Degradation of perfluoroalkyl substances using UV/Fe0 system with and without the presence of oxygen

ABSTRACT The wide presence of per- and poly-fluoroalkyl substances (PFAS) in the environment is a global concern, thus their degradation is an imminent task. In this study, oxidative and/or reductive degradation of three representative PFAS – perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), and perfluorooctane sulfonate (PFOS) was achieved using nanoscale zero-valent iron (Fe0 NPs) under ultraviolet (UV) light, both with and without the presence of oxygen. Higher degradation and defluorination rates were obtained for a longer chain PFNA compared to PFOA, and a higher removal of PFAS was achieved without the presence of O2 compared to that with O2. The degradation followed first-order reaction kinetics, and obtained the highest rates of 97.6, >99.9, and 98.5% without the presence of O2 for PFOA, PFNA, and PFOS, respectively. The degradation rates increased with an increase in the nanoparticle concentrations in the range of 1–100 mg/L. In addition to fluoride ions, shorter chain perfluorocarboxylic acids (PFCAs) were detected as the main intermediates during PFAS degradation; PFHpS and 6:2 FTS were also detected during PFOS degradation. Hydroxyl radicals (·OH) and superoxide radicals (·O2 −) were not involved in the degradation of PFOA, but likely involved in the degradation of PFOS. Emerging contaminants PFAS degradation using the UV/Fe0 system is a cost-effective technology owing to the low cost and recyclability of Fe0 nanomaterials, low energy consumption in the system, and its capability to degrade PFAS both with and without the presence of oxygen. This technology can be potentially applied to treat PFAS-contaminated waters in the environment. GRAPHICAL ABSTRACT


Introduction
Per-and poly-fluoroalkyl substances (PFAS) are a diverse group of water-soluble synthetic compounds that have alkyl chains but with part or all of the hydrogen atoms replaced by fluorine atoms, with at least one perfluoroalkyl moiety, -C n F 2n - [1][2][3]. PFAS are ubiquitous in the environment due to their large-scale manufacture, wide application, and unregulated disposal [1,[4][5][6][7]. Destruction of this group of emerging contaminants has become a hotspot in the environmental engineering field because of their possible adverse effects on living organisms and the environment [8,9]. Due to potential risks to human health, perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid (PFOS)-the two most prevalent, widely used, and most studied compounds-have been included in the U.S. EPA's third Contaminant Candidate List (CCL3) and CCL4 [10]. In May 2016, the U.S. EPA issued a lifetime drinking water health advisory of 0.070 µg/L for the sum of PFOA and PFOS, which applies to both short-term (i.e. weeks) and lifetime exposure scenarios. A trace amount of perfluorononanoic acid (PFNA) has also been detected in the environment and human blood that poses potential toxic impacts on animals and humans [11,12]. For example, up to 1.1 μg/L of PFOA, 0.6 μg/L of PFOS, and 0.4 μg/L of PFNA have been detected in wastewater effluents from wastewater treatment plants (WWTPs) that employed secondary or tertiary treatment [12,13]. Recently in the U.S., many states have set more stringent regulation limits for PFOA, PFOS, and PFNA, individually (Table 1). These stricter regulations have made the task of developing effective PFAS elimination technologies more imperative.
The PFAS treatment technologies can be generally classified into two categories: non-destructive and destructive. The non-destructive methods, such as adsorption by activated carbon and membrane filtration (reverse osmosis and nanofiltration), are effective methods for removing PFAS from water. However, the drawback of these methods is the requirement for subsequent destruction and mineralization (e.g. by incineration) of PFAS to avoid other concerns [14,15]. In comparison, destructive techniques are better options for PFAS removal, which can break down PFAS molecules to final mineralized products of CO 2 and HF. Several destructive techniques such as sonolysis [16,17], ultraviolet (UV) photolysis [18,19], and photocatalysis [19][20][21][22][23][24] have been developed for PFAS degradation. Nanoparticles (NPs) such as TiO 2 [21], In 2 O 3 [22], and β-Ga 2 O 3 [24] were used as heterogeneous photocatalysts for PFAS degradation [25]; ferric ion Fe 3 + was proven to accelerate PFAS degradation under UVC or natural light [23,25]; moreover, UV-Fenton process-using UV/Fe 2+ /H 2 O 2 system-was reported to degrade PFOA efficiently, and its defluorination process was mainly due to the interactions between PFOA and Fe 3+ ions [26].
Zero-valent iron NPs (Fe 0 NPs, nZVI) can be used as an alternative source to introduce Fe 3+ ions under UV light by enabling Fenton reaction under light but without introducing anions of, e.g. SO 4 2− , NO 3 − , Cl − , and generating iron sludges compared to photo-Fenton processes [27]. Our previous study showed that Fe 0 NPs can degrade PFOA under UVC (centered at 254 nm) light with the presence of oxygen; short-chain perfluorinated carboxylic acids (PFCAs) bearing C 4 -C 7 were detected as the intermediates, generation of fluoride ions F − was also quantified [28]. Under UVC irradiation, the degradation of PFOA (C 7 F 15 COOH) was possibly initiated by the formation of the [C 7 F 15 COO-Fe] 2+ complex from the dissociated C 7 F 15 COO − (PFO − ) with the released Fe 3+ , then following the steps of decarboxylation, hydroxylation, elimination, and hydrolysis to shorten the carbon chain [28]. However, with the presence of oxygen, the degradation rates of this novel UVC/Fe 0 NPs system for PFOA were relatively slow, possibly due to the scavenging of hydrated electrons by oxygen [25]; plus, the degradation efficiencies of this system on other types of PFAS (e.g. PFOS) and longer chain PFAS (e.g. PFNA) are still unknown. Fe 0 NPs have recently been evaluated for remediating organic contaminants of emerging concern in polluted water because of highly efficient catalytic processes they can lead to and their low cost over conventional photocatalysis [27,28].
Owing to the superior reductive reactivity of Fe 0 NPs to a variety of chlorinated organic contaminants [29], Fe 0 NPs have been used for PFAS removal under darkness. Palladium (Pd 0 )/Fe 0 NPs removed PFAS successfully but without obtaining any degradation products, when Mg-aminoclay coated nZVI was used; fluoride ions F − were detected and possible reductive degradation of PFAS in water was proposed [30][31][32]. However, the reductive degradation mechanism of PFAS in water involving Fe 0 atoms was still not clear. This study was objected to investigate the degradation of a variety of PFAS, their degradation mechanisms, and the optimal degradation conditions using Fe 0 NPs under light. In specific, this study aimed to investigate: (1) the effects of ionic headgroups (i.e. -COOH vs. -SO 3 H) and chain lengths (i.e. C8 vs. C9) on PFAS degradation; (2) the degradation mechanisms from the quenching experiment of reactive oxygen species (ROS), and the determination of concentrations of ROS and intermediates formed; (3) the optimal experimental conditions by testing with and without the presence of oxygen, different concentrations of Fe 0 NPs, and different light conditions (i.e. UVC, UVA, and visible light) in comparison to dark conditions. It was hypothesized that higher PFAS degradation rates can be obtained without the presence of oxygen in the UVC/Fe 0 NPs system.

PFAS removal
A stock solution of PFOA, PFNA, or PFOS (200 mg/L each) was prepared and spiked to a quartz vial to reach the final concentration of 1 mg/L for the investigation of the degradation mechanism-by enabling easy determination of the concentrations of intermediates formed. The PFAS concentrations studied herein were at or below the levels tested in most of the previous studies. For example, the concentrations of PFOA in its degradation study in the literature were in the range of 1-100 mg/L [21,24,26,28,31,[33][34][35][36][37][38][39][40][41][42][43][44][45][46][47], except three studies with the concentration of 0.2 mg/L [30,48], one study in the range of 0.2-5 mg/L [49], and another in the range of 0.05-1 mg/L [19]. The concentrations of PFOS studied previously were in the range of 1-186 mg/L [32,33,39,[50][51][52][53][54][55], except one study with the concentrations of 0.2 and 40 mg/L [30]. For PFNA, the concentrations studied previously were 0.2 [30], 5 [56], and 116 mg/L [57]. Fe 0 NPs (35-45 nm, 1% (w/w) polyvinylpyrrolidone (PVP)-coated, U.S. Research Nanomaterials, Inc.) were sonicated at 20 kHz for 30 min (Sonics VC505, 500 Watts Ultrasonic Processor, Sonics & Materials, Inc., Newtown, CT, U.S.A.) for proper dispersion of the particles, then added to each reaction vial to reach final concentrations of 1, 10, and 100 mg/L, respectively. The concentration of nZVI used in our study was not high as compared to those reported in the literature. For example, the concentrations of modified nZVI for PFAS removal were 1 [30], 5 [31], and 20 g/L [32] in the literature, which were at least ten times higher than that used in our study. Moreover, the concentrations of other heterogeneous photocatalysts for PFAS degradation, e.g. TiO 2 , Ga 2 O 3 , and In 2 O 3 , were at or above 0.25 g/L [24,[34][35][36]40,[45][46][47], which were higher than that used in this study. A photoreactor (Luzchem, LZC-4X) equipped with 14 light bulbs (8 W) was used to provide either UVC (centered at 254 nm), UVA (centered at 350 nm), or visible light that has a light intensity of 4.2, 3.1, and 5.4 mW/cm 2 , respectively. A tube rotator was used in the photoreactor to properly mix the nanoparticles with the solution. The experiment was also performed in the dark (by wrapping the vials with aluminum foil) for comparison. PFAS degradation under UVC light without the addition of Fe 0 NPs was used as the control. All experiments were performed at room temperature with duplicates.

PFAS sample preparation and analysis
At a certain time during a 72-h reaction period, a 100 µL sample was taken from each vial, the sample was placed in a 2 mL microcentrifuge tube, and centrifuged at 14,000 rpm (Eppendorf MiniSpin plus Microcentrifuge) for 20 min to remove the Fe 0 NPs in the suspension. Afterward, an aliquot of 20 µL of the supernatant from each microcentrifuge tube was transferred to 150 µL of methanol in a pre-washed VWR ® centrifugal filter with a 0.2 µm nylon membrane. M8-PFOA or MPFOS (Wellington Laboratories Inc., Canada) of 25 µL and 1 mg/L was spiked in the VWR filter as an internal standard. Then the filter tube was centrifuged at 3000 rpm for 5 min. The filtered solution was transferred to a liquid chromatography (LC) vial with an insert (Target glass Micro-Serts ® , Thermo Scientific) by a glass Pasteur pipette; the LC vial was capped with a pre-slit cap. Duplicate experiments were performed for all the tests.
An Agilent 1100 HPLC (Agilent Technologies, Santa Clara, CA) coupled to a Qtrap triple quadrupole/linear ion trap mass spectrometer (AB Sciex; Toronto, Canada) with an electrospray ionization (EI) detector was used to determine the concentrations of PFOA, PFNA, PFOS, and their degradation products [28,43]. An Agilent column (ZORBAX Extend C18, 3.5 µm, 80 Å, 2.1×100 mm) was used for the LC/MS/MS analysis. The injection volume was 5 µL. The mobile phase was a mixture of 2 mM NH 4 Ac in Optima™ LC/MS grade water and 2 mM NH 4 Ac in Optima™ LC/MS grade methanol with a flow rate of 0.2 mL/min.

Analysis of fluoride ion (F − ), ferrous ion (Fe 2 + ) and total Fe concentrations
The concentration of fluoride ion (F − ) was analyzed by a UV-vis spectrophotometer (Thermo Scientific, Biomate 3S) based on measuring the fluoride/lanthanum(III)/alizarin fluorine blue ternary complex at a wavelength of 620 nm, following the People's Republic of China's Environmental Protection Standards HJ 488-2009 [58,59]. This method has the advantages of high accuracy and repeatability for samples with low fluoride ion concentrations; the limit of detection is 0.02 mg/L, and the lower limit of quantification is 0.08 mg/L. After separating Fe 0 NPs from the reaction vials by precipitation and using an external magnet, Fe 2+ and total Fe (i.e. Fe 2+ + Fe 3+ species) concentrations in the supernatant were analyzed by the Ferrozine method using a UV-Vis spectrophotometer [60]. The detailed steps of F − , Fe 2+ , and total Fe concentration analyses were described in our pervious study [28].

Determination of degradation and defluorination rates
The degradation rate was determined by the change in PFAS concentration over time. Defluorination rate R was calculated by Equation (1).
C 0 represents the initial concentration of PFAS in solution, mg/L; C F − represents the concentration of F − in solution, mg/L; the factor N F corresponds to the number of fluorine atoms contained in one PFAS molecule, factor 19 is the atomic weight of fluorine, and factor M w is the molecular weight of the PFAS.

Determination of contribution of reactive oxygen species (ROS)
Tertiary butyl alcohol (TBA) and benzoquinone (BQ) were used to scavenge hydroxyl radicals (·OH) and superoxide radicals (·O 2 − ) generated in PFAS degradation, respectively. TBA and BQ, each of 1.0 mM, were added to the reaction suspension; after a certain reaction time, samples were taken to measure the concentrations of PFAS. Terephthalic acid (TPA, 0.5 mM in 2 mM NaOH solution) was employed as the probe molecule to determine the amount of ·OH generated in the degradation process. One milliliter TPA stock solution was added to the reactor, and at different time points, samples were taken and filtered by a 0.22 μm syringe filter for the measurement. With the help of an oxidant, TPA (non-fluorescent) can react with ·OH to provide a single product 2-hydroxy terephthalic acid (hTPA)(fluorescent). When O 2 is the oxidant, the maximum yield of hTPA is 35%. Therefore, the concentration of ·OH can be determined via measuring the concentration of hTPA. The method for hTPA analysis by HPLC was modified from previous studies [61][62][63][64]. A Shimadzu Prominence UPLC system (Shimadzu Co. Ltd., Japan) used was equipped with a SIL-20A autosampler (20 µL), a reverse-phase column (Brava C18-ODS, 5 μm, 4.6 mm i.d. ×250 mm), and a fluorescence detector RF-10A XL (Shimadzu, Japan). The excitation and emission wavelengths were 315 and 426 nm, respectively. The UPLC was eluted with 20 mmol/L phosphate buffer solution (pH 7.4). The column was washed with HPLCgrade water for 1 h before use and 2 h after use with a flow rate of 0.6 mL/min and a temperature at 30°C. The retention time of hTPA was 4.530 min in this study. Nirtoblue tetrazolium (NBT, 1.25 × 10 −6 M, exhibiting an absorption maximum at 259 nm) was used as a spectroscopic probe to determine the amount of superoxide radicals ·O 2 − generated in the reaction [65,66]. For the analyses of the concentration of ·O 2 − , 1 mL NBT stock solution was introduced into the reactor to obtain the target concentration. Samples for the measurement were collected at different time intervals and filtrated by a 0.22 μm syringe filter to exclude the influence of the nanoparticles. Quantitative analysis was conducted using a UV-vis spectrophotometer (Thermo Scientific, Biomate 3S).

Effect of light source on PFAS degradation
After 72 h of irradiation, PFAS were best removed under UVC irradiation compared to under other light conditions (i.e. UVA and visible light), as well as in the dark, either with or without the presence of O 2 ( Table 2). The highest removal rates achieved under UVC light were 97.6, >99.9, and 98.5% without the presence of O 2 for of PFOA, PFNA, and PFOS, respectively. PFOA has a strong spectral absorption from the deep UV-region to 220 nm and a weak, broad absorption from 220 to 270 nm [38] (Figure 1). Our study showed PFNA had a similar absorption range; however, no apparent absorption of PFOS was observed in this range ( Figure 1). The different absorption phenomena between PFOS and PFOA/PFNA may be due to their structural difference in headssulfonate and carboxylate head groups, which can also lead to different degradation pathways in the UV/Fe 0 NPs system. Under UVC irradiation alone, without the presence of the    (Table 3), total balances in most scenarios were ∼100%. Thus degradation was the dominant process in PFAS removal in the UVC/Fe 0 system; the only exception was PFOA removal without O 2 where the total mass balance was 68.0 ± 4.9% (Table 3), the fluoride loss may be due to adsorption on iron particles or the uncounted ultra-short chain PFCAs that could be lost due to evaporation from the system.

Effect of oxygen on PFAS degradation
Oxygen plays different roles in photocatalytic processes that can be performed in the presence and absence of O 2 [25,67]. In both approaches, ROS play a crucial role. Typically, oxidation reactions are conducted under ambient atmosphere, but 50% of the photogenerated charge carriers can be lost by reacting with oxygen.  Oxidation reactions can also be carried out under oxygen-free conditions where reduction reactions also take place, initiated by utilizing both the holes and electrons generated. In the case of with O 2 , the initial concentration of dissolved oxygen was 8.02 mg/L in this study. Interestingly, PFAS was removed faster without the presence of O 2 compared to with the presence of O 2 in the UV/Fe 0 NPs system ( Figure 2); for PFNA this difference was insignificant because of its fast removal under both conditions. The removal percentages were >99.9, 87.0 ± 7.0, and 78.  (Figure 2). The removal rate constant of PFOS without O 2 was much higher than that with O 2 , which may be attributed to the quenching effect of oxygen to the generated hydrated electrons (e aq − ) in PFOS degradation [25,52].
Presence of iron may further accelerate the generation of hydrated electrons under light by consuming hydroxyl radicals ·OH that were produced from water splitting under UV irradiation [52,68]. In addition, with the possibility of hydrated electrons generated from H 2 O to get involved in desulfonation, cleaving C-F bonds for defluorination, and C-C bond scission [25,33], the result that the presence of oxygen led to larger suppression of PFOS degradation as compared to PFOA degradation may be due to the restriction of all the three degradation processes involving hydrated electrons in PFOS degradation compared to the restriction of only two processes (i.e. without desulfonation) in PFOA degradation.

Effect of concentrations of Fe 0 NPs on PFAS degradation
In this study, Fe 3+ provided by Fe 0 NPs was likely involved in the photocatalytic process for each PFAS degradation. The concentration and the generation rate of Fe 3+ were controlled by the concentration of Fe 0 NPs; therefore, it is important to study the impact of the concentration of Fe 0 NPs on PFAS removal. The results showed that PFAS removal generally increased with reaction time up to 72 h, and generally improved (especially without O 2 ) with an increase in the concentration of Fe 0 NPs in the range of 1-100 mg/L under UVC ( Figure 3). From the data (Figure 3), it can be seen that 100 mg/L of Fe 0 NPs greatly shortened the time for PFAS removal, for example, 96% of PFNA and 91% of PFOS were removed after 12 h. It is likely that an increased concentration of Fe 0 NPs enhanced the concentration of Fe 3+ in the solution that induced faster degradation of PFAS. In particular, without the presence of O 2 , the removal rates of all three PFAS using 100 mg/L of Fe 0 NPs under UVC light reached to >98% in 72 h, which were much better than those cases without Fe 0 NPs ( Figure 4). However, with the presence of O 2 , the removal rates of PFOA reached to almost the same after 72 h for the cases with and without Fe 0 NPs under UVC light, although faster removal was achieved initially with Fe 0 NPs [28], which indicated the passivation of Fe 0 NPs with the presence of O 2 with time − formation of iron oxide likely reduced the generation of hydrated electrons and Fe 3+ ions involved in PFAS degradation.

Degradation products
PFOA and PFNA degradation generated short-chain PFCAs as intermediates (e.g. PFHpA, PFHxA, PFPeA, and PFBA) and F − ions ( Figure 5). With the presence of O 2 , the concentrations of the generated PFCAs initially     Information Table S1). Both Fe 2+ and Fe 3+ ions were detected in the finished water, either with or without the presence of O 2 in the UV/Fe 0 NPs system. For example, starting from 100 mg/L of Fe 0 NPs, the concentrations of Fe 2+ and Fe 3+ after PFNA degradation for 72 h were 2.3 ± 0.4 and 0.9 ± 0.6 mg/L, respectively, with O 2 ; and lower of 1.1 ± 0.1 and 0.4 ± 0.3 mg/L, respectively, without O 2 . Our previous study showed with the presence of O 2 , Fe 3+ was rapidly consumed after its generation as observed by UV-vis spectrophotometer, which indicated the possible formation of a [C 7 F 15 -COO-Fe] 2+ complex that initiated PFCA degradation through decarboxylation and defluorination pathways [28]. A previous study also showed the generation of photosensitive Fe 3+ -PFOS complex under UV irradiation that can be photolyzed to Fe 2+ and an unstable PFOS radical by the ligand-to-metal charge transfer for further PFOS degradation ( Figure 6) [25]. Without O 2 , nZVI can react with water to form Fe 2+ that may further react with hydroxyl radicals released from water splitting under UV irradiation to form Fe 3+ needed for PFAS degradation [71]. Diagram in Figure 6 shows some possible initial steps of the degradation processes that resulted in the efficient breakdown of PFOS and PFOA in the UVC/Fe 0 NPs system.

ROS in PFAS degradation
From ROS quenching experiments' results, when BQ was applied, the degradation of PFOA was comparable with that in the control (Figure 7(a)), which meant that   [26,72]. From the test, superoxide radical was generated in the UVC/Fe 0 NPs system, and its generation rate was ∼0.76 μmol/L/h during the experiment with the presence of O 2 (Figure 8(a)). With the introduction of TBA that was used to quench the generated hydroxyl radicals ·OH, the degradation rate of PFOA was also almost the same as that of the control ( Figure  7), which indicated that ·OH was also not involved in PFOA degradation in the UVC/Fe 0 NPs system with O 2 . This finding was consistent with the conclusion in previous publications that the contribution of ·OH to PFOA degradation was minimal [38,73]. The generation rate of ·OH in the system was tested to be very low and decreased with time from 13.85 to 0.02 nmol/L/h in 24 h (Figure 8(b)), which might be due to the passivation of zero-valent iron with time. For PFOS, the degradation rate with the addition of TBA or BQ was lower than that of the control (Figure 7), which indicated that different from PFOA, ·O 2 − and ·OH may be involved in PFOS degradation, e.g. in desulfonation to form PFOA. In particular, ·OH likely reacted with C 8 F 17 · that could be  formed in the desulfonation process to generate C 8 F 17 OH [70], which further can be converted to C 7 F 15 COF, then C 7 F 15 COO − for the following degradation process. ination, and C-C bond scission processes were likely quenched by the presence of oxygen to reduce the degradation and defluorination rates, whereas the presence of Fe 0 NPs may promote the generation of hydrated electrons. In addition to F − ions, shorter chain PFCAs were detected as the main intermediates for PFOA and PFNA degradation; for PFOS, trace amounts of shorter chain PFCAs, PFHpS, 6:2 FTS, and fluoride ions were detected. Also, the generated PFCAs underwent degradation with time in the system. Moreover, Fe 2+ and Fe 3+ ions were detected during the PFAS degradation in the UVC/Fe 0 NPs system both with and without O 2 that can be involved in PFAS degradation through the formation of Fe 3+ -PFAS complex, and it was found that an increase in the Fe 0 NP concentration in the range of 1-100 mg/L increased the PFAS degradation rates. In sum, the UV/Fe 0 NPs system can be used not only for PFOA, but also for the PFOS decomposition; higher degradation and defluorination rates were obtained for a longer chain PFNA compared to PFOA. Given the low cost and toxicity of Fe 0 NPs, easy recycling of the used iron particles from the treated water by exploiting their magnetic property, low energy consumption for UVC in the system, and the capability to degrade PFAS both with and without presence of oxygen, the technology of PFAS degradation using the UV/Fe 0 NPs system is very attractive in dealing with the overwhelming PFAS environmental problems, which can be potentially applied to treat various types of PFAS contaminated waters in the environment.