Graphite is used widely as a moderator or reflector in the nuclear industry. Therefore, the behavior of graphite after neutron irradiation needs to be studied. The neutron irradiation damage is assumed as the PKA damage model in three-dimensional space. We used molecular dynamics to simulate the neutron irradiation of graphite. We applied various PKA directions and incident neutron energies, up to 1500 eV, to find out defect structures.
Simulation result shows four groups of defects: structural defect, energy defect, vacancy, and near-defect structure. The results were evaluated with decision tree, a supervised machine learning technique, to classify the number of defect structures.