Open Access Review Article

Machine Learning Approach for Prediction of Glass- Forming Ability in Bulk Metallic Glasses

Eunsung Jekal*

Department of Materials, ETH Zurich, 8093 Zurich, Switzerland

Corresponding Author

Received Date: December 06, 2019;  Published Date: December 10, 2019

Abstract

Predicting the glass forming ability (GFA) to shape glass by changing the composition of alloys is not only an old conundrum of glass physics, but also an industry problem that is having a huge financial impact. Over the decades, other empirical guides for GFA predictions have been established, but efficient forecasting of good glass makers requires a comprehensive model or approach that can address as many variables simultaneously as possible. Here, we apply the support vector classification method to develop a model for predicting the GFA of binary metal alloys in a random configuration. The effect of the other inputs on the GFA was evaluated to select the optimal predictive model, indicating that the information related to liquid temperature plays a key role in the GFA of the alloy. On the basis of this model, excellent glass formers can be predicted with high efficiency. Better choice of larger databases and sophisticated input descriptors can further improve predictive efficiency. Our research results show that machine learning is very powerful and efficient and that GFA has great potential for discovering new metal glasses that are good.

Citation
Signup for Newsletter
Scroll to Top