From 2007-2017, I was a faculty member in Energy, Environmental & Chemical Engineering at Washington University in St. Louis, Missouri.
My research group emphasized the development and application of physics-based theories and models to chemical systems of importance in catalysis, electronic materials, and biological photosynthesis. We were particularly focused on technologies to upgrade carbon dioxide and methane to commodity chemicals, and design of new transparent conducting materials for thin film photovoltaics and displays.
We started by developing an understanding of “structure-property-activity” relationships, as elucidated through electronic structure calculations, to gain insight into why certain semiconductor materials are superior for given applications. One of our studies involved upgrading carbon dioxide to methanol, through initial molecular activation at point defect sites on a ceria catalyst substrate. We analyzed the change in the electronic density of states in ceria, and showed that charge transfer from the substrate to carbon dioxide is responsible for the formation of a carbonate-like ion that is prone to hydrogenation. This finding supported our notion that doped ceria, and related metal oxides that exhibit propensity towards oxygen vacancy formation, could be viable oxygen transport membranes in chemical looping combustion and other clean-burning technologies.
Concurrently, we also developed expertise in the modeling of defect physics in doped semiconductors, which enabled us to easily determine the thermodynamic conditions (e.g., temperature, pressure, chemical potential) under which they are stable relative to their parent structures. An example was the doping of zinc sulfide by aluminum, which was believed by experimentalists to be nearly impossible, yet shown to be theoretically viable if zinc sulfide adopted the hexagonal wurtzite phase. We thus used these insights to engineer defects to achieve desirable optoelectronic properties in various semiconductor materials, for eventual use in photovoltaic, display, or sensor technologies. Some of this work on artificial systems was motivated by our earlier studies of biological light harvesting in photosynthetic bacteria and dinoflagellate algae. We developed a computer model that integrates Configuration Interaction calculations with the Transition Density Cube method to calculate rates of Förster resonance energy transfer (FRET) from excited states with double excitation character; this allowed us to model the behavior of closely-packed pigment-protein systems, while treating the excited states of carotenoid pigments with accuracy that otherwise could not be captured by single-reference methods such as time-dependent DFT.
Working on these applied problems led us to the realization that numerous theoretical model limitations still exist, which would severely compromise the accuracy of electronic structure calculations and resulting property predictions. For instance, the widely-adopted models for electronic transport property calculations (e.g., mobility, conductivity, Seebeck coefficient), did not consider electron-phonon interactions, so they proved to be inaccurate when modeling trends with temperature or carrier concentration. We wanted an easy-to-use model that still provided high accuracy. We thus introduced “AMSET” (an ab initio model for mobility and Seebeck coefficient using the Boltzmann transport equation), which bridges the quantum-classical divide by: 1. Reformulating the model parameters in the semi-classical elastic scattering rate expressions to use band structure and density of states information calculated ab initio, without reliance on experimental data for model fitting, and 2. Explicitly considering the inelastic electron-phonon interactions. By implementing a full band model from the outset, we also eliminated problems that arose from only considering the band extrema (as is the case for effective mass-based models).
Another recent method development was a machine learning application to perform global optimization in surface science problems. This is particularly valuable from the standpoint of automating high-throughput calculations, where we want to rapidly and accurately search over stable molecule–surface configurations and interactions, without relying on extensive human intervention. By treating molecular adsorption as a multi-dimensional optimization over a potential energy hypersurface, we implemented a Bayesian optimization technique to explore and exploit regions of phase space, and further developed the approach that utilizes Bayesian inference to accelerate calculations on related chemical systems.
I am keeping a (mostly) up-to-date list of my peer-reviewed publications on ORCID.
My (now former) graduate and undergraduate students pursued a variety of career paths.
My Ph.D. and Postdoctoral training were instrumental in developing my interest and expertise in energy and environmental technologies. For my Ph.D. dissertation, I used Density Functional Theory calculations and ab initio molecular dynamics simulations to investigate the catalytic conversion of methanol to gasoline and olefins. I showed computationally how porous solid-acid catalysts, such as the zeolites commonly used in industry, promote this reaction. One of my key findings was that zeolite catalysts, with controlled composition and morphology, confine the reacting molecules in a small physical space and dynamically donate protons to activate the methanol molecules. By being one of the first researchers to integrate transition path methods with ab initio approaches, I demonstrated the importance of including entropic effects in the modeling of complex processes in non-biological systems.
For my Postdoctoral training, I investigated environmental mineral-water interactions and developed a protocol for accurately calculating the structure of hydrated and hydroxylated iron oxide surfaces under environmentally-relevant conditions. The motivation for this highly-interdisciplinary work was to use surface science and environmental catalysis techniques to better understand how natural waters, natural organic matter, and biological organisms interact with natural solids and environmental contaminants. One of my key findings was that the experimentally-determined mineral-water interface does not merely consist of adsorbed and heterolytically dissociated waters at undercoordinated surface sites, but also appears to be “missing” metal cations that have been reduced and consequently partition from the solid to the (free) aqueous phase during these environmental reactions. I showed that quantum mechanics could accurately predict the structure and properties of complex environmental interfaces, and established protocols for calculating the stability of such interfaces using ab initio thermodynamics.