Per- and polyfluoroalkyl substances (PFAS) are thermally persistent and can form degradation products during waste incineration. Despite this, standardised methods for measuring PFAS in flue gases have only been developed in recent years. In this study, three PFAS flue gas sampling methods were evaluated at four Swedish waste-to-energy plants. Two of the methods target semi-volatile polar PFAS (OTM-45 and a modified method based on EN 1948-1), while one method targets volatile non-polar compounds (OTM-50). Because the methods are partly designed to capture different PFAS substance groups, the comparative part of the study focuses on the two methods targeting semi-volatile polar PFAS.
The overall aim of the study is to assess how well the methods can be used to characterise emissions under real operating conditions, rather than to provide a comprehensive quantification of total PFAS emissions. Although these methods have been developed in recent years, their application at full-scale waste-to-energy plants remains limited, particularly with respect to practical implementation, comparability, and sensitivity to spatial and temporal emission variability. This study addresses this gap through a systematic, field-based evaluation of PFAS sampling methods under real operating conditions. OTM-50, which targets volatile non-polar compounds, consistently showed substantially higher levels, with total concentrations in the range of 100–1,000 µg/Nm³. Concentrations of semi-volatile polar PFAS, sampled using OTM-45 and EN 1948-1, were generally below 1 ng/Nm³. The compounds detected at the highest concentrations across all plants were chlorodifluoromethane (HCFC-22) and trichlorofluoromethane (CFC-11). These compounds are not PFAS but are included in the OTM-50 target list.
Their presence in all sampled furnaces, as well as in blank samples, indicates that multiple sources may contribute, including incomplete destruction of CFC-containing materials, possible in-stack formation via radical reactions, and background levels. For polar PFAS, the dataset was dominated by two clearly elevated samples: an OTM-45 sample at Plant B with high concentrations of 6:2 fluorotelomer sulfonate (6:2 FTS), and an EN 1948-1 sample at Plant C with elevated levels of short-chain perfluoroalkyl carboxylic acids. Both events are considered likely to represent genuine, episodic emissions in a heterogeneous flue gas, although interpretation is partly hampered by missing sample fractions. When these outliers are excluded, a relatively consistent picture emerges for total polar PFAS concentrations across all plants, with levels below 1 ng/Nm³ and broadly similar chain-length profiles. This indicates that OTM-45 and EN 1948-1 provide comparable results at these concentration levels, while capturing different local emission events and partly different PFAS substance groups.
The comparison between methods for semi-volatile PFAS highlights several important limitations. First, the ratio between OTM-45 and EN 1948-1 varied by up to two orders of magnitude between plants, and in opposite directions. Together with the strong influence of individual outliers, this demonstrates that single 2–4 h grab samples have limited representativeness for PFAS emissions that vary both spatially and temporally. Second, internal mass balance analyses show that both OTM-45 and EN 1948-1 predominantly collect PFAS in aqueous phases and on sampling train surfaces, rather than on dedicated sorbents such as XAD-2. From a practical perspective, OTM-50 proved relatively straightforward to apply logistically, but requires access to specialised sampling canisters and GC–MS infrastructure. OTM-45 and EN 1948-1 are more complex, requiring extensive handling of glassware and solvents, as well as multi-step laboratory extractions. All three methods are labour-intensive and strongly dependent on rigorous quality control.
The risk of sample contamination, as well as the loss of a single canister or a missing sample fraction, can have a substantial impact on data interpretation at low concentration levels. Overall, the study demonstrates that PFAS measurement methods can be applied at full-scale waste-to-energy plants, but that their performance and interpretability are strongly constrained by spatial variability and episodic emissions. OTM-50 is robust for the measurement of volatile fluorinated compounds, while OTM-45 and EN 1948-1 provide broadly comparable results for polar PFAS at low concentrations, subject to sensitivity related to sampling configuration and surface interactions. The observed episodic concentration peaks indicate that the methods are sensitive to transient emission events; however, the available data and study scope do not allow assessment of the underlying causes. Consequently, results from single grab samples must be interpreted with caution if they are to be used to estimate emissions over longer time periods.
Stockholm: IVL Svenska Miljöinstitutet, 2025.