Study of discharges in the CMS GEM GE1/1 station with LHC beam

The Large Hadron Collider (LHC) and the experiments installed there are being upgraded for the High Luminosity LHC project that will provide a greatly enlarged data sample in the search for physics beyond the Standard Model. During Long Shutdown 2, from 2018 to the first months of 2022, some of these upgrades were completed. In particular, the muon spectrometer of the CMS experiment was upgraded with installation of the GE1/1 detector station, based on Gas Electron Multiplier (GEM) technology. This station is positioned in the endcap region of the CMS muon system and covers the pseudorapidity range 1.55 < |η| < 2.18. On the 5th of July, 2022, Run 3 of the LHC began with collision energy of 13.6 TeV. The newly installed GEM detectors experienced high-voltage discharges when the beams were colliding inside CMS. To monitor these events and to understand how to safely operate the detectors, a study of these phenomena became necessary. The intensity of discharge current and the discharge rate were recorded in different detectors as a function of luminosity and time. It was observed that some chambers have a significantly higher discharge rate than the modal value of rate over the entire chamber population. The reasons responsible for this behavior are currently under investigation. In this contribution, an analysis of discharges will be presented, illustrating the dependence of discharge rate on different LHC beam configurations and the high voltage working point of the chambers.


Introduction
During the Large Hadron Collider (LHC) Long Shutdown 2 (LS2) phase, the muon spectrometer of the CMS experiment was upgraded with the installation of the GE1/1 station (1.55 < | | < 2.18). GE1/1, based on Gas Electron Multiplier technology, is designed to maintain the L1 trigger performance with the increase of luminosity planned for the High Luminosity LHC phase. That is done by exploiting the measurement of the bending angle of the muon trajectory between the GE1/1 detectors and the nearby Cathode Strip Chamber (CSC) detectors. To allow full azimuthal coverage, two kinds of GE1/1 detectors have been produced: short chambers (1.61 < | | < 2.18) and long chambers (1.55 < | | < 2.18), installed in CMS in a staggered fashion [1,3].

Design of HV distribution
GE1/1 detectors are triple-GEM detectors, made of three foils stacked on top of each other. To generate the electric field needed for the ionization electron multiplication, High Voltage (HV) is applied between each pair of surfaces in the detector stack, using CAEN boards A1515-tg [4]. The A1515 power supply channels feeding a potential difference between the two faces of a GEM foil are called Top electrodes, while those generating a potential difference in a gas gap are called Drift or Bot electrodes, as discussed in [5].
If a foil is damaged by a short circuit, a connection is created between its top and bottom faces and the potential difference applied on the whole foil would drop to zero, disabling completely the amplification. To prevent this, the top face of each foil has been segmented in a number of sectors: 40 for the short chambers, 47 for the long ones. In this way, when a short circuit is created in one sector, a high current flows through the protection resistors, but the unaffected sectors can still be powered. Then, due to the voltage drop across the protection resistors, a higher voltage must be provided by the power supply to reach the desired potential difference on each undamaged foil sector.
Finally, the set of voltages applied on the 7 electrodes is identified by the current eq flowing in a reference resistor divider ( = 4.7 MΩ) as discussed in [6].

Operations with LHC beams
In July 2022, CMS began operation in Run-3, collecting physics data produced by the collision of the LHC proton beams, at an energy √ = 13.6 TeV. Since the first LHC fills, the HV system started to experience frequent protection turn-offs (HV trips). The cause was discovered to be discharges inside the detector, which exceeded a current limit ( 0 threshold) of 0 = 2 μA for a time longer than tolerance time trip = 1 s.
Since then, analysis of the discharge rate was used to develop an HV configuration to use while the number of colliding bunches in the LHC was increasing. The discharge rates are shown in figure 1. For each point, the HV configuration used during LHC fill is shown. Higher discharge rates are observed for higher eq , but also sudden rises in the rate are observed when the number of colliding bunches increases. It was also observed that at a fixed eq with a fixed number of colliding bunches, the discharge rate gradually decreases and then stabilizes to a lower value.

Conclusion
The GE1/1 system faced a challenging period of operation at the beginning of Run-3. Due to the occurrence of discharges in GE1/1 detectors, the 0 threshold was raised to 0 = 10 μA (for each of the 7 electrodes needed to power the detector), to achieve stable operation, avoiding frequent HV trips which characterised the LHC fill 7923, the first one with 8 colliding bunches in CMS. Finally, the discharge rate was stabilized and the detectors were powered in the HV configuration with eq = 690 μA.    depending on the luminosity of the LHC beam. Higher discharge rates are observed for higher eq , but also sudden rises in the rate are observed when the number of colliding bunches increases. During the ramp in luminosity of the LHC, eq was varied to understand its impact on the discharge rate. During a period of time when the discharge rate was under investigation, the G3Bot electrode was turned off to protect the front-end electronics from discharge propagation. During the investigation, on 27th and 28th July, we adopted eq = 600 μA, to observe discharge rate for the standby configuration of the detector. Once the discharge rate and LHC luminosity stabilized, the detectors were powered in the configuration with eq = 690 μA.
In addition, the plot shows how the electrodes more prone to exceed the 2 μA current value during a discharge event are those powering the two faces of a foil, the Top ones. Finally, in these two months the number of GEM foils affected by short circuits moved from 30 to 37.