A new benchmark for machine learning applied to powder X-ray diffraction

Area of Science:

• Crystallography
• Materials Science
• Machine Learning

Background:

• Crystal parameter prediction from powder X-ray diffraction (PXRD) is a growing area of interest for machine learning.
• Existing PXRD datasets are often private and lack structural diversity, hindering model development.

Purpose of the Study:

• To introduce a new, public, and structurally diverse dataset for crystal parameter prediction from PXRD.
• To facilitate the development of machine learning and computer vision models for PXRD data analysis.

Main Methods:

• Generated simulated powder X-ray diffraction (PXRD) patterns for 467,861 crystal structures from the Crystallography Open Database (COD).
• Created both one-dimensional diffractograms and derived two-dimensional radial images.

Main Results:

• Established the Simulated Powder X-ray Diffraction Open Database (SIMPOD), a large-scale, public benchmark dataset.
• SIMPOD contains 467,861 crystal structures and their corresponding simulated PXRD patterns.

Conclusions:

• SIMPOD addresses the need for public and diverse datasets in PXRD analysis.
• The dataset is expected to accelerate the development of advanced models for materials analysis using PXRD.