DIAGRAM 4HE FOLLOWING POINTS WERE CONSIDERED WHEN THIS PARTICULAR DESIGN WAS DEVELOPED
3QBUJBMBSSBOHFNFOUPGUIFQOKVODUJPO
4HE FRONTSURFACE 1.2 OF THEPNJUNCTION IN A SOLAR CELL IS USUALLY CALLED ÀEMITTERÁ WHILE THE OTHER 1.2 IS CALLED ÀBASEÁ 4HESE TERMS ARE HISTORICALLY BASED ON THE DEVELOPMENTOFTHETRANSISTOR4HEFIRSTOPENQUESTIONCONCERNINGCELLDESIGNISNOW WHETHER A ÀPONNÁ I E A PTYPE EMITTER AND AN NTYPE BASE OR AN ÀNONPÁ STRUCTURE IS ADVANTAGEOUS 4HIS QUESTION CAN BE ANSWERED SIMPLY BY CONSIDERING MATERIAL PROPERTIES AND DEVICE PROCESSING PROCEDURES FOR THE ANTIMONIDES 'OOD /HMICMETALCONTACTSTOP'A3BCANBEEASILYMADEWITHOUTANYHIGHTEMPERATURE PROCESSINGWHILENTYPEMATERIALREQUIRESANANNEALINGSTEP7HENANNTYPELAYERIS USEDASEMITTERTHEANNEALINGOFTHEMETALCONTACTGRIDISLIKELYTOSHORTCIRCUITTHE PNJUNCTIONBYDIFFUSIONEXCEPTWHENTHENTYPELAYERISVERYTHICK(OWEVERATHICK NTYPEEMITTERMUSTBEEXPECTEDTOBEPROBLEMATICBECAUSE(ITCHCOCKETAL;=
REPORTASMALLMINORITYCARRIERDIFFUSIONLENGTHIN-/60%GROWNNTYPE'A3BBELIEVED TO BE DUE TO DEEP LEVEL DEFECTS ASSOCIATED WITH TELLURIUM HIGHER ÌUGER RECOMBINATIONRATESINNTYPEMATERIALANDLOWHOLEMOBILITY#ONSEQUENTLYAPTYPE EMITTERSHOULDBEUSEDINTHEDEVICES4HISPTYPEEMITTERSHOULDBETHICKENOUGHTO ENSURE THAT MOST OF THE PHOTONS ARE ABSORBED IN IT SO THAT THE MINORITY CARRIER PROPERTIESOFTHENTYPEBASEARENOTVERYIMPORTANT EXCEPTFORTHESATURATIONCURRENT DENSITYSEESECTION&URTHERMOREATHICKPTYPEEMITTERWILLBEBENEFICIALWITH RESPECTTOEXCESSIVESERIESRESISTANCEATLOWDOPINGLEVELS0OSSIBLEPROBLEMSRELATED TOINSUFFICIENTMINORITYCARRIERDIFFUSIONLENGTHSDONOTHAVETOBEEXPECTEDBECAUSE THEMOBILITIES SEESECTIONANDESTIMATEDLIFETIMES ACCORDINGTO;=SUGGEST 'JHVSF #BOE EJBHSBN PG TPMBS DFMM TUSVDUVSF 5IF SFMBUFE MBZFS TFRVFODF JT TIPXOJOGJHVSF5IFDPOUBDUMBZFSBOEUIFCVGGFSMBZFSXFSFPNJUUFEJO UIF CBOE EJBHSBN 5IF CBOE EJTDPOUJOVJUJFT BSF QMBVTJCMF FTUJNBUFT CBTFE POBWBMFODFCBOEEJTDPOUJOVJUZPG∆EV =F7GPS"M4C(B4C
0 1 2 3
-1.2 -0.8 -0.4 0.0 0.4 0.8 1.2
En er gy [ e V]
Depth [µm]
ELECTRON DIFFUSION LENGTHS OF SEVERAL MICROMETERS IN PDOPED 'A3B 4HESE CONSIDERATIONSFINALLYSUGGESTAPNJUNCTIONPROFILEASSHOWNINFIGURESAND ÌSIMILARDESIGNHASALSOBEENREPORTEDINTHELITERATUREWITHREGARDTO 'A)N ÌS3B THERMOPHOTOVOLTAIC CELLS E G ; = 4HESE POINTS COULD ALSO BE CONFIRMED IN EXPERIMENTS3OLARCELLSTRUCTURESWEREEXAMINEDTHATWERECOMPARABLETOEACHOTHER APART FROM THE POSITION OF THE PNJUNCTION -EASUREMENTS CONFIRMED A BETTER QUANTUM EFFICIENCY FIG IN ;= AND ^M6 HIGHER OPENCIRCUIT VOLTAGE FOR A STRUCTURE ACCORDING TO FIGURES AND COMPARED TO A STRUCTURE WITH THIN P EMITTERANDTHICKNBASE SEEALSOSECTION(OWEVERITMUSTFINALLYBEMENTIONED THATHIGHLYDIFFERENT'A3BBASEDSTRUCTURESAREREPORTEDINTHELITERATUREFORDEVICES PRODUCEDONTHEBASISOF:NDIFFUSIONINTOBULKGROWN4EDOPEDSUBSTRATES;=
#BDLJOUFSGBDFQBTTJWBUJPO
ÌSALREADYMENTIONEDABOVEINTERFACESWITHCRITICALRECOMBINATIONVELOCITIESWILLBE PRESENT AT THE BACK SIDE OF THE PNJUNCTION 0ARTICULARLY THE SUBSTRATEEPILAYER INTERFACE CAN BE EXPECTED TO BE A MAJOR PROBLEM DUE TO A HIGH IMPURITY CONCENTRATION4HEREARETWOPOSSIBLEWAYSTOPREVENTTHEDIFFUSIONOFPHOTOEXCITED MINORITYCARRIERSTOTHISINTERFACEAFIELDBASEDONAHOMOEPITAXIALDOPINGGRADIENT A SOCALLED ÀBACK SURFACE FIELDÁ OR À"3&Á AND A HETEROEPITAXIAL CARRIER CONFINEMENT BASED ON A WIDE BAND GAP MATERIAL "OTH APPROACHES ARE INCLUDED IN THE DEVICE STRUCTURE SHOWN IN FIGURES AND 4HE WIDE BAND GAP APPROACH IS COMPLEX BECAUSE NDOPED ÌL'A ÌS3B IS DIFFICULT TO ACHIEVE IN -/60% AND THE CARRYOVER PROBLEM OF ALUMINIUM SECTION MUST ALSO BE CONSIDERED -OREOVER A POTENTIALBARRIERFORNBKPSJUZCARRIERSALSOARISESATTHEHETEROINTERFACEEXCEPTFORVERY HIGHDOPINGCONCENTRATIONS/NTHEOTHERHANDFIGUREIMMEDIATELYSHOWSTHAT THEWIDEBANDGAPMATERIALCANBEEXPECTEDTOBEAMUCHMOREEFFECTIVEMINORITY CARRIERCONFINEMENTTHANAMEREDOPINGGRADIENTÌDOPINGSTEPMUSTBESIGNIFICANTLY HIGHERTHANTHEONEINCLUDEDINFIGURESANDINORDERTOACHIEVEACOMPARABLE CONFINEMENT5NFORTUNATELYTHISISPROHIBITEDBYTHELOWUPPERLIMITOFTHETELLURIUM CONCENTRATIONFOR-/60%GROWN'A3BÌSACONSEQUENCEANEFFECTIVEBACKINTERFACE PASSIVATIONISDIFFICULTTODESIGN
4HE ABOVE CONSIDERATIONS WERE EXPERIMENTALLY EXAMINED IN RELATED DEVICES 4HE ÀCOMBINEDPASSIVATIONAPPROACHÁ DOPING"3&PLUS ÌL'A ÌS3BCONFINEMENTWAS INCLUDEDINACELLSTRUCTUREWITHATHIN ^NM:NDIFFUSEDEMITTERANDCOMPARED TO A RELATED STRUCTURE WITHOUT ANY BACK INTERFACE PASSIVATION ÌS THESE STRUCTURES SHOWASIGNIFICANTDEGREEOFPHOTOGENERATIONINTHEBASEDUETOTHETHINEMITTERTHE BACKINTERFACEPASSIVATIONCANBEPREDICTEDTOINFLUENCETHEDEVICEPERFORMANCEÌS EXPECTED AN IMPROVED QUANTUM EFFICIENCY OVER ALMOST THE WHOLE SPECTRUM WAS FOUND FOR THE DEVICEXJUIPASSIVATION AND CONSEQUENTLY THE COMBINEDPASSIVATION APPROACHWASCONFIRMEDTOWORKFINE FIGIN;=4HESTRUCTUREACCORDINGTO FIGURESANDHASATHICKEREMITTER oMANDDIDNOTSHOWARELATEDLONG WAVELENGTH DECREASE OF THE QUANTUM EFFICIENCY WHEN THE N ÌL'A ÌS3B CONFINEMENT LAYER WAS OMITTED 4HIS IS PROBABLY CAUSED BY THE REDUCED MINORITY CARRIERGENERATIONINTHEBASELAYERSOFTHESESTRUCTURESDUETOTHETHICKEMITTERSÌSA CONSEQUENCETHEN ÌL'A ÌS3BCONFINEMENTLAYERCOULDBEOMITTEDINALLTHEDEVICE STRUCTURES DESCRIBED IN THE FOLLOWING SECTIONS AND A SIMPLER SINGLELAYER DOPING GRADIENTPASSIVATION "3&WASGENERALLYEMPLOYED
'SPOUTVSGBDFQBTTJWBUJPO
ÌS ALREADY SAID ABOVE A DEVICE WITH A THICK PTYPE EMITTER IS ADVANTAGEOUS WITH RESPECT TO PHOTOEXCITED MINORITY CARRIER DIFFUSION LENGTHS METAL CONTACT FORMATION ANDSERIESRESISTANCE(OWEVERTHESUCCESSOFTHISCONCEPTDEPENDSONALOWFRONT SURFACE RECOMBINATION VELOCITY I E A GOOD SURFACE PASSIVATION /THERWISE SURFACE LOSSESWILLDOMINATETHEPERFORMANCEOFTHESOLARCELL4HEPASSIVATIONSHOULDOPUBE BASEDONAMEREHOMOEPITAXIALDOPINGGRADIENTBECAUSEINTHISCASETHEABSORPTION INTHEPASSIVATIONLAYERWOULDBETOOHIGHÌSANALTERNATIVEANOXIDELAYERCANBE USED FOR SURFACE PASSIVATION PURPOSES IN SILICON SOLAR CELLS BUT FOR )))6 COMPOUND SEMICONDUCTORS THE DEPOSITION OF GOOD OXIDES WITH LOW INTERFACE STATE DENSITY IS PROBLEMATIC4HEREFOREMINORITYCARRIERCONFINEMENTONTHEBASISOFAWIDEBANDGAP HETEROEPITAXIALÀWINDOWÁLAYERISTHEMOSTPROMISINGALTERNATIVEFORTHEFRONTSURFACE PASSIVATION 3OME RELATED REPORTS ON THE ÌL'A ÌS3BPASSIVATION OF 'A)N ÌS3B EMITTERSCOULDBEFOUNDINTHELITERATURE;=WITH;=STATINGTHATAFITTING PROCEDURECOULDNOLONGERDETECTANYSIGNIFICANTFRONTSURFACERECOMBINATIONLOSSIN THE RELATED DEVICES 4HE ABOVE CONSIDERATIONS WERE EXPERIMENTALLY EXAMINED WITH RESPECTTO-/60%GROWN'A3BBASEDDEVICESINTHISTHESISÌSTRUCTUREACCORDINGTO FIGURE WAS DEPOSITED IN WHICH THE ÌL'A ÌS3B WINDOW LAYER WAS OMITTED SAMPLE 4HE MEASURED QUANTUM EFFICIENCIES OF THE DEVICES ARE COMPARED IN FIGURE 4HE ÌL'A ÌS3BWINDOW LAYER IMPROVES THE QUANTUM EFFICIENCY DRAMATICALLYOVERALMOSTTHEENTIRESPECTRUM
&IGUREALSOCONTAINSTHEINTERNALQUANTUMEFFICIENCYOFSAMPLE4HISSAMPLEIS EQUIVALENT TO THE STRUCTURE SHOWN IN FIGURES TO EXCEPT THAT THE ÌL'A ÌS3BBACK INTERFACE PASSIVATION LAYER WAS OMITTED 4HIS INTERNAL QUANTUM EFFICIENCY WAS CALCULATED ON THE BASIS OF THE REFLECTION OF A RELATED WAFER BEFORE THE DEVICEFABRICATIONASSUMINGACONTACTGRIDMETALCOVERAGEOFTHEFINALDEVICE4HIS
'JHVSF &YUFSOBM RVBOUVN FGGJDJFODJFT PG TBNQMF XJUI "M(B "T4C XJOEPX MBZFS BOE TBNQMF XJUIPVU BOZ GSPOU TVSGBDF QBTTJWBUJPO 5IF QBTTJWBUJOH FGGFDU PG UIF XJOEPX MBZFS JT DMFBSMZ EFNPOTUSBUFE 5IF FTUJNBUFEJOUFSOBM2&PGTBNQMF TBNQMFXJUIPVUO "M(B "T4C
#4'JTBMTPTIPXO
0.4 0.8 1.2 1.6 2.0
0 20 40 60 80 100
IQE 769
EQE 660
EQE 658
Qu an tum ef fi ci en cy [ % ]
Wavelength [µm]
VALUE IS A CAUTIOUS ESTIMATE AND PROBABLY TOO SMALL DUE TO THE THICKER CONTACT GRID FINGERSAFTERELECTROPLATING ÌS THE RELATED %1% MEASUREMENTS WERE PERFORMED BEFORE THE REMOVAL OF THE NM 'A3B CONTACT LAYER THE RELATED ABSORPTION LOSS LEADS TO A FURTHER UNDERESTIMATION OF THE REAL DEVICE PERFORMANCE (OWEVER )1%VALUES ABOVE AREACHIEVEDFORTHESPECTRALRANGEoMWHICHMUSTBECONSIDEREDAVERY GOOD VALUE FOR THESE EARLY STAGES OF THE DEVICE DEVELOPMENT 4HE BEST VALUES ARE ACHIEVED IN THE RANGE oM FOR SHORTER WAVELENGTHS THE ONSET OF THE STRONGER ABSORPTION OF THE WINDOW AND CONTACT LAYERS CAN BE SEEN WHILE LOSSES AT LONGER WAVELENGTHSCANPROBABLYBEATTRIBUTEDTORECOMBINATIONINTHEBASE4HESMALLPEAKAT NMCORRESPONDSTOALOCALREFLECTIONMINIMUMOF'A3B)TMUSTBENOTEDTHATTHE QUANTUMEFFICIENCYOFTHEDEVICEINTHEINTENDEDAPPLICATION SEESECTIONISONLYOF INTERESTFORWAVELENGTHSABOVE^NMBECAUSEPHOTONSWITHHIGHERENERGIESWILLBE ABSORBEDINTHE'AÌSBASEDSOLARCELL
&URTHERIMPROVEMENTSOFTHEDEVICECANBEEXPECTEDFORWINDOWLAYERSTHATSHOWLESS ABSORPTION REDUCEDTHICKNESSESHIGHERBANDGAPSANDIMPROVEDBASELAYERSÌDIRECT METAL CONTACT TO THE ÌL'A ÌS3B WINDOW LAYER SHOULD ALSO BE EXAMINED IN ORDER TO FACILITATETHEDEVICEFABRICATIONWITHRESPECTTOTHEREMOVALOFTHE'A3BCONTACTLAYER &NJUUFSEPQJOHMFWFM
Ì SERIES OF SOLAR CELL STRUCTURES WAS DEPOSITED WHERE THE ACCEPTOR DOPING LEVEL IN THE EMITTER WAS VARIED WHILE ALL OTHER PARAMETERS WERE HELD CONSTANT 4HE REFERENCE STRUCTUREIS WITHOUT ÌL'A ÌS3BBACKINTERFACECONFINEMENT4HEDEVICES
'JHVSF 4IPSUDJSDVJUDVSSFOUPGJEFOUJDBMTPMBSDFMMTXIFSFPOMZUIFBDDFQUPSEPQJOH MFWFMPGUIFFNJUUFSXBTWBSJFE5IFTIPSUDJSDVJUDVSSFOUXBTDBMDVMBUFEPO UIFCBTJTPGUIFLOPXODFMMBSFBBOJODJEFOU".HTQFDUSVNBOERVBO UVNFGGJDJFODZNFBTVSFNFOUT*OBEEJUJPONFBTVSFEWBMVFTBSFTIPXOUIBU XFSFBWFSBHFEPWFSBMMDPNQBSBCMFDFMMTPOPOFXBGFSJOBTPMBSTJNVMBUPS
1E160 1E17 1E18
2.0 2.5 3.0 3.5 4.0
calculated Isc on the basis of an EQE, cell area (0.13 cm2 ) and incident spectrum (AM 1.5g) measured Isc, averaged over all comparable cells on one wafer
S h ort-c irc ui t cu rren t I
SC[m A ]
Emitter doping p [cm
-3]
WERE PROCESSED AND THE OPENCIRCUIT VOLTAGES AND THE EXTERNAL QUANTUM EFFICIENCIES MEASURED /N THE BASIS OF AN INCIDENT SPECTRUM AND A GIVEN CELL AREA THE QUANTUM EFFICIENCY CAN BE TRANSLATED INTO A SHORTCIRCUIT CURRENT 4HE CORRESPONDING RESULTS ARE SHOWN IN FIGURE TOGETHER WITH MEASURED VALUES THAT WERE AVERAGED OVER ALL COMPARABLECELLSONONEWAFERINASOLARSIMULATOR4HELOWESTDOPINGLEVELCORRESPONDS TOANOMINALLYUNDOPEDEMITTERWITHABACKGROUNDCONCENTRATIONOFNATIVEACCEPTORS 4WOPROMINENTFEATURESOFTHEDATAMUSTBEDISCUSSED&IRSTLYAVERYLOWLEVELOFSILICON ACCEPTORSSEEMSTOGIVETHEBESTSHORTCIRCUITCURRENTOFALLSTRUCTURES4HISRESULTMIGHT REPRESENT AN IMPROVED MINORITY CARRIER LIFETIME DUE A REDUCED CONCENTRATION OF DEEP LEVELSASSOCIATEDWITHNATIVEACCEPTORS SEESECTION(OWEVERARELATEDCALIBRATION MISTAKEDURINGTHE%1%MEASUREMENTOFTHESAMPLECANNOTBERULEDOUTASTHEDIRECT MEASUREMENT IN A SOLAR SIMULATOR DOESNOTGIVE A SIMILARLY DISTINCT RESULT 4HE HIGHEST DOPING LEVEL GAVE A SLIGHTLY REDUCED SHORTCIRCUIT CURRENT 4HIS RESULT IS BELIEVED TO REPRESENT THE ONSET OF A STRONGER LOSS OF MINORITY CARRIERS DUE TO A HIGHER RADIATIVE RECOMBINATIONRATEWHILETHELIFETIMEATLOWERACCEPTORLEVELSISPROBABLYDOMINATEDBY DOPINGINDEPENDENT 3HOCKLEY2EAD(ALL RECOMBINATION 4HE RELATED BEHAVIOUR OF THE OPENCIRCUITVOLTAGEISSHOWNINFIGURE
ÌS THE MINORITY CARRIER DIFFUSION CURRENT IS A LINEAR FUNCTION OF THE MINORITY CARRIER CONCENTRATION EQUIN;=p⋅n=ni2MEANSTHATTHEELECTRONDIFFUSIONCURRENTIN THEEMITTERWILLFOLLOW
A diff
e N
I ∝ 1 ÌSARESULTTHEELECTRONICSATURATIONCURRENTDENSITY IS REDUCED WHEN THE EMITTER DOPING LEVEL IS INCREASED &URTHERMORE IT WAS DERIVED IN SECTION THAT THE OPENCIRCUIT VOLTAGE INCREASES WHEN THE SATURATION CURRENT DECREASES#ONSEQUENTLY7PDCANBEEXPECTEDTORISEWITHINCREASING/"ASLONGASTHE OTHERMATERIALPARAMETERSSUCHASCARRIERLIFETIMESANDMOBILITIESSTAYCONSTANT
1E160 1E17 1E18
350 400 450
Emitter doping p [cm
-3] O p e n circu it v o ltag e V
oc[m V ]
All devices without cap etching ! Best Voc at C~35 plotted.
second batch third batch
first batch
'JHVSF 0QFODJSDVJUWPMUBHFPGJEFOUJDBMTPMBSDFMMTXIFSFPOMZUIFBDDFQUPSEPQJOH MFWFMPGUIFFNJUUFSXBTWBSJFETUOEBOESECBUDIEFTDSJCFEJGGFSFOUTFUT PGFYQFSJNFOUT
4HEOPENCIRCUITVOLTAGEWASMEASUREDATACONCENTRATIONRATIOOFAPPROXIMATELY#
DEFINEDBYTHECORRESPONDINGSHORTCIRCUITCURRENTBECAUSEANOPERATIONOFTHEDEVICE INACONCENTRATINGSYSTEMISEVENTUALLYINTENDED-OREOVERSHUNTEFFECTSTENDTOLOWER 7PDATINTENSITIESAROUND#%ACHDATAPOINTREPRESENTSTHEBESTVALUEOFALLCELLSON ONEWAFER)NTHEEXPERIMENTSANINCREASEOFTHEOPENCIRCUITVOLTAGEWITHINCREASING EMITTER DOPING CAN BE OBSERVED FOR THE LOWEST DOPING LEVELS (OWEVER BETWEEN
1016
5⋅ CMAND8⋅1017CMASIGNIFICANTVARIATIONISNOLONGERFOUNDÌRELATEDANALYSIS OFTHESATURATIONCURRENTDENSITYCANBEFOUNDIN;=
4HE BATCH NUMBERS IN THE FIGURE INDICATE DIFFERENT SETS OF EXPERIMENTS AND SHOW A CERTAIN IRREPRODUCIBILITY 7HETHER THIS MUST BE ATTRIBUTED TO GROWTH ISSUES OR THE FABRICATION TECHNOLOGY COULD NOT BE CLARIFIED &OR THE GIVEN DEVICE STRUCTURE EMITTER DOPINGLEVELSAROUND1⋅1017CMAPPEARMOSTPROMISINGINTHISCONTEXT
#BTFEPQJOHMFWFM
)N ADDITION TO THE EMITTER DOPING LEVEL VARIATION OF THE PRECEDING SECTION A SERIES OF SOLAR CELL STRUCTURES WAS DEPOSITED WHERE THE DONOR DOPING LEVEL IN THE EMITTER WAS VARIEDWHILEALLOTHERPARAMETERSWEREHELDCONSTANTÌGAINTHEREFERENCESTRUCTUREIS /PENCIRCUIT VOLTAGE AND QUANTUM EFFICIENCIES OF THE RELATED DEVICES WERE MEASURED4HECORRESPONDINGRESULTSARESHOWNINFIGURESAND4HEEXTERNAL QUANTUM EFFICIENCIES ARE VERY SIMILAR FOR ALL BASE DOPING LEVELS 4HIS CAN EASILY BE EXPLAINEDBECAUSEMOSTOFTHEABSORPTIONOCCURSINTHEEMITTERPARTOFTHESTRUCTUREAND THE MINORITY CARRIER LIFETIME IN THE BASE DOES NOT SIGNIFICANTLY INFLUENCE THE DEVICE PERFORMANCE4HEOPENCIRCUITVOLTAGEINCREASESWITHINCREASINGDONORCONCENTRATIONÌS
'JHVSF &YUFSOBMRVBOUVNFGGJDJFODJFTPGEFWJDFTUSVDUVSFTCBTFEPOUIFSFGFSFODF TUSVDUVSF GJHVSF XJUIPVU "M(B "T4C CBDL JOUFSGBDF DPOGJOFNFOU /P WBSJBUJPO PG UIF RVBOUVN FGGJDJFODJFT XJUI UIF CBTF EPQJOHMFWFMJTPCTFSWFE
0.4 0.8 1.2 1.6 2.0
0 20 40 60 80 100
1E17 cm-3 2E17 cm-3 5E17 cm-3 7E17 cm-3
E x te rn al qu an tu m e ff ici en cy [ % ]
Wavelength [µm]
ASATURATIONHASNOTBEENOBSERVEDEVENHIGHERDOPINGLEVELSSHOULDBEEXAMINEDIN THEFUTUREÌSIGNIFICANTLYBETTERVOLTAGEWASACHIEVEDFORAÀFIRSTBATCHÁDEVICE SAMPLE INDICATINGREPRODUCIBILITYPROBLEMSCONCERNINGGROWTHORFABRICATIONTECHNOLOGY
"M(B "T4CIPNPKVODUJPOTPMBSDFMMT
)N SECTION THE MULTIJUNCTION APPROACH FOR EXTREMELYHIGHPHOTOVOLTAICCONVERSIONEFFICIENCIES WAS DESCRIBED )T WAS SHOWN THAT IN A JUNCTION SYSTEM THE THIRD JUNCTION SHOULD BE BASED ON A BANDGAPOFAPPROXIMATELYE6AND ÌL'A ÌS3B WAS SUGGESTED AS A PROMISING MATERIAL (OWEVER THE RESULTS PRESENTED IN SECTION HAVE SHOWN THATTHE-/60%OFHIGHQUALITY ÌL'A ÌS3BISAN EXTREMELY DIFFICULT CHALLENGE "ASED ON THE RESULTS OFSECTIONAPRELIMINARYCELLSTRUCTURE SAMPLE WITH A BAND GAP AROUND E6 COULD BE DEPOSITED4HERELATEDLAYERSEQUENCEISPRESENTED INFIGUREANDTHEBANDDIAGRAMINREALSPACE CAN BE FOUND IN FIGURE ÌS SHORT MINORITY CARRIER LIFETIMES AND BAD MOBILITIES HAVE TO BE EXPECTEDINTHEMATERIALTHEDIFFUSIONLENGTHSARE LIKELYTO BE RELATIVELY LIMITED PARTICULARLY IN NTYPE MATERIAL #ONSEQUENTLY THE LAYERS IN THE ACTIVE REGION OF DEVICE WERE CHOSEN TO BE THINNER THAN IN THE 'A3BCELLS ÌS NTYPE DOPING OF ÌL'A ÌS3B WITH HIGH ÌLCONCENTRATIONS WAS PREVENTED BY A HIGH BACKGROUND HOLE CONTAMINA 'A3BCONTACTLAYER
P%CMDNM
ÌL'A ÌS3B
P^%CMD^NM
ÌL'A ÌS3B
P^%CMD^NM
ÌL'A ÌS3B
N^%CMD^NM
'A3BBUFFERLAYER
N%CMDNM
'JHVSF
-BZFS TFRVFODF PG TPMBS DFMM TUSVDUVSF 5IF SFMBUFE CBOE EJBHSBN JT TIPXO JO GJHVSF
'JHVSF 7PDPGJEFOUJDBMTPMBSDFMMTXIFSFPOMZUIFEPOPSEPQJOHMFWFMPGUIFCBTF XBTWBSJFETUBOEOECBUDIEFTDSJCFEJGGFSFOUTFUTPGFYQFSJNFOUT
1E17 1E18
0 350 400 450
All devices without cap etching ! Best Voc at C~35 plotted.
2E17 3E17 6E17
first batch second batch
Open ci rc u it v o lt ag e V
oc[m V ]
Base doping n [cm
-3]
0.0 0.5 1.0 1.5 -1.2
-0.8 -0.4 0.0 0.4 0.8 1.2
~0.9 eV 0.72 eV
~0.9 eV
En er gy [ e V]
Depth [µm]
'JHVSF #BOEEJBHSBNNPGTPMBSDFMMTUSVDUVSF5IFSFMBUFEMBZFSTFRVFODFJT TIPXOJOGJHVSF5IFDPOUBDUMBZFSXBTPNJUUFEJOUIFCBOEEJBHSBN 5IFCBOEEJTDPOUJOVJUJFTBSFQMBVTJCMFFTUJNBUFTCBTFEPOBWBMFODFCBOE EJTDPOUJOVJUZPG∆EV =F7GPS"M4C(B4C
'JHVSF &YUFSOBMRVBOUVNFGGJDJFODZPGUIF "M(B "T4CTPMBSDFMM5IFbEBSLm NFBTVSFNFOU SFQSFTFOUT UIF TUBOEBSE UFDIOJRVF XJUIPVU BOZ CJBT MJHIU 5IF NFBTVSFNFOUT UFSNFE bUPQm BOE bCPUUPNm BSF SFMBUFE UP NPSF BEWBODFENFUIPETXIJDIBSFVTVBMMZFNQMPZFEGPSUIFBOBMZTJTPGUBOEFN DFMMT"SFMBUFESFGFSFODFDBOCFGPVOEJOUIFUFYUPOUIFGPMMPXJOHQBHF
0.4 0.8 1.2 1.6 2.0
0 20 40 60 80 100
#772-4 top #772-4 bottom #772-4 dark
E x te rn al qu an tum efficien c y [% ]
Wavelength [µm]
TIONNOWIDEBANDGAPBACKSIDEPASSIVATIONOFTHEACTIVEREGIONCOULDBEINCORPORATED INTOTHESTRUCTURE
4HECELLGAVEANOPENCIRCUITVOLTAGEOFM6ATACONCENTRATIONRATIOOFAROUND 4HE MEASURED QUANTUM EFFICIENCY IS SHOWN IN FIGURE (ERE THE ÂDARKÀ MEASUREMENT REPRESENTS THE STANDARD TECHNIQUE WITHOUT ANY BIAS LIGHT ÌN %1% OF AROUND WAS MEASURED ABOVE THE BAND GAP OF ÌL'A ÌS3B ^ E6 IE^NM"ESIDESASURPRISINGANDVERYINTERESTINGPECULIARITYWASMEASUREDAT WAVELENGTHSBETWEENANDE6IEBETWEENTHEBANDGAPSOF ÌL'A ÌS3BAND 'A3B (ERE THE ÀDARKÁ %1% OF THE DEVICE DOES NOT VANISH BUT TAKES A FINITE VALUE SLIGHTLYABOVE4HISCANPOSSIBLYBEEXPLAINEDBYABSORPTIONINTHEBUFFERLAYER$UE TOTHEIRTHERMALDISTRIBUTIONSOMEOFTHEPHOTOEXCITEDHOLESINTHE'A3BBUFFERMIGHTBE ABLE TO OVERCOME THE BARRIER INTO THE ÌL'A ÌS3B BASE (ERE THEY LEAD TO A PHOTOCURRENTANDHENCECONTRIBUTETOTHEQUANTUMEFFICIENCY
)N ORDER TO EXAMINE THIS DEVICE IN MORE DETAIL THE CELL WAS ANALYSED WITH CHARACTERISATION TECHNIQUES THAT ARE USUALLY EMPLOYED FOR THE ANALYSIS OF MONOLITHIC TANDEM CELLS 4HE CORRESPONDING MEASUREMENTS CAN BE FOUND IN ;= (OWEVER AN EXPLANATIONFORTHERELATEDDATAWITHREGARDTOTHEDEVICESTRUCTURECANNOTYETBEGIVEN