REPORT DOCUMENTATION PAGE Form Approved OMB NO. 0704-0188 The public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggesstions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA, 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any oenalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS. 1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE 3. DATES COVERED (From - To) New Reprint - 4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER A comparison of single-wall carbon nanotube electrochemical capacitor electrode fabrication methods 5b. GRANT NUMBER W911NF-07-D-0004 5c. PROGRAM ELEMENT NUMBER 611104 6. AUTHORS 5d. PROJECT NUMBER Matthew H. Ervin, Benjamin S. Miller, Brendan Hanrahan, Benjamin Mailly, Tomas Palacios 5e. TASK NUMBER 5f. WORK UNIT NUMBER 7. PERFORMING ORGANIZATION NAMES AND ADDRESSES 8. PERFORMING ORGANIZATION REPORT NUMBER Massachusetts Institute of Technology (MIT) Office of Sponsored Programs 77 Massachusetts Avenue Cambridge, MA 02139 -4307 9. SPONSORING/MONITORING AGENCY NAME(S) AND 10. SPONSOR/MONITOR'S ACRONYM(S) ADDRESS(ES) ARO U.S. Army Research Office 11. SPONSOR/MONITOR'S REPORT P.O. Box 12211 NUMBER(S) Research Triangle Park, NC 27709-2211 52900-CH-ISN.332 12. DISTRIBUTION AVAILIBILITY STATEMENT Approved for public release; distribution is unlimited. 13. SUPPLEMENTARY NOTES The views, opinions and/or findings contained in this report are those of the author(s) and should not contrued as an official Department of the Army position, policy or decision, unless so designated by other documentation. 14. ABSTRACT Carbon nanotubes (CNTs) are being widely investigated as a replacement for activated carbon in super- capacitors. A wide range of CNT specific capacitances have been reported in the literature based on experiments using different CNT materials, fabrication methods, and characterization routines; making it difficult to draw conclusions about the relative merits of the different fabrication methods. This work systematically compares four solution-based electrode fabrication methods (drop casting, air brushing, filtration, and electrospraying) and, to a 15. SUBJECT TERMS Carbon nanotube, Graphene, Supercapacitor, Fabrication method 16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 15. NUMBER 19a. NAME OF RESPONSIBLE PERSON a. REPORT b. ABSTRACT c. THIS PAGE ABSTRACT OF PAGES John Joannopoulos UU UU UU UU 19b. TELEPHONE NUMBER 617-253-4806 Standard Form 298 (Rev 8/98) Prescribed by ANSI Std. Z39.18 Report Title A comparison of single-wall carbon nanotube electrochemical capacitor electrode fabrication methods ABSTRACT Carbon nanotubes (CNTs) are being widely investigated as a replacement for activated carbon in super- capacitors. A wide range of CNT specific capacitances have been reported in the literature based on experiments using different CNT materials, fabrication methods, and characterization routines; making it difficult to draw conclusions about the relative merits of the different fabrication methods. This work systematically compares four solution-based electrode fabrication methods (drop casting, air brushing, filtration, and electrospraying) and, to a lesser extent, some solution preparation techniques to determine if there is an optimum method for fabricating electrochemical capacitor electrodes out of single-wall CNTs (SWCNTs). We have found that it is best to use CNT solutions free from additives that may be difficult to remove from the fabricated electrode. In addition, the CNT solution preparation (e.g., dilution and sonication) had little effect on the resulting specific capacitance. Large differences in performance due to the fabrication methods were not seen, and the differences that were seen could be ascribed to material loss or contamination during the deposition. A single-layer graphene electrode was also fabricated and tested to obtain an estimate of the specific capacitance potentially achievable with SWCNTs, with 550 F/g demonstrated using 1 molar (M) sulfuric acid. REPORT DOCUMENTATION PAGE (SF298) (Continuation Sheet) Continuation for Block 13 ARO Report Number 52900.332-CH-ISN A comparison of single-wall carbon nanotube ele ... Block 13: Supplementary Note © 2012 . Published in Electrochimica Acta, Vol. Ed. 0 65, (0) (2012), (, (0). DoD Components reserve a royalty-free, nonexclusive and irrevocable right to reproduce, publish, or otherwise use the work for Federal purposes, and to authroize others to do so (DODGARS §32.36). The views, opinions and/or findings contained in this report are those of the author(s) and should not be construed as an official Department of the Army position, policy or decision, unless so designated by other documentation. Approved for public release; distribution is unlimited. ElectrochimicaActa65 (2012) 37–43 ContentslistsavailableatSciVerseScienceDirect Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta A comparison of single-wall carbon nanotube electrochemical capacitor electrode fabrication methods MatthewH.Ervina,∗,BenjaminS.Millera,BrendanHanrahana,BenjaminMaillyb,TomasPalaciosb aUSArmyResearchLaboratory,2800PowderMillRoad,Adelphi,MD20783-1197,USA bMa ssach usettsInst ituteofTech nolog y,Depa rtme ntofE lectrical Eng ineeringandC omputerScience,77MassachusettsAvenue,Cambridge,MA02139,USA a r t i c l e i n f o a b s t r a c t Articlehistory: Carbonnanotubes(CNTs)arebeingwidelyinvestigatedasareplacementforactivatedcarboninsuper- Receiv ed31October2011 capacito rs.Awide range ofC NTsp ecificc apacitances ha ve beenreport ed inthelite rature ba sedon RAeccceepivteedd i4n Jr aenvuisaerdy 2f o0r1m2 4 January 2012 experimen ts using differe nt CNT material s, fabrication metho ds, an d charact eri zati on routine s; mak ing itdifficulttodrawconclusionsabouttherelativemeritsofthedifferentfabricationmethods.Thiswork Available online 24 January 2012 systematicallycomparesfoursolution-basedelectrodefabricationmethods(dropcasting,airbrushing, filtration,andelectrospraying)and,toalesserextent,somesolutionpreparationtechniquestodetermine Keywords: ifthereisanoptimummethodforfabricatingelectrochemicalcapacitorelectrodesoutofsingle-wallCNTs Carbonnanotube (SWCNTs).WehavefoundthatitisbesttouseCNTsolutionsfreefromadditivesthatmaybedifficult Graphene Supercapacitor to remove from the fabricated electrode. In addition, the CNT solution preparation (e.g., dilution and Fabricationmethod sonication)hadlittleeffectontheresultingspecificcapacitance.Largedifferencesinperformancedueto thefabricat ionm etho dswe re not seen,and thediff erencesthat weres eencouldb e ascribedtom ater ial lossorcontaminationduringthedeposition.Asingle-layergrapheneelectrodewasalsofabricatedand testedtoobtainanestimateofthespecificcapacitancepotentiallyachievablewithSWCNTs,with550F/g demonstratedusing1molar(M)sulfuricacid. Published by Elsevier Ltd. 1. Introduction storageisreferredtoaspseudocapacitancesinceitbehaveselec- tricallyasacapacitance,thoughthechargetransferreactionsare Electrochemicalcapacitors,alsoreferredtoassupercapacitors, morelikethatofabattery.Sincethereisnodielectricontheelectro- have several advantages over conventional batteries, including chemicalcapacitorelectrodes,theappliedbiasesmustremainlow highe rspecifi cpower(∼2 order sofmagnitude higher),hi ghercycle enoughth atelectro chemicalbr eak downof theele ctroly tedoes not life (millions of charge/discharge cycles), rapid charge/discharge occur.Thislimitsthevoltageratingonindividualelectrochemical tim es (secon ds to minutes), high efficie ncies (up to 98%), and capaci torce llsto∼ 1V whenu singaq ue ouselectro lytes,and∼2.7V unaltered performance in extreme heat and cold (1). However, whenusingorganicelectrolytes. electrochemicalcapacitorshavelowenergydensitycomparedto Electrochemicalcapacitorsachievelargecapacitancesbyusing batteries which is a significant disadvantage for energy storage. electrodes with very large surface areas. Carbon electrodes are Increasingelectrochemicalcapacitorenergyandpowerdensities desirablebecausetheyareconductiveandhavehighsurfacearea, willmakethemmoreusefulforportablepowerapplications. goodcorrosionresistance,andgoodthermalstability[1].Carbon An electrochemical capacitor consists of two solid dielectric- materials with improved surface area may increase the capaci- freeelectrodes,incontactwithanelectrolyte,whichstorecharge tanceofelectrochemicalcapacitors.Twomaterialsbeingstudied byadsorptionofionsontotheelectrodes.Thecapacitancedueto forthisareCNTsandgraphene.Grapheneisasingleatomiclayer the adsorption of ions onto the electrodes is referred to as elec- ofgraphite.Similarly,aSWCNTisasingleatomiclayerofgraphite trochemicaldouble-layercapacitance,sincethereisalayerofions thatcurvesbackonitselftoformatube.Multi-wallcarbonnano- ontheelectrodewithasecondlayerofcounter-ions(oppositely tubes (MWCNTs) are carbon nanotubes that are more than one chargedions)nexttotheadsorbedions.Energycanalsobestored atomic layer thick and they were not used in this study. This through reduction and oxidation (redox) reactions at the elec- studyfocusedonSWCNTssincetheyhavethelargestsurfacearea trodes,whoseratesarepotentialdependent.Thistypeofenergy to mass ratio given that any interior walls in a MWCNT con- tribute mass but not surface area. Extremely large capacitances maybeobtainableifthesematerialscanbeassembledinamanner ∗ thatoptimizestheelectrodesurfaceareathatisaccessibletothe CE-omrraeislpaodnddreinssg: aMuathttohre. wTe.Hl.:.E +r1v i3n0.C1i v3@94M 0a0il1.m7;i lfa(Mx:. H+1. E3r0v1in 3).94 1559. elect rolyte. 0013-4686/$–seefrontmatter.Published by Elsevier Ltd. doi:10.1016/j.electacta.2012.01.060 38 M.H.Ervinetal./ElectrochimicaActa65 (2012) 37–43 Thisstudydetailsinvestigationsofvarioussolution-basedelec- proprietarycomponents.ThesefilteredCNTmatelectrodeswere trode fabrication methods to determine if there is an optimum mounted in a polycarbonate holder that pressed a nickel wire method for fabricating SWCNT electrodes for electrochemical againsttheCNTmattoserveasacurrentcollector,andtheywere capacitors.Solution-basedprocessingwaschosen,asitismanu- characterizedusing1Mpotassiumhydroxideelectrolyte,anickel facturableandcompatiblewithroll-to-rollprocessing.Italsodoes foilcounter-electrode,andasilver/silverchloridepelletreference- not impose significant thermal and chemical constraints on the electrode. und erlying current col lector as dire ct growth methods at 90 0◦C For the experiments comparing CNT deposition methods, a in a reducing environment would. When SWCNTs are deposited surfactant-freeSWCNTsolutioninwaterwasobtainedfromBrewer fromsolution,theytypicallydosoinbundles.Itisnotyetclearif ScienceInc.andusedas-receivedordiluted1:3withethanol.These thisbundlingisdetrimentaltotheresultingaccessiblesurfacearea solutions were used with or without 10min probe- plus 30min and,therefore,theresultingcapacitance.Inaddition,thedeposition bath-sonication.The Brewer Science Inc. solution contains SWC- methodmayalsoaffecttheporosityoftheelectrode,whichwill NTs that were functionalized by refluxing them in strong acids, affecthoweasilytheelectrolyteionscanmoveintoandoutofthe anditissoldforspincoatingfilmsofindividualSWCNTs.Thefunc- electrode (the Warburg impedance) during the charge/discharge tionalization results in sufficient CNT solubility in water that an process. approximate ly70(cid:2)g /m lconcentr ation canbeach iev edwith out the Many CNT solution-based processing approaches have been aidofsurfactants.1mlofthesolution(4mlifdilutedwithethanol) demonstr ated withmeasured specificcap acitancesof 23–2 00F/g was d epositedusi ng fo ur dif ferentm eth od s onto2c m×2 cmsec- reported[2–9].Itisdifficulttodrawdirectcomparisonsfromthese tionsoftitaniumfoilona175–200Chotplate(exceptforthefilter studiesbecausetheyusedifferent:CNTsourcesincludingsingle- and transfer method which is done at room temperature). Drop andmu lti-walle dtub es,s olutionco mpo sitions,f abrication meth- cast ing was performe d by d ep ositin g 1 0’s of (cid:2)l (∼mm diam eter ods,andcharacterizationprotocols.AccordingtoStollerandRuoff: droplets) at a time onto the current collector and letting it dry “Methodologytoreliablymeasureamaterial’sperformanceforuse beforerepeating;airbrushingdepositeddroplets10’sofmicrons asanultracapacitorelectrodeisnotwellstandardizedwithvari- indiameterwhichformaquasi-continuousthinfilmwhichwas oustechniquesyieldingwidelyvaryingresults”[10].Inaddition, allowed to dry before continuing spraying; and electro-spraying many of the higher specific capacitances reported likely include continuouslydepositedfew-microndropletswhichmayhavedried pseudocapacitive contributions in the measured specific capaci- before reaching the current collector. For the filter and transfer tance.Suchpseudocapacitancecontributionsmayoverwhelmthe method,thesolutionwasvacuumfilteredontoa0.22micronmixed double-layer capacitance, which is the focus of this work, as celluloseesterfilterpaper,andthentheCNTmatwastransferredto double-layercapacitanceisameasureoftheaccessibleCNTsur- thecurrentcollectorbyplacingthedryfilter,CNT-sidedown,onto face area being produced by the fabrication methods tested. We thetitaniumanddissolvingthefilterwithacetonevaporfollowed havesystematicallyinvestigatedtheindividualcontributionsofthe by acetone soaks. Finally, in an attempt to remove any remain- solutionpreparationandthedepositionmethodstotheachieved ingfilterresidue,thetransferredelectrodeisthensoakedindilute double-layercapacitance. nitricacidforafewminutes.TheCNTontitaniumelectrodeswere Inordertodeterminehowhighacapacitancemightbeachieved electrochemicallycharacterizedusing1Msulfuricacidelectrolyte, withSWCNTs,amodelsystemofasingle-layergrapheneelectrode atitaniumfoilcounter-electrode,andasilver/silverchloridepellet onsapphirewastested.Adielectricsubstrateisusedtoeliminate reference-electrode. anycontributiontocapacitancefromthesubstrate.Thisshouldbea The resulting CNT electrodes were weighed with a Mettler reasonablemodelforSWCNTssincetheelectrolytecanonlyaccess Toledo XP 26 balance, imaged with a Hitatchi S4500 field emis- onesideofthegraphene(correspondingtotheoutsideofaSWCNT). sion scanning electron microscope (SEM), and electrochemically testedusingaPrincetonAppliedResearchVersaStat3potentiostat 2. Experimental in a three-electrode geometry. Electrochemical characterization consisted of cyclic voltametry (CV) performed at a 20mV/s scan TwocommercialsourcesofSWCNTswereusedinthiswork.The rateovervoltagerangesappropriatetotheelectrolyteused.The useofbindersorconductivityenhancersasistypicalwithactivated fabrication-method-comparisonelectrodesontitaniumwerealso carbon electrodes has been avoided here as they were deemed characterizedwithelectrochemicalimpedancespectroscopy(EIS), unnecessaryanditwasfearedtheywouldcomplicatetheanaly- from50mHzto100kHz,performedat0.45Vtoavoidanyofthe sis.Foracomparisonofsurfactant-freeandsurfactant-containing redoxpeaksseeninthecyclicvoltamograms. solutions, chirally pure (6,5) SWCNT powder was obtained from A single-layer graphene electrode was prepared by chemical Sigma–Aldrich. That pure semiconducting CNTs were used is of vapordepositionofsingle-layergrapheneonacoppercatalystlayer no significance to the experimental results reported here. The andthentransferringthegraphenetoaquartzsubstrate[11].An CNT powder based solutions were made by mixing approxi- approximately 15mm internal diameter alumina tube was then mately 1mg of CNTs with 5ml of water and probe sonicating glued and clamped to the graphene/quartz sample using Loctite for10minat20%amplitudeusingaSonicsandMaterialsmodel 496andapolycarbonatefixture(whichpreventedanymechani- VCX 130 probe sonicator. Two solutions were made, one with- caldamagetothegraphene.)Thisceramictubeservedasthecell out surfa ctant, and one w ith 10 0(cid:2)l of C NTspe rse AQ surf actant to contain the ele ctrolyte as wel l as defin ing the exp os ed a rea purchased from www.MKnano.com. These two solutions were of graphene. The ends of the graphene extending outside of the allowed to sit for 10min to allow any large particles to settle ceramiccellwereelectricallycontactedwiththeaidofsilverpaint. to the bottom before the top 4ml were vacuum filtered using Forthesemeasurementsa1Msulfuricacidelectrolytewasused 0.4 (cid:2)m HTTP filters f rom Mil lip ore . Afte r filterin g, the r esult- alon g wit h a platinum w ir e co unter ele ctro de and a si lver/s ilver ing CNT mat was washed by filtering an additional 75ml of chloridepelletreferenceelectrode.TheCVforthiselectrodewas DI w ater . Com merc ially obt aine d CNT so luti ons in acet one w ere performe dove ranarrow potentialw ind ow of− 50m Vto−450 mV also investigated. These solutions consisted of functionalized or toavoidanyredoxpeaks.Thecurrenttocalculatethecapacitance unfunctionalized SWCNTs suspended in acetone, using surfac- was taken as one half the difference between the oxidation and tants and/or dispersants containing: a-(nonylphenyl)-hydroxy- reduction loops at the midpoint of the potential scan range. As branchedpoly(oxy-1,2-ethanediyl)(20–50%);2,4,7,9-Tetramethyl- usual,thecapacitanceisthencalculatedasCapacitance=Current 5-decyne-4,7-diol(2–10%);and2-Butoxyethanol(<1%);andother (A)/ScanRate(V/s). M.H.Ervinetal./ElectrochimicaActa65 (2012) 37–43 39 Fig.1. Typical20mV/scyclicvoltammogramforSWCNTelectrodesproducedusing theBrewerSciencesolution.Whilethereareredoxpeakssuperimposedonthe approximatelyrectangularCVcurvethatisrepresentativeofdoublelayercapac- itance,thecurrentforcalculatingthecapacitanceismeasuredonthereductionside (lowercurve)at0.45V. 3. Resultsanddiscussion Measurement methodologies can contribute to significant differencesinmeasuredcapacitances.Therefore,standardtestcon- ditionsof20mV/shavebeenusedherefortheCVmeasurements. Whencharacterizingelectrochemicalcapacitors,oneneedstobe carefultodistinguishbetweendouble-layercapacitanceandany pseudocapacitancecontributions.Toavoidinclusionofanypseu- docapacitivecontributions,thecapacitancehasbeencalculatedfor eachelectrodeusingthecurrentmeasured,onthereductionside oftheCVloop,atapotentialthatavoidsanyredoxpeaks.InFig.1, whichisatypicalCV,thecurrentwasmeasuredonthereduction loopat0.45V(vsAg/AgCl),toavoidtheredoxpeak(s)thatappears asthepotentialisscannedmorenegatively.Thisisamuchmore conservativemeasurementofcapacitancethanresultsfrominte- gratingtheCVcurveorfromtwoelectrodemeasurementswhich would include the redox contributions. All of the measurements Fig.2. (A)SWCNTelectrodemadeusingaSWCNTsolutionthatincludedasurfac- weremadeat0.45Vfortheelectrodesmeasuredinsulfuricacid. tant.(B)SWCNTelectrodemadewithasurfactant-freesolutionofSWCNTsresulting Electrochemicalimpedancespectrawerealsomeasuredat0.45V. In order to com pare the ca pacitanc es ge nera ted by the di ffere nt in a cleaner and more porous electrode with better electrical conductivity. fabricationmethods,wecalculatedspecificcapacitance(F/g)using onlythemassoftheSWCNTsaftersubtractingthecapacitanceofa furtherprocessingisrequiredtoremovesurfactantresidues,which blank m etalfo il curr entcolle ctor. Specificcapa cita nceshaveb ee n mayrem oveCNTs fr omthes ub strateo ralterthe electrode mor- calcul atedfr om theCV curvesusi ngthem easuredma sseso fthe phol ogy. CNTs as well as masses calculated from the solution concentra- In order to investigate the effect of surfactants on the result- tions and volu me sused. Measuredm asse smo reaccura telyreflect ing e lectro des , we made ele ctrode s from surfac tan t-fr ee and perfo rma ncewhen CNTs arelostin proces sing( e.g.,oversp rayor sur factant-cont ainin g CNT solutions m ade i n house. These elec- incompletefi ltering ).Cal cula ted ma ssesaremo reac curatewh en trodes were examine d in an SEM a nd gre atl y reduc ed po rosity theremayb eresidua lcontamina tionth atad dma ss(e.g.,in com- was se en w hen surfac tan t w as u sed a s can b e seen i n Fig. 2. plete filter pap erremov al).Specificcap acita nces from atlea stthree Elec troche mical testing sho wed the C NT ele ctro de m ade wit h- dupli catee lectro deshaveb eenaver agedforrepo rting fo reach fab- out surfactants in Fig. 2b has six times the cond uctivit y and ricationm ethod. four timeslessc on tact resis tanc e,w henpr obed withanohm me- ter, as the surfactant containing electrode shown in Fig. 2a. The 3.1. SWCNTsolutioncomparison surfactant-containingelectroderesultedin14F/gversus24F/gfor the surfactant-free electrode in this case. While the differences CNTs are frequently processed as suspensions/solutions. In between surfactant-free and surfactant-containing electrodes is order to get stable dispersions, the CNTs are either chemically highly variable, due in part to how much surfactant is removed functionalizedand/ordispersedwithasurfactant.Bothapproaches byrinsingetc.,thesurfactant-freeelectrodesconsistentlyproduce havedrawbacks.Chemicalfunctionalizationcanmakethematerial betterperformance,aswehaveobservedinpreviousworkinour moresoluble,buttheadditionofthesefunctionalgroupsintroduces lab[12]. defectsthatcandecreaseconductivity.Functionalgroupscanalso Electrodes were also made using the commercial surfac- beredoxactiveandproduceundesirabledecompositionbyprod- tant/dispersantcontainingSWCNTsolutionsinacetone.Attempts uctsovertime.Surfactantscanbeusedinplaceoffunctionalization toremovetheseadditivesthroughthermalannealing,acidwash- tohelpincreasethesolubilityofCNTs,butsincetheyaregener- ing, or solvent extraction were largely unsuccessful. As a result, allynonconductive,theytypicallydegradeperformance.Therefore, theseelectrodesalsohadpoorcapacitancesduetolowaccessible 40 M.H.Ervinetal./ElectrochimicaActa65 (2012) 37–43 Fig.3. Picturesofthedifferentdepositionmethodsused(A)dropcasting,(B)airbrushing,(C)aSWCNTdeposit(black)onamixedcellulosefilterpaper,and(D)electrospraying. surfaceareaandhighresistivityduetopoortube-to-tubeelectrical approachbecauseiteliminatestheneedtoremovethesurfactant contact. altogether.Unfortunately,workingwithsurfactant-freesolutions SubsequentelectrodeshavebeenmadeusingtheBrewerSci- requirestheuseoflowerconcentrations. ence Inc. surfactant-free SWCNT in water solution. This solution yieldedmuchbetter-lookingandhigherperformingelectrodesas 3.2. Fabricationmethodcomparison willbeseenbelow.Whileitmaybepossibletoremovesurfactant residuesdependingonthesolutionused,wehaveconcludedthat Attheoutsetofthiswork,itwasassumedthattheamountof usingsurfactant/dispersant-freesolutionsofSWCNTsisthebetter SWCNTbundlingwouldbeasignificantfactorinthecapacitance Fig.4. Photographsofelectrodesfabricatedusingthefourmethods(A)dropcast,(B)airbrushed,(C)filtered/transferred,and(D)electrosprayed. M.H.Ervinetal./ElectrochimicaActa65 (2012) 37–43 41 Fig.5. SEMmicrographsofelectrodesfabricatedusingthefourmethods(A)dropcast,(B)airbrushed,(C)filtered/transferred,and(D)electrosprayed. thatcouldbeachieved,andthatitmayalsoaffectthepowerperfor- electrodeappearstohaveslightlysmallerbundles.Themicroscopic manceduetotheelectrolytemasstransferresistancetoreachthe similaritybetweentheseelectrodeswouldseemtoindicatethatthe interior bundle surfaces. Therefore, methods for producing solu- depositionmethodsinvestigatedhavelittleeffectontheresulting tionsofunbundledSWCNTsanddepositionmethodsthatprevent electrodetopographyandthereforecapacitance. re-bundlingasthesolutiondriedwouldbedesirable.Withthisin Fig.6showsthespecificcapacitancesthatresultfromtheseelec- mind, four methods for depositing the CNT solutions that might trodefabricationmethods.Theseresultswerecalculatedusingthe yielddifferentamountsofbundlingwereselectedforcomparison. measuredmassesofCNTsdeposited.ItcanbeseeninFig.6thatall Byreducingthedropletsize,itmaybepossibletoreducethenum- ofthefabricationmethods,exceptfiltrationandtransfer,yieldspe- beroftubesavailableforbundlingduringdropletdrying,andfaster cificcapacitanceswithintwostandarddeviations(95%confidence) dryingtimesmaykineticallylimitthebundling. of35F/g.Itisexpectedthatthefilteredandtransferredelectrodes DropCastingconsistsofdepositingdropletsofsolutionontothe donotperformaswellduetothepresenceoffilterpaperresidue currentcollectorandevaporatingthedropletsover1–2min,with whichincreasesthemeasuredmasswhilealsolikelyreducingthe thetotaldepositiontakingabout20min.Airbrushingspraysmuch capacitance. smallerdropletsontothecurrentcollector;andthedropletsdryin When the specific capacitances are recalculated using CNT 1–3swiththetotaldepositiontimebeingabout30min.Vacuum massescalculatedfromthesolutionconcentrationandvolume,as filtration,takesafewsecondstotal.Thelastmethod,electrospray- ing[13],requiredadilutionofthesolutionwithethanol(1:3v/v CNT/water:ethanol)todecreasesurfacetensionandincreasecon- ductivity. Electrospraying produces extremely small droplets, on the order of 4 microns in diameter for pure ethanol, and they may even dry before reaching the current collector. In addition, asthedropletsdryintransit,thechargedensityonthedroplets mayincreaseenoughthatcolumbicrepulsionwithinthedroplets couldcausethedropletstoexplode,perhapspreventingorreducing SWCNTbundling.Toensurethattheaddedethanolwasnotrespon- sibleforanydifferencesobservedwithelectrospraying,somedrop castelectrodeswerealsomadewithaddedethanol.Picturesofthe fourdepositionmethodsareshowninFig.3. The resulting electrodes were characterized photographically and through SEM imaging. Fig. 4 contains photographs showing electrodes made with the four different methods. The area that the SWCNTs were deposited over was roughly the same for the four methods. The filtration and transfer method produced the mostuniformdeposit,whiledropcastingyieldedthemostinho- mogenousdeposit.SEMimagesofthefourelectrodesareshownin Fig.5.Atamicroscopicscale,thefourelectrodeslookverymuch Fig.6. Plotofspecificcapacitancesachievedwiththevariousfabricationmethods the sam e ,w ithSWCNTb undle son theo rderof∼5– 50nm ind iame- whe n t he CN T mass is m easured wit h a balanc e. Fab rica tion met hods: (A) d rop cast – nosonication,(B)dropcast–sonicatedneat,(C)dropcast–sonicatedwithethanol, terbeingtypical.Thefilteredelectrodebundledistributionappears som ewha tskewe dto larger bundlessi zes,and theelectro sprayed ((FD)) fiellteecrterdosapnrdaytreadn –sfesorrneidca–tende awtiwthi tehtohuatnsool,n (icEa) taioirnb.rushed – sonicated neat, and 42 M.H.Ervinetal./ElectrochimicaActa65 (2012) 37–43 Fig.7. Plotofspecificcapacitancesachievedwiththevariousfabricationmethods whentheCNTmassiscalcultedfromthevolumeandconcentrationofthedeposited CNTsolution.Fabricationmethods:(A)dropcast–nosonication,(B)dropcast– sonicatedneat,(C)dropcast–sonicatedwithethanol,(D)electrosprayed–sonicated withethanol,(E)airbrushed–sonicatedneat,and(F)filteredandtransferred–neat withoutsonication. showninFig.7,onecanseethatthedropcastandelectrosprayed electrodeshavevirtuallyidenticalperformances.Inthiscase,the airbrushedandfilteredelectrodesdonotcomparewellduetoCNT lossduringdeposition.Duringairbrushingsomedropletsmaymiss thetitaniumaltogetherorbeblownoffbeforetheydryasreflected inlowermeasuredmassesthanpredicted.Infiltration,CNTsare Fig.8. EISandcapacitanceretentiontestingofatypicaldropcastCNTelectrode. visiblypresentinthefiltrate.InbothcasestheCNTmasscalculated (A)Nyquistplot,(B)capacitanceasafunctionofmeasurmentfrequency,and(C) from the solution concentration is an overestimate of the actual capacitanceretentionover1200cycles. mass ofC NTsdep osited.InFig.7 ,i tca nalsobeseen th ats onica- tionwithorwithoutaddedethanolhasnotaffectedtheelectrode themaximumpossiblespecificcapacitanceforCNTsorisfurther performance. optimizationwarranted? TheEIScharacterization,ofalloftheelectrodesfabricated,did Inanattempttoanswerthesequestions,amodelsystemofa notshowsignificantqualitativedifferencesbetweentheelectrodes. single-layergrapheneelectrodeonsapphirewastested.Thisshould Fig.8showsthetypicalEISresultsfromadropcastelectrode.The beareasonablemodelforsingle-wallCNTssincetheelectrolytecan NyquistplotshowninFig.8ashowsasmallRCloopathighfre- only access one side of the graphene (corresponding to the out- quency.Thisislikelyduetoasurfaceoxideonthetitaniumcurrent sideofaSWCNT).Fig.9showsthe20mV/sCVofthesingle-layer collectorsincetheywereonlydegreasedandnotetchedpriorto grapheneelectrode.TheCVcurveisseentobesomewhatdistorted CNTdeposition.Fig.8bshowsaplotofcapacitanceasafunction duetothehighresistanceintheelectrode.Thegraphenewaselec- offrequency.Thiselectrodehadacorresponding45degreephase tricallycontactedatbothsidesandaresistancefromoneleadto angle at 50 Hz. The capacitance retention as a function of cycle num- theoth erof4.3k(cid:2) w asm easure d.Th i sresistance isac omb inati on ber for the electrode is show in Fig. 8c where 94% of the initial ofcontactresistancebetweenthesilverpaintandthegraphene, capacitanceisretainedafter1200cycles. aswellastheresistanceofthegrapheneitself.Inaddition,inthis potentialregion,thereisaredoxcurrentduetothereductionof 3.3. Singlelayergrapheneelectrode InFigs.6and7,littledifferenceinspecificcapacitancesbasedon fabricationmethodisseen(exceptwhencontaminationispresent) which suggests that the fabrication methods investigated here havenotsignificantlyaffectedtheamountofCNTbundling.While bundlingmightbebeneficialforincreasingtheelectricalconduc- tivity of the CNT electrode, it has been assumed here that it is detrimentalduetoareductionoftheaccessiblesurfaceareaofthe electrode.However,thismaynotbethecase.Ithasbeenreported thatCNTelectrodesexpandandshrinkduringcyclingwhichcould beindicativeofatleastpartialdebundling[14].Thebundleexpan- sion may be due to electrostatic repulsion between like charged CNTs or due to oppositely charged electrolyte ions forcing their wayinbetweentheCNTs.Thereforethequestionbecomeswhatis thehighestspecificcapacitancethatcouldbeachievediftheentire CNT surface areawa sreadilyacc essib le?H ave wealrea dy ach ieved sFuiglf.u 9r.i cCaVc iodf esilnecgtlreo-llayyteera ngrdaaphsielnveer o/sni lsvaeprhcihrelo erliedcetrpoedllee ttarkeefenr aetn 2ce0 emleVc/tsr oudsien.g 1 M M.H.Ervinetal./ElectrochimicaActa65 (2012) 37–43 43 4. Conclusions AsystematiccomparisonofdifferentCNTsolutioncompositions and deposition methods has been undertaken to determine the importantfactorsinelectrodefabrication.Wehavefoundthatit isbesttouseCNTsolutionsfreefromadditivesthatmaybedif- ficult to remove from the fabricated electrode. The CNT solution compositionandprocessingismoreimportantthanthechoiceof theCNTdepositionmethodsusedhere.Intheend,thechoiceof fabrication method may be determined by other factors such as manufacturability or the efficiency with which the SWCNTs are used. Sincetheachievedspecificcapacitancesarenotapproachingthe capacitancepredictedwithasingle-layergrapheneelectrode,more work needs to be done to develop an optimized CNT electrode. Sincethedifferencesbetweenthefabricationmethodsusedhere Fig.10. EISmeasurementofcapacitanceversusfrequencyforalarge-areasingle- arenotlarge,otherapproachestooptimizingtheelectrodes,suchas layergrapheneelectrode. usingdouble-ormultiple-walltubes(whichresultinlessbundling and larger pores) and the incorporation of functional groups or oxygensincethemeasurementswereperformedinanopencell. nanoparticlestoinducespacingbetweentheCNTsshouldbeinves- Thisoxygenreductioncurrentisexpectedtobediffusionlimited tigated. andsoisexpectedtoshifttheCVcurvetowardsnegativecurrents Even if CNTs electrodes do not yield specific capacitances in butnotaffecttheareawithintheCVcurvewhichrepresentsthe cap acita nce.T hem easu redcap acit anc eof4 2(cid:2)F/cm 2isverycl ose excess of activated carbon, they may still yield improvements inpowerduetoincreasedelectrodeconductanceandoptimized totheexpectedintrinsiccapacitanceofonemolarsulfuricacid[15]. At 76n g/cm2fo rgraphen e(onlycou nti ngo nesid eofthe grap hene porosity. There may also be important electrochemical capaci- torimprovementsduetothemechanicalpropertiesofCNTs.For towardsthesurfacearea),thiscorrespondsto550F/gthatshould instance, CNTs may lend themselves to flexible, conformal, or beachievablewithSWCNTs.Thisis,atthistime,onlyanestimate integrated electrochemical capacitors that would be useful for for the specific capacitance of graphene. If the CV measurement applicationswherethereislittleavailablespace. is made a lower scan rate (e.g. 5mV/s) a higher capacitance is measured (although measurement errors get more significant at References lowerscanrates).TheEISmeasurement(Fig.10)showsthatour CVmeasurementsfallonasteepslopeinthecapacitanceversus fre quencycurvein dica ting t hatth eseres ult ssh ouldbeinter preted [[21]] XA..GZ.h Paaon,dNoalnfoo,t Aec.Fh. nHoolollgeynk2a0m(2p0, 0J. 9P)o0w6e5r6 S0o5u.rces 157 (2006) 11–27. withsome cautio n(20mV/ scor respo ndstoa pproxim a tely0.75Hz [3] M .Kaem pgen,etal.,Nano Le tt.9(2 009)1872–1876. [4] S.W .Lee,etal., J. Am .Chem .Soc . 131(20 09)671–679. in this EIS measurement). Unfortunately, the curve is not seen to [5] L.Co oper ,e tal. ,A ppl. Phys.L ett. 95(2 009)2 33104. plateauatthelowerfrequencies(whichwouldyieldthetruecapac- [6] S. -L.Chou ,et al. ,Elect roche m.Co mm un.10 (2008)1724–1727. itancel im it)d ueto thelargeti mecon stantt hatre sul tsfro mthe [7] J.H.K im,K .-W .N am,S.B.Ma,K .B.Kim,Ca rb on44(2 006)1963–1968. large r esistan ce o f th e sin gle-la yer g raphene e lectr ode. At tempt s to [[89]] LJ.. WH ua,n egt, aMl..,W Pr.oE cl.l sNwaot lr.t Ahc, aEdle. cS tcrio. cU h.Se.mA.. 1So0c6. (T2r0a 0n9s .) 12914(2900 0–92)1244914–.247. improvethismeasurementtoobtainthetruecapacitanceofsingle- [10] M .Stolle r,R.R uoff,Energ yEnviron.Sci .3(2 010)1 29 4–1301 . layergra phe neareon-goin g. Howev er, ifsin glesidedgra p heneis [11] X.L i,etal., N anoLe tt.9(20 09)4359 –43 63 . a suit able mode l fo r SWCNTs , it is alrea dy clear from t hese resul ts [[1132]] AC..MJ .J.a Awn otoren k, ,MM.Hat. eErr.vSicn i,. E4l2ec(t2r0o0 c7h)e2m6.6 S–o2c9. 7Tr.ans. 33 (2011) 117–128. thatwearenotreachingthepotentialofSWCNTelectrodeswith [14] P.W .Ruch,R .Kotz,A .W oka un,Ele ctrochim.Acta54(2009)4451–4458. thec urr ent fabr icationm etho ds,presum a blydue toCNTbun dling [15] R.Ist van,T he 20th Int ernationa lSeminaron Doub le LayerC apacitors&Hybrid ori nsufficie ntporosity resulting inreducede lect rol ytea ccessible En ergySt orag eDe vices,Deerfiel dBeach, FL, Decemb er6– 8,2010,20 10 . surfacearea.