Concentrations of perfluoroalkyl and polyfluoroalkyl substances before and after full-scale landfill leachate treatment

Abstract

Many consumer and industrial products, industrial wastes and dewatered sludge from municipal wastewater treatment plants containing per- and polyfluoroalkyl substances (PFAS) are disposed of in landfills at the end of their usage, with PFAS in these products leached into landfill leachates. On-site leachate treatment is one possible method to reduce PFAS in leachates. Many landfills are equipped with on-site leachate treatment systems, but few full-scale facilities have been systematically evaluated for PFAS concentration changes. The objective of this study was to evaluate a cross-section of full-scale on-site landfill treatment systems to measure changes in PFAS concentrations. Leachate samples were collected before and after treatment from 15 facilities and were evaluated for 26 PFAS, including 11 perfluoroalkyl carboxylic acids (PFCAs), 7 perfluoroalkyl sulfonic acids (PFSAs), and 8 perfluoroalkyl acid precursors (PFAA-precursors). Transformation of precursors was evaluated by the total oxidizable precursor (TOP) assay. Results showed no obvious reductions in total measured PFAS (Σ₂₆PFAS) for on-site treatment systems including ponds, aeration tanks, powdered activated carbon (PAC), and sand filtration. Among evaluated on-site treatment systems, only systems fitted with reverse osmosis (RO) showed significant reductions (98–99 %) of Σ₂₆PFAS in the permeate. Results from the TOP assay showed that untargeted PFAA-precursors converted into targeted short-chain PFCAs increasing Σ₂₆PFAS in oxidized samples by 30 %, on average. Overall, results of this study confirm the efficacy of RO systems and suggest the presence of additional precursors beyond those measured in this study.

1. Introduction

Per- and polyfluoroalkyl substances (PFAS) are widely distributed in the environment and humans (Ahmadireskety et al., 2021, 2022; Awad et al., 2020; Kotlarz et al., 2020; Liu et al. 2022). Because of the suspected links to adverse human health (Calafat et al., 2007; Schultes et al., 2018), PFAS removal from the environment has become a priority. For example, during 2016, the US Environmental Protection Agency (US EPA) has issued a 70 ng/L (equivalent to 0.13–0.14 nmol/L) drinking water health advisory for the sum of perfluorooctanoic acid and perfluorooctanesulfonic acid (ΣPFOA + PFOS), two widely studied PFAS (Hamid et al., 2018; US EPA, 2016). During June 2022, US EPA has since changed the health advisory to 0.004 ng/L and 0.02 ng/L for PFOA and PFOS, respectively, through new estimates of lifetime health effects from drinking water.

Landfill leachates are not original sources of PFAS but have been found to contain high concentrations of PFAS (Benskin et al., 2012; Hamid et al., 2018; Lang et al., 2017; Solo-Gabriele et al., 2020) due to their release from many PFAS-containing consumer products, industrial wastes and municipal wastewater treatment plant (WWTP) sludges disposed of in municipal solid waste landfills (Chen et al., 2020; Favreau et al., 2017; Lang et al., 2016; Schaider et al., 2017; Schultes et al., 2018; US EPA, 2018; Wang et al., 2017). Recognizing that leachates from landfills contain PFAS (Lang et al., 2017; Liu et al., 2020b; Solo-Gabriele et al., 2020), efforts are needed to manage leachate to control for potential PFAS releases, and thus to avoid impacts of PFAS to adjacent natural water bodies or off-site wastewater treatment plants (WWTP) (Masoner et al., 2020). One option is to use on-site leachate treatment processes.

There is a growing body of research related to PFAS treatment in aqueous systems, but most of this work focuses on treating contaminated groundwater (Liu et al., 2020a; Xu et al., 2021; Yan et al., 2015), domestic wastewater (Arvaniti & Stasinakis, 2015; Bao et al., 2014; Tang et al., 2006), or drinking water (Boone et al., 2019; Rahman et al., 2014). For PFAS in leachate treatment, some work is available for specific technologies, like electrochemical oxidation (Pierpaoli et al., 2021) and plasma reactors (Singh et al., 2021), evaluated at the laboratory scale, while some other work described the performance of full-scale leachate operations, especially for activated carbon and membrane filtration (Busch et al., 2010; Yan et al., 2015). However, no studies have evaluated the mainstream leachate treatment systems comprehensively based on their potential ranges of complexity. The objective of the current research was to fill this data gap by evaluating a wide cross-section of full-scale leachate management systems on their ability to reduce PFAS. Our study is unique in that concentrations of 26 PFAS were evaluated from 15 active full-scale leachate treatment operations. Treatment systems evaluated ranged from traditional pond systems to complex treatment operations that included membrane filtration processes. Measurements among the treatment systems evaluated used the same analytical approach (targeted PFAS analysis by liquid chromatography-tandem mass spectrometry (LC-MS/MS)). Such consistent analytical methods allowed for direct comparisons between the treatment systems evaluated, comparisons which are not available through other studies. In addition, collected samples were also analyzed by total oxidizable precursor (TOP) assay which was used evaluate the potential contribution of precursors beyond the 26 targeted compounds (Houtz & Sedlak, 2012; Martin et al., 2019; McDonough et al., 2019). As part of this study, we hypothesized that PFAS transformation may occur during treatment, specifically that precursors would be transformed into terminal species. The TOP assay included in this study provided the unique opportunity to assess the possible contribution of non-targeted precursors towards the contribution of terminal PFAS measured.

2. Materials and methods

2.1. Landfill treatment systems

On-site leachate treatment systems at 15 landfill facilities (called T1 to T15) located in Florida, US were evaluated (additional details in Figure S1 and Table S1). These facilities were among a total of 39 landfills that had treatment operations for which physical chemical parameters were also evaluated (Zhang et al., 2022). Four facilities (T1, T3, T5 and T6) were sampled twice. Three facilities had two or more kinds of treatment (T4, T7, and T10). In total, pond systems were measured thirteen times (n_s = 13) at nine facilities (n_f = 9), aeration tank systems were measured five times at three facilities (n_s = 5, n_f = 3), one powdered activated carbon (PAC) system consisting of three tanks in parallel was evaluated from one facility (n_s = 3, n_f = 1), two filtration systems were evaluated from two facilities (n_s = 2, n_f = 2), single-stage RO systems were measured three times at one facility (n_s = 3, n_f = 1), and one combined system was evaluated from one facility (n_s = 1, n_f = 1). This combined treatment used RO after several pretreatment steps including coagulation, sedimentation, sludge thickening, microfiltration, pH adjustment and cartridge filtration.

2.2. Sampling methods

Schematic diagrams (Figure S1) for each treatment facility are shown in the supplement based on information provided by landfill operators. Upon receiving permission from these landfill operators to collect leachates, to estimate the change of PFAS concentrations from different leachate treatment systems, grab samples were collected before and after each treatment facility to represent their influents and effluents (see circles shown in Figure S1). In addition, grab samples at intermediate points were also collected when they were accessible throughout the treatment process to track the change of PFAS concentrations.

At each sampling location, leachate was collected into a high-density polyethylene (HDPE) primary collection bottle (one liter or greater). The primary collection bottle was then used to split the leachates among four sample analysis bottles which consisted of two 250 mL HDPE bottles for PFAS analysis and two 125 mL HDPE bottles for TOP assay. In addition to the four PFAS sample analysis bottles collected per sampling location, quality control (QC) samples, including trip blanks, field blanks, and sample duplicates, were also prepared and collected. All collected samples were not filtered prior to PFAS analysis and TOP assay, and so they were analyzed as whole samples. Additional details about sample collection are provided in the supplemental text section 3.

2.3. PFAS analyses and TOP assay

PFAS can be separated into two broad categories, perfluoroalkyl acids (PFAAs) and PFAA-precursors. Compared with PFAAs, PFAA-precursors are generally less stable and degrade into related PFAAs depending upon environmental conditions (Benskin et al., 2012; Lang et al., 2017; Wang et al., 2020). Based on the functional groups in their chemical structures, two main classes within the PFAAs include the perfluoroalkyl carboxylic acids (PFCAs) and perfluoroalkyl sulfonic acids (PFSAs) (Buck et al., 2011; Lindstrom et al., 2011; Kwiatkowski et al., 2020). PFOA and PFOS, the most widely studied PFAS, are members of PFCAs and PFSAs classes, respectively.

The 26 PFAS evaluated in the current study (Σ₂₆PFAS) consisted specifically of 11 PFCAs (C₄-C₁₄, Σ₁₁PFCAs), 7 PFSAs (C₄-C₁₀, Σ₇PFSAs), and 8 precursors (Σ₈PFAA-precursors). Definitions of acronyms and chemical structures of the 26 PFAS analyzed are given in Table S2. PFAAs, including PFCAs and PFSAs, are often further classified into short-chain PFCAs (C₄-C₇, Σ₄short-chain PFCAs), long-chain PFCAs (C₈-C₁₄, Σ₇long-chain PFCAs), short-chain PFSAs (C₄-C₅, Σ₂short-chain PFSAs), and long-chain PFSAs (C₆-C₁₀ PFSAs, Σ₅long-chain PFSAs) based on the number of carbon atoms connected with fluorine in their chemical structures. Analyzed Σ₈PFAA-precursors consisted of fluorotelomer carboxylic acids (FTCAs, including 5:3 and 7:3 FTCA), perfluorooctane sulfonamides (FOSAs, including PFOSA, NMeFOSAA, and NEtFOSAA), and fluorotelomer sulfonic acids (FTSAs, including 4:2 FTS, 6:2 FTS, and 8:2 FTS).

These PFAS were analyzed in each sample by isotope dilution method using LC-MS/MS. Leachate samples were processed using solid phase extraction (SPE) followed by evaporation to dryness with nitrogen gas and reconstitution with methanol. This method and related quality control (QC) analysis complied with the DoD QSM criteria (DOD and DOE 2019). Details about sample analysis are provided in the supplemental text section 5.

To estimate the impact from untargeted PFAA-precursors in this study, the TOP assay was also used to oxidize PFAA-precursors (both among the eight measured in this study plus additional precursor compounds that were not measured) to targeted PFAAs (Houtz & Sedlak, 2012; Martin et al., 2019; McDonough et al., 2019) by strong oxidant potassium persulfate and heat (6-hour at 85 °C). In total, 22 samples were analyzed for TOP assay, including: (a) eight influent and eight effluent samples from aeration tank processes, filtration with pretreatment, and combined treatment; (b) one intermediate sample from one of the aeration processes (T15); and (c) five samples of permeates and concentrates collected from the first and the second stage of a two-stage RO system. The analysis of leachates for TOP was similar for PFAS as described above except that a split of the samples was subjected to the TOP oxidation step which was completed before the SPE process. Five oxidized laboratory procedure blanks and three oxidized laboratory control samples were also analyzed as part of QC analysis for the TOP assay. Additional details about the TOP assay and QC analysis are provided in the supplemental text sections 6 and 7.

2.4. Data analysis

For the targeted PFAS, 23 % were below detection limits. The remaining 77 % of the measurements were above detection. Details of the values below detection for each species are shown in Table S4. Bar plots show detected PFAS only and thus non-detected values were ignored in these plots. For statistical analysis comparing before and after treatment, the non-detected values were replaced with the $\mathrm{L}\mathrm{O}\mathrm{D}/\sqrt{2}$ as recommended by Verbovšek (2011). To obtain an estimate for the influent concentration for the pond systems, PFAS concentrations were estimated by the weighted average of the PFAS results for the leachate as collected at the base of the landfill cell that contributed towards the pond (with additional details in the supplemental text section 8). For a comparison between original and oxidized samples, when over 50 % of the results were below the LOQ (out of 22 samples), the p-values of related PFAS were reported as not applicable (NA) due to lack of quantifiable concentrations of PFAS. Additional details are shown in the supplemental section 8 and Table S6.

The statistical distributions of PFAS and TOP results were analyzed using Shapiro-Wilk test. Results indicated that PFAS and TOP data were not normal or lognormal distributed. To compare two paired data sets (e. g., treated to untreated, each having at least five data points), the nonparametric Wilcoxon signed rank test was used. In addition, the nonparametric Mann-Whitney U test was used for comparison between two independent data sets (total having at least five data points), like the comparison between the evaluated PFAS in intermediate samples and influent/effluent samples for the batch aeration process. The statistical software package SPSS (Statistics 26) was used to complete these tests. Statistical differences between datasets were set at p-values less than 0.05 within a 95 % confidence interval and less than 0.100 within 90 % confidence interval. For data sets having less than five data points (PAC, filtration, RO, and combined treatment presented in sections 3.2.3 to 3.2.6), descriptive statistics and plots are presented to illustrate before and after treatment concentrations.

In terms of reporting units, data are reported in the main text in units of nanomoles per liter because it allows for direct comparison between the PFAA-precursors and terminal PFAS (PFCAs and PFSAs) based upon the number of fluorine bonds. This unit of comparison is particularly useful for evaluating transformation of PFAS through the TOP assay due to the emphasis on tracking fluorine containing molecules.

Overall, to identify the differences of detected PFAS, Wilcoxon signed rank test and Mann-Whitney U were used. These tests were used to evaluate the differences among the influent, intermediate, and effluent of evaluated leachate treatment systems and between the direct LC-MS/MS and TOP assay. In addition, the percentage change of Σ₂₆PFAS, ΣPFOA + PFOS and each PFAS group were calculated based on the percent difference of PFAS concentration between the influent and effluent or between the results from direct LC-MS/MS and the TOP assay.

3. Results and Discussion:

3.1. PFAS in untreated influent

The range of Σ₂₆PFAS in all collected influents (n = 20) for individual samples was between 2.44 and 91.7 nmol/L and the average and standard deviation were 34.6 and 25.3 nmol/L, respectively (median of Σ₂₆PFAS was 23.6 nmol/L). This total detected concentration of PFAS was similar with a previous study which showed that the average of detected PFAS was about 35.3 nmol/L in the leachate from 18 US landfills (Lang et al., 2017). On the other hand, based on the average, total detected PFAS concentration in the current study was lower than that in raw leachate from a study conducted in China (246 nmol/L (Yan et al., 2015)) and was higher than that in leachate from another study conducted in Germany (14.7 nmol/L (Huset et al., 2011)).

Based on the average PFAS concentration of influents in the current study, Σ₄short-chain PFCAs, Σ₇long-chain PFCAs, Σ₂short-chain PFSAs, Σ₅long-chain PFSAs, and Σ₈PFAA-precursors represented approximately 41 %, 11 %, 9 %, 6 %, and 33 % of Σ₂₆PFAS. This suggests that the concentration of short-chain substances was usually higher than that of long-chain substances and short-chain PFCAs (C₄-C₇) was the main contributor of PFAS in these influents. The dominance of PFCAs is consistent with the results from other studies that documented concentrations in landfill leachate (Huset et al., 2011; Yan et al., 2015). These studies showed that PFCAs, based on the total detected PFAS, were from 51 % to 84 % in one study (Yan et al., 2015) and around 67 % in another study (Huset et al., 2011). Perfluorobutanoic acid (PFBA) and perfluorohexanoic acid (PFHxA) were dominant within the short-chain PFCAs (PFBA/Σ₄short-chain PFCAs = 37 %, PFHxA/Σ₄short-chain PFCAs = 36 %) and perfluorobutanesulfonic acid (PFBS) was dominant within the short-chain PFSAs (PFBS/Σ₂short-chain PFSAs = 95 %). Also, 5:3 FTCA had the highest concentration of the targeted PFAA-precursors (5:3 FTCA/Σ₈PFAA-precursors = 81 %).

Specifically, for PFOA and PFOS in the influents, the average concentrations measured in this study were 3.35 nmol/L and 0.572 nmol/L, which represented 10 % and 2 % of Σ₂₆PFAS, respectively. Based on the average of concentrations, PFOA was predominant in long-chain PFCAs (PFOA/Σ₇long-chain PFCAs = 91 %), but PFOS only represented 29 % of Σ₅long-chain PFSAs. Overall, PFAS concentrations in the current study were consistent with the results from other studies of US landfills, including state-wide studies that documented levels in Florida, Michigan, and North Carolina (Table 1). However, the average values measured in the current study were higher than for landfills in Australia and Germany.

Table 1.

Comparison of the average of PFOA and PFOS in different studies.

Landfills	PFOA (nmol/L)	PFOS (nmol/L)	Reference
This Study	3.35	0.572	NA
US	2.42	0.200	(Lang et al., 2017)
US (Florida)	3.63	1.13	(Solo-Gabriele et al., 2020)
US (Michigan)	2.87	0.574	(NTH, 2019)
US (North Carolina)	2.01	0.481	(H&H, 2020)
Australia	1.67	0.620	(Gallen et al., 2017)
Germany	0.350	0.0618	(Busch et al., 2010)

3.2. Change of PFAS concentration in different leachate treatment systems

3.2.1. Pond systems

For the pond systems (Fig. 1a), Σ₂₆PFAS decreased 16 % in the effluents based on the average molar concentration of PFAS (Table 2). Among detected PFAS, the total concentrations of detected PFAAs, including short-chain and long-chain PFCAs and PFSAs, in the influents were not statistically different when compared to the effluents (p > 0.150). Σ₈PFAA-precursors decreased 49 % in the effluents. Among these precursors, FOSAs in the effluents was significantly higher than what was found in the influents (p = 0.033), but the concentration of Σ₃FOSAs/Σ₂₆PFAS in both influents and effluents were less than 3 %, and thus it did not influence the concentration of Σ₈PFAA-precursors. In addition, there were no significant changes in other PFAA-precursors (p > 0.200). From Fig. 1b, Σ₄short-chain PFCAs/Σ₂₆PFAS and Σ₂short-chain PFSAs/Σ₂₆PFAS slightly increased from 40 % and 11 % in the influents to 44 % and 20 % in the effluents, while the changes of long-chain PFCAs and PFSAs were not obvious (less than1 % of Σ₂₆PFAS) between the influents and effluents. The increase of short-chain species from the influents to the effluents was mainly due to PFHxA (from 13 % to 20 % PFHxA/Σ₂₆PFAS) and PFBS (from 10 % to 19 % PFBS/Σ₂₆PFAS). Also, Σ₈PFAA-precursors/Σ₂₆PFAS decreased from 33 % in the influents to 20 % in the effluents, mainly caused by the decrease of 5:3 FTCA (from 26 % to 16 % 5:3 FTCA/Σ₂₆PFAS).

Fig. 1.

Panel a, average of detected PFAS classes in nmol/L for different on-site leachate treatment systems (IN = influent, OUT = effluent, Perm = permeate, Conc = concentrate). Panel a in units of nmol/L. Values in μg/L are provided in Figure S2 in the supplemental text. Panel b, detected PFAS classes percentage of total analyzed PFAS for different on-site leachate treatment systems based on the average.

Table 2.

Comparison of influent and effluent Σ₂₆PFAS in units of both ng/L and nmol/L.

Leachate Treatment	Σ26PFAS in Influent		Σ26PFAS in Effluent		Percentage Change of Σ26PFAS in the Effluent
	ng/L	nmol/L	ng/L	nmol/L	Based on ng/L	Based on nmol/L
Pond Systems	12,900	39.2	10,800	32.8	−17 %	−16 %
	(7,700)a	(26.4)	(7,440)	(21.6)
Aeration Tanks	10,200	29.2	9,720	27.8	−5 %	−5 %
	(7,640)	(22.7)	(5,970)	(18.0)
PAC	5,830	19.8	6,170	21.3	6 %	8 %
	(310)	(1.30)	(1,150)	(3.55)
Sand Filtration	14,600	44.3	12,700	39.4	−13 %	−11 %
	(12,000)	(32.9)	(9,450)	(26.7)
Two-staged RO (Influent and Permeate)	10,300	29.8	130	0.430	−99 %	−99 %
	(8,730)	(25.1)	(120)	(0.430)
Two-staged RO (Influent and Concentrate)	10,300	29.8	13,600	39.6	32 %	33 %
	(8,730)	(25.1)	(10,600)	(29.4)
Combined Treatment	740	2.44	250	0.840	−66 %	−66 %
	(20.0)	(0.100)	(10.0)	(0.0700)

^aValues in the 2nd, 3rd, 4th, and 5th columns correspond to means with standard deviations shown within parenthesis.

Specifically, ΣPFOA + PFOS decreased 16 % in the effluents based on the average molar concentration, although concentrations of PFOA and PFOS, respectively, were not statistically different between the influents and the effluents (p = 0.463 for PFOA and p = 0.173 for PFOS). The concentrations of PFOA and PFOS in the current study (PFOA was 0.385–10.4 nmol/L and PFOS was 0.0492–1.96 nmol/L) were similar to the PFOA and PFOS concentrations measured in an evaporation pond at a US landfill (22.2 nmol/L and 0.0280 nmol/L) (Allred et al., 2014), and the average of PFOA and PFOS in two leachate ponds at a Norwegian landfill (0.0990 nmol/L and 0.225 nmol/L) (Knutsen et al., 2019). Overall, the results from the current study were within the same order of magnitude reported by others. Although the concentration of PFOA and PFOS in these studies were within an order of magnitude, differences still existed likely due to the types of waste accepted by different landfills.

3.2.2. Aeration tanks

For aeration tanks, including both sequencing batch reactors (SBR) and continuous aeration systems (Fig. 1a), Σ₂₆PFAS decreased 5 % in the effluents based on the average molar concentration of PFAS (Table 2). All detected short-chain and long-chain PFAAs remained at constant concentrations between the influents and effluents (p > 0.130). Σ₈PFAA-precursors decreased 28 % overall in the effluents. However, among the precursors, FTSAs significantly increased from the influents to effluents (from 1 % to 2 % Σ₃FTSAs/Σ₂₆PFAS, p = 0.043), but the difference of FTSAs percentage of Σ₂₆PFAS was low and thus its overall impact was not large. There was no significant difference of other PFAA-precursors between the influents and effluents (p > 0.130). From Fig. 1b, Σ₄short-chain PFCAs/Σ₂₆PFAS slightly increased from 38 % in the influents to 44 % in the effluents, and other groups of PFAAs did not have obvious changes (less than 2 % of Σ₂₆PFAS) between the influents and effluents. This increase of short-chain PFCAs from the influents to the effluents was mainly contributed by PFHxA (from 19 % to 22 % PFHxA/Σ₂₆PFAS). In addition, Σ₈PFAA-precursors/Σ₂₆PFAS decreased from 30 % in the influents to 22 % in the effluents, mainly because of the decrease of 5:3 FTCA (from 24 % to 16 % 5:3 FTCA/Σ₂₆PFAS).

When comparing the results to a prior study that evaluated PFAS in aeration tanks at Florida landfills, the average PFOA and PFOS in the effluent were measured at 5.56 nmol/L and 1.53 nmol/L (Solo-Gabriele et al., 2020), which were 48 % and 68 % higher than that in the current study (average PFOA and PFOS were 3.76 nmol/L and 0.905 nmol/L). ΣPFOA + PFOS decreased 2 % in the effluents based on the average molar concentration, although this difference was not significant for either PFOA nor PFOS, (p = 0.686 for PFOA and p = 0.465 for PFOS). Overall, for Σ₂₆PFAS and PFOA + PFOS, neither the results of the prior study (Solo-Gabriele et al., 2020) nor the current one showed evidence that aeration tanks, including SBR and continuous aeration systems, reduce PFAS in the effluent.

To further understand the process of continuous aeration systems, we collected samples from the top layer of the same aeration tank in 30-minutes intervals at facility T3. This sampling before, during, and after aeration is unique and not included in prior studies (Solo-Gabriele et al., 2020). Sampling occurred during two separate four-hour aeration cycles (Fig. 2) (T3–01 was collected in Feb. 2019, and T3–02 was collected in Jul. 2020). A total of 10 samples were collected per cycle. These 10 samples included an influent and an effluent (diffusers turned off), plus 8 intermediate samples collected during the active aeration process. The difference of Σ₂₆PFAS between the influent and effluent in both cycles was less than 3 % (difference of Σ₈PFAA-precursors were within 15 %), and this provided the evidence again that aeration did not reduce the PFAS concentration in the effluent effectively.

Fig. 2.

Detected PFAS classes of aeration tanks during two separate four-hour cycles (IN = influent, IN.h0.5 to IN.h4.0 = intermediate samples collected at 0.5-hour intervals, OUT = effluent).

When comparing two cycles collected from the same facility but 17 months apart, the distribution of PFAS groups were different. Σ₈PFAA-precursors in the intermediate samples of T3–01 (collected in Feb. 2019) were lower than that of T3–02 (collected in Jul. 2020), although Σ₂₆PFAS was almost the same for influents and effluents between these cycles. The evaluation of the major ions of these samples (Zhang et al., 2022) suggests that the leachate in T3–02 was more concentrated than T3–01. This might explain the higher PFAS concentrations in the T3–02 during the aeration process.

The most notable observation during the aeration process was that the concentrations of PFAS in the intermediate samples, when the diffusers were on (times 0.5 to 4.0 h), were variable and elevated compared to the samples collected before and after the active aeration process. This is particularly striking given the lower concentrations and consistency between the “IN” and “OUT” samples when the diffusers were turned off. Specifically, average and standard deviation of Σ₂₆PFAS were 30.8 nmol/L and 9.35 nmol/L in the intermediate samples during the aeration process and were 16.5 nmol/L and 0.280 nmol/L in the influents plus effluents samples of this process. Significantly higher Σ₂₆PFAS in the intermediate samples compared with the influents and effluents (p less than 0.01) could be due to two major possibilities, the partitioning of PFAS to solids and the accumulation of PFAS in the top layer. For example, partitioning of PFAS to solids may be enhanced by microbiological uptake during the aeration process (Ebrahimi et al., 2021; Munoz et al., 2019). Microbial induced partitioning towards solids was observed in other studies, especially for long-chain PFCAs and PFSAs (Ebrahimi et al., 2021). Σ₂₆PFAS in the surface samples collected could have been impacted by the resuspension of solids during active aeration, and thus this may cause higher detected Σ₂₆PFAS in the intermediate samples compared to the effluent, which correspond to times when the aerators were turned off. For the other major possibility, prior studies have shown under laboratory conditions that PFCAs and PFSAs accumulate within the foaming layer (air water interface) during aeration (Brusseau & Van Glubt, 2019; Campbell et al., 2009; Robey et al., 2020; Sima & Jaffé, 2021), which can possibly explain the higher PFAS concentrations, especially the Σ₈PFAA-precursors, in the intermediate samples in this study. In addition, existing studies also showed that foam fractionation was effective to sequester PFAS from leachate and these PFAS-enriched foams can be further treated (Meegoda et al., 2020; OPEC, 2022). More research is needed to evaluate PFAS distributions during the aeration process. Overall, although the aeration process might impact PFAS distributions within the tank (higher at the surface foaming layer or within solids suspended during the active aeration process), the system still could not reduce the PFAS in the leachate effectively.

3.2.3. Powder activated carbon system

The system evaluated in this study directly added PAC to the aeration tank. The purpose of this addition was to remove complex non-biodegradable organics. The PAC system was not designed to change the PFAS concentration. For this PAC process (Fig. 1a corresponding to facility T7 and labeled PAC), Σ₂₆PFAS in the effluent was 8 % higher in comparison to the influent based on the average molar concentration (Table 2). The changes of short-chain PFSAs and long-chain classes were less than 2 % of Σ₂₆PFAS, which indicated there were no obvious differences between the influents and effluents. The concentration of short-chain PFCAs increased (from 59 % to 73 % Σ₄short-chain PFCAs/Σ₂₆PFAS), but the concentration of PFAA-precursors decreased (from 13 % to 3 % Σ₈PFAA-precursors/Σ₂₆PFAS) from the influents to effluents (Fig. 1b). The main changes in PFAS distribution were the increase of PFHxA (from 4 % to 14 % PFHxA/Σ₂₆PFAS) and the decrease of 5:3 FTCA (from 30 % to 23 % 5:3 FTCA/Σ₂₆PFAS) from the influents to effluents. The increase of the short-chain PFCAs was also the main reason for the slightly higher Σ₂₆PFAS concentration in the effluent for the PAC system.

Increase of short-chain PFCA (14 % Σ₄short-chain PFCAs/Σ₂₆PFAS) and decrease of PFAA-precursors (10 % Σ₈PFAA-precursors/Σ₂₆PFAS) in the effluent of the PAC system (with aeration process) suggest possible transformation of PFAS. To understand this transformation, a comparison was made between this PAC system and the results from aeration processes described in the prior section. Changes of percentage of these two PFAS groups between the influent and effluent of this PAC system were higher than that of aeration (6 % Σ₄short-chain PFCAs/Σ₂₆PFAS increased and 8 % Σ₈PFAA-precursors/Σ₂₆PFAS decreased in the effluent). This may suggest that the combination of PAC system and aeration may accelerate the transformation of PFAS.

Based on the average of ΣPFOA + PFOS in this PAC system specifically, a 9 % decrease was observed in the effluents, although there was no obvious difference of PFOA and PFOS between the influents and effluents. PFOA decreased from 1.75 nmol/L to 1.53 nmol/L, which rep resented 8 % of Σ₂₆PFAS in the influents and 7 % of Σ₂₆PFAS in the effluents. Similarly, PFOS increased slightly from 0.219 nmol/L to 0.276 nmol/L, which represented 2 % of Σ₂₆PFAS in both influent and effluent. These results of the current study contrast with prior studies which have demonstrated the effective reduction of PFOA and PFOS by activated carbon (AC) systems, including both PAC and granular activated carbon (GAC). Specifically, GAC systems reduced 60–70 % PFOA and 80 % PFOS in drinking water (McCleaf et al., 2017) and GAC combined with solvent based technology reduced over 80 % PFOA and 55 % PFOS in drinking water or groundwater (Siriwardena et al., 2021). For PAC systems, around 35 % of PFOA and 45 % of PFOS could be reduced in stabilized PAC for groundwater (Liu et al., 2020a) and the addition of PAC enhanced over 90 % PFOA and PFOS reduction for coagulation of surface water (Bao et al., 2014).

These prior studies suggested that PFOA and PFOS concentration decreased in GAC and PAC systems that treat drinking water and groundwater. One reason for the differences observed between the current study and prior studies could be due to the PAC configuration. Compared with the pseudo plug flow for GAC, PAC is a complete mix configuration. Specifically, in drinking water treatment, GAC is placed in a fixed column through which water flows. In the case of facility T7, the PAC was directly added to the aeration tank such that the liquid to PAC ratio was very high, perhaps decreasing the efficiency of PAC to reduce PFAS. Another reason for the lack of reduction of PFOA and PFOS concentration by the PAC system (T7) could be attributed to the chemical composition of landfill leachate. In contrast to drinking water and groundwater, landfill leachate contains many other chemical constituents that might be preferentially adsorbed by the PAC, thereby not allowing for reduction of PFAS for this aqueous sample type.

In addition to the bulk chemical composition characteristics of leachate, the distribution of the PFAS classes itself might have also affected the ability of PAC to reduce PFAS in the current study. A study of drinking water contaminants mentioned that the adsorption of PFAS on PAC decreased with shorter chain length (Sun et al., 2016), and thus high concentrations of short-chain PFCAs in the leachate decreased the ability of PAC to reduce PFAS. Short-chain PFCAs represented the majority (68 %) of Σ₂₆PFAS in the samples from facility T7.

Overall, this current research found no evidence (in the configuration used) that direct addition of PAC could efficiently reduce PFAS in leachate. This observation was based upon the results from one facility only, designed with a very high liquid to PAC ratio, thereby limiting the scope of the conclusions that can be drawn. Competitive absorption of organics could limit its applicability for treating leachate. More research is needed to further evaluate the efficacy of activated carbon in removing PFAS from leachates.

3.2.4. Filtration

Similar as the PAC system, the results suggest that sand filtration was not effective at reducing the Σ₂₆PFAS (Fig. 1a and labeled sand filtration). The Σ₂₆PFAS in the effluent from the sand filtration systems were only 11 % less than that measured in the influent (Table 2). Considering that sand filtration was designed to remove total suspended solid (TSS), results suggest that the decrease of PFAS may be associated with TSS in the leachate. Also similar with aeration tanks, all PFCAs and PFSAs remained at constant concentrations between the influent and effluent based on the detected concentration. Evaluating the distribution of Σ₂₆PFAS (Fig. 1b), short-chain PFCAs increased (from 42 % to 48 % Σ₄short-chain PFCAs/Σ₂₆PFAS) and PFAA-precursors decreased (from 34 % to 26 % Σ₈PFAA-precursors/Σ₂₆PFAS) from the influents to effluents again suggesting possible transformations. This possible transformation may be related to the aeration process as the leachate leaves the landfill environment and flows through the filtration system. The main contribution of the increase of short-chain PFCAs was PFBA (from 14 % to 18 % PFBA/Σ₂₆PFAS), and the main contribution of the decrease of PFAA-precursors was 5:3 FTCA (from 30 % to 23 % 5:3 FTCA/Σ₂₆PFAS).

Specifically, based on the average, ΣPFOA + PFOS decreased 7 % in the effluents, although there was no obvious difference of PFOA and PFOS between the influents and effluent. PFOA measured at 3.27 nmol/L (influent) and 3.26 nmol/L (effluent), which represented 7 % of Σ₂₆PFAS in the influents and 8 % of Σ₂₆PFAS in the effluents. Similarly, PFOS measured at 0.813 nmol/L to 0.754 nmol/L, which represented 2 % of Σ₂₆PFAS in both influent and effluent, respectively. A study that evaluated the pore size of filtration systems suggested that more PFAS would be removed with smaller pore sizes (Busch et al., 2010). The pore size of sand filtration is relatively large, thereby allowing PFAS attached to small particles, in addition to dissolved PFAS, to flow through the system and not be captured. Overall, the results suggested that sand filtration cannot effectively reduce PFAS from leachate. Given that only two leachate filtration systems were evaluated, more work is needed to reconfirm the ability of PFAS to adsorb to particulates and the ability of full-scale filtration systems to remove such particulates from leachate.

3.2.5. Two-stage reverse osmosis (RO) system

This two-stage RO system (pressure ranged from 200 to 250 psi) was design to treat leachate specifically without pretreatment steps. The landfill spray irrigates the permeate on turf on top of the landfill and injects the concentrate back into the landfill. The permeate and concentrate from the first stage were 30 % and 70 % of the influent, and the permeate and concentrate from the second stage were 50 % and 50 % of the permeate from the first stage. For this two-stage RO system (T6), the first stage was sampled twice separately (13 days apart, both were collected in March 2019), and the second stage was sampled only once. Evaluating all three sets of RO samples collectively (Fig. 1a and labeled two-stage RO), Σ₂₆PFAS in the permeate (average was 0.430 nmol/L) were 99 % lower than that in the influent (average was 29.8 nmol/L), although Σ₂₆PFAS in the concentrate (average was 39.6 nmol/L) were 33 % higher than that of influent. Σ₂₆PFAS mass towards the concentrate from both stages was around 4.59 mmol per day (equivalent to 1.59 g per day) based on the flow rate of the influent of 1.31 L/s. For the percentage of each PFAS class (Fig. 1 b), short-chain PFCAs increased from 42 % Σ₄short-chain PFCAs/Σ₂₆PFAS in the influents to 75 % Σ₄short-chain PFCAs/Σ₂₆PFAS in the permeates and PFAA-precursors decreased from 39 % Σ₈PFAA-precursors/Σ₂₆PFAS in the influents to 5 % Σ₈PFAA-precursors/Σ₂₆PFAS in the permeates. The differences of percentage of other PFAS classes between the influents and permeates were less than 2 % of Σ₂₆PFAS since other PFAS classes represented only a small fraction of the Σ₂₆PFAS.

When evaluating first versus second stage RO (Fig. 3 and labeled as two-stage RO), the results showed that the first stage reduced over 96 % of all five PFAS classes in the permeate compared with the influent (Table 2). These five PFAS classes in the concentrate were 4 % to 50 % higher than that in the influent. The results of the second stage showed that almost 100 % of all five PFAS classes had been reduced in the permeate. In the concentrate of the second stage, only short-chain PFCAs increased 25 % compared with the influent, and the changes of other PFAS classes were less than 1 %. For specific PFAS, the percentage decrease in the permeate was higher than 94 % when the original concentration of the influent was high, >0.01 nmol/L. When the original concentration in the influent was low, less than 0.01 nmol/L, the percentage decrease in the permeate was around 75 %. This may suggest that the bleed through of PFAs is finite when the detected PFAS was low in the original influent. These results suggest that the two-stage RO system reduced the PFAS concentration in the permeate effectively, with higher percent reductions in the first stage in comparison to the second stage.

Fig. 3.

Detected PFAS classes in RO systems: T6 is the two-stage RO system (IN = influent, OUT.Perm = permeate, OUT.Conc = concentrate); T10 is the combined treatment with RO system (IN = initial influent of the combined treatment, IN.RO = influent of the RO system, OUT.Perm = permeate of the RO system, OUT.Conc = concentrate of the RO system, OUT.Final = final effluent of the combined treatment).

For PFOA and PFOS specifically, 99 % were reduced in the permeate for the two-stage RO system. This result was consistent with a prior study, which showed that RO reduced over 90 % PFAAs in permeates, including 99 % reduction of PFOA and PFOS for wastewater (Tang et al., 2006). For the current study, ΣPFOA + PFOS in the permeates from the two-stage RO system were all less than 0.05 nmol/L. These results were below the 2016 health advisory for ΣPFOA + PFOS as issued by US EPA, which was 70 ng/L (equivalent to 0.13–0.14 nmol/L) (Hamid et al., 2018; US EPA, 2016). This data was not compared against the 2022 health advisory (US EPA, 2022) because it was lower than the detection limits of the analytical equipment used. In summary, the reductions were significant using RO processes, which produced permeates that met the 2016 drinking water guideline level. Additional treatment may be necessary if the permeate is to meet the 2022 drinking water guideline level.

3.2.6. Combined treatment with RO

For the combined treatment (T10) where RO served as a treatment step after coagulation, sedimentation, sludge thickening, microfiltration, pH adjustment and cartridge filtration, results (Fig. 1a and Fig. 3 both labeled combined treatment) showed that Σ₂₆PFAS in the final effluent was 66 % lower than that in the initial influent (Table 2), and the distribution of PFAS classes were almost the same (difference less than 2 % of Σ₂₆PFAS). When isolating the RO component of the combined system, results also showed that all five PFAS classes in the permeate were over 99 % lower than that in the influent. The PFAS concentrations in the concentrate were higher (over 150 %) than that in the influent of the RO system. The permeate and concentrate of this RO system were estimated at around 70 % and 30 % of the influent based on the mass balance of detected PFAS concentrations. According to the system design as reported by the operator, permeate and concentrate were set at 85 % and 15 % of the influent. The Σ₂₆PFAS mass towards the concentrate was around 4.79 mmol per day (equivalent to 1.53 g per day) based on the flow rate of the influent of 52.6 L/s. For this combined treatment process, the plant combined the pure RO permeate (0.05 nmol/L) with bypass treated leachate from microfiltration, resulting in an increase in the final effluent PFAS concentrations (0.840 nmol/L) due to the blending of these two waters. Overall results showed that the RO process within the combined system could reduce most of the Σ₂₆PFAS in the leachate.

Specifically for PFOA and PFOS, the final effluent was 70 % and 66 % lower, respectively, compared to the initial influent of the combined treatment. These results were consistent with prior studies that showed that the combined treatment with RO system was effective at removing PFAS from the leachate (Yan et al., 2015; Busch et al., 2010). ΣPFOA + PFOS in the final effluent from the combined treatment with RO were also less than 0.05 nmol/L, which was below the 2016 US EPA health advisory for ΣPFOA + PFOS (Hamid et al., 2018; US EPA, 2016).

3.2.7. Summary

Overall, pond systems, aeration tanks, PAC systems, and sand filtration in this study were found to convert PFAA-precursors into terminal species without significant decrease of Σ₂₆PFAS. Only the RO system or the combined system with RO showed significant reductions of PFAS from leachate. RO, however, does not destroy PFAS. Essentially, this process transfers PFAS from the permeate into the concentrate. These results are consistent with prior studies focused on treatment of domestic wastewater which have shown that PFAAs concentrations decreased by 90 % in wastewater treated by RO processes and by 24 % for other traditional wastewater treatment systems (Arvaniti & Stasinakis, 2015).

Because of the high PFAS concentrations in landfill leachates, even after on-site treatment, more facilities may need to rely on off-site wastewater treatment systems to further reduce PFAS. In the current study 15 of the 39 landfill facilities (or 38 %) used on-site leachate treatment systems whereas 62 % had their leachate treated off-site. This proportion is consistent with the trends nationally that suggest that most facilities utilize off-site wastewater treatment systems to handle leachate. (Feng et al., 2021; Ganesh & Jambeck, 2013; Masoner et al., 2016). Wastewater plants receiving treated leachates could expect to receive leachates with a predominance of terminal species (PFCAs and PFSAs) given the transformation of precursors observed during the leachate treatment processes observed in the current study. Wastewater plants receiving untreated leachates could expect to receive leachates with more precursors likely in forms that are not detected by traditional targeted PFAS analyses. Monitoring of incoming PFAS concentrations at wastewater plants should consider these potential differences when accepting landfill leachates given that some PFAS may go undetected, especially for leachates that are not pre-treated.

3.3. TOP assay

Results (Fig. 4) showed a significant change at the 90 % confidence interval (p = 0.058) between the Σ₂₆PFAS from the original samples (PFAS analysis) and the corresponding oxidized samples (TOP assay as given by the “-OX” suffix) when all samples were considered collectively. The average concentration showed an increase in the Σ₂₆PFAS oxidized samples of 30 % relative to the original samples (based on the molar concentration of PFAS). Although TOP assay only measures the targeted PFAAs and their related precursors (Houtz & Sedlak, 2012; McDonough et al., 2019), this still suggested that at least 30 % untargeted PFAA-precursors of Σ₂₆PFAS were not detected in the direct LC-MS/MS based on the results from the current study. Also, Σ₂₆PFAS concentration in most of the oxidized samples increased relative to the original samples. However, the concentrations of five of the 22 oxidized samples (IN3-OX in the 2nd sample, OUT7.Con2-OX, IN11-OX, IN16-OX, and OUT16-OX) decreased >10 % relative to the original samples, and Σ₂₆PFAS concentration in these five samples were lower than 30 nmol/L. This observation might be due to either analytical variability at low concentrations or over oxidation resulting in the loss of PFAS.

Fig. 4.

Comparison of detected PFAS in PFAS analysis and TOP assay. (OX = oxidized samples in TOP assay).

For PFAS classes, specifically, there was a significant difference of short-chain PFCAs between the original samples and oxidized samples (from 44 % to 68 % Σ₁₁PFCAs/Σ₂₆PFAS, p = 0.001), and this suggested that short-chain PFCAs in oxidized samples were usually higher than that in the original samples (additional details shown in Table S10). This increase in the oxidized samples was mainly caused by PFBA (from 10 % to 30 % PFBA/Σ₂₆PFAS, p = 0.002) and perfluoropentanoic acid (PFPeA) (from 9 % to 15 % PFPeA/Σ₂₆PFAS, p = 0.001), which were significantly increased in oxidized samples. For other PFAAs classes, including long-chain PFCAs (difference less than 2 % of Σ₂₆PFAS, p = 0.758), short-chain PFSAs (difference less than 1 % of Σ₂₆PFAS, p = 0.615), and long-chain PFSAs (difference less than 1 % of Σ₂₆PFAS, p = 0.638), there was no significant difference between the oxidized samples and original samples. Among long-chain PFCAs and PFSAs, PFOA (from 11 % to 9 % PFOA/Σ₂₆PFAS, p = 0.733) and PFOS (from 3 % to 2 % PFOS/Σ₂₆PFAS, p = 0.277) slightly decreased in the oxidized samples, but this decrease was not significant. For PFAA-precursors in oxidized samples, 80 % of the results were lower than the LOQ, and Σ₈PFAA-precursors significantly decreased relative to original samples (from 27 % to 7 % Σ₈PFAA-precursors/Σ₂₆PFAS, p = 0.002).

Overall, after oxidation through the TOP assay in this study, most targeted PFAA-precursors were converted into short-chain PFCAs, especially PFBA and PFPeA. PFSAs remained at relatively constant concentrations. These results were consistent with those from other studies (Houtz & Sedlak, 2012; ALGA, 2019).

4. Limitations

One of the limitations of this study was the variability of the field data. Because all samples were collected on-site, environmental factors, like antecedent rainfall and temperature, could cause possible variability in the samples. Precipitation before the sampling date may have diluted the leachate from the systems open to the atmosphere, like the pond systems and open-tank aeration processes. In addition, considering that individual flows in each pond system were variable, estimation of PFAS concentrations in the influents based on the weighted average and grab samples of the effluents from these systems may also contribute towards the variability of the results. Also, the number of samples collected from pond systems and aeration tanks were thirteen and five, respectively, which are not large. The five value was the minimum value for the Wilcoxon analysis used in the current study. For other treatment systems, only simple statistical descriptions were used because of smaller sample size. This study focused on evaluating data for a cross-section of 15 treatment facilities. Future work should focus on in-depth studies that evaluate operational variables, temporal/spatial variability, and further measurements of uncertainty. In addition, species transformation in this study is based on the change of the concentration. A more comprehensive mass balance is needed by measuring more PFAS species to further understand the details about how PFAS changes through leachate treatment systems.

In addition, only eight PFAA-precursors were targeted for analysis. Comparison between the direct LC-MS/MS analysis and TOP assay suggested that there might be additional 30 % untargeted PFAS (based on the average molar concentration of the leachates evaluated and analysis provided in the supplemental text section 10), especially for PFAA-precursors not measured in this study. Based on a prior study that measured 24 PFAA-precursors, PFAA-precursors were found to contribute around 14 % of all targeted PFAS, and it also indicated that perfluoro-4-ethylcyclohexanesulfonate (PFECHS), which did not belong to traditional PFAS classes, was also detected in the landfill leachate (Liu et al., 2020b). Therefore, more work is needed to identify the unaccounted PFAS perhaps using a larger library of targeted PFAS or utilizing non-targeted analyses.

For the TOP assay, complete oxidation during the TOP assay means PFAA-precursors should not be detected in the oxidized samples (Houtz & Sedlak, 2012; ALGA, 2019). Although there was a significant decrease of Σ₈PFAA-precursors in the oxidized samples of this study, these precursors were still detected in around 70 % of oxidized sample. This result suggested that high concentration of the organic matter was competing with the PFAS for the oxidation capacity rather than consuming it prior to oxidation of PFAS. Also, the decrease of Σ₂₆PFAS in the oxidized samples should be less than 10 % of that in the original samples based on an existing study (ALGA, 2019). Although this was consistent with most of the samples in this study, five oxidized samples still showed exceptions to this trend. A recent experiment (Guelfo personal communication, 2021) showed that if insufficient quantities of the strong oxidant (persulfate potassium) were added during the TOP assay, the total detected PFAS in the oxidized samples would be much less than that in the original samples. When more oxidant was added, more PFAS would be detected in the oxidized samples. This suggests that incomplete oxidation was the major limitation of the TOP assay to assess the PFAS in the leachate samples. Therefore, future work may focus on adjusting the amount of oxidant used in the TOP assay based upon the strength of the leachate, possibly correlating with chemical oxygen demand (COD).

5. Conclusion

Results from this current study showed that most of the leachate treatment systems, like pond systems, aeration tanks, and PAC systems, did not have the ability to reduce PFAS in the effluent. Sand filtration showed slight reduction of PFAS in the treated leachate, but its efficiency was low and thus was not observed at statistically reduced concentrations. RO was found to reduce over 98 % of the PFAS in the final permeate. ΣPFOA + PFOS in the permeate was observed to be less than 70 ng/L (0.13–0.14 nmol/L), which was below the 2016 health advisory issued by the US EPA. Although PFAS concentrations increased in the concentrate, these concentrates could be recycled through recirculation back into the landfill, sequestered through solidification, disposed via deep well or treated using emerging thermal and electrochemical technologies (Ahmed et al., 2020). Therefore, the two-stage RO system and the combined treatment with RO process evaluated in this study were able to reduce most of the PFAS in the permeate while the other systems did not demonstrate reduction of PFAS concentrations at levels of significance.

We hypothesized that PFAS transformation occurred during treatment with precursor transformation towards terminal substances. The results of this study support this hypothesis. Evidence of species transformation was observed in most on-site treatment systems with the conversion of PFAA-precursors to short-chain PFCAs. For the pond systems, aeration tanks, and PAC system, measurements suggested that the detected short-chain PFCAs increased in the effluent mainly due to PFHxA, and the measurements of sand filtration suggested this increase in the effluent was caused by PFBA. All treatment systems suggested that the decrease of PFAA-precursors in the treated leachate was caused by the reduction of 5:3 FTCA. Therefore, based on the PFAS class percentage of Σ₂₆PFAS, PFAA-precursors would transform to short-chain PFCAs in the treated leachate, which was due to the conversion from 5:3 FTCA to PFBA or PFHxA in most treatment systems.

Although aeration treatment did not result in reductions in PFAS concentrations, results show that higher concentrations of PFAS were observed in the surface layer of the aeration tank during active aeration. These higher concentrations had a larger proportion of precursors. More research is needed to evaluate the cause for the observation and whether this accumulation of PFAS in the surface layers during active aeration could be potentially exploited to enhance downstream PFAS treatment processes.

Evidence from the TOP assay evaluated in the current study also supported that PFAA-precursors were transformed into short-chain PFCAs, especially to PFBA and PFPeA, after strong oxidation. Another group of terminal species, PFSAs, remained generally constant before and after strong oxidation. The difference in the oxidized samples suggested that the fraction of untargeted PFAA-precursors (equation given in the supplemental text section 10) was at least 30 % based on molar concentration. The observed variability of the TOP assay might be caused by the incomplete oxidation. Overall, TOP results support that transformations are possible from non-targeted precursors to short-chain PFCAs, but the assay might be subject to limitations when estimating the total targeted and untargeted PFAS concentrations in landfill leachates.

Overall, results from this study confirmed that RO reduces PFAS (98 % - 99 % Σ₂₆PFAS) effectively in the permeate and also showed that for most leachate treatment systesms PFAA-precursors (e.g., 5:3 FTCA) usually converted into terminal species, especially into short-chain PFCAs. When comparing the detected PFAS measured directly via LC-MS/MS method against the results from the TOP assay, results suggest that leachate contains unmeasured precursors and efforts should focus on expanding precursor measurements in future work.