Despite the intense research efforts to convert the energy of light into renewable fuels, photocatalytic processes still have not reached the requirements for a successful application on an industrial scale. This may be attributed to the lack of profound understanding of the involved mechanisms. The scope of my PhD project is to gain fundamental insights of photocatalytic processes on metal(oxide)-semiconductor hybrid materials on a molecular scale under defined ultra-high vacuum conditions. Studies during my Master thesis have already revealed that a new mechanism for the H2 formation on TiO2(110) decorated with Pt clusters occurs in such systems. It was found that molecular hydrogen is formed by thermal recombination at the clusters, and, not as generally accepted, by the reduction of protons by photoelectrons. This concept will be applied to Pt single atoms and clusters of more abundant metals, like Ni or Cu. For the transfer of the mechanistics of alcohol reforming to full water-splitting, the deposition of an appropriate oxygen evolution co-catalyst is essential. Prominent examples for these are ruthenium oxides, nickel oxides, or copper oxides. Atomically precise clusters of these materials will serve as model co-catalysts for the oxidation side of photocatalytic reactions. By optimizing both co-catalysts on the semiconductor, it will be revealed how particular effects, such as cluster size and coverage, determine the overall photocatalytic activity. In the last part of my PhD project, GaN semiconductors will be used, because their properties, as for example their band gaps or charge carrier concentrations, can be tuned by different types of doping (n- or p-type), their doping level or alloying. These studies will reveal the importance of semiconductor-related properties on the activity of the photocatalysts. By knowing the influence of all these parameter it will eventually be possible to develop more efficient photocatalysts.