Introduction: Currently, silicon photomultiplier (SiPM) technology is used in many new radiation detection designs as an optical converter instead of the classical photomultiplier tube (PMT). SiPM consists of up to a few tens of thousands of pixels, each pixel is implemented by an avalanche photodiode. When a reverse voltage biased diode is triggered by an incoming photon or by intrinsic thermal noise, a quantity of charge is produced through avalanche breakdown. Accordingly, SiPM produces a semi-analogue output, proportional to the number of triggered diodes. SiPM technology introduces advantages such as compactness, low bias voltage and immunity to electromagnetic interference. However, as common to any semiconductor biased in a reversed voltage, the performance severely degrades by gain variations and stability issues, due to temperature changes and breakdown voltage variations [1], [2].
Methods: Fig. 1 demonstrates a system based on a 30 mm X 30 mm X 15 mm scintillator, a large area Quad-SiPM and dedicated electronics for the biasing, amplification and analysis.
Fig. 1: Detector schematics: scintillator, Quad-SiPM, electronics and MCA.
Since SiPMs are manufactured in small sizes (1mm X 1 mm to 6mm X 6 mm), a Quad 6mm X 6 mm SiPM is implemented to cover the surface of 15 mm X 15 mm scintillator. As mentioned above, practical SiPM based system must use gain over temperature stabilization. The gain is a function of the overvoltage VOV, which is a function of the bias voltage VB and the temperature dependent breakdown voltage VBR(T) [3]. The total gain is also avalanche initiation (AIP) dependent, which is itself a function of VOV.
However, the breakdown voltage is also VBR0 dependent, which is the initial VBR offset introduced due to manufacture limitation. Selected SiPM specifications are VBR = 24.5 V, with ±250 mV variations due to VBR0 and VOV range of 2.5 ÷ 6 V [4]. For small VOV variations, the gain of individual SiPM can be modeled as parabolic function of VOV (Fig. 2A). For normal distribution of VBR ~ N(µ = 24.5 V, σ = 40 mV) we can expect significant gain dissimilarity of the 4 sub-SiPM in the Quad-SiPM. By randomly generating parameters of the 4 SiPMs, the following VBR were obtained 24.516, 24.523, 24.495 and 24.472 V.
Fig. 2: (A) Gain (VOV) for the sub-SiPM; (B) Resolution (VOV) for the total Quad-SiPM.
The measured resolution, (Fig. 2B), is best at VOV of 2.7 ÷ 4.7 V. By using individual gain corrections, implemented as individual bias voltage adjustment, for matching the VOV, resolution can be improved.
Results and Conclusions:
Fig. 3: (A) Net. Peak 137Cs spectra for 4 sub-SiPMs, the total spectrum for Quad-SiPMs (black) and the enhanced spectrum (red); (B) Measured resolution (blue) and enhanced resolution (green) for Quad-SiPM.
For VOV = 4 V, each of the 4 sub-SiPM’s peak centers have different offset relative to the required 662 keV (Fig. 3A). The sum of the 4 sub-SiPMs (Quad-SiPM) is plotted as a black line. By adjusting individual bias voltage for each of the 4 sub-SiPMs, before summing, the spectra peaks are aligned so that an enhanced (red) peak is obtained. The enhanced peak (Fig. 3A) is characterized by resolution improvement from 5.5% to 5% (Fig. 3B). Overall, for small VOV, resolution enhancement of about 1.5% is achievable.