Radiation damage and recovery of SiPMs exposed to low-momentum muon beams for innovative scintillation-based muon beam monitors

Introduction

The ISIS Neutron and Muon Source features two different muon facilities[1], with a total of 7 beamlines, as shown in Figure 1. The EC muon facility is capable of delivering surface $\mu^{+}$ beams, coming from the decay of $\pi^{+}$ at rest in the production target, with high intensity and fixed momentum around 29 MeV/c. On the other hand, RIKEN-RAL is a decay muon facility: in addition to surface muons, it features a decay solenoid, which transports $\pi^{+/-}$ and allows them to decay in flight. For this reason, muons of both polarities and in momentum range 17-120 MeV/c can be delivered by RIKEN-RAL. Figure 2 shows the values of beam intensity and momenta available at RIKEN-RAL.

Image not extracted

Image not extracted

Several groups have developed beam monitors for muon beamlines. The most used design for low-momentum beams is a set of two planes of square scintillating fibres, either in a dense plane geometry[2, 3, 4], or with gaps between fibres[5, 6, 7, 8]. The first geometry allows for the highest spatial resolution, but it can be more difficult to assemble and it can interfere with the beam or block it completely; the second one is less invasive, but it needs thorough calibration to extract useful information, and in general has worse spatial resolution.

A new geometry for scintillator-based beam monitors is being designed and tested as part of a collaboration between Pavia research institutions (University, INFN and CNAO) and ISIS. It basically involves a single thin scintillator read out by SiPMs in a way to reconstruct the position and shape of the scintillation site, hence the beam spot.

The easiest geometry, presented in this paper, consists of a thin scintillator coupled with a 2D SiPM matrix. However, with this setup the SiPMs are exposed directly to the muons and might go through a radiation damage process[9, 10]. In particular, muons are expected to cause lattice defects (displace ments), increasing the number of generation-recombination centres, which eventually yields an increase of the dark count rate, hence an increase of the dark current. Thermal annealing (e.g. 24 hours at 125 °C) allows partial recombination of lattice defects, with an expected reduction of the measurable damage in terms of dark current and dark count rate. In addition, muons are not expected to modify the junction effective doping during their interaction with the nuclei, therefore, no variation in the breakdown voltage is expected. Some preliminary evidence of radiation damage of SiPMs exposed to low-momentum muon beams is presented in this work, even though extensive testing is still ongoing.

Test beam with muons at ISIS RIKEN-RAL Port4

Three days of beamtime at RIKEN-RAL Port4 were granted to carry out the studies presented in this work.

A set of Hamamatsu S14161-3050HS-08 matrices, each consisting of 64 SiPMs (8 $\times$ 8) with $3 \times3$ mm$^{2}$ active area and 50 $\mu$m cell, was used for the test. Each SiPM mounted a SAMTEC SS4-40-3.00-L-DK-TR connector to plug it into the prototype PCB, which was connected to the CAEN DT5550W-A55PET2 voltage supply and readout system through a cable. This board can operate up to 64 channels providing the bias and front-end readout through two ASIC PETIROC2A. In particular, each channel operates directly on the analog data through two simultaneous operational branches: a fast shaper for triggering (intrinsic resolution $\sim37$ ps) and a slow shaper which feeds its output to the FPGA for energy and time estimation. These values are then stored. Figure 4 shows the setup mounted on the beamline.

The setup has initially been scanned using decay $\mu^{-}$ of various momenta to study the energy deposit in the layers of the prototype. This study can be done with Muonic atom X-ray Emission Spectroscopy (μ-XES)[11, 12, 13, 14] [8], which consists of a bulk elemental characterisation of a sample by measuring the X and $\gamma$ rays emitted after $\mu^{-}$ irradiation. Negative muons stopped in a target form muonic atoms, emitting characteristic X-rays, and can then be captured by the nucleus emitting characteristic $\gamma$ rays, all used for elemental identification. As the muon energy loss is described by the Bethe-Bloch formula, it loses energy with a Bragg peak. For this reason, scanning a sample as a function of momentum lets experimenters understand its composition at the studied layer. In our case, scanning the prototype lets us find the momentum at which the energy deposit is maximised in the SiPM.

At RIKEN-RAL Port4, the scan was done using the MuX facility germanium detectors. A result of the momentum scan between 28 MeV/c and 40 MeV/c is shown in Figure 3. The prototype detector was then tested with a beam of negative muons at 50 MeV/c and a circular collimator with radius of 8 mm, resulting in the preliminary beam profile shown in Figure 5.

Image not extracted

Image not extracted

In addition, the SiPM models listed in Table 1 were tested for radiation damage effects, by exposing them to the same fluence and the same kind of muons. The maximum deposit in Si inside the plastic case (see Figure 3) is at 30–31 MeV/c, almost twice the deposit at $29\ \mathrm{MeV}/c$. However, surface muon intensity at $29\ \mathrm{MeV}/c$ is two orders of magnitude greater than that of $30$-$31\ \mathrm{MeV}/c$ decay muons. It has therefore been decided to expose SiPMs to surface muons. The beam profile can be assumed uniform over the measured beam spot ($\sim36 \times24\ \mathrm{mm}^{2}$) with intensity $10^{6}\ \mu^{+}/\mathrm{s}$, and each SiPM was exposed to the beam for 13 hours at full intensity. For this reason, the estimated fluence on each SiPM ($9\ \mathrm{mm}^{2}$) is roughly $\sim5.4 \cdot10^{9}\ \mu^{+}/\mathrm{cm}^{2}$, corresponding to $\sim5 \cdot10^{8}$ muons deposited on each SiPM. In addition, as many muons are expected to stop in the SiPM active area, possibly in the depletion region, a contribution to the dose is also given by positrons emitted during the muon decay, causing further damage.

Manufacturer	Model	Cell size ($\mu$m)
Hamamatsu	S14160-3050PS	50
Hamamatsu	S13360-3025PE	25
Hamamatsu	S13360-3050PE	50
Hamamatsu	S13360-3075PE	75
AdvanSiD	NUV	40
Onsemi (SensL)	C-series 3 mm	35
Broadcom	AFBR-S4N44C013	30
KETEK	PM3325-WB-D0	25

Image not extracted

SiPM irradiation was carried out at night, while the detector prototype testing was carried out during the day.

Each SiPM was characterised in terms of $I$-$V$ characteristic curve and waveform acquisition before and after irradiation. Each acquisition was carried out in a climatic chamber at 25 °C, the results are discussed in the next Section along with the thermal annealing.

To monitor the ongoing damage, another SiPM Hamamatsu S14160-3050PS (50 $\mu$m cell) was exposed to the beam at operating voltage, monitoring the dark current to qualitatively check the generation of radiation-induced defects in real time. The SiPM was thoroughly wrapped and checked to exclude external light contamination. For the first two nights, the dark current kept increasing, even after the daytime without beam. During the second day, the SiPM was put in a muffle furnace at 125 °C for 12 hours to check the effect of thermal annealing. Indeed, during the third night, the SiPM had partially recovered and started from a lower dark current. The plot of dark current over time is shown in Figure 6. No quantitative information can be extracted from this plot as the SiPM was exposed without active temperature-bias correction, and the absolute values could be affected by temperature and gain drift.

Image not extracted

Image not extracted

Test of SiPM annealing

After the irradiation, all eight irradiated SiPMs have undergone an established procedure[16, 17, 18] of thermal annealing consisting of heating the SiPMs at 125 °C for 24 hours, which is a good balance to have fast annealing without breaking the devices. The SiPMs have then been tested again in terms of $I$-$V$ characteristic curve and waveform acquisition.

In this work, only the results on the $I$-$V$ characteristic curves are provided, as work on the dark count rate estimation is still ongoing. The breakdown voltage is estimated from the $I$-$V$ curve by fitting $\sqrt{I(V)}$ with two linear regressions in the leakage current region and in the first part of the dark current region, after the breakdown voltage, which is defined from the intersection of the two lines[10, 4].

Figure 7 shows a comparison for all SiPMs of the breakdown voltage before irradiation, after irradiation and after annealing. No clear change is observed, within the calculated tolerances, in the three phases. This confirms that no pn junction effective doping modification is expected to be observed for positive muon irradiation. On the other hand, nuclear negative muon capture ($\mu^{-} p \rightarrow n \nu_{\mu}$) could in principle transmute Si or dopant nuclei and modify the pn junction.

Image not extracted

However, post-irradiation and post-annealing data show a small but systematic increase in breakdown voltage with respect to unexposed SiPMs, suggesting that either some effect is caused by muons or secondary particles, or some residual neutrons are also transported in the final part of the beamline. At RIKEN-RAL Port4, the CHNet-MAXI experiment[4] is currently testing CLLB scintillators, optimised to discriminate neutrons[15], which could help in understanding the cause of this effect, if observed systematically.

Conclusion

In order to monitor low-momentum ($\lesssim40$ MeV/c) muon beams with a dense detector, the research and development on thin scintillating detectors is ongoing.

The easiest geometry consists of a thin scintillator read out with a 2D SiPM matrix, for which the feasibility was tested.

The main downside of this geometry is that the SiPMs are exposed directly to the muon beam.

For this reason, various SiPM models were exposed to the RIKEN-RAL surface muon beam at ISIS, indicating that the damage is mostly recoverable through thermal annealing, which will be further investigated by our studies on the dark count rate.

Acknowledgements

The work was supported by Fondazione Angelo Della Riccia (International Exchange Grant 2024-2025 awarded to R.R.) and the Royal Society (International Exchange Grant no. IES-R2-244430 awarded to J.S.L. and A.M.), both of which are gratefully acknowledged for their contributions.

The muon beam was provided by the Science and Technology Facilities Council (STFC) through access to the ISIS Neutron and Muon Source under the CNR-STFC Agreement 2021–2027. Beamtime was allocated under proposal RB2520455.²

Roberto Bertoni (INFN Milano-Bicocca) and Sun Wen (University of Nanjing) are gratefully acknowledged for the technical support.

References