pbi licl membrane

Journal of Membrane Science 363 (2010) 149159

Contents lists available at ScienceDirect

Journal of Membrane Science

journa l homepage: www.e lsev ier .com

Pervap thr(PBI)-b tion

Yan Wana Department o ing Drb PBI Performan

a r t i c l

Article history:Received 18 MReceived in reAccepted 14 JuAvailable onlin

Keywords:PervaporationPolybenzimidazoleDual-layer hollow ber membraneEthylene glycolDewater

odityglyc

bermrepa

hollow ber membranes; and (3) PBI/PEI dual-layer hollow ber membranes. PBI at dense membraneshave the lowest separation performance due to severe swelling. PBI single-layer hollow bermembranesshow better separation performance in terms of permeation ux and separation factor but have very lowtensile strains. The dual-layer PBI/PEI hollowbermembranes have the best separation performance dueto (1) unique combination of the superior physicochemical properties of the PBI selective layer and theless swelling characteristics of the PEI supporting layer, and (2) synergistic effects ofmolecularly designed

1. Introdu

Ethylenenon-volatileing point. Itof many indbeen workidue to its goAll the aforfrom ethylelene glycolfollowed byof water isversion. Ththe productproduction

CorresponE-mail add

0376-7388/$ doi:10.1016/j.membranemorphology via dual-layer co-extrusion. The effects of spinningparameters of PBI single-layerand PBI/PEI dual-layer hollow ber membranes on pervaporation performance have been investigated.A thermal treatment of PBI/PEI dual-layer hollow ber membranes at 75 C can signicantly enhancethe separation performance. Compared with other polymeric membranes, the newly developed PBI/PEIdual-layer hollow ber membranes have much better separation factors and slightly lower uxes for theethylene glycol dehydration. It is believed that the science and engineering of designing PBI/PEI dual-layer hollow ber membranes with an ultra-thin functional separation layer and a synergic supportinglayer may open new perspective for the development of next-generation high-performance multilayermembranes for liquid separations.

2010 Elsevier B.V. All rights reserved.

ction

glycol is an important chemical widely used as theantifreeze and de-icing agent because of its low freez-is also an important raw material for the manufactureustrial products [1,3]. In addition, ethylene glycol hasng as an ideal absorbent for natural gas dehydrationod afnity to water and high boiling point (197 C) [1].ementioned applications involve separation of waterne glycol. Whats more, during the production of ethy-, a direct oxidation of ethylene to ethylene oxide isthe hydrolysis of ethylene oxide, thus a large excessused in the hydrolysis reaction to enhance the con-e excess water must be removed later to concentrate. Therefore, dewatering becomes a critical issue in theand recycle process of ethylene glycol. Table 1 lists

ding author. Tel.: +65 6516 6645; fax: +65 6779 1936.ress: [email protected] (T.S. Chung).

some physicochemical properties of ethylene glycol and water[24].

Currently, the ethylene glycol dehydration is done by con-ventional multi-stage evaporation units followed by distillationcolumns, characterizedbyhighcapital costs. Evaporation is suitablefor concentrating ethylene glycol up to a concentration of 70wt.%.However, beyond this point, the quality of the overhead productdiminishes and the energy demand increases signicantly. Distil-lation can be used to further concentrate ethylene glycol beyond70wt.% and solve the quality problem, but still consumes a largeamountof energy.Actually, the separationof ethyleneglycolwatermixture is ranked the eighthmost energy intensive distillation pro-cesses in the chemical industry because the high boiling point ofethylene glycol requires a high-pressure steamof the reboiler [5,6].Therefore, more cost effective technologies are urgently needed inthe chemical industry in order to stay competitive.

Pervaporation is the most promising technology for molecular-scale liquid/liquid separations existing in biorenery, petrochem-ical, and pharmaceutical industries because of its highly selective,economical, energy efcient and eco-friendly characteristics. Since

see front matter 2010 Elsevier B.V. All rights reserved.memsci.2010.07.024oration dehydration of ethylene glycolased membranes. 1. Membrane fabrica

ga, Michael Gruenderb, Tai Shung Chunga,

f Chemical and Biomolecular Engineering, National University of Singapore, 4 Engineerce Products, Inc., 9800-D Southern Pine Boulevard, Charlotte, NC 28273, USA

e i n f o

ay 2010vised form 12 July 2010ly 2010e 21 July 2010

a b s t r a c t

Ethylene glycol is an important commtheproduction and recycle of ethylenezole/polyetherimide (PBI/PEI) hollowThree typesofmembraneshavebeenp/ locate /memsci

ough polybenzimidazole

ive 4, 10 Kent Ridge Cresent, Singapore 117576, Singapore

in chemical industries and dewatering is a critical process inol. In thiswork,wehave developeddual-layer polybenzimida-embranes for ethyleneglycol dehydrationviapervaporation.

red;namely, (1) PBIat densemembranes; (2) PBI single-layer

150 Y. Wang et al. / Journal of Membrane Science 363 (2010) 149159

Table 1The physicochemical properties of ethylene glycol and water.

Ethylene glycol Water

Chemical strFormulaCAS numberMolar massDensity (g/cMolecular voAppearance

Melting poinBoiling poinPolarity para

(kcal/mol)Solubility pa

dphsp

Vapor pressu25 C60 C

a The molecdensity and th

only part ofundergo phenergy efcpervaporatisolve the qution in the p

In the ction has bestage, suchpolyacrylonHowever, thaqueous soldecrease intions to depsince the ptreatmentsAnother feacomprisedstrate(s) mlayer provioffers mech

Composbeen reportmeric membecause ofof the ceramalso showedehydrationbers studi[2023] havergistic sepachieved ifeven witho

Polybenmericmateof their outthermal stawith stablesesses bothwhich are c

PBI is also known to absorb 15wt.% water at equilibrium and thewater in PBI is mobile [29]. Water can preferentially permeate thePBI membrane due to the strong water afnity with PBI moleculesandamuchsmallermolecular sizeofwater.All theabovecharacter-

f PBIal foranyaembrt-ind, whiranee hyesulrm preforan

ranePEI)teriapertup-tof ketionand

otent-gennal swill dtionsepaties o

erim

ateri

PBIts, Inylacl serutions. Figylpy

DawuctureC2H4(OH)2 H2O107-21-1 7732-18-5

(g/mol) 62.068 18.015m3) 1.113 0.998lume (3)a 92.4 29.9

Odorless, colorless,syrupy liquid

Colorless,odorless liquid

t (C) 12.9 0t 197.3 100.0meter ET (30)[2]

56.3 63.1

rameter (MPa)1/2 [3]17.0 15.511.0 16.026.0 42.333.0 47.8

re (mmHg)0.07 23.761.34 149.44

ular volume is calculated by the molecular weight divided by thee Avogadro number [4].

the feed is vaporized and the remaining feed does notase change, the pervaporation process is much moreient than the conventional distillation process. Hence,on has gradually gained acceptance in industries toality and energy problems in ethylene glycol dehydra-ast 20 years.ase of pervaporation dehydration, signicant atten-en given to highly hydrophilic polymers in the earlieras poly(acrylic acid) (PAA), poly(vinyl alcohol) (PVA),itrile (PAN), sodium alginate, and chitosan, etc. [7].ese materials lack mechanical strength and stability inutions due to the excessive swelling, leading to a drasticseparation performance [8,9]. Cross-linking modica-ress the membrane swelling may not be very desirableermeation ux is often compromised. The additionalalso incur extra costs andprolongproductiondurations.sible method is to employ the composite membraneof a thin active layer and supported microporous sub-

istics omateriof somtionmsolvendryingmembextremtions, rlong-te

Thesciencemembimide (the macal prowatereffectsseparabranesmay pof nexfunctio2),we(operaon theproper

2. Exp

2.1. M

TheProducdimethThe LiCthe solplasticPolyvinof 30kade of less swelling material, where the outer surfacedes the selectivity, while the porous substrate layeranical strength and high permeability.ite membranes for pervaporation applications haveedextensively in recentyears. Ceramic-supportedpoly-branes have attracted much attention recently [1015]the unique chemical, mechanical and thermal stabilityic substrates. Many polymeric composite membranes

d much improved separation performance for ethanol[1619]. Especially, co-extrusion dual-layer hollow

ed by our group as pervaporation membranes recentlye shown promising pervaporation performance. Syn-aration performance for organic dehydration can beinner- and outer-layer materials are properly selectedut intensive thermal or chemical treatment [2023].zimidazole (PBI) is a high-performance aromatic poly-rial, suitable formanyaggressive environments becausestanding chemical resistance, mechanical strength andbility. It is a glassy polymer with high Tg (417 C) [24]mechanical properties up to 350 C. In addition, PBI pos-donor and acceptor hydrogen-bonding sites [25,26],apable of participating in specic interactions [27,28].

Polymers wDMAc, emppreparationas receivedwith deioni

2.2. The fab

Flat-shePBI/DMAcPBI/DMAc/Loriginal suping the polof a 250m75 C for 15tant lmwabelow. A Mmembrane

The 1st pimmersed irating fromwas continmaterial make it a promising pervaporationmembranethe dehydration of various organics. However, in spitedvantages, theapplicationofPBImaterial aspervapora-anes is very limited [21,30]. This is possibly because theucedphase-inversion PBImembrane is very brittle afterch makes it difcult to be fabricated into self-standings. Another disadvantage is that the PBI membrane withdrophilicity would be easily swollen in aqueous solu-ting in a decreased membrane selectivity and unstableerformance as stated above.e, the aims of this work are to (1) investigate thed engineering of fabricating dual-layer hollow berswithaPBIouter layer andananti-swellingpolyether-inner layer for ethyleneglycol dehydration; (2) examinels synergism in terms of their unique physicochemi-ies (i.e., superior hydrophilicity of PBI, extremely lowake of PEI, and their miscibility); and (3) identify they spinning parameters on membrane morphology andperformance. We also fabricated PBI at dense mem-single-layer hollow bers for comparison. This study

tially open up new perspectives for the developmenteration pervaporation membranes with an ultra-thinelective layer. In the following associated paper (Partiscuss the effects of pervaporationoperation conditionstemperature, feed concentration and vacuum pressure)ration performance, in order to investigate the intrinsicf the membrane (permeability and selectivity).

ental

als

polymer solution was provided by PBI Performancec. with a composition of 26.2wt.% PBI, 72.3wt.% N,N-etimide (DMAc), and 1.5wt.% lithium chloride (LiCl).ves the function of preventing PBI from phasing out of. Ultem 1010 polyetherimide was purchased from GE. 1 shows the chemical structures of PBI and PEI [8,24].rrolidone (PVP) (Merck, Singapore)with an averageMwasemployedas anadditive for thehollowber spinning.ere dried overnight at 120 C under vacuum before use.loyed as the solvent in bore uid and for membrane, was supplied byMerckwith analytical grade and used. Ethylene glycol of analytical grade was used to mixzed water to prepare the binary feed solution.

rication of PBI at-sheet dense membranes

et PBI dense membranes were cast from a 15wt.%polymer solution. The polymer dope solution ofiCl (15.0/84.1/0.9wt.%) was prepared by diluting thepliedPBI solution. Themembranewaspreparedbycast-ymer solution onto a glass plate with a casting knife

thickness, and then placed on a hot plate preset ath, to allow the solvent evaporated slowly. The resul-s driedwith twodifferent drying protocols as describeditutoyo micrometer was employed to measure the nalthickness.rotocol: The resultant lm togetherwith glass platewasn a water coagulation bath. After the membrane sepa-the glass plate automatically within several minutes, ituously immersed in water for 2 days with fresh water

Y. Wang et al. / Journal of Membrane Science 363 (2010) 149159 151

Fig. 1. The chemical structures of (a) polybenzimidazole (PBI) and (b) Ultem 1010 polyetherimide (PEI).

changed daily. Thewetmembranewas then dried in a Freeze-dryer(Thermo Electron Co.ModulyoD-230)with vacuumovernight afterfrozen in a freezer for 2h. Using this drying protocol, the LiCl maybe removed from the as-fabricated PBI dense membrane.

The 2ndprotocol:The resultant lmwaspeeledoff carefully fromthe glass plate and then dried between two wire meshes in a vac-uum oven, with temperature gradually increasing to 250 C at arate of 12 C/20min and hold there for 24h to remove the resid-ual solvents before cooling down naturally. The wire meshes notonly prevented the membrane from sticking to the glass plate butalso helped uniformly dry the membrane from both surfaces. Withthis drying protocol, the LiCl remains in the as-fabricated PBI densemembrane.

2.3. Spinning process and modules fabrication of hollow bermembranes

The schlow ber spThe polymesyringe pum

was employed as the bore uid in order to make a porous innersurface. The hollow ber was spun by extrusion of the polymersolution and bore uid out of the spinneret orice and subsequentphase inversion ina coagulantbathwithapre-set air gap. Bothdopeuid and bore uid were ltered through 15m sintered metal l-ters before spinning. Tap water was used as the external coagulantat room temperature. The nascent bers were rolled up by a drum,cut into segments, and then rinsed in a clean tapping water bathfor at least 3 days to remove the remaining DMAc. The as-spun hol-low bers were dried in air naturally after freeze-drying, and thenstored in ambient environment.

For single-layer hollow bers, the dope solution contains23wt.% PBI diluted from the original supplied PBI solution. Table 2lists the detailed spinning parameters of the single-layer co-extrusion process with various bore uid ow rates and take-up

. ForPBI

n; wn cond sphe 23

Table 2The spinning p

Parameters

Dope solutio 7:1.33Bore uid coDimensionsTemperatureExternal coaDope ow raAir gap distaMembrane I CBore uid o 1.5Take-up spe 5.15

Table 3The spinning p

Parameters

Outer-layerInner-layer dBore uid coDimensionsExternal coaTemperatureOuter-layerInner-layer dBore uid oMembrane IAir gap distaTake-up speematic diagrams of single-layer and dual-layer hol-inning systems have been described elsewhere [22].r solution was degassed for 24h before loading into ap (ISCO 1000). A mixture of 85/15 (w/w) DMAc/water

speeds23wt.%solutiosolutiodetailecess. T

arameters of single-layer hollow ber membranes.

Range of variables

n composition (wt.%) PBI:DMAc:LiCl (23:75.6mposition (wt.%) DMAc:water (85:15)of spinneret (mm) (inner diameter/outer diameter) 1.05/1.60(C) Ambient (232)

gulant Waterte (ml/min) 2nce (cm) 2D (PBI-S-) A Bw rate (ml/min) 1 1.25ed (m/min) 5.15 5.15arameters of dual-layer hollow ber membranes.

Range of variables

dope solution composition (wt.%) PBI:DMAc:LiCl (23:75.67:1.33)ope solution composition (wt.%) PEI (Ultem 1010):PVP:DMAc (25:5:70)mposition (wt.%) DMAc:water (85:15)of spinneret (mm) OD1/OD2/ID (1.20/0.97/0.44)gulant Water(C) Ambient (232)

dope ow rate (ml/min) 0.5ope ow rate (ml/min) 4w rate (ml/min) 2D (PBI-D-) A B Cnce (cm) 5 2 1ed (m/min) 4.60 (free fall)dual-layer hollow bers, the outer-layer dope was apolymer solution, the same as the single-layer dopehile the inner-layer dope was a 25wt.% PEI polymersisting of 25/5/70wt.% PEI/PVP/DMAc. Table 3 lists the

inning parameters for the dual-layer co-extrusion pro-wt.% PBI/DMAc solution was chosen for the outer layer

)

D E F G H2 1.5 1.5 1.5 1.55.15 4.60 9.59 16.24 21.79D E F2 2 29.59 16.24 21.79


Fig. 2. Viscosity of PBI/DMAc solutions with different PBI concentrations.

because the critical concentration of PBI/DMAc solution was about22.5wt.% as shown in the viscosityconcentration curve in Fig. 2provided by PBI [31(. Generally concentrations near or higher thanthe critical value were chosen for the preparation of asymmetricgas membrdefects [32pervaporati

The pervof hollow ing with tweffective lenand cured fration modtreatment ocation if it wimmersingfor 2h, folloair-drying a

2.4. Pervap

A staticdesign [35]at-sheet mis shown insteel permethe charactpervaporati

have been described elsewhere [36]. A feed solution of 50/50wt.%water/ethylene glycolwas used for both pervaporationunits unlessotherwise stated. Since the feed concentration varies less than0.5wt.% during the entire experiment because of a large quantityof the feed solution compared to the permeate sample, it was con-sidered to be constant. The operational temperature was 60 C. Thefeed ow rate was 1.38 l/min for the at-sheet dense membraneand 0.5 l/min for each hollow ber module. The permeate pressurewas maintained less than 3mbar by a vacuum pump. Retentateand permeate samples were collected after the membrane beingconditioned for about 2h.

The ux J was determined by the mass of permeate divided bythe product of the interval time and membrane area. The mass ofpermeate was weighed using a Mettler Toledo balance. The sepa-ration factor is dened by the equation below:

= yw,1/yw,2xw,1/xw,2

(1)

where subscripts 1 and 2 refer towater and ethylene glycol, respec-tively; yw and xw are the weight fractions of component in thepermeate and feed, and were analyzed through a Hewlett-PackardGC 7890 A with a HP-INNOWAX column (packed with cross-linkedpolyethylene glycol) and a TCD detector.

embr

mor6700llowthelatinuemb

a. Thtensare.% relbothd invree

ults

rvap

at-rvapne glanes to make sure a quasi-dense skin layer with less34] but the concept is also applicable to asymmetricon membranes [8,20,22,23].aporationmoduleswere prepared by loading one pieceber membranes into a peruoroalkoxy tubing connect-o Swagelok stainless steel male run tees with angth of around 20 cm. Both ends were sealed by epoxyor 24h at ambient temperature. At least two pervapo-ules were tested for each membrane sample. Thermalf hollow bers was carried out before module fabri-as applied. All thermal treatments were conducted by

the hollowbermembranes in a hotwater bath at 75 Cwed by naturally cooling down in the water bath andfter that.

oration study

pervaporation cell made according to Prof. Matsuuraswas used to test the pervaporation performance of theembrane at room temperature. The design schematicFig. 3. A testing membrane was placed in the stainlessation cell with an effective surface area of 15.2 cm2. Forerization of hollow ber membranes, a laboratory scaleon unit was employed and the details of the apparatus

2.5. M

Thea JSM-The hoturingwithpber mcamerusing a2 softwand 80at themetholeast th

3. Res

3.1. Pe

Thethe peethyleFig. 3. Schematic of the batch pervaporation separation systemane characterization

phology of hollowbermembraneswas observedusingF eld emission scanning electronmicroscope (FESEM).ber sample for SEM observationwas prepared by frac-membrane strip in liquid nitrogen and then coating itm. Photographsof thePBI densemembrane andhollowrane after pervaporation were taken by conventionale mechanical properties of hollow bers were testedilemeter INSTRON 5542 and analyzedwith the BluehillThe tests were carried out at room temperature (25 C)ative humidity. Each hollow ber sample was clampedends with an initial gauge length of 50mm and the testolved stretching at a rate of 10mm/min until failure. Atsamples were tested for each membrane.

and discussion

oration performance of the PBI dense membrane

sheet dense membrane is studied rstly to investigateoration performance of the neat PBI material for theycol dehydration. The separation performance of PBI

for at-sheet membranes.


Table 4Pervaporation performance of PBI dense membranes.

Membrane Operation mode Permeate (EG, wt.%) Normalized total ux (gm/m2h) Separation factor (water/EG)

Membrane 1 (freeze dried) Normal modea 14.04 1232 6.0Reverse modeb 10.08 1131 9.0

Membrane 2 (thermal treated) Normal modea 12.10 966 7.1Reverse modeb 7.05 784 13.1

a Membrane is mounted with the air-side facing the feed;b Membrane is mounted with the glass-side facing the feed.

densemembranes preparedbydifferent dryingprotocols andoper-ated in different modes is shown in Table 4. The results show thethermal treated dense membrane is of higher separation factor butlowerpermeationux,which ismainlydue toa smaller freevolumeand the elimination of micro-defects in the thin separating layer,as well as athe separatof operatiomembrane-the membrmembraneswhat lowerthe feed. T[8,3739]. Thigher hydmolecules tand (2) a leoperation m

Neverthdense mem10 and theto polyimiddense memwater but aserious swesive water aswelling anvaporationPBI hollowenhanced psections.

3.2. Pervapmembranes

Table 5PBI hollowup speeds fdense memimproved searation factmany factomuch bettemer provid

ig. 4. Photograph of the PBI dense membrane after pervaporation.

5. Photograph of the PBI single-layer hollow ber after pervaporation.

swelling than the latter. In addition, hollow bers have self-ed vacuum channels where the feed is supplied from the

ide while vacuum is applied on the lumen side. The porousy substructure of asymmetric hollow bers attributes to aelling of the selective layer, thus achieve a higher separa-ctor. As shown in Fig. 5, no severe swelling phenomenonsingle-layer hollow ber membrane after pervaporation ised. During thewhole pervaporation process of about 810h,s no obvious decline in pervaporation performance.le 5 shows the effects of bore uid rate and take-up speedpervaporation performance of single-layer PBI hollow berranes. These effects have been studied extensively on mem-for gas separation [4042] and morphology manipulations

]. Not only the bore uid rate affects the demixing process in

Table 5Pervaporation arameters.

Membrane I rmeate (H2O, wt.%) Total ux (g/m2h) Separation factor

PBI-S-APBI-S-BPBI-S-CPBI-S-DPBI-S-EPBI-S-FPBI-S-GPBI-S-Hhigher transportation resistance. Table 4 also comparesion performance of dense membranes as a functionn modes. The separation properties depend upon themounting mode (i.e., using air-side or glass-side ofane against the feed during pervaporation tests). Bothexhibit much higher separation factors with some-uxes when the bottom side is placed facing against

his phenomenon is consistent with previous studieshe possible causes are (1) the bottom surface has a

rophilicity because the hydrophilic-end groups of PBIend to localize at the bottom toward the glass plate,ss swelling of the dense-selective layer in the reversedode, as reported in our previous study [8].

eless, the pervaporation performance of at-sheet PBIbranes is very poor. The separation factor is aroundnormalized total ux is about 1kgm/m2h. Comparede materials investigated in our previous study [8], PBIbranes show an extreme low separation factor towardshigher ux. This poor performance may be due to thelling in aqueous feed solutions because PBI has impres-fnity [29]. In agreement with our hypothesis, severe

dwrinkles canbeobservedon themembranes after per-tests as shown in Fig. 4. To lower the swelling problem,ber membranes are prepared in order to achieve anervaporation performance as discussed in the following

oration performance of single-layer PBI hollow ber

shows the pervaporation performance of single-layerbers spun with various bore uid ow rates and take-or ethylene glycol dehydration. Compared with the PBIbranes, the single-layer PBI hollow bers show muchparationperformance inbothpermeationuxand sep-or. The substantial enhancement in ux arises fromrs. The conguration of asymmetric hollow bers isr than that of at dense membranes because the for-es a larger surface area, less transport resistance, and

F

Fig.

lowercontainshell sand drless swtion faof theobservthere i

Tabon themembbranes[43,44

performance of PBI single-layer hollow ber membranes with different spinning p

D Bore uid ow rate (ml/min) Take-up speed (m/min) Pe1 5.15 97.141.25 5.15 98.991.5 5.15 99.072 5.15 98.571.5 4.60 (free fall) 99.131.5 9.59 98.811.5 16.24 99.021.5 21.79 99.12348 35449 100490 109817 71592 105

1314 891372 1051147 116


the inner skon the hollothe molecuThis is due tmay increafer is facilitin an increathe as-spundirection. FPBI-S-A to Deter and redmorphologseparationtabulated inin ux posshows an umolecular ouid initialis continuoand-downfor variousthe bore uand its effefor pervapo

Table 5ration perfouid rate. Tthe separatelongationaFig. 6. SEM morphology of PBI single-layer hollow ber membranes

in and its morphology, but also plays an important rolew bers inner diameter and wall thickness, as well aslar orientation of the outer skin in the hoop direction.o the fact that the solidication rate at the inner surfacese with increasing bore uid rate since the mass trans-ated. Therefore, an increase in bore uid rate resultsse in inner diameter with a reduced wall thickness ofber and a slightly stretched outer skin in the hoop

ig. 6 illustrates the SEM morphologies of hollow berswith increasing bore uid rate. Increased inner diam-uced wall thickness can be observed. Consistent with

ical changes, clear differences can be observed in theirperformance in terms of ux and separation factor asTable 5. Increasing bore uid rate results in an increase

sibly because of a thinner wall. The separation factorp-and-down trend possibly because of the enhancedrientation in the hoop direction induced by the borely, but minor defects are created if the bore uid rateusly increased. The stress-induced orientation and up-trends in separation performance have been reportedmembranes [4446]. However, to our best knowledge,id induced molecular orientation in the hoop directioncts on separation performance have not been reportedration membranes.also compares the effect of take-up speed on pervapo-rmance for membranes made under a constant borehe ux increases with increasing take-up speed, whileion factor does not show obvious trend. Generally, thel or extensional stress upon the nascent ber is the pre-

dominant etake-up spthickness (at the outerreason contentationmaan increasechains to brchain orientor. Fig. 7 sber membthe higherlooser the i

Howevethe major dtleness. Forthe hollowwithstandTable 6 tabuspun from dall single-lathe poor e

3.3. Pervapmembranes

To enharation perfohollow bewith different bore uid ow rates.

xternal stress on the spinning line [43,46,47]. A higheed not only reduces hollow ber diameter and wallas shown in Fig. 7), but also induces chain orientationselective skin. The reduced wall thickness is the mainributing to the enhanced ux, while the enhanced ori-y attribute to the improved separation factor. However,in take-up speed also causes over-stretched polymereak at some weak points, offsetting the benets of thetation, which leads to nearly unchanged separation fac-hows the SEM morphology of PBI single-layer hollowranes as a function of take-up speed and conrms thatthe take-up speed, the denser the outer edge and thenner surface.r, in spite of the enhanced pervaporation performance,rawback of the PBI single-layer hollow bers is its brit-practical pervaporation applications, it is desired thatber has adequate mechanical strength so that it canharsh operation environments over numerous cycles.lates themechanical properties of the PBI hollowbersifferent conditions and shows the maximum strains ofyer hollow ber membranes are very low, indicatingxibility of these bers for module fabrication.

oration performance of the dual-layer PBI hollow ber

nce mechanical properties and further improve sepa-rmance of hollow ber membranes, PBI/PEI dual-layerr membranes are further developed with the aid of PEI


as the innerthe followin

(a) PEI hasation anselectiv

(b) PEI hashardly smakessupportbranes.

(c) The PEIport for

(d) An impmateriaadhesio

Table 6Mechanical str

Membrane I

PBI-S-APBI-S-BPBI-S-CPBI-S-DPBI-S-EPBI-S-FPBI-S-GPBI-S-HFig. 7. SEM morphology of PBI single-layer hollow ber membrane

supporting layer. The PEI material is chosen because ofg reasons:

been studied as a membrane material for vapor perme-d pervaporation [48], and showed preferential waterity over organic chemicals.shown very low water adsorption (1.9wt.%) [20]. Ituffers from swelling like most other polymers, whichit especially attractive to work as an anti-swellinging layer in the design of dual-layer hollow ber mem-

inner layer can provide the required mechanical sup-the relative brittle PBI outer layer.ortant factor worthy of mention is that PBI and PEIls are miscible at the molecular level [28,49], thus goodn between these two layers may be achievable.

ength of PBI single-layer hollow bers.

D Max tensilestress (MPa)

Youngsmodulus (MPa)

Max strain(mm/mm)

12.7 1614 0.01112.9 1532 0.01311.7 1506 0.01012.4 1488 0.01015.9 1436 0.01316.3 1229 0.01915.0 1175 0.01910.1 1277 0.010

Fig. 8 shlow ber s4.60m/min1250m. Touter layerlayer have acan be obseing a gooda dense strdefect-lessis very porotransport re

The intecompositedependingduring phasadditives uboth dopesinversion bformation othis study cbetween th(2) inter-peof the interpotential dspinning so

Table 7layer PBI/PEparameterslayer PBI hs with different take-up speeds.

ows the SEM morphology of the PBI/PEI dual-layer hol-pun from an air gap of 2 cm and a take-up speed of(PBI-D-B). The hollow ber has a diameter of abouthe dual-layer wall thickness is about 240m and thethickness is about 16m. Both inner layer and outersymmetric cross-sectionmorphology. Nodelaminationrved at the interface between the two layers, indicat-adhesion. The outer surface of the outer layer revealsucture at a high magnication (50,000), which is aselective layer, while the inner surface of the inner layerus which is desirable since it does not constitute muchsistance.rfacial morphology is a unique feature in dual-layermembranes. A seamless interface may be obtainedon the miscibility of both dopes and many other factorse inversion. If the solvents, non-solvents, polymers, andsed in both dopes are thermodynamically compatible,may diffuse into each other before and during phaseecause of chemical potential differences [23,50]. Thef the desired interface between PEI and PBI layers inan be attributed to several factors: (1) the miscibilityese twopolymers as provedbyprevious studies [28,49];netration occurs between these two dopes becausefacial diffusions and convections driven by chemicalifferences; (3) a common solvent (DMAc) used in thelutions of both layers.summarizes the pervaporation performance of dual-I hollow bermembranes spunwith different spinningfor ethylene glycol dehydration. Comparedwith single-ollow ber membranes, the dual-layer hollow ber


EI du

membranesparable uxthe lowsweand dry staand elongatThe observaafter pervapthebers reswelling of

Table 7air gap anddual-layer Pgas separati[20,22,46,5distance malayer; (2) ththe outer susion and focreate moreresults in ation; (4) a dare over-strhigher sepaa decline inslightly witand C; whil

hestits d

r incrr-stranc

effeof sembembtionembese ts du

Table 7Pervaporation

Membrane I

PBI-D-APBI-D-BPBI-D-CPBI-D-DPBI-D-EPBI-D-FFig. 8. FESEM images of the membrane morphology of PBI/P

have overall much higher separation factors and com-es. This improvement may arise from three factors: (1)lling characteristicsof thePEI inner layer; (2) theporoustus of the PEI inner layer; (3) the decoupling of shearional stresses in the dual-layer hollow ber fabrication.tion of the dual-layer PBI/PEI hollow ber membranesoration conrms our 1st and 2nd hypotheses becausemain their original straight shapes, indicatingno severeeach individual hollow ber membrane.also shows us the effects of spinning parameters (i.e.,take-up speed) on the pervaporation performance ofBI/PEI hollowbermembranes. The effects of air gapononperformance of hollowbers have beenwell studied

the higwherefurthebe oveperform

Thesectionber mber mseparafrom mthat thstresse1] and canbe summarized as follows: (1) a longer air gapy increase solvent evaporation, causing a denser outere surrounding moisture may induce gelation and causerface to be viscous, thusminimizing non-solvent intru-rming a denser skin; (3) a longer air gap distance maystretching by both gravity and take-up speed, which

thinner selective layer with enhanced polymer orienta-efective selective skinmay be formed if polymer chainsetched and torn apart. The former three factors lead to aration factorbut a lowerux,while the last factor causesseparation factor. As shown in Table 7, ux increases

h decreasing air gap distance formembranes PBI-D-A, Be separation factor shows an up-and-down trend, with

tions and viit will bearstretched, oalso lists thtake-up spewhile a higup speed foa thinner setor. In additperformanc

Table 8ber membdual-layer

performance of PBI/PEI dual-layer hollow ber membranes.

D Air gap (cm) Take-up speed(m/min)

OD (m) ID (m) Outer-thickn

5 4.60 1222 721 14.22 4.60 1229 752 17.51 4.60 1226 725 14.62 9.59 899 589 9.52 16.24 686 425 5.22 21.79 597 376 4.8al-layer hollow bers (PBI-D-B).

separation factor occurring at air gap of 2 cm (PBI-D-B)ense-selective layer becomes thinner and oriented. Aease in air gap to 5 cm, the dense-selective layer mayetched, causing minor defects and a lower separatione.cts of take-up speed have been discussed in the previousingle-layer hollow bers. However, dual-layer hollowranes are somewhat different from single-layer hollowranes. The former shows increased ux but decreasedfactor with an increase in take-up speed, as observedranes PBI-D-B, D, E and F in Table 7. It is due to the factwo layers experience different shear and elongationalring spinning since they have quite different composi-

scosities. Because the outer layer has a higher viscosity,more loads (i.e., elongational stresses). Thus it can beriented, and become thinner simultaneously. Table 7e membrane diameter which decreases with increasinged. Since a thin selective skin generates in a high ux,h take-up speed produces more defects, a higher take-r PBI/PEI dual-layer hollow ber membranes results inlective layer, a higher ux and a lower separation fac-ion, it seems that take-up speed may affect membranee more drastically than air gap distance.tabulates the tensile properties of dual-layer hollowranes, while Fig. 9 compares the tensile behavior ofPBI/PEI and single-layer PBI hollow ber membranes

layeress (m)

Permeate(H2O, wt.%)

Total ux(g/m2h)

Separationfactor

99.93 232 215699.96 241 228899.90 266 101699.76 492 43699.72 596 37399.67 732 303


Table 8Mechanical strength of PBI/PEI dual-layer hollow bers.

Membrane ID Max tensilestress (MPa)

Youngs modulus(MPa)

Max strain(mm/mm)

PBI-D-A 11.1 417 0.23PBI-D-B 12.7 423 0.21PBI-D-C 11.0 436 0.20PBI-D-D 9.4 325 0.24PBI-D-E 12.7 527 0.25PBI-D-F 13.2 600 0.28

Fig. 9. The typbranes.

using PBI-Slayer hollowsingle-layerexhibit highelongationtechnology

3.4. Effect o

Generallthermal trtreatmentsbecause ofing layer. Amotion of prelaxation aing. As a comorphologyresistance.

The effemembranePBI-D-C anration perfofor 2h. Thenicantly a

Acomith pr

C ha86g/eth

esear imhe rea plarat

mpaation

vapoious

Table 9Effect of the th

Membrane I

PBI-D-CPBI-D-C annPBI-D-FPBI-D-F ann

Feed compositical tensile behavior of single-layer and dual-layer hollowbermem-

-G and PBI-D-E membranes as examples. PBI-D-E dual-bers have much longer breaking strains than PBI-S-Ehollow bers, although the single-layer hollow berser tensile stress. This is a distinct proof of the improvedand exibility by means of the dual-layer co-extrusion.

Fig. 10.glycol w

PBI-D-ux (164wt.%other rAnotheaway tacts asthe sep

3.5. Codehydr

Perby varf heat post-treatment on membrane performance

y, asymmetric membranes are often exposed to post-eatments prior to industrial applications. Thermaltypically lead to enhanced separation performancethe elimination of micro-defects in the thin separat-treatment at high temperatures can promote thermalolymer chains and their interactions, facilitating chainnd rearrangement towards a denser and closer pack-nsequence, the thermally treated membrane will havewith a smaller free volumeandahigher transportation

cts of heat treatment might not be vivid if the as-spunalready has a high separation factor. Using membranesd PBI-D-F as examples, Table 9 shows their pervapo-rmance before and after thermal treatments at 75 Cux declines, while the separation factor increases sig-fter the thermal treatment. The annealed membrane

off betweealthough thability is ofRelatively,ally preferrtheoreticalloptimizatioandpoly(viethylene glnew potentods to impr

Fig. 10 slow ber mseparationilar operati64/36wt.%layer hollowmembraneshigher sepa

ermal treatment of PBI/PEI dual-layer hollow ber membranes on the pervaporation per

D Air gap (cm) Take-up speed (m/min) Permeate (H2

99.74ealed 1 4.60 99.96

99.69ealed 2 21.79 99.81

ion: EG/H2O (64/36wt.%)parisonofpervaporationperformance for thedehydrationof ethyleneevious studies.

s a separation factor of 4500 with a comparable highm2h) for the dehydration of a feed solution containingylene glycol. Similar phenomena have been observed byrches [8,23,30,5254] for various alcohol dehydrations.portant reason is that the hot water treatment washessidual DMAc solvent trapped inside the ber, where itasticizer. The ber will shrink due to solvent loss, thusion performance will be impacted.

rison with previous studies on pervaporationof ethylene glycol

ration dehydration of ethylene glycol had been studiedresearchers in recent years. Usually, there is a trade-n the permeability and selectivity in pervaporatione up-bound line is notwell dened yet. A high perme-ten accompanied with a low selectivity, and vice versa.membrane materials with high selectivity are gener-ed, as the disadvantage of low ux can be compensatedy by introducing asymmetry and suitable fabricationn. A literature search reveals that hydrophilic chitosan-nyl alcohol)-basedmembranes aremostwidely used forycol dehydration. Much research has been focused on

ial membrane materials as well as modication meth-ove membrane performance.hows a comparison between the PBI/PEI dual-layer hol-embranes and other polymeric membranes in terms offactor versus permeation ux [1,5,5565] under sim-ng conditions. Feed compositions of 50/50wt.% andethylene glycol/water are chosen for the PBI/PEI dual-ber membranes. The PBI/PEI dual-layer hollow ber

, especially those thermal treated ones present muchration factors than most other polymeric membranes.

formance.

O, wt.%) Total ux (g/m2h) Separation factor

222 1047186 4524758 592597 1004


Compared to the commercially available GFT membrane [58], thenewly developed PBI/PEI membranes have high separation factorsbut slightly lowuxes. Further efforts shouldbedevoted toenhancethe permeation ux by surface modication or spinning optimiza-tion.

4. Conclusion

In this study, PBI at-sheet dense membranes, single-layer anddual-layer hollow ber membranes are fabricated and studied forpervaporaticlusions can

(1) The PBItion permembrseparatductedwthe feed

(2) Comparmembrpermealayer anHoweveand ex

(3) Dual-ladevelopselectivbranes.balanceswellingablemeimportatwo fac

(4) The effetion perhave bebers wincreaseseparatand a lowith th

(5) A mildmembrciently.dehydramembrtreatedother pment ar

Acknowled

The auth279-597) foJing Cai Su aning. Thankfor their he

Nomenc

DMAcID

OD outer diameter of hollow berPBI polybenzimidazolePEIPAAPVAPANLiClPVPEG

EM

30)

nces

Du, Ang pombraneichainheim.Hanon, Fla.M. Hunes. Iy(viny. Fengater fembr. Hy

y(sulfdehy.hao, R07) 16ang,

ntal dmbran08) 21Qiao,meatiohols,. Peterentjes, Ceramic-supported thin PVA pervaporation membranes combin-high ux and high selectivity; contradicting the ux-selectivity paradigm,embr. Sci. 276 (2006) 42.. Peters, N.E. Benes, J.T.F. Keurentjes, Hybrid ceramic-supported thin PVAvaporation membranes: long-term performance and thermal stability indehydration of alcohols, J. Membr. Sci. 311 (2008) 7.Kreiter, D.P. Wolfs, C.W.R. Engelen, H.M. van Veen, J.F. Vente, High-perature pervaporation performance of ceramic-supported polyimidembranes in the dehydration of alcohols, J. Membr. Sci. 319 (2008) 126.. Zhu, S.S. Xia, G.P. Liu,W.Q. Jin, Preparation of ceramic-supported poly(vinylohol)-chitosan composite membranes and their applications in pervapo-ion dehydration of organic/water mixtures, J. Membr. Sci. 349 (2010).Jou, W. Yoshida, Y. Cohen, A novel ceramic-supported polymer membranepervaporation of dilute volatile organic compounds, J. Membr. Sci. 16299) 269.Yoshida, Y. Cohen, Ceramic-supported polymer membranes for pervapora-of binary organic/organic mixtures, J. Membr. Sci. 213 (2003) 145.

. Kim, K.H. Lee, S.Y. Kim, Pervaporation separation of water from ethanolough polyimide composite membranes, J. Membr. Sci. 169 (2000) 81.. Guan, T.S. Chung, Z. Huang, M.L. Chng, S. Kulprathipanja, Poly(vinyl alco-

)multilayermixedmatrixmembranes for thedehydrationof ethanolwaterture, J. Membr. Sci. 268 (2006) 113.anagishita, D. Kitamoto, K. Haraya, T. Nakane, T. Okada, H. Matsuda, Y. Ide-to, N. Koura, Separatioin performance of polyimide composite membranepared by dip coating process, J. Membr. Sci. 188 (2001) 165.on dehydration of ethylene glycol. The following con-be made from this study:

at-sheet dense membranes show very poor separa-formance for ethylene glycol dehydration due to severeane swelling. Experimental results indicate that ahigherion performance is obtained if the pervaporation is con-hen themembrane skin cast on the glass plate is facing

.ed to the at-sheet dense membranes, the hollow beranes show much better separation performance in bothtion ux and separation factor because of thin selectived porous substrate layer with less transport resistance.r, the single-layer hollow bers are of poor toughnessibility for module fabrication.yer PBI/PEI hollow ber membranes are successfullyedwith improvedmechanical toughness and enhancedity compared to the PBI single-layer hollow ber mem-The superior performance is attributed to both thed properties of the PBI selective layer and the lesscharacteristic of the PEI supporting layer. The desir-

mbranemorphologywith seamless interface is anothernt factor resulting in surprising synergism of the abovetors.cts of take-up speed and air gap distance on pervapora-formance of PBI/PEI dual-layer hollowbermembranesen studied. An increase in air gap results in hollowith a higher separation factor but a lower ux; while anin take-up speed results in hollow bers with a lower

ion factor but a higher ux. An optimal air gap of 2 cmw take-up speed of 4.60m/min result in membranes

e highest separation performance.thermal treatment of PBI/PEI dual-layer hollow beranes at 75 C can enhance separation performance ef-By comparison with previous works on pervaporationtion of ethylene glycol, PBI/PEI dual-layer hollow ber

anes developed in this study, especially the thermalmembranes, show higher separation factor than mostolymeric membranes. Further efforts for ux enhance-e needed in the future.

gement

ors thank PBI Performances Products, Inc. (R-279-000-r funding this research. Special thanks are given to Dr.nd Dr. Kai Yu Wang for their help on hollow ber spin-s also due toMr.Weijie BernardNeo andMr. Kun Chenglp on experiments.

lature

N,N-dimethylacetimideinner diameter of hollow ber

FESxw,iyw,i

ET (dphsp

Refere

[1] J.R.usime

[2] C. RWe

[3] C.MRat

[4] R.Ybrapol

[5] X.Sof wJ. M

[6] M.Npoland363

[7] P. S(20

[8] Y.Wmeme(20

[9] X.peralc

[10] T.AKeuingJ. M

[11] T.Aperthe

[12] R.temme

[13] Y.Xalcrat341

[14] J.D.for(19

[15] W.tion

[16] J.Hthr

[17] H.Mholmix

[18] H. Ymoprepolyetherimidepoly(amic acid)poly(vinyl alcohol)poly(acryl nitrile)lithium chloridepolyvinylpyrrolidoneethylene glycoleld emission scanning electron microscopeseparation factorweight fraction of component i in the feed (wt.%)weight fraction of component i in the permeate(wt.%)empirical solvent polarity parameter (kcal/mol)dispersion solubility parameter (MPa)1/2

polar force solubility parameter (MPa)1/2

hydrogen-bonding solubility parameter (MPa)1/2

total solubility parameter (MPa)1/2

. Chakma, X. Feng, Dehydration of ethylene glycol by pervaporationly(N,N-dimethylaminoethyl methacrylate)/polysulfone compositees, Sep. Purif. Technol. 64 (2008) 63.rdt, Solvents and Solvent Effects in Organic Chemistry, 2nd ed., VCH,, 1988.

sen,HansenSolubility Parameters:AUsersHandbook, CRCPress, Boca., 2007.ang, N.R. Jarvis, Separation of liquidmixtures by using polymermem-I. Permeation of aqueous alcohol solutions through cellophane andl alcohol), J. Appl. Polym. Sci. 14 (1970) 2341., R.Y.M.Huang, Pervaporationwith chitosanmembranes. 1. Separationrom ethylene glycol by a chitosan/polysulfone compositemembrane,. Sci. 116 (1996) 67.der, R.Y.M. Huang, P. Chen, Composite poly(vinyl alcohol)-one) membranes crosslinked by trimesoyl chloride: characterizationdration of ethylene glycolwater mixtures, J. Membr. Sci. 326 (2009)

.Y.M. Huang, Polymeric membrane pervaporation, J. Membr. Sci. 2872.L.Y. Jiang, T.Matsuura, T.S. Chung, S.H. Goh, Investigation of the funda-ifferences between polyamide-imide (PAI) and polyetherimide (PEI)es for isopropanol dehydration via pervaporation, J. Membr. Sci. 3187.T.S. Chung, Fundamental characteristics of sorption, swelling, andon of P84 co-polyimide membrane for pervaporation dehydration ofInd. Eng. Chem. Res. 44 (2005) 8938.rs, C.H.S. Poeth, N.E. Benes, H.C.W.M. Buijs, F.F. Vercauteren, J.T.F.


[19] D.M. Sullivan, M.L. Bruening, Ultrathin cross-linked polyimide pervaporationmembranes prepared from polyelectrolyte multilayers, J. Membr. Sci. 248(2005) 161.

[20] Y. Wang, S.H. Goh, T.S. Chung, N. Peng, Polyamide-imide/polyetherimidedual-layer hollow ber membranes for pervaporation dehydration of C1C4alcohols, J. Membr. Sci. 326 (2009) 222.

[21] K.Y.Wang, T.S. Chung, R. Rajagopalan,Dehydrationof tetrauoropropanol (TFP)by pervaporation via novel PBI/BTDA-TDI/MDI co-polyimide (P84) dual-layerhollow ber membranes, J. Membr. Sci. 287 (2007) 60.

[22] R.X. Liu, X.Y. Qiao, T.S. Chung, Dual-layer P84/polyethersulfone hollow ber forpervaporation dehydration of isopropanol, J. Membr. Sci. 294 (2007) 103.

[23] L.Y. Jiang, H. Chen, Y.C. Jean, T.S. Chung, Ultra-thin polymeric interpenetrationnetwork with separation performance approaching ceramic membranes forbiofuel, AIChE J. 55 (2009) 75.

[24] Y. Wang, S.H. Goh, T.S. Chung, Miscibility study of Torlon polyamide-imidewith Matrimid 5218 polyimide and polybenzimidazole, Polymer 48 (2007)2901.

[25] E.W. Neuse, Aromatic polybenzimidazoles: syntheses, properties, and applica-tions, Adv. Polym. Sci. 47 (1982) 1.

[26] A. Buckley, D. Stuetz, G.A. Serad, in: J.I. Kroschwitz (Ed.), Polybenzimidazoles,Encyclopedia of Polymer Science and Engineering,Wiley, NewYork, 1987, 572.

[27] T.S. Chung,A critical reviewofpolybenzimidazoles: historical development andfuture R & D, J. Macromol. Sci. Rev. Macromol. Chem. Phys. C 37 (1997) 277.

[28] M. Jaffe, Pblends, A

[29] N.W. Broabsorbed

[30] T.S. Chunbyhomog221.

[31] PBI PerforS26 solut

[32] T.S. Chunpolyether

[33] J.Z. Ren, T26 DAT hdope comperforma

[34] Y. Li, C. Ca(PES) holseparatio

[35] T. MatsuuPress, Boc

[36] R.X. Liu,polyimidEng. Sci. 6

[37] G.W. Li, Wperformaalcohol) m

[38] H. Zhangof polystySci. 281 (

[39] L.Y. Jiangmembran

[40] J.J. Qin, Ton morpseparatio

[41] T.S. Chun6FDA/6FD(1997) 15

[42] N. WidjoevolutionChem. Re

[43] N. Peng, Tmacrovoi

[44] C. Cao, T.Sin spinninsulfone) (

[45] J.J. Qin, R. Wang, T.S. Chung, Investigation of shear stress effect within a spin-neret on ux, separation and thermomechanical properties of hollow berultraltration membranes, J. Membr. Sci. 175 (2000) 197.

[46] N. Peng, F T., S. Chung, The effects of spinneret dimension and hollow berdimension on gas separation performance of ultra-thin defect-free Torlon

hollow ber membranes, J. Membr. Sci. 310 (2008) 455.[47] T.S. Chung, The limitations of using FloryHuggins equation for the states of

solutions during asymmetric hollow ber formation, J. Membr. Sci. 126 (1997)19.

[48] X.S. Feng, R.Y.M. Huang, Preparation and performance of asymmetricpolyetherimide membranes for isopropanol dehydration by pervaporation, J.Membr. Sci. 109 (1996) 165.

[49] T.S. Chung, Z.L. Xu, Asymmetric hollow ber membranes prepared from mis-cible polybenzimidazole and polyetherimide blends, J. Membr. Sci. 147 (1998)35.

[50] D.F. Li, T.S. Chung, R. Wang, Morphological aspects and structure control ofdual-layer asymmetric hollow ber membranes formed by a simultaneous co-extrusion approach, J. Membr. Sci. 243 (2004) 155.

[51] T.S. Chung, X. Hu, The effect of air-gap distance on the morphology and ther-mal properties of polyethersulfone hollow bers, J. Appl. Polym. Sci. 66 (1997)1067.

[52] X. Qiao, T.S. Chung, K.P. Pramoda, Fabrication and characterization of BTDA-TDI/MDI (P84) co-polyimide membranes for the pervaporation dehydration of

propanol, J. Membr. Sci. 264 (2005) 176.. See-Toh, F.C. Ferreira, A.G. Livingston, The inuence of membraneation parameters on the functional performance of organic solvent

oltrainnaumical. Guo,urface

. Hu, By(acryylene. Hu,rdenitnes fombr. S. Bursbyper. Purif.M. Huter mmbran. ChemJehle,

coombr. S. Namation153 (

. Guo, Cmem281 (

.Guo,Xperva07) 29. Guo,/waterct and. Chenesusinparati96) 24. Chen, E.W. Choe, T.S. Chung, S. Makhija, High performance polymerdv. Polym. Sci. 117 (1994) 297.oks, R.A. Duckett, J. Rose, I.M. Ward, J. Clements, An NMR study ofwater in polybenzimidazole, Polymer 34 (1993) 4038.g, W.F. Guo, Y. Liu, Enhanced matrimid membranes for pervaporationenous blendswith polybenzimidazole (PBI), J.Membr. Sci. 271 (2006)

mance Products, Inc., Solutions brochure of polybenzimidazole (PBI)ion, http://www.pbiproducts.com/les/SolutionsMSDS.pdf.g, S.K. Teoh, X. Hu, Formation of ultra-thin high-performancesulfone hollow ber membranes, J. Membr. Sci. 133 (1997) 161..S. Chung, D.F. Li, R. Wang, Y. Liu, Development of asymmetric 6FDA-ollow ber membranes for CO2/CH4 separation. 1. The inuence ofposition and rheology on membrane morphology and separationnce, J. Membr. Sci. 207 (2002) 227.o, T.S. Chung,K.P. Pramoda, Fabricationofdual-layerpolyethersulfonelow ber membranes with an ultrathin dense selective layer for gasn, J. Membr. Sci. 245 (2004) 53.ra, Synthetic Membranes and Membrane Separation Processes, CRCa Raton, 1994.X. Qiao, T.S. Chung, The development of high performance P84 co-e hollow bers for pervaporation dehydration of isopropanol, Chem.0 (2005) 6674.. Zhang, J.P. Yang, X.P. Wang, Time-dependence of pervaporation

nce for the separation of ethanol/water mixtures through poly(vinylembrane, J. Colloid Interface Sci. 306 (2007) 337.

, H.G. Ni, X.P. Wang, X.B. Wang, W. Zhang, Effect of chemical groupsrene membrane surface on its pervaporation performance, J. Membr.2006) 626., T.S. Chung, -Cyclodextrin containing Matrimid nanocompositees for pervaporation application, J. Membr. Sci. 327 (2009) 216..S. Chung, Effects of orientation relaxation and bore uid chemistryhology and performance of polyethersulfone hollow bers for gasn, J. Membr. Sci. 229 (2004) 1.g, E.R. Kafchinski, The effects of spinning conditions on asymmetricAM polyimide hollow bers for air-separation, J. Appl. Polym. Sci. 6555.jo, T.S. Chung, The thickness and air-gap dependence of macrovoidin phase-inversion asymmetric hollow ber membranes, Ind. Eng.

s. 45 (2006) 7618..S. Chung, K.Y.Wang,Macrovoid evolution and critical factors to formd-free hollow ber membranes, J. Membr. Sci. 318 (2008) 363.. Chung, S.B. Chen, Z.J. Dong, The study of elongation and shear ratesg process and its effect on gas separation performance of poly(etherPES) hollow ber membranes, Chem. Eng. Sci. 59 (2004) 1053.

iso[53] Y.H

formnan

[54] I. PChe

[55] R.Lof s32.

[56] C.Lpoleth

[57] C.LmobraMe

[58] M.CcolSep

[59] R.YwameEng

[60] W.fromMe

[61] Y.SporSci.

[62] R.LtionSci.

[63] R.Lfor(20

[64] R.Lcoleffe

[65] F.Rturpre(19tion membranes, J. Membr. Sci. 299 (2007) 236., B.D. Freeman, Membrane formation and modication, AmericanSociety, Washington, DC, 1999.

X. Fang, H.Wu, Z.Y. Jiang, Preparation and pervaporation performancecrosslinked PVA/PES compositemembrane, J.Membr. Sci. 322 (2008)

. Li, R.L. Guo, H.Wu, Z.Y. Jiang, Pervaporation performance of chitosan-lic acid) polyelectrolyte complex membranes for dehydration ofglycol aqueous solution, Sep. Purif. Technol. 55 (2007) 327.R.L. Guo, B. Li, X.C. Ma, H. Wu, Z.Y. Jiang, Development of novele-lled chitosanpoly(acrylic acid) polyelectrolyte complex mem-r pervaporation dehydration of ethylene glycol aqueous solution, J.ci. 293 (2007) 142.he, S.B. Sawant, J.B. Joshi, V.G. Pangarkar, Dehydration of ethylene gly-vaporationusinghydrophilic IPNsof PVA, PAAandPAAMmembranes,. Technol. 13 (1998) 47.ang, P. Shao, X. Feng, W.A. Anderson, Separation of ethylene glycol-ixtures using sulfonated poly(ether ether ketone) pervaporationes: membrane relaxation and separation performance analysis, Ind.. Res. 41 (2002) 2957.

T. Staneff, B. Wagner, J. Steinwandel, Separation of glycol and waterlant liquids by evaporation, reverse osmosis and pervaporation, J.ci. 102 (1995) 9., Y.M. Lee, Pervaporation of ethylene glycol-water mixtures. I. Perva-performance of surface crosslinked chitosan membranes, J. Membr.1999) 155..L. Hu, F.S. Pan,H.Wu, Z.Y. Jiang, PVA-GPTMS/TEOShybridpervapora-brane for dehydration of ethylene glycol aqueous solution, J. Membr.2006) 454..C.Ma,C.L.Hu, Z.Y. Jiang,Novel PVAsilicananocompositemembrane

porative dehydration of ethylene glycol aqueous solution, Polymer 4839.C.L. Hu, B. Li, Z.Y. Jiang, Pervaporation separation of ethylene gly-mixtures through surface crosslinked PVA membranes: Couplingseparation performance analysis, J. Membr. Sci. 289 (2007) 191.

, H.F. Chen, Pervaporation separation of ethylene glycol-water mix-g crosslinkedPVA-PEScompositemembranes. 1. Effects ofmembraneon conditions on pervaporation performances, J. Membr. Sci. 1097.

Pervaporation dehydration of ethylene glycol through polybenzimidazole (PBI)-based membranes. 1. Membrane fabricationIntroductionExperimentalMaterialsThe fabrication of PBI flat-sheet dense membranesSpinning process and modules fabrication of hollow fiber membranesPervaporation studyMembrane characterization

Results and discussionPervaporation performance of the PBI dense membranePervaporation performance of single-layer PBI hollow fiber membranesPervaporation performance of the dual-layer PBI hollow fiber membranesEffect of heat post-treatment on membrane performanceComparison with previous studies on pervaporation dehydration of ethylene glycol

ConclusionAcknowledgementReferences

pbi licl membrane

Documents