Relationships between per- and polyfluoroalkyl substances (PFAS) and physical-chemical parameters in aqueous landfill samples

Abstract

Variable chemistries of liquids from landfills can potentially impact levels of per- and polyfluoroalkyl substances (PFAS). The objective of the current study was to evaluate relationships between physical-chemical properties (bulk measurements, oxygen demand components, and metals) and PFAS concentrations in different types of aqueous landfill samples. Aqueous landfill samples were collected from 39 landfill facilities in Florida, United States. These samples included leachates from landfills that receive different waste types, such as municipal solid waste incineration ash (MSWA), construction and demolition debris (C&D), and municipal solid waste (MSW). Additional aqueous landfill samples were sourced from treated landfill leachate, gas condensate, stormwater, and groundwater from within and near the landfill boundaries. Results showed significant correlations (p < 0.05) between ∑₂₆PFAS and alkalinity (r_s = 0.83), total organic carbon (TOC) (r_s = 0.84), and ammonia (r_s = 0.79) for all leachate types. Other physical-chemical parameters that were significantly correlated (r_s > 0.60, p < 0.05) with PFAS included specific conductivity, chemical oxygen demand (COD), and to a lesser extent, total dissolved solids (TDS) and total solids (TS). For gas condensates, PFAS was significantly correlated with TOC. Stormwater and groundwater, within and near the landfill boundaries, had considerably lower levels of PFAS and had a minimal correlation between PFAS and physical-chemical parameters. Although PFAS concentrations and physical-chemical parameters and their correlations varied between different types of aqueous landfill samples, results suggest that physical-chemical properties can be useful indicators of relative PFAS concentrations within a leachate type. More research is needed to validate the mechanisms that relate physical-chemical parameters to PFAS concentrations in landfill leachates.

Graphical abstract

1. Introduction

Landfill leachate forms when liquids contact solid waste, via rainwater, added liquids, or the inherent moisture within the waste itself (Sivula et al., 2012). Leachate is typically managed to control impacts from organic matter, inorganic salts, organic trace pollutants, and heavy metals (Ikehata and Li, 2018; Stefanakis et al., 2014; Kjeldsen et al., 2002). Along with these traditional pollutants, per- and polyfluoroalkyl substances (PFAS) are also starting to be considered as part of landfill management practices, given the high levels observed in leachates (Chen et al., 2020; Lang et al., 2016; Liu et al., 2021a; Solo-Gabriele et al., 2020). For example, some US state government agencies (e.g., California, Michigan, and Minnesota) are currently developing their own landfill leachate pre-treatment requirements (CWB 2023; MPART 2023; MPCA, 2023).

These initiatives raise new questions about how leachates and other liquids from landfills should be measured and managed. Physical-chemical parameters in leachate are known to be different among municipal solid waste (MSW) and MSW ash (MSWA) landfills, and thus should be measured as part of standard landfill management programs (Moody and Townsend, 2017). To minimize impacts associated with the release of landfill leachates, Thakur and Medhi (2019) recommended measuring chemical oxygen demand (COD) and ammonia. Additional researchers also emphasized the need to measure heavy metals, pH, and total organic carbon (TOC) as part of landfill leachate management to meet local regulations or wastewater treatment plant (WWTP) requirements (Foo and Hameed, 2009; Gotvajn et al., 2009). Such measurements were also required before collected leachate can be discharged to a treatment system or to a receiving water body capable of assimilating the chemical constituents (Allen, 2001; Raghab et al., 2013).

Even though there are more than 9000 PFAS identified, as of March 14, 2023 only six specific PFAS (perfluorononanoic acid (PFNA), perfluorooctanoic acid (PFOA), perfluorooctane sulfonate (PFOS), perfluorohexane sulfonic acid (PFHxS), perfluorobutane sulfonic acid (PFBS), and the GenX chemical hexafluoropropylene oxide dimer acid (HFPO-DA)) are proposed for National Primary Drinking Water Regulations by the US Environmental Protection Agency (US EPA, 2022a). The US EPA currently added these same six PFAS chemicals to a list of risk-based values that help the EPA determine if response or remediation activities are needed (US EPA, 2022b). The US EPA has also begun rulemaking to revise limitations for Organic Chemicals, Plastics, and Synthetic Fibers (OCPSF) categories to address discharges from PFAS manufacturing facilities, and the initiation of detailed studies of PFAS discharges from landfills (US EPA, 2016). Thus, concerns over the elevated levels of PFAS have prompted landfill operators to not only focus on the common physical-chemical parameters in leachate but also consider PFAS levels in management plans. Currently, only a few studies have evaluated the relationships between PFAS and landfill leachate physical-chemical parameters. Limited parameters were evaluated per study, and these parameters included ammonia, bicarbonate, pH, TOC, and COD (Benskin et al., 2012; Hepburn et al., 2019; Solo-Gabriele et al., 2020).

No reviewed literature evaluated total dissolved solids (TDS), total solids (TS), and metals collectively with PFAS measurements. The inclusion of TS and TDS along with measures of TSS assists in completing the solids balance, especially for systems such as landfills that are influenced by shifts in carbon dioxide partial pressures and potential precipitation of calcium carbonates. Similarly, the inclusion of metals can provide insight into which manufactured products may be contributing towards PFAS in leachate. For example, electroplating of chrome, nickel, cadmium, zinc or lead was a common contributor as it related to PFAS use (Liu et al., 2022a). To better understand the mechanisms of PFAS releases from landfills, more emphasis is needed on measuring a broad range of physical-chemical parameters. This study is unique in the number of different physical-chemical parameters measured coupled with large sample sizes.

In addition to controlling leachate, landfill management also includes minimizing impacts to stormwater and groundwater. To the authors’ knowledge, no studies have evaluated the relationships between physical-chemical parameters and levels of PFAS in these water types. This lack of information about PFAS levels in stormwater and groundwater at landfills signals a major knowledge gap in understanding how PFAS in landfill leachates could cause increased PFAS pollution risks within the surrounding environment.

The current study focused on aqueous landfill samples collected from 39 facilities and evaluating correlations between PFAS and physical-chemical parameters in different types of landfill leachate and in other critical aqueous samples at landfills (e.g., groundwater and stormwater). Physical-chemical properties (35 measurements including five bulk measurements, four measures of oxygen demand components, and 23 metals) were compared to 26 PFAS levels in MSWA leachate; construction and demolition (C&D) leachate; MSW leachate; treated leachate (Treated); landfill gas condensate (GC); groundwater (GW); and stormwater (ST) to assess which physical-chemical parameters correlated with PFAS. The authors hypothesize that physical-chemical parameters are useful as indicators of relative PFAS levels within a particular leachate type. Such analyses serve as a starting point to evaluate the mechanisms that influence PFAS levels in leachate.

2. Methods

Aqueous landfill samples were analyzed for 26 PFAS chemicals. To simplify discussions, analyses described in the main text focused on three groups of PFAS: total PFAS (Σ₂₆PFAS), PFAS precursors (Σ₈PFAA precursors), and the sum of PFOA and PFOS (ΣPFOA + PFOS).

The collected landfill aqueous samples were classified into seven categories (MSWA leachate, C&D leachate, MSW leachate, Treated, GC, GW, and SW). MSWA leachate was collected from landfills that accepted ash from municipal solid waste incineration that aimed to reduce waste volume and generate electricity (Klein et al., 2001). C&D leachate was collected from landfills that accepted construction and demolition waste, which included concrete, soil, metals, timber, and plastics (Jambeck et al., 2008; Solo-Gabriele et al., 2020). MSW leachate was collected from landfills that accepted household waste, which typically includes food waste, green waste, cardboard, glass, paper, and plastics (Lang et al., 2017; Solo-Gabriele et al., 2020). Landfill facilities are occasionally equipped with on-site leachate treatment to meet the effluent requirements of the receiving wastewater treatment plant or other receiving water entities (Pi et al., 2009; Renou et al., 2008; Singh et al., 2012). In the current study, the effluent for landfills with on-site treatment of the collected raw leachate is termed “treated leachates.” Leachate treatment methods for the facilities evaluated included evaporation ponds, aeration tank systems, powdered activated carbon, filtration, and reverse osmosis (Zhang et al., 2022). Gas condensate samples collected in the current study were identified by the landfill operator and were collected from a vapor-liquid separator (knock-out pot) that received gas from the landfill gas collection system (Solo--Gabriele et al., 2020) or from a landfill gas flare station. Groundwater at most landfills is generally monitored by collecting samples upgradient and downgradient from the landfill at designated monitoring wells (Mor et al., 2006). In this study, upgradient and downgradient locations were defined by groundwater contours measured at the site. Among the 39 landfills evaluated, 29 landfills were equipped with complete liners and leachate collection systems to prevent the contamination of groundwater. The remaining 10 landfills had unlined cells. These cells included some active C&D waste cells and some aged, closed cells. Stormwater at landfills is typically managed through surface impoundments (Marques and Hogland, 2001). In this study, samples were collected from impoundments within the landfill boundary that were designed to collect landfill stormwater. The stormwater runoff could be generated from areas surrounding active cells or from the capped/closed cells at the landfill. Additional details about the landfill samples are available in Chen et al. (2023).

2.1. Landfill facilities

Among the 39 landfill facilities, 164 aqueous landfill samples were collected and classified into seven categories (Tables S–1). The number of samples collected from each landfill facility was based on availability, which included accessible leachate collection points and groundwater upgradient and downgradient positions from the landfill plus stormwater sampling locations.

Most landfill facilities accepted more than one type of waste (e.g., MSWA, C&D, MSW) which was sometimes separated by different landfill cells to accept each type of waste within a facility. However, landfills sometimes mix different types of waste within a landfill cell. When classifying the leachate samples by waste type, the dominant waste type with the highest proportion, determined by the landfill’s waste volume records, was used to establish the leachate category for the sample. Details about the landfills and the number of pure and co-disposed landfill cells are available in the supplemental material (Tables S–1 and Tables S–2).

2.2. PFAS and physical-chemical measurements

The 26 PFAS evaluated included perfluoroalkyl acids (PFAAs) and their precursors. The PFAAs, considered terminal PFAS (Buck et al., 2011; Chen et al., 2022; Lindstrom et al., 2011), were separated into two subcategories which included perfluoroalkyl carboxylic acids (PFCAs) and perfluoroalkyl sulfonic acids (PFSAs). The PFCAs and PFSAs were further classified into short-chain and long-chain substances by their chemical structures. In general, less than or equal to seven carbon atoms in the fluorine-carbon chain were identified as short-chain PFCAs. Short-chain PFSAs have less than or equal to six carbon atoms in the fluorine-carbon chain (Brendel et al., 2018). Precursors of PFAA are generally considered to be less stable and can degrade into terminal PFAS depending on environmental conditions (Benskin et al., 2012; Lang et al., 2017; Wang et al., 2020). Precursors of PFAAs considered in the current study included the following three groups: fluorotelomer carboxylic acids (FTCAs), perfluoroalkane sulfonamides (FASAs), and fluorotelomer sulfonic acids (FTSAs) (Buck et al., 2011). Details pertaining to the 26 PFAS (11 PFCAs and 7 PFSAs, and 8 PFAA precursors to include 2 FTCAs, 3 FASAs, and 3 FTSAs), as wells as the PFAS analysis methods are available in the supplemental material (Tables S–3).

In terms of physical-chemical analyses, measurements included bulk measurements (specific conductivity, pH, alkalinity, TDS, TS), measures of oxygen demanding components (TOC, biochemical oxygen demand-BOD, COD, ammonia due to its exertion nitrogenous oxygen demand), and metals (aluminum, antimony, arsenic, barium, beryllium, cadmium, calcium, chromium, cobalt, copper, iron, lead, magnesium, manganese, mercury, nickel, potassium, selenium, silver, sodium, thallium, vanadium, zinc). Physical-chemical parameters measured in the field included specific conductivity, pH, and temperature. For the metal measurements, sodium, potassium, calcium, magnesium, and iron were considered as major metal ions based on their relative concentration. The remaining metals were considered as minor metal ions. Due to differences in preservation methods and compatibility with bottle types, samples were distributed into multiple bottles. Sample splits with extra bottles were used for a subset of physical-chemical parameters (i.e., TOC, COD, ammonia, and metals) for sites identified as quality control samples. More details about physical-chemical parameter measurement methods and about sample collection (Tables S–4 and Tables S–5) are provided in the supplemental text.

2.3. Data analysis

Statistical analyses were conducted for aqueous sample categories with sufficient data points (i.e., MSWA: n = 17, C&D: n = 9, MSW: n = 67, Treated: n = 41, and GC: n = 13). The groundwater (n = 8) and stormwater (n = 9) categories did not have sufficient data points, and so only descriptive statistics (means, median, and ranges) were provided. For the categories with sufficient data points, the statistical distributions of PFAS and physical-chemical results were analyzed using a Shapiro-Wilk test. Initial analyses indicated that the physical-chemical and PFAS data were not normally nor lognormally distributed. As a result, nonparametric statistical analysis methods were used to evaluate the data. When at least five data points were available, the nonparametric Wilcoxon signed rank test (paired data, without continuity correction) were used to compare two or more data sets. Strong statistical differences between datasets at 95% confidence limits, corresponded to p-values less than 0.05. Correlations between the physical-chemical parameters and PFAS concentrations for data sets containing at least five data points were analyzed using the nonparametric Spearman’s rank correlations (2-tailed). The Spearman’s rank correlation coefficient was considered strong for values greater than 0.60 and weak if less than 0.40. Correlations were considered significant at 95% confidence limits for p-values less than 0.05. Non-detected values were replaced with the $LOD/\sqrt{2}$ as recommended by Verbovšek (2011) when analyzing the comparisons and correlations. Also, multiple linear regression was used to evaluate associations between individual PFAS and physical-chemical parameters in different types of landfill leachate and other aqueous samples. More details about the linear regression model development are provided in the supplemental text.

The statistical software package, SPSS (version 26), was used to complete the statistical tests, including the Shapiro-Wilk test, Spearman’s rank correlation test, outlier test, min-max normalization, and multiple linear regression analysis. In terms of the outliers, small circles were used to designate the outlier values (Figs. 1 and 2), and stars were used to designate extreme outliers. Outliers, as defined by SPSS, were 1.5 times the difference between the third and first quartiles. Extreme outliers were defined as three times the difference between the third and first quartiles. Since the data generated were non-parametric, results in subsequent discussions are based upon median values.

Fig. 1.

Physical-chemical parameter measurements in seven types of aqueous landfill samples (MSWA = municipal solid waste incineration ash, C&D = construction and demolition wastes, MSW = municipal solid waste, Treated = treated leachate, GC = gas condensate, GW = groundwater, ST = stormwater). This boxplot was used to graphically depict the parameters through their quartiles, which included the minimum (bottom error bar), first quartile (bottom of the box), median (line within box), third quartile (top of the box), maximum (top error bar), and the outliers.

Fig. 2.

PFAS measurements in different types of landfill aqueous samples This boxplot was used to graphically depict the parameters through their quartiles as described in the caption to Fig. 1.

The physical-chemical parameters were evaluated among several PFAS categories including Σ₂₆PFAS, Σ₄ short chain PFCA, Σ₇ long chain PFCA, Σ₁₁ PFCA, Σ₂ short chain PFSA, Σ₅ long chain PFSA, Σ₇ PFSA, Σ₈PFAA precursors, and ΣPFOA + PFOS. As noted earlier, discussions focused predominantly on Σ₂₆PFAS, Σ₈PFAA precursors, and ΣPFOA + PFOS, as these groupings of the data exhibited the strongest correlations or provided results relevant to regulatory guidelines. When defining the most significant parameters for correlation analysis, parameters with the highest Spearman correlation were considered first. The parameters with the highest correlations were checked against the parameters identified through multiple linear regression to confirm their contribution towards correlations with PFAS levels.

3. Results

3.1. MSWA leachate

Results shown in Fig. 1 indicate MSWA leachates (n = 17) had the highest level of all physical-chemical parameters (p < 0.05), including conductivity (med: 16,000 mg L⁻¹), TDS (med: 9800 mg L⁻¹), TS (med: 11,000 mg L⁻¹), and some major (med: 3100 mg L⁻¹ for Na, 540 mg L⁻¹ for K, 670 mg L⁻¹ for Ca) and minor metal ions (1.1 mg L⁻¹ for Ba) among all types of aqueous landfill samples. However, MSWA leachate had the lowest alkalinity level (med: 650 mg L⁻¹) among the C&D, MSW, and GC leachate samples evaluated (p < 0.05). This is consistent with previous studies, which found that landfills with high proportions of ash had leachate with lower alkalinity relative to landfills with a low proportion of ash (Moody and Townsend, 2017). The measures of oxygen demanding components in MSWA leachate were also significantly lower (med: 120 mg L⁻¹ for TOC, 53 mg L⁻¹ for BOD, 1100 mg L⁻¹ for COD, 200 mg L⁻¹ for ammonia) as compared with the C&D, MSW, and GC landfill leachates, likely due to the lower levels of organic waste in MSWA (p < 0.05). More details about the mean, median, and standard deviation for each liquid type are provided in the supplemental text (Tables S–7, Tables S–8, and Tables S–9). PFAS levels in MSWA leachates (Fig. 2) had significantly lower levels of Σ₂₆PFAS (med: 3600 ng L⁻¹), Σ₈PFAA precursors (med: 120 ng L⁻¹), and ΣPFOA + PFOS (med: 480 ng L⁻¹) among the C&D, MSW, and GC landfill leachates, but the levels were significantly higher than for groundwater and stormwater (p < 0.05).

When considering the pure waste types, the MSWA samples collected from pure ash landfill cells had lower TOC (med: 76 mg L⁻¹), ammonia (med: 83 mg L⁻¹), and Σ₂₆PFAS (med: 3100 mg L⁻¹) levels than the MSWA samples collected from ash landfill cells co-disposed with MSW and C&D waste (TOC med: 160 mg L⁻¹, ammonia med: 240 mg L⁻¹, Σ₂₆PFAS med: 9600 ng L⁻¹). However, differences were likely insignificant due to the small sample size (n = 7 for pure MSWA and n = 10 for co-disposed MSWA) once broken down by subcategory. Previous studies found that higher ash content in landfill leachate was associated with lower PFAS concentrations. The leachate from the ash-only landfills had lower PFAS concentrations than the MSW landfill leachate (Liu et al., 2022b; Solo-Gabriele et al., 2020). Also, Liu et al. (2021b) found that the short-chain PFASs, including perfluorobutyric acid (PFBA) and perfluorobutane sulfonate (PFBS), were the primary PFASs in ash leachate samples. The lower levels of PFAA precursors and long-chain PFAS in MSWA leachate suggest that these species had been lost to the atmosphere or degraded to short-chain PFAS during incineration.

In terms of the correlations between individual PFAS measurements with physical-chemical parameters in MSWA leachate (Fig. 3, Figure S-1), results showed that consistently the Σ₂₆PFAS were significantly correlated with alkalinity, TOC, and ammonia. Liu et al. (2022b) found that co-disposed ash and unburned waste usually had higher PFAS along with higher alkalinity, TOC, and ammonia than landfills containing only ash. The TOC levels between this study (3.0–860 mg L⁻¹) and the Liu et al. study (24–620 mg L⁻¹) were similar. In addition, in the current study, the TOC levels were lower in the MSWA samples collected from pure ash landfill cells (med: 76 mg L⁻¹) than in the samples collected from the ash landfill cells co-disposed with MSW and C&D waste (med: 160 mg L⁻¹). So we speculate that the correlation observed for MSWA was caused by the co-disposal of unburned waste with ash in the MSWA landfills. Among these physical-chemical parameters, alkalinity had the strongest correlation with Spearman coefficient values of 0.93 and 0.95 for Σ₂₆PFAS and Σ₈PFAA precursors, respectively, with the Σ₈PFAA precursors driving the correlation in Σ₂₆PFAS. Among the metals evaluated, one of the major ions, magnesium, exhibited strong correlations with most PFAS, with values (r_s ≥ 0.70) with Σ₂₆PFAS and Σ₈PFAA precursors. The minor ions arsenic, chromium, nickel, and vanadium also exhibited strong correlations with most PFAS (r_s > 0.60). In terms of the multiple linear regression analysis, alkalinity, and ammonia were significant independent variables for Σ₂₆PFAS, Σ₈PFAA precursors, and ΣPFOA + PFOS (Tables S–10).

Fig. 3.

PFAS correlations matrix for MSWA (municipal solid waste incineration ash), C&D (construction and demolition wastes), MSW (municipal solid waste), Treated, GC (gas condensate), and all leachate with different physical-chemical parameters. The colored matrix represents the two corresponding parameters that were significantly correlated (p < 0.05) with the Spearman correlation coefficients shown within each colored matrix. Blanks within the matrix indicate that the two corresponding parameters were not significantly correlated (p > 0.05).

3.2. Construction and demolition leachate

In terms of physical-chemical parameters, C&D leachate (Fig. 1) had the lowest measurements of specific conductivity (med: 3600 mg L⁻¹), TDS (med: 2200 mg L⁻¹), TS (med: 2500 mg L⁻¹), and major metal ions (med: 220 mg L⁻¹ for Na, 130 mg L⁻¹ for K, 3.0 mg L⁻¹ for Fe) among the MSWA, MSW and GC liquids (p < 0.05). For PFAS (Fig. 2), the C&D leachate’s level of Σ₂₆PFAS (med: 8100 ng L⁻¹), Σ₈PFAA precursors (med: 560 ng L⁻¹), and ΣPFOA + PFOS (med: 1200 ng L⁻¹) were significantly higher than MSWA leachate (p < 0.05). Results confirm that C&D materials contain and leach PFAS, and these results are consistent with components of C&D wastes such as carpet (Lang et al., 2016) and cardboard (Death et al., 2021) known to leach PFAS.

In terms of the correlations between individual PFAS measurements and the physical-chemical parameter levels in C&D leachate (Fig. 3, Figure S-2), results showed that Σ₂₆PFAS had strong and significant associations with specific conductivity, TOC, and ammonia at coefficients generally close to or above 0.90. Among these physical-chemical parameters, ammonia consistently showed the strongest correlation with different groups of PFAS. The correlation coefficients between ammonia with Σ₂₆PFAS, Σ₈PFAA precursors, and ΣPFOA + PFOS were all above 0.97 (r_s > 0.97). The range of ammonia concentrations in nine collected C&D leachate samples was between 1 mg L⁻¹ to 730 mg L⁻¹, with an average level of 270 mg L⁻¹. These strong correlations between PFAS and ammonia for C&D leachate samples again suggest (as for MSWA landfill leachate) that some sampled C&D landfills are not composed of 100% C&D waste and could have received commingled MSW waste, though all C&D landfill cells were reported to contain only pure C&D waste. Among the metals evaluated, the major ions sodium and potassium exhibited strong correlations with most PFAS (r_s > 0.90). The minor ions cobalt and nickel also exhibited strong correlations with most PFAS (r_s > 0.70). In terms of the multiple linear regression analysis, the independent parameters that described Σ₂₆PFAS and Σ₈PFAA precursors were dependent upon specific conductivity, and for ΣPFOA + PFOS, it was dependent on ammonia (Tables S–10).

3.3. Municipal solid waste leachate

For MSW leachate, specific conductivity (med: 7400 mg L⁻¹), TDS (med: 3600 mg L⁻¹), TS (med: 4100 mg L⁻¹), and major metal ions (med: 920 mg L⁻¹ for Na, 310 mg L⁻¹ for K) were significantly lower in MSW compared to MSWA leachate but were significantly higher than C&D leachate (p < 0.05). In addition, the MSW samples collected from pure MSW waste landfill cells had lower TOC (med: 270 mg L⁻¹) and ammonia levels (med: 360 mg L⁻¹) than the MSW samples collected from cells co-disposed with MSWA and C&D waste (TOC median = 480 mg L⁻¹, ammonia median = 620 mg L⁻¹). The differences were significant for both parameters (p < 0.05), which was the opposite of what was expected. The reason for this observation has yet to be discovered but may be associated with mixing waste types.

In terms of PFAS levels, theΣ₂₆PFAS of pure MSW leachate (med: 6200 ng L⁻¹) was also lower than theΣ₂₆PFAS of co-disposed MSW leachate (med: 9000 ng L⁻¹). In this case, the differences were not significant (p = 0.34). When comparing the MSW leachate against other leachate types, the MSW leachate (for all MSW samples, pure and co-disposed) had significantly higher levels of Σ₂₆PFAS (med: 8400 ng L⁻¹), Σ₈PFAA precursors (med: 2400 ng L⁻¹), and ΣPFOA + PFOS (med: 1100 ng L⁻¹) among the aqueous landfill samples (p < 0.05) except the GC leachate (Fig. 2).

For the correlations between individual PFAS measurements and the physical-chemical parameter levels (Fig. 3, Figure S-3), results for MSW leachate showed that consistently Σ₂₆PFAS were significantly correlated with specific conductivity, alkalinity, TS, TOC, and COD at coefficients generally close to or above 0.70. Comparing sample sizes among leachate categories, the MSW leachate group had a larger sample size (n_s = 67) compared to MSWA leachate (n_s = 17) and C&D leachate (n_s = 9). In addition, the concentrations of PFAS and physical-chemical parameters in MSW represented a wider range which could have led to lower correlation coefficients for MSW leachate compared to MSWA and C&D leachates. Among the metals evaluated, major ions sodium and potassium were correlated with some PFAS, especially PFAS precursors, at coefficients generally above 0.60 (r_s > 0.60). The minor ions antimony, arsenic, chromium, cobalt, nickel, and vanadium also showed significant correlations (r_s > 0.60), especially for the PFAS precursors. For the multiple linear regression analysis, alkalinity plus parameters describing oxygen demand were the parameters that were most significantly correlated with PFAS levels (Tables S–10). For Σ₂₆PFAS, alkalinity, BOD, and COD were significant independent variables. For Σ₈PFAA precursors, alkalinity and BOD were significant independent variables. For ΣPFOA + PFOS, BOD and COD were significant independent variables.

3.4. Treated leachate

Among the 41 treated leachate samples collected, results showed (Fig. 1) that treated leachate had significantly lower levels of alkalinity (med: 860 mg L⁻¹) and significantly lower levels of oxygen demanding components which include TOC (med: 160 mg L⁻¹), BOD (med: 48 mg L⁻¹), COD (med: 680 mg L⁻¹), and ammonia (med: 95 mg L⁻¹) than MSW leachate (p < 0.05). In addition, some major and minor metal ions (med: 140 mg L⁻¹ for K, 33 mg L⁻¹ for Mg, 0.079 mg L⁻¹ for Ba) were also significantly lower than MSW landfill samples (p < 0.05). For the PFAS level measurements (Fig. 2), treated leachates had significantly lower levels of Σ₂₆PFAS (med: 5700 ng L⁻¹), Σ₈PFAA precursors (med: 310 ng L⁻¹), and ΣPFOA + PFOS (med: 790 ng L⁻¹) than C&D leachate, MSW leachate, and gas condensate (p < 0.05). However, the levels of PFAS in treated leachates were significantly higher in comparison to MSWA leachate, groundwater, and stormwater (p < 0.05).

For the correlations between individual PFAS measurements and the physical-chemical parameter levels (Fig. 3, Figure S-4), treated leachate sample results showed that most physical-chemical parameters were significantly correlated with different groups of PFAS. For example, the correlation coefficients between Σ₂₆PFAS with specific conductivity, alkalinity, and TOC were close to 0.90. Among the metals evaluated, the major ions sodium, potassium, and iron exhibited strong correlations with most PFAS at coefficients generally above 0.80 (r_s > 0.80). Moreover, the minor ion cobalt exhibited strong correlations with all PFAS (r_s > 0.80). In terms of the multiple linear regression analysis for all treated leachate samples, some bulk measurements and oxygen demand components were significantly correlated with PFAS levels (Tables S–10). For Σ₂₆ PFAS, TOC, alkalinity, ammonia, and specific conductivity were significant independent variables. For Σ₈PFAA precursors, alkalinity and ammonia were significant independent variables. For ΣPFOA +PFOS, TOC and pH were significant independent variables. For treated leachate, the extreme values associated with samples from effective treatment systems (e.g., reverse osmosis treatment) could have significantly influenced the overall correlations. Since such treatment can significantly decrease the PFAS and physical-chemical parameters concentration from the sample, the multiple linear regression was rerun, excluding reverse osmosis samples. Excluding the samples treated using reverse osmosis, the alkalinity was again found to be significantly correlated with all three groups of PFAS (Figure S-4). In addition, Σ₈PFAA precursors were dependent upon ammonia, as observed when all data points were considered.

3.5. Other aqueous landfill samples

Other aqueous landfill samples including gas condensate, groundwater, and stormwater were collected and analyzed in this project. Gas condensate is water vapor formed in landfill gas collection systems when moisture in landfill gas condenses at ambient air temperature. PFAS concentrations detected in gas condensate samples were similar to those reported elsewhere (Smallwood et al., 2023). In terms of physical-chemical parameters and PFAS concentrations, gas condensates had similar measurements compared with the leachate samples (Figs. 1 and 2). Gas condensates had significantly higher levels of oxygen-demanding components (380 mg L⁻¹ for TOC, 280 mg L⁻¹ for BOD, 1700 mg L⁻¹ for COD, and 930 mg L⁻¹ for ammonia) and PFAS levels (10,000 ng L⁻¹ for Σ₂₆PFAS, 2400 ng L⁻¹ for Σ₈PFAA precursors, 1200 ng L⁻¹ for ΣPFOA + PFOS) among the aqueous landfill samples (p < 0.05). Currently, the management of gas condensate continues to be a challenge at landfill facilities since the options for proper disposal are limited, and the common management practice is to combine gas condensate with leachates (Smallwood et al., 2023). The significant concentrations of PFAS in gas condensate and potential toxicity characteristics in the hydrocarbon phase (US EPA, Briggs, 1988) suggest that separate management for gas condensate might help reduce overall risks in the environment. In contrast, groundwater and stormwater samples had significantly lower physical-chemical parameter measurements and PFAS levels than landfill leachate samples (p < 0.05). In general, PFAS levels in stormwater samples (530 ng L⁻¹ for Σ₂₆PFAS, 8.0 ng L⁻¹ for Σ₈PFAA precursors, 56 ng L⁻¹ for ΣPFOA + PFOS) were significantly higher than in groundwater samples (120 ng L⁻¹ for Σ₂₆PFAS, 8.0 ng L⁻¹ for Σ₈PFAA precursors, 21 ng L⁻¹ for ΣPFOA + PFOS) (p < 0.05). The higher PFAS level in landfill stormwater samples suggests that on-site stormwater can be contaminated through various mechanisms, including wind-blown debris, aerial deposition, and direct runoff from the landfill. First, landfills can generate wind-blown debris containing PFAS, such as food packaging, paper boxes, and other materials. This debris can be carried by the wind and deposited in nearby stormwater, polluting the water. Also, the landfill gas (containing PFAS) may dissolve in the rainwater or condense on stormwater pond surfaces. Finally, stormwater runoff from the landfill itself can also enter the stormwater system and pollute the water. PFAS was also detected in some on-site groundwater wells. Potential sources of PFAS in groundwater may include leachate or runoff that infiltrates. Although the landfills were completely lined and equipped with a leachate collection system at the bottom, they could still produce runoff during heavy rain events. This runoff can carry PFAS contaminants from the landfill towards nearby surface waters or can infiltrate into the ground and contaminate groundwater. To avoid PFAS leakage from landfills, stormwater collection systems should be managed to control heavy runoff. Also, older closed landfills sometimes do not have bottom liners or may have liners that fail. In this situation, groundwater can be directly contaminated by PFAS from landfill waste.

For the correlations between individual PFAS measurements and the physical-chemical parameter levels in the other aqueous landfill samples (Fig. 3, Figure S-5), gas condensate samples showed the highest correlation coefficients between Σ₂₆ PFAS and TOC (r_s = 0.89). Among the metals evaluated, correlations were strong (generally above 0.80) with the major ions iron and magnesium. Moreover, except for antimony and copper, most minor ions showed strong correlations with PFAS at coefficients generally above 0.60. In terms of the multiple linear regression analysis for gas condensate, TOC was shown again as the significant independent variable (Tables S–10).

To further evaluate correlations with TOC, the 13 gas condensate samples were further classified into different categories that included condensates from sump pumps (n_s = 7), condensates from flare stations (n_s = 3), and condensates from pump stations (n_s = 3). Condensates collected from the flare stations were considered pure gas condensate because they were removed from landfill leachates. All other condensate samples could be mixed with leachate. The average TOC level in gas condensates collected from sump pumps, flare stations, and pump stations were 420 mg L⁻¹, 450 mg L⁻¹, and 910 mg L⁻¹, respectively and the average Σ₂₆ PFAS in gas condensates collected from sump pumps, flare stations, and pump stations were 14,000 ng L⁻¹, 14,000 ng L⁻¹, and 18,000 ng L⁻¹, respectively. Based on these results between the average TOC and Σ26 PFAS in the three gas condensate categories, the TOC and Σ₂₆ PFAS were positively correlated.

4. Discussion

In this study, results show that PFAS concentrations were associated with environmental ambient conditions and characteristics of the different types of wastes. In terms of the environmental ambient conditions, results suggest that rainfall and stormwater dilute PFAS in leachates. Correlations were observed with sodium and potassium, which can be considered conservative ions that indicate dilution effects for the leachate. Also, a strong correlation (r_s > 0.70) was observed between Σ₂₆PFAS and ammonia and TOC. The correlation between ammonia and TOC could be an indication of the commingling of MSW within the MSWA and C&D landfills. MSW is known to generate ammonia in landfills due to the decomposition of proteins from food waste (Liao et al., 2014). Alkalinity can be attributed to elevated levels of lime and calcium carbonate in MSWA. It also can be attributed to the biodegradation processes of organic matter in MSW which result in carbon dioxide production. This carbon dioxide can be absorbed into the leachate and increase the levels of carbonates, which could result in higher alkalinities. The correlation with alkalinity could have been further influenced by the precipitation of calcium carbonates when leachate is exposed to the atmosphere. Specifically, the loss of carbon dioxide from the leachate when in contact with the atmosphere can increase pH and drive the loss of carbonate through the precipitation of CaCO₃.

Spearman correlations were higher for MSWA and C&D leachates as compared to MSW leachates. More variability was observed among MSWA and C&D leachates which in turn allowed for higher correlation coefficients compared to MSW leachates. The larger variability was likely due to varying proportions of MSW, as small contributions of MSW to MSWA or to C&D can cause a significant change in the PFAS released. Overall, the results of the study can be used to recognize landfill aqueous samples that have high PFAS levels. For MSWA leachate, this could correspond to high levels of alkalinity, TOC, and ammonia. For C&D leachate, this could correspond to high levels of conductivity, TOC, ammonia, potassium, and sodium. For MSW leachate, this could correspond to high conductivity, alkalinity, COD, and ammonia.

Different types of landfill leachates can contain PFAS derived from various sources. For example, MSW leachate comes from the decomposition of household and commercial waste in landfills. It can contain PFAS from consumer products, such as non-stick coatings on cookware, waterproofing on outdoor gear, and stain-resistant treatments on carpets and furniture. Industrial waste, such as electronic waste and chemical waste, can contain high levels of PFAS. Landfills that receive these types of waste can produce leachate that contains PFAS from these sources. Liu et al. (2022a) stated that the electroplating of chrome, nickel, cadmium, zinc, or lead was a common contributor as it related to PFAS use. The correlation between these metal ions and PFAS in the current study suggests that the manufactured products may be contributing to the PFAS in leachate. C&D waste includes materials such as insulation, roofing materials, and carpeting, which can contain PFAS. Landfills that receive C&D waste can produce leachate that contains PFAS from these materials. As a result, the sources of PFAS in leachate can vary depending on the type of waste that is disposed of in landfills.

In terms of the overall correlation analysis, all PFAS measurements had strong and significant relationships with alkalinity, TOC, and ammonia. Hepburn et al. (2019) evaluated the relationship between PFAS and typical landfill leachate indicators and concluded that there were positive correlations between Σ₁₄PFAS, PFOA, and PFCAs with ammonia and bicarbonate for aqueous samples near unlined MSW landfills. This finding is consistent with the current study regarding the relationship between PFAS with ammonia and alkalinity, especially in MSWA leachate. Previous studies have also documented significant correlations between pH and TOC for some PFAS in landfill leachate. Benskin et al. (2012) found that eight PFAS (eight of 24 PFAAs and PFAA-precursors) correlated significantly with pH in MSW landfill leachate, and Gallen et al. (2017) found that certain PFAS (PFOS, PFOA, PFHxA, PFHxS, PFHpA) were significantly correlated with pH and TOC in MSW and C&D landfill leachates. This study was consistent with these previous studies, as TOC was correlated strongly with MSWA, C&D, and MSW leachates. Although rare, studies have also shown correlations between BOD and COD with PFAS in landfill leachate. Solo-Gabriele et al. (2020) found weak and insignificant correlations between COD and PFAS in leachate collected from five different landfills, including MSWA, C&D, and MSW landfills. Pan et al. (2014) found significant correlations between specific conductivity and COD (R² = 0.62) with certain PFAS (especially for PFOS) in surface water. In terms of the treated leachate, Zhang et al. (2022) found that changes in conductivity, pH, alkalinity, and ammonia were associated with PFAS changes.

5. Conclusion

This study aimed to evaluate potential relationships between the physical-chemical properties (including bulk measurements, oxygen demanding components, and metals) and PFAS levels in different types of landfill aqueous samples. The overall results showed that PFAS had strong and significant relationships with alkalinity, TOC, and ammonia. In terms of the Spearman correlation coefficients, MSWA PFAS results exhibited the highest values for alkalinity, TOC, and ammonia; and C&D PFAS results exhibited the highest correlations with specific conductivity, TOC, and ammonia. MSW PFAS results showed the highest correlations with specific conductivity, alkalinity, TS, TOC, and COD. Treated leachate PFAS results showed the highest correlations with specific conductivity, alkalinity, and TOC. For gas condensates, TOC was the most significant variable. These results can be useful for identifying samples that have higher levels of PFAS.

The correlations observed are suggestive of the mechanisms that could be controlling PFAS levels in landfills. For example, strong correlations are believed to be indicative of the effects of rainfall dilution, possible degradation of organic materials in MSW, and the precipitation of calcium carbonate and other metals as leachate is exposed to the atmosphere. These processes can impact major ion levels and parameters indicative of organic demand, plus alkalinity and specific conductivity. Specifically, correlations with ammonia could be indicative of decomposing MSW, which probably indicate the co-disposal of MSW within MSWA. Waste products common in C&D facilities, which are known to have high PFAS, are also associated with increased TOC (e.g., cardboard and paper products) and could be responsible for the correlations observed. However, correlations do not mean causation. The higher PFAS in certain leachates are likely caused by many waste sources and chemical mechanisms.

Given the strength of the correlations with physical-chemical parameters, there is the possibility that physical-chemical parameters can be used to indicate relative PFAS levels in different types of landfill leachate. For example, alkalinity, TOC, and ammonia were significantly correlated with PFAS in landfill leachate. These physical-chemical parameters are easier to measure and can be used as triggers for the direct analyses of PFAS to evaluate if leachate is meeting specified limits. Since the underlying mechanisms governing the high correlations are not yet well understood, more research is recommended before the use of physical-chemical parameters as a crude proxy of relative PFAS levels.

HIGHLIGHTS.

• Aqueous samples from 39 landfill facilities in Florida were evaluated.

• PFAS and 35 physical-chemical parameters were analyzed.

• Alkalinity, TOC, and ammonia were correlated with PFAS.

• Some associations were dependent on leachate type.

• Associations may be due to dilution and chemical precipitation when released.

Data availability

Data will be made available on request.