Online program Overview Session: NSS-21

Abstract

Among the detector candidates for clinical photon-counting CT (PCCT), SiPM-based fast-scintillation photon-counting detectors (PCD) represent a cost-effective solution. However, their practical implementation is constrained by the incompatible trade-off between count-rate capability and energy resolution. To address this limitation, we developed a scintillation PCD designed to achieve both high counting rate and improved energy resolution. The detector consists of a 26-channel pixelated LYSO scintillator array coupled to an epitaxial-quenching-resistor (EQR) SiPM array with a 0.5-mm pixel pitch and 15-μm microcells. Benefiting from the EQR architecture, the SiPM pixels exhibit a fast recovery time of 3.2 ns, a 64% fill factor, and a photon detection efficiency exceeding 40% at 420 nm. A multi-channel FPGA-based photon-counting readout circuit with per-channel pole-zero cancellation was developed to suppress signal tailing and mitigate pulse pile-up. This PCD achieved a maximum count rate of 46.3 Mcps/mm² while maintaining an energy resolution of 33.7% at 59.5 keV. The spatial resolution of 1.86 lp/mm (537 μm) was achieved under 120-keV X-rays, which agreed closely with the 0.5-mm physical pixel pitch of the detector. A more sophisticated PCD based on a 64-channel EQR SiPM array with 0.25-mm pixel pitch and 10-μm microcells, coupled to a corresponding pixelated LYSO scintillator array, will also be presented. Those results indicate that the proposed scintillation PCD is a promising cost-effective candidate for clinical PCCT applications.

Abstract

We report on the development and characterisation of innovative Backside-Illuminated Silicon Photomultipliers (BSI-SiPMs) within the INFN IBIS and IBIS_NEXT projects. The BSI architecture relocates the photon entrance window to the back side of the device, enabling high fill factor and improved photon detection efficiency while preserving compatibility with advanced front-end electronics integration, and offering the potential for enhanced sensitivity in the VUV region.
A preliminary experimental campaign has been performed on partially processed BSI prototype devices, including electrical characterisation over a wide temperature range, dark count rate measurements, irradiation studies, annealing tests, and timing performance evaluation using both pulsed laser excitation and minimum ionising particles (MIPs) at CERN.
Irradiation up to 10⁹ n_eq/cm² results in an increase in dark count rate. Preliminary annealing studies show recovery of dark current by nearly one order of magnitude. Timing measurements yield resolutions of order ~100 ps, demonstrating the potential of BSI-SiPMs for fast timing applications in high-energy physics.

Abstract

Digital silicon photomultipliers (SiPMs) are being developed to circumvent many of the shortfalls inherent to analog SiPMs and their associated readout electronics. We design 3D-integrated digital SiPMs, called Photon-to-Digital Converters (PDCs), to simultaneously optimize the photosensitive area, remove analog noise contributions, and embed signal processing. With a low-power-optimized readout architecture providing high timing resolution, we demonstrate fully digital measurement approaches and their benefits across multiple fields, including time-of-flight medical imaging and neutron imaging. Leveraging the high-timing precision of the PDC through pulse shape discrimination, the components of an LGSO (Fast/Slow) phoswich scintillator assembly are precisely classified, while maintaining the expected energy resolution for the scintillators irradiated with calibration isotopes such as a 68Ge source. Coincidence timing resolution is also assessed for this configuration. These results position PDCs as great candidates for depth-of-interaction reconstruction in PET imaging, ToF measurement in CT imaging, and particle identification in neutron imaging.

Abstract

Segmentation of the SiPM active area into smaller, independently read out elements is one of the most promising routes to improve the single-photon time resolution (SPTR) required by next-generation time-of-flight PET systems. We report on Single-Cell Access through-silicon vias (SCA-TSVs), a custom TSV technology under development at Fondazione Bruno Kessler (FBK), Trento, and applied to the NUV-HD-MT SiPM platform. SCA-TSVs exploit the deep-trench isolation (DTI) already present in this technology to physically separate and independently access each SiPM microcell from the backside, with essentially no loss of fill factor. A backside redistribution layer (RDL) can then regroup microcells with an arbitrary segmentation level, according to application requirements. The first implementation targets the DIGILOG project, where a 3 × 3 mm² SiPM with 50 µm cells is segmented into 3 × 3- or 6 × 6-cell clusters, referred to as microSiPMs. These units are designed to be read out independently by a matched ASIC developed at EPFL and connected through chip-to-chip 3D integration. This approach preserves the SPTR of a single SPAD, approximately 25 ps FWHM, while avoiding the complexity of full single-cell readout. The first SCA-TSV learning cycle (LC1) required modifications to the original NUV-HD-MT technology, mainly an increase of the DTI depth beyond 20 µm. Front-side characterization showed that the timing performance was preserved. The first backside integration runs also revealed a reduced TSV yield. The analysis identified both layout-related and backside-process-related limitations, which are being addressed through optimized LC1 backend process splits and in the second technology learning cycle (LC2). The presentation will discuss the SCA-TSV concept, LC1 fabrication and characterization results, and the latest LC2 developments toward 3D-integrated microSiPM/ASIC chips for fast-timing applications.

Abstract

Precise detection of elementary particles such as neutrinos, gammas, or electrons in an unsegmented scintillating volume requires three-dimensional tracking with sub-nanosecond timing resolution [1], [2]. However, conventional detectors based on analog silicon photomultipliers are inherently constrained by trade-offs between spatial resolution and scalability, making them less suitable for large-scale implementations [3]. To overcome these limitations, CMOS single-photon avalanche diode (SPAD)-based detectors have emerged as a promising alternative for high-resolution particle detection [4], [5]. Here, we propose a dedicated large-format SPAD array sensor fabricated in 110 nm CMOS image sensor (CIS) technology, targeting both high timing precision and scalability. The sensor incorporates a high photon-detection-efficiency SPAD pixel architecture that triggers time-to-digital converters (TDCs) in an event-driven manner with a resolution of 200 ps to detect sparse photons generated by probabilistic events such as particle interactions. All TDCs are synchronized with a global reference clock distributed through an on-chip phase-locked loop, ensuring precise timing alignment across the entire sensor array. Furthermore, a tunable clock shifting feature is implemented to compensate for clock skew in multiple-chip operation. As a result, this chip serves as a fundamental building block for a scalable, multi-chip tiled detector system in which all timing information is globally synchronized, enabling future large-scale, high-resolution particle tracking applications.

Abstract

Single-photon avalanche diodes (SPAD) produced in CMOS technologies are increasingly being considered as photodetector candidates for future Ring imaging Cherenkov (RICH) detectors. CMOS SPAD based sensors with integrated electronics can be designed and optimized specifically for RICH detector applications, making it easier to achieve the high photon detection efficiency, excellent single photon timing resolution and hardness to neutron radiation as required in the planned high energy physics experiments. To help guide the design of first digital SiPM targeting RICH detector applications being developed in the spadRICH project, existing single SPADs as well as a 144x32 SPAD array were studied in terms of their operation after neutron irradiation. Dark count rates were measured for neutron irradiation as high as 10¹² 1-MeV neutron equivalent/cm². Thermal annealing and operation at temperatures down to liquid nitrogen were explored. In terms of cryogenic operation, afterpulsing probability and behavior of the photon detection efficiency were also studied. Simulations were used to assess the possible performance of microlenses specifically for the highly collimated Cherenkov photon light as expected in the upgraded LHCb RICH detector.

Abstract

Single-photon avalanche diodes (SPADs) fabricated in a 110 nm CMOS technology were irradiated with neutrons, protons, and electrons over a range of fluences up to 10¹¹ 1 MeV neutron-equivalent cm^-2. The dark count rate (DCR) evolution was systematically characterized as a function of displacement damage dose; online monitoring during neutron and proton irradiation allowed individual scattering events and short-term annealing dynamics to be resolved at the single-pixel level, providing direct insight into defect formation and recovery. Statistical characterization of the DCR distributions reveals that the average DCR alone is insufficient to describe radiation-induced degradation in SPAD arrays: the particle type critically determines both the magnitude and spatial uniformity of damage. These differences reflect distinct defect morphologies - clustered damage for hadrons, isolated point defects for electrons - a key aspect not captured by NIEL scaling alone.

Abstract

This work presents the first cryogenic performance evaluation of the novel Near-Ultraviolet Deep-Junction (NUV-DJ) SiPM technology developed by Fondazione Bruno Kessler (FBK). By relocating the high-electric field region deeper within the substrate, the NUV-DJ architecture achieves a state-of-the-art photo-detection efficiency of approximately 70% at 420 nm and a timing resolution of 60 ps FWHM at room temperature with a 4 x 4 mm² SiPM. We report on a comprehensive dark characterisation of these sensors from 300 K down to 77 K, extracting the activation energy (Ea​) to identify dominant carrier generation mechanisms, and evaluating the stability of correlated noise and gain. These studies serve as a foundation for a new dedicated production run for the LHCb Scintillating Fibre (SciFi) Tracker Upgrade II, in collaboration with EPFL, featuring custom 0.4 mm2 geometries and advanced afterpulsing suppression. The main goal is to assess whether the enhanced blue-green sensitivity of NUV-DJ, which is advantageous for the readout of irradiated blue-emitting scintillating fibres, can be combined with radiation tolerance and stable cryogenic operation at liquid-nitrogen temperature.