In-Situ H2O2 Cleaning for Fouling Control of Manganese-Doped Ceramic Membrane through Confined Catalytic Oxidation Inside Membrane

This work presents an effective approach for manganese-doped Al2O3 ceramic membrane (Mn-doped membrane) fouling control by in-situ confined H2O2 cleaning in wastewater treatment. An Mn-doped membrane with 0.7 atomic percent Mn doping in the membrane layer was used in a membrane bioreactor with the aim to improve the catalytic activity toward oxidation of foulants by H2O2. Backwashing with 1 mM H2O2 solution at a flux of 120 L/m2/h (LMH) for 1 min was determined to be the optimal mode for in-situ H2O2 cleaning, with confined H2O2 decomposition inside the membrane. The Mn-doped membrane with in-situ H2O2 cleaning demonstrated much better fouling mitigation efficiency than a pristine Al2O3 ceramic membrane (pristine membrane). With in-situ H2O2 cleaning, the transmembrane pressure increase (ΔTMP) of the Mn-doped membrane was 22.2 kPa after 24-h filtration, which was 40.5% lower than that of the pristine membrane (37.3 kPa). The enhanced fouling mitigation was attributed to Mn doping, in the Mn-doped membrane layer, that improved the membrane surface properties and confined the catalytic oxidation of foulants by H2O2 inside the membrane. Mn3+/Mn4+ redox couples in the Mn-doped membrane catalyzed H2O2 decomposition continuously to generate reactive oxygen species (ROS) (i.e., HO• and O21), which were likely to be confined in membrane pores and efficiently degraded organic foulants.


Introduction
Ceramic membranes have been widely used in the membrane bioreactor (MBR) for wastewater treatment [1]. However, membrane fouling remains a challenge for the wide application of ceramic membrane bioreactors [2]. During membrane filtration, organic foulants in wastewater gradually accumulate on the ceramic membrane surface and block membrane pores, causing severe membrane fouling [3]. To control membrane fouling, membrane cleaning is an essential part of the maintenance of MBR (e.g., backwashing, forward washing and ex-situ cleaning). Fouled membranes are commonly immersed in sodium hypochlorite (NaClO) solutions to remove foulants from membranes in municipal wastewater treatment [4]. However, ex-situ cleaning requires long suspension of MBR, large consumption of chemicals and complex operations. With the aim to reduce the frequency of ex-situ cleaning, in-situ chemical cleaning via backwashing with chemical solutions is employed during filtration [5]. However, traditional in-situ backwashing with chemical agents (e.g., NaClO) poses a high risk of harming the microorganisms in active sludge when the oxidants are backwashed into a mixed liquor [6]. Thus, it is interesting to explore an approach to effectively control membrane fouling but with no harm to microbes and requiring only a short stop time.
On the other hand, membrane surface modification has been studied as a means to mitigate membrane fouling [7][8][9]. With the advantages of strong catalytic activity and a low price, manganese oxides have been considered promising additives for membrane modification [10][11][12]. Byun et al. reported that a ZrO 2 ceramic membrane coated with manganese oxide on its surfaces demonstrated a higher anti-fouling tendency during the filtration of real surface water than membranes coated with TiO 2 or Fe 2 O 3 [13]. Moreover, manganese oxides have been studied as catalysts and activators for oxidants, such as ozone (O 3 ), peroxymonosulfate (PMS) and hydrogen peroxide (H 2 O 2 ), in advanced oxidation processes to generate reactive oxygen species (e.g., hydroxyl radical (HO•), superoxide radical (O 2 • − ), and sulfate radical (SO 4 • − )), which further promote the removal of organic foulants [14][15][16]. Compared with O 3 and PMS, H 2 O 2 has a higher water solubility than O 3 and constitutes a green and environmentally friendly oxidant because it releases only water as its by-product [17]. Thus, manganese oxide-doped Al 2 O 3 ceramic membrane coupling with in-situ H 2 O 2 cleaning is expected to enhance the removal of organic foulants, providing a novel strategy to control membrane fouling.
Another issue regarding in-situ chemical cleaning is the potential harm from the oxidants to the functional bacteria in activated sludge [18]. In order to avoid the undesirable harm to activated sludge, H 2 O 2 solution could be pumped into the membrane through the inner cavity to control the catalytic oxidation of organic foulants occurring within membrane pores. Confined spaces in membrane pores could achieve catalytic performances that are orders of magnitude faster than those obtained in the bulk phase [19]. We found that the ozone decomposition rate inside the membrane pores was about 428 times faster than that in the bulk phase, which was confirmed as a confinement effect of nano-scale membrane pores [20]. Similarly, it is believed that the confinement effect toward catalytic decomposition of H 2 O 2 within membrane pores could also promote the oxidation of foulants and enhance the cleaning efficiency. Overall, doping manganese oxide in an Al 2 O 3 ceramic membrane active layer is expected to bring about a double-win effect, both enhancing the membrane antifouling performance and promoting fouling mitigation via catalytic oxidation of organic foulants by H 2 O 2 within membrane pores.
In this study, an Al 2 O 3 ceramic membrane doped with Mn 2 O 3 in the membrane active layer (Mn-doped membrane) was applied in MBR for the treatment of real municipal wastewater. The fouling mitigation behavior of the Mn-doped membrane during in-situ H 2 O 2 cleaning confined in membrane pores through the inner cavity was investigated and compared with the conventional Al 2 O 3 ceramic membrane (pristine membrane). Furthermore, we discuss the fouling mitigation mechanisms in terms of surface properties, catalytic activity and the confinement effect of Mn-doped catalytic membranes.

Ceramic Membranes and Characterizations
The manganese-doped Al 2 O 3 ceramic membrane (Mn-doped membrane) and Al 2 O 3 ceramic membrane (pristine membrane) were provided by Shenzhen Huayuan Environmental Technology Co., Ltd, Shenzhen, China. Both membranes were composed of a 0.36 mm-thick active layer and 1.1 mm-thick supporting layer. The substrate (supporting layer) of both the Mn-doped membrane and the pristine membrane was made of α-Al 2 O 3 and SiO 2 . The membrane active layer of the Mn-doped membrane was made of 2% Mn 2 O 3 and 98% α-Al 2 O 3 . The pristine membrane was made by the same procedure, without the addition of any manganese oxides.
The surface morphologies and elemental compositions of ceramic membranes were analyzed by a scanning electron microscope combined with an energy dispersive spectrometer (SEM-EDS, SU8010, Hitachi, Tokyo, Japan). The XPS spectra of membrane surfaces were detected by X-ray photoelectron spectroscopy (XPS, PHI 5000 Versa Probe II, ULVAC-PHI, Chigasaki, Japan) using an Al Kα X-ray source.
Catalytic activities of the pristine and Mn-doped membranes toward H 2 O 2 were characterized by the decomposition efficiency of H 2 O 2 . To do so, 1 L of 1 mM H 2 O 2 solution was circularly filtrated with ceramic membranes under a membrane flux of 60 LMH. The catalytic stability of membranes was analyzed by cycle tests. For each cycle test, membranes were cleaned by being immersed in 1 L NaClO solution (1000 ppm available chlorine) for 24 h, then immersed in 1 L pure water for another 24 h and finally dried in an oven at 105 • C before the tests. All cycle tests were run in duplicate under a room temperature ranging from 25-28 • C.

Experimental Setup
As shown in Figure 1, the MBR system consisted of membrane modules, an air aerator and pumps for feed, permeation and cleaning. The membrane tank was 20 cm × 8 cm × 50 cm (L × W × H) and the active volume was 8 L. The surface area of the ceramic membrane was 0.044 m 2 . The ceramic membrane had two outlets in the diagonal. One was connected with the permeation pump and the other was connected with the chemical feeding pump. The latter outlet was closed during membrane filtration and opened during the chemical cleaning processes. During membrane filtration, an aerator provided air aeration for sufficient mixing of liquor in the membrane tank with an aeration rate of 21 L/h. trometer (SEM-EDS, SU8010, Hitachi, Tokyo, Japan). The XPS spectra of membrane surfaces were detected by X-ray photoelectron spectroscopy (XPS, PHI 5000 Versa Probe II, ULVAC-PHI, Chigasaki, Japan) using an Al Kα X-ray source. Catalytic activities of the pristine and Mn-doped membranes toward H2O2 were characterized by the decomposition efficiency of H2O2. To do so, 1 L of 1 mM H2O2 solution was circularly filtrated with ceramic membranes under a membrane flux of 60 LMH. The catalytic stability of membranes was analyzed by cycle tests. For each cycle test, membranes were cleaned by being immersed in 1 L NaClO solution (1000 ppm available chlorine) for 24 hours, then immersed in 1 L pure water for another 24 hours and finally dried in an oven at 105 °C before the tests. All cycle tests were run in duplicate under a room temperature ranging from 25-28 °C.

Experimental Setup
As shown in Figure 1, the MBR system consisted of membrane modules, an air aerator and pumps for feed, permeation and cleaning. The membrane tank was 20 cm × 8 cm × 50 cm (L×W×H) and the active volume was 8 L. The surface area of the ceramic membrane was 0.044 m 2 . The ceramic membrane had two outlets in the diagonal. One was connected with the permeation pump and the other was connected with the chemical feeding pump. The latter outlet was closed during membrane filtration and opened during the chemical cleaning processes. During membrane filtration, an aerator provided air aeration for sufficient mixing of liquor in the membrane tank with an aeration rate of 21 L/h.

Optimal In-Situ H2O2 Cleaning Mode Tests
To prevent H2O2 from entering the membrane tank and damaging the activated sludge during in-situ H2O2 cleaning, the optimal dosing mode of H2O2 was explored. The

Optimal In-Situ H 2 O 2 Cleaning Mode Tests
To prevent H 2 O 2 from entering the membrane tank and damaging the activated sludge during in-situ H 2 O 2 cleaning, the optimal dosing mode of H 2 O 2 was explored. The effect of the in-situ H 2 O 2 cleaning mode (i.e., backwash flux and H 2 O 2 dosage) on the concentration of H 2 O 2 in the membrane tank was studied in pure water tests (see experimental setup in Figure 1). The pristine or Mn-doped membranes were vertically fixed within the membrane tank. Prior to membrane filtration, 6 L of pure water (same volume with the fed mixed liquor during membrane fouling test) was fed into the membrane tank. Membrane filtration with pure water was conducted under a flux of 60 LMH for more than 10 min prior to backwashing. After the membrane inner cavity was fully filled with pure water, H 2 O 2 backwash tests were conducted under different modes (i.e., backwashing of H 2 O 2 solution with dosages of 1, 3 or 5 mM at a flux of 120 LMH for 1 min or backwashing of 1 mM H 2 O 2 solution at a flux of 120 or 180 LMH for 1 min). H 2 O 2 solution was backwashed into membranes through the inner cavity by a cleaning pump, while the feed and permeation pumps were closed. At 0, 1, 2, 5, 10, 15, 20 and 30 min of backwashing, water samples were collected from areas near both sides of the ceramic membrane surface using the five-spot sampling method, and the concentration of H 2 O 2 in the water samples was measured. All experiments were conducted in triplicate. In this study, backwashing with 1 mM H 2 O 2 at a flux of 120 LMH for 1 min was determined to be the optimal dosing mode, which resulted in undetected H 2 O 2 in the membrane tank during backwashing.

Fouling Mitigation Tests
The feed of the ceramic membrane reactor was the supernatant of fresh mixed liquor sampled from the aerobic tank in a municipal wastewater treatment plant in Shenzhen, China. The mixed liquid suspended solids (MLSS) of the initial mixed liquor fed into the membrane tank was 6.8 ± 0.1 g/L and the zeta potential was −12.9 ± 1.0 mV. The sludge volume after 30 min sedimentation (SV 30 ) of active sludge was 38.8 ± 1.5%.
Membrane filtration was conducted at a constant flux of 60 LMH with an on/off ratio of 9 min/1 min. The on/off filtration mode was accurately controlled by time relays. A pressure gauge measured and recorded the real-time transmembrane pressure (TMP).
For the membrane fouling mitigation, two kinds of intermittent in-situ cleaning strategies (clean with H 2 O or 1 mM H 2 O 2 ) were applied to evaluate the mitigation efficiencies of the pristine membrane and Mn-doped membrane. H 2 O or 1 mM H 2 O 2 solution was regularly pumped into ceramic membrane pores every six hours under a flux of 120 LMH for 1 min, and then H 2 O or H 2 O 2 reacted with organic foulants within membrane pores for 10 min after each clean. TMP variations during the membrane filtration were recorded and used as the indicator of membrane fouling. The TMP recovery ratio (R) after each clean was calculated using Equation (1). All filtration experiments ran for 24 h in duplicate under a room temperature ranging from 20-25 • C. R (%) = TMP before cleaning − TMP after cleaning TMP before cleaning × 100 (1) After the filtration experiment, the fouled membranes were rinsed with 100 mL ultrapure water, and then soaked in 500 mL NaClO solution (1000 ppm available chlorine) for 24 h [23,24]. The size-fractionated (≤0.45 µm) cleaning solutions were then analyzed for dissolved organic carbon (DOC). In addition, membrane permeates were collected every six hours during membrane filtration for analysis of water quality (i.e., turbidity, DOC, total nitrogen and total phosphorus). Mixed liquor properties (i.e., MLSS, SV 30 , zeta potential, specific oxygen uptake rate (SOUR) and DOC) were analyzed after membrane filtration. Detailed analytic methods can be found in our recent papers [23,24].

Characterization of Ceramic Membranes
In this study, the Mn element was successfully doped into the membrane layer of the Mn-doped membrane with an atomic percent of 0.7% ( Figure 2). The element compositions in supporting layers of the pristine membrane and Mn-doped membrane were almost the same. The average pore sizes of the pristine and Mn-doped membrane active layers were 780 nm ( Figure S1a) and 792 nm (Figure S1b), respectively. As shown in Figure 3, the Mn 2p 3/2 spectrum showed two distinct peaks at 641.6 eV and 643.2 eV in the Mn-doped membrane layer, which were consistent with the binding energy of Mn(III) and Mn(IV) [25][26][27]. This demonstrated that the doped manganese in the Mn-doped membrane existed as two oxidation states (Mn 3+ and Mn 4+ ). The existence of Mn 3+ /Mn 4+ couples would be beneficial for the catalytic activity of Mn-doped membranes [28]. tions in supporting layers of the pristine membrane and Mn-doped membrane were almost the same. The average pore sizes of the pristine and Mn-doped membrane active layers were 780 nm ( Figure S1a) and 792 nm (Figure S1b), respectively. As shown in Figure 3, the Mn 2p3/2 spectrum showed two distinct peaks at 641.6 eV and 643.2 eV in the Mn-doped membrane layer, which were consistent with the binding energy of Mn(III) and Mn(IV) [25][26][27]. This demonstrated that the doped manganese in the Mn-doped membrane existed as two oxidation states (Mn 3+ and Mn 4+ ). The existence of Mn 3+ /Mn 4+ couples would be beneficial for the catalytic activity of Mn-doped membranes [28].  Figure 4 shows the variation in H2O2 concentration with respect to filtration time with the pristine and Mn-doped membranes in a circular mode. As shown in Figure 4a, less than 1% of H2O2 was decomposed throughout two hours of filtration by the pristine membrane. This indicated that the pristine membrane caused negligible improvement in H2O2 decomposition because of the non-catalytic activity of Al2O3 and SiO2 powders toward H2O2 decomposition ( Figure S2a).

Catalytic Activity of Ceramic Membranes In H2O2 Decomposition
In comparison, more than 98% of H2O2 was constantly decomposed in permeation of the Mn-doped membrane, and the concentration of H2O2 was lower than 0.03 mM. The concentration of H2O2 in the tank with the Mn-doped membrane gradually decreased with increasing filtration time, and the decomposition efficiency of H2O2 solution reached 94.6% after one hour of filtration and 97.5% after two hours of filtration. Obviously, the tions in supporting layers of the pristine membrane and Mn-doped membrane were almost the same. The average pore sizes of the pristine and Mn-doped membrane active layers were 780 nm ( Figure S1a) and 792 nm (Figure S1b), respectively. As shown in Figure 3, the Mn 2p3/2 spectrum showed two distinct peaks at 641.6 eV and 643.2 eV in the Mn-doped membrane layer, which were consistent with the binding energy of Mn(III) and Mn(IV) [25][26][27]. This demonstrated that the doped manganese in the Mn-doped membrane existed as two oxidation states (Mn 3+ and Mn 4+ ). The existence of Mn 3+ /Mn 4+ couples would be beneficial for the catalytic activity of Mn-doped membranes [28].  Figure 4 shows the variation in H2O2 concentration with respect to filtration time with the pristine and Mn-doped membranes in a circular mode. As shown in Figure 4a, less than 1% of H2O2 was decomposed throughout two hours of filtration by the pristine membrane. This indicated that the pristine membrane caused negligible improvement in H2O2 decomposition because of the non-catalytic activity of Al2O3 and SiO2 powders toward H2O2 decomposition ( Figure S2a).

Catalytic Activity of Ceramic Membranes In H2O2 Decomposition
In comparison, more than 98% of H2O2 was constantly decomposed in permeation of the Mn-doped membrane, and the concentration of H2O2 was lower than 0.03 mM. The concentration of H2O2 in the tank with the Mn-doped membrane gradually decreased with increasing filtration time, and the decomposition efficiency of H2O2 solution reached 94.6% after one hour of filtration and 97.5% after two hours of filtration. Obviously, the  Figure 4 shows the variation in H 2 O 2 concentration with respect to filtration time with the pristine and Mn-doped membranes in a circular mode. As shown in Figure 4a, less than 1% of H 2 O 2 was decomposed throughout two hours of filtration by the pristine membrane. This indicated that the pristine membrane caused negligible improvement in H 2 O 2 decomposition because of the non-catalytic activity of Al 2 O 3 and SiO 2 powders toward H 2 O 2 decomposition ( Figure S2a).

Catalytic Activity of Ceramic Membranes in H 2 O 2 Decomposition
In comparison, more than 98% of H 2 O 2 was constantly decomposed in permeation of the Mn-doped membrane, and the concentration of H 2 O 2 was lower than 0.03 mM. The concentration of H 2 O 2 in the tank with the Mn-doped membrane gradually decreased with increasing filtration time, and the decomposition efficiency of H 2 O 2 solution reached 94.6% after one hour of filtration and 97.5% after two hours of filtration. Obviously, the Mn-doped membrane exhibited excellent catalytic activity toward H 2 O 2 decomposition. This was consistent with the finding that Mn 2 O 3 power had very high catalytic activity when it came to H 2 O 2 decomposition ( Figure S2b). Moreover, the catalytic activity of the Mn-doped membrane for H 2 O 2 decomposition did not deteriorate after chemical cleaning with NaClO (1000 ppm) (Figure 4b). The decomposition of H 2 O 2 in the tank with the Mn-doped membrane followed first-order kinetics, and the average observed rate constant was 4.6 × 10 −2 min −1 . There was no significant difference in the observed rate constants for the first, third, fifth and tenth cycle tests. During the chemical cleaning with NaClO (1000 ppm, pH > 9), the concentration of Mn in the NaClO immersion solution was below the detection limit (0.001 mg/L) of flame atomic absorption spectrometry. Mn leaching was not significantly observed during membrane chemical cleaning with NaClO solution at an alkaline pH. These results indicate the high catalytic activity and stability of the Mn-doped membrane when it came to H 2 O 2 decomposition. Mn-doped membrane exhibited excellent catalytic activity toward H2O2 decomposition. This was consistent with the finding that Mn2O3 power had very high catalytic activity when it came to H2O2 decomposition ( Figure S2b). Moreover, the catalytic activity of the Mn-doped membrane for H2O2 decomposition did not deteriorate after chemical cleaning with NaClO (1000 ppm) (Figure 4b). The decomposition of H2O2 in the tank with the Mn-doped membrane followed first-order kinetics, and the average observed rate constant was 4.6 × 10 −2 min −1 . There was no significant difference in the observed rate constants for the first, third, fifth and tenth cycle tests. During the chemical cleaning with NaClO (1000 ppm, pH > 9), the concentration of Mn in the NaClO immersion solution was below the detection limit (0.001 mg/L) of flame atomic absorption spectrometry. Mn leaching was not significantly observed during membrane chemical cleaning with NaClO solution at an alkaline pH. These results indicate the high catalytic activity and stability of the Mn-doped membrane when it came to H2O2 decomposition. It has been reported that ROS are generated from H2O2 decomposition. EPR tests were conducted to confirm the generation of ROS from H2O2 decomposition catalyzed by the pristine and Mn-doped membranes. Four characteristic peaks of DMPO-HO• with an intensity ratio of 1:2:2:1 (Figure 4c) were observed in 1 mM H2O2 solution without catalysis of the pristine or Mn-doped membranes as a result of the self-decomposition of H2O2  [32,33]. This finding directly suggested that Mn doping is a good strategy for enhancing ceramic membrane catalytic activity toward H 2 O 2 , thus decomposing H 2 O 2 into more powerful ROS, namely HO• or 1 O 2 [34].

Membrane Fouling Mitigation Performance With and Without In-Situ H2O2 Cleaning
As seen from Figure 6a, the Mn-doped membrane demonstrated an obviously lower TMP increase rate than the pristine membrane. The TMP of the pristine membrane with H2O cleaning rapidly increased to −42.1 kPa after 24 hours of filtration, while the TMP of the Mn-doped membrane with H2O cleaning only increased to −30.4 kPa. After membrane filtration, the total DOC content in the fouling layers of the Mn-doped membrane with H2O cleaning (11.0 mg) was also lower than that of the pristine membrane with H2O cleaning (12.8 mg) (Figure 6c). These results demonstrated the improved antifouling properties of the Mn-doped membrane. With H2O2 cleaning, the TMP of the pristine membrane increased to −40.5 kPa after 24-hour filtration, while the TMP of the Mn-doped membrane increased to −25.1 kPa. The total DOC content in the fouling layers decreased When the pristine and Mn-doped membranes were backwashed with 1 mM H 2 O 2 at a flux of 120 LMH for 1 min, the concentrations of H 2 O 2 in the pristine and Mn-doped membrane tanks were all below the detection limit (0.0004 mM) of the spectrophotometriciodide method. Obviously, almost all H 2 O 2 was confined within the membrane under this dosing mode, causing an undetectable change in the H 2 O 2 concentration in the membrane tank (volume of 6 L). Therefore, with the aim to protect activated sludge from being damaged by residual H 2 O 2 in the membrane tank, we conducted backwashing with 1 mM H 2 O 2 at a flux of 120 LMH for 1 min as the optimal dosing mode for in-situ confined H 2 O 2 cleaning.

Membrane Fouling Mitigation Performance with and without In-Situ H 2 O 2 Cleaning
As seen from Figure 6a It is worth noting that the H2O2 concentrations in both the pristine membrane and Mn-doped membrane tank were below the detection limit (<0.0004 mM) after in-situ H2O2 cleaning. This result demonstrated that in-situ H2O2 cleaning with the optimal mode of 1 mM H2O2 at a flux of 120 LMH for 1 min caused an undetectable change in the H2O2 concentration in the membrane tank, which was consistent with the results of pure water The TMP recovery ratio (R) achieved by each cleaning mode was calculated to investigate the efficiencies of different cleaning strategies. As shown in Figure 6b, (Figure 6b), but the overall TMP increase and the total DOC content in membrane fouling layers of the former (∆TMP = 27.7 kPa in 24 h, the total content of DOC = 11.0 mg) were much lower than of the latter (∆TMP = 37.3 kPa, the total content of DOC = 12.4 mg). This further proved that antifouling properties of membranes were more critical than the cleaning strategy for fouling mitigation when it came to membranes without catalytic activity toward H 2 O 2 .
It is worth noting that the H 2 O 2 concentrations in both the pristine membrane and Mn-doped membrane tank were below the detection limit (<0.0004 mM) after in-situ H 2 O 2 cleaning. This result demonstrated that in-situ H 2 O 2 cleaning with the optimal mode of 1 mM H 2 O 2 at a flux of 120 LMH for 1 min caused an undetectable change in the H 2 O 2 concentration in the membrane tank, which was consistent with the results of pure water tests ( Figure 5). Moreover, there was no obvious difference of mixed liquor properties (i.e., DOC, MLSS, SV 30 , zeta potential, SOUR, pH) between four membrane bioreactors tested with H 2 O or H 2 O 2 cleaning (Table S1). This verified that in-situ H 2 O 2 cleaning with the optimal mode did not affect the mixed liquor properties in the membrane bioreactor. Activated sludge of MBRs was well-protected from being damaged by H 2 O 2 . The water quality (i.e., turbidity, DOC, total nitrogen and total phosphorus) of permeation ( Figure  S3) also showed no significant difference between the pristine membrane and Mn-doped membrane with H 2 O or H 2 O 2 cleaning.
Overall, coupling the Mn-doped membrane and intermittent in-situ H 2 O 2 cleaning achieved the best fouling mitigation performance with the lowest TMP increase and the highest TMP recovery efficiency, causing no harm to the activated sludge. All the results indicated that the Mn-doped membrane exhibited a better antifouling performance than the pristine membrane during filtration of real municipal wastewater. This was attributed to the modified surface properties after Mn doping in the membrane layer, including an intensified electric charge and heightened hydrophilicity [35]. In this study, the pristine membrane layer consisted of Al 2 O 3 that was positively charged at pH 7.0, while the Mn-doped membrane layer was doped with Mn 2 O 3 that was negatively charged at the same pH. Mn doping changed the membrane surface charge from positive (pristine membrane) to negative (Mn-doped membrane). Organic foulants, especially biopolymers, were the main contributors to membrane fouling as a result of the high rejection of biopolymers by the pristine and Mn-doped membranes (Figures S4 and S5). Accordingly, the interactions between organic foulants and the ceramic membrane surface changed significantly after Mn doping. The negatively charged surface of the Mn-doped membrane changed the electrostatic attraction force between organic foulants and the pristine membrane to an electrostatic repulsion force [36][37][38]. This inhibited the adsorption and accumulation of organic foulants on the Mn-doped membrane surfaces and further in the pores. Moreover, the pure water contact angle of the Mn-doped membrane (24.1 • ) was 24.5% lower than that of the pristine membrane (31.9 • ) ( Figure S6). This meant better surface hydrophilicity was achieved by Mn doping. This was in agreement with the analysis of the O 1s spectrum ( Figure S7), which showed a greater amount of hydroxyl groups in the Mn-doped membrane. The increase of hydrogen-bond interactions between hydroxyl groups and water made the Mn-doped membrane more hydrophilic [39,40]. Higher hydrophilicity of the Mn-doped membrane weakened the hydrophobic interaction between organic compounds and membrane surface [41,42]. The biopolymer rejection rate by the Mn-doped membrane (49.0%) was lower than that by the pristine membrane (53.9%). Subsequently, fewer organic foulants attached to membrane/pore surfaces (Figure 6c) during membrane filtration, leading to a lower TMP increase rate and mitigating the Mn-doped membrane fouling [13,43]. Usually, hydrophilic membranes require a lower pressure than hydrophobic membranes to obtain the same permeation flux [44][45][46][47]. This is consistent with the fact that the operation TMP of the Mn-doped membrane was lower than that of the pristine membrane at the same flux ( Figure S8). Therefore, with a lower operation TMP and slower TMP increase rate, the final TMP of the Mn-doped membrane with H 2 O cleaning (−30.4 kPa) was 27.8% lower than that of the pristine membrane with H 2 O cleaning (−42.1 kPa) after 24 h of filtration of real municipal wastewater.

Intensified Catalytic Activity of Mn-Doped Membrane toward H 2 O 2 Decomposition
Coupling the Mn-doped membrane with in-situ H 2 O 2 cleaning exhibited the highest TMP recovery after H 2 O 2 cleaning. This was attributed to the enhanced catalytic activity toward H 2 O 2 decomposition inside the Mn-doped membrane. The existence of Mn 3+ /Mn 4+ couples (i.e., reversible transition between Mn 3+ and Mn 4+ ) in the Mn-doped membrane layer ( Figure 3) played a crucial role in improving the catalytic activity of the membranes toward H 2 O 2 decomposition and generation of ROS (i.e., HO• and 1 O 2 ) [15,28,48]. When H 2 O 2 solution was pumped into the Mn-doped membrane, available Mn 3+ /Mn 4+ couples at the manganese-based catalyst surface effectively catalyzed the decomposition of H 2 O 2 (>98% of 1 mM H 2 O 2 decomposed inside membrane pores (Figure 4a,b)) to continuously produce ROS (i.e., HO• and 1 O 2 (Figure 4c,d)) (see Equations (3) and (4) in Figure 7) [14,28,33,48]. The oxidizing powers of generated ROS, especially HO•, toward organic matters are much stronger than that of H 2 O 2 [49,50]. The adsorbed organic foulants can be efficiently oxidized by generated ROS into smaller molecules and finally removed from the Mn-doped membrane. Therefore, the total DOC content in the fouling layers of the Mn-doped membrane decreased from 11.0 mg (with H 2 O cleaning) to 9.8 mg (with H 2 O 2 cleaning) ( Figure 6c). However, the pristine membrane consisted of Al 2 O 3 and SiO 2 that had no catalytic activity toward H 2 O 2 decomposition ( Figure S2a). The decomposition efficiency of H 2 O 2 with the pristine membrane was very low (<1% within two hours) (Figure 4a) and led to a very limited amount of HO• (Figure 4c) mostly from the self-decomposition of 1 mM H 2 O 2 (see Equation (2) in Figure 7). Consequently, the low decomposition rate of H 2 O 2 inside the pristine membrane largely limited the oxidation removal efficiency of organic foulants of the fouled pristine membrane during in-situ H 2 O 2 cleaning with a short stop time (10 min). The total DOC content in the fouling layers of the pristine membrane only decreased from 12.8 mg (with H 2 O cleaning) to 12.4 mg (with H 2 O 2 cleaning) (Figure 6c). Therefore, the TMP recovery efficiency of the Mn-doped membrane after H 2 O 2 cleaning was significantly higher than that of the pristine membrane.
low decomposition rate of H2O2 inside the pristine membrane largely limited the oxidation removal efficiency of organic foulants of the fouled pristine membrane during in-situ H2O2 cleaning with a short stop time (10 minutes). The total DOC content in the fouling layers of the pristine membrane only decreased from 12.8 mg (with H2O cleaning) to 12.4 mg (with H2O2 cleaning) (Figure 6c). Therefore, the TMP recovery efficiency of the Mn-doped membrane after H2O2 cleaning was significantly higher than that of the pristine membrane.

Confined Catalytic Oxidation Performance Within Mn-Doped Membrane Pores
In addition, H2O2 solutions were pumped into the membrane and confined inside the membrane pores almost without detectable residual H2O2 entering the membrane tank. Limited reaction space within the Mn-doped membrane pores further enhanced the oxidation of organic foulants by enriched ROS [51]. It is well-known that the lifetimes of HO• (< 1 μs) and 1 O2 (~3 μs) in water are extremely short, resulting in elimination before they react with pollutants [52]. In heterogeneous catalytic reaction systems, the concentration of ROS is the highest on the catalyst surface and then rapidly decreases with increasing diffusion length [19]. Assuming the lifetimes (τ) of HO• and 1 O2 are 1 μs and 3 μs, respectively, at a neutral pH [53], the diffusion lengths (λL) of HO• and 1 O2 in the aqueous phase should be ~96 nm and ~159 nm, respectively, as calculated by  [54]. Thus, the oxidation efficiency of organic foulants by generated ROS would be significantly improved if the reaction space were restricted far below the diffusion length scale of ROS. This phenomenon was reported in previous studies as a spatial confinement effect [20,19].
Organic foulants adsorbed on the membrane/pore surfaces during membrane filtration. When H2O2 was pumped into membrane pores through the inner cavity, ROS (HO• and 1 O2) were generated on the Mn-doped membrane/pore surfaces and were restricted within the Mn-doped membrane pores (average pore size of 792 nm (Figure S1b)). The oxidation of organic foulants on the Mn-doped membrane/pore surfaces occurred once ROS were generated. As shown in Figure 8, the confined reaction space within the Mn-doped membrane pores shortened the diffusion lengths of ROS and prevented their elimination, leading to enriched ROS within the membrane pores. Because of the substantially reduced diffusion length and concentrated reactants, ROS reacted with organic

Confined Catalytic Oxidation Performance within Mn-Doped Membrane Pores
In addition, H 2 O 2 solutions were pumped into the membrane and confined inside the membrane pores almost without detectable residual H 2 O 2 entering the membrane tank. Limited reaction space within the Mn-doped membrane pores further enhanced the oxidation of organic foulants by enriched ROS [51]. It is well-known that the lifetimes of HO• (<1 µs) and 1 O 2 (~3 µs) in water are extremely short, resulting in elimination before they react with pollutants [52]. In heterogeneous catalytic reaction systems, the concentration of ROS is the highest on the catalyst surface and then rapidly decreases with increasing diffusion length [19]. Assuming the lifetimes (τ) of HO• and 1 O 2 are 1 µs and 3 µs, respectively, at a neutral pH [53], the diffusion lengths (λ L ) of HO• and 1 O 2 in the aqueous phase should be~96 nm and~159 nm, respectively, as calculated by λ L = 2 √ τ × D (where D is the diffusion coefficient; D HO• = 2.3 × 10 −9 m 2 s −1 and D1 O 2 = 2.1 × 10 −9 m 2 s −1 ) [54]. Thus, the oxidation efficiency of organic foulants by generated ROS would be significantly improved if the reaction space were restricted far below the diffusion length scale of ROS. This phenomenon was reported in previous studies as a spatial confinement effect [19,20].
Organic foulants adsorbed on the membrane/pore surfaces during membrane filtration. When H 2 O 2 was pumped into membrane pores through the inner cavity, ROS (HO• and 1 O 2 ) were generated on the Mn-doped membrane/pore surfaces and were restricted within the Mn-doped membrane pores (average pore size of 792 nm ( Figure S1b)). The oxidation of organic foulants on the Mn-doped membrane/pore surfaces occurred once ROS were generated. As shown in Figure 8, the confined reaction space within the Mn-doped membrane pores shortened the diffusion lengths of ROS and prevented their elimination, leading to enriched ROS within the membrane pores. Because of the substantially reduced diffusion length and concentrated reactants, ROS reacted with organic foulants within membrane pores with much higher efficiency than they did in the bulk phase [19,55]. In this way, organic foulants were efficiently removed, and membrane fouling was well controlled via in-situ H 2 O 2 cleaning confined in the Mn-doped membrane pores.
Overall, coupling the Mn-doped membrane with in-situ H 2 O 2 cleaning achieved the best fouling mitigation performance with the lowest TMP increase, the highest TMP recovery efficiency and the fewest DOC in fouling layers, as a result of improved surface properties and confined catalytic oxidation of organic foulants by H 2 O 2 inside membrane pores.
foulants within membrane pores with much higher efficiency than they did in the bulk phase [55,19]. In this way, organic foulants were efficiently removed, and membrane fouling was well controlled via in-situ H2O2 cleaning confined in the Mn-doped membrane pores. Overall, coupling the Mn-doped membrane with in-situ H2O2 cleaning achieved the best fouling mitigation performance with the lowest TMP increase, the highest TMP recovery efficiency and the fewest DOC in fouling layers, as a result of improved surface properties and confined catalytic oxidation of organic foulants by H2O2 inside membrane pores.

Conclusions
In-situ H2O2 cleaning confined to the inside of the membrane was applied to mitigate membrane fouling in real municipal wastewater treatment. An Mn-doped membrane was used in a membrane bioreactor for wastewater treatment with the aim to improve the membrane catalytic activity toward oxidation of foulants by H2O2. The Mn-doped membrane with in-situ H2O2 cleaning demonstrated a superior fouling mitigation performance to that of a pristine ceramic membrane with no harm to the activated sludge. After 24-hour filtration, the ΔTMP of the Mn-doped membrane with H2O2 cleaning was 22.2 kPa, which was 40.5% lower than that of the pristine membrane with H2O2 cleaning (37.3 kPa). The enhanced fouling mitigation was attributed to Mn doping that ameliorated membrane surface properties and catalyzed oxidation of foulants by H2O2 confined in the membrane pores. Mn 3+ /Mn 4+ redox couples in the active layer of the Mn-doped membrane catalyzed H2O2 decomposition continuously to generate ROS (HO• and 1 O2), which were confined within the membrane pores and efficiently degraded foulants. Overall, coupling the Mn-doped membrane with confined H2O2 cleaning achieved satisfactory fouling mitigation with a low TMP increase and high TMP recovery efficiency. The findings provide a new strategy for the rational design of antifouling membranes and fouling mitigation processes.

Supplementary Materials:
The following are available online at www.mdpi.com/xxx/s1, Figure S1. The pore size distribution of the (a) pristine and (b) Mn-doped membrane layers; Figure S2

Conclusions
In-situ H 2 O 2 cleaning confined to the inside of the membrane was applied to mitigate membrane fouling in real municipal wastewater treatment. An Mn-doped membrane was used in a membrane bioreactor for wastewater treatment with the aim to improve the membrane catalytic activity toward oxidation of foulants by H 2 O 2 . The Mn-doped membrane with in-situ H 2 O 2 cleaning demonstrated a superior fouling mitigation performance to that of a pristine ceramic membrane with no harm to the activated sludge. After 24-h filtration, the ∆TMP of the Mn-doped membrane with H 2 O 2 cleaning was 22.2 kPa, which was 40.5% lower than that of the pristine membrane with H 2 O 2 cleaning (37.3 kPa). The enhanced fouling mitigation was attributed to Mn doping that ameliorated membrane surface properties and catalyzed oxidation of foulants by H 2 O 2 confined in the membrane pores. Mn 3+ /Mn 4+ redox couples in the active layer of the Mn-doped membrane catalyzed H 2 O 2 decomposition continuously to generate ROS (HO• and 1 O 2 ), which were confined within the membrane pores and efficiently degraded foulants. Overall, coupling the Mn-doped membrane with confined H 2 O 2 cleaning achieved satisfactory fouling mitigation with a low TMP increase and high TMP recovery efficiency. The findings provide a new strategy for the rational design of antifouling membranes and fouling mitigation processes.  Figure S5. MW distribution of DOC in the physically removable fouling layer after 24 h of filtration; Figure S6. Pure water contact angles of the pristine (a) and Mn-doped membranes (b); Figure S7. XPS spectra of O 1s of the pristine and Mn-doped ceramic membranes; Figure S8. Pure water permeabilities of the pristine membrane and Mn-doped membrane; Table S1. Mixed liquor properties after membrane filtration.