Introduction

Understanding and predicting ionic transport in disordered materials is critical for the development of next-generation solid-state batteries, yet the complex and heterogeneous nature of glasses presents significant challenges. 

Research Approach

This thesis addresses the atomic-scale origins of ion mobility in glassy systems through a combination of molecular dynamics simulations, machine learning, and advanced structural representations.

Core Methodology

At the core of this work is the development and application of interatomic potentials and predictive models to capture and interpret the relationship between glass structure and ion dynamics. To enable large-scale simulations of chemically diverse glass systems, we present generalizable frameworks for machine learning interatomic potential parameterization capable of modelling glasses.

Results

These potentials accurately reproduce experimental structural properties across many glass forming systems while supporting high-throughput investigations of their ion dynamics. Using these potentials and previously established potentials, we demonstrate that ion mobility can be predicted from the atomic structure of glasses across a range of compositions and time scales.

Key Insights

Our work reveals that fast-ion dynamics are associated with specific structural motifs, such as reduced coordination environments, larger ring-like features within the network, and higher average atomic volume, offering predictive descriptors for composition–structure–property optimization.

Conclusion

Overall, this thesis provides new insights into the fundamental mechanisms of ion transport in disordered materials and delivers computational frameworks for the rational design of high-performance glassy electrolytes and cathodes.