Monte Carlo Simulation of a Neutron Detector for Orientation Measurement of Isotope Neutron Source

The orientation measurement of isotope neutron sources is critical in nuclear security and safeguards, particularly for monitoring special nuclear materials. This paper introduces a cubic neutron detection device composed of six planar neutron detectors for orientation detection in neutron source localization. A comprehensive Monte Carlo simulation was conducted, and the results showed that the planar neutron detector, based on CdTe/CdS semiconductor layers and 6Li foils, achieved a thermal neutron detection efficiency of approximately 50%. The orientation of the neutron source can be inferred from the relative neutron count ratios among the six planar detectors.To improve the measurement precision, machine learning algorithms, including k-nearest neighbor, random forest algorithm, and multi-layer perceptron were implemented using simulated data from a 252Cf neutron source at varying distances and orientations. The system achieved an angular classification accuracy exceeding 90% at 5-deg intervals, with regression-based models yielding a root-mean-square error of approximately 1 deg, confirming the effectiveness of the proposed approach for neutron source orientation detection.