Volume2008,ArticleID485467,10pagesdoi:10.1155/2008/485467
ResearchArticle
PECVD-ONO:ANewDepositedFiringStableRearSurfacePassivationLayerSystemforCrystallineSiliconSolarCells
M.Hofmann,1S.Kambor,1C.Schmidt,1D.Grambole,2J.Rentsch,1S.W.Glunz,1andR.Preu1
1Fraunhofer2Forschungszentrum
InstituteforSolarEnergySystems,Heidenhofstrasse2,79110Freiburg,Germany
Dresden-Rossendorf,BautznerLandstrasse128,01328Dresden,Germany
CorrespondenceshouldbeaddressedtoM.Hofmann,marc.hofmann@ise.fraunhofer.deReceived17January2008;Accepted6March2008RecommendedbyArminAberle
Anovelplasma-enhancedchemicalvapourdeposited(PECVD)stacklayersystemconsistingofa-SiOx:H,a-SiNx:H,anda-SiOx:Hispresentedforsiliconsolarcellrearsidepassivation.Surfacerecombinationvelocitiesbelow60cm/s(afterfiring)andbelow30cm/s(afterforminggasanneal)wereachieved.Solarcellprecursorswithoutfrontandrearmetallisationshowedimpliedopen-circuitvoltagesVocvaluesextractedfromquasi-steady-statephotoconductance(QSSPC)measurementsabove680mV.Fullyfinishedsolarcellswithupto20.0%energyconversionefficiencyarepresented.Afitofthecell’sinternalquantumefficiencyusingsoftwaretoolPC1Dandacomparisontoafull-areaaluminium-backsurfacefield(Al-BSF)andthermalSiO2isshown.PECVD-ONOwasfoundtobeclearlysuperiortoAl-BSF.Aseparationofrecombinationatthemetallisedandthepassivatedareaatthesolarcell’srearispresentedusingtheequationsofFischerandKray.Nuclearreactionanalysis(NRA)hasbeenusedtoevaluatethehydrogendepthprofileofthepassivationlayersystematdifferentstages.
Copyright©2008M.Hofmannetal.ThisisanopenaccessarticledistributedundertheCreativeCommonsAttributionLicense,whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited.
1.Introduction
Electricalrearsurfacepassivationisbecomingincreasinglyimportantincrystallinesiliconsolarcelltechnology.Themainreasonforthatistheneedforacostdecreaseinthephotovoltaicsector.Thisleadstowafersthatarebecomingthinnerandthinnerandasteadilyincreasingneedforhigherenergyconversionefficiencies.Theresultofbothtrendsisanincreasedimportanceofaneffectivelypassivatedrearsurface.InFigures1to3,simulationsonrearsurfacepassivationandcellthicknessarepresentedusingPC1D[1].Open-circuitvoltageVocandenergyconversionefficiencyηdependstronglyontherearsurfacerecombinationvelocity(SRV)Seffwhenhigh-qualitysiliconsubstratesaretakenintoaccount(τbulk=750μs,excellentoptics,frontemitter,andpassivation,Figure1).ForSeffvaluesintherangeof5∗101cm/sto5∗103cm/s,thestrongestinterdependencewasfound.Thethinnerthesolarcellsthestrongerthiseffectappears.Notethatverythinsolarcells(50μm)showahigherVoclevelatverylowSeffvalueswhilethe50μmcellssufferfromadecreasedlighttrappinginoursimulationand
convertthelightalittlelessefficientlytoelectricityatthemaximumpowerpoint.Whenlookingatmedium-efficientsolarcells(τbulk=100μs,excellentoptics,frontemitter,andpassivation,Figure2)theinfluenceoftherearpassivationincreaseswhenthecellsaregettingthinner.Wefoundthatthethinnestcellsshowevenbetterefficienciesthanthickercellswhenanexcellentrearpassivationisapplied.Low-efficiencycells(τbulk=10μs,excellentoptics,standardfrontemitter50Ω/sq,andlow-levelfrontpassivation,Figure3)showonlyaslightdependenceontherearpassivationduetoalowdiffusionlength(τbulk=10μs→approximatelyLeff=170μm).Hence,mostminoritycarriersdonotprofitfromalowrearrecombinationvelocitybutrecombineinthebulksilicon.Thethinnerthelow-efficientcellsbecome,themorepronouncedwillbetheinfluenceofagoodrearpassivation.
Agostinellietal.[2]haveshownthatsiliconsolarcellsonverythinsubstrates(downto105μm)canreachhigherefficienciescomparedtoastandardAl-BSFrearstructurewhenanadaptedrearsurfacepassivationisapplied.Kray[3]haspresentedenergyconversionefficienciesabove20%for
2
0.72Highefficiency230.7Simulatedopen-circuitvoltageVoc(V)22Simulatedefficiencyη(%)0.6824AdvancesinOptoElectronics
Highefficiency210.6620190.64180.62170.6100101102103104105106RearsurfacerecombinationvelocitySeff(cm/s)
300μm250μm200μm
(a)
150μm100μm50μm
16100101102103104105106RearsurfacerecombinationvelocitySeff(cm/s)
300μm250μm200μm
(b)
150μm100μm50μm
Figure1:Simulatedopen-circuitvoltageVoc(left)andsimulatedenergyconversionefficiencyηversustheeffectiverearsurfacerecombinationvelocitySeffforhigh-efficiencysolarcells.CalculatedusingPC1D.Parameters:NA=3∗1016cm−3,τbulk=750μs(approx.Augerlimit),texturedfront,frontemitter:120Ω/sq,Sfront=1000cm/s.
substrateswithlessthan50μmthickness.Bothpublicationsmakeuseofthepassivatedemitterandrearcell(PERC)concept[4].
FraunhoferISEhasdevelopedaneconomicallyfeasibletechnologyformassproductionofthelocalrearcontactsofPERCs[4]byimplementingalasertechnology.Thispatentedlaser-firedcontacts(LFCs)approach[5]allowsthelocalaluminiumcontactfiringthroughapassivating—andinmostcasesisolating—layersysteminapproximately1to3secondspercell.Withinthesamestep,aluminiumisalloyedintothesiliconformingap+regimeunderneaththelocalcontactcreatingaverygoodohmiccontactatthecell’srear[6,7].
RearsidesurfacepassivationschemesneedtobeadaptedtotheLFCprocessandtotheindustrialproductionenvi-ronment.Themostcommonlayersarethermallygrownsilicondioxide(SiO2),plasma-enhancedchemicalvapourdeposited(PECVD)hydrogenatedamorphoussiliconnitride(a-SiNx:H,inshort:SiN),andPECVDhydrogenatedamor-phoussilicon(a-Si:H,inshort:a-Si).Asthermallygrownsilicondioxide(SiO2)layersaremanufacturedbyatime-andenergy-intensivehightemperatureprocess,theyarenotthefirstchoiceformassproduction,althoughtheypossiblyprovideaverygoodandthermallystablepassivation[8].Unfortunately,thetemperaturelevelisnotsuitableforeverymaterialquality.Especially,low-costwafersmightdegradeduringanoxidationprocess.
SiNfilmsarefabricatedbyalowtemperatureprocessingstep(approximately300–400◦C).Theyalsoprovideaverygoodpassivationquality[9–15]butarenotasthermallystableasSiO2layers[8].ThethermalstabilityofoptimisedSiNlayerscanbeimprovedtoadegreethatissufficientforthetypicalthermalbudgetofthesolarcellproductionlineexcludingemitterdiffusion.Additionally,thecellspassivatedwithSiNfilmstypicallysufferfromaninversionlayerwhichisshuntedbythelocalrearcontacts[16,17].TheinversionlayerisduetoahighdensityoffixedpositivechargeswhichtypicallycanbefoundinSiNfilms.
Gooda-Sifilmsprovideaveryefficientsurfacepassiva-tionwithoutformingsuchaninversionlayerasa-SifilmsdonotmainlypassivateviaafieldeffectbutthroughtheloweringofthedensityofstatesattheinterfaceDit.Differentgroupshavebeenusinga-Siaspassivation,eitherassinglelayera-Sipassivation[18],a-Sistacks[19,20]ora-SistackedwithSiN[21,22],orPECVDsiliconoxide(SiOx)[23].Notonlyintrinsicbutalsodopeda-Silayershavebeenused.Themajordrawbackwhenusinga-Sipassivationlayersincrystallinesiliconsolarcelltechnologyistherelativelylow
AdvancesinOptoElectronics
0.7Mediumefficiency0.68Simulatedopen-circuitvoltageVoc(V)2122Mediumefficiency3
200.66Simulatedefficiencyη(%)150μm100μm50μm
(a)
190.64180.62170.6160.58150.56100101102103104105106RearsurfacerecombinationvelocitySeff(cm/s)
300μm250μm200μm
14100101102103104105106RearsurfacerecombinationvelocitySeff(cm/s)
300μm250μm200μm
(b)
150μm100μm50μm
Figure2:Simulatedopen-circuitvoltageVoc(left)andsimulatedenergyconversionefficiencyηversustheeffectiverearsurfacerecombinationvelocitySeffformedium-efficiencysolarcells.CalculatedusingPC1D.Parameters:NA=7∗1015cm−3,τbulk=100μs,texturedfront,frontemitter:120Ω/sq,Sfront=1000cm/s.
thermalstabilityofthepassivationproperties.Afteratypicalscreenprintedcontactfiringstepwithtemperaturesofupto∼850◦Cthepassivationpropertiesarestronglyreduced[24].Therefore,analternativepassivationlayersystemthat(i)canbedepositedquicklyandeasilyatlowtempera-tures,preferablyusingPECVD,
(ii)providesgoodpassivationwithoutasignificantshun-ting,
(iii)isthermallystableinacontactfiringprocessat
∼850◦C,
(iv)providesacompatibleinternalrearsurfacereflec-tancewouldbeaverygoodwaytoincorporateaneffectivesurfacepassivationlayersystemintoday’sindustrialsolarcells.
2.LifetimeInvestigation2.1.SymmetricSampleStructure
Asdescribedabove,thermallygrownsilicondioxidelayersprovideallmajorrequirementstobesuitablefortherearpassivationforsiliconsolarcells:goodsurfacepassivation,
goodrearreflectance,andthermalstabilityinsolarcellproductionprocesses.Onlytheprocessthroughputandcostandthepossibledegradationofthesolarcellprecursorsseemtobethemajorandcrucialproblem.
PECVDsiliconoxidelayers(SiOx)looklikeapossiblealternativetocombinethegoodlayercharacteristicsofanSiO2filmwiththebenefitsofPECVD.Regrettably,noPECVDSiOxlayerscouldbefoundbytheauthorsthatshowagoodsurfacepassivationandthermalstability,nomatterifthelayerhasbeenannealedindifferentgasambient(e.g.,forminggas)ornot.Hence,differentstacklayershavebeenunderinvestigationwiththePECVDSiOxlayerbeingthefirstlayerdeposited(sittingdirectlyonthewafer’ssurface).ItshouldbenotedthataninvestigationofdepositedoxidelayersstackedwithPECVDSiNxhasbeenpresentedbyAgostinellietal.[25].
Hoexetal.havepresentedwell-passivatingSiOxfilmsdepositedbytheexpandingthermalplasma(ETP)technique[26].
Thelifetimeinvestigationhasbeenperformedusingsiliconwafersassubstrateswiththefollowingproperties:floatzone,p-type,borondoped,1Ωcm,250μmthick,andshinyetchedsurfaceswithacrystalorientationof(100)[27].ThewaferswerecleanedwithawetchemicalRCAclean
4
0.62Lowefficiency16AdvancesinOptoElectronics
Lowefficiency15Simulatedopen-circuitvoltageVoc(V)0.6Simulatedefficiencyη(%)150μm100μm50μm
(a)
140.5813120.56100101102103104105106RearsurfacerecombinationvelocitySeff(cm/s)
300μm250μm200μm
11100101102103104105106RearsurfacerecombinationvelocitySeff(cm/s)
300μm250μm200μm
(b)
150μm100μm50μm
Figure3:Simulatedopen-circuitvoltageVoc(left)andsimulatedenergyconversionefficiencyηversustheeffectiverearsurfacerecombinationvelocitySeffforlow-efficiencysolarcells.CalculatedusingPC1D.Parameters:NA=7∗1015cm−3,τbulk=10μs,texturedfront,frontemitter:50Ω/sq,Sfront=106cm/s.
WetchemicalRCAclean
PECVDofSiOx+SiNx
PECVDofSiOx+SiNx+SiOx
Carrierlifetimemeasurement
Annealing425◦C,15minFiring850◦C,∼3s
Carrierlifetimemeasurement
Figure4:Workflowofthelifetimeexperimentonsymmetricsamplestructures.
platePECVDreactorusingaplasmaexcitationfrequencyof13.56MHzandagasmixtureofmonosilane(SiH4)andnitrousoxide(N2O)fortheSiOxlayerandmonosilane,hydrogen(H2)andnitrogen(N2)fortheSiNxlayer.Thedepositiontemperatureforalllayerswas350◦C.Theoxidelayerscanbedepositedveryquicklyatadepositionrateof∼120nm/min.Thenitridelayertakeslonger,atarateof∼8nm/min.Thicknessesof120nmfortheoxideand70nmforthenitridelayershavebeenapplied.SeeFigure4fortheprocesssequenceandFigure5forthesamplestructures.Minoritycarrierlifetimesweremeasuredusingthequasi-steady-statephotoconductance(QSSPC)method[29].ThesurfacerecombinationvelocityhasbeencalculatedfromthemeasuredlifetimesassumingthatonlyAugerrecombinationtookplaceinthesample’sbulk(i.e.,“perfectbulk”).TheAugermodelbyGlunzetal.[30]hasbeenappliedforthecalculationusingthefollowingequation[31]:
1W1W1
=++τeffτb2SeffDnπ
2−1
bathsequence[28]includingafinaldipinhydrofluoric
acid(HF)andsubsequently,thedepositiononbothsurfacesofthewafershasbeenperformed.TwodifferentPECVDstacklayersystemshavebeeninvestigated:(i)SiOxandSiNxand(ii)SiOx,SiNx,andSiOx.SiOxalwayshasbeenthefirstlayerdeposited.Thedepositiontookplaceinaparallel
,(1)
withthesiliconwafer’sbulklifetimeτb,thewaferthicknessW,theeffectivesurfacerecombinationvelocity(SRV)Seff,andthediffusionconstantforelectronsDn.
AdvancesinOptoElectronics
SiOxSiNxSiOxFZ-c-SiSiOxSiNxSiOx
5
SiNxSiOxFZ-c-SiSiOxSiNx
Figure5:Symmetricallifetimesamplestructures.Left:doublelayersamplestructureSiOx+SiNx.Right:triplelayersamplestructureSiOx,SiNxandSiOx(PECVD-ONO).
3−1000
mnco41it0a1ni∗bm5o=100
cnreΔce@af)rsu/smcve(10
itffceSffeyEticolve1
As-depositedAfterannealingAfterfiring425◦C850◦CSiO.SiNSiO.SiN.SiOFigure6:EffectivesurfacerecombinationvelocitiesofPECVDstacksofSiOx+SiNx(leftredcolumns)andSiOx,SiNxandSiOx(rightgreenstripedcolumns,PECVD-ONO)as-deposited,afterannealing(425◦C,15min,forminggas)orfiring(850◦C,∼3s).
ThestacksystemofPECVDSiOxandSiNxledtoas-depositedSeffvaluesbelow700cm/s.Subsequently,anannealinginforminggasat425◦Cfor15minuteshasbeenperformed.MinoritycarrierlifetimemeasurementsatthisstageshowedSeffvaluesbelow50cm/s.Next,athermaltreatmentcomparabletofiringofscreenprintedfrontcontactswithapeakwafertemperatureofapproximately850◦Cforabout3secondswasdonewithSeffincreasingto<70cm/s.
Forthetriplelayersystem(SiOx,SiNx,andSiOx,inshort:PECVD-ONO),thesamethermalprocessingasforthetwolayersamplestructurehasbeenapplied.Thisledtothefollowingsurfacerecombinationvelocities:as-deposited:below240cm/s,afterannealinginforminggasat425◦Cfor15minutes:below30cm/s,afterthermalprocessingcomparabletofiringofscreenprintedfrontcontactswithapeakwafertemperatureofapproximately850◦Cforabout3seconds:below60cm/s.
TheSeffresultsaresummarisedinFigure6.
Bothstacksystemsaresuitableforthesurfacepassivationofp-typeSiwafersbeforeandafterathermaltreatment
SiO2n-Sip-Si
SiOSiNxSiOxx
Figure7:Samplestructureofthesolarcellprecursors.
likethecontactfiring.Afterdepositionandafterannealing,astrongerpassivationeffectwasfoundforthetriplestackstructureSiOx+SiNx+SiOx(PECVD-ONO,Patentpend-ing).Thiseffectwillbeinvestigatedfurtherinthefuture.However,onebeneficialeffectistheincreasedoveralllayerthicknessprovidingahigherprobabilityforhydrogentodiffusetothesilicon/siliconoxideinterfaceandloweringtheamountofrecombinativelyactiveinterfacetraps.
2.2.SolarCellPrecursors
AninvestigationofthesuitabilityofthenewlydevelopedPECVD-ONOstacksystemforpassivationofsolarcellrearsidesispresentedhere.Ashortsummaryhasbeenshownin[27].
Floatzonewaferswiththefollowingcharacteristicswereused:p-type,borondoped,0.5Ωcm,250μmthick,andshinyetchedsurfacewithacrystalorientationof(100).Thewaferswereoxidised,frontsidewindowswereopenedbyphotolithography.Then,anemitterwasdiffusedintothefront(120Ω/sq.)followedbyawetchemicaletchofthephosphoroussilicateglass(PSG).Inasecondoxidationstep,anantireflectingSiO2layerwasgrownonthefront(andtheoxideonthebackwasthickened).Afterremovingtheoxideontherearandcleaningthesurfaceinawetchemicalbathsequenceof(i)HNO3and(ii)HF,thenovelpassivationstackPECVD-ONOhasbeendepositedontherearandthelifetimewasmeasured.Subsequently,thesampleswereannealedat425◦Cfor15minutesinforminggasand/orathermaltreatmentcomparabletothefiringofscreenprintedfrontcontactswasperformed.Aftereachthermalprocess,thelifetimewasmeasuredusingtheQSSPCtechnique.ThesamplestructurecanbefoundinFigure7andtheprocesssequenceinFigure8.
6OxidationOpeningfrontwindowsEmitterdiffusion120Ω/sq
PSGetchOxidationARCRemovalrearoxideDepositionrearpassivationCarrierlifetimemeasurementAnnealing“Contactfiring”425◦C,15min∼850◦C,∼3sCarriermeasurement
lifetimeCarriermeasurement
lifetimeFigure8:Processsequenceofthesolarcellprecursorexperiment.ThecolourscorrespondtothecoloursinFigure9.
Itcanbesummarisedthatthesamplestructureequalsafinishedsolarcellexceptthemetalcontactsonfrontandrear.CuevasandSintonhaveshownthatimpliedVocvaluescanbeextractedfromthemeasuredeffectiveminoritycarrierlifetimes[32,33]usinganapproximationequationforVoc:
Voc=kT
qlnnpn2,(2)
i
withBoltzmann’sconstantk,temperatureT,elementary
chargeq,free-electrondensityn,free-holedensityp,andintrinsicfree-electrondensityni.
Forp-typewafersandlongdiffusionlengths,(2)canbewrittenas[32,33]
Voc=kT
qln
Δn(Δp+NA)n2,(3)
i
withNAbeingthedensityofacceptors.
nicanbecalculatedfrom
ni=NCNVe−Eg/kT,
(4)
withsilicon’sdensityofstatesintheconductionbandNC
andinthevalencebandNVandthesiliconbandgapenergyEg.Hence,onecancalculateImpliedVocvaluesfromQSSPCmeasurements.TheimpliedVocwasextractedat1sun.Initially,Vocvaluesofapproximately631mVhavebeenobserved.Afterannealingorfiring,theimpliedVocvaluesincreasedstronglytovaluesabove680mV.Here,theresultsafterfiringwereverysimilarregardlessofwhetheranannealingwasdonepriortothefiringstepornot.Theeffectissuspectedtobeduetoanenhancementofthefrontandrearsurfacepassivation.SeeFigure9.
AdvancesinOptoElectronics
)690nus1@680CPS670SQ()V660m(CO650Vdeil640pmI630
Reardeposition
Annealing425◦C
Firing850◦C
Figure9:Solarcellprecursors.ImpliedVocvaluesextractedfromQSSPCmeasurementsat1sunshowastrongincreaseafterthermaltreatment.ColourscorrespondtoFigure8.
3.HydrogenDepthProfiling
Forthepassivationeffect,thediffusionofhydrogenfromthestronglyhydrogenatedsiliconnitridelayertotheSiOx/bulkSiinterfacemightberesponsible.Typically,ahighdensityofrecombinativelyactivestatesatthisinterfacecanbefound.Thesestates,mostlySidanglingbonds,canbesaturatedbyhydrogenandimplicitlymaderecombinativelyinactive.Therefore,thehydrogenconcentrationandthechangeinhydrogenconcentrationindifferentlayerswithinthepassivationstacksystemisofgreatinterest.
Nuclearreactionanalysis(NRA)measurementshavebeenperformedatForschungszentrumDresden-Rossendorf,Germany[34].NRAisbasedonareactionbetweenanitrogenisotope15NwhichisacceleratedtoseveralMeVandahydrogenatom1Hinthesample.Thereactiononlytakesplacewhenthenitrogenatomhasaspecifickineticresonanceenergy(6.385MeV).Atlowerenergy,nonuclearreactionwillhappen.Athigherenergy,thenitrogenatomloosesspeed(energy)whilepassingthroughthefirstatomiclayersofthefilm.Whenthenitrogenisdeceleratedtotheresonanceenergy,thereactionwillhappenifthereisahydrogenatomtoreactwith(Figure10).Thereactionofhydrogenandnitrogenwillfinallyleadtoa12Catomandanalphaparticleandagammaquantumthatareradiated.Thereactionequationsareasfollows:
15N+1H
−→16O∗−→12C∗+α,12C∗
−→12C+γ.
(5)
Countingthemeasuredgammaquantaforacertainnitrogenkineticleadstothehydrogenconcentrationandthedepthprofile[34].
SiliconwafersamplescomparabletothesymmetricallifetimesampleswereusedfortheNRAmeasurement[27].Theprocessstateofthesamplesunderinvestigationwasas-deposited,◦afterannealing(425◦C,15min)andafterfiring(850C,∼3s).Thehydrogendepthprofilingoftheas-depositedsampleshowedahydrogenconcentration
AdvancesinOptoElectronics7
HHKineticenergy∼6.385MeVHHHHKineticenergy>6.385MeV15NHHHH15NγHHHSampleγHHH(a)
Sample(b)
xFigure10:Measurementprincipleofthenuclearreactionanalysis(NRA)technique(after[35]).Theintensityofgammaradiationforaknownkineticenergyof15Nabovethethresholdenergyof6.385MeViscorrelatedwiththedepthofhydrogen.
1816Hconcentration(at%)141210864200SiOxSiNxSiOxSisubstrateEvaporatedTiPdAgfrontcontacts,Agelectroplated
Randompyramids
SiO2
n+emitterp-Si
LocalAl-BSFSiOxSiNxSiOxAlLaser-firedcontacts(LFC)
100200Depth(nm)
300400As-deposited
Annealed425◦C15minFired850◦C3s
Figure12:StructureofthefabricatedsolarcellswiththenewlydevelopedpassivationstacksystemPECVD-ONOontherearandLFCcontacts.
Figure11:ResultsofthehydrogendepthprofilingusingtheNRAmethod.SampleswithPECVD-ONOafteravariationofthermaltreatmentareshown:as-deposited,afterannealing(425◦C,15min),afterfiring(850◦C,∼3s,samplenotannealedpriortofiring).
nitridefilmtothesurroundingoxidelayerstookplace.NosignificanteffectcanbeobservedattheSiOx/c-Siinterface(seeFigure11).
4.SolarCellFabrication
ofapproximately8at%inthePECVDSiOxfilmsandapproximately16at%intheSiNxfilm.Almostnohydrogencouldbefoundwithinthecrystallinesiliconbulk.Afterannealingthehydrogendepthprofiledidnotchangemuch.Hence,nostronghydrogendiffusioncouldbeobserved.Thesamplethatwassubjectedtothe“contactfiring”showedasignificantlymodifiedhydrogenprofile.Asexpectedatthistemperaturerange,hydrogenwasverymobile.Theoverallhydrogenconcentrationwaslowered,withintheSiOxlayersapproximately5at%hydrogencouldbefoundandapeakconcentrationofapproximately8at%wasobservedwithintheSiNxlayer.Thisshowsthatastrongout-diffusion(i)fromthehydrogenatedlayerstotheambientand(ii)fromthe
Finally,solarcellswiththenewlydevelopedpassivationstacksystemPECVD-ONOwerefabricatedtoshowthequalityofthepassivationonafinalcell.
Startingthecellfabricationusingfloat-zoneSisubstrateswiththesamecharacteristicsasintheaboveexperiments,thefinalcellsexhibitevaporatedTiPdAgfrontcontacts,athermallyoxidisedantireflectioncoatingthatalsoservesasfrontpassivationlayer,a120Ω/sqn-typeemitter,fabricatedbydiffusionusingaPOCl3ambient,a1Ωcmp-typebulk,thenewlydevelopedPECVD-ONOstackpassivationsystem,andanevaporatedAllayeratthebackandlaser-firedcontactsthatledtoalocalAl-BSFunderneath(aboveiflookingatFigure12)thepointcontacts.Therearsurfacewas
8
10.9ycn0.8eicffi0.7em0.6utn0.5auql0.4anr0.3tenI0.20.10300400500600700800900100011001200Wavelength(nm)
ThermalSiO2Al-BSF
PECVD-ONOFigure13:InternalquantumefficiencyforcellsofthisbatchwiththermalSiO2andPECVD-ONOrearpassivationandhigh-qualityAl-BSFrearpassivationofanotherbatch.
Table1:Bestsolarcellsofthisinvestigation.
RearpassivationAreaVocJscFFηPECVD-ONO4.0cm2664mV38.2mA/cm278.7%20.0%ThermalSiO24.0cm2676mV38.3mA/cm280.4%20.8%
cleanedinawetchemicalbathsequenceof(i)HNO3and(ii)HF.
ThecellschemecanbefoundinFigure12.
Thefinalcellwasannealedat425◦Cinforminggasfor15minutes.Athermalprocesscomparabletothefiringofscreenprintedfrontcontacts(asusedinthelifetimeexperimentsdescribedabove)wasnotapplied.Thisprocesswouldharmthesolarcellstructureusedinthisbatch.
ThebestsolarcellwiththenovelPECVD-ONOstacksystemasrearpassivationledtoacellefficiencyof20.0%(seeTable1).
Forcomparisonalsothermallygrownsilicondioxidewasusedforrearpassivationwithinthesamesolarcellbatchthatisknownforitsexcellentrearpassivationquality.20.8%peakefficiencycouldbeobtainedforthesereferencecells.
Internalquantumefficiency(IQE)measurements(Figure13)allowforacomparisonofthequalityofthenovelrearpassivationPECVD-ONOstackwiththermallygrownSiO2aswellaswithfullareahigh-qualityAl-BSF(theAl-BSFcellsoriginatedfromadifferentbutsimilarsolarcellbatch).ItcouldbeproventhatPECVD-ONOissuperiortoAl-BSFrearsidesbutshowslowerqualitythanthermallygrownSiO2.
Toquantifytherearpassivationqualityinthefinishedsolarcell,anumericalfittotheexperimentalexternalquantumefficiency(EQE),IQE,andreflectionpropertieswaspreparedusingPC1D[1].ThegraphicalresultscanbefoundinFigure14.ThetotalfrontandrearsurfacerecombinationvelocitiescanbederivedfromthePC1Dfitas3200cm/s±500cm/s(front)and550cm/s±50cm/s
AdvancesinOptoElectronics
1,EQn0.9Ioiyt8cc0.neefli0.7creffi,eE0.6mQEu.t0.5nE.auQql0.4alannr0.3rtetexne0.2I0.10300400500600700800900100011001200Wavelength(nm)
IQEPECVD-ONO
EQEPECVD-ONOPC1DfitIQEPECVD-ONOPC1DfitReflection
EQEPECVD-ONO
Figure14:Externalquantumefficiency,IQEandreflectionvaluesforthesolarcellusingPECVD-ONOrearpassivation.Experimentaldata(symbols)andcalculatedvalues(lines)aredisplayed.ThecalculationwasdoneusingPC1D[1].
(rear),respectively.Thisincludestherecombinationatthepassivatedareaaswellasthecontactarea.FischerpresentedanequationdescribingtheSRVoflocallycontactedsolarcellrearsides[36]
SLeff=
Dπ−1
W2Wpπfarctan2WLp
f−e−W/Lp+DfWSmet
+Spass
1−f
,(6)
withtheminoritycarrierdiffusionconstantD,thewaferthicknessW,thecontactpitchLp,themetallisationfractionf,theSRVofthemetallisedsurfaceSmet,andtheSRVofthepassivatedsurfaceSpassoftherearside.Forthecellspresentedhere,thefollowingvaluesarevalid:DnL=27.1cm2/s,W=240μm,pexperimentally=1investigatedmm,andfand=modelled1%.KraytheandrecombinationGlunzhavevelocityatthelaser-firedcontactsleadingtotheexpression[37]
Smet
NA
=S0+αeβ(NA+N0),
(7)
withS0andN=−900cm/s,α=22.1cm/s,β=1.29×10−16cm3,
0=3the.40substrate×1016cm−3andNAbeingthedensityofacceptorsin(inourcase1.5×1016cm−3).Hence,byusingtheaboveequationsandtheSeffvalueextractedfromPC1D,Spasscanbecalculatedto500cm/s±50cm/s.
5.Conclusion
AnovelsurfacepassivationstackcomprisingofthreePECVDlayerswaspresented.Thestackconsistsofa-SiOx:H,a-SiNx:H,anda-SiOx:H.Lifetimemeasurementswiththe
AdvancesinOptoElectronics
novelstacksystemonbothsurfacesofp-typesiliconwafersshowedsurfacerecombinationvelocitiesof<240cm/s,afterannealingat425◦C:<30cm/s,afterfiring:<60cm/s.SolarcellprecursorswithoutmetallisationbutwithPECVD-ONOrearpassivationexhibitedimpliedVocvaluesextractedfromtheQSSPCmeasurementsof>680mV.HydrogendepthprofilingusingtheNRAmethodwasconductedresultinginhydrogenconcentrationsofapproximately8at%inthePECVDSiOxfilmsandapproximately16at%intheSiNxfilmintheas-depositedandannealedstate.Afterfiring,thehydrogenconcentrationswerestronglyloweredto5at%inthePECVDSiOxfilmsandapproximately8at%intheSiNxfilm.
Solarcellswiththenovelpassivationstackwereprocessedandshowedefficienciesofupto20.0%.IQEcharacterisationandmodellingusingPC1DledtoatotalSRVattherearsurfaceof550cm/s±50cm/s.Additionally,themodelfortherecombinationatlocallycontactedrearsidesbyFischerandKrayledtoanSRVof500cm/s±50cm/satthepassivatedarea.
ApatentonthenovelPECVD-ONOpassivationstacksystemispending.
Acknowledgments
TheauthorswouldliketothankA.Leimenstoll,A.Her-bolzheimer,andS.Seitzforcleanroomprocessing.ThelaserprocessingwassupportedbyJ.F.Nekarda,B.Fleischhauer,C.Harmel,andA.Grohe.Additionally,theauthorsthankK.Kr¨ugerforAgplating,T.Rothforlifetimemeasurements,andE.Sch¨afferandK.KordelosforI-Vmeasurements.ThisworkwaspartlysupportedbytheEUfundedproject“CrystalClear”undertheprojectnumberSES6-CT2003-502583.
References
[1]D.A.ClugstonandP.A.Basore,“PC1Dversion5:32-bit
solarcellmodelingonpersonalcomputers,”inProceedingsofthe26thIEEEPhotovoltaicSpecialistsConference,pp.207–210,Anaheim,Calif,USA,September-October1997.
[2]G.Agostinelli,P.Choulat,H.F.W.Dekkers,S.DeWolf,andG.
Beaucarne,“Screenprintedlargeareacrystallinesiliconsolarcellsonthinsubstrates,”inProceedingsofthe20thEuropeanPhotovoltaicSolarEnergyConferenceandExhibition,pp.647–650,Barcelona,Spain,June2005.
[3]D.Kray,Hocheffizientesolarzellenstrukturenf¨urkristallines
silicium-materialindustriellerqualit¨at,Ph.D.dissertation,Uni-versit¨atKonstanz,Konstanz,Germany,2004.
[4]A.W.Blakers,A.Wang,A.M.Milne,J.Zhao,andM.A.Green,
“22.8%efficientsiliconsolarcell,”AppliedPhysicsLetters,vol.55,no.13,pp.1363–1365,1989.[5]E.Schneiderl¨ochner,R.Preu,R.L¨udemann,andS.W.Glunz,
“Laser-firedrearcontactsforcrystallinesiliconsolarcells,”ProgressinPhotovoltaics:ResearchandApplications,vol.10,no.1,pp.29–34,2002.
[6]S.W.Glunz,E.Schneiderl¨ochner,D.Kray,etal.,“Laser-fired
contactsolarcellsonp-andn-typesubstrates,”inProceedingsofthe19thEuropeanPhotovoltaicSolarEnergyConferenceandExhibition,pp.408–411,Paris,France,June2004.
9
[7]F.Granek,M.Hermle,B.Fleischhauer,etal.,“Optimisation
oflaser-firedaluminiumemittersforhighefficiencyn-typesisolarcells,”inProceedingsofthe21stEuropeanPhotovoltaicSolarEnergyConferenceandExhibition,pp.777–780,Dresden,Germany,September2006.
[8]S.W.Glunz,A.Grohe,M.Hermle,etal.,“Comparison
ofdifferentdielectricpassivationlayersforapplicationinindustriallyfeasiblehigh-efficiencycrystallinesiliconsolarcells,”inProceedingsofthe20thEuropeanPhotovoltaicSolarEnergyConferenceandExhibition,pp.572–577,Barcelona,Spain,June2005.
[9]J.SchmidtandA.G.Aberle,“Easy-to-usesurfacepassivation
techniqueforbulkcarrierlifetimemeasurementsonsiliconwafers,”ProgressinPhotovoltaics:ResearchandApplications,vol.6,no.4,pp.259–263,1998.[10]H.M¨ackelandR.L¨udemann,“Detailedstudyofthecom-positionofhydrogenatedSiNxlayersforhigh-qualitysilicon
surfacepassivation,”JournalofAppliedPhysics,vol.92,no.5,pp.2602–2609,2002.
M.Hofmann,E.Schneiderl¨ochner,W.Wolke,andR.Preu,
“Siliconnitride-siliconoxidestacksforsolarcellrearsidepassivation,”inProceedingsofthe19thEuropeanPhotovoltaicSolarEnergyConferenceandExhibition,pp.1037–1040,Paris,France,June2004.
I.G.Romijn,W.J.Soppe,H.C.Rieffe,A.R.Burgers,and
A.W.Weeber,“Passivationmc-SisolarcellsusingSiNx:H:frommagictophysics,”inProceedingsofthe20thEuropeanPhotovoltaicSolarEnergyConferenceandExhibition,pp.1352–1355,Barcelona,Spain,June2005.
S.DeWolf,Rearpassivationforthinbaselow-costsilicon
solarcells,Ph.D.dissertation,KatholischeUniversit¨atLeuven,Leuven,Belgium,2005.
F.Chen,I.Romijn,A.Weeber,J.Tan,B.Hallam,andJ.
Cotter,“RelationshipbetweenPECVDsiliconnitridefilmcompositionandsurfaceandedgepassivation,”inProceedingsofthe22ndEuropeanPhotovoltaicSolarEnergyConferenceandExhibition,pp.1053–1060,Milan,Italy,September2007.
A.G.Aberle,CrystallineSiliconSolarCells:AdvancedSurface
PassivationandAnalysisofCrystallineSiliconSolarCells,TheUniversityofNewSouthWales,Sydney,Australia,1999.S.Dauwe,L.Mittelst¨adt,A.Metz,andR.Hezel,“Experimental
evidenceofparasiticshuntinginsiliconnitriderearsurfacepassivatedsolarcells,”ProgressinPhotovoltaics:ResearchandApplications,vol.10,no.4,pp.271–278,2002.
S.Dauwe,Low-temperaturesurfacepassivationofcrystalline
siliconanditsapplicationtotherearsideofsolarcells,Ph.D.dissertation,Universit¨atHannover,Hannover,Germany,2004.M.Schaper,J.Schmidt,H.Plagwitz,andR.Brendel,“20.1%-efficientcrystallinesiliconsolarcellwithamorphoussiliconrear-surfacepassivation,”ProgressinPhotovoltaics:ResearchandApplications,vol.13,no.5,pp.381–386,2005.
P.J.Rostan,U.Rau,V.X.Nguyen,T.Kirchartz,M.B.Schubert,
andJ.H.Werner,“Low-temperaturea-Si:H/ZnO/Albackcontactsforhigh-efficiencysiliconsolarcells,”SolarEnergyMaterialsandSolarCells,vol.90,no.9,pp.1345–1352,2006.S.DeWolfandG.Beaucarne,“Surfacepassivationproperties
ofboron-dopedplasma-enhancedchemicalvapordepositedhydrogenatedamorphoussiliconfilmsonp-typecrystallineSisubstrates,”AppliedPhysicsLetters,vol.88,no.2,ArticleID022104,3pages,2006.
H.Plagwitz,B.Terheiden,andR.Brendel,“Formationof
defectsattheamorphoussilicon—crystallinesiliconinterfaceduringannealing,”inProceedingsofthe22ndEuropean
[11][12][13][14][15][16][17][18][19][20][21]10
PhotovoltaicSolarEnergyConferenceandExhibition,pp.936–939,Milan,Italy,September2007.
W.Brendle,V.X.Nguyen,A.Grohe,etal.,“20.5%Efficientsiliconsolarcellwithalowtemperaturerearsideprocessusinglaser-firedcontacts,”ProgressinPhotovoltaics:ResearchandApplications,vol.14,no.7,pp.653–662,2006.
M.Hofmann,S.W.Glunz,R.Preu,andG.Willeke,“21%-efficientsiliconsolarcellsusingamorphoussiliconrearsidepassivation,”inProceedingsofthe21stEuropeanPhotovoltaicSolarEnergyConferenceandExhibition,pp.609–613,Dresden,Germany,September2006.
M.Hofmann,C.Schmidt,N.Kohn,etal.,“Detailedanalysisofamorphoussiliconpassivationlayersdepositedinindustrialin-lineandlaboratory-typepecvdreactors,”inProceedingsofthe22ndEuropeanPhotovoltaicSolarEnergyConferenceandExhibition,pp.1528–1531,Milan,Italy,September2007.G.Agostinelli,P.Choulat,H.F.W.Dekkers,E.Vermari¨en,andG.Beaucarne,“Rearsurfacepassivationforindustrialsolarcellsonthinsubstrates,”inProceedingsofthe4thIEEEWorldConferenceonPhotovoltaicEnergyConversion,vol.1,pp.1004–1007,Waikoloa,Hawaii,USA,May2006.
B.Hoex,F.J.J.Peeters,M.Creatore,M.A.Blauw,W.M.M.Kessels,andM.C.M.vandeSanden,“High-rateplasma-depositedSiO2filmsforsurfacepassivationofcrystallinesilicon,”JournalofVacuumScience&TechnologyA,vol.24,no.5,pp.1823–1830,2006.
M.Hofmann,S.Kambor,C.Schmidt,etal.,“Firingstablesurfacepassivationusingall-PECVDstacksofSiOx:HandSiNx:H,”inProceedingsofthe22ndEuropeanPhotovoltaicSolarEnergyConferenceandExhibition,pp.1030–1033,Milan,Italy,September2007.
W.KernandD.Puotinen,“Cleaningsolutionsbasedonhydro-genperoxideforuseinsiliconsemiconductortechnology,”RCAReview,vol.31,pp.187–206,1970.
R.A.Sinton,A.Cuevas,andM.Stuckings,“Quasi-steady-statephotoconductance,anewmethodforsolarcellmaterialanddevicecharacterization,”inProceedingsofthe25thIEEEPhotovoltaicSpecialistsConference,pp.457–460,Washington,DC,USA,May1996.
S.W.Glunz,D.Biro,S.Rein,andW.Warta,“Field-effectpassivationoftheSiO2-Siinterface,”JournalofAppliedPhysics,vol.86,no.1,pp.683–691,1999.
A.B.Sproul,“Dimensionlesssolutionoftheequationdescrib-ingtheeffectofsurfacerecombinationoncarrierdecayinsemiconductors,”JournalofAppliedPhysics,vol.76,no.5,pp.2851–2854,1994.
A.CuevasandR.A.Sinton,“Predictionoftheopen-circuitvoltageofsolarcellsfromthesteady-statephotoconductance,”ProgressinPhotovoltaics:ResearchandApplications,vol.5,no.2,pp.79–90,1997.
R.A.Sinton,“UserManual:WCT-100PhotoconductanceTool,”SintonConsulting.p.60,2003.
W.Rudolph,C.Bauer,K.Brankoff,etal.,“Plasticfoilsasprimaryhydrogenstandardsfornuclearreactionanalysis,”NuclearInstrumentsandMethodsinPhysicsResearchB,vol.15,no.1–6,pp.508–511,1986.
M.Schwickert,NeuaufbaueinesPelletron-BeschleunigersundUntersuchungenzumLaserhydrierenvonSilizium,Ph.D.disser-tation,Georg-August-Universit¨at,G¨ottingen,Germany,2002.B.Fischer,Lossanalysisofcrystallinesiliconsolarcellsusingphotoconductanceandquantumefficiencymeasurements,Ph.D.dissertation,Universit¨atKonstanz,Konstanz,Germany,2003.
AdvancesinOptoElectronics
[37]D.KrayandS.W.Glunz,“Investigationoflaser-firedrear-side
recombinationpropertiesusingananalyticalmodel,”ProgressinPhotovoltaics:ResearchandApplications,vol.14,no.3,pp.195–201,2006.
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33][34]
[35]
[36]
因篇幅问题不能全部显示,请点此查看更多更全内容