Machine Learning Prediction of Critical Cooling Rate for Metallic Glasses From Expanded Datasets and Elemental Features

Abstract

We use a random forest model to predict the critical cooling rate (RC) for glass formation of various alloys from features of their constituent elements. The random forest model was trained on a database that integrates multiple sources of direct and indirect RC data for metallic glasses to expand the directly measured RC database of less than 100 values to a training set of over 2,000 values. The model error on 5-fold cross validation is 0.66 orders of magnitude in K/s. The error on leave out one group cross validation on alloy system groups is 0.59 log units in K/s when the target alloy constituents appear more than 500 times in training data. Using this model, we make predictions for the set of compositions with melt-spun glasses in the database, and for the full set of quaternary alloys that have constituents which appear more than 500 times in training data. These predictions identify a number of potential new bulk metallic glass (BMG) systems for future study, but the model is most useful for identification of alloy systems likely to contain good glass formers, rather than detailed discovery of bulk glass composition regions within known glassy systems.