An improved design of the readout base board of the photomultiplier tube for future PandaX dark matter experiments
Qibin Zheng, Yanlin Huang, Di Huang, Jianglai Liu, Xiangxiang Ren, Anqing Wang, Meng Wang, Jijun Yang, Binbin Yan, Yong Yang
PPrepared for submission to JINST
An improved design of the readout base board of thephotomultiplier tube for future PandaX dark matterexperiments
Qibin Zheng, 𝑎,𝑐
Yanlin Huang, 𝑎 Di Huang, 𝑏 Jianglai Liu, 𝑏,𝑑
Xiangxiang Ren, 𝑒 Anqing Wang, 𝑒 Meng Wang, 𝑒 Jijun Yang, 𝑏 Binbin Yan, 𝑏 Yong Yang 𝑏 𝑎 Institute of Biomedical Engineering, University of Shanghai for Science and Technology,Shanghai 200093, China 𝑏 INPAC, Department of Physics and Astronomy, Shanghai Jiao Tong University,Shanghai Laboratory for Particle Physics and Cosmology, Shanghai 200240, China 𝑐 Terahertz Technology Innovation Research Institute, University of Shanghai for Science and Technology,Shanghai 200093, China 𝑑 Tsung-Dao Lee Institute, Shanghai Jiaotong University, Shanghai, 200240, China 𝑒 School of Physics and Key Laboratory of Particle Physics and ParticleIrradiation (MOE), ShandongUniversity, Jinan 250100, China
E-mail:
Abstract: The PandaX project consists of a series of xenon-based experiments that are used tosearch for dark matter (DM) particles and to study the fundamental properties of neutrinos. Thenext DM experiment PandaX-4T will be using 4 ton liquid xenon in the sensitive volume, which isnearly a factor of seven larger than that of the previous experiment PandaX-II. Due to the increasingtarget mass, the sensitivity of searching for both DM and neutrinoless double-beta decay (0 𝜈𝛽𝛽 )signals in the same detector will be significantly improved. However, the typical energy of interestfor 0 𝜈𝛽𝛽 signals is at the MeV scale, which is much higher than that of most popular DM signals.In the baseline readout scheme of the photomultiplier tubes (PMTs), the dynamic range is verylimited. Signals from the majority of PMTs in the top array of the detector are heavily saturated atMeV energies. This deteriorates the 0 𝜈𝛽𝛽 search sensitivity. In this paper we report a new designof the readout base board of the PMTs for future PandaX DM experiments and present its improvedperformance on the dynamic range.Keywords: Photomultiplier Tube; Large Dynamic Range; PandaX. Corresponding author. a r X i v : . [ phy s i c s . i n s - d e t ] D ec ontents In recent years, dark matter (DM) experiments with dual-phase xenon time project chamber (TPC)technique, such as PandaX [1], LUX [2] and XENON [3], have been leading the search for weaklyinteracting massive particles (WIMPs) with masses above ∼ . WIMPs are among theleading hypothetical particle physics candidates for DM. In these experiments, energy depositionin liquid xenon generates prompt scintillation photons (usually called S1 signal) and delayedelectroluminescence photons (S2 signal) in gaseous xenon, which leads to excellent backgroundsuppression and signal-background event discrimination. The 3D position of the energy depositioncan be reconstructed using S1 and S2 signals. This makes it possible to fiducialize the centralregion where most of the ambient radioactivity is absorbed by liquid xenon outside the fidiucialvolume. In addition, nuclear recoil events from WIMP-xenon interaction can be discriminated fromelectron recoil events from 𝛾 or 𝛽 -xenon interaction using the ratio of S2 and S1 signals. Thesefeatures together with the scalability of the detector keep this type of experiments in the forefrontof searching for WIMPs. With increasing target mass of xenon from ∼
10 kg to ∼ 𝜈𝛽𝛽 ) of Xe, given the fact that there is 8.9% of
Xe in the natural xenon.Detection of 0 𝜈𝛽𝛽 is a direct evidence of neutrinos being their own antiparticles and violation oflepton number conservation law, which are clear signs of new physics beyond the standard model ofparticle physics. Recently, PandaX-II collaboration published a first search result for 0 𝜈𝛽𝛽 of Xewith 580 kg xenon [5]. However, the obtained limit on 0 𝜈𝛽𝛽 half-life (2.4 × yr) is several ordersof magnitude weaker than the currently most stringent result (1 . × yr) from KamLAND-Zen,a dedicated 0 𝜈𝛽𝛽 experiment. This is mainly due to much higher background rate and limited– 1 –erformance of energy resolution at the MeV energy scale. PandaX-II was designed primarily forWIMP search. The typical deposited energy expected from WIMP-xenon interaction is less than 10keV. However, the most region of interest for 0 𝜈𝛽𝛽 signals is 2-3 MeV (the Q value of 𝛽𝛽 decay of Xe is 2.48 MeV). In PandaX-II, the response of photomultiplier tubes (PMTs) in the top array forS2 signals start to become nonlinear at energies higher than ∼
100 keV due to PMT saturation. Thisdeteriorates both the energy scale and resolution significantly for energies above MeV and affectsthe 0 𝜈𝛽𝛽 search sensitivity. From the study in [5], a typical S2 signal in the top PMT array fromthe energy deposition of a 2.6 MeV gamma from
Tl corresponds to 1000k photoelectrons(PEs).The PMT with the maximum number of PEs contributes to about 30%. In PandaX-4T, the width ofthe gaseous xenon region is expected to be reduced by half compared to PandaX-II since multiplescattering events can be better identified with narrower S2 signals. Therefore, for the purpose ofsearching for both DM and 0 𝜈𝛽𝛽 signals in the same detector, the dynamic range of the PMT needsto be above 150k PE, which is much higher than that of the PMT in the baseline readout scheme inPandaX-4T.In this paper, we revise the baseline design of the readout base board used for PandaX-4Texperiment to improve the PMT’s dynamic range. The base board is a printed circuit board whichnot only hosts the PMT, but also provides high voltage (HV) for each dynode and the anode of thePMT. In the baseline design, only one signal can be read out from the anode. In the new design,one additional signal can be read out from the eighth dynode (DY8) with smaller gain but largerdynamic range. The rest of the paper is organized as follows. In section 2 we describe the designof the new base board. In section 3 we describe the system of measuring the dynamic range of thePMT and present the results from both the baseline and the new base board. In section 4 the resultsare summarized.
Both experiments of PandaX-II and PandaX-4T choose Hamamatsu R11410-20 3-inch PMTs todetect scintillation photons. The typical quantum efficiency is approximately 30% for the 175 nmxenon scintillation light. Photoelectrons at the cathode are amplified through 12 dynodes and theanode. The typical gain is 5 × when the cathode and the anode is supplied with the HV of1500 V. In PandaX-II and PandaX-4T, a split positive and negative HV scheme is adopted to reducethe relative potential to ground, which reduces the risk of discharge on the feedthrough pins. Thevoltage at each dynode is set through a resistor network shown in Figure 1(a), which is the schematicof the baseline design of the base board for PandaX-4T. In the baseline design, PMT signal is readout from the anode. As suggested in [6], usually a few decoupling capacitors are added at the laststages, which provides additional charge when the electrical pulse is being read out. This in turnimproves the linearity of the PMT. Usually the decoupling capacitors are connected in series orparallel. Two capacitors (C ,C ) are connected in series at the last two dynodes. One additionalcapacitor (C ) is required to be connected in series with the load resistor (R ) of the anode since apositive HV is applied at the anode.To improve the dynamic range of the PMT, a few modifications were made based on the baselinedesign, as shown in Figure 1(b). First is the way the decoupling capacitors are connected. In the newdesign, each capacitor at the dynode is connected to ground directly. As we will show later, adding– 2 – Cathode Dy1-HV Dy2 Dy3 Dy4 Dy5 Dy6 Dy7 Dy8 Dy10 Dy12Dy11
R15C3
G1GND GND+HVDy9
C1 C2
20M 0 7.5M 20M 10M 5M 5M 5M 5M 5M 5M 5M 7.5M 7.5M10nF 10nF 10nF100K
Anode (a) Baseline design of the base board
R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11 R14R12 R13
Cathode Dy1-HV Dy2 Dy3 Dy4 Dy5 Dy6 Dy7 Dy8 Dy9 Dy10 Dy12Dy11
R20 R19 R18 R17 R16 R15C6 C5 C4 C1 C2 C3
G1GND GNDDy8 GND
20M 0 7.5M 20M 10M 5M 5M 5M 5M 5M 5M 5M 7.5M 7.5M50 50 100K 50 50 100K10nF 10nF 10nF 10nF 10nF 10nF +HVAnode (b) New design of the base board
Figure 1 . The schematics of baseline (a) and the new (b) base board. The main differences are the filteringcapacitors configuration and the additional readout at DY8. For measuring the dynamic range, the cathode,dynode 5 and the anode were connected to -700 V, GND and +800 V, respectively. – 3 –hree more capacitors from DY8 to DY10 connected in parallel to ground improves the dynamicrange of the anode significantly. Secondly, to further improve the dynamic range, another electricalsignal with smaller gain is read out from DY8 at the same time. And during the test, we observedsome damped oscillation in the falling edge of the electrical pulse from DY8. As suggested in [6], afew matching resistors are added near DY8 and the oscillation is found to be significantly reduced.
Generator Test PMT
Bifurcated
Fiber Bundle
Monitor PMT
V1725B
Dark Box
Trigger
Translation Stage
Server
AnodeAnode DY 8
Laser Diode Optical Table
Translation Stage Translation Stage
Condenser
Lens
TTL SignalAnalog SignalLightPMT SignalDigital Data Optical Fiber
Figure 2 . The system for PMT dynamic range measurement. This consists mainly of a light source andsplitting system, a dark box, a waveform digitizer and a DAQ server.
To measure the dynamic range of both the baseline and the new base boards, we set up asystem shown in Figure 2. Driven by an arbitrary waveform generator [7], the laser diode [8] emitslight with a fixed duration time and frequency. The intensity of the light can be adjusted suchthat the PMT under test can be saturated. For measuring the dynamic range, we set the durationtime to be 10 𝜇 s, which is the typical width of S2 signals at MeV energies expected in PandaX-4T.The frequency is set to be 50 Hz. To increase the light detection efficiency of the PMT, the lightis focused by a focusing lens. The focused beam of light is separated into two beams using a3-dimensional translation stage and a bifurcated optical fiber. The relative intensity ratio of thetwo separate beams is adjusted via two additional 3D translation stages before they are detected bythe monitor PMT and the test PMT, respectively. Each PMT is located in a dark box. During themeasurement, a stable output intensity ratio was adjusted to keep the anode of the monitor PMTunsaturated, which was taken as a reference to examine the dynamic range of the PMT under test.The adopted ratio was 1:10 (monitor PMT vs test PMT) and 1:30 for measuring the anode and theDY8 of the test PMT, respectively. Three signals were read out from the two anodes of both PMTsand the DY8 of the test PMT. The anode signals were decoupled from the positive HV on a sparedecoupler board from PandaX-II. Then all three signals were digitized by a waveform digitizer(CAEN V1725B, sampling rate 250 MS/s) when it received a trigger signal which was synchronouswith the LED light emission. The recorded data were sent to a server for further analysis.Figure 3 shows an example of two recorded waveforms from the anode and the DY8 of the testPMT. In this example, the amplitude of the signal from the anode was decreasing during the lightemission which means the signal was saturated. On the other hand, the signal from DY8 was notsaturated. The digitized waveform was then integrated after baseline subtraction. The obtained area– 4 –hould be proportional to the product of the gain and the charge of the PMT if it is not saturated.For the anode, the gain is measured using single photoelectron (SPE) signals in very low-intensityLED calibration runs. When the test PMT is not saturated, the charge ratio between the anode andthe DY8 is measured to be 100:1. This ratio is used later to evaluate the charge of the PMT whenDY8 is used for the measurement. Sample index (4ns) D a t a ( AD C c oun t s ) AnodeDY8 (x15)
Figure 3 . An example of two recorded waveforms from the anode and the DY8 of the test PMT, showingthat the anode was saturated while the DY8 was unsaturated. In this figure, DY8’s signal is scaled by a factorof 15 for better visuality.
As mentioned above, in the new design of the base board, the number of decoupling capacitorsand the way they are connected are different compared to the baseline design. Figure 4 showsthe relation between charge measured by the test PMT using anode and charge measured by themonitor PMT. The anode of test PMT with the baseline base board start to saturate when theincident light is above 1000 PE and gradually reaches to full saturation. The fully saturated chargeis about 4k PE. When the capacitors are connected in parallel, the full saturation happens muchlater. For the five-capacitor and three-capacitor versions, the anode current starts to deviate upwardfrom the linearity at a certain current level and gradually reaches saturation as the incident lightlevel increases. This somewhat counter-intuitive turning behavior is consistent with that describedin Ref. [6], that the increase of amplification in earlier dynode stages due to voltage redistributionovercomes the decrease of secondary emission ratios in final stages. The fully saturated charge of thefive-capacitor new design is about 40k PE, which is 10 times larger than that of the baseline design.Figure 5 shows that dynamic range of the test PMT can be further extended if the measurementfrom DY8 is used. The dynamic range can reach about 200k PE, but with some noticeable level ofnonlinearity between 150k PE and 200k PE, which can nevertheless be corrected using real data.This is the equivalent number of PEs expected from the PMT with the maximum charge for a 3.5MeV energy deposition in the liquid xenon. Thus this range includes the most region of interest for0 𝜈𝛽𝛽 signals. – 5 – harge of monitor PMT (PE)0 10 20 30 · C ha r ge o f t e s t P M T u s i ng anode ( PE ) · Figure 4 . Relation between charge of the test PMT using anode and charge of monitor PMT. This includesmeasurements from the baseline base board and the new design with different capacitor configurations,showing the improvement on the dynamic range if capacitors are connected to ground in parallel rather thanin series. Blue curve refers to the baseline design which has two capacitors connected in series to ground.In the new design, we compared the performance of different numbers of capacitors connected in parallelto ground. One capacitor (C in Figure 1(b)) , three capacitors (C ,C ,and C in Figure 1(b)), and fivecapacitors (C ,C ,C ,C , and C in Figure 1(b)). The configuration with five capacitors is chosen for the newbase board. Future PandaX DM experiments will be more sensitive in searching for both DM and 0 𝜈𝛽𝛽 signalsin the same liquid xenon detector. In PandaX-4T experiment, this requires the dynamic range ofthe PMT to be at least 150k PE, which is beyond the range provided by the readout base board ofthe baseline design. In this paper, we have presented an improved design of the readout base board.The new design uses a new configuration of decoupling capacitors. In addition, signals can be readout both from the anode and the DY8. When DY8 is used, the dynamic range of the new base boardis measured to be 200k PE which meets the requirement of PandaX-4T. A few new base boards (seephotos in Figure 6) have been assembled in the central zone of the top and bottom PMT array ofthe detector. Their performance will be evaluated using real data from PandaX-4T.
This project is supported by grants from the Ministry of Science and Technology of China (No.2016YFA0400301 and 2016YFA0400302), a Double Top-class grant from Shanghai Jiao Tong Uni-– 6 – harge of monitor PMT (PE) · C h a r g e o f t e s t P M T u s i ng DY ( P E ) · Figure 5 . Relation between charge of test PMT using DY8 and charge of monitor PMT, showing the dynamicrange of the new base board is around 200k PE. Above that the signal from DY8 is also saturated.
Figure 6 . Photos of the new base board for 3-inch PMTs Hamamatsu R11410. Left, top view. Right, bottomview. versity, grants from National Science Foundation of China (Nos. 11875190, 11505112, 11775142and 11755001), supports from the Office of Science and Technology, Shanghai Municipal Govern-ment (18JC1410200), and support also from the Key Laboratory for Particle Physics, Astrophysicsand Cosmology, Ministry of Education. This work is supported also by the Chinese Academyof Sciences Center for Excellence in Particle Physics (CCEPP). We thank Changqing Feng atUniversity of Science and Technology of China for useful discussions.– 7 – eferences [1] PandaX-II Collaboration, Dark Matter Results From 54-Ton-Day Exposure of PandaX-II Experiment,
Phys. Rev. Lett.
Phys. Rev.Lett.
Phys. Rev. Lett.
Xe with PandaX-IILiquid Xenon Detector,
Chinese Physics C ,2019 43(11)[6] Hamamatsu Photonics, PHOTOMULTIPLIER TUBES, Basics and Applications.[7] Functional generator arbstudio 1102 is from Lecroy. https://teledynelecroy.com/doc/testing-audio-devices .[8] Laser diode and most optical elements are from thorlabs. ..