ISE2010年喷雾形成前电极

Publishedonline30March2010inWileyInterScience(www.interscience.wiley.com).DOI:10.1002/pip.938

RESEARCHARTICLE

Electricalpropertiesoffinelineprintedand

light-inducedplatedcontactsonsiliconsolarcells

M.Ho¨rteis*,J.Bartsch,S.Binder,A.Filipovic,J.Merkel,V.RadtkeandS.W.Glunz

FraunhoferInstituteforSolarEnergySystems,Heidenhofstr.2,D-79110Freiburg,Germany

ABSTRACT

Thepropertiesoffine-lineprintedcontactsonsiliconsolarcells,incombinationwithlight-inducedplating(LIP),arepresented.TheseedlayersareprintedusinganaerosolsystemandanewmetallizationinkcalledSISCdevelopedatFraunhoferISE.Theinfluenceofmultiplelayerprintingonthecontactgeometryisstudiedaswellastheinfluenceofthecontactheightonthelineresistivityandonthecontactresistance.Thedependencebetweencontactresistanceandcontactheightismeasuredusingthetransferlengthmodel(TLM).Furtheron,itisexplainedbytakingSEMimagesofthemetal–semiconductorinterface,thatacontactheightoflessthan1mmoraminimuminkamountofonly4–6mgissufficienttocontactalargearea(15Á6cmÂ15Á6cm)siliconsolarcellonthefrontsideandresultsinacontactresistanceRcÂW<0Á5Vcm.Asthelineresistivityoffine-lineprintedfingersneedstobereducedbyLIP,threedifferentplatingsolutionsaretestedonsolarcells.Theobserveddifferencesinlineresistivitybetweenrf¼5Â10À8and2Â10À8VmareexplainedbytakingSEMpicturesofthegrownLIP-silver.Finally,theoptimumLIPheightfordifferentlineresistivitiesiscalculatedandexperimentallyconfirmedbyprocessingsolarcellswithanincreasingamountofLIPsilver.Copyright#2010JohnWiley&Sons,Ltd.

KEYWORDS

siliconsolarcells;metallization;aerosolprinting;light-inducedplating(LIP);finelineprinting;contactformation*Correspondence

M.Ho¨rteis,FraunhoferInstituteforSolarEnergySystems,Heidenhofstr.2,D-79110Freiburg,Germany.E-mail:matthias.hoerteis@ise.fraunhofer.deReceived20April2009;Revised10August2009

1.INTRODUCTION

Themetallizationofsiliconsolarcellsisplayinganincreasinglyimportantroleinordertoproducesiliconsolarcellsmorecosteffectively.Animprovedfrontsidemetallizationhelpstoreachthisgoalbyincreasingthecellefficiencyandbyreducingthematerialcosts.Themainlossesofametalcontactonasolarcellaretheresistivityofthemetal–semiconductorinterface,thelineconductivityofthecontactfingersandalsotheiropticalshading.Astherequirementsonthematerialsforcontactformationandcurrenttransportaredifferent,itisbeneficialtooptimizethemseparately.Forthis,so-calledtwo-layercontact,severaldepositionmethodsandmaterialshavebeeninvestigatedatISEoverthelastfewyears[1,2].Themetallizationtechniquesforthefirstlayer,thecontactlayer,canbedistinguishedintotwotypesofcontacts:(i)low-temperatureprocesseswherethemetallayerisdepositeddirectlyontheemitterandametalsilicideisformedatamoderatetemperatureofT<4008Cand(ii)high-temperatureprocesseswherethecontactmetalis

240

depositedontheanti-reflectioncoating(ARC)andfiredthroughattemperaturesabove7008C.Thecontactmaterialisusuallydepositedinaprintingstepandcontainssilverparticlesandaglassfrit.ThehightemperaturesarenecessarytoopentheARCandtosupportthereactionbetweentheglassfritincludedinthemetalpasteandthesiliconsurface.Inthiscasenometalsilicideisformedbutsmallsilvercrystallites.Theseareindirectcontactwiththeemitterandresponsibleforthequalityofthemetal–semiconductorcontact[3–6].

Thesecondlayer,whichisresponsibleforthecurrenttransportcanbeprocessedbymultipleprintingofhighlyconductivesilverinksorbyplatingofconductivematerialslikecopperorsilver.Oneverypromisingwaytocreateahigh-performancecontactunderindustriallyfeasibleconditionsisthecombinationoftheaerosolprintingtechnology(AP)forseedlayerdeposition[7]andsucceedinglight-inducedplating(LIP)[8],toproducetheconductivelayer.Duetotheadvancedoptical[9]andelectricalpropertiesofanAP-LIPcontact,solarcellefficienciesgreaterthan20%havebeenachievedona

Copyrightß2010JohnWiley&Sons,Ltd.

M.Ho¨rteisetal.Finelineprintedandlipcontacts

SiO2-passivatedandLFC-contactedhigh-efficiencycellstructure[10].

InthisworkwehaveinvestigatedsomefundamentalaspectsofAPandLIPcontactslikethecontactgeometry,lineresistivityandcontactresistanceasafunctionoftheseedlayerheight.Wedeterminedtheminimallayerthicknesssufficientforanappropriatecontactresistanceaswellastheminimumofrequiredmaterialforabefittingelectricalconductivity.

2.EXPERIMENTAL

Anaerosolsystemwasusedtodeposittheseedlayercontactstructuresonsiliconsubstrates.Thecontactmaterialisasilver-richinkforfrontsidecontactscomparabletoastandardscreenprintingpaste.Theink,calledSISC(seedlayerinkforthemetallizationofsolarcells)hasbeendevelopedatFraunhoferISEforfine-lineAP.Tostudytheeffectsofmultiple-layerprintingonthecontactgeometry,metallinesweredepositedonshiny-etchedwafers.Thelineheightwasvariedbymultipleprinting(1–15layers),seeFigure1.Theprintedlineswerefiredinanin-linefurnaceandfinallyplatedinacyanidic(CN)platingbathusingthelight-inducedprocess.Thecontactgeometry,especiallythecross-sectionwasdeterminedbyconfocalmicroscopicmeasurementsandverifiedwithmicroscopicpicturesfrompolishedcontactcross-sections.Thelineresistanceforallcontactstructureswasmeasuredusinga4-probemeasurementandthespecificlineresistancewascalculatedtakingintoaccountthedeterminedcross-sectionarea.Additionally,therelationbetweencontactresistanceandinkdensity(g/m2)wasstudiedinordertodeterminetheminimumamountofink,whichisnecessarytocontactthefrontsideofasiliconsolarcell.Theinkdensityisquantifiedbymassofdryink(withoutsolvents)percontactarea.Contactstructureswereprintedwithdifferentspeedsandheights(differentinkdensities)onrandomtexturedCz-wafers.TheshallowdiffusedemitterwithasheetresistanceofRSH¼50V/sqiscoveredbyaPECVD-SiNx(x¼1Á075)anti-reflectivelayer.Theheightofthelineswasvaried

from1to9mm.Afterprinting,thecellswerefiredinanin-linefurnaceandsubsequentlyplatedbyLIP.Thecontactresistanceofeachcontactstructurewasdeterminedbyatransferlengthmodel(TLM)measurement.TheinfluenceoftheinkdensityonthecontactresistancecouldbeanalysedbytakingSEMimagesofthemetal–semicon-ductorinterface.Forthispurpose,thefrontsidemetallization,AgandtheglassfritwereselectivelyremovedbyaHNO3/HFsequenceandtheremainingAgcrystallitescouldbeinvestigated.InordertoavoidtheremovalofthePECVDnitridelayer,theetchinginHF(5%)waslimitedforallsamplesto4min.Witvrouwaetal[11].foundanetchingrateforPECVD-SiNxcoatingsof7Á6nm/minforHF(4%)solutions,thereforewecontrolledthepresenceoftheSiNx-coatingnexttothecontactsandcouldnotobserveseriousetching.

Inasecondexperiment,westudiedtheelectricalpropertiesoftheLIP-silver.Therefore,weprintedfineseedcontactsonindustriallypre-producedlarge-areamulti-crystallinesiliconsolarcellsandfiredtheminanconveyor-beltfurnace.Theinfluenceofthreedifferentelectrolytesonthelineconductivityofacontactfingerwasstudied..Toinvestigatetheinfluenceoftheelectrolyteontheconductivityofthecontactfingerthreedifferentelectro-lyteswereusedintheLIPprocess.Oneofthemwasacyanidicelectrolyte(CN-LIP)andtwowerenon-cyanidic(NCN1andNCN2).Thecontactfingerswereisolatedbyadicingsawandtheinfluenceoftheplatingsolutiononthelineconductivitywasdeterminedbya4-probemeasure-mentbeforeandafteranannealingstep.Additionally,thecontactmicrostructurewasanalysedbySEMimages.Furtheron,10multicrystallinesolarcellsofasizeof15Á6cmÂ15Á6cmwerecontactedwithaseedlayergrid.Thecellsweredividedintwogroupsoffivecells.EachgroupwassuccessivelyplatedandconsecutivelyIV-measuredinordertodeterminetheoptimumamountofLIP-silverfortwoplatingsolutions(CNandNCN1).

2.1.Aerosol-printedcontacts

ThecontactgeometryformultipleAPlinescanbeseeninFigure2.Thelineswereprintedontopofeachother.Generally,thelinewidthismainlydeterminedbytheopeningoftheprintingnozzle.Inthiscasetheopeningwas200mmresultingin50mmwidelines.However,smallerlinewidthscanalsobeachieved[12].Theheightofthesingleprintedlineislimitedto1mm,whichisconsistentwiththediameterofthesilverpowderusedinthecontactink.Hence,finercontactscanbeobtainedwithanink-containingfinerparticles.Eachfurtherprintingstepincreasesthelineheightbyanothermicrometer.Thecontactsaregrowingfasterinheightthaninwidth,indicatedbyalinearlyincreasingaspectratio.Thespecificlineresistivitywasdeterminedfromthemeasuredlineresistancetakingintoaccountthecross-sectionalarea.Forthesingleprintedcontactnovaluecanbegivenasthecontactisnotcontinuous(seeFigure4).Forallother

241

Figure1.Aspectratios(fingerheight:width)forsingle–15layer

aerosolprintedfingersasprintedandafterfiring.

Prog.Photovolt:Res.Appl.2010;18:240–248ß2010JohnWiley&Sons,Ltd.DOI:10.1002/pip.938

FinelineprintedandlipcontactsM.Ho¨rteisetal.

Figure2.Cross-sectionmeasurementsfrom1-to15-fold(one

overeachother)printedfingersafterthecontactshavebeen

fired.

Figure4.Topviewofasingleprintedandfiredseedlayer

contact.

contacts,thevaluesareillustratedinFigure3.Aconstantincreaseinthecross-sectiontogetherwithasteadydecreaseinlineresistanceshouldleadtoaconstantresistivity;nevertheless,themeasuredresistivityisdecreasingwitheveryprintingstep.Therearetworeasons:firstofall,thesinteringbehaviourofthesilverparticlesissupportedbyahigherinkdensity,asthesilverparticlescanfindaneighbourparticleinalldirections.Multiplecurrentpathsarepossible,whichreducesthelineresistivity.Second,theinfluenceoftheglassfritonthelineresistivitydecreasesduringthefiringstep.Theglassmelts,segregatesfromthesilver,diffusestowardsthesiliconsurfaceandtheremainingpuresilverformsacontinuousconductivelayer.Thecontactcanbebuiltuplinebylineandachieveresistivitiescomparabletoscreenprintedcontacts.Inconclusion,thehighertheseedlayers,thelowerthespecificlineresistivitybeforeplating.

However,thecontactwearelookingforissupposedtobeaseedlayer,whichhastobethickenedinaLIPprocesstoreachitsfinalresistivity.Themultipleprintedcontactsareplatedwiththesameamountofsilverandthelineresistivityismeasured.InFigure3theadvantageofathin,subsequentplatedseedlayerisobvious.Ingeneraltheresistivityforallplatedcontacts,independentontheseedlayerheight,isreduced.However,thebestconductivitywasachievedforthesingleprintedseedlayerwhichinitiallywasnotmeasurable.

Ingeneral,thelessmaterialisdepositedfortheseedlayer,theloweristhetotallineresistivityafterplating.Incaseofthesingleprintedcontact,alineresistivityofrf¼1Á9Â10À8Vmismeasured,whichisalreadyclosetothevalueofbulksilver(rf¼1Á59Â10À8Vm).Theresistivityofacontactstructurewithaflatseedlayerisdominatedbytheconductivityofplatedsilver.AstheconductivityofLIP-silverismuchhigherthantheconductivityofprintedsilver,AP,orscreen-printed,itisadvisabletokeepthefractionofprintedsilverassmallaspossible.Thecross-hatchedareainFigure3representstheresistivityofscreenprintedcontacts.Aslongastheseedlayerheightisbelow10mm,thetwo-layer-contactstructureissuperiorinresistivitycomparedtoscreenprintedcontacts.

Inafurtherexperimenttheinfluenceoftheseedlayerheightonthecontactresistancewasinvestigated.Tobeindependentofprintingparameterslikelinewidth,printingspeedandmaterialoutput,avalueforthedepositedinkmass,theinkdensity(g/m2),wasdefined.Forthiswemeasuredtheinkoutputoftheprinter,thelinewidthandtheprintingspeed.Fromthisdataanareadensitycanbecalculatedby

r¼

dvÂwc

(1)

Figure3.SpecificlineresistivityasafunctionofprintinglayersafterfiringandafterCN-LIP.Thedashedlinemarkstheconduc-tivityofbulksilverandthecross-hatchedareatheresistivityof

screenprintedcontacts.

Equation(1):Inkdensity(dry)width,whereristheinkdensity,dthethroughput,vtheprintingspeedandwctheseedlayerwidth.

242

Prog.Photovolt:Res.Appl.2010;18:240–248ß2010JohnWiley&Sons,Ltd.

DOI:10.1002/pip.938

M.Ho¨rteisetal.Finelineprintedandlipcontacts

Forexample,a45mmwide,withaspeedof20mm/sandamaterialoutput(dry)ofd¼0Á2mg/minprintedline,resultsinaninkdensityofr¼3Á7g/m2.Asolarcellwithanareaof15Á6cmÂ15Á6cmandagridof80fingersand2busbarscanbecontactedusingonly4Á4mgink.InFigure4atopviewofsuchafiredseedcontactisshown.Thesurfacetextureisstillvisibleandthecontactareaisnottotallycoveredbytheink,neverthelessthisisanexampleforacontactwithacontactresistanceRcÂWbelow0Á5Vcm.ThecontactresistancewasmeasuredusingtheTLM[13,14].ThecontactstructuresweremeasuredaftertheLIPstep.TheplatedsilverisnecessaryforreliableTLMresultsasitisnotpossibletocontactandmeasureathinseedlayerdirectly.Thecontactresistanceofthenon-platedcontactscanonlybemeasuredifenoughinkispresentandtheseedlayerheightexceedsabout4mm.Inthiscase,wedidnotfindadifferenceincontactresistanceforplatedorunplatedfingers.Thepositiveeffectofaplatingsteponafiredseedlayercontact[15]couldnotbeinvestigatedinthisexperimentastheseedlayeristoothinandnoleakageglasslayerwasfoundnexttothecontact.AllmeasuredcontactresistancesareshowninFigure5.ToprovethereliabilityoftheTLMmeasurementtheemittersheetresistance,alsoobtainedbyTLMmeasurementisplotted,too.Themeasuredsheetresistanceforeveryteststructurewasaround48Æ5V/sq,whichisconsistentwiththeexpectedRSHforthesecells(45V/sqAlthoughadecreaseinthecontactresistancewithraisinginkdensitywasassumed,wefoundanopposedbehaviour.Thehigherthecontactsandthehighertheinkdensitywere,thelowerwasthecontactquality.Withincreasinginkmasspercontactareathemeasuredresistancerose.Atacontactheightof4–5mm(corre-spondingtoaninkdensityof12g/m2)orhigherthecontactresistancestayedconstantatthelevelofagoodscreenprintedcontact(aboutRcÂW$0Á8Vcm).Tounderstand

Figure6.SEMimagesofthesilvercrystallitesandthecellsurfaceunderafrontcontact.(a)Singleprintedcontactwithv¼20mm/s(lowinkdensity<10g/m2).AhighcrystallitedensityisvisiblebuttheSiNx-coatingismainlyuntouched.(b)Mediuminkdensities(10–20g/m2),thecrystallitedensityisapproxi-matelythesame,butadditionallytheSiNx-coatingisstronglyetched.(c)Highinkdensity(>20g/m2),thesurfaceandthewholeSiNx-coatingisetchedaway,butthenumberofsilver

crystallitesorimprintsisstillconstant.

Figure5.Contactresistancemeasurementsfordifferentinkdensities(calculatedfromlinewidth,materialoutputandprintingspeed)andcontactheights.TheemittersheetresistanceisadditionallymonitoredtoproofthereliabilityoftheTLM

measurement.

thecorrelationbetweeninkdensityandcontactresistivity,thesilverwasetchedoffasdescribedaboveandSEMimagesweretakenfromthemetal–semiconductorjunction(seeFigure6a–c).Aconstantlyhighdensityofsilvercrystallites,whicharemainlyresponsibleforahigh-qualitycontact,couldbefoundindependentlyofthecontactheight.Thus,itisnotthecrystallitedensitywhichdegradesthecontactqualitywitharisinginkdensity.Additionally,anincreasingopeningoftheSiNx-passiva-tionlayerwasobserved.Incontrarytothecommonunderstanding[16],thecontactresistanceincreasedwithincreasingopeningoftheSiNx-layer.ThehighertheinkdensitywasthemoreSiNxwasetchedoff.Ahigherinkdensityisequivalenttoahigheramountofreactiveinkcomponents(glass)percontactarea.Thehigheramountofglassatthecontactinterfacehastwoeffects:first,theglass

243

Prog.Photovolt:Res.Appl.2010;18:240–248ß2010JohnWiley&Sons,Ltd.DOI:10.1002/pip.938

FinelineprintedandlipcontactsM.Ho¨rteisetal.

itselfformsathickerinsulatinglayerandsecond,theglass

etchestheSiNx-layerandaneventhickerglasslayerisformedduringfiring.TheSiNx-layerisoxidizedtoSiO2bythemetaloxidesintheglass.TheSiO2issolvedintheglassandthethicknessoftheinsulatingglasslayerincreasesevenmore.Ontheonehand,thecontactareashouldbeaslargeaspossible(openedARC);ontheotherhand,theresultinginsulatingglasslayershouldbeasthinaspossible.Inconclusion,agoodcontactresistancecanbeachievedbykeepingtheSiNx-layernon-etchedapartfromthepointswheresilvercrystallitesareformed(seeFigure6a).Inthiscase,theglasslayerisverythinandmainlydeterminedbytheinitialamountofglasswhosecontentshouldbeaslowaspossible.FromourexperimentswecanconcludethatthecurrentbetweensolarcellandmetalgridcanonlyflowviatheAgcrystalliteswhileallothersuggestedcurrentpathsseemnottobesignificant[6,17].

Thus,thethinnertheprintedcontact(lowinkdensity)usingourSISCinkis,thebetterthecontactresistance.Foragoodfrontsidemetallizationtheinkdensityoftheseedlayershouldbeaslowaspossible,justenoughtoformahighdensityofsilvercrystallites,achieveagoodadhesionandagoodabilityforplating.Forinkdensitiesbelow2Á5g/m2,nocontinuouslinecouldbeachievedafterLIP.

Tosummarizetheelectricalpropertiesofatwo-layercontact,boththelineresistivityaswellasthecontactresistanceisbeneficialifkeepingtheseedlayerasthinaspossible.Thecontactformationbetweentheseedlayerandtheemitterofasolarcellismoreefficientwithathinseedlayer.Additionally,thelineresistivityofatwo-layercontactisreducedwithareducedseedlayerheightandmainlydeterminedbytheresistivityoftheLIP-silver(seeLight-InducedPlatingSection).

Figure7.Simulationofthetotallosses(opticalandelectrical)dependentontheamount(height)ofdepositedsilverfordiffer-entlineresistivities.

2.2.Light-inducedplating

Thefirstpartoftheabovesectionillustratestheimportanceofthelineresistivityofplatedsilver.Thetwo-layerconceptisonlyadvantageouscomparedtoasinglelayer(e.g.athickscreenprinted)contact,iftheplatedsilverhasahighconductivity.InFigure7,thetotallosses(electricallyandoptically)aresimulatedindependenceontheplatedsilverheightandwidth,respectively,fordifferentresistivities.Asimulation[17]wasperformedforaninfinitelythinand30mmwidelayerwitharesistivityofrc¼1mVcm2onanemitterwithasheetresistanceofRSH¼60V/sq.Forthewidthofthetotalcontactagrowthmechanismof1:2(height:width)wasassumed.Thesimulatedgridhas80fingersand3busbars(1Á5mmeach)ona15Á6cmÂ15Á6cmlargesolarcell.

Atsmallplatingheights,thelossptotisdominatedbyelectricallossesandstronglydependentontheamountofdepositedsilver.Atlargeplatingheightsthelossmechanismisdominatedbyopticallosses,duetoanincreasedfingerwidth.Thetotallosscurveisnotsymmetric,i.e.beyondtheoptimalplatingheightthe

244

dependenceofthetotallossishigheratsmallerfingerheights.Thus,concerningthetotalcellloss,itismorefavourabletoplatetoomuchsilverthantoolittle.Thesolarcellefficiencyisverystableoverabroadrangeofplatedsilvermass(seebelow).Ahighconductivityofthesilverisbeneficialintwoways:Thetotallosseswerediminishedandtheamountofrequiredsilverissmallerandtherequiredprocesstimeforplatingisshorter.Forexample,ifthelineresistivityisreducedfromrf¼3Â10À8to1Á9Â10À8Vmtheoptimumlineheightcanbereducedfrom14to11mm(21%)thelinewidthby10%andthecross-sectionfrom730to520mm2.Asthecross-sectionisnearlydirectlyproportionaltotheamountofplatedsilver,28%lesssilverisconsumed.

ThelineresistivityofAPandsubsequentLIPfingersisshowninFigure8.TheAPseedlayeristhinenough(<1mm)anditscontributiontotheresistivityisnegligibleastheplatedsilverhasaheightof15mm.ThegrowthmechanismandthemorphologyoftheLIP-silverdependamongotherthingsontheusedelectrolytes.Three

Figure8.Lineresistivityforthreedifferentplatingsolutions,beforeandafteranannealingstepof10minatT¼3508C.

Prog.Photovolt:Res.Appl.2010;18:240–248ß2010JohnWiley&Sons,Ltd.

DOI:10.1002/pip.938

M.Ho¨rteisetal.Finelineprintedandlipcontacts

differentplatingsolutionsweretested,oneCNandtwoNCNelectrolytes.Theplatingparameterslikelightintensity,appliedpotential,pHvalueandtemperaturewerekeptinanoptimizedrangeforeachelectrolyte.ThestronginfluenceoftheusedelectrolyteonthelineresistivitycanbeexplainedbySEMimages(seeFigures9and10).Alsotheinfluenceofanannealingstepbecomesclear.Both,themacroscopicshapeoftheplatedfingerandthemicrocrystallinityareinfluencedbythedifferentelectrolytes.ThesilverlayerdepositedfromtheNCN1electrolyteshowsabotryoidallymorphology.Theseveralclusterscanreachthevalueoftheheightofthefinger,thus,morethan10mmindiameter.Theclusters,althoughvery

Figure10.SEM-closeupviewofthedifferentplatedfingers:(a)non-cyanidicsolutionNCN1,(b)non-cyanidicsolutionNCN2,(c)cyanidicsolutionCN–thesingleAggrains,grownduringthe

platingprocessarevisible.

Figure9.SEMimageoftheplated,conductivelayer.(a)Foranon-cyanidic(NCN1)platingsolution.Noticeablearethelargeclusterboundariesandthebotryoidalgrowth.(b)Foranon-cyanidic(NCN2)platingsolutionand(c)foracyanidic(CN)platingsolutionthesurfacesaremuchsmootherandtheindividual

grainsarelessdistinct.

finegrained,canbedistinguished.Astheresistivityisdeterminedbythesmallestcross-section,thecurrentflowismainlylimitedbythesedeepincisions.Forthemeasurementoftheresistivitywedeterminedanaveragecross-sectionandthebotryoidalstructureislevelled.Supposedthattheresistivityofasingleclusterishighandintherangeofbulksilver,duetotheroughmorphologyandtheclusterboundaries,thelineresistivityislimited,evenifthecontactlossesacrosstheclusterboundariescanbereducedbyanannealing.Thus,alargefractionofplatedsilverdoesnotcontributetothecurrentflow.

IncaseoftheNCN2solution,theseveralAgclustersarepyramidalortrigonaltoacertainextent.Theyaremuchsmaller,below1mmanddeepincisionswhichwouldreducethecross-sectionandthemeasuredresistivityare

245

Prog.Photovolt:Res.Appl.2010;18:240–248ß2010JohnWiley&Sons,Ltd.DOI:10.1002/pip.938

FinelineprintedandlipcontactsM.Ho¨rteisetal.

notvisible.TheLIP-silvergrowshomogeneouslyatthe

surfaceattheratioof1:2(height:width)resultinginaroundish-shapedcontact.ThesurfaceroughnessisdeterminedbythesmallAgclustersandalmostalltheplatedsilvercontributestothecurrentflow.Themeasuredresistivityofrl¼2Á2Â10À8Vmcanonlyslightlybeimprovedbyanannealingstep,indicatingthedensesilverdepositionandtheexcellentconductivitybetweenthesingleAggrains.

ThebestvaluesforlineresistivitycanbeachievedwiththeCNelectrolyte.Themeasuredaveragevalueofrl¼1Á9Â10À8Vmisclosetotheresistivityofbulksilver(rl¼1Á59Â10À8Vm).Thesurfaceissmooth,i.e.noAgclusterscanbedistinguishedanymore.Possibly,thestructurerevealssingleAggrainsinaSEMimageofapolishedcross-section.Thegrainboundariesbetweenthegrainsarehardlyvisible,andaninfluenceofanannealingstepontheresistivityisbarelymeasurable.However,furtherinvestigationshavetobeperformedtorevealthecorrelationbetweenlayerstructure(ondifferentscales)andelectricalproperties.

Inconclusion,theCNplatingsolutionisthebestchoiceintermsofelectricalandopticalproperties.AlsotheprocessstabilityismuchhigherforCNelectrolytes.However,thehealthyrisksinanindustrialproductionhavetobetakenintoaccount.AgoodalternativetotheCNelectrolyteistheplatingsolutionNCN2whichresultsinalineresistivityclosetotheoneofCNplating.

mc-solarcells(15Á6cmÂ15Á6cm)withaseedlayergridwith80fingersand2busbars,APandfiredwereusedtoperformanexperimentconcerningthecellfillfactor(FF)aswellastheseriesresistance(Rs)independenceonthemassofLIP-silver(seeFigure11).ExaminedweretheLIPsolutionsCNandNCN1,respectively.AcontinuousincreaseoftheFFwiththemassofplatedsilverwasobservedtogetherwithasteadydecreaseofRs.IncaseoftheCNsolutiontheFFreacheditssaturationatasilvermassofapproximately100mgperwaferwhereassimilarFFswerereachedfortheNCN1solutionatabout130mg.

Figure12.CurrentdensityjscandcellefficiencyhasafunctionofLIP-silvermassandcontactheightfromtheconstantdecrease

injsc,aneffectivecontactwidthiscalculated.

ThelineresistanceiscontinuouslyreducedwiththequantityofdepositedLIPsilverandboth,thevaluesforRsandfortheFFareimproved.ExceedingaLIP-silvermassofabout200mganimprovementforbothelectrolyteswashardlymeasurableandtheFFswereconstantinthetestedrange.

Theshort-circuitcurrentdensity(jsc)isquitesensitivetothemassofplatedsilveranddecreaseslinearlywiththedepositedsilveramount(seeFigure12).ForbothplatingsolutionsaslopeofaboutÀ1Á6mAcmÀ2/gwasobserved.Fromthemeasuredcurrentdensities,jsc_ini(aftershortLIP)andjsc_LIP(afterLIP)aneffectivefingerwidth(weff)canbecalculated.Iftheareaoftheseedlayermetallizationisknown(Aactive_ini),theinitialnon-metalizedarea(Aactive_cell)isknownaswell.Withincreasingamountofplatedsilver,thenon-metalizedareaandalsothevalueofjscisreduced.Theaccordingopticallyeffectivemetalizedareafortheplatedfingers(Afi_eff)forallvaluesofjsccanbedeterminedbysubtractingthenon-metalizedareatogetherwiththeareaofthebusbars(ABB)fromthetotalcellarea(Atot_cell).Finally,theeffectivewidthofasinglefingercanbedeterminedaccordingtoEq.2.Itiswellknownthattheopticaleffectivewidthofacontactfingercanbeconsiderablysmallerthanitsgeometricalwidth[18].

weff¼

Afieffi;whereAfi

lfiÂnfi

cell

eff¼Atot¼

ÀAactive

cell

ÀABBandAactive

cell

jsciniÂAactive

jscLIPini(2)

Figure11.Fillfactor(FF)andseriesresistance(Rs)asafunctionoftheamountofplatedsilverforboth,NCN1andCNplating

solutions.

Equation2:Calculationoftheeffectivecontactwidth,determinedfromthejscvalues.

InFigure12thecalculatedopticallyeffectivefingerwidthfortheNCN1-LIPisshown.Therealfingerwidthwf,determinedfromamicroscopicpicture,foracontactplatedwith360mgNCN1-LIPisaboutwf¼100mm.Itseffectivewidth,determinedformthecurrentdensity,isaboutweff¼87mm.Thisreductionwillbeevenhigherifthecells

246

Prog.Photovolt:Res.Appl.2010;18:240–248ß2010JohnWiley&Sons,Ltd.

DOI:10.1002/pip.938

M.Ho¨rteisetal.Finelineprintedandlipcontacts

areencapsulatedintoamoduleduetotheinternalreflectionattheglass–airinterface[19].

TheeffectivewidthforCN-LIPcontactisevensmallerandtheaveragejscvaluesfortheCN-LIPareabout0Á2mA/cm2abovetheaveragecurrentdensityoftheNCN1-LIPeveniftheyarebuiltupwiththesameamountofLIP-silver.The0Á2mAhigherjscvaluesincaseofCN-LIPcanbetranslatedintoareductionineffectivecontactwidthby10mmwhichgivesintotalanabsoluteincreaseinthecellefficiencyofDh¼0Á1%.

Thecellefficiencyreachesitsmaximumatabout200mgforbothsolutions.Passingthispoint,thegaininefficiencyduetoanincreasingFFiscompensatedbythesteadydecreaseinjsc.ThecellefficiencyisconstantoverabroadrangeandstartstodecreaseduetothegrowingshadinglossesatLIPquantitiesof350mgforCN-LIPand450mgforNCN1-LIP,respectively.Asexpectedfromthesimulation,thecellefficiencyisquiteinvariantwithrespecttohighquantitiesofLIP-silver.

ACKNOWLEDGEMENTS

TheauthorsgratefullywouldliketothankallmembersfromISEwhocontributedtothisworkespeciallyDanielSchmidtforsolarcellprinting,JohannesSpannagelforresistanceandTLMmeasurements,DamianPyschforhis

¨fferforsupportingridsimulationandElisabethScha

measuringthecellparameters.

REFERENCES

1.GlunzSW,AlemanM,BartschJ,BayN,BayerK,BerganderR.Progressinadvancedmetallizationtech-nologyatFraunhoferISE.Proceedingsofthe33rdIEEEPhotovoltaicSpecialistsConference,SanDiego,USA,2008.

2.GlunzSW,AlemanM,BartschJ,BayN,BayerK,

¨rteisBerganderR,FilipovicA,GreilS,GroheA,Ho

¨M,MetteA,PyschD,RadtkeV,M,KnorzA,Menko

RichterP,RudolphD,RublackT,SchetterC,SchmidtD,SchultzO,WoehlR.Newconceptsforthefrontsidemetallizationofsiliconsolarcells.Proceedingsofthe21stEUPVSEC,Dresden,Germany,2006.

3.SchubertG.ThickFilmMetallisationofCrystallineSiliconSolarCells.UniversitatKonstanz:Kon-stanz,2006;142.

4.BallifC,HuijicDM,Hessler-WyserA,WillekeG.NatureoftheAg–Siinterfaceinscreen-printedcon-tacts:adetailedtransmissionelectronmicroscopystudyofcross-sectionalstructures.Proceedingsofthe29thIEEEPVSC,NewOrleans,Louisiana,USA,2002.

5.HilaliMM,ToB,RohatgiA.Areviewandunder-standingofscreen-printedcontactsandselective-emit-terformation.Proceedingsofthe14thWorkshoponCrystallineSiliconSolarCellsandModulesNREL,WinterPark,Colorado,USA,2004.

6.HuljicDM,BallifC,Hessler-WyserA,WillekeG.MicrostructuralanalysesofAgthick-filmcontactsonn-typesiliconemitters.Proceedingsofthe3rdWCPEC,Osaka,Japan:WCPEC-3OrganizingCom-mittee,December2003.

¨rteisM,GlunzSW.Metal7.MetteA,RichterPL,Ho

aerosoljetprintingforsolarcellmetallization.Pro-gressinPhotovoltaics2007;115:621–627.

8.MetteA,SchetterC,WissenD,LustS,GlunzSW,WillekeG.Increasingtheefficiencyofscreen-printedsiliconsolarcellsbylight-inducedsilverplating.Proceedingsofthe4thWCPEC,Waikoloa,Hawaii,USA,2006.

¨rteisM,GlunzSW.Determinationofthe9.WoehlR,Ho

effectiveopticalwidthofscreen-printedandaerosol-printedandplatedfingers.Proceedingsofthe23rdEUPVSEC,Valencia,Spain,2008.

¨rteisM,GlunzSW.Finelineprintedsiliconsolar10.Ho

cellsexceeding20%efficiency.ProgressinPhoto-voltaics2008;16:555–560.

11.WitvrouwaA,DuBoisaB,DeMooraP,VerbistaA,

VanHoofaC,BenderaH,BaertaK.Acomparison

247

3.SUMMARY

Theelectricalperformanceoffine-lineprintedseedlayers,subsequentlyplatedwerestudiedindetailontestsamplesandonsolarcells.Regardingtheelectricalpropertiesofatwo-layercontactitisbeneficialtoboth,thelineresistivityandthecontactresistancetokeeptheseedlayerasthinaspossible.Thefractionoftheseedlayerofthetotalcontactcross-sectionneedstobeminimizedwhilethefractionoftheLIP-silvershouldbemaximizedtoachievelowestvaluesforthelineresistivityofaboutrf¼1Á9Â10À8Vm.Additionally,theinfluenceoftheheightoftheprintedseedlayeronthecontactresistancewasinvestigated.Thecontactformationbetweentheseedlayerandtheemitterofasolarcellismoreefficientforathin(h<2mm)thanforathick(h¼2–9mm)seedlayer.InSEMimagestheinfluenceoftheseedlayerthicknessonthecontactinterfacewasinvestigated.Foralow-contactresistanceitisimportanttoetchaslittlematerialfromthewafer(SiNx-layer,Si)tokeeptheresultingglasslayerasthinaspossible.Foragoodfrontsidemetallizationtheprintedlineheightoftheseedlayershouldbeaslowaspossible,justenoughtoformalow-ohmiccontact,toachieveagoodadhesionandtoprovideagoodabilityforplating.AsthespecificlineresistivityofthetotalcontactisdeterminedbythequalityofthedepositedLIP-silver,differentplatingsolutionswereinvestigatedbyanalysingtheplatedcontacts.Differentlineresistivities,fromclosetobulksilver(CN-LIP)uptoresistivitiescomparabletoscreenprintedcontacts(NCN1-LIP)werefound.TheoptimummassofLIP-silvernecessarytometallizeasolarcellisdeterminedtheoreticallyandexperimentally.Alineresistivityofrf¼1Á9Â10À8Vmcomparedtorf¼3Á0Â10À8Vmsavesalmost30%ofsilver.Inaddition,measuringthecurrentofsolarcellsasafunctionofplatedsilver,theeffectiveopticalfingerwidthisdeterminedtobeabout15%smallerthanthegeometricalcontactwidth.

Prog.Photovolt:Res.Appl.2010;18:240–248ß2010JohnWiley&Sons,Ltd.DOI:10.1002/pip.938

FinelineprintedandlipcontactsM.Ho¨rteisetal.

12.

13.

14.

15.

betweenwetHFetchingandvaporHFetchingfor

sacrificialoxideremoval.SPIEProc.MicromachiningandMicrofabricationProcessTechnology,SantaClara,2000.

¨rteisM,MetteA,RichterPL,FidorraF,GlunzSW.Ho

Furtherprogressinmetalaerosoljetprintingforfrontsidemetallizationofsiliconsolarcells.Proceedingsofthe22ndEUPVSEC,Milan,Italy,2007.

MeierDL,SchroderDK.Contactresistance:itsmeasurementandrelativeimportancetopowerlossinasolarcell.IEEETransactionsonElectronDevices1984;ED-31:7–653.

SchroderDK,MeierDL.Solarcellcontactresistance–areview.IEEETransactionsonElectronDevices1984;ED-31:637–7.

PyschD,MetteA,FilipovicA,GlunzSW.Compre-hensiveanalysisofadvancedsolarcellcontactscon-sistingofprintedfine-lineseedlayersthickenedby

16.

17.

18.

19.

silverplating.ProgressinPhotovoltaics2008;17:101–114.

¨rteisM,GlunzSW.Peakfiringtempera-GruppG,Ho

turedependenceofthemicrostructureofAgthick-filmcontactsonsiliconsolarcells–adetailedAFMstudyoftheinterface.Proceedingsofthe20thECPVSEC,Barcelona,Spain,2005.

MetteA.Newconceptsforfrontsidemetallizationof

¨r¨tfuindustrialsiliconsolarcells.DissertationFakulta

¨tFreiburg,AngewandteWissenschaften,Universita

2007;231.

BlakersAW.Shadinglossesofsolar-cellmetalgrids.JournalofAppliedPhysics1992;71:5237–5241.

¨rteisM,GlunzSW.AnalysisoftheWoehlR,Ho

opticalpropertiesofscreen-printedandaerosol-printedandplatedfingersofsiliconsolarcells.AdvancesinOptoElectronics2008,2008;1–7.

248

Prog.Photovolt:Res.Appl.2010;18:240–248ß2010JohnWiley&Sons,Ltd.

DOI:10.1002/pip.938