Abstract
The presence of poly- and perfluoroalkyl substances (PFAS) has caused serious problems for drinking water supplies especially at intake locations close to PFAS manufacturing facilities, wastewater treatment plants (WWTPs), and sites where PFAS-containing firefighting foam was regularly used. Although monitoring is increasing, knowledge on PFAS occurrences particularly in municipal and industrial effluents is still relatively low. Even though the production of C8-based PFAS has been phased out, they are still being detected at many WWTPs. Emerging PFAS such as GenX and F-53B are also beginning to be reported in aquatic environments. This paper presents a broad review and discussion on the occurrence of PFAS in municipal and industrial wastewater which appear to be their main sources. Carbon adsorption and ion exchange are currently used treatment technologies for PFAS removal. However, these methods have been reported to be ineffective for the removal of short-chain PFAS. Several pioneering treatment technologies, such as electrooxidation, ultrasound, and plasma have been reported for PFAS degradation. Nevertheless, in-depth research should be performed for the applicability of emerging technologies for real-world applications. This paper examines different technologies and helps to understand the research needs to improve the development of treatment processes for PFAS in wastewater streams.
HIGHLIGHTS
PFOA and PFOS are frequently detected in both WWTPs, in spite of their limited use.
Most PFAS in both WWTPs show negative removal rates.
GAC and ion-exchange resins are ineffective at removing short-chain PFAS.
Positive results for EOX, ultrasound, plasma, and new generation adsorbents.
Further studies are required for the practicability of emerging technologies at real scale.
Graphical Abstract
INTRODUCTION
Poly-/perfluoro alkyl substances (PFAS) are a class of man-made chemicals that have been used in numerous products, such as coating materials, surfactants, and fire retardants, for decades. These chemicals get discharged into the environment throughout their production, application, waste disposal, and many other routes (Figure 1). Extensive use and non-biodegradable properties of these compounds have resulted in a widespread presence in the environment. Two 8-carbon PFAS, perfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS), are well known and widely used and have been detected in surface water (Gobelius et al. 2018), seawater (Brumovský et al. 2016), groundwater (Cousins et al. 2016), and tap water (Scher et al. 2018) in many locations around the world.
Due to their strong C–F bonds, PFAS are reported to be very persistent when released into the environment and therefore tend to bioaccumulate in the food chain. When exposed to eight or more carbon chain length PFAS, vertebrate animals suffer instabilities in the liver and pancreas (Li et al. 2017; Mahapatra et al. 2017). In particular, PFOA affects lipid and glucose metabolism, mitochondrial dysfunction, and oxidative phosphorylation in zebrafish, mice, and humans (Nelson et al. 2010; Abbott et al. 2012; Yan et al. 2015). The toxicity of PFAS is closely linked to the length of their carbon chain (Kleszczyński et al. 2007; Ulhaq et al. 2015), which can directly influence their removal and bioaccumulation rates (Kudo et al. 2006). In vitro toxicity tests in mammalian cells show this positive relationship between the chain length and the toxicity (Mulkiewicz et al. 2007). Shorter chain PFAS can be excreted faster from the body and therefore have been shown to result in minor accumulation in the liver tissues and serum of mice (Kudo et al. 2001).
Over the last decade, many manufacturers have phased out the production of C8-based PFAS, and the United States and European Union have regulated or limited the production of PFOA and PFOS. Therefore, industrial releases for PFOA and PFOS have decreased in developed countries since 2006 due to regulations and voluntary phaseout. However, industrial environmental releases of PFAS have increased in many other countries, like China, due to an expansion of the fluorochemical industry. Consequently, since C8 productions have been restricted, the production of short-chain PFAS (C2–C7) has increased (Wang et al. 2013b; Houtz et al. 2016). The degree of the environmental impact of a PFAS source would depend upon the scale of PFAS release, specific compounds, and their concentrations. The identification of PFAS sources to surface water is important to establish effective strategies for controlling their release. However, there is a lack of literature review on the occurrence, type, and concentration of PFAS in both municipal and industrial wastewater treatment plants (WWTPs) which seem to be the major point sources. Table 1 shows PFAS which have been reported in the literature to be present in municipal and industrial WWTP effluents.
PFAS and their structures which have been detected in municipal and industrial WWTP influents and effluents
Class . | Compound . | Abbreviation . | Chemical formula . | Molecular weight (g/mol) . | pKa . | Concentration range . | Identified source . | Reference . |
---|---|---|---|---|---|---|---|---|
Perfluorocarboxylates (PFCAs) | Perfluorohexanoate | PFHxA | F(CF2)5CO2 | 314.05 | −0.16 | 165–847 ng/L | Industrial influent | Kim et al. (2021), Kunacheva et al. (2011) |
Perfluoroheptanoate | PFHpA | F(CF2)6CO2− | 364.06 | −2.29 | 662–1,143 ng/L | Industrial effluent | Kunacheva et al. (2011) | |
Perfluorooctanoate | PFOA | F(CF2)7CO2− | 414.07 | −0.5 to 4.2 | 103–443 ng/L | Municipal effluent | Chen et al. (2016c) | |
Perfluorononanoate | PFNA | F(CF2)8CO2− | 464.08 | 0 | 0.05–4.9 ng/L | Municipal influent | Dauchy et al. (2017) | |
Perfluorodecanoate | PFDA | F(CF2)9CO2− | 514.08 | −5.2 | 0.1–8.3 ng/L | Municipal effluent | Dauchy et al. (2017) | |
Perfluorotridecanoate | PFTrDA | F(CF2)12CO2− | 664.11 | 0.52 | 15.2 ng/L | Industrial effluent | Kim et. al. (2021) | |
Perfluoroundecanoate | PFUnDA | F(CF2)10CO2− | 564.09 | −5.2 | 0.031–<2 μg/L | Industrial effluent | Kunacheva et al. (2011) | |
Perfluorododecanoate | PFDoA | F(CF2)11CO2− | 614.10 | 0.52 | 0.039–<2 μg/L | Industrial effluent | Kunacheva et al. (2011) | |
Perfluoropentadecanoate | PFPA | F(CF2)14CO2− | 764.12 | n/a | 8.8–20 ng/L | Municipal influent | Houtz et al. (2016) | |
Perfluorohexadecanoate | PFHxDA | F(CF2)15CO2− | 814.13 | 0.52 | <MLQ −22.8 ng/L | Municipal effluent | Sinclair & Kannan (2006) | |
Perfluorobutanoate | PFBA | F(CF2)3CO2− | 214.04 | n/a | 15–23 ng/L | Municipal effluent | Houtz et al. (2016) | |
Perfluorosulfonates (PFSAs) | Perfluorobutane sulfonate | PFBS | F(CF2)4SO3− | 300.10 | −3.31 | 1.29–195 ng/L | Municipal influent | Pan et al. (2016) |
Perfluorohexane sulfonate | PFHxS | F(CF2)6SO3− | 399.11 | −3.45 | 2.6–15 ng/L | Municipal influent | Kunacheva et al. (2011) | |
Perfluorooctane sulfonate | PFOS | F(CF2)8SO3− | 500.13 | <1 | <0.44–461.7 | Municipal effluent | Chen et al. (2012), Gallen et al. (2018), Pan et al. (2016) | |
Fluorotelomer alcohols (FTOHs) | 6:2 fluorotelomer alcohol | 6:2 FTOH | F(CF2)6CH2CH2OH | 364.10 | >7 | 0.89–12.29 ng/L | Municipal influent | Dauchy et al. (2017), Zhao et al. (2013) |
8:2 fluorotelomer alcohol | 8:2 FTOH | F(CF2)8CH2CH2OH | 464.12 | >7 | 0.144–10.3 ng/L | Municipal influent | Dauchy et al. (2017), Zhao et al. (2013) | |
10:2 fluorotelomer alcohol | 10:2 FTOH | F(CF2)10CH2CH2OH | 526.14 | >7 | 0.07–1.1 ng/L | Municipal influent | Dauchy et al. (2017), Zhao et al. (2013) | |
Fluorotelomer carboxylates (FTCAs) | 6:2 fluorotelomer carboxylate | 6:2 FTCA | F(CF2)6CH2CO2− | 358.08 | n/a | 8.62 μg/L | Municipal effluent | Gremmel et al. (2017) |
Fluorotelomer sulfonates (FTSAs) | 6:2 fluorotelomer sulfonate | 6:2 FTS | F(CF2)6CH2CH2SO3− | 428.17 | n/a | 242.5 μg/L | Industrial influent | Gomez-Ruiz et al. (2017a) |
8:2 fluorotelomer sulfonate | 8:2 FTS | F(CF2)8CH2CH2SO3− | 528.18 | n/a | 874 ng/L | Industrial influent | Gomez-Ruiz et al. (2017a) | |
Fluorotelomer sulfonamidealkylbetaine (FTABs) | 6:2 fluorotelomer sulfonamidealkylbetaine | 6:2 FTAB | C6F13C2H4SO2 − NHC3H6N+(CH3)2CH2COO− | 556.14 | n/a | 45.5 mg/L | Industrial influent | Dauchy et al. (2017) |
Fluorotelomer iodides (FTIs) | 6:2 fluorotelomer iodide | 6:2 FTI | F(CF2)6CH2CH2I | 473.90 | n/a | 0.74–3.5 μg/L | Industrial effluent | Bach et al. (2016) |
Perfluorooctane sulfonamides (FOSAs) | N-ethyl perfluorooctane sulfonamidoacetic acid | NEtFOSAA | C8F17SO2N(C2H5) CH2CO2H | 585.20 | n/a | 4.3–5.9 ng/L | Municipal influent | Boulanger et al. (2005) |
Chlorinated polyfluorinated ether sulfonates | 6:2 chlorinated polyfluorinated ether sulfonate (F53-B) | 6:2 Cl-PFAES | C8F16ClO(SO3K) | 570.6 | n/a | 10–50 ng/L | Industrial effluent | Ruan et al. (2015), Wang et al. (2013a) |
Perfluoroalkyl ether carboxylic acids with one ether group (mono-ether PFECAs) | Hexafluoropropylene oxide dimer acid (GenX) | HFPO-DA (GenX) | C6F11O3H | 330.2 | n/a | Total PFAS concentration of 40,000 ng/L | Industrial influent | Hopkins et al. (2018) |
Perfluoro-2-methoxyacetic acid | PFMOAA | C3HF5O3 | 180.0 | n/a | ||||
Perfluoroalkyl ether carboxylic acids with multiple ether group (multi-ether PFECAs) | Acetic acid, 2-[difluoro(trifluoromethoxy)methoxy]-2,2-difluoro- | PFO2HxA | C4HF7O4 | 246.0 | n/a | |||
perfluoro-3,5,7-trioxaoctanoic acid | PFO3OA | C5HF9O5 | 312.0 | n/a | ||||
2-[[[difluoro(trifluoromethoxy)methoxy]-difluoromethoxy]-difluoromethoxy]-2,2-difluoroacetic acid | PFO4DA | C6HF11O6 | 378.1 | n/a |
Class . | Compound . | Abbreviation . | Chemical formula . | Molecular weight (g/mol) . | pKa . | Concentration range . | Identified source . | Reference . |
---|---|---|---|---|---|---|---|---|
Perfluorocarboxylates (PFCAs) | Perfluorohexanoate | PFHxA | F(CF2)5CO2 | 314.05 | −0.16 | 165–847 ng/L | Industrial influent | Kim et al. (2021), Kunacheva et al. (2011) |
Perfluoroheptanoate | PFHpA | F(CF2)6CO2− | 364.06 | −2.29 | 662–1,143 ng/L | Industrial effluent | Kunacheva et al. (2011) | |
Perfluorooctanoate | PFOA | F(CF2)7CO2− | 414.07 | −0.5 to 4.2 | 103–443 ng/L | Municipal effluent | Chen et al. (2016c) | |
Perfluorononanoate | PFNA | F(CF2)8CO2− | 464.08 | 0 | 0.05–4.9 ng/L | Municipal influent | Dauchy et al. (2017) | |
Perfluorodecanoate | PFDA | F(CF2)9CO2− | 514.08 | −5.2 | 0.1–8.3 ng/L | Municipal effluent | Dauchy et al. (2017) | |
Perfluorotridecanoate | PFTrDA | F(CF2)12CO2− | 664.11 | 0.52 | 15.2 ng/L | Industrial effluent | Kim et. al. (2021) | |
Perfluoroundecanoate | PFUnDA | F(CF2)10CO2− | 564.09 | −5.2 | 0.031–<2 μg/L | Industrial effluent | Kunacheva et al. (2011) | |
Perfluorododecanoate | PFDoA | F(CF2)11CO2− | 614.10 | 0.52 | 0.039–<2 μg/L | Industrial effluent | Kunacheva et al. (2011) | |
Perfluoropentadecanoate | PFPA | F(CF2)14CO2− | 764.12 | n/a | 8.8–20 ng/L | Municipal influent | Houtz et al. (2016) | |
Perfluorohexadecanoate | PFHxDA | F(CF2)15CO2− | 814.13 | 0.52 | <MLQ −22.8 ng/L | Municipal effluent | Sinclair & Kannan (2006) | |
Perfluorobutanoate | PFBA | F(CF2)3CO2− | 214.04 | n/a | 15–23 ng/L | Municipal effluent | Houtz et al. (2016) | |
Perfluorosulfonates (PFSAs) | Perfluorobutane sulfonate | PFBS | F(CF2)4SO3− | 300.10 | −3.31 | 1.29–195 ng/L | Municipal influent | Pan et al. (2016) |
Perfluorohexane sulfonate | PFHxS | F(CF2)6SO3− | 399.11 | −3.45 | 2.6–15 ng/L | Municipal influent | Kunacheva et al. (2011) | |
Perfluorooctane sulfonate | PFOS | F(CF2)8SO3− | 500.13 | <1 | <0.44–461.7 | Municipal effluent | Chen et al. (2012), Gallen et al. (2018), Pan et al. (2016) | |
Fluorotelomer alcohols (FTOHs) | 6:2 fluorotelomer alcohol | 6:2 FTOH | F(CF2)6CH2CH2OH | 364.10 | >7 | 0.89–12.29 ng/L | Municipal influent | Dauchy et al. (2017), Zhao et al. (2013) |
8:2 fluorotelomer alcohol | 8:2 FTOH | F(CF2)8CH2CH2OH | 464.12 | >7 | 0.144–10.3 ng/L | Municipal influent | Dauchy et al. (2017), Zhao et al. (2013) | |
10:2 fluorotelomer alcohol | 10:2 FTOH | F(CF2)10CH2CH2OH | 526.14 | >7 | 0.07–1.1 ng/L | Municipal influent | Dauchy et al. (2017), Zhao et al. (2013) | |
Fluorotelomer carboxylates (FTCAs) | 6:2 fluorotelomer carboxylate | 6:2 FTCA | F(CF2)6CH2CO2− | 358.08 | n/a | 8.62 μg/L | Municipal effluent | Gremmel et al. (2017) |
Fluorotelomer sulfonates (FTSAs) | 6:2 fluorotelomer sulfonate | 6:2 FTS | F(CF2)6CH2CH2SO3− | 428.17 | n/a | 242.5 μg/L | Industrial influent | Gomez-Ruiz et al. (2017a) |
8:2 fluorotelomer sulfonate | 8:2 FTS | F(CF2)8CH2CH2SO3− | 528.18 | n/a | 874 ng/L | Industrial influent | Gomez-Ruiz et al. (2017a) | |
Fluorotelomer sulfonamidealkylbetaine (FTABs) | 6:2 fluorotelomer sulfonamidealkylbetaine | 6:2 FTAB | C6F13C2H4SO2 − NHC3H6N+(CH3)2CH2COO− | 556.14 | n/a | 45.5 mg/L | Industrial influent | Dauchy et al. (2017) |
Fluorotelomer iodides (FTIs) | 6:2 fluorotelomer iodide | 6:2 FTI | F(CF2)6CH2CH2I | 473.90 | n/a | 0.74–3.5 μg/L | Industrial effluent | Bach et al. (2016) |
Perfluorooctane sulfonamides (FOSAs) | N-ethyl perfluorooctane sulfonamidoacetic acid | NEtFOSAA | C8F17SO2N(C2H5) CH2CO2H | 585.20 | n/a | 4.3–5.9 ng/L | Municipal influent | Boulanger et al. (2005) |
Chlorinated polyfluorinated ether sulfonates | 6:2 chlorinated polyfluorinated ether sulfonate (F53-B) | 6:2 Cl-PFAES | C8F16ClO(SO3K) | 570.6 | n/a | 10–50 ng/L | Industrial effluent | Ruan et al. (2015), Wang et al. (2013a) |
Perfluoroalkyl ether carboxylic acids with one ether group (mono-ether PFECAs) | Hexafluoropropylene oxide dimer acid (GenX) | HFPO-DA (GenX) | C6F11O3H | 330.2 | n/a | Total PFAS concentration of 40,000 ng/L | Industrial influent | Hopkins et al. (2018) |
Perfluoro-2-methoxyacetic acid | PFMOAA | C3HF5O3 | 180.0 | n/a | ||||
Perfluoroalkyl ether carboxylic acids with multiple ether group (multi-ether PFECAs) | Acetic acid, 2-[difluoro(trifluoromethoxy)methoxy]-2,2-difluoro- | PFO2HxA | C4HF7O4 | 246.0 | n/a | |||
perfluoro-3,5,7-trioxaoctanoic acid | PFO3OA | C5HF9O5 | 312.0 | n/a | ||||
2-[[[difluoro(trifluoromethoxy)methoxy]-difluoromethoxy]-difluoromethoxy]-2,2-difluoroacetic acid | PFO4DA | C6HF11O6 | 378.1 | n/a |
PFAS review papers published to date have mostly evaluated PFAS occurrence in drinking water, surface, and groundwaters, or only municipal WWTPs and there is a lack of studies on their presence, concentrations, and fate in industrial WWTPs. Hence, the purpose of this review is to deliver comprehensive information on the occurrence of PFAS in both municipal and industrial wastewater, and compare the type and concentration of PFAS detected in both industrial and municipal WWTPs. Providing a detailed assessment of PFAS problems linked to both industrial and municipal wastewater, this literature review emphasizes the occurrence of a broad group of PFAS in these sources. Furthermore, assessment of treatment technologies in the literature typically focused on novel technologies (i.e., plasma) or only a specific class of technology (i.e., chemical oxidation, electrochemical technologies, and advanced oxidation). In addition, those reviews mostly considered PFAS remediation from surface water or groundwater. This paper reviews both recent and emerging technologies in removing PFAS from wastewater streams with a scale of technology ranging from laboratory studies to full-scale treatment systems. Literature gaps and recommendations for future research were compiled using these data.
PFAS SOURCES IN MUNICIPAL AND INDUSTRIAL WASTEWATER
PFAS can be found in many customer products, including food packaging materials, stain-resistant carpets, furniture, paints, non-stick cookware, cleaning products, waxes, and fire-fighting foams. When these PFAS-containing products are used in our daily life, they can enter municipal wastewater treatment plants (WWTPs) and landfills. Moreover, many PFAS sorb on biosolids that are produced in WWTP operations. Land application of PFAS-contaminated biosolids was observed to pollute soil, groundwater, and surface waters. In addition, some biosolids are directly sent to landfills and many landfills send their leachates to municipal WWTPs. Similarly, manufacturing facilities or industries such as chrome plating, electronics, and oil recovery that use PFAS significantly contribute to the occurrence of PFAS in industrial wastewater samples. Because conventional wastewater treatment processes are not sufficient to remove and degrade these compounds, PFAS in both municipal and industrial WWTPs have received a great amount of attention. Although some fluorotelomers (FTs) can get oxidized during treatment processes, they are converted to recalcitrant perfluorocarboxylic acids (PFCAs) (Bach et al. 2016; Kim et al. 2021). Thus, WWTP effluents may contain a relatively high amount of PFAS, and effluent discharge to surface waters may result in toxic effects on aquatic organisms and bioaccumulation of PFAS. Effluents from WWTPs in different countries have been stated to have PFOA and PFOS concentrations measured as <0.28–1,057 ng/L and <0.44–462 ng/L, respectively (Bossi et al. 2008; Guo et al. 2010; Kunacheva et al. 2011; Chen et al. 2012, 2016a, 2016c; Zhang et al. 2013; Subedi & Kannan 2015; Hamid & Li 2016; Pan et al. 2016; Dauchy et al. 2017; Gremmel et al. 2017; Gallen et al. 2018; Nguyen et al. 2019).
The occurrence of PFAS in industrial wastewater
In the literature, there are limited studies that examine the discharge of PFAS from industrial WWTPs to evaluate the occurrence of PFAS and their total discharge loads, especially for emerging PFAS. However, understanding PFAS pollution from the PFAS manufacturing and utilizing industries are crucial in regulating the sources of such compounds. PFAS are not effectively removed in WWTPs. Kunacheva et al. (2011) reported similar findings when conducting a study in two industrial zones (located in Central and Eastern Thailand) that have biological treatment processes. The study shows that ten different known PFAS, namely, PFOA, PFOS, PFPA, PFHxA, PFHpA, PFHxS, PFNA, PFDA, PFUnDA, and PFDoA, were identified in both industrial WWTPs. While total average concentrations of PFAS in industrial wastewater samples ranged between 674 and 847 ng/L in influent, effluent samples ranged between 662 and 1,143 ng/L. The abundant compounds were determined as PFOA, PFOS, PFNA, PFDA, PFUnDA, and PFHpA accounting for more than 87% of total PFAS, while other identified PFAS were less than 10%. Concentrations of most PFAS were higher in effluent samples than in influent samples, demonstrating no removal of most PFAS in the treatment process (Kunacheva et al. 2011), or conversion of precursors materials to PFAS. Another study investigated mass flows and the fate of 51 legacy PFAS in the WWTP of a fluorochemical manufacturer in France (Dauchy et al. 2017). Accordingly, one PFCA (PFHxA) and nine FTs, namely 6:2 FTOH, 8:2 FTOH, 10:2 FTOH, 12:2 FTOH, 6:2 fluorotelomer sulfonic acid (6:2 FTSA), 8:2 FTSA, 6:2 fluorotelomer iodide (6:2 FTI), 6:2 fluorotelomer sulfonamide alkylbetaine (6:2 FTAB and M4), were identified in influent samples. 6:2 FTAB was found to be the dominant PFAS, reaching a concentration of 45.5 mg/L. It was found that mass flows of PFCA considerably increased after biological treatment leading to the conclusion that the degradation of precursors was forming PFCA. However, PFCA concentrations decreased remarkably after the air flotation process due to sorption onto flotation bubbles and sludge. But, PFCA concentrations were still high in the final effluent (from 21 to 247 g/day). Another example of the increasing tendency of mass flow was observed for PFCAs from four industrial WWTPs in Denmark. Among them, the textile industry contributed to the highest concentration of PFCAs. This was reported as a result of FTOH degradation, which is widely used as saturating agents for textiles (Bossi et al. 2008). However, a recent study reported that while 6:2 FTS was detected in 12 different industrial influents at concentrations reaching up to 145 ng/L, it was detected in only one effluent sample. It was concluded that it was sorbing onto the sludge (Kim et al. 2021).
In the last decade, emerging PFAS have also been detected together with legacy PFAS. For example, emerging PFAS such as 6:2 chlorinated polyfluorinated ether sulfonate (F-53B), which is a PFOS replacement chemical, has been detected in electroplating industry wastewater at similar levels as PFOS (10–50 ng/L) in China (Wang et al. 2013a; Ruan et al. 2015). In addition, GenX (PFOA replacement compound) and other emerging PFAS including perfluoro-2-methoxyacetic acid (PFMOAA), acetic acid, 2-[difluoro(trifluoromethoxy)methoxy]-2,2-difluoro- (PFO2HxA), perfluoro-3,5,7-trioxaoctanoic acid (PFO3OA), Nafion byproduct, and 2-[[[difluoro(trifluoromethoxy)methoxy]-difluoromethoxy]-difluoromethoxy]-2,2-difluoroacetic acid (PFO4DA) have been detected in industrial wastewater with a total concentration of about 40,000 ng/L (Hopkins et al. 2018). Additionally, a recent study reported the identification of 37 new PFAS (6 classes), mostly chlorine-substituted PFCAs, together with 12 legacy PFAS (2 classes; PFCAs and PFSAs), 41 previously discovered PFAS (7 classes including perfluorinated dicarboxylates and perfluorinated ethers/alcohols) in the nearby Yangtze River (Wang et al. 2018). It was mentioned that only one PFAS class (per- or polyfluorinated monocarboxylates) was removed after the treatment. By contrast, the concentration of four classes of PFAS increased after the treatment process. The tendency of rising PFAS concentrations in effluents could be ascribed to existing PFAS precursors in wastewater and their biodegradation during the activated sludge process. There is also a possibility that compounds with a short carbon chain (C4) may be degraded into their PFCA or PFSA homologs (D'eon et al. 2009). Some scientists have stated that laboratory-scale experiments support the field studies indicating biodegradation of some precursors in aerobic conditions can contribute significant amounts of PFAS (Rhoads et al. 2008; Kibambe et al. 2020). However, further research on this process is still required to understand the actual behavior and chemical reactions during biodegradation of PFAS. Also, the identification of precursors is critical for the information of secondary sources of PFAS, such as PFOA and PFOS. Furthermore, it is still essential that numerous unregulated and emerging PFAS in the aquatic environment should be determined and quantitated. The behavior of PFAS may differ during the treatment process due to different functional groups, physicochemical properties, existing ions, and other compounds in wastewater. In some cases, long-chain PFAS have a greater tendency to sorb to sediment, while sorption of short-chain PFAS is more difficult due to their higher water solubility (Wang et al. 2020b). For instance, in contrast to increasing PFAS concentrations in effluents, a study stated that the typical concentrations of PFAAs were lower in the effluent (106 ng/L) than in influent (411 ng/L), associated with the paper and textile industries (Kim et al. 2016). Also, some authors reported that while PFOS levels showed a decreasing trend in effluent samples, PFOA concentration increased similar to the other studies mentioned above (Schultz et al. 2006; Guo et al. 2010). The higher normalized organic carbon partition coefficient of PFOS may be the reason for this tendency as compared to that of carboxylate analog, which points to the favored uptake of some PFAS by the sludge or biosolids (Guo et al. 2010). In addition, existing cations in wastewater may provide the required ion bridges for the sorption of negatively charged PFAS. For example, PFAS sorption to the sludge (biosolids) increased with higher cation concentrations and it was greater in the presence of divalent cations compared to monovalent cations (Ebrahimi et al. 2021). This could be ascribed to electrostatic interaction between a divalent cation and the anionic PFAS head group, as well as with a negative charge on the sludge. It was also observed that PFOS accumulated in higher amounts (400 ng/g) in sludge in comparison with other PFAS with accumulation ranging between 2.9 and 327.7 ng/g (Kunacheva et al. 2011). Dauchy et al. 2017 also observed a remarkable decrease in the concentration of some PFAS after the flotation tank as the compounds were adsorbed onto air bubbles. Remarkably high concentrations of PFAAs have been detected in the sludge from electronics (91 ng/g) and chemical (81.5 ng/g) industries (Dauchy et al. 2017). These results show that PFAS removal occurs primarily by sorption on sludge during conventional treatment. In summary, four factors affect the fate of PFAS in WWTPs: (1) physicochemical properties of PFAS, (2) existing cations in wastewater, (3) precursors of PFAS in wastewater, and (4) produced sludge or biosolids. It should also be noted that the behavior of a single-compound system is different from the behavior of a multi-component system, and multi-component interactions play a significant role in the removal rates of pollutants. For example, sorption capacities of seven PFAS, namely PFOA, PFOS, PFBA, PFBS, PFHxA, PFHxS, and PFDoA, onto sludge were evaluated, and the authors mentioned that the sorption of PFAS greatly diminished in the multi-component system in comparison with individual PFAS in the single-component system (Zhou et al. 2010; Zhao et al. 2013). This behavior might be attributed to hydrophobic properties, ionic head groups, and chain length of PFAS. In summary, the examination of the fate of numerous and different PFAS classes at several wastewater treatment steps (i.e., activated sludge and flotation) indicates that biodegradation is responsible for the increasing PFAS concentrations due to the transformation of PFAS precursors to perfluoroalkyl acids (PFAAs) in WWTPs. Besides, sorption onto sludge and/or bubbles may occur resulting in decreased PFAS concentrations in the effluent. Further research studies are needed to increase our knowledge of these fate mechanisms.
The occurrence of PFAS in municipal wastewater
Municipal WWTPs are a significant source of PFAS which can bioaccumulate in food chains and eventually result in human consumption. Clara et al. (2009) indicated that more than half of the total PFAS detected in the Danube River, Eastern Europe can be ascribed to municipal wastewater discharges (Clara et al. 2009). In addition, PFOS detected in rivers in Japan is correlated to the discharges from municipal WWTPs (Murakami et al. 2008; Zhang et al. 2013). Another study showed that municipal WWTP effluents were the main source of the PFAS discharged into Lake Ontario (Boulanger et al. 2005).
PFAS and their precursors for domestic use are usually categorized as fluorophosphonate (FP)-based products, perfluorooctane sulfonyl fluoride (POSF)-based products, and FT-based products. FP-based products may contain small amounts of releasable PFAS residuals by means of PFAS-based emulsifiers usage (Larsen et al. 2006), such as polytetrafluoroethylene (PTFE)-coated non-stick cookware, polyvinylidene fluoride (PVDF) coatings, and woven fiberglass cloths. Non-polymeric POSF-based materials contain releasable precursors of PFAS and are indirectly released into the environment, while their impurities, such as coated paper and package products and pesticides based on perfluorooctane sulfonamides (FOSAs), are released directly into the environment. Non-polymeric FT-based materials contain 0.04–3.8% of releasable unbound FTs (indirect environmental releases) and <1–100 ppm of PFAS impurities (direct environmental releases) (Dinglasan-Panlilio & Mabury 2006).
A study reported the results of a screening survey of PFAS from six municipal and four industrial WWTPs and landfill sites in Denmark (Bossi et al. 2008). PFOA and PFOS were observed as dominant species in all municipal WWTPs, including both influent and effluent waters. It was stated that PFOA and PFOS levels were higher in effluent samples, except at one municipal WWTP which provided complete removal of both compounds. Increasing concentration levels of PFOA and PFOS in effluent samples was also observed in other studies (Sinclair & Kannan 2006; Clara et al. 2009; Zhang et al. 2013). According to a study (Zhang et al. 2013) conducted in China, 16 PFAS were detected in municipal wastewater samples, and their concentrations ranged between 0.04 and 91 ng/L for influent and 0.01–107 ng/L for effluent. PFOA had the highest concentrations for both influents (2–91 ng/L) and effluents (3–107 ng/L). They postulated that these high PFOA concentrations may be due to the surrounding industries. The second leading compound was PFOS with concentrations of 1–32 ng/L in the influent and 1–67 ng/L in the effluent. It is also reported that many investigated WWTPs showed higher concentrations of PFAS (PFOA, PFOS, PFNA, PFHpA, PFBS, and PFDA) in effluent than in influent samples, with a cumulative increase of 32–290 times. This trend is linked to the PFAS precursors in wastewater and their biodegradation to form PFAS. Accordingly, PFOS may come from biodegradation of N-ethyl-perfluorooctane sulfonamido acetate (EtFOSAA) which can transform to N-ethylperfluorooctane-1-sulfonamide (EtFOSA), which is converted to FOSA and finally to PFOS. PFOA may be formed due to the biodegradation of 8:2 FTOH and polyfluoroalkyl phosphate esters (Martin et al. 2004). Another reason for the increasing trend in effluents may also be the desorption of PFAS from solids (Zhang et al. 2013).
Xiao et al. (2012) conducted experimental data analysis for the contribution of PFAS from municipal WWTPs (Xiao et al. 2012). They identified PFAS in 37 WWTPs in more than 40 cities in Minnesota, USA. Substantially high levels of PFHxA, PFOA, and PFOS in influents have been found in almost half of the WWTPs surveyed. Per capita discharges were measured as 0.30–7.6, 0.17–1.4, 0.67–5.1, 0.35–7.8, and 0.17–1.4 μg/capita/d for PFOA, PFNA, PFHxA, PFOS, and PFHpA, respectively. A considerable increase in influent samples for PFHxA and PFOA was detected in 59% of the WWTPs studied, indicating high-level inputs of precursors. Indeed, the authors modeled the fate of one precursor (8:2 FTOH) using fugacity analysis which showed that 9.0% of 8:2 FTOH is biodegradable to PFOA or other analogs during the activated sludge process. These high PFAS levels in influents were observed especially in WWTPs situated in industrial areas. Furthermore, a noteworthy correlation was observed in PFAS concentrations and population in the influent water for some WWTPs, demonstrating PFAS were primarily from local sources and the use of PFAS-based products by the residents has a large impact (Xiao et al. 2012).
Another study investigated PFAS discharge from municipal WWTPs, including river and coastal release straight to the Bohai Sea, China. Accordingly, they sampled effluents from 15 drain channels including 10 river outfalls, 3 WWTPs, and 2 industrial WWTPs. Ten PFCAs (C4-C12) and four PFSAs (C4-C10) were identified in all effluents with levels of 103–443 ng/L. While PFHxA and PFOA were the main PFAS in the effluents from municipal drain channels and rivers, PFOA and PFHxS were leading PFAS in the effluents of industrial drain outlets. Furthermore, the average concentrations of PFOA and PFHxA found in rivers and municipal drain channels were much higher than those in seawater samples, demonstrating wastewater from drain outlets was a major pollution source of PFHxA and PFOA in the Bohai Sea (Chen et al. 2016a).
A study conducted in 2014 detected PFAS in eight municipal WWTP which discharge effluents to San Francisco Bay, USA. It was mentioned that the concentrations of short-chain PFCAs, such as PFBA and PFHxA, increased compared to municipal wastewater effluents sampled in 2009. Most of the effluents showed similar trends for total PFAS concentration. While PFHxA had the highest average concentration with 24 ng/L, the concentrations of PFOA, PFOS, and PFBA ranged between 15 and 23 ng/L. However, effluents from two WWTPs contained higher concentration levels for all PFAS (i.e., the median concentration of PFOS was 420 ng/L). Similar to previously mentioned studies, increasing concentrations were observed for effluents due to the precursors of PFAS. They observed that PFAS precursors were the reason for 33–63% of the total concentration of PFAS for all effluents (Houtz et al. 2016).
Similar to industrial wastewater, emerging types of PFAS were also detected in municipal wastewater samples. For instance, emerging PFAS including H–PFCAs, H-PFESAs, and PFSMs were detected at WWTPs in China, and their concentrations were equal to or higher than legacy PFAS in WWTPs (Wang et al. 2020d). While some classes of PFAS were removed, the class of perfluoroalkyl ether alcohol (PFEA) increased during the treatment due to biodegradation of other ether-PFAS.
With regard to the aforementioned studies, emerging and previously discovered PFAS seem to be a challenging target since PFAS have been found in municipal and industrial WWTPs. Despite the restriction and regulation of the manufacturing of C8-based compounds (i.e., PFOA and PFOS), PFOA and PFOS are still identified as the dominant compounds in most cases. This may be related to the persistency, accumulative behavior, and other physicochemical properties of these compounds (Pan et al. 2016; Li et al. 2020). The biodegradation of precursors and abiotic transformations may also contribute to the occurrence of PFOA and PFOS (Eriksson et al. 2017; Chen et al. 2019). The main gap in the literature is that there is inadequate experimental data on the related physicochemical properties of the compounds, such as their solubilities in water, soil–water partitioning coefficients, and bioaccumulation factors. This information would be valuable to model the fate and transport of PFAS during and after treatment. Aside from PFOA and PFOS, short-chain PFCAs (C4–C7) and PFSAs (C4) are the most detected PFAS in municipal and industrial WWTP influents and effluents as a result of the replacement of C8-based products. Longer chain PCASs (C7–C12) and one PFSA (C6) are also identified in many WWTPs, and other fluorinated compounds, such as FTOHs and FTSAs, also seem to be present in WWTPs. More research is required to identify and examine the fate of emerging PFAS classes.
RECENT TREATMENT TECHNOLOGIES FOR PFAS
Conventional treatment processes, such as biological activated sludge, cannot provide efficient removal and/or degradation of PFAS, since they are resistant to biodegradation. Although activated carbon adsorption and ion-exchange processes are being used at full scale and pilot scale for the removal of PFAS mostly for drinking water and groundwater, currently there are some studies focusing on these technologies for PFAS removal from wastewater at various Technology Readiness Level (TRL) stages, as reported in the literature. This paper makes an attempt to summarize the key features of these technologies.
Activated carbon adsorption
Carbon adsorption process is well known and the most commonly used treatment technology for PFAS removal. In this process, four mechanisms have been reported: (1) dispersion in the liquid phase, (2) mass transfer through the solid phase, (3) pore diffusion inside an adsorbent, and (4) electrostatic interface between pollutant and exchange site (Qiu 2007; Park et al. 2020). Most full-scale treatment systems using activated carbon are focused on PFAS removal from impacted drinking water sources. There is a lack of information on the removal of PFAS from industrial and municipal wastewaters. A more complex matrix of wastewater may limit its direct usage for wastewater treatment, but carbon-based adsorption can be utilized after pretreatment systems and/or reclamation of wastewater effluents. For example, Thompson et al. (2011) examined the removal of different PFCAs and PFSAs from treated water of WWTP in Australia. Plant A used adsorption via biologically activated carbon (BAC) and other treatment methods (including filtration and ozonation) to purify recycled water. PFOA, PFHxA, PFOS, and PFHxS were the most abundant compounds in effluent water. While PFOS and other long-chain PFAS removal occurred in Plant A during the adsorption/filtration processes, PFOS and shorter chain PFAS were not removed completely (Thompson et al. 2011). Similarly, the effectiveness of powdered activated carbon (PAC) was examined at a large scale to remove emerging pollutants, including PFOA and PFOS, from wastewater. While PFOS was poorly removed by PAC adsorption with removal ratios ranging between 6 and 52%, negative removals were reported for PFOA because concentrations were continuously higher in effluents compared to influents, and which was attributed to potential contamination due to the usage of Teflon® pipes (Mailler et al. 2015). Another study reported pilot-scale research on the removal of PFBA, perfluoropentanoic acid (PFPnA), PFHxA, PFHxS, PFHpA, PFOA, PFOS, PFNA, and PFDA from effluent water by granular activated carbon (GAC) made from peach pit and spruce pine wood biochars. GAC was found to be very effective in the removal of PFPnA, PFHxA, PFOA, and PFOS. PFOS was removed to a higher extent on GAC (61%) than PFOA (35%), while both are of the same chain length. PFOS has a higher molecular weight and has a lower water solubility (0.57 g/L) compared to PFOA (3.4 g/L), thus it characteristically shows a higher sorption affinity to solids because of its hydrophobic sulfonic acid group (Inyang & Dickenson 2017).
By contrast, some bench-scale studies were conducted using activated carbon to remove several PFAS from wastewater effluents. For example, a study investigated the removal of PFOA, PFHxA, and PFHpA at concentrations of 0.29 mmol/L, 0.1 mmol/L, and 0.11 mmol/L, respectively, from actual wastewater using coal-based activated carbon adsorption (Du et al. 2015). They observed more than 88% PFOA removal with the adsorbent dose of 0.3 g/L, while a dose of 1.9 g/L was required to provide 90% removal of other PFCAs. This indicates the presence of co-existing ions and other organics in wastewater significantly affected the efficiency of activated carbon adsorption for other PFCAs. Another study evaluated the performance of GAC (bituminous coal) for the removal of 13 PFAS from wastewater (Rostvall et al. 2018). GAC showed 95% average removal efficiency for the target PFAS because of its very high specific surface area (950 m2 g/L) and average pore size of 2.2 nm. It was also mentioned that there was no specific relationship between removal efficiencies and the functional group using GAC due to the low number of treated bed volumes. FOSA was an exception, possibly due to its neutral sulfonamide head group (Rostvall et al. 2018). GAC (bituminous coal and coconut shell) efficacy was also tested by Appleman et al. (2013) for the removal of PFAAs in the presence of natural organic matter (NOM). GAC was less efficient for the shorter chain PFCAs, such as PFBA and PFPeA, as increased concentrations were observed in treated samples. This was attributed to the competitive effects of other adsorbing species, such as longer chain PFAAs and DOM, which caused desorption of adsorbed PFBA and PFPeA. This conclusion was verified with higher removal efficiencies in lower DOC values. The presence of DOM affected GAC performance for PFAA removal because it occupied pores and sorption sites of GAC (Appleman et al. 2013).
A relationship between chain length/functional group and treatment efficiency is usually observed during GAC treatment. For example, PFAS with carboxylate groups showed lower removal efficiency than those with sulfonate functional groups. In addition, the time for column breakthrough decreased with decreasing chain length, indicating that agglomeration and/or micelle formation for long-chain PFAS was poorer for the PFCAs than PFSAs. Current concerns about the adsorption of PFAS are that longer chain PFAS potentially tend to be adsorbed by GAC whereas short-chain PFAS do not, and the presence of competing ions reduces the removal efficiencies (McCleaf et al. 2017). Therefore, more attention should be given to enhancing GAC efficiency and lowering the operation cost by the addition of pretreatment or using other sorbents.
Ion exchange
Ion-exchange resins, particularly anion-exchange resins for PFAS, can efficiently remove most of the PFAS from water due to their molecular structure and their simultaneous removal mechanisms of anion exchange and adsorption. PFAS compounds have two functional components, namely fluorinated carbon chain which is hydrophobic with a non-ionic tail and negatively charged anionic head. The anion-exchange resin consists of neutral co-polymers (mostly plastics) and exchange sites with a positive charge. The PFAS’ hydrophobic tail is adsorbed to the hydrophobic backbone on the resin, and the anionic head of the PFAS attracts to the ion-exchange sites of the resin, which is positively charged. The typical structure of anion-exchange resin is given in Figure 2.
The removal efficiency of PFAS by anion-exchange resins strongly depends on the type and concentration of inorganic content in water and the affinity of resin for PFAS (Lampert et al. 2007; Deng et al. 2010; Maimaiti et al. 2018; Zeng et al. 2020). Some resins have less sensitivity for inorganic ions and more affinity for PFAS. While most of the PFAS molecules have an anionic head, some PFAS molecules have a cationic head, and others have both anionic and cationic heads (zwitterionic). Bench-scale column tests should be conducted to find out the most suitable resin for PFAS removal.
A study explored the effect of resin functional groups during PFOS removal (Chularueangaksorn et al. 2013). The polystyrene divinylbenzene resins were more hydrophobic than polyacrylic resins. The hydrophobicity of the resins increased the hydrophobic interactions between the PFOS molecule and the resin. Hence, polystyrene divinylbenzene structure and the gel type increased the adsorption of PFOS (Chularueangaksorn et al. 2013). Conversely, Deng et al. (2010) observed that the resin matrix exhibited a substantial effect on the PFOS sorption capacity from wastewater in bench-scale experiments. Resins made of polyacrylic showed higher and faster removal of PFOS than the polystyrene resins as a result of a more hydrophilic structure. The sorption capacity of PFOS on polyacrylic resins was determined to be 4–5 mmol/g. The authors also observed that the amount of adsorbed PFOS on the resins was more than the release of chloride from resins, demonstrating the contribution of other interactions, such as adsorption, apart from anion exchange in the removal mechanism. It was also stated that the presence of co-existing ions such as sulfate and chromium in wastewater inhibited the sorption of PFOS due to a competing effect on the exchange sites (Deng et al. 2010).
Functional group of PFAS removal could affect their removal via ion-exchange resins. Strong base anion (Cl and OH ionic form) resins were used in a study to remove PFOS and PFOA from processed wastewater in batch and continuous flow column experiments (Lampert et al. 2007). PFOS concentration was reduced to below detection limits after treatment, while PFOA was reduced to low levels. The PFOA and PFOS loadings onto the resin were measured as 686 mg/g and 151 mg/g, respectively. According to the isotherm tests, PFOA/chloride isotherm showed an incurvated shape, indicating the exchange of PFOA with chloride was more satisfactory than the exchange of PFOS. PFOA/PFOS isotherm also showed that PFOS was favored over PFOA (Lampert et al. 2007). Poly(N-[3-(dimethylamino)propyl]acrylamide, methyl chloride quaternary, and (DMAPAA-Q) hydrogel matrix was reported to remove long- and short-chain PFAS, including PFBA, PFBS, and GenX, from treated wastewater samples in bench-scale studies (Ateia et al. 2019b). Equilibrium time ranged between 60 and 120 min for complete removal of PFAS and was mentioned that higher removal was observed for sulfonated PFAS as compared to carboxylic PFAS, regardless of chain length (Ateia et al. 2019b).
Ion-exchange resins usually offer more efficient adsorption for short-chain PFAS as compared to activated carbon. Du et al. (2015) reported higher adsorption capacities for PFOA, PFHpA, and PFHxA by ion-exchange resin (IRA67) compared to a bamboo-derived AC at acidic pH in perfluorooctanesulfonyl fluoride (PFOSF) washing wastewater. Ion-exchange resins have provided efficient removal of PFOA and PFOS. More research is needed to reduce the operating cost and enhance the removal effectiveness of different types of PFAS in the presence of competing organics, ions, and co-contaminants in wastewater.
EMERGING TREATMENT TECHNOLOGIES FOR PFAS
The physical and chemical properties of PFAS, such as high energy carbon–fluorine bonds, hydrophobicity, and thermal stability, pose challenges to many treatment technologies. In addition, many of those technologies have been developed and designed for other classes of pollutants, not for PFAS. Thus, the existing treatment options for PFAS are inadequate in number as compared to other pollutants. Although recent technologies, such as carbon adsorption and ion exchange, show high removal efficiencies, secondary pollutants and/or high cost of regeneration procedures make it difficult to apply these technologies. Thus, innovative technologies are being developed, focusing on PFAS degradation rather than transferring from one phase to another. For this reason, advanced oxidation processes (AOPs), such as electrooxidation (EOX), photocatalysis, plasma, and ultrasound, are receiving increasing attention.
Many innovative technologies are being investigated in laboratories, but the scale-up also must be considered to deal with challenges when translating from laboratory-scale to pilot- and full-scale applications. Considering the extremely high persistence and mobility of PFAS, emerging treatment technologies should be designed to achieve efficient degradation for even very low concentration levels.
Electro-oxidation
Electro-oxidation (EOX) processes have attracted considerable attention in the last few years due to their simplicity and efficiency of the degradation of various harmful contaminants. This technology may also be called ‘green technology’ since little or no chemicals are added during water and wastewater treatment. The electrode material is one of the most significant factors in the EOX process; thus, the development of novel electrode materials is a key factor to increase treatment efficiency. EOX can be used for treating different kinds of wastewaters, disinfecting drinking water, and degrading harmful target pollutants (Sirés & Brillas 2012; Brillas & Martínez-Huitle 2015; Gomez-Ruiz et al. 2017a; Barışçı et al. 2018a, 2018b; Turkay et al. 2018; Barisci & Suri 2020; Barisci & Suri 2021). EOX involves two mechanisms that occur simultaneously during the oxidation of pollutants: (1) direct electron transfer to the anode surface, and (2) indirect oxidation with powerful reactive oxygen species (ROS) generated from water oxidation, such as hydroxyl radical or active oxygen (Sirés et al. 2014; Brillas & Martínez-Huitle 2015; Schaefer et al. 2015). Ozone and hydrogen peroxide can also be produced during the EOX process. Depending on the existing components in water or the chemicals added to the water, indirect oxidation may occur via other oxidants, such as chlorine and hypochlorite (also called mediated oxidation).
Gomez-Ruiz et al. (2017a, 2017b) investigated the treatment of eight PFAS from industrial wastewater by EOX using the BDD anode in bench-scale experiments (Gomez-Ruiz et al. 2017b). The industrial WWTP received wastewater from four manufacturing plants, with one of them producing PFAS, and contributed 3–17% to the overall flow treated in the industrial WWTP. The total concentration of PFAS in the industrial effluent was 1,652 μg/L, in which 6:2 FTAB and 6:2 FTSA were the most significant contributors with 92% (w/w). In their experiments, a BDD anode with an active surface area of 0.007 m2 was used to treat 2 L of industrial effluent with an applied current density of 50 mA/cm2. After 10 h of process time, 99.7% total PFAS removal occurred while more than 90% TOC removal was achieved. All analyzed PFAS, except 6:2FTSA, were found to be below detection limits after treatment, while 6:2 FTSA was the only compound identified in the treated water. The energy consumption was estimated as 256 kWh/m3 to achieve 99.7% removal. Barisci & Suri (2020) studied short- and long-chain PFCA degradation in different water matrices, i.e., ultrapure water, surface water, and domestic wastewater effluent using BDD anodes at bench-scale. More than 95% degradation was reported for all long-chain PFCAs (except C7), regardless of chain length in ultrapure water. Here, 37% defluorination and 74% TOC removal were reported in the same study during the treatment of 10 mg/L PFOA in ultrapure water. In contrast to long-chain PFCAs, strong chain-length dependence was observed for short-chain PFCAs. It was also mentioned that water matrix did not have an effect for short-chain PFCAs; however, degradation efficiencies were in the order of ultrapure water > WWTP effluent > river water for long-chain PFCAs. The reason for this behavior was explained by higher concentrations of nitrite and bromide, which are known ·OH scavengers, in river water as compared to WWTP effluent. However, when a Ti/RuO2 electrode was used, to study the electrochemical degradation of PFCAs and PFSAs in a laboratory-scale study (Barisci & Suri 2021), the effect of PFAS chain length on the degradation rate was reported to be significant. For example, the degradation of short-chain PFAS ranged between 16 and 67% while it was 64–91% for long-chain PFAS. The TOC removals were 64 and 39% for PFOA and PFOS, respectively. Recently, the Magnéli phase titanium sub-oxide (Ti4O7) electrode has been reported as a promising candidate for the EOX process due to its chemical inertness, high conductivity, and relatively low production cost (Walsh & Wills 2010). This material, also known as Ebonex® (Atraverda Ltd, UK), has the generic formula of TinO2n−1 (3 ≥ n ≥ 10) with >2.0 V and ∼− 1.4 V anodic and cathodic polarization, respectively. Unlike BDD anode, which is not effective at high pH, the Magnéli phase anode is extremely stable over a wide range of pH, including very high and low values. However, recent applications of a Magnéli phase electrode have been limited to deionized water at bench-scale for degradation of PFOA and PFOS. Only one study examined Magnéli phase electrodes for degradation of PFAAs from still bottoms recovered from regenerated saturated ion-exchange resin media. Still bottom samples had a TOC value ranging between 13,280 and 201,529 mg/L with a chloride content of 62.7 mg/L. The degradation ratios of monitored PFAS including PFOA, PFOS, PFHxS, PFHpA, and perfluoro-1-heptanesulfonate (PFHpS) were 98.9%, 94.8%, 95.0%, 56.9%, and, 96.8%, respectively, while a 79.5% TOC reduction was achieved after 40 h of process time. However, the increase in short-chain PFAS (PFBA, PFBS, PFPeA, and PFHxA) concentration was observed which was concluded to possible incomplete degradation of precursors and long-chain PFAAs. The authors also mentioned that no considerable adsorption of PFAS on the magnéli phase anode was detected and the removal of PFAS was mostly attributed to electrochemical reactions (Wang et al. 2020a).
In the literature, there are numerous scale-up studies for the EOX process of wastewater containing many organic contaminants. However, limited scale-up studies for PFAS-containing wastewater have been reported. In one study, a pilot-scale EOX process was used to treat industrial wastewater containing PFAS (Fath et al. 2016). The process was automated and the pilot-scale reactor consisted of 36 lead (Pb93Sn7) electrodes having an active surface area of 700 mm × 400 mm connected parallelly. The process successfully degraded C6–C8 PFAS with up to 99% efficiency. However, the process was reported to be less efficient for short-chain PFAS (PFBS). It was reported that PFAS degradation produced hydrogen fluoride under the acidic operating conditions (Fath et al. 2016). Operational cost was calculated as US$4500/year (140 L/h, 10 ppm PFAS with 95% degradation) in this study. Both laboratory- and pilot-scale studies show that this technology could be used for efficient degradation of PFAS-containing wastewater in full scale, but more research should be conducted to determine optimum operating conditions and to provide a cost-effective process. Furthermore, the presence of chloride in wastewater may lead to the formation of toxic byproducts, i.e., chlorate and perchlorate, during EOX treatment of PFAS (Yang et al. 2019), and which should also be considered before using this technology.
Sonochemical degradation (ultrasound)
The sonochemical process uses an acoustic field to produce chemical reactions in an aqueous media. Cavitation occurs through the collapse of bubbles formed in the solution by sound waves. This results in high temperatures at cavity collapse points and at the bubble–water interface (Fu et al. 2007; Suri et al. 2007; Cheng et al. 2008; Suri et al. 2008; Andaluri et al. 2012; Hao et al. 2014; Shende et al. 2019; Shende et al. 2021a, 2021b). Characteristic ultrasound frequencies range between 20 and 1,000 kHz during the sonochemical degradation of PFAS.
Vecitis et al. (2010) reported the ultrasound treatment of AFFF-contaminated wastewater (FC-600), containing PFOS (3,650 ± 710 mg/L), PFHS (820 ± 140 mg/L), PFBS (254 ± 13 mg/L), PFOA (161 ± 13 mg/L), and PFHA (53 ± 13 mg/L). The primary constituent in FC-600, PFOS, was removed by more than 90% in 2 h with a frequency of 505 kHz. Additionally, the TOC decreased by almost 50% from 270 to 140 ppm after 120 min of process time (Vecitis et al. 2010). Moreover, ultrasound treatment of AFFF stockpiles has been also examined using a dual-transducer system that consists of supersonic and megasonic frequencies (Kucharzyk et al. 2017). The system would be scaled up for the treatment of a large volume of PFAS solutions. This novel approach will be useful for managing the inventory of AFFF products kept at defense sites to develop a system which could generate high temperatures for PFAS thermal degradation in presence of select reactive oxidizing chemicals. The authors contend that this approach could lead to lower energy requirements (Kucharzyk et al. 2017).
Based on these studies and data in the literature, more research is required for the application of ultrasound technology at pilot- and field-scale to treat wastewater influents/effluents with a high concentration of PFAS. Due to its high energy needs, the technology is more favorable for treating small wastewater volumes but at high PFAS concentrations. This technology could be used in combination with other processes such as physical separation and/or other AOPs to avoid the negative effects of competing ions and organics and provide higher PFAS degradation efficiencies at a lower cost.
Plasma-based technologies
Plasma-based treatment technology uses only electricity, converting water in an air or oxygen environment into a mixture of extremely oxidative species, such as OH•, H2O2, HO2•, H•, O2•−, O3, H2 and hydrated electrons (e−aq) (Thagard et al. 2009; Joshi & Thagard 2013). These species have a very short lifetime. However, the electrical discharges take place close to the water surface in the reactor (i.e., just above the water level), and some of these radical species may get into the water and destroy the pollutants (Burlica et al. 2006). It must be mentioned that there are approaches to generate plasma inside the aqueous solution but is not very efficient. Plasma-based water/wastewater treatment technologies can have diverse discharge types, namely pulsed corona, direct current (DC) pulseless corona, gliding arc, dielectric barrier, DC arc, and DC glow discharges. Different oxidative species with different degradation efficiencies can be formed during plasma water treatment based on the type of applied discharge. For example, in some cases, reactive nitrogen species (RNS), including peroxynitrite and peroxynitrate, can also be generated (Lewis et al. 2020). For pollutant degradation, plasma is usually generated through a high voltage between two electrodes: one grounded electrode within or contacting the polluted water and one high voltage electrode. Up until now, several attempts have been made for the utilization of plasma to degrade PFAS from synthetic and groundwater samples (Yasuoka et al. 2011; Takeuchi et al. 2013; Stratton et al. 2017; Singh et al. 2019; Lewis et al. 2020) using different types of electrical discharge plasma reactors with different electrode configurations. Studies investigating reactive species generated during plasma treatment show that hydroxyl and other superoxide radicals, which are characteristically the main plasma-based oxidizing agents, do not play an important role in PFOA degradation. Instead, it is reported that hydrated electrons comprise a substantial fraction of the PFOA transformation. Some linear chain PFCAs with carbon chain-length C4–C7 have been identified as byproducts during degradation of PFOA and PFOS in plasma-based treatment in the literature (Singh et al. 2019). PFOA, PFBS, and PFHxS were also identified as byproducts of the degradation of PFOS. In addition, numerous PFOA- and PFOS-related intermediates at trace amounts and gaseous byproducts in small amounts have also been identified where defluorination ratios ranged between 58 and 77%. Those identified byproducts, including short-chain PFCAs, suggest a stepwise degradation of the PFAS molecule, following degradation of perfluoroalkyl radicals, intermediates, and perfluoro alcohols/ketones (Singh et al. 2019; Lewis et al. 2020).
A recent study investigated a bench-scale plasma reactor for degradation of PFAS from landfill leachate samples containing six short-chain, five long-chain perfluoroalkyl acids (PFAAs), and eight PFAA precursors at concentrations ranging 102–105 ng/L. The concentrations of PFOS and PFOA were 2,000 and 3,000 ng/L, respectively. Here, 500 mL of samples were treated by plasma, resulting in higher degradation efficiencies of longer chain PFAAs compared with shorter chain PFAAs. Both PFOS and PFOA were degraded by 90% in 10–75 min. Other long- and short-chain PFAAs were degraded by >99% (Singh et al. 2020a). Another bench-scale batch plasma reactor was used to degrade PFAS from ion-exchange regenerant still bottom solutions containing a high concentration (∼100 mg/L) and a low concentration (<1 μg/L) of PFAS. In the plasma reactor for the high concentrations, the precursors of PFAS and long-chain PFAAs were degraded by >99.9% in 2 h, where short-chain PFAAs were degraded by >99% in 6 h of treatment. To degrade additional PFAAs, another plasma reactor was used for low concentrations. In this reactor, the authors added a cationic surfactant (cetrimonium bromide) which promoted the removal of short-chain PFAAs (below detection limits) in 1.5 h. Overall, >99% of PFAS from wastewater was removed during the treatment with corresponding fluorine recoveries of 47–117% (Singh et al. 2020b).
Plasma-based processes achieved efficient degradation of PFAS in synthetic water, contaminated groundwater, and wastewater samples. This technology seems to be less affected by competing ions and other organic contaminants present in real wastewater samples, as chemical reactivity mostly occurs in the plasma–liquid interface in comparison with other AOPs, including ultrasound and persulfate. According to these, plasma-based processes are promising for PFAS degradation from wastewater. However, further studies should be done to confirm its efficiency for wastewater treatment, degradation of intermediates, and scale-up. In addition, safety aspects need to be considered due to the high voltage discharges needed to generate the plasms.
Heterogenous photocatalysis using nanomaterials
Direct photolysis has been recognized as a promising option for the removal of a variety of organic pollutants, however, direct photolysis for the degradation of PFAS in wastewater is of low efficiency due to the strong energy of C−F bonds of PFAS (Panchangam et al. 2009). In comparison, heterogeneous photocatalysis has been developed to degrade PFAS by applying certain catalysts with photon (hv) absorption capacity. During heterogeneous photocatalysis, the pollutant and catalyst occur in various stages and the process can be defined as absorbing hv and then producing electrons (e−) and positively charged electron–hole pairs (h+), which can migrate to the photocatalysts’ surface and react directly or indirectly with pollutants (Suri et al. 1993; Crittenden et al. 1997; Suri et al. 1999). It has been demonstrated that heterogeneous photocatalysts such as gallium oxide (Ga2O3) and indium trioxide (In2O3) are much more effective for the degradation of PFAS compared to direct photolysis (Chen et al. 2011; Li et al. 2012, 2013, 2014; Tan et al. 2020; Xu et al. 2020). However, light penetration and separation or immobilization of catalysts on support materials is limiting the application of this technology (Thiruvenkatachari et al. 2008).
4.4.1. Ga2O3-based nanomaterials
β-Ga2O3 is an extensive bandgap semiconducting material (Eg = 4.8 eV) (Zhao & Zhang 2009) that finds multipurpose use in electronic devices with high temperature and gas sensors (Lin et al. 2007). Various nanostructures of Ga2O3 have been synthesized, such as nanoplates, nanosheets, nanowires, and nanobelts (Huang & Yeh 2010; Yan et al. 2010; Muruganandham et al. 2012). For example, Shao et al. (2013a, 2013b) synthesized sheaf-like Ga2O3 consisting of nanoplates with high crystallinity for PFOA removal from secondary effluent (Shao et al. 2013a). The sheaf-like Ga2O3 had a high specific surface area (36.1 m2/g) and a large number of nanopores (2–4 and 8 nm). The feasibility of sheaf-like Ga2O3 photocatalysis to degrade PFOA in a municipal wastewater plant in Beijing, China has been tested. 81% of PFOA was degraded within 3 h by photocatalysis processes in the presence of sheaf-like Ga2O3. However, the PFOA degradation reduced in the secondary effluent compared to ultrapure water due to the impeding effect of bicarbonate and organic constituents since bicarbonate concentration was approximately 3,000 times higher than the concentration of PFOA in the secondary effluent. Due to the competition between bicarbonate and PFOA for the surface of Ga2O3, PFOA adsorption was inhibited (Li et al. 2012). The authors also mentioned that after adjusting the pH of the secondary effluent as 4.3 to transform bicarbonate into carbonic acid, the degradation of PFOA increased to 100% from 81% after 3 h. However, PFOA degradation was still significantly slow as compared to results for pure water, and this was attributed to adsorption and attenuation of UV penetration by organic matters in the secondary effluent. The same research group also investigated the needle-like Ga2O3 for PFOA degradation from wastewater; 100% PFOA degradation was achieved with an initial concentration of 500 μg/L in wastewater. The required treatment time was 180 min which was much longer compared to ultrapure water (Shao et al. 2013b).
In2O3-based nanomaterials
Electron–hole pair separation and band gap are important properties to improve photocatalyst performance. Nanocrystals such as In2O3 have lower band gaps (2.9–3.0 eV) and larger available surface areas (Li et al. 2013). Compared to TiO2 nanoparticles, In2O3 nanocrystals have a higher activity to degrade PFAS because the transformation of photogenerated holes to OH• is extremely lower for In2O3 nanocrystals (Vecitis et al. 2009). Accordingly, OH• generated in In2O3 solution is less than the amount produced in TiO2 solution with the same UV dosage (Li et al. 2013). This implies that the reaction between photogenerated holes in the valence band of In2O3 and the interface of water or hydroxyl group would be slower. Therefore, most of the photogenerated holes on the surface of In2O3 would not be converted to OH• but conserved under UV radiation. This was confirmed by other authors demonstrating that direct oxidation due to the generated hole provided dramatically higher PFOA decay than oxidation with OH• (Kutsuna et al. 2006; Vecitis et al. 2009).
Li et al. (2012) investigated photocatalysis of PFOA in the presence of an In2O3 catalyst. They mentioned that while 69% PFOA (initial concentration of 100 μmol/L) was degraded in pure water by In2O3 within 180 min, the efficiency was reduced to only 10% in wastewater (Li et al. 2012). However, once the pH value was adjusted to 4 for the secondary effluent and ozone gas was concurrently added (∼8 mg/L.h), almost similar PFOA degradation (69%) was achieved. The ozone addition can promote the efficiency of PFOA degradation in wastewater because it can promote the reaction of photo-produced h+ with PFOA instead of recombination with photo-produced e− (Zsilák et al. 2014). In other words, ozone acts as an electron acceptor and reduces the recombination of electron and hole. PFOA degradation using In2O3 catalyst in simulated wastewater showed that wastewater constituents such as sulfate, chloride, and NOM significantly affected PFOA degradation. However, it was reported that when pH was adjusted to 2, PFOA degradation was 97.6% with complete defluorination (Jiang et al. 2016). The results show that PFAS degradation in wastewater could be significantly diminished by bicarbonate and NOM, however their adverse effects can be reduced by pH adjustment and the addition of ozone. It must be stated that changing the pH of large volumes of water may not be practically (economically) feasible.
New generation adsorbents
New nanoscale generation of adsorbents (nano-adsorbent) generally has advantages as compared to conventional adsorbents due to their larger and more reactive specific surface areas. The PFAS adsorption mechanism on nanoscale adsorbents includes the electrostatic interface, hydrogen bonds, hydrophobic effect, and ion exchange, which can be different from conventional adsorbents. Amongst them, both hydrophobic and electrostatic interactions are believed to be the key reasons for the adsorption of PFAS (Du et al. 2014; Muruganandham et al. 2015). Electrostatic interaction can occur between the negatively charged PFAS molecule and positively charged nano-adsorbents. Each fluorine atom in the PFAS molecule has a high electronegativity, giving it the tendency to attract electrons (Xiao et al. 2011). Therefore, along with the negative charge due to the ionization of the PFAS molecule, the strong electronegativity of its fluorine atoms also produces negatively charged shells covering the PFAS molecule, which can enhance the electrostatic force. PFAS usually consists of a hydrophilic head and a hydrophobic tail, as mentioned previously. Because of the hydrophobic interaction, the hydrophobic fluorinated tail which is negatively charged can be adsorbed onto the hydrophobic nano-adsorbents’ surface, in spite of electrostatic repulsion since some nano-adsorbents have a negative charge such as carbon nanotubes, (Fujii et al. 2007). Aside from the electrostatic and hydrophobic interactions, self-aggregation, such as the generation of hemi-micelles and/or micelles, may also increase the adsorption of PFAS. Considering these factors, many innovative nano-adsorbents have been synthesized and evaluated to improve PFAS adsorption. Most of the studies conducted at laboratory scales provide information that is helpful in developing field-scale treatment processes for PFAS removal from wastewaters.
Covalent organic frameworks
Covalent organic frameworks (COFs) are crystalline, covalent porous polymers containing various light elements (C, H, O, N, and B) (Peng et al. 2016). COFs display several attractive properties including large surface area, configurable porosity, outstanding stability, low mass densities, adaptable function, and faster charge carrier mobility (Zhang et al. 2020). Different synthesis methods of COFs have been utilized by compression reactions using covalent bonds such as imine (Chandra et al. 2013), hydrazine (Uribe-Romo et al. 2011), keto-enol (Kandambeth et al. 2012), and boroxine (Wan et al. 2009) linkages. Recent studies specify that mainly hydrophobic interactions are the key forces that provide the adsorption of hydrophobic PFOA on COFs (Yu et al. 2008; Kawano et al. 2013; Ateia et al. 2018; Ateia et al. 2019a), because the major skeleton of COFs is typically hydrophobic, which favors for PFOA adsorption. The sorption capacity usually increases with increasing perfluorocarbon chain length because of the higher hydrophobic properties of PFAS (Zhang et al. 2011; Ji et al. 2018; Xiao et al. 2019). However, evidence indicates that electrostatic reactions of the PFOA carboxylate headgroup and the cationic amine groups of the sorbent also take place during the adsorption process (Niu & Cai 2009; Mitra et al. 2016; Ateia et al. 2019b). For other PFAS having more hydrophilic character, electrostatic interactions play a predominant role depending on pH for the adsorption of PFAS on COFs (Du et al. 2014; Gao et al. 2017). In general, the long-chain PFAS have high hydrophobicity which provides higher adsorption affinity when compared to hydrophilic short-chain PFAS. Hence, while long-chain PFAS can be adsorbed via electrostatic and hydrophobic interactions, short-chain PFAS can be adsorbed mostly by electrostatic interactions (Senevirathna et al. 2010; Wang et al. 2019).
Recently, several applications of COFs for PFAS removal have been conducted at a bench-scale. Ji et al. (2018) investigated 2D amine-functionalized COF for the adsorption of PFOA, PFOS, and ammonium perfluoro-2-propoxypropionate (GenX) (Ji et al. 2018). It was reported that COF with partial amine incorporation exhibited more than 90% removal of PFAS, signifying the role of synergetic interaction between hydrophobic surfaces and polar groups for PFAS removal. A very recent study reported that the cationic COF with quaternary ammonium displayed 2.06 mmol/g of adsorption capacity for GenX which was more effective than traditional activated carbons and resins (Wang et al. 2020c). Another study reported the application of hybrid hydrogel matrix COF (DMAPAA-Q) for the removal of 16 PFAS including PFBA, PFBS, and GenX from industrial wastewater effluents (Ateia et al. 2019b). The results displayed fast removal kinetics at equilibrium times of 60–120 min. A higher removal was obtained for sulfonated PFAS compared to carboxylic PFAS, and no chain-length effect was observed. It was also mentioned that hydrogel COF showed an additional advantage over conventional adsorbents since no desorption was observed when the process time was extended to 24 h. Furthermore, the removal efficacy was similar at different pH range of 4–10 (Ateia et al. 2019b).
COFs were assessed in batch and column application modes for PFAS removal. Batch tests are useful and essential for the fast evaluation of adsorption kinetics and capacity of the COFs. However, batch application modes could be limited in real-scale applications for PFAS adsorption from industrial and municipal wastewater because the recovery of COFs is difficult in large-scale applications. Hence, more column tests, where solid to water volume is much higher compared to batch tests, should be performed since column tests can specify the performance of COFs in real-scale applications.
Metal–organic frameworks
Metal–organic frameworks (MOFs) are an innovative class of porous and crystalline materials composed of organic bridging linkers and metal/metal oxide-containing nodes (Barpaga et al. 2019). MOFs have recently gained attention due to their high surface area, tunable porosity, very high stability, easily adaptable functionalities, and reusability with relatively simple washing procedures compared to AC (Xue et al. 2016; Moreton et al. 2020).
PFAS sorption mechanism onto MOFs mostly relies on both electrostatic and hydrophobic interactions (Merino et al. 2016). The solution pH and other constituents of water affect electrostatic interactions between MOFs and PFAS molecules (Barpaga et al. 2019). In addition, as known, the hydrophobicity of PFAS molecule increases with the presence of more C–F bonds (Chen et al. 2016b). For example, Barpaga et al. (2019) synthesized chromium and iron analogs of the MOF for PFOS sorption at a laboratory scale. It has been reported that the sorption capacity of chromium analog MOF was higher as compared to iron analog MOF due to stronger interaction between the metal node and sulfur moieties of PFOS molecule (Barpaga et al. 2019). Another study developed an aluminum-based MOF for PFOA uptake from water at a laboratory scale and sorbent performance was compared with AC (Hakim Mohd Azmi et al. 2021). The much stronger adsorption capacity of PFOA (340 mg/g) compared to AC (120 mg/g) was observed, and which was attributed to the electrostatic bonds formed between the anionic functional group of PFOA and the aluminum-based MOF. A fluorinated zirconium-based MOF also presented high adsorption capacities for both PFOA (1.13 mmol/g) and PFOS (0.51 mmol/g) owing to the high interaction of fluoride moieties of the MOF and the PFAS molecule. Sini et al. (2018) employed zirconium-based MOF to display an improved affinity for PFOA and PFOS with fluorinated functionalities in the pores of MOF that can provide fluorine–fluorine interactions.
Although MOFs represented promising performances, these materials are still at a very early stage of development and testing. In addition, there is a lack of studies on MOF applications for domestic and industrial wastewater, thus, more research is required before they become a feasible technology for field-scale applications for wastewater treatment.
Carbon nanotubes
Carbon nanotubes (CNTs) are a promising nanoscale adsorbent that can provide effective PFAS removal due to their physicochemical properties, adsorption affinity, and exceptional structure. Although no field tests have been reported for wastewater treatment, Deng et al. (2012) studied bench-scale adsorption of PFOA, PFOS, PFHxS, PFHxA, PFBA, and PFBS on unchanged CNTs, namely, single-/multiwalled (MWCNT) CNTs and functionalized CNTs (MWCNT-OH and MWCNT-COOH). The authors observed that the PFAS adsorption on CNTs diminished for shorter chain PFAS. The main adsorption mechanism was the hydrophobic interaction between PFAS and the walls of CNTs. Hence, CNTs with hydrophilic -OH and -COOH groups significantly lowered the adsorption capacity as compared to the unchanged CNTs. Nevertheless, electrostatic repulsion could hinder the adsorption capacity, as both CNTs and PFAS had negative charges. Furthermore, the authors did not observe the formed hemi-micelles and micelles on the CNT surface. Consequently, the authors stated that the adsorption of PFAS on CNTs was less effective than other well known adsorbents, such as active carbon and resins (Deng et al. 2012). The oxygen content could be increased in MWCNTs to 18% to increase the sorption efficiency (Li et al. 2011b). A different study also showed that MWCNTs with electrochemical assistance were able to enhance the adsorption of PFAS from water containing CaCl2 and NaN3 (Li et al. 2011a). This enhancement was attributed to the improved electrostatic interaction between PFAS and MWCNTs. Comparing MWCNTs without electrochemical assistance, the kinetic model presented that the adsorption rates of PFOS and PFOA initially increased 41- and 60-fold, respectively. This study showed a promising process for the removal of PFAS (Li et al. 2011a). The data from existing studies might help to apply CNTs for field-scale wastewater treatment systems.
Organically modified silica
One of the next-generation adsorbents involves organically modified silica adsorbents which can provide effective removal of PFAS. Silica-based adsorbents have available pores with mesoscale which allow modifiers to be suitably mounted on their surfaces, giving the adsorbents high adsorption capacity and selectivity. Several compounds, such as 3-aminopropyltriethoxysilane, 1,8-bis(dimethylamino)naphthalene, hexadecyl trimethyl ammonium bromide, and cross-linked alkoxysilanes have been considered as a coating for the silicas to allow for higher adsorption of PFAS (Edmiston 2013).
β-Cyclodextrin (β-CD)-coated silica adsorbents were synthesized and investigated for PFOA removal (Bhattarai et al. 2014). The adsorbent prepared by using a crosslinking agent of hexamethylene diisocyanate was reported to remove more than 90% of PFOA. The authors also mentioned that increasing β-CD loading enhanced PFOA removal (Bhattarai et al. 2014). Another study investigated the removal of PFOA and PFOS by β-cyclodextrin–ionic liquid with the magnetic Fe3O4 nanoparticle adsorbent (Badruddoza et al. 2017). The sorption capacities were reported as 13,200 and 2,500 μg/g for PFOS and PFOA, respectively, and the higher PFOS removal was ascribed to hydrophobicity and functional groups of PFOS (Abu et al. 2017).
A novel modified organo-clay adsorbent (matCARE™) has been developed and evaluated for the PFOS removal from AFFF-polluted wastewater (Das et al. 2013) in laboratory batch experiments. It took only 1 h to reach equilibrium with significantly higher adsorption capacity (0.09 mmol/g) compared to activated carbon (0.07 mmol/L). This material was then used at field-scale in AFFF wastewater treatment plants in Australia (Arias Espana et al. 2015). For example, almost 1 million liters of wastewater were effectively treated at one of the WWTPs with an average influent concentration of 0.02 mmol/L and 0.004 mmol/L for PFOS and PFOA, respectively, to below detection limits, using matCARE™ cartridges (1.2 kg each).
Tables 2 and 3 summarize recent and emerging technologies on PFAS removal and advantages/disadvantages for each technology, respectively.
Summary of recent and emerging treatment technologies for PFAS degradation/removal
Treatment technology . | Target PFAS . | Water matrix . | Initial concentration . | Degradation/sorption . | Scale . | Reference . | |
---|---|---|---|---|---|---|---|
Recent technologies | AC | PFOA, PFOS, PFHxS, PFHxA, | Municipal WWTP effluent | 3.7–16 ng/L | PFOS and short-chain PFAS not removed | Full scale | Thompson et al. (2011) |
PAC | PFOA, PFOS | Municipal WWTP effluent | 25–44 ng/L | 6–52% for PFOS, PFOA not removed | Full scale | Mailler et al. (2015) | |
GAC | PFOA, PFOS, PFNA, PFDA, PFHpA, PFHxA, PFHxS, PFPnA, PFBA | Municipal WWTP effluent | 1 μg/L | Complete removal except PFOS (40%) | Pilot scale | Inyang & Dickenson (2017) | |
AC | PFOA, PFHxA, PFHpA | Industrial WWTP effluent | 0.1–0,29 mmol/L | 88–90% | Laboratory scale | Du et al. (2015) | |
GAC | 13 PFAS | Municipal WWTP effluent | 2 μg/L | 95% on average | Laboratory scale | Rostvall et al. (2018) | |
GAC (Calgon Filtrasorb® 300 | PFAAs | DI water + NOM | 1 μg/L | >20% breakthrough for all PFAAs and negative removal for PFBA and PFPeA | Laboratory scale | Appleman et al. (2013) | |
Ion exchange (PFA 300) | PFOS | DI water | 0.01–1 mg/L | 99%, 455 mg/g adsorption capacity | Laboratory scale | Chularueangaksorn et al. (2013) | |
Ion exchange (IRA67) | PFOS | Simulated industrial WW | 400 mg/L | 4–5 mmol/g adsorption capacity | Laboratory scale | Deng et al. (2010) | |
Ion exchange (A714) | PFOA, PFOS | Simulated industrial WW | 1,000–3,000 mg/L | PFOS below detection limit, 99.6% PFOA removal | Laboratory scale | Lampert et al. (2007) | |
Ion exchange | 11 short-chain, 5 long-chain including PFBA, PFBS, GenX | Surface water | 1 μg/L | 80–100% | Laboratory scale | Ateia et al. (2018) | |
Ion exchange (IRA67) | PFOA, PFOS, PFBA, PFBS, PFHxA, PFHxS | Simulated AFFF-impacted groundwater | 0.1–0.5 mmol/L | 90% of PFOS, 10% of PFBA (PFOS > PFHxS > PFOA > PFBS > PFHxA > PFBA) | Laboratory scale | Maimaiti et al. (2018) | |
Emerging technologies | EOX (BDD) | PFBA, PFPeA, PFHxA, PFHpA, PFOA, FTSAs, 6:2 FTCA, 6:2 FTAB | Industrial WWTP effluent | ΣPFAS = 1,642 μg/L | Σ99.7% | Laboratory scale | Gomez-Ruiz et al. (2017b) |
EOX (Ti/RuO2) | PFOA, PFOS and PFAAs | AFFF-impacted groundwater | 0.7–65 μg/L | 98% for PFOS and 58% for PFOS | Laboratory scale | Schaefer et al. (2015) | |
EOX (BDD) | Short- and long-chain PFCAs | DI water, river water and municipal WWTP effluent | 200 μg/L each | >92 and 75% defluorination for long-chain PFCAs in DI water and effluent | Laboratory scale | Barisci & Suri (2020) | |
EOX (TiRuO2) | PFCAs and PFSAs | DI water | 100 μg/L each | 16–67% for short-chain PFAS, 64–91% for long-chain PFAS. | Laboratory scale | Barisci & Suri (2021) | |
EOX (magnéli phase) | PFOA, PFOS, PFHxS, PFHpA, PFHpS | Still bottom wastewater | Not mentioned | 56.9%–98.9% | Laboratory scale | Wang et al. (2020a) | |
EOX (Lead) | C6–C8 PFAS | Industrial WWTP effluent | 1,000–20,000 μg/L | 99%, less effective for short-chain | Pilot scale | Fath et al. (2016) | |
Ultrasound (505 kHz) | PFOA, PFOS, PFHS, PFBS, PFHA | AFFF-contaminated wastewater | 53 ± 13–3,650 ± 710 mg/L | 90% removal in 2 h with 50% TOC removal | Laboratory scale | Vecitis et al. (2010) | |
Ultrasound (354–612 kHz) | PFOA, PFOS | DI water and groundwater below a landfill | ∼100 μg/L | Rate constant reduced by 56–61% in groundwater | Laboratory scale | Cheng et al. (2008) | |
Plasma-based treatment | PFOA, PFOS | DI water | 8.3 mg/L | Complete removal in 120 min | Laboratory scale | Singh et al. (2019) | |
Plasma-based treatment | Five long-chain, six short-chain PFAAs and eight PFAA precursors | Landfill leachate | 102–3,000 ng/L | Both PFOS and PFOA were removed by 90% in 10–75 min. Other long- and short-chain PFAAs were removed by >99% | Laboratory scale | Singh et al. (2020a) | |
Plasma-based treatment | Short- and long-chain PFAAs | Still bottom wastewater | ∼0.01–100 mg/L | >99% of short- and long-chain PFAAs were degraded in 2–6 h | Laboratory scale | Singh et al. (2020b) | |
Emerging technologies | Photocatalysis with needle-like Ga2O3 | PFOA | Municipal wastewater influent | 500 μg/L | 100% PFOA degradation was achieved in 180 min which was much longer compared to ultrapure water | Laboratory scale | Shao et al. (2013b) |
Photocatalysis in the presence of In2O3 catalyst | PFOA | Municipal wastewater effluent | 100 μmol/L | While 69% PFOA was degraded in pure water within 180 min, the efficiency was reduced to only 10% in wastewater, however, with pH adjustment and ozone addition similar PFOA degradation (69%) was achieved | Laboratory scale | Li et al. (2012) | |
Photocatalysis in the presence of In2O3 catalyst | PFOA | Simulated municipal wastewater | 100 mg/L | PFOA degradation was 97.6% with a complete defluorination at pH 2 | Laboratory scale | Jiang et al. (2016) | |
Electro assisted MWNTs | PFOA, PFOS | DI water with ionic strength | 100 μg/L | 92% | Laboratory scale | Li et al. (2011a) | |
Organo-clay (matCARETM) | PFOS | AFFF-contaminated wastewater | 0.6 mmol/L | 0.09 mmol/L adsorption capacity | Laboratory scale | Das et al. (2013) | |
Organo-clay (matCARETM) | PFOA, PFOS | Municipal wastewater influent | 0.04–0.02 mmol/L | Complete removal | Full scale | Arias Espana et al. (2015) | |
Photocatalysis with sheaf-like Ga2O3 | PFOA | Municipal wastewater effluent | 500 μg/L | 81% removal within 3 h with 66% defluorination, 100% after adjusting pH to 4.3 | Laboratory scale | Shao et al. (2013a) | |
COF | PFOA, PFOS, GenX and other PFAS | DI water | 100 μg/L | > 90% | Laboratory scale | Ji et al. (2018) | |
COF | GenX and HFPO-TA | DI water | 0.0756 mmol/L | 100% for HFPO-TA and 80% for GenX | Laboratory scale | Wang et al. (2020c) | |
COF | 16 PFAS including PFBA, PFBS, and GenX | Industrial wastewater effluent | 1,000 ng/L | Fast removal kinetics in 60–120 min with no desorption in 24 h | Laboratory scale | Ateia et al. (2019b) | |
MOF | PFOS | DI water | 1–100 mM | 100% | Laboratory scale | Barpaga et al. (2019) | |
MOF | PFOA and PFOS | DI water | 100–1,000 mg/L | > 90% | Laboratory scale | Sini et al. (2018) |
Treatment technology . | Target PFAS . | Water matrix . | Initial concentration . | Degradation/sorption . | Scale . | Reference . | |
---|---|---|---|---|---|---|---|
Recent technologies | AC | PFOA, PFOS, PFHxS, PFHxA, | Municipal WWTP effluent | 3.7–16 ng/L | PFOS and short-chain PFAS not removed | Full scale | Thompson et al. (2011) |
PAC | PFOA, PFOS | Municipal WWTP effluent | 25–44 ng/L | 6–52% for PFOS, PFOA not removed | Full scale | Mailler et al. (2015) | |
GAC | PFOA, PFOS, PFNA, PFDA, PFHpA, PFHxA, PFHxS, PFPnA, PFBA | Municipal WWTP effluent | 1 μg/L | Complete removal except PFOS (40%) | Pilot scale | Inyang & Dickenson (2017) | |
AC | PFOA, PFHxA, PFHpA | Industrial WWTP effluent | 0.1–0,29 mmol/L | 88–90% | Laboratory scale | Du et al. (2015) | |
GAC | 13 PFAS | Municipal WWTP effluent | 2 μg/L | 95% on average | Laboratory scale | Rostvall et al. (2018) | |
GAC (Calgon Filtrasorb® 300 | PFAAs | DI water + NOM | 1 μg/L | >20% breakthrough for all PFAAs and negative removal for PFBA and PFPeA | Laboratory scale | Appleman et al. (2013) | |
Ion exchange (PFA 300) | PFOS | DI water | 0.01–1 mg/L | 99%, 455 mg/g adsorption capacity | Laboratory scale | Chularueangaksorn et al. (2013) | |
Ion exchange (IRA67) | PFOS | Simulated industrial WW | 400 mg/L | 4–5 mmol/g adsorption capacity | Laboratory scale | Deng et al. (2010) | |
Ion exchange (A714) | PFOA, PFOS | Simulated industrial WW | 1,000–3,000 mg/L | PFOS below detection limit, 99.6% PFOA removal | Laboratory scale | Lampert et al. (2007) | |
Ion exchange | 11 short-chain, 5 long-chain including PFBA, PFBS, GenX | Surface water | 1 μg/L | 80–100% | Laboratory scale | Ateia et al. (2018) | |
Ion exchange (IRA67) | PFOA, PFOS, PFBA, PFBS, PFHxA, PFHxS | Simulated AFFF-impacted groundwater | 0.1–0.5 mmol/L | 90% of PFOS, 10% of PFBA (PFOS > PFHxS > PFOA > PFBS > PFHxA > PFBA) | Laboratory scale | Maimaiti et al. (2018) | |
Emerging technologies | EOX (BDD) | PFBA, PFPeA, PFHxA, PFHpA, PFOA, FTSAs, 6:2 FTCA, 6:2 FTAB | Industrial WWTP effluent | ΣPFAS = 1,642 μg/L | Σ99.7% | Laboratory scale | Gomez-Ruiz et al. (2017b) |
EOX (Ti/RuO2) | PFOA, PFOS and PFAAs | AFFF-impacted groundwater | 0.7–65 μg/L | 98% for PFOS and 58% for PFOS | Laboratory scale | Schaefer et al. (2015) | |
EOX (BDD) | Short- and long-chain PFCAs | DI water, river water and municipal WWTP effluent | 200 μg/L each | >92 and 75% defluorination for long-chain PFCAs in DI water and effluent | Laboratory scale | Barisci & Suri (2020) | |
EOX (TiRuO2) | PFCAs and PFSAs | DI water | 100 μg/L each | 16–67% for short-chain PFAS, 64–91% for long-chain PFAS. | Laboratory scale | Barisci & Suri (2021) | |
EOX (magnéli phase) | PFOA, PFOS, PFHxS, PFHpA, PFHpS | Still bottom wastewater | Not mentioned | 56.9%–98.9% | Laboratory scale | Wang et al. (2020a) | |
EOX (Lead) | C6–C8 PFAS | Industrial WWTP effluent | 1,000–20,000 μg/L | 99%, less effective for short-chain | Pilot scale | Fath et al. (2016) | |
Ultrasound (505 kHz) | PFOA, PFOS, PFHS, PFBS, PFHA | AFFF-contaminated wastewater | 53 ± 13–3,650 ± 710 mg/L | 90% removal in 2 h with 50% TOC removal | Laboratory scale | Vecitis et al. (2010) | |
Ultrasound (354–612 kHz) | PFOA, PFOS | DI water and groundwater below a landfill | ∼100 μg/L | Rate constant reduced by 56–61% in groundwater | Laboratory scale | Cheng et al. (2008) | |
Plasma-based treatment | PFOA, PFOS | DI water | 8.3 mg/L | Complete removal in 120 min | Laboratory scale | Singh et al. (2019) | |
Plasma-based treatment | Five long-chain, six short-chain PFAAs and eight PFAA precursors | Landfill leachate | 102–3,000 ng/L | Both PFOS and PFOA were removed by 90% in 10–75 min. Other long- and short-chain PFAAs were removed by >99% | Laboratory scale | Singh et al. (2020a) | |
Plasma-based treatment | Short- and long-chain PFAAs | Still bottom wastewater | ∼0.01–100 mg/L | >99% of short- and long-chain PFAAs were degraded in 2–6 h | Laboratory scale | Singh et al. (2020b) | |
Emerging technologies | Photocatalysis with needle-like Ga2O3 | PFOA | Municipal wastewater influent | 500 μg/L | 100% PFOA degradation was achieved in 180 min which was much longer compared to ultrapure water | Laboratory scale | Shao et al. (2013b) |
Photocatalysis in the presence of In2O3 catalyst | PFOA | Municipal wastewater effluent | 100 μmol/L | While 69% PFOA was degraded in pure water within 180 min, the efficiency was reduced to only 10% in wastewater, however, with pH adjustment and ozone addition similar PFOA degradation (69%) was achieved | Laboratory scale | Li et al. (2012) | |
Photocatalysis in the presence of In2O3 catalyst | PFOA | Simulated municipal wastewater | 100 mg/L | PFOA degradation was 97.6% with a complete defluorination at pH 2 | Laboratory scale | Jiang et al. (2016) | |
Electro assisted MWNTs | PFOA, PFOS | DI water with ionic strength | 100 μg/L | 92% | Laboratory scale | Li et al. (2011a) | |
Organo-clay (matCARETM) | PFOS | AFFF-contaminated wastewater | 0.6 mmol/L | 0.09 mmol/L adsorption capacity | Laboratory scale | Das et al. (2013) | |
Organo-clay (matCARETM) | PFOA, PFOS | Municipal wastewater influent | 0.04–0.02 mmol/L | Complete removal | Full scale | Arias Espana et al. (2015) | |
Photocatalysis with sheaf-like Ga2O3 | PFOA | Municipal wastewater effluent | 500 μg/L | 81% removal within 3 h with 66% defluorination, 100% after adjusting pH to 4.3 | Laboratory scale | Shao et al. (2013a) | |
COF | PFOA, PFOS, GenX and other PFAS | DI water | 100 μg/L | > 90% | Laboratory scale | Ji et al. (2018) | |
COF | GenX and HFPO-TA | DI water | 0.0756 mmol/L | 100% for HFPO-TA and 80% for GenX | Laboratory scale | Wang et al. (2020c) | |
COF | 16 PFAS including PFBA, PFBS, and GenX | Industrial wastewater effluent | 1,000 ng/L | Fast removal kinetics in 60–120 min with no desorption in 24 h | Laboratory scale | Ateia et al. (2019b) | |
MOF | PFOS | DI water | 1–100 mM | 100% | Laboratory scale | Barpaga et al. (2019) | |
MOF | PFOA and PFOS | DI water | 100–1,000 mg/L | > 90% | Laboratory scale | Sini et al. (2018) |
Advantages and disadvantages of the recent and emerging technologies used for the treatment of PFAS from wastewater
Treatment technology . | Advantages . | Disadvantages . |
---|---|---|
Activated carbon adsorption | Technically simple process and adaptable to many existing treatment plants Can be applied for large-scale operations Inexpensive process Extensive range of available commercial products Highly effective process with fast adsorption kinetics High quality of the treated effluent | Nondestructive method Not effective for short-chain PFAS Efficiency reduces in the presence of other constituents Costly reactivation process Requirement of further disposal of the adsorbent if not reactivated |
Ion-exchange resins | Technically simple operation Wide range of commercially available products Easy operation and maintenance Easy integration to other wastewater treatment processes Possibility of regeneration and reuse High quality of the treated effluent | Nondestructive method Economic limitations (high capital and maintenance cost of the selective resin and time-consuming regeneration process) Clogging of columns in the presence of particulates and organic matter, and may require a physicochemical pre-treatment Performance sensitive to pH Not very effective for short-chain PFAS Requirement of further disposal of the resin if not regenerated |
Electrooxidation | Destructive technology High mineralization efficiency No need for addition of external chemicals in most cases Effective for many different types of PFAS including short-chain Relatively inexpensive operation cost Simple operation Rapid degradation | High initial cost of the electrodes Cost of the maintenance (depends on electrode inertness and stability) Anode passivation and sludge deposition on the electrodes which can reduce the process efficiency in continuous mode of operation Foam formation during wastewater treatment Possible toxic byproducts formation (chlorate and perchlorate) in the presence of chloride Mostly suitable for small- or medium-sized communities |
Ultrasound | Destructive technology No need for addition of external chemicals No secondary pollution Simple operation No sludge production | High operational cost Management of generated heat energy Scale-up challenges Mostly suitable for small waste streams |
Plasma-based technologies | Destructive technology Inexpensive operation cost High mineralization efficiency Effective for short-chain PFAS No need for addition of external chemicals | Scale-up is challenging Toxic byproduct formation Safety issues due to high voltage discharge for plasma generation |
Photocatalysis using nanomaterials (Ga2O3 and In2O3-based) | In situ production of reactive radicals Little or no consumption of chemicals Mineralization of the pollutants No sludge production Rapid degradation | Separation of catalyst material from treated water Limited light penetration in solution for reaction to occur Mass transfer limitations Susceptible to inactivation from presence of NOM and ions |
New generation adsorbents (COFs, MOFs, CNTs, and organically modified silica) | High chemical stability (COFs and MOFs) Pre-designable porous structure (COFs and MOFs) Effective for low concentrations Ease of operation High surface areas Tunable functionalities | Currently limited to laboratory scale Narrow pH range (MOFs)Relatively high synthesis cost (COFs and MOFs) Poor dispersibility in water (CNTs) Not effective for short-chain PFAS |
Treatment technology . | Advantages . | Disadvantages . |
---|---|---|
Activated carbon adsorption | Technically simple process and adaptable to many existing treatment plants Can be applied for large-scale operations Inexpensive process Extensive range of available commercial products Highly effective process with fast adsorption kinetics High quality of the treated effluent | Nondestructive method Not effective for short-chain PFAS Efficiency reduces in the presence of other constituents Costly reactivation process Requirement of further disposal of the adsorbent if not reactivated |
Ion-exchange resins | Technically simple operation Wide range of commercially available products Easy operation and maintenance Easy integration to other wastewater treatment processes Possibility of regeneration and reuse High quality of the treated effluent | Nondestructive method Economic limitations (high capital and maintenance cost of the selective resin and time-consuming regeneration process) Clogging of columns in the presence of particulates and organic matter, and may require a physicochemical pre-treatment Performance sensitive to pH Not very effective for short-chain PFAS Requirement of further disposal of the resin if not regenerated |
Electrooxidation | Destructive technology High mineralization efficiency No need for addition of external chemicals in most cases Effective for many different types of PFAS including short-chain Relatively inexpensive operation cost Simple operation Rapid degradation | High initial cost of the electrodes Cost of the maintenance (depends on electrode inertness and stability) Anode passivation and sludge deposition on the electrodes which can reduce the process efficiency in continuous mode of operation Foam formation during wastewater treatment Possible toxic byproducts formation (chlorate and perchlorate) in the presence of chloride Mostly suitable for small- or medium-sized communities |
Ultrasound | Destructive technology No need for addition of external chemicals No secondary pollution Simple operation No sludge production | High operational cost Management of generated heat energy Scale-up challenges Mostly suitable for small waste streams |
Plasma-based technologies | Destructive technology Inexpensive operation cost High mineralization efficiency Effective for short-chain PFAS No need for addition of external chemicals | Scale-up is challenging Toxic byproduct formation Safety issues due to high voltage discharge for plasma generation |
Photocatalysis using nanomaterials (Ga2O3 and In2O3-based) | In situ production of reactive radicals Little or no consumption of chemicals Mineralization of the pollutants No sludge production Rapid degradation | Separation of catalyst material from treated water Limited light penetration in solution for reaction to occur Mass transfer limitations Susceptible to inactivation from presence of NOM and ions |
New generation adsorbents (COFs, MOFs, CNTs, and organically modified silica) | High chemical stability (COFs and MOFs) Pre-designable porous structure (COFs and MOFs) Effective for low concentrations Ease of operation High surface areas Tunable functionalities | Currently limited to laboratory scale Narrow pH range (MOFs)Relatively high synthesis cost (COFs and MOFs) Poor dispersibility in water (CNTs) Not effective for short-chain PFAS |
CONCLUSIONS
PFAS consist of fluorinated hydrophobic fluorocarbon chains attached to a hydrophilic functional group and are very stable chemicals. The exclusive features of PFAS have led to their widespread industrial and domestic applications, predominantly in surfactants, fire-fighting foams, textile, and food-packing papers. These compounds do not undergo hydrolysis or photolysis, and their biodegradation under typical environments is very weak, which makes them difficult to remediate. Thus, the development of replacement chemicals for these compounds is vital. Several producers have begun to substitute long-chain PFAS with short-chain PFAS. However, short-chain PFAS have been linked to serious health and environmental effects, similar to their homologs. Regarding legislation of PFAS, while the United States Environmental Protection Agency (US EPA) has issued health advisory as 70 ng/L for the combination of PFOA and PFOS in drinking water, similar to the Australian guidelines, the Germany Ministry of Health projected guidance of 300 ng/L for both PFOA and PFOS. Furthermore, several US states have PFAS guidelines that range from 13 to 1,000 ng/L for PFOA and PFOS. PFAS regulations seem to be challenging due to considerable differences in characteristics among PFAS compounds, and lack of monitoring and toxicity studies. The occurrence of PFAS in industrial and domestic WWTPs has been extensively reviewed in this paper. Investigating the role of WWTPs will lead to a better understanding of the universal cycling of recalcitrant PFAS and in possible environmental regulations. This review suggests that industrial WWTP effluents are a major source of PFAS as compared to domestic wastewater effluents. Despite the phaseout of C8-based PFAS, PFOA and PFOS are still the dominant compounds in most WWTPs. Thus, well known and emerging precursor compounds that breakdown to PFCAs should also be identified and their physicochemical properties should be explored for improving the knowledge for the fate and transport of PFAS.
To mitigate associated negative impacts on humans and aquatic organisms, several treatment technologies have been reported. Commonly used PFAS treatment technologies include carbon adsorption and ion-exchange resins. Both, GAC and ion-exchange processes are efficient for the removal of PFAS from water; however, the presence of competing organics and ions can negatively affect these processes. These technologies have limited capacity to remove short-chain PFAS, and which can render them expensive. Research continues to improve the effectiveness of these processes and decrease operational cost. Emerging technologies, such as electrooxidation, plasma, sonochemical degradation, and new generation adsorbents, are being developed as viable options with promising results. Positive preliminary results at the laboratory scale and some pilot-scale studies were reported. Additional studies are desirable to conduct an assessment of the feasibility of emerging technologies at a larger scale and for complex wastewater streams. Due to the difficulty of in degradation and co-occurrence of many types of PFAS (molecular structure, size, charge, and functional groups), combined or hybrid treatment approaches may be required for effective removal/degradation of PFAS. Combined, hybrid or sequentially arranged treatment technologies may provide higher degradation efficiencies for PFAS, their precursors, and additional co-contaminants, and reduce the overall operational cost.
ACKNOWLEDGEMENT
This article is based upon research supported by the National Science Foundation (NSF), Water and Environmental Technology (WET) Center, and Temple University. The opinions expressed in this article are of the authors and do not necessarily reflect the views of WET Center or Temple University.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.