ATLAS Muon Spectrometer Upgrade for the HL-LHC Era’s Challenges

Abstract

The High-Luminosity Large Hadron Collider (HL-LHC) project aims to improve the performance of the LHC by increasing the proton–proton collision luminosity. New physics discoveries will be possible starting in 2027. The HL-LHC aims to improve the integrated luminosity by a factor of 10 concerning the current running LHC’s design value. The HL-LHC project foresees delivering proton–proton collisions at 14 TeV CM (Center of Mass) energy providing the integrated luminosity to a value of 3 ab⁻¹ for the ATLAS and CMS experiments, 50 fb⁻¹ for LHCb, and 5 fb⁻¹ for ALICE. The increased integrated luminosity for the above LHC experiments will provide the potential to discover rare processes while improving these measurements’ signal-to-noise (S/N) ratio statistics. The ATLAS muon spectrometer has been upgraded to face the challenges of the luminosity at the HL-LHC run. The new sub-detectors are as follows: The New Small Wheel (NSW) has replaced the Cathode Strip Chambers (CSC) discs at the internal part of the ATLAS end cups. The new integrated small Monitored Drift Chambers (sMDT) with the Resistive Plate Chambers (RPC) are installed at the outer end of the ATLAS BI (Barrel Inner) layer, in the barrel–endcap transition region, at 1.0 < |η| < 1.3, where η is the pseudo-rapidity (pseudo-rapidity η is a commonly used spatial coordinate describing the angle of a particle relative to the beam axis, defined as $\eta =-ln\left[tan\left({\displaystyle \frac{\theta }{2}}\right)\right]$, where θ is the angle between the vector momentum $\overrightarrow{p}$ and the positive direction of the beam axis). The NSW is an innovative technological achievement, including the MicroMegas (MM) gas detectors in large areas and small-strip Thin Gap Chambers (sTGC), enabling high p_T (high p_T is the high value of the particles’ transverse momentum versus the beam collision axis) trigger and muon detection. The muon reconstruction, the background rate, other spectrometer parameters, and the NSW performance are also presented.

1. Introduction

The ATLAS experiment [1] at the Large Hadron Collider (LHC) concludes a set of multi-purpose particle detector subsystems of cylindrical geometry located at one of the LHC interaction points (Figure 1).

The ATLAS experiment was designed and constructed to exploit the entire physics potential of the LHC energy era, including the discovery and spectroscopy of the Higgs boson plus the research for physics beyond the Standard Model (SM). The highest, ever proton–proton collision energy at 14 TeV with the highest luminosity during the HL-LHC running will allow for the discovery of new physics effects at the TeV scale. The ATLAS experiment is divided into three basic detector subsystems: the Inner Detector (Pixel, SCT, TRT), inside a solenoid magnet of 2T magnetic field; the Calorimetry complex (Liquid Argon and Tile); and the muon spectrometer (MS), containing the unique and largest superconducting toroidal magnet, providing an average magnetic field of 0.5 T. The ATLAS is the largest detector ever built in HEP collider experiments, extending to overall dimensions of 44 m in length and 25 m in diameter.

The ATLAS muon spectrometer (MS), capable of detecting and tracking muons with high spatial resolution, is the largest muon spectrometer ever constructed [2]. The MS provided the four muons detection for the Higgs boson discovery during the 2012 run along with other major measurements for Standard Model (SM) research. The placement of the MS modules follows the superconducting toroidal magnet eight-fold azimuthal symmetry along the z-axis. Each barrel sector, of the eight in total, has three muon station names, respectively, inner, medium, and outer. Both the MS endcaps have three disc-shaped muon stations, starting with the innermost muon disc at z = ±7.4 m [3].

The LHC Run 3 (a new data-taking period, from July 2022 for experiments at the Large Hadron Collider (LHC), after more than three years of upgrade and maintenance work) is expected to have an increased luminosity by a factor of 2, compared to the nominal value, and it is underlined that during the HL-LHC run, the expected luminosity will be increased by a factor of 10 with respect to the current running LHC’s design value. The ATLAS muon spectrometer was upgraded during phases I and II, to cope with the increase in background rates, while still maintaining the high efficiency for tracking and triggering. The Cathode Strip Chambers (CSC), composed of multi-wire proportional chamber technology and Monitored Drift Tubes (MDT), being mounted to the Muon Small Wheel (MSW) located at the endcaps of the detector [4], had to be replaced with two New Small Wheels (NSW side-A and side-C). Additionally, upgraded small-diameter Monitored Drift Tube (sMDT) chambers, small-strip Thin Gap Chambers (sTGC), and a new layer of thin-gap Resistive Plate Chambers (RPC) were designed and made ready for installation to integrate the MS upgrade phases I and II.

The technical specifications and the relative performance of the above novel detectors for the MS upgrade are presented in the next paragraphs, in addition to the improved potentiality of the MS muon track reconstruction, the suppression of the background rates, and the foreseen physics analysis from the LHC run 3 and thereafter.

2. LHC Upgrade to HL-LHC

The aim of the High-Luminosity (HL) upgrade of the LHC is to deliver an integrated luminosity of at least 250 fb⁻¹ per year, considering a run duration of a minimum of 160 days of LHC operation [5], in each of the two high-luminosity general-purpose detectors, ATLAS and CMS [6]. The HL-LHC timeline and the ATLAS and CMS upgrade phases I and II are presented in Figure 2 [7].

The ambitious goal of the entire HL-LHC operation is to extend the luminosity from a few hundred fb⁻¹ to 3000 fb⁻¹ by its commissioning in 2027.

The ATLAS and CMS [6] detectors are under upgrade to handle the increased pile-up and the number of events per bunch crossing, of at least 140 and, ultimately, 200. The corresponding values of the instantaneous luminosity are expected to be approximately 10 × 10³⁴ cm⁻² s⁻¹ [5,8], for operation with 25 ns beams consisting of 2760 bunches at 7 TeV per proton beam, and for an inelastic cross-section of σ_in = 81 mb [9]. The detectors should be able to handle a peak line density of pile-up events of about 1.3 events per mm per bunch crossing and finally larger values with limited reduction of the detection efficiency [8,9].

The HL-LHC project [10,11] has been designed to achieve a peak luminosity as far as possible to 2 × 10³⁵ cm⁻² s⁻¹, considerably higher than the maximum luminosity imposed by the acceptable event pile-up. The LHC upgrade design foresees control of the instantaneous luminosity during the physics fill to accumulate the required integrated luminosity [9], for a performance efficiency of at least 50% [5,12], as obtained and exceeded during LHC Run 2 [12,13].

3. ATLAS Muon Spectrometer Upgrade

Figure 3 shows the 2D design of the quarter part of the muon spectrometer [14] in the y-axis perpendicular to the z-axis (beam direction). The barrel part is built up of three concentric cylindrical layers. The inner, middle, and outer layers are located at the y-axis at distances of about 5 m, 8 m, and 10 m, respectively. Each of the three layers consists of Monitored Drift Tube (MDT) chambers (shown in dark green) providing precision measurements for the muon momenta. The MDT chambers are equipped, at the middle and outer layer, with thin-gap Resistive Plate Chambers (RPC) (shown in red) providing a high-p_T muon trigger system, which operates out of a coincidence of all three layers.

At the endcap area, the MDT chambers are mounted on the barrel, perpendicular to the beam axis, whereas the Thin Gap Chambers (TGC) have been installed for the muon trigger. The muon system is embedded in a toroidal magnetic field provided by eight superconducting coils in the barrel region (η < 1.7) and two additional superconducting magnet systems in the endcap region.

The high- p_T muon trigger acceptance is limited to 72%, due to the absence of muon chambers in the areas located at η = 0, as shown in Figure 3, due to the space reserved for maintenance of the inner detector and the calorimeters [14]. In addition, the toroidal magnet coils prevent the muon chambers at the regions of η = 0.4, 0.75, and 1.0. The ATLAS muon spectrometer upgrade consists of two phases named phase I and II, respectively. Phase I, integrated during 2017–2021, is dedicated to the construction, installation, and commissioning of the NSW plus an initial part of 12 small-diameter Monitored Drift Tubes (sMDT) in the middle layer of the feet region of the ATLAS muon spectrometer.

After the upgraded phase I, the thin-gap Resistive Plate Chambers (RPC) provide, at |η| < 1.05, fast timing trigger signals for the muon identification, while the Monitored Drift Tube (MDT) chambers provide precision tracking. In the endcap, at 1.05 < |η| < 2.7, the muon trigger comes from Thin Gap Chambers (TGC) and the precision tracking from MDTs. The MDTs at the barrel area are in three concentric stations, each with a single tube resolution of 80 μm, plus three stations of RPCs on the middle and outer MDT stations. The MDTs are arranged around the beamline, covering 2π in alternating large and small chamber sizes, referred to as relative sectors. The Νew Small Wheel (NSW) [15] is in the innermost end cup region consisting of Micromegas (MM) and small-strip (sTGC) detectors.

In the upgraded phase II, to be integrated during the period 2024–2028, the non-negligible muon trigger rate should be prevented from charged particles not emerging from the pp interaction point, which has been observed in a small region between the barrel and endcap parts of the ATLAS muon spectrometer. Additionally, the installation of 16 new integrated chambers sMDT + RPC for muon tracking and triggering at the two ends of the toroid magnet coils in the small sectors of the barrel inner layer (BIS) was completed.

3.1. New Small Wheel (NSW)

The motivation for the New Small Wheel design, construction, and installation is to remove the background rate concern associated with higher luminosities during the HL-LHC era. The NSW will cope with the higher number of pile-up events per bunch, crossing large amounts of cavern background rates, up to 20 kHz/cm², to be induced, affecting a large |η| region of the muon spectrometer.

The NSW must ensure efficient tracking at a high particle rate up to L = 5 × 10³⁴ cm⁻²s⁻¹ and large |η|, with a position resolution of <100 μm, and additionally, it will be integrated into the Level-1 trigger [15].

The NSW, with a disc detector of 10 m diameter, consists of eight large and eight small sectors (wedges), shown in Figure 4 [16]. Each sector has a spacer in the middle, where one double Micromegas (MM) wedge (back-front each one) is mounted, covered by 2 sTGC wedges. The NSW wedge (MM and sTGC) is a detector quadruplet. The MM detectors were designed and constructed, for the first time, on O(m²) dimensions, for their installation at the ATLAS experiment. The total NSW area is about 2400 m², whereas the granularity for each sub-detector technology with the number of channels, MM: ~2.1 M, sTGC: ~280 k (strip) + 46 k (pads) + 28 k (wires) pics. The NSW meets the following requirements:

• 15% p_T resolution at ~1 TeV;
• 97% segment reconstruction efficiency for muon p_T > 10 GeV with 30 μm spatial resolution;
• Segment measurements with up to 1 mrad angular resolution for the trigger;
• 50 μm spatial resolution per layer;
• Alignment precision readout elements < 30 μm.

The MM and sTGC are considered as the main detector technologies of Micro-Mesh gas-filled structures [17], and both measure the effects of the drifting electrons and ions separated by muons ionizing in the gas. Both technologies present different characteristics:

• Small-strip Thin Gaps (sTGC) comprise two cathode boards with 25 gas gaps that contain gold-coated tungsten wires between the cathode boards filled with CO₂+n-Pentane (Figure 2).
• MicroMegas (MM) consists of a planar electrode structure with a thin metal micromesh electrically dividing the gas volume into two parts: a conversion/drift gap (5 mm) and a very thin amplification gap, 120 μm thick. The final choice selected is an operating ternary gas mixture (non-flammable) of Ar:CO2:iC4H10 (93:5:2) vol%.

Time roadmaps for the two NSW detector (A and C side) designs, production, integration, commissioning, and installation in the ATLAS cavern are presented in Figure 5 [18].

3.1.1. The ATLAS Micromegas Detector

The ATLAS Micromegas (MM) operation is based on the charged particle detection via ionization of the MM’s gas atoms, producing about 100 electron or muon pairs/cm for minimum ionizing particles. The concept of Micromegas detectors was invented in the 1990s [19] and was further developed in the following years [20]. An R&D program started in 2007, to investigate the potentiality of this detector technology for its use in the harsh LHC environment, aiming for the ATLAS muon system upgrade plans. The major problem of this effort was the large size of the MM detectors and the subsequent desired minimization of discharges that required a detector design with a surface of several square meters. A solution for the spark problem was to add a layer of resistive strips above the readout strips separated by a thin layer of insulator [21]. The solution adopted for the large-area MM detector construction is described in the ATLAS NSW Technical Design Report [17]. Performance studies of the anode strips protected with the resistive-strip bulk Micromegas technology have been investigated with radioactive sources and high-energy beams to suppress the potential discharge effect [22].

The produced free electrons in the drift gap pass through the grounded micromesh presenting a transparency of 95%. These electrons form avalanches in the region between the micromesh and anode, the amplification region, due to the high electric field, as shown in Figure 6. The dashed line presents the grounded micromesh dividing the MM gas area to the Conversion/Drift Gap of 5 mm size and the Amplification Gap of 128 μm size. The dotted arrows at the left and the solid line at the right of the plot show the applied electric field direction. The applied high voltage on electrodes and the unequal distances between electrodes and micromesh determine the electric field in the two regions. Readout electronics collect the induced charge on anode strips from the moving electrons. Quick evacuation of the produced ions is achieved through the mesh, allowing high-rate capabilities.

The ATLAS Micromegas (MM) detector has a surface of 2–3 m² size with four active volumes. Each volume has three planar structures: cathode, grounded mesh, and segmented anode. The drift gap region is 5 mm with an electric field E = 480 V/cm, and the amplification region is only 0.12 mm with an electric field of 43 kV/cm. The MM operates in proportional mode, i.e., primary electrons multiplied with a gain of ~5000–10,000, with capacitive coupling between resistive and readout strips. The detector dead time is very short, almost negligible. A position reconstruction method was developed for inclined tracks, called the “micro-TPC method” (μTPC); the impact of the unavoidable presence of pillars and the relative alignment of readout and resistive strips on the Micromegas performance has been quantified. The Micromegas detectors will also contribute to the trigger system in ATLAS, to study its time response.

In the μTPC method, the measured time t_i (Equation (1)) is used for each hit at position z_i of the primary ionization in the drift space. A linear fit of z_i as a function of the strip position x_i defines the trajectory of the track, then the track angle θ and the x-coordinate of the track at the middle plane of the drift space x_half. This method provides the best procedure and x_trk value for inclined tracks.

𝑧_𝑖 = 𝑣_d (𝑡_i − 𝑡_0,i)

(1)

where t_0,i is the t₀ of the strip i, corresponding to the time of a primary ionization being very close to the mesh. The t₀ values are independent of i, for good accuracy. An example of an event display of a track with an incidence angle of 40° is shown in Figure 7 [22].

Test measurements obtained by the Micromegas (MM) prototype have proved the agreement of that small effect measured for perpendicular tracks, with relative simulation results. The MM performance is not affected significantly for inclined tracks by the pillars. The possible rotation between resistive and readout strips was measured to be ≤20 μm, with a negligible contribution to the spatial resolution. The time resolution for the hits in a cluster is obtained as a function of the track impact angle in a range (11–20 ns) [22].

The gas leak system of the NSW MM detectors, developed and implemented during the production period, ensures the absence of leaks, which is important for MM’s performance. In particular, the novel Flow Rate Loss (FRL) method was introduced for the gas tightness of the MM quadruplets and wedges [23]. The FRL is a reliable and precise method, also, for the variable gas volume, with tested and verified performance at CERN following extensive end-to-end validation tests.

The MM prototyped baseline devices and the portable standalone ones were operationally stable during the integration and commissioning period, in terms of gas tightness. The obtained overall final statistical distribution of the leak rate of MM quad modules results in a Gamma-shaped distribution according to the operated parametrization presenting a relatively large rms deviation, which is based on the conditions of the barometric pressure impact and the laboratory seasonal temperature variations. A precise and low-cost percent oxygen monitoring technique, based on Oxygen Probe Analyzers (OPA), can sign the drop-off of the MM wedge’s performance. It is underlined that the implementation of the above FRL and OPA methods and prototype setups have been performed for similar gas detectors and other applications [23].

A standard variation in sequential QA/QC procedures was implemented to all NSW MM wedges during integration. The final performance operation test was measured in a Cosmic Ray Stand (CRS), where couples of single MM wedges were tested to be accepted for installation at the NSW detector system at CERN.

The MM wedges operate at about 3 mbar, low static gauge pressure, with an upper typical limit of 10 mbar. Any higher pressure may cause deformations due to volume expansion and subsequent damage risks. The essential protection of the MM wedges against any unexpected gauge pressure increases during their operation test at CRS and their installation at NSW was foreseen. A differential safety mechanism (DSM) was designed to protect the MM wedges against unexpected and sudden increases in the static gauge pressure in the volume [24]. The DSM provides an asynchronous digital sequential system with an adjustable voltage comparator and many digital logic circuits for driving a three-way solenoid valve at the correct time, interrupting the pressure increase, and maintaining the wedges to the atmospheric pressure levels. The DSM is a low-cost, reliable solution with a differential safety mechanism, for the MM wedges, acting rapidly to any abnormal event or intervention in the input or output of the gas line [24].

3.1.2. Small-Strip Thin Gap Chambers (sTGC)

The Thin Gap Chamber is a gaseous detector with a geometrical configuration of a multi-wire proportional chamber operated in a limited proportional mode. The anode wires and anode strip signals are placed in orthogonal directions, providing two-dimensional position measurements of the impinging particle [25]. The TGC detector [17] is used for the muon trigger system in the endcap regions of the ATLAS experiment. The chamber is made of a plane of gold-plated tungsten wire anodes sandwiched by two cathode planes, as shown in Figure 8 [26].

The detector gaps are filled with a CO2 and n-pentane gas mixture, with a volume ratio 55:45, respectively. The high voltage, usually 2800 V, applied between the cathode and anode applies a strong electric field. The gas molecules are ionized by energetic charged particles passing through the detector and the produced primary electrons and ions drift along the electric field lines to the anode and cathodes, respectively. The electric field is mainly concentrated around the anode wires, and the primary electrons will be multiplied in an electron avalanche as they approach the anode wires due to the strong electric field around them. The motion of the electron-ion pairs produced in the avalanche induces a signal in the cathode and anode according to the Shockley–Ramo theorem [27].

The cathodes include two planes. A graphite–epoxy layer coated on the inner face of the external drift panel faces the wires and serves for slow charge evacuation. The resistive layer and cathode–anode gap is an equivalent RC circuit that limits the speed of charge dispersion to the ground, reducing the intensity of occasional discharges. The resistivity of the graphite is in the order of 100 kΩ to MΩ. On the external side, a copper plane is segmented in pads or strips, working as signal-pickup electrodes [28].

An sTGC quadruplet is the four sTGC modules stacked together. Different logic coincidence scheme implementations optimize the efficiency and background rejection. The centroid of the strip clusters derives precise location information. The total segment information is forwarded to the back-end trigger electronics for integration with the trigger from other muon detectors. The hit information from pads, wires, and strips is also used for the offline muon track reconstruction. Three sTGC quadruplets are mechanically assembled to make wedges of two different types (large and small) depending on the sTGC module size.

The block diagram of the TGC electronics for the phase II upgrade is shown in Figure 9 [25]. The TGC signals are amplified, shaped, and discriminated by the amplifier shaper-discriminator (ASD) boards and fed into the PS board (front-end board).

The TGC electronic system is responsible for the endcap muon trigger of ATLAS in the region 1.05 < η < 2.4, upgrading the readout and trigger electronics for the HL-LHC. The TGC system involves many front-end electronics (1’434) with timing signals distributed via optical fibers and reconstructed, individually. Effective bunch-crossing identification in TGC trigger electronics requires clock tuning well below O(1) ns. A methodology was developed for measuring the clock phase for individual reconstructed signals across the entire system and aligning the clock phase remotely and automatically in situ.

The reproducibility of fixed latency clock distribution is an accuracy of ~50 ps. All systematic skew observed by clock phase adjustment or considered during clock phase measurement is monitored with the measurement method. It is concluded that a clock phase adjustment of 1’434 front-end electronics can be performed with an accuracy of ~700 ps, which is sufficient for the ATLAS trigger requirement.

3.1.3. Trigger System for the NSW

A system of 128 detectors is included in the NSW trigger electronics and DAQ dataflow, leading to about 2.4 million readout channels and separate trigger, readout, and configuration/monitoring paths for both sTGC and MM detectors. The NSW trigger electronic system is divided into the on-detector electronics front-end boards (FEB) with radiation-tolerant Application-Specific Integrated Circuits (ASICs), etc., and the off-detector electronics at the cavern area, with Data Driver Cards, Routers, Read Out Drivers (ROD), Detector Control System (DCS), etc. [29,30]. A block diagram of electronics for the sTGC trigger (green color frame) and MM trigger (red color frame) procedure with the special pathed of the dedicated electronic units is presented in Figure 10 [31].

The ATLAS Local Trigger Interface (ALTI) runs the control software on a single-board computer (SBC). The primary function of ALTI is to provide the interface between the Level-1 Central Trigger Processor (CTP) and the TTC optical broadcasting network to the front-end electronics for each of the ATLAS sub-detectors. The Central Trigger Processor is responsible for making the initial trigger decision (Level 1 trigger accept signal) by identifying interesting particle candidates coming from the Level-1 calorimeter and Level-1 muon trigger systems, reducing the event rate to a maximum of 100 kHz [31,32].

The ALTI, a custom-made 6U VME64x module, provides, for test purposes, an artificially generated pulse pattern, which contains the TTC information. In addition, ALTI provides the bunch-crossing clock, the main timing signal coming from the LHC, at a frequency of 40 MHz for data synchronization [32].

Two sTGC trigger levels were introduced as follows: (i) the pad trigger, fast coarse information and seeds of strip trigger, deployed in the 2023 acquisition campaign, and (ii) the strip trigger, which reconstructs strip clusters and uses them for precise measurement of the segment angles. The latter is under commissioning needed for the HL-LHC. The merged MM and sTGC trigger segments are forwarded to the sector logic, Figure 11 [32]. Finally, an independent MM trigger was introduced to be deployed in 2024.

3.2. Small Monitored Drift Tube (sMDT)

The high-luminosity upgrades of the LHC of neutrons and gamma rays background rates of the order 14 kHz/cm² are expected to exceed the rate capability of the current ATLAS precision muon tracking Monitored Drift Tube (MDT) chambers, with an initial drift tube diameter of 30 mm. Smaller-diameter Drift Tube (sMDT) Chambers with 15 mm tube diameter were developed for upgrading the ATLAS muon spectrometer (MS).

About 96 sMDT chambers are ready for installation at the phase II upgrade. The BIS1-6 chamber layout and envelopes have been defined with new 96 thin-gap RPC muon trigger chambers, as presented in Figure 12 [33]. Each sMDT chamber has 500 tubes, and each tube has a diameter of 15 mm diameter and a length of 1.5 m. An additional 14 similar sMDT chambers of the same type were built for additional regions of the ATLAS MS and were operational during LHC Run 2. Sixteen similar chambers were integrated for the ATLAS MS BIS78 upgrade area. The BIS78 (Barrel Inner Small MDTs, sectors 7 and 8) chambers will replace the existing Monitored Drift Tubes (MDT), used for the muon measurement in this area, with muon chambers of integrated smaller-diameter tubes (sMDT) and a new generation of thin-gap RPCs, capable of higher rates, providing a robust standalone muon trigger confirmation. The existing BIS7 and BIS8 MDT chambers will be replaced by 16 new smaller-diameter (sMDT) chambers and two thin-gap Resistive Plate Chamber (RPC) triplets, for the phase II BI upgrade.

An sMDT prototype chamber was operational in a muon test beam at CERN and with cosmic muons at high background irradiation rates of up to 95 kHz/cm² and 1400 kHz/tube. The test results provide the required track reconstruction efficiency and spatial resolution of the sMDT chambers at background rates well beyond the maximum expected values at high-luminosity LHC. The sensing wire locations in the prototype chamber were verified to within a few microns of precision with cosmic rays with a wire positioning accuracy of <20 μm [34]. The designed sMDT chambers’ track reconstruction efficiency and spatial resolution at background rates well beyond the HL-LHC maximum expected values are demonstrated by the test results presented in this paper [34].

The sMDT chambers were developed with standard industrial aluminum tubes with 0.4 mm wall thickness. The assembled drift tubes passed standard QA-QC procedures for gas tightness, high-voltage stability, and wire tension using semi-automated facilities in a temperature-controlled clean room at a typical rate of 100 tubes per day. A new chamber assembly method was developed for chamber gluing in one working day. The sMDT chamber end-plug design is more reliable than the initial MDT chamber one, providing higher wire positioning accuracy [34].

In Figure 13 (left), a full sMDT prototype chamber constructed and tested is shown, in a muon beam at CERN and with cosmic muons at high background irradiation rates of up to 95 kHz/cm² and 1400 kHz/tube, and in Figure 13 (right), the upgraded trigger and readout chains for both MDT and sMDT chambers are shown. The MDT on-chamber electronics provide all muon hit signals to the ATLAS counting room [35].

The sMDT chamber operating parameters of the ATLAS MS remain unchanged. The chamber assembly time was reduced to one working day and the sense wire positioning accuracy improved by a factor of 2 to <10 μm [36].

The sMDT chambers to be installed in MS regions provide much higher rate capability, contrary to the large footprint of initial MDT chambers. The sMDT chamber replacement has space for additional RPC trigger chambers [36]. The wire positions in the end plugs are measured within a few microns of precision by a coordinate measuring machine. The automated Coordinate Measurement Machine (CMM) is used to measure the position of each end-plug external reference surface and, therefore, of each wire deriving wire position accuracy, which locates the wires with better than 5 μm precision. Space charge effects degrading the spatial resolution are almost eliminated up to the maximum expected background rates. The resolution and muon detection efficiency are limited due to the current readout electronics and can be further improved [37]. The HV supply used for the dark current measurement has a sensitivity of 0.5 nA. The results of the HV leakage current measurement showed that, for most of the drift tubes, the HV leakage current was below the measurement sensitivity. The gas leakage rate of the chambers was tested by measuring the pressure drop over a maximum of 3 days for each multilayer. The gas leakage rate is required to be less than the value in Equation (2):

$$2N_{tubes}·{10}^{-8}{\displaystyle \frac{bar·l}{s}}$$

(2)

at a reference temperature of 20 °C, where 2N_tubes denotes the number of end plugs in the multilayer.

The ATLAS specification for the upper limit of the leak rate can be converted to a pressure drop per hour equal to 0.281 mbar/h. For this test, the chamber is filled with the nominal gas mixture Ar:CO₂ (93:7) at an absolute pressure of 3 bar. The pressure is measured using a baratron capacitance manometer once after the filling and a second time at least 24 h later, while the gas temperature is measured using temperature sensors placed in the chamber [38].

A trigger signal is provided by muons passing a scintillator placed above the chamber perpendicular to the tubes. Alternatively, a periodic trigger signal with a frequency of 800 Hz is used to measure the electronic noise. The trigger signal is sent to the TTC system with an internal clock of 40 MHz. The TTC system distributes trigger signals and control information through the CSM to the chamber front-end electronics, which starts the data acquisition. The CSM and mezzanine cards are programmed and configured via the Embedded Local Monitor Board (ELMB). The mezzanine cards used for the tests are equipped with the final ASD-6 chip design that is set to be used in the phase II upgrade.

Figure 14 shows the sMDT cosmic ray data used to determine the efficiency and resolution collected over at least 24 h with the nominal threshold value using the scintillator to trigger the readout [37]. The ASDs are used in the TOT mode. Muon tracks are reconstructed from the recorded hits, requiring at least four hits per track. The efficiency is determined by performing the track reconstruction while leaving out a single layer of drift tubes.

For every reconstructed track, we tested whether the drift tube in the layer after the test was traversed by a muon candidate providing a hit. The efficiency is the ratio between the number of hits associated with a reconstructed track and the number of reconstructed tracks traversing this drift tube. The procedure is repeated leaving out another layer of drift tubes. The average, for all chambers, tube efficiency for each tested chamber is 0.9891 ± 0.0003 [38].

3.3. Thin-Gap Resistive Plate Chambers (RPC)

The present ATLAS Resistive Plate Chambers (RPC) system is a tracking detector providing the first-level trigger in the ATLAS barrel. The RPC system is constituted of six concentric cylindrical layers providing independent space-time measurements along the track, with 1 ns × 1 cm resolution.

The major upgrade for the HL-LHC program, consisting of three additional full coverage layers of new-generation thin-gap RPCs, will be installed in the inner barrel region, as shown in Figure 15 [30]. The new system has 1 mm gas gaps instead of 2 mm gas gaps, expected to increase the trigger acceptance from ~70% to ~96%. In addition, a redundancy of the RPC system increases the trigger selectivity and improves the resolution of the particle velocity up to 0.5%, due to the increased time resolution and lever arm [39].

The 276 new triplet thin-gap RPCs have been built for the phase II upgrade [14]. Similar chambers are currently being developed for the BIS78 upgrade and a singlet RPC chamber. It is expected that the RPC detector will have a lower operating high voltage for high-luminosity LHC runs, to reduce the aging effects.

The RPC trigger schemes will use nine measurement planes, provided by four layers of RPC chambers: one BI (Barrel Inner) triplet (RPC0), two BM (Barrel Middle) doublets (RPC1 and RPC2), and one BO (Barrel Outer) doublet (RPC3). These schemes are based on the four layers of RPC chambers:

• 3/3 chambers. Hits in at least three out of four planes of the RPC1 + RPC2 chambers and in at least one out of two planes of RPC3. This is equivalent to the present high-pT trigger.
• 3/4 chambers. The previous requirement is in a logical OR with the requirement of hits in at least two planes out of three in RPC0 and at least three planes out of six in RPC1 + RPC2 + RPC3.
• 3/4 chambers + BI-BO. The previous requirement is in a logical OR with the requirement of at least two hits in RPC0 and at least one hit in RPC3. This enhances the trigger coverage in the regions where no BM RPCs are installed due to the mechanical support structure of the toroid coils.

Figure 16a shows the efficiency of the three-out-of-three (3/3) RPC chamber trigger corresponding to the high-p_T trigger as a function of the pseudo-rapidity η, used in Run 2, while Figure 16b shows the efficiency of the three-out-of-three (3/3) new RPC chamber trigger corresponding to the high-p_T trigger applied in Run 3. The overall RPC trigger efficiency, between Run 2 and Run 3, increases from 65% to 90% [30].

The BIS1-6 chamber envelopes with the chamber layout and the installation strategy were defined and optimized. In addition, an improvement is foreseen for a lighter dielectric filler material for the readout strip panels. It is planned to read out strips from both ends to obtain η and φ coordinates simultaneously.

New electronics are required to read out the thin-gap RPC signals, as the effective threshold for front-end electronics should be 0.1 mV instead of the 1 mV used for the initial RPC chambers in the barrel region. A new ASIC front-end integrating discriminator TDC and serializer was designed using the SiGe process. A prototype was developed, integrating the discriminator and TDC plus a DCT board that receives the RPC data and provides the muon hits to the sector logic board.

3.4. DCS for the New Small Wheel (NSW) [Micromegas (MM) and the sMDT Chambers]

The NSW detector requires the development of a sophisticated Detector Control System (DCS), considering its complexity and long-term operation. The DCS allows the detector to function consistently and safely with a direct interface to all sub-detectors and the technical infrastructure of the experiment via the ATLAS overall DCS [30]. The NSW DCS works following closely the architecture and command structure of MS DCS to facilitate expert operations. The top node of both MM and sTGC propagates the current state and receives commands from the ATLAS overall DCS [40].

The NSW High-Voltage Scheme (HVS) is provided by many hardware units installed at Point 1 (P1) of the ATLAS, US15 area. The HVS hardware consists of three mainframes, 16 slots each, HV Boards, and cables for their connection. The mainframes are filled with two types of HV Boards, the A7038STP and A7038STN board (positive and negative polarity) and the A7038AP Board.

The HV validation tests for the MM chambers were completed successfully, under specific conditions with stable behavior, where all the channels were turned on for at least 10 min without trip or failure. Extensive tests of the even or the odd channels turned on for 2 h and then all channels turned on for one hour, resulting in a total minimum 4 h test procedure completed without any issues, leading to the successful qualification of the boards [41].

The DCS power supply, HV for the chambers, and LV for the front-end electronics (FE) of the sMDT chambers were developed and tested. A new Low-level Structure was created, as shown in Figure 17: data points, alarms, archive, and a new Main Finite State Machine (FSM) Panels for better visualization. The State Management Interface (SMI) is the FSM toolkit incorporated in the framework, which allows for the description of any subsystem as a collection of objects, each object operating as an FSM.

The architecture adopted for modeling the structure of sub-detectors, subsystems, and hardware components in a fashion consistent using hierarchical (tree-like) structures is composed of two types of nodes: the Device Units (DU), which are capable of monitoring and controlling devices they correspond to, and the Control Units (CU) or Logical Units (LU), which are considered to contain FSM capable of modeling and controlling the subtree below them. The nodes of the tree are CUs and DUs and always appear as leaves of the tree, as shown in Figure 18. The Device Units (DU) are in the lowest part of the tree, the Logical Units (LU) are in the middle part of the tree and the Control Unit (CU) appears in the upper part of the tree, i.e., DCS.

Additionally, a new MDT power supply panel was designed for the shared LV channels, having easy access to the state and the status of the shared channels (Figure 19). The new naming scheme was properly installed in the main MDT libraries and scripts, establishing the connection with the Power Supply project. The new Low-Voltage (LV) Intermediate Control Station (ICS) is used to power the Low-Voltage Distributor (LVDB) boards of the NSW, and, through them, the readout and trigger boards of the system provide data and function safely. The Supervisory Control System monitors the NSW LV unit with the rest of the projects. The NSW DCS separate configuration/monitor, readout and trigger path consists of 2.4 million readout channels that result in 100,000 DCS parameters [41]. The electronics and the temperature monitoring units are the most valuable interfacing systems for normal NSW operation.

4. The Muon Spectrometer Operation Parameters

The muon spectrometer operation parameters, track reconstruction and spatial resolution, plus the signal and background rates, were simulated and tested experimentally, after every step of the upgrade procedure.

4.1. Muon Track Reconstruction and Resolution

The muon spectrometer (MS) detector and electronics components were, initially, designed for, at least, 10 years of operation at a luminosity of about 1 × 10³⁴ cm⁻² s⁻¹, providing an integrated luminosity of 1000 fb⁻¹. The detector hit rates and radiation doses were measured directly at the start of the LHC operation. The MS electronic components were tested for radiation tolerance and rate capability, considering the levels corresponding to the expected doses and rates predicted by simulations multiplied by the conservative safety factors. At the HL-LHC operation corresponding to an integrated luminosity of 3000 fb⁻¹, the new components produced for the phase II upgrade are foreseen to be qualified for the ultimate HL-LHC of an integrated luminosity of 3000 fb⁻¹ [30].

The upgraded MS must be able to maintain good performance over the full rapidity and momentum range despite the larger detector occupancies and to operate at instantaneous luminosities up to 7.5 × 10³⁴ cm⁻² s⁻¹ and at the nominal HL-LHC pile-up ⟨μ⟩ = 200 (⟨μ⟩ = 0, equivalent to no pile-up. LHC Run 1 ⟨μ⟩ = 20; Run 2+3 ⟨μ⟩ = 80; HL-LHC Run ⟨μ⟩ = 200 [42]) without significant performance degradation. The di-muon resonances such as Z bosons or high-mass Z0 are used to quantify the impact of the upgrades on the reconstruction.

For a successful physics program at the HL-LHC, the muon identification, muon reconstruction performance and identification efficiency, fake rates, and momentum resolution should not degrade significantly due to the higher particle rates and backgrounds.

The muon trigger should be able to trigger on single muons with high efficiency, with a rate of less than 40 kHz for a threshold of p_T > 20 GeV. In addition, lower p_T thresholds, as low as p_T > 4 GeV, should be viable for multi-muon and combined triggers. The sharpness of the turn-on curve of the trigger, and, thus, the rate of fake triggers from low-momentum muons, which dominate the total muon trigger rate, is limited by the spatial resolution of the RPC and TGC trigger chambers in the bending coordinate. The NSW detector system is fully integrated into the ATLAS simulation and reconstruction software Athena [43,44]. The double Gaussian fit to track or layer residual provides the extracted muon track resolution, in Figure 20 [31].

The reconstruction of the muon tracks follows the procedure from the muon hits, detected by the TGC and MM clusters for the hits clusterization, then pattern finding seed for transverse coordinate for the 2D segment, in addition to azimuthal coordinate in fit to 3D track segment, combining NSW segments with segments in other muon subsystems, leading to the ATLAS muon spectrometer tracks and combining with tack segment on ATLAS inner detector, leading to a combined muon track dataset [31,45,46].

The cluster position reconstruction is obtained by taking the charge centroid, although the improved cluster reconstruction methods under study are promising, as shown in Figure 21 [31].

The reconstruction of single-layer resolution is affected by residual layer–layer and global misalignment. An optical-based alignment system has been installed and commissioned to track the position and deformations of the NSW detectors [17].

A substantial improvement in resolution is expected once all effects are considered and corrected. An improvement in the single-layer position reconstruction is achieved with new time-based reconstruction algorithms.

Further improvements are in progress from the improved cluster position reconstruction methods by using MM time information and sTGC fit charge profile with Gaussian parabola. In Figure 22a, the Micromegas (MM) chamber position resolution as a function of angle θ is presented for muon tracks with a reconstructed p_T > 15 GeV at pp collisions energy 13.6 TeV. This resolution has been extracted and corrected by comparison of the cluster position on two neighboring layers corrected by the track angle for small and large sectors under the B-field effects. The cluster position is reconstructed using the charge-weighted mean value of the strip position and not all corrections are applied yet. The worsening of the resolution for higher angles is expected. In Figure 22b, the sTGC chamber position resolution as a function of angle θ is presented for muon tracks with a reconstructed p_T > 15 GeV at pp collisions energy 13.6 TeV. The cluster position is reconstructed using charge-weighted mean strip position, and not all corrections are applied yet. For the black points, no pedestal was subtracted from the charge read out by the front-end chip; for the red points, a pedestal of 32 ADC counts was subtracted for each channel; and for the green points, the pedestal was individually calibrated for each channel and subtracted in the reconstruction processing. The nominal resolution of the detector is expected to be better in the future, avoiding the effects of the residual misalignments and the built geometry that are currently under study [46].

4.2. The Signal and Background Rates

The expected signal rates in the NSW during the HL-LHC operation are presented in Figure 23, for various parts of the MM chambers, obtaining the values by scaling the measured rates in ATLAS on 2022 data at sTGC. The particle rate during the HL-LHC operation is expected at a luminosity of 7.5 × 10³⁴ cm⁻² s⁻¹ to be 18 kHz/cm², which is within the design goal of the NSW (20 kHz/cm²) [17].

A dedicated simulation of the cavern background in the MS is performed by recovering the particles that can be rejected by the time and energy cuts and passing them through the full ATLAS simulation. To improve the simulation of the cavern background, the GEANT4 physics lists, which are optimized for low-energy processes, were used. Cavern background hits are then overlaid on the standard events in the same way as normal pile-up events [30].

The background hits in the MS originate from low-energy photons and neutrons, as well as from charged particles, and are dominated by the rate of real muons. In the current detector configuration, the background rates are highest in the New Small Wheel detectors, namely MM, sTGC, and sMDT chambers.

The prediction of background rates in the NSW detectors primarily relies on an accurate measurement of the background rates with the present detector, applying scaling factors for luminosity and the HL-LHC filling scheme. Monte Carlo simulations are used for projecting the situation after the phase I upgrade when the new shielding disc (NJD) in front of the NSW replaces the current shielding disc (JD).

5. Performance and Physics

The HL-LHC physics program includes precision measurements of the Higgs boson properties in all its production and decay modes plus improved measurements of all relevant Standard Model (SM) processes and parameters, the study of rare physics processes, and searches for phenomena beyond the Standard Model.

Hence, it is mandatory for the muon spectrometer (MS), having a fundamental role in the ATLAS physics program, to be upgraded, enabling it to cope with the high event pile-up and background expected in the muon chambers. The muon reconstruction in the muon spectrometer combined with the new inner tracker (ITk) benefits from the superior momentum resolution expected from the ITk concerning the Run-2 detector [3] for muons with a transverse momentum, p_T, below a few hundred GeV. This is particularly important for the physics potential of key channels at the HL-LHC, e.g., Higgs sector processes.

The muon physics results are directly dependent on the optimum muon track reconstruction. The combined muon reconstruction, as can be shown, schematically, in Figure 24 [48], uses information from the muon spectrometer (MS), the Inner Detector (ID), and the Calorimeter. It is underlined that there are five categories of reconstructed muons, considered for the final muon reconstruction:

• The standalone muons (MS-extrapolated) were identified using a track reconstruction only by the MS.
• The combined muons (CB) were identified using information from both ID and MS.
• The segment-tagged muons (ST), where the ID track matches in position an angle to a segment in the MS, is particularly useful for low pT muon tracks and in regions with reduced acceptance.
• The calo-tagged muons (CT), where the ID track matches an energy deposit in the calorimeter consistent with a minimum ionizing particle, improve the muon track purity in the region of limited MS.

The inside-out muons have the track in the MS matched to the segment in the silicon tracker. The QCD background is estimated from the data using a template fit in the invariant mass spectrum, using the same charged muons (SC), as shown in Figure 25. For the signal, a template is created using all Standard Model processes that can produce an opposite-charged muon pair (OC), and it is then fit to the data, after having subtracted the QCD background.

The physics working points require a definition of having the NSW segment to help the reconstructed muon, e.g., requires three stations for high p_T muon. We defined the OR between the MM and sTGC detectors of having four out of eight layers with a hit on track in either technology, MM or sTGC, as input to make use of the high redundancy, resulting in an average efficiency of the four out of eight layers more than 95%, as shown in the third disc on the right with the label sTGC OR MM in Figure 26. The lower efficiency of MM chambers in the peripheral area compared to the sTGC chambers is due to the residual misalignment effects to be improved and the smaller opening angle as the MM chambers installed behind the sTGC chambers.

The two technologies, sTGC and MM are independent and complementary, providing good overall coverage [45].

The accepted muon tracks for the physics analysis were considered with p_T > 15 GeV, and their reconstruction should be a combination of ID and MS detectors or standalone MS. Two types of efficiencies were obtained.

The obtained clusters are on track and contribute to the final product for physics analysis, and the clusters within 5 mm of the track provide essential information for debugging reconstruction and initial misalignment, mis-cabling, etc. Both the MM and sTGC chambers resulted in an efficiency better than 90%, at least in the regions not affected by LV/HV readout instability problems.

These NSW performance problems of the efficiency due to the various detector and readout issues can affect the single-layer efficiency. It should be noted that redundancy is key to maintaining a high overall efficiency for the detector. It is, necessarily, important to have four measurements able to seed and reconstruct a segment in the NSW, independently. Figure 27 shows the average single-layer efficiency vs. run number for the MM (a) and the sTGC (b) chambers. The efficiency of the two detector MM + sTGC combination appears in (c).

Simulation studies showed that the total number of signal and background events expected after the full event selection for an integrated luminosity of 3000 fb⁻¹, as well as the significance and precision of the measurement, are shown in

Table 1. The expected signal and background event yields, and precise measurement, after the full event selection for an integrated luminosity of 3000 fb⁻¹.

	Nsig	Nbkg	Δμ/μ (%)
Without forward muon tagging (mm channel)	1731	840	6.2
With forward muon tagging (mm channel)	1793	700	5.1
Without forward muon tagging (combined)	5498	5200	4.6
With forward muon tagging (combined)	5598	4800	4.0

, for the di-muon channel separately and all channels combined.

The high-η tagger affects both the signal acceptance, as signal leptons in the forward region may be selected, as well as the background rejection, where the dominant WZ background is reduced with the veto on additional leptons in the event. The main effect of the high-η tagger is to provide a stronger veto on the presence of a third lepton, thus significantly reducing the WZ background, while the increase in signal acceptance is relatively modest due to the central distribution of signal leptons.

6. Conclusions

The ATLAS muon spectrometer (phase I and phase II) was presented with the major technical characteristics of subparts and other important detector subsystems, such as DCS, muon track reconstruction, and resolution, resulting in higher performance for physics studies. The New Small Wheel (NSW) is ATLAS’ largest LS2 upgrade project. The NSW employs, for the first time, two new large area sub-detector technologies: the Micromegas (MM) and the small-strip Thin Gap Chambers (sTGC) with a new DAQ system at a large scale. The hardware was fully commissioned and installed in the ATLAS cavern during LHC’s Long Shutdown 2, providing an outstanding achievement. The NSW has successfully contributed to the ATLAS Run-3 acquisition campaign. As the trigger and DAQ commissioning is being finalized, the NSW already shows significant improvements on the Level-1 muon trigger and tracking in the ATLAS forward region. There are intense and continuous efforts to understand and improve the MS system’s performance.

In parallel, the small-diameter Monitored Drift Tubes (sMDT) and the thin-gap Resistive Plate Chambers (RPC) final production is under installation and commissioning into the ATLAS cavern for preparation, ready for the HL-LHC era with the upgrade of other major ATLAS detector parts. The ATLAS collaboration is looking forward to the restart of the HL-LHC collisions and finding new physics.