Machine-learning-assisted modeling of alloy ordering phenomena at the electronic scale through electronegativity.

This article focuses on modeling the ordering phenomena in high-entropy alloys (HEAs) using machine learning (ML) based on electronegativity scales. By constructing a large dataset of 4,000 alloy compositions characterized by features derived from ten different electronegativity scales, the study employs a decision forest algorithm to predict the alloy's ordered state with 94% accuracy. The model interprets ordering as a function of electron density fluctuations, particularly emphasizing the second central moment of electronegativity, with the Nagle electronegativity scale identified as highly informative. This approach demonstrates that complex ordering in multicomponent alloys can be effectively described through ML models linking electronic-level features to macroscopic alloy properties, offering insights for alloy design without relying solely on traditional phenomenological models.