Emerging investigator series: photocatalytic treatment of PFAS in a single-step ultrafiltration membrane reactor

Abstract

Amidst the discovery of widespread per- and polyfluoroalkyl substances (PFAS) contamination and growing concerns of prolonged exposure even at low levels, many water treatment facilities are adopting reversed osmosis and nanofiltration processes to address these pollutants. Yet, these technologies are not sustainable, generating highly concentrated brines and requiring high operational pressures and energy inputs. Meanwhile, ultrafiltration (UF) membranes operate at less than 1 bar of transmembrane pressure (TMP) but are considered ineffective at removing organic pollutants. However, surface modifications make it possible to remove PFAS via UF. This study investigated the use of an adsorptive, photocatalytic, iron-enhanced titanium nanotube activated carbon composite coating on UF membranes to simultaneously remove and degrade PFAS in situ. In a photo-membrane reactor (PMR) under UV irradiation, the membranes removed up to 80% of the initial PFOA within 2 hours and the average removal over two 8-hour operation cycles was 69%. Although PFOA removal decreased to 35% when tested on a mixed PFAS solution, 46% of PFOS was still removed and 95% of the adsorbed PFOA was destroyed, while short-chain PFAS were removed to a lesser degree. This work provides a proof-of-concept of the PMR technology, which with further development could provide a single-step treatment for aqueous PFAS contamination in groundwater and pretreated surface and wastewaters.

---

Water impact

Sustainable solutions to PFAS contamination require its destruction, not just its removal from water. Combining ultrafiltration and advanced oxidation processes, a single-step reactor destroys aqueous PFAS in situ, allowing long-term reuse of the effective photocatalyst and consequently increasing its treatment capacity from a few milliliters to 2 L. This proof-of-concept paves the way for further development of PFAS-degrading technology.

1. Introduction

Per- and polyfluoroalkyl substances (PFAS) are a class of chemical compounds defined by their molecular structure, which consists of a fluorinated carbon chain tail and a functional group head. Many industrial and consumer products rely on PFAS precursor compounds due to their high thermal stability and surfactant properties.^1,2 However, research has correlated PFAS with suppression of the immune system as well as reproductive and developmental health risks in humans.^3–5 Of all the PFAS, polyfluoroalkyl carboxyl acids (PFCAs) and polyfluoroalkyl sulfonic acids (PFSAs) are the most studied and regulated groups, including perfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS), respectively.¹ Therefore, these chemicals have been largely banned in Europe and North America; nonetheless, as highly persistent chemicals, they will continue to circulate and accumulate in the environment, making the release of PFAS a global concern.⁶ Indeed, aqueous PFAS have been detected in waterways, oceans, and even remote regions of the Artic,^7,8 where slow mixing and sedimentation into the deep oceans are the only recognized environmental sinks.

Further complicating the problem, alternative PFAS have been developed to replace the banned compounds. Replacement PFAS often have a shorter carbon chain, which was thought to be less toxic, yet new research challenges this assumption.^9,10 Moreover, the shorter-chain PFAS (i.e., less than 6 or 7 perfluorinated carbons for PFSAs and PFCAs, respectively¹) are known to be more mobile in the environment, more difficult to capture using adsorptive technologies, and more resistant to degradation.^11–13 Thus, municipal waste landfills and wastewater treatment facilities (WWTFs), where conventional treatment processes are ineffective against PFAS,^14,15 release these contaminants to the environment and are thereby considered point sources.

Available water treatment technologies rely on separation to remove PFAS, but cannot destroy PFAS, thereby generating highly contaminated waste streams. While activated carbon (AC) and anion exchange (AIX) resins can successfully remove PFOA and PFOS,^11,16 the rising occurrence of short-chain PFAS limits their effectiveness.^17,18 Furthermore, spent AC and AIX media are regenerated using extremely high temperatures and organic solvents, respectively,^19,20 neither which are green processes. Alternatively, nanofiltration (NF) and reverse osmosis (RO) membranes are superior at removing PFAS of various molecular weights, including short-chain compounds such as perfluorobutane sulfonate (PFBS; C4),^21–23 especially when combined with adsorptive pretreatments.²⁴ However, NF and RO are energy intensive and generate large volumes of concentrated brine, which is difficult to handle. Thus, there is an urgent need to overcome the limitations of the current technologies by developing new, more sustainable treatment methods.

One possibility is ultrafiltration (UF), which operates at lower pressures, generates less concentrated brine, and requires significantly less energy than NF or RO. Due to their larger pore size, UF membranes are generally considered ineffective for PFAS removal,^11,25 yet surface modifications have been shown to improve PFAS rejection.²⁶ Olimattel et al. showed that multilayer coatings of polyallylamine hydrochloride and polyacrylic acid increased a commercial UF membrane's rejection of PFOA and PFOS by 30%.²⁷ By modifying UF membranes with an active layer of adsorptive photocatalyst, it should be possible to not only capture but also degrade PFAS in situ.

Several experiments have shown suspended photocatalysts' potential to degrade aqueous PFAS via generation of radical oxidation species (ROS).^13,28 Yet, photocatalyst development should also consider its sustainability, in terms of raw material availability, regeneration as opposed to one time use, and affordability. Due to its stable chemical properties, low toxicity, and low cost, the most widely used photocatalyst is titanium dioxide (TiO₂), but alone its effectiveness on PFAS is limited, requiring extensive treatment time.^29,30 Therefore, research into composite photocatalysts has aimed to enhance photoactivity by creating heterojunctions with other metals and to improve adsorption by combining with adsorptive materials. For example, the hydrothermal synthesis of titanium nanotubes on adsorptive activated carbon (TNT@ACs) in conjunction with post-transition metals has been proven to improve direct hole transfers, minimize recombination of electron–hole pairs, and keep PFAS in contact with the photocatalyst during degradation.^31–34 The most frequently proposed degradation mechanism is that the process is initiated by direct hole oxidation, which radicalizes the PFAS compound. Then, degradation proceeds with cleavage of the functional head group from the perfluoroalkyl chain, stepwise defluorination and chain shortening.^29,35,36 Although previous research has focused on PFOA and PFOS, analysis of degradation products indicates that short chain byproducts are also defluorinated under prolonged photocatalytic treatment after several hours.^37–39 However, the extensive treatment time and the need to separate the suspended photocatalysts after treatment present critical challenges to reusing the photocatalyst and this technology's general application.

These challenges can be overcome by immobilizing the adsorptive photocatalyst on an UF membrane, thereby combining catalytic treatment and separation in a single step. Yet, traditionally enclosed membrane cells must also be redesigned to allow UV light to irradiate the photocatalytic surface. Such a novel photocatalytic membrane reactor (PMR) can accept a continuous flow of contaminated water, while trapping the PFAS on the photocatalyst's surface under UV irradiation, which allows for their complete mineralization into benign components, e.g., fluoride ion (F⁻), carbon dioxide (CO₂), sulfate (SO₄²⁻), and water (H₂O). To demonstrate the technology, four photocatalysts were synthesized, immobilized to membranes, and compared for PFAS degradation in both batch experiments and filtration tests in the PMR. Further development of the photocatalytic membranes could meet the need for a zero-pollution technology to destroy PFAS, while being compatible with the mild operating conditions found at most water treatment facilities. This work provides a proof-of-concept of the PMR technology, which with further development could treat aqueous PFAS contamination in groundwater and pretreated surface and wastewaters.

2. Materials & methods

2.1 Chemicals & materials

See the ESI† for a detailed list of chemicals and materials (Text S1).

2.2 Synthesis of composite membranes

2.2.1 Photocatalyst synthesis. Iron-, indium-, and gallium-enhanced titanium nanotubes on activated carbon (TNT@AC) were selected as the adsorptive photocatalysts and synthesized. The hydrothermal synthesis and doping methods have been documented in detail by others^31,33,40 and are summarized in the ESI† (Text S2). Briefly, TNT@AC was formed from powdered active carbon and titanium dioxide mixed under basic conditions, then baked at 130 °C for 72 hours. The respective metal salts were added to TNT@AC particles in aqueous solution, then the precipitated solids were calcinated at 550 °C, washed and dried.

2.2.2 Photocatalytic membrane synthesis. Flat sheet polyethersulfone (PES) membranes with a molecular weight cut-off (MWCO) of 20 000 Da were purchased from Mann & Hummel (Nadir® UP020 P). Active sites were added to the membrane surface via a plasma-activated acrylic-acid (AA) graft or a polydopamine (PDA) graft. Then, the photocatalyst was suspended in a 50 : 50 ethanol–ultrapure water solution and deposited via vacuum filtration onto the grafted membranes. Membrane modifications are described in the ESI† (Text S3) and a summary of the synthesized membranes is provided (Tables S1 and S2†). For comparison, the as-prepared photocatalyst was also loosely loaded on a commercial nanofiltration (NF) membrane (Alfa Laval NF, thin film composite of polyester, MWCO = 300 Da) and tested in the reactor.

2.3 Batch tests

2.3.1 Light source. For the batch photodegradation experiments, two UV light sources were selected to excite electrons in the TiO₂ photocatalysts. The LZC-4Xb photoreactor (LuzChem Research, Canada) is a closed photoreactor, equipped with 14 lamps (8 W each) emitting 254 nm light at a measured power density of 4.49 mW cm⁻². Alternatively, three UV chips (PureFize Technologies AB, Sweden) were used, due to their flexible placement and minimal heat generation. The UV chips emit broadband UV irradiation (255–300 nm; peak wavelength: 262 nm) but use less power (500 mW); the total power density emitted from the 3 chips is 6.5 mW cm⁻².

2.3.2 Batch slurry tests. Batch tests were conducted in slurry, where the photocatalysts were suspended in lab-contaminated water and irradiated, to compare photodegradation and defluorination efficiencies. Details of the batch experiments are provided in the ESI† (Text S4).

2.3.3 Static membrane tests. Batch testing of the photocatalytic UF membranes was conducted to compare their photodegradation and defluorination efficiencies. A coated membrane was placed in a Petri dish containing 40 mL of PFOA solution (100 ppb) and covered with a ¼ inch thick quartz cover to mimic the PMR conditions. Three UV chips were placed directly on the quartz cover and the setup was irradiated for 3 hours. After irradiation, samples were prepared for F⁻ and PFAS analyses. The remaining solution was collected and stored in polypropylene (PP) tubes at 4 °C for trace metals analysis via inductively coupled plasma optical emission spectroscopy (ICP-OES, Optima 4300 DV, PerkinElmer Inc.). Between treatment cycles, membranes were rinsed with ultrapure water and stored in the dark at 4 °C.

2.4 Photocatalytic membrane reactor

2.4.1 Reactor design and setup. A lab-made reactor, consisting of a stainless-steel cell with a ¼ inch thick quartz window, threaded inlet and outlet tubes (6 mm OD), nitrile O-rings and silicon seals, was used to test the photocatalytic membrane (45 mm diameter). A highly porous PP sheet (6 mm, Vyon® HP) supported the membrane from below, with approximately 1 cm distance between the quartz window and the membrane. Pressure-rated silicon tubing (St. Gobain, 6 mm and 10 mm inner diameter) connected the reactor to the feed tank (5 L HDPE carboy, Kautex), gear pump (MCP-Z drive with Z-200 pump head, Ismatec), turbine flow meter (0.05 to 10 L min⁻¹, RS Components A/S), pressure sensors (IPS-M12 Series with 16 bit ADS1115, RS Components A/S), stainless steel needle valve (¼ inch, ESSKA Teknik AB), and shut-off valves (Festo, HE Manual). All elements were rated for at least 5 bars of pressure. A lab-programmed data logger (Raspberry Pi 4) collected flow meter and pressure sensor readings every 1.5 seconds. The 3 UV chips were placed on top of the quartz window. The PMR was enclosed in a cabinet to prevent interference from other light sources.

2.4.2 Continuous PMR tests. After inserting the photocatalytic membrane, ultrapure water flooded the PMR to prime the system. Meanwhile, 2 L of feed solution was prepared by diluting PFOA stock solution to 10 ppb with ultrapure water. Then, the feed tank was connected to the system and slowly pumped to remove any air. To maximize PFAS' interaction with the membrane, the setup was operated in dead-end mode by closing the outlet valve and recirculating the permeate (Fig. 1). The pump rate was adjusted to the desired pressure (i.e., 0.8 ± 0.05 bar). Finally, the UV chips were switched on and the experimental time (t = 0) began. Each trial ran for two consecutive, 8-hour cycles (each equivalent to 4 hours of irradiation, when operating the light in on/off mode), where the feed solution was renewed each cycle.

Fig. 1 Schematic of the photocatalytic membrane reactor setup operating in dead-mode.

At regular intervals, 2 mL samples were collected from the permeate line. Initial and final feed samples were collected from the feed tank, while PMR drain samples were taken at the end of the membrane trial to evaluate the internal concentration. The cumulative removal rate (R) was calculated using the following equation, where C_p and C_f represent the permeate and initial feed concentrations, respectively:

While the data logger continuously monitored flow and pressure, the permeate was periodically collected in a graduated cylinder to verify the flux. The following equation was used to calculate the permeate flux (J_v) from the volumetric flow rate of the permeate (dV_p/dt), and the area of the membrane (A_m):

Between experiments, the system was cleaned, as described in the ESI† (Text S5). A sample was collected at the end of the cleaning cycle to verify that there was no contamination of the system. The same procedure was used to test the PMR treatment of a mixed PFAS solution, consisting of 20 ppb of each PFOA, PFOS, PFBA, and PFBS, as well as the NF membrane tests. NF membranes were not grafted, and the photocatalyst was loosely loaded to the surface by vacuum filtration. All PMR experiments were conducted in duplicate and data is presented as an average.

2.5 Analysis & characterization

Descriptions of the techniques used to analyze water samples and to characterize the photocatalytic membranes are provided in the ESI† (Text S6 and Table S3).

3. Results and discussion

3.1 Batch photodegradation of PFOA

All synthesized photocatalysts exhibited consistently high adsorption rates, with 90% or more of initial PFOA removed from solution after 30 minutes (Table S4†). After photo-irradiation, shorter chain PFPrA was only observed at less than 1 ppb in the irradiated solutions, while >5 ppb of PFHxA and PFBA were observed in solid phase extracts of the photocatalysts, in addition to adsorbed PFOA. Meanwhile, the ion chromatograph revealed F⁻ as a degradation product, along with acetate (CH₃COO⁻) and formate (HCOO⁻). The Fe/TNT@AC samples had the highest defluorination rates: 34.3% defluorination when irradiated by the PureFize® UV chips and 53.9% defluorination when irradiated in the LuzChem® photoreactor (Fig. 2). The lower defluorination rate by the UV chips is explained by the lower energy input from the broadband UV light, which did not sufficiently excite the photocatalyst. Meanwhile, both In/TNT@AC and Ga/TNT@AC samples had insignificant defluorination rates, i.e., <10%, under both light sources. No F⁻ peak was observed for the TNT@AC samples.

Fig. 2 Defluorination rates for PFOA solution (C₀ = 200 ppb) treated with different photocatalysts (1 g L⁻¹) under 4 hours of UV irradiation from the LuzChem® or PureFize® light source.

Several controls were conducted in conjunction with the current study, where PFOA was treated with either UV light alone (Table S4†) or adsorption onto the catalyst without irradiation (Fig. 2). For UV only treatment, there was no significant change in PFOA concentration though trace amounts of perfluorohexanoic acid (PFHxA) were observed; furthermore, no F⁻, CH₃COO⁻ or HCOO⁻ was detected. Similarly, neither F⁻ nor CH₃COO⁻ was detected in the unirradiated samples; however, Cl⁻ was observed, probably due to the hydrochloride acid used for pH adjustment during catalyst synthesis, and a trace amount of HCOO⁻ was also observed, which could be related to organic groups present on the activated carbon's surface. Meanwhile, PFAS were below the detection limits, indicating that the catalysts had adsorbed all the initial PFOA. In conjunction with the batch experiments, these controls proved that irradiation of the synthesized catalyst was necessary for photodegradation and defluorination to proceed.

The detection of some shorter-chain PFAS byproducts but no F⁻ was expected for the plain TNT@AC, indicating incomplete degradation. Meanwhile, the metal-enhanced TNT@AC can facilitate the attachment of PFAS molecules to the photocatalyst through the creation of positively charged centers (e.g., Fe^2+/3+) and oxygen vacancies, thereby improving photoactivity through direct charge transfers and suppressed recombination of electron–hole pairs.^32,36 For instance, Wang et al. (2008) proposed that complexation with Fe³⁺ precedes the degradation process of PFOA. After complexation, photolysis yields Fe²⁺ and PFOA radical, thereby initiating decarboxylation and progressive chain shortening, while the remaining Fe²⁺ ions are quickly oxidated to Fe³⁺.⁴¹ Other researchers have documented the complexation of PFOS and Fe³⁺via the changed electronic state of the complex, as detected by electron spin resonance (ESR) spectrometry, and indirectly through reduced UPLC-MS response for PFOS in ferric sulfate solution.⁴² Similar bidentate bonding between indium oxide and PFOA's carboxylate group has also been observed,⁴⁰ albeit a rather low defluorination rate by In/TNT@AC.

In addition to F⁻, IC analysis detected HCOO⁻ and CH₃COO⁻ (anionic form of formic and acetic acids). While formic acid formation has been reported by others, who postulated that formic acid was a byproduct from the decarboxylation of PFOA,^41,43–45 CH₃COO⁻ has not previously been described as a byproduct of oxidative PFOA photodegradation. The photodegradation mechanism leading to its formation requires further clarification.

Due to its superior defluorination performance, additional tests were run on the Fe/TNT@AC, including a batch degradation experiment with a lower initial PFOA concentration of 50 ppb. Following adsorption and UV irradiation, no PFAS was detected in the sample, but F⁻ and CH₃COO⁻ were. The calculated defluorination results were 109–122%, which indicates very high to complete defluorination, when PFAS recovery error is considered.⁴⁶ Therefore, Fe/TNT@AC was selected as the most effective photocatalyst.

3.2 Static membrane tests

When the photocatalytic membranes were tested, both grafting methods showed a diminishing trend of PFOA removal from the first to third treatment cycle. In the first irradiation treatment cycle, AA-grafted membranes removed 34–46% of PFOA, while PDA-grafted membranes removed 37–41% of PFOA (Fig. 3). In the second cycle, the PDA-grafted membranes generally performed better than the AA-grafted membranes, exhibiting 30–38% removal versus only 16–17% removal. However, in the third cycle, the adsorption capacities of the In/TNT@AC and Ga/TNT@AC coated membranes were exhausted, exhibiting zero to negative removals for both grafting methods. During the third-cycle run, the Fe/TNT@AC-PDA membrane removed 4% PFOA, which was higher than the AA-graft (1.4% removal), but still insignificant after accounting for instrument variability. While PFOA was measured in the treated solutions, short-chain byproducts were not detected. Therefore, it can be concluded that all readily accessible adsorption sites are occupied by the third cycle and that, despite UV irradiation, these sites were not regenerated quickly enough to adsorb more PFOA.

Fig. 3 Percent removal of PFOA (C₀ = 200 ppb) after 4 hours of UV irradiation of various photocatalysts on membranes grafted with acrylic acid (a) or polydopamine (b), tested cyclically.

The higher PFOA removal rate by the PDA-grafted membranes than the AA-grafted membranes is attributed to the higher photocatalyst loading achieved using this method. The AA-grafted membranes had an average catalyst loading of 0.13, 0.18, and 0.19 mg cm⁻² for the Fe-, In-, and Ga/TNT@AC, respectively, whereas PDA-grafted membranes had an average catalyst loading of 0.94, 0.63, and 0.63 mg cm⁻² for the Fe-, In-, and Ga/TNT@AC, respectively. In comparison to the slurry tests, where 90–100% of PFOA was removed using 40 mg of catalyst, AA-grafted membranes removed up to 46% PFOA with just 3–4 mg of catalyst, while PDA-grafted membranes removed 30–40% PFOA during the first two cycles with 10–15 mg of catalyst. Tran et al. (2020) achieved similar catalyst loading (0.13–0.70 mg cm⁻²) of titania–graphene oxide nanoparticles applied to the PVDF membrane, yet found that higher catalyst loading did not correlate with increased photoactivity, presumably due to light blocking effects.⁴⁷ Accordingly, photoactive sites (e.g., TNTs and metal dopants) not located at an exposed surface may adsorb PFAS but are not photo-accessible and cannot be regenerated.³¹ Thus, the static membrane tests showed that significant adsorption can still be achieved using drastically less catalyst, but sustained removal is constrained by the time required to regenerate active sites.

Trace metal analyses of the treated solutions did not detect Ti leaching in most scenarios. However, for the Ga/TNT@AC-PDA coated membrane, up to 0.96 mg L⁻¹ of Ti was leached. Minimal leaching was also observed from the TNT@AC-PDA and In/TNT@AC-PDA coated membranes, at 0.11 and 0.10 mg L⁻¹, respectively. The Fe/TNT@AC-AA membrane exhibited some leaching of Fe (0.12 mg L⁻¹) and Ti (0.15 mg L⁻¹) in the first cycle, but no further leaching was observed in subsequent cycles.

As a result of the larger range of catalyst loadings achieved and generally higher PFOA removal, the PDA graft was selected as the preferred immobilization method. Furthermore, no metal leaching was observed for this method, when combined with the superior catalyst, Fe/TNT@AC, so this combination was selected for the subsequent membrane characterization and PMR studies.

3.3 Photocatalyst and membrane characterization

Further characterization of the best performing photocatalyst (Fe/TNT@AC) and grafting method (PDA-graft) was conducted. SEM images of the Fe/TNT@AC particles (Fig. 4a) showed a fluffy appearance on the flat sheet-like surfaces of the activated carbon, consistent with previous reports.^31,33,40 The uncoated PES membrane exhibited a flat surface (Fig. 4b), where pores were typically too small to see in detail.^48,49 For the coated membrane, the accumulation of bulky catalyst particles was observed over the dark patches, where the membrane surface is uncoated (Fig. 4c). From SEM scans, the smallest Fe/TNT@AC particles were on the order of 50 nm wide. Thus, it is unlikely that the photocatalyst particles clogged pores or reduced the permeate flux, as the UF pore size was on the order of 10–20 nm.

Fig. 4 SEM images of Fe/TNT@AC catalyst particles (a), uncoated membrane (b) and coated membranes (c). SEM–EDS analysis of the elemental composition of the Fe/TNT@AC yielded the elemental composition plot (d).

SEM–EDS analysis of the Fe/TNT@AC particles estimated the relative weight of iron to be 0.7%, slightly below the 1% target (Fig. 4d). As main components of the catalyst, carbon and titanium had the highest relative weights, while oxygen was present primarily in functional groups on the activated carbon surface but may also be present as metal oxides. Finally, aluminum and silica represented impurities in the activated carbon or metal components.

Topographical AFM scans of the pristine, grafted, and coated membranes allowed for an investigation of pore sizes, coating particle sizes, and surface roughness. In the center of the pristine membrane image (Fig. 5a), a double pore was observed, while at least 3 distinct pores were captured in the PDA-grafted membrane image (Fig. 5c). The pores appeared oblong, which is consistent with the method of stretching membranes to form pores, while diameters narrow with depth (Fig. 5b and d). Although no change was observed in pore diameter, the PDA-graft resulted in a rougher surface than the pristine membrane, as indicated by shadows and globular structures in 2D topographical images (Fig. 5a and c) and roughness height histograms (Fig. S1 and S2†). Meanwhile, AFM images revealed that the catalyst particles are stacked on top of each other. Due to the large particle size and stacking, the surface roughness increased significantly, resulting in unresolved magnetic striations (Fig. 5f), seen as lines across the image (Fig. 5e), and increased recorded heights (Fig. S3†). The rougher surface may increase PFOA's interactions with the surface, providing access to additional adsorption sites; however, it could also increase the membrane's fouling potential.

Fig. 5 Atomic force microscopy (AFM) topographical scans of the pristine membrane (a and b), the PDA-grafted membrane (c and d), and Fe/TNT@AC-coated membranes (e and f). a, c and e are 2D images and b, d and f are surface profiles, which correspond to the blue line in the image read left to right.

The porometer results estimated the pore size of the unmodified membranes to be 20–24 nm (Table 1); however, the instrument was operated at its pressure limit (e.g., 13 nm diameter equivalent). The measurements of the dopamine grafted membrane were less conclusive and without a definitive cutoff for the smallest pore size, suggesting that the graft may reduce the diameter or reshape some pores. After coating with Fe/TNT@AC, the largest pore diameter encountered was 40 nm, though the mean size was still 24 nm. The larger pore observed could be due to the effects of drying, which can crack the membrane, during the vacuum deposition of the catalyst. Nonetheless, the observed pore sizes ruled out size exclusion of PFOA as the main removal mechanism by the membranes.

Table 1 Summary of membrane characteristics for the pristine, grafted, and coated UF membranes

Sample	Smallest pore diameter (nm)	Largest pore diameter (nm)	Surface potential (mV)	Water contact angle
Unmodified membrane	20	24	−29.3	74–64°
PDA-grafted membrane	Inconclusive	21	−29.0	71–67°
Fe/TNT@AC coated membrane	23	40	−16.8	84–67°

---

Streaming surface potential analysis indicated that there was no significant change between the unmodified and the PDA-grafted PES membrane samples, −29.3 and −29.0 mV, respectively (Table 1). Negative surface charge is a common issue for polymeric membranes, which typically repels PFAS anions. However, the positive charges in the catalytic coating raised the overall surface potential to −16.8 mV. Finally, water contact angle measurements were used to classify the membranes as hydrophobic or hydrophilic (Table 1). While all membranes were characterized as hydrophilic, the coated membranes also had the highest contact angle measurements, up to 84°, indicating the addition of hydrophobic groups from the aromatic skeleton of the activated carbon. By introducing localized positive charges and hydrophobic interactions, the photocatalytic coating enabled PFAS adsorption onto the membrane surface.

3.4 PMR tests

Given that concentrations encountered in water treatment applications are several orders of magnitude lower than in batch photodegration tests, the initial PFOA concentration was set to 10 ppb for the PMR studies. Meanwhile, the targeted catalyst loading was set to 20 mg per membrane; higher loadings were limited by the coating becoming too loose and fragile, resulting in significant catalyst losses from the setup during the PMR startup. To determine the extent to which overall removal was due to adsorption versus degradation, PFAS extractions were performed on the photocatalytic membranes after PMR tests; the results are presented in section 3.5.

The coated UF membranes removed 80% of PFOA in the first 2 hours of the first cycle, with removal rates diminishing to 41% after 8 hours of operation. In the second cycle, the average removal rate was 69%, remaining relatively constant throughout 8 hours of operation (Fig. 6a). These PFOA removal rates are significantly higher than those observed for commercial UF membranes, which are typically below 25% removal.^11,25,27 Meanwhile, the TMP and the permeate flux were stable during both cycles (Fig. 6b). The TMP averaged 0.83 bar, while the permeate flux averaged 222 LMH. Thus, nearly the entire 2 L of feed solution was circulated through the setup during each cycle. Neither fouling nor a loss of permeability over the operation period was observed.

Fig. 6 Cumulative PFOA removal (C₀ = 10 ppb) by coated ultrafiltration membranes (control is the unmodified ultrafiltration membrane) (a) and corresponding permeate flux and transmembrane pressure (b); and by loaded nanofiltration membranes (control is the unmodified nanofiltration membrane) (c) and corresponding permeate flux and transmembrane pressure (d).

In comparison, NF membranes loaded with Fe/TNT@AC were run in the same PMR setup and removed 87–100% of PFOA, with an average removal of 95% (Fig. 6c). However, the permeate flux was significantly lower, 12 LMH, when operating with an average TMP of 0.95 bar (Fig. 6d). At this low flux, only about 130 mL passed through the setup per cycle.

Controls were conducted using both unmodified UF and NF membranes in the PMR. In the UF control, the initial 39% removal diminished to 0% after 4 hours, after which a negative removal rate was observed. The NF control exhibited a similar trend, with an initial 89% removal rate decreasing to less than 20% removal after 4 hours. One explanation for diminishing rejection rates is the aggregation of amphiphilic PFAS in pores, eventually leading to diffusion across the membrane.^50,51 Additionally, exposure to UV light can degrade the polymeric membranes, thereby increasing the effective pore size. Discoloration and delamination of the thin film barrier from the support material were observed on the uncoated NF membrane after 24 hour UV exposure, but no visible changes were observed for the uncoated UF membrane. Thus, the photocatalytically active layer both shielded the polymeric surface from destructive UV radiation and added an adsorptive layer to trap PFAS.

When a feed solution of mixed PFAS was run through the PMR, the removal of PFOA was slightly lower and showed a similar removal rate for PFOS. Over time the removal rates increase to 68% of PFOA and 79% of PFOS in the first cycle (8 h), but then begin to diminish in the second cycle to 17% and 23%, respectively, indicating that the adsorptive sites are saturated and are not regenerated fast enough for more uptake (Fig. 7a). Over two cycles, the average removal of PFOA and PFOS was 35% and 46%, respectively, while short-chain compounds PFBA and PFBS were 1% and 6%, respectively. Permeate flux and pressure remained constant throughout operations (Fig. 7b), as in the single-analyte PMR treatment tests.

Fig. 7 Cumulative removal of various PFAS (a), permeate flux and transmembrane pressure (b) for the coated ultrafiltration membrane operated in the PMR for two 8-hour cycles.

3.5 PFAS mass balance

Following the filtration tests, a mass balance of PFAS was conducted in order to understand how the contaminants was removed in the PMR system. The total quantity of PFAS removed included both the PFAS adsorbed onto the membrane and degraded in the PMR. The degraded portion includes detected PFAS byproducts, either adsorbed onto the catalyst surface or in the permeate solution, as well as F⁻. However, only known PFAS byproducts (Table S3†) could be quantified, and even theoretical F⁻ concentrations were below the detection limits. Additional losses may be a result of instrument error or adsorption to lab equipment, although no PFAS was detected in any of the rinse water used to clean the setup in between trials.

After two consecutive cycles, adsorbed PFAS was extracted from the photocatalytic membranes. The total PFOA loaded to the system was estimated by multiplying the initial feed concentration and the permeate flux, if it was less than the total feed volume (2 L). When PFAS adsorbed onto the loaded NF membrane was extracted and quantified, only 30% of the total PFOA load was recovered, meaning 57% of the PFOA was unaccounted for and likely completely degraded, as shorter-chain PFAS intermediates were not identified in the extract nor the permeate solution.

Similarly, analyses of solid phase extractions from the Fe/TNT@AC-coated UF membranes revealed that 2%, 1%, and 4% of the total PFOA, PFBA and PFBS was adsorbed, compared to up to 44% of PFOS (Fig. 8). No other known PFAS byproducts were observed, indicating that 95% of the adsorbed PFOA had been destroyed, while the PFOS was mainly removed through adsorption but not degraded. These results are consistent with trends reported in the literature, namely, that long-chain PFCAs adsorb more easily to activated carbon than short-chain counterparts and that PFSAs are preferentially adsorbed, in some cases competing with PFCAs, but are generally more difficult to degrade.

Fig. 8 Mass balance for the mixed PFAS solution, comparing the portions adsorbed and degraded to the average removal by the polydopamine-grafted, Fe/TNT@AC-coated PES ultrafiltration membrane.

3.6 Mechanism of PFAS degradation

To elucidate the degradation mechanism, batch quenching experiments were performed to determine the effective photogenerated radicals. The batch photodegradation experiments with Fe/TNT@AC were repeated in the Luzchem® photoreactor reactor, but this time with the addition of either isopropanol, potassium iodine (KI), or benzoquinone to scavenge for hydroxyl radicals (OH˙), holes (h⁺), or superoxide radicals (O₂˙⁻), respectively. The quenching agent concentrations used were 5 mM isopropanol, 1 mM KI, and 1 mM benzoquinone, which ensured complete consumption of the generated radical species.^31,33,34

The IC analysis showed 42%, 32% and 26% defluorination for the unquenched, isopropanol-quenched, and KI-quenched samples, respectively (Fig. S7†). While the decline in the defluorination rate, i.e., about 10% for 1 mM isopropanol, matches the results of OH˙ quenching previously described for this catalyst,³¹ the 16% inhibition by KI is significantly less pronounced than results in previous studies. Nevertheless, the results imply that h⁺ and OH˙ were important radical species during PFOA defluorination, consistent with the previous observation of h⁺ as the primary radical species and OH˙ as the secondary radical species.^28,52 The Fe/TNT@AC was able to adsorb all the initial PFOA, but the addition of benzoquinone reduced PFOA removal down to 78% suggesting the competition for adsorption sites between benzoquinone and PFOA. Therefore, it is not clear whether the non-detectable fluorine ions stemmed from the inhibition of superoxide radicals. Besides the direct oxidation by h⁺, the OH˙ may be involved in the oxidation state of the iron active sites: Fe³⁺ can complex with PFAS and, when irradiated, may produce Fe²⁺ and a perfluoroalkyl radical, mediated through metal-to-ligand electron transfer.^41,53 However, the reduced iron must be re-oxidized, potentially by OH˙, for this mechanism to continue.

4. Conclusions

The current study has shown that it is possible to remove and degrade PFAS using a photocatalytic UF membrane in a PMR system operating at a low pressure and under mild conditions. Immobilization of the photocatalyst on the membrane surface resulted in significant PFOA and PFOS removal, even though the catalyst loading represented only a fraction of the dose used in the batch slurry tests, and no additional steps were required to cyclically reuse the catalyst. The carbon-based photocatalyst coating seems to mitigate the fouling without a significant drop in permeate flux. Moreover, because the composite photocatalyst is effective at adsorbing and holding PFOA and PFOS in place during photodegradation, up to 95% of the PFOA removed by the modified UF membranes was destroyed in situ, though PFOS was more recalcitrant to degradation.

However, the PFAS removal efficiency decreased as PMR operations continued, plateauing or diminishing in the second cycle. After initial adsorption, the rate of degradation must keep up with the permeation rate so that photoactive sites are regenerated and available to capture more PFAS from the feed solution. Adjusting the permeate flux to increase the residence time, and thereby the contact time, could potentially improve the consistency of PFAS removal over time.⁵⁴ Alternatively, investigating and optimizing the wavelengths emitted by the UV light source could contribute to more efficient photocatalysis. Additionally, improving the photocatalyst's degradation efficiency and kinetics could overcome this challenge, making the PMR technology appropriate for continuous treatment. Decreasing the photocatalyst's particle size would further improve the photocatalytic coating's robustness, making it less susceptible to particle loss due to rubbing or abrasion. Finally, other strategies, such as operating PMRs in series, must be considered to capture short-chain PFAS from aqueous solution, as the PFOA and PFOS in the mixed PFAS solution outcompeted the PFBA and PFBS for adsorption sites.

Further development of the PMR technology, as a proof-of-concept, could provide a potential solution to achieve the complete mineralization of PFAS in aqueous solutions. Future research should investigate the potential of the PMR technology to degrade other micropollutants and its effectiveness in treating real environmental samples. When considering more complex water matrices, surface fouling may arise, which could be mitigated by operating the PMR in crossflow mode, increasing the membrane's longevity in real world applications.

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported by Aarhus University Centre for Water Technology (AU-WATEC) Start-Up Fund from Grundfos Foundation, Aarhus University Research Foundation Starting Grant (No. AUFF-E-2019-7-28), Novo Nordisk Fonden (No. NNF20OC0064799), and Independent Research Fund Denmark Sapere Aude Award (No. 1051-00058B). The authors would like to thank Professor Lars Rosengaard Jensen (Aalborg University) for his efforts to characterize the fabricated membranes using Raman spectroscopy.

References

1. R. C. Buck, J. Franklin, U. Berger, J. M. Conder, I. T. Cousins and P. de Voogt, et al., Perfluoroalkyl and polyfluoroalkyl substances in the environment: Terminology, classification, and origins, Integr. Environ. Assess. Manage., 2011, 7(4), 513–541.
2. Z. Wang, J. M. Boucher, M. Scheringer, I. T. Cousins and K. Hungerbühler, Toward a comprehensive global emission inventory of C4–C10 perfluoroalkanesulfonic acids (PFSAs) and related precursors: Focus on the life cycle of C8-based products and ongoing industrial transition, Environ. Sci. Technol., 2017, 51(8), 4482–4493.
3. S. E. Fenton, A. Ducatman, A. Boobis, J. C. DeWitt, C. Lau and C. Ng, et al., Per- and polyfluoroalkyl substance toxicity and human health review: Current state of knowledge and strategies for informing future research, Environ. Toxicol. Chem., 2021, 40(3), 606–630.
4. A. J. Blomberg, Y. H. Shih, C. Messerlian, L. H. Jørgensen, P. Weihe and P. Grandjean, Early-life associations between per- and polyfluoroalkyl substances and serum lipids in a longitudinal birth cohort, Environ. Res., 2021, 200, 111400.
5. L. Dalsager, N. Christensen, U. Halekoh, C. A. G. Timmermann, F. Nielsen and H. B. Kyhl, et al., Exposure to perfluoroalkyl substances during fetal life and hospitalization for infectious disease in childhood: A study among 1,503 children from the Odense Child Cohort, Environ. Int., 2021, 149, 106395.
6. I. T. Cousins, J. C. DeWitt, J. Glüge, G. Goldenman, D. Herzke and R. Lohmann, et al., The high persistence of PFAS is sufficient for their management as a chemical class, Environ. Sci.: Processes Impacts, 2020, 22(12), 2307–2312.
7. H. Joerss, Z. Xie, C. C. Wagner, W. J. von Appen, E. M. Sunderland and R. Ebinghaus, Transport of legacy perfluoroalkyl substances and the replacement compound HFPO-DA through the Atlantic Gateway to the Arctic Ocean: Is the Arctic a sink or a source?, Environ. Sci. Technol., 2020, 54(16), 9958–9967.
8. L. Ahrens, S. Taniyasu, L. W. Y. Yeung, N. Yamashita, P. K. S. Lam and R. Ebinghaus, Distribution of polyfluoroalkyl compounds in water, suspended particulate matter and sediment from Tokyo Bay, Japan, Chemosphere, 2010, 79(3), 266–272.
9. M. I. Gomis, Z. Wang, M. Scheringer and I. T. Cousins, A modeling assessment of the physicochemical properties and environmental fate of emerging and novel per- and polyfluoroalkyl substances, Sci. Total Environ., 2015, 505, 981–991.
10. S. Brendel, É. Fetter, C. Staude, L. Vierke and A. Biegel-Engler, Short-chain perfluoroalkyl acids: environmental concerns and a regulatory strategy under REACH, Environ. Sci. Eur., 2018, 30(1), 1–11.
11. T. D. Appleman, C. P. Higgins, O. Quiñones, B. J. Vanderford, C. Kolstad and J. C. Zeigler-Holady, et al., Treatment of poly- and perfluoroalkyl substances in U.S. full-scale water treatment systems, Water Res., 2014, 51, 246–255.
12. M. Ateia, A. Maroli, N. Tharayil and T. Karanfil, The overlooked short- and ultrashort-chain poly- and perfluorinated substances: A review, Chemosphere, 2019, 220, 866–882.
13. F. Li, J. Duan, S. Tian, H. Ji, Y. Zhu and Z. Wei, et al., Short-chain per- and polyfluoroalkyl substances in aquatic systems: Occurrence, impacts and treatment, Chem. Eng. J., 2020, 380, 122506.
14. U. Eriksson, P. Haglund and A. Kärrman, Contribution of precursor compounds to the release of per- and polyfluoroalkyl substances (PFASs) from waste water treatment plants (WWTPs), J. Environ. Sci., 2017, 61, 80–90.
15. X. C. Hu, D. Q. Andrews, A. B. Lindstrom, T. A. Bruton, L. A. Schaider and P. Grandjean, et al., Detection of poly- and perfluoroalkyl substances (PFASs) in U.S. drinking water linked to industrial sites, military fire training areas, and wastewater treatment plants, Environ. Sci. Technol. Lett., 2016, 3(10), 344–350.
16. N. Belkouteb, V. Franke, P. McCleaf, S. Köhler and L. Ahrens, Removal of per- and polyfluoroalkyl substances (PFASs) in a full-scale drinking water treatment plant: Long-term performance of granular activated carbon (GAC) and influence of flow-rate, Water Res., 2020, 182, 115913.
17. S. P. Albu, A. Ghicov, J. M. Macak, R. Hahn and P. Schmuki, Self-organized, free-standing TiO₂ nanotube membrane for flow-through photocatalytic applications, Nano Lett., 2007, 7(5), 1286–1289.
18. P. McCleaf, S. Englund, A. Östlund, K. Lindegren, K. Wiberg and L. Ahrens, Removal efficiency of multiple poly- and perfluoroalkyl substances (PFASs) in drinking water using granular activated carbon (GAC) and anion exchange (AE) column tests, Water Res., 2017, 120, 77–87.
19. B. Sonmez Baghirzade, Y. Zhang, J. F. Reuther, N. B. Saleh, A. K. Venkatesan and O. G. Apul, Thermal Regeneration of Spent Granular Activated Carbon Presents an Opportunity to Break the Forever PFAS Cycle, Environ. Sci. Technol., 2021, 55(9), 5608–5619.
20. T. H. Boyer, A. Ellis, Y. Fang, C. E. Schaefer, C. P. Higgins and T. J. Strathmann, Life cycle environmental impacts of regeneration options for anion exchange resin remediation of PFAS impacted water, Water Res., 2021, 207, 117798.
21. E. Steinle-Darling and M. Reinhard, Nanofiltration for trace organic contaminant removal: Structure, solution, and membrane fouling effects on the rejection of perfluorochemicals, Environ. Sci. Technol., 2008, 42(14), 5292–5297.
22. J. Xiong, Y. Hou, J. Wang, Z. Liu, Y. Qu and Z. Li, et al., The rejection of perfluoroalkyl substances by nanofiltration and reverse osmosis: influencing factors and combination processes, Environ. Sci.: Water Res. Technol., 2021, 7(11), 1928–1943.
23. E. W. Tow, M. S. Ersan, S. Kum, T. Lee, T. F. Speth and C. Owen, et al., Managing and treating per- and polyfluoroalkyl substances (PFAS) in membrane concentrates, AWWA Water Sci., 2021, 3(5), e1233.
24. V. Franke, M. Ullberg, P. McCleaf, M. Wålinder, S. J. Köhler and L. Ahrens, The Price of Really Clean Water: Combining Nanofiltration with Granular Activated Carbon and Anion Exchange Resins for the Removal of Per- And Polyfluoralkyl Substances (PFASs) in Drinking Water Production, ACS ES&T Water, 2021, 1(4), 782–795.
25. B. K. Pramanik, S. K. Pramanik, D. C. Sarker and F. Suja, Removal of emerging perfluorooctanoic acid and perfluorooctane sulfonate contaminants from lake water, Environ. Technol., 2017, 38(15), 1937–1942.
26. S. Das and A. Ronen, A Review on Removal and Destruction of Per- and Polyfluoroalkyl Substances (PFAS) by Novel Membranes, Membranes, 2022, 12(7), 662.
27. K. Olimattel, L. Zhai and A. H. M. A. Sadmani, Enhanced removal of perfluorooctane sulfonic acid and perfluorooctanoic acid via polyelectrolyte functionalized ultrafiltration membrane: Effects of membrane modification and water matrix, J. Hazard. Mater. Lett., 2021, 2, 100043.
28. M. Gar Alalm and D. C. Boffito, Mechanisms and pathways of PFAS degradation by advanced oxidation and reduction processes: A critical review, Chem. Eng. J., 2022, 450.
29. S. C. Panchangam, A. Y.-C. Lin, K. L. Shaik and C.-F. Lin, Decomposition of perfluorocarboxylic acids (PFCAs) by heterogeneous photocatalysis in acidic aqueous medium, Chemosphere, 2009, 77(2), 242–248.
30. J. Zhong, Y. Zhao, L. Ding, H. Ji, W. Ma and C. Chen, et al., Opposite photocatalytic oxidation behaviors of BiOCl and TiO2: Direct hole transfer vs. indirect OH oxidation, Appl. Catal., B, 2019, 241, 514–520.
31. F. Li, Z. Wei, K. He, L. Blaney, X. Cheng and T. Xu, et al., A concentrate-and-destroy technique for degradation of perfluorooctanoic acid in water using a new adsorptive photocatalyst, Water Res., 2020, 185, 116219.
32. D. Leonello, M. A. Fendrich, F. Parrino, N. Patel, M. Orlandi and A. Miotello, Light-Induced Advanced Oxidation Processes as PFAS Remediation Methods: A Review, Appl. Sci., 2021, 11(18), 8458.
33. Y. Zhu, T. Xu, D. Zhao, F. Li, W. Liu and B. Wang, et al., Adsorption and solid-phase photocatalytic degradation of perfluorooctane sulfonate in water using gallium-doped carbon-modified titanate nanotubes, Chem. Eng. J., 2021, 421, 129676.
34. Y. Zhu, H. Ji, K. He, L. Blaney, T. Xu and D. Zhao, Photocatalytic degradation of GenX in water using a new adsorptive photocatalyst, Water Res., 2022, 220, 118650.
35. R. Dillert, D. Bahnemann and H. Hidaka, Light-induced degradation of perfluorocarboxylic acids in the presence of titanium dioxide, Chemosphere, 2007, 67(4), 785–792.
36. B. Xu, M. B. Ahmed, J. L. Zhou, A. Altaee, M. Wu and G. Xu, Photocatalytic removal of perfluoroalkyl substances from water and wastewater: Mechanism, kinetics and controlling factors, Chemosphere, 2017, 189, 717–729.
37. T. Xu, H. Ji, Y. Gu, T. Tong, Y. Xia and L. Zhang, et al., Enhanced adsorption and photocatalytic degradation of perfluorooctanoic acid in water using iron (hydr)oxides/carbon sphere composite, Chem. Eng. J., 2020, 388, 124230.
38. L. Duan, B. Wang, K. Heck, S. Guo, C. A. Clark and J. Arredondo, et al., Efficient photocatalytic PFOA degradation over boron nitride, Environ. Sci. Technol. Lett., 2020, 7(8), 613–619.
39. Y.-C. Chen, S.-L. Lo and J. Kuo, Effects of titanate nanotubes synthesized by a microwave hydrothermal method on photocatalytic decomposition of perfluorooctanoic acid, Water Res., 2011, 45(14), 4131–4140.
40. J.-M. Arana Juve, F. Li, Y. Zhu, W. Liu, L. D. M. Ottosen and D. Zhao, et al., Concentrate and degrade PFOA with a photo-regenerable composite of In-doped TNTs@AC, Chemosphere, 2022, 300, 134495.
41. Y. Wang, P. Zhang, G. Pan and H. Chen, Ferric ion mediated photochemical decomposition of perfluorooctanoic acid (PFOA) by 254nm UV light, J. Hazard. Mater., 2008, 160(1), 181–186.
42. L. Jin, P. Zhang, T. Shao and S. Zhao, Ferric ion mediated photodecomposition of aqueous perfluorooctane sulfonate (PFOS) under UV irradiation and its mechanism, J. Hazard. Mater., 2014, 271, 9–15.
43. X. Li, P. Zhang, L. Jin, T. Shao, Z. Li and J. Cao, Efficient photocatalytic decomposition of perfluorooctanoic acid by indium oxide and its mechanism, Environ. Sci. Technol., 2012, 46(10), 5528–5534.
44. M. Ohno, M. Ito, R. Ohkura, A. E. R. Mino, T. Kose and T. Okuda, et al., Photochemical decomposition of perfluorooctanoic acid mediated by iron in strongly acidic conditions, J. Hazard. Mater., 2014, 268, 150–155.
45. J. Eke, L. Banks, M. A. Mottaleb, A. J. Morris, O. V. Tsyusko and I. C. Escobar, Dual-functional phosphorene nanocomposite membranes for the treatment of perfluorinated water: An investigation of perfluorooctanoic acid removal via filtration combined with ultraviolet irradiation or oxygenation, Membranes, 2020, 11(1), 18.
46. F. Xiao, P. C. Sasi, B. Yao, A. Kubátová, S. A. Golovko and M. Y. Golovko, et al., Thermal stability and decomposition of perfluoroalkyl substances on spent granular activated carbon, Environ. Sci. Technol. Lett., 2020, 7(5), 343–350.
47. M. L. Tran, C. C. Fu, L. Y. Chiang, C. T. Hsieh, S. H. Liu and R. S. Juang, Immobilization of TiO₂ and TiO₂-GO hybrids onto the surface of acrylic acid-grafted polymeric membranes for pollutant removal: Analysis of photocatalytic activity, J. Environ. Chem. Eng., 2020, 8(5), 104422.
48. B. Malczewska and A. Żak, Structural changes and operational deterioration of the UF polyethersulfone (PES) membrane due to chemical cleaning, Sci. Rep., 2019, 9(1), 422.
49. A. Razmjou, J. Mansouri, V. Chen, M. Lim and R. Amal, Titania nanocomposite polyethersulfone ultrafiltration membranes fabricated using a low temperature hydrothermal coating process, J. Membr. Sci., 2011, 380(1), 98–113.
50. E. Steinle-Darling, E. Litwiller and M. Reinhard, Effects of sorption on the rejection of trace organic contaminants during nanofiltration, Environ. Sci. Technol., 2010, 44(7), 2592–2598.
51. X. Chen, A. Vanangamudi, J. Wang, J. Jegatheesan, V. Mishra and R. Sharma, et al., Direct contact membrane distillation for effective concentration of perfluoroalkyl substances – Impact of surface fouling and material stability, Water Res., 2020, 182, 116010.
52. J. M. Arana Juve, B. Wang, M. A. Ibrahim, M. S. Wong and Z. Wei, Complete defluorination of per- and polyfluoroalkyl substances — dream or reality?, Curr. Opin. Chem. Eng., 2023, 41, 100943.
53. H. Hori, A. Yamamoto, K. Koike, S. Kutsuna, I. Osaka and R. Arakawa, Photochemical decomposition of environmentally persistent short-chain perfluorocarboxylic acids in water mediated by iron(II)/(III) redox reactions, Chemosphere, 2007, 68(3), 572–578.
54. S. Lotfi, K. Fischer, A. Schulze and A. I. Schäfer, Photocatalytic degradation of steroid hormone micropollutants by TiO2-coated polyethersulfone membranes in a continuous flow-through process, Nat. Nanotechnol., 2022, 17(4), 417–423.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ew00224e

---