Optimization of pulse shape discrimination based on frequency domain for SiPM-based NaIL scintillator detector

Frequency gradient analysis (FGA) neutron-gamma pulse shape discrimination (PSD) based on Fourier transform has proved to be an effective method for discrimination between two types of pulses by simply constructing a discrimination parameter by subtracting the amplitude of the 1st frequency component from the 0th frequency component. However, the discrimination parameter constructed by the above method could not meet the requirements at some lower sample rates, and the dark noise caused using optoelectronic devices such as miniaturized SiPMs array further erodes the difference between the two frequency components. This paper proposes a new discrimination parameter construction method based on FGA for neutron-gamma discrimination, which is referred as mFGA (modified FGA). This discrimination parameter uses the 0th frequency component and other frequency components to enhance the discrimination performance. A neutron detector module consisting of a 25.4 mm× 25.4 mm× 50.8 mm NaI:Tl,⁶Li (NaIL) scintillator and an array of 8 × 8 SiPMs was established to verify the method of mFGA. The module measured a ²⁵²Cf neutron source under three shielding conditions: no shield, 5 cm high-density polyethylene (HDPE), and 10 cm lead brick. The proposed mFGA method was used to calculate the figure of merit (FoM) under each shielding condition and compared to the traditional frequency gradient method (FGA) and the traditional charge comparison method (CCM). The results show that the discrimination performance with the mFGA method is significantly improved. Taking the HDPE shielding group as an example, the performances of the three methods are FoM-FGA =1.14, FoM-mFGA =2.75, FoM-CCM =0.92. At the same time, with the increase of energy, due to the neutron gamma pulse becoming similar, and the discrimination performance of FGA and CCM decrease significantly. The discrimination performance of the mFGA method is relatively stable in all energy regions. Therefore, the proposed method is superior to the traditional FGA of the 0th frequency minus the 1st frequency and has a good application prospect for neutron-gamma discrimination with detectors of poor signal quality.