The use of light energy to drive chemical reactions has gained increasing interest for a variety of applications in the last decades. This habilitation thesis comprises a selected number of original papers from the author's own research in this field, particularly concerning semiconductor photocatalysis. Semiconductor photocatalysis has so far been explored for two main applications: The decontamination of air, water or surfaces from unwanted pollutants and the synthesis of value-added chemical products. While in principle the design criteria and required properties of photocatalysts and devices for these two applications are quite different, they also share a number of fundamental aspects. A universally important aspect is the design of efficient photocatalyst materials, which on the one hand absorb the light energy and convert it efficiently to usable chemical potential energy and on the other hand exhibit good catalytic properties so that this energy can be effectively transferred to the target molecules. Modification with metals and metal ions is a very prominent and powerful tool to improve the catalyst properties. This procedure may improve the catalyst's activity and selectivity to certain products or even alter the spectrum of usable light. However, the induced properties are very sensitive to the concentration and location of the added metals. Therefore, a fundamental model was developed that is able to predict the optimal metal concentration for a given system with good precision.Photocatalysis can be used to degrade and thereby remove harmful pollutants such as nitrogen oxides from the air. Of particular interest in the optimization of this application is the selectivity of the photocatalytic reaction. Unlike in chemical synthesis where unwanted side-products are mainly a nuisance for product isolation and purification, potentially toxic intermediates and side-products may pose severe risks in environmental applications where the population may be directly exposed to them. However, as shown herein, modification with metal ions as co-catalysts is a very effective way to engineer the selectivity and suppress the formation of side-products.Being a light-driven process, the photocatalyst's activity is dynamic with respect to the locally absorbed photon flux, which due to its exponential decay often varies by several orders of magnitude throughout the studied system. This highly inhomogeneous distribution in catalyst activity adds another layer of complexity to the process and requires specific approaches to model the kinetics and optimize these processes. Herein, a new holistic approach for this problem is presented and evaluated on the basis of two exemplary cases. Furthermore, the integration of photocatalytic reactions into combined reaction cascades with traditional chemical or biocatalytic reaction steps is discussed and kinetically analyzed.As a consequence of the exponential decay of light intensity, photocatalytic processes cannot be performed efficiently in traditional reactor types, particularly at larger scales. Herein, the fundamental challenges and basic design criteria for photoreactors are discussed. Also, a new concept for powering photocatalytic reactions based on internal illumination with wirelessly powered light sources is presented.