Performance evaluation of emerging block copolymer membranes for oil-water separation

Membrane processes such as microfiltration (MF) and ultrafiltration (UF) are known to be the best advanced technologies for water reuse application. Numerous research efforts have been conducted in areas of modifying commercial MF/UF products or synthesizing novel materials promising enhanced oil-water separation performances. Block copolymer (BCP)-based membranes have recently gained increased popularity due to their improved water permeabilities. This study applies a comprehensive testing protocol for performance evaluation of two emerging poly (styrene-block-methyl methacrylate) BCP membranes developed by the project team. Tests mimicking industrial conditions were conducted by using a representative synthetic produced water and operating repeat tests. Both BCP membranes (referred to as A & B) were found to possess high permeabilities of 5538 and 12,424 LMH/bar, respectively. Membrane B showed higher organic rejection at 79% against 74% rejection obtained for membrane A. The novel membranes were then compared to a relevant commercial product. Lower permeability at 3831 LMH/bar and slightly higher rejection performance (within ~ 10%) were obtained for the commercial membrane as compared to the BCP membranes. Test results obtained for those novel membranes being still in the development stage will be utilized in future studies investigating further optimization of the membrane structure and oil-water separation performance.


Introduction
The oil & gas (O&G) industry produces large volumes of oily wastewaters that are typically treated by conventional technologies such as flotation, centrifugation, and gravity separation to remove free and dispersed forms of the total oil and grease (TOG). Nevertheless, the pressure on applying advanced treatment technologies has increased recently due to the stringent environmental regulations restricting the deep-well injection volumes and the discharge limits of oil-inwater. With such drivers toward water recycling, various oil and gas companies have started seeking for opportunities of treating and reusing generated byproduct waters [1]. One of the best-known advanced treatment technologies applied for reuse application are membrane processes including microfiltration (MF) and ultrafiltration (UF). Key advantages associated with applying polymeric MF/UF include their low energy consumption, compactness, and high removal of organics. The main performance limitation for such processes is fouling which reduces their respective life expectancy [2][3][4].
There are several MF/UF membrane chemistries being applied for the treatment of oily wastewaters such as Polyacrylonitrile (PAN), polyvinylidene difluoride (PVDF), polyether sulfone (PES), and polysulfone, etc. [5]. Various studies have proven the effectiveness of polymeric membranes in treating oily wastewaters in terms of their high oil removal efficiency, low fouling propensity, and recyclability. For example, an oil removal of > 99.7% was achieved by a tubular MF membrane with PAN chemistry. The flux was found to be restored after applying membrane cleaning using citric acid and caustic detergent [6]. Another example is the use of a hydrophilic UF membrane having PAN chemistry and a molecular weight cut off (MWCO) of 20 kDa at bench level for the treatment oily wastewater from refinery. TOG removal of 99% was obtained [7]. Numerous research studies have focused on the modification of commercial MF/UF products or synthetizing novel alternatives to enhance oil-water separation performances. The surface properties of commercial membranes can be enhanced by adding nanomaterials such as carbon nanotubes, mesoporous silica, halloysite nanotubes and metal oxide nanoparticles (zinc oxide, silver oxide, cobalt oxide, iron oxide etc.) to increase the membrane hydrophilicity and antifouling performances [8][9][10][11][12][13][14]. Recently, a specific type of emerging UF membranes consisting of block copolymer (BCP) films gained a wide attention due to their improved water fluxes [8,[15][16][17][18][19][20][21][22]. Such membranes are prepared based on a nano-reinforced polymers and etched BCP. The process involves application of annealing process followed by etching with ultraviolet (UV) resulting in vertically oriented cylindrical BCP embedded on a membrane substrate. Depending on the application, the polymers such as polymethylmethacrylate (PMMA), vinyl pyridine (PVP), dimethylacrylamide (DMA), polydimethylsiloxane (PDMS) and styrene-isoprene-styrene (SIS) can be selected as candidate BCP coated to the membrane substrate (polystyrene, polysulfone, polvinylidene fluoride) to form a UF membrane [22]. These membranes possess high density arrays of well-defined nanopores which leads to high selectivity, hydraulic permeability and high throughput in comparison to commercial membranes [23][24][25]. Another significant advantage of a BCP membrane is the ability to manipulate the nanochannel for required pore size (10-100 nm) and ultra-high pore density (~ 10 10 /m 2 ). Studies have also reported the mixing of the BCP with ionic liquids (IL) [26,27]. They are added to the BCP to increase the width of the cylindrical nanopore array and optimize the pore size and thermal stability of the BCP membrane [26]. Ionic liquids are molten salts possessing a melting point < 100 °C. They have unique features such as low vapor pressure, heat stability, high ion conductivity, and non-combustibility. Few common ionic liquids are 1-butyl-3-methyl imidazolium tetrafluoroborate and 1-butyl-3-methylimidazolium hexafluorophosphate [28].

Graphical Abstract
In our earlier study [29], a comprehensive MF/UF evaluation protocol was developed, filling in the gap of having a robust bench testing procedure used to assess the performance of commercial and emerging MF/UF membranes while mimicking industrial conditions. Such procedure incorporated using a representative synthetic produced water (PW) to evaluate the membrane performance in terms of permeability, fouling, rejection, and cleanability. The validity of the procedure was proven using various commercial products through assessing the impact of changing membrane pore size/MWCO, chemistry, and vendor on performance. The procedure was also successfully applied on some novel membranes at their initial development stage to be evaluated and compared to relevant commercial products. This study builds on the previous evaluation and focuses on applying the same thorough procedure to assess multiple newly developed BCPbased membranes.

Objectives
This paper presents an evaluation of newly developed BCP-based membranes at bench scale level targeting industrial oil water separation application. The performance of those new materials was then compared to a relevant commercial product under representative industrial testing conditions.

Commercial membranes
To be able to illustrate the variation in performance trends for the newly developed membranes that are expected to show variation in permeability depending on the alignment of the BCP vertical cylinders (i.e. pores), multiple commercial membranes were tested to determine the impact of pore size variation on membrane performance. Table 1 lists the selected products. Three commercial membranes with PES chemistry at multiple pore sizes were acquired from M1. PES chemistry was selected as a comparison basis because the synthesized BCP films were layered on a commercial PES support. Based on the obtained performance findings for the newly developed membranes, a relevant commercial membrane, having PES chemistry, was also tested and used as benchmark for performance comparison.

Emerging membranes
The emerging membranes were developed by our partners from University of Houston and were shipped to our laboratories for performance testing [22]. The first generation of the BCP-based membranes were initially prepared using a poly (styrene-block-methyl methacrylate) (PS-b-PMMA) BCP as the active film (~ 100 nm) layered on a commercial 8.0 μm PES support. Table 2 lists the second enhanced generation of such membranes involved using a similar BCP (PS-b-PMMA). The subscripts 55 k and 22 k represent the molecular weights of the individual blocks PS and PMMA, respectively. Adjustments in the second generation included the mixing of the BCP with an ionic liquid. Two membranes with varying IL concentrations labeled as A and B were prepared and tested. The single PES layer used for the first generation was not able to perform as expected when tested under applied pressure. Therefore, an approach of using a dual 8.0 μm PES support sandwiching the active film was applied. Details of the BCP film preparation were presented in earlier studies [22,27].

Bench scale setup
As shown in Fig. 1, evaluation tests were conducted using a custom-built bench scale unit, incorporating the learning from previous publications [14,24,25,27]. The unit consists of a 50 mL Amicon stirred cell (Millipore, USA) with an active membrane area of 13.5 cm 2 . The cell is capable of fitting flat sheet membranes with a diameter of 44.5 mm, and it is placed on top of a stirred plate to ensure continuous mixing of the feed solution during filtration. The set-up has a continuous feed flow from a 1-gal pressurized reservoir (Sterlitech, USA), and it is capable of operating at pressures ranging from 1 to 5 bars, adjusted using a pressure regulator connected to a nitrogen gas cylinder. If the applied pressured exceeds 5 bars, a pressure relief valve will trigger to safeguard the Amicon cell. A pressure transducer (Omega Engineering, USA) was installed at the outlet of the feed reservoir (just before entering the cell) to monitor the inlet pressure. The permeate from the cell is collected in a 1 L glass beaker located on top of an analytical balance that is used to calculate the water flux based on the weight difference over time. All operating conditions and process parameters, including pressure, flux, weight readings are displayed in the user interface and logged by the acquisition system (NI, USA).

Chemicals and analytical methods
Deionized water (DI) at a resistivity of ≈18 MΩ-cm was supplied via a Milli-Q ultrapure water system (integral, 10, Millipore). Several salts were acquired from Sigma Aldrich to prepare the brine for the synthetic PW including sodium chloride, calcium chloride dihydrate, magnesium chloride hexahydrate, potassium chloride, sodium sulfate, ammonium chloride, and sodium bicarbonate all at 99% purity. Crude oil from O&G operation at °API of 38.7 and density @ 25 °C of 0.825 g/ml was used as the oil source in the synthetic PW. Oil-in-water emulsions were formed using sodium dodecyl sulfate (SDS) that is obtained from Thermo Fischer Scientific. To be used for membrane chemical cleaning, sodium hydroxide pellets -NaOH were acquired from Thermo Fischer Scientific.
In terms of analytical methods, removal of organics was measured through total organic carbon (TOC) measurements using TOC-V, Shimadzu.

Synthetic PW solution preparation
Evaluation tests were conducted using a synthetic PW solution mimicking the characteristics and behavior of real PW as reported by [30]. This representative recipe includes using a brine of low salinity (~ 3800 mg/L of total dissolved solids (TDS)), a medium grade crude oil, and a surfactant to form the oil-in water emulsions. A ratio of 5:1 (oil-to surfactant) was used targeting tertiary treatment application. To prepare the test solution, 500 mL of brine was prepared. Then SDS was added at mass of ~ 5 mg followed by the crude oil at volume of 0.03 ml. Magnetic stirring was applied on the mixture for 30 minutes followed by sonication for another 30 minutes. The solution was poured into a glass separatory funnel to settle for 1-2 hours. The aqueous fraction was separated from the excess free oil and the final solution was used in testing. Concentrations of the salts used to prepare the bine as well as the particle size distribution of the synthetic PW being at an average of ~ 4.6 μm were reported earlier [29]. Table 2 summarizes the water quality of the synthetic PW solution.

MF/UF experimental protocol
The previously developed procedure [29] was applied to evaluate membrane performance in terms of permeability, fouling, rejection and cleanability. The protocol consists of the below steps:

Characterization
After soaking the membrane in DI water for around 24 hours, the membrane sample was cut into the proper size and placed in the Amicon cell (diameter: 44.5 cm). The characterization test was performed using DI water, 3 bar pressure, and magnetic stirring speed of 560 rpm until obtaining a stable flux measurement. Comparable conditions were applied for all baseline tests. Flux acquired in this test is described as the "clean membrane flux" which was later used as the benchmark for comparison with the fouled membrane.

Fouling & rejection
The synthetic PW was used to perform the fouling tests targeting 50% PW volume reduction. The test was performed by using 100 mL and 50 mL of the synthetic PW transferred to the feed reservoir and the Amicon cell, respectively. After that, the membrane was tested at 3 bar pressure under stirring at 560 rpm. The flux resulted from this test is referred to as the "synthetic PW flux". To assess organics rejection, feed and permeate samples were analyzed for TOC.

Cleaning and recovery
The "fouled membrane flux" was then measured by performing a DI water baseline. After that, two chemical cleaning steps were applied to recover lost flux. Caustic cleaning using 50 mL of a ~ 1 mM NaOH and pH of ~ 11.5 at ~ 25 °C. Surface cleaning at 560 rpm for 15 minutes was performed. The efficiency of NaOH cleaning was measured by performing a DI water baseline as noted in section 3.4.1. SDS surface cleaning was performed after that using a 50 mL ~ 10 mM solution and pH of ~ 9.4 at 35 °C. Another baseline was performed to measure the "chemical cleaning flux" and determine the total flux loss attributed to fouling. Equation 1 was used to calculate the total membrane flux loss. Where J o is the "clean membrane" flux and J f is the final membrane flux after "chemical cleaning" in LMH/ bar. After characterization, three repeat cycles of fouling and cleaning were performed on the membrane. This approach is considered more robust for measuring the oil-water separation efficiency. Figure 2 shows a graphical representation of the testing protocol.

Results and discussion
The newly developed BCP membranes in the current study were synthesized using a PS-b-PMMA BCP mixed with varying concentrations of IL. Playing with those parameters resulted in membranes having different size distributions of the vertical cylinders thus impacting the permeability of the membrane and consequently its fouling propensity, rejection, and cleanability. To be able to interpret the effect of changing pore size on membrane performance, results obtained from studying such variation using commercial membranes of similar PES chemistry were employed.

Performance of commercial membranes
The impact of pore size on membrane performance was assessed using by testing PES membranes of 0.2 and 8.0 μm membranes all manufactured by M1. The initial characterization test results are shown in Fig. 3, both membranes showed stable fluxes when tested under pressure. Results obtained from the three test cycles are presented in Fig. 4. Higher water fluxes were obtained at increased membrane pore size, yet such increase was found to be associated with decreased organics rejection. The 0.2 μm resulted in average rejection of 87% against 67% obtained for the 8.0 μm membrane as indicated in Table 3. The lower rejection obtained by the 8.0 μm is attributed to the pore size of the oil-water emulsions in the synthetic feed water measured at average droplet size of 4.6 μm. Smaller oil emulsions were not rejected by the 8.0 μm membrane resulting in its lower TOC rejection. The 0.2 μm was able to reject those smaller particles which resulted in increased fouling tendency and reduced cleaning efficiency. A total flux loss of 26% was measured for the 0.2 μm membrane whereas the 8.0 μm showed less fouling tendency and negligible membrane   flux loss. Another commercial PES membrane at MWCO of 500 kDa was also tested using the comprehensive testing procedure as shown in Figs. 5 and 6. The membrane showed stable water flux measurements throughout the characterization test at average specific flux of 3831 LMH/bar. The repeat test cycles resulted in average TOC rejection of 87% and total flux loss of only 1% as indicated in Table 3.

Performance of emerging membranes 4.2.1 Clean membrane flux
It is important to determine the membrane initial stable flux prior to testing with the synthetic PW. The initial characterization test is utilized in assessing the machinal stability of the BCP film when tested under pressurized operating conditions. The first generation of such membranes was tested in our previous study at which it was found that the applied pressure resulted in significant drop in permeability with time due to the gradual deformation of the vertical porous BCP cylinders. To address this limitation, the BCP film was sandwiched between two layers of the 8.0 PES support. A performance comparison between the first and second membrane generations in terms of initial characterization is shown in Fig. 7. Results for the newly developed BCP membranes using the sandwich configuration showed stable performances throughout the test period signifying the added protection from the dual support layers.
Modified membranes A and B showed stable specific water fluxes of 12,242 and 5538 LMH/bar, respectively. The modified dual support design was proven effective in eliminating the BCP film deformation upon applying pressure.

Oil-water separation performance
The developed testing procedure was applied to assess the oil-water separation performance of the BCP membranes using the synthetic PW. Figure 8 and Table 4 compare the results from the three test cycles performed on membranes A and B. Membrane A showed initial membrane flux of 12,424 LMH/bar and average TOC rejection of 74%. The membrane was also found to yield a total flux loss of ~ 16%. Membrane B, initially having approximately 50% less flux compared to membrane A, showed better TOC rejection at 79%. Such performance trend obtained for a lower membrane permeability showing better TOC rejection was found consistent with commercial products (section 4.1). Although membrane B showed higher rejection than membrane A, the membrane showed negligible flux loss from the three operated test cycles. As proven previously [29], the fouling behavior for UF membranes is governed by the dominant occurring fouling mechanism. The BCP film improved the rejection of the commercial support being initially at 69% up to 79% for membrane B. Upon screening several commercially available products, a relevant PES membrane at MWCO of 500 kDa was selected for performance comparison against membranes A and B. The novel membranes showed higher water fluxes against the commercial membrane possessing initially a flux of 3831 LMH/bar. Although both emerging membranes had initially higher fluxes, they were able to reject ~ 74% to 79% of organics present in the feed water. This was not very far off from the organic rejection obtained for the commercial membrane at 87% while possessing a lower permeability. Based on that, the decision on membrane selection will be dependent of the target application. For instance, the novel products specifically membrane B can be applied in cases where high permeability, minimum membrane fouling or cleaning frequency, and moderate rejection performances are required. Future optimization is underway for improving the permeability and organics rejection of the novel BCP membranes.

Conclusions
This paper applies a comprehensive testing protocol to evaluate the performance of newly developed BCP-based UF membranes targeting oil-water separation application. The procedure mimics industrial testing conditions through using a representative synthetic PW recipe and operating repeat testing cycles. Two BCP-based membranes referred to as A and B were tested using the robust procedure. The early generation of such membranes, prepared using a single support layer, showed limitation in the measurement of the stable flux when operating under pressure. Hence, the modified version included sandwiching the active film between two support layers. The initial characterization test comparing the previous and the new membrane designs confirmed the added protection from the dual support layers by achieving stable water flux measurements under pressurized conditions.  Membrane B having a measured flux of 5538 LMH/bar showed better organic rejection at 79% when compared to the higher permeability membrane A (12,424 LMH/ bar). Such performance was proven to be consistent with results obtained for commercial products. Variation in fouling tendency and flux loss were both verified to be governed by the membrane fouling mechanism. Test results revealed that membrane B showed minimum flux loss in comparison to membrane A being at 16%. A relevant commercial membrane was also selected and tested for comparison with the novel membranes. The commercial product showed lower permeability compared to the novel membranes which was associated with slightly higher rejection performance (within ~ 10%). With all performance parameters considered, membrane selection will depend on the target application which will potentially involve a compromise between permeability, fouling, or rejection. Since those membranes are still in the development stage, future studies can investigate optimizing the IL concentration as well as applying surface modification to enhance the membrane organic rejection.