Environmental Forensics of Industrial Wastewater based on Non-Target Screening

Abstract

The discharge of industrial wastewater into surface water requires constant monitoring of the water quality to comply with specifications and quality requirements. Furthermore, it must be ensured that individual approvals for the production plants are obtained when industrial wastewater is discharged into the wastewater treatment plant (WWTP). For this reason, the influent of the WWTP is continuously monitored for trace organic compounds (TrOCs). These compounds are usually determined using liquid (LC) or gas chromatographic (GC) target methods coupled to low resolution mass spectrometers. However, these methods monitor only a small part of TrOCs in wastewater. Only a limited number of compounds can be detected in a single run and many compounds are ignored in the analysis as they are not part of the target list. Thus, unknown TrOCs neither can be detected nor identified in wastewater samples using these methods, even if they are present in high concentrations. Therefore, high resolution mass spectrometers (HRMS) have become more and more common in water analysis to carry out more extensive monitoring by detecting both known and unknown compounds. Besides, in combination with a non target screening (NTS), the generated HRMS data additionally enable the identification of unknown compounds. The application of LC HRMS in NTS related to industrial wastewater data is described in this work. Sample treatment procedures and an analytical LC HRMS method are developed which enable the sensitive and reliable monitoring of TrOCs in a broad polarity range. Additionally, the development of a reliable data processing algorithm for NTS is part of this work. A large amount of data is produced in LC HRMS that cannot completely be evaluated. Thus, prioritisation methods are required, enabling data reduction. As a result, three prioritisation strategies were developed, which make it possible to extract relevant features (a combination of a particular mass to charge ratio, the associated retention time and intensity) from the data for identification. The relevance of each feature depended on the prioritisation strategy. The first prioritisation method selects these features, whose intensities followed rising or falling trends over time series measurements. As a result, influences on industrial wastewater through the different production processes in an industrial park were recognised. This method was carried out by principal component analysis (PCA) and group wise PCA (GPCA). 130 of initially 3303 detected features were prioritised in the WWTP influent samples. In addition to prioritisation, the introduced method enabled componentisation (grouping of several features into one TrOC). As proof of concept, one feature with an increasing trend over five months was identified as N methylpyrrolidone. In the second trend related prioritisation method, the time series investigations were linked to spatial trends. For this purpose, several sampling sites before and after the WWTP were sampled and analysed over five months. This allows evaluating the treatment procedure of industrial wastewater over time. Besides, site specific features were detected. In future studies, these features could serve as a fingerprint in the monitoring of the wastewater streams. In contrast to the first two, the latter prioritisation method shows a more technical approach. TrOCs, which were repeatedly detected in the influent of the WWTP by the routine monitoring, but which were initially not identified (‘known unknowns’) are prioritised and identified by an (offline) two dimensional LC coupled to two kinds of detection techniques (ultra violet detection and MS). The identification of these ‘known unknowns’ is of high interest for the operators of the WWTP. LC UV peaks from wastewater samples were fractionated manually in the first dimension and elucidated in the second dimension, the LC HRMS, by NTS. By applying this method, the analysis of only one sample fraction led to sampling purification and therefore, data reduction. As an example, the ‘known unknown’ with the retention time of 41.1 minutes and the maximum UV absorption of 240 nm in the first dimension was successfully identified as a dichlorodinitrophenol isomer. All in all, this work shows that the use of HRMS data, in combination with NTS and the application of the presented prioritisation methods, extends the monitoring of industrial wastewater and permits an evaluation of the WWTP processes. It indicates great potential for future establishment in routine monitoring.