An Experimentally Driven Automated Machine Learned lnter-Atomic Potential for a Refractory Oxide

TL;DR

This paper presents an automated, experimental-driven machine learning approach to develop inter-atomic potentials for refractory oxides, specifically HfO2, reducing development time and effort through active learning and experimental data integration.

Contribution

It introduces a combined experimental and simulation method using active learning to efficiently generate multi-phase potentials for refractory oxides.

Findings

Successfully developed a multi-phase potential for HfO2 from room temperature to liquid state.

Significantly reduced model development time and human effort.

Demonstrated the effectiveness of experimental data-driven active learning in potential generation.

Abstract

Understanding the structure and properties of refractory oxides are critical for high temperature applications. In this work, a combined experimental and simulation approach uses an automated closed loop via an active-learner, which is initialized by X-ray and neutron diffraction measurements, and sequentially improves a machine-learning model until the experimentally predetermined phase space is covered. A multi-phase potential is generated for a canonical example of the archetypal refractory oxide, HfO2, by drawing a minimum number of training configurations from room temperature to the liquid state at ~2900oC. The method significantly reduces model development time and human effort.