Pxrd-bench advances crystal structure discovery from x-ray data
PXRD-Bench: How Close Are We to Determining Crystal Structures from Powder Diffraction?
Artificial Intelligence
Summary
Figuring out the arrangement of atoms in a crystal from powder X-ray pictures is tricky but important for studying new materials. The authors created PXRD-Bench, a big test set to better measure how well computer models can do this, including for mixtures and unknown materials. They also built a strong model called Bragglite that improves accuracy compared to earlier attempts. Their work shows where current methods work well and where challenges remain, especially with mixtures and unseen crystal types.
What this means in practice
- •For materials scientists: Automatically identify crystal structures of new compounds from their powder X-ray diffraction data, even without prior composition knowledge.
- •For chemical manufacturing engineers: Improve material characterization workflows by detecting multiple crystal phases in mixtures from PXRD measurements using data-driven models.
Authors
Dongyeop Woo, Nofit Segal, Nayoung Kim, Minkyu Kim, Rafael Gomez-Bombarelli, Sungsoo Ahn
Abstract
Determining crystal structures from powder X-ray diffraction (PXRD) would enable routine characterization of newly synthesized materials, including those absent from reference databases. Generative models offer a promising approach, but existing evaluations remain fragmented and largely assume single-phase samples with known composition. To fill this gap, we introduce PXRD-Bench, a unified benchmark that extends evaluation to unknown per-element atom counts and recovery of both structures in two-phase mixtures. It provides approximately 4.1 million training structures with simulated PXRD, tests generalization to unseen compositions and structural prototypes, and evaluates recovery from experimental PXRD after excluding structurally matching training entries. To provide a strong reference baseline, we also introduce Bragglite, a PXRD-to-structure model developed through a systematic study of PXRD modeling choices. It achieves a 44.2\% Top-1 match rate on experimental RRUFF patterns with known composition, 24.6\% above the strongest prior baseline. Further analysis yields several key findings: some models fail to benefit from experimental PXRD, while generalization to unseen structural prototypes and recovery from two-phase mixtures remain challenging. Together, our benchmark, systematic studies, and Bragglite clarify the capabilities and limitations of current PXRD-to-structure models.