HeliECat · A Helicopter View on Electrocatalysis

Converting carbon dioxide, an unwanted by-product of industrial processes and combustion engines, into valuable chemical compounds is key for a cyclic carbon economy and can be achieved by the electrochemical CO2 reduction reaction (CO2RR). Despite the immense importance of this reaction in the fight against climate change, no catalyst to date meets the industrial requirements for product selectivity and catalyst stability. Theoretical models and simulations have played a fundamental role in the understanding of these and similar catalytic processes and aided the development of more efficient catalysts for this reaction. These models, however, assume few often idealized surface structures, ignoring the fact that a variety of structures is present under working conditions that define the catalyst properties in terms of efficiency, selectivity, and stability. With HeliECat, I will for the first time establish an approach that offers a holistic view of the catalytic processes, including side reactions and degradation mechanisms. To achieve this ambitious goal, I will combine an efficient semiempirical model for structural exploration with accurate embedded cluster models for kinetic studies. This combination shifts the focus from idealized structures to the diversity of the catalyst surface under realistic conditions. I will then model the interconversion of accessible surface structures with deep reinforcement learning techniques. The analysis of the emerging connected kinetic reaction networks will ultimately allow me to simultaneously study the catalytic activity, stability, and selectivity in silico. I will study the CO2RR on a Cu/Au catalyst surface and demonstrate that automated kinetic modeling, considering structural diversity and realistic thermodynamic boundary conditions, is the key to a predictive simulation of catalyst behavior under actual working conditions. HeliECat will hence pave the way for realistic in silico catalyst screening.