Nuclear Instruments and Methods in Physics Research A - F9

Nuclear Instruments and Methods in Physics Research A 477 (2002) 299–303Radiation hard cryogenic silicon detectorsL. Casagrande a, *, M.C. Abreu a , W.H. Bell b , P. Berglund c , W. de Boer d , E. Borchi e ,K. Borer f , M. Bruzzi e , S. Buontempo g , S. Chapuy h , V. Cindro i , P. Collins j ,N. D’Ambrosio g ,C.DaVi!a k , S. Devine b , B. Dezillie l , Z. Dimcovski h , V. Eremin m ,A. Esposito n , V. Granata j,k , E. Grigoriev d,h , F. Hauler d , E. Heijne j , S. Heising d ,S. Janos f , L. Jungermann d , I. Konorov n ,Z.Li l , C. Louren-co j , M. Mikuz i ,T.O. Niinikoski j , V. O’Shea b , S. Pagano g , V.G. Palmieri j , S. Paul n , S. Pirollo e ,K. Pretzl f , P. Rato a , G. Ruggiero g , K. Smith b , P. Sonderegger j , P. Sousa a ,E. Verbitskaya m , S. Watts k , M. Zavrtanik ia LIP, Lisbon, Portugalb Department of Physics and Astronomy, University of Glasgow, UKc Low Temperature Laboratory, Helsinki University of Technology, Finlandd IEKP, University of Karlsruhe, Germanye Dipartimento de Emergetica, Universit "a di Firenze, Italyf LHEP, University of Bern, Switzerlandg Universit "a Federico II di Napoli, Dipartimento di Fisica e INFN, Italyh Department de Radiologie, Universit!e de Gen"eve, Switzerlandi JSI, Experimental Particle Physics Department, Ljubljana, Sloveniaj CERN, Geneva, Switzerlandk University of Brunel, Brunel, UKl Brookhaven National Laboratory, Upton, USAm Ioffe Physico-Technical Institute, St. Petersburg, Russian Technical University of Munich, Garching, GermanyThe CERN-RD39 CollaborationAbstractIt has been recently observed that heavily irradiated silicon detectors, no longer functional at room temperature,‘‘resuscitate’’ when operated at temperatures below 130 K. This is often referred to as the ‘‘Lazarus effect’’. The resultspresented here show that cryogenic operation represents a new and reliable solution to the problem of radiationtolerance of silicon detectors. r 2002 Elsevier Science B.V. All rights reserved.PACS: 07.77.n; 85.30.zKeywords: Cryogenic silicon detectors; Lazarus effect*Corresponding author. CERN-EP Division, 1211 Geneva 23, Switzerland.E-mail address: luca.casagrande@cern.ch (L. Casagrande).0168-9002/02/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved.PII: S 0168-9002(01)01863-0

300L. Casagrande et al. / Nuclear Instruments and Methods in Physics Research A 477 (2002) 299–3031. IntroductionFinely segmented silicon detectors are the mostcommon choice for close to interaction pointtracking in high energy physics experiments. Dueto the harsh environment, the effect of radiationinduced damage in silicon represents a key issue inview of their long-term operation. Moderatecooling is applied to irradiated silicon detectorsto lower the leakage current and to inhibit reverseannealing [1]. However, this kind of operationdoes not prevent the changes in effective dopingconcentration, implying a dramatic increase of thedepletion voltage. When the radiation fluenceapproaches 10 15 n=cm 2 ; standard detectors becomeunusable [2]. Sophisticated multi-guard-ring designscan be used to increase the maximum biasvoltage which can be applied to a detector. Silicondetectors based on oxygenated bulk show asmaller increase of the depletion voltage whenirradiated with charged particles [3].An alternative way to improve the radiationhardness of silicon detectors is their operation atcryogenic temperatures. Indeed, it has been shown[4] that heavily irradiated silicon detectors, nolonger functional at room temperature, recovertheir performance when operated at temperaturesbelow 130 K. This experimental observation isusually referred to as the ‘‘Lazarus effect’’.At very low temperatures, the de-trapping rateof charged carriers from radiation induced deeplevels is strongly affected by the reduced thermalenergy, leading to the situation in which animportant fraction of traps remains filled, henceinactive. This prevents further trapping of carriersgenerated by particles traversing the detector and,therefore, the signal is not reduced. Moreover, thecharged traps reduce the effective doping concentrationthus reducing the full depletion voltage ofthe detector. This qualitative interpretation is inagreement with the available observations [5].Furthermore, at cryogenic temperatures the leakagecurrent becomes negligible. This facilitates thedesign of the front-end readout electronics interms of noise requirements, and a simple readoutscheme (like DC coupling) can be used. Inthis paper we present data obtained on silicondiodes irradiated at room temperature to fluencesof 5 10 14 and 2 10 15 1 MeV eq. neutrons=cm 2 :We also present preliminary results on theirradiation of a silicon pad detector with450 GeV protons during operation at 80 K.2. Results on samples irradiated at roomtemperatureWe have investigated the performance of twosilicon (Al/p þ /n/n þ /Al) diodes irradiated at roomtemperature with neutrons to fluences of 5 10 14and 1 10 15 n=cm 2 ; respectively. The 400 mmthick smaples has a sensitive area of 5 5mm 2surrounded by a guard-ring. Their resistivitybefore irradiation was around 4.5 kO cm.The measurements were performed in a cryostatdescribed elsewhere [6]. The samples were exposedto a radioactive source ( 90 Sr) and a trigger diodewas used to select minimum ionising particles(MIPs). The detector signal was read out by acharge amplifier, the first stage of which was aGaAs FET operational down to 1 K mountedinside the cryostat. The noise of the amplifier wasabout 1500 electrons FWHM. The signal wasshaped (1 ms shaping time) and sent to a multichannelanalyser. The recorded charge spectrawere fitted with a Landau distribution. The chargecollection efficiency (CCE) was determined as themost probable value from the fit, normalised to thesame value obtained for a non-irradiated detectorof the same thickness operated above full depletion.The errors shown in the figures include thesystematic errors due to the uncertainty on thedetector thickness.The current–voltage characteristic of the irradiatedsamples at 77 K showed a negligible leakagecurrent (o1 nA) even when operated in forwardbias up to 250 V. Fig. 1 shows the CCE of thesamples as a function of the temperature. Themeasurements were only performed whenthe leakage current was reduced to such a valuethat the noise was below 2000 electrons FWHM.In the case of the sample irradiated with5 10 14 n=cm 2 operated in forward bias, this wasonly achieved for temperatures below B140 K. Inreverse bias, the CCE improves with decreasingtemperature until it reaches a maximum around

L. Casagrande et al. / Nuclear Instruments and Methods in Physics Research A 477 (2002) 299–303 301Fig. 1. CCE as a function of the temperature in reverse and forward bias operation after fluences of (a) 5 10 14 n=cm 2 and (b)1 10 15 n/cm 2 :figures correspond to measurements in stableconditions. In fact, it has been observed [6] thatCCE in reverse bias decreases with time after thebias voltage is set, until it reaches a stable value. Inconventional reverse bias operation, for the sampleirradiated with the higherst fluence, an MIP signalof about 6500 electrons can be obtained. Inforward bias there is no time dependence of theCCE and B100% CCE can be achieved with200 V after a fluence of 5 10 14 n=cm 2 : The sameresult can be obtained in reverse bias in presence ofshort wavelength light [6].3. Results from a sample irradiated at 80 KFig. 2. CCE at T=77 K as a function of the bias voltage for thediodes irradiated at room temperature.130 K. In forward bias, 2–3 times higher values ofthe CCE can be achieved also at higher temperatures,as long as the leakage current does notexceed tolerable values.Fig. 2 shows the voltage dependence of the CCEin reverse and forward bias. The data shown in theA3 3 silicon pad detector matrix was placedinside a cryostat in the CERN-SPS 450 GeVproton beam line. Each pad had a sensitive areaof 1:5 1:5mm 2 : The sensor thickness was400 mm. Two movable scintillators were used fortriggering single protons for the CCE measurements.We used a liquid nitrogen continuous flowcryostat. The sample holder was kept in anultrahigh vacuum chamber and was connected toa cold copper finger. The sample was then cooledby thermal conduction. The cryostat had two

302L. Casagrande et al. / Nuclear Instruments and Methods in Physics Research A 477 (2002) 299–303windows for the beam made of 100 mm thickstainless steel. In correspondence of the windows,the cryostat sample chamber had two 50 mm thickAl sheets acting as thermal radiation shield.The beam was always centred on one of thepads. It had Gaussian shape in both transversecoordinates, with s x Bs y B 1 mm at the detector.Therefore, the dose across the irradiated pad canbe considered as rather uniform. The detectorsignal was read out by a charge amplifier placedoutside the cryostat. The noise of the amplifier wasB2500 electrons FWHM, determined mainly bythe 30 cm coaxial cable between the detector andthe amplifier input. The signal was then fed into anoscilloscope with histogramming function, whichrecorded the charge spectra. The charge collectionefficiency of the pad was measured using a lowintensity beam, in between the high beam intensityperiods of irradiation.Fig. 3 shows the CCE of the irradiated pad as afunction of the bias voltage. We assumed thatbefore irradiation the pad was fully depleted andwe normalised to this vlaue the CCE measuredafter the irradiation. After a dose of1:5 10 15 protons=cm 2 a CCE of 60% (mostCCE (%)1007550250Before irradiation0 50 100 150 200Bias Voltage (V)After 1.5×10 15 p/cm 2Fig. 3. CCE as a function of the bias volatge for the padirradiated at 80 K.probable value) at 200 V bias volatge has beenmeasured. It is worth mentioning that, contrary tothe case of the irradiation at room temperature,these data show a negligible time dependence ofthe CCE.One expects that stable defect generation in thesilicon bulk is different from room to cryogenictemperatures. Since the detector was alwaysoperated at 80 K, we can reasonably assume thatthere is no annealing effect in the data. A completeunderstanding of the phenomenon is envisagedand work is currently in progress on thissubject.4. ConclusionsWe have shown that the CCE of heavilyirradiated silicon detectors recovers when theoperating temperature is below 130 K. In thecase of conventional reverse bias operation, anMIP signal of B6500 electrons can be obtainedfrom a 400 mm thick detector irradiated with 1 10 15 n=cm 3 : Furthermore, at very low temperaturesforward bias operation becomes possible. Inthis case, a CCE of B100% and B60% can beachieved for samples irradiated with 5 10 14 and1 10 15 n=cm 2 ; respectively.Preliminary results on a silicon pad detectorirradiated at 80 K indicate that there are nopathologies associated with the irradiation in thecold. It is important to stress that in the case ofcryogenic operation, since the reverse annealingdoes not play a significant role [6], the devices onlyneed to be cooled during operation and canotherwise be kept at room temperature.AcknowledgementsThis work was supported by the U.S. Departmentof Energy (contract DE-AC02-98CH10886),the Swiss Federal program MEDTECH (project3893.1), and the Swiss National-fondszur Foerderung der wissenschaftlichen Forschung.L.C. acknowledges financial support from theEC (TMR programme, contract ERBFM-BICT961204).

L. Casagrande et al. / Nuclear Instruments and Methods in Physics Research A 477 (2002) 299–303 303References[1] ATLAS Inner Detector TDR, CERN/LHCC/97-16/17,1997;CMS Tracker Project TDR, CERN/LHCC/98-6, 1998.[2] RD48 Collaboration Status Report, CERN/LHCC 98-39,1998.[3] A. Ruzin, et al., Radiation hardness of Si detectorsprocessed on various oxygen enriched wafers, Rose Meeting,CERN-LEB 99-8.[4] V.G. Palmieri, et al., Nucl. Instr. and Meth. A 413 (1998) 475.[5] B. Dezillie, V. Eremin, Z. Li, IEEE Trans. Nucl. Sci. NS-46(1998) 221.[6] K. Borer, et al., Nucl. Instr. and Meth. A 440 (2000) 5.