Antifouling Ability of Hydrophilic PVDF-TiO2 membrane Evaluated by Critical Flux and Threshold Flux

The PVDF flat-membrane was modified by hydrophilic nano-TiO2, which blending by ultrasonication and mechanical stirring pretreatment in phase inversion method. To evaluate the permeate productivity and anti-fouling capacity of protein solution, both the critical flux (JCW) and threshold flux (JTH) of PVDF and PVDF-TiO2 membrane were firstly measured by Advanced Constant Pressure-step Method in cross-flow filtration apparatus. Some evaluation indicators were utilized to analyze the results, such as Flux vs. Time and TMP vs. Time Curves, flux decline rate (dFlux/dt) and TMP-Fluxave curve. Two type fluxes were compared, results exhibited that hydrophilic PVDF-TiO2 modified membrane possessed a higher level of both Jcw and JTH and better anti-protein fouling ability after testing by Advanced Constant Pressure-step Method.


Introduction
Membrane separation technology was widely used in both water and wastewater treatment, especially for pressure-driven membranes which consisted of four types, porous micro-filtration (MF) or ultra-filtration membranes (UF), or dense nano-filtration (NF) or reverse osmosis (RO) membrane [1]. For porous MF and UF, they were gradually accepted in practical application, due to low-pressure and energy consumption. During practical filtration process, reaction of membrane and contaminants occurred by physical, chemical or biological effects, easily leading to fouling problems limited the application scope. Fouling problem of UF had become one research hotspot. There were three ways to solve the aforementioned problems, treatment of feeds [2], hydrophilic modification of membrane [3], and operation control by critical flux (CF) and threshold flux (TF) [4]. The first route was achieved by extra auxiliary methods, and the latter two routes did not require additional cost investments.
In last decades, various novel hydrophilic additives were attractive for UF membrane fabrication, owing to some advantages, such as rich hydrophilic functional groups and brilliant modification performance [5]. As compared to carbon nanotube [6] and graphene oxide [7], inorganic metal-based oxide was seemingly more available by commercial purchased or self-preparation. Except for Al2O3 [8], Ag [9], SiO2 [10] and so forth, nano TiO2 had attracted lots attention because of high hydrophilicity of less toxicity which commonly used in UF modification [11,12].
For operation control, CF or TF operation was often realized by no or low fouling rate, to obtain a less cleaning frequency and longer membrane life in practical significance [13]. After the CF concept proposed in 1995 [14], some researchers found that ideal state of CF, no fouling, might be in very low flux level or did not exist, especially the composition of feeds was complicated [15]. Then, a mild concept of low fouling rate operation named TF, was put forward in 2011 [4]. References investigation showed four major research directions of CF/TF operation, the existence [16], new measuring methods [17], the practical significance [18], and the relationship of operation and fouling [19].
Regarding to hydrophilic modified UF membrane, particularly of nano-TiO2 modification, some research gaps put in an appearance. Did the ideal CF operation of modified membrane exist or not. And did the modified membrane had a higher CF and TF level to promote productivity, when it owned a better anti-fouling ability. The purpose of this study was to give better understanding on the anti-fouling capacity of nano-TiO2 modified UF membrane, from CF and TF perspective. For this, one modified membrane was fabricated by PVDF polymer and nano-TiO2 particle with assist of novel promote pretreatment. The CF and TF values of PVDF and modified PVDF-TiO2 membranes were firstly measured by Advanced Constant Pressure-step Method. Fouling behavior of two PVDF membranes were analyzed by fouling resistance.

Fabrication of pristine and modified membranes
Both pristine PVDF and PVDF-TiO2 membrane were fabricated by phase inversion method which clearly described in our previous study [20]. Pristine PVDF membrane was made with weight ratio of PVDF : PVP : DMAc =19:1.3:79.7, and PVDF-TiO2 membrane with weight ratio of nano-TiO2 : PVDF : PVP : DMAc=1:19:1.3:78.7. The nano-TiO2 was added into DMAc solvent with direct ultrasonication pre-treatment for 1 h, the mixture was kept for 25℃ by constant temperature circulating water tank. The direct ultrasonication was realized by Cell disruption instrument which named new model Ultrasonic Processor (FS-450, China).
Then, PVDF polymer and PVP were added into mixture for mechanical stirring (24 h) and statically placed in water bath at 55℃ (24 h) to finish degassing stage. Certain amounts of casting solution was poured onto clear glass and then casted by thin-film casting doctor blade (~200 μm). Certain volume of deionized water was put as coagulation bath to complete the phase inversion process. The pristine membrane was marked as PVDF, and the modified membrane was marked as PVDF-TiO2(MSU2).

Filtration set-up and determination procedures
One cross-flow filtration set up was used, three sensors were put near influent and effluent position to measure real-time pressure. The feeds was protein solution with bovine serum albumin concentration of 30mg/L, and cross-flow velocity was 15 cm/s. The bovine serum albumin was measured by ultraviolet radiation spectrophotometry at a λ of 280 nm. The effective area of membrane cell was 42 cm 2 , and membrane was compacted at 2 bar for 20 min before CF and TF testing.
Based on the strict concept of CF, Field [14] proposed two types, one was the strong form (JSW) corresponding to feed solution were pure water, another one was the weak form (JCW) corresponding to feeds consisted of water and foulants. For the concern of antifouling ability evaluation, only JCW was measured in our study, which tested by the Advanced Constant Pressure-step Method. This method [21] applied the TMP up to a maximum TMP and back analogous to the initial TMP step ( Fig. 1(a)). The upping TMP step named as X, and the initial TMP step named as 1-X. The operating order was step 1 → step 2 → step 1-2 → step 3 → step 1-3→…. For instance, fouling occurred at step 4 and permeate flux decline was observed here, revealing that the critical flux was situated between step 3 and step 4. At same time, the permeated flux wouldn't be same at step 1-3 and step 1-4. It meant the critical TMP should be below step 4, and the critical flux should be measured at the corresponding critical TMP. If no fouling occurred between step 3 and step 4, the permeated flux of those two steps shouldn't be decreased with filtration time and it also should be same at step 1-3 and step 1-4. The evaluation indicators of the Advanced Constant Pressure-step Method was shown in Fig. 1(b). The ideal CF was corresponding to no fouling, thus the permeate flux should keep same with no decline and dFlux/dt should be zero. The dFlux/dt was permeate flux decline rate in one pressure step, which equaled to the slope (k) of permeate flux curve (Equation (1)). The Fluxave meant the average flux level in one pressure step, which was calculated by average value collected one time per minute in Equation (2). The TF could be estimated by dFlux/dt kept for low value.
Flux(t)=k×t+b (1) Where, Flux(t) was the real-time permeate flux collected by electronic balance (L/(m 2 ·h)), and t was the filtration time (min), k was the slope of fitting linear, b was one constant.
Fluxave=∑(Fluxi n +…+Fluxf n )/15 (2) Where, Fluxi n was the initial flux at pressure step n, and Fluxf n was the final flux at pressure step n, 15 was the collected times at one pressure step.

Fouling resistance
After measuring JCW and JTH, the fouling resistance was calculated by resistance-inseries model put in Equation (3) [22]. Where, the J was permeate flux, the △P was the pressure average value in one step, the μ was permeate viscosity of water at 25 ℃, and Rt was the total fouling resistance. Rt was also the sum of Rm, Rc and Rf. The Rm was intrinsic resistance related to membrane structure and characterization. The Rc was the gel or cake layer resistance formed by foulants concentration polarization and precipitation, belonged to reversible resistance removed by physical cleaning that fouled membrane was flushed with tap water for 5 min of each side. The Rf was the adsorption or blockage resistance formed by foulants adsorption into membrane surface or inside pores, which belonged to the irreversible fouling. J=△P/(μ×Rt)=△P/(μ×(Rm+Rc+Rf)) 3 Results and discussion

The dFlux/dt
Additionally, the JCW could be estimated by dFluxt/dt equalled to zero, which was demonstrated in Fig. 3. It was clearly that no zero dFlux/dt of both TMP ascending phase and relaxation phase was found, which might related to instrument precision and high accuracy required for operation. Depending on the dFlux/dt≈0 reflected no fouling occurred, the JCW was estimated as following. For PVDF membrane, when TMP step raised from step 1 to step 2, the dFlux/dt decreased to less than -0.05, which showed the irreversible fouling occurred. And for PVDF-TiO2(MSU2), when step 1 raised TMP to step 2, the dFlux/dt seemed remained near zero. When step 2 decreased to step 1-2, the dFlux/dt decreased to near -0.05, which illustrated that fouling occurred between step 1 and 2, then showed at step 1-2 due to fouling hysteresis effect. Therefore, based on the result obtained by dFlux/dt, the critical TMP of PVDF and PVDF-TiO2(MSU2) was less than and equal to 15 kPa, their JCW were ＜5.43 and 13.88 L/(m 2 ·h), respectively.

The TMP-Fluxave Curves
The JCW also could been evaluated by the TMP-Fluxave curve, according to the Espinasse's studied on the JCW measured by Alternating Constant Pressure-step Method [23]. Set as an example, the first three TMP step of PVDF membrane was step 1→ step 2→ step 1-2, i) if the reversible fouling formed between step 1 and step 2 which could be removed by hydraulic scour of cross flow, the Fluxave of step 1-2 should be close enough to it at step 1, the Fluxave of first three TMP steps should be fitted in linear; ii) or if irreversible fouling formed between step 1 and step 2 which couldn't been removed by cross flow, the Fluxave of step 1-2 would be less than it of step 1, and this would not continue to be linear relationship. For PVDF, from step 1 to step 1-2, the Fluxave was 5.43, 6.66 and 5.26 L/(m 2 ·h), and the R 2 of fitting linear was 0.9915 which displayed that only reversible fouling formed between 15 and 20 kPa. When TMP raised to step 3 then back to step 1-3, the first five points of the TMP-Fluxave curve would not maintain in linear, because the R 2 decreased to 0.9681 which was less than 0.99. For PVDF-TiO2(MSU2), first three points from step 1 to step 1-2 could be in linear due to the R 2 was 1. And it same to PVDF, the linear could not keep until step 3, owning to the R 2 decreased to 0.9867. Herein, the critical TMP of PVDF and PVDF-TiO2(MSU2) were same of 20 kPa at step 2, and the JCW were 6.66 and 15.39 L/(m 2 ·h), respectively.

Threshold flux estimation
Relying on the TF concept put by Field [4], the fouling rate between low and high fouling region was the key-point. According to the dFlux/dt (Fig. 3), we found that dFlux/dt of two membranes were decreased with TMP increased, then it seemly kept for one pseudosteady state in few steps and dropped dramatically. It looked like dFlux/dt=-0.15 was one demarcation, when dFlux/dt was less than -0.15, the dFlux/dt jump appeared and started to drop dramatically. We used this dFlux/dt jump to differentiate low and high fouling region. Therefore, the threshold TMP of PVDF and PVDF-TiO2(MSU2) was 40 and 65 kPa, and the JTH were 9.08 and 25.60 L/(m 2 ·h).

Comparison of JCW and JTH of PVDF and PVDF-TiO2(MSU2)
On the basis of previous results, the summary of PVDF and PVDF-TiO2(MSU2) were listed in Table 1. When feeds was protein solution, the JCW and JTH of PVDF and PVDF-TiO2(MSU2) did exist. The critical pressure of PVDF was ＜15~20 kPa, the JCW was ＜ 5.43~6.66 L/(m 2 ·h). And for PVDF-TiO2(MSU2), the critical pressure was 15~20 kPa, but JCW was improved to 13.88~16.92 L/(m 2 ·h). The threshold pressure of PVDF and PVDF-TiO2(MSU2) was 40 and 65 kPa, respectively, and their JTH were 9.08 and 25.6 L/(m 2 ·h). The hydrophilic modified membrane had 2 or 3 times of both JCW and JTH than PVDF membrane, no matter it was evaluated by which indicators. It was inferred that adding nano-TiO2 was benefit for enhancing productivity of PVDF membrane, when operating at JCW condition. At same time, hydrophilic modified membrane was surmised brilliant antiprotein fouling ability, when operating at threshold pressure condition it had higher JTH.

Fouling mechanism
The fouling resistance of Rt, Rc and Rf was shown in Fig. 5. For PVDF-TiO2(MSU2) at JCW, the reversible fouling Rc close to 1% of the Rt, which could be neglected. However for PVDF membrane, the irreversible fouling Jf could not been ignored at JCW, and it was increased at JTH level. It was indicated that ideal status of critical flux condition was not found, but the operating results of PVDF-TiO2(MSU2) was close to it. It was probably due to the irreversible protein contamination occurred easily of hydrophobic PVDF, the ideal JCW was rather small and not detected in our study because instrument limitation. The antiprotein fouling ability of PVDF-TiO2(MSU2) was enhanced effectively when nano-TiO2 was added, the JCW level was improved and could be detected. When flux increased to JTH, the Rt, Rc and Rf of two membranes were increased with different degree, especially for PVDF. It attributed to the driven force was higher than that at JCW, the shear force provided by cross-flow was weaker than protein sedimentation effect, so that concentration polarization layer or pore blockage might formed quickly at threshold TMP. For modified PVDF-TiO2(MSU2) membrane, the increased degree of Rc was higher than that of PVDF, due to the threshold TMP was 65 kPa which higher than 30 kPa of PVDF. At same time, no matter at flux level of JCW or JTH, Rt of PVDF-TiO2(MSU2) was smaller than that of PVDF membrane, which illustrated that fouling degree was severer of PVDF. In brief, it was evidently that hydrophilic modification by nano-TiO2 was profit to alleviate protein fouling at permeate flux level of both JCW and JTH.

Conclusion
When feed solution was protein, the critical flux and threshold flux of PVDF and hydrophilic modified PVDF-TiO2(MSU2) were successfully evaluated by the Advanced Constant Pressure-step Method depending on some indicators. As compared to PVDF, the PVDF-TiO2(MSU2) owned greater value at both JCW and JTH which easily guaranteed the productivity at operating optimization. Additionally, the nano-TiO2 was helpful to alleviate the irreversible fouling degree of PVDF membrane at JCW and JTH, especially decreasing the Rf of PVDF-TiO2(MSU2) at JCW level which leading the operation close to ideal status of critical flux.