SODAS—Autonomous Characterization of Defects in High Entropy Materials Under Extreme Conditions via Graph Neural Networks

High entropy alloy processing often occurs under extreme dynamic conditions leading to a multitude of unique structural defect environments. These structural environments generally occur under rapid temperature changes, and therefore non-equilibrium conditions, which results in drastic changes in the material’s structure over time. Computational techniques such as molecular dynamics simulations can help probe the atomic regime under these extreme conditions. However, characterizing the resulting atomistic structures has proved challenging due to the intrinsic levels of disorder present. Here, we leverage our SODAS platform, a graph neural network framework that can map global state information onto local atomic environments in an autonomous manner. This paradigm represents a powerful tool to encode global information, such as the level of disorder, onto local atomic environments, providing a pathway to characterize atomistic simulations in an intuitive and interpretable manner. In this work we showcase the power of SODAS at characterizing the time-evolution of defects in high entropy alloys during simulations representing additive manufacturing processes, providing experiments with detailed information regarding how processing conditions drive defect formation.