Study on the factors affecting solid solubility in binary alloys: An exploration by Machine Learning

Abstract The formation of solid solution alloy systems happens to two kinds of atoms with similar radii to comply with Hume-Rothery rules as a common feature. In recent years, as a useful tool, Machine Learning (ML) has been widely used in material science research to obtain useful information, including material preparation and process and so on. In this work, we use the method of Support Vector Machine (SVM) to predict solid solubility with a small dataset in order to provide evidence for the correctness of the Hume-Rothery rules and find factors of solid solubility. The results indicate that the main factors of solid solubility include three traditional Hume-Rothery factors. The solid solubility can also be evaluated by Support Vector Regression (SVR) with Radial basis function (RBF) kernel.