aa r X i v : . [ phy s i c s . i n s - d e t ] M a r Characterization of VUV4 SiPM for Liquid Argondetector
L. Wang a , M.Y. Guan a,b,c , H.J. Qin d , C. Guo a,b,c ∗ , X.L. Sun a,b,c † C.G. Yang a,b,c , Q. Zhao b,c , J.C. Liu b,c , P. Zhang a,b,c , Y.P. Zhang a,b,c ,W.X. Xiong b,c , Y.T Wei b,c , Y.Y. Gan b,c , J.J. Li e , a State Key Laboratory of Particle Detection and Electronics, Beijing, China b Experimental Physics Division, Institute of High Energy Physics, Beijing, China c School of Physics, University of Chinese Academy of Science, Beijing, China d School of Internet of Things Engineering, JiangNan University, Wuxi, China e School of Nuclear Science and Engineering, North China Electric Power University,Beijing, China
Abstract
Particle detectors based on liquid argon are recently become recognized asan attractive technology for dark matter direct detection and coherent elas-tic neutrino-nucleus scattering measurement. A program using a dual-phaseliquid argon detector with a fiducial mass of 200 kg to detect coherent elas-tic neutrino-nucleus scattering at Taishan Nuclear Power Plant has beenproposed. SiPM will be used as the photon sensor because of its high radio-purity and high photon detection efficiency. S13370-6050CN SiPM, made byHamamatsu, is an alternative for the detector. In this paper, the charac-terisation of S13370-6050CN SiPM, including the cross talk and after pulseprobabilities at liquid argon temperature and the temperature dependenceof break down voltage, dark count rate and relative quantum efficiency werepresented.
Keywords:
Coherent Elastic Neutrino-nucleus Scattering, Liquid Agron,SiPM ∗ Corresponding author. Tel: +86-01088236256. E-mail address: [email protected](C. Guo). † Corresponding author. Tel: +86-01088236069. E-mail address: [email protected](X.L. sun)
Preprint submitted to Journal of Instrumentation March 3, 2021 . Introduction
Coherent elastic neutrino-nucleus scattering (CE ν NS), which was firsttheorised by Freedman in 1974 [1], is the dominant for neutrinos with ener-gies less than 100 MeV. The COHERENT collaboration achieved the firstmeasurement of CE ν NS by using a 14.6 kg CsI(Na) crystal detector to de-tect the neutrinos from the spallation neutron source at Oak Ridge NationalLaboratory (ORNL) in 2017 [2]. Recently, they reported the first detectionof CE ν NS on argon using the CENNS-10 liquid argon detector [3]. A dual-phase liquid argon time projection chamber (TPC) with 200 kg fiducial masswas proposed to measure ¯ ν e -Ar CE ν NS process in China [4]. The detectoris expected to be adjacent to the JUNO-TAO experiment [5], which is about35 m from a reactor core of Taishan Nuclear Power Plant. Since the energyof the recoil nucleus is concentrated in the sub-keV region and the event rateis very low, single photon detection ability and high detector radio-purity, in-cluding the target and detector components, become the basic requirementsof the detector.Silicon photo-multipliers (SiPM), which were originally developed in Rus-sia in the mid-1980s [6], got rapid development in the past few years andexpected to be a possible replacement of conventional photo-multiplier tubes(PMT). Compared with PMTs, SiPMs have lower radioactive backgroundand higher photon quantum efficiency, which make them very promising asthe photon sensors for low background experiments. However, the gain ofSiPM is usually ∼ , thus a suitable amplifier must be used to realize themeasurement of single photon. A set of SiPM and preamplifier which can beused at liquid argon temperature is crucial for our experiment.The S13770-6050CN [7] SiPM reported in this paper is specially developedfor cryogenic experiments by Hamamatsu and it has been widely used inliquid xenon (LXe) detectors [8][9]. However,the liquid argon (LAr) detectorsrequire a lower temperature of 87 K and its performances have not beenstudied in detail.The main purpose of this work is to study the application prospect ofS13370-6050CN SiPM in our liquid argon detector. Due to our experimentis currently in the R&D stage, we believe in the field of SiPM application inLAr detector, DarkSide-20 [18] is of a good reference. According to Ref. [18],the SiPM used in a dual-phase liquid argon detector needs to meet at leastthe following two criteria:A. The dark count rate (DCR) should be less than 0.1 Hz/mm to keep2ulse shape discrimination effective;B. The total correlated noise probability, namely the sum of direct crosstalk(DiCT), delayed crosstalk (DeCT) and after pulse(AP), should be less than60% for energy reconstruction of S2 signal.In addition to the measurement of DCR and correlated noises, the changesof break down voltage and relative quantum efficiency (QE) from 87 K upto room temperatures are also introduced in this paper.
2. Experimental Setup
Figure 1: Scheme of the experimental setup.
The scheme of the experimental setup is shown in Fig. 1. A 6 mm*6 mmSiPM (Fig. 2 left-side) chip is placed in a stainless steel (SST) chamber whichis put in a dewar (Fig. 2 right-side). A liquid nitrogen tank is standing on theground nearby for cooling the system by a copper tube. Two Pt100 sensorshave been used in this system. One is located inside the SST chamber tomonitor the temperature of the SiPM in real time, and the other is located inthe inter-layer between the dewar and the SST chamber as the input signalof the temperature control system. It intermittently injects liquid nitrogenby controlling the switch of the electromagnetic valve, so as to control thetemperature. The accuracy of the temperature control system is ± igure 2: Left: A 6 mm*6 mm SiPM connected with electronics. Right:A stainless steel(SST) chamber which is put in a dewar. LED to monitor its stability [11]. The LED is driven by a pulse genera-tor (Tektronix, AFG31102, 100M, 2CH). During the whole experiment, theLED and the silicon photocell rest at room temperature. Before being sentinto the oscilloscope (LeCroy 610Zi, 250MHz sampling frequency) for pulsesrecording, signals from the SiPM are amplified by a LMH6629 preamplifierwhich worked at the same temperature as the SiPM and signals from the sili-con photocell are amplified by a warm amplifier (model AD825 LF353). TwoRAGOL DP831A DC power supplies provide ± bd ), the LED is driven bythe pulse generator with periodic square pulses of 1 kHz and the oscilloscopeis triggered by synchronous pluses. Signals of the LED trigger and SiPM arerecorded and the time window is set to 500 ns.(B) While measuring the relative quantum efficiency (QE), the setupsare basically same as case A except that the time window is set to 5 µ s andsignals from the silicon photocell are also recorded.(C) While measuring the correlated signal probabilities, the LED is off.4nly the SiPM signals are recorded and the oscilloscope is at self-triggermode. The time window is set to 500 µ s because we aim to analyze thecorrelated signals which follow behind the primitive pulses.(D) While measuring DCR, the signals are directly sent to a LTD (CAEN,N844). The output of the LTD is sent to a scaler (CAEN, N145) for counting.All the signals are recorded by the oscilloscope and the data samplingrate is set to 500 MS/s.LMH6629 is a high-speed, ultra-low noise amplifier designed for the appli-cations requiring wide bandwidth with high gain and low noise [12]. Althoughthe official claims that the amplifier works at 233 K to 398 K, the applica-tion reported in Ref. [13] shows that it can even work at 60 K. Therefore,in this work, LMH6629 is chosen to amplify the SiPM signals. The circuitdiagram, which inherits from Ref. [13], is shown in Fig. 3 and the values ofthe components used in this work is presented in Tab. 1. Figure 3: Circuit diagram of the readout. Cal is the calibration signal which is generatedby the pulse generator. Junction J C is connected while testing the amplifier stability,otherwise it is disconnected.Table 1: The values of the components shown in fig. 3 used in this work. Resistance R in R s R f R out R − R + Value 10 kΩ 10 Ω 2 kΩ 50 Ω 2 kΩ 5 Ωcapacitance C in C s C f C out Value 100 nf 30 nf 10 nf 10 nf5 . Stability test of LMH6629
The tests reported in Ref. [13] show that the amplifier LMH6629 can workat 60 K, but its cryogenic performance, especially the gain stability around87 K, is not clear yet. Thus, before measuring the characterisation of theSiPM, the stability of LMH6629 amplifier need to be estimated to guaranteethe linearity of the SiPM outputs.The circuit diagram shown in Fig. 3 is used in this measurement, butthe SiPM is removed and junction J c is connected. The input signals, whichare periodic square pulses with 1 kHz frequency, are generated by the pulsegenerator mentioned in Sec. 2. The width of the input pulse is 100 ns and theamplitude is 2 mV. Two identical and synchronised signals are generated bya AFG31102 pulse generator from two output channels. One is directly sentto the oscilloscope for pulses recording, while the other is amplified beforebeing sent to the oscilloscope. Ten thousands events were recorded in eachtemperature to calculate the charges of the input and output pulses. Thecharge ratio of the two signals is used to calculate the gain of the amplifier.Fig. 4 shows the gain variation along with temperatures and the stability iswithin 2%, which is calculated according to (Maximum-Minimum)/Average.Systematic uncertainties from the temperature controller have been presentedin the X-axis. Temperature [K]80 82 84 86 88 90 92 94 96 98 100 A m p li f i e r G a i n Figure 4: Amplifier gain VS temperature. Gain is calculate by the charge ratio of signalsbefore and after amplification. The errors of the X-axis are the accuracy of the temperaturecontrol system and the errors of the Y-axis are statistical only.
Besides temperatures, the input charge may also affect the amplifier gain.According to Ref. [7], the typical gain of S13370-6050CN SiPM is 5.8 × ,thus one photoelectron (P.E.) corresponds to a charge of ∼ Input Charge [pC]0 1 2 3 4 5 6 7 8 A m p li f i e r G a i n Figure 5: Amplifier gain VS input charge. The errors of X-axis are statistical only whilethe errors of Y-axis include the statistical and systematic errors, the systematic errors arederived from ±
4. Data Analysis
Charge [pC]-7 -6 -5 -4 -3 -2 -1 A m p li t ude [ m V ] -20-18-16-14-12-10-8-6-4 (a) Primary events(c) DeCT events(d) After pulse events(b) DiCT events Figure 6: Scatter distribution of charge VS amplitude of S13370-6050CN SiPM operatedat 87 K with an over voltage of 3 V. Amplitude and charge are negative because the pulsesare negative. Events in the black circle are dark noises, in the red circle are after pulses,in the blue circle are delayed crosstalks and in the orange circle are direct crosstalks. over ), which is the difference between the supplying bias and thebreakdown voltage. The events shown in Fig. 6 consist of the following fourparts [14–16]:A. Dark noise : In the absence of light, dark noise, which is producedby carriers generated in the depletion region, constitutes the main part ofthe signal. The dark noise is SPE event thus has an amplitude centred at1 P.E.A. and a charge centred at 1 P.E.C.. A sample pulse of dark noise isshown in Fig. 7-(a).B. Direct crosstalk: Direct crosstalk (DiCT) is caused by a photon whichis emitted by the movement of accelerated carriers in the strong field duringthe primary avalanche triggering a neighbouring pixel within an extremelyshort time interval. Thus, a DiCT signal has an amplitude centred at 2 P.E.A.and a charge centred at 2 P.E.C., which is just similar to a 2 P.E. event. Asample pulse of DiCT is shown in Fig. 7-(b).C. Delayed crosstalk: Delayed crosstalk (DeCT) occurs when the photonsproduced in the primary avalanche absorbed in the non-depleted region ofa neighbouring cell. Before triggering a second avalanche, the carriers haveto diffuse into the high-field region, thus the time difference between thetwo avalanches is around several nanosecond. Based on its mechanism, theamplitude of DeCT should be between 1 P.E.A. to 2 P.E.A. while the chargeshould be centred at 2 P.E.C.. An example pulse of DeCT is shown in Fig. 7-(c).D. After pulse: After pulse (AP) is observed when a secondary electronis trapped by some sort of impurity during the primary avalanche. Thetrapped electron is then released after a characteristic time from nanosecondto microsecond and finally results a second avalanche. In principle, the timeinterval and energy distribution depend on the location of the trapped elec-tron. The AP events should have an amplitude around 1 P.E.A. while itscharge should be between 1 P.E.C. to 2 P.E.C. because the second avalanchedischarge starts in the middle. The example pulse of AP is shown in Fig. 7-(d). 8 ime [ns]9950 10000 10050 10100 10150 10200 10250 10300 V o l t age [ m V ] -7-6-5-4-3-2-101 (a) Time [ns]9950 10000 10050 10100 10150 10200 10250 10300 V o l t age [ m V ] -14-12-10-8-6-4-202 (b) Time [ns]9950 10000 10050 10100 10150 10200 10250 10300 V o l t age [ m V ] -10-8-6-4-202 (c) Time [ns]9950 10000 10050 10100 10150 10200 10250 10300 V o l t age [ m V ] -8-6-4-202 (d) Figure 7: Example pulses for different events. (a) SPE event. (b) Direct crosstalk (DiCT).(c) Delay crosstalk (DeCT). (d) After pulse (AP).
As described in Sec. 4.1, different kinds of signals can de identified withthe charge and amplitude information. In order to better estimate the chargeand time information of DeCT and AP events, a pulse fitting program hasbeen developed.As can be seen from Fig. 7-(c) and Fig. 7-(d), the delayed pulse overlaps onthe primary SPE pulse, thus the total waveform can be described accordingto Eq. 1, which is a superposition of two Landau Distributions [17] and aconstant with seven parameters.
F it ( x ) = A Landau ( x, p , w ) + A Landau ( x, p , w ) + BL (1)where A , A represent the amplitudes, p , p represent the locations andw , w represent the widths of the two pulses, respectively, BL represents thebaseline. The first term of Eq. 1 is for describing the primary pulse and thesecond term is for the delayed pulse. Fig. 8 shows an example of the fittingresult of an AP event. DeCT events have the similar topological structurewith AP events, thus Eq. 1 can be used to fit both kinds of signals.9 eak1 1.004e+004width1 9.806Amp1 32.47peak2 1.01e+004width2 9.232Amp2 12.33baseline 9.693e-015 Time [ns]9950 10000 10050 10100 10150 10200 10250 10300 V o l t age [ m V ] -5-4-3-2-101 peak1 1.004e+004width1 9.806Amp1 32.47peak2 1.01e+004width2 9.232Amp2 12.33baseline 9.693e-015 Figure 8: A sample of AP event fitted according to Eq. 1. The black curve is the totalwaveform, the red dashed curve is the primary pulse, the blue curve is the after pulse.Baseline is not plotted because it is very small.
5. Results
DCR and total correlated noise probability changes along with the overvoltage, therefore the break down voltage must be accurately measured.
The breakdown voltage (V bd ) is the bias point at which the electric fieldstrength generated in the depletion region is just sufficient to create Geigerdischarge. When the bias is V bd , the gain of a SiPM, which is defined as inEq. 2, drops to exactly zero. G = Qe = R I ( t ) dtq e = R V out dtR load G Amp q e (2)where G is the gain of the SiPM, Q is the charge of one avalanche, q e isthe charge of an electron, I is the current, V out is the output amplitude ofthe signal, R load is the matching resistance of the oscilloscope, which is 50 Ω,G Amp is the gain of the preamplifier, which is around 33 in this test.V bd can be calculated as the intercept of linear fits with X-axis which ispresented in Fig. 9. Fig. 10 shows the change of V bd along with tempera-ture. Above 120 K, V bd decreases linearly with temperature, which is about0.5 V/10 K, while below 120 K the decrease tends to be slow.10 perating Voltage [V]40 42 44 46 48 50 G a i n × Figure 9: Gain curves at different temperature, errors in the X-axis is the accuracy of thebias supplier, which is 1 mV, errors in the Y axis are statistical only.
Temperature [K]80 100 120 140 160 180 200 220 V bd [ V ] Figure 10: Change of V bd along with temperature, the errors in the X-axis are the accuracyof the temperature system, which is ± The event rate of dark noises is called dark counting rate (DCR), which iscaused by thermally generated electrons that trigger an avalanche in activevolume. DCR can be measured with a counting system with a threshold of0.5 P.E. level. The change of DCR with threshold is shown in Fig. 11 andthe event rates at 0.5 P.E., 1.5 P.E., and 2.5 P.E. are indicated in the plotby the vertical dotted lines.As discussed in Ref. [14–16], DCR depends on both temperature andV over . The study of those dependences, shown in Fig. 12, indicate that theDCR of the tested SiPM is ∼ at 87 K with an over voltage of4 V. 11 hreshold [mV]0 5 10 15 20 25 30 ] DCR [ H z / mm Figure 11: DCR of the SiPM operated at V over =4 V at 298 K as a function of thediscriminator threshold. The locations of 0.5 P.E., 1.5 P.E., and 2.5 P.E. are indicated inthe plot by vertical dotted lines separately.
Temperature [K]100 150 200 250 300 ] DCR [ H z / mm -3 -2 -1
10 Vover [V]2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 ] DCR [ H z / mm Figure 12: Left: Change of DCR along with temperatures, V over is set to 4 V during thetest, errors of the X-axis are the temperature accuracy of ± over at 87 K, the errors shown inthe X-axis are caused by the temperature variation and errors in the Y-axis are statisticalonly. Correlated signals are the general term of DiCT, DeCT and AP, thedefinition and characteristics are introduced in Sec. 4.1. Fig. 13 shows thechanges of the probabilities of DiCT, DeCT and AP with different V over .At 87K, the total correlated signal probability is ∼
10% at V over = 3 V,which could satisfy the requirements of using SiPM in a dual phase LArdetector [18].The time interval distribution between correlated pulses (DeCT and AP)and the primary DCR is shown in Fig. 14. X-axis is the time differencebetween the delayed pulse (p ) and the primary pulse(p ), which is calculated12 verVoltage [V]2 2.5 3 3.5 4 4.5 5 R a t e [ % ] AP DeCT DiCT
Figure 13: The changes of the probabilities of DiCT, DeCT and AP along with V over at87K. The errors in the X-axis are caused by the temperature variation and the errors inthe Y-axis is statistical only. as p -p of Eq. 1. Y-axis is the charge of the delayed pulse, which is theintegral of the blue line shown in Fig. 8. The plot shows that there is adirect proportion between the delay time and pulse energy for AP events,which possibly indicates that the further the electron is trapped from anode(higher energy), the faster the electron is released from the impurity (lesstime interval). Delay Time [ns]0 20 40 60 80 100 120 140 C ha r ge [ p C ] (a) After Pulses(b) DeCT Figure 14: The time difference between correlated signals and primary DCR VS the chargeof the correlated signals (DeCT and AP). For better understanding, the charge is taken asabsolute value. X-axis is the time difference between the delayed pulse and the primarypulse, Y-axis is the energy of the delayed pulse. Events in the red circle are APs and in theblack circle are DeCTs. The events in the time range of 80-100 ns with charge less than1 pC is due to failed fitting, which are actually normal DCR events with a little largerbaseline jitter. .4. Relative quantum efficiency One of the advantages of S13370-6050CN SiPM is that it can directlydetect the photons emitted by liquid argon, of which the wavelength is peakedat 128 nm. But the official also claims that it has the largest quantumefficiency at ∼
500 nm [7]. For liquid argon detectors, some [19, 20] prefernot to read the 128 nm photons directly, one reason is that the photon sensorshave very low quantum efficiency at 128 nm and another reason is that thewavelength shifter, 1,1,4, 4-tetraphenyl-1,3-butadiene (TPB) [21], has morethan 100% photon conversion efficiency [22], which means that more thanone 420 nm photons will be emitted when one 128 nm photon is absorbed.According to our current experimental scheme, TPB will also be used. InRef. [7], the quantum efficiency is measured at 298 K, while according toRef. [23], the quantum efficiency at room temperatures and liquid argontemperature may be very different. In this work, a 425 nm LED is used tomeasure the relative quantum efficiency with an over voltage of 3 V. Theresult, plotted in Fig. 15, indicates that the quantum efficiency difference ofS13370-6050CN SiPM between 300 K and 87 K is ∼ Temperature [K]100 150 200 250 300 R e l a t i v e Q E Figure 15: The relative quantum efficiency at different temperatures. The results arenormalised to the point of 300K. Errors in the X-axis are the temperature accuracy of ±
6. Summary
A program of using liquid argon to detect ¯ ν e -Ar CE ν NS at Taishan Nu-clear Power Plant has been proposed and SiPM will be used as the photonsensors. In order to study the possibility of using S13370-6050CN SiPM in14ur LAr detector, a cryogenic system has been developed and the tempera-ture dependences of V bd , DCR, correlated noises and relative QE has beenmeasured in detail. Our results show that the DCR is below 0.003 Hz/mm and the total correlated noise is less than 10% at 87 K with 3 V over voltage,which indicate that the S13370-6050CN SiPM made by Hamamatsu is a goodcandidate for our liquid argon detector.
7. Acknowledgements
This work is supported by National Key R&D Program of China (GrantNo. 2016YFA0400304) and National Natural Science Foundation of China(Grant No. 12020101004).The authors would like to thank Alessandro Razeto, Paolo Musico andGemma Testera of INFN for their help during the design of the readoutsystem.
ReferencesReferences [1] D.Z. Freedman, Coherent effects of a weak neutral current, Phys. Rev.D 9, 1389 (1974).[2] COHERENT collaboration, Observation of Coherent Elastic Neutrino-Nucleus Scattering, Science 357, 1123-1126 (2017).[3] COHERENT collaboration, First Detection of Coherent ElasticNeutrino-Nucleus Scattering on Argon, arXiv:2003.10630v6.[4] Yu-Ting Wei, Meng-Yun Guan, Jin-Chang Liu, Ze-Yuan Yu, Chang-GenYang, Cong Guo, Wei-Xing Xiong, You-Yu Gan, Qin Zhao, Jia-Jun Li,Prospects of detecting the reactor ¯ ν ee