Demonstration of ThGEM-Multiwire Hybrid Charge Readout for Directional Dark Matter Searches

Sensitivities of current directional dark matter search detectors using gas time projection chambers are now constrained by target mass. A ton-scale gas TPC detector will require large charge readout areas. We present a first demonstration of a novel ThGEM-Multiwire hybrid charge readout technology which combines the robust nature and high gas gain of Thick Gaseous Electron Multipliers with lower capacitive noise of a one-plane multiwire charge readout in SF$_6$ target gas. Measurements performed with this hybrid detector show an ion drift velocity of $139~\pm~12~\text{ms}^{-1}$ in a reduced drift field $\text{E/N}$ of $93~\text{Td}~(10^{-17}~\text{V cm}^{2})$ at a gas gain of $2470\pm160$ in 20 Torr of pure SF$_\text{6}$ target gas.


Introduction
Detection and characterisation of dark matter (DM) -thought to be Weakly Interacting Massive Particles (WIMPs) [1,2,3] in a direction sensitive nuclear recoil detector with a suitable target material, is a major goal of the DM search community [4,5,6,7]. This technology offers the potential to discriminate WIMP candidate events with galactic signature from terrestrial backgrounds/artefacts and hence, can probe below the so-called neutrino floor [8,9,10]. The use of low pressure gas Time Projection Chamber (TPC) technology, in which ionisation electrons from the nuclear recoil tracks are drifted to a charge readout plane and recorded for reconstruction, offers a route to achieving this goal. This is with potentials for low energy threshold and low background operations, including active electron recoil discrimination in the low WIMP mass parameter space.
The CYGNUS consortium is extensively exploring the feasibility of this technology for a large-scale experiment with aim to search for WIMPs beyond the so-called neutrino floor [4,11]. This builds on previous R&D and DM search results by multiple directional efforts, including DRIFT [12] , NEWAGE [13], MIMAC [14], D 3 [15], DM-TPC [16] and CYGNO [17] collaborations. A feature of interest in DRIFT, for instance, is the use of CS 2 gas for primary ionization charge transport through negative ions (NI) drift, rather than drifting electrons for minimal and thermal scale diffusion [7,18]. Primary ionization electrons from interactions in the TPC attach rapidly to the electronegative CS 2 to form anions. These anions are drifted towards the readout plane where they are field ionized by the inhomogeneous high electric field in this regionthereby inducing signal amplification by electron avalanche [18]. The use of NI drift substantially reduces blurring of the tracks by diffusion [18,19,20], and hence saves cost by allowing the possibility for longer drift distances relative to the conventional electron drift concepts.
Recently, it has been discovered that SF 6 [21,22], which has lower toxicity with improved handling over CS 2 [23], can also serve as a negative ion TPC gas. This is with a further advantage of formation of a minority charge carrier specie SF − 5 , in addition to the main SF − 6 charge carrier specie [24]. Measurement of the arrival time difference between these charge species at the readout plane, allows for identification of the absolute perpendicular distance between an event interaction vertex and the charge readout plane. This characteristic is vital for full rejection of background events emanating from the surfaces of the detector materials. Such event fiducialisation power has been demonstrated using a controlled admixture of O 2 gas in a CS 2 :CF 4 based target gas [7,25].
The higher 19 F content in SF 6 (relative to CF 4 ) offers a further advantage for improved WIMP-nucleon spin-dependent sensitivity [26,27]. Studies indicate that stronger avalanche fields are required near the readout planes to achieve field ionization of SF 6 anions for electron avalanche due to the higher electron affinity of SF 6 relative to the CS 2 gas [24]. These strong avalanche field are outside the operational range of more fragile electrode configurations in the conventional multiwire proportional counter (MWPC) geometry as used in DRIFT [28,29,30]. However, Thick Gaseous Electron Multipliers (ThGEMs) [31,32] have been demonstrated to produce gains of order 10 3 in SF 6 gas [24]. Initial results show that a gain of 10 4 can be achieved with a triple thin GEM setup [33].
The combination of the high gas gain from ThGEMs with low capacitance of the multiwires offers a route to achieving lower operational threshold with a 3-d reconstruction ability. Hence, the possible signal-to-noise ratio that can be achieved in operations with non-hybrid ThGEM or MWPC based TPC technologies with SF 6 target gas can be surpassed.
In this work, we present for the first time, a demonstration of a ThGEM-Multiwire hybrid charge readout technology as a possible candidate for next generation large area, low threshold TPC-based directional dark matter detectors. In this hybrid configuration, the field ionization of the anions occur in the ThGEM while induced charge signals from the avalanche electrons are read out using wires coupled at a mm-scale distance above the ThGEM.

Design and Construction of the ThGEM-Multiwire Hybrid Detector
The ThGEM-Multiwire hybrid detector technology combines the robust nature and high gas gain of ThGEM readouts with low capacitive noise and the ability to achieve better event track granularity with multiwires. The hybrid detector used in this work was made from a circular, 1 mm thick GEM (sourced from CERN) of 5 cm fiducial diameter coupled to a 2 cm × 2 cm, one-plane multiwire readout [34,35]. An illustration of the detector configuration with typical operational voltages and a pictorial view of the detector is shown in Figure 1.
The hole diameter and pitch of the ThGEM was 0.56 mm and 0.8 mm, respectively as shown in Figure 1(a). Either side of the holes were enclosed by an additional 0.04 mm rim, etched on the copper-clads to prevent electrostatic hole-edge discharges and ensure that electric field lines are centred on the ThGEM holes for optimal ion collection and field ionization for the avalanche process. The rim size can affect the performance of a ThGEM based detector. For instance, increasing the rim size of the ThGEM from 0.04 mm to 0.09 mm in a Garfield simulation [36], resulted in 86% loss of initial electrons into the copper cladding and dielectric FR4 material. It is important to point out that this is one of the early set of ThGEMs produced by CERN, so it does not represent an optimal design. The one-plane Multiwire readout was made using 100 µm diameter stainless steel wires, placed on a custom made printed circuit board at 1 mm pitch. This was then mounted on the induction side of the ThGEM at a ThGEM-Multiwire separation of 1 mm.
A field cage was designed and constructed to maintain a uniform drift field [37,38] within the 2 cm × 2 cm × 5 cm detector volume as shown in Figure 1   was achieved by stepping down the high-voltage applied to the copper-plate cathode through a series of five, 33 MΩ resistors connected to four series of copper field rings. To complete the circuit, the last field cage ring was connected to ground through the fifth successive resistor. The detector was then built by mounting the ThGEM-Multiwire readout on the field-cage with the ThGEM (charge transfer) side of the readout facing the drift volume as shown in Figure 1.
The full data flow path from the read-out wire to storage disk is shown in Figure  2. During an operation, the charge transfer side of the ThGEM was biased to ensure that the drift field is maintained before the drifting anions are field-ionized for electron avalanche. The avalanche process and signal multiplication was induced by setting the opposite (induction) side of the ThGEM to a sufficient and more positive potential. By biasing the wire potential to 0, avalanche electrons induce equivalent current [39,40] on the wires as they follow the field lines to the induction side of the ThGEM as shown in Figure 3. This induced current i can be defined as: where E A is the avalanche field, v A is the avalanche electron velocity and e is the electronic charge. The induced charge Q over a given time t can then be determined using: These induced charge signals were used as avalanche electrons can reattach to SF 6 to form anions while drifting from the induction side of the ThGEM to the charge collection wires. Garfield simulations were used to determine the optimal operational voltage configurations for the detector. For instance, typical operational drift field range, induction and charge transfer voltages of the ThGEM and wire bias voltage of −600 V cm −1 to − 700 V cm −1 , +600 V, −450 V and 0, respectively was used. Electric field lines from this voltage configuration end on the induction side of the ThGEM as seen in Figure  3(b). The induced current on the wires were successively amplified using Cremat CR-111 pre-amplifiers and CR-200-4µs gaussian shaping amplifiers. As shown in Figure  2, a pair of grounded inverse parallel FDH-300 diodes were added between the preamplifiers and the wires to protect the amplifiers from power surge. Amplified signals were digitised using an NI 5751 digitizer controlled through a NI PXI-7953R FlexRIO FPGA and saved to disk for analyses. Due to the capacity of the digitizer, only 16 wire channels were instrumented. An FPGA [41] and LabVIEW [42] based data acquisition system (DAQ) was developed for online data quality monitoring and run control. A pictorial view of the experimental setup stand is depicted in Figure 4. The white dual channeled ISEG NHQ 238L NIM cassette high voltage power supply shown in   Figure 4 were used to bias the Cremat amplifiers which are located inside the vacuum vessel to minimise noise distortions. A Leybold Ceravac CTR-101 pressure gauge was used to monitor the pressure of the 96 l vacuum vessel.

Detector Commissioning and Performance
Gain measurements were done to test the detector performance using X-rays from the electron capture decay of 55 Fe source to 55 Mn. As shown in Figure 1(a) the source was positioned close to the copper-plate cathode to irradiate the detector fiducial volume. To achieve this, the source was bonded to the tip of a 5 cm M6 nylon studding glued to a Neodymium disc magnet and attached to the inside wall of the vessel. This was magnetically coupled to a second magnet on the outside vessel wall which was used to control the source position in the vessel. Ionization electrons from X-ray interactions with the target gas through photoelectric effect -attach to the electronegative SF 6 to form anions. As described in Sections 1 and 2, these anions drift in a uniform field to the ThGEM for field ionization and electron avalanche. Signal pulses induced on the wires from this process were amplified and recorded to disk at a frequency of 1 MHz per channel, without any hardware trigger as the pulses were small (for instance, >5 mV in 30 Torr of SF 6 ). All signal pulses on each of the 16 wires with amplitude >5 mV threshold were analysed in the 55 Fe runs to reject pedestal and electronic noise. Pulses with >3 V amplitude were not included in the analysis to remove sparks and events that saturate the amplifiers and the digitizer.
Background events are expected to be minimal due to the short exposure time (about 2 h) of this source run. To determine the pulse area, any charge that passed the analysis threshold on each of the signal channels were integrated from the 10 µs time before the pulse rising edge crosses the threshold to the 10 µs time after the pulse falling edge crosses the threshold. The energy spectrum of events that passed the analysis cuts is shown in Figure 5. The peak of a gaussian fit on the observed spectrum of the X-ray data is 2140 ± 374 mV µs. To understand the detector gas gain, the amplifier gain was calibrated using test pulses of 14 mV (minimum output voltage of the pulser) and 20 mV to 90 mV amplitude at 10 mV interval. To do this, each of the test pulse signals was connected to the test input of the pre-amplifiers. The mV-scale test pulses were converted to charge signals through a 1 pF test capacitor of the pre-amplifiers. The charge output of the pre-amplifier was then coupled to the shaping amplifier for further amplification and shaping. The shaped pulse signals were then digitized, saved to disk and analysed using the same analyses algorithm used in the X-ray data shown in Figure 5. As in the X-ray data, a gaussian was fitted on each of the amplifier calibration data and the peak of the fits were extracted and analysed as a function of the expected detector gas gain from 5.9 keV X-rays from 55 Fe exposures. Results from the calibration pulse analyses are shown as a function of the expected detector gas gain in Figure 6. The expected gas gain shown in Figure 6 was determined by converting each  of the observed test charge to their equivalent number of ion pairs (NIPs) using: where NIPs o is the observed number of ion pairs, C is the test capacitance of the amplifier, V is the test pulse in mV and e is the electronic charge. The expected number of ion pairs (NIPs e ) from an 55 Fe X-ray can then be determined using: where E is the 5.9 keV energy of 55 Fe X-rays while W is the mean energy required to create an electron-ion pair in SF 6 target gas -measured to be 35.45 eV in Ref. [43]. This implies that the 55 Fe X-ray will produce 166.4 electron-ion pairs after an interaction with the target gas before the electron avalanche. Hence, the detector gas gain can be defined as the ratio of the NIPs o to the NIPs e . Using the gradient and intercept of the linear fit in Figure 6, the gain for the integral charge from the 55 Fe analyses was found to be 1030 ± 180. Note that this measurement was performed at a drift field of 600 V cm −1 with 10.5 kV cm −1 avalanche field in 30 Torr of pure SF 6 which is equivalent to a reduced electric drift field E/N of 62 Td (10 −17 V cm 2 ) and hence, it does not represent the optimal gain of the detector. Here, E is the electric drift field and N is the reduced gas density, for more on this N parameter, see Section 4.

Drift Velocity and Mobility Measurements for SF 6 Anions
To further understand the detector performance and demonstrate the ionization tracking capabilities of the detector, the 55 Fe source discussed in Section 3 was replaced with an 241 Am source which emits 5.5 MeV alphas as it decays to 237 Np. As discussed in Sections 2 and 3, ionization electrons along the interaction track of the ionizing alpha, as it traverses the detector fiducial volume attach to the electronegative SF 6 to form anions. These anions were drifted in the uniform electric field to the high field region of the detector readout for field ionization and subsequent electron avalanche. Signal pulses induced on the wires by this process were pre-amplified, shaped, digitized and saved to disk for analyses. For the drift velocity and mobility measurements, seven runs were performed at different E/N configurations. A constant reduced avalanche field E A /N of 1425 ± 1 Td (10 −17 V cm 2 ) was maintained throughout these alpha runs except in the last run where 1437 ± 1 Td (10 −17 V cm 2 ) E A /N was used to test the response of the ThGEM.
An example of a raw alpha track as seen on the LabVIEW based DAQ is shown in Figure 7. It can be seen that most of the main signal pulses on each of the DAQ channels is preceded by a smaller pulse at ∼100 µs temporal separation with pulse amplitude that is about 2.5% of that of the main pulse. This charge carrier temporal separation and pulse amplitude ratio are consistent with expectations from SF − 6 main and SF − 5 minority charge carriers as discussed in Ref. [24]. These minority charge carriers were not analysed further as their studies is subject of future work. The clear observation of delays between induced charge signals on adjacent wires due to the expected slow negative ion drift properties is of more importance to the work reported here. The wire shown as Channel 0 on the DAQ is closer to the alpha source. The 210 µs delay between channel 0 and channel 15 signals in Figure 7 is due to the drift times of the anions which depends on the incident angle (within the source subtended solid angle) of the alpha track. Events with >40 mV pulse amplitude threshold were analysed further. This is to remove pedestal and electronics noise from the analysis. As  target gas. As discussed in Section 3, the detector gas gain for each of the alpha runs was measured. To do this, a cumulative integral charge in a fixed time window, around the trigger time for a given signal channel was computed. The maximum from this computation was recorded as the channel charge integral. The sum of this charge integral over all the 16 signal channels is the total integral charge for a given alpha track. This method helps to remove the effect of the pedestal noise from the integral charge computations. Samples of total integral charge distributions as observed from two of the drift velocity and reduced mobility measurement runs are shown in Figure 9. There are no visible pedestal noise in Figure 9 as seen in Figure 5(a) due to the lower exposuretimes in the alpha runs because the source activity is 2 orders of magnitude higher than that of the X-ray runs in Section 3. Also, as described above, the charge threshold and cuts applied in the alpha analyses are more stringent for the pedestal noise than in the X-ray data analyses. As described in Section 3, the peak of the gaussian fits in Figures 9(a) and 9(b) were extracted and used as the effective track integral charge for the gain computations. The mean integral charge extracted from the fits in Figure 9, are 132.70 ± 1.49 V µs and 158.20 ± 1.73 V µs for E/N runs of 68 Td (10 −17 V cm 2 ) and 87 Td (10 −17 V cm 2 ), respectively. Similar analyses were performed on the remaining 5 alpha runs taking at different E/N configurations.
To convert these results to a gain measurement, a SRIM simulation was performed to determine the fraction of the 5.5 MeV alpha energy that was deposited within the detector fiducial volume as the tracks are expected to be longer than the detector width. This was found to be 0.22 MeV in average, which translates to 4% of the total alpha  energy. Using this average alpha deposited energy and calibration results from analyses of the pulse calibration runs described in Section 3, the gas gain in each of the alpha runs were determined. The gas gain results from these measurements are shown as a function of E/N in Figure 10. It can be deduced that the 1030 ± 180 gas gain measured in Section 3 from X-ray events at 62 Td (10 −17 V cm 2 ) E/N is consistent with 1156 ± 88 gain obtained from alpha tracks in Figure 10, when extrapolated to the 1087 ± 1 Td (10 −17 V cm 2 ) E A /N used in the X-ray measurement. The observed gain depicted in Figure 10 increases with E/N as expected due to the difference in the operational negative ion transparency in the high field region, closer to the ThGEM. It can be seen in Figure 10 that the observed gas gain plateaued from the 75 Td (10 −17 V cm 2 ) to 87 Td (10 −17 V cm 2 ) E/N runs. This is consistent with expectations from reaching the optimal negative ion drift field transparency for the given E A /N setup. As discussed earlier, the observed rise in the detector gas gain for the E/N run of 93 Td (10 −17 V cm 2 ) is due to the higher E A /N configuration in this run. However, the drift velocity and mobility of anions are independent of the gas gain so these observed plateau and rise in the detector gas gain should not affect our measurements. These results indicate that a gas gain of 2.5 × 10 3 is feasible with the ThGEM-multiwire hybrid setup in SF 6 target gas.
To extract the drift velocity and mobility of the SF 6 anions, only alpha tracks with incident angle of 9 • were selected. As shown in Figure 11, these tracks are expected to induce signals only on the first 8 wires of the detector. Hence, events that recorded hits on these 8 wires, only were selected for further analysis. A hit here is charge induced on a wire that produced a pulse amplitude that passed the analyses threshold. Between  0.2% to 2.3% of the total events on disk in the alpha runs passed these cuts and were used in the drift velocity and mobility measurements. Using the incident angle of the alpha track, the distance d between the charge charge transfer side of the ThGEM and the mean interaction vertex for events that induced signals on each of the first 6 wires was determined. It can be seen in Figure 11 that there is no drift distance for most probable ionization charge that could end on the 7 th and 8 th wires due to the proximity of the interaction vertex to the field ionization region and readout plane, respectively. Hence, signals on these two wires were not included in the drift velocity and mobility computation. The difference between the distance ∆d (arrival times ∆t) for any two successive wires were computed as ∆d 1 (∆t 1 ), ∆d 2 (∆t 2 ), ∆d 3 (∆t 3 ), ∆d 4 (∆t 4 ) and ∆d 5 (∆t 5 ), see Figure 11 for an illustration. The drift velocity v d is an average from the 5 measurements, given by: where n is the number of ∆d and ∆t measurements which is 5 in this case. The mobility µ can then be defined in terms of v d as: where E is the drift field of the anion. This can be converted to the reduced mobility µ 0 by considering the operational gas density in each of the measurements for better comparison with other measurements using: As mentioned in Section 3, N is the reduced gas density defined as ρA Mm in cm −3 , where ρ is the gas density for the gas pressure used in a given measurement, A is the Avagadro's constant 6.0221×10 23 mol −1 and Mm is the molar mass of SF 6 given as 146.06 g mol −1 . The N 0 is the N parameter computed at the standard temperature (0 • C) and pressure (1 atm or 760 Torr) which can be evaluated to obtain 2.6868 × 10 19 cm −3 . Results from the v d and µ 0 measurements using Equations 5 and 7, respectively are depicted in Figure 12 as a function of E/N configurations for a given run in 20 Torr of SF 6 target gas. It can be seen in Figure 12(a) that the observed drift velocity increases with E/N as expected. These observed v d results are consistent within errors with SF − 6 results in Ref. [24] for the relevant E/N used in these measurements. The large error in the 93 Td (10 −17 V cm 2 ) E/N results is mainly due to lower statistics in that measurements as only 20 events passed the analyses cuts. This is due to a lower source exposure time. Apart from this, every other measurement in Figure 12 had between 50 to 300 events after the implementation of the cuts. The observed reduced mobility for the respective E/N values in Figure 12

Conclusion
A ThGEM-Multiwire based hybrid time projection chamber (TPC) detector was designed, built and tested for the first time. Gas gain measured from the performance tests of the hybrid detector are in the range of 1120 ± 90 to 2470 ± 160 at a reduced drift field E/N range of 56 Td (10 −17 V cm 2 ) to 93 Td (10 −17 V cm 2 ) in 20 Torr of pure SF 6 target gas. Using the hybrid detector, the drift velocity and reduced mobility of SF 6 anions were measured at this E/N range. The observed drift velocity (reduced mobility) results were found to be between 80 ± 2 m s −1 and 139 ± 12 m s −1 (0.53 ± 0.01 cm 2 V −1 s −1 and 0.56 ± 0.05 cm 2 V −1 s −1 ) in this E/N range. The drift velocity and reduced mobility results from these measurements are consistent within errors with other published measurements in Refs. [24,45,46,47].
Hence, the ThGEM-Multiwire technology has the potentials to serve as a robust, low noise charge readout with known fine grain track resolution in the future massive directional dark matter detector know as CYGNUS-TPC.