Characterization of New Silicon Photomultipliers with Low Dark Noise at Low Temperature
PPrepared for submission to JINST
Characterization of New Silicon Photomultipliers with LowDark Noise at Low Temperature
K. Ozaki, a S. Kazama, b , c M. Yamashita, a Y. Itow, a , b S. Moriyama d , e a Institute for Space-Earth Environmental Research, Nagoya University, Nagoya, Aichi, 464-8601, Japan b Kobayashi-Maskawa Institute for the Origin of Particles and the Universe, Nagoya University, Nagoya,Aichi, 464-8601, Japan c Institute for Advanced Research, Nagoya University, Nagoya, Aichi 464-8601, Japan d Institute for Cosmic Ray Research, University of Tokyo, Kashiwa, Chiba 277-8582, Japan e Kavli Institute for the Physics and Mathematics of the Universe, Kashiwa, Chiba 277-8583, Japan
E-mail: [email protected], [email protected]
Abstract: Silicon photomultipliers (SiPMs) have a low radioactivity, compact geometry, lowoperation voltage, and reasonable photo-detection efficiency for vacuum ultraviolet light (VUV).Therefore it has the potential to replace photomultiplier tubes (PMTs) for future dark matter exper-iments with liquid xenon (LXe). However, SiPMs have nearly two orders of magnitude higher darkcount rate (DCR) compared to that of PMTs at the LXe temperature ( ∼
165 K). This type of highDCR mainly originates from the carriers that are generated by band-to-band tunneling effect. Tosuppress the tunneling effect, we have developed a new SiPM with lowered electric field strength incooperation with Hamamatsu Photonics K. K. and characterized its performance in a temperaturerange of 153 K to 298 K. We demonstrated that the newly developed SiPMs had 6–54 times lowerDCR at low temperatures compared to that of the conventional SiPMs.Keywords: Dark Matter detectors (WIMPs, axions, etc.); Photon detectors for UV, visible and IRphotons (solid-state); Solid state detectors a r X i v : . [ phy s i c s . i n s - d e t ] J u l ontents In the past three decades, numerous terrestrial experiments have been conducted to search fora faint interaction between weakly interacting massive particles (WIMPs) and ordinary matter.Among them, experiments using dual-phase (liquid/gas) xenon time projection chambers (TPCs)are leading the search for WIMPs with masses ranging from a few GeV/c to a few TeV/c [1–4].For future experiments with larger detector mass such as DARWIN [5], it is important to furtherdecrease the backgrounds for reaching the sensitivity limited by atmospheric and supernova relicneutrinos (neutrino floor) [6]. In the current experiments with liquid xenon (LXe), photomultipliertubes (PMTs) were used to detect the prompt primary scintillation (S1) with a wavelength of ∼
175 nm [7] and secondary electro-luminescence of ionized electrons (S2), which are generatedfollowing an interaction between a WIMP and a xenon nucleus. However, PMTs have severalimportant shortcomings, namely, their residual radioactivity levels [8, 9], bulkiness, and stabilityat cryogenic temperatures [10]. Therefore several alternative technologies are under considerationfor future dark matter experiments with LXe. Recently, Silicon photomultipliers (SiPMs), sensitiveto vacuum ultraviolet light (VUV), have been developed by Hamamatsu Photonics K. K. [11] andFondazione Bruno Kessler [12]. It was reported that one of these SiPMs (Hamamatsu S13371 [13])exhibited low intrinsic radioactivity [14]. Therefore SiPMs have the potential to replace PMTsfor direct dark matter experiments. However, as listed in table 1, currently available SiPM has ahigh dark count rate (DCR) of 0.1–0.8 Hz/mm at the LXe temperature ( ∼
165 K) [15], which is O (10–100) times higher than that for PMTs (Hamamatsu R11410) used in the current dark matterexperiments with LXe [16]. Such a high DCR of SiPMs can generate numerous fake S1 signals dueto accidental coincidences. In the XENON1T experiment, 3-fold coincidence with 100 ns time-window is required to create S1 signals [17]. Assuming the same requirement is applied for a futuremulti-ton LXe experiment with 1900 PMTs [5], this would result in an accidental coincidence rate of O (1) Hz for PMTs, which is similar to the fake S1 rate observed in the XENON1T experiment [18].– 1 –herefore it is necessary to reduce the DCR of SiPMs at least to the level achieved for PMTs ( ∼ ) for low energy threshold and background level [19]. Table 1 . Detector parameters for SiPMs and PMTs used in LXe experiments [13, 15, 16].
Photo-sensor SiPM PMTS13370 R11410-21Operation voltage ∼
50 V ∼ ∼ × ∼ × Photo-detection efficiency at 175 nm ∼
24 % ∼
35 %DCR at LXe temperature ( ∼
165 K) 0.1–0.8 Hz/mm ∼ We developed a new SiPM (S12572-015C-SPL, which is hereafter referred as SPL) as a prototypesensor with a lower DCR in cooperation with Hamamatsu. This SiPM is similar to the commerciallyavailable SiPM (S12572-015C-STD, which is hereafter referred as STD [20]), but its internal electricfield structure was modified to reduce the DCR as in [12]. Dark pulses of SiPMs mainly originatefrom the carriers generated thermally [21] and by the band-to-band tunneling effect [22]. The firstcomponent has a strong temperature dependence, while the second one has a weak dependence ontemperature [23]. The high DCR of SiPM near the LXe temperature originates from the carriers dueto the band-to-band tunneling, and the lowering of the internal electric field strength can suppress itseffect, which enables to reduce the DCR [24]. Table 2 shows the comparison of detector parametersfor SPL and STD. From the table we see that SPL and STD have the same properties of the activearea, number of pixels, pixel pitches, and fill factor. Also, we can see from the table that operationvoltage and single photoelectron (p.e.) gain are different because of the modified electric fieldstructure. It is noted from table 2 that both the SiPMs are not sensitive to VUV light.
Table 2 . Comparison of detector parameters between S12572-015C-SPL and S12572-015C-STD.
SiPM S12572-015C-SPL S12572-015C-STDOperation voltage ∼
100 V ∼
65 VGain at over-voltage = 6 V 1.6 × × Active area 3 mm × µ mFill factor 53 %Trench No trenchSpectral response range 320 – 900 nm– 2 – C power supply (GDP-43035)LED Amplifier (LNA-650)CuHeater DRS4 evaluation boardPC
Trigger
Function generator (AFG-31051)SiPMsDC power supply (KEITHLEY 2400 SMU)Vacuum Pump
Gas N Digital indicating controller (Chino DB1000D) Pt 100
Pulse-tube Refrigerator (PC150U)
Inner chamber Outer vacuum chamber
Pt 100
Temperature monitor (OMRON K3HB-H)
Vacuum
Figure 1 . Block diagram of the setup of this experiment.
The experiment was performed in the Kamioka underground laboratory at a depth of approximately1,000 meters. Figure 1 shows a block diagram of the setup that was used for this experiment. Thissetup consisted of an inner chamber filled with gas nitrogen along with an outer vacuum chamber.SiPMs were installed in the inner chamber and on a readout board, and the readout schematic isshown in figure 2, left. Voltage (V bias ) from a source meter (KEITHLEY 2400 Source Meter) wasapplied to the SiPMs. Signals were amplified with a low noise amplifier (RF Bay LNA-650) andread out using a DRS4 evaluation board [25], which acquired data with a rate of 1 GS/s and alength of 1024 data points per time window. Square pulses from a function generator (TektronixAFG-31051) were used to turn on an LED and trigger the data acquisition via the DRS4 board. Weoperated two SPLs labeled as SPL-1(-2) and two STDs labeled as STD-1(-2). Figure 2 (right), showsthe typical waveforms of SPL-1 (STD-1) acquired with DRS4 at a temperature of 163 (164) K andbias voltages of 94.6 (60.0) V. Temperature was measured by two Pt-100 sensors and was controlledwith a pulse-tube refrigerator (ULVAC CRYOGENICS PC150U) and a heater connected to a digitalindicating controller (Chino DB1000D). We operated the setup at a temperature range of 153 K to298 K, which was stable within ± To investigate the single p.e. gain, resolution, and breakdown voltage ( V br ) properties of bothSiPMs, we measured their responses to the blue light ( λ ∼
375 nm) from an LED. The function– 3 – kΩ 51 Ω51 Ω10 nF10 nF0.1 μF SiPMSiPM GNDGND GNDBias Voltage SignalSignal
Time [ns] A m p li t u d e [ m V ] SPL-1STD-1
Figure 2 . (Left): Readout schematic of SiPMs. It is possible to readout two SiPMs using a common biasvoltage. (Right): Typical waveforms of single p.e. for SPL (red) and STD (blue) at a temperature of 163 Kand 164 K, and a bias voltage of 94.6 V (SPL) and 60.0 V (STD), respectively.
250 0 250 500 750 1000
Pulse area [mV ns] P u l s e s / b i n Figure 3 . Pulse area spectrum of SPL-1 at a temperature of 163 K and a bias voltage of 94.6 V with a fittedGaussian function (red line). The first and second peaks correspond to pedestal and single p.e., respectively. generator inputs square pulses with a frequency of 460 Hz and a width of 30 ns to the LED. Weobtained 40,000 triggered events for each temperature and bias voltage. Figure 3 shows a pulse areaspectrum of SPL-1 at a temperature of 163 K and a bias voltage of 94.6 V.At this temperature, single p.e. gains of SPL-1 and STD-1 at a bias voltage of 94.6 V (SPL) and60.0 V (STD) are estimated to be 1.6 × and 2.0 × , respectively. The single p.e. resolutions( σ p . e . µ p . e . ) of SPL-1 and STD-1 are measured to be 17.9 % and 18.8 %, respectively, where µ p . e . ( σ p . e . ) is mean (standard deviation) of the single p.e. pulse area obtained by fitting with Gaussianfunction as shown in figure 3. Figure 4 shows single p.e. gain of SPL-1 and STD-1 as a functionof bias voltage. The single p.e. gain ( G ) can be expressed as G = C cell ( V bias − V br ) q , (4.1)where C cell is a cell capacitance and q is an electric charge. The slope of a linearly fitted functionin figure 4 corresponds to C cell / q . The cell capacitance of SPL-1 and STD-1 are estimated tobe 4.3 fF at 163 K and 5.4 fF at 164 K, respectively. Breakdown voltage ( V br ), where the gaincollapses to zero, is shown in figure 5 as a function of temperature. We can see from the figure 5– 4 – Bias voltage [V] G a i n
298 K268 K 238 K208 K 183 K 163 K
58 60 62 64 66 68 70
Bias voltage [V] G a i n
298 K268 K 238 K208 K 183 K 164 K
Figure 4 . Single p.e. gain of SPL-1 (left) and STD-1 (right) as a function of bias voltage for temperaturesranging from 153 K to 298 K.
150 175 200 225 250 275 300
Temperature [K] B r e a k d o w n V o l t a g e [ V ] SPL-1STD-1
Figure 5 . Breakdown voltage of SPL-1 and STD-1 as a function of temperature. that V br of SPL-1 and STD-1 decrease with a slope of 88.3 mV/K and 61.2 mV/K, respectively.Also, approximately at the LXe temperature, V br of SPL-1 and STD-1 is estimated to be 88.8 V and54.2 V, respectively. As discussed in section 2, V br of SPL is higher than that of STD. For all thecharacteristics described above, no significant sample dependence is observed. DCR is defined as the number of pulses per second whose height is larger than 0.5 p.e. In thisstudy, it was normalized by the active area and the fill factor for each SiPMs. At a temperaturebetween 198 K and 298 K, we acquired data with random trigger because of high values of DCR.Below 198 K, data were obtained with self-trigger with a threshold of 25 mV. Figure 6 (left) showsa pulse height distribution of SPL-1 at 163 K and an over-voltage of 6.0 V, where over-voltage isdefined as V bias − V br . The peak around 60 mV corresponds to the mean of the single p.e. pulseheight. A red dashed line in figure 6 shows the self-trigger threshold of 25 mV. The remaining noisecontribution is discarded by requiring that pulse area is larger than µ p . e . − σ p . e . . For the dataacquired with self-trigger, deadtime of the DRS4 evaluation board ( ∼
20 40 60 80 100 120
Pulse Height [mV] P u l s e s / m V Half p.e. pulse heightSelf-trigger threshold
Over Voltage [V] D a r k C o un t R a t e [ H z / mm ] SPL-1 298 KSPL-2 298 KSTD-1 298 KSTD-2 298 K SPL-1 163 KSPL-2 163 KSTD-1 164 KSTD-2 164 K
Figure 6 . (Left): Pulse height distribution of SPL-1 acquired with self-trigger at a temperature of 163 Kand an over-voltage of 6.0 V. Red dashed and solid lines show a self-trigger threshold of 25 mV and a halfp.e. pulse height, respectively; (Right): Dark count rate of SPL and STD as a function of over-voltage at 298K, 163 K (SPL), and 164 K (STD).
150 175 200 225 250 275 300
Temperature [K] D a r k C o un t R a t e [ H z / mm ] SPL-1 V over =5.0VSPL-2 V over =5.0VSTD-1 V over =5.0VSTD-2 V over =5.0V
150 175 200 225 250 275 300
Temperature [K] D a r k C o un t R a t e [ H z / mm ] SPL-1 V over =7.0VSPL-2 V over =7.0VSTD-1 V over =7.0VSTD-2 V over =7.0V Figure 7 . Dark count rate of SPL and STD as a function of temperature at over-voltages of 5 V (left) and7 V (right). discussed in section 2, between 200 K and 300 K, the contribution from thermally generated carriersis dominant; therefore, the DCR shows rapid decrease with temperature. However, at temperaturesbelow 200 K, the contribution from carriers originating from the band-to-band tunneling effectis dominant; therefore, the DCR has less temperature dependence. The DCR for SPL-1(-2) atapproximately 165 K is measured to be 0.11–0.24 (0.018–0.071) Hz/mm depending on over-voltage, less by a factor of 6–54 compared with that of STD, indicating that the changing inner fieldstructure reduced the DCR.Cross-talk probability (CTP) is calculated as N . N . , where N . and N . are the number of pulsesper second whose pulse heights are larger than 1.5 p.e. and 0.5 p.e., respectively [26]. Cross-talkoccurs when multiplied electrons in a cell enter and fire a neighboring cell. Figure 8 shows the CTPsfor SPL and STD at different temperatures as a function of over-voltage. It can be seen from thefigure 8 that CTP increases with over-voltage for both SiPMs but has a low temperature dependence.At approximately the LXe temperature and an over-voltage of 6.0 V, the CTPs of SPL-1 and STD-1are estimated to be 33.4 % and 31.5 %, respectively. No significant sample differences in the CTPare observed. – 6 – .0 5.5 6.0 6.5 7.0 Over Voltage [V] C r o ss - t a l k p r o b a b ili t y [ % ] SPL-1 298 K SPL-1 163 K
Over Voltage [V] C r o ss - t a l k p r o b a b ili t y [ % ] STD-1 298 K STD-1 164 K
Figure 8 . Cross-talk probability of SPL-1 (left) and STD-1 (right) as a function of over-voltage.
SiPM is a good candidate for photo-detectors in future dark matter experiments with LXe becauseof its low radioactivity. However, it has a high DCR, and it is necessary to reduce it at least to thesame level as it is for the PMTs ( ∼ ) used in LXe experiments [16]. With the help ofHamamatsu, we developed a dedicated SiPM (SPL) with lowered electric field for suppressing theband-to-band tunneling effect and operated it in a temperature range of 153–298 K. As a result, bymodifying the inner electric field structure, the DCR at the LXe temperature could be reduced bya factor of 6–54 compared to that of STD that depended on over-voltage. Currently, both SPL andSTD are not sensitive to the LXe scintillation light, but we are developing a dedicated SiPM sensitiveto VUV light with the help of Hamamatsu. The DCR of S13370, a commercially available SiPMfor VUV light detection, was measured to be 0.1–0.8 Hz/mm at the LXe temperature dependingon over-voltage [15]. With the same technology developed in this work, it might be possible thatthe DCR of S13370 can be reduced to the same extent, enabling SiPMs to be used in future darkmatter experiments using LXe. Acknowledgments
We thank Hamamatsu Photonics K. K. for this fruitful collaboration and the production of theSiPMs used in this study. We gratefully acknowledge the cooperation of Kamioka Mining andSmelting Company. This work was supported by DAIKO FOUNDATION, the Japanese Ministryof Education, Culture, Sports, Science and Technology, Grant-in-Aid for Scientific Research, JSPSKAKENHI Grant Number 19H05805 and 20H01931, and the joint research program of the Institutefor Cosmic Ray Research (ICRR), the University of Tokyo.
References [1] E. Aprile et al. (XENON Collaboration),
Dark Matter Search Results from a One Ton-Year Exposureof XENON1T , Phys. Rev. Lett.
Dark Matter Results from 54-Ton-Day Exposure of PandaX-IIExperiment , Phys. Rev. Lett. – 7 –
3] D.S. Akerib et al. (LUX Collaboration),
Results from a Search for Dark Matter in the Complete LUXExposure , Phys. Rev. Lett.
A direct dark matter search in XMASS-I , Physics Letters B (2019) 45–53.[5] J. Aalbers et al. (DARWIN Collaboration),
DARWIN: towards the ultimate dark matter detector , JCAP11 (2016) 017.[6] Ciaran A. J. O’Hare,
Dark matter astrophysical uncertainties and the neutrino floor , Phys. Rev. D High-accuracy measurement of the emission spectrum of liquid xenon in the vacuumultraviolet region , Nuclear Instruments and Methods in Physics Research A (2015) 293-297.[8] E. Aprile et al. (XENON Collaboration),
Lowering the radioactivity of the photomultiplier tubes forthe XENON1T dark matter experiment , Eur. Phys. J. C
546 (2015).[9] K. Abe et al.(XMASS Collaboration),
Development of low radioactivity photomultiplier tubes for theXMASS-I detector , Nuclear Instruments and Methods in Physics Research Section A , (2019)171–176.[10] E. Aprile et al. (XENON Collaboration), The XENON1T Dark Matter Experiments , Eur. Phys. J. C
881 (2017).[11] S. Ogawa et al. (MEG II Collaboration),
Liquid xenon calorimeter for MEG II experiment withVUV-sensitive MPPCs , Nuclear Inst. and Methods in Physics Research, A (2017) 528–532.[12] F. Acerbi et al.,
Silicon photomultipliers: technology optimizations for ultraviolet, visible andnear-infrared range , Instruments
Characterisation of Silicon Photomultipliers for Liquid Xenon Detectors , 2018
JINST P10022.[15] G. Gallina et al.,
Characterization of the Hamamatsu VUV4 MPPCs for nEXO , Nuclear Inst. andMethods in Physics Research, A , (2019) 371–379.[16] P. Barrow et al., Qualification tests of the R11410-21 photomultiplier tubes for the XENON1Tdetector , 2017
JINST P01024.[17] E. Aprile et al. (XENON Collaboration),
The XENON1T data acquisition system , 2019
JINST P07016.[18] E. Aprile et al. (XENON Collaboration),
XENON1T dark matter data analysis: Signal andbackground models and statistical inference , Phys. Rev. D Direct Dark Matter Search with XENON1T and Developments for Multi-Ton Liquid XenonDetectors
A New Analytical Diode Model Including Tunneling and Avalanche Breakdown , IEEE Trans. Electron Devices (1992) 2090–2098.[22] R.H. Haitz, Mechanisms Contributing to the Noise Pulse Rate of Avalanche Diodes , J. Appl. Phys. (1965) 3123. – 8 –
23] G. Collazuol et al.,
Studies of silicon photomultipliers at cryogenic temperatures , Nuclear Inst. andMethods in Physics Research, A (2011) 389âĂŞ392.[24] M. Ghioni et al.,
Large-area low-jitter silicon single photon avalanche diodes , Quantum Sensing andNanophotonic Devices V
Opto-semiconductor Handbook