Machine learning based approach for solving atomic structures of nanomaterials combining pair distribution functions with density functional theory

TL;DR

This paper introduces a machine learning algorithm that integrates DFT calculations with PDF analysis to determine unknown nanomaterial structures, overcoming the need for initial structural motifs and enabling the identification of metastable configurations.

Contribution

The novel approach combines machine learning, DFT, and PDF analysis for structure determination without prior motifs, advancing nanomaterials characterization.

Findings

Successfully determines unknown crystal structures from PDF data

Can identify metastable configurations in nanomaterials

Operates with on-the-fly trained machine learning model

Abstract

Determination of crystal structures of nanocrystalline or amorphous compounds is a great challenge in solid states chemistry and physics. Pair distribution function (PDF) analysis of X-Ray or neutron total scattering data has proven to be a key element in tackling this challenge. However, in most cases a reliable structural motif is needed as starting configuration for structure refinements. Here, we present an algorithm that is able to determine the crystal structure of an unknown compound by means of an on-the-fly trained machine learning model that combines density functional theory (DFT) calculations with comparison of calculated and measured PDFs for global optimization in an artificial landscape. Due to the nature of this landscape, even metastable configurations can be determined.