Rosen Research Group
Research
Our research program is organized around two major questions at the intersection of chemical engineering and materials science.
What governs the stability of a material and the series of reactions that lead to its decomposition?
Shared foundation
Across both areas, we develop the datasets, benchmarks, machine learning models, and open-source software needed for accurate and reliable simulation across the computational materials science community.
Research area 01
Electronic structure and emergent reactivity
Our group uses quantum-mechanical simulations and machine learning to uncover how unusual electronic configurations in solid-state materials can give rise to new forms of chemical reactivity. In particular, we are interested in studying how spatially confined electrons in solid-state materials activate otherwise strong chemical bonds at mild conditions. To achieve these goals, we also develop new machine learning algorithms to model the spatiotemporal flow of electrons during chemical reaction events.
Research area 02
Stability and decomposition
We are developing and applying computational methods to identify what factors govern the stability of solid-state materials and to uncover the reaction mechanisms that drive their decomposition when pushed beyond their limits. Through this research area, we are developing and building upon foundation machine-learned interatomic potentials to describe how solid-state materials respond to high temperature and high pressure conditions. In doing so, we seek to capture the complex reaction dynamics of decomposition and the factors governing phase stability.
Crosscutting area
Machine-learned interatomic potentials
Machine-learned interatomic potentials extend quantum-mechanical accuracy toward larger length and time scales, but their reliability depends on the data, benchmarks, and computational infrastructure behind them. We develop training datasets, experimentally grounded benchmarks, higher-fidelity models, and reproducible open-source workflows for the broader materials modeling community.