Characterization of afterpulse in SiPMs with single-cell readout as a function of bias voltage and fluence

TL;DR

This study investigates the afterpulse effect in silicon photomultipliers (SiPMs) using a single-cell readout structure, developing new analysis methods to measure afterpulse probability and time constants across different irradiation levels.

Contribution

The paper introduces three validated analysis methods for characterizing afterpulses in SiPMs, enabling detailed study of device behavior post-irradiation.

Findings

Afterpulse probability is below 6% for overvoltages of 3-5 V.

The afterpulse time constant is under 10 ns.

Irradiation does not significantly affect afterpulse probability or time constant.

Abstract

We present a detailed investigation of the afterpulse effect in silicon photomultipliers (SiPMs), using a dedicated structure with single-cell readout. This enables direct measurement of intrinsic device properties and observation of individual pulses even after irradiation. Three independent analysis methods to quantify afterpulse induced events were developed and validated by Monte Carlo simulations. The first method is based on charge integration, while the other two methods use multiple linear regression to reconstruct transient waveforms and accurately identify individual pulse positions. These positions are then used either as direct event counts or to construct time interval distributions, enabling comprehensive characterization of the afterpulse probability and providing insights into the dynamics of trapping in silicon. Using this framework, we measured three SiPM samples: one fresh reference device and two irradiated devices exposed to reactor neutron fluences of 2e12 and 1e13 cm^-2. We report systematic measurements of the afterpulse probability and time constant as functions of bias voltage and irradiation fluence. For overvoltages in the range of 3 to 5 V above breakdown, the afterpulse time constant is found to be below 10 ns and the afterpulse probability below 6%. Both quantities show no significant dependence on irradiation fluence.