Journal of Membrane Science 378 (2011) 551561

Contents lists available at ScienceDirect

Journal of Membrane Science

Synthesis and characterization of carbon nanotubes/poly vinyl alcohol nanocomposite membranes for dehydration of isopropanol
Yaser Shirazi, Maryam Ahmadzadeh Toghy, Toraj Mohammadi
Research Centre for Membrane Separation Processes, Faculty of Chemical Engineering Iran University of Science and Technology (IUST), Narmak, Tehran, Iran

a r t i c l e

i n f o

a b s t r a c t
Carbon nanotubes (CNTs) were synthesized by CVD method using cyclohexanol and ferrocene as carbon precursor and catalyst, respectively. Nitric acid was used for purication and functionalization of CNTs. TEM image of CNTs before and after purication revealed that acid treatment could remove encapsulated catalyst particles. Afterward, highly pure and functionalized CNTs were incorporated in poly (vinyl alcohol) (PVA) to synthesize PVACNTs nanocomposite membranes. FESEM characterization was carried out to investigate dispersion of different CNTs loading in PVA matrix. The results showed that 2 wt.% CNTs loading is better dispersed in the polymer and increasing CNTs loading more than 2 wt.% agglomerates CNTs. Degree of swelling results for nanocomposite membranes showed that present of CNTs in the PVA membranes reduce the degree of swelling. Furthermore, pervaporation (PV) results revealed that incorporating of CNTs in PVA matrix increases signicantly water selectivity due to rigidication of the polymer chains. The water selectivities for the pristine PVA and 2 wt.% CNTs loading nanocomposite membranes were evaluated as 119 and 1794, respectively. 2011 Elsevier B.V. All rights reserved.

Article history: Received 8 February 2011 Received in revised form 19 May 2011 Accepted 22 May 2011 Available online 27 May 2011 Keywords: Nanocomposite membrane Carbon nanotubes Poly (vinyl alcohol) Pervapotation Characterization

1. Introduction The exceptional mechanical properties and low weight of CNTs make them potential lling materials in polymer composites. CNTs can improve the strength and stiffness of polymers, as well as add multifunctionality (such as electrical conductivity) to polymer based composite systems [15]. The CNTs are ideal reinforcing bers for composites due to their high aspect ratio and high in axis strength [6]. The CNTs have been synthesized by various methods e.g., electric arc discharge, laser evaporation and chemical vapor deposition (CVD) [79]. CVD appears to be a promising method to synthesize CNTs because of its low cost, simple conguration and high exibility in adjustable parameters for controlling the CNTs structures [10]. The CNTs intrinsically have excellent separation properties and can be used as potential llers in fabrication of nanocomposite membranes. Skoulidas et al. [11] studied atomic simulation of transport diffusivity of light gases in CNTs. They found that transport rates in CNTs are order of magnitude faster than zeolites. This high transport rates is the result of inherent smoothness of CNTs. The rst experiment of CNTs in membrane process was reported by Hinds et al. [12] in 2004. They incorporated an array of aligned

Corresponding author. Tel.: +98 21 77240496; fax: +98 21 77240495. E-mail address: [email protected] (T. Mohammadi). 0376-7388/$ see front matter 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2011.05.047

CNTs across a polymer lm to form a well-ordered nanoporous membrane structure. After that researchers considered to incorporate CNTs as ller in mixed matrix membranes (MMMs) [1321]. Tang et al. [22] studied water transport behavior of chitosan/CNT membranes. Their results showed that water ux could increase 4.6 times with adding 10 wt.% CNTs to the neat chitosan. Peng et al. [23] incorporated CNTs, which was dispersed by -cyclodextrin grinding method into PVA membranes. These membranes were applied to PV separation of benzene/cyclohexane mixtures, and showed excellent PV properties. Although CNTs have excellent separation, electrical and mechanical properties, synthesis of mixed matrix membranes using these materials has the following problems. First, during the CNTs synthesis, some impurities are produced and they must be puried by chemical or physical methods [24]. Second, the synthesized CNTs are generally close ended and they must be cut during the purication processes or separately [25,26]. Third, dispersion and dissolution of the as-grown CNTs are not usually sufcient in various organic solvents and different polymers and interaction of the interface between CNTs and polymer matrix is weak [14,27]. Therefore, synthesis of highly pure, open ended and functionalized CNTs is important to synthesize excellent MMMs. Uniform dispersion and improved CNTs/matrix wetting and adhesion are critical issues in fabrication of nanocomposites. The lack of afnity between CNTs and polymers inhibits load transfer from polymer to CNTs. It is difcult to fabricate homogeneous nanocomposites with CNTs. Spinks et al. [28] showed

552

Y. Shirazi et al. / Journal of Membrane Science 378 (2011) 551561

that mechanical properties of chitosan/CNTs is related to the dispersion properties of CNTs in polymer. They improved dispersion of CNTs by sonic agitation, and then centrifugation to remove their aggregates. Aroon et al. [20] investigated effect of chitosan as a functionalization agent on performance and separation properties of PI/MWCNTs MMMs. Their results showed that Helium permeability of the PI/as-grown CNTs MMMs decreased from that of the PI membrane because the CNTs were closed ended, thus acting as impermeable llers. Furthermore, they showed that addition of 1 wt.% chitosan-functionalized CNTs into PI membranes increased CO2 and CH4 permeabilities and also CO2 /CH4 selectivity [20]. Membrane-based PV is an energy-intensive method of separating liquid mixtures compared to conventional distillation or cryogenic separations particularly involving azeotropic and closely boiling mixtures due to its high separation factors and ux rates [29]. Hydrophilic polymeric materials have been widely investigated for PV dehydration, among which PVA is one of the most commonly studied, because of its superior hydrophilic property, good lm-forming ability and high chemical stability. However, PVA membranes have to be modied to restrict their swelling in aqueous solutions. So far, modication methods, such as crosslinking, blending and using inorganic llers have been applied by many researchers, and the water insolubility of PVA membranes was improved dramatically, or the swelling of the membranes was reduced [3032]. Many inorganic types of ller were incorporated into PVA membranes to make MMMs such as zeolites, fumed silica, TiO2 , CMS and graphite [3236]. No previous published study on PVA membranes loaded with CNTs to be used in PV separation of waterisopropanol mixtures could be found in the literature. In this work, CNTs was synthesized by CVD system. Since CNTs usually agglomerate due to van der waals force, they are extremely difcult to disperse in PVA solutions. In order to prevent agglomeration in the PVA solutions, the CNTs were chemically treated to produce stabilized dispersions. Characterization of CNTs before and after purication was carried out to conrm formation of highly pure CNTs. The dispersion properties of CNTs were improved by acid treatment and also sonication. The functionalized CNTs were incorporated into the PVA solutions to prepare PVACNTs nanocomposite membranes. Dispersion properties of CNTs in the membranes were investigated by FESEM to compare different loadings of the CNTs in the membranes. Furthermore, PVACNTs nanocomposite membranes were used in PV experiments for dehydration of isopropanol aqueous mixtures. 2. Experimental 2.1. Materials PVA (MW: 145,000) with a degree of hydrolysis of more than 98%, isopropanol (IPA), glutaraldehyde GA (50%) and hydrochloric acid were all purchased from Merck. Ferrocene as catalyst (purity 98%, B.D.H.) and cyclohexanol (purity 98%, Fluka) as carbon precursor were used to synthesize the CNTs. Deionized water, with a conductivity of 20 S/cm, was produced in the laboratory using Permionics pilot plant (Vadodara, India). 2.2. Synthesis of CNTs The CNTs was synthesized via chemical vapor deposition (CVD) of a gas mixture evaporated from the catalyst powder and the liquid carbon source. The CVD system consisted of a horizontal stainless steel reactor (70 cm long, 3.2 cm in diameter) housed in a one stage cylindrical furnace. A ask (steel container) containing

Fig. 1. Schematic diagram of the experimental CVD set-up.

reagents was connected to the reactor. The reagents were prepared by dissolving ferrocene (purity 98%, B.D.H.) in cyclohexanol with mass ratio of 1:20. A schematic diagram of the experimental set-up is shown in Fig. 1. Evaporation of the reagents was performed using an oil bath. Nitrogen was used as carrier gas connected to the reactor nearby the reagent inlet to carry the gas mixture of precursors towards the center of the furnace, where pyrolysis of the gases took place. The choice of cyclohexanol as carbon source was selected due to its hexagonal carbon ring structure and its low boiling point (160 C). Furthermore, it is a good solvent for ferrocene and this allows constitution of a homogeneous solution. Ferrocene was used as catalyst, since it is a good precursor for production of iron nanoparticles, which play the role of catalyst in formation of the CNTs [37,38]. The CNTs were characterized before and after purication. The morphology and structure of the CNTs were characterized by FESEM (Hitachi, S4160), TEM (Philips, CM200), Raman (WITec, TS150) and FTIR spectroscopy (SHIMADZU, IRsolution 8400S). 2.3. Purication of CNTs For purication, the CNTs were placed in the reactor. Air with ow rate of 250 cm3 min1 was injected into the reactor. Temperature of the reactor was raised until 442 C and stayed at that temperature for 30 min. The reactor was then switched off and cooled until room temperature and the CNTs were taken from the reactor. Amorphous carbon was removed after oxidation. To omit the metal catalyst, the oxidized CNTs are soaked in nitric acid solution and reuxed at 120 C for 3 h and then the CNTs were centrifuged and washed with distilled water for many times until a pH of 7 was obtained. Reuxing with the acid solution resulted in dissolving the metal catalyst in the acid solution and consequently removing the metal catalyst from the CNTs. 2.4. Membrane preparation PVACNTs nanocomposite membranes were synthesized by solution casting and solvent evaporation technique. PVA was dissolved in deionized water at 90 C then stirred for a period of time to form a homogeneous solution. The solution was ltered through a glass ber lter to remove insoluble impurities; afterward the solution was vacuumed to remove any bubbles. Simultaneously the acid treated CNTs were dissolved in deionized water, and then exposed to ultrasonication for 4 h. Fig. 2 represents the membrane preparation procedure. In situ crosslinking was performed by adding GA and hydrochloric acid to the above solution and further stirring.

Y. Shirazi et al. / Journal of Membrane Science 378 (2011) 551561

553

Fig. 2. Membrane preparation procedure.

The resulting solution was cast on a clean glass plate in a cleaning room to prepare the membrane. The membrane was dried at ambient temperature and peeled off from the glass plate, then annealed. The membrane thickness measured using a micrometer screw gage was around 25 1.0 m. In the synthesized nanocomposite membranes, PVA-X, X; the mass ratio of CNTs to PVA was varied at 0, 0.5, 1, 2 and 4%. 2.5. PV experiments The prepared membranes were evaluated in PV separation using a setup as shown in Fig. 3. Mixtures of IPAwater were prepared and hold in a tank as feed. The feed solution was circulated by a pump (Talow magnet pump) through the membrane cell. Effective membrane area was 27 cm2 with a at sheet shape. The downstream pressure was maintained at about 1 mbar using a vacuum pump (Vac Torr 25, USA), while the feed pressure was atmospheric. In fact, partial vacuum was applied to the permeate side

as a driving force for water to evaporate easily. The permeate vapor was collected in a cold (liquid nitrogen) trap to freeze the permeate. The collected permeate sample was weighed to determine the ux. To achieve the steady state ux, the membrane was swelled in the feed for 2 h before performing the PV tests. To assess the experimental errors, each test was repeated three times and the average results were reported. Concentrations of the feed and permeate were measured using Abbes refractometer (2 W AJ, Japan) by comparing them with a standard graph of refractive index versus feed mixture composition calibration curve. Alternatively, to ascertain the exact compositions, the analyses of feed and permeate samples were performed by Gas Chromatography provided with a thermal conductivity detector (TCD) equipped with a DEGS or Tenax packed column of 1/8 in. i.d. having 2 m length. The GC response was calibrated for the column and for known compositions of isopropanol/water mixture. The results of GC analysis were agreed with the refractometer measurements.

Fig. 3. Schematic diagram of the PV apparatus: (1) feed tank; (2) cooling coil; (3) thermo couple; (4) heater; (5) feed pump; (6,7) valve; (8) rotor ow meter; (9) membrane cell; (10) membrane; (11) pressure gage; (12) vacuum regulator; (13) permeate collection trap; (14) liquid nitrogen trap; (15) vacuum pump.

554

Y. Shirazi et al. / Journal of Membrane Science 378 (2011) 551561

Permeation properties of the membranes were characterized by PV selectivity (PV ), total permeation ux (JP ) and PV separation index (PSI) using the following equations, respectively: PV = JP = yA /yB xA /xB (1) (2) (3)

WP At

PSI = JP (PV 1)

where yA and yB are mass fractions of water and IPA in the permeate, respectively and xA and xB are mass fraction of water and IPA in the feed, respectively. The permeate ux, JP (kg/m2 h), was calculated using WP , the permeate mass (kg); A is the effective membrane area (m2 ) and t is the permeation time (h). 2.5.1. Swelling studies The extent of membrane swelling was determined by the sorption experiments. These experiments are helpful to investigate the interactions of the membranes with the liquid penetrants. The swelling experiments were performed in 20 wt.% water containing feed mixtures. The PVA-X nanocomposite membranes (dimensions of 3 cm 3 cm) were dried completely at 80 C for 8 h and weighed. Then, these membranes were immersed in the waterIPA mixture in a sealed vessel at 30 C for 3 days to allow them to reach their equilibrium swelling. The swollen membranes were weighed using a digital microbalance (Sartorius, TE214S) sensitive to 0.1 mg, as quickly as possible after wiped with tissue papers. Each run was performed at least three times, and the results were averaged. The degree of swelling (DS) of the membrane was calculated by DS (%) = Ws Wd Wd 100 (4)

where Ws and Wd are mass of the swollen and the dry membranes, respectively. 2.6. Characterization Field emission scanning electron microscope (FESEM) was used for analysis of the CNTs and the PVA-X nanocomposite membranes morphologies. The PVA-X nanocomposite membranes were cryogenically fractured in liquid nitrogen to examine cross sectional morphology of the membranes. The surface and cross section of membranes were coated with gold for SEM observations. Nanostructure of the CNTs was determined by transmission electron microscope (TEM). Raman spectroscopy was used to investigate the structural properties of the CNTs before and after purication. FTIR was also used to investigate the functional groups of the CNTs and the PVA-X nanocomposite membranes. In order to study the dispersibility behavior of different CNTs samples, 10 mg of the CNTs before and after purication was dispersed in 20.0 ml of water with the aid of an ultrasonic homogenizing system (Heilscher, UP400S) for 5 min. The dispersion was characterized by optical microscopy (Hund WETZLAR, MEDICUS HF). 3. Results and discussion 3.1. Characterization of CNTs 3.1.1. The as-grown CNTs Fig. 4 illustrates the FESEM and TEM image of CNTs synthesized by the CVD method. As observed, CNTs have appropriate density and are entangled. In the TEM image (Fig. 4b), black spots indicate deactivated catalysts. Also TGA/DTA results, as presented in our previous work [39], revealed that amount of amorphous carbon and deactivated catalysts are low in the synthesized CNTs. Obviously,

Fig. 4. (a) FESEM and (b) TEM images of as-grown CNTs.

one can observe that the synthesized CNTs are free of amourphous carbon and also deactivated metalic partilces are signicantly low. This can be attributed to presence of oxygen atoms in cyclohexanol [39]. Oxygen atoms in cyclohexanol can oxidize amorphous carbon and consequently the synthesized CNTs are free of amorphous carbon. This conclusion is consistent with the results of experiments in which adding 2 mol% of acetic acid to carbon precursor (xylene) declined amorphous carbon [40]. Generally, deposition of amorphous carbon on the catalyst surface increases formation of amorphous carbon. However, using cyclohexanol as carbon precursor reduces the formation of amorphous carbon. Presence of oxygen atoms in cyclohexanol causes the in situ oxidation of amorphous carbon [39]. This reduction of amorphous carbon prevents the catalyst surface to be passive. TEM image of a CNT (Fig. 4b) conrms that the CNTs are multi-wall and also have highly pure structure. As observed, the outer and inner diameter of CNTs is around 60 and 13 nm, respectively. 3.1.2. The puried CNTs Purication of the CNTs is very important in their structural modication. Oxidation and acid treatment are the most common methods for the CNTs purication [41]. Although using cyclohexanol as carbon precursor reduces amounts of amorphous carbon and deactivated catalyst tremendously, there are a few impurities which should be removed to obtain highly pure CNTs. Oxidation can remove just amorphous carbon and is not effective for omitting metal catalyst. Acid treatment is an effective method for purication of the CNTs and opening their tips (removing deactivated catalyst particles).

Y. Shirazi et al. / Journal of Membrane Science 378 (2011) 551561

555

Fig. 5. TEM images of (a) as-grown CNTs (b) CNTs treated by 8 M HNO3 .

Fig. 5 illustrates TEM images of the CNTs before and after acid treatment by HNO3 8 M. As can be observed, acid treatment can remove metallic particles trapped in the CNTs walls. As reported before, oxidation of CNTs can remove their encapsulated catalyst particles and as a result open CNTs inner cavities [42]. It is well known that CNTs inner cavities have remarkable smooth surfaces and as a result water molecules can slide through them readily [12,43]. From raw soot, amorphous carbons and carbon particles can be eliminated more easily than CNTs due to their faster oxidation reaction rates. The amorphous carbons are multi-shell structure and have many hanging bonds with high energy in their fringe, and these hanging bonds can be easily oxidized. Therefore, highly pure CNTs can be obtained after oxidation and acid treatment. Fig. 6 represents the Raman spectra of as-grown and acid treated CNTs. The D and G bands originate from defect and graphite, respectively. Ratio of the defect bands to the graphite is presented by ID /IG parameter. However, acid treatment with HNO3 8 M can reduce amorphous carbon without damaging the structure of CNTs significantly. According to the Raman spectra, acid treatment with HNO3 8 M can reduce D band which is attributed to acid treatment which can oxidize amorphous carbon and also remove catalyst particles. The values of ID /IG parameter which presents the ratio of defect to graphite, for the as-grown and the acid treated CNTs are 0.22 and 0.17, respectively. 3.1.3. The functionalized CNTs As mentioned, application of CNTs is facing several challenges which are mainly due to the hydrophobic nature of CNTs and formation of long bundles stabilized by van der waals interactions

between CNTs. Functionalization of CNTs is an effective way to prevent CNTs aggregation which can improve afnity between CNTs and solvent. Acid treatment not only increases purity of CNTs, but also improves dispersion and stabilization of CNTs within polymer matrixes. Fig. 7 illustrates the FTIR results of as-grown and acid treated CNTs. The FTIR spectrum of as-grown CNTs is presented in Fig. 7a. As observed, the as-grown CNTs show a peak at 1535 cm1 which is assigned to the C C bond. The weak peak at approximately 1726 cm1 is attributed to the C O bonds [44]. The FTIR results reveal that the as-grown CNTs using cyclohexanol as carbon precursor have weak functional groups and as a result slight hydrophilic properties. The FTIR spectrum of CNTs treated by HNO3 8 M presents two peaks at approximately 1726 and 3425 cm1 which are attributed to acidic carbonyl and hydroxyl groups, respectively [45,46]. This indicates formation of the carboxyl groups on the surface of CNTs. The peak at 3425 cm1 is assigned to OH bonds, and signicant sharp peak at 3425 cm1 indicates the signicant formation of OH bonds and this consequently causes the more hydrophilicity after acid treatment. The peak at 1282 cm1 is characterized as the acidic CO bonds [44]. Fig. 8 presents the dispersion of CNTs in water aided with sonication. Fig. 8a illustrates the dispersion of as-grown CNTs after 5 min sonication. The inherent hydrophobic surface of CNTs and presence of numerous interactions between the CNTs make dispersion of CNTs in water very difcult and exposure to sonication does not improve their dispersability [44]. Fig. 8b presents the dispersion of as-grown CNTs after a week. As observed, as-grown CNTs tend

Fig. 6. Raman spectra of as-grown CNTs and CNTs treated by 8 M HNO3 .

Fig. 7. FTIR spectra of (a) as-grown CNTs and (b) CNTs treated by 8 M HNO3 .

556

Y. Shirazi et al. / Journal of Membrane Science 378 (2011) 551561

Fig. 8. Optical micrographs showing the dispersion of (a) as-grown CNTs with 5 min sonication (b) as-grown CNTs with 5 min sonication after a week (c) acid treated CNTs with 5 min sonication (d) acid treated CNTs with 5 min sonication after a month.

to aggregate, and stopping sonication starts their agglomeration. However, the nitric acid treated CNTs shows appropriate dispersion in water due to the presence of OH groups on the outer surface of CNTs which form hydrogen bonding with water molecules (Fig. 8c). Furthermore, formation of the C O and CO bonds during acid treatment causes more interaction with water molecules and this results in good dispersion of the CNTs in water. Fig. 8d presents the dispersion of nitric acid treated CNTs after a month. The results reveal that the nitric acid treated CNTs can be dispersed in water perfectly, and for more than a month, the dispersion is still stable. 3.2. Characterization of PVA nanocomposite membranes 3.2.1. Morphology of PVA-X membranes Dispersed inorganic phase as well as continuous organic phase affects nanocomposite membrane separation properties and morphology. In order to investigate the dispersion of functionalized CNTs in nanocomposite membranes, FESEM inspections were carried out. Fig. 9 illustrates the FESEM images of surface morphology of PVA-X nanocomposite membranes at various CNTs loading. Obviously, PVA-0 is free of CNTs particles and by increasing CNTs loading the number of lighting dots which present the CNTs tips increase. By comparing FESEM images of different PVA-X nanocomposites membranes, it can be concluded that membranes containing 2 wt.% CNTs exhibits better dispersion of CNTs. Increasing CNTs loading more than 2 wt.% can agglomerate the CNTs. It is known that the CNTs like to agglomerate because of van der waals forces between them [47]. As a result, increasing CNTs loading enhances probable agglomeration of CNTs. In Fig. 9e red dotted cycles exhibit the agglomerated CNTs. (For interpretation of the references to color in this text, the reader is referred to the web version of the article.)

To further investigate the internal structure of the nanocomposite membranes, cross section FESEM observations of PVA-X nanocomposites are shown in Fig. 10. The bright dots which are illustrated by arrows in Fig. 10a are the CNTs tips. The well adhesion between the acid treated CNTs and the PVA matrix is due to high compatibility between the hydrophilic CNTs and the hydrophilic PVA matrix. Therefore, no non-selective defects (voids) can be found at the interface between the PVA and the CNTs additives. Red dotted square are shown in Fig. 10b demonstrate that some belt like CNTs were observed which interconnects two polymer lumps. (For interpretation of the references to color in this text, the reader is referred to the web version of the article.) Moreover, the middle part of the tube is thinner than the two ends adhered to the two lumps of polymeric matrix. This also indicates that the interaction of CNTs with PVA matrix is very strong, the same result claimed by Chen et al. [48]. FESEM image of enlarged morphology of selected region is presented in Fig. 10c to show that the ber shape materials in the polymer matrix can be referred to the CNTs or a bundle of CNTs. On the other hands, the diameter of these bers is around 60 nm which is compatible with the CNTs diameter. As the TEM image of pristine CNTs illustrates (Fig. 4b) the outer diameter of CNTs is around 60 nm. 30 CNTs particles of PVA-2 and PVA-4 nanocomposite membranes were selected randomly to measure the outer diameter of CNTs. As observed in Fig. 11a, the mean of outer diameter of CNTs for PVA-2 nanocomposite membranes is 58 nm with standard deviation of 1.85. This indicates that the size of CNTs particles in the PVA matrix is close. Comparing the outer diameter of CNTs in the PVA matrix with that of the as-grown CNTs, one can nd that the outer diameter of CNTs in the PVA matrix is close to that of the as-grown CNTs. However, in the PVA-4 membrane, the mean of outer diameter of CNTs is 175 nm (Fig. 11b). The results conrm that in PVA-4

Y. Shirazi et al. / Journal of Membrane Science 378 (2011) 551561

557

Fig. 9. FESEM surface images of (a) PVA-0 (b) PVA-0.5 (c) PVA-1 (d) PVA-2 and (e) PVA-4.

nanocomposite membranes CNTs tend to agglomerate and make the bundles of CNTs. 3.2.2. FTIR results of PVA-X membrane The ATR-IR spectra of PVA-0 and PVA-2 nanocomposite membranes in the range of 5002000 cm1 are presented in Fig. 12. The absorption peak at 10001100 cm1 is assigned to the stretching vibration of CO and COC groups in PVA [49]. The FTIR spectra of acid treated CNTs showed that the CNTs is functionalized by carboxyl and hydroxyl groups, and this helps the CNTs to be well dispersed in water and also improves the interaction between the PVA matrix and CNTs. The bands of 16401685 cm1

region are likely due to conjugation of C O with C C bonds or interaction between localized C C bonds and carboxylic acids and ketones [14]. As observed, PVA-0 membrane contains C O bonds and as mentioned, functionalized CNTs also contains C O bonds, this conrms that PVA can interact strongly with functionalized CNTs. The FTIR spectra of PVA-0 and PVA-2 are similar. This can be attributed to the fact that the functional groups of CNTs are the same as those of PVA membrane. However, the intensity of absorption at 17001750 cm1 which is assigned to the stretching vibration of C O increases by adding CNTs to PVA. This can be attributed to the carbonyl groups of acid treated CNTs.

558

Y. Shirazi et al. / Journal of Membrane Science 378 (2011) 551561

Fig. 11. Histogram showing the outer diameter of 30 CNTs in PVA matrix; (a) PVA-2 (b) PVA-4.

results showed that adding 2 wt.% CNTs into the PVA membrane increases the modulus signicantly from 2.1 to 3.09 GPa. This can be due to the high Youngs modulus of CNTs and also due to the strong interaction between PVA and CNTs. It is well known that CNTs can increase the mechanical properties of PVA as well as other polymers [18,28,48]. Rigidication consequently decreases mobility and free volume of the polymer chains. Moreover, presence of the CNTs in the PVA matrix can reduce the surface hydrophilicity

Fig. 10. FESEM cross sectional images of nanocomposite membranes (a) PVA-1 (b) PVA-2 (c) enlarged morphology of selected region in PVA-2 indicating diameter of CNTs into PVA matrix.

3.2.3. Swelling results of PVA-X membrane In PV, membrane swelling controls transport of permeating molecules under the inuence of chemical potential gradient. Degree of swelling (DS) of various PVA-X membranes in 20 wt.% water in IPA mixture at 30 C are illustrated in Fig. 13. As observed, the PVA-0 membrane shows higher DS compared with the PVACNTs nanocomposite membranes. Obviously, increasing CNTs content in PVA matrix is responsible for the DS reduction. This can be attributed to the fact that the CNTs increase rigidication [50]. Tensile properties were also measured using universal testing machine (tensile machine model: STM 50). The results showed that incorporating CNTs into the PVA matrix increases strength and Youngs modulus of the PVA-X membranes. The

Fig. 12. The FTIR spectra of PVA-0 and PVA-2 membranes.

Y. Shirazi et al. / Journal of Membrane Science 378 (2011) 551561

559

Fig. 13. Degree of swelling (DS) of different membranes in 20 wt.% water in IPA mixture at 30 C.

Fig. 15. Effects of CNTs loading on PSI of PVA-X nanocomposite membranes (10 wt.% IPA aqueous solution at 30 C).

of the nanocomposite membranes which is compatible with the less DS of PVA-X nanocomposite membranes [50]. PVA in crystal is more difcult to be dissolved in water than its amorphous state because of stronger intermolecular hydrogen bonds among the PVA molecules in the crystal state [50,51]. CNTs can act as nucleation sites and this results in the growing PVA crystallinity [48,50,51]. Increasing the PVA nanocomposite crystallinity affects the surface hydrophilicity of the PVA nanocomposite membranes. However, DS increases from PVA-2 to PVA-4 nanocomposite membranes due to agglomeration of CNTs in the membranes. Aggregation of the CNTs decreases the compactness of PVA membrane, and also the reduction of polymer chain packing increases free volume of the polymer membrane [36,52]. As a matter of fact, the aggregated CNTs reduce compacting of the polymer chains and the penetrants can permeate through the membrane readily. 3.3. PV results 3.3.1. Effect of CNTs on PV results Fig. 14 shows selectivity and permeation ux of PVA-X membranes versus CNTs loading. Operating condition was xed at 10 wt.% water in IPA mixture and feed temperature of 30 C for all the PVA-X membranes. The results show that there is a considerable increase in selectivity for the PVA-X nanocomposite membranes when compared with the pristine PVA membranes. This can be attributed to the fact that the CNTs in the PVA matrix acts as reinforcing bridge elements, as observed in Fig. 10b, and thus, it makes PVA chains more rigid, thereby giving a reduced DS with a simultaneous ux reduction. Moreover, rigidication can reduce the polymer free volume and the results can be higher selectivity and less permeation ux. The pristine PVA membrane with the highest DS also has the highest permeation ux, as observed.

Obviously, permeation uxes of the different nanocomposite membranes follow similar trends with the DS results. This means that the membranes with higher DS, exhibit higher permeation ux. There are two effects which inuence permeation ux of the nanocomposite membranes. First, nanochannels of CNTs increase permeation of water [52]. It is known that CNT nanochannels can increase transport of water because of their inherent smoothness [12]. Second, present of CNTs in the polymer matrix can increase polymer rigidication which causes free volume reduction and this reduces permeability. It is worthy to mention that adding 0.5 wt.% of CNTs to PVA decreases permeation ux signicantly. This can be attributed to free volume reduction of the membranes. It is well known that molecular transport through dense polymeric membranes is inuenced by the presence of free volume [53]. However, increasing CNTs content up to 2 wt.% decreases permeation ux slightly. This can be due to transport of water molecules through the nanochannels of the opened tip CNTs. As a matter of fact, the CNTs give a pathway for transport compared to solid diffusion through pure polymer [18]. This may reduce the effect of free volume reduction on permeation ux. As observed, in PVA-4 nanocomposite membrane, permeation ux increases. This can be due to agglomeration of the CNTs in the PVA matrix. The agglomerated CNTs increase free volume and consequently penetrants can permeate through the membranes more readily. The DS results also conrm the higher permeation ux for PVA-4 nanocomposite membrane. Furthermore, as observed, in PVA-4 nanocomposite membrane, selectivity decreases signicantly because of the nonselective voids which are the result of the CNTs agglomeration. PV separation index (PSI) of PVA-X nanocomposite membranes is illustrated in Fig. 15. As can be observed, PSI values increase with increasing the CNTs loading up to 2 (wt.%) and then decrease. The reduction of PSI value for PVA-4 nanocomposite membrane can be due to the sharp reduction of selectivity. As mentioned this is due to CNTs agglomeration. As a result, PVA-2 nanocomposite membrane can be recommended as an optimized nanocomposite membrane. 3.3.2. Comparison of synthesized membrane performance with the literature Table 1 compares performance of the nanocomposite membrane with that of reported membranes. The results show that incorporation of CNTs into PVA can exert a great effect on water selectivity and the synthesized membrane performs better than these reported by others. Although permeation ux of PVACNTs nanocomposite membranes is lower than some other PVA membranes, water selectivity of PVACNTs nanocomposite membranes is much higher than that of other nanocomposite membranes. The PSI values for different PVA membranes were also listed in Table 1.

Fig. 14. Effect of CNTs loading on PV performance of PVA-X nanocomposite membranes in 10 wt.% water in IPA mixture at 30 C.

560

Y. Shirazi et al. / Journal of Membrane Science 378 (2011) 551561 [13] G. Nechifor, S.I. Voicu, A.C. Nechifor, S. Garea, Nanostructured hybrid membrane polysulfonecarbon nanotubes for hemodialysis, Desalination 241 (2009) 342348. [14] J.-H. Choi, J. Jegal, W.-N. Kim, Fabrication and characterization of multi-walled carbon nanotubes/polymer blend membranes, Journal of Membrane Science 284 (2006) 406415. [15] S. Kim, T.W. Pechar, E. Marand, Poly(imide siloxane) and carbon nanotube mixed matrix membranes for gas separation, Desalination 192 (2006) 330 339. [16] H.-H. Tseng, I.A. Kumar, T.-H. Weng, C.-Y. Lu, M.-Y. Wey, Preparation and characterization of carbon molecular sieve membranes for gas separation the effect of incorporated multi-wall carbon nanotubes, Desalination 240 (2009) 4045. [17] P.S. Rao, M.-Y. Wey, H.-H. Tseng, I.A. Kumar, T.-H. Weng, A comparison of carbon/nanotube molecular sieve membranes with polymer blend carbon molecular sieve membranes for the gas permeation application, Microporous and Mesoporous Materials 113 (2008) 499510. [18] J. Deng, X. Zhang, K. Wang, H. Zou, Q. Zhang, Q. Fu, Synthesis and properties of poly(ether urethane) membranes lled with isophorone diisocyanategrafted carbon nanotubes, Journal of Membrane Science 288 (2007) 261 267. [19] S. Kim, L. Chen, J.K. Johnson, E. Marand, Polysulfone and functionalized carbon nanotube mixed matrix membranes for gas separation: theory and experiment, Journal of Membrane Science 294 (2007) 147158. [20] M.A. Aroon, A.F. Ismail, M.M. Montazer-Rahmati, T. Matsuura, Effect of chitosan as a functionalization agent on the performance and separation properties of polyimide/multi-walled carbon nanotubes mixed matrix at sheet membranes, Journal of Membrane Science 364 (2010) 309317. [21] S.S. Hosseini, T.S. Chung, Carbon membranes from blends of PBI and polyimides for N2 /CH4 and CO2 /CH4 separation and hydrogen purication, Journal of Membrane Science 328 (2009) 174185. [22] C. Tang, Q. Zhang, K. Wang, Q. Fu, C. Zhang, Water transport behavior of chitosan porous membranes containing multi-walled carbon nanotubes (MWNTs), Journal of Membrane Science 337 (2009) 240247. [23] F. Peng, Z. Jiang, C. Hu, Y. Wang, L. Lu, H. Wu, Pervaporation of benzene/cyclohexane mixtures through poly(vinyl alcohol) membranes with and without [beta]-cyclodextrin, Desalination 193 (2006) 182192. [24] P.X. Hou, S. Bai, Q.H. Yang, C. Liu, H.M. Cheng, Multi-step purication of carbon nanotubes, Carbon 40 (2002) 8185. [25] Y. Liu, L. Gao, J. Sun, S. Zheng, L. Jiang, Y. Wang, H. Kajiura, Y. Li, K. Noda, A multistep strategy for cutting and purication of single-walled carbon nanotubes, Carbon 45 (2007) 19721978. [26] Y. Wang, L. Gao, J. Sun, Y. Liu, S. Zheng, H. Kajiura, Y. Li, K. Noda, An integrated route for purication, cutting and dispersion of single-walled carbon nanotubes, Chemical Physics Letters 432 (2006) 205208. [27] R Andrews, M.C. Weisenberger, Carbon nanotube polymer composites, Current Opinion in Solid State and Materials Science 8 (2004) 3137. [28] G.M. Spinks, S.R. Shin, G.G. Wallace, P.G. Whitten, S.I. Kim, S.J. Kim, Mechanical properties of chitosan/CNT microbers obtained with improved dispersion, Sensors and Actuators B: Chemical 115 (2006) 678684. [29] V. Dubey, L.K. Pandey, C. Saxena, Pervaporative separation of ethanol/water azeotrope using a novel chitosan-impregnated bacterial cellulose membrane and chitosan-poly(vinyl alcohol) blends, Journal of Membrane Science 251 (2005) 131136. [30] Y. Zhu, S. Xia, G. Liu, W. Jin, Preparation of ceramic-supported poly(vinyl alcohol)chitosan composite membranes and their applications in pervaporation dehydration of organic/water mixtures, Journal of Membrane Science 349 (2010) 341348. [31] M.N. Hyder, R.Y.M. Huang, P. Chen, Correlation of physicochemical characteristics with pervaporation performance of poly(vinyl alcohol) membranes, Journal of Membrane Science 283 (2006) 281290. [32] Z. Huang, H.-m. Guan, W.l. Tan, X.-Y. Qiao, S. Kulprathipanja, Pervaporation study of aqueous ethanol solution through zeolite-incorporated multilayer poly(vinyl alcohol) membranes: effect of zeolites, Journal of Membrane Science 276 (2006) 260271. [33] S.J. Lue, D.T. Lee, J.Y. Chen, C.H. Chiu, C.C. Hu, Y.C. Jean, J.Y. Lai, Diffusivity enhancement of water vapor in poly(vinyl alcohol)-fumed silica nano-composite membranes: correlation with polymer crystallinity and freevolume properties, Journal of Membrane Science 325 (2008) 831839. [34] M. Sairam, M.B. Patil, R.S. Veerapur, S.A. Patil, T.M. Aminabhavi, Novel dense poly(vinyl alcohol)-TiO2 mixed matrix membranes for pervaporation separation of waterisopropanol mixtures at 30 C, Journal of Membrane Science 281 (2006) 95102. [35] H. Sun, L. Lu, F. Peng, H. Wu, Z. Jiang, Pervaporation of benzene/cyclohexane mixtures through CMS-lled poly(vinyl alcohol) membranes, Separation and Purication Technology 52 (2006) 203208. [36] F. Peng, L. Lu, C. Hu, H. Wu, Z. Jiang, Signicant increase of permeation ux and selectivity of poly(vinyl alcohol) membranes by incorporation of crystalline ake graphite, Journal of Membrane Science 259 (2005) 6573. [37] B.C. Satishkumar, A. Govindaraj, C.N.R. Rao, Bundles of aligned carbon nanotubes obtained by the pyrolysis of ferrocene-hydrocarbon mixtures: role of the metal nanoparticles produced in situ, Chemical Physics Letters 307 (1999) 158162. [38] R. Andrews, D. Jacques, A.M. Rao, F. Derbyshire, D. Qian, X. Fan, E.C. Dickey, J. Chen, Continuous production of aligned carbon nanotubes: a step closer to commercial realization, Chemical Physics Letters 303 (1999) 467474.

Table 1 Comparison of the synthesized membrane performance with the literature (operating condition: 10 wt.% water in IPA mixture at 30 C). Membrane PVA/APTEOS PVA/(APTEOS + TEOS) PVA/KA PVA/NaA PVA/CaA PVA/NaX PVAFe PVAsilicalite PVA-1 PVA-2 Flux (kg/m2 h) 0.0265 0.06 0.179 0.183 0.193 0.216 0.084 0.084 0.096 0.079 Selectivity 1580 891 410 328 233 233 144 1295 817 1794 PSI 41.84 53.4 73.21 59.84 44.77 50.11 12.01 108.67 78.79 141.32 Reference [54] [55] [56]

[57] [58] Present work

One can observe that the PSI value of PVACNTs is satised by comparing with that of other PVA membranes. 4. Conclusion Cyclohexanol as carbon source and ferrocene as catalyst was used to synthesize CNTs via CVD method. As-grown CNTs were characterized using SEM, TEM and Raman spectra. The results showed that the CNTs have high purity and free of amorphous carbon due to in situ oxidation of amorphous carbon in presence of oxygen atoms of cyclohexanol. Nitric acid was used to functionalize the as-grown CNTs. As CNTs are intrinsically hydrophobic, acid treatment can improve their hydrophilic property and helps them to better disperse in the PVA matrix. FESEM observation conrms that the acid treated CNTs interact with PVA well. Moreover, the results of FESEM indicate when the CNTs loading increases more than 2 wt.% in the PVA matrix, the CNTs start to agglomerate because of van der waals interaction between them. The PV results indicate that the agglomerated CNTs in the PVA matrix decrease water selectivity and diversely increase permeation ux of nanocomposite membranes. The swelling results of PVA-X nanocomposite membranes indicate that incorporating the CNTs in the PVA matrix reduces swelling of the membranes and improves water selectivity signicantly. References
[1] J. Sandler, M.S.P. Shaffer, T. Prasse, W. Bauhofer, K. Schulte, A.H. Windle, Development of a dispersion process for carbon nanotubes in an epoxy matrix and the resulting electrical properties, Polymer 40 (1999) 59675971. [2] E. Kymakis, G.A.J. Amaratunga, Single-wall carbon nanotube/conjugated polymer photovoltaic devices, Applied Physics Letters 80 (2002) 112114. [3] P.A.O. Muisener, L. Clayton, J. DAngelo, J.P. Harmon, A.K. Sikder, A. Kumar, A.M. Cassell, M. Meyyappan, Effects of gamma radiation on poly(methyl methacrylate)/single-wall nanotube composites, Journal of Materials Research 17 (2002) 25072513. [4] R.H. Baughman, A.A. Zakhidov, W.A. de Heer, Carbon nanotubes the route toward applications, Science 297 (2002) 787792. [5] M.S. Han, Y.K. Lee, H.S. Lee, C.H. Yun, W.N. Kim, Electrical, morphological and rheological properties of carbon nanotube composites with polyethylene and poly(phenylene sulde) by melt mixing, Chemical Engineering Science 64 (2009) 46494656. [6] P. Calvert, Nanotube composites a recipe for strength, Nature 399 (1999) 210211. [7] T.W. Ebbesen, P.M. Ajayan, Large-scale synthesis of carbon nanotubes, Nature 358 (1992) 220222. [8] T. Guo, P. Nikolaev, A.G. Rinzler, D. Tomanek, D.T. Colbert, R.E. Smalley, Self-assembly of tubular fullerenes, Journal of Physical Chemistry 99 (1995) 1069410697. [9] J. Kong, A.M. Cassell, H.J. Dai, Chemical vapor deposition of methane for singlewalled carbon nanotubes, Chemical Physics Letters 292 (1998) 567574. [10] B. Chen, P. Wu, Aligned carbon nanotubes by catalytic decomposition of C2 H2 over NiCr alloy, Carbon 43 (2005) 31723177. [11] A.I. Skoulidas, D.M. Ackerman, J.K. Johnson, D.S. Sholl, Rapid transport of gases in carbon nanotubes, Physical Review Letters 89 (2002) 185901. [12] B.J. Hinds, N. Chopra, T. Rantell, R. Andrews, V. Gavalas, L.G. Bachas, Aligned multiwalled carbon nanotube membranes, Science 303 (2004) 6265.

Y. Shirazi et al. / Journal of Membrane Science 378 (2011) 551561 [39] Y. Shirazi, M.A. Toghy, T. Mohammadi, A. Pak, Effects of different carbon precursors on synthesis of multiwall carbon nanotubes: purication and functionalization, Applied Surface Science 257 (2011) 73597367. [40] A.A. Khodadadi, M. Maghrebi, Y. Mortazavi, A. Sane, Y. Shirazi, M. Rahimi, Z. Tsakadze, S. Mhaisalkar, Effect of acetic acid on amorphous carbon removal along a CNT synthesis reactor, Journal of Optoelectronics and Advanced Materials 11 (2009) 16111617. [41] C.N.R. Rao, A. Govindaraj, G. Gundiah, S.R.C. Vivekchand, Nanotubes and nanowires, Chemical Engineering Science 59 (2004) 46654671. [42] W. Mi, Y.S. Lin, Y. Li, Vertically aligned carbon nanotube membranes on macroporous alumina supports, Journal of Membrane Science 304 (2007) 17. [43] A. Noy, H.G. Park, F. Fornasiero, J.K. Holt, C.P. Grigoropoulos, O. Bakajin, Nanouidics in carbon nanotubes, Nano Today 2 (2007) 2229. [44] M.V. Naseh, A.A. Khodadadi, Y. Mortazavi, F. Pourfayaz, O. Alizadeh, M. Maghrebi, Fast and clean functionalization of carbon nanotubes by dielectric barrier discharge plasma in air compared to acid treatment, Carbon 48 (2010) 13691379. [45] S.M. Chen, W.M. Shen, G.Z. Wu, D.Y. Chen, M. Jiang, A new approach to the functionalization of single-walled carbon nanotubes with both alkyl and carboxyl groups, Chemical Physics Letters 402 (2005) 312317. [46] Y.H. Yan, J. Cui, M.B. Chan-Park, X. Wang, Q.Y. Wu, Systematic studies of covalent functionalization of carbon nanotubes via argon plasma-assisted UV grafting, Nanotechnology 18 (2007) 115712. [47] J.U. Park, S. Cho, K.S. Cho, K.H. Ahn, S.J. Lee, Effective in situ preparation and characteristics of polystyrenegrafted carbon nanotube composites, Korea Australia Rheology Journal 17 (2005) 4145. [48] W. Chen, X.M. Tao, P. Xue, X.Y. Cheng, Enhanced mechanical properties and morphological characterizations of poly(vinyl alcohol)carbon nanotube composite lms, Applied Surface Science 252 (2005) 14041409. [49] R. Guo, X. Ma, C. Hu, Z. Jiang, Novel PVAsilica nanocomposite membrane for pervaporative dehydration of ethylene glycol aqueous solution, Polymer 48 (2007) 29392945.

561

[50] M. Naebe, T. Lin, W. Tian, L. Dai, X. Wang, Effects of MWNT nanollers on structures and properties of PVA electrospun nanobres, Nanotechnology 18 (2007) 225605. [51] O. Probst, E.M. Moore, D.E. Resasco, B.P. Grady, Nucleation of polyvinyl alcohol crystallization by single-walled carbon nanotubes, Polymer 45 (2004) 44374443. [52] Y.C. Liu, J.W. Shen, K.E. Gubbins, J.D. Moore, T. Wu, Q. Wang, Diffusion dynamics of water controlled by topology of potential energy surface inside carbon nanotubes, Physical Review B Condensed Matter and Materials Physics 77 (2008) 125438. [53] B. Vijaya Kumar Naidu, M. Sairam, K.V.S.N. Raju, T.M. Aminabhavi, Pervaporation separation of water+isopropanol mixtures using novel nanocomposite membranes of poly(vinyl alcohol) and polyaniline, Journal of Membrane Science 260 (2005) 142155. [54] Q.G. Zhang, Q.L. Liu, Y. Chen, J.H. Chen, Dehydration of isopropanol by novel poly(vinyl alcohol)silicone hybrid membranes, Industrial and Engineering Chemistry Research 46 (2007) 913920. [55] S. Razavi, A. Sabetghadam, T. Mohammadi, Dehydration of isopropanol by PVAAPTEOS/TEOS nanocomposite membranes, Chemical Engineering Research and Design 89 (2011) 148155. [56] Z. Gao, Y. Yue, W. Li, Application of zeolite-lled pervaporation membrane, Zeolites 16 (1996) 7074. [57] M. Sairam, B.V.K. Naidu, S.K. Nataraj, B. Sreedhar, T.M. Aminabhavi, Poly(vinyl alcohol)-iron oxide nanocomposite membranes for pervaporation dehydration of isopropanol 1,4-dioxane and tetrahydrofuran, Journal of Membrane Science 283 (2006) 6573. [58] S.G. Adoor, B. Prathab, L.S. Manjeshwar, T.M. Aminabhavi, Mixed matrix membranes of sodium alginate and poly(vinyl alcohol) for pervaporation dehydration of isopropanol at different temperatures, Polymer 48 (2007) 54175430.