Graph-based machine learning beyond stable materials and relaxed crystal structures

TL;DR

This paper evaluates the performance of graph-based machine learning models, specifically CGCNN, on theoretical and unstable crystal structures, and explores transfer learning to improve predictions for materials discovery.

Contribution

It extends the application of CGCNN to unstable and partially relaxed structures and investigates transfer learning from related datasets to enhance model performance.

Findings

Models trained on similar data transfer effectively.

Pre-training on stable materials does not significantly improve performance.

Transfer learning reduces computational effort in phase diagram generation.

Abstract

There has been a recent surge of interest in using machine learning to approximate density functional theory (DFT) in materials science. However, many of the most performant models are evaluated on large databases of computed properties of, primarily, materials with precise atomic coordinates available, and which have been experimentally synthesized, i.e., which are thermodynamically stable or metastable. These aspects provide challenges when applying such models on theoretical candidate materials, for example for materials discovery, where the coordinates are not known. To extend the scope of this methodology, we investigate the performance of the Crystal Graph Convolutional Neural Network (CGCNN) on a data set of theoretical structures in three related ternary phase diagrams (Ti,Zr,Hf)-Zn-N, which thus include many highly unstable structures. We then investigate the impact on the performance of using atomic positions that are only partially relaxed into local energy minima. We also explore options for improving the performance in these scenarios by transfer learning, either from models trained on a large database of mostly stable systems, or a different but related phase diagram. Models pre-trained on stable materials do not significantly improve performance, but models trained on similar data transfer very well. We demonstrate how our findings can be utilized to generate phase diagrams with a major reduction in computational effort.