Algorithm that directly generates crystal structures from X-ray diffraction without database

A study conducted by Joohwi Lee et al. was published in npj Computational Materials.

In the field of materials science, the identification of crystal structures using X-ray diffraction patterns (XRD) is a fundamental task, because the material properties are determined by the type of constituent elements and the arrangement of atoms. Researchers have been traditionally relying on established databases to find matching XRD patterns to determine crystal structures. However, a significant challenge arises when dealing with novel and complex materials for which no similar XRD pattern exists in the database. In response to this challenging issue, this study introduces an innovative approach: a crystal structure creation scheme designed to reproduce a given XRD pattern. This method, which is named "Evolv&Morph", is an inverse design strategy consisted of evolutionary algorithm and crystal morphing. The effectiveness of this method is demonstrated across sixteen distinct crystal structure systems, encompassing both simulated and real-world powder target XRD patterns. Beyond its immediate utility in crystal structure determination, Evolv&Morph holds promise as a versatile tool for materials design. By allowing researchers to tailor materials to exhibit specific properties, this approach opens doors to the creation of novel materials with desired specific characteristics.

Our code is available at
https://github.com/ToyotaCRDL/EvolvMorph

Title: Creation of Crystal Structure Reproducing X-ray Diffraction Pattern without Using Database
Authors: Lee, J., Oba, J., Ohba, N., Kajita, S.
Journal Name: npj Computational Materials
Published: August 5, 2023
https://doi.org/10.1038/s41524-023-01096-3