Lawrence Livermore National Laboratory (LLNL) scientists have devised a groundbreaking method that combines machine learning and X-ray absorption spectroscopy to accelerate the characterization of materials. In a recent study published in Chemistry of Materials, LLNL researchers Wonseok Jeong and Tuan Anh Pham introduced an innovative approach to predict the structure and chemical composition of heterogeneous materials, focusing on amorphous carbon nitrides.
The research sheds light on the intricate local atomic structure of these materials and marks a crucial milestone in establishing a rapid characterization framework for complex materials. The atomic structure analysis of heterogeneous materials, like carbonaceous residues from high explosives detonation, has been a challenging task for scientists due to its labor-intensive nature and reliance on empirical parameters.
To tackle this challenge, the LLNL team developed machine-learning potentials to efficiently navigate the vast configuration space of amorphous carbon nitrides. This neural-network-based model helps identify representative local structures within the material, offering insights into the evolution of these structures with varying chemical compositions and densities.
By combining these machine-learning potentials with high-fidelity atomistic simulations, the researchers established correlations between local atomic structures and spectroscopic data, allowing them to interpret experimental X-ray absorption near-edge structure (XANES) data effectively. This correlation forms the basis for extracting essential chemical information from complex spectra.
Jeong, the first author of the paper, highlighted the study’s objective of addressing the long-standing challenge of characterizing disordered materials by integrating computational methods with experimental techniques. Meanwhile, Pham, the project’s principal investigator, emphasized the approach’s potential for predicting elemental speciation in diverse material systems, such as carbonaceous residues, to enhance detonation models.
The collaborative effort among researchers from various disciplines underscores the interdisciplinary nature of LLNL research. This innovative approach not only advances our understanding of materials but also paves the way for real-time interpretation of spectroscopic measurements across different material classes.
The study’s findings represent a significant advancement in materials science, providing a robust framework for elucidating the atomic speciation of disordered systems. The versatility of the approach allows for its adaptation to investigate other material classes and experimental characterization methods, indicating potential for technological innovation and scientific discovery.
Co-authors of the paper include Wenyu Sun, Marcos Calegari Andrade, Liwen Wan, Trevor Willey, and Michael Nielsen, contributing to a comprehensive exploration of materials design and characterization frontiers through innovative methodologies and collaborative research efforts.
