Reliable and scalable synthesis of newly identified structural and functional materials is the main bottleneck for realizing new technologies. We use a unique combination of simulation-based "computational synthesis" with deep reinforcement learning to predictively synthesize novel materials with extraordinary properties.

Materials expose new functionality and hidden phases when driven out of their equilibrium states. We integrate electronic structure theory, machine-learning simulations and analysis with ultrafast pump-probe experiments to understand and control materials under photo-excitation, shock and impact loadings.

Processes at material surfaces and interfaces control their mechanical, chemical and electronic behavior. We develop high-fidelity learning-driven multi scale models in conjunction with surface characterization to construct digital twins for tribology, reactivity, and electrochemistry.