UC Berkeley

The goal of this dissertation is to explore the capability of first-principles modeling of diffusion in complex materials. Atom diffusion in ferritic alloys and electronic diffusion in alpha-MoO3 are chosen as examples to demonstrate the computational study of diffusion properties in solid-state materials using first-principles modeling frameworks.

We first study the atomic diffusion problems for the development of high-temperature creep-resistant Fe-based multicomponent alloys. The temperature dependent self and solute diffusion coefficients in bcc Fe are calculated using density-functional theory, including the computation of diffusion prefactors and activation energies. For the self diffusivity, a spin-wave methodology is used for modeling the paramagnetic state to account for the effect of magnetic disorder on diffusion activation energy. Calculated self diffusion coefficients are shown to accurately reproduce the experimental measurements, including the anomaly in the Arrhenius plot near the Curie temperature. The solute impurity diffusion coefficients of the transition metal solutes (Ti-Zn, Nb-Cd and Ta-Au) are further calculated and shown to be comparable to available experimental measurements for most solutes. Our calculations show a general solute impurity diffusivity trend with minimum values for a given transition-metal row corresponding to solutes in the middle of series. Further we find a trend that diffusion of 5d solutes are slower than 3d which are slower than 4d. The results suggest that Co, Re, Os and Ir are the slowest diffusing solute species in bcc Fe, and these elements may be effective additions for slowing coarsening rates in precipitation-strengthened ferritic alloys. Additionally, some initial work for developing an automated computational tool for calculating equilibrium point defects in intermetallic compounds is established, to assist future first-principles calculations of diffusion coefficients in these ordered alloy phases.

First-principles modeling is further employed to study adiabatic diffusion of electron small polarons in alpha-MoO3, a material that has received significant attention for electrode applications in batteries and electrochemical supercapacitors. Density functional theory based calculations with van der Waals corrections (empirical dispersion corrections and van der Waals functionals) and self-interaction error corrections (Hubbard-U correction and hybrid functionals) are used to obtain accurate atomic and electronic structures of alpha-MoO3, respectively. After obtaining the atomic structure of an isolated electron small polaron structure, we present a computational scheme for calculating polaron adiabatic hopping barriers in the nearest-neighbor directions. Results suggest strong polaron diffusion anisotropies in crystalline alpha-MoO3. The effects of lithium-polaron binding and lattice relaxation are further studied in order to understand their effects on electron mobilities during Li intercalation.

Overall, the results presented in this dissertation demonstrate the predictive capabilities of first-principles modeling for studying diffusion problems in complex materials. The computational framework presented here can be extended to other advanced materials of interest.