AbstractHerein, tools are presented for quantifying the search space for inorganic materials. By restricting the search to stoichiometric compounds, and by limiting the stoichiometry of each element to a maximum value, the space becomes finite for binary, ternary and quaternary element combinations. Hierarchical workflows are then used to target specific materials properties such as thermodynamic stability and bandgap. The workflows consist of modular screening steps that are also developed within this thesis, and are based on a mixture of heuristic chemical rules and data-driven approaches. The classic chemical heuristics used include electronegativity and oxidation state, along with more recently developed metrics such as the solid state energy scale. The data-driven screening steps include a machine learning model that predicts bandgap from chemical composition, a probabilistic model to predict likely oxidation state combinations, and a previously reported ionic substitution model that assigns crystal structures to compositions. Finally, these workflows are applied to the search spaces of metal chalcohalides and quaternary metal oxides, with top candidates identified and further characterised using high-throughput first-principles calculations.
|Date of Award||13 Feb 2019|
|Supervisor||Keith Butler (Supervisor), Aron Walsh (Supervisor) & Benjamin Morgan (Supervisor)|
Materials discovery using chemical heuristics and high-throughput calculations
Davies, D. (Author). 13 Feb 2019
Student thesis: Doctoral Thesis › PhD