|About this Abstract
||Materials Science & Technology 2019
||Ceramics and Glasses Simulations and Machine Learning
||Exploring Molecular Dynamics Descriptors to Improve Machine Learning Predictions of Glass Forming Ability
||Benjamin Thomas Afflerbach, Lane Schultz, Izabela Szlufarska, Dane Morgan
|On-Site Speaker (Planned)
||Benjamin Thomas Afflerbach
Molecular dynamics simulations can give extensive information on atomic behavior in both the supercooled liquid and glassy state. In this work we explore the use the descriptors from molecular dynamics simulations for predicting critical cooling rate, Rc, a key quantity for assessing glass forming ability. To assure that the Rc values and descriptors are consistent, both are calculated from the same interatomic potential, where the Rc values are assessed by identifying the onset of crystallization during rapid cooling simulations. We have generated a database of ~100 Rc values from various compositions of 11 binary metallic alloys and have calculated a range of descriptors previously proposed in literature. We explore how these descriptors relate to the Rc values using machine learning regression models that predict Rc as a function of the descriptors.