Analysis of discharge events in the CMS GE1/1 GEM detectors in presence of LHC beam

In July 2022, the experiments installed on the Large Hadron Collider (LHC) accelerator ring started a new data taking phase, Run-3. Before this period, an upgrade campaign took place during the so called Long Shutdown 2 (2018–2022). In particular, the muon system of the CMS experiment has been upgraded with the installation of a new gas detector station, GE1/1, based on Gas Electron Multiplier technology (GEM). The CMS experiment has scheduled the installation of two additional GEM stations: GE2/1 and ME0. The aim of the GEM stations is to maintain the performance of the muon system, reached during the last data taking phase (Run-2), with the increase of instantaneous luminosity expected at the LHC. The installed GE1/1 station covers the pseudorapidity region 1.55 <|η|< 2.18. GE1/1 participated in the CMS Run-3 data taking from July to November 2022 and, in this period, its detectors were exposed for the first time to the radiation produced during the collisions of the LHC beams, with a center of mass energy of 13.6 TeV. Since the first days of Run-3, where only a few bunches collided in CMS, the GE1/1 detectors started to experience a significant number of discharges, affecting their smooth operation during the data taking. In this talk, we present an analysis of discharges, which started by simply counting the number of discharges occurring per detector. An interesting phenomenon observed was that the discharge rate can vary a lot among the installed detectors; this is due to the fact that the occurrence of discharges is driven by the manufacturing of the GEM foils used in the detectors. Imperfections in their manufacturing can lead to large variation in the discharge rate. In addition, the rate can vary in time, due to the fact that discharges can produce damage at the site generating them, deactivating a particular hole or, in the worst case, producing a short circuit in the GEM foil, deactivating a part of the foil amplification region. We present the evolution of the discharge rate in time and its dependence on the HV working point and on the luminosity delivered by the LHC beams. This contribution illustrates the actions taken to mitigate the discharge rate and to assure a smooth detector operation.


The CMS GE1/station
During the Long Shutdown 2 of the Large Hadron Collider (LHC), the CMS experiment was upgraded by the installation of 144 detectors based on the Gas Electron Multiplier technology, composing the GE1/1 detector station.This is the first of three GEM stations to be installed in CMS (GE1/1, GE2/1 and ME0).Their purpose is to maintain the performances of CMS in the future High-Luminosity phase of the LHC.GEM detectors are indeed able to sustain a high radiation flux and the purposes of their installation are to increase the redundancy of the muon spectrometer and thus keep under control the trigger rate in the endcap region.The ME0 station will in addition extend the pseudorapity coverage of the muon spectrometer up to || = 2.8 [1,3].
Each of the GE1/1 detectors provides an azimuthal angular coverage of 10 • , 72 per each CMS endcap, installed in a staggered fashion and organized in pairs in a module called Super-Chamber.Two kind of detectors were manufactured: short and long detectors, covering respectively the pseudorapidity regions 1.61 < || < 2.18, and 1.55 < || < 2.18.

High Voltage distribution for GE1/1 detectors
In GE1/1 detectors, the multiplication of the signal produced by ionising particles crossing the gas medium, is provided by three GEM foils stacked one above the other.The multiplication of primary ionisation electrons inside the GEM holes is enabled by the intense electric field present in the holes, generated by applying a high potential difference between the top and bottom face of the foil.The board used for the application of High Voltage (HV) is CAEN A1515-tg [4].The board channels used to generate the potential difference applied between the two faces of a given foil are called Top, while those used to generate the electric field for transport of electrons inside the gas gaps are called Drift and Bot, as explained in ref. [5].
Each A1515-tg board is equipped with 14 HV channels and the design of HV power system consists in powering both detectors (Layers) of each Super-Chamber using the same group of 7 HV channels, splitting the cable going from the board to the detectors.This design implies that the same channel of the board powers the same electrode of both connected detectors.Due to the fact that A1515-tg is a multichannel board where the voltage of each electrode can be set independently, it was chosen to use the current  eq flowing in the reference resistor divider, presented in [6], to easily identify -1 -

JINST 19 C02003
with a single number the set of 7 voltages to operate the detector.The proportions between the voltages are indeed set by the reference resistor divider, used during the detector quality control phases.
During their operation, GEM foils can develop damages, such as the generation of short circuits, consisting in an electrical connection between the top and bottom face of the GEM foil.This connection leads to a drop to 0 V the potential difference applied between the two faces, preventing the multiplication of ionisation electrons.To avoid the complete deactivation of the foil in case of occurrence of such an event, the top face of the foil has been segmented in 40 (47) sectors in short (long) detectors and the scheme used to distribute the voltage from the power supply to each GEM foil is illustrated in figure 1.On the other hand, the bottom face of the foil is a unique copper plane, not segmented.In this way, the generation of a short circuit in one HV sector leads to: • the deactivation of multiplication just in the affected sector; • the decrease of the voltage actually applied to the sectors, because of the current flowing in the resistors.
Figure 1.HV distribution scheme adopted to power GEM foils in GE1/1 detectors.

Discharges in the presence of LHC beam collisions
The operation of GEM detectors in the presence of LHC beam during the Run-3 data taking period started in July 2022, and the occurrence of discharges in GE1/1 detectors was observed from the very beginning.The discharge is due to the breakdown of the gas medium, in particular inside a GEM hole, due to the presence of a big charge during the multiplication of primary ionisation electrons.This can indeed be generated by highly ionising particles traversing the gas medium, such as neutrons.
In addition, the presence of background radiation crossing the detector leads to the flow of a current on the foil, whose intensity varies with the luminosity of the colliding beams, as showed in figure 2. On top of this baseline current, there are current spikes which identify the occurrence of discharges.
The occurrence of discharges is dangerous both for the potential of creating damages in the front-end electronics and for the possibility of generating damages in the GEM foil holes.In extreme cases, a damage in a GEM hole leads to the production of a short circuit.Because of these two main reasons, the rate of discharges has to be carefully monitored during the detectors' operation.
The rate of discharges can vary first of all from detector to detector, because the intense localized electric field enabling the breakdown of the gas medium is favoured by the presence of imperfections in the manufacturing of the GEM foils.This is well illustrated by figure 3.
-2 -  -3 - In addition, it's also useful having a view of the distribution of the amplitude of the spikes of currents observed in correspondence of discharges per each electrode.Figure 4 illustrates as the electrodes more prone to the production of discharges are those powering the foils (Top electrodes).the baseline current of each electrode, in a given example LHC fill.Some parameters describing the properties of the LHC fill are reported in the top bar of the plot, such as the number of bunches colliding in CMS (Bunches), the single beam energy (Energy), the total luminosity delivered by the LHC during the fill (DelLumi), the peak luminosity (PeakLumi), the peak and average pileup (respectively PeakPU and AvgPU), the rate average rate of luminosity delivered per hour (DelLumiPerHour) and the timestamps for the start and end of the data considered.In addition, the parameters Endcap_P and Endcap_M identify the HV configuration correspondent to the current  eq flowing in the reference resistor divider, adopted respectively for detectors installed in the negative and in the positive endcap.The label on the bottom right of the plot identifies the number of single detectors powered by HV (Layers) and the number of independent HV channels used to power the detectors (Pairs).

Short circuits in GE1/1 detectors
The main contexts of generation of short circuits from operation of the detectors, since their installation in CMS, proved to be GEM foil HV training, CMS disk movement, ramp of CMS magnet and generation of discharges in the presence of LHC beam collisions.To mitigate the generation of shorts in the presence of magnetic field variation a dedicated procedure was adopted, illustrated in [5].It was observed as shorts can be both generated and heal in time: the evolution in time of the number of GEM foils in GE1/1 detectors, with at least one HV sector affected by a short circuit, is illustrated in figure 5.It can be clearly noticed as the foil with the highest number of shorts is GEM3, the last in the avalanche multiplication of the primary ionisation charge, with around a factor 2 of affected foils respect to GEM1 and GEM2.This higher charge involved in the process could provide indeed enough energy to a discharge for the generation of a short, but a thorough investigation is ongoing on this.

Evolution of discharge rate per hour per chamber during operations
GE1/1 operation during first months of Run-3 showed an important evolution in time, illustrated in figure 6.The plot illustrates the strategies adopted during the operations to mitigate the discharge rate, consisting in the variation of the working point and the temporary deactivation of the HV electrode closest to the readout plane (G3Bot), meant to protect the electronics during this investigation phase.The investigation showed a clear higher discharge rate in correspondence of higher working points, due to the higher detector gain.In addition, it was observed both sudden increases of the rate in correspondence of increase of the number of colliding bunches, but also a gradual decrease of the rate in time at fixed gain and number of bunches.This last effect should be due to the conditioning of the GEM foils, by removing imperfection of GEM holes, or their deactivation or decrease in gain by modification of the hole shape by effect of discharges occurrences.

Conclusion
The GE1/1 operation in Run-3 showed the need of a thorough study on the phenomenon of discharges occurrence in the presence of beam collisions, correlated to the extreme event of short circuit generation in the GEM foil.The study allowed to tune the HV working point at a value  eq = 690 µA, stabilizing the discharge rate around 1-2 discharges per hour per chamber at the maximum, with 2400 colliding bunches.

Figure 2 .
Figure 2. Current flowing in the power supply channel (G3Top) powering the third foil of a GE1/1 detector in the presence of collisions of LHC beams.The right plot is a zoomed view of the left one, clearly showing the evolution of the baseline current, reproducing the trend of luminosity of colliding LHC beams.

Figure 3 .
Figure 3. Rate of discharges per hour, observed in a given LHC fill in each GE1/1 detector.The same number of discharges is observed for both layers of a given Super-Chamber, since they are powered by the same cable and because of this the two discharges of the two layers cannot be disentangled.To count the discharges, in this case a cut of 2 µA. on the amplitude of the discharge respect to the baseline current was applied.

Figure 4 .
Figure 4. Number of discharges counted per electrode as function of the amplitude of the discharge respect to the baseline current of each electrode, in a given example LHC fill.Some parameters describing the properties of the LHC fill are reported in the top bar of the plot, such as the number of bunches colliding in CMS (Bunches), the single beam energy (Energy), the total luminosity delivered by the LHC during the fill (DelLumi), the peak luminosity (PeakLumi), the peak and average pileup (respectively PeakPU and AvgPU), the rate average rate of luminosity delivered per hour (DelLumiPerHour) and the timestamps for the start and end of the data considered.In addition, the parameters Endcap_P and Endcap_M identify the HV configuration correspondent to the current

Figure 5 .
Figure 5. Evolution of the number of GEM foils with at least one HV sector affected by one short circuit, from the beginning of Run-3.

Figure 6 .
Figure 6.Evolution of the discharge rate during the first two months of Run-3 operations.