Abstract:
Per- and polyfluoroalkyl substances (PFAS) are a large group of anthropogenic organic chemicals that are characterized by their carbon fluorine bonds. Due to their unique characteristics that include high stability, oil- and water-repellent properties and more, they are used in countless consumer products and industrial processes. While in perfluorinated compounds every C-H bond is substituted by a C-F bond, polyfluorinated compounds are only partially fluorinated and consist also of a hydrocarbon part. The fully fluorinated moieties of PFAS, usually perfluoroalkyl or perfluoroether chains, are extremely persistent in the environment. As a result of this persistence combined with the extensive use, PFAS can be detected everywhere in the environment as well as in humans. While some polyfluorinated compounds are partially transformed in the environment, they eventually form persistent end products such as perfluoroalkyl acids (PFAAs) which is why they are also referred to as precursors. The sheer number of individual PFAS makes their analytical detection very challenging. Since PFAS are industrial chemicals, the availability of reference standards is limited. Therefore, chromatographic techniques coupled to high-resolution mass spectrometry (HRMS) are necessary for a more comprehensive characterization of the PFAS in the environment.
In the first part of this dissertation, methods for a prioritization and subsequent identification of PFAS in both environmental and consumer product samples were developed and validated in selected case studies. Since during so-called non-target screening (NTS), large datasets are acquired, an efficient prioritization is crucial to isolate the analytes of interest from background signals and other detected compounds. In case of PFAS, their intrinsic properties were used to separate them from other organic compounds in the HRMS datasets. The novel MD/C-m/C approach that uses the chemical mass defect (MD) and the mass (m) normalized to the carbon number (C) was theoretically evaluated with ~490,000 chemical formulas from online database to show that a wide range of PFAS can efficiently be separated from non-PFAS in the presence of other substances. PFAS with at least 55 mass percent of fluorine, a F/C ratio > 0.8, or a H/F ratio < 0.8 were separable from other compound classes showing the great potential to remove unwanted compounds from HRMS datasets. To also prioritize PFAS fragmentation spectra, important fragment mass difference (or neutral losses) of common PFAS were determined, evaluated, and used to detect MS/MS spectra and identify PFAS in both extracts of coated papers and soils. This approach, combined with other NTS techniques was used to identify and semi-quantify several novel PFAS in a highly contaminated agricultural soil from north-western Germany where over 70 PFAS were detected that were previously unknown on this site. The total concentration was estimated to be > 30 mg/kg total identified PFAS. Furthermore, the contamination was almost entirely dominated by perfluorinated compounds including SF5-perfluoroalkyl sulfonic acids which indicated a unique source of contamination.
Eventually, several existing and here developed NTS techniques were combined into PF∆Screen, an open-source Python-based NTS tool for a vendor-independent prioritization and partial annotation of PFAS in HRMS raw data. The functionality of PF∆Screen was demonstrated by its application to four contaminated soils from south-western Germany where over 80 PFAS were identified including novel transformation products.
In the second part, a photocatalytic oxidation method (PhotoTOP) was developed that allows the quantitative conversion of unknown precursors to their terminal end products perfluoroalkyl carboxylic acids (PFCAs). The PhotoTOP uses the production of hydroxyl radicals from irradiation of TiO2 (UV/TiO2) for the conversion of precursors in different samples and is complementary to the total oxidizable precursor (TOP) assay. Known precursors could be quantitatively oxidized to PFCAs and the PhotoTOP was shown to be able to almost completely conserve the perfluoroalkyl chain lengths of the oxidized precursors. This allows the prediction of chain-lengths of unknown precursors and was demonstrated by oxidation of pre-characterized PFAS-coated paper samples. A second promising advantage of the PhotoTOP is the absence of salts which simplified subsequent sample preparation and makes further direct injection with electrospray ionization MS possible.
To this end, the PhotoTOP was compared with the performance of complementary methods such as the direct TOP assay, hydrolysis (total hydrolysable precursors) and fluorine sum parameters (extractable organic fluorine and total fluorine; measured by the Federal Institute for Materials Research and Testing (BAM)) to characterize non-extractable side-chain fluorinated polymers (SFPs) in functional textiles. It was shown that several textiles contained high concentrations of short and long-chain perfluoroalkyl side chains that are not extractable and therefore not amenable to mass spectrometry without prior chemical conversion.