We seek to connect materials exploration with models that explain and predict material behavior. This research direction combines candidate generation, atomistic modeling and property evaluation to identify promising compositions and structures for further investigation.
Generative Search of Material Candidates
Generative methods can propose structures and compositions for evaluation, while first-principles calculations provide physically grounded information about their properties. Our work includes combining diffusion-based generative models with first-principles calculations to explore metal-doped oxides for syngas conversion.
Simulation-Guided Evaluation
The value of a candidate depends on the conditions and mechanisms relevant to its intended function. We aim to connect structural search with stability, reactivity and property predictions, using transferable interatomic models where their applicability has been established.
Connecting Computation and Experiment
Our longer-term aim is to develop workflows in which simulations help prioritize informative experimental tests and new measurements guide subsequent modeling. This is an evolving research direction toward a more predictive materials-discovery process.