Capturing Short-Range Order in High-Entropy Alloys with Machine Learning Potentials

Computational investigations of chemistry-microstructure relationships require atomistic models that act at the appropriate length scales while capturing chemical-bond intricacies, such as short-range order (SRO). Here we consider various approaches for the construction of training data sets for machine learning potentials (MLPs) for metallic alloys and evaluate their performance in capturing SRO and its effects on materials quantities of relevance for microstructural evolution. Based on this analysis we systematically derive design principles for the rational construction of MLPs that capture SRO. The resulting approach is demonstrated to have high physical fidelity by comparing the predictions directly to experimental measurements, such as enthalpy of SRO formation and SRO domain size.