Using EXAFS data to improve atomistic structural models of glasses.

Daniel T Bowron

doi:10.1107/s1600577518002072

Abstract

Quantitative characterization of the atomic structure of multi-component glasses is a long-standing scientific challenge. This is because in most cases no single experimental technique is capable of completely resolving all aspects of a disordered system's structure. In this situation, the most practical solution for the materials scientist is to apply multiple experimental probes offering differing degrees of insight into a material's properties. This powerful and widely adopted approach does, however, transfer the characterization challenge to the task of developing a coherent data analysis framework that can appropriately combine the diverse experimental insight into a single, data-consistent, structural model. Here, taking a terbium metaphosphate glass as an example system, it is illustrated how this can be achieved for X-ray diffraction and extended X-ray absorption fine-structure (EXAFS) spectroscopy data, using an empirical potential structure refinement approach. This methodology is based on performing a Monte Carlo simulation of the structure of a disordered material that is guided to a solution consistent with the provided experimental data, by a series of pairwise perturbation potentials operating on a classical reference potential foundation. For multi-component glasses the incorporation of EXAFS data into the resulting bulk structural models is shown to make a critical contribution that is required to properly account for the increase in local structural order that can develop in the melt-quench process of glass formation.

Highlights

For a glass formed from N atomic components, NðN þ 1Þ=2 atomic pair correlation functions are required to properly account for the material’s microscopic structure as measured by X-ray, or neutron, diffraction
In this study only one X-ray diffraction experiment was performed to drive an atomistic modelling process, and this single data set only provides a primary constraint on the ultimate structural solution that is formed from a weighted sum of the partial structure functions
To get round this limitation the technique of empirical potential structure refinement (Soper, 1996) has been used, as the method provides the ‘missing’ information required to solve the structure, using estimations derived from a reference potential based atomistic model that forms the starting point for the structure refinement process

Summary

Introduction

For a glass formed from N atomic components, NðN þ 1Þ=2 atomic pair correlation functions are required to properly account for the material’s microscopic structure as measured by X-ray, or neutron, diffraction. Techniques have been developed that allow diffraction experiments to deliver a degree of chemically specific insight into a material’s partial pair correlation functions, for example using anomalous X-ray scattering (Fuoss et al, 1981) or neutron diffraction with isotopic substitution (Enderby et al, 1966). Both of these approaches are generally very difficult, or impossible, to apply to non-ideal or technically unsuitable samples.

Methods

Findings

Discussion

Conclusion