Evaluation of 3D pixel silicon sensors for the CMS Phase-2 Inner Tracker

TL;DR

This study validates 3D pixel silicon sensors for the CMS Phase-2 Inner Tracker by beam-testing FBK- and CNM-manufactured sensors bump-bonded to RD53A and CROCv1 readout chips. The devices demonstrate excellent hit efficiency (>96–97%) and robust charge collection at fluences up to $\sim2.6\times10^{16}\,n_{eq}/\mathrm{cm}^{2}$, with spatial resolutions of a few micrometers in the non-irradiated case and around $5$–$6\,\mu\mathrm{m}$ after irradiation. Rotation of the sensor relative to the beam mitigates column-related inefficiencies at normal incidence, preserving performance across incident angles. The results from two production sites (FBK and CNM) and two readout chips (RD53A and CROCv1) support the feasibility of the CMS Inner Tracker upgrade, with integration planning continuing toward the end of 2026 and consistency with ATLAS 3D-pixel findings. The work also identifies operational voltage margins and masking behaviors that inform sensor deployment under HL-LHC conditions.

Abstract

The high-luminosity upgrade of the CERN LHC requires the replacement of the CMS tracking detector to cope with the increased radiation fluence while maintaining its excellent performance. An extensive R\&D program, aiming at using 3D pixel silicon sensors in the innermost barrel layer of the detector, has been carried out by CMS in collaboration with the FBK (Trento, Italy) and CNM (Barcelona, Spain) foundries. The sensors will feature a pixel cell size of \mbox{$25\times100~μm^2$}, with a centrally located electrode connected to the readout chip. The sensors are read out by the RD53A and CROCv1 chips, developed in 65~nm CMOS technology by the RD53 Collaboration, a joint effort between the ATLAS and CMS groups. This paper reports the results achieved in beam test experiments before and after irradiation, up to a fluence of approximately \mbox{\SI{2.6e16}{n_{eq}/\cm^{2}}}. Measurements of assemblies irradiated to a fluence of \mbox{\SI{1e16}{n_{eq}/\cm^{2}}} show a hit detection efficiency higher than 96\% at normal incidence, with fewer than 2\% of channels masked, across a bias voltage range greater than \SI{50}{V}. Even after irradiation to a higher fluence of \mbox{\SI{1.6e16}{n_{eq}/\cm^{2}}}, similar performance is maintained over a bias voltage range of \SI{30}{V}, remaining well within CMS requirements.

Figures (35)

• Figure 1: A slice of a quarter of the upgraded CMS IT layout in the longitudinal view. It consists of three substructures: TBPX, TFPX, and TEPX. Planar pixel modules with two and four readout chips are depicted in green and orange, respectively. The 3D pixel modules with two readout chips for the innermost TBPX layer are shown in black. The innermost ring of the last TEPX disc (shown in brown) features modules with four readout chips and sends data to the beam luminosity system. The thin lower black line represents the outer radius of the beam pipe.
• Figure 2: Thermal simulation for planar (black) and 3D (orange/blue) pixel sensor modules in the innermost layer of the barrel. The comparison is done for a simulated fluence of 2e16n_eq/^2, assuming operation voltages of $600V$ and $140V$ for planar and 3D pixel sensors, respectively. The minimum temperature the coolant can reach at the modules is depicted with a vertical black line (T${_\mathrm{CO2}} = -33° C$).
• Figure 3: Cross-section of a typical (a) CNM and (b) FBK 3D pixel silicon sensor, showing the high resistivity and low resistivity layers, together with the $p^+$ and $n^+$ columns 3D3D_2.
• Figure 4: Electron microscope images of an FBK sensor: (a) top view and (b) cross-section. Only the $n^+$ columns are visible in the latter.
• Figure 5: Schematic view of the linear AFE.