Machine learned feature identification for predicting phase and Young's modulus of low-, medium- and high-entropy alloys

Abstract The growth in the interest and research on high-entropy alloys (HEAs) over the last decade is due to their unique material phases responsible for their remarkable structural properties. A conventional approach to discovering new HEAs requires scavenging an enormous search space consisting of over half a trillion new material compositions comprising of three to six principal elements. Machine learning has emerged as a potential tool to rapidly accelerate the search for and design of new materials, due to its rapidity, scalability, and now, reasonably accurate material property predictions. Here, we implement machine learning tools, to predict the crystallographic phase and Young's modulus of low-, medium- and high-entropy alloys composed of a family of 5 refractory elements. Our results, in conjunction with experimental validation, reveal that the mean melting point and electronegativity difference exert the strongest contributions to the phase formation in these alloys, while the melting temperature and the enthalpy of mixing are the key features impacting the Young's modulus of these materials. Additionally, and more importantly, we find that the entropy of mixing only negligibly influences the phase or the Young's modulus, reigniting the issue of its actual impact on the material phase and properties of HEAs.