Understanding And Predicting The Chemical Properties Of Complex Oxides Using First-Principles Methods

Transition metal oxides are at the forefront of several applications in catalysis, energyconversion and storage, and are bound to play a pivotal role in our transition to a sustainable energy future. However, vast differences in the electronic structure among different transition metal oxides make them highly complex materials to understand and predict chemical properties for a given application. This thesis provides insights into improving our understanding of these complex oxides through a first-principles approach. Further, the goal is to leverage this understanding to improve the predictability of chemical properties of these materials as it pertains to heterogeneous catalysis. Through the extensive use of density-functional theory calculations, a detailed analysis of the electronic structure, and ab-initio molecular dynamics coupled with enhanced sampling methods, this thesis aims at better understanding perovskite and rutile oxides for their applications as electrocatalysts in water-splitting and as catalyst supports in thermal catalytic applications. Electrocatalysts for water-splitting pose the twin issues of activity and stability in that the most active catalysts are unstable and the most stable are inert. To resolve this, hetero-structured oxide systems which involve the doping of a semiconducting inert host oxide with 3d, 4d and 5d metals were studied as candidates for electrochemical water-splitting using a computational high-throughput screening approach. It was found that certain dopants such as Ru, and Ir improved the activity of an otherwise inert material making them viable options as water-splitting catalysts. The introduction of dopants to a semiconducting oxide also served as a unique perturbation of their electronic structure, resulting in the development of a physics based adsorption model and the identification of an electronic structure descriptor that captures the adsorption trends across these doped semiconducting oxides accurately. Understanding the changes to a catalyst surface is also vitally important to better design these materials. Dissolution of state-of-the-art perovskite and rutile oxides used in water splitting is a persistent problem that shortens their lifetime. Using a combination of ab-initio thermodynamics, molecular dynamics and enhanced sampling methods this thesis provides an atomistic understanding of the dissolution process, reaching length and time-scales otherwise inaccessible to experimental techniques. The dissolution path for the (110) surface of three different rutile oxides RuO2, IrO2 and TiO2 are identified using this approach. Further, the differences in surface stability of the perovskite oxides SrRuO3, SrIrO3 and SrTiO3 are determined and discussed. Similar to the issue of dissolution seen with electrocatalysts, supported metal catalysts which form the backbone of thermal catalytic applications, as well as in energy conversion such as solid-oxide fuel cells, pose the issue of sintering and coking, once again reducing their operational lifetimes. Novel synthesis strategies such as the exsolution of transition metals doped into perovskite oxide supports have been suggested as ways to mitigate these problems. Although several successful metal-host combinations have been identified successfully, there remains a lack of mechanistic understanding of the process. In this thesis, following a similar approach used to model dissolution, the exsolution of Pt from different perovskite titanate hosts was modelled to gain an atomistic understanding. It was found that the diffusion of Pt to the surface of the perovskite was dependent on the exposed facet of the host, which may now be used towards the predictive synthesis of supported metal catalysts. Taken together, through a combination of electronic structure insight and the dynamics of the process under operating conditions, this thesis provides a comprehensive understanding of complex oxides, as well as a path towards their bottom-up design