See discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/258000808Selectivity of chemical oxidation attack of single-wall carbon nanotubes in solutionARTICLEinPHYSICAL REVIEW B NOVEMBER 2003Impact Factor: 3.74 DOI: 10.1103/PhysRevB.68.193412CITATIONS28DOWNLOADS49VIEWS1564 AUTHORS, INCLUDING:Enzo MennaUniversity of Padova74 PUBLICATIONS 868 CITATIONS SEE PROFILEMoreno MeneghettiUniversity of Padova146 PUBLICATIONS 2,848 CITATIONS SEE PROFILEAvailable from: Enzo MennaRetrieved on: 12 August 2015Selectivity of chemical oxidation attack of single-wall carbon nanotubes in solutionEnzo Menna, Federico Della Negra, and Michela Dalla FontanaDepartment of Organic Chemistry, CNR ITMPadova, University of Padova, 1, Via Marzolo, 35131 Padova (Italy)Moreno Meneghetti*Department of Physical Chemistry, University of Padova, 2, Via Loredan, 35131 Padova (Italy)Received 5 December 2002; revised manuscript received 1 August 2003; published 21 November 2003Chemical oxidation of single wall carbon nanotubes SWNTs, produced with the HiPco method, is studiedbyresonanceRamanscattering. Theanalysisoftheradial breathingmodemakesit possibletoassigntheobserved bands to metallic and semiconducting SWNTs and to suggest the chiral indices of SWNTs contrib-uting to the spectra. On this basis we observe that the most important parameter which determines the attackof the nanotubes is their diameter: Small diameter tubes, due to the stress induced by the curvature, are rstattacked and destroyed. Some reactivity is also observed for larger diameter metallic nanotubes, which, how-ever, are not easily destroyed.DOI: 10.1103/PhysRevB.68.193412 PACS numbers: 61.48.c, 78.30.Na, 33.20.FbAmong the most interesting nanostructures useful fornanotechnology, single-wall carbon nanotubes SWNTshavecharacteristicproperties, inparticular relatedtotheirelectronicstructure.1,2ThesynthesisofSWNTsproducesavarietyofnanotubeswithavarietyofdiametersandsemi-conducting or metallic properties. However synthetic proce-dures are not able, at the moment, to separate different typesofnanotubes. Manipulationofisolatedandwell character-ized SWNTs is important for many applications such as na-noelectronics and photonics and their solubilization is an im-portant step.SolublederivativesofSWNTshavebeenobtainedwithdifferent approaches,3many of which start by oxidizing themicrometerlongtubesproducedbythesynthesis. Theoxi-dation is obtained by a strong attack with concentrated sul-phuricandnitricacidsundersonicationandtheresult isashortening of the nanotubes to some hundreds of nanometersand their functionalization with carboxilic acid groups. Theshortened nanotubes are then further oxidized with sulphuricacid and hydrogen peroxideetching reaction without soni-cation in order to purify them from amorphous carbon.4,5We present a resonant Raman study of the strong SWNToxidation attack in solution. Raman is a powerful techniquefor SWNTcharacterization since a strong resonance en-hancement of thespectra, whichprovidesbothvibrationaland electronic structural information, is observed. Character-istic features of the Ramanspectrumare foundat about15001600 cm1tangential bands andcanberelatedtocarbon-carbon stretching.6,7The low frequency region, usu-ally below400 cm1, shows the radial breathing modeRBM, atotallysymmetricmodecharacteristicofanano-tube, andwhosefrequencydependsontheinverseof thetubediameter.8,9This spectral featuremadeit possibletosuggest the chiral indices (n, m) of isolated SWNTs.10This isa very important result because the chiral indices determinethe properties of any single nanotube. We will use the RBMbands, inparticular, andshowhowit is possibletogaininformation on the oxidative attack on the basis of an assign-ment of all the bands observed in the spectrumbelow350 cm1.The SWNTs studied in the present paper are HiPcoSWNTs obtained by a disproportionation of CO molecules athighpressureinthepresenceofFenanoparticlesproducedbythedecompositionofFe(CO)5.11TheHiPcomethodispromising for large scale production using a continuous owgas-phase process. SWNTs obtained by this method are veryinterestingsincetheycorrespondtonanotubeswithawidevariety of diameters, in particular nanotubes with smalldiameters,12thusenablingustostudy, inthesamesample,the behavior of many tubes. It will be shown that small di-ameter tubes are easily destroyed, as already noted by usingmolecular oxygen.13,14Thechemical oxidationinsolution,studied here, also makes it possible, within the same step, toremove almost completely the Fe nanoparticles encagedwithin the tubes during their synthesis. This is a useful resultin view of the need for samples produced on a large scale forexample for nanoelectronic applications like densely packedeld-effect transistorsFET.15Puried HiPco SWNTs were purchased fromCarbonNanotechnologies Inc.. For the oxidative reaction we used asuspension of the pristine sample in a 3:1 H2SO4(98%)/HNO3(65%) solutionat55Candundersonicationat 45 kHz. We analyzed the reaction mixture after some min-utes and up to 4 h. The solutions were ltered and thesamples etched for half an hour with a 4:1 H2SO4(98%)/H2O230%solutiontoremoveallcarbonparticlesproduced by the rst reaction.4Sharper Raman spectra wereactually obtained, in particular for samples with longer oxi-dationtimes. Yieldof theoverall procedureisreportedinFig. 1. After4htherewasa9%yieldandRamanspectrashowedthat thenanotubeswerestill present. Ontheotherhand, high-resolution transmission electron microscope HR-TEM measurements showed, for sample oxidizedfor 35min, that Fenanoparticleshadbeenalmost completelyre-moved. However, wewerenotabletondthatbundlesofnanotubes,presentinpristinesample,weredisentangled. Alower solubility observed for samples after the oxidative pro-cedureisinlinewiththelatterobservationsinceitcanberelated to a pronounced aggregation of nanotubes.PHYSICAL REVIEW B 68, 193412R 20030163-1829/2003/6819/1934124/$20.00 2003 The American Physical Society 68 193412-1Raman spectra were recorded using the 1.96-eVline632.8 nm of an He-Ne laser and a TRIAX 320 Jobin Yvonspectrometer equipped with a charge-coupled deviceCCDdetector and a microscope. A 50 magnication and15103mW on an area of 510 m2were used. Figure 2shows the spectrum of pristine HiPco nanotubes in the RBMregion. Many bands can be observed between 150 and350 cm1whichcanberelatedtoSWNTs withdifferentdiametersandwhoseelectronicstructuresareinresonancewith the laser excitation line. It is well known that nanotubeelectronic structures can be accounted for, at least at a rstapproximation and for the lowest energies, by using a tightbinding scheme.1,16,17This founds experimental conrmationwith the measurement of the density of states by using scan-ning tunnelling microscopy and spectroscopy STS.18,19Thevan Hove singularities in the density of states determine themost relevant electronic transition energies of the nanotubes.In the energy region around 1.96 eV, a resonance conditioncanbefoundfor theE22Sof SWNTswithsemiconductingproperties and a diameter between 0.75 and 0.94 nm and forthe E11Mof SWNTs with metallic properties for which (n-m)isamultipleof 3 andadiameter between1.10and1.46nm.6RBMbandscanbetentativelyassignedconsideringthepossibleresonanceoftheexcitinglaserlinewiththeelec-tronictransitioncharacteristicofSWNT. Withthisaimwecalculated the van Hove singularities of individual nanotubesusing the tight binding parameter 02.9 eV. This 0 valuemade possible the best and most consistent interpretation ofthe Raman spectra.6,10A different value used for the interpre-tation of STS data,19seems to derive from different experi-mental conditions characteristic of this technique.20In TableIwehaveidentiedthechiral indices(n, m) ofthenano-tubes with E22Stransition energies within 0.15 eV from theexciting laser line, which seems adequate to include all thetubes in resonance considering that a broadening of the vanHove peaks can be present due to intertube interactions.21InTable II are reported the calculated data for the correspond-ing metallic tube with E11M1.960.15 eV.Reported in the same table are also: The lowest frequencylimit which is considered to be appropriate for isolatedSWNT,8RBM(cm1)224/d(nm), and the highest limitvalid for strong interacting tubes, such as those in bundles,FIG. 1. HiPco SWNT reaction yield after oxidation for differenttimes and half an hour etching.FIG. 2. Raman spectrum of pristine HiPco SWNT. Exciting lineat 1.96eV. Near thebands arereportedthefrequencies and, inparentheses, the full width at half maximum obtained by tting withLorentzian line shapes as shown by dashed lines. The assignment tometal and semiconducting nanotubes comes from the analysis of thevan Hove singularities of individual nanotubessee the text.TABLE I.E22Selectronic transition energies of a semiconductingSWNT with a diameter d in the range 0.750.94 nm and in reso-nance with the exciting laser line at 1.96 eV. is the chiral angle ofthenanotube.Thefrequencyrange, RBM, reportsthelimitsofthe RBM frequency for isolated and interacting nanotubessee thetext.(n, m) dnm deg RBM (cm1) E22SeV7,5 0.829 24.5 270297 1.947,6 0.895 27.5 250275 1.858,3 0.782 15.3 286315 2.028,4 0.840 19.1 267293 2.019,1 0.757 5.2 296325 2.0510,3 0.936 12.7 239263 1.8411,1 0.916 4.2 245269 1.91TABLE II. E11Melectronic transition energies of metallic SWNTwith diameter d in the range 1.101.46 nm and in resonance withthe exciting laser line at 1.96 eV. is the chiral angle of the nano-tube. The frequency range, RBMreports the limitsofthe RBMfrequency for isolated and interacting nanotubessee the text.(n, m) dnm deg RBM (cm1) E22MeV9,9 1.238 30.0 181199 1.9810,7 1.175 24.2 191209 2.05 2.1211,5 1.126 17.8 199219 2.10 2.2611,8 1.312 24.8 171188 1.85 1.9012,6 1.260 19.1 178195 1.90 2.0113,4 1.222 13.0 183201 1.93 2.1113,7 1.396 20.2 161176 1.73 1.8114,2 1.199 6.59 187205 1.95 2.1914,5 1.354 14.7 165182 1.76 1.9015,0 1.191 0.0 188207 1.96 2.2215,3 1.326 8.95 169186 1.78 1.9616,1 1.312 3.0 171188 1.79 2.0017,2 1.436 5.5 156171 1.65 1.8218,0 1.429 0.0 157172 1.66 1.83BRIEF REPORTS PHYSICAL REVIEW B 68, 193412R 2003193412-2calculated with RBM(cm1)246/d(nm) where the propor-tionality constant is raised by 10%.25Therefore the fre-quency interval can account for samples which contain iso-latedSWNTsornanotubesinabundle. Itcanbeobservedthat metallic tubes in resonance with the exciting laser lineare more numerous than semiconducting nanotubes also be-causeofthedoublepeakscalculatedforallmetallictubes,excluding the armchair (n, n)ones.Onecanseethatthespectrumisaccountedforandthatthe bands above 250 cm1are associated with semiconduct-ing tubes whereas those below this frequency with metallicones. Seven Lorentzian line proles allowed the tting of theexperimental data. The tentative assignment of the spectrum,based on the data in Tables I and II, is reported in Table III.The linewidths of the bands are consistent with the fact thatmore than one nanotube contributes to a single band, givingthat the natural temperature independent linewidthof theRBM is estimated to be of the order of 3cm1.22Only theband at 218 cm1has one associated nanotube, although ithasawidthof9cm1, andit seemstobeasymmetricinparticular when the reaction of oxidation is in progressseebelow. Further investigations are needed to clarify this spec-tral region. Two achiral tubes, one armchair9,9 and onezigzag15,0, aresuggestedtocontributetothemostin-tensebandat 193 cm1. All othernanotubesprovedtobechiral tubes apart from the 18,0 zigzag tube which seems tocontributetotheweakbandat166 cm1. Itisnoteworthythat one can suggest associating one or more nanotubes withall the bands.Figure 3 shows the spectra of the RBM region for samplesobtained after a given time of oxidative reaction followed byhalf an hour of etching. A relative comparison of the spectrahas been obtained by considering that the bands decrease inintensitybecausenanotubesaredestroyedduringthereac-tion. Therefore to nd the relative decrease of the intensitiesof the bands of a sample which reacted for a given time, wehave normalized them to those of the sample which reactedfor the previous shorter time. One band of the sample whichreacted for the longer time was taken as a reference consid-eringthat all theothersmust havealower intensity. Theband at 218 cm1shifted at 223 cm1for the sampleswhichreactedwasusedasareferenceforall thespectraunless for those with 65, 120, and 240 reaction times, forwhichthebandat 203 cm1foundat 193 cm1for thepristinesample wasused. Theintensityof thebandsob-tained by tting the spectra with Lorentzian line shapes, andnormalized to the intensity they have for the pristine sample,are reported in Fig. 4. One can observe a progressive disap-pearance of the bands starting from those at higher frequen-cies. However, one can clearly see that also the intensity ofthe band at 193 cm1is reduced to half the value it has forthepristinesample, after onlyafewminutesof chemicalTABLE III. Assignment of the RBM bands observed with the 1.96-eV 632.8-nm laser exciting line. Fullwidth at half maximumFWHM and relative intensitiesI are obtained by a tting with Lorentzian lineshapes. A*indicatesthat theexcitationlineat 1.96eVisstronglyresonant within0.02eV withtheelectronic transition of the nanotube.RBM (cm1) FWHM (cm1) I (n, m)a166 22 0.16 M:13,7,14,5,17,2,18,0193 12 1.00 M:9,9*,10,7,12,6,13,4,14,2*,15,0*218 9 0.45 M:11,5251shoulder 5 0.16257 9 0.98 S:7,6,10,3,11,1283 6 0.37 S:7,5*,8,4296 9 0.18 S:7,5*,8,3,9,1aM and S are for the metal and semiconductor, respectively.FIG. 3. Raman spectra of the RBM spectral region for the HiPcoSWNT after the oxidative reaction for the time in minutes reportednear each spectrum and half an hour etching reaction.FIG. 4. Intensities of the RBM bands reported in Fig. 3 relativeto their values for the pristine sample as a function of the time ofthe oxidative reaction.BRIEF REPORTS PHYSICAL REVIEW B 68, 193412R 2003193412-3oxidation. This indicates that the SWNTs contributing to thatband are easily destroyed or strongly modied by the oxida-tivereaction. Onecanalsondthat all thebandsshift tohigher frequencies, and this agrees with a larger aggregationof the shortened SWNTs which we observed to be lesssoluble than those of the pristine sample.Fromtheassignment of thespectrumgivenabove, onecan deduce that the rst SWNT to be strongly attacked arethe semiconducting nanotubes with small diameters. This canbe understood since nanotubes with a small diameter have alarge stress, due to their curvature, which makes them morereactive. We think, therefore, that in this case it is the diam-eters of the nanotubes that are important for their reactivity.However one observes that the metallic tubes contributing tothebandat193 cm1arealsoeasilyattacked. A particularreactivityof metallic tubes towardoxidationwas not re-ported to our knowledge. We only observe that, according totheaboveassignment, (n, m) chiral tubeswithmn) aresuggestedtocontributetoall theRBMbands, but achiraltubes, like armchair 9,9 and zigzag 15,0, are suggestedto contribute only to the band at 193 cm1. Statistics on thetype of tubes more easily destroyed by chemical oxidation insolution, showedthat tubeswithalowmchiral indexaremoreeasilyattacked.19Ourobservationcanbeinterpreted,therefore, with a rapid oxidation of the 15,0 metallic tubes.To support this conclusion one can remember that it is alsoknown, from the oxidation rate of graphite sheet, that zigzagsites are more reactive than armchair sites.23To understandthe reactivity of the metallic nanotubes one must recall thattheoxidizednanotubeshavebeensubsequentlyfunctional-ized with polyethyleneglycol chains by the formation of anamide bond.5Raman spectra of the functionalized nanotubes,not reported in the present paper, show that the small diam-eter nanotubesarealwaysabsent but that theratioof theintensity of the 193-cm1now at 199 cm1) to the218-cm1nowat222 cm1) bandsisfoundtobethatofthe pristine nanotubes. This allows one to conclude that thelarger diameter metallic nanotubes are not destroyed by theoxidative procedure and that the lower intensity of their bandin the Raman spectra, after the oxidation, must be due to astrong variation of their electronic structure and therefore ofthe resonance condition of the Raman spectra. The recoveryof theintensityof their bandsinthefunctionalizednano-tubes, where they are believed to be separated by the poly-mer chains, and the above recalled observation of the lowersolubility of the oxidized samples, suggests that the aggrega-tion of the oxidized nanotubes determines an interactionamong the nanotubes which is very important in determiningtheir electronic properties. One can also suggest that a broad-eningof thevanHovesingularitiesof thelarger diametermetallic nanotubes can be a consequence of their interaction,and that this determines different resonance conditions. Thereasonfor thelower intensityof thebands relatedtothemetallictubes must befound, therefore, intheresonanceconditions and not in their destruction. Other excitation laserfrequencieswillbeconsidered, andwillallowustowidenthe understandingof the oxidative reactionof the HiPconanotubes.In conclusion we have analyzed the RBM Raman featuresof HiPco SWNT and shown that a tentative assignment, alsoof chiral indices, can be suggested with a contribution of asmall number of individual nanotubes with metallic andsemiconductingproperties. The assignment allowedus tofollow the nanotubes chemical oxidation in solution and tond that the diameter of the nanotubes, and not other prop-erties, is the most important parameter whichdeterminestheir reactivity and destruction.The authors thank G. Marcolongo for technical assistanceand Stefano Polizzi for HR-TEMmeasurements. We ac-knowledgeinterestingdiscussionswithR. Bozio, G. Scor-rano, andM. Maggini. TheworkwassupportedbyMIURalsoContract No. MM03198284 andCNRlegge95/95Nanotecnologie.*Email address: Moreno.Meneghetti@unipd.it1R. Saito, G. Dresselhaus, and M. S. Dresselhaus, Physical Prop-ertiesof CarbonNanotubes Imperial CollegePress, London,1998.2R.H. Baughman, A.A. Zakhidov, and W.A. de Heer, Science 297,7872002.3S. Niyogi et al., Acc. Chem. Res. 35, 11052002.4J. Liu et al. Science 1253, 2801998.5F. Della Negra, M. Meneghetti, and E. Menna, Fullerenes, Nano-tubes, and Carbon Nanostructures 11, 252003.6A. Jorio et al., Phys. Rev. B 65, 1554122002.7A. Jorio et al., Phys. Rev. Lett. 90, 1074032003.8S. Bandow et al., Phys. Rev. Lett. 80, 37791998.9J. Kurti, G. Kresse, andH. Kuzmany, Phys. Rev. B58, R88691998.10A. Jorio et al., Phys. Rev. Lett. 86, 11182001.11P. Nikolaev et al., Chem. Phys. Lett. 313, 911999.12A. Kukovecz et al., Eur. Phys. J. B 28, 2232002.13W. Zhou et al., Chem. Phys. Lett. 350, 62001.14I.W. Chiang et al., J. Phys. Chem. 105, 82972001.15P.G. Collins, M.S. Arnold, and P. Avouris, Science 292, 7062001.16H. Kataura et al., Synth. Met. 103, 25551999.17J.W. Ding, X.H. Yan, andJ.X. Cao, Phys. Rev. B66, 0734012002.18J.W.G. Wildoer et al., NatureLondon 391, 591998.19T.W. Odom et al., J. Phys. Chem. B B104, 27942000.20M.S. Dresselhaus et al., Carbon 40, 20432002.21A.M. Rao et al., Phys. Rev. Lett. 86, 38952001.22A. Jorio et al., Phys. Rev. B 66, 1154112002.23J.M. Thomas,ChemistryandPhysicsofCarbonDekker,NewYork, 1966, Vol. 1, p. 121.24H. Kuzmany et al., Eur. Phys. J. B 22, 3072001.25There are many indications that the RBM frequency of interactingnanotubes is higher thanthat of isolatednanotubes, howeverthere is not agreement on the upshift see Refs. 21 and 24. Wehave used a large estimation of the upshift which is of the sameorderoftheresultobtainedinthestudyofanisolatedSWNTstrongly interacting with a substratesee Ref. 10.BRIEF REPORTS PHYSICAL REVIEW B 68, 193412R 2003193412-4