Spectroscopic properties of doped and defective semiconducting oxides from hybrid density functional calculations.

CONSPECTUS

Very rarely do researchers use metal oxides in their pure and fully stoichiometric form. In most of the countless applications of these compounds, ranging from catalysis to electronic devices, metal oxides are either doped or defective because the most interesting chemical, electronic, optical, and magnetic properties arise when foreign components or defects are introduced in the lattice. Similarly, many metal oxides are diamagnetic materials and do not show a response to specific spectroscopies such as electron paramagnetic resonance (EPR) spectroscopy. However, doped or defective oxides may exhibit an interesting and informative paramagnetic behavior. Doped and defective metal oxides offer an expanding range of applications in contemporary condensed matter science; therefore researchers have devoted enormous effort to the understanding their physical and chemical properties. The interplay between experiment and computation is particularly useful in this field, and contemporary simulation techniques have achieved high accuracies with these materials. In this Account, we show how the direct comparison between spectroscopic experimental and computational data for some selected and relevant materials provides ways to understand and control these complex systems. We focus on the EPR properties and electronic transitions that arise from the presence of dopants and defects in bulk metal oxide materials. We analyze and compare the effect of nitrogen doping in TiO2 and ZnO (two semiconducting oxides) and MgO (a wide gap insulator) and examine the effect of oxygen deficiency in the semiconducting properties of TiO2-x, ZnO1-x, and WO3-x materials. We chose these systems because of their relevance in applications including photocatalysis, touch screens, electrodes in magnetic random access memories, and smart glasses. Density functional theory (DFT) provides the general computational framework used to illustrate the electronic structure of these systems. However, for a more accurate description of the oxide band gap and of the electron localization of the impurity states associated with dopants or defects, we resorted to the use of hybrid functionals (B3LYP), where a portion of exact exchange in the exchange-correlation functional partly corrects for the self-interaction error inherent in DFT. In many cases, the self-interaction correction is very important, and these results can lead to a completely different physical picture than that obtained using local or semilocal functionals. We analyzed the electronic transitions in terms of their transition energy levels, which provided a more accurate comparison with experimental spectroscopic data than Kohn-Sham eigenvalues. The effects of N-doping were similar among the three oxides that we considered. The nature of the impurity state is always localized at the dopant site, which may limit their application in photocatalytic processes. Photocatalytic systems require highly delocalized photoexcited carriers within the material to effectively trigger redox processes at the surface. The nature of the electronic states associated with the oxygen deficiency differed widely in the three investigated oxides. In ZnO1-x and WO3-x the electronic states resemble the typical F-centers in insulating oxides or halides, with the excess electron density localized at the vacancy site. However, TiO2 acts as a reducible oxide, and the removal of neutral oxygen atoms reduced Ti(4+) to Ti(3+).