2005-JMS-The role of membrane surface charge and solute physico-chemical properties in the rejection

Theroleofmembranesurfacechargeandsolutephysico-chemicalpropertiesintherejectionoforganicacidsbyNFmembranes

ChristopherBellona,J¨orgE.Drewes∗

EnvironmentalScienceandEngineeringDivision,ColoradoSchoolofMines,Golden,CO80401-1887,USAReceived12November2003;receivedinrevisedform17September2004;accepted24September2004

Availableonline21November2004

Abstract

Theobjectiveofthisstudywastoinvestigatethecontributionofelectrostaticinteractionstotherejectionofselectedorganicacidsbynanofiltrationmembranes.Effectivemembranesurfacecharge(determinedbytwomethodsofzetapotentialmeasurements)oftwocommercialnanofiltrationmembraneswasexaminedatdifferentpHvaluesandfeedwaterchemistries.Sixdifferentorganicacidswereselectedforthisstudyrepresentingtypicalphysico-chemicalpropertiesofemergingorganictracepollutants.Findingsofthisstudyindicatedthattherejectionofnegativelychargedorganicacidswaslargerthanexpectedbasedonstericexclusionandwasprimarilydrivenbythesurfacechargeofthemembraneandcorrelatedwiththedegreeofionizationofthesolute.IncreasingfeedwaterpHresultedinanincreasednegativesurfacecharge,anincreasedpercentageofsolutesinthedeprotonatedstateandanincreasedrejectionthroughelectrostaticrepulsion.Therejectionofthepharmaceuticalresidueibuprofen,anorganicacidwithhydrophobicproperties,wasalsopHdependent,butatpHvaluesbelowthepKaibuprofenadsorbedandpartitionedthroughthemembrane.AtpHabovethepKa,adsorptionofibuprofenwasminimalduetoanincreasedsolubilityandthedominantroleofelectrostaticrepulsion.Thepresenceofcalciuminthefeedwaterloweredtheeffectivemembranesurfacecharge(asdeterminedbyelectrophoreticmobilitymeasurements)ofbothmembranestested,however,rejectionofnegativelychargedorganicsolutesdecreasedonlyformembraneswithamolecularweightcut-offlargerthanthesolutemolecularweight.©2004ElsevierB.V.Allrightsreserved.

Keywords:Nanofiltration;Electrostaticrepulsion;Membranesurfacecharge;Zetapotential;Organictracepollutants

1.Introduction

Apartialrejectionoforganicmicropollutantssuchaspes-ticides,endocrinedisruptingcompounds,orpharmaceuticalresiduesisbecominganimportantconcernduringwaterandwastewatertreatmentapplicationswherenanofiltration(NF)orreverseosmosis(RO)treatmentisemployed.Therefore,understandingthefactorsdrivingtherejectionoforganicmicropollutantsisessentialinordertoassesstreatmentefficienciesapriori.PreviousstudiesontherejectionoforganicsbyNFandultra-lowpressurereverseosmosis(ULPRO)membranesreportedthattheretentionofsolutesdependsuponbothsoluteproperties(i.e.,size,polarityandcharge)andmembraneproperties(i.e.,poresize,chargeand

∗

Correspondingauthor.Tel.:+13032733401;fax:+13032733413.E-mailaddress:jdrewes@mines.edu(J.E.Drewes).

hydrophobicity)[1–8].Recentstudiesontherejectionofor-ganiccompoundsbyROandNFmembraneshavereportedahighrejectionofnegativelychargedorganicsolutesasaresultofelectrostaticinteractionsbetweenmembraneandsolute[1,7,9].Investigations,however,onthemechanismsbehindtherejectionoforganicacids,especiallytheroleofmem-braneandsoluteproperties,islimited.OzakiandHuafang[9]andBergetal.[1]reportedthattherejectionoforganicacidsincreasedwithincreasingpHrelativetothesolutesdisassoci-ationconstant(pKa).NumerousstudiesfocusingontheroleofmembranesurfacechargehavereportedthatmostthinfilmcompositemembraneshaveanegativechargeatneutralpHduetodeprotonatedacidicfunctionalgroupswhichareaddedduringthemanufacturingprocesstoincreasetheselectivityandfluxofULPROandNFmembranes[1,9–15].MembranesurfacechargehasbeenfoundtovarybetweenmembranesandisdependentuponfeedwaterchemistryincludingpH,

0376-7388/$–seefrontmatter©2004ElsevierB.V.Allrightsreserved.doi:10.1016/j.memsci.2004.09.041

228C.Bellona,J.E.Drewes/JournalofMembraneScience249(2005)227–234

Table1

MembranecharacteristicsMembranetypeManufacturerClassifiedasMaterial

MWCO(manufacturer)NaClrejection(%)Testpressure(psi)

RecommendedpHrange(manufacturer)

Membranecharge(pH7)Purewaterpermeability(L/(m2daykPa)at25◦C)

Contactangle(◦)

MorphologydescriptionMeanroughness(nm)Surfaceouterporesize(nm)

Standarddeviationofporesize(nm)Surfaceporosity

NF-90

Dow/FilmtecNF

Polyamide20095704–11Negative2.49

NF-200

Dow/FilmtecNF

Polyamide30060654–11Negative1.2

electrolyteconcentrations,andthepresenceofnaturalor-ganicmatter[12–19].Therejectionofinorganicionssuchassodium,magnesium,perchlorate(ClO4−),andheavymetalsisreportedtoincreaseduringconditionsfavoringamembranesurfacechargethatcanelectrostaticallyrepulsetheionbeingstudied[12,13,15–17].Otherresearchershavereportedthatincreasingtheconcentrationofdivalentcationsinfeedwatercanbothreducethechargeofamembraneandtherejectionofinorganicions[13,16,17].RecentstudiesbyChildressandElimelech[12],Yoonetal.[16,17]andHagmeyerandGim-bel[20]haveattemptedtorelatemembranesurfacechargepropertiestoinorganicionseparationperformance.However,littlehasbeenreportedintermsofmembranesurfacechargeanditsroleintherejectionofnegativelychargedorganicsolutes.

ThescopeofthisstudyistomechanisticallyinvestigatetheeffectofmembranesurfacechargeoftwocommercialNFmembranesontherejectionofselectedorganicacidsrep-resentingatypicalrangeofphysico-chemicalpropertiesofemergingcontaminantssuchaspharmaceuticalresiduesandendocrinedisruptingcompounds.ThesurfacechargeofthetwoNFmembraneswasinvestigatedasafunctionoffeedwaterpHandcalciumconcentrations.Therejectionofsixor-ganicacidsrepresentingdifferentmolecularweights(MW),molecularsizes(molecularwidth,MWd),acidities(i.e.,pKa),andhydrophobicproperties(measuredastheoctanol–waterdistributioncoefficient,Kow)wasstudiedfordifferentfeedwatermatrices.

63.2

Rough,fibrils,pores63.8638281.59E−03

30.3

Smooth,defects5.19

Non-detectableNon-detectableNon-detectable

2.Experimentalapproach2.1.Membranesutilizedinthisstudy

Table1.TheNF-90membranewasdevelopedtoachieveahighremovalofsalts,nitrate,ironandorganiccompoundsfromfeedwater.TheNF-200membraneisclassifiedasasofteningmembranethatcanremoveahighpercentageoforganiccompounds.Allmembraneexperimentswerecarriedoutatafeedpressureof550kPa(80psi)andafeedflowrateof500mL/minresultinginasuperficialfeedchannelvelocityofapproximately11cm/sandarecoveryof9%fortheNF-90and3%fortheNF-200,respectively.TypeIwater(Milli-Qgrade)wasusedtopreparesyntheticwatersolutionsforsinglesoluterejectionexperiments.2.2.Soluteanalysis

Twothinfilmcompositepolyamidemembranes(NF-90andNF-200)wereobtainedfromDow/Filmtec(Midland,MI)andevaluatedinthisinvestigation.Bothmembranesweretestedasflatsheetmembranespecimensinacom-mercialSEPAIIcrossflowfiltrationtestcell(Osmonics,Minnetonka,MN).Priortoexperiments,specimenswerethoroughlyrinsedinTypeIwater(Milli-Qgrade)andstoredat5◦C.Thecrossflowfiltrationtestcellpermittedtestingofasinglemembranecut-out(9.5cm×14.6cm)undertangentialfeedflowconditionswithachannelheightof0.79mm.Allrelevantmembranepropertiesarepresentedin

Table2

Physico-chemicalpropertiesoftargetcompoundsCompound

2,4-Dihydroxybenzoicacid

2-Naphthalenesulfonicacid,sodiumsalt

1,5-Naphthalenedisulfonicacid,disodiumsaltIbuprofenGlutaricacidAceticacid

a

Theorganicacidsolutesevaluatedduringthisstudywerechosentorepresentarangeofmolecularsizeparameters(MWandMWd)aswellasaciddissociationconstant(pKa)valuesinordertoinvestigatetheinfluenceofsizeandchargeonrejection(Table2).Additionally,theorganicacidseval-uatedduringthisstudyhavemolecularsizeandacidicchar-acteristicssimilartomanyemergingtraceorganicsfoundinwaterandwastewater.AHewlett-Packard1040Afluores-cencedetector(PaloAlto,CA)wasemployedtoquantify1,4-dihydroxybenzoicacid,2-naphthalenesulfonicacid,1,5-

MW(g/mol)15420828820613260

˚aMWd(A)5.38

5.467.135.043.163.08

pKa

3.1,9.1,15.6n/an/a4.91

4.34,5.414.76

logKow

1.630.63−3.15

3.72(pH10),−0.4(pH1)−0.29−0.17

EstimatedusingchemicalmolecularmodelingsoftwarepackageHyperchem7.0(Hypercube,Gainesville,FL).

C.Bellona,J.E.Drewes/JournalofMembraneScience249(2005)227–234

Table3

AnalyticalmethodandfeedconcentrationfortargetcompoundsCompound

AnalyticalmethodExcitation

wavelength(nm)Emission

wavelength(nm)229

Targetfeed

concentration(mg/L)2,4-DihydroxybenzoicacidFluorescence3142-Naphthalenesulfonicacid,Fluorescence320sodiumsalt

1,5-NaphthalenedisulfonicFluorescence286acid,disodiumsaltIbuprofenFluorescence225GlutaricacidDOCNAAceticacid

DOC

NA

naphthalenedisulfonicacid,andibuprofen.Thespecificexci-tation/emissionwavelengthsforeachcompoundaresumma-rizedinTable3.GlutaricacidandaceticacidwerequantifiedusingaSievers800TotalOrganicCarbonanalyzer(Ionics,Boulder,CO).Feedwatersolutionswerepreparedinavol-umeof1–3LbyspikingtargetcompoundstoTypeIwater.Allmembranespecimenexperimentswereconductedusingsinglesolutesolutions.2.3.Solutionchemistries

Rejectionexperimentsforbothmembraneswerecon-ductedatdifferentpHvaluesrangingfrom3to9.Analyticalgradehydrochloricacid(HCl)andsodiumhydroxide(NaOH,FisherScientific,Pittsburg,PA)wereusedforpHadjustment.Feedwaterhardnessconcentrationswerepreparedtorepre-sentarangeofdifferentwatertypes(nohardness,moderatelyhard,andhard).Calciumsulfate(CaSO4,FisherScientific)wasaddedtothefeedwatertoachievethedesiredlevelofhardnessatconcentrationsof0mMCa2+,1mMCa2+,and3mMCa2+,respectively.2.4.Zetapotentialmeasurements

TheeffectivesurfacechargeoftheNF-90andNF-200membraneswasdeterminedusingtwozetapotentialmea-surementtechniquesincludingelectrophoreticmobilitymea-surementsandstreamingpotentialmeasurements.

Electrophoreticmobilitymeasurementswereperformedwithanelectrophoreticmeasurementapparatus(ELS-8000Photal,OtsukaElectronics,Japan)withaplatesamplecell.Polystyrenelatexparticles(diameter520nm,OtsukaElec-tronics)coatedwithhydroxypropylcellulose(HPC)andamolecularweightof300,000g/mol(ScientificPolymerProducts,Japan)wereusedasmobility-monitoringparticles.Theseparticlesweredispersedina0.01MNaClsolutiontopreventinteractionswith,oradsorptionontothequartzcellsurfaceduringmeasurement[18].Zetapotentialmea-surementswereconductedwithmembranespecimens2+inabackgroundsolutionof10mMNaCland3mMCaandpHrangingfrom3to11.

Surfacestreamingpotentialmeasurementsweredeter-minedbyastreamingpotentialanalyzer(EKA,Brookhaven

3901.5–1.83701.5–1.83901.5–1.82751.5–1.8NA8.8–13.2NA

10–15

InstrumentsCorp.,NewYork)followingtheprocedurede-scribedbyChildressandElimilech[21].Streamingpotentialmeasurementswereperformedusingasolutionof10mMKClatpHrangingfrom3to11.

3.Resultsanddiscussion

3.1.Membranezetapotentialmeasurements

ResultsfromtheelectrophoreticmobilitysurfacechargemeasurementsandstreamingpotentialsurfacechargemeasurementsfortheNF-90andNF-200atdifferentpHand10mMNaCl(electrophoretic)andKCl(streaming)areshowninFig.1.Fromelectrophoreticmobilitymeasure-mentsconductedinthepresenceof10mMNaCl,theNF-200membraneexhibitsanisoelectricpointclosetopH3,whiletheNF-90doesnotdisplayanisoelectricpointinthepHrangeinvestigated(pH3–9).Resultsobtainedduringstream-ingpotentialexperiments,however,indicatethattheNF-90hasasignificantlylessnegativeeffectivesurfacechargeandanisoelectricpointclosetopH4.Streamingpotentialmea-surementsoftheNF-200indicateaneffectivesurfacechargethatislessnegativethanduringelectrophoreticmobilitymeasurementsandanisoelectricpointclosertopH4.Recentstudieshavedeterminedthatzetapotentialvaluesobtainedusingelectrophoreticmobilitymeasurementsandstreamingpotentialmeasurementscanbesignificantlydifferentwhen

Fig.1.ZetapotentialmeasurementsfortheNF-90andNF-200performedbyelectrophoreticmobility(EM)measurementsat10mMNaClandstreamingpotential(SP)measurementsat10mMKCl.

230C.Bellona,J.E.Drewes/JournalofMembraneScience249(2005)227–234

measuringNFandROmembranespecimens[18,22].Ingen-eral,effectivesurfacechargevaluesobtainedwithstreamingpotentialmeasurementsareoftenlessnegativethaneffectivesurfacechargevaluesobtainedwithelectrophoreticmobilitymeasurements[18,22].Itistheorizedthatmembranesurfacemorphologycaninfluencethedistanceoftheshearplaneduringstreamingpotentialmeasurements[23]andcouldinteractwiththemobilityofparticlesduringelectrophoreticmobilitymeasurements[18].Bowenetal.[23]notedthatstreamingpotentialmeasurementsaremostaccurateforsmoothflatsurfaces.TheeffectivesurfacechargeresultsfortheNF-200showsmallerdifferencesamongthetwomethodsemployedthanfortheNF-90.ItistheorizedthatduetothesmoothsurfaceoftheNF-200(Table1),bothmethodsemployeddisplayedsimilarisoelectricpointsalthoughstreamingpotentialmeasurementsshowedamuchsmallernegativesurfacechargeathigherpHvalues.TheroughsurfacemorphologyoftheNF-90couldhelpexplainwhymeasurementsbetweenthetwomethodsindicatedsignificantlydifferentsurfacechargecharacteristics.Bothmethods,however,indicatethattheNF-90membraneismorenegativelychargedthantheNF-200atpHgreaterthan5inthepresenceof10mMNaClorKCl.

Recentstudiesinvestigatingtheinfluenceofionicstrengthonmembranesurfacechargehavefoundthatresultsobtainedwithelectrophoreticmobilitymeasurementsagreewiththeelectricaldoublelayercompactiontheory[18,22]and,ingeneral,themeasuredsurfacechargebecamelessnegativewithincreasingionicstrength.Otherstudies,usingsurfacestreamingmeasurements,havefoundthatsurfacechargeval-uespassthroughamaximumwithincreasingionicstrength[18,22,24,25].Therefore,fortheinvestigationoftheinflu-enceofcalciumadditionsontheeffectivesurfacechargeofthemembranesemployedinthisstudy,electrophoreticmo-bilitymeasurementswereusedtoquantifyzetapotentialval-ues.FortheNF-90,thepresenceofdivalentcations(3mMCa2+)resultedinareductionoftheeffectivemembranesur-facechargebyapproximately20–25%atallpHvaluesstud-iedwhenemployingelectrophoreticmobilitymeasurements(Fig.2).Thisreductionofthezetapotentialinthepresenceof

Fig.2.ElectrophoreticmobilityzetapotentialmeasurementsoftheNF-90andNF-200withabackgroundelectrolytesolution(10mMNaCland3mMCa2+).

Fig.3.Rejectionof2-naphthalenesulfonicacidand1,5-naphthalenedisulfonicacidbyNF-90andNF-200vs.feedwaterpH(electrophoreticmobilityzetapotentialofNF-90andNF-200asdeterminedat10mMNaClshownasdottedlines).

CaSO4wasalsoobservedfortheNF-200membrane.Numer-ousstudieshavereportedthatpositivelychargedionssuchassodium,calcium,magnesium,andcationicsurfactantscanbindtothenegativelychargedmembranesurfaceresultinginareducednegativesurfacecharge[11,12,14,15,19,25].3.2.SelectedorganicacidrejectionatvaryingfeedwaterpHvalues

Rejectionexperimentswereconductedwithtwonega-tivelychargedorganicacids(2-naphthalenesulfonicacidand1,4-dinaphthalenesulfonicacid)atdifferentpHwiththeNF-90andNF-200membranes.ResultsoftheseexperimentsarepresentedinFig.3alongwithelectrophoreticmobilitysurfacechargemeasurements.SincethepKavaluesofthetwocompoundsarewellbelowthepHrangeinvestigated(i.e.,compoundsarenegativelychargedatpH3–9),thein-creaseinrejectionobservedatapHbetween3and5forbothmembranescanlikelybeattributedtoanincreaseinthenegativeeffectivemembranesurfacecharge(asquan-tifiedbybothzetapotentialmeasurements)resultinginanincreaseddegreeofelectrostaticrepulsion.Dependingonwhichsurfacechargemeasurementtechniqueisconsidered,atlowpH,thereducedsurfacechargeoftheNF-90andNF-200membraneappearstolimittheamountofelectrostaticrepulsionbetweenthemembraneandsoluteandsievingef-fectsduetothesizeofthecompoundalsoappearimportantsince1,5-dinaphthalenesulfonicacidisrejectedtoanhigherdegreethan2-naphthalenesulfonicacid.Therelativelylowrejectionobservedfor1,5-dinaphthalenesulfonicacidand2-naphthalenesulfonicacidbybothmembranesatpH3,also,seemstobedrivenprimarilybylowelectrostaticrepulsionre-sultinginimprovedmasstransferduetotheslightlypositivemembranesurfacecharge,especiallyifstreamingpotentialmeasurementsfortheNF-90aretakenintoaccount[7].AtpH

C.Bellona,J.E.Drewes/JournalofMembraneScience249(2005)227–234231

Fig.4.Speciationof1,4-dihydroxybenzoicacidandrejectionbyNF-90andNF-200vs.feedwaterpH.

valuesof5,7and9,therejectionof2-naphthalenesulfonicacidand1,5-dinaphthalenesulfonicacidremainedatabout90%forbothmembranestestedalthoughthenegativesur-facechargecontinuestoincreaseforbothmembranesinthispHrange.VanderBruggenetal.[7]concludedthattherejec-tionofnegativelychargedorganicsoluteswithamolecularsizeclosetotheporesizeofaNFmembraneismoredrivenbysievingthanelectrostaticrepulsion.Researchers[11,19,26]havereportedthatincreasingthenegativesurfacechargeofamembranecanincreasetheMWCOasaresultofelectrostaticrepulsionwithinthemembraneporesormembraneswelling.Fregeretal.[26]foundthatamaximumintherejectionoflactatewasfoundatneutralpH,sincetheincreaseofchargerepulsionathigherpHwascancelledbyadecreasedsiev-ingeffectthroughmembraneswellingasthepHincreased.Itishypothesizedthatforthetwomembranestestedinthisstudy,theincreaseinpermeabilityasaresultofincreasedsurfaceelectronegativitymayoffsettheexpectedincreaseinelectrostaticrepulsionbetweenthemembraneandsolute.AdditionalexperimentswereconductedtoinvestigatetheeffectofpHonthespeciationandrejectionofaceticacid,glutaricacid,and1,2-dihydroxybenzoicacid.TheseorganicacidsarecharacterizedbypKavalueswithinthepHrangestudiedandtheirspeciationasafunctionofpHisillustratedinFigs.4–6.TheeffectofincreasingfeedwaterpHfrom3to7onrejectionoftheorganicacidsofinterestresulted

Fig.5.SpeciationofaceticacidandrejectionbyNF-90andNF-200vs.feedwaterpH.

Fig.6.SpeciationofglutaricacidandrejectionbyNF-90andNF-200vs.feedwaterpH.

inasignificantincreaseinrejection,whichcloselyfollowsthepercentageofthedeprotonatedspeciesforeachofthecompounds.OzakiandHuafang[9]andBergetal.[1]re-portedasimilareffectforaceticacidandmecoprop,respec-tively,whererejectionofthesesolutesincreasedasthepHapproachedthepKa.Thegreatestchangeintherejectionofthethreeorganicacidsoccurredasthemonoproticnegativelychargedspeciesbecamedominant(>60%).Foraceticacid,thegreatestchangeinrejectionoccurredbetweenpHof5and7fortheNF-200andbetween4and7fortheNF-90.Forglutaricacidand1,2-dihydroxybenzoicacid,thegreat-estchangeinrejectionoccurredbetweenpH3and5.ThelargeincreaseobservedintherejectionoftheorganicacidsbetweenpH3and7islikelycausedbyacombinationofthesolutesbecomingmoredeprotonatedandthemembranechargebecomingincreasinglynegative.AtlowpHvalues(3–6),however,therejectionofthesecompoundsissignifi-cantlylowerbytheNF-200thantheNF-90.TheNF-90re-jectsthesecompoundsathigherlevelsthantheNF-200duetothesmallerMWCO/poresizeoftheNF-90andthegreaternegativechargeatpHvaluesgreaterthan4.TherejectionofionizablecompoundsbytheNF-90reachesanapproximatemaximumoncefullyionized,whiletheapproximatemaxi-mumrejectionbytheNF-200occursaroundpH7.FortheNF-90,nofurtherincreaseofrejectionoccurredaboveapHof5for2-naphthalenesulfonicacid,1,5-dinaphthalenesulfonicacid,and1,2-dihydroxybenzoicacid.Foraceticacid,glu-taricacid,and1,2-dihydroxybenzoicacid,rejectionbyboththeNF-90andtheNF-200remainedrelativelyconstantinapHrangebetween7and9.Additionally,therejectionofallcompoundsisapproximatelythesamebetweentheNF-90andNF-200atapHof9evenwithsmalldifferencesinef-fectivemembranesurfacechargeandMWCOvalues.Basedontheseexperiments,itappearsthatanincreasinglynegativesurfacechargecanonlyrejectnegativelychargedsolutestoacertainlevelbeforetheeffectisoffsetbyporeexpansionormembraneswelling.ItshouldbenotedthatalthoughthetwomembranesstudiedhavedifferentMWCOvalues,theNF-90andNF-200havesimilarzetapotentialvaluesatpH7and9,whichmightexplainwhytherejectionofthesecompoundsremainedverysimilarforthetwomembranes.

232C.Bellona,J.E.Drewes/JournalofMembraneScience249(2005)227–234

Fig.7.SpeciationofibuprofenandrejectionbyNF-90andNF-200vs.feedwaterpHatpH7.

3.3.RejectionofibuprofenatvaryingfeedwaterpHvalues

Ibuprofenisapharmaceuticalresiduecommonlyoccur-ringinwaterofimpairedquality.IbuprofenhashydrophobicpropertiesbelowitspKavalueof4.91.Speciationandre-jectionofibuprofenbyNF-90andNF-200areillustratedinFig.7asafunctionoffeedwaterpH.Incontrasttorejec-tionresultsreportedfortheotherorganicacids,thehighestrejectionofibuprofenoccurredatpH3.Onceibuprofenbe-comesincreasinglynegativelychargedwithincreasingpH,therejectiondecreasedslightlyatpH5beforeitincreasedagainbeyondapHof7.Sch¨aferandWaite[27]reportedasimilarfindingfortherejectionofthesteroidestronebycer-tainNFandROmembranes,whererejectiondecreasedasthepHapproachedthepKaofestrone.Itwashypothesizedthattheinitialseparationmechanismwasadsorptionofestroneontothemembrane,whereasathigherpHvalueselectrostaticrepulsionbetweenthemembraneandsolutedecreasedtheadsorptioncapability.TheroleofadsorptionofhydrophobicibuprofenwasexaminedthrougharejectionexperimentthatwasperformedwiththeNF-200atpH3forapproximately50h.Inthefirsthouroftheexperiment,thepermeateconcen-trationwaslow(approximately0.4mg/L)andtherejectionhigh(above90%)(Fig.8).Within10h,however,aconsid-erableamountofmasswaslostfromthefeed,apparently

Fig.8.FeedandpermeateconcentrationsandrejectionofibuprofenbyNF-200atpH3.

Fig.9.FeedandpermeateconcentrationandrejectionofibuprofenforNF-200.

adsorbedtothemembrane,andtherejectiondecreasedtoap-proximately30%.Kimuraetal.[28]reportedasimilartrendfornaphthalenerejectionbyROandNFmembranes.Kimuraetal.[28]andBoussaheletal.[19]hypothesizedthatalthoughadsorptioncanresultininitialrejection,theadsorbedsolutecanpartitionanddiffuseacrossamembraneandreducere-jectionconsiderablythroughpartitioningintothepermeateduringlong-termoperation.Afterthefeedandpermeatecon-centrationstabilized,thefeedwaterpHwasadjustedto7after550minofoperation.Within40minfollowingthepHadjustment,thepermeateconcentrationequaledthatofthefeedconcentration.TheexperimentwasrepeatedatpH7forapproximately425minwithanewmembranespecimen(Fig.9).Therejectionandsoluteconcentrationsobservedinthepermeateandconcentrateremainedconstantthroughouttheexperimentindicatingnolosstoadsorptionandpointingtoelectrostaticrejectionasthemainremovalmechanism.Foribuprofen,anorganicacidwithhydrophobicproperties,thesolutionchemistryofthefeedwaterdeterminesthemech-anismofrejection.AtfeedwaterpHvaluesbelowthepKa,ibuprofenispredominatelyremovedbyadsorptionandabovethepKaitisremovedbyelectrostaticrepulsion.3.4.EffectofhardnessonrejectionoforganicacidsRecentstudieshavereportedthatincreasedfeedwaterionstrength,especiallyintheformofdivalentcations(Ca2+andMg2+)candecreasethemembranesurfacechargeandsubsequentlyresultinareducedrejectionofinorganicions[13,16,17].Rejectionexperimentswithtargetorganicacidswererepeatedinthepresenceof1mMcalciumand3mMcalciumaddedintheformofCaSO4tothefeedwater.Oftheorganicacidstested,onlytherejectionofaceticacidbyNF-90wasreducedbycalciumaddition(Fig.10).Calciumadditionreducedtherejectionofaceticacid,DHBandNSAbytheNF-200(Fig.11).Boussaheletal.[19]reportedthatmembraneswithlargerpores,liketheNF-200,areaffectedmorebyin-organicionsthantightermembranes.SinceelectrophoreticmobilityzetapotentialmeasurementsoftheNF-90andNF-200showedthatthemembranesbecomelessnegativeinthepresenceofcalciumions(Fig.2),thedecreaseinrejection

C.Bellona,J.E.Drewes/JournalofMembraneScience249(2005)227–234233

Fig.10.InfluenceofcalciumionsontherejectionoforganicacidsbyNF-90atpH7.

Fig.11.InfluenceofcalciumionsontherejectionoforganicacidsbytheNF-200atpH7.

bytheNF-200foraceticacid,1,4-dihydroxybenzoicacid,and2-naphthalenesulfonicacidcanbeexplainedbyade-creaseinelectrostaticinteractionbetweenmembraneandso-lute.Itishypothesizedthat,chargedorganiccompoundswithMWclosetotheMWCOofamembranearelessaffectedbydecreasedelectrostaticinteractionssincestericexclusionalsoplaysadominantroleintherejectionofthesecom-pounds.TheremovalofchargedorganiccompoundswithaMWsmallerthantheMWCOofamembrane,however,canbeaffectedbythepresenceofcalciumionsandthereducednegativesurfacechargeofamembrane.

4.Conclusions

TherejectionofnegativelychargedorganicacidsbyNFmembranesresultedinalargerrejectionthanexpectedbasedonsteric/sizeexclusionsduetoelectrostaticrepulsionbe-tweensoluteandmembraneasdrivingfactorforrejection.Thedegreeofrejectionofthecompoundstesteddependeduponthesurfacechargeofamembrane,thedegreeofdeproto-nationofthecompound,andthepresenceofdivalentcations.Discrepancieswereobservedbetweeneffectivemembrane

surfacechargevalueswhenmeasuredbyelectrophoreticmo-bilitymeasurementsandstreamingpotentialmeasurements.IncreasingfeedwaterpHresultedinanincreasednegativesurfacechargeofthemembrane,anincreasedpercentageofsolutesinthedeprotonatedstate,andanincreasedrejectionthroughelectrostaticrepulsion.AlthoughtheNF-90andNF-200havedifferentMWCOvalues,similarsurfacechargevaluesledtosimilarorganicacidrejectionsathigherfeedwaterpHvalues.Ibuprofen,anorganicacidwithhydropho-bicproperties,wasfoundtoadsorbtothemembranestestedatfeedwaterpHvaluesbelowthepKa.Initialrejectionduetoadsorptionoverestimatedtherejectionofibuprofensincepartitioningoccurredatlongeroperatingtimes.However,atfeedwaterpHvaluesabovethepKaofibuprofen,adsorptionwasnotdominantandrejectionremainedconstantovertheexperimentandcouldbeconsideredelectrostaticinnature.TheadditionofcalciumionstothefeedwaterdecreasedtherejectionofsoluteswithaMWsignificantlysmallerthantheMWCOofthemembrane.

Acknowledgements

TheauthorsthanktheWateReuseFoundation(WRF)foritsfinancial,technical,andadministrativeassistanceinfund-ingandmanagingtheprojectthroughwhichthisinformationwasdiscovered.ThecommentsandviewsdetailedhereinmaynotnecessarilyreflecttheviewsoftheWateReuseFoun-dation,itsofficers,directors,affiliatesoragents.Theau-thorsarealsogratefultoCentral/WestBasinMunicipalWa-terDistrictforitsfinancialandtechnicalsupport.WethankDow/Filmtecforprovidingmembranespecimens.Theau-thorsalsothankDr.JaewonCho,K-JIST(SouthKorea),Dr.GaryAmyandTae-UkKim,UniversityofColorado,aswellasDr.PeiXu,ColoradoSchoolofMinesfortheirsupport.

References

[1]P.Berg,G.Hagmeyer,R.Gimbel,Removalofpesticidesandother

micro-pollutantsbynanofiltration,Desalination113(1997)205–208.[2]Y.Kiso,T.Kon,T.Kitao,K.Nishimura,Rejectionpropertiesof

alkylphthalateswithnanofiltrationmembranes,J.Membr.Sci.182(2001)205–214.

[3]Y.Kiso,Y.Sugiura,T.Kitao,K.Nishimura,Effectsofhydropho-bicityandmolecularsizeonrejectionofaromaticpesticideswithnanofiltrationmembranes,J.Membr.Sci.192(2001)1–10.

[4]Y.Kiso,Y.Nishimura,T.Kitao,K.Nishimura,Rejectionproperties

ofnon-phenylicpesticideswithnanofiltrationmembranes,J.Membr.Sci.171(2000)229–237.

[5]Y.Kiso,T.Kitao,J.Kiyokatsu,M.Miyagi,Theeffectsofmolecu-larwidthonpermeationoforganicsolutethroughcelluloseacetatereverseosmosismembrane,J.Membr.Sci.74(1992)95–103.

[6]K.Kosutic,B.Kunst,Removaloforganicsfromaqueoussolutions

bycommercialROandNFmembranesofcharacterizedporosities,Desalination142(2002)47–56.

[7]B.VanderBruggen,J.Schaep,D.Wilms,C.Vandecasteele,Influ-enceofmolecularsize,polarityandchargeontheretentionofor-ganicmoleculesbynanofiltration,J.Membr.Sci.156(1999)29–41.

234C.Bellona,J.E.Drewes/JournalofMembraneScience249(2005)227–234

[8]B.VanderBruggen,J.Schaep,W.Maes,D.Wilms,C.Vande-casteele,nanofiltrationasatreatmentmethodfortheremovalofpesticidesfromgroundwaters,Desalination117(1998)139.

[9]H.Ozaki,Li.Huafang,Rejectionoforganiccompoundsbyultra-lowpressurereverseosmosismembrane,WaterRes.36(1)(2002)123–130.

[10]M.J.Ariza,A.Canas,J.Malfeito,J.Benavente,EffectofpHonelec-trokineticandelectrochemicalparametersofbothsub-layersofcom-positepolyamide/polysulfonemembranes,Desalination148(2002)377–382.

[11]A.Braghetta,F.A.Digiano,W.P.Ball,Nanofiltrationofnaturalor-ganicmatter:pHandionicstrengtheffects,J.Environ.Eng.123(7)(1997)628–0.

[12]A.E.Childress,M.Elimelech,Relatingnanofiltrationmembrane

charge(electrokinetic)characteristics,Environ.Sci.Technol.34(2000)3710–3716.

[13]H.Ozaki,K.Sharma,W.Saktaywin,Performanceofanultra-low

pressurereverseosmosismembrane(ULPROM)forseparatingheavymetal:effectsofinterferenceparameters,Desalination144(2002)287–294.

[14]S.S.Deshmukh,A.E.Childress,ZetapotentialofcommercialRO

membranes:influenceofsourcewatertypeandchemistry,Desalina-tion140(2001)87–95.

[15]J.Tanninen,M.Nystrom,Separationofionsinacidicconditions

usingNF,Desalination147(2002)295–299.

[16]Y.Yoon,G.Amy,J.Cho,N.Her,J.Pellegrino,Transportofper-chlorate(ClO4−)throughNFandUFmembranes,Desalination147(2002)11–17.

[17]J.Yoon,G.Amy,Y.Yoon,P.Brandhuber,J.Pellegrino,Rejec-tionoftargetanions,hexavalentchromium(CrO42−),perchlorate(ClO4−),andarsenate(H2AsO4−/HAsO42−),bynegativelychargedmembranes,in:ProceedingsoftheAmericanWaterWorksAssoci-ationMembraneTechnologyConference,Atlanta,GA,2003.

[18]Y.Shim,H.-G.Lee,S.Lee,S.-H.Moon,J.Cho,EffectsofNOMand

ionicspeciesonmembranesurfacecharge,Environ.Sci.Technol.36(2002)38–3871.

[19]R.Boussahel,A.Montiel,M.Baudu,Effectsoforganicandinor-ganicmatteronpesticiderejectionbynanofiltration,Desalination145(2002)109–114.

[20]G.Hagmeyer,R.Gimbel,Modelingtherejectionofnanofiltration

membranesusingzetapotentialmeasurements,Sep.Purif.Technol.15(1999)19–30.

[21]A.E.Childress,M.Elimilech,Effectofsolutionchemistryonthe

surfacechargeofpolymericreverseosmosisandnanofiltrationmem-branes,J.Membr.Sci.119(1996)253–268.

[22]L.Ricq,A.Pierre,J.-C.Reggiani,J.Pagetti,A.Foissy,Useof

electrophoreticmobilityandstreamingpotentialmeasurementstocharacterizeelectrokineticpropertiesofultrafiltrationandmicrofil-trationmembranes,ColloidsSurf.A:Physicochem.Eng.Aspects138(1998)301–308.

[23]W.R.Bowen,T.A.Doneva,J.Austin,G.Stoton,Theuseofatomic

forcemicroscopytoquantifymembranesurfaceelectricalproperties,ColloidsSurf.A:Physicochem.Eng.Aspects201(2002)73–83.[24]J.M.M.Peeters,M.H.V.Mulder,H.Strathmann,Streamingpotential

measurementsasacharacterizationmethodfornanofiltrationmem-branes,ColloidsSurf.A:Physicochem.Eng.Aspects150(1999)247–259.

[25]M.Elimilech,A.E.Childress,Zetapotentialofreverseosmosismem-branes:implicationsformembraneperformance,WaterTreatmentTechnologyProgramReportNo.10,U.S.BureauofReclamationFinalReport,1996.

[26]V.Freger,T.C.Arnot,J.A.Howell,Separationofconcentratedor-ganic/inorganicsaltmixturesbynanofiltration,J.Membr.Sci.178(2000)185–193.[27]A.I.Sch¨afer,T.D.Waite,Tracecontaminantremovalusinghybrid

processesinwaterrecycling,in:H.H.Hahn,E.Hoffman,H.Ode-gaard(Eds.),ChemicalWaterandWastewaterTreatmentVII,IWAPublishing,London,2002,pp.319–330.

[28]K.Kimura,G.L.Amy,J.E.Drewes,Y.Watanabe,Adsorptionof

hydrophobiccompoundsontoNF/ROmembranes—anartifactlead-ingtooverestimationofrejection,J.Membr.Sci.1–2(2003)–101.