Capturing experimental properties in computationally efficient faceted titania nanoparticle models

Abstract

Understanding the surface chemistry of nanostructured TiO2 has long been a priority to improve photochemical device efficiency. Faceted nanoparticles, characterized by known facets not at thermodynamically ideal ratios, are particularly challenging to model due to the large number of chemical and computational parameters that must be chosen for which there is no experimental guidance. This research supplies a modeling framework for faceted TiO2 nanoparticles that provides rationale for such decisions. By performing full density‐functional theory optimization and characterization on a series of inter‐related anatase TiO2 nanoparticles displaying {101}, (001), and {010} facets with sizes up to 202 TiO2 units, parameter space is mapped with regard to particle size, shape, defects, and optimization protocol. Specifically, it is shown that pre‐optimization is necessary in order to achieve a sufficiently delocalized electronic structure, and the increased reorganization afforded by removing higher coordinated Ti atoms compensates the high formation energy of creating these defects. Furthermore, by characterizing differently shaped nanoparticles with the same number of TiO2 units, this research provides direct observation of shape effects on nanoparticles.

Lessons Learned from Catalysis to Qubits: General Strategies to Build Accessible and Accurate First-Principles Models of Point Defects

“…These “rattled” structures allowed the optimizer to escape local minima and relax to a lower energy ground state structure. Such symmetry breaking has been shown to improve the optimization behavior in TiO 2 nanoparticles previously . As a general rule of thumb, it was suggested that structural perturbations of around −40% to +20% to provide optimal results …”

“…Such symmetry breaking has been shown to improve the optimization behavior in TiO 2 nanoparticles previously. 32 As a general rule of thumb, it was suggested that structural perturbations of around −40% to +20% to provide optimal results. 29 In addition to random atomic perturbations, systematic modification of the starting geometry also has the potential to alter the optimization pathway.…”