Dynamics in Ceria and Related Materials from Molecular Dynamics and Lattice Dynamics

Sammanfattning: In discussions of heterogeneous catalysis and other surface-related phenomena, the dynamical properties of the catalytic material are often neglected, even at elevated temperatures. An example is the three-way catalyst (TWC), used for treatment of exhaust gases from combustion engines operating at several hundred degrees Celsius. In the TWC, reduced ceria (CeO2-x) is one of the key components, where it functions as an oxygen buffer, storing and releasing oxygen to provide optimal conditions for the catalytic conversion of the pollutants. In this process it is evident that dynamics plays a crucial role, not only ionic vibrations, but also oxygen diffusion.In this thesis, the structure and dynamics of several ionic crystalline compounds and their surfaces have been studied by means of Molecular dynamics (MD) simulations and Lattice dynamics (LD) calculations. The main focus lies on CeO2-x, but also CeO2, MgO and CaF2 have been investigated.The presence of oxygen vacancies in ceria is found to lead to significant distortions of the oxygen framework around the defect (but not of the cerium framework). As a consequence, a new O-O distance emerges, as well as a significantly broadened Ce-O distance distribution.The presence of oxygen vacancies in ceria also leads to increased dynamics. The oxygen self-diffusion in reduced ceria was calculated from MD simulations in the temperature range 800-2000 K, and was found to follow an Arrhenius behaviour with a vacancy mechanism along the crystallographic <100> directions only.The cation and anion vibrational surface dynamics were investigated for MgO (001) using DFT-LD and for CaF2 (111) in a combined LEED and MD study. Specific surface modes were found for MgO and increased surface dynamics was found both experimentally and theoretically for CaF2, which is isostructural with CeO2.Many methodological aspects of modeling dynamics in ionic solids are also covered in this thesis. In many cases, the representation of the model system (slab thickness, simulation box-size and the choice of ensemble) was found to have a significant influence on the results.