No abstract

No abstract

The bulk conductivity of polycrystalline 1 mol-% Fe-doped rutile TiO2 has been measured as a function of p H2O and temperature under oxidizing and reducing conditions. From the p H2O-dependency of the conductivity, it is concluded that protons are significant positive defects and, furthermore, that mixed p-type electronic and protonic, and n-type electronic conduction dominate under oxidizing and reducing conditions, respectively. H2O/D2O isotope exchange confirmed that protons are significant charge carriers under wet oxidizing conditions below approximately 600 °C. Thermodynamic parameters for the hydration reaction were obtained by modeling the experimental p H2O and temperature dependencies, assuming that the acceptor dopant (Fe3+) is charge compensated by protons and oxygen vacancies, and from ab initio density functional theory (DFT) calculations. The experimental data yield standard enthalpy changes of hydration of −130 ± 16 kJ/mol, whereas the calculated values are somewhat more negative; −155 to −162 kJ/mol. Based on such favorable thermodynamics of hydration, it is concluded that protons will be the dominating positive defect in TiO2 under most conditions of practical interest.

Li and H are important electrically active impurities in ZnO and this work presents a detailed experimental and computational study of the behavior of H and Li in ZnO and their effect on its defect structure. We employ AC conductivity measurements as a function of temperature and partial pressure of O2, H2O, and D2O, which is combined with first principles density functional theory (DFT) calculations and thermodynamic modeling (TDM) of finite temperature defect structures in undoped and Li doped ZnO. Undoped ZnO is dominated by protons as hydroxide defects (OHO •), oxygen vacancies (vO ••), and electrons under a large variety of atmospheric conditions, and we also predict from DFT and TDM the substitutional hydride ion (HO •) to dominate concentration-wise under the most reducing conditions at temperatures above 500 °C. The equilibrium concentrations of defects in ZnO are small, and dopants such as Li strongly affect the electrical properties. Experimentally, Li doped ZnO is found to be n-type under all available atmospheric conditions and temperatures, with an n-type conductivity significantly lower than that of as-grown ZnO. The n-type conductivity also increases with decreasing p O2 and with increasing p H2O. The observed electrical properties of Li doped ZnO are attributed to dominance of the ionic defects LiZn /, OHO •, Lii •, vO ••, and the neutral complexes (LiZnOHO)× and (LiZnLii)×. Although Li doping lowers the Fermi level of as-grown ZnO significantly, low formation energy of the ionic donors, and passivation of LiZn / in the form of (LiZnOHO)× and (LiZnLii)×, prevents realization of significant/stable p-type activity in Li doped ZnO under equilibrium conditions.

The electrical conductivity and thermoelectric power of ZnO have been examined in hydrogen-containing atmospheres up to 550 °C. The type and concentration of charge carriers (here electrons) and their charge compensating defects (here protons) were determined from the thermoelectric power, and electron charge mobility was evaluated by combination with the measured conductivity. Above approximately 450 °C, the defects are in equilibrium with the surroundings, and the concentration of protons and electrons increases with temperature and is proportional to p H 2 1/4 , in accordance with the defect thermodynamics from early literature. Below approximately 450 °C, the concentration of hydrogen appears to become frozen-in. This results in a high internal hydrogen pressure during further cooling, which, for instance, may crack single crystals. The local strain from the presence of frozen-in neutral H 2 species is suggested to cause an observed modest reduction in the mobility and conductivity of electrons below the freezing-in temperatures. The levels of defect concentrations and electron mobility are 1 order of magnitude off compared with established literature when based on our thermoelectric power applying standard theory. This discrepancy is in the order of either a reduction in the assumed effective mass of electrons from the commonly used 0.23m 0 to 0.075m 0 or the removal of the 5/2k B term in the expression used for the entropy of a free electron gas.