CERN Accelerating science

Precision timing upgrade of the CMS detector

by Joel Butler (Fermilab), Keti Kaadze (Kansas State University), Tommaso Tabarelli (CERN)

Soon after 2027, the LHC will enter the high luminosity era, known as HL-LHC. It will begin operations at a stable luminosity of 5.0×1034 cm−2 s−1, resulting in much higher collision rates than currently achievable, but with a pileup of 140 collisions during a bunch crossing. In an ultimate scenario, it will operate at 7.5×1034 cm−2 s−1 luminosity with 200 pileup collisions per bunch crossing. The CMS detector must be upgraded to retain its excellent performance in terms of efficiency, resolution, and background rejection for all final state particles and physical quantities at these higher rates, increased pileup, and much higher integrated radiation doses. Maintaining the detector’s performance in the demanding conditions of the HL-LHC is  crucial to extend the direct search for physics Beyond the Standard Model (BSM) and the program of precision measurements to look for deviations from the predictions of the standard model (SM).

In the current LHC conditions, with pileup of 40-50, charged particles from pileup can be effectively excluded using tracking. Specifically, the current pileup mitigation strategy is to remove charged tracks inconsistent with the vertex of interest, usually the one with the highest activity or transverse momentum. However, the contamination of charged particles from pileup scales with pileup density, increasing to >1 vertex/mm, while the increase in the spatial overlap of tracks and energy deposits from these extra collisions will degrade the identification and the reconstruction efficiency for the interesting one. The first step of the response to high pileup is to improve the CMS Tracker by providing smaller detection elements (silicon pixels and strips) resulting in improved spatial resolution so that charged tracks can be assigned to the correct vertices.  Even then, however, some significant level of misassignment of tracks to vertices persists.

Another challenge for CMS is to assign neutral showers to the correct vertex and accurately measure their energy in the face of very high pileup. Showers, from neutral particles, e.g. γ and neutrons, are recorded through the energy they deposit in the calorimeters, which  in CMS  do not have directional information and cannot be associated with specific vertices, confusing the interpretation of the events. Energy from pileup interactions must be removed with special statistical inference techniques, which still suffer from some contamination from neutral energy from other collisions in the bunch.

CMS plans to further reduce the effects of pileup by the use of precision timing using the fact that the individual collisions within a bunch crossing are not simultaneous, but instead occur at slightly different times. The proton bunches travelling at the LHC –nearly at the speed of light- still need about a nanosecond to fully pass through each other. This results in a time distribution with an RMS of 180-200 picoseconds, approximately independent of where along the collision axis the interaction occurs. It is this time difference that CMS will exploit to provide additional capability to correctly associate tracks and showers to vertices. Further upgrades, in addition to the new tracker, are needed: a new dedicated detector for precision timing of minimum ionizing particles (MIPs), the MIP timing detector (MTD); and electronics upgrades to enhance the timing capabilities of all the calorimeters. The first of these upgrades is the subject of  the remainder of this article.

The MTD is a hermetic  detector that will surround the entire Outer Tracker and will measure the time-of-arrival of each charged particle. The time resolution is  expected to be ~35 ps on all charged tracks at the beginning of the HL-LHC. By the end of HL-LHC operations, after the detector has experienced about 3000 fb-1 of collisions, the resolution will be degraded by approximately a factor of two. The proposed design and adopted technologies must meet a number of different technical requirements including radiation and magnetic field tolerance, low deadtime, high granularity (low occupancy), low cost per unit area, and must be far along in their R&D programs to allow us to meet the schedule requirements of the upgrade. Five technologies were investigated and studied in dedicated beam tests and radiation exposures. Crystal scintillators read out with silicon photomultipliers (SiPMs) were chosen to instrument the barrel region of CMS and Low-Gain-Avalanche-Detectors (LGADs), silicon sensors with internal gain in the neighborhood of 10-30, emerged as the best technology for the endcap timing layers. A simplified layout of the MTD is shown in the following image.

A schematic view of the GEANT geometry of the timing layers implemented in the CMS software for simulation studies comprising a barrel layer (grey cylinder), at the interface between the tracker and the ECAL, and two silicon endcap (orange and light violet discs) timing layers in front of the endcap calorimeter. 

The Barrel Timing Layer (BTL) will cover the pseudorapidity region up to |η| = 1.48 with a total active surface of about 40 m2. The fundamental detection cell is based on a very small bars LYSO:Ce crystals of 57 mm length, and transverse dimensions of 3 mm width.  The crystals’ radial thickness vary from 3.7 mm (|η| < 0.7) to 2.4 mm (|η| > 1.1), to equalize the slant depth crossed by all particles starting from the interaction point. Each bar is read out by two SiPMs, one on each end, with  cross-sectional areas matched to the faces of the bars.  They are located at the outer edge of the Outer Tracker support cylinder about 1.1 m from the beams over a length of about 5 m. Groups of 16 bars (32 SiPMs) are read out by a custom ASIC. The average of the time from the two SiPMs  gives the time-of-arrival of the charged particle independent of its position along the 57 mm of the bar and the time difference gives an approximate distance along the bar. There are approximately 166,000 bars. The good timing resolution of  ~30 ps of the design has been already demonstrated in tests of small prototypes consisting of crystals read out with SiPMs both at CERN and Fermilab. Moreover they were also proven to be radiation tolerant up to a neutron equivalent fluence of 3 × 1014 cm−2 when cooled to below −30oC.

Overview of the BTL showing the hierarchical arrangement of the various components, bars, modules, Readout Units, and trays, inside the Tracker support tube.

The MTD system is closed by the Endcap Timing Layer (ETL)  which  has disks at each end of the barrel. There are two disks on each end, providing two timing measurements per track. LGADs provide high time resolution over the pseudorapidity range from about |η| = 1.6 to |η| = 2.9. The LGADs consist of 1.3 x 1.3 mm2 pads of silicon. A 16 x 16 array of pads is read out by a custom ASIC which measures the arrival time of particles at each pad and the energy deposited in it. The total area of all four disks is about 15 m2 and there are nearly 9 million pads in total. The need to choose a different technology than for the barrel is imposed by the radiation tolerance limitations. Ongoing studies of the radiation tolerance of LGADs indicate a promising performance of about 30 (50) ps at fluences of 3 x1015 cm-2 corresponding to |η| = 3.0, at the beginning (end) of the HL-LHC operation. The same technology is also proposed for a fast-timing layer under consideration for the very forward region (2.4 < |η| < 4.8) of the ATLAS experiment.  The ETL will be installed in a thermally isolated volume on the nose of the Endcap ECAL, positioned about 3 m from the interaction point. The disks have a radius of 1.2 m and are split vertically in a “clam-shell” arrangement, allowing then to be extracted from the detector without even removing the beam pipe.

Cross-sectional view of the endcap timing layer along the beam axis. The interaction point is to the left of the image. Shown are two ETL disks populated with modules on both faces, along with the support structure. The grey sections are the active areas of the modules with LGAD sensors. Each orange bar represents a service hybrid.

The CMS physics programme at the HL-LHC targets a wide range of studies, including precise measurements of the Higgs boson properties and direct BSM searches. All these studies will benefit from the improved efficiency for isolated objects. A very crucial measurement to be performed is the di-Higgs production, which will provide a direct measurement of the Higgs self-coupling. In this case, precision timing could increase the signal yields for constant background in HH→ bbγγ by 22% while similar improvements are predicted for other Higgs boson signatures, ranging from 15–20% in the case of HH→ 4b to 20–26% for H→ 4µ.

Impact on signal efficiency for HH → bbγγ for no-timing, barrel only timing, and barrel plus endcap timing scenarios. The quantity yHH is the rapidity of the di-Higgs system. 

Moreover, the sensitivity to searches for new physics depends on measuring the missing transverse energy (EmissT ). In that sense, tails are equally important for tracking resolution and the proposed detector could achieve a nominal |∆z| < 1 mm window providing very high efficiency in isolation sums. The gain in the EmissT resolution with track timing leads to a reduction of ∼ 40% in the tail of the distribution above 130 GeV, which approximately offsets the performance degradation for SUSY searches because of the higher pileup.

Diagram for top-squark pair production and decay (left), η and mass distribution for a 700 GeV neutralino with three different lifetimes reconstructed from the kinematic closure of the secondary vertex using time information with 30 ps resolution (right). 

In addition, the improved track-time reconstruction opens whole new capabilities that allow CMS to look for phenomena that are outside the capability of the current detector. These include searches for neutral long lived particles (LLPs), postulated in many extensions of the SM. Heavy particles that travel a distance in the tracking volume and then decay into lighter standard model particles produce very delayed signals in the MTD. Time-of-flight between the collision vertex and the MTD can also be used to identify charged hadrons, such as pions, kaons, or protons, at relatively low transverse momentum of a few GeV/c, typical of most particles produced in Heavy Ion collisions, providing new opportunities to study those collisions as well. 

Thus, the capabilities offered by the MTD, combined with other planned CMS upgrades will contribute to improvements in many physics analyses during the HL-LHC and deepen our understanding of the validity of the SM and the physics of the ElectroWeak sector, enable searches for new LLPs and enhance our knowledge of the collisions of Heavy Ions. A detailed technical design report for the MTD was submitted earlier this year and approved recently by the CERN Research Board.

Further Reading: 
[1] A MIP Timing Detector for the CMS Phase-2 Upgrade, Technical Design Report, CMS Collaboration