Journal of Membrane Science 493 (2015) 224–231

Contents lists available at ScienceDirect

Journal of Membrane Science

http://d0376-73

n CorrE-m

journal homepage: www.elsevier.com/locate/memsci

Highly permeable and robust membranes assembled from block-co-polymer-functionalized carbon nanotubes

Xueping Yao, Jie Li, Zhaogen Wang, Liang Kong, Yong Wang n

State Key Laboratory of Materials-Oriented Chemical Engineering, College of Chemical Engineering, Nanjing Tech University, Nanjing 210009, Jiangsu,PR China

a r t i c l e i n f o

Article history:Received 12 May 2015Received in revised form1 July 2015Accepted 2 July 2015Available online 3 July 2015

x.doi.org/10.1016/j.memsci.2015.07.00888/& 2015 Elsevier B.V. All rights reserved.

esponding author. Fax: þ86 25 8317 2292.ail address: yongwang@njtech.edu.cn (Y. Wan

a b s t r a c t

The fabrication of homogenous integrated layers of assembled carbon nanotubes (CNTs) on poroussubstrates for advanced membranes via vacuum filtration remains a significant challenge due to theaggregation of CNTs. Here we report the composite membranes with both high flux and good mechanicalrobustness produced from functionalized CNTs through direct coating with amphiphilic block copolymer.Chemically crosslinked CNT layers are tightly adhered to the substrate, rendering a strong mechanicalstability to the composite membranes. The thickness of the separation layer as well as the permselec-tivity of the membranes can be readily tuned simply by changing the deposition amount of CNTs. Themembranes with a moderate nanofiber coverage are capable of complete removal of 10-nm-particleswhile offer a remarkable pure water flux higher than 5.7�103 L/(m2 h bar), suggesting their high effi-ciency in the purification of liquids contaminated by tiny colloids or particulates.

& 2015 Elsevier B.V. All rights reserved.

1. Introduction

As one efficient and environmentally friendly strategy, mem-brane technology manages to provide a variety of solutions in allwalks of life including seawater desalination, wastewater reuseand groundwater treatment [1–3]. Among the pressure-drivenmembranes in aqueous applications, namely micro-, ultra-, na-nofiltration and reverse osmosis membranes, ultrafiltration (UF)membranes which are able to reject nanometer-sized particulatesand colloids under relatively lower driving forces have beenwidely applied in industry and tremendous efforts are con-tinuously made to further increase the permeance of UF mem-branes at negligible expense of selectivity [4]. For UF membraneswhich are based on size-sieving mechanism, there are some cru-cial factors that determine the separation performances, includingthe surface morphology, pore size and the size distribution, etc.[5]. The trade-off effect between selectivity and permeability ofmost membranes urgently needs to be broken by altering thesefactors, e.g. increasing the porosity and conducting surface mod-ification [4].

Compositing UF membranes of large porosity and narrow poresize distribution as selective layers onto a macroporous supportingsubstrate can be considered as a promising strategy for the fabri-cation of energy-efficient membranes. As one alternative origin for

g).

the selective layers, one-dimensional nanofibers (including alsonanowires and nanotubes) with high aspect ratios lead to largespecific surface area, high porosity of the nanofiber membranes,and more importantly, the consequent superiority in permeation.Moreover, with such advantages as three-dimensional continuousporous structures and small effective pore sizes, the high perme-ability and simultaneously high retention of the nanofiber mem-branes might be well guaranteed. A number of polymeric nanofi-bers membranes have been applied in some areas. Though pro-mising, the low mechanical strength greatly obstructs their furtherapplications. With the mechanical strength of tensile modulus ashigh as 1–2 TPa and tensile strength of 200 GPa, carbon nanotubes(CNTs) can meet the mechanical strength demand of the mem-branes under relatively harsh conditions [6]. Commercial feasi-bility of the preparation for CNTs-based separation membraneshas been realized thanks to the scale-up and manufacture whichreduces the cost of CNTs [7].

Vacuum filtration has been greatly applied in the past decadesto fabricate composite membranes with nanofibers as the buildingblocks for the selective layers due to its simplicity [8–10]. How-ever, two major factors need to be considered for such syntheticroute, namely the good dispersibility in solutions of the nanofibersbefore filtration and the strong adhesion between the nanofibersafter filtration. As for CNTs with strong intertube interactionspresent locally along the tubes, they have the tendency to inter-weave, and therefore form an inhomogeneous and unstable dis-persion. Purification of CNTs and addition of surfactant and poly-mers will lead to highly dispersed CNTs; and uniform CNTs

www.elsevier.com/locate/memscihttp://dx.doi.org/10.1016/j.memsci.2015.07.008http://dx.doi.org/10.1016/j.memsci.2015.07.008http://dx.doi.org/10.1016/j.memsci.2015.07.008http://crossmark.crossref.org/dialog/?doi=10.1016/j.memsci.2015.07.008&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.memsci.2015.07.008&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.memsci.2015.07.008&domain=pdfmailto:yongwang@njtech.edu.cnhttp://dx.doi.org/10.1016/j.memsci.2015.07.008

X. Yao et al. / Journal of Membrane Science 493 (2015) 224–231 225

membranes can be generated upon vacuum filtration [11,12]. Anumber of works have been reported for the preparation ofmembranes or filters composed of randomly stacked CNTs [13–15].In most cases, CNTs without any pretreatment were directly fil-trated on the porous substrates and usually large amount of CNTswere used to generate CNTs layers thick enough to maintainheavily entangled, integrated layers. It was found that these thickCNTs layers can be easily peeled off from the substrate and theywere typically used as free-standing, symmetric membranes[13,14]. Such thick and symmetric structures would significantlycompromise permeance. Recently, thin layers of single-walledCNTs (SWCNTs) have been filtrated on porous substrates to formcomposite membranes exhibiting very high permeance [16]. Un-fortunately, SWCNTs require very tedious and low-efficient dis-persion treatment and their cost is extremely high. Therefore, it ishighly appealing to assemble cheap multiwalled CNTs on macro-porous substrates to form thin but robust selective layers, allowinghighly permeable and selective filtrations.

In our previous work, homogeneous dispersion of multiwalledCNTs with excellent stability has been successfully produced uponthe addition of amphiphilic block copolymers, polystyrene-block-poly(4-pyridine) (PS-b-P4VP, shortly S4VP), into the dispersion ofCNTs in acetic acid [17]. Here in this work, for the first time wedemonstrated the fabrication of block-copolymer-functionalizedCNTs composite membranes. This simple yet effective methodinvolves vacuum filtrating the selective layers, well dispersed CNTswith active P4VP chains onto the macroporous polyethersulfone(PES) substrate membranes. The composite membranes exhibitedhighly controllable permeability via changing the mass of S4VP/CNTs nanofibers. Moreover, the mechanical properties of CNTs andthe chemical crosslinking of the P4VP chains stabilize the CNTslayers, which provide the composite membranes greater potentialin industrial applications. Compared to other membranes usingnanofibers as the building blocks, our membranes are dis-tinguished for their easy and affordable availability of the buildingmaterials (CNTs) and the mechanical robustness. Specifically, suchmembranes are expected to find interesting applications in theseparation of nanosized colloids, particulates, and proteins with

Scheme 1. The preparation process of the composite membranes: dispersion of CNTs witdispersed CNTs (b) to generate stable S4VP/CNTs dispersions (c); the vacuum filtrationcrosslinking of the composite membranes (e).

similar sizes.

2. Experimental

2.1. Materials

Block copolymer S4VP (Mn(PS)¼12,000 g/mol, Mn(P4VP)¼1700 g/mol, Mw/Mn¼1.09), polyethylene glycol (PEG, Mn¼203000 g/mol, Mw/Mn¼1.14) were purchased from PolymerSource Inc., Dorval, Canada and used without further treatment.CNTs (inside diameter: 4.570.5 nm, outside diameter: 1071 nm,length: 3–6 μm) with purity higher than 98% and 1,4-diiodobutane(DIB) with purity higher than 99% were purchased from Sigma-Aldrich. PES membranes with an average pore diameter of 450 nm(Tianjin Jinteng Instrument Co. Ltd.) were used as substrates.Acetic acid and ethanol with purity Z99.5% were obtained fromlocal suppliers. Monodisperse gold colloidal nanoparticles with adiameter of 10 nm were purchased from British BiocellInternational.

2.2. Preparation of S4VP/CNTs composite membranes by vacuumfiltration

The general strategy for fabricating composite membranes isshown in Scheme 1. The preparation of the S4VP-coated CNTs wasdescribed detailed in our previous work [17]. Briefly, a suspensionof CNTs in acetic acid (5 mg in 10 g) was prepared and sonicatedfor 30 min with a power of 350 W. Then 10 mg of S4VP was addedand sonicated for another 2 h with the power of 100 W. For thepreparation of composite membranes, the as-synthesized 0.15 wt%S4VP-coated CNTs solution was cooled down to room temperatureand diluted for 50 times with ethanol as the stock solution. Thedispersion formed a depositing layer on the surface of the sub-strate membrane by vacuum filtration. In a typical experiment,one piece of PES substrate membrane was immersed in ethanol for10 min and then fixed in a glass vacuum filter (Model 16306,Sartorius AG) connected to a suction filtration flask. 3 g of ethanol

h the assistance of sonication (a); the addition of S4VP block copolymers into highlyof S4VP/CNTs dispersions onto the macroporous substrates (d) and the chemical

X. Yao et al. / Journal of Membrane Science 493 (2015) 224–231226

was added for filtration to thoroughly prewet the PES membrane.Various amounts of S4VP/CNTs dispersions ranging from 5 to 20 gwere filtered through the PE S substrate membranes under a re-duced pressure of 3 kPa. 6 mL of ethanol was subsequently filteredthrough the membrane for 3 times to wash away the residualacetic acid in the composite membrane. The composite mem-branes were air dried for 30 min at the temperature of 60 °C andthen crosslinked in the vapor of DIB at 80 °C for 2 h under vacuum[18,19]. After crosslinking, the composite membranes were cooleddown to room temperature and immersed in ethanol for 30 minand then thoroughly washed with fresh ethanol to remove theresidual DIB. The membranes were finally dried in air at roomtemperature and were ready for usages.

2.3. Characterizations

The surface morphology of composite membranes was ob-served on a Hitachi S4800 field emission scanning electron spec-troscope (FESEM) operated at 5 kV. A thin layer of Pt/Pd alloy wassputter coated on the samples to avoid surface charging duringSEM observations. The detailed morphology of the S4VP-coatedCNTs was examined on a JEM-2100 transmission electron spec-troscope (TEM) operated at 200 kV. Water contact angles of dif-ferent composite membranes were obtained from a contact anglegoniometer (DropMeterA-100P, MAIST). Fourier transform infraredspectroscopy (FTIR) was recorded on a Thermo Nicolet 8700FT-IRspectrometer in the transmission-mode (32 scans, 4 cm�1). X-rayphotoelectron spectroscopy (XPS) was performed on a ThermoScientific ESCALAB 250 XPS system using a monochromatic Al Kαline X-ray source. The as-prepared S4VP/CNTs stock solution waspre-filtered by a PVDF membrane with a nominal pore diameter of0.22 μm to eliminate the S4VP micelles possibly coexisting withthe coated CNTs, and then deposited on silicon substrates for XPScharacterizations after drying. Scratch tests were carried out on ananoindentation instrument (Nano-Test Vantage, Micromaterials)to analyze the interfacial adhesion of between the PES substratesupport and the S4VP-coated CNTs layer in the composite mem-branes. Indentation tests were also performed on this instrumentto measure the mechanical properties of the composite mem-branes with the S4VP/CNTs layer and with the pure polymer layerassembled from S4VP micellar nanofibers, respectively. The lattermembrane was prepared following our previous method [9].

2.4. Filtration experiments of composite membranes

The water flux and the retention to PEG and gold nanoparticlesof the composite membranes were determined using a stirredfiltration cell (Amicon 8003, Millipore Co., Billerica, MA) under thepressure of 0.02 MPa. Typically, the composite membrane with adiameter of 25 mm was placed in the filtration cell with the S4VP/

Fig. 1. TEM images of the S4VP-coated CNTs (a, b) and the photogr

CNTs layer side of the membrane as the permeating side. For thePEG retention test, 5 mL of PEG solution at the concentration of0.5 g/L was filtered. After filtration for 1–2 min, about one half ofthe liquid was left in the cell and we terminated the filtration. Theconcentration of filtration solution as well as feed solution wascharacterized by Gel Permeation Chromatography (Water 515).The concentration of gold nanoparticles in the feed, permeate andconcentrated solutions were obtained by a UV–vis spectro-photometer (NanoDROP 2000C) at the wavelength of 530 nm.

3. Results and discussions

3.1. Morphologies of S4VP-coated CNTs nanofibers

We first tried to directly filter the as-received, pristine CNTsonto macroporous substrate membranes to produce a selectivelayer. However, the efforts failed as the pristine CNTs cannot forma stable suspension in any solvent we investigated, imposing dif-ficulties in performing the filtration process. Even for the portionof CNTs deposited on the substrate membranes did not establish ahomogeneous layer throughout the substrate surface and theywere easily detached from the substrate. Such challenges as lowdispersibility and insufficiently adhesion between nanotubes canbe attributed to the inert surface and large length-diameter ratioof CNTs [20]. Therefore, it is a prerequisite to render CNTs a gooddispersibility before we can use CNTs as building blocks for theselective layer of membranes. We previously demonstrated thatthe amphiphilic block copolymer S4VP, were effective in coatingCNTs with the assistance of ultrasonication, and the coated CNTsexhibited excellent dispersibility in polar solvents. In this work,S4VP-coated CNTs were used as active building blocks to buildselective layers on macroporous substrates to prepare highlypermeable composite membranes. With the fibrous-like mor-phology and narrow-distributed diameter of �20 nm, the soni-cation-assisted S4VP/CNTs solutions display the advantages suchas uniform dispersion and high stability, which serve as excellentbuilding blocks for the generation of composite membranes withhigh performances. On one hand, the mechanical strength of thecomposite membranes would be improved as the interior of thecomposite fibers are fulfilled with CNTs. On the other hand, withexterior layer composing of pyridine-enriched P4VP block copo-lymers, chemical crosslinking will take place to enhance the in-teractions between fibers and eventually improve the stability ofthe composite membranes.

As shown in Fig. 1(a), we easily discerned from the TEM imagesthe core/shell structures of the coated CNTs: a darker coresheathed with a uniform light shell. Fig. 1(b) is the high resolutionTEM of the composite nanofiber, from which we can clearly allo-cate the dark core with crystalline lattices as the CNTs and the

aph of the coated CNTs dispersed in ethanol (the inset of (a)).

X. Yao et al. / Journal of Membrane Science 493 (2015) 224–231 227

light shell is the polymer. Such a core/shell structure clearly in-dicates that CNTs are coated by the copolymer. Furthermore, fromFig. 1(b) we can determine that the integrated diameter of thecoated CNTs is approximately �20 nm and the thickness of thecopolymer layer is �5 nm. The coated CNTs form a stable solutionin polar solvents, e.g. acetic acid, ethanol and their mixtures. Theillustration picture in inset of Fig. 1(a) was the solution of coatedCNTs prepared in acetic acid followed by dilution with ethanol atthe ratio of 1:50. The solution maintains its clear and semi-transparent appearance for months at ambient environment,suggesting that the addition of S4VP greatly improves the dis-persion of CNTs in polar solvents because of the presence of activeP4VP blocks on the outermost surface of the coated CNTs.

3.2. Surface properties of S4VP/CNTs composite membranes

We then applied the vacuum filtration technique to deposit thecoated CNTs on the macroporous PES substrates. As CNTs used inthis work have a length of 3–6 μm which is one order of magni-tude larger than the pore diameter of the substrate membrane, theCNTs originally dispersed in the solution are completely collectedon the surface of the substrate membrane by forming a “filtrationcake”. The complete retention of CNTs by the PES membrane isalso evidenced by the colorless appearance of the filtrate which isin stark contrast to the dark color of the feeding solution con-taining CNTs. The as-prepared CNTs layer does not have a strongadhesion to the PES substrate and the CNTs are easily detachedfrom the substrate and dispersed again in water and other polarsolvents. The S4VP-coated CNTs layer must be fixed on the sub-strate before it can be used in separation. To this end, we exposedthe as-prepared CNTs/PES composite structures to the vapor of DIBto crosslink the P4VP blocks which existed on the outermost layerof the S4VP-coated CNTs. The crosslinking reaction is based on thequaternization of pyridyl groups on the P4VP blocks of S4VP by the

Fig. 2. Survey (a), N 1s (b), and I 3d (c) XPS spectra

bifunctional alkyl halide, DIB. When the two ends of a DIB mole-cule react with pyridyl groups anchored to different tubes, thecrosslinking occurs, connecting two different but adjacent CNTs.The CNTs are densely packed, allowing the P4VP chains from dif-ferent tubes to be frequently contacted. Therefore, the crosslinkingtaking place between different tubes should be accounted for aconsiderable portion in all types of quaterinzation involved in thissystem. As a result, the CNTs are tightly bound with each other andfixed as an integrated layer to the PES substrate after crosslinking.

We first used FT-IR to characterize the composite membranesbefore and after chemical crosslinking. A characteristic peak ofP4VP after quaternization reaction is expected to appear at thewavenumber of 1645 cm�1; however, this peak is overlapped bythe characteristic peak of pristine CNTs (spectra not shown). Al-ternatively, we utilized XPS with higher sensitivity to analyze thesurface composition of the membranes before and after cross-linking. Fig. 2(a–c) displays the survey, the corresponding high-resolution N 1s and I 3d spectra of the pristine and crosslinkedCNTs layers, respectively. As shown in Fig. 2(a), the survey of thepristine CNTs shows three main emission peaks: 284.6 eV for C 1s,399.0 eV for N 1s, and 532.1 eV for O 1s. The presence of a smallpeak corresponding to O 1s was due to the surface absorption ofH2O and CO2, which is very common in XPS analysis [21,22]. SinceN element is only originated from the pyridyl groups of S4VP, theXPS results confirmed again the successful coating of S4VP ontoCNTs.

After the CNTs were crosslinked via the quaternization reaction,the emission peaks of C 1s and O 1s remained prominent whereasthat of N 1s is weakened. The peak intensity of O 1s is enhancedcompared with the pristine one before crosslinking, which mightbe ascribed to the stronger adsorption of water of the quaternizedP4VP. After quaternization, the pyridyl groups in P4VP blocks areconverted into quaternary ammonium salt, which has a strongerhydrophilicity than the pristine P4VP block and adsorbs more

of pristine and crosslinked S4VP/CNTs layers.

X. Yao et al. / Journal of Membrane Science 493 (2015) 224–231228

water vapor from ambience which is difficult to be completelyremoved in the XPS tests. Interestingly, the peak intensity of N 1sdecreases after crosslinking as it splits into two peaks, which couldbe evidently discerned in the high resolution spectrum shown inFig. 2b. In contrast to a single stronger peak of the N 1s spectrumof the CNTs before crosslinking, the N 1s peak recorded from thequaternized sample showed two separated peaks, located at399.0 eV and 402.1 eV, respectively. The peak position of 399.0 eVN 1s was the same as the pristine sample, which corresponds tothe uncharged pyridyl rings of the P4VP chains. The other peakemerging at 402.1 eV was typically found for cationized nitrogenatoms of ammonium groups [23,24], which confirms the suc-cessful quaternization reaction. Moreover, the two separated peaksof nitrogen also indicate that the quaternization reaction does notcompletely convert all pyridyl groups into pyridinium cations. Thepeak area ratio can be calculated as [Nþ]/([Nþ]þ[N])¼0.34,

Fig. 3. Surface (a, b) and cross-sectional (c, d) SEM images of the composite membranimages of composite membranes with different deposition amounts of nanofibers: (e) 7

suggesting that the degree of the quaternization reaction wasapproximately 34%. In addition, as shown in Fig. 2c the emissionpeaks from iodine located at the binding energy of 618.4 eV and629.7 eV can also be detected from the high-resolution XPSspectrum of the crosslinked sample, representing I 3d5/2 and I3d3/2, respectively [25,26]. This indicates the incorporation of DIBmolecules which are the only source of iodine into the sample andfurther confirms the occurrence of the quaternization reactions.

3.3. Morphologies of S4VP/CNTs composite membranes

Fig. 3 exhibits the morphologies of the S4VP/CNTs nanofibersdeposited on the substrate membranes after filtration with variousamounts of fibers dispersed in ethanol. At relatively low amount offiltrated nanofibers, for instance, 36.6 μg/cm2, the massive net-work of nanofibers maintained the complete coverage of the

e with a nanofibers deposition amount of 36.6 μg/cm2. Cross-sectional (e–g) SEM3.2, (f) 109.8 and (g) 146.4 μg/cm2.

Fig. 4. Friction–distance curves of the composite membrane with a nanofiber de-position amount of 36.6 μg/cm2 before (upper line) and after (lower line) cross-linking (a), and photographs of this membrane subjected to ultrasonication chal-lenge before (b) and after (c) crosslinking.

Fig. 5. Water flux and rejection of PEG for composite membranes as a function ofdifferent deposition amount of S4VP/CNTs nanofibers.

X. Yao et al. / Journal of Membrane Science 493 (2015) 224–231 229

substrate membrane with a uniform and smooth surface. As can beseen from Fig. 3(c, d), a composite structure consisting of nanofi-brous selective layer and macroporous supporting layer can beclearly observed. The enlarged cross-sectional SEM image (Fig. 3(d)) of composite membranes displayed an interconnected selec-tive layer with the thickness of 750 nm. Such a selective layer withthe nanoporous structure can be attributed to the closely stackedand densely entangled nanofibers. The S4VP/CNTs nanofibers aretightly adhered and vertically compacted on the membrane sur-face without cracks or defects. The selective layers maintained thecoverage of all the macropores when the amount of filtrated fibersincreased to 146.4 μg/cm2 (Fig. 3(g)). The apparently thicker se-lective layer with similar morphology indicated the selective layerthickness could be tuned by controlling the amount of stackedfibers on the surface of composite membranes. For instance, thethickness of selective layer increased to 1.6 μm when the deposi-tion amount of nanofibers reached 73.2 μg/cm2. The selective layerthickness continuously increased to 2.4 and 2.9 μm with furtheraddition of deposited nanofibers, i.e., 109.8 and 146.4 μg/cm2, re-spectively. These cross-sectional SEM images also clearly show theexcellent uniformity in the thickness of the deposited S4VP/CNTslayers.

3.4. Mechanical properties of S4VP/CNTs composite membranes

For a composite membrane, there must be a interfacial adhe-sion strong enough between the selective layer and the under-neath substrate to avoid the exfoliation of the top layer from thesupport [27]. Peeling will tend to occur when the interfacial stresssurpasses a specific value, not to mention the exfoliation in theusage under harsh conditions. Considering the natural function-ality of P4VP on the nanofiber surface, this part will concentrate onthe interfacial adhesion between the nanofibers after crosslinkingthe surface P4VP and nano-scrach was applied. The critical loadincreased from 5.24 mN of the uncrosslinked membrane to11.96 mN of the crosslinked one, implying that the interfacial ad-hesion doubled after the crosslinking (results were shown in Fig. 4(a)). Figs. 4(b and c) were the photographs showing the compositemembrane before and after crosslinking challenged by ultra-sonication under the power of 100 W for 10 min in the ultrasonicbath, respectively. They clearly showed that the fibrous layer waswell preserved on the substrate for the crosslinked membraneafter ultrasonication (as shown in Fig. 4(b)); however, the fibrouslayer largely peeled from the substrate for the uncrosslinkedmembrane. The overall adhesion strength between the nanofibers,the fibrous layer and the substrate were improved through thesimple crosslinking method. The high adhesion strength guaran-teed the excellent permeation and separation performance of thecomposite membrane.

3.5. Permselectivity of the S4VP/CNTs composite membranes

Permeability and selectivity are important characteristics forthe evaluation of membrane performances. Fig. 5 displays waterflux and rejection of PEG versus deposition amount of S4VP/CNTsnanofibers. The pristine PES membrane exhibited a substantiallylarge flux of 4.4�104 L/(m2 h bar) and a negligible rejection toPEG, which possessed a molecular weight of 20,300 g/mol. Thewater flux of the composite membrane with a deposition amountof 36.6 μg/cm2 nanofibers decreased about 87% comparing to thatof the original PES substrate membrane. At the meantime the PEGrejection rate of this composite membrane drastically increased to52%. Such fluctuation in permeation and rejection was due to thesignificant decrement in effective pore size of the compositemembranes upon the filtration of S4VP/CNTs micellar fibers andthe establishment of the thin selective layers. Further increase of

the S4VP/CNTs fibers deposition amount to 109 μg/cm2 led to aneven higher rejection of 87% and a flux of 2.7�103 L/(m2 h bar).The rejection rate to PEG remained at �90% whilst water fluxdecreased despite of the continuous enhancement in the deposi-tion amount of S4VP/CNTs fibers. The results indicated an excellentpermeability of the composite membranes, which might be as-cribed to the particular selective layers of S4VP/CNTs fibers withthin thicknesses and active surfaces, leading to reduced flow re-sistance [28]. Moreover, the S4VP/CNTs nanofibers with a uniformdiameter of 20 nm and narrow size distribution were utilized inthis work, which implies remarkably improved rejection ability ofthe composite membranes can be anticipated since the selectivelayers were bestowed a relatively smaller pore size.

Many factors, such as surface charging, surface roughness andsurface hydrophilicity affect the permeability and selectivity of themembrane. As has been widely reported before, CNTs display ahydrophobic property due to its inert surface [29], which might

Fig. 6. Photographs of the water droplet remained on the surface of pure CNTs composite membrane (a) and S4VP/CNTs membrane after crosslinking the P4VP by DIB (b).

Fig. 8. Flux of composite membrane with the nanofibers deposition amount of146.4 μg/cm2 as a function of pressure.

X. Yao et al. / Journal of Membrane Science 493 (2015) 224–231230

cause poor permeability for the fabrication of separation mem-branes. To investigate the surface hydrophilicity of the membrane,contact angles were measured on the composite membranes be-fore and crosslinking. As shown in Fig. 6(a), the contact angle ofthe composite membrane before crosslinking was �141°. After lcrosslinking, the contact angle decreased to 104°. The reduction ofthe contact angle indicated that the surface hydrophilicity of thecomposite membrane considerably improved, which assigned tothe synergetic effect of P4VP block on the surface of the S4VP-coated CNTs and the quaternization reaction. Such process mayimprove the surface hydrophilicity of the P4VP chains which hasbeen verified in the XPS analysis. Coating of S4VP over CNTs alongwith quaternization moderately enhanced the hydrophilicity andwe note that their hydrophilicity may further be improved usingother post treatment, for example, coating of metal oxides byatomic layer deposition [30]. Therefore, a composite membranewith exceptional permeability compared to conventional CNTscomposite membranes was effectively fabricated through thecoating of S4VP following by crosslinking of the CNTs.

The modification of CNTs by coating S4VP empowered the su-perior permeability and selectivity performance of compositemembrane. Furthermore, the rigid nature of CNTs endowed thecomposite membrane with an excellent mechanical strength [31].We investigated the changes of flux as a function of filtration timeto study its mechanical stability. As shown in Fig. 7, the flux ofcomposite membrane with 109.8 μg/cm2 deposited nanofibersshowed a slight reduction in the first 30 min of filtration process,which should be attributed to the slight compaction of the nano-fibers stacked on the substrate surface [32]. With the enduring ofthe filtration, the flux almost kept constant. Fig. 8 shows that thereis a linear correlation between flux and the applied transmem-brane pressures (TMPs) in the range of 0.2–2 bar for the compositemembrane with 146.4 μg/cm2 deposited nanofibers. Therefore, thecomposite membrane indicated a great mechanical stability of thecomposite membranes with the addition of CNTs. Furthermore, wecompared the mechanical properties of the S4VP/CNTs layer pre-pared in this work and a pure polymer layer assembled from S4VPmicellar nanofibers prepared in our previous work [9] by the

Fig. 7. Water flux of composite membrane with the nanofibers deposition amountof 109.8 μg/cm2 as a function of time under the pressure of 0.02 MPa.

nanoindentation method. We found that the rigidity of the S4VP/CNTs layer increased 3 times compared to the pure S4VP layer,while the elasticity modulus increased by 77%, indicating that theaddition of CNTs could significantly improve the mechanicalproperties of the selective layers.

It has been demonstrated in the permeation tests that the size-selective separation performance of composite membranes washighly tunable simply by controlling the amount of deposited fi-bers. The separation efficiency was characterized by monodispersegold nanoparticles with a diameter of 10 nm. Fig. 9 shows theabsorption spectra variation of gold nanoparticle solutions. Thecharacteristic absorption band of gold at 530 nm [9] diminishedafter permeation whilst intensified on the separation side of thecomposite membrane, indicating an entire rejection of the com-posite membrane with an nanofiber deposition amount of36.6 μg/cm2. Visual color changes of the gold nanoparticle solu-tions from pink representing the feed stock into colorless as for thepermeate solution whereas the pink color darkened in the con-centrate solution corresponding to the increased absorption

Fig. 9. UV–vis absorption spectra of the concentrate, feed, and permeate solutionof monodisperse gold nanoparticles with a diameter of 10 nm separated by acomposite membrane with the deposited nanofiber amount of 36.6 μg/cm2. Insetshows photograph of the corresponding feed, concentrate, and permeate goldnanoparticle solution (from left to right). (For interpretation of the references tocolour in this figure, the reader is referred to the web version of this article.)

X. Yao et al. / Journal of Membrane Science 493 (2015) 224–231 231

intensity also confirmed the conclusion (Fig. 9 inset). Yu andcoworkers prepared membranes from stacked carbonaceous fibersand obtained a pure water flux of 1.1�103 L/(m2 h bar) and aneffective rejection to gold particles with the diameter of 25 nm[33]. In comparison, here we successfully improved the water fluxfor more than 5 times and the prepared composite membraneswere capable of completely rejecting even smaller nanoparticleswith the diameter down to 10 nm. This concurrent improvementin permeability and rejection should be ascribed to the synergeticeffect of the smaller diameter of the S4VP/CNTs fiber, the thinnerthickness of the S4VP/CNTs fiber layer, as well as the enhancedwater wettability of the selective layer.

4. Conclusions

In this work we report a robust methodology fabricatingcomposite membranes with superior mechanical strength andseparation performances simply by assembling S4VP/CNTs nano-fibers onto a macroporous substrate. The adhesion of compositemembranes was successfully improved via chemical crosslinking.Moreover, the addition of CNTs significantly enhanced the me-chanical strength of the composite membranes. With a particularlyuniform and well distributed diameter of 20 nm, the compositemembranes were endowed with small effective pore sizes leadingto tightened rejections. The permeability and selectivity of thecomposite membranes could be easily tuned by altering the de-position amount of S4VP/CNTs nanofibers. The composite mem-brane with the nanofiber deposition amount of 36.6 μg/cm2 iscapable of completely reject gold nanoparticles as small as 10 nmwhile offers a remarkable pure water flux higher than5.7�103 L/(m2 h bar). Such advanced composite membranes withultrathin selective nanofibrous layers are expected to find inter-esting applications where tiny particulates or colloidal substancesare require to remove or concentrate with high efficiency and lessenergy input.

Acknowledgment

Financial support from the National Basic Research Program ofChina (2015CB655301), the Doctoral Program Foundation of In-stitutions of Higher Education of China (20123221110003), theJiangsu Natural Science Funds for Distinguished Young Scholars(BK2012039), and the Project of Priority Academic Program De-velopment of Jiangsu Higher Education Institutions (PAPD) isgratefully acknowledged.

References

[1] R.V. Linares, Z. Li, M. Abu-Ghdaib, C.-H. Wei, G. Amy, J.S. Vrouwenvelder,Water harvesting from municipal wastewater via osmotic gradient: an eva-luation of process performance, J. Membr. Sci. 447 (2013) 50–56.

[2] M. Elimelech, W.A. Phillip, The future of seawater desalination: energy, tech-nology, and the environment, Science 333 (2011) 712–717.

[3] T. Melin, B. Jefferson, D. Bixio, C. Thoeye, W. De Wilde, J. De Koning, J. van derGraaf, T. Wintgens, Membrane bioreactor technology for wastewater treat-ment and reuse, Desalination 187 (2006) 271–282.

[4] M. Mulder, Basic Principles of Membrane Technology, 2nd ed., Kluwer Aca-demic Publishers, Dordecht, 2004.

[5] M. Radjabian, V. Abetz, Tailored pore sizes in integral asymmetric membranesformed by blends of block copolymers, Adv. Mater. 27 (2015) 352–355.

[6] J.H. Du, J. Bai, H.M. Cheng, The present status and key problems of carbonnanotube based polymer composites, Express Polym. Lett. 1 (2007) 253–273.

[7] N.G. Sahoo, S. Rana, J.W. Cho, L. Li, S.H. Chan, Polymer nanocomposites based

on functionalized carbon nanotubes, Prog. Polym. Sci. 35 (2010) 837–867.[8] S. Karan, S. Samitsu, X.S. Peng, K. Kurashima, I. Ichinose, Ultrafast viscous

permeation of organic solvents through diamond-like carbon nanosheets,Science 335 (2012) 444–447.

[9] X.P. Yao, Z.G. Wang, Z.M. Yang, Y. Wang, Energy-saving, responsive membraneswith sharp selectivity assembled from micellar nanofibers of amphiphilicblock copolymers, J. Mater. Chem. A 1 (2013) 7100–7110.

[10] H. Huang, Z. Song, N. Wei, L. Shi, Y. Mao, Y. Ying, L. Sun, Z. Xu, X. Peng, Ultrafastviscous water flow through nanostrand-channelled graphene oxide mem-branes, Nat. Commun. 4 (2013) 345–350.

[11] K. Yurekli, C.A. Mitchell, R. Krishnamoorti, Small-angle neutron scatteringfrom surfactant-assisted aqueous dispersions of carbon nanotubes, J. Am.Chem. Soc. 126 (2004) 9902–9903.

[12] Y. Deng, Q. Hu, Q. Yuan, Y. Wu, Y. Ling, H. Tang, One-pot synthesis of molecularbottle-brush functionalized single-walled carbon nanotubes with superiordispersibility in water, Macromol. Rapid Commun. 35 (2014) 97–102.

[13] G. Viswanathan, D.B. Kane, P.J. Lipowicz, High efficiency fine particulate fil-tration using carbon nanotube coatings, Adv. Mater. 16 (2004) 2045–2049.

[14] Á. Kukovecz, R. Smajda, Z. Kónya, I. Kiricsi, Controlling the pore diameterdistribution of multi-wall carbon nanotube buckypapers, Carbon 45 (2007)1696–1698.

[15] S. Roy, V. Jain, R. Bajpai, P. Ghosh, A.S. Pente, B.P. Singh, D.S. Misra, Formationof carbon nanotube bucky paper and feasibility study for filtration at the nanoand molecular scale, J. Phys. Chem. C 116 (2012) 19025–19031.

[16] Z. Shi, W.B. Zhang, F. Zhang, X. Liu, D. Wang, J. Jin, L. Jiang, Ultrafast separationof emulsified oil/water mixtures by ultrathin free-standing single-walledcarbon nanotube network films, Adv. Mater. 25 (2013) 2422–2427.

[17] X. Yao, J. Li, L. Kong, Y. Wang, Surface functionalization of carbon nanotubes bydirect encapsulation with varying dosages of amphiphilic block copolymers,Nanotechnology (2015) (in press).

[18] Q. Lou, P.S. Chinthamanipeta, D.A. Shipp, Mechanism of titania deposition intocylindrical poly(styrene-block-4 vinyl pyridine) block copolymer templates,Langmuir 27 (2011) 15206–15212.

[19] S. Akasaka, H. Mori, T. Osaka, V.H. Mareau, H. Hasegawa, Controlled in-troduction of metal nanoparticles into a microdomain structure, Macro-molecules 42 (2009) 1194–1202.

[20] P.-C. Ma, N.A. Siddiqui, G. Marom, J.-K. Kim, Dispersion and functionalizationof carbon nanotubes for polymer-based nanocomposites: a review, Compos.Part A 41 (2010) 1345–1367.

[21] Y. Sui, X.L. Gao, Z.N. Wang, C.J. Gao, Antifouling and antibacterial improvementof surface-functionalized poly(vinylidene fluoride) membrane prepared viadihydroxyphenylalanine-initiated atom transfer radical graft polymerizations,J. Membr. Sci. 394 (2012) 107–119.

[22] K. Andersson, A. Nikitin, L.G.M. Pettersson, A. Nilsson, H. Ogasawara, Waterdissociation on Ru(001): an activated process, Phys. Rev. Lett. 93 (2004)196101.

[23] Q. Wang, S. Samitsu, I. Ichinose, Ultrafiltration membranes composed of highlycross-linked cationic polymer gel: the network structure and superior se-paration performance, Adv. Mater. 23 (2011) 2004–2008.

[24] B.P. Tripathi, N.C. Dubey, S. Choudhury, F. Simon, M. Stamm, Antifouling andantibiofouling pH responsive block copolymer based membranes by selectivesurface modification, J. Mater. Chem. B 1 (2013) 3397–3409.

[25] X.L. Zhou, F. Solymosi, P.M. Blass, K.C. Cannon, J.M. White, Interactions ofmethyl halides (Cl, Br and I) with Ag(111), Surf. Sci. 219 (1989) 294–316.

[26] C. Furlani, G. Mattogno, G. Polzonetti, R. Barbieri, E. Rivarola, A. Silvestri, Re-lationship between XPS core binding-energies and atomic charge in adductsof Sn-IV derivatives with pyrazine, and comparison with mossbauer isomer-shift data, Inorg. Chim. Acta 52 (1981) 23–28.

[27] K. Huang, G. Liu, Y. Lou, Z. Dong, J. Shen, W. Jin, A graphene oxide membranewith highly selective molecular separation of aqueous organic solution, An-gew. Chem. Int. Ed. 53 (2014) 6929–6932.

[28] Q. Yu, Y. Mao, X. Peng, Separation membranes constructed from inorganicnanofibers by filtration technique, Chem. Rec. 13 (2013) 14–27.

[29] N. Kemnade, C.J. Shearer, D.J. Dieterle, A.S. Cherevan, P. Gebhardt, G. Wilde,E. Eder, Non-destructive functionalisation for atomic layer deposition of metaloxides on carbon nanotubes: effect of linking agents and defects, Nanoscale 7(2015) 3028–3034.

[30] F. Li, X. Yao, Z. Wang, W. Xing, W. Jin, J. Huang, Y. Wang, Highly porous metaloxide networks of interconnected nanotubes by atomic layer deposition, NanoLett. 12 (2012) 5033–5038.

[31] X.Y. Lin, P. Liu, Y. Wei, Q.Q. Li, J.P. Wang, Y. Wu, C. Feng, L.N. Zhang, S.S. Fan, K.L. Jiang, Development of an ultra-thin film comprised of a graphene mem-brane and carbon nanotube vein support, Nat. Commun., 4, (2013) 2920.

[32] W.A. Phillip, M. Amendt, B. O’Neill, L. Chen, M.A. Hillmyer, E.L. Cussler, Dif-fusion and flow across nanoporous polydicyclopentadiene-based membranes,ACS Appl. Mater. Interfaces 1 (2009) 472–480.

[33] H.W. Liang, L. Wang, P.Y. Chen, H.T. Lin, L.F. Chen, D. He, S.H. Yu, Carbonaceousnanofiber membranes for selective filtration and separation of nanoparticles,Adv. Mater. 22 (2010) 4691–4695.

http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref1http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref1http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref1http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref1http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref2http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref2http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref2http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref3http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref3http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref3http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref3http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref4http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref4http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref5http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref5http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref5http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref6http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref6http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref6http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref7http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref7http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref7http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref8http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref8http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref8http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref8http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref9http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref9http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref9http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref9http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref10http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref10http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref10http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref10http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref11http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref11http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref11http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref11http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref12http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref12http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref12http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref12http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref13http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref13http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref13http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref14http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref14http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref14http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref14http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref15http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref15http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref15http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref15http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref16http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref16http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref16http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref16http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref17http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref17http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref17http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref18http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref18http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref18http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref18http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref19http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref19http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref19http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref19http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref20http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref20http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref20http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref20http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref21http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref21http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref21http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref21http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref21http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref22http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref22http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref22http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref23http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref23http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref23http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref23http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref24http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref24http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref24http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref24http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref25http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref25http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref25http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref26http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref26http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref26http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref26http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref26http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref27http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref27http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref27http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref27http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref28http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref28http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref28http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref29http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref29http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref29http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref29http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref29http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref30http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref30http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref30http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref30http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref31http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref31http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref31http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref32http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref32http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref32http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref32http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref33http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref33http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref33http://refhub.elsevier.com/S0376-7388(15)30038-7/sbref33

Highly permeable and robust membranes assembled from block-copolymer-functionalized carbon nanotubesIntroductionExperimentalMaterialsPreparation of S4VP/CNTs composite membranes by vacuum filtrationCharacterizationsFiltration experiments of composite membranes

Results and discussionsMorphologies of S4VP-coated CNTs nanofibersSurface properties of S4VP/CNTs composite membranesMorphologies of S4VP/CNTs composite membranesMechanical properties of S4VP/CNTs composite membranesPermselectivity of the S4VP/CNTs composite membranes

ConclusionsAcknowledgmentReferences