Affiliation：Graduate School of Engineering Science, Osaka University

Abstract：A general-purpose neural network interatomic potential (NNIP) for the α-iron is presented aiming for understanding the mechanism of deformation of BCC iron at the atomic scale. It is trained using an extensive reference database produced by density functional theory (DFT) calculations. Beyond the properties of perfect α-iron crystal, this new NNIP can describe the interactions between various defects in α–iron, including vacancies, surfaces, grain boundaries, and dislocations. Our demonstrations show that this NNIP can be applied to, but not limited to, the study of dynamic behaviour of the various defects and these interaction in systems in first-principles computation accuracy.