Composition optimization of FeCrAl alloys for enhanced mechanical properties via machine learning potential molecular dynamics

Abstract

FeCrAl alloys serve as promising nuclear fuel cladding materials due to their exceptional oxidation resistance and irradiation damage tolerance. These superior properties stem from their optimal chemical composition, particularly the concentrations of Cr and Al as critical alloying elements. Consequently, exploring the compositional space of FeCrAl alloys using efficient and accurate methods is crucial for designing high-performance variants. In this work, we developed a machine learning interatomic potential for FeCrAl alloys based on the Deep Potential (DP) framework. Utilizing this potential, we performed DP molecular dynamics simulations to systematically investigate the elastic properties and uniaxial tensile behavior in FeCrAl alloys with varying compositions. These studies elucidate that the Cr content exhibits a relatively minor influence on the mechanical properties of FeCrAl alloys, while the Al content shows a significant impact instead. Based on our study, a composition of 13–14 wt. % Cr and approximately 3–4 wt. % Al is beneficial for enhancing the mechanical properties of the FeCrAl alloys.