A computational high throughput search of symmetric tilt grain boundaries in cerium oxide

Cerium dioxide is an important solid electrolyte in energy applications. Interfaces affect its properties, e.g. grain boundary blocking effect, and to begin engineering such properties, we need to control the stability of interfaces. Phenomenological models for interfaces may predict energy as a function of structural parameters, but we need more substantial quantitative results. Here, we apply high-throughput computing to provide a systematic atom level representation of the structures and energetics of grain boundaries for CeO₂. This search is based on 160 symmetrically independent Coincidence Site Lattice (CSL) mirror-tilt grain boundaries arising from surfaces with Miller indices {hkl} where h, k, and l = 0–9. For each boundary, we perform a “scan” of all possible structures by searching the configurational space of the two adjoining surfaces, which provides a measure of the “configurational availability” of structures accessible via doping or thermal activation. We demonstrate that for known interfaces, structures have been experimentally observed. We elucidate the relationships amongst CSL parameters and formation and cleavage energies. There is a general rule that low formation energies are correlated to low Miller indices, and although largely true, we found also low formation energies for high Miller index {hkl} boundaries, even for the comparatively simple fluorite-structured CeO₂. Within different classes of grain boundary, formation energies appear to follow the Bulatov-Reed-Kumar model, while cleavage energies do not. All grain boundary structures are presented to facilitate and assist experimental characterization. This computational approach is general and could be applied to any material and any grain boundary class.