The preparation of highly porous structures from filamentary nickel

Thepreparationofhighlyporousstructuresfrom

filamentarynickelpowders

A.Y.Zaitseva,∗,D.S.Wilkinsona,G.C.Weatherlya,†,T.F.Stephensonb

b

McMasterUniversity,1280MainStreetWest,Hamilton,Ont.,CanadaL8S4L7INCOTechnicalServices,2060FlavelleBoulevard,Mississauga,Ont.,CanadaL5K1Z9

Received10March2003;accepted30March2003

a

Abstract

Porousnickelstructuresareinhighdemandforbattery,catalystandfiltermaterialsapplications.Traditionallysuchstructuresaremadebysinteringfinefilamentarynickelpowders.However,thestrengthofsuchstructuresisratherlow,whencompared,forexample,withNifoamsofsimilardensity.Inthisresearchwehaveappliedcolloidalprocessingtechniquestoimprovethepowderparticledistributionand,hencethestrengthofthefinalsinteredstructure.BydispersingNipowderinwaterpriortointroducingabinder,betterseparationofparticlesandbreak-upofconglomeratesisachieved.Theadditionofdispersantsfurtherimprovestheparticledistribution.Thenewmethodalsoenhancescontroloftheslurryviscosityandthegreendensityofthenickelporousbody.Thestructureofbatteryplaquepreparedaccordingtothenewtechniqueisthereforemoreuniformthanthestructureoftheconventionalplaque.Moreover,thetensilestrengthofaplaquepreparedbythenewmethodisincreasedby50–70%.©2003ElsevierB.V.Allrightsreserved.

Keywords:Porousnickel;Sintering;Batteries;Positiveelectrode

1.Introduction

Nickelpowdershavingahighlyirregularfilamentaryshapecanbefabricatedandsinteredtoproducehighlyporousstructures,whichareinhighdemandforbattery,catalystandfiltermaterialsapplications.Inadditiontohighporositythesematerialsofferexcellentcorrosionresistanceandgoodelectricalconductivity.Oneofthemostimportantrequirementsforthesetypesofstructuresisthecombinationofhighstrengthwhilemaintainingalevelofhighporos-ity,especiallyforthecaseofrechargeablebatteries.Thecharge-carryingcapacityofbatteriesisstronglydependentontheamountofactivemassimpregnatedintheporousplaquewhilebatterylifetimeissensitivetothestrengthoftheporousstructure.Thismustbesufficienttowithstandrepeatedswellingandshrinkingduringbatterychargingandrechargingcycleswithoutfailure[1].

Thetraditionalapproachtothemanufactureofsuchstruc-turesisbysinteringoffinefilamentarynickelpowders.Thedominanttechnologyincommercialproductioninvolvesahighlyautomatedandeconomicalslurrycoatingprocess

Correspondingauthor.†Deceased.

E-mailaddress:azaitsev@inco.com(A.Y.Zaitsev).

0378-7753/$–seefrontmatter©2003ElsevierB.V.Allrightsreserved.doi:10.1016/S0378-7753(03)00534-2

∗

ontoaperforatedsubstrate.Thoughdetailedinformationonthetechnicalaspectsofthisprocessislimiteddueitspro-prietarynature,itisknownthatelectrodeplaqueproductionprocessbasedontheuseoftheIncoType255filamentarynickelpowderinvolvesseveralmainsteps[2]:

•preparationofaslurrybymixingthenickelpowderinanaqueouscellulosesolution,

•coatingofaperforatedmetalsubstrate(typicallymadeofnickel-coatedsteel)withtheslurryusingaverticaltapecastingprocess(twindoctorbladesareusedtocontrolthecoatingthickness),

•dryingtheslurryinanovenattemperaturesbelow100◦C,and

•sinteringinareducingatmosphereatatemperatureofabout1000◦C.AnexampleofthesinterednickelplaqueproducedbysuchaprocessisgiveninFig.1.Theplaquewasproducedbyacommercialvendorandisthoughttobetypicalofthoseemployedinindustrialpractice.Averysimilarmicrostruc-tureisobservedinplaquesproducedinourlaboratorybytheprocessjustdescribed.

Thestructureofthisplaqueincludeslargevoids(uptoabout100␮mindiameter).Thereareseveraldisadvantagesinproducingsuchanon-uniformstructure,including:

254A.Y.Zaitsevetal./JournalofPowerSources123(2003)253–260

Fig.1.Microstructureofasinteredindustrialplaque(×125).

•decreasedbatteryefficiencyduetodifficultyindischarg-ingandrechargingtheactivemassfarfromtheconduct-ingphase,and

•decreasedstrengthduetounevenstressconcentrations.Hence,theperformanceofabattery(orfilter)preparedus-ingthisproceduremaybesignificantlydegradedcomparedtoitsfullpotential.

2.Improvingplaquemicrostructure

Ananswertotheproblemindicatedaboveisproposedinthispaper.Thesolutioninvolvesimprovementsinparticledistributionthatprovideamoreevenandhence,strongerstructure,withoutsacrificingthedesiredoverallporosityofabout80%.Alteringthedetailsofthesinteringprocess(e.g.temperature,duration,sinteringatmosphere)wasfoundtohaveinsignificantimpactonthehomogeneityofthefinalmicrostructure.Wehavetherefore,focusedondevelopinganewapproachtoslurrypreparationusinganadaptationofcolloidalprocessingtechniquestopromoteauniform,dis-persedpowderdistribution.Incolloidalprocessingoneusessurface-activechemicalsasdispersantagentsinanappropri-atesolvent(inthiscaseallprocessingwasaqueous-based)alongwithefficientmixing.Dispersantsbecomeattachedtothesurfaceoftheparticlesleadingtotheemergenceofrepulsiveforcesbetweenthem.Asuggestedapproachtoprocessingisoftenusedinceramicsforwhichtheaverageparticlesizeisaboutamicronorless[3].Itwasnotclearhowwellsuchanapproachmightworkforfilamentarynickelpowdersforwhichthedispersantmustovercomethehighermaterialdensity,largeparticlesize(typicalparticlelengthisabout30␮m),andirregularshape.Despitethesemisgivingsitwasfeltthatthisapproachofferedthepoten-tialofsignificantimprovementinstrengthandhomogeneityofsinterednickelelectrodes.

AflowchartfortheinitialstagesoftheconventionalelectrodeplaquepreparationprocessisgiveninFig.2.

Fig.2.Conventionalmethodofplaquepreparation.

Inthefirststepnickelpowderismixeddirectlywiththecellulose–watersolution.Celluloseinthisapplicationplaystheroleofabinderbetweenparticles.Theadditionofcel-luloseisnecessaryinordertoproduceslurryofsufficientviscositytoenabletheverticaltapecastingprocesstopro-ceed.Duetothefilamentarystructureofthesenickelpow-dersandrelativelyhighviscosityofthecellulosesolutiontheparticlesmaybeneverwelldispersed.Moreover,whileintensivemixingcanhelpdisperseparticlesreasonablywell,highshearstressmaycauseafilamentbreakage.Thebreak-ageofthefilamentsleadstothereductionofthesinteredmaterialporosity.Intheprocessofthestudyitwasfoundthatthesimpleadditionofdispersantagentsintothesolu-tionduringmixingdoesnotleadtoimprovementsinthestructure,probablyduetothecompetitiveabsorptionofthecellulosecontainedinthestartingsolution.Thus,thedisper-santisunabletobecomeattachedtothesurfaceanddeveloprepulsiveforcesbetweenparticles.Suchphenomenonisawellknownintheceramicprocessing[4].

Wehavethereforedevelopedanewprocessingsequence(Fig.3)ensuringthatthepowderiswelldispersedandthedispersantisattachedtotheparticlesurfacebeforethebinderphaseisadded.Moreover,inthenewprocess,theinitialmixingoperationcanbeperformedusinghigh-speedstirringwithoutfilamentsbreakageduetothelargeexcessofwaterpresent.

Theproposedmethodofslurrypreparationshouldnotonlyallowgoodcontactbetweendispersantandpowderbutalsoshouldenablecontroloftheslurryrheologicalproper-tiesoverawiderangeofdensitiesbyadjustingboththecon-centrationofcelluloseinthesolutionaswellastheamountofwaterused.

3.Experimental

Awiderangeofchemicalsexiststhatmightbesuitableforuseasdispersingagentstoimprovenickelpowderdis-persion.TheefficiencyofsomeofthesewasdeterminedbymeasuringtheZ-potentialofIncoType255nickelpowderinaqueousmedia.Theconcentrationsofdispersantswere

A.Y.Zaitsevetal./JournalofPowerSources123(2003)253–260

255

Fig.3.Improvedmethodofplaquepreparation.

variedfrom0to1%(byweight)basedontheamountofnickel.

Dispersantsusedinthisstudywerechosenfromthosecommonlyusedintheceramicsandpaintindustry.TheseincludedseveralfromtheDarvanseries(DarvanC,Darvan821A,Darvan811,providedbyR.T.VanderbittCompany)andthePolacrylseries(A40-43N,A70-40NandB55-50N,providedbyPolacrylcompany).Allofthesearewater-basedandcontaindifferentformsofpolyacrylateammoniumsalts.NickelplaqueswerepreparedaccordingtotheflowchartsillustratedinFigs.2and3.Inthelattercase,dispersantconcentrationsselectedonthebasisoftheZ-potentialmea-surementswereused.Nickelpowderwasgraduallymixedwithwatercontainingaknownconcentrationofdispersant,inordertoobtaingoodcontactbetweenthemoleculesofdispersantandnickelparticles.Thequantityofwaterwastwotothreetimesthatwhichwasnecessarytoproducefinalslurryoftherequiredviscosity.Afterstirring,thepowderwasallowedtosettleandtheexcesswaterwasdecantedtoachievethedesiredslurrycomposition,towhichthecellu-losesolutionwasadded.Table1showsthecompositionsofslurriesusedintheseexperiments(dispersantnotincluded).Whilecommercialbatteryplaquesalwayscontainametalsubstrate,wewishedtomanufactureplaqueswithoutsub-stratessothatthestrengthofthesinterednickelbodiescouldbemeasureddirectly.Therefore,theslurrywascastintoagreentapewithoutasubstrateusinghorizontaltapecastingmachine.Aftertapecastingsamplesweredriedatroomtem-peratureandsinteredbetween750and1050◦Cfor10mininareducingatmosphereof15%H2and85%N2.Plaques

Table1

Slurrycomposition(per100g)Component

Ni(g)H2O(g)

Cellulose(g)

OriginalprocessNi

42.5

Cellulosesolution(3.7%)55.4

2.1Total

42.555.4

2.1

NewprocessNi42.5

H2O

39.7Cellulosesolution(11.8%)15.7

2.1Total

42.5

55.4

2.1

madebythisprocessusingtheoriginalprocesswerefoundtohaveasimilarmicrostructureanddensitytothecommer-cialmadebycastingontoasubstrate(seeFig.1).Wearethereforereasonablyconfidentthathorizontaltapecastingdoesnotalterthepropertiesofthematerialtoanysignificantextentandthereforeoffersaviablemethodforstudyingtheeffectofprocessmodificationsontheplaquestrength.

ImagesofthesamplesurfacewereobtainedusinganElec-troScan2020environmentalSEM.Theimagesofthefrac-turesurfaceforquantitativeanalysisoffracturepointswereobtainedusingPhilips515SEM.Tensilestrengthmeasure-mentswereconductedusinganInstron4411loadframewithloadsupto50N.X-rayphotoelectronspectroscopyanalysisofsinteredplaqueswasperformedusingaPHI5500ESCAsystem.

4.Resultsanddiscussion

TheresultsofZ-potentialmeasurementsareshowninFig.4.TheZ-potentialincreasesrapidlywithincreasing

Fig.4.TheinfluenceofdifferentdispersantsontheZ-potentialofnickelpowder.

256A.Y.Zaitsevetal./JournalofPowerSources123(2003)253–260

Fig.5.Theeffectofdispersantsontheplaqueappearance.(a)DarvanC,(b)Darvan821A,(c)PolacrlylA40-43N.

concentrationandthenlevelsofforevendecreases.Opti-mumbenefitscanbeachievedwithaslittleas0.1%disper-santconcentration.Darvan821AandDarvanC,basedonammoniumpolyacrylate,appearedtobethemosteffectivedispersantofthosestudied.

Themicrostructuresofplaquespreparedwithvariousdis-persantsareshowninFig.5.Thereisaclearimprovementinthemicrostructureintermsofbetteruniformity,ascom-paredtotheplaquedepictedinFig.1.Similarresultswereobservedwithotherdispersants.

Despitetheobservedimprovementinmicrostructurethetensilestrengthexhibitedbyalloftheplaquesmadeus-ingtheDarvandispersantswasverylowuptofourtimeslowerthanthatoftheplaquespreparedwithoutdispersants(1.2MPaversus3–5MPa).Moreover,theseplaqueswerequitebrittle.

DetailedSEMimages(Fig.6)suggestedthattheneckgrowthbetweennickelparticlesduringthesinteringpro-cessaswellasthesurfacemorphologyofthenickelhadbeenadverselyaffectedbytheadditionofthesedispersants.Fig.6ashowsthatthefracturesurfaceofthenickelfila-mentexhibitsbrittledeformation.Fig.6bindicatesclearlythatsinteringdidnotoccurproperlyasboundariesbetweengrainsarequitevisibleandthesurfaceofthestructureisrough.Themostprobablecauseofsuchbehaviorwascon-taminationofthepowdersurfacebythedispersantsastheirpresencewastheonlydifferencebetweenconventionalandnewcompositionofsinteredplaques.

Todeterminethenatureofthecontamination,X-raypho-toelectronspectroscopy(XPS)ofthesurfaceofthepowdersinteredwithandwithoutdispersantswasperformed.TheresultsarepresentedinFig.7.

Thelowergraphwasobtainedfromasamplethatcon-tainedthedispersantDarvan821A,whiletheuppergraphwastypicalofasampleprocessedwithoutadispersant.Thelowergraphshowsthepresenceofsulfurintheform

Fig.6.Surfacesofsinterednickelpowderparticles.(a)WiththeuseofDarvanC,(b)withtheuseofDarvan821A.

A.Y.Zaitsevetal./JournalofPowerSources123(2003)253–260257

Fig.7.TheXPSdataforsinterednickelpowderintherangeof158–178eV.

ofsulphites(rightpeak)andsulphides(leftpeak).Thesulfur-containingpeaksaretheonlydifferencesbetweensamplesintheentirerangeofbindingenergies.Thus,themostprobablecauseofthedecreasedmaterialstrengthwasthepresenceofsulfurthatwasintroducedtotheslurrywiththedispersant.BulksulfuranalysisofthedispersantDarvan821Ashowedthatitcontainedabout1.3%sulfur.Sulfuriscommonlyusedtoterminatethereactionandcontrolthemolecularweightintheproductionofmanydispersants,includingsomeofthose(i.e.alloftheDarvanvariants)usedinthisstudy.

Thetensilestrengthofsinteredplaquesthatwerepre-paredusingsulfur-containingdispersantscanbeplottedasthefunctionofthesulfurcontent.TheresultsforDarvan821areshowninFig.8.SpecificstrengthplottedbyY-axiswasdefinedasthemeasuredtensilestrengthnormalizedbytherelativedensityoftheplaques.Inallcaseswefoundastrongcorrelation,whichclearlydemonstratesasignificantinfluenceofsulfuronthestrengthofporousnickelmateri-als,evenatverylowconcentrations.Thus,inordertopre-pareastrongstructurebysinteringofnickelpowder,sulfurcontaminationshouldbeavoided.

Toeliminatetheinfluenceofsulfuronthestrengthofasinterednickelbody,anumberofsulfur-freedispersantswereselectedforfurtherstudy.TheprocedureforplaquepreparationremainedasshowninFig.3.ThedispersantsusedwerePolacrylA40-43N,A70-40NandB55-50Nwithconcentrationsfrom0.05to2%basedontheweightofnickel.ThetensilestrengthandporosityofsinteredplaquesispresentedinFig.9,forplaquesmadebothbythecon-ventionalprocessandforthosemadeusingtheA40-43Ndispersantwiththemodifiedmixingsequence.Arangeofporositylevelswasachievedforeachcasebyalteringthesinteringtemperature.Theresultsforothersulfur-freedis-persantsweresimilar.Resultsobtainedbyimplementingthenewmixingsequence,butwithouttheuseofanydispersant,

Fig.8.Influenceofsulfuronthestrengthofsinterednickelpowder.Percentagesrefertothedispersant/nickelratiointheslurry.

258A.Y.Zaitsevetal./JournalofPowerSources123(2003)253–260

Fig.9.Strength,porositydependenceofsinterednickelplaques.Thetemperaturesreferredtothoseatwhichtheplaqueswerefired.

alsoshowedconsiderableincreaseinstrengthincomparisontotheconventionalmethod,thoughslightlylowerthanthestrengthobtainedusingdispersants.

Ascanbeseen,therelativeincreaseinthestrengthoftheplaquepreparedusingthenewmixingsequenceandsulfur-freedispersantaccountsfor50–70%overtheentirerangeofporositiesstudied.

AtypicalexampleofafracturedligamentispresentedinFig.10.

IncontrasttoFig.6theimagepresentedinFig.10showsahighlyductilefractureprocess,asonewouldexpectforpurenickel.Itcanbenoticedthatduringapplicationoften-sileforcetheneckofthefilamentwasdeformedandbecamethinnerbeforethefailureoccurred.Whilethisexplainsthestrengthdifferencesobservedbetweentheplaquespro-ducedwithsulfur-freedispersantsandthoseproducedusingsulfur-contaminateddispersants,itdoesnotexplainthe

Fig.10.Theexampleofafracturepoint.

significantincreaseinstrengthovertheplaquesproducedbytheoriginalprocess(forwhichthenickelligamentsalsoproducedductilefailures).Themostprobablecauseofthisimprovementisduetochangesatthemacroscopiclevelinthehomogeneityoftheparticledistributioninthesinteredplaque.

Inordertomorequantitativelyunderstandhowchangesintheprocedureofslurrypreparationaffectthestructureofthesinterednickelplaqueanditscorrelationtostrengthin-creases,thefracturesurfacesofseveralplaqueswerestudiedindetailusingSEM.Foreachmaterialstudiedabout80im-ages(60␮m×60␮meach)weretakenofrandomlychosenareasacrosseachfracturesurface.Thetotalareasoftheseimagesaccountedforabout7.5%oftheentirefracturesur-face.Theseimageswereprocessedusingimaginganalysissoftwareandthetotalareaandnumberofactualfractureligamentswasdeterminedforeachimage.TheresultsofmeasurementsandcalculationsareshowninTable2.

Theincreaseinthemeasuredstrengthoftheplaquepre-paredwiththenewmethodcanbeunderstoodfromthehighertotalareaoffractureligaments(line3)andthein-creasednumberoffractures(line4).Thisdatasuggeststhatthenewplaquesformamoreinterconnectedstructure,whichrequiresthatmoreligamentsbebrokenastheplaquefails.Anotherinformativenumberderivedfromthedataistheareaoffracturesurfaceperfractureligament(line5)Thisnumberisabout1.5timeslowerintheplaquepreparedwiththenewmethod.Itmeansthatwiththenewmethodmoreligamentsresistthetensileforce.Fromtheresultsaboveandpreviouslyestablishedtensilestrengthsofsamples,thestrengthofnickelligamentswasalsocalculated.Thisap-pearedtobeequalto402and388MPafortheconventionalandmodifiedplaque,respectively.Derivednumberscorre-latewellwiththetensilestrengthofpurenickelintheliter-aturethatiscloseto400MPa[5].

A.Y.Zaitsevetal./JournalofPowerSources123(2003)253–260

Table2

FractographicanalysisS.no.12345

Measuredandcalculateddata

Strengthofthesample(MPa)

Areastudied,At,withSEM(␮m2)

PercentageoftheareaoccupiedbyfractureligamentsNumberoffracture,nf,ligamentspermm2Fracturesurfaceareaperligament(␮m2)

Conventionalmethod3.08

2.17×1050.7662860350

259

Newmethod4.77

1.96×1051.234800208

Fig.11.Cumulativedistributionoffracturepoints.

Anotherimportantfactorinthedevelopmentoftheim-provedstrengthisthedistributionoffractureligamentsthroughtheareaoffracture.OneindicationoftheimproveddistributionisgiveninFig.9,wheretheporosityachievedbythenewprocessissomewhatlessforagiventemper-aturethanfortheconventionalprocess.TheincreaseinhomogeneityshowninFig.5suggeststhatnickelligamentshavefilledinthelargepores.Tofurtherunderstandthechangeinligamentsdistribution,thenumberoffractureligamentswasobtainedforeachsampleoutofabout80images60␮m×60␮m.Afrequencydistributionbasedonthesenumberswasderivedfortheareaofimagesthatcontainedagivennumberoffractureligaments.TheresultsareexpressedintheformofacumulativedistributionplotasshowninFig.11.

Asseenfromthegraph,thedistributionoffracturelociintheplaquepreparedbythemodifiedtechniqueisclosertotheclassicGaussdistributionanditisshiftedtowardsalargeraveragevalue.Thehighermedianvalueindicatesthatthenewprocessingtechniqueresultsinmorefracturedligamentsperunitarea,whiletheshapeofthedistributionisanindicatorofplaqueuniformity.Moreoverthegraphindicatesalargereductioninthenumberofregionsthatcontainasmalldensityofligaments.Indeed,theminimumligamentdensityisaround3000ligamentspermm2forthenewprocess,ascomparedtolessthan1000fortheoriginal

process.Thehigherligamentdensityfollowsasadirectconsequenceofremovingthelargepores.

Theimproveduniformityleadstoanincreaseoftheplaquestrengthby50–70%atthesameporositylevel.Astrengthincreaseofthismagnitudecanonlyoccurifthein-terconnectivityoftheparticlesintheplaquehasincreased.However,eventhisincreasedstrengthisaround10timeslessthanthestrengththatonewouldcalculatebasedonthetheoryofuniformfoams[6].Thesesinteredplaquesthere-foredonotformacontinuous,regularcellularstructure.Indeedmanyoftheparticlesremainunconnectedovermostoftheirlengthtootherparticles.Fromamechanicalper-spectivetheseligamentsareofnovalue.However,theystillcontributeelectricalconductivitytotheelectrode;nickelfilamentsserveasconductingprobesintotheimpregnatedactivemass.Aprocessingmethodthatwouldresultinthestrengthtoporosityratioforthesematerialscomparabletowhatonecanachieveinnickelfoamsappearstobeunlikely.However,thisratiohasbeendemonstratedtobeimprovedsubstantiallythroughprocesscontrol.

5.Conclusions

Anewapproachtoslurrypreparationhasbeendevelopedinthemanufactureofporousnickelplaquesfromfilamen-

260A.Y.Zaitsevetal./JournalofPowerSources123(2003)253–260

tarynickelpowders.Theapproachprimarilyinvolvesanim-provedsequenceforthemixingoftheslurrycomponentswiththeappropriateuseofdispersants.Abetterdistributionofnickelpowderparticlesinthegreenandsinteredmate-rialhasbeenproduced.Asmallamountofsulfurwasfoundtohaveadramaticdeleteriouseffectonthestrengthofthesinterednickelpowder.Therefore,onlysulfur-freechemi-calsshouldbeusedinprocessingthesematerials.Throughacombinationofthenewtechniqueofslurrypreparationandtheselectionofsulfur-freedispersants,thestrengthtoporosityratioofsinterednickelplaqueswasincreased.Anal-ysisoffracturesurfaceshasbeenusedtoquantifytheef-fectofthenewprocessintermsofincreaseduniformityoftheligamentspacingandincreaseddensityofconnectedligaments.

References

[1]V.A.Ettel,NewmaterialsforNiCadbatteries,in:Proceedingsof

thePaperPresentationatNickel-CadmiumBatteryConferenceNiCad96oftheInternationalCadmiumAssociation,NiCad96,Barcelona,Spain,30September–3October1996.

[2]V.A.Tracey,Nickelpowdersintosinteredstructuresforthealkaline

battery:porositystudies,Ind.Eng.Chem.Prod.Res.Dev.25(4)(1986)582.

[3]G.D.Parfitt,DispersionofPowdersinLiquids,AppliedSciencePub-lishersLtd.,London,1973,pp.1–43.

[4]D.W.Richerson,ModernCeramicEngineering,MarcelDekker,New

York,1982,p.191.

[5]W.Betteridge,NickelanditsAlloys,EllisHorwood,Chichester,UK,

1984,pp.68–69.

[6]L.J.Gibson,M.F.Ashby,CellularSolids:StructureandProperties,

CambridgeUniversityPress,Cambridge,1988.