Abstract For next-generation X-ray free-electron lasers (XFELs), achieving low-emittance electron beams (below 0.2 mm·mrad for a 100 pC bunch charge) is critical for enhancing lasing performance. At the Institute of High Energy Physics, a C-band photoinjector test platform, featuring a 3.6-cell RF gun operating at 5.712 GHz, has been developed to deliver such high-brightness beams. To characterize the beam emittance, we have evaluated three different schemes based on single-slit and dual-slit scanning methods. Moreover, to address the impact of the dynamic errors, the positional accuracy of the slit and the spatial resolution of the beamlet profile monitor have been quantified. The single-slit method, optimized with a 10 μm slit width and 0.1 m drift distance, results in a measurement error of 14%. To improve the accuracy, we propose two novel dual-slit approaches: parallel and orthogonal scanning. Numerical simulations using ASTRA and Python demonstrate that both methods suppress the space-charge effect and are able to control the emittance-measurement error below 10%. The orthogonal-slit-scan scheme, mimicking a moving square micro-hole, offers four-dimensional phase-space reconstruction with simpler hardware, making it the recommended approach. These results represent a step forward towards the precise characterization of the high-brightness electron beam from a high-gradient photocathode RF gun.
Abstract Future high-energy physics experiments require tracking detectors with improved timing and spatial resolution, combined with a low material budget. The CASSIA project addresses these requirements by developing a monolithic pixel detector with internal amplification, achieved through implementation of gain layers in a commercial 180 nm CMOS imaging process. The first prototype, CASSIA1, includes four 3×3-pixel matrices and 24 single-pixel structures with varying gain layer geometries and electrode spacings, which affect gain, noise and breakdown behaviour. CASSIA1 was characterised using TCT and source measurements, with temperature-dependent noise studies conducted in a controlled climate chamber. Measurements confirm operation in both LGAD and SPAD regimes, soft-breakdown behaviour allowing fine gain tuning, and low dark count rates suitable for high-SNR applications. The results of this characterisation campaign informed both the improved sensor design and the in-pixel front-end electronics for CASSIA2. The next-generation CASSIA2 chip, also submitted in 180 nm CMOS, will feature larger matrices with a fully scaled front-end, combining mixed digital and analogue readout. Its design will be presented, highlighting the challenges of integrating LGAD structures within CMOS circuitry. The design stages, from in-pixel circuit placement to top-level digital readout architecture, are discussed.
Abstract The next generation of high-energy physics experiments demands dedicated R&D to meet increasingly stringent requirements. In this context, the proposed ALICE 3 experiment at the Large Hadron Collider, scheduled to start data taking in 2036 (Run 5), aims to achieve excellent particle identification capabilities, with a dedicated silicon pixel time-of-flight detector targeting a time resolution of about 20 ps. To this end, traditional LGADs with different thicknesses have been tested, demonstrating the possibility to meet the ALICE 3 requirements for thinner sensors. However, a monolithic approach would simplify production and reduce costs by integrating the sensor and readout electronics into a single substrate. In this contribution, the main results and future outlook for the development of LGAD in CMOS technology (CMOS-LGAD) are presented. The first prototypes produced within the INFN ARCADIA project, were characterized in the laboratory using a laser source to estimate the contribution of electronic jitter to the overall time resolution; this contribution is about 40 ps, but it decreases to 22 ps in a higher power consumption configuration. Initial tests with charged particle beams have demonstrated the possibility of achieving time resolutions of 75 ps and a global efficiency larger than 95%.
Abstract A high-threshold, fast-rise-time multi-channel synchronous pulse trigger source has been developed, based on analog-digital signal processing and high-speed semiconductor switching technology. Operating in external trigger mode, the system provides 15 V threshold level and an amplitude level of 15 V in critical signal chains through optimized circuit design and selective component sourcing. The study proposes and implements a hundreds-of-millisecond self-locking mechanism using a D flip-flop closed-loop circuit. This mechanism secures input signals through logical timing relationships, preventing multiple false triggers within short intervals and increasing system reliability. Furthermore, a multi-channel parallel buffer structure combined with a push-pull BJT-driven PMOS shaping circuit is designed to increase driving capability and reduce single-pulse rise time. The system ultimately delivers four synchronous pulse outputs via a distribution network, enabling multi-device synchronization. Experimental results show that the system achieves an external trigger threshold of 15 V and a self-locking duration of 201 ms. Under a 50 Ω load, it delivers pulse signals with an amplitude of 21.7 V, a pulse width of 101.09 ns, and a rise time of 1.52 ns. The peak-to-peak jitter ( T p-p ) of the triggering pulses is 0.33 ns, and the channel-to-channel temporal skew ( T p-p ) across the four output channels is 54 ps. These specifications indicate the system's capability to meet the demands for triggering in complex electromagnetic environments.
Abstract The upgrade of the LHC to the high luminosity LHC (HL-LHC) by the end of this decade will impose significant challenges on the detectors of the LHC experiments. Increased luminosity of up to 7.5 × 10 34 cm -2 s -1 with up to 200 simultaneous p-p interactions per bunch crossing and foreseen run-times equivalent to up to 4000 fb -1 make it necessary to develop new detectors that can cope with the corresponding radiation damage, occupancy, and bandwidth needs. Among other detector upgrades, ATLAS will replace its entire inner tracking system with a new, all-silicon inner tracker (ITk) with a 5-layer hybrid pixel detector at its heart. This new pixel detector will feature a sensitive surface of about 13 m 2 and deploy several silicon sensor technologies as well as innovative concepts like serial detector powering and evaporative CO2 cooling to unprecedented scales. The ITk pixel project has finished its design and prototyping period and the different detector components are either in the pre-production or production phase. This proceedings will give a comprehensive overview of the detector design, the overall project status and the biggest challenges towards production. It will include lessons learned from module pre-production, and experience with the RD53 front-end chip. Recent results of close-to-real system-level tests as well as the remaining project timelines will also be discussed.
Abstract N-channel and P-channel metal oxide semiconductor (NMOS and PMOS) transistors of various dimensions were fabricated in a commercial 130 nm complementary metal oxide semiconductor (CMOS) process. The total ionizing dose (TID) effect of these transistors is evaluated through X-ray irradiation up to a total dose of 10 Mrad. Results indicate that for PMOS transistors with a fixed channel width, a shorter channel length leads to more pronounced radiation-induced performance degradation. For NMOS transistors with a fixed channel length, a narrower channel width results in more significant performance degradation. This study provides experimental insights into radiation-hardening design for CMOS devices at this technology node.
Abstract The Mu3e Experiment aims to observe charged lepton flavor violation via the μ → e + e - e + channel with a goal of single event sensitivity down to 2 × 10 -15 on the branching ratio in Phase 1. The detector is being actively constructed and commissioned at the Paul Scherrer Institute (PSI), where it will be subjected to 10 8 muons per second at the piE5 beamline. Due to the low momentum nature of the electron tracks, an ultra-thin detector with excellent vertex reconstruction performance and momentum resolution is required in order to reduce multiple scattering effects. Using HV-MAPS based MuPix11 sensors at thicknesses down to 50 µm, the so-called Vertex Detector is made up of two layers of “ladders” with six chips each, concentric around the target. High Density Interconnects (HDIs) made up of aluminium traces and polyimide layers provide the electrical connections. A tool-assisted manual assembly procedure and dedicated air-cooled quality control (QC) setups have been developed for the production of the Vertex ladders. This article will provide an overview on the assembly process and QC testing for the second iteration of the detector for a planned upcoming run in 2026.
Abstract Within the INFN OpTiME project, we present the latest characterization results of a novel 3D silicon sensor developed as a core component of a multistage photo-detector system for fast timing with resolutions below the ps threshold. The presented device is based on FBK double-sided 3D silicon technology and is optimized for low-energy electrons and photon detection. Extensive electrical and functional tests have been performed to assess its performance, including studies of its charge collection and time response using Transient Current Technique (TCT). The results from the characterisation campaign show an intrinsic time resolution below 1 ps, displayed during TCT scan as well as the presence of a stable intrinsic charge gain without the need for a dedicated gain layer up to a factor of 10 at 200 V bias voltage, indicating the potential for radiation-tolerant silicon sensors with internal amplification. While its primary application is in fast-timing detectors, we also explore its suitability as a compact, high-resolution photosensor. Such versatility could enable its use in fields requiring precise timing, such as at pulsed fs X-ray facilities. In this work we will present the overall characterisation of the device, its performances after TCT scan and show its future applications as a fast-timing sensor.
Abstract We developed a compact detector system with high energy resolution and count rate capability for plasma-profile gamma cameras on the HL-3 tokamak and performed comprehensive performance and adaptability evaluations. The detector unit employs ∅ 1”× H 2” LaBr 3 :Ce crystals coupled to a SiPM array and customized low-noise pulse-shaping preamplifiers. Experimental results demonstrated the following: (1) Owing to the crystal's excellent intrinsic energy resolution and the complete charge collection of the SiPM current signal, the detectors achieved energy resolutions of <2% @ 4.4 MeV, thereby effectively supporting fuel-ratio and fast-ion-behavior analyses. (2) Pole-zero cancellation reduced the output pulse width to approximately 100 ns, resulting in negligible spectrum degradation at 1 Mcps and enabling operation under HL-3's anticipated high gamma flux. (3) Because the 2”-thick LaBr 3 :Ce crystal provided sufficient attenuation, the detector still yielded well-resolved gamma spectra after irradiation to a fluence of 1 × 10 10 n/cm 2 from a Deuterium-Tritium (D-T) accelerator source, equivalent to withstanding 100 high-parameter experiments similar to JET (∼ 2 × 10 28 n/s). All 20 detectors of the gamma camera exhibited highly consistent energy resolutions of 3.36% at 662 keV and 1.88% at 4.4 MeV, with standard deviations of ± 0.13% and ± 0.05%, respectively. Moreover, accelerated aging tests confirmed that the detectors are suitable for reliable deployment in the upcoming HL-3 experiments over two years, with energy resolution degradation of less than 0.5%. Consequently, this high-performance LaBr 3 :Ce detector array represents a reliable core module for the gamma-ray diagnostic system on the HL-3 device. This design also provides a reference for other high-flux, high-resolution gamma-ray spectroscopy applications.
Abstract Accurate luminosity measurements are essential for stable machine operation and precision physics at the High-Luminosity LHC (HL-LHC), where pileup levels exceeding 140 proton-proton interactions every 25 ns impose stringent requirements on fast and radiation-tolerant instrumentation. The Beam Monitoring detector for ATLAS (BMA) is an additional luminosity detector designed for HL-LHC. It is based on Low-Gain Avalanche Diodes (LGADs), which provide intrinsically fast signal rise times, high signal-to-noise ratio (SNR), and strong radiation tolerance — key features for resolving individual 25 ns bunch crossings in the HL-LHC environment. This contribution presents the BMA detector architecture, including the LGAD sensor layout and the readout chain, all designed to operate under the high particle fluences expected at the installation position. The amplification of the detector signal occurs far from the detector, thereby reducing radiation damage to the electronics, but still maintaining a high SNR. We report results from laboratory characterisation as well as the performance of prototype BMA detectors installed in ATLAS during the 2025 data-taking period. The LGADs performance — such as efficiency, gain stability, and gain degradation after irradiation — is analysed. The results provide an assessment of the LGAD technology as a beam-monitor and luminosity detector in the LHC environment, indicating that the gain and the particle detection efficiency, relevant for bunch-by-bunch luminosity measurements, can be retained under HL-LHC conditions.
Abstract To meet the demand for flexible control of the longitudinal profile of the magnetic field in undulators, an electromagnetic undulator powered by multiple independent power supplies is proposed. In this scheme, each magnetic pole pair is supplied with an independently controlled excitation current, enabling arbitrary K ( z ) distributions. To address the accumulation of field integrals and the associated electron beam orbit distortion caused by longitudinal field variations, a compensation method based on local current balancing between adjacent periods is introduced. This method suppresses the angular deflection and transverse orbit offset of the electron beam. Simulations demonstrate that for both strong linear tapering and complex non-monotonic K profiles, the proposed compensation method, combined with end-pole correction, effectively suppresses the accumulation of the first and second field integrals, thereby maintaining electron beam orbit straightness. The proposed electromagnetic undulator provides a precise and flexible experimental platform for FEL efficiency enhancement, inverse free-electron laser (IFEL) acceleration, and ultrashort broadband THz pulse generation.
Abstract We report recent developments and tests with beams and cosmic rays of the gaseous photomultiplier (GasPM). The GasPM is a photosensor that combines a photocathode with the avalanche-multiplication mechanism of a resistive-plate chamber, offering excellent time resolution and cost-effective scalability. In addition, the GasPM provides precise and efficient Cherenkov-based charged-particle identification if combined with a radiator. Our primary use case aims at an upgrade of the Belle II detector to suppress beam-induced background photons, preferably detected off-collision time, that degrade the performance of the electromagnetic calorimeter. In 2022 we achieved a promising single-photon time-resolution of 25 ps at 3.3 × 10 6 gain, using a picosecond-pulse laser and a LaB_6 photocathode. An electron-beam test in 2023 demonstrated the Cherenkov application using a MgF 2 window attached to a CsI photocathode, showing 70 ps time resolution. This work aims at addressing the principal GasPM limitations. We target ultraviolet-photon emission during excitation and de-excitation of the gas molecules, which yields a secondary signal that overlaps the primary signal, spoiling time resolution (photon feedback). We design and execute an improved beam test. Along with several GasPM configuration changes, we introduce a new 10 GSPS frequency digitizer to better discriminate primary from secondary signals thus enabling the study of photon feedback. We also conduct a cosmic-ray test using a LaB 6 photocathode, which possibly offers a more robust alternative to CsI against damage caused by ions drifting back onto the photocathode.
Segmentation of complex structures in X-ray tomographic data is a fundamental task in biomedical research, but it often requires large amounts of precisely annotated data, making fully supervised approaches costly and difficult to scale. In this study, weakly supervised deep learning is investigated as a strategy to reduce annotation effort while maintaining accurate segmentation. A two-dimensional convolutional neural network based on the nnU-Net framework was adapted to a weak supervision setting using sparse dot-based annotations, complemented by a limited number of fully segmented images. The approach was evaluated on high-resolution microCT slices of rat kidneys, targeting the segmentation of renal glomeruli, which are small, low-contrast anatomical structures. Results indicate that weak supervision provides a meaningful learning signal, enabling reliable localization of glomeruli even in the absence of dense labels. Incorporating a small set of high-quality annotations substantially improves segmentation performance, approaching that of a fully supervised model. These findings highlight the potential of weakly supervised learning as an annotation-efficient strategy for the analysis of complex structures in X-ray tomographic data, and suggest that alternative loss formulations tailored to sparse annotations may further enhance performance.
Abstract Trigger Systems (TrigSys) are an indispensable component of High-Energy Physics (HEP) experiments, serving as the initial decision layer to filter scientifically relevant events from the massive data rates produced by particle collisions. The robustness of these systems is paramount, as their performance critically dictates the efficiency and purity of the final physics dataset. However, TrigSys classifiers, especially those involving complex tasks like electron identification based uniquely on calorimetry, are highly susceptible to performance degradation caused by subtle shifts in the underlying probability density functions (PDFs) of detector response models. To address this vulnerability, a nonparametric information geometry framework is introduced for diagnosing and quantifying classifier instability under model transitions. In this work, we propose a validation model for calorimeter-based triggers, specifically focusing on electron signatures. Two trigger classifiers were constructed to simulate a transition from an initial physics model ( v 0) to an updated model ( v 1), both explicitly calibrated to preserve identical signal efficiencies. Using the exponential connection derived from the reference model ( v 0), the minimal geodesic distortions between the PDFs representing electron shower shapes are rigorously quantified. Even under equivalent signal efficiency calibration, the model transition is shown to induce fine-grained structural shifts in the underlying distributions. The main contribution of this work is the provision of a precise, non-parametric diagnostic tool capable of measuring these fundamental PDF distortions using geometric principles. By quantifying the stability of the trigger's underlying data representation, this novel methodology offers a significant advance over traditional efficiency metrics, enabling physicists to assess the true operational discrepancy and ensure the long-term integrity of HEP data acquisition.
Abstract The pursuit of ultrafast timing resolution is a central goal in modern particle physics and medical imaging. While bismuth germanate (BGO) has been largely superseded in Time-of-Flight applications by faster scintillators due to its slow scintillation decay, it possesses an untapped potential: the generation of prompt Cherenkov radiation. Through Geant4 simulations, we anticipate that coupling BGO with ultrafast photodetectors for the detection of Cherenkov light can achieve a timing resolution of 30 ps. In a series of beam tests conducted with proton, muon, and electron beams, using Fast-timing Photomultiplier Tubes (FPMTs) coupled to BGO crystals, we have characterized the production of Cherenkov light in BGO and achieved a coincidence time resolution (CTR) of 26.6 ps (σ). This study further demonstrates the potential of BGO crystals as a Cherenkov radiator for ultrafast timing applications when combined with FPMTs and high-precision readout electronics. Further optimization of optical coupling and data acquisition systems is expected to improve performance.
Abstract The Phase-2 upgrade of the CMS Level-1 Trigger (L1T) for the High-Luminosity LHC (HL-LHC) requires real-time execution of computationally intensive algorithms like particle-flow (PF) reconstruction and pileup-per-particle identification (PUPPI), within a fixed latency budget of 12.5 μs on FPGA hardware. Implementing these algorithms in High-Level Synthesis (HLS) introduces non-trivial challenges: quadratic complexity in PUPPI, data-dependent memory access patterns that prevent full parallelization, and iterative loop structures that inflate pipeline depth and resource consumption. This paper presents the HLS firmware design and optimization strategies developed for the Hadron Forward (HF) calorimeter, covering the forward pseudorapidity region 3.0 < | η | < 5.2, where pileup is most severe and no tracking information is available. We describe four concrete techniques, hierarchical parallel reduction for seed identification, mask-based array access for cluster formation, odd-even sorting for candidate selection, and lookup-table-based trigonometric computation, each addressing a specific HLS synthesis bottleneck. The parallel reduction reduces seed-finding latency from 110 to 7 clock cycles while cutting flip-flop usage by 69%; the mask-based cluster formation eliminates dynamic array indexing, reducing LUT consumption by 67%; and replacing CORDIC with a 72-entry lookup table reduces DSP usage by 89% and latency by 43%. Together, these strategies enable the complete PF and PUPPI firmware to meet the real-time processing constraints of the HF trigger, demonstrating that HLS is a viable and efficient development framework for complex trigger firmware at the HL-LHC.
Abstract This paper presents a conceptual scheme for high-power-gain standing wave resonant testing of ceramic windows during the fabrication of fundamental power couplers. The proposed method aims to enable efficient, low-cost performance evaluation of this critical and fragile component at an early stage. By constructing a short resonant cavity with the ceramic window as the device under test, the number of electric field antinodes is minimized to the theoretical limit of one, enabling the maximum possible power gain for a standing wave resonant test system. Electromagnetic simulations demonstrate that the system can produce fields on the ceramic window equivalent to those in a conventional standing wave test with a power gain of approximately 700, which corresponds to approximately 2800 relative to a conventional traveling wave test. Multiphysics simulations confirm that the thermal effects on the ceramic window under the proposed scheme are essentially equivalent to those under conventional testing. This approach enables early-stage high-power testing of ceramic windows at dramatically reduced cost and power requirements, offering a solution for rapid process iteration during R&D and cost-effective defect screening in mass production.
Abstract Recognizing the potential for artificial intelligence (AI) and machine learning (ML) to enhance operations and create more research opportunities, the Electron-Ion Collider (EIC) is being envisioned and designed as a large-scale AI-ready state-of-the-art facility. In particular, infrastructure is being planned to support three core areas of AI/ML capabilities, referred to as Edge, End-to-End, and Bottom-Up. These capabilities must further be manifested in a flexible, efficient, reliable, safe, and secure controls ecosystem and operations environment. We present the latest on our plans and work in realizing this vision.
Abstract The 7th international workshop on new Photon-Detectors (PD2025) was held in Bologna, Italy, in December 2025, bringing together experts in photon-detector technologies and their applications. The workshop focused on recent advances in photosensors, associated technologies, and their impact across a wide range of scientific domains. The programme featured invited and contributed talks, poster sessions, and dedicated opportunities for interaction between academia and industry. This article summarises the main aspects of the workshop, including its scientific highlights, organisation, and outcomes.
Abstract This contribution presents a parametric analysis and TCAD simulations of the radiation resistance of compensated LGADs. In such designs, the gain layer is engineered through compensation of acceptor and donor dopants. Following radiation-induced deactivation of donor and acceptor atoms in the gain layer, the goal is to preserve the multiplication mechanism beyond the radiation tolerance of current LGAD technology of 2×10 15 n eq /cm 2 for 50 µm-thick sensors. The conditions under which compensated LGADs may achieve improved radiation hardness with respect to conventional LGAD technology are presented. Starting from assumptions regarding acceptor and donor deactivation rates, the optimal initial concentrations that maximize the extent of LGAD radiation tolerance is computed. If, as preliminary observations suggest, the donor removal rate exceeds that of the acceptor, compensated LGADs would need to be designed to operate initially under a high external bias close to the single-event burnout limit.
