ANN-based pulse shape discrimination for background reduction in neutron depth profiling

This study presents an artificial neural network (ANN)-based pulse shape discrimination (PSD) algorithm developed to enhance the accuracy of neutron depth profiling (NDP) analysis in solid-state electrolytes (SSEs). A key innovation of the proposed approach lies in effectively suppressing low-energy background signals, which typically compromise analytical precision. To overcome labeling challenges in the low-energy region, the training data was augmented with white noise, enhancing the ability of the model to discriminate the emulated low-energy signals. Compared with the conventional rise-time method, the ANN-based PSD algorithm demonstrated a 77 % higher area under the curve value indicating higher discrimination performance. Furthermore, explainable artificial intelligence techniques were employed to interpret and evaluate the performance of the algorithm. This underscored the reliability and potential of ANN-based PSD in advancing the high precision of NDP analysis of SSEs.