We Are Quantum-Chemical Engineers
The Rosen Research Group applies recent advances in quantum-mechanical calculations, high-throughput computing, and machine learning to study porous materials at the edge of stability. Many of the most useful materials do not exist in their thermodynamic ground state. Instead, they are often metastable—higher-energy, kinetically trapped structures whose unique properties can make or break their utility. Our work spans two interconnected areas:
- Materials pushed to their electronic limits: we predict how unusual electronic configurations give rise to fundamentally new forms of chemical reactivity.
- Materials pushed to their thermodynamic and kinetic limits: we identify the factors that dictate whether a material is stable, how it decomposes, and under what conditions it persists.
Our work is motivated by applications in energy and sustainability, including the design of new materials for heterogeneous catalysis, energy storage and conversion, and separation processes. We are also committed to developing open-source software and openly accessible datasets that power modern machine learning methods for materials chemistry.
Source: C&EN
Engineering Exotic Electronic States for Unique Chemical Reactivity
The transfer and transport of electrons govern chemical reactivity across nearly every application area, from catalysis to energy conversion. Exotic electronic configurations are inherently rare, but the materials that host them often display unprecedented chemical reactivity. We use quantum-mechanical calculations to understand how porous materials can be designed with electrons that exhibit unusual spatial and energetic configurations — materials where the electrons, rather than the atoms, are the chemically active species. In tandem, we are developing new classes of machine learning models that go beyond the atom to directly predict the complex behavior of electrons in solid-state materials.
Modeling Thermodynamic and Kinetic Stability Limits
Metastable materials are, by definition, not in their thermodynamic ground state. This raises a fundamental question: under what conditions can they exist, and when do they break down? We develop and apply machine learning-enhanced atomistic simulations to answer these questions, predicting decomposition mechanisms, mapping finite-temperature phase behavior, and establishing the stability limits of porous materials under realistic conditions. We are particularly interested in developing and applying machine-learning interatomic potentials to access the thermal properties of materials that are beyond the reach of traditional quantum-chemical methods.
Source: Matter
Source: PBS
Democratizing Computational Materials Science
High-quality data and accessible software are the backbone of modern computational materials science. We curate open datasets of quantum-mechanical properties and develop open-source scientific software to make rigorous, high-throughput materials simulations accessible to the broader community. We believe the tools and data behind our work are as important to share as the results themselves. For more about our software infrastructure, datasets, and machine learning models, visit our Software page.