Machine Learning Approach for Prediction and Understanding of Glass-Forming Ability

TLDR

Predicting glass‑forming ability across alloy compositions is a challenging problem with significant industrial impact, yet comprehensive models that handle many variables simultaneously are lacking. The study aims to develop and improve efficient predictive models for GFA of binary metallic alloys using machine learning, specifically support vector classification, and to enhance prediction efficiency through larger databases and refined descriptors. Support vector classification was applied to random alloy compositions to build predictive models, evaluating various input descriptors, particularly liquidus temperatures. The models identified liquidus temperature as a key descriptor and achieved high efficiency in predicting good glass formers, demonstrating that machine learning can effectively discover new metallic glasses with strong GFA.

Abstract

The prediction of the glass-forming ability (GFA) by varying the composition of alloys is a challenging problem in glass physics, as well as a problem for industry, with enormous financial ramifications. Although different empirical guides for the prediction of GFA were established over decades, a comprehensive model or approach that is able to deal with as many variables as possible simultaneously for efficiently predicting good glass formers is still highly desirable. Here, by applying the support vector classification method, we develop models for predicting the GFA of binary metallic alloys from random compositions. The effect of different input descriptors on GFA were evaluated, and the best prediction model was selected, which shows that the information related to liquidus temperatures plays a key role in the GFA of alloys. On the basis of this model, good glass formers can be predicted with high efficiency. The prediction efficiency can be further enhanced by improving larger database and refined input descriptor selection. Our findings suggest that machine learning is very powerful and efficient and has great potential for discovering new metallic glasses with good GFA.