Carbon nanotube thin film transistors by droplet electrophoresis

(1)

Publisher’s version / Version de l'éditeur:

Materials Today Communications, 10, pp. 72-79, 2017-01-17

READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE. https://nrc-publications.canada.ca/eng/copyright

Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la

première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez pas à les repérer, communiquez avec nous à PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca.

Questions? Contact the NRC Publications Archive team at

PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca. If you wish to email the authors directly, please see the first page of the publication for their contact information.

NRC Publications Archive

Archives des publications du CNRC

This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. / La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version acceptée du manuscrit ou la version de l’éditeur.

For the publisher’s version, please access the DOI link below./ Pour consulter la version de l’éditeur, utilisez le lien DOI ci-dessous.

https://doi.org/10.1016/j.mtcomm.2017.01.002

Access and use of this website and the material on it are subject to the Terms and Conditions set forth at

Carbon nanotube thin film transistors by droplet electrophoresis

Lefebvre, J.; Ding, J.

https://publications-cnrc.canada.ca/fra/droits

L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.

NRC Publications Record / Notice d'Archives des publications de CNRC:

https://nrc-publications.canada.ca/eng/view/object/?id=6f84f2de-8615-457c-abcc-686d3e0f9405

https://publications-cnrc.canada.ca/fra/voir/objet/?id=6f84f2de-8615-457c-abcc-686d3e0f9405

(2)

ContentslistsavailableatScienceDirect

Materials

Today

Communications

jo u r n al h om ep age :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

Carbon

nanotube

thin

film

transistors

by

droplet

electrophoresis

J. Lefebvre

∗

_,

_J.

_Ding

NationalResearchCouncil,1200MontrealRoad,Ottawa,ON,K1A0R6,Canada

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received1November2016

Receivedinrevisedform5January2017 Accepted6January2017

Availableonline17January2017

a

b

s

t

r

a

c

t

Spontaneouslychargedaerosoldroplets,eachcontainingafewsingle-walledcarbonnanotubeswrapped inconjugatedpolymerswereprecipitatedonatargetsubstrateusingelectrostaticforces.Nanotube net-workswereassembledonavarietyofdielectricsurfacesincludingpolymerswithlowsurfaceenergy: Goodtransistorperformancewasachievedinallcases.Underproperregimeofelectrostaticandflow fields,patternswereproducedinamasklessfashionandfeaturesizesbelow150␮mweredemonstrated. CrownCopyright©2017PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCC BY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Withits of ease of fabrication combined with performance, thinfilmsofrandomlyorganizedcarbonnanotubes(mainly single-walled−SWCNT-)haveattractedalotofinterestinconductiveand transistorapplications[1–3].Themethodstoobtainsuchfilmsare moreorlesssophisticateddependingontheendgoal:Fromdry transferandsoakingtoself-assembly;[4–9]Orfromsprayingto inkjetoraerosoljetprinting[10–19].ForSWCNTbasedtransistors inprintedelectronics,requirementsonthefilmmorphologyare quitestringentifmobilityinexcessof10cm2_/Vs_is_to_be_achieved

whilemaintaining a low Off-state current.Moreover, a suitable depositionmethodshouldallowforpatternswith10–100␮m res-olutionwithgooduniformityacrossmacroscopicareas,aswellas scalabilityandthroughputasprimaryconsiderations.Thepresent workwasmotivatedbythelimitednumberofprintingmethods capableofachievingtheserequirements.

In order to explore new territories, we asked the question whetheritwouldbepossibletoefficientlydelivercarbon nano-tubesafewatatimetoatargetsurface.Intheformationofarandom network,thissituationwouldappearidealsinceindividualdroplets containingoneorafewnanotubeswouldevaporateindependently of one another: Bundling/coffee staining would be minimized. Commonprintingmethodsoperateaway fromthisregimewith concentratedcarbonnanotubesolutionsbeingused.Those meth-odsrely onefficientspread of material toachieve desired film morphologies.

Aerosolgenerationprovidesunexploredopportunitiesfor car-bonnanotubenetworkandthinfilmassembly.Withsub-micron

∗ Correspondingauthor.

E-mailaddress:Jacques.lefebvre@nrc-cnrc.gc.ca(J.Lefebvre).

droplet generators commercially available, the few nanotube regimeis reachedwithreasonable carbonnanotube concentra-tions.Forexample,with1–2␮mdiameterdroplets(1–4fL),asingle nanotubeperdropletrequiresconcentrationsoforderof1␮g/mL, amodest10–50xdilutioncompared todispersions usedbyour group for inkjet printing [19]. The highthroughput of aerosol generators, >L/s,should besufficient tocover cm2 _areas _with

10–100nanotubes/␮m2_in_seconds._In_addition_to_accessing_a_new

dilutionregime,adifferentflowregimefordropletdeliverycanbe explored:Laminar.Incontrastwithaerosoljetprinting,droplets canbecarriedatvelocitiesoforderofcm/s,morethanhundred timesslower.Inthisregime,externalforcesneedtoacton micro-droplets topromotetheyield of depositedmaterial;Otherwise dropletswillbeefficientlycarriedawayfromsurfaces.Electrostatic forcesrepresentaconvenientchoice:Thefieldcanactondroplets tocountertheactionofthelaminarflowandmovethemtoatarget substrate.

In this paper, we demonstrate that sc-SWCNT thin film transistorswithgoodperformancecanbeassembled from elec-trophoreticallytransportedmicron-sizedroplets,eachcontaining oneorafewcarbonnanotubes.Nanotubenetworkswere assem-bledonavarietyofdielectricsurfacesincludingpolymerswithlow surfaceenergywherehysteresis-freetransportismeasured. We alsodemonstratethat underpropercontrolofthegasflow and electricfield,patterneddepositionwasachievedwithoutuseofa mask,withfeaturesizebelow150␮m.

2. Experimental

2.1. Single-walledcarbonnanotubedispersion

The sc-SWCNTs dispersions were prepared from enriched plasma sc-SWCNTs (diameter around 1.3nm), using a hybrid

http://dx.doi.org/10.1016/j.mtcomm.2017.01.002

(3)

process previouslyreported byourgroup [20,21]. In brief,raw material from Raymor Nanotechnologies (RN-000) was mixed withpoly(9,9-didodecylfluorene),PFDDintolueneatapolymerto SWCNTratioof1.0/1.0andananotubeconcentrationof0.8mg/mL. Themixturewashornsonicatedfor30min(Bransonsonifier250) followed by 30min centrifugation (relative centrifuge force of ≈18,000g). Thisstep wasrepeated 5 timesusing the sediment fromthepreviouscentrifugationandthefivesupernatantsolutions werecombinedandaddedwithsilicagelatausageof1mg/mL. Themixturewassonicatedinabathsonicator(Branson2510)for 40min,followedbystandingfor3h,andthen30min centrifuga-tion(relativecentrifugeforceof∼18,000g).Thesupernatantwas filteredusingaTeflonmembranewith0.2␮mporesize(Sartorus StedimBiotech)andrinsedwithtoluenetoremoveexcess poly-mer.The collectedsc-SWCNTswere dispersedin toluene using bathsonicationfor5–10minataconcentrationof0.48mg/mLand apolymertonanotubeweightratioof2.4.The>99.9%purity solu-tionisdominatedbySWCNTchiralitiesaround(10,9)[20,22,23]. TransistorsmadefromsoakingonSiO2/Sisubstratesshowhigh

mobilitycombinedwithhighOn/Offratio(averagenanotubelength is1.3␮m[20,24]).Priortoaerosoldeposition,dilutesolutions start-ingfroma 480␮g/mL werepreparedin20–25mLquantitiesby addingtolueneandsonicatingfor60mininaBranson1250. 2.2. Carbonnanotubedeposition

Theformationofananotubenetworkisbasedonusing electro-staticforcestoattractchargedaerosoldropletstoatargetsubstrate. Anaerosolgenerator,241PGfromSonaerInc.,breaksananotube containingsolutionintomicron-sizedroplets(calculated hydrody-namicdiameterof1.7␮mforwaterand1.3␮mfortoluene[25,26]. MeasuredvaluesprovidedbySonaerforwaterare:Massmedian dia.:3.7␮m;Sautermeandia.:2.9␮m;Volumemeandia.:1.8␮m; Areameandia.:1.4␮m;Numbermeandia.:1.2␮m).Dropletsare carried to a deposition chamber using a low flow rate N2 gas

(50–150sccmin¼ inchTeflontubing).Theyenterthechamber throughabrassnozzlesettoahighvoltage(StandfordResearch PS350forDCvoltagesandfunctiongenerator/amplifiercombofor ACvoltages)withrespecttothesampleholder.Voltagesinthe 100–2500V rangehavebeenapplied.Thedistancebetweenthe nozzleandthesampleisabout2mmfortheresultspresentedhere butlargergapshavebeenused.Thechambersitsonahotplateheld at75◦_C;_This_temperature_is_sufficiently_high_to_quickly_evaporate

dropletsandsufficientlylowtoavoidthermalgradientsand con-vection.Gases(N2andtoluene)areexhaustedthroughaN2cold

traptocollectresidualparticlesbeforebeingventedtoahood. 2.3. Transistorfabrication

SubstrateswereSiO2/SiwiththedopedSiactingasacommon

backgateelectrodeandSiO2asthegatedielectric(100nm).Wafers

werelimitedlycleaned,withonlya30minexposuretoUV-O3.

Sam-plesproducedintheaerosolprocesswereinitiallyinsulating.Even asmallamountofexcesspolymer(2.5:1weightratiowithrespect toSWCNT)wassufficienttopreventconductionandarinsingstep wasrequired.Soonafterthedepositionstep,sampleswerebathed intoluenefor5minfollowedbyisopropanolanddriedwithN2

gas.Opticalabsorptionobtainedfromthoroughlyrinsed disper-sionshavepolymer:nanotubeweightratioof1orslightlyless.A 5–10minbakeat150◦_C_was_also_performed,_for_consistency_with

oursoakingprocess.Source-draincontacts(Ti/Au)wereevaporated througha shadowmask(Tecan,UK)toproducetransistorswith dimensionsfrom12×220,20×220and 40×220␮m2.Atthose lengthscales,currentscaleswithchannellengthandwasnot lim-itedbycontactresistance.AphotoofthesampleusedinFig.3is

showninFig.S5.Samplesfabricatedonpolymerdielectricshave beendescribedinRef.[27].

2.4. Thinfilmcharacterization(optical,SEM,Raman,electrical) 2.4.1. Optical

Macro-photographypicturestakenunderdifferentillumination conditions arepresented throughoutthis manuscript. Since the filmsareultrathin,specialattentionwasgiventoillumination con-ditions.Apowerfulfibreilluminatorsourceworkedbesttoreveal thefilmthroughscatteredlight.Insomecases,fluorescencefrom thepoly-fluorenewasfounduseful,aswellaslightinterference seenasangledependentcolorationchangesonthinoxidelayers. Severalexamplesarepresented in Figs.S4–S8.Allphotos were takenpriortotherinsingstepafterwhichthefilmswereonlyvisible toawell-trainedeye.

2.4.2. SEM

AHitachi4700electronmicroscopewasoperatedunder opti-mum conditions to observeSWCNTs in charge contrast mode: 0.5–1kVaccelerationvoltageand4mmworkingdistance. 2.4.3. Ramanscattering

Bothcommercialandhome-builtRamansystemsoperatingat 532nmexcitationwavelengthwereusedforthiswork.This wave-lengthisresonant withhigheropticaltransitions insc-SWCNTs withdiametersaround1.3nm[23].

2.4.4. Electricaltesting

Transistor transfer characteristics (current versus gate bias) wereacquired in airambient usinga home-builtprobe station (voltage source,ammeter, functiongenerator and oscilloscope). Thesource-drainbiaswassetto0.5Vandthecurrent(Keithley 6517A)outputfedtotheinputofadigital oscilloscope(Agilent DSO6012A).A110mHzsawtoothvoltagefromafunctiongenerator (Agilent33220A)wasappliedtothebacksideofthesample(Si sub-stratecoveredwith100nmthermaloxideorpolymerdielectric). Thisvoltagewasalsofedasinputtotheoscilloscope.Holemobility wascalculatedfromtheslopeofthetransfercurvesusingthe par-allelplatecapacitormodel.Deviceconfigurationanddimensions aregiveninFig.S6(e).

3. Results

3.1. Electrostaticprecipitation

Fig. 1 summarizes the main findings. The setup,illustrated in Fig.1(a)is in many ways similarto otheraerosol processes (see Fig.S1forphotos ofthe setup):a particlegeneratorturns a carbon nanotube dispersion into droplets. Here, we used a highfrequency(2.4MHz)ultrasonicgeneratortoproduce1–2␮m diameterdroplets.Generateddropletsweretransportedtoa depo-sitionchamber usingN2 ascarriergas.A lowflowrateensures

dropletswerecarriedinalaminarfashion,withminimumdroplet coalescenceexpected. Tubingwasalso keptveryshort tolimit dropletevaporation:weestimatedropletsshouldtakeonlyafew secondstotravel thetotal distancebetweenthegeneratorand thesubstrate.Atthedepositionchamber,tubingisconnectedto a nozzle withdifferent configurations of aperture (single hole, manyholesorslitaperture).Typicaldimensionsforslitopenings were≈0.5×20mm2_._Substrates_were_heated_sufficiently_to_quickly

evaporatethedroplets.Weexpectthatdropletsevaporatemainly oncetheyreachthesubstratesurface,sincethetimetocoverthe smalldistancebetweenthenozzleandthesubstrateisveryshort (≪1s).

(4)

Fig.1. Carbonnanotubesthinfilmbydropletprecipitation.(a)Schematicillustrationofthedepositionsystemandprocess.Usingacombinationofnozzle(yellow-orange) aperture,electrostaticfieldandsampletranslation,patterneddepositionisachieved.SeeFig.S1forseveralphotographsoftheexperimentalsetup.(b)Scanningelectron microscopyimageofanetworkofSWCNTs.(c)Transistorcharacteristicforasc-SWCNTthinfilmonSiO2/Si(1000nmoxide).Channellengthandwidthare12and220␮m, respectively.Mobilityiscalculatedusingthestandardplatecapacitormodel.7␮A/Visequivalentto0.03␮A/␮mforthesedevices.

The main attributes of the method is summarized in three points:1–2␮mdiameterchargeddroplets,eachcontainingafew carbonnanotubes;Alowflowratetocarrydropletsinalaminar fashion;Anelectrostaticfieldtoprecipitatedropletsonasubstrate. Thesubstratecanbeaconductororaninsulator(thicknessesupto 150␮mhavebeentested).Thenozzlecanbeaconductorconnected toavoltagesourceoratribo-staticallychargedinsulator(for exam-pleTeflonrubbedwithanaluminumfoil).Thispaperfocuseson solutionsofpolymerwrappedsc-SWCNTsintoluenebutunsorted SWCNTsinchlorobenzenehavealsobeentested,withqualitatively similarresultswithregardstofilmmorphology.

Fig.1(b)showsanSEM(scanningelectronmicroscopy)example ofarandomnetworkof SWCNTsonaSiO2/Sisubstrate.In

sev-eralways,filmsresembleothersmadebysoakingordropcasting andconsistofasparseassemblyof≈1␮mlongnanotubes (aver-agenanotubelengthis1.3␮m[20])witharealdensitiesoforder 10–50␮m−2.Incontrastwithsoaking,smallareaswith“missing” materialareseeninthenetworkatlowercoverages;nanotubes aremostoftennotstraightbutareslightlycurved,aconsequence ofthenanotubeconformationinsidesphericaldroplets.Although SEM provides a good indication that the network morphology issuitable forthin filmtransistors,this isconfirmedwith elec-tronicperformancedata.Fig.1(c)presentsatransfercurvefrom thebest films achieved thus far: the hole field effect mobility isquiterespectableat >10cm2_/Vs _and_On/Off_ratio_in _excess_of

10,000.Thesevaluesarelowerthanfilmsobtainedinourgroup bysoakingorinkjetprintingwheremobilityinexcessof30cm2_/Vs

hasbeenachieved[19,20,24].Recentworkonnetworksof poly-merwrappedsc-SWCNTsobtainedbyinkjetprintingalsoshowed goodmobility(closeto10cm2_/Vs)_for_both_electrons_and_holes_in

atopgatedtransistor[18].Here,severalparametershaveyetto beoptimizedandmayaccountforthelowerperformance com-paredtootherdepositions methodsexploredinourgroup.The differencemayalsobeintrinsic tothefilmmorphology, where straighternanotubesareseen forsoaking andinkjet,with con-sequentlyfewerinter-nanotubejunctionstobridgesource-drain electrodes.Improvementsinthefilmmorphologyisexpectedwith

largerdropletswhichcanbecontrolledwiththechoiceofaerosol generatororanoptimizednanotubedispersion.

Intermsofthroughput,centimetersquareareasaretypically coveredinafewminutes,evenatthelowestconcentrations pre-sentedhere.At1␮g/mL,ittakes5␮Lofsolutiontocoveracm2 with100Nanotubes/␮m2.Weestimatethat1–2mLisaerosolized perhourofoperationusing241PGgenerator.With1minbeing typicalofa depositionrun,≈10–20␮Listherefore requiredfor one run. The yield from this calculation is reasonably high at ≈25–50%.Later,wewillshowthatcollecteddropletsare sponta-neouslychargedandthisyieldsuggeststhattheyrepresentagood fractionofgenerateddroplets.Ifthenumberofneutraldropletswas fairlyhigh,theyieldofdepositedmaterialwouldbemuchlower, unlessdielectrophoreticforceswereatplay.Usingexternalmeans tointentionallychargeddropletsshouldmaximizeprocessyield andthroughput.

InadditiontoproducingsuitablenanotubenetworksonSiO2

dielectric, good adhesion was found even on surfaces with verylow surface energy.For example,bottom gateair exposed thin film transistors have been fabricated on several polymer dielectrics,includinghydrophobicTeflon-AFandpoly(vinyl phe-nol)/poly(methylsilsesquioxane)blend(PVP-pMSSQ)wheregate hysteresiswasgreatlysuppressed[27].Resultsfromsixpolymer dielectricsincludingTeflon-AFandPVP-pMSSQareshownas sup-portinginformation (Fig.S2). Wespeculate that“oneatatime” deliveryofnanotubesinthisaerosolmethodpreventssignificant reorganizationofthenetworkduringandafteritsformation.Once formed,thecollectivearrangementofnanotubescannotbeeasily disrupted,eveninarinsingstep.

Intermsofuniformityovermacroscopicscales,mappingof tran-sistorperformanceacrossan entiresample (Fig.S3) shows +/− 10% variabilityonthemobility withinmm2 areas. Thisnumber islikelylimitedonlybytheabilitytoproperlyengineerand con-trolthedeliveryofaerosoltothesurface.Inanycase,thispaper wishestohighlighttheuniqueattributesofthisaerosolprocess: depositioninafewnanotuberegime,electrostaticprecipitationof droplets,andpatternedmasklessdepositionunderlaminarflow

(5)

andelectrostaticfields.Moreover,themethodaddressesan impor-tantchallengewithprintingcarbonnanotubeinks:adhesionon polymersurfaces.Thisisoftenmitigatedbytreatingthesurface usingaplasma[18,19,28],a processwhichinducestrapsinthe dielectricmaterialand consequentlydegradestransistor perfor-mance.Theaerosolprocesspresentedhereallowsfordeposition onanysurfaces,withoutanyharshsurfacepretreatment.

Fig.2highlightsthecriticalroleofelectrostaticintheassembly ofcarbonnanotubenetworksbyaerosol.Forthisandthemajorityof experimentspresentedhere,aslitshapenozzlewasused(Fig.S1).

Fig.2(a)(seeFigs.S4–S5forthecompletedataset)showsaseriesof filmsdepositedunderincreasinglyhighervoltages.Fig.2(b-g)show correspondingSEMimages.Under theactionof anelectrostatic field,an initiallysparsenetworkof carbonnanotubesis substi-tutedwithadensematofhighlyconnectednanotubes.Inabsence ofappliedvoltage(seeFig.S4(a)),alimitedamountofmaterialis found.Evenattheverylowflowrateofcarriergas(50sccmN2,

whichcorrespondstoalinearvelocityaround1cm/sforour geom-etry.sccmstandsforstandardcubiccentimeterperminute),the micronsizedropletsarecarriedawayfromregionsdirectlyunder theslit-shapenozzleandonlyasmallamountendsuponthe sub-strate.Uponapplyingamoderatevoltage(±100VinFig.2(b–c)), nanotubematerialstartstoconcentrateunderthenozzleandthe amountofdepositedmaterialissignificantlyincreasedcompared to0V.Twolobesofmaterialareapparentandtheyextendafew millimetersonbothsidesofthenozzlelongaxis.Asthevoltageis increasedto±400V,thesidelobesgraduallydisappearand mate-rialconcentratesfurtherandfurtherunderthenozzle.InFig.2(f–g), ±1600Vwereappliedand narrowstripesofmaterialarebeing deposited.Qualitatively,thepolarityoftheappliedvoltagedoesnot significantlyinfluencematerialdeposition.Itisworthemphasizing thatbythenatureofthedroplets,specificallytheirsizeandthe numberofnanotubestheycontain,wehaveobservednoevidence ofcoffeestainatthemacro-scale.

Inordertogainagreaterphysicalinsightoftheprocess,wehave performedRamanscatteringmeasurementsonthethreesamples inFig.S4(a–c).InFig.2(h),theG-bandintensityisplottedasa func-tionofappliedvoltage,withintensitynormalizedtoaccountforthe varyingdurationbetweenthethreesampledepositiontimes.The datashowsalinearincreaseofdepositedmaterialupto≈700V followedbyasub-linearincreaseandsaturationathighvoltages. Averygoodsymmetryisobservedbetweenpositiveandnegative voltages,butwithaslightasymmetryfavoringhighercoveragesat positivevoltages.Usingthe0and2000Vdata,closetoa hundred-foldenhancementhasbeenestimated.

3.2. Fewnanotubesperdroplet

The initialpremise of this work wasthat the best network morphologyshouldbeachievedinaregimewherenanotubesare depositedone byone.Using reasonable parametersthat reflect oursc-SWCNTmaterial(seeSupportingInformationunder“ONE NANOTUBEperDROPLET”),wehavecalculatedacross-over con-centration where each droplet should contain on average one nanotube or less.We then proceeded toassemble carbon nan-otubenetworksusingfourdilutions,two belowand twoabove 3␮g/mL,thecross-overvalue.Thedurationofthedepositionwas adjustedtoobtainnominallythesameamountofmaterial.SEM micrographsarepresentedinFig.3(a–d).Startingwiththehighest dilution,0.9␮g/mL,wefindanetworkwithexcellentuniformity withindividualnanotubesbeingslightlycurled.Asthe concentra-tionincreases,clumpingstartstobecomesignificantwithseveral openingsexposingtheunderlyingSiO2/Sisubstrate.Foreach

dilu-tion,imageswerealsotakenawayfromthenozzleareatorevealthe morphologyofindividualdroplets.Atthehighestconcentration, 14.1␮g/mL, micro-coffeestainsformwithatleast10–12

nano-tubesclumpedtogether.Astheconcentrationisreduced,individual dropletsappeartocontainfewerandfewernanotubesandlittle evidenceofmicro-coffeestainscanbefound.

Ramanmeasurementsperformedonthissamplesupportour SEMobservations.Inordertorevealtheeffectofdilution,the inten-sityfrom24neighboringlocationswasaveraged.Fig.3(e)showsthe averageintensityandthestandarddeviation(asanerrorbar)forthe fourconcentrationsinthisstudy(twodepositiontimeswerelooked at toconfirmtheexpectedlinearity). Wefirst observethat the amountofmaterialisessentiallyflatforthefirsttwoconcentrations andthenincreasestoalmostdoubleat14.1␮g/mL.Thisresult indi-catesstrongerelectrostaticforceswhenmorenanotubesoccupya singledroplet.Atthesametime,thestandarddeviationmorethan doublesindicatingthatuniformityworsensdramatically, consis-tentwithSEMresults.BothSEMandRamandatademonstratethat below3␮g/mL,goodnetworkmorphologyisobtained:Thatregime correspondstooneorfewcarbonnanotubesbeingdepositedfor eachdropletreachingthesubstrate.Furthersupportforoperating inthisdilutionregimeisgainedfromlookingattransistor perfor-mance.

Fig.3(f)showstransfercharacteristicsobtainedfromthefour networksinFig.3(a–d).Atthelowestconcentration,good tran-sistorperformanceisfoundwithgoodcurrent/mobilitycombined withgoodOn/Offratios:3.1cm2_/Vs_and_>103_,_{respectively.}_Higher

numbershavebeenfoundonotherdeviceswithmobilityinthe 10–20cm2_/Vs_range_with_On/Off_>104 ₍_Fig.₁_(c))._For_the_sample

setinFig.3,bettermobilitynumberswouldhavebeenobtained withaslightlylongerdepositiontime.

TheexperimentinFig.3highlightsthecriticalimportanceof concentrationonnetworkmorphologyandtransistorperformance. Fromthehighestdilution,increasingnanotubeconcentrationdoes not impact performance so strongly: going from 0.9␮g/mL to 2.2␮g/mLleadstosomereductioninmobilitywithsimilarOn/Off ratio.Inthosenetworks,nanotubesarenotsignificantlybundled; otherwisemobilityinexcessof10cm2_/Vs_with_good_On/Off_ratio

wouldnotbepossible(asshowninFig.1(c)).Asthe concentra-tionincreasesfurther,thenanotubenetworkbecomespatchierand consequentlycontainsadecreasingnumberofactivepathwaysfor current flow.Thegate isstillquiteeffectiveatmodulating cur-rent:Thenetworkisstill largelyinterconnectedwithindividual SWCNTsandbundlespresentinthenetworkdonotproduce perco-latingpathsbythemselves.Astheconcentrationincreasesfurther to5.6and14.1␮g/mL,transistorperformancedegrades substan-tiallywithaweakcurrentmodulationseeninthegatesweeps. Patchescontaininganincreasingnumberofnanotubesbecomethe dominantpathfor conduction.SEMin Fig.3(d)shows that sig-nificantpatch overlapexists and individualnanotube junctions appear insmallnumber.Transportinnanotubescoalesced dur-ingdropletevaporation(micro-coffeestain)issimilartodevices madeofbundleswherereducedgateactionisobserved[29,30]. WeshouldhighlightthatevenwithtwicetheamountofSWCNTs, the14.1␮g/mLfilmdoesnotmatchthecurrentcarryingcapacityof the0.9␮g/mLfilm,whichisonaveragesparserbutpresentsmuch greateruniformity.Thesesampleswillbecharacterizedfurtherto obtaindistributionsofbundlesizeusinganAFMbasedprotocol

[31].

3.3. Masklesspatterning

Wewouldliketoaddressanessentialrequirementfor depo-sition methods in printed electronics: patternability. Aerosol processesare verymuch compatiblewithshadowmasking and oneexampleisshowninFig.S8.Thesmallsizeofdroplets com-binedwithefficientsolventevaporationproducespatternswith littlematerialbeingdrawn underthemask.Withcurrent tech-nologyinmaskfabrication,featuresizesdownto10␮marefairly

(6)

Fig.2. Carbonnanotubethinfilmsbydropletprecipitation.(a)Macro-photographyofSiO2/Sisubstrateswithcarbonnanotubethinfilmsdepositedunderincreasingvoltages

usinganozzlewithaslitaperture.Betweeneachcondition,thesampleistranslatedlaterallytoobtainnon-overlappingdeposits.Samplesizeisabout22×28mm2_._Colored

circlescorrespondtoapproximatelocationforelectronmicroscopyimagesin(b)to(g).(b)-(g)Scanningelectronmicroscopyimagestakenfromfilmsdepositedatdifferent voltages.AcompletesetofimagesisshowninFig.S4.Solutionconcentrationwas3␮g/mL.anddepositiontime,60s.(h)AmountofdepositedSWCNTdeterminedbyG-band Ramanintensityasafunctionofappliedvoltage.Macro-photographyofthesamplesisshowninFig.S5.

realistic.Althoughthiscanproveusefulinproofofconceptstudies, itistechnologicallymoreviabletohaveaprocesscapableof pro-ducingpatternsinamasklessfashion.Fig.4showsthreeexamples ofhowthisispossiblebyengineeringboththegasflowandthe electrostaticfield.

Fig.4(a)demonstrates howtightlytheelectrostatic fieldcan focusthematerial.Optically,stripesdepositedatthehighest volt-agesappearverycrispwithdimensionswellunderamillimeter.

Fig.4(b)presentsaRamanintensityprofiletakenacrossastripeand revealsaFWHM(Full-widthathalf-maximum)of100␮m.Even withtheoverspreadofmaterialbeyond200␮m,wecansafelysay thattransistorswithchannelwidthbelow150␮mcanbefabricated usingthissimple slit-shapenozzle.Patterningin thetransverse directioncanalsobeachievedbyusinganozzlecomprising mul-tipleholes.AnexampleofsuchadepositionisshowninFig.4(c). Individualislandsofmaterialaswellaslinescanbeobtainedwith

(7)

Fig.3. Effectofsolutionconcentrationoncarbonnanotubethinfilmmorphologyanddeviceperformance.Macro-photographypicturesofthesampleareshowninFig.S6. Depositionwasperformedat1600V.(a–b)Scanningelectronmicroscopyimagesoffilmsobtainedfromcarbonnanotubesolutionsofincreasingconcentration.Mainimages aretakenfromareaswithhighestcoveragewhileinsetimagesaretakenfromlowcoverageareas.Thedurationofthedepositionwasadjustedtoobtainnominallythesame amountmaterial(75,30,12and4.8s,respectively).(e)Amountofcarbonnanotubedeposited(usingRamanG-bandintensity)asafunctionofsolutionconcentrationfortwo setsofdurations(75,30,12and4.8s,and50,20,8and3.2s).EachdatapointisobtainedfromaveragingRamanspectralintensitiesfrom24neighboringlocations.Theerror barrepresentsonestandarddeviationandisameasureofthinfilmuniformity.(f)Deviceperformancedataobtainedfromfilmsin(a)–(d).Forwardandreversegatesweeps areshown(12␮A/Visequivalentto0.05␮A/␮mforthesedevices).Forthe0.9␮g/mL,amobility3.1cm2/Vsiscalculatedfromthelinearportionofthetransfercurve(light gray).AnON/Offrationinexcessof103_was_obtained_for_this_specific_device._The_substrate_was_SiO

2/Si(100nmoxide);thechannellengthandwidthwere12and220␮m, respectively.

dimensions of order 500␮m. Here, in addition tothe focusing actionoftheelectrostaticfield,theabsenceofflowmixing inher-enttothelaminarflowregimepreventsmaterialfromdepositing inareasbetweentheholes.Improvementsin nozzledesignand flowengineeringshouldallowfeaturesizesof100␮morlessin bothlateralandtransversedirections.Anotherexampleoflaminar flowdepositionisshowninFig.S9-10.Inparticular,Fig.S10shows depositionusingmuchfinermasksfabricatedusingaphotoetching process(Tecan,UK).Withthe100␮mholesand140holesperinch linedensity(180␮mperiod),limitedcrosstalkbetween neighbor-ingdepositswasobserved.SEMimagesrevealedexcellentspatial selectivitywithveryfewcarbonnanotubesbetweendeposits(Fig. S10(b–d)).Thisresultindicatesapotentialforlinedensitiesof300 perinch,areferenceinprintingtechnology.

Averyappealingperspectiveformasklesspatterningisthrough electricfield engineering.Inthepreviousexamples,theelectric field remained minimallypatterned. However,we can possibly attributethefocusingofmaterialtothenozzlecurvatureand/or

toapertures(slitorholes).Goingbeyondthenozzle,wehave per-formedafewinitialtestsasproof-of-conceptsanddemonstrated thatuponpatterningthegroundelectrodes,depositionof nano-tubesoccursatspecificlocations.Fig.4(d)showsoneexampleof carbonnanotube islandsdepositedonathinnylonfilm.During deposition,thenylonfilmwasrestingoverthemask(seeFig.S11), aTeflonslabwithcopperwireinclusionsconnectedtoground.Even ifthisexperimentusedaslit-shapenozzlewithuniformdeliveryof droplets,depositionoccurredmainlyoverthecopperareas.Thisis asimpleconsequenceofthehigherfield(and/orfieldgradient)in thoseareascomparedtothesupportmaterial(Teflon)wherethe groundis(spatially)approximately2mmlower.

4. Mechanism

It is in principle quite trivial to determine the mechanism leadingtodropletprecipitation,eitherfromelectro-(EP)or dielec-trophoresis (DEP) [32]. In the former, mass transport occurs if

(8)

Fig.4. Masklessdepositionofcarbonnanotubefilms.(a)Depositionfromaslit-shapenozzlesimilartoFigs.1–3.(b)Ramanintensityprofileacrossastripe(redcircle)of carbonnanotubematerial.Thelightgraylineisalorentzianprofile.(c)Depositionpatternobtainedfromanozzlewithmultipleapertures.Fromlefttoright,thesampleis immobilefortwodurationsandthencontinuouslytranslatedfor3mm,andthenimmobileagainfortwomoresteps.(d)Depositionofcarbonnanotubesonanylonfilm usingasampleholderwithapatternedground.MoredetailsonpatterneddepositionarepresentedinFig.S8–S11.

dropletsaccumulateelectricalchargesupontheircreationinthe ultrasonicgenerator.Thesymmetryinthedepositionbetween pos-itiveandnegativevoltagesleadsustoconcludethatbothpositively andnegatively chargeddroplets are present,in equal amounts. Giventhissymmetry,itmightappearthatDEPforcesare respon-siblefordropletprecipitation.Inthiscase,dropletsdonotrequire holdingacharge.However,sufficientelectricalpolarizability cou-pledwitha fieldgradient isnecessary.Our electrode geometry howeverdoesnotseemtosupporttherequiredfieldgradient.In ordertocompletelyexcludeone ofthetwo processes,we per-formedadepositionunderanACratherthanDCvoltage.Acommon approachinEPandDEP,theactionofatimeoscillatingforce aver-agestozeroforchargeddroplets(EP)whilefordipoles(DEP),a finitemagnitudeisexpectedfromthequadraticdependenceonthe field.Fig.5showstheresultsofACdepositionperformedat increas-ing frequencies (square wave modulation): At low frequencies belowafewHz,efficientdepositionoccurswithanamountof mate-rialsimilartoaDCdeposition.Asthefrequencyincreasesabove 8Hz, substantial reduction of deposited material is seen while above30Hz,essentiallynomaterialisvisible.Thatresultstrongly supports electrophoresis.If dielectrophoresis wasan important contributiontotheprocess,materialdepositionwouldbeseenat allfrequencies.Thefrequencycutoffforelectrophoresisrepresents ameasureofhowquicklydropletsareattractedtothesurface,of orderof100ms.Foranestimatedflowrateof1cm/s,that corre-spondstoa1mmdistance,oforderofthegapbetweenthenozzle andthesample.

Inthepresentexperiment,upto2000Vwasappliedacrossa 2mmgap,producingafieldinexcessof1MV/m.Foradroplet car-ryingasinglecharge,thecalculatedforceis0.16pNwhichseems likeasmallnumber.However,ataflowrateof1cm/s,weestimate

Fig.5.Mechanismofdropletprecipitation.Aerosoldepositionperformedunder alternatingvoltage(500Vamplitude)atvariousfrequencies(0.5–516Hz)is com-pared with depositionat constant voltage. Concentration and duration were 2␮g/mLand90s,respectively.Sampledimensionis25×30mm2,approximately.

at0.18pNtheforcerequiredtodeflecta1␮mdropletatthe sub-strateboundary(seeSupportingInformationunder“FORCESona DROPLET”).Thesimilaritybetweenthetwonumbersisquite strik-ing:However,itshouldonlybeseenasindicativethatonlyfew chargesneedtoaccumulateonadroplettoefficientlyperturbits trajectory.

In Ref.[33],electrostatic precipitationwasappliedin a CVD (chemicalvapor deposition)synthesis process tocollectcarbon nanotubesdirectlyfromthereactor.Itwasfoundthatfieldsoforder

(9)

100kV/mweresufficienttomaximizethecollectionyieldof nano-tubesholdingonaverage4unitcharges.Here, thehigherfields reportedcanbeattributedpartlytothegreatermassofdroplets comparedtofreenanotubes,ofcourseifthechargestateis compa-rable.Inthepresentmethod,themechanismofchargegeneration andtheroleplayedbycarbonnanotubesisnotknown.

5. Conclusion

Wedescribedanaerosolprocessthatusesdroplet electrophore-sisforcarbonnanotubedeposition.Electrostaticforcesgenerated onspontaneously chargeddroplets aresufficientlyimportantto drawnanotubestowardsatargetsubstrate,evenwhenthedroplets areloadedwithoneorafewcarbonnanotubes.sc-SWCNT assem-bledintonetworksonSiO2/Sishowedgoodperformanceintermsof

mobilityandOn/Offratio.Centimetersquareareascanbecovered inminutes,producinguniformfilmsonthemostdifficultsurfaces:

[27]Carbonnanotubenetworkswereassembledonsixdifferent polymerdielectricsand transistorsinbottomgateconfiguration showedsimilarperformancetoSiO2/Si,however,oftenwith

lit-tletonogatehysteresis.Otherthanencapsulationandtopgating, thisaerosolmethodoffersanalternativeroutetohysteresisfree devices,andpossiblyimprovedperformancefornanotubebased gassensors.

Currently,theprintingthroughputfortransistorisestimatedat ≈100s−1_and_a_thousand_to_million_fold_increase_should_be_within

reach. Thatwouldqualifythe methodinthemedium through-putcategory.Patterningofcarbonnanotubefilmsisdemonstrated usingashadowmask.Moreimportantlyforprintedelectronics, bothgasflowandelectricfieldcanbeengineeredtoproduce pat-terns in a maskless fashion. Similar to electrospray, conformal depositionon3Dsurfacesappearsrealistic.Themethodhasthe potentialtoprovideanalternative pathtoa fullyprinted,fully additiveprocesswithseveraluniqueattributes.

Acknowledgements

ThisworkwasperformedwithintheMaterialsThrustofthe PrintableElectronicsprogram,underthedynamicleadershipofDr. P.R.L.Malenfant.SpecialthanksgotoJ.McKeeforhishelp machin-ingseveralcomponentsinthisproject,P.Finnieforsomeofthe Ramandata,J.LapointeforSEMtraining,L.Chenformetal evapo-rationandN.Duforpolymerdielectrics.

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound, intheonlineversion,athttp://dx.doi.org/10.1016/j.mtcomm.2017. 01.002.

References

[1]M.Shiraishi,T.Takenobu,T.Iwai,Y.Iwasa,H.Kataura,M.Ata,Chem.Phys. Lett.39(2004)110–113.

[2]L.Hu,D.S.Hecht,G.Grüner,NanoLett.4(2004)2513–2517.

[3]S.Park,M.Vosguerichian,Z.Bao,Nanoscale2013(1727)5.

[4]K.Higuchi,S.Kishimoto,Y.Nakajima,T.Tomura,M.Takesue,K.Hata,E.I. Kauppinen,Y.Ohno,Appl.Phys.Express6(2013)085101.

[5]M.A.Meitl,Y.Zhou,A.Gaur,S.Jeon,M.L.Usrey,M.S.Strano,J.A.Rogers,Nano Lett.4(2004)1643–1647.

[6]Q.Cao,S.-j.Han,G.S.Tulevski,Y.Zhu,D.D.Lu,W.Haensch,Nat.Nanotechnol. 8(2013)180–186.

[7]Q.Cao,S.-j.Han,G.S.Tulevski,Nat.Commun.5(2014)5071.

[8]T.Toda,H.Frusawa,M.Furuta,Jpn.J.Appl.Phys.52(2013)03BB09.

[9]S.Fujii,T.Tanaka,H.Suga,Y.Naitoh,T.Minari,K.Tsukagoshi,H.Kataura, Phys.StatusSolidiB247(2010)2750–2753.

[10]R.C.Tenent,T.M.Barnes,J.D.Bergeson,A.J.Ferguson,B.To,L.M.Gedvilas,M.J. Heben,J.L.Blackburn,Adv.Mater.21(2009)3210–3216.

[11]S.L.Guillot,K.S.Mistry,A.D.Avery,J.Richard,A.-M.Dowgiallo,P.F.Ndione,J. vandeLagemaat,M.O.Reeseb,J.L.Blackburn,Nanoscale7(2015)6556.

[12]M.Jeong,K.Lee,E.Choi,A.Kim,S.-B.Lee,Nanotechnology23(2012)505203.

[13]A.Falco,L._Cina`,G.Scarpa,P.Lugli,A.Abdellah,ACSAppl.Mater.Interfaces6 (2014)10593–10601.

[14]P.Beecher,P.Servati,A.Rozhin,A.Colli,V.Scardaci,S.Pisana,T.Hasan,A.J. Flewitt,J.Robertson,G.W.Hsieh,F.M.Li,A.Nathan,A.C.Ferrari,W.I.Milne,J. Appl.Phys.102(2007)043710.

[15]T.Takenobu,N.Miura,S.-Y.Lu,H.Okimoto,T.Asano,M.Shiraishi,Y.Iwasa, Appl.Phys.Express2(2009)025005.

[16]Y.Nobusa,Y.Takagi,S.Gocho,S.Matsuzaki,K.Yanagi,T.Takenobu,Jpn.J. Appl.Phys.51(2012)06FD15.

[17]M.Ha,J.-W.T.Seo,P.L.Prabhumirashi,W.Zhang,M.L.Geier,M.J.Renn,C.H. Kim,M.C.Hersam,C.D.Frisbie,NanoLett.13(2013)954–960.

[18]S.G.Bucella,J.M.Salazar-Rios,V.Derenskyi,M.Fritsch,U.Scherf,M.A.Loi,M. Caironi,Adv.Electron.Mater.(2016),http://dx.doi.org/10.1002/aelm. 201600094.

[19]C.M.Homenick,R.James,G.P.Lopinski,J.Dunford,J.Sun,H.Park,Y.Jung,G. Cho,P.R.L.Malenfant,ACSAppl.Mater.Interfaces8(2016)27900–27910.

[20]J.Ding,Z.Li,J.Lefebvre,F.Cheng,G.Dubey,S.Zou,P.Finnie,A.Hrdina,L. Scoles,G.P.Lopinski,C.T.Kingston,B.Simard,P.R.L.Malenfant,Nanoscale6 (2014)2328.

[21]J.Ding,Z.Li,J.Lefebvre,F.Cheng,J.L.Dunford,P.R.L.Malenfant,J.Humes,J. Kroeger,Nanoscale7(2015)15741.

[22]Z.Li,J.Ding,P.Finnie,J.Lefebvre,F.Cheng,C.T.Kingston,P.R.L.Malenfant, NanoRes.8(2015)2179.

[23]P.Finnie,J.Ding,Z.Li,C.T.Kingston,J.Phys.Chem.C118(2014)30127.

[24]Z.Li,J.Ding,J.Lefebvre,P.R.L.Malenfant,Org.Electron.26(2015)15.

[25]http://www.sonaer.com/241TV.PDF.

[26]P.Kulkarni,G.J.Deye,P.A.Baron,J.AerosolSci.40(2009)164–179.

[27]J.Lefebvre,J.Ding,Z.Li,F.Cheng,N.Du,P.R.L.Malenfant,Appl.Phys.Lett.107 (2015)243301.

[28]Z.Liu,J.Zhao,W.Xu,L.Qian,S.Nie,Z.Cui,ACSAppl.Mater.Interfaces6 (2014)9997.

[29]L.Hu,D.S.Hecht,G.Grüner,NanoLett.4(2004)2513–2517.

[30]T.Kim,G.Kim,W.I.Choi,Y.-K.Kwon,J.-M.Zuo,Appl.Phys.Lett.96(2010) 173107.

[31]D.Vobornik,S.Zou,G.P.Lopinski,Langmuir32(2016)8735–8742.

[32]J.Voldman,Annu.Rev.Biomed.Eng.8(2006)425–454.

[33]M.Y.Zavodchikova,T.Kulmala,A.G.Nasibulin,V.Ermolov,S.Franssila,K. Grigoras,E.I.Kauppinen,Nanotechnology20(2009)085201.