CF4 plasma-modified superhydrophobic PVDF membranes for directcontact membrane distillation

Chi Yang a, Xue-Mei Li a, Jack Gilron b, Ding-feng Kong a, Yong Yin a, Yoram Oren b,Charles Linder b, Tao He a,n

a Lab for Membrane Materials and Separation Technology, Shanghai Advanced Research Institute, Chinese Academy of Sciences, Shanghai 201203, Chinab Desalination and Water Treatment, Zuckerberg Institute for Water Research, Ben Gurion University, Beersheba, Southern District 84105, Israel

a r t i c l e i n f o

Article history:Received 30 September 2013Received in revised form31 December 2013Accepted 6 January 2014Available online 13 January 2014

Keywords:Membrane distillationPlasmaContact angleSuperhydrophobicPVDF

a b s t r a c t

High performance superhydrophobic PVDF flat sheet membranes for direct contact membrane distilla-tion (DCMD) were prepared via CF4 plasma surface modification. The performance of the virgin and CF4modified membranes was examined using sodium chloride solution (4 wt%) as feed in a DCMD process.Results reveal that CF4 surface modification yielded superhydrophobic membranes (water contact angle:162.41) with about 30% higher water flux than the virgin PVDF membrane. The underlying mechanismthat leads to the significant improvement in membrane performance was investigated in detail withrespect to membrane morphology, gas permeability, liquid entry pressure, and contact angle measure-ments. Results indicate that the combination of increased gas permeability and superhydrophobicitycontributes mostly to the improved water flux.

& 2014 Elsevier B.V. All rights reserved.

1. Introduction

Membrane Distillation (MD) is a thermally driven membraneseparation process, where only vapor molecules are able to passthrough a porous hydrophobic membrane. This process is drivenby the vapor pressure difference [1]. MD is promising for waterpurification, brine management [2,3], heavy metal removal [4],food and beverage industry [5,6] and purification of pharmaceu-tical products. Direct contact membrane distillation (DCMD),among different MD operation variations, has attracted broadinterests due to its simplicity and ease of implementation, possi-bility of utilizing waste heat, and potential applications in a widerange of areas [7]. However, MD processes have not yet gainedlarge industrial-scale application because of several technicalissues, including the availability of suitable membrane materials,membrane fouling, membrane module design as well as uncertainenergy recovery and total costs [4], among which the developmentof a high flux and antifouling membrane for MD is still a keyproblem. The performance of the membrane is affected by variousfactors including membrane hydrophobicity, material chemistryand morphology-related characteristics including pore size and

porosity, and pore tortuosity. Being hydrophobic is a basic require-ment for the successful implementation of the MD process, which,as an important interfacial property, is directly related to themembrane performance stability and is reflected by membranepore wetting and membrane fouling.

Hydrophobization is a key step for the preparation of MDmembranes including direct phase separation [8–14] and surfacetreatment using perfluorodecyltriethoxysilane [15], TiO2 nanopar-ticles deposition [16], hydrophilic sodium alginate hydrogel [17],plasma polymerization of fluoropolymer [18–20] and other treat-ment methods [21]. However, after surface modification, most ofthe membranes showed a significant flux reduction which wasattributed to a compact layer of high mass transfer resistance tothe water vapor resulting from the surface modification. Althoughsuperhydrophobic membranes have been reported, the advantagesof them being superhydrophobic have not always been demon-strated nor fully understood. For example, while Razmjou et al.[16] prepared superhydrophobic membranes from commercialPVDF membrane (0.45 μm) by TiO2 nanoparticles deposition onthe membrane surface, a slight decrease of flux was observed aftermodification.

We have reported previously an increase of the water flux ofhydrophilic flat sheet and hollow fiber polyethersulfone (PES)DCMD membranes by CF4 plasma surface modification [22]. Thisplasma surface modification is different from plasma polymeriza-tion reported by Kong et al. [23,24]. Plasma polymerizationinvolves the polymerization of monomers and deposition of the

Contents lists available at ScienceDirect

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

Journal of Membrane Science

0376-7388/$ - see front matter & 2014 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.memsci.2014.01.013

n Correspondence to: Lab for Membrane Materials and Separation Technology,Shanghai Advanced Research Institute, Chinese Academy of Sciences, Bld. no.2-513, 99 Haike Road, Pudong, Shanghai 201210, China. Tel.: þ86 21 20325162;fax: þ86 21 20359025.

E-mail addresses: lixm@sari.ac.cn (X.-M. Li), het@sari.ac.cn (T. He).

Journal of Membrane Science 456 (2014) 155–161

polymeric layer on the support membrane surface, where a thickdeposited coating layer may add extra mass transfer resistance tothe membrane and thereby lower the water flux of the membrane.Instead, when we used CF4 plasma surface modification, thesurface modification was realized by etching, replacement andpossibly some polymer deposition [22,25]; thus it may not reducethe membrane flux after surface modification. However, in ourprevious report, the membranes before surface modification weresuperhydrophilic with a water contact angle of 0ο, which showedno water flux in the DCMD process, making it impossible to justifythe CF4 plasma treatment effect on the flux or mass transportresistance of the MD membrane. The application of the CF4 plasmatreatment on a hydrophobic membrane is, therefore, highly desir-able in order to assess CF4 plasma treatment effects on the masstransport resistance of the membranes.

In this report, commercial Millipore hydrophobic PVDF mem-branes were used as the base membranes to prepare super-hydrophobic membranes via the CF4 plasma modification. Thesurface modification was investigated in terms of plasma treat-ment time. The membranes were characterized by contact angle,gas permeability, and DCMD performance to elucidate the under-lying mechanism for flux enhancement after surface modification.

2. Experimental

2.1. Materials

Commercial polyvinylidene fluoride (PVDF) flat-sheet mem-branes GVHP (Millipore, nominal pore size: 0.22 μm, porosity:75%, and thickness: 125 μm) were used in this study. Sodiumchloride (NaCl, Analytical grade) was supplied by SinopharmChemical Reagent Co. Ltd. Deionized water was used in the directcontact membrane distillation (DCMD) processes.

2.2. Membrane preparation by CF4 plasma treatment

Plasma surface treatment was conducted on an IoN 40 plasmasystem (PVA TePla Co. Ltd.) equipped with parallel plate electrodescoupled with the RF plasma reaction system. The membranes werefirst placed on the plates. A pretreatment with argon plasma wascarried out to clean the substrates at 45 W for 15 s. CF4 gas wasthen introduced to the chamber at a flow rate of 250 standardcubic centimeter per minute (SCCM), and CF4 plasma dischargewas processed at 150 W for 5–60 min. Upon completion of the CF4plasma process the chamber was purged with N2 for 10 min underatmospheric pressure in order to reduce the physical adsorption ofCF4 gas on the surfaces. The samples were taken out and kept in aclean sample holder for 24 h before usage. After each CF4 deposi-tion, the chamber was cleaned by O2 plasma for 15 min, at 200 Wto avoid CF4 deposition on the electrodes.

2.3. Membrane characterization

2.3.1. SEMScanning electron microscopy samples were prepared by cryo-

genic breaking, followed by drying under vacuum at 30 1C over-night and then coated with a thin layer of gold. Field emissionscanning electron microscopy (FESEM HITACHI Japan S-4800) wasutilized for image acquisition.

2.3.2. Pore size and gas permeabilityPore size and gas permeability analyses were conducted with

Capillary Flow Porometry (Porolux 1000). The membrane was pre-wetted with commercial low surface tension liquid Porefil (surfacetension of 16 dyne/cm based on the supplier0s data sheet). After

mounting the sample onto the test cell, the measurement wasmanaged with a program consisting of wet-run and dry-run. Thewet-run was realized by replacing the wetting liquid within acertain pore size by compressed air at certain pressure till themembrane was dried out (wet-run). Then the air flow rate of themembrane was tested by decreasing the air pressure (dry-run).The bubble point was determined as the pressure at whichsignificant flow of air was detected. The mean pore size wasdetermined by fitting the wet curve with a half-dry curve,according to published work by Khayet and Matsuura [26].

2.3.3. Water contact angleWater contact angles were measured by a contact angle

goniometer (Maist Drop Meter A-100P) equipped with a highspeed CCD camera. A water droplet of 5 μL was deposited on themembrane surface for contact angle measurements, and eachreported value was the average of five measurements.

2.3.4. LEPLiquid entry pressure (LEP) is a measure of the ability of a

hydrophobic membrane against pore wetting. LEP was measuredusing a dead-end filtration set-up with deionized water. Pressureon the feed side was increased step-wise while allowing it tostabilize for a couple of minutes after each increment (0.05 bar).The pressure at which the first water droplet gets through themembrane is taken as the LEP.

2.4. Direct contact membrane distillation

A schematic illustration of the DCMD process and the testmodule is shown in Fig. 1. The PVDF membrane was tightlyclamped between two perplex plates. The effective operation areaof the membrane was 100 mm�30 mm. The aqueous sodiumchloride (NaCl) solution (4 wt%) was used as the feed and thepermeate was deionized water. To maintain a constant feedconcentration, additional deionized water was added based onthe amount of water transferred across the membrane. The feedinlet temperature (Tf) was varied in the range of 50–70 1C, and thedistillate inlet temperature (Tp) was maintained at 21.170.4 1C.The inlet and outlet temperatures of the feed and permeate werelogged by the computer. Two independent peristaltic pumps wereused to maintain a flow rate of 600 mL/min.

The permeate was collected as an overflow and logged by acomputer. The water permeate flux was calculated according to

Jw ¼ΔmAt

ð1Þ

where Jw, Δm, A and t represent the water permeate flux(kg/m2 h), mass of permeate (kg), effective membrane surfacearea (m2) and the time duration (h), respectively.

The salt rejection R was determined according to the followingequation:

R¼ 1�Cp

Cfð2Þ

where Cf and Cp represent the NaCl concentration in the feed andpermeate, respectively. An online conductivity meter (EC-4300RS,supplied by SUNTEX Instrument Ltd.) was used to monitor theconductivity of permeate. The salt concentration in the permeatewas determined based on a calibration curve of the conductivityand salt concentration.

C. Yang et al. / Journal of Membrane Science 456 (2014) 155–161156

3. Results and discussion

3.1. CF4 plasma surface modification

The plasma surface treatment process of PVDF membranes issimilar to our previous reports [22] and the surface modificationeffect was first monitored by water contact angle measurements.Fig. 2 illustrates the surface contact angle change of the PVDFmembrane against the treatment time and their correspondingphotographs of a water droplet (5 μL) sitting on top of themembranes. The membrane surface with a more compact layer(smaller pores and higher water contact angles) is denoted as thetop surface and the side with looser structure as the bottomsurface. The contact angle of the virgin PVDF membranes was130.271.11 and 121.670.41 at top and bottom surface, respec-tively. The top surface water contact angles were 147.070.91,151.370.81, and 162.471.21 at the CF4 plasma treatment time of5 min, 10 min and 15 min, respectively. Thereafter, the contactangle remained nearly constant. A similar trend in the contactangle change of the bottom surface is observed as illustrated inFig. 2.

Wettability of a surface is an interfacial property which isdetermined by the surface tension of the solid substrate and waterand is measured at the interface where air, liquid and the solidmeet. For a flat solid, the surface contact angle is determined byYoung0s equation. However, for surfaces with air pockets, thesurface contact angle is determined by the percentage of the pores(air) and the solid (here the membrane material). For a hydrophilicsubstrate, the presence of air pockets may make the surface morehydrophilic according to Wenzel0s prediction (Fig. 3a) [27]. For ahydrophobic surface, the presence of air pockets may make thesurface more hydrophobic (Fig. 3b) and even superhydrophobic(Fig. 3c). Fluorination is often used for surface modification inorder to prepare anti-sticking or self-cleaning surfaces becausefluorinated materials have a low surface tension. In our previousstudies, unsaturated polyester resins and hydrophilic polyether-sulfone membranes were made hydrophobic by CF4 plasmafluorination. The surface modification mechanism was illustratedas replacement of CH bonds with CF and CF3 with possiblepolymeric materials deposition [22], which did not change thesurface structure of the base material. Here again, we have usedXPS to monitor the fluorination process by the atomic ratiobetween fluorine and carbon (F/C ratio) as shown in Table 1. Forthe Virgin membrane, an F/C ratio of 0.89/1 was found at both thetop and bottom surfaces, which deviates about 11% from theexpected 1/1 ratio. This is nevertheless not so surprising becausecarbon contamination is unavoidable in the natural environment,

which causes a higher amount of detected carbon elements. Whenthe membrane was treated for 5 min, the F/C ratio increased to1.44 at both bottom and top surface membranes. This increaseindicates that more fluorine elements are present at membranessurfaces, agreeing to the contact angle observation that plasmatreatment increased the hydrophobicity. With further elongationof the plasma treatment time of 15 min the F/C ratio increased to1.66/1.62 at the top/bottom surface, and reached to 1.80/1.72 at20 min of treatment time, respectively. The change of the F/C ratiowith treatment is expected and conforms to our previous report.While it would be interesting to explore the fluorination in thecross-section of the membrane, the MD performance of a mem-brane is more closely related to the non-wetting surface propertiesof the membrane. It is nevertheless to be expected that the insidestructure of the membrane should show a lower degree offluorination because the plasma media must penetrate into themembrane pores before surface modification [22].

The surface morphology of the membranes was monitoredbefore and after surface modification. Virgin membrane has ratheropen and rough surface at both the top and bottom surfaces(Fig. 3d and e). With the plasma treatment for 15 min, the topsurface morphology of the membranes appears to be slightly moreopen (Fig. 3f), but the bottom surface appears unchanged (Fig. 3g).It is noteworthy that SEM cannot provide quantified informationof the membrane pore size. Thereby, based on SEM images it canbe seen that there was no visible materials deposition on themembrane surface. Moreover, the cross-section of the membraneswas analyzed; there was no obvious difference in the morphology

Fig. 2. Water contact angles of top/bottom PVDF membrane surfaces as a functionof CF4 plasma treatment time.

Fig. 1. Schematic illustration of the direct contact membrane distillation process and the test module. (1) membrane; (2) MD module; (3) digital thermometer; (4) flowmeter; (5) peristaltic pump; (6) feed tank; (7) permeate tank; (8) thermostatic bath; (9) cooler bath; (10) conductivity meter; (11) digital balance; (12) level controller; and(13) computers.

C. Yang et al. / Journal of Membrane Science 456 (2014) 155–161 157

particularly from the snapshot view of the top skin layer observed,as shown in a separate file as “Supplementary Data”.

3.2. DMCD performance

The membrane performance was evaluated in a DCMD processusing 4 wt% NaCl water solution as the feed and deionized water atthe permeate side (Fig. 4). Membranes treated at the sameconditions are considered as the same. For the same membrane,the flux increased with Tf, which agrees with the general principleof the MD process in that higher temperature can generate highervapor pressure, thus higher vapor pressure gradient and even-tually higher permeation. For comparison of the plasma treatmenteffect, it can be seen that with the increase of treatment time, theDCMD flux increases, and at about 15 min, the flux reaches about a

maximum, thereafter the membrane flux decreases with longerplasma treatment time. However, even after 60 min of treatment,the plasma modified membrane still shows a higher water fluxthan the virgin membrane. For example, at the Tf of 64 1C, thewater flux was 25.5 kg/m2 h for the virgin membrane and 32.8 and27.5 kg/m2 h for the CF4 modified membranes with 15 and 60 minof treatment, respectively. A maximum flux enhancement of about

Fig. 3. Schematic illustration of wetting states on rough surfaces and the top surface SEM images of the PVDF membranes. (a) Hydrophilic substrate, (b) hydrophobicsubstrate, (c) superhydrophobic substrate, (d, e) top and bottom surface images of Virgin membrane, and (f, g) top and bottom surfaces of the membrane CF4 plasma treatedfor 15 min.

Table 1Atomic ratio of F/C change of the PVDF membranes with CF4 plasmatreatment time.

Fluorination time (min) Top Bottom

0 0.89 0.895 1.44 1.44

15 1.66 1.6220 1.80 1.72

Fig. 4. DCMD flux of the CF4 plasma treated PVDF membranes at different feedinlet temperatures. Feed: NaCl (4 wt%), flow rate: 0.17 m/s; and Tp: 21.170.4 1C,flow rate: 0.17 m/s.

C. Yang et al. / Journal of Membrane Science 456 (2014) 155–161158

30% is observed for the plasma treated membrane. In addition, thesalt rejection for all membranes is greater than 99.98%, indicatingthat negligible salt leakage exists.

In general, the flux pattern against the plasma treatment timeshowed improved MD performance up to 15 min treatment, but adecline thereafter. When we look at the contact angle change ofthe membrane with treatment time, we find that the contact angleat 15 min was reaching close to the highest point (Fig. 2) incomparison to the DCMD water flux pattern (Fig. 4). This observa-tion seems to indicate that for these membranes, increasinghydrophobicity may lead to increased membrane DCMD flux. Ingeneral, an MD membrane performance should be determined bythe porosity, gas permeability, tortuosity and the operating con-ditions. So far, there have not been reports on the water contactangle relationship with membrane permeability.

According to our previous report [22], during the CF4 plasmatreatment, elemental fluorine elements are introduced to the PVDFmembrane surface leading to a membrane with more fluorineatoms; the gaseous species penetrate into the membrane pores orchannels inside the porous structure of the membrane, therebymodifying the interior of the membrane. Although this helps avoidpore wetting during membrane distillation, it does not necessarilyexplain a flux change. In order to understand the underlyingmechanism for flux enhancement and decline with treatmenttime, other effects of plasma modification must be examined.

3.2.1. Liquid entry pressure (LEP)The LEP of the membranes was determined experimentally

and is plotted against the treatment time for each membrane.As shown in Fig. 5, the LEP change of membranes follows anapproximately linear increase in the treatment time till 30 minwith an increase from 2.4 bar for the virgin membrane to 3.1 barfor the modified membrane at 30 min. Thereafter, the LEP for themembranes remains nearly unchanged.

LEP is an indication of the ability of a hydrophobic membraneagainst wetting in the MD process. Therefore, a membrane with ahigher LEP is expected to perform better than that with a lowerone. However, the measured LEPs of the membranes are all higherthan the vapor pressure difference of the DCMD processes; hence,LEP is not considered to be a major factor for the flux enhance-ment after CF4 plasma surface modification.

3.2.2. Pore size and gas permeabilityPlasma modification involves etching, replacement and deposi-

tion, which may alter pore size, porosity and eventually gaspermeability. The mean pore size of the surface modified mem-branes was determined by a porometer and is plotted againsttreatment time. As presented in Fig. 6, the mean pore diameter ofthe membranes increased slightly from 0.232 μm for the virgin

membrane to 0.246 μm for the surface modified membrane at150 W for 15 min and thereafter the pore size stayed nearlyconstant, a trend very much similar to the contact angle change.Interestingly, the maximum pore size also increased from0.293 μm for virgin membrane to 0.306 μm for the membranetreated for 15 min. This slight change of the membrane pore sizeobserved here confirms that plasma treatment can alter themembrane pore structure; however, the change is not significantwhen compared to membrane MD permeability.

In a membrane distillation process, the water vapor diffusesacross the porous membrane structure, driven by a vapor pressuregradient. This process is very similar to a gas permeation through aporous media at a low pressure gradient. Therefore, the gaspermeability is related to the mass transfer resistance. Fig. 7 showsthe gas permeability (ο) of the plasma modified PVDF membranesat 0.1 bar, which is in the same range as the vapor pressure.Compared to the DCMD flux, the gas permeability change shows asimilar trend as the MD flux: at 15 min, the gas permeabilityincreased about 30%. Further increase in the treatment time leadsto a slight decline. The results indicate that the gas permeabilitychange might be responsible for the enhanced MD flux for thesurface modified membrane. In order to verify this, the molar ratiobetween the MD flux (mol/m2 h) (at the feed inlet temperature of54 1C) and the gas permeability (mol/m2 h) is presented in Fig. 7(■). It should be noted that the average molecular weight of air istaken as 29. The ratio is nearly constant with variations of lessthan 6% within the treatment time. It is generally believed that theDCMD flux is mainly controlled by the diffusion resistance acrossthe membrane matrix, which is reflected by the gas permeability.Because the gas permeability has changed, the MD flux changes aswell, which might explain the flux enhancement effect of the CF4plasma modified membranes.Fig. 5. Effect of CF4 plasma treatment time on the LEP of the PVDF membranes.

Fig. 6. Mean pore size of the PVDF membranes as a function of CF4 plasmatreatment time.

Fig. 7. Gas permeability (ο) and molar ratio of MD flux (54 1C)/gas permeability (■)as a function of CF4 plasma treatment time for PVDF flat sheet membranes.

C. Yang et al. / Journal of Membrane Science 456 (2014) 155–161 159

3.2.3. Effect of argon plasma treatmentA question remains unsolved: what does the superhydropho-

bicity of the membrane contribute to its MD performance? Inorder to validate the superhydrophobic effect of the membrane,Argon plasma was utilized for surface treatment. Unlike CF4plasma, Argon plasma does not introduce new elements to thePVDF membrane surface. However, the argon plasma treatmentmay feature the same surface morphology, surface porosityetc. Therefore, the difference of the membranes after argontreatment and CF4 treatment probably reflects the influence ofthe surface chemistry or superhydrophobic characteristics onmembrane performance.

Control experiments with Ar plasma treatment was conductedfollowing a similar procedure to CF4 plasma treatment with Argonpretreatment (45 W, 15 s) and then glow discharge at 150 W in anArgon atmosphere for 5–60 min. The gas permeability of theresulting membranes with different treatment time was thenmeasured (Fig. 8). Compared to CF4 treatment, a similar gaspermeability increase trend is observed for Ar plasma treatmentup to 42% at 30 min of treatment time. Thereafter, a level-off isobserved. This observation confirms that surface plasma treatmentcan reduce the membrane mass transfer resistance. Because of itsinertness, Ar plasma treatment is a pure etching process and no Arspecies can be introduced to the surface, which can be confirmedby the contact angle of the membrane change. After surfacemodification, the contact angle of the membranes decreased withtreatment time, however, even after 60 min of treatment, themembrane stays hydrophobic, indicating that there was no sig-nificant chemical composition change on the membrane.

The MD performance of a PVDF membrane treated by Arplasma (150 W for 15 min) was evaluated and compared withboth virgin and a CF4 plasma treated membrane (150 W for15 min) as illustrated in Fig. 9. The Ar treated membrane showed

a DCMD flux of 16.6870.14 kg/m2 h, slightly lower than that ofthe virgin membrane (17.11 kg/m2 h) with a salt rejection of99.97%. Interestingly, although both plasma treated membraneshave similar gas permeability, their DCMD performance differsdramatically from each other. It should also be pointed out that thewettability of both membranes differed dramatically from eachother as well. Consequently, this comparison indicates that MDperformance of a membrane is not solely determined by the gaspermeability but also by the wettability of the membrane. More-over, superhydrophobic membranes yielded higher water flux; theunderlying mechanism is proposed in further paragraphs. Itshould be noted that superhydrophobic surfaces are potentiallyantifouling; therefore, the performance of these superhydrophobicmembranes against fouling is an interesting subject, which isunder investigation and will be addressed in the future.

According to the gas permeability test, both plasma treatmentscan lead to reduced mass transfer resistance across the membrane;however, the surface wettability of the membranes is quitedifferent. The performance difference in MD by different plasmacan then only be ascribed to the different scenario at the water/membrane interface. The wetting state of the membrane with thevirgin membrane and the superhydrophobic membrane is illu-strated in Fig. 10. The membrane surface is denoted as a roughsurface because of the presence of membrane pores. A red dottedline is marked as the solid, water and air interface. When themembrane is hydrophobic, the area for water–air interface isdenoted as S0 (Fig. 10a), which is responsible for the watervaporizing. After plasma treatment, with the CF4 plasma treat-ment, the surface becomes superhydrophobic. For the superhy-drophobic membrane a “lifting up” of the gas–liquid interface tothe top edge of the membrane is expected, which will result in aneven larger effective evaporation area S1 (Fig. 10b). This increase ineffective interface area is responsible for the vaporization of thewater molecules and thereby the MD flux is enhanced. At 15 minof CF4 plasma treatment, the contact angle reached the maximum,because the fluorination reached saturation. Thereafter, because ofthe decrease in gas permeability, the flux did not benefit furtherfrom the superhydrophobic effect and stayed nearly unchanged. Incomparison, for Argon treated membranes, although the masstransfer resistance is reduced with plasma treatment, the contactangle is also reduced. Accordingly, because of the increasedwetting, a smaller water–air interface is expected as illustratedin Fig. 10c, which leads to reduced effective vaporization area andthereby reduced MD flux. Another effect related to this is that withincreased wetting, water reaches the porous surface structure,generating a certain thickness of stagnant solution, which willenhance temperature polarization and reduce flux. This effect isstill under investigation.

Fig. 8. Effect of Ar plasma treatment time on gas permeability and water contactangle for the PVDF flat sheet membranes.

Fig. 9. Comparison of flux and salt rejection for virgin, Ar and CF4 plasma modifiedPVDF membranes. Feed: NaCl (4 wt%), Tf: 57.8 1C, flow rate: 0.17 m/s; and Tp:21.3 1C, flow rate: 0.17 m/s.

C. Yang et al. / Journal of Membrane Science 456 (2014) 155–161160

4. Conclusions

Hydrophobic PVDF flat sheet membranes were successfullyconverted to superhydrophobic ones by CF4 plasma treatmentwith an average flux enhancement of 730% and a salt rejection of99.98%. The mechanism of flux enhancement by CF4 plasmatreatment was explored. It was found that gas permeability isdirectly linked to the mass transfer resistance of a porous mem-brane. Yet, the MD performance is also controlled by the wett-ability of the membrane. The advantages of superhydrophobicmembranes are proposed to be superhydrophobic surfaces con-tribute larger air, membrane, and water three-phase interface, andconsequently a larger effective water vaporization surface area andthereby enhance significantly the membrane performance.Furthermore, because of the non-wetting and self-cleaning prop-erties of the superhydrophobic membranes, they are potentiallyuseful for treatment of highly complicated feed waters withcomplicated compositions including shale gas drilling flow backfluids. However, the influence of salinity and various organicfoulants on the stability of the superhydrophobic membranesurface requires systematic research. Currently, we are conductinga long-term stability measurement of the membrane in a syntheticsolution with defined salinity and foulants.

Acknowledgments

The authors are grateful for the partial financial support fromNational Natural Science Fund China (Project nos. 21176119 and51206114), the National Key Basic Research Program of China (973Program) (Project nos. 2012CB932800 (TH) and 2009CB623402).

Appendix A. Supplementary information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.memsci.2014.01.013.

References

[1] A. Alkhudhiri, N. Darwish, N. Hilal, Membrane distillation: a comprehensivereview, Desalination 287 (2012) 2–18.

[2] A.M. Alklaibi, N. Lior, Membrane-distillation desalination: status and potential,Desalination 171 (2005) 111–131.

[3] F. Lagan, G. Barbieri, E. Drioli, Direct contact membrane distillation: modellingand concentration experiments, J. Membr. Sci. 166 (2000) 1–11.

[4] P.P. Zolotarev, V.V. Ugrozov, I.B. Volkina, V.M. Nikulin, Treatment of wastewater for removing heavy metals by membrane distillation, J. Hazard. Mater.37 (1994) 77–82.

[5] V.N. Bui, M.H. Nguyen, J. Muller, Challenge and Opportunity of Direct ContactMembrane Distillation in Liquid Food Concentration, Chemeca 2006: Knowl-edge and Innovation, Auckland, N.Z., Engineers Australia, 2006, pp. 321–326.

[6] V.D. Alves, I.M. Coelhoso, Orange juice concentration by osmotic evaporation andmembrane distillation: a comparative study, J. Food Eng. 74 (2006) 125–133.

[7] M.S. El-Bourawi, Z. Ding, R. Ma, M. Khayet, A framework for better understandingmembrane distillation separation process, J. Membr. Sci. 285 (2006) 4–29.

[8] M. Khayet, C.Y. Feng, T. Matsuura, Morphological study of fluorinated asym-metric polyetherimide ultrafiltration membranes by surface modifying macro-molecules, J. Membr. Sci. 213 (2003) 159–180.

[9] M. Khayet, T. Matsuura, J.I. Mengual, Porous hydrophobic/hydrophilic compo-site membranes: estimation of the hydrophobic-layer thickness, J. Membr. Sci.266 (2005) 68–79.

[10] M. Khayet, J.I. Mengual, T. Matsuura, Porous hydrophobic/hydrophilic compo-site membranes: application in desalination using direct contact membranedistillation, J. Membr. Sci. 252 (2005) 101–113.

[11] M. Qtaishat, M. Khayet, T. Matsuura, Novel porous composite hydrophobic/hydrophilic polysulfone membranes for desalination by direct contact mem-brane distillation, J. Membr. Sci. 341 (2009) 139–148.

[12] M. Qtaishat, T. Matsuura, M. Khayet, K.C. Khulbe, Comparing the desalinationperformance of SMM blended polyethersulfone to SMM blended polyether-imide membranes by direct contact membrane distillation, Desal. Water Treat.5 (2009) 91–98.

[13] M. Qtaishat, D. Rana, M. Khayet, T. Matsuura, Preparation and characterization ofnovel hydrophobic/hydrophilic polyetherimide composite membranes for desali-nation by direct contact membrane distillation, J. Membr. Sci. 327 (2009) 264–273.

[14] M. Qtaishat, D. Rana, T. Matsuura, M. Khayet, Effect of surface modifyingmacromolecules stoichiometric ratio on composite hydrophobic/hydrophilicmembranes characteristics and performance in direct contact membranedistillation, AIChE J. 55 (2009) 3145–3151.

[15] Z.D. Hendren, J. Brant, M.R. Wiesner, Surface modification of nanostructuredceramic membranes for direct contact membrane distillation, J. Membr. Sci.331 (2009) 1–10.

[16] A. Razmjou, E. Arifin, G. Dong, J. Mansouri, V. Chen, Superhydrophobicmodification of TiO2 nanocomposite PVDF membranes for applications inmembrane distillation, J. Membr. Sci. 415–416 (2012) 850–863.

[17] J.B. Xu, S. Lange, J.P. Bartley, R.A. Johnson, Alginate-coated microporous PTFEmembranes for use in the osmotic distillation of oily feeds, J. Membr. Sci. 240(2004) 81–89.

[18] H. Lee, F. He, L. Song, J. Gilron, K.K. Sirkar, Desalination with a cascade of cross-flow hollow fiber membrane distillation devices integrated with a heatexchanger, AIChE J. 57 (2011) 1780–1795.

[19] B. Li, K.K. Sirkar, Novel membrane and device for direct contact membranedistillation-based desalination process, Ind. Eng. Chem. Res. 43 (2004) 5300–5309.

[20] B. Li, K.K. Sirkar, Novel membrane and device for vacuum membranedistillation-based desalination process, J. Membr. Sci. 257 (2005) 60–75.

[21] Y. Liao, R. Wang, A.G. Fane, Engineering superhydrophobic surface on poly(vinylidene fluoride) nanofiber membranes for direct contact membranedistillation, J. Membr. Sci. 440 (2013) 77–87.

[22] X. Wei, B. Zhao, X.-M. Li, Z. Wang, B.-Q. He, T. He, B. Jiang, CF4 plasma surfacemodification of asymmetric hydrophilic polyethersulfone membranes fordirect contact membrane distillation, J. Membr. Sci. 407 (2012) 164–175.

[23] Y. Wu, Y. Kong, X. Lin, W. Liu, J. Xu, Surface-modified hydrophilic membrane andits applications in membrane distillation, J. Funct. Polym. 4 (1991) 199–206.

[24] Y. Wu, Y. Kong, X. Lin, W. Liu, J. Xu, Surface-modified hydrophilic membranesin membrane distillation, J. Membr. Sci. 72 (1992) 189–196.

[25] G.L. Li, X. Wei, W.J. Wang, T. He, X.M. Li, Modification of unsaturated polyesterresins (UP) and reinforced UP resins via plasma treatment, Appl. Surf. Sci. 257(2010) 290–295.

[26] M. Khayet, T. Matsuura, Pervaporation and vacuum membrane distillationprocesses: modeling and experiments, AIChE J. 50 (2004) 1697–1712.

[27] R.N. Wenzel, Resistance of solid surfaces to wetting by water, Ind. Eng. Chem.28 (1936) 988–994.

Fig. 10. Schematic illustration of wetting states and gas–liquid interfaces of membranes with different hydrophobicities. (a) Virgin hydrophobic membranes,(b) superhydrophobic membranes, and (c) hydrophilic membranes (For better interpretation of this figure, the reader is referred to the web version of this article as incolour.).

C. Yang et al. / Journal of Membrane Science 456 (2014) 155–161 161