# PhD Theses - Haiming Lu

## Atomistic Simulation of Defects and Thermal Conductivity in Oxides

### Abstract

Atomic scale computer simulation techniques are used to predict the thermal conductivity and defect properties in Li2O and Bi4Ti3O12 ceramics. Li2O represents a simple model system while Bi4Ti3O12 is a complex layered oxide ceramic with potential engineering applications due to its highly anisotropic thermal conductivity. The efficacy of the available pair potential models for Li2O are evaluated by comparing with experimental properties such as lattice parameters, elastic constants, thermal expansion and bulk modulus. Other properties, such as the Li ion superionic transition temperature, activation energy and diffusion coefficients for lithium diffusion are also compared with experimental data. The empirical potential set of Chroneos et al. [1] was found to best replicate the experimental diffusion data; therefore this potential was used to calculate the reaction energies of the intrinsic disorder properties and then used in the investigation of the thermal properties of Li2O. The thermal conductivity of Li2O was calculated using the velocity exchange methodology within molecular dynamic simulations. The calculated thermal conductivity was compared to experimental data. A number of corrections have been suggested to improve the agreement between the simulated thermal conductivity and the equivalent experimental value. These corrections and the underlying physics are also discussed within the context of Li2O. The extent of the isotope effect on thermal conductivity was investigated by comparing a supercell, in which there is a random distribution of 6Li and 7Li atoms, to a supercell where all the atoms are given a fractional average mass. The results show that the effect of a non-homogeneous distribution of Li mass can lead to a significant decrease in the thermal conductivity at low temperatures but this effect gradually decreased up to the superionic transition temperatures, where it disappears altogether. Two different structures have been proposed for low temperature Bi4Ti3O12, one with a monoclinic space group (B1a1 ) and the other with a very similar structure but an more symmetric orthorhombic space group (B2cb). A combination of empirical pair potential and first principles density functional simulations are employed to compare the two structures. Both techniques suggest the monoclinic structure has the lower energy of the two candidate structures. The enthalpy of formation of Bi4Ti3O12 is calculated using first principles simulation. The macroscopic properties of a material are determined by the concentration and behaviour of point defects such as vacancies and interstitials. Therefore, density functional theory (DFT) simulations were performed to investigate the formation energies and defect volume of vacancy defect in Bi4Ti3O12. The pair potential set of Islam et al. [2] and Snedden et al. [3] which replicate the structural properties of Bi4Ti3O12 were used to predict the anisotropic thermal conductivities of Bi4Ti3O12 in x, y and z direction. The sound velocity and mean free paths were calculated and used to explain the temperature independent thermal conductivity in the z direction and the anisotropic thermal conductivity. Finally, the Debye temperature as a function of temperature was computed.