Enhanced Degradation of Sulfamethoxazole (SMX) in Toilet Wastewater by Photo-Fenton Reactive Membrane Filtration

Pharmaceutical residuals are increasingly detected in natural waters, which made great threat to the health of the public. This study evaluated the utility of the photo-Fenton ceramic membrane filtration toward the removal and degradation of sulfamethoxazole (SMX) as a model recalcitrant micropollutant. The photo-Fenton catalyst Goethite (α-FeOOH) was coated on planar ceramic membranes as we reported previously. The removal of SMX in both simulated and real toilet wastewater were assessed by filtering the feed solutions with/without H2O2 and UV irradiation. The SMX degradation rate reached 87% and 92% respectively in the presence of UV/H2O2 for the original toilet wastewater (0.8 ± 0.05 ppb) and toilet wastewater with a spiked SMX concentration of 100 ppb. The mineralization and degradation by-products were both assessed under different degradation conditions to achieve deeper insight into the degradation mechanisms during this photo-Fenton reactive membrane filtration. Results showed that a negligible removal rate (e.g., 3%) of SMX was obtained when only filtering the feed solution through uncoated or catalyst-coated membranes. However, the removal rates of SMX were significantly increased to 67% (no H2O2) and 90% (with H2O2) under UV irradiation, respectively, confirming that photo-Fenton reactions played the key role in the degradation/mineralization process. The highest apparent quantum yield (AQY) reached up to approximately 27% when the H2O2 was 10 mmol·L−1 and UV254 intensity was 100 μW·cm−2. This study lays the groundwork for reactive membrane filtration to tackle the issues from micropollution.


Introduction
Pharmaceutical residuals are increasingly detected in natural waters and effluent from wastewater treatment plants (WWTPs) [1]. This raises public health concerns even though most

Batch Degradation Experiments under Different Conditions
To examine the contributions of membrane adsorption, UV photolysis and photo-Fenton reaction toward the removal of SMX, a series of bath experiments were carried out with/without UV 254 irradiation, H 2 O 2 , and the presence of α-FeOOH catalyst on the membrane. Briefly, 30 mL of the SMX solution with the initial concentration of 12 mg·L −1 was prepared. Then, the ceramic membrane (47 mm in diameter and 2.5 mm in thickness) with or without the coating of α-FeOOH was placed on the bottom of the 90-mm petri dish as shown in Figure S1 in the Supporting Information (SI). The distance between the UV lamp and the surface of the liquid was 2.5 cm to obtain approximately the exposure intensity of 400 ± 1 µW·cm −2 . The petri dish was mildly agitated on a rotational shaker to thoroughly mix up the solution. The dose of H 2 O 2 was consistently 10 mM for all the experiments unless indicated. 0.5 mL samples were taken at different times (0, 1, 5, 10, 20, 30, and 60 min) and filtered before the analytical measurement of the SMX concentrations by a high-performance liquid chromatography (HPLC, WATERS e2695, USA) as detailed in the Supporting Information.

Operation of Continuous Filtration Experiments
The removal and degradation of SMX were also assessed in a dead-end mode filtration through the catalyst-coated ceramic membrane. The membrane filtration module was made of polytetrafluoroethylene (PTFE) that is highly resistant to chemical oxidation or UV irradiation. The available membrane surface area was approximately 17.34 cm 2 with an overhead space of 1.9 ml (0.2 cm in depth) and a quartz window allowing the UV light illumination ( Figure S2). A UVL 214-Watt lamp (Analytikjena Company, Beverly, MA, USA) provides a monochromatic UV 254 irradiation of 401 µW·cm −2 on the surface of the α-FeOOH-coated membrane. More detailed illustration of the photocatalytic membrane filtration was reported previously [24,25].

Degradation of SMX Spiked in Toilet Wastewater via Photocatalytic Membrane Filtration
Real toilet water was taken from railway stations in China and filtrated by mixed cellulose ester (MCE) membranes with a nominal pore size 0.45 µm. The background SMX concentration was determined to be 0.8 ± 0.05 ppb. To accurately assess the removal of SMX in the real toilet water, SMX was spiked to reach a final concentration of 100 ppb in the tested water by adding 0.1 mL of the SMX stock solution (1 mg·L −1 ) to 1 L of the real toilet water that was pre-filtrated with 0.45-µm glass fiber membrane filters. The pH of the real toilet water varied slightly (5.7-7.3), which was adjusted to 7 with phosphate buffer or NaOH. Other major real toilet water quality parameters are shown in Table 1.

Analysis of Photocatalytic Degradation Mechanisms
To examine the photocatalytic degradation pathways for SMX, influent and effluent samples from the photo-Fenton reactions were analyzed for degradation byproducts using liquid chromatography−electrospray ionization mass spectrometry equipped with an electrospray ionization source (ESI) or LC−ESI−MS (Agilent1290-6430, USA).

Statistical Analysis
The following experiments were carried out at least with triplicate independent sampling or testing: (1) DI water permeation test; (2) degradation assessment of SMX in batch mode; (3) the concentration measurement of SMX and TOC. SEM images in Figure 1 are typical results selected from at least five sample locations, while the presented results in Figures 2-4 are usually presented with average values with standard deviation as error bars. For the filtration studies, permeate samples were taken at multiple sampling times to obtain representative results, which were shown as average ( Figures 5 and 6). However, three repetitions of filtration tests were conducted to confirm the observations. t-testing was used to examine the significance of data variations we observed in Figure 5 in different filtration conditions using at a significant level of 0.05. Figure 1 shows the SEM images of the coated membranes from the top and cross-sectional views. As the catalyst coating density increased from 0.5 to 6 µg·g −1 , the surface coverage apparently increased, and the resulting pores seemed to decrease as compared between Figure 1a,c. The cross-sectional images show that the coating thickness varied from 5 to 8 µm accordingly. Figure S3 shows the water permeability under various TMPs for ceramic membranes before and after catalyst coating. The permeate flux (L m −2 h −1 , LMH) was calculated by the Darcy's equation under different TMPs as detailed in Section S3. The water permeability for the pristine ceramic membranes with a nominal pore diameter of 0.14 µm are determined to be more than 44.0 LMH·psi −1 . For coated membranes, the water permeability reduced to 10, 13, and 20 LMH·psi −1 under heavy coating, medium coating and low coating. The inherent membrane resistance (R m ) for the pristine ceramic membranes was 0.8 × 10 10 m −1 . With the catalyst coating, the values of R m increased to 3.45 × 10 10 m −1 , 2.74 × 10 10 m −1 , and 1.98 × 10 10 m −1 under heavy coating, medium coating, and low coating conditions. Nanomaterials 2020, 10, x FOR PEER REVIEW 4 of 11 chromatography−electrospray ionization mass spectrometry equipped with an electrospray ionization source (ESI) or LC−ESI−MS (Agilent1290-6430, USA).

Statistical Analysis
The following experiments were carried out at least with triplicate independent sampling or testing: (1) DI water permeation test; (2) degradation assessment of SMX in batch mode; (3) the concentration measurement of SMX and TOC. SEM images in Figure 1 are typical results selected from at least five sample locations, while the presented results in Figures 2-4 are usually presented with average values with standard deviation as error bars. For the filtration studies, permeate samples were taken at multiple sampling times to obtain representative results, which were shown as average (Figures 5 and 6). However, three repetitions of filtration tests were conducted to confirm the observations. t-testing was used to examine the significance of data variations we observed in Figure 5 in different filtration conditions using at a significant level of 0.05. Figure 1 shows the SEM images of the coated membranes from the top and cross-sectional views. As the catalyst coating density increased from 0.5 to 6 µg·g −1 , the surface coverage apparently increased, and the resulting pores seemed to decrease as compared between Figure 1a,c. The crosssectional images show that the coating thickness varied from 5 to 8 µm accordingly. Figure S3 shows the water permeability under various TMPs for ceramic membranes before and after catalyst coating. The permeate flux (L m −2 h −1 , LMH) was calculated by the Darcy's equation under different TMPs as detailed in Section S3. The water permeability for the pristine ceramic membranes with a nominal pore diameter of 0.14 µm are determined to be more than 44.0 LMH·psi −1 . For coated membranes, the water permeability reduced to 10, 13, and 20 LMH·psi −1 under heavy coating, medium coating and low coating. The inherent membrane resistance (Rm) for the pristine ceramic membranes was 0.8 × 10 10 m −1 . With the catalyst coating, the values of Rm increased to 3.45 × 10 10 m −1 , 2.74 × 10 10 m −1 , and 1.98 × 10 10 m −1 under heavy coating, medium coating, and low coating conditions.   Figure 2 compares the degradation rates of SMX on catalyst-coated membranes under different conditions. The removal rate of SMX was negligible when the SMX solution was only exposed to Nanomaterials 2020, 10, 180 5 of 11 catalyst-coated membrane, implying that the surface adsorption of SMX on coated membrane was minor. Similarly, the SMX degradation was also negligible if only H 2 O 2 was present in the solution. When the membrane was present with addition of H 2 O 2 , the SMX removal slightly increased to a stable level of over 5% after 10 min. By contrast, the SMX removal was significantly improved under UV irradiation, which alone led to a progressive SMX degradation as shown by the purple triangle data. With the combination with UV/H 2 O 2 or the coated membrane/UV/H 2 O 2 , the SMX removal efficiencies were substantially increased. UV irradiation alone appeared to cause SMX degradation or photolysis, especially in the presence of the catalyst-coated membranes, on which UV photocatalytic reactions may occur. The results in Figure 2a were fitted using a first-order degradation kinetics [38]. The corresponding rate constants (k) and the squared correlation coefficients (R 2 ) are summarized in Table 2. The highest reaction rate constant was obtained when using the coated membrane under UV/H 2 O 2 , confirming that photo-Fenton reaction on the membrane was the primary factor for the enhanced degradation of SMX [24,25]. Figure 2b shows the TOC changes in the SMX solution under different reaction conditions. No mineralization of SMX was obtained when the solution was exposed to the catalyst-coated membrane or H 2 O 2 only. By contrast, a TOC removal rate of 90% at 60 min when UV/H 2 O 2 were both applied to the catalyst-coated membrane. However, the mineralization of SMX was reduced to 49% and 42% at 20 min if the solution was only exposed to UV/H 2 O 2 or to the catalyst-coated membrane under UV irradiation only. Nanomaterials 2020, 10, x FOR PEER REVIEW 5 of 11 Figure 2 compares the degradation rates of SMX on catalyst-coated membranes under different conditions. The removal rate of SMX was negligible when the SMX solution was only exposed to catalyst-coated membrane, implying that the surface adsorption of SMX on coated membrane was minor. Similarly, the SMX degradation was also negligible if only H2O2 was present in the solution. When the membrane was present with addition of H2O2, the SMX removal slightly increased to a stable level of over 5% after 10 min. By contrast, the SMX removal was significantly improved under UV irradiation, which alone led to a progressive SMX degradation as shown by the purple triangle data. With the combination with UV/H2O2 or the coated membrane/UV/H2O2, the SMX removal efficiencies were substantially increased. UV irradiation alone appeared to cause SMX degradation or photolysis, especially in the presence of the catalyst-coated membranes, on which UV photocatalytic reactions may occur. The results in Figure 2a were fitted using a first-order degradation kinetics [38]. The corresponding rate constants (k) and the squared correlation coefficients (R 2 ) are summarized in Table 2. The highest reaction rate constant was obtained when using the coated membrane under UV/H2O2, confirming that photo-Fenton reaction on the membrane was the primary factor for the enhanced degradation of SMX [24,25]. Figure 2b shows the TOC changes in the SMX solution under different reaction conditions. No mineralization of SMX was obtained when the solution was exposed to the catalyst-coated membrane or H2O2 only. By contrast, a TOC removal rate of 90% at 60 min when UV/H2O2 were both applied to the catalyst-coated membrane. However, the mineralization of SMX was reduced to 49% and 42% at 20 min if the solution was only exposed to UV/H2O2 or to the catalyst-coated membrane under UV irradiation only.    Figure 3 shows that approximately 2% of SMX was removed by the uncoated membrane with less than 1% of TOC reduction, which indicates that the contributions from the size exclusion or membrane surface adsorption for SMX are negligible. Meanwhile, filtration through the catalyst-coated membrane slightly increased the removal rate of SMX to about 11%. By contrast, when the catalyst-coated membrane was only exposed to H 2 O 2 , 20% of SMX and 13% of TOC were was removed, indicating that the degradation of SMX was slightly enhanced but the mineralization was still minor. In the presence of catalyst on the membrane and UV irradiation, both the removal rates of SMX and TOC were significantly increased to 40% and 22% respectively, which agrees with the results from the batch experiments. Furthermore, when applying UV irradiation and H 2 O 2 onto the catalyst-coated membrane, the removal of SMX reached the highest level (almost 58%), confirming that the degradation/mineralization was primarily attributed to the photo-Fenton reactions. Nanomaterials 2020, 10, x FOR PEER REVIEW 6 of 11 Table 2. First-order degradation kinetics rate constants of SMX under different experimental conditions in Figure 2a with the uncoated and coated ceramic membranes.  Figure 3 shows that approximately 2% of SMX was removed by the uncoated membrane with less than 1% of TOC reduction, which indicates that the contributions from the size exclusion or membrane surface adsorption for SMX are negligible. Meanwhile, filtration through the catalystcoated membrane slightly increased the removal rate of SMX to about 11%. By contrast, when the catalyst-coated membrane was only exposed to H2O2, 20% of SMX and 13% of TOC were was removed, indicating that the degradation of SMX was slightly enhanced but the mineralization was still minor. In the presence of catalyst on the membrane and UV irradiation, both the removal rates of SMX and TOC were significantly increased to 40% and 22% respectively, which agrees with the results from the batch experiments. Furthermore, when applying UV irradiation and H2O2 onto the catalyst-coated membrane, the removal of SMX reached the highest level (almost 58%), confirming that the degradation/mineralization was primarily attributed to the photo-Fenton reactions.   Figure 4 compares the degradation rates of SMX in toilet wastewater with/without spiked SMX on catalyst-coated membranes under different filtration conditions. Clearly, the degradation rates of SMX in the real toilet water well align with in the spiked toilet water, though the SMX removal rates were Nanomaterials 2020, 10, 180 7 of 11 higher in the spiked toilet water than in the real toilet water. The removal rate of SMX was negligible when the SMX solution was filtrated only by adding H 2 O 2 to the feed solution. By contrast, the SMX removal was significantly improved under UV irradiation, which alone led to a progressive SMX degradation to 50% and 60% in raw and spiked toilet wastewater respectively. With the combination with UV/H 2 O 2 , the SMX removal efficiencies were substantially increased to 87% and 92% for the raw and spiked toilet wastewater respectively.

Assessment of SMX Removal in Toilet Wastewater
Apparently, the background constituents in the raw toilet wastewater such as dissolved organic matters with a TOC of 1712 ± 18 mg·L −1 , suspended solids (983 ± 9 mg·L −1 ) and ammonia nitrogen (1218 ± 16 mg·L −1 ) did not negatively affect the photodegradation of SMX on the ceramic membrane surface, although these species may potential sequester and consume photogenerated radicals. This is probably because the operating photo-Fenton reaction exhibited a greater electron scavenging or transferring rates than the level imposed from the filtering wastewater. The rate (e − ·s −1 ) of electrons (J P ) transferred from valence band to the conduction band on photocatalyst can be calculated by where η is the apparent quantum yield (e.g., 5-15%), the UV intensity was 400 µW·cm −2 , the effective UV-exposure area was about 12.56 cm 2 and the band gap of FeOOH catalyst was 2.5 eV (1 eV = 1.6 × 10 −19 J). The total electron loading rate (J e ) from the influent is a function of the flow rate (Q) and concentrations of equivalent electrons where i refers to the electron donor species (SMX, TOC, and NH 4 + -N); n is the number of the total electrons from electron donor species (e − mole −1 ); (e.g., for SMX, n = 42 e − ·mole −1 , for TOC, n was taken as 8 e − mole −1 ; and for NH 4 + -N, n = 8 e − mole −1 if completely oxidized to nitrate); C i is the concentration of electron donor species (mg·L −1 or mol·L −1 ); and Q is the feed flow rate (3.5 µL·s −1 ). Applying our filtration and experimental conditions, we determined that J e (5.89 × 10 −6 e − ·s −1 ) was much smaller than J P (1.26 ± 0.63 × 10 15 e − ·s −1 , assuming η = 15%), which explains why the background pollutants did not significantly affect the removal rates of SMX. This result also agrees with the previous studies showing that TOC and ammonia could be efficiently oxidized by photo-Fenton reactions [39,40].
Nanomaterials 2020, 10, x FOR PEER REVIEW 7 of 11 3.3.2. Assessment of SMX Removal in Toilet Wastewater Figure 4 compares the degradation rates of SMX in toilet wastewater with/without spiked SMX on catalyst-coated membranes under different filtration conditions. Clearly, the degradation rates of SMX in the real toilet water well align with in the spiked toilet water, though the SMX removal rates were higher in the spiked toilet water than in the real toilet water. The removal rate of SMX was negligible when the SMX solution was filtrated only by adding H2O2 to the feed solution. By contrast, the SMX removal was significantly improved under UV irradiation, which alone led to a progressive SMX degradation to 50% and 60% in raw and spiked toilet wastewater respectively. With the combination with UV/H2O2, the SMX removal efficiencies were substantially increased to 87% and 92% for the raw and spiked toilet wastewater respectively.
Apparently, the background constituents in the raw toilet wastewater such as dissolved organic matters with a TOC of 1712 ± 18 mg·L −1 , suspended solids (983 ± 9 mg·L −1 ) and ammonia nitrogen (1218 ± 16 mg·L −1 ) did not negatively affect the photodegradation of SMX on the ceramic membrane surface, although these species may potential sequester and consume photogenerated radicals. This is probably because the operating photo-Fenton reaction exhibited a greater electron scavenging or transferring rates than the level imposed from the filtering wastewater. The rate (e − ·s −1 ) of electrons (JP) transferred from valence band to the conduction band on photocatalyst can be calculated by JP = η × UV intensity × surface area/band gap (1) where η is the apparent quantum yield (e.g., 5-15%), the UV intensity was 400 µW·cm −2 , the effective UV-exposure area was about 12.56 cm 2 and the band gap of FeOOH catalyst was 2.5 eV (1 eV = 1.6 × 10 −19 J). The total electron loading rate (Je) from the influent is a function of the flow rate (Q) and concentrations of equivalent electrons where i refers to the electron donor species (SMX, TOC, and NH4 + -N); n is the number of the total electrons from electron donor species (e − mole −1 ); (e.g., for SMX, n = 42 e − ·mole −1 , for TOC, n was taken as 8 e − mole −1 ; and for NH4 + -N, n = 8 e − mole −1 if completely oxidized to nitrate); Ci is the concentration of electron donor species (mg·L −1 or mol·L −1 ); and Q is the feed flow rate (3.5 µL·s −1 ). Applying our filtration and experimental conditions, we determined that Je (5.89 × 10 −6 e − ·s −1 ) was much smaller than JP (1.26 ± 0.63 × 10 15 e − ·s −1 , assuming η = 15%), which explains why the background pollutants did not significantly affect the removal rates of SMX. This result also agrees with the previous studies showing that TOC and ammonia could be efficiently oxidized by photo-Fenton reactions [39,40].

Analysis of Photocatalytic Degradation Mechanisms
The surface sites (≡FeIII(OH)) on α-FeOOH catalyst are considered to catalyze the generation of hydroxyl radicals and peroxide anions via photo-Fenton reactions [24]. As reported previously [41], ·OH stoichiometricaly reacts with p-Chlorobenzoic acid (pCBA) in a mole ratio of 1:1, and degradation of ·H through side reactions with other potential contaminants in our reaction system can be ignored. Our results show that UV or H 2 O 2 alone did not decrease the pCBA concentration significantly, which similarly occurred to coated membrane with or without the addition of H 2 O 2 . When exposed to UV alone or the catalyst-coated membrane, the pCBA concentration started to decline much faster. Furthermore, when UV/H 2 O 2 was applied with or with the catalyst-coated membrane, the pCBA concentration declined sharply, indicative of the generation of OH via photocatalysis or photo-Fenton reactions. The detailed measurement of pCBA has been reported elsewhere [24].
To gain a better understanding of mineralization mechanisms, the oxidation byproducts of SMX were identified by LC−ESI−MS. Figure 5 shows the chromatograms of the treated permeate samples at different reaction times (0, 5, 30, and 60 min). Four main oxidation intermediates and SMX are identified based on their characteristic peaks at different m/z values. Figure 6 illustrates the hypothetical degradation or transformation pathways (A, B, and C) of SMX in the photo-Fenton oxidation reaction. In pathway A, the SMX forms monohydroxylated sulfamethoxazole by direct attack of HO· on the aromatic moiety of SMX and/or hydrolysis of unstable radical cation SMX· + formed by interaction with SO 4 − [42]. Subsequently, the sulfonamide bond is cleaved by reactive oxygen species to produce monohydroxylated sulfanilic acid and 3-amino-5-methyl-isoxazole [43,44]. In pathway B, cleavage of the S-N bond, which leads to the hydrolyzation of SMX into C 6 H 8 NO 2 S and C 5 H 7 N 2 O, respectively. In pathway C, the H were replaced by the O forming the 5-(4-methoxy phenyl)-1 and 4-oxadiazoles-2-mercaptan [45].

Analysis of Photocatalytic Degradation Mechanisms
The surface sites (≡FeIII(OH)) on α-FeOOH catalyst are considered to catalyze the generation of hydroxyl radicals and peroxide anions via photo-Fenton reactions [24]. As reported previously [41], ·OH stoichiometricaly reacts with p-Chlorobenzoic acid (pCBA) in a mole ratio of 1:1, and degradation of ·H through side reactions with other potential contaminants in our reaction system can be ignored. Our results show that UV or H2O2 alone did not decrease the pCBA concentration significantly, which similarly occurred to coated membrane with or without the addition of H2O2. When exposed to UV alone or the catalyst-coated membrane, the pCBA concentration started to decline much faster. Furthermore, when UV/H2O2 was applied with or with the catalyst-coated membrane, the pCBA concentration declined sharply, indicative of the generation of OH via photocatalysis or photo-Fenton reactions. The detailed measurement of pCBA has been reported elsewhere [24].
To gain a better understanding of mineralization mechanisms, the oxidation byproducts of SMX were identified by LC−ESI−MS. Figure 5 shows the chromatograms of the treated permeate samples at different reaction times (0, 5, 30, and 60 min). Four main oxidation intermediates and SMX are identified based on their characteristic peaks at different m/z values. Figure 6 illustrates the hypothetical degradation or transformation pathways (A, B, and C) of SMX in the photo-Fenton oxidation reaction. In pathway A, the SMX forms monohydroxylated sulfamethoxazole by direct attack of HO· on the aromatic moiety of SMX and/or hydrolysis of unstable radical cation SMX· + formed by interaction with SO4 − [42]. Subsequently, the sulfonamide bond is cleaved by reactive oxygen species to produce monohydroxylated sulfanilic acid and 3-amino-5-methyl-isoxazole [43,44]. In pathway B, cleavage of the S-N bond, which leads to the hydrolyzation of SMX into C6H8NO2S and C5H7N2O, respectively. In pathway C, the H were replaced by the O forming the 5-(4methoxy phenyl)-1 and 4-oxadiazoles-2-mercaptan [45].

Conclusions
This study evaluated the photo-Fenton ceramic membrane filtration toward the removal and degradation of SMX as a model recalcitrant micropollutant. The removal of SMX in feed water as well as the raw toilet wastewater in the presence of H2O2 and UV irradiation was over 80%. The

Conclusions
This study evaluated the photo-Fenton ceramic membrane filtration toward the removal and degradation of SMX as a model recalcitrant micropollutant. The removal of SMX in feed water as well as the raw toilet wastewater in the presence of H 2 O 2 and UV irradiation was over 80%. The background constituents such as TOC or ammonia in raw toilet wastewater did not appear to affect the photodegradation of SMX. The TOC removal and degradation by-products analysis revealed three possible SMX degradation pathways with four main oxidation intermediates. The research findings laid groundwork toward the application of photo-Fenton reactive membranes for such as wastewater treatment and water reuse.
Supplementary Materials: The following are available online at http://www.mdpi.com/2079-4991/10/1/180/s1, Figure S1: Schematics of the batch experiment of Photo-Fenton degradation. Figure S2: Schematic of the filtration unit with the Photo-Fenton ceramic membrane. Figure S3: The pure water fluxes of the pristine and the catalyst-coated ceramic membranes. Figure S4: (a) The degradation kinetics of SMX and (b) The apparent quantum yield (QY) under different UV 254 irradiation intensities. Initial SMX concentration: 20 mg·L −1 ; H 2 O 2 concentration was 10 mmol·L −1 ; and the catalyst on the ceramic membrane was 2 µg·g −1 .