<%BANNER%>

Enhanced Interpretation Models for Impedance of Lithium Ion Batteries


PAGE 1

ENHANCEDINTERPRETATIONMODELSFORIMPEDANCEOFLITHIUMIONBATTERIESByNELLIANNPEREZ-GARCIAADISSERTATIONPRESENTEDTOTHEGRADUATESCHOOLOFTHEUNIVERSITYOFFLORIDAINPARTIALFULFILLMENTOFTHEREQUIREMENTSFORTHEDEGREEOFDOCTOROFPHILOSOPHYUNIVERSITYOFFLORIDA2006

PAGE 2

Idedicatethisworktomygrandparents.ThesepeoplearemyheroesandIfeelveryproudtobepartoftheirfamily.

PAGE 3

ACKNOWLEDGMENTSTheauthorwishestoexpresshersinceregratitudeandappreciationtovariousindividualswhohelpedinthedevelopmentofthiswork.Tomyadvisor,Dr.MarkE.Orazem,goesmyappreciationforhisguidance,patience,anddiscussionsrelatedtotheanalysisofthiswork.Iamverygratefulthathebelievedinmeandgavemetheopportunitytodevelopmyexperimentationandmodelingskills.Ithanksmycommitteeandgroupmembersforalltheirhelpandinputinthedevelopmentofthiswork.SpecialthanksgotoVickyHuangandShirleyKellyforalltheirsupport.AlsoIwishtoacknowledgetheUniversityofFloridaGraduateMinorityFellowshipandtheNationalAeronauticsandSpaceAdministrationNASAforfundingthisproject.SpecialthanksgotoWilliamBennett,MichelleManzo,andMarlaPerez-Davisforalltheirhelpinthisproject.Finally,andmostimportant,Iwouldliketothankmyparents,OscarandCarmen,forallthesacricestheymadetosupportmeinthisproject;andmysister,Maricarmen,formotivatingmealongthisjourney.Iwouldliketothankmyhusband,Jose,forhissupportandlove.Thesepeoplegavemethecouragetopursuemydreams,andIamalwaysgoingtobegratefulforhavingthembymyside. iii

PAGE 4

TABLEOFCONTENTS page ACKNOWLEDGMENTS ............................. iii LISTOFTABLES ................................. vi LISTOFFIGURES ................................ viii ABSTRACT .................................... xiii CHAPTER 1INTRODUCTION .............................. 1 2CHARACTERISTICSOFLITHIUM{BASEDBATTERIES ....... 3 2.1HistoryofBatteries .......................... 3 2.2CharacteristicsofLithium-BasedBatteries ............. 5 2.2.1AnodeMaterials ........................ 7 2.2.2CathodeMaterials ....................... 8 2.2.3Electrolyte ........................... 9 2.2.4Electrode{ElectrolyteInterface ................ 12 3ELECTROCHEMICALIMPEDANCESPECTROSCOPYTHEORY .. 15 3.1ImpedanceResponseofElectricalCircuits ............. 16 3.2ExperimentalConsiderations ..................... 17 3.3ErrorStructure ............................ 19 3.3.1Kramers{KronigRelations .................. 19 3.3.2MeasurementModelApproach ................ 19 3.4ImpedanceSpectroscopyDataAnalysis ............... 20 3.5ImpedanceResponseModels ..................... 23 4EXPERIMENTALDESIGNANDERRORANALYSIS .......... 26 4.1ExperimentalSystem ......................... 26 4.2ExperimentDesign .......................... 29 4.3MeasurementModelAnalysis .................... 30 4.3.1SteadyStateExperimentsatConstantTemperature .... 31 4.3.2SteadyStateExperimentsatIncreasingTemperatures ... 33 4.3.3TransientExperimentsatConstantTemperature ...... 37 4.3.4BlockingElectrodes ...................... 38 iv

PAGE 5

5EXPERIMENTALRESULTSANDPRELIMINARYANALYSIS .... 40 5.1SteadyStateExperimentsatConstantTemperature ........ 41 5.2SteadyStateExperimentsatIncreasingTemperatures ....... 44 5.3TransientExperimentsatConstantTemperature .......... 49 5.4Conclusions .............................. 54 6BLOCKINGELECTRODEANALYSIS .................. 57 6.1ExperimentalResults ......................... 57 6.2GraphicalAnalysis .......................... 57 6.3ComparisontoDiskElectrode .................... 61 6.4Conclusions .............................. 63 7MODELDEVELOPMENT ......................... 64 7.1Chemistry ............................... 64 7.2Physics ................................. 67 7.3ModelDevelopment .......................... 69 7.3.1SteadyState .......................... 74 7.3.2SinusoidalSteadyState .................... 75 7.3.3MassTransfer ......................... 76 7.3.4ImpedanceExpressions .................... 79 7.4Results ................................. 80 7.5Conclusions .............................. 85 8CONCLUSIONS ............................... 86 8.1Experiments .............................. 86 8.2Model ................................. 87 REFERENCES ................................... 88 BIOGRAPHICALSKETCH ............................ 94 v

PAGE 6

LISTOFTABLES Table page 2{1Chemicalcompositionofpassivationlmsformedonlithiummetalsurfaceincontactwithdierentelectrolytes. ............. 13 2{2ChemicalcompositionofpassivationlmsontheLimetalincontactwithadditivesusedtoenhancethecyclingeciency. ......... 14 4{1Summaryoftheexperimentaldesign.Thenumberofsequentialmea-surementsispresented. ......................... 30 4{2ErrorstructuremodelsdeterminedforCellUF-CTe.Thiscellwasex-posedtoatemperaturerangevaryingfrom60Cto85C. ...... 35 4{3ErrorstructuremodelsdeterminedforCellNA{AB.Thiscellwasex-posedtoatemperaturerangevaryingfrom80Cto100Cat10Cintervals. ................................. 39 5{1Averagevaluesobtainedusinggraphicaltechniquesattheinterme-diatefrequencyrangeforCellNA{ATithroughNA{DTi.Sixrepli-cateswereusedtocalculatetheaverageparameters ......... 44 5{2AveragevaluesoftheparametersobtainedusinggraphicaltechniquesatthehighfrequencyrangeforCellNA{ATitroughNA{DTi.Sixreplicateswereusedtocalculatetheaverageparameters ....... 44 5{3AveragevaluesofthelowfrequencytailslopesfromcellsUF{BTe,UF{CTe,andUF{FTe.Sixreplicateswereusedtocalculatetheav-erageparameters. ............................ 46 5{4Slopesobtainedusinggraphicaltechniquesattheintermediatefre-quencyrangefromcellsUF{BTe,UF{CTe,andUF{FTe.Sixrepli-cateswereusedtocalculatetheaverageparameters ......... 47 5{5AveragevaluesoftheparametersobtainedusinggraphicaltechniquesatthehighfrequencyrangefromcellsUF{BTe,UF{CTe,andUF{FTe.Sixreplicateswereusedtocalculatetheaverageparameters 50 5{6ValuesofthelowfrequencytailslopesfromCellUF{DTrandUF{ETr. 53 5{7ValuesofthelowandintermediatefrequencytailslopesobtainedfromCellUF{DTrandUF{ETr. ....................... 53 vi

PAGE 7

5{8AveragevaluesoftheparametersobtainedusinggraphicaltechniquesatthehighfrequencyrangeforCellUF{DTr. ............. 55 6{1ParametervaluesextractedfromFigures 6{2 and 6{3 forCellNA-BB. 61 7{1Reactionmechanismsofcommonlithiumbatterycontaminants.Wa-terwastheonlycontaminantconsideredinthiswork. ........ 65 7{2SummaryoftheparametersusedtocalculatetheimpedanceresponseofthemodelspresentedinSection 7 ................. 81 vii

PAGE 8

LISTOFFIGURES Figure page 2{1Schematicofthearticialelectricalorgan"orthePile"developedbyAlessandroVoltain1799. ...................... 4 2{2Schematicofarechargeablelithiumbattery. .............. 6 2{3Schematicrepresentationofthelithium{polymerelectrolyteinter-phase.aOverallinterphasebSmallsegmentoftheinterphase. 13 3{1Passiveelementsthatserveasacomponentforanelectricalcircuit. 16 3{2Smallsignalanalysisofanelectrochemicalnon{linearsystem. .... 18 3{3Electrodecongurationinanelectrochemicalcell. ........... 18 3{4Impedance{planeorNyquistrepresentationofimpedancedataforCellNA{ATi.Theopensymbolsrepresentmeasurementstaken15hourafterthecellpreparation.Thehalf{fullsymbolsrepresentmeasurementstakenapproximately48hoursafterthecellprepara-tion. ................................... 21 3{5Schematicrepresentationofasimplereactivesystem. ......... 21 3{6Boderepresentationofimpedancedata,magnitudeandphaseangle,forCellNA{ATi,respectively.Theopensymbolsrepresentmea-surementstaken15hourafterthecellpreparation.Thehalf{fullsymbolsrepresentmeasurementstakenapproximately48hoursaf-terthecellpreparation. ........................ 22 3{7PlotsoftherealandimaginarypartoftheimpedanceasafunctionoffrequencyforCellNA{ATi.Theopensymbolsrepresentmea-surementstaken15hourafterthecellpreparation.Thehalf{fullsymbolsrepresentmeasurementstakenapproximately48hoursaf-terthecellpreparation. ........................ 22 4{1Schematicofthecellsandwich.Thecellwasbuildusingathin,insu-latingmasktoseparatetheelectrodes.Themaskwasintendedtopreventshortcircuit. .......................... 27 4{2Schematicoftheexperimentalsetupusedfortheelectrochemicalim-pedancetests. .............................. 28 viii

PAGE 9

4{3SchematicrepresentationofaVoigtcircuitusedasameasurementmodel. .................................. 31 4{4MeasurementmodelpredictionofimpedancedataforCellNA{ATi.Theopensymbolsrepresentmeasurementstaken15hoursaftertheirpreparation.Thehalf-fullsymbolsrepresentmeasurementstakenapproximately48hoursafterthecellpreparation.Thesolidlinerepresentsthemeasurementmodelt. .............. 32 4{5ErrorstructureforCellNA{ATimeasurementstaken15and48hoursaftertheirpreparation,respectively.Theopensymbolsrepresenttherealpartoftheimpedance.Thehalf-fullsymbolsrepresentimag-inarypartoftheimpedance.Thesolidlinerepresentstheerrorstructuremodelt. ........................... 33 4{6MeasurementmodeltoftheimaginarypartoftheimpedancedataforCellNA{ATi.Theopensymbolsrepresentmeasurementstaken15hoursaftertheirpreparation.Thehalf-fullsymbolsrepresentmeasurementstakenapproximately48hoursafterthecellprepara-tion.Thesolidlinerepresentsthemeasurementmodeltandthedashedlinesrepresentthe95:4%condenceinterval. ........ 34 4{7MeasurementmodelpredictionoftherealpartoftheimpedancedataforCellNA{ATi.Theopensymbolsrepresentmeasurementstaken15hoursaftertheirpreparation.Thehalf-fullsymbolsrepresentmeasurementstakenapproximately48hoursafterthecellprepara-tion.Thesolidlinerepresentsthemeasurementmodelpredictionandthedashedlinesrepresentthe95:4%condenceinterval. .... 35 4{8ErrorstructureforCellUF{CTemeasurementsa0CbCc70Cd5Ce80CfC.Theopensymbolsrepresenttherealpartoftheimpedance.Thehalf-fullsymbolsrepresentimaginarypartoftheimpedance.Thesolidlinerepresentstheerrorstructuremodelt. .................................... 36 4{9ErrorstructureforCellNA{ATimeasurementstaken15and48hoursaftertheirpreparation,respectively.Theopensymbolsrepresenttherealpartoftheimpedance.Thehalf-fullsymbolsrepresentimag-inarypartoftheimpedance.Thesolidlinerepresentstheerrorstructuremodelt. ........................... 37 4{10ErrorstructureforCellNA{ABmeasurementsa80CbCc100C.Theopensymbolsrepresenttherealpartoftheimpedance.Thehalf-fullsymbolsrepresentimaginarypartoftheimpedance.Thesolidlinerepresentstheerrorstructuremodelt. .......... 38 ix

PAGE 10

5{1EquivalentcircuitmodelusedtottheimpedancedataobtainedfromtheLijPEO-LiTFSIsystem.Figurereproducedfrom:R.Bouchet,S.Lascaud,andM.Rosso.AnEISStudyoftheAnodeLijPEO-LiTFSIofaLiPolymerBattery.JournalofElectrochemicalSoci-ety.Vol.150,No.10,pp.A1385{A1389,2003. ........... 40 5{2EectiveCPEcoecientdenedbyequation 5.1 forCellNA{ATi.Theopensymbolsrepresentmeasurementstakenapproximately15hoursafterthecellpreparation.Thehalf{fullsymbolsrepresentmeasurementstakenapproximately48hoursafterthecellprepa-ration. .................................. 42 5{3IR-correctedBodeplotsforCellNA{ATi;aphaseangle,bmod-ulus.Theopensymbolsrepresentmeasurementstakenapproxi-mately15hoursafterthecellpreparation.Thehalf{fullsymbolsrepresentmeasurementstakenapproximately48hoursafterthecellpreparation. ............................ 43 5{4Impedance-planeorNyquistrepresentationofimpedancedataforCellUF{CTewithtemperatureasaparameter.Sixreplicatesofeachmeasurementarepresentedforeachtemperature. ....... 45 5{5BodeRepresentationofimpedancedataforphaseangleandmagni-tudeforCellUF{CTe,respectively,withtemperatureasapara-meter.Sixreplicatesofeachmeasurementarepresentedforeachtemperature. .............................. 45 5{6PlotsoftheimaginaryandrealpartoftheimpedanceasfunctionsoffrequencyforCellUF{CTewithtemperatureasaparameter.Sixreplicatesofeachmeasurementarepresentedforeachtemperature. 46 5{7AveragevaluesofthelowfrequencytailslopesfromcellsUF{BTe,UF{CTe,andUF{FTe.Sixreplicateswereusedtocalculatetheaverageparameters. .......................... 47 5{8AveragevaluesoftheintermediatefrequencytailslopesfromcellsUF{BTe,UF{CTe,andUF{FTe.Sixreplicateswereusedtocal-culatetheaverageparameters. ..................... 48 5{9AveragevaluesofthehighfrequencytailslopesfromcellsUF{BTe,UF{CTe,andUF{FTe.Sixreplicateswereusedtocalculatetheaverageparameters. .......................... 48 5{10EectiveCPEcoecientdenedbyequation 5.1 forCellUF{CTe.Sixreplicatesofeachmeasurementarepresentedforeachtempera-ture. ................................... 49 x

PAGE 11

5{11IR-correctedBodeplotsforCellUF{CTe;aphaseangle,bmodu-lus.Sixreplicatesofeachmeasurementarepresentedforeachtem-perature. ................................ 50 5{12Impedance-planeorNyquistrepresentationofimpedancedataforCellUF{DTrwithtemperatureasaparameter.Impedancespectraweretakenevery30secondsfor48hours,ittookapproximatelytwohourstocompleteonespectrum. ................. 51 5{13BodeRepresentationofimpedancedataforphaseangleandmagni-tudeforCellUF{DTr,respectively.Impedancespectraweretakenevery30secondsfor48hours,ittookapproximatelytwohourstocompleteonespectrum. ........................ 51 5{14PlotsoftheimaginaryandrealpartoftheimpedanceasfunctionsoffrequencyforCellUF{DTr.Impedancespectraweretakenevery30secondsfor48hours,ittookapproximatelytwohourstocom-pleteonespectrum. ........................... 52 5{15EectiveCPEcoecientdenedbyequation 5.1 forCellUF{DTr. 54 5{16IR-correctedBodeplotsforCellUF{DTr;aphaseangle,bmodulus. 56 6{1CompleximpedanceplaneorNyquistrepresentationfortheresponseofa430StainlessSteeldiskinPEO/LiTSFIelectrolytewithcon-centrationasaparameter.Sixreplicatesofeachmeasurementarepresentedforeachtemperature. .................... 58 6{2PlotsoftheimaginaryandrealpartoftheimpedanceasfunctionsoffrequencyforCellNA-BBwithtemperatureasaparameter.Sixreplicatesofeachmeasurementarepresentedforeachtemperature. 59 6{3EectiveCPEcoecientdenedbyequation 6.3 forCellNA-BB.Sixreplicatesofeachmeasurementarepresentedforeachtempera-ture. ................................... 60 6{4DimensionlessanalysisfortheimpedanceresponseofastainlesssteeldiskinPEO/LiTFSIelectrolyteswithtemperatureasaparameter. 62 6{5Dimensionlessimaginarypartoftheimpedanceasafunctionofdi-mensionlessfrequencyforadiskelectrodewithlocalcapacitancedispersion. ................................ 62 7{1Impedance-planeorNyquistrepresentationofimpedancedataforCellUF-DTr.Verylargeimpedanceatlowfrequenciescanbeob-servedsuggestingthatadisconnectionofthesystemwhichmaybeattributedtogasevolution. ...................... 66 xi

PAGE 12

7{2CellUF-ATelithiumelectrodesurface.Thesurfaceoftheelectrodewasalmostcompletelycoveredwithawhitetoblueoxidelayer. .. 67 7{3CellUF-BTelithiumelectrodesurface.Theedgesoftheelectrodewerecoveredwithawhitetoblueoxidelayer,whiletheinnersur-facekeptamuchbrightermetalliclustre. ............... 68 7{4ScanningelectronmicrographsofCellUF-ATelithiumelectrodesur-face.aMagnication=35.bMagnication=200. ....... 70 7{5ScanningelectronmicrographsofCellUF-BTelithiumelectrodesur-face.Magnication=350. ....................... 71 7{6Commerciallithiumfoilnativelayercomposition. ........... 71 7{7Schematicofthesimpliedsystem.Twosurfaceswereidentiedinwhichthereactionstakeplace.Thelithiumoxidation/reductionreactiontakeplaceatthelithiummetalsurfaceandtheLiOHandLi2Oformationreactionstakeplaceintheoxidelayersurface. ... 72 7{8Schematicofthesimpliedsystem.Twosurfaceswereidentiedinwhichthereactionstakeplace.Thelithiumoxidation/reductionreactiontakeplaceatthelithiummetalsurfaceonly,buttheLiOHandLi2Oformationreactionscantakeplaceinboththelithiummetalsurfaceandtheoxidelayersurface. .............. 73 7{9Impedance{planeorNyquistrepresentationofimpedancedataofasymmetriccell.aExperimentalResult.bModel1.cModel2. 82 7{10Imaginaryplotsofimpedancedataofasymmetriccell.aExperi-mentalResult.bModel1.cModel2. .............. 83 7{11Imaginaryrepresentationofimpedancedataofasymmetriccell.aExperimentalResult.bModel1.cModel2. ........... 84 xii

PAGE 13

AbstractofDissertationPresentedtotheGraduateSchooloftheUniversityofFloridainPartialFulllmentoftheRequirementsfortheDegreeofDoctorofPhilosophyENHANCEDINTERPRETATIONMODELSFORIMPEDANCEOFLITHIUMIONBATTERIESByNelliannPerez-GarciaMay2006Chair:MarkE.OrazemMajorDepartment:ChemicalEngineeringElectrochemicalimpedancespectroscopyexperimentsandmodelswereusedtofacilitateanin-depthunderstandingofthephysicalprocessesthatcontrolsymmetriclithiumcoincells.Theexperimentalcomponentwasintendedtoobtaingoodqualitydatatoguideinthemodeldevelopment.Thequalityoftheresultsobtainedweretestedusingthemeasurementmodelapproachdevelopedinourgroup.Experimentsperformedonsymmetriclithiumcoincells,withtimeandtemperatureasthetestedvariables,showedthatmorethanonetimeconstantcanbeidentiedinthesystem.Thedecreaseoftheimpedanceresponsewithtimeisconsistentwithdissolutionofnativeoxidelayersonthelithiumsurface,asproposedintheliterature.Theresultsshowedthatthetimerequiredforcompletedissolutionoftheoxidelayerswasontheorderofseveraldays.Asthetemperatureofthecellswasincreasedtheimpedanceresponsedecreased.Thisresultisconsistentwithtwohypotheses,thatthehigh-frequencycapacitiveloopisduetoreactionsatthelithiummetalsurfaceorthattheloopisduetoadielectricbehavioroftheelectrolyte. xiii

PAGE 14

Experimentsusing430stainlesssteelelectrodeswereperformedtotestthehypothesisthattheloopwasassociatedwithadielectricbehavioroftheelectrolyte.Theresultsobtainedprovedthatthehigh-frequencycapacitiveloopisassociatedinsteadwithreactionsatthelithiumelectrodeandnotwithadielectricbehavioroftheelectrolyte.Thelow-frequencybehaviorwasconsistentwiththatexpectedforblockingelectrodes.However,athighfrequencies,theimpedancewasinuencedbyanon-uniformcurrentdistribution.Theobjectiveofthemodelingworkwastondabridgebetweeneasily-usedcircuitmodelsandmorecomplexphysico-chemicalmodels.EISisnotastand-alonetechnique,andadditionalinformationwasneededtodevelopsuchamodel.Anextensiveliteraturereviewwasdonetoinvestigatethereactionsandmechanismsoftransportthatcontrolthesymmetriclithiumcoincells.Asimpliedsetofreactions,whichincludetheoxidationreductionreactionoflithiumandthereactionsduetothepresenceofwaterasacontaminant,waschosen.TheproductsofthereactionschosenincludeLiOH,LiO2,andH2.Independentobservationssupportedthechoiceofreactions.Theimpedancedataandbrokensealsinthecellsprovideevidenceofhydrogenevolution.MicroscopicandScanningElectronMicroscopeSEMmicrographsprovidedevidenceoftheformationofsolidlmsonthelithiumsurface,consistentwiththeformationofLiOHandLiO2.Also,themicrographsshowedthatthelmformedattheperimeterofthecellwhichisevidenceofanon-uniformcurrentdistribution.Twopossiblemechanismsoftransportweremodelled.Bothmodelsprovidedevidencethatthedistributionoftimeconstantsobservedinthehigh-frequencycapacitiveloopcanbeattributedtoreactionsatthelithiumelectrode.Themodelsdevelopedareconsistentwiththeimpedanceresponseandaresupportedbyindependentobservations. xiv

PAGE 15

CHAPTER1INTRODUCTIONRechargeablelithiumbatteriesareapromisingenergysourceforconsumer,industrial,andmilitaryapplications.Lithiumbatteriesareappealingfortheseapplicationsbecauselithiummetaloersalowweightanode.99gmol-1withhighnegativepotential-3.05V/NHEandhighspeciccapacity.76Ahg-1.1Theoverallperformanceofalithiumrechargeablebatterydependsonthechoiceofcathode,anode,electrolyte,andtheelectrode-electrolyteinterfacialproperties.2SolidpolymerelectrolytesSPEsoerimprovedlithiumbatterysafetyascomparedtoliquidelectrolytes,sinceSPEsdonotcontainanyvolatileorreactivecomponents.Ontheotherhand,SPEsareknowntoresultinbatterieswithinsucientperformance,sueringfromlargevoltagelossesintheelectrolyteduringpassageofacurrent.3Forthisreason,newmaterialsarebeingdevelopedforthenextgenerationoflithium-basedbatteries.4Thesearchforoptimalchemistryandbatterydesignforagivenapplicationcanbecostlyandtime-consuming.Mathematicalmodelingcanachievemanyofthesegoalsaswellasimprovethefundamentalunderstandingoftheprocessesinsidethebattery.5ElectrochemicalImpedanceSpectroscopyEISisanattractiveexperimentaltechniquetostudylithiumbatteriesbecauseitisnon{destructiveandavarietyofcapacitiveandresistivepropertiesofthematerialcanbedeterminedinasinglebriefexperiment.Interpretationofimpedancedataisgenerallyguidedbyuseofequivalentelectricalcircuits.Bothsimplelumped-parametercircuitsandmorecomplexnite-transmission-linecircuitshavebeenused.Theformulationoftheoverallcircuitnetworkinvolvesthesemi-empiricaladditionofcircuitelementsuntilthemodelprovidesanadequatedescriptionoftheimpedanceresponse. 1

PAGE 16

2 Upontheestablishmentofanappropriatenetwork,adeductiveprocessisusedtorelatethepropertiesofthesystemtotheindividualcircuitelements,whichcomprisetheoverallnetwork.Onelimitationofsuchelectriccircuitmodelsisthatanunambiguousinterpretationofthedataisnotpossible.Forparticularimpedancespectra,morethanonecircuitnetworkcanbeconstructedwhichwouldprovideequivalentstatisticallyvalidts,andeachofthesemodelscanarisefromextremelydierentinterpretations.Thisreducestheecacyofcircuitnetworksforinterpretationofthephysicalpropertiesandthekineticprocessesofthesystem.Alessambiguousapproachforinterpretingimpedancedataistodevelopphysico-chemicaldeterministicmodelsbasedonthegoverningequations,whichdescribethephysicsandchemistryofthesystemandthatcanberegressedtotheexperimentaldata.Theobjectiveofthisworkwastondabridgebetweenthecomplexphysico-chemicalmodelswithmorecommonlyusedequivalentcircuitmodels.Theworkcombinedexperimentswiththedevelopmentofmodelsthatcouldbeappliedtolithium-basedbatterysystems.ThisinvestigationwasperformedaspartofNASA'sPolymerEnergyRechargeableSystemsPERSprogram.Theprogramwasestablishedtodevelopthenextgenerationoflithium-based,polymerelectrolytebatteriesforaerospaceapplications.Thegoaloftheprogramistodevelopaspace-qualied,advancedbatterysystemembodyingasolidpolymerelectrolyteandlithium-basedelectrodetechnologiesandtoestablishworld-classdomesticmanufacturingcapabilitiesforadvancedbatterieswithimprovedperformancecharacteristicsthataddressNASA'sfutureaerospacebatteryrequirements.6

PAGE 17

CHAPTER2CHARACTERISTICSOFLITHIUM{BASEDBATTERIESLithium{basedbatterieshavebeenthefocusofresearchforthelastdecades.Thesebatteriesoersignicantadvantagessuchasreducedweightandvolumeoftheenergystoragesystem,improvedreliability,lowerpowersystemlifecyclecosts,higherenergydensity,andamorebenignenvironmentalimpact.2Forthesereasonstheyareconsideredfordiverseapplicationsthatincludeportableelectronicdevices,suchascellularphonesandlaptopscomputers,electricvehicles,spaceexplorations,uninterruptedpowersupplies,andmedicaldevices.Lithium-ionbatterieshavereplacedlead-acid,nickel-metal,andnickel-cadmiumbatterysystemsforseveralapplications. 2.1 HistoryofBatteriesTherstbatteryreportedinhistorywasinventedbyAlessandroVoltaintheyear1799.7Herstnameditarticialelectricalorgan,"becausehewastryingtoreproducethebehaviorofthetorpedosh.HisinventionwaslaterknownasthePile."Thearticialelectricalorgan"wastheresultofacontroversybetweenVoltaandhiscolleagueLuigiGalvani,amedicaldoctoratBolognaUniversity.7,8Galvaniobservedthatheobtainedmuscularcontractioninafrogbytouchingoneendofthelegwithonemetalandtheotherendwithadierentmetal.Heconcludedthattheanimalelectricuid"wasthesourceofthecontraction.Volta,ontheotherhand,believedthatthecontactofdissimilarmetalswasthetruesourceofstimulation.Totesthishypothesis,hebuiltthearticialelectricalorgan."Theapparatusconsistedofamoistenedporouspadsandwichedbetweentwodissimilarmetals.AdetailedsketchofitispresentedinFigure 2{1 .Themetalsusedweresilverandzinc,andthediskwasmoistenedwithwaterforsome 3

PAGE 18

4 Figure2{1:Schematicofthearticialelectricalorgan"orthePile"developedbyAlessandroVoltain1799.FigurereproducedfromA.P.Chagas,The200YearsoftheElectricPile,"QuimicaNova,23000427{429.7 cellsandasalinesolutionforothers.Voltaobservedandreportedthatthesalinesolutionworkedbetterthanpurewaterinthemoisteneddisks,butheshowednointerestinexploringtheroleoftheelectrolytesincehebelievedthatthemetal{metalinterfacewasthesourceofelectricity.8Intheyearsthatfollowed,othermeansofproducingelectricitywereinvented.TheBritishresearcherJohnFrederichDanielldevelopedanarrangementinwhichacopperplatewaslocatedatthebottomofawide{mouthedjar,andacastzincpiecewaslocatedatthetop.Twoelectrolyteswereemployed.Asaturatedcoppersulphatesolutioncoveredthecopperplateandextendedhalfwayuptheremainingdistancetowardthezincpiece.Thenazincsulphatesolutionwascarefullypouredabovethecoppersulphateandimmersedthezinc.Thetwosolutionswereseparatedbyaporousclaycylinderseparator.Itwasoneoftheearliestpracticallaboratory"electricalsources,butitwasnotmuchusedoutsidethelaboratory.8,9Amongallthesystemsinvented,themostsuccessfulweretheonesdevelopedbyGrovein1839andBunsenin1842.Groveaddedanoxidizingagenttopreventaccumulationofhydrogenatthecathodewhereitreducedthe

PAGE 19

5 voltagethecellproduced.BunsenimprovedGrovescellbysubstitutingcheapcarbonfortheexpensiveplatinumcathode.9Noneofthesetechnologiessurvivestoday.TherstenduringinventioncamefromRaymondGastonPlante,whodevelopedthelead-acidcellin1859.ThenextmajordevelopmentwasthewetcellbytheFrenchengineerGeorgesLaclanche.Heusedacathodeofmanganesedioxidemixedwithcarbonandananodeofzincintheformofarod.Theelectrolytewasasolutionofammoniumchloride.Thischemicalsystemisusedinashlightbatteries.9TheproductionofbatterieswasgreatlyincreasedduringtheFirstWorldWarasameansofpoweringtorchesandeldradios.Othermilestonesinbatterypro-ductionincludethewidespreadradiobroadcasting,whichbroughtbattery{operatedwirelessintomanyhomes.Duringtheinter{waryears,batteryperformancewasgreatlyenhancedthroughbetterselectionofmaterialsandmethodsofmanu-facture.Batterieshavenowbecomeanessentialpartofeverydaylife.Theyarepowersourcesformillionsofconsumer,business,medical,militaryandindustrialappliancesworldwide. 2.2 CharacteristicsofLithium-BasedBatteriesLithiumisthemostchemicallyreactivemetalandprovidesthebasisfortoday'smostcompactpowersystems.Nearlyallhigh{densitystoragesystemsuselithiumbecauseithasaspeciccapacityof3860Ampere{hourperkilogramofmassAh/kgcomparedto820Ah/kgforzincand260Ah/kgforlead.Dependingonthecathode,cellswithlithiumanodescanproduce1:5voltsto3:6voltspercell.Thethreeprimarycomponentsofanybatteryarethecathode,theelectrolyte,andtheanode.Oneproblemwithusinglithiummetalastheelectrodeisthatitistooreactive.Itreactsviolentlywithwaterandcanigniteintoame.Also,asaconsequenceofthereaction,almisformedonthemetalliclithium.This,inturn,resultsincontinuouslyincreasingelectrodesurfaceareaandmoreextensive

PAGE 20

6 Figure2{2:Schematicofarechargeablelithiumbattery.Figurereproducedfrom:G.NazriandG.Pistoia,LithiumBatteries:ScienceandTechnology,Norwell,MA:KluwerAcademicPublishers,2004. surfacereactions.Theinsulating,complexdepositsthatbuilduponthesurfaceoftheelectrodeleadtothegrowthofunevendendrites.Topreventproblemscausedbythereactivityoflithium,batterymakersreplacelithiummetalwithlithiuminitsionicstate.Inthiscell,lithium{ionsareintercalatedintotheactivematerialoftheelectroderatherthanbeingplatedoutasametal.Atthebeginningofdischarge,thenegativeelectrodeischargedwithlithiumionsLi+whilethepositiveelectrodeisreadytoacceptlithiumions.Duringdischarge,thelithiumionsleavethenegativeelectrodeandenterthesolutionphase,whileinthepositiveelectroderegionlithiumionsinthesolutionphaseintercalateintotheactivematerial.Thisbehaviorresultsinaconcentrationgradient,whichdriveslithiumionsfromthenegativetothepositiveelectrode.10,11AschematicrepresentationofatypicalLi{ioncellisshowninFigure 2{2 .Mucheorthasbeenexpendedinthelastdecadetoimprovethesebatteries,withresearchdevotedtoallthreecomponentsofthecell.Thesehaveresultedinsignicantimprovementsintheperformanceandcost.

PAGE 21

7 Thenewestbatterytechnologyisthelithiumsolidpolymercell,inwhichtheelectrolyteisintegratedintoapolymerplasticseparatorbetweentheanodeandthecathode.Theelectrolyteisapolymercompositecontainingalithiumsalt.Althoughtheenergydensityofthesolidpolymerelectrolytecanbesimilartothatoftheordinarylithiumcells,theyhavetheadvantagethattheycanbeshapedtotthespaceavailable.Inaddition,solidpolymerbatteriesareenvironmentallyfriendly,lighterbecausetheyhavenometalshellandsaferbecausetheycontainnoammablesolvent.However,theoverallperformanceofalithiumrechargeablebatterydependsonthechoiceofpositiveelectrode,negativeelectrode,electrolyte,andtheelectrode{electrolyteinterfacialproperties.2 2.2.1 AnodeMaterialsTheessentialreactionofmetalliclithiumanodeisverysimpleLi$Li++e)]TJ/F15 11.955 Tf 161.954 -4.936 Td[(.1but,inspiteofthissimplicity,thepracticalapplicationofLimetaltorechargeableanodehasbeenverydicultbecauseLimetaltendstodepositasdendriticstructureduringcharge,andthedisorderedmetallicdepositgivesrisetoapoorcoulombiceciency.Extensiveresearchhasbeendedicatedtoobtainbetteranodes.12{16Theresearchhasconcentratedinthreeareas:improvementsofcarbonaceousmaterials;utilizationofmetalliclithiumwithstableelectrolytes;andsearchfornewmaterials.10Mostoftheimprovementsinanodesactuallyimplementedoverthelastdecadehaveinvolvedcarbonaceousmaterials,inconcertwithlayeredoxides.Thereasonsthatliebehindthecommercialsuccessofcarbon{basednegativeelectrodesincludetherelativelylowinherentcostofcarbon,itsexcellentreversibilityforlithiuminsertion,andtheformationofaprotectivesurfacelmwithmanyelectrolyte

PAGE 22

8 solutions.Thechoiceofanodeisbasedontheneedforfastinsertionkineticsandaredoxpotentialvs.lithiumwhichprovidesasucientlylargecellvoltage.Despitetheadvancesmadeincarbons,ithasbeenrecognizedforsomeyearsthatnewanodematerialsareneeded.Anincreaseintheoperatingvoltageabovecurrentvaluesforcarbonaceousmaterialsishighlydesirabletoinhibitlithiummetaldepositionthatcanoccuratfastrates.Alargenumberofalternativepossibilitiesfornegativeelectrodeshavebeenrecentlyreportedintheliteraturethatmayleadtonewdirectionsinmeetingtheanodechallenge.ThegoalofdevelopingrechargeablebatterieswithametallicLianodehasbecomemoredenite.OneoftheeldsinwhichaLimetalanodeisattractingmuchinterestisthatofpolymerbatterysystems.Toobtainasatisfactoryperfor-mancefrommetallicLianodeitisnecessarytooptimizechemicallyandphysicallytheinterfacialstructurebetweenLianodeandtheelectrolyte.Itisnecessarytochooseappropriateelectrolytestopromotetheformationofapassivatinglayerthatwillprotectagainstsidereactions,butthatdonotimpedethediusionofLiionsacrosstheinterfaceduringcycling.AnotherapproachtotheimprovementofLicyclingeciencyisincludingsomeadditivesintheelectrolyteformulation.Thismethodcancontrolormodifythethickness,morphologyandchemicalcompositionoftheLisurfacelayer.10AdetaileddescriptionisprovidedinSubsection 2.2.4 .Besidesthechemicalfactorssuchaselectrolytesandadditives,somephysicalfac-torshaveaninuenceontheLiinterfacestructure.Thetemperatureandstackingpressureontheanodehaveaninuenceinthesurfacemorphology,whichcanimprovethecyclingeciency. 2.2.2 CathodeMaterialsOverthelastdecadetheresearchoncathodematerialshasbeenintense,andremarkablesuccesshasbeenmetatthepracticalandatthetheoreticallevel.1,17

PAGE 23

9 Theclassicalcriteriaforselectinganecientcathodearebasedonthermodynamic,kinetic,andpracticalconsiderationsinclude 1. ThecathodemustintercalateorinsertLi+. 2. ThecathodehastohavelowelectronenergyandlowsiteenergyforLi+. 3. TheelectrodepotentialshouldhavelimitedvariationsasafunctionofLi+content. 4. ThenumberofsitesforLi+hastobehighandthehostingmoleculehastofeaturelowweightandhighdensity. 5. ThecouplediusionofelectronsandLi+inthehost,asafunctionofconcentrationgradient,mustbefastenoughtograntagoodratecapability. 6. TheLi+intercalation/insertionhastobereversibletoallowcyclability. 7. Thecathodemustbestabletowardstheelectrolyteovertheentireoperatingvoltagerange. 8. Thecathodematerialshouldbeeasytosynthesize,shouldbenontoxic,andoflowcost. 9. Thesynthesismustbereproducibleandshouldallowtheproductionofamaterialwiththedesiredparticlesize. 10. Thematerialhastobeeasilyprocessedintoapracticalelectrode. 11. Afavorableinterfaceshouldbeformedbetweenthecathodeandtheelectrolyte. 2.2.3 ElectrolyteTheselectionoftheelectrolyteisdeterminedbytheelectrodereactions.Thebasicrequirementsofasuitablepolymerelectrolyteforlithiumbatteriesare10 1. Highionicconductivitytominimizethecellresistanceandresistiveheatingofthedevice. 2. Highchemicalstabilitytopreventdecompositionofelectrolyteonthesurfaceofahighlyreducinganodematerialandahighlyoxidizingcathodematerial. 3. Electrochemicalstabilitytotoleratethehighvoltagedierencebetweentheanodeandthecathode>4V. 4. Lowmeltingpointtoprovidesucientconductivityatsub{ambienttemperaturesandpreventsolidicationanphaseseparation.

PAGE 24

10 5. Highboilingpointtoprovidesafetyandpreventexplosionsresultingfromhighpressurebuild{upinthecell. 6. Non{toxicitytobeacceptedenvironmentallyforeaseofhandling,massproduction,andwastetreatment. 7. Lowcost,safety,andeaseofprocessing.Althoughchemicalstabilityofthebulkelectrolyteisarequisiteforlongbatterylife,themostcommonelectrolytesliquidsandpolymerarenotstableattheopencircuitpotentialforlithiummetal.However,theformationofsurfacelmsattheelectrodesurfacelimitstheextentofthesereactionsandreducestheratestolevelsthatallowtheusefulcycleandcalendarlifeofthebatterytobeachieved.10,18{20ButitisimportanttounderstandthatthesesurfacelmsprovidelayerswithdierenttransportpropertiesfromthebulkelectrolyteandthroughwhichthelithiumionsmustpassSeeFigure 2{2 .TheselayersareusuallyreferredtoasthesolidelectrolyteinterfaceorSEI,andtheyusuallyplayacrucialroleinboththecyclelifeandalsointheratecapabilityofthebattery.AdetaileddescriptionoftheSEIispresentedinSection 2.2.4 .Theweightoftheelectrolyteisamajorcontributiontothatofthecompletebattery.21Battery{makersareinterestedinndingwaysofpackingbatterypowerintosmallerpackagesthatenableelectronicsmanufacturertocontinuetheirdriveforminiaturization.Forthisreasonionicallyconductingpolymershavebeenthefocusofmuchfundamentalandappliedresearchformanyyears.2,3,18,19,22{27Sincetheintroductionofpolymerelectrolytesinthe1970'smanyresearchershaveattempttoidentifysolidpolymerelectrolytesformulationswhichwouldcombinehighionicconductivity,goodmechanicalproperties,andgoodinterfacialbehaviorwithlithiummetalelectrodes.2,24ThemostcommonpolymerelectrolyteinvestigatedhasbeenbasedonpolyethyleneoxidePEOwhichiscommerciallyavailableinarelativelypurestateatreasonablecost.10TheeortswerefocusedonthebulkpropertiesofpolymerelectrolytesbasedoncombinationsofPEOwith

PAGE 25

11 variouslithiumsaltsandonthecharacteristicsofelectrochemicalprocessesattheelectrode{polymerelectrolyteinterface.Thisledtothedevelopmentofhighionicconductivitypolymersystemsthatarealmostcompetitivewithliquidelectrolytesatelevatedtemperatures.However,atelevatedtemperatures,PEO{basedelectrolytesaswellasotherpolymer{basedelectrolytesaremostlyintheiramorphousstate,inwhichthemechanicalpropertiesarerelativelypoor.24Theadditionofceramicllershasshownanimprovementofthemechanicalpropertiesoftheelectrolyte.Also,theceramicllersinduceanenhancementoftheLitransferencenumberandoftheionicconductivity.24,28Ithasalsobeenshownthatthepreparationprocedureoftheelectrolyteandtheenvironmentplayamajorrolewithrespecttotheadditionoftheller.24Theimplementationofpolymerelectrolytesincommercialbatteriesrequiresanincreaseinionicconductivityatambienttemperature.Alotofresearchhasbeendonetondpolymerelectrolytesthathavehighionicconductivityatambienttemperature.Forexample,KimandSmotkinstudiedtheeectofplasticizersontransportofelectrochemicalpropertiesofPEO{basedelectrolytesforlithiumrechargeablebatteries.2Theresultsshowedthattheplasticizedelectrolyteshavehigherconductivityandhigherdiusioncoecientattheselectedtemperatures.Also,theinterfacialresistancebetweenthepolymerelectrolyteandtheLimetalsurfacedecreased.But,thecationictransferencenumberwasreducedwiththeuseofsomeplasticizers.Thedesignofpolymerelectrolytesforlithiumbatteriesisacomplexproblemthatneedstobeapproachedwithacombinationofsophisticatedmodeling,diagnostictechniquesandsignicantlygreatersyntheticeortthanhasbeenappliedinthepast.Onlythencantheratesofsidereactionsberelatedtolifetimeissues.10

PAGE 26

12 2.2.4 Electrode{ElectrolyteInterfacePeledwasthersttointroduceandnametheelectronicallyprotective,butconductive,interfaciallayerassolidelectrolyteinterfaceorSEI.18,19AccordingtoPeled,theSEIisaheterogenouslm,consistingofamosaicofnumerousindividualparticlesofdierentchemicalcompositionbeinginpartialcontactwitheachotheratthegrainboundaries.Thechemicalcompositionoftheseparticlesdependsonthetypeofsolvent,orsolventmixture,typeofelectrolytesaltandontheimpurities.ThelithiumconductionthroughtheSEIisprovidedbyionictransportmainlyatthegrainboundariesofinorganiccompoundssuchasLiCl,LiFandLi2Oornearthesurfaceoftheparticlesvialatticedefects.Aschematicrepresentationoftheelectrode{electrolyteinterphaseispresentedinFigure 2{3 .InFigure 2{3a anativeoxidelmcanbeobserved.ItiswellacceptedthatcommerciallithiumfoilshaveanativeoxidelayerthatiscomposedprimarilyofaLi2OlayerandLi2CO3/LiOHlayer.10,29Thepolymerelectrolyteshavearoughsurface,andso,incontactwithlithium,somespikespenetratetheoxidelayerandthelithiummetalsuchthataSEIisformedattheLi/SPEinterface.Inotherareas,softercontactsbetweentheSPEandlithiumareformed.IntheseareastheSEIformsonthenativeoxidelayer,orasaresultoftheretreatoflithiumduringcorrosion,thenativeoxidelayerbreaksandthegapislledbytheformedSEI.Thenetresultisthatonlyafractionofthelithiumsurfaceisinintimatecontactwiththepolymerelectrolyte.Atopencircuitpotential,theratelimitingstepforthedeposition{dissolutionprocessisLi+transportintheSEI.Thecomponentsoftheinterfacelayerdependonthenatureoftheelectrolytesandtheadditivesaddedtotheelectrolyte.Tables 2{1 and 2{2 summarizetheeectofseveralliquidelectrolytesandadditivesinthecompositionoftheSEIlayer.

PAGE 27

13 a bFigure2{3:Schematicpresentationofthelithium{polymerelectrolyteinter-phase.aOverallinterphasebSmallsegmentoftheinterphase.Figurerepro-ducedfrom:E.Peled,D.Golodnitsky,GArdelandV.Eshkenazy,TheSEIModel{ApplicationtoLithiumPolymerElectrolyteBatteries,SolidStateIonics40952197-2204. Table2{1:Chemicalcompositionofpassivationlmsformedonlithiummetalsurfaceincontactwithdierentelectrolytes. Electrolyte PassivationFilm Withoutelectrolytenativelm LiOH,Li2CO3,Li2O GeneralPCsystems CH3CHOCO2LiCH2OCO2Li PC/LiPF6 LiF,Li2O,LiOH PC/LiClO4 Li2CO3,LiOH,Li2O,LiCl,ROCO2Li, LiCHClCHCl,LiCH2CHClCH2Cl GeneralECsystems CH2OCO2Li2 GeneralDMCsystems CH3OCO2Li GeneralDECsystems CH3CH2OCO2Li,CH3CH2OLi SO2/LiAlCl4 Li2S2O4,Li2SO3,LiSnO6,Li2S2O5 Tablereproducedfrom:G.NazriandG.Pistoia,LithiumBatteries:ScienceandTechnology,Norwell,MA:KluwerAcademicPublishers,2004.

PAGE 28

14 Table2{2:ChemicalcompositionofpassivationlmsontheLimetalincontactwithadditivesusedtoenhancethecyclingeciency. Group Additive MajorActingMechanism HF FormationofLiFlayer Inorganic AlI3,MgI2,SnI2 FormationofLi-alloylayer S)]TJ/F21 7.97 Tf 6.586 0 Td[(2x Formationofprotectinglm 2Me-furan,2Me-THF,Pyridine Surfaceadsorption,formationof derivatives,Sipyridylderivatives organicprotectinglm,solvation ofLi+ Organic Cetyltrimetylammoniumchloride Electrodesurfaceadsorption Nonionicsurfactants,CrownEthers Electrodesurfaceadsorption, solvationofLi+ Benzene Electrodesurfaceadsorption CO2 FormationofLi2CO3layerGas N2O,CO Formationofprotectinglm Tablereproducedfrom:G.NazriandG.Pistoia,LithiumBatteries:ScienceandTechnology,Norwell,MA:KluwerAcademicPublishers,2004.

PAGE 29

CHAPTER3ELECTROCHEMICALIMPEDANCESPECTROSCOPYTHEORYImpedancecanprovidevaluableinsightsintothebehaviorofalargevarietyofsubstances,componentsandsystems.Inmanymaterialstheimpedanceresponsevariesasthefrequencyoftheappliedvoltagechanges.Thismaybeduetothephysicalstructureofthematerial,tochemicalprocesseswithinit,ortoacom-binationofboth.Thus,ifameasurementofimpedanceoverasuitablefrequencyrangeismade,andtheresultsareplottedonsuitableaxes,itispossibletorelatetheresultstothephysicalandchemicalpropertiesofthematerial.Thistechniqueisknownaselectrochemicalimpedancespectroscopyandisusetoinvestigatematerialsandchemicalmechanisms.Electrochemicalimpedancespectroscopyisameasurementoftheconductiveanddielectricpropertiesofelectroactivesystemsoverawiderangeoffrequencies.Itspopularityandapplicabilityhasincreaseddramaticallyoverthepast30yearswiththeintroductionoffast{responsepotentiostatsandfrequencyresponseana-lyzers.Impedancespectroscopyhasbeenappliedextensivelyinelectrochemistry,especiallyinbatteryandsensorresearch.30,31Thischapterprovidesabasicunderstandingoftheconceptsthatareinvolvedinelectrochemicalimpedancespectroscopy.Detaileddiscussionsofthetechnicalandtheoreticalissuesassociatedwithelectrochemicalimpedancespectroscopyareavailableelsewhere.30{41 15

PAGE 30

16 Figure3{1:Passiveelementsthatserveasacomponentforanelectricalcircuit. 3.1 ImpedanceResponseofElectricalCircuitsImpedanceisdenedtobetheratioofthevoltagephasor,V,tothecurrentphasor,I.Ohm'slawcanbeexpressedinphasorformas42,43Z=V I.1TheimpedanceZindicatesoppositiontocurrentowandithastheunitsofohms.Impedanceisingeneralacomplexnumberandnotafunctionoftime.Itcompletelydenestheelement'sbehaviorinanaccircuit.ElectricalcircuitscanbeconstructedfromthepassiveelementsshowninFigure 3{1 .Forapureresistor,astheoneshowninFigure 3{1 a,theimpedanceresponseisexpressedasZresistor=R.2fortheinductorshowninFigure 3{1 bZinductor=j!L.3andforthecapacitorshowninFigure 3{1 cZcapacitor=1 j!C.4Theimpedanceresponseofresistors,inductors,andcapacitorsareusedtocon-structtheimpedanceresponseofcircuits.

PAGE 31

17 3.2 ExperimentalConsiderationsEISisasmall-signaltechniqueinwhichasinusoidalcurrentorpotentialperturbationisimposedonthesystemofinterestandthecorrespondingpotentialorcurrentresponseismeasured,Figure 3{2 .Comparisonoftheinputandoutputsignalsprovidestheimpedanceatagivenperturbationfrequency.AnappealingfeatureofEISisthatsystemswithcharacteristictimeconstantdistributedoverawiderangeoftimescalescanbestudied.Avarietyofcapac-itiveandresistivepropertiesofthematerialcanbedeterminedinasinglebriefexperimentbyanalyzingtheresponseoverawidefrequencyrange.Also,EISisanon{destructivemethodfortheevaluationofawiderangeofmaterials,includingcoatings,anodizedlms,corrosioninhibitors,batteriesandfuelcells.32,33Themaindicultyintheimpedancestudiesoriginatefromthecomplexityofelectrochemicalsystems.Animportantrequirementforasuccessfulperformanceoftheinvestigationisthatthesystemmustbelinear.34Electrochemicalsystemsareusuallynon{linearbuttherequirementforlinearitycouldbefulllediftheamplitudeoftheinputsignalissmallenoughtoallowlinearizationoftheresponse.Inaddition,theinputsignalshouldbesucientlysmallastoavoidirreversiblechangesinthesystempropertiesSeeFigure 3{2 .Thefrequencyrangemustbewideenoughtoattainthehighfrequencylimitoftheimpedanceequaltotheelectrolyteresistance.Theelectrochemicalcellplaysamajorroleinthesuccessoftheimpedanceexperiments.Thecellcanconsistof2,3or4electrodesimmersedinanelectrolyticsolutioncontainedinavessel,Figure 3{3 .Theelectrochemicalinterfacewhoseimpedanceismeasuredisgenerallylocatedbetweentheelectrolyteandtheworkingelectrode.32Thecounterelectrodeisusedtoallowcurrenttoowthroughthecell.Itisusuallyalargesurfaceareaelectrodewithverylowimpedance.Whenconcentrationgradientsintheelectrolytesolutionaecttheimpedanceresponse

PAGE 32

18 Figure3{2:Smallsignalanalysisofanelectrochemicalnon{linearsystem. ofthesystem,a3{electrodecellshouldbeused.Thereferenceelectrodeshouldbeplacedasclosetothesystemaspossible.A4{electrodecellisusedwhentransportthroughamembraneisstudied.Thepotentialdierenceacrossthemembraneismeasuredbymeansoftworeferenceelectrodes,oneoneachsideofthemembrane.Themodulationtechniquealsoplayanimportantroleinthesuccessoftheexperiments.Therearetwomodulations,potentiostaticandgalvanostatic.The Figure3{3:Electrodecongurationinanelectrochemicalcell.

PAGE 33

19 potentiostaticmodulationisthestandardapproach;inthismodulationlinearityiscontrolledbythepotential.Thegalvanostaticmodulationisgoodfornonstationarysystems,likecorrosionanddrugdeliver.However,thismodulationrequiresavariableperturbationamplitudetomaintainlinearity.Additionalconsiderationsincludetheuseofshortleads,goodandshieldedwires,andtheuseofafaradaiccagewhentheimpedanceofthesystemislarge,resultinginsmallcurrents. 3.3 ErrorStructureInterpretationoftheimpedancedatarequiresaprocessmodelandaquanti-tativeassessmentoftheerrorstructure.Theerrorstructureofthemeasurementisusedimplicitlyinregressionanalysisandhasasignicantinuenceonthequalityandamountofinformationthatcanbeextractedfromimpedancedata.Thesto-chasticerrorscanalsoinuencetheKramers-Kronigrelationsfordeterminingtheinternalconsistencyofthedata.40 3.3.1 Kramers{KronigRelationsTheKramers-Kronigrelationsareintegralequations,whichconstraintherealandimaginarycomponentsoftheimpedanceforsystemsthatsatisfyconditionsofcausality,linearity,andstability.Thismeansthatthesystemresponsetoaperturbationcannotprecedetheperturbation,thatthesystemrespondslinearlywithrespecttotheperturbation,andthattheperturbationstothesystemdonotgrow.Inprinciple,theKramers-Kronigrelationscanbeusedtodeterminewhethertheimpedancespectrumofagivensystemhasbeeninuencedbybiascaused,forexample,bytime-dependentphenomenaorinstrumentalartifacts. 3.3.2 MeasurementModelApproachThemeasurementmodelapproachdevelopedbyAgarwaletal.isusedtoidentifytheerrorstructureofthemeasurements.Theobjectiveofthemeasurementmodeltoolboxistoaccountformeasurementerrorsintheanalysisofelectrochem-icalimpedancespectroscopy.Themeasurementmodelhasbeenusedtodetermine

PAGE 34

20 whethertheresidualerrorsintheregressionareduetoaninadequatemodel,tofailureofdatatoconformtotheKramers-Kronigassumptions,ortonoise.SincethemodelsatisestheKramers-Kronigrelations,itcanbeusedtoidentifyportionsofthespectrumthatwereinconsistentwiththeserelations.Themeasurementmodelprovidesmorethanapreliminaryanalysisofimpedancedataintermsofthenumberofresolvabletimeconstantsandasymptoticvalues.Themeasurementmodelcanbeusedasalterforlackofreplicacythatallowsaccurateassessmentofthestandarddeviationofimpedancemeasurements.Thisinformationiscriticalforselectionofweightingstrategiesforregression,providesaquantitativebasisforassessmentofthequalityofts,andcanguideexperimentaldesign.Theer-roranalysisprovidesinformationtobeincorporatedintoregressionofprocessmodels.Theanalysisconsistsoftwoparts;therstoneisapreliminaryanalysistodeterminetheerrorstructureandthesecondoneistoassessconsistencywiththeKramers-Kronigrelations.AdetaileddescriptionofthemeasurementmodelapproachandtheresultsobtainedfromtheexperimentsarepresentedinChapter 4 3.4 ImpedanceSpectroscopyDataAnalysisGraphicalmethodsprovidetherststeptowardinterpretationandevaluationofimpedancedata.44Impedancedataareusuallyrepresentedinimpedance{planeorNyquistformat,asshowninFigure 3{4 .ThisplotshowstheimpedanceresponseofasymmetricallithiumbasedbatterycellwithelectrolytecompositionPEO16LiTFSI.Inthisplot,thelimitsoftherealpartoftheimpedancerepresenttheelectrolyteresistanceReathighfrequencyandthesumoftheelectrolyteresistanceandthechargetransferresistanceRe+Rfatlowfrequency.Onedisadvantageofthistypeofformatisthatthefrequencydependenceishidden.Nyquistplotsareverypopularbecausetheshapeofthelocusofpointsyieldsinsightintopossiblemechanismsorgoverningphenomena.Forexample,ifthelocus

PAGE 35

21 Figure3{4:Impedance{planeorNyquistrepresentationofimpedancedataforCellNA{ATi.Theopensymbolsrepresentmeasurementstaken15hourafterthecellpreparation.Thehalf{fullsymbolsrepresentmeasurementstakenapproximately48hoursafterthecellpreparation. Figure3{5:Schematicrepresentationofasimplereactivesystem. ofpointstracesaperfectsemicircletheimpedanceresponsecorrespondstoasingleactivation-energy-controlledprocess,suchasthatshowninFigure 3{5 .Adepressedsemicirclerepresentsasystemwithcharacteristictimeconstantdistributedoverawide-rangeoftimescales.OthercommonrepresentationsaretheBodeplotsshowninFigure 3{6 andplotsoftherealandimaginarypartoftheimpedanceasfunctionsoffrequencyshowninFigure 3{7 .Bodeplotshavetheadvantagethatthefrequencydepen-denceisshowninthegraph.Frequencyisgenerallypresentedonlogarithmicscaletorevealtheimportantbehaviorseenatlowerfrequencies.Inelectrochemical

PAGE 36

22 a bFigure3{6:Boderepresentationofimpedancedata,magnitudeandphaseangle,forCellNA{ATi,respectively.Theopensymbolsrepresentmeasurementstaken15hourafterthecellpreparation.Thehalf{fullsymbolsrepresentmeasurementstakenapproximately48hoursafterthecellpreparation. a bFigure3{7:PlotsoftherealandimaginarypartoftheimpedanceasafunctionoffrequencyforCellNA{ATi.Theopensymbolsrepresentmeasurementstaken15hourafterthecellpreparation.Thehalf{fullsymbolsrepresentmeasurementstakenapproximately48hoursafterthecellpreparation.

PAGE 37

23 systemsBoderepresentationhasseriousdrawbacks.Forexample,theinuenceoftheelectrolyteresistanceconfoundstheuseofphaseangleplotstoestimatecharac-teristicfrequencies.However,ifanaccurateestimateoftheelectrolyteresistanceisavailable,amodiedBoderepresentationthatcanbeusedtoavoidthedrawbacks.Theplotsoftherealandimaginarypartoftheimpedanceasfunctionsoffrequencyhavetheadvantagethatthecharacteristicfrequenciescanbeeasilyidentied.Whenthefrequencyisplottedinlogarithmicscalefortheimaginarypartoftheimpedance,theslopesatlowandhighfrequencyare+1and)]TJ/F15 11.955 Tf 9.299 0 Td[(1,respectively,forasimplereactivesystemastheoneshowninFigure 3{5 .Departurefrom1providesanindicationofdistributedprocesses.Observationofmultiplemaximashowsthatthedatamustbeinterpretedintermsofmorethanoneprocess.AdetailedanalysisoftheexperimentaldataispresentedinChapter 5 3.5 ImpedanceResponseModelsInterpretationofimpedancedataisgenerallyguidedbyuseofequivalentelectriccircuits.Bothsimplelumped-parametercircuitsandmorecomplexnite{transmission-linecircuitshavebeenused.Theformulationoftheoverallcircuitnetworkinvolvesthesemi-empiricaladditionofcircuitelementsuntilthemodelprovidesanadequatedescriptionoftheimpedanceresponse.Upontheestab-lishmentofanappropriatenetwork,adeductiveprocessisusedtorelatethepropertiesofthesystemtotheindividualcircuitelements,whichcomprisetheoverallnetwork.Alimitationofsuchelectriccircuitmodelsisthatanunambigu-ousinterpretationofthedataisnotpossible.Forparticularimpedancespectra,morethanonecircuitnetworkcanbeconstructedwhichwouldprovideequivalentstatisticallyvalidts;andeachofthesemodelscanarisefromextremelydierentinterpretations.Thisreducestheecacyofcircuitnetworksforinterpretationofthephysicalpropertiesandthekineticprocessesofthesystem.Alessambiguousapproachforinterpretingimpedancedataistodevelopphysico-chemicalmodels

PAGE 38

24 basedonthegoverningequations,whichdescribethephysicsandchemistryofthesystem.Todevelopsucharenedmodelitisnecessarytoidentifythereactionmechanismorifnotknownassumekineticandtransportmechanisms.Then,expressionsforthecurrentcontributionswillbewritten.Thefaradaiccurrentdensity,if,canbeexpressedintermsofasteadytime-independentvalueandanoscillatingvalueasif= if+Refeifej!tg.5TheoscillatingcomponentofthecurrentcanbecalculatedusingaTaylorseriesexpansionaboutthesteady{statevalueaseii;f=@f @Vci;keV+X@f @ci;0V;cj6=i;keci;0+X@f @kV;ci;l6=kek.6whereVistheinterfacialpotential,ci;0isthelocalconcentrationofthebulkspeciesandkisthesurfacecoverageoftheabsorbedspecies.Thechargingcurrentisgivenby i= if+cdldV dt.7andei=eif+j!cdleV.8Toobtainanexpressionforimpedanceitisnecessarytosolveequation 3.6 .Usingthereactionmechanismortheassumedmechanisms,equationsthatrelateconcentrationandsurfacecoveragetovoltagecanbeintroducetoobtainanequationofcurrentintermsofinterfacialvoltage.Thentheimpedanceofthesystemisobtainedbycomparingtheoscillatingtermofthecellpotential,eV,andtheoscillatingtermofthecurrent,ei.ThecellpotentialisdenedasU=iRe+V.9

PAGE 39

25 andeU=eiRe+eV.10whereReistheelectrolyteresistance.TheimpedanceofthesystemisdenedtobeZ=eU ei.11Substitutingequations 3.8 and 3.10 intoequation 3.11 ,anexpressionfortheimpedanceofthesystemintermofelectrolyteresistance,interfacialpotentialandchargingcurrentisobtainedasZ=Re+eV ei.12Ifthekineticandtransportmechanismswereassumedtheimpedancerelationobtainedfromequation 3.12 iscomparedtotheexperimentalvalues.Ifagoodtisobtainedtherelationisacceptedasthemodelforthesystem.However,ifagoodtisnotobtainednewkineticandtransportmechanismsareassumedandalltheprocedureisrepeateduntilagoodtisobtained.TheprocessmodelsdevelopedfortheexperimentaldataispresentedinChapter 7

PAGE 40

CHAPTER4EXPERIMENTALDESIGNANDERRORANALYSISThereexistsarichliteratureonresearchofbatterysystems,both,exper-imentalandmodelingwork.Dierenttechniqueshavebeenusedtostudythemechanismsoftransportinsidebatteries.Someofthetechniquesincludeconstantcurrentdischarge,pulsedischarge,cyclicvoltammetry,abusetests,andimpedancespectroscopy.1{3,5,6,11,12,14,15,17,25,26,28,45{63InthisworkEISwasusedtostudysymmetriclithiumcells.Thegoalwastodevelopamodelthatdescribesthephysicsandchemistryofthesystem,butthatcanberegresstotheexperimentaldata.Thismodelwouldbetherststeptowardthedevelopmentofrenedequivalentcircuitmodelsforlithium-basedbatterysystems.Experimentswereperformedtoprovidereliableimpedancedatathatwereusedtotestthemodels.Inthischapteradetaileddescriptionoftheexperimentalcomponentofthisworkisprovided.Itincludeadescriptionoftheexperimentalsystemandexperimentaldesign,aswellas,adescriptionofthemethodusedtotestthequalityofthedata.AdetaileddescriptionofthemodeldevelopmentispresentedinChapter 7 4.1 ExperimentalSystemThesystemusedinthisworkwasasymmetricLijSPEjLicoincell.Allcellassemblywasdoneindryroomatmosphere)]TJ/F15 11.955 Tf 9.299 0 Td[(35Fdewpoint,nominal.AschematicdiagramofthebatterycellassembleisshowninFigure 4{1 .Atwolithium{metal{electrodesystemwasusedsinceresearchershavestatedthatatwo{electrodesystemwillintroducemuchlesserrorthanwhatwouldresultfromanill{placedthirdreferenceelectrode.2,17Thecellwasbuiltusingathin,insulating 26

PAGE 41

27 Figure4{1:Schematicofthecellsandwich.Thecellwasbuildusingathin,insu-latingmasktoseparatetheelectrodes.Themaskwasintendedtopreventshortcircuit. masktoseparatetheelectrodesandtoprovideawelldenedreservoirforthemeltedelectrolyte.ThismaskwasintendedtopreventshortcircuitsthatwereobservedinearliercoincellworkdoneatNASAGlenResearchCenter.6ThemaskwascenteredontopoftheanodeandcoveredwiththeSPEdisc.ThelithiumelectrodewasthencenteredontopoftheSPEdiscbeforeclosingthecell.Awavespringprovidedaninternalpressureof18psi.Themaskoutsidediameterwasoversizedtoassistinthealignmentoftheelectrodesandwichinthecellcan.Theinsidediameterestablishedanapparentactiveareaof1:27cm2.TheelectrolytesystemusedinthisresearchwaspolyethyleneoxidePEOplusalithiumimidesalt:1ether{oxygen:Li.PEOhasaregularrepeatedstructure)]TJ/F15 11.955 Tf 9.299 0 Td[(CH2CH2O)]TJ/F15 11.955 Tf 9.299 0 Td[(,iscrystallineandhasafusiontemperatureTfof339K.TheimidesaltformulaisLiNSO2CF32.Thissaltisformallycalledlithiumbis-triuoromethanesulfonyl-imide,butmostpeopleabbreviatethistoLiTFSI.TheelectrolytewaspreparedusingPEOAldrich6E5Mv,LiTFSIMHQ-115andacetonitrileasthesolvent.PEOandacetonitrilewereusedasreceived.ThegradeofthePEOcontains200to500ppmofButylhydroxyltolueneBHTasstabilizer.LiTFSIsaltwasdriedfor4hoursundervacuumat160C.PEOwas

PAGE 42

28 Figure4{2:Schematicoftheexperimentalsetupusedfortheelectrochemicalim-pedancetests. pre-dissolvedasa12:3w/wsolutioninacetonitrilebeforeadditionof23:4w/wLiTFSI-acetonitrilesolution.Quantitieswereestablishedtoprovidea16:1EO:Liratioafterevaporationoftheacetonitrilesolvent.Thisisequivalentto28:9w/wLiTFSIinthesolidelectrolyte.SPElmswereroutinelypreparedbycastingtheabovesolutioninTeonPetridishesandallowingthesolventtoevaporateovernight.Castingandsolventevaporationstepswereconductedatdryroomatmosphere<1%RH.CuredSPElmsweredriedovernightundervacuumat45Candstoredinadesiccatorbeforeuse.Thisprocessproducedrubbery,freestandinglmswithastrongself-adherentcharacter.Drylmdensityatroomtemperaturewas1:26g/cm3,givingasaltconcentrationof1:27moleLiTFSI/liter.45SomeoftheexperimentswereconductedatNASAGlenResearchCenterandtheotherswereconductedattheUniversityofFloridaDepartmentofChemicalEngineering.ASolartron1250FrequencyResponseAnalyzerinterfacedwithaSolartron1286Potentiostatwasusedtocollectthedata.FortheexperimentsconductedatNASAfacilities,Z-plotwasusedforexperimentcontrolanddataac-quisition,whilesoftwarewrittenin{houseinLabviewwasusedfortheexperimentsconductedattheUniversityofFlorida.ThebatterycellswereplacedinaThermalProductSolutionsT-UJRenvironmentalchambertoensureconstanttemperaturethroughouttheexperimentsSeeFigure 4{2

PAGE 43

29 4.2 ExperimentDesignTheexperimentalworkwasdividedintofourcategories.TherstsetofexperimentswereconductedattheNASAGlenResearchCenter.Theyweresteadystateimpedancemeasurementsthatweretakenapproximately15and48hoursaftereachcellwasprepared.Opencircuitpotentialmeasurementsoffourcellsweremadewithsixreplicatesofeachscan.Thecellswerestoredataconstanttemperatureof80Cforabout10hours.Thenthetemperaturewasdecreasedto60Candkeptconstantthroughouttheexperiments.Thefrequencyrangeofthemeasurementsvariedfrom1MHzto0:02Hz.ThesecondandthirdsetofexperimentswereconductedattheUniversityofFlorida,DepartmentofChemicalEngineering.Thesecondsetofexperimentsweresteady{stateimpedancemeasurementswherethetemperaturewasincreasedfrom60Cto90Cin5Cincrements.Open{circuitpotentialmeasurementsoffourcellswereperformedwithsixtoeightreplicatesofeachscan.Thefrequencyrangeofthemeasurementsvariedfrom65KHzto0:003Hz.Thethirdsetofexperimentsweretransientimpedancemeasurementsperformedataxedtemperature.Themeasurementsweretakenconsecutivelyfor48hours.Twocellsweretested.Thetemperaturesusedwere60Cand70C.Thefrequencyrangeofthemeasurementsvariedfrom65kHzto0:003Hz.Intheseexperiments,thetemperatureinsidetheenvironmentalchamberwasequilibratedtothedesiredvalueandthenthecellswereplacedinsidethechamber.Twenty{twototweny{foursequentialimpedancespectraobtainedattheopen{circuitpotentialwereanalyzed.ThelastsetofexperimentswereconductedatNASAGlenResearchCenter.Theyweresteadystateimpedancemeasurementsusingblockingelectrodes.Theexperimentsweredesignedtotestthehypothesisthattheimpedanceresponseofthesymmetriclithiumcellswasduetotheelectrolyteandnottothereactiononthelithiummetalsurface.ThecellconsistedofPEO{LiTFSIelectrolyte

PAGE 44

30 Table4{1:Summaryoftheexperimentaldesign.Thenumberofsequentialmea-surementsispresented. Group TemperatureC Timehour 25 60 65 70 75 80 85 90 100 1 NA-ATi 15 6 48 6 NA-BTi 15 6 48 6 NA-CTi 15 6 48 6 NA-DTi 15 6 48 6 2 UF-ATe 8 UF-BTe 8 6 7 6 6 6 6 UF-CTe 6 6 12 6 6 6 6 UF-FTe 6 6 6 6 6 6 6 3 UF-DTr 24 UF-ETr 22 4 NA-AB 6 6 6 NA-BB 6 6 6 6 sandwichedbetweentwosteelelectrodes.Thecellswasexposedtoanincreasingtemperaturerangevaryingfrom80Cto100Cin10Cintervals.Sixsequentialimpedancespectraobtainedattheopen{circuitpotentialwerecollectedperset.Thefrequencyrangeofthemeasurementsvariedfrom65kHzto0:003Hz.Table 4{1 summarizestheexperimentsdesign. 4.3 MeasurementModelAnalysisThemeasurementmodelanalysiswasusedinthisworktotestthequalityoftheimpedancedata.TheanalysisconsistsofidentifyingtheerrorstructureofthemeasurementsandthenusingthaterrorstructuretodeterminewhetherthedataconformtotheKramers-Kronigassumptions.Todeterminetheerrorstructureofthemeasurements,aVoigtmodelwasttedtotheexperimentaldata.Voigtelementsinserieswiththesolutionresistance,i.e.,Z=R0+XRk 1+j!k.1

PAGE 45

31 Figure4{3:SchematicrepresentationofaVoigtcircuitusedasameasurementmodel. wereaddedasshowninFigure 4{3 .Withasucientnumberofparameters,theVoigtmodelisabletoprovideastatisticallysignicantttoabroadvarietyofimpedancespectra.64,65However,themodelshouldnotbeassociatedwiththesetofdeterministicortheoreticalparametersforagivensystem.Theuseofthemeasurementmodelforanalysisoftheerrorstructurerequiresreplicatemeasurementsoftheimpedanceresponseusingthesamemeasurementfrequenciesforeachreplicate.Mostelectrochemicalsystemsofpracticalimportancearenonstationaryandtheresultingtrendingbiasesdirectcalculationofstatisticalquantitiessuchasmeanandvariance.Toeliminatethecontributionofthedriftfromscantoscan,themeasurementmodelwasregressedtoeachscanindividually.ForthesystemtobeconsistentwiththeKramers-Kronigrelations,ithastoevolvesucientlyslowlythatthechangeduringonecompletescanisinsignicant. 4.3.1 SteadyStateExperimentsatConstantTemperatureTheimpedanceresponseandthemodelobtainedusingthemeasurementmodeltechniqueofCellNA{ATiareshowninFigure 4{4 .Twodatasetsarepresentedinthisgure,onewasobtainedapproximately15hoursafterthecellpreparationandtheotherwasobtainedapproximately48hoursafterthecellpreparation.ThersttenpointsathighfrequencywereeliminatedbecausethosepointswerecorruptedbyinstrumentartifacteectsandwerenotconsistentwiththeKramers-Kronigrelations.Themodelagreedwellwiththeexperimentaldata.

PAGE 46

32 Figure4{4:MeasurementmodelpredictionofimpedancedataforCellNA{ATi.Theopensymbolsrepresentmeasurementstaken15hoursaftertheirpreparation.Thehalf-fullsymbolsrepresentmeasurementstakenapproximately48hoursafterthecellpreparation.Thesolidlinerepresentsthemeasurementmodelt. TheerrorstructuresforthetwodatasetsareshowninFigure 4{5 .Theseplotsshowtheerrorstructurefortherealandimaginarypartoftheimpedance,aswellastheerrorstructuremodel.Theerrorstructuremodelwascalculatedtobe=0:001035Zj+0:000467Zr)]TJ/F15 11.955 Tf 11.955 0 Td[(50:1038.2Thismodelwascalculatedusingtheerrorstructuresofthefourcells.Forthemeasurementstaken15hoursafterthecellpreparation,thestandarddeviationsoftherealandimaginarypartoftheimpedanceatlowfrequencieswerenotequal,Figure 4{5a .Thislackofequalityofstandarddeviationfortherealandimaginarypartsoftheimpedancesuggeststhatthedatawereinuencedbynon-stationarybehavior.Thestandarddeviationoftherealandimaginarypartoftheimpedancewasequalforthemeasurementstaken48hoursafterthecellpreparation,Figure 4{5b .ThesecondpartofthisanalysiswastoassessfortheKramers-Kronigconsis-tency.AVoigtelementmodelwasttedtotheimaginarypartoftheimpedance.Theexperimentallydeterminederrorstructurewasusedtoweightthemodel.Whenanadequatedescriptionwasobtained,themodelwasusedtopredicttherealpartofthespectrum.Fromtheestimatedvarianceofthemodelparameters,a

PAGE 47

33 a bFigure4{5:ErrorstructureforCellNA{ATimeasurementstaken15and48hoursaftertheirpreparation,respectively.Theopensymbolsrepresenttherealpartoftheimpedance.Thehalf-fullsymbolsrepresentimaginarypartoftheimpedance.Thesolidlinerepresentstheerrorstructuremodelt. 95{percentcondenceintervalwasfoundforthemodelpredictions.Ifthepredicteddataarewithinthe95{percentcondenceinterval,thesystemcanberegardedasbeingstationaryduringthecourseoftheexperiment.Ifasignicantportionofthepredicteddatafallsoutsidethe95{percentcondenceinterval,thesystemislikelytobenon-stationary.36{39,41,66Otherwise,ifanydatapointfallsoutsidethe95{percentcondenceintervalitwillbeeliminatedformthedataset.OnlytheportionsofthespectrathatareconsistentwiththeKramers-Kronigrelationswereusedforfurtheranalyses.TheresultsfortheKramers-KronigconsistencycheckareshowninFigures 4{6 and 4{7 .Themodelpredictedeverydatapointwithinthe95%condenceintervalwhichmeansthat,withintheerrorstructureofthemeasurement,alldatacanbesaidtobeconsistentwithKramers-Kronigrelations. 4.3.2 SteadyStateExperimentsatIncreasingTemperaturesThesameprocedurewasfollowedforthesteadystateexperimentswherethetemperaturewasincreasedsequentially.Thetemperaturerangestudiedvariedfrom60Cto85C.Temperaturesabove85Cwerenotanalyzedbecausethebehavior

PAGE 48

34 a bFigure4{6:MeasurementmodeltoftheimaginarypartoftheimpedancedataforCellNA{ATi.Theopensymbolsrepresentmeasurementstaken15hoursaftertheirpreparation.Thehalf-fullsymbolsrepresentmeasurementstakenapproximately48hoursafterthecellpreparation.Thesolidlinerepresentsthemeasurementmodeltandthedashedlinesrepresentthe95:4%condenceinterval.

PAGE 49

35 a bFigure4{7:MeasurementmodelpredictionoftherealpartoftheimpedancedataforCellNA{ATi.Theopensymbolsrepresentmeasurementstaken15hoursaftertheirpreparation.Thehalf-fullsymbolsrepresentmeasurementstakenapprox-imately48hoursafterthecellpreparation.Thesolidlinerepresentsthemea-surementmodelpredictionandthedashedlinesrepresentthe95:4%condenceinterval. Table4{2:ErrorstructuremodelsdeterminedforCellUF-CTe.Thiscellwasex-posedtoatemperaturerangevaryingfrom60Cto85C. TemperatureC ErrorStructureModel 60 =8:7422E)]TJ/F15 11.955 Tf 9.299 0 Td[(4jZrj 65 =1:4348E)]TJ/F15 11.955 Tf 9.299 0 Td[(3jZjj+1:1404E)]TJ/F15 11.955 Tf 9.299 0 Td[(3jZrj+1:1541E)]TJ/F15 11.955 Tf 9.298 0 Td[(5jZj2=Rm 70 =9:2728E)]TJ/F15 11.955 Tf 9.299 0 Td[(4jZrj+2:8481E)]TJ/F15 11.955 Tf 9.298 0 Td[(2 75 =5:0172E)]TJ/F15 11.955 Tf 9.299 0 Td[(4jZjj+6:0620E)]TJ/F15 11.955 Tf 9.299 0 Td[(4jZrj 80 =1:1159E)]TJ/F15 11.955 Tf 9.299 0 Td[(3jZjj+3:5791E)]TJ/F15 11.955 Tf 9.299 0 Td[(4jZrj 85 =7:6544E)]TJ/F15 11.955 Tf 9.299 0 Td[(3jZrj ofthecellswasnotstableatthosetemperatures.TheerrorstructuresforcellUF{CTeareshowninFigure 4{8 .Theseplotsshowtheerrorstructurefortherealandimaginarypartoftheimpedance,aswellastheerrorstructuremodel.Agoodagreementbetweenthemodelandtheexperimentaldataforthe60Cwasnotpossible,howevertherealandimaginaryerrorswereequal.TheerrorstructuremodelsforthesixdatasetsarepresentedinTable 4{2 .Thesemodelswerecalculatedusingtheerrorstructuresofthesixreplicatesmeasuredateachtemperature.Itwasnotpossibletoobtainasinglemodelforallthetemperatures.

PAGE 50

36 a b c d e fFigure4{8:ErrorstructureforCellUF{CTemeasurementsa0CbCc0CdCe80CfC.Theopensymbolsrepresenttherealpartoftheimpedance.Thehalf-fullsymbolsrepresentimaginarypartoftheimpedance.Thesolidlinerepresentstheerrorstructuremodelt.

PAGE 51

37 Figure4{9:ErrorstructureforCellNA{ATimeasurementstaken15and48hoursaftertheirpreparation,respectively.Theopensymbolsrepresenttherealpartoftheimpedance.Thehalf-fullsymbolsrepresentimaginarypartoftheimpedance.Thesolidlinerepresentstheerrorstructuremodelt. AlldatapointsinthesetwereconsistentwithKramers-Kronigrelations.ThesameprocedurewasrepeatedforCellsUF{BTeandUF{FTe. 4.3.3 TransientExperimentsatConstantTemperatureThisanalysiswasveryimportantforthetransientexperiments.TheanalysishelpedusidentifywhichportionsofthespectrawereconsistentwiththeKramers-Kronigrelationsandcouldbeusedtoidentifymeaningfulparametersasfunctionsoftime.Thisanalysisalsohelpedidentifyhowmuchtimewasneededtoreachsteadystateconditions.TheerrorstructuresforcellUF{DTiareshowninFigure 4{9 .Thisplotshowstheerrorstructurefortherealandimaginarypartoftheimpedance,aswellastheerrorstructuremodel.Theerrorstructuremodelwascalculatedtobe=0:00058896Zj+0:0016651Zr)]TJ/F15 11.955 Tf 11.955 0 Td[(29:00.3Thismodelwascalculatedusingtheerrorstructuresofthetenofthetwenty{fourmeasurements.

PAGE 52

38 a b cFigure4{10:ErrorstructureforCellNA{ABmeasurementsa0CbCc00C.Theopensymbolsrepresenttherealpartoftheimpedance.Thehalf-fullsymbolsrepresentimaginarypartoftheimpedance.Thesolidlinerepresentstheerrorstructuremodelt. 4.3.4 BlockingElectrodesThisanalysiswasalsodonefortheblockingelectrodemeasurements.TheerrorstructuremodelsforcellNA{ABareshowninFigure 4{10 andinTable 4{3 .Thesemodelswerecalculatedusingtheerrorstructuresofthesixreplicatesmeasuredateachtemperature.Thisprocedurewasrepeatedforeverydataset.DatapointfoundtobeinconsistentwiththeKramers{Kronigrelationswerenotconsideredforfurtheranalysis.

PAGE 53

39 Table4{3:ErrorstructuremodelsdeterminedforCellNA{AB.Thiscellwasex-posedtoatemperaturerangevaryingfrom80Cto100Cat10Cintervals. TemperatureC ErrorStructureModel 80 =3:6053E)]TJ/F15 11.955 Tf 9.298 0 Td[(4jZjj+1:7754E)]TJ/F15 11.955 Tf 9.298 0 Td[(7jZj2=Rm 90 =4:5841E)]TJ/F15 11.955 Tf 9.298 0 Td[(3jZrj+1:9697E)]TJ/F15 11.955 Tf 9.299 0 Td[(7jZj2=Rm 100 =4:3598E)]TJ/F15 11.955 Tf 9.298 0 Td[(3jZrj+1:8417E)]TJ/F15 11.955 Tf 9.299 0 Td[(7jZj2=Rm

PAGE 54

CHAPTER5EXPERIMENTALRESULTSANDPRELIMINARYANALYSISBodeplotsandcomplex-impedance-planeplotsarethetypicalgraphicalrepresentationswhenaspecicmodelisnotpostulated.However,additionalplots,likeIR-correctedBodeplots,log-logplotsoftheimaginarycomponentoftheimpedance,andeectivecapacitanceplotscanprovideguidanceforthedevelopmentofappropriatephysicalmodels.Equivalentcircuitmodelshadbeenpublishedintheliteraturethatprovideadequatetstotheimpedancedata.Forexample,R.Bouchet,S.Lascaud,andM.RossodevelopedanequivalentcircuitfortheLijPEO-LiTFSIsystem,asshowninFigure 5{1 .63Theirworkwasfocusedontheevolutionoftheimpedancespectraofsymmetriccells,LijPEO-LiTFSIjLi,withagingtimeat90C.Intheirexperimentalresults,theydistinguishfourfrequencydomains, 1. f>105Hz:correspondstothebulkoftheelectrolyte.Thispartofthespectrawasperturbedbyhighfrequencyinductiveeectsduetoconnections. 2. 105{102Hz:attributedtotheformationofasurfacelayer. 3. 102{10:1Hz:notveryreproducible. Figure5{1:EquivalentcircuitmodelusedtottheimpedancedataobtainedfromtheLijPEO-LiTFSIsystem.Figurereproducedfrom:R.Bouchet,S.Lascaud,andM.Rosso.AnEISStudyoftheAnodeLijPEO-LiTFSIofaLiPolymerBattery.JournalofElectrochemicalSociety.Vol.150,No.10,pp.A1385{A1389,2003. 40

PAGE 55

41 4. 10:1{10)]TJ/F15 11.955 Tf 7.085 -4.338 Td[(4Hz:Warburgimpedancecharacteristicofthetransportofchargedparticles.Theproposedequivalentcircuitassumedthatthephenomenacorrespondingtothefourfrequencydomainsareinseries.Theauthorsclaimedthattheproposedmodelcanreproducehisexperimentaldata.ThegraphicalanalysispresentedinthissectioncanhelpacceptorrejectmodelsliketheoneproposedbyR.Bouchet,S.Lascaud,andM.Rosso.Inthissectionweusegraphicalanalysestoobtainanunderstandingofthephysicsofthesystem. 5.1 SteadyStateExperimentsatConstantTemperatureThetraditionalNyquistrepresentationoftheimpedancedataobtainedfromCellNA{ATiareshowninFigure 3{4 and 3{6 .Themeasurementsweremadeapproximately15and48hoursafterthecellpreparation.Thesiximpedancereplicatesateachtimearepresentedinthisplot.Theelectrolyteresistancevalueobtainedfromthisplotis62:0{cm2fortherstdatasetand59:7{cm2forthesecond.TheshapeoftheNyquistplotindicatesthatthesystemhavecharacteristictimeconstantsdistributedoverawide{rangeoftimescales.ThealternaterealandimaginaryrepresentationsarepresentedinFigure 3{7 .Theslopesatintermediateandhighfrequencywerecalculatedfortheimaginaryrepresentation.Theslopesatintermediateandhighfrequencywereingoodagreementforthesixreplicates.Theseslopeshavethevalueof)]TJ/F22 11.955 Tf 12.793 0 Td[(,anddeparturefrom1providesindicationofdistributedprocesses.Areductioninthevaluesofatintermediatefrequencyandanincreaseinthevaluesofathighfrequencywereevidentforthemeasurementstaken48hoursafterthecellpreparation.Alinewithaverageslopeof0:740:0086wasttedtotheintermediatefrequencyfortherstdatasetandalinewithaverageslope0:700:0117wasttedfortheseconddataset.Alinewithaverageslope)]TJ/F15 11.955 Tf 9.299 0 Td[(0:620:0081

PAGE 56

42 Figure5{2:EectiveCPEcoecientdenedbyequation 5.1 forCellNA{ATi.Theopensymbolsrepresentmeasurementstakenapproximately15hoursafterthecellpreparation.Thehalf{fullsymbolsrepresentmeasurementstakenapproxi-mately48hoursafterthecellpreparation. wasttedtotherstdatasetathighfrequencyandalinewithaverageslope)]TJ/F15 11.955 Tf 9.298 0 Td[(0:640:0091wasttedtotheseconddataset.Thedierenceinthesetwoslopesindicatesthatmorethanonecapacitivetimeconstantcanbedistinguishedinthesystem.ThevalueofcanbeusedtoobtainanapparentconstantphaseelementCPEcoecientQeQe=cos 2 )]TJ/F22 11.955 Tf 9.299 0 Td[(Zjff1)]TJ/F23 7.97 Tf 6.586 0 Td[(.1TheresultingvaluesoftheQearepresentedinFigure 5{2 .Theabsenceofaclearlyidentiableasymptotemaybeattributedtohigh{frequencyinstrumentalartifacts.Thelasttenpointsathighfrequencyweredisregardedbecauseitisbelievedthattheywerecorruptedbysuchinstrumentalartifacts.ThevaluefortheCPEcoecientprovidedinTable 5{2 representstheaverageoverthevaluesfortheremaining10highestfrequencies.ThevalueofcanalsobeusedtondthesolutionresistanceReusedin=tan)]TJ/F21 7.97 Tf 6.586 0 Td[(1Zj Zr)]TJ/F22 11.955 Tf 11.955 0 Td[(Re.2

PAGE 57

43 a bFigure5{3:IR-correctedBodeplotsforCellNA{ATi;aphaseangle,bmodulus.Theopensymbolsrepresentmeasurementstakenapproximately15hoursafterthecellpreparation.Thehalf{fullsymbolsrepresentmeasurementstakenapproxi-mately48hoursafterthecellpreparation. toyieldtheexpectedasymptoticvalueforadjustedphaseanglegivenby1=)]TJ/F15 11.955 Tf 9.299 0 Td[(90)]TJ/F22 11.955 Tf 11.955 0 Td[(.3TheelectrolyteresistancecanalsobeusedtoobtaintheadjustedmagnitudeoftheimpedanceasshowninjZj=q Zr)]TJ/F22 11.955 Tf 11.955 0 Td[(Re;est2+Zj2.4TheIR{correctedphaseangleandmagnitudeasafunctionoffrequencyarepresentedinFigure 5{3 .Thevalueoftheelectrolyteresistanceobtainedfromthismethodwasapproximately61:20:841{cm2fortherstdatasetandapproximately58:60:011{cm2forthesecond.ThesevaluesareingoodagreementwiththeonesobtainedfromtheNyquistplot.Also,thehigh-frequencyslopeinthemagnitudeoftheimpedanceplotwasequaltotheslopeusedtocalculatethevalue.

PAGE 58

44 Table5{1:Averagevaluesobtainedusinggraphicaltechniquesattheintermedi-atefrequencyrangeforCellNA{ATithroughNA{DTi.Sixreplicateswereusedtocalculatetheaverageparameters MeasurementTime/h 15 48 NA{ATi 0:260:009 0:300:012 NA{BTi 0:35)]TJ/F15 11.955 Tf 11.955 0 Td[(0:31 0:370:0002 NA{CTi 0:280:004 0:390:0002 NA{DTi 0:310:007 0:38)]TJ/F15 11.955 Tf 11.956 0 Td[(0:36 Table5{2:AveragevaluesoftheparametersobtainedusinggraphicaltechniquesatthehighfrequencyrangeforCellNA{ATitroughNA{DTi.Sixreplicateswereusedtocalculatetheaverageparameters MeasurementTime/h 15 48 /dimensionless 0:380:008 0:360:009NA{ATi Qeff/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 24:80:801 18:90:010 Re/cm2 61:20:841 58:60:011 /dimensionless 0:370:011 0:350:002NA{BTi Qeff/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 22:61:581 17:80:452 Re/cm2 61:70:048 59:80:021 /dimensionless 0:420:017 0:390:006NA{CTi Qeff/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 27:60:079 27:60:833 Re/cm2 49:20:019 44:20:036 /dimensionless 0:350:007 0:360:003NA{DTi Qeff/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 27:7)]TJ/F15 11.955 Tf 11.955 0 Td[(23:9 20:00:875 Re/cm2 41:80:016 45:60:018 ThesameprocedurewasappliedtoCellsNA{BTithroughNA{DTi.AsummaryoftheparametersobtainedfromCellsNA{ATithroughNA{DTiarepresentedinTables 5{1 and 5{2 .Ingeneral,theaverageresultofthevalueforthehighfrequencyrangewas0:370:005and0:330:040fortheintermediatefrequencyrange.Morevariabilitywasobservedinthevalueofatintermediatefrequencythanathighfrequency.Also,thevalueoftheeectivecapacitance,forthefourcells,decreasedwithtime. 5.2 SteadyStateExperimentsatIncreasingTemperaturesNyquistandBoderepresentationsoftheimpedancedataobtainedfromCellUF{CTeareshowninFigures 5{4 and 5{5 ,respectively.Thetemperaturerange

PAGE 59

45 Figure5{4:Impedance-planeorNyquistrepresentationofimpedancedataforCellUF{CTewithtemperatureasaparameter.Sixreplicatesofeachmeasurementarepresentedforeachtemperature. a bFigure5{5:BodeRepresentationofimpedancedataforphaseangleandmagnitudeforCellUF{CTe,respectively,withtemperatureasaparameter.Sixreplicatesofeachmeasurementarepresentedforeachtemperature.

PAGE 60

46 a bFigure5{6:PlotsoftheimaginaryandrealpartoftheimpedanceasfunctionsoffrequencyforCellUF{CTewithtemperatureasaparameter.Sixreplicatesofeachmeasurementarepresentedforeachtemperature. Table5{3:AveragevaluesofthelowfrequencytailslopesfromcellsUF{BTe,UF{CTe,andUF{FTe.Sixreplicateswereusedtocalculatetheaverageparameters. Temp./C UF{BTe UF{CTe UF{FTe 60 0:3330:01 0:2980:01 0:2960:01 65 0:3510:02 0:3210:02 0:3370:02 70 0:3960:13 0:3420:13 0:3510:13 75 0:3910:01 0:4110:01 0:3990:01 80 0:3890:01 0:3870:01 0:3430:01 85 0:4080:02 0:3380:02 0:3880:02 analyzedwas60Cto85Cinintervalsof5C.Siximpedancereplicatesateachtemperaturearepresentedinthisplot.TheNyquistplotrepresentsasystemwithcharacteristictimeconstantdistributedoverawide-rangeoftimescales.TherealandimaginaryrepresentationsarepresentedinFigure 5{6 .Theslopeofthelowfrequencytailincreasedwithtemperaturefromavalueof0:2980:01at60C,to0:4110:01at75C,anddecreasedto0:3380:02at85C.AsimilarbehaviorwasobservedincellsUF{BTeandUF{FTe.TheaveragevaluesofthelowfrequencytailslopesarepresentedinFigure 5{7 andTable 5{3 .TheaverageslopesoftheimaginaryrepresentationatintermediatedecreasedwithincreasingtemperatureasshowninFigure 5{8 .Athighfrequencies,theaverageslopedof

PAGE 61

47 Figure5{7:AveragevaluesofthelowfrequencytailslopesfromcellsUF{BTe,UF{CTe,andUF{FTe.Sixreplicateswereusedtocalculatetheaverageparame-ters. Table5{4:Slopesobtainedusinggraphicaltechniquesattheintermediatefre-quencyrangefromcellsUF{BTe,UF{CTe,andUF{FTe.Sixreplicateswereusedtocalculatetheaverageparameters Temp./C UF{BTe UF{CTe UF{FTe 60 0:8260:006 0:7590:020 0:6260:072 65 0:8350:003 0:7470:031 0:6630:036 70 0:7800:018 0:7200:033 0:7070:019 75 0:7860:018 0:7110:016 0:6930:022 80 0:7720:003 0:7120:028 0:6220:017 85 0:7370:011 0:6770:035 0:6110:070 theimaginaryrepresentationremainconstantuntilatemperatureof75CwasreachedandthendecreasedasshowninFigure 5{9 .TheaverageresultsobtainedforallthecellsarepresentedinTables 5{4 and 5{5 .Theslopeoftheimaginaryrepresentationathighfrequencyrangevariesfrom)]TJ/F15 11.955 Tf 9.299 0 Td[(0:690:05to)]TJ/F15 11.955 Tf 9.299 0 Td[(0:740:01,whichcorrespondstovaluesrangingfrom0:31to0:26.Thedierenceintheslopeatintermediateandhighfrequencywassmallerintheseexperiments.ThevalueofQeathighfrequencywascalculatedandtheresultingvaluesarepresentedinFigure 5{10 .Thevehighestfrequenciespointsweredisregarded

PAGE 62

48 Figure5{8:AveragevaluesoftheintermediatefrequencytailslopesfromcellsUF{BTe,UF{CTe,andUF{FTe.Sixreplicateswereusedtocalculatetheaverageparameters. Figure5{9:AveragevaluesofthehighfrequencytailslopesfromcellsUF{BTe,UF{CTe,andUF{FTe.Sixreplicateswereusedtocalculatetheaverageparame-ters.

PAGE 63

49 Figure5{10:EectiveCPEcoecientdenedbyequation 5.1 forCellUF{CTe.Sixreplicatesofeachmeasurementarepresentedforeachtemperature. becausetheywerebelievedtobecorruptedbyinstrumentartifacts.ThevaluesfortheapparentCPEcoecientprovidedinTable 5{5 representstheaverageoverthevaluesfortheremaining5highestfrequencies.ThevalueofQeseemstoreachamaximumaround75C.Thereisalsomorevariabilityinthisvalueasthetemperatureincreased.WewerenotabletocalculateameaningfulaveragevalueofQeforcellUF{FTe.ThesolutionresistanceRe,thecorrectedphaseangle,andthecorrectedim-pedancemagnitudewerecalculated.TheIR{correctedphaseangleandmagnitudearepresentedinFigure 5{11 .Theslopeofthehighfrequencyasymptotehasavalueof1)]TJ/F22 11.955 Tf 12.217 0 Td[(.ThevaluesfortheelectrolyteresistanceareprovidedinTable 5{5 .Thesevaluesareingoodagreementwiththevaluesobtainedusingtheimpedanceplanerepresentation. 5.3 TransientExperimentsatConstantTemperatureImpedance-planeandBoderepresentationsobtainedfromCellUF{DTrareshowninFigures 5{12 and 5{13 ,respectively.Thetemperaturewasheldconstantat70Cthroughouttheexperiment.Impedancespectraweretakenevery30

PAGE 64

50 a bFigure5{11:IR-correctedBodeplotsforCellUF{CTe;aphaseangle,bmodu-lus.Sixreplicatesofeachmeasurementarepresentedforeachtemperature. Table5{5:AveragevaluesoftheparametersobtainedusinggraphicaltechniquesatthehighfrequencyrangefromcellsUF{BTe,UF{CTe,andUF{FTe.Sixreplicateswereusedtocalculatetheaverageparameters Temp./C UF{BTe UF{CTe UF{FTe 0:240:01 0:260:01 0:320:0160 Qe/M)]TJ/F21 7.97 Tf 6.587 0 Td[(1cm)]TJ/F21 7.97 Tf 6.586 0 Td[(2s1)]TJ/F23 7.97 Tf 6.587 0 Td[( 11:601:15 12:981:54 24:702:05 Re/cm2 37:182:55 37:351:22 50:381:23 0:250:02 0:270:04 0:270:0565 Qe/M)]TJ/F21 7.97 Tf 6.587 0 Td[(1cm)]TJ/F21 7.97 Tf 6.586 0 Td[(2s1)]TJ/F23 7.97 Tf 6.587 0 Td[( 14:062:23 16:695:33 21:6911:76 Re/cm2 30:340:67 32:251:60 38:961:80 0:240:1 0:270:03 0:180:1070 Qe/M)]TJ/F21 7.97 Tf 6.587 0 Td[(1cm)]TJ/F21 7.97 Tf 6.586 0 Td[(2s1)]TJ/F23 7.97 Tf 6.587 0 Td[( 13:271:87 19:434:94 15:2720:94 Re/cm2 24:291:05 27:771:11 32:603:57 0:220:01 0:310:02 0:200:0175 Qe/M)]TJ/F21 7.97 Tf 6.587 0 Td[(1cm)]TJ/F21 7.97 Tf 6.586 0 Td[(2s1)]TJ/F23 7.97 Tf 6.587 0 Td[( 11:781:20 29:096:74 23:3026:85 Re/cm2 20:041:22 18:850:97 26:211:53 0:210:01 0:310:05 0:150:0480 Qe/M)]TJ/F21 7.97 Tf 6.587 0 Td[(1cm)]TJ/F21 7.97 Tf 6.586 0 Td[(2s1)]TJ/F23 7.97 Tf 6.587 0 Td[( 10:021:22 32:9015:72 10:814:16 Re/cm2 13:160:50 15:152:69 24:641:23 0:260:01 0:250:10 0:370:0385 Qe/M)]TJ/F21 7.97 Tf 6.587 0 Td[(1cm)]TJ/F21 7.97 Tf 6.586 0 Td[(2s1)]TJ/F23 7.97 Tf 6.587 0 Td[( 13:681:34 26:0930:94 222:12126:0 Re/cm2 11:140:36 14:160:78 19:402:08

PAGE 65

51 Figure5{12:Impedance-planeorNyquistrepresentationofimpedancedataforCellUF{DTrwithtemperatureasaparameter.Impedancespectraweretakenevery30secondsfor48hours,ittookapproximatelytwohourstocompleteonespectrum. a bFigure5{13:BodeRepresentationofimpedancedataforphaseangleandmagni-tudeforCellUF{DTr,respectively.Impedancespectraweretakenevery30secondsfor48hours,ittookapproximatelytwohourstocompleteonespectrum.

PAGE 66

52 a bFigure5{14:PlotsoftheimaginaryandrealpartoftheimpedanceasfunctionsoffrequencyforCellUF{DTr.Impedancespectraweretakenevery30secondsfor48hours,ittookapproximatelytwohourstocompleteonespectrum. secondsfor48hours,ittookapproximatelytwohourstocompleteonespectrum.TheNyquistplotrevealsadepressedsemicircle.TherealandimaginaryrepresentationsarepresentedinFigure 5{14 .Theslopeofthelowfrequencytaildecreasedwithtimefromavalueof0:350:02to0:240:01.However,thelowfrequencytailslopesforcellUF{ETrvariedrandomlyfrom0:380:04to0:300:03.Thisdiscrepancycanbeattributedtothefactthatthemeltingpointoftheelectrolyteisapproximately60C,thusthereismoreinstabilityinthesystematthattemperature.TheaveragevaluesofthelowfrequencytailslopesarepresentedinTable 5{6 .Theslopesoftheimaginaryrepresentationatintermediatefrequenciesdecreasewithtimeandtheslopeathighfrequenciesoscillatewithtimebetween)]TJ/F15 11.955 Tf 9.299 0 Td[(0:74and)]TJ/F15 11.955 Tf 9.299 0 Td[(0:79,whichcorrespondtovaluesrangingfrom0:26to0:21.TheaverageresultsobtainedfromCellUF{DTrandUF{ETrarepresentedinTable 5{7 and 5{8 .Inthisexperimenttheslopeatintermediateandhighfrequencyweredierent,suggestingthatmorethanoneCPEareaectingthesystem.Thisresultisinagreementwithpreviousexperiments.

PAGE 67

53 Table5{6:ValuesofthelowfrequencytailslopesfromCellUF{DTrandUF{ETr. No. UF{DTr UF{ETr 2 0:350:02 0:350:023 0:330:02 0:370:044 0:330:02 0:350:025 0:330:01 0:370:026 0:310:01 0:320:047 0:320:01 0:360:038 0:310:01 0:380:049 0:300:01 0:340:0310 0:280:01 0:300:0319 0:240:01 0:360:06 Table5{7:ValuesofthelowandintermediatefrequencytailslopesobtainedfromCellUF{DTrandUF{ETr. No. UF{DTr UF{ETr 2 0:830:01 0:790:013 0:820:01 0:740:014 0:810:01 0:720:015 0:810:01 0:690:026 0:790:01 0:670:027 0:790:01 0:650:038 0:770:02 0:640:039 0:780:01 0:620:0310 0:770:01 0:610:0319 0:720:02 0:420:05

PAGE 68

54 Figure5{15:EectiveCPEcoecientdenedbyequation 5.1 forCellUF{DTr. ThevalueofQeathighfrequencywascalculatedandtheresultingvaluesarepresentedinFigure 5{15 .ThevaluefortheapparentCPEcoecientprovidedinTable 5{8 representstheaverageoverthehighestfrequenciesvalues.ThesolutionresistanceRe,thecorrectedphaseangle,andthecorrectedim-pedancemagnitudewerecalculated.TheIR{correctedphaseangleandmagnitudearepresentedinFigure 5{16 .Theslopeofthehighfrequencyasymptotehasavalueof1)]TJ/F22 11.955 Tf 12.217 0 Td[(.ThevaluesfortheelectrolyteresistanceareprovidedinTable 5{8 .Thesevaluesareingoodagreementwiththevaluesobtainedusingtheimpedanceplanerepresentation. 5.4 ConclusionsTheplotspresentedhereprovideusefulguidestomodeldevelopment.Thedierenceintheslopesatintermediateandhighfrequenciesintheplotsoftheimaginarycomponentoftheimpedancesuggestthatmorethanonecapacitivetimeconstantcanbeidentiedinthesystem.Thehigh-frequencycapacitiveloopcanbeattributedtoadielectricbehavioroftheelectrolyteortoreactionsatthelithiummetalsurface.Thismeansthatanequivalentcircuitsuchastheoneproposedby

PAGE 69

55 Table5{8:AveragevaluesoftheparametersobtainedusinggraphicaltechniquesatthehighfrequencyrangeforCellUF{DTr. No. UF{DTr UF{ETr 0:26 0:322 Qe/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 12:52 21:39 Re/cm2 35:48 60:24 0:26 0:303 Qe/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 13:95 20:62 Re/cm2 33:36 68:65 0:24 0:294 Qe/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 12:01 23:16 Re/cm2 32:90 60:61 0:21 0:295 Qe/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 9:40 24:90 Re/cm2 33:05 59:66 0:23 0:296 Qe/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 11:80 31:89 Re/cm2 33:43 63:53 0:22 0:307 Qe/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 10:87 23:74 Re/cm2 33:23 56:37 0:22 0:298 Qe/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 11:25 26:59 Re/cm2 32:60 57:14 0:23 0:309 Qe/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 13:08 23:86 Re/cm2 31:80 60:56 0:19 0:3110 Qe/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 9:42 25:76 Re/cm2 33:43 54:57 0:29 0:3019 Qe/M)]TJ/F21 7.97 Tf 6.586 0 Td[(1cm)]TJ/F21 7.97 Tf 6.587 0 Td[(2s1)]TJ/F23 7.97 Tf 6.586 0 Td[( 28:05 39:37 Re/cm2 26:59 56:12

PAGE 70

56 a b Figure5{16:IR-correctedBodeplotsforCellUF{DTr;aphaseangle,bmodulus. Bouchetetal.maydescribethesystemcorrectlyifthecapacityloopisattributedtoadielectricbehavioroftheelectrolyte.Thedecreaseoftheimpedanceresponsewithtimeisconsistentwithdissolu-tionofnativeoxidelayersonthelithiumsurface,asproposedintheliterature.Theresultsshowedthatthetimerequiredforcompletedissolutionoftheoxidelayerswereontheorderofseveraldays.Asthetemperatureofthecellswasincreasedtheimpedanceresponsedecreased.Thisresultisconsistentwitheitherhypothesis,thehigh-frequencycapacitiveloopisduetoreactionsatthelithiummetalsurfaceorduetoadielectricbehavioroftheelectrolyte.

PAGE 71

CHAPTER6BLOCKINGELECTRODEANALYSISTheimpedancedatapresentedinChapter 5 wereobtainedusingtwolithiumelectrodesinasymmetricconguration.Toexplorethehypothesisthatthehigh-frequencycapacitiveloopseenintheimpedancespectrawereassociatedwiththedielectricresponseoftheelectrolyte,asecondsetofexperimentswasperformedusingstainlesssteelelectrodes,againinasymmetricconguration. 6.1 ExperimentalResultsImpedancemeasurementswereobtainedatfourdierenttemperatures.Theresultsobtainedat25C,80C,90C,and100CarepresentedinFigure 6{1 withtemperatureasaparameter.Theresultsindicatethatthesteelelectrodesactedasblockingelectrodeswithalocalcapacitydispersion.Thelocalcapacitydispersioncanbecausedbyroughnessontheelectrodesurface.Atlowfrequencies,theimpedanceresponseofthesteelelectrodescanbeexpressedasZf=Re+1 j2fQ.1wheretheparametersandQareconstants.When=1,Qrepresentsthecapacitanceofthedoublelayer.When6=1,thesystemshowsbehaviorthathasbeenattributedtosurfaceheterogeneity.Athighfrequencies,thedatashowadispersionthatmaybeassociatedwithanon-uniformcurrentandpotentialdistribution. 6.2 GraphicalAnalysisHuangetal.67developedamathematicalmodelforablockingdiskelectrodeembeddedinaninniteinsulatingplane.Themathematicalmodelinvolvessolving 57

PAGE 72

58 Figure6{1:CompleximpedanceplaneorNyquistrepresentationfortheresponseofa430StainlessSteeldiskinPEO/LiTSFIelectrolytewithconcentrationasaparameter.Sixreplicatesofeachmeasurementarepresentedforeachtemperature. Laplace'sequationforpotentialinthefrequencydomain.TheirworkprovidesaguideforanalysisofimpedancedatathatissimilartotheoneexplainedinChapter 5 .TherealandimaginaryrepresentationsarepresentedinFigure 6{2 asfunc-tionsoftemperature.Thedierenceintheresponsebetweentheexperimentsdoneat25Candtheonesdoneat80C,90C,and100Ccanbeattributedtothechangeinconductivityassociatedwiththemeltingpointoftheelectrolyte,approximately60C.Avalueforthesolutionconductivitywasextractedfromtherealpartoftheimpedancefollowing=4r0limf!1Zr)]TJ/F21 7.97 Tf 6.586 0 Td[(1.2Therealpartsoftheimpedancecorrespondingtothevaluesobtainedfromequation 6.2 arepresentedinFigure 6{2a asdashedlines.TheresultingvaluesforconductivityarepresentedinTable 6{1

PAGE 73

59 a bFigure6{2:PlotsoftheimaginaryandrealpartoftheimpedanceasfunctionsoffrequencyforCellNA-BBwithtemperatureasaparameter.Sixreplicatesofeachmeasurementarepresentedforeachtemperature.

PAGE 74

60 Figure6{3:EectiveCPEcoecientdenedbyequation 6.3 forCellNA-BB.Sixreplicatesofeachmeasurementarepresentedforeachtemperature. Atlowfrequencies,theslopeoftheimaginarypartoftheimpedancehasavalueof)]TJ/F22 11.955 Tf 9.298 0 Td[(.ThevaluesofobtainedfromFigure 6{2b arealsopresentedinTable 6{1 .Forablockingelectrode,thevalueofQcanbeobtainedfromthelow-frequencyasymptotewherethecurrentdistributionassociatedwiththeelectrodegeometrydoesnotinuencetheimpedanceresponse.ThevalueofQwascalcu-latedusingQ=)]TJ/F22 11.955 Tf 10.494 8.088 Td[(sin=2 Zjf.3andtheresultingvaluesarepresentedinFigure 6{3 .TheextractedvaluesforQaremarkedbydashedlinesinFigure 6{3 andarereportedinTable 6{1 .Asthetemperatureincreases,thevaluesforQalsoincrease.At100CthevaluesforQincreasedwithtime.

PAGE 75

61 Table6{1:ParametervaluesextractedfromFigures 6{2 and 6{3 forCellNA-BB. Temp./C Q/M)]TJ/F21 7.97 Tf 6.587 0 Td[(1cm)]TJ/F21 7.97 Tf 6.586 0 Td[(2s /mhocm)]TJ/F21 7.97 Tf 6.586 0 Td[(1 25 0:8240:003 0:400:0166 2:9710)]TJ/F21 7.97 Tf 6.586 0 Td[(059:08910)]TJ/F21 7.97 Tf 6.586 0 Td[(07 80 0:7040:008 8:450:132 0:0188:53410)]TJ/F21 7.97 Tf 6.587 0 Td[(05 90 0:7280:014 11:40:617 0:0251:01110)]TJ/F21 7.97 Tf 6.587 0 Td[(04 0:749 14:3 0:027 0:721 14:8 0:028 0:701 16:8 0:028100 0:726 18:9 0:026 0:693 19:3 0:026 0:708 19:3 0:027 6.3 ComparisontoDiskElectrodeThedimensionlessimaginarypartoftheimpedanceZjr0ispresentedinFigure 6{4 asafunctionofdimensionlessfrequencyK=Q!r0 .4TheresultspresentedinFigure 6{4 canbecomparedtotheresultsobtainedbyHuangetal.67showninFigure 6{5 fordierentvaluesof.AsuperpositionofthedataisevidentinbothFigures 6{4 and 6{5 atlowfrequency.Thissuperpositionistypicalofblockingelectrodes.Theslopeatlowfrequencyhasavalueof)]TJ/F15 11.955 Tf 9.299 0 Td[(1becausetheinuenceoftheCPEparameterisincorporatedinthedenitionofthedimensionlessfrequencyinequation 6.4 .Achangeintheslopefromavalueof)]TJ/F15 11.955 Tf 9.299 0 Td[(1appearsinFigure 6{4 atfrequencieshigherthanK=610)]TJ/F21 7.97 Tf 6.586 0 Td[(3.AsimilarchangeinslopeappearsinFigure 6{5 atfrequenciesK>1.Forthediskelectrode,thechangeinslopeatfrequenciesK>1isduetothenon-uniformcurrentdistributionassociatedwiththediskgeometry.Thegeometryofthecoincellsusedforthepresentworkissignicantlydierentthanthatofadiskelectrodewithacounterelectrodeinnitelyfaraway.Thereisapotentialforanon-uniformcurrentdistributioninthebuttoncell,butthe

PAGE 76

62 Figure6{4:DimensionlessanalysisfortheimpedanceresponseofastainlesssteeldiskinPEO/LiTFSIelectrolyteswithtemperatureasaparameter. Figure6{5:Dimensionlessimaginarypartoftheimpedanceasafunctionofdimen-sionlessfrequencyforadiskelectrodewithlocalcapacitancedispersion.Theoreti-calresultderivedbyHuangetal..

PAGE 77

63 geometryoftheelectrode-insulatorinterfaceisnotwelldened.ThechangeofslopeatK=610)]TJ/F21 7.97 Tf 6.586 0 Td[(3maybeassociatedwithanon-uniformcurrentdistribution. 6.4 ConclusionsTheimpedanceresponseofthesymmetricbuttoncellwithsteelelectrodeswasthatofablockingelectrodewithlocalCPEbehaviorassociatedwithacapacitydispersion.Thehigh-frequencycapacitiveloopseenintheresultsofthesymmetriclithiumcellwerenotevidentintheresultsfromthecellwithsteelelectrodes.Thus,thehigh-frequencycapacitiveloopseenintheimpedancespectraofChapter 5 arenotassociatedwiththedielectricresponseoftheelectrolyte.Theseloopsmustbedescribedinsteadbyamodelthattakesintoaccountthereactionsatthelithiumelectrode.WhenconductivityandtheCPEparametersQandaretakenintoaccount,theimaginaryimpedanceresponsesforallthedatacollectedwiththesymmetricstainlesssteelcellaresuperposedatlowfrequencies.Thisresultwasseenevenwhenthelow-temperature25Cdatawereincluded,forwhichtheelectrolytewassolidandtheconductivitywasthreeordersofmagnitudelowerthanforthemeltedelectrolyte.ThissuperpositionprovidesevidencethatthesystembehaviorisindeedthatofablockingelectrodeandthatnoFaradaicreactionneededtobeconsidered.Athighfrequencies,whenthecontributionofthenon-uniformcurrentdistributionismoreimportant,theexperimentalresultsdeviatedfromthetheoreticalbehavior.However,thisdeviationwasanticipatedsincethegeometryoftheexperimentalsystemdiersfromthegeometryofthesystemusedtodevelopthetheoreticalresults.

PAGE 78

CHAPTER7MODELDEVELOPMENTTheresultspresentedinChapter 6 demonstratedthatthehigh-frequencycapacityloopobservedintheexperimentalresultswereduetoreactionsatthelithiumelectrodes.TheelectriccircuitmodelproposedbyBouchetetal.doesnotdescribethesystemcorrectlybecauseheattributedthecapacitylooptoadielectricbehavioroftheelectrolyte.Stillhewasabletotadequatelyhisexperimentaldata.Thisisthelimitationoftheelectriccircuits,morethanonecircuitnetworkcanbeconstructedwhichwouldprovideequivalentstatisticallyvalidts;andeachofthesemodelscanarisefromextremelydierentinterpretations.Physico-chemicalmodelsaremoreecientforinterpretationofthephysicalpropertiesandthekineticprocessesofthesystem.However,thephysico-chemicalmodelshavethedisadvantagethattheycannotbeeasilyregressedtotheexperimentaldata.Inthepresentworkwedevelopedamodelthattakeintoaccountthephysicsandchemistryofthesystemandcanbeuseforregressionanalysis.Todevelopsuchmodelitwasnecessarytohavecomplimentarysupportingin-formationconcerningthephysicsandchemistryofthesystem.Microscopeimages,scanningelectronmicroscopeSEMmicrographsandinformationpublishedintheliteraturewereusedtoguidethemodeldevelopment. 7.1 ChemistryItiswellacceptedintheliteraturethatthereactionmechanismforthelithiumelectrodesisextremelycomplexandmustberepresentedbyalargenumberofreactions.10,68,69Theelectrolytecompositionandthepresenceofadditivesandcontaminantsaectthemechanismsoftransport.AccordingtoAurbachand 64

PAGE 79

65 Table7{1:Reactionmechanismsofcommonlithiumbatterycontaminants.Waterwastheonlycontaminantconsideredinthiswork. Contaminant ReactionMechanisms CO2 CO2+e)]TJ/F15 11.955 Tf 10.987 -4.338 Td[(+Li+!CO2Li CO2Li+CO2!O=C-OCO2Li O=C-OCO2Li+e)]TJ/F15 11.955 Tf 10.987 -4.338 Td[(+Li!CO"+Li2CO3# 2LiOH#+CO2!Li2CO3# Li2O#+CO2!Li2CO3# H2O H2O+e)]TJ/F15 11.955 Tf 10.987 -4.339 Td[(+Li+!LiOH+1=2H2 LiOH+Li++e)]TJ/F25 11.955 Tf 10.987 -4.339 Td[(!Li2O+1/2H2 H+e)]TJ/F15 11.955 Tf 10.987 -4.339 Td[(+Li+!LiH N2 N2+6e)]TJ/F15 11.955 Tf 10.986 -4.339 Td[(+6Li+!2Li3N O2 O2+e)]TJ/F15 11.955 Tf 10.987 -4.338 Td[(+Li+!LiO2 LiO2+e)]TJ/F15 11.955 Tf 10.986 -4.338 Td[(+Li+!Li2O2 Li2O2+2e)]TJ/F15 11.955 Tf 10.987 -4.339 Td[(+2Li+!2Li2O HF HF+e)]TJ/F15 11.955 Tf 10.986 -4.338 Td[(+Li+!LiF+1=2H2 Tablereproducedfrom:G.NazriandG.Pistoia,LithiumBatteries:ScienceandTechnology",Norwell,MA:KluwerAcademicPublishers,2004. Schechter10thepossiblereactionsbetweenLi+andtheLiTFSIsaltareLiNSO2CF32+ne)]TJ/F15 11.955 Tf 9.742 -4.936 Td[(+nLi+!Li3N+Li2S2O4+LiF+C2FxLiy.1andLiNSO2CF32+2e)]TJ/F15 11.955 Tf 9.742 -4.937 Td[(+2Li+!Li2NSO2CF3+CF3SO2Li.2andthecommoncontaminantspresentinthistypeofcellincludeCO2,H2O,N2,O2,andHF.10,68,69ThereactionmechanismsofthecommoncontaminantsaresummarizedinTable 7{1 .Astheexactcomposition,sizeanddistributionoftheparticlesthatcomposedtheSEIseeSection 2.2.4 areunknown,asimpliedsystemwasused.Thereactionsconsideredforthesimpliedsystemweretheoxidation/reductionreactionoflithium,LiLi++e)]TJ/F15 11.955 Tf 161.954 -4.936 Td[(.3

PAGE 80

66 Figure7{1:Impedance-planeorNyquistrepresentationofimpedancedataforCellUF-DTr.Verylargeimpedanceatlowfrequenciescanbeobservedsuggestingthatadisconnectionofthesystemwhichmaybeattributedtogasevolution. andthereactionsduetothepresenceofwaterasacontaminant,H2O+e)]TJ/F15 11.955 Tf 9.741 -4.936 Td[(+Li+!LiOH+1=2H2.4LiOH+Li++e)]TJ/F25 11.955 Tf 10.405 -4.936 Td[(!Li2O+1=2H2.5IthasbeenreportedintheliteraturethatthesecellscontaintracesofwaterandtheformationofLiOHisexpectedatthelithiuminterface,whichsupporttheselectionofequations 7.4 and 7.5 aspartoftheproposedreactionmechanism.70Also,theimpedancedataandbrokensealsprovidedevidenceofgasevolution.AnimpedancespectrumofcellUF-DTrtakenapproximately14hoursaftertheexperimentstartedispresentedinFigure 7{1 .Theimpedanceresponseofthecellwasasexpectedforasystemwithdistributedtimeconstants,asshownintheinsert.However,alargeimpedancewasobservedatlowfrequencies.Thisisusuallyattributedtoadisconnectionofthesystem,whichmaybeassociatedwithgasevolution.Thisbehaviorwasobservedinmorethanonecell.

PAGE 81

67 Figure7{2:CellUF-ATelithiumelectrodesurface.Thesurfaceoftheelectrodewasalmostcompletelycoveredwithawhitetoblueoxidelayer. 7.2 PhysicsTheformationofpassivatinglmsonthelithiumsurfacehasbeenreportedextensivelyintheliterature.10,18{20PeledwasthersttointroduceandnametheSEI.18,19AccordingtoPeled,theSEIisaheterogenouslm,consistingofamosaicofnumerousindividualparticlesofdierentchemicalcompositionbeinginpartialcontactwitheachotheratthegrainboundaries,asexplainedinSection 2.2.4 .Someofthecellsusedinthisresearchwereopenedaftertheexperiments.MicroscopicimagesofcellsUF-ATeandUF-BTearepresentedinFigures 7{2 and 7{3 .Awhitetoblueoxidelayerwasobservedonthelithiumsurface.CellUF-ATe

PAGE 82

68 a bFigure7{3:CellUF-BTelithiumelectrodesurface.Theedgesoftheelectrodewerecoveredwithawhitetoblueoxidelayer,whiletheinnersurfacekeptamuchbrightermetalliclustre.

PAGE 83

69 wasalmostcompletelycoveredwiththelayer,whileincellUF-BTe,theoxidelayerwasobservedintheperimeterofthecells,asshowninFigure 7{2 and 7{3a ,respectively.Formostofthecells,theoxidelayerwasmoreevidentonthelithiumedgethaninsidethebattery.Thisbehaviorhasbeenreportedintheliterature70andisevidenceofanon-uniformcurrentdistribution.SEMmicrographsweretakentogetanunderstandingofthestructureofthislayer.Thelayerisanheterogenouslmthatappearstobeamosaicofindividualparticles,asshowninFigures 7{4 and 7{5 .Unfortunately,itwasnotpossibletodeterminethecompositionofthelayer.Severaloptionswereconsideredaspossiblemechanismsoftransport.Allofthemechanismsconsideredtakeintoaccountthepresenceofthepassivatinglmsattheelectrodessurface,asdescribedintheliterature.10,18{20AccordingtoNaudinetal.,29commerciallithiumfoilshaveanativelayerthatiscomposedofapproximately10{100nmLi2Oinsidelayerand1{20nmofLi2CO3/LiOHoutsidelayer,asshowninFigure 7{6 .TherstmechanismispresentedinFigure 7{7 .Inthismodel,theonlyreactiontakingplaceattheelectrodesurfaceistheoxidation/reductionreactionofLi.TheLiOHandLi2Oformationreactionstakeplaceonthesurfaceofthepassivatinglayer.Onthesecondmechanismitwasassumedthatthelithiumoxidation/reductionreactiontakeplaceatthelithiummetalsurfaceonly,buttheLiOHandLi2Oformationreactionscantakeplaceinboththelithiummetalsurfaceandtheoxidelayersurface.AschematicdiagramofthesystemisshowninFigure 7{8 7.3 ModelDevelopmentTheprocedureexplainedinSection 3.5 wasusedtodeveloptheimpedanceresponsemodel.InthissectionthemodelforthemechanismsshowninFigure 7{8

PAGE 84

70 a bFigure7{4:ScanningelectronmicrographsofCellUF-ATelithiumelectrodesur-face.aMagnication=35.bMagnication=200.

PAGE 85

71 Figure7{5:ScanningelectronmicrographsofCellUF-BTelithiumelectrodesur-face.Magnication=350. Figure7{6:Commerciallithiumfoilnativelayercomposition,asdescribedbyNaudinetal.29

PAGE 86

72 Figure7{7:Schematicofthesimpliedsystem.Twosurfaceswereidentiedinwhichthereactionstakeplace.Thelithiumoxidation/reductionreactiontakeplaceatthelithiummetalsurfaceandtheLiOHandLi2Oformationreactionstakeplaceintheoxidelayersurface.

PAGE 87

73 Figure7{8:Schematicofthesimpliedsystem.Twosurfaceswereidentiedinwhichthereactionstakeplace.Thelithiumoxidation/reductionreactiontakeplaceatthelithiummetalsurfaceonly,buttheLiOHandLi2Oformationreactionscantakeplaceinboththelithiummetalsurfaceandtheoxidelayersurface.

PAGE 88

74 isexplainedindetail.OnlythenalexpressionforthemechanismshowninFigure 7{7 ispresented. 7.3.1 SteadyStateTheequationforthefaradaiccurrentdensity,ii;f,canbeexpressedintermsofasteadytime-independentvalueandanoscillatingvalueforeachofthethreereactionsconsideredinthesystem.Thesteadystatevaluesare ia;Li;0=Fka;Li;0expba;Li;0V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLi7.6and ic;Li;0=)]TJ/F22 11.955 Tf 9.298 0 Td[(Fkc;Li;0 cLi+;Li;0exp)]TJ/F22 11.955 Tf 9.299 0 Td[(bc;Li;0V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLi;0.7forthelithiumoxidation/reductionreactiontakingplaceattheelectrodesurface.WhenLiOHandtheLi2Oareformedinthelithiummetalthesteadystatecurrentcontributionsare iLiOH;0=)]TJ/F22 11.955 Tf 9.299 0 Td[(Fkc;LiOH;0 cH2O;0exp)]TJ/F22 11.955 Tf 9.299 0 Td[(bc;LiOH;0V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLiOH;0.8and iLi2O;0=)]TJ/F22 11.955 Tf 9.298 0 Td[(Fkc;Li2O;0exp)]TJ/F22 11.955 Tf 9.298 0 Td[(bc;Li2O;0V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLi2O;0.9Thesteadystatecurrentsattheoxidelayersurfaceare iLiOH;=)]TJ/F22 11.955 Tf 9.298 0 Td[(Fkc;LiOH; cLi+;LiOH; cH2O;exp)]TJ/F22 11.955 Tf 9.298 0 Td[(bc;LiOH;V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLiOH;.10and iLi2O;=)]TJ/F22 11.955 Tf 9.298 0 Td[(Fkc;Li2O; cLi+;Li2O;exp)]TJ/F22 11.955 Tf 9.298 0 Td[(bc;Li2O;V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLi2O;7.11fortheLiOHandLi2Oformationreactions,respectively.Thesubscript0andrepresentthelithiummetalsurfaceandtheoxidelayersurface,respectively.Noticethatthediusionoflithiumforthereactionsthatinvolvewaterisconsideredwhen

PAGE 89

75 thereactionstakeplaceintheoxidelayerbutisnotconsideredwhenthereactionstakeplaceinthelithiummetalsurface. 7.3.2 SinusoidalSteadyStateTheoscillatingcomponentofthecurrentwasderivedusing,eii;f=@f @Vci;keV+X@f @ci;0V;cj6=i;keci;0+X@f @kV;ci;l6=kek.12where~Vispotential,~ci;0isconcentrationand~kissurfacecoverage.Inthismodelthesurfacecoveragewasassumedtobeconstant,sothecurrentisonlydependentonpotentialandconcentrationoflithiumandwater.Solvingequation 7.12 withrespecttoeachofthesteady{statecurrentcontributionswouldgivetheoscillatingcurrentcontributionsofthereactionsconsideredinthesystem.Theseexpressionsareeia;Li;0=R)]TJ/F21 7.97 Tf 6.587 0 Td[(1a;Li;0eV.13where,R)]TJ/F21 7.97 Tf 6.586 0 Td[(1a;Li;0=Fba;Li;0ka;Li;0expba;Li;0V)]TJ/F22 11.955 Tf 11.956 0 Td[(VLi;0.14andeic;Li;0=R)]TJ/F21 7.97 Tf 6.586 0 Td[(1c;Li;0eV+)]TJ/F22 11.955 Tf 9.298 0 Td[(Fkc;Li;0exp)]TJ/F22 11.955 Tf 9.299 0 Td[(bc;Li;0V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLi;0ecLi+;Li;0.15where,R)]TJ/F21 7.97 Tf 6.587 0 Td[(1c;Li;0=Fbc;Li;0kc;Li;0 cLi+;Li;0exp)]TJ/F22 11.955 Tf 9.298 0 Td[(bc;Li;0V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLi;0.16forthelithiumoxidation/reductionreactiontakingplaceattheelectrodesurface.TheoscillatingcurrentcontributionsoftheLiOHandLi2OformationreactionsinthelithiummetalareeiLiOH;0=R)]TJ/F21 7.97 Tf 6.587 0 Td[(1LiOH;0eV+)]TJ/F22 11.955 Tf 9.298 0 Td[(Fkc;LiOH;0exp)]TJ/F22 11.955 Tf 9.298 0 Td[(bc;LiOH;0V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLiOH;0ecH2O;0.17where,R)]TJ/F21 7.97 Tf 6.586 0 Td[(1LiOH;0=Fbc;LiOH;0kc;LiOH;0 cH2O;0exp)]TJ/F22 11.955 Tf 9.298 0 Td[(bc;LiOH;0V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLiOH;0.18

PAGE 90

76 andeiLi2O;0=R)]TJ/F21 7.97 Tf 6.587 0 Td[(1Li2O;0eV.19where,R)]TJ/F21 7.97 Tf 6.587 0 Td[(1Li2O;0=Fbc;Li2O;0kc;Li2O;0exp)]TJ/F22 11.955 Tf 9.299 0 Td[(bc;Li2O;0V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLi2O;0:.20Whenthereactionsoccurintheoxidelayer,theoscillatingcurrentcontribu-tionsareeiLiOH;=R)]TJ/F21 7.97 Tf 6.587 0 Td[(1LiOH;eV+)]TJ/F22 11.955 Tf 9.298 0 Td[(Fkc;LiOH; cH2O;exp)]TJ/F22 11.955 Tf 9.299 0 Td[(bc;LiOH;V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLiOH;ecLi+;LiOH;+)]TJ/F22 11.955 Tf 9.299 0 Td[(Fkc;LiOH; cLi+;LiOH;exp)]TJ/F22 11.955 Tf 9.299 0 Td[(bc;LiOH;V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLiOH;ecH2O;.21where,R)]TJ/F21 7.97 Tf 6.586 0 Td[(1LiOH;=Fbc;LiOH;kc;LiOH; cLi+;LiOH; cH2O;exp)]TJ/F22 11.955 Tf 9.298 0 Td[(bc;LiOH;V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLiOH;.22andeiLi2O;=R)]TJ/F21 7.97 Tf 6.586 0 Td[(1Li2O;eV+)]TJ/F22 11.955 Tf 9.299 0 Td[(Fkc;Li2O;exp)]TJ/F22 11.955 Tf 9.299 0 Td[(bc;Li2O;V)]TJ/F22 11.955 Tf 11.956 0 Td[(VLi2O;ecLi+;Li2O;.23where,R)]TJ/F21 7.97 Tf 6.586 0 Td[(1Li2O;=Fbc;Li2O;kc;Li2O; cLi+;Li2O;exp)]TJ/F22 11.955 Tf 9.299 0 Td[(bc;Li2O;V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLi2O;.24 7.3.3 MassTransferAsecondequationwasneededtocorrelateconcentrationoflithiumandwatertothecorrespondingcurrentcontributions ii=)]TJ/F22 11.955 Tf 9.299 0 Td[(niFDid ci dyy=0.25Thisequationcanbeexpressedintermsoftheoscillatingcontributionsaseii=)]TJ/F22 11.955 Tf 9.299 0 Td[(niFDideci dyy=0.26

PAGE 91

77 Itisconvenienttowriteequation 7.26 intermsofthedimensionlesssolutiontotheappropriatediusionorconvectiveequationeii=)]TJ/F22 11.955 Tf 9.299 0 Td[(niFDieCi;0 ie0i.27wherei=eci=eci;0.iwasevaluatedassumingdiusionthroughaninnitelayeras1 e0i=)]TJ/F15 11.955 Tf 22.553 8.088 Td[(1 q j! Di.28Thedenitionofiwasthensubstitutedintoequation 7.27 toobtaintherelationshipbetweenconcentrationandcurrent,eii=niFDieci;0 ir j! Di.29Thenalcurrentexpressionasafunctionofpotentialforthereactionsthattakeplaceinthelithiummetalsurfaceareeia;Li;0=Ra;Li;0)]TJ/F21 7.97 Tf 6.587 0 Td[(1eV.30andeic;Li;0=Rc;Li;0+zd;Li;0)]TJ/F21 7.97 Tf 6.586 0 Td[(1eV.31where,zd;Li;0=Ra;Li;0kc;LiOH;0exp)]TJ/F22 11.955 Tf 9.298 0 Td[(bc;LiOH;0V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLi;0Li;0 DLi+q j! DLi+.32forthelithiumoxidation/reductionreactionandeiLiOH;0=RLiOH;0+zd;H2O;0)]TJ/F21 7.97 Tf 6.587 0 Td[(1eV.33where,zd;H2O;0=RLiOH;0kc;LiOH;0exp)]TJ/F22 11.955 Tf 9.298 0 Td[(bc;LiOH;0V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLiOH;0H2O DH2Oq j! DH2O.34

PAGE 92

78 andeiLi2O;0=RLi2O;0)]TJ/F21 7.97 Tf 6.587 0 Td[(1eV.35fortheLiOHandLi2Oformationreactions.ForthereactionsthattakeplaceintheoxidelayersurfacethenalcurrentwasexpressedaseiLiOH;=RLiOH;+zd;H2O;+zLi;2)]TJ/F21 7.97 Tf 6.587 0 Td[(1eV.36wherezH2O;=RLiOH;kc;LiOH; cLi+;LiOH;exp)]TJ/F22 11.955 Tf 9.299 0 Td[(bc;LiOH;V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLiOH;H2O; DH2Oq j! DH2O.37andzLi;2=RLiOH;kc;LiOH; cH2O;LiOH;exp)]TJ/F22 11.955 Tf 9.298 0 Td[(bc;LiOH;V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLiOH;Li+;2 DLi+q j! DLi+.38fortheLiOHformationreactionsandfortheLi2OformationreactionsthenalcurrentwasexpressedaseiLi2O;=RLi2O;+zd;Li;3)]TJ/F21 7.97 Tf 6.587 0 Td[(1eV.39wherezd;Li;3=RLi2O;kc;Li2O;exp)]TJ/F22 11.955 Tf 9.299 0 Td[(bc;Li2O;V)]TJ/F22 11.955 Tf 11.955 0 Td[(VLi2O;Li+;3 DLi+q j! DLi+.40Thetotalcurrentdensitywasexpressedintermsofthefaradaicandchargingcurrentsaseii=eii;f+j!CdeV.41wherethefaradaiccurrenteifforthereactiontakingplaceinthelithiumsurfaceisgivenbyei0;f=eia;Li;0+eic;Li;0+eiLiOH;0+eiLi2O;0;.42

PAGE 93

79 eifforthereactiontakingplaceintheoxidelayersurfaceisgivenbyei;f=eiLiOH;+eiLi2O;;.43andCdisthedoublelayercapacitance. 7.3.4 ImpedanceExpressionsThedierencebetweenthepotentialmeasuredatthesurfaceandthepotentialmeasuredsomedistanceawayisgivenbyeU=eiRe+eV.44Usingequation 7.44 theimpedanceforthelithiumsurfaceandtheoxidelayerwerecalculatedasZ=eU ei.45wheretheimpedanceofthelithiumsurfacewascalculatedtobeZ0=Req+1 R0)]TJ/F21 7.97 Tf 6.586 0 Td[(1+Rc;Li;0+zd;Li;0)]TJ/F21 7.97 Tf 6.586 0 Td[(1+RLiOH;0+zd;H2O;0)]TJ/F21 7.97 Tf 6.587 0 Td[(1+j!Cd1.46whereR)]TJ/F21 7.97 Tf 6.586 0 Td[(10=Ra;Li;0)]TJ/F21 7.97 Tf 6.586 0 Td[(1+RLi2O;0.47andtheimpedanceoftheoxidelayersurfacewascalculatedtobeZ=Re+1 RLiOH;+zd;H2O;+zd;Li;2)]TJ/F21 7.97 Tf 6.586 0 Td[(1+RLi2O;+zd;Li;3)]TJ/F21 7.97 Tf 6.586 0 Td[(1+j!Cd2.48TheelectrolyteresistancetermintheZ0wasreplacedbyanequivalentresistancethatcorrespondtoReq=Re+Rfilm.49whereRfilmistheresistanceacrosstheoxidelayer.TheimpedanceofthelithiumsurfaceforthesystempresentedinFigure 7{7 wascalculatedtobeZ0=Re+1 Ra;Li;0)]TJ/F21 7.97 Tf 6.587 0 Td[(1+Rc;Li;0+zd;Li;0)]TJ/F21 7.97 Tf 6.586 0 Td[(1+j!Cd1.50

PAGE 94

80 andtheimpedanceoftheoxidelayersurfacewascalculatedtobeZ=Re+1 RLiOH;+zd;H2O;+zd;Li;2)]TJ/F21 7.97 Tf 6.586 0 Td[(1+RLi2O;+zd;Li;3)]TJ/F21 7.97 Tf 6.586 0 Td[(1+j!Cd2.51TheoverallimpedanceforbothsystemswasnallycalculatedtobeZT=1 Z)]TJ/F21 7.97 Tf 6.587 0 Td[(10+Z)]TJ/F21 7.97 Tf 6.587 0 Td[(1.52 7.4 ResultsInthissectionresultsofcellUF-BTeexposedat80Carepresentedandtheresultsarecomparedtothemodels.Theobjectofthiscomparisonwastoshowthatthemodeldevelopedcandescribequalitativelythefeaturesseenintheexperimentaldata.ThemodelledresultswerecalculatedusingMatlab.TheparametersusedineachmodelarepresentedinTable 7{2 .ThetraditionalNyquistorimpedance{planerepresentationispresentedinFigure 7{9 fortheexperimentaldataandthemodels.Allplotsindicatethatthesystemhavecharacteristictimeconstantsdistributedoverawide{rangeoftimescales.Also,intheplotsalowfrequencytailisobserved.However,inFigure 7{9a asmoothtransitionandalowerslopeareobservedbetweenthehighfrequencycapacitiveloopandthelowfrequencytail,whereasinFigures 7{9b and 7{9c asharptransitionandasteeperslopeareobserved.TheimaginaryrepresentationisshowninFigure 7{10 fortheexperimentaldataandthemodels.Thegeneralbehaviorofbothplotsissimilar.However,theslopeofthegraphathighfrequencyinFigure 7{10a is)]TJ/F15 11.955 Tf 9.299 0 Td[(0:7653,whileforFigures 7{10b and 7{10c is)]TJ/F15 11.955 Tf 9.299 0 Td[(1.Thisdierenceisexplainedduetoadistributionoftimeconstantsintheexperimentalsystem.Thisdistributionisalsoevidentinthenon-uniformcoverageofthecellssurfacelms.TherealrepresentationisshowninFigure 7{11 fortheexperimentaldataandthemodels.Thegeneralbehaviorofbothplotsissimilar.

PAGE 95

81 Table7{2:SummaryoftheparametersusedtocalculatetheimpedanceresponseofthemodelspresentedinSection 7 Parameter Model1 Model2 DLi 10)]TJ/F21 7.97 Tf 6.587 0 Td[(9 10)]TJ/F21 7.97 Tf 6.587 0 Td[(9 DH2O 10)]TJ/F21 7.97 Tf 6.587 0 Td[(10 10)]TJ/F21 7.97 Tf 6.586 0 Td[(10 kc;Li;0 100 100 ka;Li;0 0.001 0.001 kc;LiOH;0 100 kc;Li2O;0 100 kc;LiOH; 100 100 kc;Li2O; 100 100 bc;Li;0 19.7 19.7 ba;Li;0 0.197 0.197 bc;LiOH;0 19.7 bc;Li2O;0 19.7 bc;LiOH; 19.7 19.7 bc;Li2O; 19.7 19.7 V 0.8 0.8 VLi;0 0.1 0.1 VLiOH;0 0.01 VLi2O;0 0.01 VLiOH; 0.01 0.01 VLi2O; 0.01 0.01 cLi+;Li;0 0.1 0.1 cH2O;0 0.01 cH2O; 0.01 0.01 cLi+;LiOH; 0.03 0.03 cLi+;Li2O; 0.03 0.03 Ra;Li;0 300 300 Rc;Li;0 200 200 RLiOH;0 150 RLi2O;0 150 RLiOH; 150 300 RLi2O; 150 200 Re 20 20 Req 140 180 Li;0 1 H2O;0 1 Li+;2 1 1 H2O; 1 1 Li+;3 0.5 0.5 Cd1 0.000001 0.000001 Cd2 0.000001 0.000001

PAGE 96

82 a b cFigure7{9:Impedance{planeorNyquistrepresentationofimpedancedataofasymmetriccell.aExperimentalResult.bModel1.cModel2.

PAGE 97

83 a b cFigure7{10:Imaginaryplotsofimpedancedataofasymmetriccell.aExperi-mentalResult.bModel1.cModel2.

PAGE 98

84 a b cFigure7{11:Imaginaryrepresentationofimpedancedataofasymmetriccell.aExperimentalResult.bModel1.cModel2.

PAGE 99

85 7.5 ConclusionsTwomodelsweredevelopedthatareabletodescribequalitativelythebehaviorobservedintheexperiments.Theselectionofthereactionsmechanismusedtodevelopedthosemodelswassupportedbyindependentobservations.Theimpedancedataandbrokensealsinthecellsprovidedevidenceofhydrogenevolution.MicroscopicandScanningElectronMicroscopeSEMmicrographsprovidedevidenceoftheformationofsolidlmsonthelithiumsurface.ThisisconsistentwiththeformationofLiOHandLiO2.Also,themicrographsshowedthatthelmformedatperimeterofthecellwhichprovidesevidenceofnon-uniformcurrentdistribution.Bothmodelsprovidedevidencethatthedistributionoftimeconstantsobservedinthehigh-frequencycapacitiveloopisduetoreactionsatthelithiumelectrode.Noneofthemodelledresultsallowfortheidenticationofmorethanonecapacitivetimeconstants.Itisnecessarytoaccountforthedistributedreactivityevidentintheexperimentaldata.ThedistributionofRCtimeconstantsshouldbeexploredbeforearigorousregressionofthemodelsisdone.

PAGE 100

CHAPTER8CONCLUSIONSElectrochemicalimpedancespectroscopyexperimentsandmodelswereusedtofacilitateanin-depthunderstandingofthephysicalprocessesthatcontrolsymmetriclithiumcoincells. 8.1 ExperimentsPreliminarygraphicalanalysisoftheexperimentsperformedonsymmetriclithiumcoincellsprovidedusefulguidestomodeldevelopment.NyquistandBoderepresentations,aswellas,IR-correctedBodeplots,log-logplotsoftheimaginarycomponentoftheimpedance,andeectivecapacitanceplotwereused.Theslopesinthelog-logplotsoftheimaginarycomponentoftheimpedancearedirectlyrelatedtotheapparentconstant-phase-elementcoecient.Thedierenceintheslopesatintermediateandhighfrequenciesdemonstratedthatmorethanonecapacitivetimeconstantcanbeidentiedinthesystem.Adecreaseoftheimpedanceresponsewithtimewasobserved.Thisisconsis-tentwithdissolutionofnativeoxidelayersonthelithiumsurface,asproposedintheliterature.Theresultsshowedthatthetimerequiredforcompletedissolutionoftheoxidelayerswasontheorderofseveraldays.Ahigh-frequencycapacitiveloopwasobservedinalltheexperimetalresults.Thisloopcanbeattributedtoadielectricbehavioroftheelectrolyteortoreac-tionsatthelithiummetalsurface.Asthetemperatureofthecellswasincreasedtheimpedanceresponsedecreased.Thisresultisconsistentwitheitherhypothesis,thatthehigh-frequencycapacitiveloopisduetoreactionsatthelithiummetalsurfaceorthattheloopisduetoadielectricbehavioroftheelectrolyte. 86

PAGE 101

87 Theimpedanceresponseofthesymmetricstainlesssteelcellswasconsistentwithablockingelectrodeatlowfrequencies,andathighfrequencies,thebehaviorwasconsistentwithnon-uniformcurrentdistribution.Thehigh-frequencycapaci-tiveloopseenintheresultsofthesymmetriclithiumcellwerenotevidentintheresultsfromthecellwithsteelelectrodes.Thus,thehigh-frequencycapacitiveloopseenintheimpedancespectraofChapter 5 isnotassociatedwiththedielectricresponseoftheelectrolyte.Thisloopmustbedescribedinsteadbyamodelthattakesintoaccountthereactionsatthelithiumelectrode. 8.2 ModelTwomodelsthattakeintoaccountthephysicsandchemistryofthesystemandthataresuitableforregressionanalysisweredeveloped.Bothmodelswereabletoqualitativelydescribethebehaviorofobservedintheexperiments.Theselectionofthereactionsmechanismusedtodevelopedthosemodelswassupportedbyindependentobservations.Theimpedancedataandbrokensealsinthecellsprovideevidenceofhydrogenevolution.MicroscopicandScanningElectronMicroscopeSEMmicrographsprovidedevidenceoftheformationofsolidlmsonthelithiumsurface,thisisconsistentwiththeformationofLiOHandLiO2.Also,themicrographsshowedthatthelmformedatperimeterofthecellwhichisevidenceofnon-uniformcurrentdistribution.Bothmodelsprovidedevidencethatthedistributionoftimeconstantsobservedinthehigh-frequencycapacitiveloopisduetoreactionsatthelithiumelectrode.Noneofthemodelledresultsallowfortheidenticationofmorethanonecapacitivetimeconstants.Itisnecessarytoaccountforthedistributedreactivityevidentintheexperimentaldata.ThedistributionofRCtimeconstantsshouldbeexploredbeforearigorousregressionofthemodelsisdone.

PAGE 102

REFERENCES [1] A.Hjelm,T.Eriksson,andG.Lindbergh,ElectrochemicalinvestigationofLiMn2O4cathodesingelelectrolyteatvarioustemperatures,"ElectrochimicaActa,vol.48,pp.171{179,2002. [2] Y.KimandE.Smotkin,Theeectofplasticizersontransportandelec-trochemicalpropertiesofPEO-basedelectrolytesforlithiumrechargeablebatteries,"SolidStateIonics,vol.149,pp.29{37,2002. [3] P.GeorenandG.Lindbergh,Characterizationandmodellingoftransportpropertiesinlithiumbatterypolymerelectrolyte,"ElectrochimicaActa,vol.41,pp.577{587,2001. [4] J.M.Tarascon,S.Grugeon,M.Morcrette,S.Laruelle,P.Rozier,andP.Poizot,Newconceptsforthesearchofbetterelectrodematerialsforrechargeablelithiumbatteries,"ComptesRendusChimie,vol.8,pp.9{15,2005. [5] M.Doyle,J.Meyers,andJ.Newman,Computersimulationsoftheim-pedanceresponseoflithiumrechargeablebatteries,"JournaloftheElectro-chemicalSociety,vol.1471,pp.99{110,2000. [6] W.BennetandM.Manzo,Preliminaryevaluationsofpolymer-basedlithiumbatteryelectrolytesunderdevelopmentforthepolymerenergyrechargeablesystemsprogram,1stInternationalEnergyConversionEngineeringConfer-ence,Portsmouth,VA,2003. [7] A.PereiraChagas,The200yearsoftheelectricpile,"QuimicaNova,vol.23,pp.427{429,2000. [8] F.Decker,Voltaandthepile,"ElectrochemistryEncyclopedia,2005,UniversityofRome"LaSapienza". [9] Batteryandfuelcell,"BritannicaStudentEncyclopedia.EncyclopediaBritannicaPremium,2005,http://www.britannica.com/ebi/article-197279.AcessedOctober23,2005. [10] G.NazriandG.Pistoia,LithiumBatteries:ScienceandTechnology,KluwerAcademicPublishers,Norwell,MA,2004. [11] P.Gomadam,J.Weidner,R.Dougal,andR.White,Mathematicalmodellingoflithium-ionandnickelbatterysystems,"JournalofPowerSources,vol.110,pp.267{284,2002. 88

PAGE 103

89 [12] K.Jung,S.Pyun,andJ.Lee,AninvestigationoflithiumtransportthroughvanadiumpentoxidexerogellmelectrodebyanalysesofcurrenttransientandAC-impedancespectra,"ElectrochimicaActa,vol.49,pp.4371{4378,2004. [13] D.Aurbach,B.Markovsky,A.Shechter,Y.Ein-Eli,andH.Cohen,AcomparativestudyofsyntheticgraphiteandLielectrodesinelectrolytesolutionsbasedonethylenecarbonate-dimetylcarbonatemixtures,"JournalofElectrochemicalSociety,vol.14312,pp.3809{3820,1996. [14] T.S.OngandH.Yang,SymmetricalcellforelectrochemicalACimpedancestudiesoflithiumintercalationintographite,"ElectrochemicalandSolidStatesLetters,vol.47,pp.A89{A92,2001. [15] S.ZangandP.Shi,ElectrochemicalimpedancestudyoflithiumintercalationintoMCMBelectrodeinagelelectrolyte,"ElectrochimicalActa,vol.49,pp.1475{1482,2004. [16] M.Itagaki,N.Kobari,S.Yotsuda,K.Watanabe,S.Kinoshita,andM.Ue,Insituelectrochemicalimpedancespectroscopytoinvestigatenegativeelectrodeoflithium-ionrechargeablebatteries,"JournalofPowerSources,vol.135,pp.255{261,2004. [17] A.HjelmandG.Lindbergh,ExperimentalandtheoreticalanalysisofLiMn2O4cathodesforuseinrechargeablelithiumbatteriesbyelectrochemicalimpedancespectroscopyEIS,"ElectrochimicaActa,vol.47,pp.1747{1759,2002. [18] A.Lisowska-Oleksiak,Theinterfacebetweenlithiumandpolyethylene-oxide,"SolidStateIonics,vol.119,pp.205{209,1999. [19] E.Peled,D.Golodnitsky,G.Ardel,andV.Eshkenazy,TheSEImodel-applicationtolithium-polymerelectrolytebatteries,"SolidStateIonics,vol.40,pp.2197{2204,1995. [20] P.Novak,F.Joho,R.Imhof,J.Panitz,andO.Haas,Insituinvestigationoftheinteractionsbetweengraphiteandelectrolytesolutions,"JournalofPowerSources,vol.81-82,pp.212{216,1999. [21] D.PletcherandF.C.Walsh,IndustrialElectrochemistry,ChapmanandHallLtd.,London,2ndedition,1990. [22] M.Broussely,Ph.Biensan,F.Bonhomme,Ph.Blanchard,S.Herreyre,K.Nechev,andR.J.Staniewicz,MainagingmechanismsinLiionbatteries,"JournalofPowerSources,vol.146,pp.90{96,2005. [23] W.Gorecki,M.Jeannin,E.Belorizky,C.Roux,andM.Armand,PhysicalpropertiesofsolidpolymerelectrolytePEOLiTFSIcomplexes,"JournalofPhysics,vol.7,pp.6823{6832,1995.

PAGE 104

90 [24] G.B.Appetechi,S.Scaccia,andS.Passerini,Investigationonthestabilityofthelithium-polymerelectrolyteinterface,"JournalofTheElectrochemicalSociety,vol.14712,pp.4448{4452,2000. [25] K.Murata,S.Izuchi,andY.Yoshiba,Anoverviewoftheresearchanddevelopmentofsolidpolymerelectrolytebatteries,"ElectrochimicaActa,vol.45,pp.1501{1508,2000. [26] P.GoerenandG.Lindbergh,Characterizationandmodellingofthetransportpropertiesinlithiumbatterygelelectrolytes.PartI:ThebinaryelectrolytePC/LiClO4,"ElectrochimicaActa,vol.49,pp.3497{3505,2004. [27] I.Rey,J.C.Lassegues,J.Grondin,andL.Servant,InfraredandRamanstudyofthePEO-LiTFSIpolymerelectrolyte,"ElectrochimicaActa,vol.43-11,pp.1505{1510,1998. [28] F.S.Fiory,F.Croce,A.D'Epifanio,S.Licoccia,B.Scrosati,andE.Traversa,PEObasedpolymerelectrolytelithium-ionbattery,"JournalofEuropeanCeramicSociety,vol.24,pp.1385{1387,2004. [29] C.Naudin,J.L.Bruneel,M.Chami,B.Besbat,J.Grondin,J.C.Lessegues,andL.Servant,Characterizationoflithiumsurfacebyinfraredandramanspectroscopy,"JourmalofPowerSources,vol.124,pp.518{525,2003. [30] R.PottsandR.Guy,Eds.,MechanismsofTransdermalDrugDelivery.DrugsandthePharmaceuticalSciences,MercelDekker,Inc,NewYork,NY,1997. [31] J.R.Macdonald,Ed.,ImpedanceSpectroscopy.EmphasizingSolidMaterialsandSystems,JohnWileyandSons,Inc.,Hoboken,NJ,1987. [32] C.Grabrielli,Identicationofelectrochemicalprocessesbyfrequencyresponseanalysis,"Solartroninstrumentationgroupmonograph004-83,TheSolartronElectronicGroupLtd.,Farnborough,UK,1984. [33] H.ShihandT.Lo,Electrochemicalimpedancespectroscopyforbatteryresearchanddevelopments,"Tech.Rep.,SolartronAnalytical,Farnborough,UK,1996. [34] Z.StoynovandD.Vladikova,DierentialImpedanceAnalysis,MarinDrinovAcademicPublishingHouse,Soa,Bulgaria,2005. [35] C.Gabrielli,Useandapplicationsofelectrochemicalimpedancetechniques,"Tech.Rep.,TheSolartronElectronicGroupLtd.,Farnborough,UK,1990. [36] P.Agarwal,M.E.Orazem,andL.Garcia-Rubio,Applicationofmeasurementmodelstoimpedancespectroscopy,"JournaloftheElectrochemicalSociety,vol.142,pp.4159{4168,1995.

PAGE 105

91 [37] P.Agarwal,M.E.Orazem,andL.Garcia-Rubio,ApplicationoftheKramers-Kronigrelationsinelectrochemicalimpedancespectroscopy,"ElectrochemicalImpedance:AnalysisandInterpretation,ASTMSTP1188,pp.115{139,1993. [38] M.Durbah,M.E.Orazem,andL.Garcia-Rubio,SpectroscopyapplicationsoftheKramers-Kronigtransforms:Implicationsfortheerrorstructureidentication,"JournaloftheElectrochemicalSociety,vol.1441,pp.48{55,1997. [39] M.E.Orazem,P.Shukla,andM.Membrino,Extensionofthemeasurementmodelapproachfordeconvolutionofunderlyingdistributioinsforimpedancemeasurements,"ElectrochimicaActa,vol.47,pp.2027{2034,2002. [40] S.Carson,M.E.Orazem,O.Crisalle,andL.Garcia-Rubio,Ontheerrorstrucutureofimpedancemeasurements.simulationofFRAintrumentation,"JournalofElectrochemicalSociety,vol.15010,pp.E477{E490,2003. [41] M.E.Orazem,Asystematicapproachtowarderrorstructureidenticationforimpedancespectroscopy,"JournalofElectroanalyticalChemistry,vol.572,pp.317{327,2004. [42] H.Young,UniversityPhysics,Addison-WesleyPublishingCompany,Reading,MA,8thedition,1992. [43] S.Boctor,ElectricCircuitAnalysis,PrenticeHall,EnglewoodClis,NJ,2ndedition,1992. [44] M.E.Orazem,N.Pebere,andB.Tribollet,Anewlookatgraphicalrepre-sentationofimpedancedata,"Paperpresentedatthe206thmeetingoftheElectrochemicalSociety,Honolulu,Hawaii,October3-8,2004. [45] B.Bennett,IntercalationexperimentswithMCMBcarboninPEO-basedelectrolyte,"NASAInternalReport-211990,Cleveland,OH,2003. [46] J.Newman,K.Thomas,H.Hafezi,andD.Wheeler,Modellingoflithium-ionbatteries,"JournalofPowerSources,vol.119-121,pp.838{843,2003. [47] C.SchmidtandP.Skarstad,Developmentofanequivalent-circuitmodelforthelithium/iodinebattery,"JournalofPowerSources,vol.65,pp.121{128,1997. [48] J.TheveninandR.Muller,Impedanceoflithiumelectrodeinpropylenecarbonateelectrolyte,"JournalofTheElectrochemicalSociety,vol.134,pp.273{280,1987. [49] M.Doyle,T.Fuller,andJ.Newman,Theimportanceofthetransferencenumberinlithium/polymercell,"ElectrochimicaActa,vol.39,pp.2073{2081,1994.

PAGE 106

92 [50] R.SpotnitzandJ.Franklin,Abusebehaviorofhigh-power,lithium-ioncells,"JournalofPowerSources,vol.113,pp.81{100,2003. [51] D.Aurbach,Y.Talyosef,B.Markovsky,E.Markevich,E.Zinigrad,L.Asraf,J.Gnanaraj,andH.Kim,DesignofelectrolytesolutionsofLiandLi-ionbatteries:Areview,"ElectrochimicaActa,vol.50,pp.247{254,2004. [52] P.Ramadass,B.Haran,R.White,andB.Popov,MathematicalmodellingofcapacityfadeofLi-ioncells,"JournalofPowerSources,vol.123,pp.230{240,2003. [53] P.Nelson,I.Bloom,K.Amine,andG.Henriksen,Designmodellingoflithium-ionbatteryperformance,"JournalofPowerSources,vol.110,pp.237{244,2002. [54] B.Markovsky,A.Rodkin,Y.Cohen,O.Palchik,E.Levi,D.Aurbach,H.Kim,andM.Schmidt,ThestudyofthecapacityfadingprocessofLi-ionbatteries:Majorfactorsthatplayarole,"JournalofPowerSources,vol.119-121,pp.504{510,2003. [55] B.DixonandR.Morris,Computationalchemistry:Designandexperimentalvericationofpre-designedheeropolymerelectrolytesforrechargeablelithiumbatteries,"JournalofPowerSources,vol.119-121,pp.856{858,2003. [56] A.Crespi,C.Schmidt,J.Norton,K.Chen,andP.Skarstad,Modellingandcharacterizationoftheresistanceoflithium/SVObatteriesforimplantablecardioverterdebrillators,"JournalofTheElectrochemicalSociety,vol.1481,pp.A30{A37,2001. [57] S.Buller,M.Thele,E.Karden,andR.DeDoncker,Impedance-basednon-lineardynamicbatterymodellingforautomotiveapplications,"JournalofPowerSources,vol.113,pp.422{430,2003. [58] A.Salkind,P.Singh,A.Cannone,T.Atwater,X.Wang,andD.Reisner,Impedancemodellingofintermediatesizelead-acidbatteries,"JournalofPowerSources,vol.116,pp.174{184,2003. [59] P.Notten,O.Ouwerkerk,A.Ledovskikh,H.Senoh,andC.Iwakura,Hydrideformingelectrodematerialsseenfromakineticperspective,"JournalofAlloysandCompounds,vol.356-357,pp.759{763,2003. [60] E.Deiss,F.Holzer,andO.Hass,ModellingofanelectricallyrechargeablealkalineZn-Airbattery,"ElectrochimicaActa,vol.47,pp.3995{4010,2002. [61] A.Ledovskikh,E.Verbitskiy,A.Ayeb,andP.Notten,Modellingofrecharge-ableNiMHbatteries,"JournalofAlloysandCompounds,vol.356-357,pp.742{745,2003.

PAGE 107

93 [62] R.Marsh,S.Vukson,S.Surampudi,B.Ratnakumar,M.Smart,M.Manzo,andP.Dalton,Li-ionbatteriesforaerospaceapplications,"JournalofPowerSources,vol.98,pp.25{27,2002. [63] R.Bouchet,S.Lascaud,andM.Rosso,AnEISstudyoftheanodeLi/PEO-LiTFSIofaLipolymerbattery,"JournaloftheElectrochemicalSociety,vol.1500,pp.A1385{A1389,2003. [64] M.E.Orazem,Asystematicapproachtowarderrorstructureidenticationforimpedancespectroscopy,"JournalofElectroanalyticalChemistry,vol.572,pp.317{327,2004. [65] M.E.Orazem,Interpretationofimpedancespectroscopy,"Manuscriptinpreparation.DraftNovember2005. [66] M.E.Orazem,UserManualfortheMeasurementModelToolboxforImpedanceSpectroscopy,UniversityofFlorida,ChemicalEngineeringDepartment,ElectrochemistryLaboratory,Gainesville,FL,2000. [67] V.M.Huang,V.Vivier,M.E.Orazem,N.Pebere,andB.Tribollet,TheimpedanceresponseofablockingelectrodewithlocalCPEbehavior,"Manuscriptinpreparation.DraftApril2006. [68] D.Aurbach,Reviewofselectedelectrode-solutioninteractionswhichdeterminetheperformanceofLiandLi-ionbatteries,"JournalofPowerSources,vol.89,pp.206{218,2000. [69] P.Arora,R.E.White,andM.Doyle,Capacityfademechanismsandsidereactionsinlithium-ionbatteries,"JournaloftheElectrochemicalSociety,vol.145,pp.3647{3667,1998. [70] I.Rey,C.Lessegues,P.Baundry,andH.Majastre,StudyofalithiumbatterybyconfocalRamanmicrospectrometry,"ElectrochimicaActa,vol.43-11,pp.1539{1544,1998.

PAGE 108

BIOGRAPHICALSKETCHNelliannPerez-GarciareceivedtheBachelorofSciencedegreeinchemicalen-gineeringfromtheUniversityofPuertoRico,MayaguezCampusinMay1999.ShebeganhergraduatestudiesattheUniversityofFloridainAugust1999.Intheyear2000shejoinedProfessorMarkE.Orazem'selectrochemicalengineeringresearchgrouptopursueaMasterofSciencedegree.Uponcompletionofthedegreerequire-mentsinMay2003,NellianncontinuedhergraduatestudiestopursueaDoctorofPhilosophydegree.InJanuary2006NelliannbeganworkingforLockwoodGreeneinAtlanta,Georgia,wheresheplanstocontinueherdevelopmentasachemicalengineer. 94


Permanent Link: http://ufdc.ufl.edu/UFE0013779/00001

Material Information

Title: Enhanced Interpretation Models for Impedance of Lithium Ion Batteries
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013779:00001

Permanent Link: http://ufdc.ufl.edu/UFE0013779/00001

Material Information

Title: Enhanced Interpretation Models for Impedance of Lithium Ion Batteries
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013779:00001


This item has the following downloads:


Full Text











ENHANCED INTERPRETATION MODELS FOR IMPEDANCE OF
LITHIUM ION BATTERIES















By

NELLIANN PEREZ-GARCIA


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2006















I dedicate this work to my grandparents. These people are my heroes and I

feel very proud to be part of their family.















ACKNOWLEDGMENTS

The author wishes to express her sincere gratitude and appreciation to various

individuals who helped in the development of this work. To my advisor, Dr. Mark

E. Orazem, goes my appreciation for his guidance, patience, and discussions related

to the ci i1,Ji -i of this work. I am very grateful that he believed in me and gave

me the opportunity to develop my experimentation and modeling skills. I thanks

my committee and group members for all their help and input in the development

of this work. Special thanks go to Vicky Huang and Shirley Kelly for all their

support.

Also I wish to acknowledge the University of Florida Graduate Minority

Fellowship and the National Aeronautics and Space Administration (NASA) for

funding this project. Special thanks go to William Bennett, Michelle Manzo, and

Marla Perez-Davis for all their help in this project.

Finally, and most important, I would like to thank my parents, Oscar and

Carmen, for all the sacrifices they made to support me in this project; and my

sister, Maricarmen, for motivating me along this journey. I would like to thank

my husband, Jose, for his support and love. These people gave me the courage to

pursue my dreams, and I am albv- -i going to be grateful for having them by my

side.















TABLE OF CONTENTS
page

ACKNOWLEDGMENTS ................... ...... iii

LIST OF TABLES ...................... ......... vi

LIST OF FIGURES ................... ......... viii

ABSTRACT ...................... ............ xiii

CHAPTER

1 INTRODUCTION .................... ....... 1

2 CHARACTERISTICS OF LITHIUM-BASED BATTERIES ....... 3

2.1 History of Batteries .......................... 3
2.2 ('C! ''o :teristics of Lithium-Based Batteries ............. 5
2.2.1 Anode Materials ........... ............ 7
2.2.2 Cathode M materials ....................... 8
2.2.3 Electrolyte . . . . . . 9
2.2.4 Electrode-Electrolyte Interface ....... ........ 12

3 ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY THEORY 15

3.1 Impedance Response of Electrical Circuits . . ..... 16
3.2 Experimental Considerations .......... .... 17
3.3 Error Structure .................. ....... .. .. 19
3.3.1 Kramers-Kronig Relations .............. .. .. 19
3.3.2 Measurement Model Approach . . ..... 19
3.4 Impedance Spectroscopy Data Analysis . . 20
3.5 Impedance Response Models ................ .... 23

4 EXPERIMENTAL DESIGN AND ERROR ANALYSIS . ... 26

4.1 Experimental System ................ ... ... .. .. 26
4.2 Experiment Design ............... ...... .. 29
4.3 Measurement Model Analysis ............. .. .. 30
4.3.1 Steady State Experiments at Constant Temperature . 31
4.3.2 Steady State Experiments at Increasing Temperatures 33
4.3.3 Transient Experiments at Constant Temperature . 37
4.3.4 Blocking Electrodes ..... ........... ...... 38









5 EXPERIMENTAL RESULTS AND PRELIMINARY ANALYSIS .... 40

5.1 Steady State Experiments at Constant Temperature ....... 41
5.2 Steady State Experiments at Increasing Temperatures ...... 44
5.3 Transient Experiments at Constant Temperature . ... 49
5.4 Conclusions .................. ........... .. 54

6 BLOCKING ELECTRODE ANALYSIS ................. .. 57

6.1 Experimental Results .................. ..... .. 57
6.2 Graphical Analysis .................. ....... .. 57
6.3 Comparison to Disk Electrode ............ .. .. .. 61
6.4 Conclusions .................. ........... .. 63

7 MODEL DEVELOPMENT .................. ..... .. 64

7.1 C(' C i,-Iriy . . . . . . ... .. 64
7.2 Physics ................... ..... ....... .. 67
7.3 Model Development .................. ....... .. 69
7.3.1 Steady State .................. ..... .. .. 74
7.3.2 Sinusoidal Steady State ............ .. .. .. 75
7.3.3 M ass Transfer .................. ..... .. 76
7.3.4 Impedance Expressions ............ .. .. .. 79
7.4 Results ................... ..... ....... 80
7.5 Conclusions .................. ........... .. 85

8 CONCLUSIONS .................. ............ .. 86

8.1 Experiments .................. ........... .. 86
8.2 M odel .................. .............. .. 87

REFERENCES ...... ........... ............ .. 88

BIOGRAPHICAL SKETCH .................. ......... .. 94















LIST OF TABLES
Table page

2-1 C('! .i,, I composition of passivation films formed on lithium metal
surface in contact with different electrolytes. ........... 13

2-2 C('! i i,, I composition of passivation films on the Li metal in contact
with additives used to enhance the cycling efficiency. . ... 14

4-1 Summary of the experimental design. The number of sequential mea-
surements is presented .............. ..... .. 30

4-2 Error structure models determined for Cell UF-CTe. This cell was ex-
posed to a temperature range varying from 600C to 850C. ..... 35

4-3 Error structure models determined for Cell NA-AB. This cell was ex-
posed to a temperature range varying from 80C to 100C at 10C
intervals. .................. .............. ..39

5-1 Average a values obtained using graphical techniques at the interme-
diate frequency range for Cell NA-ATi through NA-DTi. Six repli-
cates were used to calculate the average parameters . ... 44

5-2 Average values of the parameters obtained using graphical techniques
at the high frequency range for Cell NA-ATi trough NA-DTi. Six
replicates were used to calculate the average parameters ...... ..44

5-3 Average values of the low frequency tail slopes from cells UF-BTe,
UF-CTe, and UF-FTe. Six replicates were used to calculate the av-
erage parameters. ............... ....... 46

5-4 Slopes obtained using graphical techniques at the intermediate fre-
quency range from cells UF-BTe, UF-CTe, and UF-FTe. Six repli-
cates were used to calculate the average parameters . ... 47

5-5 Average values of the parameters obtained using graphical techniques
at the high frequency range from cells UF-BTe, UF-CTe, and UF
FTe. Six replicates were used to calculate the average parameters .50

5-6 Values of the low frequency tail slopes from Cell UF-DTr and UF-ETr. 53

5-7 Values of the low and intermediate frequency tail slopes obtained from
Cell UF-DTr and UF-ETr. ............. .... 53









5-8 Average values of the parameters obtained using graphical techniques
at the high frequency range for Cell UF-DTr. . ...... 55

6-1 Parameter values extracted from Figures 6-2 and 6-3 for Cell NA-BB. 61

7-1 Reaction mechanisms of common lithium battery contaminants. Wa-
ter was the only contaminant considered in this work. . ... 65

7-2 Summary of the parameters used to calculate the impedance response
of the models presented in Section 7. ............... 81















LIST OF FIGURES
Figure page

2-1 Schematic of the o11 iii i I electrical ois;, or the "Pile" developed
by Alessandro Volta in 1799. .................. ... 4

2-2 Schematic of a rechargeable lithium battery. ...... . . 6

2-3 Schematic representation of the lithium-polymer electrolyte inter-
phase.(a) Overall interphase (b) Small segment of the interphase. .13

3-1 Passive elements that serve as a component for an electrical circuit. .16

3-2 Small signal analysis of an electrochemical non-linear system. ... ..18

3-3 Electrode configuration in an electrochemical cell. ......... .18

3-4 Impedance-plane or Nyquist representation of impedance data for
Cell NA-ATi. The open symbols represent measurements taken
15 hour after the cell preparation. The half-full symbols represent
measurements taken approximately 48 hours after the cell prepara-
tion ...... ............. ................ .. 21

3-5 Schematic representation of a simple reactive system. . . 21

3-6 Bode representation of impedance data, magnitude and phase angle,
for Cell NA-ATi, respectively. The open symbols represent mea-
surements taken 15 hour after the cell preparation. The half-full
symbols represent measurements taken approximately 48 hours af-
ter the cell preparation. .................. ..... 22

3-7 Plots of the real and imaginary part of the impedance as a function
of frequency for Cell NA-ATi. The open symbols represent mea-
surements taken 15 hour after the cell preparation. The half-full
symbols represent measurements taken approximately 48 hours af-
ter the cell preparation. .................. ..... 22

4-1 Schematic of the cell sandwich. The cell was build using a thin, insu-
lating mask to separate the electrodes. The mask was intended to
prevent short circuit. .................. ..... 27

4-2 Schematic of the experimental setup used for the electrochemical im-
pedance tests. .. .. ... .. .. .. .. ... .. .. .. .. 28









4-3 Schematic representation of a Voigt circuit used as a measurement
model ...... ............. .............. .. 31

4-4 Measurement model prediction of impedance data for Cell NA-ATi.
The open symbols represent measurements taken 15 hours after
their preparation. The half-full symbols represent measurements
taken approximately 48 hours after the cell preparation. The solid
line represents the measurement model fit. ............ .. 32

4-5 Error structure for Cell NA-ATi measurements taken 15 and 48 hours
after their preparation, respectively. The open symbols represent
the real part of the impedance. The half-full symbols represent imag-
inary part of the impedance. The solid line represents the error
structure model fit. ............... ....... 33

4-6 Measurement model fit of the imaginary part of the impedance data
for Cell NA-ATi. The open symbols represent measurements taken
15 hours after their preparation. The half-full symbols represent
measurements taken approximately 48 hours after the cell prepara-
tion. The solid line represents the measurement model fit and the
dashed lines represent the 95.!' confidence interval. . ... 34

4-7 Measurement model prediction of the real part of the impedance data
for Cell NA-ATi. The open symbols represent measurements taken
15 hours after their preparation. The half-full symbols represent
measurements taken approximately 48 hours after the cell prepara-
tion. The solid line represents the measurement model prediction
and the dashed lines represent the 95.!' confidence interval. . 35

4-8 Error structure for Cell UF-CTe measurements (a)600C (b)650C (c)70C
(d)75C (e)80C (f)85C. The open symbols represent the real part
of the impedance. The half-full symbols represent imaginary part
of the impedance. The solid line represents the error structure model
fit .................. ................. 36

4-9 Error structure for Cell NA-ATi measurements taken 15 and 48 hours
after their preparation, respectively. The open symbols represent
the real part of the impedance. The half-full symbols represent imag-
inary part of the impedance. The solid line represents the error
structure model fit. ............... ....... 37

4-10 Error structure for Cell NA-AB measurements (a)800C (b)900C (c)1000C.
The open symbols represent the real part of the impedance. The
half-full symbols represent imaginary part of the impedance. The
solid line represents the error structure model fit. . .... 38









5-1 Equivalent circuit model used to fit the impedance data obtained from
the Li|PEO-LiTFSI system. Figure reproduced from: R. Bouchet,
S. Lascaud, and M. Rosso. An EIS Study of the Anode Li PEO-
LiTFSI of a Li Polymer Battery. Journal of Electrochemical Soci-
ety. Vol. 150, No. 10, pp.A1385-A1389, 2003. . . 40

5-2 Effective CPE coefficient defined by equation (5.1) for Cell NA-ATi.
The open symbols represent measurements taken approximately 15
hours after the cell preparation. The half-full symbols represent
measurements taken approximately 48 hours after the cell prepa-
ration ...... ............. .............. .. 42

5-3 IR-corrected Bode plots for Cell NA-ATi; a) phase angle, b) mod-
ulus. The open symbols represent measurements taken approxi-
mately 15 hours after the cell preparation. The half-full symbols
represent measurements taken approximately 48 hours after the
cell preparation. .................. ..... 43

5-4 Impedance-plane or Nyquist representation of impedance data for
Cell UF-CTe with temperature as a parameter. Six replicates of
each measurement are presented for each temperature. ...... ..45

5-5 Bode Representation of impedance data for phase angle and magni-
tude for Cell UF-CTe, respectively, with temperature as a para-
meter. Six replicates of each measurement are presented for each
temperature. ............... .......... 45

5-6 Plots of the imaginary and real part of the impedance as functions of
frequency for Cell UF-CTe with temperature as a parameter. Six
replicates of each measurement are presented for each temperature. 46

5-7 Average values of the low frequency tail slopes from cells UF-BTe,
UF-CTe, and UF-FTe. Six replicates were used to calculate the
average parameters. .................. ..... 47

5-8 Average values of the intermediate frequency tail slopes from cells
UF-BTe, UF-CTe, and UF-FTe. Six replicates were used to cal-
culate the average parameters. ................... .... .48

5-9 Average values of the high frequency tail slopes from cells UF-BTe,
UF-CTe, and UF-FTe. Six replicates were used to calculate the
average parameters. .................. ..... 48

5-10 Effective CPE coefficient defined by equation (5.1) for Cell UF-CTe.
Six replicates of each measurement are presented for each tempera-
ture ................... ........... .... 49









5-11 IR-corrected Bode plots for Cell UF-CTe; (a) phase angle, (b) modu-
lus. Six replicates of each measurement are presented for each tem-
perature. ............... ............ .. 50

5-12 Impedance-plane or Nyquist representation of impedance data for
Cell UF-DTr with temperature as a parameter. Impedance spectra
were taken every 30 seconds for 48 hours, it took approximately
two hours to complete one spectrum. ................ 51

5-13 Bode Representation of impedance data for phase angle and magni-
tude for Cell UF-DTr, respectively. Impedance spectra were taken
every 30 seconds for 48 hours, it took approximately two hours to
complete one spectrum. ............... .... 51

5-14 Plots of the imaginary and real part of the impedance as functions of
frequency for Cell UF-DTr. Impedance spectra were taken every
30 seconds for 48 hours, it took approximately two hours to com-
plete one spectrum. ............... ...... 52

5-15 Effective CPE coefficient defined by equation (5.1) for Cell UF-DTr. 54

5-16 IR-corrected Bode plots for Cell UF-DTr; a) phase angle, b) modulus. 56

6-1 Complex impedance plane or Nyquist representation for the response
of a 430 Stainless Steel disk in PEO/LiTSFI electrolyte with con-
centration as a parameter. Six replicates of each measurement are
presented for each temperature. ................ .... 58

6-2 Plots of the imaginary and real part of the impedance as functions of
frequency for Cell NA-BB with temperature as a parameter. Six
replicates of each measurement are presented for each temperature. 59

6-3 Effective CPE coefficient defined by equation (6.3) for Cell NA-BB.
Six replicates of each measurement are presented for each tempera-
ture ................... ........... .... 60

6-4 Dimensionless analysis for the impedance response of a stainless steel
disk in PEO/LiTFSI electrolytes with temperature as a parameter. 62

6-5 Dimensionless imaginary part of the impedance as a function of di-
mensionless frequency for a disk electrode with local capacitance
dispersion .................... ........... .. 62

7-1 Impedance-plane or Nyquist representation of impedance data for
Cell UF-DTr. Very large impedance at low frequencies can be ob-
served si-.:. -i- ii; that a disconnection of the system which may be
attributed to gas evolution. .............. ...... 66









7-2 Cell UF-ATe lithium electrode surface. The surface of the electrode
was almost completely covered with a white to blue oxide 1 iVr. .. 67

7-3 Cell UF-BTe lithium electrode surface. The edges of the electrode
were covered with a white to blue oxide 1~-.vr, while the inner sur-
face kept a much brighter metallic lustre. . . ...... 68

7-4 Scanning electron micrographs of Cell UF-ATe lithium electrode sur-
face. (a) Magnification = 35. (b) Magnification = 200. ...... ..70

7-5 Scanning electron micrographs of Cell UF-BTe lithium electrode sur-
face. Magnification = 350. .................. .... 71

7-6 Commercial lithium foil native lv.-r composition. . . 71

7-7 Schematic of the simplified system. Two surfaces were identified in
which the reactions take place. The lithium oxidation/reduction
reaction take place at the lithium metal surface and the LiOH and
Li20 formation reactions take place in the oxide l v-r surface. .72

7-8 Schematic of the simplified system. Two surfaces were identified in
which the reactions take place. The lithium oxidation/reduction
reaction take place at the lithium metal surface only, but the LiOH
and Li20 formation reactions can take place in both the lithium
metal surface and the oxide l1 v-r surface. . . ...... 73

7-9 Impedance-plane or Nyquist representation of impedance data of a
symmetric cell. (a) Experimental Result. (b) Model 1. (c) Model 2. 82

7-10 Imaginary plots of impedance data of a symmetric cell. (a) Experi-
mental Result. (b) Model 1. (c) Model 2 ... . 83

7-11 Imaginary representation of impedance data of a symmetric cell. (a)
Experimental Result. (b) Model 1. (c) Model 2 . . ... 84















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Phil. .. 1hi

ENHANCED INTERPRETATION MODELS FOR IMPEDANCE OF
LITHIUM ION BATTERIES

By

Nelliann Perez-Garcia

i ,v 2006

C'!I ,i': Mark E. Orazem
Major Department: C', iii,. I Engineering

Electrochemical impedance spectroscopy experiments and models were used

to facilitate an in-depth understanding of the physical processes that control

symmetric lithium coin cells. The experimental component was intended to obtain

good quality data to guide in the model development. The quality of the results

obtained were tested using the measurement model approach developed in our

group.

Experiments performed on symmetric lithium coin cells, with time and

temperature as the tested variables, showed that more than one time constant

can be identified in the system. The decrease of the impedance response with

time is consistent with dissolution of native oxide lV-. -i on the lithium surface,

as proposed in the literature. The results showed that the time required for

complete dissolution of the oxide l-,--ir was on the order of several d v-. As the

temperature of the cells was increased the impedance response decreased. This

result is consistent with two hypotheses, that the high-frequency capacitive loop is

due to reactions at the lithium metal surface or that the loop is due to a dielectric

behavior of the electrolyte.









Experiments using 430 stainless steel electrodes were performed to test the

hypothesis that the loop was associated with a dielectric behavior of the electrolyte.

The results obtained proved that the high-frequency capacitive loop is associated

instead with reactions at the lithium electrode and not with a dielectric behavior of

the electrolyte. The low-frequency behavior was consistent with that expected for

blocking electrodes. However, at high frequencies, the impedance was influenced by

a non-uniform current distribution.

The objective of the modeling work was to find a bridge between easily-

used circuit models and more complex physico-chemical models. EIS is not a

stand-alone technique, and additional information was needed to develop such a

model. An extensive literature review was done to investigate the reactions and

mechanisms of transport that control the symmetric lithium coin cells. A simplified

set of reactions, which include the oxidation reduction reaction of lithium and the

reactions due to the presence of water as a contaminant, was chosen. The products

of the reactions chosen include LiOH, LiO2, and H2. Independent observations

supported the choice of reactions. The impedance data and broken seals in the

cells provide evidence of hydrogen evolution. Microscopic and Scanning Electron

Microscope (SEM) micrographs provided evidence of the formation of solid films

on the lithium surface, consistent with the formation of LiOH and LiO2. Also,

the micrographs showed that the film formed at the perimeter of the cell which is

evidence of a non-uniform current distribution.

Two possible mechanisms of transport were modelled. Both models provided

evidence that the distribution of time constants observed in the high-frequency

capacitive loop can be attributed to reactions at the lithium electrode. The models

developed are consistent with the impedance response and are supported by

independent observations.















CHAPTER 1
INTRODUCTION

Rechargeable lithium batteries are a promising energy source for consumer,

industrial, and military applications. Lithium batteries are appealing for these

applications because lithium metal offers a low weight anode (6.99 g mol-1) with

high negative potential (-3.05 V/NHE) and high specific capacity (3.76 A h g-1).1

The overall performance of a lithium rechargeable battery depends on the choice of

cathode, anode, electrolyte, and the electrode-electrolyte interfacial properties.2

Solid polymer electrolytes (SPEs) offer improved lithium battery safety as

compared to liquid electrolytes, since SPEs do not contain any volatile or reactive

components. On the other hand, SPEs are known to result in batteries with

insufficient performance, suffering from large voltage losses in the electrolyte during

passage of a current.3 For this reason, new materials are being developed for the

next generation of lithium-based batteries.4 The search for optimal chemistry

and battery design for a given application can be costly and time-consuming.

Mathematical modeling can achieve rn' i of these goals as well as improve the

fundamental understanding of the processes inside the battery.5

Electrochemical Impedance Spectroscopy (EIS) is an attractive experimental

technique to study lithium batteries because it is non-destructive and a variety of

capacitive and resistive properties of the material can be determined in a single

brief experiment. Interpretation of impedance data is generally guided by use of

equivalent electrical circuits. Both simple lumped-parameter circuits and more

complex finite-transmission-line circuits have been used. The formulation of the

overall circuit network involves the semi-empirical addition of circuit elements

until the model provides an adequate description of the impedance response.









Upon the establishment of an appropriate network, a deductive process is used

to relate the properties of the system to the individual circuit elements, which

comprise the overall network. One limitation of such electric circuit models is

that an unambiguous interpretation of the data is not possible. For particular

impedance spectra, more than one circuit network can be constructed which would

provide equivalent statistically valid fits, and each of these models can arise from

extremely different interpretations. This reduces the efficacy of circuit networks for

interpretation of the physical properties and the kinetic processes of the system.

A less ambiguous approach for interpreting impedance data is to develop physico-

chemical deterministic models based on the governing equations, which describe the

physics and chemistry of the system and that can be regressed to the experimental

data.

The objective of this work was to find a bridge between the complex physico-

chemical models with more commonly used equivalent circuit models. The work

combined experiments with the development of models that could be applied to

lithium-based battery systems.

This investigation was performed as part of NASA's Polymer Energy

Rechargeable Systems (PERS) program. The program was established to develop

the next generation of lithium-based, polymer electrolyte batteries for aerospace

applications. The goal of the program is to develop a space-qualified, advanced

battery system embodying a solid polymer electrolyte and lithium-based electrode

technologies and to establish world-class domestic manufacturing capabilities for

advanced batteries with improved performance characteristics that address NASA's

future aerospace battery requirements.6















CHAPTER 2
CHARACTERISTICS OF LITHIUM-BASED BATTERIES

Lithium based batteries have been the focus of research for the last decades.

These batteries offer significant advantages such as reduced weight and volume of

the energy storage system, improved reliability, lower power system life cycle costs,

higher energy density, and a more benign environmental impact.2 For these reasons

they are considered for diverse applications that include portable electronic devices,

such as cellular phones and laptops computers, electric vehicles, space explorations,

uninterrupted power supplies, and medical devices. Lithium-ion batteries have

replaced lead-acid, nickel-metal, and nickel-cadmium battery systems for several

applications.

2.1 History of Batteries

The first battery reported in history was invented by Alessandro Volta in the

year 1799.7 He first named it 11 il i i. electrical organ," because he was trying

to reproduce the behavior of the torpedo fish. His invention was later known as

the "Pile." The o 1I111 i electrical ,i i;, was the result of a controversy between

Volta and his colleague Luigi Galvani, a medical doctor at Bologna University.7'

Galvani observed that he obtained muscular contraction in a frog by touching

one end of the leg with one metal and the other end with a different metal. He

concluded that the "animal electric fluid" was the source of the contraction. Volta,

on the other hand, believed that the contact of dissimilar metals was the true

source of stimulation. To test his hypothesis, he built the ,111 i i id electrical

organ." The apparatus consisted of a moistened porous pad sandwiched between

two dissimilar metals. A detailed sketch of it is presented in Figure 2-1. The

metals used were silver and zinc, and the disk was moistened with water for some























Figure 2-1: Schematic of the "artificial electrical ,ii, i, or the "Pile" developed by
Alessandro Volta in 1799. Figure reproduced from A.P. ('!i I, -, "The 200 Years of
the Electric Pile," Quimica Nova, 23 (2000) 427-429.7


cells and a saline solution for others. Volta observed and reported that the saline

solution worked better than pure water in the moistened disks, but he showed no

interest in exploring the role of the electrolyte since he believed that the metal

metal interface was the source of electricity.8

In the years that followed, other means of producing electricity were invented.

The British researcher John Frederich Daniell developed an arrangement in which

a copper plate was located at the bottom of a wide-mouthed jar, and a cast

zinc piece was located at the top. Two electrolytes were employ, 1d A saturated

copper sulphate solution covered the copper plate and extended halfway up the

remaining distance toward the zinc piece. Then a zinc sulphate solution was

carefully poured above the copper sulphate and immersed the zinc. The two

solutions were separated by a porous clay cylinder separator. It was one of the

earliest practical "laboratory" electrical sources, but it was not much used outside

the laboratory.8'9 Among all the systems invented, the most successful were the

ones developed by Grove in 1839 and Bunsen in 1842. Grove added an oxidizing

agent to prevent accumulation of hydrogen at the cathode where it reduced the









voltage the cell produced. Bunsen improved Groves cell by substituting cheap

carbon for the expensive platinum cathode.9

None of these technologies survives tod i,. The first enduring invention came

from Raymond Gaston Plant6, who developed the lead-acid cell in 1859. The next

n, i i development was the wet cell by the French engineer Georges Laclanch6.

He used a cathode of manganese dioxide mixed with carbon and an anode of zinc

in the form of a rod. The electrolyte was a solution of ammonium chloride. This

chemical system is used in flashlight batteries.9

The production of batteries was greatly increased during the First World War

as a means of powering torches and field radios. Other milestones in battery pro-

duction include the widespread radio broadcasting, which brought battery-operated

wireless into many homes. During the inter-war years, battery performance was

greatly enhanced through better selection of materials and methods of manu-

facture. Batteries have now become an essential part of everydi- life. They are

power sources for millions of consumer, business, medical, military and industrial

appliances worldwide.

2.2 ('1 i i''teristics of Lithium-Based Batteries

Lithium is the most chemically reactive metal and provides the basis for

tod 'i's most compact power systems. Nearly all high-density storage systems use

lithium because it has a specific capacity of 3860 Ampere-hour per kilogram of

mass (Ah/kg) compared to 820 Ah/kg for zinc and 260 Ah/kg for lead. Depending

on the cathode, cells with lithium anodes can produce 1.5 volts to 3.6 volts per cell.

The three primary components of any battery are the cathode, the electrolyte,

and the anode. One problem with using lithium metal as the electrode is that it

is too reactive. It reacts violently with water and can ignite into flame. Also, as

a consequence of the reaction, a film is formed on the metallic lithium. This, in

turn, results in continuously increasing electrode surface area and more extensive











Load

Interfaces (SEI)



Li VxO
Electrolyte or
....Li. Mnx.02
-X


Figure 2-2: Schematic of a rechargeable lithium battery. Figure reproduced from:
G.Nazri and G. Pistoia, Lithium Batteries: Science and T I. I,,...I/; (Norwell,MA:
Kluwer Academic Publishers, 2004).

surface reactions. The ilv-il iv: complex deposits that build up on the surface

of the electrode lead to the growth of uneven dendrites. To prevent problems

caused by the reactivity of lithium, battery makers replace lithium metal with

lithium in its ionic state. In this cell, lithium-ions are intercalated into the active

material of the electrode rather than being plated out as a metal. At the beginning

of discharge, the negative electrode is charged with lithium ions (Li+) while the

positive electrode is ready to accept lithium ions. During discharge, the lithium

ions leave the negative electrode and enter the solution phase, while in the positive

electrode region lithium ions in the solution phase intercalate into the active

material. This behavior results in a concentration gradient, which drives lithium

ions from the negative to the positive electrode.0' 11 A schematic representation

of a typical Li-ion cell is shown in Figure 2-2. Much effort has been expended

in the last decade to improve these batteries, with research devoted to all three

components of the cell. These have resulted in significant improvements in the

performance and cost.









The newest battery technology is the lithium solid polymer cell, in which the

electrolyte is integrated into a polymer plastic separator between the anode and

the cathode. The electrolyte is a polymer composite containing a lithium salt.

Although the energy density of the solid polymer electrolyte can be similar to that

of the ordinary lithium cells, they have the advantage that they can be shaped to

fit the space available. In addition, solid polymer batteries are environmentally

friendly, lighter because they have no metal shell and safer because they contain

no flammable solvent. However, the overall performance of a lithium rechargeable

battery depends on the choice of positive electrode, negative electrode, electrolyte,

and the electrode-electrolyte interfacial properties.2

2.2.1 Anode Materials

The essential reaction of metallic lithium anode is very simple


Li Li+ + e- (2.1)


but, in spite of this simplicity, the practical application of Li metal to rechargeable

anode has been very difficult because Li metal tends to deposit as dendritic

structure during charge, and the disordered metallic deposit gives rise to a poor

coulombic efficiency. Extensive research has been dedicated to obtain better

anodes.12-16 The research has concentrated in three areas: improvements of

carbonaceous materials; utilization of metallic lithium with stable electrolytes; and

search for new materials.10

Most of the improvements in anodes actually implemented over the last decade

have involved carbonaceous materials, in concert with 1 iv, 1 oxides. The reasons

that lie behind the commercial success of carbon-based negative electrodes include

the relatively low inherent cost of carbon, its excellent reversibility for lithium

insertion, and the formation of a protective surface film with many electrolyte









solutions. The choice of anode is based on the need for fast insertion kinetics and a

redox potential vs. lithium which provides a sufficiently large cell voltage.

Despite the advances made in carbons, it has been recognized for some years

that new anode materials are needed. An increase in the operating voltage above

current values for carbonaceous materials is highly desirable to inhibit lithium

metal deposition that can occur at fast rates. A large number of alternative

possibilities for negative electrodes have been recently reported in the literature

that may lead to new directions in meeting the anode challenge.

The goal of developing rechargeable batteries with a metallic Li anode has

become more definite. One of the fields in which a Li metal anode is attracting

much interest is that of polymer battery systems. To obtain a satisfactory perfor-

mance from metallic Li anode it is necessary to optimize chemically and physically

the interfacial structure between Li anode and the electrolyte. It is necessary to

choose appropriate electrolytes to promote the formation of a passivating li. -r that

will protect against side reactions, but that do not impede the diffusion of Li ions

across the interface during cycling. Another approach to the improvement of Li

cycling efficiency is including some additives in the electrolyte formulation. This

method can control or modify the thickness, morphology and chemical composition

of the Li surface 1-.-r.1o A detailed description is provided in Subsection 2.2.4.

Besides the chemical factors such as electrolytes and additives, some physical fac-

tors have an influence on the Li interface structure. The temperature and stacking

pressure on the anode have an influence in the surface morphology, which can

improve the cycling efficiency.

2.2.2 Cathode Materials

Over the last decade the research on cathode materials has been intense, and

remarkable success has been met at the practical and at the theoretical level.1'17









The classical criteria for selecting an efficient cathode are based on thermodynamic,

kinetic, and practical considerations include

1. The cathode must intercalate or insert Li+.

2. The cathode has to have low electron energy and low site energy for Li+.

3. The electrode potential should have limited variations as a function of Li+
content.

4. The number of sites for Li+ has to be high and the hosting molecule has to
feature low weight and high density.

5. The couple diffusion of electrons and Li+ in the host, as a function of
concentration gradient, must be fast enough to grant a good rate capability.

6. The Li+ intercalation/insertion has to be reversible to allow cyclability.

7. The cathode must be stable towards the electrolyte over the entire operating
voltage range.

8. The cathode material should be easy to synthesize, should be non toxic,
and of low cost.

9. The synthesis must be reproducible and should allow the production of a
material with the desired particle size.

10. The material has to be easily processed into a practical electrode.

11. A favorable interface should be formed between the cathode and the
electrolyte.

2.2.3 Electrolyte

The selection of the electrolyte is determined by the electrode reactions. The

basic requirements of a suitable polymer electrolyte for lithium batteries areo1

1. High ionic conductivity to minimize the cell resistance and resistive heating
of the device.

2. High chemical stability to prevent decomposition of electrolyte on the
surface of a highly reducing anode material and a highly oxidizing cathode
material.

3. Electrochemical stability to tolerate the high voltage difference between the
anode and the cathode (> 4 V).

4. Low melting point to provide sufficient conductivity at sub-ambient
temperatures and prevent solidification an phase separation.









5. High boiling point to provide safety and prevent explosions resulting from
high pressure build-up in the cell.

6. Non-toxicity to be accepted environmentally for ease of handling, mass
production, and waste treatment.

7. Low cost, safety, and ease of processing.

Although chemical stability of the bulk electrolyte is a requisite for long

battery life, the most common electrolytes (liquids and polymer) are not stable

at the open circuit potential for lithium metal. However, the formation of surface

films at the electrode surface limits the extent of these reactions and reduces the

rates to levels that allow the useful cycle and calendar life of the battery to be

achieved.10', 1820 But it is important to understand that these surface films provide

1-.. -i~ with different transport properties from the bulk electrolyte and through

which the lithium ions must pass (See Figure 2-2). These lv'--is are usually

referred to as the solid electrolyte interface or SEI, and they usually pll i, a crucial

role in both the cycle life and also in the rate capability of the battery. A detailed

description of the SEI is presented in Section 2.2.4.

The weight of the electrolyte is a 1n i i r contribution to that of the complete

battery.21 Battery-makers are interested in finding v-i-, of packing battery power

into smaller packages that enable electronics manufacturer to continue their drive

for miniaturization. For this reason ionically conducting polymers have been the

focus of much fundamental and applied research for many years.2',3,18,19,22-27

Since the introduction of polymer electrolytes in the 1970's many researchers

have attempt to identify solid polymer electrolytes formulations which would

combine high ionic conductivity, good mechanical properties, and good interfacial

behavior with lithium metal electrodes.2'24 The most common polymer electrolyte

investigated has been based on p.. ,li- Iylene oxide (PEO) which is commercially

available in a relatively pure state at reasonable cost.o1 The efforts were focused

on the bulk properties of polymer electrolytes based on combinations of PEO with









various lithium salts and on the characteristics of electrochemical processes at the

electrode-polymer electrolyte interface. This led to the development of high ionic

conductivity polymer systems that are almost competitive with liquid electrolytes

at elevated temperatures.

However, at elevated temperatures, PEO-based electrolytes as well as other

polymer-based electrolytes are mostly in their amorphous state, in which the

mechanical properties are relatively poor.24 The addition of ceramic fillers has

shown an improvement of the mechanical properties of the electrolyte. Also, the

ceramic fillers induce an enhancement of the Li transference number and of the

ionic conductivity.24'28 It has also been shown that the preparation procedure of

the electrolyte and the environment p li' a iin .i role with respect to the addition

of the filler.24

The implementation of polymer electrolytes in commercial batteries requires

an increase in ionic conductivity at ambient temperature. A lot of research has

been done to find polymer electrolytes that have high ionic conductivity at ambient

temperature. For example, Kim and Smotkin studied the effect of plasticizers

on transport of electrochemical properties of PEO-based electrolytes for lithium

rechargeable batteries.2 The results showed that the plasticized electrolytes have

higher conductivity and higher diffusion coefficient at the selected temperatures.

Also, the interfacial resistance between the polymer electrolyte and the Li metal

surface decreased. But, the cationic transference number was reduced with the use

of some plasticizers.

The design of polymer electrolytes for lithium batteries is a complex problem

that needs to be approached with a combination of sophisticated modeling,

diagnostic techniques and significantly greater synthetic effort than has been

applied in the past. Only then can the rates of side reactions be related to lifetime

issues.10









2.2.4 Electrode-Electrolyte Interface

Peled was the first to introduce and name the electronically protective, but

conductive, interfacial l1 -r as solid electrolyte interface or SEI.1819 According to

Peled, the SEI is a heterogenous film, consisting of a mosaic of numerous individual

particles of different chemical composition being in partial contact with each other

at the grain boundaries. The chemical composition of these particles depends

on the type of solvent, or solvent mixture, type of electrolyte salt and on the

impurities. The lithium conduction through the SEI is provided by ionic transport

mainly at the grain boundaries of inorganic compounds such as LiC1, LiF and Li20

or near the surface of the particles via lattice defects. A schematic representation

of the electrode-electrolyte interphase is presented in Figure 2-3. In Figure 2-3(a)

a native oxide film can be observed. It is well accepted that commercial lithium

foils have a native oxide l-iv-r that is composed primarily of a Li20 l- v r and

Li2CO3/LiOH l-iv-, .10,29 The polymer electrolytes have a rough surface, and so,

in contact with lithium, some spikes penetrate the oxide l- v r and the lithium

metal such that a SEI is formed at the Li/SPE interface. In other areas, softer

contacts between the SPE and lithium are formed. In these areas the SEI forms

on the native oxide 1.- -r, or as a result of the retreat of lithium during corrosion,

the native oxide l-iv-r breaks and the gap is filled by the formed SEI. The net

result is that only a fraction of the lithium surface is in intimate contact with

the polymer electrolyte. At open circuit potential, the rate limiting step for the

deposition-dissolution process is Li+ transport in the SEI.

The components of the interface 1l- r depend on the nature of the electrolytes

and the additives added to the electrolyte. Tables 2-1 and 2-2 summarize the effect

of several liquid electrolytes and additives in the composition of the SEI lI--r.




























(a) (b)

Figure 2-3: Schematic presentation of the lithium-polymer electrolyte inter-
phase.(a) Overall interphase (b) Small segment of the interphase. Figure repro-
duced from: E. Peled, D. Golodnitsky, G Ardel and V. Eshkenazy, The SEI Model
Application to Lithium Polymer Electrolyte Batteries, Solid State lonics 40 (1995)
2197-2204.





Table 2-1: ('!i ,, I composition of passivation films formed on lithium metal
surface in contact with different electrolytes.

Electrolyte Passivation Film
Without electrolyte (native film) LiOH, Li2CO3, Li20
General PC systems CH3CH(OCO2Li)CH2OCO2Li
PC/LiPF6 LiF, Li20, LiOH
PC/LiClO4 Li2CO3, LiOH, Li20, LiCl, ROCO2Li,
LiCHC1CHC1, LiCH2CHClCH2CI
General EC systems (CH20CO2Li)2
General DMC systems CH3OCO2Li
General DEC systems CH3CH2OCO2Li, CH3CH2OLi
SO2/LiAlC14 Li2S204, Li2SO3, LiS,O6,Li2S205

Table reproduced from: G.Nazri and G. Pistoia, Lithium Batteries: Science and
T ,, ,I,., ,. (Norwell,MA: Kluwer Academic Publishers, 2004).


Pol'mter
Ele.tfcWte


Metal




N
"___ _


Freshly Formed SEI





















Table 2-2: C'! ii I. composition of passivation films on the Li metal in contact
with additives used to enhance the cycling efficiency.

Group Additive Major Acting Mechanism
HF Formation of LiF l-1 v,-

Inorganic All3, MgI2, SnI2 Formation of Li-alloy liv.r

S-2 Formation of protecting film
2Me-furan, 2Me-THF, Pyridine Surface adsorption, formation of
derivatives, Sipyridyl derivatives organic protecting film, solvation
of Li+

Organic Cetyltrimetylammonium chloride Electrode surface adsorption

Nonionic surfactants, Crown Ethers Electrode surface adsorption,
solvation of Li+

Benzene Electrode surface adsorption
CO2 Formation of Li2COs 13,-i
Gas
N20, CO Formation of protecting film

Table reproduced from: G.Nazri and G. Pistoia, Lithium Batteries: Science and
T 1. ,, .,.i.,/; (Norwell,MA: Kluwer Academic Publishers, 2004).















CHAPTER 3
ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY THEORY

Impedance can provide valuable insights into the behavior of a large v ,i i I v of

substances, components and systems. In many materials the impedance response

varies as the frequency of the applied voltage changes. This may be due to the

physical structure of the material, to chemical processes within it, or to a com-

bination of both. Thus, if a measurement of impedance over a suitable frequency

range is made, and the results are plotted on suitable axes, it is possible to relate

the results to the physical and chemical properties of the material. This technique

is known as electrochemical impedance spectroscopy and is use to investigate

materials and chemical mechanisms.

Electrochemical impedance spectroscopy is a measurement of the conductive

and dielectric properties of electroactive systems over a wide range of frequencies.

Its popularity and applicability has increased dramatically over the past 30 years

with the introduction of fast-response potentiostats and frequency response ana-

lyzers. Impedance spectroscopy has been applied extensively in electrochemistry,

especially in battery and sensor research.30'31

This chapter provides a basic understanding of the concepts that are involved

in electrochemical impedance spectroscopy. Detailed discussions of the technical

and theoretical issues associated with electrochemical impedance spectroscopy are

available elsewhere.30-41













(a) (b) (c)

Figure 3-1: Passive elements that serve as a component for an electrical circuit.


3.1 Impedance Response of Electrical Circuits

Impedance is defined to be the ratio of the voltage phasor, V, to the current

phasor, I. Ohm's law can be expressed in phasor form a,-' 43


Z = (3.1)
I

The impedance Z indicates opposition to current flow and it has the units of

ohms. Impedance is in general a complex number and not a function of time. It

completely defines the element's behavior in an ac circuit.

Electrical circuits can be constructed from the passive elements shown in

Figure 3-1. For a pure resistor, as the one shown in Figure 3-1(a), the impedance

response is expressed as

Zresistor = R (3.2)

for the inductor shown in Figure 3-1(b)


Zinductor = jL (3.3)


and for the capacitor shown in Figure 3-1(c)

1
Zcapacitor (3.4)


The impedance response of resistors, inductors, and capacitors are used to con-

struct the impedance response of circuits.









3.2 Experimental Considerations

EIS is a small-signal technique in which a sinusoidal current or potential

perturbation is imposed on the system of interest and the corresponding potential

or current response is measured, Figure 3-2. Comparison of the input and output

signals provides the impedance at a given perturbation frequency.

An appealing feature of EIS is that systems with characteristic time constant

distributed over a wide range of time scales can be studied. A variety of capac-

itive and resistive properties of the material can be determined in a single brief

experiment by analyzing the response over a wide frequency range. Also, EIS is a

non-destructive method for the evaluation of a wide range of materials, including

coatings, anodized films, corrosion inhibitors, batteries and fuel cells.32,33

The main difficulty in the impedance studies originate from the complexity of

electrochemical systems. An important requirement for a successful performance

of the investigation is that the system must be linear.34 Electrochemical systems

are usually non-linear but the requirement for linearity could be fulfilled if the

amplitude of the input signal is small enough to allow linearization of the response.

In addition, the input signal should be sufficiently small as to avoid irreversible

changes in the system properties (See Figure 3-2). The frequency range must

be wide enough to attain the high frequency limit of the impedance equal to the

electrolyte resistance.

The electrochemical cell pl .- a in .i wr role in the success of the impedance

experiments. The cell can consist of 2, 3 or 4 electrodes immersed in an electrolytic

solution contained in a vessel, Figure 3-3. The electrochemical interface whose

impedance is measured is generally located between the electrolyte and the working

electrode.32 The counter electrode is used to allow current to flow through the

cell. It is usually a large surface area electrode with very low impedance. When

concentration gradients in the electrolyte solution affect the impedance response











10



------------


Potential


SEo + AE sin ot


Figure 3-2: Small signal analysis of an electrochemical non-linear system.


of the system, a 3 electrode cell should be used. The reference electrode should be

placed as close to the system as possible. A 4-electrode cell is used when transport

through a membrane is studied. The potential difference across the membrane is

measured by means of two reference electrodes, one on each side of the membrane.

The modulation technique also 1p i, an important role in the success of the

experiments. There are two modulations, potentiostatic and galvanostatic. The


2-electrode
Working
Electrode


Counter
Electrode


3-electrode
Working
Electrode


Ref 1



Counter
Electrode


4-electrode
Working
Electrode


Ref 1
Ref2


Counter
Electrode


Figure 3-3: Electrode configuration in an electrochemical cell.


10 + Al sin col









potentiostatic modulation is the standard approach; in this modulation linearity is

controlled by the potential. The galvanostatic modulation is good for nonstationary

systems, like corrosion and drug deliver. However, this modulation requires a

variable perturbation amplitude to maintain linearity. Additional considerations

include the use of short leads, good and shielded wires, and the use of a faradaic

cage when the impedance of the system is large, resulting in small currents.

3.3 Error Structure

Interpretation of the impedance data requires a process model and a quanti-

tative assessment of the error structure. The error structure of the measurement is

used implicitly in regression analysis and has a significant influence on the quality

and amount of information that can be extracted from impedance data. The sto-

chastic errors can also influence the Kramers-Kronig relations for determining the

internal consistency of the data.40

3.3.1 Kramers-Kronig Relations

The Kramers-Kronig relations are integral equations, which constrain the real

and imaginary components of the impedance for systems that satisfy conditions

of causality, linearity, and stability. This means that the system response to a

perturbation cannot precede the perturbation, that the system responds linearly

with respect to the perturbation, and that the perturbations to the system do not

grow. In principle, the Kramers-Kronig relations can be used to determine whether

the impedance spectrum of a given system has been influenced by bias caused, for

example, by time-dependent phenomena or instrumental artifacts.

3.3.2 Measurement Model Approach

The measurement model approach developed by Agarwal et al. is used to

identify the error structure of the measurements. The objective of the measurement

model toolbox is to account for measurement errors in the analysis of electrochem-

ical impedance spectroscopy. The measurement model has been used to determine









whether the residual errors in the regression are due to an inadequate model, to

failure of data to conform to the Kramers-Kronig assumptions, or to noise. Since

the model satisfies the Kramers-Kronig relations, it can be used to identify portions

of the spectrum that were inconsistent with these relations. The measurement

model provides more than a preliminary analysis of impedance data in terms of

the number of resolvable time constants and .,-vmptotic values. The measurement

model can be used as a filter for lack of replicacy that allows accurate assessment

of the standard deviation of impedance measurements. This information is critical

for selection of weighting strategies for regression, provides a quantitative basis

for assessment of the quality of fits, and can guide experimental design. The er-

ror analysis provides information to be incorporated into regression of process

models. The analysis consists of two parts; the first one is a preliminary analysis

to determine the error structure and the second one is to assess consistency with

the Kramers-Kronig relations. A detailed description of the measurement model

approach and the results obtained from the experiments are presented in C'! lpter

4.

3.4 Impedance Spectroscopy Data Analysis

Graphical methods provide the first step toward interpretation and evaluation

of impedance data.44 Impedance data are usually represented in impedance

plane or Nyquist format, as shown in Figure 3-4. This plot shows the impedance

response of a symmetrical lithium based battery cell with electrolyte composition

PEO16LiTFSI. In this plot, the limits of the real part of the impedance represent

the electrolyte resistance Re at high frequency and the sum of the electrolyte

resistance and the charge transfer resistance Re + Rf at low frequency. One

disadvantage of this type of format is that the frequency dependence is hidden.

Nyquist plots are very popular because the shape of the locus of points yields

insight into possible mechanisms or governing phenomena. For example, if the locus











-300 -115 Hz


S-200


N' -100 1 Hz


0
I I I a I a I I I a I I

0 100 200 300 400 500 600 700 800

Z / cm2


Figure 3-4: Impedance-plane or Nyquist representation of impedance data for Cell
NA-ATi. The open symbols represent measurements taken 15 hour after the cell
preparation. The half-full symbols represent measurements taken approximately 48
hours after the cell preparation.

Rf







Cdl

Figure 3-5: Schematic representation of a simple reactive system.


of points traces a perfect semicircle the impedance response corresponds to a single

activation-energy-controlled process, such as that shown in Figure 3-5. A depressed

semicircle represents a system with characteristic time constant distributed over a

wide-range of time scales.

Other common representations are the Bode plots shown in Figure 3-6 and

plots of the real and imaginary part of the impedance as functions of frequency

shown in Figure 3-7. Bode plots have the advantage that the frequency depen-

dence is shown in the graph. Frequency is generally presented on logarithmic scale

to reveal the important behavior seen at lower frequencies. In electrochemical



















600


400


..,,L ..,.,,L ,, .. .._ .. ., ..., ...o,,,, ,
103 102 10 100 101 102 103 104 105 106
f/Hz


-30 -


-101


103 102 101 100 101 102 103 104 105 101
f/Hz


Figure 3-6: Bode representation of impedance data, magnitude and phase angle,
for Cell NA-ATi, respectively. The open symbols represent measurements taken
15 hour after the cell preparation. The half-full symbols represent measurements
taken approximately 48 hours after the cell preparation.


10 10 10 100 10 102 10 10 10 10
f/Hz


01 10
10 10:


101 100 101 102 103 104 105 101 107
f/Hz


Figure 3-7: Plots of the real and imaginary part of the impedance as a function
of frequency for Cell NA-ATi. The open symbols represent measurements taken
15 hour after the cell preparation. The half-full symbols represent measurements
taken approximately 48 hours after the cell preparation.


_ _ _ _ _









systems Bode representation has serious drawbacks. For example, the influence of

the electrolyte resistance confounds the use of phase angle plots to estimate charac-

teristic frequencies. However, if an accurate estimate of the electrolyte resistance is

available, a modified Bode representation that can be used to avoid the drawbacks.

The plots of the real and imaginary part of the impedance as functions of frequency

have the advantage that the characteristic frequencies can be easily identified.

When the frequency is plotted in logarithmic scale for the imaginary part of the

impedance, the slopes at low and high frequency are +1 and -1, respectively,

for a simple reactive system as the one shown in Figure 3-5. Departure from 1

provides an indication of distributed processes. Observation of multiple maxima

shows that the data must be interpreted in terms of more than one process.

A detailed analysis of the experimental data is presented in C'! lpter 5.

3.5 Impedance Response Models

Interpretation of impedance data is generally guided by use of equivalent

electric circuits. Both simple lumped-parameter circuits and more complex finite

transmission-line circuits have been used. The formulation of the overall circuit

network involves the semi-empirical addition of circuit elements until the model

provides an adequate description of the impedance response. Upon the estab-

lishment of an appropriate network, a deductive process is used to relate the

properties of the system to the individual circuit elements, which comprise the

overall network. A limitation of such electric circuit models is that an unambigu-

ous interpretation of the data is not possible. For particular impedance spectra,

more than one circuit network can be constructed which would provide equivalent

statistically valid fits; and each of these models can arise from extremely different

interpretations. This reduces the efficacy of circuit networks for interpretation of

the physical properties and the kinetic processes of the system. A less ambiguous

approach for interpreting impedance data is to develop physico-chemical models









based on the governing equations, which describe the physics and chemistry of the

system. To develop such a refined model it is necessary to identify the reaction

mechanism or if not known assume kinetic and transport mechanisms. Then,

expressions for the current contributions will be written.

The faradaic current density, if, can be expressed in terms of a steady time-

independent value and an oscillating value as

if = if + Re{ife } (3.5)

The oscillating component of the current can be calculated using a Taylor series

expansion about the steady-state value as

tif = + (8c j i,O + sk V (3.6)
cf) i,Ok V,cj+i,Of V,c ,O1k

where V is the interfacial potential, cio is the local concentration of the bulk

species and Ok is the surface coverage of the absorbed species. The charging current

is given by
dV
i = if + cdl (3.7)

and

i if + joCdlV (3.8)

To obtain an expression for impedance it is necessary to solve equation (3.6).

Using the reaction mechanism or the assumed mechanisms, equations that relate

concentration and surface coverage to voltage can be introduce to obtain an

equation of current in terms of interfacial voltage. Then the impedance of the

system is obtained by comparing the oscillating term of the cell potential, V, and

the oscillating term of the current, i. The cell potential is defined as


U = iRe + V


(3.9)









and

U iRe + V (3.10)

where Re is the electrolyte resistance. The impedance of the system is defined to be

U
z = (3.11)


Substituting equations (3.8) and (3.10) into equation (3.11), an expression for the

impedance of the system in term of electrolyte resistance, interfacial potential and

charging current is obtained as
V
S= Re + (3.12)

If the kinetic and transport mechanisms were assumed the impedance relation

obtained from equation (3.12) is compared to the experimental values. If a good fit

is obtained the relation is accepted as the model for the system. However, if a good

fit is not obtained new kinetic and transport mechanisms are assumed and all the

procedure is repeated until a good fit is obtained.

The process models developed for the experimental data is presented in

C'!h ,pter 7.















CHAPTER 4
EXPERIMENTAL DESIGN AND ERROR ANALYSIS

There exists a rich literature on research of battery systems, both, exper-

imental and modeling work. Different techniques have been used to study the

mechanisms of transport inside batteries. Some of the techniques include constant

current discharge, pulse discharge, cyclic voltammetry, abuse tests, and impedance

spectroscopy.1-3, 5, 6, 11, 12, 14, 15, 17, 25, 26, 28, 4563

In this work EIS was used to study symmetric lithium cells. The goal was to

develop a model that describes the physics and chemistry of the system, but that

can be regress to the experimental data. This model would be the first step toward

the development of refined equivalent circuit models for lithium-based battery

systems.

Experiments were performed to provide reliable impedance data that were

used to test the models. In this chapter a detailed description of the experimental

component of this work is provided. It include a description of the experimental

system and experimental design, as well as, a description of the method used to

test the quality of the data. A detailed description of the model development is

presented in C'!i pter 7.

4.1 Experimental System

The system used in this work was a symmetric Li|SPE|Li coin cell. All cell

assembly was done in dry room atmosphere (-35F dew point, nominal). A

schematic diagram of the battery cell assemble is shown in Figure 4-1. A two

lithium-metal-electrode system was used since researchers have stated that a two

electrode system will introduce much less error than what would result from an

ill-placed third (reference) electrode.2'17 The cell was built using a thin, insulating












cover / gasket assembly

0_ wavespring
spacer
lithium disc (%" diameter)
SPE disc (%" diameter)
S polyester mask
lithium disc (%" diameter)

Scan



Figure 4-1: Schematic of the cell sandwich. The cell was build using a thin, insu-
lating mask to separate the electrodes. The mask was intended to prevent short
circuit.


mask to separate the electrodes and to provide a well defined reservoir for the

melted electrolyte. This mask was intended to prevent short circuits that were

observed in earlier coin cell work done at NASA Glen Research Center.6 The mask

was centered on top of the anode and covered with the SPE disc. The lithium

electrode was then centered on top of the SPE disc before closing the cell. A wave

spring provided an internal pressure of 18 psi. The mask outside diameter was

oversized to assist in the alignment of the electrode sandwich in the cell can. The

inside diameter established an apparent active area of 1.27 cm2

The electrolyte system used in this research was p1l.. Ilylene oxide (PEO)

plus a lithium imide salt (16 : 1 ether-oxygen:Li). PEO has a regular repeated

structure -CH2CH20-, is crystalline and has a fusion temperature Tf of 339

K. The imide salt formula is LiN(SO2CF3)2. This salt is formally called lithium

bis-(trifluornii. I1 i,. 1-ii lf..i.l)-imide, but most people abbreviate this to LiTFSI.

The electrolyte was prepared using PEO (Aldrich 6E5 Mv), LiTFSI (3M HQ-

115) and acetonitrile as the solvent. PEO and acetonitrile were used as received.

The grade of the PEO contains 200 to 500 ppm of Butyl hydroxyl toluene (BHT)

as stabilizer. LiTFSI salt was dried for 4 hours under vacuum at 160C. PEO was

















Figure 4-2: Schematic of the experimental setup used for the electrochemical im-
pedance tests.

pre-dissolved as a 12.3 w/w solution in acetonitrile before addition of 23.4 w/w

LiTFSI-acetonitrile solution. Quantities were established to provide a 16 : 1 EO:Li

ratio after evaporation of the acetonitrile solvent. This is equivalent to 28.9 w/w

LiTFSI in the solid electrolyte.

SPE films were routinely prepared by casting the above solution in Teflon

Petri dishes and allowing the solvent to evaporate overnight. Casting and solvent

evaporation steps were conducted at dry room atmosphere (< 1. RH). Cured SPE

films were dried overnight under vacuum at 450C and stored in a desiccator before

use. This process produced rubbery, freestanding films with a strong self-adherent

character. Dry film density at room temperature was 1.26 g/cm3, giving a salt

concentration of 1.27 mole LiTFSI/liter.45

Some of the experiments were conducted at NASA Glen Research Center and

the others were conducted at the University of Florida Department of C'!, iii- dl

Engineering. A Solartron 1250 Frequency Response Analyzer interfaced with a

Solartron 1286 Potentiostat was used to collect the data. For the experiments

conducted at NASA facilities, Z-plot was used for experiment control and data ac-

quisition, while software written in-house in Labview was used for the experiments

conducted at the University of Florida. The battery cells were placed in a Thermal

Product Solutions T-U JR environmental chamber to ensure constant temperature

throughout the experiments (See Figure 4-2).









4.2 Experiment Design

The experimental work was divided into four categories. The first set of

experiments were conducted at the NASA Glen Research Center. They were steady

state impedance measurements that were taken approximately 15 and 48 hours

after each cell was prepared. Open circuit potential measurements of four cells

were made with six replicates of each scan. The cells were stored at a constant

temperature of 800C for about 10 hours. Then the temperature was decreased to

60C and kept constant throughout the experiments. The frequency range of the

measurements varied from 1 MHz to 0.02 Hz.

The second and third set of experiments were conducted at the University of

Florida, Department of Chemical Engineering. The second set of experiments were

steady-state impedance measurements where the temperature was increased from

60C to 900C in 5C increments. Open-circuit potential measurements of four cells

were performed with six to eight replicates of each scan. The frequency range of

the measurements varied from 65 KHz to 0.003 Hz. The third set of experiments

were transient impedance measurements performed at a fixed temperature. The

measurements were taken consecutively for 48 hours. Two cells were tested. The

temperatures used were 60C and 700C. The frequency range of the measurements

varied from 65 kHz to 0.003 Hz. In these experiments, the temperature inside the

environmental chamber was equilibrated to the desired value and then the cells

were placed inside the chamber. Twenty-two to tweny-four sequential impedance

spectra obtained at the open-circuit potential were analyzed.

The last set of experiments were conducted at NASA Glen Research Center.

They were steady state impedance measurements using blocking electrodes. The

experiments were designed to test the hypothesis that the impedance response

of the symmetric lithium cells was due to the electrolyte and not to the reaction

on the lithium metal surface. The cell consisted of PEO-LiTFSI electrolyte







30

Table 4-1: Summary of the experimental design.
surements is presented.


The number of sequential mea-


Group Temperature (C) Time (hour) 25 60 65 70 75 80 85 90 100
1 NA-ATi 15 6
48 6
NA-BTi 15 6
48 6
NA-CTi 15 6
48 6
NA-DTi 15 6
48 6
2 UF-ATe 8
UF-BTe 8 6 7 6 6 6 6
UF-CTe 6 6 12 6 6 6 6
UF-FTe 6 6 6 6 6 6 6
3 UF-DTr 24
UF-ETr 22
4 NA-AB 6 6 6
NA-BB 6 6 6 6


sandwiched between two steel electrodes. The cells was exposed to an increasing

temperature range varying from 80C to 100C in 10C intervals. Six sequential

impedance spectra obtained at the open-circuit potential were collected per set.

The frequency range of the measurements varied from 65 kHz to 0.003 Hz. Table

4-1 summarizes the experiments design.

4.3 Measurement Model Analysis

The measurement model analysis was used in this work to test the quality of

the impedance data. The analysis consists of identifying the error structure of the

measurements and then using that error structure to determine whether the data

conform to the Kramers-Kronig assumptions. To determine the error structure

of the measurements, a Voigt model was fitted to the experimental data. Voigt

elements in series with the solution resistance, i.e.,


Rk (4.1)
Z Ro+ t k
1 + jeo-k










Tl=RlC1 T2=R2C2 Tk=RkCk TK=RKCK



Ro,
R1 R2 Rk RK


Figure 4-3: Schematic representation of a Voigt circuit used as a measurement
model.

were added as shown in Figure 4-3. With a sufficient number of parameters, the

Voigt model is able to provide a statistically significant fit to a broad v i' i. Iv of

impedance spectra.64'65 However, the model should not be associated with the set

of deterministic or theoretical parameters for a given system.

The use of the measurement model for analysis of the error structure requires

replicate measurements of the impedance response using the same measurement

frequencies for each replicate. Most electrochemical systems of practical importance

are nonstationary and the resulting trending biases direct calculation of statistical

quantities such as mean and variance. To eliminate the contribution of the drift

from scan to scan, the measurement model was regressed to each scan individually.

For the system to be consistent with the Kramers-Kronig relations, it has to evolve

sufficiently slowly that the change during one complete scan is insignificant.

4.3.1 Steady State Experiments at Constant Temperature

The impedance response and the model obtained using the measurement

model technique of Cell NA-ATi are shown in Figure 4-4. Two data sets are

presented in this figure, one was obtained approximately 15 hours after the cell

preparation and the other was obtained approximately 48 hours after the cell

preparation. The first ten points at high frequency were eliminated because those

points were corrupted by instrument artifact effects and were not consistent with

the Kramers-Kronig relations. The model agreed well with the experimental data.










300 .
250
200
0 150
100
50

0 200 400 600 800
Z / cm2


Figure 4-4: Measurement model prediction of impedance data for Cell NA-ATi.
The open symbols represent measurements taken 15 hours after their preparation.
The half-full symbols represent measurements taken approximately 48 hours after
the cell preparation. The solid line represents the measurement model fit.


The error structures for the two data sets are shown in Figure 4-5. These plots

show the error structure for the real and imaginary part of the impedance, as well

as the error structure model. The error structure model was calculated to be


a = 0.001035Z + 0.000467(Zr 50.1038) (4.2)


This model was calculated using the error structures of the four cells.

For the measurements taken 15 hours after the cell preparation, the standard

deviations of the real and imaginary part of the impedance at low frequencies were

not equal, Figure 4-5(a). This lack of equality of standard deviation for the real

and imaginary parts of the impedance -i-i. -I that the data were influenced by

non-stationary behavior. The standard deviation of the real and imaginary part

of the impedance was equal for the measurements taken 48 hours after the cell

preparation, Figure 4-5(b).

The second part of this a'n 1,-i ; was to assess for the Kramers-Kronig consis-

tency. A Voigt element model was fitted to the imaginary part of the impedance.

The experimentally determined error structure was used to weight the model.

When an adequate description was obtained, the model was used to predict the

real part of the spectrum. From the estimated variance of the model parameters, a








33




00 01 ee e 0 1-



--- -- a-- a%- -- -- --- --- ---1 o-- & s- a o -o -
10 10 10 0 102 10 10 10 100 10 102 10 10 10 10





Figure 45: Error structure fr Cell NAATi measurements taken 15 and 48 hours







The solid line represents the error structure model fit.
Se e e. e



















being stationary during the course of the experiment. If a significant portion of

the predicted data falls outside the 95opercent confidence interval, the system is

likely to be non-stationarylf-f39,41,66 Otherwise, if any data point falls outside the

95-percent confidence interval it will be eliminated orm the data set. Only the

portions of the spectra that are consistent with the Kramers-Kronig relations were

used for further analyses.



and 4t7. The model predicted every data point within the 95'-, confidence interval

which means that, within the error structure of the measurement, all data can be

said to be consistent with Kramers-Kronig relations.
s9c 0



001 ae





















4.3.2 Steady State Experiments at Increasing Temperatures




















The same procedure was followed for the steady state experiments where the

temperature was increased sequentially. The temperature range studied varied from

60C to 85C. Temperatures above 85C wee 5 er not cnid env.e because the behavior








34












250

200

150

100

50

0

103 102 101 100 101 102 103 104 105 106
f/ Hz

(a)


250

200 -

150

100

50



103 102 10- 100 101 102 103 104 105 106
f/ Hz

(b)


Figure 4-6: Measurement model fit of the imaginary part of the impedance data for
Cell NA-ATi. The open symbols represent measurements taken 15 hours after their
preparation. The half-full symbols represent measurements taken approximately 48
hours after the cell preparation. The solid line represents the measurement model
fit and the dashed lines represent the 95.!' confidence interval.











800 -
800 -
700
700
600
600
500
4 E 500 .
O 400 o
2 2400
N 300 N 300
200 200
100 100
0 0
102 101 100 101 102 103 104 105 106 102 10 100 101 102 103 104 105 106
f/Hz f//Hz

(a) (b)

Figure 4-7: Measurement model prediction of the real part of the impedance data
for Cell NA-ATi. The open symbols represent measurements taken 15 hours after
their preparation. The half-full symbols represent measurements taken approx-
imately 48 hours after the cell preparation. The solid line represents the mea-
surement model prediction and the dashed lines represent the 95.!' confidence
interval.

Table 4-2: Error structure models determined for Cell UF-CTe. This cell was ex-
posed to a temperature range varying from 600C to 85C.

Temperature (C) Error Structure Model
60 a- 8.7422E-41Z|
65 a 1.4348E-3|Zj + 1.1404E-3|Zl + 1.1541E-51Z12/R
70 a = 9.2728E-41Z| + 2.8481E-2
75 a = 5.0172E-4 Zj + 6.0620E-4|,Z
80 = 1.1159E-3|Zj + 3.5791E-4|Zr
85 a = 7.6544E-3| Z


of the cells was not stable at those temperatures. The error structures for cell

UF-CTe are shown in Figure 4-8. These plots show the error structure for the

real and imaginary part of the impedance, as well as the error structure model.

A good agreement between the model and the experimental data for the 60C

was not possible, however the real and imaginary errors were equal. The error

structure models for the six data sets are presented in Table 4-2. These models

were calculated using the error structures of the six replicates measured at each

temperature. It was not possible to obtain a single model for all the temperatures.



















10





E
0
o

10

ad


104 103 102 101 100 101 102 103 104 1C

flHz

(a)


A

A AA
A AA fAA AA A

A A
Sa A *A A A
A
A A A
A A'
AA A






103 102 101 100 101 102 103 104

flHz


o



*0 eo ,o*

*, 0
6 eA S

00



103 102 101 100 101 102 103 104

f/Hz

(e)


E 101


o1
C0





102


103 102 10 10 10' 102 103 104
flHz


104 10 10 101 100 101 102 103 104 10

f/Hz

(d)


10




0 10
E



10~
ea


44v











10, 102 101 100 10 102 10 104 105

flHz

(f)


Figure 4-8: Error structure for Cell UF-CTe measurements (a)600C (b)650C

(c)700C (d)750C (e)800C (f)850C. The open symbols represent the real part of

the impedance. The half-full symbols represent imaginary part of the impedance.

The solid line represents the error structure model fit.


* S.D r


r Dr
Sr0 C
Sa' *

BC o


e
00


10









100




(N
E
0



1a





102


10'
E
o



06 10,
U


V i V v
v v




v
,-







37






0 10-1 No .1
E *EE!M
m Bm




oNn
SReal
> 10-2 i n Imaginary



10-4 10-3 10-2 101 100 101 102 103 104 10
f/Hz

Figure 4-9: Error structure for Cell NA-ATi measurements taken 15 and 48 hours
after their preparation, respectively. The open symbols represent the real part of
the impedance. The half-full symbols represent imaginary part of the impedance.
The solid line represents the error structure model fit.


All data points in the set were consistent with Kramers-Kronig relations. The

same procedure was repeated for Cells UF-BTe and UF-FTe.

4.3.3 Transient Experiments at Constant Temperature

This analysis was very important for the transient experiments. The analysis

helped us identify which portions of the spectra were consistent with the Kramers-

Kronig relations and could be used to identify meaningful parameters as functions

of time. This analysis also helped identify how much time was needed to reach

steady state conditions. The error structures for cell UF-DTi are shown in Figure

4-9. This plot shows the error structure for the real and imaginary part of the

impedance, as well as the error structure model. The error structure model was

calculated to be

a = 0.00058896Z + 0.0016651(Z 29.00) (4.3)

This model was calculated using the error structures of the ten of the twenty-four

measurements.












10
104
103
102
101
10,
10
102
103

103 102 101 100 101 102 103 104 105 101
f/Hz

(a)


e






oOo



102 101 100 10' 102 103 104 105 106
f/Hz

(b)


103

102

E 101
& 100

10A
10
06 102 AAA ,

103
10 4 .... ..... ..... .... ..... ..... .... ..
102 101 100 101 102 103 104 105
fl Hz

(c)


Figure 4-10: Error structure for Cell NA-AB measurements (a)800C (b)90C
(c)1000C. The open symbols represent the real part of the impedance. The half-full
symbols represent imaginary part of the impedance. The solid line represents the
error structure model fit.


4.3.4 Blocking Electrodes

This analysis was also done for the blocking electrode measurements. The

error structure models for cell NA-AB are shown in Figure 4-10 and in Table

4-3. These models were calculated using the error structures of the six replicates

measured at each temperature.

This procedure was repeated for every data set. Data point found to be

inconsistent with the Kramers-Kronig relations were not considered for further

analysis.































Table 4-3: Error structure models determined for Cell NA-AB. This cell was ex-
posed to a temperature range varying from 800C to 1000C at 100C intervals.


Temperature (C)
80
90
100


Error Structure Model
a = 3.6053E-41Zjl + 1.7754E-71|Z2/R,
a = 4.5841E-3|Zj + 1.9697E-71ZI2/R,
a = 4.3598E-3|Zj + 1.8417E-71ZI2/R,















CHAPTER 5
EXPERIMENTAL RESULTS AND PRELIMINARY ANALYSIS

Bode plots and complex-impedance-plane plots are the typical graphical

representations when a specific model is not postulated. However, additional

plots, like IR-corrected Bode plots, log-log plots of the imaginary component

of the impedance, and effective capacitance plots can provide guidance for the

development of appropriate physical models.

Equivalent circuit models had been published in the literature that provide

adequate fits to the impedance data. For example, R. Bouchet, S. Lascaud, and M.

Rosso developed an equivalent circuit for the Li|PEO-LiTFSI system, as shown in

Figure 5-1.63 Their work was focused on the evolution of the impedance spectra of

symmetric cells, LilPEO-LiTFSIILi, with aging time at 90C. In their experimental

results, they distinguish four frequency domains,

1. f > 105 Hz: corresponds to the bulk of the electrolyte. This part of the

spectra was perturbed by high frequency inductive effects due to connections.

2. 105 102 Hz: attributed to the formation of a surface I iT.

3. 102 10.1 Hz: not very reproducible.

Connections 1- Electrolyte 2- Passive Layer 3 4- Diffusion
Rb R1 R2
Le R Rd, Id



Cb

Figure 5-1: Equivalent circuit model used to fit the impedance data obtained from
the Li|PEO-LiTFSI system. Figure reproduced from: R. Bouchet, S. Lascaud, and
M. Rosso. An EIS Study of the Anode Li|PEO-LiTFSI of a Li Polymer Battery.
Journal of Electrochemical Society. Vol. 150, No. 10, pp.A1385-A1389, 2003.









4. 10.1-10-4 Hz: Warburg impedance characteristic of the transport of charged

particles.

The proposed equivalent circuit assumed that the phenomena corresponding to the

four frequency domains are in series. The authors claimed that the proposed model

can reproduce his experimental data.

The graphical analysis presented in this section can help accept or reject

models like the one proposed by R. Bouchet, S. Lascaud, and M. Rosso. In this

section we use graphical analyses to obtain an understanding of the physics of the

system.

5.1 Steady State Experiments at Constant Temperature

The traditional Nyquist representation of the impedance data obtained from

Cell NA-ATi are shown in Figure 3-4 and 3-6. The measurements were made

approximately 15 and 48 hours after the cell preparation. The six impedance

replicates at each time are presented in this plot. The electrolyte resistance value

obtained from this plot is 62.0 f- cm2 for the first data set and 59.7 f cm2 for the

second. The shape of the Nyquist plot indicates that the system have characteristic

time constants distributed over a wide-range of time scales.

The alternate real and imaginary representations are presented in Figure 3-7.

The slopes at intermediate and high frequency were calculated for the imaginary

representation. The slopes at intermediate and high frequency were in good

agreement for the six replicates. These slopes have the value of (1 a), and

departure from 1 provides indication of distributed processes. A reduction

in the values of a at intermediate frequency and an increase in the values of a

at high frequency were evident for the measurements taken 48 hours after the

cell preparation. A line with average slope of 0.74 0.0086 was fitted to the

intermediate frequency for the first data set and a line with average slope 0.70

0.0117 was fitted for the second data set. A line with average slope -0.62 0.0081












10

S104

C: 103


102

10o ....-, .....-, ....-,, .....-, .....-, ....-., .....- ....- ....
10-3 10-2 10-1 10 101 102 103 104 105 106
f/Hz

Figure 5-2: Effective CPE coefficient defined by equation (5.1) for Cell NA-ATi.
The open symbols represent measurements taken approximately 15 hours after
the cell preparation. The half-full symbols represent measurements taken approxi-
mately 48 hours after the cell preparation.


was fitted to the first data set at high frequency and a line with average slope

-0.64 0.0091 was fitted to the second data set. The difference in these two slopes

indicates that more than one capacitive time constant can be distinguished in the

system. The value of a can be used to obtain an apparent constant phase element

(CPE) coefficient Qeff
cos(aw)
e =ff 2 -(5.1)
-Q Z(f ) (27rf)1-

The resulting values of the Qeff are presented in Figure 5-2. The absence of a

clearly identifiable .,-,iii !l. .1,! may be attributed to high-frequency instrumental

artifacts. The last ten points at high frequency were disregarded because it is

believed that they were corrupted by such instrumental artifacts. The value for the

CPE coefficient provided in Table 5-2 represents the average over the values for the

remaining 10 highest frequencies.

The value of a can also be used to find the solution resistance Re used in


Stan- ( Z (5.2)
Z,r R,







43


-80 C 103

-60
102
-40

-20 101

0
O -

10 102 101 10 10 10 10 10 10 10 10 10 10 100 10 102 10 10 10 106
f/Hz f/Hz
(a) (b)

Figure 5-3: IR-corrected Bode plots for Cell NA-ATi; a) phase angle, b) modulus.
The open symbols represent measurements taken approximately 15 hours after
the cell preparation. The half-full symbols represent measurements taken approxi-
mately 48 hours after the cell preparation.


to yield the expected ..i-mptotic value for adjusted phase angle given by


*(oo) -90(1- a) (5.3)


The electrolyte resistance can also be used to obtain the adjusted magnitude of the

impedance as shown in


|Z|* (Z Ret)2 + (Z)2 (5.4)


The IR-corrected phase angle and magnitude as a function of frequency are

presented in Figure 5-3. The value of the electrolyte resistance obtained from

this method was approximately 61.2 0.841 Q-cm2 for the first data set and

approximately 58.6 0.011 O cm2 for the second. These values are in good

agreement with the ones obtained from the Nyquist plot. Also, the high-frequency

slope in the magnitude of the impedance plot was equal to the slope used to

calculate the a value.







44

Table 5-1: Average a values obtained using graphical techniques at the intermedi-
ate frequency range for Cell NA-ATi through NA-DTi. Six replicates were used to
calculate the average parameters

Measurement Time / h 15 48
NA-ATi 0.26 0.009 0.30 0.012
NA-BTi 0.35 0.31 0.37 0.0002
NA-CTi 0.28 0.004 0.39 0.0002
NA-DTi 0.31 0.007 0.38 0.36

Table 5-2: Average values of the parameters obtained using graphical techniques
at the high frequency range for Cell NA-ATi trough NA-DTi. Six replicates were
used to calculate the average parameters


Measurement Time / h 15 48
a / dimensionless 0.38 0.008 0.36 0.009
NA-ATi Qeff / MQ-1 cm-2 s1-" 24.8 0.801 18.9 0.010
Re / Q cm2 61.2 0.841 58.6 0.011
a / dimensionless 0.37 0.011 0.35 0.002
NA-BTi Qeff / M -1 cm-2 s1-" 22.6 1.581 17.8 0.452
Re / Q cm2 61.7 0.048 59.8 0.021
a / dimensionless 0.42 0.017 0.39 0.006
NA-CTi Qeff / M -1 cm-2 s1-" 27.6 0.079 27.6 0.833
Re / Q cm2 49.2 0.019 44.2 0.036
a / dimensionless 0.35 0.007 0.36 0.003
NA-DTi Qeff / M -1 cm-2 s1-" 27.7- 23.9 20.0 0.875
Re / Q cm2 41.8 0.016 45.6 0.018


The same procedure was applied to Cells NA-BTi through NA-DTi. A

summary of the parameters obtained from Cells NA-ATi through NA-DTi are

presented in Tables 5-1 and 5-2.

In general, the average result of the a value for the high frequency range was

0.37 0.005 and 0.33 0.040 for the intermediate frequency range. More variability

was observed in the value of a at intermediate frequency than at high frequency.

Also, the value of the effective capacitance, for the four cells, decreased with time.

5.2 Steady State Experiments at Increasing Temperatures

Nyquist and Bode representations of the impedance data obtained from Cell

UF-CTe are shown in Figures 5-4 and 5-5, respectively. The temperature range

















-200 205 55 H
o 650C
-150 _301 7 Hz A 70C
E D v 75C
o -100 8 442 84 Hzdz 0 9541 Hz 0 80
Co fl ,400 H 0c 3017Hz n / o 85C
-50 ..0650 Hz oO/ or /


0 100 200 300 400 500 600 700

Z / cm2


Figure 5-4: Impedance-plane or Nyquist representation of impedance data for Cell
UF-CTe with temperature as a parameter. Six replicates of each measurement are
presented for each temperature.


102 10 100 101 102 103 104 105
fl Hz


103 102 101 100 101 102 103 104 105
f/ Hz


Figure 5-5: Bode Representation of impedance data for phase angle and magnitude
for Cell UF-CTe, respectively, with temperature as a parameter. Six replicates of
each measurement are presented for each temperature.











i, fz ml:7,


76 7010
E~0 65'C"


N A 700

10 858



10 102 10 10 10 102 10 10 10 10 12 10 10 0 1 0 1 02 10 10 105
a N A 70'C
v 750
101 o 80'C
101 o 85C

103 102 101 100 101 102 103 104 105 103 102 101 100 101 102 103 104 105
f Hz fl Hz
(a) (b)

Figure 5-6: Plots of the imaginary and real part of the impedance as functions of
frequency for Cell UF-CTe with temperature as a parameter. Six replicates of each
measurement are presented for each temperature.

Table 5-3: Average values of the low frequency tail slopes from cells UF-BTe, UF
CTe, and UF-FTe. Six replicates were used to calculate the average parameters.

Temp. / C UF-BTe UF-CTe UF-FTe
60 0.333 + 0.01 0.298 0.01 0.296 0.01
65 0.351 0.02 0.321 0.02 0.337 0.02
70 0.396 0.13 0.342 0.13 0.351 0.13
75 0.391 0.01 0.411 0.01 0.399 0.01
80 0.389 0.01 0.387 0.01 0.343 0.01
85 0.408 0.02 0.338 0.02 0.388 0.02


analyzed was 600C to 85C in intervals of 5C. Six impedance replicates at each

temperature are presented in this plot. The Nyquist plot represents a system with

characteristic time constant distributed over a wide-range of time scales.

The real and imaginary representations are presented in Figure 5-6. The slope

of the low frequency tail increased with temperature from a value of 0.298 0.01

at 60C, to 0.411 0.01 at 75C, and decreased to 0.338 0.02 at 85C. A similar

behavior was observed in cells UF-BTe and UF-FTe. The average values of the

low frequency tail slopes are presented in Figure 5-7 and Table 5-3. The average

slopes of the imaginary representation at intermediate decreased with increasing

temperature as shown in Figure 5-8. At high frequencies, the average sloped of


F ""' ""'' ""' ""' "~B~ ""'~ ""'~~







47

0.5 .

a UF-BTe
e UF-CTe
.. A UF-FTe
0.4 A
C) A

A
A e



0


60 70 80 90

Temperature / C

Figure 5-7: Average values of the low frequency tail slopes from cells UF-BTe,
UF-CTe, and UF-FTe. Six replicates were used to calculate the average parame-
ters.

Table 5-4: Slopes obtained using graphical techniques at the intermediate fre-
quency range from cells UF-BTe, UF-CTe, and UF-FTe. Six replicates were used
to calculate the average parameters

Temp. / oC UF-BTe UF-CTe UF-FTe
60 0.826 0.006 0.759 0.020 0.626 0.072
65 0.835 0.003 0.747 0.031 0.663 0.036
70 0.780 0.018 0.720 0.033 0.707 0.019
75 0.786 0.018 0.711 0.016 0.693 0.022
80 0.772 0.003 0.712 0.028 0.622 0.017
85 0.737 0.011 0.677 0.035 0.611 0.070


the imaginary representation remain constant until a temperature of 750C was

reached and then decreased as shown in Figure 5-9. The average results obtained

for all the cells are presented in Tables 5-4 and 5-5. The slope of the imaginary

representation at high frequency range varies from -0.69 0.05 to -0.74 0.01,

which corresponds to a values ranging from 0.31 to 0.26. The difference in the

slope at intermediate and high frequency was smaller in these experiments.

The value of Qeff at high frequency was calculated and the resulting values

are presented in Figure 5-10. The five highest frequencies points were disregarded
















1W


u.o
>' : [


Co
o e









[ UF-BTe
SF-CTe


SUF-FTe


0.5 A





Figure 5 8: Average values of the intermediate frequency tail slopes from cells

UF-BTe, UF-CTe, and UF-FTe. Six replicates were used to calculate the average

parameters.


0.8



0.7



0.6


60 70 80 90


Temperature / C


Figure 5-9: Average values of the high frequency tail slopes from cells UF-BTe,

UF-CTe, and UF-FTe. Six replicates were used to calculate the average parame-

ters.


A
A






A
A

a UF-BTe
e UF-CTe
A UF-FTe







49



108 E 600C
o 65C
107 O A 700
6
U) 106 o v 75C
o 80'C
E 10 o 85C
104

r 103
103
102
10
100.... .".. ..
10-3 10-2 101 100 101 102 103 104 105
f/Hz

Figure 5-10: Effective CPE coefficient defined by equation (5.1) for Cell UF-CTe.
Six replicates of each measurement are presented for each temperature.


because they were believed to be corrupted by instrument artifacts. The values

for the apparent CPE coefficient provided in Table 5-5 represents the average

over the values for the remaining 5 highest frequencies. The value of Qeff seems to

reach a maximum around 750C. There is also more variability in this value as the

temperature increased. We were not able to calculate a meaningful average value of

Qeff for cell UF-FTe.

The solution resistance RP, the corrected phase angle, and the corrected im-

pedance magnitude were calculated. The IR-corrected phase angle and magnitude

are presented in Figure 5-11. The slope of the high frequency .,-vmptote has a

value of 1 a. The values for the electrolyte resistance are provided in Table 5-5.

These values are in good agreement with the values obtained using the impedance

plane representation.

5.3 Transient Experiments at Constant Temperature

Impedance-plane and Bode representations obtained from Cell UF-DTr are

shown in Figures 5-12 and 5-13, respectively. The temperature was held constant

at 700C throughout the experiment. Impedance spectra were taken every 30












-bu
-70 60
o 650C
-60 A 700C
v 750C
-50 0 80C g
o 850 C a

-30 6 8 8a
-30
20 8
10


103 102 101 100 101 102 103 104 105
flHz


10" 10' 10 10' 10 10' 10" 10" 10-
flHz


Figure 5-11: IR-corrected Bode plots for Cell UF-CTe; (a) phase angle, (b) modu-
lus. Six replicates of each measurement are presented for each temperature.





Table 5-5: Average values of the parameters obtained using graphical techniques at
the high frequency range from cells UF-BTe, UF-CTe, and UF-FTe. Six replicates
were used to calculate the average parameters

Temp. / C UF-BTe UF-CTe UF-FTe
a 0.24 0.01 0.26 0.01 0.32 0.01
60 Qeff / MQ-1 cm-2s1- 11.60 1.15 12.98 1.54 24.70 2.05
Re / 2 cm2 37.18 2.55 37.35 1.22 50.38 1.23
a 0.25 0.02 0.27 0.04 0.27 0.05
65 Qeff / M2-1 cm-2s1-" 14.06 2.23 16.69 5.33 21.69 11.76
Re / 2 cm2 30.34 0.67 32.25 1.60 38.96 1.80
a 0.24 0.1 0.27 0.03 0.18 0.10
70 Qeff / M -1 cm-2sl-" 13.27 1.87 19.43 4.94 15.27 20.94
Re / 2 cm2 24.29 1.05 27.77 1.11 32.60 3.57
a 0.22 0.01 0.31 0.02 0.20 0.01
75 Qeff / M -1 cm-2s1-" 11.78 1.20 29.09 6.74 23.30 26.85
Re / 2 cm2 20.04 1.22 18.85 0.97 26.21 1.53
a 0.21 0.01 0.31 0.05 0.15 0.04
80 Qeff / M2-1 cm-2s1-" 10.02 1.22 32.90 15.72 10.81 4.16
Re / 2 cm2 13.16 0.50 15.15 2.69 24.64 1.23
a 0.26 0.01 0.25 0.10 0.37 0.03
85 Qeff / M -1 cm-2sl-" 13.68 1.34 26.09 30.94 222.12 126.0
Re / 2 cm2 11.14 0.36 14.16 0.78 19.40 2.08



















120


0 50 100 150 200 250 300 350 400

Z / Q cm2
r


Figure 5-12: Impedance-plane or Nyquist representation of impedance data for Cell
UF-DTr with temperature as a parameter. Impedance spectra were taken every 30
seconds for 48 hours, it took approximately two hours to complete one spectrum.


-30 -


0 [ 2 I

10 10 10 10 100 10 102 10 104 105
flHz


E 102
0
o
N




101 L
10, 10-


10 102 10- 100 101 102 103
flHz


Figure 5-13: Bode Representation of impedance data for phase angle and magni-
tude for Cell UF-DTr, respectively. Impedance spectra were taken every 30 seconds
for 48 hours, it took approximately two hours to complete one spectrum.


10 10 10


1 1 1 --, 1 1 1 -, 1

p~f~8







52

. .. .10 ...-. .- ..- -- -. .. --. -..-. ..-- -- -- r. ..

102
101

C 10-- '

N N g
101


104 103 10 10 100 10 102 3 10 10 1010 104 103 102 101 100 101 102 103 104 105 106
flHz f/Hz
(a) (b)

Figure 5-14: Plots of the imaginary and real part of the impedance as functions of
frequency for Cell UF-DTr. Impedance spectra were taken every 30 seconds for 48
hours, it took approximately two hours to complete one spectrum.


seconds for 48 hours, it took approximately two hours to complete one spectrum.

The Nyquist plot reveals a depressed semicircle.

The real and imaginary representations are presented in Figure 5-14. The

slope of the low frequency tail decreased with time from a value of 0.35 0.02

to 0.24 0.01. However, the low frequency tail slopes for cell UF-ETr varied

randomly from 0.38 0.04 to 0.30 0.03. This discrepancy can be attributed to

the fact that the melting point of the electrolyte is approximately 600C, thus there

is more instability in the system at that temperature. The average values of the

low frequency tail slopes are presented in Table 5-6. The slopes of the imaginary

representation at intermediate frequencies decrease with time and the slope at high

frequencies oscillate with time between -0.74 and -0.79, which correspond to a

values ranging from 0.26 to 0.21. The average results obtained from Cell UF-DTr

and UF-ETr are presented in Table 5-7 and 5-8. In this experiment the slope at

intermediate and high frequency were different, ~-.--1 -ii.-; that more than one CPE

are affecting the system. This result is in agreement with previous experiments.














Table 5-6: Values of the low frequency tail slopes from Cell UF-DTr and UF-ETr.

No. UF-DTr F-ETr
2 0.35 0.02 0.35 0.02
3 0.33 0.02 0.37 0.04
4 0.33 0.02 0.35 0.02
5 0.33 0.01 0.37 0.02
6 0.31 0.01 0.32 0.04
7 0.32 0.01 0.36 0.03
8 0.31 0.01 0.38 0.04
9 0.30 0.01 0.34 0.03
10 0.28 0.01 0.30 0.03
19 0.24 0.01 0.36 0.06


Table 5-7: Values of the low and
Cell UF-DTr and UF-ETr.


intermediate frequency tail slopes obtained from


No. UF-DTr UF-ETr
2 0.83 0.01 0.79 0.01
3 0.82 0.01 0.74 0.01
4 0.81 0.01 0.72 0.01
5 0.81 0.01 0.69 0.02
6 0.79 0.01 0.67 0.02
7 0.79 0.01 0.65 0.03
8 0.77 0.02 0.64 0.03
9 0.78 0.01 0.62 0.03
10 0.77 0.01 0.61 0.03
19 0.72 0.02 0.42 0.05







54


108
10
S10'

E 1o5
10

S10,

0)
S101

100
10'3 102 100 10 102 103 104 10,
fl Hz

Figure 5-15: Effective CPE coefficient defined by equation (5.1) for Cell UF-DTr.


The value of Qeff at high frequency was calculated and the resulting values are

presented in Figure 5-15. The value for the apparent CPE coefficient provided in

Table 5-8 represents the average over the highest frequencies values.

The solution resistance Re, the corrected phase angle, and the corrected im-

pedance magnitude were calculated. The IR-corrected phase angle and magnitude

are presented in Figure 5-16. The slope of the high frequency ..i-mptote has a

value of 1 a. The values for the electrolyte resistance are provided in Table 5-8.

These values are in good agreement with the values obtained using the impedance

plane representation.

5.4 Conclusions

The plots presented here provide useful guides to model development. The

difference in the slopes at intermediate and high frequencies in the plots of the

imaginary component of the impedance -ii-.-. -i that more than one capacitive time

constant can be identified in the system. The high-frequency capacitive loop can be

attributed to a dielectric behavior of the electrolyte or to reactions at the lithium

metal surface. This means that an equivalent circuit such as the one proposed by















Table 5-8: Average values of the parameters obtained using graphical techniques at
the high frequency range for Cell UF-DTr.

No. UF-DTr UF-ETr
a 0.26 0.32
2 Qeff / MQ-1 cm-2s-" 12.52 21.39
Re / 2 cm2 35.48 60.24
a 0.26 0.30
3 Qef / M-1 cm-2s1-" 13.95 20.62
Re / Q2 cm2 33.36 68.65
a 0.24 0.29
4 Qff / MQ-1 cm-2s1-" 12.01 23.16
Re / Q2 cm2 32.90 60.61
a 0.21 0.29
5 Qff / MQ-1 cm-2s1-a 9.40 24.90
Re / Q2 cm2 33.05 59.66
a 0.23 0.29
6 Qff / MQ-1 cm-2s1-" 11.80 31.89
Re / 2 cm2 33.43 63.53
ca 0.22 0.30
7 Qeff / M-1 cm-2s1-" 10.87 23.74
Re / Q2 cm2 33.23 56.37
a 0.22 0.29
8 Qff / M2-1 cm-2s1-a 11.25 26.59
Re / 2 cm2 32.60 57.14
a 0.23 0.30
9 Qef / M2 -1 cm-2s1-a 13.08 23.86
Re / Q2 cm2 31.80 60.56
a 0.19 0.31
10 Qeff / M2-1 cm-2s1-" 9.42 25.76
Re / 2 cm2 33.43 54.57
oa 0.29 0.30
19 Qeff / M-1 cm-2s1-a 28.05 39.37
Re / Q2 cm2 26.59 56.12












-80

102 -40
E

1010 1%
40

10 .. .. .. .. .. .. 80 "
10' 102 101 100 101 102 103 104 105 103 102 101 100 101 102 103 104 105
flHz flHz
(a) (b)

Figure 5-16: IR-corrected Bode plots for Cell UF-DTr; a) phase angle, b) modulus.


Bouchet et al. may describe the system correctly if the capacity loop is attributed

to a dielectric behavior of the electrolyte.

The decrease of the impedance response with time is consistent with dissolu-

tion of native oxide l.- ris on the lithium surface, as proposed in the literature. The

results showed that the time required for complete dissolution of the oxide l., r? S

were on the order of several d4 i- As the temperature of the cells was increased the

impedance response decreased. This result is consistent with either hypothesis, the

high-frequency capacitive loop is due to reactions at the lithium metal surface or

due to a dielectric behavior of the electrolyte.















CHAPTER 6
BLOCKING ELECTRODE ANALYSIS

The impedance data presented in Chapter 5 were obtained using two lithium

electrodes in a symmetric configuration. To explore the hypothesis that the high-

frequency capacitive loop seen in the impedance spectra were associated with the

dielectric response of the electrolyte, a second set of experiments was performed

using stainless steel electrodes, again in a symmetric configuration.

6.1 Experimental Results

Impedance measurements were obtained at four different temperatures. The

results obtained at 250C, 800C, 900C, and 1000C are presented in Figure 6-1 with

temperature as a parameter. The results indicate that the steel electrodes acted as

blocking electrodes with a local capacity dispersion. The local capacity dispersion

can be caused by roughness on the electrode surface.

At low frequencies, the impedance response of the steel electrodes can be

expressed as
1
Z(f) Re + (6.1)

where the parameters a and Q are constants. When a 1= Q represents the

capacitance of the double l-V-r. When a / 1, the system shows behavior that

has been attributed to surface heterogeneity. At high frequencies, the data show

a dispersion that may be associated with a non-uniform current and potential

distribution.

6.2 Graphical Analysis

Huang et al.67 developed a mathematical model for a blocking disk electrode

embedded in an infinite insulating plane. The mathematical model involves solving








58

10 .

8- a

6
EP

4 0
I 250C
2 o 0 800C
S A 90C
So 100C
oAl
012345
0 1 2 3 4 5
Z/MQ


Figure 6-1: Complex impedance plane or Nyquist representation for the response
of a 430 Stainless Steel disk in PEO/LiTSFI electrolyte with concentration as a
parameter. Six replicates of each measurement are presented for each temperature.


Laplace's equation for potential in the frequency domain. Their work provides a

guide for win 1. i-; of impedance data that is similar to the one explained in C'! plter

5.

The real and imaginary representations are presented in Figure 6-2 as func-

tions of temperature. The difference in the response between the experiments

done at 250C and the ones done at 800C, 900C, and 1000C can be attributed to

the change in conductivity associated with the melting point of the electrolyte,

approximately 600C.

A value for the solution conductivity was extracted from the real part of the

impedance following
r -1
S= 4ro lim Z, (6.2)


The real parts of the impedance corresponding to the values obtained from

equation (6.2) are presented in Figure 6-2(a) as dashed lines. The resulting values

for conductivity are presented in Table 6-1.














106


105


104


N 103


102


N
a


10-3 10-2 10-1 100 101 102 103 104 10 106

fl Hz

(a)



107
10 7 250C
10O o 800C
10 A 900C

10 5 1000C

104

103

102

101

10 0
10-3 10-2 10-1 100 101 102 103 104 105 106

fl Hz

(b)


Figure 6-2: Plots of the imaginary and real part of the impedance as functions of
frequency for Cell NA-BB with temperature as a parameter. Six replicates of each
measurement are presented for each temperature.


E 250C
o 800C
1 A 900C
r a o 10000C








r I

, o- .o ----- oo-- a .oo o oo- .-- -
.- J .- .. -_ .. .. J ..J .


j


101







60

45
40 250C o
o 80C
35 90C o



E o
S 30 o 100C _o -

25 -



S10
0 --.

0 4- -I3- B Bo-"O B -E- --lB R aB o- B B E fl -BB f" O nn-"rn ] ---

10-2 10 100 10 102 103 104 105
f/Hz

Figure 6-3: Effective CPE coefficient defined by equation (6.3) for Cell NA-BB. Six
replicates of each measurement are presented for each temperature.


At low frequencies, the slope of the imaginary part of the impedance has a

value of -a. The values of a obtained from Figure 6-2(b) are also presented in

Table 6-1.

For a blocking electrode, the value of Q can be obtained from the low-

frequency .,-vmptote where the current distribution associated with the electrode

geometry does not influence the impedance response. The value of Q was calcu-

lated using
Q sin(aw/2) (6.3)
Q= (6.3)
Zj (2fpa
and the resulting values are presented in Figure 6-3. The extracted values for Q

are marked by dashed lines in Figure 6-3 and are reported in Table 6-1. As the

temperature increases, the values for Q also increase. At 100C the values for Q

increased with time.







61

Table 6-1: Parameter values extracted from Figures 6-2 and 6-3 for Cell NA-BB.

Temp. / C a Q/\ Q.-cm-2sa K/mho cm-1
25 0.824 0.003 0.40 0.0166 2.97 x 10-05 9.089 x 10-07
80 0.704 0.008 8.45 0.132 0.018 8.534 x 10-05
90 0.728 0.014 11.4 0.617 0.025 1.011 x 10-04
0.749 14.3 0.027
0.721 14.8 0.028
0.701 16.8 0.028
100 0.726 18.9 0.026
0.693 19.3 0.026
0.708 19.3 0.027


6.3 Comparison to Disk Electrode

The dimensionless imaginary part of the impedance Zjrro is presented in

Figure 6-4 as a function of dimensionless frequency


K = Q ro (6.4)


The results presented in Figure 6-4 can be compared to the results obtained by

Huang et al.67 shown in Figure 6-5 for different values of a. A superposition of the

data is evident in both Figures 6-4 and 6-5 at low frequency. This superposition

is typical of blocking electrodes. The slope at low frequency has a value of -1

because the influence of the CPE parameter a is incorporated in the definition of

the dimensionless frequency in equation (6.4).

A change in the slope from a value of -1 appears in Figure 6-4 at frequencies

higher than K = 6 x 10-3. A similar change in slope appears in Figure 6-5 at

frequencies K > 1. For the disk electrode, the change in slope at frequencies K > 1

is due to the non-uniform current distribution associated with the disk geometry.

The geometry of the coin cells used for the present work is significantly different

than that of a disk electrode with a counterelectrode infinitely far away. There

is a potential for a non-uniform current distribution in the button cell, but the



















LC

N,

100




10-2
10.5 10-3 10-1 101 103

K

Figure 6-4: Dimensionless -in i1 -i for the impedance response of a stainless steel
disk in PEO/LiTFSI electrolytes with temperature as a parameter.


0
- 100

10-1

10-2

10-3


104 L_
10-5


10-4 10-3 10-2 10-1 100 101 102 103
Dimensionless Frequency K


Figure 6-5: Dimensionless imaginary part of the impedance as a function of dimen-
sionless frequency for a disk electrode with local capacitance dispersion. Theoreti-
cal result derived by Huang et al..









geometry of the electrode-insulator interface is not well defined. The change of

slope at K = 6 x 10-3 may be associated with a non-uniform current distribution.

6.4 Conclusions

The impedance response of the symmetric button cell with steel electrodes was

that of a blocking electrode with local CPE behavior associated with a capacity

dispersion. The high-frequency capacitive loop seen in the results of the symmetric

lithium cell were not evident in the results from the cell with steel electrodes.

Thus, the high-frequency capacitive loop seen in the impedance spectra of C'! lpter

5 are not associated with the dielectric response of the electrolyte. These loops

must be described instead by a model that takes into account the reactions at the

lithium electrode.

When conductivity K and the CPE parameters Q and a are taken into

account, the imaginary impedance responses for all the data collected with the

symmetric stainless steel cell are superposed at low frequencies. This result was

seen even when the low-temperature 25C data were included, for which the

electrolyte was solid and the conductivity was three orders of magnitude lower than

for the melted electrolyte. This superposition provides evidence that the system

behavior is indeed that of a blocking electrode and that no Faradaic reaction

needed to be considered. At high frequencies, when the contribution of the non-

uniform current distribution is more important, the experimental results deviated

from the theoretical behavior. However, this deviation was anticipated since the

geometry of the experimental system differs from the geometry of the system used

to develop the theoretical results.















CHAPTER 7
MODEL DEVELOPMENT

The results presented in C'! ipter 6 demonstrated that the high-frequency

capacity loop observed in the experimental results were due to reactions at the

lithium electrodes. The electric circuit model proposed by Bouchet et al. does not

describe the system correctly because he attributed the capacity loop to a dielectric

behavior of the electrolyte. Still he was able to fit adequately his experimental

data. This is the limitation of the electric circuits, more than one circuit network

can be constructed which would provide equivalent statistically valid fits; and each

of these models can arise from extremely different interpretations. Physico-chemical

models are more efficient for interpretation of the physical properties and the

kinetic processes of the system. However, the physico-chemical models have the

disadvantage that they cannot be easily regressed to the experimental data. In

the present work we developed a model that take into account the physics and

chemistry of the system and can be use for regression analysis.

To develop such model it was necessary to have complimentary supporting in-

formation concerning the physics and chemistry of the system. Microscope images,

scanning electron microscope (SEM) micrographs and information published in the

literature were used to guide the model development.

7.1 C!. In-I ry

It is well accepted in the literature that the reaction mechanism for the lithium

electrodes is extremely complex and must be represented by a large number of

reactions.10'68,69 The electrolyte composition and the presence of additives and

contaminants affect the mechanisms of transport. According to Aurbach and









Table 7-1: Reaction mechanisms of common lithium battery contaminants. Water
was the only contaminant considered in this work.

Contaminant Reaction Mechanisms
C02 C02 + e- + Li+ -- C02Li
CO2Li + CO2 -2 OC-OCO2Li
O=C-OCO2Li + e- + Li -- COT + Li2CO3s
2LiOHJ + CO2 Li2CO3s
Li201 + CO2 -- Li2CO3s
H20 H20 + e- + Li+ LiOH + 1/2H2
LiOH + Li+ + e- Li20 + 1/2H2
H + e- + Li+ -LiH
N2 N2 + 6e- + 6Li+ 2Li3N
02 02 + e- + Li+ LiO2
Li2 + e- + Li+ --Li202
Li202 + 2e- + 2Li+ -- 2Li2O
HF HF + e- + Li+ LiF + 1/2H2

Table reproduced from: G.Nazri and G. Pistoia, "Lithium Batteries: Science and
Te. !iii,., ; (Norwell,MA: Kluwer Academic Publishers, 2004).


Schechtero1 the possible reactions between Li+ and the LiTFSI salt are


LiN(SO2CF3)2 + ne- + nLi+ -- Li3N + Li2S204 + LiF + C2FxLiy



LiN(SO2CF3)2 + 2e- + 2Li+ -- Li2NSO2CF3 + CF3SO2Li


and


(7.1)



(7.2)


and the common contaminants present in this type of cell include CO2, H20, N2,

02, and HF.10'68'69 The reaction mechanisms of the common contaminants are

summarized in Table 7-1.

As the exact composition, size and distribution of the particles that composed

the SEI (see Section 2.2.4) are unknown, a simplified system was used. The

reactions considered for the simplified system were the oxidation/reduction reaction

of lithium,


Li ; Li+ + e


(7.3)












40 x10 -

2 Ox1 0

00 |

N_
2 Oxl10'


-40x10 I I I II
-5x10 -4x109 -3x109 -2x109 -1x109 0 1x109
Z / cm2


Figure 7 1: Impedance-plane or Nyquist representation of impedance data for Cell
UF-DTr. Very large impedance at low frequencies can be observed s i'-'.-. --i;i,' that
a disconnection of the system which may be attributed to gas evolution.


and the reactions due to the presence of water as a contaminant,


H20 + e- + Li+ LiOH + 1/22 (7.4)




LiOH + Li+ + e- Li20 + 1/2H (7.5)

It has been reported in the literature that these cells contain traces of water

and the formation of LiOH is expected at the lithium interface, which support the

selection of equations (7.4) and (7.5) as part of the proposed reaction mechanism.70

Also, the impedance data and broken seals provided evidence of gas evolution.

An impedance spectrum of cell UF-DTr taken approximately 14 hours after the

experiment started is presented in Figure 7 1. The impedance response of the

cell was as expected for a system with distributed time constants, as shown in the

insert. However, a large impedance was observed at low frequencies. This is usually

attributed to a disconnection of the system, which may be associated with gas

evolution. This behavior was observed in more than one cell.











IL lwb v




















4jr-


rNt

A6y


Figure 7-2: Cell UF-ATe lithium electrode surface. The surface of the electrode
was almost completely covered with a white to blue oxide l-,r.


7.2 Physics

The formation of passivating films on the lithium surface has been reported

extensively in the literature.10' 18-20 Peled was the first to introduce and name the

SEI.', 19 According to Peled, the SEI is a heterogenous film, consisting of a mosaic

of numerous individual particles of different chemical composition being in partial

contact with each other at the grain boundaries, as explained in Section 2.2.4.

Some of the cells used in this research were opened after the experiments.

Microscopic images of cells UF-ATe and UF-BTe are presented in Figures 7-2 and

7-3. A white to blue oxide 1l-v-r was observed on the lithium surface. Cell UF-ATe




























































Figure 7-3: Cell UF-BTe lithium electrode surface. The edges of the electrode
were covered with a white to blue oxide l1I-V, while the inner surface kept a much
brighter metallic lustre.









was almost completely covered with the 1I- r, while in cell UF-BTe, the oxide 1,i r

was observed in the perimeter of the cells, as shown in Figure 7-2 and 7-3(a),

respectively. For most of the cells, the oxide l1-v-r was more evident on the lithium

edge than inside the battery. This behavior has been reported in the literature70

and is evidence of a non-uniform current distribution.

SEM micrographs were taken to get an understanding of the structure of this

lI-V r. The 1-- r is an heterogenous film that appears to be a mosaic of individual

particles, as shown in Figures 7-4 and 7-5. Unfortunately, it was not possible to

determine the composition of the l.,-r.

Several options were considered as possible mechanisms of transport. All

of the mechanisms considered take into account the presence of the passivating

films at the electrodes surface, as described in the literature.10' 18-20 According to

Naudin et al. ,29 commercial lithium foils have a native l1-v-r that is composed of

approximately 10-100 nm Li2O inside l1- -r and 1-20 nm of Li2CO3/LiOH outside

lI- r, as shown in Figure 7-6.

The first mechanism is presented in Figure 7-7. In this model, the only

reaction taking place at the electrode surface is the oxidation/reduction reaction

of Li. The LiOH and Li20 formation reactions take place on the surface of the

passivating l,- r.

On the second mechanism it was assumed that the lithium oxidation/reduction

reaction take place at the lithium metal surface only, but the LiOH and Li20

formation reactions can take place in both the lithium metal surface and the oxide

l1-,-r surface. A schematic diagram of the system is shown in Figure 7-8.

7.3 Model Development

The procedure explained in Section 3.5 was used to develop the impedance

response model. In this section the model for the mechanisms shown in Figure 7-8


























































Figure 7-4: Scanning electron micrographs of Cell UF-ATe lithium electrode sur-
face. (a) Magnification = 35. (b) Magnification = 200.







































Figure 7-5: Scanning electron micrographs of Cell UF-BTe lithium electrode sur-
face. Magnification = 350.









---_i jSurface
Layer




Li20 layer (10-100 nm)1
E Li2CO3/LiOH layer (1-20 nm)1


Figure 7-6: Commercial lithium foil native l-1~-r composition, as described by
Naudin et al.29















































Figure 7-7: Schematic of the simplified system. Two surfaces were identified in
which the reactions take place. The lithium oxidation/reduction reaction take place
at the lithium metal surface and the LiOH and Li20 formation reactions take place
in the oxide l-,-r surface.






















































Figure 7-8: Schematic of the simplified system. Two surfaces were identified in
which the reactions take place. The lithium oxidation/reduction reaction take place
at the lithium metal surface only, but the LiOH and Li20 formation reactions can
take place in both the lithium metal surface and the oxide 1I-. r surface.


Electrolyte











Lithim Meal Srfac









is explained in detail. Only the final expression for the mechanism shown in Figure

7-7 is presented.

7.3.1 Steady State

The equation for the faradaic current density, i,f, can be expressed in terms

of a steady time-independent value and an oscillating value for each of the three

reactions considered in the system. The steady state values are


ia,Li, = Fka,Li,oexp(ba,Li, (V VLi)) (7.6)

and

ic,Li,0 = -Fkc,Li,OCLi+,Li,oexp(-b,Li,o(V VLi,o)) (7.7)

for the lithium oxidation/reduction reaction taking place at the electrode surface.

When LiOH and the Li20 are formed in the lithium metal the steady state current

contributions are


iLiOH,O = -Fkc,LiOH,OCH2O,oexp(-bc,LiOH,o(V VLiOH,o)) (7.8)

and

Li20,0 = -Fkjc,Li2o,oexp(-bc,Lio2,o(V VLi2,0)) (7.9)

The steady state currents at the oxide l-~.-r surface are


iLiOH, = -FOH,iOH,CLi+ ,LiOH,6CH2O,6exp(-bc,LiOH,6(V VLiOH,6)) (7.10)

and

iLio0, = -Fkc,Li2o,SCi+ Li2o,6exp(-bc,Li2o,(V VLi20,6)) (7.11)

for the LiOH and Li20 formation reactions, respectively. The subscript 0 and 6

represent the lithium metal surface and the oxide 1lv. r surface, respectively. Notice

that the diffusion of lithium for the reactions that involve water is considered when









the reactions take place in the oxide li. r but is not considered when the reactions

take place in the lithium metal surface.

7.3.2 Sinusoidal Steady State

The oscillating component of the current was derived using,

if Vc (7.12)
) Y= V Oc ^o +V ci (7,12)

where V is potential, ci,o is concentration and Ok is surface coverage. In this model

the surface coverage was assumed to be constant, so the current is only dependent

on potential and concentration of lithium and water. Solving equation (7.12) with

respect to each of the steady-state current contributions would give the oscillating

current contributions of the reactions considered in the system. These expressions

are

la,Li, R, i,0oV (7.13)

where,

a,Li, Fba,Li,oka,Li,oexp(ba,Li, (V- VLi,o)) (7.14)

and

ic,Li,O Rci,V + (-Fk,Li,oexp(-bc,Li,o(V- VLi,o))CLi+,Li,0 (7.15)

where,

jci,o FbcLi,okc,Li,O +,L,oexp(-b,Li,o(V VLi,o)) (7.16)

for the lithium oxidation/reduction reaction taking place at the electrode surface.

The oscillating current contributions of the LiOH and LiO2 formation reactions in

the lithium metal are


ILiOH,0 R LiOH,OV + (-Fkc,LiOH,oexp(-bc,LiOH,o(V VLiOH,0)))CH20,0 (7.17)

where,


RLiOHO Fbc,LiOH,okc,LiOH,OCH2o,oexp(-bc,LiOH,o(V VLiOH,o))


(7.18)









and

2Li20,0 RLi20oV (7.19)

where,

RL-oo FbcLi2o,okc,Li2o,oexp(-bLi2o,o(V VLi2o,o)). (7.20)

When the reactions occur in the oxide 1 i- V, the oscillating current contribu-

tions are

iLiOH,6 RLiOH,6V + (-Fkc,LiOH6CH2iOexp( cLOH,6(V VLiOH,6 )))cLi+,LiOH,6
+ (-Fkc,LiOH,6CLi ,LiOH,6exP(-bc,LiOH,6(V VLiOH,6))) H20,6 (7.21)


where,


RLiOH, Fbc,LiOH,6akc,LiOH,6CLi ,LiOH,6CH20,aexp(-bc,LiOH,6(V- VLiOH,6)) (7.22)

and

Li2o0,6 RLilo,6V + (-Fkc,Li20,aexp(-bc,Li20,a(V- VLi2o0,)))Li+,Li20,6 (7.23)


where,


R 0,6 = Fbc,Li2,ac,Li2O,aLi+,Li2o,aexp(- b,Li2o,6(V VLi20,6)) (7.24)

7.3.3 Mass Transfer

A second equation was needed to correlate concentration of lithium and water

to the corresponding current contributions


ii = -nzFD i (7.25)
y=0

This equation can be expressed in terms of the oscillating contributions as


ii = -nzFD 0 (7.26)
?dy o









It is convenient to write equation (7.26) in terms of the dimensionless solution to

the appropriate diffusion or convective equation


i = -niFDi (0) (7.27)


where 0i = /~jo. 0i was evaluated assuming diffusion through an infinite l--r as

1 1
(7.28)
0/(0)

The definition of Oi was then substituted into equation (7.27) to obtain the

relationship between concentration and current,

SCi W
ii = niFDi (7.29)
6i Di

The final current expression as a function of potential for the reactions that

take place in the lithium metal surface are


ta,Li, = Ra,Li,-1V (7.30)

and

ic,Li, (Rc,Li,o + Zd,Li,o) V (7.31)

where,
R,Li, a,Li,okc,LiOH,exp(-bc,LiOH,O(V VLi,))Li,O (7.32)
zd,Li,o = (7.3--2)
DLi+

for the lithium oxidation/reduction reaction and


iLiOH,0 (RLiOH,O + Zd,H20,0)- (7.33)

where,
RLiOH,0kc,LiOH,oexp(-bc,LiOH,o(V VLiOH,0))H20 (7.34)
Zd,H20,0 3H2O
D ^DHgO












Li2O,O = RLioo- V


(7.35)


for the LiOH and Li2O formation reactions. For the reactions that take place in the

oxide liv-r surface the final current was expressed as


ILiOH,S (RLiOH, + d,H20,6 + ULi,62)-1V


(7.36)


ZH20,6


and


ZLi,62


RLiOH,6kc,LiOH,SCLi+ ,LiOH,exp(- bc,LiOH, (V -
DHO20 DHo




RLiOH,6 kc,LiOH,65CH20,LiOH,6exp(-bc,LiOH,6 (V -
DLi+VJ
V Li


VLiOH,6))6H20,6





VLiOH,6))Li+,62


for the LiOH formation reactions and for the Li2O formation reactions the final

current was expressed as


iLi2O,6 = (RLi2O,6 + Zd,Li,63)- V


where
R Li2o,6kc,Li2o,6exp(-bc,Li2o,6(V VLi2o,S))Li+63 (7.40)
Zd,Li,63 (7.4)
DLi+ DLi

The total current density was expressed in terms of the faradaic and charging

currents as


ii ii, + jWCdV


where the faradaic current if for the reaction taking place in the lithium surface is

given by


(7.42)


io,f = ia,Li,o + ic,Li,o + iLiOH,O + Li2o,o,


and


where


(7.37)


(7.38)


(7.39)


(7.41)









if for the reaction taking place in the oxide l1i-r surface is given by


t6,f = iLiOH,6 + tLi20,6, (7.43)

and Cd is the double l1v.r capacitance.

7.3.4 Impedance Expressions

The difference between the potential measured at the surface and the potential

measured some distance away is given by


U =Re + V (7.44)


Using equation (7.44) the impedance for the lithium surface and the oxide l1i-r

were calculated as

Z = (7.45)
1
where the impedance of the lithium surface was calculated to be


Zo= R (7.46)
( R)-1 + (Rc,Li,0 + Zd,Li,O)-1 + (RLiOH,O + Zd,H20,0)-1 + j'Cdl (

where

RO = Ra,Li,O- + RLLi20,o (7.47)

and the impedance of the oxide 1-i-r surface was calculated to be

1
Za = Re + (7.48)
(RLiOH,6 + Zd,H20,6 + Zd,Li,6)-1 + (Li20, + Zd,Li,)63) + jCd2

The electrolyte resistance term in the Zo was replaced by an equivalent resistance

that correspond to

Req (Re + Rti,,,) (7.49)

where Rfil is the resistance across the oxide I-,ir. The impedance of the lithium

surface for the system presented in Figure 7-7 was calculated to be

1
zo = Re + (7.50)
(Ra,Li,o)-1 + (Rc,Li,0 + zd,Li,0)-1 + jcddl









and the impedance of the oxide l--r surface was calculated to be

1
Z6 = R, + (7.51)
(RLiOH,6 + Zd,H20,6 + Zd,Li,62)-1 + (RLi20, + Zd,Li,63)-1 + jCd d2

The overall impedance for both systems was finally calculated to be


ZT = (7.52)
Zo- + Z,-

7.4 Results

In this section results of cell UF-BTe exposed at 80C are presented and

the results are compared to the models. The object of this comparison was to

show that the model developed can describe qualitatively the features seen in

the experimental data. The modelled results were calculated using Matlab. The

parameters used in each model are presented in Table 7-2.

The traditional Nyquist or impedance-plane representation is presented in

Figure 7-9 for the experimental data and the models. All plots indicate that the

system have characteristic time constants distributed over a wide-range of time

scales. Also, in the plots a low frequency tail is observed. However, in Figure 7

9(a) a smooth transition and a lower slope are observed between the high frequency

capacitive loop and the low frequency tail, whereas in Figures 7-9(b) and 7-9(c) a

sharp transition and a steeper slope are observed.

The imaginary representation is shown in Figure 7-10 for the experimental

data and the models. The general behavior of both plots is similar. However, the

slope of the graph at high frequency in Figure 7-10(a) is -0.7653, while for Figures

7-10(b) and 7-10(c) is -1. This difference is explained due to a distribution of

time constants in the experimental system. This distribution is also evident in the

non-uniform coverage of the cells surface films.

The real representation is shown in Figure 7-11 for the experimental data and

the models. The general behavior of both plots is similar.









Table 7-2: Summary of the parameters used to calculate the impedance response of
the models presented in Section 7.


Parameter Model 1 Model 2
DLi 10-9 10-9
DH20 10-10 10-10
kc,Li,O 100 100
ka,Li,O 0.001 0.001
kc,LiOH,O 100
kc,Li20,0 100
kc,LiOH,S 100 100
kc,Li20, 100 100
bc,Li,O 19.7 19.7
ba,Li,O 0.197 0.197
bc,LiOH,O 19.7
bc,Li20,o 19.7
bc,LiOH,6 19.7 19.7
bc,Li20,6 19.7 19.7
V 0.8 0.8
VLi,O 0.1 0.1
VLiOH,0 0.01
VLi20,o 0.01
VLiOH,6 0.01 0.01
VLi20,6 0.01 0.01
CLi+,Li,O 0.1 0.1
CH20,0 0.01
CH20, 0.01 0.01
CLi+,LiOH,6 0.03 0.03
CLi+,Li20,S 0.03 0.03
Ra,Li,O 300 300
Rc,Li,O 200 200
RLiOH,O 150
RLi20,o 150
RLiOH,6 150 300
RLi20,6 150 200
Re 20 20
Req 140 180
HLi,O 1
6H20,0o 1
6Li+,2 1 1
6H20,S 1 1
6Li+,63 0.5 0.5
Cdl 0.000001 0.000001
Cd2 0.000001 0.000001

















60


40
(N
E
20

rsT


(N
0

rsT


(N
E
C-)


1.155 kHz

S 20 40 0 80 100 120 140

Z cm2





0 20 40 60 80 100 120 140
Z / Q cm2

(a)




1.264kHz






0.5647Hz

0 20 40 60 80 100 120 140

Z / Q cm2

(b)


200.4Hz
40


20


0
O.O01Hz

0 20 40 60 80 100 120 140

Z / I cm2
r

(C)


Figure 7-9: Impedance-plane or Nyquist representation of impedance data of a
symmetric cell. (a) Experimental Result. (b) Model 1. (c) Model 2.









































06 102 102
fl Hz


10" 10'


07 103 10' 105 109
f/Hz


N 103

N
10,


10 1r \- 1________
107 103 101 105 109
fl Hz

(c)


Figure 7-10: Imaginary plots of impedance data of a symmetric cell. (a) Experi-
mental Result. (b) Model 1. (c) Model 2.


10






E
0101
a
N



105
1


- -. -. -.- -.- -.-































10" 10, 1
Z /I cm2

(a)



102


0-


10'

E
0

N


101


103 101
fl Hz


103 101 105 109
fl Hz
(b)


105 109


Figure 7-11: Imaginary representation
Experimental Result. (b) Model 1. (c)


of impedance
Model 2.


data of a symmetric cell. (a)


10'

E

N


101


* ]
[]
[]
[]
B]
[]
\


- - - - - - - -


//









7.5 Conclusions

Two models were developed that are able to describe qualitatively the behavior

observed in the experiments. The selection of the reactions mechanism used

to developed those models was supported by independent observations. The

impedance data and broken seals in the cells provided evidence of hydrogen

evolution. Microscopic and Scanning Electron Microscope (SEM) micrographs

provided evidence of the formation of solid films on the lithium surface. This is

consistent with the formation of LiOH and LiO2. Also, the micrographs showed

that the film formed at perimeter of the cell which provides evidence of non-

uniform current distribution.

Both models provided evidence that the distribution of time constants

observed in the high-frequency capacitive loop is due to reactions at the lithium

electrode. None of the modelled results allow for the identification of more than one

capacitive time constants. It is necessary to account for the distributed reactivity

evident in the experimental data. The distribution of RC time constants should be

explored before a rigorous regression of the models is done.















CHAPTER 8
CONCLUSIONS

Electrochemical impedance spectroscopy experiments and models were used

to facilitate an in-depth understanding of the physical processes that control

symmetric lithium coin cells.

8.1 Experiments

Preliminary graphical analysis of the experiments performed on symmetric

lithium coin cells provided useful guides to model development. Nyquist and Bode

representations, as well as, IR-corrected Bode plots, log-log plots of the imaginary

component of the impedance, and effective capacitance plot were used. The slopes

in the log-log plots of the imaginary component of the impedance are directly

related to the apparent constant-phase-element coefficient a. The difference in

the slopes at intermediate and high frequencies demonstrated that more than one

capacitive time constant can be identified in the system.

A decrease of the impedance response with time was observed. This is consis-

tent with dissolution of native oxide lV- -ir on the lithium surface, as proposed in

the literature. The results showed that the time required for complete dissolution of

the oxide lV- -ir was on the order of several d-i,-

A high-frequency capacitive loop was observed in all the experimental results.

This loop can be attributed to a dielectric behavior of the electrolyte or to reac-

tions at the lithium metal surface. As the temperature of the cells was increased

the impedance response decreased. This result is consistent with either hypothesis,

that the high-frequency capacitive loop is due to reactions at the lithium metal

surface or that the loop is due to a dielectric behavior of the electrolyte.