Score-based diffusion models for accurate crystal-structure inpainting and reconstruction of hydrogen positions

TL;DR

This paper introduces a novel application of score-based diffusion models combined with computer vision techniques to accurately inpaint and reconstruct hydrogen positions in crystal structures, significantly improving over previous methods.

Contribution

The authors develop a cross-domain approach that leverages diffusion models and image inpainting techniques for precise crystal-structure reconstruction, especially hydrogen positions.

Findings

Achieved over 97% success rate in structural matching or stability prediction.

Outperformed unconditioned diffusion models and DFT-based approaches.

Enabled faster and more accurate hydrogen atom placement in crystal structures.

Abstract

Generative AI models, such as score-based diffusion models, have recently advanced the field of computational materials science by enabling the generation of new materials with desired properties. In addition, these models could also be leveraged to reconstruct crystal structures for which partial information is available. One relevant example is the reliable determination of atomic positions occupied by hydrogen atoms in hydrogen-containing crystalline materials. While crucial to the analysis and prediction of many materials properties, the identification of hydrogen positions can however be difficult and expensive, as it is challenging in X-ray scattering experiments and often requires dedicated neutron scattering measurements. As a consequence, inorganic crystallographic databases frequently report lattice structures where hydrogen atoms have been either omitted or inserted with heuristics or by chemical intuition. Here, we combine diffusion models from the field of materials science with techniques originally developed in computer vision for image inpainting. We present how this knowledge transfer across domains enables a much faster and more accurate completion of host structures, compared to unconditioned diffusion models or previous approaches solely based on DFT. Overall, our approach exceeds a success rate of 97% in terms of finding a structural match or predicting a more stable configuration than the initial reference, when starting both from structures that were already relaxed with DFT, or directly from the experimentally determined host structures.