| About this Abstract |
| Meeting |
MS&T24: Materials Science & Technology
|
| Symposium
|
Advanced Characterization of Materials for Nuclear, Radiation, and Extreme Environments V
|
| Presentation Title |
Machine Learning and Molecular Dynamics-Coupled X-Ray Absorption Spectroscopy for Disordered Multicomponent Systems |
| Author(s) |
Stephen Lam, Nicholas Marcella, Omar Oraby, Rajni Chahal, Santanu Roy, Vyacheslav Bryantsev, Anatoly Frenkel |
| On-Site Speaker (Planned) |
Stephen Lam |
| Abstract Scope |
Disordered multicomponent materials are used as advanced fluids, waste storage media, structural ceramics, and more. Yet, interrogating their local structure and chemistry (multimodal bonding distributions across multiple species) remains a formidable challenge. While X-ray absorption spectroscopy (XAS) can probe specific elements in the material, analysis of its spectra is still challenging when multiple coordination complexes are present. Meanwhile, simulating structurally diverse systems with ab initio molecular dynamics (MD) is also extremely challenging due to high computational cost. Here, we overcome these limitations by utilizing machine learning (ML) to 1) accelerate MD to enhance sampling, and 2) train ML models that directly extract information from XAS spectra. This method is deployed to analyze experimental XAS data for NaF-ZrF4 and NiCl2-KCl-ZnCl2 molten salts. Coordination complexes beyond the first solvation shell for the first time, presenting a significant advance over current XAFS analysis methods and a pathway towards real-time monitoring of disordered materials. |