Levels of per-and poly ﬂ uoroalkyl substances (PFAS) in various wastewater-derived fertilizers – analytical investigations from di ﬀ erent perspectives †

Solid wastewater-based fertilizers were screened for per-and poly ﬂ uoroalkyl substances (PFAS) by the extractable organic ﬂ uorine (EOF) sum parameter method. The EOF values for ten sewage sludges from Germany and Switzerland range from 154 to 7209 m g kg − 1 . For thermal treated sewage sludge and struvite the EOF were lower with values up to 121 m g kg − 1 . Moreover, the application of PFAS targeted and suspect screening analysis of selected sewage sludge samples showed that only a small part of the EOF sum parameter values can be explained by the usually screened legacy PFAS. The hitherto unknown part of EOF sum parameter contains also ﬂ uorinated pesticides, pharmaceutical and aromatic compounds. Because these partly ﬂ uorinated compounds can degrade to (ultra-)short PFAS in wastewater treatment plants they should be considered as signi ﬁ cant sources of organic ﬂ uorine in the environment. The combined results of sum parameter analysis and suspect screening reveal the need to update current regulations, such as the German fertilizer ordinance, to focus not solely on a few selected PFAS such as per ﬂ uorooctane sulfonic acid (PFOS) and per ﬂ uorooctanoic acid (PFOA) but consider an additional sum parameter approach as a more holistic alternative. Moreover, di ﬀ usion gradient in thin-ﬁ lms (DGT) passive samplers were utilized as an alternative simpli ﬁ ed extraction method for PFAS in solid wastewater-based fertilizers and subsequently quanti ﬁ ed via combustion ion chromatography. However, the DGT method was less sensitive and only comparable to the EOF values of the fertilizers in samples with >150 m g kg − 1 , because of di ﬀ erent di ﬀ usion properties for various PFAS, but also kinetic exchange limitations.


Introduction
Per-and polyuoroalkyl substances (PFAS) are synthetic chemical and include more than 10 000 compounds. 1 PFAS have been used extensively in a variety of products and industries due to their inert chemical stability and resistance to degradation by heat or acids. 2 Thousands of industrial and military installations have been found to contain contaminated soil surfaces and groundwater resources. 3][17][18][19][20][21] The German legislation banned sewage sludge/biosolid application on agricultural land through the amendment of the German fertilizer ordinance 22 and by 2029 sewage sludge will be totally prohibited from agricultural application.While environmental exposure of organic pollutants like PFAS and antibiotics is no longer desirable, phosphorus (P) from sewage sludge should still be recycled in WWTPs of cities with a population larger than 50 000 residents.To produce high-quality Pfertilizers for a circular economy, German sewage sludge has to undergo treatment. 23Currently, plant-available P-fertilizers from recycled materials (e.g.sewage sludge, sewage sludge ash, wastewater precipitates) can be produced using a variety of treatment approaches including precipitation, leaching, and thermal treatment. 24,25During thermal treatment, the sewage sludge is heated to varying temperatures to destroy pathogens and organic matter.Since there is a lack of universal regulation, some thermal treatments (e.g.low-temperature conversion/ pyrolysis) are conducted at temperatures of 400-500 °C which might not be sufficient for a complete PFAS decomposition.Other incineration techniques may reach higher temperatures in the range of 850-900 °C, at which most PFAS decompose. 26,27owever, in the absence of hydrogen sources, incomplete degradation takes place, yielding shorter chained and more volatile PFAS. 26,28urrently, only the sum amount of peruorooctane sulfonic acid (PFOS) and peruorooctanoic acid (PFOA) is monitored via the German fertilizer ordinance 23 with a limit of 100 mg per kg dry matter by German regulation.Due to the strong diversity of industrial PFAS usage 29 this limitation is not adequate to ensure safe application of novel recycled P-fertilizers from WWTPs on agricultural elds.Moreover, wastewater and sewage sludge contain also uorinated pesticides 30 and pharmaceuticals 31,32 which potentially break down to form ultrashort PFAS in WWTPs. 33In addition, the European Union has proposed to ban the use of all (non-essential) PFAS to reduce pollutant entry to the environment. 34tate-of-the-art method liquid chromatography tandem mass spectrometry (LC-MS/MS) is the most comprehensive tool for characterizing many PFAS in environmental samples.However, the high number of compounds in this group renders this type of comprehensive analysis with justiable efforts prohibitive.In order to determine the concentration of "total" PFAS in wastewater-based fertilizers, novel PFAS screening methods are needed.In previous work of several researchers, the sum parameter extractable organic uorine (EOF) by combustion ion chromatography (CIC) was successfully applied for water, sewage sludges, sediments and soil samples. 5,19,35n this work, our goal was on the one hand to analyse PFAS in wastewater-based fertilizer using EOF coupled to CIC for "total" PFAS followed by comparing subset of results with classical LC-MS/MS target analysis, and one selected sample by HR-MS suspect screening.On the other hand, we compared the EOF method with diffusive gradients in thin-lms (DGT) passive sampler, 36 which also can be used for extraction of solid-liquid mixtures (e.g.soil, sewage sludge) 37 (see Fig. 1).
Because the DGT technique is based on diffusion, in contrast to chemically extraction of the EOF method, this technique might underestimate the amount of PFAS due to sorption and resupply on sludge particles.Moreover, the PFAS adsorption on the DGT binding layer was investigated via Fourier-transform infrared (FT-IR) and uorine K-edge X-ray absorption nearedge structure (XANES) spectroscopy, respectively.

Sampling
Ten dried sewage sludge (SL) samples from various wastewater treatment plants in Germany and Switzerland, six sewage sludge ashes (SSA) from Germany, six thermally treated SL and SSA samples with different additives (temperatures: 700-1050 °C), two low-temperature converted (LTC) SL samples (pyrolysis, temperature: 400 °C) 38 and two struvite samples from Germany and Canada were analysed in this study.

Sample extraction and preparation for quantitative EOF analysis
For determination of the sum parameter EOF, all samples were extracted and prepared in triplicate according to our previously reported work. 19Detailed information can be found in the ESI.†

Sum parameter analysis (EOF and DGT)
All extraction samples (EOF and DGT) were quantied by CIC similar to previously reported work. 19The methanol extracts of all samples (500 mL) were injected into quartz wool lled ceramic boats before being measured by CIC.All samples were measured in duplicates to maintain quality in data consistency.In order to quantify the correct absorption volume, an internal standard of known concentration was added to the absorption solution before each combustion step.Quantication of the samples was enabled using an eleven-point calibration curve from 1 to 20 mg per L F − (R 2 = 0.995) for low uoride containing samples and a six-point curve from 10 to 500 mg per L F − (R 2 = 0.999) for higher uoride value detection.The calibration solutions were prepared from KF stock solution and calibration was performed by combustion of the respective aliquots via the CIC instrument.All EOF values are given in mg per kg dry weight.

Targeted PFAS analysis by LC-MS/MS
Quantitative LC-MS/MS analysis was performed for the SL samples 1, 4, 5, 6, 8 and 10 according to German standard DIN 38414-14:2011-08, 39 since they revealed elevated levels of EOF per sample.Therefore, 0.5 g of each SL sample was weighed in conical polypropylene (PE) tubes and subsequently extracted with 2.5 mL methanol in an ultra-sonic bath for 30 min.Prior to that, 50 mL of the 13 C-labelled recovery standard solution (Wellington Laboratories, Canada) was added.Aer settling for 1 h, an 500 mL aliquot of the supernatant was ltered using a 0.45 mm cellulose lter and diluted with 500 mL ultra-pure water (Milli-Q).All samples were extracted in duplicates.
Analyses for SL samples were performed using an Agilent 1260 HPLC and an AB SCIEX TSQ 6500 as mass selective detector.7 mL of the samples were injected and separated on a Nucleodur C 18 Pyramid pre-column (8 mm × 3 mm, 3 mm) and a Nucleodur C 18 Pyramid column (125 mm × 2 mm; 3 mm) (both Macherey-Nagel, Düren, Germany) at 35 °C and a ow rate of 0.3 mL min −1 using the following gradient program using water with 10 mM ammonium acetate solution (eluent A) and methanol (eluent B): the eluent composition of 75% A and 25% B at the beginning changed at 9 min to 25% A and 75% B. The ion source was operated at 425 °C and with an ion spray voltage of 1200 V and all measurements were executed in the multireaction mode (MRM).

Diffusive gradients in thin-lms (DGT) extraction
DGT devices (window size: 2.54 cm 2 ; 0.8 mm agarose diffusion layer) with a weak anion exchanger (WAX; thickness 0.5 mm) binding layer 36 from DGT Research, Lancaster, UK were used for the experiments.The fertilizer samples were incubated at 100% water holding capacity for one hour, then the DGT devices were deployed in duplicate on the samples for 24 h at 23 °C.Aer deployment, DGTs were removed from the samples and disassembled, and the binding layers were obtained and eluted in 3 mL of MeOH for 24 h.The separated eluates were slowly evaporated to dryness by N 2 gas and subsequently reconstituted in 1.5 mL of fresh MeOH.Blank DGT devices were kept in the refrigerator until the processing and extraction of the DGT devices.
Furthermore, DGT devices were loaded with 200 mL solutions (50 mg F per L) of various PFAS compounds (Na-TFMS, PFOS, Na-TFA, PFOA, HFPO-DA and PFOPA).The DGT devices were deployed for 24 h at 23 °C in plastic bakers in constantly agitated solutions.Aer deployment, the binding layers of the DGT devices were dried at room temperature and spectroscopically investigated as described below.

Fourier-transform infrared (FT-IR) spectroscopy
FT-IR spectra of the DGT binding layers were collected with the PerkinElmer 2000 FT-IR spectrometer.The binding layers were measured with the SensIR DuraSamplIR II attenuated total reection (ATR) module (spectral resolution 8 cm −1 ; 32 scans were coadded per spectrum).All FT-IR spectra were blanksubtracted and normalized (min-max to 1495-1382 cm −1 region) with the soware OPUS (Bruker, Version 7.0).

Fluorine K-edge X-ray absorption near-edge structure (XANES) spectroscopy
Fluorine K-edge XANES spectra of the DGT binding layers were collected on the PHOENIX II beamline of the Swiss Light Source (SLS, Villigen, Switzerland).The experiments were conducted at room temperature under a high vacuum (10 −6 mbar).Bulk-XANES spectra were collected from an area of approx. 2 × 3 mm at the sample over the range 660-780 eV in uorescence mode, using a silicon dri diode (SDD, manufacturer: Ketek).The collected spectra were normalized, and background corrected using the Athena soware from the Demeter 0.9.26 package (Ravel and Newville 2005).

Extractable organic uorine (EOF) of wastewater-based fertilizers
First the EOF values of all 26 wastewater-based fertilizers were analysed.Fig. 2, Tables S1 and S2  The EOF values of the SLs mainly range between 154 and 538 mg kg −1 except for SL1 which showed an elevated EOF value of 7209 mg kg −1 due to high PFAS and other organouoride contamination.These EOF values are in good agreement with EOF values for previously reported north European SL samples. 5or the SSA samples the EOF values were lower and values between LOQ (approx.60 mg kg −1 ) and 121 mg kg −1 could be detected.From SSA4 and SSA5 two samples out of the triplicate were negative aer blank correction.The rst three SSA samples (SSA1 to SSA3) were taken from a grate ring incinerator and samples SSA4 to SSA6 aer incineration via uidized bed combustion.This is an indication that the type of SL incineration may affect the PFAS degradation rate.Additionally, we analysed the activated carbon adsorber of the off-gas cleaning aer the electric precipitator from the mono-incineration facility SSA4. 40This material was quantied with an EOF value of 157.9 ± 10.4 mg kg −1 which indicates that aer incineration some organouoride compounds are still detectable in the offgas stream.
The various fertilizers derived from thermal treatment of SLs and SSAs, partly contain organo uorinated compounds with EOF values up to 88 mg kg −1 .For the pyrolyzed SLs from LTC no EOF values above the LOQ could be detected.Moreover, the two wastewater-based struvite fertilizers contain 96 and 112 mg per kg EOF, respectively.Altogether, relevant amounts of EOF could be systematically detected in almost all types of fertilizers from wastewater.

Sum parameter analysis EOF vs. targeted PFAS analysis
The EOF parameter provides a good overview on the amount of PFAS and other uorinated compounds in SLs and wastewaterbased fertilizers.However, it cannot give any information on the specic type of PFAS in these fertilizers.In addition, the German fertilizer ordinance 23 only prescribes a limit value based on the sum of the PFAS targets PFOS and PFOA.Therefore, several selected SL samples were analysed with regard to PFAS targets according to German Standard DIN 38414-14:2011-08, 39 presented in Fig. 3 and Table S3.† In contrast to the EOF values, the sum of PFAS target values are relatively low for all SLs.It has to be noted that the EOF values refer to the amount of uorine in the extract solution.Since many PFAS contain approx.60% (w/w) uorine the EOF values of PFAS would be expected to be higher.Surprisingly, the sum of PFOS and PFOA for SL1 is only approx.11 mg kg −1 even if the EOF has a very high value of 7209 mg kg −1 .However, this agrees with previous examples of Aro et al. 5 who showed that only a small portion of organouorides can be identied in SLs from several north European countries.Moreover, all analysed SLs in the present study would be opened to be used as fertilizers in Germany since their values lie below the ordinance limit of 100 mg kg −1 (sum of PFOS and PFOA).Since the total sum of all analysed PFAS of SL1 by LC-MS/MS was approx.104 mg kg −1 compared to its EOF value (7209 mg kg −1 ), we applied a LC-HRMS based suspect screening approach for qualitative iden-tication of additional PFAS and other uorinated compounds in this SL sample.

PFAS suspect screening analysis
The PFAS suspect screening approach aimed to tentatively identify PFAS that could contribute to the hitherto unknown part of the EOF value.Fig. 4 shows the compound classes of identied PFAS and uorinated compounds in SL1, while Table S4 † lists the individual identied compounds.
The majority of the detected uorinated compounds are legacy PFAS such as short-and long-chain peruorocarboxylic acids (PFCA), peruorosulfonic acids (PFSA), polyuoroalkyl phosphate esters (PAPs) and peruorophosphonic acids (PFPA).Furthermore, the emerging PFAS hexauoropropylene oxide dimer acid (HFPO-DA, the acid of GenX) could be identied and was classied as "others" (see Table S4 †).Moreover, uorinated  S1. † From SSA4 and SS5 two samples (EOF) out of the triplicate were negative after blank correction, hence, values were omitted.Some of the EOF values were published previously 19 see Table S1; † from SL2, SL8 and SSA1 one sample each (DGT) was negative after blank correction, hence, values were omitted.LTC = low-temperature conversion (pyrolysis).pesticides (ufenacet and udioxonil), pharmaceutical (ufenamic acid and pitavastatin) as well as aromatic compounds (4,4 ′ -diuorobenzophenone and 1-cyclopropyl-6,7-diuoro-1,4dihydro-8-hydroxy-4-oxo-3-quinoline-3-carboxylic acid) were also identied, which are all included in the EOF parameter.Our ndings agree well with previous work of others 18,41,42 who were able to detect a broad range of different PFAS and uorinated compounds in SLs/biosolids from France, Sweden and the US by LC-HRMS suspect screening.

Diffusive gradient in thin-lms (DGT) technique for PFAS extraction
Moreover, the DGT technique was applied for PFAS extraction from the wastewater-based fertilizers.Fig. 2 and Table S2 † show the PFAS-DGT values for selected fertilizers (orange spots).In contrast to the EOF values which are stated as weight in mass concentration of the sample, the PFAS-DGT values are presented in absolute values of ng PFAS-uorine per DGT binding layer.Similar to the EOF value, the DGT extraction of SL1 yielded the highest amount of 2.2 ng absolute uorine.
Overall, there is a correlation between the EOF and the PFAS-DGT values which has a correlation coefficient R 2 of 0.98 (see Fig. 5).Since for both applied extraction methods (EOF and DGT) weak anion exchange resin-based materials (WAX; for EOF a combination of WAX and graphitized carbon black) were used, a comparable pool of PFAS and uorinated compounds was accessed in the fertilizer samples.However, the R 2 -value of the correlation in Fig. 5 is highly dependent on the value of SL1.Without SL1 the R 2 -value is only 0.42.Because wastewater-based fertilizers contain various PFAS 42 and different PFAS have different diffusion properties this could be an explanation for the dispersion of the DGT values.Moreover, DGT is sampling the freely available/labile portion of PFAS in the sample and Huang et al. 43 showed that pH and texture of a samples are important properties which controls the labile pool size of PFAS.They showed also that the rate of supply of PFAS to DGT devices in soil was controlled by the kinetics of release from the solid phase to the solution phase, which may also apply for wastewater-based fertilizers.
The kinetic exchange limitations for PFAS of the DGT method might be also the reason that for some wastewaterbased fertilizers (SSA2, the thermal treated SL (TT1) and stru-vite1) the DGT method did not show results above the blank DGT value of approx.0.11 ng per binding layer which correspond to an EOF value of approx.150 mg kg −1 .Therefore, the EOF method is in this case more sensitive than the DGT method.

FT-IR spectroscopy of PFAS loaded DGT binding layers
In order to prove that PFAS and uorinated organic compounds adsorb to the WAX containing binding layer of the DGT device, several approaches for direct spectroscopic analysis on the binding layer are known. 44,45Therefore, the DGT devices were saturated using various aqueous PFAS solutions (all 50 mg F per L).Aerwards the binding layers were dried and spectroscopically investigated by FT-IR and uorine K-edge XANES spectroscopy, respectively.Fig. 6 (top) shows the FT-IR spectra of various PFAS adsorbed to the DGT binding layer where the blank binding layer spectra was subtracted (the raw data are presented in Fig. S1 †).
The ultrashort-chain PFAS compounds TFMS and TFA show a lower signal-to-noise ratio in the FT-IR spectrum compared to the long-chain PFAS which indicate a lower amount of adsorbed PFAS to the DGT binding layer.All PFAS (except TFA and TFMS) show a strong n(CF 2 ) band at approx.1250 cm −1 , which is in good agreement with previous works. 46,47In opposition, TFA and TFMS show the strong n(CF 3 ) band at lower wavenumber of approx.1203 cm −1 and 1223 cm −1 , respectively, which also agrees with previous ndings. 48,49Furthermore, there are other detectable IR bands of each PFAS which belong to the functional groups of each molecule. 46,50For example, the IR band at 1215 cm −1 of the DGT-PFOS spectrum can be attributed to the  R-SO 3 vibration of the sulfonate group. 46Thus, FT-IR spectroscopy is a viable method to identify pure PFAS compounds adsorbed to the DGT binding layer.
Additionally, the DGT devices were deployed to the sludges SL1 to SL3, because they have the highest EOF values, and the binding layers investigated by FT-IR spectroscopy (see Fig. 6 bottom).As observed before, all three samples show a n(CF 2 ) vibration around 1250 cm −1 .However, the additional IR bands of the three SL samples are identical and may be attributed to other organic molecules in the SL.Therefore, it was impossible to identify individual PFAS compounds in the SLs by the DGT/ FT-IR spectroscopy approach.

Fluorine K-edge XANES spectroscopy of loaded DGT-layers
As reported previously by Yan et al. 51 PFAS adsorption to clay can be investigated by means of surface spectroscopic methods such as the scanning transmission X-ray microscopy (STXM)-XANES technique.Based on these results, we investigated the PFAS adsorption to the saturated DGT binding layer also by uorine K-edge bulk-XANES spectroscopy (see Fig. 7), because is surface sensitive with a penetration depth of max. 3 mm at the uorine edge energy.
However, for the saturated DGT binding layer with the ultrashort-chain PFAS TFMS, TFA and HFPO-DA no spectra could be collected.One possible reason is that due to the ultrahigh vacuum these volatile PFAS evaporate quickly.Moreover, the radiation beam warms up the sample which probably enhance the evaporation.Hence, only for the PFAS compounds PFOS, PFOA and PFOPA-uorine K-edge XANES spectra could be collected for the loaded binding layers.Although the signalto-noise ratio is rather low, the uorine edge of these samples is higher than the one of the DGT blank.Therefore, the presence of an organic uorine compound on the DGT binding layer could be conrmed. 19Thus, for non-IR transparent samples, like activated carbon, the uorine K-edge XANES spectroscopy technique is a useful tool to analyse the adsorption of PFAS.

Conclusions
Currently, different treatment methods for nutrient recycling from WWTPs are under development in Germany.Both EOF, as well as target analytical data showed that SLs and currently available wastewater-based fertilizers contain PFAS.Additional LC-HRMS suspect screening indicate that SLs contain a broad range of PFAS and other uorinated organic compounds, such as common pesticides and pharmaceuticals.PFAS sum parameter (e.g.EOF/CIC) and suspect screening by LC-HRMS are useful techniques to identify PFAS contaminations in fertilizer samples, especially for research.Target analysis is normally applied for quantication of specic PFAS and used for regulation.But since the number of known PFAS already exceeds 10 000, the currently ordinance limit of 100 mg per kg PFOS + PFOA in the German Fertilizer Ordinance is no longer up to date.With the planned REACH restriction of the EU that is covering the use of all PFAS (except those of essential use) it should also be considered to use a PFAS sum parameter limit in some form for fertilizer regulation. 5,52As an example, an approach for "total" PFAS was already implemented in the EU drinking water directive. 53The sum parameter EOF includes besides legacy PFAS also uorinated pesticides and pharmaceuticals, which can also end up as ultrashort-chain PFAS in the WWTPs. 33Other sum parameters like the total oxidisable precursor assay (TOPA), do not include uorinated pesticides and pharmaceuticals but are limited on available 13 C-marked internal standards for LC-MS/MS.
Furthermore, we presented the potential of the simplied liquid extraction with the DGT passive sampler as a monitoring tool for PFAS in wastewater-based fertilizers, when DGT extraction gets analysed with CIC analysis.However, in comparison with the EOF/CIC method the DGT/CIC values were more widely spread because wastewater-based fertilizers contain various PFAS including uorinated pharmaceutical and pesticides 42 which have different diffusion properties, but also kinetic exchange limitations have been observed for the DGT method previously. 43Therefore, the DGT approach has the advantage that almost no sample preparation is necessary, but the EOF method is more sensitive to analyse low amounts of "total" PFAS in wastewater-based fertilizers.
Moreover, we found that thermal treatment of SL in different form 54 reduces the amount of PFAS in the fertilizer product, however, there is still PFAS pollution detectable to some extent.The lowest PFAS values were detected for the pyrolyzed products which is also in good agreement with previous studies. 55,56owever, it has to be kept in mind that thermal treatment of PFAS containing fertilizers may produce volatile ultrashortchain PFAS 57,58 or organouoride compounds like CF 4 , CHF 3 etc. 59which have a high global warming potential, and may aerwards enter the atmosphere, depending on the used lter systems in thermal treatment plants. 26,28aper Environmental Science: Advances
† show the EOF and DGT Paper Environmental Science: Advances Open Access Article.Published on 08 September 2023.Downloaded on 9/27/2023 11:55:22 AM.This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.values of the various SLs and wastewater-based fertilizers.The DGT values will be discussed in the fourth subsection.

Fig. 2
Fig. 2 EOF (green bars; mg kg −1 ) and DGT results (orange points; ng PFAS-F per binding layer) of various fertilizers from wastewater-based materials.The LOQ is for the EOF values.The exact standard deviation of SL1 is displayed in Fig. 5 and TableS1.† From SSA4 and SS5 two samples (EOF) out of the triplicate were negative after blank correction, hence, values were omitted.Some of the EOF values were published previously19 see TableS1; † from SL2, SL8 and SSA1 one sample each (DGT) was negative after blank correction, hence, values were omitted.LTC = low-temperature conversion (pyrolysis).

Fig. 6
Fig. 6 Normalized and background subtracted FT-IR spectra of loaded DGT binding layers with PFAS (top) and DGT binding layers applied to wetted SL samples (bottom).The strong n(CF 2 ) and n(CF 3 ) band, respectively, is marked in yellow.Spectral cutout from 1400 to 1100 cm −1 .

Fig. 7
Fig. 7 Fluorine K-edge bulk-XANES spectra of loaded DGT binding layers with PFAS and corresponding PFAS reference compounds.