Electronic structure of Ga$_{1-x}$Cr$_{x}$N and Si-doping effects studied by photoemission

X-rayabsorptionspectroscopy

G.S.Song,1M.Kobayashi,1J.I.Hwang,1T.Kataoka,1M.Takizawa,1A.Fujimori,1,2T.Ohkouchi,2Y.Takeda,2T.Okane,2Y.Saitoh,2H.Yamagami,2,3F.-H.Chang,4L.Lee,4H.-J.Lin,4D.J.Huang,4C.T.Chen,4S.Kimura,5M.Funakoshi,5S.Hasegawa,5andH.Asahi5

arXiv:0805.3299v1 [cond-mat.mtrl-sci] 21 May 2008DepartmentofPhysicsandDepartmentofComplexityScienceandEngineering,

UniversityofTokyo,7-3-1Hongo,Bunkyo-ku,Tokyo,113-0033,Japan

2

SynchrotronRadiationResearchUnit,JapanAtomicEnergyAgency,Sayo-gun,Hyogo679-5148,Japan

3

DepartmentofPhysics,FacultyofScience,KyotoSangyoUniversity,Kyoto603-8555,Japan

4

NationalSynchrotronRadiationResearchCenter,Hsinchu30076,Taiwan

5

TheInstituteofScientificandIndustrialResearch,

OsakaUniversity,8-1Mihogaoka,Ibaraki,Osaka567-0047,Japan

(Dated:May21,2008)

TheelectronicstructureofthemagneticsemiconductorGa1−xCrxNandtheeffectofSidopingonithavebeeninvestigatedbyphotoemissionandsoftx-rayabsorptionspectroscopy.WehaveconfirmedthatCrinGaNispredominantlytrivalentsubstitutingforGa,andthatCr3dstatesappearwithinthebandgapofGaNjustabovetheN2p-derivedvalence-bandmaximum.AsaresultofSidoping,downwardshiftsofthecorelevels(exceptforCr2p)andtheformationofnewstatesneartheFermilevelwereobserved,whichweattributetotheupwardchemicalpotentialshiftandtheformationofasmallamountofCr2+speciescausedbytheelectrondoping.PossibilityofCr-richclustergrowthbySidopingarediscussedbasedonthespectroscopicandmagnetizationdata.

PACSnumbers:75.50.Pp,75.30.Hx,78.70.Dm,79.60.-i

1

Dilutedmagneticsemiconductors(DMSs)withCurietemperatures(TC’s)significantlyaboveroomtempera-tureareneededforthedevelopmentofspintronicsde-vicesforpracticalapplications.Ga1−xCrxNhasbeenpredictedtheoreticallytobeferromagneticviadoubleex-change(DE)mechanism[1]andindeedferromagnetism(FM)aboveroomtemperaturehasbeenobserved[2,3].X-rayabsorptionandphotoemissionstudieshavere-vealedstronghybridizationbetweentheCr3dstatesandthevalencepband[4],andtheformationofCr3d-derivedstateswithinthebandgapofGaN[5,6],whichsug-gestedahighpotentialasanintrinsicDMS.However,highresistivity[7]andtheabsenceofdensityofstatesatFermilevel(EF)[5,6]havethrowndoubtontheDEmechanismoftheFMinGa1−xCrxN.Ontheotherhand,thesubstitutionaltransition-metal(TM)-richclus-tersofnanometersizesembeddedinthehostsemiconduc-torhavebeenobservedinZn1−xCrxTe[8]andsuggestedasanotheroriginofferromagnetisminDMSs[9].SuchTM-richclustershavealsobeenobservedinotherIII-V-basedwide-gapDMSs,Ga1−xMnxN[10],In1−xCrxN[11],andGa1−xFexN[12].

InordertoclarifythemechanismoftheFMinGa1−xCrxN,whethertheDEmechanismortheCr-richclusterformation,theinvestigationofthecarrier-concentrationdependenceoftheelectronicstructureto-getherwiththemagneticpropertiesisexpectedtoyieldessentialinformationasinthecaseofN-andI-dopedZn1−xCrxTe[13,14].Recently,Funakoshietal.[15]reportedasuccessfulcontrolofcarrierconcentrationinGa1−xCrxNbydopingSi,whichisexpectedtoreplaceGaandhenceactsasadonor.Inthiswork,wereporton

photoemission(PES)andx-rayabsorptionspectroscopy(XAS)studiesofSi-dopingeffectontheelectronicstruc-tureofGa1−xCrxNsampleswhosemagneticpropertieshavebeencharacterized.TheresultsindicatethatSidopingcausesanupwardchemicalpotentialshiftandtheformationofnewstatesnearEF(presumablyderivedfromCr2+).WesuggestthatCr-richclusterformationmaybeenhancedbySidoping,andmaystabilizeFMorsuperparamagnetism(SPM)inGa1−xCrxN.PossibleoriginsoftheobservedFMorSPMshallbediscussedbelow.

Ga1−xCrxNthinfilmsweregrownbyradio-frequency(rf)plasma-assistedmolecularbeamepitaxyon2µmGaN-templatesubstrateswhichhadbeengrownbymetal-organicchemicalvapordeposition.ElementalGa,Cr,Si,andrfplasma-enhancedN2wereusedassources.Afterthermallycleaningthesubstratesurfaceat700◦Cfor15min,ahigh-temperatureGaNbufferlayerwasgrownat700◦C.Finally,aGa0.98Cr0.02N(GaCrN)orSi-dopedGa0.98Cr0.02N(GaCrN:Si)layerwasgrownatarelativelylowtemperatureof700◦CinordertoincreasetheCrsolubility.TheGaflux,N2flowrate,Crcelltem-perature,andSicelltemperatureweresetat1.6×10−7Torr,1.5SCCM,980◦C,and1100◦C,respectively.Thenominalcarrierconcentrationswere1×1018and2×1019/cm3forGaCrNandGaCrN:Si,respectively,andtheSiconcentrationshouldbesomewhatlargerthanthem.ThethicknessoftheCr-dopedlayerwasabout19nm.Forcomparison,Si-dopedGaNwithoutCrdoping(GaN:Si)wasalsogrownbythesameprocedure.ToavoidtheoxidationofsamplesurfacesandtoperformthePESandXASmeasurementswithoutsurfacecleaning,2nm-

0.2 5 K(c) 10 K0.4GaCrN0 50 K(a))r cC CW/ 300 KBµ(GaCrN0.2)T Curie-Weiss fit ·nro-0.2Ci/taBzµ0i(t e(d)n0.2 5 KWgC0.40.8 < H < 2 TGaCrN:Sia M 300 Kc0(b) H ⊥ [0001]GaCrN:Si0.2 c CW-0.2H ⊥ [0001] Curie-Weiss fit-0.4-0.200.20.40.00100200300Magnetic Field (T)Temperature (K)FIG.1:(ColorOnline)MagnetizationofGaCrNandGaCrN:Si.(a),(b)M-Hcurvesatvarioustemperatures.(c),(d)Curie-Weiss(CW)componentofthehigh-fieldmagneticsusceptibilityasafunctionoftemperature.

thickcappinglayers(GaNforGaCrNandGaN:SiforGaCrN:Si,respectively)weredepositedonthesamplesurfaces.MagnetizationwasmeasuredusingaSQUIDmagnetometer(MPMSXL,QuantumDesign,Co.,Ltd.).X-rayphotoemissionspectroscopy(XPS)measure-mentswereperformedusingaMg-KαsourceandaGammadataScientaSES-100hemisphericalanalyzer.Resonantphotoemissionspectroscopy(RPES)measure-mentswereperformedatBL-23SUofSPring-8usingsyn-chrotronradiation.PhotoelectronswerecollectedusingaGammadataScientaSES-2002hemisphericalanalyzer.ThephotoemissionspectrawerereferencedtotheAu4fpeakandtheEFofgoldwhichwasinelectricalcon-tactwiththesamplesforXPSandRPES,respectively.XASmeasurementswereperformedattheDragonBeam-lineBL-11AofNationalSynchrotronRadiationResearchCenterinthetotal-electron-yieldmode.Allthespectraweretakenatroomtemperature.Thetotalresolutionofthespectrometerincludingthetemperaturebroadeningwas∼800meVand∼200meVforXPSandRPES,re-spectively.ThemonochromatorresolutionforXASwasE/∆E>10,000.Thebasepressureofspectrometerwasbelow3×10−8Pa.

ThemagnetizationdataofthetwoCr-dopedsamplesareshowninFig.1.Panels(a)and(b)showsM-Hcurvesatvarioustemperatures.Here,thelinearcom-ponentshavebeensubtracted.SuchaM-HcurveisknownasancharacteristicfeatureofSPM.OnecanseethattheTCexceeds400Kforbothsamples.Thesatura-tionmagnetizationoftheGaCrNslightlydecreaseswithincreasingtemperature,whilethatoftheSi-dopedonehardlydecreaseswithtemperature,asshowninpanels(a)and(b),respectively.Since,ingeneral,saturationmagnetizationisconstantontemperaturefarbelowTC,theobservationsuggeststhatSidopinginGa1−xCrxNcausesanincreaseofTC.Figure1(c)and(d)showsthetemperaturedependenceofhigh-field(0.82

magneticsusceptibilityandCurie-Weiss(CW)fit.ThefittingwasmadeassumingaCWtermplusatempera-tureindependentconstant∂M/∂H=NC/(T−Θ)+χ0.Here,Nisnumberofthemagneticions,gisthegfactor,C=(gµB)2S(S+1)/3kBistheCurieconstant(as-sumingS=3/2andg=2),andθistheWeisstemper-ature.Afterthefitting,theχ0termhasbeenremoved.FractionsoftheCratomsshowingaCW-paramagneticbehaviorare∼26%forGaCrNand∼16%forGaCrN:Si.ThedecreaseoftheCWcomponentwithSidopingin-dicatesthatthenumberofisolated(paramagnetic)Cratomsdecreases,probablybecausetheywereabsorbedbytheCr-richclusters.

Figure2(a)-(d)showsthecore-levelXPSspectraofGaCrN,GaCrN:Si,andGaN:Si.Inpanel(a),thetwopeaksineachspectrumareduetotheCr2p3/2-2p1/2spin-orbitdoublet.TheCr2pspectrumofCr2O3(Cr3+)[17]isalsoshownforcomparison.ThepeakpositionforGaCrNisclosetothatforCr2O3,indicatingthattheva-lenceofCrinGaCrNiscloseto3+,thatis,theCratomisdopedintoGaNasaneutralimpurityifCrsubstitutesforGa,consistentwiththeearlierreports[6,7,18].Thecore-levelpeaksshowninpanels(b)-(d)areshifteddown-wardsinGaCrN:SiandGaN:SicomparedwithGaCrN,indicatingthattheSidopingcausesanupwardshiftofthechemicalpotentialµ.Furthermore,onenoticesthatthecore-levelpeaksofGaCrN:SiareshiftedupwardscomparedwithGaN:Si,suggestingthatCractsasanac-ceptorandcompensatestheelectroncarriers,consistentwiththepreviousreportsthatCrdopingmakesemicon-

)sCr 2p XPS2ptGa 2pi3/2(b)(c)N 1s(d)Ga 3d3/2(a)nu2p.1/2bra( GaCrNytGaCrNisn)GaCrN:SisettinnIGaN:Siu.GaCrN:Sibra(112411201116 ytBinding Energy (eV)400396222018i)sCr2O3(Cr3+)nVeet(n Iy0.4(e)g Cr 2p3/2reGaCrN and n0.2 Ga2p3/2EGaCrN:Si ka0.0 N 1se Ga 3dP e-0.2vitale-0.4595590585580575570RGaCrNGaCrN:SiGaN:SiBinding Energy (eV)n-type carrier concentrationFIG.2:(ColorOnline)Core-levelXPSspectraofGaCrN,GaCrN:SiandSi-dopedGaN(GaN:Si).(a)Cr2p,(b)Ga3d,(c)N1s,and(d)Ga2p3/2corelevels.ElectronmeanfreepathsinGaPfortheCr2p,Ga3d,N1s,andGa2p3/2corelevelsare∼2.0,∼3.1,∼2.3,and∼0.7nm,respectively[16].Verticallinesrepresentthepeakpositionofeachcorelevel.In(a),theCr2pcore-levelXPSspectrumofCr2O3[17]isalsoshownforcomparison.ThespectraofGaCrNandGaCrN:Siareoverlaidatthebottompanel.(e)Shiftsofthecore-levelpeaksofGaCrN:SiandGaN:SirelativetothoseofGaCrN.Then-typecarrierconcentrationincreasesinorderofGaCrN,GaCrN:Si,GaN:Si.

ductingGaNsamplesinsulating[7,19].TheamountofthepeakshiftofeachcorelevelrelativetothatinGaCrNisplottedinpanel(e).TheeffectofelectrontrappingbyCrappearsintheCr2pspectrumofGaCrN:Siasshownatthebottomofpanel(a):onecanseetheformationofasmallhumpfortheSi-dopedsampleandtheoppositepeakshifttotheothercorelevels.Ingeneral,thebindingenergy(EB)ofacore-levelspectrumdecreaseswithde-creasingthevalenceoftheelementduetotheincreasedscreeningeffectbyvalenceelectrons.Wethereforecon-siderthatelectronssuppliedbytheSidopingaretrappedatCrsiteandconvertthoseatomsfromCr3+toCr2+.TheCr2pXASspectraofGaCrNandGaCrN:SiareshowninFig.3(a).Forcomparison,thoseofCr2O3andZn1−xCrxTe[20]arealsoshown.Thepeakpositionsandthelineshape3+ofthespectrumofGaCrNagreewiththoseofCr2O3(Cr),andthusonecanconcludethatthevalenceofCrinGaCrNandGaCrN:Siispredominantly3+.ComparingthespectraofGaCrNandGaCrN:Si,onecanseeaslightdifferenceat∼577eVand585eV.ThesmalldifferenceimpliesthatonlyasmallamountofCr2+speciesexist.Thedifferencecorrespondswelltotheover-allspectralweightdistributionofZn1−xCrxTe(Cr2+).Figure3(b)and(c)showsthevalence-bandRPESspec-traofGaCrNandGaCrN:SirecordedintheCr2p-3dexcitationregion.Thedifferencespectraobtainedbysubtractingtheoff-resonancespectrum(A)fromtheon-resonancespectra(BorC)shownatthebottomofpanels(b)and(c)representtheCr3dpartialdensityofstates(PDOS).OnecanclearlyseethatCr3d-derivedemissionisseenmostlywithinthebandgapofGaNasreportedinRefs.[5,6].AlsotheexistenceofCr-relatedstatesat∼reported340meVinaboveRef.the[19]valence-bandfromaphotoluminescencemaximum(VBM)study.wasThepeakpositionoftheCr3dPDOSofGaCrNrela-tivetotheVBMwellagreeswiththisvalue.Here,wehaveestimatedthepositionoftheVBMbytheintersec-tionbetweenthezerolevelandalinearextrapolationofthelowEBedgeofthevalence-bandspectrum[21].ThedistancebetweentheEFandVBMwasestimatedtobe∼panels2.9eV(b)forandGaCrN(c).

and∼3.1eVforGaCrN:SiasshownFigure3(d)and(e)showscomparisonoftheon-resonancespectra(C)ofthethreesamplesandcompari-sonoftheCr3dPDOS(C-A)ofGaCrNandGaCrN:Si.Inpanel(e),oneclearlyfindsthatnewstatesareformedaroundEBverticalarrow.∼1ConsideringeVbytheSithedoping,resultsasofXPSindicatedandXAS,byaweattributethisnewfeaturetoaCr2+origin.ThenewstatesarecreatedasdeepasEBsmallchemicalpotentialshiftof∼∼1200eVmeV,inspitequiteofdif-theferentfromtheconventionalrigid-bandbehavior.WhiletheentirevalencebandisshiftedtowardshigherEBbySidoping,theCr3dPDOSstaysatthesameEB,asindicatedbyaverticaldashedlineinpanel(d).Thefor-mationofthenewstatesandtheabsenceofshiftsintheCr3dPDOSmaybeexplainedbythestrongCoulombrepulsionbetweentheCr3delectrons[22].Wenotethat

3

noCr3dPDOSwasobservedatEF.InFig.3(f),weshowaschematicdiagramoftheelectronicstructureofGaCrN,GaCrN:Si,andGaN:Siwherethechemicalpo-tentialshift,theupwardshiftofCr3dPDOS,andtheformationofthenewstatesarepresented.

Intheprecedingparagraphs,wehaveconsideredchangesintheelectronicstructurebySidopingasaresultofCr2+formation.Inordertodiscusstherela-tionshipbetweentheobservedchangesintheelectronicstructureandthemagnetismofGa1−xCrxNcausedbySidoping,weconsidertheformationofCr-richclustersasfollows.IfCrionsarerandomlydistributedinGaN,mag-neticinteractionbetweenCrionswouldbeweakbecausetheinteractionwouldbeshort-ranged[23].Consideringtheresultsofthemagnetizationmeasurements,whichim-plytheexistenceofCr-richclustersanditsgrowthbySidoping,andtheresultsofthePESandXASmeasure-

)stin(a)BC XASOn resonance (C)(d)uGaCrN:SiCr 2p .bArGaCrNa(GaCrN y)tsiCr3+ts2O3(Cr)innGaCrN:Sieu.tZnn0.85Cr0.15TebIrGaN:Sia(570575 ytPhoton Energy (eV)580585590595isnet Cr 2p-3d RPES nI)st(b)GaCrNCr 3d PDOS (C-A)(e)iC GaCrNn2.9 eV GaCrN:Siu .bABra( ytiVBM121086420snetnIBCAABinding Energy (eV) GaCrNEg ~3.4 eV(f))(c)GaCrN:Si20CEsGa 4stCr3+ inAB3.1 eVBN 2pu .GaCrN:Sibra( ytVBMEiBN 2pCr3+ Ga 4ssnetCr2+ nIBCAAGaN:SiEBN 2p12100Ga 4s Binding Energy (eV)8642µFIG.3:(ColorOnline)Cr2pXASspectra(a)andCr2p-3dRPESspectraofGaCrN(b)andGaCrN:Si(c)takenwithphotonenergiesA,BandC(hν:A∼571.5eV,B∼576eV,C∼and577.5C-AeV)areasshownshownatinthepanelbottom(a).Theofpaneldifference(b)andspectra(c).B-ATheVBMsestimatedfromtheoff-resonancespectraareindicatedbyverticaldashedlines.(d)Comparisonoftheon-resonancespectraofthedifferentsamplestakenwithphotonenergyC.ThespectrahavebeennormalizedatthepeakintensityoftheN2p-derivedvalencebandatEBdependenceoftheCr3dPDOS.Arrow∼4.5-5indicateseV.(e)aSi-dopingnewCr3dPDOSfeatureinducedbytheSidoping.(f)SchematicdiagramoftheelectronicstructuresofGaCrN,GaCrN:Si,andGaN:Si.Theshiftsofchemicalpotentialµ,theupwardshiftofCr3dPDOSpeak,andtheformationofthenewstatesofCr2+originareshown.Here,thebandgap∼3.4eVofhexagonalGaNisalsoindicated.

4

ments,whichsuggesttheincreaseofCr2+stateswithSidopingandtheabsenceofCr3dPDOSatEF,onecanimaginethatthecoexistenceofCr3+andCr2+ionswithintheCr-richclustermayinduceshort-rangeDEwithintheCr-richclustersandstabilizeferromagnetism.However,ifthereisferromagneticspinalignmentwithintheCr-richclustersviaDE,thesaturationmagnetiza-tion(MS)wouldincreasewithincreasingclustersizesig-nificantly.Inthepresentstudy,wecouldnotobservesuchanincreaseofMSandMSremainedsmall(∼0.25µB/Cr)asshowninFig.1(a)and(b).Onepossiblerea-sonfortheabsenceofshort-rangeDEistheinsufficientamountofCr2+.Actually,theconcentrationofSiandCris∼1019cm−3and∼1021cm−3,respectively,andthereforethereareonlyfew%ofCr3+isconvertedtoCr2+.

AnotherscenariotoexplainthepossibleincreaseofTCbySidopingisthegrowthofantiferromagnetic(AFM)Cr-richclusters.SincethespinmomentoftheCr3+ionis3µB,MS∼0.25µB/CrmeansthattheamountofferromagneticCratomsinGaCrNandGaCrN:Siisonly∼8%.TheamountofparamagneticCratomsesti-matedfromtheCWfitis∼26%and∼16%forGaCrNandGaCrN:Si,respectively.Thisindicatesthat∼66%and∼76%forGaCrNandGaCrN:SitheCratomscouldnotbedetectedbythepresentmagnetizationmeasure-mentsandareprobablyantiferromagneticallycoupled.Recently,Cuietal.[24]havecalculatedthemagneticmo-mentandthestabilityofsubstitutionalCr-richclusterswithseveralsizesinGa1−xCrxNbasedonfirst-principlesdensityfunctionaltheory,andhaveshownthatlargeCr-richclustersareenergeticallyfavoredandCratomsintheselargeCr-richclustersarecoupledantiferromagnet-ically.ThecalculatedmomentsoftheCr-richclusterswithvarioussizesare0.06-1.47µB/Cr,correspondingwelltotheobservedMS∼0.25µB/Cr.Kimetal.[7]havereportedthattheMSofGa1−xCrxNdecreaseswithincreasingCrconcentration,whichcanbeinterpretedasduetothesizeincreaseoftheantiferromagneticCr-richclusters.AsforZn1−xCoxO,forwhichevensmallerMShasbeenreported,Dietletal.[25]attributedtheoriginofthesmallMStotheuncompensatedmomentsoftheCo-richAFMnanoclusters.Thesetheoreticalandex-perimentalreportsleadustospeculatethattheuncom-pensatedmomentsfromAFMCr-richclustersandthein-creaseoftheclustersizemaybetheoriginoftheobservedmagnetismanditsSi-dopingdependenceinGa1−xCrxN.Also,themagneticanisotrpy(factorof2-3inthesatura-tionmagnetization)observedforthepresentGaCrNandGaCrN:Sisamplesmayberesponsiblefortheobservedapparentferromagneticbenhavior.

WenotethattheenhancementofFMbydonoratomdopingisseeninI-dopedZn1−xCrxTe[8].InthecaseofZn1−xCrxTe,inundopedsamples,becauseZnTeisknownasanativep-typesemiconductor,I-dopingmakestheCrionbecomesneutral2+.Kurodaetal.[8]at-tributedtheoriginoftheenhancementofFMtotheCr-richclusterformationintheI-dopedsamplescausedtoattractiveforcesbetweenneutralCrimpurities(Cr2+)asopposedtotherepulsiveforcesbetweenchargedCrimpu-rities(Cr3+).Inthepresentcase,theSidopingmakesafractionoftheCrionsnegativelycharged,andthereforethechargeneutralitycannotbeadrivingforceoftheCr-richclusterformationintheSi-dopedGaN.MicroscopyobservationofGa1−xCrxN:Siarethereforehighlydesiredinfuturestudies.

Insummary,wehaveinvestigatedtheSi-dopingdepen-denceoftheelectronicstructureofGa1−xCrxNanditsrelationshiptomagnetismusingXPS,XAS,andRPES.TheupwardchemicalpotentialshiftinducedbySidop-ingintoGa1−xCrxNanddownwardonebyCrdopingintoGaN:Siwereobserved.ThenewstatesofCr2+characterwerefoundtoformnearEFinCr3dPDOSspectrabytheSidoping.ThegrowthofAFMCr-richclustersisproposedtoexplaintheobservedchangeofmagnetismbytheSidoping.

ThisworkwassupportedbyaGrant-in-AidforSci-entificResearchinPriorityAreagCreationandControlofSpinCurrenth(19048012)fromMEXT,Japan.TheexperimentatSPring-8wasapprovedbytheJapanSyn-chrotronRadiationResearchInstitute(JASRI)ProposalReviewCommittee(ProposalNo.2007A3832).WealsothanktheMaterialDesignandCharacterizationLabo-ratory,InstituteforSolidStatePhysics,UniversityofTokyo,fortheuseoftheSQUIDmagnetometer.

[1]K.SatoandH.Katayama-Yoshida,Jpn.J.Appl.Phys.

40,L485(2001).

[2]M.Hashimoto,Y.K.Zhou,M.Kanamura,H.Katayama-Yoshida,andH.Asahi,J.Cryst.Growth251,327(2003).[3]J.J.Kim,M.Makino,P.P.Chen,T.Suzuki,D.C.Oh,

H.J.Ko,J.H.Chang,T.Hanada,andT.Yao,Phys.Stat.Sol.C0,2869(2003).

[4]M.Hashimoto,Y.K.Zhou,M.Kanamura,H.Katayama-Yoshida,andH.Asahi,J.Cryst.Growth273,149(2004).[5]J.J.Kim,H.Makino,K.Yamazaki,A.Ino,H.Na-matame,M.Taniguchi,T.Hanada,M.W.Cho,andT.Yao,Curr.Appl.Phys4,603(2004).[6]J.I.Hwang,Y.Ishida,M.Kobayashi,Y.Osafune,T.

Mizokawa,A.Fujimori,Y.Takeda,K.Terai,S.-I.Fuji-mori,Y.Saitoh,Y.Muramatsu,A.Tanaka,T.Kondo,H.Munekata,M.Hashimoto,H.Tanaka,S.Hasegawa,,andH.Asahi,Phys.Stat.Sol.B243,1696(2006).

[7]J.J.Kim,H.Makino,M.Sakurai,D.C.Oh,T.Hanada,

M.W.Cho,T.Yao,S.Emura,andK.Kobayashi,J.Vac.Sci.Technol.B23,1308(2005).

[8]S.Kuroda,N.Nishizawa,K.Takita,M.Mitome,Y.

Bando,K.Oshuch,andT.Dietl,Nat.Mater.6,440(2007).

[9]K.Sato,H.Katayama-Yoshida,andP.H.Dederichs,

Jpn.J.Appl.Phys.44,L948(2005).

[10]S.Dhar,O.Brandt,A.Trampert,L.D¨aweritz,K.J.Friedland,K.H.Ploog,J.Keller,B.Beschoten,andG.G¨untherodt,Appl.Phys.Lett.82,2077(2003).

[11]L.Gu,S.Y.Wu,H.Liu,R.K.Singh,N.Newman,andD.J.Smith,J.Magn.Magn.Mater.290,1395(2005).[12]

A.Bonanni,M.Kiecana,C.Simbrunner,T.Li,M.Saw-icki,M.Wegscheider,M.Quast,H.Przybyli´nska,A.Navarro-Quezada,R.Jakiela,A.Wolos,W.Jantsch,andT.Dietl,Phys.Rev.B75,125210(2007).

[13]N.Ozaki,I.Okabayashi,T.Kumekawa,N.Nishizawa,S.Marcet,S.Kuroda,andK.Takita,Appl.Phys.Lett.87,192116(2005).

[14]N.Ozaki,N.Nishizawa,S.Marcet,S.Kuroda,O.Eryu,andK.Takita,Phys.Rev.Lett.97,037201(2006).[15]M.Funakoshi,Masterthesis(OsakaUniversity,2006).[16]S.Tanuma,C.J.Powell,andD.R.Penn,Surf.Inter.Analys.11,57(1988).

[17]T.Uozumi,K.Okada,andA.Kotani,J.Electron.Spec-trosc.Relat.Phenom.78,103(1996).

[18]

M.Hashimoto,S.Emura,H.Tanaka,T.Honma,N.Ume-saki,S.Hasegawa,andH.Asahi,J.Appl.Phys.100,

5

103907(2006).

[19]S.Shanthi,S.Kimura,M.S.Kim,S.Kobayashi,Y.K.Zhou,S.Hasegawa,andH.Asahi,Jpn.J.Appl.Phys.45,3522(2006).

[20]

M.Kobayashi,Y.Ishida,J.I.Hwang,G.S.Song,A.Fu-jimori,C.-S.Yang,L.Lee,H.-J.Lin,D.-J.Huang,C.T.Chen,Y.Takeda,K.Terai,S.-I.Fujimori,T.Okane,Y.Saitoh,K.Kobayashi,A.Tanaka,H.Saito,andK.Ando,J.Supercond.Nov.Magn.20,467(2007).

[21]S.A.Chambers,T.Droubay,T.C.Kaspar,andM.Gutowski,J.Vac.Sci.Technol.B22,2205(2004).

[22]K.Morikawa,T.Mizokawa,A.Fujimori,Y.Taguchi,andY.Tokura,Phys.Rev.B54,8446(1996).

[23]K.Sato,P.H.Dederichs,andH.Katayama-Yoshida,Hy-perfineInteractions160,57(2005).

[24]X.Y.Cui,J.E.Medvedeva,B.Delley,A.J.Freeman,N.Newman,andC.Stampfl,Phys.Rev.Lett.95,256404(2005).

[25]

T.Dietl,T.Andrearczyk,A.Lipinska,M.Kiecana,M.Tay,andY.Wu,Phys.Rev.B76,155312(2007).