With the vision of a realistic simulation of material properties and processes and the expectation of a targeted design of catalytic materials Maria Veronica Ganduglia-Pirovano performs computational modelling and simulations employing state-of-the-art quantum mechanical calculations. Her interests are to model the complex structure of functional materials and to understand –at the molecular level– the nature of the (active) surface of a solid catalyst as well as the elementary steps of chemical reactions that take place on it.
Abstract:
Cerium-Oxide-Bound CO Vibrations: Beyond DFT's Comfort Zone
In cerium oxide-based catalysis, surface structure plays a crucial role in the catalyst's reactivity. The vibrational frequency of surface-adsorbed carbon monoxide (CO) surface serves as a precise marker for identifying active defect sites and determining the exposed crystal facets. To analyze spectroscopic data accurately, reference data for well-characterized single crystal surfaces and a precise theoretical assignment of the vibrations are essential. Significant efforts have been made to gather comprehensive IRRAS (infrared reflection absorption spectroscopy) data for CO adsorption on all three low-index single-crystal ceria surfaces, both in their oxidized and reduced forms, at saturation coverage. However, applying theoretical methods, many of which rely on density functional theory (DFT) with the generalized gradient approximation (GGA) for exchange and correlation, has encountered significant challenges in accurately describing CO adsorption on oxides. Recently, there has been a growing interest in combining machine learning (ML) techniques with DFT calculations for interpreting spectroscopic data. In this study, we demonstrate that by employing DFT with the HSE06 hybrid functional yields excellent agreement with experimental data. Achieving this high level of consistency, requires meticulously adjustment of the model to closely align with experimental conditions. These conditions encompass factors such as surface orientation, the presence of oxygen vacancies, and the extent of CO coverage. Our study reveals that CO is highly sensitive to the precise structure of ceria surfaces. We also explain the limitations of conventional DFT by highlighting its inability to accurately capture facet- and configuration specific donation and backdonation effects, which control the changes in the C─O bond length upon CO adsorption and the CO force constant. Our findings establish a robust theoretical foundation for the accurate interpretation of experimental results. Furthermore, we emphasize the importance of understanding the origin of DFT data before blindly incorporating them into ML models.
