Towards Structural Reconstruction from X-Ray Spectra

TL;DR

This paper demonstrates that machine learning can predict structural features of amorphous GeO₂ from X-ray spectra, enabling analysis of pressure-induced structural changes solely from spectral data.

Contribution

It introduces a novel approach combining machine learning and spectral analysis to infer structural information from X-ray spectra of amorphous materials.

Findings

Machine learning accurately predicts spectral moments from Coulomb matrix descriptors.

Spectral-significance-guided dimensionality reduction enables inverse mapping to structural features.

The method reproduces pressure-induced coordination changes in amorphous GeO₂.

Abstract

We report a statistical analysis of Ge K-edge X-ray emission spectra simulated for amorphous GeO$_2$ at elevated pressures. We find that employing machine learning approaches we can reliably predict the statistical moments of the K$\beta''$ and K$\beta_2$ peaks in the spectrum from the Coulomb matrix descriptor with a training set of $\sim 10^4$ samples. Spectral-significance-guided dimensionality reduction techniques allow us to construct an approximate inverse mapping from spectral moments to pseudo-Coulomb matrices. When applying this to the moments of the ensemble-mean spectrum, we obtain distances from the active site that match closely to those of the ensemble mean and which moreover reproduce the pressure-induced coordination change in amorphous GeO$_2$. With this approach utilizing emulator-based component analysis, we are able to filter out the artificially complete structural information available from simulated snapshots, and quantitatively analyse structural changes that can be inferred from the changes in the K$\beta$ emission spectrum alone.