Many promising catalytic technologies, such as processes for efficiently converting carbon dioxide to fuels and the activation of nitrogen to produce fertilizer, are based on chemical reactions at interfaces. The rates of chemical reactions depend on their reaction energies and activation energies, which are affected by the interfaces at which they occur. Gaining a better understanding of these effects is critical for engineering these interfaces for optimal performance. From a computational standpoint this requires capturing detailed electronic structure effects from the metal and also the many intermolecular interactions and long-range electrostatic effects from the electrolyte, all of which must in principle be sampled over a thermodynamic ensemble of configurations. Furthermore, as chemical reactions and experimental systems become increasingly complex, more generalizable and automated approaches to rigorously calculate the free energy surfaces of reactions are needed.
We address several of these challenges using density functional theory models of metal/vacuum and metal/electrolyte interfaces. We use global optimization algorithms to provide new insights into the effect of the interface on reaction energies and other physical quantities which are critical in modeling catalytic processes. We demonstrate the application of state-of-the-art enhanced sampling methods such as adaptive biasing force and a recently developed machine learning-based approach, and make comparisons of these more general approaches to conventional, harmonic approximation-based approaches. In addition to specific physical insights, this work demonstrates the value of systematic, generalizable algorithms for studying increasingly complex models of important chemical reactions.
