Research

Deep Neural Network Potentials

Our goal is to harness the power of AI/ML to enable computational models to represent reality as closely as possible, help uncover emergent phenomena, and obtain a foundational understanding of the system for the atomistic design of materials. We develop deep neural network potentials for molecular dynamics simulations of complex reactive systems in the condensed phase. We are interested in all aspects of their development, such as the efficient curation of training data, robust uncertainty estimation, and the inclusion of long-range electrostatics.

Catalyst Dynamics

The ephemeral nature of active sites has been known for a long time, yet computational models have largely focused on ideal pristine surfaces to model catalysts. We focus on applying deep learning-accelerated molecular simulations and enhanced sampling methods to uncover the interdependence between catalyst surface dynamics and the catalyzed reactions, opening up the next frontier in heterogeneous catalysis.

Environmental Geochemistry

Mineral-water interfaces play a crucial role in everything from aiding the possible origins of life and its sustenance through various environmental processes to quite possibly ensuring its survival from the adverse effects of climate change. We focus on understanding and uncovering the chemistry at mineral oxide-water interfaces using a combination of deep learning-accelerated molecular simulations and enhanced sampling methods with an emphasis on carbon capture and storage applications.

Surface and Interface Science

Surfaces and Interfaces are the "most happening" places for complex processes. Solid-liquid interfaces, in particular, are extremely challenging for both computational and experimental approaches due to the collective dynamics of atoms at the interface. We focus on characterizing the structure and chemistry at complex interfaces with an emphasis on surface hydrophobicity for applications in self-cleaning devices, and protective coatings.

Electronic Structure and Chemisorption models

The electronic state of the catalyst principally dictates the "enthalpic" component of the reaction free-energy.  We focus on identifying the surface electronic structure features that provide the optimal interaction with the adsorbates while taking into account the surface dynamics of the catalyst under operating conditions. Combining these insights with solid-state physics, we develop analytical chemisorption models for complex catalytic systems.