Defluorination of Aqueous Perfluorooctanesulfonate by Activated Persulfate Oxidation

Activated persulfate oxidation technologies based on sulfate radicals were first evaluated for defluorination of aqueous perfluorooctanesulfonate (PFOS). The influences of catalytic method, time, pH and K2S2O8 amounts on PFOS defluorination were investigated. The intermediate products during PFOS defluorination were detected by using LC/MS/MS. The results showed that the S2O8 2− had weak effect on the defluorination of PFOS, while the PFOS was oxidatively defluorinated by sulfate radicals in water. The defluorination efficiency of PFOS under various treatment was followed the order: HT (hydrothermal)/K2S2O8 > UV (ultraviolet)/K2S2O8 > Fe2+/K2S2O8 > US (ultrasound)/K2S2O8. Low pH was favorable for the PFOS defluorination with sulfate radicals. Increase in the amount of S2O8 2− had positive effect on PFOS defluorination. However, further increase in amounts of S2O8 2− caused insignificant improvement in PFOS defluorination due to elimination of sulfate radicals under high concentration of S2O8 2−. CF3(CF2)nCOOH (n = 0–6) were detected as intermediates during PFOS defluorination. Sulfate radicals oxidation and hydrolysis were the main mechanisms involved in defluorination process of PFOS.

Persulfate anion (S 2 O 8 22 ) is a strong oxidizing agent with a redox potential of 2.0 V, and can be reduced to sulfate anions as shown below (Eq.1) [21]: S 2 O 8 22 can be activated to produce sulfate radicals (SO 4 N2 ) with higher redox potential of 2.6-3.1 V, which are very reactive with a wide range of contaminants. There are two general ways of activating S 2 O 8 22 : homolysis of the peroxide bond using heat, ultrasound or light (Eq.2) and an oxidation reduction process (analogous to the Fenton reaction) with electron donors, including e 2 (aq) from radiolysis of water or low-valent metals such as Fe 2+ and Ag + (Eq.3).
Previous studies show that SO 4 N2 is more prone to redox reactions than hydroxyl radicals (NOH) under neutral conditions, as summarized in the following reactions (Eq. [4][5][6]: . N2 is a strong oxidizing radicals and has been known to react with electron-rich moieties through several reaction pathways, including electron exchange, hydrogen abstraction, and direct oxygen transfer. Because of comparative stability of persulfate, strong oxidation characteristics, and pH-independent effectiveness, persulfate and sulfate radical oxidation have been utilized for the oxidative degradation of a number of organics [22][23][24][25][26]. It also had been reported that perfluorooctanoic acid (C 7 F 15 COOH, PFOA) and other short-chain perfluorocarboxylic acid (C n F 2n+1 COOH, PFCAs) could be decomposed by sulfate radicals [27][28][29][30]. The results show that PFOA and other PFCAs were effectively decomposed to F 2 and CO 2 by using SO 4 N2 , and almost all of the initial S 2 O 8 22 was transformed to SO 4 22 during the reaction. However, no studies have been reported on persulfate oxidation of PFOS and perfluoroalkyl sulfonate (C n F 2n+1 SO 3 2 ). The present study made first attempt to evaluate the PFOS defluorination effect of S 2 O 8 22 or SO 4 N2 using the activated K 2 S 2 O 8 oxidation systems under air atmosphere. The influence of catalytic method (ultraviolet, ultrasound, hydrothermal, ferrous ion), time, pH and K 2 S 2 O 8 concentration on PFOS defluorination were investigated. The concentrations of intermediate products were detected using LC/MS/MS. Finally, the possible defluorinaiton mechanism of PFOS with sulfate radicals was proposed.

Methods
PFOS stock solution (100 mg/L) was prepared with ultrapure water and then stored in a refrigerator (5uC) before used. The reactions were conducted in batch experiments in polytetrafluoroetylene (PTFE) reactor, due to their superior non-sorption property. Ultraviolet (UV)-activated K 2 S 2 O 8 oxidation system was shown in Fig. 1A: Two 15W low-pressure mercury lamps (Direct-immersion, H model, SUNSHINE, China) emitting light at 254 nm were placed in the reactor. The photoelectric conversion efficiency was 35%,38%. The effective radiating length of the UV tube was 9.5 cm (with a radiating surface area of 113.04 cm 2 ) and UV fluence rate was around 40 mW/cm 2 . During the irradiation, the UV device was switched on for the first 15 min to warm up the UV lamps. The reactor was placed on constant temperature magnetic stirrer at desired temperature during the reaction time. Ultrasound (US)-activated K 2 S 2 O 8 oxidation system was shown in Fig. 1B: The ultrasonic apparatus consisted of an ultrasonic generator and an oscillator operated at an applied (calorimetric) power of 100 (86)W (TOSO model, China, 40 kHz). The average power density delivered to the reactor was 2W/cm 2 . The solution temperature was maintained by a thermostatic container. A sink was placed in the a thermostatic container to keep the temperature of both ultrasound transducer and reactor. During the US process, the thermostatic container device was switched on for the first about 20 to 30 min to keep desired temperature (20uC). According to results of control experiments, the temperature of reaction solution remained about 20uC, was no more than 21uC during the reaction (measured by using mercury's thermometer). Hydrothermal (HT)-activated K 2 S 2 O 8 oxidation system: The reactor was placed on constant temperature magnetic stirrer at desired temperature during the reaction time. And the reactor was closed during the reaction process to prevent loss of H 2 O and gas product by volatilization. Ferrous ion (Fe 2+ )-activated K 2 S 2 O 8 oxidation system: the addition quantity of Fe 2+ catalyst was 3 mM. The reactor was placed on constant temperature magnetic stirrer at desired temperature during the reaction time. In all K 2 S 2 O 8 oxidation systems, an aqueous (Milli-Q) solution (100 mL) of PFOS (100 mg/L; 0.186 mM) were introduced into the reactor. The samples were withdrawn from the reactor at fixed time intervals and quickly quenched in iced water to end reaction [28,29]. Control experiments also had been conducted to make sure the reactions could be quenched in iced water (Fig. S1). The PFOS and F 2 concentrations were determined by LC/MS/MS and IC, respectively. All the results discussed in this article were average values form at least two experiments. The initial pH values were adjusted at about 7.0 by using standard NaOH and HCl solutions except for ''solution pH effect''.
LC/MS/MS. The concentrations of PFOS and the intermediates were measured using liquid chromatography tandem mass spectrometry. Liquid chromatography was performed on an HPLC apparatus equipped with an Agilent model 1200 series (RRLC/6410B Triple Quard MS, USA) and was used for the LC separation of PFOS. The HPLC separation was carried out at 30uC using a gradient composed of solution A (10 mM ammonium acetate solution adjusted to pH 4 with the addition of acetic acid) and solvent B (acetonitrile, GR). The gradient expressed as changes in solvent B was as follows: 0 to 2.0 min, a linear increase from 10% to 20% B; 2.0 to 4.0 min, 20% to 45% B; 4.0 to 5.0 min, 45% to 60% B; 5.0 to 6.0 min, 60% to 95% B; 6.0 to 6.1 min, 95% B to 10% B, hold at 4 min. The flow rate was 0.3 mL/min. Ionization was achieved by electro spray in the negative-ion mode. The electro spray conditions were as follows: nitrogen curtain gas flow: 10 Table 1.
The sample solution and standard solution were injected into the liquid chromatography tandem mass spectrometry, respectively. MRM spectra of standard solutions of nine kinds of PFCs were showed in Fig. 2. The linear regression equations between the concentration of each PFCs and peak area of characteristic peaks were built with the correlation coefficient R 2 more than 0.99. The linear ranges of each PFCs were 0.001-1 mg/L. Ion Chromatography. The concentrations of F 2 and SO 4 22 were determined by an anion-chromatography system (Dionex, ICS-1000, USA) consisting of a degasser, a sampler (1 mL injection volume), a guard column (AS4A-SC 4650 mm, Dionex), a separation column (AS4A-SC 46250 mm, Dionex), a column heater (30uC), and a conductivity detector with a suppressor. A mixture solution containing 3.5 mM Na 2 CO 3 and 1.0 mM NaHCO 3 as the mobile phase was delivered at a flow rate of 1.0 mL?min 21 .The lowest detection limit of F 2 and SO 4 22 was 0.02 mg/L and 0.1 mg/L, respectively.

Results and Discussion
Defluorinationof PFOS by K 2 S 2 O 8 oxidation The defluorination of PFOS (100 mg/L, 0.186 mM) in each system with the initial K 2 S 2 O 8 amount of 5 g/L (18.5 mM) were shown in Fig. 3.
The S 2 O 8 22 had weak effect on the defluorination of PFOS. The defluorination efficiency (moles of F 2 formed)/(moles of fluorine content in initial PFOS) of PFOS were only 1.45% in solo K 2 S 2 O 8 system after 12 h, indicating that S 2 O 8 22 was not an efficient oxidant for degrading PFOS at the room temperature of 20uC. In contrast, the K 2 S 2 O 8 showed different defluorination efficiencies of PFOS in the presence of catalyst, such as UV, US, HT and Fe 2+ . It should be mentioned that these treatments had poor effects on PFOS defluorination when they are used alone. Thus, it can be concluded that the sulfate radical oxidation was mainly responsible for the defluorination of PFOS in activated K 2 S 2 O 8 oxidation. On the basis of data (Fig. 3) of this research, the varying tendencies of F 2 concentration were fitted using firstorder kinetics equation. According to the fitting results, the apparent rate constants of F 2 in UV/K 2 S 2 O 8 , US/K 2 S 2 O 8 , HT/ K 2 S 2 O 8 and Fe 2+ /K 2 S 2 O 8 were 0.016, 0.004, 0.023 and 0.010 h 21 , respectively. Thus, PFOS was defluorinated faster in HT/K 2 S 2 O 8 system than others. The highest PFOS defluorination efficiency reached 22.52% in HT/K 2 S 2 O 8 system after 12 h. Temperature has an important influence on free radical and its inducing oxidization reactions. The PFOS move more quickly at high temperature, which increases their probability to react with SO 4 N2 . Thus, high temperature could not only activate S 2 O 8 22 but also promote the reaction process, and HT/K 2 S 2 O 8 system had showed better defluorination effect than other activated K 2 S 2 O 8 oxidation systems. The formation of SO 4 22 in each system with the initial K 2 S 2 O 8 amount of 5 g/L (18.5 mM) was also shown in Fig. 4.
In activated K 2 S 2 O 8 oxidation, SO 4 N2 radical anions were formed from S 2 O 8 22 , which was then react with PFOS. The SO 4 22 could be formed through two main ways. One way is that one-electron transfer from reducible agent such as PFOS, S 2 O 8 22 and H 2 O to SO 4 N2 (total concentration = 37 mM) and the other way is sulfonic acid group deprivation to form PFOS (total concentration = 0.186 mM). By contrast, the SO 4 22 generation form sulfonic acid group deprivation could be negligible, therefore we could assume that all the SO 4 4 22 . Based on these data (Fig. 2), the concentrations of SO 4 22 were fitted using first-order kinetics equation. According to the fitting results, the apparent rate constants of SO 4 22 21 with an increase in the ultrasound frequency from 50 to 400 kHz [31]. However, in present study, low frequency of US (40 KHz) could not effectively decomposes

Effect of initial solution pH
Effects of initial pH (3,11) on defluorination of PFOS 0.186 mM (100 mg/L) under each activated K 2 S 2 O 8 oxidation treatment was shown in Table 2.
During the experimental runs, pH of the PFOS solution dropped as the reaction progressed due to formation of lager amount of SO 4 22 . Change of pH in K 2 S 2 O 8 solution during reaction process in activated oxidation systems was shown in Fig. 5. With the reaction going on, pH of the solution dropped, and showed different changing tendency with different initial pH in each activated persulfate oxidation system. The pH of US/ K 2 S 2 O 8 system slowly decreased as compared with other three systems. It inferred that US had a weaker activating capability than UV, HT and Fe 2+ . Moreover, there was no obvious change in pH in Fe 2+ /K 2 S 2 O 8 system in 120 min at initial pH of 11.0. Thus, a fixed initial pH was just maintained at the initial stage of PFOS defluorination in UV/K 2 S 2 O 8 and HT/K 2 S 2 O 8 systems, and later reaction was conducted under the acidic condition. Even so, PFOS defluorination was varied with initial pH under each activated K 2 S 2 O 8 oxidation treatment. As shown in Table 2, the PFOS defluorination efficiency decreased with the increase of initial pH under all treatment, especially for Fe 2+ /K 2 S 2 O 8 .
The PFOS defluorination efficiency increased with deceasing of solution pH, because additional sulfate radicals were formed due to acid catalyzation [32]. In addition, sulfate radicals may react with OH 2 to form more hydroxyl radical under alkaline conditions [28], as shown by Eq. (8) below: The generated hydroxyl radicals generally attack organic molecules through the H-atom abstraction to form water. However, PFOS contain no hydrogen to be abstracted, therefore hydroxyl radicals have a very poor reactivity with PFOS in aqueous solution and slow down the PFOS decomposition rate. Thus, alkaline conditions were unfavorable for the defluorination of PFOS by sulfate radicals. The pH of PFOS solution decrease during the defluorination process, but high initial concentration of OH 2 need to consume SO 4 N2 , which could weaken the oxidant effect of activated K 2 S 2 O 8 oxidization. Thus, the PFOS defluorination efficiency decreased with the increase of initial pH under all treatment. In addition, compared with UV/K 2 S 2 O 8 and HT/ K 2 S 2 O 8 system, the more decrease in PFOS defluorination under Fe 2+ /K 2 S 2 O 8 treatment under alkaline conditions could be explained by the formation of Fe(OH) 2 which cause the deactivation of catalyst.

Effect of S 2 O 8 22 amounts
The effect of initial S 2 O 8 22 amount (0-12.5 g/L) on defluorination of PFOS (100 mg/L) under each activated K 2 S 2 O 8 oxidation treatment was shown in Fig. 6.
The defluorination efficiency increased under all treatments when the initial amount of S 2

PFOS defluorination by-products
We measured the concentrations of PFOS and its intermediates under UV/K 2 S 2 O 8 treatment during UV irradiation. Fig. 7 shows the MRM spectra of assigned ionic compounds in the degraded solutions after 4 h by UV/K 2 S 2 O 8 oxidation.
The concentrations of PFOS and the intermediates at different irradiation time were determined using external standard method. The results were shown in Fig. 8.
The concentrations of PFOA, PFHpA and PFPA first increased and then decreased during the reaction process, and the concentrations of PFHxA, PFBA, PFPrA and TFA gradually increased with the increasing of time. As intermediates, PFCAs were not only the hydrolysis product but also the reactant of following reactions. The formed perfluorinated carboxylic acid undergo a further degradation and shortening of the perfluorocarbon chain. Based on the reaction dynamics of continuous reaction, the concentration of PFCAs should first increase and then decreased during the reaction process. From the results, it can infer that PFHxA, PFBA, PFPrA and TFA could not reached the  . Because of the lack of standards of some compounds we could not measure their actual concentration. The unknown fluorines are those not to be confirmed by above analysis. Thus, the irradiation-time dependence of the mass balance of fluorine was also studied in this study. The mass concentration of the inorganic fluorine in the solutions was analyzed using IC at different reaction time. The organic fluorine referred mainly to the fluorine in PFOS and intermediates which had been detected by LC/MS/MS and their mass concentration was calculated from mass ratio between fluorine and parent PFCs (Eq.12).
The result of analysis of mass balance of fluorine was given in Table 3. The sum of the fluorine in PFOS, PFCAs and inorganic fluorine accounted for greater than 95% of the fluorine from the degraded PFOS at any point in time during UV/K 2 S 2 O 8 system. According to the results of analyses of the intermediates and the mass balance of fluorine, it can be inferred that PFCAs were the main intermediates.

PFOS defluorination mechanisms
Previous investigations have provided that PFOS could lose electrons to anode or strong oxidizers, but the mechanism was still in controversy. Kimberly et al. found that an alternative oxidation mechanism is the direct transfer of electrons from PFOS to the BDD (Boron-Doped Diamond) anode as shown in Eq.13 [19].
Lin et al. reported that hydroxyl radicals attack the perfluoroanion and become quenched into hydroxide ions, leaving perfluorinated radicals for continuing chain reactions that lead to further decomposition [20]. Liu et al. found that the first oxidative attack of PFOS by permanganate may have occurred at the C-S and C-C bonds, which released the -CF 2 units and oxidation product of SO 4 22 to form the shorter chain perfluoroalkyl sulfonates and perfluorochemical products subject to further transformation reactions such as hydrolysis [2].
As another most extensive and typical of PFCs, PFOA oxidation mechanism with SO 4 N2 have previously reported. Both Hori et al. and Lee et al. reported that the PFOA oxidation mechanism with SO 4 N2 was described as follows (Eq.14-18) [27,28]: Based on the results discussed above, the mechanism of oxidative defluorination of PFOS with sulfate radicals could be illuminated. The defluorination of PFOS was as Fig. 9: The oxidation potential of SO 4 N2 is 2.5-3.1eV, while the C-F bond in PFOS is the most oxidation resistant bond (E0 = 3.6 eV) and is difficult to be dissociated by SO 4 N2 . Therefore, the first oxidative attack of PFOS by SO 4 N2 may have occurred at the C-S bonds, therefore, SO 4 Noxidize PFOS to form C 8 F 17 N, the unstable C 8 F 17 N may react with H 2 O to form unstable C 8 F 17 OH, which undergo HF elimination to form C 7 F 15 COF. Moreover, C 7 F 15 COF further undergo hydrolysis, resulting in the formation of short chain PFOA(C 7 F 15 COOH). PFOA is first formed by dissociation of two fluorine of PFOS, and long-chain PFCAs are decomposed stepwise to form short-chain perfluorocarboxylic acid such as PFHpA, PFHxA, PFPA and PFBA are formed successively by similar reaction. Moreover, previous investigations have provided that PFOA and other short-chain perfluorocarboxylic acid could be effectively degraded by sulfate radicals [28]. Hori et al. found that the defluorination of PFOA (1.35 mM) reached 73.8% in UV/ K 2 S 2 O 8 system after 12 h of irradiation. And they also reported that the defluorination of PFOA (0.374 mM) was 77.5% in HT/ K 2 S 2 O 8 system at 80uC after 12 h [33]. In addition, Lee et al. found that short-chain PFCAs are easier to degrade and mineralize than long-chain PFCAs [28]. Thus, the SO 4 N2 /PFCAs reaction could compete with the SO 4 N2 /PFOS reaction. Our results show that PFOS was much more stable and more difficult to be defluorinated by sulfate radicals. It can be inferred that SO 4 N2 /PFCAs could be more likely to occurred than the SO 4 N2 / PFOS.

Conclusion
The S 2 O 8 22 showed weak effect on defluorination of PFOS while sulfate radicals could oxidatively decompose PFOS. The defluorination efficiencies were varied with the K 2 S 2 O 8 oxidation activated by different treatment. The defluorination performance of treatments follow the orders: HT/K 2 S 2 O 8 . UV/K 2 S 2 O 8 . Fe 2+ /K 2 S 2 O 8 . US/K 2 S 2 O 8 . Acidic conditions was favorable for the defluorination of PFOS by sulfate radicals. The defluorination efficiency increased with increases of the initial amount of S 2 O 8 22 under all activated K 2 S 2 O 8 oxidation treatment. However, further increase in the initial amounts of S 2 O 8 22 did not resulted in further increase in PFOS defluorination efficiency. The main intermediates during the degradation of PFOS were short-chain PFCs. Oxidization and hydrolysiswere the main defluorination mechanism of PFOS. PFOS was first lost electron to sulfate radicals and then desulfonated to C 8 F 17 radical. During the hydrolytic reaction, long-chain PFCAs are decomposed stepwise to form short-chain PFCAs. Figure S1 The effects of low temperature (0uC) on the defluorination of PFOS (0.186 mM) with S 2 O 8 22 (18.5 mM) and Fe 2+ (3 mM) in ice water were investigated. IC spectra of F 2 , S 2 O 8 22 and SO 4 22 before and after reaction were shown in Fig. S1. The results showed that no F 2 and SO 4 22 was observed after 20 min, and there was less change in concentration of S 2 O 8 22 . Thus, it could be considered that S 2 O 8 22 was stable in iced water, and the ice water could quench the formation SO 4 N2 . Due to the time for sampling was controlled in 10 min, we can considered that this method for quenching reaction could ensure the accuracy of the F 2 detection. (TIF)