Conventional TiO2 based photocatalysts oxidize NOx to nitrates which do not automatically desorb and have to be washed away from the catalyst surface. To avoid this, the research described in this thesis aims to design new photocatalysts that can photo-reduce NO into N2 and O2. Previous efforts in the literature have shown that catalysts prepared by depositing isolated titanium oxide clusters (TiO4, tetrahedra) inside the cavities of zeolite-Y using ion beam implantation technology show high activity for N2 and O2 formation. However, large scale application of zeolites using ion beam technology is economically unattractive. We aim to create a large number of oxygen vacancies in TiO2 to ‘artificially’ create TiO4 species. The basic principles of photocatalysis and the overall approach taken in this thesis have been described in Chapter 1. The following chapters describe how a series of Fe-doped TiO2 nanoparticles were prepared and systematically investigated for creating and stabilizing oxygen vacancies for changing the photocatalytic selectivity from the oxidation to the reduction of NO. The synthesis and characterization of Fe-doped TiO2 nanoparticles with different Fe concentrations have been described in Chapter 2. A simple, template free sol-gel method was used for synthesizing pure TiO2 and Fe-doped TiO2 nanoparticles. After drying the as-synthesized samples at 373 K , most iron is located at the surface. However, after firing at 773 K, the iron diffuses into the bulk of the TiO2 up to a 10% Fe doping concentration. No evidence for phase segregation was found. For the XRD patterns, the anatase (101) peak position shifts towards smaller d values with increasing Fe concentration. The observed shift is linear with the amount of Fe. This is in accordance with Vegard’s law and this indicates the incorporation of Fe into TiO2 lattice. However, the decrease in lattice constant is contrary to what one might initially expect, since the radius of Fe3+ ions (0.65 ?) is slightly larger than Ti4+ radius (0.61 ?) for six-fold coordinated ions. The formation of Fe4+, can be ruled out from Fe K-edge XANES spectra. Instead, we attribute the decrease in lattice spacing to a high concentration of oxygen vacancies. These oxygen vacancies are formed in order to compensate the effective negative charge of the Fe acceptor dopant. The presence of oxygen vacancies is further confirmed by Raman spectra. A similar linear shift at the strongest anatase (Eg mode) is observed by comparing Fe-doped TiO2 with reduced TiO2-x from Parker et al.. This confirms that oxygen vacancies are indeed present in our material. The surprisingly high solubility of Fe is mainly due to the fact that the ionic radii of Ti4+ and Fe3+ are nearly equivalent. In addition, there is an energetically favorable Coulomb attraction between the negatively charged Fe acceptor and the positively charged oxygen vacancies. Intriguingly, XANES measurements indicate that the coordination geometry of Ti is changed from octahedral to tetrahedral at high oxygen vacancy concentrations. The tetrahedrally coordinated TiO4 clusters are presumably present at the surface, where the lattice symmetry constraints are more relaxed. The phenomenon that the coordination geometry can be changed by creating large number of oxygen vacancies provides a new way for designing highly selective photocatalyst. The first step of photocatalysis, NO adsorption and release at Fe3+ sites in Fe/TiO2 nanoparticles, is studied by in-situ Diffuse Reflection Infrared Fourier Transformed spectroscopy coupled with Mass Spectrometry (DRIFT, Chapter 3). In this chapter, Fe3+ ions are found to be highly effective NO adsorption sites after in-situ heat treatment. Upon exposure to trace amounts of H2O, up to ~ 89 ?mol/g NO can be released from the surface of 10% Fe-doped TiO2. This can be explained by the much larger dipole moment of H2O as compared to that of NO. In addition, the phenomenon that adsorbed H2O can effectively block NO adsorption explains why Fe3+-NO species are often not observed. In addition, a new IR band at 1840 cm-1 is assigned to the stretch vibration of N-O bond over Fe3+ site. This is the first clear evidence for the presence of Fe3+-NO on Fe-doped TiO2 nanoparticles. While easy replacement of NO by H2O molecules is clearly undesirable for practical use of photocatalysts, the controllable adsorption and release of NO may lead to new NO storage and release applications based on Fe-doped TiO2. In order to identify the contribution of oxygen vacancies to catalytic selectivity, photocatalytic measurements of pure TiO2 after annealing in air (unreduced TiO2) and in a 2% H2/Ar atmosphere (reduced TiO2) are compared. The photocatalytic activities are now evaluated by a combination of an NOx analyzer and gas chromatography (GC). With UV light irradiation, 1% NO photo-reduction was detected after reaching catalytic equilibrium, while no photo-reduction can be found for unreduced TiO2. This indicates that oxygen vacancies contribute to photo-reduction selectivity. To further prove the role of oxygen vacancies, Fe doped TiO2 with different oxygen vacancy concentrations are further evaluated for NO decomposition. For 1% Fe-doped TiO2, 3% NO photo-reduction is found in air after achieving the photocatalytic equilibrium. The reduction activity is a three times improvement over reduced TiO2 at the same experimental conditions. Furthermore, a 4.5% NO photo-reduction is found in pure N2 atmosphere. This is a higher efficiency than that in air. This strongly supports our interpretation of the photo-reduction process in Fe-doped TiO2. In this case, the formation of gas phase NO2 is now completely suppressed. This is consistent with our observation from in-situ DRIFT spectra. In addition, we find that the photoreduction activity increases with an increasing concentration of oxygen vacancies. The direct evidence for NO photo-reduction is provided by GC measurements. For 1% Fe-doped TiO2, almost the same amount of N2 and O2 are produced after reaching catalytic equilibrium. This excludes the possibility of N2O formation. Pure TiO2 shows no formation of N2 and O2 at the same experimental conditions. Two possible explanations are provided for the formation of N2 and O2 in chapter 4. One of the possible explanations is that a small amount of tetrahedrally coordinated Ti-oxides are formed. However, the formation of tetrahedrally coordinated Ti would require two oxygen vacancies to be present on the same Ti sites. As positively charged oxygen vacancies would repel with each other and very low oxygen vacancy concentrations are present up on 1% Fe doping, the formation of tetrahedrally coordinated Ti is not very likely. A more likely explanation is that oxygen vacancies may act as catalytic centers for trapping O atom of NO molecules, thus lowering the bond energy of NO. The trapped nitric molecules have no net charges. This allows mobile oxygen vacancies to closely approach these species, after which a second NO species can be adsorbed directly next to the first one. When both nitrogen end-species combine to form N2, the energy released may easily knock out the trapped oxygen species, resulting in the release of O2. Although the mechanism of photocatalytic reduction is still not yet exactly clear, the contribution of oxygen vacancies to NO photo-reduction selectivity provides a promising way for designing new and highly selective photocatalysts. To confirm the reaction mechanism from Chapter 4, the photon-assisted NO adsorption-desorption and reaction species on the surface of the samples are investigated in Chapter 5. The combination of DRIFT spectra and on-line NOx analysis confirms that Fe dopant can actually suppress the formation of NO2 during photocatalytic process. The suppression of NO2 formation is caused due to the reduction of Fe3+ to Fe2+ by photo-generated electrons. In addition, the photon-assisted NO adsorption shows a new IR band at 1805 cm-1, which is attributed to N=O stretch vibration in a Fe2+-(NO)2 complex. After turning off the UV light, on-line NOx analysis shows that the NO desorbs again. XPS data indicates that this coincides with the re-oxidation of Fe2+ to Fe3+ presumably by photo-generated hydroxyl radicals. The re-oxidation triggers the release of NO due to its weak bonding to Fe3+, which is consistent with the results presented in chapters 3 and 4.