Development and Performance of a Sealed Liquid Xenon Time Projection Chamber
Yuehuan Wei, Jianyu Long, Francesco Lombardi, Zhiheng Jiang, Jingqiang Ye, Kaixuan Ni
PPrepared for submission to JINST
Development of a Sealed Liquid Xenon Time ProjectionChamber with a Graphene-Coated Electrode
Yuehuan Wei, Jianyu Long, Francesco Lombardi, Zhiheng Jiang, Jingqiang Ye,Kaixuan Ni Department of Physics, University of California San Diego, La Jolla, CA, 92093, USA
E-mail: [email protected] , [email protected] Abstract: The liquid xenon (LXe) time projection chamber (TPC) technology is leading thedark matter direct detection in a wide range of dark matter masses from sub-GeV to a few TeV.To further improve its sensitivity to sub-GeV dark matter and its application in reactor neutrinomonitoring via coherent elastic neutrino-nucleus scattering (CE ν NS), more understanding andsuppression of single/few electrons background rate are needed. Here we report on the design andperformance of a sealed LXeTPC with a graphene-coated fused silica window as the cathode. Thepurpose of the sealed TPC is for isolating the liquid xenon target volume from the majority of out-gassing materials in the detector vessel, thus improving the liquid xenon purification efficiency andreducing the impurity-induced single/few electrons background. We investigated the out-gassingrate and purification efficiency using the data from the sealed TPC with a simple purification model.The single electron signals from the photoionization of impurities in LXe are obtained and theircorrelation with the LXe purity is investigated. The photo-electron emission rate on the graphene-coated electrode is compared to that from stainless steel, the electrode material typically used inLXe detectors. We discuss the possible further improvement and potential applications of the sealedTPC for the next generation liquid xenon experiments for dark matter and neutrino physics.Keywords: Dark Matter, Weakly Interacting Massive Particles (WIMPs), Liquid Xenon TimeProjection Chamber (LXeTPC), Sealed TPC, Graphene-Coated, Electrode Corresponding authors Current institution: Department of Physics, University of Chicago Current institution: Department of Physics, University of Coimbra a r X i v : . [ phy s i c s . i n s - d e t ] J u l ontents Co 53.2 Time evolution of purification 73.3 Single electron detection 93.3.1 Single electron identification 93.3.2 Photoionization on electrodes and LXe impurities 11
The technology of Liquid Xenon Time Projection Chamber (LXeTPC) is widely used in rare eventsearches, such as direct detection of dark matter (DM) [1–3] and neutrino experiment [4]. TheLXeTPC-based detectors are leading the searches for DM in form of weakly interacting massiveparticles (WIMPs) with masses above a few GeV [1]. LXe and also liquid argon (LAr) detectorshave been used to search for sub-GeV DM via their scattering on electrons [5–8], but furtherimprovement on sensitivity requiring more understanding and suppression of background at singleand a few electrons level. A low background and single-electron sensitive LXe detector also haspotential applications to detect coherent elastic neutrino-nucleus scattering (CE ν NS) from reactoror solar neutrinos [9].The LXeTPC detects primary scintillation signals (S1) in liquid xenon. The ionized electronsare drifting into the gas phase from the liquid xenon, where electroluminescence is produced undera strong electric field ( ∼
10 kV/cm), called proportional scintillation light (S2). Both S1 and S2are recorded by photomultiplier tubes (PMTs). The z coordinate of the interaction can be extractedfrom time difference between S1 and S2, while the ( x , y ) coordinates can be reconstructed from thelight pattern of S2 signal on top PMT array. The S2/S1 ratio is used to discriminate between nuclearand electronic recoils for background reduction. Benefits from the amplification of S2 signals, thistechnology can detect very low energy events in a few electrons level [7]. However, the backgroundrate at single/few electrons level is still the limiting factor to improve the sensitivity. Photoionizationon metal surfaces of TPC material, such as electrodes, and impurities in LXe are believed to bethe main background source at the single/few electrons level. To improve our understanding andin an attempt to suppress the single/few electrons background, a sealed TPC with a cylindrical– 1 –crylic field cage and fused silica windows is developed. In an attempt to further suppress theelectrode-induced single electrons, a graphene-coated electrode is used.In this paper, the design and operation of the sealed TPC are described in Sec. 2, including aCOMSOL [10] simulation for the design optimization. We investigate the purification efficiencyand out-gassing rate based on the LXe purity evolution and a simple purification model, in Sec. 3,followed by a discussion on the rates and production mechanisms of single electrons and theircorrelation with the LXe purity and electrode materials. The schematic design of the sealed TPC is shown in figure 1 (left). A 3D mechanical design isshown on the right which reflects the actual size and structural details of each component. Atarget volume of ∼ Φ
63 mm × h
59 mm, corresponding to 0 . O ) respectively. The mesh pitch is 2 mmand the wire diameter is 0.1 mm, resulting in an optical transparency of 92%. Two pieces of fusedsilica windows ( O and O ) are placed at the top and bottom of the acrylic cylinder for isolating theLXe target volume from outside. Teflon o-rings are placed between the acrylic flange and fusedsilica windows for sealing. For the bottom fused silica window, a single-layer graphene was coatedon the surface to serve as the cathode electrode ( O ) with ∼
96% transparency [11]. The coatingwas done by Graphene Laboratories Inc. [12]. To make a better HV connection, gold powder wascoated on the edge of the fused silica window as shown in figure 2 (right). Four field-shapingrings ( O ) are connected via a resistor chain to ensure a uniform drift field between the negativelybiased cathode and the gate at ground potential. The distance between the bottom shaping ringand cathode is doubled due to insufficient installation space caused by protrude structure ( O ) forholding the bottom silica window, thus a resistor with twice resistance value is used for this region.A thin layer of PTFE ( O ) is placed on the inner wall of the TPC to increase the light collection. Four1" PMTs (Hamamatsu R8520, O ) are placed above the top fused silica window for signal detection.The liquid surface is precisely kept in the middle of anode and gate electrodes at 2.5 mm abovegate, controlled by a open slit ( O ). During detector operation, the purified xenon is continuouslyfilling into the inner TPC through an acrylic tube ( O ). Once the liquid surface reach the position ofthe slit, the overflowed LXe falls into S.S vessel where the TPC sited, and then being pumped outfor purification. The purified xenon, through a SAEN getter, is re-liquefied and fed into the TPC.Although the TPC is not fully sealed due to the small slit at the liquid overflow position, the purifiedxenon is always fed directly into the LXe target without mixing to the xenon outside of the TPC,thus improving the purification efficiency. The gas gap is automatically obtained at the position ofthe liquid xenon overflow without the need of extra liquid level control.– 2 – Xe in tube (Acrylic)
LXe out Fused Silica windowLXe surfaceAcrylic wallLXe surface (Exterior)
LXeGXe
ScreeningAnode
LXe out
GateShaping ringsPTFE layer
Graphene-coated Cathode (on Fused Silica window)
PMTsLXe in E e E d D = 63.1 mmh = 58.7 mm
124 986 35 7 Figure 1 . (Left) The conceptual design of the sealed TPC. (Right) The 3D TPC design: 1 - Four 1" PMTson top array. 2 - Fused silica window. 3 - TS, anode and gate electrodes. 4 - Slit for liquid level control.5 - Field-shaping rings in copper. 6 - PTFE sheet for increasing light collection. 7 - Acrylic tube for liquidxenon filling. 8 & 9 - Graphene cathode: graphene-coated on fused silica window. 10 - “Protrude” structurefor holding the fused silica window.
Figure 2 . (Left) Etched S.S mesh as TS, anode and gate electrodes. (Right) Graphene-coated fused silicawindow as cathode with gold-coated on the edge for HV connection.
An electric field simulation was performed to investigate the TPC field uniformity using a finiteelement analysis software package COMSOL [10]. 3D model was built for precise estimation offield distribution. Limited to computing power, the etched mesh were modeled approximately asparallel wires with diameter of 0.1 mm while the graphened-coated cathode was modeled as a plate.– 3 –he field distribution from simulation is shown in figure 3. The anode and cathode are set at 4.0 kVand -1.5 kV respectively, which is the HV setting for all the presented data in the paper. The targetvolume of TPC was divided into two parts, center ( r < R/2) and edge ( r > R/2) according to theevents distribution as shown in figure 5 (left). R is radius of the target volume. Most of events arelocalized in the edge region of the TPC due to the high self-shielding power of LXe, so the fielddistribution in this region was investigated more closely. The drift and extraction field as a functionof z in edge region is shown in left and right of figure 4, respectively. The black solid line is themean field in each z bin. The dashed-vertical lines in figure 4 (left) represent the positions of theelectrodes, shaping rings (SRs) and the protrude structure. The high field near the gate is causedby the field leaking from the high field region between gate and anode, while the high field aroundprotrude is due to the dielectric constant difference between LXe and acrylic. To avoid the effectof field non-uniformity, the region from the middle of SR1 and SR2 to SR4 is used for the eventsselection in our main analysis, which gives a drift field of (255 ±
8) V/cm as shown in figure 4 (left).In right of figure 4, the positions of the electrodes and liquid surface are marked as dashed-verticallines, (5.3 ± Figure 3 . The simulated electric field in TPC by COMSOL with anode and cathode at 4 kV and -1.5 kV,respectively.
Figure 4 . Drift (left) and extraction field (right) as a function of z in the edge region of figure 3. – 4 – .2 Detector operation For stable operation of the TPC in a high purity LXe environment, a cryogenic system and apurification system have been built. The cryogenic system consists of a pulse tube refrigerator(PTR) from Iwatani corporation, a heat exchanger and a vacuum insulated S.S vessel. The PTRconsists of a PDC08 cold head and an SA115 helium compressor, providing 40 W of cooling powerat 165 K. During operation liquefied xenon drips through a 1/4" S.S tube into the vessel wherethe TPC is located. Benefit from the sealed structure, only a small amount of xenon is requiredto keep a stable operation. In total 1.13 kg xenon is used with 0.54 kg sensitive target inside thesealed TPC. The vessel is connected to a gas purification system, which contains a SAES Getter(model PS3-MT3-R-2) for removing impurities such as water and oxygen from xenon gas down topart-per-billion (ppb) level. The detailed configuration of the cryogenic and purification system canbe found in [13].To investigate the detector performance, a 122 keV Co gamma ray source was placed ex-ternally for detector calibration and electron lifetime monitoring. The source was placed outsidethe detector at the same height as the TPC center during calibration. Without LXe surroundingthe sealed TPC, the low energy gamma rays can easily reach the sensitive LXe target volume thusproduce sufficient interaction rate for detector performance study. To record the signals, the PMTwaveforms are digitized by CAEN V1720 FADC with a sampling frequency of 250 MS/s after alow noise amplifier ( × Co Two-phase xenon TPC allows the reconstruction of 3-dimensional (3D) positions based on thedetection of S1 and S2 signals. A simple ( x , y ) position reconstruction based on center-of-gravitymethod is used. Figure 5 (left) shows the reconstructed position of the events from Co sourcewhich is placed on the right-bottom in ( x , y ) coordination. Most of events are localized on the edgeof the TPC, due to the strong self-shielding property of LXe. The reconstruction capability is quitedegraded for edge events due to its insensitive to S2 pattern. The z position can be inferred from thedrift time of ionized electrons, i.e. the time difference between S1 and S2. In figure 5 (right), thelocation of gate and cathode electrodes, denoted as red dashed lines, can be clearly observed in thedrift time distribution. The locations of the four field-shaping rings also can be identified (magentadashed lines) which shows the impact of non-uniform shaping rings in the design. The valley isprobably caused by the electrons attachment on the TPC wall around shaping rings locations. Thedescending step around ∼ µ s is caused by a protrude structure as shown in figure 1 (right) forholding the bottom silica window. – 5 – .6 0.4 0.2 0.0 0.2 0.4 0.6 X [a.u] Y [ a . u ] Edge Inner E v e n t s / b i n Drift time [ s] C o un t s / b i n Gate
SR1 SR2 SR3 SR4 Protrude
Cathode
Figure 5 . (Left) Reconstructed ( x , y ) position based on a simple center-of-gravity method for Co gammarays interacting in the TPC. The red dashed circles denote the radius R and R /2 of the LXe target volumerespectively. (Right) The drift time distribution. The red dashed lines indicate the location of gate andcathode electrodes respectively. The magenta dashed lines show the location of field-shaping rings while thegreen dashed line is the location of the protrude structure. S1 [PE] L o g ( S [ P E ]) E v e n t s / b i n Drift time [ s] L o g ( S [ P E ]) = 455.5 ± 96.5 us 10 E v e n t s / b i n Figure 6 . (Left) Correlation of the observed S1 and S2 signals in number of photo-electrons (PE) from Cogamma rays interacting in the liquid xenon target. (Right) S2 as a function of drift time from these events. Anexponential fit was applied to the mean S2 values (magenta dots) from each drift time bin in the range of [10,26] µ s to obtain the electron lifetime (see text). The drift time window is selected to avoid the non-uniformdrift field outside the range shown in figure 4. A representative dataset is shown in figure 6. Left plot shows the Log (S2) - S1 space from Co. The location of the 122 keV peak gives a mean S1 of 16.3 PE and S2 of 11,922 PE. Thephoton detection efficiency ( g
1) and the electron detection factor ( g
2) are obtained by comparingthe observed S1 and S2 values with the yield of 122 keV monoenergetic peak in the Noble ElementSimulation Technique (NEST) calculator [14]. Based on NEST calculator, 5685 photons and 3344electrons are expected from 122 keV gamma line at 255 V/cm drift field, giving g / γ and g / e − in this detector setup. It is expected that this detector geometry would result in amuch lower light collection efficiency, compared to the dark matter detectors such as XENON1T [15]and LUX [16], due to the lack of a bottom PMT array and additional light transmission loss throughthe fused silica window. Nevertheless, the resulting g t d ) is shown in the right plot, where an exponential fit ofS2( t d ) = S2(0) · exp(– t d / τ ) was applied to derive the electron lifetime ( τ ). The magenta dots are thegaussian mean of S2 in each t d bin. The middle part of TPC in t d range of [10, 26] µ s , correspondsto the (255 ±
8) V/cm field region in figure 4 (left), was selected for the fitting, and resulting in ameasured electron lifetime ( τ ) of (456 ± µ s as indicated by the red dashed line. The S2 wasslightly dropped with the drift time approaching to gate and near the bottom caused by the fielddeformation mentioned in Sec. 2.1. The data with drift time below 10 µ s and above 26 µ s was notincluded in the electron lifetime fitting. To investigate the purification efficiency of the sealed TPC, we monitor the electron lifetime atdifferent gas xenon circulation and purification speed. The getter used (SAES model PS3-MT3-R-2) has a maximum purification speed of 5 standard-liter-per-minute (SLPM) for nitrogen, heliumand argon, but 3 SLPM for xenon according to the specification. During the detector operation, thexenon flow rate through getter was set at different values to investigate the purification efficiencyand understand the xenon purity. The time evolution of electron lifetime τ was monitored for severaldays using the calibration data and divided into four periods as shown in figure 7. - : - : - : - : May E l e c t r o n L i f e t i m e [ s ] F = 2 SLPM 5 SLPM
P-1 P-2 - : - : - : - : June bypass Getter 5 SLPM bypass Getter
P-3 P-4
Figure 7 . Electron lifetime as a function of time for different purification settings. A simple purificationmodel (blue lines, see text) is used to fit the data to obtain parameters relevant to the out-gassing and liquidpurity. The highest electron lifetime obtained is a little over 0.5 ms, with large uncertainties due to the limiteddrift time window. Data were taken in 2019.
A simple purification model was developed to describe the electron lifetime evolution duringour run. During the gas xenon circulation and purification, the impurity concentration in the targetLXe can be described as, M ρ dn ( t ) dt = R o − n ( t ) η F (3.1)– 7 –here n ( t ) is the oxygen-equivalent electro-negative impurity concentration in unit of part-per-billion (ppb) in xenon. M is the total mass of LXe inside the TPC ( ∼ ρ is the density of xenon gas (5.9 × − kg/liter). R o is a constant “effective” out-gassing rate fromdetector materials into the LXe target. Note that the out-gassing here is not the same concept invacuum, but including the impurities that are diffused/dissolved/leaked into the LXe target. F is thexenon gas circulation flow rate. We added the η term to describe the purification efficiency. In casethe impurity in the xenon is not fully removed by the getter, or if additional impurities are added tothe purified xenon gas before it reaches the target volume inside the TPC, the η term will be lessthan 1.The time evolution of the electron lifetime, at a constant purification speed F , thus can beobtained as, τ ( t ) = n e − ρη FM t + R o η F (cid:16) − e − ρη FM t (cid:17) [ µ s ] (3.2) n is the initial (t = 0) concentration. The constant 368 is the electric field dependent termwhich can be derived from Ref. [17] to relate the electron lifetime and impurity concentration inLXe as, τ ( t ) ≈ n ( t ) [ µ s ] . The electron lifetime would reach a plateau after reaching an equilibrium, τ ( t → ∞) = × η FR o [ µ s ] (3.3)Each of the four periods in figure 7 is fit with the relevant purification model equations. Thefitting results are summarized in Tab. 1. The period of P-1 is fit with Eq. 3.2 with F = 2 SLPMwhile the P-2 is fit with a constant value of Eq. 3.3 suppose the maximum electron lifetime has beenreached after the flow rate being switched to F = 5 SLPM. The model fitting gives 0.51 ± ± η for P-1 and P-2 respectively. Less than 100% purification efficiency is obtained inP-1 due to the LXe outside the TPC which is largely contaminated with higher out-gassing volumecompared to the inside of the sealed TPC. The further drop of purification efficiency from P-1 to P-2indicates the degrade of the getter’s purification capability after reaching its maximum specified flowrate (3 SLPM) for xenon. The effective impurity out-gassing rate R o of ( . ± . ) × − liters/sis obtained from the fit.The period of P-4 is fit with Eq. 3.2. A η of (0.21 ± τ in June is mostly caused bythe operations on circulation pipes for the purification system upgrade. A higher electron lifetimewas achieved in June compared to May, which indicates that the out-gassing outside the TPC issmaller in June compared to May after ∼ R o of ( . ± . ) × − liters/s.During the period of P-3, the getter was bypassed for understanding the purification processbut the flow was kept at 5 SLPM in the bypass pipes, thus the impurity concentration in LXe can bedescribed by Eq. 3.4 and solution in 3.5. M ρ dn ( t ) dt = R o + n out F (3.4)– 8 – able 1 . The fitting results from figure 7. Note that the R o from P-1 fitting is used in Eq. 3.3 for P-2 fitting. Periods Circulation n [ppb] R o [10 − liter/s] η n out [ppb]P-1 2 SLPM through getter 66.2 ± ± ± ± ± ± ± ± ± τ ( t ) = n + ( R o + n out F ) ρ M t [ µ s ] (3.5)where n out is the impurity concentration in the LXe outside the sealed TPC where the xenonis taken out through the circulation. In this diagnostic run, impurity from outside of the sealed TPCis brought inside quickly reducing the liquid xenon purity thus lowering the electron lifetime.By fitting the period of P-3 with Eq. 3.5, (2.7 ± × − liter/s of ( R o + n out F ) was obtained.The R o is very small according to the values obtained from P-1 and P-4 periods. An impurityconcentration of (32.4 ± F and negligible R o . This indicates the LXe outside the sealed TPC containsmuch higher impurity than that inside. In the current design of the sealed TPC, the LXe in thesensitive target is always first mixed with the LXe outside, then passing through the getter, effectivelyincreasing the burden of the getter to purify the LXe outside the target and reducing the purificationefficiency. In addition, the impurity gas molecules from the outside the sealed TPC can “leak” intothe LXe target through the small slit between the anode and gate. That “leak” certainly increases thevalue of R o . A LXe detector with the target volume completely sealed will not only have reduced R o but also improved purification efficiency, thus electron lifetime can be further improved. Acompletely sealed TPC would require a different cryogenic and purification system design and willbe pursued in the future. Understanding and controlling the background at single and a few electrons level is critical to thesensitivity of sub-GeV dark matter electron scattering or CE ν NS of low energy reactor neutri-nos. There are two types of single/few electrons background. One is the prompt single electronbackground within one max drift time window, typically at tens or hundreds µ s following a largesignal (S1 or S2). The other type is the delayed single electrons that can extend to hundreds of msfollowing a large signal (S1 or S2). In this study, we investigate the prompt single electron rateand its correlation with impurity levels in LXe or different electrodes. The study of delayed singleelectron emission [18] requires much lower background and will be pursued in the future. The raw data for single electron study was acquired with window length of 80 µ s with 50% posttrigger. Most of the events were triggered on S2 peak, thus ∼ µ s time window after S2 can beused for extracting the single electrons from the photoionization by S2 light. Figure 8 shows the– 9 –elay time of the small S2s after the main one. Three clear peaks can be observed, indicating thelocation of the gate, protrude and cathode respectively. The peaks (red dashed lines) around gateand cathode electrodes are considered to be from the photoionization of electrode surface by themain S2. The peak (green dashed line) around the protrude structure is most probably caused by theelectrons accumulation on the edge of the acrylic material. The other region is the photoionizationof main S2 light on the impurities in the bulk LXe, and overlapped with the PMT after-pulses. Delay time [ s] C o un t s / b i n Gate Protrude
Cathode
Figure 8 . Delay time of the small S2s, corresponding to single or a few electrons, after the main S2 from arecoil event. The red dashed lines indicate the location of the gate and cathode electrodes respectively, whilethe green dashed line is the location of the protrude structure for holding the fused silica window as shownin figure 1 (right).
S2 [PE] C o un t s / b i n Detection efficiency1 e- 2 e- 3 e- 4 e- = 4.7 ± 0.1 PE/e- = 2.3 ± 0.1 PE/e- fitdata 0.00.20.40.60.81.0 E ff i c i e n c y S2 [PE] C o un t s / b i n Detection efficiency1 e- 2 e- 3 e- 4 e- = 4.3 ± 0.3 PE/e- = 1.8 ± 0.1 PE/e- fitdata 0.00.20.40.60.81.0 E ff i c i e n c y Figure 9 . Typical small S2s spectrum within 2 µ s time window for gate (left) and cathode (right) electroderespectively. The spectrum was fit with a sum of 4 Gaussians, supposing that the spectrum comprises a sumof one to four electrons, multiplied by a function (see text) to take into account the detection efficiency. To extract the single electron signals, the events within 2 µ s time window around the peak of thegate (or cathode), as shown in the pink shaded region in figure 8, were selected. The correspondingS2 spectra are shown in figure 9. The spectra are fit using a sum of several Gaussian functions withmean i µ and standard deviation √ i σ , where i represents 1 ∼ f ( S ) = e −( S − A )/ B + with A and B as free parameters.This fit assumes that the S2 spectrum comprises a sum of one to a few electrons S2 signals with– 10 –he first Gaussian ( µ , σ ) provides the detected PEs per electron which has been extracted to GXe.The efficiency curve reflects the efficiency of the S2 peak-finder algorithm. The single electrongain ( µ ± σ ) of 4.7 ± − and 4.3 ± − were derived from the data for gate and cathodeelectrodes respectively, which is consistent with each other. In addition, the single electron gain canalso be estimated based on the g ∼
90% [19]. The single electron gain will be 4.0 PE/e − considering g / e − from Sec. 3.1,which is consistent with the fitting results. The single electron rate in each µ s delay time after the main S2 is defined as the number of singleelectrons produced per primary S2 electrons after the electron lifetime correction. The number ofprimary S2 electrons is the electrons that have been extracted to the gas phase. The single electronrate as a function of its delay time is shown in figure 10 (left) for runs with different electron lifetime(or liquid purity).To understand the photoionization mechanisms from different materials, the single electronrate from gate, cathode and LXe bulk are investigated separately. For electrons generated on thegate and cathode electrodes, we used the average rate within 2 µ s delay time window around thepeak rate. For electrons generated on the impurities in the bulk LXe, we chose the rate within [10,26] µ s delay time window in the center of the TPC to avoid the potential field distortion in top andbottom TPC region. The corresponding delay time regions are the shaded-bands in figure 10 (left).The rates from each components are shown in figure 10 (right). The single electron rate from LXebulk decreases with the increasing of electron lifetime due to the purity improvement, while therate from the gate and cathode are less affected by the LXe impurities as expected. Delay time [ s] P r o d u c e d e r a t e [ / p r i m a r y S e / s ] : 41.0 ± 3.1 s : 91.0 ± 9.1 s : 162.7 ± 7.3 s : 151.9 ± 22.8 s : 187.4 ± 22.5 s 50 75 100 125 150 175 200 Electron lifetime [ s] P r o d u c e d e r a t e [ / p r i m a r y S e / s ] GateCathodeLXe bulk
Figure 10 . (Left) The single electron rate as a function of the delay time at different electron lifetime. seemore explain in text. (Right) The single electron rate from gate, cathode and LXe bulk region, as a functionof its electron lifetime. The rate is the average value from the corresponding shaded-band in the left plot.
Although the single electron rate has been corrected with electron lifetime, it still decreaseswith the delay time. This can be explained by the smaller number of S2 photons hitting the lowerpart of the TPC. A simple Solid Angle (SA) model can be used to model the photon emission fromthe location where S2 is produced. The S2 photons are mostly produced around anode wires due– 11 –o the strong local electric field, and emitted isotropically. In the SA model, we assume the S2 isproduced in the center of the anode electrode, thus the solid angle inside TPC gradually decreaseswith increasing of TPC depth. One example single electron rate as a function of its delay time isshown in figure 11 (left). The rate between the delay time of [10, 26] µ s was fit with the SA model.The fitting is slightly deviated from the data caused by the photon reflection on the thin PTFE sheeton TPC wall. This distribution indicates that the single electrons from the LXe bulk are most likelyassociated with the neutral impurities [18] which are more evenly distributed in LXe than negativeions caused by electrons attached to the impurities. The larger deviation from the SA model forevents near the gate electrode is most likely caused by the negative ions accumulation at the upperregion of the TPC, with a higher photoionization probability than the neutral impurities. Anotherpossible reason is the inaccurate drift time correction due to the field distortion as described inSec. 3.1. Delay time [ s] P r o d u c e d e r a t e [ / p r i m a r y S e / s ] ×10 = 91.0 ± 9.1 s Fitted with Solid Angle0 10 20 30 40012 ×10 impurity [ppb] P r o d u c e d e r a t e [ // m ] ×10 LXe bulk
Figure 11 . (Left) The single electron rate as a function of its delay time for data with 91.1 µ s electronlifetime, fit with the solid angle model in range of [10, 26] µ s delay time. The inset is a zooming of the fitregion. (Right) Photoionization rate as a function of the impurities in the LXe bulk. The impurity in ppb iscalculated from the electron lifetime (see text for details). The data-points is fit with a linear function. The photoionization probability can be estimated for different electrode materials and differentLXe impurities level as following. The photons that produced by one extracted electron in GXe iscalculated by the empirical equation: N γ = α ( ε e / p − β ) px . α is the amplification factor and β is thethreshold of reduced field for proportional light production. α = 120, and β = 1.3 kV/cm/atm [20]are used in our calculation. The electric field ε e of 9.6 kV/cm in S2 production region is from theCOMSOL simulation. In our measurement, the gas pressure ( p ) is controlled and kept at 2.2 atm.The gap ( x ) for proportional light production is 0.25 cm. In the end, 202 γ /e − of N γ can be derivedbased on the parameters above.Considering the solid angle, 92% gate transparency, and 96% transparency of the single-layergraphene cathode [11], the number of photons hitting the electrodes can be estimated. Basedon the number of electrons emitted from the electrodes, and photons received accordingly, thephotoionization probability of the electrodes can be obtained. The photoionization probability ofthe stainless steel gate is 7.7 × − e − / γ , which is consistent with the value of ∼ × − e − / γ in Ref. [21] for stainless steel. The photoionization probability of graphene is estimated to be– 12 –.0 × − e − / γ , which is slightly lower than the stainless steel. The result is reasonable consideringthe slightly higher work function of graphene (4.5 ∼ ∼ ∼ × − e − / γ / m . In our measurement, the photoionized electrons within [10, 26] µ s delaytime from each extracted electron can be calculated by integrating the histogram in figure 10 (left).The number of photons cross this LXe region can be estimated using the solid angle coverageand the gate transparency. Assume a 5 cm mean path length for photons within the [10, 26] µ s delay time window taking into account the reflection on TPC wall, the photoionization rate in LXewas calculated and shown in figure 11 (right) for different electron lifetimes. A linear fit resultsa value of 4.5 × − e − / γ / m for every ppb impurity concentration, which is compatible with thecalculation from the photoionization cross section on oxygen. The results is also close to theestimated (5-20) × − e − / γ / m from the LUX experiment [18]. In this work, we demonstrated a semi-sealed LXeTPC with a graphene-coated fused silica windowas the cathode electrode. The effectiveness of the LXe purification is studied using the electronlifetime evolution data with a simple purification model. It shows a much cleaner liquid xenon purityin LXe target inside the sealed chamber, compared to that outside, achieving an electron lifetime of0.5 ms, with uncertainties limited by the drift length of the TPC. We expect further improvementof purification efficiency with a fully-sealed TPC, which will be developed in the future.We investigated the single electron rate caused by the photoionization of scintillation photonson electrode materials and impurities in LXe. The graphene shows a little improvement, in termsof photoionization probability, compared to stainless steel. The photoionization rate of impuritiesin LXe shows that the prompt single electrons are caused mainly by the oxygen impurities in thebulk LXe.The TPC concept demonstrated here serves as a basis for further development of the sealedTPC technique aiming to significantly improve the purification efficiency of LXe in the target, andto reduce the single/few electrons background for future experiments to search for sub-GeV darkmatter via electron scattering [25], reactor neutrino detection via the CE ν NS process [26], and thenext generation multi-purpose liquid xenon experiment, e.g. DARWIN [27], for dark matter andneutrino physics.
Acknowledgments
This material is based upon work supported by the U.S. Department of Energy, Office of Science,Office of High Energy Physics under Award Number DE-SC0018952.– 13 – eferences [1] E. Aprile et al. Dark Matter Search Results from a One Ton-Year Exposure of XENON1T. Phys. Rev.Lett., 121(11):111302, 2018.[2] Xiangyi Cui et al. Dark Matter Results From 54-Ton-Day Exposure of PandaX-II Experiment. Phys.Rev. Lett., 119(18):181302, 2017.[3] D.S. Akerib et al. Results from a search for dark matter in the complete LUX exposure. Phys. Rev.Lett., 118(2):021303, 2017.[4] G. Anton et al. Search for Neutrinoless Double- β β electronic recoils in liquid xenon using LUX calibrationdata. JINST, 15(02):T02007, 2020.[17] Bakale U. G et al. Effect of an Electric Field on Electron Attachment to SF , N O , and 0 In LiquidArgon and Xenon. J. Phys. Chem., 80(23):2556–2559, 1976.[18] D.S. Akerib et al. Investigation of background electron emission in the LUX detector. arXiv:2004.07791, 4 2020.[19] Jingke Xu, Sergey Pereverzev, Brian Lenardo, James Kingston, Daniel Naim, Adam Bernstein,Kareem Kazkaz, and Mani Tripathi. Electron extraction efficiency study for dual-phase xenon darkmatter experiments. Phys. Rev. D, 99(10):103024, 2019.[20] E Aprile, K. L. Giboni, P. Majewski, Ni Kaixuan, and M Yamashita. Proportional light in a dualphase xenon chamber. IEEE Transactions on Nuclear Science, 51(5):1986–1990, 2004. – 14 –
21] J. Laulainen, T. Kalvas, H. Koivisto, J. Komppula, and O. Tarvainen. Photoelectron emission frommetal surfaces induced by vuv-emission of filament driven hydrogen arc discharge plasma. 2015.[22] Young-Jun Yu, Yue Zhao, Sunmin Ryu, Louis E. Brus, Kwang S. Kim, and Philip Kim. Tuning thegraphene work function by electric field effect. Nano Letters, 9(10):3430âĂŞ3434, Oct 2009.[23] R. G. Wilson. Vacuum Thermionic Work Functions of Polycrystalline Be, Ti, Cr, Fe, Ni, Cu, Pt, andType 304 Stainless Steel. Journal of Applied Physics, 37(6):2261–2267, May 1966.[24] C. E Brion, K. H. Tan, M. J Wiel, and P. E. Leeuw. Dipole oscillator strengths for thephotoabsorption, photoionization and fragmentation of molecular oxygen. J. Electron Spectrosc.Relat. Phenom., 17:101–119, 1979.[25] A. Bernstein et al. LBECA: A Low Background Electron Counting Apparatus for Sub-GeV DarkMatter Detection. J. Phys. Conf. Ser., 1468(1):012035, 2020.[26] D. Yu Akimov et al. First ground-level laboratory test of the two-phase xenon emission detectorRED-100. JINST, 15(02):P02020, 2020.[27] J. Aalbers et al. DARWIN: towards the ultimate dark matter detector. JCAP, 11:017, 2016.21] J. Laulainen, T. Kalvas, H. Koivisto, J. Komppula, and O. Tarvainen. Photoelectron emission frommetal surfaces induced by vuv-emission of filament driven hydrogen arc discharge plasma. 2015.[22] Young-Jun Yu, Yue Zhao, Sunmin Ryu, Louis E. Brus, Kwang S. Kim, and Philip Kim. Tuning thegraphene work function by electric field effect. Nano Letters, 9(10):3430âĂŞ3434, Oct 2009.[23] R. G. Wilson. Vacuum Thermionic Work Functions of Polycrystalline Be, Ti, Cr, Fe, Ni, Cu, Pt, andType 304 Stainless Steel. Journal of Applied Physics, 37(6):2261–2267, May 1966.[24] C. E Brion, K. H. Tan, M. J Wiel, and P. E. Leeuw. Dipole oscillator strengths for thephotoabsorption, photoionization and fragmentation of molecular oxygen. J. Electron Spectrosc.Relat. Phenom., 17:101–119, 1979.[25] A. Bernstein et al. LBECA: A Low Background Electron Counting Apparatus for Sub-GeV DarkMatter Detection. J. Phys. Conf. Ser., 1468(1):012035, 2020.[26] D. Yu Akimov et al. First ground-level laboratory test of the two-phase xenon emission detectorRED-100. JINST, 15(02):P02020, 2020.[27] J. Aalbers et al. DARWIN: towards the ultimate dark matter detector. JCAP, 11:017, 2016.