Beam test results of IHEP-NDL Low Gain Avalanche Detectors(LGAD)
S. Xiao, S. Alderweireldt, S. Ali, C. Allaire, C. Agapopoulou, N. Atanov, M. K. Ayoub, G. Barone, D. Benchekroun, A. Buzatu, D. Caforio, L. Castillo García, Y. Chan, H. Chen, V. Cindro, L. Ciucu, J. Barreiro Guimarães da Costa, H. Cui, F. Davó Miralles, Y. Davydov, G. d'Amen, C. de la Taille, R. Kiuchi, Y. Fan, A. Falou, A. S. C. Ferreira, M. Garau, J. Ge, A. Ghosh, G. Giacomini, E. L. Gkougkousis, C. Grieco, S. Guindon, D. Han, S. Han, M. Holmberg, A. Howard, Y. Huang, M. Jing, Y. Khoulaki, G. Kramberger, E. Kuwertz, H. Lefebvre, M. Leite, A. Leopold, C. Li, Q. Li, H. Liang, Z. Liang, B. Liu, J. Liu, A. Luthfi, F. Lyu, S. Malyukov, I. Mandić, L. Masetti, M. Mikuž, I. Nikolic, L. Polidori, R. Polifka, O. Posopkina, B. Qi, K. Ran, B. J. G. Reynolds, C. Rizzi, M. Robles Manzano, E. Rossi, A. Rummler, S. Sacerdoti, G. T. Saito, N. Seguin-Moreau, L. Serin, L. Shan, L. Shi, N. F. Sjostrom, A. Soares Canas Ferreira, J. Soengen, H. Stenzel, A. J. Szadaj, Y. Tan, S. Terzo, J. O. Thomas, E. Tolley, A. Tricoli, S. Trincaz-Duvoid, R. Wang, S. M. Wang, W. Wang, W. Wang, K. Wu, X. Shi, T. Yang, Y. Yang, C. Yu, X. Zhang, L. Zhao, M. Zhao, Z. Zhao, X. Zheng, X. Zhuang
BBeam test results of IHEP-NDL Low Gain AvalancheDetectors(LGAD)
S. Xiao a,b , S. Alderweireldt c , S. Ali d , C. Allaire c , C. Agapopoulou e , N.Atanov f , M. K. Ayoub a , G. Barone g , D. Benchekroun h , A. Buzatu d , D.Caforio i , L. Castillo Garc´ıa j , Y. Chan k , H. Chen l , V. Cindro m , L. Ciucu d , J.G. da Costa a , H. Cui a,b , F. Dav´o Miralles j , Y. Davydov f , G. d’Amen g , C. dela Taille n , R. Kiuchi a , Y. Fan a , A. Falou e , A. S. C. Ferreira c , M. Garau o , J.Ge l , A. Ghosh p , G. Giacomini g , E. L. Gkougkousis j , C. Grieco j , S.Guindon c , D. Han q , S. Han a,b , M. Holmberg r , A. Howard m , Y. Huang a , Y.Huang k , M. Jing a,b , Y. Khoulaki h , G. Kramberger m , E. Kuwertz c , H.Lefebvre s , M. Leite t , A. Leopold u , C. Li l , Q. Li l , H. Liang l , Z. Liang a , B.Liu a , J. Liu a , A. Luthfi c , F. Lyu a , S. Malyukov f , I. Mandi´c m , L. Masetti v ,M. Mikuˇz m , I. Nikolic u , L. Polidori v , R. Polifka w , O. Posopkina c , B. Qi a,b ,K. Ran a,b , B. J. G. Reynolds x , C. Rizzi c , M. Robles Manzano v , E. Rossi g ,A. Rummler c , S. Sacerdoti e , G. T. Saito t , N. Seguin-Moreau n , L. Serin e , L.Shan a , L. Shi a , N. F. Sjostrom r , A. Soares Canas Ferreira c , J. Soengen v , H.Stenzel i , A. J. Szadaj r , Y. Tan a,b , S. Terzo j , J. O. Thomas y , E. Tolley x , A.Tricoli g , S. Trincaz-Duvoid u , R. Wang v , S. M. Wang d , W. Wang z , W.Wang l , K. Wu a,b , X. Shi a , T. Yang a,b , Y. Yang a , C. Yu a,b , X. Zhang q , L.Zhao l , M. Zhao a , Z. Zhao l , X. Zheng l , X. Zhuang a a Institute of High Energy Physics, Chinese Academy of Sciences, 19B Yuquan Road,Shijingshan District, Beijing 100049, China b University of Chinese Academy of Sciences, 19A Yuquan Road, Shijingshan District,Beijing 100049, China c CERN, Esplanade des Particules 1, 1211 Geneva 23 d Academia Sinica, Nangang District, Taipei e LAL, IN2P3-CNRS and Universit´e Paris Sud, 91898 Orsay Cedex, France f Joint Institute for Nuclear Research, Joliot-Curie street 6, Dubna, 141980 Russia g Brookhaven National Laboratory (BNL), Upton , NY 11973, U.S.A. h University of Hassan II Casablanca, Casablanca, Morocco i Justus Liebig University Giessen, Ludwigstrasse 23, Giessen, Hesse 35390 j Institut de F´ısica d’Altes Energies (IFAE), Carrer Can Magrans s/n, Edifici Cn,Universitat Aut´onoma de Barcelona (UAB), E-08193 Bellaterra (Barcelona), Spain k National Tsing Hua University, Kuang-Fu Road, Hsinchu, Taiwan l Department of Modern Physics and State Key Laboratory of Particle Detection andEmail address: [email protected] (X. Shi)
Preprint submitted to NIMA May 18, 2020 a r X i v : . [ phy s i c s . i n s - d e t ] M a y lectronics, University of Science and Technology of China, Hefei 230026, China m Jozef Stefan Institut (JSI), Dept. F9, Jamova 39, SI-1000 Ljubljana, Slovenia n Omega Group, Brookhaven National Laboratory, BNL o University of Cagliari, Via Universit´a 40, Cagliari, Sardinia 09124 p University of Iowa, 116 Calvin Hall, Iowa City, IA 52242 q Novel Device Laboratory, Beijing Normal University, No. 19, Xinjiekouwai Street,Haidian District, Beiing 100875, China r KTH Royal Institute of Technology in Stockholm, Kungliga Tekniska H¨ogskolan, SE-10044 s Royal Holloway University Of London, University Of London, Egham TW20 0EX t Instituto de F´ısica - Universidade de S˜ao Paulo (USP), R. do Mat˜ao, 1371, CidadeUniversit´aria, S˜ao Paulo - SP 05508-090 - Brazil u LPNHE, Sorbonne Universit´e, Universit´e de Paris, CNRS/IN2P3, Paris; France v University of Mainz, Saarstr. 21, Mainz, Rhineland-Palatinate 55122 w Charles University In Prague, Ovocn´y trh 3-5, Prague 116 36 x Ohio State University, Columbus, Ohio y Southern Methodist University, 6425 Boaz Ln, Dallas, TX 75205 z Nanjing University, 22 Hankou Road, Nanjing, Jiangsu 210093, China
Abstract
To meet the timing resolution requirement of up-coming High LuminosityLHC (HL-LHC), a new detector based on the Low-Gain Avalanche Detec-tor(LGAD), High-Granularity Timing Detector (HGTD), is under intensiveresearch in ATLAS. Two types of IHEP-NDL LGADs(BV60 and BV170) forthis update is being developed by Institute of High Energy Physics (IHEP)of Chinese Academic of Sciences (CAS) cooperated with Novel Device Labo-ratory (NDL) of Beijing Normal University and they are now under detailedstudy. These detectors are tested with 5
GeV electron beam at DESY. ASiPM detector is chosen as a reference detector to get the timing resolutionof LGADs. The fluctuation of time difference between LGAD and SiPM isextracted by fitting with a Gaussian function. Constant fraction discrimina-tor (CFD) method is used to mitigate the effect of time walk. The timing2esolution of 41 ± ps and 63 ± ps are obtained for BV60 and BV170 re-spectively. Keywords:
LGAD, timing resolution, electron beam, CFD
1. Introduction
Silicon detectors have been widely used in physics experiments to trackthe positions of particles for decades. Millions of detectors form large scalescientific facilities, such as ATLAS. On these detectors, tracks could be recon-structed for the identification of each collision and corresponding secondaryparticles. The pileup on ATLAS will go up to ∼
200 after the LHC upgrade.During this upgrade, the luminosity will be increased by a factor of 5 to 10.One way to supress these pileup events is to add timing information in theexisting 3D tracking to form the 4D tracking.About 10 µm spatial resolution and ∼ ps timing resolution are requiredfor High-Luminosity LHC (HL-LHC). And the ATLAS high-granularity tim-ing detector (HGTD) is proposed to provide the timing information for the4D tracking detector. A new type of silicon detector is the low-gain avalanchedetector (LGAD) [1, 2, 3]. A highly doped p layer structure is added to nor-mal PIN diode (Fig. 1).To study the properties of IHEP-NDL LGADs, the detectors are testedat DESY using a 5 GeV electron beam in room temperature on beam lineT22 as shown in Fig. 2[4]. The electron beam is converted bremsstrahlungbeams from carbon fibre targets in the electron-positron synchrotron DESYII with up to 1000 particles per cm and energies from 1 to 6 GeV with anenergy spread of ∼
5% and a divergence of ∼ mrad .3 igure 1: Schematic of an n-on-p LGAD operated under reverse bias voltage.Figure 2: Schematic Layout of a beam test at DESY. . Properties of IHEP-NDL detector The detectors to be tested, BV60 and BV170, are provided by IHEP-NDL.2 by 2 pads each detector are both fabricated with multiple inactive guardrings (GR) (Fig. 3), four pads behaving almost the same and no marks on thesurface could be used to distinguish these pads. No electrodes for these sixGRs in this first version of IHEP-NDL LGADs are connected. The detectorhas a size of 3 . × . mm . The physical thickness of detector is 300 µm ,while the epitaxial layer thickness is 33 µm . Figure 3: Schematic Layout of a IHEP-NDL detector. 2 by 2 pads each detector arefabricated with floating six GRs, no marks for distinguishment.
The Current-Voltage (I-V) and Capacitance-Voltage (C-V) measurementsare carried out with a probe station at room temperature, GR floating(Fig. 4).The reversed bias voltages are applied to the LGADs. Only absolute valueof the bias voltage is discussed both in the following text and plots for sim-plification. Depletion voltage, breakdown voltage and foot voltage could be5xtracted from the plots. The foot voltage is the bias voltage to completelydepleted the gain layer. BV60 depletes at 50 V and breaks down at 110 V ,the foot voltage locating at 24 V . BV170 depletes at 110 V and breaks downat 160 V , the foot voltage locating at 21 V . Bias Voltage (V) C u rr e n t ( A ) HGTD
Preliminary (a) IV ´ ] - [ F - C apa c i t an c e BV60BV170
Footvoltage TotaldepletionvoltageBias Voltage (V) C apa c i t an c e - ( F - ) HGTD
Preliminary (b) C − Figure 4: The IV curve of IHEP-NDL LGADs and C − vs bias voltage for IHEP-NDLLGADs. The timing resolution of LGADs could be describe as Eq. 1. σ = σ Landau noise + σ Signal distortion + σ T ime walk + σ T DC + σ Jitter (1) σ Landau noise is effected by the non-uniform energy deposition caused bythe beam particles. σ Signal distortion represents the uncertainty caused bythe non uniform drift velocity and weighting field according to the Ramo-Shockley’s theorem [1]. σ T ime walk is caused by the arrival time shift amongpulses with different amplitudes. Constant fraction discriminator (CFD)method is used to reduce the effect from time walk. Three cases, 20%,50% and 70%, will be discussed. σ T DC is the contribution of TDC bin-ning, which is usually below 10 ps and therefore it will be neglected. σ Jitter
6s from the noise on the signal or electronics and could be estimated by σ Jitter =N/(dV/dt) ≈ t rise /(S/N) [1]. The jitter term is evaluated using pico-second laser and fast sampling rate oscilloscope[5] to be 10 ps .Another important parameter of LGAD is the gain, usually calculating bythe collected charges of LGAD and corresponding PIN. All structure shouldbe the same except the gain layer. No PINs are fabricated in this batch, sogain measurements will be reported in next papers.
3. Experimental set-up
As shown in Fig. 5, the LGAD is wire bonded to a ∼ cm square read-outboard, a 5 G Hz silicon phosphate coupled to a 2 G Hz amplifier [6], developedat the University of California Santa Cruz (UCSC). The EUDET-type beamtelescope [7] is consist of six pixel detector planes equipped with MIMOSAsensors. It is used to acquire native data by Trigger Logic Unit(TLU), clusterevent entries and form hits on the telescope planes, align the telescope planesand fit the tracks. MIMOSA planes and FEI4 are fixed as shown in Fig. 6,and DUTs (detectors under test) can be moved according to the locationrequirements. For BV60 and BV170, DUTs are located D . mm , D . mm and D . mm ( ± mm ). Negative voltages areapplied to IHEP-NDL LGADs, and a positive voltage to reference SiPM. Forsimplification, only absolute values of bias voltages applied on LGADs willbe discussed. 50 V , 90 V and 130 V are applied to BV170 while 50 V , 70 V and90 V to BV60. 26 . V bias voltage is also applied to the reference SiPM.7 igure 5: Tested detector is wire bonded to UCSC read-out board in size of ∼ cm . A5 G Hz silicon phosphate is coupled to a 2 G Hz amplifier.Figure 6: Sketch of positions in the setup.
4. Results
The oscilloscope records all waveforms generated during the measure-ment, and the main information of detectors. The EUDET-type beam tele-scope [7] is used to acquire the native data. An analysis framework, PyAna[8], is used to reconstruct the HGTD beam test data. This framework takesthe raw oscilloscope data as input and generates a root file containing a treewith various variables as output. The collected charge could be calculated by Q = I × t . For inconstant current, the charges estimated by Q = (cid:80) ni ( I i × ∆ t ).∆ t is a constant time interval between the data acquisition. CFD methodis studied by the time during which period the amplitude is higher than apercentage of the maximum amplitude. The time is used only to investigatethe timing resolution in the next section.The amplitude and collected charge is fitted with a Landau function con-8oluted with a Gaussian function. The most probable value (MPV) is takenas amplitude or collected charge at certain bias voltage. A Gaussian func-tion is used to describe the noise distribution, and the sigma of this fittedGaussian function is taken as the noise. S/N is calculated using amplitudeover noise.The performance of these two detectors are shown in Fig 7. The higherbias voltage leads to higher amplitude and more collected charges. The noiseremains almost the same at different bias voltages, resulting in S/N a similartrend with amplitude. BV60 collects more charges than 4 f C , but BV170fails to do so at tested voltages. But the properties of BV170 at higher biasvoltages are still worth being studied, for example, above 150 V . 130 V issomehow too far from the breakdown voltage, which is probably the mainreason causing the bad behavior of BV170. As mentioned above, SiPM is taken as a reference detector in order to cal-culate the timing resolutions. Three different combinations can be obtainedwith BV60, BV170 and SiPM. And these time differences(based on differentCFD valus) can be described well with Gaussian functions. The parame-ter sigma of the fitted Gaussian function could be extracted as combinedresolutions for the three systems.By assuming that the two detectors in a combination are non-correlatedand taking the system containing BV60 and SiPM as an example, the timingresolution of BV60 could be calculated as σ = (cid:113) σ − σ (2)9 ias voltage(V)50 60 70 80 90 100 110 120 130 A m p li t ude ( m V ) PulseHeight of NDL Detectors
HGTD Test Beam
Preliminary
BV170BV60 (a) Amplitude
Bias voltage(V)50 60 70 80 90 100 110 120 130 N o i s e ( m V ) Noise of NDL Detectors
HGTD Test Beam
Preliminary
BV170BV60 (b) Noise
Bias voltage(V)50 60 70 80 90 100 110 120 130 S / N S/N of NDL Detectors
HGTD Test Beam
Preliminary
BV170BV60 (c) S/N
Bias voltage(V)50 60 70 80 90 100 110 120 130 C ha r ge ( f C ) Collected Charge of NDL Detectors
HGTD Test Beam
Preliminary
BV170BV60 (d) ChargeFigure 7: (a) Amplitude distribution for BV60 and BV170. (b) Noise distribution forBV60 and BV170. (c) S/N distribution for BV60 and BV170, calculating using amplitudeover noise. (d) Collected charge for BV60 and BV170. BV60 collects more charges than4 f C . σ and δ as timing resolution and uncertainty of BV60, σ and δ asSiPM, with an uncertainty δ = (cid:115) ( σ δ ) + ( σ δ ) σ − σ (3)Same to other two cases, BV60 and BV170 ( σ with δ ), BV170 andSiPM. Then calculating for the σσ = (cid:114) σ + σ − σ ,σ = (cid:114) σ − σ + σ ,σ = (cid:114) − σ + σ + σ δ = (cid:112) ( σ δ ) + ( σ δ ) + ( σ δ ) σ ,δ = (cid:112) ( σ δ ) + ( σ δ ) + ( σ δ ) σ ,δ = (cid:112) ( σ δ ) + ( σ δ ) + ( σ δ ) σ (5)Timing resolutions for BV60 and BV170 are shown in Fig 8. The timingresolutions of these two detectors are getting better as bias voltage goesup. The effect from CFD values is shown clearly. The CFD50, black solidline(color online) for BV60 and red solid line for BV170, leads to smallertiming resolutions than CFD20 and CFD70. All the timing resolutions ofBV60, BV170 and SiPM are listed in Table 1, Table 2 and Table 3 for threedifferent bias voltages. 11 ias voltage(V)50 60 70 80 90 100 110 120 130 T i m i ng r e s o l u t i on ( p s ) Timing Resolution of NDL Detectors
HGTD Test Beam
Preliminary
BV170 CFD20 BV60 CFD20BV170 CFD50 BV60 CFD50BV170 CFD70 BV60 CFD70
Figure 8: Timing resolution for BV60 and BV170 as function of voltage. The timingresolutions of these two detectors are getting better as bias voltage goes up. The CFD50leads to smaller timing resolutions than CFD20 and CFD70.Table 1: Timing resolution for detectors at bias voltage: 50 V for BV60, 50 V for BV170,and 26 . V for SiPM. CFD[%] BV60[ps] BV170[ps] SiPM[ps]20 113 ± ± ± ± ± ±
870 122 ± ± ± able 2: Timing resolution for detectors at bias voltage: 70 V for BV60, 90 V for BV170,and 26 . V for SiPM. CFD[%] BV60[ps] BV170[ps] SiPM[ps]20 65 ± ± ±
250 67 ± ± ±
270 71 ± ± ± Table 3: Timing resolution for detectors at bias voltage: 90 V for BV60, 130 V for BV170,and 26 . V for SiPM. CFD[%] BV60[ps] BV170[ps] SiPM[ps]20 45 ± ± ±
150 41 ± ± ±
170 43 ± ± ± . Conclusion IV and CV tests have been carried out on BV60 and BV170. The voltagepoints applied on the beam test are determined according to the depletionvoltages and breakdown voltages. The amplitude of BV60 is higher than thatof BV170, while they have similar noise, which resulting in higher S/N thanBV170. The BV60 could collect more charges than 4 f C . More investigationof BV170 on collected charge at higher bias voltages near to breakdownvoltage are still worth being studied, for example, 150 V or 155 V . DifferentCFD values do effect on timing resolution, CFD50 leading to the best resultscomparing with CFD20 and CFD70. The timing resolution of BV60 is 41 ± ps , and BV170 is 63 ± ps .The reference SiPM has been used in several beam tests at CERN andDESY. Long-term exposure to particles results in a degraded performance,70 ± ps at 26 . V . And this could be improved by using a new SiPM.Irradiation hardness is another issue that matters as well. IHEP-NDLLGADs have been sent to China Institute of Atomic Energy (CIAE) andCyclotron and Radioisotope Center (CYRIC) for proton irradiation up to2 . × neq/cm . IV, CV, gain measurement and beam test will all becarried out on these irradiated LGADs. Detailed researches can be expectedin near future. Acknowledgement
The measurements leading to these results have been performed at thebeam test Facility at DESY Hamburg (Germany), a member of the HelmholtzAssociation (HGF). Thanks to Beijing Normal University for the detector14roduction, and DESY for the beam test setup and shifts. Thanks all col-leagues involved in these processes. This work was supported by the NationalNatural Science Foundation of China (No. 11961141014), the State Key Lab-oratory of Particle Detection and Electronics (SKLPDE-ZZ-202001), and theHundred Talent Program of the Chinese Academy of Sciences (Y6291150K2).
ReferencesReferences [1] Hartmut F-W Sadrozinski et al , 4D tracking with ultra-fast silicon de-tectors, Reports on Progress in Physics, 81 (2018) 026101[2] G. Pellegrini et al , Technology developments and first measurements ofLow Gain Avalanche Detectors (LGAD) for high energy physics appli-cations, Nuclear Instruments and Methods in Physics Research A, 765(2016) 12-16[3] Gabriele Giacomini et al , Development of a technology for the fabri-cation of Low-Gain Avalanche Detectors at BNL, Nuclear Instrumentsand Methods in Physics Research A, 934 (2019) 52-57[4] R. Diener et al. , The DESY II beam test facility, Nuclear Instrumentsand Methods in Physics Research A, 922 (2019) 265-286[5] Y. Yang et al , Charaterization of the first prototype IHEP-NDL LGADsensor, arXiv:1912.13211 156] N. Cartiglia et al. , Beam test results of a 16 ps timing system based onultra-fast silicon detectors, Nuclear Instruments and Methods in PhysicsResearch A, 850 (2017) 83-88[7] H. Jansen et alet al