PreprinttypesetinJINSTstyle-HYPERVERSION 5 1 0 2 A Leakage Current-based Measurement of the b e Radiation Damage in the ATLAS Pixel Detector F 2 1 ] t e d - IgorV. Gorelovonbehalfofthe ATLASCollaboration s n aDepartmentofPhysicsandAstronomy,UniversityofNewMexico, i s. Albuquerque,NM87131,USA. c E-mail: [email protected] i s y h p ABSTRACT: Ameasurementhasbeenmadeoftheradiation damageincurred bytheATLASPixel [ Detector barrel silicon modules from the beginning of operations through the end of 2012. This 2 translatestohadronicfluencereceivedoverthefullperiodofoperationatenergiesuptoandinclud- v 2 ing 8TeV. Themeasurement isbased ona per-module measurement ofthe silicon sensor leakage 9 current. The results are presented as a function of integrated luminosity and compared to predic- 1 7 tions bytheHamburg Model. Thisinformation canbeused topredict limitson thelifetime ofthe 0 PixelDetectorduetocurrent, forvariousoperating scenarios. . 1 0 5 KEYWORDS: Simicrostrip andpaddetectors; Pixelateddetectors; Radiation-hard detectors. 1 : v i X r a Contents 1. Introduction 1 2. MeasuredQuantities 2 3. TheATLASCurrentMeasurementSubsystem 3 4. CurrentMeasurements, 2011-2012 4 5. Precision andSystematicUncertainties 5 6. LifetimeEstimate 6 7. SummaryandProspects 7 1. Introduction TheATLASPixelDetector[1]isconstructedfromplanarsiliconpixelmodulesarrangedinabarrel and disk geometry. It is centered on the interaction point at the heart of the ATLAS Detector [2] attheLargeHadronCollider(LHC).Alinearrelationship betweenleakage currentandahadronic particle fluenceisexpected toapplytothepixelsiliconsensorsandisgivenby D I=a F V, (1.1) eq · · whereD IisthedifferenceinleakagecurrentatfluenceF relativetothevaluebeforeirradiationof eq thephysicalvolumeV,anda isthecurrent-relateddamagecoefficient[3]. Thegoalofthisstudyis tomeasuretheleakagecurrentsofarepresentative sampleofsensorsintheATLASPixelDetector inordertomonitorandunderstandthesensordamageresultingfromincreasingradiationdose. The measured results are compared to predictions based on the Hamburg Model [4] in which reverse- bias current is generated by the introduction of defect levels in the band gap due to non-ionizing energyloss(NIEL)ofthrough-going particles. Thedefectsmigrateandrecombinethroughseveral processes that result inboth short-term beneficial annealing (which reduces the current) andlong- term anti-annealing (which increases the current even in the absence of further radiation). The modelcanbecalibratedtothedatatopredictthelifetimeofthesensorsandthePixelDetectorasa wholeforvariousoperating scenarios. The detector is instrumented as three barrels, specifically the Layer-0 at r = 50.5mm, the Layer-1 atr=88.5mm,andtheLayer-2 atr=122.5mm,comprising intotal 1456modules. The endcap areas are instrumented with three disks located at z= 495mm, 580mm and 650mm ± ± ± in each of the forward and backward regions, comprising an additional 288 modules. The Pixel –1– Detector and the other elements of the ATLAS Inner Detector span a pseudorapidity range h < | | 2.51. Thepixelsensormodules aremountedonmechanical/cooling supports, calledstaves, inthe barrel region. Thirteen modules are mounted on a stave with identical layouts for all layers. The temperature is maintained stable by an evaporative cooling system. The entire 1.7m2 sensitive ∼ area of the ATLAS pixel detector is covered with 1744 identical modules. Each module has an activesurfaceof6.08 1.64cm2. Thedetailsofthepixelsensorgeometryandlayoutcanbefound × in[1]. Theradiation fieldin the region of the detector has been predicted with the aid of a FLUKA software package [5]. The FLUKA transport code to cascade the particles through the ATLAS detectormaterialhasbeenusedtogetherwithPYTHIA 6.2 orPHOJETgeneratingproton-proton minimum-bias eventsasaninputtoFLUKA [1]. Weassumethatthedominantradiation damagetypeisdisplacement defectsinthebulkofthe pixel sensor, initiated by hadronic species and caused by NIEL.Surface ionization isneglected in this treatment. Charged pions are expected to dominate the bulk damage for the radii covered by thePixelDetector. Albedoneutrons originating intheouterATLASdetectors alsocontribute. The resulting displacement defects increase the reverse leakage current, degrade the charge collection efficiency, and change the effective doping concentration which directly determines the depletion voltage. WedefinetheeffectivefluenceF asthenumberofparticles causing damageequivalent eq to that of 1MeV neutrons traversing 1cm2 of asensor’s surface. The fluence F accumulated by eq ATLASpixeldetector, andmeasuredinunitsof cm 2,isexpected tobeproportional tointegrated − luminosity L dt,measuredin pb 1. − R 2. Measured Quantities Theparametera inEquation1.1hasbeenmeasuredunderavarietyofconditions,forexample[3]. The linear form applies to the leakage currents drawn by sensors past their beneficial annealing periods. Atthebeginning ofdata-taking andduring beneficial annealing periods, thesensors draw currents at levels comparable to dark currents; thus any hardware implementations for leakage current measurementshouldprovideaccesstothelowestcurrent rangepossible. The reverse bulk generation current depends on the sensor temperature T. The current mea- surements I(T)arenormalized tothereference temperature T (0 C)with R ◦ E I(T)=I(T )/R(T), whereR(T)=(T /T)2 exp g (1/T 1/T) , (2.1) R R R · (cid:18)−2k − (cid:19) B whereE =1.21eVistheenergyofthesiliconbandgap,k istheBoltzmannconstant,andTandT g B R aremeasuredinK[6]. Themeanpixel module temperature for allbarrel pixel modules isshown versus timein Fig- ure 1 [7]. There were a few exceptions when the temperature rose to ambient due to cooling interruptions and calibration scans. Temperature is read out from every module continuously and 1ATLASusesaright-handedcoordinatesystemwithitsoriginatthenominalinteractionpoint(IP)inthecentreof thedetectorandthez-axisalongthebeampipe. Thex-axispointsfromtheIPtothecentreoftheLHCring,andthe y-axispointsupward.Cylindricalcoordinates(r,f )areusedinthetransverseplane,f beingtheazimuthalanglearound thebeampipe.Thepseudorapidityisdefinedintermsofthepolarangleq ash = lntan(q /2). − –2– averaged over30-minute intervals. Temperature data(aswellasother slowcontrol data) arecom- mittedintotheATLASDetectorControlSystem(DCS)database. 25 ATLAS Preliminary 20 15 C (cid:176)e, 10 ur 5 at er 0 p m -5 e T -10 -15 Mean Pixel Module Temperature: Layer-0 / Layer-1 / Layer-2 -20 02/07/11 01/01/12 01/07/12 31/12/12 Date, dd/mm/yy Figure1: Themeantemperature( C)ofallbarrelpixelmodules,versustime(days)[7]. ◦ 3. The ATLAS Current Measurement Subsystem The ATLAS High Voltage Patch Panel 4 (HVPP4) serves as a fan-out point for the bias voltages delivered by Type II boards from Iseg [8] high voltage power supplies to the pixel modules. The leakagecurrentismonitoredatthepixelmodulegranularitylevelbyaCurrentMeasurementBoard (CMB)system. ACMBismountedoneveryTypeIIfan-outboard. Thebiasvoltagetothesensors isprovidedbytheIsegchannels. Inthepresentdata-taking era,whentheradiation damageofthesensorsisrelatively low,6or 7 pixel modules are fed by one Iseg power supply channel. At some point after inversion of the sensors, before the current drawn by 6 or 7 pixel modules reaches the Iseg limit, power supplies can be added until the system provides one Iseg power supply channel per pair of pixel modules. TheCMBsystem allows simultaneous measurement ofthe leakage currents in4pixel modules in the current range from 0.04m A to 2mA with a precision of better than 20% per module. ATLAS uses21CMBsinLayer-0, 16inLayer-1,and16inLayer-2. Theanalogcurrentmeasurementsare digitized by the 64-channel ATLASEmbedded Local Monitoring (ELMB)board and sent via the ControllerAreaNetwork(CAN)bustotheDCSdatabase(see[1]fordetailsontheTypeIIfan-out board, ELMB, CAN, and DCS). The building block of the CMB circuit is a current to frequency converter optically coupled to a frequency to voltage converter as shown in Figure 2. Each CMB holds4 2currentmeasurementcircuitsandprovidesthecurrentmeasurementfor4pixelmodules. × Twosimilarcircuits withdifferent gainsareallocated perCMBchannel serving thesamemodule. The channels are isolated from each other and from the pixel module readout system. For each pixel module, the leakage current range [10 8 10 5]A is covered by the high-gain channel, and − − − therange[10 6 2 10 3]Aiscoveredbythelow-gainchannel. TheoutputCMBvoltagerangeis − − − · definedbyoneoffiveavailable ELMBinputvoltage ranges, whichare[0 25]mV,[0 100]mV, − − [0 1]V, [0 2.5]V, and [0 5]V. The output range [0 5]V is necessary to comply with the − − − − specifications of the digital ELMB board. The 16-bit ADC of the ELMB provides a resolution of (ELMB range)/(216 1). The range [0 1]V was used for data taken through 2011; the range − − –3– Figure2: TheCurrentMeasurementBoardcircuit . was changed to [0 5]V in 2012 as the high-gain channels reached saturation, corresponding to − pixelleakagecurrentsgreaterthan10 5A. AfterLHCshutdownof2013-2014,thedata-takingwill − resume with the low-gain channels. Before installation, each CMB is calibrated offline assuming a linear model. The calibrated gains of the channels are stored in the database. The pedestals are re-calibrated insitu. 4. Current Measurements, 2011-2012 Figure 3 shows the leakage current recorded by the CMB system from the beginning of ATLAS data-takingthroughtheendof2012,asafunctionofintegratedluminosity, formodulesinallthree PixelDetectorbarrellayers[7]. InstallationoftheCMBsystemproceededconcurrently withearly data-taking, and this is the reason that some Layer-2 data are missing from the early part of the run. Itisforeseen thatthedisks willbeinstrumented withcurrent monitorboards priortothestart of data-taking in 2015 (Run 2). The points at which LHC and cooling were off, hence beneficial annealingofthesiliconwasunsuppressed,areapparentasdiscontinuitiesinthisfigure. Thiseffect, producing thesawtoothfeatures, isduetotheincrease oftheeffectivedoping concentration ofthe silicon duetoannealing ofacceptors [3,4]. –4– Figure3: ATLAS Pixel module leakage current versus integrated LHC luminosity [7]. The cur- rents are averaged over their layer for all mod- ulesequipped withCurrentMeasurement Boards within the layer. The current is continuously monitored by the ATLAS Detector Control Sys- tem. A prediction based on the Hamburg Model is included. Discontinuities are due to annealing duringLHCandcooling stops. Figure3alsoincludesaprediction[4]oftheleakagecurrentforthedetailedATLASgeometry and LHC luminosity profile based on the Hamburg Model. The predictions of the fluence F eq accordingtotheluminosityprofilearemadeusingtheFLUKAtransportcode[5]. Thepredictions are uniform w.r.t. the azimuthal angle f and suppose a forward-backward symmetry w.r.t. the polar angle q , i.e. F (h )=F ( h ). At the end of Run 1 with L dt =30fb 1, the Layer-0 eq eq − − moduleshavereceivedatotalfluenceF =7.0 1013cm 2,whileLRayer-1andLayer-2havebeen eq − · irradiated to F =3.0 1013cm 2 and F ∼=1.8 1013cm 2 respectively. The uncertainty of the eq − eq − · · leakage current p∼redictions comprises the u∼ncertainty of the calculations of the hadronic particle spectra initiated by pp interactions in the ATLAS detector and the uncertainty of the Hamburg modelindescribing radiation damage effectsinflicted byNIELofhadrons crossing thevolumeof the silicon pixel sensors. Figure 4a shows the distribution in azimuthal angle of modules whose currentissampledbyCMBsintheATLASPixelDetector. Figure4bshowsthedataforeachoffour quadrants inthe azimuthal angle astheyaredefined byFigure 4a[7]. Nosystematic difference in a module leakage current is observed w.r.t. to the quadrant where the module islocated. Figure 5 shows the instrumented module distribution, segmented into five sectors in pseudorapidity. The pseudorapidity acceptance of each sector of each layer is indicated. Plots in Figure 6a through Figure 6e, show the leakage current data for each of the 5 ranges in pseudorapidity [7]. Figure 6f shows the current data versus pseudorapidity for the sectors defined by Figure 5, for integrated luminosity 25fb 1 [7]. Thepossible slight asymmetry ofthedatapoints inFigure6fisthought to − beduetotheasymmetricbeamprofile. Yetthedistribution isinagreementwithintheuncertainties withtheleakagecurrentpredictionswhicharesymmetricalaccordingtothefluencesimulation[5]. 5. Precisionand SystematicUncertainties Thecontributions totheuncertainty onthecurrentinclude theCMBandELMBprecision oncurrentmeasurements: 12%forbothranges. • thenumberofcurrentmeasurements (twoperhour). • theuncertainty ontheluminosity, 1.8%in2011and2.8%in2012[9]. • –5– (a)Adiagrammofmodules. (b)ModuleleakagecurrentversusintegratedLHCluminosity. Figure 4: (a) Sampling of 4 quadrants in azimuthal angle with modules instrumented by CMBs. (b) The module leakage current in each of the 4 quadrants in azimuthal angle [7]. The currents areaveraged overeachlayerforallmodules equipped withCurrentMeasurement Boards withintheparticular quadrant ofthelayer. Predictions basedontheHamburgModelareincluded. Figure 5: A diagram of the dis- tribution of instrumented mod- ulesineachoffivepseudorapid- ityranges. the temperature uncertainty, which, at less than 0.3 C, contributes to the uncertainty on the ◦ • current with3.4%. 6. Lifetime Estimate Thetemperature-correctedcurrentreadingspermodulecanbeextrapolatedtopredicttheamountof currentthepixelmoduleswilldrawafteracertainintegratedluminosityhasbeencollectedwiththe ATLASpixeldetector. Asleakagecurrentcanleadtoexcessivepowerandthermalrunaway,limits on the module lifetime due to current can be determined from the limit on the bias voltage that canbeapplied. AsingleIsegpowersupplychannelcansustainamaximumcurrentof <4000m A fortwomodules. Thereareotherconsequences ofradiation-induced leakage current up∼onmodule –6– (a) (b) (c) (d) (e) (f) Figure6: (a)through(e): ATLASPixelmoduleleakagecurrentineachof5pseudorapiditysectors, versus integrated LHC luminosity [7]. The currents are averaged over their layer for all modules equipped with Current Measurement Boards within the layer. The current is continuously mon- itored by the ATLAS Detector Control System. Predictions based on the Hamburg Model are included. Discontinuities are due to annealing during LHC and cooling stops. (f): the module leakagecurrentindefined5pseudorapidity bins,fortheintegratedluminosityequalto25fb 1[7]. − lifetime, including the noise, bias, and threshold limitations of the front end chips as well as the finitecapacity ofthecoolingsystem;whileimportant, thesearenottreatedhere. Weextrapolate the present rate ofcurrent increase withintegrated luminosity, assuming opti- misticallythatthemoduleswillbeexposedfor10daysto20 Candotherwisemaintainedat 13 C ◦ ◦ − in future years. The model assumes that the bias voltage is raised throughout LHC operation as neededtokeepthebulkfullydepleted,untilamaximumallowedvoltageof600Visreached. Only the effects of proton runs are considered. Figure 7 shows the implied leakage current per module versus date and versus integrated luminosity for Layer-0 in this scenario [7]. This extrapolation predicts a leakage current of about 500m A per Layer-0 module (of thickness 250m m) after the accumulation ofabout450fb 1 ofdeliveredluminosity, whichcorresponds toJanuary2024inthe − scenario proposed here. 7. Summary andProspects Wehave described the principles ofradiation damage monitoring using the current measurements provided by the circuits of the ATLAS Pixel high voltage delivery system. We have observed radiation damage in the ATLAS Pixel Detector with characteristics in agreement with the Ham- burg Model. The dependence of the leakage current upon the integrated luminosity has been pre- –7– 0.6 e0.5 ATLAS Preliminary ul (cid:176) od0.4 Leakage Current per Module at -13C : Layer-0 m per 0.3 A] Figure 7: Predicted leakage m0.2 [Ileak0.1 current per Layer-0 module at 13 C, versus date (upper) and ◦ 0 − 31/12/12 31/12/14 31/12/16 31/12/18 31/12/20 31/12/22 integratedluminosity(lower)for Date [dd/mm/yy] the scenario in which modules 0.6 will be exposed for 10 days to e0.5 ATLAS Preliminary odul0.4 Leakage Current per Module at -13(cid:176)C: Layer-0 20◦C and otherwise maintained per m0.3 at−13◦C[7]. A] m0.2 [Ileak0.1 0 0 50 100 150 200 250 300 350 400 450 Integrated L [fb-1] sented. Alinearbehavior oftheresponse hasbeenobserved. Thecurrentdistribution inazimuthal quadrants is symmetric as expected. The shape of the current distribution in pseudorapidity is in agreement with a fluence calculation based on the FLUKA transport code within the assigned systematic uncertainties. Extrapolation of leakage current with time predicts survival of Layer-0 up to 450fb 1 of data. This current monitoring system will be expanded to encompass pixel disk − modulesinRun2andbeyond. Acknowledgments Theauthorexpresseshisgrateful appreciation tothecolleagues fromATLASPixelDetector groupforfriendly cooperation andsupport throughallphasesoftheproject onpixelsensor leakage currentmonitoring andmeasurement. References [1] G.Aadetal.,ATLASpixeldetectorelectronicsandsensors,JINST3,P07007(2008). [2] ATLASCollaboration,TheATLASExperimentattheCERNLargeHadronCollider,JINST3, S08003(2008). [3] M.Moll,E.FretwurstandG.Lindström[CERN-ROSE/RD48Collaboration],Leakagecurrentof hadronirradiatedsilicondetectors-materialdependence,Nucl.Instrum.Meth.A426,87(1999). [4] O.Krasel,Chargecollectioninirradiatedsilicon-detectors,Ph.D.Thesis,UniversityofDortmund, DortmundGermany(2004),http://hdl.handle.net/2003/2354. [5] S.Baranov,M.Bosman,I.Dawson,V.Hedberg,A.NisatiandM.Shupe,EstimationofRadiation Background,ImpactonDetectors,ActivationandShieldingOptimizationinATLAS, ATL-GEN-2005-001,ATL-COM-GEN-2005-001,CERN-ATL-GEN-2005-001. –8– [6] A.Chilingarov,TemperaturedependenceofthecurrentgeneratedinSibulk,2013JINST8P10003, availableathttp://dx.doi.org/10.1088/1748-0221/8/10/P10003. [7] ATLASCollaboration,ALeakageCurrent-basedMeasurementoftheRadiationDamageinthe ATLASPixelDetector,ATL-INDET-PUB-2014-004,http://cdsweb.cern.ch/record/1752122.The figuresareavailablealsoat http://atlas.web.cern.ch/Atlas/GROUPS/PHYSICS/PUBNOTES/ATL-INDET-PUB-2014-004 [8] IsegSpezialelektronikGmbH,Rossendorf,Germany,http://www.iseg-hv.com. [9] ATLASCollaboration,Improvedluminositydeterminationin ppcollisionsat√s=7TeVusingthe ATLASdetectorattheLHC,Eur.Phys.J.C73,no.8,2518(2013)[arXiv:1302.4393[hep-ex]]. –9–