Alleviation of Ultrafiltration Membrane Fouling by ClO2 Pre-Oxidation: Fouling Mechanism and Interface Characteristics

In order to alleviate membrane fouling and improve removal efficiency, a series of pretreatment technologies were applied to the ultrafiltration process. In this study, ClO2 was used as a pre-oxidation strategy for the ultrafiltration (UF) process. Humic acid (HA), sodium alginate (SA), and bovine serum albumin (BSA) were used as three typical organic model foulants, and the mixture of the three substances was used as a representation of simulated natural water. The dosages of ClO2 were 0.5, 1, 2, 4, and 8 mg/L, with 90 min pre-oxidation. The results showed that ClO2 pre-oxidation at low doses (1–2 mg/L) could alleviate the membrane flux decline caused by humus, polysaccharides, and simulated natural water, but had a limited alleviating effect on the irreversible resistance of the membrane. The interfacial free energy analysis showed that the interaction force between the membrane and the simulated natural water was also repulsive after the pre-oxidation, indicating that ClO2 pre-oxidation was an effective way to alleviate cake layer fouling by reducing the interaction between the foulant and the membrane. In addition, ClO2 oxidation activated the hidden functional groups in the raw water, resulting in an increase in the fluorescence value of humic analogs, but had a good removal effect on the fluorescence intensity of BSA. Furthermore, the membrane fouling fitting model showed that ClO2, at a low dose (1 mg/L), could change the mechanism of membrane fouling induced by simulated natural water from standard blocking and cake layer blocking to critical blocking. Overall, ClO2 pre-oxidation was an efficient pretreatment strategy for UF membrane fouling alleviation, especially for the fouling control of HA and SA at low dosages.


Introduction
Formed as a result of the interaction between the hydrological cycle, the biosphere, and the lithosphere, natural organic matter (NOM) is commonly found in surface water and groundwater, with diverse distribution patterns [1]. Recently, scholars have observed that the content of NOM in water bodies is increasing, accompanied by severe seasonal changes, and can absorb ultraviolet rays to protect harmful substances, such as pathogens, and reduce the self-purification capacity of surface water [2,3]. In addition, Bond [4] et al. reported that NOM is the precursor of disinfection byproducts such as trihalomethanes (THMs) and haloacetic acids (HAAs). Therefore, NOM can inevitably pose a threat to the operation of water plants. Some components of NOM are collectively referred to as humus, because they have no separate chemical formulae [5]. Lipczynska-Kochany et al. [2] found that humus could be divided into humin, humic acid (HA), and fulvic acid (FA) according to molecular weight, among which HA is considered to be the dominant component of humus. Nevertheless, HA is prone to forming stronger toxic complexes with heavy metals

Experimental Setup
The dosages of AR-grade ClO 2 were 0.5, 1, 2, 4, and 8 mg/L. The experimental sewages were pre-oxidized for 90 min. Then, 60 mL of solution was taken out for filtration experiments. Flat ultrafiltration membranes of polyethersulfone (PES, 100 kDa) were used in the filtration experiments, which were obtained from the MICRODYN-NADIR company (Wiesbaden, Germany). The ultrafiltration device maintained a constant pressure (100 kPa) with a dead-end ultrafiltration system, without stirring of the feed solution. The device consists of an ultrafilter cup, an electronic balance (NV2201ZH, Auhaus, Shanghai, China), a nitrogen cylinder, a pressure gauge, a control valve, and a computer terminal. The cleaned ultrafiltration membrane was placed at the bottom of the ultrafiltration cup, and the pressure in the ultrafiltration cup was kept low enough that the ultrafiltration cup did not leak air or water. Under this pressure, the liquid in the ultrafilter cup flowed through the silicone tube into the beaker placed on the electronic balance. The balance was set to record the weight data at intervals of 3 s. The weight data recorded by the electronic balance were then transmitted to the computer terminal through the data line in order to obtain the ultrafiltration weight data.

Membrane Fouling Assessment and Mechanism Analysis
The resistance of the ultrafiltration membrane is composed of inherent resistance, reversible resistance, and irreversible resistance [30]. It is generally believed that the reversible resistance is caused by the cake layer on the membrane surface. The reversible resistance can be greatly reduced by gently wiping the cake layer on the membrane surface with wet sponge. Irreversible resistance is usually caused by membrane pore narrowing. The specific calculation can be deduced by the Darcy formula [31,32], as shown in Equation (1): where J represents the filtration flux (L/m 2 ·h), µ represents the dynamic viscosity (Paxs), ∆P represents the operating pressure (Pa), R tot represents the total membrane resistance (m −1 ), R i represents the membrane's inherent resistance (m −1 ), R b represents irreversible resistance (m −1 ), and R c represents the reversible resistance (m −1 ). Ultrapure water filtration was used before each filtration until the flux was stable, and the steady membrane flux was denoted as J 0 . All filtration tests were performed at least 3 times. The R i can be calculated according to Equation (2): At the end of filtration, the average flux of the last 3 mL was denoted as J 1 to calculate the total membrane resistance, as shown in Equation (3):

of 17
In order to calculate the irreversible resistance of the membrane, after each filtration cycle, the ultrafiltration membrane was reset in the ultrafiltration device after gently wiping the cake layer on the membrane surface with a wet sponge under water flushing, and pure water was filtered again, i.e., the stable membrane flux after cleaning was obtained, denoted as J 2 . Therefore, the irreversible resistance R b can be calculated according to Equation (4). The reversible resistance can be calculated by subtracting the inherent resistance and irreversible resistance from the total membrane resistance.
In order to further investigate the membrane fouling mechanism during ultrafiltration, four classical fouling models were introduced, including a complete blocking model, critical blocking model, standard blocking model, and cake layer filtration model. The complete blocking model assumes that the pores of the ultrafiltration membrane are completely blocked, so that the transmembrane resistance increases. The standard blocking model assumes that the small-particle-size foulants adhere to the inner sides of the membrane pores, narrowing them and reducing the flow capacity. The critical blocking model is between the complete blocking model and the standard blocking model. The cake layer filtration model refers to the formation of a cake layer by foulants deposited on the membrane surface. Ho and Zydney [33]'s differential form model was introduced in this study to analyze and fit the flux data. The advantage of this method is that it can more intuitively analyze the fouling mechanism under different filtration times or volumes. MATLAB was used to conduct mathematical modeling for the imported data, and cubic polynomial fitting was performed for the flux data. The formulae were as follows: where t represents the filtration time (s), J represents the filtration flux, and V represents the total filtration volume (mL). After finding the turning point by fitting the curve, the data before and after the change point were fitted respectively to determine the leading mechanism in different groups of fouling, according to the obtained N value: when n = 2, it is mainly a complete blocking model; when n = 1.5, standard blocking plays the dominant role; when n = 1, it is critical blocking; when n = 0, it is cake layer filtration [34].

Analytical Methods
The ultraviolet absorbance (UV 254 ) was determined using an ultraviolet spectrophotometer (T6, PUXI, Beijing, China), and dissolved organic carbon (DOC) was determined using a total organic carbon analyzer (TOC-VCSH, Shimadzu, Kyoto, Japan). pH was measured with a pH meter (868-2, Orion, Shanghai, China). Because the humic and tryptophan compounds in the target foulant have fluorescent excitation properties, the fluorescent compounds in the solution can be determined using the three-dimensional fluorescence excitation-emission matrix (EEM) spectra. A fluorescence spectrometer (F-7000, Hitachi, Tokyo, Japan) was used for determination. The excitation light (Ex) wavelength range was 200-450 nm, and the emission light (Em) wavelength range was 250-550 nm; the scanning intervals were 5 nm and 1 nm, respectively. The sample solution was filtered through a 0.45 µm microfiltration membrane, and the pH value was adjusted to 7.0 before the EEM test. The contact angle of the liquid drops to the membrane surface was measured using a contact angle tester (JYSP-360, Beijing, China). Based on the extended Derjaguin-Landau-Verwey-Overbeek (XDLVO) theory, diiodomethane, ultrapure water, and glycerol were selected as test liquids to calculate the interfacial free energy. Computational methods of cohesion and adhesion free energy can be found in a previous work [35]. Scanning electron microscopy (Gemini SEM 300, Zeiss, Germany) was used to observe the morphology and microstructure of foulant particles on the ultrafiltration membrane surface, and the detailed working conditions were similar to those in previous studies [36,37]. The alleviating effect of ClO 2 pre-oxidation on ultrafiltration membrane fouling is shown in Figure 1. As shown in the figure, the decrease in membrane flux caused by raw humic acid water was more obvious than that after ClO 2 pre-oxidation. At the end of the first, second, and third filtration cycles, the specific flux (J/J 0 ) decreased to 0.36, 0.31, and 0.28, respectively. ClO 2 pre-oxidation significantly improved the flux, and the optimal flux mitigation was obtained when the dosage was 2 mg/L. At the end of each cycle of 2 mg/L ClO 2 , the specific flux J/J 0 rose to 0.45, 0.40, and 0.38, respectively. However, when the ClO 2 dosage was increased to 8 mg/L, the J/J 0 at the end of each filtration cycle was decreased to 0.41, 0.37, and 0.34, respectively. Compared with the dosage of 2 mg/L, the ability to recover J/J 0 at 8 mg/L was limited; thus, it could be seen that there was no positive correlation between ClO 2 dosage and flux recovery, which might be related to the membrane fouling mechanism [38].

ClO2 Pre-Oxidation for SA Fouling
As depicted in Figure 2a, the decrease in membrane flux caused by SA is more obvious than that caused by humic acid. The specific flux remained flat or even lower at the end of each cycle as the ClO2 dose increased from 0 mg/L to 8 mg/L. When the ClO2 dosage was 2 mg/L, although the flux decrease was relieved to some extent, the specific flux at the end of each filtration cycle was close to the value of sodium alginate raw water. Compared with humic acid, flux recovery of sodium alginate after hydraulic backwash was more apparent. As shown in Figure 2b, sodium alginate mainly caused severe reversible resistance, with a higher proportion than that in humic acid, which could be eliminated by high-intensity backwashing. With the dosage of ClO2 above 2 mg/L, the irreversible fouling resistance decreased, indicating that ClO2 pretreatment alleviated the irreversible fouling caused by sodium alginate to a certain extent. On the other hand, the reversible fouling caused by sodium alginate increased significantly with the dosing of ClO2, which was the most serious when the dosage was 8 mg/L, showing an increased reversible fouling resistance of 2.19 × 10 11 m −1 compared with the sodium alginate raw water.
It could be seen that reversible fouling with pretreatment of ClO2 played a leading In addition, after primary and secondary backwashing, the flow recovery rate was lower by 73% and 51%, respectively, indicating that the flow recovery rate of hydraulic backwashing was effectual, and that reversible fouling was dominant in HA fouling. As shown in Figure 1b, the total fouling resistance and reversible fouling resistance caused by raw humic acid water were the highest in the three cycles, and the total fouling resistance of different ClO 2 dosage groups decreased at first, and then increased over three filtration cycles, during which the optimal fouling resistance alleviation was achieved when the dosage was 2 mg/L. On the other hand, the proportion of irreversible fouling resistance increased, and the decline in the degree of resistance was relatively lower-even when the dosage was 4 mg/L, there was an increase at the end of the third filtration cycle. These results indicated that ClO 2 pre-oxidation had a good effect on improving permeate flux and alleviating reversible fouling during humic acid filtration, and that the filtration perfor- As depicted in Figure 2a, the decrease in membrane flux caused by SA is more obvious than that caused by humic acid. The specific flux remained flat or even lower at the end of each cycle as the ClO 2 dose increased from 0 mg/L to 8 mg/L. When the ClO 2 dosage was 2 mg/L, although the flux decrease was relieved to some extent, the specific flux at the end of each filtration cycle was close to the value of sodium alginate raw water. Compared with humic acid, flux recovery of sodium alginate after hydraulic backwash was more apparent. As shown in Figure 2b, sodium alginate mainly caused severe reversible resistance, with a higher proportion than that in humic acid, which could be eliminated by high-intensity backwashing. With the dosage of ClO 2 above 2 mg/L, the irreversible fouling resistance decreased, indicating that ClO 2 pretreatment alleviated the irreversible fouling caused by sodium alginate to a certain extent. On the other hand, the reversible fouling caused by sodium alginate increased significantly with the dosing of ClO 2 , which was the most serious when the dosage was 8 mg/L, showing an increased reversible fouling resistance of 2.19 × 10 11 m −1 compared with the sodium alginate raw water.
It could be seen that reversible fouling with pretreatment of ClO 2 played a leading role in membrane fouling induced by sodium alginate. This was consistent with previous research results, which could be attributed to the fact that sodium alginate can rapidly form a gel layer or cake layer on the membrane surface, causing membrane surface fouling [40].

ClO2 Pre-Oxidation for SA Fouling
As depicted in Figure 2a, the decrease in membrane flux caused by SA is more obvious than that caused by humic acid. The specific flux remained flat or even lower at the end of each cycle as the ClO2 dose increased from 0 mg/L to 8 mg/L. When the ClO2 dosage was 2 mg/L, although the flux decrease was relieved to some extent, the specific flux at the end of each filtration cycle was close to the value of sodium alginate raw water. Compared with humic acid, flux recovery of sodium alginate after hydraulic backwash was more apparent. As shown in Figure 2b, sodium alginate mainly caused severe reversible resistance, with a higher proportion than that in humic acid, which could be eliminated by high-intensity backwashing. With the dosage of ClO2 above 2 mg/L, the irreversible fouling resistance decreased, indicating that ClO2 pretreatment alleviated the irreversible fouling caused by sodium alginate to a certain extent. On the other hand, the reversible fouling caused by sodium alginate increased significantly with the dosing of ClO2, which was the most serious when the dosage was 8 mg/L, showing an increased reversible fouling resistance of 2.19 × 10 11 m −1 compared with the sodium alginate raw water.
It could be seen that reversible fouling with pretreatment of ClO2 played a leading role in membrane fouling induced by sodium alginate. This was consistent with previous research results, which could be attributed to the fact that sodium alginate can rapidly form a gel layer or cake layer on the membrane surface, causing membrane surface fouling [40].

ClO 2 Pre-Oxidation for BSA Fouling
The specific flux of different ClO 2 doses dropped sharply during the initial filtration cycle, after which the flux curve became smooth in the third cycle (Figure 3a). At the end of the first, second, and third filtration cycles, bovine serum protein caused a severe decrease in flux, with specific fluxes dropping to 0.46, 0.32, and 0.17, respectively. Moreover, the flux was difficult to recover after physical cleaning. Although the flux in each cycle increased slightly after hydraulic flushing with the dosing of ClO 2 , the flux decreased more significantly with the increase in the dose during the first two filtration cycles, indicating that ClO 2 pre-oxidation had no alleviating effect on flux reduction caused by bovine serum protein.
As shown in Figure 3b, irreversible and reversible fouling resistance caused by BSA at the end of filtration accounted for 62.8% and 27.2% of the total resistance, respectively. It could be concluded that irreversible resistance played the dominant role in membrane fouling caused by bovine serum protein. Thus, the membrane flux was not well alleviated after hydraulic cleaning; however, after the pre-oxidation with ClO 2 (0.50-8 mg/L), both reversible and irreversible fouling were not only not alleviated, but even aggravated, at dosages below 1 mg/L. In addition, Cheng et al.'s experimental results also showed that lower doses of PMS as an oxidant aggravated membrane fouling by BSA [38]. These results indicate that ClO 2 pre-oxidation cannot improve the flux and reduce the fouling resistance during bovine serum albumin filtration. A possible reason for the limited fouling mitigation effect of pre-oxidation on BSA-especially compared to the mitigation effect for HA and SA-is the high molecular weight of BSA, which cannot be efficient degraded by a low-dosage oxidation process.
cating that ClO2 pre-oxidation had no alleviating effect on flux reduction caused by bovine serum protein.
As shown in Figure 3b, irreversible and reversible fouling resistance caused by BSA at the end of filtration accounted for 62.8% and 27.2% of the total resistance, respectively. It could be concluded that irreversible resistance played the dominant role in membrane fouling caused by bovine serum protein. Thus, the membrane flux was not well alleviated after hydraulic cleaning; however, after the pre-oxidation with ClO2 (0.50-8 mg/L), both reversible and irreversible fouling were not only not alleviated, but even aggravated, at dosages below 1 mg/L. In addition, Cheng et al.'s experimental results also showed that lower doses of PMS as an oxidant aggravated membrane fouling by BSA [38]. These results indicate that ClO2 pre-oxidation cannot improve the flux and reduce the fouling resistance during bovine serum albumin filtration. A possible reason for the limited fouling mitigation effect of pre-oxidation on BSA-especially compared to the mitigation effect for HA and SA-is the high molecular weight of BSA, which cannot be efficient degraded by a low-dosage oxidation process.

ClO2 Pre-Oxidation for NOM Fouling
In actual water treatment, humus, proteins, and polysaccharides generally exist together. Therefore, in order to further explore the influence of ClO2 as a pre-oxidant on membrane fouling caused by compound foulants, raw water mixed with humic acid, sodium alginate, and bovine serum protein was prepared. Figure 4 shows the influence of ClO2 pre-oxidation on the membrane fouling of simulated natural water. At the end of three cycles of direct filtration, the specific flux of ultrafiltration membrane decreased to

ClO 2 Pre-Oxidation for NOM Fouling
In actual water treatment, humus, proteins, and polysaccharides generally exist together. Therefore, in order to further explore the influence of ClO 2 as a pre-oxidant on membrane fouling caused by compound foulants, raw water mixed with humic acid, sodium alginate, and bovine serum protein was prepared. Figure 4 shows the influence of ClO 2 pre-oxidation on the membrane fouling of simulated natural water. At the end of three cycles of direct filtration, the specific flux of ultrafiltration membrane decreased to 0.11, 0.10, and 0.09, respectively. As can be seen from the figure, when the dose was 1 mg/L, ClO 2 pre-oxidation had the most obvious effect on improving flux compared with other doses, which was more prominent in the second and third filtration cycles. Unlike the treatment of model foulants, the membrane flux could be effectively recovered by hydraulic flushing. As shown in Figure 4b, similar to the treatment of raw water with individual forms of organic matter, the mitigation of membrane fouling was not improved with the increase in oxidizer dosage. On the contrary, low doses (1-4 mg/L) could effectively alleviate both reversible and irreversible resistance. These results showed that low ClO2 pre-oxidation could not only improve the specific flux of mixed simulated natural water, but also reduce the membrane fouling-especially the irreversible fouling [41].   Unlike the treatment of model foulants, the membrane flux could be effectively recovered by hydraulic flushing. As shown in Figure 4b, similar to the treatment of raw water with individual forms of organic matter, the mitigation of membrane fouling was not improved with the increase in oxidizer dosage. On the contrary, low doses (1-4 mg/L) could effectively alleviate both reversible and irreversible resistance. These results showed that low ClO 2 pre-oxidation could not only improve the specific flux of mixed simulated natural water, but also reduce the membrane fouling-especially the irreversible fouling [41]. Figure 5a shows the influence of pre-oxidation of ClO 2 at different dosages on DOC in raw water. As shown in the figure, with the increase in ClO 2 dosage (0.5-8 mg/L), the influent DOC values were 3.98 mg/L, 3.97 mg/L, 3.96 mg/L, 3.90 mg/L, and 3.51 mg/L, respectively. It was found that the dosing of ClO 2 caused a very limited improvement in DOC rejection, which is consistent with the results of previous studies on ozone degradation of organic foulants [41]. Compared with DOC removal, the UV 254 value changed more obviously, which was because unsaturated bonds and aromatic rings preferentially reacted with oxidants and free radicals in water (Figure 5b) [42]. As shown in the figure, with the increase of ClO 2 dosage, the UV 254 value of raw water decreased from 0.155 to 0.129, and the removal rate was 16.70%. When the dosage range of ClO 2 was between 2 and 8 mg/L, the absorbance remained virtually unchanged. In general, after adding ClO 2 , the removal rate of UV 254 by the ultrafiltration membrane slightly decreased. When compared with previous work, it can be confirmed that ClO 2 oxidation has a better removal effect than chlorine oxidation in terms of DOC and UV 254 [43]. The humic and tryptophan compounds in the target contaminant have fluorescent excitation properties. Therefore, these two substances in the solution can be determined using the EEM. Three-dimensional fluorescence spectra of effluents under different treatment conditions are shown in Figure 6. The three-dimensional fluorescence spectrograms have three distinct characteristic peaks: T, A, and C. The characteristic peaks of humuslike organics are generally considered as A and C, which mainly appear in the range of excitation wavelength 250-350 nm and emission wavelength 350-500 nm, while the characteristic peaks of tryptophan protein (T) mainly appear in the range of excitation wavelength 225-280 nm and emission wavelength 310-340 nm [44]. As shown in Figure 6a, humic acid raw water had an obvious characteristic peak A at Ex/Em = 250-400/380-540 nm, representing humus-like organics. When the dosage of ClO2 was 2 mg/L, the fluorescence value of humus-like organics decreased. With the increase of ClO2 dosage (4-8 mg/L), the fluorescence response value of peak A increased appropriately, and reached a peak value at the dosage of 4 mg/L. This study has shown that some functional groups that can stimulate fluorescence are hidden in humus water samples, and these hidden functional groups can be activated under the action of oxidants, resulting in the increase in the fluorescence values of humic The humic and tryptophan compounds in the target contaminant have fluorescent excitation properties. Therefore, these two substances in the solution can be determined using the EEM. Three-dimensional fluorescence spectra of effluents under different treatment conditions are shown in Figure 6. The three-dimensional fluorescence spectrograms have three distinct characteristic peaks: T, A, and C. The characteristic peaks of humuslike organics are generally considered as A and C, which mainly appear in the range of excitation wavelength 250-350 nm and emission wavelength 350-500 nm, while the characteristic peaks of tryptophan protein (T) mainly appear in the range of excitation wavelength 225-280 nm and emission wavelength 310-340 nm [44]. As shown in Figure 6a, humic acid raw water had an obvious characteristic peak A at Ex/Em = 250-400/380-540 nm, representing humus-like organics. When the dosage of ClO 2 was 2 mg/L, the fluorescence value of humus-like organics decreased. With the increase of ClO 2 dosage (4-8 mg/L), the fluorescence response value of peak A increased appropriately, and reached a peak value at the dosage of 4 mg/L.  Figure 7 shows the effects of the ClO2 ultrafiltration process on fluorescent substances of simulated natural water. As with the results of humic acid, the addition of ClO2 reduced the intensity of peak A and produced an excitation effect on the fluorescent substances concealed in raw water. When the ClO2 dose was 8 mg/L, peak A, representing humic This study has shown that some functional groups that can stimulate fluorescence are hidden in humus water samples, and these hidden functional groups can be activated under the action of oxidants, resulting in the increase in the fluorescence values of humic analogs [45]. As can be seen from Figure 6b,d,f,h, peak A intensity weakened after filtration by the PES ultrafiltration membrane, indicating that the ultrafiltration membrane has certain interception and adsorption capacity for small-molecule humus-like organics that can stimulate fluorescence [46]. Figure 7 shows the effects of the ClO 2 ultrafiltration process on fluorescent substances of simulated natural water. As with the results of humic acid, the addition of ClO 2 reduced the intensity of peak A and produced an excitation effect on the fluorescent substances concealed in raw water. When the ClO 2 dose was 8 mg/L, peak A, representing humic substances, was significantly reduced, and the fluorescence value of peak A after UF was also lower than the other conditions, corresponding to the result of UV 254 , indicating that the ClO 2 preferentially attacked unsaturated bonds and aromatic ring substances in humic acids [42].

Effect of ClO 2 Pre-Oxidation on Interface Characteristics and Fouling Mechanisms
The surface of the ultrafiltration membrane was characterized by scanning electron microscope (SEM), and the mechanism of ultrafiltration membrane fouling caused by ClO 2 was further analyzed by fitting the interfacial free energy and membrane fouling model. As shown in Figure 8a, the surface of the original ultrafiltration membrane was clean and smooth, with no accumulation of foulant. However, after filtration of humic acid and simulated natural water, the surface morphology of the membrane was obviously different due to the accumulation of foulants [38]. The membrane surfaces after filtration of HA and simulated natural water are shown in Figure 8b,e, respectively, where we can observe obvious particle deposition on the membrane surfaces and a relatively dense fouling layer/cake layer. Figure 8c,d show the SEM images of membrane surfaces having filtered pre-oxidized humic acid solution with ClO 2 dosages of 2 mg/L and 8 mg/L, respectively. It can be seen that compared with Figure 8b, the membrane surface of the two pretreatment conditions was relatively smooth, and the surface foulant particles were also reduced in quantity.
As shown in Figure 8f, when the dosage of ClO 2 was 2 mg/L, the foulants on the membrane surface had a boundary outline except for a few places, and the rest of the membrane surface was relatively smooth. When the ClO 2 dosage was 8 mg/L, the membrane surface showed a large accumulation of foulants; this may have been due to the agglomeration of organic substances in raw water after pre-oxidation, which reduced the membrane water capacity and increased the membrane resistance, which was not conducive to the stable operation of the membrane.
The XDLVO theory was used for interface analysis, and the condensation free energy and adhesion free energy of the ultrafiltration membrane surface with various ClO 2 doses were calculated. The results are shown in Tables 1 and 2, in which ∆G LW , ∆G AB , and ∆G EL are the van der Waals force condensation/adhesion free energy, polar force condensation/adhesion free energy, and electrostatic condensation/adhesion free energy, respectively. In addition, ∆G 131 represents the condensation free energy between the interfaces, ∆G 132 represents the adhesion free energy of the interface, and their values are calculated through the various interface forces. "∆G > 0" indicates that the material interface is in a stable state, which is repulsive; "∆G < 0" indicates that the material interface is in an unstable state, and is attractive [47]. As shown in Table 1, ∆G LW and ∆G EL slightly changed with the increase in ClO 2 dosage, indicating that polar force dominated. In addition, the ∆G 131 was −5.36 mJ/m 2 , ∆G AB was −1.63 mJ/m 2 , and the value of ∆G AB increased to 14.10 mJ/m 2 with the increase in the ClO 2 dose (from 2 mg/L to 8 mg/L), indicating that the surface polarity increased and the hydrophilicity was improved. However, when ClO 2 was added, the adhesion free energy of ∆G 132 between humic acid and the UF membrane increased from −8.28 to 8.91, indicating that the attraction between humic acid and the UF membrane changed to repulsion. substances, was significantly reduced, and the fluorescence value of peak A after UF was also lower than the other conditions, corresponding to the result of UV254, indicating that the ClO2 preferentially attacked unsaturated bonds and aromatic ring substances in humic acids [42].  As shown in Figure 8f, when the dosage of ClO2 was 2 mg/L, the foulants on the membrane surface had a boundary outline except for a few places, and the rest of the membrane surface was relatively smooth. When the ClO2 dosage was 8 mg/L, the membrane surface showed a large accumulation of foulants; this may have been due to the agglomeration of organic substances in raw water after pre-oxidation, which reduced the membrane water capacity and increased the membrane resistance, which was not conducive to the stable operation of the membrane. The XDLVO theory was used for interface analysis, and the condensation free energy and adhesion free energy of the ultrafiltration membrane surface with various ClO2 doses  As shown in Table 2, the ∆G EL and ∆G LW between the ultrafiltration membrane and the simulated natural water did not change much with the dosage of ClO 2 , which was still dominated by polar force. When the ClO 2 dosage was 8 mg/L, unlike with humic acid, the value of ∆G 131 was −13.16 mJ/m 2 , indicating that the foulants were in a state of mutual attraction. At the same time, the value of ∆G 132 was −0.07 mJ/m 2 , indicating that the adhesion free energy between the foulants and the membrane was small and attractive, facilitating deposition on the membrane surface. This confirmed that the obvious foulant profile of the simulated natural water on the membrane surface in the SEM figure was the result of mutual attraction between the mechanical mixture and the ultrafiltration membrane, and was consistent with the obvious attenuation of membrane flux and serious membrane fouling.
In order to further investigate the effect of ClO 2 pretreatment on the membrane fouling mechanism, the membrane fouling model based on d 2 t/dV 2 and dT/dV (Ho and ZyD-Ney, 2000) was adopted, where the t and V represent the membrane filtration time and the total membrane filtration volume, respectively. As can be seen from Figure 9, the values of "D 2 ", "T/D", "V 2 ", and "DT/dV" decreased significantly after ClO 2 pre-oxidation. However, the value of "DT/dV" increased during the treatment of BSA, indicating that the filtration time increased during ultrafiltration. It was confirmed that severe membrane fouling was produced during the pretreatment of BSA with ClO 2 , which is consistent with its flux curve. As shown in Figure 9a, the n value of HA without pre-oxidation was 0.40 at the initial stage of filtration, and −1.02 at the later stage of filtration. At this time, the fouling mechanism of the membrane surface caused by the blocking of membrane holes changed [48], which is consistent with Wang's results of UV/chlorine treatment of humic acid [43]. It could be concluded that the membrane fouling mechanism was mainly a result of membrane hole blockage and the cake layer. Figure 9b shows the fitting curve of HA after preoxidation with 2 mg/L ClO2; it can be seen that the fitting curve is similar to that in Figure  9a, indicating that the fouling mechanism was consistent with and without ClO2 pre-oxidation. When SA was pre-oxidized at 2 mg/L ClO2, the membrane fouling mechanism was not changed, although the fouling was alleviated. It is noteworthy that after oxidizing simulated natural water with 2 mg/L ClO2, the fouling mechanism changed from cake layer fouling to critical blocking, which alleviated membrane fouling.  As shown in Figure 9a, the n value of HA without pre-oxidation was 0.40 at the initial stage of filtration, and −1.02 at the later stage of filtration. At this time, the fouling mechanism of the membrane surface caused by the blocking of membrane holes changed [48], which is consistent with Wang's results of UV/chlorine treatment of humic acid [43]. It could be concluded that the membrane fouling mechanism was mainly a result of membrane hole blockage and the cake layer. Figure 9b shows the fitting curve of HA after pre-oxidation with 2 mg/L ClO 2 ; it can be seen that the fitting curve is similar to that in Figure 9a, indicating that the fouling mechanism was consistent with and without ClO 2 pre-oxidation. When SA was pre-oxidized at 2 mg/L ClO 2 , the membrane fouling mechanism was not changed, although the fouling was alleviated. It is noteworthy that after oxidizing simulated natural water with 2 mg/L ClO 2 , the fouling mechanism changed from cake layer fouling to critical blocking, which alleviated membrane fouling.

Conclusions
In this work, the effects of ClO 2 pre-oxidation on the treatment of various typical organic foulants-and their mixture simulating natural water-with an ultrafiltration membrane were investigated, and the fouling mechanism was analyzed by membrane surface morphology, interfacial free energy, and membrane fouling models. It was found that this method was not only low-cost and easy to operate, but also could effectively alleviate membrane fouling caused by simulated natural water in the appropriate dosage range. The main conclusions are as follows: ClO 2 pre-oxidation at a lower concentration (1-2 mg/L) can alleviate the decrease in membrane flux caused by humus, polysaccharides, and mixed organic foulants, among which the highest traffic recovery rate reached 73%, 89%, and 86%, respectively, although its effect in terms of reducing membrane irreversible resistance was limited. In addition, the experimental results showed that ClO 2 pre-oxidation faced some difficulty in alleviating the membrane flux decrease caused by protein substances. Furthermore, the removal effect of DOC in raw humic acid water by ClO 2 dosing was not ideal. ClO 2 oxidation activated the hidden functional groups in the raw water, resulting in an increase in the fluorescence value of humic analogs and a good removal effect on the fluorescence intensity of BSA. In general, the dosage of ClO 2 was inversely proportional to the mitigation effect of membrane fouling, as well as the irreversible fouling, in the treatment of simulated natural water.
The interfacial free energy analysis showed that the attraction between organics increased, and they easily deposited onto the membrane surface, resulting in a serious flux decline and membrane resistance increase when the dosage was 8 mg/L. However, at 1 mg/L pre-oxidation, the polarity force between the membrane and organic matter could be improved, thus increasing the repulsion force and, in turn, alleviating membrane fouling.
The fitting results of the membrane fouling model showed that ClO 2 could not effectively change the fouling mechanism caused by typical organic matter in water, and that the pre-oxidation of BSA could accelerate the formation of the cake layer, thus prolonging the membrane filtration duration. However, at a low concentration (1 mg/L), the membrane fouling mechanism induced by simulated natural water could change from standard blocking and cake layer blocking to critical blocking.