Abstract
This thesis describes the application of advanced computational and experimental techniques to the detailed analysis of the defect chemistry and ion transport properties of mixed metal oxides. Attention is focused on two complex oxide materials containing tetrahedral and octahedral units: Sc2(WO4)3, which is a proposed trivalent cation conducting material; and LiFePO4, which is a potential lithium battery cathode material. Computer simulations of Sc2(WO4)3 first reproduce the complex crystal structure, with the calculated unit cell parameters within 0. 8% of experimental diffraction data. Frenkel and Schottky defect energies have been calculated suggesting that such intrinsic defects are not significant within the structure. Vacancy migration (O2- or Sc3+) has been calculated to be unfavourable. Modelling the pathways of interstitial O2- and Sc3+migration suggests that either mechanism is possible, although the process to create these defects is still not clear. Isovalent doping onto the Sc3+ site is shown to be energetically favourable for a range of ions (e.g. Ga3+, In3+, Yb3+). In terms of experimental doping studies, a range of strategies have been tried to introduce either vacancies or interstitial ions. These attempts were unsuccessful showing that aliovalent doping on either cation site is extremely difficult and so deviations from the ideal stoichiometry appear unfavourable. Isovalent doping was favourable for a range of ions (e. g. Ga3+, Al3+, Fe3+, In3+) which support the modelling results. Impedance data suggests that the main conduction mechanism is ionic rather than electronic in agreement with previous studies. The simulation model of LiFePO4 shows good reproduction of the observed olivine-type crystal structure. The most favourable intrinsic defect is the Li-Fe "anti-site" pair in which an Li ion and an Fe ion are interchanged. This type of anti-site defect or "intersite exchange" has been observed in olivine silicates. The lowest Li migration energy is found for the pathway along the [010] channel, with a non-linear, curved trajectory between adjacent Li sites. Trends in dopant substitution energetics of a range of cations with charges varying from +2 to +5 are also examined. Low favourable energies are found only for divalent dopants on the Fe site (such as Mn), which is in accord with experimental work. Our results suggest that, on energetic grounds, LiFeP04 is not tolerant to aliovalent doping (e.g., Al3+, Ga3+, Zr4+, Ti4+, Nb5+, Ta5+) on either Li or Fe sites.