In vitro and in vivo remediation of per- and polyfluoroalkyl substances by processed and amended clays and activated carbon in soil

Abstract

Remediation methods for soil contaminated with poly- and perfluoroalkyl substances (PFAS) are needed to prevent their leaching into drinking water sources and to protect living organisms in the surrounding environment. In this study, the efficacy of processed and amended clays and carbons as soil amendments to sequester PFAS and prevent leaching was assessed using PFAS-contaminated soil and validated using sensitive ecotoxicological bioassays. Four different soil matrices including quartz sand, clay loam soil, garden soil, and compost were spiked with 4 PFAS congeners (PFOA, PFOS, GenX, and PFBS) at 0.01–0.2 μg/mL and subjected to a 3-step extraction method to quantify the leachability of PFAS from each matrix. The multistep extraction method showed that PFAS leaching from soil was aligned with the total carbon content in soil, and the recovery was dependent on concentration of the PFAS. To prevent the leaching of PFAS, several sorbents including activated carbon (AC), calcium montmorillonite (CM), acid processed montmorillonite (APM), and organoclays modified with carnitine, choline, and chlorophyll were added to the four soil matrices at 0.5–4 % w/w, and PFAS was extracted using the LEAF method. Total PFAS bioavailability was reduced by 58–97 % by all sorbents in a dose-dependent manner, with AC being the most efficient sorbent with a reduction of 73–97 %. The water leachates and soil were tested for toxicity using an aquatic plant (Lemna minor) and a soil nematode (Caenorhabditis elegans), respectively, to validate the reduction in PFAS bioavailability. Growth parameters in both ecotoxicological models showed a dose-dependent reduction in toxicity with value-added growth promotion from the organoclays due to added nutrients. The kinetic studies at varying time intervals and varying pHs simulating acidic rain, fresh water, and brackish water suggested a stable sorption of PFAS on all sorbents that fit the pseudo-second-order for up to 21 days. Contaminated soil with higher than 0.1 μg/mL PFAS may require reapplication of soil amendments every 21 days. Overall, AC showed the highest sorption percentage of total PFAS from in vitro studies, while organoclays delivered higher protection in ecotoxicological models (in vivo). This study suggests that in situ immobilization with soil amendments can reduce PFAS leachates and their bioavailability to surrounding organisms. A combination of sorbents may facilitate the most effective remediation of complex soil matrices containing mixtures of PFAS and prevent leaching and uptake into plants.

1. Introduction

Per- and polyfluoroalkyl substances (PFAS) are comprised of fluorinated carbon chains attached to functional groups (such as carboxylic acids, sulfonic acids, and alcohols) and are often described in two groups (long- and short-chain). They have unique physicochemical characteristics due to the larger size of the fluorine atoms compared to the hydrogen atoms and the strength of carbon-fluorine bonds, which contribute to their high thermal, chemical, and biochemical stability (Ghisi et al., 2019; Kabiri et al., 2021). Their widespread use has caused them to be ubiquitously detected in the environment, leading to growing concerns about their negative effects on the health of humans and animals, and the environment (Teymourian et al., 2021). For example, PFAS are found in soil all over the world, and the concentration has reached mg/kg levels at some highly contaminated sites (Sormo et al., 2021; Wang et al., 2020a; Li et al., 2019). They can enter soil via multiple pathways including land application of biosolids, irrigation with PFAS-contaminated water, leaching from landfilled wastes, release from PFAS-containing aqueous film forming foam (AFFF) at fire-training facilities, and pesticide applications (Wang et al., 2020a; Sorengard et al., 2019a). Given the important role of soil matrices in influencing the environmental fate of hydrophobic organic contaminants, a better understanding of the transport of PFAS in soil-groundwater systems is required to develop effective remediation strategies (Li et al., 2019; Sima and Jaffe, 2021).

The most common approach for treating PFAS-contaminated soil currently relies on adsorption (stabilization and immobilization) technologies (Kabiri et al., 2021; Kah et al., 2021; Navarro et al., 2023; Melo et al., 2022). While activated carbon (AC) has been shown to exhibit sizable adsorption capacities for longer chain PFAS, the effectiveness of AC declines precipitously for shorter chain PFAS species and precursors (Liu et al., 2020), as well as for carboxylates compared to sulfonates with the same chain length (Barth et al., 2021). Few studies have investigated the sorption of PFAS to clay-based sorbents for remediation purposes and most have focused on long-chain PFOA or PFOS in laboratory batch experiments (Teymourian et al., 2021; Sörengård et al., 2020; Hearon et al., 2022). Additionally, the spent sorbents loaded with PFAS can potentially be disposed of at landfills, provided the sorbed contaminants remain sequestered and certain risk criteria are met. The results from laboratory batch studies have indicated the possible desorption of PFAS from carbonaceous sorbents. This phenomenon is more likely to occur for PFAS with shorter chain lengths at high pH, and in the presence of other PFAS or other anionic sorbates (Sorengard et al., 2019b). Our knowledge of the dynamics of PFAS adsorption processes on engineered clay-based sorbents is limited, and even less is known about their desorption processes (Kabiri et al., 2021; Kah et al., 2021). These factors are essential for determining the long-term efficacy of any in-situ remediation treatment.

The use of in situ PFAS adsorbents created from naturally abundant geogenic materials is a promising option due to their low costs and toxicity. Clay minerals, such as montmorillonite, are mostly hydrophilic and negatively charged; therefore they poorly adsorb the anionic and hydrophobic PFAS compounds (Melo et al., 2022). However, adding cationic surfactants and modifying the hydrophilic silicate surface are strategies that facilitate the effective uptake of organic contaminants by mineral-based adsorbents (Teymourian et al., 2021; Cai et al., 2022; Aly et al., 2019; Tan et al., 2022; Ateia et al., 2019; Li et al., 2023).

A better understanding of PFAS sorption to materials could assist in the development of in situ stabilization remediation technologies for soil. Therefore, the objectives of the study were to (i) determine the mobilization of long-chain and short-chain PFAS in 4 soil matrices using a series of established leaching tests, (ii) evaluate the sorption efficacy of 4 PFAS to different clay-based adsorbent materials and AC, (iii) assess the sorption-desorption kinetics of PFAS from spent sorbents, and (iv) validate the reduced PFAS toxicity in soil and water leachates using sensitive ecotoxicological models. Overall, this study aimed to assess the effectiveness of different sorbents in reducing the leachability of PFAS in complex soil matrices under a variety of leaching conditions.

2. Materials and methods

2.1. Analytical PFAS standards, soil, and sorbent materials

The target PFAS analytes, including perfluorooctanoic acid (PFOA), perfluorooctanesulfonic acid (PFOS), hexafluoropropylene oxide dimer acid (HFPO-DA, GenX), and perfluorobutane sulfonic acid (PFBS) (purity >98 %) were purchased from Sigma-Aldrich (St. Louis, MO). A PFAS mixture containing equal concentrations of the above 4 analytes was used in all studies. Three isotopically labeled internal standards (IS): ¹³C₄-PFOA, ¹³C₄-PFOS, and ¹³C₃-HFPO-DA (purity >99 %) were used (Wellington Laboratories, Guelph, ON).

Garden soil was obtained from The Scotts Miracle-Gro Company (Marysville, OH) contained compost, processed forest products, sphagnum peat moss, a wetting agent, and fertilizer with 0.09 % total nitrogen, 0.05 % available phosphate, and 0.07 % soluble potash. Compost was obtained from Foxfarm Soil & Fertilizer Company (Humboldt County, CA). It contained aged forest products, sphagnum peat moss, perlite, sandy loam, and fertilizer with 0.30 % total nitrogen, 0.45 % available phosphate, 0.05 % soluble potash, and 1.00 % calcium from fish emulsion, crab meal, shrimp meal, earthworm castings, bat guano, kelp meal, and oyster shell (Hearon et al., 2022; Hearon et al., 2021). Blank (clean) clay loam soil was purchased from LGC Standards (Middlesex, UK) and white quartz sand (−50 + 70 mesh) was purchased from Sigma Chemical Company (St. Louis, Missouri) (Herrera et al., 2004). All soils were air-dried at 105 ^◦C for 24 h, homogenized and sieved (< 2 mm). These soil matrices were chosen based on their different physicochemical properties (Table S1) to represent a wide range of soil.

Calcium montmorillonite (CM) clay obtained from BASF (Ludwig-shafen, Germany), has a generic formula of (Na,Ca)_0.3(Al, Mg)₂Si₄O₁₀(OH)₂ nH₂O and a cation exchange capacity equal to 89.2 cmol/kg (Grant and Phillips, 1998; Phillips et al., 2019; Wang et al., 2020b). Acid processed montmorillonite (APM) clay was synthesized based on a previously described method (Phillips and Wang, n.d.). The molecular structure of APM has been reported as a mixture of delaminated clay aggregates and amorphous and cross-linked silica (Wang et al., 2021a; Wang and Phillips, 2020; Wang et al., 2019a). Organoclays resulting from amendments of L-carnitine, choline, and chlorophyll at 100 % cation exchange capacity were previously developed in our laboratory (Rivenbark et al., 2022; Wang et al., 2019b; Wang et al., 2021b; Wang and Phillips, 2023). A virgin powdered activated carbon (AC) derived from a selected grade of coconut shell, purity >99 %, was obtained from General Carbon Corporation (Paterson, NJ). This carbon was selected from 6 studied carbons with various origins due to its high surface area and high carbon content.

All soil and sorbents were characterized for their surface area, pH at the point of zero charge (pHpzc), density, moisture, expansibility in water, hydrophobicity, and particle size (Tables S1 and S2). The external surface area was calculated by the Brunauer-Emmett-Teller (BET) with nitrogen absorption using a Micromeritics 3Flex Adsorption Analyzer. All samples were activated at 200 ^◦C for 4 h before assessing the absorption of nitrogen at 77 K. The total surface area was measured by the ethylene glycol monoethyl ether (EGME) method (Bu et al., 2019). Metal analysis of the soil was conducted by an inductively coupled plasma-mass spectrophotometer (ICP-MS) NexIon 300 (PerkinElmer) (Wang et al., 2021a). The background levels of PFAS in the soil was non-detectable (limit of detection at 0.1 ng/mL). Their analysis by SEM, XRD, and FTIR were previously published (Wang and Phillips, 2023; Wang et al., 2020c; Wang and Phillips, 2022; Wang et al., 2023).

2.2. PFAS treatment and adsorption in soil

PFAS mobility was tested with 1 g of each soil matrix (sand, clay loam soil, garden soil, and compost) spiked with the PFAS standard mixture at 0.01, 0.05, 0.1, and 0.2 μg/mL/PFAS in 0.25 mL of methanol and left in a fume hood overnight to allow the methanol to evaporate. Deionized water was applied at a liquid/solid (l/s) ratio equal to 2:1, and the wet soils were agitated at 50 rpm for 24 h (Sorengard et al., 2019a). This allowed the spiked PFAS to equilibrate with the wet soil matrix and to simulate the extraction of field-collected or aged soil samples. A 5 μL aliquot of 50 % NH₄OH/50 % methanol was added to all pre-cleaned polypropylene tubes to limit the loss of PFAS by evaporation. Leaching assessments using 3 consecutive extraction methods (described below) were conducted after 24 h of equilibration.

To assess PFAS adsorption by sorbents, the starting concentration was 0.1 μg/mL for each compound. This PFAS mixture in 0.5 mL of methanol was spiked in 1 g of the soil mixture containing equal weights of each soil matrix and left in a chemical fume hood overnight for evaporation of methanol. Deionized water was applied at an l/s ratio of 10, and the sorbent treatments were employed at 0, 0.5 %, 1 %, 2 %, and 4 % w/w dry weight. The soils in polypropylene centrifuge tubes were mixed on an end-over-end rocker at 50 rpm for 7 days. During this incubation period, soil was sealed in air-tight tubes to maintain sample moisture conditions. Incubation conditions were stable in an indoor environment with no direct exposure to sun or extreme weather conditions. The soil not treated with sorbents was subjected to the same conditions. This standard batch leaching test was performed to simulate aggressive, worst-case scenario conditions for leaching (i.e., shaking for one week at a high liquid/solid ratio) and to test the longevity and stabilization of sorbent treatments in soil. LEAF water method was used to extract PFAS after sorbent treatments.

For PFAS mobilization from soil at different pHs, deionized water at pH 4, 7, and 8.4 (simulating acid rain, fresh water, and brackish water, respectively) were added to the soil mixture containing 0.1 μg/mL PFAS and 1 % individual sorbent and mixed for 7 days. In the time course study, soils containing 0.1 μg/mL PFAS and 1 % individual sorbent were agitated for 2, 4, 7, 14, and 21 days to measure the adsorption kinetics and rate of dissociation. All other conditions remained the same. The experimental results of sorption kinetics were analyzed by fitting the data to the pseudo-first-order, pseudo-second-order, and Elovich kinetic models. All models and parameters are presented in the Supplementary Material (Table S3).

2.3. Leaching tests

The leachability assessment was conducted in the first PFAS mobility study using 3 different test protocols, which were selected to represent a range of scenarios, including short- and long-term leaching potentials. First, each soil matrix spiked with PFAS for 24 h was left to settle for filtration through 1.2 μm glass microfiber filters (Whatman, grade GF/C) (Sormo et al., 2021). These filtrates were referred to as the water leachate and were analyzed for PFAS by LC-MS/MS with the addition of 0.25 μM IS.

Afterward, soil residues were subjected to a methanol extraction (Hearon et al., 2022; Zhang and Liang, 2022a). Briefly, 2 mL of methanol with 1 % NH₄OH was added to each 1 g soil. The mixture was sonicated in an ice bath for 30 min, vortexed for 30 min at 70 rpm, and centrifuged at 4000 rpm for 10 min. The above extraction steps were repeated twice, and the extracts were combined (~5 mL) and cleaned by solid-phase extraction using a Strata C18-E (55 μm, 70 Å) column. Briefly, SPE cartridges were conditioned with 2 mL of methanol and water. Then, 1 mL of each sample was acidified with 7 μL of concentrated phosphoric acid and was loaded into the cartridge. The acidification step is necessary as the optimum working pH range of the cartridge is pH 6–7. The cartridge was then rinsed with 1 mL of 5 % methanol in water. Samples were eluted using 2 mL of 50 % acetonitrile/50 % methanol at 1–2 mL/min. All the samples were then placed under a gentle stream of nitrogen gas until dry, and resuspended in 250 μL of 5 % NH₄OH in 60 % acetonitrile/40 % methanol. ¹³C-PFAS IS were added to each vial at 0.25 μM. This was referred to as the methanol extract.

Finally, the desorption behavior of PFAS from treated and untreated soil was examined using the U.S. EPA LEAF Method 1313 (Leaching Environmental Assessment Framework) (Sorengard et al., 2019a; Sörengård et al., 2022) with modifications. Briefly, the remaining soil following the water and methanol extraction in the PFAS mobility study was added to water at an l/s ratio of 10 in a polypropylene tube. Soils were agitated at 50 rpm for 7 days to equilibrate and then centrifuged at 3000 rpm for 15 min. The supernatant was filtered through cellulose syringe filters (0.45 μm, Sartorius) into 2.0 mL auto-injector glass vials (Eppendorf, Germany) and then analyzed for PFAS after the addition of ¹³C-PFAS IS. This was referred to as the LEAF water leachate.

2.4. PFAS separation and analysis

Elutes containing 4 target PFAS were analyzed by liquid chromatography tandem mass spectrometry (LC-MS/MS) with an Acquity BEH C18 column (2.1 × 50 mm, 1.7 μm) at 40 °C (Hearon et al., 2022; Wang et al., 2021b). Quantitation was based on recoveries of 3 isotopically labeled IS. A gradient elution using 20 mM ammonium acetate (eluent A) and acetonitrile (eluent B) was carried out at a flow rate of 0.6 mL/min. The gradient program for elution was 10 % eluent B (initial), 10 %–55 % (0 to 0.1 min), 55 %–99 % (0.1 to 4.5 min), 99 % (4.5 to 5 min), and 99 %–10 % (5 to 6.5 min). The injection volume was 10 μL for each analysis. The mass spectrometer was used with an electrospray ionization interface (ESI) and operated in a negative ion mode. The spray voltage was maintained at 4.5 kV. The source temperature was kept at 450 ^◦C. The monitored precursor and product ions (m/z) for PFOA, PFOS, GenX, and PFBS were 413 to 369, 499 to 80, 285 to 168.9, and 298.9 to 80, respectively. The cone voltage (mV) for PFOA, PFOS, GenX, and PFBS was 20, 40, 45, and 40, respectively. The mass spectrometer was operated under the multiple reaction monitoring (MRM) mode. Nitrogen gas was used as the collision and curtain gas, and argon gas was used as the nebulizer and heater gas. Empower analyst software was used to control the LC/MS-MS system and acquire the data. A 7-point calibration curve ranging from 0.1 to 1000 ng/mL PFAS in water showed good linearity and accuracy with R² > 0.99, limit of detection at 0.1 ng/mL, and recovery at 86 %–105 % calculated from 0.25 μM IS for each PFAS. This validated the PFAS extraction methods and detection methods on LC-MS/MS, and the PFAS standards were run with each sample analysis.

PFAS that was leached in the mobility study was reported as mass (μg) and recovery percentage (%). The treatment efficacy of sorbents was reported as percentage reduction (%) in leaching from treated soil relative to untreated soil. Triplicate (n = 3) samples were used for the analysis of initial spiked soil concentrations and laboratory blanks.

2.5. Lemna minor

To validate the aqueous bioavailability of PFAS, filtrates after the LEAF water treatment from sorbent-treated and untreated soil were evaporated under nitrogen gas and reconstituted in 2 mL of Lemna growth media. They were exposed to an aquatic floating plant (L. minor) that has been shown to absorb PFAS contaminants and thus is susceptible to PFAS toxicity in water (Zhang and Liang, 2022b).

Lemna minor (duckweed) was procured from AquaHabit (Chatham, UK) and cultured under cool white, fluorescent lights (400 ft-c intensity) with a light-to-dark cycle of 16 h/8 h at 25 ^◦C (Wang and Phillips, 2023; Wang et al., 2022). Three lemna plants with 3 fronds each were randomly selected and exposed to LEAF water leachates from soil for 7 days. Lemna was inspected daily for changes in frond number and surface area of surviving plants, which were analyzed by Image J (NIH, Bethesda, MD, USA). On Day 7, the plants were collected from individual dishes and homogenized in 1.5 mL of 80 % acetonitrile. The chlorophyll content was extracted after 48 h (4 ^◦C, dark) and quantified using UV/Visible scanning spectrophotometry (Shimadzu UV-1800, Kyoto, Japan) at 663 nm. The growth rate and inhibition percentage were calculated following standard OECD guidelines (Organisation for Economic Co-operation and Development, 2004) (Table S3).

2.6. Caenorhabditis elegans

Nematodes are sensitive indicators of PFAS toxicity and were used in this study as living models to investigate the bioavailability of PFAS in soil (and pore water) after the LEAF water extraction (Qi et al., 2022). Wildtype (Bristol N2) C. elegans and E. coli strains NA22 and OP50–1 were purchased from the Caenorhabditis Genetics Center (CGC, University of Minnesota). Nematodes were grown on 8P media seeded with 8 × 10⁸ cells/mL of E. coli NA22 and were maintained at 20 ^◦C (Wang et al., 2021c). Age-synchronized L1 nematodes were obtained by a bleaching and washing process. Egg culture solutions were rocked on a rocking platform (VWR, Radnor, PA) at 2.5 rpm in the dark for 18 h at 20 ^◦C.

After the incubation period, 100 nematodes were counted and transferred to a Petri dish, where they were exposed to 0.3 g of treated and untreated soil for 72 h. Soil was kept in the dish with a closed lid and moistened daily with K media. Afterward, C. elegans were separated from soil using the soil-agar isolation method, where soil was transferred to the center of an NGM agar plate seeded with strips of E. coli OP50–1 on each side (Kim et al., 2015). The percentage of recovered nematodes in each group was measured by counting the number of nematodes attracted to the E. coli strips, indicating the survival rate. Locomotion of 10 nematodes in each group was measured using the nose touch assay under an Olympus SZ61 zoom stereomicroscope (Olympus, Waltham, MA). Then, the separated nematodes were transferred to a new agar plate and left to mature at 20 ^◦C for 24 h. The body length of individual nematodes was measured using the CellSens Entry (standard version 3) software (Olympus, Waltham MA) after nematodes were paralyzed with 25 mM sodium azide. Relative body lengths were calculated as a percentage of the medium control group, which was adjusted to 100 %.

For reference, the sensitivity of C. elegans to PFAS was confirmed by exposing 5000 L1 nematodes to PFAS in aqueous K-media complete solution at 1–75 μg/mL for 2 days (Wang et al., 2021c; Organisation for Economic Co-operation and Development, 2006). The same toxicity endpoints were measured as mentioned above.

2.7. Statistical analysis

As part of QA/QC, blank samples consisting of the leaching solution used (without soil) were prepared with every leaching test. All leaching tests were conducted in triplicate. Samples (leachates) were kept at 4 ^◦C prior to PFAS analysis or exposure to bioassays. One-way ANOVA and Tukey’s Test post-hoc analysis were used to assess means between different treatment groups. The level of significance was set at p ≤ 0.05.

3. Results and discussion

3.1. PFAS recovery in soil

To investigate the distribution of PFAS in soil, each soil matrix was spiked with 4 PFAS congeners at 4 concentrations and subjected to 3 continuous extractions using water, methanol, and LEAF water methods. The recovery of each PFAS from a combination of the 3 extractions is shown in Fig. S1. The results showed a concentration-dependent trend, where the total amount extracted increased as the spike concentration increased for each PFAS in all soil matrices.

In the absence of sorbent treatment, a significant proportion of PFAS in the soil, especially in the clay loam soil and sand, was easily leachable in water (dark blue bars in Fig. S1). PFAS leaching in water generally increased with decreasing perfluoroalkyl chain lengths, such as GenX (log Kow = 1.3) and PFBS (log Kow = 1.8), due to their higher polarity and mobility in water. PFBS recovery was calculated based on ¹³C-PFOS due to a lack of the respective IS and possibly resulted in higher than 100 % recovery. Leaching of PFOA (containing a carboxylate head group) tended to be higher than PFOS (containing a sulfonate head group) in water, leading to PFOS being mostly retained in the soil until extracted by methanol (orange bars in Fig. S1). Most PFAS were extracted from soil either by water (e.g., GenX, PFBS, and PFOA) or methanol (e.g., PFOS), except from the compost. Compost can hold PFAS tightly compared to other soils, so most PFAS were retained and only extracted by the LEAF method at the end (light blue bars in Fig. S1). This method uses a high amount of water for 7 days. These results show that PFAS sorption and leaching behavior in solid matrices is dependent on their surface activity, including the chain length of their tail group and their functional head group type. These trends are consistent with the literature (Sima and Jaffe, 2021; Askeland et al., 2020). Because water is an efficient extraction method for most PFAS (except PFOS) and is the most realistic and environmentally relevant solution, only the LEAF water leaching method was used in the following sorbent remediation studies.

The overall recovery of each compound was determined by combining the recoveries from the 3 extractions of PFAS at 4 concentrations to study the effect of soil types (Fig. 1). The recovery occurred in the order of quartz sand > clay loam soil > compost ≥ garden soil for PFAS. This difference in recovery indicated that PFAS sorption and leachability in different soil types was in alignment with the organic carbon content (Table S1) and suggested competitive sorption and interference by soil-based organic substances. PFAS sorption in soil can be affected by many complex factors, but it is dependent on the combination of specific PFAS congeners and soil geochemical properties (Askeland et al., 2020).

Fig. 1.

Recovery of PFAS leachates from sand, clay loam soil, compost, and garden soil. Each PFAS was sequentially extracted by water, methanol, and LEAF water methods, and the total recovery was averaged among 4 spiked concentrations (0.01, 0.05, 0.1, 0.2 μg/mL) of PFAS incubated for 24 h in each matrix.

3.2. PFAS sorption

A soil mixture containing an equal weight of garden soil, compost, clay loam soil, and quartz sand was used to represent a realistic and complex soil matrix in the environment to test the application of soil amendments. Figs. 2 and 3 show the percent reduction of each PFAS and mixture of PFAS (ΣPFAS) in the leachates after treatment compared to leachates of the untreated soil, respectively. Following the amendment of soil with various sorbents at 4 doses, a reduction in the concentration of PFAS leachates was observed at all treatment levels, even the 0.5 % rate. In most cases, a dose-dependent binding was shown as the inclusion of 2 % and 4 % sorbents that reduced PFAS leaching more than 1 % and 0.5 % inclusions (Fig. 2).

Fig. 2.

Reduction of each PFAS congener by sorbents administered at 0.5 %, 1 %, 2 %, and 4 % in the soil mixture. PFAS leachates were extracted via the LEAF method using water.

Fig. 3.

Reduction of ΣPFAS by sorbents administered at 0.5 %, 1 %, 2 %, and 4 % in the soil mixture. PFAS leachates were extracted via the LEAF method using water.

Strong sorption of PFAS to AC-based materials has been reported in previous studies on contaminated soil, drinking water, and sewage sludge (Navarro et al., 2023; Sörengård et al., 2020). In this study, AC was the most effective sorbent in binding ΣPFAS with the removal rates ranging from 73 % - 97 % with 0.5 % and 4 % inclusions, respectively (Fig. 3). This AC was a medical grade, virgin powder derived from coconut shell, and it showed a higher efficacy compared to other carbons in the literature. For example, ash carbon showed a similar reduction of ΣPFAS at inclusion rates of 4 %, 10 %, and 25 % (Bu et al., 2019). Despite their high removal efficacy, the origin and specific surface area of carbonaceous materials are crucial besides other variable factors, and thus the standardization and stabilization of carbon sorption efficiency is challenging.

Montmorillonite clay-based sorbents were also tested as they have been demonstrated to be good sorbents for PFAS when modified with organic amendments. For example, retention of >99.3 % of PFOS has been reported in soil amended with modified montmorillonite at 10 % w/w (Sörengård et al., 2020). As shown in Fig. 2, all tested clays showed a significant reduction of each PFAS. Among them, montmorillonite amended with chlorophyll and montmorillonite processed by acid (APM) showed the highest sorption efficacy (Fig. 3). APM was synthesized to simulate an active carbon structure with high porosity, surface area, and heterogeneous binding sites (Wang and Phillips, 2020; Wang et al., 2019c; Wang and Phillips, 2019). It is positively charged at environmentally relevant conditions and therefore strong interactions through electrostatics are possibly involved in the adsorption of PFAS. Amending montmorillonite with organic compounds such as L-carnitine, choline, and chlorophyll increased the hydrophobicity, surface area, and zeta potential of the materials. These made them similar to other organoclays (Table S2), and thus enhanced the attraction of PFAS possibly through hydrophobic interactions. Despite being negatively charged (Table S2, pHpzc), these organoclays were able to reduce the leaching of PFAS, suggesting the importance of hydrophobic interactions in the sorption of PFAS particularly in the case of the negatively charged sorbents. Due to differences in binding mechanisms, combining sorbents to target the hydrophobic tail and hydrophilic head groups may be a good strategy to achieve higher performance for a wide range of PFAS.

3.3. Lemna minor

Previous studies showed that the Lemna minor (duckweed) assay is a sensitive toxicity indicator and has identified dose-dependent relationships for PFAS toxicity on vegetative growth in aqueous lemna growth media, LM (Hearon et al., 2022; Zhang and Liang, 2022b). In this study, we expanded on this traditional application to evaluate the sorbents’ efficacy in reducing PFAS in water leachates from soil. Filtrates after the LEAF water extraction from untreated and treated soil, were reconstituted in LM and exposed to the plant for 7 days. Compared to the LM blank, water leachates from 0.1 μg/mL PFAS spiked soil showed significant toxicity on plant growth with lower daily frond number and surface area (Fig. 4). Specifically, a decrease of 76 %, 36 %, and 78 % was shown on the frond number, surface area, and chlorophyll content on day 7, respectively. Leachates from each sorbent treatment at doses of 0.5 %, 1 %, 2 %, and 4 % were individually exposed to L. minor and the growth parameters for each sorbent were averaged from the 4 doses and are presented in Fig. 4. Results on leachates from all sorbent-treated soils showed increased growth in frond number, surface area, and chlorophyll content, indicating that all sorbents reduced PFAS toxicity in water leachates. This remediation was dose-dependent and sorbent inclusion at 4 % showed the highest growth that was similar to the media blank (Fig. S2). Interestingly, although AC showed the highest sorption percentage in Fig. 3, amended montmorillonite minerals showed higher growth in general. This has been shown previously and is possibly because these clay amendments added nutrients and promoted growth. For example, L-carnitine is important in fatty acid metabolism and cell homeostasis in plants and can promote cell division and plant growth (Hearon et al., 2022). Choline chloride can promote plant growth by increasing leaf numbers, chlorophyll levels, and sucrose content in the leaves (Zheng et al., 2016). This suggests that organic montmorillonite clays from natural sources containing GRAS materials could be used as value-added soil supplements in addition to PFAS remediation.

Fig. 4.

PFAS in LEAF water leachates as indicated by changes in L. minor frond number (A) and surface area of surviving plants (B) over 7 days, and chlorophyll content on day 7 (C). All data on sorbent groups were averaged from treatments at 4 doses (0.5 %, 1 %, 2 %, and 4 %) and showed reduced toxicity compared to the leachates from 0.1 μg/mL PFAS spiked soil (*p ≤ 0.05 compared to LM; ^#p ≤ 0.05 compared to PFAS for all 3 parameters).

3.4. Caenorhabditis elegans

Caenorhabditis elegans is a reliable toxicological model that is sensitive to a wide range of contaminants at environmentally relevant concentrations with established toxicity testing methods. The remaining soil after the LEAF water extraction were exposed to nematodes for 72 h before they were separated from the soil using the feed attraction method on NGM agar plates. Approximately 85–90 % of nematodes in the blank groups were recovered (Fig. 5), and only minimal (15–25 %) nematodes exposed to PFAS-spiked soil were recovered. Separated nematodes were tested by nose-touching for their response and measured for their relative body length to the blank control, which was adjusted to 100 %. PFAS-contaminated soil resulted in significantly decreased locomotion and body length, showing PFAS toxicity on these soil invertebrates. The toxicity of PFAS to nematodes has been previously studied and demonstrated in their metabolic, reproductive, and neurologic systems (Feng et al., 2022; T et al., 2023). For comparison, a PFAS toxicity test in water was run following the previous protocol (Wang et al., 2022; Wang et al., 2021c), and showed dose-dependent toxicity at μg/mL levels (Fig. S3). However, exposure to soil showed more severe toxicity at even lower PFAS levels than in water, possibly because C. elegans are more sensitive in their natural habitats and in the presence of other factors such as contaminants or organic matter in soil. This media-dependent sensitivity has been reported for C. elegans with other contaminants (Kim et al., 2020). Similar to results in L. minor, exposure to sorbent-treated soil protected C. elegans and increased their recovery rates, locomotion, and body length. Importantly, this protection was dose-dependent, and a 4 % inclusion of sorbents delivered significantly higher protection that was similar to the blank control, especially for the recovery rate. A higher dose with a longer treatment duration or a combination of sorbents may be required to further promote the well-being of nematodes in locomotion and body length. While AC showed the highest recovery rate, amended montmorillonite clays showed higher responses in locomotion and body length. It is possible that a mixture of clays and carbons will be needed as soil supplements to provide a variety of binding sites and activities for the remediation of PFAS in complex matrices. Such products have been reported (Melo et al., 2022; Sorengard et al., 2019b; Zhang and Liang, 2022a), including RemBind^®, PlumeStop^®, and Intraplex A^®, and have shown high efficacy in in vitro tests, while more comparison tests in real environmental matrices and validation in ecotoxicological models are needed before application.

Fig. 5.

PFAS in soil residues after extraction as indicated by C. elegans recovery % (A), locomotion (B), and relative body length (C). The results showed reduced toxicity in sorbent-treated soil compared to the PFAS-spiked soil.

3.5. Sorption kinetics

The storage and disposal of spent sorbents loaded with PFAS is feasible only if the PFAS remains sequestered for a long period and under a range of varying conditions. Therefore, PFAS-spiked soils were treated with 1 % sorbents for 7 days at pH 4, 7, and 8.4 which simulated acid rain, fresh water, and brackish water. PFAS were extracted using the LEAF water method and the results were averaged from the reduction of each PFAS congener. Fig. 6 shows the significantly reduced total PFAS leachates from soil treated with all sorbents, compared to untreated soil. Similar to the dosimetry study, AC, APM, and montmorillonite-chlorophyll clays consistently reduced PFAS the most effectively at 3 pH conditions. The reduction for all sorbents was relatively consistent at all pHs tested, indicating stable sorption. This is possibly related to the surface charge on sorbents that was not significantly altered based on their pHpzc (Table S2). It has been reported in the literature that increasing pH resulted in greater desorption and leaching of PFAS (Kah et al., 2021), but it only happended when pH values exceeded 9, which are rarely encountered in the natural environment (Kabiri et al., 2021). Overall, at pH 4 to 8.4, which is the normal range of pH values for most soil, all remediation agents resulted in a consistent reduction of all PFAS in leachates.

Fig. 6.

Reduction of ΣPFAS by 1 % sorbents in the soil mixture at pH 4, 7, and 8.4. PFAS leachates were extracted via the LEAF method using water.

A time course study was carried out by incubating treated and untreated soil for 21 days. PFAS were extracted by LEAF water treatment at intervals on days 2, 4, 7, 14, and 21, and the reduction of each PFAS congener was compared to the corresponding spiked soil. The results from the time course of adsorption are shown in Fig. 7. All sorbents showed a curved binding plot. Most of the reaction reached equilibrium within 2 days, suggesting a high affinity for sorption. Despite the fast reaction, dissociation of bound PFAS occurred with time. The sorption of PFAS on day 21 decreased by 14–33 % (GenX), 7–31 % (PFBS), 9–43 % (PFOA), and 12–35 % (PFOS) among all sorbents, compared to the highest sorption on day 2 or 4. All desorbed fractions were <50 %, suggesting that new sorbents should be reapplied every 21 days to prevent further dissociation (Askeland et al., 2020). All sorbents showed similar trends with minor differences, although AC and APM showed slightly higher sorption in general. Perfluorocarbon chain length showed no significant correlation (p > 0.05) with any of the sorbents.

Fig. 7.

Adsorption of GenX (A), PFBS (B), PFOA (C), and PFOS (D) onto 1 % sorbents in the soil mixture at pH 7 for a total of 21 days. PFAS leachates were extracted on days 2, 4, 7, 14, and 21 via the LEAF method using water.

The above time course results were analyzed by 3 common nonlinear kinetic models, including pseudo-first-order, pseudo-second-order, and Elovich (Table S3). Based on the adjusted correlation coefficient values (R² ≥ 0.85) and the comparison between the binding capacity derived from experiments (q_e, exp) and calculated from the models (q_e, cal), the adsorption of all 4 PFAS onto all sorbents best fit the pseudo-second-order kinetic model. Therefore, the parameters for the pseudo-second-order kinetic model are shown in Table 1. This suggested that the adsorption kinetics of PFAS were mainly dependent on diffusion-limited processes and affected by heterogeneous distributions of pore sizes and continual partitioning of PFAS. The driving force for adsorption was the concentration gradient, and therefore, the rate of adsorption was proportional to the driving force. Compared among the sorbents, AC constantly showed the highest capacity in sorption, especially for PFBS. The expected effect of adding sorbents as soil supplements will be to adsorb PFAS and possibly other contaminants in soil, thus reducing their bioavailability and uptake to plants.

Table 1.

Adsorption constants of each PFAS onto sorbents calculated by the nonlinear pseudo-second-order model and the experimental binding capacities.

	GenX				PFBS
	qe (exp)	qe (cal)	K2	${\text{R}}_{\text{adj}}^2$	qe (exp)	qe (cal)	K2	${\text{R}}_{\text{adj}}^2$
AC	0.95	0.81	1.0 × 104	0.99	0.92	0.91	5.8 × 10−1	0.94
CM	0.93	0.74	8.3 × 107	0.99	0.81	0.66	6.4 × 105	0.86
APM	0.96	0.74	6.2 × 107	0.90	0.9	0.74	4.2 × 107	0.99
CM-carnitine	0.90	0.72	8.4 × 107	0.94	0.9	0.62	1.7 × 108	0.85
CM-choline	0.96	0.65	2.0 × 108	0.91	0.83	0.65	7.7 × 107	0.89
CM-chlorophyll	0.93	0.75	8.5 × 107	0.99	0.9	0.64	9.3 × 107	0.99
	PFOA				PFOS
	qe (exp)	qe (cal)	K2	${\text{R}}_{\text{adj}}^2$	qe (exp)	qe (cal)	K2	${\text{R}}_{\text{adj}}^2$
AC	0.98	0.88	1.1 × 104	0.95	0.92	0.75	8.8 × 108	0.98
CM	0.96	0.85	6.7 × 10−1	0.96	0.89	0.69	9.7 × 107	0.93
APM	0.96	0.75	2.0 × 108	0.98	0.82	0.70	1.8 × 108	1
CM-carnitine	0.98	0.87	7.3 × 10−1	0.97	0.94	0.69	9.1 × 107	0.96
CM-choline	0.98	0.74	1.2 × 100	0.98	0.80	0.64	1.1 × 108	0.99
CM-chlorophyll	0.91	0.73	1.2 × 108	0.93	0.90	0.73	8.5 × 107	0.99

4. Conclusion

Immobilization of PFAS in soil can reduce the leaching of hazardous pollutants into the surrounding environment and be taken up by native organisms. In this study, the efficiency of sorbent amendments to immobilize PFAS in an impacted soil mixture was investigated by quantifying the reduction of PFAS in water leachates and validated with sensitive living organisms. Individual soil matrices including quartz sand, clay loam soil, garden soil, and compost were spiked with 4 common PFAS congeners (PFOA, PFOS, GenX, and PFBS) at 0.01–0.2 μg/mL each. PFAS leachates were extracted through a sequential 3-step extraction method that showed a high and dose-dependent recovery. Sorbents were added to a soil mixture that was spiked with 0.1 μg/mL/PFAS at 0.5–4 % and showed a dose-dependent reduction of each PFAS congener with activated carbon (AC). AC reduced 73–97 % of total PFAS in the LEAF water leachates. The effect of sorbents in reducing PFAS bioavailability was validated by exposing the water leachates to a floating plant (Lemna minor) and the remaining soil to a soil nematode (Caenorhabditis elegans). Both models showed sensitivity to PFAS toxicity by reduced growth response and recovery rates. Soil treated with sorbents showed a dose-dependent reduction in toxicity. Importantly, organoclays amended with nutrients including carnitine, choline, and chlorophyll not only protected the living organisms from PFAS toxicity but also promoted their growth, suggesting valued-added potential as soil amendments. Results of a kinetic study with varying pHs simulating acid rain, fresh water, and brackish water showed stable sorption of total PFAS. A time course study to test the duration of sorption showed PFAS dissociation up to 43 % on day 21. While good for short-term remediation, further work is warranted to enhance the stability of the surface complex. This might be possible with the addition of fungi and bacteria that can act symbiotically with the sorbents to consume the dissociated products.

This study has shown that carbon and clay-based sorbents can be used to reduce the leaching of PFAS from contaminated soil and also decrease their effects on living organisms. The results suggest that the addition of a mixture of diverse toxin binding materials could facilitate the sorption and inactivation of PFAS in soil due to physical and chemical differences in sorbents. Future laboratory work will benefit from large-scale tests on undisturbed soil in order to obtain more information about the suitability of the method for field application.

Data availability

Data will be made available on request.