In ceria-based catalysis, the shape of the catalyst particle, which determines the exposed crystal facets, profoundly affects its reactivity. The vibrational frequency of adsorbed carbon monoxide (CO) can be used as a sensitive probe to identify the exposed surface facets, provided reference data on well-defined single crystal surfaces together with a definitive theoretical assignment exist. We investigate the adsorption of CO on the ${\mathrm{CeO}}_2(110)$ and (111) surfaces and show that the commonly applied $\mathrm{DFT}(\mathrm{PBE})+U$ method does not provide reliable CO vibrational frequencies by comparing with state-of-the-art infrared spectroscopy experiments for monocrystalline ${\mathrm{CeO}}_2$ surfaces. Good agreement requires the hybrid DFT approach with the HSE06 functional. The failure of conventional density-functional theory (DFT) is explained in terms of its inability to accurately describe the facet- and configuration-specific donation and backdonation effects that control the changes in the $\mathrm{C}─\mathrm{O}$ bond length upon CO adsorption and the CO force constant. Our findings thus provide a theoretical basis for the detailed interpretation of experiments and open up the path to characterize more complex scenarios, including oxygen vacancies and metal adatoms.

• Received 27 May 2020
• Revised 4 September 2020
• Accepted 9 November 2020

DOI:https://doi.org/10.1103/PhysRevLett.125.256101