Performance studies of Micromegas electronics in a high radiation environment at the CERN Gamma Irradiation Facility (GIF++)

The ATLAS experiment at CERN undergoes a series of upgrades, in line with the upgrades (Phase-I and Phase-II) of the Large Hadron Collider (LHC), to cope with the steadily increasing instantaneous luminosity that ultimately is expected to reach 7.5×1034cm−2 s−1 at the High Luminosity LHC (HL-LHC) era. The most challenging upgrade project of the ATLAS experiment concerns its Muon Spectrometer’s inner wheel-shaped detection stations (Small Wheels) located close to and on either side of the pp beams’ interaction point. Another type of detection system (New Small Wheels) replaced the two Small Wheels so that the ATLAS maintains its performance while preserving the acceptance of critical physics signatures under the harsh conditions of higher data and radiation rates (up to 25kHz/cm2 at the HL-LHC accelerator). The NSWs are a complex detection system (each wheel comprises 2.4x106 readout channels)that employ two novel micro-pattern gaseous detector technologies, the Micromesh Gaseous Structure (Micromegas, MM) and the small strip Thin Gap Chambers (sTGCs), mainly to improve the trigger performance and to provide precise spatial measurements. New electronic boards residing on the detectors were designed and manufactured, hosting high radiation and magnetic field tolerant custom-made ASICs.This paper presents extensive tests carried out at the new CERN Gamma Irradiation Facility (GIF++) by combining a muon beam (of 12kHz) with a 137Cesium radiation source of adjustable rate (up to ten times the expected rate at the HL-LHC conditions), to assess the performance of the Micromegas chambers and their readout electronics. Data from NSW Micromegas modules of different detection areas, having different positions on the wheels, operating with different high voltage and filling gas have been read out and analyzed by placing the electronics on PCBs located at various distances from the pp beam pipe. The spatial resolution of the precision coordinate is found to be well below the required 100μm, and the excellent alignment of the precision layers has been verified. Results confirm that all types of Micromegas modules tested have a linear response up to a rate around four times the expected at HL-LHC.


Introduction
The Large Hadron Collider (LHC) [1] at CERN to extend its physics potential is scheduled to increase its instantaneous luminosity.Its first upgrade (Phase-I) was accomplished during the long technical shutdown (2019-2022), and during its current operation (run3), an instantaneous luminosity of 2-3×10 34 cm −2 s −1 , is expected to reach.After its next (Phase-II) upgrade (2025-2027), the High Luminosity (HL-LHC) will begin its operation with a foreseen maximum instantaneous luminosity of 7×10 34 cm −2 s −1 .
Due to the increased LHC instantaneous luminosity, the higher data and trigger rates and the higher beam-induced background would degrade the ATLAS [2] subdetectors' efficiency and raise the fake muon triggers, especially in regions closer to the beam pipe.Therefore, to keep its excellent performance in terms of efficiency, resolution, and background rejection while preserving the acceptance of critical signatures for many physics channels, the ATLAS experiment has to upgrade its detectors and electronics.
The challenge of the ATLAS upgrades is its Trigger [3] and Data AcQuisition [4] system (TDAQ) and trigger algorithms to maintain low transverse momentum trigger thresholds at manageable trigger rates in a much higher background environment.For this, the forward regions (end-caps) of the ATLAS Muon Spectrometer [5] are of particular importance.
The Muon Spectrometer is arranged in three coaxial cylindrical detector systems around the beam axis (barrel region) and four parallel plates on both sides of the barrel, perpendicular to the beam axis (end-caps).The expected hit rate at the innermost end-caps (the "Small Wheels") is expected to be up to 15 kHz/cm 2 , necessitating, therefore, its replacement with another type of detection system, named "New Small Wheels" (NSWs) [6].
Each NSW accommodates two novel detector technologies; the small-strip Thin Gap Chambers (sTGC) [7], mainly for triggering, and the Micro-Mesh Gaseous Structure (MicroMegas, MM), mainly for the precise measurement of the muon's position and momentum [8].
This work aimed to investigate the performance of the NSW's MM detectors in a high radiation of up to ten times the rate expected at the HL-LHC conditions ( 25kHz/cm 2 ).
The MM operation principle and the NSW layout are briefly described in sections 2 and 3.Then, section 4 describes the MM trigger and readout electronics used for the tests.Section 5 presents the hardware testing setup and the tests performed.Finally, section 6 summarizes the results of the tests.

The Micromegas Detector
The MM detectors (figure 1) are gaseous detectors that consist of three planes, the cathode (drift), the anode (readout), and a thin mesh plane (micromesh).The micromesh divides the gas volume into two regions; the conversion/drift gap (5 mm) and the amplification gap (128 µm).Charged particles entering the conversion/drift region ionize the gas and create electronion pairs.The produced ions drift towards the cathode while the electrons drift towards the micromesh.Because of the much higher electric field of the amplification region, the created electrons pass through the mesh to the amplification region, where electron avalanches occur.The induced charge on the readout strips forms a fast pulse read out by the front-end electronics.The resistive strips, separated from the readout strips by an insulating layer, serve to make the chamber discharge tolerant [9].Both the anode and cathode are based on printed circuit boards (PCB) [10].3. The New Small Wheel project The NSW system [6] is composed of two wheels, each one placed at the corresponding end-cap (A or C) of the ATLAS Muon spectrometer, and employs MM and sTGC detectors.Each NSW wheel comprises eight large and eight small sectors, trapezoidal shaped and fitted in a metal circle structure, with a diameter of 10 meters (figure 2a).Each sector comprises two MM wedges mounted on both sides of a metallic central spacer frame and sandwiched by two sTGC wedges (figure 2b).Each wedge of a sector is assembled radially from two separate trapezoidal shape quadruplet modules (figure 2c), trapezoidal shaped and off different sizes, each consisting of four detection layers (quadruplet).The SM1 (LM1) module of the Small (Large) sector having a smaller radius (closer to the beampipe), whereas the SM2 (LM2) is the one with the higher radius.The chambers' areas of the small (SM1, SM2) and the large (LM1 and LM2) modules are 2m 2 and 3m 2 , respectively.The MM quadruplets are composed of two readout panels (named eta and stereo) that carry the readout boards and three drift panels comprising the drift electrode and the micro-mesh, constructed as described in [11].Each of the two readout panels has on each side two detection layers.On the eta readout panel, the readout strips of the layers (named η Layer 1 and 2) are oriented parallel to the trapezoid bases for measuring the η (precision) coordinate.In contrast, on the stereo panel, the strips of the layers (named stereo Layer 3 and 4) are tilted (±1.5 • ) relative to the η-strips for measuring the azimuthal coordinate.Each readout layer in the SM1 and LM1 modules is segmented into five readout boards (PCBs numbered 1,2,3,4,5), while in the SM2 and LM2 modules, each readout layer has three readout boards (PCBs numbered 6,7,8), as illustrated in figure 3. MM chambers with different detection areas and positions relative to the beam pipe (an SM1 and an LM2 module) and with their readout electronics placed on all the detection layers of selected PCBs have been tested to investigate their performance, as described in section 5.2.

The Micromegas Electronics
The NSW electronics (figure 4) are classified into two categories; on-detector and off-detector electronics.The on-detector electronics reside on the detectors and are inside the detector area (cavern), whereas the off-detector are located in the USA area (a few meters outside the cavern).The Micromegas' readout electronics used for the tests are shown in the lower part of figure 4.These are a) the front-end boards [12] and b) the Level-1 Data Driver Card (L1DDC) [13].
The front end boards host a) eight Venetios Micromegas (VMM3) [14] that provide the peak amplitude and the time relative to the bunch crossing clock or other trigger signals via 64 channels, where each channel is connected to a readout strip, (b) one Read Out Controller (ROC) [15] that transmits the data to the readout path and (c) one Slow Control Adapter (SCA) [16] used for the configuration and monitoring of all the on-detector boards.
The main radiation tolerant ASICs that the L1DDC hosts are the GigaBit Transceiver (GBTX) [17], a high-speed serializer/deserializer ASIC responsible for the bidirectional data transmission between the front-ends and the off-detector network interface system (Front-End LInk eXchange, FELIX) [18].
Data from multiple VMM3s are collected by the ROC ASIC, which transmits them through serial links at 320Mb/s to the GBTX ASIC of the L1DDC board.Then, the GBTX transmits them via a single fiber to FELIX and finally to the Read Out Devices (ROD) [19] and the Detector Control System (DCS) [20].

The Micromegas tests at the GIF++
Extensive tests have been carried out in the new CERN Gamma Irradiation Facility (GIF++) [21] located in the North Area of the Super Proton Synchrotron (SPS).The GIF++ combines a high-energy particle beam (80-150 GeV) in the SPS H4 beam line with a 14 TBq 137 Cesium source of 662keV to simulate the radiation intensity expected at the HL-LHC.A system of absorbers can tune the flux of the emitted gamma rays.

The Experimental Setup
The experimental setup in the SPS H4 Experimental Hall (figure 5 ) consists of the MM modules, two scintillators (SC1, SC2) to provide the trigger, a 14 TBq Caesium ( 137 Cs) gamma source, and an 80-150 GeV muons beam with a rate of 12kHz/cm 2 rate).Three modules of different types (SM1 and LM2), operated with isobutane gas and a small prototype (Tz) with an active area of 10 x 10 cm 2 were tested.Tz uses one layer, and the readout is provided by one board connected to 256 strips.Due to its small capacitance, it has an ideal linearity behavior and was used for reference.The modules were positioned sequentially and vertical with respect to the beam, between the scintillators.The distances with each other are shown are shown in figure 5.
The irradiator was placed at about 1m distance from the SM1 module (corresponding to a gamma rate of (250kHz/cm 2 ), and its rate was attenuated down to the rate expected at HL-LHC 25 kHz/cm 2 by using a set of absorbers.For the SM1 module, the MMFE8 boards were placed on PCBs 2 and 3, whereas for the LM2, on PCBs 6 and 7.
In total, thirty-three MMFE8s were used (sixteen in each module and one in the Tz chamber), and four L1DDCs for obtaining the Level-1 accept (L1) data.The muon beam was crossing the whole Tz chamber and the PCB 2 and 6 of the SM1 and LM2 modules, respectively.

Results
Tests under all source and muon beam operation combinations have been performed, and measurements under different values for the high voltage and the VMM3s' parameters have been taken to investigate the modules' performance.
The following paragraphs present the results of the measurements taken with the VMM3s'gain set to 9mV/fC and the integration time of the pulses equal to 200ns.All chambers were filled with isobutane gas mixture and operated at a high voltage of 520V.
To evaluate the spatial resolution of the precision coordinate, the modules were irradiated with the muon beam (12kHz/cm 2 rate) in the absence of the 137 Cs source.Their readout planes were perpendicular to the beam.Figure 6 shows the beam profile reconstructed using stereo layers of the SM1 and LM2 modules.A passivation area at the SM1 module can be noticed.For the evaluation of the spatial resolution of the precision coordinate, the cluster position on two precision layers has been reconstructed using the charge centroid method, as described in [22].Assuming that the spatial resolutions of the two layers are equal, the combined spatial resolution can be estimated by the difference in the cluster position on the two layers.The distribution of the difference was fitted with a Gaussian function figure 7, and the derived standard deviation (σ) was divided by √ 2 [23].The core distribution of the difference (95% of the events analyzed) that gives an estimation of the intrinsic detector resolution, was fitted with a single Gaussian (represented by a black line in figure 7), while a double Gaussian fit was applied to account for the tails (red lines in figure 7).A resolution of 64µm (84 µm) for the SM1 (LM2) module has been obtained, below the 100µm required for the precision coordinate.It can also be noticed that the mean value of the position distributions (for both SM1 and LM2 modules) is almost equal to zero, indicating a good alignment of the precision layers.(For the LM2, the observed offset is about 50µm, and can be considered negligible compared to the 450µm pitch of the strips).The response of the detectors under high radiation rates has been evaluated using the muons beam and the 137 Cs source simultaneously.As a metric, the hits occupancy as a function of the source rate has been studied.
The theoretical values expected for the hit occupancy (HOth) are calculated using the formula: HO th = Rate × sectorP itch × P CBLength × N umberOf HitsP erCluster × ReadOutW indow (1) where Rate is the radiation rate of the 137 Cesium source (in kHz/cm 2 ), sectorPitch is the gap between two consecutive strips, PCBLength is the length of the PCBs, NumberOfHitsPerCluster is the Number of Hits that form a cluster, and ReadOutWindow is the duration of the integration of a pulse performed by the VMM3.According to the previous formula, the dependence of hit occupancy versus the rate is linear (figure 8).Given that the SectorPitch is 0.0425cm (0.045cm) for the SM1 (LM2), the PCBLength is 68cm, the NumberOfHitsPerCluster equals 5, and the ReadOutWindow is 200ns, the expected hit occupancy at the HL-LHC rate (20 kHz/cm 2 ) is 5.8%.For the hits occupancy study, the maximum source's intensity (250kHz/cm 2 ) at a distance of 1m from the modules was attenuated down to the expected at the HL-HC rate using a set of absorbers and was calculated as:

HEP-2022
where the Average Number of Strip Hits was measured by analyzing the Level-1 data received during a particular test.
Figure 9 shows the hit occupancy as a function of the source's inverse attenuation for the SM1 and LM2 modules and the prototype Tz.As seen, the SM1 and LM2 modules saturate approximately at a rate of 25% of the maximum source's intensity which corresponds to four times the rate expected at HL-LHC.
The saturation of the modules at high rates is due to the faster gas ionization, consequently in higher electron production in the chambers' conversion region.The charge of the strips (because of their long length) cannot evacuate fast enough, leading to a significant number of strip hits in a given time interval.Since the VMM3s can read only one hit in that interval, any other hit during that time window is missing.Indeed, as shown in figure 9c, the Tz chamber has a linear behavior since its strips have a shorter length.The impact of the high voltage on the performance of the detectors has also been studied.In the tests, the muon beam was off, and a random trigger of approximately Hz was used.Figure

Conclusion
The performance of two Micromegas modules (SM1 and LM2) and their readout electronics have been tested at the CERN GIF++ facility using a high energy muon beam of 12kHz/cm 2 rate and a 137 Cesium source of rates up to ten times the rate expected in the forthcoming HL-LHC.For tracks perpendicular to the modules' planes, a spatial resolution of 64µm (84 µm) for the SM1 (LM2) module has been obtained, well below the 100µm required for the precision coordinate.Extensive tests performed in the presence of both the muon beam and the irradiator indicate that the hit occupancy preserves its linearity up to a rate of about four times the rate expected at HL-LHC.
The saturation of the electronics without the 137 Cs irradiator was also studied for different high voltage values.As expected, the high voltage is a trade-off parameter between high values, essential for the efficient ionization of the gas, and low ones, so that the electronics' saturation occurs at higher rates.

Figure 2 :
Figure 2: The NSWs layout: a) the large and the small NSW sectors b) View of a sector with two sTGC and two MM quadruplets on both sides of the spacer frame c) dimensions of a wedge

Figure 4 :
Figure 4: The NSW electronics: trigger path of the sTGCdetecors (up), trigger path of the MM detectors (middle), and data acquisition path of both the MM and the sTGC detectors (down).

Figure 7 :
Figure 7: The spatial resolution of the coordinate for perpendicular tracks on the SM1 (left) and the LM2 (right) module.

Figure 8 :
Figure 8: The values expected for the hit occupancy (%) of SM1 and LM2 modules as a function of the radiation rate.(The difference of the hit occupancy values of SM1 and LM2 are too small to be visible).

Figure 9 :
Figure 9: Hit occupancy versus the inverse attenuation of the source's intensity for all four detection layers of (a) the SM1 module (IP-1 to IP-4), (b) the LM2 module (HO-1 to HO-4), and (c) the Tz chamber.

Figure 10 :
Figure 10: Hit occupancy, for the SM1 module, as a function of the radiation rate using different high voltage values.