhile mountingthe m [ 1100 pure CF detector in order to be appropriate for operating without µ200 1900000 pure CO42 agnadscpiarcrtuiclaletiobne,atmoffaacciilliittaietes.thusthemeasurementsatlaser d in 800 ea 700 e spr 600 2.5 Performanceoftheprototype m 500 n ti 400 The gas mixture was chosen to be 90% Ne plus 10% o ctr 300 Ethane, a mixture that gives high gain with good energy ele 200 resolutionandhasshowngoodbehaviorfor sealed mode 100 0 operation [10], though it is not optimal for timing. We 1 10 100 pre−amplification in 200µm usedapulseddeuteriumlampinordertomonitorthegain stabilitywithoutgascirculation.Dataweretakenoverone Figure3.Electrontimespreadduetodiffusionforseveralgases weekshowingonlyrandomfluctuations(arisingfromthe asafunctionoftheachievedpre-amplificationina200μmgap. lampintensity)andnosignificantdropofthegain,demon- strating the possibility to perform measurementswithout renewingthegas. 2.4 TheMicromegasprototype s nt candle @ 20 cm Aprototypedetectorhasbeenconstructed(figure 4)inor- ou250 candle @ 30 cm C dertostudythetimingpropertiesoftheproposedscheme. 200 It has been designed to be compatible with operation in bothsemitransparentandreflectivemode. 150 100 50 0 0 100 200 300 400 500 Amplitude [mV] Figure5.Pulseamplitudespectraforacandleplacedat∼20cm and∼30cmdistancefromthedetectorwithachargepreampli- Figure4.Schematicviewofthedetectorchamberandexpanded fier. Theshapeofthespectrumdoesnotchange,aclearindica- viewoftheMicromegas-spacers-crystalassembly. tionforsingleelectronpulses. TheMicromegaswasconstructedusingbothBulkand The sensitivity of the detector to single photons was Microbulk technologies[5, 6]. It was a single-anodede- verified using a candle flame, which is emitting continu- tector of 1 cm diameter (bigger than the maximum pixel ously light from the IR down to the VUV, the spectrum size of a possible tracker). The amplification gap was stoppingat∼200nmduetotheabsorptionoftheair[10]. 50 μm forthe Microbulkand 128μm for the bulk. Kap- The quantum efficiency of the aluminium is sufficient to ton rings, 50 μm thick, were used in orderto supportthe haveseveralphotoelectronspersecondwhenthedistance MgF orQuartzcrystalonwhichthephotocathodeisde- 2 to the flame isnotverylong. Two spectra takenata dis- posited and define the driftgap of 200 μm. The top ring tance of 20 and 30 cm for the same time duration are was metalized on the top surface, in order to apply the shown in figure 5. The amplitude distribution does not biasvoltagetothephotocathode. Athin(10nm)Allayer changewiththedistance,whiletheratedoes,provingthat wasdepositedonaquartzcrystalandservedasphotocath- thepulsescorrespondtosinglephotons.Theresultingvery ode. This simplified configuration is sufficient to study goodenergyresolutionis dueto the single stage amplifi- thetimingpropertiesofthedetector,usingaveryfastUV cationintheMicromegasgap. laser. Forchargedparticledetection,thequartzhastobe replacedwithMgF ,whilethephotocathodewillbeaCsI 2 layer (100-300 nm) deposited on a metallic substrate (5- 3 Time resolutionmeasurements 50nm).Theexactconfigurationwillbeoptimizedinterms ofagingpropertiesandofthemaximizationofthenumber The time response of the bulk Micromegas detector in ofproducedphotoelectrons. semitransparent mode was studied at LIDyL laboratory The Micromegas-crystal assembly is encased in a (CEA/Saclay). An Optical Parametric Oscillator (Coher- stainless-steel chamber. A quartz window, transparentto entMIRA-OPO),pumpedbyaTi:sapphirelaser(Coherent EPJWebofConferences MIRA900)provided120fspulsesat550nm. Therepeti- taioPnurlsaete-Pwicaksevra(rCieodhbeertewnete9n2900k)H. zTahneds4e.c7o5nMdhHazrmbyonuiscinagt σ [ps]t dddaaatttaaa ssseeerrriiieeesss 123,,, <<<NNNpp..ee..>>> === 5513130..40 × p.e. 2o7u5tpnumtewnearsggyewnearsa0te.1d5i-n1ap5JmfomraBbBeOamcrdyisatmale.tTerh<et1ypmicmal. 0.5 N>p.e.1000 dspartae asde rfireosm 4 ,d i<ffNups.ieo.>n (=M 1a.g2b6oltz) < 1D avalanche simulation ThoughtheextractionefficiencyfortheAlphotocath- on, uti odeatthosewavelengthsissmall,theintensityofthelaser ol (up to 106 photons/pulse) allows for the production of a me res sufficientnumberofphotoelectrons(upto1000p.e./pulse n ti o for10nmAl,dependingalsoonthebeamtuningandthe phot 100 − appliedMicromegasdriftfield).Thenumberofphotoelec- gle n trons per laser pulse was controlled by using a series of si 2 3 4 5 6 7 8 9 10 11 12 light attenuators: fine electroformednickelmeshes (100- E [kV/cm] drift 2000LPI)withopticaltransmissionvaryingbetween10% -25%,givingattenuationfactorsof4,5,10andtheircom- Figure7. Dependenceofthemeasuredtimeresolutionwiththe binations. Thepitchofthemesheswasvariedinorderto driftfield,scaledtothesinglephotoelectroncase.Thedifference avoidsystematiceffectswhencombiningseveralofthem. betweenthethinandthicklinesindicatestheimprovementdueto pre-amplification,accordingtoastochastic1Davalanchemodel. Thelaserbeamwassplitintwo,partofitilluminating our prototype and the other part an ultra fast Si detector (ThorlabsDET10A/M).Bothsignalswererecordedbya2 GHzoscilloscopeoperatingat10GS/s. Thetimejitterof thephotodiodewasoftheorderof13ps,beingappropriate astriggerandasreferenceforthemeasurement. Thesig- nal from the Micromegaswas filtered for high frequency ps] noiseusinga movingaverage. Thephotocathodeandthe σ [t100 filtered Micromegassignal times were calculated by tak- on, uti ing a constant fraction of each signal. Best results were ol es obtainedfora30%fractionoftheamplitude(figure6). me r data (Edrift=10kV/cm) ti spread from diffusion (Magboltz) spread from diffusion × 0.65 1D avalanche simulation 400 <<NNpp..ee..>> == 5511..44 ,, σσCCFF 3300%% == 2266..99 ppss <<<<NNNNpp..ee..>>>> ==== 11113.3.22..0606 ,,,, σσσσCCFF 3300%% ==== 51512727.4.488.. 5 5pp ppssss 101 10 100 pp..ee.. CCFF 3300%% number of photoelectrons, N PPDD ((11..2266 pp..ee..)) ,, σσ == 1122..66 ppss p.e. 300 CCFF 3300%% Figure8. Dependenceofthemeasuredtimeresolutionwiththe meannumberofphotoelectrons,forfixedamplificationanddrift 200 fields.Aresolutionof200pspersinglephotoelectronand27ps for 50photoelectronshasbeenachieved. 100 4 Conclusions −7.5 −7 −6.5 −6 t [ns] A Micromegas photodetector, equipped with a UV- Figure6.Distributionofthetimeatthe30%fractionoftheam- transparentcrystalandametallicphotocathodedeposited plitudefortheMicromegaspulses,fordifferentmeannumberof on the bottom of the crystal has been constructed in or- photoelectrons.Thetimespreadofthephotodiodeisalsoshown. der to investigatethe potentialof MPGDs forfast timing tracking of minimum ionizing particles. The aim of the firsttestswastheinvestigationoftheintrinsictimecharac- teristicsofthedetector. AveryfastUVlaser(timespread Data were taken for different values of the drift field 120 fs) was used to producethe photoelectronsin a con- in combinationwith the attenuators, in order to maintain trolledway.Withagasmixtureconsistingof90%Neand the signal within a reasonable dynamic range. The time 10%Ethanewemeasuredatimeresolutionoftheorderof resolutionimproveswiththedriftfieldintensityduetothe 200 ps for single photoelectronsand 27 ps for 50 photo- decrease of the longitudinaldiffusion but also due to the electrons. Simulationsshowthatthisalreadysatisfactory preamplificationathighfieldvalues. Theresultsaresum- resultcanbefurtherimprovedbyoptimizingthegasmix- marizedinfigures7and8,incomparisonwithastochastic ture. A more exhaustive description of these results will 1D avalanchesimulation(thatwill be detailed in a forth- be publishedelsewhere. Tests on a particle beam, with a comingpublication). MgF crystalandCsIphotocathodewillbemadein2016. 2 MPGD2015 Thecurrentprojectiscarriedoutwithintheframework [4] I.Giomataris,Nucl.Instrum.Meth.A419(1998)239. oftheRD51commonProject"FastTimingforHighRate [5] I.Giomatarisetal.,Nucl.Instrum.Meth.A560(2006) Environments:AMicromegasSolution". 405-408 [6] S.Andriamonjeetal.,JINST5(2010)P02001 References [7] J.Séguinotetal.,Nucl.Instrum.Meth.A297Issues1-2 (1990)133-147. [1] S. White, proc Picosecond Workshop, Clermond- [8] http://www.korth.de/index.php/162/items/21.html Ferrand2014,arXiv:1409.1165[physics.ins-det]. [9] P. Laporte et al., Optical Society of America Journal [2] C. Williams et al., Nucl.Instrum.Meth. A594 (2008) (ISSN0030-3941),73,(1983)1062-1069. 39-43. [10] A. Peyaud et al., Nucl.Instrum.Meth. A787 (2015) [3] Y. Giomataris, P. Rebourgeard, J.P. Robert, G. 102-104. Charpak,Nucl.Instrum.Meth.A376(1996)29.