Pa

Sa

b

a

ARRAA

KPPHVD

1

wlas(sMboophueh

df

(

0d

Journal of Membrane Science 364 (2010) 219–232

Contents lists available at ScienceDirect

Journal of Membrane Science

journa l homepage: www.e lsev ier .com/ locate /memsci

reparation of hollow fibre membranes from PVDF/PVP blends and theirpplication in VMD

. Simonea,b, A. Figoli a,∗, A. Criscuoli a,∗, M.C. Carnevaleb, A. Rosselli a, E. Drioli a,b

Institute on Membrane Technology, ITM-CNR, c/o University of Calabria, Via P. Bucci Cubo 17/C, 87030 Rende (CS), ItalyDepartment of Chemical Engineering and Materials, University of Calabria, Via P. Bucci Cubo 42/A, 87030 Rende (CS), Italy

r t i c l e i n f o

rticle history:eceived 5 March 2010eceived in revised form 28 July 2010ccepted 9 August 2010vailable online 14 August 2010

eywords:

a b s t r a c t

Microporous hydrophobic poly(vinylidene fluoride) (PVDF) hollow fibres were prepared by the dry–wetspinning technique. N,N-Dimethylformamide (DMF) was used as solvent, while water and poly(vinylpyrrolidone) (PVP) were used as pore forming additives. Mixtures of DMF or ethanol in water wereemployed as bore fluids.

The influence of different parameters on the fibres characteristics was explored. Particular attentionwas focused on the PVP concentration and on the composition of the bore fluid. The obtained fibres

olyvinylidene fluoride (PVDF)olyvinyl pyrrolidone (PVP)ollow fibresacuum membrane distillationesalination

exhibit good structure, excellent mechanical properties, high porosity (up to 80%) and an average poresize ranging from 0.12 to 0.27 �m. Fibres were tested in vacuum membrane distillation (VMD) config-uration, using distilled water as feed. The effect of membrane properties on the water vapor fluxes wasinvestigated. For the prepared membranes, the measured fluxes ranged between 3.5 and 18 kg/m2 h at50 ◦C and 20 mbar vacuum pressure. Some selected fibres were assessed for long term stability. Constantperformances were observed up to 2 months. Furthermore, preliminary VMD tests on salty water were

to ver

also performed, in order

. Introduction

Nowadays, it is widely recognized that, besides the globalarming, water shortage is becoming a major emergency. In

ast years, membrane distillation (MD) was proposed as a validlternative to traditional desalination techniques, such as multi-tage flash vaporization (MSFV), or coupled to reverse osmosisRO) (integrated membrane systems), for its lower energy con-umption and lower influence of osmotic pressure, respectively.embrane distillation is a separation method in which a hydropho-

ic and microporous membrane is used with a liquid feed phasen one side of the membrane and a condensing, permeate phasen the other side. The driving force for transport is the partialressure difference across the membrane. The principles of MDave been reviewed by many authors [1–4]. The possibility ofsing alternative energy sources, such as geothermal and solarnergy, as well as to exploit low grade or waste energy, has been

ighlighted [5–7].

Although based on the same principle, MD processes can beivided in different categories, depending on the method appliedor establishing the required driving force. Among the differ-

∗ Corresponding authors. Tel.: +39 0984 492027/492118; fax: +39 0984 402103.E-mail addresses: a.figoli@itm.cnr.it (A. Figoli), a.criscuoli@itm.cnr.it

A. Criscuoli).

376-7388/$ – see front matter © 2010 Elsevier B.V. All rights reserved.oi:10.1016/j.memsci.2010.08.013

ify fibres suitability for seawater desalination.© 2010 Elsevier B.V. All rights reserved.

ent MD configurations, vacuum membrane distillation (VMD) hasattracted increasing interest for various applications beside seawa-ter desalination, i.e.: removal of volatile organic compounds (VOCs)from water, concentration of aqueous solutions, and separationof non-volatile components form water such as ions, colloids andmacromolecules. The application areas range from environmentalwaste clean-up to food processing [8–18]. In VMD, the liquid feedis brought in contact with one side of a hydrophobic membrane,while vacuum is applied at the permeate side. Vapor condensationtakes place outside of the module.

VMD results in many advantages, with respect to conventionalseparation techniques, and, from an economic point of view, iscomparable to alternative membrane processes, such as pervapo-ration [19]. With respect to other MD configurations, VMD allowsto reach higher partial pressure gradients and, hence, higher fluxesand plant productivity. It performs better than DCMD in terms oftransmembrane fluxes, energy consumption/permeate flow ratiosand evaporation efficiency [20].

Recently, MD potentialities, in terms of productivity, costs andefficiency, have been examined in details [21]. Authors investigatedthe effect of different parameters on the MD performance. Mem-

brane critical features resulted to be the thermal conductivity andthe porosity.

The heat lost by conduction through the membrane reduces thethermal gradient, which is the driving force of the process, therebynegatively affecting the transport.

220 S. Simone et al. / Journal of Membrane Science 364 (2010) 219–232

Table 1Separation performance of PVDF flat sheet (FS) and hollow fibre (HF) membranes for VMD reported in the literature.

MD process Membrane Membraneproperties

Dope composition J/◦C Ref

VMD/pure H2O FS Pore size 0.22 �mPorosity 70.5%

DMAC, H2O 1.39 kg/m2 h(feed 25 ◦C)

[44]

VMD/pure H2O HF Avg. Pore size0.031 �mEffective porosity1516 m−1

DMAC, LiCl/H2O 0.5 kg/m2 h(feed 50 ◦C)

[46]

A cuum membrane distillation.

fith

rii

ccorsa

gdfrs

uaisc

ptrm

P

ds

mp

Table 3Hollow fibre spinning base conditions.

Spinning conditions

Bore fluid EtOH:H2O 30:70; DMF:H2O 15–35:85–65Bore fluid injection rate (ml/min) 7–20Bore fluid temperature (◦C) 50Coagulation bath H2O 100%Coagulation bath temperature (◦C) 20Polymeric dope temperature (◦C) 85Polymeric dope flow rate (g/min) 12

permeability, produce hydrophilic membranes and prevent foul-

TP

bbreviations: DMAc: dimethylacetamide; FS: flat sheet; HF: hollow fibre; VMD: va

Membranes having higher porosity ensure better molecular dif-usion and, therefore, higher flux. The effect of membrane thicknesss more complex. Thinner membranes offer lower resistance to theransport, but, membranes of reduced thickness leads to highereat losses by conduction.

Different papers have pointed out that the search for new mate-ials and the development of suitable membranes are of primarymportance for the future commercialization of MD, in order tomprove the process productivity and reduce its costs.

The production of membranes suitable for MD has been dis-ussed in different works. They can be divided in two mainategories: (a) surface modification to turn hydrophilic flat sheetr hollow fibre membranes into hydrophobic ones and (b) prepa-ation of flat membranes or fibres, from binary or ternary dopes,tarting from a hydrophobic polymer, and using a suitable solventnd pore forming or other additives.

Surface modifications of hydrophobic membranes, like coating,rafting or incorporation of macromolecules usually aim at the pro-uction of a membrane having reduced transport resistances. Inact, the hydrophilic part provides mechanical support with lowesistance to transport, leading to high fluxes, while the modifiedurface ensures the required hydrophobicity [22–34].

In recent years, polyvinylidene fluoride (PVDF) became a pop-lar membrane material due not only to its good hydrophobicitynd excellent chemical stability, but also to the feasibility of form-ng hollow fibres and membranes via a phase inversion method;everal examples of PVDF membrane preparation for various appli-ations are already reported in literature [35–38].

There are different studies that have examined the possibility ofroducing PVDF flat sheet and hollow fibre membranes for applica-ion in MD [39–55]. Table 1 shows the VMD transmembrane fluxeseported in literature, with both PVDF flat sheet and hollow fibreembranes.In the recent papers [52–55], different approaches to improve

VDF membranes morphology and performances were proposed.Teoh and Chung [52] prepared hydrophobic polyvinyli-

ene fluoride–polytetrafluoroethylene (PVDF–PTFE) hollow fibres

howing improved hydrophobicity and macrovoid-free structure.

Wang et al. [53] produced mixed matrix PVDF hollow fibreembranes by adding hydrophobic cloisite clay particles to the

olymeric dope. These fibres have high porosity but a layer of

able 2VDF dopes composition.

Membrane dope 1 2

PolymerPVDF Solef 6020 (%) 20 20

SolventDMF (%) 68 65

AdditivesPVP (K-17) (%) 6 9H2O (%) 6 6

Air gap (cm) 25Room temperature (◦C) 25

nanoscale pores thus ensuring high fluxes, good thermal insula-tion and reduced risk of membrane pore wetting because of thehigh LEPw (water liquid entry pressure).

Bonyadi and Chung [54] obtained highly porous and macrovoid-free PVDF hollow fibre membranes co-extruding the polymericdope and the solvent by means of a triple spinneret. The solventflow at the fibres outer surface prevented the formation of a denseskin, increased the outer surface porosity of the PVDF fibres andeliminated the formation of macrovoids.

Finally, Hou et al. [55] proposed the synergic use of two poreforming agents (LiCl and PEG 1500) which led to fibres with highporosity and good hydrophobicity.

The aim of this research work was the production of micro-porous hydrophobic PVDF hollow fibres suitable for VMD, byusing water and poly(vinylpyrrolidone) as pore forming addi-tives.

Different works already reported about the use of water as poreforming additive for preparing both flat sheet and hollow fibremembranes for MD [44], sometimes in combination with otheradditives, such as LiCl [46].

PVP is widely used in preparing membranes and fibres toadjust pore size and pore size distribution, increase membrane

ing. It is mainly combined in blends with polysulfone (PSU),polyethersulfone (PES) and polyvinylidene fluoride (PVDF) to pre-pare membranes for ultra- (UF) and microfiltration (MF) to be usedin: biomedical applications (e.g. dialysis), water purification, waste

3 4 5 6

20 20 20 20

63 59 74 65

11 15 0 156 6 6 0

S. Simone et al. / Journal of Membrane Science 364 (2010) 219–232 221

heme

wal

bh

dp

wsa

a

2

2

owpTftcAtm

ir1ed

Fig. 1. Simplified sc

ater treatment and food processing (e.g. beer and wine). It is char-cterized by high biocompatibility, especially with blood, and veryow toxicity.

On the contrary, the use of PVP for preparing membranes toe applied in MD was never reported in the literature, due to itsydrophilic properties.

In this work, the effect of the PVP percentage in the polymericope and of the bore fluid composition on fibres morphology anderformance was examined.

The produced fibres were tested in a VMD plant using distilledater as feed. Some selected fibres were assessed for long term

tability; to this purpose, VMD tests on pure water were performedt different time intervals for a period of 2 months.

Preliminary VMD runs using synthetic seawater as feed werelso performed.

. Materials and methods

.1. Fibre preparation

For the preparation of the polymeric dopes, poly(vinylidene flu-ride) (PVDF Solef® 6012, supplied by Solvay chemical company)as dissolved in N,N-dimethylformamide (DMF), adding water andoly(N-vinylpyrrolidone) (K-17) (Mw 7–11 kDa) as pore formers.he water amount was of 6% (w/w), while PVP was increasedrom 0 to 15% (w/w). In one experiment only PVP was added tohe polymeric dope as pore former. The dope was kept underonstant stirring at 85 ◦C until the mixture was homogeneous.fterwards, the stirring was stopped allowing the release, at a con-

rolled temperature, of the gas bubbles incorporated during theixing.The dope was, then, loaded in a stainless steel reservoir pressur-

zed by pure nitrogen. The nitrogen pressure on the polymeric dopeeservoir was adjusted to have a constant dope flow rate of around2 g/min. Hollow fibres were spun using a spinneret an inner diam-ter (i.d.) of 600 �m and an outer diameter (o.d.) of 1600 �m. Moreetails about the spinning set-up were reported elsewhere [56].

of the VMD set-up.

The polymeric dopes compositions used in hollow fibres prepara-tion are listed in Table 2; spinning conditions for all experimentsare listed in Table 3.

2.2. Fibres post-treatment

The fibres were collected and washed with plenty of water toensure complete coagulation and removal of the residual solvent.

They were subsequently treated with a solution of sodiumhypochlorite 4000 ppm buffered to pH 7, in order to remove PVP,as suggested in the literature [57–62].

Fibres were then washed thoroughly and placed in a solutionof glycerol to 30% for 3–4 h. Finally, they were left to dry at roomtemperature.

2.3. SEM observations

The membrane’s morphology was observed by using a scanningelectron microscope (Quanta FENG 200, FEI Company). Fibres cross-sections were prepared by freeze fracturing the samples in liquidnitrogen. The internal and external surfaces were also observed.

2.4. Mechanical properties

The tensile strength of the fibres was measured by means ofa ZWICK/ROELL Z 2.5 test unit. Each sample was stretched uni-directionally at a constant rate of 5 mm/min; the initial distancebetween the clamps was of 50 mm. Five specimens were testedfor each sample. The breaking elongation and elastic or Young’smodulus were determined.

2.5. Bubble point and pore size distribution

Fibres bubble point and pore size distribution were determinedby using a PMI Capillary Flow porometer (Porous Materials Inc.,USA), following the procedure described in literature [49]. Mea-surements were carried out on fibres not treated with glycerol,

222S.Sim

oneet

al./JournalofMem

braneScience

364 (2010) 219–232Table 4Properties of the produced PVDF hollow fibres.

Hollow fibre Bore fluidcomposition

Bore fluid flowrate (ml/min)

Outer diameter(mm)

Inner diameter(mm)

Membranethickness (mm)

Emod(N/mm2)

�break (%) Bubblepoint(bar)

Largest detected porediameter (�m)

Average poresize (�m)

Porosity (%)

(a) Dope 1 (PVP 6%, H2O 6%)1a EtOH 30% 10 1.502 ± 0.053 1.124 ± 0.036 0.189 ± 0.089 104.64 148.59 2.11 0.29 0.14 76.61 ± 0.561b EtOH 30% 14 1.711 ± 0.035 1.335 ± 0.01 0.188 ± 0.045 137.42 171.31 2.47 0.25 0.16 75.03 ± 0.411c DMF 15% 14 1.712 ± 0.030 1.371 ± 0.031 0.170 ± 0.061 149.70 126.13 2.55 0.24 0.15 77.86 ± 2.021d DMF 15% 10 1.76 ± 0.045 1.37 ± 0.02 0.194 ± 0.065 155.31 120.05 3.67 0.17 0.16 78.03 ± 1.571e DMF 25% 10 1.67 ± 0.033 1.34 ± 0.03 0.166 ± 0.063 162.22 119.90 2.34 0.27 0.17 75.83 ± 2.451f DMF 25% 14 1.568 ± 0.037 1.187 ± 0.026 0.190 ± 0.063 157.02 164.54 2.79 0.22 0.14 76.50 ± 1.041g DMF 35% 14 1.714 ± 0.023 1.346 ± 0.017 0.184 ± 0.04 172.34 175.66 2.24 0.28 0.16 74.5 ± 1.311h DMF 35% 10 1.69 ± 0.05 1.38 ± 0.03 0.157 ± 0.08 170.48 145.47 2.76 0.23 0.16 77.72 ± 1.78

(b) Dope 2 (PVP 9%, H2O 6%)2a EtOH 30% 14 1.678 ± 0.056 1.338 ± 0.013 0.170 ± 0.068 88.90 142.48 1.29 0.48 0.15 80.87 ± 0.732b EtOH 30% 17 1.73 ± 0.02 1.33 ± 0.02 0.198 ± 0.04 97.78 155.84 1.42 0.44 0.14 78.64 ± 1.52c DMF 25% 17 1.76 ± 0.015 1.35 ± 0.011 0.203 ± 0.026 92.78 142.65 1.26 0.49 0.16 80.95 ± 0.362d DMF 25% 14 1.899 ± 0.029 1.519 ± 0.041 0.190 ± 0.07 73.54 134.18 1.25 0.50 0.15 79.09 ± 1.782e DMF 35% 14 1.775 ± 0.041 1.435 ± 0.039 0.170 ± 0.08 103.32 129.30 1.57 0.40 0.16 80.62 ± 1.272f DMF 35% 17 1.83 ± 0.039 1.5 ± 0.013 0.163 ± 0.052 120.75 174.47 2.39 0.26 0.16 76.84 ± 1.85

(c) Dope 3 (PVP 11%, H2O 6%)3a EtOH 30% 14 1.64 ± 0.047 1.238 ± 0.027 0.201 ± 0.074 103.68 147.75 2.12 0.29 0.17 82.54 ± 0.543b EtOH 30% 17 1.694 ± 0.012 1.347 ± 0.078 0.174 ± 0.09 119.68 186.67 2.17 0.29 0.15 77.04 ± 0.353c DMF 25% 17 1.73 ± 0.01 1.387 ± 0.033 0.175 ± 0.043 100.20 206.75 2.12 0.29 0.14 76.86 ± 0.823d DMF 25% 14 1.721 ± 0.041 1.341 ± 0.029 0.190 ± 0.07 100.64 197.76 2.59 0.24 0.15 82.07 ± 0.123e DMF 35% 14 1.661 ± 0.026 1.24 ± 0.038 0.211 ± 0.064 118.28 150.82 2.60 0.24 0.18 81.5 ± 0.083f DMF 35% 17 1.67 ± 0.02 1.25 ± 0.043 0.208 ± 0.063 127.01 180.42 1.99 0.31 0.16 81.5 ± 0.14

(d) Dope 4 (PVP 15%, H2O 6%)4a EtOH 30% 10 1.602 ± 0.058 1.230 ± 0.014 0.186 ± 0.072 125.82 230.11 2.56 0.24 0.21 83.27 ± 2.154b EtOH 30% 14 1.694 ± 0.053 1.347 ± 0.033 0.174 ± 0.086 140.75 220.68 2.60 0.24 0.22 81.23 ± 3.134c DMF 25% 7 1.597 ± 0.033 1.098 ± 0.034 0.249 ± 0.067 128.38 167.51 3.12 0.20 0.15 80.56 ± 1.674d DMF 25% 14 1.69 ± 0.046 1.287 ± 0.042 0.202 ± 0.088 159.87 196.21 2.47 0.25 0.16 78.09 ± 0.764e DMF 25% 20 1.759 ± 0.045 1.497 ± 0.026 0.131 ± 0.071 168.80 194.09 2.55 0.24 0.20 78.03 ± 1.644f DMF 35% 7 1.798 ± 0.044 1.443 ± 0.033 0.178 ± 0.077 125.77 183.93 2.56 0.24 0.15 78.85 ± 0.474g DMF 35% 14 1.767 ± 0.051 1.490 ± 0.028 0.139 ± 0.079 148.98 216.99 4.10 0.15 0.13 78.45 ± 0.864h DMF 35% 20 1.767 ± 0.065 1.490 ± 0.023 0.139 ± 0.088 199.23 192.82 4.19 0.15 0.13 77.54 ± 1.12

(e) Dope 5 (PVP 0%, H2O 6%)5a EtOH 30% 10 1.47 ± 0.019 1.15 ± 0.01 0.161 ± 0.029 116.72 130.41 3.15 0.20 0.15 73.62 ± 1.675b EtOH 30% 14 1.6 ± 0.03 1.26 ± 0.038 0.17 ± 0.068 87.59 184.99 2.98 0.21 0.14 71.21 ± 0.565c DMF 25% 14 1.61 ± 0.034 1.28 ± 0.024 0.165 ± 0.058 120.98 233.78 3.45 0.18 0.12 70.55 ± 1.045d DMF 25% 10 1.544 ± 0.012 1.24 ± 0.012 0.152 ± 0.024 129.48 175.76 3.05 0.20 0.16 74.22 ± 1.545e DMF 35% 10 1.642 ± 0.048 1.24 ± 0.013 0.202 ± 0.061 119.45 112.07 2.75 0.23 0.15 75.17 ± 0.785f DMF 35% 14 1.69 ± 0.055 1.285 ± 0.032 0.203 ± 0.087 80.48 152.90 2.55 0.24 0.15 71.2 ± 2.045g EtOH 45% 10 1.557 ± 0.05 1.4 ± 0.035 0.159 ± 0.085 86.58 96.92 3.15 0.20 0.16 73.82 ± 1.57

(f) Dope 6 (PVP 15%, H2O 0%)6a EtOH 30% 10 1.531 ± 0.011 1.401 ± 0.013 0.065 ± 0.024 84.02 382.70 1.15 0.54 0.23 82.79 ± 2.386b DMF 25% 10 1.467 ± 0.035 1.350 ± 0.013 0.058 ± 0.048 86.55 412.25 1.28 0.49 0.23 83.59 ± 1.086c DMF 25% 14 1.577 ± 0.021 1.522 ± 0.012 0.027 ± 0.033 83.54 256.65 2.15 0.29 0.25 82.31 ± 1.66d DMF 25% 20 1.674 ± 0.03 1.528 ± 0.019 0.073 ± 0.049 93.52 411.70 1.65 0.38 0.26 77.76 ± 1.66e DMF 35% 20 1.554 ± 0.032 1.408 ± 0.014 0.073 ± 0.046 95.50 391.79 2.05 0.30 0.27 77.43 ± 1.366f DMF 35% 14 1.509 ± 0.012 1.329 ± 0.012 0.090 ± 0.024 87.70 282.20 1.75 0.35 0.26 77.56 ± 1.976g DMF 35% 10 1.454 ± 0.035 1.293 ± 0.017 0.081 ± 0.052 89.43 412.34 1.85 0.34 0.25 80.81 ± 2.48

S. Simone et al. / Journal of Membrane Science 364 (2010) 219–232 223

F DF hov 35% (

uupU

2

[lpt

ε

ig. 2. Effect of the bore fluid on macrovoid formation in the microstructure of PV/v). Bore fluids: (a) EtOH 30% (v/v); (b) DMF 15% (v/v); (c) DMF 25% (v/v); (d) DMF

sing isopropanol (surface tension 21.7 dynes/cm) as wetting liq-id. The bubble point and pore size distribution were determinedrocessing data by the software Caprep (Porous Materials Inc.,SA).

.6. Fibres porosity

Membrane porosity was determined by gravimetric method43], measuring the weight of liquid (here kerosene, as suggested initerature, Refs. [51,53,54]) contained in the membrane pores. The

orosity of the hollow fibre membranes (ε) can be calculated usinghe following formula [43]:

={

(w1 − w2)/Dk

(w1 − w2)/Dk + w2/Dp

}× 100%

llow fibres spun from a polymeric dope containing 6% of PVP (K-17) and H2O (6%,v/v). Bore fluid injection rate: 14 ml/min.

where w1 is the weight of the wet membrane; w2 is the weight ofthe dry membrane; Dk is the kerosene density (0.82 g/cm3) and Dp

is the polymer density (1.78 g/cm3, as reported in Solvay technicalsheets).

2.7. CAM measurements

The hydrophobicity of the produced fibres was assessed by mea-surements of the contact angle to water. In particular, the dynamiccontact angle of the PVDF hollow fibre membranes was measured

by using a tensiometer (DCAT11 Tensiometer, Dataphysics, Ger-many) as suggested in the literature [49,53]. A fibre glued to theholder was hung from the arm of an electro-balance, and thenput, by cyclic immersions, into DI water. The contact angle wascalculated from the wetting force based on the Wihelmy method.

224 S. Simone et al. / Journal of Membrane Science 364 (2010) 219–232

F 2O (6fl

2

2

fiomoo

2

w

wb3

f

J

ws

i

ig. 3. Effect of the PVP percentage on the microstructure of PVDF hollow fibres. Huid: DMF 35% (v/v) and injection rate: 14 ml/min.

.8. VMD tests

.8.1. Membrane modules preparationMembrane modules were prepared by inserting three hollow

bres in a glass device, sealing both extremities with epoxy glue. Inrder to remove glycerol from fibres, before each test, membraneodules were washed with double distilled water at 40 ◦C and 1 bar

f pressure for 3 h. Modules were completely dried by sending airvernight before being connected to the VMD plant.

.8.2. VMD plantThe VMD set-up used in this study was already described else-

here [20]. Fig. 1 shows a simplified scheme.The feed was recirculated inside the hollow fibres, while vacuum

as applied at the shell side. VMD tests were performed using dou-le distilled water as feed at: feed temperature 50 ◦C; feed flow rate2 l/h; vacuum pressure 20 mbar.

The distillate flux, J (kg/m2 h), was calculated according to theollowing equation:

= Md

A.t

here Md is the mass of the permeate; A is the active membraneurface; t is the time interval.

A was calculated taking into account the number of membranesnserted in the module, their length and internal diameter.

%, v/v); PVP (K-17) percentage: (a) 9% (w/w); (b) 11% (w/w); (c) 15% (w/w). Bore

VMD experiments on pure water were performed over a periodof almost 2 months in order to test the stability of produced hollowfibres.

VMD tests were also performed feeding a salty solution,prepared in lab and simulating seawater, having the following com-position: NaCl 23.27 g/l, Na2SO4 3.99 g/l, MgCl2·6H2O 11.28 g/l, KBr0.095 g/l, CaCl2·2H2O 1.47 g/l, NaHCO3 0.1932 g/l, KCl 0.6635 g/l,Na2CO3·H2O 0.0072 g/l.

3. Results and discussion

In this work, an extensive study of how the addition of PVP tothe polymeric dope affects fibres properties and, then, their per-formance in VMD, was carried out. The effect of the bore fluidcomposition was also considered. The main features of the pro-duced PVDF hollow fibres are reported in Table 4a–f.

3.1. Morphology examination by SEM

Fibres morphology was investigated using a scanning electronmicroscope (SEM). Looking at the images of fibres cross-sections,

presented in Figs. 2–4, it can be seen that fibres show a sponge-likestructure, with some small macrovoids on the surfaces.

The obtained structures are related to the mechanisms of thecoagulation process. As reported in the literature, during the hollowfibre coagulation, the bore fluid causes the polymer precipita-

S. Simone et al. / Journal of Membrane Science 364 (2010) 219–232 225

Fig. 4. Effect of the bore fluid composition and PVP amount on the microstructure of PVDF hollow fibres. All pictures were taken at 1000× magnification grade. H2O (6%,v/v); PVP (K-17) percentage: (a) 0% (w/w); (b) 6% (w/w); (c) 9% (w/w); (d) 11% (w/w); (e) 15% (w/w). Bore fluid injection rate: 14 ml/min.

226 S. Simone et al. / Journal of Membrane Science 364 (2010) 219–232

F ); PVPfl

ttelofiec

acc

ig. 5. Fibres cross-sections observed a high magnification (10,000×). H2O (6%, v/vuid injection rate: 14 ml/min.

ion and formation of the inner surface; while, the formation ofhe outer surface starts during fibre falling along the air gap andnds in the coagulation bath. The composition of both coagu-ation media (internal and external), the air gap, the presencef hydrophilic additives and the temperature are all affectingbre’s final structure, since they influence the solvent/non-solventxchange velocity, that is the main variable during the hollow fibre

oagulation.

In our case, the morphology of the produced PVDF hollow fibresre discussed taking into account the effect of the different con-entrations of PVP in the polymeric dope and of the bore fluidomposition.

(K-17) percentage: (a) 6% (w/w); (b) 9% (w/w); (c) 11% (w/w); (d) 15% (w/w). Bore

Regarding the fibres outer layer, when the nascent fibre reachesthe water bath, it undergoes to solidification in a short time-period.This causes a limited growth of the polymer-lean phase, resulting insmall macrovoids near the outer edge. The presence of finger-likestructures at the outer surface is modulated by the PVP percentage.

Cavities of bigger size can be observed, at the outer surface byincreasing the PVP percentage from 6 to 9% (Figs. 2d and 3a). Further

increase of PVP percentage to 11 and 15% leads to a reduction oreven to a elimination of fingers formation (Fig. 3b and c). Our find-ings are in agreement with literature. It is reported that the additionof PVP, which is frequently used as pore former additive [63–65],could induce the formation of finger-like structures and macrovoids

S. Simone et al. / Journal of Membrane Science 364 (2010) 219–232 227

Ffi(

botrfapcpcTt

it

obsbSE3

cc

fb

fia2

s

wuata

ifipp

The analysis of mechanical properties led to interesting resultsthat are summarized in Table 4a–f. In particular, the values ofYoung’s modulus reached peaks of 160–170 N/mm2.

ig. 6. Effect of the PVP percentage on the mechanical properties of PVDF hollowbres. Results obtained for three different bore fluids with the same injection rate14 ml/min).

oth at the outer and the inner surfaces [66]. However, PVP effectn membrane morphology is strongly influenced by its concentra-ion. At low concentration, the size of cavities is larger, but it iseported to reduce when the concentration of PVP is increased. Inact, on one side, PVP increases the dope thermodynamic instability,ccelerating the solvent and non-solvent diffusion rates during thehase inversion process. Big cavities and macrovoids develop as aonsequence of the rapid demixing. On the other hand, PVP is com-atible with PVDF and increases the dope viscosity. At higher PVPoncentration, the solvent/non-solvent exchange can be hindered.his counterbalances the thermodynamic effect of PVP, avoidinghe formation of macrovoids, as also reported in literature [67,68].

Comparing fibres internal surfaces reported in Figs. 2d and 3a–ct can be also noticed how the increase of PVP concentration leadso a higher surface porosity of the fibres internal layer.

In order to study the effect of the bore fluid, aqueous solutionsf ethanol or DMF at different concentrations (from 15 to 45%) haveeen tested. As it results from the SEM pictures presented in Fig. 2,ome macrovoids at the inner surface are formed when using theore fluid having the highest water percentage (DMF 15%, Fig. 2b).ome macrovoids can be still detected when using DMF at 25% ortOH at 30%, but they are completely eliminated when using DMF5% as bore fluid. Internal surfaces look porous in all cases.

For a better understanding of the effect of the bore fluidomposition on hollow fibres morphology, SEM pictures of theross-sections are compared in Fig. 4.

For PVP 0% in the dope, there are macrovoids at both fibres sur-aces. They reduce, but not disappear, only when using DMF 35% asore fluid.

For PVP 6-9-11% there are finger-like macrovoids mainly at thebres outer surfaces. Some finger-like structures can be detectedt the fibres inner layer when using DMF 15% (see Fig. 2b) or DMF5% as bore fluid.

For PVP 15% there are only small macrovoids at the fibres outerurface, for all bore fluids.

These results can be discussed referring to literature, whereater and mixtures of water and various alcohols or solvents aresed as bore fluids [36,69,35]. In general, it is known that large voidsre formed when using pure water as the internal coagulant, whilehe void formation can be eliminated by using aqueous mixtures oflcohols or solvents.

Furthermore, Sukitpaneenit and Chung [70] recently reported

n details the effect of different non-solvents on PVDF hollowbres morphology and properties. Since PVDF is a semi-crystallineolymer, both liquid/liquid and solid/liquid demixing can takelace during membrane coagulation, depending on the non-solvent

Fig. 7. Effect of the PVP percentage on the porosity of PVDF hollow fibres. Resultsobtained for three different bore fluids with the same injection rate (14 ml/min).

strength. Water is a strong coagulant; hence, precipitation is fast,by liquid/liquid demixing, resulting in finger-like macrovoids anda small portion of interconnected cellular type structure. Whenusing “softer” coagulants, as methanol, ethanol or isopropanol, itcan be clearly noticed that macrovoids disappear while membranestructure shifts to globule type. The observed spherical globulesare made of semi-crystalline PVDF. They grow in a process ofsolid/liquid demixing, that can take place only when coagulationis delayed, leaving enough time to induce crystallization.

Looking at fibres cross-sections at very high magnification(Fig. 5) it can be seen that, for all bore fluids, fibres structure isof cellular type. This means that the water percentage in the borefluid is enough high to ensure liquid/liquid demixing in all casesand that, even if coagulation is delayed, there is insufficient time toinduce crystallization.

It can be concluded that, in our case, the bore fluid compositionaffects mainly macrovoid formation at the fibres inner layer; how-ever, as observed in the SEM pictures, at high PVP concentration,this effect is less evident.

The influence of both PVP and bore fluid on the mechanical prop-erties of the produced fibres is discussed in more details in nextsection.

3.2. Mechanical tests

Fig. 8. VMD flux vs. εrp/ı. Results obtained for three different bore fluids at the sameinjection rate (14 ml/min). For each dataset the H2O percentage in the dope is 6%.

228 S. Simone et al. / Journal of Membrane Science 364 (2010) 219–232

Table 5Selection of results of VMD tests performed on some fibres in order to analyse the effect of porosity and thickness. Feed double distilled water;Tfeed 50 ◦C; vacuum pressure 20 mbar.

Fibre PVP (%) Bore fluid Flux (VMD)kg/(m2 h)

1b 6 ETOH 30%, 14 ml/min 8.501f DMF 25%, 14 ml/min 10.31g DMF 35%, 14 ml/min 9.80

2a 9 ETOH 30%, 14 ml/min 17.02b ETOH 30%, 17 ml/min 16.72c DMF 25%, 17 ml/min 18.52d DMF 25%, 14 ml/min 12.82e DMF35%, 14 ml/min 17.92f DMF35%, 17 ml/min 13.2

3a 11 ETOH 30%, 14 ml/min 16.73b ETOH 30%, 17 ml/min 15.13c DMF 25%, 17 ml/min 15.23d DMF 25%, 14 ml/min 14.13e DMF 35%, 14 ml/min 14.03f DMF 35%, 17 ml/min 15.3

4b 15 ETOH 30%, 14 ml/min Water leakage4d DMF 25%, 14 ml/min 12.34g DMF 35%, 14 ml/min 15.3

5b 0 ETOH 30%, 14 ml/min 3.55c 0 DMF 25%, 14 ml/min 5.7

DMF

ETOHDMF

nsrrt

tes

up

titvp81lDm

ticii

ao

aClss

3.4. Bubble point and pore size distribution

5f 0

6a 156c 15

The mechanical properties of a membrane are strongly con-ected to its structure. Membranes having a good sponge-liketructure should be stronger; while, the presence of macrovoidseduces membranes mechanical strength [71]. It was alreadyeported in the literature that the addition of PVP could improvehe fibres mechanical strength [56].

The formation of macrovoids is linked to the PVP percentage inhe dope and to the bore fluid composition. In particular, as alreadyxplained in Section 3.1, the percentage of PVP strongly affects theize of cavities developed at the fibres outer surface.

The plots presented in Fig. 6 compare the Young’s Modulus val-es of fibres spun from polymeric dopes containing increasing PVPercentages, but coagulated with the same bore fluid.

It can be noticed that, when PVP percentage increases from 0o 6%, there is an evident increase of the Young’s Modulus. Thiss connected to the increase of the total percentage of solids inhe polymeric dope, which, in turn, affects the polymeric dopeiscosity and, hence, fibres mechanical strength. Then, when PVPercentage reaches 9%, the Young’s Modulus drops from 137.42 to8.90 N/mm2, in the case of fibres produced using EtOH 30%, from57.02 to 73.54 N/mm2 when DMF 25% is used as internal coagu-

ant, and from 172.34 to 103.32 N/mm2 for fibres coagulated withMF 35%. Further increase of the PVP concentration improves fibresechanical properties.This can be explained considering the PVP effect on fibres struc-

ure. While, for small amounts, the increase in PVP concentrationnduces the formation of large cavities and macrovoids, higher con-entrations promote the formation of sponge-like structures, thusmproving also fibres mechanical properties. This is also completelyn agreement to what observed with SEM pictures.

The elongation-at-break percentages show a similar behaviour,s also reported in Fig. 6. In this case, very good results werebtained for the highest PVP concentration: εbreak reaches 230.11%.

Regarding the effect of the bore fluid, it is very difficult to findcommon trend among all the produced fibres. Sukitpaneenit and

hung [70] evidenced that a delayed demixing during fibres coagu-

ation can suppress macrovoid formation but, if the microstructurehifts to globular, polymer crystallites are loosely packed, with sub-equent reduction of the Young’s modulus.

35%, 14 ml/min 4.3

30%, 10 ml/min 1025%, 14 ml/min 10.9

As already discussed in Section 3.1, in our case, water percentagein the bore fluid is so high that fibres structure is always of cellulartype, and, therefore, no PVDF crystallization-solid/liquid demixingtakes place during coagulation.

In our case, fibres mechanical properties are strongly connectedto the size of macrovoids, which are only moderately affected by thebore fluid composition. Therefore, it can be concluded that, in ourcase, the effect of PVP percentage in the dope on fibres propertiesprevails with respect to that of the bore fluid composition. The samehappens for the membrane porosity.

3.3. Porosity

Experimental results show that fibres porosity ranges from 71to 83%, as it is shown in Table 4a–f. As already discussed, porosityis a crucial parameter if fibres are designed to be used in MD.

In Fig. 7 the dependence of the fibres porosity on the PVP con-centration is illustrated, taking into account three different borefluids (EtOH 30%, DMF 25%, and DMF 35%).

PVP is a hydrophilic additive that is combined with the poly-mer to promote the formation of porous structures. Therefore, ingeneral, membrane porosity should increase by increasing the per-centage of PVP added to the polymeric dope. In fact, fibres porosityclearly increases for PVP percentage moving from 0 to 9% in allcases. A further increase of the PVP percentage brings about to aslight reduction of the porosity. This effect can be explained takeninto account that fibres porosity is measured on the basis of the voidfraction. Membranes having finger-like structures and macrovoidscould have higher void fractions with respect to spongiform mem-branes. As observed in section 3.2, the increase of PVP percentagefrom 9 to 15% reduces the sizes of cavities at the outer surface, thusreducing the membrane void fraction.

From Table 4a–f, it can be found that the maximum diameterof the membrane pores ranges from 0.15 �m (Fibre 4h) to 0.54 �m(Fibre 6a), as obtained from bubble point data. The membrane aver-

S. Simone et al. / Journal of Membrane Science 364 (2010) 219–232 229

Fig. 9. VMD flux dependence on time (days) and module storage. Fibre 4d.

Fig. 10. VMD fluxes measured during tests performed on Fibre 1f feeding distilled water and synthetic seawater.

Fig. 11. VMD fluxes measured during tests performed on Fibre 4d feeding distilled water and synthetic seawater.

2 mbran

a6

nmd

3

wt(4d

3

twitettaImomo

Bdd

ndlpahoa

atas

3

btOw

s4oHswp

30 S. Simone et al. / Journal of Me

ge pore size falls in the range 0.13 �m (Fibre 4h)–0.27 �m (Fibree).

In general, for each fibre, the pore size distribution is veryarrow. The most dispersed one is that of Fibre 2d (Diameter ataximum pore size distribution: 0.15 �m, largest detected pore

iameter: 0.50 �m.).

.5. CAM measurement

All PVDF fibres batch were screened by using double distilledater, in order to evaluate their hydrophobic character and, hence,

heir applicability in MD. Contact angle values ranging from 82 ± 1◦

for fibres spun form dope 1) to 91 ± 1◦ (for fibres spun from dope) were obtained. Therefore, the contact angle to water does notecrease when increasing the PVP percentage in the dope.

.6. VMD tests on distilled water

The PVDF hollow fibres produced were tested in a VMD plant. Athe beginning, experiments were carried out using double distilledater as the feed. This allowed to assess membrane hydrophobic-

ty and performance. Comparing the results listed in Table 5 withhose reported in Table 1, for VMD, it can be seen that, in gen-ral, transmembrane flux values comparable to or even better thanhose reported in literature were obtained. In fact, VMD flux fluc-uates around 15 kg/h m2. The obtained results confirm that PVPnd water play a synergic action in enhancing fibres performances.n fact, the fluxes obtained when working with fibres from experi-

ent 5 (H2O 6%, PVP 0% in the dope) are the lowest observed. On thether hand, fluxes obtained when working with fibres from experi-ent 6 (H2O 0%, PVP 15% in the dope) are generally lower than those

btained when using fibres containing both pore forming additives.Only in one case (Fibre 4b), a liquid water leakage was observed.

y repeating VMD tests on Fibre 4b several times, some waterroplets were noticed at the shell side. This is probably due to someefects in the fibre rather than hydrophobicity loss.

VMD fluxes are strongly affected by fibres porosity and thick-ess. In general, membranes should ensure good moleculariffusion (high porosity, reduced thickness) minimizing the heat

oss (high void fraction, higher thickness). From Table 5, the worsterformances were observed for fibres having reduced thicknessnd low porosity (Fibres 5b, c and f, VMD flux 3.5, 5.7 and 4.3 kg/m2

respectively). On the other hand, the best performances werebtained working with fibres having high porosity and thicknessround 0.17 mm (Fibre 2e, VMD flux 17.9 kg/m2 h).

Fig. 8 reports the flux vs. the ratio: εrp/ı, which takes intoccount the membrane porosity, thickness and pore size, all proper-ies that are the result of an interplay between the PVP percentagend the bore fluid type. The flux increases with this ratio and aomehow linear trend is registered.

.7. Long term performances

In order to evaluate the hollow fibre stability with time, a mem-rane module (Fibre 4d) was tested at different time intervals upo 2 months. VMD flux dependence on time is reported in Fig. 9.n the x-axis the time (days) and the module storage way (dried orith water inside the fibres) are indicated.

Looking at the plot, it can be noticed that the performance istable over 15 days. At day 23 the flux seems to drop, but aftermeasurements, the initial performance is restored. At the end

f the test, a slight reduction (around 7%) of the flux is observed.owever, it can be concluded membrane performances are quite

table over time. Moreover, no membrane damage and/or leakageere detected. These results are quite interesting, but, it has to beointed out that they refer to tests carried out with pure water as

e Science 364 (2010) 219–232

feed. As further analysis, the long term performance of the preparedmembranes with scaling water as feed should be also evaluated.

3.8. Tests on synthetic seawater

VMD experiments on simulated seawater were carried out usingtwo different types of PVDF hollow fibres (Fibre 1f, PVP 6%, borefluid DMF 25%, injection rate 14 ml/min; Fibre 4d, PVP 15%, borefluid 25%, injection rate 14 ml/min). The experimental tests wereperformed in sequence: a first test with distilled water, then dif-ferent tests on seawater (without any distilled water flushing inbetween), then, another test with distilled water and so on.

As it can be seen from Figs. 10 and 11, VMD flux measured duringtests performed on salty water is only slightly lower than that mea-sured when feeding double distilled water, but it is constant withtime. Moreover, pure water fluxes measured after experimentson simulated seawater are similar to that measured at the begin-ning of the experiment. For all tests, the salt rejection was around99.98%.

4. Conclusions

In this work, microporous hydrophobic PVDF hollow fibres wereproduced by the phase inversion method, by using H2O and PVP aspore forming additives. This is the first time that PVP is used as poreforming additive for preparing membranes designed for MD.

The PVP effect on fibres structure can be resumed as follows: forsmall amounts, the increase in PVP concentration induces the for-mation of large cavities and macrovoids, increasing the membranevoid fraction and reducing its strength; higher concentrations pro-mote the formation of sponge-like structures, thus improving alsofibres mechanical properties, but slightly reducing the porosity.Referring the effect of the bore fluid, it was found an influence onmacrovoids formation at the fibres inner surface. However, its effectis negligible with respect to that of PVP.

The water flux values obtained in VMD tests made on dis-tilled water as feed, were comparable to or even better than thosereported in the literature.

The produced fibres were assessed for long time stability in pro-longed VMD runs, showing good performance up to 2 months. Somepreliminary VMD tests on simulated seawater were also performed,in order to test if fibres could be applied in desalination. A slightreduction of the transmembrane flux was observed, but the fluxwas constant with time. On the other hand, when working withpure water, the initial performances were restored.

Acknowledgements

The authors acknowledge the financial support of the EuropeanCommission within the 6th Framework Program for the grant tothe Membrane-Based Desalination: An Integrated Approach project(acronym MEDINA). Project no.: 036997.

The authors are grateful to Dr. M. Davoli (Department of EarthScience, University of Calabria, Via P. Bucci, 87030 Rende (CS), Italy)for the use of SEM microscope.

The authors are grateful to Prof. R. Wang (SMTC, Singapore) forthe dynamic contact angle measurements.

References

[1] K.W. Lawson, D.R. Lloyd, Membrane distillation, Journal of Membrane Science

124 (1997) 1–25.

[2] E. Curcio, E. Drioli, Membrane distillation and related operations—a review,Separation & Purification Reviews 34 (2005) 35–86.

[3] M.S. El-Bourawi, Z. Ding, R. Ma, M. Khayet, A framework for better understand-ing membrane distillation separation process, Journal of Membrane Science285 (2006) 4–29.

bran

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

S. Simone et al. / Journal of Mem

[4] E. Drioli, A. Criscuoli, E. Curcio, Membrane Contactors: Fundamentals, Applica-tions and Potentialities, Membrane Science and Technology Series 11, Elsevier,Amsterdam, Boston, 2006.

[5] C. Cabassud, D. Wirth, Membrane distillation for water desalination: how tochose an appropriate membrane, Desalination 157 (2003) 307–314.

[6] Y. Xu, B.K. Zhu, Y.Y. Xu, Pilot test of vacuum membrane distillation for seawaterdesalination on a ship, Desalination 189 (2006) 165–169.

[7] J. Koschikowski, M. Wieghaus, M. Rommel, Solar thermal-driven desalinationplants based on membrane distillation, Desalination 156 (2003) 295–304.

[8] F. Banat, S. Al-Asheh, M. Qtaishat, Treatment of waters colored with methyleneblue dye by vacuum membrane distillation, Desalination 174 (2005) 87–96.

[9] A. Criscuoli, J. Zhong, A. Figoli, M.C. Carnevale, R. Huang, E. Drioli, Treatmentof dye solutions by vacuum membrane distillation, Water Research 42 (2008)5031–5037.

10] F.A. Banat, J. Simandl, Removal of benzene traces from contaminated waterby vacuum membrane distillation, Chemical Engineering Science 51 (1996)1257–1265.

11] M.S. EL-Bourawi, M. Khayet, R. Ma, Z. Ding, Z. Li, X. Zhang, Application of vacuummembrane distillation for ammonia removal, Journal of Membrane Science 301(2007) 200–209.

12] T. Mohammadi, M. Akbarabadi, Separation of ethylene glycol solution by vac-uum membrane distillation (VMD), Desalination 181 (2005) 35–41.

13] H.J. Huang, S. Ramaswamy, U.W. Tschirner, B.V. Ramarao, A review of sep-aration technologies in current and future biorefineries, Separation andPurification Technology 62 (2008) 1–21.

14] S. Al-Asheha, F. Banat, M. Qtaishat, M. Al-Khateeb, Concentration of sucrosesolutions via vacuum membrane distillation, Desalination 195 (2006) 60–68.

15] T. Mohammadi, O. Bakhteyari, Concentration of l-lysine monohydrochloride(l-lysine–HCl) syrup using vacuum membrane distillation, Desalination 200(2006) 591–594.

16] M.A. Izquierdo-Gil, J. Abildskov, G. Jonsson, The use of VMD data/model totest different thermodynamic models for vapor–liquid equilibrium, Journal ofMembrane Science 239 (2004) 227–241.

17] V. Soni, J. Abildskov, G. Jonsson, R. Gani, Modeling and analysis of vacuum mem-brane distillation for the recovery of volatile aroma compounds from blackcurrant juice, Journal of Membrane Science 320 (2008) 442–455.

18] Z.P. Zhao, F.W. Ma, W.F. Liu, D.Z. Liu, Concentration of ginseng extracts aqueoussolution by vacuum membrane distillation. 1. Effects of operating conditions,Desalination 234 (2008) 152–157.

19] F. Banat, F.A. Al-Rub, K. Bani-Melhem, Desalination by vacuum membrane dis-tillation: sensitivity analysis, Separation and Purification Technology 33 (2003)75–87.

20] A. Criscuoli, M.C. Carnevale, E. Drioli, Evaluation of energy requirementsin membrane distillation, Chemical Engineering and Processing 47 (2008)1098–1105.

21] S. Al-Obaidani, E. Curcio, F. Macedonio, G. Di Profio, H. Al-Hinai, E. Drioli,Potential of membrane distillation in seawater desalination: thermal efficiency,sensitivity study and cost estimation, Journal of Membrane Science 323 (2008)85–98.

22] Y Wu, Y. Kong, X. Lin, W. Liu, J. Xu, Surface-modified hydrophilic membranesin membrane distillation, Journal of Membrane Science 72 (1992) 189–196.

23] Y. Kong, X. Lin, Y. Wu, J. Chen, J. Xu, Plasma polymerization of octafluorocy-clobutane and hydrophobic microporous composite membranes for membranedistillation, The Journal of Applied Polymer Science 46 (1992) 191–199.

24] M. Khayet, T. Matsuura, Application of surface modifying macromolecules forthe preparation of membranes for membrane distillation, Desalination 158(2003) 51–56.

25] M. Khayet, J.I. Mengual, T. Matsuura, Porous hydrophobic/hydrophilic com-posite membranes: application in desalination using direct contact membranedistillation, Journal of Membrane Science 252 (2005) 101–113.

26] S.R. Krajewski, W. Kujawski, M. Bukowska, C. Picard, A. Larbot, Application offluoroalkylsilanes (FAS) grafted ceramic membranes in membrane distillationprocess of NaCl solutions, Journal of Membrane Science 281 (2006) 253–259.

27] B. Li, K.K. Sirkar, Novel membrane and device for vacuum membranedistillation-based desalination process, Journal of Membrane Science 257(2005) 60–75.

28] J. Xua, M. Furuswa, A. Itoa, Air-sweep vacuum membrane distillation using finesilicone, rubber, hollow-fibre membranes, Desalination 191 (2006) 223–231.

29] Z Jin, D.L. Yang, S.H. Zhang, X.G. Jian, Removal of 2,4-dichlorophenol fromwasterwater by vacuum membrane distillation using hydrophobic PPESK hol-low fibre membrane, Chinese Chemical Letters 18 (2007) 1543–1547.

30] Z. Jin, D.L. Yang, S.H. Zhang, X.G. Jian, Hydrophobic modification ofpoly(phthalazinone ether sulfoneketone) hollow fibre membrane for vacuummembrane distillation, Chinese Chemical Letters 19 (2008) 367–370.

31] Z. Jin, D.L. Yang, S.H. Zhang, X.G. Jian, Hydrophobic modification ofpoly(phthalazinone ether sulfone ketone) hollow fibre membrane for vacuummembrane distillation, Journal of Membrane Science 310 (2008) 20–27.

32] Z.D. Hendren, J. Brant, M.R. Wiesnera, Surface modification of nanostructuredceramic membranes for direct contact membrane distillation, Journal of Mem-brane Science 331 (2009) 1–10.

33] M. Qtaishat, D. Rana, M. Khayet, T. Matsuura, Preparation and characteriza-tion of novel hydrophobic/hydrophilic polyetherimide composite membranesfor desalination by direct contact membrane distillation, Journal of MembraneScience 327 (2009) 264–273.

34] M. Qtaishat, M. Khayet, T. Matsuura, Novel porous composite hydropho-bic/hydrophilic polysulfone membranes for desalination by direct con-

[

e Science 364 (2010) 219–232 231

tact membrane distillation, Journal of Membrane Science 341 (2009)139–148.

35] S.P. Deshmukh, K. Li, Effect of ethanol composition in water coagulation bathon morphology of PVDF hollow fibre membranes, Journal of Membrane Science150 (1998) 75–85.

36] D. Wang, K. Li, W.K. Teo, Preparation and characterization of Polyvinylidenefluoride (PVDF) hollow fibre membranes, Journal of Membrane Science 163(1999) 211–220.

37] J. Kong, K. Li, Preparation of PVDF hollow-fibre membranes, Journal of AppliedPolymer Science 81 (2001) 1643–1653.

38] M. Khayet, C.Y. Feng, K.C. Khulbe, T. Matsuura, Preparation and characterizationof polyvinylidene fluoride hollow fibre membranes for ultrafiltration, Polymer43 (2002) 3879–3890.

39] Y. Fujii, S. Kigoshi, H. Iwatani, M. Aoyama, Selectivity and characteristics ofdirect contact membrane distillation type experiments. I. Permeability andselectivity through dried hydrophobic fine porous membranes, Journal of Mem-brane Science 72 (1992) 53–72.

40] Y. Fujii, S. Kigoshi, H. Iwatani, M. Aoyama, Y. Fusaoka, Selectivity and character-istics of direct contact membrane distillation type experiments. II. Membranetreatment and selectivity increase, Journal of Membrane Science 72 (1992)73–89.

41] J.M. Ortiz de Zárate, L. Pena, J.I. Mengual, Characterization of membrane dis-tillation membranes prepared by phase inversion, Desalination 100 (1995)139–148.

42] M. Tomaszewska, Preparation and properties of flat-sheet membranes frompolyvinylidene fluoride for membrane distillation, Desalination 104 (1996)1–11.

43] C.S. Feng, B. Shi, G. Li, Y. Wu, Preparation and properties of microp-orous membrane from poly(vinylidene fluoride-cotetrafluoroethylene) (F2.4)for membrane distillation, Journal of Membrane Science 237 (2004)15–24.

44] M. Khayet, T. Matsuura, Preparation and characterization of polyvinylidenefluoride membranes for membrane distillation, Industrial & Engineering Chem-istry Research 40 (2001) 5710–5718.

45] M. Khayet, K.C. Khulbe, T. Matsuura, Characterization of membranes for mem-brane distillation by atomic force microscopy and estimation of their watervapor transfer coefficients in vacuum membrane distillation process, Journalof Membrane Science 238 (2004) 199–211.

46] B. Wu, K. Li, W.K. Teo, Preparation and characterization of poly(vinylidene flu-oride) hollow fibre membranes for vacuum membrane distillation, Journal ofApplied Polymer Science 106 (2007) 1482–1495.

47] B. Wu, X. Tan, K. Li, W.K. Teo, Removal of 1,1,1-trichloroethane from waterusing a polyvinylidene fluoride hollow fibre membrane module: vacuum mem-brane distillation operation, Separation and Purification Technology 52 (2006)301–309.

48] C. Feng, K.C. Khulbe, T. Matsuura, R. Gopal, S. Kaur, S. Ramakrishna, M. Khayet,Production of drinking water from saline water by air-gap membrane distilla-tion using polyvinylidene fluoride nanofiber membrane, Journal of MembraneScience 311 (2008) 1–6.

49] K. Wang, T. Chung, M. Gryta, Hydrophobic PVDF hollow fibre membranes withnarrow pore size distribution and ultra-thin skin for the freshwater produc-tion through membrane distillation, Chemical Engineering Science 63 (2008)2587–2594.

50] C. Kuo, H. Lin, H. Tsai, D. Wang, J. Lai, Fabrication of a high hydrophobic PVDFmembrane via nonsolvent induced phase separation, Desalination 233 (2008)40–47.

51] S. Bonyadi, T.S. Chung, Flux enhancement in membrane distillation by fabrica-tion of dual layer hydrophilic–hydrophobic hollow fibre membranes, Journalof Membrane Science 306 (2007) 134–146.

52] M. Teoh, T. Chung, Membrane distillation with hydrophobic macrovoid-freePVDF–PTFE hollow fibre membranes, Separation and Purification Technology66 (2009) 229–236.

53] K.Y. Wang, S.W. Foo, T. Chung, Mixed matrix PVDF hollow fibre membraneswith nanoscale pores for desalination through direct contact membrane distil-lation, Industrial & Enginnering Chemistry Research 48 (2009) 4474–4483.

54] S. Bonyadi, T. Chung, Highly porous and macrovoid-free PVDF hollow fibremembranes for membrane distillation by a solvent-dope solution co-extrusionapproach, Journal of Membrane Science 331 (2009) 66–74.

55] D. Hou, J. Wang, D. Qu, Z. Luan, X. Ren, Fabrication and characterization ofhydrophobic PVDF hollow fibre membranes for desalination through directcontact membrane distillation, Separation and Purification Technology 69(2009) 78–86.

56] F. Tasselli, J.C. Jansen, F. Sidari, E. Drioli, Morphology and transport propertycontrol of modified poly(ether ether ketone) (PEEKWC) hollow fibre mem-branes prepared from PEEKWC/PVP blends: influence of the relative humidityin the air gap, Journal of Membrane Science 255 (2005) 13–22.

57] J.J. Qin, Y.M. Cao, Y.Q. Li, Y. Li, M.H. Oo, H. Lee, Hollow fibre ultrafiltrationmembranes made from blends of PAN and PVP, Separation and PurificationTechnology 36 (2004) 149–155.

58] B. Jung, J.K. Yoon, B. Kim, H.W. Rhee, Effect of molecular weight of polymeric

additives on formation, permeation properties and hypochlorite treatment ofasymmetric polyacrylonitrile membranes, Journal of Membrane Science 243(2004) 45–57.

59] J.J Qin, M.H. Oo, Y. Li, Development of high flux polyethersulfone hollow fibreultrafiltration membranes from a low critical solution temperature dope viahypochlorite treatment, Journal of Membrane Science 247 (2005) 137–142.

2 mbran

[

[

[

[

[

[

[

[

[

[

[ical properties of PVDF hollow fiber membranes from aspects of phaseinversion, crystallization and rheology, Journal of Membrane Science 340(2009) 192–205.

32 S. Simone et al. / Journal of Me

60] L.S. Wan, Z.K. Xu, Z.G. Wang, Leaching of PVP from polyacrylonitrile/PVPblending membranes: a comparative study of asymmetric and dense mem-branes, Journal of Polymer Science Part B: Polymer Physics 44 (2006)1490–1498.

61] S. Rouaix, C. Causserand, P. Aimar, Experimental study of the effects ofhypochlorite on polysulfone membrane properties, Journal of Membrane Sci-ence 277 (2006) 137–147.

62] I.M. Wienk, E.E.B. Meuleman, Z. Borneman, T. van den Boomgaard, C.A. Smol-ders, Chemical treatment of a polymer blend: mechanism of the reaction ofhypochlorite with poly(vinyl pyrrolidone), Journal of Polymer Science Part A:Polymer Chemistry 33 (1995) 49–54.

63] R.M. Boom, I.M. Wienk, T. van den Boomgaard, C.A. Smolders, Microstructuresin phase inversion membranes. Part 2. The role of a polymeric additive, Journalof Membrane Science 73 (1992) 277–292.

64] R.M. Boom, T. van den Boomgaard, C.A. Smolders, Mass transfer and thermody-

namics during immersion precipitation for a two-polymer system: evaluationwith the system PES–PVP–NMP–water, Journal of Membrane Science 90 (1994)231–249.

65] M.J. Han, S.T. Nam, Thermodynamic and rheological variation in polysulfonesolution by PVP and its effect in the preparation of phase inversion membrane,Journal of Membrane Science 202 (2002) 55–61.

[

e Science 364 (2010) 219–232

66] C.A. Smolders, A.J. Reuvers, R.M. Boom, I.M. Wienk, Microstructures in phase-inversion membranes. Part 1. Formation of macrovoids, Journal of MembraneScience 73 (1992) 259–275.

67] N. Chen, L. Hong, Surface phase morphology and composition of the castingfilms of PVDF–PVP blend, Polymer 43 (2002) 1429–1436.

68] K.W. Lee, B.K. Seo, S.T. Nam, M.J. Han, Trade-off between thermodynamicenhancement and kinetic hindrance during phase inversion in the preparationof polysulfone membranes, Desalination 159 (2003) 289–296.

69] A. Bottino, G. Capannelli, S. Munari, Effect of coagulation medium on propertiesof sulphonated polyvinylene fluoride membranes, Journal of Applied PolymerScience 30 (1985) 3009–3022.

70] P. Sukitpaneenit, T.S. Chung, Molecular elucidation of morphology and mechan-

71] N. Peng, T.S. Chung, K.Y. Wang, Macrovoid evolution and critical factors to formmacrovoid-freee hollow fibre membranes, Journal of Membrane Science 61(2008) 363–372.