Ongoing industrial growth and technological progress have contributed to the release of various substances into the environment, including heavy metals, inorganic compounds, and polar or non-polar organic pollutants. Therefore, the proper treatment and monitoring of industrial wastewater is essential for mitigating environmental pollution and safeguarding public health. Wastewater treatment plants (WWTPs) play a vital role in reducing the environmental impact of industrial effluents. However, their efficiency often may be challenged by the presence of unknown and emerging pollutants. During routine monitoring, this category of pollutants remains unaccounted for, as traditional approaches focus primarily on sum parameters or target compounds. This thesis investigated the application non-target screening (NTS) via liquid chromatography-high resolution mass spectrometry (LC-HRMS) in combination with novel data analysis techniques, for improved industrial wastewater monitoring, offering the detection of both known and unknown contaminants. It aimed to establish an advanced, routine and preventative monitoring framework for assessing the influent, thereby increasing WWTP’s efficiency.

The thesis initially focused on the first step of NTS data analysis, feature extraction. Various open-source and commercial feature extraction software tools, namely MarkerView, MZmine3, OpenMS, SIRIUS and XCMS were evaluated for their performance in time trend analyses and identifying patterns within an industrial wastewater matrix. This comprehensive comparison incorporated several techniques such as Spearman’s rank correlation and other multivariate statistical approaches, including principal component analysis (PCA) and sparse principal component analysis (SPCA), which is applied here for NTS data for the first time. Despite optimized parameters, the results showed notable differences in feature recall rates, with software under- and overestimating the number of detected features due to their underlying mathematical frameworks. These discrepancies directly molded the structure of the resulting data matrices, which in turn led to inconsistent outcomes in the subsequent statistical analyses. As a prioritization approach with SPCA, 70% and 90% sparsity levels were applied to identify features causing the highest variance, and to clearly identify intensity patterns in the datasets. XCMS, OpenMS, and MZmine3 demonstrated better harmony in their outputs, consistently prioritizing more common features. The study also concluded that the use of at least two peak detection tools can reduce the number of false negatives and can improve the robustness of marker detection in complex longitudinal datasets.

Building on this foundation, a direct application of this powerful technique was realized through the development of a prioritization tool tailored for monitoring the wastewater system of a chemical industrial park, with the goal of protecting the WWTP and preventing the potential release of contaminants. Developed in MATLAB, this tool enables simultaneous spatial and temporal prioritization of features detected in the WWTP influent. The first module of this tool allows the creation of wastewater feature fingerprints of localized wastewater sources, which in this study involved diverse chemical plants. Therefore, each feature fingerprint corresponded to a specific production line, allowing the tracking of features in the overall influent back to their emitting plant. The second module carries out a temporal feature prioritization in consecutive influent samples based on their intensity trends. A 7-day study of the WWTP’s influent following the creation of fingerprints of 36 plants revealed that although the majority of the features could be traced to their release source, a percentage of the features were not accounted for, due to various reasons such as  in-sewer transformations. By combining these feature prioritization techniques, the tool allows for early detection of pollution sources and fluctuations in wastewater composition, providing WWTP operations with a proactive strategy to mitigate pollution events before they affect treatment efficiency. It is worth mentioning that the application of this novel tool is not limited to chemical industrial parks and may be applied to any water source apportionment.

Another major contribution of this study was the development of an extended LC-HRMS workflow for detecting highly polar compounds, which are often overlooked by conventional reversed-phase chromatography. A zwitterionic hydrophilic interaction liquid chromatography (ZIC-HILIC) method was optimized and validated for industrial wastewater analysis, significantly enhancing the detection capabilities for polar contaminants. The results demonstrated that incorporating ZIC-HILIC into routine NTS workflows increased the range of detectable substances, providing a more comprehensive overview of wastewater composition.