Desalination 367 (2015) 134–144

Contents lists available at ScienceDirect

Desalination

j ourna l homepage: www.e lsev ie r .com/ locate /desa l

Preparation and properties of nanocomposite polysulfone/multi-walledcarbon nanotubes membranes for desalination

Arsalan Khalid a, Abdulhadi A. Al-Juhani a, Othman Charles Al-Hamouz b, Tahar Laoui c,Zafarullah Khan c, Mautaz Ali Atieh d,⁎a Department of Chemical Engineering, King Fahd University of Petroleum & Minerals, Dhahran 31261, Saudi Arabiab Chemistry Department, King Fahd University of Petroleum & Minerals, Dhahran 31261, Saudi Arabiac Department of Mechanical Engineering, King Fahd University of Petroleum & Minerals, Dhahran 31261, Saudi Arabiad Qatar Environment & Energy Research Institute (QEERI), Doha, Qatar

H I G H L I G H T S

• Dodecylamine functionalized multi-walled carbon nanotubes (DDA-MWNTs) were prepared• Nanocomposite Polysulfone/DDA-MWNTs membranes were casted by phase inversion method• DDA-MWNTs enhance the interfacial compatibility between polysulfone and nanotubes• Nanocomposite membranes displayed enhanced fouling resistance and flux recovery

⁎ Corresponding author.E-mail address: mhussien@qf.org.qa (M.A. Atieh).

http://dx.doi.org/10.1016/j.desal.2015.04.0010011-9164/© 2015 Published by Elsevier B.V.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 8 March 2015Received in revised form 1 April 2015Accepted 2 April 2015Available online xxxx

Keywords:Polysulfone membranesMulti-walled carbon nanotubesDodecylaminePolymer compatibilityDesalinationMembrane fouling

Novel nanocomposite membranes for application in desalination were fabricated from polysulfone (PSf) andmulti-walled carbon nanotubes (MWNTs) functionalized with dodecylamine (DDA) and anti-fouling propertiesof the membranes were studied during filtration of bovine serum albumin (BSA) solutions. Phase inversion pro-cess with dimethylacetamide as solvent and polyvinylpyrrolidone as a porogen was used to prepare flat sheetnanocomposite PSf/DDA-MWNTs membranes. Before embedding MWNTs in the polymer matrix, they weretreated with HNO3 for the introduction of carboxylic groups on nanotube surface and then modified with DDA.The prepared DDA-MWNTs were characterized using scanning electron microscope, infrared spectroscopy andthermal gravimetric analysis. The long alkyl chains of DDA functionalizedMWNTs seem to enhance the interfacialadhesion and compatibility between inorganic nanotubes and PSf matrix. The changes in surface hydrophilicity,morphology and roughness of fabricated nanocomposite membranes as a function of DDA-MWNTs loadingwereevaluated by using contact angle goniometry and atomic force microscopy. The prepared nanocomposite mem-branes possessed significantly higher permeabilitywith improved protein fouling resistance than the pristine PSfmembrane during filtration of BSA solutions. The membrane prepared with 0.5 wt.% loading of DDA-MWNTsdisplayed the highest flux recovery (83%) and lowest total flux loss (29%)with reduced irreversible fouling resis-tance (17%).

© 2015 Published by Elsevier B.V.

1. Introduction

Polysulfone (PSf), a thermoplastic film forming material, has been aresearch interest of membrane manufacturers for a long time due to itshigh thermal strength, solubility in many aprotic polar solvents, andchemical durability over a wide pH range [1,2]. PSf is used for the prep-aration of micro- and ultrafiltration membranes, as well as support

material for fabrication of composite nanofiltration and reverse osmosismembranes [2,3]. Nevertheless, the hydrophobic nature of PSf mem-branes results in their susceptibility to membrane fouling.

Membrane organic fouling is often caused by adsorption and deposi-tion of natural organic matter (NOM) on the membrane surface or intothe pores leading to pore plugging phenomena and formation of cakelayer on membrane surface [4]. Membrane surface chemistry plays acrucial role in the performance ofmembrane as the electrostatic and hy-drophobic interactions between the organic foulant and the membranecause organic fouling [5,6]. Proteinmolecules that have hydrophobic re-gions are attracted towards hydrophobic membrane [7]. Therefore,

135A. Khalid et al. / Desalination 367 (2015) 134–144

surface modification of membrane, i.e., increasing the hydrophilicity, iswidely believed as a beneficial route to enhance protein fouling resis-tance [8].

Carbon nanotubes (CNTs) have been widely explored in chemistryand material science owing to its unique properties such as high aspectratio, low density, high chemical, thermal, and mechanical strengthsand remarkable electrical and optical properties [9–13]. The potentialadvantages of CNTs integration in polymer membranes to manipulatemembrane properties through nanotubesmodification include improvedpermeability and solute rejection, decreased fouling tendency, increasedtensile strength and electrical conductivity along with controlled poresize, surface chemistry and polymer crystallinity [14–20].

Up to date, various studies have succeeded to prepare PSf/CNTsnanocomposite membranes for flux increment and porous structurecontrol [14,20–22]. Brunet et al. [22] formed PSf/CNTs composite mem-branes using a phase inversionmethod. Dispersion of 4wt.% CNTs insidethe polymer matrix was assisted by the addition of polyvinylpyrroli-done (PVP). However, mechanical strength of the prepared membranewas reduced due to the presence of CNT agglomerates. Electricallyconductive PSf microporous membrane was fabricated with pristinemulti-walled carbon nanotubes (MWNTs) using sonication method[16]. The result indicated that MWNTs at 3 wt.% loading are homoge-neously dispersed in themembranewhile nanotubes start agglomeratingat 6 wt.% content of MWNTs in the casting solution. It can be concludedfrom the above mentioned studies that the dispersion of MWNTs inpolymer matrix and affinity of MWNTs for polymer were critical factorsin determining the prepared membrane properties.

Choi et al. [14] obtained PSf/MWNTs membranes by immersionprecipitation and the prepared membranes displayed higher surfacehydrophilicity due to addition of carboxylic functionalized CNTs. Thecomposite membrane depicted a slightly higher water flux and soluterejection than the unmodified PSf membranes. Ultrafiltration mem-branes were also prepared by functionalizing MWNTs with isocyanateand isophthaloyl chloride functional groups followed by MWNTs dis-persion into the PSf casting solution [21]. Suppression of protein adsorp-tion over membrane surface was observed in static conditions and itwas concluded that composite PSf/MWNTs membrane might mitigatemembrane fouling.

Recently, Lannoy et al. [20] investigated an effect of the degree ofMWNTs carboxylization onwaterflux, hydrophilicity and tensile strengthof the prepared compositemembranes. Higher degree ofMWNTs carbox-ylation resulted in higher hydrophilicity, but also reduced tensile strengthand increased CNTs leakage from the membrane. The study indicated aneed for enhancing the compatibility of CNTs with polymer membranematrix through targeted CNTs functionalization.

Although CNTs possess very good mechanical and electrical proper-ties, their large scale exploitation for fabrication of polymer nanocom-posite membranes still encounters some problems [23]. The principlechallenge is the difficult debundling and dispersion of nanotubes in or-ganic solvents because of nanotubes aggregation due to strong van derWaal's attraction. Non-homogeneous dispersion of inorganic CNTsthroughout the polymer matrix is also an obstacle along with poor in-terfacial bonding between CNTs and polymer [9–12]. Moreover, the ad-verse toxicological effects of CNTs on humans and the environment arestill under research [24], but it is advised tominimize CNTs loss into theenvironment from economic and environmental points of view. There-fore, these challenges need to be addressed during the fabrication ofnanocomposite membrane with enhanced compatibility betweenCNTs and polymer matrix.

Chemical functionalization is one of the few techniques to improvethe uniform dispersibility of CNTs in organic solvents and interfacialbonding with polymer chains [12,25]. The objective of this study is tochemically functionalize MWNTs with dodecylamine (DDA) as a longchain aliphatic amine provides numerous bonding and/or entangle-ment sites to polymer matrix with superior interfacial compatibilityand affinity for PSf [26,27]. These effects can minimize CNTs leaching

from PSf matrix during membrane fabrication and application. DDAalso facilitates better dispersion and debundling of hydrophobic CNTsin organic solvents [28–30]. DDAmodifiedMWNTs have already provedpotential antibacterial properties [26] and might be effective for miti-gating of membrane fouling with organic compounds such as proteinssince protein and bacterial deposition on the membrane surface bearsome common features [32,33].

To the best of our knowledge, no study has been presented on or-ganic fouling behavior of PSf membranes with embedded function-alized DDA-MWNTs. Using immersion precipitation procedure weprepared PSf membranes with different loadings of functionalizedDDA-MWNTs. Scanning electron microscope (SEM), Fourier trans-form infrared spectroscopy (FTIR), and thermal gravimetric analysis(TGA) were used for the characterization of DDA-MWNTs. To char-acterize composite PSf/DDA-MWNTs membranes, porosity, surfacemorphology, hydrophilicity and water fluxes were studied. The anti-fouling behavior of fabricated membranes was evaluated in crossflow filtration with solutions of bovine serum albumin (BSA) as amodel organic foulant.

2. Experimental

2.1. Materials and reagents

PSf and PVP with molecular weight of 35 kDa and 40 kDa were pur-chased from Sigma-Aldrich and used for membrane casting as a basepolymer and a pore forming agent, respectively. Dimethylacetamide(DMAC) and DDA were provided by Acros Organics (Belgium). Polyes-ter support Novatexx-2413 (as shown in Fig. 1) was ordered fromFreudenberg Filtration Technologies (Germany). Millipore deionized(DI) water (18 MΩ·cm resistivity) was used as non-solvent in theimmersion precipitation coagulation bath. Bovine serum albumin(BSA) with a molecular weight of ~68 kDa was also purchased fromSigma-Aldrich. MWNTs (purity N95%) of 10–30 μm in length andoutside/inside diameter of 10–20 nm and 5–10 nm, respectively weresupplied from Chengdu Organic Chemicals Co. Ltd. (China). Nitric acidwith 69% purity was used for oxidation of MWNTs. Petroleum ether(Sigma-Aldrich) was used for washing of functionalized MWNTs.

2.2. Preparation of functionalized MWNTs

2.2.1. Acid treatment of MWNTsIn order to introduce carboxylic group (COOH) on the surface of

MWNTs, 1 g of purified MWNTs were dispersed in 75 ml of HNO3

using sonication (Branson, 3510) for 30 min. The mixture was thentransferred to a round bottom flask and refluxed under vigorousstirring at 120 °C for 48 h [34]. The mixture was allowed to cooldown at ambient temperature and diluted with DI water. Theoxidized MWNTs (O-MWNFs) were washed with DI water at leastsix times for 5 min by using centrifuge (CR22GΙΙΙ, Hitachi) at12,000 rpm to remove the excess nitric acid. Thereafter O-MWNTswere dried under vacuum at 85 °C overnight.

2.2.2. MWNTs modification with DDATo functionalize MWNTswith DDA, 10 g of DDAweremelted first in

a round bottom flask on a hot plate at 90 °C and 1 g of O-MWNTs wasthen added. The mixture was allowed to stir for about 10min, followedby the addition of a few drops of H2SO4 as catalyst. The reactionproceeded for 5 h under nitrogen atmosphere. The resulting mixturewas repeatedly washed and precipitated with petroleum ether untilno more DDA could be observed in the supernatant solution. Afterthat, the precipitated DDA-MWNTs were decanted with acetone as afinal washing step before drying overnight under vacuum at 85 °C. Re-action scheme for the preparation of DDA modified MWNTs is present-ed in Fig. 2.

Fig. 1. SEM images of polyester support Novatexx-2413.

136 A. Khalid et al. / Desalination 367 (2015) 134–144

2.3. Characterization of functionalized MWNTs

The morphology of pristine and functionalized MWNTs was ob-served using Field Emission Scanning Electron (FE-SEM) microscope(MIRA3 TESCAN) with accelerating electron voltage of 15 kV. The sam-pleswere coatedwith a gold layer of 5 nm thickness using Ion Sputter Q150R S (Quorum Technologies).

Fourier Transform Infrared (FTIR) spectra of MWNTs samples wererecorded using NICOLET 6700 (Thermo Scientific) FTIR spectrometerat 32 scans. Thermal gravimetric analysis (TGA) of MWNTs sampleswas performed using SDT Q600 TGA-DTA (TA instrument). Samplesof 5–10 mg weight were heated in aluminum pans up to 800 °C at2 °C/min rate under N2 flow rate.

2.4. Preparation of nanocomposite PSf/DDA-MWNTs membranes

The phase inversion technique and casting solutions, which con-tained 15 wt.% PSf, 5 wt.% PVP and 0.1–1.0 wt.% of DDA-MWNTs in80 wt.% DMAC, were used to prepare PSf/DDA-MWNTs membranes.Higher molecular weight PVP results in the suppression of macrovoidsand reduces water flux. Therefore, PVP with an optimum molecularweight of 40 kDa was chosen [35,36].

A required amount of DDA-MWNTs was sonicated in DMAC for 1 hto fully disperse nanotubes in the solvent. After PVP addition, whichwraps around nanotubes to reduce van derWaal's interaction, themix-ture was continuously stirred for at least 30 min prior to PSf dissolving.The casting solution was stirred for 24 h at 60 °C. The prepared solutionwas dark in color, homogeneous and without any macroscopic agglom-eration. Itwasdegassed for 30min and left under vacuum for 24 h to en-sure removal of any air bubble from the solution followed by the knifecasting at a constant shear rate and thickness of 14 mm/s and 200 μm,

Fig. 2. Reaction scheme for the prepa

respectively. The solutionwas cast at ambient temperature on commer-cial polyester no-woven support which was attached to the glass plateby double side sticker. The cast film is instantaneously immersed in anon-solvent water bath at 22 °C for 1 day for the complete phase inver-sion process. Leaching of CNTswas not observed from the casted film byvisualization. For preparation of pristine PSFmembranes, the casting so-lution was stirred for 5 h at 60 °C until the solution is clear transparent.Other casting procedures are similar as mentioned above.

2.5. Characterization of nanocomposite PSf/DDA-MWNTs membranes

The surface morphology of PSf nanocomposite membranes wasstudied by FE-SEM as described in Section 2.3. The pore size of the fab-ricated membranes was evaluated by means of the imaging software‘Image Pro Premier’.

Water contact angles on the membrane surface were measuredby DM-501 device (Kyowa Interface Science Co.). 2 μL of DI water wasdropped on the membrane and at least 10 water contact angles wereaveraged, which were taken on different surface locations.

The overall porosity of membranes was determined accordingly to[37]. Themembranes were soaked in awater bath for 24 h, then carefullyremoved and swipedwithfilter paper to remove superficialwater prior tobeingweighed. After that, membraneswere dried in a vacuumover nightat 60 °C and again weighed. The porosity was calculated as follows [37]:

P %ð Þ ¼ W1−Woð Þ= ρAhð Þ � 1000 ð1Þ

where P is the membrane porosity (%), ρ is the density of water at roomtemperature (g/cm3), A is the surface area (cm2), h is the membranethickness (mm), andW1 andWo arewet and dryweights (g), respective-ly. Three values were taken to avoid experimental error and average isbeing reported.

ration of DDA modified MWNTs.

Fig. 3. Schematic representation of experimental setup for filtration unit.

137A. Khalid et al. / Desalination 367 (2015) 134–144

The morphology and roughness of the membrane surface werestudied using atomic force microscope (Bruker Co., Germany). Thedriedmembrane sampleswere scanned in a tappingmode at room tem-perature in air. The roughness parametersweremeasured over 5× 5 μmscan size and estimated in terms of average roughness (Sa) [38,39].

2.6. Filtration experiments

The filtration experiments were conducted in a cross flow modeusing Sterlitech CF-042 membrane cell with an effective membrane

Fig. 4. SEM images of pristine MWNTs (a

area of 42 cm2. (See Fig. 3.). The filtration experiments were performedat constant transmembrane pressure of 1 bar. After filtration of DI waterfor 1 h the water flux was measured as follows:

Jo ¼ V=AΔt ð2Þ

where Jo is the volumetric flux (l/m2·h), A is the effective membranearea (m2), Δt is the permeation time interval (h) and V is permeatevolume (l).

and b) and DDA-MWNTs (c and d).

Fig. 5. FTIR spectra of as received MWNTs (a), O-MWNTs (b) and DDA-MWNTs (c).

138 A. Khalid et al. / Desalination 367 (2015) 134–144

BSA solution of 200 mg/l concentration in phosphate buffer (pH 7)was used for filtration fouling experiments. After filtration of BSA solutionfor 1 h the membrane flux J1 was measured. Thereafter the fouled mem-branes were flushed with DI water for 10 min and water flux was mea-sured again (J2). Antifouling ability of the membranes was evaluated bymeans of several parameters such as flux recovery ratio (FRR), total foul-ing (Rt), reversible (Rr), and irreversible (Rir) membrane fouling accord-ingly to [40]:

FRR %ð Þ ¼ J2Jo� 100 ð3Þ

Rr %ð Þ ¼ J2− J1Jo

� �� 100 ð4Þ

Fig. 6. TGA analysis of MWNTs and O

Rir %ð Þ ¼ Jo− J2Jo

� �ð5Þ

Rt %ð Þ ¼ Jo− J1Jo

� �� 100 ð6Þ

where Jo is the initialwaterflux, J1 is thefluxduring BSAfiltration, and J2 isthe water flux of the fouled membrane after its flushing with water. Theaverage flux of three replicates is reported with standard deviation.

-MWNTs (a), DDA-MWNTs (b).

Fig. 7. Possible hydrogen bonding between the amide group of DDA-MWNTs and the sul-fonic group of PSf.

139A. Khalid et al. / Desalination 367 (2015) 134–144

3. Results and discussion

3.1. Characterization of functionalized MWNTs

3.1.1. Scanning electron microscope analysisSurface morphologies of the raw and modified adsorbents were

observed using FE-SEM. Fig. 4(a, b, c and d) displays the FE-SEM imagesof raw and functionalize carbon nanotubes at low and high magnifica-tions. The diameters of the CNTs varied from20 to 40 nmwith an averagediameter of 24 nm. It can be observed that, there are no changes on sur-faces of raw andmodified CNTs after acid treatment and functionalizationwith amine groups. The only major differences that can be seen from theSEM images that; the samples after treatment with acid became morepure, less containment and highly pores compared to raw CNTs whichare highly dense samples. The observation of the clear individuality of

Fig. 8. Photo of PSf/O-MWNTs (a) and PSf/DDA-MWNTs (b) nanocomposite membranes at differight).

CNTs may be attributed to the effect of functional groups that decreasethe interaction forces between the CNTs.

3.1.2. FTIR analysisFig. 5 represents the FTIR spectra of as received MWNTs, O-MWNTs

and DDA-MWNTs. For the as-receivedMWNTs (a) and O-MWNTs (b), apeak at 3440 cm−1 can be assigned to the O–H stretch from carboxylgroups, while the characteristic peak at 1741 cm−1 corresponds to thecarbonyl stretching mode of the carboxylic groups [14]. The aromaticring stretched in the region of 1450–1510 cm−1. All types of MWNTsshow peaks between 1300 and 1100 cm−1, which are ascribed to thephenyl-carbonyl C–C stretch bonds [34]. The presence of carboxylicgroups in commercial MWNTs can be expected because a purificationstage was used by the manufacturer [34,41]. The hydroxyl stretchingat 3436 cm−1 might also result from ambient atmospheric moistureabsorption.

The successful chemical modification of MWNTs by DDA is confirmedby strong peaks at 3443 cm−1, 2920 cm−1, 2851 cm−1 and 1635 cm−1 inFig. 5, spectra (c) [28–30]. For DDA-MWNTs, the sharp and broad stretchat 3443 cm−1 corresponds to N–H bonds (stretching oscillations) inamide group (O_C–N–H). The peak at 1635 cm−1 is also due to forma-tion of amide linkage, while a peak appearing at 1744 cm−1 is indicativeof carbonyl group (C_O) in amide group present on DDA-MWNTs.The relatively strong peaks at 2920 cm−1 and 2851 cm−1 associatedwith C–H stretch mode are indicative of the presence of a long alkylchain of DDA on MWNTs surface.

3.1.3. Thermal gravimetric analysisFig. 6 demonstrates the thermal degradation behavior of MWNTs,

O-MWNTs and DDA-MWNTs under nitrogen atmosphere. The negligi-ble weight loss before 100 °C can be related to the removal of moistureand/or volatile organic materials associated with washing steps duringmaterial synthesis procedure. The pristineMWNTs showed degradationbehavior in single step, % weight loss starts after 600 °C. As comparedto MWNTs, O-MWNTs exhibit the initial degradation temperature at156 °C and continue to degrade till 584 °C with total loss of approxi-mately 4% which indicates loss of carboxylic group. Degradation further

rentMWNTs loadings of 0%, 0.1%, 0.25%, 0.5%, and 1.0% in the casting solution (from left to

55

60

65

70

0 0.25 0.5 0.75 1

Co

nta

ct A

ng

le (

θ)

Concentration of DDA-MWNTs (wt. %)

Fig. 9.Water contact angles of fabricated PSf/DDA-MWNTs nanocomposite membranes atdifferent DDA-MWNTs loadings in the casting solution.

140 A. Khalid et al. / Desalination 367 (2015) 134–144

past 584 °C indicates nanotubes itself loss. On the contrary, the DDA-MWNTs showed more than 50% degradation between the temperaturerange of 156 °C and 426 °C in two stages. The first stage is related to thedegradation of attached DDA groups obvious by peak at 248 °C whilethe latter is linked with residual oxygenated groups that did not reactwith DDA (peak at 358 °C) [29].

Fig. 10. Surface morphology of the fabricated nanocomposite PSf/DDA-MWNTs membranes a(c) and 1.0% (d).

3.2. Characterization of nanocomposite PSf/DDA-MWNTs membranes

3.2.1. Compatibility of DDA-CNTs with PSf matrixAs mentioned earlier, homogeneous dispersion and compatibility of

CNT additive with polymeric matrix is very crucial for efficient utiliza-tion of nanotubes properties, long term stability and durability of theprepared composite membranes. Lannoy et al. [20] have already studiedthe leaching effect of carboxylated CNTs from PSf membrane and quanti-fied nanotubes loss using a UV–vis spectrophotometer. In our case, as PSf/O-MWNTs cast film was immersed in coagulation bath (water), two ob-servations were recorded. First, black plumes emanating on different sur-face locations of precipitatedmembranewere visually evidencedwhich isin accordancewith the findings of [20]. In actual, nanotubes leached fromthe casted solution towards aqueous interface due to the improved hy-drophilicity and greater affinity of O-MWNTs for water [14,20]. Carbonnanotubes are migrated to membrane surface and leached out of mem-brane subsequently. Second, the CNTs loss is also dependent on the vis-cosity of the dope solution. Higher viscosities of cast solution cause lessnanotube loss as movement is retarded due to delayed interdiffusion ofsolvent andnon-solvent during phase inversion process [42,43]. The cast-ing solution consisting of 0.25% O-MWNTs exhibited higher CNTsleaching as compared to the casting solution with 0.5% and 1.0%O-MWNTs loading.

On the contrary, DDA-MWNTs have an enhanced compatibility withPSf matrix comparingwith O-MWNTs. During the immersion precipita-tion of PSf/DDA-MWNTs casting solution, no small black plumes of CNTswere originating from the precipitated polymer into the coagulation

t different DDA-MWNTs loadings (wt.%) in the casting solutions: 0% (a), 0.25% (b), 0.5%

8

10

12

14

16

x (L

/hr.

m2)

Water

BSA

(b) 1%

(a) 0%

Fig. 11.AFM images of fabricated pristine PSfmembrane (a) andnanocomposite PSf/DDA-MWNTs membrane at 1.0% DDA-MWNTs loading in the casting solutions (b).

141A. Khalid et al. / Desalination 367 (2015) 134–144

bath as opposed to O-MWNTs. This proves that the DDA-MWNTs en-hanced the interfacial adhesion and stability of modified CNTs withPSf matrix. This might be explained by the fact that long alkyl chain inDDA molecule attached on the MWNTs surface has higher affinity forpolymer and hence the enhanced tendency of entanglement is expected[26,27]. Also, as reported previously [15,44,45], the presence of alkylgroup offers multiple sites for hydrogen bonding between the amidelinkage of modified CNTs and sulfonic group of PSf (Fig. 7).

The photo of pristine PSf and DDA-MWNTs embedded PSf mem-branes are shown in Fig. 8. All active membrane sides were darker incolor as compared to the support side (not shown here). By increasingthe percentage ofMWNTs in the neat PSfmembrane, the top surface be-comes darker in color, suggesting the transportation ofMWNTs towardsthe top layer of the membrane surface. It can be clearly observed thatsurface images of PSf/O-MWNTs nanocompositemembranes are darkerthan PSf/DDA-MWNTs supporting the argument of improved hydrophi-licity and greater affinity of O-MWNTs for water coagulation bath priorto leaching out of membrane.

3.2.2. Contact angle measurementsAs seen in Fig. 9, water contact angle on the surface of nanocompos-

ite membranes reduces with an increase of MWNTs loading in the cast-ing solution for membrane fabrication. For example, the pristine PSfmembrane has a contact angle of 66° while the membrane preparedwith 0.25 wt.% of DDA-MWNTs loading indicated the lowest contactangle of 59°. Obviously, wrapping of PVP around DDA-MWNTs reducesthe hydrophobicity of the carbon nanotubes [16,46] and this effect leadsto smaller contact angle values on the membrane surface. It should benoted, however, that at DDA-MWNTs loadings higher than 0.25 wt.%the contact angle slightly increases. Probably at such conditions it isnot easy to control the arrangement of CNTs into polymermatrix, there-fore irregular positioning of CNTs did not enhance the membrane hy-drophilicity [15,21,47].

3.2.3. Surface morphology and porosityThe surface morphology of PSf/DDA-MWNTs nanocomposite mem-

branes is presented in Fig. 10. As seen SEM images depict the nodularstructure of the fabricated membranes. Owing to hydrophilic effect offunctionalized MWNTs [21], we expect the spinodal demixing whenfast diffusion of solvent and non-solvent during coagulation step ofphase inversion process facilitates CNTs collocation in the membranesurface to form a surface nodular structure [42]. The same observationswere reported previously by Qiu et al. [21] and Vatanpour et al. [48]. Asseen in Table 1 the nanocomposite membrane casted at 0.25 wt.% DDA-MWNTs loading has the highest values of pore size and surface porosity.However, when DDA-MWNTs loading further increases, the membranepore size and porosity start decreasing. These findingsmay be explainedby the delayed solution demixing and enhanced kinetic hindrance dueto increased viscosity of the casting solution. The increased viscositysupports out diffusion of solvent from the solution over the inside diffu-sion of non-solvent (water) into the cast film, resulting in lower poresize and membrane porosity [42,43].

The surface topography was also characterized by 3D AFM imagesat scan area of 5 × 5 μm as shown in Fig. 11. The surface roughnesswas evaluated by probing three different random locations on themem-brane. The highest and lowest locations on the membrane surface are

Table 1Total porosity and equivalent diameter of surface pores of PSf/DDA-MWNTs membranes.

Membrane(DDA-MWNTs loading)

Porosity,% Equivalent pore diameter (nm)

PSf-1 (0%) 39.24 ± 0.77 3.92 ± 0.25PSf-2 (0.1%) 42.14 ± 0.72 4.33 ± 0.62PSf-3 (0.25%) 54.57 ± 0.61 6.03 ± 0.95PSf-4 (0.5%) 48.70 ± 1.23 5.12 ± 0.72PSf-5 (1.0%) 46.04 ± 0.69 3.15 ± 0.45

represented by bright and dark regions in the images, respectively.Upon adding of 1wt.% DDA-MWNTs in the casting solution, a lower sur-face roughness (Ra of 6.277) is observed as compared to bare PSf mem-brane (Ra is 9.244). This finding might be explained by the fact that ahigh MWNTs loading leads to an increase in viscosity of the casting so-lution. This hinders the exchange rate of solvent and non-solvent diffu-sion during immersion precipitation process, and hence, smoothermembrane surface is formed.

0

2

4

6

0.00 0.25 0.50 0.75 1.00

Flu

Concentration of DDA-MWNTs (wt. %)

Fig. 12. Fluxes of PSf/DDA-MWNTsmembranes fabricated at different DDA-MWNTs loadingsduring filtration of DI water and 0.2 g/l BSA solution.

0

20

40

60

80

100

0.00 0.10 0.25 0.50 1.00

4335 39

17 23

51 48 50

2939

5765

61

8377

Per

cen

tag

e (%

)

Concentration of DDA-MWNTs (wt. %)

Reversible Resistance (%) Irreversible Resistance (%)

Total Resistance (%) Flux Recovery Ratio (%)

Fig. 13. Fouling resistances and flux recovery of nanocomposite membranes (%).

142 A. Khalid et al. / Desalination 367 (2015) 134–144

3.3. Filtration experiments

3.3.1. Pure water flux of membranesThe pure water flux of PSf membranes fabricated at different DDA-

MWNTs loadings in the casting solutions is presented in Fig. 12. Asseen the nanocomposite membranes showed a significant increase influx (145% to 275%) compared with the pristine PSf membrane. Theorder ofwater fluxwith respect to DDA-MWNTs loadingwas as follows:0.25% N 0.5% N 0.1% ≥ 1.0% N 0%. These findings might be explainedbased on the porosity data of the fabricated membranes. As seen inTable 1 the porosity of the composite membranes increases with an in-crease of DDA-MWNTs loading from 0 to 0.25 wt.%. Higher membranesporosity results in higher water flux values for the composite membrane.However, at higher CNTs loading (N0.25wt.%) a denser membrane struc-ture of lower porosity is formed (Table 1) due to the increased viscosity ofthe casting solution. This results in lower water flux of the membranesamples.

It should be also noted that MWNTs loading in the casting solutionlargely affects the pore size of fabricated nanocomposite membranes(Table 1). The highest average pore sizewas obtained for themembranecasted at 0.25 wt.% DDA-MWNTs loading, that correlates well withmembrane water flux. It should be also noted that despite its smallerpore size the membrane casted at 1.0 wt.% DDA-MWNTs loading had ahigher water flux than a pristine PSf membrane, obviously because ofhigher porosity for nanocomposite membrane.

The increase in water flux for PSf/DDA-MWNTs membranesmight also be attributed to the hydrophilicity of the prepared mem-branes. As seen in Fig. 9 the hydrophilicity of nanocomposite mem-branes increases with increase of DDA-MWNTs content from 0 to0.25 wt.% which causes an increase in water flux values. HoweverDDA-MWNTs loadings higher than 0.25wt.% lead tofluxdecline probablydue to uncontrolled and irregular positioning of CNTs in the polymerma-trix which does not enhance membrane hydrophilicity as previously de-scribed in Section 3.2.2.

3.3.2. Membrane performance at filtration of BSA solutionsAs seen in Fig. 12, the membrane flux at filtration of BSA solution

is lower compared to pure water flux obviously because of adsorption/deposition of protein molecules on the membrane surface.

However, all PSf/DDA-MWNTs composite membranes performedbetter than the neat PSf membrane during the filtration tests; probablydue to improved hydrophilicity of the composite membranes (Fig. 12).It is well known that higher hydrophilicity weakens the membrane

foulingwith organic foulants, thus enhancing the anti-fouling resistanceandflux performance [49]. The smoothening of the surface roughness ofcomposite membranes as seen from the AFM data (Fig. 11) may alsocontributes to improved flux of the composite membranes because itis well recognized that higher surface roughness, causes large mem-brane fouling [49–51].

3.3.3. Anti-fouling performance of PSf/DDA-MWNTs nanocompositemembranes

To evaluate the anti-fouling performance of the fabricated PSfmembranes, flux recovery, total fouling resistance, reversible andirreversible fouling resistances were quantified during filtration experi-ments with BSA solution as described in Section 2.6.

As shown in Fig. 13, the total fouling membrane resistance (Rt)during filtration of BSA solutions decreased from 51% for the pris-tine PSf membrane to 29% for the nanocomposite membrane with0.5 wt.% DDA-MWNTs loading. However, at lower DDA-MWNTsloading of 0.1–0.25 wt.%, values of total flux losses were approxi-mately similar.

No significant percentage change in reversible fouling resistances ofnanocomposite membranes was observed, but the irreversible resis-tance of PSf/DDA-MWNTsmembranes decreasedwith CNTs loading as-suming that the irreversiblemembrane fouling controls the total foulingresistance of the membrane. The pristine PSf membrane exhibits thehighest irreversible resistance value (43% out of 51% total resistance)obviously because of higher hydrophobicity, which facilitates the pro-tein fouling [52].

As seen in Fig. 13, the flux recovery ratio of PSf/DDA-MWNTs mem-branes is higher than for the pristine PSf one. The nanocompositemem-brane fabricated at 0.5 wt.% DDA-MWNTs loading displayed the highestflux recovery (83%) and lowest total flux loss (29%) with reduced irre-versible resistance (17%). It should be noted that fouling resistances ofthe membrane with 1.0 wt.% DDA-MWNTs loading were higher com-pared with the membrane of 0.5 wt.% DDA-MWNTs content, obviouslybecause of the allocation of MWNTs mainly in the polymer body ratherthan on the surface of the first membrane [53].

The improved anti-fouling properties of PSf/DDA-MWNTs mem-branes can be attributed to their improved hydrophilicity and surfaceroughness. Reduced hydrophobic interaction between the BSA mole-cules and the membrane surface due to the improved hydrophilicityas well as smoothening of the membrane surface might contribute toenhanced anti-fouling performance of the fabricated nanocompositemembranes [54].

143A. Khalid et al. / Desalination 367 (2015) 134–144

4. Conclusion

Novel nanocomposite membranes were fabricated from PSf andMWNTs functionalized with DDA. The amide functionalized MWNTsprovide better segregation and dispersion of nanotubes in the castingsolutions as well as improve interfacial compatibility and stability ofcomposite membranes by binding carbon nanotubes with PSf matrix.Itwas shown that addition of DDA-functionalizedMWNTs in the castingsolutions significantly affects porosity, pore size, surface roughness andhydrophilicity of the fabricated membranes. It was found that nano-composite PSf/DDA-MWNTs membranes displayed enhanced foulingresistance and flux recovery than the pristine PSf membrane during fil-tration of BSA solutions. The composite membrane prepared with0.5 wt.% of MWNTs loading displayed the highest flux recovery (83%)and lowest total flux loss (29%) with reduced irreversible fouling resis-tance (17%).

Acknowledgment

The authorswish to thankfully acknowledge King Fahd University ofPetroleum & Minerals, Dhahran, Saudi Arabia for providing financialsupport for this research via Project No. R5-CW-08 under the Centrefor Research Collaboration in Clean Water and Clean Energy at KFUPMand MIT.

References

[1] M. Mulder, Basic Principles of Membrane Technology, Kluwer Academic Publishers,1996.

[2] L.K. Wang, J.P. Chen, Y.T. Hung, N.K. Shammas, Membrane and Desalination Technolo-gies, Humana Press, 2011.

[3] T.A. Tweddle, O. Kutowy,W.L. Thayer, S. Sourirajan, Polysulfone ultrafiltrationmem-branes, Ind. Eng. Chem. Prod. Res. Dev. 22 (1983) 320–326.

[4] Q. Shi, Y. Su, S. Zhu, C. Li, Y. Zhao, Z. Jiang, A facile method for synthesis of pegylatedpolyethersulfone and its application in fabrication of antifouling ultrafiltrationmembrane, J. Membr. Sci. 303 (2007) 204–212.

[5] A.V.R. Reddy, H.R. Patel, Chemically treated polyethersulfone/polyacrylonitrile blendultrafiltration membranes for better fouling resistance, Desalination 221 (2008)318–323.

[6] I.H. Huisman, P. Prádanos, A. Hernández, The effect of protein–protein and protein–membrane interactions on membrane fouling in ultrafiltration, J. Membr. Sci. 179(2000) 79–90.

[7] J.F. Blanco, J. Sublet, Q.T. Nguyen, P. Schaetzel, Formation and morphology studies ofdifferent polysulfones-based membranes made by wet phase inversion process, J.Membr. Sci. 283 (2006) 27–37.

[8] D. Rana, T. Matsuura, Surface modifications for antifouling membranes, Chem. Rev.110 (2010) 2448–2471.

[9] Z. Spitalsky, D. Tasis, K. Papagelis, C. Galiotis, Carbon nanotube-polymer composites:chemistry, processing, mechanical and electrical properties, Prog. Polym. Sci. 35(2010) 357–401.

[10] S. Bose, R.A. Khare, P. Moldenaers, Assessing the strengths and weaknesses ofvarious types of pre-treatments of carbon nanotubes on the properties of polymer/carbon nanotubes composites: a critical review, Polymer 51 (2010) 975–993.

[11] R. Andrews, M.C. Weisenberger, Carbon nanotube polymer composites, Curr. Opin.Solid State Mater. Sci. 8 (2004) 31–37.

[12] M. Mohammad, K.I. Winey, Polymer nanocomposites containing carbon nanotubes,Macromolecules 39 (2006) 5194–5205.

[13] Michael F.L. De Volder, Sameh H. Tawfick, Ray H. Baughman, A. John Hart, Carbonnanotubes: present and future commercial applications, Science 339 (2013)535–539.

[14] J.H. Choi, J. Jegal, W.N. Kim, Fabrication and characterization of multi-walled carbonnanotubes/polymer blend membranes, J. Membr. Sci. 284 (2006) 406–415.

[15] E. Celik, H. Park, H. Choi, H. Choi, Carbon nanotube blended polyethersulfone mem-branes for fouling control in water treatment, Water Res. 45 (2011) 274–282.

[16] Yaocihuatl Medina-Gonzalez, Jean-Christophe Remigy, Sonication-assisted prepara-tion of pristine MWCNT–polysulfone conductive microporous membranes, Mater.Lett. 65 (2011) 229–232.

[17] H. Wu, B. Tang, P. Wu, Novel ultrafiltration membranes prepared from a multi-walled carbon nanotubes/polymer composite, J. Membr. Sci. 362 (2010) 374–383.

[18] H.A. Shawky, S.R. Chae, S. Lin, M.R. Wiesner, Synthesis and characterization of a carbonnanotube/polymer nanocomposite membrane for water treatment, Desalination 272(2011) 46–50.

[19] C.F. de Lannoy, D. Jassby, D.D. Davis, M.R. Wiesner, A highly electrically conductivepolymer-multiwalled carbon nanotube nanocomposite membrane, J. Membr. Sci.415–416 (2012) 718–724.

[20] Charles-François de Lannoy, Elif Soyer, Mark R. Wiesner, Optimizing carbonnanotube-reinforced polysulfone ultrafiltration membranes through carboxylicacid functionalization, J. Membr. Sci. 447 (2013) 395–402.

[21] S. Qiu, L. Wu, X. Pan, L. Zhang, H. Chen, C. Gao, Preparation and properties offunctionalized carbon nanotube/PSF blend ultrafiltration membranes, J. Membr.Sci. 342 (2009) 165–172.

[22] L. Brunet, D.Y. Lyon, K. Zodrow, J.-C. Rouch, B. Caussat, P. Serp, J.-C. Remigy, M.R.Wiesner, P.J.J. Alvarez, Properties of membranes containing semi-dispersed carbonnanotubes, Environ. Eng. Sci. 25 (2008) 565–576.

[23] M.A. Aroon, F. Ismail, M.M. Montazer-Rahmati, T. Matsuura, Effect of chitosan as afunctionalization agent on the performance and separation properties of polyimide/multi-walled carbon nanotubes mixed matrix flat sheet membranes, J. Membr. Sci.364 (2010) 309–317.

[24] T. Xia, M. Kovochich, J. Brant, M. Hotze, J. Sempf, T. Oberley, C. Sioutas, J.I. Yeh, M.R.Wiesner, A.E. Nel, Comparison of the abilities of ambient and manufactured nano-particles to induce cellular toxicity according to an oxidative stress paradigm,Nano Lett. 6 (2006) 1794–1807.

[25] K. Balasubramanian, M. Burghard, Chemically functionalized carbon nanotubes,Small 1 (2005) 180–192.

[26] S. Kim, L. Chen, J.K. Johnson, E. Marand, Polysulfone and functionalized carbon nano-tube mixed matrix membranes for gas separation: theory and experiment, J.Membr. Sci. 294 (2007) 147–158.

[27] L. Cao, H. Chen, M. Wang, J. Sun, X. Zhang, F. Kong, Photoconductivity study of mod-ified carbon nanotube/oxotitanium phthalocyanine composites, J. Phys. Chem. B 106(2002) 8971–8975.

[28] V.T. Le, C.L. Ngo, Q.T. Le, T.T. Ngo, D.N. Nguyen, M.T. Vu, Surface modification andfunctionalization of carbon nanotube with some organic compounds, Adv. Nat. Sci.Nanosci. Nanotechnol. 4 (2013).

[29] M.C.F. Soares, M.M. Viana, Z.L. Schaefer, V.S. Gangoli, Y. Cheng, V. Caliman, M.S.Wong, G.G. Silva, Surface modification of carbon black nanoparticles bydodecylamine: thermal stability and phase transfer in brine medium, Carbon 72(2014) 287–295.

[30] S.S. Kish, A. Rashidi, H.R. Aghabozorg, L. Moradi, Increasing the octane numberof gasoline using functionalized carbon nanotubes, Appl. Surf. Sci. 256 (2010)3472–3477.

[32] R. Chapman, E. Ostuni, M. Liang, G. Meluleni, E. Kim, L. Yan, G. Pier, S. Warren, G.Whitesides, Polymeric thin films that resist the adsoption of proteins and theadhesion of bacteria, Langmuir 17 (2001) 1225–1233.

[33] A. Matin, Z. Khan, S.M.J. Zaidi, M.C. Boyce, Biofouling in reverse osmosis membranesfor seawater desalination: phenomena and prevention, Desalination 281 (2011)1–16.

[34] F.A. Abuilaiwi, T. Laoui, M. Al-harthi, M.A. Atieh, Modification and functionalizationof multiwalled carbon nanotube (MWCNT) via fischer esterification, Arab. J. Sci. Eng.35 (2010) 37–48.

[35] B. Chakrabarty, A.K. Ghoshal, M.K. Purkait, Preparation, characterization and perfor-mance studies of polysulfone membranes using PVP as an additive, J. Membr. Sci. 315(2008) 36–47.

[36] Hideto Matsuyama, Taisuke Maki, Masaaki Teramoto, Kiyotaka Kobayashi, Effect ofPVP additive on porous polysulfone membrane formation by immersion precipita-tion method, Sep. Sci. Technol. 38 (2003) 3449–3458.

[37] Q.-Z. Zheng, P. Wang, Y.-N. Yang, D.-J. Cui, The relationship between porosity and ki-netics parameter ofmembrane formation in PSF ultrafiltrationmembrane, J. Membr.Sci. 286 (2006) 7–11.

[38] M.M.A. Shirazi, A. Kargari, S. Bazgir, M. Tabatabaei, M.J.A. Shirazi, Characterization ofelectrospun polystyrene membrane for treatment of biodiesel's water-washingeffluent using atomic force microscopy, Desalination 329 (2013) 1–8.

[39] M.M.A. Shirazi, D. Bastani, A. Kargari, M. Tabatabaei, Characterization of polymericmembranes for membrane distillation using atomic force microscopy, Desalin.Water Treat. 51 (2013) 6003–6008.

[40] Y. Wang, T. Wang, Y. Su, F. Peng, H. Wu, Z. Jiang, Remarkable reduction of irrevers-ible fouling and improvement of the permeation properties of poly (ether sulfone)ultrafiltration membranes by blending with pluronic F127, Langmuir 21 (2005)11856–11862.

[41] F. Avilés, J.V. Cauich-Rodríguez, L. Moo-Tah, A. May-Pat, R. Vargas-Coronado, Evalu-ation of mild acid oxidation treatments for MWCNT functionalization, Carbon 47(2009) 2970–2975.

[42] M. Han, S. Nam, Thermodynamic and rheological variation in polysulfone solutionby PVP and its effect in the preparation of phase inversion membrane, J. Membr.Sci. 202 (2002) 55–61.

[43] I.M. Wienk, R.M. Boom, M.A.M. Beerlage, A.M.W. Bulte, C.A. Smolders, H.Strathmann, Recent advances in the formation of phase inversion membranesmade from amorphous or semi-crystalline polymers, J. Membr. Sci. 113 (1996)361–371.

[44] C. Rong, G. Ma, S. Zhang, L. Song, Z. Chen, G. Wang, P.M. Ajayan, Effect of carbonnanotubes on the mechanical properties and crystallization behavior of poly(etherether ketone), Compos. Sci. Technol. 70 (2010) 380–386.

[45] X. Wei, Z. Wang, J. Wang, S. Wang, A novel method of surface modification topolysulfone ultrafiltration membrane by preadsorption of citric acid or sodiumbisulfite, Membr. Water Treat. 3 (2012) 35–49.

[46] Michael J. O' Connell, Peter Boul, Lars M. Ericson, Chad Huffman, Yuhuang Wang,Erik Haroz, Cynthia Kuper, Jim Tour, Kevin D. Ausman, Richard E. Smalley, Reversiblewater-solubilization of single-walled carbon nanotubes by polymer wrapping, Chem.Phys. Lett. 324 (2001) 265–271.

[47] J. Yin, G. Zhu, B. Deng, Multi-walled carbon nanotubes (MWNTs)/polysulfone (PSU)mixed matrix hollow fiber membranes for enhanced water treatment, J. Membr. Sci.437 (2013) 237–248.

[48] V. Vatanpour, S.S. Madaeni, R. Moradian, S. Zinadini, B. Astinchap, Novel anti-bifouling nanofiltration polyethersulfone membrane fabricated from embeddingTiO2 coated multiwalled carbon nanotubes, Sep. Purif. Technol. 90 (2012) 69–82.

144 A. Khalid et al. / Desalination 367 (2015) 134–144

[49] V. Kochkodan, N. Hilal, A comprehensive review on surfacemodified polymermem-branes for biofouling mitigation, Desalination 356 (2015) 187–207.

[50] E.M. Vrijenhoek, S. Hong, M. Elimelech, Influence of membrane surface propertieson initial rate of colloidal fouling of reverse osmosis and nanofiltration membranes,J. Membr. Sci. 188 (2001) 115–128.

[51] V. Kochkodan, D. Johnson, N. Hilal, Polymeric membranes: surface modification forminimizing (bio) colloidal fouling, Adv. Colloid Interf. Sci. 206 (2014) 116–140.

[52] P. Déjardin, Proteins at Solid–Liquid Interfaces, Springer, 2006.

[53] E. Celik, L. Liu, H. Choi, Protein fouling behavior of carbon nanotube/polyethersulfonecomposite membranes during water filtration, Water Res. 45 (2011) 5287–5294.

[54] V. Vatanpour, S.S. Madaeni, R. Moradian, S. Zinadini, B. Astinchap, Fabrication andcharacterization of novel antifouling nanofiltration membrane prepared from oxi-dized multiwalled carbon nanotube/polyethersulfone nanocomposite, J. Membr.Sci. 375 (2011) 284–294.