An analysis on the fate of PFAS in still bottoms during electrochemical oxidation with a Ti 4 O 7 anode

Most previous studies on the degradation of per- and polyfluoroalkyl substances (PFASs) by destruction technologies, such as electrochemical oxidation (EO), focused on the transformation of target PFASs, while a more complete analysis is necessary to verify PFAS mineralization and understand the pathways. This work attempts to analyze the fate of PFASs during EO treatment of still bottoms (SB), a concentrated waste stream resulting from ion exchange resin (IXR) treatment of PFAS-contaminated groundwater. The EO treatment is performed with a Ti 4 O 7 anode in a closed reactor in an attempt to analyze all products, including (1) their distribution in solid, liquid, and gas phases, (2) the fate of fluorine by quantifying both fluoride and total organic fluorine (TOF), and (3) semi-quantification of nontarget PFAS by high-resolution mass spectrometry. The results reveal that PFAS accounted for only 33% TOF in the original SB, with target PFAS responsible for merely 34% TOF removal by EO treatment, following which target and nontarget PFAS account for almost all TOF. Data suggest that most unaccounted TOF in the original SB was likely due to undetected PFAS masked by microplastics. Additionally, fluoropolymers may also be responsible. Most fluorine released from PFAS during EO treatment was recovered from the anode and precipitate. The study highlights challenges in tracking the fate of fluorine in destructive treatment of PFAS, especially regarding capturing volatile products in the context of complex environmental samples, while it showcases the capabilities of EO treatment in breaking down target and nontarget PFAS and their precursors.

Per- and polyfluoroalkyl substances (PFASs) are a class of manufactured compounds, comprised of at least one perfluoroalkyl moiety, -C n F 2n or -CF 3 (Ritscher et al., 2018; Buck et al., 2011; Arora and Verma, 2026). Many PFASs are highly stable (O'Hagan, 2008) and both hydrophobic and lipophobic (Buck et al., 2012). The stability and versatility of PFASs have led to their wide use in industry, firefighting, and consumer products (Glüge et al., 2020; Arora and Verma, 2024). They are widely detected in soil, water, air, and human serum because of their extensive uses and high persistence in the environment (Wang et al., 2022a; Wang et al., 2023a; Kuo et al., 2023). Studies show that PFAS exposure is linked to various health risks, including cancer (Zahm et al., 2024; Verma et al., 2025a; Verma et al., 2025b). This has prompted concern among the public, actions from the regulatory authorities (EPA, 2023; Verma and Arora, 2025), and development and application of the treatment technologies (Blotevogel et al., 2023).

Several technologies aim to destroy PFASs, including electrochemical oxidation (EO), plasma, advanced reduction, hydrothermal alkaline treatment, and supercritical water oxidation (Wang et al., 2022a). EO has emerged as a favorable treatment technology for PFASs in water, partially because it does not require demanding operating conditions but can degrade PFASs at a relatively lower energy consumption (Lin et al., 2018; Wang Y. et al., 2020; Wang et al., 2022b). PFAS degradation by EO involves direct electron transfer (DET), converting PFAS on the anode surface to their free radical forms that are more amenable to subsequent oxidation processes facilitated by reactive oxygen species, such as hydroxyl free radicals (Lin et al., 2018; Shi et al., 2019), that are also formed on the anode surfaces. Porous Ti 4 O 7 has recently emerged as a promising anode material for EO applications in water/wastewater treatment because of such feature such as large porous surface area, ease of fabrication, and robustness in harsh conditions (Lin et al., 2018; Wang et al., 2022b; Wang et al., 2021; Liang et al., 2022), offering advantages over some other possible anode materials, such as boron-doped diamond (BDD) and Ti/TiO 2 -NTs/PbO 2 (Lin et al., 2018; Lin et al., 2012; Liang et al., 2020). EO treatment with Ti 4 O 7 has been demonstrated with great performance in treating PFAS-contaminated waters in both small-scale experiments and large-scale tests (Wang et al., 2021; Liang et al., 2022; Wang L. et al., 2020; Liang et al., 2018; Wang et al., 2023b).

However, PFASs are often present at low concentrations in large volumes of water, making it challenging for any treatment aiming to destroy PFAS. To address this challenge, EO has been coupled with regenerable ion exchange resin (IXR) procedures, which has been proved significantly more efficient than direct EO treatment (Wang et al., 2021; Liang et al., 2022). The PFAS-laden IXR is regenerated using organic solvent/brine solutions that are then distilled to reclaim the organic solvent. The residue waste solution, known as still bottoms (SB), contains high concentrations of PFAS and brine. This regenerable IXR approach is more effective at removing PFAS than traditional granular activated carbon (GAC) and is being increasingly implemented (Wisckol, 2021; McFarland, 2024; Kenoyer, 2024). There have been a few studies on the destruction of concentrated PFAS in both real and synthetic SB samples (Wang et al., 2021; Liang et al., 2022; Maldonado et al., 2021; Singh et al., 2020; McIntyre and Hart, 2023; Fuller et al., 2024). These studies reported PFAS destruction using EO, plasma, UV, and sonochemistry, but covered only limited scopes beyond target PFAS analysis. Although the total oxidizable precursor (TOP) assay was often included in these studies, none has systematically investigated PFAS transformation by EO treatment of real SB samples, focusing on the fate of fluorine using techniques such as total organic fluorine (TOF) and nontarget PFAS screening that are deemed critical to assessing a PFAS destruction process (Smith et al., 2024).

This study aimed to analyze the fate of PFASs in a real SB sample during EO treatment with a Ti 4 O 7 anode in three ways: (1) examining PFAS degradation products in solid, liquid, and gas phases by conducting the experiment in a closed reactor, (2) investigating fluorine distribution by quantifying both fluoride and total organic fluorine (TOF), and (3) including nontarget PFAS screening using high-resolution mass spectrometry (HRMS) along with target PFAS analysis. The results, and the lack thereof in some cases, underline the need for multi-phase PFAS monitoring and inform future PFAS mass-balance endeavors.

Chemicals

PFAS standards and isotope-labeled PFAS used as extracted internal standards (EIS) (≥98%) and non-extracted internal standards (NIS) (≥98%) were purchased from Wellington Laboratories (Ontario, Canada). Their nomenclature and correspondence are shown in Supplementary Table S1 in Supporting Information. Sodium hydroxide and ammonium acetate of the ACS reagent grade (≥97%) were purchased from Sigma-Aldrich (St. Louis, United States). Methanol (≥99.9%) and ethanol (≥99.9%) of the HPLC grade were purchased from Fisher Scientific (Pittsburgh, United States).

SB sample

An SB sample was obtained through regeneration of an IXR used for removing PFAS from groundwater at a fire training site of a military base, as described in previous studies (Wang et al., 2021; Liang et al., 2022). An IXR column was first loaded by pumping through contaminated groundwater of approximately 61,000 bed volumes (BVs), followed by elution using 2-BV 2% w/v NaCl solution and 5-BV solution consisting of 80% v/v methanol and 2% w/v NaCl. Methanol was afterward recovered by distillation, leaving behind a 3-BV SB sample. Compared with the 61,000-BV contaminated groundwater, the PFAS in the obtained SB sample was approximately 20,333× concentrated. As seen in Supplementary Table S2, PFOS and PFOA were the most prominent PFAS species in the resultant SB material, which also contained high concentrations of total organic carbon. Its high conductivity was partly supported by a high concentration of chloride.

EO experiment

The setup used in the EO experiment is an electrolytic cell (8.5 cm L × 7.5 cm W × 10.5 cm H) placed inside a closed chamber (19 cm Ø × 19 cm H). A rectangular Ti 4 O 7 ceramic plate (10 cm L × 5 cm H) was used as the anode. The characteristics of the Ti 4 O 7 electrode can be found in a previous report (Wang et al., 2021). Two 304 stainless steel plates of the same size were used as cathodes placed on each side of the Ti 4 O 7 anode at a 3-cm gap. Two beakers of trap solutions, one containing 100 mL EtOH and the other 100 mL 0.1 M NaOH, were placed inside the closed chamber to absorb possible airborne products. During the EO treatment experiment, a 100-mL SB sample was placed in the reactor and stirred at 700 rpm, with a current density of 20 mA/cm 2 supplied to the electrodes by an adjustable DC power source. The schematic is shown in Supplementary Figure S1. The power source for EO was shut down after 320 h, and the solution was continuously stirred until the foam collapsed and the solution reached near homogeneity. The volumes of the treated SB and two trap solutions were measured. Duplicate samples were taken from the SB and trap solutions for subsequent analysis.

Total oxidizable precursor assay

The TOP assay was based on the method developed by Houtz and Sedlak (2012), in which samples were exposed to hydroxyl radicals generated by thermolysis of persulfate under basic pH conditions. For each sample, an aliquot of 10 mL was used as a control, and another aliquot was amended with 60 mM potassium persulfate and 150 mM NaOH in 50 mL plastic centrifuge tubes. The tubes were placed in a water bath at 85 °C for 6 h to produce enough hydroxyl radicals to oxidize perfluoroalkyl acid (PFAA) precursors to perfluoroalkyl carboxylic acids (PFCAs). The samples were cooled to room temperature prior to subsequent analysis to determine PFCA concentrations.

Fluorine analysis

Fluorine analysis includes measurements of fluorine in the forms of fluoride ion and organofluorine. The fluoride concentration in solution was measured using a fluoride ion selective (ISE) electrode (Supplementary Figure S1), the procedures for which can be found in our previous study (Wang et al., 2022b). Potential interference on the ISE measurement from the SB matrix was dealt with by adding the total ionic strength adjustment buffer in an amount equal to the aliquot. The fluorine bound on the Ti 4 O 7 anode was recovered through thermal treatment and converted to fluoride ion form for measurement (Supplementary Figure S1), following the procedure described in a previous study (Wang et al., 2022b), in which the method was proven by the control with limited uncertainty. In brief, the Ti 4 O 7 anode was placed in a furnace that was heated from room temperature to 930 °C at a rate of 10 °C/min and was held at 930 °C for 30 min. N 2 gas feed to the furnace throughout the process was maintained at 3 L/min, with the exhaust gas passing through 100 mL 0.01 M NaOH solution for fluorine capture. Fluorine bound to the precipitates that were formed during EO treatment was also recovered and quantified using the same thermal treatment procedure. The organofluorine in the samples before and after EO treatment was measured by Eurofins Scientific (Lancaster, PA) in the form of total organic fluorine (TOF) by combustion ion chromatography (CIC), as shown in Supplementary Figure S1.

PFAS analysis

SB samples were diluted with DI water 10–100×, from which 0.5 mL was taken and mixed with 0.5 mL methanol containing internal standards for PFAS analysis. For the trap solutions, a 0.5-mL aliquot was directly sampled and mixed with 0.5 mL methanol containing the internal standards for PFAS analysis. The concentrations of target PFAS compounds (Supplementary Table S2 for a full list) were analyzed using a Waters ACQUITY UPLC system coupled with Xevo TQD tandem mass spectrometry (UPLC–MS/MS) (Milford, MA) and quantified using calibrations by the ratios between the target PFAS and that of its corresponding EIS (Supplementary Table S1), as described in our previous study (Wang et al., 2022b). Nontarget PFAS analysis was conducted using a Fourier-transform ion cyclotron resonance mass spectrometer (FTICR–MS), following the procedures that were described in our previous study (Wang et al., 2023c) (see Text S1 for more information). To analyze fluorine fate and distribution in nontarget PFAS, TOF contribution from nontarget PFAS was semi-quantified through the methodology explained in detail in Text S2, in which the calculation of nontarget PFAS TOF was demonstrated following the equation that was based on respective peak intensities, molecular weights, and total fluorine of all the target PFAS. Target and nontarget analysis are also illustrated in Supplementary Figure S1.

Microplastic analysis

The study identified microplastics using an optical photothermal IR (O-PTIR) spectroscopy (mIRage®-LS, Photothermal Spectroscopy Corp.), equipped with a Quantum Cascade Laser (QCL) and a photo-detector capable of spectroscopic detection ranges of 1000–3000 cm −1. Spectra were recorded in reflection mode with a 532 nm −1 probe laser. Experiments were conducted in a 0% relative humidity environment through continuous purge with N 2 gas, with the background recorded under the same conditions. The original SB sample was filtered onto a 0.4-micron glass fiber filter, 4–6 drops of Nile red solution were applied, and it was analyzed using fluorescence and IR. A laboratory blank through the same procedures was included. Five grids (1,026 × 1,026 micron each) were analyzed for the sample. Microplastics were identified on OpenSpecy through which the IR spectra were uploaded, processed, and matched against a comprehensive, crowdsourced reference library. A common literature threshold of 0.7 was applied for the match score to qualify positive identification (Clough et al., 2024).

EO treatment of still bottom in a closed reactor

The concentrations of target PFASs in the SB sample before and after EO treatment for 320 h are shown in Figure 1A (n = 2). The total concentration of the target PFASs decreased from 37.7 mg/L to 8.3 mg/L (calculated from the result shown in Figure 1A), accounting for a 78% reduction. The dominant species, PFOS, PFOA, and PFHxS, were reduced by 85%, 75%, and 75%, respectively, while the reductions of the short-chain PFHxA, PFPeA, and PFBA were 67%, 78%, and 72%, respectively. Assuming pseudo-first-order reactions based on the previous studies (Wang et al., 2021; Liang et al., 2022; Wang L. et al., 2020; Wang et al., 2023b), the rate constant for each target PFAS was calculated (Supplementary Table S3) and was found to be lower than that reported in a previous study involving EO treatment of the same SB solution under similar conditions but in an open reactor (Liang et al., 2018).

Intense foaming was observed during the EO, which also occurred in those previous studies and was credited for slower reaction rates. Therefore, slower reactions in this study may be related to different oxygen evolution behavior in the closed reactor. It is known that PFAS are mainly destroyed on the anode surface via direct electron transfer (DET) at an anodic potential higher than the oxygen evolution potential (Lin et al., 2018; Yang, 2020), although the cell voltages were not logged and anodic potentials were not measured in this study. In the closed reactor, it is safe to assume that oxygen partial pressure continued to increase during the EO reaction, as oxygen was produced by water oxidation on the Ti 4 O 7 surface. To test the closed condition on the oxygen partial pressure, dissolved oxygen (DO) in DI water after 30-min EO was used as a proxy. Applying the same EO conditions, DO exhibited an increase from 8.80 (±0.02) mg/L to 12.84 (±0.18) mg/L in the closed condition, compared with an increase from 8.80 (±0.02) mg/L to 9.37 (±0.08) mg/L in the open condition. According to Henry’s Law (Mackay and Shiu, 1981), the closed condition led to 36.9% higher oxygen partial pressure. The increase in oxygen partial pressure makes it harder for produced oxygen to leave the active sites on the anode surface due to equilibrium (Henry, 1803). This leaves fewer reactive sites on the Ti 4 O 7 surface available for PFAS reaction and thus slows down their degradation.

Note also that the longer-chain PFAS generally exhibited greater reaction rate constants than shorter-chain ones, which is consistent with the results of the SB study in an open reactor (Liang et al., 2018) and that with laboratory-spiked solutions (Wang Y. et al., 2020). Partial degradation of precursors to form PFAAs may also contribute to lower reaction rate constants of shorter-chain PFAAs. This possibility was explored in the following TOP assay. The potential impact of methanol on SB in PFAS transformation, byproduct formation, and analytical recoveries during EO is also worth mentioning. A previous study reported a 31.4% decrease in PFOA degradation with 1 M methanol in the solution (Zeidabadi et al., 2024). The moderate effect of the high methanol concentration was partly explained by the fact that the continuous generation of hydroxyl free radicals during EO contributed to PFAS degradation.

Another study on EO found that 100 mM methanol modestly suppressed the formation of ClO 3 − and ClO 4 − due to the methanol quenching effect on hydroxyl free radicals (Wang L. et al., 2022). Another study involving the same SB batch had to spike methanol to 1 M to limit the formation of chlorinated byproducts (Liang et al., 2022). Although ClO 3 − and ClO 4 − were not measured, the methanol content in this study is approximately 100 mg/L (Wang et al., 2021), far below 1 M, above which the impact on EO was deemed significant (Liang et al., 2022).

The target PFAS amounted to 37.7 mg/L in the original SB sample, comprising 24.6 mg/L TOF. As shown in Figure 1B (n = 2), it contributed to 29% of the TOF measured in the SB (84.5 mg/L) prior to the EO treatment. Nontarget PFAS were identified with Confidence Level 3 (i.e., with which spectral evidence is sufficient to propose a, but not to ascertain the, structure (Schymanski et al., 2014)), shown in Supplementary Table S4. The contribution to TOF from nontarget PFAS was approximately 4% in the original SB sample (Figure 1B). There was 67% TOF unaccounted for in the original SB sample. Target PFAS accounted for 88% of the organofluorine from PFAS in the original SB sample, while the remaining 12% was from nontarget PFAS (Hour 0, Figure 1B).

After EO treatment, 84.5 mg/L TOF exhibited a reduction of 55.5 mg/L (or 66%) to 29 mg/L (Figure 1B). The total concentration of the target PFAS decreased from 37.7 mg/L to 8.3 mg/L (calculated from the result shown in Figure 1A), a removal of 29.4 mg/L. The reduction of TOF contributed from the removal of the target PFAS is 19.2 mg/L (Figure 1C, n = 2). No fluoride was detected in the treated SB, but 13.8 mg/L fluoride was recovered from the Ti 4 O 7 anode through thermal treatment, as described in Section 2.5. A previous study using the same thermal treatment of the Ti 4 O 7 anode confirmed more than 99% fluorine recovery from the NaOH solution using spiked PFOA and less than 1% fluorine carryover in the blank sample (Wang et al., 2022b). Assuming full recovery, 13.8 mg/L fluoride is equivalent to 72% of the 19.2 mg/L removed TOF from target PFAS (Figure 1C); however, it cannot be completely ruled out that some fluorine was lost during the thermal treatment recovery. It is most likely that the organofluorine in the SB (including PFAS) was transformed to the form of a Ti-F bond on the Ti 4 O 7 anode during the process of defluorination facilitated by the EO reaction (Wang et al., 2022b) because of the strong affinity between titanium and fluorine (Dean, 1999).

Our earlier study indicated that nearly all fluorine released from PFOS degradation by EO with a Ti 4 O 7 anode was bound by a Ti-F bond on the Ti 4 O 7 anode surface (Wang et al., 2022b). The nickel-coated steel clips used for electric contactors in the EO reactor were eroded during EO due to capillary wicking, by which the water moved upward through the capillary-like micro-pores of the partially submerged porous Ti 4 O 7 electrode and reached and corroded the clips, resulting in the formation of red precipitate in the solution, which was speculated to be Fe(OH) 3 based on its red appearance. It should be noted that such precipitate is not a feature of the EO treatment and can be conceivably avoided with contactors having better corrosion resistance. The precipitate went through the same thermal treatment described in Section 2.5 immediately after recovery to prevent any potential loss, and 3.3 mg/L fluoride was recovered, which was equivalent to 17% of the 19.2 mg/L removed TOF from target PFAS (Figure 1C). Given that no fluorine was used for the fabrication of the Ti 4 O 7 anode (Lin et al., 2018), it is assumed that the recovered fluoride solely resulted from defluorination fluorine that was first transformed to fluoride form in aqueous phase and then adsorbed to the Fe(OH) 3 precipitate, as Fe(OH) 3 has been shown to adsorb fluoride strongly by charge (Kumar et al., 2009). Therefore, the TOF removed from the SB by EO (55.5 mg/L) was partially transformed as fluorine bound to Ti on the anode and that bound to the precipitate, while there was still approximately 66% TOF reduction unaccounted for (Figure 1C). The explored aspects of TOF mass balance during EO are further illustrated in a Sankey diagram (Figure 1D).

No fluoride was detected in the NaOH trap solution placed in the closed chamber during EO treatment, and none of the target PFAS were detected by the UPLC-MS/MS. Both target and nontarget PFAS were detected in the EtOH trap solution (Supplementary Table S6) by FTICR–MS but not by the UPLC-MS/MS target analyses. It was calculated that target and nontarget PFAS accounted for, respectively, 42% and 58% of TOF in the EtOH trap solution. By assuming that the concentration of each FTICR–MS-detected target PFAS was no greater than the UPLC–MS/MS limit of detection (LOD) (Supplementary Table S2), all the PFAS contributed 33.3 mg/L to TOF at maximum, which accounted for 0.06% of the 55.5 mg/L removed TOF. Given that there are volatile PFAS capable of evading the EtOH trap (Droz et al., 2025), the airborne PFAS could have contributed significantly to the removal of TOC during the EO treatment using Ti 4 O 7 and thus reshaping the distribution shown in Figure 1C. Future studies should consider qualifying such distributions by validating trap performance with the organic solvent, for example, through multi-phase trapping, for volatile PFAS.

The remaining 8.3 mg/L target PFAS in the SB after treatment comprised 5.4 mg/L TOF, accounting for approximately 19% of the total 29 mg/L remaining TOF in the treated SB sample, based on the UPLC-MS/MS analysis (the left bar at Hour 320, Figure 1B), leaving 81% of the remaining TOF unaccounted for. However, based on the FTICR–MS analysis, target PFAS accounted for 12% of the TOF from the initial target PFAS, while nontarget PFAS were responsible for 88% in the post-EO treatment SB sample. The result points to the nontarget PFAS as the source of unaccounted TOF in the treated SB (the right bar at Hour 320, Figure 1B). The dominant species among the nontarget PFAS in the treated SB (Supplementary Table S7) include C 4 HF 9 O 2 S (e.g., nonafluorobutane-1-sulfinic acid-related), C 2 F 6 O 2 S (e.g., pentafluoroethane-1-sulfonyl fluoride-related), C 9 H 6 F 12 O 2 (e.g., ethyl 7H-perfluoroheptanoate-related), C 7 H 5 F 9 O 2 [e.g., 3-(perfluorobutyl)propanoic acid)-related], and C 8 H 5 F 13 (e.g., perfluorohexylethane-related), all of which are components in fabricating polymers (Farnham et al., 1997; Ie et al., 2012; Ma et al., 2013; Sakoh et al., 2021; Suzuki et al., 2021; Shahinpoor et al., 2001) found in consumer products such as paints and medicine.

To investigate the possible presence of fluorine-containing microplastics, the original SB sample was filtered through a 0.22-μm syringe filter and measured for TOF. The result showed less than 5% TOF difference from the unfiltered SB sample, which only ruled out the possibility of major TOF contribution from micro-scale fluorine-containing particles. However, nanoscale particles are still a possible source of TOF. The unaccounted TOF source was further explored by microplastic screening by O-PTIR. The screening identified 113 particles of 15 microplastic types in five scanned grids in the original SB sample (Supplementary Table S5), with background considered. Although all the identified microplastics are larger than 1 micron (>50% below 10 microns, not shown), a size well within the resolution limits of O-PTIR, their prevalence posits the likely presence of sub-micron particles, for example, nanoplastics.

Based on these results, two probable sources of the unaccounted 67% TOF in the original SB sample (Figure 1B) emerge. One is the ionic PFAS “masked” by nanoparticles, such as nanoplastics, that is, PFAS that were adsorbed and absorbed to nanoplastics and evaded detection by either LC-MS/MS or FTIR–MS. Past studies suggest that such adsorption and absorption can occur on both the outer and inner surfaces of polymeric nanoparticles (Cortés-Arriagada, 2021). The amount of PFAS involved is often considerably large. A study showed PFAS concentrations adsorbed on MPs in a lake were hundreds of times higher than the lake background levels (Scott et al., 2021). Another study on how PFAS attached to polyethylene microplastics of different sizes found that smaller particles had a higher PFAS absorption capability, which showed less restrained intraparticle diffusion in smaller particles due to a greater surface area and a thinner boundary layer (Cormier et al., 2022). Such a finding further boosts the likelihood of significant undetected PFAS trapped in the nanoparticles, such as nanoplastics.

Even though there was no fluoropolymer among the microplastics identified by O-PTIR, the existence of nano-sized fluoropolymer cannot be effectively ruled out. Their likely presence can be argued based on the emergent nontarget PFAS after the EO treatment, which were not there before, as shown in Supplementary Table S4, in which identified PFAS on FTICR-MS spectrum in the original SB solution are listed. This is arguably attributed to their release from polymers through electrochemical depolymerizations during the EO treatment (Weber and Ramasamy, 2020). As for the unaccounted TOF, fluoropolymers are likely a major source, given the fact that most PFAS in SB originated from the aqueous film-forming foam (AFFF) historically used for fire training at a military base.

As indicated above by the TOF result between the filtered and the nonfiltered SB samples, the preexisting fluoropolymers are likely of nanoscale sizes. Few studies have examined the particle size of fluoropolymers, but, in the dispersed aqueous phase, it is believed to range from 10 1 to 10 2 μm (Jing et al., 2020). A fluoropolymer is a key component of alcohol-resistant AFFF (AR-AFFF), which forms a gelatinous polymeric layer that blocks oxygen supply and helps suppress fires (Malik et al., 2024). Fluorinated AFFF, including AR-AFFF, such as that manufactured by ChemGuard, has been used by all branches of the U.S. military for firefighting and related training since the 1960s (Moody and Field, 2000; LeBlanc, 2025). Components in AFFF formulations have been shown to transport in the vadose zone and groundwater after application to the ground surface (Ruyle et al., 2023; Backe et al., 2013).

There are other sources of fluoropolymers. For example, they have been found in sludge and biosolids from wastewater treatment plants (WWTPs) (Fredriksson et al., 2020; Letcher et al., 2020). Fluoropolymers do not ionize well in conventional MS-containing workflows, such as LC-MS/MS and FTICR–MS, and thus could not be identified or quantified in this work. However, it has been shown that they can be captured in TOF analysis by CIC (Hua et al., 2025), and it is thus likely that fluoropolymers may be a major source of unaccounted TOF in the original SB sample (i.e., 67% TOF, Hour 0, Figure 1B). Another possible source of the unaccounted TOF is fluorinated pharmaceuticals, in which fluorine provides unique properties that improve drug efficacy (Purser et al., 2008; Gillis et al., 2015). Fluorine-containing drugs account for approximately a 20% share in the drug market (Inoue et al., 2020). Over 300 fluorinated medicines have been approved for use by 2020 (Heravi and Zadsirjan, 2020), and they are reported to contribute significantly to unaccounted TOF in WWTP sludge and effluent (Aro et al., 2021; Spaan et al., 2023). Similar to what might have happened to fluoropolymers, these fluorinated pharmaceuticals may also be picked up by IXR from groundwater when WWTP effluent is returned to the water cycle and released into waterways. Additionally, chemicals in biosolids used for land application can leach from soil and reach groundwater.

EO can effectively break down fluorine-containing pharmaceuticals, as shown in a recent study by Liu et al. (2025). Based on the treated SB results from FTICR–MS (the right bar at Hour 320, Figure 1B), it is speculated that (1) nanoparticles, for example, nanoplastics, were a likely sink of a significant amount of PFAS, (2) EO treatment was responsible for the release of most detected nontarget PFAS (the right bar at 320 H, Figure 1B) from the fluoropolymers and other fluorine-containing nano particles and pharmaceuticals, which evaded FTICR–MS screening prior to EO treatment (Hour 0, Figure 1B); (3) nontarget PFAS release from polymers and their subsequent defluorination lagged behind the defluorination of those target PFAS because the depolymerization rate constants by EO were orders of magnitude smaller than the PFAS degradation rate constants by EO (Wang Y. et al., 2020; Bawareth et al., 2018a; Bawareth et al., 2018b). Some of the detected prominent nontarget PFAS species, such as C 2 F 6 O 2 S, C 9 H 6 F 12 O 2, and C 8 H 5 F 13 (Supplementary Table S7), were less polarizable than the target PFAS and thus less amendable to EO degradation (Wang Y. et al., 2020); (4) the recovered fluorine from the anode, precipitate and air accounts for 89% recovery of the 19.2 mg/L TOF removal assumedly originated from the target PFAS defluorination (Figure 1C).

Several factors could contribute to the 36.5 mg/L of TOF that was not accounted for (equivalent to 66%, Figure 1C). Some PFAS may simply be absorbed to the dissolved organic matter due to hydrophobic and electrostatic interactions and were excluded during sample preparation (i.e., filtration) (Mejia-Avendaño et al., 2020). The TOF contribution from fluorinated pharmaceuticals was unknown but potentially significant (Aro et al., 2021; Spaan et al., 2023). The thermal fluorine recovery method used in this study may not be adequate to fully recover fluorine bound on the anode and/or precipitate. In addition, the EtOH trap may not be sufficient to capture some volatile fluorinated compounds (e.g., ultrashort-chain PFAS) that may be formed during EO treatment. This factor may be improved by using a less polar absorption solvent, such as hexane (Reichardt, 1965). Cryotrapping, that is, a cold trap, could also have been included for volatile compound retention (Werkhoff and Bretschneider, 1987).

The importance of capturing airborne PFAS during treatment has been demonstrated in recent studies. One study reported PFOA and PFOS loss in biochar through evaporation with steam generated during pyrolysis, although the operation temperature was 300–700 °C, much higher than the putative operation temperature in the closed EO reactor (Kim et al., 2015). Another study reported that a portion of tested PFCAs, including those involved in this study, such as PFBA, PFBS, PFOA, and PFOS, transformed into volatile hydroperfluoroalkanes during hydrothermal liquefaction operated at 250 °C–350 °C and pressures of 10–25 MPa (Wen et al., 2024), conditions that are far more conducive to such reactions than those used in this study. Notwithstanding that study’s finding that 60%–90% organic fluorine in PFOA went through volatile transformation, it highlighted the importance of capturing volatile products from PFAS treatment under any heated and pressurized conditions.

The analysis of airborne fluorine is not complete without headspace sampling. This is partly because some studies showed that even gaseous adsorption techniques designed for gaseous fluorine often fail to capture the most volatile organic fluorine fraction, such as hydrofluorocarbons (Young et al., 2021; Lin et al., 2022; Ye et al., 2024). The challenge highlights the potential exposure risk posed by highly evasive and volatile PFAS. In addition to volatile PFAS, it could have been helpful to include more thorough screening of non-PFAS organic compounds. Only four well-known ones were checked on the FTICR-MS spectrum of the EtOH trap in this work, but none were found. They were fluoromethane (CH 3 F), difluoromethane (CH 2 F 2), trifluoromethane (CHF 3), and 1,1,1,2-tetrafluoroethane (C 2 H 2 F 4). Headspace samples targeting both PFAS and non-PFAS compounds should be included in any future effort to close fluorine mass balance during PFAS treatment. Analytical methods targeting total gaseous fluorine should be considered, such as one that relies on thermalizing air or special gas to decompose organic fluorine to release F atoms that react to form F − (Ye et al., 2024; Furlani et al., 2023). Regarding testing the hypothesis related to nanoparticles, future efforts to be considered can include (1) measuring TOF of the reject on the 0.22-µm filter, (2) isolating dissolved organofluorine from colloidal material via ultrafiltration, and (3) testing the recovery of TOF analysis from representative fluoropolymers such as PTFE. The efforts aiming to narrow the gap in fluorine budget will be crucial to assess EO mineralization efficiency in the future.

TOP assay

The results of the TOP assay for the SB before and after 320-h EO treatment are shown in Figure 2. Prior to EO treatment, there were a total of 10.5 mg/L PFAA precursors, most of which transformed to PFHxA, PFBA, PFPeA, and PFOS (Figure 2). After the TOP treatment, they contributed a 28.0% increase to the total target PFAS concentration, equivalent to 7.7 mg/L or a 31.3% increase in fluorine. The 28% precursor-to-target PFAS equivalence in SB suggests complexity of the precursors, but not necessarily anything definitive about the precursor structure or sources (Chiang, 2019). On the one hand, for instance, studies have found that linear PFCAs can be converted through a TOP assay from the precursors produced during both electrochemical fluorination (ECF) and fluorotelomerization processes (Dauchy, 2025). On the other hand, PFSAs after the TOP assay can only come from precursors that contain sulfonic groups, which are used only during the ECF process, not fluorotelomerization (Cioni et al., 2023; Cioni et al., 2022). Given that target PFAS accounted for only 29% of the total TOF in the SB prior to EO treatment, the TOF would have been substantially underestimated if solely based on a TOP assay (i.e., 24.6 + 7.7 mg/L), versus 86 mg/L measured by TOF analysis, which is consistent with the previous report (Schaefer et al., 2018). Note that the 7.7 mg/L TOF increase was only a fraction of the 58 mg/L unaccounted TOF in the original SB sample (Figure 1B), accounting for approximately 13.3%. This suggests that the vast majority of unaccounted TOF was not amenable to oxidative conversion during the TOP assay. These results underscore insignificant TOF contribution from PFAS, both target and precursors, and the limitations and pitfalls of applying a TOP assay when it comes to complex samples such as SB. The modest presence of the PFAA precursors suggests that the unaccounted TOF (Hour 0, Figure 1B) mainly comes from other organofluorine compounds, such as fluorinated polymers, as discussed in the previous section. The TOP assay shows that most PFAA generated in samples were C6 and shorter, indicating minor presence of long-chain PFAA precursors. This is consistent with previous reports on the presence of precursors in AFFF-impacted water samples (Houtz et al., 2013; Houtz et al., 2016; Ruyle et al., 2020). After EO treatment, PFAA precursor concentrations decreased from 10.5 mg/L to 0.4 mg/L, all of which were transformed to PFOS (Figure 2). The PFAA precursors accounted for a 4.7% increase in total PFAS concentration and a 4.6% increase in TOF. Compared to those prior to EO treatment, which are respectively 28.0% and 31.3%, EO is demonstrated to degrade a wide range of PFAS, including PFAA precursors.

This study attempted a systematic examination of the fate of PFAS in a field SB sample during EO treatment. The results suggest that the sample contained nanoscale PFAS-containing polymers, including fluoropolymers, that led to underestimation of PFAS and unaccounted TOF. After EO treatment of SB, more nontarget PFAS emerged, with moieties often used to fabricate polymers. This is believed to be due to EO-induced depolymerization. The nontarget PFAS contributed a significant amount of TOF in the treated SB. It is thus postulated that small-size PFAS-containing polymers were responsible for most of the unaccounted TOF in the original SB. Future research with means and resources can test whether nanoplastics play a role in PFAS-masking or contribute directly to TOF.

PFAA precursors had a moderate presence in the field SB sample, equivalent to approximately 28% of the total target PFAS concentration, according to a TOP assay. Their presence was vastly reduced, with the remainder accounting for less than 5% of the remaining PFAS after EO treatment. This indicates the capability of EO in treating nontarget PFAS, including PFAA precursors. Assuming that defluorination of target PFAS was mostly responsible for the fluorine recovered after EO treatment in this experiment, the study verified that Ti 4 O 7 -based EO treatment can effectively degrade PFAS by breaking down the C-F bond in PFAS. Most of the fluorine from degraded PFAS ended up in the solid phase, either accumulating on the Ti 4 O 7 anode or forming precipitates. Volatiles captured in the EtOH trap solution were found to be negligible, indicating minimal airborne release under the experimental conditions or production of highly volatile and elusive PFAS, which warrants further investigation to confirm. These findings provide critical insight into the intricate composition of SB, and, by extension, the broader challenges associated with PFAS contamination in groundwater. This study highlights the importance of understanding PFAS transformation pathways to develop more effective remediation strategies.