Taking a look at the surface: m -XRF mapping and ﬂ uorine K-edge m -XANES spectroscopy of organo ﬂ uorinated compounds in environmental samples and consumer products †

For the ﬁ rst time, m -X-ray ﬂ uorescence ( m -XRF) mapping combined with ﬂ uorine K-edge m -X-ray absorption near-edge structure ( m -XANES) spectroscopy was applied to depict per-and poly ﬂ uoroalkyl substance (PFAS) contamination and inorganic ﬂ uoride in sample concentrations down to 100 m g kg − 1 ﬂ uoride. To demonstrate the matrix tolerance of the method, several PFAS contaminated soil and sludge samples as well as selected consumer product samples (textiles, food contact paper and permanent baking sheets) were investigated. m -XRF mapping allows for a unique element-speci ﬁ c visualization at the sample surface and enables localization of ﬂ uorine containing compounds to a depth of 1 m m. Manually selected ﬂ uorine rich spots were subsequently analyzed via ﬂ uorine K-edge m -XANES spectroscopy. To support spectral interpretation with respect to inorganic and organic chemical distribution and compound class determination, linear combination (LC) ﬁ tting was applied to all recorded m -XANES spectra. Complementarily, solvent extracts of all samples were target-analyzed via LC-MS/MS spectrometry. The detected PFAS sum values range from 20 to 1136 m g kg − 1 dry weight (dw). All environmentally exposed samples revealed a higher concentration of PFAS with a chain length > C 8 ( e.g. 580 m g kg − 1 dw PFOS for Soil1), whereas the consumer product samples showed a more uniform distribution with regard to chain lengths from C 4 to C 8 . Independent of quanti ﬁ ed PFAS amounts via target analysis, m -XRF mapping combined with m -XANES spectroscopy was successfully applied to detect both point-speci ﬁ c concentration maxima and evenly distributed surface coatings of ﬂ uorinated organic contaminants in the corresponding samples.


Introduction
Per-and polyuoroalkyl substances (PFASs) are industrially synthesized chemicals which include more than 10 000 compounds predominantly used in the formulations of numerous consumer goods and industrial applications. 1[10] The ongoing production of new, yet unrestricted PFAS alternatives has become a major challenge for environmental routine analyses, since the state-of-the-art methods liquid chromatography tandem mass spectrometry (LC-MS/MS) and gas chromatography coupled with mass spectrometry (GC-MS) rely on structural information and the availability of isotopically labelled standards of the targeted compound. 11,12Simultaneously, unspecic analytical methods such as combustion ion chromatography (CIC), [13][14][15] high resolution graphite furnace molecular absorption spectroscopy (HR-GF-MAS), [16][17][18] or inductively coupled plasma mass spectrometry (ICP-MS) 19,20 gradually gain signicance, since PFASs and other uorinated organic compounds can be determined as sum parameters, e.g.EOF (extractable organic uorine) or AOF (adsorbable organic uorine).There is a current debate on whether compounds of low uorine content, such as pesticides or pharmaceuticals, should be included in the denition of PFASs as a class of substances, as their contribution to sum values of uorinated organic pollutants in environmental samples could be of great importance. 21,22n addition, several non-destructive methods for the qualitative detection of PFASs and total uorine (TF) in environmental samples have been presented so far. 23While X-ray uorescence spectroscopy (XRF), 24 instrumental neutron activation analysis (INAA) 25 and particle-induced gamma-ray emission (PIGE) spectroscopy [26][27][28][29] can be utilized to determine the total uorine content of an environmental sample, X-ray photoelectron spectroscopy (XPS) allows for supplementary structural identication of organouorine compounds. 30,31reviously, we introduced the application of uorine K-edge Xray absorption near edge structure (XANES) spectroscopy as an alternative surface analytical method of PFAS contaminated solid matrices. 32he combination of m-XRF mapping and m-XANES spectroscopy has been successfully demonstrated for detection of a wide range of inorganic trace elements in environmental media [33][34][35] but proves difficult for light elements as in organic contaminants. 36In our study we applied m-XRF mapping in combination with uorine K-edge m-XANES spectroscopy to determine PFAS contamination, spatially resolved in environmental and exemplary consumer product samples.Compared to classical XPS surface analysis, m-XANES spectroscopy allows for a higher penetration depth of the analytical beam (approx. 1 mm), thus enabling detection of uorinated species at the immediate surface and below.More importantly, m-XRF mapping can be utilized to identify uorine-specic hotspots in the samples, prior to measurement.The use of the combined technique signicantly improves the detection ability of minor compound levels in environmental samples. 37Moreover, LC-MS/MS analysis of organic extracts of the investigated samples was conducted, in order to complement the surface analysis with quantied target analytical data and to highlight the sensitivity of uorine K-edge m-XANES spectroscopy.

Samples
Two different soil samples from known PFAS-contaminated remediation sites (Soil1 and Soil2) and three sewage sludge samples (SL1-3) from selected active wastewater treatment plants in Germany were analyzed.The initial samples were air dried at room temperature for 30 days and homogenized aerwards.Additionally, we investigated representative consumer product samples as potential sources of PFAS for environmental pollution.Therefore, two PFAS coated textile samples from used outdoor jackets (Textile1 and Textile2), one sample originating from a surface treated food contact paper (Paper1) as well as one sample from a purchased permanent baking sheet (Sheet1) were analyzed.Moreover, two PFAS coated fabric samples (Fabric1 and Fabric2) were freshly prepared via treatment of uorine-free fabric (Fabric_blank) with two different PFAS-based impregnation products (Spray1 and Spray2) to inspect PFAS surface deposition.All samples were stored in sealed polyethylene bags or containers prior to the investigation.
Sample preparation for uorine m-XANES spectroscopy All solid reference compounds were prepared as a thin layer of a few milligrams spread on a uorine-free carbon tape (uoride levels of the tape were checked via m-XRF mapping).All liquid compounds were mixed in excess with small amounts of epoxy resin adhesive (UHU® Plus Endfest), spread out to thin layers, cured for 24 hours under a fume hood and subsequently xed on carbon tape.The soil and sludge samples were pressed into small pellets for easier handling and to maintain a smooth surface prior to the measurements.All samples from consumer goods were cut out with a methanol-rinsed sharp blade to pieces of approx.5 × 5 mm and were carbon taped directly on the sample plate.Aerwards, the uorine K-edge micro-XANES spectra of each spot were complemented with linear combination (LC) tting.LCF was used to estimate the uorine compound composition of a recorded XANES spectrum based on calculated linear combination tting of manually selected reference spectra. 38All recorded reference uorine K-edge bulk-XANES spectra, calculated LCF spectra and more details on the m-XRF/m-XANES measurements are provided in the ESI.†

Sample extraction and preparation for quantitative analysis with LC-MS/MS
According to DIN 38414-14: 2011-08, 0.5 g of each soil and sludge sample was weighed in conical polypropylene (PP) tubes and subsequently extracted with 2.5 mL methanol in an ultrasonic bath for 30 min.Prior to that, 50 mL of 13 C-labelled internal standard solution was added. 39Aer settling for 1 h, an aliquot of 500 mL supernatant was ltered using a 0.45 mm cellulose syringe lter and diluted with 500 mL ultra-pure water (Milli-Q).All samples were extracted in duplicate.
For each textile and paper sample, a slightly modied approach was applied.Cut out pieces of uniform sized material were weighed in PP falcon tubes (15 mL), spiked with 100 mL of a 13 C-labelled internal standard mixture and topped with 5 mL of pure methanol.Aer that, the tubes were placed in an ultrasonic bath at 60 °C for 2 h and subsequently concentrated with a gentle nitrogen stream to a volume of 500 mL.The solutions were transferred to Eppendorf® vials and diluted with 500 mL of ultra-pure water (Milli-Q).Aer vortexing, all samples were ltered using 0.45 mm nylon syringe lters.As mentioned before, all samples were extracted in duplicate.
Liquid chromatography tandem mass spectrometry (LC-MS/ MS) analysis was performed partly on an Agilent 1260 HPLC with an AB SCIEX TSQ 6500 as a mass selective detector and an Agilent 1290 Innity II UHPLC, using negative ion mode.Further details are reported in the ESI.†

LC-MS/MS QA/QC
The isotopically labelled internal standard species were obtained from Wellington Laboratories as a prepared solution with a concentration of 50 mg mL −1 with an uncertainty of ±2.5%.For quality assurance and control an independently certied reference material (Chiron) containing all target PFAS with a concentration of 5 mg mL −1 with an uncertainty of ±5% was used.All organic solvents, reagents, and modiers used for extraction and for LC-MS/MS analysis were tested with respect to a possible blank value.All dilution and spiking steps were gravimetrically controlled.

m-XRF mapping and uorine K-edge m-XANES spectroscopy of various sample types
We applied m-XRF mapping in combination with uorine Kedge m-XANES spectroscopy to analyze low concentrated uorine compounds in environmental matrices.In order to yield the m-XRF surface maps for every sample, suitable areas of 300 × 300 mm 2 were manually selected, scanned for uorine uorescence intensity and subsequently recorded (Fig. 1 .Subsequently, F K-edge m-XANES spectra were collected at uorine hotspots.To support the interpretation of spectral data, linear combination (LC) tting was applied to all recorded F K-edge XANES spectra (see also Table S1 and Fig. S24 †).The used algorithm linearly combines up to four out of 12 reference spectra gathered from bulk-XANES measurements for each spectral t.Therefore, LC reference compounds were individually selected based on the curve shape and location of the energetic maxima.To point out the differences in the spot-specic analysis, comparative bulk-XANES spectra (2 × 3 mm beam size) were recorded for all samples.An overview of all measured uorine bulk-XANES spectra of the reference compounds is displayed in the ESI (Fig. S18-S23 †).

Environmental samples
Fig. 1 shows the m-XRF maps of uorine and oxygen (le-hand side) and uorine K-edge m-XANES spectra (right-hand side) of selected uorine hotspots of Soil1 and SL1 (top and bottom).
The element specic m-XRF maps for Soil1 (Fig. S5 †) suggest an overall high presence of uorine containing species within the sample, in which several uorine hotspots were identied.For the selected spots P1 and P2, the respective m-XANES spectra revealed a dominance of inorganic uorides, shown by the characteristic maxima at 688 and 692 eV, respectively.Similar energetic maxima can be observed for the metal uorides CaF 2 or uoroapatite.The presence of mineral uorides in soil samples is well studied and explains the high intensity of the uorine response at certain sites compared to other matrices (compare Fig. S5-S9 vs. S10-S16 †). 40It has to be noted that a dominant presence of inorganic uorides lowers the identi-ability of organouorine compounds.Nonetheless, the spectral data of P3 to P6 show a much smoother curve of the F-K edge region, indicating a minor presence of metal uorides and thus a higher occurrence of uoroorganic compounds (see also Fig. S19-S23 † for spectral comparison).
Very similar count rates were recorded for the elemental m-XRF maps of Soil2 (Fig. S6 †), indicating a comparable distribution of uorine species.Aer manual selection of the uorine hotspots P1-P6 in the maps, m-XANES spectra were recorded.For positions P1 and P2, spots of high uorine concentration were revealed (see Fig. S1 †).Their uorine m-XANES spectra show two distinct maxima at 691 and 694 eV, which again is in good conformance with spectral data of inorganic uorine compounds such as AlF 3 or Na 2 SiF 6 .As before, the spots P3-P6 exhibit two much broader maxima at 691 and 694 eV, respectively, thus pointing towards an increased presence of organo-uorine compounds (see Fig. S1 †).The low solubility of many common sources of uoride may explain the uneven distribution of uoride in our soil samples. 41Linear combination t analysis agreed well with the suspected compound distribution for both soil samples and supported the increased presence of organouorinated compounds in these parts (see also Fig. S24 and Table S1 †).
To our surprise, recording of the uorine specic m-XRF maps of all different sludge samples SL1-3 revealed a signicant presence of uorinated compounds (see Fig. S7-S9 †).Aer application of uorine K-edge m-XANES spectroscopy to pre-selected spots for sample SL1, a large variety of spectral features could be identied (Fig. 1 bottom).The spectrum recorded at position P1 exhibits two maxima at 687 and 692 eV, in good agreement with the spectral data recorded for synthetic uoroapatite (Ca 5 (PO 4 ) 3 F).This nding is supported by the applied LC t, suggesting a good conformance with uoroapatite and AlF 3 as inorganic uorides (see Fig. S24 and Table S1 †).In comparison, the locations P2 and P3 exhibit m-XANES spectra with a single broad maximum at 692 eV, which is in good agreement with the reference spectrum for the tri-uoromethylated pharmaceutical uoxetine, also proposed as an organic uorinated compound in the respective LC t.Very recent work showed the signicance of uorinated drugs and pesticides in the uorine mass balance of sewage sludge samples. 42Moreover, the spectra of spots P4 and P5 exhibit additional XANES features as observed in the spectra of 4-FBA, tolyluanid or Na-TFMS (see Fig. S19-S23 †).Furthermore, m-XANES spectra for sewage sludge samples SL2 and SL3 were recorded (Fig. S2 and S3 †).For SL2, all uorine spots show spectral data which indicate the occurrence of organic uorine compounds.As observed before, LC t analysis could conrm the higher proportion of organic compounds in the sample (see Fig. S24 and Table S1 †).According to the m-XRF map, sample SL3 revealed several spots of high uorine content aer prescanning (P1-P5, Fig. S3 and S9 †).Although the spectra of P1, P2 and P5 exhibit a slightly broadened maximum at 691 eV, they differ due to their low signal intensity, and thus, a clear iden-tication was not possible.The LC t analysis for P1, P2 and P5 suggested a mixed inorganic to organic uorine composition with a higher presence of inorganic compounds (see Fig. S24 and Table S1 †).In contrast, P3 and P4 exhibit two distinct maxima at 688 and 691.5 eV which can be clearly attributed to inorganic uorides. 43These ndings could be conrmed by their respective LC t spectra, suggesting AlF 3 and CaF 2 as the dominant species and organouorinated compounds as minor uoroorganic discharge.The presence of both inorganic 44 and organic uorinated species in sludge samples is not surprising and well reported by others previously. 5,6nsumer product samples Similarly, m-XRF map images were obtained for selected consumer products, as they were reported to emit substances of health and environmental concern. 45One sample was taken from a food contact material (Paper1), two textile samples were taken from used outdoor jackets (Textile1 and Textile2) and one sample was taken from a 'non-stick' labelled permanent baking sheet (Sheet1).Subsequently, all samples were investigated via uorine K-edge m-XANES spectroscopy (Fig. 2

and S4 †).
The uorine m-XRF map of the sample Paper1 reveals one signicant uorine containing spot contrasted by a very low background signal in the sample (Fig. 2 top and S10 †).For position P1 one distinct energetic F K-edge in the respective m-XANES spectrum and two maxima at 689 and 691.5 eV were detected that can be correlated with organically bound uorine compounds.The respective LC ts of the m-XANES spectrum correlate with a mixture of PTFE and 6:2-FTOH based on the given reference XANES spectra (see Fig. S24 and Table S1 †).To the best of our knowledge, the presence of inorganic uorides is unlikely and has not been reported in similar samples before.Overall, these results are in good agreement with previous studies on PFAS-treated food contact materials. 25,46he C, F and O m-XRF maps of Textile1 show a pattern correlating with the bre structure of the fabric (middle part of Fig. 2).Upon closer inspection of the count rate differences in the uorine m-XRF map, the PFAS surface impregnation appears to be not entirely homogenous (see also Fig. S11 †).This might be explained by inhomogeneous application of the coating, abrasion during use or numerous washing-cycles.The uorine m-XRF map displays three uorine spots of high uorine count rates (P1-P3; Fig. S11 †), whose m-XANES spectra are similar in appearance and exhibit a broadened maximum at 691 eV.All measured spectra agree well with reference data of organic uorine compounds and no inorganic uorides were identied.The respective LC t analysis suggests PFPrA and Na-TFMS to be comparable uorine species.Similar results were obtained for the sample Textile2, where two spots of high intensity (P1 and P2) and one of lower intensity (P3) were identied (Fig. S4 and S12 †).Both m-XANES spectra of P1 and P2 exhibit a maximum at 692 eV, which can be attributed to organic uorine compounds.Their LC t suggests a mixture of compounds structurally related to PFPrA, PTFE and 6:2-FTOH for P1.For P2 a combination of PFPrA and Na-TFMS was proposed as uorinated species (see also Fig. S24 and Table S1

†).
As expected for a purely uoropolymer-based sample, the C, F, O-m-XRF maps of Sheet1 show a different outcome.Compared to the previously discussed samples, the elemental m-XRF-map of Sheet1 is highly dominated by uorine atoms at the surface, indicating a dense and homogenous surface distribution of PFAS (Fig. 2 bottom and S13 †).This was further demonstrated by recordings of m-XANES point spectra at spots of high uorine count rate (see P1-P3), resulting in identical m-XANES spectra.Interestingly, the m-XANES spectrum recorded at P4, representing the lower end intensity within the uorine m-XRF map, yielded an identical spectrum (see also bottom part of Fig. 2).As expected, the data clearly prove the presence of a highly homogenous uoropolymer surface.
In comparison, the respective m-XRF maps of the surface of Paper1 and Textile1 yield a much more inhomogeneous distribution and much lesser spot concentration of uorinated organic compounds.This can be explained by the lack of a uoropolymer base material and the use of a PFAS coating based on the different water and grease repellency requirements of the product materials.Moreover, similar uorine distributions between textile and paper samples were found in previous investigations. 30As shown for Sheet1, the m-XRF method allows for a clear distinction between partly covered surface treatments and a material surface based on uoropolymer matrices.Altogether, the spot-by-spot ndings of all sample types correlate well with the recorded uorine bulk-XANES spectra, indicating the presence of differently scattered PFAS coatings.

Surface impregnated fabric samples
To further demonstrate the surface imaging possibilities of our analytical method, two different samples of the same PFAS-free textile (Fabric_blank) were manually spray-coated with two commercially available surface impregnation products (Spray1 and Spray2).The subsequent analysis via m-XRF mapping resulted in the recording of a signicant before-and-aer image of the investigated fabric sample (Fabric1 and Fabric2).The respective m-XRF maps and selected uorine m-XANES spectra are displayed in Fig. 3.
As depicted in Fig. 3 the recording of the m-XRF elemental map of the untreated fabric sample (Fabric_blank) resulted in an evenly distributed surface of carbon and oxygen atoms (see also Fig. S14 †).At this stage, the count rate of the uorine m-XRF map can be described as background noise with rates between 0 and 64 counts per second.The recorded uorine XANES spectra at P1 and P2 conrm the absence of uorinated compounds at the sample surface.Aer application of the rst, recently purchased impregnation product (Spray1), a different m-XRF map was recorded for Fabric1 (Fig. 3 mid and S15 †).The respective uorine m-XRF map of the sample clearly shows the presence of high amounts of uorinated species in areas on the sample surface.The selected spots P1-P3 on the m-XRF map result in uorine m-XANES spectra similar to those of other organic uorine compounds.Note that the XANES spectra at P4 only show a background signal, clearly revealing the uneven distribution of the spray-coating application at the sample surface.For comparison, an impregnation product from the mid 2010's was applied on the blank fabric sample (Spray2).The subsequently recorded m-XRF map for Fabric2 is depicted in Fig. 3 bottom (see also Fig. S16 †).The uorine m-XRF map shows a much more intense response compared to that of Fabric1, with uorine count levels ranging from 218 to 12 141 counts per second.In fact, all recorded uorine m-XANES spectra of selected high and low spots P1-P3 indicate the presence of organouorine compounds.This nding reveals that the impregnation product used for Fabric2 leads to a more even surface distribution of uorinated chemicals compared to the product used for Fabric1.This result might be based on the various chemical formulations used for both products which in turn are based on their production year and eld of application.More details on the PFAS composition are discussed in the target analytical section later.

m-XRF mapping sensitivity in correlation with analyte concentration
In our previous work, we approached the "detection limit" for uorine K-edge bulk-XANES spectroscopy at approx. 10 mg kg −1 dry weight based on the uorine amount by measuring reference samples of PFOS in sand of various concentrations. 32For the m-XANES spectroscopy sensitivity approach, we used four uoridefree sand samples spiked with PFOS methanol solutions at various concentration levels (100 000, 10 000, 1000, and 100 mg kg −1 dw).Therefore, all samples were overlaid with the respective PFAS solutions and subsequently le in a fume hood to evaporate.To disperse PFOS in the samples prior to measurement, the spiked sand samples were carefully homogenized using a mortar and pestle.To exclude background contamination, untreated sand samples were investigated for potential uoride blank values, prior to the measurements.Although the sensitivity of the bulk XANES method is identical, the m-XRF/m-XANES approach allows the identication of uorinated hot spots on the surface due to the implemented microfocus.In case of the m-XANES spectra we observed that the spectral quality based on the signalto-noise ratio correlates with decreasing PFOS concentrations in the samples.Nonetheless, the m-XRF elemental mapping approach allows for identication of the organouorine compound at all levels, even for the lowest concentrated sample (100 mg kg −1 dw).As depicted in Fig. S17, † the uorine m-XRF maps of all four samples provide detectable uorine-hotspots independent of their spiked concentration.All recorded m-XANES spectra clearly indicate the presence of PFOS at the selected points (see Fig. 4).As for the m-XRF maps, the shape of the m-XANES spectrum is independent of the initial PFOS concentration of the spiked samples.It should be noted that m-XANES cannot analyze lower concentrations than bulk XANES for ideal PFOS distribution in quartz.However, for environmental samples where PFAS hotspots are the common case, m-XANES shows an advancement.These ndings have to be considered as a sensitivity check and not be mistaken as the limit of detection, since F K-edge m-XANES spectroscopy cannot be regarded as a quantitative analysis method and standardized LOD protocols are not applicable to this method yet.

PFAS quantication of organic solvent extracts via LC-MS/MS analysis
To get a more detailed reference of the respective PFAS composition in every sample, target analytical investigations were performed.Therefore, LC-MS/MS spectrometry was applied to analyze methanol extracts of Soil1-2, SL1-3, Textile1-2, Paper1, and Sheet1 as well as Fabric1 and 2. The method included the identication of up to nine per-uoroalkylcarboxylic acids (PFCAs) and ve per-uoroalkylsulfonic acids (PFSAs) (see Tables S2-S5 †), respectively.All measured concentrations and further details can be reviewed in Table S6.† As depicted in Fig. 5, the target analysis revealed targeted PFASs for all types of samples.
Both soil samples showed the highest sum amounts of all investigated samples (1136.01 and 313.38 mg kg −1 dw).Targeted analysis identied long chain PFAS $C 8 such as per-uorododecyl acid (PFDA; 351.86 mg kg −1 dw for Soil1 and 240.01 mg kg −1 dw for Soil2) and peruorooctanesulfonic acid (PFOS; 579.96 mg kg −1 dw for Soil1) in both soil samples, whereas shorter chain length PFAS (C 4 -C 7 ) exhibited rather low concentrations.This observation can be explained by the lower hydrophilicity and mobility of the compounds $C 8 , leading to an enrichment in the topsoil. 47The sludge samples (SL1-3) showed sum PFAS concentrations between 20 and 104 mg kg −1 dw mainly composed of PFAS with chain length < C 8 .A recent study found PFCAs and PFSAs, especially with chain length C 5 to C 8 as the predominant compound class in sewage sludge samples, which ts well with our ndings. 48ower PFAS concentration could be observed for sample Paper1.Here, the total PFAS amount of 20.2 mg kg −1 dw can mainly be attributed to the sulfonic acids PFBS and PFOS.Peruorobutanoic acid was identied as the only PFCA compound in this sample, which correlates with previous studies. 30,49Unsurprisingly, slightly higher amounts of PFAS were identied in the selected fabric samples (Textile1-2), yielding 33.3 and 145 mg kg −1 dw as the sum of PFAS, respectively.The textile extracts were mainly composed of PFCAs with chain length between C 4 and C 8 (71 and 83%), whereas sulfonic acids were only identied in lower concentrations.Among others, these compounds were previously identied as surface coatings and side chain functionalized uoropolymers applied in waterproof apparel and medical fabrics. 50,51Similar observations were made for target analysis of manually applied surface spray coatings on samples Fabric1 and Fabric2.Both sample extracts showed elevated PFAS values of 108.16 and 61.31 mg kg −1 dw, respectively.Whereas Fabric2 treated with an older spray coating product exposed predominantly legacy PFAS ngerprints, target analysis of Fabric1 revealed high amounts of HFPO-DA (87.38 mg kg −1 dw), an industrial surrogate for the recently prohibited PFOA. 52The predominant verication of shorter chain PFCAs in coated textile extracts might correlate with the industrial shi from sulfonic acid use on textiles towards PFCAs and their respective precursors. 53Compared to all other analyses, the target analysis of sample Sheet1 exclusively gave a measurable level of PFBA as the detectable analyte.Again, this nding indicates an alteration in application towards short chain uorinated detergents in the production of peruoro polymer-based products such as permanent baking sheets.Altogether, the LC-MS/MS spectral analysis yielded detectable amounts of targeted PFASs with various compositions for all single product compartments.Specically, for samples with uorine-free backgrounds such as Fabric1 and Fabric2, the target analytical data can be utilized to interpret the recorded m-XANES spectra, making it suitable as a complementary analytical method.

Conclusions
We have shown that m-XRF mapping can be used to uniquely visualize hotspots and the spatial distribution of uorinated compounds on the surface of environmental and consumer product samples.For the rst time, the visualization of PFASs and other organic uorine compounds was successfully achieved aer being applied at the surface of a uorine-free sample.
To distinguish between inorganic uorides and uoroorganic compounds at specic surface locations, uorine K-edge m-XANES spectroscopy was applied for selected spots.The success of the analytical investigation is independent of the material composition, as proven by scanning various PFAS containing matrices.Fluorine K-edge m-XANES measurements require X-ray synchrotron radiation and a specialized setup only available in very few facilities.Due to their crystallinity, XANES spectra of inorganic uorides generally exhibit more characteristic features compared to those of the predominantly amorphous organic compounds. 37This fact complicates organouorine m-XANES analysis in samples with a high uoride salt concentration, such as soil and sewage sludge matrices.Nevertheless, we have shown that organouorinated compounds exhibit characteristic XANES spectral patterns depending on the chain length (short vs. long) and uorine content (peruorinated and partially uorinated), and they can thus be distinguished from inorganic uorides.Although the method does not allow specic analyte assignment, the linear combination of XANES spectra can be used in part to conrm the presumed composition of the sample.We showed that the quality of the calculated ts is strongly dependent on the spectral data available for LC-t analysis, and thus cannot be interpreted as a fully reliable theoretical backup yet.Manual pre-selection of suitable reference spectra based on a similar looking curve progression can signicantly increase the quality of the LC ts, and thus the  S6.† overall quality of the analysis.Since the method is not yet able to identify single types of analytes with detailed precision, a complementary analytical method is required.We completed our m-XANES data with state-of-the-art LC-MS/MS analysis methods to prove the presence of substantial amounts of PFASs in all samples.
Most signicantly, we have been able to demonstrate an unprecedented application of uorine m-XRF mapping and m-XANES spectroscopy that allows for precise surface analysis of uorinated organic matrices.Although still under development, this method should be regarded as an important contribution to the future evaluation of PFASs in surface coatings of consumer products or contaminated environmental samples.

Fig. 2
Fig. 2 m-XRF maps of samples Paper1 (top), Textile1 (middle) and Sheet1 (bottom) for fluorine (red), carbon (green) and oxygen (blue) and corresponding F K-edge m-XANES spectra; all m-XRF maps span 300 × 300 mm 2 , step size 10 mm, and the color scale is arbitrary; black spots in the m-XRF map are low in intensity; the excitation energy for all measured m-XANES spectra was 690 eV; spectra of manually selected references were added for comparison.PFA = perfluoroalkoxy polyalkanes; PTFE = polytetrafluoroethylene; Na-TFMS = sodium trifluormethyl sulfonate; PFPrA = perfluoropropanoic acid.

Fig. 4
Fig. 4 Normalized m-XRF map of fluorine (red), carbon (green) and oxygen (blue) (left) and corresponding F K-edge m-XANES spectra (right) of (a) 100 000 mg kg −1 based on the fluorine amount of PFOS in quartz sand, (b) 10 000 mg kg −1 F of PFOS in quartz sand, (c) 1000 mg kg −1 F of PFOS in quartz sand and (d) 100 mg kg −1 F of PFOS in quartz sand.The bulk-XANES spectrum of PFOS is displayed in black for better comparison.All m-XRF maps span 300 × 300 mm 2 , 10 mm step, and the color scale is arbitrary; black spots in the m-XRF map are low in intensity.

Fig. 5
Fig. 5 (a) Normalized percentage distribution of measured PFCAs for all investigated samples; (b) normalized percentage distribution of all measured PFSAs for all investigated samples; (c) cumulative sum amount of extracted targeted PFASs displayed in mg kg −1 dw of the respective samples (no conversion to fluorine equivalent concentrations).The displayed standard deviation error bars were calculated via variance addition of the single values shown in TableS6.†