ContentslistsavailableatScienceDirect
Materials
Today
Communications
jo u r n al h om ep ag e : w w w . e l s e v i e r . c o m / l o c a t e / m t c o m m
Vertically
aligned
carbon
nanotube
field
emitter
arrays
with
Ohmic
base
contact
to
silicon
by
Fe-catalyzed
chemical
vapor
deposition
M.
Morassutto
a,
R.M.
Tiggelaar
a,
M.A.
Smithers
b,
J.G.E.
Gardeniers
a,∗aMesoscaleChemicalSystems,MESA+InstituteforNanotechnology,UniversityofTwente,P.O.Box217,7500AEEnschede,TheNetherlands bNanolab,MESA+InstituteforNanotechnology,UniversityofTwente,P.O.Box217,7500AEEnschede,TheNetherlands
a
r
t
i
c
l
e
i
n
f
o
Articlehistory: Received12April2016 Accepted19April2016 Availableonline20April2016 Keywords:
Alignedcarbonnanotubes Thermalcatalyticchemicalvapour deposition
Fecatalyst
a
b
s
t
r
a
c
t
Inthisstudy,densearraysofalignedcarbonnanotubesareobtainedbythermalcatalyticchemicalvapor deposition,usingFecatalystdispersedonathinTalayer.Alignmentofthecarbonnanotubesdepends ontheoriginalFelayerthicknessfromwhichthecatalystdispersionisobtainedbythermaltreatment priortothesynthesisstep,aswellasonthesynthesistemperature,whichinthisstudywasvaried between680and740◦C.Fortheintendedapplicationofthenanotubesasfieldemitters,itisessential thatelectricalconductivitybetweenthetubesandthesubstrateisguaranteed.Inanovelapproach,a noblemetallayer,of100nmPtorAu,wasappliedbetweentheTalayerandthesupport,whichensures anOhmiccontactthroughoutthemetalmultilayerwhichconnectsthesiliconsubstratewiththeCNT fibers,becauseoxidationofTaatitsinterfacewithSiisprevented.Foroptimizeddensearraysofaligned nanotubeswithlowOhmicresistance,fieldemissioninnitrogenatmosphereisdemonstrated,withfield enhancementfactorsofuptoca.2000fornanotubesgrownusingFe/TacatalystdepositedonAuandPt. ©2016TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
The promising applications of carbon nanotubes (CNTs) in differentfields,duetotheirexceptionalchemical,mechanical, ther-malandelectricalproperties,hasinitiatedasignificantresearch effortoverthelasttwentyyears[1–3].Methodsfrequentlyapplied tosynthesizevertically alignedCNTs(VACNTs)are thermal cat-alyticchemicalvapordeposition(TCCVD)[4]andplasmaenhanced chemicalvapordeposition(PECVD)[5],whichbothpermitthe for-mationofdensearraysofVACNTsondifferentsubstratesbytuning andcontrolofsynthesisparameters.InTCCVD,whichisthemethod usedinthisstudy,avarietyofmetalcatalyst/supportcombinations hasbeenused,includingCo,NiandFe,onalumina(Al2O3), sil-ica(SiO2)andoxidizedSi[6,7].Ofthemanycatalysts,Fe(withor withoutaTaadhesionlayer)hasgainedattentionduetothe possi-bilitytoformVACNTsonelectricallyconductivesupports[8,9],for whichimportantapplicationscanbefoundinfieldemissionand vacuummicroelectronicdevices,sinceVACNTsbelongtothebest electronemitterscurrentlyavailable[10–12].Sinceadense emit-terarrayisexpectedtogivethehighestemissioncurrentdensity, itisofhighrelevancetoidentifyandoptimizetheTCCVD
param-∗ Correspondingauthor.
E-mailaddress:j.g.e.gardeniers@utwente.nl(J.G.E.Gardeniers).
etersthataffectthealignmentofCNTsonathinfilmofFe/Ta,and topointoutcorrelationsbetweenfactorsinfluencingvertical align-mentandfactorsdeterminingelectricalproperties.Inthisworkwe performastudyoftheinfluenceofthehydrocarbonsourceandthe synthesisparameterstemperatureandtimeofpretreatmentand growthstagesontheverticalalignmentandelectricalproperties ofVACNTs.Wewilldemonstratethatanoblemetallayerinserted betweenthesupportandtheTalayerhasabeneficialeffect,and willshowfieldemissionwiththesynthesizedCNTarraysin1atm nitrogengas.
2. Experimental
2.1. Samplepreparation
Fe thin films were deposited on wafers of fused silica (UV grade,100mmdiameter,500m thick;MarkOptics)orSi (p+-typeborondoped,(100)-orientation,resistivity0.01–0.024cm, 100mmdiameter,thickness525m,singlesidepolished;Okmetic, Finland).Thesupports werecleanedfor10mininfuming100% nitricacid(UN2031OMGroup)and15mininboiling69%nitric acid(51153574BASF),followedbyrinsingindemineralized(DI) water,andspindrying.Onthewafers8×8mm2squares,centered withrespecttoasamplesizeof10×10mm2,weredefinedinOlin 907-17photoresistbystandardUV-lithography.Priortoloading http://dx.doi.org/10.1016/j.mtcomm.2016.04.007
2352-4928/©2016TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4. 0/).
Table1
CompositionofFe-basedthinfilmsonvarioussupports.
Sampleseries Felayer(nm) Talayer(nm) Interlayermetal(nm) Adhesionlayerforinterlayer(nm) Substrate
A1-A8 2 10 none none p+Si(100)
B1-B10 4 10 none none p+Si(100)
C1-C4 6 10 none none p+Si(100)
D1-D8 3 10 none none p+Si(100)
E1-E3 3 10 none none fusedsilica
F1-F3 3 10 Au,100 Ta,10 p+Si(100)
G1-G3 3 10 Pt,100 Ta,10 p+Si(100)
inthemetaldepositionsystem,siliconsupportswereimmersed for1minin1%hydrofluoricacid(D252MHoneywell)toremove thenativeoxide,followed byDIwater rinsingandspin drying. ElectronbeamevaporationofFethinfilms,withcompositionsas giveninTable1,wasdoneinaBalzersBAK600system.Metalfilms weredepositedatapressurebelow10−5Pa,andifmultiple lay-ersweredepositedonthesamesubstrate,thiswasdonewithout breakingthevacuum.Filmthicknesswasmonitoredwithanin-situ quartzcrystalmonitor.Evaporationmaterialpuritywas99.95%for Ta,99.95%forFe,99.99%forPtand99.999%forAu.After evap-oration,ultrasoniclift-offofthephotoresistlayerwasperformed inacetone(20min;VLSI51150924,BASF)followedbyimmersion inisopropanol(10min;VLSI51152037,BASF),rinsinginDIwater (10min)andspindrying.Alayerofphotoresistwasspin-coated onthemetalsurfacetoprotectitfromcontaminationand dam-ageduringdicingof10×10mm2samples(DiscoDAD-321dicing machine).Afterdicing,thephotoresistwasremovedvia ultrasoni-cationinacetone.
2.2. Substratepretreatmentandsynthesisof(VA)CNTs
Samples were positioned flat, with deposited metals facing upwards,inaquartztubethree-zoneheatedreactor(Elicra Elec-trowarmteB.V.).Thisreactorhasaflattemperaturezonewitha variationintemperatureoflessthan10Koverca.18cmofreactor length.Inordertoguaranteeawell-definedandequaltemperature ateachsamplesurface,allsampleswereplacedinthiszone,with apositionaltoleranceof±0.5mm.Nomorethan6samplesperrun wereloaded.
As-deposited continuous Fe thin films were annealed in a gas mix of hydrogen (99.999%, specs<1ppm O2, <3ppm H2O; Indugas/Praxair)and nitrogen (N2; 99.999%, specs: <1ppm O2, <3ppmH2O;Praxair)atatmosphericpressuretodewetthefilm, leadingtometallicnanoparticleswhichsubsequentlyactasstarting pointsforVACNTgrowth.Immediatelyafterthisreductionstep(to becalled“pretreatmentstage”),thereactionmixturewasadapted toinitiatecarbonnanostructureformation(the“growthstage”), inwhichethylene(C2H4;99.96%,specs:<5ppmO2,<1ppmH2O; Praxair)wasusedascarbonsourceandnitrogenasadilutinginert gas.IncaseswherethefocuswasonSEMimagingtostudythe effectofthepretreatmentstageoncatalystparticlesizeand den-sity,thecarbonsourcegaswasleftoutandonlyN2gasflowwas applied,atthesametemperatureandforthesameperiodasinthe growthstageofCNTsdevelopingfromthesamethicknessFe start-inglayer.Varioussettingsoftemperatureandgasflowcomposition wereappliedforthegrowthstage(Table2).Allexperimentswere repeatedatleastthreetimes.Ifnotmentionedotherwise,presented resultsarereproducibleandshownimagesarecharacteristicforthe outcomeofsynthesisexperimentsatgivenconditions.
2.3. AnalysisofCNTs
ThemorphologyofgrownCNTswasinvestigatedusing high-resolutionscanningelectronmicroscopy(HR-SEM;Zeiss,Merlin) andtransmissionelectron microscopy(TEM;Philips,CM300ST).
Samples with catalyst particles after pretreatment were ana-lyzedwithHR-SEM.In ordertodeterminethesizedistribution ofnanoparticles(NPs)formedonasamplesurfaceafterthe pre-treatmentstage,topviewSEMimageswereanalyzedwithImageJ software (using the toolbox“analyze particles”). Of each SEM-image,withamagnificationof50.000×,fiveareasof1m×1m wereconsidered.Theseareaswereconvertedtograyscalevalues (8-bit),followedbybinarization,thresholding,andidentificationof theedgesofparticlesanddiscretization,inordertoobtainparticle sizeandsizedistribution.
TheelectricalresistanceofsampleswithVACNTswasmeasured withasourcemeasurementunit(2410,Keithley).Contactpadsto sampleswithVACNTswerecreatedviasputteringofa100nmthick Aulayeronthebacksideofthesilicon(thisdepositionwasalready donepriortoCNT-synthesis),andviagluingacopperwiretothe VACNTs,withconductivepaste(CW2400Circuitworks)appliedas a 2.5mm diameterdot. Thisglue was alsousedto contactthe backsideAupatchwithacopperwire.Afterelectricalanalysisthe conductiveepoxygluewasremovedinordertoverifythatitdid notpenetratethroughtheCNT-film,whichwouldhaveresultedin adirectelectricalpathtotheTaandsiliconsubstrateunderneath theVACNTs.DuringremovaloftheglueconsistentlytheCNTswere tornofthesample,whilethemetalremainedonthesamplesurface, whichprovesthattheepoxygluedropwassufficientlyviscousto penetrateonlyintothetoplayeroftheVACNTarray.
FieldemissionfromVACNTsinN2 atmosphereat 1atmwas testedforthreedifferentsampleconfigurations(samplesID’s:D7, F3andG3),whereeachsamplewasplacedat500mfromaflat stainlesssteel(AISI316L)anode.Therequiredelectricalfieldwas generatedusingthesourcemeasurementunitmentionedabove, anddatawererecordedbyacustomizedLabviewprogram.
3. Resultsanddiscussion
3.1. GrowthofVACNTs
Inordertounderstandiftheinitialphaseofthesynthesis proce-dureaffectsverticalalignment,CNTsweregrownon3nmFeonTa onSi,at720◦Cforvarioussynthesistimes(samplesD1,D2,D3,D4, Tables1and2).TheresultsinFig.1indicatearandomorientation ofCNTsforgrowthtimesupto15min.Especiallyfortheshortest growthtimes,2.5and5min,notallFeparticleshavebeenactivein formingCNTs,whichpointsoutanincubationtimeforCNTgrowth. Becauseoftheextensivepretreatmentstage(seealsonextsection) thatprecededthegrowth,itisexpectedthatcompletebreaking upoftheoriginalcontinuousFefilmintoactivenanoparticleshas alreadyoccurred.Furthermore,Fecatalystcarburization (satura-tionoftheiron leadingtoanironcarbide, suchasFe3C, which proceedsCNTformation)wasreportedtooccuratamuchfaster timescale,oftheorderof0.1s[13].Therefore,thetimeeffectin Fig.1canonlybeattributedtothepresenceofathinoxideshell ontheparticles,whichrendersthemtemporarilyinactive[14].The ironoxidemaybereducedbyhydrogen(atoms)generatedthrough decompositionofethylene,or,aswasproposedbyBayeretal.[15], byTaactingasasolid-statereducingagent.
Table2
Conditionsofpretreatmentandsynthesisstageforeachsample.
Pretreatmentstagea Growthstageb
SampleID Temp(◦C) Time(min) Temp(◦C) Time(min) N2flow(ml/min) C2H4flow(ml/min)
A1,B1,C1 680 180 680 60 100 50 A2,B2,C2 680 180 680 60 50 D1 720 180 720 60 100 50 D2 720 180 720 2.5 100 50 D3 720 180 720 5 100 50 D4 720 180 720 15 100 50 A3,B3,C3 720 180 720 60 50 A4,B4,C4 720 180 720 60 100 50 B5 680 180 680 60 50 B6 680 120 680 60 50 B7 680 60 680 60 50 B8 680 180 680 60 100 50 B9 680 120 680 60 100 50 B10 680 60 680 60 100 50 A5 680 180 680 60 50 A6 700 180 700 60 50 A7 720 180 720 60 50 A8 740 180 740 60 50 E1,D5,F1,G1 720 180 720 60 100 50 E2,D6,F2,G2 720 180 720 60 50 E3 720 120 720 60 100 50 D7 720 180 720 60 100 50 F3 720 150 720 60 100 50 G3 720 110 720 60 100 50 D8 720 180 720 30 100 50
aPretreatment stage: both temperature ramp at +10◦C/min and annealing during the time mentioned in the third column were performed in a flow of
40ml/minN2+10ml/minH2.
bAftergrowthsamplesarecooledatarateof10◦C/mininaflowof100ml/minN2.
Fig.1.CNTssynthesizedon3nmFe/10nmTathinfilmsonsiliconsupportsat720◦Catvariousgrowthtimes:(a),2.5min(sampleD2),(b)5min(sampleD3),(c)15min
(sampleD4),d)1h(sampleD1).
In another study Bayer et al. [9] established a correlation betweensynthesistemperatureand verticalalignment ofCNTs, withthebestalignmentat550◦Canddecreasingalignmentabove 600◦C.ThetwomainprocessesinvolvedarethereactionofTatoits oxidesorcarbides,whichiscatalyzedbyFe,andthereductionofFe, enhancedbyTa.DuetodiffusionofFeviagrainboundariesinthe crystallineTa2O5thatformsathighertemperatures,Feisremoved fromthesurfaceandlessdense,andtherewithlessaligned,CNT arraysaregenerated.TheoxygenneededforTaoxideformation
(whichisamorphousatlowtemperatures)originatesfromresidual O2inthebackgroundgasesandfromexposuretoairin-between filmdepositions(inBayer’scaseTawassputteredandFe evapo-ratedindifferentvacuumsystems).
InourstudywehavetakencaretodeposittheTaandFefilm combinationsinoneandthesamesystem,withoutbreakingthe vacuum.Unfortunately,itcouldnotbeavoidedthatthemetalfilms becameexposedtoatmosphericairconditionsduringtransferfrom themetaldepositionsystemtotheTCCVDreactorforCNT
depo-Fig.2. (a-c-e)CNTssynthesizedonFe/Tafilmsonsiliconat680◦C/1h(samplesA1,B1,C1);(b-d-f):Fe/Tafilmsonsiliconexposedtosameconditions,butwithoutcarbon
source(samplesA2,B2,C2).Allsampleswerepretreatedat680◦C/3hinaH2 N2mix.(a/b)2nmFeon10nmTa,(c/d)4nmFeon10nmTa,(e/f)6nmFeon10nmTa;(g)
SizedistributionofFe-NPson1m×1mfootprintonsilicon.
sition.This willinevitably have resultedin oxide formationon theouterFelayer,butmostlikely notontheTa layer,because itisentirelycoveredbytheFelayerwhichisstillcontinuousat thisstage.DuringtheTCCVDofCNTsresidualoxygen-containing speciesarepresentinourreactorinppmconcentrations, originat-ingfromtheusedprocessgases(seeSection2.2).Togetherwith themuchhighersolid-statediffusionratesatthepretreatmentor synthesistemperatures,incombinationwiththeexposureoftheTa filmtothegasatmosphereduetodewettingoftheFelayer,thiswill
withoutdoubtleadtoTaoxidation.Bayeretal.haveproventhat acrystallineTa2O5phaseisformedundersuchconditions[9],but wedidnotfurtherverifythisinourwork.Bayeretal.attributethe decreaseinCNTyieldathighertemperaturestothefastdiffusion ofFeawayfromthesurface,whichoccursviathegrain bound-ariesinpolycrystallineTa2O5.Itisnoteworthythoughthatwestill obtaingood CNTalignmentattemperaturessignificantlyhigher thanBayer etal., whichwe attribute toalower oxygenpartial
Fig.3.Pretreated2nmFe/10nmTafilmsonsiliconfor3h:(a)680◦C(sampleA5),(b)700◦C(sampleA6),(c)720◦C(sampleA7),(d)740◦C(sampleA8)(e)Sizedistribution
ofFe-NPson1m×1mfootprintonsiliconuponpretreatmentatvarioustemperatures;(f)Averageheightoftheparticlesonsiliconuponpretreatmentatvarious temperatures.
pressureinoursystem,leadingtoahighertemperatureatwhich crystallineTa2O5formationbecomesprominent.
Asis seeninFig.1,thetendencytowardsverticalalignment increasesforincreasinggrowthtimes,untilforasufficientlylong synthesistime(1h)theCNTsexhibitalignedgrowth.Thisresult confirmsthatalignmentofCNTs,grownbyTCCVD,arisesfromthe crowdingeffectofneighboringCNTs.
3.2. InfluenceofFethicknessonalignedgrowth
AdifferentFefilmthicknessisexpectedtoresultindifferent catalystparticlesizeandsurfacedensityafterdewetting,andwill thereforeaffectthedegreeofCNTcrowding,whichisthedriver foralignedgrowth.TheinfluenceofFelayerthicknesswasstudied byexposingsampleswithlayersof2,4and6nmofFetoidentical conditions,i.e.a3hpretreatmentstepat680◦Cunderareducing atmosphere,followedby1hgrowthat680◦C(seeTable2).
Fig.2a–eindicatethatinthestudiedrangeanincreaseinthe FelayerthicknessisbeneficialfortheformationofVACNTs.The effectofFelayerthicknessshouldhoweverbeanalyzedtakinginto
accountthepretreatmentandgrowthconditionsandtheinfluence of the support onto which the catalyst film is deposited. Dur-ingpretreatmentinareducingatmosphere,morphology,sizeand compositionofcatalyticFe-NPswillchangeduetothreemain pro-cesses:(i)Fe2O3presentonthefilmsurfaceisreducedtometallicFe bytheH2atmosphere;(ii)becauseofsurfaceenergyminimization, theFefilmdewetsandformsNPs;and(iii)theformedNPs agglom-eratetolargerparticlesbyOstwaldripeningorsintering,whichis alsodrivenbysurfaceenergyminimization,andoccurstypically abovetheTammantemperature(halfthemeltingtemperature), whichisca.630◦CforFe.
TheincreaseofNPsizebyagglomerationevidentlyleadstoa decreaseinthenumberandtherewithtoalowersurfacedensityof Fe-NPs.Forequalpretreatmentconditions,anoriginallythickerFe layergivesahighersurfaceNPdensity,consequentlywithasmaller gapbetweentheagglomeratedNPs.ThiscanbeseenintheSEM imagesofFig.2b,dandfofsamplesafterannealingwithoutgrowth, whichshowanincreaseinNPdensitywithstartingfilmthickness, whereasparticlesizeanalysisinFig.2gdemonstratesanincreased densityofNPswithdiameters8–30.Theincreaseinthedensity
Fig.4.(a-c-e)CNTssynthesizedonFe-basedthinfilmsonsiliconsupportsat720◦C/1h(samplesA4,B4,C4);(b-d-f):Fe/TafilmsonsiliconpretreatedinH2/N2at720◦C/3h
(samplesA3,B3,C3);(a/b)2nmFeon10nmTa,(c/d)4nmFeon10nmTa,(e/f)6nmFeon10nmTa;(g)SizedistributionofFe-NPson1m×1mfootprintonsilicon upon720◦C/3hpretreatmentinH2/N2.
ofthisspecificsizeofparticlesisassociatedwithatransitionto verticalalignment ofCNTs(Fig.2a,cand e).Inagreementwith thisistheobservationinFig.1thatFeparticlessignificantlylarger than30nmseemtostayinactiveatthesubstratesurface.Wedo notexcludethattheselargerparticlesbecomeactivelaterinthe growthstage,whentheyloosetheirprotectiveoxygenshell,aswas suggestedbyNessimetal.[14].
3.3. Influenceofpretreatmentconditions
Experimentswerecarriedoutfora2nmFelayerthicknessand 3hpretreatmenttimeinH2flow,atdifferentpretreatment tem-peratures.Ifthetemperatureisincreasedfrom680to740◦C,the surfacedensityofparticlesfirstincreasesandatthehighest tem-peraturelargerparticlesareseenduetoanagglomerationeffect whichapparentlygivesabimodaldistribution(Fig.3).Alsofor4 and6nmFelayersanincreaseofthepretreatmenttemperature
Fig.5. (a-c-e)CNTssynthesizedon4nmFe/10nmTathinfilmsonsiliconsupportsat680◦C/1h(samplesB8,B9,B10);(b-d-f):4nmFe/10nmTafilmsonsiliconpretreated
at680◦C(samplesB5,B6,B7);(a/b)1hpretreatment,(c/d)2hpretreatment,(e/f)3hpretreatment;(g)SizedistributionofFe-NPson1m×1mfootprintonsiliconupon pretreatmentat680◦Cfordifferenttimes.
from680◦Cto720◦CwasobservedtocreateahigherdensityofNPs (datanotshown).Onewouldexpectthistoleadtobetteralignment oftheCNTs,whichisindeedthecaseforfilmsofallthicknesses, compareFig.2a,c,ewithFig.4a,c,e,respectively.
Besidespretreatmenttemperature,pretreatmenttimein reduc-ingH2atmospherealsohasaneffectonNPformation.Nessimetal. [14]demonstratethattheinitialrateofNPformationislowbecause Fe2O3onthesurfaceoftheFefilmblocksdewettingandcoarsening. Asubstantialincreaseinthecoarseningrateoccursduring
reduc-tioninH2at770◦C,whichlevelsoffafterca.25minannealingto giveFeparticlesonAl2O3withanalmostconstantradiusof curva-ture(intheircase35nm,fromastartinglayerthicknessof1.2nm). InFig.5SEM imagesareshown ofoursampleswitha4nm Fe layer,pretreatedfor1h,2hor3hat680◦CinaH2/N2atmosphere. TheNPsizedistribution(Fig.5g)keepschangingwith pretreat-menttime,suggestingthatanequilibriumasdescribedbyNessim hasnotyetbeenreachedevenafter3hannealing.Direct compar-isonwithNessim’sworkisdifficult,becauseweusedathickerFe
Fig.6. Zoom-inSEMimageofthesampleofFig.5a(sampleB8),whichshowsa carbonmicrostructureddepositnearthesubstrate-CNTinterface,whichhasnot developedintowell-definedCNTs.
film,adifferentsubstrateandperhapsmostimportantly,alower pretreatmenttemperature,whichmaymeanadifferentFe oxida-tiondegree[16].RemarkableinthisrespectisalsotheworkofIn etal.whodidnotfindextensiveripeningwhentheychangedtheir pretreatmenttimeinreducingatmospherefrom17minto1h[17]. Theseauthorshavetakencaretousegaseswithoxygen-containing speciesbelow1ppblevels,anddidapretreatmentatina40vol-% hydrogeninheliummixtureattemperaturessimilartoours.
Foratemperatureof680◦C,VACNTsareformedonlyfora pre-treatmenttimeof3h(Fig.5e),forshortertimesrandomlyoriented CNTsarefound.Anexplanationforthisresultcouldbethat1h pre-treatmentinagasmixtureof20%H2inN2isnotenoughtoremove alloxidefromtheFesurfaceatthelowertemperatureof680◦C. However,anadditionalphenomenonalsoplaysarolehere,and thatisFe-carbideformation.For1hpretreatmentalayerof car-bonaceousrandomlyorientedmicrostructuresisobservedatthe substratebetweentherandomlyorientedCNTs(Fig.6),whichwas notobservedfora2nmFelayerwith3hpretreatmentthatgave roughlythesameamountofparticlesonthesurface(Fig.2).Nessim etal.reportedthattheabsenceofhydrogenduringsynthesis nega-tivelyaffectstheformationrateofCNTsasaconsequenceofcarbide layerdeposition[14].Thehydrogengeneratedbydecomposition ofthecarbonsourcegasmaynotalwaysbeenoughtoremove thislayer.ThequestionnowishowthepretreatmentinH2 atmo-spherewithoutanycarbonsourcecaneffectcarbideformationin thegrowthstage.
Theaffinityof Ta forgases likeH2 and O2 is awell-studied topic.Shleifmanetal.demonstratedthat(inanultrahighvacuum chamber)athinfilmofTaabsorbsH2[18],andUchidaandFromm demonstratedthataTalayeralsoabsorbsH2whenitssurfaceis cov-eredwithaFefilm[19].Asanexplanationforthecarbide-related phenomenadiscussedabove,wethereforeproposethatduring pre-treatmentabsorptionofH2bytheTalayeroccurs.ThisH2impedes carbidedepositionontheFe-NPsduringthegrowthstage.Atoo lowamountofabsorbedH2availablefortheFe-NPsdecreasesthe numberofactivesitesfornucleationandgrowthofCNTs,resulting inalowerdensityofCNTs,whichthenwillhaverandomgrowth directionsduetoabsenceofstericimpedance.Apretreatmenttime of1hat680◦CseemsnotsufficienttoabsorbenoughH2.
Weconcludethat,foragivenFefilmthickness,increasingthe pretreatmenttimefrom1hto3hatfixedtemperatureshiftsthe sizedistributiontowardslargerparticlesmorethanincreasingthe pretreatmenttemperaturefrom680◦Cto720◦Cwithfixed anneal-ingtime.Ourobservationsareinagreementwithreporteddataon Fesurfacemobility[20]andNi/Tadewetting[21].Intheworkthe authorsstatethatthepresenceofTafavorsdewettingofNi coat-ings,comparedtoNidirectlyonoxidizedsiliconsupports,andit isverylikelythatasimilareffectoccursforFecoatingsonTa.This notionmotivatesourfollowinginvestigationoftheinfluenceofthe
supportandthecompositionofFe-basedmetallicthinfilmsonthe synthesisofVACNTs.
3.4. Synthesisstage:effectsofadhesionlayerandsupport
AcombinatorialapproachwasusedbyNgetal.toanalyzethe influenceofvariousadhesionmetalsunderneaththinlayersofFe, Ni,CoandFe/Nialloycatalysts[22].Theseauthorsconcludethat thetypeofintermediatelayerinfluencestheappearanceofgrown CNTs,whichareverticallyalignedforFe/NicatalystonAland ran-domlyorientedintheothercases.Nessimetal.employedSEM analysistoinvestigate theeffect ofpalladium (Pd)asadhesion layerforNicatalystandTaasadhesionpromoterforFe[23],and foundthatTa,independentofitslayerthickness,isthemore sta-bleinterlayerbecauseitdoesnotexhibitgrainboundarymobility, incontrasttoPdwithsubstantialgraingrowth,leadingtomoreNi incorporation,thethickerthePdfilm.Burtetal.studiedCNTgrowth onNi/Alfilmsdepositedonbareandoxidizedsiliconsupportswith SEM[24],whereasTiggelaaretal.andThakuretal.investigated CNT-growthandpretreatmenteffectsofNi/TaandNi/Tithinfilms onoxidizedsiliconandfusedsilicasupports[21,25]bymeansof HR-SEMandRamanspectroscopy.Commonconclusionofthis lit-eratureisthattheadhesionlayerinfluencescatalystparticlesize anddistribution,byaffectingboththesurfaceandthesub-surface mobilityofthecatalyticmetaluponreductionpretreatment.
SincegoodadhesionofsynthesizedCNTsisanecessitytoobtain properelectricalcontact, wehave maintaineda 10nm thickTa interlayerinallsubsequentexperiments.Fe/Ta(3nm/10nm)was depositedontosilicon,fusedsilica,andsiliconcoatedwith100nm AuorPt,whichbothalsohada10nmTaadhesionlayerunderneath. SamplesofeachconfigurationwereexposedtoN2/H2for3h, fol-lowedby1hCNTgrowth,allatatemperatureof720◦C.Resultsof theseexperimentsareshowninFig.7.
Clearly,alignmentisreasonablygoodforFe/TaandFe/Ta/Au/Ta films onSi (Fig. 7c, g), but poor for Fe/Taon fused silica and Fe/Ta/Pt/TaonSi(Fig.7a,e).SEM-imagesofpretreatedsamples showthatidenticalFe/Tafilmsonsiliconandfusedsilicadewet dif-ferentlyuponidenticalpretreatmentconditions(Fig.7b,d),which isconfirmedbythesizedistribution(Fig.7i).FeonAugivesresults similartoFeonSi,whereasinthecaseofFeonPtaloweramount ofparticlesisfound(Fig.7i),whichmostlikelyisthereasonfor thelowerdegreeofalignmentinthiscase.Ifasufficientdensityof catalystparticlesisobtained,alignedgrowthonasamplewithaPt interlayerbecomesfeasible(seeFig.8cbelow).
SampleF3inFig.8dbelowisthesamplewiththebestvertical alignmentthatwaspossiblewithourTCCVDsystem.Foreven bet-teralignment,synthesis methodswhichemployelectricalfields perpendiculartothesubstratesurface,likePECVD,maybemore suitable.TheplasmainPECVD,ortheelectricalfielditself,maythen beappliedtocreateadirectionalityinthegrowthofCNTs.Similar conceptsaredifficulttoapplytoTCCVD,wheretheonlycontrol parametersthatwouldgivesomedirectionalityareconcentration orthermalgradients(boundarylayers)perpendiculartothe sub-stratesurface,whichthenhavetobematchedwiththestericeffects thatthegrowingCNTsposeuponeachother.
Bytuningthepretreatmenttime,itispossibletoobtainVACNTs atgrowthconditionsof1h/720◦Cfor3nmFe/10nmTaon differ-entsupports,exemplifyingourcontrolofdewettingofFe-based thinfilmsandsubsequentCNT-growthonvarioussupports.For example,Fig.8ashowsVACNTson3nmFe/10nmTaonfusedsilica pretreatedat720◦Cfor2h,whereasonasiliconsupporta pretreat-menttimeof3hisrequiredtoobtainsuchVACNTs(Fig.8b).Ascan beseeninFig.9,whichisacharacteristicTEMimageofasample preparedatsimilarconditionsastheoneofFig.8b,butwithhalf thegrowthtime,theCNTsaremultiwall.WhenaAu/TaorPt/Ta layerisimplementedonthesiliconsupport,apretreatmenttime
Fig.7. (a-c-e-g)CNTssynthesizedon3nmFe/10nmthinfilmsonvarioussupportsat720◦C/1h(samplesE1,D5,G1,F1);(b-d-f-h):3nmFe/10nmTafilmsonvarious
supportspretreatedat720◦C/3h(samplesE2,D6,G2,F2);(a/b)fusedsilicasubstrate,(c/d)siliconsubstrate,(e/f)siliconcoatedwith100nmPt,(g/h)siliconcoatedwith 100nmAu;(i)SizedistributionofFe-NPson1m×1mfootprintonvarioussupportsupon720◦C/3hpretreatment.
Fig.8. CNTssynthesizedon3nmFe/10nmTathinfilmsonvarioussupportsat720◦C/1h,pretreatmentconditions:(a)fusedsilica120min/720◦C(sampleE3),(b)silicon
180min/720◦C(sampleD7),(c)100nmPtonsilicon110min/720◦C(sampleG3),(d)100nmAuonsilicon150min/720◦C(sampleF3).
Fig.9.TEMimageofCNTssynthesizedon3nmFe/10nmTathinfilmsonsilicon at720◦C,forgrowthphaseof30mininN2/C2H4(sampleD8).Thelengthofthe
scale-barintheimageis20nm.
ofrespectively110min(Fe/Ta/Pt/Ta,Fig.8c)or2.5h(Fe/Ta/Au/Ta, Fig.8d)resultsinVACNTs.
3.5. ElectricalconductivityofVACNTs
Theelectricalresistanceofthesampleswiththebestaligned VACNTs(synthesizedat720◦C/1hon3nmFe/10nmTaon vari-oussupports)wasmeasured.It consistsofaseriesresistanceof thousandsof VACNTsnanowire resistorsin parallel,the silicon substrateand(ifapplicable)theintermediateAu/TaorPt/Tafilm. Table3andFig.10showthatlowOhmicresistanceisfoundfor sam-plescontainingametallayerunderneaththeFe/Tacatalystsystem, incontrasttosampleswheretheFe/Taisdirectlydepositedonto silicon,whichshowahighresistanceandslightnonlinearityover theinvestigatedvoltagerange.
Table3
ElectricalresistanceofsampleswithVACNTsgrownat720◦C/1h.
Sampleconfiguration Resistance
(Ohm) standard deviation (Ohm) 3nmFe/10nmTa 36 19 3nmFe/10nmTa/100nmPt/10nmTa 1.7 1.0
3nmFe/10nmTa/100nmAu/10nmTa 1.4 0.9
Fig.10.CurrentvoltagecurvesforCNTssynthesizedon3nmFe/10nmTathin filmsonvarioussupportsat720◦C/1h,withpretreatmentconditions:onlysilicon 180min/720◦C(sampleD7;opencircles);100nmPtonsilicon110min/720◦C (sam-pleG3;closedtriangles);100nmAuonsilicon150min/720◦C(sampleF3;closed circles).
Theaverageconductivityofsampleswiththeseriesofinterfaces CNT/Fe/Ta/Au/Ta/SiorCNT/Fe/Ta/Pt/Ta/Siisatleast30timeshigher thanthatofsampleswiththeinterfaceseriesCNT/Fe/Ta/Si,withno significantdifferencebetweenAuandPtinterlayers.Themeasured lowerresistanceinournoblemetalmultilayerscomparedtothat ofBayeretal.isduetothediffusionbarrierformedbyPtandAuin ourcase,whichpreventsoxidationoftheTalayeronSi,whereasin Bayer’scaseforFe/Taonsiliconduringsynthesisattemperatures above600◦CtheTalayeroxidizedtotheinsulatingTa2O5[9].The
Fig.11.(a)FieldemissiontestforsampleswithVACNTssynthesizedon3nmFe/10nmTathinfilmsonvarioussupportsat720◦C/1h.Pretreatmentconditionsare110min
at720◦Cfor100nmPtonsilicon(sampleG3;opensymbols),150minat720◦Cfor100nmAuonsilicon(sampleF3;closedsymbols),and180min/720◦Cwithoutnoble
metallayeronsilicon(sampleD7;opentriangles).(b)Samedatasetasin(a),inaFowler-Nordheimplot,whereJiscurrentdensity(measuredcurrentdividedbyCNTsample area)andEisappliedelectricfield(appliedvoltagedividedbyelectrodegap).
useofPttoprotectTa/Si(orrather,tantalumsilicide/Si)Ohmic contactsfromoxidationathightemperatureshasbeendiscussed indetailbyGueyeetal.intheirworkonhigh-temperature Through-SiliconVias[26].
Themeasuredresistanceisanaverageofatleastthousandsof VACNTelectricallyaddressedinparallel.Averyroughestimateof theresistivityoftheCNTforest,basedonaCNTlengthofca.10m (seeFig.8d),acontactareaofca.20mm2andaresistanceofca. 1.5givesavalueofca.300cm.Anestimationoftheresistance ofanindividualnanotubeisvirtuallyimpossiblebecausethe den-sityofCNTscannotbeestimateddirectlyfromourSEMimages. Thereforeourdatacannotbecomparedwithreporteddata,which aremostlybasedonAFMprobing,amethodbymeansofwhicha singleCNT(orafewCNTs)canbeelectricallycharacterized[8]. 3.6. FieldelectronemissionfromVACNTs
FieldemissionfromaforestofVACNTstoaflatsteelelectrode wasperformedinN2gasat1atmospherewithanelectrodedistance of 500m. The characteristic field emission currents (Fig. 11a) showasteepincreasewhen thevoltagereaches560and620V respectively(theso-calles“onsetpotential”),inthecaseofVACNTs grownwiththeintroductionofAuandPtasmetalintermediate layer.Arapidincreaseincurrentatvoltagesabove800Vwasalso observedforsampleswithCNTsgrownwithouttheintroduction ofametallayerbetweentheTaandthesilicon,butinthiscase itwasnotpossibletoobtainreproducibleandstablecurrents.The observedcurrent-voltagecurvesinFig.11aarere-plottedinFig.11b inaccordancewithFowler-Nordheimtheory,fromwhichthe fol-lowingrelationbetweenfieldemissioncurrentandelectricfield strengthcanbederived:
I=Aa∅−1F2exp
−b∅1.5/F (1)whereAistheareaofemission,Fisthelocalelectricfieldstrength (whichwetakeequaltotheappliedfieldstrengthE),∅isthelocal workfunction (forwhich we havetaken thevalueof graphite, 4.6eV), and a and b are constants (a≡e3/8h
p=1.541434× 10−6AeVV−2,b≡4
3(2me) 1
2/ep=6.830888×109eVm−1whereme istheelectronmass,eistheelementarypositivecharge,andpis thePlank’sconstantdividedby2(p=hp/2)).Thelocal electri-calfieldFattipswithasmallradiusofcurvatureisoftenhigher thanthemacroscopicappliedelectricfield(Fm),theratioofthese twofieldsisdefinedasthefield-enhancementfactor=F/Fm.
Aswasreportedin literature[27],thepresenceofN2 in the experimentalenvironmentdoesnotinfluencethefieldemission current,thereforetheobtainediV-curveswereanalyzedusingthe standard Fowler-Nordheimtheoryforfield emissionin vacuum [28].Enhancementfactorsof971and2136arefoundforarraysof nanotubesgrownusingFe/Tacatalystdepositedonplatinumand gold,respectively.Nofieldemissionwasobservedforvoltagesup to1kVoltforthesamplewithoutnoblemetalinterlayer,although onemayarguethatthelast2measurementpointsinthecurvein Fig.11a,at980Vandabove,indicatetheonsetofemission. How-ever,becausethesepointsarequiteclosetothelimitoftheapplied voltagesource,weconsiderthemlessreliable.
TheobserveddifferenceinelectricalresistancefortheAuand Ptcasesof17–21%(butadmittedlywithalargemeasurementerror ofca.60%)isinlinewiththeobservedrelativelylargedifferencein theonsetpotentialforemission,ofca.60V,i.e.ca.10%ofthetotal appliedvoltage.Thisobservationwasreproducible,andwasfound forasignificantnumberofdifferentfieldemitterchipsofthesame typesasstudiedhere.ItshouldhoweverbenotedthattheCNT seriesresistanceisnottheonlyparameterthatdefinestheonset potential,alsothefieldemissioncharacteristicsoftheCNTs(tip shapeandtipdensity)contribute.Therefore,anotherexplanation fortheobservedbehaviormaybethattheCNTsonAuandPtare slighlydifferent,possiblyintermsofCNTdensity,whichisdifficult toobserveintheSEMimagesaspresentedinthisstudy.Anemitter densitydifferenceleadstodifferentfieldemittershielding,which isexpressedinthefieldenhancementfactor[29].
4. Conclusion
Inthisworkweanalyzedtheinfluenceofgrowthand pretreat-mentparametersonthealignmentofMWCNTs,usingFecatalyst onaTaunderlayer,andethyleneasthecarbonsource.Sizeand densityofcatalystparticlesarethedominantfactorsinthe forma-tionofverticallyalignedCNTs,andthesedependstronglyontime andtemperatureofareductionpretreatmentstepandthe origi-nalFelayerthickness:increasingpretreatmenttimefrom1hto3h affectshasalargereffectonparticlesizedistributionthan increas-ingpretreatmenttemperaturefrom680◦Cto720◦C,orchanging thethicknessofthecatalystlayerfrom2to6nm.Introductionofan AuorPtlayerbetweensiliconsubstrateandTalayersignificantly improvestheoverallconductivityofthesample,leadingto VAC-NTswithalowOhmicresistance,andfieldemissioncharacteristics withfieldenhancementfactorsabove2000.
Acknowledgements
The research for this paper is financially supported by the Netherlands’Organizationfor ScientificResearch(NWO),within theNWO-Nanoprogram,projectno.1141.WethankBertGeerdink oftheCatalyticProcessesandMaterialsgroupatouruniversityfor assistancewiththeCNTdepositionsetup.
References
[1]P.Serp,M.Corrias,P.Kalck,Carbonnanotubesandnanofibersincatalysis, Appl.Catal.A—Gen.253(2003)337–358.
[2]R.S.Ruoff,D.C.Lorents,Mechanicalandthermalpropertiesofcarbon nanotubes,Carbon33(1995)925–930.
[3]P.Avouris,Z.H.Chen,V.Perebeinos,Carbon-basedelectronics,Nat. Nanotechnol.2(2007)605–615.
[4]Z.W.Pan,S.S.Xie,B.H.Chang,C.Y.Wang,L.Lu,W.Liu,W.Y.Zhou,W.Z.Li,L.X. Qian,Verylongcarbonnanotubes,Nature394(1998)631–632.
[5]Z.P.Huang,J.W.Xu,Z.F.Ren,J.H.Wang,M.P.Siegal,P.N.Provencio,Growthof highlyorientedcarbonnanotubesbyplasma-enhancedhotfilamentchemical vapordeposition,Appl.Phys.Lett.3(1998)3845–3847.
[6]N.Nagaraju,A.Fonseca,Z.Konya,J.B.Nagy,Aluminaandsilicasupported metalcatalystsfortheproductionofcarbonnanotubes,J.Mol.Catal. A—Chem.181(2002)57–62.
[7]C.J.Lee,J.Park,J.A.Yu,Catalysteffectoncarbonnanotubessynthesizedby thermalchemicalvapordeposition,Chem.Phys.Lett.360(2002)250–255.
[8]G.D.Nessim,M.Seita,K.P.O’Brien,A.J.Hart,R.K.Bonaparte,R.R.Mitchell,C.V. Thompson,Lowtemperaturesynthesisofverticallyalignedcarbonnanotubes withelectricalcontacttometallicsubstratesenabledbythermal
decompositionofthecarbonfeedstock,NanoLett.9(2009)3398–3405.
[9]B.C.Bayer,S.Hofmann,C.Castellarin-Cudia,R.Blume,C.Baehtz,S.
Esconjauregui,C.T.Wirth,R.A.Oliver,C.Ducati,A.Knop-Gericke,R.Schlögl,A. Goldoni,C.Cepek,J.Robertson,Support-catalyst-gasinteractionsduring carbonnanotubegrowthonmetallicTafilms,J.Phys.Chem.C115(2011) 4359–4369.
[10]W.A.deHeer,A.Chatelaine,D.Ugarte,Acarbonnanotubefield-emission electronsource,Science270(1995)1179–1180.
[11]J.-M.Bonard,J.-P.Salvetat,T.Stöckli,L.Forró,A.Châtelain,Fieldemission fromcarbonnanotubes:perspectivesforapplicationsandcluestothe emissionmechanism,Appl.Phys.A69(1999)245–254.
[12]W.Zhu,C.Bower,O.Zhou,G.Kochanski,S.Jin,Largecurrentdensityfrom carbonnanotubefieldemitters,Appl.Phys.Lett.75(1999)873–875.
[13]R.Sharma,E.Moore,P.Rez,M.M.J.Treacy,Site-specificfabricationofFe particlesforcarbonnanotubegrowth,NanoLett.9(2009)689–694.
[14]G.D.Nessim,A.J.Hart,J.S.Kim,D.Acquaviva,J.H.Oh,C.D.Morgan,M.Seita,J.S. Leib,C.V.Thompson,Tuningofvertically-alignedcarbonnanotubediameter andarealdensitythroughcatalystpre-treatment,NanoLett.8(2008) 3587–3593.
[15]B.C.Bayer,M.Fouquet,R.Blume,C.T.Wirth,R.S.Weatherup,K.Ogata,A. Knop-Gericke,R.Schlögl,S.Hofmann,J.Robertson,Co-catalyticsolid-state reductionappliedtocarbonnanotubegrowth,J.Phys.Chem.C116(2012) 1107–1113.
[16]E.Teblum,Y.Gofer,C.L.Pint,G.D.Nessim,Roleofcatalystoxidationstatein thegrowthofverticallyalignedcarbonnanotubes,J.Phys.Chem.C116 (2012)24522–24528.
[17]J.B.In,C.P.Grigoropoulos,A.A.Chernov,A.Noy,Growthkineticsofvertically alignedcarbonnanotubearraysincleanoxygen-freeconditions,ACSNano5 (2011)9602–9610.
[18]D.E.Shleifman,D.Shaltiel,I.T.Steinberger,Thermallystimulatedhydrogen desorptionfromzirconiumandtantalum,J.AlloysCompd.223(1995)81–86.
[19]H.Uchida,E.Fromm,Kineticsofhydrogenabsorptionbytitanium,tantalum, tungsten,ironandpalladiumfilmswithandwithoutoxygenpreabsorptionat 300K,J.Less-CommonMet.95(1983)139–146.
[20]C.E.KrillIII,L.Helfen,D.Michels,H.Natter,A.Fitch,O.Masson,R.Birringer, Size-dependentgrain-growthkineticsobservedinnanocrystallineFe,Phys. Rev.Lett.86(2001)842–845.
[21]R.M.Tiggelaar,D.B.Thakur,H.Nair,L.Lefferts,K.Seshan,J.G.E.Gardeniers, Influenceofthinfilmnickelpretreatmentoncatalyticthermalchemicalvapor depositionofcarbonnanofibers,ThinSolidFilms534(2013)341–347.
[22]H.T.Ng,B.Chen,J.E.Koehne,A.M.Cassell,J.Li,J.Han,M.Meyyappan,Growth ofcarbonnanotubes:acombinatorialmethodtostudytheeffectofcatalysts andunderlayers,J.Phys.Chem.B107(2003)8484–8489.
[23]G.D.Nessim,D.Acquaviva,M.Seita,K.P.O’Brien,C.V.Thompson,Thecritical roleoftheunderlayermaterialandthicknessingrowingverticallyaligned carbonnanotubesandnanofibersonmetallicsubstratesbychemicalvapor deposition,Adv.Funct.Mater.20(2010)1306–1312.
[24]D.P.Burt,W.M.Whyte,J.M.R.Weaver,A.Glidle,J.P.Edgeworth,J.V. Macpherson,P.S.Dobson,Electrochemistryatcarbonnanotubeforests: sidewallsandclosedendsallowfastelectrontransfer,J.Phys.Chem.C113 (2009)15133–15139.
[25]D.B.Thakur,R.M.Tiggelaar,J.G.E.Gardeniers,L.Lefferts,K.Seshan,Growthof carbonnanofibercoatingsonnickelthinfilmsonfusedsilicabycatalytic thermalchemicalvapordeposition:ontheuseoftitanium,titanium–tungsten andtantalumasadhesionlayers,Surf.Coat.Technol.203(2009)3435–3441.
[26]R.Gueye,T.Akiyama,D.Briand,N.F.deRooij,Fabricationandformationof Ta/Pt-Siohmiccontactsappliedtohigh-temperatureThroughSiliconVias (TSVs),Sens.ActuatorsA191(2013)45–50.
[27]C.Kim,Y.S.Choi,S.M.Lee,J.T.Park,B.Kim,Y.H.Lee,Theeffectofgas adsorptiononthefieldemissionmechanismofcarbonnanotubes,J.Am. Chem.Soc.124(2002)9906–9911.
[28]R.H.Fowler,L.Nordheim,Electronemissioninintenseelectricfields,Proc.R. Soc.A119(1928)173–181.
[29]J.R.Harris,K.L.Jensen,D.A.Shiffler,J.J.Petillo,Shieldinginungatedfield emitterarrays,Appl.Phys.Lett.106(2015)201603.
