Machine Learning in Sustainable Materials

We develop next-generation machine-learning frameworks to enable the predictive modeling of complex materials across various length and timescales. Our interests span machine-learned interatomic potentials, ML-powered coarse-grained simulations, and multimodal models that combine simulation and experimental data to accelerate materials discovery. We actively collaborate with experimentalists to translate our computational insights to scalable, impact-driven materials solutions.

One of our research focuses is on understanding and designing efficient electrochemical energy storage materials through molecular-level modeling of electrode–electrolyte interfaces and advanced electrolyte systems. We study polymer, aqueous, and liquid electrolytes to uncover the structure–dynamics relationships that govern stability, ion transport, and performance in batteries. We integrate physics-based simulations with data-driven approaches, and aim to enable the rational design of safe, efficient, and next-generation energy storage technologies.

We investigate how the immediate chemical environment surrounding charged ions and water molecules influences the rate and efficiency of energy conversion reactions. By using molecular simulations, we uncover how ion–water interactions influence reaction pathways at electrochemical interfaces. Our work focuses on key clean-energy reactions, such as carbon dioxide conversion and green hydrogen production, to guide the design of more efficient and sustainable catalysts.