Open Access

The emission of anthropogenic trace substances into the aquatic environment continuously poses challenges to water suppliers. The contamination of raw waters with organic trace substances requires complex water treatment processes to secure drinking water quality. The routine monitoring of these raw waters as well as the behavior and fate of organic trace substances during different treatment processes is of great interest to recognize and counter potential dangers at an early stage. Chromatographic separation techniques coupled to triple quadrupole mass spectrometers are conventionally used for the reliable monitoring of traces of known polar substances. However, such analytical techniques usually fail to recognize unknown compounds. This weakness presents a serious restriction with regard to the monitoring of treatment processes, since transformation products are often not - or not sufficiently - characterized and are thus only detected sporadically. Non-target screening using liquid chromatography coupled to high-resolution mass spectrometry (LC-HRMS) allows the detection of thousands of compounds within a single run and covers known as well as unknown substances. Compared to the established analytical techniques, this is a decisive advantage for the monitoring of raw and process waters during water treatment. While the analytical technique LC-HRMS has undergone significant developments in recent years, the algorithms for data processing reveal clear weaknesses. This dissertation therefore deals with reliable processing strategies for LC-HRMS data. The first part of this work seeks to highlight the problematics of false positive and false negative findings. Based on repeated measurements, various strategies of data processing were assessed with regard to the repeatability of the results. To ensure that real peaks were barely or not removed by the filtering procedure, samples were spiked with isotope-labeled standards. The results emphasize that the processing of sample triplicates results in sufficient repeatability and that the signal fluctuation across the triplicates emerged as a powerful filtering criteria. The number of false positives and false negatives could be significantly reduced by the developed strategies which consequently improve the validity of the data. The second part of this thesis addresses the development of processing strategies particularly aimed at assessing water treatment processes. The detected signals were tracked across the treatment process and classified based on their fold changes. A more reliable signal classification was achieved by implementing a recursive integration approach. Special integration algorithms allow a reliable signal classification even though the signal to be compared was below the intensity threshold. Different combinations of replicates of process influents and effluents were processed for evaluating the repeatability. The good repeatability was indicated by the results of both the plausibility checks and the ozonation process (ozonation of pretreated river water) and thus points to high reliability. The developed procedure enables the assessment of water treatment processes based on the changes in the pattern of all detected signals and offers a more comprehensive picture of the treatment efficiency. Particularly with regard to transformation products, existing knowledge gaps can be reduced by this approach, albeit the entire variety of chemicals cannot be covered completely. The applicability of the developed strategies to real world applications is demonstrated in the last part of this work. Besides the prioritization of the generated results, the main focus was the identification of recognized compounds. The developed strategies clearly improve the validity of the underlying data. The combination of LC-HRMS analysis with reliable processing strategies opens up multiple possibilities for a more comprehensive monitoring of water resources and for the assessment of water treatment processes. The processing strategies and validation concepts may be easily transferred to other research fields.