SiPM non-linearity studies in beam tests with scintillating crystals

SiPM non-linearity studies in beam tests with scintillating crystals
Notice: This research summary and analysis were automatically generated using AI technology. For absolute accuracy, please refer to the [Original Paper Viewer] below or the Original ArXiv Source.

High-granularity homogeneous electromagnetic calorimeters based on scintillating crystals and silicon photomultipliers (SiPMs) are a promising option for future $e^{+}e^{-}$ Higgs factories, where both excellent energy resolution and a very large dynamic range are required. In this work, the non-linear response of high-pixel-density SiPMs with pixel pitches of 6–10~$μ$m coupled to BGO and BSO crystals is studied under realistic beam conditions. A dual-end readout scheme with an attenuated reference SiPM was employed to precisely calibrate the deposited energy and the corresponding number of photoelectrons over a wide dynamic range. Beam tests were carried out at the CERN SPS H2 beamline using high-energy electrons, with a tungsten pre-shower and variable incident angles to enhance energy deposition. The measurements directly quantify the non-linear response of SiPMs to scintillation light over an extended dynamic range. For BGO-coupled Hamamatsu SiPMs, deviations from linearity of about 20% are observed at $5\times10^{5}$ photoelectrons, while larger deviations are measured for the tested NDL devices and for configurations with faster BSO scintillation.


💡 Research Summary

This paper presents a comprehensive experimental study of the non‑linear response of high‑pixel‑density silicon photomultipliers (SiPMs) when coupled to dense scintillating crystals, specifically bismuth germanate (BGO) and bismuth silicate (BSO). The motivation stems from the stringent requirements of future electron‑positron Higgs factories (e.g., CEPC), which demand electromagnetic calorimeters that simultaneously achieve percent‑level energy resolution and a dynamic range extending from minimum ionizing particles up to electromagnetic showers of order 180 GeV.

Because a SiPM contains a finite number of micro‑cells, its output saturates as the incident photon flux approaches the total pixel count. However, the long scintillation decay times of BGO (≈300 ns) and BSO (≈100 ns) allow individual pixels to recover and fire multiple times within a single light pulse, effectively increasing the usable dynamic range beyond the nominal pixel number.

To quantify this effect under realistic beam conditions, the authors built a dual‑end readout system. One end (the “reference” SiPM) is equipped with a neutral‑density filter that attenuates the light to ≈1 % transmission, ensuring that this channel remains linear throughout the measurement. The reference SiPM thus provides an absolute calibration of the deposited energy and the number of photons reaching the opposite, unattenuated SiPM (the “device under test”, DUT).

A tungsten pre‑shower absorber was placed upstream of the crystal to initiate shower development and increase the energy deposited in the crystal. By rotating the crystal relative to the beam, incident angles of 30°, 60°, and 90° were explored; Geant4 simulations showed that smaller angles substantially increase the absorbed energy while reducing the optimal pre‑shower thickness.

Four SiPM models were tested: two Hamamatsu devices (pixel pitch 10 µm, active areas 3 × 3 mm² and 6 × 6 mm²) and two NDL devices (pixel pitches 6 µm and 10 µm). They were coupled to three crystal bars (a large BGO bar 40 × 1.5 × 1.5 cm³ and two smaller 12 × 2 × 2 cm³ bars of BGO and BSO). The front‑end electronics provided simultaneous high‑gain and low‑gain channels, digitized with a 1 GHz, 1.25 GS/s oscilloscope. A charge‑injection calibration characterized the pre‑amplifier’s transfer function, which was modeled with a linear segment for small signals and an exponential segment for large signals.

Calibration proceeded in four steps: (1) pre‑amplifier charge‑to‑voltage conversion, (2) SiPM gain determination using a fast LED, (3) relative light‑collection efficiency calibration via the reference SiPM, and (4) absolute energy scale setting with minimum‑ionizing particles. After these calibrations, the DUT SiPM output could be expressed directly in photo‑electron (p.e.) counts.

Results show that for the Hamamatsu S14160‑6010PS SiPM coupled to BGO, the response deviates from linearity by about 20 % at 5 × 10⁵ p.e. The NDL devices, which have fewer pixels, exhibit larger deviations (≈30 % or more) at the same photon count. When coupled to the faster BSO scintillator, the non‑linearity is even more pronounced because the shorter decay time reduces the benefit of pixel recovery. Systematic uncertainties arise from pre‑amplifier non‑linearity (≈2 %), temperature variations (≈0.5 %/°C), light‑collection efficiency differences (≈3 %), and the modeling of the pre‑shower optimization (≈5 %).

The study demonstrates that the combination of slow‑decay crystals and ultra‑dense SiPMs can effectively mitigate pixel saturation, maintaining non‑linearity below 10 % up to photon counts of order 10⁶. This validates the feasibility of using SiPM‑readout homogeneous crystal calorimeters in future high‑energy e⁺e⁻ colliders, offering both the required dynamic range and the high energy resolution. The authors suggest further work on temperature‑compensated operation, multi‑channel cross‑calibration, and comparison with faster scintillators to fully optimize the technology for collider applications.


Comments & Academic Discussion

Loading comments...

Leave a Comment