Accurate Fe-He machine learning potential for studying He effects in BCC-Fe

Nanostructured Ferritic Alloys (NFAs) containing oxide nanoparticles are prospective advanced structural materials in future fusion reactors. Lacking fusion neutron sources with an adequate flux, modeling is crit-ical to understand radiation damage in the NFAs, including helium effects. Machine learning interatomic potentials (MLPs) have been demonstrated to be capable of describing atomic interactions in chemically complex systems with comparable accuracy to density functional theory (DFT) but with much less com-putational costs. Hence, MLPs are chosen to study He bubble accumulation in chemically complex NFAs. As a preliminary step, we develop a Fe-He potential, based on-10,0 0 0 atomic configurations. The de-veloped MLP accurately predicts the bulk properties of BCC-Fe compared to DFT, including the lattice constant (2.832 A), elastic constants ( c 11 = 271 GPa, c 12 = 141 , c 44 = 93 ), and phonon frequencies (max-imum error < 3.5%). The MLP predicts He binding energies in Hen V bubbles and He n clusters with a mean absolute error (MAE) of only-70 meV, which is-3 to-5 times smaller than the MAE of bind-ing energies estimated using empirical potentials. The MLP also accurately predicts the migration barrier of interstitial He (64 meV). Furthermore, the MLP correctly predicts the maximum number of He atoms (critical size) at which a Hen V bubble ( Hen cluster) can kick-out a self-interstitial atom (SIA), forming a bigger Hen V 2 bubble ( Hen V bubble). Subsequently, we use the MLP to perform molecular dynamics sim-ulations to investigate the effect of temperature on the critical size of Hen V bubbles and He n clusters. We find that the critical size decreases with increasing temperature - indicating that smaller bubbles and clusters can kick-out an SIA at higher temperatures, thereby escalating the He bubble growth. The current Fe-He MLP can be further developed to include all chemical interactions in NFAs. (c) 2022 Elsevier B.V. All rights reserved.

Automated summary

Nanostructured Ferritic Alloys (NFAs) containing oxide nanoparticles are potential advanced structural materials for fusion reactors. Due to the lack of fusion neutron sources with sufficient flux, modeling is crucial for understanding radiation damage and helium effects in NFAs. Machine learning interatomic potentials (MLPs) are chosen to study helium bubble accumulation in NFAs.