La lecture en ligne est gratuite
Le téléchargement nécessite un accès à la bibliothèque YouScribe
Tout savoir sur nos offres
Télécharger Lire

Nitride nanostructures for spintronic applications [Elektronische Ressource] / Ralf Schuber. Betreuer: H. Kalt

De
123 pages
Nitride nanostructures for spintronic applicationsZur Erlangung des akademischen Grades einesDOKTORS DER NATURWISSENSCHAFTENvon der Fakult at fur Physik des Karlsruher Instituts fur TechnologiegenehmigteDISSERTATIONvonDipl. Phys. Ralf Schuberaus L orrachTag der mundlic hen Prufung: 1. Juli, 2011Referent: Prof. Dr. H. KaltKorreferent: Prof. Dr. Th. SchimmelParts of this work have already been published or accepted for publishing:R. Schuber, M. M. C. Chou, P. Vincze, Th. Schimmel, D. M. Schaadt, "Growth ofA-plane GaN growth on (010) LiGaO by plasma-assisted molecular beam epitaxy",2Journal of Crystal Growth 312 (2010) 1665-1669.R. Schuber, M. M. C. Chou, D. M. Schaadt, "Growth of M -plane GaN on (100)LiGaO by plasma-assisted molecular beam epitaxy", Thin Solid Films 518 (2010)26773-6776.R. Schuber, Y. L. Chen, C. H. Shih, T. H. Huang, P. Vincze, I. Lo, L. W. Chang, Th.Schimmel, M. M. C. Chou, D. M. Schaadt, "Growth of non-polar GaN on LiGaO2by plasma-assisted MBE", Journal of Crystal Growth 323 (2010) 76-79.R. Schuber, M. M. C. Chou, P. Vincze, Th. Schimmel, D. M. Schaadt, "Growthof M - and A-plane GaN on LiGaO by plasma-assisted MBE", accepted in AIP2Conference Proceedings (2010).C. H. Shih, T. H. Huang, R. Schuber, Y. L. Chen, L. W. Chang, I. Lo, M. M. C.Chou, D. M. Schaadt, "Microstructure of M - and A-plane GaN on LiGaO grown2by plasma-assisted MBE", Nanoscale Research Letters 6 (2011) 425.R. Schuber, P. R. Ganz, F. Wilhelm, A. Rogalev, D.
Voir plus Voir moins

t:Korreferen

t:Referen

Tagderm¨undlichenPr¨ufung:

2011Juli,1.

KaltH.Dr.Prof.

himmelScTh.Dr.Prof.

Dipl.Phys.RalfSchuber
horracL¨aus

onv

DISSERONTIAT

genehmigte

ZurErlangungdesakademischenGradeseines

DOKTORSDERNATURWISSENSCHAFTEN

vonderFakult¨atf¨urPhysikdesKarlsruherInstitutsf¨urTechnologie

Nitridenanostructuresforspintronicapplications

Partsofthisworkhavealreadybeenpublishedoracceptedforpublishing:

R.Schuber,M.M.C.Chou,P.Vincze,Th.Schimmel,D.M.Schaadt,”Growthof
A-planeGaNgrowthon(010)LiGaO2byplasma-assistedmolecularbeamepitaxy”,
JournalofCrystalGrowth312(2010)1665-1669.
R.Schuber,M.M.C.Chou,D.M.Schaadt,”GrowthofM-planeGaNon(100)
LiGaO2byplasma-assistedmolecularbeamepitaxy”,ThinSolidFilms518(2010)
6773-6776.R.Schuber,Y.L.Chen,C.H.Shih,T.H.Huang,P.Vincze,I.Lo,L.W.Chang,Th.
Schimmel,M.M.C.Chou,D.M.Schaadt,”Growthofnon-polarGaNonLiGaO2
byplasma-assistedMBE”,JournalofCrystalGrowth323(2010)76-79.
R.Schuber,M.M.C.Chou,P.Vincze,Th.Schimmel,D.M.Schaadt,”Growth
ofM-andA-planeGaNonLiGaO2byplasma-assistedMBE”,acceptedinAIP
ConferenceProceedings(2010).
C.H.Shih,T.H.Huang,R.Schuber,Y.L.Chen,L.W.Chang,I.Lo,M.M.C.
Chou,D.M.Schaadt,”MicrostructureofM-andA-planeGaNonLiGaO2grown
byplasma-assistedMBE”,NanoscaleResearchLetters6(2011)425.
R.Schuber,P.R.Ganz,F.Wilhelm,A.Rogalev,D.M.Schaadt,”Localelectronic
structureofCu-dopedGaNinvestigatedbyXANESandXLD”,PhysicalReviewB
155206.(2011)84

Inlischeaddition,Gesellscresultshaftwe(DPG)represenandthetedatfollothewingspringinternationalmeetingsoftheconferences:DeutschePhysika-

16thInternationalConferenceonMolecularBeamEpitaxy,Berlin,Germany2010.
30thInternationalConferenceonthePhysicsofSemiconductors,Seoul,Korea2010.
InternationalConferenceonSuperlatttices,NanostructuresandNanodevices,Peking,
2010.ChinaEuroMBE,L’alpd’Huez,France2011.

AbbreviationsofList

2DEGAESAHEAFMBEPBMPCTEDMSELOGEXAFSFETFIBFWHMHEMTHVPEOLALDLEDLGOMBEMCDMOCVDVPEMOMQWAMBEPPIDQCSEQDQWRBSRHEEDRMBErms

tAwo-ugerdeimensionallectronspelectronectroscopgyas
anomalousHalleffect
atomicforcemicroscopy
beamequivalentpressure
boundmagneticpolaron
coefficientofthermalexpansion
dilutemagneticsemiconductors
eepitaxialxtendedlxay-raeryoavergbsorptionrowthfinestructure
ffoield-cusedeiffectontbeamransistor
fhullighweidthlectronathmalfobilitmyaximtumransistor
hLiAydridelO2vaporphaseepitaxy
llighasertediodemittingdiode
LiGaO2
mmagneticolecularcbeamircularedicpitaxyhroism
metalorganicchemicalvapordeposition
metalorganicvaporphaseepitaxy
multiplequantumwell
plasmaassistedmolecularbeamepitaxy
proportionalintegralderivative
qQuanuantumtumdCotonfinedStarkEffect
ellwtumuanqRutherfordbackscattering
rreactiveflectionemhigholecularenergybeameelectronpitaxydiffraction
rootmeansquare

ii

SEMSIMSSQUIDTCETCTEMTMVSMXANESXLDXRDXMCD

scanningelectronmicroscopy
secondaryionmassspectroscopy
superconductingquantuminterferencedevice
Trichloroethylene
eraturetempCurietransmissionelectronmicroscopy
etalmransitiontvibratingsamplemagnetometer
x-rayabsorptionnearedgestructure
x-raylineardichroism
iffractiondy-raxx-raymagneticcirculardichroism

iii

Contents

ductionIntro1

2Ashortintroductiontospintronics
2.1Thebasicideaofspintronics.......................
2.2Currentstatusondilutemagneticsemiconductors(DMS).......
2.3Theroleofnitridesinspintronics....................
2.3.1Mn-dopedGaN..........................
2.3.2Gd-dopedGaN..........................
2.3.3Cu-dopedGaN..........................

3PropertiesandfabricationofGaN
3.1BasicparametersandcrystalstructureofGaN.............
3.2InternalfieldsinGaN...........................
3.3FabricationofGaN............................
3.3.1Choiceofsubstrate........................
3.3.2Molecularbeamepitaxyandthesystemusedinthiswork..

4Non-polarGaNfilms
4.1GrowthofM-planeGaNon(100)γ-LiAlO..............
24.1.1Propertiesofthesubstrateγ-LiAlO..............
24.1.2GrowthparameteroptimizationforM-planeGaNgrowthon
(100)γ-LiAlO..........................
24.1.3SummaryofGaNgrownonγ-LiAlO..............
24.2Growthofnon-polarGaNonβ-LiGaO.................
24.2.1Propertiesofthesubstrateβ-LiGaO..............
24.2.2Growthofβ-LiGaO.......................
24.2.3GrowthofM-planeGaNon(100)β-LiGaO..........
24.2.4GrowthofA-planeGaNon(010)β-LiGaO..........
24.2.5SummaryandconclusionofGaNgrownonβ-LiGaO.....
2

5CuincorporationinGaN
5.1Experimentalprocedure.........................
5.2Resultsandconclusion..........................
5.2.1γ-CuGasurfacecompounds..................
49

2

661015171920

252527313336

4141414249515153536576

78798182

6

5.3

5.2.35.2.2X-raX-rayylinearAbsorptiondichroismNearofEdgeCu-dopStreductureGaNof...Cu-dop..ed.
5.2.4SummaryCofonsideringtheCustrainincorpinγoration-Cu9Gastudy4incompGaNounds..........

rySumma

...GaN......

....

....

....

1

84909496

98

1ionductIntro

Thefieldofspintronics,i.e.theapplicationoftheelectron’sspininadditiontoits
chargeforfuturesuperiordevicefunctionality,isnowadaysofgreatinterestboth
forresearchandcommercialpurposes[1].Spintronicsinmetalsisalreadywidely
usedincomputerstoragedevices[2];recentlythenotionofbringingspintronicsto
devicesbasedonsemiconductortechnologyhastriggeredextensiveinvestigations
onamaterialclasscalleddilutemagneticsemiconductors(DMS).DMSaresemi-
conductorsdopedwithelementsthatchangethehostsemiconductormaterialto
exhibitmagneticproperties.Theprospectofbeingabletohandlethechargecarri-
ers’spinforinformationtransporthasledtoalargenumberofnewdevicedesigns
likethespinfield-effecttransistor(FET)andthespinlightemittingdiode(LED)
[3,4],promisingtorevealnewapplicationsoranimprovementofexistingelectrical
devices.

Beforethebenefitsofnewspintronicsapplicationscanbeexploitedanumberof
issueshavetobesolved,likefindinganappropriatematerialforefficientroomtem-
peraturespin-injectionandanappropriatesemiconductormaterialwithproperties
promisingthedesireddevicefunctionality.Inordertocontributetotheunderstand-
ingandfabricationofefficientspintronic-relatedmaterials,thisthesisaimstotackle
twopoints.Firstly,thematerialsystemofgroupIII-nitrides,especiallygalliumni-
tride(GaN)isinvestigatedforbeinganadequatematerialsystemforthe”active”
regionofadevice.Secondly,theunderstandingofaprospectivespin-alignercandi-
dateisimprovedbyansweringthequestionwhereCuisincorporatedinCu-doped
GaN,aroomtemperatureDMS.

itsGaNhasmagnificenbecometoneopticalofandthemostelectricalimpproportantertiesitissemiconductorsnowusedofinourage[applications5].Becausesuchofas
lighting(projectors,TVs,automobiles,displays,trafficlightsandgenerallighting)
[p6o,w7er],aslaserswellas(gamehighconsolesfrequencyandpowerBlue-Rayelectronplayicsers)[7].[7],Fig.de1.1tectorsdepicts[8,a9],andcompanhighy’s
websiteenormousadvertisingapplicabilittheyandmtheultitudefavoforableapplicationspropertiesofofGaNGaNincouldtheirgreatlybusiness.suppTheort
thesuccessofspintronicsdevices,ifthesewerebasedonGaN.

spGaNondsistoaadirwaectvbandelengthgapnear355materialnmwithwhichaisbandonthegapvioletofabsideoutof3.5visibleeV.Thislight.corre-One

3

Figure1.1:istakExamplesenfromofthewapplicationsebsiteoofftheGaNasaNitronexcommercialCorporationpro[duct.10].Thegraphic

oftheprobablymostquotedpreferencesofthenitridematerialsystemistheband
gaprangeexhibitedbyAlN,GaNandInN,namelyfromtheUV(6.2eVforAlN)to
theIR(0.7eVforInN).Thisleadstothepossibilityoffabricatingbandgapsinthe
completevisiblerangebyformingternaryorquaternaryalloysofAl,Ga,InandN.
Notonlytheopticalpropertiesarepromising;thematerialalsoexhibitsveryhigh
electronmobilitywhichisessentialforhigh-power,high-frequencyelectronicappli-
cations,likee.g.inhighelectronmobilitytransistors(HEMT)[11].Highelectron
mobilityatroomtemperatureisusuallyaround1500cm22/VsinAl0.25Ga0.75N/GaN
heterostructuresandhaveb12eensho−2wntoreach75000cm/Vsat4.2Kwithasheet
electrondensityof1.53×10cmforatwo-dimensionalelectrongas(2DEG)in
Al0.05Ga0.95N/GaN[12].Furthermore,thehighchemical,thermalandmechani-
calstabilityofGaNisofbenefittomanyapplications,makingthematerialmore
worthwhileforinvestigation.
GaNisalsoamaterialwhichexhibitsapolaraxisalongthe[0001]c-direction.Due
tothehexagonalcrystalsymmetryandanaturaldiversionofGaNfromtheperfect
wurtzitestructure,GaNshowsspontaneousandpiezoelectricpolarizationalongthis
polaraxis.Thisfactleadstothepresenceofpolarizationchargesatinterfaces
andisespeciallycriticalwhenmultilayeredheterostructuresarefabricatedinc-
direction.Thepolarizationchargescauseelectricfieldsinsidethestructuresleading
toadistortionoftheelectronicbandstructureandthesocalledquantumconfined
Starkeffect(QCSE)instructuresoflowdimensionality.Themainconsequencesare
areductionoftheoscillatorstrengthduetoaspatialseparationofelectronsand
holesaswellasadecreaseintheenergyoftheradiativetransition.

4

OneapproachtoavoidtheunwantedconsequencesoftheQCSEisthegrowthof
epitaxiallayerswithnon-polarsurfaces,suchastheM-andA-planes,wherethe
internalelectricfieldsrunperpendicularinsteadofparalleltothecrystalgrowth
direction.IthasbeenexperimentallydemonstratedthatinsuchacaseGaNlight
emittingdevicesshowhigherefficiency[13]makinggrowthonthesecrystalplanes
e.attractivhighlyBecausenon-polarGaNsubstratesforhomoepitaxyareonlyavailableinverylimited
sizesandextremelyexpensive,alternativesubstratesforheteroepitaxyofnon-polar
GaNareneeded.Inthecontextofthisthesis,anewplasmaassistedmolecularbeam
epitaxy(PAMBE)systemespeciallyaimedatthegrowthofnitridesemiconductors
wassetupandbroughttooperationforgrowthofGaN.Usingthismachine,growth
ofnon-polarGaNwasinvestigatedontwodifferentsubstrates,namelyLiAlO2and
LiGaO2.Thegrowthparameterswereoptimizedtowardsaflatsurfacemorphology
andhighcrystalqualityoftheresultingM-andA-planeGaNfilms.
Forapplicationsusingthespinofchargecarriers,theirspinsneedtobecorrectly
orientedandinjectedintothe”active”materialinanefficientway,i.e.without
lossofthespinorientation.Therefore,anadequatespin-alignerhastobefound.
Forspininjectionintoasemiconductorthemoststraightforwardapproachwould
betouseaDMS[14,15].OneimportantdesiredpropertyofaDMSforfeasible
spintronicdeviceoperationisferromagnetismatroomtemperature.Manydifferent
DMSmaterialshavebeeninvestigatedinthepastyears;GaN,however,gained
specialattentionduetothepredictedpossibleroomtemperatureferromagnetism,
whendopedwithmagneticelements[16].AlthoughCuisanon-magneticelement,
Wuetal.[17]theoreticallypredictedferromagnetismaboveroomtemperaturein
Cu-dopedGaN,whichwaslaterconfirmedexperimentallyinanumberofreports
onCu-dopedGaN,e.g.[18–20].
TheexactmechanismgoverningferromagneticbehaviorinCu-dopedGaNandDMS
ingeneralisstillunclear.Clarificationofthemany,partlyseeminglycontradicting
experimentalresults,isnecessaryforunderstandingthesituationinthisnewclass
ofmaterials.Oneverybasicquestiontobeansweredinthiscontextisthelocation
ofCuintheGaNmaterial,i.e.howandwhereCuatomsareincorporatedinthe
GaNcrystal.Thisissueisaddressedinthisworkwiththehelpofx-rayabsorp-
tionexperiments,whichwereperformedattheEuropeanSynchrotronRadiation
Facility(ESRF).Theresultsoftheanalysisoftheexperimentaldataarecompared
tocalculationsperformedforCu-dopedGaNusingacommercialsoftwarepackage
]).21[(FDMNESTheoutlineofthisthesisisasfollows.Therelevantpartsofthefieldofspintronics
tothisworkarediscussedinchapter2.Hereinanintroductiontothebasicideasof
spintronics,thecurrentstatusofDMSresearchandtheroleofthemostimportant
GaNDMS,includingCu-dopedGaNarepresented.Thesubsequentthirdchapter

5

givesanoverviewofimportantpropertiesandpeculiaritiesofthematerialofcentral
interest,namelyGaN.Thisincludesadescriptionofitscrystalstructure,theim-
portantroleofinternalelectricfieldsandhowtheycanbeavoidedinGaN.Details
onhowGaNcanbefabricatedinthelaboratoryandadescriptionofthemachine
usedforproducingGaNsamplesinthisworkisprovidedintheconcludingsection
.3.3.2

βGro-LiGaOwthof2isnon-pdescribolaredGaNincbyhapterP4AMBEonaccompanied(100)γby-LiAlOan2,analysis(100)ofβthe-LiGaOpro2ducedandGaN(010)
offilms.identifyingChapterthe5presenplacementsthetofexpCuinerimenGaNtalandduringPtheoreticalAMBEgroinvwthofestigationsCu-dopandedresultsGaN.

2Ashortintroductiontospintronics

Theaimofthischapteristogivethereaderabriefintroductiontothesubject
ofspintronicsandtobuildabridgeofthisfieldtotheuseofgroupIII-nitrides
asaconvenientandadequatematerialsystemforthebasisofspintronicdevices.
Insection2.1theverybasicprinciplesofspintronicsareintroducedandthefield
isnarroweddowntospintronicsinsemiconductorsandtheimportantcategoryof
dilutemagneticsemiconductors(DMS).Thesubsequentsection2.2willoutlinethe
currentstatusofworkthathasbeendoneinthisfield.Concludingthischapter
thespecificadvancesandtheroleofnitridesinspintronicsandspecificallythe
introductionofCuasanon-magneticimpuritymakingGaNferromagneticwillbe
.2.3sectionindiscussed

2.1Thebasicideaofspintronics

Spintronicsisatermbasicallyputtogetheroutoftwoparts:spinandelectronics.
Itsmeaningliesexactlyinthisconnection,namelythatnotonlythechargebutalso
thespinofelectricalchargecarriersisusedintransportinginformation.Agreat
achievementinthefieldofspintronicswasthediscoveryofthegiantmagnetore-
sistance(GMR)inmagneticlayersofmetals[2,22].TheGMReffectnowadays
usedineverydaylife(e.g.datastorageincomputerharddiscs)wasconsideredof
suchimportance,thattheNobelPrizewasawardedforitsdiscoveryin2007.In
thiswork,however,theusageofadifferentclassofmaterialsinspintronicsshallbe
focusedon:theclassofsemiconductors.
Thediscoveryandresearchonsemiconductorshavemade1possibleahugenumber
ofdevicesandhaverevolutionizedourtechnologicalworld.Especiallytheoptical
propertiescombinedwiththeelectricbehaviorofsuchmaterialsareofgreatinterest
andhavebeenappliedinlightemittingdiodes,solarcells,laserdiodes,quantum

1Interestingly,inthebeginningsofsemiconductorresearchtheenormousimpactofthisfieldcould
notatallbeanticipated.Inthiscontext,afamousstatementofWolfgangPauliinalettertohis
assistantatthetime,RudolfPeierls,in1931iscited,wherehewrites:ӬUberHalbleitersollteman
nichtarbeiten,dasisteineSchweinerei,werweiss,obes¨uberhauptHalbleitergibt.”whichloosely
translatedmeans:”Oneshouldnotworkonsemiconductors,thatisamess,whoknowswhether
therearesemiconductorsatall.”[23]

7

Figure2.1:SchematicoriginaldesignofaspinFETfrom[25].InaDMSbased
spinFETtheironelectrodesarereplacedbyamagneticsemiconductor,
polarizingtheelectronswhichareinjectedintothechannel.Thespin
characterofthechargecarriersinthechannelcanbealteredbyagate
voltagetoturnthecurrentonoroff.

cascadelasers,phototransistors,photomultipliers,charge-coupledimagingdevices
andopticalfibercommunication.Theadvantagesofspintransportadditionally
tofasterorinsteadinformationofchargeprocessingtranspatortlesspromiposweesrtheconsumptionimprovemen[3],tofbutalsoexistingleadsdevicestonewfor
deviceconceptslikespinFETs,spinLEDs,spintunnelingjunctionsandmanymore
[3,4,24].
Itwasanoveldesignforaspintransistorusingthespinofthecarrierstomodulate
theelectriccurrent,proposedbyDattaandDas[25]in1990,thatkickedoffa
newmotivationforthesearchofadequatematerialswithpropertiessuitablefor
spinpsemiconductorolarizedspinelectrons,tronicfromaapplications.spinaligningTheideasourceofelectroDattade,andcouldDasbedescribinjectededinthatto
asemiconductorchannelandreadoutbyamagneticallypolarizeddrainelectrode.
theFig.2.1magneticshowsascmomentshematicofedesignlectronsofaandspindrainFET.areWhilealigned,thecurrenthetiscurrentmaximaliscwhhangeden
byinfluencingtheelectronspininthechannelbyapplyingagatevoltage.The
interactionoftheelectricfieldfromthegateontheelectrons’spinsinthechannel
happensviatheRashbaspin-orbitcoupling[25,26].Theresultingmisalignmentof
intheacshhargeutdowncarriers’ofthespincurrenwitht.respTheecttobiastheneededmagneticforaorieshnutdotationwnofisthepredicteddraintoresultsbe
muchlowerthanthatrequiredforaconventionalchargecontrolledFET.
Now,thequestiontobeansweredforaccomplishingspintronicsinsemiconductors,
is:Whicharetherequirementsforamaterialforsuchfunctionaldevices?Inthecase
ofaspinFET,thespinoftheelectronsmustbealigned(needforroomtemperature
ferromagneticmaterialwithhighspinpolarizationofthecarriers),thenefficiently
injectedintoamaterial(lowresistanceattheinterface),transportedwiththeability
tocontrolthespin(thespinmustbecarriedbychargecarrierswithlongcoherence

8

Figure2.2:SchematicofaspinLEDbasedonGaN.Efficientspininjectioncanbe
obtainedbychoosingaGaNbasedDMSforspininjectore.g.Cu-doped
GaN.

time)andfinallythespinaswellasthecurrentmustbedetected.Thefirststeps,
i.e.thespinalignmentandinjectionintoasemiconductormaterial,arenecessary
formanydevices.InthecaseofaspinLEDthelaststepofspinandcurrent
pdetectionolarizedisphotons.replacedbEfficienythetwaysradiativofeprorecomvidingforbinationthisofaretheelectrousenofholepairs,heterostrucyieldingtures
ofsemiconductorssuchasmultiplequantumwells(MQW)orquantumdots(QD)
an(seeaFig.ppropriate2.2).Thismaterialworkforaimsspintotaccarriagekletheandquestionradiativofeantransefficienitiontforthespin-alignerbasisandof
LEDs.spinelikdevices

Fortheinjectionofspin-polarizedcarriersintoasemiconductoraferromagnetic
metalcouldbeused.Thesuccessofsuchasystemwashoweversuppressed,two
ofthereasonsbeingthemismatchinconductivitiesbetweenthemagnetandthe
semiconductorandtherelativelylowelectronspinpolarizationofthemetallicfer-
romagnet[27].Calculationshaveshown,thatonlyabout0.1%ofthecurrentina
2Delectrongascontactedbytwoferromagneticmetalsarespin-polarized,whilefer-
romagneticsemiconductorswouldovercometheproblemoflossofspin-polarization
duetothegiantZeemansplittingwhichwouldforcethecurrentcarryingelectronsto
aligntheirspinstothelowerZeemanlevel[15].Hence,theapproachofusingmag-
neticsemiconductors,promisinggreatersuccessandbeingmorepractical,yieldinga
higherefficiencyandlowerresistancebarrier,i.e.betterconductivity,betterFermi

9

levelmatchingandbetterproductionpossibilitiesforcleanersurfacesisdiscussed
inliterature.Here,thedilutedmagneticsemiconductors(DMS)havereceivedthe
mostinterestbecauseoftheireasyimplementationintheexistingsemiconductor
technologyandtheprospectofusingonebasismaterialsystemforthewholede-
vice,whereelectronictransitionsfromonematerialcomponenttothenextwould
preservetheirspin-polarizationtoalargeextent.
Dilutemagneticsemiconductors,initiallycalledsemimagneticsemiconductors[28],
aresemiconductingmaterialsthataredilutedwithmagneticionsoftransitionmet-
als(Sc,Ti,V,Cr,Mn,Fe,Co,Ni;thecaseofCuisdiscussedinmoredetailin
section2.3)orrareearths(e.g.Eu,Gd,Er)substitutingcationsofthehostma-
terialthusrenderingthecompletesemiconductorferromagnetic.Amongdifferent
desirablepropertiesforthesematerialsoneofthemainpointsforprospectivede-
viceapplicationintheabsenceofanexternalmagneticfieldisthenecessityfora
highCurietemperature(TC)inferromagneticDMS,namelyatroomtemperature
e.voaborSpininjectionfromaDMSintoanon-magneticsemiconductorhasalreadybeen
showninafewmaterialsystemsandisinvestigatedextensively.Azero-magnetic
fieldelectricalspininjectioninanall-GaAsbasedspinLED,usingferromagnetic
GaMnAsasspin-aligner,hasbeenreportedbyOhnoetal.[29],injectingspinpolar-
izedholesintotheferromagneticstructure.AnearlyreportonBexMnyZn1−x−ySe
asaspin-alignertoinjectspin-polarizedelectronsintoaGaAs/AlGaAsMQWLED
structurehasshownaninjectionefficiencyof90%[30].ZnMnSeisknowntoexistin
paramagnetic,spinglassorantiferromagneticphases,dependingontheMnconcen-
trationandthetemperature[31,32]andspininjectionexperimentsthereforeneed
thepresenceofahighexternalmagneticfieldandlowtemperature.Ithasbeenre-
portedthattheelectronconductivityisimportantforspinpolarizationinn-typeZn-
MnSe[33],whichiswhyCldopingtocontroltheelectrondensityisusedfrequently
inZnMnSe.ThebenefitsoftheZnMnSesystem,however,havebeendemonstrated
byanalmostperfectspininjectioninitializationintoseveraldifferentInAs/GaAs
quantumdots(QD)simultaneously[34].ForQDensembles,uptoabout60%spin
polarizationhasbeenachievedinInGaAsQDs[35]whereZn0.95Mn0.05S0.1Se0.9:Cl2
wasusedasaspin-aligner;InGaAsQDsexcitedbyshortelectricalpulsesinthe
nanosecondrangethroughtheDMSZn0.95Mn0.05Se:Clshowaspin-polarizationde-
greeofupto41%[36].Also,spin-injectionthroughZn0.80Mn0.20SeintoCdSQDs
hasbeenreportedbyopticalmeanswhichshowsaninjectedspinpolarizationof
].37[32%AlthoughinvestigationsonDMSalreadygobacksome30years,theinterestinthis
fieldcontinuestopersistandisenormous.Tounderlinethisstatementthenumbers
2TheadmixtureofSulfurintendstoimprovethematchingofthelatticeconstantsbetweenZnMnSe
andGaAsforbetterepitaxialgrowth.

10

ofcitationsonimportantpapersishuge.Untiltoday3forexample,theDatta&
Daspaper[25]onthespinFEThasbeencited2065times(193timesintheyear
2010);thepaperofDietletal.[16]aboutthemeanfieldZenermodelformagnetic
semiconductors(discussedbelow)wasreferenced3678times(438timesin2010)and
ashortreviewbyOhno[38]onthistopic2492times(213timesin2010).

2.2Currentstatusondilutemagnetic
(DMS)rssemiconducto

Whileintheorythequestionoftheusefulnessofmagneticsemiconductorsseems
undoubtedlyclear,theexperimentalrealizationofdeviceslagsbehind.Thisiscon-
sideredtobeareasonoffindingtherightmaterialthatcanbealteredinsucha
way,thatthedesiredpropertiesappear.Aboveall,overcomingthelowCurietem-
peratureofmostmaterialsystemswasconsideredachallenge.Besidesexperimental
effortstofindingthebestDMSwhichwillbebrieflysummarizedfirstinthissection,
anterials,outlineofresultingtheoreticinthealpmoersistendelstappliednotiontotopredicimprotvtheeourpropknoertieswledgeofofapproptheriatephysicsma-
involvedandpushingforwardinthedirectionofspintronicdevices,willfollow.
scienAlreadytistsinintheexclatehanging1970’sparttheofideatheofmmetalakingatomasinasemiconductorcompoundmagneticsemiconductorengaged
withmagneticatoms.Biginterestinthesestructuresstartedtorise,whenan
anomalouslylargespinsplittingwasobservedinmagnetotransmissionexperiments
bonetwHg1een−xMnthexTfreee[39c],hargewhichwcarriersasinandterpretedthelotocalizedoriginfromd-electronstheexcofthehangeMninions.teractionA
largeFaradayrotationandaMagneticCircularDichroism(MCD)signalmeasured
inCd1−xMnxTe[40]wasattributedtothesamemechanismandsubsequentlythe
emerged.DMSoffieldTheseearlyexamplesofDMS,nowsometimestermed”traditional”DMS[41],be-
longtothefamilyofA1II−xMnxBVI,whereAIIandBVIstandforCd,Zn,HgandS,
Se,Te,respectively[28].However,thelowcriticaltemperaturesinthesematerials
[42],makingthemimpracticalforspintronicdevicesledtoaquickspreadofthefield
tonewmaterialsystemslikeA1II−xFexBVIorA1II−xCoxBVI[42,43].Rejuvenationin
atDMSrelativresearcelyhhighwasthtempeoperatureseningofwasatonewbematerialobserved:systemtheIIinI-Vwhichsemiconductors.ferromagnetismA
firstrromagnetismeportof[44]MBEwasgrofollownwIned1−bxyMnxmanAsyonInAspublicationsandonGaAsIII-VandsubstratesII-VIshowingmaterials,fer-

3ThecitationnumbersaretakenfromtheISIWebofKnowledgeonthe12.April,2011usingthe
WebofSciencedatabasewithConferenceProceedings.

11

Figure2.3:IllustrationoftheRKKYinteractioninasemiconductor(GaMnAs)
andametal(Cu:Mn)adaptedfrom[61].Theoscillatorybehaviorof
tthewoMnspinions.exchangeThegraphconstanwtasiscasholculatedwnasausinganfunctionelectronoftheconcendistancetrationof
of8.5×1022cm−3inthecaseofCuandaholeconcentrationinof
3.5×1020cm−3forGaMnAs.Thegraphshowsthatalongrange
ferromagneticcouplingispossibleinthesemiconductormaterial.

likeGaMnAs[45],GaN,ZnO,ZnSeandalargenumberofreviewarticles[3,24,
46–52]describingtherapidgrowthofthefield.
TheneedforhighCurietemperatures(TC)inDMShaskeptthesearchfornew
materialsactiveandindeedabigimprovementcouldbemadeforexampleontheTC
reportedforInMnAsandGaMnAs,whichcouldberaisedfrominitiallybelow10K
[53]and110K[38]to343K[54–56]and185K[57],respectively.ForGaMnAsthis
wasrelatedtotheprogressincontrollingcrystalquality,namelyreducingthenumber
ofunintentionalchargeandmomentcompensatingdefectsthroughoptimizedgrowth
andpost-growthannealingprocedures[58].AmongthehighestTCvaluesreported
forDMSaretheonesforGaMnN(TC=940K[59])andZnCoO(TC=790K[60]),i.e.
theyarewellaboveroomtemperature.However,itisstillunclearwhatthedriving
forcesbehindtheferromagneticbehaviorare.Thisleadsustoaveryimportant
questioninthisfield,theanswerofwhichisessentialforreliablycontrollingthe
magneticpropertiesinDMS:WhatmechanismmakesaDMSwork?
Therehavebeenmanyattemptstodescribethemechanismresponsibleforferro-
magnetisminsemiconductors.Todate,thesituationisstillnotclearandnosingle
theoryexiststoexplaintheoriginandthelargescatterofdataofthemagnetic
propertiesofsemiconductors.ThreemodelsofferromagneticmechanismsinDMS
discussedinliteraturewillbepresentedinthefollowing.

12

ThesuccessofthemeanfieldZenermodelintroducedbyDietletal.[16]wasfounded
ontheabilitytoaccountfortheexperimentallyobservedCurietemperaturesof
Ga1−xMnxAs4andZn1−xMnxTe.Thescientistsappliedthismodeltoavarietyof
DMStopredicttheirCurietemperature,aswillbediscussedinsection2.3.The
basicideaofthismodelisthatferromagneticinteractionbetweenlocalizedspins
oftransitionmetal(TM)ionsismediatedviaholesintroducedbyshallowacceptor
levels,providedbytheMnions.Thedevelopedtheoryisbasedonthedouble-
exchange(Zenermodel)[62,63]andtheRuderman-Kittel-Kasuya-Yoshida(RKKY)
[64–66]mechanismtoexplainthelongrangemagneticcouplingofthemoments.
Doubleexchangeisthetermusedtodescribetheferromagneticcouplingbetween
twodifferentlychargedtransitionmetalions(e.g.Mn3+andMn4+)separatedbyan
atomwithaclosedshell(e.g.Cl−orO2−).Thisisdonebythetransferofanelectron
fromoneMniontothecentralatomandasimultaneoustransferfromthecentral
atomtothenextMnion,wherethelowestenergyofthesystemcorrespondstoa
parallelalignmentofthespinsofthetransitionmetalions[63],henceaferromagnetic
couplingisestablished.Theexplanationofthelongrangecouplingisthengivenby
theRKKYinteraction,wherethespinorientationoffreeelectronsisinfluencedby
thelocalmagnetizationofthemagneticionsthroughthecombinedeffectofcoulomb-
andspindependentexchange-screening,originatingfromthePauliprinciple.This
so-calledindirectsuper-exchangeofmagneticmomentsismediatedbyconduction
carriers,whichexperienceanoscillatoryspinpolarization,dependingonthedistance
betweenthemagneticionsinthecrystal.Fig.2.3,wheretheexchangecoupling
constantisplottedasafunctionoftheseparationofMnions,demonstratesthe
longrangeferromagneticinteractioninthesemiconductorGaMnAs;butalsoshows
thatthecouplingismuchweakerthaninadopedmetalsuchasCu:Mn.Amajor
prerequisitetothiskindofinteractionisalargecarrierconcentrationandahigh
dopingconcentrationinsemiconductorswhichisnotnecessarilythecaseintheDMS
sampleswhichhavebeenfabricated.TheobservationofferromagnetisminDMS
withratherlowdopingconcentrationthereforedemandedforthedevelopmentof
dels.moealternativOnesuchmodel,alsoaccountingforferromagnetisminlowdopedp-andn-type
DMS,wasputforwardbyCoeyetal.[67],explainingthatferromagneticcoupling
ismediatedbyelectronsfromshallowdonorstates,whichcreateaspin-splitim-
puritybandwhensurpassingadonordefectconcentrationthresholdandcoupleto
the3dmomentsofTMtoformboundmagneticpolarons(BMP).Thesepolarons
themselvescoupletoeachotherleadingtoaferromagneticinteractionbetweenthe
magneticionsandwhenthepolaronpercolationthresholdisreachedthewholesam-
plebecomesferromagnetic.AnillustrationofthismechanismisseeninFig.2.4.
TheauthorsstatethattheCurietemperaturedependsonboththemagneticimpu-
rityandthedonorconcentrationinthematerial.Thismodelisalsosupportedby
4MeasuredTC=110K,calculatedTC=120K.

13

Figure2.4:Representationofmagneticpolarons.Smallgraycirclesrepresentcation
sites.Anionsitesarenotshown;anionvacanciesareindicatedby
squares.Blackcirclesstandforimpurity(dopant)atoms.Adonor
electroninitshydrogenicorbit(largebluecircles)couplesantiparallel
toimpuritieswitha3dshellthatishalf-fullormorethanhalf-full.
].67[romF

experimentalresults,e.g.[60],wherechargecarriermediatedmagneticcouplingis
ruledoutandthepredictedTCbythistheoryliesinthevicinityofthemeasured
5.one[68],AnotherwhomoassumedelofaconferromagnetismtributiontointheIII-VindirectmaterialsexchangewasinpropteractionosedbybypLitvinovolarizationetal.
oftheacceptorvalencestatestobandthevelectronsalenceviaband.virtualThismoelectrondelexcitaexplainstionthefromoriginofmagneticferromagneticimpurity
behaviorforhighandlowcarrierconcentrationbyuseofapercolationapproach
(likeinthecaseofBMPs[67])ratherthanameanfieldtheory(likeinthecaseof
themeanfieldZenermodel[16]).
DespitethepersistenteffortstoclarifythepictureofferromagnetisminDMSalsoon
thepuzzlingexperimenresultstalaresideandobtained.togivForeinputexample,tothenoneparoftheameterstwousedoppinosingthemomodelsdels,relyingrather

5ThepredictedTCinthisstudywas822KwhiletheexperimentgaveTC∼790K.

14

Figure2.5:RoomtemperaturemagnetizationforZnOfilmswith5%Coandvarying
Aldoping,asafunctionofthemobilecarrierdensity.Thegraphis
].75[fromadapted

oncarriermediatedinteractionorapercolationmechanismcanexperimentallybe
evidenced.Whilesomereportsseenocorrelationbetweencarrierconcentration
andferromagneticbehaviorinDMS[69–71],leadingtotheconclusionthatfree
carriermediatedmechanismsdonotapply,otherpublicationsshowanincreaseof
ferromagnetismwithincreasingfreecarrierconcentration[72–74],indicatingthe
oppositeconclusion.Interestingly,acombinationofbothmechanismsinTM-doped
ZnOwassuggestedbyBehanetal.[75],sayingthatthecarriermediatedexchange
canaccountforferromagnetisminsampleswithhighfree-carrierconcentration,
whiletheBMPmodelcanbeappliedforsampleswithlowfree-carrierconcentration.
Theobservedparamagneticbehaviorintheintermediateconcentrationisexplained
bythedestructionofpolaroncouplingbyhoppingoffreecarriers.Fig.2.5illustrates
thisbehaviorinBehanetal.’s[75]dataofthesamplemagnetizationasafunction
ofthefreecarrierconcentration.
Inaddition,amoreandmoreemerginginfluenceofferromagneticbehaviorinDMS
wasattributedtotheroleofstructuraldefectsinsemiconductors,boththeoretically
[76,77]andexperimentally[19,70,71,78–80].However,exactmechanismsstill
remainunclear.AdiscussionofdefectsrelatedtoferromagnetisminMn-,Gd-and
Cu-dopedGaNispresentedinthefollowingsection.

15

Table2.1:CalculatedCurietemperaturesofzinc-blende(exceptwherenotedoth-
erwise)p-typesemiconductorscontaining5%Mnand3.5×1020holes
percm3[16].GaNisthecandidatewiththehighestexpectedTC.
CurieSemiconductor[K]eraturetemp120Si72Ge212AlP121AlAsGaNGaN(wurtzite)(zinc-blende)400426
101GaP107GaAs42GaSb62InP44InAs315(wurtzite)ZnO44ZnSe31eZnT

2.3Theroleofnitridesinspintronics

GroupIII-Nitridesareagroupofsemiconductorswithamultitudeofbeneficial
properties.Theseincludehighchemical,thermalandmechanicalstability,thefact
avthatailable,noneoofwingthetocomptheonhighentsareabundtoxicanceandofatnitrogenleastoneonofourtheplanet.compTheonentswideistunreadilying
abilityoftheirdirectbandgapbetween0.7eVforInNover3.4eVforGaNupto
ha6.2veeVforthereforeAlNbmakeenesubthesejecttonitridesgreatavineryterestattractivbotheformaterialresearchforandopticalfordevicescommercialand
oses.purpInspiteoftheadvantageouspropertiesofnitrides,theywerenotamongthefirst
inthissemiconductorsmaterialtohowbeevercameattemptedabtooutbewhenmadeDietletal.ferromagnetic.[16]setAnuptheirincreasedintheoreticalterest
modelofferromagnetisminDMSbasedontheZenermodel[62,63]andcalculated
Curietemperaturesthatcouldbeachievedinvarioussemiconductors(seeTable2.1).
andTheirferromagnetismextraordinaryforresultsGaNabindicatedoverotheomhighesttempTCerature!ofallInthetheirconsideredpapertheymaterialsshow
thatthemeanfieldZenermodel,usuallyappliedtodescribeferromagnetismin
andtransitionZnMnTme,etals,andisableappliedtothisexplainmodelthetoexptheerimendifferentaltCurieDMStemplistedineraturesTableof2.1.GaMnAs

16

Ofcoursethepictureisnotaseasyasitseems,for,althoughproductionpossibilities
forhighqualityGaNhavegreatlyimprovedovertheyears,theprerequisitesforroom
temperatureferromagnetismarenoteasilymatched.Ifferromagneticcouplingis
indeedmediatedbyfreechargecarriers,asthemeanfieldZenermodelofDietlet
al.suggests,oneofthemaindifficultiesforGaNDMSmaylieintheachievementof
thenecessaryhighholeconcentrationandmobility.GaNisknowntobeintrinsically
n-type[81],mostlyduetoGavacancies[5,pp.929],andovercomingthedifficultyof
efficientlymakingGaNp-typeissometimesevenseenas”theHolyGrail”inGaN
deviceproduction[82].
Anothercriticalaspectforspintronicdevicesistheelectronspinrelaxationtime
inthespintransportmaterial.Motivatedbytheoreticallypredictedlongspinlife
timesinbulkGaNtobeinthemicrosecondrange[83],whichwasconsistentwith
theexpectationoflongspinrelaxationtimesinamaterialwithweakspin-orbit
interactionandawidebandgap[84],GaNseemedtofulfillverypromisingproperties
foraspintransportmaterial.Experimentalinvestigationsusingspin-dependent
pumpandprobereflectancemeasurementhowevershowedaveryfastspinrelaxation
inbulkGaN[85](470fsat150Kand250fsat225K)andalsobulkInGaN[86],
wherenopolarizationoftheelectronscouldbeobserved,limitedbyatimeresolution
setup.theoffs)(350AfirstreportoftimeresolvedphotoluminescencemeasurementsonanInGaNmul-
tiplequantumwell(MQW)structurebyJulieretal.[87]statesarelativelylong
spinrelaxationtimeof∼100psat10K,whilesubsequentanalysesofdifferent
InGaN/GaNMQWstructuresforspinLEDapplicationreportedonverymuch
shorterspinrelaxationtimesmeasuredbycircularlypolarizedpump-probespec-
troscopy[84,86,88–90],i.e.200psat200Kand70psat295K[91],forother
sampleseven450fsat5K[84].Thesefastrelaxationtimesarebelievedtoresult
fromconsiderablebiaxialstrainattheheterostructureinterfacesleadingtolarge
piezoelectricfieldsinwurtziteGaNgrownalongthec-axis,distortingtheQWpo-
tential.Thetherebybreakingoftheconfiningsymmetryofthequantumwellleads
toalargeRashbaspin-orbitsplitting[90]causingamixingofthespinstatesinthe
conductionbandandhencefastspinrelaxation.
Substantiallylongerspin-relaxationtimesareexpected[4]forstructureslackingthe
abovedescribedinternalelectricfields,asinforexamplecubicsymmetrylattices.
However,thecubicsymmetriccrystalstructureofGaNisonlyametastablephase
andcannoteasilybeobtainedinhighphasepurity.Anotherwaytoavoidthe
electronicbandbendingcausedbytheinternalelectricfieldsistotiltthepolaraxis
ofthewurtzitematerialinsuchaway,thatitliesparalleltothegrowthplanerather
thaningrowthdirection.Thiscanbeachievedbygrowthontheso-calledM-and
A-planesofGaN,aswillbedescribedinmoredetailinchapter3.2.

17

AveryinterestingobservationinthiscontextwasmadebyNagaharaetal.[91,92]
who(inadditiontoashortspinlifetimeinlowIn(7.1%)contentMQWsamples)
measuredlong-lived(220±40ps)andtemperatureindependentelectronspinstates
inhighInfraction(10.6%)samples.Thisresultwasattributedtotheformation
ofInquantumdots(QD)asaconsequenceofthecompositionalfluctuationsinthe
InGaNQWs,meaningthatanInphaseseparationledtoInQDsimbeddedinthe
MQWstructure.DuetotheconfiningpropertiesofQDscausingadiscretedensity
ofstatesandthelackofacontinuumofstatesavailableforspinrelaxation,the
scatteringprocessesinQDsaresuppressedandalongerspinlifetimeisexpected
comparedtoQWstructures[93].Further,asdiscussedinsection3.2theeffects
ofinternalelectricfieldsinnitridesarelesspronouncedinsmallerstructures,asis
demonstratedbyinvestigationsofsampleswithdifferentQWthicknesses,e.g.in
].95,94[Fueledbytheprospectsofanall-GaNbasedspintronicdeviceandthepredicted
roomtemperatureferromagnetism,alotofresearchwasdirectedintoGaN-based
DMS.Themajorityofwork,doneinthisfield,investigatedferromagnetisminMn-
dopedGaN,butotherdopingmaterialssuchasCr,Fe,Co,GdandCuwerealso
considered.Inthefollowingashortaccountofresearchresultswillbegivenforthe
exemplarycasesofMn,GdandCudopedGaN.Indeed,highCurietemperatures
werefoundfordifferentdopingconcentrationsforsamplesproducedeitherbydirect
incorporationofthedopantduringagrowthprocesslikeMBEorionimplanted
samples,wherethedopantswereintroducedintoabulksample.

GaNedMn-dop2.3.1

ForMBEgrownGaN:MnsamplesaverylargescatterofCurietemperatureswas
foundrangingfrom10-25K6[96]to”TChigherthanroomtemperature7”[97]
upto940K[59].ThehighestreportedTCof940Kwasestimated8bySonoda
etal.[al.98][59]foundfortheir,ferromagnetismprobablyinn-typ13.7%e,9Ga0.Mn-dop91Mned0.09NGaNsampluptoe.∼750SimilarlyK,,butDharseeanet
antiferromagneticbehaviorinahomogeneous7.6%Mn-dopedGaNsample,ascribed
toaMn-MninteractionintheinsulatingGaMnNsample.

6andSampleawithhomogeneous7%Mnincorpfraction,orationofmeasuredMn,byrelyingAugeronXRDElectronSpmeasuremeectroscnopts.y(AES)depthprofiling,
7Fconortentsamples3-9%inv(measuredestigatedbybyAES).SupNoerconductingMnrelatedQuanclusterstumInwereterferencefoundbyDeviceXRDor(SQUID)XTEM.withMn
8ThetemperaturelimitoftheSQUIDsetup,usedtomeasurethemagnetizationofthesamplewas
K.7509MeasuredbySecondaryIonMassSpectroscopy(SIMS).

18

ThehugerangeinTCofMBEgrownGaMnNcanpossiblybetracedbacktothe
presenceofnanoscaleMn-clustersinsidesomeoftheGaNsamples.Thereishowever
acontroversytothismatter,namelywhileDharetal.ascribetheferromagnetic
behavioratthehighesttemperaturestotheformationofminisculeMn-richclusters
formedduringtheMBEgrowththataretoosmalltobedetectedinX-rayDiffraction
(XRD),Sonodaetal.denytheexistenceofMnclustersrelyingonXRD[59],
RutherfordBackScattering(RBS)andExtendedX-rayAbsorptionFineStructure
(EXAFS)[99]measurements.WhileDharetal.mentionthatMnGaphasescan
reachCurietemperaturesgreaterthan600◦(TC=748KreportedforMn66.7Ga33.3
[100])andthatthenitridecompoundMn4Nisferrimagnetic[101]uptoTC=738K
[102],theSonodacollaborationclaimsthatthehighTCvalueof940Kcannot
beexplainedbysuchclusters.Tofurtherinvestigateandclarifythefactsinthis
discussionAndoetal.[103]investigatedoneofSonodaetal.’ssamples,namely
sampleBofref.[59]10with6.8%Mncontent,whichalsoshowedferromagnetismat
atemperaturehigherthanroomtemperature[103],byMagneticCircularDichroism
(MCD).Thesampleonlyshowedaparamagneticbehavioranditisconcluded,that
theobservedferromagneticbehaviormustbeassignedtoanunidentifiedmaterial,
alsointhefilm,butundetectablebyXRD.Thediscussionclearlyshowstheneed
forabigdiversityofexperimentalmethodstotackletheintricatequestionofwhat
isgoingoninthesematerials.
CommontoalltheabovementionedreportsisadiscussionaboutMnclustersgiving
risetothehighCurietemperatures.Almostallgroupsstatethatsuchsecond-
phaseinclusionscouldnotbefoundbyXRDorsometimesevenbyTEM.However,
asstatedin[98],[103],[104]and[51]thesizeofMnclusterscanbetoosmall
tobedetectedbythesemethods.Further,themagneticcharacterizationofthe
sampleswasmainlyperformedbySQUID,theVibratingSampleMagnetometer
(VSM)orusingtheAnomalousHallEffect(AHE).Theseareallmethodsprobing
thetotalmagnetizationinthesample,asopposedtoelementspecificmethods,where
clusteringeffectscanbedifferentiated.
GroupsusingotherproductionmethodsforMn-dopedGaNfindCurietemperatures
of250K[105]and300K[106]11forionimplantedsamplesand228-370K[107]
forMOCVDgrownGaNsampleswithpostgrowthdopingbysolidstatediffusion
fromaMnsource.Itisstatedinthesereports(e.g.[107])thatthemagneticbe-
haviordependsstronglyonthegrowthconditions,post-growthproceduresand,as
canbeseenbythespreadofexperimentaldata,theyalsodependonthefabrication
method.Especiallytheeffectofpost-growthannealingatdifferenttemperatures
influencesthemagneticpropertiesofthesampleseverelyandraisesthequestionof
MnclusteringagainascanbeseenintheexampleofBaiketal.[104]whofound

10UnfortunatelythisisnotsampleA(9%Mn),whichshowedthehighCurietemperatureof940K.
11InthiscaseMnionswereimplantedinp-typecubicGaN.

19

ferromagneticbehaviorintheirMnimplantedGaNsamplesonlyafterannealingat
800◦C.BytheuseofSynchrotronRadiationPhotoemissionSpectroscopy(SRPES)
themagneticbehaviorwasattributedtotheformationofGa-Mnbonds,probably
aMn3Gaphase,whichcouldnotevenbe◦observedinSynchrotronXRD.Byequiv-
alentmeanstheyfoundannealingat900Ctoreducethemagneticbehaviordue
totheformationofMn-Ncompounds,suchasMn3N2andMn6N2.58,whichcouple
antiferromagnetically[102],hencereducingthemagneticmoment.
Ithasbeenproposedthatferromagneticpropertiescanbeinfluencedbytheforma-
tionofvacancyorvacancycomplexdefects[76,77]thatarealteredduringannealing.
Theexistenceofsuchcomplexdefectshasexperimentallybeenverifiedinapositron
annihilationstudyofvacancytypedefectsinMn-dopedGaNgrownbyMOCVD
[108]whereanewkindofdefect,identifiedasaVN-MnGacomplex,wasfound.At
thisstage,itishoweveruncleartowhatextentthesedefectsplayaroleinthe
ferromagneticbehaviorofMn-dopedGaN.Thequestionaboutdefectrelatedferro-
magnetismisalsotreatede.g.inZnObasedDMS[109].Inthiswork,however,we
shallstaywiththenitridesanddiscussthisissueinGd-andCu-dopedGaNinthe
followingtwosections.
Manyquestionsconcerningthemechanismsandmicroscopicoriginsofferromag-
netisminthesenewmaterialsystemscontinuetodriveresearchfurtherandhave
ledtodifferentmethodsofinvestigationandthedevelopmentofnewideasforan
explanationoftheexperimentaldataasalsothenextexampleofGd-dopedGaN
w.showill

eGd-dop2.3.2GaNd

Asinvestigationsdigdeeperintotheunderstandingofthevaluesforthediffering
Curietemperatures,theunderstandingofmechanismsforferromagnetisminthese
systemsiscontinuouslyimproved.Toillustratethis,thecaseofGd-dopedGaN
shallbebrieflysketchedinthefollowing.
pIner2005GdDharatometinal.Gd-[69dop]repedortedGaNansamplesenormousgrownbmagneticyreactivmomenetofmolecularuptobeam4000µepi-B
taxy(RMBE).Themagneticinvestigationofthesamplesinthisworkwasdoneby
SQUIDfortemperaturesupto360Kwhereallsamplesstillshowedamagnetic
hsuredysteresisvaluecurvofe,360K.meaningPossiblethattheclustersCurieliketempGdNeratureorGdliesabcouldoveonlythemaximaccountumformea-TC
valuesupto60K[110]and293K[111],respectively.Thismaterialsystemwasthen
investigatedbyanelementspecificinvestigationmethod,namelyX-rayMagnetic
inCircularGd-dopDicedGaN.hroismThey(XMCD)found[112only],bayveryNeyweteakal.[113magnetic]topprobeolarizationtheofGdferromagnetismatoms,

20

whicidenhtifiedcouldverynotsmallexplainGdtheclustershugealreadymagneticinamomen0.05%trepGd-doportededearlier.GaNFsample,urther,whictheyh
couldobservedgiverisemagnetictothedatadetectedabove360XMCDKbysignal.SQUID.However,Hence,thisitwascannotstatedaccounthattfortherethe
mdeed,ustbetheothernotionmecthathanismsdefecintsvcolvouldedinplayleadinganimptotheortanftroleerromagneticinthebmecehahanvior.ismIn-of
ferromagnetisminthismaterialwasexpressedbyDharetal.intheconclusionof
ionsanotherwasstudyincreased[78]bywhereantheyorderoffoundmathatgnitudethewheneffectivGdeionsmagnweticeremomeimplannttedperinGdto
GaNratherthanincorporatedduringepitaxialgrowth.Sinceimplantationmeth-
odsincreasethedensityofdefectsinacrystal,themeasureddrasticincreaseof
magneticmomentperTMatom,intheas-implantedsamples,wasassumedtobe
related.defectThiskindofuncertaintyofwheretheferromagnetismreallycomesfromhasledto
asomewhatskepticalviewonthereportsofnewDMSwithroomtemperaturefer-
romagnetism,andthecallformoreexperimentalevidenceprobingdifferentaspects
ofthemagneticinteractionsinsidethematerialhasrisentoclarifytheoriginofthe
vior.ehabferromagneticHerewefindasomewhatsimilarsituationasinthecaseofMn-dopedGaN,where
somecomponentofthematerial,otherthanthesimpleinteractionofthesubstitu-
tionaldopedmagneticions,isneededfortheexplanationofthemagneticdata.One
wecouldcanslighseetlythebeeffectsremindedandofconsequthesearcences,hforbutdarkwhenmatteraskinginwhatthedarkuniverse,matterofis,whicwhe
onlyclearhaexpveerimenunconfirmedtalindicationtheoriesoforwhatcanitexpactuallyerimenis,tallyisstillexcludemissing.someofthem,buta
pIntheossibilitiesfolloofwingferrfinalomagneticpartofbehathisviorcinhapternitridesconcernedwillbewithdescribCu-doped.ItedwillGaN,bepossiblefurther
tomagneticexcludeoneclustersmaduejortosourcetheofdopant,confusionwhichisandalreadydifficultyone,ofnamelythethemainmotivformationationsof
forusingCu-dopinginDMS.

Cu-dop2.3.3GaNed

hoCuwisever,notawhenmagneticimplantedelemenontGaandmaysubstitutionalatfirstnotsitesinseemGaNtoitfulfillbtheecomesDMSacritermagneticia,
impurityowingtoitsCu2+state.IthashencebeenconsideredasadopantforGaN
DMS.Investigationsaresimplifiedbecauseofthelackofdisturbancesintheforma-
tionofmagneticclusters.NeitherCu-Cuclusters,norCu-GaorCu-Ncompounds
havebeenseentoexhibitferromagnetism,providingaverypromisingapproachfor
answeringthequestionoftheoriginofferromagnetisminDMSmaterials.

21

Figure2.6:[114GaN]forsup5.6%ercellCuwith72concenatomstration.usedPinositionsthe1andcalculations2ofoftheRosaCuetatomsal.
incorpindicateoration.the”close”Thescandphematicositionsistak1enand3fromthe[114”far”].configurationofCu

WFirstuetinal.[terest17],inwhereCu-dopfirstedGaNprinciplesasaDensitDMSywFasunctionalinitiatedbTheoryya(DFT)theoreticalcalculationsstudyof
showed100%spinpolarizationoftheconductioncarriersandanexpectedCurie
htempybridizationeratureofbet350weenK.theTheseCu3dpropandertiestheNwere2pseenstates,tobeleadingthetoaresultspinofpstrongolarizationp-d
oftheNanions,themselvescouplingtootherdopantsandhenceprovidingforan
indirectmagneticcouplingamongthedopants.

Oneyearlater,RosaandAhuja[114]revealedinDFTcalculations,thatCu-doped
GaNdoesnotshowrobustferromagnetismandisthereforenotsuitableasaspin-
alignerinspintronicapplications.These,atfirstsight,rathercontradictingtheo-
reticalresultsdemonstratedtheneedforexperimentaldataonCu-dopedGaN.In
fact,thepredictionsof[17]and[114]donotcontradicteachother.RosaandAhuja
explainthatWuetal.’sresultsarereproducedingoodagreementwhenchoosinga
largerseparationbetweentwoCuatomsintheGaNsupercell(”far”configuration),
buttheenergeticallymorefavorable”close”configurationoftwoCuatomsinGaN
leadtoasignificantdecreaseofthemagneticmomentsonboththeCuandtheN
atoms.Fig.2.6illustratesthemeaningof”far”and”close”configuration.Further,
itisstatedthattheinfluenceofdefectsinthecrystal(whichwasnotinvestigated)
couldcontributetoanenhancementorreductionofthemagneticmoments.

22

Anotherreport,usingab-initiocalculationsbasedonDFT,onCu-dopedGaN,sup-
ptheortingcrystalthefieldthesisofsplitting,astrongwhichp-dahCuionybridizationexperiencesinmoreinthedetailtetrahedralasaenconsequencevironmenoft
ofGaNwaspublishedbyLeeetal.[115].Theauthorsevaluatetheprospectsof
GaN:CuandGaN:MnassuitableDMSforspintronicpurposesandconcludethat,
becauseofamoredelocalizedcharacterofthecarriersinCu-dopedGaNresulting
essentiallyfromthesmallerdifferenceofelectronegativitiesbetweenCuandNas
comparedtoMnandN,thereisalongrangevalencebandsplittingleadingtoa
longrangemagneticinteractionbetweenthedopantsinGaN:Cu,whilethevalence
bandsplittinginGaN:Mnanditsmagneticmomentisratherconcentratedatthe
transitionelement.ThesefindingsleadtheauthorstothestatementthatGaN:Cu
fulfillsthecriteriaofasuccessfulDMSmaterialandisevenmorepromisingthan
GaN:Mn.

Indeed,aboveroomtemperature,ferromagnetismwasfoundinnumerousinvesti-
pgationsersistence[18–of20a,80,measured116,117]magneticofGaN:Cu.momentupAstoan400K,example,i.e.wellFigureabov2.7eroshoomwstem-the
perature,inasamplegrownbyplasmaassistedmolecularbeamepitaxy.Thefirst
aexpstudyerimentalwhereCuevidencewasforimplantedferromagnetismintoaGaNinbulkCu-dopedcrystalGaN,[116ho].wevAfterer,ancamenealing,from
btheehaviorsampleswasshoobservwededroinomtemsamplespwheratureichwerenotferromagnetism.annealedHoorwever,annealednoatatoferromagneticohigh
oftemptheeratureobserved(leadingferromagnetotism,clusteringbutofstateCu).sthat,Thesincestudydoannealingesnotaffectsexplainthethedamageorigin
recoters)verydonotafterconimplantributetation,tothedamageobservedintheferromagnetismsamplesoratrosecondaryomtempphaseserature.(CuItclus-is
informsterestingmagnetictonote,impurities,thatintheleadingcaseoftoanGaMnNamandbiguousGaGdNoriginofclusteringthemofagneticthesidopangnals,ts
theclusteringofCuions,ontheotherhand,seemstodiminishtheferromagnetic
behaviorofthesamples.

Shortlyafterthisreport,Seongetal.[18]publishedapaperonroomtemperature
ferromagnetismobservedinCu-dopedGaNnanowires.Theauthorsexplainthat
btheirehavior.nanoFwiresurtherarethedefectystatefreethatandCudefectsatomscantakhenceepartnotinbethetheoriginwurtziteoftheGaNmagneticlattice,
suggestingasubstitutionalsiteoccupationofCuatoms.WithXMCDmeasurements
attheL-edgeofCuamagneticmomentcouldbemeasuredattheCuions,proving
theparticipatingroleofCutotheobservationofferromagnetisminthenanowires.

Upsultstowerethispoinconsistentthet,roomsituationtempseemederaturetobevferromagnetismeryencouraging.hadbeenExpestablishederimentalandre-
ofnodisturbexplainingingeffectsferromagneticfrombeclustershaviorincouldbCu-dopeedobservGaNed.wHoaswevfounder,tothebesoundmorewcom-orld

23

Figure2.7:SQUIDmeasurementsofthemagnetizationasafunctionoftemperature
ofaCu-dopedGaNsample,showingferromagneticbehaviorwellabove
roomtemperature.Graphtakenform[117].

plicatedclaimingandthatthediscussionsoriginofabouttheferromagnetismmagneticintheseoriginsystarose,emse.g.couldbyYnotangbeetexplaial.[ned19]
bythep-dhybridizationofCuandNatoms[17],duetothelowdopantandcar-
rierconcentrationspresentintheirsamples,whichcouldthereforenotaccountfor
thelongrangeinteractionnecessaryfortheferromagneticbehavior.Yangetal.
thehadseenobservedGavmagacancneticyrelatedmomentssignalsarepinrobablytheiraPLresultspectraofdefectandinsteadrelatedargue,structures,that
sucprohvideasGaforvtheacancieslong(VrangeGa)orCu-Nferromagneticvacancyin(CuteractionGa-[76V]N)andhacomplexes,veevenwhicbeenhcouldpro-
posedargumentotswsolelyereacsuppcounortedtforbytheferromagneticfindingsofbehaMadhvioruetinal.[undop118]edwhoGaNcomm[77].unicatedThese
inanarticlethatundopedGaNnanoparticlesexhibitroomtemperatureferromag-
netism.TheauthorsascribetheferromagneticbehaviorofthepureGaNparticles
towhendefectsparticleatthesizessurface,increase.ItstatingisthatarguedthatferromagnetismtheformationanddefectenergyofdensitGayvacanciesdecrease
maybeferromagnetismdramaticallyhasnotydifferenetbteenfornanoobservedsizeinparticlesundopedthanbulkforGaN.bulkGaNandhence

Likeintheoreticalstudies,seemingcontradictionshaverisenintheexperimental
findings.Somestudiesexcludedefectrelatedferromagnetismwhileothersstrongly
pointtowardsit.Althoughtheunderstandingoftheactualmechanismsatworkis
stillincomplete,thenatureofapossibleanswertothequestionsisalreadysketched.
WhilenotgoingintodetailsaboutZnO,butjustacknowledgingthesimilarities(e.g.

24

crystalstructure,bandgap)toGaN,studiesonZnObasedDMShintstronglyto
thepresenceofamultitudeofmechanismsandoriginsofferromagnetisminDMS
].109,75[OtherstudiesofCu-dopedGaNbySunetal.[80,119,120]investigatingCuim-
plantationinnon-polarA-planeGaNgrownonR-planesapphirewithsubsequent
annealingfoundferromagneticbehavioruptoatemperatureof380K[120].They
studiedtheinfluenceoftheimplantationdoseandannealingofthesamplesand
pointoutthatCumostlyoccupiesinterstitialsitesinas-implantedsamplesbut
changestoGasubstitutionalsitesafterannealing.Inaddition,thedefectsinthe
latticeareinfluencedbytheannealingstep;thecommonassumptionisadamage
recovery.Becausenoferromagneticbehaviorwasfoundinanyoftheas-grownor
as-implantedsamplesinthesestudies[120]norhastherebeenanyreportabout
undopedbulkGaNbeingferromagnetic,itisconcludedthatonlydefectscannotbe
responsiblefortheobservedmagnetism.However,theauthorsalsoclaimthesame
argumentasYangetal.[19],namelythatthedopingconcentrationinthesamplesis
toolowtoexplaintheferromagnetismbythep-dhybridizationmodel[17]andhence
pleadforanexplanationwhereCu-defectinteractionordefectmediatedinteraction
betweentheCuionsshouldbeconsidered.
AninterestingaspectoftheworkofSunetal.[120],whichstronglyrelatestothe
topicofthisthesis,isthecomparisonofthemagneticbehaviorofCuimplanted
non-polarA-planeGaNtopolarC-planeGaN,findingthatthenon-polarsample
yieldsafourtimeshighersaturationandremanentmagnetizationthanthepolar
one.Thisisspeculativelyattributedtotheobservationofahighersubstitution
incorporationefficiencyofCuinthenon-polarGaNsample.
Recently,aninvestigationofvacancytypedefectsinCu-dopedGaN[121]gavein-
sightstotheannealingeffectsinthematerial.Sampleswereproducedbyin-diffusion
ofCufromthesurfaceintoabulkGaNwaferat873Kandwereexpectedtobe
oversaturatedwithCuwhenquenchedtoroomtemperatureafter96h.Aseriesof
samples,producedinthisway,wasannealedatdifferenttemperaturesupto850K
andinvestigatedbyvacancytypedefectsensitivepositronannihilationexperiments.
Theresearchersfound,thatvacancy-typedefectsareclearlyintroducedintothelat-
ticewhenannealinguptoatemperatureof550K,whileathighertemperaturesthe
resultsindicatetwopossibilities,namelyahealingofthedefectsoranenlargement
ofthevacanciesintolargevacancyclusters.Theobservedeffectswererelatedtothe
out-diffusionofCufromthesamples.
Inordertogettothebottomofthequestionsonthistopicitisofgreatimportanceto
clarifythemostbasicpropertiesofthesamplesinvolved.Oneofthesefundamental
propertiesisthequestionofhowandwhereCuatomsareincorporatedinthehost
GaNmatrix.ThisshallbediscussedindetailonaseriesofMBEgrownGaN:Cu
samplesinchapter5.

3PropertiesandfabricationofGaN

ToinvestigateGaNasaprospectiveDMSmaterialitisimportanttofirstgetan
understandingoftheGaNhostmaterialanditsrelatedbenefitsandissues.The
followingchaptergivesanoverviewofkeypropertiesandparametersofGaNrele-
vantforthepresentstudy,aswellasanintroductionintoitsfabricationprocessand
therelateddifficulties.StartingwiththemostbasiccharacteristicsofGaNcrystals
thefocuswillcontinueontheimportantinternalelectricfieldsinGaNinsection
3.2,theirconsequencesandhowtheycanbecircumvented.Movingontotheman-
ufacturingofGaNandsomeofitschallenges(section3.3),theimportanttopicof
substratesforGaNepitaxywillalsobediscussedinsection3.3.1.Finalizingthis
chapterwillbeashortdescriptionofthemethodofmolecularbeamepitaxy,the
materialdepositionmethodusedinthiswork.

3.1BasicparametersandcrystalstructureofGaN

TherearetwophasesofGaNwithdifferentcrystalstructures.Oneofthemis
thestablemetahexagonalstablecubicwurtzitezinc-blendephaseandallphase.propThisertieswork,discussedhoweviner,theconcenfollotrawingtesonconcernthe
C64GaNv−inP63thismclat(#186er)crystalorusingstructure.thePearsonThespacesymbolgrouphP4.ThesymmetryunitofcellwurtziticonsistscofGaNtwiso
NandtwoGaatoms;eachoftheatomsexperiencesatetrahedralbondconfiguration
withsymmetricfourofthecrystals,otherfouratomcospeciesordinates(see(a1Fig.,a2,3.1a3(a)).andFcor)thearemostdescriptionconvenofienhexagotlyusednal
andcrystalplanesaredescribedbytheBravais-Millerindicesconsistingoffourdigits
most(hkil),impwhereortanh,tkplanesandlareareshothewnwithMiller-indtheiricesBravand−ais-Milleri=h+indicesk.InandFig.the3.1name(b)forthe
thefamilyofequivalentplanes,e.g.M-planeforall{1¯100}planes.
GaNThereCisa-plane.distinctThecrystaldifferencesurfacebetweenwiththetdirectionwo[0001]directionsisalso[0001]termedand[000Ga-p¯1]ofolaritthey
orbase(+c)oftheandtetragonconstitutesarethefacingstatedownwherewardthetowthreeardbtheondsofsubstrate.GatoTheNbucaseildingisjustthe
NaturnedtomsaroundfacetheforN-psubstrate.olarityNorote,(-c),that[000it¯1],doeswhicnothismatterwhenwhettheherthreethebtopondsofsurfacethe

Figure

3.1:

(a)

hematicSc

of

linessolidkblac

the

GaN

wurtzite

crystal

structure,

outlinetheunitcell;thetetrahedral

dicatedbythegreypyramidsandthelattice

(b)wn.shoalso

discussed

planes

crystalHexagonal

and

their

els.lab

structure

of

from

].122[

symmetry

constanandats

GaN

wingsho

the

26

The

in-is

arec

most

27

ofthematerialisterminatedbyNorGaatoms;ratherthesymmetryofbonds
inthecrystalischanged.Effectsofthedifferentpolarities,suchashigherthermal
ofstabilitfilmsygrandownhighinerN-pincorpolaritoyharationvebrateseenofobservdopanedts,[123but].aInroughercurrentandpresearcoorerh,qualiteffortsy
areundertakentoexpandthesmoothfilmgrowthregimetoN-polarityinorderto
exploititsbenefits.Thisisalsodiscussedinthecontextofp-typedopinginsection
.3.3DuetothepartlyionicbondingnaturebetweenNandGa,thewurtzitestructureof
GaNispolarinthec-direction[0001],i.e.ithasasinglepolaraxisperpendicularto
theC-plane(0001).Nitrogenexhibitsahigherelectronegativitythangalliumand
thuscausesaslightdeviationoftheGaNcrystalstructurefromtheidealwurtzite
structure.Theconsequencesofthisfactisdiscussedindetailinthenextsection
).3.2(TheparameterslatticetoconstanbetsconsideredandcoeffiforcientsepitaxiaoflgrothermalwthexofpansionGaN.While(CTE)aarelargimpeortanlatticet
mismatchbetweenGaNandthesubstratecausesstressandstrainintothefilm
orevenmayleadtoanon-epitaxialdeposition.AmismatchoftheCTEmaybe
responsibleforadditionalstrainorinducecrackingoftheepitaxialfilmwhencooling
downthesamplefromgrowthtemperaturetoroomtemperature.Theirvaluesalong
withotherparameterinformationonGaNaregiveninTable3.1.
GaNisopticallyveryinterestingduetoitsdirectbandgapofEg=3.437eV[131]at
roomtemperature(290K)orEg=3.503eV[126]atlowtemperature(1.6K).This
incorrespthenearondstoultraawavioletvelengthregion.ofλHence=361.7GaNnmbandelongs353.9tothenm,respfamilyectivofelywide,whicbadhgaplies
semiconductorsandisconsideredasaconvenientmaterialforblueandgreenLED
andlaserdiode(LD)applications.

GaNinfieldsInternal3.2

Oneofthebigissuesinnitridesemiconductorsarethepolarizationeffectsthatare
inherentlypresentintheirwurtzitestructureandevokethepresenceofelectricfields
insidethesemiconductor[132,133].TheoriginofthepolarizationofGaNcanbe
dividedintotwoparts,namelythespontaneous-andthepiezoelectricpolarization.
Whilethepiezoelectricpolarizationcomesaboutduetostrainandstressinthe
crystalstructure,especiallyatheterostructureinterfaces,leadingtoadistortionof
thecrystalsymmetryandthereforeyieldinganetpolarization,thespontaneous
polarizationisalwayspresentandresultsfromthenaturaldeviationofthenitrides’
crystalstructurefromtheperfectwurtzitesymmetry.AsaresultwurtziteGaN
accommodatesapolaraxisrunningalongthe[0001]c-axis.

28Table3.1:PropertiesofGaNat300K.
ReferencealueVarameterPmcP6groupSpace3Latticeconstantsat(25◦C)[˚A]
ac3.1890(3)5.1864(2)[124]
atlinearhighT≈thermal600−expan1000sionKco[1/K]efficients
a6.20±0.4×10−−66[125]
c5.70±0.5×10
atBand1.6Kgap[eV]3.503[126]
3.437K290atMolarmass[g/mol]83.7267[5]
Density[g/cm3]6.15[127]
Meltingpointat1atm[K]2550calculated[128]
Debyetemperature(T=12-1025K)[K]
ca898868±±2420[125]
Electronaffinity[eV]4.1[127]
Young’smodulusofelasticity[GaP]210±23[129]
C11Elasticconstants[GPa]390±15[130]
CC4433105398±±1020
10123C66±CC1312106145±±2020
FordeviceapplicationslikeanLED,itisnecessarytobuildaheterostructureof
differentmaterialsontopofoneanotherinMQWorQDstructures.Insuchstruc-
turesgrownalongthec-direction,i.e.paralleltothepolaraxisofthewurtzite
crystal,thepolarizationcauseschargestobuildupattheinterfaces,whichgiverise
tointernalelectricfieldsasindicatedinFig.3.2(a),andresultingintheso-called
QuantumConfinedStarkEffect(QCSE).Theeffectoftheinternalelectricfieldon
theelectronicbandstructurecausestheelectronicbandsinthesemiconductorto
deviatefromtheflatbandstructure,whichwouldbethecaseinabsenceofelectric
fields.TheelectronicbandbendinginaGaN/InGaN/GaNstructure,ascanbe
seeninFig.3.2(b),causestheelectronsandholestospatiallyseparateintheQW
orQDleadingtoareducedwavefunctionoverlapandhencetoalowerradiative

29

Figure3.2:(a)Schematicindicationoftheinfluenceofelectricalpolarizationina
GaN/InGaN/GaNMQWstructuregrownonpolar(0001)GaN.The
totalpolarizationchangeoftheindividuallayersleadstotheaccumu-
lationofchargesattheinterfaces,evokinginternalelectrostaticfields.
PSPandPPZdenotespontaneousandpiezoelectricpolarization,re-
spectively.σ1andσ2standforthedifferentstressattheinterfaces.
(b)Calculatedbandprofilefora10-period(3nmIn0.15Ga0.85N)/(7nm
GaN)MQWindicatingtheelectron-holewavefunctions.CBandVB
aretheconductionbandandvalenceband,respectively.Figureadapted
].135[from

transitionprobability.Owingtotheprolongedradiativetransitionlifetime,nonra-
diativede-excitationchannelsgainmoreimportanceandtheprobabilityofcarriers
capturedbynonradiativecentersisincreased.Afurtherconsequenceoftheband
distortionisaredshiftinthetransitionenergycomparedtoflatbands,ascaneasily
bededucedfromgraph3.2(b).TheQCSEisalsoseentodependonthequantum
wellthicknesswhereareductionoftheradiativetransitionprobabilityandared
shiftwithincreasingQWthicknessisobserved[94,95,133,134].
Waystocircumventtheeffectsofinternalelectricfieldsthatarebuiltupathet-
erostructureinterfacesinnitrideshavebeenproposed.Onewayistomakeuseof
thecubiczinc-blendestructureofnitrides,wherethepolarizationeffectscancelout
duetotheunderlyingcrystalsymmetry1.InthisapproachhoweverGaNsamples
ofthethermodynamicallylessstablezinc-blendephaseoftenshowfractionswith

1Inzinc-blendestructurefoursymmetryequivalentpolaraxisalongthe111directionscanceleach
others’polarizationcontributions.

30

Figure3.3:(a)Schematicindicationoftheinfluenceofelectricalpolarizationin
aGaN/InGaN/GaNMQWstructuregrownonnon-polar(1¯100)GaN.
Nointernalelectrostaticfieldsarepresentalongthegrowthdirection.
(b)GaN)CalculatedMQWbandindicatingprofiletheforaelectron-hole10-periodw(3avnmInefunctions.0.15Ga0.85CBN)/(7andnmVB
aretheconductionbandandvalenceband,respectively.Figureadapted
135[from].

wurtzitesymmetry[136,137].Adifferentansatz,whichistheonethatthisthesisis
basedon,istogrowwurtziteGaNinadirectiondeviatingfromthe[0001]direction
alsoreferredtoassemi-polarornon-polarfaces.Crystaldirectionsperpendicularto
thec-axis,likethe1¯100or11¯20directions,resultinginM-orA-planeGaN,re-
spectively,aretermednon-polar,whileanyotherinclinationsofthecrystalc-axisto
thesubstratearecalledsemi-polar.Semi-polargrowthofGaNreducestheelectric
fieldsingrowthdirection,whereasinnon-polarstructures,theintrinsicspontaneous
polarizationofthecrystalliesparalleltothefilmsurfaceandnoelectricfieldsare
builtupthatdistorttheelectronicbandstructure.Further,thepiezoelectricpo-
larizationvanishes[138]forgrowthdirectionstilted90◦fromthec-axis,resulting
inflatbandsattheinterfacesofQWs.ThissituationisdepictedforanM-plane
GaN/InGaN/GaNstructureinFig.3.3,equivalenttoFig.3.2forC-planeGaN.
Indeed,theeffectsoftheQCSEcouldexperimentallybeobservedwhencomparing
structuresgrownonpolarandnon-polarplanes.InGaN/AlGaNMQWstructures
grownon6H-SiC(0001)andγ-LiAlO2(100),resultinginGaN(0001)andGaN
(1¯100),respectively,Waltereitetal.[13,139]demonstratedthatthetransition
energywassignificantlyblueshifted(by∼120meV)andthattheradiativelifetime
isroughlytentimesshorterforthenon-polarsample.InanotherstudyCravenet

31

al.[95]investigatedthePLdependenceonthequantumwellwidthofnon-polar
andpolarGaN/AlGaNMQWsandsawasignificantredshiftwithincreasingwell
width.Eventhoughthenon-polarsampleshowedahigherdislocationdensity,
itprovedtoyieldenhancedrecombinationefficiencyascomparedtotheC-plane
MQWstructure.Alltheseresultsareinverygoodagreementtowhatisexpected
theoreticallyfromtheinfluenceoftheQCSE.

3.3FabricationofGaN

icallyGalliumlargenitrideinteresthasbduringeenundertheinpastv20yestigationearsformainlymanyfueyledears,bybutitshascommercialgainedspappli-ecif-
cationammoniaprospoverectshot[140].galliumItwandaswsynasthesizedreferredfortoasthefirst”gallictimenitride”.in1932The[141fi]rstbychemicalpassing
vapordepositiononahexagonalsapphiresubstrate,toproducealargeepitaxial
lationyerproofGaNcesseswbaspecameerformemoredandin1969morebyadvMaruskancedatheandpurityTietjenleve[ls81].andAscrystalthequalitfabrica-y
increased.However,thedensityofdefectswasstillmuchlargerthaninothercom-
parablesemiconductorcompoundslikeZnSeandnotmuchhopewasgiventothis
material.AmajorcontributiontothesuccessofGaN,showingthepossibilityofef-
ficientGaNbasedLEDdevices,wasmadebyShujiNakamurawhodevelopedbright
blue[142]andgreenlightemittingdiodesandpresentedthefirstbluesemiconductor
laserbasedonGaN[140].
ForfrequenthetsynonesthesisnowofbeingGaNhmanydrideyvdepaporositionphasetechepitaxyniqueshav(HVPE),ebeenapplied,metalorganicthevapmostor
phaseepitaxy(MOVPE),metalorganicchemicalvapordeposition(MOCVD)and
themolecularstoicbhiometriceamepitaxyratiofor(MBE).GaNgroThewth,necbessecauseityofofhightheloNwvaporsolubilitypressuresofNintoGa,assurefor
aforlongantimeepitaxiallimitedprocesstheisqualitofyandfundamenavtalailabilitimpyoroftancebulktoGaN.thesuccessClearly,ofthehighsubstratequality
oftheepitaxialmainlayersreasons(seeforsectionthe3.3.1).difficultiesTheinlackofobtainingbulkGaNhighqualitsubstratesyGaNisthereforematerialoneand
necessitatedtheuseandinvestigationofalternativesubstratesforheteroepitaxial
growthofGaN.Anumberofsubstrateshavebeentriedandhavebeenevaluated
forgrowthofGaN(furtherdetailsaregiveninsection3.3.1).
Note,thatdependingonthecrystalorientationofGaNthatisaimedtobegrown;
onlyspecificcrystalplanesofdifferentmaterialscanbeconsideredforepitaxial
orgrowth.semi-polarThedevorienelopmentationstofofCGaN,-planeduetoGaNitshassimplerearlierprorootsductionthanpossibtheility,non-prisingolar
fromthefactthatC-planeGaNisthefastestgrowthdirectionandhenceisthe

32

crystalphaseeasiestobtainable,butalsoimpactedbytheavailabilityandquality
ofhexagonalsapphiresubstratesforitsgrowth.
Recently,triggeredbythegreatdemand,GaNtemplatesandbulksubstrateshave
becomecommerciallyavailable2.However,thepriceofthesesubstratesforhomoepi-
taxyisstillapproximately100timeshigherthanforasubstrateforheteroepitaxy.
Forexample,aonesidedepi-readypolishedM-planeGaNsubstrateof9×13mm
andthickness350±100µmissoldfor2553e3,whereasaonesidedepi-readysub-
strateof10×10mmand40.5mmthicknessforM-planeGaNheteroepitaxysuchas
(100)LiAlO2costs26.5e.
Furthermore,largerwafersizes(e.g.round2inchwafers)areonlyavailablefor
C-planeGaN.ThereasonbeingthatotherorientedGaNsubstratesarecutfromC-
planeorientedGaNandverythickGaNcrystalsaredifficulttoproduce.Therefore,
heteroepitaxyofGaNisstillwidelyusedandinvestigatedandwillexclusivelybe
consideredinthiswork.
OfcoursetherearealsootherissuestobedealtwithintheproductionofGaN,
especiallylookingtowarddevicerealization.Thisisforexamplep-typedopingof
GaN,whereMgorZnhavebeenconsideredtoprovideacceptorlevels.Thein-
corporationofacceptoratomsisnotthemainobstacle,thedifficultyarisesfrom
thelowholeconcentrationandthelargeresistivityinp-typematerial.Oneprob-
lemtowardslow-resistivityp-typedopingusingMgistheacceptor-compensationof
holesbye.g.atomichydrogen[144,145].ThisisespeciallycriticalifNH3isused
inthegrowthofGaNasNsupply(asisthecaseine.g.MOCVDorammonia-
MBEalsocalledRMBE).Theusageofplasma-assistedMBE,whichwasusedin
thestudiesofthiswork,beingadepositionmethodperformedinanultrahigh
vacuumenvironment,hastheadvantageofloweringthiseffecttoaminimum.An-
otherdifficulty,however,isthelargeactivationenergiesofthedopantsinGaN
(theactivationenergiesareevenhigherinAlN).TheMgrelatedacceptorlevelsin
GaNlieat∼125meV−170meV[146–151]abovethevalanceband,whereasfor
Zntheactivationenergyisevenhigher,beingaround370meV[152].Still,there
areideasandconceptstoovercometheseissuessuchasmodulationdoping[153]or
polarization-inducedholedoping[154].
TheideaofusingN-polarityinC-planeGaN,asopposedtoGa-polarity,forabetter
incorporationofdopantshasraisedinteresttoovercomecertainproblems.Forex-
ample,theabovementionedpolarization-inducedholedopingneedsN-polarityGaN
2ExamplesofcompaniessellingfreestandingbulkGaNareinAsia:Sumitomo,Nichia,Sony,
markHitacethi-cablemostlyandnon-orMitsubishisemi-polarChemical;GaNisinfoEucusedrope:onbyLumiLOG,Kyma,TInlustraopGaNTandechnologiesAmmono;anindtheOxforUSd
Instruments(TDI)[143].
3Priceinquiryonthe18.April2011atAmmono.
4Priceinquiryonthe3.May2010atCrystec.

33

andthedissociationtemperatureisincreasedforN-polarInGaN[155].However,
crystalqualityofsuchlayersisinferiortotheGa-polaritygrownfilms[123],dueto
areducedadatommobilityontheNterminatedface.Further,ithasbeenshown
thatimpurities,i.e.alsodopants,areincorporatedinhigherabundanceinsemi-
andnon-polarfilms[120,156].
Whileresearchispushinglimitsonmanyfrontiers,thisworkfocusesonthetaskof
findingpromisingsubstratesforadefectpoorgrowthofGaNwithasmoothsurface.
Inthefollowingsectiononlythesubstratesusedfornon-polarGaNgrowthwillbe
discussed.Theshortreviewofsubstrateswillbefollowedbyadescriptionofthe
MBEprocesstogrowGaNwithspecialfocusoftheMBEmachineusedinthiswork.

substrateofChoice3.3.1

Forepitaxialgrowthprocessesthechoiceofaconvenientandadequatesubstrate
isacrucialtask.Thesubstrateinfluencespropertiesoftheepitaxialfilmsuchas
butdefectisalsodensities,importantroughnesswithofresptheecttofilm,fuabilitrtheryproofwecessingtting,ofi.e.thengrownucleationmaterial.properties,This
means,thataspectssuchasconductivity,transparencyorchemical,thermaland
themechanicalsubstrateisstabilitymandatoryhavetobandeaconsidered.prerequisitetoTherefore,successfulathoroughepitaxialinvgrowthestigationofanofy
material.engivAveryimportantaspectwithregardtothepossibilityofepitaxialgrowthanda
desiredlowdefectdensityofthegrownfilmisamatchingofthelatticeconstantsand
thecoefficientsofthermalexpansionbetweenthefilmandthesubstrate.Generally
thelatticemismatchbetweenthesubstrateandthegrowthmaterialisconsidered
asthemainindicatorofthefeasibilityofepitaxialgrowth.Asaruleofthumba
lowlatticemismatchshouldleadtolittlestrainwhereasalargelatticemismatch
increasestheinducedstrainandhencepreferentiallyleadstomisfitdislocations
atdefectstheinsubstratethecrystal-epitaxialandtolapyoerorincrystaterface.lqualitThisy.inHerebturnyamaydifferenleadtotiationelongahastedto
bestrainmadeleadsbettowaeencrackingcompressivofetheandcrystaltensilewhilestrain.compreItshassivebeenstrainisstatedmorethatlikelytensileto
berelievedincrystaldefects[5].Acrackinguponcoolingofthesamplefromthe
growthtemperaturetoroomtemperaturemayalsobeinducedbyalargemismatch
ofthecoefficientofthermalexpansion(CTE)ofthesubstrateandtheepitaxial
r.eylaThechoiceofasubstrateisnotmerelydependentonitsphysicalandchemical
propertiesbutalso,ashasbeenmentionedabove,reliesontheattainabilityofsuch
material.DuetothehighpriceandlowabundanceofGaNsubstratesforgrowth

34

ofnon-polarGaN,alotofeffortisinvestedinsearchofanadequatesubstratefor
heteroconsideredepitaxyfor.tTheableepitaxial3.2shogrowsawthofcompilationnon-polarofGaN.selectedsubstratesthathavebeen
forWhilesubstratesextensiveforAstudies-orofM-planesubstratesGaNforCgrowth-planehasGaNonlygrobwtheenexisttriggered[157],theroughlysearc10h
yearsagobythepublicationofWaltereitetal.[139]wheretheauthorsexperi-
Mmen-planetallyshoGaNwedgrothewnbonγeneficial-LiAlO2.effectsInoftheavfollooidingwingtheonlyQCSEsubstrates(seesectionforthe3.2)grousingwth
ofnon-polarGaNshallbeconsidered.
Non-nativesubstratesusuallyleadtoaconsiderableamountofextendeddefects,
suchasthreadingdislocations,stackingfaultsorinversiondomainboundariesin
thebinationgrowncencrystal.ters,usuallyTheseindefectsformproofducestatesinsidescatteringtheasbadwellgapastherebnon-radiativyloweeringrecom-the
quantumefficiency.Evidentlyareductionofdefectsiswantedandalotofresearch
isaimedatexactlythispoint.
Thebiggestdrawbackofnon-polarGaNisthestillmuchlargerdefectdensities,
lyingcomparedinthetoCC-plane,-planeposeGaNa[158].prominenFtoremostdefecttheinthefrequennon-ptlyoolarccufilmrringorienstackingtation[faults,158,
fect159].Vdensitariousyinthesubstratesfilms,anandimptecortanhniquestonehavbebeingeeninvEpitaxialestigatedLateraltoOvreduceerGrothewthde-
5tion(ELOG)densit.yThisfromtecusuallyhnique,10allo8−w10ing10acm−2significantolevtelsofreduction106cmof−2[the5],howthreadingeverdislrequiresoca-
additionalstepsintheproductionprocesslikepatterning,makingitmorecomplex
andconcenexptrateensivofe.theInthdirectisstudygrowththeofGaNELOGtecfilms.hniquewasnotapplied,wetherefore
Non-psapphireolar[160GaN,161gro],wthSiC[has162,pre163]viouslyandbγeen-LiAlO2attempted[164,on165].substratesAlsoGaNlikeMR-plane-plane
substrateshavebeenconsideredinafewstudies[166,167],despitetheextremely
highnon-pcoolarstandGaNarelimitedtowobtainafersizesphaseavaipuritlable.y,flatSomesuimprfacesortanandtaaspreductiectsinontheofthegrowthdefectof
density.Previoustothisstudynon-polarGaNhasnotbeenreportedonLGO.Due
tothelowlatticemismatchofthenon-polarGaNplanestotheappropriateLGO
grosurfaceswthof(seeGaNTableon3.2LGO)lowwillbdefectediscusseddensitiesinarechapterexp4ected..TheinvestigationofMBE

5ForthetechniqueofepitaxiallateralovergrowththeacronymsELOandLEOarealsocommon.

]stands1−K6−CTE’-GaN[10X[%]hGaNmismatcLatticetodirectionSubstrateIn-planeGaNrelationship’abbreviationthetableethInGaN.olarnon-pofwthgroepitaxialheterotheforSubstrates’theforsymmetry
GaNEpitaxialtomaterials.othertheorfsimilarlyandGaNof’-planeX
tsconstanLattice˚A][
Crystal3.2:SubstrateableT

]172[116
]168[3.5.1....]168[46.16.
7=8=]170[15=7=10=21=13=4=4=
acacabcac
ααααααααα
]173[0.191.76-10.3516.07-1.852.921.762.923.521.193.582.253.580.093.520.34∼∼

20]¯20]¯20]¯20]¯¯101]¯100]¯100]
[11[1[1[11[010][001][010][010][001][100][001][11[0001][1[0001][11

¯100]¯20]¯10]¯100]¯100]¯20]¯100]¯20]¯100]¯20]
[0001][0001][0001][1[11[10[1[1[11[1[0001][0001]
[0001]22222-Al-Al-SiC[11[1-SiC[11-SiC
O3O-LiAlOγ-LiAlOγ-LiGaOβ-LiGaOβ-LiGaOβ
322MCCARon-GaNon-GaN(100)on-GaN(100)on-GaN(100)on-GaN(010)on-GaN(001)on-GaNon-GaNon-GaNon-GaN
ACMCMACMAC

]168[7589.991.
4=a12=c
Hexagonal3c¯R3O2-Alα

]169[1687.2679.]171[402.372.007.
5=6=5=6=5=
cbaacetragonalTP4bicOrthorhomPna2
2112122-LiAlO-LiGaOγβ

]168[0806.1173.
=a3=c15
mcHexagonal3P66H-SiC

35

36

3.3.2Molecularbeamepitaxyandthesystemusedinthiswork

ThesamplesusedinthisworkwereallgrownbyPlasma-AssistedMolecularBeam
Epitaxy(PAMBE).Herebyageneratedfluxofmaterial(themolecularbeam)isused
todepositatomsormoleculestoformanoverlayer(epitaxy)onasubstrateina
controlledfashion.Whentwoormorefluxesofdifferentmaterialsareprovided,they
canreactatthesubstrateandformacompound.Thematerialisusuallysupplied
throughevaporationbyheatingacrucibleinwhichthehighpuritymaterialisplaced.
Gaseouselementsontheotherhandneednotbeevaporatedbutcanbeletintothe
growthchamberthroughaflowcontrolunit.However,becausethegaseouselements
formstrongbindingmolecules,likeO2andN2,itispurposefultoactivatethemwith
someenergy,sotheycanmorereadilydissociateandtakepartinnewcompound
formation,havingreachedthesurfaceofthesubstrate.Inplasma-assistedMBE
thisactivationisdonebycreatingaplasmaofthedesiredgasinsideaso-called
plasmacell.Here,themoleculesareexcitedbeforetheyarereleasedintothegrowth
chamberanddirectedatthesubstrate.
OneoftherequirementsofaMBEprocessistheconditionofanUltraHighVacuum
(UHV).Forone,thisisnecessarytoprovideforthelongmeanfreepathofmaterial
fromthematerialsourcestothesubstrateandtominimizetheincorporationof
otherelementsinthesamplefromthesurrounding.Theincorporationofsuch
defectsisoftenreferredtoasunintentionaldoping.MBEisseenasamaterial
depositiontechniqueofsampleswithlowimpuritylevelsandagoodcontrolover
compositionanddopingofthestructurebysimplyblockingthematerialfluxesby
rapidshutteroperation.IntrinsicpropertiesoftheMBEprocesscomparedwith
otherepitaxialmethodslikeMOCVDorHVPEaretheonaverageslowergrowth
rateandthepossibilityofgrowingatlowertemperatures.Alowergrowthrateis
usefulforproducingverythinlayersofmaterialandallowsforaprecisecontrolof
structures.Lowtemperaturesareofvaluewhengrowingcrystalswhichareinstable
athightemperatures,suchasInAsandInN,whenthermallydrivendiffusionof
atomsfromorintothesurroundinglayersisundesired,orwhenusingsubstrates
whicharethermallyunstableathightemperatures.
BecauseoftheuseofUHVconditionsitisalsopossibletoapplyin-situanalysis
suchasReflectionHighEnergyElectronDiffraction(RHEED)inMBE,whichis
oftenstatedasoneofthebigadvantagesovere.g.MOCVD.ForRHEED,elec-
tronswithanenergyofapproximately20keVarediffractedatthesamplesurface
underasmallangleofincidence,e.g.5◦.Thediffractionpatternisdisplayedona
fluorescentscreen,whereinformationofthecrystalquality,growthmodeandsur-
facereconstructionscanbeextracted[174].Whilegrowthofe.g.GaAsisusually
conductedinAsrichconditions,i.e.limitedbythesupplyofGa,nitridegrowth
needsslightlyGarichconditions,meaninganexcessofGa,usuallyinformofabout
twomonolayersofliquidGafloatingonthesamplesurface[175],toproduceflat

Figure3.4:PictureoftheMBEmachineusedinthisexperimentalstudy.

37

andsmoothsurfaces.ThissituationexplainswhyRHEEDoscillationscannotbe
observedinnitride(asopposedtoarsenide)growthforthedeterminationofgrowth
brates.eamasaRHEEDfunctionoscillationsoftheareinformatiotensitnyofvnewariationsatomicofthelaydiffersonractedtheelectronsampleRHEEDsurface.
Nevertheless,innitridegrowth,RHEEDdiffractionpatternsareofgreathelpwhen
evaluatingthesurfaceroughness,thecrystaldirectionandlatticeconstantsandthe
approximateGacoverageofthesample.

DescriptionoftheMBEsystemusedinthisworkThespecificMBEchamber
usedinthisworkisaRIBERCompact21TMBEmachinesituatedintheclean
roomoftheCenterforFunctionalNanostructuresattheKarlsruheInstituteof
Technology.ItispartofanMBEclusterof2systemsconnectedbyatunnel;one
mainlydedicatedtothegrowthofIII-Varsenides,theotheroneassignedtothe
growthofnitridesemiconductors.Thefollowingdescriptionwillonlyconsiderthe
latterofthetwosystems.Fig.3.4showsaphotographoftheMBEsystemusedin
.studythisMyworkonthisMBEmachinestartedwithbuildingupthe”nitride”system.This
includedinstallationofthecomponentsdescribedbelow,likee.g.theeffusionand
plasmacells,transferrods,massspectrometer,RHEEDsetup,loadingchamber,
cryogenicvacuumpumpandtheelectricalrackwithallthesurveillanceandcon-

38

trolunitslikeforinstancethetemperatureandshuttercontrolsforthecellsand
substrate.Afterallnecessarycomponentshadbeenattachedtothesystem,the
requiredstablevacuumattained,thecellsfilledwithmaterial,theProportional-
Integral-Derivative(PID)controlvaluesofthetemperaturecontrollersadjusted,the
MBEcontrolsoftware6installedandacompletebake-outofthesystemperformed,
theexperimentalsetupwasreadyforfirstgrowthrunsandcommissioningexperi-
mentswereconducted.Thewellknownsubstrateofγ-LiAlO2fornon-polarGaN
growth[13]waschosenforgrowthconditionandparameteroptimization,whichwill
beshortlysketchedinchapter4beforepresentingtheresultsonthenewsubstrate
LiGaO2.Beforerevertingtotheactualgrowthofnitrides,inthefollowinglastpart
ofthischapter,theMBEsystemdedicatedtoIII-nitridegrowthwillbedescribed.
ThemainbodyoftheMBEmachineconsistsofthreechambers:aloadingcham-
ber,agrowthchamberandatransferchambertoavoiddirectcontactbetweenthe
loadingchamber(oftensubjecttoairexposure)andthegrowthchamber.Thethree
chambersareconnectedviavacuumtightshutters.
Theloadingchamberrepresentsthesystem’sconnectiontothesurroundingenviron-
ment.Itissubjecttofrequentexposuretoairwhensamplesaretakenoutornew
substratesareintroducedintothevacuum.Whenopeningtheloadingchamber,
thecavityisconstantlyfloodedwithpurenitrogentolessencontaminationfrom
air.Thechamberispumpedbyaturbomolecularpumpconnectedtoascrollfore
pump.Thepressureismeasuredbyacompactfullrangepressuregauge,consisting
ofaPiraniandcoldcathodeionizationgauge.Thischamberisalsousedasafirst
desiccationandoutgasingstationfornewlyintroducedmaterial.Forthispurpose2
bakinglampsareinstalledandcontrolledviaaPIDcontroller,connectedtoather-
mocoupleinthevacuumandapowersupply,forsmoothtemperaturecontrolinside
thechamber.Thepressuregenerallyreachesapprox.109mbaraftertheoutgasing
step(usuallyperformedat130◦Cfor60min).
ThetransferchamberispumpedbyaTiiongetterandTisublimationpumpleading
toapressureofabout1011mbar,asmeasuredbyanionizationpressuregauge.Two
transferrodsmagneticallyandalmostfrictionlesssupportedareattachedtothis
chamber,toprovideforthehorizontalandverticalmanipulationofthesamples
fromonechambertothenext.Forthetransportofsamples,onlyoneshutterata
timeisopenedtoavoiddirectcontactbetweentheloadingandgrowthvolumes.
ThegrowthchamberispumpedbybasicallyfourvacuumpumpsincludingaTiion
getterpump,aTisublimationpump,aHecryogenicpumpoperatingbetween12K
and14Kandaliquidnitrogencryopanelsurroundingtheentireepitaxyvolume.
Aconstantcoolingofthecryopanelshieldisassuredbyaliquidnitrogen(LN2)
bathlocatedabovethechamber,supplyingtheMBEmachineconstantlywithLN2

6CustomizedwrittenprogrambyDr.DanielSchaadt.

39

viavacuumisolatedpipes.Theaveragebackgroundpressureinthegrowthcham-
ber,−10measuredbyanionizationgaugeoftheBayard-Alperttype,isapproximately
10mbar.Amassspectrometer(PfeifferVacuumPrisma80,QMS200)isinserted
inthechamberforvacuumanalysisandleakagedetection.In-situgrowthanalysisis
performedbyRHEEDmeasurementsusingtheelectrongunandfluorescentscreen
installedinthesystem.AcameraisinstalledinfrontoftheRHEEDscreentoview
anddocumentdiffractionpatternsofsamples.
ThesourcesattachedtotheMBEsystemcontainIn,Al,Ga,Cu,Si,N2andMg
orZn.Theirspatialorderinthechambercanbeseenintheschematicdrawing
inFig.3.5.ThesolidsourcesareplacedinPBNcruciblesmountedinstandard
effusioncellsequippedwiththermocoupleelements.Fortemperaturemonitoring
andcontrolthecellsareconnectedtoPIDcontrollers(Eurotherm2408)whichin
turngivetheappropriatesignalstothepowersuppliesofeachcell.Eachofthese
cellsisequippedbyitsownshuttertoblockthemolecularbeamfromthecell.The
N2supplyismanagedviaagasflowcontrolunitpositionedinfrontofawater
cooledOxfordHD25RFradiofrequencyplasmasource.Deflectionplatesoperating
atavoltageof600Varefixedattheoutputoftheplasmacelltodiverthighly
energeticionsgeneratedintheplasma.Thisisdonetopreventtheenergeticions
fromreachingthesamplesurfacetherebyriskingdamagingit.
Substratesizesofupto2inchcircularwaferscanbeprocessedinthisMBEsystem.
ThesubstrateisplacedonaMolybdenumholderwhich,whenintroducedtothe
growthchamberisplacedinasubstratemanipulatorwiththesamplesurfaceori-
enteddownward.Thesubstratemanipulatorcanbecontinuouslyrotatedtoprovide
forauniformdepositionofthematerialandheateduptoamaximumtemperature
of1000◦C;theheatingisalsomanagedviaaPIDcontroller.Inordertosimul-
taneouslycutofforstartagrowthprocessinvolvingseveralsources,ashutteris
installeddirectlybelowthesubstratemanipulator.
Awaytogetanideaoftheamountofmaterialprovidedforsamplegrowthisto
measuretheBeamEquivalentPressure(BEP)ofthemolecularoratomicbeams
comingfromthesource.Thismeasurementisdonebyanionizationgaugewhichis
placedjustunderneaththesubstrateshutterbyamanipulatorarmwhenmeasuring.
Duringgrowth,thisgaugeisretractedtoapositioninsideoneoftheflanges.The
pressureindicatedbythegaugeisameasurefortheamountofprovidedfluxof
thespecificmaterial.Itcanbeusedtoreproducegrowthconditionsmorereliably.
Anothermeasureforthereestablishmentofearliergrowthconditionsisthecell
temperature.However,dependingonthematerialcontentofthecell,thefluxmay
varyataconstantcelltemperature.
Electronicunitsinthesystemareconnectedtoacomputerforsupervisionand
controlofthegrowthparameters,suchaspressuremonitoring,vacuumanalysis,
camerafortheRHEEDobservationandtemperatureandshuttercontrolofthe

Figure

3.5:

40

Scusedhematicinthisofwtheork.Forarrangemenreasonstinsideofclaritthey,grothewthvchamacuumberofpumps,theMBEBEP
pressuregauge,massspectrometeraswellastheflangesandwindows
omitted.erew

baccellskedandupthewithasubstratenumberheater.ofuninTheterruptiblesystemispowattacerhedsupplytoan(UPS)emergencyunits.powergrid

4Non-polarGaNfilms

Non-polarGaNisusedasadescriptionreferringtotheabsenceofelectricfields
runningalongthegrowthdirectionofthematerial.Thisdirectsawaytoavoidthe
QuantumConfinedStarkEffectandallitsassociatedconstraints,asdescribedin
section3.2.Thecurrentchapterisdedicatedtotheinvestigationofthegrowthof
non-polarGaNfilmsondifferentsubstratesandsubstrateorientationsbyplasma-
assistedmolecularbeamepitaxy.Theseare(100)γ-LiAlO2,(100)β-LiGaO2and
(010)β-LiGaO2.Inthefirstpartofthischapter(sec.4.1),inordertocommission
thenewlybuiltMBEsystem,theresultsofgrowthonγ-LiAlO2,asubstrateknown
toyieldM-planeGaN[176],aredisplayed.Subsequently,GaNgrowthandgrowth
resultsonthenewsubstrateβ-LiGaO2arepresented(sec.4.2).

4.1GrowthofM-planeGaNon(100)γ-LiAlO2

TofindappropriategrowthconditionsandparametersforGaNgrowth,M-plane
GaNwasgrownon(100)γ-LiAlO2(LAO)inthenewlyinstalledMBEsystem.The
aimwastoobtainflatsurfacemorphologyofGaNlayerswithgoodcrystalquality.
Ashortcharacterizationofthesubstrateinthenextsection(4.1.1)willbefollowed
bythedescriptionofgrowthparameteroptimization(section4.1.2).

4.1.1Propertiesofthesubstrateγ-LiAlO2

Interestingly,thefirststudyofGaNgrowthon(100)LAOresultedinpolarC-
planeGaNfilms[177].Later,however,itwasshownthat,ifthegrowthparameters
weretuned,M-planeGaNfilmscouldbegrownon(100)LAO[164,176,178–180].
ThecrystalstructureofLAOwasdeterminedtobeP41212withlatticeconstants
a=5.1687andc=6.2679[169].TheLAOcrystaliscompletelytransparentinthe
wavelengthrangebetween200nmand4µm,itsmeltingpointisaround1700◦C
anditcanbegrownbytheCzochralskitechnique[170],i.e.largecrystalscanbe
pulledoutofameltusingaseedcrystal[181].Despiteitsrelativelyhighmelting
pointthecrystalhasbeenreportedtodecomposealreadyattemperaturesabove
900◦C[182].ThefactthatLAOisetchedbyanumberofacids[183]hasbeen

42

Figure4.1:Ballandstickmodelofthe(100)LiAlO2substratesurface.Theshaded
areasindicatepossiblenucleationsitesforC-andM-planeGaN.From
].135[

usedtoobtainafreestandingGaNsubstratebyGaNgrowthon(100)LAOand
subsequentremovaloftheLAObyetching[166].The(100)faceofLAOexhibitsa
certainpolarity,i.e.twodifferentqualitiesofGaNhavebeenfoundtogrowonthis
substrate[164].Thisisdiscussedindetailin[135,164].
ThelatticematchofGaNandLAOmakebothC-andM-planeGaNgrowthpossible
on(100)LAO(seeTable3.2)asillustratedinFig.4.1wherethe(100)LAOsurface
andpossiblesitesforC-andM-planeGaNgrowtharedepictedschematically.The
simultaneousgrowthofM-andC-planeGaNon(100)LAOhasalsobeenobserved
inthepresentstudyandposesanissuewhenaimingathighpurityofasingleGaN
phase[184–186].

4.1.2GrowthparameteroptimizationforM-planeGaNgrowth
-LiAlOγ(100)on2

Forepitaxialgrowthofnitridesthereareamanifoldofgrowthconditionsthatin-
fluencethecrystalqualityoftheepitaxiallayer.Someexamplesarethesubstrate
temperature,theprovidedmetaltonitrogenratio,thegrowthtime,thenitrogen
plasmapower,the1substratemorphology,pre-growthtreatmentofthesubstrate,like
annealing,nitridationandwet-chemicalcleaningoretching.

1tempNitridationeratures.isaThistermprousedceduretodeisscribewidelytheusedexpforosureofnitridethegrosubstratewthontoactivsapphireatedsubstratesnitrogen[at187,elev188ated].

43

Inthefollowing,thetuningofoneoftheseparameters,namelytheinfluenceof
thesubstratetemperature,willbeexemplified.Thereafter,somesamplesgrownat
optimizedgrowthconditionswillbepresented.
Priortogrowth,theLAOsubstratesweretypicallycleanedbysubsequentlydipping
thesubstratesintoTrichloroethylene(TCE),AcetoneandMethanolforoneminute
andrinsedfortensecondswithdeionizedwater.Thereaftertheywereblowdried
withnitrogen,attachedtoaSiwaferwithindium,placedintoamolybdenumholder
andintroducedintotheMBEmachine.TheSiwafer’sfunctionistoenhancethe
heatabsorptionofthesubstrateheaterduringgrowth;Indiumisusedtoprovide
forahomogeneoustemperaturecoupling.OnceinjectedintotheMBEsystem,the
substrateswereoutgasedat130◦Cfor60minandthentransferredintothegrowth
er.bhamc

InfluenceofthesubstrategrowthtemperatureToseetheimpactofthesub-
stratetemperatureonthesurfacemorphologyoftheGaNlayers,thesubstrate
temperaturewastunedfrom600◦Cto800◦C,therebyleavingallotherparame-
tersconstant.Alltemperaturevaluesgivenforthesubstratetemperatureinthis
workcorrespondtothethermocouplereadingofthethermoelementattachedtothe
substrateheater.Theactualtemperatureofthesampleisthereforelowerbyap-
proximately30◦Cto50◦C2.Fig.4.3displaysaseriesofsampleswhichweregrown
withaconstantN2fluxof0.3◦sccm3ataplasmaforwardpowerof450Wanda
gallium◦celltemperatureof930Cfor20min.Aftergrowth,thesampleswereleft
at900Cfor5mintoevaporateremnantGafromthesurface.
AscanbeseenontheScanningElectronMicroscopy(SEM)imagesinFig.4.3,a
mixtureofcolumnarhexagonalC-planecrystalsandM-planeislandsarevisibleat
alowsubstratetemperaturewhereasthesurfacemorphologydevelopsintoahigher
purityM-planefilmaroundtemperaturesbetween625◦Cand700◦C.Thisdevel-
opmentofmorphologychangeprobablyresultsfromahighersurfacemobilityand
longermeanfreelengthofpathofadatomsonthesample,beforegettingincorpo-
ratedintothefilmathighertemperatures,henceleadingtoamoreuniformmaterial
distribution.Athightemperatures(Fig.4.3(d)and(e))however,othereffectscome
toplay.Thiscanbeafasterevaporationrateofatomsfromthesamplesurfacelead-
ingtoachangedGa/Nratio,resultinginachangeofthegrowthconditionsfrom

2Aninternalcalibrationcanbeattainedbyobservingthe7×7surfacereconstructioninRHEED
patternsofa(111)Sisubstrate,whichappearsaround1100±15K[189],i.e.at∼827◦C.
3TheGacellhastwofilamentsforheatingtheGafilledcrucible,whichcanbecontrolledseparately,
oneatthebaseofthecell(body)andoneclosertothecrucibleexit(tip).Inthisstudythe
temperaturedifferencebetweenbodyandtipwasalways50◦C.Inthisthesisonlythehighertip
temperaturevaluewillbegiven.

44

Figure4.2:SchematicillustrationofthemainRHEEDazimuthsthatcanbeseen
inciproM-calandvCalueof-planetheGaN.latticeRHEEDspacingperpmeasuremenendiculartsaretothesensitiveelectrontobtheeamre-
direction.

galliumtonitrogenrich.AtasubstratetemperatureTSub=750◦CM-planeis-
landsappearagainsurroundedwithC-planeorientedGaN.Thehighamountof
C-planeGaNgrownatsubstratetemperaturescanbeseenintheRHEEDpictures,
whichshowthetypicaltwo[1¯100]and[11¯20]azimuthsobservedonC-planeGaN.
ToillustratetheRHEEDobservations,Fig.4.2showsthemainazimuthsthatcan
beseenonM-andC-planeGaNschematically.Thespacingofthestreaksonthe
RHEEDscreenisproportionaltothereciprocalvalueofthelatticeconstantsofthe
crystalperpendiculartotheRHEEDbeamdirection.Notethatthe[11¯20]azimuth
correspondstodifferentdistances,dependingonthecrystalgrowthdirection.The
M-planeGaNvanishescompletelyatTSub=800◦CandtheRHEEDpatternshows
clearandbrightdiffractionoffaC-planeGaNsurface.
ThelowerSEMpicturesinFig.4.3(b)and(c)showanadmixtureoftheC-plane
orientationinformofsmallhexagonalcrystalshapes,whichpreferentiallynucleate
atcracksorinhomogenetiesinthefilm.Byfinetuningthegrowthparametersitis
aimedtocompletelygetridoftheseC-planeinclusions.Indeed,noneofsuchC-
planecrystallitescanbeseeninthesamplesshowninFig.4.3(d)and(e).Moreover,
theM-planelayersinFig.4.3(b),(c)and(e)showcracksinthelayersurface,
whicharepossiblyduetothesuboptimalGa/Nratioatthissubstratetemperature
andtheshortgrowthtimeusedinthissampleseries4.Theroughcircularareas
thatcanbeseenthroughouttheSEMimagesofFig.4.3areso-calledfootprints

4FilmqualityhasbeenobservedtobeinfluencedbytheGaNlayerthickness[165],i.e.thegrowth
time.

45

Table4.1:Tplesable(seenofrmsinFig.roughness4.3(b)-(e))determinedgrownbyonAFMγfor-LiAlOM2at-planedifferenGaNtsasub-m-
stratetemp◦eratures.Theflattestsurfacemorphologyisachievedfor
C.675=TSub

Growthrmsforanarearmsforanarea
temperature[◦C]of10×10µm2[nm]of5×5µm2[nm]
8.010.96256.18.66503.55.36757.18.6700

ofGadropletspresentonthesamplesurfaceduringgrowth.Areductionofthese
featuresiswanted,however,somewhatdifficulttoac◦hieve.However,theroughness
issignificantlyimprovedinthecaseofTSub=675C(Fig.4.3(d))comparedto
thesamplesgrownatothertemperatures.Thisisalsoreflectedintherootmean
square(rms)roughnessvaluesobtainedbyAtomicForceMicroscopy(AFM)foran
areaof10×10µm2listedinTable4.1.

(a),

the(a),-planeC◦M◦forwwindowthgroA(g).C
◦eraturestemphigheranderwloatwhileC
◦800and)(fC
700toC750600oferaturestempsubstratetdifferenatwngrossampleariousvofimagesRHEEDandSEM625625arounderaturestempforseenebcanGaNpictures.RHEEDthefrominferredebcantationorienfilmThe.largeisGaNoffraction-planeC
◦(e),C◦700(d),C◦675(c),C◦650(b),C◦

4.3:

Figure

46

47

GaNsamplesgrownon(100)γ-LiAlO2usingoptimizedgrowthconditionsBy
tuningthevariousgrowthparameterssmoothM-andC-planeGaNsampleswere
grownon(100)LAO.TheresultscanbeseeninFig.4.4andFig.4.5.Anarrow
rockingcurvefullwidthathalfmaximum(FWHM)statesahighcrystalquality
withanarrowdistributionoftilttothemaincrystalorientation.StreakyRHEED
patternsareanindicationofaflatsurfaceandcanbeseeninboththeM-and
films.GaN-planeCTheM-planeGaNsampleofFig.4.4wasgrownatasubstratetemperatureof725◦C
for120min,resultinginathicknessofroughly200nm(measuredbySEM).TheGa
celltemperaturewasTGa=931◦Candthenitrogenplasmacelloperatedat450W
withanitrogenfluxof0.3sccmresultinginaBEPGa/Nratioofapprox.5.7.
ThetypicalslatelikemorphologyoftheM-planeorientationisshownintheSEM
imageofFig.4.4.ThesurfaceisessentiallyclosedandshowsnocracksorC-plane
inclusions.InthelowermagnificationSEMimage,ontheleftside,Gadroplets
arevisibleonthesurface.TwoRHEEDazimuthsatthetopofFig.4.4displaythe
reciprocallatticespacingoftheaGa(left)andcGa(right).Fromthestreakinessofthe
patternsaflatsurfaceisexpected;the[0001]azimuth,however,showsdiscontinuities
inthestreaks,whichpointstowardathreedimensionalmodulationofthesurface,
droplets.ofpresencethei.e.X-raydiffraction(XRD)performedonthesampleconfirmsthehighM-planephase
purityofthesampleastheω−2θscan(leftsideinFig.4.4)onlydisplaysthepresence
ofthe(1¯100)GaNand(200)LAOorientedplanesatω=16.14◦andω=17.34◦,
respectively,andnoindicationofthe(0002)GaNreflectionusuallyfoundnearω=
17.28◦canbeseen.TherightXRDscaninFig.4.4depictsarockingcurvescan
acrossthe(1¯100)GaNreflectionandrevealsaFWHMofabout817arcsec,which
isagoodvalue;comparabletoorsmallerthanresultsreportedinliteraturefor
M-planeGaNgrowthonLAO[139,164,165,190–192].
RHEED,SEMandXRDdataofaC-planeGaNsampleisdepictedinFig.4.5.By
growingatanominalsubstratetemperatureof725◦Cfor60mina90nm(measured
bySEM)thickGaNfilmwasobtained.Inthiscasethesubstratewasmountedonto
theSiwaferinadifferentwaycomparedtotheM-planesampledescribedabove,so
thatabetterthermalcouplingoftheSitothesubstrateandhenceahighersubstrate
temperaturecanbeassumed.Theothergrowthparameterswereleftunchanged,so
thatTGa=931◦C,thenitrogenplasmacellwasoperatedat450Wwithanitrogen
sccm.0.3offluxFromtheRHEEDpictures,displayedatthetopofFig.4.5,thedominantcrystal
orientationcanalreadybededuced,asthedistancesbetweenthevisiblestreaks
correspondtothereciprocalvaluesofaGaforthe[1¯100]and√3aGaforthe[11¯20]
azimuth.Thediffractionpatternsarestreakyandhenceindicateaflatsurface

Figure

4.4:

48

ThisfiguredisplaysRHEED,SEMandXRDresultsobtainedfora
200nmthickM-planeGaNfilmgrownon(100)LAO.Thetoptwo
picturesdisplaystreakyelectrondiffractionpatternsoftheM-plane
GaNsurfaceintwodirections,[0001]and[11¯20],indicatingasmooth
surface.TheSEMimages,takenattwodifferentmagnificationsshow
thetypicalM-planesurfacemodulationandthepresenceofGadroplets
onthefilm.Thelowertwographs,displayingXRDresults,showthe
highphasepurityintheω−2θscan(left)byastrong(1¯100)GaN
andnoreflectionofthe(0002)GaNplaneandalowFWHMvalueof
817arcsec;agoodvalueforM-planeGaNgrowthon(100)LAO.

49

where,morphologyincon.trastThistoobservtheMation-planecanfilm,againbehexagonallymadeinshaptheedSEMcrystallitespicturescaninbeFig.seen4.5in
onareasthewheresamplethesurfaceGaNfilmandisvisiblenotinfullythelocoalesced.werThemagnifiedsizeofSEMtheimGaageissdropletsmallerpresenthant
ontheM-planeGaN,supportingtheconclusionofahighersubstratetemperature
case.thisinThex-raydiffractionmeasurements,depictedatthebottomofFig.4.5clearlycon-
firmthedominatingC-planeorientationoftheGaNfilm.Threepeaksarevisible
intheω−2θscanwhich,fromthelefttotheright,areidentifiedtooriginfromthe
(1¯100)GaN,(0002)GaNand(200)LAOplanes.AlthoughtheexistenceofM-plane
GaNisproveninfilmbythismeasurement,itisonlypresenttoaverysmallfraction
paseaks,canbewhichconcludedis4000/100,fromtheinmeaningtensitanyratiosadmofixturetheofattributedroughlyC2.5%-planeofMand-planeM-planeGaN
intheotherwiseC-planeorientedfilm.TheinsetintheleftXRDgraphshowsan
ωthe−2θcostscanofaplowerfoerinrmedtensitwithy.aThetriplescancrystalshowstheanalyzerclearlytoacsephievearatedahigherdoublepeakresolutionoftheat
(0002)GaNand(200)LAOplanesatω=17.25◦andω=17.34◦,respectively.The
rightpartofXRDdatainFig.4.5depictstherockingcurveofthe(0002)GaNpeak
takeninthetriplecrystalanalyzerconfigurationandexhibitsaFWHMof1946arc-
sec.ThisisaratherlargevalueforC-planeGaNindicatingfaircrystallinequality
ofthicthekerfilm.GaNfilmsUsingareothergrowngro(onwththemethoorderds,oflik200eµMOm),VPErepoandrtedHVPE,(0002)GaNwheremFWHMuch
rockingcurvevaluesinliteratureare2200arcsec[193],1400arcsec[194],1270arcsec
[GaN195]groandwthwithon(100)thoroughLAOgrobywthMBEhasoptimizationonlyb<een400reparcsecorted[196once].Howgivingever,aCFWHM-plane
ofthe(0002)GaNreflectionof2700arcsec[177]whichissignificantlyhigherthan
theoneshowninFig.4.5.

4.1.3SummaryofGaNgrownonγ-LiAlO2
Observationofveryhighphasepurity,asmoothsurfaceandlowFWHMvaluesof
evMap-planeoratedGaNoretcfilmshed,onLAtheseOwGaeredropletsaccompaniedleaveabytherougherpresencesurfaceofbGaehinddroplets.thantheIf
adjacentfilmthatwasnotcoveredbyGa.Itwasdifficulttoobtainhighphase
purity,highcrystalqualityandasmoothsurfacewithoutGadroplets.
inThevestresuligations.tsobtainedTheforsurfaceM-planemorphologyGaNonandLAOFWHMarevsimilaraluestoofthepreviouslyXRDreprocokingrted
curvesarecomparabletoreportsinliterature.
ForhighqualityGaNC-planeepitaxiallayersgrownheteroepitaxially,substrates
likeforexample(001)LiGaO2andmainlyhexagonalsapphiresubstratesareused.

Figure

4.5:

50

TheRHEED,SEMandXRDresultsobtainedfora90nmthickC-
planeGaNfilmgrownon(100)LAO.ThetoptwoRHEEDpictures
shoSEMwimastreakygesofCtw-planeodifferenazimtuthsofmagnificationsGaN,indicatingshowahexagonalsmoothCsurfac-planee.
graphsstructuresintheandlowtheerpartpresenofcetheoffigureGashodropletswtheonpresencethefilm.ofbothTheC-XRDand
M-planeGaNinthesample.However,theM-planefractionamounts
toshownonlyinabtheoutlow2.5er%.rightThepart,rochaskingacurveFWHMofofthe1946(0002)arcsec.GaNreflection,

51

RockingcurveFWHMvaluesofthe(0002)GaNreflectionon(001)LiGaO2arere-
portedaround100-300arcsec[193,197],thebestvaluebeing41.7arcsec[193],and
forGaNgrownonsapphirebelow100arcsec,e.g.51arcsecin[198].However,
C-planeGaNon(100)LAOhasonlybeenreportedfewtimes[177,193,195,196]
andthecrystalqualityachievedinthisstudyisclosetothemajorityofthereported
onesandbetterthantheGaNfilmsalsosynthesizedbyMBE.
AftertheinfluencesofanumberofgrowthparametersonthebehaviorofGaN
growthinthenewlycommissionedMBEmachinehadbeeninvestigatedandgrowth
ofGaNhadbeenestablishedandunderstoodonLAO,growthofGaNonthenew
substrateLGOcouldproceed.Thisisdescribedindetailinthenextchapter.

4.2Growthofnon-polarGaNonβ-LiGaO2
Theprocessofsamplepreparation,growthandcharacterizationofnon-polarGaN
filmsonLGOisdescribedinthissection.Itstartsoffwithimportantproper-
tiesofthesubstrateusedinthisstudy(sec.4.2.1).Thereafter,abriefreporton
thefabricationofthesubstrates(sec.4.2.2)isgiven,followedbyadescriptionof
thegrowthproceduresleadingtonon-polarGaNthinfilmson(100)LiGaO2and
(010)LiGaO2withapresentationoftheircharacterizationresultsinsections4.2.3
and4.2.4,respectively.

4.2.1Propertiesofthesubstrateβ-LiGaO2
Largemethodw[afer199].sizesTheofthestructureLiGaO2ofLiGsubstrateaOcanwasbeachievdeterminededbyastheorthoCzocrhomhralskibic,grospacewth
2◦◦◦[171group].Pna2While1the(#33),(001)withfacelatticeshowsparametershexagonala=5atomic.402A,barrangemen=6.372t,Athe,c=(100)5.007andA
(010)facesareatomicallyorderedinarectangularfashion.Examinationofthe
possibleepitaxialrelationshipsofGaNonLGOleadstotheconclusionthatclosely
(100)latticeLGOmatc5,hedA-planerelationsGaNbetonween(010)LGOLGO6andandGaNCare-planefoundonfor(001)MLGO-plane7[173GaN,200on].
ThismakesLGOa¯uniquesubstrate¯forgrowthofthethreeprominentGaNcrystal
orientations[0001],[1100]and[1120],dependingontheLGOsurfaceorientation.
attenBecausetionofandthewloaswinvlatticeestigatedmismatcforhbtheetwheteroeenLGOepitaxyandofGaN,GaN,thisasansubstratealternativreceievedto
5latticemismatch[010]LGO[11¯20]GaN:0.07%and[001]LGO[0001]GaN:3.58%
67avlatticeeragemismatclatticehmismatc[100]LGOh0.9%[1¯100]GaN:2.22%and[001]LGO[0001]GaN:3.58%

52

Table4.2:TheaveragelinearcoefficientsofthermalexpansionofLiGaO2.After
].157[ReferenceTemperatureα[100]α[010]α[001]
range[K][×10−6K][×10−6K][×10−6K]
Nanamatsuetal.[215]293-473697
Neumannetal.[216]293-1113n.a.n.a.n.a.
293-47312.615.77.5
Ishiietal.[217]293-10731.711.04.0
RawnandChaudhuri[172]293-142310.1±0.221.1±0.313.6±0.2
293-4737.0±0.315.6±0.410.7±0.1

sapphire8,alreadyinthesecondhalfofthe1990’sbyMOCVD[201,202],MBE
[200,203–208],MOVPE[209,210],HVPE[211],PLD[212,213]andbyreaction
withconductedammoniaonat(001)elevLGOatedaimingtemptoeratureimpro[v214e].theHowcrystalever,allqualitofyoftheseC-planestudieswGaN.ere
Beforethispresentstudy,therehasbeennoreportaboutGaNgrowthoneither
oftheotherorientationsofLGO.ThegrowthofM-planeGaNon(100)LGOand
A-planeGaNon(010)LGObyPAMBEwillbeshowninsection4.2.3and4.2.4,
.elyectivrespThemeltingpointofLiGaO2liesapproximatelyat1570◦C-1585◦C[5,157,215]
makingitinprinciplesuitableforgrowthathightemperature.However,ithas
beenreportedthatthesubstratesurfacedeteriorateswhenheatedtotemperatures
above800◦C[200]andsecondaryphasesappearattemperaturesaround1173◦C
[sion172].repThereortedisinanunceliterature,rtainytyetonallthethedatadataonshowLGO’salargecoefficienanisotroptsofyofthermalthethermalexpan-
expansion.Thismayposesomedifficultyingrowthofthickcrackfreelayerson
topofLGO.Table4.2givesacompilationofresultsobtainedbydifferentgroups
determiningtheaveragelinearCTEofLGO.InadditionitwasfoundthattheCTE
dependquitestronglyonthetemperature[172,216]andacomparisonofthelinear
CTEisonlyapproximatelylegitimateinthelowertemperaturerange.Insection
4.2.3thethermallyinducedstrainintotheGaNfilmisevaluated.
LGOishydrolytic,i.e.itdissolvesinwater,andiseasilyetchedbysolutionswith
differentpH[5,157].Doolittleetal.[218]showed,thatanentireLGOsubstrate
couldbeetchedawayinlessthan5mininabasesolutionleavingtheGaNlayer
unharmed.Thisledtothedemonstrationofremovingthesubstrateandtransferring
theGaNfilmontoaGaAssubstrate.Thecompleteremovalofthesubstratecould
alsobeusedtoattainfreestandingGaN,whichwouldbeagreatbenefitforvarious

8latticemismatchhexagonalsapphiretoC-planeGaN:∼13.9%[157]

53

applications.However,thereisalsoadownsidetothechemicalinstabilityofLGO,
becausepolishingthesubstrateisdifficult,leadingtothenecessityoffindingan
adequatesubstratepretreatmentforsmoothsurfaces.
ThebandgapofLGOisat5.6eV[203],correspondingtoawavelengthof∼221nm.
ItismaterialoftransparenLGOtisapprotransparenximatelytinfromtheitswholebandgapopticalupandto6aµmlarge[5],fractionmeaningofthethe
toinfraredheatsptheectrumsubstrateoftoaspelectromagneticecifictempwaves.eratureThisforfactgrowth.requiressomethoughtofhow

4.2.2Growthofβ-LiGaO2

TheLGObulkcrystalwasgrownattheNationalSunYat-senUniversityinTaiwan
thencutandpolishedbyacommercialvendor.Thecrystalgrowthwascarriedout
inaCzochralskipullingfurnace.Thestartingrawmaterials(Li2CO3andGa2O3
powderswithapurityofatleast99.99%)weremixedmechanicallyaccordingto
99.99%stoichiometricratio.Afterplacingtherawmaterialsinaniridiumcrucible
coveredwithaniridiumlidtoreducethetemperaturegradient,thecruciblewas
heatedtoapproximately1650◦Ctomelttherawmaterials.Duringgrowth,nitrogen
gaswascontinuouslysuppliedtopreventtheoxidationoftheIrcrucible.Theseed
hada(001)c-axisorientation.Arotationrateof10-20rpmwasusedtocontrol
thegrowthconditions.BecauseofthehighvaporpressureoftheLGOmelt,ahigh
crystalpullingrateof2-4mm/hwasappliedinordertoavoidlossofthemelt.Due
todifficultiesinobtainingtransmissionelectronmicroscopyimagesonthesetypes
ofsubstrates,owingtotherapiddestructionofLGOunderhighlyenergeticelectron
irradiation,onlylimitedstructuralinformationhasbeenobtainedsofar.Thecrystal
ofwhichthesubstratesusedinthisworkwerecutfromshowedasmallinclusionat
thebottom4ofthe5c−rystal.3Notwinboundarieswerefoundandadislocationdensity
ofabout10∼10cmwasestimated[219].
Thesubstrateswerecutintwodifferentorientations.Onesethadasurfaceori-
entationof(100)whileanothersetwascutparallelto(010).Inthefollowingthe
growthofM-planeGaNon(100)LGO(section4.2.3)andA-planeGaNgrowthon
(010)LGO(section4.2.4)willbepresented.Eachofthesesectionsisdividedintwo
parts;thefirstpartisconcernedwiththeactualgrowthofthefilmsandthesecond
partwiththecharacterizationoftheGaNfilmsthathavebeengrown.

4.2.3GrowthofM-planeGaNon(100)β-LiGaO2

aFiguredotted4.6rectashowsngle.aballTheandgraystickrectanmoglesdelofdisplatheyp(100)ossibleLGOnsurfaceucleationwithsitesitsforunitMcell-planeas

54

Figure4.6:Ballandstickmodelofthe(100)LGOsurface.Thedottedrectangle
signalsthesizeoftheLGOunitcell,whiletherectangleswithcontin-
uouslinesindicatepossiblenucleationsitesfor(1¯100)GaN.

GaNgrowth.Thelatticemismatchfortherelationship[010]LGO[11¯20]GaNand
[001]planeLGOofLGO[0001]alsoGaNbisenefitsonlyfrom0.07%thelacandkofa3.58%,metal–nrespectivon-metalely.pGroolaritwthy,onwhicthehis(100)an
issuethathastobetakenintoaccountwhengrowingon(001)LGO.

Duetotransparencyofthesubstrateinthevisibleandnearinfraredwavelength
rangethesubstratewasmountedontoaSiwaferusingathinlayerofIntoprovidea
homogeneousthermalcoupling.The◦substrates,insertedintomolybdenumholders,
weretransferredfirstintooutgasedtheforgro60wthcminhamatber.130ActivCinatedtheloadnitrogenlockwaschamsuppberliedandbythethereafterradio
frequencyplasmacelloperatedatasteadygasflowof0.3sccmand450Wforward
power.Agrowthrateofroughly100nm/hwasused.Growthwasmonitoredin-situ
theusingsampleRHEED.fortoToolongpreventwiththedamagetoelectronthebeam.sample,carewastakentonotirradiate

Initialgrowthshowedalmostcompletepeelingoffoftheepitaxialfilmfromthesub-
strate.Thismayhaveanumberofreasonssuchasdamagecausedtothesubstrate
byitsirradiationwithhighlyenergeticelectronsfromRHEEDmeasurements.More-
over,anenhancedmisfitoftheanisotropicCTEbetweenLGOandGaNathigh
temperaturecouldfacilitatethelift-offoftheepitaxialfilmfromitssubstrate.The
exactcalculateasdirectiontheredepisaendenlargetamounscattertofofexpstresserimencausedtalbdaytathisonthemechanismdeterminationisdifficofulttheto

55

Figure4.7:Non-linearbehaviorofthethreecoefficientsofthermalexpansionof
LiGaO2inthetemperaturerangebetween293Kand1100K.Graph
takenfrom[216].

CTEforLGO[172,215–217].However,fromthedatareporteditisapparent,that
thereisastronganisotropyintheLGOvaluesfortheCTE.Inaddition,theCTEin
aal.[temp216]eratueachreofrangethe3of700differenKtareCTEnotinconstanLGOtshoandwsasapoindifferentedtoutbnon-linearybNeumannehavioret
inthetemperaturerangebetween293Kand1100K.Thisisseeninthegraphof
depFig.4.7endence,whereofthethetempCTEoferaturthee.threeThereforecrystalthedirectionscalculationa,ofb,theandcstressarecauseddepictedusingin
inexact.isCTEtheRather,a◦directcomparisonofthelatticeparametersnearthegrowthtemperature
bGTyRa=wn700andCandChaudhatrouriom[172te]mpanderatureNeuma(RnnT)etisal.more[216].reliableHeretheusingdataresultingpresenlatttedice
mismatchnearGTissubjectonlytosmallvariations.ThelatticeparametersofGaN
werecalculatedbytheCTEgivenin[125]forhightemperatures.Thedifference
toin1.18%latticeandmismatcforhGaNbetcwoneenLGOGTcatandR0.50%TfortoGaN0.61%.aonInLGObothb/2thethen[11¯lies20]atGaN0.93%and
the[0001]GaNdirectionthestressinducedfromtheLGOlatticeontotheGaN
latticedirectionwhenthancoinolingthecdown-directionisofcompressivGaN,e.acracSincekingmoreofthestressGaNisfilmcausedisinexptheectead-

56

Figure4.8:SEMimagedemonstratingthecrackingin[0001]directionandlifting
offofanM-planeGaNfilmon(100)LGO.Thepicturewastakenwith
a45◦tilttowardsthesubstratesurface.

alongthec-direction.Thecracksinc-directionareconfirmedbySEMpictures
(seeFig.4.8)wheretheorientationofthefilmisinferredfrompreviousreportsof
M-planegrowthondifferentsubstrates,e.g.[166,176,220,221],andexperiences
withM-planeGaN.Fromtherelativelysmallchangeinlatticemismatchcausedby
thermalexpansionitisunlikelythatthismechanismissolelyresponsibleforthe
peelingoffoftheepitaxialfilmfromthesubstrate.However,acontributionofthe
anisotropicthermalexpansionofGaNtothisphenomenoncanbeexpected.
AprocedureforgrowthofM-planeGaNonLGOwasfound.However,dueto
thelimitedavailabilityofsubstratesthegrowthconditionswerenotthoroughly
optimized.TheGaNfilmhadcracksattheedgeswhentakenoutofthegrowth
chamber,probablycausedbyatoohightemperatureintheareaofcontactbetween
thesubstrateandtheMoholder.However,thecenterofthefilmremainedstable
anddidnotlift-offorshowcracks.PriortoGaNdepositionthesubstratewas
annealedat800◦Cfor60min.Thereafter,growthofGaNcommencedatslightly
Garichconditionsatasubstratetemperatureof700◦C.Thetemperaturesgiven
correspondtothethermocouplereadings.
Thesamplewascharacterizedex-situbyXRDinaBrukerAXSD8Discoverdiffrac-
tometer.Aceramicx-raytubeoftypeKFLCu2Kwasusedtogeneratethechar-
acteristicCuKαline.ScanningelectronmicroscopywasperformedonaLEO1530
systematan1operatingvoltageof10keV.AFMimagesweretakenwithaSiN
tipincontactmodewithanAutoprobeCPheadfromParkScientificInstruments
calibratedwithacommercialcalibrationsample.

57

CharacterizationofGaNgrownon(100)LiGaO2ObservationoftheRHEED
patternsoftheLGOsubstratebeforeandafterannealingshowedaremarkable
improvementofthesurface.InFig.4.9twoazimuthsservetoillustratethedevel-
opmentofthereflectionpatterns.Whilethetoptwopictures,(a)and(b),showthe
reflectionpatternsofthesubstratein[010]and[001]directionbeforeannealing,part
(c)and(d)displaythesituationafter1hourannealingat800◦Cforthesamedirec-
tionsrespectively.Becauseoftheweakreflectionsignalfromthesubstratebefore
annealingitwasimpossibletofindtheexact[010]and[001]azimuths.Thepictures
givenin(a)and(b)thereforeonlyroughlycorrespondtothesedirections.Thetwo
lowestpictures,(e)and(f),inthisfigureshowRHEEDimagesofthesampleafter
growth,takenatthesameanglesas(c)and(d),respectively.Theyshowstreaky
imagesofM-planeGaNalongthe[11¯20]and[0001]directions,wherethespacingof
thestreakscorrespondtothecandalatticeconstantrespectively.Comparingthe
twoRHEEDimagesofLGOafterannealingwiththetwolowerpicturesofGaN,the
epitaxialrelationshipisdemonstratedinaccordancewithFig.4.6.Inotherwords,
fromthespacingsoftheRHEEDstreaksinFig.4.9ahighdegreeoflatticematchof
M-planeGaNon(100)LGOisevidentforaGaNonbLGO/2andcGaNoncLGO.From
thestreakinessoftheRHEEDimages,thefilmisexpectedtobeflatandsmooth.
AspottyRHEEDpatternwouldbeanindicationofaroughsurfaceoreventhree
dimentionalgrowth(likefoundwhengrowingquantumdots).
Surfacescratchesonthesamplebeforeandaftergrowtharevisibleinopticalmicro-
scopeimages(Fig.4.10).Thesescratchesmostprobablyoriginatefrommechanical
polishingofthesubstrate.Fig.4.11showsthesurfacemorphologyoftheGaNlayer
bymeansofSEM(a)andAFM(b).Theepitaxialfilmhasathicknessof390nmas
measuredbySEM.Theslate-likesurfacemodulationistypicalforM-planeGaN.
NoindicationofC-planeGaNcouldbeobservedatanypointwhenperforming
microscopyonthefilm.Thesurfacescratchesfromthesubstratearevisibleasline
discontinuitiesintheM-planeGaNfilminbothimages.Despitethesescratches,
asmallrmsroughnessof2.9nmisfoundinanareaof10×10µm2.Part(c)of
Fig.4.11showsthedepthprofileofthelineindicatedin(b)overarangeof10µm.A
peaktovalleyroughnessbelow5nminregionswithoutscratchesandsurfaceheight
variationsupto25nmatthelocationofscratchesisdisplayed.Itisexpectedthat
thesurfaceroughnesswillimprovewhenpolishingproceduresandgrowthconditions
areoptimized,therebyalsoincreasingtheGaNfilmqualitytremendously.
Toexaminethecrystalorientation,qualityandstrainstateoftheepitaxialGaN
film,x-raydiffractionscanswereperformed.Fig.4.12(a)showsanω−2θscaninthe
rangeω=13◦−38◦wherethe(1¯100)GaN,(100)LGOand(2¯200)GaN,(200)LGO
peakscanbeseen.Thisscanwasperformedinadoublecrystalconfigurationwith
aresolutionof0.005◦.Noothermajorpeakisobservedinthespectrum.Onlyan
extremelysmallhintofapeakaroundω=17.16◦withamaximumof7counts
couldbenoticed,whichmaybeattributedtoaverysmallfractionofC-plane

58

Figure4.9:(a)and(b):RHEEDimagesoftwoLGOazimuths,separatedby90◦
beforeannealingthesubstrate.ThesameazimuthsofLGOafteran-
nealing,shownin(c)and(d),aremuchmorepronounced,whichindi-
catesnamelya[11clean¯20]andsurface.[00The01],twafterogrocorrespwthondingcanbeMseen-planein(e)GaNand(fdirections,).A
streakypatternisobservedduetoaflatsurfaceofthefilm.Thespacing
ofthetheastreakslatticein[11¯constan20]tisdirectiongivengbivyesthethecGaNseparationlatticeofconstanstreakstinwhereas[0001]
GaNdirection.

theGaN.(1¯Ho100)wevGaNer,pthiseaksignalwithaisfarmaximtooumweakinfortensityfurtherof∼900analysiscountsanditiswithsaferesptoectstateto
avdetailederyhighmeasuremenphasetpuritofytheof(1M¯100)-planeGaNGaN.andP(200)art(bLGO)inpFig.eaks4.12withashowsresolutionamoreof
0.001◦usingatriplecrystalanalyzer.Thestrainstatecanbeestimatedfromthis
(200)measuremenLGOpteak.usingtheDisregardingrelativetwistshiftoroftilttheinside(1¯100)theGaNGaNpeakcrystal,withtherespGaNecttofilmtheis
relaxedtoroughly80%comparedtothemaximalpossibletransversedeformation
causedbythelatticemisfitbetweenLGOandGaNinthebasalplane.
Todeterminethefilmquality,therockingcurveofthe(1¯100)GaNpeakwasmea-
sured.Theresultisgiveninpart(c)ofFig.4.12whereaFWHMwithalarge
vexpalueectedof∼am2500uchloarcsecwerisvaluefound.fortheAlthoughFWHMgrowthconsideringconditionsthewerequalitnotyoftheoptimizedsamplewe
thersurfaceduefromtoSEMirregularitiesandAFMpropagateddata.TheintolargethefilmFWHMfromvaluescratcishesonprobablythemainlysubstrateei-

59

Figure4.10:Opticalmicroscopeimagesofthesubstratesurfacebeforegrowth(a)
andofthesamplesurfaceaftergrowth(b)atamagnificationof100×.
Thescratchesonthesubstrate,probablycausedbymechanicalpol-
ishing,areclearlyvisibleinbothcases.Thespotsoccurringmostlyin
thelowerpartofbothpicturesareduetoimpuritiesonthemicroscope
lens.

surface(seeFig.4.10)orcausedbycontributionstotheBraggreflectionfromthe
crackededgesoftheGaNfilm.Yet,thevalueismuchbetterthanarecentlyre-
portedstudyofM-planeGaNon(100)LGO,reportingaFWHMof4121arcsecfor
the(10¯10)GaNreflection[222].
AplanviewTEMsampleoftheM-planeGaNfilmwaspreparedbymechanical
polishingandsubsequentAr-ionmilling.Twocross-sectionalTEMsampleswere
fabricatedbyfocusedionbeam(FIB),onecutparalleltotheC-planeandone
paralleltotheA-planeasisillustratedschematicallyinFig.4.13.Fig.4.14shows
atypicalFIBcutsampleinanSEM.Thethinpartindicatedinthepictureisthe
regionforTEMinvestigation.ThesampleswereanalyzedusingaFEITecnaiF20
TEMoperatedwithanelectronaccelerationvoltageof200kV.
TheepitaxialrelationshipoftheM-planeGaNsamplecanbededucedfromFig-
ure4.15(a)and(b),wherethediffractionpatternstakenfromtheGaNfilmand
theLGOsubstrate,seeninFig.4.15(c),aredepicted.Fromthepatterns,itisclear
that[11¯20]GaN[010]LGO,and(1¯100)GaN(100)LGO.InFigure4.15(c),showinga
brightfieldimageofthesamplecutparalleltotheC-plane,ahighdensityofthread-
ingdislocationsisapparent.Thebrightareasinthesubstratelocateddirectlyat
theinterfaceofthesubstrateandtheepitaxiallayerrepresentholesintheTEM
samplecausedbytheelectronirradiationofthetransmissionelectronmicroscope.
SinceLGOisverysensitivetoelectronbombardment,itisverydifficulttoobtain

60

Figure4.11:AnSEMimageofthesamplesurfaceisseenin(a)displayingachar-
acteristicM-planeGaNslate-likemodulation.A6×10µm2AFM
scanisshownin(b),witha10µmlonghorizontallinedepictingthe
profilegivenin(c).Heightvariationsinthelineprofilecorrespondto
scratchesinthesamplesurfaceasindicatedbythedottedlines.

goodqualityhigh-resolutionimagesofthismaterial.Thisissuewasalsodiscussed
regardingRHEEDmeasurementsinthegrowthprocedureabove.
LookingatthesamplecutparalleltotheA-planeinFig.4.16(b),ahighdensity
ofpartialdislocationsassociatedwithahighdensityofstackingfaultscanbeseen.
Figure4.16(b)displaystheelectrondiffractionpatternoftheGaNfilmseenin
Figure4.16(a).Thediffractionspotsshowstreaksalongthe[0001]direction,giving
strongevidencetowardahighdensityofstackingfaultsinthefilm.
TheplanviewTEMsampleoftheGaNfilmallowsforanestimateofthedensityof
threadingdislocationsandstackingfaults.Acentereddarkfieldimageofoneregion
oftheplanviewsampleisseeninFigure4.17.Thenumeroussmalldotsrepresent
threadingdislocationsthathavereachedthesurface,whilestackingfaults,appearing
aslines,arefoundrunningperpendiculartothe[0001]directionasindicated,i.e.,
theylieintheC-plane.Theelongatedspotsinthediffractionpatterninsetin
Figure4.17pointtowardthepresenceoftwistedmosaicblocksinthefilm.The
twistanglecanbeashighas5◦.Inthissample,thethreadingdislocationdensity
wasfoundtobeontheorderof1×1011cm−2andthestackingfaultdensityaround
2×105cm−1.

Figure

4.12:

Pofarta390(a)

nmshowsthickanωGaN−

61

la2θyerx-ragroywnscanoninLGO.doubleOnlyMcrystal-planeconfigurGaNationand

(100)LGOrelatedpeaksareobservedstatingahighphasepurityof

theGaNM-plane.Ahigherresolutionω−2θscanwithatriplecrystal
analyzerintherangeω=16◦−16.8◦isseenin(b).(c)presentsthe
rockingcurveofthe(1¯100)GaNreflection,theFWHMofwhichlies

around2500arcsec.

Figure

4.13:

Figure

Illustration

4.14:

GaN

of

TEM

the

TEM

sample

samples

cut

yb

a

fo

aredprep

cused

Ga

for

ion

analysis.

b

eam.

62

Figure

4.15:

artsP

[0001].

direction

wingvie

with

sample,

-plane

M

the

fo

data

TEM

degree

of

ingthread

cationsdislo

is

visible

in

in

the

GaN

A

high

ec-resp

theelectrondiffractionpatternsoftheGaNfilm

(a)and(b)display

andtheLGOsubstrateseeninthebrightfieldimagein(c),

.elytiv

Thededucedrelationshipitaxialep

isrlierea

confirmed.

film.

63

Figure

4.16:

TEM

data

diffraction

imagefilm

in

the

of

the

M

-plane

and(a)pattern

(b).

sample.

The

data

sample,

with

viewing

direction

[11

¯20].

thecorrespondingbrightfieldM-plane

wssho

a

high

abundance

of

ingkstac

64

GaN

GaN

faults

65

Figure4.17:cationsInplane(spTEMots)samandplestacofkingtheMfaults-plane(lines)GaNreacsamhingple.totheThreadingsurfacedislo-are
twistvisible.presenThetiinnsettheshowscrystallinetheGaNfilmbydiffractionthecircularlypatternandelongatedindicatesspots.a

Thedislocationdensityinthissampleishigherthanvaluesreportedintheliterature
∼109cm−2[158]forM-planeGaN,e.g.grownonLiAlO2byMOVPE.However,
growthparametershadnotbeenfullyoptimizedonLGOandthefilmthickness
hereimpactistonwicetheasthreathindingasindislo[158].cationThedensitthicyknessasisofmenthetionedfilmisin[b223eliev]edwheretohathevevsomealues
for8M−2-planeGaNgrownonLiAlO2aregivenas109cm−2nearthesubstrateand
10cmnearthesurface.Thestackingfaultdensityinoursampleisalittlehigher
thanthebestvalueof104cm−1[158]reportedinliterature,butisinthesameorder
asthemostfrequentlyreportedvalues.

4.2.4GrowthofA-planeGaNon(010)β-LiGaO2

ThegrowthprocedureandconditionsforA-planeGaNgrowthonLGOwerees-
sentiallythesameasforM-planeGaN.Howeversmallchangesweremadeandfor
clarity,thewholeprocedureisdescribedhereagain.
TheepitaxialrelationshipofA-planeGaNonthe(010)surfaceofLGOisshownin
Figure4.18byaballandstickmodelofthe(010)LGOsurface.Thegrayrectangles
suggestpossiblenucleationsitesforA-planeGaNgrowth.Thelatticemismatch

66

Figure4.18:Ballandstickmodelofthe(010)LGOsurface.Thedottedrectan-
glesignalsthesizeoftheLGOunitcell,whiletherectangleswith
continuouslinesindicatepossiblenucleationsitesfor(11¯20)GaN.

¯for3.58%,therespectivrelationshipely.Gro[100]wthLGOonthe[1100](010)GaNplaneandof[001]LGO,LGOsimilarly[0001]GaNtogroiswth2.22%onandthe
(100)LGOplane,benefitsfromthelackofametal-non-metalpolarity.
aSiBeforewaferintrousingductionathintoInthelayMBEertocprohambvideer,fotherLGOhomogeneoussubstratesthermalweremouncoupling.tedonTheto
◦priorsubstrates,totransferplacedintointothegromolybwthdencumhamber.holders,Duwringeregrowthoutgasedactivforated60minnitrogenatN1302∗wasC
suppliedbytheplasmacell,operatedatagasflowof0.3sccmand450Wforward
powerforallsamples.Growthproceduresweremonitoredin-situusingRHEED.
ToestablishA-plane◦GaNgro◦wththesubstratetemperaturewasoptimizedinthe
torangeGabricethweenconditions.650CSevanderal800Cpre-groandwththeGatreatmenfluxwtsasofvtheariedsubstratefromslighintlytheGagroricwthh
chamberwerestudied.ThesewereGadesorption,nitridation,Gadesorptionand
nitridation,annealingandnotreatmentofthesubstrate.
◦GaNAnnealingdepthositionesubstrateresultedatin800successfulCfor60grominwthinofAthe-planegrowthGaNchamonberLGO.directlyGrobwtheforeof

67

GaNcommencedatslightlyGarichconditionsatasubstratetemperatureof700◦C.
Thetemperaturesgivencorrespondtothethermocouplereadings.Agrowthrateof
roughly60nm/hwasused.IrradiationofthesamplewiththeRHEEDbeamwas
reducedtoaminimumtopreventpossibledamagetothesample,i.e.RHEEDwas
onlyturnedonafter30minofGaNgrowthandinordertotakeRHEEDpictures.
Shortlyaftertakingthesampleoutofthegrowthchamber,theGaNfilmshowed
smallcracksattheedges.Atoohightemperatureintheareaofcontactbetween
thesubstrateandtheMoholderprobablyisthereasonforthesefeatures.However,
approximatelymorethan98%ofthefilmremainedstableanddidnotlift-offor
ks.cracwsho

CharacterizationofGaNgrownon(010)LiGaO2Samplesfrominitialgrowth
therunscaseshowofedGaNpeelingrogwnoffonofthe(100)LGO.epitaxialFollofilmwingfromsevtheeralsubstratedifferentaswsubstrateasalsoseenpretreat-for
mentapproachestopreventthis,wefoundthatannealingofthesubstrateleadto
thedesirableresult.WhileRHEEDpatternsoftheuntreatedsubstrateweredim
andhintedtoacontaminatedsurface,substratecleaningbyGadesorption,nitrida-
tionorbothtogethershowednoobservableeffect.WhenahighGafluxwasused,
assoRHEEDonasobservtheGaationsshshoutterwedwasanopened,increasinglyindicatedpbyolycrystallineringlikteypeelongadiffractiontedhorizonpatterntal
ots.spAmisfitpossibleofthereasonanisotropicforthepthereelingmaloffoexpansionfthecoepilaefficienyertsmabyetwhaveenebeenLGOtheandGaNenhancedat
hightemperaturessimilarasdiscussedinsection4.2.3.Moreover,damageofthe
substrate’ssurfacemayhavebeencausedbyimpinginghighlyenergeticelectrons
fromRHEEDmeasurements.ThepolycrystallineRHEEDpatternsmayresultfrom
bendingofthepartlypeelingofffilm,therebygivingtheimpressionofpolycrys-
.ytallinitInthediscussed.followingonlyCharacterizationtheofsamplethethatsamplewaswgroaswndoneatex-situoptimizedbyXRD,conditionsSEM,willAFMbe
TEM.andInFig4.19(a)-(e)RHEEDmeasurementsoffiveazimuthsaredepicted.Thetop
bpicturesottomshopictureswdiffractiondisplayelectpatternsronofthediffractionsubstratepatternsshortlyafterb3heforegrogrowthwthofGwhileaNthefor
thesameangles,respectively.Thefiveazimuthsofboth,theLGOandGaNcrys-
tal,couldreliablybeidentifiedintwowaystherebyprovidingavaluablemethod
ofconsistencycheck.First,therelativeseparationinanglewascomparedtothe
theoreticalangleseparation.Thesecondmethodinvolvedcomparingthequotientof
latedstreakvalues.separationThepofostthegrofivewthazimpicturesuthstoshoonewstreakyanotherimageswiththeofAcorresp-planeGaNondinghincaltingcu-

Figure

4.19:

68

¯pairRHEEDofpicturesimagesof(a)-(e)fivewazimasuthstakenofat(010)theLGOsameandangles,(1120)theGaN.topEaconeh
beforeandthebottomoneaftergrowthrespectively.Thecloselattice
matcindicatehbetawsmoeenoththesubandstflatrateandsurface.the(ffilm)isshowsevidenat.schematicStreakydrapatternswing
¯ofazimtheuths,[1120]similarGaNtohosurfacewitwasindicatingdonefortheM-planedirectionsGaNofbtheyWobservaltereited
et.allowsal[the139].assignmenCalculationtofanddirectionstomeasurementhetpictures.ofanglesandspacings

69

Figure4.20:(a)scratcSEMhespictureoriginatingofthefromGaNthesurfacesubstrateusingarea10kobservablemagnification.andinterceptDeep
AthelargecontinnuumitbyerofofthescratcGaNhesfilm.int(b)he1kfilmaremagnificationapparent.oftheGaNfilm.

towardsaflatandsmoothfilm.Thecorrespondenceofstreakseparationofthe
LGOandGaNforthesameangledisplaysthecloselatticematchbetweenthesub-
strateanditsepitaxiallayer.Fig.4.19(f)showshowthedirectionsoftheobserved
RHEEDpatternscorrespondtoazimuthsofboththe(010)LGOand(11¯20)GaN
surface.ThethicknessoftheGaNepitaxiallayerwasmeasuredinSEMtobe170nm.In
Fig.4.20,part(a)and(b)showSEMpicturesoftheGaNsamplesurfaceusing
magnificationsof10kand1k,respectively.ScratchesintheGaNfilm,presumably
causedbymechanicalpolishingofthesubstrateareclearlyvisibleinbothimages.
ThesesurfaceirregularitiesarealsoobservedinAFMimagesshowninFig.4.21.
ThedepthprofileofthehorizontallineindicatedintheAFMscanisgraphedbelow.
Whilethepeaktovalleyroughnesscanbeashighas75nminpartswith’deep’
scratches,anrmsvalueof10nmisobtainedonanareaof8×8µm2whenexcluding
thedeepestscratchesfromanalysis.Throughouttheperformanceofmicroscopyno
indicationofC-planeorM-planeGaN,i.e.theirtypicalcrystalshape,couldbe
observedinthefilm.Itisexpectedthatthesurfaceroughnesswillimprovewhen
polishingproceduresandgrowthconditionsareoptimized,therebyalsoincreasing
theGaNfilmquality.
ThecrystalorientationoftheepitaxialGaNfilmwasexaminedbyx-raydiffraction.
Fig.4.22(a)showsanω−2θscaninthewiderangeofω=12◦−78◦,performed
withoutanalyzer.AlltheapparentmainpeakscouldbeattributedtoLGOplanes.
Tworegionsofthespectrum,thatshowadditionalinformationwhenzoominginto
themaredisplayedinFig.4.22(b)and(c).Here,reflectionsofA-planeGaNplanes

Figure

4.21:

Anprofile

about
alongprofile

7along

10
horizonthe

2mµhorizon

AFM
blatalhorizon

image
kcblaline.

ofline.

Anabout7×10µm2AFMimageof
profilealongthehorizontalblackline.
scratcestdeeptheexcludingasareas

areas

as

excluding

the

deepest

theThe

sample
roughnessrmsThe

surfaceroughness

Thermsroughness
10ishesm.n

scratches

is

10

m.n

with
8aof

ofa

70

depth2mµ8

88×

µm

71

canbeseen.NotethatapartfromdiffractionpeaksattributedtoA-planeGaN
andLGOnootherpeakswereobservable.Thismeans,ahighdegreeofphase
(040)purityLGOwasfoundreflectioninandthedoGaNesfilm.thereforeThenot(11allo¯20)wGaNforanpeakexactisconvdeterminationolutedwithofthethe
¯pandeaklopwerositionintensitofythegivingGaNaslighsignal.tlyThelarger(22error40)GaNwhenreflectiondetermininghasitsapeakbroaderposition.shape
However,astrainstateoftheGaNepitaxialfilmcouldbeestimatedfromtherelative
orshifttiltoftheinside(22the¯40)GaNGaNpcrystal,eakwiththerespGaNectfilmtowtheas(080)relaxedLGOtopabeak.out50%Disregardingcomparedtwistto
themaximalpossible¯transversedeformationcausedbythelatticemisfitbetween
(010)LGOand(1120)GaN.
Twocross-sectionalTEMsampleswerefabricatedfromtheA-planeGaNsample.
withMecthehanicalCp-planeolishingastheandsampleAr-ionsurfacemilling,wwhileereusedFIBincuttingthewaspreparationusedtoofprotheducesamtheple
samplewiththeM-planeastheTEMsamplesurface(seeFig.4.23forillustration).
TheTEM,eacsampleshopweratereedanalyzedwithanusingelectronaJEOLaccelerati3010onTEMvoltageaswofell200asakV.FEITecnaiF20
TheA-planeGaNsamplecutperpendiculartothe[0001]direction,showninFig-
ure4.24,displaysahighdensityofthreadingdislocations.TheimagesinFig.4.24(a)
allelandto(b)theare[11brigh¯20]tfieandldtheimages[1¯100]takenindirection,thetworespbeamectively.conditionThewitinseththeingFig.vector4.24par-(a)
ofdisplaAys-planetheGaN.diffractionDislocationspatternthatofthehaveaGaNlaburgersyervshoectowingrparallelevidencetoof[11¯the20]grocanwthbe
observedinFig.4.24(a);thesearemixedandedgethreadingdislocations.InFig-
ureburgers4.24v(b)ectorsbothpureparalleltoscrew[000and1]edgeand[11dislo¯20],cationrespsareectivoutely[of5],conandtrastcansincethereforetheyhavnote
beseen.OwingtothemuchlowerdensityofvisibledislocationsinFigure4.24(b),
wecanstatethatmostdislocationsareofeitheredgeorscrewtypeandaminority
bpictureselongstoasthestraighmixedtlinestype.asInFig.indicated.4.24inTheseversionhaveandomainbinclinationoundariesof60◦appwithearinrespbothect
totheinterface,i.e.theylieontheothertwo{11¯20}planesofGaN.
Comparingin-zonebrightfieldimagesoftheM-andA-planefilms,thethreading
1dislo×1011cationcm−2.densityofthetwofilmsisonthesameorderofmagnitude,i.e.,around

Figure

4.22:

72

(a)GaNωla−y2erθgrowidewnonrange(010)x-rayLGO.scanwithoutDiffractionpanalyzereaksofofasev170eralnmLGOthicask
wellasA-planeGaNplanescanbeseen.Thereisnoindicationof
thediffractionregionωp=eaks27.of75◦an−y29.other75◦orienwheretedthecrystalclearlyplanes.visible(b)rightzoshomoulderinto
¯ωof=the72.(04075◦)−LGO76.75p◦.eakHere,canbetheattributeddiffractiontop(11eaks20)ofAGaN.-plane(c)zoGaNominandto
LGOliefurtherapartallowingforasimplestrainstateanalysis.

Figure

4.23:

Illustration

of

the

TEM

samples

prepared

for

analysis.

73

Figure

4.24:

tBrigh

field

images

thewithcondition

thesefordirection

ofpresence

are

visible.

of

g

the

GaNand(a)

=the(11A¯20)-plane(a)GaNand

sample(1=g

in(b).

gsample=(1¯100)taken(b).in

theThe

owtviewingThe

74

eambviewing

viewingThe

imagesis[0001].Threadingdislocationsandthe

ersionvin

domain

oundaries,b

gorderinb

kingstac

,sfault

Figure

4.25:

75

TEMdataoftheA-planesample,withviewingdirection[1¯100].(a)
DiffractionpatternoftheGaNfilmoftheA-planeGaNsample.(b)
and(c)showbrightfieldimagesofthesamesampletakeninthe
twobeamconditionswithg=(0002)andg=(11¯20),respectively.
Threadingdislocationsandstackingfaultscanbeseen,however,much
fewerstackingfaultsarepresentthanintheM-planeGaNfilm.

76

Fig.4.25(b)and(c)showbrightfieldimagesoftheA-planeGaNsamplecutper-
pendiculartothe[1¯100]directiontakeninthetwobeamconditionwithg=(0002)
andg=(11¯20),respectively.Damagetothesubstratebytheelectronbeamisagain
seen.Threadingdislocationsaswellasstackingfaultsarevisible.Slightlymoredis-
locationsappearintheimagewithg=(0002)indicatingthattherearemorepure
screwthreadingdislocationsthanedgedislocations.IncomparisontotheM-plane
GaNsample,muchfewerstackingfaultsarepresentintheA-planefilm.Thisobser-
vationisalsoconfirmedbythemissingstreaksinthediffractionpattern,shownin
Fig.4.25(a).Thismeansthatthestackingfaultdensityislowerthan∼105cm−1,
andthereforelowerthanvaluesreportedpreviouslyforA-planeGaNe.g.grownon
R-planesapphirebyMOCVD,3.83×105cm−1[160]orbyPAMBE,∼2×105cm−1
].224[

4.2.5SummaryandconclusionofGaNgrownonβ-LiGaO2

IthasbeenshownforthefirsttimethatM-planeandA-planeGaNcanbegrown
epitaxiallyon(100)LiGaO2and(010)LiGaO2,respectively,usingplasma-assisted
MBE.Thermalannealingofthesubstratebeforegrowthprovedtobeaprocedure
wellsuitedtocleanitssurfaceandseemstobenecessaryforobtainingstableGaN
films.Thisisacomfortablesolutionofcleaningasubstrate’ssurface,duetothelack
ofelaboratecleaningorpretreatmentstepspriortoepitaxialgrowth.Thegrowth
conditions,likepre-growthannealingandgrowthtemperaturewerealsousedina
recentlypublishedstudy,whichalsoshowsgrowthofM-planeGaNonLGO[222].
StreakyRHEEDpatternsaftergrowthinadditiontotheSEMandAFMdata
showverysmoothM-andA-planeGaN,despitethepresenceofpolishingrelated
surfacescratchesonthesubstrates.TheXRDresultsindicateahighdegreeofphase
purityforbothfilmorientationsandshowedthatthestrainwasrelaxedto80%for
theM-planeandto50%fortheM-planeGaNfilmascomparedtothemaximal
possiblestrainthatcouldhavebeeninducedduetothelatticemismatchbetween
filmandsubstrate.Ahighphasepurityofthegrownmaterialisaverydesirable
propertyofthefilmandsometimesposesgreatdifficultytoachieve.However,asthe
resultsshownaboveprove,growthofGaNonLGOstatestheeasinessofobtaining
highphasepurityofGaN,whichiscloselyrelatedtothecloselatticematchofthe
GaN.withsubstrateTheanalysisoftheM-andA-planeGaNfilmsbytransmissionelectronmicroscopy
showthattheepitaxialrelationshipofthefilmwiththesubstrateisinagreement
withtheresultsobtainedfromRHEEDandXRD.Threadingdislocationsandstack-
ingfaultsareseentobethemaindefectsinthefilms.F11orthe−2caseoftheM-plane
GaNsample,athreadingdislocationdensityof1×10cmandstackingfault
densityofabout2×105cm−1werefound.TheA-planesampleshowsathreading

77

dislocationdensityonthesameorder;however,amuchlowerstackingfaultdensity
ascomparedtotheM-planesample.Althoughthethreadingdislocationdensity
isvsubstrateseryhighusedandforthestheteroackingepitaxyfaultitcandensitiesbeareexpinectedantoequivobtainalentsmorangeotherasforsurfacesother
andsubstrateahigheraswellcrystalasgroqualitwthyofconditionstheGaNhavefilbmseenwhenthoroughlymechanicaloptimized.polishingOnceofthethe
technologicalissueofpolishingforasmoothsubstratesurfacehasbeensolvedit
wnon-pouldbolarevGaNeryinonterestingLGO.tooptimizethegrowthprocedureforthefabricationof

5CuincorporationinGaN

Cu-dopedgroupIIInitrides,asdiscussedinsec.2.3.3,haveearnedgreatinterest
duetotheexperimentalevidenceofferromagnetismobservedinGaN[117]and
AlN[225]atroomtemperature,henceservingasaprospectivecandidatefora
magneticspin-aligningelement,materialexcludesinspintheptronics.ossibilitTheyofusagemagneticofCuasaclustersdopanbuiltt,bywhicthehisadopannon-t.
CuThereforeorGaN,theobutccisaurrenceresultofoftheferromagnetisminteractioninbetGaN:Cuweendotheesnothostoriginmaterialfromandeitherthe
t.dopanatomsTheoreticalonworksubstitutionstudyingalGasites,ferromagnetice.g.[17,Cu-dop114,ed115,GaN226].hasForforemostinstance,WassumeduetCual.
[17]explainthatincorporationscenariosotherthanCusubstitutingGasitesarenot
takenintoaccountbecausetheyareenergeticallyunfavorable.However,sincemost
growthprocessesofdopedGaNdonottakeplaceinathermodynamicequilibrium
enSeongvironmenetal.t,[the18]ostatingccupationofunfaferromagnetismvorableinenergeticCu-dopedstatesGaNmaynanonotwiresbeshoexcluded.wby
anomalousx-rayscatteringthatCuatomsdoindeedsubstituteontoGasites,but
doextennottGacommensitestareonoccupinitrogenedorandinwhetherterstitialNsiteorointerstitialccupation.ItsitesisalsostillplayuncleararoletoinwhatCu
incorporation.Sunetal.[119]forexample,seethatthetotalmagnetizationoftheir
CuimplantedsamplesincreaseswithhigherdosesbutthemagneticmomentperCu
atomdecreases.ThisisspeculatedtostemfromanincreasedincorporationofCu
respatomsonsibleoninforaterstitialdiminishingsites.Hoofwevtheer,expmagneticerimenproptalertiesevidandenceofsubstituinterstitialtionalpCubositionseing
enhancingthemcanonlybeobtainedifthepositioningofCuintheGaNmatrix
isknomagneticwn.Tpropogeterties,insighthetintofundamenthetalinfluencesquestionofofthewhperelacemenandthowofCuCuistotheincorpsystemsorated
inGaNmustbeanswered.
ForthispurposeX-rayAbsorptionNearEdgeStructure(XANES)andX-rayLinear
Dichroism(XLD)measurementsofaseriesofCu-dopedGaNsampleswithdifferent
dopingelectronsconcenfromthetrations1swereorbitalinvofGaestigatedandatCu,theGarespandectivCuely,wK-edge.ereTexcitedhisbmeansyx-rathatys
thefromainnersyncelectronhrotronsbareeammainlyline.Atexcitedtheloinwtoestunoenergies,ccupiedi.e.statesattheoftheabsorption4porbitals.edge,

79

WhileinXANEStheedgepositionandshapeissensitivetotheformalvalence
state,ligandtype,andcoordinationenvironment,whichcanbeusedtoidentify
phaseselectronic[227],propXLDertiesofprothevidesCuaconatomsvinenientthetoolnitridetoprobhost,esincetheloitcalisproporstructuraltionalandto
thehighlyanisotropicsensitivetounotheccuploiedcaldensitsymmetryyofcandharge[giv228es].aTheuniqueXLDspsignatureectruministhecasethereforeof
wurtzitecomparingstructuthemreandeasurednodatasignaltosimamplitudeulationsforpcubicerformedcrystalbyuseosymmetryfthe[228].FDMNESBy
codesubstitutes[21],theoronincorpinterstitialorationsitesfractioncanbofeCudeduced.atomsinthenitridehostasGaorN
InatthisthecsynchapterhrotrontheexpinerimenGrenobletalproandceduredetailsofthetothesamplesimulationspreparation,(sec.5.1measurem)willenbtse
described.Thereafter,theexperimentalfindingsandtheirinterpretationwillbe
presentedinsection5.2byafirstsurfaceanalysisofthesamples(sec.5.2.1)followed
byhelptheofthepresentationcalculatedofsptheectraXANESinsec.and5.2.2XLD.dataFinallywhic,hsectionarethen5.3interpretedsummarizeswiththe
.studythisoffindings

5.1Experimentalprocedure

Theinvestigatedsamplesweregrownbyplasma-assistedmolecularbeamepitaxy
byPhilippGanzusingthesystemdescribedinsection3.3.2.AseriesofsixGaN:Cu
sampleswithvaryingnominalCuconcentrationfrom0%to2.69%,inthefollow-
inglabeledA(0%),B(0.50%),C(1.35%),D(1.77%),E(2.22%)andF(2.69%),was
produced.Theterm”nominalconcentration”usedherereferstothefluxratioof
CuatomstothetotalmetalatomfluxofCuandGa.Assumingsticking-and
incorporation-coefficientstobethesameforCuandGaandequaltounity,these
numberswouldrepresentanestimateoftheactualdopingconcentrationofCu.
However,theseassumptionsarenotjustifiedandhence,thenominalconcentration
canonlybeseenasameasureoftherelativefluxprovidedduringgrowth.Aswillbe
describedlater,asignificantpartoftheprovidedCuatomsformmetalcompounds
inthesamples,i.e.notalloftheprovidedCuisavailableasdopingmaterialinthe
GaNfilm.Inthisseriesofsamples,sampleCwith1.35%nominalCudopingtakes
inaspecialpositionduetothemeasuredferromagneticbehavior.Thedataisnot
shownhere,butitsmagneticbehavioraswellasthegrowthprocedureappliedis
verysimilartothecasereportedearlier[20].
Toobtainbetterthermalcouplingduringgrowth,C-planeorientedsapphiresub-
stratesweremountedontosiliconusingindiumasanadhesive.Eachsubstratewas
firstoutgasedinabakingchamberat130◦Cfor60minbeforeintroducingitinto

80

Figure5.1:SchematicrepresentationofpartofaGaN4×4×4supercell.(a),(b)
and(c)showhowtheCuatomsareplacedintheGaNmatrixoninter-
stitial,NandGasites,respectively.

◦thestepbgroeforewthchamnitridationber.Itwascommencedthenforheated180tomin890atC200for◦30C.AminthinforaAlNfinalbufferoutgasinglayer
in(thictroknessduced≈prior22±to4thenmasgrowthmeasuredofthebyepitaxialScanningCu-dopElectronedGaNlaMicroscopyer.yAgro(SEM))wthwrateas
ofroughly70nm/hwasused,resultinginGaNfilmthicknessesintherangeof
130-150nm.Afterunloadingthesamplefromthemolecularbeamepitaxy(MBE)
etcsystem,hingofitwtheascutsamplesintowasparts,poneerformedoftowhichremowvaseetcmetahedlcompandoneoundsleftfromunetcthehed.samplesThe
andobtainanincluded65%understandingHNO3ofthetreatmeneffetctforof5etcminhingandontherinsingsamples,withtheirdeionizedsurfaceswater.wTereo
investigatedbySEM.
FandorallCu,i.e.samplesaroundtheXANES10367eVandandthe8979XLDeV,spectrarespwectivereely.measuredTheseatthemeasuremenK-edgestsofwereGa
performedwithhardx-raysfromtheID12beamlineattheEuropeanSynchrotron
RadiationFacility(ESRF)detectingthetotalfluorescenceyield.Theexperimental
chambercontainseightsiliconphotodiodesarrangedinasemispherearoundthe
samplethatareequippedwith10µmthickNifilters.Aquarterwaveplatewasused
towhenflipptheerforminglinearptheolarizationXLDofmeasurementhesyncts,hrotronalsolighdescribtedfrominvdetailerticalinto[229].horizonThetal
angleofincidencewasbetween14◦and17◦withrespecttothesamplesurface.For
theilluminatedanalysisofsampletheareaXANESismucandhXLDlargerspthanectrathewesizehaveoftoakeepsurfaceincompmindoundthatandthe
alsomuchlargerthanthemeandistancebetweenthecompounds.Thex-raysignals
locollectedcally.Inareaddition,thereforetheanavharderagex-raofystheofthesamplesyncanddohrotronnotareprobeassumedpartsoftotpheenetratesample
thesamplescompletely,leadingtoaprobingofthewholevolumeofthesamples
insteadofonlythesurfacelayers.

81

theXANESmultipleandXLDscatteringspectrawformalismeresimwithinulatedthemusinguffin-tintheapproFDMNEScoximation.de[21T]oobtainapplyinga
dopingconcentrationofCuinGaNwhichiscomparablewiththeCuconcentration
inwiththeaexpcalculerimenationtalradiussamples,of10aA˚wuwasrtziteusedsup.GaNercelloflattice4×4×4,constani.e.tsaoftotala=of3.256188A˚atoms,and
c=5.185A˚wereused[124].Forthex-rayabsorptionspectraeightdifferentdoping
atomconcenintrationthessupwereercell)upcalculated,to6.25%.thelowTheestvpariationossibleofrangingdopingfromconcen0.78%tration(within1Cuthe
supercell,howeverhasalmostnoinfluenceonthecalculatedspectra.Duetothis
tofact,folloforw.reasonsConfigurationsofclarity,ofonlyCuinthethe1.56%GaNdopcrystaledspectrasubstitutingareGplottedainatoms,theNaanalysistoms
orasinterstitialsinthelargestvoidoftheGaNmatrix(seeFig.5.1)wereperformed.
ThetwoCuatomsareseparatedbythelargestpossibledistancetoavoidpossible
effects.clustering

conclusiondanResults5.2

DuringgrowthoftheCu-dopedGaNsamplesCuisnotonlyincorporatedinthe
filmbutalsoformscubic[230],non-magneticγ-Cu9Ga4structuresaccordingtoX-
RayDiffraction(XRD)[117]andTransmissionElectronMicroscopy(TEM)analysis
[231].BecauseofthedifferentplacementpossibilitiesoftheCuatomsinthesample
(intheGaNfilmorintheγ-Cu9Ga4structures)acleardistinctionoftheoriginof
theCusignalpresentintheXANESandXLDspectramustbeclarified.Forthis
purposemeasurementsonasgrownsamplesaswellasonsampleswherethesurface
compoundshadbeenremovedbywetchemicaletchingwerecompared.
Inafirststep,anestimateoftheCuincorporationintheGaNfilmandtheγ-Cu9Ga4
compoundsisdonebymeansofSEMsurfaceanalysis.Thisgivesabetterunder-
standingforthefollowinganalysisbecausetheintensityoftheabsorptionsignalis
proportionaltotheamountofatomspresentinaspecificchemicalconfiguration.
SinceXLDprobestheanisotropyofthevalencecharge,CuatomsinGaNgivea
wurtzitetypesignature,ife.g.substitutingGasites,butshouldgivenosignal
contributionwhenboundinγ-Cu9Ga4compounds,asaresultofthecompounds’
cubicsymmetry[232].Inasecondstepthex-rayabsorptionspectraareexamined.
FromthesurfacesensitiveSEManalysisthemajorityofCu,whichisincorporated
inthesample,isexpectedtobefoundintheγ-Cu9Ga4compounds.Thisobser-
vationisstillvalidafteretching,wheremostoftheγ-Cu9Ga4couldberemoved,
howeverleavingbehindbetween12%and35%ofthecompounds,dependingonthe
sample.Duetothehenceexpectedbiginfluenceofγ-Cu9Ga4compoundsonthe
absorptionspectrawefirstturntolookattheXLDspectra,whereonlyanon-cubic
coordinationofCugivesanon-zerosignal,tomakesureCuisreallyincorporated

82

inXLDthesignalGaNatfilm.theWCeushallK-edge,seeinprothevingfolloCuwingothatccupationwedointheindeedGaNobservcrystal.eanToon-zerogain
theconfidenceXANEStosptheectra,answwhicerhofarethenotactualdominatedCubplacemenythetCuinsignGaN,alweincorpthereafteroratedinanalyzeGaN
butaredefinitelyinfluencedbyit.XANESandXLDsimulationsofCu-dopedGaN,
ppoundserformedareforCucomparedplacementothetexponGa,erimenNtalanddataintoterstitialextractsites,numandbersofforγ-Cuthe9Ga4fractionscom-
ofdifferentlyoccupiedsitesinGaNbyCu.

5.2.1γ-Cu9Ga4surfacecompounds

AwithfirstrespectestimatetotheforfilmthecanamounbetdeduofCucedpresenestimatingtinthetheCusurface9Ga4compislandoundscoverage(Cu9Gausing4)
ofSEMthesamimagesples[(Fig.231].5.2)Anandimpanaortanvetrageassumptioheightnestimateincludedofin220thenmfollofromwingTEMevaluation,pictures
namelythatalltheprecipitatesvisibleonthesamplesurfacesareonlycomposed
ofγ-Cu9Ga4,isjustified◦bythefactthattheyareallformedduringgrowth,i.e.at
atemperatureof790C(bythermocouplereading).Thiscanbeassumedbecause
bysomeTEMofthedatacomp[231].oundsFhaurther,vebaeenomeanvergroCuwntobyGaathinratioGaNwithfilm32.78%andisGasuppintheseorted
compoundsisassumedinaccordancewiththephasediagramgivenin[233,234].
Adecreaseinsurfacecompoundsonthesamplescouldbeachievedbyetching.This
isclearlyseenforsampleFinSEMimagestakenoftheasgrown(Fig.5.2(a))
andcirclestheetcsurroundhedthe(Fig.5.2structures,(b))issample.anexampTheleleftofhopartwtheoftheestimatedpictures,surfacewherecovblaceragek
bofyCuCu99GaGa44wwasastak8.22%eninoftotheaccountotaltinsamplethecalcarea).ulationsThe(insamplethiscaseaftertheetchingareacoshovwereded
andoped88%samplesdecreasewithofthetheircompnominalounds.CuTableconcen5.1trationgivesandanoinveacerviewhcaseofallthethevaluesCu-
ofthevolumepercentageofCu9Ga4inthesamplebeforeandafteretchingalong
withtherelativeandabsolutedecreaseoftheamountofcompounds.Thelasttwo
columnsgivethecalculatedratioofCuatomsboundinthecompoundswithrespect
tofilm,theasittotalisknometalwnatomonlyconfortentsampleoftheC).Thesamplefact(excludingthatthetheamounamounttandofsizeCuofinthethe
Cuconcen9Ga4trationcompisoundsclearlyonretheflectedsampleinthensurfaceumbersincreasesofTablewith5.1.increasingnominalCu
NotethatthecalculatedratioofCu,formingcompounds,tothetotalmetalcontent
isalreadyhigherthantheprovidedCutometalfluxratioduringgrowth.Thisis
aratioplausibasislebuiltoutcomeintobtheecausesample.theproItvidedisafluxclearratioofindicationCuandthatGatheissticnotking-thesameand

Figure

5.2:

83

SEMimageofsampleF(2.69%)beforeetching(a)andafteretching(b).
Ontheleftsidesofthepicturesthesurfacecoverageoftheprecipitates
isestimatedbyapproximatingthestructures’areawiththeareaof
multiplesmallcircles.Forthissample,an88%decreaseinthesurface
coverageofthesurfacestructuresiscalculatedfromthepre-etchedto
thepost-etchedconditionofthesample.

84

Table5.1:ListingoftheCu-dopedGaNsampleswiththeestimatedγ-Cu9Ga4cov-
eragebySEMmeasurements.
samplenominalVCu9Ga4/VsampleVCu9Ga4VCu9Ga4#Cuat.inCu9Ga4/
conc.[%]decreasedecrease#metalat.[%]
3][mm[%][%]unetchedetchedunetchedetched
3.47n.a.1.460.50BC1.353.190.8179.743.47×10−47.261.95
D1.775.723.0265.434.13×10−412.276.89
E2.226.392.1574.846.48×10−413.515.01
F2.6911.441.5688.0315.88×10−421.783.68

incorporation-coefficientsofCuandGaonGaNaredifferent.Whilethesticking
coefficientforCucanbeassumedtobenearunity,becauseofitslowvaporpressure
atthegrowthtemperatureusedinthisstudy[235,236],itishigherforGa[237].It
isthereforeimpossibletodeduceareliablenumberfortheCuconcentrationinthe
GaNfilmbythiskindofanalysis.However,usingresultsobtainedbywavelength
dispersivex-rayspectroscopy(WDXS)ona1.38%nominallyCu-dopedsample,
where0.26%ofthemetalatomsintheGaNfilmwereCuatoms[231],wecandeduce
a’real’dopingof0.24%CuforsampleCassumingthesameCuincorporationrate
forboththesesamples.Thismeans,consideringthepresenceofCu9Ga4compounds,
thatapproximately3.1%oftheCuatomsinthesampleareincorporatedintothe
GaNfilm,whereastheremainingroughly96.9%arefoundinthecompounds.The
situationappearsreversedwhencalculatingthepercentageofGainthefilmandin
thecompounds,wheretheestimatednumbersare96.5%and3.5%,respectively.
AsafirstoutcomeoftheseestimateswecanstatethatthemajorityofCuthatis
presentinthesamplesisincorporatedinthecompoundofγ-Cu9Ga4.Byetching
itwaspossibletoremovemostofthecompounds,butnotto100%.TheXANES
spectraarethereforeexpectedtobedominatedbysignalsoriginatingfromthecom-
pounds.XLDhowever,isinsensitivetocubiccoordinationofboundatomsand
shallthereforebefreeofasignaloriginatingfromthecompounds.Inthefollowing
theanalysisofXLDattheGaK-edge,followedbytheCuK-edgewillbeshown.
Thereafter,insection5.2.3,theXANESspectraatbothedgeswillbediscussed.

5.2.2X-raylineardichroismofCu-dopedGaN

XLDsamplesatatthetheGaGa(Fig.K-edge5.3In(a),XLD(b))spandectraCuK-edgerecorded(Fig.for5.3the(c),etc(d))hedweandcanunetcclearlyhed
seeones.anincreaseRegardingofthetheGaXLDedgetsignalheforincreasetheetcofhedsignalsamplesamountstocomparedanavtoeragetheofunetcabouthed

Figure

5.3:

The

measured

XLD

ectrasp

depicted

here

erew

normalized

to

the

85

k-bac

(b)groundareatthetheXLDlowestspectraandofthethehighestGaK-edgeenergiesofpriorthetospandectrum.afteretc(a)hingand

ofthesamples,respectively.(c)and(d)showtheXLDspectra

pCueaksK-edgearebmarkeforeed

afterand”yb”.×

hingetc

of

the

es,sampl

elyectivresp.

theof

Bragg

86

Figure5.4:etcThehedinandtensityunetcincreasehedofsamplestheisXLDshopwneaksasatathefunctionGaofK-edgethebetwnominaleen
dopingconcentration.TheincreaseisduetotheremovalofCu9Ga4
compounds(seetextfordiscussion).

7%to33%,dependingonthesample.ThisbehavioroftheXLDsignals’intensity
increaseduetoetchingforthedifferentsamplesandpeakscanbeseeninFig.5.4.
Theintensityvariationbetweenetchedandunetchedsamplescanbeexplainedby
thefactthatafteretchingcubicsurfacecompounds(Cu9Ga4)havebeenremoved
andahigherpercentageofthesignalcollectedoriginsfromthewurtzitefilm,hence
givingmoreintensepeaksintheXLDspectrum.Thenon-uniformityintheincrease
oftheintensitycanbeascribedtothedifferentamountofcompoundsremovedfrom
thedifferentsamples,aswasalsoseenintheSEManalysisabove(seetheabsolute
VCu9Ga4decreasecolumninTable5.1).Thetendencyofhighernominallydoped
samplesgivinglessincreaseoftheXLDsignalispresumablyduetotheshortetching
timeapplied.ThismeansthatbecausemoreCusupplyduringgrowthleadstomore
compounds,alongeretchingtimeisneededtocompletelyremovethecompounds.
However,forallsamplestheetchingtimewasconstantleadingonlytoapartial
removalofthecompoundsandinconclusiontoahigherpercentageofremaining
Cu9Ga4onthesampleshavingbeenexposedtohigherCufluxes.
ThesharpBraggpeaksinthegraphsofFig.5.3aremarkedby”×”andoccur
becausesomecrystallatticeplanesinthesamplessatisfiedtheBraggcondition
withrespecttothepositionofsomeofthedetectorsandyieldnoinformationwhich
isofinteresthere.Apartfromtheslightvariationinintensity,anexaminationof
differencesintheGaK-edgeXLDsignalindependenceoftheCudopingcannot
beobserved.However,comparingtheexperimentalXLDdataoftheGaK-edgeto
simulations(Fig.5.5(b)),wecanstatethatonlyaverysmallpercentageofCuatoms

87

islocatedonNsubstitutionalsites.Thisconclusionisdrawnfromthequalitatively
greatlydifferingcurveprogressionofthesimulateddatafor2Natomshavingbeen
replacedby2Cuatomsina4×4×4supercellcalculation.Aclearanalysisof
paossiblequanattitativthisestage,measuredueoftoCutheinatomsaccuracyoccupinyinmogdeliGa,ngNortheseinspterstitialectra.Thesiteseffectisnotof
theinaccuracycanbeshowninFig.5.5(a)wheretheexperimentalandsimulated
dataofpureGaNaredepicted.Althoughafairlygoodqualitativeagreementis
Gaapparenorint,itterstitialshowssitesthatintheFig.fit5.5is(b).notgoodenoughtodiscriminatebetweenCuon

ofXLDtheatsamplestheCurecordedK-edgeattheTheCumuchloK-edgewersignalcomparedtotonoisetheratioGaeofdgethedataXLDissignalsclearly
visibleinFig.5.3(c)and(d).Nevertheless,theincreaseofthesignalfromthe
ofunetccomphedoundstothepresenetcthedinthesamplessamples,isapparentherebt,ybagaeinginrespresultingonsiblefromforalesslowerabsorptfractionion
ofx-raysinthecubicstructure(notcontributingtotheXLDsignal).Ingraph(c)
ofFigure5.3,sampleBseemstodivertsignificantlyfromtherestofthesamples.
Thisishowevernotconsideredtobeofphysicalorigin;ratherthebehavioristraced
thebacktopresencedifficultoftwiesoinintensereliablyBraggpnormalizingeaks(marktheeddataby”×acquired”intheforthisgraph)sampleandtheduelotow
forstatisticsimpracticablecomparedintotensittheydatacomparisons.oftheotherDuetosamples,thegenerallythereforelowbeingstatisticsrespofonsiblethe
dataamountforofallCuinsamples,theGaNresultingfilm,afromcomparisonlimitedofbeamthetimesignals’andinatensitieslowiwithncorpresporatection
tothedifferentdopingconcentrationisunfeasible.However,aclearstatementof
thepresenceofanXLDsignalispossible,leadingtotheconclusionthatCuisreally
incorporatedintheGaNfilmandisnotmerelypresentintheformofCu9Ga4
ounds.compcult,AlthoughanideatheofstatisticswhereofmosttheofexptheCuerimenatomstaldataaremakplacedesainquantheGaNtitativehostlatanalysisticediffi-can
beextractedwhencomparingthenon-zeroXLDexperimentsignaltothesimula-
tionsgraphis(Fig.5.6missing).Winetheobservexpethaterimenthetalindata.tensefirstHenceanegativlargeeppeakerceninthetageCuofonCuNonsitesN
sitescansafelybeassumedtobelimitedtoaverysmallamount.Similarly,theCu
onGaandinterstitialsitesmakeacontributiontothisfirstnegativepeak,butwith
muchlowerintensity,thereforenotcontradictingwiththepossibilityofasignificant
pdata.ercenIntageofterestinglythese,theconfigurabestmtionsatcbhingeingsppectraresenttobuttheexpunresolverimenedtalinthecurveexpistheerimenXLDtal
signatureofCuoninterstitialsiteswheretheshapeofthemaximacorrespondwell
withthetwopeaksinFig.5.6(d).

ones.(a)showstheundoped

differ-forectrasp

graphofsampleC(1.35%)incomparisontocalculated

ltaerimenexpthe

entCusitepossibilities.Themisfitoftheredcurveto

atoms.

88

Figure

5.5:

data

indicates

very

little

N

site

ccupationo

of

the

Cu

simulated

dmeasure

the

the

to

signal

XLD

K-edge

Ga

spectra,while(b)displays

redusmea

the

of

Comparison

Figure

5.6:

(a)

to

eddop

(c)

tdepic

the

Cu

K-edge

XLD

GaNonGa,Nandinterstitial

ofdata

tpresen

Csample

in

(d)

ear

ulationsim

results

for

1.56%

89

Cu-

sites,(d)showstheexperimental

(1.35%).Asaguidetotheeyethetwomainpeaks

edmark

yb

dotted

lines

houtthroug

the

other

graphs.

90

Figure5.7:ExpdifferenerimentlytaldopedXANESsamples.spectra(b)ofandthe(d)Gasho(a)wtheandCuexpe(c)rimenK-edgetalforXANESthe
dataferentofsitessampleforCtheGa(1.35%)andCualongK-edge,withcalcrespulatedectivelysp.ectraforCuondif-

5.2.3X-rayAbsorptionNearEdgeStructureofCu-dopedGaN

XANESattheGaK-edgeTheexperimentalXANESdataofthesamplesat
theGaK-edge(Fig.5.7(a))showhardlyanyvariationforthedifferentdoping
concentrations.Thisisexpected,becausetheamountofCuincorporatedinthe
latticeissmallleadingtoevensmallerdifferencesintheincorporationamountfrom
onesampletoanotherwhichshouldnotaffecttheGasignalofthefilmagreat
deal.ThelargespikesinthehigherenergyregionofsamplesA(0.00%),B(0.50%)
andD(1.77%),markedby”×”inthegraphareagainBraggpeakswhichareonly
avoidablewhentiltingthesample.Here,however,thepeaksarewellawayfromthe
regionofinterestinthespectrum,therebynotcausinganyconfusionidentifyingthe
signal.XANESreal

91

InFig.5.7(b)thecomparisonofoneoftheexperimentalcurves(sampleC)with
thesimulateddataisshown.Here,simulationresultsareshownforpurewurtzite
GaN(0%Cu)aswellasforGaNwith2Cuatomsina4×4×4supercell(1.56%
Cu)replacingGaatoms,Natomsandsituatedasinterstitials.
ThecomparisonofthecalculateddatawiththeexperimentaloneinFig.5.7(b)
displaysgeneralagreementbetweenthetwoinanumberofpoints.Themainpeak
around10383eVisverywelldescribedbyallsimulatedcurvesapartfromthe
curveforCuonNsites,whichisshiftedtohigherenergiesby2eV.Onemight
beadmixturetemptedoftoCuoexplainccupationtheofhighNsitesenergyinGaN.shoulderThisofthecaseismainreppeateeakdbyforaptheossiblenext
peakataround10395eV,whereagainthehighenergyshoulderintheexperimental
datacouldbeexplainedbythesubstitutionsofCuonNsites.However,asthese
shoulder-peaksintheexperimentalspectrumarealsopresentintheundopedGaN
spectrum,wenofurtherneedtoanalyzetheoriginofthesefeatureswithrespect
toCudoping.Weassumethelackofprecisematchingcomesaboutbecauseofthe
doesinaccuracynotofnoticeablethetheoryaffectandthespclaimectrum.thattheFlourther,wptheercensimtageofulationsCuinfailthetoGaNdescribfilme
moredetailsofthemeasurement,mostnoticeablythelowenergyrisingedgeofthe
spectrumwhereadistinctpeakintheexperimentaldataisseen.Thispeakalso
appearsinthecalculationsbutisratherashoulder.Itismostpronouncedwhen
CuoconcludedccupiestoinbedueterstitialtoCu,sitesforinthethesameGaNlattice;argumenitstasexactaboveorigin,holds,howevnamelyer,thatcannotitbise
alsopresentintheundopedsample.Gaboundinγ-Cu9Ga4compoundscannotbe
seentoinfluencethespectrumasthedoublepeakfeatureinthecalculatedCu9Ga4
spectrumiscompletelyabsentintheexperimentaldata.ThecalculationofCu9Ga4
wasperformedforoneunitcell(52atoms)andaclustercalculationradiusof10˚A.

XANESattheCuK-edgeTurningtotheXANESspectraattheCuK-edge
(Fig.5.7(c)),asinthecasefortheGaK-edge,thereisnosignificantdifference
visiblebetweenthesampleswithdifferentCuconcentrations.Thisoutcomeisnot
surprisingrememberingtheconsiderationsofsection5.2.1involvingtheSEMim-
ages,wherewestatedthatotherthanfortheGasignal,theCusignalmainlyhas
itstheGaNoriginfilm.fromThethespCuectrumincorpoforatedsampleinBthemaCuy9Gaseem4tocompdifferoundsfromandthenotrest,frombutCuwithin
theregardothers,tothealsolowerapparenstatisticstintheinlothewerexpsignalerimentotalnoisedataforratio,thisthesamplesmalldeviationcomparedtoto
therestofthesamplescanbeconsideredinsignificant.
Fig.5.7(d)depictstheexperimentaldataaswellasthesimulateddatafor1.56%
Cu-dopedGaNwhereCuissittingonGaorNsubstitutionalsites,asinterstitials
orisboundinCu9Ga4.Itcanclearlybeseenthattheshapeoftheexperimental

92

Figure5.8:FittingoftheexperimentalXANESgraphofsampleC(1.35%)with
differentcalculatedCusitepositions.

datamostaccuratelyfitsthecalculatedCu9Ga4graph.However,theshoulderin
therisingedgeofthemeasureddataisnotaccountedforintheCu9Ga4curve.
ThisaspectcanbeexplainedbyaslightportionofNsitesbeingoccupiedbyCu
atoms.UnfortunatelyascanbeseenbythestronglyCu9Ga4dominatedshapeofthe
experimentaldata,signalsofanadmixtureoftheCucomponentcomingfromthe
CuatomsincorporatedintheGaNfilmarehighlysuppressed,i.e.anexperimental
curvefitwithacontributionofoneormoreoftheGaN:Cucurvescanbeestablished
butissubjecttoalargeuncertainty.
Althoughanaccuratefitisnotpossiblefortheentirespectrum,atendencyofCusite
occupationcanbeextractedwhendeconvolutingtheXANESCuK-edgespectrum
(Fig.5.8)byconsideringthefourpossiblesignalorigins(CuonGa,N,interstitial
sitesorinCu9Ga4).Agoodoverallfitoftheexperimentaldatawasachievedfora
placementofCuatomsat4.0%Ga,0.6%Nand1.1%interstitialsites;theremaining
94.3%ofthesignalisattributedtoCu9Ga4.ThismeansthatforCuinthefilm
70%isonGa,20%oninterstitialand10%onNsites.Althoughthesenumbersare
valuesforthebestfitwhendeconvolutingtheexperimentalspectrumtheerrorbars
areratherlarge,i.e.similarfitscanbeobtainedwithadifferentoccupationofCu
inGaN.Tofindarangewherethisispossible,thedatawasfitwiththemaximum
tolerableCufractionineachpositionofCuonGa,Nandinterstitialsites.Theresult
isthatCucanoccupybetween0-6.2%Ga,0-0.8%Nand0-6.8%interstitialsites

93

Figure5.9:p(a)ossibleshowssitesthe(Ga,simN,ulatedinXLDterstitial)datathatofCuGaN:Cucanoccupwithy,amaccordingixtureotofthethe
Fig.same5.8.fractions(b)displaofconysthetributionsexperimenastaldescribXLDedforcurvetheofXANESsamplespC.ectrain

inGaNwithinthesensitivityofourmeasurementandisto93.2-100%presentin
Cu9Ga4compounds.
GoingbacktothesimulationofXLDspectraattheCuedgeandplugginginthe
valuesofCuoccupationonthedifferentGaNsitesextractedfromXANESweexpect
toseeagoodfitwiththeexperimentaldataiftheXANESfitwasreasonable.The
simulation,wherethecompositionofthesignalisthesameastheonedetermined
bytheXANESdata,andtheexperimentaldataareshowninFig.5.9(a)and(b),
respectively.ItcanbeseenthatthecalculationdescribesthemeasuredXLDdata
wellwithinthesensitivityofthemeasurement.Further,theorderofmagnitudeof
thesignal’sintensityiswelldescribedinthiscasewherethemajorityofthesignal
isassumedtobeabsorbedbytheCu9Ga4compounds.Thefactthatthefitresults
oftheXANESspectracorrespondwelltotheXLDspectraisanimportantcheck
forconsistencyandmakesthedata,inspiteofthelowsignaltonoiseratio,reliable.
TheresultsfromthisanalysisroughlycorrespondtotheresultsfromtheSEM
considerationsabove,i.e.fromXANESweconcludethat5.7%oftheCuinthe
sampleisinthefilmwhereasfromacombinedSEMandWDXSanalysisitwas
anestimatedfractionof3.1%.Duetothenumerousassumptionsleadingtothese
numbersthediscrepancyisnotsurprising.Nevertheless,importantinformation
canbegained:Cuisincorporatedinthefilm;CuisnotonlyincorporatedonGa
sites;CuoccupiesNsitestoameasurableextent.Thisisafindingthathasnot
beenpreviouslyreportedandmayleadtonewtheoreticalconsiderationsofdifferent
defects,otherthanonlyCusubstitutionsonGasites.

5.2.4Consideringstraininγ-Cu9Ga4compounds

94

TEMinvestigationshaverevealedepitaxialgrowthofCu9Ga4compoundsonGaN
[231]whichindicatesthepossibilityofCu9Ga4beingstrainedandtherebydeviating
fromitscubicstructure.ThismeansthatstrainedCu9Ga4couldalsoberesponsible
fortheXLDsignalthathasbeenobserved.ThereforeXANESandXLDspectraof
strainedCu9Ga4isalsoinvestigated.Differentmagnitudesofbiaxialcompressive
strainhavebeenconsideredrangingfrom0.5%to20%.
AchangeintheCu9Ga4structureisexpectedtoinfluencetheXANESspectrumat
theCuedge.Indeed,theshapeofthegraphandthemainpeakpositionchanges
withvaryingstrain.Bestagreementtothemeasureddataisgivenintherangeof
13-15%strain,whichisasignificantamountofstrainandshouldbeexperimentally
observablebye.g.XRD.However,XRDaswellasinTEMmeasurements,where
suchalargestrainofthecompoundsshouldbevisible,shownoindicationofstrain
inthecompoundswithinthemeasurementsensitivity.
AfurtherindicationthatacontributionofstrainedCu9Ga4tothespectraishighly
improbableistheincreaseoftheXLDsignalfortheetchedsamplescomparedtothe
unetchedones,thathasbeendiscussedaboveandisdepictedinFig.5.4.Ifstrained
Cu9Ga4wereresponsibleforalargepartoftheXLDsignal,adecrease(ratherthan
anincrease)ofXLDsignalintensityafteretching,duetopartialremovalofthe
Cu9Ga4compounds,wouldbeexpected.
Inadditiontothesearguments,XLDspectraofsimulationswereperformedforup
to20%strainedCu9Ga4togetanestimateoftheinfluenceontheXLDspectraof
strainedCu9Ga4.TheresultsareshowninFig.5.10(a)and(c)togetherwiththe
experimentaldataofsampleC(b)forcomparison.Noneofthecalculatedspectrais
ingoodagreementwiththeexperimentalcurveandhencethepossibilityofstrained
Cu9Ga4,alsofromthisviewpoint,isverysmall.
AlltheseobservationsshowthatthepresenceofstraininCu9Ga4,duetoepitaxial
growthonGaN,isnegligiblefortheanalysisofthepresenteddata.However,in
ordertoestimatethesensitivityofthefitpresentedinFig.5.8ontheshapeofthe
simulatedCu9Ga4spectrum,theexperimentalXANESspectrumattheCuedge
wasfitted.Fig.5.11showsthisfitconsidering14%biaxialcompressivestrained
Cu9Ga4.ThechangeinCuoccupationdistribution,withthebestfitvaluegiven
inbrackets,isasfollows:0-10.6%(3.3%)forCuonGa,0-0.5%(0.5%)forCu
onN,0-7.7%(2.2%)forCuoninterstitialsitesand89.4-100%(93.9%)forCu
inmetalcompounds.RegardingCuinthefilmonly,thismeansthat54.5%ison
Ga,36.4%oninterstitialand9.1%onNsites.Introducingstraintothecompound,
thedifferenceofGaorinterstitialsiteoccupationplaysalessimportantroleforthe
graph.theofeshap

Figure

5.10:

Comparison

of

the

XLD

signals

of

strained

Cu9

Ga4

theexperimentalcurve(b)ofsampleC(1.35%)at

Noneofthecalculatedspectracorrespondwellwith

one,

indicating

the

absence

of

strain

in

.GaCu49

(a)

the

the

and

Cu

(c)

95

and

K-edge.

talerimenexp

96

Figure5.11:FittingoftheexperimentalXANESgraphofsampleC(1.35%)with
differentCusitepositions,wheretheCu9Ga4compoundsaresubject
to15%biaxialcompressivestrain.

Comparingthefittingresultsofthestrainedandunstrainedscenario(Fig.5.8and
5.11),wefindthatinbothcasesmostlyGapositionsareoccupiedbyCu,followed
byinterstitialsandabout10%arepresentinN-likecoordination.ThevaluesforGa
andinterstitialsiteoccupationaresubjecttoaratherlargeerror;however,because
ofthegreatlydifferingshapeofthespectrumforCuatomsonNsites,thisvalueis
consistentinbothscenarios.

5.3SummaryoftheCuincorporationstudyinGaN

AseriesofetchedandunetchedCu-dopedGaNsampleswithdifferentdopingcon-
centrationshavebeeninvestigatedbyXANESandXLDmeasurementstoevaluate
whereCuatomsareincorporatedintheGaNmatrix.Theexperimentaldatashows
noapparentdependenceonthenominalCuconcentrationinthesamples.Theob-
servedXLDsignalattheCuK-edgeprovestheincorporationofCuinGaNand
indicatesthatinterstitialpositionsinGaNareoccupiedbyCu.TheXANESspectra
attheCuK-edgearedominatedbythesignalcomingfromγ-Cu9Ga4compounds.
WiththeknowledgethattheXANESspectramustalsobeaffectedbyCupresent
inGaN,afitoftheexperimentalXANESdatawasperformed,takingintoaccount
apossibleplacementofCuintheGaNcrystalonGa,Nandinterstitialsites.The

97

sites,findingsaboutshow20%thatcanabelargefoundmajoritonyinoftheterstitialCuandatoms10%(∼on70%)Nsites.takeGaXLDsimsubstitutionalulations
spofectra,GaN,withcomparedaCutositetheoexpccupationerimentalscenariodataasconfirmextractedthewithresultstheinhelpaofconsistenthetXANESway.
AcontributionofstrainedCu9Ga4totheXLDsignalisruledoutbyanumberof
ofobservsimulatedations.XLDTheinspectracreaseforofXLDstrainedsignalCuinGatensitwithyaftertheexpetchingerimenandtalthedatamismatcindicateh
49theabsenceofstraininCu9Ga4.ThisisalsoconsistentwithearlierXRDandTEM
results,whichfoundnoindicationoflargestraininthecompounds.
Althoughthefindingshereclearlypointintoadirection,onehastobecautiouswhen
forPattemptingAMBEtogrowngeneralizesamplestheinresults.GarichWeshouconditionsldkeepwithinmindcertainthatprothevidedresultsCutoapplyGa
fluxratiosunderspecificgrowthparameters.Becauseofthedifferenttechniquesand
stepmethoindsavstudyingailablethisforthematerialsyn.InthesisvofestigationsCu-dopofeddifferGaNentthesamplepresentgrowwthorkstatesconditionsone
(Nrichcondition)andsampletreatments(post-growthannealing)arecurrently
underwaytogetinsightintotheinfluenceoftheseonthecharacteristicsofGaN:Cu
samples.Nevertheless,thefindingthatalargefractionofCuatomsinGaNoccupiesGasites
hassuppalsoortsbtheeenvshoaliditwnyofthatatheoreticalsignificantstudiesportionbasedofondopanthistsisassumption.situatedonHoinwevterstitialer,it
theandmagneticnitrogenbsites,ehaviorgivofingGaN.risetoTotheanswerquestionthisifquestioandnhowandtothesegainCuaatommoresthoroughinfluence
understandingofthematerialsproperties,futuretheoreticalstudiesareencouraged
toalsoconsiderCuatomsonpositionsotherthanGasites,whichareshowntobe
occupiedinthiswork.

rySumma6

ofWithinnitridetheframewsemiconductorsorkofwthisastbuilthesis,andanewbroughtMBEtoopsystemerationfordedicatedroomtotempthegroeraturewth
ofspinchargetronicscarriers,applicationshighspinsuchpasolaspinrization,LEDs.i.e.Foreffectivsuchedevicesspin-alignmenefficient,traswecomellasbinationhigh
GaN,aspin-injectionroomtempefficiencieseratureareferromagneticdesirable.DMS,Therefore,wasincorpstudiedorationandgroofwthCuinofnon-pCu-dopolared
GaN,leadingtostrongeremissionefficienciesduetotheabsenceofpolarization
chargesalongthegrowthdirection,therebyavoidingtheQuantumConfinedStark
Effect,wasinvestigated.
LiAlOsubstrateswereusedtofindappropriategrowthconditionsandparameters
fornon-p2olarGaNgrowthonthenewlyinstalledsystem.However,LiGaOsub-
strateswerethemainfocusofthisthesis,duetothepromisingimpactof2close
latticematchoftwodifferentLiGaO2surfaceorientationstonon-polarM-andA-
planeGaN,respectively.Non-polarGaNgrowthonLiGaO2byMBEhadnotbeen
investigatedpriortothisstudy.
GoodqualityGaNgrowthonLiAlOsubstrateswasaccompaniedbythepresenceof
Gadropletsonthesamplesurface,2whichhowever,couldeasilyberemoved.M-plane
theGaNbestcouldrepbeortgroedwnresultswithinveryliterature.highpurFityurthermandgooore,dpcrystalolarCqual-planeity,GaNcomparablecouldbtoe
achievedbyalterationofthegrowthparameters,butgrowthofthisGaNcrystal
Cdirection-planewphaseasseenpurittoybecouldlessbeacsuitablehievedforwithgroawthcrystalonLiAlOqualit2.ybNevetterertheless,thanrep97orted.5%
inliteratureforMBEgrownC-planeGaN.
GrosubstratewthofLiGaOnon-p.olarDepM-endingandAon-planetheGaNselectedwasLiGaOstudiedonsubstratethewellsurfacelatticeorienmatctationhed,
22i.e.pureand(100)flatornon-p(010),olarM-orGaNAw-planeasgrocouldwn.beDislogrocatiown,nrespdensitiesectivelyw.ereVshoerywnhightobephaseon
thesameorderascomparablereportedvalues.Stackingfaultswerefoundinhigh
shodensitieswinganintheimproMvemen-planetsamplecomparedbuttowrepereortsinconsiderablyliteraturelessabinoutthestacAking-planefaultssamplein
A-planeGaNgrownonR-planesapphire.

99

TheuseofLiAlO2asasubstratefornon-polarGaNbenefitsfromthepersistent
investigationsonthissubstrateoverthepastdecade,meaningthatmanufacturing
andsurfacepolishingissueshavebeensubjecttoalongerdevelopmentandaremore
refinedthanforLiGaO2,usedforthefirsttimetogrownon-polarGaNinthecontext
ofthisthesis.LiGaO2presentsaverypromisingcandidateforhighphasepurityand
aflatandsmoothmorphologyofthegrownnon-polarGaNplanes.NoGadroplets
wereobservedontheGaNsurface;stillgoodcrystalqualitywasevident.However,
thepresenceofscratchesonthesubstratesurfacesindicatedthatpolishingissuesof
LiGaO2arenotcompletelysolvedyet.
Nevertheless,comparingnon-polarGaNgrowthonLiAlO2andLiGaO2,thelatter
seemstobeamoreconvenientsubstratefornon-polarGaNgrowth,duetothehigh
achievablephasepuritywhichcomesnaturally,duetotheselectivelatticematchof
thedifferentsurfacestoGaN.However,progressonthematterofsubstratepolishing
hastobemadefirst,beforetheinvestigationofGaNgrowthonthissubstratecan
yieldhighercrystalquality.
Aprospectivecandidateforaroomtemperaturespin-aligningmaterialinspintron-
icsapplicationsCu-dopedGaNgrownbyMBE,wasinvestigatedwithrespectto
theplacementofCuatomsinGaNbyx-rayabsorption.Simulationsusingthe
FDMNESprogramwereperformedinconjunctionwiththeanalysisoftheexper-
imentalXANESandXLDdata.Thecombinationofexperimentandcalculations
givesinsightintotheCuincorporationinGaN.Itwasfoundthatthemajority
(around70%)ofCuintheGaNfilmissituatedonGasubstitutionalsites,roughly
20%ispresentoninterstitialand10%onNsites.
TheseresultsshowthatCuisincorporatedonGasitesasassumedbytheoretical
studiesandgivesvaliditytothoseinvestigations.However,notonlyisCupositioned
onsubstitutionalcationsitesbutalsotosomepartonothersitesinthecrystal.
Therefore,theoreticalstudiesexaminingthemagneticbehaviorofGaN:Cushould
alsoconsidertheCuoccupationofinterstitialandnitrogensites,whichmightyield
insightintounravelingthemechanismofferromagnetisminthismaterial.Inthis
sense,thefindingspresentedhereprovideadditionalinformationwhichmayserve
toaidtheunderstandingofthematerialofCu-dopedGaN.
Inconclusion,forfutureworkitwouldbeveryinterestingtocombinethebene-
fitsofnon-polarGaNwiththepromisingpropertiesofCu-dopedGaNforefficient
spintronicsdevices.Non-polarInGaNquantumdotsimbeddedinnon-polarGaN,
grownon(100)or(010)LiGaO2couldserveastheactivematerialinaspinLED,
whereasahighdegreeofspinpolarizationandanefficientspininjectionatroom
temperaturecouldberealizedbyferromagneticCu-dopedGaN.

Bibliography

1.2.3.4.5.6.7.8.9.10.11.12.

13.14.

Awschalom,D.D.&Flatte,M.E.Challengesforsemiconductorspintronics.
English.NaturePhysics3,153–159.issn:1745-2473(2007).
Binasctoresistanceh,G.,inGrla¨unyberederg,P.,magneticSaurenbacstructuresh,F.&withanZinn,W.tiferromagneticEnhancedinterlamagneyer-
exchange.PhysicalReviewB39,4828–4830(1989).
Wolf,S.etal.Spintronics:Aspin-basedelectronicsvisionforthefuture.
English.Science294,1488–1495.issn:0036-8075(2001).
PIEEearton,ProceS.eetdingsal.-Spin152,tronics312–322.deviceissn:concepts.1350-2409Cir(2cuits,005).DevicesandSystems,
Morko¸c,H.HandbookofNitrideSemiconductorsandDevices(NewYork:
2008).CH,Wiley-VZorpette,G.optoelectronics/let-LetThereBethere-Light¡be-light/1¿(2002).http://spectrum.ieee.org/semiconductors/
GaN¡http://www.ostendo.com/gan/applications.php#leds¿(2011).
18,Szw40eda,–41.R.GaNissn:and0961-1290SiC(2detectors005).forradiationandmedicine.III-VsReview
PanasonicdevelopsaGalliumNitride(GaN)terahertzdetectorwithhighsen-
sitivity¡http://www.physorg.com/news196524107.html¿(2010).
ApplicationofGaN¡http://www.nitronex.com/gan_applications.html¿.
AlNShealy,J.sub-bufferR.etlaal.yer.AnJournalAlGaN/GaNofPhysics:high-electron-moCondensedbilitMattery14,transistor3499with(2002).an
Manfra,plasma-assistedM.J.etal.molecularHighbeammobilityepitaxyAlGaN/GonaNsemi-insulatingheterostructuresGaNgrowntemplatesby
preparedbyhydridevaporphaseepitaxy.JournalofAppliedPhysics92,
(2002).338–345Waltereit,P.etal.Nitridesemiconductorsfreeofelectrostaticfieldsforeffi-
cientwhitelight-emittingdiodes.Nature406,865–868(Aug.2000).
D:Jansen,ApplieR.dThePhysicsspin-v36,alveR289transistor:(2003).areviewandoutlook.JournalofPhysics

15.16.17.18.19.20.21.22.23.24.25.26.27.28.29.

101

Schmidt,G.,Ferrand,D.,Molenkamp,L.W.,Filip,A.T.&vanWees,B.J.
Fundamentalobstacleforelectricalspininjectionfromaferromagneticmetal
intoadiffusivesemiconductor.PhysicalReviewB62,R4790–R4793(2000).
Dietl,T.,Ohno,H.,Matsukura,F.,Cibert,J.&Ferrand,D.Zenermodelde-
scriptionofferromagnetisminzinc-blendemagneticsemiconductors.English.
Science287,1019–1022.issn:0036-8075(2000).
Wu,R.Q.etal.Cu-dopedGaN:Adilutemagneticsemiconductorfromfirst-
principlesstudy.AppliedPhysicsLetters89,062505(Aug.2006).
Seong,H.K.etal.Room-temperatureferromagnetisminCudopedGaN
nanomores.NanoLetters7,3366–3371(Nov.2007).
Yang,X.L.etal.Structural,optical,andmagneticpropertiesofCu-implanted
GaNfilms.JournalofAppliedPhysics105,053910(Mar.2009).
Ganz,P.R.,S¨urgers,C.,Fischer,G.&Schaadt,D.M.Cu-dopedGaN
grownbymolecularbeamepitaxy.JournalofPhysics:ConferenceSeries200,
(2010).062006Joly,Y.X-rayabsorptionnear-edgestructurecalculationsbeyondthemuffin-
tinapproximation.PhysicalReviewB63,125120(2001).
Baibich,M.N.etal.GiantMagnetoresistanceof(001)Fe/(001)CrMagnetic
Superlattices.PhysicalReviewLetters61,2472–2475(1988).
Pauli,W.WissenschaftlicherBriefwechselMitBohr,Einstein,Heisenberg
U.A./ScientificCorrespondencewithBohr,Einstein,HeisenbergA.O.(ed
Meyenn,K.V.)1930-1939,94(Springer,1985).
Liu,C.,Yun,F.&Morko¸c,H.FerromagnetismofZnOandGaN:Areview.
English.JournalofMaterialsScience-MaterialsinElectronics16,555–597.
issn5).(2000957-4522:Datta,S.&Das,B.Electronicanalogoftheelectro-opticmodulator.Applied
PhysicsLetters56,665–667(1990).
Bychkov,Y.A.&Rashba,E.I.Oscillatoryeffectsandthemagneticsuscepti-
bilityofcarriersininversionlayers.JournalofPhysicsC:SolidStatePhysics
(1984).603917,Schmidt,G.&Molenkamp,L.W.Spininjectionintosemiconductors,physics
andexperiments.SemiconductorScienceandTechnology17,310(2002).
Furdyna,J.K.Dilutedmagneticsemiconductors:Aninterfaceofsemiconduc-
torphysicsandmagnetism(invited).JournalofAppliedPhysics53,7637–
(1982).7643Ohno,Y.etal.Electricalspininjectioninaferromagneticsemiconductor
heterostructure.English.Nature402,790–792.issn:0028-0836(1999).

30.31.32.33.34.35.36.37.38.39.40.41.42.

102

Fiederling,R.etal.Injectionanddetectionofaspin-polarizedcurrentina
light-emittingdiode.English.Nature402,787–790.issn:0028-0836(1999).
Furdyna,J.K.,Samarth,N.,Frankel,R.B.&Spalek,J.Staticmagnetic
susceptibilityofZn1−xMnxSe.PhysicalReviewB37,3707–3709(1988).
Furdyna,J.K.,Liu,X.,Sasaki,Y.,Potashnik,S.J.&Schiffer,P.Ferromag-
neticIII-Mn-Vsemiconductormultilayers:Manipulationofmagneticprop-
ertiesbyproximityeffectsandinterfacedesign(invited).JournalofApplied
(2002).7490–749591,PhysicsL¨offler,W.etal.Dopingandoptimalelectronspinpolarizationinn-ZnMnSe
forquantum-dotspin-injectionlight-emittingdiodes.AppliedPhysicsLetters
(2010).05211396,L¨offler,W.etal.Parallelpreparationofhighlyspin-polarizedelectronsin
singleInAs/GaAsquantumdots.AppliedPhysicsLetters90,232105(2007).
Hetterich,M.etal.inAdvancesinSolidStatePhysics(edHaug,R.)103–114
(SpringerBerlin/Heidelberg,2009).doi:10.1007/978-3-540-85859-1_9.
¡http://dx.doi.org/10.1007/978-3-540-85859-1_9¿.
Asshoff,P.etal.Spin-polarizationdynamicsinelectricallyexcitedsingle
InGaAsquantumdots.JournalofPhysics:ConferenceSeries200,062002
(2010).Dagnelund,D.etal.Efficiencyofopticalspininjectionandspinlossfrom
adilutedmagneticsemiconductorZnMnSetoCdSenonmagneticquantum
dots.PhysicalReviewB77,035437(2008).
Ohno,H.Makingnonmagneticsemiconductorsferromagnetic.English.Sci-
ence281,951–956.issn:0036-8075(1998).
Bastard,G.,Rigaux,C.&Mycielski,A.Giantspinsplittinginducedbyex-
changeinteractionsinHg1−kMnkTemixedcrystals.PhysicaStatusSolidi(b)
79,585–593.issn:1521-3951(1977).
Komarov,A.,Ryabchenko,S.,Terletskij,O.,Zherou,I.&Ivanchouk,R.
Magneto-OpticalInvestigationsandDoubleOptico-MagneticResonanceof
ExcitionBandinCdTe-Mn2+.Russian.ZhurnalEksperimentanoiiTeoreich-
eskoiFiziki73,608–618.issn:0044-4510(1977).
Kossut,J.&Gaj,J.A.IntroductiontothePhysicsofDilutedMagneticSemi-
conductors1sted.(edsKossut,J.&Gaj,J.A.)doi:10.1007/978-3-642-
2010).(Springer,815856-Furdyna,J.K.Dilutedmagneticsemiconductors:Issuesandopportunities.
JournalofVacuumScience&TechnologyA:Vacuum,Surfaces,andFilms4,
(1986).2002–2009

43.44.45.46.47.48.49.50.51.52.53.54.55.56.

103

Furdyna,J.K.Dilutedmagneticsemiconductors.JournalofAppliedPhysics
(1988).R29–R6464,Munekata,H.etal.DilutedmagneticIII-Vsemiconductors.PhysicalReview
(1989).1849–185263,ettersLOhno,H.etal.(Ga,Mn)As:Anewdilutedmagneticsemiconductorbasedon
GaAs.AppliedPhysicsLetters69,363–365(1996).
Ohno,H.,Munekata,H.,vonMoln´ar,S.&Chang,L.L.NewIII-Vdiluted
magneticsemiconductors(invited).JournalofAppliedPhysics69,6103–6108
(1991).Chambers,S.&Farrow,R.Newpossibilitiesforferromagneticsemiconduc-
tors.English.MRSBulletin28,729–733.issn:0883-7694(2003).
Jonker,B.,Erwin,S.,Petrou,A.&Petukhov,A.Electricalspininjection
andtransportinsemiconductorspintronicdevices.English.MRSBulletin
28,740–748.issn:0883-7694(2003).
Macdonald,A.,Schiffer,P.&Samarth,N.Ferromagneticsemiconductors:
movingbeyond(Ga,Mn)As.English.NatureMaterials4,195–202.issn:
(2005).1476-1122Bonanni,A.Ferromagneticnitride-basedsemiconductorsdopedwithtransi-
tionmetalsandrareearths.SemiconductorScienceandTechnology22,R41
(2007).Avrutin,V.,Izyumskaya,N.,Oandzguandr,U.,Silversmith,D.&Morko¸c
and,H.FerromagnetisminZnO-andGaN-BasedDilutedMagneticSemicon-
ductors:AchievementsandChallenges.ProceedingsoftheIEEE98,1288–
1301.issn:0018-9219(2010).
Sato,K.etal.First-principlestheoryofdilutemagneticsemiconductors.Re-
viewsofModernPhysics82,1633–1690(2010).
Munekata,H.III-Vbaseddilutedmagneticsemiconductors:Carrier-induced
magnetism.AdvancedMaterials7,82–85.issn:1521-4095(1995).
Nov´k,J.etal.DependenceofCurietemperatureonsurfacestraininInM-
nAsepitaxialstructures.AppliedSurfaceScience256.ProgressinApplied
Surface,InterfaceandThinFilmScience-SURFINTSRENII,5672–5675.
issn).(20100169-4332:May,S.,Blattner,A.,Eam,D.&Wessels,B.Directobservationofroomtem-
peraturemagnetismin(In,Mn)Asthinfilmsbymagneticforcemicroscopy.
AppliedSurfaceScience252,3509–3513.issn:0169-4332(2006).
Telek,P.etal.Design,preparationandpropertiesofspin-LEDstructures
basedonInMnAs,175–178(2010).

57.58.59.60.61.62.63.64.65.66.67.68.69.70.71.

104

Olejn´ık,K.etal.Enhancedannealing,highCurietemperature,andlow-
voltagegatingin(Ga,Mn)As:Asurfaceoxidecontrolstudy.PhysicalReview
(2008).05440378,BJungwirth,T.etal.Prospectsforhightemperatureferromagnetismin(Ga,Mn)As
semiconductors.PhysicalReviewB72,165204(2005).
Sonoda,S.,Shimizu,S.,Sasaki,T.,Yamamoto,Y.&Hori,H.Molecularbeam
epitaxyofwurtzite(Ga,Mn)Nfilmsonsapphire(0001)showingtheferromag-
neticbehaviouratroomtemperature.JournalofCrystalGrowth237-239,
1358–1362.issn:0022-0248(2002).
Song,C.etal.Giantmagneticmomentinananomalousferromagneticinsu-
lator:Co-dopedZnO.PhysicalReviewB73,024405(2006).
Krenn,H.&Granitzer,P.MagnetischeHalbleiter:Spintronik.Physikinun-
sererZeit33,218–225.issn:1521-3943(2002).
Zener,C.InteractionbetweenthedShellsintheTransitionMetals.Physical
(1951).440–44481,eviewRZener,C.Interactionbetweenthed-ShellsintheTransitionMetals.II.Fer-
romagneticCompoundsofManganesewithPerovskiteStructure.Physical
(1951).403–40582,eviewRRuderman,M.A.&Kittel,C.IndirectExchangeCouplingofNuclearMag-
neticMomentsbyConductionElectrons.PhysicalReview96,99(1954).
Kasuya,T.ATheoryofMetallicFerro-andAntiferromagnetismonZener’s
Model.ProgressofTheoreticalPhysics16,45–57(1956).
Yosida,K.MagneticPropertiesofCu-MnAlloys.PhysicalReview106,893–
(1957).898Coey,J.,Venkatesan,M.&Fitzgerald,C.Donorimpuritybandexchange
indiluteferromagneticoxides.English.NatureMaterials4,173–179.issn:
(2005).1476-1122Litvinov,V.I.&Dugaev,V.K.FerromagnetisminMagneticallyDopedIII-V
Semiconductors.PhysicalReviewLetters86,5593–5596(2001).
Dhar,S.,Brandt,O.,Ramsteiner,M.,Sapega,V.F.&Ploog,K.H.Colossal
MagneticMomentofGdinGaN.PhysicalReviewLetters94,037205(2005).
Khare,N.,Kappers,M.,Wei,M.,Blamire,M.&MacManus-Driscoll,J.
Defect-InducedFerromagnetisminCo-dopedZnO.AdvancedMaterials18,
1449–1452.issn:1521-4095(2006).
Song,C.etal.Evidenceofstructuraldefectenhancedroom-temperaturefer-
romagnetisminCo-dopedZnO.JournalofPhysics:CondensedMatter19,
(2007).176229

72.73.74.75.76.77.78.79.80.81.82.83.84.85.86.

105

Lee,K.J.etal.MagneticpropertiesofGaMnNgrownviamolecularbeam
epitaxyusingasingleprecursor.PhysicaStatusSolidi(b)241,2854–2857.
issn4).(2001521-3951:Xu,X.H.etal.Carrier-inducedferromagnetisminn-typeZnMnAlOand
ZnCoAlOthinfilmsatroomtemperature.NewJournalofPhysics8,135
(2006).Kittilstved,K.2+R.etal.DirectKineticCorrelationofCarriersandFerromag-
netisminCo:ZnO.PhysicalReviewLetters97,037203(2006).
Behan,A.J.etal.TwoMagneticRegimesinDopedZnOCorrespondingto
aDiluteMagneticSemiconductorandaDiluteMagneticInsulator.Physical
ReviewLetters100,047206(2008).
Dev,P.,Xue,Y.&Zhang,P.Defect-InducedIntrinsicMagnetisminWide-
GapIIINitrides.PhysicalReviewLetters100,117204(2008).
Hong,J.LocalmagneticmomentinducedbyGavacancydefectinGaN.
JournalofAppliedPhysics103,063907(2008).
Dhar,S.etal.FerromagnetismandcolossalmagneticmomentinGd-focused
ion-beam-implantedGaN.AppliedPhysicsLetters89,062503(2006).
Hsu,H.S.etal.Evidenceofoxygenvacancyenhancedroom-temperature
ferromagnetisminCo-dopedZnO.AppliedPhysicsLetters88,242507(2006).
Sun,L.etal.Structural,morphological,andmagneticcharacteristicsofCu-
implantednonpolarGaNfilms.English.AppliedSurfaceScience256,1361–
1364.issn:0169-4332(2009).
Maruska,H.P.&Tietjen,J.J.Thepreparationandpropertiesofvapor-
depositedsingle-crystallineGaN.AppliedPhysicsLetters15,327–329(1969).
Jena,D.ConferencetalkattheInternationalConferenceforthePhysicsof
Semiconductors(ICPS)inSeoul,Korea2010.
Krishnamurthy,S.,vanSchilfgaarde,M.&Newman,N.Spinlifetimesofelec-
tronsinjectedintoGaAsandGaN.AppliedPhysicsLetters83,1761–1763
(2003).Brown,J.etal.FastspinrelaxationinInGaN/GaNmultiplequantumwells.
PhysicaStatusSolidi(b)243,1643–1646(2006).
Kuroda,T.etal.SubpicosecondexcitonspinrelaxationinGaN.Applied
PhysicsLetters85,3116–3118(2004).
Tackeuchi,A.etal.NospinpolarizationofcarriersinInGaN.PhysicaE:
Low-dimensionalSystemsandNanostructures7,1011–1014.issn:1386-9477
(2000).

87.88.89.90.91.92.93.94.95.96.97.98.99.100.

106

Julier,M.etal.SlowSpinRelaxationObservedinInGaN/GaNMultiple
QuantumWells.PhysicaStatusSolidi(b)216,341–345.issn:1521-3951
(1999).Buyanova,I.A.etal.OntheoriginofspinlossinGaMnN/InGaNlight-
emittingdiodes.AppliedPhysicsLetters84,2599–2601(2004).
Chen,W.M.etal.EfficientspinrelaxationinInGaN/GaNandInGaN/GaMnN
quantumwells:Anobstacletospindetection.AppliedPhysicsLetters87,
(2005).192107Kyrychenko,F.V.etal.InvestigationofaGaMnN/GaN/InGaNstructure
forspinLED.AIPConferenceProceedings772,1319–1320.issn:0094243X
(2005).Nagahara,S.,Arita,M.&Arakawa,Y.Long-livedelectronspinsinInxGa1−xN
multiquantumwell.AppliedPhysicsLetters86,242103(2005).
Nagahara,S.,Arita,M.&Arakawa,Y.Notemperaturedependenceofspin
relaxationinInGaNphase-separatedquantumdots.AppliedPhysicsLetters
(2006).08310188,Gotoh,H.,Ando,H.,Kamada,H.,Chavez-Pirson,A.&Temmyo,J.Spin
relaxationofexcitonsinzero-dimensionalInGaAsquantumdisks.Applied
PhysicsLetters72,1341–1343(1998).
Grandjean,N.,Massies,J.&Leroux,M.Self-limitationofAlGaN/GaNquan-
tumwellenergybybuilt-inpolarizationfield.AppliedPhysicsLetters74,
(1999).2361–2363Craven,M.D.,Waltereit,P.,Speck,J.S.&DenBaars,S.P.Well-widthde-
pendenceofphotoluminescenceemissionfroma-planeGaN/AlGaNmultiple
quantumwells.AppliedPhysicsLetters84,496–498(2004).
Overberg,M.E.etal.Indicationofferromagnetisminmolecular-beam-epitaxy-
derivedN-typeGaMnN.AppliedPhysicsLetters79,1312–1314(2001).
Thaler,G.T.etal.Magneticpropertiesofn-GaMnNthinfilms.Applied
PhysicsLetters80,3964–3966(2002).
Dhar,S.etal.Originofhigh-temperatureferromagnetismin(Ga,Mn)Nlayers
grownon4H-SiC(0001)byreactivemolecular-beamepitaxy.AppliedPhysics
(2003).2077–207982,ettersLSasaki,T.etal.MagneticandtransportcharacteristicsonhighCurietemper-
atureferromagnetofMn-dopedGaN.JournalofAppliedPhysics91,7911–
(2002).7913Bither,T.A.&Cloud,W.H.MagneticTetragonaldeltaPhaseintheMn-Ga
Binary.JournalofAppliedPhysics36,1501–1502(1965).

101.102.103.104.105.106.107.108.109.110.111.112.

113.

107

Takei,W.J.,Shirane,G.&Frazer,B.C.MagneticStructureofMn4N.Phys-
icalReview119,122–126(1960).
Yang,H.,Al-Brithen,H.,Trifan,E.,Ingram,D.C.&Smith,A.R.Crystalline
phaseandorientationcontrolofmanganesenitridegrownonMgO(001)by
molecularbeamepitaxy.JournalofAppliedPhysics91,1053–1059(2002).
Ando,K.Magneto-opticalstudiesofs,p-dexchangeinteractionsinGaN:Mn
withroom-temperatureferromagnetism.AppliedPhysicsLetters82,100–102
(2003).Baik,J.M.,Jang,H.W.,Kim,J.K.&Lee,J.-L.Effectofmicrostructural
changeonmagneticpropertyofMn-implantedp-typeGaN.AppliedPhysics
(2003).583–58582,ettersLTheodoropoulou,N.etal.MagneticandstructuralpropertiesofMn-implanted
GaN.AppliedPhysicsLetters78,3475–3477(2001).
Chitta,V.A.etal.+RoomtemperatureferromagnetismincubicGaNepilayers
implantedwithMnions.AppliedPhysicsLetters85,3777–3779(2004).
Reed,M.L.etal.Roomtemperatureferromagneticpropertiesof(Ga,Mn)N.
AppliedPhysicsLetters79,3473–3475(2001).
Yang,X.L.etal.Positronannihilationin(Ga,Mn)N:Astudyofvacancy-
typedefects.AppliedPhysicsLetters94,151907(2009).
Potzger,K.&Zhou,S.Non-DMSrelatedferromagnetismintransitionmetal
dopedzincoxide.PhysicaStatusSolidi(b)246,1147–1167.issn:1521-3951
(2009).Leuenberger,F.etal.X-raymagneticcirculardichroismattheGdL2,3ab-
sorptionedgesinGdNlayers:Theinfluenceoflatticeexpansion.Physical
ReviewB73,214430(2006).
Iwata,N.,Okamoto,T.&E.,T.VariationofCuiretemperaturewithhydro-
staticpressureandanomalouscompressibilityonGd.English.Journalofthe
physicalsocietyofJapan24,948–&.issn:0031-9015(1968).
Rogalev,A.,Wilhelm,F.,Jaouen,N.,Goulon,J.&Kappler,J.-P.X-raymag-
neticcirculardichroism:Historicalperspectiveandrecenthighlights.Lecture
NotesinPhysics697(edsBeaurepaire,E,Bulou,H,Scheurer,F&Kap-
pler,J.)4thMittelwihrSchoolonMagnetismandSynchrotronRadiation,
Mittelwihr,France,Oct10-15,2004,71–93.issn:0075-8450(2006).
Ney,A.etal.Gd-dopedGaNstudiedwithelementspecificity:Verysmall
polarizationofGa,paramagnetismofGdandtheformationofmagneticclus-
ters.JournalofMagnetismandMagneticMaterials322.Proceedingsofthe
JointEuropeanMagneticSymposia,1162–1166.issn:0304-8853(2010).

114.115.116.117.118.119.120.121.122.123.124.125.126.127.128.

108

Rosa,A.L.&Ahuja,R.WeakferromagnetisminCu-dopedGaN.Applied
PhysicsLetters91,232109(Dec.2007).
Lee,S.C.,Choi,J.H.,Lee,K.R.&Cha,P.R.OriginofFerromagnetismand
Long-rangeInteractionsofCuinGaN:ChemicalBondingandElectronega-
tivityApproaches.JournalofTheKoreanPhysicalSociety55,1013–1017
2009).(Sept.Lee,J.H.etal.Room-temperatureferromagnetismofCu-implantedGaN.
AppliedPhysicsLetters90,032504(Jan.2007).
Ganz,P.R.,Fischer,G.,S¨urgers,C.&Schaadt,D.M.Cu-dopednitrides:
Promisingcandidatesforanitridebasedspin-aligner.JournalofCrystal
Growth323,355–358(Oct.2010).
Madhu,C.,Sundaresan,A.&Rao,C.N.R.Room-temperatureferromag-
netisminundopedGaNandCdSsemiconductornanoparticles.PhysicalRe-
(2008).20130677,BviewSun,L.L.etal.Theimpactofimplantationdoseonthecharacteristicsof
diluted-magneticnonpolarGaN:Cufilms.MaterialsLetters63,2574–2576
2009).(Dec.Sun,L.L.etal.Strongroom-temperatureferromagnetisminCu-implanted
nonpolarGaNfilms.JournalofAppliedPhysics106,113921(Dec.2009).
Elsayed,M.etal.Cudiffusion-inducedvacancy-likedefectsinfreestanding
GaN.NewJournalofPhysics13,013029(2011).
Wurtzitecrystalstructureofwithcoordinationpolyhedra¡http://upload.
wikimedia.org/wikipedia/commons/8/8e/Wurtzite_polyhedra.png¿
(2008).Che`ze,C.Conferencetalkatthe16thEuropeanMolecularBeamEpitaxyCon-
ference(EuroMBE)inL’Alpd’Huez,France.2011.
Leszczynski,M.etal.Latticeparametersofgalliumnitride.AppliedPhysics
(1996).73–7569,ettersLRoder,C.,Einfeldt,S.,Figge,S.&Hommel,D.Temperaturedependenceof
thethermalexpansionofGaN.PhysicalReviewB72,085218(2005).
Monemar,B.FundamentalenergygapofGaNfromphotoluminescenceexci-
tationspectra.PhysicalReviewB10,676–681(1974).
Bougrov,V.,Levinshtein,M.,Rumyanstev,S.&Zbrilov,A.Propertiesof
AdvancedSemiconductorMaterials(edsLevinshtein,M.E.,Rumyantsev,S.
&Shur,M.S.)(Wiley&Sons,2001).
Harafuji,K.,Tsuchiya,T.&Kawamura,K.Moleculardynamicssimulation
forevaluatingmeltingpointofwurtzite-typeGaNcrystal.JournalofApplied
(2004).2501–251296,Physics

129.130.131.132.133.134.135.136.137.138.139.140.141.142.143.

109

Kucheyev,S.O.etal.NanoindentationofepitaxialGaNfilms.AppliedPhysics
(2000).3373–337577,ettersLPolian,A.,Grimsditch,M.&Grzegory,I.Elasticconstantsofgalliumnitride.
JournalofAppliedPhysics79,3343–3344(1996).
Monemar,B.etal.RecombinationoffreeandboundexcitonsinGaN.Physica
StatusSolidi(b)245,1723–1740.issn:1521-3951(2008).
Bernardini,F.,Fiorentini,V.&Vanderbilt,D.Spontaneouspolarizationand
piezoelectricconstantsofIII-Vnitrides.PhysicalReviewB56,R10024–
(1997).R10027Leroux,M.etal.QuantumconfinedStarkeffectduetobuilt-ininternalpo-
larizationfieldsin(Al,Ga)N/GaNquantumwells.PhysicalReviewB58,
(1998).R13371–R13374Bernardini,F.&Fiorentini,V.Macroscopicpolarizationandbandoffsetsat
nitrideheterojunctions.PhysicalReviewB57,R9427–R9430(1998).
Sun,Y.-J.GrowthandCharacterizationofM-planeGaNand(In,Ga)N/GaN
multiplequantumwellsPhDthesis(Humboldt-Universit¨atzuBerlin,2004).
Lei,T.etal.Epitaxialgrowthofzincblendeandwurtziticgalliumnitride
thinfilmson(001)silicon.AppliedPhysicsLetters59,944–946(1991).
Powell,R.C.,Lee,N.-E.,Kim,Y.-W.&Greene,J.E.Heteroepitaxialwurtzite
andzinc-blendestructureGaNgrownbyreactive-ionmolecular-beamepitaxy:
Growthkinetics,microstructure,andproperties.JournalofAppliedPhysics
(1993).189–20473,Romanov,A.E.,Baker,T.J.,Nakamura,S.,Speck,J.S.&Group,E.U.
Strain-inducedpolarizationinwurtziteIII-nitridesemipolarlayers.Journal
ofAppliedPhysics100,023522(2006).
Waltereit,P.etal.GrowthofM-planeGaN(1¯100):Awaytoevadeelectrical
polarizationinnitrides.PhysicaStatusSolidiA-AppliedResearch180,133–
2000).(July138Nakamura,S.&Fasol,G.TheBlueLaseDiode(Springer,1997).
Johnson,W.C.,Parson,J.B.&Crew,M.C.NitrogenCompoundsofGallium.
III.TheJournalofPhysicalChemistry36,2651–2654(1932).
Nakamura,S.,Mukai,T.&Senoh,M.Candela-classhigh-brightnessInGaN/Al-
GaNdouble-heterostructureblue-light-emittingdiodes.AppliedPhysicsLet-
(1994).1687–168964,tersRoussel,P.BulkGaNMarket2009.¡\url{http://www.i-micronews.com/
¿.t_cs_BulK\%20GaN_web.pdf}upload/Rapports/yole_flyer_repor

144.145.146.147.148.149.150.151.152.153.154.155.156.157.

110

Brandt,M.S.,Johnson,N.M.,Molnar,R.J.,Singh,R.&Moustakas,T.D.
Hydrogenationofp-typegalliumnitride.AppliedPhysicsLetters64,2264–
(1994).2266Zavada,J.M.,Wilson,R.G.,Abernathy,C.R.&Pearton,S.J.Hydrogena-
tionofGaN,AlN,andInN.AppliedPhysicsLetters64,2724–2726(1994).
Akasaki,I.,Amano,H.,Kito,M.&Hiramatsu,K.PhotoluminescenceofMg-
dopedp-typeGaNandelectroluminescenceofGaNp-njunctionLED.Journal
ofLuminescence48-49,666–670.issn:0022-2313(1991).
Tanaka,T.etal.p-typeconductioninMg-dopedGaNandAl0.08Ga0.92N
grownbymetalorganicvaporphaseepitaxy.AppliedPhysicsLetters65,593–
(1994).594G¨otz,W.,Johnson,N.M.,Walker,J.,Bour,D.P.&Street,R.A.Activa-
tionofacceptorsinMg-dopedGaNgrownbymetalorganicchemicalvapor
deposition.AppliedPhysicsLetters68,667–669(1996).
Kim,W.etal.Mg-dopedp-typeGaNgrownbyreactivemolecularbeam
epitaxy.AppliedPhysicsLetters69,559–561(1996).
Nakayama,H.etal.ElectricalTransportPropertiesofp-GaN.JapaneseJour-
nalofAppliedPhysics35,L282–L284(1996).
Eiting,C.,Grudowski,P.&Dupuis,R.Growthoflowresistivityp-typeGaN
bymetalorganicchemicalvapourdeposition.English.ElectronicsLetters33,
1987–1989.issn:0013-5194(1997).
Monemar,B.,Lagerstedt,O.&Gislason,H.P.PropertiesofZn-dopedVPE-
grownGaN.I.Luminescencedatainrelationtodopingconditions.Journal
ofAppliedPhysics51,625–639(1980).
Polyakov,A.Y.etal.Electricalandopticalpropertiesofmodulation-doped
p-AlGaN/GaNsuperlattices.AppliedPhysicsLetters79,4372–4374(2001).
Simon,J.,Protasenko,V.,Lian,C.,Xing,H.&Jena,D.Polarization-Induced
HoleDopinginWideBand-GapUniaxialSemiconductorHeterostructures.
(2010).60–64327,eSciencNath,D.N.,G¨ur,E.,Ringel,S.A.&Rajan,S.Molecularbeamepitaxyof
N-polarInGaN.AppliedPhysicsLetters97,071903(2010).
Sawaki,N.etal.Selectivegrowthandimpurityincorporationinsemipolar
GaNgrownonSisubstrate.GalliumNitrideMaterialsandDevicesV7602
(edsChyi,J.-I.etal.)760203(2010).
Liu,L.&Edgar,J.H.Substratesforgalliumnitrideepitaxy.MaterialsScience
&EngineeringR-Reports37,61–127(Apr.2002).

158.159.160.161.162.163.164.165.166.167.168.169.170.171.

111

Neumann,W.etal.StructureinvestigationsofnonpolarGaNlayers.Journal
ofMicroscopy-Oxford237,308–313.issn:0022-2720(2010).
Shih,C.-H.etal.Microstructureofnon-polarGaNonLiGaO2grownby
plasma-assistedMBE.acceptedinNanoscaleResearchLetters.¡\url{http://
www.nanoscalereslett.com/imedia/1001019518468975_article.pdf}¿
(2011).Craven,M.D.,Lim,S.H.,Wu,F.,Speck,J.S.&DenBaars,S.P.Struc-
turalcharacterizationofnonpolar(11¯20)a-planeGaNthinfilmsgrownon
(1bar102)r-planesapphire.AppliedPhysicsLetters81,469–471(2002).
Haskell,B.A.etal.Structuralandmorphologicalcharacteristicsofplanar
(11¯20)a-planegalliumnitridegrownbyhydridevaporphaseepitaxy.Applied
PhysicsLetters83,1554–1556(2003).
Wagner,B.etal.Growthandcharacterizationofpendeo-epitaxialGaN(11¯20)
on4H-SiC(11¯20)substrates.JournalofCrystalGrowth290,504–512.issn:
(2006).0022-0248Sun,Q.etal.Microstructuralevolutioninm-planeGaNgrowthonm-plane
SiC.AppliedPhysicsLetters92,051112(2008).
Sun,Y.etal.Impactofnucleationconditionsonthestructuralandoptical
propertiesofM-planeGaN(1¯100)grownonγ-LiAlO2.JournalofApplied
Physics92,5714–5719.issn:0021-8979(2002).
Zou,J.&Xiang,W.Nonpolarm-anda-planeGaNthinfilmsgrownon
γ-LiAlO2substrates.JournalofCrystalGrowth311,3285–3288.issn:0022-
(2009).0248Chen,C.Q.etal.GaNhomoepitaxyonfreestanding(1¯100)orientedGaN
substrates.AppliedPhysicsLetters81,3194–3196(2002).
Wernicke,T.etal.SurfacemorphologyofhomoepitaxialGaNgrownonnon-
andsemipolarGaNsubstrates.PhysicaStatusSolidi(b)248,574–577.issn:
(2011).1521-3951Ambacher,O.GrowthandapplicationsofGroupIII-nitrides.Journalof
PhysicsD:AppliedPhysics31,2653(1998).
Marezio,M.Crystalstructureandanomalousdispersionofγ-LiAlO2.Acta
Crystallographica19,396–&(1965).
Cockayne,B.&Lent,B.TheCzochralskigrowthofsinglecrystallithium
aluminate,LiAlO2.JournalofCrystalGrowth54,546–550.issn:0022-0248
(1981).Marezio,M.CrystalstructureofLiGaO2.English.ActaCrystallographica18,
481–&.issn:0108-7673(1965).

172.173.174.175.176.177.178.179.180.181.182.183.184.185.186.

112

Rawn,C.J.&Chaudhuri,J.HightemperatureX-raydiffractionstudyof
LiGaO2.JournalofCrystalGrowth225,214–220.issn:0022-0248(2001).
Matyi,R.J.,Doolittle,W.A.&Brown,A.S.Highresolutionx-raydiffraction
analysesofGaN/LiGaO2.JournalofPhysicsD-AppliedPhysics32,A61–A64
1999).y(MaIchimiya,A.&Cohen,P.I.ReflectionHigh-EnergyElectronDiffraction(Cam-
bridgeUniversityPress,2004).
Koblm¨uller,G.,Fern´andez-Garrido,S.,Calleja,E.&Speck,J.S.Insitu
investigationofgrowthmodesduringplasma-assistedmolecularbeamepitaxy
of(0001)GaN.AppliedPhysicsLetters91,161904(2007).
Waltereit,P.etal.GrowthofM-planeGaN(1¯100)onγ-LiAlO2(100).Journal
ofCrystalGrowth218,143–147.issn:0022-0248(2000).
Hellman,E.S.,Liliental-Weber,Z.&Buchanan,D.N.E.EpitaxialGrowth
andOrientationofGaNon(100)γ-LiAlO2.MRSInternetJournalofNitride
SemiconductorResearch2,article30(1997).
Sun,Y.J.,Brandt,O.&Ploog,K.H.GrowthofM-planeGaNfilmson
γ-LiAlO2(100)withhighphasepurity.JournalofVacuumScienceandTech-
nologyB21,1350(2003).
Gerlach,J.W.etal.High-qualitym-planeGaNthinfilmsdepositedonγ-
LiAlO2byion-beam-assistedmolecular-beamepitaxy.AppliedPhysicsLetters
(2006).01190288,Jun,Z.etal.Growthstudiesofm-GaNlayersonLiAlO2byMOCVD.English.
ChinesePhysics15,2706–2709.issn:1009-1963(2006).
Czochralski,J.EinneuesVerfahrenzurMessungderKristallisations-geschwindigkeit
derMetalle.Zeitschriftf¨urPhysikalischeChemie92,219–221(1918).
Mogilatenko,A.etal.MechanismofLiAlO2decompositionduringtheGaN
growthon(100)γ-LiAlO2.JournalofAppliedPhysics102,023519(2007).
Hsu,C.etal.WetchemicaletchingofLiGaO2andLiAlO2.English.Electro-
chemicalandSolidStateLetters4,C35–C38.issn:1099-0062(2001).
Hsu,C.etal.PhotoluminescencestudiesofGaNepilayer-nanocrystalsgrown
onγ-LiAlO2substrate.PhysicaE:Low-dimensionalSystemsandNanostruc-
tures40.13thInternationalConferenceonModulatedSemiconductorStruc-
tures,1971–1974.issn:1386-9477(2008).
Lo,I.etal.LinedefectsofM-planeGaNgrownonγ-LiAlO2byplasma-
assistedmolecularbeamepitaxy.AppliedPhysicsLetters92,202106(2008).
Lo,I.,Hsieh,C.-H.,Hsu,Y.-C.,Pang,W.-Y.&Chou,M.-C.Self-assembled
GaNhexagonalmicropyramidandmicrodisk.AppliedPhysicsLetters94,
(2009).062105

187.188.189.190.191.

192.

193.194.195.196.197.198.

113

Widmann,F.,Feuillet,G.,Daudin,B.&Rouvi`ere,J.L.Lowtemperature
sapphirenitridation:AcluetooptimizeGaNlayersgrownbymolecularbeam
epitaxy.JournalofAppliedPhysics85,1550–1555(1999).
Namkoong,G.etal.Theimpactofsubstratenitridationtemperatureand
bufferdesignandsynthesisonthepolarityofGaNepitaxialfilms.Journalof
CrystalGrowth252,159–166.issn:0022-0248(2003).
Telieps,W.&Bauer,E.The(7×7)<->(1×1)phasetransitiononSi(111).
SurfaceScience162,163–168.issn:0039-6028(1985).
Cheng,A.-T.,Su,Y.-K.,Lai,W.-C.&Chen,Y.-Z.MetalorganicVaporPhase
EpitaxyGrowthofm-PlaneGaNUsingLiAlO2Substrates.JapaneseJournal
ofAppliedPhysics47,3074–3076(2008).
Hang,D.,Chou,M.M.,Chang,L.,Dikme,Y.&Heuken,M.Growthandchar-
acterizationofm-planeGaN-basedlayersonLiAlO2(100)grownbyMOVPE.
JournalofCrystalGrowth311.TheProceedingsofthe4thInternational
AsianConferenceonCrystalGrowthandCrystalTechnology,452–455.issn:
(2009).0022-0248Chou,M.M.etal.GrowthbehaviorofnonpolarGaNonthenearlylattice-
matched(100)γ-LiAlO2substratebychemicalvapordeposition.Journalof
CrystalGrowth311.TheProceedingsofthe4thInternationalAsianConfer-
enceonCrystalGrowthandCrystalTechnology,448–451.issn:0022-0248
(2009).Reed,M.,Kryliouk,O.,Mastro,M.&Anderson,T.Growthandcharac-
terizationofsingle-crystallinegalliumnitrideusing(100)LiAlO2substrates.
JournalofCrystalGrowth274,14–20.issn:0022-0248(2005).
Mogilatenko,A.etal.English.inPhysicaStatusSolidi(c)-Currenttopicsin
solidstatephysics,vol5,no122008(edStutzmann,M)5(2008),3712–3715.
Richter,E.etal.Freestandingtwoinchc-planeGaNlayersgrownon(100)
γ-lithiumaluminiumoxidebyhydridevapourphaseepitaxy.physicastatus
solidi(c)3,1439–1443.issn:1610-1642(2006).
Richter,E.etal.OptimizationofHVPEgrowthoffreestandingc-planeGaN
layersusing(100)γ-LiAlO2substrates.physicastatussolidi(c)4,2277–2280.
issn7).(2001610-1642:Losurdo,M.etal.ThesurfacemodificationandreactivityofLiGaO2sub-
stratesduringGaNepitaxy.JournalofCrystalGrowth264,139–149(Mar.
2004).Balakrishnan,K,Okumura,H&Yoshida,S.Studyontheinitialstagesofhet-
eroepitaxialgrowthofhexagonalGaNonsapphirebyplasmaassistedMBE.
English.JournalofCrystalGrowth190,244–249.issn:0022-0248(1998).

199.200.201.202.203.

204.205.206.207.208.209.210.

114

Ishii,T.,Tazoh,Y.&Miyazawa,S.Single-crystalgrowthofLiGaO2fora
substrateofGaNthinfilms.JournalofCrystalGrowth186,409–419(Mar.
1998).Johnson,M.A.L.etal.GrowthofGaN,InGaN,andAlGaNfilmsandquan-
tumwellstructuresbymolecularbeamepitaxy.JournalofCrystalGrowth
175-176,72–78.issn:0022-0248(1997).
Kung,P.etal.Metalorganicchemicalvapordepositionofmonocrystalline
GaNthinfilmsonβ-LiGaO2substrates.AppliedPhysicsLetters69,2116–
(1996).2118Kryliouk,O.etal.MOCVDgrowthofGaNfilmsonlattice-matchedoxidesub-
strates.MaterialsResearchSocietySymposiumProceedings449(edsPonce,
F.,Moustakas,T.,Akasaki,I&Monemar,B.)SymposiumonIII-VNitrides,
Boston,Dec02-06,1996,123–128.issn:0272-9172(1997).
Nicholls,J.etal.GrowthandopticalpropertiesofGaNgrownbyMBEon
novellattice-matchedoxidesubstrates.MaterialsResearchSocietySymposium
Proceedings395(edsPonce,F.,Dupuis,R.,Nakamura,S&Edmond,J.)1st
InternationalSymposiumonGalliumNitrideandRelatedMaterials,Boston,
Nov27-Dec01,1995,535–539.issn:0272-9172(1996).
Tazoh,Y.,Ishii,T.&Miyazawa,S.GaNThinFilmGrowthonLiGaO2Sub-
stratewithaMulti-DomainStructure.JapaneseJournalofAppliedPhysics
(1997).L746–L74936,Andrianov,A.etal.PhotoluminescencefromGaNfilmsgrownbyMBEon
LiGaO2substrate.English.SemiconductorScienceandTechnology12,59–63.
issn).(19970268-1242:Doolittle,W.A.etal.MBEgrowthofhighqualityGaNonLiGaO2.Journal
ofElectronicMaterials27,L58–L60(Aug.1998).
Okada,M.etal.ECR-MBEgrowthofGaNonLiGaO2.JournalofCrystal
Growth189-190,213–217.issn:0022-0248(1998).
Seo,S.W.etal.GaNmetal-semiconductor-metalphotodetectorsgrownon
lithiumgallatesubstratesbymolecular-beamepitaxy.AppliedPhysicsLetters
(2001).1372–137479,Duan,S.K.,Teng,X.G.,Han,P.D.&Lu,D.C.Growthandcharacterization
ofGaNonLiGaO2.JournalofCrystalGrowth195,304–308(Dec.1998).
Pohl,U.,Knorr,K.&Blsing,J.MetalorganicVaporPhaseEpitaxyofGaN
onLiGaO2SubstratesUsingTertiarybutylhydrazine.PhysicaStatusSolidi
(a)184,117–120.issn:1521-396X(2001).

211.212.213.214.215.216.217.218.219.220.221.222.223.

115

Kang,S.etal.CharacterizationofAlGaN/GaNstructuresonvarioussub-
stratesgrownbyradiofrequency-plasmaassistedmolecularbeamepitaxy.
English.JournalofElectronicMaterials30,156–161.issn:0361-5235(2001).
Takahashi,H.,Fujioka,H.,Ohta,J.,Oshima,M.&Kimura,M.Structural
propertiesofGaNgrownonLiGaO2byPLD.JournalofCrystalGrowth259,
36–39.issn:0022-0248(2003).
Sakurada,K.,Kobayashi,A.,Kawaguchi,Y.,Ohta,J.&Fujioka,H.Low
temperatureepitaxialgrowthofGaNfilmsonLiGaO2substrates.Applied
PhysicsLetters90,211913(2007).
Kisailus,D.&Lange,F.F.GrowthofepitaxialGaNonLiGaO2substrates
viaareactionwithammonia.JournalofMaterialsResearch16,2077–2081
2001).(JulyNanamatsu,S.,Doi,K.&Takahashi,M.Piezoelectric,ElasticandDielec-
tricPropertiesofLiGaO2.JapaneseJournalofAppliedPhysics11,816–822
(1972).Neumann,H.,Pirl,E.&Kuhn,G.Thermal-ExpansionofLiGaO2.English.
JournalofMateralsScienceLetters6,495–496.issn:0261-8028(1987).
Ishii,T.,Tazoh,Y.&Miyazawa,S.LiGaO2singlecrystalasalattice-matched
substrateforhexagonalGaNthinfilms.JournalofCrystalGrowth189-190,
208–212.issn:0022-0248(1998).
Doolittle,W.A.etal.GrowthofGaNonlithiumgallatesubstratesforde-
velopmentofaGaNthincompliantsubstrate.Papersfromthe16thnorth
americanconferenceonmolecular-beamepitaxy16,1300–1304(1998).
Chou,M.privatecommunication2010.
Armitage,R.,Horita,M.,Suda,J.&Kimoto,T.m-planeGaNlayersgrownby
rf-plasmaassistedmolecularbeamepitaxywithvaryingGa/Nfluxratioson
m-plane4H-SiCsubstrates.JournalofAppliedPhysics101,033534(2007).
Schaadt,D.,Brandt,O.,Trampert,A.,Sch¨onherr,H.-P.&Ploog,K.Epitaxial
growthof6H-AlNonM-planeSiCbyplasma-assistedmolecularbeamepitaxy.
JournalofCrystalGrowth300.FirstInternationalSymposiumonGrowthof
Nitrides-ISGN-1,127–129.issn:0022-0248(2007).
Shih,C.-H.,Lo,I.,Pang,W.-Y.,Wang,Y.-C.&Chou,M.M.Characterization
ofM-planeGaNfilmgrownonβ-LiGaO2(100)byplasma-assistedmolecular
beamepitaxy.ThinSolidFilms519,3569–3572.issn:0040-6090(2011).
Liliental-Weber,Z.,Jasinski,J.&Zakharov,D.GaNgrowninpolarand
non-polardirections.Opto-Electron.Rev.12,339–346(2004).

224.225.226.227.228.229.230.231.232.233.234.235.236.237.

116

Rajpalke,M.K.etal.Structuralandopticalpropertiesofnonpolar(11¯20)
a-planeGaNgrownon(1¯102)r-planesapphiresubstratebyplasma-assisted
molecularbeamepitaxy.ScriptaMaterialiaInPress,CorrectedProof,–.
issn).(20111359-6462:Ran,F.-Y.,Subramanian,M.,Tanemura,M.,Hayashi,Y.&Hihara,T.Ferro-
magnetisminCu-dopedAlNfilms.AppliedPhysicsLetters95,112111(2009).
Kang,B.S.,Heo,C.M.,Lyu,K.K.&Yu,S.C.First-principlesStudyon
MagnetismofCuinGaN.JournalofMagnetics14,114–116(Sept.2009).
Newville,M.FundamentalsofXAFStech.rep.(ConsortiumforAdvanced
RadiationSources,UniversityofChicago,Chicago,IL,2004).
Wilhelm,F.etal.HardX-rayLinearDichroismUsingaQuarterWavePlate
forStructuralCharacterizationofDilutedMagneticSemiconductors.AIP
ConferenceProceedings879,1675–1678(2007).
Sarigiannidou,E.etal.Intrinsicferromagnetisminwurtzite(Ga,Mn)Nsemi-
conductor.PhysicalReviewB74,041306(2006).
Stokhuyzen,R.,Brandon,J.K.,Chieh,P.C.&Pearson,W.B.Copper-
gallium,γCu9Ga4.ActaCrystallographicaSectionB30,2910–2911(1974).
Chang,L.,Huang,T.H.,Ganz,P.R.&M.,S.D.Formationmechanismof
islandsonCu-alloyedGaNgrownbyplasmaassistedmolecularbeamepitaxy
submittedtoJournaloftheElectrochemicalSociety.2011.
St¨ohr,J.&Siegmann,H.Magnetism:FromFundamentalstoNanoscaleDy-
2006).Berlin,(SpringernamicsTheLandholt-B¨ornsteinDatabase¡http://www.springermaterials.com/
¿.docs/pdf/10086090_1074.htmlLi,J.-B.etal.Athermodynamicassessmentofthecopper-galliumsystem.
Calphad32,447–453.issn:0364-5916(2008).
Hersh,H.N.TheVaporPressureofCopper.JournaloftheAmericanChem-
icalSociety75,1529–1531(1953).
Hersh,H.N.TheVaporPressureofCopper-Correction.Journalofthe
AmericanChemicalSociety75,6361–6361(1953).
Cochran,C.N.&Foster,L.M.VaporPressureofGallium,StabilityofGal-
liumSuboxideVapor,andEquilibriaofSomeReactionsProducingGallium
SuboxideVapor.JournalofTheElectrochemicalSociety109,144–148(1962).

tswledgemenknoAc

117

Iwishtoexpressmythanksandappreciationtothosewhohelpedmethrough-
outtheyearsincarryingouttheresearchworkinpreparingthisthesis.Iamvery
to:eciallyespgrateful

atProf.theDr.InstituteHeinzforKalt,mAppliedydoPhctoralysicssupattheervisor,whoKarlsruheacceptedInstitutemeofasTaecPh.D.hnology.studenHist
carefulcoachingthroughoutthisperiodwashighlybeneficialtome.
Prof.Dr.ThomasSchimmelforthewillingnesstoevaluatethisthesisasco-referee.
Dr.andforDanielhisSccomphaadt,etentforguidancegivingtmethehroughoutopptheortunitywholetobedurationpartofofhismyresearcPh.D.hthesis.team
Heenabledmyparticipationonseveralnationalandinternationaltopicalconferences
aswellasvisitstoourfellowscientificcollaboratorsinTaiwan.Hewasconsiderably
helpfulinprovidingmewithessentialinformationcontributingtothesuccessful
completionofmywork.
upHeinricandhmainReimertainingforhistheMBEexceptionalmachine.technicaldedicationandassistanceinbuilding
PhilippGanzforfruitfuldiscussionsonthegrowthofnitridesandforthegreat
teamworkontheXANESandXLDmeasurementsofCu-dopedGaN.
Dr.nitrides,Yen-LiainngpreparingChenforthemanyTEMdiscussionssamplesonbytheFIB,grohiwthsofenormousnitrides,helphelpingincoungrowingtless
issuesduringmyvisitstoTaiwan.
Yu-ChiHsuforagrandcollaborationonthegrowthofnon-polarGaNonLAOand
creatingapleasantatmosphereformystayinTaiwan.
MariusB¨urkleforthegreatassistanceandmanydiscussionsonthex-rayabsorption
calculationsofCu-dopedGaN.
MatthieuHelfrichandDr.DongzhiHufornumerousdiscussionsaswellasforthe
terrificteamworkwithinourgroup.
AllformerandpresentmembersofDr.Schaadt’sMBEgroup.Especially,Iwould
liketomentionMareikeTrunkandRaimundV¨ohringer.

118

Prof.Dr.IkaiLoforfacilitatingmystayinhisgroupandforlettingmeusethe
laboratoryinTaiwantomybenefit.

menProf.ts.Dr.PaulLiuwenVinczeChangforforAFMverymeasuremenhelpfultsadviceofAand-planediscussionsGaNgroonwntheonTEM(010)expLGO.eri-

Teng-HsingHuang,Dr.Yen-LiangChenandDr.Cheng-HungShihfortheaccurate
preparationofTEMsamplesandfortakingTEMimagesoftheGaNsamplesgrown
LGO.on

JacquesHaweckerforpatientlycarryingoutcountlessSEMinvestigationsonnon-
samples.GaNolarp

labTheoratoriesmembersandoftheProf.Dr.friendlyH.wayKalt’sofgroupmeetingforanytheirconcerns.considerablehelpfulnessinthe

AllPh.D.studentsandmembersoftheDepartmentofPhysicsandtheDepartment
ofUniversitMaterialysinTScienceaiwanandfortheirwOpto-electroniconderfulEngineerinhospitalitgyofinthemakingNationalmystaSunyYthereat-Sena
one.tpleasan

TheKarlsruheHouseofYoungScientists(KHYS)forfinancingthescientificex-
changewiththePhysicsDepartmentandtheDepartmentofMaterialsScienceand
Opto-electronicEngineeringoftheSunYat-SenUniversityinKaohsiung,Taiwan.

ToDr.FabriceWilhelmandDr.AndreRogalevfortheirkindsupportattheEuro-
peanSynchrotronRadiationFacility(ESRF)inGrenoble.

loMyvinglyfamilyforthroughouttakingallwsteighagestoffofmmyylifeshouldersleadingtowhenevtheercomneededpletionandofsuppthisortingthesis.me