2/4/2025 | 3:30 PM - 5:30 PM | Regency B

Coarsening materials microstructures: Evolving information complexity and a role for machine learning

Abstract

The temporal evolution of microstructural features in metals and ceramics has been the subject of intense investigation over many years because deviations from normal grain growth behavior that lead to abnormal grain growth (AGG) are ubiquitous and strongly dictate observed mechanical and magnetic properties. In this talk, I examine the kinetics of evolving synthetic and experimental microstructures, as quantified by their embodied information, to identify temporal signatures of the onset of AGG.  This is accomplished by tracking the information content of coarsening microstructures via selected metrics and measures of shared information and interaction strength.  As the information content of a system is a proxy for the entropy, a thermodynamic description of grain growth is also outlined.  Finally, I describe how machine learning strategies focusing on salient features permit one to identify precursors of AGG and thereby allow the prediction of such rare events well in advance of their occurrence.