The impact of high frequency high current perturbations on film formation at the negative electrode electrolyte interface

warwick.ac.uk/lib-publications

Original citation:

Uddin, Kotub, Somerville, Limhi, Barai, Anup, Lain, Michael J., Rajan, Ashwin T., Jennings,

Paul A. and Marco, James. (2017) The impact of high-frequency-high-current perturbations

on film formation at the negative electrode-electrolyte interface. Electrochimica Acta, 233 .

pp. 1-12.

Permanent WRAP URL:

http://wrap.warwick.ac.uk/86932

Copyright and reuse:

The Warwick Research Archive Portal (WRAP) makes this work of researchers of the

University of Warwick available open access under the following conditions.

This article is made available under the Attribution-NonCommercial-NoDerivatives 4.0 (CC

BY-NC-ND 4.0) license and may be reused according to the conditions of the license. For

more details see:

http://creativecommons.org/licenses/by-nc-nd/4.0/

A note on versions:

The version presented in WRAP is the published version, or, version of record, and may be

cited as it appears here.

The

impact

of

high-frequency-high-current

perturbations

on

fi

lm

formation

at

the

negative

electrode-electrolyte

interface

Kotub

Uddin

a,

*

,

Limhi

Somerville

b

,

Anup

Barai

a

,

Michael

Lain

a

,

T.R.

Ashwin

a

,

Paul

Jennings

a

,

James

Marco

a

a_{WMG,InternationalDigitalLaboratory,TheUniversityofWarwick,CoventryCV47AL,UK} b

JaguarLandRover,BanburyRoad,WarwickCV350XJ,UK

ARTICLE INFO

Articlehistory:

Received17November2016

Receivedinrevisedform2March2017 Accepted3March2017

Availableonline7March2017

Keywords:

Highfrequencycurrent Lithiumion

Degradation Cellautopsy

Electrochemicalmodelling

ABSTRACT

LongtermageingexperimentalresultsshowthatdegradationresultingfromcoupledDCandACcurrent waveformsleadtoadditionaldegradationoflithium-ionbatteriesabovethatexperiencedthroughpure DCcycling.More profoundly,suchexperimentsshowadependencyofbatterydegradation onthe frequencyofACperturbation.Thispaperaddressestheunderlyingcausalityofthisfrequencydependent degradation.Cellautopsytechniques,namelyX-rayphotoelectronspectroscopy(XPS)ofthenegative electrodesurfacefilm,showgrowthofsurfacefilmcomponentswiththesuperimpositionofanAC waveform.XPSresultsshowthathighfrequencyACperturbationsleadtotheincreasedformationofa passivatingfilm.Inordertodeterminethecauseofthisincreasedfilmformation,aheterogeneous electrochemicalmodelfortheLiNiCoAlO2/C6lithiumionbatterycoupledwithgoverningequationsfor

theelectricaldouble-layerandsolidelectrolyteinterfacefilmgrowthisdeveloped.Simulationresults suggestthattheincreasedgrowthofsurfacefilmisattributedtofrequencydependentheatgeneration. ThisisduetoionkineticsinthedoublelayerwhicharegovernedbythePoisson-Boltzmannequation. Additionalthermalandreferencecellrelaxationexperimentsareundertakenthatfurthercorroborates theconclusionthatheatgenerationwithinthebatteryisafunctionoftheACexcitationfrequency throughresistivedissipationandtheentropyofthecellreaction.

©2017TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1.Introduction

Lithium-ion(li-ion)cellsarerecognisedasacentraltechnology intheprocessofachievingacleanandsustainableenergyfuture throughtheelectrificationofroadtransportandtheuseof grid-connected energy storage to underpin the increased use of renewableenergy sources [1,2]. Within each application, a key enablingtechnology is thedesign and integrationof thepower electronicsubsystems that are requiredto manage theflowof energy [3]; suchsubsystems areknown to generateundesired electrical noise [4–6]. For example, Fig. 1 shows the current waveformfromcharginga NissanLeaf vehicleusinga commer-ciallyavailable3 kWEltekValere bi-directionalvehiclecharger.

Fig.1 highlightsthemeasuredharmoniccontentpresentonthe electricalinputtothebattery.

Highfrequencycurrentoscillations,orripple,ifunhinderedwill addafurtherperturbativeloadontothevehicle’sbatterysystem.In

recentwork,Uddinet.al.[7]measuredtheelectricalnoiseonthe DC linkofahighvoltagebusonapre-production serieshybrid electricvehicleduringaregenerativebreakingevent[7].Thisdata was used to define the current waveforms that batteries are exposed toduring typicalelectricoperation. Longterm battery degradation resulting from cycling batteries under such wave-forms was studiedin [7]. The authors showthat exposing the batterytocoupleddirectcurrent(DC)andalternatingcurrents(AC) lead to additional battery degradation that is related to the frequency of the perturbative ripple current. Through the experimental datapresented,theauthorswereabletoquantify theimpactofACexcitationatthecell-level.However,theauthors werenotabletoprovideanyreasoningforthebatterydegradation. Furthermore,giventhenoveltyoftheresultpresented[7],itwas notpossibletoexplainthecausalityofdegradationduetoACripple excitationdirectlyfromliterature.

Thispaperextendstheresearchpresentedin[7]byseekingto explaintheunderlyingcausalityofdegradationresultingfromhigh frequencyripplecurrent.Thisisachievedthroughacombinationof furtherexperimentalanalysisandnovelmathematicalmodelling.

* Correspondingauthor.Tel.:+4402476150736;fax:+4402476524307.

E-mailaddress:[email protected](K.Uddin).

http://dx.doi.org/10.1016/j.electacta.2017.03.020

0013-4686/©2017TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/). ContentslistsavailableatScienceDirect

Electrochimica

Acta

X-rayphotoelectronspectroscopy(XPS)isemployedtostudythe degradedcells used inourpreviouswork[7].A heterogeneous electrochemical model based on continuum theory has been developed and used to investigate the dependency of cell degradationonthefrequencyofanACperturbation.Themodel relates battery degradation to the frequency dependant heat generationin the electrical double layer resulting from super-imposedACloadswhichisknowntopromotemorepronounced growthof solidelectrolyteinterface(SEI)film[8].Thermaland reference electrode relaxation experiments are presented that demonstratetheprocesses involved,tofurther understandand explainbatterydegradationwhenthecellisexposedtoACcurrent excitation.

This paper is structured as follows: in the next section, experimental methodologies pertaining to results presented in this paper are agglomerated and presented. In Section 3, a heterogeneous electrochemical model that accounts for ion dynamics in the electrical double-layer through the Poisson-Boltzmann equation is presented. The model suggeststhat the causeofcelldegradationundercoupledACandDCwaveformsis theincreasedheatgeneration.Themodelresultsaresubstantiated byexperimental results presentedinSections4. Longtermcell ageingtestresults,wherecellswerecycled1200timesunderanAC andDCcoupledwaveform,arepresented;XPSautopsyresultsfor thesecells showa frequencydependantgrowth ofsurfacefilm. Finally,inSection5weconcludeandstatefurtherwork.

2.Experimental

2.1.LongTermAgeingTest

It isbeyondthescopeofthispapertoexplain,indetail,the originalexperimentalageingstudy.Thisworkispresentedin[7].A full description of the aims and objectives, the methods and equipmentemployedarepresented.However,forcompletenessa briefsummary is provided in this subsection, highlighting key aspectsofthework.

Fifteen commerciallyavailable 3.0 Ah18650-typecells were cycled[7].EachcellhadaLiC6negativeelectrode,Li(Ni0.822Co0.148

Al0.030)O2(NCA) positiveelectrode,separated bya polyethylene

separator, sandwiched between two current collectors and

immersed in an electrolyte solution comprising of ethylene carbonate(EC),ethylmethylcarbonate(EMC),vinylenecarbonate (VC)andLiPF6salt.

Each cellwas cycled usinga DC current signal(IDC) super-positionedwithanACperturbationIACsin(

v

t)suchthattheoverall current(I)waveformwas:

I¼IDCþIACsinð

v

tÞ ð1Þ

witharesultingcellpotential:

V¼VDCþVACsinð

v

t

u

Þ ð2Þ

where

u

isaphaseshiftbetweenthecellpotentialandcurrent waveform. The test signals were generated using a bespoke amplifiertogeneratetheACwaveformandaBitrodecellcyclerto generatetheDCloadprofile.Inaddition,aTektronixnon-contact currentprobeandoscilloscopewasusedfordataacquisitionand testmonitoring.ALAUDAheatingandcoolingsystemwasusedto ensurethattheambienttemperaturewasmaintainedat25_C.

TheDCportionofthecyclestartedwitha0.8Ccycledischarge (where Ccycle is the de-rated battery C-rate, i.e., the value of retaineddischargecapacityafteranageingcharacterisationtest) from95%SOCcycleto65%SOCcycle(whereSOCcycleisthestateof chargedefined byCcycle).Followingarestperiodof10minutes, each cellwasthen chargedusing astandard ConstantCurrent– ConstantVoltage(CC-CV)protocol,inwhichthecellwascharged (CC)totheuppercellpotentiallimitof4.1V(correspondingto95% SOCcycle)atwhichpointchargingcontinuedusingaCVmethod untilthevalueofcurrentreducedto0.15A.Superimposedontothe DCcyclewasoneoffourACcurrentexcitations,offrequency:10 Hz,55 Hz,254Hz and14.8kHz–withapeak-to-peakcurrentof 1.2Ccycle.Toimprovetherobustnessofthetestmethodandthe efficacyoftheresults,threecellsofthesametypewereexercised usingeachcurrentwaveformdefinedabove.

Ageingcharacterisationwascarriedoutafter300,600,900and 1200completecharge–dischargecycles.Thecharacterisationtests involveda1Ccapacityretentiontests;pulsepowertestsusing10 secondpulsesat20%,40%,60%,80%and100%ofthemanufacturers recommendedmaximumcontinuouschargeanddischargecurrent at 90%, 50% and 20% state of charge (SoC); electrochemical impedancespectroscopytestswerealsocarriedoutat90%,50% and20%SoCusingaSolartronModuLabEISSystem.

Fig.1.ThelefthandpanelshowsthemeasurementofacoupledDC-ACcurrent.Thescopesettingswere:100msperdivision,with106

2.2.CellAutopsy

Threesamplesweretakenfromeachcell,approximately160 mmfromthetopofthelatitudinalaxis(wheretopisdefinedasthe startofthejellyroll)alongthefirsthalfofthelengthofthejellyroll asopened.Thispositionwaschoseninordertoavoidthecentral bandonthelongitudinalaxisofan18650- typecellwhich has unusuallyhighresistancesattributedtopoorelectrodewetting[9]. Thissample locationalsoavoidstheinner windings ofthecell which ispronetode-lamination [9].Allthree samplesshowed good agreement in peak positions and widths. To ensure that sampleinhomogeneitydoesnotaffectresults,alargeellipticalwas usedequatingtoananalysedelectrodeareacontainingbetween 500and1000graphiteparticles.

XPS provides information on the chemical environment of elementalspeciesfromthetop<10nmofasamplesurface.Since electrodesamplesareusuallycoveredwithanumberofdeposits, inthisworkthesamplewassputteredat500eVwithanAr+ion sputter gun for 90seconds to remove the salts, solvents and depositsfromthetopoftheelectrodesurface.Thesurfacefilmwas deemed ready for analysis if the relative peak areas after an episodeof sputteringdidnot change.Thissametechniquewas employedinarecentstudypublishedin[9].Asshowninthework of Somerville et. al., it is possible to determine the total contribution tothe chemical environments of the surface film (e.g., C-O, because it surrounds the active material particles compared to graphite C-C), providing a relative surface film thickness[9].Afteracquisition,allspectrawerereferencedtotheC 1speakat 284.4eV(aftersputtering).Thepeakswereassigned usingShirleybackgroundsandmixedGaussian-Lorentzian(Voigt) lineshapes.

Sampleswereanalysedin duplicatewitha KratosAxis Ultra DLD spectrometer and a Thermo k-alpha+ _{spectrometer. Both}

maintainedabasepressureofcirca.21010_{mbarduringanalysis.}

XPSmeasurementswereperformedusingamonochromaticAlk_a x-ray sourceand conducted at roomtemperatureat a take-off angleof90 withrespecttotheparallelsurface.The corelevel spectrawere recordedusinga pass energyof 20eV(resolution approx.0.4eV),acrossanellipticalareaofcirca.300

_m

m700

_m

m.

2.3.CellPotentialRelaxation

Cell potential relaxation in the electrolyte adjacent to the electrodewasinvestigatedusingthreeelectrodecellscontaining threelithiummetalelectrodes(working,counterandreference). Lithium electrodes were preferred over intercalation electrode materialsbecauseifthelatterhadbeenused,thecellpotential relaxationtransientwouldalsoreflectsolidstatediffusion,either towardsorawayfromtheparticlesurface.TheSwagelok1three electrode cells, which were used for this test, contained three layersofseparator;aglass_fibrelayer(GF/A)tominimisetheriskof dendrites, with porous polyethylene layers on either side. The electrolytewas1moldm3_LiPF

6inEC:EMC(3:7),witha2wt%

(weightpercentage)VC.

The cells were tested on a Bio-Logic VMP3 multi-channel potentiostat.Aftersomeconditioningchargeanddischargepulses, individual tests were used for each measurement. A single frequency galvanostatic impedance was performed, with an appliedcurrentof+0.1mAcm2_or0.1mAcm2_{.Thefoursingle}

frequenciesusedwereselectedtomatchtheotherexperimentsi.e.

10 Hz, 55 Hz, 254 Hz and 14.8 kHz. The number of repeat

measurements was adjusted in the range 300–500. The AC perturbationwas appliedforone minute ineach case. Thecell potential relaxation curves following theAC perturbationwere thencomparedforthefourfrequencies.

Cell potential relaxation experiments are used to further deduce information on ion concentration at the electrode-electrolyteboundarywhichisknowntoimpactheatgeneration thoughtheheatofmixing.Experimentalresultsfromthisstageof theresearcharepresentedinanddiscussedinSection6.3.

3.Theory

In this section, we develop an electrochemical model that accuratelydescribesthehighfrequencydynamicsonalithiumion battery.Thismodelisthenemployedtounderstandthetheoretical impactofcoupledACandDCwaveformsonlithium ionbattery degradationwithinthecontextofelectrochemistrytheory.

3.1.Modelformulation

Withinthefieldofelectrochemicalenergystorage,DoyleFuller and Newman’s model [19] based on principles of transport phenomena,electrochemistryandthermodynamicsiswell estab-lished.Themodelsolvesfortheelectrolytephaseconcentration:

@

ð

e

eceÞ

@

t ¼r D ef f e rce

þ1t

0

þ

F asj ð3Þ

solidphaseconcentration:

@

ð

e

scsÞ

@

t ¼

Ds r2

@

@

r r

2

@

cs

@

r

ð4Þ

electrolytephasepotential:

r

k

ef f e r

f

e

þr

k

ef f e;DrlnðceÞ

h i

¼asj ð5Þ

andsolidphasepotential:

r:

s

ef f s r

f

s

¼asj ð6Þ

where c is concertation,

f

is the electrode potential,

e

is porosity,Deff_,

_s

eff_andkeff_{aretheeffectivediffusioncoefficient,} conductivityandchargetransfercoefficient,respectively,t0

þisthe

transferencenumber,asisthespecificsurfaceareaoftheelectrode and j isthe current density;thesubscripts eand s denote the electrolyteand solidphase respectively. Doyle,Fullerand New-man’smodelconsistsofthreeprimarydomains(seeFig.2):the negativeelectrode(inthisworkLiC6),theseparator(inthiswork

polyethylene)andthepositiveelectrode(inthisworkLiNiCoAlO2)

permeated by electrolyte solution. During electrical discharge, lithiumionsthatoccupyinterstitialsiteswiththeLiC6electrode

diffusetothesurfacewheretheyreact(de-intercalate)andtransfer fromasolidintoaliquidphase.Theionsthendiffuseandmigrate throughtheelectrolytesolutionviatheseparatortothepositive electrode where theyreact(intercalate) and occupy interstitial siteswithinthemetaloxidematerial.Duringelectricalcharging the reversereactions takeplace, in summary:lithium ions de-intercalatefromthepositiveelectrode(metaloxide)and interca-latewithinthenegativeelectrode(carbon).Duetothewidespread acceptanceoftheDoyleFullerandNewmanmodel,itsufficeshere to refer interested readersto Refs.[19–22]for more details of modelformulationandboundaryconditions.

In theDoyle Fullerand Newmanelectrochemicalmodel the reactionkineticsaremodelledbytheButler-Volmerequation

j¼asj0 exp

a

aF RT

h

exp

a

cF RT

h

ð7Þ

positiveelectrode(

h

p)andSEIas(

h

SEI)[23]:

h

n¼

f

s

f

eU Rf ilmjLiþ

as j ð8Þ

h

p¼

f

s

f

eU ð9Þ

h

SEI¼

f

s

f

eUref Rf ilmje

as j: ð10Þ

ThepassivatingfilmresistanceRfilmisseparatedintoelectronic andionicconurbationsbecausetheSEIlayerispermeableto Li-ionsbutimpermeabletoelectrolyteandelectrons.Ontheother hand, the anodic reaction xLi+_+xe+Li

1xC6$LiC6 takes place

inside the electrode particle; the ions that participate in the reactionpassthroughtheSEIreducingtheirelectricalpotential. Thisisrepresentedbythelastterminequation(8).

It is wellunderstoodthat theButler-Volmer equation (7) is unable,in thehighfrequencyregime,toreproduce thecoupled resistanceandsurface-capacitancebehaviourobservedin electro-chemicalimpedancespectroscopytests[20].Toaddressthisissue, theauthorsextendtheoriginalmodel,employingthetechnique presentedbyXiaoand Choe[24].The electrolyteadjoiningthe particleisdividedintotworegions:thebulkregionandthedouble layer(DL) which itselfis composed of thediffuse layerwith a thicknessof110nmandtheSternplanewithathicknessequalto thesumofthediametersofananionandasolventmolecule[25].A schematicdiagramoftheseregionsaredepictedinFig.2.Lithium ionstravelthroughtheDLandparticipateinchemicalreactionsat thesurfaceofthesolidparticleswhiletheanionsthatdonottake partinthechemical reactionsareblocked outsideof theStern plane[24].Duetotheensuingelectrolytepotential,theDLregionis assumednottobechargeneutralwhilethebulkischargeneutral. Furthermore,thetransport ofcations (lithiumions) andanions (PF6-)intheDLisassumedtobegovernedbytheirrespectivelocal chemicalpotential gradientsdeterminedbyelectrostatic poten-tials and ion concentration gradients. In this formulation, ion kineticsintheDLregionisgovernedbythePoisson-Boltzmann

equation

r2

f

DLþ F Ec

þ_c

ð Þ¼0 ð11Þ

whereEisthepermittivityofdielectricsolvent,FisFaraday’s constant and the subscripts+and_–relate to cation and anion respectively.Chargeconservationisgivenby

@

cDL

@

t ¼

b

Fr2

@

@

r r

2_c

DL

@m

DL

@

r

ð12Þ

wherethesubscriptDLdenotesthedoublelayer,

b

istheion mobilityand

m

isthechemicalpotentialdefinedas

m

DL¼F

f

DLþRTln cDL

: ð13Þ

Insteadofelectroneutrallity,asassumedbyDoyle,Fullerand Newmanforthebulkregion,herethePoisson-Boltzmannequation isusedtocalculatethechemicalpotentialgiveninequation(13). Formoredetaileddiscussiononthederivationofequations(11) -(13),readersaredirectedtoRef.[24].Modelparametersusedin thisstudyareprovidedintheAppendix.

3.2.NumericalMethod

areiterateduntiltheresidualerrorreducestobelowthespecified thresholdvalue.Thisstudyassumes1.01003A/cm3astheerror limitforcurrentdensitiesonbothelectrodes.

3.3.Electrochemicalmodelresults

Duringoperation, abattery typicallyworksatcellpotentials

beyond the thermodynamic stability range of the organic

electrolyte.Therefore,electrolytedecomposition,whichincludes theoxidationandreduction ofelectrolyteonthesurfaceof the cathodeandanodeoccurs.ThefactorsaffectingthesubsequentSEI layerformationonthenegativeelectrodeincludecarbonsubstrate, electrolytecomposition,additivesandtemperature[26].Assuming that the pertinent cause of degradation is the increase in passivating surface film–aswill bedemonstrated in Section 4– andgiventhatthecellsusedinthestudywereidentical,i.e.,carbon substrate,electrolytecompositionandadditiveswerethesame;it ispostulated thatdiscrepancies in batterydegradation resulted fromvaryingheatgenerationratesthatwerepresentwhen the cellswereexcitedbydifferentwaveforms.

The cell potential response and cation concentrationin the sternplaneunderacurrentloadoftheformofequation(1)with varyingfrequenciesisshowninFig.3.Ionkineticsisdrivenbythe chemical potential

m

(c.f., equation (13)) which includes both concentration and potential terms. Under the AC-DC coupled waveformsconsideredinthiswork,thegradientofthechemical potentialisalwaysone-directional,i.e.,eitherpositiveornegative. Since thepotential gradientis increasing,theamplitude of the oscillating concentration in the stern plane is also increasing (though not visible in Fig. 3 due to the short time increment presented).Withregardstocellpotential,thediscrepancyinthe amplitude of the cell potential response when galvanostatic waveformswithdifferentfrequenciesareappliedonacellresults from frequency dependant impedance of the electrochemical system.

The Butler-Volmer equation governs ion kineticsin the low frequencyregime(<1Hz);thePoisson-Boltzmannequationonthe otherhandgovernshighfrequencykineticsintheDLregionand hencealsothedependencyofheatgenerationonfrequencywhen

v

10Hz. ThesimulatedheatgenerationintheDLis shownin

Fig.4,whichhighlightsdifferencesinsimulatedheatgeneration arisingfromcurrent perturbations of differentfrequencies.The

heat generation is caused by the deviation of the surface overpotentialfromtheequilibriumstate,i.e.,resistivedissipation. Localheatgenerationp½W=cmofbothionsinDLiscalculated astheproductofchemicalpotentialgradientandionflux:

pDL¼4

p

r2

@m

DL

@

r N

DL ð14Þ

wheretheionfluxinthedoublelayerNDLmol=s=cm2

isgivenby

[24]:

NDL¼

b

_F cDL

@m

DL

@

r ð15Þ

where

b

ionmobility.Theheatgenerationforthewholecellcan thenbeobtainedbyintegrationofpDL.ThecationfluxNþDL,whichis similarforallfourfrequencies,isdeterminedbyloadcurrent.The lithium ion chemical potential

m

þDL however, can be different between frequencies because it is affected by the electrostatic potential

f

DLwhichinteractswithanionconcentration.Theextra heatgenerationoflithium ionat14.8kHzisthuscaused bythe higher electrostatic potential

f

DL induced by the transient processesofanionsintheDL.

Themodellingresultsshowthat,fortheNCA/C6cellstudiedin

thiswork,aperturbationfrequencyof14.8kHzleadstothelargest heatgeneration.Furthermore,whileaperturbativefrequencyof 254Hzgenerateslargeranionsourcedheatthan

v

_<254Hz,the totalheatgenerationisdominatedbythedynamicsofthecations; heatgenerationduetoanionsisconsequentlymarginalised.

For the cell consideredin this study, the ion mobility

b

+ is

3.241010_cm2_V1_s1 _and

_b

is 1.851010cm2V1s1. The magnitudeofcationfluxisthuslargerthantheanionflux.This is related to the larger reservoir of cations in the electrode, facilitatingarelativelymoresustainablecationfluxcomparedwith anionflux.Theconcentrationofanionsinthedoublelayeronthe otherhandislargerthancationconcertation,asfoundinRef.[24]. Electrochemical modelling results presented in this section suggestthatheatgeneratedduetoresistivedissipationwithinthe DLisrelatedtoionkinetics(i.e.,ACexcitationfrequency)described by the Poisson-Boltzmann equation. Since heat is known to acceleratethe growth of SEI,the modellingresults are usedto postulatethatthegrowthofSEIfilmisdependentonthefrequency ofanACperturbation.Inthenextsection,experimentalresultsare presentedtoverifythefrequencydependantgrowthofSEI.

Fig.3.ThelefthandpanelshowscellpotentialresponsetoacurrentperturbationoftheformIDC+IACsin(vt).Therighthandpanelshowslithiumconcentrationinthestern

4.ResultsandDiscussion

4.1.LongTermAgeingTestResults

Electrochemicalimpedancespectroscopy(EIS)results compar-ingtestcellspriortocyclingandafter1200cyclesareshownin

Fig.5.TheresultspresentedinFig.5areforsinglecellswiththe medianvalueforresistancerisefromthesubsetofcellsineach frequencyband[7].Usinganalogieswithelectricalcircuitmodels, thepronouncedadditionalsemi-circleinthehighfrequencyregion of theNyquist plotforthe 254Hzand 14.8kHz ACexcitations indicatethatthereisanadditionalincreaseinresistancecoupled withsurfacelayercapacitancescomparedwiththecaseof55Hz, 10 HzandpureDCcycling.

Thefallin1Cdischargecapacityandtheriseinresistanceare showninFig.6.Capacityandpowerfadeduetosuperpositioned AC perturbationswithafrequencyof55 Hzand 10Hz areonly marginallydifferentfrompureDCcycling.Ontheotherhand,at higherfrequencies(14.8kHz),thereisanadditionalcirca.2.5%and 5%capacityandpowerfaderespectively.Giventhatautomotive end of life is defined by 20% capacity fade and a doubling of impedance[10–12],higherfrequencyACperturbationscanresult inanextra12.5%lossofusefulcapacityand 5%lossofrequired powercomparedwithlowerfrequencyACperturbations, poten-tiallyshorteningtheoperationallifetimeofanautomotivebattery byuptoone-and-a-halfyears[13].

ThebroadobservationthatcouplinganACperturbationtoaDC loadcausesmorebatterydegradationthanapureDCloadisclear, sincetheenergythroughput(Ethro)underagalvanostaticloadof theform(1)is:

Ethro¼IDCVDCtIACVDC

cosð

v

tÞþ

v

þIDCVAC

sinð

u

Þsinð

v

tÞcosð

u

Þcosð

v

tÞ

v

þIACVAC cosð

u

Þt

2

cos_ð

u

_Þsin_ð2

v

t_Þþsin_ð

u

_Þcos_ð2

v

t_Þ 4

v

;

ð16Þ

Inahighfrequencyregime(

v

>>1),equation(3)reducesto

EthroffiIDCVDCtþIACVACcosð

u

Þ t

2 ð17Þ

whichisdependentonfrequencythroughthephaseangle

u

(as will be shown in Section 6). Equations (16) and (17) clearly demonstratethatelectricalenergythroughputishigherwhenAC wavesaresuperimposedontoDCloads(i.e.,whenIAC6¼0and

p

/ 2<

u

<

p

/2).Thecorrelationbetweenincreasedenergythroughput andincreasedbatterydegradationiswellestablished[14].Inthis sense,theresultthatcouplinganACexcitationtoaDCloadleadsto increaseddegradationconformswithwell-establishedtheory.Itis importanttonotehowever,thatenergythroughputisnottheonly stressfactorwhichacceleratesbatterydegradation.Elevatedcell temperatureisknowntoresultinmoreseveredegradationmodes

[8].Arguably,themorenoteworthyresult–intermsofnovelty–is thedependencyofbatterydegradationonfrequencywhenIAc6¼0. Theremainderofthispaperexploresthisdependencyfurther.

While the results presented in Figs. 5 and Fig. 66, at a componentlevel, showthat coupledAC perturbationslead toa levelofdegradationthatexceedsthatexperiencedwithDCcycling in isolation, thefrequency dependencyof this degradationand fundamentalcausesofdegradationarenotyetfullyunderstood.In thefollowingsubsection,postmortemexaminationofthecycled cells is presented to elucidate the mechanism of degradation exacerbatedbysuperpositioningACcycling.

4.2.XPSCellAutopsyResults

TocharacterizethesurfacefilmformationresultingfromDC-AC coupled cycling, XPS analysis was completed on the graphite electrodes of four cells–(a)a control sample cell that was not cycled,(b)acellafter1200DCandACcoupledcyclesat10Hz,(c)a cellafter1200DCandACcoupledcyclesat55Hzand(d)acellafter 1200DCandACcoupledcyclesat14.8kHz.TheXPSresultsforthe C1sspectraarepresentedinFig.7.Theassignmentofchemical environmentsareconsistentwiththosefoundinliterature[15–

17].Fig.8showstheareaforthenon-graphitechemical environ-ments(concentration)relativetothefrequencyoftheACripple. However, it excludes the Li-C component in the cells. This is becauseFig. 7 shows a largeLi-Ccomponent in the‘new cell’ comparedtoallsubsequentcells.IthasbeenproposedthatLi-Cis formedfromreactionofdimethylcarbonatewithLi+_followedbya

radicalinducedchainreactionwithlithiumthatisfacilitatedbythe presence of H2O and thus would primarily form directly after

manufacture[17].Thisworksupportsthatassertionbyshowinga

much higher quantity of Li-C in the new cell and lower

concentrations in each subsequent cell after testing. However, whentheLi-Cpeak isincludedincalculatingthepercentageof surfacefilm,thenewcellshowsagreaterpercentageofsurface filmformationthanthecellcycledat55Hz,eventhoughthelatter hadundergonetestingfor6months.Becausethecellperformance does not improve in the same period, its inclusion skews

subsequent results. Therefore, the values recorded in Fig. 8

discounttheLi-Cpeakascontributingtosurfacefilmeventhough itisnotgraphiteactivematerial.

Fig.7showstheformationofesters(O-C=O)andethers(C-O). Thisisconsistentwiththechemicalreductionoftheelectrolyte solventsatthesurfaceofthenegativeelectrode.Anincreaseinthe concentrationofthesecompoundsisconsistentwithsurfacefilm formation[18].Fig.8showsthattheconcentrationofsurfacefilm increaseswiththeACfrequencyused.However,thecellsubjectto 55Hzhasonly19%surfacefilmcomparedto35%at10Hz;dueto lowerconcentrationsofbothO-C=OandC-Ocomponents(Fig.7). Thecellcycledat14.8kHzcontainsmuchhigherconcentrationsof surfacefilm,primarilyduetoincreasesinchemicalcomponentsat binding energies consistent withan O-C=O chemical environ-ment.

Theseresultsestablishadependencyofpassivatingfilmgrowth onthe frequencyof anAC perturbation.Based onXPSanalysis alonetheworstcyclingconditionforcellsisathighfrequencies, reducingtohavelessimpactatlowerfrequencies.However,there isaperceivedanomalyatoraround55Hz.

Thus far, electrochemical modelling results presented in Section3,suggestthatheatgenerationwithinthecellisrelated totheACexcitationfrequencyofacurrentload.Longtermageing testresultsforcellscycledunderanumberofACandDCcoupled waveforms showthat capacity and powerfadeis progressively morepronouncedforhigherACexcitationfrequencies.XPSresults Fig.5.ThelefthandpanelshowsinitialEISresultsforallcellsmeasuredat50%SOC;therighthandpanelshowsEISresultsforcellsatdifferentfrequenciesafter1200cycles.

Fig.6. Thelefthandpanelshowsaverage1Ccapacityfadeafter1200cycles;therighthandpanelshowsaveragepowerfade(impedancerise)resultsforcellsatdifferent frequenciesafter1200cycles.Theerrorbarsindicatethemaximumandminimummeasuredvalues.Foraddedconfidence,thesemeasurementswerecarriedouttwice,once usingtheBitrodecellcyclerandasecondtimeusingtheMaccor4200Modelcellcycler;theresultsforeachcellwasthesametowithintheerroroftheequipment.For14.8

corroboratethisfindingandfurtherestablishesthatthepertinent modeof degradation is thegrowth ofpassivating _films onthe negative electrode. Thus, a causal link is proposed between frequencydependantheatgenerationandSEI filmformation.In thefollowingsubsectionstherefore,resultsanddiscussionrelating toheatgenerationwithinthelithium-ioncellsispresented.

4.3.CellSurfaceTemperatureMeasurements

ItiswellestablishedthatthegrowthoftheSEIlayerisstrongly dependentonthetemperature[8].Thedifferenceincellsurface temperature resulting from the various current waveforms considered in this work is shown in Fig.9. The differences in cell surface temperature resulting from superpositioning AC harmonics is qualitatively aligned with cell ageing results presented in Section 4.1 and with the growth of surface film measured from XPS results presented in Section 4.2. More specifically, a 55 Hz excitation exhibited the lowest cell Fig.7.XPSanalysisresultsoftheC1sspectrumofanewcellandthreecellssubjectedto1200DCcycleswithdifferentimposedACfrequencies(10Hz,55Hzand14.8kHz) withtheresultsofpeakfittingshowingpossiblechemicalstates(C-C,C-O.C-Li,O-C=OandROCO2Li).

temperature (see Fig. 9) in agreement with XPS results that showedleastsurface_filmgrowthatthisexcitationfrequency(see

Fig.8).

4.4.HeatGenerationWithinCells

There are four sources of heat generation in a lithium ion battery under operation, which include: heat generated from resistivedissipation,theentropyofthecellreaction,sidereactions andtheheatofmixing.Theenergybalancefortheelectrochemical systemisexpressedas[27]:

_

Q¼Ið

f

s

f

eUÞþIT

@

U

@

t þ

X

i

D

Havg_i ri

þZ X j

X

k

HjkHavgjk

_@

_cjk

@

tdv ð18Þ

The first term on the right-hand side of equation (18) I (

f

s

f

eU)representsirreversibleresistivedissipationcausedby thedeviationofthesurfaceoverpotentialwhichisthedifference between the solid phase potential (

f

s) and electrolyte phase potential (

f

e)fromthevolume averaged opencircuitpotential (OCP)duetoaresistanceofthepassageofcurrent(I).Thisterm

differswiththatofRef.[27]inthattheheatgeneratedbycelltabs arenotconsidered.Thesecondtermisthereversiblerateofheat generationduetoelectrochemicalreactions,definedasIT@U

@t ¼T@ S

@t whereSisthetotalentropyofthesystemandTisthebulkcell temperature.Thethirdtermistheheatarisingfromchemicalside reactions where theenthalpyand rateof chemical reactioniis

D

Havg_i andrirespectively.Thelasttermistheheatofmixingdueto the formation and subsequent relaxation of concentration gradients,wherecjkistheconcentrationofspeciesjinphasek, Hjk andHavgjk arethepartialmolarenthalpyandaveragedpartial molarenthalpyofspeciesjinphasekrespectively,anddvisthe differentialvolumeelement.

Themagnitudeofheatgeneratedbyresistivedissipationand entropicreactionarecomparable[27].InFig.10wepresentthe impedanceofthecellwhichquantifiesresistivedissipationwhich, at a cell level, can be represented as I2_Z

cell. The entropic

contributiontoheatgenerationisquantifiedthroughchangein entropyDS

nF ¼@ U

@t wherenisthenumberofelectronsinvolvedina half-cellreactionandFistheFaradayconstant.Bothoftheseterms are dependent on lithium concentration in the lattice, i.e., stoichiometry.Furthermore,whileresistivedissipationisalways exothermic,inprincipleentropicheatcanbeeitherexothermicor endothermic, dependingontheentropyofthereactionandthe directionofcurrentandSoC.

Theresistance(realpartoftheimpedance)inFig.10showsthat batteryimpedanceisfrequencydependentandconsequentlyheat generation within the battery will be a function of excitation frequency. In agreement with electrical results presented in Section3,the14.8kHzexcitationfrequencyis showntohavea higherresistancevaluethanacurrentperturbationat10Hzwhich itself hasa higher resistance than a 55 Hz AC waveform. This corroborates the cell surface temperature results presented in

Fig.9 andexplainswhy 55Hz resulted inless passivating film formation.IntheBodephaseplot,thepeakoccurringcloseto105

Hzcorrespondstothesurfacefilm.SEIfilmgrowthoccursatboth boundariesoftheSEI[17,28],althoughtheseperatorsidegrowth maybedominant.UnderelevatedtemperaturestherateofSEIfilm growthatbothboundariesisincreased.FortheinnerSEIboundary, thisoccursbecausethediffusivity(D)ofionsislesshinderedat elevatedtemperatures.Thistemperaturedependencyisdescribed Fig.9.Showingcellsurfacetemperatureriseundera600seconddischargeusing

thedifferentcurrentwaveformsconsideredinthiswork.

bytheArrheniuscorrection,

D¼Dref Eact_{R Tref}1 _T1

ð19Þ

whereEactistheactivationenergy.TheBodephaseplotalso showsashortplateauregionbetweenthefrequencylimits102 101_{Hzcorrespondingtothesemi-infinitediffusionregion.}

Cell surface temperature measurements presented in Fig. 9

suggestthat heat generationwithin a cell is related tothe AC excitationfrequency.Equation(18)proposesfoursourcesofheat generation, namely resistive dissipation, electrochemical reac-tions, chemical reactions and the heat of mixing. Impedance spectroscopy results presented in Fig. 10 show that resistive dissipationwithinthe cellis frequency dependant. In the next subsection,areferenceelectrodecellpotentialrelaxation experi-mentisemployedtoinvestigatetheheatofmixing.Cellpotential relaxation is indicative of ion concentration at the electrode-electrolyteboundary,suchthatarelativelylongerrelaxationtime represents higher ion concentration. If there is a significant differenceinionconcertationattheelectrode-electrolyte bound-ary when differentexcitation frequencies areapplied, this will revealanunderlyingfrequencydependenceoftheheatofmixing. Furthermore,assumingfirst-ordersolventdecompositionkinetics, ahigherreactantconcentrationatthenegativeelectrodeboundary will,itself,facilitateanincreasedrateofSEIgrowth[29].

4.5.CellPotentialRelaxationExperimentResults

In an electrochemical system that is constantly perturbed, thermodynamicpropertiesaredeterminedbyhowthechemical distributionrespondstochanges.Therateofchangeofenthalpy (H)ofthesystemisgivenby

dH

dt ¼Q_ IV: ð20Þ

In Fig.11, theelectrode potentialafter a60 second DC step currentof0.1mA/cm2superpositionedwithanACperturbationof frequencies 10 Hz, 55 Hz, 254 Hz and 14.8 kHz is shown. The correlationbetweentheelectrodepotentialandsurface concen-trationisgivenby

Df

¼ð1tþÞRT_Fln

cejÞx¼Le csjÞx¼x’

!

ð21Þ

wherecejx¼Le

istheconcentrationofelectrolyteatthesurfaceof thenegativeelectrodeandcs|x=xistheconcentrationoflithiumat

x = x’ in the negative electrode [30]. A slower cell potential relaxationthereforeindicatesahighersurfaceconcentrationand consequentlyamorepronouncedgrowthrateofSEIthrough:1) theheatofmixing,and2)alargerionconcentrationgradient.

ThecellpotentialrelaxationforacoupledDCandACloadwith varyingexcitationfrequencies,showninFig.11,displaysnegligible differencesbetweenthewaveforms.Thissuggeststhattheheatof mixing will not be a significant distinguishing factor. The cell potentialrelaxationresultalsosuggeststhattheconcentrationof the reacting electrolyte species at the electrolyte-electrode interfaceisnotthereasonforintensifiedSEIformationathigher frequencies.

5.Conclusion

Thispaperhasstudiedthecausalityofincreasedcell perfor-mance degradationdue tosuperimposing current rippleonDC loads.Inparticular,therelationshipbetweenthefrequencyofan ACperturbationandtheresultingbatterydegradationmeasured. Firstly,equation(16)establishedthatforacurrentloadoftheform I=IDC+IACsin(

v

t)thecellenergythroughoutishigherthanwhen IAC=0indicatingthat,justfromanenergythroughoutperspective, superpositioning an AC perturbation will lead to increased degradation. Secondly, XPS results provide explicit evidence showing that the mechanism of degradation is the increased growth of surface film. In addition, XPS results highlight a dependency of surface film formation on the frequency of the coupled AC perturbation, corroborating the electrical ageing characterisation data. Using a heterogeneous electrochemical model, thefrequency dependentformation of surface filmwas attributedtoheatgenerationwithinthedoublelayer,whereion dynamicsis governed by thePoisson-Boltzmannequation. This causalityofbatterydegradation,i.e.,theprominenceofresistive dissipation and its frequency dependence, was established by measuringcell surface temperaturefor different current wave-forms (i.e.,varying frequency

v

).The results agreedwithboth electrical ageing characterisation and autopsy results. The processesthatgenerateheatwithinthecellwerealsodiscussed. Resistive dissipation was explicitly shown to be frequency dependentandthedistinguishingfactorinheatgenerationunder highfrequencysuperpositionedACloads.Inconclusion,ifanAC perturbationiscoupledtoaDCwaveform,therateofSEIgrowthis morepronounced.TheincreasedSEIgrowthisrelatedtoincreased resistivedissipationwhichisdependentonthefrequencyoftheAC perturbation.

5.1.FurtherWork

Anumberofopportunitiesexistwheretheresearchpresented heremaybefurtherextendedandrefined.Theresultscollected from both long term cycling experiments and autopsy studies suggestthatdegradation–forthecellspecimenconsideredinthis work–islowestwhenanACperturbationof55Hz,corresponding to the frequency exhibiting the lowest cell impedance in EIS results, is applied. More experimental and theoretical work is required to understand the precise nature of this effect and whetherthisfindingcanbeextendedtootherbatterytechnologies andcelltypes(i.e.,otherthanNCA/C6and18650-tyoecells).

The second areaof furtherwork, is toreduce the potential impactofcell-to-cellvariations(c.f.,Fig.5)byexpandingthescope oftheexperimentalstudytoencompassagreaternumberofcells ofagiventype.Expandingtheexperimentalprogrammeshould alsoinclude using cells from a broadercross-section of manu-facturersandchemistries. Thiswillidentifyiftheexperimental resultspresentedhereandinRef.[7]aretransferabletoothercell technologies.

In the work presented here (and in Ref. [7]), isolated AC perturbationsareconsidered.Inreality,asisshowninFig.1,there isacouplingofmultipleACperturbations.Anexperimentalstudy ontheimpactofsuchcomplexcouplingsbetweenDCandmultiple frequencyAC perturbations are required.Suchan experimental studyshouldalsoassesstherelativedegradationassociatedwith varyingthepeak–peakamplitudeoftheACcurrentwaveformin additiontotheexcitationfrequency.

Finally, given that the aim of this study was to assess the causality and impact of varying frequency for AC current perturbations,furtherresearchshouldbeundertakentodevelop andvalidateprognosticageingmodelswhichareabletoquantify thegradation.

Acknowledgements

ThisresearchwassupportedbyEPSRCgrants(EP/M507143/1) and(EP/N001745/1)andtheWMGCentreHighValue Manufactur-ingCatapult(fundedbyInnovateUK)inpartnershipwithJaguar LandRover(JLR).

AppendixA.Appendix

Negative Electrode

Separator Positive Electrode

Thicknessd(cm) 50104

25.4104

36.4104

Particleradius,Ls(cm) 1104 1104

Activematerialvolumefraction 0.580 0.500 Electrolytephasevolumefractionee 0.332 0.5 0.330

Maximumsolidphaseconcentration

cs,max(molcm3)

16.1103

23.9103

Stoichiometryat0%SOC 0.126 0.936

Stoichiometryat100%SOC 0.676 0.442 Averageelectrolyteconcentrationce

(molcm3

)

1.2103 _1.2103 _1.2103

Exchangecurrentdensityj0(Acm3) 3.6103 2.6103

Charge-transfercoefficientsaa,ac 0.5,0.5 0.5,0.5

SEIlayerfilmresistance,VSEI(V

cm2

)

100 100

SolidphaseLidiffusioncoefficient,Ds

(cm2_s1₎

21012 _3.71012

Solidphaseconductivity,s(Scm1

) 1.0 0.1

ElectrolytephaseLi+

diffusion coefficient,De(cm2s1)

2.6106

2.6106

2.6106

Bruggemanporosityexponent,p 1.5 1.5 1.5 Electrolyteactivitycoefficient,f 1.0 1.0 1.0 ReferencecellpotentialUref(V) 0 0

MolecularweightMp(kgmol1) 7.3104

DensityofSEILayerre(kgcm3) 2.1103

1.51012

(Continued)

Negative Electrode

Separator Positive Electrode

Sidereactionexchangecurrent densityj0s(Acm3)

ConductivityofSEILayerkp(Scm1) 1104

Densityofelectrolytere(kgcm3) 1123106 1123106 1123106

Densityofsolidphasers(kgcm3) 1343106 2328.5

106

ElectrolytethermalConductivityle

(Wcm1

K1

)

3.39102

3.39102

3.39102

Solidphasethermalconductivityls

(Wcm1

K1

)

3.39102

3.39102

Cationmobilityb+ 3.241010 3.24

1010

3.24

1010

Anionmobilityb+ 1.851010 1.85

1010

1.85

1010

References

[1]M.Armand,J.-M.Tarascon,Buildingbetterbatteries,Nature451(7179)(2008 Feb)652–657.

[2]B.Scrosati,etal.,Lithium-ionbatteries.Alookintothefuture,EnergyEnviron. Sci.4(9)(2011)3287.

[3]T.Hutchinson,S.Burgess,G.Herrmann,Currenthybrid-electricpowertrain architectures:Applyingempiricaldesigndatatolifecycleassessmentand whole-lifecostanalysis,Appl.Energy119(2014)314–329.

[4]D.M.Erickson,W.Robert,Fundamentalsofpowerelectronics,SpringerScience &BusinessMedia,2007.

[5]M.Yilmaz,P.T.Krein,ReviewofBatteryChargerTopologies,ChargingPower Levels,andInfrastructureforPlug-InElectricandHybridVehicles,IEEETrans. PowerElectron.28(5)(2013May)2151–2169.

[6]C.Mi,F.Z.Peng,K.J.Kelly,andV.Hassani,“Topology,design,analysisand thermalmanagementofpowerelectronicsforhybridelectricvehicle applications,”J.Electr.HybridVeh.,1(3),pp.276-294.

[7]K.Uddin,A.D.Moore,A.Barai,J.Marco,Theeffectsofhighfrequencycurrent rippleonelectricvehiclebatteryperformance,Appl.Energy178(2016)142–

154.

[8]J.Vetter,P.Novák,M.R.Wagner,C.Veit,K.-C.Möller,J.O.Besenhard,M.Winter, M.Wohlfahrt-Mehrens,C.Vogler,A.Hammouche,Ageingmechanismsin lithium-ionbatteries,J.PowerSources147(1)(2005)269–281.

[9]L.Somerville,J.Bare,S.Trask,P.Jennings,A.McGordon,C.Lyness,I.Bloom,The effectofchargingrateonthegraphiteelectrodeofcommerciallithium-ion cells:Apost-mortemstudy,J.PowerSources335(2016)189–196. [10]D. Andre, C. Appel, T. Soczka-Guth, D.U.Sauer, Advanced mathematical

methodsofSOCandSOHestimationforlithium-ionbatteries,J.PowerSources 224(2013)20–27.

[11]M.Ecker,N.Nieto,S.Käbitz,J.Schmalstieg,H.Blanke,A.Warnecke,D.U.Sauer, CalendarandcyclelifestudyofLi(NiMnCo)O2-based18650lithium-ion batteries,J.PowerSources248(2014)839–851.

[12]A.Farmann,W.Waag,A.Marongiu,D.U.Sauer,Criticalreviewofon-board capacityestimationtechniquesforlithium-ionbatteriesinelectricandhybrid electricvehicles,J.PowerSources281(2015)114–130.

[13]R.Spotnitz,Simulationofcapacityfadeinlithium-ionbatteries,J.Power Sources113(1)(2003)72–80.

[14]S.Bourlot,P.Blanchard,S.Robert,Investigationofagingmechanismsofhigh powerLi-ioncellsusedforhybridelectricvehicles,J.PowerSources196(16) (2011)6841–6846.

[15]R.Dedryvère,S.Laruelle,S.Grugeon,L.Gireaud,J.-M.Tarascon,D.Gonbeau, XPSIdentificationoftheOrganicandInorganicComponentsoftheElectrode/ ElectrolyteInterfaceFormedonaMetallicCathode,J.Electrochem.Soc.152(4) (2005)A689.

[16]S.Leroy,F.Blanchard,R.Dedryvère,H.Martinez,B.Carré,D.Lemordant,D. Gonbeau,SurfacefilmformationonagraphiteelectrodeinLi-ionbatteries: AFMandXPSstudy,Surf.InterfaceAnal.37(10)(2005Oct)773–781. [17]D. Aurbach, Review of selected electrode–solution interactions which

determinetheperformanceofLiandLiionbatteries,J.PowerSources89 (2)(2000)206–218.

[18]D.Aurbach,Y.Talyosef,B.Markovsky,E.Markevich,E.Zinigrad,L.Asraf,J.S. Gnanaraj,H.-J.Kim,DesignofelectrolytesolutionsforLiandLi-ionbatteries:a review,Electrochim.Acta50(2)(2004)247–254.

[19]M.Doyle,J.Newman,Theuseofmathematicalmodelinginthedesignof lithium/polymerbatterysystems,Electrochim.Acta40(13–14)(1995Oct) 2191–2196.

[20]J.S.Newman,K.E.Thomas-Alyea,Electrochemicalsystems,J.Wiley,2004. [21]K.Smith,C.-Y.Wang,Solid-statediffusionlimitationsonpulseoperationofa

lithiumioncellforhybridelectricvehicles,J.PowerSources161(1)(2006Oct) 628–639.

ElectrochemicalBatteryModelParameters,Batteries2(2)(201626-Apr)13 MultidisciplinaryDigitalPublishingInstitute,.

[23]P.Ramadass,B.Haran,P.M.Gomadam,R.White,B.N.Popov,Developmentof FirstPrinciplesCapacityFadeModelforLi-IonCells,J.Electrochem.Soc. 151(2) (2004)A196.

[24]M. Xiao, S.-Y. Choe, Impedance model of lithium ion polymer battery consideringtemperatureeffectsbasedonelectrochemicalprinciple:PartIfor highfrequency,J.PowerSources277(2015Mar)403–415.

[25]D.C.Grahame,TheElectricalDoubleLayerandtheTheoryofElectrocapillarity, Chem.Rev.41(3)(1947Dec)441–501.

[26]M.Lu,H.Cheng,Y.Yang,Acomparisonofsolidelectrolyteinterphase(SEI)on theartificialgraphiteanodeoftheagedandcycledcommerciallithiumion cells,Electrochim.Acta53(9)(2008)3539–3546.

[27]K.E.Thomas,J.Newman,ThermalModelingofPorousInsertionElectrodes,J. Electrochem.Soc.150(2)(2003)A176.

[28]Y.Choquette,G.Brisard,M.Parent,D.Brouillette,G.Perron,J.E.Desnoyers,M. Armand,D.Gravel,N.Slougui,SulfamidesandGlymesasAproticSolventsfor LithiumBatteries,J.Electrochem.Soc.145(10)(1998)3500.

[29]M.B.Pinson,M.Z.Bazant,TheoryofSEIFormationinRechargeableBatteries: CapacityFade,AcceleratedAgingandLifetimePrediction,J.Electrochem.Soc. 160(2)(2012Dec)A243–A250.