AMPLITUDECONTROLOFSPIN-TRIPLETSUPERCURRENTIN SUPERCONDUCTOR/FERROMAGNET/SUPERCONDUCTORJOSEPHSON JUNCTIONS By WilliamM.Martinez ADISSERTATION Submittedto MichiganStateUniversity inpartialentoftherequirements forthedegreeof Physics|DoctorofPhilosophy 2015 ABSTRACT AMPLITUDECONTROLOFSPIN-TRIPLETSUPERCURRENTIN SUPERCONDUCTOR/FERROMAGNET/SUPERCONDUCTOR JOSEPHSONJUNCTIONS By WilliamM.Martinez Whenaconventionalsuperconductor(S)isplacedincontactwithaferromagnet(F), thedecaylengthofthepaircorrelationsintheferromagnetisveryshort,ontheorderof anminastrongferromagnetsuchasCoorFe.Thisisduetothespin-polarizednature oftheferromagnet,whereasthespinsofCooperpairsinaconventionalsuperconductorare anti-alignedinaspin-singletstate.However,in2001,theoristspredictedthatlong-range paircorrelationsinaspin-tripletstatecouldbegeneratedthroughmagneticinhomogeneity. Withparallelspins,thedecaylengthofthesecorrelationsextendsinprincipletothatofa superconductor-normalmetalsystem,whichcanbeontheorderofamicronattly lowtemperature. Thishasbeenobservedexperimentallybyseveralgroups,commonlythroughthe useofextrinsicmagneticinhomogeneityinsampleswithmultiplemagneticlayers.Josephson junctionmeasurementshavedemonstratedcriticalcurrentsordersofmagnitudelargerin sampleswiththisinhomogeneitycomparedtosampleswithout.However,theabilityto reliablycontrolthespin-stateofthepaircorrelationsinasinglesamplehasyettoberealized. ThegoalofthisworkistoperformmeasurementsonJosephsonjunctionsinwhichthe inhomogeneitycanbemanipulated.OurapproachistofabricateS/F'/F/F"/SJosephson junctionswherewecancontroltherelativemagnetizationorientationsofallthreeferromag- neticlayers.Inordertorealizethiscontrol,wehadtoperformstudiestocharacterize variousmagneticmaterials,mostnotablyaNiFealloysimilartoPermalloyandCo/Ru/Co, asyntheticantiferromagnet.StudiesoftheNiFehavedemonstrateditsabilitytobeused asaspin-tripletgenerator.MeasurementshavealsobeentakenofNiFetodetermine howeasilyitsmagnetizationdirectioncanberotatedinanexternalWehavealso measuredthemagnetichardnessofCo/Ru/Cosyntheticantiferromagnetsasafunctionof theCothickness.BykeepingtheCothin,wecanminimizetherotationofthislayerunder theofsmallappliedmagnetic Usingtheseresults,wedemonstrateamplitudemodulationofthesupercurrentin S/F'/F/F"/SJosephsonjunctionswhichisdependentonthemagnetizationdirectionof NiFe.ThroughtheuseofanexternalthemagnetizationoftheNiFeF"layercanbe rotatedwithrespecttothemagnetizationoftheCo/Ru/CoFlayer.Byrotatingintoand outofanon-collinearstate,wehavedemonstratedtheabilitytotunethesupercurrentfrom aspin-triplettoaspin-singletstate,elyturningthesupercurrentinthesejunctions \on"and Thisthesisisdedicatedtomyparents,AnitaandMichaelMartinez. iv ACKNOWLEDGMENTS Inwritingtheseacknowledgments,IamremindedofhowfortunateIamtohavehadthe supportofsomanyamazingpeople.Therearefartoomanytonamethemall,andIsincerely apologizeinadvanceforleavinganyoneout. Firstandforemost,Iwouldliketothankmyadviser,ProfessorNormanBirge.Since thephysicsclassIhadasanundergraduatealmost11yearsago,Ihaveturnedtoyou foradviceonawiderangeoftopics.Yourguidanceandmentoringhavebeeninvaluable.I sincerelythankyouforyourtimeandpatiencethroughtheyears,aswellasyourconstant assistanceduringthestumblesalongtheway.Iamhonoredtohavehadyouasanadviser. IalsowanttothankProfessorWilliamPratt,whoseexperimentalknowledgeandengi- neeringskillsknownoequal.Aconstantinthelab,Icannotthankyouenoughfor allyourhelp,especiallyregardingQD-IIandQD-VII.Imightstillbetryingtorepairthose probesifitweren'tforyou. Itisnoexaggerationtosaythatmyresearchwouldnotbepossiblewithoutthehelpof Dr.RezaLoloeeandDr.BaokangBi.IcannotbelievehowmuchIhavelearnedfromyou both,andamsothankfulforallyourtimeandpatience.Ihavelookeduptoyoubothfor solongandwithpridecallyounotonlymymentorsbutalsomyfriends. Ihavereceivedalotofguidanceandencouragementfrommycommitteemembers:Pro- fessorsWilliamHartmann,CarloPiermarocchi,KirstenTollefson,andRemcoZegers.When variousprojectsweren'tpanningoutasexpected,thesoundadviceIreceivedfromthemkept meontrack. Myresearchgrouphasbeenamazing,andIhavelearnedsomuchfromvariousmem- bers,pastandpresent,andIhopeIcanpassonevenhalfofit:MichaelCrosser,Charles v Moreau,GassemAlzoubi,TruptiKhaire,MazinKhasawneh,YixingWang,EricGingrich, BethanyNiedzielski,JosephGlick,VictorAguilar,AlexCramer,BrandonEwert,Patrick Quarterman,KevinBarry,KevinWerner,KurtBoden,andIlyaBeskin.Beyondmentoring, you'vemadethelabmuchmorefun.Thatalsogoesforallmyothercoworkers,especially SeanWagner,IanDayton,JessWest,ConnorGlosser,andBaiNie.You'veallmadethe basementabrighterplace. IwouldberemissifIdidn'tacknowledgethehelpfromvariousfacultyandin BPS.Somuchofourequipmentismadeoralteredinthemachineshop,whichhasan unbelievablyfriendlyandhelpfulIwanttorecognizeTomPalazzolo,TomHudson, JimMuns,andRobBennettforthecountlesshoursspentmachiningthepartswerely on.Whileexperimentalistscouldn'tgetbywithoutthemachineshop,noneofthestudents couldgetthroughgraduateschoolwithoutthehelpofoursecretaries,whoalltoooftengo unrecognizedfortheirwork.Fortheirhelpandpatiencealongtheway,Iwanttothank CathyCords,DebbieBarratt,andKimCrosslan,myworkmom. Speakingofmoms,Ihavetoacknowledgemyfamily,includingtheStockmeiersand Maga~nasinthatcategory.Thisisespeciallyextendedtomyparents,MikeandAnita.They havestoodbymeandhelpedoutmorethanIcansay,evenwhenIhaven'tbeensopatient inreturn.MybrotherJohndeservesequalthanks,ashealwaysknewhowtogetthesmile backonmyfacewheneveritstartedtofade. Inadditiontothosepreviouslymentioned,Ihavetothankmyfriendsthathavepulled methroughmorethantheyrealize.TothoseI'vemetingraduateschool,especiallyKatie Corker,PaulCurran,SteveFawley,EricMacaulay,AaronMosier,andKellySmith,Iknow weallhavelimitedfreetimeinourprograms.Whetherplayingboardgamesorhaving philosophicaldiscussionsoverbeer,itmeansalotthatyou'dspendyourswithme.With vi that,therecentadditionofmurdermysterieswithmyneighborsBenLosethandNickiLarson havebeenagreatwaytowrapupaweekend.Iwanttothankmyteammates,especially KortneyCooper,DanielCouvertier,AaronChrisProkop,andSteveQuinn.Running aroundonthecourtsorinthesunlighthasbeenabreathoffreshairandawelcomechange fromtheworkroutine. Iamfortunateenoughtostillbeclosewithmanyfriendsfrommypast,andtheyhave allbeensourcesofbothenjoymentandencouragement.TothoseI'vemetasanundergrad, especiallyPatrickBreen,Craig,CullenDembowski,AndrewKeller,TylerNeal,and CalebWojcik,aswellasthosefromevenbefore,includingPeterButzen,PaulCasperson, PatDineen,MonicaFumorolo,AlexMohr,LoganPrice,MarieRusso,MaggieRyan,Jack Scacco,TimSchram,ArjunVenkataswamy,andTerryWhittington,Icannotexpresshow muchyoumeantome. SpecialthanksmustgotomyLittle,DeveontyeBrown.Whilewehaven'tbeenableto hangoutasmuchaseitherofuswouldhaveliked,IknowIcancountonyoutohelpme switchoutofworkmodefromtimetotime. Despiteintomanyoftheabovecategories,individualrecognitionhastobepaidto AimeeShore.Amazinglysupportive,especiallyinthesepastfewmonths,theterm\neigh- bor"doesn'tseemtodoitjustice.Youandthedrawingonmyfridgeareconstantreminders oflifeoutsideofthebasement. Whetherexplicitlymentionedorapologeticallyomitted,thankyouall. ThisworkhasbeensupportedbyUS-DOEgrantDE-FG02-06ER46341. vii TABLEOFCONTENTS LISTOFTABLES .................................... xi LISTOFFIGURES ................................... xii Chapter1Introduction ............................... 1 1.1Motivation.....................................1 1.2ThesisStructure..................................3 Chapter2Theory ................................... 6 2.1Ferromagnetism..................................6 2.1.1BalancingAct:EnergiesandDomains.................8 2.1.2Magnetization...............................11 2.2Superconductivity.................................14 2.3Proximity..................................16 2.3.1Superconductor/NormalMetal......................16 2.3.2Superconductor/Ferromagnet......................19 2.4JosephsonJunctions...............................23 2.4.1RCSJModel................................26 2.4.2FraunhoferPattern............................31 2.5Spin-TripletPairCorrelations..........................34 Chapter3RelevantPreviousWork ........................ 39 3.1InitialEvidenceofSpin-TripletPairCorrelationsinS/FSystems......39 3.2ReproducibleGenerationoftheSpin-TripletSupercurrent..........40 3.3CharacterizationofSpin-MixerLayers.....................44 3.3.1Ni.....................................44 3.3.1.1Generation...........................44 3.3.1.2SwitchField..........................44 3.3.2NiFeMo..................................45 3.3.2.1Generation...........................47 3.3.2.2SwitchingField.........................49 3.4ExternalWork..................................50 3.4.1Spin-TripletJosephsonJunctions....................52 3.4.2RotatingMagnetizationinPseudoSpinValves.............53 3.4.3FieldDependentSpin-TripletAmplitudeModulation.........53 3.5AdditionalBackgroundInformation.......................56 Chapter4SampleFabrication ........................... 57 4.1FabricationEquipment..............................57 viii 4.1.1Lithography................................57 4.1.1.1Photolithography........................58 4.1.1.2ElectronBeamLithography..................59 4.1.2DepositionandMilling..........................60 4.1.2.1ThermalEvaporation.....................61 4.1.2.2Sputtering............................63 4.1.2.3IonMilling...........................67 4.1.3Microscopy................................69 4.1.3.1ScanningElectronMicroscopy.................69 4.1.3.2AtomicForceMicroscopy...................72 4.2Procedure.....................................74 4.2.1WafertoChip...............................76 4.2.2FabricatingBaseLayer..........................76 4.2.3andMillingJosephsonJunctionPillars............79 4.2.4FabricatingTopLeads..........................82 Chapter5Measurement ............................... 84 5.1QuickDipperII..................................84 5.2SQUIDElectronics................................89 5.2.1OverviewofSQUIDs...........................89 5.2.2SQUIDCurrentComparatorCircuit..................91 5.3System.......................................93 Chapter6Characterization ............................. 95 6.1UseofNiFe....................................95 6.1.1Generation................................95 6.1.2SwitchingField..............................97 6.2Co/Ru/Co.....................................99 6.2.1AnisotropicMagnetoresistance......................101 6.2.2AMRProcedure..............................102 6.2.3AMRData................................103 Chapter7ControlofSpin-TripletSupercurrent ................ 109 7.1Procedure.....................................109 7.1.1InitialFraunhoferPattern........................109 7.1.2Switching:Zero-FieldMeasurements..................113 7.1.3Switching:FraunhoferMeasurements..................113 7.1.4Switching.............................114 7.2RightPlaceattheRightTime..........................114 7.3EarlyAttempts..................................115 7.3.1RotatetheSAF..............................115 7.3.2LargeNiFeMoPillars...........................119 7.3.3SmallNiFeMoPillars...........................124 7.4NiFePillars....................................126 7.4.1FirstEvidence...............................126 ix 7.4.2AFabricationHiccup...........................131 7.4.3Reproducibility..............................133 7.4.4QuantitativeAnalysisofRatios.................141 Chapter8Conclusions ................................ 144 8.1Overview......................................144 8.2SummaryofResults...............................145 8.3FutureWork....................................147 APPENDIX ........................................ 149 BIBLIOGRAPHY .................................... 178 x LISTOFTABLES Table7.1Tableof R k and R ? forCo(4)SAF...................142 Table7.2Tableofquantitativeratioanalysis................143 xi LISTOFFIGURES Figure2.1Magnetizationcurves...........................12 Figure2.2Cartoonofpairleakagemodels.....................17 Figure2.3Interfacebandstructure.........................20 Figure2.4Paircorrelationfunctions.........................22 Figure2.5AsimpleJosephsonjunction.......................25 Figure2.6Schematicofaresistivelyandcapacitivelyshuntedjunctionmodel..26 Figure2.7Tiltedwashboardmodel.........................29 Figure2.8CharacteristicI-VcurveforanoverdampedJosephsonjunction...31 Figure2.9RepresentationofacircularJosephsonjunction............32 Figure2.10Airypattern................................34 Figure2.11Structureforspin-tripletgeneration...................36 Figure2.12Tripletjunctionphasecartoon......................37 Figure3.1Spin-tripletpillarschematic.......................41 Figure3.2Plotofinitialspin-tripletevidence....................42 Figure3.3Schematicoflayermagnetization....................43 Figure3.4NiDCSQUIDMagnetometerdata...................46 Figure3.5ExampleofFraunhoferpatternforNiFeMogenerationsamples....48 Figure3.6 I c R N vsNiFeMoThickness.......................49 Figure3.7Ellipsepatterngeometry.........................50 Figure3.8NiFeMoSwitchingField.........................51 Figure3.9JunctionusedbytheBlamiregroup...................55 xii Figure3.10DatatakenandsimulationsrunbytheBlamiregroup.........55 Figure4.1Photoofsputteringchamber.......................65 Figure4.2Schematicofionmillchamber......................68 Figure4.3SchematicofHitachiSEM........................71 Figure4.4Cartoonsofsamplefabricationsteps..................75 Figure4.5Schematicofbaseleadmask.......................77 Figure4.6Imageofma-Npillars...........................80 Figure4.7Imagesofma-N.........................81 Figure4.8ImageofVernieralignmentmarks....................83 Figure5.1SchematicofmeasurementelectronicsinQuickDipper-II.......86 Figure5.2Top-downrepresentationsofleadgeometryacrosssample.......87 Figure5.3MountingQD-II..............................88 Figure5.4SQUIDLoop...............................90 Figure5.5Schematicofmeasurementset-up....................93 Figure6.1ExampleofFraunhoferpatternforNiFegenerationsamples.....97 Figure6.2 I c R N vsNiFeThickness.........................98 Figure6.3NiFeswitching...........................100 Figure6.4SAFcartoon................................100 Figure6.5mechanism...........................101 Figure6.6CartoonsofSAFmagnetizationdueto.............102 Figure6.7Resolutionissuesoflock-in.................103 Figure6.8Circuitdiagramofaratiotransformer.................104 Figure6.9AMRdataforCo(2)andCo(4).....................105 xiii Figure6.10AMRdataforCo(6)andCo(8).....................106 Figure6.11AMRSummary..............................107 Figure7.1Representationsofmagnetizationdirectionintheonstate......111 Figure7.2Representationsofmagnetizationdirectioninthestate......112 Figure7.3InitialFraunhoferpatternmeasurementsforNi/SAF/Nisamples...116 Figure7.4InitialswitchingbehaviorofNi/SAF/Nisamples............116 Figure7.5On-andmeasurementsforNi/SAF/Nisamples.......117 Figure7.6InitialFraunhoferpatternmeasurementsforNi/SAF/NiFeMosamples.120 Figure7.7Low-statemeasurementsofNi/SAF/NiFeMosamples.........121 Figure7.8High-statemeasurementsofNi/SAF/NiFeMosamples.........122 Figure7.9I-VcurvesforNiFeMoandNiFe.....................125 Figure7.10InitialmeasurementsforNi/SAF/NiFe{Sample1...........127 Figure7.11Sample1switchingdata.........................128 Figure7.12FraunhoferpatternsmeasuringtheswitchinSample1........129 Figure7.13OswitchingforNi/SAF/NiFe{Sample1.............130 Figure7.14InitialandFraunhofermeasurementsforNi/SAF/NiFe{Sample 1......................................131 Figure7.15Fielddependenceofonswitching..................132 Figure7.16InitialFraunhofermeasurementsofNi/SAF/NiFe{Sample2.....134 Figure7.17Ni/SAF/NiFeswitchmeasurement{Sample2..........134 Figure7.18Fraunhoferpatternsmeasuringtheswitch{Sample2.........135 Figure7.19OswitchingforNi/SAF/NiFe{Sample2.............135 Figure7.20InitialFraunhofermeasurementsofNi/SAF/NiFe{Sample3.....137 xiv Figure7.21Ni/SAF/NiFeswitchmeasurement{Sample3..........137 Figure7.22NeaFraunhoferpatternsmeasuringthestate{Sample3.138 Figure7.23Fraunhoferpatternsmeasuringtheswitch{Sample3.........138 Figure7.24InitialandFraunhofermeasurements{Sample3.........139 Figure7.25OswitchingforNi/SAF/NiFe{Sample3.............139 Figure7.26Cartoonofmagnetizationdirectionandrelativeanglesforthe stateusedinratioanalysis....................142 FigureA.1Cartoonofperpendicularmagneticanisotropy.............153 FigureA.2ImagesoflateralJosephsonjunction..................154 FigureA.3SEMimageofalateralgeometryproximitysample.......154 FigureA.4MFMresultsofNinanowires......................156 FigureA.5Cartoonsoflateralgeometryfabricationsteps.............158 FigureA.6Imagesofsidewallissues.........................160 FigureA.7Cartoonsofresist.......................161 FigureA.8Imagesofjunctiononconcerns..................162 FigureA.9Imagesofresistdeformation.......................163 FigureA.10Plotsofionmillcharacteristics.....................164 FigureA.11ImagesofQD-VII.............................167 FigureA.12Diagramofvariousloadlineschemes..................169 FigureA.13CircuitdiagramforlateralgeometryJosephsonjunctionmeasurements.169 FigureA.14S/N/SJosephsonjunctionmeasurements................170 FigureA.15TemperaturedependenceonS/N/Scriticalcurrent..........171 FigureA.16S/F/SandS/F'/F/F'/SJosephsonjunctionmeasurements......171 xv FigureA.17Datafrom Rvs:T proximitymeasurements............173 FigureA.18ProximityinJosephsonjunctionsamples.............175 FigureA.19Proximitygeometries.......................177 xvi Chapter1 Introduction 1.1Motivation Attheturnofthe20thCentury,experimentalistswereinvestigatingtheoftemperature ontheelectricalresistanceofmetals.Theyhadfoundthat,asthetemperaturedecreases, theresistanceofthemetalalsodecreases.However,duetothelimitedtemperaturerange ofrefrigerationsystemsavailableatthetime,thequestionof\Whathappenstoametal's resistanceasitapproachesabsolutezerotemperature?"hadremainedonefortheorists. However,in1908,HeikeKamerlinghOnnessuccessfullyHelium,whichcon- densesat4.2K.Thisnovelcryogenicliquidopenedaregimeoflow-temperaturephysics thathadyettoberealized.In1911,Onnesdippedasampleofmercuryintoliquidheliumto measuretheresistanceasitdroppedtowards0K,discoveringsuperconductivityinthepro- cess[1].Althoughittookdecadestodevelopatheoryforthisdropinresistance[2],research involvingsuperconductivityhasbecomeamajoraspectofcondensedmatterphysics. Sincethediscoveryofsuperconductivityjustoveracenturyago,numerous haveemerged,manyleadingtotechnologicalapplicationsusefultosociety.Superconducting magnetsareusedinawiderangeoffrommedicalusesinMRImachinestoaccelerator labs,wherenuclearandhighenergyphysicistsaroundtheworldprobetheveryfoundation ofmatter. Althoughtheymaynotbeinthepubliceyeasmuchasthosementionedabove,researchis 1 constantlybeingdonetonewapplicationsforsuperconductorsinsociety.Forevidence ofthis,weneedtolooknofurtherthanthelimitationsofsilicon-basedcomputing[5]. Whiletraditionalcomputingisreachingitslimit,thedesireforhigherpoweredcomputers continuestogrow,evidencedbyfrequentbenchmarkingcompetitions[6].Inanto moveawayfromthesilicon-basedtransistorsinconventionalcomputing,alotofworkis beingdonetorealizequantumcomputing.Thefoundationofquantumcomputingrevolves aroundsuperconductingqubits,quantumdotsthatremaininasuperpositionoftwostates. Althoughonecompanyhasclaimedtohavedevelopedtheiterationofthesecomputing \holygrails"[7][8],criticsandsupportersalikeagreethatquantumcomputingisinits infancy. Superconductinglogiccircuits,includingreliablemagneticrandomaccessmemory (MRAM),isanothertypeofappliedsuperconductivitybeingsought[9][10].Inprinciple, superconductingmemorycanbeachievedbyeithercontrollingthesupercurrentamplitude orthephaseofaJosephsonjunction.Theworkdescribedinthisthesisisalsoaimedatre- alizingcontroloverthesupercurrentamplitudeinsuperconducting-ferromagneticJosephson junctions,theculminationofdecadesoftheoreticalandexperimentalresults[11]. Inthiswork,weuseconventionalsuperconductors(S)thatcarryspinsingletCooper pairs.InJosephsonjunctionswithanormalmetal(N)barrier,supercurrentcanbemeasured eveninjunctionswithbarriersasthickasamicron.Whenaferromagnet(F)isinsertedasa barrierbetweentwoSelectrodes,theCooperpairsdieoutveryquickly,anofthespin- polarizationofferromagnets[12].However,previoustheoreticalandexperimentalresults haveshownthatinJosephsonjunctionswithmultiplemagneticlayers,i.e.S/F 1 /F 2 /F 3 /S junctions,itispossibletogeneratespin-tripletpaircorrelationsfromthespin-singletCooper pairs[13].Thesecorrelationshavemuchlongerlengthscales,reminiscentofthoseofS/N/S 2 systems. Aswillbediscussedinthenextchapter,thistripletgenerationisdependentonmagnetic inhomogeneitybetweenferromagneticlayers.IfadjacentFlayershavecollinearmagneti- zationdirections,therewillbenotripletgenerated.Thespin-singletpaircorrelationswill dephaserapidly,andthecriticalcurrentwillbequitesmall.If,however,thereisnon- collinearitybetweenalladjacentFlayers,alargespin-tripletsupercurrentcanbemeasured acrossthejunction.Therefore,ifonecancontroltheorientationofthemagnetizationsof theFlayers,theabilitytotunetheamplitudeofcriticalcurrentfollows.Controloverthese parameters,themagnetizationandcriticalcurrentamplitude,isthefocusofthisthesis. Whiletheworkhereinhasbeeninvestigatedlargelyforthesakeofdiscoveringnew andnoveladvancementsforscience,Iwillmakethenaiveclaimthatitmaybeusefulin anyapplicationthatrequiresasystemwithbinarycontrolofasupercurrentamplitude, suchastheMRAMmentionedabove.Regardlessoffutureoutcomesorlackthereof,itis withexcitementthatIcanpresenttheresultsofthiswork,demonstratingreliablecontrolof criticalcurrentamplitudeinsuperconductingJosephsonjunctionswithferromagneticbarrier layers. 1.2ThesisStructure Theintentofthisthesisistodemonstrateourmostrecentattemptstocontrolthespin- tripletsupercurrentinsuperconducting-ferromagneticJosephsonjunctions.Thereisalot ofbackgroundinformationthatneedstobeconveyedandwillbediscussedinthe followingway: Chapter2willdiscussthetheoreticalbackgroundofthesystemsemployedinourwork. 3 Itwillstartwithanoverviewofferromagnetismandsuperconductivityandmoveontothe physicsattheinterfaceofthesematerials.ThebackgroundprinciplesofJosephsonjunctions willfollow,withanemphasisontheshort-rangenatureofthesupercurrentinjunctionswith aferromagneticbarrier.Thechapterwillwrapupwithadiscussionofvariousmechanisms togeneratelong-rangesuperconductingpaircorrelations,capableofbeingdetectedthrough thickferromagneticlayers. Chapter3willtakealookatthepreviousexperimentalworkthatfacilitatedthisresearch. Theresultsthereinwillbemostlycomposedofrelevantcontributionsfrompreviousand currentgroupmembers;theseincludetherealizationoflong-rangespin-tripletcorrelations, thecharacterizationofNiasahardferromagnet,andthecharacterizationofNiFeMoasboth aspin-tripletgenerationlayeraswellasasoftmagnetwhosemagnetizationcanbeswitched easily.Thischapterwillalsofeatureworkdonebyanothergroupinvestigatingverysimilar physics;junctionstheyhavecreatedhavedemonstrated,inacertainmeasurementscheme, theabilitytoswitchfromspin-singlettospin-tripletsupercurrent. Chapter4willfocusonsamplefabrication.Itwillfeatureananalyticallookatthe equipmentusedforpreparation,deposition,andcharacterizationofsamples,allofwhichhave beenmadein-house.ItwillalsofeaturetherecipeusedtocreateourJosephsonjunctions. LikeChapter4,Chapter5willbemechanicalinnatureasitdiscussestheequipment necessarytomeasureoursamples.Thisincludesthefunctionalityofthequick-dippersystem usedtosubmergesamplesinliquidheliumandtheprinciplesofaSQUID-basedmeasurement system.Theformerisnecessarytogeneratetherequiredmagneticatthesample whilethelatterisahigh-precisionmeasurementtoolthatyieldsexcellentsignal-to-noise capabilities. Chapter6isincludedtodemonstratetheimportantpropertiesofindividuallayersinour 4 Josephsonjunctions.Withthreetmagneticlayersbeingutilizedinthesesamples,it isimportanttounderstandhoweacharebyanexternalmagneticThechapter willbeprimarilyconcernedwithtwolayers:NiFeandCo/Ru/Co.(TheNiFeusedinthis workdeviatesmarginallyfromthecommonalloyPermalloy(Py),whichconsistsof80%Ni, 20%Fe.)Discussionoftheformerwillfocusonitsspin-tripletgenerationcapabilitiesand theeaseofrotatingitsmagnetizationinthesamemannerashadbeendoneforNiFeMo. Incontrast,discussionofthelatterwillfocusonitsabilitytobeusedasarelativelyhard ferromagnet,demonstratedinthemagnetization'slackofresponsetoanexternalat leastwhentheissmall. Finally,inChapter7,thedatademonstratingourabilitytocontrolspin-tripletsupercur- rentinferromagneticJosephsonjunctionswillbepresented.Thiswillincludetheprocedure utilizedtoobtainthesemeasurementsanddiagrammaticrepresentationofeachlayer'smag- netizationdirectionatvarioustimesofmeasurement,animportantunderstandingnecessary togleananyrelevantinformationfromthedata. Chapter8willbeabriefsummaryoftheworkthatprecededit,aswellasaquicklook atfutureresearchareasthatemergeasaresult. Forthoseinterested,theappendixwillfeatureimportantaspectsofaprojectintendedto probetherangeofthespin-tripletsupercurrentinsuperconducting-ferromagneticJosephson junctions.Tothedismayofthisauthor,despiteevidenceofsuccessfulfabrication,the supercurrentinthisgeometryhasnotyetbeenobservedasofthetimeofwritingthisthesis. Whilenotentirelyunderstood,possiblereasons,aswellaspotentialdirectionsforfuture research,willbeanalyzed. 5 Chapter2 Theory Traditionally,superconductors(S)andferromagnets(F)areconsideredsystemswithop- posingformsoforder.Thespin-polarizednatureofferromagnetsdirectlycontraststhe anti-parallelspinnatureofCooperpairs.Andyet,theworkdescribedinthisdissertationis fundamentallyreliantontheinterplaybetweenthesesystems.Thissectionwilldiscussthe principlesofthesematerials,theinteractionattheirinterface,andthebackgroundofhow experimentalinterestemergedfromsystemsofsuchopposingorder. 2.1Ferromagnetism Magnetismisaphysicalpropertywithwhicheveryreaderlikelyhasexperience,fromatool withwhichonecanitemstothekitchenrefrigeratortoanavigatingaidintheform ofacompassneedle.However,despitetheofmagnetismbeingknownformillennia, thefundamentalpropertiesofferromagnetsarecomplexenoughtowarrantresearchinterest today. Ferromagnetsarematerialswhosemagneticmomentsarepredominantlyalignedparallel totheirnearestneighbor,creatingamicroscopicmagneticmoment ~ = ge 2 m ~ J (2.1) 6 where ~ J istheorbitalorspinangularmomentum, e isthechargeofanelectron, m isthe massofanelectron,and g isthegyromagneticratio,suchthat g =1forpurelyorbital motionand g =2forpurelyspin(else1 0oranti-parallel if J ab < 0.Ferromagnetshave J ab > 0. 2.1.1BalancingAct:EnergiesandDomains Amaterialwilldevelopamagneticstructuresuchthatitsenergyisminimized.While exchangeisamajorsourceofthetotalmagneticenergy,thereareseveralothertypesof interactionthatalsothemagnetizationofasample.Theseincludemagnetostatic, magnetocrystalline,magnetostrictive,andZeemanenergies. Magnetostaticenergyisdirectlyrelatedtotheshapeofthematerial.Todeterminethe magnetostaticenergy,onemustintegratetheinteractionbetweenthemagnetizationand internal ~ H ,whichpointinoppositedirectionsduetothedemagnetizingnatureofthe internal[18].Thelargertheregionofalignedspins,thelargertheinternalbecomes inoppositionofthismagnetization.Dependingonsizeandshapeofthematerial,thisenergy canbecomequitelarge,forcingthesystemtootherwaystominimizeinternalenergy (discussedbelow).Ifthesystemhasafavoredmagnetizationdirection,itiscalledtheeasy axis.Aspherehasperfectshapesymmetryaboutallaxesandthereforehasnopreferential magnetizationdirection.Incontrast,alongstraightwirehasuniaxialsymmetryinwhich theenergyisminimizedbypointingthemagnetizationalongthewire.Athindisc,whichis symmetricintwoplanesbutverythininthethird,willminimizeenergywithaneasyaxis layingintheplaneofthedisc.Thisgeometryistheoneutilizedinthiswork,andalthough itthemagnetizationin-plane,themagnetostaticenergyforcesnofurtherfavored directionwithinit. 8 Themagnetocrystallineenergyisanofmagnetocrystallineanisotropywhicharises fromthecrystalstructure.Toorderthisisanofspin-orbitcoupling,andthe orbitalmotionofelectronscancoupletotheelectricinherentinthecrystal[18][19].In ordertomagnetizationinthisway,however,theremustbeasymmetryinboththe electricandorbitals.Otherwise,nodirectionisfavorable,andmagnetizationcanalign randomly.Forexample,incertainalloyssuchasNiFe(Permalloy),thelocationsoftheNi andFeatomscaninduceamagnetocrystallineanisotropy.Thedirectionofthisanisotropy canbesetbygrowingNiFeinanexternalmagneticAsthegrows,the positionsofNiandFewithinthecrystalpreferentiallyfavortheasymmetrydesired.While weakcomparedtotheexchangeinteraction,theofmagnetocrystallineanisotropycan apreferentialdirectionwhenothersmaynotexist.Consideringtheundetermined easyaxisintheexampleofathin-discabove,controllingthecrystalgrowth,andthus magnetocrystallineanisotropy,willtheeasyaxisinasingledirection. Magnetostrictionisanofthestressesonthemagneticmaterial.Theenergyfrom thesestressesrelateshowaferromagneticsamplewillexpandorcontractduetoitsmag- netization.Conversely,straininducedduringlmgrowthcaninducemagneticanisotropy. Theshapeofthesystemcanchangeduetothisenergy,whichbringsaboutyetanother competitoroftheexchangeenergy. TheZeemanenergyariseswhenasampleisexposedtoanexternalInorder tominimizetheinternalmagneticenergy,momentdirectionandmagneticstructuremay changeinthepresenceofaThisisdiscussedmorethoroughlyinSection2.1.2. Clearly,ferromagneticsystemsareverycomplex.Therefore,whentryingtominimizethe internalenergy,itisnecessarytoconsidermuchmorethansolelytheparalleloranti-parallel natureofnearbymoments.Inbulkmaterials,thesecompetingenergiescanovercomethe 9 exchangeenergy,minimizingthetotalenergyinthesystemthroughthecreationofdomains. Theexistenceofdomainswasoriginallypredictedin1907[20],andtheexperimental ofthemwereobservedin1919[21].Bycreatingdomains,areasofalignedspin thataretfromoneanother,themagnetostaticandmagnetostrictiveenergiescan becomeminimizedatthecostofanincreaseoftheexchangeenergyattheedgeofthese domains[22].Theseedges,calleddomainwalls,areregionswherethespinsofnearby momentsundergoachangeofdirection. Domainscanbedirectlyobservedanumberofways.Theobservationoccurred whenmagnetitedepositsorientedalongsurfacedomainwallsofaNisample[23].Whilethis demonstratedtheboundaryregions,thedomainsthemselveswerelaterimagedoptically usingpolarizedlight[24].Theplaneoflightpolarizationisbythedomaindirection, andthereforecanbeusedtodirectlyimagesurfacedomains. Tominimizethemagnetostaticenergy,whichisdirectlyanalogoustotheexternalmag- neticamaterialwillcreateitsdomainstructureinordertoreduceitsexternal Domainswhicheliminatethisarecalledclosuredomains[19],andmaterialswithper- pendiculareasyaxeswillformclosuredomainswithnoadditionalenergycost.Otherwise, closuredomainswouldnotbepointingalongtheeasyaxis,therebyincreasingthemagne- tocrystallineenergy.Thedomainsizewillalsobebythemagnetostrictiveenergy. Domainsthathavemagnetizationsunalignedwillexpand/contractthecrystalin ways.Thiseaddsanelasticstrainenergy,whichissmallerforsmallerdomains.Tokeep domainssmallwould,however,requiremoreofthem,whichincreasestheexchangeenergy. Withsomanycompetingenergyterms,determiningthedomainstructureofamaterialis nosimpletask. Thesizeofthewallisanothercomplexconsideration[19].Anabruptswitchisunfavor- 10 ablewithrespecttoexchangeenergy,whichwouldpreferaslowrotationofspin.Iflong enough,nearestneighborsaremarginallymisaligned,keepingenergybetweenthesemoments low.However,shapeandcrystallinestructurefavoralignmentalongtheeasyaxis(ifoneis present).Becauseofthis,slowrotationofspinsmaynotbeasfavorable,insteadpromoting anarrowdomainwall. Ifthesampleiskeptsmallenough,itispossibletocreateferromagneticsthatare singledomain.TheNiFeswitchinglayerdiscussedinthiswork(seeSections6.1,7.4)has beenpatternedsmallenoughtoensurethis. 2.1.2Magnetization TheZeemanenergyariseswithinamaterialwhenitisplacedinanexternalThis additionalenergytermwillalterthedomainstructure,causingmagneticdomainstomove orevenberemovedcompletely.Thisprocesswillbedescribedhereandisdemonstratedin Figure2.1a. Considerabulkferromagnetthathasdomainspointinginalldirections.Whilemini- mizingtheinternalenergy,domainshaveformedthatresultedinnonetmagneticmoment. Recallthatminimizingtheinternalenergyisanalogouswithminimizingexternali.e. enclosingthemagnetizationcompletelyalsominimizestheinternalenergy.Atthispoint, thissampleisconsideredatitslowestenergystateandunmagnetized. Asanexternaldisapplied,magneticmomentswillattempttopointinthesame directionasthisThecanbeassubtleasincreasingthesizeofdomainsthat pointparalleltotheexternaltomoreextremesuchascompleterotationor eliminationofdomainsdependingontheeldandsizeofindividualenergyparameters.For example,along,thinwirewillmovedomainsalongit'slengthmoreeasilythanrotating 11 (a) (b) Figure2.1 Magnetizationcurves. a)InitialapplicationofH,willalterthedomain structure(segmentedgreensections).Domainswithmomentsinthedirectionofthe willincreaseinsizewhileopposingmomentswilldecrease.Eventuallyonlyonedomainwill remainandrotateinthedirectionofappliedsaturatingthemagnetization(M s ).b) Afterinitialmagnetization,measuringMvsHwilldemonstratehysteresis.Themagneti- zationthatremainsat H =0iscalledremnantmagnetization(M R ),whilethethat cancelsthemagnetizationiscalledthecoercive(H c ). 12 themagnetizationoutofthewireitself,whichwouldforcethemagnetizationtopointalong ahardaxis.Withinacertainlimit,theprocessisreversible.Decreasingtheexternal stillallowsforthedomainstorotateandmoveinanattempttoresetbacktoits original,unmagnetizedstate.However,eventuallythedomainscanbecomepinned,either throughcrystalirregularitiesortheeliminationofdomainwalls.Oncethisoccurs,itisnot possibletoreturntoanunmagnetizedstate,atleastnotbyremovalofthedalone(while demagnetizingprocessesexist,includingheatingthematerialaboveitsCurietemperature, theyarenotusefulforthisworkandwillnotbediscussedfurther).Afterremovalofthe externalthatwhichremainsiscalledtheremnantmagnetizationandiscreatedbythe ferromagnet[22]. Oncepinningbegins,furtherincreasesofexternalthesizeoftheremnant bymoving,removing,orrotatingmoredomainsthatalsobecomepinned.Eventually, however,allmomentswillpointinthedirectionoftheandnomoremagnetizationcan beinduced.Thisupperlimitisaptlycalledthesaturationmagnetization,M S . AsshowninFigure2.1b,futuremeasurementofthemagnetizationasafunctionof willyieldahysteresisloop.Therequiredtodemagnetizethesample(i.e.whenthe curvecrossesM=0)iscalledthecoercive( H c ).Thesehysteresisloopsandmagnetization measurementsareimportantwhendeterminingthehardness/softnessofmagneticmaterial (i.e.larger H c ! hardermagnet),anecessaryconsiderationwhenchoosingmaterialstobe usedinsamples. 13 2.2Superconductivity Intheearly1900s,experimentalistswereexploringthelimitsoftransportinmetalsatever coldertemperatures.Ithadbeenwellknownthatresistancedropswiththetemperature, butwhathappenedwhen T ! 0 K hadlongbeenacuriosity.In1911,aidedbythediscovery ofliquidhelium,HeikeKamerlinghOnnesmeasuredtheresistanceofmercuryasitcooled. Whathefoundwastheevidenceofsuperconductivity,adropinresistanceto0ohm at4.2K[1].Thisdiscoverywasremarkableandunexpected,somuchsothathisresults couldn'tbefullyexplainedforalmosthalfacentury. Usingaspectsofvariousearlytheories,especiallyLeonCooper's1956paperonelectron pairs[25],JohnBardeen,LeonCooper,andJohnRobertScdevelopedamicroscopic theorytoexplainthisphenomenain1957,nowcalledBCSTheory[2].Thetheoryisbased aroundtheprincipleideathat,insuperconductors,itbecomesenergeticallyfavorablefor electronstopairtogether.OneelectronneartheFermisurfacecreatesavirtualphonon, aregionofpositivechargesinthelatticeofthematerial.This,inturn,attractsanother electron(withoppositespin)withtpotentialenergytoovercomeCoulombrepulsion betweenthetwo.Thebindingenergy(E)oftheseelectronsis E =(2.5) whereisthematerial-dependentsuperconductinggap[26].Scatteringevents,suchas phononscattering,electronrepulsion,latticedefects,etc.,arenotlargeenoughtoovercome thisbindingenergy,andisthusthereasonthatRdropscompletelyto0ohm. Theabruptnessofthedropoccursbecausethesuperconductinggapenergybecomes 14 greaterthanthatofthermalexcitations.WithinBCStheory,therelationshipbetween superconductingenergygapandtemperatureis =1 : 764 k B T c (2.6) where k B istheBoltzmannconstantand T c isthecriticaltemperature,belowwhichthe materialbecomesasuperconductor[26]. One(andveryimportant)aspectofCooperpairsinthevastmajorityofsuper- conductorsisthattheirquantummake-upconsistsofopposite-spin,opposite-momentum electrons.Theseelectronsarecalledspin-singletandcanbedescribedas j 0 ; 0 i = 1 p 2 ( j"#ij#"i ) : (2.7) Thegeneralformofthisnotationiswrittenas j s;m s i where s isthetotalspinquantum numberand m s isthesecondaryspinquantumnumber( s m s s ). Asidefromhavingnoresistance,bulksuperconductorsalsoexhibitperfectdiamagnetism, knownastheMeissner[27].Inthepresenceofamagneticthesurfaceofthe superconductorwillgeneratescreeningcurrents.Thiscausesthemagnetictodecay exponentiallytozerointhesuperconductingregion[26],eliminatingtheinthebulkof thesuperconductor.Therepulsionofmagneticisbutanotherexampleofthecompeting natureofsuperconductorsandmagnetism.Thescreeninglengthwillbediscussedagain inSection2.4.2. ItshouldbenotedthatBCSTheoryonlyaccountsfortraditional,s-wavesupercon- ductorswithCooperpairsinthespin-singletstate.Whiletherearemanyothertypesof 15 superconductors,theworkhereinonlyutilizess-wavesuperconductors,andthusothersare beyondthescopeofthisthesis. 2.3Proximity Despitebeingawareofthesematerialsforacenturyormore,thereisacontinuedinterestin studyingtheindividualfundamentalpropertiesofsuperconductors,ferromagnets,andnor- malmetals(N).However,whatisofgreaterconcerntothisthesisistheinterplaybetween theseverydtmaterialswhentheyareplacedincontactwitheachother,calledthe proximityOverall,itdescribestheamountofCooperpair\leakage"fromasupercon- ductortoanothermaterial,althoughthedetailsofthearehighlymaterialdependent. 2.3.1Superconductor/NormalMetal Whathappenswhenyouplaceanormalmetalincontactwithasuperconductor?Atenergies wellabovethebandgap,quasi-particlestatesexistinthesuperconductor,andsingleelectron transportcanoccur.Itisinsteadatelectronenergieswithinthesuperconductorbandgap (if e has j E j < thatuniquephysicscanbeobserved. Attheseenergies,electronsinthesuperconductorhavepairedwithspin-oppositecoun- terpartsandformedCooperpairs.Therearenostatesthatsingleelectronsfromthenormal metalcanoccupyinthesuperconductor,andsochargesof e cannotenterthesuperconductor (asshowninthebandstructureofFigure2.2).Thisdoesn't,however,preventtransportto occurinthesesystems.Instead,electronsinthenormalmetalwillenterthesuperconductor andoutahole,aprocessknownasAndreev[28](Figure2.2a),the ofwhichisthatachargeof 2 e entersthesuperconductor.Thisisequivalenttosayingtwo 16 (a)Andreev (b)InverseAndreev (c)CooperpairleakagefromS intoN. Figure2.2 Cartoonofpairleakagemodels. EnergybanddiagramsneartheFermienergyfor NandS(leftandrightofeachrespectively).Therearenosinglestatesforelectrons inNtoenterwithinthesuperconductinggapinS,andalltransportbetweenthetwomust exchangeachargeof 2 e .ThiscanbemodeledaselectronfromNenteringS, ahole(a)andcreatingaCooperpair.ThisprocessisknownastheAndreevThe inverseAndreevmodelsaCooperpairinSenteringNasaholeincidentonthe interface,asingleelectron(b).ThiscanalsobemodeledasaCooperpair leakingintothenormalmetal,transferringbothelectronsintoN(c). 17 electrons,oraCooperpair,enterthesuperconductor.Thisprocesswillthereforeexchange 2 e foreverytransportevent. Theinverseprocessconsistsofaholeincidentontheinterface,anelectron,and thusaCooperpairleavingthesuperconductor(Figure2.2b).Inthiscase,onecanalsolook attheprocessasCooperpairsleakingintothenormalmetal(Figure2.2c).Thecharacteristic decaylengthscale,orcoherencelength( ˘ ),describeshowfarthetwoelectronspropagate intoNbeforetheylosephasecoherencewitheachother.Thetimespanofthecoherence isas ˝ = ~ ,where ˇ 2 ˇk B T afteraveragingoverthethermaldistribution.If themeanfreepathismuchlongerthanthethicknessofthematerialthroughwhichthe electrontransmits,theparticlewilltravelattheFermivelocity, v F ,andissaidtobeinthe clean(ballistic)limit.Ifthematerialisthickerthanthemeanfreepath,likeinthesamples discussedherein,theelectronthroughthematerialviascatteringaccording tothecot D .Thislimitiscalledthedirtye)limit.Takingthese limitsintoaccount,weobtaina\normalmetalcoherencelength" ˘ N = ~ v f 2 ˇk B T (2.8) inthecleanlimitand ˘ N = s ~ D 2 ˇk B T (2.9) intheelimit[32][33]. Thenormalmetalcoherencelengthisdependentonthethermalenergyandmaterial used,butinsomesystemsatlowenoughtemperature( T ˝ 1 K ), ˘ N cangetaslongasa . 18 2.3.2Superconductor/Ferromagnet Thebiggestbetweenanormalmetalandaferromagnetisthatnormalmetals havenopreferentialspinwhilespinsarepolarizedinaferromagnet.Arigorousapproachto analyzingthestrongmagnetichaveonCooperpairswouldrequireustofollowthe workdonebyFludeandFerrell[29],andLarkinandOvchinkov[30][31],commonlyreferred toasFFLOorLOFF.Fortunatelythough,forourneeds,averyapproachcanbe takenbymodelingtheferromagneticbandstructureasanormalmetalinamagnetic TheresultantZeemansplittingisdemonstratedinFigure2.3whichplotskineticenergy( E k ) vswavevector( k )attheinterface.Inthenormalmetal,thereisonlyonebandstructure forallspins.However,intheferromagnet,thetwobandsarebytwicetheexchange energy( E ex )oftheferromagnet.Therefore,attheFermienergy(horizontalblacklinein the j k F j willbeslightlytfortheup-spinelectronrelativetothedown-spin electron. Asdiscussed,electronsinthesuperconductorareboundasCooperpairsandhaveoppo- sitespinandmomentum.Ifweconsidertheseelectronsastheypassthroughtheinterface betweenSandF,thebandstructureofFwillcausethepaircorrelationstopickupa center-of-massmomentumshift ~ Q = ~ ( k " F k # F )(2.10) whichcanbeto Q = 2 E ex ~ v F (2.11) ifweignoretheFermivelocitybetweenthespin-bands,i.e. E ex ˝ E F [34].For j"#i pairsincidentnormaltotheinterface,movinginthe x -direction,theywillbeshiftedby 19 (a) (b) Figure2.3 Interfacebandstructure. Inthecaseofanormalmetal(a),thereisnospin- polarization,andbothup-anddown-spinelectronsenterthesameband.Inthesim cartoonrepresentationofaferromagneticbandstructure(b),thebandseparationof2 E ex causestheup-anddown-spinelectronstoentertbands. 20 Q whilethe j#"i pairsareshiftedby Q ,i.e. j i F = 1 p 2 exp iQx ~ j"#i exp iQx ~ j#"i : (2.12) Ifthepairisincidentontheinterfaceatanangle ,thensincethecomponentsofmomen- tumparalleltotheinterfaceareconserved,theperpendicularcomponentmustbeshifted by Q= cos toconserveenergy[34].Averagingoverallpossibleangleswillyieldthepair correlationfunctionamplitudeandcoherencelength( ˘ F ) = sin x ˘ F x ˘ F ˘ F = ~ v F 2 E ex (2.13) inthecleanlimitand =exp x ˘ F sin x ˘ F ˘ F = r ~ D E ex (2.14) inthedirtylimit. TocomparethisresultwiththatofS/N,weneedtolookcloserat E ex .Wecanesti- mate E ex ˇ k B T C ,where T C istheCurietemperatureoftheferromagnet.Therefore,the relationshipbetweencoherencelengthscanbeapproximatedas ˘ N ˘ F ˇ E ex k B T ˇ T C T : (2.15) 21 (a) (b) Figure2.4 Paircorrelationfunctions. Ina)wecanseetheS/Npaircorrelationfunction whileb)demonstratesthesameforS/Fsystems.Boththeshortenedcoherencelengthand theoscillatorynatureareadirectresultoftheexchangeenergypresentinferromagnetsbut absentinnormalmetals. LookingatCurietemperaturesofcommonferromagnets,we T C;Fe =1043K, T C;Co = 1388Kand T C;Ni =627K[14].Taking T C ˇ 1000 K asanestimateandmeasuringat T =4K,thecoherencelengthisroughly250timessmallerinferromagnetsthaninnormal metalsinthecleanlimit.Duetothelargeexchangeenergy, ˘ F canonlyextenduptoafew nm,incontrasttothe mlengthobtainableinS/Nsystems. ThecorrelationfunctionsareplottedforbothS/NandS/FsystemsinFigure2.4.An interestingobservationisthat,inadditiontotheshortcoherencelength,weobtainan oscillatingcomponentinthecorrelationamplitudeintheS/Fcase. Because ˘ F issoshort,theexperimentsintendedtoobservetheoscillationsinthepair- correlationwavevector(Figure2.3b)werequitechallenging.Asanoftheoscillations, thesuperconductingcriticaltemperatureofanS/Fbilayerispredictedtoalsooscillateas afunctionofthethicknessoftheferromagnet.Combiningtheoffabricating thinferromagnetsampleswiththepresenceofmagneticallydeadlayersforverythinF [35],anyobservationsofoscillationsinthecriticaltemperature[36]wereto understand.Twoexperimentalbreakthroughswereachievedin2001.Byusingferromagnetic 22 alloyswithlower E ex ,andthuslonger ˘ F ,experimentalistswereabletoobserveachange insigninthestructureofthedensityofstatesinNb/Pd 1 x Ni x systems[37].Further observationsofoscillationwerereportedinsystemswithasecondsuperconductingelectrode, S/F/SJosephsonjunctions[38][39](foradiscussiononJosephsonjunctions,seethefollowing section). 2.4JosephsonJunctions Wehaveseenwhathappenswhenmaterialsareplacedincontactwithasuperconductor. What,then,happensifyouputsuperconductorsonbothsidesofanon-superconducting AstheorizedbyBrianJosephson,itisconceivablethatCooperpairswilltunnelcompletely throughthebarrierbetweenthesuperconductors[40][41].Thesesystemsbecameknownas Josephsonjunctions,andalthoughinitiallydemonstratedwithaninsulatingbarrier(I),the materialseparatingthesuperconductinglayerscanbeI,N,orF. TherearetwomainequationsgoverningtheowofsupercurrentinJosephsonjunctions, knownastheJosephsonequations,whichcanbedevelopedbyexaminingthesuperconduct- ingwavefunction ~r;t )= p n s ( ~r;t ) e ( ~r;t ) (2.16) where n s isthelocalCooperpairdensity(= y and isthephaseofthesuperconducting condensate.ThiswavefunctionevolvesintimeaccordingtotheSchrodinger-equationinan electromagnetic i ~ @ ~r;t ) @t = 1 2 ( ~ i r q A ( ~r;t )) 2 ~r;t )+ q ˚ ( ~r;t ~r;t ) : (2.17) 23 WealsohaveanexpressionfortheCooperpaircurrentdensity J s = q ~ 2 y r r y ) q 2 y A (2.18) where A isthevectorpotential, ˚ isthescalarpotential, istheemassofaCooper pairand q = 2 e ,theechargeofaCooperpair. Substituting2.16into2.18yields J s = q n s ~ ( r q ~ A ) : (2.19) ThecurrentthereforedependsonthedensityoftheCooperpairsaswellasthegauge- invariantphasegradient( r q ~ A ).Wecancalculatethegauge-invariantphase ( ' )betweenthesuperconductors(whichweshalllabel1and2asshowninFigure2.5)asa pathintegralthroughthejunction ' = 2 Z 1 ( r q ~ A ) d ~ l = 2 1 q ~ 2 Z 1 A d ~ l: (2.20) Withnocurrent,weshouldexpectthephaseandgradienttoalsobe0.Wealso expectthatthesystemshouldactindependentlyofphaserencesof2 ˇ .Fromthesecon- clusions,assumingtheJosephsoncouplingisweakandcombiningtheleadingmultiplicative factorsas J c (themaximumcriticalcurrentinthejunction),Josephsonobtainedthe Josephsonequation[41] J s = J c sin ': (2.21) 24 Figure2.5 AsimpleJosephsonjunction. S 1 andS 2 arethetwosuperconductingelectrodes, labeledforclarityforthederivationoftheJosephsonequations,andareseparatedbya barrierlayer. AnotherderivationofEqn2.21hasbeendonebyFeynman[42]. Ifweassume n s tobeconstantintime,wecansubstitute2.16into2.17toget ~ @ @t = 2 n 2 s q 2 J 2 s + q ˚: (2.22) Insertingthisintothetimeevolutionof ' ,we @' @t = @ 2 @t @ 1 @t q ~ @ @t 2 Z 1 A d ~ l = q ~ ( ˚ 1 ˚ 2 ) q ~ @ @t 2 Z 1 A d ~ l = q ~ 2 Z 1 ( ˚ @ A @t ) d ~ l: (2.23) 25 Figure2.6 Schematicofaresistivelyandcapacitivelyshuntedjunctionmodel. Recognizingtheintegrandas E = ˚ @ A @t (2.24) wecanintegratetothesecondJosephsonequation,yieldingthevoltage-phaserelation @' @t = 2 ˇ 0 V (2.25) where 0 = h 2 e ,themagneticquantum.For j J j J c ,V 6 =0. 2.4.1RCSJModel Asstatedintheprevioussection,for j I j I c 0 theparticlewillfreelyslidealongthecurveastherearenostableequilibrium points. Itiscommontoalsoseeequation2.27writtenas I I c 0 =sin ' + 1 Q! p @' @t + 1 ! 2 p @ 2 ' @t 2 (2.33) wherewehaveintroducedboththeplasmafrequency ! p = 2 ˇI c 0 0 C 1 = 2 (2.34) andqualityfactor Q = ! p RC: (2.35) Q canalsobewrittenas Q = 1 = 2 c ,where c istheStewart-McCumberdampingparameter [26]. Foroverdampedjunctions,theregimeinwhichtheJosephsonjunctionsinthiswork exist, c ˝ 1.Withsmall C ,wecanignoreitscontributiontoequation2.27 I = I c 0 sin ' + 0 2 ˇR @' @t (2.36) 28 Figure2.7 Tiltedwashboardmodel. Aparticleistrappedintheenergywellwhen I =0 (blackcurve)or II c therearenostablestates andtheparticlewillslidealongthewellfreely(greencurve). 29 whichwecanrewriteas d' dt = 2 ˇI c 0 R 0 I I c 0 sin ' : (2.37) For I>I c 0 , d' dt isalwayspositive,meaningthephaseisconstantlychanging.Thesizeof thisisperiodicwithsin ' ,windingslowerifsin ' ispositiveandfasterifitisnegative [26].AccordingtoEqn2.25,non-constant ' correspondstoaninstantaneousvoltageinthe junction.IfweintegrateEqn2.37,wecanthetime-averagedvoltageas j V j = R N Re [( I 2 I 2 c 0 ) 1 = 2 ](2.38) [26].Here R N isthenormal-stateresistanceofthejunction.Ifwelookqualitativelyat thisequation,weseethatitmatcheswiththepreviouslydiscussedphenomenaofJosephson junctions.Thatis,when j I j 1),thesystem exhibitssignsofhysteresis.Thatistosaythecurrentatwhichthesystemjumpsfrom superconductingregimetothatofthenormalstate( I c 0 )islargerthanthatatwhichit 30 Figure2.8 CharacteristicI-VcurveforanoverdampedJosephsonjunction. Theredcurve indicatestheOhmicresponseforthesame R N . goesfromnormaltosuperconducting,calledtheretrappingcurrent( I r 0 ˇ 4 I c 0 =ˇQ )[26]. Intheextremecase( c ˛ 1),thecurrentcanbeincreasedfrom I =0untilitnolonger superconducts( I>I c 0 ),buttoreturntoitssuperconductingstate,thecurrenthastobe decreasedentirelyto I =0. 2.4.2FraunhoferPattern Datafromthisworkisprimarilyconcernedwiththemaximumcriticalcurrentinoursample. AsitwillbediscussedinChapter5,ourmeasurementsystemisquiteadeptatmeasuringI-V curves.However,becausethereareferromagneticlayerswithinourjunctions,itisnecessary toconsiderhowJosephsonjunctionsrespondtoamagnetic LetusconsideraJosephsonjunctionwithcirculargeometry(radiusR),suchasthe oneshowninFigure2.9andgeometricallysimilartothosethatweremeasuredforthis thesis.Thematerialofthenon-superconductingbetweenthetwosuperconductorsisnot 31 Figure2.9 RepresentationofacircularJosephsonjunction. Thedashedblacklinesrepresent theextentoftheLondonpenetrationdepthwhenthejunctionisinthepresenceofamagnetic relevant,butlet'sconsideritasanormalmetal(N)inthisexamplewithathickness d .Inthe presenceofanexternalmagneticwithdensityandvectorpotential ~ B ex =(0 ;B y ; 0) and A =(0 ; 0 ; B y x ),respectively,thesuperconductorwillgeneratescreeningcurrentsto suppress ~ B ex ,resultinginaninternal B ( z )= B y exp z L (2.39) where L = r m e 0 n s e 2 (2.40) istheLondonpenetrationdepth, m e and e arethemassandchargeofanelectron,respec- tively, 0 isthemagneticpermeabilityconstant,and n s isthedensityofCooperpairs[26]. TheinsideNcanbeconsideredtobeconstant B N ( z )= B y : (2.41) 32 Lookingbackatthegauge-invariantphase ' = 2 1 2 ˇ 0 2 Z 1 A d ~ l (2.42) wecanthephaseacrossthejunctionasafunctionofpositionxas ' ( x )= ' 0 + 2 ˇ 0 B y x (2 L + d )(2.43) withasupercurrentdensity J ( x;y )= J c ( x;y )sin 2 ˇ 0 B y x (2 L + d )+ ' 0 : (2.44) Inordertothecriticalcurrent,wemustintegrateovertheentirejunction, I = RR J ( x;y ) dxdy .Forthecirculargeometrywehavedescribed, I =2 J c R Z R p R 2 x 2 sin 2 ˇ 0 B y x (2 L + d )+ ' 0 dx (2.45) afterintegratingovery.MaximizingEqn.2.45withrespectto ' 0 andintegratingoverx, weobtain I c =2 I c ( B =0) J 1 ( ˇ 0 ) ˇ 0 (2.46) aftercompletingtheintegration,where I c ( B =0)= ˇJ c R 2 , J 1 istheBesselfunctionofst kindandorder,and= B y 2 R (2 L + d ).ThisresultisknownasanAirypatternanditis plottedinFigure2.10.Thisresultisforthecaseofacircularjunction,buttraditionallythis iscalculatedforarectangularjunction,forwhichtheresultiscalledaFraunhoferpattern. 33 Figure2.10 Airypattern. Throughoutthisthesis,theterm\Fraunhoferpattern"isusedinplaceof\Airypattern" duetocolloquialism,butIwillnoteherethat,technically,thesearetfunctions. 2.5Spin-TripletPairCorrelations AswesawinSection2.3.2,thecoherencelengthinS/Fsystemsisverysmall,andthusany proximityshouldbeveryshortranged.Why,then,had3tgroupsreported evidenceoflongrangeproximity[44][45][46]ordersofmagnitudelargerthanexpected inS/Fsystems?Whiletheobservedphenomenacouldbeexplainedinpartbysurfaceand interfacethequestionremainedlargelyunanswereduntil2001whentwotheoretical groupsindependentlyproposedasolution[47][48]. Electronsinspin-singletCooperpairshaveoppositespinandwillexperiencethe ofthebandsplittingwhenenteringtheferromagnet,resultinginashortcoherencelength. If,however,thetwoelectronsintheferromagnethavethesamespin,theywouldobserveno bandsplittingandnofrom E ex .Tothem,theferromagnetwouldseemasifitwerea normalmetalsystem,andthusthecoherencelengthwouldincreasedramatically. 34 Thesurprisingpredictionof2001isthat,inthepresenceofmagneticinhomogeneityinan S/Fsystem,itispossibletogeneratelong-rangespin-triplet(m s = 1)paircorrelationsfrom thespin-singlet(m s =0)Cooperpairsinsidethesuperconductor.Forclarity,thestandard notationforthesestatesarebelow,intheform j s;m s i . j 0 ; 0 i = 1 p 2 ( j"#ij#"i ) j 1 ; 1 i = j""i j 1 ; 0 i = 1 p 2 ( j"#i + j#"i ) j 1 ; 1 i = j##i (2.47) Itispossibletoobservethisinsampleswithintrinsicinhomogeneity,suchasthrough domainwalls[49]orspiralmagnetization[50].However,controllingtheinhomogeneity extrinsically[51]hasprovenfarmorereproducible,andasitisthemethodusedinthiswork, itwillbethefocusofthisdiscussion.Itcanbeachievedfromwell-engineeredstructures withmultiple,non-collinearferromagneticlayers( F 1 ;F 2 ;::: ). Theformalapproachtodemonstratetheabilitytocreate j 1 ; 1 i or j 1 ; 1 i tripletcom- ponentsisquitecomplexandheavilyreliantonGreen'sfunctions.Thankfully,Matthias Eschrigdevelopedasimpler,albeitless-rigorous,explanationwhichwillbefollowedhere [52][53].(Forreference,Figure2.11depictsthesystemutilizedinthisdiscussion.)Inthis example,thesecondferromagnethasitsmagnetizationdirectionrotated relativetothe Asstatedpreviously,Cooperpairsinsidethesuperconductorarespin-singlet, j 0 ; 0 i . WhenaCooperpairpassesfromStoF 1 ,duetotheexchangeenergyoftheferromag- net,thepaircorrelationwillpickupacenterofmassmomentumshift, Q = k " k # and 35 Figure2.11 Structureforspin-tripletgeneration. Theisonlyrealizedfor 6 = nˇ . evolveto j i 1 = 1 p 2 exp iQx ~ j" ; #i exp iQx ~ j# ; "i =cos Qx j 0 ; 0 i F 1 + i sin Qx j 1 ; 0 i F 1 (2.48) whichincludesbothspin-singletaswellasashort-rangespin-tripletpaircorrelations. Asthisensembleentersthesecondferromagnetthespin-singletcorrelationswillevolve inthesamewayasabove,yieldingnolong-rangecomponents.Therefore,wewillignore theircontributionfromhereout.However,assumingthatF 1 isthinenoughthattheshort- rangespin-tripletpaircorrelationshaven'tfullydecayed,the j 1 ; 0 i componentwillundergo atraditionalbasisrotationattheF 2 interface,resultingin j 1 ; 0 i F 1 = sin p 2 j 1 ; 1 i F 2 +cos j 1 ; 0 i F 2 sin p 2 j 1 ; 1 i F 2 : (2.49) Havinggeneratedcomponentsofparallelspin,paircorrelationswith j m j =1represent 36 (a)0-junction (b) ˇ -junction Figure2.12 Tripletjunctionphasecartoon. Themagnetizationofeachlayerisdepictedby conventionalrepresentations.Therelativerotationofthemagnetizationdirectionbetween adjacentmagneticlayerswilldeterminethephaseofthejunction.A0-junctionwillmaintain rotationdirectionwhilea ˇ -junctionwillswitchrotationdirections. twoelectronsinthesamespinbandinF 2 .Inthisway,aslongastheymaintaintheirspin, thesepairsactthesameinaferromagnetastheywouldinanormalmetal,increasingthe coherencelengthdramatically. Inordertodetecttheselong-rangespin-tripletcomponentsely,werelyonmea- suringtheJosephsoncharacteristics,whichrequiresustoplaceanothersuperconductorafter F 2 .However,thesuperconductorcanonlycarryspin-singletCooperpairs.Duetotherapid decayofboththe j 0 ; 0 i andthe j 1 ; 0 i components,therewillbenomorepaircorrelationsthat canrotatebacktospin-singletattheF 2 /Sinterface.Therefore,beforecappingthejunction withthealsuperconductor,onemoreferromagnetlayer(F 3 )mustbeadded.This ferromagneticlayerwillprovideanotherbasisrotation,onceagaingeneratingsome j 1 ; 0 i componentsthatcanevolvebacktospin-singletCooperpairsattheSinterface. 37 Spin-tripletJosephsonjunctionsamples,therefore,haveanS/F 1 /F 2 /F 3 /Sstructure(of- tenwrittenhereinasS/F'/F/F"/S),aspredictedbytheoristsin2007[54].Basedonthe chiralityofrotationofmagnetizationdirectionofadjacentmagneticlayerswithin,these junctionsaresaidtohaveeithera0or ˇ spin-tripletphase.Ifthedirectionofrotation betweenF'andFisthesameasthatforFandF",thesystemisina0-state.Iftherotation changesdirection,thesystemisina ˇ -state.ThisisdemonstratedinFigure2.12. Inadditiontothephase,theabilitytoturnonorthespin-tripletgenerationmech- anismalsoexistsasafunctionofmagnetizationdirection.IfanytwoneighboringFlayers havecollinearmagnetization,thelongrangeterms,withanamplitudeproportionaltosin , arenotgeneratedandonlythestandardspin-singletdecaycanbeobserved.However,ifall neighboringlayershavenon-collinearmagnetization,itistheoreticallypossibletorealizea criticalcurrentenhancementofseveralordersofmagnituderelativetothatfoundintradi- tionalS/F/Sjunctionswithcomparablethicknesses.AsshowninEqn2.49,the m s = 1 paircorrelationsaremaximallygeneratedwithorthogonalmagnetization,minimallywith parallelmagnetization.Therefore,throughmanipulationoftherelativemagnetizationangle betweenneighboringferromagneticlayers,itshouldbepossibletocontrolthesizeofthe criticalcurrentmeasuredinS/F 1 /F 2 /F 3 /SJosephsonjunctions,whichfollows I c / sin sin ' (2.50) where istheanglebetweenF 1 andFand ' istheanglebetweenFandF 2 .Itisthis realization,theabilitytoturnthecriticalcurrent\on"andbymanipulatingthemag- netizationdirectionofindividualferromagneticlayers,thatmotivatestheworkundertaken inthisdissertation. 38 Chapter3 RelevantPreviousWork Therehasbeenalotofexperimentalworkdoneinthespin-tripletsupercurrentsince itsdiscovery,andtoignorethatwouldbelieitsimportanceherein.Theworkthatisdirectly relevanttothisthesiswillbediscussedinthischapter. 3.1InitialEvidenceofSpin-TripletPairCorrelations inS/FSystems Asmentionedinthepreviouschapter,inthelate1990s,threetgroupsreportedlarge decreasesoftheresistanceinferromagneticwiresattachedtoasuperconductingelectrode [44][45][46].Thesizeoftheproximityineachpublicationwasmuchlargerthanone wouldexpectinS/Fsystems,andthegroupsclaimedtheseresultsasevidenceoflong-range proximitywhichhadnotheoreticalunderstandingatthetime. Afterthetheoreticalpredictionsof2001,severalgroupsreportedevidenceofsupercurrent inwideferromagneticbarrierJosephsonjunctionsin2006.OnegroupmadeJosephson junctionsonCrO 2 [55].Beingahalf-metallicferromagnet,CrO 2 iscompletelyspin-polarized. Therefore,nospin-singletsupercurrentshouldbeobservable,andanysupercurrentmust arisefromlong-rangespin-tripletpaircorrelations.Asecondgroupfabricated\Andreev interferometers"withHoasaferromagneticwire[56].Theymeasuredtheresistanceof 39 theHoasafunctionofsuperconductingphaseoftheelectrodesandobserved modulations,claimingthiswasaresultoflong-rangespin-tripletcorrelationsintheHo.Ho isaconicalferromagnet,sothenecessarymagneticinhomogeneityoccursinternallyasthe magnetizationspiralsinthedirectionofelectronpropagation. Despitetheseresults,theacceptanceoflong-rangespin-tripletcorrelationsinS/Fsystems hadyettobefullyappreciated.Intheproximitymeasurementsofthe90s,toomany parameterswerestillunaccountedfor,includingthechangeofresistanceattheS/Finterface. IntheS/F/Sexperiments,neithergroupcouldcontroltheinhomogeneitypresentintheir system,thesourceofthespin-tripletcomponentspropagatingthroughtheirjunctions.This wasalargeconcern,especiallyforthegroupworkingwithCrO 2 ,whosawsample-to-sample variationsincriticalcurrentofuptotwoordersofmagnitude. 3.2ReproducibleGenerationoftheSpin-TripletSu- percurrent Priorto2010,skepticswithinthecommunityremainedconcernedbythelackofreproducible andconcreteevidencesupportinglong-rangepaircorrelations.Withhopesofrealizinga reliablemethodtogeneratespin-tripletcomponents,TruptiKhaireandMazinKhasawneh, membersofourgroup,grewS/F'/F/F"/SJosephsonjunctions[57][58](Figure3.1),relying ontheextrinsicmagneticinhomogeneityofmultipleferromagneticlayers[54](seeSection 2.5).(Inthisthesis,F'andF"arealsoreferredtoasthe\spin-mixer"layer.) Datafromtheseexperiments,summarizedinFigure3.2,clearlydemonstratetheemer- genceofspin-tripletsupercurrent.InthissampleswithoutanF'layerweremeasured todemonstratethetypicaldecayofthespin-singletinS/F/SJosephsonjunctions.However, 40 Figure3.1 Spin-tripletpillarschematic. F'andF"arethespin-mixerlayers,initiallyaweak ferromagneticalloy(PdNiorCuNi).Layerthicknessnottoscale.Figureadaptedfrom[57] whenfabricatedwithPdNi(4nm)asthespin-mixinglayer,thecriticalcurrentbecamemuch larger,surpassingtwoordersofmagnitudeforatotalCothicknessof20nm(Figure3.2). WorkstartedbyTruptiKhaireandcontinuedbyCarolineKloseandYixingWangdemon- stratedanenhancedsupercurrentinsimilarS/F'/F/F"/SJosephsonjunctionstotheone inFigure3.1bymagnetizingtheminlargeexternalmagneticAnythatmaybe trappedintheNbwasremovedbyraisingthetemperatureofthesampleabovethecritical temperature( T c )ofNbbeforemeasuringtheFraunhoferpattern.Thiscanbeseen inreferences[59]and[61].Inthedata,notonlyhasthesizeofthecriticalcurrentbeen enhanced,buttheFraunhoferpatternhasalsobecomemorepronounced.Insomesamples, magnetizingthejunctionbeforemeasurementdemonstrateda20-foldincreaseinthecritical current[59]. Thisenhancementisduetotheorganizeddomainstructurethatemergesaftermagneti- zation,asdemonstratedinFigure3.3.Becausethesejunctionshavelargeareas,theyhave multipledomainswithinthepillar,eachpointinginit'sowndirection(a).Thedomainscan beorganizedbyapplyingalargemagnetic(b),whichwillalignthemparalleltothe 41 Figure3.2 Plotofinitialspin-tripletemergence. Thex-axis, D Co ,representsthetotal thicknessofCointheSAF( D Co =2 d Co ).TheblackcurveiswithoutanF'layer,so measuresthetypicaldecayinanS/F/SJosephsonjunction.Theredandbluecurvesshow enhanced I c R N thatresultsfromgenerationofspin-tripletcomponents.Thebluecurvehas alsobeenmagnetizedformoreenhancement.Theblack,red,andbluepointsweretakenby MazinKhasawneh[57],TruptiKhaire[57],andYixingWang[60],respectively. 42 (a) (b) (c) (d) Figure3.3 Schematicoflayermagnetization. Initially,domainswillpointrandomlythrough- outeachlayer(b).Applyingalargeeldwillcausealldomainsto(mostly)aligntoit(c). Whenremoved,thespin-mixerdomainsmagnetizationwillremainparalleltothewhile theSAFwillspinperpendiculartoit(d).Thisorganizationofthedomainsgivesrise totheenhancementofspin-tripletcriticalcurrent. Therefore,whentheisremoved(c),thespin-mixerlayerswillremainalignedin thisdirection,buttheCo/Ru/Cowillperpendiculartothem(seeSection6.2for moreinformationonthemechanism). AsmentionedinSection2.5,S/F'/F/F"/SJosephsonjunctionswillhaveanoverallphase relatedtothechiralityofmagnetizationdirectionsbetweenadjacentferromagneticlayers. Intheunmagnetizedstate,thesysteminitializesasamixtureof0and ˇ phases.However, aftermagnetizing,thesystemfullyentersa ˇ phase,enhancingthecriticalcurrent.Beyond aligningthelayersfromarandomorientation,magnetizingthesamplealsocauseseach adjacentlayertohaveorthogonalmagnetization,theoptimalsituationforlong-rangespin- tripletgeneration(aswasdiscussedinSection2.5). 43 3.3CharacterizationofSpin-MixerLayers Itisnecessarytothoroughlycharacterizeeachferromagneticlayer,andoriginalworkdone toaccomplishthisisdiscussedinSection6.However,twoferromagneticmaterialsthatwere usedinthisproject,NiandNiFeMo,werecharacterizedbypreviousgroupmembers,and willbediscussedinthissection. 3.3.1Ni 3.3.1.1Generation WhenourgroupstartedfabricatingJosephsonjunctionsthatdemonstratedspin-tripletsu- percurrent,Niwasoneofthematerialsusedasaspin-mixer.Assuch,itsabilityto generatespin-tripletpaircorrelationshadbeenoptimizedearlyonbyCarolineKlose[59] andYixingWang[61].ThebluepointsinFigure3.2plotthecriticalcurrentmeasuredin samplesfabricatedwithNiasaspin-mixerlayer,whichdemonstratestheincreasedcriti- calcurrentwhencomparedtosamplesmadewithoutaspin-mixerlayer(blackpoints).In magnetizedsamples,thisenhancementwasmaximizedforNithicknessesabout1.0-1.5nm [62]. 3.3.1.2SwitchField AsmentionedinSection3.2,anenhancementofthecriticalcurrentinspin-triplet-super- currentsampleshasbeenobservedbymagnetizingthembeforemeasurement.Theonsetof thisrequiredmagneticwithmagnitudesof50-100mT,dependingonthickness. Sampleswiththicknessesof1.0and1.5nm,closetothe1.2nmNithicknessusedinthis 44 work,demonstratedsupercurrentenhancementstartingat ˇ 100mT. NisampleswerealsofabricatedandmeasuredinaQuantumDesignDCSQUIDMagne- tometer,ameasurementdevicewhichiscapableofmeasuringMvsHcurvessimilartothe onedemonstratedinFigure2.1.ThesedataareplottedinFigure3.4forNithicknessesof1.0 and1.5nm[62].Toobtainapproximatelythesametotalmagnetizationbetweensamples, multipleNilayersaredepositedineachsample,separatedbyCulayers.Fromthese datawecandeterminethatthecoercivefor1.2nmNi,whichwillfallbetweenthatof 1.0and1.5nmsamples,isbetween40-60mT. Additionally,themagneticrequiredtodemagnetizespin-triplet-supercurrentsam- pleshasbeenmeasured[61].Aftermagnetizingthesamplesinapositivemagnetic incrementallylargernegativewereapplied,measuringthecriticalcurrentaftereach iteration.Thedatasuggestthat1.5nmNirequires ˇ 60mTtodemagnetize. Measuredthreeseparateways,theseexperimentsdemonstratethehardnessofNiwhen usedasathinspin-mixerlayer.Allmeasurementssuggestthatittakesatleast40mT todemagnetizeNi,amuchlargerthanthattheonethatisrequiredtorotatethe magnetizationofsofterlayers. 3.3.2NiFeMo Becausemyworkreliesonrotationofmagneticlayers,softerferromagnetsthanNiorCo weresought.OneoptionforasofterferromagnetwasNiFeMo,amaterialthatgroup-member BethanyNiedzielskihadbeenworkingwithatthetime. SimilartothepaircorrelationofS/FsystemsdiscussedinSection2.3.2,S/F/SJosephson junctionswilleitherbeconsidereda0ora ˇ junction[63],anoftheoscillatingnature ofthepaircorrelationwavefunctioninS/Fsystems(Figure2.4b).Controlofjunctionphase 45 (a) d Ni =1 : 0nm (b) d Ni =1 : 5nm Figure3.4 NiDCSQUIDMagnetometerdata. Nilayerthicknesslabeledforeachplot.Layers wererepeatedtomaintainapproximatetotalmagnetizationbetweensamples. H c ˇ 60,40 mTforNithickness1.0,1.5nm,respectively.DatatakenbyCarolineKlose[62]. 46 canalsobedoneinS/F'/F/F"/Striplet-junctionsduetothechiralityofrotationofthe magnetizationlayers(seeSection2.5).Thisbi-modalitycanbequiteusefulinmemory applications[64],anddeterminingreliablecontrolofspin-singletandspin-tripletJosephson junctionswastheinitialgoalofBethany'swork. Althoughourmotivationsweret,manyoftheresultsfromherworkwereincor- poratedintheonsetofthisproject,especiallyregardingNiFeMoasaspin-mixerlayer. 3.3.2.1Generation TodeterminetheabilityofNiFeMotogeneratespin-tripletsupercurrent,sampleswere fabricatedconsistingofNb(150)/Cu(5)/Ni(1.2)/Cu(10)/Co(6)/Ru(0.75)/Co(6)/Cu(10)/ NiFeMo(x)/Cu(5)/Nb(20)/Au(15)/Nb(150)/Au(10),wherexvariedfrom0.8-2.4nm(aswell assampleswith0nmNiFeMotouseasacontrol).Thesewerelargearea,circularjunctions withdiametersmeasuringbetween3and48 .Thesesamplesunderwentthetypical Fraunhoferpatternmeasurement,anexampleofsuchapatterncanbefoundinFigure3.5. Inthiscase,Niwasleftastheotherspin-mixerasitsoptimalthicknesstogeneratetriplet supercurrenthadalreadybeenmeasured. Forthesemeasurements,thesizeofthecriticalcurrentisnotasrelevantasthe I c R N . I c scalesproportionallywithareawhile R N scalesinversely.Therefore,thisproductyieldsa characteristicparameterforaJosephsonjunctiondependentonlyonthematerialsusedand shouldaccountforsample-to-samplevariationsinthingslikeareaandresistance The I c foreachrunistakenasthemaximumvalueoftheFraunhoferpattern,while R N can bedeterminedastheslopetheI-Vcurveasymptoticallyapproachesfarfromthesupercurrent regime.Toobtainamoreaccuratevalue,thisslopeisgenerallydeterminedfromanI-Vcurve withverysmallcriticalcurrent,oftenathightosuppressthecriticalcurrentasmuch 47 aspossible. The I c R N productisplottedagainstNiFeMothicknessinFigure3.6[65].AsNiFeMo getsthicker,thesupercurrentissuppressedinthesamewayitwouldbeinatypicalS/F/S Josephsonjunction.Thereforeweseealineardecayonalog-linearplot.Forverythin samples,wealsoseeadecreasein I c R N .WecanseefromEqn2.48thattheamountof j 1 ; 0 i paircorrelationsgeneratedisproportionaltosin Qx .Therefore,thespin-mixerlayer cannotbetoothin,asthelong-rangetripletcomponentsaregeneratedfromthe j 1 ; 0 i pair correlations.ThepeakofFigure3.6isdeterminedtobetheidealthicknessofNiFeMofor generatingspin-triplet.Inthiscase,NiFeMoisoptimallyabout1.0nmthick. Figure3.5 ExampleofFraunhoferpatternforNiFeMogenerationsam- ples. Theblackandredcurvesindicatemeasurementstakenwithdif- ferentsweepdirections,asdepictedbyarrows.Samplecomposition: Nb(150)/Cu(5)/Ni(1.2)/Cu(10)/Co(6)/Ru(0.75)/Co(6)/Cu(10)/NiFeMo(0.8)/Cu(5)/ Nb(20)/Au(15)/Nb(150)/Au(10)[65]. 48 Figure3.6 I c R N vsNiFeMoThickness. Themaximum I c R N foreachsampleislabeled asapoint,andmultiplesamplesweremeasuredforeachthickness.Samplecomposition: Nb(150)/Cu(5)/Ni(1.2)/Cu(10)/Co(6)/Ru(0.75)/Co(6)/Cu(10)/NiFeMo(x)/Cu(5)/ Nb(20)/Au(15)/Nb(150)/Au(10),0.8nm x 2.4nm[65]. 3.3.2.2SwitchingField Fortheworkshewasdoing,Bethany'sprojectinvolvingNiFeMomovedawayfromspin- tripletandintospin-singletsamples,whichconsistedofJosephsonjunctionswithasingle ferromagneticlayer.Inordertobetterthedirectionofmagnetization,thesesamples weremadewithanellipticalgeometry.Theellipse,whichispatternedwithaspectratioof 2.5andtotalareaof0.5 2 ,issmallenoughtobesingledomain.Thisshouldmakethe 180-degreeswitchinmagnetizationdirectionveryabrupt,whiletheaspectratiocreatesa shapeanisotropythatthemagnetizationdirectionalongthelongaxis.Anexample oftheseellipsesandtheirmagnetizationcanbeseeninFigure3.7. Stillinterestedinthemaximumcriticalcurrent,Fraunhoferpatternsweremeasuredfor theseellipticalsingle-layersamples.AnexampleofsuchasamplecanbeseeninFigure 49 (a) (b) Figure3.7 Ellipsepatterngeometry. Duetothesmallsize,theellipticalpillarsaresingle domain,withmagnetizationpointingalongthelongaxisduetoshapeanisotropy. 3.8[66].Criticalcurrentmeasurementsatthesameinthepositiveandnegativesweep directiondonotoverlapthewholetimeduetothechangeofNiFeMomagnetizationdirec- tion.Dependingonhowthemagnetizationisaligned,thethroughthejunctionchanges direction,resultinginashiftinthecentralpeakofthepattern.Theshiftalternatesbetween positiveandnegativedependingonthedirectionofthemagnetization(blueandred curves,respectively).Wecanseethatmeasurementsofbothsweepdirectionsoverlapwhen themagnitudeoftheisgreaterthan5mT,implyingthattheNiFeMomagnetization directionisthesameforthoseThesedatainformusthattherequiredto NiFeMomagnetization180degreesisabout5mT.Lookingforward,rotationofacircu- larlypatternedNiFeMopillarshouldrequirenomorethan5mTaswell,atleastforsingle magnetic-layerjunctions. 3.4ExternalWork Whileourgrouphasreportedalotofinterestingworkinthiswearenottheonly oneslookingatspin-tripletsupercurrentandtheresearchpossibilitiestherein,norarewe theonlygrouplookingatrotationoflayers.Forexample,quiteabitofworkisbeingdone 50 Figure3.8 NiFeMoSwitchingField. WhenFraunhoferpatternsoverlap,NiFeMoispointing inthesamedirectionforeachsweep.Themagnetizationpointsintheoppositedirection inregionswherethepatternsdeviate.TheswitchingthatwhichtheNiFeMo sothatbothmagnetizationspointthesamedirection,isthatwhichbringsthepatterns backtogether.Thisoccursatabout5mTinthissingletsample.DatatakenbyBethany Niedzielski[66]. 51 investigatingtheofmagnetizationdirectiononsuperconductingcriticaltemperature. 3.4.1Spin-TripletJosephsonJunctions Shortlyaftertheinitialreportingofspin-tripletsupercurrentobservationsinS/F'/F/F"/S Josephsonjunctions(asdiscussedinSection3.2above),severalothergroupsreportedphe- nomenologicallysimilarresults.UsingHoasaspin-mixerlayer,onegroupmeasuredjunc- tionswithHo/Co/Howhichdemonstratedalargeenhancementincriticalcurrent,with littledecayastheCothicknessincreased[67].Theyalsowereabletoshowthatthisef- fectwasdependentonthethicknessofHo,aresultcausedbythespiralmagnetizationof Ho.AnothergroupusedtheHeusleralloyCu 2 MnAlinS/I/F/SJosephsonjunctionsand demonstratedanincreaseincriticalcurrentforcertainthicknesses[68].Startingat7nm Cu 2 MnAl,thecriticalcurrentdecayratechangesdramatically,almostplateauing.Thisis duetoaninhomogeneousmagnetizationattheinterfacebetweenthealloyandadjacent materials.However,withthicknessesabove10.3nm,thestrengthofthebecomestoo greatandtheinterfacesnolongermaintaininhomogeneity,causingthecriticalcurrentto fallbacktosizesreminiscentofsingletdecay. Twomoregroupsdemonstratedlong-rangesupercurrentinS/F/SJosephsonjunctions. Inoneexperiment,superconductingelectrodesweredepositedonsinglecrystalCowires viafocusedion-beam(FIB)deposition[69].Intheother,CrO 2 sampleswerefabricated onvarioussubstrates[70],similartotheworkdonein[55].Althoughthesematerialsand geometrieshavenocharacteristicmagneticinhomogeneity,thefabricationprocedurefor eachrequiresratheraggressiveetchingtechniques,especiallyFIB.Itispresumedthat,while creatingacleaninterfacebetweenmaterials,thesetechniquescanalsodamagethesurface ofthemagnet,creatinginhomogeneousmagnetizationattheinterface. 52 Alloftheseresultswereremarkableevidencethatmagneticinhomogeneity,regardlessof whetheritemergesextrinsically,intrinsically,orattheinterface,willgeneratespin-triplet paircorrelations,bolsteringthetheoreticalclaimsmadealmostadecadeprior. 3.4.2RotatingMagnetizationinPseudoSpinValves Theoretically,S/F/F'pseudospinvalveswithanSlayerthinnerthantheBCScoherence lengthcandemonstrateachangeincriticaltemperature( T c )dependentontherelativemag- netizationofthetwoferromagneticlayers[71].Astheanglebetweenthetwomagnetization directionsincreasesfrom0to90degrees,theofspin-tripletgenerationincreases, allowingformoresupercurrenttoleakintotheFlayers.Thisadditionalchannelforpair leakagedecreasesthecriticaltemperatureofthesuperconductor. Thiswasreportedexperimentallyin2012witha T c = T c (90 ) T c (0 )of 50mK[72].Severalothergroupsreportedsimilarresults[73][74],with T c = 120mK beingthelargestatthetime[75].Recently,agroupmeasured j T c j > 1K[76],byfarthe largestreportedatthetimeofwritingthisthesis. 3.4.3FieldDependentSpin-TripletAmplitudeModulation Moredirectlyrelatedtotheprojectdescribedinthisthesisisworkbeingdonebythe BlamiregroupattheUniversityofCambridge[77].Theyarealsohopingtomanipulate thespin-tripletsupercurrentbyrotatingmagneticandrecentlypublishedsomework demonstratingthatphenomena.However,itgreatlyfromtheworkdescribedinthe restofthisthesisinsomeimportantareas;oneverynotableisthattheyare onlyabletoseetheastheysweeptheexternalmagneticThismeansthat, 53 inthepresenceofnoexternaltheyarenotabletomeasurebothstates,spin-triplet supercurrentonand Samplesarefabricatedwithrectangulargeometry,usingPermalloy(Py),aNiFealloy consistingof80%Niand20%Fe,astheirspin-mixerlayerandCoastheircentralferromagnet [Nb(250)/Cu(5)/Py(1.5)/Cu(5)/Co(5.5)/Cu(5)/Py(1.5)/Cu(5)/Nb(250)](Figure3.9).The magnetizationsofCoandPywillaligncollinearlyinalargemagneticTherefore,at thosethecriticalcurrentisentirelyshortrangedandimmeasurableintheirsystem. Thekeytogeneratingspin-tripletgenerationinthesesamplesisthatthemagnetization ofthePylayersandColayerrotateattthresholds.Therefore,bysweepingfrom largepositivetonegativetherotationofthePyrelativetoCowillcreateanon- collinearity,generatingspin-tripletsupercurrent.Becauseofthistheyareabletodetectan increaseincriticalcurrentatcertainvalues.Tovalidatetheirclaims,theywereable tomimictheirdata(Figure3.10a)throughtheuseofmicromagneticsimulations(Figure 3.10b).Theyalsosimulatedthemagnetizationofeachmagneticlayerastheeldwasswept, whichisdemonstratedinFigure3.10c. Whilethisenhancedcriticalcurrentcanbeheldaslongastheexternalremains, anychangestothewillcontinuetorotatethemagnetization,potentiallyremovingthe non-collinearityanddiminishingtheofspin-tripletsupercurrent.Toresetthemagne- tizationsandrecoverthespin-triplet,alargeneedstoonceagainbeappliedandswept backtotheeldwherethemagnetizationsbecomemaximallyunaligned.Unfortunately,this doesnotoccurattheremnantstate,andthusthesystemisnotbi-modalunless inthepresenceofaspexternal 54 (a) (b) Figure3.9 JunctionusedbytheBlamiregroup [77].In(a),allmagnetizationdirectionsare alignedandthereforenocriticalcurrentisobserved.In(b),theF'layersrotaterelativetoF, thuscreatingaspin-tripletsupercurrenttopassthroughtheFlayer.Adaptedbypermission fromMacmillanPublishersLtd: NatureCommunications 5 ,4771,copyright2014.[Available online{http://www.nature.com/nature] (a)FraunhoferpatterntheBlamiregroup measuredasitsweptfrompositivetonega- tivefromzero,shownbythe dottedline,artifactofmeasurementsys- tem,asshowntohavezeocriticalcurrent forsome(inset). (b)Simulationsofexpectedcriticalcurrent withthissamplegeometry.Greencurveac- countsforsignofsin ˚ 1 sin ˚ 2 ( ˚ istherel- ativemagnetizationangleforneighboring magneticlayers). (c)Magneticsimulationsofindividuallayermagnetizationatvariousmomentsofthe Fraunhoferpatternsweep.(i-vindicatethemomentindicatedin(b).).Inbothiand v,theisenoughtorotateallmagnetizationsco-linear.However,intheprocessof rotating,certainmomentshavenon-colinearity. Figure3.10 DatatakenandsimulationsrunbytheBlamiregroup [77].Adaptedbyper- missionfromMacmillanPublishersLtd: NatureCommunications 5 ,4771,copyright2014. [Availableonline{http://www.nature.com/nature] 55 3.5AdditionalBackgroundInformation Whilethepreviousworkmentionedabovecoversafairamountoftheworkrelevanttothis thesis,itshouldcomeasnosurprisethatitismerelythetipoftheicebergwhenitcomes toS/Fsystems,spin-tripletpaircorrelations,andtorotatemagnetization.Ifmore informationisdesired,threerecentreviewarticleshavebeenrecentlypublishedtowhichI directthereader[78][79][11]. 56 Chapter4 SampleFabrication Toinvestigatethephenomenadiscussedinthiswork,micro-andnano-scaledevicesmust befabricatedandmeasured.Thisrequirestheuseofsophisticatedequipmentandtimeto optimizetheprocess.Inthischapter,Iwilldescribetheequipmentusedandtheresulting processdevelopedinordertofabricatesamples. 4.1FabricationEquipment Theequipmentusedtofabricateoursamplesandoptimizethefabricationprocesswillbe discussedinthissection,andwillbestructuredinroughlythesameorderasfabrication takesplace:lithographytechniqueswillbediscussedinSection4.1.1;materialdeposition andmillingtechniqueswillfollowinSection4.1.2;microscopytechniquesusedtocharacterize issuesandoptimizetheprocesswillconcludethediscussioninSection4.1.3. 4.1.1Lithography InthissectionIwilldiscussbothphoto-,oroptical,lithography(Section4.1.1.1)aswellas electronbeamlithography(EBL)(Section4.1.1.2).Whiletheequipmentisveryt, thebasicprincipleisthesame.Lithographyisaprocessthatutilizespolymers,calledresist, thatreactinoneoftwoways:positivelyornegatively.Apositiveresistisonethatwill developifexposedtolight;negativeresistdevelopsunlessitisexposed.Afterexposingonly 57 thedesiredpattern,theunwantedresistcanbedevelopedaway,leavingroomtodeposit material,asdiscussedinSection4.1.2.Thelithographyprocessisawaytomakeastencil withveryfeatures. 4.1.1.1Photolithography Photolithographyisaverycommon,t,andreliabletechniqueinsamplefabrication, usingUVlighttoreactwithaphoto-sensitiveresist,typicallyS1813.WeuseanABM MaskAlignerthatemitsUVanddeepUV(405nmand365nmlight,respectively).With thislithographer,wecanmakefeaturesdownto ˘ 3 .Featuresthanthiswillbe discussedinthefollowingsection. Toexposeasppattern,itmustbecreatedonaphotomaskwhichistypically apieceofglasswithachromeplateononeside.Thechromeplateiscutintoadesired patternthatwillallowlighttopassthrough,exposingtheresistbelowtotheUVlightand transferringfeatures.Becausetheseplatesaremadefromdurablechrome-on-glass,they reliablyreproducethesamepatternforeachexposure. Achiporwaferisplacedandalignedbelowthemask.Oftenalignmentrequiresnomore thanvisualinspectiontomakesurethesubstrateisunderthepattern.However,whenit becomesimportant(asinSection4.2.4),thealignerhasanopticalmicroscopewithupto 20xthatcanbeusedtodetermineproperalignment. Afterthesubstrateisaligned,itisbroughtintocontactwiththemasktoreducethe amountofedlightdistortingfeatures.Propercontactcanbedeterminedby seeingathininterferencepatternintheresist,visiblethroughtheopenareasofthe mask.Onlynowisthesubstratereadytobeexposed,whichistimedautomaticallyas bytheuser,about10-12secondsforthiswork.Afterexposure,thesubstratesare 58 readytoundergodevelopmenttoremoveunwantedresist,leavingonlythedesiredpattern. 4.1.1.2ElectronBeamLithography Despitetakingquiteabitmoretime,therearemultipleadvantagestousingelectronbeam lithographyinsteadofopticallithography.Firstly,linewidthsasnarrowas50nmcanbe resolvedinEBLsystems,allowingformuchpatternstobefabricated.Also,EBL patternsaredesignedinCADsoftware,meaningthat,unlikewiththeexpensiveandrigid masksusedinopticallithography,EBLpatternscanbeeasilymoandtheyoftenare. ThelithographerweuseisaJEOL840ScanningElectronMicroscope(SEM).Thewrit- ingcapabilitiesoftheJEOLwillbehighlightedhere;foramoredetaileddescriptionofthe imagingcapabilitiesofSEMs,seeSection4.1.3.1.UnliketheUVsensitivepolymersmen- tionedintheprevioussection,theresistsinEBLreacttoelectrons.Theseelectronsare generatedwithatungstenhot-cathodetandacceleratedthrough35kV,resultingin abeamresolutionofabout8nm.Towriteourpatterns,anexternalprogramcalledNanome- terPatternGenerationSystem(NPGS,colloquiallyknownas\Nabity")controlstheraster ofthebeamanddirectsitaccordingtothepattern.Givenadesireddoseand beamcurrent,NPGSdeterminesthedwelltimeofthebeamoneachpoint,typicallyonthe orderofmicroseconds.Typicaldosesareabout2-4nC/cmforlinesand200-800 C/cm 2 for areas. Itshouldbenotedherethatdosingisacrucialstepandneedstobethoroughlytestedin advance.Theareaisn'texactlythesamesizeasthebeam,asbackscatteringthe substratecreatesaslightteardropshapeintheresist.Theamountofspreadisafunctionof thebeamcurrent,butshouldalwaysbeconsideredwhenadjustingpatterns.Becausethere isabitofproximitydosingthatoccurs,nearbypatternsmayeachotherandalterthe 59 necessarydosestoobtainthedesiredpattern. Alignmentisgenerallyanotherconcernwhenwritingfeatures,soitisworthmen- tioninghere.Patternsaretypicallywrittenat1000xbutthedoseobtained bytherasterrequiredtoimagethesampleatthisisintenseenoughtoexpose theresist.Therefore,alignmentmarksaregenerallydepositedalongwithpreviousfeatures. Twosetsofthesearewrittenaroundtheareaofinterest,onesetwithintheofviewat 200xandonesetwithintheofviewat1000x.Imagingat200xistypicallynotintense enoughtoexposeresist,sopositioningthesealignmentmarksonthescreenwillnot dosetimes.(Isaytypicallybecause,ifleftlongenough,evenlowintensitycangiveenough dosetochangenecessarydoses,butallowabletimesatthistionarelongenough thatnoissueshouldarise.) Withthesamplepositionedat200x,theuserblanksthebeamitawayfrom thesample)andgivescontrolofthebeamtoNabity.Afteraligningat200xandwiththe beamstillblanked,thecanbeturnedupto1000xtomorepreciselyalignthe samplewhilekeepingthebeamfromsensitiveareas.Thisalsoensuresalignmentisdoneat thesameatwhichthepatternisbeingwritten.Withalignmentcomplete,the userpatterncannowbewrittenontothesample.Likeintheprevioussection,after thewritingisdone,substratescanbeplacedintodevelopertoremoveallbutthedesired pattern. 4.1.2DepositionandMilling Iflithographywasdonecorrectly,oneshouldhaveasubstratethathasadesiredpatternfor materialdepositionbytheremainingresist.Wetypicallyusetwotypesofadditive deposition,thermalevaporation(Section4.1.2.1)andsputtering(Section4.1.2.2),andone 60 typeofsubtractivepatterning,ionmilling(Section4.1.2.3).Themillingremovesunwanted materialanddeimportantfeatures,andcanbethoughtofasnegativedeposition. Independentofthedepositiontechnique,thesamplesgothroughaprocessto dissolvetheremainingresistand,withit,removetheunwantedsputteredmaterial.Theresist usedwilldeterminewhichchemicalandprocessisappropriatefordevelopment.Sp detailscanbefoundinSection4.2andthecorrespondingsubsections. TocontinuetheartistandstencilanalogyfromSection4.1.1,thedepositionstepwould beapplyingthepaintonthestencil,andwouldberemovingthestenciltoleavebehind justthepaintedwork. 4.1.2.1ThermalEvaporation Ifheatedtotlyhightemperature,metalswillmeltandradiateatoms.Thesewill propelsphericallyawayfromthesource,andeventuallydepositonwhatevermaterialthey hit.Thisisthefundamentalprincipleofthermalevaporation,andthemoreenergyapplied tothemetal,themorematerialwillevaporatefromit. Likemostthermalevaporators,ourEdwardsAuto306Turboappliesheatbypushing acurrent(usuallyaboutafewamps)throughhighlyresistiveboats,typicallytungstenor molybdenum.Withthedesiredmaterialfordepositionsittinginandmakingthermalcontact withtheseboats,thecurrentintheresistiveboatswillgenerateenoughheattomeltand evaporatethemetal. Asmentioned,thematerialwillradiatesphericallyawayfromtheboat.However,with tdistancefromtheboat,momentaofthematerialbecomesmostlyparallel,and depositioncanbeassumedtobecollimated.Toachievethis,thesubstrateismountedtens ofcentimetersawayfromtheboatssuchthatthematerialisincidentonthesubstratesurface 61 (90degreesbetweentheplaneofthesubstrateandthemomentaofthematerial).Thisangle, withthenaturalcollimationofthedeposition,yieldsverysharpedgesofthematerialonthe substrate.However,attimesataper(usefulifdepositingasecondlayerinthefuture)or angleevaporation(usefultodepositparticularoverlapregions)aredesired.Toachievethis, thesamplesurfacecanbetilteduptoabout60degreesaswellasrotatedaboutit'splanar axis. Ithasbeenmentionedthattheevaporatedmaterialiscollimatedoverseveraltensof centimeters,butthisassumesthemeanfreepathislargeenough.Atlowenoughpressures, ormoreimportantlyinthe10 5 torrregimeatwhichtheevaporatorisoperated,themean freepathoftheevaporatedmoleculesisontheorderofseveralmeters.Thesepressuresare achievedbyacombinationofmechanicalandturbopump;theformerusedasaroughing pump,bringingthepressuredownfromatmosphericpressurewhilethelatterbringsitlow enoughtostartevaporation.ThistransitionisseamlessbecausetheAuto306switchesfrom roughingtoturboautomatically. Pumpingdowntolowenoughpressurescantakeseveralhours,butifonedesiresspeeding theprocessup,theevaporatorhasaMeissnertrap,orcold-trap,usedtofreezeoutwater inthesystem.Whenbroughttoliquidnitrogentemperatures,watermoleculesthatmake contacttothetrapwillsticktothetrap,elyremovingthemfromtheenvironment. Usingthistrapcansavehoursfortheuser,elydroppingthepumpingtimebya factorof2.However,becausethewaterisjustfrozenoutandnotremovedfromthesystem completely,thecold-trapneedstoremaincoldforthedurationoftheevaporation. Todeterminethethicknessonthesample,athicknessmonitor(FTM)ismounted inthechamber.Athicknessmonitorisacrystaloscillatorthatrespondstomass loading.Thatis,asmorematerialisdepositedonthecrystal,itsthicknessincreases.Being 62 asensitivepiezoelectric,thiscausesthefrequencyofoscillationsinthecrystaltodecrease. Bymeasuringthischange,andwithaknownsurfaceareaandmaterialdensity,thethickness depositedonthemonitorcanbedetermined.TheFTMisfromtheverticaltoavoid shadowingthesubstrateholder,butbecausetheevaporationisnearlyspherical,this positiondoesnotthethicknessreadingmuch.However,becausematerialdensityis distancedependent( ˆ 1 R 2 ),therearestillcalibrationconcerns.Toensurecalibrationis accurate,acanbegrowninthesystemandmeasuredinamicroscope,suchas anatomicforcemicroscope(seeSection4.1.3.2).WhencomparingthisresulttotheFTM, thecalibrationratiocanbeed. 4.1.2.2Sputtering Whiletherearemanyadvantagestothermalevaporation,forthisworkthattechniqueis primarilyusedtodepositSiOxintheionmillchamber,aswellastotestdosesandpatterns inthethermalevaporator.Ferromagnetsarestrictlyforbiddenintheevaporatorlestit becomecontaminated.Therefore,weneedanotherdepositiontechnique,forwhichweturn tosputtering.Whereevaporationutilizesheattothermallyvaporizemetal,sputteringrelies onatomsfromametaltargetbeingreleasedviacollisionswithionsinaplasma. Argongasisbledintothesystemneartheguns,theareaswherematerialissputtered from.Eachguncanhaveitsowntarget,thematerialdesiredfordeposition.Theargon moleculesthatarepumpedininteractwithelectricandmagneticintersectingabove thetarget,creatingaplasmaofpositivelychargedArgonions(Ar + )andelectrons.TheAr + ionsarethenattractedtothetargetbyapplyinganegativebiasvoltagetothetarget.The ensuingcollisionisenergeticenoughtoknockatomsofthetargetfree,whichspreadaway fromtheguntowardsthesubstrateabove. 63 Inthesputteringsystem,therearetwotypesofguns:DCtriodemagnetronandDC magnetron.Bothfunctionasdescribedabove,butusetsourcesofelectricand magneticTheyarealsotsizes,with2.5"targetsforthetriodegunand1" targetsfortheother.Themainbetweentheseisthebeamthus theacceptabletoleranceofsamplepositionabovethegun.Forthesmallestguns,the issharpenoughthatsamplesneedtobepositionedabovethegunwithinaquarterinchof center.Itisthereforeimportanttocalibratethepositionofallsampleandguncombinations forthe16-sample,8-gunsystem(4DCtriodemagnetronguns,3DCmagnetronguns,and 1ionmill),aprocessthatwascompletedbymyselfaswellasVictorAguilar(whenanew sputteringprogramuserinterfacewascreated).OursputteringchamberisshowninFigure 4.1. Thesputteringprocessisnotacollimatedone.Becausecollisionsrequireawofgas, thepressureismuchhigherthaninanevaporationsystem;typicallylow10 3 torr.Therefore themeanfreepathofthesputteredmaterialisnotlongerthanthe ˘ 10cmbetweentarget andsample.Duetotheamountofundercutintheresistthisisnotalargeconcern forthesamplesdiscussedhere.However,thiswasamajorissuewiththeworkdiscussedin theappendixandwillbediscussedfurtherthere. BecausedepositedmaterialisgeneratedincollisionswiththeionizedAratoms,therate willbedependentonthefrequencyandenergyoftheseevents.Thethreemainwaysto increasethisrateare:increaseArgasw,resultinginmoreionizedArabovethetarget; increasetargetvoltage,attractingmoreions;increaseionizationcurrent,increasingproba- bilityofionizingArgonatoms.Ofthese,onlythelastiscompletelyfavorable.Byincreasing thewofargon,theamountofcollisionsbetweengasandsputteredmateriallowersthe meanfreepath,thusloweringtheamountofmaterialthatreachesthesample.Increasing 64 Figure4.1 Photoofsputteringchamber. Thetriodeguns,locatedinthenortheast,southeast, southwest,andnorthwestcorners,areshownattstagesoftargetinstallation.Start- ingattheNEandrotatingclockwise:unloadedgun;targetloadedinthegun;aluminum housingloadedoverthetarget;fullyloadedgunwithachimneyoverthealuminumhousing. Themagnetrongunsarelocatedattheeast,west,andsouth.Thesoutherngunisunloaded, whiletheeastandwestareloadedwithAuandCu,respectively.Theionmillisshownat thenorth,withitscapjustnorthofit.Tothefarwest,thetoolusedtoopenandclosethe shutters,thewobblestick,isshown. 65 thetargetvoltagecouldincreasetheamountofionsattracted,itcouldalsoincreasethe energyofcollisions.Themoreenergeticsputteredatomscouldaltertheconsistencyofthe deposition,resultinginalessidealnaldevice.Increasingthecurrent,though,increasesthe likelihoodofionizingAr.ThisincreasedamountofAr + increasesthefrequencyofcollisions withoutsaturatingtheenvironmentoradjustingtheenergy. ThedepositionrateforeachgunisdeterminedbyanFTMandischeckedimmediately beforethesputteringprogramisrunforeachchip.Thisworksmuchlikedescribedabovein Section4.1.2.1.Themainisthattherateisnotbeingconstantlymeasuredduring deposition,butcheckingaftereachchiptypicallyshowsconsistentratesthroughouttherun. Sincethechamberholdsupto16samples,butonlyoneisfabricatedatatime,thereisa needtoprotecttheotherchips.Topreventmaterialfrombeingerrantlysputtered,thereare twoshutters:oneforthechamber,whichissituatedbetweenthetargetsandthesamples; anotherforeachsampleholder,whicharemanuallyopenedandclosedbytheuser.This happensimmediatelybeforeandaftereachrunsothatonlythedesiredsampleisexposedto deposition.Oncethesample-holdershutterisopened,aclevercombinationofthechamber shutterplateandsamplepositionisrequiredsothatthesampleneverpassesoverexposed guns,changingthestructureandmaterialdeposited. Becauserun-to-runcontaminationwasaconcernwhendiscussingevaporation,itwould beincompletetonotmentionsomethingaboutithere.Topreventmaterialfrompast sputteringrunscontaminatingthedeposition,allexposedcommonsurfaceshavetheexcess previouslydepositedmaterialremoved.Forthesampleholders,thatinvolvessoakingparts inaNitricAcidbathfor30-40minutes,addinghacid(HF)tothesampleshutter partsifnecessary.Chimneysaroundthegunsandthechambershutterplatearewrappedin aluminumfoilbeforesputtering.Therefore,aftersputtering,thefoilisremovedandchimneys 66 re-wrappedwithacleanlayer.Thegunpartsthemselvesarededicatedtoaspmaterial, sonocontaminationarisesfromthese. 4.1.2.3IonMilling Thedepositiontechniquesmentionedabovecanbethoughtofasadditive,asmaterialis beinggrownonthesubstrate.Ionmillingisasubtractiveprocess,asmaterialismilledaway fromthesample.Althoughitisnotexactlyadepositiontechnique,Ihaveincludedithereas itisvitaltosamplefabrication;thepatternandgeometryoftheJosephsonjunctionpillars arethroughionmilling. Asinthesputteringsystem,ionmillingusesionizedargongasandacceleratesittowardsa target,scatteringmaterialfromthattargetintheensuingcollision.However,thedistinction ariseswhenconsideringwhattheionsarecollidingwith.Inasputteringsystem,thetarget isthematerialtobedepositedonthesample,collisionsbeingthemechanismtoknockthem looseandsendthemtowardsthesample.Intheionmillitisthesampleitselfwithwhich theionscollide,materialtoberemovedfromit. Therearetwoionmillsthatareusedfrequentlyinsamplefabrication:oneinthesput- teringchamberandoneinadedicatedmillingchamber.Theoneinthesputteringchamber isprimarilyusedasan insitu cleaningtechniquebeforesputteringandtheoneintheinde- pendentchamberisusedtomillandpatternsamples.Whiletheyarefunctionallythesame, spdetailsinthissectionwillfocusontheindependentchambermillunlessotherwise noted. Argongasisfedintothedischargechamber,asectionofthemillthathousesacathode andanode.Applyingavoltagebetweenthetwo,typicallybetween150and300 V,andacurrenttothecathodegeneratesanelectronbeamthatwillionizeargongas.Once 67 Figure4.2 Schematicofionmillchamber. Thethicksolidlinerepresentstheionmillhousing whilethedashedlinerepresentsthegridtheionspassthroughtoreachthesample.User controllableparameterslabeledinthediagram:V a {AcceleratingVoltage;V b {Beam Voltage;V d {DischargeVoltage;I e {Neutralizercurrent[80]. ionized,Ar + atomsareacceleratedbyapotentialappliedtotheaccelerating grid,typicallybetween50-100V.Thisgridisperforatedtoadjustthebeamand somewhatcollimateit.Tomaintaincollimationofthelike-chargedargonatomsinthe beam,aneutralizerislocatedjustabovethegridthataddselectronstothebeam. Todeterminethemillingrate,anFTMandagoldsputteringgunarelocatedinthe millingchamberaswell.Aftergettinggolddepositedonit,theFTMismovedoverthemill andtheratecanbedeterminedasanegativedeposition.Byknowingthedesiredmilldepth andrate,totalmilltimeiscalculatedbytheuser. Amajorconcernisthatmillingratesarematerialdependent.Whilewecanmeasurethe 68 millingrateforgold,everymaterialthatneedstobemilledthroughwillhaveat rate.Toaccountforthis,ratesforeachmaterialaremeasuredinadvancebymeasuring milleddepthinanatomicforcemicroscope,asdiscussedinSection4.1.3.2.Usingthisratio, thethicknessofeachlayercanbeconvertedtoanethickness,relativetogold,and timedappropriately. Alsolocatedinthemillchamberisasiliconmonoxide(SiOx)thermalevaporator.The evaporatorfunctionsasdescribedinSection4.1.2.1.ItisusedtodepositSiOxonthesample aftermillingandisolatesfuturetopleadsfromthebottomleadseverywhereexceptthrough thepatternedpillars.ThisisdescribedfurtherinSection4.2.3. 4.1.3Microscopy Tooptimizethefabricationprocess,itisimportanttobeabletoseethedetailsanddif- ferencesthatchangeasweadjustvariouslithographyordepositionparameters.Thissection willtalkabouttwomainonesweuseduringfabricationanalysisandoptimization:Scanning ElectronMicroscopy(Section4.1.3.1)andAtomicForceMicroscopy(Section4.1.3.2). 4.1.3.1ScanningElectronMicroscopy Thepillarjunctionsdiscussedinthisworkhavediametersof1 orless,sotocharacterize andoptimizethefabricationprocess,animagingresolutionfarbelowthat(andthatof opticalmicroscopes)isneeded.Toimageoursamples,weutilizeaHitachiS-4700IIScanning ElectronMicroscope(SEM)withanupperlimitof500,000x.Whilethere aremanyimagingtechniquespossiblewithmostSEMs,ourHitachidisplayssignalsfrom secondaryelectronsandwillbethemaintechniquedescribedhere. AswiththeJEOLmentionedpreviously(Section4.1.1.2),primaryelectronsaregenerated 69 fromanelectronsource.UnliketheJEOL,whichusesahot-cathodet,theHitachi usesthesharptipofaemissiongun,acold-cathodet.Ifexposedtoanelectric anelectroncloudwillformaroundthistip.Theseelectronsarethenfunneledaway throughtheWehlnetcap,anegativelychargedenclosurewithanopeningpointingtowards theimagingchamber.Thesearefurtheracceleratedtowardtheanode,apositivelycharged platewithaholeinit,throughwhich(someof)theelectronspass.Whileaccelerating voltagesof30kVarepossible,ourmachineistypicallyrunat15kV,allowingforresolutions asas1.5nm.Ontheirwaytothesample,thisbeamofelectronswillpassbyanumber ofcondenserlenses,magneticlenseswhichcoandfocusthebeam. Justbeforethebeamgetstothesample,itpassesthroughtioncoilsandanobjec- tivelens.Theformerisusedtopositionthebeamonthesamplewhilethelatterfocuses itonetime,creatingaveryandresolvedelectronbeamdirectlyontothesample. Therearealsoeightelectromagneticcoilsthatdeterminethestigmation,ameasureofcir- culardistortionofthebeam.Afterpassingthroughonenalaperture,whichremovesany unfocusedelectrons,thisfocusedandsharpenedbeamhitsandscattersonsamplesurface. Thereisenoughenergyfromthisimpingementtoscatterlow-energyelectrons(3-5eV) fromthematerialinthesample.Initiallyscatteredinalldirections,theyareattractedtoa chargedFaradaycage,partofanEverhart-Thornleydetector.Onceinsidethecage,these electronsgetacceleratedintoascintillator,thelightofwhichisinaphotomulti- plier.Duetothefocusofthebeam,thesignalfromthephotomultipliergivesanee brightnessforjustonepointonthesample,elyonepixeloftheimage.Thebeam isthereforerasteredacrossthesample,withsignalsbeingprocessedforeachpositionofthe beam(andcorrespondingpixelonthescreenfortheuser). Anadvantagetothisimagingtechnique,whichgivesrisetotheabilitytofurtherchar- 70 Figure4.3 SchematicofHitachiSEM. A)Electrongun;B)Beammonitoraperture;C)First condenserlens;D)Objectiveaperture;E)Secondcondenserlens;F)coils;G) Objectivelens;H)Detector;I)Samplestage.Stigmationcoils(notpictured)arelocated leftof(H). 71 acterizethesample,comesfromtheresultofthisscattering.Thesecondaryelectronsthat scatterfromthesampleleavevacanciesatlowerenergystates.Higherenergyelectronscan transitiontothesevacancies,emittinganx-ray,calledacharacteristicx-ray,inthepro- cess.Theenergiesofx-raysemittedthiswayareuniquetotheelementwhichemittedit,so measuringthisenergycangiveinformationonthematerialpresentandtheconcentration thereof.FortunatelytheHitachisystemusedinourcleanroomisequippedwithanenergy- dispersivex-rayspectroscopysystem(EDSorEDX)todojustthat,whichisveryusefulto determineandalloyconcentrationsinoursamples. TheS-4700IIhasanumberofotherfeatures,butthelastoneIwilldiscusshereisthat ofitstiltandrotationcontrols.Oursystemcantiltasampleupto50degreeswithholders thatcanloadsamplesin-planeoredge-ontothebeam.This,withtheabilitytorotatethe sampleholder,allowsforimagingatanyangle.Whilemostimagingisdonewithoutany tilting,itbecomesaveryusefultoolwheninvestigatingundercutortaperingon edges.LiketheEDSsystem,thisprovesmorehelpfulintheearlyofdevelopment thanwhenmakingsamplestomeasure,butwithoutitoursamplefabricationprocesswould befarlessregulated. 4.1.3.2AtomicForceMicroscopy AtomicForceMicroscopy(AFM)isatechniquetoprimarilymeasurethesurfacemorphology ofasample.Naively,atomicforcemicroscopy(AFM)canbebestdescribedastheactof draggingapointoverasurfaceandmeasuringtheverticalofthetip.This descriptionmightbeappropriateifitwereabletoconveythesensitivityofthistechnique: errorsofmeasurementareontheorderofhalfanangstrom. ForAFMslikeours,aDimension3100ScanningProbeMicroscope(SPM),this 72 point"isasharptip(about4 mdiameterforus,butinsomesystemsitcanbeatomically sharp)ontheendofacantileverarm(about100-200 mlong).Thisleverarmhasaforce constant(inN/m)whichdeterminestheamountofsurfaceforcesneededtoit.Of thesesurfaceforces,theinteractionswiththegreatestareelectrostatic,dipole-dipole, andvanderWaalsforces.Alaserthatshinesonthebackofthecantileveris intoadetectionsystemofphotodiodes,andminorchangesinintensitiescanbeinterpreted astipThroughthismechanism,veryresolution,liketheaforementioned half-Angstrom,canbeobtained. Whilethedescriptionaboveisaroughideaofhowcontactmodeworks,amoresophis- ticatedmeasuringtechnique,tappingmode,wasusedinthedevelopmentofthiswork.In thismode,insteadofkeepingthetipstaticanddraggingitoverthesurface,itisvibrated nearitsresonantfrequency(usuallyaround300kHz).Interactingwiththesurfaceatevery approach,theoscillationscanbecomedampened.Theheightofthetipisthereforeadjusted tomaintainoptimumoscillationamplitude.Asthisscansthesurfaceofasample,these heightadjustmentsarethedirectmeasureofthesurface WhileAFMsystemsareaveryewayofmeasuringasurfacetheycan alsobeusedtodetermineothersamplecharacteristics.Theonethatplayedthebiggestrole inthisworkisthatofmagneticforcemicroscopy(MFM).Whenusingthischaracterization technique,specialtipsthatarecoatedinamagneticmaterialareused.TheseMFMtipshave thesamepropertiesasAFMtipswhenitcomestothesurfacewhiletheforceconstant willagaindeterminehowmuchorhowlittleforceisneededtodthetip.However,unlike standardAFMtips,themagneticmaterialcoatingonthetipgivesrisetoanothersource ofmagneticforces.Whilethecoatingcanbesoftorhard(dependingonuser needs),attractions(repulsions)betweenthemagneticmaterialinthesampleandthetipwill 73 registeras\low"(\high")heightadjustments.WhenrunninginMFMmode,asurfacescan iscarriedoutjustlikeinAFMmode,followedbyasecondaryscanprobingthesemagnetic interactions.Thetipisliftedslightlyabovethesurface,followingthepreviouslymeasured measuringonlythemagneticonthesecondpass.Whencomparingthis tothestandardsurfacescan,thesystemcandeterminelocationofpolesinthemagnetic material,whichhasbeenusefulforthisworkwhenstudyingdomainstructuresordetermining sizeneedsforsingledomainpillars. 4.2Procedure Attheriskoftrivializingtheneededtocreatefunctioningsamples,aprocess isbelow: 1.Protectanddicewafer 2.ebaselayerwithphotolithography 3.Sputterbasemultilayerandliftresist 4.Patternsubmicronpillarswithelectronbeamlithography 5.Ionmill,evaporatesiliconmonoxide,andliftresist 6.etopleadswithphotolithography 7.Sputtertoplayersandliftresist TheentireprocessisshowninFigure4.4.Wheneverpossible,sampleswereprotected fromcontaminants,suchasdustparticles,byprocessingtheminsideacleanroomenviron- ment.Whenevernecessarytobringthesesamplesoutofthecleanroom,suchassputteringor 74 (a)Depositionofbasestack. (b)Patterningofma-Nresist. (c)MillingthroughNiFelayer. (d)SiOxdeposition. (e)ma-N (f)Topleaddeposition. Figure4.4 Cartoonsofsamplefabricationsteps. Layerthicknessesaredrawnforclarityand arenottoscale.Colorsfrombottomup:Darkblue{Sisubstrate;Lightblue{Nb;Orange {Cu;Blue{F'(typicallyNi);Lightpurple{Co;Red{Ru;Green{F"(typicallyNiFe); Yellow{Au;Black{ma-Nresist;Purple{SiOx. ionmilling,thesampleswerekeptfreefromcontaminantswithmultipleprotectivebarriers, suchassealingtheminnitrogenbagsandisolatingthemwithshutters. 75 4.2.1WafertoChip Wefabricatesampleson < 100 > p-typeBoron-dopedSiliconchipswithresistivity1-10-cm. These3"wafersmustbedicedinto1/2"x1/2"substratesonthedicingsaw.Toprotect thechipsfromsilicadustandothercontaminants,aprotectivelayerofS1813,atypical photoresist,isspuncoatontothewaferbeforedicing.Unlikeduringthelithographysteps describedinSection4.1.1,consistencyoftheS1813isunimportantsolongasresistcovers theentirewafer. Afterdicing,chipsarecleanedthoroughlybeforeuse.Todoso,weplacetheminacetone, warmedto90 Conahotplate.Thechipsarethenultrasonicatedfor5minutes,followed by5minutesofultrasonicationinisopropylalcohol(IPA)toremoveacetoneresidue.These chipsarethenrinsedinde-ionized(DI)waterbeforebeingblowndrywithdryN 2 gas.After ensuringeachchipiscleanbylookingatitintheopticalmicroscope,thesubstratesare readyforprocessing.Ifthereisanyremainingresidue,previouscleaningstepsarerepeated untilthechipisclean. 4.2.2FabricatingBaseLayer ChipsarespuncoatwithS1813at5000rpmfor50seconds,resultinginaphotoresist thicknessof1.3 m.Theyarethenbakedona110 Chotplatefor1minutetoremove solvent.Coatedchipscannowbeexposedthroughthebaseleadmask.VisualizedinFigure 4.5,thismaskpatternsthreenecessaryfeaturesofthebaselead:thebasewirethatwill makeupourJosephsonJunctionpillars,asdiscussedinthissection;alignmentmarksfor theelectron-beamlithographystep,asdiscussedinSection4.2.3;andalignmentmarksused duringthephotolithographyofthetopleads,asdiscussedinSection4.2.4.Afterensuring 76 Figure4.5 Schematicofbaseleadmask. Asidefrompatterningthebaseleadwire,optical alignmentmarks(Vernieralignmentmarksspreadaroundthechip)andEBLalignment marks(smallcrossesnearthewire)arealsodepositedforupcomingfabricationsteps. propercenteringviaopticalmicroscope,eachchipisexposedtoUVlightfor11seconds. Todevelop,thechipsaredippedinchlorobenzenefor5minutes,agitatingforthe 10and30secondsofthedip.ThisprocesshardensthesurfaceoftheS1813,making itlesssusceptibletodevelopmentandgivingthephotoresistpitsundercut.Theyare thenwaftedinMF352solutionfor45secondstodevelop,stoppingdevelopmentbywafting inDIwaterfor30secondsandblowingdrywithN 2 gas.Allsamplesaretheninspected underanopticalmicroscopetoensuredevelopmenthasbeencompleted.Ifresidualresist remainsinthepattern,thesampleisreturnedtotheMF352forafewsecondslonger,and repeateduntilthepatterndevelopmentisthoroughly 77 Nowreadyfordeposition,thesechipsareloadedintosampleholdersforthesputtering chamber.ThefullJosephsonjunctionstack,Nb(100)/Cu(5)/Ni(1.2)/Cu(10)/Co(4)/ Ru(0.75)/Co(4)/Cu(10)/NiFe(1)/Cu(5)/Nb(20)/Au(15)(frombottomtotop),issputtered inthisstep.Depositingtheentiremulti-layerinthiswayneverbreaksvacuum,ensuring thattheinterfacesbetweenlayersareascleanaspossible.However,theprocesstherefore requiresionmillingasasubtractivetechniquetothepillars.Thiswillbediscussedin thenextsection.Thetop20nmNbisenoughtosuperconduct,andiscoatedin15nmAu topreventoxidationandexposurewhenremovedfromthesputteringchamber. Attachedtothesputteringholdersisapermanentrare-earthmagnet,creatingain whichthemultilayerstackisgrown.Giventhatthepillarsarecircular,thereisnofavorable directionforthemagnetizationtopoint(seeSection2.1.1).Growingitinaisan attempttoconstrainit,atleastsomewhat.Themagnetocrystallinestructuremaydevelop ananisotropyfavoringthedirection.WhilethismightmaketheNiFepointmoreeasily inthegrowthdirection,whichisalsothe\on"direction,itsimultaneouslymakesitabit hardertopointintheperpendicular,ordirection.GiventhattheNiFewilldevelop ananisotropyregardless,justonethatwouldberandom,growinginaisworththe Atleastoneofthedirectionsisfavoredinthisway,andastheysay,\betterthe devilyouknowthanthedevilyoudon't." Tothephotoresistmask,chipsareplacedintoacetonewarmedto90 Cforat least10minutes.Whenitappearsthattheprocessiscomplete,chipsareplacedin theultrasonicatoronceagaintoensureallresidualS1813orsputteredmaterialisremoved. TheyarethenrinsedandultrasonicatedwithIPAtoremoveanyresidualacetoneandblown drywithN 2 gastoremoveallIPAresidue. 78 4.2.3andMillingJosephsonJunctionPillars Atthispoint,theverticalJosephsonjunctionrunsalongtheentirelengthofwire,sowe mustmaskonlyourdesiredpillarsandionmilleverythingelse.Becausethedimensions ofthe0.3-1.0 mdiametercircularpillarsaretoosmallforopticallithography,werelyon electronbeamlithography(EBL)topatterntheionmillmask.Todoso,wespinasingle layerofnegativee-beamresist,ma-N2401,at3000rpmfor40seconds,bakingitfor120 secondsona90 Chotplate. ThesechipsarethenplacedintheJEOL840SEMforpatterning.Alignmentbecomes crucialinthesesteps,somarksdepositedduringtheprevioussputteringrunareusedto ensurethepillarsarelocatedaccurately.Teststhatwererunpreviouslyshowthatadose of500 C/cm 2 iscientexposureforthema-N.Patternednearthepillarsaretwouseful features:apinwheelandalarge,3 mdiameterdisk.Thepinwheelhelpsdeterminethe qualityoftheSEMstigmationwhilethediskhelpsduringasdiscussedlaterinthis section.AfterEBL,chipsaredevelopedinAZ300MIFfor30secondsandrinsedinDI waterfor20secondstostopdevelopment.Theexposedma-Nshouldstillremain,asshown inFigure4.6. Thesesamplesarethenloadedintoionmillholderswitha5mmx5mmmask,exposing thecenterformilling.Overheatingtheresistmakesalmostimpossible,soasmalldrop ofpumpoilisplacedbetweenthesampleandtheheatsinktoensuregoodthermal coupling.Atrelativelyhighmillingenergy,300V,samplesaremilledhalfwaythroughthe secondCulayer(fromthetop),whichpatternsthePyaswell.ThisallowsforeasierPy switching,asdiscussedinSection6.1.2.Whileeachchiptakesabout5minutesofmilling, mainlyduetotheslowrateofmillingNb,eachchipismilledfornolongerthan2minutes 79 Figure4.6 Imageofma-Npillars. Thedotsontheverticalgoldwirearepillarsmadeof ma-Nresist.Theresistwillpreventthewireunderneathfrombeingmilled,the circularjunctions. atatime,onceagaintopreventtheresistfromoverheating. Aftermillingiscompleted,50nmofSiliconMonoxide(SiOx)isdepositedtoisolatethe bottomwirefromthefuturetopleads.AssumingnopinholesexistintheSiOx,thisforces anycurrentthroughthepillarandthegeometryofthejunction.Becausethepillars aresosmall,someSiOxmaycreepalongthesidesoftheresist,prohibitingcleanTo trytoreducethelikelihoodofissues,thesamplesareloadedintoside-millholders. Theserotatethesamplesothatthereisa3-degreeanglebetweenthemillandthesurfaceof thesubstrate.Eachsampleismilledatthisglancinganglefor2minutes,ed180degrees sothatbothsidesofthepillarareexposed,andmilledforanother2minutes. Tothema-N,thechipsarethenplacedinabeakerofPGremoverwarmedona 110 Chotplate.Afterabout10minutesinthewarmremover,thechipisrubbedvigorously withacottonswabtobreakapartanyremainingSiOxcoveringthepillarresist.Thebeaker 80 (a)Unsuccessfulofma-N. (b)Successfulofma-N. Figure4.7 Imagesofma-N Inbothimages,thetoppillarhassuccessfullylifted However,in(a),thebottompillarisstillcoveredinresist. 81 iscoveredinaluminumfoiltoraisethetemperatureofthePGremoverslightlyhigherandlet sitforanother5minutesbeforeputtingitintheultrasonicatorfora5minutes.Chips arethenrinsedinDIwater,blowndrywithN 2 gas,andinspectedunderthemicroscope. Becausethepillarsareoftentoosmalltoseeclearly,thelargerpillarpatternedontheside willshowwhetherthesampleneedsmoretimeintheornot.Ifso,theprocessis repeated,andsometimesleftinPGremoverovernight.Atthispoint,anyfurtherattempts havebeenshowntobefruitless,andanycloggedpillarsareed.Whatasuccessful lookslikeunderthemicroscopeisshowninFigure4.7b. 4.2.4FabricatingTopLeads Thesamplesarenowreadyfortopleaddeposition,sotheyareonceagainspunandbaked withaS1813monolayer,describedinSection4.2.2.Exposureanddevelopmentarethesame hereaswell,butwithanaddedemphasisonalignment.Becausealignmentforthetopleads isonceagaincrucial,alotoftimeisspentadjustingthepositionandrotationofeachchip sothattheVernieralignmentmarks(Figure4.8)arealigned. Afterdevelopment,thesamplesareplacedintheplasmaetchertocleanthesurfaceof thepillarandclearremainingresistresidue.Theetcherispreviouslycleanedbyrunning itemptyat300Wand500mTorrO 2 gasfor5minutes.Aftercleaningthesystem,thechips areexposedtoan100W,500mTorrO 2 plasmaetchfor90secondsbeforebeingloadedonce againinthesampleholdersforsputtering. Afterpumpingdownthesputteringchamber,thesamplesareionmilledatalowenergy (175V)tocleananyremainingresidueonthesurfaceofthepillar.Thisisdone insitu withthesputteringprocess,whichdepositsNb(150)/Au(20)ontopofthepillars. onceagainisthesameasdescribedinSection4.2.2.Aftertheprocess,thechipsare 82 Figure4.8 ImageofVernieralignmentmarks. completedandreadytobemeasured.Theentiresamplefabrication(tomakeeightchips, eachwithsixpillars)typicallytakesbetweentwoandthreeweeks. 83 Chapter5 Measurement Whenconductingresearchatlowtemperatures,therequiredtemperaturerangeforthe projectwilldeterminewhatrefrigerationtechniqueisrequired[81].Fortemperaturesdown to4.2K,thesimplestsolutionistodipsamplesdirectlyintoliquidhelium-4.Bypumping onit,theevaporationofhelium-4canbringtemperaturesdowntoabout1K.Ifwepump onhelium-3,temperaturesofabout0.3Kcanbeachieved.However,belowthis,useofa dilutionfridgeisrequired.Byusingamixtureof 3 Heand 4 He,experimentalistscanmeasure totemperaturesaslowas2mK. Asthetemperaturerangedecreases,thecost,bothintermsofmoneyandtime,torun theexperimentsincreasesdramatically.Theworkdiscussedintheappendixrequiredatleast asystemthatcouldachieve0.3K,perhapsevencolder.Thankfully,theworkdiscussedin thebulkofthisthesiscouldbedoneat4.2K,andthemeasurementschemeandequipment usedforthisexperimentwillbedescribedinthischapter. 5.1QuickDipperII Inordertobemeasured,thedevicesinthisworkmusthaveaccesstocurrentandvoltage leads,for4-terminalmeasurements,aswellasexternalmagneticinatleasttwoor- thogonaldirections,torotatethemagnet.Sincenoneofthesearepresentinaliquidhelium storagedewar,specialprobesneededtobedeveloped.Inthe1990s,Dr.WilliamPratt,Jr. 84 meticulouslyengineeredandcreatedtheseprobesinhouse,calledQuickDippers,whichcan bedippedintostandard60-literliquidheliumstoragedewars.Thereareseveralofthese probesweusetodaythataredesignedforvariouspurposes;themostbasicsystemhasthe abilitytoattachupto6currentand6voltageleadsonasubstrate,usefultoswitchbetween tsamplesonthesamechip,whileoneofthemostcomplexhasaseparateHe-3pot andcharcoalpumptobringsampletemperaturesdownto0.4Kandbelow. Forthiswork,QuickDipperII(QD-II)wasprimarilyusedasithashadbeenbuiltwith twoorthogonalmagneticcoils:alongitudinalcoilthatcansupplyaalongthedipper (heretoforthreferredtoasthey-orlongitudinal-axis)andatransversecoilthatsuppliesa acrossthedipper(x-ortransverse-axis). Thelongitudinalcoilhasacoilconstantof22.58(+/-0.15)mT/Awhilethetransverse coil'sis8.74(+/-0.04)mT/A,allowingforgreaterthan450and170mT,respectively, whenattachedtothe20AKepcoBipolarOperationalPowerSupply(BOPS).Considering theonlytimeduringtheexperimentthatweapplylargerthan20mTtooursampleiswhen magnetizing(doneat260mT),theseldsaremuchmorethant.Asuperconducting persistenceloopallowsforcurrenttowwithoutaconstantexternalsupply,potentially decreasingthenoiseinthesystem.Thisswitchneedstobecomeheated(drivingitnormal) tochangetheBecausethisisasensitiveswitch,themeasurementprogramsdiscussed latercontrolitautomatically,preventingusersfromdestroyingitaccidentally. Mountingasampleinthesystemisabitmorecomplexthanmostoftheotherdippers, butisnottobecomeaccustomedto.Uptoahalf-inchsamplecanbemounted bystplacingitonthebrasssampleplateandtyingitdownwithstring.Thisisdone tominimizeanymovementorshiftsofthesampleasitcoolsinthedewar.Currentand voltageleadsarethenpressedontothesuperconductingpadswithindiumsolder,asshown 85 Figure5.1 SchematicofQuickDipper-II. Electronicsformagneticcoilshavebeenomitted. 86 (a)Schematicofchipgeometry. (b)Cartoonrepresentationoflead orientationacrosspillar. Figure5.2 Top-downrepresentationsofleadgeometryacrosssample. Thecurrentisdriven fromthebaselead(red)throughtheJosephsonjunction(green)andalongthetopsuper- conductinglead(blue).Voltageismeasuredacrossthepillar. inFigures5.2and5.3b,withonecurrentandonevoltageleadoneachsideofthepillar(top andbottomleads).Althoughthetopleadgeometryallowsforit,eachwiredoesnotneedits ownpadbecausetheleadsaresuperconducting.Aslongasthewiresdon'tsharethesame solderjointornormalleads,thevoltagedropwillonlybemeasuredacrossthepillar. Oncetheleadsaresecure,thelongitudinalandtransversecoilsthathadbeenpreviously pulledbacktoallowaccesstothesampleplate(Figure5.3c)canbeslidintopositionover thesample(Figure5.3d).Guiderodsarethenunscrewedfromthebaseandreplacedwith screwsthattightenthecoilstothesampleplate.Thisisdoneoneatatimetoprevent rotationofthecoilsandtanglesintheleadwires.Atthispoint,thelongitudinalmagnetis set,butthetransversecoilsarestillloose.Thetwocoilsaredesignedtosplayout,giving themroomtoslideupanddownthedipper.However,thesemustbetiedtogetherwhen mountingtoensurethateachcoilisstaticwhencreatingamaintainingtheexpected atthesample. Afterthecoilsareinplace,alloftheslackinthemagneticleadsneedstobeprotected. 87 (a)QD-IIbeforemounting. (b)SamplemountedonQD-II (c)QD-IImagnetcoilsreadytobeslidup. (d)QD-IImagnetcoilsslidtocoverthesample. (e)Closingtheclamshelltoprotectthemagnet leads. (f)FullyclosedQD-II. Figure5.3 MountingQD-II. Thesampleismountedonthedipper(atob),thenthe magneticcoilsareslidoverthesample(ctod).Afterthetransversecoilsaretiedtogether (e),theclamshellisclosedaroundthemagneticleadsandtiedtosecureit(etof). 88 Aclamshell-likeprotectiveenclosureisplacedaroundtheseleadsandcoversallelectronics stillexposed,fromthebaseofthedippertothetopofthecoils(Figure5.3e).Thispre- ventsanyleadsfromgettingtangledaroundsomethingiftheymoveinthedewarduring dipping/removaloftheprobe.Thisclamshellisthentieddownwithalotofstring,keeping itshuttight.Lastly,theknotsaresecuredwithadabofGEVarnishtomakesuretheycan't comeuntied,makingtheshelluseless(Figure5.3f).Oncethevarnishdries,thesamplecan thenbedippedintothedewar. Alongthepathofthevoltageleads,thereareanumberofotherelementsthatare characteristictoeachdipper.Theyincludereferenceandfeedbackresistors( R ref and R FB ) andasuperconductingquantuminterferencedevice(SQUID)currentcomparatorcircuit. ThesecanbeseeninFigure5.1andwillbediscussedinthefollowingsections. 5.2SQUIDElectronics 5.2.1OverviewofSQUIDs Superconductingquantuminterferencedevices(SQUIDs)arecircuitelementsconsistingof superconductingloopswithasmallJosephsonjunctiononeachlimb,asshowninFigure 5.4.Thesesystemsareverysensitivetomagneticandthereforeveryusefultoolsfor datacollection.Theyareinnowaythefocusofthiswork,butthedatacollectedhereinare takenwithaSQUIDcomparatorcircuit,andthereforeabriefdescriptionofSQUIDphysics isuseful. AsingleJosephsonjunctionhasaperiodicityinitscriticalcurrentbasedonthrough thejunction,asdiscussedin2.4.2.Byputtingtwojunctionsinparallel,wemaintainthis 89 Figure5.4 SQUIDLoop. ThecurrentthatenterstheloopsplitsbetweenbothJosephson junction(blue)branches.Theintheloopdeterminestheinterferencebetweenthe junctionsandthusthesizeofthecriticalcurrentthroughtheloop. periodicity,butobtainasecondaswell[82].Thisadditionaloscillationisduetothe passingthroughthesuperconductingloop(Fig.5.4).Lookingatapatharoundtheloop, a ! b ! c ! d ! a ,weexpressionsthatareanalogoustoaJosephsonjunctionina magnetic[83](seeSection2.4).IfthesuperconductingisthickerthantheLondon penetrationdepth,theintegralofcurrentdensityalongtheentireloopwillvanish,andwe areleftwithonlythephasegradienttermofEqn2.19. Integratingovertheentireloop,thetotalgauge-invariantphase ' =2 ˇn . ThereforewefromEqn2.20, 2 ˇn = ' 2 ' 1 2 ˇ 0 I C A d ~ l (5.1) where ' n isthephaseinthe nth Josephsonjunction.Theintegralofthevector potentialalongtheenclosedpathyieldsthewithintheloop,andthusEqn5.1canbe 90 as ˚ 2 ˚ 1 =2 ˇn + 2 ˇ 0 : (5.2) FromthediscussioninSection2.4,weknowthecurrent-phaserelationshipis J n = J C n sin ˚ n (5.3) withtotalcurrent J = J C 1 sin ˚ 1 + J C 2 sin ˚ 2 = J C 1 sin ˚ 1 + J C 2 sin ˚ 1 + 2 ˇ 0 : (5.4) Ifwemaximizethiscurrentwithrespecttophase,andconsiderthesimplestcasewhere J C 1 = J C 2 ,wecanthesuperconductingcurrentintheloopandtheexternalmagnetic arerelatedby J =2 J C cos ˇ 0 : (5.5) ThisresultimpliesthatthesupercurrentinaSQUIDwilloscillatefromamaximumtoa minimumwithinonequantum, 0 = h 2 e =2 : 0678 ::: 10 15 Wb!WithpracticalDC SQUIDs,changesaslowas10 6 0 canbedetected[84]. 5.2.2SQUIDCurrentComparatorCircuit Withsensitivityatthislevel,it'snosurpriseSQUIDcircuitshavefoundtheirwayinto measurementsystems.Forourexperiments,weuseanRFSQUID[85].Stillaverysensitive 91 device,capableofmeasuringchangesindownto10 5 0 ,RFSQUIDsintheir measurementschemecomparedtothatofDCSQUIDsmentionedabove.RFSQUIDscouple asingleJosephsonjunctiontoan LC circuit.Anoscillatingcurrentintheinductorgenerates anoscillatingvoltageintheSQUID,whichisperiodicinappliedwithaperiod 0 . OursystemusesaQuantumDesign2010SQUIDControlthattalkstotheSQUIDin theQuickDipper,tryingtokeepthevoltageacrossthesampleandreferenceresistorequal. Asthecurrentthroughthepillarexceedsthecriticalcurrent,avoltagedropdevelopsin thesample.This,inturn,generatesacurrentintheloopcreatedbythesample( R s ),the referenceresistor( R ref ),andtheinductorbetweenthem(seeFigure5.1).Throughthe transformerthatcouplesthislooptoit,theSQUIDloopexperiencesachangeinmagnetic whichisreadintheelectronicsbox. Inanattempttocancelthisx,theSQUIDelectronicsboxoutputsavoltagewhich, afterdroppingacrossthefeedbackresistor( R FB ),dropsacross R ref .Theelectronicsbox tunesthevoltagein R ref untilitexactlymatchesthatwhichisacross R s ,whicheliminates thecurrentintheinductor.BymeasuringtheoutputvoltageoftheSQUIDelectronicsand knowingthat(forQD-II) R ref =126 and R FB =wecandeterminethevoltagein thesampleby V s = V out R ref R ref + R FB ˇ V out R ref R FB : (5.6) 92 5.3System WiththecomplexitiesoftheQuickDipperandtheSQUIDdevicecovered,therestofthe measurementsystemisrelativelystraightforwardandisdepictedinFigure5.5. Figure5.5 Schematicofmeasurementset-up. Colorcode:redwires{magneticcontrol;blue {currentcontrol;green{SQUIDoutput/voltagemeasurement. Theexternalcurrenttothesampleissuppliedbyananalogpowersupplydrivenby 12Vmotorcyclebatteries.Theseareverystableaslongastheyarecharged,whichisdone continuouslywhennotoperating.Duringoperation,thechargingaddsalittlenoise,sofor verysensitivemeasurementsitisadvantageoustodisconnectthebatteriesfromthecharging source.ThispowersupplysystemwascreatedinhousebyDanEdmunds. Asthecurrentincreasesanddrivesthesamplenormal,theSQUIDComparatorCircuit adjustsitsoutputtomatchthatacrossthesample(asdescribedinSection5.2.2).The SQUIDoutputvoltageismeasuredwithatHP34401Adigitalmultimeterwhichcanberead bythecomputer. ThecurrentissteppedwithcounterpartvoltagereadtomeasureanentireI-Vcurve foraspmagneticThisisdrivenbysupplyingacurrenttotheappropriate 93 magneticcoil(transverseorlongitudinal)viatheKepcoBOPsdiscussedpreviously(Section 5.1)andisreadbythecomputerbypassingthecurrentthroughasmallresistor(0.1or 0.01andmeasuringthevoltagedropwithaFluke45DigitalMultimeter. Whenadjustingtheasmallamountofheatisappliedtothepersistentswitch, drivingitnormalandpassingthecurrentintothemagneticcoils.Thisisturnedoncethe desiredisreached,limitingtheationsandnoisefromthemagneticcurrent withinthemeasurement. Oncecompleted,themagneticeldissteppedandanewcurveismeasured,whichisdone repetitivelyuntilanI-Vcurveistakenfortheentirerangeofdesiredexternalmagnetic Thesedatacanthenbeanalyzedandtotheappropriatefunction(asshowninEqn2.38) tothecriticalcurrentforeachcurve.Plotting I c vs 0 H yieldstheFraunhoferpattern forthesampleinitscurrentmagneticstate,whichcanthenbeadjusted(i.e.magnetization rotated)andtheprocessrepeatedtodetermineanewpattern. Allofthemeasurements,aswellascriticalcurrentextrapolationprogram,areautomated withLabVIEWprogramsdevelopedinhouse.ThesewereoriginallyprogrammedbyNate VerhanovitzandlatermobyTruptiKhaire,YixingWang,EricGingrichandmyself. Thisautomation,besidesalleviatingthemonotonyoftheroutine,isparticularlyimportant whenconsideringthepersistentswitchcontrol.Withoutappropriatecurrentmatchingbefore andafteropening,thepersistentswitchcanvaporize,renderingthedipperlittlemorethan anelongatedpaperweightuntilitisrepaired. 94 Chapter6 Characterization Whileevidenceofspin-tripletswitchingshouldbeclearinthedata,wemustbecarefulas othertscouldgiveasimilarsignature.Tobesurethatthecauseofthesupercurrent modulationsareduetotherotationofanindividuallayer'smagnetization,itisimportant tocharacterizethepropertiesofthevariousmaterials. 6.1UseofNiFe ThestsamplesmadeforthisexperimenthadNiFeMoasthesoftferromagneticlayer, characterizationofwhichwasdiscussedinSection3.3.2.AddressedinSection7.3.3,these sampleshadNiFeMoasbothspinmixerlayers.Presumably,ifthemagnetizationsofboth layersrotated,thewouldbequitelarge.However,samplesmadewiththismaterial didnotbehavequiterightfortworeasons:thenecessarytorotateNiFeMointhefull systemseemedtobeabithigherthanexpectedandanyformofswitchingseemedtobevery messy;also,thesignalwasverysmall.Atthispoint,NiFewasconsideredasaspin-mixer, butinordertojustifyitsuse,itneededtobecharacterizedastheNiFeMohadbeen. 6.1.1Generation Inordertocomparethespin-tripletgenerationcapabilitiesofNiFetoNiFeMo,thesample geometriesandstructure(materialthickness,etc)hadtomatch.Withthisinmind,samples 95 ofNb(150)/Cu(5)/Ni(1.2)/Cu(5)/Co(6)/Ru(0.75)/Co(6)/Cu(5)/NiFe(x)/Cu(5)/Nb(20)/ Au(15)/Nb(150)/Au(10)werefabricatedusingthegroup'sstandardphotolithographypat- tern,wherexvariesfrom0.8-2.4nm.Withthismask,6circularpillars(diametersof3 , 6 ,2x12 ,24 ,48 )aremilledfromawidebaselead(300 ).Whilethereare anumberofdbetweenthisstructureandthoseusedintheexperiment,the relativeamountofspin-tripletwecangenerateisnotbyanyofthese Duetofabricationissues,noteverypillarhadacceptablenormalstateresistances.How- ever,everyNiFethicknesshadatleastonepillarthatwasmeasurable,allowingforacomplete mappingofthespin-tripletgenerationforthesethicknesses. InthesamemannerthedatainSection3.3.2.1weretaken,aFraunhoferpatternwas measuredforeachNiFethickness.AnexampleofthesemeasurementsisshowninFigure 6.1.DuetothesoftnessoftheNiFeandthesizeofthepillar,webelievemuchofthenoise inthepatternisduetodomainsrotatingindependentlythroughthesweepasopposedtoall atonce.However,theexpectedwidthofthecentralpeakisabout8mT,sotheminimaof theblackcurveat1.2and 6mTlikelythecentrallobeoftheFraunhoferpattern. Themaximummeasuredcriticalcurrentistakenfromeachmeasurement,multiplied bythesample'snormalstateresistance,andplottedagainstthicknessoftheNiFe(Figure 6.2).Asexpected,the I c R N decayswithincreasingNiFethickness.Thedecreaseatlow thicknessesisduetoalackofmagnetizationforNiFethicknessesoflessthan1.0nm,similar towhatwasobservedinNiFeMosamples(seeSection3.3.2.1). Itislikelythatthevaluesobtainedfromthesedataarelessthanthemaximumpossible criticalcurrentduetothepoorqualityoftheFraunhoferpatterns.However,becausethis measurementisintendedtodeterminetheoptimumNiFethicknesstogeneratespin-triplet Cooperpairs,thisunderestimateisnotlikelytoalterthetrendandnotamajorconcern. 96 Figure6.1 ExampleofFraunhoferpatternforNiFegenerationsamples. TheredandblackcurvesrepresentFraunhoferpatternstakenwith enteepdirections,asdepictedbyarrows.Composition: Nb(150)/Cu(5)/Ni(1.2)/Cu(5)/Co(6)/Ru(0.75)/Co(6)/Cu(5)/NiFe(0.8)/Cu(5)/Nb(20)/ Au(15)/Nb(150)/Au(10). Fromthedata,wechoosea1.0nmastheoptimumNiFethickness. 6.1.2SwitchingField Withthethicknessdetermined,thenextthingtooutishowlargeanexternalis requiredtorotate1.0nmNiFe.Theeasiestwaytomeasurethisproperty,asseeninSection 3.3.2.2,isbyfabricatingellipticalpillarsandmeasuringFraunhoferpatternsinalong thelongaxisoftheellipse.Thesesampleshadthesamegeometryasthosediscussedin Section3.3.2.2aswell,sothemagnetizationwasstilluseful(seeFigure3.7). 97 Figure6.2 I c R N vsNiFeThickness. Composition:Nb(150)/Cu(5)/Ni(1.2)/Cu(5)/Co(6)/ Ru(0.75)/Co(6)/Cu(5)/NiFe(x)/Cu(5)/Nb(20)/Au(15)/Nb(150)/Au(10).0 : 8 nm x 2 : 4 nm . 98 Onemajorbetweenthesemeasurementsisthat,becausethesewerespin-triplet samples,theyhadthreemagneticlayers(here,andelsewhereinthisthesis,weareconsidering theCo/Ru/Cosyntheticantiferromagnet(SAF)tobeasinglelayer),withnon-collinearity betweenadjacentlayersgeneratingthemeasuredsupercurrent.Allmagnetizationswould pointalongthelongaxisifpatternedasanellipse,soonlythetopNiFecouldbepatterned. Thisresultedinlessclearswitchingsignatures,whichcanbeseeninthedata.However, becausetheNiFeshouldstillswitchabruptly,itsswitchingisstillevident.Thisis showninFigure6.3;thereisaratherclearchangeincriticalcurrentaround 10mT(10 mT)inthenegative(positive)sweepdirection. 6.2Co/Ru/Co AsimportantasitistodeterminetherequiredtorotateNiFe,itisequallyimportantto determinehowmuchcanbeappliedwithoutrotatingtheCo/Ru/Cosyntheticantifer- romagnet(SAF).ASAFisamultilayeroftwoferromagnetsseparatedbyathinnon-magnet, typicallyanormalmetal(Figure6.4).DuetothebandstructureofRu,thereexistsalong rangeexchangecouplingbetweenColayers,causingthemtoalignanti-parallel[86]. Inanexternalthemagneticlayerswillstarttobendinthedirectionoftheeld. Ifstrongenough,thiswillcausethemagnetizationtopointinascissor-likemannerinthe directionoftheAstheisturneddown,thescissoringbetweenlayerswillrelease andthemagnetizationwillrealignanti-paralleltoeachother.Thisprocess,knownasspin- willcausethemagnetizationofthetwoferromagneticlayerstoalignperpendicular tothedirectionoftheexternalasshowninFigure6.5. 99 Figure6.3 NiFeswitching TheabruptdropincriticalcurrentisduetotheNiFeellipse magnetizationdirection.Theblackcurveismeasuringfrom60to 60mTwhile theredcurveisopposite.Composition:Nb(150)/Cu(5)/Ni(1.2)/Cu(10)/Co(4)/Ru(0.75)/ Co(4)/Cu(10)/NiFe(1)/Cu(5)/Nb(20)/Au(15)/Nb(150)/Au(10). Figure6.4 SAFcartoon. ThecutawayshowsthetwoColayers.Rulayernotshown. 100 (a)SAFscissoringwithlargeex- ternal (b)SAFstartingto withreducedexternal (c)Fullanti-alignmentafterex- ternalremoved. Figure6.5 mechanism. InalargeenoughexternalmagneticSAFmagnetiza- tionsscissorinthedirectionoftheAsthedecreases,theSAFstartstorelease, anti-parallelwitheachotheruntiltheiscompletelygoneandthemagnetiza- tionsareanti-aligned.Theorientationofthemagnetizationsisindependentoftheir initialorientations,butratherdependsonlyonthedirectionof ~ H 6.2.1AnisotropicMagnetoresistance TomeasuretherotationoftheSAFduetoexternalwerelyonaphenomenaknownas anisotropicmagnetoresistance(AMR).Thisarisesfromtheinterplaybetweenthemagneti- zationofthematerialandthespin-orbitinteractionofthecurrent.Inshort,theresistivity ( ˆ ),andthusresistance,ofthematerialisdependentontheanglebetweenthemagnetization andcurrent( ' ), ˆ ( ' )= ˆ ? +( ˆ k ˆ ? )cos 2 ': (6.1) Theresistanceismaximumforparallelorientationandminimumforperpendicularorienta- tion. BecauseofthemechanismintheSAFs,thismeansthatanexternalpar- allel(perpendicular)tothecurrentwillresultinperpendicular(parallel)magnetization. ThisisdemonstratedinFigure6.6.Alsoshownisthemountinginastandard4-terminal measurementofasamplewiththisgeometry. 101 (a)SAFmagnetizationperpendicular relativetocurrentduetolongitudinal (y-) (b)SAFmagnetizationparallelrelative tocurrentduetotransverse(x-) Figure6.6 CartoonsofSAFmagnetizationdueto I ,V showntoindicatemounting geometryforresistancemeasurements. 6.2.2AMRProcedure MeasurementsofAMRarenomorecomplicatedthanapplyingaknowncurrentandmeasur- ingthevoltageresponseofthesample.ToseethedastheSAFrotates,amagnetic canbeappliedandremoved,measuringthechangeinoutputvoltagecausedbythe Thesystemisinitializedina260mT(thesameourtrolsamples willbeinitializedat)butonlymeasuredupto160mT,themaximumofthetransverse coil. Thecurrenttothesampleisdrivenfromtheoutputvoltageofalock-in(Stan- fordResearchSystemsModelSR830DSP)connectedtoavariableballastresistor.Bymea- suringtheoutputvoltage,theresistanceissimpletodeterminewithOhm'sLaw: R = V out I in . However,theofresistancewithparallelcurrent-magnetizationorientation( R k ) 102 (a)AMRwithincreasingperpendicular (b)AMRwithincreasingparallel Figure6.7 Resolutionissuesoflock-in Minorchangesinthesampleresistanceare maskedbythesensitivityceilingofthelock-in. andresistanceoftheperpendicularorientation( R ? )isverysmall,between0.05%and0.6%, dependingontheCothickness.Therefore,theresolutionofthelock-inis1part per10,000,or0.01%.Whiletheoveralltrendismeasurable,minorchangesintheresistance astheincreasescangounnoticed(seesteppingsignatureinFigure6.7). Toobtainbetterresolution,aratiotransformer(SingerRT-61)isconnectedtothecircuit asavoltagedivider.Theresultisasubtractionofthenormaloutputvoltagemeasuredat thesample.Byremovingmostofthesignal,thesensitivityofthelock-incan beincreased,measuringtofarbetterresolution.Thecircuitdiagramforthisset-upis demonstratedinFigure6.8. 6.2.3AMRData Nowthatthemeasurementsystemistuned,samplesofvariousCothicknessescanbemea- sured.Fabricationofthesesamplesusedamechanicalmasktothesamplegeometry. SamplesweregrownwithstructureofNb(2.5)/Cu(3)/Co(x)/Ru(0.75)/Co(x)/Cu(3)/ Nb(2.5),wherexis2,4,6,or8nm.Thecopperlayersarenecessaryinfabrication,asCo 103 Figure6.8 Circuitdiagramofaratiotransformer. AandBgototheinputterminalsofa lock-in. growssmootheronCuthanonNb.Theyare,though,keptthintolimitparallelchannels forthecurrent.Niobiumisalsokeptthintopreventitfrombecomingsuperconducting. Twosamplesofeachthicknessweremade,andonesampleofeachwasgrowninanexternal magneticwhiletheotherwasnot.Becauseallofourtrolsamplesaregrown inaitisusefultoknowthe(ifany)anexternalhas. ThedataisplottedinFigures6.9and6.10.Aisappliedineithertheperpendicular orparalleldirection,thenremoved.Inzeroresistanceismeasuredandplottedrelative totheapplied.Initially,alarge(260mTisappliedparallelto I ,initializingthe systeminthewaycontrolsamplesaremeasured(seeChapter7).Theissteppedin 10mTincrementsto160mT,stintheperpendiculardirection,followedbytheparallel direction(blackcurves).Toobtainmoreresolveddataatlowtheprocessisfollowed againwithsmallerstepsforsmall(redcurves). Fromthedata,itisevidentthatrotationoftheSAFisatrueconcern,especiallyas thethicknessincreases.ItalsoappearsthatgrowinginasoftenstheSAF(makesit easiertorotate),butthecauseofthisisnotentirelyunderstood.KeepingtheSAFthinis important,butifit'stoothin,thesuppressionofspin-tripletsupercurrentwillbetooweak. 104 (a)Co(2)withHappliedperpendiculartoI.Up- percurvesweregrowninamagnetic (b)Co(2)withHappliedparalleltoI.Upper curvesweregrowninamagnetic (c)Co(4)withHappliedperpendiculartoI. Lowercurvesweregrowninamagnetic (d)Co(4)withHappliedparalleltoI.Lower curvesweregrowninamagnetic Figure6.9 AMRdataforCo(2)andCo(4). Eachmeasurementwasundertakentwice, startingwiththeblackcurve. Fortheworkdescribedherein,aCothicknessof4nmwaschosenasagoodbalancebetween hardnessandthickness. Thereisaslightceforsomesamplesbetweentheandsecondmeasurement (blackandredcurves,respectively),butonlyintheperpendiculardirection(i.e.the timeitisrotated).Webelievethisisduetoslightchangesinthedomainstructureofthe bulksheetsofCo.Thesystemisinitializedintheparallelstatetolarge(260mT) butanydomainchangesastheSAFisrotatedupto160mTmaynotreinitialize 105 (a)Co(6)withHappliedperpendiculartoI.Up- percurvesweregrowninamagnetic (b)Co(6)withHappliedparalleltoI.Upper curvesweregrowninamagnetic (c)Co(8)withHappliedperpendiculartoI.Up- percurvesweregrowninamagnetic (d)Co(8)withHappliedparalleltoI.Upper curvesweregrowninamagnetic Figure6.10 AMRdataforCo(6)andCo(8). Eachmeasurementwasundertakentwice, startingwiththeblackcurve. 106 (a)Summaryofsamplesgrowninamagnetic (b)Summaryofsamplesgrownintheabsenceof amagnetic Figure6.11 AMRSummary. Therelativepercentchangeinresistanceinthe20mT comparedtothefull160mT.Onlythesecondrunofeachsampleisusedforcalculations. Green:H ? I;Blue:H k I. fully.However,thedataseemsreproducibleaftertherotation,evidentintheand secondparallelmeasurementsaligning. SincewecanrotatetheNiFewithinthe20mT,knowingtherelativerotationof theSAFisusefulwithinthatrange.Calculatingtheresistancechangeinthe20mT relativetothefull160mT,wecanobtainapercentchange,whichisplottedinFigure6.11. Duetotheslightofthemeasurements,asnotedabove,onlythesecondsweep valueswereusedinthesecalculations. ItisclearthatrotationoftheSAFisarealconcernwhenmeasuringthesesamples.It seemsasthoughanyevenifverylow,isenoughtomovethemagnetizationalittle bit.WhiletheissmallerforthinnerCo,andrelativelysmalliftheiskeptbelow 20mT,itappearsthatthemagnetizationwillneverbefullystable.Thismeansthatthe magnetizationdirectionsofadjacentlayersmayneverbecompletelycollinear,andtherefore thespin-tripletpaircorrelationsmayneverfullyvanish.However,iftherotationiskept small,theamplitudeofpaircorrelationsshouldalsobeverysmall,andthereforeweshould 107 beabletorealizeamplitudecontrolinS/F/SJosephsonjunctions. 108 Chapter7 ControlofSpin-TripletSupercurrent Tobetterunderstandthedatapresentedhere,thischapterwillbeginwithadiscussionofthe procedurefollowedtomeasureoursamples.Afterwards,tojumpdirectlytothesuccessful samplesanddata,whichshowtheabilitytocontrolthespin-tripletsupercurrentinS/F/S Josephsonjunctions,wouldleavemanyunansweredquestionsastohowthedecisionsleading tothosesamplesweremade.Theinitialandstumblesthatdirectlyledtooursuccesses willbediscussedinthesection.Theldatawillfollow. 7.1Procedure Becausesomuchofthefollowingdiscussionreliesonrotatingmagneticlayers,itisuseful totalkaboutthemeasurementprocedurehere.Thissection,withtheaidofcartoonrep- resentations,willhopefullyhelpthereaderbecomefamiliarwiththepotentiallyconfusing discussionaboutmagnetizationdirection. 7.1.1InitialFraunhoferPattern Thesystemisinitializedbyapplyingalargemagnetic(260mT)longitudinally alongthesample.ThisalignstheNiandNiFemagnetizationsinthedirectionoftheapplied whiletheCo/Ru/CoSAFspinsothateachlayer'smagnetizationisperpendic- ulartotheappliedasdepictedinFigure7.1.Becauseadjacentmagnetizationlayers 109 areperpendiculartoeachother(Ni ? Co/Ru/Co ? NiFe),thereismaximumspin-triplet generationandthusthepillarinitiatesintheonstate. AmagneticofthissizeislargeenoughtotrapsomexintheNbtopandbottom leads,addinganunknownelementtoourmeasurements(nottomentionmakingtheFraun- hoferpatternlooklousy).Toremovethetrappedweslowlyraisethedipperslightly outoftheheliumtoallowtheNbtogonormal,vbymeasuringsampleresistance withahand-helddigitalmultimeter(DMM).AssoonastheNbgoesnormal,thesample resistancejumpsup,atwhichpointtheDMMisremovedandthesampleisredippedinto thedewar.Althoughthisstepseemsunsophisticated,itisnotacarelessoneasitisvital thatthetemperatureofthesamplebekeptascoldaspossible.Raisingthetemperature abovetheCurietemperatureofanymagneticlayerwillcauseittoloseitsmagnetization, forcingtheusertore-initializeandremoveagain. Atthispoint,asimpleFraunhofermeasurementistakenfrom20to 20mTandback again(assuminginitialmagnetizationwaspositive,orderofsweepsedifnegative). Thisisdonetomakesurethesampleisbehavingnormallyasjudgedbyanumberofits characteristics:thatthepeakoftheFraunhoferpatternhasalargeenoughcriticalcurrent andisn'tshiftedtoofarcenter,thatthenormalstateresistancematcheswhathadbeen previouslymeasured,thatitrespondstoamagneticandthatisshowssignsofswitching theNiFemagnetizationdirection180degrees(seeFigure7.1c).Ifanyofthosecriteriaare notmet,thesampleisreinitializedandmeasuredagain.Iftheresultsconsistentlyshow poorbehavior,thesampleisremovedandanewsampleismountedinstead. 110 (a) (b) (c) Figure7.1 Representationsofmagnetizationdirectionintheonstate. Blue:Ni;Pink: Co/Ru/CoSAF;Green;NiFe.(a)representsaskewedviewofthestack,while(c)and(d) representthetop-downview.(d)alsoshowstheswitchingoftheNiFeasitswitchesin positiveandnegativebutremainingorthogonaltotheSAF. 111 (a) (b) (c) Figure7.2 Representationsofmagnetizationdirectioninthestate. Blue:Ni;Pink: Co/Ru/CoSAF;Green;NiFe.(a)representsaskewedviewofthestack,while(c)and(d) representthetop-downview.(d)alsoshowstheswitchingoftheNiFeasitswitchesin positiveandnegativebutremainingparalleltotheSAF. 112 7.1.2Switching:Zero-FieldMeasurements Atthispoint,thecriticalcurrentatzeroshouldbelargeandnearthepeakofthe Fraunhoferpattern.AswerotatetheNiFe90 tobecomeparallelwiththeCo/Ru/CoSAF (Figure7.2),thespin-tripletcomponent,andthusthetotalcriticalcurrent,shoulddrop dramatically.Tocheckforthis,weapplyamagneticinthetransversedirectionand measurethecriticalcurrentatzeroAsthemagnetizationstartstorotate,thecritical currentshoulddropuntilitors(whenthepillarisfullyrotated). TheexactsameprocedureisdoneinreversetomeasuretherotationoftheNiFeasa longitudinalisapplied.Inthiscase,astheNiFerotatesbacktoitsinitialdirection,the criticalcurrentshouldincreaseuntilitlevelsatamaximum.Thisiswhenthepillaris fullyrotatedlongitudinally,bringingthesystembacktoitsonstate. 7.1.3Switching:FraunhoferMeasurements Tobettercomparewiththeinitialdata,thissamerotationismeasuredasaFraunhofer pattern.Ifsuccessful,thecriticalcurrentshoulddropasthetransverseisincreased. Atwhichpoint,afullsweepcanbetakentoshowthatthesamplestayslowduringthe entiretyofthesweep.Thissweepdoneinthestateshouldshowadramaticerence whencomparedtotheinitial(on-state)Fraunhoferpattern. Forcompleteness,thissamemeasurementisdoneagaininthelongitudinaldirection.The criticalcurrentshouldstartinalowstate,eventuallyrotatingtothehighstateathigher Afterrotation,thecriticalcurrentsintheFraunhoferpatternshouldbemuchlarger thaninthestate,andifdonefullywithoutothermagnetizationlayers,this patternshouldmatchtheinitialpatterntaken. 113 7.1.4Switching Whiletheaboveshouldbeenoughtodemonstratetheswitchingcapabilitiesofthepillars, onemeasurementisundertaken:measuringonlythecriticalcurrentaftereachiteration ofswitchingthespin-tripletsupercurrentonandTodothis,alargeenoughtoswitch theNiFe(asmeasuredinSection7.1.2)isapplied,alternatingbetweenthelongitudinaland transversedirections.ely,thisshouldturnthespin-tripletonandresultingina largeandsmallcriticalcurrent,respectively,whenmeasuredinzeroeld.Ifthesampleis behavingproperly,thecriticalcurrentsshouldalsobereproduciblebetweenconcurrenton )measurements. 7.2RightPlaceattheRightTime Asmentionednumeroustimesalready,theworkdiscussedhereinrequiresrotatingthemag- netizationofspferromagneticlayersbetweentwoaxes.Amainreasonithadneverbeen attemptedbefore,despitethepioneeringspin-tripletpaircorrelationworkanddiscovery4 yearsprior(seeSection3.2),isthatwehadnowaytocontrolanexternalmagneticin twodirections.ThereforetheimportanceofQD-IIanditsabilitytogeneratethenecessary magnetic(asdiscussedinSection5.1)cannotbeoverstated.Finishingthemagnetic wiringoftheprobe,completedjustbeforetheonsetofthisproject,wasthekeytoopening thisareaofinvestigation. Anotherfacetthatallowedforrelativelyrapidresultsisthatspin-tripletsupercurrenthas beenamajorfocusofthegrouprecently,meaningmuchoftheworknecessarytooptimizethe fabrication,magneticlayercharacterization,etc.hadbeendonepreviously.Whileindividual 114 aspectsstillneededtobetakenfurther(seeChapter6),alotofthedetailshadbeen tunedpriortomyinitialinvolvement. 7.3EarlyAttempts 7.3.1RotatetheSAF Knowingittobenaive,theattempttomeasurecontrolofthespin-tripletwasdone withsamplesmadebyYixingWangthathadNiforbothspin-mixerlayersandaCo/Ru/Co SAFasthecentralferromagnet[Nb(150)/Cu(5)/Ni(1.2)/Cu(5)/Co(6)/Ru(0.75)/Co(6)/ Cu(5)/Ni(1.2)/Cu(5)/Nb(20)/Au(15)/Nb(150)/Au(20)].Awareofthemagnetichardnessof Ni,asitrequiresa > 200mTtofullyinitializethemagnetizationdirection,thehope wastorotatetheSAF,causingtheCo/Ru/Cotospinparallelorperpendiculartothe Ni,andthusputtingthejunctionintoanoronstate,respectively. Iclaimthiswasanaivetask,butinrealityitwasundertakenwithabitofforethought. Firstwealreadyhadaccesstosampleslikethis. If thistypeofsampleworked,there wouldbenoreasontospendtimefabricatingduplicates.Secondly,whilewewerenot expectingthemtoworkoutright,wehopedwewoulddiscoverpotentialconcernsandbe abletoamelioratethemwhilefabricatingthenextsamples. Whatwefoundwasapromisingtrendwithoutconcreteresults.TheinitialFraunhofer patterns,Figures7.3aand7.3b,showthattheproducedbytheNishiftedthepeakaway from0mT.Whilewehopedtogetapeaknear0mT,wecouldalwaysdevelopaprogramthat allowsustomeasureinasmallandhitthepeak.Soinsteadofgettingtooconcerned withthatrightaway,wedecidedtopressonwiththerestofthemeasurements. 115 (a) (b) Figure7.3 InitialFraunhoferpatternmeasurementsforNi/SAF/Nisamples. In(a)and(b), theisappliedintheonanddirections,respectively. (a) (b) Figure7.4 InitialswitchingbehaviorofNi/SAF/Nisamples. Allmeasurementsaretakenin zeroafterarotatingisappliedinthex(a)ory(b)direction. 116 (a)transvmeasurement. (b)lmeasurement. (c)On-statetransvmeasurement. (d)On-statemeasurement. Theredandblackcurvesaremeasuredinthe samesweepdirection.Thelackofoverlapis evidenceofsomedomainmotionintheNi. Figure7.5 On-andmeasurementsforNi/SAF/Nisamples. Measuringinzero,wecanseethattheFraunhoferpatternpeakdecreasedwitha largex-,or(Figure7.4a),andincreasedagainwithasimilar-magnitudey-,oron-, (Figure7.4b),aswehoped. WehadevenbeenabletomeasuretheFraunhoferpatternsinthelowandhighstates, asshowninFigure7.5.However,wedidn'tknowenoughaboutthestabilityofthenickel andhowitisectedbyaslargeas30mT.CouldthebehaviorshowninFigure7.4be duetosomecombinationofNiandCo/Ru/Corotation?Isthereanywaytobesurethat 117 theCo/Ru/Coisrotatedfully,either? RunningtheFraunhoferpatternasecondtime,asshownastheredcurveinFigure 7.5dtellsusthatwehaveenoughNidomainsmovingtocauseaproblem,asthereisashiftin thepeak(andzeros)betweenthetworuns.Evenifonlyasmallfractionofdomainsrotated inthatitisapparentlyenoughtoaltertheinternalofthejunction,movingthe centralpeakposition.Thisappearsasadecreaseincriticalcurrentatzerold,inthiscase byafactorofroughly3.Sinceourcriticalcurrentismeasuredata(typically 0mT),movementoftheFraunhoferpeakcausesalotofproblemsmovingforward.To makesurewearealwaysatthepeak,wewouldhavetomaptheentire2-Dspacefor everymeasurement.Otherwisewecouldneverbesureifwewerecontrollingthespin-triplet supercurrent,asdesired,orjustmovingthepositionofthecentralpeakintheFraunhofer pattern. Weknewthatafull2-Dmapwasn'tfeasible.Somovingforward,althoughthedata werenotexactlywhatwehopedfor,itdidrevealthethreemainissuesweknewwehadto address: 1.ArewealteringtheNi,theCo/Ru/Co,orboth?Canwelimitourrotationtoonly oneofthose? 2.Areweobserving\switching"duetomovingthecentralpeakratherthantuningthe spin-tripletsupercurent? 3.Aretherotationscleanenoughtobesureminimaarerealandnotjustnoisefrom individualdomainsmovingaround? Asolutiontotheissuewassimple,makingittheobviousonetotry.Therequired tofullyrotatetheCo/Ru/CoSAFwastoohightobesureweweren'tsimultaneously 118 theNi,soitbecamenecessarytoincorporateasofterferromagnettorotatewhile leavingtheothersalone. 7.3.2LargeNiFeMoPillars Beingundertakenconcurrentlywiththisproject,othergroupmemberswerelookingatcon- trollingthe0- ˇ phaseofferromagneticJosephsonjunctions,asmentionedinSection3.3.2, measurableinsuperconductingSQUIDloops.Injunctionscontainingtwoorthreeferromag- neticlayers,thephasecanbeedbyrotatingonelayermagnetizationby180 (Fig2.12). Thereforethepropertiesofsoftferromagents,suchastheabilitytogeneratespin-triplet supercurrentandtherequiredswitching(thenecessarytothemagnetization 180degrees),werealreadybeinginvestigated(seeSection3.3.2).Ofthesampledmaterials, NiFeMopillarsseemedtomeetourneedsasitgeneratesenoughcriticalcurrentandhasalow switchingInthehopeofrevealingresults,thedecisionwasmadetomeasurethesesam- ples[Nb(150)/Cu(5)/Ni(1.2)/Cu(10)/Co(6)/Ru(0.75)/Co(6)/Cu(10)/NiFeMo(1)/Cu(5)/ Nb(20)/Au(15)/Nb(150)/Au(20)]insteadoffabricatingnewones.Whilewestillhadother potentialconcerns(namelyquestions2and3above),ifthesesamplesworked,fabrication wouldagainbeunnecessarytimeandiftheydidn't,perhapswecouldoutwhy anditwhilefabricatingthenextbatch.Onceagain,thissimplecheckrevealedvaluable informationbeforemakingnewsamples. Asbefore,weseethatthecenteroftheFraunhoferpatternisshiftedfrom0mT,but nowwecanclearlyseethattheNiFeMoisitsmagnetizationdirection.Thiscan beseeninthatthetwolongitudinalsweepdirections(Figure7.6a),onestartingfrom 5 mTandincreasingandtheotherfrom2mTanddecreasing(directiondenotedbycolored arrows),havepeaksthathavemovedrelativetoeachother.AstheNiFeMoswitches,its 119 (a) (b) Figure7.6 InitialFraunhoferpatternmeasurementsforNi/SAF/NiFeMosamples. Both(a) and(b)areFraunhoferpatternsmeasuredintheonstate.However,duetotheshiftofthe centrallobeof0mT[blackcurvein(a)],themeasurementinthex-directionremained lowtheentiresweep. magnetizationwilleitheraddtothatoftheNi,movingtheshiftfurtherfrom0mT,or cancelit,bringingitbacktowards0mT.Thisisapromisingresultasittellsusthatweare actuallymovingmagnetizationwithinthepillar,asdesired.Again,theinitialsweepinthe transversedirection(Figure7.6b)islowthroughout,includingat0mT,duetotheshiftof thepeakcenter. Figure7.7ashowsthemeasurementsofthecriticalcurrentat0mTaswerotatethe NiFeMotothelowstate,whileFigures7.7b,and7.7cshowtheFraunhofersweepsinthe longitudinalandtransversedirections.Weonceagaincanseesomeamountof180-degree intheX-direction,butthesizeofthispeakistroubling.Weexpectedittobelow, assumingwehadrotatedtheNiFeMo.Apparently,wemayhavejustrotateditenoughto havemovedthepeak,butnotfullyturnedthespin-tripletsupercurrent. WhentryingtorotatetheNiFeMolongitudinallybacktotheinitialstate(Figure7.8a), thingsbecomeevenmoretroubling.Thecriticalcurrentneverseemstoincrease,even whenapplyingmorethantwicewhatwasusedtorotateinthetransversedirection. 120 (a)Zeroswitchingmeasurement (b)transvmeasurement. (c)longitudimeasurement. Thethird(blue)datawasmeasuredtoen- surethestateofthesamplehadn'tchanged duringthemeasurement. Figure7.7 Low-statemeasurementsofNi/SAF/NiFeMosamples. 121 (a)switchingmeasurement (b)On-statetransvmeasurement. Thethird(blue)datawasmeasuredtoen- surethestateofthesamplehadn'tchanged duringthemeasurement. (c)On-statemeasurement. Figure7.8 High-statemeasurementsofNi/SAF/NiFeMosamples. 122 Fraunhofersweeps(Figures7.8band7.8c)onceagainshowapeakinthetransverse butnotinthelongitudinalThispeakalsomovesbetweentheup-anddown-sweep directions,despiteneverexperiencingmorethana30mTexternalThis,aswellasthe switchingdatashowingmovementatlow(Figure7.7a),impliestheNiFeMo ismovingveryeasily,buttheFraunhoferdatatakenthroughoutimpliesthatthisswitching isneithercleannorconsistent. Intheprevioussection,Imentionedmeasuringinasmalltomakesurewewere alwaysatthepeak.However,withtheapparenteaseofmovingthemagnetizationand thusinternalandpeakposition,measuringinanysortofcouldbeenoughto changemagnetizationonceagain.Thatistosay,sincethepeaksinFigure7.7bare about2-3mT,itwouldbeidealtokeepthissmallexternalonforallmeasurements. However,becausethereisroughlya20%decreaseincriticalcurrentwithinthe2mT (seeFigure7.7a),thiscouldbeenoughtodisruptmagnetizationthroughouttheentire experiment.Wealsorealizedthat,ifrotatinglongitudinallyortransverse,wecanonly measuretheFraunhoferinthatdirection,unlesstheiskept very low.Becauseofthis, futuremeasurementswereonlytobedoneinonedirection(asdescribedinSection7.1)to avoidanyminorrotation. Whenusingsoftmagneticlayersthatcanmovemagnetizationeasily,theonlysolutionto theissuewouldbetoextendthewidthoftheFraunhoferpattern.Thiswouldensure thatwearecloseenoughtothepeaktomeasurehighonthecentrallobe,evenifnotatthe maximum,forallmeasurementsinnoexternalAccordingtoEqn2.46,wecanseethat thevaluesof B thatproduceminimainthecriticalcurrentareinverselyproportionaltothe sampleradius R ,andaregivenbythezerosoftheBesselfunction.Therefore,thewidthof thecentralpeak,inmT,isdeterminedbythediameterofthesample.Solvingforoursample 123 geometrywecancalculatethat,forpillardiametersof1 ,thepeak-to-zeroFraunhofer patternwidthisroughly12mT,andevenlargerforsmallerpillardiameters.Thiswouldbea solutiontothesecondissueraisedintheprevioussection,limitingourconcernswithmoving theFraunhoferpeak.Inaddition,magneticlayersofthissizeshouldbesingledomain.This shouldmaketheswitchinglessambiguous,resultinginamuchcleanerFraunhoferpattern ingeneralandsolvingthethirdissuediscussedabove. 7.3.3SmallNiFeMoPillars Withacleardirection,sampleswiththestructureofNb(100)/Cu(5)/NiFeMo(1)/Cu(10)/ Co(4)/Ru(0.75)/Co(4)/Cu(10)/NiFeMo(1)/Cu(5)/Nb(20)/Au(15)/Nb(150)/Au(20)were fabricated.Thediameterofthesepillarswere0.2,0.3,and0.5 .Themillingdepth, i.e.howmanyferromagnets(0-3,labeledP0-P3)werepatternedintosingledomaincircles, wasalsovariedoneachchip.WhilejustpatterningtheNbabovethejunctionisenoughto thejunctionsize,actuallymillingthroughtheferromagnetscouldensureanabsence ofdomainwallsaswellasalterthenecessarysizeoftheswitchingWhilealldepths wereinvestigated,issuessuchasdipolarcouplingandsofteningtheSAFbecameappar- ent.Therefore,thefollowingdiscussion(inthisandfuturesections)assumesonlythetop ferromagneticlayer(P1)waspatterned. Inadditiontothepillardiameter,therewereseveralotherbetweenthese samplesandtheprevious.AllspinmixerlayerswerenowNiFeMotoenhancethewe werehopingtosee.BecausechangingNitoNiFeMoreducesthespin-tripletgenerationand thusthecriticalcurrent,Co(6)/Ru(0.75)/Co(6)wasdecreasedtoCo(4)/Ru(0.75)/Co(4)to compensate.ThebaseNbdroppedfrom150nmto100nmduetotheNbroughnessrelative toitsthickness.AFMmeasurementsshowedadramaticdecreaseinsurfaceroughness,from 124 (a)I-VcurveforNiFeMo/SAF/NiFeMo samples. (b)I-VcurveforNi/SAF/NiFesamples. Figure7.9 I-VcurvesforNiFeMoandNiFe. ThecriticalcurrentinNiFeMo/SAF/NiFeMo samplesistoolowtoelymeasure.AtypicalI-VcurveforNi/SAF/NiFeisshownfor comparison,demonstratingamuchlargercriticalcurrent. 0.61nmrmsto0.35nmrms[65]for150nmand100nmNb,respectively.Becauseallother metallayersaregrownonthisNbbaselayer,itisimportanttokeeptheNbsurfacesmooth. Thisensuresbetterferromagneticlayers.Lastly,thepreviousrunhad10nmCubetween ferromagneticlayerswhiletheonebeforehad5nm.Itwasatthispointdecidedthatall futuresampleswouldmaintain10nmofCubetweenferromagnetsinanattempttodecrease couplingfromnearbymagnets.Despitemybestandattemptsatforethought, theresultsfromthesesampleswerefruitless. AsshowninFigure7.9a,themaximumcriticalcurrentobtainedinthesesampleswas about400nA,muchlowerthanweexpectedtheSQUIDmeasurementsystemcouldmeasure. Forcomparison,Figure7.9bdisplaysanI-VcurvefromaNi/SAF/NiFesample,whichhasa criticalcurrentcloserto50 A.Whilehappytoseethatoursystemwasmoresensitivethan expected,thedatacollectedfromthesesampleswerefarfromconcreteandfartoocloseto ournoiseorforcomfort. 125 7.4NiFePillars Althoughmovingintherightdirectionwithrespecttomagneticswitchingproperties,afew changesstillhadtobemaderegardingourmaterialselectionandgeometry.Atthispoint,the workdescribedinChapter6wascarriedout,andNiFewaschosenasagoodcandidateforthe softferromagneticlayerasitshowedlargertripletsupercurrentandbetterswitchingbehav- iorthanNiFeMo.Toenhancethesupercurrentevenmore,thebasespin-mixerlayerwasonce againreplacedwithNi,meaningwewerethesizeofthesupercurrentratio byrotatingonlyonespin-mixerlayerrelativetotherestinexchangeforalarger(andmeasur- able)criticalcurrent.Thesizeofthepillarswasalsoincreasedto0.5,0.7,and1.0 diam- eter,hopingtoincreasethecriticalcurrentevenmore.Stillbeingsingledomain,thesejunc- tionswouldalsohaveawideenoughcentralpeaktocontlymeasureclosetothemaxi- mum,evenifmovingtheFraunhoferpeakremainedaconcern.Thesesampleswerefabricated asfollows:Nb(100)/Cu(5)/Ni(1.2)/Cu(10)/Co(4)/Ru(0.75)/Co(4)/Cu(10)/NiFe(1)/Cu(5)/ Nb(20)/Au(15)/Nb(150)/Au(20). 7.4.1FirstEvidence Toourfortune,beitduetodiligenceorluck,thesamplewemeasuredwiththisgeom- etryshowedusexactlywhatwewanted,asshowninFigures7.10aand7.10b.Theinitial Fraunhoferpatterndisplayednothingnegativelynoteworthy.Theforwardandbackward directionsdidnotoverlapperfectly,butconsideringtheNiFeispointingatway foreachsweep,thisisnotasurprise.Overlapdidoccurattheextremes,wherethetwo curveshavethesamemagnetization.Inaddition,thecriticalcurrentinthewidecentral peakisgreatlyabovethatofthelobes;eventhoughthecriticalcurrentatzeroldisnot 126 (a) (b) Figure7.10 InitialmeasurementsforNi/SAF/NiFe{Sample1. Datain(a)and(b)arethe same,plottedwithtranges.Measurementsto 20mTagreewellwiththose takento 40mT. themaximum,itislargeenoughtodistinguishwhetherthespin-tripletsupercurrentison or ItshouldbenotedthatthedatashowninFigures7.10aand7.10barethesame,but plottedwithtranges.Bymeasuringoutto 40mT,weobservemorelobes oftheFraunhoferpattern.However,toavoidunnecessaryrotationofmagnetizationwithin ferromagneticlayers,wekeeptheexternalaslowaspossible.Fromthecomparisonof thedataweshowthatitisnotnecessarytouselargerthan20mT,despitethedata notdemonstratingasclearofaFraunhoferpattern. Thenextstep,asdiscussedinSection7.1,istomeasurethesizeofthecriticalcurrent atzeroaswerotatetheNiFemagnetization.EvenifweweremovingtheFraunhofer peak,aswasaconcernbefore,itswidthensuresweareobtainingadirectmeasurementof criticalcurrent,andthusthemagnitudeofspin-tripletsupercurrent,nearthemaximum.As weincreasetheeldinthetransverse)direction(Figure7.11a),thecriticalcurrentgoes down,andcomesbackupastheincreasesinthelongitudinal(on)direction(Figure 7.11b).AccordingtoFigures7.11aand7.11b,theNiFerotateswithin20mT.Thiscorre- 127 (a) (b) Figure7.11 Sample1switchingdata. Thetripletcriticalcurrentturnsasthemagneti- zationofNiFeisrotatedbythetransversemagnetic(a).Thetripletcriticalcurrentis turnedbackonwiththeapplicationofalongitudinal(b). spondswiththedatafromSection6.1.2,alsoshowing180-degreeswitchingwithin20mT. Lastly,AMRdata(seeSection6.2.3)showsthatCo/Ru/CoSAFswith4nmCothickness shouldn'trotatemuch( < 10%)withinthe20mT,either. Althoughthechangesinthecriticalcurrentforeachsuccessivearenotconstant,and thecriticalcurrentevenplateausforcertainranges,giventheirsmallsizeitispossible thatthepillarsaren'tperfectlyroundorhavesomeotherdefects.Assuch,themagnetization mayenergeticallyfavorableminimawhilerotating,buteventuallyitdoescontinueuntil fullyrotated. Therequiredtorotatethemagnetizationbacktoitsoriginal(on)stateislessthan thatrequiredtorotateittoitsstate.Thisbehaviorisconsistentwiththemagnetocrys- tallineanisotropyinducedbythegrowth(seeSection2.1.1).Becausethesamples aregrowninamagnetic(seeSection4.2.2),theinternalmagnetizationfavorstheon direction,makingiteasiertorotatetowardthisdirection.Inadditiontothemag- netocrystallineanisotropy,whiletheferromagneticlayersaremostlydecoupledbytheCu spacersbetweenthem,straysduetodipolarcouplingbetweenlayerscouldstill 128 (a)On-toFraunhoferpatternsweeps. (b)toon-stateFraunhoferpatternsweeps. Figure7.12 FraunhoferpatternsmeasuringtheswitchinSample1. Theblackcurvesmeasure theNiFemagnetizationasitisrotatedintothestatefor(a)andintotheonstatefor(b). Redandbluecurvesineacharesubsequentmeasurements,demonstratingthatthesample remainsinthatstate. havean ThenextstepistomeasureinandobtaintheFraunhoferpatternasthemagne- 129 Figure7.13 switchingforNi/SAF/NiFe{Sample1. Redandblackcurvesmeasure theintdirections(positiveandnegative).On/ratio ˇ 7. tizationrotates,measuringinforthepatterninitsstateaswell.AsFigure7.12a shows,oncethesampleisinitsstate,itstaystherefortheentiretyofthemeasurement, notreturninguntilbeingrotatedagain.Thisisalsotrueforturningthetripletsupercurrent backon(Figure7.12b),maintainingthehigh-tripletstate,asseeninthe Thelaststepistodemonstratetheabilitytoturnthesamplefromontoandback againbysimplyapplyingthenecessarytorotate.Applyinga20mTintheon )direction,measuringthecriticalcurrentshouldshowusahigh(low)value.Thisis demonstratedinFigure7.13.Thereseemstobeabitofatrainingperiodinthecouple ofswitches,whichisnotfullyunderstood. Theratioofcriticalcurrentsizebetweenonandisabout7forthissample. Thatratio,alongwiththeshapeoftheFraunhoferpatternsforonandmeasurements, isttoclaimthattheabilitytocontroltripletsupercurrentinthesesampleshas beenrealized.Forcompleteness,theinitialFraunhoferpatterns(longitudinal)andthe 130 Figure7.14 InitialandFraunhofermeasurementsforNi/SAF/NiFe{Sample1. Black andRedcurvesdemonstratetheinitialsweepswhileblueandpinkdemonstratethe sweeps,takenafterbeingswitchedbetweenstatesnumeroustimes. onesareplottedoneachotherinFigure7.14.Althoughthesizeofthecriticalcurrentis notexactlythesameforeachpoint,thegeneraltrendisthesame.Mostdeviationsoccur aftertheswitchingbegins,whichisalargelyuncontrollableregime(untiltheswitchingis completed).Therefore,atthispointarenottooalarming. 7.4.2AFabricationHiccup However,onesampledoesnotatrueexperimentmake.Moresampleswerethenfabricated andmeasuredwiththesamegeometry.Duringthissputteringrun,theCotargetcameloose fromitshousing,potentiallycontaminatingtheColayerswithsomeindium,makingthe SAFsofter.ThesesamplesallhadeitherpoorinitialFraunhofermeasurements,implying somemagnetizationlayerhadn'tsetproperly,orshowedevidenceofSAFrotationmuch earlierthanexpected.ThiscausedustolookcloserattheSAFdata,decidingitwouldbe prudenttotrytokeeptheaslowaspossibleasopposedtousing20mTforevery sample.Even20mTisenoughtopotentiallyrotatetheSAFalittle,sowewanttolimit ourtoonlywhatisnecessarytorotatetheNiFe. 131 Figure7.15 Fielddependenceofswitching. switchingdatawastakenformul- tiplesizes,aslisted(inmT).Theeveniterationsarealwaysinthelongitudinal,oron, direction.At20mT(blue),thelongitudinaldirectionbecomesthedirection,likelydue toSAFrotationinthepillar. However,despitebeingabitdisappointing,thesesampleswerestillmeasuredtosee theeofSAFrotationonourmeasurements.Themostusefuldatatakenfromthis experimentwaslookingattheswitchingforerentFigure7.15showsthis switchingforfourerentTheinitialmeasurement(5mT)showsasmall whichgetslargerwithincreasedd(9mT).However,startingat20mT,furtherincreases startstorotatemorethanNiFeMo,withaswitchbetweenwhatbecomestheonand states.(Forallswitchingdata,eveniterationsshouldbetheonstatewhileoddare Thisbetweenwhatishighandlowremainsthatwayforallsubsequent(up to45mT). WecanconcludefromthisresultthattheSAFisrotatinginthesemorethanwe anticipated.WhetherthisisduetosofteningoftheSAFfromindiumcontaminationor someothersource,itreallydriveshometheneedtokeepoursmallenoughtoprevent 132 asmuchSAFrotationaspossible. 7.4.3Reproducibility AftercleaningtheCoanditshousingtopreventfurthercontaminationandsecuringitin placetoensureitdidn'thappenagain,newsampleswerefabricatedwiththesamestructure asbefore.TheywereonceagainmeasuredasexplainedinSection7.1,andtheirdatais plottedinFigures7.16-7.25.InantominimizeCo/Ru/Comagnetizationrotation, thesizeofthewaskeptassmallaspossible,whichcanbeseeninFigures7.17and7.21. This,aswellassomepotentialconcernswithkeepingthetoosmall,willbedetailed below. TheFraunhoferpatternsthatweremeasuredwhilethemagnetizationwasrotating(Fig- ures7.18and7.23)demonstratethesamecharacteristicsweobservedinSample1.The samplesdemonstrateaswitchfrom\on"to(a)andviceversa(b)betweensubsequent Fraunhoferpatternmeasurements.Inadditiontothese,anextrameasurementwastaken withSample3.Afterthemagnetizationwasrotatedintotheonorstate,averynarrow Fraunhoferpatternineachdirection(xandy)wasmeasurednearzero(Figure7.22). Thiswasdonetoshowthatthestateisstable,eveninthepresenceofasmallwhich isdemonstratedbythereversibilityofallmeasurements.Onmeasurementsofthesetwo samples(Figures7.19band7.25),demonstratingratiosof19forSample2and5forSample 3. Asmentionedabove,thesizeoftheexternalusedwhilemeasuringSamples2and3 waslimitedasmuchaspossibletopreventCo/Ru/Corotation.Todeterminethislimit,the measurementsinzeroweredoneasnormal,steppingthetransverseupto20mT, althoughtheminimummayoccurbefore20mT.Theatwhichtheminimumisfoundis 133 Figure7.16 InitialFraunhofermeasurementsofNi/SAF/NiFe{Sample2. (a)switchtostate{Sample2. (b)switchtoonstate{Sample2. Figure7.17 Ni/SAF/NiFeswitchmeasurement{Sample2. Switchinginzeroin (a)and(b),measuringtherequiredtorotatethemagnetizationtotheandonswitch, respectively.Blackcurvesin(a)and(b)measurerotationto20mT,whichincludesrotation oftheSAFat19mTin(a).TheredcurveskeepthelowenoughtoavoidasmuchSAF rotationaspossible. 134 (a)On-toFraunhoferpatternsweeps. (b)toon-stateFraunhoferpatternsweeps. Figure7.18 Fraunhoferpatternsmeasuringtheswitch{Sample2. Theblackcurvesmeasure theswitchfromontostate(a)andviceversa(b).Theredandbluecurvesineachare subsequentmeasurements,demonstratingthatthesampleremainsinthatstate. (a)switchingat15mT. (b)switchingat16mT. Figure7.19 switchingforNi/SAF/NiFe{Sample2. Eveniterationshavethe appliedinthelongitudinaldirection,odditerationshaveappliedtransversely.Curves in(a)arealltakenat15mT,butforappliedintdirections(positiveand negative).Thelackofreproducibilityiseliminatedwhentheisincreasedslightlyto16 mT(b).ratioforthissampleis ˇ 19. 135 thethatisusedforallsubsequentswitchinginthesample.Anexampleofthisstepcan beseeninFigure7.17a.Therunmeasuresinzerouptoa20mTexternallyapplied transverse(black).Above18mT,though,thecriticalcurrentincreases,implyingsome rotationoftheSAF.Asecondmeasurementwasundertakentodemonstratemonotonic decreaseofcriticalcurrentto15mT(red)inthetransversedirection.Measurementsinthe longitudinaldirectionshowthatthesamplereturnstotheonstatewithinthatrangeaswell (Figure7.17b).Therefore,theutilizedtorotatethesampleissetto15mT. Whilewewanttokeeptheaslowaspossible,limitingthetoexactlythe minimumvaluehasitsdetrimentsaswell.ThismaybeenoughtorotatetheNiFe90 degrees,butmightnotbeenoughtofullyit180degrees.Sample3demonstratedthis plottedinFigures7.23band7.24.LookingatthedatainFigure7.21b,sweepingto 10mTshouldbeenoughtoresetthesampleintotheonstate.Therefore,aftersweeping from0-10mT(blackcurve,Figure7.23b),thesampleshouldbeintheonstate.Thisis inthesubsequentsweepfrom10to 10mT(redcurve).However,whensweeping backfrom 10mT(bluecurve),weobservewhatlookslikebehavior.Ibelieve thisisduetotheNiFenotfullyrotating180 at 10mT;whensweepingback,adjacent magnetizationdirectionsarenotorthogonal,andthusnotamaximum.However,resetting thismeasurementto 20mT,thetowhichitwasinitiallymeasured,weseethesample isbackintheonstateasexpected. ThisisalsoplottedinFigure7.24.Thefourcurvesplottedareasfollows:black{initial measurementfrom20to 20mT;red{initialmeasurementfrom 20to20mT,takenafter 180 duetoblack;blue{comparisontoinitialon-statemeasurement(black)from10to 10 mT,takenafter90 rotation;pink{comparisontoinitialon-statemeasurement(red)from 20to20mT,takenafter180 duetoblue.Eventhoughthebluecurvewastakenafter 136 Figure7.20 InitialFraunhofermeasurementsofNi/SAF/NiFe{Sample3. (a)switchtostate{Sample3. (b)switchtoonstate{Sample3. Figure7.21 Ni/SAF/NiFeswitchmeasurement{Sample3. 137 (a)Fraunhoferpattern sweeps. (b)On-stateFraunhoferpattern sweeps. Figure7.22 Near-zerFraunhoferpatternsmeasuringthestate{Sample3. Theblack curvesineacheitherturnsthetripletsupercurrent(a)oron(a).Thecritical currentisthenmeasuredinalldirections,x-andy-,tosmalltoshowstabilityofthe magnetizations.Thisensuresthesamplereallyisinthisstateandstableinsmall (a)On-toFraunhoferpatternsweeps. (b)toon-stateFraunhoferpatternsweeps. Figure7.23 Fraunhoferpatternsmeasuringtheswitch{Sample3. Theblackcurvesmeasure theswitchfromontostate(a)andviceversa(b).Theredandbluecurvesin(a)are subsequentmeasurements,demonstratingthatthesampleremainsinthestate.Thered curvein(b)remainsintheonstate,butdoesn'tthemagnetizationfully.Therefore,the bluecurveisn'tintheonstate.Thisisresetwhenalargerisapplied(pinkcurve). 138 Figure7.24 InitialandFraunhofermeasurements{Sample3. Theoverlapsbetween blackandblueaswellasbetweenpinkandreddemonstratethesamplehasreturnedtoits initialstateafternumerousswitches. Figure7.25 switchingforNi/SAF/NiFe{Sample3. ratio ˇ 5. 139 rotationfromtoonstateinonly10mT,weseethattheblackandbluecurvescompare nicely.Thistellsusthat10mTisenoughtorotatethesamplebacktoitsonstate.However, thepinkcurverequiresalarger( 20mT)tocomparewellwiththeinitialredcurve. Thisimpliesthat180 requiresmorethanthatof90 . Keepingthetoolowcanalsotheswitchingmeasurements.Alldatataken priortothisalswitchingmeasurementwereacquiredusingaslowlyincreasingwhich willincrementallyrotatetheNiFemagnetization.However,themeasurementisdone byapplyingthefullnecessaryinalternatingdirections,transverseandlongitudinally. Itispossiblethat,torotate90degreesinonestep,alargerisrequiredthanwhat isnecessarytoslowlyrotate90degrees.UsingSample2asanexampleofthison- switchingat15mTwasinconsistentandsmall(Figure7.19a).However,byincreasing theswitchingonlyslightlyto16mT,theswitchingwaspronouncedandreproducible (Figure7.19b). Overall,Samples2and3behavedinthesamewayasSample1,showingthesame switchingcharacteristicsandbehavior,withratiosofabout19and5,respectively. Potentialcausesforthencesinratios,aswellasamorethoroughanalysisofthesize oftheratio,willbediscussedinthenextsection.However,itisimportanttonotethatthe magnitudeoftheon-statecriticalcurrent,andmoreimportantly I c R N ,isconsistentacross allthreesamples.Thisimpliesthatthevariabilitybetweensamplesismorelikelycaused bymagneticpropertiesthanfabricationissuesorsampleinconsistencies.Additionally,when measuringcurrentsassmallasweobtainedwheninthelowstate,slightvariationsofthis currentcanhavelargeimpactsontheratio.Theimportantresultisthatthesamplehas changedfromahigh-toalow-tripletsupercurrentstate,orfromaspin-triplettoaspin- singletstate,andisstablewithoutthepresenceofamagneticld.Theisreproducible 140 betweensamples,andwithinasamplethesizeoftheisalsoreproducible. 7.4.4QuantitativeAnalysisofRatios Itishardtoignorethattheratiosforthishavefairlylargesample-to-samplevariation. Whileitistruethatthesizeoftheshouldbethesame,itisnotusefultoputtooa pointonit.Therearemanyesbetweeneachrunthatcanaccountfordiscrepancies: theinitialmagnetizationstateoftheNicanvary,andremovalisnotaprecisepractice; smallinroughnessorgrowthconditionscouldthehardnessoftheNiFe, etc.Moreimportantlyisthattherelativemagnetizationdirectionsplayaverycrucialrole intheamplitudeofsupercurrent. Duetominorbetweensamples,itisimpossibletodetermineexactlyhow muchwearerotatingtheNiFeandnottheNiortheCo/Ru/Co.Whenfullyintheonstate, alladjacentmagnetizationlayersareassumedtobeorthogonal( ' = =90 ).Recalling Eqn2.50,whichstates I c / sin sin ' ,wecanseethatsmalldeviationsfromthisangle marginallythecriticalcurrentamplitude.Incontrast,whenfullyinthestate, oneangleisorthogonal( ' =90 )whiletheotheriscollinear( =0 ).Thisstateisnever completelyachievedduetosmallrotationsintheSAF(seeSection6.2),andsmalldeviations in ( )willhavealargeronthecriticalcurrentamplitudethandeviationsin ' ( ' ), i.e. j sin( ) sin( ) j > j sin( ' ) sin( ' ) j forthesamesizedeviation. Itisthereforetoanalyzethetheoreticalratioweexpectinthesesamples. However,usingthedatafromAMRmeasurementsoftheCo/Ru/CoSAF,wecandetermine aroughestimateofspin-tripletgenerationintheonandstatesfromEqn2.50.Inthis analysis,inthestate,wewillassumethatthemagnetizationoftheNiFehasrotated90 relativetothatoftheNilayer.Obtainingresistancesfromthedata(lowercurveinFig6.9c) 141 Figure7.26 Cartoonofmagnetizationdirectionandrelativeanglesforthestateusedin ratioanalysis. andsolvingfor ' inEqn6.1,wecandeterminetherotationofthemagnetizationoftheSAF relativetothatofNias ' =arccos s R ' R ? R k R ? ! (7.1) where R k and R ? aregiveninTable7.1.Therelative-magnetizationanglebetweenNiFe andtheSAFis =90 ' .Thesedirectionsofmagnetizationforeachlayerandrelative anglesthereofbetweenadjacentlayersaredemonstratedinFig7.26.Intheonstate,wewill assume ;' =90 . R k 52.9044 R ? 52.7624 Table7.1 Tableof R k and R ? forCo(4)SAF. Theratiocanbewrittenas I c on I c off = 1 sin off sin ' off : (7.2) Eqn7.2yieldsratiosof3.1,3.6,and5.0,respectively.ThisdataissummarizedinTable7.2. 142 Ineachsample,wemeasuredratiosaslargeorlargerthanthosedeterminedbythisanalysis, implyingwehaverotatedtheSAFlessinthesesamplesthanaworst-caseanalysisyields. Sample SwitchField(mT) R( ' ) ' ( ) ExpectedRatio MeasuredRatio 1 20 52.7797 69.58 3.1 7 2 16 52.7745 73.01 3.6 19 3 10 52.7684 78.12 5.0 5 Table7.2 Tableofquantitativeratioanalysis. 143 Chapter8 Conclusions 8.1Overview Theearlytheoriespredictinglong-rangespin-tripletpaircorrelationsinsuperconductor/ ferromagnet/superconductor(S/F/S)Josephsonjunctionsspurredexperimentaliststore- alizethisphenomenon.Earlyresultsprovedpromising,andthedevelopmentofthree- ferromagnetic-layerstructures,writtenhereinasS/F 1 /F 2 /F 3 /SorS/F'/F/F"/S,granted researchersamethodtomakereproduciblesampleswithwhichexplorationofthis waspossible.Oncetheabilitytocreatesamplesthatdemonstratedlong-rangespin-triplet supercurrent(aswellasthosethatdidnot)becameclear,adesiretoturnthistripletsu- percurrent\on"andormeasurealargeorsmallcriticalcurrent,respectively,inthe samesampledevelopedwithinthecommunity.Thatwasthefocusofthiswork{tocreate ferromagneticJosephsonjunctionsinwhichthespin-tripletcriticalcurrentamplitudecould beadjusted.Relyingontherelativemagnetizationsbetweenneighboringlayers,wewere abletodemonstratereproducibleswitchingbyrotatingtheNiFespin-mixerlayerintoand outofcollinearitywiththeCo/Ru/Cosyntheticantiferromagnet(SAF). 144 8.2SummaryofResults Attheonsetofthisproject,asamplethatcoulddemonstratethisswitching,andstably maintainthestateinzerohadyettoberealized.Toobservethiswereliedoncon- trollingmagnetizationdirectionsofindependentferromagneticlayerswithinS/F'/F/F"/S Josephsonjunctions.Althoughsomeofthesamplesmeasuredhadbeenmadepreviously, noneofthemhadundergoneswitchingmeasurements.Inadditiontoinvestigatingthe withpreviouslymadesamples,alotofoptimizationandcharacterizationneededtobedone beforesuccessfulsamplescouldbefabricatedandmeasured.Thisprocessandtheresults obtainedaresummarizedbelow. ThesamplesmeasuredhadNiastheF'andF"layersandCo/Ru/CoSAFastheF layer.KnowingNitobeahardferromagnet,wehopedtorotatetheSAFinordertocontrol thecollinearityoftheferromagneticlayers.However,toachievethisrotation,measurements requiredverylargeThisresultedinmovementofNidomainsaswell,providing irreproducibledata.Fromtheseresults,itwasdeterminedthatasofterferromagnetwas neededfortherotatinglayer. NiFeMo,asofterferromagnetthanNi,waschosentoreplaceNiasonespin-mixinglayer (F").Niremainedastheotherspinmixinglayer(F')whilemaintainingCo/Ru/CoastheF layer,bothofwhichwerehardferromagneticlayers.AlthoughtheNiFeMolayerrotatedat lowerthaninprevioussamples,thesampledimensionspreventedusfromsuccessfully measuringswitching.IdeallythecentrallobeoftheFraunhoferpattern,ameasure ofthecriticalcurrentrelativetotheinthejunction,wouldbewideenoughsuchthat thecriticalcurrentisstilllargeatzeroHowever,thewidthofthecentrallobeina 3 -diametersamplewastoonarrowrelativetotheofthecentralpeak,preventing 145 alargecriticalcurrenttobemeasuredatzeroToobtainawidercentrallobe,the Josephsonjunctiondiameterneededtobedecreased,whichconvenientlyalsoallowedfor themagneticlayersintheJosephsonjunctiontobesingledomain.Allfuturesampleswere madewithsmallsamplediameters, 1 . Whilefabricatingsmallersamples,wealsoreplacedtheF'layerwithNiFeMo.Thiswas donewiththeintentofrotatingbothspinmixerlayers,increasingthesizeoftheeas describedinEqn2.50.Unfortunately,thecriticalcurrentwastoosmalltobemeasured. Additionally,betweenthissetofsamplesandtheprevious,theexpectedswitchingbehavior ofNiFeMoseemedtochange{aworryingresultwhentryingtoeliminatesample-to-sample variability.Atthispoint,beingcomfortablewiththeinitialinvestigation,wedecidedto narrowourscopetoafewmaterials:NiFe,Co/Ru/Co,andNi.Layercharacterizationwas necessarytodeterminetheoptimalthicknessesofeachlayerbeforefullsampleswerecreated. TripletgenerationmeasurementswithNiFedemonstratedthatitwouldbeassuitablea spin-mixerlayerasNiFeMo,withcomparableorhigher I C R N valuesandalowswitching ( j 0 H switch jˇ 10mT).Also,anisotropicmagnetoresistance(AMR)datafromCo/Ru/Co sampleswithvariousCothicknessesassuredusthatthinnerColayerwouldgrantahard syntheticantiferromagnet.Althoughthinnersamplesprovedharder,thereisalsolessshort- rangesupercurrentdecayinsampleswiththinFlayerscomparedtothosewiththickF layers.ToabalancebetweentheseopposingCo(4),wheretheparenthetical numberisthelayerthicknessin nm ,waschosenforeachlayerintheSAF. SampleswerethenmadeandmeasuredwithoneNiFespin-mixer(F")asarotating softlayer.Niwastheotherspin-mixer(F'),asitgeneratesmorespin-tripletcompo- nentsthanNiFe,andCo/Ru/CowasthecentralSAF.Usingthesmallergeometry(sample diameter=1 )andthiscomposition(F':Ni;F:Co/Ru/Co;F":NiFe),thesesamples 146 demonstratedtheabilitytocontrollong-rangespin-tripletamplitudeandmaintainthestate inzerod.Multiplesamplesacrossvarioussputteringrunsdemonstratedtheswitching wesought,withoratiosaslargeas I c on =I c off ˇ 19. 8.3FutureWork Despitefeelingcomfortablewiththeresultspresentedhere,samplesinthisgeometryun- doubtedlyhavea\GoldilocksZone"ofsorts.Themagneticmustbehighenoughto rotatetheNiFewhilestayinglowenoughnottorotatetheSAF.However,wefeelthatthe erangeissmallforthesesamples. Asstatedpreviously,SAFcharacterizationresultsimplythatthinnerCorequiresmore torotate,generatingahardersyntheticantiferromagneticlayer.However,thethinner thecentralferromagnet,thelesssuppressionofspin-singletsupercurrentispresent.Weare intheprocessofmakingandmeasuringsampleswithCo(3);perhapstheslightlyharder SAFwillgivealargerregionofacceptablemagneticthatminimallyrotatethemag- netization,thuswideningtheGoldilocksZone. Inaddition,itwouldbeidealiftheNiFeweresofterthanitisinthesesamples.Giventhat NiFeisthelastmagneticlayergrowninthestructure,surfaceroughnessfromtheunderlying layerscouldbecomeaverylargefactorastohowitbehaves.Althoughtheisnot large,roughnessmeasurementshaveshownthatgrowinga[Nb/Al] n multi-layerforthe bottomelectrode,asopposedtotheNb(100)currentlyinthesystem,providesasmoother surfacefortherestofthestack.Thiswould,however,requirebreakingvacuumduringthe run,asthereareonlysevensputteringgunsinthesputteringchamber,andeightmaterials areneededforthistypeofmulti-layergrowth.Beforeopeningthesystemtoswitchguns, 147 itwouldthereforebenecessarytodepositanAucappinglayertoprotectthemultilayer. Despitetheaddedsteps,thesmoothergrowthcouldpromotebettercharacteristicsofthe ferromagneticlayers,perhapsevenenoughtolowertheswitchingoftheNiFeslightly.As anotheroption,inordertomaintainan insitu fabricationofalllayers,[Nb/Au] n multilayer samplesarealsobeingconsidered.Whiledemonstratingsmoothergrowthcomparedto Nb(100),sampleswithmultilayersof[Nb/Au] n arenotassmoothasthosewith[Nb/Al] n . However,astheirgrowthdoesnotrequirebreakingvacuum,theystillmayyieldsmoother samplelayers.Samplesofbothtypesofmultilayersarecurrentlybeingmadeasofthetime ofwritingthisthesis. AnotheroptionforfutureworkistoremovetheNiF'layerandreplaceitwithNiFe.As mentionedpreviously,rotatingbothspin-mixerlayerswouldtheoreticallyenhancethe ratio.Atonepointduringtheexperiment,thishopewasabandoned,butwiththeprocedure andfabricationthoroughlyoptimized,thismaystillbeworthpursuing.Inaddition,NiFe hasslightlymoretripletgenerationcapabilitiesthanNiFeMo,andwithathinnerSAF,the criticalcurrentmaybecomelargeenoughtomeasure. OneaspectIparticularlyinterestingisthepotentialuseofthisswitchinginpractical applications.Havingan\on"andtripletstatecanbeinterpretedasa\0"or\1"state asabinarymemory.However,Josephsonjunctionswiththisgeometry,i.e.acircle,don't havebi-modalstability;thereisnopreferentialdirectionintheplaneoftheferromagnet,so theinitialstatecouldpointanydirection.Ifwecouldengineerasystemthathadapreferred initialdirection,thistypeofapplicationcouldbecomerealized. 148 APPENDIX 149 RangeofSpin-TripletPairCorrelations inS/F/FSystems Quiteabitofmytimeasagraduatestudenthadbeenspenttryingtodeterminethelength- scaleofthelongrangespin-tripletpaircorrelationsinS/F/SJosephsonjunctions.Tomy dismay,allresultsfromthisworkhavebeenfarfromconclusive.Thisappendixwilldescribe thatwork,thetroublesencounteredalongtheway,andwheretheprojectisnow.Hopefully thisdocumentwillhelpanyfuturestudentwhomaystartasimilarundertaking. Therearetwomainaspectsofthisproject:measurementsofsupercurrentinlateral geometryJosephsonjunctionsandmeasurementsoflongrangeproximitytinferromag- neticwires.TheinitialintentwastomeasuretheJosephsonjunctioncriticalcurrentas afunctionofjunctionlength,elymappingoutthespin-tripletdecaylength.The proximitymeasurementswerecarriedoutbymeasuringthechangeinresistanceas afunctionoftemperatureandwirelength.Fromsuchdata,thecoherencelengthofspin- tripletpaircorrelationsinCoshouldbedeterminable.Whiletheproximityproject onlyemergedfromthefrustratingresultsoftheJosephsonjunctions,thesetwoprojectswill bediscussedsimultaneouslythroughoutthisappendix.Inlieuofachronologicalstory,I hopethisorganizationallowsforbettercomprehensionoftheworkandresultsdescribed. 150 Motivation ItiscommonforthoseinvestigatingS/Fsystemstoclaimthatapairofelectronsinaspin- singletstatewith m s = 1experienceaferromagneticmaterialasanormalmetalbecause bothelectronsareinthesame(majorityorminority)spinband(seeSections2.3.2and2.5), aclaimIhaveevenmademyselfthroughoutthisthesis.However,thereislittleexperimental evidenceastothespatialextenttowhichspin-tripletcorrelationswillpersist.Theexchange energybetweenbandsmaynolongerbeanissue,butotherphenomenascattering, spin-orbitscattering)maystillplayaroleintransport.Therefore,thisprojectwasintended todeterminethelengthscalespin-tripletpaircorrelationscantravelbeforedecaying. Conceptually,determiningthislength-scaleshouldbenomorethangrowinga thickerCo/Ru/CoSAFinthemiddleofaS/F'/F/F"/Sjunction.However,anumberof complicationsarisefromthisattempt.AsCoisgrownthicker,itslatticeorientationswitches fromface-centeredcubic(FCC)tohexagonalclosedpack(HCP),causingthemagnetization tobecomeinhomogeneous[87].InadditiontoalteringthemagnetizationoftheCo,the transitionmaycausemorescattering,shorteningthelengthcompared toCowithoutalatticetransition,evidentinthedatain[90].Thelengthis asthedistanceinamaterialanelectroncanbeforeitencountersaspin- collision,i.e.anupelectronbecomesadownorviceversa,andisgiveninthedirty limitby l F sf = q D F ˝ F sf (A.1) where D F istheconstantinaferromagnetand ˝ F sf isthemeantimebetween events[88].Thisisdependentonanumberoffactors,includingmaterialand 151 temperature.ThelengthinCohasbeenmeasuredatvarioustemperatures, withresults(innm) =38 12( T =300 K )[89] l F sf =59 18( T =77 K )[89] 40( T =4 : 2 K )[90] : (A.2) Thevalueat4.2Kisanestimateextrapolatedfromthinsamplesduetoanapparentchange inlengthassamplesweregrownthicker[90]. IssuesarisingfromfabricatingjunctionswiththickCohavebeenobservedwhenmea- suringtheFraunhoferpatterns[91].Usingthegeometrymentionedthroughoutthisthesis (seeSection3.1andFigure3.1),criticalcurrentmeasurementsinjunctionswith d Co < 15 nm,andthereforetotalthickness D Co < 30nm,demonstratedcleanFraunhoferpatterns. However,at D Co 30nm,thepatternsbecametoomessytodetermineareliablepeak valueofthecriticalcurrent. Alternativematerialswerealsousedintheverticalgeometry,includingCo/Nimultilay- ersthathaveperpendicularanisotropy,i.e.withthemagnetizationpointinginthesame directionasthecurrent(FigureA.1)[92].Inthosejunctions,theFraunhoferpatternswere stillverydistortedinsampleswith18[Co/Ni]layers,oratotalcentralferromagnetthickness of11.2nm.Fromtheseresults,itwasclearthatanewgeometrywouldneedtobedeveloped todeterminetherangeofspin-tripletpaircorrelationsinferromagneticsamples. 152 FigureA.1 Cartoonofperpendicularmagneticanisotropy. TheFlayerinthisJosephson junctiondemonstratesthisanisotropyasitsmagnetizationpointsoutofplane. SampleGeometryandFabrication InordertokeepCothin,alateral,orplanar,geometrywaschosenforthisproject.Thisis representedinFigureA.2.Seeingnoreasontochangethematerial,bothJosephsonjunction andproximitysampleswerecreatedwithCoasFandNiasthespin-mixinglayer(s). TheproximitysampleswerealsotestedwithNiasabasewire.SeparateFandS layerswereisolatedfromeachotherbyCuorAuspacinglayers.Thestructurefor thesejunctionshadawireof[Co/Au](x)withjunctionlengthsbytwoelectrodesof Ni(1.5)/Cu(5)/Nb(55)forJosephsonjunctionsamplesanda[Co/Au](x)orNi(x)wirewith Ni(1.5)/Cu(5)/Nb(60)contactforproximitysamples.Forthebasewires,boththeCo andAuweregrown15nmthickandtheNiwas30nm,butduetothelateralgeometry, thedistancebetweentheelectrodes(x)determinesthelengthofthejunction,sohasbeen writtenaboveas[Co/Au](x)orNi(x).Forproximitysamples,thelengthofthewire wastypically1or5 whiletheelectrodesintheJosephsonjunctioncouldbeasnarrow 153 (a) (b) FigureA.2 ImagesoflateralJosephsonjunction. IntheSEMimage(a),thebrightvertical wireistheCo/AubilayerandthedimmerhorizontalwiresaretheNi/Cu/Nbmultilayer. Acartoondepictionofthesideviewisshownin(b).TheNi/Cu/Nbelectrodesthe junctionlength,L. as50nm,buttypicallywere100-150nm. FigureA.3 SEMimageofalateralgeometryproximityctsample. Thehorizontalwireis theCo/Aubilayer,attachingto4leadsfora4-terminalmeasurement.Theverticalwireis theNi/Cu/Nbmultilayer. Considerations Manyiterationsofthesppatternswereattempted,culminatingwithFiguresA.2and A.3.Throughcleverchoicesofgeometry,manypotentialissuesareavoided,someofwhich arelistedbelow: 154 1.Cothicknessisalongthelengthofthewire,sotherearenostructuretransitions andthereforenodirectlengthlimitationsinJosephsonjunctions. 2.Fabricatingnarrow,longwirescreatesanaturaluniaxialshapeanisotropyasmentioned inSection2.1.1. 3.With ~ M pointingalongthewire,thecurrentandmagnetizationbecomecollinear. This,inprinciple,avoidsalargemagneticcontributiontotheFraunhoferpatterns fortheJosephsonjunctionsamples. 4.Beingabletothemagnetizationdirectionthroughwirefeaturesallowsusto createanaturalnon-collinearity,moresponewithorthogonalorientation, betweenneighboringferromagneticlayers. 5.Withawiderbasewireawayfromthejunction,thereisalowerprobabilityofbreaks alongthewire.Thisallowsfor4terminalinterfaceresistancestobemeasured. Toavoidadverseatdomainwalls,fabricationwasdoneinawaytopromotesingle domainwires.Todeterminethedomainstructure,Co,Co/Au,andCo/Cu/Auwireswere fabricatedandmeasuredwithMFM.ThisworkwasdonewithCharlesMoreauatAlbion Collegeinamannerverysimilartoworkdoneinhispreviouspaper[93].Co/Aumultilayers demonstratedthehighestlikelihoodofsingledomainstructureinthewire,especiallyafter magnetization.ThisresultisfavorablebecauseitalsocapstheColayer,preventingitfrom oxidizingbetweenfabricationsteps.WealsotestedsampleswithNiasabaselayer,andthe MFMwastestedforthoseaswell.They,too,showedsingledomainproperties(FigureA.4). Aninitialworrywasthatanypositiveresultscouldbeattributedbycriticsassuper- currentwingthroughtheAulayer.Toaccountforthat,sampleswithandwithoutthe 155 FigureA.4 MFMresultsofNinanowires. Thedarkandlightdotsattheendsofthewires, butnowhereelse,demonstratethesingledomainnatureofthewires.Thefringepatternin theimageisanartifactoftheAFM. 156 spin-mixingNilayerswouldbecreated.Similartosamplesusedintheinitialspin-triplet discovery(Section3.1),thereshouldbealargeincriticalcurrent(forJosephson junctions)oranappreciableinresistance(forproximitysamples)betweensamples withandwithoutNi. However,theseconcernswerelargelyquelledwiththeoretical[94]andexperimental[95] resultsregardingferromagnetic-normalmetalbilayers.Whenanormalmetalisgrownona ferromagnet,thenormalmetaldecreasestheexchangeenergyoftheferromagnet.However, thestructureasawholeactsasaweakferromagnetwithareducedexchangeenergy.This wasdemonstratedwithS/N/Scriticalcurrentmeasurements(N=Cu)thatwerecompared toS/F-N/Scriticalcurrents(F=Fe,N=Cu)[95].Theresultsshowedadramaticdecrease incriticalcurrent,reminiscentofthespin-singletsuppressioninS/F/SJosephsonjunctions. Therefore,inoursamples,eventheAuchannelintheCo/Aubilayerisspin-polarized, limitingtherangeofspin-singletpaircorrelations. Fabrication Thesamplefabricationsteps,demonstratedinFigureA.5,arethesamebetweenthetwo geometrieswithonlythepatternwrittenduringtheEBLstepchanging.Theprocedure follows: 1.SpincoatwaferwithLOR-5BandS1813bi-layer 2.Exposewafertoopticalpattern,thelargepadsandleads 3.EvaporateTi/Auandliftresist 4.Protectanddicewafer 157 (a)Depositthebasewire (b)Magnetizethebasewire (c)Depositthetopleads FigureA.5 Cartoonsoflateralgeometryfabricationsteps. Colorsfrombottomup:Dark blue{Sisubstrate;Lightpurple{Co;Yellow{Au;Blue{Ni;Orange{Cu;Lightblue{ Nb. 5.Liftresistandcleanchip 6.SpincoatMMA/MAAEL9andPMMAC2bi-layer 7.ebasepatternwithEBL 8.Sputterbasemultilayer,Co/Au,andliftresist 9.ProtectthechipwithS1813andmagnetizesample 10.Liftresistandcleanchip 11.SpincoatPMMAC2mono-layer 12.netoplayerpatternwithEBL 13.Lightionmill( ˘ 2nm) insitu ,sputtertopmultilayer,Ni/Cu/Nb/Au,andliftresist DetailsofmostofthesestepscanbefoundinChapter4.Initially,theentirestructurewas tobemadeintwoEBL/sputteringsteps,withthetoplayersbeingpatternedassmallwires atthejunctionbutextendingtothelimitsoftheJEOL,writingattmin ordertopatternthelargepadsforsamplemountingaswell.However,theinitialattemptsat thisprocessrevealedamajorissue:the5nmAu(atthetimekeptthintopreventachannel forsupercurrent)didnothaveenoughcontrastintheSEM,makingalignmentimpossible. 158 Instead,aphotomaskwasdevelopedwhichcontainedalignmentmarks,leads,andmounting pads,doneonawaferscaletoexpeditethefabricationprocess. Themagnetizationstepwasaddedtoincreasetheprobabilityofcreatingsingledomain wires.Becausethisisdoneoutsideofthecleanroom,chipsarespunwithS1813 protectingitfromdustandothercontaminants.Thechipsareplacedinanelectromagnet whichisbroughtupto ˘ 150mT.Thechipsarethenbroughtbackintothecleanroomwhere theresistisremovedinacetone,andthechipsarereadyfortherestofthefabricationprocess. Theionmillstepbeforethesecondsputteringrunisdonetoensuretheinterfaceisclean. Forthisprojectespecially,theresistanceattheinterfacewasamajorconcern.Becausethe materialsarenotsputtered insitu ,thereisthepossibilityofcontamination,forexample fromresistresidue,attheinterfacethatasimpleionmillingstepcanalleviate. FabricationIssues Thegeometryattemptedinthisprojecthadneverbeenusedbeforeandwasthereforede- velopedfromscratchduringthisprocess.Assuch,manycomplications,adjustments,and restartsfollowed,asonewouldexpect.Ihavealreadytouchedononeofthem,inthatthe initialpatternhadnowaytoalign,andthereforeanentirelynewprocesswasdeveloped toincorporateopticalbaseleads.Whatfollowsareothermaincomplicationsthatarose, althoughIdonotclaimthisisanexhaustivelist. Sidewall Thesputteringsystemisanuncollimatedsystem,asdiscussedinSection4.1.2.2.However, theextentofthespreadanditshadn'tbeenquanuntilverynarrowwireswere 159 (a)Sidewallbuildupduetowidesputtering angle. (b)Sidewalleliminatedthroughcollima- tion. FigureA.6 Imagesofsidewallissues. sputtered.WhilenotanissueforsoftmaterialssuchasAu,sputteredNbwiresdemonstrated alargesidewalldeposition,sometimescollapsingonthejunctionbutoftenstandingrigid. ThisisshowninFigureA.6a.Thishappensbecause,duetotherelativethicknessesofthe upperandlowerresistlayersinthebilayer,thealloweddepositionanglesofthesputtered materialaresowidethatresistdepositsontheedges.Acartoonrepresentationofthisis demonstratedinFigureA.7a. Toeliminatethesidewall,thesputteredmaterialmustbebettercollimated.Thiscan bebychangingtheresistasdemonstratedinFigureA.7b[96],orbyaddinga mechanicalcollimator,along,narrowtubethatisplacedoverthesampleduringdeposition thatgeometricallylimitsthespreadofmaterial(FigureA.7c).Whileattemptstomanipulate theresistprwillappearagain,mechanicalcollimationwasaddedtotheprocessing, eventuallybecomingapermanentaspectofsputteringthebaselayer.FigureA.6bshows thetheofcollimatingduringthesputteringstep. 160 (a)EL9/C2bilayer (b)EL6/C4bilayer (c)EL9/C2bilayerwitha collimator FigureA.7 Cartoonsofresistpr Theblacklinesrepresenttheanglesallowedbasedon resistgeometry.(a)demonstratesuncollimatedsputtering,while(b)showsself-collimation fromtheresist(c)demonstratestheabilitytoreducespreadfromamechanical mask. 161 (a) (b) FigureA.8 Imagesofjunctionconcerns. Inthetopdownview(a),therearefaint whitelinesbetweenthetopleads,leftandrightofthejunction.Theselinesaretheedgesof thetopleads,andareincontactatthejunction.Thesearemoreclearintheangledview, which,inthissample,showsseparationbetweentheelectrodes,buttheiredgesare stillunclear. Junction Eventhoughthecollimatorsreducedthespreadenoughtoeliminatesidewallbuildup,there wasstillabitofspreadinthedepositedmaterial.Whilethisisusefulforthebaselayerdepo- sition,ofwhichrigidedgescouldcauseproblemswhengrowingthetoplayer,theelectrodes depositedduringthestepneedtobewellWithoutdistinctedges,howcanwe knowwhatthelengthoftheJosephsonjunctionis?Inaddition,howcanweguaranteethere isn'tashortintheelectrodes?ThisconcernisdemonstratedinFigureA.8. Atthispoint,self-collimationfromresistwasconsideredandattempted.Various combinationsofresistswereused,twoofwhichbeingattemptedthemost:MMA/MAAEL9 withPMMAC2,theoriginalcombination,andMMA/MAAEL6withPMMAC4,which bothdecreasesthebaselayerthicknessandalsoincreasesthatofthetoplayer.Onceagain, thesecanbeseeninFigureA.7.Withtheresisttakingcareofthecollimation, themechanicalcollimatorscouldberemoved,increasingthesputteringratedramatically. Whileinprinciplethisshouldhaveworked,theresultingsamplesoftendemonstratedodd 162 (a) (b) (c) FigureA.9 Imagesofresistdeformation. Allimagesaretakenaftersputteringbutbefore In(a),weseetheshapeandstructureofthejunctionwithoutanydeformation.In (b)wecanseethewireshavebeenpinched,theextentofthebucklingvisiblein(c),which wastakenata45-degreeangle. deposition,andsamplereproducibilitybecameamassiveconcern.Whilenotimmediately obvious,thiswascausedbydeformationoftheresistwheninthesputteringchamber.At somepointduringsputtering,strainsduetotemperatureexcursions,eitherfromcooling ormorelikelyoverheating,causedbucklingbehaviors,asseeninFigureA.9.Asuggested solutionwastoutilizeaPMMAmonolayer[97].Althoughthiseliminatestheundercut, potentiallyleavingrigidedgesinthedepositedmaterial,itallowsforveryprecise ofthesputteredlayer.Afterattemptingthis,theremovedanyofourconcernsand progresscouldcontinue. 163 (a)Plotofmillratevsbeamvoltage. (b)Plotofmillratevs.position,relativetoac- celeratorvoltage. FigureA.10 Plotsofionmillcharacteristics. Themillraterelativetobeamenergyisplotted in(a),demonstratingthelowerthresholdofappreciablebeamvoltage.Thebeamis plottedin(b),demonstratingtheamountofspreadpresentinthemillrelativetoaccelerating voltage.Themillrateswithanacceleratingvoltageof50Vweremeasuredtwice,represented bytwosetsofdatain(b). MillDamageandSputteringAlignment Initially,sampleswereionmilledwiththesameparametersusedtomillandnepillar dimensions.However,samplesmadethiswaywereunsuccessful,andwefoundevidenceof damagetothebasewireduetotheionmill.Furtherreadingledustorealizethebeam energyplaysamajorroleinhowmillingtakesplace,aswellasthepotentialdamagecaused [98],soweneededtocharacterizeourmill.Thisincludedthebeamenergyaswellasthe beam Todeterminetheacceptablerangeofbeamenergy,wemeasuredthemillingratevsbeam voltage,displayedinFigureA.10a.Wewantedtomillatlowenergiestoavoidunnecessary damage,butnotrightatthethreshold,limitingtheofthecleaningentirely.Typically thismeantthemillwassetbetween125and135V,butvoltagesashighas175Vhadbe usedattimes. 164 Wethenhadtodeterminetheofthemillwhich,accordingtothemanual[99],is largelybytheacceleratingvoltage.Tomeasurethis,werotatedthesampleholder overthemillwhiletheFTMmeasuredthemillingrateforeachpositioninthesteppingmotor. TheresultsareplottedinFigureA.10b.Fromthis,itwasdeterminedthattheaccelerating voltageshouldbeturnedupto100V,yieldingaslightlybroaderMoreimportantly, becausethebeameformillingissonarrowweneededtorecalibratethesputtering chamber,whichalsoimproveddepositionduetothelimitinganglesofthecollimators.Todo this,RezaLoloeeandImeasuredthepositionofeverysampleholder/targetguncombination andreprogrammedthesystemaccordingly.ThiswasdoneonceagainbyRezaandVictor Aguilarwhenanewsputteringprogramwasdeveloped. MeasurementSetup QuickDippers Liketheotherworkdescribedinthisthesis,thelateralgeometrysampleswerealsodipped intoaliquidheliumstoragedewaronvariousQuickDipperprobes.However,theprobes usedforthesemeasurementshavebeenhighlymotoobtainmuchlowertemperatures, andareexplainedhere.Astherearemanymanuals,checklists,andprocedures,written byWilliamPratt,RezaLoloee,JosephGlick,andmyself,accessibletoanyoneusingthese systems,Iwillforgodiscussionsonmountingandoperationanddiscussonlytherelevant aspectsofeachdipper.However,Iwillmentionthatthesesamplesaremuchmoresensitive tostaticdischarge,andthereforegroundingstrapsmustbewornandtheprobemustbe groundedwhenmounting. 165 QuickDipperV(QD-V)isasemi-permanentreattachedtoaliquidheliumstorage dewar.Itiselyahollowcylindricalshell,isolatedfromthedewarwithavacuum jacket,thatisplaceddeepintothedewar.Itcanbebyopeningacapillarytube, allowingliquidheliumtowfromtheoutsideinwhilethelevelismonitoredbyCernox resistorsthatchangeresistancewhencoveredwithhelium.QD-Visalsoattachedtoa roughingpump.Throughevaporativecoolingwhilebeingpumpedon,thetemperatureof theliquidheliumdrops,reachingaminimumtemperatureof ˘ 1.1Kinthissystem.This temperaturewasoncemeasuredwithaconductancebridge,althoughrecentlyaLakeshore Model350TemperatureControllerhasreplacedit. QuickDipperVI(QD-VI)isaverybasicprobethathas6voltageleadsand6current leadsrunningfromtheoutsidetothesample.Ithas,however,beendesignedtobedipped intoQD-V.Byusingthiscombination,itispossibletomeasuresamplesfrom1.2Kto4.2 Kreliably. QuickDipperVII(QD-VII)isaremarkableprobethattookyearstoengineer.Itfeatures avacuumcananda 3 Hepot.BydippingitintoandloweringthetemperatureofQD-V, the 3 Heliqueabovethesampleholder.Byusingacharcoalpillonamagneticarm,this liquid 3 Hecanbepumpedon,decreasingthetemperatureofthesampledownto0.32K. Measurement ThemeasurementtechniquesusedfortheJosephsonjunctionsandtheproximity samplesquiteabit,andassuchIwilldescribethemseparately.However,all measurementsare4terminal,utilizingalock-intoobtainminimalnoise.Whiletheinitial proximitymeasurementsweretakenbyhand,LabVIEWprogramshavebeendeveloped byVictorAguilartoautomatethedatacollection.TheJosephsonjunctionprogramswere 166 (a) (b) FigureA.11 ImagesofQD-VII. Theentiredipperisshownin(a),includingthesystem requiredtopumpoutthevacuumcan.Acloseupofthemountingareaisshownin(b). 167 writtenandmobyseveralpastandpresentgroupmembers. TheJosephsonjunctiondataweretakeninastandardtialresistancemeasure- ment.Usingtheconceptofloadlines[100](FigureA.12),voltagessuppliedbyaTektronix AFG3022BFunctionGeneratorandSRSModelSR850DSPLock-inAmwereadded together,fromwhichacurrentsourcewascreatedviaaballastresistor(typicallysetto10 fortheseexperiments).ThecircuitisshowninFigureA.13.Thefunctiongeneratorsup- pliedthesystemwithaveryslow(typically2mHz)sawtooth-patternedvoltage,ely supplyingaDCcurrenttothesample.Ontopofthis,afast(typically98Hz),smallAC signalwasaddedfromthelock-inThismeasurementschemeallowsustomeasure dV dI vs I curves.ely,thiscanbethoughtofasmeasuringtheinstantaneousresistance ofthesampleateverycurrentincrement.InrelationtoFigureA.12,thiscanbepicturedas slowlyslidingtheredlinealongthex-axis,measuringtheslopeofthecurveateachpoint.If desired,amoretraditionalI-Vcurvecanbeobtainedbyintegratingtheoutputwithrespect tocurrent. Theimportantdataforproximitychipsarethatofresistancevstemperature( R vs T ).Asthesuperconductingpaircorrelationspenetrateintotheferromagneticwire,the resistanceofthewireshoulddrop.Thelongerthecoherencelength,themoretheresistance shouldchange.Therefore,themeasurementdoesnotneedtobeanymorethanastandard4 terminalresistancemeasurement.However,becausetheissmallrelativetothenormal resistanceofthewire,aratiotransformerisused(seeSection6.2.2andFigure6.8).By subtractingthenormalresistanceofthewire,amuchmoreresolvedmeasurementofthe changesinresistanceasanoftemperaturearepossible. 168 FigureA.12 Diagramofvariousloadlineschemes. Thefollowingexamplesareformeasuring dV dI vs I inaJosephsonjunction:if R B ˝ R s (blue),detailsinthenormalstateandnear thecriticalcurrentarewellresolved,butthesuperconductinggapisalmostentirelymissed; if R B ˛ R s (green),measurementswithinthesuperconductinggaparewellresolved,but notthedetailsnearthecriticalcurrent.Agoodbalanceisobtainedfor R B ˇ R s ,whichcan measurepointswithinthegap,nearthenormaltransition,andwhilethesampleisnormal. FigureA.13 CircuitdiagramforlateralgeometryJosephsonjunctionmeasurements. The ballastresistor( R B )turnsthevoltagesfromthelock-inandfunctiongeneratorintoAC andDCcomponentsofthecurrent,respectively.Theoutputismeasuredwiththelock-in. Often,nopre-ampwasused,sog=1,althoughapre-ampwithg=100wasusedonoccasion. 169 (a) (b) FigureA.14 S/N/SJosephsonjunctionmeasurements. dV/dImeasurement(a)andinte- gratedI-Vcurveofthesamedata(b).WidthoftheAuwireis ˘ 100nm,withNbelectrode separationof ˘ 100nm; T =0 : 33K. Data Insamplesmadewithnoferromagneticmaterial,themeasuredcriticalcurrentbecamelarger withdecreasedtemperature,asexpected.ThisisshowninFiguresA.14aandA.15.These resultsleftustthatwecouldfabricatejunctionswithferromagneticlayersinthis geometry.Addingtmaterialshouldnotchangethat. However,changingthematerialhadaverylargeInsampleswithoutF'layers, weobservednocriticalcurrent(FigureA.16a).Thisisnotasurprise,giventhatwewere generatingnospin-tripletpaircorrelations.WhenweaddedtheNispin-mixerlayer,com- pletingtheS/F'/F/F'/Swestillobservednopositiveresults(FigureA.16b). Thedatashownarenotuniquetotheparticularsample.Measuredsamplesacrossvarious sputteringrunsandfabricationtechniques,madeoveryearsoffabrication,allshowsimilar trends. Someothergroupshaveattributedsimilartrendsintheirdatatochargeimbalance,a consequenceofinjectingelectronsintoquasiparticlestatesabovethesuperconductinggap 170 FigureA.15 TemperaturedependenceonS/N/Scriticalcurrent. WidthoftheAuwireis ˘ 100nm,withNbelectrodeseparationof120nm.Temperaturesarelistedfromtheinside curveout:Lightblue{3.55K;Darkblue{2.55K;Yellow{1.91K;Green{1.55K;Purple {1.22K. (a)S/F/SJosephsonjunctionmeasurement. (b)S/F'/F/F'/SJosephsonjunctionmeasure- ment. FigureA.16 S/F/SandS/F'/F/F'/SJosephsonjunctionmeasurements. In(a),thereisno spin-mixer,socriticalcurrentisneitherexpectednorobserved.AddingNispin-mixerlayers (b)shouldshoweithercriticalcurrentorproximity(lowerresistanceatlowdrive). However,thisisnotobserved.S-Nb;F-Co/Aubilayer;F'-Ni. T =0 : 34Kforbothsamples. 171 [101].Theirresultsmimicourtrendratherclosely.However,wedonotbelievethe seeninoursamplesisduetothis.Chargeimbalanceismostimportantnear T c orunder strongcurrentdrive.ThecurrentthatthisoccursatinourS/F'/F/F'/Ssamplesis muchlowerthanthecriticalcurrentinourS/N/Sjunctions. Thedatafromproximitysamples,whichshouldbemoreclearastheyareless complicatedsystems,wereevenmoreconfusing(FigureA.17).Fabricationofthesesamples weredonethesamewayasintheJosephsonjunctions,sometimeseveninthesamesputtering runs.Despitethis,whileanS/N/SJosephsonjunctiondemonstratedsupercurrent,the proximitymeasuredinanormalmetalwasunreliable.TheS/Njunctionshownin FigureA.17ahasthelargestdecreaseinresistance,about10%by0.3K.However,the superconductingleadcoversabout10%oftheAuwire,whichopensapathwayforcurrent totravelthroughthesuperconductor,asdemonstratedinFigureA.19b.Thisregionof superconductivityshoulddroptheresistanceofthatlengthto0.Takingintoaccountthis \shorting"throughthesuperconductor,inadditiontotheproximityweexpecttosee, wethereforeexpectthedecreaseinresistanceoftheentirewiretobemuchlargerthanwe measureoursamples. Theresultsinsampleswithferromagneticmaterialareevenlessclear.Inthosewith nospin-mixerlayer,i.e.S/Fsamples,wehavemeasuredaverysmalldropinresistance, lessthan1%(FigureA.17b).WhenaddingasecondFlayerasaspin-mixer,thedata aremysteriousandinconsistent.AsshowninFigureA.17c,weobserveexcursionsfrom thenormalresistanceofthewireathighertemperatures( ˘ 9K)whencomparedtoother samples( ˘ 5K).Theredoesnotseemtobeaconsistentdropinresistance,butrather multiplejumps,implyingthepresenceofmultipleTheresistancealsoincreasesand plateausbelow ˘ 4K.Inonesample(FigureA.17d),theresistanceneverdecreasesbutrather 172 (a)Au/Nb{1 longx100nmwide (b)Co/Au/Nb{1 x100nmwide (c)Co/Au/Ni/Cu/Nb{1 x100nm wide (d)Co/Au/Ni/Cu/Nb{5 x100nm wide FigureA.17 Datafrom Rvs:T proximityctmeasurements. anincreaseismeasuredasthetemperaturedecreased(FigureA.17d).Theseresultsarenot wellunderstood. Othergroupshavereportedresistancechangesthataremuchlargerthanthosemeasured inoursamples.Ina40nmlongConanowire,onegroupobserved0resistance,i.e.aJoseph- sonwhenmeasuringthewirewithsuperconductingWelectrodes[102].Eliminating anyJosephsont,theyreplacedWwithPtelectrodes,andaddedasmallWcontactbe- tweenthevoltageleads,inducingsuperconductingproximityInthesesamples,they reporteda\normalmetalcoherencelength"ofhundredsofnminCo.Itshouldbenoted thatthesesamplesdidnothaveaspin-mixinglayer,butitislikelythattheFIBusedin 173 thefabricationprocesscreatedaspin-mixinginterface.Similarresultswerealsoreported recently,againinConanowiresincontactwithanarrowWstrip,about220nmwide[103]. Althoughthesuperconductorwasincontactwiththeferromagneticwireover ˘ 6%ofits length,thedropinresistancebetween5.2Kand2.4Kwas22%.Thisresultisnotonly muchlargerthaninourS/Fsamples,butlargerthaninourS/Nsamplesaswell. WeeventriedtomeasuretheproximityectintheJosephsonjunctionsamples,the datafromanS/F'/F/F'/SsampleareshowninFigureA.18.Thetrendsareonceagain similartothoseobservedbyothergroups:theincreaseatverylowtemperatureissimilar tothere-enteranceobservedinthe90s[104];thesharppeaknear6Kissimilarto anmeasuredinotherConanowire/superconductorsystems[102],whichwasclaimed topossiblybeduetochargeorspinimbalance,butnotconclusivelyknown.Webelievethe rapidchangeinresistancebetween6and7Kisduetonon-uniformcurrentdensitythat emergeswhentheelectrodesarenormal,i.e. T>T c .Withthegeometryasitis,when theNbisnon-superconducting,thepaththecurrenttakesmaynotoverlapentirelywith thevoltageleads,droppingthemeasuredresistancesubstantially.Overall,whilethetrends presentinourdataaresimilartothoseseeninotherwork,wearenotconvincedthe arethesame. DiscussionandOutlook Theresultsoftheseprojectsareconfusingand,ifImaysay,disappointing.Theresults obtainedthroughfabricationofS/N/Ssamples,especiallythosethatdemonstratetypical temperaturedependenceoftheJosephsoncriticalcurrent,implythatoursamplefabrication isoptimized.However,noJosephsonjunctionsampleswithferromagnetsdemonstrated 174 FigureA.18 ProximityctinJosephsonjunctionsamples. successfulresults.Thesamecanbesaidabouttheproximityresults. Someothergroupshaveseensignaturessimilartooursintheirdata,claimingchargeor spinimbalanceasapossibleexplanation.However,wedonotbelievethisisthecaseinour samples,atleastnotthefullpicture.Intheend,thereasonforthesamplesbehavingas theydoisstillamysterytous.Despitereasontothinkthattheyareclean,theproblem likelyarisesfromsomeissueattheinterface.Whatthatissueis,however,isunknown. Whatdoesthatmeanforthefutureofthisproject?Eithermaterialorgeometrychanges maybeinourfutureyet.ImustacknowledgetheworkdiscussedbeforebyGolikova etal. regardingF-Nbilayers[95].Theirsamplesweregrownwithane-beamevaporator, equipmentthatwedonothaveaccesstoatourfacility.However,slightadjustmentstotheir patterning,entirelybasedaroundawaytointroduceanon-collinearmagnetization layer,couldbeallthatisrequiredtomakelateralS/F'/F/F'/Ssamples. 175 Separately,ifwecouldndamaterialthathasperpendicularanisotropyandisagood spin-tripletgenerator,wecouldavoidfabricationandinterfaceissuesbydoingallofthe depositioninonestep.Iftheentiremultilayerisdepositedinthesamewayourpillarsare, wecouldmillasmallgapinthewire,millingthroughallofthelayersexcepttheF layer.Inthisway,wewouldhavenon-collinearity(F'layermagnetizationpointsoutof plane,basewiremagnetizationpointsalongthewire),wecouldcontrolthelengthofthe junction,andeveryinterfacewouldbedeposited insitu .However,thismagicmaterialhas notbeenfoundyet.PerhapsonedayastudentwillpickupwhereIleftandsuccessfully measuresupercurrentintheselateralgeometryJosephsonjunctionsusingthepossibilities above. Asfortheproximitysamples,alotmoretroubleshootingneedstobedone.Part oftheconcernisthatwemaybeseeingshortingthroughthesuperconductingwirewhere itcrossestheferromagneticwire.Toeliminatethisconcern,wecouldattemptt geometriesthatavoidwirecrossing,oneexampleofwhichisdemonstratedinFigureA.19c. 176 (a)Currentproximityctsamplegeom- etry (b)Cartoondemonstratingpossibleshort- ingofcurrentthroughsuperconducting wire (c)Potentialfutureproximitygeom- etry FigureA.19 Proximityctgeometries. Thecurrentgeometryisdisplayedin(a).Dueto thesuperconductormakingcontactwiththecurrent-carryingwire,shortingthroughScould causearesistancedropunrelatedtotheproximityect(b).Bymovingthesuperconductor awayfromthewire,asdemonstratedin(c),amoredirectmeasureoftheproximityis possible. 177 BIBLIOGRAPHY 178 BIBLIOGRAPHY [1]H.KammerlingOnnes, LeidenComm. 120b,122b,124c (1911). [2]J.Bardeen,L.N.Cooper,J.R.Sc Phys.Rev. 108 ,1175(1957). [3]J.OrensteinandA.J.Millis, Science 288 ,468(2000). [4]T.Tohyama Jpn.J.Appl.Phys. 51 010004(2012). [5]S.Hassan,Humaira,M.Asghar, Proc.ICCSN'10 559(2010). [6]High-performancecomputingresultscanbefoundathttp://www.top500.com. [7]M.W.Johnson etal. , Nature , 473 ,194(2011). [8]T.F.R˝nnow etal. , Science , 345 ,420(2014). [9]D.S.Holmes,A.L.Ripple,M.A.Manheimer, IEEETrans.Appl.Supercond. 23 , 1701610(2013). [10]K.K.LikharevandV.K.Semenov, IEEETransactionsonAppliedSuperconductivity 1 ,3(1991). [11]N.O.Birge,toappearin Phil.Trans.R.Soc.A. [12]A.I.Buzdin, Rev.Mod.Phys. 77 ,3(2005). [13]F.S.Bergeret,A.F.Volkov,K.B.Efetov, Rev.Mod.Phys. 77 ,4(2005). [14]C.Kittel, IntroductiontoSolidStatePhysics,SeventhEd. JohnWiley&Sons,New York(1996). [15]D.J. IntroductiontoQuantumMechanics,SecondEd. PearsonEducation, UpperSaddleRiver(2005). 179 [16]W.Heisenberg, Z.Physik 49 ,619(1928). [17]N.W.Ashcroft,N.D.Mermin, SolidStatePhysics .Holt,RinehartandWinston,New York(1976). [18]K.H.J.Buschow, HandbookofMagneticMaterials,vol.23 .North-Holland,Oxford (2015). [19]NicolaA.Spaldin, MagneticMaterials,FundamentalsandApplications .Cambridge UniversityPress,Cambridge(2011). [20]P.Weiss, J.Phys. 6 ,661(1907). [21]H.Barkhausen, Z.Phys. 20 ,401(1919). [22]L.Solymar,D.Walsh,R.R.A.Syms, ElectricalPropertiesofMaterials,NinthEdition . OxfordUniversityPress,Oxford(2014). [23]F.Bitter, Phys.Rev. 38 ,1903(1931). [24]H.J.Williams,F.G.Foster,E.A.Wood, Phys.Rev 82 ,119(1951). [25]L.N.Cooper, Phys.Rev. 104 ,1189(1956). [26]MichaelTinkham, IntroductiontoSuperconductivity,SecondEdition .DoverPublica- tions,Inc.,NewYork(1996). [27]W.Meissner,R.Ochsenfeld, Naturwissenschaften 21 ,787(1933). [28]A.F.Andreev, Sov.Phys.JETP 19 ,1228(1963). [29]P.Flude,R.A.Ferrell, Phys.Rev. 135 ,A550(1964). [30]A.I.Larkin,Y.N.Ovchinnikov, Zh.Eksp.Teor.Fiz. 47 1136(1964). [31]A.I.Larkin,Y.N.Ovchinnikov, Sov.Phys.{JETP 20 762(1965). [32]C.J.Lambert,R.Raimondi, J.Phys.:Condens.Matter 10 ,901(1998). 180 [33]W.Belzig,F.K.Wilhelm,C.Bruder,G.Schon,A.D.Zaikin, SuperlatticesandMi- crostructures 25 ,1251(1999). [34]E.A.Demler,A.B.Arnold,M.R.Beasley, Phys.Rev.B 55 ,15174(1997). [35]T.Muhge,N.N.Garifyanov,Y.V.Goryunov,G.G.Khaliullin,L.R.Tagirov,K. Westerholt,I.A.Garifullin,H.Zabel, Phys.Rev.Lett. 77(9) ,1857(1996). [36]J.S.Jiang,D.Davidovic,D.H.Reich,C.L.Chien, Phys.Rev.Lett. 74(2) ,314(1995). [37]T.Kontos,M.Aprili,J.Lesueur,X.Grison, Phys.Rev.Lett. 86(2) ,304(2001). [38]V.V.Ryazanov,V.A.Oboznov,A.Y.Rusanov,A.V.Veretennikov,A.A.Golubov, J.Aarts, Phys.Rev.Lett. 86(11) ,2427(2001). [39]T.Kontos,M.Aprili,J.Lesueur,F.Genet,B.Stephanidis,R.Boursier, Phys.Rev. Lett 89(13) 137007(2002). [40]B.D.Josephson, Phys.Lett. 1 ,251(1962). [41]B.D.Josephson, Adv.Phys. 14 ,419(1965). [42]R.P.Feynman,R.B.Leighton,M.Sands, TheFeynmanLecturesonPhysics,VolIII. Addison-WesleyPublishingCompany,Massachusetts(1965). [43]V.Ambegaokar,B.I.Halperin, Phys.Rev.Lett. 22 ,25(1969). [44]M.Giroud,H.Courtois,K.Hasselbach,D.Mailly,B.Pannetier, Phys.Rev.B 58(18) , R11872(1998). [45]M.D.Lawrence,H.Giordano, J.Phys.:Condens.Matter 11 ,1089(1999). [46]V.T.Petrashov,I.A.Sosnin,I.Cox,A.Parsons,C.Troadec Phys.Rev.Lett. 83(16) , 3281(1999). [47]F.S.Bergeret,A.F.Volkov,K.B.Efetov, Phys.Rev.Lett. 86(18) ,4096(2001). [48]A.Kadigrobov,R.E.Shekhter,M.Jonson, LowTemp.Phys. 27(9-10) ,760(2001). [49]Y.V.Fominov,A.F.Volkov,K.B.Efetov, Phys.Rev.B 75 ,104509(2007). 181 [50]T.Champel,T.ofwander,M.Eschrig, Phys.Rev.Lett. 100 ,077003(2008). [51]A.F.Volkov,F.S.Bergeret,K.B.Efetov, Phys.Rev.Lett. 90 ,117006(2003). [52]M.Eschrig,Tofwander, NaturePhysics 4 ,138(2008). [53]MatthiasEschrig,\Spin-polarizedsupercurrentsforspintronics." PhysicsToday 64 , 43(2011). [54]M.Houzet,A.I.Buzdin, Phys.Rev.B 76 ,060504(R)(2007). [55]R.S.Keizer,S.T.B.Goennenwein,T.M.Klapwijk,G.Miao,G.Xiao,A.Gupta, Nature 439 ,825(2006). [56]I.Sosnin,H.Cho,V.T.Petrashov,A.F.Volkov, Phys.Rev.Lett. 96(15) ,157002 (2006). [57]T.S.Khaire,M.A.Khasawneh,W.P.Pratt,Jr.,N.O.Birge, Phys.Rev.Lett. 104 , 137002(2010). [58]M.A.Khasawneh,W.P.Pratt,Jr.,N.O.Birge, Phys.Rev.B 80 ,020506(R)(2009). [59]C.Klose etal. , Phys.Rev.Lett. 108 ,127002(2012). [60]Y.Wang,Ph.D.Thesis,MichiganStateUniversity,UnitedStates(2010). [61]Y.Wang,W.P.Pratt,Jr.,N.O.Birge, Phys.Rev.B 85 214522(2012). [62]C.Klose,BachelorsThesis,UnivatKonstanz,Germany(2010). [63]V.A.Oboznov,V.V.Bol'ginov,A.K.Feofanov,V.V.Ryazanov,A.I.Buzdin, Phys. Rev.Lett. 96 ,197003(2006). [64]B.Baek,W.H.Rippard,S.P.Benz,S.E.Russek,P.D.Dresselhaus, Nat.Commun. 5 ,3888(2014). [65]B.M.Niedzielski,unpublisheddata. [66]B.M.Niedzielski,E.C.Gingrich,R.Loloee,W.P.Pratt,Jr.,N.O.Birge, Supercond. Sci.Technol. 28 ,085012(2015). 182 [67]J.W.A.Robinson,J.D.S.Witt,M.G.Blamire, Science 329 ,59(2010). [68]D.Sprungmann,K.Westerhold,H.Zabel,M.Weides,K.Kohlstedt, Phys.Rev.B 82 , 060505(R)(2010). [69]J.Wang,M.Singh,M.Tian,N.Kumar,B.Z.Liu,C.Shi,J.K.Jian,N.Samarth,T. E.Mallouk,M.H.W.Chan, NaturePhysics 6 ,389(2010). [70]M.S.Anwar,F.Czeschka,M.Hesselberth,M.Porcu,J.Aarts, Phys.Rev.B 82 , 100501(2010). [71]Y.V.Fominov,A.A.Gulobov,T.Y.Karminskaya,M.Y.Kupriyano,R.G.Deminov, L.R.Tagirov, JETPLett. 91 ,308(2010). [72]P.V.Leksin,N.N.Garif'yanov,I.A.Garifullin,Y.V.Fominov,J.Schumann,Y. Krupskaya,V.Kataev,O.G.Schmidt,B.Buchner, Phys.Rev.Lett. 109 ,057005 (2012). [73]A.A.Jara,C.Safranski,I.N.Krivorotov,C.Wu,A.N.Malmi-Kakkada,O.T.Valls, K.Halterman, Phys.Rev.B 89 ,184502(2014). [74]M.G.Flokstra,T.C.Cunningham,J.Kim,N.Satchell,G.Burnell,P.J.Curran,S. J.Bending,C.J.Kinane,J.F.K.Cooper,S.Langridge,A.Isidori,N.Pugach,M. Eschrig,S.L.Lee, Phys.Rev.B 91 ,060501(R)(2015). [75]X.L.Wang,A.DiBernardo,N.Banerjee,A.Wells,F.S.Bergeret,M.G.Blamire,J. W.A.Robinson, Phys.Rev.B 89 ,140508(R)(2014). [76]A.Singh,S.Voltan,K.Lahabi,J.Aarts, Phys.Rev.X 5 ,021019(2015). [77]N.Banerjee,J.W.A.Robinson,M.G.Blamire, Nat.Commun. 5 ,4771(2014). [78]M.G.BlamireandJ.W.A.Robinson, J.Phys.:Condens.Matter 26 ,453201(2014). [79]J.LinderandJ.W.A.Robinson, Nature 11 ,307(2015). [80]CTI-Cryogenics, 8200 TM Compressor:Installation,Operation,andServicingInstruc- tions ,HelixTechnologyCorporation(1995). [81]F.Pobell, MatterandMethodsatLowTemperatures,SecondEd. Springer-Verlag, Berlin(1996). 183 [82]R.C.Jaklevic,J.Lambe,A.H.Silver,J.E.Mercereau, Phys.Rev.Lett. 12 ,159 (1964). [83]J.Clarke,A.I.Braginski(Eds.), TheSQUIDHandbook,Vol1. Wiley-VCH,Weinheim (2004). [84]R.Kleiner,D.Koelle,F.Ludqig,J.Clarke, ProceedingsoftheIEEE 92 ,10(2004). [85]A.H.Silver,J.E.Zimmerman, Phys.Rev. 157 ,317(1967). [86]S.S.P.Parkin,N.More,K.P.Roche, Phys.Rev.Lett. 64 2304(1990). [87]B.Dassonneville,H.Y.T.Nguyen,R.Acharyya,R.Loloee,W.P.Pratt,Jr.,J.Bass, IEEETransactionsonMagnetics 46 ,6(2010). [88]J.Bass,W.P.Pratt,Jr., J.Phys.:Condens.Matter 19 ,183201(2007). [89]L.Piraux,S.Dubois,A.Fert,L.Belliard, Eur.Phys.J.B 4 ,413(1998). [90]A.C.Reilly,W.-C.Chiang,W.Park,S.Y.Hsu,R.Loloee,S.Steenwyk,W.P.Pratt, Jr.,J.Bass, IEEETransactionsonMagnetics 34 ,4(1998). [91]M.A.Khasawneh,T.S.Khaire,C.Klose,W.P.Pratt,Jr.,N.O.Birge, Supercond. Sci.Technol. 24 ,024005(2011). [92]E.C.Gingrich,P.Quarterman,Y.Wang,R.Loloee,W.P.Pratt,Jr.,N.O.Birge, Phys.Rev.B 86 ,224506(2012). [93]C.E.Moreau,J.A.Caballero,R.Loloee,W.P.Pratt,Jr.,N.O.Birge, J.Appl.Phys. , 87 9(2000). [94]T.Yu.Karminskaya,M.Yu.Kupriyanov, JETPLetters 85 ,6(2007). [95]T.E.Golikova,F.Hubler,D.Beckmann,I.E.Batov,T.Yu.Karminskaya,M.Yu. Kupriyanov,A.A.Golubov,V.V.Ryazanov, Phys.Rev.B 86 ,064416(2012). [96]PrivatecommunicationwithDr.BaokangBi,MSU. [97]PrivatecommunicationwithVincentMourik,Delft,atAPSMarchMeeting2013, Baltimore. 184 [98]E.G.SpencerandP.H.Schmidt, JVST 8 ,5(1971). [99] OnecmIonMillManual ,CommonwealthScien(1985). [100]P.Horowitz,W.Hill, TheArtofElectronics .CambridgeUniversityPress,Cambridge (1989). [101]P.Cadden-Zimansky,Z.Jiang,V.Chandrasekhar, NewJ.Phys. 9 ,116(2007). [102]J.Wang,M.Singh,M.Tian,N.Kumar,B.Liu,C.Shi,J.K.Jain,N.Samarth,T.E. Mallouk,M.H.W.Chan, NaturePhysics 6 ,389(2010). [103]M.Kompaniiets,O.V.Dobrovolskiy,C.Neetzel,W.Ensinger,M.Huth, JSupercond NovMagn 28 ,431(2015). [104]P.Charlat,H.Courtois,Ph.Gandit,D.Mailly,A.F.Volkov,B.Pannetier, Phys.Rev. Lett. 77 ,24(1996). 185