Effect of CNT content on physicochemical properties and performance of CNT composite polysulfone membranes

Accepted Manuscript Title: Effect of CNT content on physicochemical properties and performance of CNT composite polysulfone membranes Authors: Liheng Xu, Jinsong He, Yang Yu, J. Paul Chen PII: DOI: Reference:

S0263-8762(17)30084-9 http://dx.doi.org/doi:10.1016/j.cherd.2017.01.031 CHERD 2561

To appear in: Received date: Revised date: Accepted date:

29-3-2016 6-1-2017 31-1-2017

Please cite this article as: Xu, Liheng, He, Jinsong, Yu, Yang, Chen, J.Paul, Effect of CNT content on physicochemical properties and performance of CNT composite polysulfone membranes.Chemical Engineering Research and Design http://dx.doi.org/10.1016/j.cherd.2017.01.031 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Effect of

CNT

content

on physicochemical

properties

and

performance of CNT composite polysulfone membranes

Liheng Xu1,2, Jinsong He2, Yang Yu2, J. Paul Chen2*

1

Department of Environmental Engineering, China Jiliang University, Hangzhou 310018, China. 2 Department of Civil and Environmental Engineering, National University of Singapore, Singapore 119260, Singapore *

Corresponding author. E-mail address: [email protected]; [email protected]

Abstract: This study was to research the influences of carbon nanotubes (CNTs) on the physicochemical properties and filtration performance of CNT/polymer composite membranes. Two types of CNT embedded polysulfone (PSF) membranes were fabricated by using raw CNTs (rCNTs) and oxidized CNTs (fCNTs), respectively. The leakage of CNTs during the membrane fabrication was observed; the loss of fCNTs in the fCNT membrane (M-fCNT) was more severe than that of rCNTs in the rCNT membrane (M-rCNT). The porosity, surface hydrophilicity, thermal stability and strain stress of the membranes were dependent upon the CNT content. The CNTs composite membranes exhibited typical asymmetric membrane structure including a thin dense top surface and a porous bulk layer with abundant finger-like pores, and the mean pore size of top surfaces increased with the incorporation of CNTs. The overall porosities of M-rCNT and M-fCNT membranes were 54 and 68%, respectively.

M-fCNT membranes exhibited slightly lower thermal stability and mechanical strength than PSF membrane (M-PSF) and M-rCNT. The pure water flux of the composite membrane was dramatically enhanced by the loading of CNTs, as indicated by 2 times higher for M-rCNT2.0 and M-fCNT2.0 than M-PSF. The membranes were further studied by the filtration of 2-naphthol containing solution; M-fCNT2.0 had significantly higher rejection ability than that of the pristine PSF membrane.

Key words: carbon nanotubes (CNTs); composite membrane; polysulfone (PSF); rejection; ultrafiltration.

1. Introduction In recent years, membrane separation technology has attracted increasing attention in the treatment of drinking water as well as wastewater due to its advantages of high stability and efficiency, low environmental footprint and being free of chemicals [1, 2]. The conventional operation of the technology based on a membrane makes it superior to other methods such as adsorption and distillation. According to the operational pressure, membrane-based liquid separation technologies could be sorted into microfiltration (MF), ultrafiltration (UF), nanofiltration (NF) and reverse osmosis (RO). NF and RO were reported to be efficient in rejecting organic micro-pollutants [3, 4], while the main constraint of these technologies is the high energy consumption due to the high operational pressures [1, 5, 6]. Meanwhile, the rejection performance

of MF and UF was limited by their large pore sizes although their operational pressures were relatively lower. To achieve a more cost-effective and efficient membrane separation technology, the development of advanced membrane materials with a relatively high rejection and low operational pressure is crucial to be needed. A composite membrane composed of a polymer matrix and inorganic filler seemingly provides a promising approach to enhancing the permeability and separation properties of the membrane. Many inorganic materials are currently being actively researched as candidates used in the fabrication of composite membranes, including metal oxides [7, 8], zeolites [9], clay [10] and carbon materials [11, 12]. Carbon nanotubes (CNTs) possess remarkable advantages including high mechanical strength, high surface area and strong adsorption affinity towards organic matters and microbes. Many researchers have added CNTs to various polymers such as poly (vinyl alcohol) (PVA) [13], PSF [14] and polyvinylidene fluoride (PVDF) [15] to fabricate different composite membrane materials. The properties of the membranes, such as surface properties, pore structure and tensile strength, were altered by the presence of CNTs even at very low loadings, resulting in higher antifouling and organic matter rejection [2, 16, 17]. Recently, researchers have found that the characteristics of incorporated CNTs remarkably influence the properties of the membranes, and consequently change the membrane performance. Murthy et al. [18] fabricated a composite membrane with three types of functionalized CNTs, namely, oxidized, amide and azide CNTs, and it

was found that the

amide-functionalized CNTs resulted in the best rejection in metal removal from

drinking water. Huang et al. [19] studied the influence of the CNT diameter on the blend membrane performance and found that the membrane modified with larger diameter CNT was more effective in removing larger organic macromolecules from feed water. Wiesner et al. [20] found that the benefits and disadvantages coexisted for the blend membranes as the oxidation degree of the loading CNTs changed. To achieve superior performance of CNT composite membranes, the alteration and manipulation of the membrane properties by the incorporating CNTs content should be extensively explored. In this paper, PSF membranes composited with different contents of raw CNTs (rCNTs) and oxide-functionalized CNTs (fCNTs) were fabricated by the phase inversion method. The morphology, porosity, surface hydrophilicity, thermal stability, and mechanical strength of the composite membranes were studied to investigate the influences of the CNTs content, and their permeability and rejection of organic pollutants were also investigated using a UF experimental system.

2. Materials and methods 2.1 Materials CNTs were purchased from Shenzhen Nanotech Port Co. (China). The reported outer diameter was 10-20 nm, and the length was 1-2 μm. The CNTs as received were referred to as rCNTs in this experiment. The oxidized CNTs (referred as fCNTs) were prepared from rCNTs via acid oxidization process reported in the literature [21]. Briefly, 500 mg of CNTs was added to 250 mL of mixed acid of concentrated H2SO4

and HNO3 (volume ratio of 3:1). The mixture was sonicated at 60°C for 7 h, followed by extensive washing and filtrating in deionized water (DI water) until the filtrate was neutral. The fCNTs were then dried at 50°C. Polysulfone

(PSF,

P-1700)

was

purchased

from

Solvay

(USA).

Polyvinylpyrrolidone (PVP) with a molecular weight of 40,000Da was purchased from Sigma-Aldrich (USA). 1-Methyl-2-pyrrolidone (NMP) and 2-naphthol (99% purity) were purchased from J&K Chemical (China).

2.2 Preparation of membranes The composite flat plate PSF membranes were prepared by the phase inversion method. First, an appropriate amount of rCNTs or fCNTs was uniformly dispersed in 85 mL of NMP as solvent under ultrasonic for 4 h. Then, 15 g of PSF and 3 g of PVP were added into above mixed solution, by shaking for 24 h to form a homogeneous casting solution. After that, the casting solution was degassed through 1-h ultrasonication and at least 5-h stationary store at room temperature. The obtained solution was then cast onto a glass plate using an Elcometer 3570 micrometric film applicator which was set to a constant distance of 200 μm. After exposure to air for 30 seconds, the glass dish was immersed into a DI water coagulation bath at room temperature. Once the membrane was peeled off from the glass dish, it was transferred to another DI water bath and stored for at least 48 hours to remove the NMP solvent. The prepared membranes were stored in a DI water bath, and were dried at 50°C before characterization.

The membranes were referred to by their CNT content. M-rCNTx and M-fCNTx refer to the PSF membranes composited with rCNTs and fCNTs, respectively. The value of x refers to the rCNT or fCNT dosage in the form of wt% PSF. The membrane without any mixed CNTs was named as M-PSF. To experimentally determine the actual loading amount of CNTs in the composite membranes, a certain mass of membrane was dissolved in NMP under ultra-sonication, and the absorbance of the obtained solution at 600 nm wavenumber was then determined using a UV-vis spectrophotometer (UV1800, Shimadzu, Japan) with a NMP solvent as reference solution. The content of rCNT or fCNT was calculated according to Lambert-Beer law.

2.3 Characterization of membranes The Raman spectra of the prepared membranes were collected on a Nicolet Almega Raman Spectrometer (Thermal, USA) with a 532 nm laser source. The morphology of the membranes were investigated by scanning electron microscopy (SEM) using an SU8000 instrument (Hitachi, Japan). The membrane samples were sputter coated with Au atoms, and the surface and cross-section of the membranes were studied by SEM at 5.0kV. ImageJ open-source image analysis software was adopted to calculate the surface porosity, mean pore size and pore size distribution form the SEM image. Thermogravimetric (TG) and differential thermogravimetric (DTG) analyses of the membranes were performed using a SDT Q600 instrument (TA Co., USA) under a nitrogen atmosphere, in the temperature region from room temperature to 700°C.

Stress-strain measurements of the composite membranes were performed with an RTW10 electronic universal testing machine (Reger, Shenzhen, China) at 25°C, with a crosshead speed of 10 mm/min and an initial gauge length of 20 mm. The Young’s modulus of the membranes was calculated from the stress-strain data. The overall porosity of the membranes was measured by the gravimetric method. After measuring the dry weight of the membranes, they were subsequently immersed in DI water overnight to ensure water penetration into the membrane pores. After removing the excess water from the membrane surface with tissue paper, the wet weight of the membrane was measured. The overall porosity was calculated as follows. Overall porosity =

(

−

)

(1)

where mw and md are the weight of the wet and dry membranes, respectively; ρwater is the density of pure water at 25°C, and V is the volume of the dry membrane. The static water contact angle of the top surface of the membranes was monitored with a JC2000C contact angle goniometer (Zhongchen Co., China) to evaluate the hydrophilicity of the membrane. Ultrapure water was dropped onto the surface of the membrane sample placed on the monitoring platform, and the contact angle was recorded.

2.4 Flux test and filtration experiment A dead-end filtration system (Millipore Micro 8010, USA) was employed to test the membrane flux as this method is most commonly used in membrane water treatment.

The membrane effective area was 4.9 cm2. The membrane was pre-pressured at 20 psi using pure water for 1 h, and then the pure water flux was measured at the pressure of 15 psi. The pure water flux (J) was calculated as follows J=

∆

(2)

where V is the volume of permeated pure water (L), A is the effective area of the membrane (m2), and t is the permeation time (h). 2-naphthol was chosen as a representative contaminant in this study to evaluate the rejection performance of the composite membranes. The filtration test of the organic contaminant was carried out with the same dead-end experimental system. After pre-pressurizing at 20 psi for 1 h, the contaminant solution with a feed concentration of 4.0 mg/L was filtrated at 15 psi. The concentration of the permeated solution was determined on-line with a UV1800 detector at 273nm wavelength, and the corresponding permeated volume was read from the collector at the terminal.

3. Results and discussions 3.1 Formation of CNT composite membranes To fabricate CNT composite polymer membranes, different amounts of rCNTs and fCNTs were dispersed in a PSF casting solution and then cast into membranes by the phase inversion method. The resulted composite membranes are shown in Figure 1. It is clearly observed that both the rCNTs and fCNTs are well dispersed in the PSF polymer matrix forming composite membranes. The remarkable difference in color between the M-rCNT and M-fCNT membranes was noticed, indicating the different

actual CNT content of the composite membranes even with the same designed CNT dosage. This may be attributed to the CNTs leakage in the formation process of the membranes. During the phase inversion process in the water coagulation bath, some of the nanotubes were found migrating from the membranes into the water along with the NMP solvent. To quantify the leakage level of the CNT, the actual CNT content in the membranes was determined, as shown in Table 1. Although CNT leakage occurred in every membrane of the two studied series, the loss percentage dramatically varied according to the type and loading amount of CNTs. The leakage of the fCNT was clearly more severe than that of the rCNT. Approximately 40.7% of the fCNT leached out when the dosage amount reached 2.0 wt% of the polymer, while for rCNT only 6.55% of the rCNTs leached out for the same dosage. Compared to rCNT, fCNT possessed a superior dispensability in the NMP solvent as well as in water, due to their surface functional groups. The superior dispensability consequently resulted in easier leakage of fCNTs during the membrane formation. Lannoy et al. [20] found that a greater degree of carboxylation resulted in a greater CNT loss during membrane formation. For the M-rCNT membranes, the loss percentage of the rCNT sharply decreased as the rCNT dosage increasing. When the dosage rose from 0.2 to 2.0 wt%, the loss percentage decreased from 21.5% to 6.55%. This may be due to the fact that increasing rCNT content strengthens the Van de Walls forces between the nanotubes and/or between the nanotube and the polymer [22]. Raman spectroscopy can be used as a useful method to characterize the properties

of CNTs composite membranes [23-25]; the Raman spectra of the CNT composite membranes are shown in Figure 2. The intense Raman bands of PSF at 3081cm-1 and 1605 cm-1 are clearly located in the spectra of each blend membrane. The weak Raman bands of 1364 cm-1 and 2712 cm-1 in the spectra of the composite membranes were appointed to be the D band and the overtone of 2D of CNT [26], and the intensities of these two peaks varied along with the CNT content in the membranes as shown. It is noted that the frequencies of D band and the overtone shift towards higher wavenumber in composite membranes than that in free CNT (1348 cm-1 and 2679 cm-1, respectively); this indicates that the microenvironment of nanotubes changed as being dispersed in polymer matrix.

3.2 Morphology of CNT composite membranes The morphology of the CNTs composite membranes was examined by SEM. Figure 3 shows representative SEM images of the surfaces and cross-sections of the membranes. As shown, both the M-PSF and CNTs composite membranes exhibited typical asymmetric membrane structure including a thin dense surface and a porous bulk layer with abundant finger-like pores. Pores were investigated spreading evenly on the compact top surfaces. The pore size distributions of the surfaces could be calculated from the SEM images (Figure S1). The insertion of CNTs into PSF membrane resulted in broader pore size distribution (2.5-60 nm) compared to M-PSF (2.5-40 nm), and the ratio of large pores were also increased. The mean pore diameters were 4.8 nm, 9.5 nm and 5.6 nm for M-PSF, M-rCNT2.0 and M-fCNT2.0,

respectively, and the surface porosity was found to be 2.5, 3.9 and 5.9%, respectively. The blending of carbon nanotubes in the membranes obviously caused the significant increasing in both surface pore size and porosity. The leakage of carbon nanotubes during the membrane formation process may account for the larger pore size and the higher surface porosity of the composite membranes. It was noticed that the surface porosity of M-fCNT2.0 was higher than that of M-rCNT2.0 in spite of the mean pore size of M-fCNT2.0 was lower than that of M-rCNT2.0, indicating that the higher leakage ratio of fCNTs in the membrane formation process formed more pores. Several clear worm-like nanotubes were found dispersed lying on the M-rCNT2.0 top surface as pointed by the white arrows in Figure 3. The nanotubes were also observed on the M-fCNT2.0 top surface, while the lengths of the nanotubes were clearly shorter than the rCNT due to the oxidation process of fCNT. Focusing on the enlarged images of the inner walls, the CNTs were found to be embedding in the polymer or crossing the pores in M-rCNT2.0 and M-fCNT2.0.

3.3 Physicochemical properties of CNT composite membranes The incorporation of CNTs in the polymer matrix may change the physicochemical properties of membranes including porosity, surface hydrophilicity, thermal stability and mechanical strength. The overall porosity of the CNT composite membranes, shown in Figure 4(a), varied from 54% to 68% according to the CNTs content in this experimental region. There were no distinct differences in overall porosity between M-rCNT and M-fCNT membranes. As the dosage amount of CNTs increases, the

overall porosity increased, followed by a decrease as the dosage exceeding 0.5 wt%. As for M-rCNT2.0 and M-fCNT2.0, the overall porosities were 54% and 57% respectively, lower than that for the M-PSF (62%). The variation of membrane overall porosity induced by the embedding of CNTs is also observed in other reported studies [27]. The viscosity of the casting solution of membrane increased as the increasing of CNTs content in the solutions. Subsequently, the exchange rate of solvent and water was decreased in the phase inversion process, which subsequently resulted in a reduction in the formation of marcoviods. The pore filling by CNTs was also assumed to be one of the causes of the reduced porosity [18]. The surface hydrophilicity of the composite membrane also varied with the incorporated CNT content. The contact angles of pure water on the top surface of the CNT composite membranes in the air are shown in Figure 4(b). The contact angle was obviously decreased by the presence of CNT, from 74.1° of M-PSF to 65.2° and 64.9° of M-rCNT0.2 and M-fCNT0.2 respectively. It was indicated that the surface hydrophilicity was enhanced by the mixing of CNTs in the PSF membranes. Hydrophobicity of membrane surface is affected by both intrinsic hydrophobicity of the material and the surface structure of the membrane [28]. As compared, the contact angle of the M-fCNT membrane was slightly lower than that of the M-rCNT with the same CNT dosage. This was understandable in view of the hydrophilic groups (-COOH) on the fCNT surface. Figure 5 shows the DTG curves of the CNT composite membranes under a nitrogen atmosphere. The degradation temperatures inferred from the curves are listed in Table

2. The thermal stability of composite membranes is directly related with the structure and the loading level of CNTs [15]. The enhanced thermal stability of polymer membranes by the presence of CNTs was observed in previous studies [18, 29]. Opposite effect of rCNT and fCNT loading on the thermal stability of membranes were observed in this study. For the M-rCNT membranes, the degradation temperature slightly increased as the rCNTs dosage above 1.0% (from 523.6°C for M-PSF up to 524.7°C for M-rCNT1.0). The embedding of rCNTs into PSF matrix seems to enhance the thermal stability of the resulting membranes. For the M-fCNTs membranes, the degradation temperature slightly decreased compared to the neat M-PSF, as summarized in Table 2, and the degradation temperature sharply decreased to 512.4 °C for M-fCNT2.0. The tensile strength of the membrane materials influences their usage in membrane technology operations. The CNT composite membranes were stretched under constant stress through their elastic region until the break point, and the results are shown in Table 2. Increased rCNT loading leads to an initially increased Young’s modulus, followed by a decrease. The maximum Young’s modulus was 3.36 MPa for M-rCNT0.5, which was obviously larger than that of M-PSF (2.62 Mpa). The high mechanical strength of the CNTs would be beneficial to the membranes, while a large amount of nanotubes in the polymer matrix can cause the membrane elasticity to decline. The apparent Young’s modulus is a balance of these two opposing effects. For the M-fCNT membranes, the Young’s modulus was generally less than that of M-PSF membrane in this experimental region. The greater loss of fCNTs during the

membrane formation may have changed the structure of polymer matrix [20], and thus declined the mechanical strength of the membranes.

3.4 Filtration performance of membranes Permeability is a vital property of a membrane that directly affects its practical application. Figure 6 shows that the pure water flux (J) of a membrane was dramatically enhanced by the presence of rCNTs or fCNTs. The pure water flux increased with the increasing content of CNTs in the low dosage level regardless of the type of CNTs. The flux of M-rCNT1.0 reached 322 L/m2h, 2.5 times as high as that of M-PSF. When the content of nanotubes further increased, the flux slightly decreased, while it was still much higher than that of M-PSF in this experimental region. The overall porosity and the top surface properties of a membrane were proposed to be significant factors that affect the membrane flux [14, 15]. In this study, the water fluxes of composite membranes showed a similar variation tendency with the overall porosity to the dosage CNTs content. Meanwhile, it was noticed that the pure water fluxes of M-rCNT2.0 and M-fCNT2.0 maintained 2 times higher than that of M-PSF, in spite of their lower overall porosity than M-PSF. This result is mainly ascribed to the enhancement of the surface properties of M-rCNT2.0 and M-fCNT2.0. As discussed above, the surface porosity of M-rCNT2.0 and M-fCNT2.0 is much higher than that of M-PSF, and the mean pore diameters in the top surfaces of M-rCNT2.0 and M-fCNT2.0 are also larger than that of the M-PSF. The larger pores and the higher porosity are account for the higher fluxes of the M-rCNT2.0 and

M-fCNT2.0. It was indicated that the surface properties of composite membrane played an important role in the permeability of membrane. 2-Naphthol was adopted as a representative organic pollutant in water to examine the rejection performance of the CNT composite membranes. Concentrations of 2-naphthol in the permeate solution through composite membranes using a dead-end ultrafiltration experimental system were determined. Figure 7 shows the filtration curves in the form of C/C0 versus filtration volume, in which C and C0 refer to the concentrations of the penetrant and feed solutions, respectively. Adsorption and size exclusion were assumed to be two main mechanisms in the filtration process [30]. It was obviously conducted from the gradually increasing filtration curves in Figure 7 that the adsorption of 2-naphthol acted as the main removal mechanism in the first stage for all the tested membranes. The rejection (calculated as 1-C/C0) of 2-naphthol reached a constant as the filtration volume increasing, meaning a stage of size exclusion as dominant role. As compared, the raising of penetrant concentration as the filtration volume for M-fCNT2.0 was much slower than that for M-PSF in the initial stage. The rejection of M-PSF reached a constant at about 1000 times of bed volume, while that of M-fCNT2.0 reached a constant at nearly 1800 times of bed volume. The adsorption capacity was obviously higher than that of M-PSF. The steady rejection of M-fCNT2.0 in the experiment region was kept about 0.24, obviously superior to that of M-PSF (with rejection of 0.11). The filtration performance of M-fCNT2.0 was remarkably superior to that of M-PSF. The presence of oxidized fCNTs in the PSF matrix provided 2-naphthol with an adequate hydrophilic environment and functional

groups, and the retaining capacity was consequently enhanced. M-rCNT2.0 demonstrated a similar filtration curve for 2-naphthol to M-PSF, indicating the similar filtration performances. The larger mean pore size of top surface of M-rCNT2.0 compared to M-fCNT2.0 and M-PSF may withdraw the rejection performance towards 2-naphthol. Taking into consideration the 2 times higher flux of M-rCNT2.0 than M-PSF, M-rCNT2.0 obviously possesses higher efficiency than M-PSF in the ultrafiltration process.

4. Conclusions Composite PSF membranes with varying contents of rCNTs and oxidized fCNTs were fabricated via the phase inversion method in this study. The losses of CNTs were found to be unneglectable during the composite membranes formation, especially for the fCNT composite membranes. The composite membranes exhibited a similar structure, composed of a thin dense top surface and a porous bulk layer with abundant finger-like pores. Carbon nanotubes were found to be dispersedly embedding in the polymer matrix. The porosity, surface hydrophilicity, thermal stability and strain stress of the CNT composite membranes were also found to be dependent on the CNT content. The pure water flux of the PSF membrane was dramatically enhanced by the loading of CNTs in the experimental region. M-fCNT2.0 exhibits a significantly higher rejection ability for 2-naphthol than that of the pristine PSF membrane and M-rCNT2.0 in the ultra-filtration process.

References 1. 2.

3.

4.

5. 6.

7.

8.

9.

10.

11.

12.

13.

14.

Elimelech, M. and Phillip, W.A., The future of seawater desalination: energy, technology, and the environment. Science, 2011. 333: p. 712-717. Ismaila, A.F.; Goha, P.S.; Sanipa, S.M., and Aziz, M., Transport and separation properties of carbon nanotube-mixed matrix membrane. Separation and Purification Technology, 2009. 70: p. 12-26. Wang, X.; Li, B.; Zhang, T., and Li, X., Performance of nanofiltration membrane in rejecting trace organic compounds: Experiment and model prediction. Desalination, 2015. 370: p. 7-16. Kim, J.-H.K.S.L., Chung-Hak; Kwon, Heock-Hoi; Lee, Sangho, A novel nanofiltration hybrid system to control organic micro-pollutants: application of dual functional adsorbent/catalyst. Desalination, 2008. 231: p. 276-282. Semiat, R., Energy issues in desalination processes. Environ.Sci.Technol., 2008. 31: p. 8193-8201. Ludwig, H., Energy consumption of reverse osmosis seawater desalination possibilities for its optimisation in design and operation of SWRO plants. Desalination and Water Treatment, 2010. 13: p. 13-25. He, J.; Matsuura, T., and Chen, J.P., A novel Zr-based nanoparticle-embedded PSF blend hollow fiber membrane for treatment of arsenate contaminated water: Material development, adsorption and filtration studies, and characterization. Journal ofMembraneScience, 2014. 452: p. 433-445. He, J.; Siah, T.-S., and Paul Chen, J., Performance of an optimized Zr-based nanoparticle-embedded PSF blend hollow fiber membrane in treatment of fluoride contaminated water. Water Research, 2014. 56: p. 88-97. Varoon, K.; Zhang, X.; Elyassi, B.; Brewer, D.D.; Gettel, M.; Kumar, S.; Lee, J.A.; Maheshwari, S.; Mittal, A.; Sung, C.Y.; Cococcioni, M.; Francis, L.F.; McCormick, A.V.; Mkhoyan, K.A., and Tsapatsis, M., Dispersible exfoliated zeolite nanosheets and their application as a selective membrane. Science, 2011. 334: p. 72-75. Villaluenga, J.P.G.; Khayet, M.; López-Manchado, M.A.; Valentin, J.L.; Seoane, B., and Mengual, J.I., Gas transport properties of polypropylene/clay composite membranes. Eur. Polym. J., 2007. 43: p. 1132-1143. Li, X.; Ma, L.; Zhang, H.; Wang, S.; Jiang, Z.; Guo, R.; Wu, H.; Cao, X.; Yang, J., and Wang, B., Synergistic effect of combining carbon nanotubes and graphene oxide in mixed matrix membranes for efficient CO2 separation. Journal ofMembraneScience, 2015. 479: p. 1-10. Kim, W. and Nair, S., Membranes from nanoporous 1D and 2D materials: a review of opportunities, developments, and challenges. Chem.Eng.Sci., 2013. 104: p. 908-924. Yaser, S. and Toraj, M., Effects of CNTs Content on Physicochemical and Pervaporation Separation Properties of PVA Membranes. Separation Science and Technology, 2013. 48: p. 716-727. Choi, H.-H.; Jonggeon, J., and Kim, W.-N., Fabrication and characterization

15.

16.

17.

18.

19.

20.

21. 22.

23.

24.

25.

26.

of multi-walled carbon nanotubes/polymer blend membranes. Journal of Membrane Science, 2006. 284: p. 406-415. Silva, T.L.S.; Morales-Torres, S.; Figueiredo, J.L., and Silva, A.M.T., Multi-walled carbon nanotube/PVDF blended membranes with sponge- and finger-like pores for direct contact membrane distillation. Desalination, 2015. 357: p. 233-245. Sears , K.; Dumée, L.; Schütz , J.; She, M.; Huynh, C.; Hawkins, S.; Duke, M., and Gray, S., Recent Developments in Carbon Nanotube Membranes for Water Purification and Gas Separation. Materials, 2010. 3: p. 127-149. Ahn, C.H.; Baek, Y.; Lee, C.; Kim, S.O.; Kim, S.; Lee, S.; Kim, S.-H.; Bae, S.S.; Park, J., and Yoon, J., Carbon nanotube-based membranes: Fabrication and application to desalination. Journal of Industrial and Engineering Chemistry, 2012. 18: p. 1551-1559. Shah, P. and Murthy, C.N., Studies on the porosity control of MWCNT/polysulfone composite membrane and its effect on metal removal. Journal of Membrane Science, 2013. 437: p. 90-98. Ajmani, G.S.; Goodwin, D.; Marsh, K.; Fairbrother, D.H.; Schwab, K.J.; Jacangelo, J.G., and Huang, H., Modification of low pressure membranes with carbon nanotube layers for fouling control. water research, 2012. 46: p. 5645-5654. Lannoy, C.-F.; Soyer, E., and Wiesner, M.R., Optimizing carbon nanotube-reinforced polysulfone ultrafiltration membranes through carboxylic acid functionalization. Journal of Membrane Science, 2013. 447: p. 395-402. Xu, L. and Zhang, M., Construction of carbon nanotube-based microcapsules by self-assembly. Environ Chem Lett, 2014. 12(2): p. 359-364. Britz, D.A. and Khlobystov, A.N., Noncovalent interactions of molecules with single walled carbon nanotubes. Chemical Society Reviews, 2006. 35: p. 637-659 . Hu, S.Y.; Zhang, Y.; Lawless, D., and Feng, X., Composite membranes comprising of polyvinylamine-poly(vinylalcohol) incorporated with carbon nanotubes fordehydration of ethyleneglycol by pervaporation. Journal of Membrane Science, 2012. 417-418: p. 34-44. Akieh, M.N.; Latonen, R.-M.; Lindholm, S., and Ralph, S.F., Electrochemically controlled ion transport across polypyrrole/multi-walled carbon nanotube composite membranes. Synthetic Metals, 2011. 161: p. 1906-1914. Kim, S.; Fornasiero, F.; Park, H.G.; In, J.B.; Meshot, E.; Giraldo, G.; Stadermann, M.; Fireman, M.; Shan, J.; Girgoropoulos, C.P., and Bakajin, O., Fabrication of flexible,aligned carbon nanotube/polymer composite membranes by in-situ polymerization. Journal ofMembraneScience, 2014. 460: p. 91-98. Pimenta, M.A.; Dresselhaus, G.; Dresselhaus, M.S.; Cancado, L.G.; Jorio, A., and Saito, R., Studying disorder in graphite-based systems by Raman spectroscopy. Physical Chemistry Chemical Physics, 2007. 9: p. 1276-1291.

27.

28.

29.

30.

Sengur, R.; Lannoy, C.F.; Turken, T.; Wiesner, M., and Koyuncu, I., Fabrication and characterization of hydroxylated and carboxylated multiwalled carbon nanotube/polyethersulfone (PES) nanocomposite hollow fiber membranes. Desalination, 2015. 359: p. 123-140. Shirazi, M.M.A.; Bastani, D.; Kargari, A., and Tabatabaei, M., Characterization of polymeric membranes for membrane distillation using atomic force microscopy. Desalination and Water Treatment, 2013. 51: p. 6003-6008. Chen, D.; Wang, M.; Zhang, W., and Liu, T., Preparation and characterization of poly(vinylidene fluoride) nanocomposites containing multiwalled carbon nanotubes. Journal of Applied Polymer Science, 2009. 113(1): p. 644-650. Steinle-Darling, E.; Litwiller, E., and Reinhard, M., Effects of sorption on the rejection of trace organic contaminants during nanofiltration. Environ. Sci. Technol., 2010. 44: p. 2592-2598.

Figure1. Pictures of the CNT composite membranes

M-PSF

M-fCNT2.0

Intensity

M-fCNT0.5

M-rCNT2.0 M-rCNT0.5

rCNT

4000

3500

3000

2500

2000

1500 -1

Raman shift (cm ) Figure 2. Raman spectra of the CNT composite membranes.

1000

500

Figure 3. SEM images of M-PSF (top), M-rCNT2.0 (middle) and M-fCNT2.0 (bottom). Images from left to right: top surfaces of membrane, membrane cross side, amplified images, and insight images of the membrane inner walls.

Figure 4. The overall porosity (a) and contact angle (b) of the composite membranes varying with the CNT contents.

0.75

M-PSF M-rCNT M-fCNT

overall porosity

0.70

0.65

0.60

0.55

0.50

0.45 0.0

0.5

1.0

1.5

wt% dosage of CNTs in PSF

2.0

85

M-PSF M-rCNT M-fCNT

75

.

Contact Angle ( C)

80

70

65

60

55 0.0

0.5

1.0

1.5

wt% dosage of CNTs in PSF

2.0

(a)

0.00

DTG (mg/min)

-0.05

-0.10

-0.15

-0.20 350

M-PSF M-CNT0.2 M-CNT0.5 M-CNT1.0 M-CNT2.0

400

450

500

550

600

650

?

Temperature( C) (b)

0.00

DTG (mg/min)

-0.05

-0.10

-0.15

-0.20 350

M-PSF M-fCNT0.2 M-fCNT0.5 M-fCNT1.0 M-fCNT2.0

400

450

500

550

600

650

¡£

Temperature( C) Figure 5. DTG curves of PSF membranes composited with (a) and (b) fCNTs.

450 400

2

Pure water flux (L/m h)

350

M-rCNT M-fCNT M-PSF

300 250 200 150 100 50 0 0.0

0.2

0.5

1.0

2.0

wt% dosage of CNTs in PSF Figure 6. The pure water flux of the composite membranes varying with the CNT contents

1.2 M-PSF M-rCNT2.0 M-fCNT2.0

1.0

C/C0

0.8

0.6

0.4

0.2

0.0 0

1000

2000

3000

4000

Number of bed volume

Figure 7. Filtration curves of 2-naphthol by the CNT composite membranes and neat PSF membrane. Initial concentration of 2-naphthol was 4.0 mg/L, the surface of the membrane was 4.9 cm2, and the operational pressure was 15 psi.

Table 1. The dosage and determined contents of CNT in composite membranes

Membrane M-PSF

Dosage content of Determined content of Losing CNT CNT CNT (mg/g-PSF) (mg/g-membrane) (%) / 0 0

M-rCNT0.2 2.00

1.57

21.5

M-rCNT0.5 5.00

4.21

15.8

M-rCNT1.0 10.00

8.77

12.3

M-rCNT2.0 20.00

18.69

6.55

M-fCNT0.2 2.00

1.38

31.0

M-fCNT0.5 5.00

3.09

38.2

M-fCNT1.0 10.00

6.02

39.8

M-fCNT2.0 20.00

11.86

40.7

extent

of

Table 2. Thermal and mechanical properties of the CNT composite membranes

Membrane

Degradation temperature (℃)

Tensile (MPa)

M-PSF

523.6

2.93

2.62

M-rCNT0.2

523.3

3.41

2.98

M-rCNT0.5

523.2

2.60

3.36

M-rCNT1.0

524.7

2.67

2.16

M-rCNT2.0

524.4

2.96

2.61

M-fCNT0.2

522.1

2.85

2.54

M-fCNT0.5

523.4

2.51

2.02

M-fCNT1.0

522.6

3.20

2.14

M-fCNT2.0

512.4

2.71

2.78

strength

Young’s modulus (MPa)