Journal Search Engine
Search Advanced Search Adode Reader(link)
Download PDF Export Citaion korean bibliography PMC previewer
ISSN : 1226-0088(Print)
ISSN : 2288-7253(Online)
Membrane Journal Vol.34 No.1 pp.38-49

Carbon Nanosphere Composite Ultrafiltration Membranes with Anti-Biofouling Properties and More Porous Structures for Wastewater Treatment Using MBRs

Jaewoo Lee*,**,***,****
*Department of Polymer-Nano Science and Technology, Jeonbuk National University, Jeonju 54896, Republic of Korea
**Department of Bionanotechnology and Bioconvergence Engineering, Jeonbuk National University, Jeonju 54896, Republic of Korea
***Department of JBNU-KIST Industry-Academia Convergence Research, Jeonbuk National University, Jeonju 54896, Republic of Korea
****Polymer Materials Fusion Research Center, Jeonbuk National University, Jeonju 54896, Republic of Korea
Corresponding author(e-mail:;
December 10, 2023 ; December 10, 2023 ; December 21, 2023


Wastewater treatment using membrane bioreactors has been extensively used to alleviate water shortage and pollution by improving the quality of the treated water discharged into the environment. However, membrane biofouling persistently holds back an MBR process by reducing the process efficiency. Herein, we synthesized carbon nanospheres (CNSs) with many hydrophilic oxygen groups and utilized them as an additive to prepare high-performance ultrafiltration (UF) membranes with hydrophilicity and porous pore structure. CNSs were found to form crescent-shaped pores on the membrane surface, increasing the mean surface pore size by about 40% without causing significant defects larger than bubble points, as the CNS content increased by 4.6 wt%. In addition, the porous pore structure of CNS composite membranes was also attributable to the CNS’s isotropic morphologies and relatively low particle number density because the aforementioned properties contributed to preventing the polymer solution viscosity from soaring with the loading of CNS. However, too porous structure compromised the mechanical properties, such that CNS2.3 was the best from a comprehensive consideration including the pore structure and mechanical properties. As a result, CNS2.3 showed not only 2 times higher water permeability than CNS0 but also 5 times longer operation duration until membrane cleaning was required.

분리막 생물반응기를 활용한 폐수처리를 위한 생물오염방지 특성 및 다공성 구조를 가진 탄소나노구체 복합 한외여과막

이 재 우*,**,***,****
*전북대학교 고분자나노공학과
**전북대학교 대학원 바이오나노융합공학과
***전북대학교 대학원 JBNU-KIST 산학연융합학과
****전북대학교 고분자소재융합연구센터


본 연구에서는 분리막 생물반응기(membrane bioreactor, MBR)에서 발생되는 생물막오염 완화에 탁월한 효과를 가진 분리막을 개발할 목적으로, 친수성 산소 기능기가 많은 탄소나노구체(carbon nanosphere, CNS)를 합성한 뒤, 이를 첨가 제로 활용하여 친수성과 다공성 기공 구조를 갖는 고성능 한외여과막을 제조하였다. CNS는 막 표면에 초승달 모양의 기공을 형성하였고, CNS 함량을 4.6 wt%까지 증가시킴에 따라 최대기공 크기보다 큰 결함을 야기하지 않으면서 평균 표면 기공 크 기를 약 40% 증가시키는 것으로 나타났다. 또한, CNS 복합막의 다공성 기공 구조는 CNS의 등방성 형태와 상대적으로 낮은 입자 수밀도 덕분에 CNS 첨가에 따른 고분자 용액의 점도 급등이 방지됐기 때문이라고 판단된다. 그러나 너무 다공성이 커 지게 되면 기계적 물성이 저하되므로, 기공구조와 기계적 성질을 포함한 종합적인 고려를 했을 때 CNS2.3이 가장 우수하다 고 관측되었다. CNS2.3은 CNS0에 비해 수투과도가 2배 이상 높을 뿐만 아니라, MBR 공정에서 분리막 세정이 요구될 때까 지의 운전 시간도 5배 이상 연장시킨 것으로 확인되었다.

    1. Introduction

    The water shortage problem has been increasingly prominent due to anthropogenic climate change and growing demographic pressure[1-3]. Thus, many people have steadily sought various water treatment processes to alleviate water deficiency. For example, thermal desalination like distillation has been intensively used to get fresh water from seawater by separating water vapor from liquid seawater including various kinds of salts in the Middle East[4-6]. However, they require high energy consumption because the process demands the phase change from liquid to gas. For that reason, membrane-based water treatment processes have attracted much attention as a more energy-efficient process than thermal desalination or other conventional water treatment processes because they do not need phase change to separate pure water from seawater or wastewater. Among several water and wastewater treatment approaches using membranes, a membrane bioreactor (MBR) process, which includes microorganisms capable of consuming organic matter in wastewater as nutrients, is one of the most widely used strategies [7-9].

    However, an MBR process has inherent weaknesses in that microorganisms reduce the process efficiency by causing membrane fouling, while it has to utilize microorganisms that are beneficial to remove organic matter in wastewater. To overcome the inherent limitation, we need to develop novel membranes that can effectively mitigate membrane fouling and thereby improve the efficiencies of MBR processes. We can begin understanding how to design novel membranes to address membrane fouling from the concentration polarization model, which is a widely held conventional wisdom. According to the concentration polarization model stemming from a mass balance for the solute, as shown in Fig. 1, the solute flux can be represented by subtracting the back-diffusive flow term from the convective flow term (Equation (1)).

    J S = C P J V = C J V D e f f d C d x

    where JS is the solute flux (g m−2 s−1), CP is the solute concentration (mg L−1) in a permeate solution, C is the solute concentration (mg L−1), Deff is the solute’s effective diffusion coefficient (m2 s−1), x is the distance from the membrane surface (m), and JV is the volumetric water flux (m s−1).

    Equation (1) can be rewritten as follows.

    D e f f d C d x = J V ( C C P )

    If we integrate Equation (2) from the membrane surface to the end of the boundary layer after transposing dx on the left side and ‘C – CP’ on the right side, we can get the following equations successively.

    C B C G D e f f d C C C P = 0 δ J V d x

    J V = D e f f δ ln C G C P C B C P

    where CG is the solute concentration on the gel layer surface, CB is the solute concentration in the bulk solution, and δ is the boundary layer thickness. If we can suppose that the CP can be negligible, Equation (4) can be simplified as follows.

    J V = D e f f δ ln C G C B

    From Equation (5), we can obtain a few useful tips about the strategy to improve membrane efficiency on the condition that the solid concentration in the gel layer does not undergo any change. First, if the temperature is high, water flux can increase because of a larger diffusion coefficient. Next, a low feed concentration can lead to a low bulk concentration, leading to high water flux. Lastly, the water flux can increase if a boundary layer becomes thinner. However, the temperature and feed concentration are usually variable depending on environmental factors like the season or wastewater source. Therefore, we have no choice but to decrease the boundary layer thickness. In an MBR process, the boundary layer thickness can be reduced by increasing an aeration intensity, which is not appropriate because increasing an aeration intensity can cause high operating costs. For this reason, we need to find out how to enhance membrane performance without increasing aeration intensity, and this is why we need to develop novel membranes for wastewater treatment.

    Furthermore, we need to consider that the mass transfer of colloidal suspensions could be driven from the membrane surface toward the bulk stream by other forces inducing back diffusive transport than the concentration gradient[12], given that individual bacteria in MBRs corrspond to colloidal suspension. Specifically, colloidal suspensions are known to go through the augmented back diffusive transport owing to drag torque, inertial lift, charge repulsion, and shear-induced diffusion[ 11], as shown in Fig. 2. This phenomenon is called ‘tubular pinch effect’, which is the collective concept related to the enhancement of the back-diffusive transport of colloidal particles[13,14]. Among the parameters in terms of the tubular pinch effect, the inertial lift and shear-induced diffusion are likely to be enhanced by adjusting process parameters such as a crossflow rate. On the other hand, the charge repulsion can be enhanced by material approaches such as membrane surface modification. For example, carbon nanomaterials possessing oxygen functional groups such as carboxyl groups can contribute to charge repulsion because the functional groups can induce charge-charge repulsion with bacteria known to have negatively charged cell membranes[15].

    In addition, hydrophilicity is also a very important property when we try to mitigate the adhesion of bacteria. Generally, bacteria or other foulants tend to irreversibly adsorb on a hydrophobic surface to minimize interfacial energy. On the other hand, hydrophilic surfaces no longer offer any significant thermodynamic advantage because of the already low interfacial energy. Therefore, a hydrophilic and negatively charged surface is more appropriate to prevent bacterial adhesion. Lastly, it is advisable to enlarge the surface pore size within reasonable bounds that do not compromise a particle removal rate to further improve membrane performance by alleviating membrane resistance. In this study, carbon nanospheres (CNSs) with hydrophilic functional groups were used to modify ultrafiltration (UF) membranes to meet the aforementioned characteristics. The characteristics of CNSs were investigated to correlate hydrophilic carbon nano additives with the modified membrane performance. Moreover, to dig deeper into the behavior of CNSs during phase separation, the modified membranes’ surface pore size and sublayer morphologies were investigated at the different amounts of CNS incorporated in the membrane. In particular, the morphologies of nano additives were correlated with the viscosity of a polymer solution and the tendency of demixing. Lastly, as an ultimate performance test, an optimal CNS composite membrane was compared with the control membrane during MBR operation.

    2. Material and Methods

    2.1. Preparation of CNSs

    Unless specified otherwise, all the chemicals and materials were used without further purification after being received from Sigma-Aldrich and Merck. CNSs were prepared via a hydrothermal method, as reported previously[16]. 40 mL of a 0.5 M glucose solution (D(+)-glucose, Junsei Chemical Co., Ltd., Japan) was prepared and sonicated for 30 min. Afterward, the solution was located in a Teflon-lined autoclave, and then hydrothermal treatment was carried out at 160°C for 12 h. After hydrothermal treatment, the synthesized CNSs were collected and rinsed successively with deionized (DI) water and ethanol several times. The washed CNSs were dried in an oven at 60°C overnight.

    2.2. CNS characterization

    In order to determine whether CNSs possess oxygen functional groups, the chemical characteristics of the as-prepared CNS were examined by a Fourier transform infrared (FTIR) spectrometers (Nicolet 6700, Thermo Scientific, USA) equipped with an attenuated total reflection (ATR). In addition, the bulk zeta potential of CNSs was measured by an electrophoretic light scattering spectrometer (ELS-Z2, Ostuka Electronic, Japan). Lastly, the morphologies of the CNS were observed by transmission electron microscopy (TEM; LIBRA 120, CarlZeiss, Germany). The viscosities of the polymer solution containing different concentrations of CNS were evaluated at a share rate of 1 s‒1 using an advanced rheometric expansion system (ARES; Rheometric Scientific ARES, Rheometric Scientific, USA).

    2.3. Membrane fabrication

    Polysulfone (PSf; Solvay Korea, Korea) UF membranes were fabricated using a polymer solution consisting of 15 wt% PSf and 85 wt% N, N-Dimethylformamide (DMF) via non-solvent induced phase separation (NIPS), as reported previously[17,18]. In detail, a polymer solution was coated on a non-woven fabric and immediately immersed in tap water. The solidified membrane was stored overnight in a non-solvent bath for entire liquid-liquid demixing. To prepare CNS composite membranes, different concentrations (0, 1, 2, 4, 8, and 16 mg mL‒1) of CNSs were added to a polymer solution. The as-prepared polymer solutions were solidified in the same way as the above-mentioned NIPS method. The content of CNS incorporated in the solidified CNS composite membranes corresponded to 0, 0.6, 1.2, 2.3, 4.6, and 8.8 wt% with respect to the solidified PSf mass. Accordingly, the as-prepared CNS composite membranes were marked as CNS0, CNS0.6, CNS1.2, CNS2.3, CNS4.6, and CNS8.8, respectively.

    2.4. Membrane characterization

    The surface morphologies and cross-sectional sublayer structures of the as-prepared CNS composite membranes were observed by means of a field-emission scanning electron microscope (FESEM; JSM- 7600F, JEOL, Japan) after being coated with platinum in a sputter coater (Cressington 108auto, Cressington Scientific Instruments Ltd., UK). As for cross-sectional FESEM images, they were fractured in liquid nitrogen ahead of observation. The surface pore size and bubble point of the as-prepared CNS composite membranes were measured using a capillary flow porometer (CFP-1500AEL, PMI, USA) and 1,1,2,3,3,3-hexafluorupropene as known as a Galwick solution (PMI; surface tension: 16 dyn cm−1) based on a wet up-dry down method. The mechanical properties of the CNS composite membranes were assessed with a universal testing machine (Instron, USA). The steady-state pure water flux (LMH, L m−2 h−1) of the as-prepared CNS composite membranes was measured at 1 bar using a stirred cell (Model 8010, Amicon Corp., USA) after membrane compaction for 4 h.

    2.5. MBR operation

    A continuous MBR operation was conducted to determine the feasibility of the CNS composite membranes from the perspective of anti-fouling properties and long-term stability. To feed microorganisms in an MBR, the synthetic wastewater was prepared in the following composition (mg L−1), as reported previously[ 7]: glucose, 400; yeast extract, 14.0; bactopeptone, 115; (NH4)2SO4, 105; KH2PO4, 21.8; MgSO4∙ 7H2O, 32.0; FeCl3∙6H2O, 0.13; CaCl2∙2H2O, 3.25; MnSO4∙5H2O, 2.88; NaHCO3, 256. The details of the operating conditions are provided in Table 1. The as-prepared CNS composite membranes were simultaneously evaluated after being installed in the same MBR. While operating an MBR, the transmembrane pressure (TMP, kPa) was recorded as an indicator of the degree of membrane fouling.

    3. Results and Discussion

    3.1. Characterization of the as-prepared CNSs

    The as-prepared CNSs were found to have a spherical shape with a diameter of about 590 (20, n = 4) nm, as shown in a TEM image (Fig. 4). As for the chemical characteristics identified by FTIR spectroscopy (Fig. 5), the CNSs feature a variety of oxygen functional groups comparable to graphene oxide (GO), which was prepared by the modified Hummers method used in the previous literature[15]. Thanks to the myriad oxygen functional groups, the bulk zeta potential of the CNSs was about −27 mV (2 mV, n = 6), which was half of GO’s zeta potential (−52 mV (1 mV, n = 2)). The difference in the bulk zeta potential was assumed to be attributed to the specific surface area’s difference arising from the nanomaterials’ shapes. In other words, GO is known to have a high specific surface area owing to its atomic thickness. On the other hand, the CNS should have a lower specific surface area since only part of the materials were exposed to the external environment while a considerable portion was underneath the surface. However, although the potential of the CNS to endow a membrane with negative charge was limited compared to GO, the CNS’s oxygen functional groups were thought to confer hydrophilic and antifouling properties to a membrane similarly to other types of nanomaterials with oxygen groups[15, 19, 20].

    3.2. Surface pore size of the CNS composite membranes

    After adding CNS particles to polymer solutions at different concentrations, we investigated the surface pore size and sublayer structure of the CNS composite membranes to clarify the CNS’s behavior during phase separation. As for the pore size analysis, the mean surface pore size was found to increase from 20.9 to 37.7 nm as the CNS content increased from 0 to 8.8 wt% (Fig. 6). Intriguingly, unlike the gradually increased mean surface pore size, the bubble point data showed a spike at the CNS8.8 (Fig. 6). To clarify the reason for the distinctive phenomenon observed in the bubble point data, we explored the surface morphologies of the CNS composite membranes. According to the surface morphology investigation (Fig. 7(a) to 7(f)), CNS particles were found to form crescent-shaped pores on the surfaces of the CNS0.6 to CNS4.6 to the extent that they did not compromise the membrane integrity in terms of particle removal. Note that the crescent- shaped gap between the membrane matrix and CNS ranged from about 35 to 55 nm (Fig. 7(g)), which was still much smaller than their bubble points (about 120 μm). In stark contrast, when it comes to the CNS8.8 (Fig. 7(h)), it was confirmed to have even bigger gaps (250~450 nm) on the surface than its average bubble point (about 150 nm). The significantly large gaps observed on the CNS8.8’s surface were presumed to arise from a too-high concentration of the CNS upon consideration that a bundle of CNSs resulted in a group behavior capable of making the formation of irregular overlapped pores (i.e., irregular doublet or triplet crescent pores) much more prominent. That could explain the spike in the CNS8.8’s bubble point data.

    3.3. Sublayer structure of the CNS composite membranes

    Another point worthy of note is the sublayer structure change induced by the CNS addition. As shown in Fig. 8(a), the CNS0’s sublayer entirely consisted of a dense sponge-like structure, whereas it included a few macrovoids of tens-of-micrometers in the middle of the sublayer. The macrovoids are known to stem from density-driven convection occurring while going through a significant local density change during rapid desolvation of DMF[21, 22], which is less favorable for PSf. What attracted our eyes to the sublayer structure of the CNS composite membranes was that more macrovoids formed with the loading of CNSs (Fig. 8(b) to 8(f)). Especially, it was evident from CNS4.6 and CNS8.8 that the inflow of the non-solvent became noticeably remarkable, thereby leading to the significant advance of the non-solvent front and the resulting formation of finger-like structure in the sublayer, although DMF was used as a solvent. Such a transition was thought to be attributable to the crescent- shaped pores capable of permitting the non-solvent inflow faster than the outflow of solvent as the CNS content increased.

    Meanwhile, the viscosity is the second factor inducing the more porous sublayer structure. To be specific, an excessive amount of additives is known to cause rheological hindrance and the resulting delayed demixing during phase separation, forming a dense pore structure in the end[15,23-25]. This phenomenon has been commonly found in the application of GO to membrane fabrication via NIPS. Indeed, the viscosity of a 15 wt% PSf solution soared almost 4 times as the GO content increased from 0 to 8.8 wt% (Fig. 9(a)). In stark contrast, it was quite a surprise to see that the polymer solution containing different concentrations of CNSs maintained the viscosity regardless of the CNS content (Fig. 9(a)). This distinct difference in the rheological behaviors between GO and CNS could be understood in line with the Brownian rotations observable in nanoscale phenomena. Briefly, the Brownian rotations of colloidal particles can cause additional friction in the fluid, contributing to the shear stress of the fluid and the resulting increase in viscosity[26]. In this regard, it is worth noting that the number of particles per unit mass of GO, which is a kind of two-dimensional anisotropic colloid, is much larger than that of CNS because it means that GO is likely to significantly increase the polymer solution viscosity due to the high particle number density per unit volume of the polymer solution. Thanks to the favorable behavior of CNS to suppress an increase in the polymer solution viscosity, high loading of CNS did not cancel out the more porous pore structure induced by CNSs.

    3.4. Water permeability of the CNS composite membranes

    As such, the addition of CNS brought about a more porous surface and sublayer, which was assumed to reduce membrane resistance and enhance the water permeability of the CNS composite membranes. Indeed, the water flux of the CNS composite membranes was improved with the CNS loading until the CNS content increased up to 2.3 wt% (Fig. 9(b)). However, the pure water flux of the CNS4.6 and CNS8.8 rather decreased over the left shoulder. On top of that, the increasing rate of the pure water flux declined as the operating pressure increased (Fig. 9(c)). The water flux trend of the CNS composite membranes could be influenced by their mechanical properties. In detail, the membranes for a pressure-driven filtration process need to meet the requirement for high mechanical properties endurable for high pressure and the resulting membrane compaction over a certain level regardless of the type of membrane processes[27-29]. With this in mind, we examined the mechanical properties of the CNS composite membranes. According to the modulus data (Fig. 9(d)), the moduli of the CNS composite membranes were higher than or comparable to the control membrane within the CNS content of 0.6 to 2.3 wt%. However, the CNS4.6 and CNS8.8 exhibited lower moduli than the CNS0, which is attributed to the fact that the two membranes were too porous. In light of this fact, we chose CNS2.3 as an optimal CNS composite membrane from the perspective of pore structure, water permeability, and mechanical properties.

    3.5. The CNS composite membranes’ anti-biofouling properties evaluated in MBR operation

    The CNS2.3’s anti-biofouling properties were compared with the CNS0 during continuous MBR operation. Since an MBR is operated under constant flux conditions, the degree of membrane fouling is revealed by the TMP profile (i.e., the higher the TMP profile, the higher the membrane fouling). Accordingly, how long we delay a TMP increase is crucial from the perspective of efficient MBR operation in that extending a cleaning period cuts down on the cost incurred by the material purchase (e.g., chemicals and membrane modules) and personnel expenses for membrane maintenance (e.g., membrane cleaning and replacement). Note that membrane cleaning is required when the TMP reaches 30 kPa[7]. Based on this fact, we evaluated the membrane filterability of the CNS0 and CNS2.3. As shown in Fig. 10, the CNS2.3 (122 h) effectively prolonged the time required to reach 30 kPa almost by 5 times compared to the CNS0 (24.5 h). This delayed TMP increase could stem from not only the porous structure induced by the crescent-shaped pores but also the CNS’s oxygen functional groups capable of endowing the membrane surface with hydrophilic properties and charge repulsion.

    Meanwhile, one may wonder whether it is okay for the CNS2.3 to be used beyond 30 kPa upon consideration that they are more fouling-resistant owing to their hydrophilicity and porous structures. To answer the question, we measured the mass and proportion of the total attached biomass (TAB), extracellular polymeric substances (EPS), and cells. According to our autopsy, the CNS2.3 (37%) exhibited a slightly higher ratio of EPS to TAB than the CNS0 (29%) as its TMP reached 50 kPa. This result indicates that a longer operation than an advisable level can exacerbate membrane fouling because EPS accumulates consistently for a longer time after being secreted by microorganisms in a biofilm. From the result, we can conclude that CNS2.3 needs to be used until the TMP reaches 30 kPa, although it possesses anti-biofouling properties.

    4. Conclusion

    In this study, CNSs were prepared via an environmentally friendly hydrothermal method to improve the filtration performance of a UF membrane by endowing the membrane with hydrophilicity and porous structures. The as-prepared CNS were found to be hydrophilic and formed crescent-shaped pores. The crescent- shaped pores not only enlarged the surface pore size but also made a sublayer pore structure more porous gradually with the loading of CNS. Another point worthy of note is that CNS did not increase a polymer solution viscosity owing to its isotropic morphologies and relatively low particle number density per unit volume of polymer solution compared to different types of nano additives, contributing to preventing rheological hindrance. Among the CNS composite membranes, CNS2.3 exhibited the best water permeability (2 times higher than CNS0) since it balanced the porous pore structure and mechanical properties in contrast to CNS4.6 and CNS8.8, which could not endure membrane compaction properly due to too porous structure. Lastly, CNS2.3 successfully mitigated membrane biofouling by extending the cleaning period by 5 times in continuous MBR operation until the TMP reached 30 kPa.

    Declaration of Competing Interest

    The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.


    This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (grant number: 2021R1I1A3044565). This work was also supported by the Commercialization Promotion Agency for R&D Outcomes (COMPA) grant funded by the Korean Government (Ministry of Science and ICT) (RS-2023-00304743).



    Concentration polarization model (adapted from [10]).


    The effect of forces and torques on the back diffusive transport of a charged spherical particle dispersed in a viscous liquid going through laminar flow in the proximity of a flat porous surface (adapted from [11]).


    A schematic illustration of a continuous MBR system (adapted from [7]).


    TEM images of the (a) CNS and GO.


    FTIR speactra of of the CNS and GO.


    Mean pore sizes and bubble points of the CNS composite membranes (n = 3).


    Surface SEM images of the (a) CNS0, (b) CNS0.6, (c) CNS1.2, (d) CNS2.3, (e) CNS4.6, and (f) CNS8.8. Magnified surface SEM images of the (g) CNS4.6 and (h) CNS8.8. Scale bar: 3 μm for (a) to (f), 200 nm for (g), and 500 nm for (h), respectively.


    Cross-sectional SEM images of the (a) CNS0, (b) CNS0.6, (c) CNS1.2, (d) CNS2.3, (e) CNS4.6, and (f) CNS8.8. Scale bar: 30 μm.


    (a) Viscosity of different concentrations of the as-prepared GO and CNS (n = 3). The average size of GO nanosheets used in this study was around 230 (40) nm. (b) Pure water flux (n = 3) and (c) increasing rate of the pure water flux of the CNS composite membranes measured at different operating pressures. (d) Modulus of the CNS composite membranes (n = 3).


    TMP profiles of the CNS0 and CNS2.3 measured during continuous MBR operation.


    MBR Operating Conditions

    The Mass of TAB, EPS, and Cells Obtained from Biofilms Formed until the TMP Values of Each Membrane Module Reached 50 kPa


    1. J. Lee, J. H. Jang, H.-R. Chae, S. H. Lee, C.-H. Lee, P.-K. Park, Y.-J. Won, and I.-C. Kim, “A facile route to enhance the water flux of a thin-film composite reverse osmosis membrane: incorporating thickness-controlled graphene oxide into a highly porous support layer”, J. Mater. Chem. A, 3, 22053-22060 (2015).
    2. J. Lee, F. Zhou, K. Baek, W. Kim, H. Su, K. Kim, R. Wang, and T.-H. Bae, “Use of rigid cucurbit 6. uril mediating selective water transport as a potential remedy to improve the permselectivity and durability of reverse osmosis membranes”, J. Membr. Sci., 623, 119017 (2021).
    3. J. Lee, C. Ahn, S. Chang, J. Lee, and T.-H. Bae, “Overcoming the trade-off between water flux and salt rejection of an RO membrane via simultaneous optimization using highly porous microstructured support and sulfonated porous organic polymerbased polyamide active layer”, J. Membr. Sci., 687, 122085 (2023).
    4. Y. J. Lim, J. Lee, T.-H. Bae, J. Torres, and R. Wang, “Feasibility and performance of a thin-film composite seawater reverse osmosis membrane fabricated on a highly porous microstructured support”, J. Membr. Sci., 611, 118407 (2020).
    5. S. Lee, H.-J. Kim, M. Tian, G. Khang, H.-W. Kim, T.-H. Bae, and J. Lee, “Silk fibroin-coated polyamide thin-film composite membranes with anti- scaling properties”, Desalination, 546, 116195 (2023).
    6. Z. M. Ghazi, S. W. F. Rizvi, W. M. Shahid, A. M. Abdulhameed, H. Saleem, and S. J. Zaidi, “An overview of water desalination systems integrated with renewable energy sources”, Desalination, 542, 116063 (2022).
    7. J. Lee, Y.-J. Won, D.-C. Choi, S. Lee, P.-K. Park, K.-H. Choo, H.-S. Oh, and C.-H. Lee, “Micro-patterned membranes with enzymatic quorum quenching activity to control biofouling in an MBR for wastewater treatment”, J. Membr. Sci., 592, 117365 (2019).
    8. W. Rongwong, J. Lee, K. Goh, H. E. Karahan, and T.-H. Bae, “Membrane-based technologies for post-treatment of anaerobic effluents”, npj Clean Water, 1, 21 (2018).
    9. J. He, S. Xia, W. Li, J. Deng, Q. Lin, and L. Zhang, “Resource recovery and valorization of food wastewater for sustainable development: An overview of current approaches”, J. Environ. Manage., 347, 119118 (2023).
    10. S.-H. Yoon, “Membrane bioreactor processes: Principles and applications”, CRC press, Boca Raton, United States (2015).
    11. S.-H. Yoon, C.-H. Lee, K.-J. Kim, and A. G. Fane, “Three-dimensional simulation of the deposition of multi-dispersed charged particles and prediction of resulting flux during cross-flow microfiltration”, J. Membr. Sci., 161, 7-20 (1999).
    12. M. C. Porter, “Concentration polarization with membrane ultrafiltration”, Ind. Eng. Chem. Prod. Res. Dev., 11, 234-248 (1972).
    13. G. Segré and A. Silberberg, “Behaviour of macroscopic rigid spheres in Poiseuille flow Part 1. Determination of local concentration by statistical analysis of particle passages through crossed light beams”, J. Fluid Mech., 14, 115-135 (1962).
    14. G. Segre and A. Silberberg, “Behaviour of macroscopic rigid spheres in Poiseuille flow Part 2. Experimental results and interpretation”, J. Fluid Mech., 14, 136-157 (1962).
    15. J. Lee, H.-R. Chae, Y. J. Won, K. Lee, C.-H. Lee, H. H. Lee, I.-C. Kim, and J.-m. Lee, “Graphene oxide nanoplatelets composite membrane with hydrophilic and antifouling properties for wastewater treatment”, J. Membr. Sci., 448, 223-230 (2013).
    16. X. Song, P. Gunawan, R. Jiang, S. S. J. Leong, K. Wang, and R. Xu, “Surface activated carbon nano spheres for fast adsorption of silver ions from aqueous solutions”, J. Hazard. Mater., 194, 162- 168 (2011).
    17. H. M. Tran, D. Kwak, U. Lee, S. Chang, D. T. Tran, and J. Lee, “A hydrophilic near dissolved organic matter microfiltration (NDOM MF) membrane prepared using multifunctional porogen synthesized via metal-free atom transfer radical polymerization for highly efficient microplastic removal”, Chem. Eng. J., 480, 147564 (2023).
    18. J. Lee, Y. J. Lim, J. H. Low, S. M. Lee, C.-H. Lee, R. Wang, and T.-H. Bae, “Synergistic effect of highly porous microstructured support and co-solvent assisted interfacial polymerization on the performance of thin-film composite FO membranes”, Desalination, 539, 115947 (2022).
    19. H. E. Karahan, K. Goh, C. J. Zhang, E. Yang, C. Yildirim, C. Y. Chuah, M. G. Ahunbay, J. Lee, S. B. Tantekin-Ersolmaz, Y. Chen, and T. H. Bae, “MXene materials for designing advanced separation membranes”, Adv. Mater., 32, e1906697 (2020).
    20. Q. Zhang, J. Wang, Z. Zhou, K. Huo, M. Dou, C. Han, X. Wang, S. M. Hasan, and B. Gao, “Preparation of GO@ DCN nanocomposite modified PVDF membranes for improved membrane performance and exploration of formation mechanism”, J. Membr. Sci., 683, 121859 (2023).
    21. J. Lee, R. Wang, and T.-H. Bae, “High-performance reverse osmosis membranes fabricated on highly porous microstructured supports”, Desalination, 436, 48-55 (2018).
    22. H. Lee, W. B. Krantz, and S.-T. Hwang, “A model for wet-casting polymeric membranes incorporating nonequilibrium interfacial dynamics, vitrification and convection”, J. Membr. Sci., 354, 74-85 (2010).
    23. K. Ho, Y. Teow, A. Mohammad, W. Ang, and P. Lee, “Development of graphene oxide (GO)/multiwalled carbon nanotubes (MWCNTs) nanocomposite conductive membranes for electrically enhanced fouling mitigation”, J. Membr. Sci., 552, 189-201 (2018).
    24. A. Khan, T. A. Sherazi, Y. Khan, S. Li, S. A. R. Naqvi, and Z. Cui, “Fabrication and characterization of polysulfone/modified nanocarbon black composite antifouling ultrafiltration membranes”, J. Membr. Sci., 554, 71-82 (2018).
    25. J. Wei, Y. Li, L. Setiawan, and R. Wang, “Influence of macromolecular additive on reinforced flat-sheet thin film composite pressure-retarded osmosis membranes”, J. Membr. Sci., 511, 54-64 (2016).
    26. A. M. Wierenga, A. P. J. C. Philipse, S. A. Physicochemical, and E. Aspects, “Low-shear viscosity of isotropic dispersions of (Brownian) rods and fibres; A review of theory and experiments, Colloids Surf. A Physicochem. Eng. Asp., 137, 355-372 (1998).
    27. J. Lee, H. Yoon, J. H. Yoo, D.-C. Choi, C. H. Nahm, S. H. Lee, H.-R. Chae, Y. H. Kim, C.-H. Lee, and P.-K. Park, “Influence of the sublayer structure of thin-film composite reverse osmosis membranes on the overall water flux”, Environ. Sci. Water Res. Technol., 4, 1912-1922 (2018).
    28. L. Nie, K. Goh, Y. Wang, J. Lee, Y. Huang, H. E. Karahan, K. Zhou, M. D. Guiver, and T. H. Bae, “Realizing small-flake graphene oxide membranes for ultrafast size-dependent organic solvent nanofiltration”, Sci. Adv., 6, eaaz9184 (2020).
    29. U. Lee, E. S. Jang, S. Lee, H. J. Kim, C. W. Kang, M. Cho, and J. Lee, “Near dissolved organic matter microfiltration (NDOM MF) coupled with UVC LED disinfection to maximize the efficiency of water treatment for the removal of Giardia and Cryptosporidium”, Water Res., 233, 119731 (2023).