Multiconfiguration Pair-Density Functional Theory with Quantum Embedding Predicts Correct CO Adsorption Sites on Copper Facets

The adsorbed states of CO on copper are ubiquitous in copper-mediated heterogeneous catalysis and CO2 reduction, as they represent the initial structures or critical intermediates in reaction mechanisms. However, accurately determining CO adsorption energies and identifying the lowest-energy binding sites on various copper facets present unexpected challenges for density-functional theory with local exchange-correlation functionals. Previous work has shown that all widely used semilocal Kohn-Sham density functionals, including the Perdew-Burke-Ernzerhof (PBE) functional and the M06-L functional, fail to predict the correct, most favorable binding sites of CO on copper surfaces. These functionals consistently favor hollow sites rather than the experimentally observed on-top sites. In this work, we demonstrate that quantum embedded multiconfiguration pair-density functional theory (emb-MC-PDFT), combined with the PBE functional, quantitatively and correctly predicts both the most favorable binding sites and the corresponding binding energies from first-principles for CO adsorption on copper (111), (110), and (100) facets.