Efficient Electronic-Structure Methods Toward Catalyst Screening: Projection-Based Embedding Theory for CO₂ Reduction Reaction Intermediates

Catalyst screening is a demanding task for computational chemistry since the profound diversity of surface structures under operando conditions is accompanied by high demands on the accuracy to predict the relevant kinetics. Embedding approaches that allow researchers to focus the computational effort on the chemically active regions of interest are promising tools in the pursuit of balancing accuracy and efficiency. However, for metallic catalysts, the required separation of the system into an active part treated with highly accurate methods and an environment is technically hard to achieve due to the delocalization of electrons in the conducting surface. Therefore, studies analyzing the potential of embedding methods for heterogeneous (electro-)catalyst screening are scarce. In this contribution, we demonstrate that simple embedding approaches are indeed achievable for studying metallic catalysts if i) the active orbital space is held consistent over a reaction coordinate and ii) the nonadditive exchange-correlation functional used to calculate the embedding potential includes a fraction of exact exchange to mitigate delocalization errors. We verify the approach for a set of open- and closed-shell CO₂ reduction reaction intermediates on different adsorption sites of a Cu(111) surface represented by cluster models to demonstrate that catalyst screening with embedding approaches is achievable.