A comparison of small-batch clustering and charge-comparison methods for n/γ discrimination using a liquid scintillation detector

In this study, a novel neutron and γ-ray (n/γ) discrimination method based on a small-batch clustering algorithm has been developed to improve the discrimination performance of an EJ-301 liquid scintillator detector. The Principal Component Analysis (PCA) of the particle waveform shows that the main feature distribution of the original signal presents an approximate ellipse, which prompts us to adopt a clustering algorithm based on Gaussian Mixture Model (GMM). The classification ability of the small-batch clustering for discriminating n/γ pulse waveforms is verified by a mixed radiation field from a 252Cf isotope source, and then compared with the discrimination results of the charge comparison (CC) method. The results show that the proposed method based on the small-batch clustering provide a higher Figure of Merit (FoM) for n/γ discrimination compared to the CC method, in which FoM increases from 1.26, 1.30, 1.39, and 1.45 for the CC method to 1.35, 1.48, 1.52 and 1.68 respectively for the proposed method during the energy intervals 400–600 keV, 600–800 keV, 800–1000 keV, and >1000 keV for the mixed field of 252Cf. The results infer that the small-batch clustering method can be used as reference for pulse shape discrimination in neutron detection.