Abstract

Nuclear materials undergo continuous microstructural changes induced by irradiation, particularly by high-energy neutrons. These changes, which originate at the atomic scale, result in the degradation of the mechanical, electrical, thermal and optical properties of materials.

Predicting this dynamic microstructure and connecting it to macroscopic changes is one of the great challenges in nuclear materials modelling due to the vast number of variables involved, ranging from the complexity of materials to irradiation conditions such as dose, dose rate, temperature and stress gradients.

Significant advances have been achieved through the combination of molecular dynamics, kinetic Monte Carlo and cluster dynamics models. These simulations highlight the critical role of primary damage, namely the defect distribution formed on picosecond timescales.

The current status of these simulations will be reviewed, with particular emphasis on how Artificial Intelligence could contribute to the development of more accurate models through improved interatomic potentials and more efficient parameter selection using Large Language Models.