Characterisation of resistive MPGDs with 2D readout

Micro-Pattern Gaseous Detectors (MPGDs) with resistive anode planes provide intrinsic discharge robustness while maintaining good spatial and time resolutions. Typically read out with 1D strips or pad structures, here the characterisation results of resistive anode plane MPGDs with 2D strip readout are presented. A µRWELL prototype is investigated in view of its use as a reference tracking detector in a future gaseous beam telescope. A MicroMegas prototype with a fine-pitch mesh (730 line-pairs-per-inch) is investigated, both for comparison and to profit from the better field uniformity and thus the ability to operate the detector more stable at high gains. Furthermore, the measurements are another application of the RD51 VMM3a/SRS electronics.


Introduction
In recent years, various experiments and detector R&D lines started the use of Micro-Pattern Gaseous Detectors (MPGDs) with resistive elements, mainly because of their discharge robustness [1][2][3][4].Additionally, parameters such as the signal induction, i.e. the spread of the charge over a given number of readout channels, and rate-capability can be tuned to the desired values of the experiment by adjusting the design of resistive elements.This is most prominently employed in MPGDs with a single amplification stage, such as MicroMegas (MM) [5] and µRWELL [6].
In this paper, the results of a characterisation of these two technologies are shown.The studies are conducted in view of various R&D aspects.At first, there is the technology of the readout anode, a 2D strip structure underneath a single layer of resistive material.So far, most resistive MPGDs employ either 1D strip structures underneath the resistive layer, 1D resistive strips or resistive pads.The second aspect is specific to the µRWELL detector that has been investigated.It is a prototype of what is supposed to be used as reference detectors for particle trajectory reconstruction in a new beam telescope of the DRD1 collaboration [7,8] -the preceding RD51 collaboration [9,10] provided for its joined test beam campaigns at the H4 beam line of CERN's SPS two beam telescopes, one based on MicroMegas and one based on triple-GEM detectors [11].While a previous prototype with 2D X-Y strip readout showed good performance [12], it showed an unequal signal sharing between the top strip layer and the perpendicular bottom strip layer.For the new prototype which was investigated for the here presented studies, this imbalance has been addressed.The third point is specific to the investigated MicroMegas detector.The particular detector is equipped with a fine mesh in the amplification stage with 730 line-pairs-per-inch (LPI), corresponding to a mesh  cell pitch of 35 µm.The motivation for investigating a mesh structure like this is a more uniform electric field and thus the possibility to operate the detector at higher gains.While the stability measurements are part of a separate study -here only the results on the spatial resolution, time resolution and charge behaviour are presented -it can be noted that the MicroMegas detector could be operated at gains more than twice as high as the µRWELL detector.Another more general aspect is the application of the new VMM3a/SRS front-end electronics [13] of the RD51 collaboration to a wider range of detectors.

Devices under test
Both detectors (sketched in Fig. 1) have an active area of 10 × 10 cm 2 with a 2D X-Y-strip readout with 256 strips in each direction.The strip pitch is 400 µm.In both detectors, the anode is a layer of Diamond-Like Carbon (DLC) with similar surface resistivities -40 MΩ/sq for the µRWELL and 37 MΩ/sq for the MicroMegas.The drift region of each detector had a width of 3 mm.Although both detectors were operated with a negative high voltage on the cathode, the MicroMegas detector was operated with a grounded mesh and a positive high voltage on the DLC anode, while the µRWELL was operated with a negative high voltage at the top layer of the Well structure and a grounded DLC anode.
When performing gain scans, the drift field was kept constant at around 2.4 kV/cm for the µRWELL, which corresponds to 724 V difference between drift cathode and Well, and 0.9 kV/cm for the MicroMegas detector, which corresponds to 275 V between cathode and mesh.The amplification voltages were varied from 460 to 570 V for the µRWELL and 580 to 700 V for the MicroMegas -at higher voltages, the detectors started to show instabilities in operation due to discharges leading to high voltage trips.When performing drift scans, the amplification voltage was kept constant at 680 V for the MicroMegas, which corresponds to an effective gain of around 40 000 at 0.9 kV/cm drift field, and at 540 V for the µRWELL, which corresponds to a gain of around 15 000 at 2.4 kV/cm drift field.

Beam telescope and readout electronics
Both detectors have been characterised as Devices Under Test (DUTs) with the RD51 VMM3a/SRS beam telescope [11].It consists of three COMPASS-like triple-GEM detectors [14] with an active area of 10 × 10 cm 2 almost equally spaced with a total lever arm of around 1 m to provide the position information.In addition, it contains three scintillators with Photo-Multiplier Tubes (PMTs) connected to a NIM coincidence unit as a time reference.All the MPGDs in the telescope have been operated with the same gas mixture of Ar/CO 2 (70/30 %).
For the readout of the detectors, including the output of the NIM coincidence unit, the AT-LAS/BNL VMM3a front-end ASIC [15] in its integration [16] of the RD51 Scalable Readout System (SRS) [17] has been used.It provides the acquired charge per channel (10-bit ADC, effectively 7-bit) in a continuous, multi-channel self-triggered readout mode with around 1 ns time resolution and a MHz rate-capability.The adjustable analogue front-end parameters of the VMM3a have been set to 200 ns peaking time, 9 mV/fC electronics gain and around 10 000 electrons threshold per channel for both the DUTs and the reference detectors.The threshold level might seem high, but it should be considered that the VMM3a is operated in its self-triggered readout, meaning that each signal above the Threshold Level (THL) will be processed and become part of the data stream, i.e. a THL that is set too low will results in a lot of external pick-up noise being part of the raw data.Furthermore, in terms of the dynamic range of the VMM3a, the THL is still a low value of less than 2 % per channel.

Charge behaviour
With the previous µRWELL prototype showing an imbalance of the charge collection between the top and the bottom readout strips [12], the first point of investigation is the charge sharing and collection behaviour of the detectors.For this, the charge distribution (Landau distribution) is generated from all recorded interactions that can be assigned to a reconstructed particle trajectory from the beam telescope.Then the mean value of this measured distribution is taken and plotted depending on the effective gain of the detector (Fig. 2a).It can be seen that both strip planes collect almost the same amount of charge, resulting in a charge-sharing ratio between the top and bottom  strips of each detectors close to one.Another observation is that the mean charge measured with the Landau distribution increases linearly with the gain for the µRWELL detector, while for the MicroMegas detector, the increase starts to flatten out at higher gains.This is due to the ADC saturation of individual readout channels which at higher gains is more likely to occur.Another observation is that the total collected charge for the µRWELL is significantly less than for the MicroMegas detector, despite the same effective gain.This is due to the signal induction from the anode to the readout strips.While the effective detector gain was determined from the current measured on the resistive anode layer, the front-end electronics only measures the induced current in the capacitively coupled readout strips.This is important to note, as the gain at which a discharge might occur in the detector is the one determined via the signal current on the anode plane.
This explanation is strengthened by the behaviour of the cluster size (number of channels above the THL), with the mean cluster size being plotted against the effective detector gain (Fig. 3a).It can be seen that the cluster size scales accordingly, with the bottom strips of the µRWELL collecting slightly more charge, also having a slightly larger cluster size.And an almost doubled cluster size for the MicroMegas detector at similar gains, leading to almost twice the amount of measured charge.This behaviour is also reflected in the detector efficiency (Fig. 3b), which is defined as Here,  tracks is the number of particle interactions and trajectories reconstructed with the reference tracking detectors and  det is the number of particle interactions that have been recorded in the DUT and that can be assigned to an existing trajectory.Due to the difference in signal induction between the investigated µRWELL and MicroMegas detectors, reflected by the smaller cluster size and measured signal amplitude for the µRWELL, higher detector gains are needed to reach comparable cluster sizes and induced signal charges and thus comparable efficiencies.In addition to the dependence on the effective gain of the detector, also the dependence of the measured charge depending on the electric drift field was studied, representing the charge collection behaviour by the amplification stage.The results are shown in Fig. 2b.Both detectors show the expected characteristic behaviour, with the charge collection of the µRWELL having a much broader peak at higher electric fields (e.g. as shown in [6]), while the MicroMegas shows a more well define peak at lower electric fields (e.g. as shown in [18]).

Spatial resolution 4.1 Basic results
The spatial resolution of the DUTs is extracted from the width of the residual distributions that are generated from the difference Δ =  ref −  ′ between the reference particle position  ref that is provided by the reconstructed trajectory from the reference tracking detectors and the interaction point  ′ reconstructed in the DUTs.To determine the width, two overlapping Gaussian functions are fitted to the residual distribution, with the mean value being identical, but a weight factor  to account for the different scales and different standard deviations  to account for the core and the tails of the residual distribution.The final width is thus defined as The spatial resolution is then obtained by quadratically subtracting the contribution from the uncertainty on the track reconstruction, as described in [18][19][20].
Using the Centre-Of-Gravity (COG) to determine the position of the particle interaction within the detector, the gain dependence of the spatial resolution as shown in Fig. 4a is obtained.For both detectors, the spatial resolution improves with increased detector gain.Taking the cluster size behaviour (Fig. 3a) into account, this is related to more charge information being available and distributed over more than a single readout strip.This also explains the better spatial resolution of the MicroMegas compared to the µRWELL, as it has a larger charge collection and cluster size.A similar dependence can be observed when plotting the drift behaviour (Fig. 4b).The spatial resolution is inversely proportional to the charge collection depending on the electric drift field (Fig. 2b), i.e. the more charge collected, the better the spatial resolution.At very high gains (Fig. 4a), the spatial resolution starts to decrease for the MicroMegas detector.This is due to the saturated readout channel that also caused the flattening of the measured charge-gain-dependence (Fig. 2a).The position reconstruction by COG loses accuracy when the relative amount of charge is not correctly represented anymore because the corresponding readout channel is in saturation.
Another observation is that the spatial resolution for the µRWELL is almost the same for the two readout planes, as it is expected from equal charge sharing, while for the MicroMegas the behaviour between the top and bottom strips deviates.This was found to be the result of the so-called 'readout modulation', as illustrated by Fig. 5.It shows the distribution of the reconstructed interaction points with high granularity.This makes a peak structure visible which originates from the readout pattern modulated into the position distribution that is expected to be uniform due to the detector's uniform irradiation.Due to this modulation effect by the discrete readout structure in combination with a threshold-based zero-suppressed readout electronics, the interaction points are more likely to be reconstructed to the central strip for odd-strip-count clusters and in-between the two central strips for even-strip-count clusters.For a more detailed description of this effect, it is referred to [21], while it should be noted that the effect has already been observed with Multi-Wire Proportional Chambers [22].This behaviour affects the accuracy of the position determination, with a stronger effect leading to a worse spatial resolution.This is what can be conducted from Fig. 5, where the modulation effect is stronger for the top strips than for the bottom strips.It seems to be an intrinsic  behaviour of the detector as it is also observed in laboratory measurements using an 55 Fe source.A possible explanation that was found for this behaviour is the way the signal induction.With the top strips being much thinner than the bottom strips, the distribution of the induced signal charge changes compared to the bottom strips, with more charge being in the cluster's central strip for the top strips and less charge being acquired in the outer strips.

Improving the spatial resolution
While so far, the results have been obtained with COG to calculate the position, previous studies [11,21] showed that a different weighting of the charge in the COG formula with  the index of the readout strip with a signal above THL and  the weighting factor reduces the modulation effect and thus improves the spatial resolution.This is because of low amplitude signals in the tails of the charge distribution that go above the THL on one side of the charge distribution but not on the other one.Thus, a certain fraction of charge information to reconstruct the position is lost and the reconstruction of the interaction point is forced towards the signal above the THL.To reduce the weight of the charge distribution's tails and thus mitigate the bias in the calculation of the interaction point,  > 1 can be selected.In previous studies with triple-GEM detectors [11],  = 2 was used due to its simplicity and its proximity to the optimal solution obtained from simulation studies [23,24].Thus, it was investigated, if this  2 weighting can also be used to improve the spatial resolution of µRWELL and resistive MicroMegas detectors.
In addition, an orthogonal approach was investigated, making use of a hardware feature of the VMM3a front-end ASIC, its Neighbouring-Logic (NL).By default, the neighbouring-logic is turned off, but when enabled, the NL triggers the acquisition of an induced signal with an amplitude  below the THL, if the neighbouring channel has a signal that surpasses the THL.This allows to obtain more charge information and thus to improve the reconstructed position.The results of these two methods, including their combination, are shown in comparison with the COG reconstruction in Fig. 6.It can be seen that the  2 weighting has a large impact and significantly improves the spatial resolution, in both detector cases.On the other hand, the effect of the NL is only visible at low gains, i.e. at low efficiencies and low signal-to-threshold ratios, where the relative amount of collected charge on the total cluster charge is large, with a large fraction of the induced signal charge being still below the THL.At higher gains, this effect gets reduced, which is also reflected in the larger cluster size, i.e. on more channels a signal above THL was acquired.Thus, the tails of the measured charge distribution contain less induced signal charge and the probability of acquiring electronics noise increases.As a result, the spatial resolution decreases.These two outcomes are exactly in line with what has been observed in previous studies [11,20].

Time resolution
As the last part of the characterisation studies, the time resolution of the two detectors is investigated.For this, the interaction time of the particle measured in the DUTs is compared with the reference time measured with the scintillator/PMT/NIM-coincidence-unit combination, both acquired with VMM3a/SRS.The reference time is provided as a constant amplitude signal -due to the NIM output of the coincidence unit -on a single channel of the VMM3a.The measured interaction time in the DUTs is defined as ( top +  bottom ) . (5.1) The interaction timestamps for each plane,  top and  bottom , correspond to the time of the signal within each cluster with the largest peak amplitude.For comparison purposes, also the results from measurements with a small-pad resistive MicroMegas detector from the RHUM project have been added [3].
With both timestamps found, their difference Δ =  ref −  ′ is calculated.Fitted to the resulting distribution is a single Gaussian function, with its width  Δ .The final time resolution is then this width, from which the time resolution contributions of the VMM3a (around 2 ns at the used peaking time of 200 ns) and the scintillator/PMT/NIM-logic (around 1.5 ns) are quadratically subtracted.The obtained time resolutions have been plotted against the electric drift field (Fig. 7a) and the electron drift velocity (Fig. 7b) which was calculated through Magboltz [25].Added for reference, are data from measurements with a small-pad resistive MicroMegas detector from the RHUM project [3].All data points show the same trend of an improved time resolution with increasing drift velocity.Two aspects should be noted though in regard to the used gas mixtures.The RHUM data have been obtained with a mixture of Ar/CO 2 /iC 4 H 10 (93/5/2 %) for the low drift velocities and Ar/CF 4 /iC 4 H 10 (88/10/2 %) for the high drift velocities (> 6 cm/µs).In the case of the µRWELL, which has been filled with Ar/CO 2 (70/30 %), the drift velocity does not change significantly at fields above 2.5 kV/cm -a linear increase of around 1 cm/µs over a range of 2.5 kV.This explains the observed saturation behaviour of the time resolution.In the case of the MicroMegas detector, this saturation behaviour could not be observed, because of the high-voltage power supply, which did not allow it to go to larger drift voltages.
In addition to the observation that the time resolution in the three detectors follows the same trend and that the results are compatible with each other, two other points are shown by the measurements.Especially highlighted by the results from the resistive plane MicroMegas detector, it becomes clear that the working point for the best detector performance in terms of charge collection and spatial resolution (here at 0.9 kV/cm) is not necessarily the working point for the best time resolution (here > 3 kV/cm).Secondly, due to the capabilities of VMM3a/SRS, both types of detectors, as well as the reference timing detectors, could all be read out by the same front-end electronics, with the corresponding data being contained in a single data stream.This shows the versatility of the readout system, but also how it simplifies the data-taking and analysis process as it is not necessary to rely on additional high-precision timing electronics with an additional data stream that requires additional effort in the offline data analysis.

Conclusion and outlook
In this paper, the results from characterising two single-stage resistive plane MPGDs, a µRWELL detector and a MicroMegas detector, with 2D strip readout have been presented.Due to the capabilities of the new VMM3a/SRS readout electronics, it was possible to study simultaneously the charge behaviour, the spatial resolution and the time resolution of these two detectors.
One immediate observation was the difference in the amplitude of the induced signal measured by the front-end electronics between the two detectors, despite them being operated at the same gain.This shows the importance of a good signal coupling between the anode and readout structure as otherwise, the detector efficiency can be still not sufficient although the detectors might be operated close to the breakdown voltages.Both detectors showed good performance though.
The µRWELL detector was investigated as a prototype detector for a future beam telescope of the DRD1 collaboration.With time resolutions of 10 ns, spatial resolutions of better than 70 µm in Ar/CO 2 (70/30 %) and equal charge sharing between the two readout strip planes, the detector fulfils the requirements to be used in the new telescope.The MicroMegas detector was built with a 730 line-pairs-per-inch (LPI) mesh, corresponding to a mesh cell pitch of 35 µm, in order to increase the operation stability.Although the stability studies will be presented separately, the detector showed a good performance -e.g.spatial resolutions of around 50 µm -while being operated stably at gains of more than 50 000.In addition, the results demonstrated that the optimal working point to achieve the best spatial resolution and charge collection, is not necessarily the best working point to achieve the best time resolution.

Figure 1 :
Figure 1: Sketched cross-section of the MicroMegas (a) and the µRWELL (b) amplification stages, as used in the here presented measurements.

Figure 2 :
Figure 2: Average measured charge, i.e. the calculated mean of the energy loss distribution after amplification, depending (a) on the effective gain of the two investigated detectors or (b) on the drift field, in each read out detector plane.

Figure 3 :
Figure 3: In (a), the mean of the cluster size distribution (distribution of the number of channels above the THL for each recorded interaction) is shown, depending on the effective gain of the detectors.In (b), the detector efficiency as defined in Eq. (3.1) is shown, depending on the effective gain.

Figure 4 :
Figure 4: Dependence of the spatial resolution of the two DUTs on (a) their effective detector gain and on (b) the electric drift field, determined for both readout planes individually.

Figure 5 :
Figure 5: Distribution of the reconstructed interaction points at high granularity for the MicroMegas detector at high detector gains (a) and low detector gains (b).

Figure 6 :
Figure 6: Dependence of the spatial resolution on the effective detector gain for the two different methods to determine the interaction point, with and without NL enabled.Here only the results of the bottom strips are shown, but the observed behaviour is the same for the top strops.

Figure 7 :
Figure 7: Measured time resolution for the two different detectors, depending on the electric drift field (a) and the electron drift velocity (b).For comparison purposes, also the results from measurements with a small-pad resistive MicroMegas detector from the RHUM project have been added[3].