Single Electron Spectra in RPWELL-based detectors
Purba Bhattacharya, Andrea Tesi, Dan Shaked-Renous, Luca Moleri, Amos Breskin, Shikma Bressler
PPrepared for submission to JINST
Single-Electron Spectra in RPWELL-based detectors
Purba Bhattacharya, 𝑎, , Andrea Tesi, 𝑎 Dan Shaked-Renous, 𝑎 Luca Moleri, 𝑏 AmosBreskin, 𝑎 Shikma Bressler. 𝑎 𝑎 Department of Particle Physics and Astrophysics, Weizmann Institute of Science, 7610001 Rehovot, Israel 𝑏 Technion - Israel Institute of Technology, 3200003 Haifa, Israel
E-mail: [email protected]
Abstract: Single-electron spectra are the key ingredient in the efficient detection of single UV-photons. In this work, we investigated the shape of single-photoelectron spectra in single- anddouble-stage Resistive Plate WELL (RPWELL) detector configurations, operated in Ne / CH andAr / CH . Discharge-free operation was reached over a broad dynamic range, with charge gainsof 10 -10 . Compared to the usual exponential ones, the observed Polya-like charge spectra pavethe way towards higher single-electron detection efficiencies. The latter was evaluated here, usingexperimental data combined with numerical simulations. The effects of the gas mixtures, electricfield configuration, and detector geometry on the Polya spectra and their related “ 𝜃 ” parameter arepresented.Keywords: Charge transport and multiplication in gas; Electron multipliers (gas); Micropatterngaseous detectors (MSGC, GEM, THGEM, RETHGEM, MHSP, MICROPIC, MICROMEGAS,InGrid, etc), Photon Detector for UV, visible and IR photons (gas) (gas-photocathodes, solid-photocathodes) Corresponding author. Presently at Dept. of Physics, University of Calcutta. a r X i v : . [ phy s i c s . i n s - d e t ] F e b ontents 𝜃 dependence on THGEM voltage 95.3 Double-stage RPWELL with different gas mixtures 105.3.1 Effect of quencher concentration 105.3.2 Effects of the carrier gas 11 Gas-avalanche detectors, introduced originally for the needs of particle-physics experiments, havebecome a prominent subject of research also in a variety of other fields [1]. To this extent, significantefforts have been dedicated to the development of single-electron [2, 3] and single-photoelectronUV-photon gaseous detectors; the latter, mainly in the context of Ring Imaging Cherenkov Counters(RICH) [4, 5], with gaseous and solid photocathodes. Such gaseous photomultipliers (GPM) [6, 7]provide a cost-effective solution suitable for the coverage of large areas with good spatial andtemporal resolution, and low sensitivity to magnetic fields. GPM detectors have been developedalso for imaging scintillation and electroluminescence photons in noble gases [8] and liquids [9].In this context, Gaseous Photodetectors [6] employing Multiwire [10] and Drift [11] Cham-bers, Multi-step avalanche chambers [12], cascaded GEMs [13–15] and THGEM-based [16–19]detectors have been playing an important role in experiments. As demonstrated in [20], the morerecent “hybrid” CsI-coated THGEM-Micromegas has shown to be an efficient upgrade for theCOMPASS-RICH-I. Furthermore, THGEM-based detectors were demonstrated to have moderate(sub-millimeter) localization resolution [21] and about 10 nsec time resolution, which comply withthe requirements of most RICH devices. Though, previous studies with THGEM detectors haveshown high gas gains for single-photoelectron detection [17], the latter could be considerably lowerin the presence of intense particle background [22].In some applications, highly ionizing background environment can result in the formation oflarge avalanches, often leading to electric discharge. The latter can damage the readout electronics– 1 –nd the detector’s electrodes; it often introduces significant dead-time. The Thick Resistive WELL(RWELL) [22] and the Resistive Plate WELL (RPWELL) detectors [23] were introduced to preventoccasional discharges and mitigate their potential destructive effects. In the RPWELL (Fig. 1),the single-sided CsI-coated THGEM electrode is coupled to a segmented readout anode through athin plate of high bulk resistivity (10 –10 Ω cm). Ionization electrons induced by X-rays or UV-induced photoelectrons from a photocathode deposited on the top surface (e.g. CsI) are collectedinto the WELL holes where they undergo avalanche multiplication. Signals are induced capacitivelythrough the resistive plate onto a patterned readout anode, in direct contact with the resistive plate.The RPWELL and its properties have been studied extensively in the laboratory and with particlebeams [24–26]. The studies demonstrated a high particle detection efficiency, over a broad dynamicrange and at high particle-flux range, in a discharge-free operation mode at charge gains up to ∼ .Compared to single-stage configurations, higher maximal achievable gains and lower dischargeprobabilities can be achieved in a multi-stage one. In the present work, a double-sided THGEMfollowed by a RPWELL was investigated. This leads to higher detector gains at lower voltage biasper single THGEM and RPWELL element and thus to higher operation stability. (a)(b) Figure 1 . A schematic view of the (a) single-stage RPWELL detector and (b) double-stage RPWELL-baseddetector with a double-sided THGEM pre-amplification stage. – 2 –he single-photon detection efficiency (PDE) of a gaseous photon detector with a photocathodefollowed by an amplification element is provided by: 𝜖 eff ( 𝜆 ) = QE ( 𝜆 ) 𝜖 extr 𝜖 coll 𝜖 thresh (1.1)Here, 𝑄𝐸 ( 𝜆 ) is the wavelength-dependent quantum efficiency value in vacuum of the photo-cathode. 𝜖 𝑒𝑥𝑡𝑟 is the photoelectron extraction efficiency into the gas [27, 28], 𝜖 𝑐𝑜𝑙𝑙 is the efficiency totransfer the extracted photoelectron into the amplification region [17]. 𝜖 𝑡ℎ𝑟𝑒𝑠ℎ is the single-electrondetection efficiency - the probability that the collected photoelectron will generate a signal abovea given threshold. 𝜖 𝑡ℎ𝑟𝑒𝑠ℎ is strongly related to the shape of the electron charge spectrum whichreflects the physical processes governing the formation and development of the single-electronavalanche.In the present study, we investigated the UV-induced single-photoelectron charge spectragenerated in Ne / CH and Ar / CH gas mixtures with CsI-coated single RPWELL and double-stageCsI-coated THGEM followed by an RPWELL. In section 2 we elaborate on the theory associatedwith the formation of single-electron charge spectra. The experimental setup is detailed in section3, followed by the introduction of the numerical analysis framework in section 4. The results arepresented in section 5 and discussed in section 6. Electron avalanches in gas develop stochastically; the mean-free-path between successive interac-tions and, accordingly, the electron energy available, can vary considerably between successiveinteractions. This is dictated by the cross sections of the different electron interactions with gasmolecules (elastic and inelastic). As a result of this, the avalanche size (e.g. detector chargegain) fluctuations determine the shape of single-photoelectron charge spectra (see detailed study byAlkhazov [29]). Thus, these statistical fluctuations set a physical limit on the single-photoelectrondetection efficiency and localization resolution. Furthermore, large avalanche fluctuations increasethe probability for high-charge events, with a potential for occasional discharges.The shape of the single-electron charge distribution depends on the gas mixture, the reducedelectric field ( E / p ; p being the gas pressure and E the electric field), the electron initial momentumand the distance over which the avalanche develops. Gain fluctuations can be described quantita-tively in terms of the probability P n ( r , p ) that an electron with initial momentum p released at aposition r initiates an avalanche resulting in n electrons in the detector.According to [30, 31], under a moderate uniform field, an estimate of the single-electronavalanche distribution can be carried out with the assumption that the probability of ionization byan electron depends only on the electric field strength and is independent of its previous history.Yule-Furry statistics states that the probability P ( n , x ) of a single primary electron to produce anavalanche with n electrons, while propagating from the origin to a point x , follows an exponentiallaw: P ( n , x ) (cid:39) ( x ) e − n¯n ( x ) (2.1)– 3 –ere, ¯n is the mean number of avalanche electrons. We define f to be the relative variance ofthe mean value. For an exponential distribution, f ∼ and the photon detection efficiency fallsexponentially with increasing threshold.At higher field values, the avalanche size distribution was found to depart from the monoton-ically falling exponential; it exhibits a rounded peak. The probability of ionization by an electroncan no longer be considered totally independent of its history [30, 31]. The assumption that all theelectrons take part in the multiplication process with equal probability must be abandoned. Theionization mean-free-path becomes comparable to that for excitation and other inelastic processes.The charge distribution under these circumstances, introduced by Byrne, is known as Polya distri-bution [32, 33]. Its derivation assumes that the ionization probability per unit path-length dependson the current size of the avalanche through a dimensionless parameter 𝜃 . The probability of asingle primary electron to produce an avalanche with n electrons, while propagating from the originto a point x , follows a “peaked” Polya distribution [30–32]: P ( n , x ) = (cid:20) ( + 𝜃 ) n¯n ( x ) (cid:21) 𝜃 e (cid:104) ( + 𝜃 ) n¯n ( x ) (cid:105) (2.2)Relative to an exponential decay, detectors providing single-photoelectron Polya-like spectraare expected to have superior detection efficiency for a given electronics threshold. For each detectorconfiguration, the relative variance f and most probable avalanche size are given by [34, 35]: f = + 𝜃 (2.3)and n mp = ¯n 𝜃 + 𝜃 (2.4)respectively. Thus, the larger the 𝜃 -value, the smaller the variance and the closer the most probablecharge value is to the mean. Namely, for a given charge gain and electronics threshold the largerthe 𝜃 -value the better the single-electron detection efficiency is expected to be.The effect of inelastic and ionizing collision on the avalanche size distribution can be understoodin terms of a simple model discussed in [32]. During the electron drift, after each interaction, it eitherionizes or loses its kinetic energy by other, non-ionizing inelastic collisions. At electron energiesclose to the ionization threshold, the cross-section for ionization is still significantly smaller thanthe sum of the inelastic cross-sections (including excitations). With increasing energy, ionizationgradually becomes the dominant process; thus, with increasing the E / p - the relative frequencyof ionizing collisions is enhanced. The shape of the distribution in this case is determined by theionization yield Y which is defined as: Y = N ion N ion + N inel + N exc (2.5)Here, N ion , N exc and N inel are the number of ionizations, excitation, and other inelastic collisions,respectively.The relative variance f is, then, related to Y by: f ≈ − Y1 + Y (2.6)– 4 – Experimental setup
In the present work, a single and a double-stage RPWELL-based detector have been investigated.The setup consisted of UV- and X-ray sources, a vessel containing the detector elements and areadout system. The detector vessel was equipped with two windows - a 50 𝜇 m thick Kapton onefor the X-rays and a quartz one for the UV photons.Two radiation sources were used: a self-triggered homemade H discharge lamp emitting ∼
160 nm UV photons and a Fe source emitting 5 . -lamp pulse ratewas controlled by the voltage supplied to the lamp (here, 3 . ∼ / mm .The detector vessel was flushed with either Ne / CH (2%, 5%, 10% and 15% CH ) or withAr / to study the effect of gas mixtures and the quencher concentration on the single-electronspectrum.Details of the two, 30 ×
30 mm FR4-made detector configurations are depicted in Fig. 1.The single-stage detector is a 0 . ∼
300 nmthick) on its top THGEM-electrode surface; it is preceded by a 5 mm drift gap and a drift cathode.The double-stage detector has a CsI-coated 0 . . . .
96 mm and hole-rim 0 . . Ω cm). The resistive plate wascoupled to a Cu-coated readout anode using 3MTM Electrically Conductive Adhesive Transfer Tape9707.The transfer electric field 𝐸 𝑇 𝑟 𝑎𝑛𝑠 𝑓 𝑒𝑟 value in the double-structure, was set to of 0 . / cm;that of The drift field 𝐸 𝐷𝑟𝑖 𝑓 𝑡 , was set to 0 . / cm, to collect the ionization electrons induced byX-rays; for an efficient detection of UV-photons, 𝐸 𝐷𝑟𝑖 𝑓 𝑡 value was set to zero [37]. The electrodeswere polarized through low-pass filters (LPF), by CAEN N1471H HV power supplies. In allconfigurations, the anode was kept at ground potential and the signals were recorded through acharge-sensitive pre-amplifier (CAEN A1422). They were further processed through an Ortec 572A linear amplifier with 𝜇 sec shaping time. The acquisition was performed either by a TektronixMSO 5204B Mixed Signal Oscilloscope or by a multi-channel analyzer (MCA Amptek 8000A).In the current study, the measured vacuum QE-value of the CsI photocathode was ∼
15% at160 nm; no efforts were made to enhance its value. Its assembly in the detector vessel was performedunder controlled N atmosphere. Taking into account potential charging-up effects [38], some gain-stabilization was necessary priorto measurements. Therefore, single-photoelectron charge spectra were measured according to the– 5 –ollowing protocol: • Switching “off” the source and voltages; • Flushing the detector with the selected gas mixture at 50 SCCM for 2 hours; • Switching “on” the voltages and the source(s) and waiting for 30 minutes for the gain stabi-lization; • Data collection for 30 minutes; • Switching off the source and measuring the noise for 30 minutes;The signal spectra were obtained by subtracting the measured noise spectra from the measureddata. In our experimental configuration, the electronic noise was ∼ electrons.Each experimental spectrum was fitted to a generic Polya distribution (eqn. 2.2). The meangain value was extracted with its corresponding value for the Polya parameter 𝜃 . Experimentally,for each detector configuration, the relative single-photon detection efficiency was estimated (fora given electronic threshold) by dividing the number of counts at a given voltage for a particularconfiguration by the number of counts at the maximal voltage for the same configuration - normalizedto the number of lamp triggers, 150,000 per experiment. The statistical error in all the measurementsis at the level of 1%. The error bars are too small to be seen in the plots presented below.In addition, a set of Monte Carlo simulations was carried out using the same analytical function(eqn. 2.2) with experimentally obtained detector gain and 𝜃 values as input parameters. For 10,000events, a histogram was filled, representing the expected single-electron spectrum for given gainand 𝜃 values. The absolute single-electron detection efficiency was estimated by counting eventswith total number of electrons above a given threshold.It should be worth mentioning here that the reported detector gain values presented in this studycould vary due to charging-up effects in presence of X-rays [37]. By comparing the single-electronspectra before and after X-ray irradiation ( ∼
30 Hz / mm ) it was observed that the gain is reduced bya factor of ∼ . 𝜃 parameter varied by a factor ∼ .
2. A detailed study on the chargingup effects is beyond the scope of the present work; it will be discussed elsewhere.
Polya-like distributions were recorded with single photoelectrons, in the single and double-stageRPWELL-based detectors (shown on Fig. 1), with different gas mixtures and voltage configurations.The effects of the latter on the Polya parameters 𝜃 and f and on the single-electron detection efficiencyare discussed. UV-induced single-electron spectra were measured with the double-stage and single-stage RPWELLin Ne / at different voltage settings. The charge spectra, at a gain of ∼ . × , are shown inFig. 2(a). In the double-stage structure, the voltage difference across the THGEM ( Δ 𝑉 𝑇 𝐻 𝐺𝐸 𝑀 ) waskept constant at 700 V; it was set to maximize the extraction efficiency of the photoelectrons and– 6 – a) (b)(c) (d)
Figure 2 . (a) Single-electron spectra recorded in the single and double-stage RPWELL detectors (of Fig. 1),at similar charge gains of . × . (b) Variation of gain vs Δ 𝑉 𝑅𝑃𝑊 𝐸𝐿𝐿 , (c) 𝜃 and (d) relative gain variance f as a function of gain in Ne / , in the single and double-stage RPWELL-based detectors. For double-stage configuration Δ 𝑉 𝑇 𝐻𝐺𝐸 𝑀 =
700 V, 𝐸 𝑇 𝑟 𝑎𝑛𝑠 𝑓 𝑒𝑟 = . / cm. 𝐸 𝐷𝑟𝑖 𝑓 𝑡 = / cm. The spectra in(a) are normalized for equal number of counts. their collection into the holes [38]. The RPWELL voltage ( Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 ), in the double-stage, was setto 575 V to get similar total-gain value to that of the single-stage detector (at Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 =
950 V).The gain variation with Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 is shown in Fig. 2(b) for the two detector configurations.Each spectrum was fitted with eqn. 2.2 to yield the values of the gain and 𝜃 . The double-stageRPWELL reached a maximal gain with UV-photons of ∼ . × ( ∼ × with X-rays); in thesingle-stage detector the gain was limited to ∼ . × - both under stable discharge-free operationconditions.As expected, the 𝜃 -value increases with the gain (Fig. 2(c)). The relation between 𝜃 and thegain, measured with the two detectors, exhibits the same linear trend. The comparison with thedouble-stage detector shows that the 𝜃 value is lower than that of the single-stage RPWELL, forequal gain. It is due to the charge development and “saturation” in a single hole. However, as thedouble-stage detector reaches higher gain values, the maximum achievable 𝜃 parameter is higher,reaching values ∼ . f ” is plotted as a function of the gain in Fig. 2(d). As discussed– 7 –arlier, with the increasing gain value (thus, increasing 𝜃 parameter), “ f ” decreases. For a single-stage detector, at a fixed amplification gap, increasing the electric field ( Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 ) leads toan increase of the relative frequency of the ionizing collisions; therefore, according to eqn. 2.5,resulting in a reduction of the relative variance [32].It was also found that for the same gain value, the relative gain variance is higher for adouble-stage configuration than for the single-stage one. This is expected, since in a double-stageconfiguration, the electric field per stage is lower than in a single-stage one, resulting in a decreaseof the ionization probability. Also, the loss of electrons while transferred from the pre-amplificationstage and the RPWELL amplification one gives rise to the additional avalanche-size fluctuations.It should be mentioned here that, for the double-stage configuration under X-ray irradiation, themaximum applied Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 was limited to 650 V, with a Δ 𝑉 𝑇 𝐻 𝐺𝐸 𝑀 =
700 V. At these voltageconfigurations, the maximal achievable gain was ∼ × , with a corresponding 𝜃 parameter forsingle electrons of ∼ .
38 (Fig. 2(c)). (a) (b)
Figure 3 . Double-stage RPWELL detector: (a) Calculated single-electron efficiency as a function of 𝜃 , for dif-ferent avalanche-electron thresholds in Ne / . (b) Comparison of the relative single-electron detectionefficiency between experimental and simulation results, as function of Δ 𝑉 𝑅𝑃𝑊 𝐸𝐿𝐿 . Δ 𝑉 𝑇 𝐻𝐺𝐸 𝑀 =
700 V, 𝐸 𝑇 𝑟 𝑎𝑛𝑠 𝑓 𝑒𝑟 = . / cm. 𝐸 𝐷𝑟𝑖 𝑓 𝑡 = / cm. In both plots, the statistical error bars are too small to beseen. Numerically-estimated single-electron detection efficiency values are plotted in Fig. 3(a) asa function of 𝜃 for different avalanche-electrons thresholds. Note that even at low 𝜃 values, e.g. ∼ .
25, a high efficiency ( > electrons or less. Thecomparison of the relative efficiency between the experimental estimate and the numerical one ispresented in Fig. 3(b). The good agreement validates the numerical analysis method.A comparison of numerically-estimated efficiencies for different gain values in the single anddouble-stage detectors is given in Table 5.1. The avalanche-electron threshold was fixed to 10 electrons. The efficiency of a detector having an exponential distribution, for the same gain values,was estimated for comparison. It was found that for the same gain value, the single-stage detectorprovides better efficiency than that of the double-stage detector due to the higher 𝜃 -value. However,with the increase of the gain, the difference reduces. Note that the efficiency value for an exponentialdistribution at the gain of 2 . × is a hypothetical one. It was reached with the double-stage– 8 –ain Exponential PolyaEfficiency Detector 𝜃 Efficiency Comments6 . × .
3% Single-Stage 0.34 86 .
3% Stable under UVDouble-Stage 0.20 86 .
3% and X-ray2 . × .
9% Single-Stage 0.42 98 .
6% Stable under UVDouble-Stage 0.30 97 .
4% Stable under UV and X-ray2 . × .
5% Double-Stage 0.59 99 .
9% Stable under UV
Table 1 . Comparison of efficiency in single and double-stage RPWELL-based detectors in Ne / . Theelectron threshold is 10 electrons. RPWELL, due to its “charge-quenching” property; it would be hard to reach in other detectorconfigurations. 𝜃 dependence on THGEM voltage We studied the effect of Δ 𝑉 𝑇 𝐻 𝐺𝐸 𝑀 on the gain and on the 𝜃 parameter for a fixed Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 . Thedrift and transfer fields were kept at 0 and 0 . / cm, respectively. (a) (b) Figure 4 . A double-RPWELL configuration in Ne / . (a) gain variation as a function of Δ 𝑉 𝑅𝑃𝑊 𝐸𝐿𝐿 (with Δ 𝑉 𝑇 𝐻𝐺𝐸 𝑀 =
700 V) and of Δ 𝑉 𝑇 𝐻𝐺𝐸 𝑀 (with Δ 𝑉 𝑅𝑃𝑊 𝐸𝐿𝐿 =
750 V and 500 𝑉 ); (b) corresponding 𝜃 variation as a function of the total gain. 𝐸 𝑇 𝑟 𝑎𝑛𝑠 𝑓 𝑒𝑟 = . / cm and 𝐸 𝐷𝑟𝑖 𝑓 𝑡 = / cm. Fig. 4(a) shows the gain dependency on Δ 𝑉 𝑇 𝐻 𝐺𝐸 𝑀 for two different values of Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 andwith Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 for a fixed Δ 𝑉 𝑇 𝐻 𝐺𝐸 𝑀 . Δ 𝑉 𝑇 𝐻 𝐺𝐸 𝑀 was limited to 800 V before the occurrenceof occasional discharges, resulting in a maximum gain of ∼ . × for Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 =
500 V.For the two other configurations, with higher applied Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 -values, gains up to ∼ . × were reached under stable conditions. The dependency of the 𝜃 parameter on the gain (Fig. 4(b)) isroughly the same for the three configurations. – 9 – a) (b)(c) (d) Figure 5 . Double-RPWELL configuration in Ne with CH -quencher concentrations of 2%, 5%, 10%and 15%; (a) single-electron spectra for a total-gain value for each ∼ . − . × - reached by ad-justing Δ 𝑉 𝑅𝑃𝑊 𝐸𝐿𝐿 and Δ 𝑉 𝑇 𝐻𝐺𝐸 𝑀 values; (b) gain variation vs Δ 𝑉 𝑅𝑃𝑊 𝐸𝐿𝐿 ; Δ 𝑉 𝑇 𝐻𝐺𝐸 𝑀 adjusted tomaximum value below discharge onset; (c) corresponding 𝜃 and (d) relative gain variance f vs total gain. 𝐸 𝑇 𝑟 𝑎𝑛𝑠 𝑓 𝑒𝑟 = . / cm and 𝐸 𝐷𝑟𝑖 𝑓 𝑡 = / cm. The dependence of single-electron spectral shape and, consequently, of the 𝜃 parameter, on thequencher concentration has been studied in Ne / CH mixtures, having 2, 5, 10 and 15% CH concentrations. Fig. 5(a) depicts the single-electron spectra for equal gain of ∼ . × for thefour-quencher concentrations. One can notice the more pronounced Polya peak at the lower CH concentrations.Fig. 5(b) shows the trends of the gain with respect to Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 for the four differentconcentrations. In all measurements, Δ 𝑉 𝑇 𝐻 𝐺𝐸 𝑀 was set to the highest value allowing for stable,discharge-free, operation to maximize the electron extraction efficiency from the photocathode.The dependence of 𝜃 parameter and relative variance f -parameter on the gain, at different quencherconcentrations, are shown in Fig. 5(c) and Fig. 5(d), respectively. In presence of the quencher, theratio of ionizing versus other inelastic collisions is shifted in favor of the other inelastic scattering– 10 –ain Exponential PolyaEfficiency CH Concentration 𝜃 Efficiency Comments6 . × .
9% 2% 0.28 89 .
5% Stable operation5% 0.19 88 .
2% of double-stage10% 0.16 87 .
9% under UV and15% 0.02 85 .
5% of X-ray4 . ×
98% 2% 0.45 99% Stable operation5% 0.41 99% of double-stage10% 0.25 99% under UV15% 0.17 99%
Table 2 . Comparison of efficiency as a function of quencher amount in double-stage RPWELL-baseddetectors in Ne / CH . The electron threshold is 10 electrons. [32]. Therefore, the avalanche size distribution broadens up (according to eqn. 2.5 and 2.6) withthe increase of quencher, thus, the relative variance “ f ” increases. This in turn, results in higher 𝜃 parameter values for lower quencher concentrations - as also depicted in Fig. 5(c).The single-electron detection efficiency for two different values of gain in the four-quencherconcentration is given in Table 5.3.1. With an avalanche-electron threshold of 10 electrons, for thelower gain value of ∼ × , the efficiency in Ne / is ∼
4% higher than that of Ne / .This rather small difference is due to the higher 𝜃 value in lower quencher concentrations. However,with the increase of gain, the difference between the measured 𝜃 values decreases and the detectorreaches full efficiency for quencher concentrations of 2 − The dependence of the 𝜃 parameter on the carrier gas, for a fixed CH quencher concentration werestudied. Fig. 6(a) shows single-electron spectra in Ne / and Ar / , for similar gainvalues of ∼ , adjusted by setting different Δ 𝑉 𝑇 𝐻 𝐺𝐸 𝑀 values. The gain curves as function of Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 are shown in Fig. 6(b). The operation in Ar / required higher Δ 𝑉 𝑇 𝐻 𝐺𝐸 𝑀 and Δ 𝑉 𝑅𝑃𝑊 𝐸 𝐿𝐿 values. The maximum achievable gain is higher in the Ne-based mixture, by at least anorder of magnitude, as previously observed in [21, 38]. The dependence of 𝜃 and f on the gain inthese two gas mixtures are shown in Fig. 6(c) and Fig. 6(d), respectively. In Ne / , the f valueis slightly lower than in Ar / ; it agrees with measurements in Micromegas detectors [32, 34].In Ar-based mixtures, due to the lower threshold for excitation and larger inelastic cross-section,the ionization yield is lower, thus enhancing avalanche fluctuations.The numerically estimated single-electron detection efficiency for two different values of gainin Ne / and Ar / is given in Table 5.3.2. For the lower gain value of ∼ × , theefficiency in Ne / is slightly higher than in Ar / due to the higher 𝜃 value. However,with the increase of gain, the difference between the efficiencies is smaller.– 11 – a) (b)(c) (d) Figure 6 . A double-RPWELL configuration in Ne / and Ar / : (a) single-electron spectraat equal total charge gain of ∼ , adjusted by the Δ 𝑉 𝑅𝑃𝑊 𝐸𝐿𝐿 and Δ 𝑉 𝑇 𝐻𝐺𝐸 𝑀 values. (b) gain varia-tion vs Δ 𝑉 𝑅𝑃𝑊 𝐸𝐿𝐿 , (c) corresponding 𝜃 and (d) relative gain variance f . 𝐸 𝑇 𝑟 𝑎𝑛𝑠 𝑓 𝑒𝑟 = . / cm and 𝐸 𝐷𝑟𝑖 𝑓 𝑡 = / cm. Gain Exponential PolyaEfficiency Gas 𝜃 Efficiency Comments4 . ×
78% Ne / / . ×
96% Ne / / Table 3 . Comparison of efficiency as a function of carrier gas in double-stage RPWELL-based detectors.The electron threshold is 10 electrons. In this work, we investigated single-electron spectra obtained with single-stage and double-stageRPWELL-based detectors. The goal was to evaluate the potential advantage of their operation inconditions yielding Polya-like single distributions; the latter result from charge-avalanche saturationin the RPWELL holes. The detectors’ response was single-photoelectrons emitted from a CsI– 12 –hotocathode deposited on the multiplier’s surface. Their performance was studied in various gasmixtures and electric-field settings.Operating in Ne / , the single-stage RPWELL detector reached gains of ∼ . × ina discharge-free mode; the single-electron Polya-like distribution had a 𝜃 -value of ∼ .
36. Thedouble-stage THGEM/RPWELL detector reached gains of ∼ . × in stable conditions; thePolya-like distribution reached a 𝜃 -value of ∼ . 𝜃 and f as a function ofgain in both RPWELL detector configurations. Note, however, that for equal gain, the single-stagedetector yielded more pronounced Polya-like distributions, with higher 𝜃 -value and thus narrowerrelative gain variance f . This is expected since in a double-stage configuration, the electric fieldper stage is lower than in a single-stage one, resulting in a decrease of the ionization probability(gain fluctuations) with respect to other inelastic processes. In addition, avalanche-electron lossesbetween the two stages result in additional avalanche-size fluctuations.The experimental values of 𝜃 and gain, followed by numerical calculations, yielded the single-electron detection-efficiency values; they are presented as function of the threshold (number ofavalanche-electrons). For a given threshold of 10 electrons, the numerical estimation suggests thatfor a same gain of ∼ × , the efficiency in the single-stage detector ( ∼ ∼ . 𝜃 value. Also, for the single-stage detector, relativeto an exponential distribution at the same gain, the expected single-electron detection efficiency isabout ∼
6% higher (91% relative to 86%). However, with the increase of the gain, for the samethreshold, both detectors attain similar efficiency values. The double-stage detector yielded anefficiency of ∼
97% at a gain of ∼ × , with 𝜃 = .
3. In the presence of 5.9 keV x-ray, themaximal achievable stable gain is limited to ∼ × , with a 𝜃 value of ∼ .
38 and 99% detectionefficiency.The amount of quencher added to the carrier gas has a major role in reducing instabilities,which are mainly due to photon-induced secondary avalanches. It also affects the electron-transportparameters and enhances the extraction efficiency of photoelectrons from the photocathode in asingle-photon detector (the matter is out of the scope of this work). We studied the effect of the CH quencher concentration on the detector properties, gain, 𝜃 and the relative gain variance f . It wasobserved that with increasing quencher concentration, 𝜃 -value decreased, and f increased - for agiven gain value. With an electron threshold of 10 , for a gain of 6 × , the 2% CH concentrationyielded higher efficiency (89 . (85 .
5% efficiency) due to a lower 𝜃 valuein the latter case. But, with the increase of gain, the quencher concentration did not affect theefficiency significantly, for the same threshold. Relative to an exponential distribution at a similargain of ∼ × , the single-electron efficiency improved by ∼
5% for 2% quencher concentrationand by ∼
1% for 15% quencher concentration.The dependence of 𝜃 and f on the carrier gas was evaluated for a given CH quencher concen-tration. Significant avalanche fluctuations were observed in Ar / ; it is due to low ionizationyield resulting in higher relative gain variance. Ne / , with lower avalanche fluctuations,showed a higher breakdown limit - thus a lower discharge probability. Therefore, Ne / CH provedto be superior to Ar / CH in detecting single-photoelectrons. The numerical study suggested thatdepending on the electron threshold, at lower gains, the detector operation in Ne / CH yields higherefficiency than in Ar / CH ; with increasing gain, both mixtures reach similar efficiency values,– 13 –.g. ∼
98% at 2 . × . Relative to an exponential distribution at lower gain, the single-electronefficiency improved by ∼
1% for Ar / and by ∼
4% for Ne / . Acknowledgments
This research was supported in part by the Nella and Leon Benoziyo Center for High Energy Physicsat the Weizmann Institute of Science and Grant No 713563 from the Israeli Science Foundation(ISF). This work is supported by Sir Charles Clore Prize. Special thanks to Martin Kushner Schnurfor supporting this research.
References [1] F. Sauli,
Gaseous Radiation Detectors: Fundamentals and Applications , Camb. Monogr. Part. Phys.Nucl. Phys. Cosmol. vol. 36 (2013) pp 1.[2] G. Malamud, A. Breskin and R. Chechik,
A Study of Relativistic Charged Particle Identification byPrimary Cluster Counting , Nucl. Instrum. Method A vol. 372 (1996) pp 19.[3] A. Pansky, A. Breskin, R. Chechik and G. Malamud,
Detection of X-ray-Fluroscence ofLight-Elements by Electron Counting in A Low-Pressure Gaseous Electron Multiplier , Nucl. Instrum.Method A vol. 330 (1993) pp 150.[4] J. Senguinot and T. Ypsilantis,
A Historical Survey of Ring Imaging Cherenkov Counters , Nucl.Instrum. Method A vol. 343 (1994) pp 1 and references therein.[5] F. Tessarotto,
Evolution an Recent Developments of the Gaseous Photon Detectors Technologies , Nucl. Instrum. Method A vol. 912 (2018) pp 278 and references therein.[6] R. Chechik and A. Breskin,
Advances in Gaseous Photomultipliers , Nucl. Instrum. Method A vol. 595 (2008) pp 116.[7] A. Di Mauro,
Status and Perspectives of Gaseous Photon Detectors , Nucl. Instrum. Method A vol. 766 (2014) pp 126.[8] J. Nickles et al.,
A Gas Scintillation Counter With Imaging Optics and Large Area UV-Detector , Nucl.Instrum. Method A vol. 477 (2002) pp 59.[9] L. Arazi et al.,
First Results of A Large-Area Cryogenic Gaseous Photomultiplier Coupled to ADual-Phase Liquid Xenon TPC , J. Instrum. , vol. 10 (2015) pg P10020.[10] Piuz et al., Final Tests of the CsI-based Ring Imaging Detector For the ALICE Experiment , Nucl.Instrum. Method A vol. 433 (1999) pp 178.[11] E. Albrecht et al.,
Operation, Optimisation, and Performance of the DELPHI RICH detectors , Nucl.Instrum. Method A vol. 433 (1999) pp. 178.[12] R. Baur et al.,
The CERES RICH Detector System , Nucl. Instrum. Method A vol. 343 (1994) pp. 87.[13] D. Mormann et al.,
Operation Principles and Properties of the Multi-GEM Gaseous PhotomultiplierWith Reflective Photocathode , Nucl. Instrum. Method A vol. 530 (2004) pp. 258.[14] T. Meinschad et al.,
GEM-based Photon Detectors for RICH Applications , Nucl. Instrum. Method A vol. 535 (2004) pp. 324.[15] W. Anderson et al.,
Design, Construction, Operation and Performance of A Hadron Blind Detectorfor the PHENIX Experiment , Nucl. Instrum. Method A vol. 646 (2011) pp. 35. – 14 –
16] R. Chechik, A. Breskin and C. Shalem,
Thick GEM-like Multipliers – A Simple Solution for LargeArea UV-RICH Detectors , Nucl. Instrum. Method A vol. 553 (2005) pp. 35.[17] C.D.R. Azevedo, M. Cortesi, A. V. Lyashenko, A. Breskin, R. Chechik, J. Miyamoto, V. Peskiv, J.Escada, J. F. C. A. Veloso, J. M. F. dos Santos,
Towards THGEM UV-Photon Detectors for RICH: onsingle-Photon Detection Efficiency in Ne / CH and Ne / CF , J. Instrum. , vol. 5 (2010) pg P01002.[18] V. Peskov, M. Cortesi, R. Chechik and A. Breskin, Further Evaluation of a THGEM UV-photonDetector for RICH – Comparison with MWPC , J. Instrum. , vol. 5 (2010) pg P11004.[19] M. Alexeev et al., THGEM-based Photon Detectors for the Upgrade of COMPASS RICH-I , Nucl.Instrum. Method A vol. 732 (2013) pp. 732.[20] J. Agarwala et al.,
Novel MPGD-based Detectors of Single Photons in COMPASS RICH-I ; Nucl.Instrum. Method A vol. 912 (2018) pp. 158.[21] A. Breskin et al.,
A Concise Review on THGEM Detectors , Nucl. Instrum. Method A vol. 598 (2009)pp. 107.[22] S. Bressler, L. Arazi, L. Moleri, M. Pitt, A. Rubin and A. Breskin,
Recent Advances with THGEMDetectors , J. Instrum. , vol. 8 (2013) pp. C12012.[23] A. Rubin et al., First Studies with the Resistive-Plate WELL Gaseous Multiplier ; J. Instrum. vol. 8 (2013) pp. P11004.[24] S. Bressler et al.,
First In-Beam Studies of a Resistive-Plate WELL Gaseous Multipliers ; J. Instrum. vol. 11 (2016) pp. P01005.[25] L. Moleri et al.,
In-Beam Evaluation of a Medium-Size Resistive-Plate WELL Gaseous ParticleDetector , J. Instrum. vol. 11 (2016) pp. P09013.[26] S. Bressler et al.,
Novel Resistive-Plate WELL Sampling Element for (S)DHCAL ; Nucl. Instrum.Method A vol. 958 (2020) pp. 162861.[27] A. Breskin,
CsI UV Photocathodes: History and Mystery , Nucl. Instrum Method A vol. 371 (1996)pp. 116.[28] J. Escada et al.,
Measurements of Photoelectron Extraction Efficiency from CsI into Mixtures of Newith CH , CF , CO and N ; J. Instrum. vol. 4 (2009) pp. P11025.[29] G.D.Alkhazov,
Statistics of Electron Avalanches and Ultimate Resolution of Proportional Counters , Nucl. Instrum. Method A vol. 89 (1970) pp. 155.[30] W. Blum, W. Riegler, L. Rolandi,
Particle Detection with Drift Chambers; Second edition,Springer-Verlag Berlin Heidelberg (2008).[31] G. F. Knoll,
Radiation Detection and Measurement; Third edition, NewYork Wiley (2000).[32] Heinrich Schindler,
Microscopic Simulation of Particle Detectors , CERN-THESIS-2012-208 (2012).[33] J. Byrne,
Statistics of Electron Avalanches in the Proportional Counter , Nucl. Instrum. Method vol. 74 (1969) pp. 291.[34] T. Zerguerras et al.,
Understanding Avalanches in a Micromegas from Single-Electron ResponseMeasurement , Nucl. Instrum. Method A vol. 772 (2015) pp. 76.[35] J. Derre et al.,
Fast Signals and Single Electron Detection with a MICROMEGAS Photodetector , Nucl.Instrum. Method A vol. 449 – 15 –
37] M. Cortesi et al.,
THGEM operation in Ne and Ne / CH , J. Instrum. vol. 4 (2009) pp. P08001.[38] D. Shaked Renous et al.; Gain Stabilization in Micro Pattern Gaseous Detectors: Methodology andResults,
J. Instrum. vol. 12 (2017) pp. P09036.(2017) pp. P09036.