MALAYSIAJEOL(MALAYSIA) SDN.BHD.508, Block A, Level 5, Kelana Business Center, 97, Jalan SS 7/2, Kelana Jaya, 47301 Petaling Jaya, Selangor, MalaysiaTel. 60-3-7492-7722Fax. 60-3-7492-7723
MEXICOJEOL DE MEXICO S.A. DE C.V.Arkansas 11 Piso 2Colonia NapolesDelegacion Benito Juarez, C.P. 03810Mexico D.F., MexicoTel. 52-5-55-211-4511Fax. 52-5-55-211-0720
PAKISTAN (Karachi)ANALYTICAL MEASURING SYSTEM (PVT) LTD.(AMS LTD.)14-C Main Sehar Commercial Avenue Lane 4,Khayaban-e-Sehar, D.H.A-VII, Karachi-75500, PakinstanTel. 92-21-35345581/35340747Fax. 92-21-35345582
PANAMAPROMED S.A.Parque Industrial Costa del Este Urbanizacion Costa del EsteApartado 0816-01755, Panama, PanamaTel. 507-303-3100Fax. 507-303-3115
SOUTH AFRICAADI Scientific (Pty) Ltd.370 Angus Crescent, Northlands Business Park, 29 Newmarket RoadNorthriding, Ranburg, Republic of South AfricaTel. 27-11-462-1363Fax. 27-11-462-1466
TAIWAN JIE DONG CO., LTD.7F, 112, Chung Hsiao East Road, Section 1, Taipei, Taiwan 10023Republic of ChinaTel. 886-2-2395-2978Fax. 886-2-2322-4655
For NMR & Mass Spectrometer ProductsWidetron Technologies Corp. No.8-2., No.77, Sec.2, Zhonghua E Rd., East Dist., Tainan City 701, Taiwan(R.O.C.)Tel. 886-6-289-1943Fax. 886-6-289-1743
(For Mass Spectrometer Products) Tech Max Technical Co., Ltd.5F., No.11, Wuquan 2nd Rd., Wugu Dist.,New Taipei City 248, Taiwan (R.O.C.)Tel. 886-2-8990-1779Fax. 886-2-8990-2559
For Semiconductor Products:JEOL TAIWAN SEMICONDUCTORS LTD.11F-1, No. 346, Pei-Da Road, Hsin-Chu City 300,Taiwan, Republic of ChinaTel. 886-3-523-8490Fax. 886-3-523-8503
UAEMRS Group Dubai P.O.Box: 171612,Rashid Abdulla Al Noami BuildingSuite No: 306, Damascus Street,Al Qusais-3, Dubai, U.A.E.Tel. 971-4-2581141Fax. 971-4-2581151
Gulf Bio Analytical LLCOffice No.902, Saphire Tower,Al Ittihad Road, Near DNATA Building,Deira Dubai, UAE(P.O.Box 28832)Tel. 971-4250-7300Fax. 971-4250-7400
JEOL Gulfc/o Gulf Bio Analytical LLCOffice No.902, Saphire Tower,Al Ittihad Road, Near Dnata Building,Deira Dubai, UAE(P.O.Box 28832)Tel. 971-4-2579228Fax. 971-4-2579448
USAJEOL USA, INC.11 Dearborn Road, Peabody, MA 01960, U.S.A.Tel. 1-978-535-5900Fax. 1-978-536-2205/2206
JEOL USA, INC. WEST OFFICE5653 Stoneridge Drive Suite #110Pleasanton, CA 94588, U.S.A.Tel. 1-925-737-1740Fax. 1-925-737-1749
VENEZUELAGOMSA Service and Supply C.A.Urbanizacion Montalban III- Residencias Don Andres - Piso 7 - Apartomento 74Avenida 3, entre calles 7 y 6Montalban, Caracas, VenezuelaTel. 58-212-443-4342Fax. 58-212-443-4342
VIETNAMTECHNICAL MATERIALS AND RESOURCES IMPORT-EXPORT JOINT STOCK COMPANY(REXCO)Hanoi Branch,No. 13-Lot 12 Trung Yen, Trung Hoa Street, Cau Giay Dist, Hanoi, VietnamTel. 84-43-562-0516Fax. 84-43-853-2511
Development of Aberration Corrected Differential Phase Contrast (DPC) STEM
Damien McGrouther, Maria-Jose Benitez, Sam McFadzean, and Stephen McVitie
SUPA School of Physics & Astronomy, University of Glasgow
In this article we demonstrate that aberration correction for STEM probes has been achieved for field-free Lorentz STEM imaging of magnetic samples, and that, an order of magnitude improvement of spatial resolution has been obtained. We believe, that our achieved <1 nm spatial resolution is currently the best in the world for direct imaging of magnetic structure by electron microscopy.
Introduction
Correctors for spher ical aberration (Cs) ofelectron lenses have resulted in a step-change inperformance forbothTEMandSTEMinstruments,makingatomicscaleimagingandanalysisofmaterialsroutinely possible. Working in collaboration withJEOLandpartnersweshowthataberrationcorrectionand other technologies have enabled an order ofmagnitude improvement in the capability to imagemagnetic behaviour in thin nano-scale structuresi mpor ta nt for cu r rent a nd f utu re i n for mat iontechnologies. T he i mag i ng of mag net ic s t r uc tu re i n theelectronmicroscopehasa longhistory thatextendsbackto the1950's.Thegroupof imagingtechniquesused togeneratemagnetic contrast are collectivelyknown as "Lorentz microscopy" [1] as they can beunderstood in terms of the classical Lorentz force(F= -e(v×B) )experiencedby thebeamelectronstraversingthespecimen.Thinmagneticsamplesalsoexhibitaquantuminteractionwiththebeam,via theAharonov-Bohm effect, whereby the phase of thepassingelectronwavesarealtered.Thus, for imagingmagneticsamplestheLorentztechniquesareabranchofphasecontrastmicroscopy.InTEMmode,FresnelandFoucault techniqueshavebeenextensivelyused.Bothareeffectiveingeneratingimagesshowingstrongmagneticcontrastbutarelimitedinspatialresolutionor linearity.TEMHolographic techniqueshavebeenquite successfuland shown tobecapableofhigherresolutionimagingofmagneticinduction[2],however,theirbasis isoff-line imagereconstructionand theycannotbeappliedtoallsamplegeometries. AttheUniversityofGlasgow,wehavedevelopedtheSTEMbasedLorentzimagingmodeofDifferentialPhaseContrast (DPC)over the last30years. In this
article we demonstrate that in collaboration withJEOL, CEOS GmbH, Gatan Inc., Deben Ltd andUniversity of Warwick, that aberration correctedDPC STEM has been achieved enabling the studyof magnetic structure with world-leading spatialresolutionbetter than1nanometre.Furthermore, incontrast toholographic techniques, these imagesareavailableinrealtimeatnearvideoframe-rates.
Experimental
On a JEOL JEM-ARM200FCS TEM/STEMequipped with cold field emission gun (C-FEG),a CEOS Cs STEM probe corrector and HR pole-piece, severalmajordevelopmentshavebeenmadeto successfully realise DPC mode imaging. AfterdescribingthegeneralconceptoftheDPCmodewewilldealwitheachoftherequireddevelopmentsinturn. Figure 1 depicts the setup required for DPCmode imaging,where the focusedelectronprobe israster-scannedacrossthespecimenwiththescatteredtransmittedconeofelectronsbeingdetectedinthefar-fieldbyasegmentedSTEMdetector.If thespecimenismagneticandcontainsregionswithcomponentsofthemagnetic inductionBoriented in the specimenplane,thenitcanbeshownthatthebeamisdeflectedthroughanangle:
eλß = — ∫B × ndzh
where e is the charge on the electron, λ it'swavelength,Bthemagneticinductioninthespecimenandn theunit vectoralong theelectron trajectory.TheclassicalLorentzdeflection inducedbya typicalmagnetic sample is relatively weak. The deflectionangle,ß,isintherange1-100micro-radiansandisverymuchsmallerthantypicaldiffractionscatteringangleswhicharegenerally>3milli-radians.Thesegmenteddetector is used to detect such Lorentz deflection
of the beam by measuring difference signals fromoppositequadrants.Analternative interpretationoftheinteractionofathinmagneticsampleonthebeamisthattheelectronbeam'swavefunctionafterpassingthrough the region containing magnetic inductionbecomesphase-shiftedduetothequantummechanicalAharonov-Bohmeffect [3].Thinking in these terms,bytheactionoftakingdifferencesignals,thegradientof thephase-changedue to the sample ismeasuredand hence the technique produces images showingdifferentialphasecontrast. Theprocessof combining the signals from thedetectorsegmentstoproduceliveDPCSTEMimagesisdescribedinmoredetaillater.
STEM probe formation for magnetic imaging.
In standardSTEMmode, thenormallyexcitedobjective lens (OL),whichproduces focusedSTEMprobeswithsemi-convergenceangles,α=3-30mradand has enabled imaging of Si-dumb-bel ls withinformationat0.67Å.However, thenormallyexcitedOL also subjects the sample to a magnetic field ofstrength ~2 Tesla. A field of such strength would
completely saturate the vast majority of magneticsamples,obliteratinganymagneticdomainstructureof interest.Thus,DPCSTEMmodeimagingmustbeperformed with the OL completely de-excited andthesampleresiding in field-freeornear to field-freeconditions.This iseasilyachievedby switching themicroscopeinto"LOWMAG"modewhereby,theOLgoesoffandSTEMprobeformation iscontrolledbyacombinationof thevariablecondenser (CL3)andcondensermini(CM)lenses.Inthisscenario,and intheabsenceofanaberrationcorrector, thediameteroftheSTEMprobewouldbedictatedbythesphericalaberration coefficient of the CM. In collaborationwithus, JEOLandCEOShavedevelopeda specialoptical configuration for the aberration correctorthat compensates for the Cs of the CM and haveenabled a magnification range that extends up to2.0million times.Figure 2 showsan image for theresultingRonchigraminthismodethatexhibitsaflat,aberrationcorrected, region thatextendsout to3.2milli-radiansemi-convergenceangle.Usinga70µmcondenseraperture,correspondingtothefulldiameteroftheflatregion,STEMimagingofatestsample,Aunano-particles,showninFig. 3(a),demonstratedthatparticleswithsizesof theorderof1nmandsmallercould be resolved. In fact, Fig. 3(b) demonstrates
thatthesmallestparticlethatcouldberesolvedhadawidthoftheorderof0.7nm.TakingtheFastFourierTransformofFig.3(a)(insetofFig.3(c))andformingaradiallyaveraged lineprofileshowedthatFig.3(a)containsinformationcontentuptoamaximumspatialfrequencyof1.8nm-1.AlloftheseobservationsareinagreementwithcalculationsmadebyCEOSGmbHwhoexpected that theCScoefficientof theCMlensshouldbereducedtotheorderofseveralmicronsandshouldresult inaFWHMprobediameterof0.8-1.0nmwithspatial resolutiondefinedasbeinghalf thisvalue,0.4.-0.5nm.
While it would be desirable to always operatewith the highest spatial resolution, inevitably atrade-offhas tobemade.Recallingthat theLorentzdeflectionangle,ß, canbeas smallasa fewmicro-radians, around 1000× smaller than the optimumprobesemi-convergenceangleα.HighersensitivitytosmallLorentzdeflectionscanbeobtainedbyreducingαat theexpenseofspatialresolution.Thiscaneasilybedonebychangingtothesmallest10µmcondenseraperturewhich reducesα to450micro-radiansbutmeansthat thespatialresolutionbecomes limitedbydiffraction.Lorentz sensitivity canbe increasedby
a further twotimes,α furtherreducedto215micro-radians,bythecombinedadjustmentof theCL3lensandtheCEOScorrector'sadapterlenselement(ADL).Under these conditions spatial resolutionhasbeenmeasuredtobeintherange3-5nm.
Nulling and application of in-situ magnetic fields to the specimen
In "LOW MAG" mode, a lthough the OL iscompletelyde-excited, the sample still resides inamodestmagneticfield,~150Oedirectedperpendicularto itsplane.This is theresultofremanencefromtheferromagnetic pole-pieces. For many thin film in-planemagnetisedsamples theout-of-planeorientedremanent f ield has l itt le inf luence on the staticmagneticstructure.Itsstrengthisgenerallyverymuchweaker than the strength of the in-plane magneticanisotropy. In-situ studies of magnetic reversalbehaviour canbeaccomplishedbyusingeither theremanentfield,orastrongerfieldappliedbypartiallyexciting theOL.Tilting thesample(generallyup to+/-30degreesispossible)canthenbeusedtonucleateand grow magnetic domains eventually leading toreversal and saturation of the film. For ultra-softmagnetic samples,where thecoercivity isverymuchless than the remanent field strength, it isdesirableto be able to reduce to near zero the strength ofthe remnant field. Utilising a system developed atUniversityofWarwickwecanmeasure thestrengthoftheremanentfieldatthesampleplaneusingaHall-probeTEMrodandapplyareversecurrent through
DevelopmentofthesegmenteddetectorandDPCimageacquisition systemhas requiredanextensivecollaboration involvingourselves,JEOL,GatanInc.,DebenLtdandAndrewArmitDesigns. T he ge omet r y of t he seg mente d dete c toremployed isdepicted inFig. 4(a).Itconsistsofeightsegments arranged into an inner sol id quadrant(INT0 to INT3) and an outer annular quadrant(EXT0toEXT3).DPCSTEMimaging,detectingthedisplacementofthetransmittedelectrondisc, ismostsimplyrealisedbyusingacameralengththatprojectsthe transmitted electron disc onto only the innerquadrants.However, inpreviouswork [4],wehaveshown that forpolycrystallinemagnetic thin films,strongandunwantedelectrostaticphase fluctuationsariseduetodiffractionfromthenano-scalecrystalliteswithvaryingorientations.Byutilizingacamerallengththatprojects thetransmittedelectrondiscacross theouterannular (aswellas the inner )quadrants, thehigherspatial frequencyelectrostaticfluctuationscanbe"filtered"fromthelowerspatialfrequencymagneticdomainanddomainwallfeatures. Conversionof charge signals fromthedetectorsegments into video-level voltage signals has beenachieved through thedevelopmentof theSuperFast8-channel, 2 MHz bandwidth amplifier by Deben
&AndrewArmitdesigns.TheSuperFastamplifieris controlled by software and has a wide range ofsettingsenablingselection, foreachchannel,of inputresistance/capacitancefornoisereduction/bandwidthselection and gain. "On the fly" arithmetic mixingofchannels, ispossibleandcanbeused toview livedifference signals between segments. However, weprefer toperformsuch imagearithmeticonacquireddigital imagesandutilize theSuperFastamplifier topassthesegmentsignalsunaltered. Commonly on an advanced STEM instrumentsuch a s the J EM-A R M 2 0 0 F, c ombi ned i mageacquisition and point-wise analysis (via ElectronEnergyLossSpectroscopy(EELS)orX-rayEnergyDispersive Spectrometry (EDS)) are controlled byGatan'sModel788DigiscanIIsystemthroughDigitalMicrographsoftware.OnourJEM-ARM200FCS,theDPCdetectoradds8 segment signals to thealreadylengthylistofsignalstobeacquiredfromthecommonSTEM detectors (JEOL ADF1, ADF2, BF, GatanModel806HAADF,Model807BF/ADF)andcurrentmeasurementfromtheCFEG.Thus, inall,a totalof13signalswererequiredforacquisition,althoughnotallwouldbeusedatanyone time.Gatandevelopeda solution for this by implementing hardware andsoftware that al lowed 4 Digiscan II boxes to beoperatedinparallel.ThiswasachievedinsuchawaythatforrecentreleasesofDigitalMicrographsoftware(fromGMSversion2.3.X),thiscapability isnowpartofthestandardsoftware-base. LiveDPC imaging,enablingmagnetic contrasttobevisualised,hasbeenachievedthroughacontrolpalette,Fig.4(b),created in theDigitalMicrographscripting languageby theauthor.Byclicking“Start/Stop”or“GrabFrame”buttons,callsareissuedtotheDigiscanIIboxesthatstart/stoptheimagingprocess.Theindividualsegmentimagesarevisiblebutmagneticcontrastcanonlybeseenbydisplayinglivedifferenceimagesbetweenopposite segments.Twoorthogonaldirection components are necessary to reconstructmagnetic orientations and these are achieved byviewing the image pairs “INT0 – INT2”, “INT1 –INT3” ifusingthe innerquadrant(“EXT0–EXT2”& “EXT1 – EXT3” if using the outer quadrant) .Basedontheseimagepairslivecolourimagesshowingmagneticorientationsarealsopresented.Quantitativedetermination of Lorentz def lection, and hence,BS×t, the product of the magnetic induction timesthethicknessofthespecimen,ispossiblebyrelativelystraightforward post-processing of the recordedDPCimages.It is the intentionto includeareal-timecapability so thatDPC images canbecalibrated intermsofquantitativeLorentzdeflection.
Investigation of magnetic samples
In this sectionwepresentresultsobtainedfromapplyingtheDPCsystemtoinvestigatethepropertiesandbehaviourofmagneticspecimens insomeofourcurrentresearch.
Iron nanostructures
Thefabricationofnano-scalemagneticstructures
is a lengthy procedure, most commonly achievedby multi-step lithographic techniques in which theshapestobecreatedarewrittenintoasensitiveresistfollowedbychemicaldevelopment,metallisationand"lift-off"steps.Alternatively, rapiddirect-writingofmagneticnano-structurescanbeachieved infocusedionbeamandscanningelectronmicroscope (SEM)systemswhereaneedlebasedsystemisusedtoinjectan organometallic precursor gas into the region ofthebeam-scanning [5,6].UsingSEM, suchelectronbeaminduceddeposition(EBID)hasbeenemployedtocreaterectangularironelements,Fig. 5,andpillarswithdiameteraround50nanometres,Fig. 6. Fig. 5(a) and (b) show greyscale DPC images(obtainedusingSpotL1anda10microncondenseraperture)thathighlightthegroundstatearrangementofmagnetisation in theapproximately600nm×400nm×40nmthick rectangularelementswhichwerefabricatedonSi3N4supportmembranes.Asdescribedearlier,Fig.5(a)&(b),areproducedbysubtractingthevideosignalsfromopposingsegmentsontheDPCdetectorandproduceapairofimageswithorthogonaldirectionsof sensitivity. Inside theelements strongblackandwhitecontrastcanbeseenthatcorrespondsto the magnetic domain structure, while outsideof the elements "noisy" phase contrast from thethin carbon coating used for charge dissipation isobserved.Thearrangementof themagneticdomainorientations intheelement ismosteasilyunderstoodby forming the colour map in Fig. 5(c). From thisFig.itcanbeseenthattheelementhasformedaflux-closingmulti-domainLandau typepattern inwhichthe magnetisation tends to be oriented parallel totheelementedgesandcirculatesaroundtwovorticeswithintheelementsinterior.Thespatialextentofeachofthevorticesisdictatedbythemagneticpropertiesofthematerial,specificallytheexchangestiffnessandthesaturationmagnetisation.Frompolycrystallinealloysinvolving Co, Ni and Fe, vortex widths have beenmeasuredtorangefrom7-15nanometreswide[7].InFig.5(d),wehaveutilisedthehighspatialresolutionaffordedbyaberrationcorrectionof theCMlens tomeasureaDPCintensityprofile fromthepositionoftheredlineinFig.5b.Fig.5d,showsthatfortheEBIDFeelement(withapproximatechemicalcomposition60%iron,40%carbon)thatthemeasuredwidthofthevortexcoreis13.6nm. Narrow pillar-like magnetic structures can beformedbyallowingtheelectronbeamintheSEMtodwellona single location.Suchpillarshaveprovedhighlyeffectiveasmagnetically switchable trappingsites when directly written on top of out-of-planemagnetisednanostrips (seereference [8] fora fullerexplanation). For these pillars, due to their smalldiameter,~50nm,DPC imaginghasbeenused tomeasurethemagneticfieldstrengthneededtoswitchthepillarsmagnetisationdirection.Fig.s6(a)and(b)showscolourDPCmaps thatdepict thenanopillarsgrown on the edge of a grid support. The colourcontrast inside thepillars isnotsimply interpretablein terms of magnetic structure as it is dominatedbyelectrostaticphase gradients from the changingthicknessassociatedwiththeircircularcross-section.In the free-space region immediately surroundingthetipofthepillar, locatedinsidethedashedellipse,colourcontrastrelatingtothepillar'sde-magnetising
7 JEOL News Vol. 49 No. 1 (2014)
fieldscanbeobserved.Starting inFig.6(a),movinginaclockwisedirectionaroundthe tipof thepillar,thecolourcontrastchangesfrombluetoredtoyellow.Referringtothecolour-wheel inset, this indicatesthemagnetic fieldsemanatingfromthetiparedivergentandthusitcanbeinferredthatthepillarismagnetisedin an upwards direction. A field of strength 1000Oewas thenappliedto thepillar in-situbypartiallyexcitingtheOLlensandtiltingthespecimentonear30degrees.Afterde-exciting theOLandreturningthespecimento itsuntiltedstate, thecolourmap inFig. 6(b)wasobtained.Again, thedirectionof themagnetisationinthepillarwasinferredbyexaminingthecolourcontrastassociatedwithde-magnetisationfield from thepillar tip. InFig. 6(b) it canbe seenthatthecolourcontrasthasalteredandchangesfromyellow togreen tobluewhenmoving inaclockwise
sense around the tip. This indicates that the de-magnetisation fields are now convergent at the tipand infers that themagnetisationdirectionhasbeenswitched toadownwardsorientationby theappliedfield.
FIBcross-sectionof themultilayer (~50nm thick)withthestructureNiFe/(FeMn/NiFe)×10grownonanoxidisedSisubstratewithacapping layerof5nmofTa.TheNiFelayershadathicknessof16.5nmandtheFeMnlayers12.8nm. Initiallythesamplewasimmersedinalargefield(around 1000Oe) tohaveall the layers in parallela l ignment. The DPC image component showingthemagnetic inductionparallel to the interfaces isshown inFig.7 (a)where theNiFe layersappearasbright stripes in the imageand theFeMn layersaregray indicatingnonet inductioncomponent in theseregions.Thevariation incontrastwithin the stripesis a consequence of the granular structure of thefilmand thisgives rise toadiffractioncontributionto the phase contrast image. A linetrace from thearea indicatedbytherectangle inFig.7(a) isshown,thisaveragesthesignalovera74nmwidthtoreducetheeffectsof thediffractioncontrast fromthegrainstructure. The profile shows the magnetised layervariationvery clearlywhereeachmagnetic layer isaround16-17nmwide(i.e.depositedthicknessofthefilm)and theAF layer is13nmwide.By tilting thesample in theobjective lens field themagnetic statecouldbealteredwith individual layersswitchingandanexampleofthestatepartwaythroughthereversalprocess is shown in Fig. 7 (b) where seven of theelevenmagnetic layershaveswitched theirdirectionofmagnetisation,onepartially.ThiscanalsobeseenbycomparingtheaveragedlinetracesforFig.s7(a)&(b)forthetwodifferentstates.Thelinetracesshowtheinduction in theferromagnetic layerveryclearlyandindeed the interfacebetween the ferromagneticandantiferromagnetic layersshowsthetransitionwhichisontheorderof1-2nm.
Reduction of magnetization in nano-scale regions by ion irradiation
Obtainingquantitativemeasurementsregardingthestrengthofmagnetic inductionfromDPCimagesisusuallystraightforward.WehavebeeninvestigatingtheuseofFIBbased ion-irradiation to control thestrength of magnetisation in Cr(3 nm)/Ni80Fe20(10nm)/Cr(5 nm) f i lms deposited on Si 3N4 electrontransparentmembranes.Thekeyaim foruswas tocreateandcharacterisenarrowirradiatedlinedefectsthat couldactas trapping sites fordomainwalls inmagneticnanowires [9,10].Fig. 8 showsquantitativeDPCimagingofa line irradiatedatadoseof8×1015ionscm-2.Thecomponentsofmagneticinductionweremappedparallel,Fig.8(a),andorthogonal,Fig.8(b),totheirradiatedline.InFig.8(a)theirradiatedlineisobservedasalowerintensityfeaturewhileitisinvisibleinFig.8(b),thelatterisconsistentwiththecomponentof magnetic induction being continuous across aninterfaceasproved fromMaxwell’sequations,eventhough the magnetization is discontinuous. TheintensityprofilefromtheregionindicatedinFig.8(a)isplotted inFig.8(c)wheretheverticalaxisdisplaysquantitativemeasurementoftheLorentzdeflectionofthebeam.Quantitativedeterminationofthedeflectionisachievedbydividing thedifference imagesby the"SUM" image (i.e. summing the images from allsegments). Since the diameter of the transmittedelectron disc incident on the segmented detectorrelatestothebeamconvergencesemi-angle,α,whichisknown,thentheLorentzdeflection,ß,canbeeasilyberecovered.InFig.8(c), thequantitativeprofileshowsameasuredLorentzdeflectionangleof~4.3µradfortheunirradiatedregion.This isasexpected.Fora10nmthickNi80Fe20filmwithBS=1Tesla,thenthetotalbeamdeflectionshouldbeß=6.5µrad.WiththeDPC
9 JEOL News Vol. 49 No. 1 (2014)
sensitivitycomponentsbeingorientedat45degreesto the direction of the mean magnetisation in thefilm,thenthemeasuredß isreducedbysin(45)=0.7,yieldingß=4.4µrad.The irradiationdoseof8×1015ionscm-2hasresultedinalineofwidth50nm,withameasureddeflectionof1.3µrad,corresponding toareductionof70%reductioninMS.
Summary
In summary,our collaborativedevelopmentofan aberration corrected STEM Differential PhaseContrast system has demonstrated quantitat iveimagingofmagneticstructurewithspatialresolutioninthe1-6nanometrerange.Asfarasweareaware,apartfromultra-highvacummbased scanning tunnellingmicroscopy(UHV-STM)ofatomicsurfaces,weknowofnoothertechniquesthatcurrentlyenablesmagneticimagingat this lengthscale.Excitingly,weenvisagethat further improvementscanbemade.All resultspresentedherewereobtainedwithabeamenergyof200keV.RecentlywehavebegunperformingDPCat80keVwhichshouldleadtoarounda4×improvementinmagneticsensitivityandwhichwillbeessential forinvestigatingnewphenomena inultrathin,1-5atomthick, magnetic layers.Furthermore,DPC imagingis not just limited to magnetic samples. Materials
and fi lms containing intrinsic electric f ields andpolarisationexert similar influenceon theelectronbeam.However, theexcitingprospecthere is thattheOLneednotbede-excitedforsuchwork.OperatinginthemoreusualaberrationcorrectedOLONmode,atomic resolution DPC investigations are enabled.Weenvisagethatsuchpowerful imagingmaybenefittheunderstandingofchargedistributions inbonding,across interfaces and at surfaces and lead to thediscoveryofnewaspectsofmaterialsphysics.
Acknowledgments
Thedevelopmentsreportedandprocurementofthemicroscopewereenabledbyjointfundingfrom the University of Glasgow and the ScottishFundingCouncil(throughtheScottishUniversitiesPhysicsAlliance(SUPA)). TheauthorswouldliketotaketheopportunitytoexpresstheirgratitudetoallstaffmembersofJEOL,GatanInc.,CEOSGmbH,DebenLtd.,UniversityofWarwickand toAndrewArmit for their invaluableefforts in this collaboration.Wearegrateful to thefollowingcollaborators for samples:EBIDFenano-structuresfromthegroupofH.J.M.SwagtenatTUEindhoven, Netherlands; multilayer ferromagneticsamplesfromthegroupofR.M.BowmanatQueens
University, Belfast, UK; tri-layered Cr/Ni80Fe20/CrsamplesfromthegroupofC.H.Marrows,UniversityofLeeds,UK. We also acknowledge funding support fromtheUKEPSRC,grantnumberEP/I013520/1,whichfundedoneoftheauthors(M-J.B.)andenabledmuchofthedevelopmentwork.
D. McGrouther, S. McVitie, J. N. Chapman,Physical Review Letters100,029703(2008).
[8] J.H.Franken,M.A. J. vanderHeijden,T.H.Ellis,R.Lavrijsen,C.Daniels,D.McGrouther,H. J. M Swagten, B. Koopmans, accepted forpublication in Advanced Functional Materialsdoi:10.1002/adfm.201303540.
Atomic-Resolution Characterization Using the Aberration-Corrected JEOL JEM-ARM200CF at the University of Illinois – Chicago
Robert F Klie, Ahmet Gulec, Arijita Mukherjee, Tadas Paulauskas, Q i a o Q i a o , X u e R u i , R u n z h e T a o , C a n h u i W a n g , T a d D a n i e l , Patrick J. Phillips, Alan W. Nicholls
University of Illinois at Chicago
Modern aberration-corrected scanning transmission electron microscopes (STEM) provide a multitude of characterization techniques that can be applied across a wide-range of length and temperature scales. At the University of Illinois at Chicago, the aberration-corrected, cold-field emission JEOL JEM-ARM200CF is capable of atomic-resolution imaging, electron energy-loss (EEL) as well as energy-dispersive X-ray (XEDS) spectroscopy in a temperature range between 80 – 1,300 K. The capabilities of this instrument will be demonstrated using a number of studies focusing on both structural and chemical properties of materials including NbH, SrTiO3, CdTe and ferritin protein. The central theme of these studies is microscope versatility, which is realized through the ability to perform atomic-scale chemical characterization while retaining high spatial imaging resolution. Of particular interest to many studies is the visualization of light elements, such as N, O or H, using simultaneous high-angle annular dark field (HAADF) and annular bright field (ABF) imaging. Novel in-situ capabilities will be demonstrated using graphene liquid cells. Finally, we will demonstrate the effects of a new silicon drift detector, the Oxford X-MaxN 100TLE on performing atomic-column resolved XEDS mapping at varying length scales and energy resolution.
Introduction
Abetterunderstandingof themechanisms thatinfluenceamaterial’spropertyontheatomic level isessential fordevelopingnewfunctionalnano-devicesforelectronics,energyandbiologicalapplications, itisessentialthatwegainanunderstanding.Evenmoresothanonthemacroscopic level,whenacomponenthas been miniatur ized for use in nano-devices,surfaces, interfaces,and individualdefectsdominateitspropertiesbychanging thematerial’s structural,compositionalandbondingbehavior. Ifworkingandreliabledevicesaretobesuccessfullyfabricated,itisimperativethatweunderstandthestructure-propertyrelationshipsofthesedefects,interfacesandsurfaces. Over the last decade, analytical transmissionelectron microscopy (TEM) and scanning TEM(STEM) have emerged as the pr imary tools forexploring the structure-property relationships ofnovelnano-materials [1-15], inparticular theatomicand electronic structures of defects, interfaces, ornano-particles [16-20]. The scanning transmissionelectron microscope (STEM) fitted with a probe-side aberration corrector, such as the JEOL JEM-
2200FSor theJEOLJEM-ARM200F [21-24]yieldsunparalleled spatial resolution of both heavy andlight chemical species. With spatial resolution ofaberration-correctedhigh-resolutionphasecontrastimaging(HRTEM)andhighangleannulardarkfield,HAADF, (or Z-contrast) imaging in aberration-correctedSTEMsnowreachingthefundamentallimitof theBohr radius, a0~50pm, the focus is shiftingto increasingthechemicalresolutionandvisualizinga l l const ituent atomic spec ies , whi le reta in ingspatial resolution.To thatend,electronenergy lossspectroscopy (EELS) and energy-dispersive X-rayspectroscopy(XEDS)arebeingpushed to thepointwhereatomically-resolvedchemicalmapsarepossiblewithboth techniques [25, 26].The implementationof silicon drift detectors (SDD) with larger activedetector areas in XEDS have greatly aided thisadvancement, leading to thedemonstrationofsingleatomsensitivity[27]. Many material systems are utilized in devicesunder conditions (i.e. temperatures, pressures orfields) that are substantially different from thosepresent inside an electron microscope column.In a conventional atomic-resolution transmissionelectronmicroscope,atomic-resolutionstudieshavepreviouslybeenlimitedtoambienttemperaturesandunderhighvacuum(PO2=10-5Pa)duetothe inabilitytoget in-situholders into the smallgapofanultra-high-resolution objective lenspole-piece. Over the
lastdecade, the fieldhasadvancedsignificantly,andin-situ capabilities are now available to study thedynamicbehaviorofmaterialsinenvironmentsotherthan the high-vacuum of the TEM column. Novelsample stage designs have enabled in-situ heatingup to 1,300 K with atomic resolution, in-situ gasand liquidexperimentswithnano-meter resolution,as well as time-resolved imaging with better thannanosecondresolution. Thispaperwilldemonstrate thecapabilitiesofthe JEOL JEM-ARM200CF at the University ofIll inois at Chicago, a probe aberration-correctedSTEMequippedwithacold-fieldemissionelectronsource,apost-columnEELspectrometerandthenewOxfordX-MaxN100TLESDD-XEDSdetector.Wewilldemonstrate thatatomic-resolution imagingoflightelements,includingN,OandHispossibleusingABFimaging,while thecold-fieldemissionelectronsourceallowsforEELSwithanenergyresolutionofbetter than400meV. Inaddition,wewillhighlightournewlydevelopedgraphene liquid cell approachofencapsulatingbiological samples, suchas ferritinproteins in a l iquid between two single layers ofgraphene.
Experimental
The JEOL JEM-ARM200CF installed in theResearchResourcesCentre(RRC)at theUniversityofIllinoisatChicago(Fig. 1)isequippedwithacoldfield emission gun, 5 annular detectors, a GatanEnfina EELS spectrometer and the new OxfordX-Ma x N 10 0 T L E SDD EDS detector a nd wasinstalled inour laboratory in late2011.The spatialresolution thatcannowberoutinelyachievedusingthe JEOL JEM-ARM200CF exceeds 70 pm at 200kVprimaryenergy(seeFig. 2(a))and100pmat80kVprimaryenergy.TheenergyresolutionoftheEEL
spectraatanyenergybetween80and200kV is350meV(seeFig.2(b)).Thestabilityof the instrumentwasmeasuredusinga long-exposureHAADFimagewithout drift correction, and was determined tobetter than 150 pm/min. In addition to the doubletiltholders,wealsotookdeliveryofaGatandouble-ti lt l iquid He cooling stage, a Protochips Adurodouble tilt heating stage, a Fischione tomographystage,aswellasaNanofactorySTM-TEMstageandaProtochipsPoseidon liquid flowcell. Inaddition,wehavestillaccess to theGatandouble tiltheatingand the double tilt LN2 cooling stage, which werepurchasedin1998foruseintheUICJEM-2010F. Inthispaper,wewillshowresultsfromseveralofthese in-situstages, includingtheGatanLN2coolingstageandthehome-madegrapheneliquidcell.Theseresultsaremeant todemonstrate thewide rangeofexperimentsthatcanbeconductedonadailybasisintheARM200CF.Atthispoint,itisimportanttonotethat switchingbetweenTEMandSTEMmodecanbedonewithoutdisturbingthealignmentorstabilityof the instrument significantly. For example, whenswitching fromTEMtoSTEMmode, it ispossibleto achieve atomic resolution nearly immediatelyafter thecomaandastigmatismhavebeenmanuallycorrected. At the highest imaging resolution, wenotice that during the f irst 60 -120 minutes afterenteringSTEMmode, someadjustmentof the focusandstigmatorisnecessary.OncetheARM200CFhasbeeninSTEMmodefortwohours,theinstrumentiscompletelystable. TheUICJEOLJEM-ARM200CFislocatedinanopen-accessuser-facility,whichallowsqualifieduserstousetheinstrument24hours,7daysperweekonanhourlyuser-chargebasis[28].Itis,therefore,crucialthatanyoftheexperimentalsetupsrequiredbytheuserarenotaffectingtheperformanceoftheinstrumentationforthesubsequentusers.Todate,wehavenotexperiencedanysignificantinfluenceofanyin-situexperimentsor
Here, we will discuss the results from severalexperimentsconductedusingtheJEOLARM200CFattheUniversityofIllinoisatChicago.Thematerialsthat were characterized include SrTiO3 thin filmsonGaAs,ß-NbH,polycrystallineCdTeand ferritinproteinsinagrapheneliquidcell.
Low voltage characterization of SrTiO3/GaAs interfaces
Over thepast fewyears,ultrathinmetal-oxidefilmsonpolarsemiconductorsurfaceshavereceivedmuch attent ion due to the occurrence of novelinterfacial properties, including ferroelectricity,superconductivityand thepresenceofan interfacial2 - d i mensiona l elec t ron ga s [ 29 -3 4 ] . T he f i r s tsuccessful growth of SrTiO3 thin fi lms on Si wasreported more than two decades ago, and varioustechniques have been used to assemble a layer-by-layercrystallineoxide filmandavoid the formationoftheamorphouslayerattheinterface[35,36]. T he SrTiO 3/ Ga A s i nter faces were stud ied
exper i menta l ly and theoret ica l ly w ith var ioustech n iques in order to develop a f u ndamenta lunderstandingofthestructurepropertyrelationshipsand ithasbeenreportedthat theSrTiO3filmprefersto be SrO terminated at the interface, regardlessof the growth condition, and a submonolayer ofTi,hereafter referred toasaTipre-layer,betweentheoxideand semiconductorcan release theFermilevel pinningafter the thin film deposition.[37-39]However,due to theoxygendeficiencyof the film,it was not possible to perform atomic resolutionelectron energy-loss spectroscopy (EELS) on thehighlybeamsensitiveSrTiO3/GaAsinterface[40]. Us i ng t he J EOL A R M 2 0 0 C F at U IC , weconducted a deta i led study of the atom ic a ndelectronic structures of the SrTiO3/GaAs interfacewith and without a Ti pre-layer using H A A DFimagingandEELSat80kVprimaryelectronenergy[41] . Figure 3 shows atomic-resolution Z-contrastimagesofa4monolayer thinSrTiO3 filmonGaAs(001),wheretheepitaxyisapparentwithSrTiO3(001)|| GaAs (001) and SrTiO3[110] || GaAs[100] . Theinterface between the SrTiO3 film and the GaAs-support in Fig. 3 appears sharp with no obviousinterfacialdiffusion.Moreover, the images suggestthat theoxide filmsstartwithaSrO layerat theAsterminated GaAs interface, regardless of the factthat0.5MLofaTipre-layerwasdepositedon theGaAssurfaceprior toSrTiO3growth for thesampleshown inFig.3(a)).Asopposed to theexperiments
conducted at 200 kV primary electron energy, theSrTiO3/GaAsinterfacesarenowsufficientlystabletowithstandatomicresolutionimagingandspectroscopyforextendedperiodsof time.Evenseveralhoursofanalysis on a particular region of a sample has sofarnot shownanysignsofbeamdamage,neither inZ-contrastimagingnorinspectroscopy. The acquired Ti L- and O K-edges are shownin Fig. 4 and Fig. 5, respectively acquired using aconvergencesemi-angleof30mrad,acollectionsemi-angleof35mrad,witha0.1eV/pxldispersionanda3sdwelltime.ThetopspectruminFig.4andFig.5istakenfromaSrTiO3bulkspecimenthusprovidingaTi4+finestructureasareference.InFig.4,thecrystalfield splitting of Ti L3 and L2 edges can be clearlyresolvedinallthespectratakenfromtheSrTiO3thinfilm,whichindicatesthattheTivalenceiscloseto4+throughout the films inboth filmswithandwithouttheTiprelayer.[42]To furtherassess theTivalencestate, and thereby the oxygen stoichiometry of thefilms, theenergyscalehasbeencalibratedusing theOK-edgeonset,asshowninFig.4. It cannowbe seen that in the filmwithout theTi pre-layer, the Ti L-edges for each location inthe film are slightly shifted towards lower energy,again indicating a slight decrease in the Ti. Morespecif ical ly, both L3 and L 2 edges shift to lowerenergies as the electron probe approaches to theSrTiO3/GaAs interface, indicatingadecreaseof theTivalence from4+ toamixture stateof3+and4+[42] . Such phenomena could be caused by oxygenvacancies,orbecause theTi-Obondingstateon theTiO2columnsinthethinfilmchangedtoTi-Asonthesurfaceof thesubstrate. Inaddition, it is interestingtonotehere thataTi signal isnoticeableat least2layersintotheGaAssupport,indicatingthatsomeof
theTihasdiffusedintotheGaAssupportduringthefilmsynthesis. In Fig. 5, the acquired O K-edges are f ittedto their 5th nearest neighbors and compared withthe reference spectrum taken from bulk SrTiO3.The colored spectra correspond to the coloredrectangles in Fig. 3. It is immediately noticeablethat theOK-edge fine-structure in theSrTiO3 thinfilms is significantlydifferent fromthebulkSrTiO3,especiallyforthefilmgrownwithouttheTiprelayer.Morespecifically,thepre-peak,whichstronglyscales,with thepresenceofoxygenvacancies, is suppressedinthespectratakenfromthefilms[43,44],indicatingthepresenceofoxygenvacancies.Furthermore,neartheSrTiO3/GaAs interface, the fine-structureof theOK-edgeexhibitsseveralpeaks thatwerenot foundin thebulkSrTiO3,nor in the filmsSrTiO3 spectra.Theseadditionalpeaks indicate theoxygenbondingwitharsenicon theGaAssurfaceduring the initialstepsofthethinfilmsynthesis.Finally,theintegratedO K-edge intensity disappears completely in theGaAssupport,eveninthelocationswhereaTisignalwas found. This further indicates that only the Tidiffusesintothesupport,whiletheoxygenremainsintheSrTiO3filmandontheGaAssurface. Our low-kV imaging and spectroscopy study,therefore, shows, thatwhile theatomic structureofthe interfaces does not reveal any evidence of theTi prelayer at the SrTiO3 interface, the electronicstructureof the filmsappearssignificantlydifferent.The films grown without the Ti pre-layer appearsto be more oxygen def ic ient , and exhibit stonginteract ion between the GaAs support and theinterfacial oxygen, potentially forming As2O3. Thedepositionof theTipre-layerappear toalleviate theoxidationof the substrateandconsequently lift the
Niobium, a 4d t ransit ion meta l , has foundmany appl icat ions including hydrogen storage,heterogeneous catalysis, dielectric coatings, andsuperconducting devices, such as superconductingradio-frequency (SRF) cavities [45] . In the nextgeneration accelerators, the performance of SRFcavities at moderate (i.e. 16 -19 MVm-1) and highelectric (i.e.>35MVm-1) fieldgradientsarecritical[46].Nevertheless,evenafterdecadesof research,asolid understanding of the microstructural defects
l imiting the medium and high field performanceis st i l l missing. What appears clear, however, isthat Niobium hydride, if present, can be a majorc ont r ibutor to the deg radat ion of the qua l it yfactor, Q, since hydride precipitates can only besuperconduct ing by prox imity ef fect , and theirpremature transition to thenormalstatewill leadtostronglosses[47,48]. Here, we present an atomic-resolution studyof the formation of ß-NbH precipitates at roomtemperatureinNbgrainsnearthecavitysurfaces.[49]Inaddition,wedemonstrate thatatomic-resolutionimagingisstillpossibleatLN2temperatures,althoughspatialdriftlimitsthedwelltimeperpixeltoaround16µs/pixel. Figure 6 shows a pair of HAADF and ABF
i mages , ac qu i red s i mu lta neously, of a ß -N bHprec ipitate in the [110 ] or ientat ion. W hi le theH A A DF image appears very simi lar to that ofpureNb, theABF image reveals additional atomiccolumns,whichcanbeidentifiedashydrogenatomiccolumns. The insert in Fig. 6 shows the proposedstructure of ß-NbH [110] as well as the calculatedimage contrast using the multi-slice method. Thecontrast in the image calculationsagreeswith thatmeasure in theexperimentalABF images.UsingacombinationofABFimaging,selectiveareaelectrondiffractionandEELS,weconfirmed that thephaseimageisß-NbH[110]. AccordingtothephasediagramforNb-hydride,there exist several other phases depending on thelocal hydrogen concentrat ion and temperature.Therefore, in order to ful ly understand the roleof hydride precipitates on the mid and high fieldgradientperformanceofNbbasedSRFcavities,animagingmethodcapableofatomic-resolutionat lowtemperature(e.g.LN2temperature) isneeded.UsingtheGatandouble tiltLN2stage,we imagedbulkNb[110] (Fig. 7(b)) and show that achieving atomic-resolution ispossibleat that temperature.However,dueto the increasemechanicalvibrationsasaresultof thenitrogenboilingoff,wehad todecrease thepixel dwell t ime to 16 µs /pixel and average over
severalimagestoachieveapresentablesignaltonoiseratio.Ifnecessary,moresophisticatednoisereductionprocedurescanbeemployed if repeatedacquisitionofshort-exposuretimeimagesisnotpossible.
XEDS analysis of defects in poly-crystalline CdTe
P o l y c r y s t a l l i n e C d Te t h i n f i l m b a s e dphotovoltaicdevicesarepresent leaders in thin filmsolar technology ([50] , [51]). Commercial successofCdTebased devices stems from the nearly idealband gap of the mater ial which very effectivelycouplestooursun’slightspectrumaswellaseaseofmanufacturingandlowcostofthemodules.However,to improve theconversionefficiencybeyond20%, itis critical tominimize theharmfuleffectsof grainboundar ies and latt ice defects in CdTe. Directatomic-scalestructuralandchemical investigation isdesirable inorder to identifyatomicconfigurationswhich can act as carr ier recombination centers.Likewise, it isnecessary to confirm thatpassivantsintroducedintoCdTeareabletodiffuseandbindtothetargetdefects. Besides arbitrarily oriented grain boundaries,stackingfaultsandƩ3twinboundaries,bothlyingon
Fig.6 SimultaneousHAADFandABFimagesat200kVofß-NbH[110]clearly showing the hydrogenatomiccolumnsintheABFimage.
17 JEOL News Vol. 49 No. 1 (2014)
{111}planes,areverycommonplanardefectsinCdTe.These defects have the correct nearest-neighborbonding,howeverstackingdisorderbecomesapparentin second (and beyond) nearest neighbors. Whilethese interfaceshavebeendeemedbenignas farascarrier recombination centers are concerned, theirroleinattractingimpuritiesandanti-sitepointdefectsremains unknown. In Fig. 8, we show an HAADFimageandXEDS imageofa stacking fault inCdTealong [110] . In this projection, Cd and Te atomiccolumnsformadumbbell-likestructurewith162pmcolumn separation. The XEDS data clearly showsatomicresolutionandthepolarityoftheplanardefectcanbereadilyidentified.Moreprecisely,thedirectionof thedumbbellsat the stacking fault is rotatedby250°about[110].Figure8showsthattheterminatingcolumnoftheleftsideoftheinterfaceisTe,followedbyCd.Suchdatawill allowus toquantifypossiblechanges instoichiometryacross twinboundariesandstacking faults,detect thepresenceofdopants,anddetermine theatomic structureofdislocationcoresterminatingsuchdefects. HAADF image and atomic-column resolvedXEDSmapofaLomer-Cottrelldislocationalong[-1-10]zone-axis inCdTearepresented inFig. 9.Morespecifically, two intrinsic stacking faultsare seen intheHAADF imageand theXEDSspectrum image
shows the integrated intensity of the Cd and TeL-peaks in the stacking faults.Thedislocationcoreis locatedat thevertexof two intersecting stackingfaults,and iscomposedof threeCdatomiccolumnsandasingleTecolumn(Cd3Te),ascanbeseenfromthe spectrum imagemap shown inFig. 8. It shouldbepointedouthere thatwithout theavailabilityoftheXEDSspectrum image, the identificationof theatomiccolumns in thedislocationcorewouldnotbepossible, since theatomicnumbers (Z) forCdandTeare tooclose tobedistinguishedusingHAADFimagingalone.
Ferritin in a graphene liquid cell
Nanoparticle growth, chemical reactions orbiochemical activity often occur in the presenceof a liquid. To study liquid sample in an electronmicroscope, several liquidcelldesignshavebecomecommercially available in recent years that enablematerialstobeimagedinacarefullycontrolledliquidenvironmentwithinthevacuumofaTEM.However,allsufferfromafewkeylimitationsthatdonotallowforultrahigh-resolutionimagingorspectroscopy:[52]1)twoSi3N4layers(50-500nmthick)usedaselectrontransparent windows and 2) the thickness of the
l iqu id surrounding the sample. In these l iqu idcells, the imaging resolution is usually limited tonanometers. Electron energy-loss spectroscopy(EELS) is degraded by multiple scattering eventsin the thick window layers, and the strong core-losssignalsassociatedwith thepresenceofSiandN[53]. Inaddition to the increased sample thickness,radiationdamage isa fundamentally limiting factorwhen examining beam sensitive materials and /orhydrous samples in TEM. It has been shown thatcoatingthespecimenwithcarbon,metalorgrapheme[54 -58] , or lower ing the temperature [57] havepositiveeffectsagainstradiationdamagebyreducingelectrostaticcharging,mass loss, lossofcrystallinity,or defect formation rate [54 -58] . These studiessuggestthatitispossibletoreduceradiationdamageto below breakage of covalent bonds. However,further reductionof radiationdamage isneeded forcharacterization of biological samples, since manybiologicalstructuresandfunctionsarerelatedto themuchweakerhydrogenbonds. We have developed a biocompatible approachof encapsu lat i ng l iqu id conta i n i ng sa mples i nmonolayers of g raphene. T h is not on ly a l lowsbiological samples to be directly imaged at atomicresolution inanative liquidstatewithout limitationsfrom the window thickness (see Fig. 10), but alsoenables nm-scale analysis using EELS to quantifyreactions inanaqueousenvironment [59]. Ithas tobe pointed out here that any imaging or chemicalanalysis of graphene liquid cells required that theelectronbeamenergy is loweredbelow100kV,andtheimagesshowninFig.10aretakenat80kV. F u r t her, we h ave s how n t h at t he energ ydepositedby the incomingelectrons isdissipatedbygraphenefromtheareairradiatedatarateequivalentto thebeamcurrentof severalelectronsperÅ2persecond[59].Thiswouldthereforeprovideareduction
of radiation damage, allowing for high resolutioni m a g i ng a nd s p e c t ro s c opy of b e a m s en s i t ivemater ia ls . Detai ls , such as indiv idual Fe atoms(see Fig. 10) or polypeptide of unstained protein,are resolved in a liquid environment. By carefullycontrolling the inducedelectrondose rate,wehaveshown that react ions, such as l iquid / gas phasetransition (bubble formation and condensation),or nanoparticle/ nanowire growth can be initiatedat selected locations in GLC, and recorded at nmresolution.This techniquealsoallowsus toperforma quantitative study of radiation damage effect ondifferent encapsulated samples, such as water orprotein,byobservinglocalreactionprocesses.
Conclusion
Nearly twoyearsafter thedeliveryof thenewJEOLARM200CFtoUIC,wehavedemonstratedthecapabilitiesof the instrument toperformwithsub-Åand sub-eVresolution inavarietyofenvironmentsusinga200kVprimaryenergyelectronbeam.Thespatialresolutionisdecreasedin1.0Åforlowenergyimagingat80kVwithoutany loss in theanalyticalcapabi l it ies of the instrument. In addit ion, theARM200CFiscapableofimagingsamplesinaliquidenvironment,atelevatedorcryogenic temperaturesineitherTEMorSTEMmode.
Acknowledgments
T he authors acknowledge suppor t for th i swork fromtheNationalScienceFoundation [DMR-0846748]and theUSDepartmentofEnergy(DOE-EE0005956).TheacquisitionoftheUICJEOLJEM-ARM200CF is supported by a MRI-R2 grant from
Fig.9 a)Atomic-column resolved XEDSmapoverlaidontopoftheZ-contrastimage in the<110>projection.L-Cdislocationcore(circled)isassociatedwiththetwodashedintrinsicstackingfaults.
19 JEOL News Vol. 49 No. 1 (2014)
the National Science Foundation [DMR-0959470].SupportfromtheUICResearchResourcesCenterisalsoacknowledged.
[ 8 ] Bel l , A .T. , T he i mpact of na nosc ienc e onheterogeneouscatalysis.Science,299(5613):p.1688-1691(2003).
[ 9 ] McKee,R.A.,F.J.Walker,andM.F.Chisholm,Crystal l ine oxides on si l icon: The f irst f ivemonolayers.Physical Review Letters,81(14):p.3014-3017(1998).
[10] McKee,R.A.,F.J.Walker,andM.F.Chisholm,Physical structure and inversion charge at asemiconductor interfacewithacrystallineoxide.Science,293(5529):p.468-471(2001).
[11] Ya n , Y. , S . J . Pe n nyc o o k , a nd A . P. Ts a i ,Direct imagingof local chemicaldisorderandcolumnarvacancies in idealdecagonalAl-Ni-Coquasicrystals.Physical Review Letters,81(23):p.5145-5148(1998).
[12] Shan, Z.W., et al., Grain boundary-mediatedplasticity in nanocrystall ine nickel. Science ,305(5684):p.654-657(2004).
[25] Klie,R.F., et al.,Examining the structureandbonding in complex oxides using aberration-corrected imaging and spectroscopy. Physical Review B,85(5):p.7(2012).
[26] D'A l fonso, A. J. , et a l . , Atom ic -resolut ionchemicalmappingusingenergy-dispersivex-rayspectroscopy.Physical Review B,81(10)(2010).
[38] Liang, Y., J. Curless, and D. McCready, Banda l ig n ment at epita x ia l SrTiO 3- Ga A s ( 0 01)heterojunction.Applied Physics Letters,86(8)(2005).
[39] Liang, Y., et al., Hetero-epitaxy of perovskiteoxidesonGaAs(001)bymolecularbeamepitaxy.Applied Physics Letters,85(7):p.1217-1219(2004).
[50] Tr iboulet , R . and S . P. , CdTe and RelatedCompounds ; Physics , Defects , Hetero- andNano-structures,CrystalGrowth,SurfacesandApplicationsed.Elsevier(2010).
[51] Chen, L . , et a l . , From atom ic st ructure tophotovoltaic properties in CdTe solar cel ls.Ultramicroscopy,134:p.113-125(2013).
[52] deJonge,N.andF.M.Ross,Electronmicroscopyof specimens in liquid.Nature Nanotechnology,6(11):p.695-704(2011).
[53] Holtz, M.E., et al., In Situ Electron Energy-LossSpectroscopy inLiquids.Microscopy and Microanalysis,19(4):p.1027-1035(2013).
[54] Salih,S.M.andV.E.Cosslett,REDUCTION IN ELECTRON-IRRADIATION DAMAGE TO ORGANIC COMPOUNDS BY CONDUCTING COATINGS.PhilosophicalMagazine,1974.30(1):p.225-228(1974).
[56] Fryer,J.R.andF.Holland,HIGH-RESOLUTIONE L E C T R O N - M I C R O S C O P Y O FMOLECULAR-CRYSTALS .3.RADIATIONPROCESSES AT ROOM-TEMPERATURE.Proceedings of the Royal Society of LondonSeriesa-MathematicalPhysicalandEngineeringSciences,393(1805):p.353-&(1984).
[57] Zan,R.,etal.,ControlofRadiationDamage inMoS2 by Graphene Encapsulation. Acs Nano,7(11):p.10167-10174(2013).
[58] Algara-Sil ler, G., et al., The pristine atomicstructure of MoS2 monolayer protected fromelectronradiationdamagebygraphene.Applied Physics Letters,103(20)(2013).
[59] Wang, C., et a l . , High-Resolution ElectronMicroscopy and Spectroscopy of Ferritin inBiocompat ible Graphene Liqu id Cel ls andGrapheneSandwiches.Advanced Materials,26:p.3410–3414(2014).
21 JEOL News Vol. 49 No. 1 (2014)
Quantitative Characterization of Magnetic Materials Based on Electron Magnetic Circular Dichroism with Nanometric Resolution Using the JEM-1000K RS Ultra-High Voltage STEM
Shunsuke Muto1, Jan Rusz2, Kazuyoshi Tatsumi1, Roman Adam3, Shigeo Arai1, Vancho Kocevski2, Peter M. Oppeneer2, Daniel E. Bürgler3 & Claus M. Schneider3
1Division of Green Materials, EcoTopia Science Institute, Nagoya University, Japan2Department of Physics and Astronomy, Uppsala University, Sweden3Peter Grünberg Institut, Forschungszentrum Jülich GmbH, Germany
Electron magnetic circular dichroism (EMCD) allows the quantitative, element-selective determination of spin and orbital magnetic moments in a manner similar to its better-established X-ray counterpart, X-ray magnetic circular dichroism (XMCD). As an advantage over XMCD, EMCD measurements are performed using transmission electron microscopes, which are routinely operated at sub-nanometer resolution. However, because of the low signal-to -noise ratio of the EMCD signal, it has not yet been successful to obtain quantitative information from EMCD signals at the nanometer scale. In the present article, we demonstrate a new approach to EMCD measurements that takes most of the higher accelerating voltage of the incident electrons, which considerably enhances the applicability of the technique. The statistical analysis introduced here yields robust quantitative EMCD signals. In the present scheme, quantitative magnetic information can be routinely obtained using electron beams of only a few nanometers in diameter, without imposing any restrictions on the crystalline order of the specimen.
Introduction
Electronmicroscopy,whenused incombinationwithseveraltypesofspectrometers,providesnotonlymagnified images with spatial resolutions down totheatomicscalebutalso localchemical informationsuchaschemicalcompositionandelectronicstates.Inparticular,electronenergy-lossspectroscopy(EELS),whichanalyzeselectronstransmittedthroughasample,can be used to explore a rich variety of materialproperties. Itsemerging significance innanometricanalysis is growing further, in competitionwith itscounterpart, the X-ray absorption spectroscopy,measuredatsynchrotronX-raysources. Modern synchrotron X-ray sources capableof delivering intense radiation with a well-definedpolarizationhaveprovided insight intonotonly thestructural and chemical but also magnetic aspectsof solids [1].Theusefulnessof synchrotron X-raysinmagnetismcanbeattributed to thediscoveryofan importantphenomenoncalledasX-raymagnetic
circulardichroism(XMCD)[2-4].XMCDoriginatesfromthedependenceof theabsorptioncross-sectionon the sample magnetization with respect to thephotonhelicity.Spinandorbitalmagneticmomentscan be quantitatively determined in an element-specific manner using a simple integration of theXMCDspectraemployingsumrules[5,6].AnanalogtoXMCDistheelectronmagneticcirculardichroism(EMCD)technique,inwhichelectronsaretransmittedthroughamagneticsample inatransmissionelectronmicroscope (TEM) [7,8] . EELS measured at corelevels indifferentpositions in thediffractionplanecan then be employed to extract element-selectivemagneticinformation. Since thepossibleexistenceofEMCDwas firstoutlined in 2003 [7] using methods reminiscent ofsymmetry-selectedEELS[9], recent theoreticalandexperimentalprogress in theEMCDtechniquehasled to improvements in its spatial resolution [10,11],facilitated theoretical understanding [12-16] , andalsogiven rise to the firstquantitative studies [17-19]of spinandorbitalmagneticmoments.AlthoughTEM has been expected to offer a much higherspatialresolutionthanthatcurrentlyobtainablewithXMCD,high-resolutionquantitativeEMCDhasnotyet been achieved, because of its inherent low net
Division of Green Materials, EcoTopia Science Institute,NagoyaUniversity,Furo-cho,Chikusa-kuNagoya464-8603,Japan
signalstrength.BecauseEMCDismeasuredatcore-leveledges, thesignal strengthexhibitsapower-lawdecayasafunctionofenergyloss.Anotherchallengeis the necessity to measure EMCD at diffractionangles thatdonot coincidewithBragg spots in thediffractionplane,which reduces the signal strengthevenfurther.Moreover,beamandsampleinstabilities,coupled toapotentialdamageof the samplebyanintenseelectronbeam, limit theacquisition time. Inmostexperimentsperformed todate, a largebeamcurrentwasusedinanoptimizedgeometry—namely,the 2-beam [8,14,17,20,21] or 3-beam [11,17,19,22]Braggcondition—which in turnnecessitates theuseofasingle-crystallinespecimeninapreciseandstableorientation. Inour recentarticle [23],wehad introducedanew scheme for measuring EMCD by utilizing anultra-high voltage scanning transmission electronmicroscope(UHV-STEM)equippedwithEELS.Theconventionalprocedureofoptimizing the signal-to-noise ratio (SNR) ina fixedgeometry is inherentlypronetobeamdamageandstability issues.Thereforewerapidlycollecteda largenumberof independentspectraand, inaddition, introducedanewstatisticaltechniquefortheanalysis,efficientlyovercomingtheaforementionedrestrictionsandreducing the lateralresolutionofquantitativeEMCDto thenanometerrange.Thepowerof thetechniquewasdemonstratedbyanalyzingapolycrystallineironsample[23].
T h e o r e t i c a l b a c k g r o u n d a n d simulation
T he u nderly i ng idea for the measu rementscheme involves the decomposition of the inelastictransition matrix elements (mixed dynamic form-factor;MDFF)usingadipoleapproximation intoalinear combination of non-magnetic and magneticterms,includingtheiranisotropies[15].Inamaterialwith a cubic crystal structure, such as bcc iron,the anisotropic terms are negligible, and a simpleexpressionfortheMDFFS(q, q',E)remains:
where q and q’ are momentum transfer vectors, Eis theenergy loss,and N(E), MZ (E) are thewhite-linenon-magneticsignalandmagneticEMCDsignal,respectively. In this equation, it has been assumedthat the sample ismagnetically saturatedalong thez-axisinthemagneticfieldoftheobjectivelens. At an arbitrar y relat ive or ientat ion of thecrystallineaxes, incomingbeam, anddetector, thescattering cross-section∂2
where A(Ω) and B(Ω) are coefficients that dependon the d i f f rac t ion a ng le Ω . C onsequent ly, weestablish that forabcc-ironpolycrystalline samplewith sufficiently large grains, every spectrum is al inear combination of a white-l ine non-magneticsignal N(E) and a magnetic EMCD signal MZ (E),albeitwitha priori unknowncoefficientsA(Ω)andB(Ω). Onthebasisofthetheoreticalmethodoutlinedabove, we performed simulat ions to obta in thedichroic signaldistribution in thediffractionplanefor a randomly oriented polycrystalline sample. Abcc iron supercell of 4 × 4 × 18 lattice dimensions(approximately1nm×1nm×5nm)containing576ironatomswasmelted,annealed,andslowlycooledin accordance with classical molecular dynamics,considering the interatomic potential proposed byMendelev et al. [24] . The procedure was repeatedmultiple times, and in some cases, we obtained amodel structure with several crystalline grains ofdifferentorientations(Fig.1(a)).Thecorrespondingelectron dif fraction pattern is presented in Fig.1(b).Wecalculated theelectronic structureof thismodel structure using density functional theoryand estimated an average magnetization of 2 .19μ B per atom. For calculat ions of the dynamicaldiffraction, this model structure was periodically
repeatedinthex,andydimensionsandilluminatedbyaplanewave.ThecalculatedFe-L3edgeenergy-filtered diffraction pattern of the magnetic signalalong thez-direction [16] ispresented inFig. 1(c).Approximately 600 beams were considered in thesimulation [12,25]. The approach proposed in thisarticlehasbeen inspiredby simulateddistributionsofdichroic signals in thediffractionplane [12,15],wh ic h s u g ge s t t h at E MC D i s p re s ent a l mo s teverywhere in the diffraction plane, despite non-trivialvariationsinstrengthandsign.
Experimental
A 30 -nm-th ick bcc i ron layer and a 3 -nm-t h ic k A l c ap l ayer (to prevent t he ox id at iono f Fe) were d e p o s i t e d o n 5 0 - n m - t h i c k S i 3N 4membranes by thermal evaporation in an ultra-h ig h vacuu m molecu la r bea m epit a x y (U H V-M BE) system. The d isordered structure of themembranes (nanocrystalline or amorphous) led toapolycrystallinemorphologyof themetallicFe/Alfilms. Air exposure after the deposition oxidizedtheAlcap layer toadepthof1.5 to2nm.BecausetheAllayerwas3-nmthick,aclosedAlOx layerwasmaintainedeveninthepresenceofsurfaceroughness,as is likely forapolycrystalline film.Thegrain sizewas~30nm,which iscomparable to thethicknessoftheironlayer(Fig. 2(a)). Because oxidation of the iron film could havesubstantialeffectsontheintensityratiooftheL3andL2edges [26], the filmwasexaminedwithEELStoprobeifanyoxidationoftheironoccurredbeforeorafter theEMCDmeasurements.Thiswasachievedby taking advantage of the fact that the oxygen Kedgecanbeeasilydistinguishedbetweenaluminumand iron oxides.Nevertheless, no ironoxideswereobservedwithin thedetection limitofEELS(<1at%). ThemeasurementswereperformedusingEELSspectral imagingwithanultra-highvoltagescanningtransmission electron microscope (UHV-STEM),
the JEOL JEM-10 0 0K RS of Nagoya University(Fig. 2(b)).TheSTEMwasoperatedat 1MV,andit was equipped with an equivalent Gatan ImageFilter (GIF) Quantum specially designed for thisacceleratingvoltage.The samplewasheated to200°Cduringtheexperimentsothatmeasurementswerecontamination-free.Thefullwidthathalfmaximum(FWHM)of the zero-losspeak (ZLP)was~2.6eVbecause the beam current had been increased toensureanincreaseinspectralcounts. Aschematicdiagramofourexperimental setupispresentedinFig. 3.Theelectronbeamwasfocusedto~5nm,and thesamplewas scannedwith15×15pixelswitha scanstepof20nm.Thisconfigurationensured that theelectronbeam illuminatedrandomgrains, typicallyoneora fewgrainsateachscannedposition. The convergence semiangle of the probewas ~1 mrad. EELS spectra were recorded with adispersionof0.5eV/channel,exposure timeof30 sforeachspot,andacollectionsemiangleof~1mrad.The detector aperture was placed adjacent to thetransmitted beam in the diffraction plane to avoidincludingthe intensetransmittedbeam.Undertheseconditions,theFe-L3peakintensitiesrangedbetween8,000 and 12,000 counts per spectrum. The sameexperiment was repeated independently for threedifferentareas.
Results
The acquired dataset forms the starting pointforourstatisticalprocedureleadingtotheextractionof the EMCD spectrum. In the measurement, weuseanultra-high-voltage (1MV)electronbeam tosignificantlyreduce theeffectofmultiplescattering,which could otherwise distort the EMCD signal[18,19,27].Our theoretical simulations indicate thatthenetEMCDsignalatanacceleratingvoltageof1MVis~20%larger thanthatobtainedat200kVforaspecimenthicknessbetween25and40nm[28],asshowninFig. 4. The obtained spectral image data-cubes were
first treatedbyaligning thedrifted peakpositionsandremovingX-rayspikes (verybright/darkspots).Wethenappliedthepre-edgebackgroundsubtractiontoextracttheFe-L2,3peaks,afterwhichweappliedalow-passfilterwithawindowwidththatwasvariablefrom3.5to10eV. We d i d n o t a p p l y t h e F o u r i e r- r a t i o d e -convolution to remove the plural scattering effect,becausetheplasmonpeak-to-zerolosspeakintensityratio was less than 6%, causing only a negligibled i f ference to the f i na l m L/ m S rat io w ith i n theexperimentalaccuracy [18].Thereducedsizeof theerror,attributedtothemultiplescattering,isanotheradvantageofusingamegavolt-STEMinaddition totheimprovedstrengthoftheEMCDsignalcomparedwith loweraccelerationvoltages, asmoreexplicitlydiscussedin[28]. Each measured spectrum was normalized byscal ing the post-edge intensity, integrated overthe range 740 –750 eV, to one. The effect of thisnormal izat ion on the quantitat ive analysis wasdiscussed inaprevious study [29]. In the first step,we examined the difference spectra for al l pairsof the spectra in the dataset. As described above,each spectrum can be considered as consisting oftwocomponents,amagneticone(theEMCD)andanonmagneticone.Whenaspectrumisacquired inarandomorientation, theweightof each componenti s a l so random ; however, once the spectra arebackground-subtracted and normalized post-edge(where themagneticcomponent isalwayszero), thenonmagnetic component will be identical in everyspectrumandwill thereforebeeliminatedby takingthedifferencespectrum. Afterpre-processing the225measured spectraineachdataset,wecalculatedadifferencespectrumfor every pair of spectra present in the dataset,which yielded a total of 25,200 difference spectraper dataset. Each of the difference spectra weresubsequentlyexaminedforthepresenceofanEMCDsignalusingcriteria thatprobe the relative signsofintensityaroundtheL3andL2edges:
where ∫L 3 dE represent s the EMC D s ig na lintensity integratedover theFe-L3peakregionandI (atL3peak) denotes theEMCDsignal valueattheL3peakposition.Allthedifferencespectrathatmetourselectioncriteriawerefirstadjusted insignsuch that the L3 signal was positive (cf., Fig. 7(a))and then summed.Of theentire set, approximately20–25% of the pairs always displayed signs of theEMCD signal and thus met our selection criteria.The selected subset (~ 5,000 difference spectra)originated from more than 100 independent rawspectra among the entire set of 225 spectra, whichensures that the stat ist ica l noise is reduced bya factor of 10 compared to the average noise ofindividualrawspectra. Thefinalsummedsignal intensitiesandprofilesfor the threedatasetswerenearly identical (seeFig. 5(a)).Minordifferenceswereobserved in thepre-edgeandpost-edgebackgroundregions,whichwererelated to varying fluctuations between individual
datasets. These differences can be attr ibuted toinaccuracies in the extraction of the power-lawbackgroundsignal,which iscausedbyextrapolationfromanoisypre-edgesignal. We averaged the three accumulated EMCDsignals toobtain the finalEMCDspectrum,whichis presented in Fig. 6. The EMCD signal intensityfractionwasestimated tobe2–2.3%of theFe-L2,3signal intensity (Fig.5(b)),which is consistentwiththetheoreticalestimate(seeFig.4).ThefinalEMCDsignal isa resultofaveragingover~15,000differentspectra and was thus observed to be statisticallyrobust. AfterextractionofanEMCDsignalwithagoodSNR, theEMCDsumrules [12]wereapplied to thesignal toevaluate theorbitalmoment/spinmomentratio,whichisgivenby[13,14]:
mL=
2∫L3(E)dE+∫L2(E)dE =
2q . . . (4)─ ─.──────────......──
mS ..3∫L3(E)dE -2∫L2(E)dE 9p -6q
whereqisanenergyintegraloftheEMCDspectrum(E) over both edges and p is an energy integralover theL3edgeonly.Without lossofgenerality,wecanrescale theEMCDspectrumor its integral suchthatp=1 (Fig.6(a)).Thus, theorbital tomagneticmoment ratiomL/mSbecomesa functionofq only.The application of the low-pass filter can possiblylead to the overlap and cancellation of the EMCDsignalsof theL3andL2peaks,whichcouldequallyreducethemagnitudeofbothpeaksandconsequentlyenhance the effect of their difference. Moreover,theapplicationof the low-pass filter canaffect thepost-edgenormalization [29].The lattercouldalterthe m L/m S ratio determined using the sum rules.ThemL/mS ratiowashenceplottedasa functionofthe low-pass filter window width, as illustrated inFig. 6(b).Theminimalandmaximalq valueswereextractedfromthepost-edgeregionbetween740-750eVenergyloss,fromwhichtheminimalandmaximalm L/m S were ca lcu lated. In Fig. 6 (b) , the mL/m Svalues with error bars are plotted as the middlepointsof theseupperand lowerbounds.Theshadedareaisboundedbylinearregressionlines(greysolidlines)at the lowerandupperboundvalues.ThefinalmL/mSwasevaluatedtobetheinterceptofthelinearregressionline(greybrokenline)ofthemiddlepoints+/-halfof the intervaldefinedby the interceptsoftheboundlines.Bylinearextrapolationtozerofilterwidth,we finallyobtainedmL/mS=0.0429±0.0075,which is ingoodagreementwith thevalueobtainedbyXMCDforbcc iron,0.043[3].Toourknowledge,this result is the firstquantitativeEMCDdetectionperformedonapolycrystallinefilm.
Discussion
Themainerrorsourceforourextractionmethodof theEMCDspectrumstemsfromthe imperfectionof theemployeddetectors: therecordedrawspectramay includesignificant fractionsofsystematicerrorsthat originate from instrumental instabilities andconcomitant intr insic l imitations, in addition tostatisticalrandomnoise. For the reasons statedabove,weobserved that
(2014) Vol. 49 No. 1 JEOL News 26
conventional statistical signal extraction methods,such a s t he one ba s e d on mu lt iva r iate c u r veresolution technique [29],werenotefficient. Itwasratheressential that thedataare first sortedso thatonly thedatasetscontainingsignificantsignal levelsareselectedonthebasisoftheselectioncriteria(Eq.(3))appliedtothedifferencespectra. Toeliminate thepossibility thatourproceduremight lead to a fake EMCD-l ike signal prof i leex tracted f rom a random noisy data ar ray, weperformedanequivalent setofmeasurementsonanantiferromagneticNiO film, fromwhichnoEMCDsignal was expected because of the cancellation ofmagnetic signals from symmetrical ly equivalentatoms with antiparallel moments. The sample wasaNiOpolycrystalline film(grainsize~30nm, film
thickness ~ 30 nm), and the measurements wereperformed under the same conditions (at 165°C,which isbelowtheNéel temperature)as for thebcciron film. Subsequently, the same data processingprocedure was appl ied to the Ni L2 , 3 white-l inespectra. The noise level was estimated by taking thesquare-rootof the sumof squaresof thedifferencespec t ra bet ween t he raw a nd med ia n f i l tered(averagedoverevery10channels)spectradividedbythenumberofchannelsused for theanalysis.Then,theSNRwasestimatedbydividing the intensityattheL3peakpositionof theaveragedEMCDsignalby the noise level. The estimated SNR was 2-3 forbcc iron,whereas itwas less than0.5 forNiO.Theaveraged signal extracted from the NiO data was
actuallyobserved toexhibit abimodalEMCD-likeprofile;however,theprofilesignificantlyvarieduponchanging the integrationwidth for the first selectioncriterion. Additionally, the q-value from the sumrulewassometimespositiveandsometimesnegative,withoutacleartrendwithrespecttothefilterwidth.The bimodal prof i le is a result of the select ionrequirements(Eq.(3)),whichextractallEMCD-likedifference-spectranaturallyoccurring in the setofcompletelyrandomdifference-spectra.However, thesignalprofilesextractedfromtheironfilmexhibiteda stable feature.Thesituationcanbebestobservedfromtheoverlaid typicaldifferencespectrameetingthe selectioncriteria for the twocases,as shown inFig.7(a)and(b).For the iron film,onecanclearly
recognize an approximate EMCD signal profi le,whereas for theNiOfilm, thespectravisuallydifferonlyslightlyfromrandomnoise. Asanadditionaltestofthestatisticalrobustnesso f t he p o lyc r y s t a l l i ne i ron E MC D re su l t , i ncontrast to the averaged NiO signals originatingfromnoise,wevaried theL2,3 peakenergies in theneighborhood of their experimental values in ourextraction procedure. At every combination of theL2,3 peak energies, an averaged EMCD-like signalwas constructed. For every extracted signal, wecalculated its norm as a sum of squares within theedgeintegrationintervals.IfatrueEMCDsignalwaspresentinthedata,thissumofsquareswouldhavealocalmaximumnear thephysicallycorrectvaluesof
thepeakenergies. Fig. 7(c) and (d) present maps of the normsca lcu lated for a med ia n f i l ter of 7 eV a nd a nintegration range ±5 eV around the assumed peakenergies,with theL3energyas its abscissaand theL2 energy as its ordinate. A green circle denotesthe position of the L3/L2peak energies, where theexperimental white-line spectra exhibit maximalcounts.Thestrikingdifferencebetweenthetwoplotsclearly demonstrates the validity of our method.Thepolycrystalline irondata revealadistinctpeakaroundtheexpectedvaluesofthepeakenergies,withamaximumwithin1eVof theexperimentalvalues.Conversely, theNiOspectradonotexhibitanysuchfeaturearoundtheexpectedpeakenergies.
Conclusion
I n th i s a r t ic le we demonst rated a methodu t i l i z i n g U H V- S T E M - E E L S , t h a t l e a d s t ostatisticallysignificantEMCDspectra.Weemphasizethat, even thougha5-nmnanoprobewasused, thespectraldifferenceswerecomputed fromtheentiredata stack, rendering the effectively sampled areato be of the order of 100 nm. This area could bereducedbychanging thescanningpattern,providedthat sufficiently randomorientationswere includedineachdataset.Consequently,ourapproachallowsquantitativeEMCDstudiesofnon-singlecrystallinesamplesatthenanoscale,andhencepavesthewayforawiderangeofapplicationsofEMCDexperimentsinthefieldofnano-magnetism.
Acknowledgments
We a re ver y g ratef u l to the eng i neer s , i nparticular Drs. M. Ohsaki and S. Ohta, of JEOLLtd. for theirdedicatedefforts inbuilding thenewUHV-STEM at Nagoya University. A part of thisworkwassupportedbyaGrant-in-AidonInnovativeAreas"NanoInformatics"(grantnumber25106004)fromtheJapanSocietyof thePromotionofScience,SwedishResearchCouncil,andSTINT.
Department of Electronic Science and Engineering, Kyoto University
In this article, recent progress in broad-area photonic-crystal lasers based on photonic band-edge effect is described. It is shown that unique beam patterns can be generated by designing photonic crystal structures. Moreover, it is demonstrated that watt-class high-power, high-beam-quality, surface-emitting, lasing oscillation has been successfully achieved. These results represent an important milestone for innovation in the field of lasers because it provides a route towards overcoming limitations in applications that suffer from low beam quality, which opens the door to a wide range of applications in material processing, laser medicine, nonlinear optics, sensing and so on.
Introduction
It iswellknownthat semiconductordistributedfeedback lasers possess a one-dimensional lattice,and that the forward-propagating wave undergoesBraggreflectiondueto thisgrating,beingdiffractedto theoppositedirection.Theresulting forwardandbackward-propagatingwavescouplewitheachothertogeneratea standingwave, forminga cavity.Thisis equivalent to the fact that in a one-dimensionalphotonic crystal the cavity loss is smallest at thebandedges,whichareatbothendsof thephotonicbandgap,givingrisetoastatethatcausesoscillation.When this idea is extended to photonic crystalswith two-dimensional periodicity, one can makeuse of the coupling of optical waves due to Braggreflectionwithin thetwo-dimensionalplane inorderto forma standingwavestate thatcovers theentiresurface of the plane [1,2] . As a result, it becomespossible to obta in an osc i l lat ion mode with anelectromagnetic field distribution that is perfectlydefinedateach latticepoint in the two-dimensionalcrystal.Theopticaloutput canbediffracted in thedirectionperpendicular to theplaneof thecrystal,thus real izing a surface-emitting characteristic.Two-dimensional photonic crystals hence enabletheconstructionof surface-emitting lasers inwhichnotonly the longitudinalmodeof lasing isdefined,but also the beam pattern, usually referred to asthe transverse mode. Fur thermore, it becomespossible to realizeanovel laser thatoscillates inasingle longitudinaland transversemode,nomatterhow large the surface area is, which surpasses aconventionalconceptinthefieldoflaserresearch. T he f i r st sem iconductor laser to be basedon this principle was real ized in 1999 [1] . Since
then, in addition to the demonstration of room-temperature continuous lasing, it has been shownthat two-dimensionalphotoniccrystalscangeneratebeamswithcontrolledpolarizationandpatterns; forexample, a doughnut-shaped beam can be formed,which is expected to be focusable to sizes smallerthan thewavelength [2-6].Theothernotable recentdevelopmentsusing thisprincipleare therealizationofacurrent-injection-typeblue-violetsurface-emittingoperationandelectronicallybeam-steeringoperation[7,8].Wediscuss thecurrent stateof theart in thefollowingsections.
Basic Device Structure and Operation Principle
Figure 1 showsanexampleofa laserbasedonthe two-dimensional photonic crystal band-edgeeffect. This laser consists of two wafers, A and B;waferA includesanactive layer for the injectionofelectrons and holes, and a photonic crystal as theupper-most layer. The integration of wafers A andB results in thephotonic crystalbeing sandwichedto complete the device. As shown in the insert ofFig. 1, this photonic crystal has a square latticestructureandisdesignedsuchthattheperiodicity intheГ-Xdirectionmatches theemissionwavelengthin theactive layer. In thisdesign, lightpropagatinginacertainГ-Xdirection isBraggdiffracted to theopposite (-180°) direction, as well as to the -90°and 90°directions; the fourequivalent lightwavespropagating intheГ-Xdirectionthencoupletoforma two-dimensional cavity. More precisely, higher-orderBlochwaves inaddition to these fundamentalfourwavesare involved for theconstructionof two-dimensionalcavitymode(seeFig. 2(a)).Figure2(b)showsthephotonicbandstructureofthiscavity.ThelasingmodeoccursatthebandedgesindicatedbythereddotsattheГ-pointsofthefourbands,A,B,Cand
Broad Area Coherent Oscillation and Beam Pattern Control
To construct the device shown in Fig.1, thephotonic crystal was fabricated by electron beamlithographyanddryetchingtechniqueandembeddedin the device by a wafer bonding technique. Thedevicesuccessfullyoscillatedcoherentlyinthebroadareaas shown inFig. 3. It isapparent thata single-wavelengthoperationwasachievedacross thedevicedespitethelargelasingareaof150×150μm. Such an ability to realize broad-area coherentoscillationenablesus toproduceveryuniquebeampatterns,whichcannotbe realizedbyconventionalsemiconductor lasers. Because the pattern of thesurface-emittedbeamfromaphotoniccrystal lasercanbedeterminedby theFourier transformationof
its two-dimensional electromagnetic distribution,the beam pattern can be tailored by varying theelectromagneticdistribution in the two-dimensionalplane, that is,bychanging thecoupling stateof thelightthatpropagatesinvariousdirectionsinthetwo-dimensionalplane.Oneeffectivemethodofachievingthis is tovary the shapesand spacingof the latticepoints in the photonic crystal. Figure 4 (a) and (b)show the electromagnetic field distribution in theunit latticeofacrystalwhentheholesplacedat thelattice points are circles and equilateral triangles,respectively.Changing the shapeof theholes fromcirculartotriangularremovesthefour-foldrotationalsymmetry in theelectromagnetic fielddistribution;thereisnosymmetryinthex-directionfortriangularholes.Figure4(c)–(g)showtheelectromagneticfielddistributionsover theentire crystal in caseswhereshiftsofthelatticepointswereintroducedinordertoincreasethe latticespacing ineither the longitudinalor transversedirections.Figure4(c) represents thecasewithnoshift,whereasFig.4(d)– (g) representincreasing numbers of shifts. It is apparent thatshifting the lattice spacing reverses the polarityoftheelectromagnetic fielddistributionat theposition
Surface emitting region
Active layer Lower clad
Photonic crystal
Upper clad
Substrate
Electrode
Electrode
Carrier block
Contact layer
A
B
- X
- M
2 /a
(a) (b)
M X X k//
Freq
uenc
y (c/a
)
X M
Band Edges
k//
A
B
C
D
Fig.1
Fig.2
Schematic of an exampleofa laserbasedonthetwo-dimensionalphotoniccrystalband-edgeeffect.The insetshows the photonic crystalwithasquarelatticestructure.
of theshift.Further increasing thenumberof shiftsrepeats the reversalof theelectromagnetic field. Itis clear that the electromagnetic field distributionin the plane can be controlled in various ways byappropriatedesignofthephotoniccrystal. We fabricated devices with various differentphoton ic c r yst a l s t r uc t u res , a s show n i n F i g . 5 (a) –5(f) . Al l of these devices exhibited lasingoscillationat roomtemperaturewitha stable singlemode.Theright-handpanelsofFig.5(a)–9(f) showthe corresponding measured beam patterns. Aninteresting array of patterns was obtained rangingfroma singledoughnut shape to twofolddoughnut,fourfolddoughnut,andregularcircular shapes.Thebeam divergence was extremely narrow, reflectingthe fact that these are large area coherent laserosci l lations. The device in Fig. 5(a) has regularcircularholes,andthecorrespondingelectromagneticfield distribution exhibits well-defined rotationalsymmetry as shown in Fig. 4 (a) . When the laserl i g ht c o r re s p o nd i n g t o t h i s e l e c t ro m a g net i cf ield d i s t r ibut ion i s output to f ree spac e, theelectromagnetic f ield at the center of the beamcancelsouttoyieldadoughnut-shapedbeam.Ontheotherhand,triangularlatticeholes(Fig.5(f))removetherotational symmetryof theelectromagnetic fielddistribution,as shown inFig.4(b).Thecancellationeffectat the centerof thebeam inFig. 4(a) is alsolost, yieldinga clean circularpattern. In this case,the polar izat ion is a lso d i f ferent , being l inear.Introducing such a nonsymmetrical effect is a keyfactor in achieving high optical output power byenabling a greater optical extraction efficiency intheperpendiculardirectionasdescribed in thenext
section.Note that in thenextsection,asignificantlyasymmetric structure in the formof right-isosceles-triangle-shaped air holes was employed for muchhigherpoweroperation.
In the previous section, the photonic crystalwas embedded in the device by a wafer-bondingtechnique,wherebondedinterfacemaycontainmanydefectstateswhichabsorbthelasinglightandmakeitdifficulttorealizehighpoweroperation.Toavoidsuchdegradationofperformance,wechangedthemethodtointroducethephotoniccrystal intothedevicefromawaferbondingtoacrystalgrowthtechniquesuchasorganometallicvaporphaseepitaxy(OMVPE)[10,11],andfoundthattheairholesofphotoniccrystalcanberetainedevenbythecrystalgrowthtechnique.Figure 6 (a)showstheschematicof thedevicefabricatedbytwo-step OMVPE. Note that the growth directionwas downward. Figure 6 (b) shows a plan-v iewscanning electron microscope (SEM) image of thephotoniccrystalwithright-isosceles-triangle-shapedairholesbefore thecrystalgrowth.A typicalcross-sectionalSEMimageofarowofairholesembeddedbyOMVPE is shown inFig. 6(c). It is clearly seenthat theairholesweresuccessfullyembedded in thedevice,wheretheairholesbecomenarrowertowardsthelowersideofthedevice,whereastheupperpartsoftheairholesmaintainanalmostuniformshapewithverticalsidewalls. I thendescribe the lasingcharacteristicsof the
33 JEOL News Vol. 49 No. 1 (2014)
Fig.6 (a) Schematic of the devicefabricatedbytwo-stepOMVPE.(b) p lan-view SEM imageof the photonic crystal withright-isosceles-triangle-shapedair holes formed by electronbeamlithography(JEOLJBX-6300FS) and dry etching. (c)Atypicalcross-sectionalSEMimage of a row of air holesembeddedbyOMVPE.
fabricateddevice [12]under the room-temperature( RT ) c o nt i nu ou s -wave ( C W ) c o nd i t io n . T hecorresponding experimental results are shown inFig. 7(a)–(c).AmaximumCWoutputpowerof1.5Wat2.5Awasachievedwithanarrow,single-lobedbeamof lowdivergence.Whentheoutputpowerwasless than0.5W, thebeamqualitywasquantitativelyevaluated by measuring the value of M 2. For anidealGaussianbeam,M2 isknowntobeunity,but itincreaseswhen thebeamquality isdegradeddue tothetransversemultimode.MeasurementsofM2wereperformedunderroom-temperatureCWconditions,andwefoundthatM2waskeptalmostat~1.0inboththe x- and y-directions up to a power up to 0.5 W,indicatingthatafundamentalsingletransversemodeismaintained.Note that thebeamdivergenceanglewaslessthan3°evenat1.5Wpowerlevel. Lasers with such a narrow beam divergenceshouldenableuniqueapplicationsthatdonotrequireany lens. We examined the direct irradiation of asheet of paper placed 8.5 cm from the PCSEL todemonstrate such lens-free potential under CWoperationat25°C.Thelightoutputwassetto0.86Watacurrentof1.7A.Thepaperwasburnt,formingasmallhole immediatelyafter radiation,as shown inFig. 8.Althoughthisisjustasimpledemonstration,itshowsthepotentialoflens-freeapplications.
Conclusion (or Summary)
Ihave described the current statusand recentdevelopments in the fieldofphotonic-crystal lasers.Ithasbeenshownthat theband-edgeeffectof two-dimensional photonic crystals enables large-areasingle longitudinal and transverse mode lasingoscillation,aswellascompletecontroloverthebeampatterns obtained. It has been also described thata device with an output power exceeding 1.5 wattunderCWconditionatRT.Ourworkrepresentsan
important milestone for innovation in the field oflasersbecauseitprovidesaroutetowardsovercominglimitations inapplicationsthatsuffer fromlowbeamquality, which opens the door to a wide range ofapplications inmaterialprocessing, lasermedicine,nonlinearoptics,sensingandsoon.
Acknowledgments
The author thanks members of Noda’s Lab.,Kyoto Un iver s it y, a nd Roh m a nd Ha ma matsuPhotonics for the col laboration. This work wassuppor ted i n pa r t by JST, ACC EL & C R E ST,C-PhoST,MEXT,Japan.
References
[ 1 ] M. Imada, S. Noda, A. Chutinan, T. Tokuda,M. Murata, and G. Sasaki : "Coherent two-dimensionallasingactioninsurface-emittinglaserwithtriangular-latticephotoniccrystalstructure,"Appl. Phys. Lett.,vol.75,pp.316-318(1999).
[ 2 ]S. Noda, M. Yokoyama, M. Imada, A. Chutinan,M.Mochizuki,“PolarizationModeControlofTwo-Dimensional Photonic Crystal Laser by Unit CellStructureDesign,”Science,vol.293,pp.1123-1125(2001).
[ 3 ]M . I mada , A . C hut i na n , S . Noda , a nd M .Mochizuki , "Mult id irect ional ly d istr ibutedfeedbackphotoniccrystallasers",Physical Review B,Vol.65, No.19,pp.195306(2002).
[ 4 ]K. Sakai, E. Miyai, T. Sakaguchi, D. Ohnishi,T. Okano, and S. Noda, "Lasing band edgeidentification for a surface-emitting photonic-crystal laser,"IEEE Journal of Selected Area in Communications,vol.23, no.7,pp.1330-1334(2005).
[11] K . H i rose, Y. Ku rosa ka , A . Wata nabe, T.Sugiyama,Y.Liang,andS.Noda,“HighpowerPhotonic-CrystalSurface-EmittingLasers,”The10thConferenceonLasersandElectro-OpticsPacificRim(CLEO-PR2013),ThI1-4(2013).
[12] M. Nishimoto, K. Ishizaki, K. Maekawa, K.Kitamura, and S. Noda, "Air-Hole RetainedG row t h by Mol e c u l a r B e a m E p i t a x y fo rFabricatingGaAs-BasedPhotonic-CrystalLasers",Applied Physics Express,vol. 6, no. 4, 042002,(2013).
Electron Microprobe Study of the Yinxu (Anyang) Bronze of Academia Sinica Collection
Yoshiyuki Iizuka1 and Junko Uchida2
1Institute of Earth Sciences, and 2Institute of History and Philology, Academia Sinica
To understand bronze casting technology in ancient China, a series of electron microprobe study has been carried out on bronze objects from the Yinxu (Anyang) in the Academia Sinica collection. Because oxidation parts of bronzes do not preserve the original structures and chemical compositions, non damaged bronze’s interior in polished cross-sections were carefully selected, and then their micro-structure and chemical compositions were investigated. Observed metallurgical structures of bronze are divided in two types; dendrite and granular (annealed) structures. Although the granular structured bronze is not common, it suggests that the heat treatment technique has already been applied in the Yinxu Period. Oxygen was also measured to confirm its condition of the oxidation by EPMA, and 73 samples of arms and vessels were discriminated as well-preserved samples. Most of the Yinxu bronzes are tin (Sn) -bronze with a little amount of lead (Pb). An overall result indicates that the bulk Cu/(Cu+Sn) ratios of the bronzes range from 0.79 to 0.89, and chemical compositions are rather different in type of usages. It indicates that the chemical compositions (mixture ratios of Cu:Sn) of the bronzes were already intentionally controlled for their usages in the Yinxu Period.
Introduction
Institute of History and Philology, AcademiaSinica, which was established in 1928 for modernarchaeologicalstudies,performedexcavationworksfor15timesinYinxu(殷墟 )ofAnyang,HenanProvince,theCentralPlaininChina.Excavationprogramsweresuspendedin1937duetochaoticsituations.Themostofexcavatedmaterialswere transferred,anda largequantityofbronzeobjectsfromYinxuhasbeenstoredintheInstitutesince1949,nowatTaipei. Yinxu is theplacewheretheoraclebonescriptswerediscoveredandisthoughttobeanancientcapitalin theLateShangDynasty (ca. 14c.BC-11c.BC), intheBronzeAgeofChina.TheYinxubronzesof theAcademia Sinica collection were excavated fromaristocratic tombs in theXiaotun palacearea, androyal tombs in theXibeigangareaandthecollectioncontainsallkindsofbronzeobjectsfromallphasesoftimesequences throughtheYinxuPeriod.Althoughthecollectionisoneofthemostpreciousandvariablefor study of theBronze culture,only little amountof bronze was studied by scientific approaches. Tounderstand technologica l innovat ion of bronzecasting in theEastAsia, theYinxu'smaterials areextremely importantbecause itusedtobethecenterof bronze manufacture at early Bronze Age in the
EastAsia. Information fromtheYinxubronzesandfurther comparison study of other ages, areas andtechnologywould indicateevolutionof the Bronzeculture.Since2007,theauthorshavelaunchedaseriesof investigationof theYinxubronzecollectionusingelectron microprobe techniques to reveal bronzecasting technology inAnyangof theShangDynasty.Here we report on analytical methods of ancientbronzeandimplicationsofbronzecultureintheShangDynasty.
A n c i e n t b r o n z e s a n d s a m p l e preparations
The bronze, the first alloy of human kind, iscomposedoftwometallicelementsofcopper(Cu)andtin(Sn).Bronzeobjectwasmanufacturedbypouringmoltenalloy intoamold.MeltingpointsofCuandSnareapprox.1085ºCand232ºC, respectively,andmeltingpointsdecreasewith increasingofSncontentinbronze.Figure 1ashowsthephasediagramofCu-Sn(tin-bronze) system.Ofmoltenbonzewithin90-80wt.%ofCu(10-20wt.%ofSn), theprimarysolidphaseofbronzeisα-phasewhenitreachedtheliquidustemperature. The α-phase generates segregation-solidification(dendrite:Fig.1b-d)duringtemperaturefallingandthensecondaryδ-phaseappearsincoolingrateofnormalcasting.Innormalcasting,crystallizationofSn-enrichedphases,εandη,doesnotoccurbecausetemperaturefallstoolowtoreact.Inotherwords,only
dendriteα-phaseandα+δeutecticphasesareobservednormally inancientbronze’s interiorsasreportedbyGettens[1]andWan[2], andanyphasecontain lessthan77wt.%ofCu(>33wt.%ofSn)doesnotexistinnormalcastingbronze. It is well known that a green patina forms onbronzesurface for long timeofburial.Thepatina iscomposedofcopper-, tin-and leadoxidesand theircarbonates.Withoutexception,thesurfaceofancientbronzes was oxidized. Although various analyticalapproaches have been applied for study of bronzechemistrysuchasX-rayfluorescence(XRF),XR-EDSandchemicaldissolutionmethodonthesurfaces,onlysurfaceanalysisisimpossibletoinvestigateitsoriginalchemistryandcastingtechnology. TheYinxucollectioncontainsa largequantity(probablymorethan20,000)ofbronzefragmentsbutmanyarenotable toapply forconservationwork. Itis,however, stillvaluableand isable tochoosesomeappropriate samples to investigate metallurgicalmicrostructureandchemistryfromcross-sections.The
studiedsampleswereselected fromentirephasesoftheYinxuPeriod(MiddleShangtoLateShang)andvarioususages.AtleastintheHanDynasty,sixkindsofusagesare recognized inbronzeobjects,namely,vessels, instruments, arms, tools, ornaments andchariot items.Tounderstandtheirchemicalcharacterof theYinxubronzes,ritualvessels(Jue, Ding, Zun, Gu, Pou, and Hu), arms (helmets,daggers,knives,arrow-andspear-heads)andchariotornamentswereselected for this study.ExcavatedhelmetswereonlyfromHPKM1004tombintheYinxu.
Analytical procedure of the Yinxu bronzesSample preparation
micro-diamondsaw.Toavoidthermalandmechanicaldamagesformetallurgicalstructure,slowrotationspeedof thediamondsawwasoperated in100r.p.m.withdistilledwaterforcoolingduringthecutting.Cleanedsamplesbyethanolweremountedinacold-mounting(roomtemperaturecuredforeighthours)epoxyresinwith1-inchdiametermoldandexposedsurfaceswerepolishedwithdiamondpaste, and then finishedbycolloidalsilicasolution.
SEM
Polishedcross sectionwas initiallyobservedbyanopticalmicroscopewiththereflecting light.ThenScanningElectronMicroscopes(JEOLW-SEMJSM-6360LVandFE-SEMJSM-7100F)wereusedtoobservemetallurgical structure by back-scattered electronimages,whichrepresentmeanatomicabundancebycontrast inblackandwhite images.Semi-quantitativeanalyses were conducted by an energy dispersivespectrometer (Oxford InstrumentsLtd)usedunderthebeamconditionsof15kiloVolt(kV),and0.1nanoAmpere(nA) for theaccelerationvoltageandbeamcurrent,respectively,inthevacuumconditionof25Pa(Pascal).Bulkchemicalcompositionsweredeterminedbymeanvalueof10 to20-areasof120μm×90μm(1000timesinthemagnificationofSEMimage).X-raycounting timewas for100seconds.ThequantitativedatawerecorrectedbyZAFmethodwithchemical-knownpuremetalsandsyntheticalloysforCu(coppermetal),Sn(tinmetal),Sb(antimonymetal),Ag(silvermetal),As(galliumarsenide:GaAs),Zn(zincmetal),Pb(crocoite:PbCrO4),Bi(bismuthmetal),Fe-Co-Ni(NBS868metalstandard)andS(pyrite:FeS2).
EPMA
IntheX-rayenergydispersiveanalysisonbronze,tin(Sn-L lines)is interferenceelementforoxygen(O-kα)analysis.Then,quantitativechemicalanalysisofcopper, tin, lead and oxygen was made by EPMAs(JEOLW-EPMAJXA-8900RandFE-EPMAJXA-850 0F) which equipped wave-length d ispersivespectrometers (WDS). Operated beam conditionswere20kV,10nA,and5μmde-focusedbeamfortheacceleration voltage, beam current and beam size,respectively. The measured X-ray intensities werecorrectedbymetalPRZmethodusing the standardcalibrationof chemical-knownstandardmetalsandoxideswiththefollowingdiffractingcrystals:coppermetal forCu-KαwithLiFcrystal, tin-metal forSn-LαwithPETHcrystal,crocoite(PbCrO4)forPb-Mß withPETHcrystalandtin-oxide(SnO2)forO-kawithLDE1Hcrystal.X-raypeaksaretheirbothupperandlowerbaselineX-rayswhicharecounted for20and10s,respectively.Toobtainbulkchemistryfromeachsample,analyzedpointswererandomlyselected100to225pointsat50μmintervalswithX-Ydirectionsbythemappingpoint tableconversion.Bothsecondary-and back-scattered electron images were used toavoiddamagedandweatheredareas.Then thebulkchemistry,especiallybulkCu/(Cu+Sn)ratios(Cu#),werecalculatedasmeanvalues.Chemicaldistribution(mapping)analysisofCu,Sn,Pb,Oandsomeothers
(AsandSb)wasalsoperformedbyFE-EPMAatthecondition of 20 kV and 30 nA for the accelerationvoltageandbeamcurrent,respectively.
ResultsMetallurgical structure of the bronze’s interior
ThethicknessofstudiedYinxubronzesismostly2 to3mm.Manyofbronzeswereseriouslyoxidizednotonlyat their surfacesbutalso in their interiorsoccasionally. Such samples were not appropriateto investigate their metal lurgical structure. Weattemptednear200bronzefragments,but95sampleswereabletostudytheircrosssections. Figure1 (b-e) showsa representativedendritestructureintheYinxubronzeobject(helmetF1fromHPKM1004)withback-scatteredelectronmicrographanditschemicaldistributionsofCu,SnandPb.Lead(Pb) is segregated from the bronze phases whicharecomposedofαphase, theprimarycrystalphase,andα+δ eutecticphases.Pb isa fusingagentand isconducting melting point to be lowered. Pb was inmolten bronze at the high temperature, but is notdistributedinsolidbronze. The bronze micro-structures are divided intotwo types, suchasdendriteandgranular structures(Fig. 2). The dendrite is substantially present butgranularstructureisobservedonly5casessofar.Forstructurecomparison,westudyexperimentalbronzeobjects simultaneously.Thedendrite isobserved innormalcastingbronzes,whereasgranularorchemicalhomogeneitystructureisobtainedfromexperimentalproductsafter thermal treatment suchasannealingortempering.Theresultsindicatethatatleastakindof thermal treatment method has already appliedin theYinxuPeriodof theShangDynasty, it isveryuncommonthough.
Chemical composition of the Yinxu bronzes
Accuracy of bulk Cu/(Cu+Sn) ratios (Cu#) ofbronzes was confirmed by chemical known bronzealloys.Figure 3 showsperformed resultsby4-typeof analysis methods. Some standards were leadedbronzes(3to10wt.%inoriginalweights).Lead(Pb)isfusingagentandbehavesasvolatilegasduringtheexperiments.ThusPbcontentswereexpected tobedecreasefromtheoriginals.However,theCu#intheleadedbronzeswerewellmaintained.Overallresultsareacceptablewithin0.02intheCu#[3]. Figure 4 showsPbcontentwith theCu#of thestudied95bronzesbytheusages.Themostofbronzesarerangingbetween0.77and0.89intheCu#,anddonotcontainmuchPb(lessthan2wt.%).Thebronzes,containedmorethan0.9 inCu#,are four(4)arsenic(As)bronzes(Fig. 5a),one(1)antinomy(Sb)bronzefromthearms(Fig.5b),andPb-bronze(2-ornaments:R014314 and R007306) which show less Sn wereidentified.Excludingsuchbronzes(>0.9inCu#),themostofstudiedYinxubronzesareshownless-contentofPbandmorethan75%of thestudiedbronzesareshownless than2wt%inPb.Themeanvalueof theCu# and Pb of helmets from the HPKM1004 are0.838and1.56wt.%, respectively, in the studied30
samples. On the other hand, the #Cu of the ritualvessels ranges0.78-0.88which is relatively lowerCu(orhigherSn)range.Mostofthevesselscontainsomeamounts of Pb (up to 7 wt.%). In the ornaments,Cu#rangeswidelyfrom0.77to0.98.Asamplewhichhighlydecoratedwith inlaidof turquoise (R017653)showsthehighestPbcontent(10.5wt.%)inthisseriesofanalysis. In the chariot items, sampleR006919containsratherhighPbas4.8wt.%,andthe#Cuissimilartotheotherritualvesselsas0.814.Ontheotherhand,socalledabow-shapedornament (R001768), containsless Pb (0.3wt.%), and the#Cu is only0.766. It isso far the lowestvalueof#Cuanddiffers from theratioofotherarms.This is thought tobeakindofarms,butitschemicalcompositionindicatesdifferentconcern to categorize the objects. Cavities or voidspaceswerewellobserved in the interiorofhigh-Pbbronzesgenerally.Lesscavitiesareobserved in theinterioroflow-Pbbronzes.ItseemsthatconditionofpreservationisrelativelybetterinlowerPbbronzes.
EPMA results with Oxygen analysis
Presenceofoxygeninbronzes’interiorindicatesconditionofpreservation.Thus inspectionofoxygenisuseful todiscriminate theiroriginalchemistry forfurther discussion. Figure 6 shows representativeresults of mapping analysis on two samples, anoxidized and a well-preserved bronze helmet ofhel mets , f rom the H PK M10 0 4. It i s obv iouslyimpossible to indicate their state of the oxidationf rom the back-scattered images (BEI) on bothsectionsbecausethedendriteswereclearlyidentified.However,oxygenmappingsdemonstrate themicro-dendrite structure was oxidized in the helmet-07.QuantitativespotanalysisresultsareshowninFig. 7.Oxidizedhelmets (Hel-06and -07) shows scatteredO rangeup to25wt.% and theCu/(Cu+Sn) ratiosfrom1.0 to0.3whichare inconsistentwithnormal
castingbronze,whereaswell-preservedhelmets(Hel-05and-08)doesnotcontainOandallCu#rangeareconsistentwithα-phaseandα+δeutecticphase(from94to72wt.%ofCu:seeFig.1a).TheresultsindicatethatoxidationprocessvaryitsCu#fromtheoriginal. In theEPMAoverall results,73bronzeobjectswereconfirmedaswell-preservedsamplesandkeepreliablechemistry todiscuss itsoriginalbulkCu:Snrat ios. Figure 8 shows distr ibutions of the Cu #(bulk Cu/[Cu+Sn] ratio) by usages. Each Cu# wascalculatedby100-225spotsanalysisbyEPMA. The helmets from the HPKM1004 (grey) areranging between 0.80 and 0.89, and most of thehelmetsarebetween0.83and0.86,andmeanvalueis0.843 in theCu#.Thearms(green)arealsowide inrangefrom0.82to0.89intheCu#.Thedatacontainsvarioustypesofarmsthough.Ontheotherhand,theCu#of theritualvessels(red)arerangingfrom0.80to0.86.Differencesinthebronzechemistryseemnotclear in thevariationof the time sequence throughtheYinxuPeriod.
Discussion
It is wel l studied about physical propertiesof Cu-Sn alloy in the modern metallurgy that thecolor, theBrinellhardness, the tensile strength,andtheelongationofbronzevarywith ratiosofCu:Sn(Fig. 9). In general, bronze is getting harder withincreasingofSncontentfrom15wt.%butitisgettingbrittlebecausethetensilestrengthandtheelongationare significantly decreasing with Sn content morethan 20 wt.% . It seems that bronze gains mosttoughnessaround85-80wt.%ofCu(15-20wt.%Sn).EstimatedviscosityofmoltenbronzeisalsoshowninFig.9 thatviscosity isdecreasingwith increasingofSncontent[4]. Sofarweobtained73bronzechemistriesand46dataof themwerehelmets fromtheHPKM1004.Asshown inFig.8, the#Cuofhelmets showsrelatively
uniformedandmostofhelmetsdistributebetween0.83 and 0.86, and mean value is 0.843. Their #Curatios seem comparable to high toughness rangeof bronze. On the other hand, the Cu # range ofthe ritual vessels is slightly Sn-enriched and Pb aswell (seeFig.4).Thevesselsareusuallydecoratedwith fine relieveson the surface.Thesephenomenaindicate that theSnandPbwereadded intentionallyinorder to increase theviscosityofmoltenbronzefor casting, which might be easy to pour into finedecorationmolds.Variouskindsofarmswerestudiedbut some of arms seem ritual objects with surfacedecoration insteadof realweapons. Itmightbe thereason that the Cu# of arms distributes in widerrangefrom0.82to0.89.Fromthesecircumstances,it
islikelythatcraftsmenalreadyunderstoodcharactersofbronzeatthetime. In the ancient Chinese classic of the ZhouDynastyofRitesofZhou<周礼考工記Zhou Li Kao Gong Ji>,the Six Formulaofmixtureratios<金有六斉Liu Qi>werestandardizedfordifferentusagesofbronze.Ithasdebatedthatdescriptionwasinthe4thtothe3rdCenturyBConthebasisoftheknowledgeofthe9thtothe7thCenturyBCwhichisrepresentedof the period of Zhou Dynasty. According to thedescr iption, it is thought that di f ference of thebronzealloycomponents (ratiosofCu:Sn)mightbecontrolledsince long timeago inChina.However ithasnotbeenconfirmedwithchemicalcompositionsofancientbronzes.
RepresentativeresultsofoxygencontentswiththeCu/(Cu+Sn)ratiosfromtheinteriorsofthebronzehelmetsfromtheHPKM1004byEPMAspot(quantitative)analysis.Diamonds:oxidizedbronzes(Helmets-06and-07).Triangles:well-preserved bronzes (helmets-05 and -08). n:numbersofanalyticalspots.
41 JEOL News Vol. 49 No. 1 (2014)
Wan[2]proposed2-waysinterpretationofthe Six Formulabasedonweighingratios,shownascases-Aand-BinTable 1.Becauseatermofcopperwasnotpresentat the time, theancient sentences in the Six formulaweredescribedbronzemixture ratioswith“Metal”and“Tin”.Heassumed“metal” in caseofbronze(case-A)andcopper (case-B)andestimatedthesixmixtureratiosinweightintheCu#from83.3to50wt.%incase-A,andfrom85.7 to66.7wt.%incase-B. In modern metal lurgy, it is well known thatbronze which contained less than 66.7 wt.% of Cu(more than 33.3 wt.% of Sn) is unable to cast. Inthe metallurgical point of view, a bronze which iscomposed of only δ phase (68.2-66.8 wt.% of Cu)isalsonotexist.TheworldmosthighestSn (lowestCu)bronze objecthas been recordedas 32.6wt.%ofSn(67.4wt.%ofCu)fromKerala,SouthIndia[5].As shown in the results,mostof theYinxubronzesare constructed by α and α+δ phases. The lowestCu# is0.783.Ontheotherhand, thehighestCu#of83.3and85.7wt.% incase -Aand -B, respectively.However there are not reliable because the well-preservedbronzesaredispersedup to the rangeof
0.89 in theCu#.Therefore it is suggestedthatWan’s2-hypothesesareinconsistentinreality. Hori[6]investigatedtheancientbalance-weightsfromtheAncientCentralAsiaand itwasconfirmedthattheCentralAsianweightsystemwasestablishedaround 4000BC and is the oldest in the world. Onthe other hand, Chinese weight system might beestablished in later than 1000BC which might becomparable to Post-Shang Dynasty. Qiu et al. [7]interpretedancientChinesearticlesthatvolumeunitby the decimal system was already established inthePre-QinPeriod (by221BC).Theyalsopointedout inhistoricalpointof view that theappearanceof the weight unit is later than the establishmentofunitsof lengthandvolume.TwoPb ingots wereexcavated from the Xiaotun E-16 pit at Yinxu [8]butanyweighingandbalance-weights toolwerenotdiscovered yet from Anyang. It is supposed that,therefore,weighingsystemwasnotestablishedyetintheYinxuPeriod. Volumetricsystemisanothermeasurementway.ThedensityofCuandSn(ß-Sn)are8.94and7.365(g/cm3) respectively. Thus the mass (or weight) ofCu and Sn are different in even the same volume.
Volumetric ratios『金 Metal』 as Bronze 『金 Metal』 as Copper 『金 Metal』 as Copper
Assumingthatmixtureratiosof the Six Formulaarebased on the volumetric ratios, their weight ratiosaresuggestedfrom88to71wt.%withinfew%stepsin theCu#(case-C inTable1).Fromtheanalyticalresults, the#Cuofhelmets fromtheHPKM1004arerelativelyuniformedwithin3wt.%andtheserangesarealsoreflectedtothephysicalpropertyofbronze.FurtherthesuggestedrangeoftheCu#from88to71wt.%ofCumightbe reliable for casting.Thus it islikelythatvolumetrichypothesisisprobable. In types I and III of the Six Formula instructthe ratio forBells&Cauldrons (Cu6:Sn1= 88wt.%of Cu), and daggers (Cu4 : Sn1= 83 wt.% of Cu),respectively,whicharerepresentedvesselsandarms.Thus it indicates that thevesselsareenriched inCuthanthearms,however,theanalyticalresultsshowedcontradict ion that the arms are enr iched in Cuthan thevessels. It isalso inconsistentwithphysicalpropertyofbronze.
Conclusion
Weattemptednear200bronzeobjects fromtheYinxu (Anyang) in theAcademiaSinica collectionto invest igate their metal lurgical structure andchemistry.Mostof theYinxubronzesare tin (Sn)-bronzewitha littleamountof lead (Pb).BulkCu/(Cu+ Sn) rat ios (Cu # in weight) of each bronzefragment were calculated by average of 100 to 225EPMA quantitative (spots) analysis with oxygenand lead, followedby semi-quantitativeanalysisbySEM-EDS . The Cu# from non-oxidized interiorrepresents itsoriginalCu#,whereas theratioscouldsignificantlybeshiftedafteroxidation.Weconfirmedthat 73 samples of arms (helmets, daggers, arrow-andspear-heads)andvesselswerewellpreserved.Anoverall result indicates that theoriginalCu#rangesfrom 0.79 to 0.89. No variation is verified in timesequence during the Yinxu Period. By the usages,theCu#ofhelmetsarebetween0.84and0.89, and0.845 inaverage,and it is relativelyuniformedwhencompared with other various usages. The most ofhelmets do not contain much Pb (< 2 wt. %), andless cavities are observed in the interior. PhysicalpropertyofbronzeisvariedbytheirCu#,andhelmetchemistriesare fit its toughnesschemicalrange.ThevesselsshowrelativelylowerinCu#(high-Snbronze)andcontainsomeamountofPb(upto5-6wt.%).Inthecontrastwiththehelmets, thevesselsareusuallydecorated with fine relieves on the surface. SincelowerCu#(orhigherSncontent)andadditionofPbreduceviscosityofmoltenbronze, itmightbeeasyto cast with fine decoration molds. It is suggestedfromtheresultsofchemicalanalysis that theCu#ofbronzeswas intentionally controlledbypurposeofusageswithin few%in theYinxuPeriod.Basedonthis series studyof theYinxubronzes,weproposedahypothetical interpretationof the Six Formulabyvolumetric ratiosmightbeprobable, insteadof theweight-base interpretation.However, theCu#of theYinxubronzesare inconsistentwiththeratioswhichweredescribedintheancientChineseclassic.
Acknowledgments
We thank Dr. Kwang-tzuu Chen and Ms. Yu-yun Lin of Institute of History and Philology fortheir kind support. Professors Haruhisa Mifuneand Takekazu Nagae of University of Toyama areappreciated toprovideexperimentalbronzesamplesand valuable discussion. We thank Ms. Ya-t ingHsu, Mr. Yu-shiang Wang and Ms. Hui-ho Hsiehof Institute of Earth Sciences for their technicalsupportofbronzeanalysis.ThisstudyissupportedbyNationalScienceCouncil (Taiwan)andtheInstituteofHistoryandPhilologyofAcademiaSinica.
References
[1] Gettens R.J. 1969 The Freer Chinese BronzesVolume I I Technica l Studies . SmithsonianInstitutionFreerGalleryofArt,OrientalStudies,No.7.WashingtonDC.
[2] WanChia-pao (1970)Apreliminary reportonthemetallographicexaminationsofShangbronzehelmets. Institute of History and Phi lologyAcademia Sinica Special Publications, No.60.pp.4 8 . Ta ipei (in Chinese w ith an Eng l i shsummary).
[3] IizukaY.,J.Uchida(2013)ChemicalcompositionsofYinxu(Anyang)bronzeobjectsintheAcademiaSinicacollectionandits implicationsforAncientChineseCasting techniques.Bulletinof JapanSociety of Chinese Archaeology, 13 :23-47 (inJapanesewithChineseabstract).
[4] KozlovL.Y.,L.M.Romanov,N.N.Petrov(1983)Pred ict ion of mult icomponent meta l meltsviscosity. IzvestiyaVyssh.Uch.Zav.,ChernayaMetallurgiya,3:7-11.
[5] M i f u n e H . ( 2 0 1 0 ) C o m p a r i s o n o f t h emanufacturing technology of high-tin bronzetools inmodernAsia.InAsian high-tin bronzes: Production technology and regional characteristic.pp.125-135(ISBN978-4-9905066-1-2).
[7] Qiu Guangming, Qiu Long, Yang Ping (2001)UnitofWeight(Chapter4)inHistory of Chinese Technology.ScienceandTechnologyPress,pp.25-31.Beijing(inChinese).
[8] Chen Kwang-tzuu (1991) Analysis and studyon Lead Ingots from Yinxu. In Archaeology and Historical Culture. Commemoration of the Eightieth Anniversary of Kao Ch'u-hsun,pp.355-388.Cheng-ChungBook.Taipei(inChinese).
[10] S c o t t D . A . ( 1 9 9 1 ) M e t a l l o g r a p h y a n dmicrostructureofancientandhistoricmetals.TheJ.PaulGettyMuseum,pp.155.LosAngels(ISBN0-89236-195-6).
43 JEOL News Vol. 49 No. 1 (2014)
Elucidation of Deterioration Mechanism for Organic Solar Cells– Toward Highly Efficient Solar Cells –
Kazuhiro Marumoto
University of Tsukuba
We report on an electron spin resonance (ESR) study of polymer solar cells to investigate accumulated charge carriers in these devices under device operation from a microscopic viewpoint. Light-induced ESR (LESR) signals and device characteristics were simultaneously measured using the same device under simulated solar irradiation. From the ESR analysis, the molecules where photogenerated hole carriers were accumulated are clearly identified as poly(3-hexylthiophene) (P3HT). Moreover, the simultaneous measurements of ESR and device characteristics demonstrate a clear correlation between the increased LESR intensity and deteriorated device performance. The ESR study reveals that the deep trapping sites for photogenerated hole carriers are located at interfaces between a hole buffer layer poly(3,4-ethylenedioxythiophene):poly(4-styrenesulfonate) (PEDOT:PSS) and an active layer P3HT:[6,6] -phenyl C61-butyric acid methyl ester (PCBM), which can be ascribed as the main mechanism for the reversible deterioration of the device performance of polymer solar cells.
Introduction
Organic thin-film solar cells are a promisingalternativesourceofelectricalenergybecauseoftheirprintableand flexibledevicestructure, lightweight,and low-costproduction.[1-3]Solar cellshavebeeninvestigatedusingavarietyofmethodswith theaimofimprovingtheirperformance.[4-10]Therehasbeena significant amount of interest in the high powerconversionefficiency (PCE)ofmore than10%thatwasrecentlyobtained inorganic thin-filmsolarcellsdue to theirpotentialpractical applications.[11,12]Inaddition todeviceperformance, thedurabilityofsolarcells isan importantproblemfor thepracticaluseofsolarcells.Fordurabilitystudies,degradationof solarcellsdue toextrinsic factors suchasoxygenand water has been repor ted , where mater ia l sand device structures were irreversibly degraded.[13-15] Moreover, in addition to the irreversibledegradation, the reversible initial deterioration ofdeviceperformancewithoutmaterialdegradationhasbeenreportedforapolymersolarcellusingmethodsthatemploy thermal stimulated current (TSC)andcurrent density (J ) -voltage (V ) measurements .[16,17] The studied polymer solar cell was a bulk-heterojunction(BHJ) thin-filmsolarcellwithblendfilmsofa conductingpolymer regioregularpoly(3-hexylthiophene) (P3HT) and a soluble ful lerene
[6,6]-phenylC61-butyricacidmethylester (PCBM),whichhasbeenwidely studiedasa typicalpolymersolar cell. [1-10,16,17] The polymer solar cell has adevicestructureof indium-tinoxide(ITO)/poly(3,4-ethylenedioxythiophene):poly(4-styrenesulfonate)(PEDOT:PSS)/P3HT:PCBM/Al,wherePEDOT:PSSis a typical hole buffer layer and has been widelyusedto improvedeviceperformance.[1-13,16,17]Thepreviousstudiesreportedthatthedeviceperformancegraduallydeterioratedduringdeviceoperationandthatthedeteriorationcouldberecoveredviathermalannealing.[16,17]Inotherwords,thedeteriorationhasbeenascribedtoanaccumulationofphotogeneratedchargecarriersduringdeviceoperation rather thantothedegradationoforganicmaterialsand/ordevicestructures.[16,17]TheTSCstudy indicated that theaccumulationsiteswerelocatedat interfacesbetweenactive layers and electrodes with various trappinglevels.[16]However,amoredetailedstudyclarifyingmoleculesand these siteswherechargecarriersareaccumulated(trapped)withoutmoleculardegradationhasnotyetbeenconducted,whichwillbeextremelyimportant for fur ther dev ice per formance anddurabilityimprovements. Electronspinresonance(ESR) isonepromisingmethod for such a microscopic character izationof charge-accumulation sites because it is a highlysensitive and powerful approach that is capable ofinvestigatingorganicmaterialsat themolecular level.[18-21]Thismethodhastheadvantageofbeingabletodirectlyobserveaccumulatedchargecarrierswithoutdetrappingcarriersviathermalstimulation,asusedin
theTSCmethod.ThisESRmethodhassuccessfullyclarifiedthemicroscopicpropertiesofcharge-carrierstates in organic devices, including spin states andtheirspatialextentofwavefunctioninpentacene,[18]fu l lerene (C 6 0) , [19] and P3H T[19, 21] at dev iceinterfaces.TheESRmethodhasalsobeenappliedtopentacene/C60heterojunctionorganic thin-filmsolarcells and their organic layered films; these studiesdirectlyobservedchargeformationinpentacenelayersduringdevicefabricationunderdarkconditions.[22,23]However, theaccumulationofphotogeneratedchargecarriers inpolymersolarcellsunderdeviceoperationhasnotyetbeen investigatedusingtheESRmethod.Suchan investigation fromamicroscopicviewpointwouldbeuseful inclarifying theaccumulationsites,whichwouldhelp to improvedevicedurabilityandperformance. In this news, we report on an ESR study ofP3H T: PCBM polymer solar cel ls to invest igateaccumulatedchargecarriers in thesedevicesunderdeviceoperation.[24]Wemeasuredlight-inducedESR(LESR)signalsanddevicecharacteristics(short-circuitcurrentandopen-circuitvoltage)simultaneouslyusingthe same device under simulated solar irradiation.From t he E SR a na lys i s , t he mole cu les wherephotogeneratedholecarrierswereaccumulatedareclearlyidentifiedasP3HT.Moreover,thesimultaneousmeasurements of ESR and device characteristicsdemonstrateaclearcorrelationbetweentheincreasedLESRintensityanddeteriorateddeviceperformance.WiththeuseoforganiclayeredfilmsofPEDOT:PSS/P3HT:PCBM, theESRstudy reveals that thedeeptrapping sites for photogenerated hole carriers arelocated at PEDOT:PSS /P3HT:PCBM interfaces,which can be ascribed as the main mechanism forthe reversible initial deterioration of the deviceperformanceofpolymersolarcells.
Experimental
CommerciallyavailableP3HT(Sigma-Aldrich,Plexcore OS 1100, regioregularity: 98.5%), PCBM(FrontierCarbon,nanomspectraE100,purity:99.2%),and PEDOT:PSS (Clevios PAI4083) were used tofabricate the solarcells.Thechemical structuresof
P3HT,PCBM,PEDOT,andPSSareprovidedinFig. 1.ThedevicestructurewasITO/PEDOT:PSS(≈40nm)/P3HT:PCBM(≈160nm)/Pd(1.2nm)/LiF(0.6nm)/Al(100nm).PEDOT:PSSfilmswerefabricatedbyspin-coatinganaqueoussolutionontoITO-coatedquartzsubstrates,followedbyannealingat140°Cfor10minunderanAratmosphere.SolutionsofP3HTandPCBM(1:0.8w/w)dissolvedino-dichlorobenzene(3.4wt.%)werestirredwithamagneticstirrerfor30minat70°C,andwerespincoatedontopofthePEDOT:PSSfilmsat1000rpmfor75stoformP3HT:PCBMfilms.TheareaoftheinterfacebetweenP3HT:PCBMandPEDOT:PSSwas3mm×≈14mm.Then,Pd,LiF,andAllayersweredepositedonto theP3HT:PCBMblend film to formtheanodeusingaconventionalvacuumsublimationtechniqueundervacuumconditionsbelow2×10-4Pa.Finally, thedeviceswereannealedat140 °C for30minunderanAratmosphereandvacuum-sealedafterwiringinanESRsampletube.ThedeviceperformancewasconfirmedtoimprovebytheuseofPdontheanodeside.WealsoobtainedsimilarESRresultstothoseofthepresentstudybyusingdeviceswithoutPdandLiF. TheJ-V characteristicswereevaluatedusinganAgilent Technology B1500A semiconductor deviceanalyzerundersimulatedsolar irradiation(AM1.5G)witha100mWcm-2 intensityat290KunderanAratmosphere. ESR measurements were performedusing a JEOL RESONANCE JES-FA200 X-bandspectrometerundervacuumconditionsat290K.Thenumberof spins,g factor,and linewidthof theESRsignalwerecalibratedusingastandardMn2+markersample.Theabsolutevalueofthenumberofspinswascalculatedusingasolution(220μl)of4-hydroxy-2,2,6,6-tetramethylpiperidin-1-oxyl(TEMPOL)asastandard.Thecalibrationofthegfactorwasperformedbyusinga software program of the JEOL RESONANCEESRsystemconsideringhigherorder correction totheeffectiveresonancefield.Itscorrectnesswasalsoconfirmed by using 2,2-diphenyl-1-picrylhydrazyl(DPPH)asanother standard sample.SimultaneousmeasurementsofLESRandthedevicecharacteristicsJsc andVocwereperformedusingaKeithley2612Asourcemeterundersimulatedsolar irradiation(AM1.5G)witha100mWcm-2intensitywithaBunkoukeikiOTENTOSUN-150BXMsolarsimulatorat290K.Thesimultaneousmeasurementsunder theopen-circuit
45 JEOL News Vol. 49 No. 1 (2014)
conditionwereperformedimmediatelyafterperformingsimultaneousmeasurementsunder the short-circuitconditionusingthesamedevice.
Results and Discussion
Firstly, we describe the device characteristicsof our solar cells. When we fabricated solar cellswithanactiveareaof 2mm×2mmusingan ITOsubstratewithaconventionalsizeof20mm×20mm,our devices had the following typical parameters:a short-circuit currentdensity (Jsc)of≈7.1mAcm-2,anopen-circuitvoltage(Voc)of≈0.66V,a fill factor(FF) of≈0.60, and a PCE of≈2.8%. These valuesareequivalent to thosereportedbyothergroups.[3-9,16,17] Thus, we have confirmed standard deviceoperation using our device fabr ication method.However, the device parameters of the cel ls forthe ESR experiments deteriorated, compared withthose of the above cells because of the followingdifficulties.Firstly,theuseofanarrowITOsubstratewithasizeof3mm×20mm(seeFig.1cofRef.22)wasnecessarybecauseweneededtoinsertthedeviceintoanESRsample tubewithan innerdiameterof3.5 mm. Secondly, to overcome the low signal-to-noise (SN) ratio of the LESR signal due to device
operation, we needed to fabricate a device with alarge active area; we adopted an active area of 2mm×10mmin this study.Thus, the filmqualityoftheasymmetricactiveareadeterioratedbecauseofthe variation in film thickness, which was causedby thedifficultiesof spin coatingusing thenarrowITOsubstrate.Asaresult, thedevicecharacteristicsdeteriorated;Jsc andVocdecreased toapproximately2.2 mA cm-2 and 0.35 V, respectively. Despite thisdecreased device performance, our devices for theESR experiments exhibited similar performancedeteriorationunderdeviceoperationasthatreportedby other groups. [16 ,17] T herefore, the presentmicroscopic characterization by ESR analysis isconsidered tobe useful for clarifying theessentialproblemsofperformancedeteriorationunderdeviceoperation,asdemonstratedbelow. Next,wepresentthelight-inducedESR(LESR)signalsofthedevicesunderdeviceoperation.Figure 2demonstratesthedependencesoftheLESRsignalsofthesamedeviceonthedurationofsimulatedsolarirradiation under: a) short-circuit condition, andb) open-circuit condition. Here, the LESR signalswereobtainedby subtracting theESRsignalunderdark conditions from that under simulated solarirradiation.Theverticalaxisisplottedusingaunitofpeak-to-peakESR intensityof theESRsignal fora
standardMn2+markersample,IMn,which isattachedto the inside wall of the ESR cavity of the JEOLRESONANCEESRsystem. Inourexperiment,weuseda continuous-wavemethodwithamodulationfrequencyof100kHzfortheexternalmagneticfield.Thus, thephotogeneratedchargecarrierswithshortlifetimeslessthan10μscannotbeobservedusingthepresentESRmethod.Freelydiffusivechargecarriersto electrodes contribute the standard operation ofthedevice.Therefore,theobservedLESRsignalsaredue to the accumulation of photogenerated chargecarriers with lifetimes longer than 10 μs, namely,trappedphotogeneratedcarriers.Asrevealed inFig.2a,b, gradual increases in theLESR intensitiesareclearly observed under both operation conditions.The increased rate of the LESR signal under theshort-circuit condition is larger than thatunder theopen-circuit condition. The g factor and peak-to-peak ESR linewidth, Hpp, are determined to be:a)g=2.0022andHpp=0.25mT,andb)g=2.0023andHpp=0.24mT.Thesevaluesareconsistentwiththose of hole carriers (positive polarons) in P3HT
filmsinorganicfield-effectdevices.[19,21]Therefore,the LESR observation directly demonstrates thatthe observed spin species are photogenerated holecarriersaccumulatedinP3HTmolecules. To clarify the locations of accumulated holesfurther, we performed LESR measurements fororganic layered films of ITO/P3HT:PCBM, ITO/PEDOT:PSS, and ITO/PEDOT:PSS/P3HT:PCBM.For the ITO/P3HT:PCBM and ITO/PEDOT:PSS/P3HT:PCBM films, we observed the same LESRsignalas thatobservedinFig.2a,b.However, fortheITO/PEDOT:PSS films,wedidnotobserve suchaLESR signal. Moreover, the ESR signal of PCBMhasbeenreported tohaveg-tensorvaluesof1.9982- 2.00058, [25-27] which differ radically from thatobserved inFig.2a,b.Furthermore, theESRsignalsofPEDOT:PSSandPEDOThavebeenreported toexhibitabroaderESR linewidth than thatobservedin Fig. 2a,b. [28,29] Therefore, we can ascribe theLESRsignalsobserved inFig.2a,b toaccumulatedhole carriers in P3HT. We here comment that thereason for theabsenceofLESRsignal fromradial
47 JEOL News Vol. 49 No. 1 (2014)
anion (electron)onPCBM is ascribed to the rapidspin relaxat ion of fu l lerene’s electron at roomtemperature.[19,27] Next, we present the results of simultaneousmeasurements of LESR and device characteristicsusing the samedevice.Topresent theESRresults,we u se t he nu mber of sp i n s , N s p i n , due to t heaccumulation of photogenerated hole carriers inP3HT varied from that before irradiation at eachexperiment,whichwasobtainedby integrating theLESR signal twice at approximately 320.5-322 mTand comparing this value with the standard Mn2+marker sample.Fig.2c,d illustrate thedependencesof N spin and the device parameters Jsc and Voc onthe duration of the simulated solar i rradiation,respectively.These results indicate thatan increasein Nspin clearly correlates with the deterioration ofthedeviceperformance.That is,NspinmonotonicallyincreasesandJscandVocconcomitantlydecreaseasthedurationofsimulatedsolarirradiationincreases.Thisclearcorrelationdemonstrates that theaccumulationofphotogeneratedholecarriersinP3HTdeterioratesthedeviceparametersJscandVoc.To thebestofourknowledge, this is the first instance in which sucha clear correlation between the microscopic ESRcharacteristics and macroscopic device parametershasbeenobserved.Thechargeaccumulationaffectsaninternalelectricfieldinthedevice,whichpreventscurrent flowandcreatesadditionalpotential in thecells,asdisscussedlater. T h e i n f l u e n c e o f t h e i n c r e a s e d c h a r g eaccumulation (carrier density) on Jsc andVoc wereinvestigated in detail by other groups. [16,17] Thestudies showed the decreases in Jsc andVoc duringdeviceoperation,whichareconsistentwith thoseofourdevice.Note that the reversibilityof thechargeaccumulationand thedeviceparametersJscandVocwas confirmed from ESR and J-V characteristicsafter irradiationoff,which isconsistentwith thatofthepreviousworks.[16,17]Wecommentthatthelong-livedaccumulatedchargecarriers inpolymer filmshave been reported by several groups, which werecausedbydeeptrappingsduetolowtemperatures[30]orextrinsic factorssuchasoxygenunder irradiation.[27] The inf luence of traps due to such extrinsicfactorsonJscandVochavebeenalso studied.[13,14]The inf luence is an interesting problem for thepresent resultsbecauseourdevice isvacuumsealedinanESRsample tube,which isbeyondthepresentresearchscopeandisopenforfurtherstudy. Next,wediscuss theaccumulation sitesof thephotogenerated hole carriers in more detai l. Aspresented in Fig. 2c,d, the accumulation rate ofthe photogenerated hole carriers under the short-circuitcondition is larger than thatunder theopen-circuit condition.Under the short-circuit condition,an internal electric field exists in the BHJ activelayer in the device, which causes the migration ofphotogenerated hole carr iers to ITO electrodesthrough the interface between the PEDOT:PSSand P3HT:PCBM layers. Thus, one can considerthat the d i f ference for the accumulat ion ratesbetween the short- and open-circuit condit ionscan be ascribed to the effect of the PEDOT:PSS/P3HT:PCBM interfaces. To clarify the effect ofthe PEDOT:PSS/P3HT:PCBM interfaces on hole
accumulation, we examined the transient responseofLESRfor layeredthin filmsofITO/P3HT:PCBMandITO/PEDOT:PSS/P3HT:PCBMuponsimulatedsolar irradiation.Figure 3a,b present the transientresponses of N spin for: a) ITO/P3HT:PCBM, andb) I T O / P E D O T: P S S / P 3H T: P C BM . For I T O /P3HT:PCBM, Nspin monotonically increases undersimulatedsolar irradiation.When irradiationceases,Nspin sharply decreases to a small value, and then,thesmall remainingcomponentgraduallydecreases(see Fig. 3a) . However, for I TO / PEDOT: PSS /P3HT:PCBM, a different behavior was observedcomparedwith thatof ITO/P3HT:PCBM.That is,alargeremainingcomponentwasclearlyobservedafterturningofftheirradiation(seeFig.3b).Thelifetimeof theremainingcomponent isextremely long,morethan 40 h, even at room temperature. Therefore,this resultdemonstrates thatPEDOT:PSS insertionbetween ITO and P3HT:PCBM causes the holeaccumulationsiteswithdeeptrappinglevelsinP3HTatPEDOT:PSS/P3HT:PCBMinterfaces.Using thisholeaccumulation, thedifferentratesof increaseforNspinobserved inFig.2canbereasonablyexplainedbecauseholecarriersareeasilyaccumulatedat thePEDOT:PSS /P3HT:PCBM interfaces due to theflowof short-circuit currents.Weherecommentonthe sharply increasing and decreasing componentsuponirradiationonandoff,respectively,inFig.2c,d,and3a,b.TheseLESRsignalsaredue to shallowlyt rapped photogenerated holes in P 3H T, whichaccumulation sites are probably located at bulkmaterials. T h e f i n d i n g s f r o m t h e E S R s t u d y a r esummarized in Fig. 3c, which helps to explain theholeaccumulationat the interfacesunder simulatedsolar irradiation.Firstly, theholeaccumulationsiteswithdeep trapping levelsare formedbydepositingP3HT:PCBMfilmsonaPEDOT:PSS layer(see left-handside inFig.3c).Secondly,photogeneratedholecarriers are accumulated at deep trapping sites inP3HT molecules at the PEDOT:PSS/P3HT:PCBMinterfaces (see the right-hand sideofFig. 3c).Thisholeaccumulation is enhancedby the current flowattheinterfacesbecauseoftheinternalelectricfieldundertheshort-circuitcondition.Wehereexplainanenergy-level shift in thehighestoccupiedmolecularorbital (HOMO) of P3HT at the interfaces shownin Fig. 3c. This energy-level shift is related to theinterfacialelectricdipole layer,whichcanbecausedby an electron transfer f rom P 3H T to PEDOTbecauseoftheenergydifferencebetweentheHOMOofP3HT(4.7-5.1eV)[31,32]andtheworkfunctionofPEDOT:PSS (5.3 eV).[8,22] This electron transferforms holes in P3HT under the dark condition, asdiscussed inapreviousESRstudyonpentacene/C60heterojunction solar cells with PEDOT:PSS hole-buffer layers.[22]Weconfirmedthisadditionalholeformation in the present study by measuring ESRsignals of layered f i lms of quartz / P3H T:PCBMand quartz /PEDOT:PSS/P3HT:PCBM under darkconditionsbeforesimulatedsolarirradiation.Thatis,thenumberof spinsofP3HTunderdarkconditionsforquartz/PEDOT:PSS/P3HT:PCBMwasmeasuredtobe3.2×1012,whichislargerthanthatof4.6×1011forquartz /P3HT:PCBM. This result demonstrates theadditionalholeformationinP3HTduetotheelectron
(2014) Vol. 49 No. 1 JEOL News 48
Fig.3 Transient responsesofNspin fororganic layered thin filmsof: a) ITO/P3HT:PCBM,andb) ITO/PEDOT:PSS/P3HT:PCBMuponsimulatedsolarirradiationat290K.TheNspinisobtainedfromtheaveragedLESRsignalofP3HTunderirradiationfor1hat290K,andisplottedateachaveragedtimeover1h.c)EnergydiagramsatthePEDOT:PSS/P3HTinterfaces,whichschematicallyexplaintheaccumulationofphotogeneratedholecarriersinP3HTmolecularsitesaftersimulatedsolarirradiation.
transfer from P3HT to PEDOT:PSS under darkconditions.Note that thishole formationunderdarkconditions isnotpresented inFig.3c topresent theholeaccumulation inP3HTclearlyunder simulatedsolarirradiationmentionedabove. In the fol lowing, we discuss the correlationshowninFig.2c,d inmoredetail.Firstly,weexplainthedecreaseinVocusinganinterfacialelectricdipolelayerdue toaccumulatedholesat thePEDOT:PSS/P3HT:PCBMinterfaces.Whenaninterfacialelectricdipole layerdue toaccumulatedcharges is formed,a vacuum-level shift occurs at the interface. [33]Suchvacuum-level shiftdecreasesVocof solar cellsbecauseoftheenergy-levelshift formoleculesat theinterface.[34]Wehereevaluatetheinterfacialelectricdipole length (d) due to the hole accumulation atPEDOT: PSS / P 3H T: PC BM i nter face us i ng thecapacitanceformulaQ=CV.LetSbetheareaattheinterfaces,NspinandVocbetheincreaseinNspinand
thedecrease inVocdue to theholeaccumulationattheinterfaces,respectively,ebetheelementalcharge,0be thepermittivity invacuum,rbe thedielectricconstantsofP3HT:PCBMmaterials,[35]then,dmaybeexpressedasfollows:
Using the exper imenta l va lues of S = 0 .4 cm 2,Nspin = 1.2×1012, and Voc = 0.11 V, d is evaluatedto be approximately 1 nm. This length probablycorresponds to the length of alkyl-side chains inP 3H T; [36] the a lkyl- side cha ins are insu latorswithout� electrons. It should be noted that Vocis proportional to N spin, which well explains theexperimentalresultshowninFig.2d. Next,weexplainthedecreaseinJscusingcharge-carrierscatteringduetotheholeaccumulation.Such
49 JEOL News Vol. 49 No. 1 (2014)
scattering is considered tobe independentofothercharge-carrier scatteringmechanisms in solarcells,and thenwemayuse theMatthiessen’s rule for themobilityμinthecellsasfollows:[37]
Here,weuse twomobilityconstituents,μSCandμHA,related to thecharge-carrier scattering insolarcellswithoutandwiththeholeaccumulation,respectively.The latter depends on Nspin with a proportionalityconstantc.UsingEquation(2),wemayexpress thecurrentdensity j insolarcellsusingchargedensitynandinternalelectricfiledEinthecellsasfollows:
Equation(3)showsthatjdecreasesasNspinincreases.Thisbehaviorwellexplains theexperimental resultshown in Fig. 2c. Therefore, the charge-carr ierscattering inducedby theaccumulatedholes in thecells decreases Jsc during device operation undersimulatedsolarirradiation. F i n a l l y, we c o m m e nt o n t h e d i s o r d e r e dmolecular orientation for the hole-accumulationsites. Figure 4 shows the ESR and LESR signalsof P3HT in an organic layered thin fi lm of ITO/
PEDOT:PSS/P3HT:PCBM under dark conditionand simulated solar irradiation, respectively.Thesesignalswereobtainedbysubtracting theESRsignalof PEDOT:PSS from that of ITO / PEDOT:PSS /P3HT:PCBM to present the ESR signal of P3HTclearly. Under dark condition, the ESR signal ofP3HT was observed, which can be ascribed to thehole formation due to the charge transfer betweenP3HT and PEDOT:PSS, as mentioned above. [22]Theobtainedgvalueof2.0018means that thehole-formationsitesareattributabletoP3HTmoleculesinorderedlamellastructuresfromtheanisotropyofthegvaluesforP3HTmolecules.[21]However,theLESRsignalofP3HTaftersimulatedsolarirradiationshowsthedifferentgvalueof2.0022fromthat fororderedP3HTmolecules.This findingclearlydemonstratest h at t he mole c u l a r or ient at ion for t he ho le -accumulationsites isdisorderedandisdifferentfromthat in theordered lamellastructures.Note that thedisorderedmolecularorientationisfurtherconfirmedbythedetailedanisotropyfortheLESRsignal.
Conclusion
We have fabr icated polymer solar cel l s ofI TO / PEDOT: PSS / P 3H T: PCBM / Pd / Li F /A l andperformed simultaneous measurements of LESRand device characteristics under simulated solarirradiation to clarify the deterioration mechanism
of device performance dur ing device operationat the molecular level. We observed a monotonicincrease in LESR signals under simulated solarirradiation,which isascribedto theaccumulationofphotogeneratedholecarriers inP3HTof thedeviceunder operating condition from the microscopicviewpointforthefirsttime.Wealsoobservedaclearcorrelationbetween the increase inNspindue to theholeaccumulationandthedeteriorationofthedeviceparametersJscandVoc.Thiscorrelationdemonstratesthat the hole accumulation in P3HT causes theinitialdeteriorationof thedeviceperformance.Thesitesofholeaccumulationwithdeep trapping levelswere identif ied as being formed in P3HT at thePEDOT:PSS/P3HT:PCBMinterfaces fromthestudyof organic layered films, which explains the morerapidlyincreasingrateofLESRforthedevicesundertheshort-circuitconditionthanundertheopen-circuitcondition.Thus,modificationsat thePEDOT:PSS/P3HT:PCBMinterfacesarerecommendedtoimprovedevice durability by preventing hole accumulationunder device operation. For other solar cells, wealso propose improvements in device durabi l ityby reducing the charge accumulation in the cellsduring device operation based on the microscopicinformation obtained from this and future ESRstudies.
Acknowledgments
TheauthorthanksT.NagamoriandM.Yabusakifor their collaboration. This research was partlysupportedbyGrants-in-Aid forScientificResearch(24560004 and 22340080) from the Japan Societyfor the Promotion of Science (JSPS) and by JST,PRESTO.
References
[1] C.J.Brabec,Sol. Energy Mater. Sol. Cells,83,273(2004).
[2] P. Kopola, T. Aernouts, R. Sliz, S. Guillerez,M. Ylikunnari, D. Cheyns, M. Valimaki, M.Tuomikoski,J.Hast,G.Jabbour,R.Myllylä ,A.Maaninen,Sol. Energy Mater. Sol. Cells,95,1344(2011).
[31] M. C. Scharber, D. Mühlbacher, M. Koppe, P.Denk,C.Waldauf,A.J.Heeger,C.J.Brabec,Adv. Mater.18,789(2006).
[32] R. J. Davis, M. T. Lloyd, S. R. Ferreira, M. J.Bruzek,S.E.Watkins,L.Lindell,P.Sehati,M.Fahlman,J.E.Anthony,J.W.P.Hsu,J. Mater. Chem.21,1721(2011).
Super High Resolution Imaging with Atomic Resolution Electron Microscope of JEM-ARM300F
H. Sawada, N. Shimura, K. Satoh, E. Okunishi, S. Morishita, T. Sasaki, Y. Jimbo, Y. Kohno, F. Hosokawa, T. Naruse, M. Hamochi, T. Sato, K. Terasaki, T. Suzuki, M. Terao, S. Waki, T. Nakamichi, A. Takano, Y. Kondo, T. Kaneyama
EM Business Unit, JEOL Ltd.
Through the technology evolved in the R005 project and the JEM-ARM200F, we developed an atomic resolution electron microscope of the JEM-ARM300F as a new platform for a super high-resolution instrument. The developed microscope is equipped with an ultra-stable cold field emission gun and spherical aberration correctors for probe and image forming systems. The stability of the microscope in TEM was tested by a lattice fringe of a crystal specimen and Young’s fringe for a thick specimen showing beyond (50 pm) -1 spatial information in these images. Ga-Ga dumbbells separated by 63 pm for a GaN [211] specimen was resolved in high angle annular dark field STEM imaging. Sub-50 pm imaging was demonstrated in STEM using a Ge [114] specimen.
Introduction
Sinceanaberrationcorrectionsystemhasbeenpractically established [1-3] , electron microscopesw ith the system drast ica l ly en hanced ana lysi scapabi l ity in a scanning t ransmission electronm i c ro s c o py ( S T E M ) a nd s t r u c t u r a l s t u d y i ntransmissionelectronmicroscopy(TEM).DuringtheR005project(Projectleader:Prof.KunioTakayanagi20 04 -20 09), Tokyo Inst itute of Technology andJEOL Ltd. real ized resolution of 0.05 nm [4,5] ,with a developed 300 kV high resolution electronmicroscopeequippedwithprobe-andimage-formingaberrationcorrectors [3].Foranelectron source,acoldfieldemitterwasusedtoachievehigh-brightnessand narrow energy spread. 63-pm resolution wasdemonstrated using a GaN crystal l ine specimenobserved from the [211] or ientation [6] . Sub-50pm resolution was demonstrated [7] using 47-pmseparated atomic columns of Ge-Ge with Ge[114]specimen[8,9].Liatomiccolumnwasdetectedwithannularbright field (ABF) imaging technique forasampleofLiV2O4[10]. For 200 kV microscopes, spherical aberrationc or rec tors of C E SCOR a nd C ETCOR (C EOSGmbH) were instal led into the JEM-210 0F andtheJEM-2200FS in2003 [11].And theJEM-2100Fwith a STEM aberration corrector imaged atomicstructures of segregat ions at gra in boundar ies[1 2 ] . Nex t , we develop e d a 2 0 0 kV h ig h - end
microscope for commercial use in 2009 ; it is theJEM-ARM200F,whichisequippedwithasphericalaberration corrector for a probe-forming systemin the standard conf igurat ion. The microscopedemonstrated sub-Ångström imaging with Ge andSi specimensobserving from the [112]orientation.Light elements were v isua l i zed by A BF with adeveloped STEM detecting system using a brightfield aperture [13] . Atomic resolution analysis byelectron energy loss spectroscopy (EEL S) andenergy dispersive spectroscopy (EDS) were alsodemonstrated. Refining these technologies cultivated in theR005projectandtheJEM-ARM200F,wedevelopedanewatomicresolutionmicroscopeof300kVJEM-A R M 30 0F. In th is paper, the features and theperformancesofthemicroscopeareintroduced.
Instrumental featuresAppearance of JEM-ARM300F
F i g u re 1 shows appea ra nc es of t he J E M-ARM300F. There are four types of configurationsdependingoncorrectorequipment;basicversionwhichhasnocorrector,STEMcorrectorversion[Fig.1(a)],TEMcorrectorversion,anddoublecorrectorversion[Fig.1(b)].Thewidthofanoperation table is sameas that of the JEM-ARM200F, whereas the JEM-ARM300FishigherthanJEM-ARM200F;theheightof theSTEM-versionof theJEM-ARM300F is3.16mandthatofthedoubleversionis3.44m.TheJEM-ARM300Fisusableatahigheracceleratingvoltageupto300kVandbetterresolutionisattainable,asshownlater.Then,wenamethemicroscope“GRANDARM”
3-1-2 Musashino, Akishima, Tokyo, 196-8558, Japan.
6-SIPs & TMP pumping system and high performance cold field emission gun
Toachievehighqualityvacuumat thespecimen
stage, we employed a differential pumping systemcomposed of six sputter ion pumps (SIP) and aturbomolecularpump(TMP),asshowninFig. 2(a).The specimen stage is pumped with a 150 L/s SIP,the intermediate lens (IL) and the condenser lens(CL)systemsarepumpedwith two individual20L/s SIPs. The pre-evacuation of a specimen holderis performed with the TMP, whereas the TMP is
stoppedduringobservation.TheTMPisalsousedtopumpthecolumnduringthebakingofcolumnand/ora liquidnitrogen tank foranti-contamination.Withthispumpingsystem,apressurelevelof2~10×10-6Paistypicallyachievableatthestagemeasuredwiththeioncurrentofthe150L/sSIP. The developed CFEG is evacuated with nonevaporablegetters(NEG),a200L/sSIP(withnoblepump) and two 20 L/s SIPs. NEG pumps the gasaround theemitter,and the200L/sSIPpumps theaccelerating tube,and the20L/sSIPspumps the1stand 2nd intermediate chambers [14] . The pumpingsystem enables ach iev i ng a n u lt i mate cu r rentstability of the CFEG. Figure 2(b) shows currentstability,aftera flashingof theemitterandasettinganemissioncurrent tobe10μA.Total time for theprocedures of f lashing and auto-emission takesapproximately1min.Owingtoanultra-highvacuumof4~10×10-9Pameasured at theaccelerating tube,theprobecurrentkeeps90%of initialcurrentevenafter fourhours.Thus, the6-SIPs&TMPpumpingsystem is very effective to keep a specimen stage
T he J E M -A R M 3 0 0 F i s equ ipped w it h a naber rat ion c or rec tor, wh ich was developed i nthe R005 project [3] . The spherical aberration iscompensated by two three-fold astigmatism fields[1]generated in twododeca-poles (Fig. 3(a)).Asafurtheroptical innovation,anelectron trajectory isexpanding toward a specimen in the corrector andlargelyexpandingbetween thecondenser-mini lensand the transfer condenser-mini lens (CM-CMT)orbetween theobjective-mini lensand the transferobjective-mini lens (OM-OMT) [3].Theexpandingtrajectoryenablesus toreducedisturbance inaboveelements resulting that extra chromatic aberrationand noise from the corrector can be reduced. Weca l l the system as ETA (Expanding TrajectoryAberration)opticalsystem(Fig.3(b),(c)).
Foracorrectorcontrol software,wedevelopedt he cor rector system modu le (J EOL COSMO :correctorsystemmodule),whichmeasuresaberrationsup to5thorder.For tuningof aprobe-forming lenssystem,underandoverdefocusedRonchigramsarerecorded tomeasureaberrationsby the segmentedRonchigramauto-correlationfunctionmatrix(SRAM)method [15] (Fig. 4 (a-d)).For tuningofan image-forming lens system,diffractogramtableau [16]areusedforaberrationmeasurement(Fig.4(e,f)).
Highly stabilized column
Figure 5(a) showsapowerspectrumofFouriert ra n s for m f rom t he T E M i mage of a S i [110 ]specimenatanacceleratingvoltageof300kV,wherelattice informationbeyond50pmcanbeconfirmed.Young’s fringe test includingnon-linear termusinga thick specimenofgoldparticlesonacarbon film
isshowninFig.5(b).Fringesareextendedtospatialinformation better than (50 pm) -1. These resultsindicated that themechanical andelectric stabilityof themicroscope realized the capabilityof 50pmresolution. Figure5(c) isahigh-resolutionHAADFSTEMimageofSi [110]at300kVwithanacquisition timeof10sandFig.5(d)isonewithanacquisitiontimeof80s.Ahigh-resolutionimagewitha longacquisitiontime in Fig. 5(d) shows little distortion, indicatingthat the scanning system and the stage are highlystabilizedagainstnotonlyhighfrequencydisturbancebutalsolowfrequencyfluctuation.
Detecting System
The microscope is equipped with a viewingchamberandadetectingchamber(Fig. 6(a)and(b)),where fourSTEMdetectorscanbeattached totally:high-angleannulardark field (HAADF), low-angleannulardarkfield(LAADF),annularbrightfield(ABF),
Themicroscopecanbeoperatedatwiderangeacceleratingvoltages from80 to300kV.Figure 7(a-c)showhigh-resolutionHAADFimagesofaSi[110]crystall ine specimen. Observation at 300 kV canimageasub-Ångströmstructure,asshownlater,and80kV imaging isuseful for softmaterials to reducethe specimen damage. At 80 kV, 136 pm imagingis achievable, as shown in Fig. 7 (c). Operation at160 kV is optionally available (Fig. 7(b)), and 160kV imaging isuseful fora semiconductingmaterialbecause of h igh-resolut ion and relat ively smal ldamagecomparedwithhighervoltage.Lowervoltageoperationat60kVisoptionallyavailable.
Developed two objective pole pieces and ultra-high resolution imaging
We newly developed two types of pole piecesof an objective lens for the JEM-ARM300F. Theyare FHP (full high resolution pole piece) for theultra-highresolutionconfigurationandWGP(widegap pole piece) for the high resolution analyticalconfiguration. FHP is designed [3] for ultra-highresolut ion obser vat ion w ith a smal l chromat icaberrat ion ; a chromat ic aberrat ion coef f ic ientva lue of 1. 35 m m i n ST EM / T EM at 3 0 0 kV i ss ign i f icant ly sma l l . WGP i s usef u l for a h igh-
sensitive EDS analysis with a large solid angle ofthe detection, for a thicker special holder owingto a large space inside a pole piece gap, and for ahigher ti lt angle of a specimen holder comparedwithFHP. U lt ra-h ig h resolut ion ST EM i mages wereobtained using GaN and Ge crystalline specimenusing the developed STEM ETA corrector. 63 pmseparation was clearly detected in a raw image(Fig.8(a))andan intensity lineprofileofHAADFat300kVusingtheFHPpolepiece(Fig.8(d)).Spotssmaller than(63pm)-1 informationweredetected inthepowerspectruminFig.8(b). Asafurtherchallenging,asub-50pmresolutionwas imaged inHAADFusing theFHP(Fig. 9).47-pm separation was confirmed in the intensity lineprofileinFigure9(f)andaspatialinformationof(47pm)-1wasdetectedinthepowerspectruminFig.9(d).Themicroscopeof“GRANDARM”hasacapabilityto imagesub-50pmresolution.Thestabilitymustbeeffectiveinnotonlyultra-high-resolutionimagingbutalsorobustdataacquisitionforanalysisandstructurestudy.
Summary
We d e v e l o p e d a n e w a t o m i c r e s o l u t i o nmicroscopeof the JEM-ARM300F.Thedevelopedcold fieldemissiongun showedultra-highemissionstability due to an ultra-high vacuum around theemitter.Stabilityof themicroscopewas confirmedusing the power spectrum of T EM images andan atomic-resolut ion ST EM image with a longacquisitiontime.Ultra-highresolutionSTEMimagesweredemonstrated,byusingthedevelopedcorrectorand objective lens. The microscope wil l be newplatformforanatomicresolutionstudy.
The authors thank Prof. K. Takanayagi , Y.TanishiroofTokyoInstituteofTechnologyandothermembers of JST CREST R005 project for adviceto establish the R005 microscope and develop theelements of ETA correctors, ultra-high-stabilizedcold f ield emission gun, and F H P. The authorsappreciate Emeritus Prof. M. Tanaka of TohokuUniversity for organizing of the CREST project.Theauthors thankProf.Y. IkuharaandN.Shibataof theUniversityofTokyo for collaborationof thedevelopmentoftheGRANDARM.
[ 3 ]Hosokawa,F.,Sawada,H.,Kondo,Y.,Takayanagi,K., Suenaga, K. ; Development of Cs and Cccorrectors for transmissionelectronmicroscopy.Microscopy,62,23-41(2013).
[ 5 ]Takayanagi, K., Kim, S., Lee, S., Oshima, Y.,Tanaka,T.,Tanishiro,Y.,Sawada,H.,Hosokawa,F., Tomita, T., Kaneyama T., and Kondo Y. ;Electronmicroscopyatasub-50pmresolution.J. Electron Microsc.,60,S239-S244(2011).
[ 6 ]Sawada,H.,Hosokawa,F.,Kaneyama,T.,Ishizawa,T.,Terao,M.,Kawazoe,M,,Sannomiya,T.,Tomita,T.,Kondo,Y.,Tanaka,T.,Oshima,Y.,Tanishiro,Y.,Yamamoto,N.,andTakayanagi,K.;Achieving63pmresolutioninscanningtransmissionelectronmicroscopewithsphericalaberrationcorrector.Jpn. J. Appl. Phys.,46,L568-570(2007).
[ 7 ] Sawada,H.,Tanishiro,Y.,Ohashi,N.,Tomita,T., Hosokawa, F., Kaneyama, T., Kondo, Y.,and Takayanagi, K. ; STEM Imaging of 47-pm Separated Atomic Columns by SphericalAberration-CorrectedElectronMicroscopewith300 kV Cold Field Emission Gun. J. Electron Microsc.,58,357-361(2009).
[ 8 ]O'Keefe, M., Allard, L., Blom, D. ; HRTEMimagingofatomsatsub-Angstromresolution.J. Electron Microsc.,54,169-180(2005).
[ 9 ]Erni, R., Rossell, M. D., Kisielowski, C., andDahmen,U. ;Atomic-resolution imagingwithasub-50-pmelectronprobe.Physical Review Letter,102,096101(2009).
[10] Oshima,Y.,Sawada,Hosokawa,F.,Okunishi,E.,Kaneyama,T.,Kondo,Niitaka,S.,Takagi,H.,Y.,Tanishiro,Y.,andTakayanagi,K.;DirectimagingoflithiumatomsinLiV2O4bysphericalaberration-corrected electron microscopy. J. Elec t ron Microsc.,59,457-461(2010).
[11] Sawada, H., Tomita, T., Naruse, M., Honda,T., Hambr idge, P. , Hartel , P. , Haider, M.,Hetherington,C. J.D.,Doole,R.C.,Kirkland,A.I.,Hutchison,J.L.,Titchmarsh,J.M.,Cockayne,D.J.H. ;Experimentalevaluationofasphericalaberration-correctedTEMandSTEM.J. Electron
[13] Okunish i , E . , Sawada, H., and Kondo, Y. ;Experimentalstudyofannularbrightfield(ABF)imaging using aberration-corrected scanningtransmission electron microscopy (STEM).Micron,43,538-544(2012).
[15] Sawada, H., Sannomiya, T. , Hosokawa, F. ,Na ka m ich i , T. , Ka neya ma, T. , Tom ita , T. ,Kondo, Y., Tanaka, T., Oshima, Y., Tanishiro,Y.,andTakayanagi,K. ;Measurementmethodofaberration fromRonchigrambyautocorrelationfunction.Ultramicroscopy108,1467-1475(2008).
Advanced Analysis of Active Materials in Li-Ion Battery by XPS and AES
Kenichi Tsutsumi, Masahide Shima and Akihiro Tanaka
SA Business Unit, JEOL Ltd.
Introduction
AugerElectronSpectroscopy(AES)andX-rayPhotoelect ron Spectroscopy (X PS) are w idelyusedasanalyticalmethods thatperformelementalanalysisofthespecimensurfacedowntoonlyafew-nanometer region. Inparticular, these twomethodsenable direct detection of lithium which is a lightelementand furthermore,providebothquantitativeevaluation and chemical bonding-state evaluation.Thus, themethodshavebeenused for researchanddevelopmentofvariousmaterials inaLi-ionbattery.However, when examining the number of booksand papers covering the topics of a Li-ion batteryusing the Internet literature searching systems (ex.http://scholar.google.com),documentswhichpresentresearch achievements by the use of XPS reach asmanyas2000,butthosebyAESare issuedonly500under the same searching conditions. This may bepartlydue to thedifferencebetween thenumberoftheXPSusersandtheAESusers.Butoneessentialreason is that the AES users misunderstand AEStobedifficult toanalyzeLiandhard tobeused forthe Li investigation. In fact, the kinetic energy ofan Auger electron of Li KVV is about 50 eV. Thismeans thatanAuger spectrumofLi is locatedatalargebackgroundconsistingof secondaryelectronsand also it overlaps with Auger peaks of theotherelements.Thus,dependingonthespecimen,itmaybedifficulttoidentifytheLipeakfromAugerspectrum.AnotherreasonisthattheescapedepthofanAugerelectronofLi isveryshort.TheintensityofLiKVVpeaks decreasesdrasticallywhenasmallvolumeofcontaminants (only1 to2nmthickness) coveringaspecimen(containingLi), thusmaking itdifficult todetectLi.Ontheotherhand,inthecaseofXPS,thekineticenergyofaphotoelectronemitted fromLi isapproximately1200to1400eV,whichismuchhigherthan thatof thecorrespondingAugerelectron.Forthis reason,Li1s iseasilydetected inXPSspectrumeven if thespecimensurface is rathercontaminated.FurthermoreinXPS,theoverlapoftheLipeakwiththeotherpeaksis less, thusfacilitatingthedetectionofLi.Lithiumpeakoverlapsespeciallywith thatof
transitionmetal suchasMn,Fe,Co,Ni,etc.,whichare frequently-used transitionmetals for theLi-ionbattery. However in the case of XPS, only a Fe3ppeakoverlapswithaLi1speak.That is, in thecaseofAES,preventingdepositionofcontaminantsontothe specimensurface isofprime importanceat theprocessofpre-treatmentof the specimen. If apre-treatment technique is improper, even if the samespecimenisanalyzedbyXPSandAES,therehappenmanycaseswhereLi isdetectedbyXPSbut isnotdetectedbyAES. In this context, Li seems to be an unsuitableelement tobeobservedbyAES.Butweemphasizethefollowingpoints.Iftheoperatorunderstandsandgains a proper specimen pre-treatment techniqueand precautions to be taken during AES analysis,thedetection intensityofLi itself ishigher thanthatanalyzedbyXPS. Inaddition,AEShascapacity tomeasure the Li distribution in a single particle ofLi-ion battery materials even at magnifications ofseveraltensofthousandsoftimestoseveralhundredsof thousandsof times.That is,AESandXPShaveadvantagesanddisadvantages inLianalysis toeachother.Thus,a creativeuseofAESandXPS,whichisbasedonthe fullunderstandingof the featuresofthetwoanalyticalmethods,willplayaneffectiverolein Li analysis. In this article, we focus on analysistechniques for Li-ion battery and report on ouranalysisresults.
Pre-Treatment Technique for Lithium Detection in Surface Analysis
Precautions when handling materials containing Lithium
Due to its characteristics, a Li ion in Li-ionbattery mater ia l moves easi ly toward mater ia ls(irrespectiveof solidand liquid)whichare locatednea r the L i - ion bat ter y mater ia l s . T herefore,precautionshave tobe takenwhenperformingpre-treatmentofthespecimen.The followingexampledescribesaLiCoO2particle,one of Li-ion battery mater ia l . W hen the A ESanalysis is applied to a clean particle surface, itis possible to clearly observe the Li peak in thedifferential spectrumas shown inFig. 1.Butwhenultrasoniccleaning(byethylalcohol,etc.) isapplied
3-1-2 Musashino, Akishima, Tokyo, 196-8558, Japan.
Kenichi Tsutsumi, Masahide Shima and Akihiro Tanaka
SA Business Unit, JEOL Ltd.
to the specimenaimedat removingcontaminationsonthespecimensurface,theLipeakbecomeshardlydetected f rom the surface of LiCoO 2 part icles .This is considered toarise from thedecreaseofLiconcentrationon the surface,asa resultof the factthatLielements in theregionof theparticlesurfaceare eluted into an ethyl alcohol solvent during thecleaning.AsshowninFig.1, theLipeakdisappearsandthissubstanceisfoundtobecobaltoxide.Thereis high possibility that this phenomenon arises fornot only water but also the other organic solvents;therefore, AES and XPS which analyze the topsurfaceof thespecimenrequirespecialprecautions.Consequently, in surface analysis by AES or XPS,wepointoutthat it isnotrecommendedtocleantheparticle surface with inorganic or organic solventsexceptforspecialcases. IftheLiCoO2particlesurfacehascontamination,we recommend the followingway.Donotclean thespecimen surface with a solvent and expose a newclean surfacebygroundingLiCoO2powderswithapestleinamortarforanalysis.
Pre-treatment technique and precautions for powder specimen at Auger analysis
Concerningspecimenpre-treatment techniques
and the tips for analysis when surface analysis isapplied toapowderspecimen,weexplained in2009J EOL EPM A / Sur face ana lysis Users’ Meet ingdocument [1] . Thus in this article, we introduce“Method for spraying a powders onto a carbon-specimenstub”. Thecarbon-specimenstubpresentedhere is thecarbonspecimenstub(PartsNo. :600154386)thatissoldbyJEOLLtd., shown inFig. 2.Thedimensionsof this stub are 10 mm in diameter and 5 mm inheight. In thepre-treatmentprocess, this specimenstub ispolishedwithsandpaperwith itsgritnumberabout1000 toexposea relatively flat surface.Afterthat,grooveswitha little smalldepthare fabricatedon the surface with a head-pointed tool (precisionscrewdriver, etc.). After spraying the powder ontothiscarbon-specimenstub,slightlyscrubthecarbon-specimen stub surface for filling the powders intothe groove with paper (medical paper, etc.) . Theremaining powders on the stub surface must beremovedwithablower. Aftermountingthecarbon-specimenstubonthestandard specimenholderwithout spring inside thestandardspecimenholder(shown inshown inFig. 3left),preparationforAESiscompleted. This specimen pre-treatmentenables particleswith various configurations to be placed on thecarbon-specimenstub.Forexample,asisshowninthe
Beforecleaning
Beforecleaning
Aftercleaning
Aftercleaning
Lightlypolish thesurfacewithsandpaper.
Make grooves on the mount with ahead-pointedtool.
Rub a powder specimen into thegroovesandremoveextrapowders.
right-sidesecondaryelectron imageofshown inFig.3right,alargesurfaceareaisseenwherepowdersarepermeatedwithhighparticledensityinthefabricatedgrooves. Inaddition,on thesurface that ispolishedby sandpaperwith its gritnumberabout1000, finesurface irregularities or f ine polished marks areproduced. In this state, singleparticlesandclustersexistintheirregularitiesorthepolishedmarks.Sincevarious-shapeparticlesexiston the specimen stub,it becomes possible to perform analyses at severalstateswithonlyone specimen stub.Forexample,asingleparticleisselectivelyanalyzedandtheaveragecompositionofaregionwithhighparticledensity isalsoanalyzed.
Pre-treatment technique and precautions for powder specimen at XPS analysis
W he n a n i n s u l a t i n g p owd er s p e c i m e n i smeasuredbyXPS, therearise severalproblems thattheinsulatingspecimenisinfluencedbychargingandadifficulty in fixing the specimenon the specimenholder.Forcomparisonwiththeotherfixingmethods,wereported in2006JEOLEPMA/SurfaceAnalysisUsers’Meetingdocument[2].Themostsuitablepre-treatment method for XPS is to mold a specimeninto a pellet (Fig. 4) . To perform pellet molding,a relatively large volume of powder specimen isrequired;however, the specimensurfacemoldedbythismethod is flat, thus leading toadvantages thatXPSmeasurementsensitivityishighandit iseasyto
suppresstheinfluenceofcharging.
Detection and Quantification of Lithium in Surface Analysis
Detection and sensitivity of lithium
In AES, rather surpr isingly, the sensit iv ityof Li is not so low, enabl ing us to complete theanalysis ina shortperiodof time fornotonlypointanalysis but also Auger mapping. We comparedthe detection sensitivity for Li in AES with thatin X PS. As i s show n in F ig . 5 , when the peakintensities in thestandardspectraacquiredbyAESis compared between C (carbon) and Li under thesame measurement conditions, the peak intensityof Li is approximately 4 times higher than that ofC. In the case of XPS as well, the peak intensitiesof pure mater ia ls should d i rect ly be comparedin principle; however, it is hard to acquire the ListandardspectruminXPS.Duetothislimitation,weperformedcomparisonofthedifferenceinionizationcrosssectionsexcitedwithanAlKα(1486.6eV)X-raysource as the difference in the sensitivity betweenCandLi (right toFig.5).As is indicated inFig.5,compared to thepeak intensityofC,Liprovidesavery low sensitivity, approximately 1/18 that of C.That is,when theLi intensity is comparedbetweenA ES and X PS based on the C intensity as thestandardpeak intensity, it is revealed thatAEShasmuchhigher sensitivity thanXPS,approximately72
Kinetic energy of Li photo electron excited by AlKα: 1437 eV
Fig.5
Fig.6
C o m p a r i s o n o flithiumsensitivitybetweenAESandXPS.
Meanfreepathofanelectroninasolidspecimen.
times. However ingeneral, it is likely toberecognizedthat XPS allows easier detection of Li than AESdoes. This misunderstanding leads to the fact thatXPS is frequently used for surface analysis of Licompared to AES as mentioned in “Introduction”in thisarticle.This fact isowingto thedifferenceofescapedepthbetweenAESandXPS,whichislargelydue to thedifferenceofkineticenergybetween theLiAugerelectronandtheLiphotoelectron.Figure 6shows theenergydependenceof themean freepathofanelectroninasolid.TheLiKVVAugerelectrononlypossessesakineticenergyofapproximately50
eV.Ontheotherhand,Ontheotherhand,thekineticenergyofLipeakobtainedusing theAlKα line inXPSprovidesmore than1400eV.This result leadstoa largedifferenceof themean freepathbetweentheAugerelectron(inAES)and thephotoelectron(inXPS).When theescapedepthof theelectron isestimatedtobeapproximately3timesthemeanfreepath,theescapedepthoftheLiKVVAugerelectronis less than2nm.This implies that theescapedepthofAugerelectroninAESisapproximately1/3timesthatofphotoelectron inXPS.Thus, it is found thatdepositionofvery small amountof contaminationsonto the specimen surface g ives r ise to a large
Peak intensity ratio in AES 10kV, 10nA, E/E:0.5 %
ILi / IC 4 Peak intensity ratio
ILi / IC 1/18 Peak intensity ratio
Photoelectron cross section in XPSX-ray: AlKα (1486.6 ev)
Fig.7 Images of lithiummetal (left: beforec u t t i n g , r i g h t :immediately aftercutting).
White portion
Blackportion
Metal glazeportion
White portion
Blackportion
Metal glazeportion
63 JEOL News Vol. 49 No. 1 (2014)
decrease in the peak intensity and there is highpossibility thatLicannotbedetected.That is,whenperforming AES for Li, specimen pre-treatment isprime importance inaddition to theanalysis itself,andsufficientcare is required for thespecimenpre-treatment. Next,weshowspectraofLichemicalcompoundsobta ined by actua l measu rement w ith an X PSinstrument [3].First,ameasurementresultofmetalLi is presented. Normally, to avoid oxidation ofmetal Li, this material is preserved by immersingit in paraff in oi l. However, Li reacts with waterevenat roomtemperatureand transformto lithiumhydroxide;finallytransformstolithiumcarbonatebyabsorbingCO2intheair.Asaresult,whenpreservedLiisextractedtotheair,thecolorofthesurfaceofLiturnstowhite(Fig. 7).CuttingLiusingarazoratthisstateenablesasurfacewhichcontainsmetalglazetobeexposed. However in theair, aportion showingmetalglazechanges itscolor toblackwithinseveral10sandfurtherchangestowhite. For thewhiteportionwhereLi transformed tolithiumcarbonate, evenafter the specimen surfacewasetchedbyionsputteringinalongperiodoftime,oxygenandcarbondidnotvanish;thereforemetalLispectrawereunable tobeacquired.Figure 8 showsthe spectra acquired from the l ithium carbonateportionbyXPSmeasurement. Since Li metal reacts with water in the air totransformtooxide,abulkofLi iscut inaglovebagfilledwithdrynitrogenandthecutLibulkissubjecttoXPSwithLimetalglazekept.After thisprocess,a spectrum corresponding to metal Li and lithiumoxidewereobtained.Figure 9 showsLi1sacquiredfrom threekindsof chemical compounds includingl i t h iu m ox ide, l i t h iu m c a rbonate a nd l i t h iu mphosphate. Table 1 lists spectral peak positions ofthose chemical compounds.Aspresentedhere, theuseofXPSenablesus toclearlyobserveLi spectraandalso toobtain the changeof chemicalbondingstatesofLiitself.
Uptotheprevioussections,weexplainedaboutthe specimen pre-treatment and precautions aboutdetectingLi. If theabove-explainedprecautionsaretakenintoconsideration,arelativelyskilledoperatorforAESdoesnothaveadifficulty in analyzingLiby not only XPS and but also AES. However forthenextchallenge, that is,Liquantificationand itsaccuracy, there isnoestablishedmethodatpresent.In this section, as thequantificationmethodofLi,therelativesensitivityfactormethodandtheabsoluteintensityquantificationmethodaredescribed. Lihasonly twooccupiedorbitalsof1sand2s.In particular, the 2s orbital forms a covalent bondorbital with the bonded-counterpart element, thusdepending on the bonding state, the peak positionand thepeakshapeof theAuger spectrumchangesgreatly. Furthermore, the Li KV V peak around50eVexistson the largebackgroundof secondaryelectrons, the valence-electron peak of the otherelementsoverlapwith theLiKVVpeak.Therefore,
theuseoftherelativesensitivityfactormethod,whichhasbeenusedasaquantitativeanalysismethod forAES, makes it very difficult to correctly estimateconcentrationofelementscontainedinthespecimen. Thepeak intensitiesofAESappearaspeak-to-peak in the differential spectrum (Fig. 10). In therelativesensitivityfactormethodforAES,theatomicconcentration ratio is obtained in the fol lowingway. A measured peak intensity is multiplied bythe relativepeak intensity ratio (relative sensitivityfactor) among pure substances, then the intensitycorrection is executed. After that, the summationof those corrected values is normalized to 100%.Finally,theresultantnormalizedvalueisregardedtobetheatomicconcentrationratio.Ifonlytherelativesensitivityfactorisobtained,simplecalculationofthepeak intensitiesenablesus toobtainaquantitativeanalysis result easi ly. This method is a widely-recognized,generalquantitativeanalysismethodforsurface analysis. However in AES, the differentialpeak intensity is significantlysensitive to thechangeof peak shape. This is due to the change of thechemical statesand theoverlapof thepeaksof theotherelements,whichgivesrisetoalargeerrorinthequantitativeanalysis result.These largeerrorscausevery seriousproblemwhenquantifyingLi, typicallyin the case of a very low concentration Li (atomicconcentration ratio: 5 % or less), thus the peak ofLicannotbedetectedasan independentpeak.TheexistenceofLiinthespecimenisnotoftendiscoveredwithoutacurvefittingwhichshowsLipeakburiedinthepeaksoftheotherelements. Inthiscontext,weproposethatwhenperformingquantitative analysis of Li, instead of using therelativesensitivity factormethod, it is reasonable touse theabsolute intensityquantificationmethod forthequantitativeanalysisofLi.TheprocedureofthismethodisdescribedinSteps(1)to(4)below.(1) MeasuretheAugerspectraof thespecimenwith
takingcareaboutpretreatment.(2) Deconvolute the measured spectra into the
respective components based on curve-fittingcalculationsusingthestandardspectra.
(3) Convert thespectral intensitiesof therespectivecomponents into the corresponding atomicconcentrationswith the intensity ratiobetweent he s p e c t r a l i nt en s i t ie s o f t he re s p e c t ivecomponentsatthecorrespondingchemicalstatesandthoseofthestandardspectraacquiredunderthesamemeasurementconditions.
(4) Examine the quantification accuracy with thecalculatedtotalatomicconcentration.
InStep (1), it isnotalwaysnecessary to selecta high energy resolution for the Auger spectra. Itis possible to calculate the quantification resultseven with the energy resolution of 0.5% with thestandardspectraofthesamemeasurementcondition.Off course, it is effective to calculate with highenergy resolution spectra in the caseof containingdifferentchemicalstatesspecies[4].ItisimportanttomeasuretheAugerspectrawiththeenergyresolutionsuitable for thepurposeofanalysisand toperformmeasurement by taking care about the specimendamagedue toelectron-beam irradiationorAr-ionirradiation. InStep(2),curvefittingcalculationisperformed
(2014) Vol. 49 No. 1 JEOL News 64
O1s
Li1s C1s
O
C Li
CH
531.0 eV
289.7 eV 54.6 eV CO3
wide
IAg
ICu
Fig.8 XPSmeasurementr e s u l t o f t h el i t h i u m w h i t eportion.
Table 1 Spectral peak positions of lithium related substances.
Fig. 10 Definition of peak intensity for the relative sensitivity factor method.
by using the least squares method based on thestandard spectra. In this article, we omit to showthe detailed calculation method, so please refer toReference [5] “Peak Deconvolution Analysis inAugerElectronSpectroscopy”.An important issuethat shouldbenotedhere is to surelydifferentiatethe spectra at curve f it t ing for subtract ing thebackground. In the case of the Auger spectra, theAugerelectron itself isoneof secondaryelectrons;therefore, it is impossible to clearly dist inguishbetweenthebackgroundandtheAugerspectra,bothof them are contained in the N(E) spectra beforedifferentiation.Executionofdifferentiationimprovesthe calculation accuracy by simply subtracting thesecondary-electron background, thus making itpossible toperformcurve-fittingcalculationfocusedonthedifferenceofthepeakshape. InStep (3), theabsolute intensityquantitativecalculationisperformedbycomparingtheintensitiesof the respective components acquired in Step (2)withtheintensitiesofthestandardspectraunderthesamemeasurementconditions.Here,Fig. 11presentstheprocedureoftheabsoluteintensityquantificationmethodforthemeasuredspectraofLiCoO2particles
showninFig.1asanexample.Here,acomplexoxideofLiCoO2particlesisassumedtobeamixedoxideofLi2OandCo3O4. AugerspectrumofLiCoO2particlesindicatedinFig.11isacquiredunderthemeasurementconditionsofprobecurrent(10nA,10kV)andenergyresolutionof0.5%.Withthisenergyresolution, it isdifficulttodistinguish thedifferenceof thechemical statesduetotheCo2+andCo3+becauseCospectrahavealmostthe same shape. Thus, it is possible to use CoOor Co3O4 as the standard spectra for the absoluteintensity quantification method for LiCoO2. Here,thequantitativecalculationwasperformedwith thestandardspectraofcobaltoxide(Co3O4)andlithiumoxide(Li2O). In the regionof30 to60eVwhere theLipeakisdetected, theCopeak isalsodetected; therefore,when curve f itting is applied using the standardspec t ra of L i 2O a nd C o 3O 4, L i 2O c omponent sare extracted. The peak intensity at the presentmeasurement is 1078countsand thepeak intensityof the standard spectrum of Li2O under the samemea su rement c ond it ion s i s 312 7 c ou nt s . T h i sresult shows that if the absolute intensity of Li2O
Co3O4 component contained in the actuallymeasured spectrum
Co3O4 standard spectrum acquired under the same conditions of the actually measured spectrum
Li2O component contained in the actually measured spectrum
Li2O standard spectrum acquired under the same conditions of the actually measured spectrum
Deconvoluted result of chemical states of O contained in the actually measured spectrum
componentsaftercurve fitting reaches3127counts,theintensityequalsto2/3=66.66…%,whichistheLiatomicconcentrationofthestandardspecimen.Fromthis result, the Li concentration measured by theabsolute intensityratio is foundtobeapproximately23.0%. Similarlyinthiscase,theenergyrangeof600to850eVwhereCoisdetected,allowsonlythreepeaksofCo toexist, thus theCoatomic concentration iscalculatedtobeapproximately24.5%bycomparisonwith the absolute intensity of the Co3O4 standardspectrum. On theotherhand, theOpeaksaround470 to530eVexistinsuchawaythattheOpeakbondedtoLioverlapswith theOpeakbonded toCo.But therespectivepeakpositionsaredifferentandtherefore,when the corresponding two standard spectra aresubject to curve-fitting calculations, the absoluteintensit ies separated to two components can beobtained.Asaresult,whentheatomicconcentrationsoftherespectivecomponentsarecalculatedfromtheobtainedabsolute intensity ratio, theconcentrationofObondedtoLi iscalculated tobeapproximately12.1%,whereastheconcentrationofObondedtoCo
isobtainedtobeapproximately41.7%. Table 2 l ists the total atomic concentrationand the respective atomic concentrations obtainedfrom theabsolute intensity ratio.Whenwe see thetotal value of atomic concentrations, the total isapproximately 100%, indicating that the error isonly1.3%.In the relative sensitivity factormethod,the concentrations are inevitably normalized, thusmaking it difficult to discuss the error about thequantitativecalculationresult.Tothecontrary,whentheabsolute intensityquantificationmethod isused,the obtained total atomic concentration enablestheerror tobeestimated.Furthermore, theatomicconcentrationratioamongtherespectiveelements isalsoanimportantpoint.TheconcentrationratioofLitoObondedtoLiisapproximately2:1,thusallowingustoestimatetheformationofLi2O.Inaddition,theresultreveals thatLiCoO2 is formedbecauseLi:Co:O(totalatomicconcentration)equalsto1:1:2. Aswehavepresented,even if thepeaksof theotherelementsoverlapwith the targetpeak like inthe caseof theLipeak, it ispossible toobtain theconcentration of the target element by applyingtheabsolute intensityquantificationmethod to the
Fig.12 Li-ionbatterypowderm a t e r i a l m o l d e dinto a pellet and itsbackscatteredelectronimage.
Ato
mic
con
cent
ratio
n%
Ato
mic
con
cent
ratio
n%
67 JEOL News Vol. 49 No. 1 (2014)
spectrasubjecttocurvefittingcalculation. OntheotherhandinthecaseofXPS,whentheLiatomicconcentration iscalculatedby therelativesensitivity factormethod,a relativelyhigh-accuracyresult can be obtained because of the fol lowingreasons. The change of the Li spectrum, whichdepends on the difference of its chemical bondingstates,leadsonlytotheshiftoftheXPSpeakaswasdescribedintheSectionof“DetectionandsensitivityofLi”.Inaddition,thequantitativecalculationbasedontherelativesensitivityfactormethodisappliedforthearea intensityof thespectra.However, it shouldbenotedthat theunevennessof thespecimenhas toberemindwhenXPSisused[6].Also in thecaseofthepowderspecimen,thereishighpossibilitythatthespecimenitselfhasunevenness. AsisshowninthebackscatteredelectronimageinFig. 12, thepowderspecimenmoldedintoapelletis found to have an unevenness of approximately100μm.But, specimenobservation innormalXPSisperformedwithanopticalmicroscope(OM), thusmaking itdifficult to judgethespecimenunevennesswhichcannotbeobservedwithOM.In suchacase,it becomes possible to reduce the influence of theunevenness by sufficiently increasing the analysisdiameter.Figure 13 shows thequantitativeanalysisresults inwhichquantitativeanalysiswasperformedby setting the analysis diameter to 100 μm and 3mm for the specimen shown in Fig. 12. From Fig.13,whentheanalysisdiameter is set to100μm, it isfound that theatomicconcentrationdependson theanalysisposition, thus theLi atomic concentrationdiffers greatly. On the other hand, in the case oftheanalysisdiameterof3mm, suchadifference inthe quantification value is not found. Thus, whenanalyzing the specimen by XPS, it is necessary totakeaccountofthespecimenunevenness.
Comparison of Auger analysis of powder material for Li-ion battery with the other analytical methods
In this Section, we introduce actual analysisexamplesofAESforLi-ionbatterypowderspecimenandcomparetheAESresultwiththeresultsobtainedbytheotheranalyticalmethods.Aspecimentreatedin thepresentSection isanNMCpowder specimen
(hereinafter called NMC specimen) composed ofMn, Co and Ni equal to 1:1:1 (shown in Fig. 14).This specimen was subject toanalysis afterproperspecimenpre-treatmentwasapplied. First,prior toanalyzing the specimenbyAES,we used an Energy-Dispersive X-ray FluorescenceSpectrometer (XRF) instrument (JEOL ElementAnalyzer JSX-3100RII) toexamine whether ornottheaveragecompositionratiowasMn:Co:Ni=1:1:1,whichwasadesignatedvalue.Ingeneral, inordertoexamine theaverage compositionofLi-ionbatterypowder mater ial , ICP-MS (Inductively CoupledPlasmaMassSpectrometry)orICP-AES(InductivelyCoupled Plasma Atomic Emission Spectrometry)based on the wet analysis is mainly used. The wetanalysis has a big mer it of h igh quant i f icat ionaccuracy,butalsohasademerit thatall volumeofthespecimenneedstobedissolvedinaacidsolutionor an alkali solution and it requires a substantialtimeforsolutionadjustment; therefore, it isdifficultto treat a number of samples. Thus, in order toexamine whether or not the average compositionratioofMn,CoandNi in theLi-ionbatterypowderspecimen (except for Li concentration) is correct,the XRF method is efficiently used as a primaryscreening technique because the XR F analyzer(Element Analyzer) enables us to obtain a high-accuracy quantitative analysis result in only a fewminutes. Figure 15 shows a spectrum acquired bytheXRFmethodandanalysis resultsobtainedwiththis spectrum.TheXRFanalysis results reveal thatthe atomic concentrations of Mn, Co and Ni areapproximately 13% each and those elements arecontainedwiththeratioof1:1:1. Next, the specimenpowderswere molded intoa pellet to measure the concentration of Li thataveragely spreads over the surface of the powderspecimen. Then, we used a JEOL JPS-9200 basedon X-ray P hoto ele c t ron S p e c t ro s c opy ( X P S )for performing top-surface analysis of an NMCspecimen.Figure 16 shows thequantitativeanalysisresult.ComparedtotheresultobtainedbyXRF,thedetectedconcentrationofNi ishigher than thoseofMnandCo,andalsoitisfoundthatasmanyas14.8%Liexistsatthetopsurfaceofthespecimen. Next,inordertoanalyzetheaveragecompositionofthespecimensurfaceusingAES,wemeasuredthe
Fig.14 NMC spec imen for Li - ionbattery actually used in ourexperiment (Mn,CoandNi=1:1:1).
averagespectruminanareaof50μmdiameter(Fig. 17).Figure17 shows thequantitativeanalysis resultbasedontheabsoluteintensityquantificationmethodexplained in the Section of Lithium quantificationin an average area. In the present result, the totalvalueofeachatomicconcentration reaches96.7%,indicating that theerror isapproximately3.3%.Asa reference,Table 3 listsacomparisonresultof thenormalized quantification values obtained by theconventional relative sensitivity factormethodandtheabsolute intensityquantificationmethod,exceptfor C originating from contamination. It is foundthat the Li concentration obtained by the relativesensitivity factormethod is larger thanthatobtainedby the absolute intensity quantif ication methodbecause theMVVpeaksofMn,CoandNioverlap
withtheLipeak. W hen compar ing the quant itat ive ana lysisresults of XPS (Fig. 16) with AES (Fig. 17), thedifferenceintheCconcentrationisremarkable.Thismay be due to the difference of the analysis areabetween XPS and AES. When the average atomicconcentrations ina largeareaareanalyzed,XPS ismoreadvantageousbecauseXPScananalyzea largeareaand thismethod isnot likely tobe influencedbylocalizedcontamination.Ontheotherhand,AESaims to analyze a local area in principle and theanalysisarea is confined to several100μmsquaresat amaximumdue to the input lens system. In thepresentexperiment,werandomlymeasuredanareawhichwasconsidered tohaveaveragecomposition.But incidentally, an area containing many organic
substanceswasanalyzed, thus the Cconcentrationwas considered to be high. When comparing thenormalizedquantificationvaluesofAESandXPS,which were obta ined by the absolute intensityquanti f ication method and excluded C element,the quantification values of the respective atomicconcentrations in AES are close to those in XPS,and for Li, the difference between AES and XPSis confined to approximately 5%. Consequently,whenquantificationofLi isperformedusingAES,we recommend the use of the absolute intensityquantificationmethodbecause thismethodenableshigheraccuracyquantitativeanalysis. Next, aswe introduced in theSectionof“Pre-treatment technique and precautions for powderspecimenatAugeranalysis”,weshowtheobservationresult of a secondary electron image (SEI) of anNMCspecimensprayedonacarbon-specimenstubinFig. 18. Figure18revealstheexistenceofmanyparticleswithdifferentsizeandalso,foreignparticlesthatareconfirmedtobeorganicsubstances.Forconvenienceof analysis , particles with large diameters weredefinedas“point-1”, thosewith smalldiametersas“point-2”,and foreignparticlesas“point-3”.Augerspectra forpoint-1 topoint-3are shown inFig. 18.
From the results, itwasdiscovered that theatomicconcentration ratioofMn,CoandNiwasdifferentbetween “point-1” and “point-2”, and also the Liconcentrationsamong the twopointsquitediffered.In “point-3”, no Li was detected, therefore, thepowdersinthisareaareconsideredtobeonlyforeignmaterials.Generally,a largevolumeof such foreignparticles are mixed in the Li-ion battery powderspecimen.ThepresentobservationofSEI(secondaryelectron image) and BEI (backscattered electronimage) accounts for the fact that a quantificationvalueofhighCconcentration isobtaineddependingon the analysis area as shown in Fig. 17. Thus, inthecaseof theLi-ionbatterypowderspecimen, theanalysisresultsofXPSwhichaveragelycoverawidearea might be different from that of AES whichperforms local area analysis. Taking this fact intoconsideration, only either XPS or AES does notprovide sufficientdata forquantitativelyanalyzingthespecimenmoreaccurately.Thatis,itisnecessarytoobtainthedatafromtheaverageareaandthedatafromthe localarea for comprehensive judgmentontheanalysisresultsofthespecimen. Afterthisexperiment,wepreparedacrosssectionof an NMC specimen with a JEOL Cross SectionPolisher (CP) and performed analysis (Fig. 19) .
In order to prepare the cross section of a powderincludingLisuchasLi-ionbatterymaterialforAES,severalprecautionsneedtobetaken.Firstly,inordertoprevent thediffusionofLi, theuseof thecarbon-specimenstub isamust to fix thepowders (insteadof theuseof resin).Secondly, inorder topreservethespecimenstate in thepowderspecimen,across-section fabrication device must be used. TypicallyuseddevicesareaCPoran IonSlicer.These toolsmill thespecimenwith inertgas ionswhichdoesnotchemically react with the elements in the powderspecimen.Thirdly, inorder topreventalterationofpropertiesonthecross-sectionedsurface,aTransferVessel isused,whichcanpreserveand transfer thespecimenwithoutexposingthespecimentotheair.Inparticular,sinceanAESperformselementalanalysisand chemical state analysis of the specimen topsurface(onlyafewnanometersdowntothesurface),muchcareisrequired. A secondary electron image (SEI) in Fig. 19reveals that the part icles in the specimen haveporous-like structures.SomeLi-ionbatterypowdermaterialspossess suchporous-like structures in theparticles(showninFig.19)at theprocesswhererawmaterialparticlesaremixedandfired.CPcanmillacrosssectionofparticleeven includingaporous-likestructure.Thus,CP ispowerfulandsuitable for theanalysisofsuchLi-ionbatterypowdermaterials.WeacquiredanAugerspectrumforanareaenclosedwitharedcircleinFig.19andperformedthequantitativeanalysisbasedontheabsoluteintensityquantificationmethod.Asa result, itwas found thatMn,CoandNiexistwitharatioofapproximately1:1:1andwhenconverting the result into theatomic concentrationexcept for C, the resultant concentration becomesapproximately13to14%.Theseconcentrationvaluesareclose to thoseobtainedbyXRF inFig.15, thusthe values are considered to indicate the atomicconcentrationsoftheparticles. Next, as is shown in an SEI in Fig. 20, SEM-EDS analysis was performed for the same areameasuredwithAESshown inFig. 19.TheanalysisresultsareindicatedinthetableinFig.20.Comparedto thequantification resultofAES, it is found thatdifferenceoftheatomicconcentrationsbetweenMn,CoandNiisonly2%;thusindicatingthesamevaluestoeachother. ThequantitativeanalysisresultsaresummarizedinTable 4.Forboththesurfaceanalysisandthebulkanalysis, the results obtained by AES, which arebasedontheabsoluteintensityquantificationmethod,demonstrate highaccuracy in comparisonwith theotheranalysismethods.Thatis,thedifferenceintheatomicconcentrationsofMn,CoandNi isconfinedwithinapproximately2%andalso, theerror in thequantification value of Li is estimated to be a few%.Thus, it shouldbeemphasized that theabsoluteintensityquantificationmethodprovides substantialaccuracy for thequantitativeanalysisofLi,and theuseofAESwillbeapplied tovariouspurposes forinvestigationofLi-ionbatterypowdermaterial,fromlocalareaanalysistotheaverageareaanalysis.
Conclusions
Nowadays, v igorous development of Li- ionbattery is carried out by researchers around theworld and a large number of reports have beenissued. Under this circumstance, knowledge aboutLi behav iors and Li d istr ibut ions with var iouschemical statesata localarea in theLi-ionbatteryis strongly demanded, which are the basics of thedevelopment of their electr ical properties. AESand X PS are only a few methods that can meetthoseneeds.Thosemethodsenableus todetectandquantify Li. Furthermore, commercially availableAES instruments can perform mapping in a nano-area. However especia l ly for AES, there ex istsmisunderstanding that it is difficult to analyze Li,thustheAESinstrumentarenotusedverymuchforsuchpurposes.Astheauthorsintroducedinthisarticle,theanalysisof Li by AES is very effective like the analysis byXPS.These two instrumentscandetectLiwithhighsensitivity. In addition, when using the absoluteintensity quant i f icat ion method, A ES enablesone to obtain reliable quantitative analysis resultscomparable to theotheranalysismethods.Sufficientutil ization of AES and XPS can complement thedataobtainedwith instruments thatcannotdirectlymeasureLi(TEM,SEM,EPMA,XRF,etc.).Bytheuseofoveralldatawithsuchanalyticalmethods, theauthorswillexpectLidistributionsandLibehaviorstoberevealed.
References
[ 1 ] Ts u t s u m i , K . ; A p p l i c a t i o n s o f A E S fo rdevelopmentofLi ionbatteriesandCIGSsolar-cel ls , 20 09 JEOL EPMA / Surface AnalysisUsers’ Meeting document, AP100 (2009) (inJapanese).
[ 4 ]Tsutsumi, K., Shima, M., Tanaka, A., Tazawa,T. ; Quantitative Chemical State Analysis ofTinOxides(Sn,SnO,SnO2)withHigh–EnergyResolutionAES,J. Surf.Sci . Soci . Japan,Vol. 33, No. 8,pp.431-436,(2012)(inJapanese).
[ 5 ]Tsutsumi, K. ; Peak Deconvolution Analysisin Auger Electron Spectroscopy, JEOL news,Vol.37, No. 1,p66(2002).
[ 6 ]Shima,M. ; ImportanceofAverageLargeAreaAnalysis in XPS; oversight problems in microa rea a na lys i s , 2 012 J EOL EPM A / Su r fac eAnalysisUsers’Meetingdocument,AP92(2012)(inJapanese).
73 JEOL News Vol. 49 No. 1 (2014)
Characteristic Features and Applications of a Newly Developed Wavelength Dispersive Soft X-ray Emission Spectrometer for Electron Probe X-ray Microanalyzers and Scanning Electron Microscopes
H. Takahashi*1, T. Murano*2, M. Takakura*2, N. Handa*3, M. Terauchi**,M. Koike***, T. Kawachi***, T. Imazono***, N. Hasegawa***, M. Koeda****,T. Nagano****, H. Sasai****, Y. Oue****, Z. Yonezawa**** and S. Kuramoto****
*1 Global Business Promotion Division, JEOL Ltd.*2 SA Business Unit, JEOL Ltd.*3 Advanced Technology Department, JEOL Ltd.** Institute for Multidisciplinary Research for Advanced Materials,
Tohoku University*** Quantum Beam Science Directorate, Japan Atomic Energy Agency**** Optical Components BU Device Department, SHIMADZU Corp.
1. Introduction
A new wave-length dispersive spectrometer(WDS) to detect ultra-soft X-rays for soft X-rayem is s ion spec t rosc opy (SX E S) ha s f i r s t beendesigned and developed for transmission electron
microscope (T EM) by Terauchi et . a l . [1] Thisspectrometermaybecalledawavelengthdispersivesoft X-ray emission spectrometer (WD-SXES). Itconsistsofanaberration-corrected,concave,varied-l ine-spacing grat ing as a wavelength dispersiveelement, X-ray ref lection mirrors and a charge-coupleddevice(CCD)asadetector.Terauchiandhisgrouphavesuccessfullymadeaseriesofinvestigationon chemica l st ructures of va lence electrons inlightelementsand their compoundswith thisWD-SXES installed toaTEM[2, 3, 4].Encouragedby
2-1-1 Ohtemachi, Chiyoda-ku, Tokyo 100-0004, Japan.
A new wavelength dispersive soft X-ray emission spectrometer (WD-SXES) consisting of a pair of newly designed varied-line-spacing gratings: JS50XL and JS200N, X-ray focusing morrors and a detector of a charge-coupled device has been developed for soft X-ray emission spectroscopy. This WD-SXES, which covers nominally the X-ray energy range between 50 and 210 eV, has successfully been installed for commercial use to electron probe X-ray microanalyzers and scanning electron microscopes. The high energy resolution Spectrum mapping software has been developed using a parallel detection system. The observed area is flexibly applied from micron square to 9 × 9 cm2. The energy resolution of this WD-SXES was evaluated to be 0.2 eV from the Fermi edge of the Al-L spectrum. The Li-K and Li-satellite-K emission spectra due to lithium metal and lithium fluoride were observed for the first time in EPMA. The detection limit of lithium was estimated in a 5 mass% Li-Al alloy to be 40 ppm in mass. The Li-K and Li-satellite-K emission spectra from samples of a lithium ion battery anode were measured and mapped in an area as large as 30 × 80 mm2 under three different charging conditions. A clear distinction in spectra due to chemical states of lithium shows a high potential for the characterization of lithium ion battery anodes. For the trace element analysis of boron, carbon and nitrogen in steel, their linear calibration curves could successfully be obtained in the composition range between 10 and 100 ppm in mass. This method could detect their concentrations as small as 10 ppm in mass. The trace element analysis of light elements in other materials than steel could possibly be performed at this level of concentrations. Effects of compiling spectra obtained with the WD-SXES are described for fulfilling the need for comparing observed spectra with the ones compiled to identify the chemical state.
(2014) Vol. 49 No. 1 JEOL News 74
their success,wehavedevelopedanewWD-SXESinstalled to electron probe X-ray microanalyzers(EPMAs): JXA-8100, JXA-8500F, JXA-8230 andJXA-8530Fand twoscanningelectronmicroscopes(SEMs):JSM-7800FandJSM-7100FasshowninFig. 1.Characteristic featuresanda fewapplicationsof theWD-SXESinstalledtoEPMAsandSEMsarebrieflyreportedinthispaper.
2. Characteristic features
Soft X-ray emission spectroscopy (SXES) hasaninherentdifficultyduetolowefficiencyofthesoftX-rayemissionanddetection.Thisdifficultymet inTEM-SXES can partly be overcome by installingthe WD-SXES to EPMAs, which have a capacityproducing far larger probe currents. In fact, theacquisitionconditionparameter(AQ)definedas thebeam current times the acquisition time is usuallyexpressedasaunitofnA-min forTEMwhereasasaunitofμA-min forEPMA.Oneotherdifferencein the acquisition conditions between TEM- andEPMA-SX ES is the accelerat ing voltage of theelectronprobe.Itwassofarfixedat100kVforTEM-SXES [2] whereas it can be varied at a few kV forEPMA-SXES. One of the intensity measurementsof accelerating voltage dependence of soft X-rayemissionspectraisshowninFig. 2.Inthisfigure,theAL-Lspectrumwasmeasuredwithaprobecurrentof100μAformeasuringtimeof60satsixacceleratingvoltages: 1, 2, 3, 5, 10, 15and20kV.Theoptimumvalue for the Al-L spectrum was experimentallydeter m i ned to be 5 kV. I n t he fol low i ng , t heaccelerating voltages were selected to be a few kVdependingonthespectra. The two varied-line-spacing gratings: JS50XLand JS200N have been developed for all of TEM,EPMA and SEM. Nominally, the former JS50XLcovers the X-ray energy range between 50 and 170eV whereas the latter JS200N between 70 and 210eV.Experimentally theenergyrangeofJS50XLcan
beextendeddown to46eV.Thebothgratingshavebeen set in one channel as a pair in EPMAs andSEMsso that theWD-SXESforEPMAsandSEMscovers the X-ray energy range between 50 and 210eV. In this energy range, spectra such as Li-K (54eV),Al-L(72eV),Si-L(92eV),P-L(119eV)andB-K(182eV)are located.Becausethevaried-line-spacinggrating isusedasawavelengthdispersiveelement,the X-rays with higher energy than the one in therangepreviouslydescribedcanalsobedetectedasthehigherorderspectraso that theenergyrangeupto700eVcanpracticallycovered, In thecaseof theKemission,C(277eV),N(392eV),O(525eV)andF(677eV)canbewelldetected.LemissionspectraofTi,V,Cr,MnandFe,andMemissionspectraofZr,Nb,Mo,Ag,Cd,SnandSb,canbealsoobserved.NemissionspectraofHf,Ta,W,Re,Pt,Au,Bi,a fewLanthanidessuchasU,cansuccessfullybeobservedquiterecently. One other characteristic feature of the WD-SXESisrobustnessofthesystemandeasinessofdataacquisition.Thevaried-line-spacinggratingandCCDdetector are both set rigidly without any movablepar t , resu lt ing in an excel lent reproduc ibi l ity.Furthermore, spectra within the covered energyrange by the wavelength dispersive elements canbemeasuredandacquired simultaneously, just likeordinaryenergydispersivespectroscopy.Thisparalleldetectionof spectraenables toacquirea spectrummap fromwhichelementalmapsandchemical statemapscanbedrawn. T he s i z e of a rea of a na ly s i s ra nge s f romnanometer to micrometer in diameter in TEM. Itbecomes three orders of magnitude larger fromm icrometer to m i l l imeter in EPM A and SEM.When the scanning stage is used in EPMA, thearea as larger as 90 × 90 mm 2 can be analyzed.T he i nsta l l ment of W D - SX E S to EPM A s a ndSEMs makes SXES serve not only as a method offundamental researchon the chemical structureofvalence electrons in elements and compounds butalsoasamethodofthecharacterizationoffunctional
3-1. Emiss ion Spec tra of L i th ium and Lithium Compounds
3-1-1. Li-K emission spectrum of lithium metal and Li -satelli te - K peak obtained from lithium fluoride
Figure 3 (a) shows a Li-K emission spectrumobtained from lithium metal. This spectrum wasacquiredunderthefollowingcondition;fivedifferentpositionsweremeasuredwithaprobecurrentof0.1μAatanacceleratingvoltageof15kV;themeasuringtimeatonepositionwas30ssothatthetotaltimeofthemeasurementforthespectrumwas150s;AQwas0.25μA-min.Theextensionof thedetection rangedown to 50 eV made the Li-K emission spectrumfullyobservedwithitspeakaround54eV.ThisLi-Kemissionspectrumdueto lithiummetalwasthefirstoneobservedinEPMA. Figure 3 (b) shows a spectrum obtained fromlithium f luoride in the same energy range as theone inFig.3(a).Atpresent,wehavenot succeededtoobserveagenuineLi-Kspectrumdue to lithiumfluoride.Noappreciableintensitypeakwasobservedin the energy range shown in Fig. 3 (b) under acondition of a broad beam with a size of 30 μm indiameter even after 30 s. But once the beam was
focused, thespectrumshown inFig.3 (b) started tobeobserved.Itsacquisitionconditionwasthesameastheone inFig.3 (a),except theacceleratingvoltagedecreasedto5kV.InadditiontothespectrumduetotheLi-Kemissionshown inFig.3(a), it showsaLi-satellite-Kpeakwasusedtomonitorchemicalstateoflithiuminalithiumionbatteryanodeinthefollowingsection.
3-1-2. Li-K and Al-L emission spectra of a 5mass% Li-Al alloy
(Energy resolution and detection limit)
Figure 4 showsa softX-rayemission spectrumofa5mass%Li-Alalloyinanenergyrangebetween47and100eV.Thespectrumwasacquiredunderthefollowingcondition,whichisalsoshowninthefigure;fivedifferentpositionsweremeasuredwithaprobecurrent of 2 μA at an accelerating voltage of 5kV;themeasuring timeatonepositionwas60 s so thatthe total timeof themeasurement for the spectrumwas300s.Underthiscondition,AQwas10μA-min.At a higher energy range around 70 eV, the Al-Lemissionspectrumwasbeautifullyobservedandatalowerenergyrangearound54eV, theLi-Kemissionspectrumwasalsoobserved.Becausethepeakofthelatter spectrumwas small, anenlarged spectrum isshownat thebottom leftof in the figure.TheAl-Lemissionspectrumshowsasharpedgecorrespondingto theFermi-edge (EF)at thehighestenergyof thepeak.Theenergyresolutionwasevaluated tobe0.2eVfromthisportionofthespectrum.Theleftsideoftheprofile fromtheFermi-edgecorresponds to thedensityof stateofvalenceband.Fromthespectrum
intensity of the Li-K emission, the detection limit(CDL)oflithiumwasestimatedtobe40ppminmassbasedonthefollowingequation:
wherePandBarepeakandbackground intensities(counts/s), t is time(s)andC is theconcentrationoftheelement[5].
3-1-3. Change of chemical states of lithium due to charging in a Lithium-ion battery anode
Three samples A, B and C of a l ithium-ionbattery anode with different charging conditions:0% (A). 30% (B) and 100% (C), were supplied assealed separately in threeglovebags filledwithArgas. Theywere transferred toa sampleholder in aglove-bag typeofwrapper filledwithArgas,whichcovered both the sample holder and the insertionchamberoftheEPMAasshowninFig. 5(a).Thoughcumbersome, thisprecautionwasnecessary toavoidchemical reactions of the sample surface in theordinary atmosphere. The three samples, each ofwhichhada lengthof50mmandawidthof30mm,weremounted to the sampleholder sideby sideasshowninFig.5(b).Alargeareawithasizeof16×50
mm2was selected formappingof chemical stateoflithium.TheareaconsistedofA,BanbCwasdividedinto50×10pixelssothattheinterpixeldistancewas1.0 mm in the length direction and 1.6 mm in thewidthdirection.TheintensitiesofsoftX-raysemittedbyanelectronprobewithadiameterof1μmwitha current of 0.8 μA at an accelerating voltage of 2kV was acquired for 60 s /pixel in an energy rangebetween48and165eV.ThreemapsconstructedbyusingtheX-raysinthreeenergyrangescorrespondingtoLi-satellite-K(49-51eV),Li-K(53-55.5eV)andC-Kofthe4thorder(nominally66-70eV)areshowninFig. 6.Themapdue to theLi-satellite-Kon thelefthandsideapparentlycorrespondedtotheamountof charging; the intensityofLi-satellite-Kemissionincreasedwiththeincreaseintheamountofcharging.Themap in themiddle in Fig. 6 indicated that themetal l ic l ith ium formed local ly and distr ibutedheterogeneouslyunder the fully chargedcondition.The map due to C-K of the 4th order on the righthandsideinFig.6indicatedtheintensitydistributiondue tographite;oneother importantconstituentoftheanode.Under thefullydischargedcondition, theintensitywasstrongest.Whentheamountofchargingincreased by 30%, it decreased substantially. Butit stayed the same level,even though theamountofchargingincreasedto100%. Threetypicalspectracorrespondingtothethreedifferent conditions were shown in Fig. 7. Figures
7(a)and7(b) show the spectra inoneenergy rangebetween 48.5 and 59.5 eV and in the other energybetween 62 and 74.5 eV, respectively. The formerenergy range covered the Li-satellite-K and Li-Kemission with overlapped C-K of the 5th order,whereas the latterC-Kof the4thorder.Theywerevisually selected fromthe typicalpixelsobserved inthethreemapsshowninFig.6.AsshowninFig.7(a),theLi-satellite-KandtheLi-Kwerenotobserved inthefullydischargedanode,whereastheLi-satellite-Kwasobserved inthechargedanodesataround50eVandtheLi-Kwasclearlyobservedinthefullychargedanode.ThechangeintheintensityoftheC-Kthe4thorder shown inFig.7(b)was the sameasexplainedinFig.6.One thing tobenoticedwas theobservedshoulder around 69.5 eV due to the� bond in theprofileofthefullychargedanode.
3-2. Trace element analysis of nitrogen in steel
The WD-SXES has a very high potential foranalyzing trace elements in steel. Sl ightly extraendeavorandprecautionextend the rangeof traceelementanalysis to the lowerconcentrations. In theordinaryEPMA,thecalibrationmethodisusuallyusedin lowerconcentrationrangeswhichneedaseriesofreliablestandardspecimens.Thestandarddeviationsofcalibrationcurvesdeterminetheconcentrationlimitaswellastheaccuracyofmeasuredconcentrations.WiththeWD-SXES, thebetter intensity ratioofpeak tobackgroundcanbeobtained.Furthermore,thebetterenergyresolutionoftheWD-SXESproducesagenuinetarget spectrum well separated from overlappedspectraoftentakenplace,andwelldefinedbackgroundintensitiesnotfarawayfromthespectrumsothattheintensityofthetargetspectraoftheelementofinterestveryclosetothegenuineonecanbeobtained. Aspreliminaryexperiments, calibrationcurvesofboron,carbonandnitrogen insteelsweretriedtobe obtained within their each concentration rangebetween10and100ppminmass.Thesignalofeach
characteristicX-rays from the traceelements;B-K,C-K of the 2nd order and N-K of the 2nd order,could beunambiguouslyobserved.The calibrationcurves for the three trace elements were l inear.Thesestraight lines indicated that theconcentrationrange of 10 ppm in mass could be detected withthismethod.Theresultsof thepresentexperimentsstrongly suggest that the similar method can alsobeused for theelementalanalysisof traceelementanalysisinsemiconductors.
3-3. Compilation of spectra
Oneof themainpurposes for thedevelopmentof the WD-SXES was to extend the l imit of thelower energy end of the wavelength d ispersiveX-ray emission spectroscopy in EPMA so that inthe early stage of the development attention wasmainlyfocusedonthelowerenergyrangeofspectra.Oneof the successful results is thedetectionof theLi-K spectrum and its application to monitor thechemical state of lithium in a lithium ion batteryanodedescribed in theprevioussection.Becauseofthehigherenergy resolutionof theWD-SXES, theshapesof the spectrawell reflect thedetailed stateof thevalenceelectrons.Naïveapproach touse thespectrumshapesfortheidentificationofthechemicalstate of the element concerned is to compare theobserved shapeswith theonespreviouslyobtainedfor the standard specimens. Tentative compilationhas been published as an appendix in the leaf letprepared for the first JEOLseminaronSoftX-rayEmissionSpectroscopy [6]. It containsanumberofspectraofelementssuchasLi,Be,B,C,N,O,F,Mg,Al,Si,PandSinthestatesofeitherpureelementsorcompounds. We are now tr y ing to add more spectra inthe compilation. One of the efforts is to measureand compile spectra of the standard specimens forEPMA. There are 32 pure elements commerciallyavailableasthestandardspecimens.Theirspectraintheenergy rangebetween50and210eVhavebeen
79 JEOL News Vol. 49 No. 1 (2014)
measured using both dispersion elements; JS50XLandJS200N in the followingcondition;acceleratingvoltage:5kV,probecurrent:100nA,andacquiringtime: 5 × 60 s. One of the examples of the spectraobtained in this series of the systematic effort isshown inFig. 8.EachspectrumofmetalPtandAuin the sixth row of periodic table of the elementsshowonedistinctpeakconsistingof two sub-peaksintheenergyrangebetween120and150eVtogetherwith a few additional peaks in the lower energyrange.As faras thepresentauthorsareconcerned,thesepeakshavenotexplicitlybeenreported in theliterature, especially in the energy range shown inthe figure.Basedon the tableonX-raywavelengths[7] , the peaks in those spectra were assigned asshown in the figure. Roughly speaking, the shapesof two spectra look rather similar, except the shiftof the peak position to higher energy by about 50eV with increasing the atomic number from Pt toAu.Thesimilarsystematicenergyshiftof thepeakshasbeenobserved inspectraofa seriesof thesixthrow elements; starting from Hf, continuing Ta, W,Re,Os, Ir,PtuntilAu.Although the shapesof thepeaksofPtandAuaresimilar,slightdifferencemay
benoticed.Many features in these seriesof spectraincluding thepeaksdescribedaboveareawaited tobeexplained.
4. Conclusion
A newly developed wavelength dispersive softX-rayemission spectrometer (WD-SXES)with twokindsofgratings(JS50XLandJS200N)wereinstalledto EPMAs. It covered the energy range between50and210eV for commercialuse.Experimentallytheenergy rangecouldbeextendeddown to46eV.Theenergyresolutionof thisWD-SXESwas0.2eVwhichwasevaluatedfromtheFermiedgeoftheAl-Lspectrum. TheLi-KandLi-satellite-Kemissionspectradueto lithiummetaland lithiumfluoridewereobservedfor the first time inEPMA.Thedetection limitoflithiumwasestimatedina5mass%Li-Alalloytobe40ppminmass.TheLi-KandLi-satellite-Kemissionspectra fromsamplesofa lithium ionbatteryanodeweremeasuredandmappedinanareaas largeas16×50mm2underdifferentchargingconditions.Aclear
distinctioninspectraduetochemicalstatesoflithiumshows a high potential for the characterization oflithiumionbatteryanodes. For the traceelementanalysisofboron,carbonandnitrogen in steel, their calibrationcurvescouldsuccessfully be obtained in the composition rangebetween10and100ppminmass.Thismethodcoulddetect their concentrations as small as 10 ppm inmass. The trace element analysis of light elementsin other mater ia ls than steel could possibly beperformedatthislevelofconcentrations. EffortsofcompilingspectraobtainedwithWD-SXESareunderway.The tentativecompilationwaspublished as an appendix in the leaf let preparedfor the f i rst J EOL Seminar on X-ray EmissionSpectroscopy. During the present undertaking,quitea fewobservedpeaksofheavyelementshavebeen found that theyhavenotbeenreported in thel iterature. It is emphasized that many aspects ofspectraawaitexplanation.
Acknowledgments
T h i s wo rk i s s u p p o r t e d a s a p ro g r a m ofCollaborative Development of Innovative Seeds(Practicabilityverification stage)by JapanScienceandTechnologyAgency.
References
[ 1 ] M. Terauchi, H. Yamamoto and M. Tanaka,Journal of Electron Microscopy,50,101(2001).
[ 2 ]M. Terauchi, M. Koike, K. Fukushima and A.KimuraJournal of Electron Microscopy,59,251(2010).
[ 3 ]M . Terauch i , H . Ta ka ha sh i , N. Ha nda , T.Murano, M. Koike, T. Kawachi, T. Imazono,M. Koeda, T. Nagano, H. Sasai, Y. Oue, Z .YonezawaandS.Kuramoto,Journal of Electron Microscopy,61,1(2012).
[ 4 ]M . Terauch i , H . Ta ka ha sh i , N. Ha nda , T.Murano, M. Koike, T. Kawachi, T. Imazono,M. Koeda, T. Nagano, H. Sasai, Y. Oue, Z .YonezawaandS.Kuramoto,JEOL News,Vol.47, No. 1,23(2012).
[ 5 ]J. I.Goldstein,D.E.Newbury,P.Echlin,D.C.Joy,A.D.Romig,Jr.,C.E.Lyman,C.FioriandE.Lifshin,Scanning Electron Microscopy and X-ray Microanalysis2nded.PlenumPress,NewYork,p.499(1992).
[ 6 ]H. Takahashi, T. Murano, N. Handa and H.Ishikawa, Developavelength Dispersive SoftX-ray Emission Spectrometer for Section 1Microscopes,paperpresented inJEOLSeminaronSoftX-rayEmissionSpectroscopy(2013).
[ 7 ]J. A. Bearden, Review of Modern of Modern Physics,39,78(1967).
Analysis of Organic Thin Films by the Laser Desorption/Ionization Method Using the JMS-S3000 “SpiralTOF”
Takaya Satoh
MS Business Unit, JEOL Ltd.
Laser Desorption/ Ionization-Time of Flight Mass Spectrometry (LDI-TOFMS) is generally used for analysis of organic compounds because this technique generates little fragmentation of molecular ions at ionization. It makes possible to obtain information on molecular weights and molecular structures in organic compounds. In particular, a technique which uses the matrix compounds for enhancing ionization efficiency is well known as Matrix-Assisted Laser Desorption/Ionization-Time of Flight Mass Spectrometry (MALDI-TOFMS). This technique is widely used in the bio markets owing to its capability of ionizing proteins and peptides with the molecular weights of several thousands to several hundreds of thousands. The MALDI-TOFMS is also utilized for analysis of synthetic polymers. In many cases, LDI-TOFMS and MALDI-TOFMS have been used to estimate the molecular weights of organic compounds in solution. But very recently, techniques of imaging mass spectrometry, which controls the laser irradiation position by two-dimensional scan to acquire mass spectra for visualizing localization of chemical compounds with specific molecular weights, have been improved. The application of this innovative technique is increasingly spreading in the bio markets. The technology of Imaging Mass Spectrometry has been advancing for analyzing biological tissue sections, but in the future, it is expected to develop toward the material science markets. It is noted that various surface analytical techniques are already available in the material science markets. In order to study the advantages of LDI-TOFMS as one of effective surface analysis tools, it is essential to consider the complementary analysis of LDI-TOFMS with the existing surface analytical techniques. In this article, the advantages of using LDI-TOFMS for analyzing organic light-emitting diode material thin films, in accordance with comparison with Time-of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS), X-ray Photoelectron Spectroscopy (XPS) and Scanning Electron Microscopy/Energy-Dispersive X-Ray Spectroscopy (SEM/EDS), have been studied. In addition, since LDI-TOFMS is a destructive analytical technique, the influence on the sample surface caused by LDI-TOFMS was also examined.
Introduction
The surface analytical techniques irradiatean electron beam, an ion beam or X-ray on thesu r fac e of t he sa mple for i nvest igat ion of i t smorphology and physical characteristics based onthe interactions between the beam and substancesex ist ing on the sample sur face. To observe thesample morphology, an optical microscope and anelectronmicroscopearemainlyused.To study thesample characteristics, a wide range of techniquesis avai lable depending on the incident particles(beam)andthesignalstobedetected.TheyincludeElectron Probe Microanalysis (EPM A), AugerElectronSpectroscopy (AES),X-rayPhotoelectron
Spectroscopy (XPS)andTime-of-FlightSecondaryIon Mass Spectrometry (TOF-SIMS) . In recentyears,electronicdevicesarefrequentlycomposedoforganic compounds suchasorganic semiconductor,organic light-emitting diode (OLED) and organicfilmsolarcell,andtheuseofthemwillbeexpectedto further expand. It is increasingly important toinspect organic-compounds and their degradationm e c h a n i s m i n t h e p ro d u c t s . A m o n g s u r f a c eanalytical techniques,AESandXPSarecapableofobtainingchemicalbondingstatesorinformationonfunctionalgroupsinchemicalcompounds,butthosetechniques have a difficulty in structural analysisof organic compounds. The TOF-SIMS is a massspectrometry technique well known as a surfaceanalytical technique. By using the dynamic SIMS,fragmentationofthemolecularionsislikelytooccurat ionization,thusmakingitdifficulttoapplySIMStoanalyzeorganiccompounds.Recently,techniqueswhich utilize metallic clusters or gas clusters as a
3-1-2 Musashino, Akishima, Tokyo, 196-8558, Japan.
primaryionbeamattachedtoTOF-SIMShavebeensucceeded to ionize more softly. These techniquesareexpected toexpand theTOF-SIMSapplicationsfororganiccompounds. T h is a r t ic le repor ts on Laser Desor pt ion /Ionization-TimeofFlightMassSpectrometry(LDI-TOFMS). As a technique that uti l izes the laserdesor pt ion mecha n i sm, Matr i x-A ss i sted L DI-TOFMS (MALDI-TOFMS) is in widespread use,which enables ionization of a variety of chemicalcompounds by properly combining a sample withmatrix compounds enhancing ionization. Around2000year,thenumberofinstalledMALDI-TOFMShasbeendramatically increasedaiming toanalyzeproteins and peptides. Moreover in the materialanalysis fields, MALDI-TOFMS has been utilizedfor the analysis of synthetic polymers. The use ofthematrix isessentialforthemeasurementoftheselarge-molecularweightorganiccompounds,thustheionization techniqueusing laser is generally called“MALDI”.But,therearemanychemicalcompoundswhichcanbe ionizedonlywith laser irradiation. Inthis case, the used ionization technique is simplycalled“LDI”. Mostofmass spectrometry techniquesanalyzesamplesinsolution.MALDI-TOFMSalsomixesthesample solventand thematrix solvent tocrystallizethem by dropping on a target plate. By irradiatingthe co-crystal of matrix and sample compoundsw it h t he u lt rav iolet , M A L DI-TOF M S ion i zesvariousorganiccompoundscontainedinthesampleand performs mass separation. In recent years,imaging mass spectrometry that adopts MALDI-TOFMS[1,2],whichcanacquire informationaboutlocalizationonthesamplesurface, isdemonstratingtechnologicalimprovementandtherefore,theuseofthisunique technique is increasingly spreading. InImaging Mass Spectrometry, the matrix is sprayeduniformlyontothesamplesurfaceandmassspectraare acquired while the laser irradiation positionto the sample is two-dimensionally scanned. Thisprocess allows acquisition of information on two-d imensiona l d ist r ibut ions of spec i f ic chemica lcompounds. Imaging Mass Spectrometry has beenexpanding its applications to thebiomarkets fromitsdawn, includingproteins,peptides, lipids,drugsand theirmetabolites.Mostof the subjects for thistechnique are biological t issue sect ion. On theother hand, in accordance with the establishmentofbiological tissue sectioning techniques, theusersgradually start to study theapplicationof ImagingMassSpectrometrytothematerialanalysismarketsand also, this technique is expected to visualizeinformation on localization of organic compoundsona thin filmora solid surface. In order tomakeImaging Mass Spectrometry more effective in thematerial analysis markets, it is very important tocarryout complementaryanalysiswith theexistingsurface analytical techniques. In this article, thefundamental exper iments using the JMS-S30 0 0“SpiralTOF” to examine LDI-TOFMS as one ofsurface analytical techniques were reported. Thecomparisonwith informationobtainedbyXPSandTOF-SIMS and the inf luence of laser irradiationonto the surface of the organic thin film made ofOLEDmaterialwereexamined.
Sample
For c omplement a r y a na lys i s a mong L DI-T O F M S , T O F - S I M S a n d X P S , N , N ' - D i(1-naphthyl) -N, N' -d iphenylbenzidine (α-NPD),which is a material for a hole transport layer ofan OLED, was deposited onto a Si substrate with60 0 nm thick (hereinafter, cal led “α-NPD/ Si”) .In addition, in order to examine the inf luence ofLDI on the sample surface, the author prepareda different sample of another Si substrate wherea material for a hole transport layer of an organicE L ( 4 , 4 ’, 4 ’ ’ -Tr i s [ 2 - n a p ht hy l ( p h e ny l ) a m i n o ]t r iphenyla m i ne (2 -T NATA) of a t h ick nes s of70 0 nm) was deposited onto the substrate andfurthermore,α-NPDofa thicknessof1300nmwasdeposited onto the prepared layer (hereinafter,called“α-NPD/2-TNATA/Si”).
Analyses of Organic Thin Film Using LDI-TOFMS, TOF-SIMS and XPS
T he J MS - S3 0 0 0 “Spi ra lTOF ” was used asan LDI-TOFMS. Figure 1(a) shows the externalview of the SpiralTOF. The biggest feature of theSpiralTOFisadoptingaJEOLoriginally-developedspiral ion trajectory (Fig. 1(b)) and this trajectoryis formedby fourhierarchicalelectrostatic sectors.The f l ight distance of 1 cycle is 2 .093 m and theSpiralTOFachievesaneffectiveflightdistanceof17mat8 cycles.Here, themass resolutionofTOFMSis proportional to the f light distance. The generaleffective f light distance of the reflectron TOFMSis approximately a few meters, but the SpiralTOFwhich has an effective f light distance of 17 m canachieve the world-highest mass resolution amongMALDI-TOFMSs. Furthermore, the electrostaticsectorswhich forms the spiral ion trajectorymakesit possible to eliminate the fragment ions duringtheir f light, thus a mass spectrum with little noisecanbeacquired.Byattaching theTOF-TOFoption[4], it ispossibletoperformstructuralanalysiswiththeTandemMassSpectrometer(MS/MS).Thehighenergy CID (collision-induced dissociation) couldprovide much structural information rather thanlowenergyCIDused inmajorMS/MS instruments.TheSpiralTOFisequippedwithaNewportNd:YLF(349 nm) as an ionization laser source. The laseri rradiation diameter onto the sample surface isapproximately 20 μm and the laser intensity is 60μJ at 100 % laser setting.α-NPD and 2-TNATAare ionized without requiring the matrix, so theexperimentswereperformedbyacquiring themassspectra using LDI-TOFMS. Figure 2 (a) shows amass spectrum (m/z 10 to 800) acquired by fixingthe laser irradiationpositionon theα-NPD/Si andby accumulations of 250 t imes. Only molecularions ofα-NPD are observed in the mass spectrumand it is found that ion fragmentation is very littleat the ion i zat ion. Using the TOF-TOF opt ion,the author acquired a product ion spectrum byselecting the observed molecular ions. Fig. 2 (b)showstheobservationresultofthecreatedfragmentions and the est imated f ragmentat ion posit ionof fragmentation. By the use of the High-Energy
83 JEOL News Vol. 49 No. 1 (2014)
CID technique, sufficiently much information wasobtainedtoestimatethemolecularstructure. TheArgasclusterionbeamsourceattachedtoJEOLJMS-T100LP“AccuTOFLC-plus”developedin MatsuoGroupatKyotoUniversity [5]wasusedfor TOF-SIMS experiments. Figure 3 (a) shows itsexternalview.Fig.3(b)showsamassspectrum(m/z0to800)whichwasacquiredwiththeprimaryionbeamofArcluster ions (acceleratingvoltage:10kV) thatirradiates on the “α-NPD/Si”. The molecular ionpeakofα-NPD([M]+•)wasobserved.However, thefragment ions were also observed with noticeableabundance in low mass range, (m/z 100 to 500),compared to LDI-TOF MS. This may be due totwo reasons. One is the fragment ions generatedfromα-NPD at the ionization. It was consideredreasonable because the pattern of the production spectrum in Fig. 2 (b) is relatively similar tothe mass spectrum acquired with TOF-SIMS. Onthe other hand, the measurement region in depthdirectionbyTOF-SIMSisconfinedtoonly10nmorless from the top surfaceof the sample, thusmanychemicalbackgroundpeaksproduced fromsurfacecontamination.SinceArclusterionsareusedfortheprimary ionbeam, themass spectrumachieves thelittlest fragmentationamongTOF-SIMSs.Howevercompared to LDI-TOFMS, it should be taken intoconsideration the inf luence of the fragmentationor a rem a rkable i n f luenc e of sa mple - su r fac econtamination on the mass spectrum. TOF-SIMSmakes itpossible toperformhighspatial resolution
m a p p i n g a n d d e pt h p ro f i l i n g by m o n i t o r i n gmolecular ionsormajorfragmentions.Forexampleduring the mapping, a spatial resolution of 1 μmor less is achieved, indicating that this resolutionperformanceishigherthanthatobtainedbyImagingMass Spectrometry using the present M A LDI-TOFMS(typicallya few tensofmicrometers).But,whentakingaccountofthefactthatmanyfragmentions and the background or ig inat ing f rom thesurfacecontaminationareobserved, this techniqueisapplicableonly to the ionsofmajorcomponents.The chemical compounds deriving from the majorcomponents in degradations expected to be minorcomponents ; therefore, distinction with fragmentions or with surface contamination may becomedifficult. T h e J E O L J P S - 9 0 1 0 w a s u s e d f o r X P Sexperiments.Figure 4(a) shows itsexternalviewoftheJPS-9010.Theanalysisareawasset tobe1mmdiameter. Fig. 4(b) and (c) show the measurementresu lt of α -N PD / Si . A spectrum shown in Fig.4(b) is a wide spectrum (energy resolution: 1.7 eVequivalent to Ag3d5/ 2) and the peaks of C and Nwhichareconstituentelementsofα-NPDareclearlyobserved.InthespectrumobtainedbyXPS,whichisatop-surfaceanalysis instrument likeTOF-SIMS,aSipeakoriginatingfromasubstrateisnotobserved.Furthermore,anarrowspectrum(energyresolution:0.5eVequivalenttoAg3d5/2)wasacquiredfromthevicinity of the C peak. It was able to understandt he p e a k s i nc lud i ng t he i n for m at ion on C - C
bonding and the C-N bonding. As compared tomass spectrometry techniques (LDI-TOFMS,TOF-SIMS,etc.),XPSprovidesnon-destructiveanalysisandalsocanperformquantitativeanalysiswhich isdifficult inmassspectrometry causedby ionizationuncertainty.However,whenthesampleisanorganiccompound formed by a combinat ion of l imitedelements, it is not easy to quantitatively analyzemixturesinthecompoundswithanXPSinstrument.Inparticular, it is estimated that,when the limitedelements are mixed as minor components wherethe compositionofadegradationproductdoesnot
changelargely,theseparationoftheirspectralpeaksbecomesmoredifficult. As described above, the chemical informationf rom L DI -T OF M S w it h b ot h T OF - SI M S a ndX PS, which are the ex ist ing sur face analy t ica ltechniques, are compared.TheadvantagesofLDI-TOFMS in the analysis of organic compounds arethefollowings.LDI-TOFMSenablesonetoconfirmmainlymolecular ions fromthemass spectrumandalso,makesitpossibletoperformstructuralanalysisthrough MS/MS analysis. These powerful featuresplay a significant role especially in the analysis of
(a)
(b)
[M]+
85 JEOL News Vol. 49 No. 1 (2014)
organicmixtureswhichexistonthesamplesurface.Also in degradat ion analysis , th is technique isexpectedtoallowtheanalysisofaminorcomponentwhich is a degradation product created from themajorcomponent.
Influence of Laser Irradiation on the Sample Surface
was conf irmed. Figure 5 (a) shows the externalview of the instrument used for this experiment,JEOL SEM JSM-7001FTTLLV equipped with theOX FORD Instruments AZtec Energy StandardX-Max50. Fig. 5 (b) shows an SEM image of ani r rad iat ion sc a r a f ter the sa mple su r fac e wa sirradiatedwithalaserbeamundertheconditionsoflaser intensity40%and thenumberof laser shotof250.From thisSEM image, theablationoforganicthin-filmlayerswasobservedatadiameterof35μmin the scar after the laser irradiation. In addition,Fig.5(c)and(d)respectivelyshowtheanalysisresultof EDS spectra acquired from the irradiation scar
and fromanareaonwhichorganic thin-film layersexistneartheirradiationscar.ThepeaksofSiandCwereobservedfromtheformerandlatterspectrum,respectively, indicating that the laser irradiationallowspenetrationoftheorganiclayersintothescartobeconfirmed. Mass spectra of α -N PD / 2-T NATA / Si wereacquired under the conditions of laser irradiationposition fixedand laser intensity40%.Figure 6(a)shows the plot diagram of ion intensity variationsof molecular ions ofα-NPD and 2-TNATA withre spe c t to t he nu mber of la ser shot . T he ionintensity ofα-NPD on the upper layers decreasedas thenumberof laser shot increases.On theotherhand, 2-TNATA on the lower layers started to beobserved in themass spectrumwhen thenumberoflasershotreached100.Fig.6(b)and(c)respectivelyshow the accumulation mass spectrum acquiredwith the number of laser shot of 0 to 50 and 100to150are shown inFig. 6(b)and (c), respectively.The fragment ions are hardly observed in bothspectrum and the 2-TNATA is clearly appeared inonlyFig.6(c).However,α-NPDontheupper layerswasstillobservedevenafter2-TNATAonthelowerlayers started to be observed. , It is expected thatasthenumberoflasershotincreases,theionizationregion spreads in the plane direction as well as inthedepthdirection.Thevariationof thenumberoflaser shot for the appearance of 2-TNATA in themass spectrum according to the laser intensity isshowninFig. 7.Itisfoundthat,asthelaserintensityincreases,2-TNATAappearsevenwhenthenumberof laser shot isdecreased.This result indicates that
the influenceofdepth isaffectedby thenumberoflasershotandthelaserintensity. From these results, the author found that theinfluenceof laser irradiationon the sample surfacechanges greatly depending on the laser irradiationcond it ions ( la ser i ntensity a nd the nu mber oflaser shot) . For the depth direction, the presentex per i ment s i nd ic ate t hat t he c omprehen s iveinformation on regions between 100 nm and 1 μmisobtained.Compared to the measurement resultsobtainedby top-surfaceanalytical techniques suchas XPS and TOF-SIMS, the present depth regionsareconsiderablylarge.Whenincreasingthenumberof laser shot and the laser intensity, the ionizationregion increases fornotonly in thedepthdirectionbutalsointheplanedirection,thuscareisrequiredformapping.
Summary
T h i s a r t ic le repor ted on c ompa r i son a ndexaminationoforganic thin-filmanalysis forLDI-TOFMS,TOF-SIMSandXPS.TheXPSandTOF-SIMShaveadifficultyinapplyingthetechniquestomulti-component samples.This isbecauseXPScanobtain information only on elements and chemicalbond i ng states , a nd TOF- SI MS ma kes a massspectrum complicated caused by fragment ions.To the contrary, LDI-TOFMS can mainly observemolecular ions, thus it is suitable for the analysisof multi-components. In degradation analysis oforganic chemical compounds inelectronic parts, it
Fig.7 The number of laser shot at which 2-TNATA starts to be observed when the laser intensity is changed. As the laser intensity increases, the influence of laser irradiation on the sample surface becomes large.
(2014) Vol. 49 No. 1 JEOL News 88
isexpected that the totalelementcompositionratiodoes not change largely, thus LDI-TOFMS can beusedasanimportanttoolforidentifyingdegradationcomponents because this technique enables one toconfirm molecular ions and to perform structuralanalysisbyMS/MS.Inaddition,itmaybeconsideredthat the amount of the degradation product is notso largecompared tooriginal compound; therefore,the use of LDI-TOFMS allows one to expect clearanalysis because LDI-TOFMS produces almost nofragmentionsattheionization. Fu r ther more, the SEM obser vat ion resu ltwhich revealed the sample-sur face states a f terlaser irradiation clarified that the comprehensiveinformation of 10 0 nm or more was obtained inthedepthdirectionofa thin film inanorganicELmaterial.Influenceof laserirradiationonthedepthdirection depends on the laser intensity and thenumber of laser shot. When LDI-TOFMS is used,the information of depth direction is considerablylarger than that obtained by XPS and TOF-SIMS,inwhich the typicalanalysisdepth is10nmor less.In the analysis of thin films having structures inthe depth direction using XPS or TOF-SIMS, itis often combined with ion etching because theyare top-surface analytical technique. In this case,it is possible to perform depth profiling with highresolution to depth direction. On the other hand,whenLDI-TOFMS isused, clearacquisitionof theinformationinthedepthdirectionisratherdifficultcompared to XPS and TOF-SIMS, but it may beconsidered thatLDI-TOFMScanclassify chemicalcompoundscontainedinthesamethin-filmlayer. Now, the use of the (M A) L DI-TOF MS i smaking it possible to acquire two-dimensionald i s t r ibut ion s o f c hem ic a l c omp ou nd s on t hespecimen surface in accordance with the progressof Mass Imaging technologies. In the future, byaccumulating the knowledge about ionization ofsamples of thin f i lms and the inf luence of laser
irradiation on the sample surface, LDI-TOFMSwill widely be applied as one of powerful surfaceanalyticaltechniques.
Acknowledgments
We wou ld l i ke to ack nowledge A s so c iateProfessor J. Matsuo and his group at QuantumScience and Engineering Center, Kyoto Universityfor providing their organic thin-films samples andTOF-SIMSmassspectra.
References
[1] Caprioli,R.M.,Farmer,T.B.,Gile,J.:Molecularimaging of biological samples: localization ofpeptidesandproteinsusingMALDI-TOFMS.Anal. Chem.69,4751–4760(1997).
[2] Jungmann, J.H., Heeren, R.M.A. : Emergingtechnologies in mass spectrometry imaging. J. Proteomics75,5077–5092(2012).
[3] T. Satoh, T. Sato, J. Tamura, “Developmentof a h igh-Per for mance M A L DI-TOF massspectrometerutilizinga spiral ion trajectory”:J. Am. Soc. Mass Spectrom ., 18 , 1318 –1323,(2007).
[4] T.Satoh,TSato,A.Kubo,J.Tamura,“TandemTime-of-Flight Mass Spectrometer with HighPrecursor Ion Selectiv ity Employing SpiralIon Trajectory and Improved Offset ParabolicReflectron”: J. Am. Soc. Mass Spectrom., 22 ,797-803,(2011).
[5] K . Ich i k i , J . Ta mu ra , T. Sek i , T. Aok i , J.M at suo, “Development of ga s c lu s ter ionbeam irradiation system with an orthogonalaccelerat ion TOF instrument” Surface and Interface Analysis,45(1),522-524(2013).
Table1 Comparison of LDI-TOFMS with the other surface analytical techniques.
ProbeDetectedsignal
Spatialresolution
Depthdirection
Chemical information
Energy-Dispersive X-ray Spectroscopy (EDS)
Electron X-ray 1 µm < 1 µm Element
Auger Electron Spectroscopy (AES) Electron Auger electron 10 nm < 10 nm Element, Chemical bonding states
X-ray Photoelectron Spectroscopy(XPS)
X-ray Electron 10 µm < 10 nm Element,Chemical bonding states,Functional group
Time-of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS)
Ion Ion 100 nm < 10 nm Element,Partial molecular structure
Laser Desorption/Ionization-Time of Flight Mass Spectrometry(LDI-TOFMS)
N u c l e a r m a g n e t i c r e s o n a n c e ( N M R )spectroscopy is an extremely important analyticaltechnique, but its sensitivity is intrinsically verylow. (Sensitivity in NMR measurements is definedby the signal-to-noise (S/N) ratio.) This is due tothe fact that NMR utilizes radio frequencies, andelectromagneticwavesatthesefrequencieshaveverylowenergy.BothUltra-violet (UV)andvisible lightspectroscopyhavemuchhighersensitivitythanNMRas theyutilizemuchhigherenergyelectromagneticwaves,soallowanalysisofmicrosamples. T hus a relat ively la rge a mou nt of sa mplei s r e q u i r e d fo r N M R a n a l y s i s c o m p a r e d t oother spectroscopic techniques. However, NMRspectroscopy is of prime importance in chemicalanalysis because of the wide range of informationit offers, often unobtainable by any other means,i nc lud i ng the deter m i nat ion of the molecu la rstructureofunknownsamples. But the low sensitivity of NMR is inevitablya major drawback in it s use. In order to makestructural analysis of e.g. natural products wheresampleamountsare limited,a realisticpracticality,variousmethodshavebeendevisedtoachievehighersensitivity.Establishedmethods,e.g.pulse&FourierTransfor m (F T) , of fer improved accumulat ioneff iciency. The use of higher magnetic f ields byresearchersand instrumentmanufacturershasalsoincreased sensitivityand theseapproachesarenowcommonlyimplemented. In addition, development of a wide range ofprobes has been on-going with the aim of moreefficientdetectionofNMRsignals.Forexample, a10mm(largediameter)probe increases the samplevolume to be detected and so increases the signal;this is good if there is sufficient sample available.A nother exa mple i s a 3 1m m d ia meter (sma l l -diameter)probewhich isgood for smallamountsofsamplesbykeeping thesolutionmoreconcentrated.Anotherdevelopment isacapillary-typeprobeusing
ahigh-sensitivitysolenoidcoilinsteadofthestandardHelmholtz coi l which is used in probes in mostsuperconductingFT-NMRsystems. Asdemonstratedinthisarticle, insteadofusinga general purpose 5 mm diameter NMR probe, awide rangeofmethodsareavailable forenhancingsensitivity. However, it must be noted that thesemethodsoffercomplementaryfeaturesandshouldbeselectedaccording to the requirementsofboth thesampleandthestudy. However,whentryingto increase thesensitivityusinga ‘normal’ sample, thesemethodsmaynotbeapplicable.Forexample,forsamplesoflowsolubilityandlimitedsamplequantity,itisinappropriatetouseeitherthelarge-diameterorsmall-diameterprobe.Toovercome this limitation,wehavedevelopedprobesthatoperateatultra-lowtemperatures.
Ultra-Low Temperature Probes and Sensitivity
Ultra-low temperature probes cool var iouscomponents (detectioncoil, etc.)used in theNMRsystem,toverylowtemperaturesusingliquidHelium(4.2K)orliquidnitrogen(77K)ascoolant.Thecooleddetectioncircuits,whicharecritical components insignaldetection,notonlyincreasethecoilsensitivitybut also reduce thermal noise. By increasing thesignal intensity and reducing noise, sensitivity isgreatlyenhanced(Fig. 1). With these probes, the detection coil and thepre-amplifier(signalamplificationcircuit)arecooledseparately. As the temperature is reduced, so theelectrical resistanceof thematerialsused for thesecircuitsalso reduces, resulting inan increase in theQ value. At the same time, thermal noise is alsoreduced, so thesensitivityof theprobe is increased,the increase being inversely proportional to theabsolutetemperature.(Equation 1) As is clear inEquation1, thedetectioncircuitscooled down to very low temperatures improvegreatly the sensitivity. InUltraCOOL™probes, thedetectioncoil is cooledwith liquid-helium,whereasinSuperCOOL™probes thedetectioncoil iscooledwithliquid-nitrogen.
Metal at ambient temperatureThethermalmotionsoftheatomsandtheelectronsareterrific.
The thermal noise and the electric resistance ofcircuitarelarge.
Metal at very low temperatureThethermalmotionsoftheatomsandtheelectronsarequiet.
The thermal noise and the electric resistance ofcircuitaresmall.
Forcoolingtheprobeandthedetectioncircuits,two systems are avai lable. The ‘Closed System’circulates cooled helium gas ; the ‘Open System’introduces liquidnitrogendirectly into thedetectioncoi l . The UltraCOOL™ probe uses the ‘ClosedSystem’,whereas theSuperCOOL™probemayuseeitherthe‘ClosedSystem’orthe‘OpenSystem’.The‘OpenSystem’ requiresa supplyof liquidnitrogen,but does not require a coolant-circulation system,thus reducing both the initial probe cost and themaintenancecost.Inbothprobesystems,thesampletemperature is independentof thecooleddetectioncircuits.IntheUltraCOOL™probe,thetemperaturedifferencebetweenthedetectioncoilandthesampleisapproximately300°C. It is important that thecoilis thermally isolated from the sample (Fig. 2 andFig. 3).Therefore, the insideof theprobeneeds tobekeptathighvacuumtoensurethermalinsulation.The UltraCOOL™ probe achieves a sensit iv ityimprovementof 4 to5 times thatof theequivalentro o m - t emp er at u re p rob e . I n t he c a s e o f t heSuperCOOL™ probe, a sensitivity improvement of2 to3 timeshigher than thatofa room-temperatureprobeisachieved. As S / N improves by the square root of thenumber of accumulations, with an improvementi n s en s i t iv i t y of 4 to 5 t i me s , t he nu mb er ofaccumulations required to measure a spectrum tothe same S/N ratio is therefore only 1/16 to 1/25.Thisofferssignificant timesavingsasmeasurementswh ich convent iona l ly requ i red severa l days tocompletemay takeonlya fewhours.This leads toagreat improvement in operational efficiency of theNMRsystem(Fig. 4andFig. 5).Inthemeasurementex a mple s how n i n t he f i g u re b elow, t he 1 3 Cmeasurement acquired with a conventional room-temperatureproberequiredalargenumberoftimes.UseoftheUltraCOOL™probeenabledconfirmationofallofthesignalsinonescan(Fig. 6). I n t h e I N A D E Q UAT E m e a s u r e m e n t ( afa mou s ly i n s en s i t ive ex p er i ment wh ich g ive sinformation on 13C-13C bonding), almost all of the
13C-13Cbondsaredetected inanacceptedperiodoftime(43hours)using theUltraCOOL™probe(Fig. 7).Thesameexperimentusinga room-temperatureprobewould take1075hours (25 times longer thantheUltraCOOL™probe).Measurementperiodsof45-daysaregenerallynotacceptable! TheUltraCOOL™probeandtheSuperCOOL™probe are also capable of stable high-temperaturemeasurementsup to150°C.Measurementsat150°Cmeansa temperaturedifferencegreater than400°Cbetween the coil (near the sample)and the sampleitself.However,detectionofNMRsignals is stableeven when measurements require a long period oftime(Fig. 8).
Summary
T h e n ew l y d eve l o p e d U l t r a C O O L™ a n dSuper C OOL™ probes of fer g reat ly en ha nc edsensitivity over conventional NMR probes, thusdramaticallyshorteningthemeasurementtime.Thesenewprobescanbeused forNMRmeasurementsathigh temperatures with operability comparable tothatofconventionalroom-temperatureprobes.Thus,the two new probes are expected to support high-temperatureanalysis invarious fields, including theanalysisofpolymersamples.
Acknowledgments
We would l ike to acknowledge Dr. Y. Godaat theNational InstituteofHealthSciences forhissupport for our measurements with the 800 MHzUltraCOOL™ probe, and for provision of samplesand data. A part of the development of the ultra-low temperature probes is now under progress asS-Innovation (Strategic Promotion of InnovativeResearch and Development) supported by JapanScienceandTechnologyAgency(JST).
NewNMRspectroscopicmethodscontinuetobedeveloped formanydifferentpurposes,particularlyfor research. However, to achieve good results ,the complex pulse sequences often require highprecisionRFcontroldependingontheapplicationorpropertiesofthesamplestobeanalyzed.Atthesametime, thedemand for routineNMRmeasurements,forexampleinthefieldsofqualitycontrolandsimpleanalysis, is remarkable, thus leading to requests forNMRtobeamore‘user-friendly’technique. Older, conventionalNMRsystemsusedanalogtechnologies thatwouldhave led to relatively largeinstruments.However,theuseofdigitaltechnologieshasbeenadvancingandthishasenableddevelopmentof next-generated NMR systems with increasedfunct iona l ity, h igher per formance and greaterexpandabilityaswellasproviding improvedgeneralversatility. Inordertomeetthesedemandsandtoanticipatefuture development of NMR measurements, JEOLR ESONA NC E I nc has developed a new N M Rsystem, the JNM-ECZ series (* Notice) . Buildingon theexperienceof the JNM-ECAII/ECXII/ECSseries, the JNM-ECZ series uses fully integratedcutting-edge digital technologies. In this report,some of the hardware features of the JNM-ECZspectrometers(ZETA)areintroduced.
ECZ Series Spectrometers (NMR Spectrometer ZETA)
TheECZseriesof spectrometersareequippedw it h Sma r t Tra n sc eiver System ( S T S) , a newtechnologywhichachieveshigh-precisiondigitalRFthusgivingperformancewhichgreatlysurpassesthatofcurrentlyexistingspectrometers.Thebasicdesignwillenable thespectrometers tooperateatultrahighfrequencies exceeding 1.2 GHz. The high quality
performance is complemented by the cutting-edgedesigninblack. The ECZ spectrometers are controlled by thebuilt-inSpectrometerControlComputer(SCC),andtheSCCiscontrolledbythehostcomputerconnectedth roug h a n Ether net l i n k . T he host computerprovides thedirectuser interface,but theSCCcanoperate stand-alone. This prevents the danger ofmeasurement omission should a communicationproblem oc cu r between the SCC a nd the hostcomputer.TheSCCincorporatesalargememoryandharddisk, thusemphasizing thesecurereliabilityofthepulse-programmedandmeasureddata.
Two “Z” (ECZR/ECZS Series Spectrometers)
Thereare twospectrometers in theECZseries.Thespectrometershavecustomizedfeatures tomeetdifferentapplicationneeds.
ECZR series spectrometer (Fig. 1)
T he h ig h- end mo del EC Z R spe c t rometeris conf igured pr imar i ly for a research or ientedworkplace. With a highly f lexible and expandableconfiguration tomeet thedemandsofvariousNMRmeasurements, the ECZR is compact comparedto currently available spectrometers, and achievesoverwhelmingperformance.
ECZS series spectrometer (Fig. 2)
The entry model ECZS spectrometer has thesamebasic functions,performanceandcapabilityofthe ECZR spectrometer. Furthermore, the ECZSspectrometer isevenmorecompact than theECZRspectrometerandoffersgoodgeneralversatility.ThemainconsoleoftheECZSspectrometerisamazinglysmall, less than 1/2 that of the current ECS seriesspectrometer.3-1-2 Musashino, Akishima, Tokyo, 196-8558, Japan
The ECZ ser ies spectrometers bui ld on thehighlysuccessful systemarchitectureof theECAII/ECXII/ECS series spectrometersbutareequippedwith a highly advanced STS (Smart TransceiverSystem) developed u s i ng cut t i ng- edge d ig it a ltechnologies. STS allows the construction of high-precisionRFcontrolwithhigh-speeddigitalcircuitsmounted on a small logic device. Because of this,the ECZ series dramatically improves the digitalfunctionsandperformanceof theRFtransmitter&receiver system. Inaddition, theECZseriesutilizea compact spectrometer that integrates the basicfunctionsof theconventionalNMRsystem intooneboard(Fig. 3).
Multi-sequencer control
EachDDS(DirectDigitalSynthesizer) for theRF transmitter& receiver system is independentlycontrol led at high speed by the respective slavesequencers. This independent control by the slavesequencers is comprehensively managed by themaster sequenc er. T h i s mecha n ism a l lows forhighly flexiblecontrol, thusallowing thecreationofversatile pulse sequences. For example, the ECZRseries spectrometercancontrolover30 sequencers,
that is,morethan3 times thoseof theECAIIseriesspectrometer.Thus, theECZRserieswillbeable tosupport many kinds of measurement methods thatmayberequiredinthefuture.
High-precision digital control
T he t ime resolut ion for each of f requencymo du l at ion , ph a s e mo du l at ion a nd i nt en s i t ym o d u l a t i o n , w h i c h a r e s i m u l t a n e o u s l y a n dindependently controlled as digital RF signals bythe sequencers, is as small as 5 ns. This ultimatehigh t ime resolution al lows for the control of aduration(modulationtimewidth)of5ns(minimum).This corresponds to approximately 10 to 20 timesimprovement when compared to that offered bycurrentlyavailablespectrometers.Moreoveraseachcharacteristic is accurately controlled, the overallperformanceisalsoimproved(Fig. 4).Thisimproveddigital-control performance further enhances theeffectiveness of phase and intensity modulationpulses such as adiabatic pulse schemes (frequentlyusednowadays).Alsoinordertomakemeasurementsrequiring ultra-high speed control, e.g. in recentsolid-stateNMR,theSTSof theECZspectrometersprovideshighaccuracyincontrollingthegatesignalsandtheexternalinput&outputtriggersignals.
In this system, conventional R F osci l lat ionand transmission functions are highly integrated.The new RF system can output up to 4 differentfrequencies for each RF transmitter channel. Inaddition, an expansion of the variable frequencyoffsetrangeallowstheECZspectrometerstosupportcomplicated measurements such as simple tr ipleresonancewithinthestandardconfiguration.TheRFdetectionsystemisequippedwithasequencercontrolfunction comparable to that of the RF oscillationsystem. This makes it possible for the ECZ seriesspectrometers tocarryoutdynamicmodulationsoffrequency and phase with or without synchronism,thusimplementingimportantcutting-edgesolidstateNMRmethodswhichhave recentlybeenpublished.In addition, DQD (Digital Quadrature Detection)provides a way to reduce artifacts including QD(QuadratureDetection) imageandthecentre ‘spike’at0Hz,therefore,theimproveddigitalRFcontroloftheECZspectrometersmakes theanalysisofNMRspectraclearer.
Analog RF control
IntheRFtransmitter&receiversystem,ahybridsystem that combines under-sampling with super-heterodyneandover-samplingwithdirectconversionis achieved using a h igh-speed D /A (Digita l toAnalog)converteroperatingat800Mspsandhigh-speedA/D(AnalogtoDigital)converteroperatingat100Msps.Thismakesthetransmissionandreceivingefficientdependingon theRFsignalsand linkedbyanoptimizedfilteringmechanism.
PFG control and Digital-lock control
STS is also used for both PFG (Pulse FieldGradient) and lock control, and provides digitalhighperformancecomparable to thatofoscillationanddetectionofRFsignals. Inparticular, theECZspectrometers allow for lock-control with higherprecision and better flexibility provided by a lock-feedback mecha n i sm based on d ig ita l c ont roltechnologies. This digital-control design enablesthe ECZ spectrometers to provide magnetic-fieldcorrection according to the environment of the
Fig.4 RFwaveformsgeneratedbymodulationcontrol.
Frequencymodulation(0to10MHz)
Intensitymodulation(0to100%)
Phasemodulation(0to180degrees)
(2014) Vol. 49 No. 1 JEOL News 98
instrumentandsamples.TheECZspectrometerscanalso use the lock transmitter & receiver system inapplicationmeasurements.
Touch panel display (Head amplifier chassis)
The head amplifier chassis displays functionsrelated to the super conductingmagnet (SCM)andtheprobemounted in theSCM.Ontopof theheadamplifier chassis, a large-screen (5”) touch paneldisplay is mounted, providing an intuitive multi-functional user interface. The spectrometer candisplaytheRFreflectiondiporareflectionvalue(bardisplay)duringprobetuning,andtheresiduallevelofmagnetcryogenscanbedisplayed in real time(Fig. 5).This function improvesvisualusabilityaswellasoperability.
Summary
The new JNM-ECZ ser ies of spectrometershasbeendevelopedwithbasicdesignconcepts that
supportsexcellent functionality,highperformance,a re h ig h ly expa ndable a nd of fer h ig h genera lversatility. Furthermore, the JNM-ECZ series hasextremely h igh potentia l that supports f lex ibleapplications for the future development of NMRmeasurements.Weexpect thatour innovativeJNM-ECZ series will meet and exceed a wide range ofdemands invarious scientific fieldsand to serve incutting-edgeresearcharoundtheworldaswellas ingeneral-purposeanalyses.
* NoticeInstrument specifications are subject to change without notice.
