Single-wall carbon nanotubes with diameters approaching 6 nm obtained by laser vaporization

Carbon 40 (2002) 417–423

Single-wall carbon nanotubes with diameters approaching 6 nmobtained by laser vaporizationa , b b a*Sergei Lebedkin , Peter Schweiss , Burkhard Renker , Sharali Malik ,

c c c a,cFrank Hennrich , Marco Neumaier , Carsten Stoermer , Manfred M. Kappesa ¨Forschungszentrum Karlsruhe, Institut f ur Nanotechnologie, D-76021 Karlsruhe, Germanyb ¨ ¨Forschungszentrum Karlsruhe, Institut f ur Festkorperphysik, D-76021 Karlsruhe, Germany

c ¨ ¨Institut f ur Physikalische Chemie, Universitat Karlsruhe, D-76128 Karlsruhe, Germany

Received 20 February 2001; accepted 28 April 2001

Abstract

Single-wall carbon nanotubes (SWNTs) with large diameters from 2 to 5.6 nm were prepared by pulsed laser vaporizationof carbon rods doped with Co, Ni and FeS in an atmosphere of Ar:H . The SWNT material was characterized by SEM,2

HRTEM, Raman, IR, UV–VIS–NIR absorption spectroscopy and thermogravimetric analysis. 2002 Elsevier ScienceLtd. All rights reserved.

Keywords: A. Carbon nanotubes; B. Laser irradiation; C. Electron microscopy, Raman spectroscopy, Thermal analysis (DTA and TGA)

1. Introduction composition of the target /process gas. For example, laservaporization of carbon doped with Pd, Rh is known to

The established methods for the preparation of SWNTs produce smaller diameter nanotubes [9]. Numerous studiessuch as laser vaporization of C:Ni:Co targets [1,2] and of the effect of various additives on the yield and morphol-electric arc vaporization of C:Ni:Y rods [3] in an inert gas ogy of nanotubes have been performed using electric arcatmosphere (Ar) as well as pyrolysis of CO [4] and vaporization, which was historically the first method tohydrocarbons (CH ) [5–7] on metal catalyst nanoparticles produce SWNTs [10,11]. It was found that the addition of4

produce carbon tubes with typical peak diameters of |1.2– some non-catalytic elements (e.g., S, Bi, Pb) to Ni, Co1.7 nm. Among these preparative methods, laser vapor- promoted growth of nanotubes with large diameters up toization remains the most advantageous with regard to high |6 nm [12,13]. On the other hand, a solar oven vapor-yield of SWNTs, homogeneity of the product, and precise ization of carbon doped with Ni, Co and S yielded SWNTscontrol of process parameters. The diameter distribution of with a broad diameter distribution, including a substantialSWNTs produced by this method is particularly narrow fraction of nanotubes with diameters smaller than |1 nmand within |1.2–1.5 nm. The preparation of SWNTs of as judged from the Raman spectra [14]. Hydrogen can alsosimilarly high quality but with larger diameters would be influence the growth of nanotubes: SWNTs with diametersof great interest. They can be used for comparative studies around 2 nm have been produced from graphite mixedof the properties of SWNTs vs. tube diameter and as with Co, Ni and FeS and vaporized by electric arc in atemplates for nanowires and other nanostructures. pure hydrogen atmosphere [15].

The diameter distribution of SWNTs produced by laser In this work, we have studied the effect of hydrogen (asvaporization is affected by the process temperature and gas a H admixture to Ar) and sulfur (as a FeS additive to2

pressure, however the variation is limited by a decrease in carbon targets) on SWNTs produced by pulsed laserthe nanotube yield [8]. A larger effect on the diameter vaporization of carbon targets doped with Ni, Co metaldistribution can be achieved by changing the chemical catalysts. Our results show that the introduction of hydro-

gen, sulfur or, particularly, both leads to the efficientformation of SWNTs with large diameters approaching 6*Corresponding author. Fax: 149-7247-826368.

E-mail address: [email protected] (S. Lebedkin). nm. The SWNT materials have been characterized by

0008-6223/02/$ – see front matter 2002 Elsevier Science Ltd. All rights reserved.PI I : S0008-6223( 01 )00119-1

418 S. Lebedkin et al. / Carbon 40 (2002) 417 –423

several methods including electron microscopy, Raman graphite powder (Alfa Aesar). Both were mixed with Co,and optical absorption spectroscopy and thermogravimetric Ni or Co, Ni, FeS powders (1 at.% each) and cold-pressedanalysis. at 150 or 500 MPa, respectively. Cylindrical pre-forms (13

mm diam., 10–25 mm length) were then heated in argon to10508C. In this fashion quite rigid rods were obtained from

2. Experimental mesophase carbon [16]. However, (uneven) laser vapor-ization eroded their surface into fine needles, indicating a

2.1. Laser vaporization setup non-homogeneous bulk structure. Annealing the targets upto |17008C for |20 min (by resistive heating in vacuum)

The laser vaporization setup is shown schematically in eliminated this problem. Also, SWNTs produced from theFig. 1. A Nd:YAG laser (Continuum Powerlite, Q- annealed rods contained less amorphous carbon accordingswitched, 1064 nm, 0.5 J /pulse, 30 Hz) irradiated a side to Raman spectroscopy. This might be due to a lower levelsurface of a rotating and axially translating composite of impurities in the annealed targets (elemental analysiscarbon rod (target) placed in a [70 mm T-like quartz tube showed that concentrations of O, N decreased from 1.1 andinside a hinge oven (Linn High Therm). This configuration 0.8 to 0.03 and 0.09 wt.%, respectively, whereas theis more practical for continuous operation than rastering a amounts of Co, Ni and FeS decreased only slightly), orlaser beam across the target [2]. The apparatus can run due to a more homogeneous structure and, consequently, awithout interruption until the target is fully consumed different vaporization regime. Targets prepared from(when its diameter becomes comparable to that of the laser briquetting graphite powder were much less robust me-beam (|6 mm)). Vaporization was performed at an oven chanically than those from mesophase carbon pitch andtemperature of 11508C in 0.5 bar Ar or Ar15 vol.% H required careful handling. On the other hand, an even laser2

flowing at |80 sccm. A small part of the gas flow was vaporization of these targets and a good yield of SWNTsinjected into a side arm of the quartz tube in order to were obtained without requiring additional high-tempera-protect the laser irradiation input window (anti-reflection ture annealing. Furthermore, in contrast to the pitch,coated) from particle deposition. SWNT material was briquetting graphite is practically pure carbon (,0.01collected on a filter downstream of the oven, in a cold wt.% of O, N, S and ,0.001 wt.% of ash). This type ofdownstream region of the quartz tube and along the tubing composite carbon target was used in most experimentsto the filter (Fig. 1). Generally, the laser beam was reported here.unfocused. When it was focused to a |2 mm spot,significant ablation of the target (macroscopic particles)and a lower yield of SWNTs were observed. A typical 3. Results and discussionunfocused vaporization rate was |0.15 g/h in pure Ar, butit was |10 times lower in Ar15 vol.% H . The following characterization methods were applied to2

as-prepared SWNT samples.2.2. Preparation of composite carbon targets

3.1. Electron microscopyThe composite carbon targets were prepared from

mesophase carbon pitch or briquetting grade natural SEM micrographs of the SWNT materials obtained by

Fig. 1. Laser vaporization setup for preparation of SWNTs. A particular feature of this setup is the side surface irradiation of a rotating andaxially translating composite carbon target with a fixed laser beam. The target is mounted on a ceramic rod and placed in the middle of aT-like quartz tube in a hinge oven.

S. Lebedkin et al. / Carbon 40 (2002) 417 –423 419

laser vaporization of C:Ni:Co:FeS and C:Ni:Co targets in 3.3. Raman spectroscopyAr:H (Samples 1 and 2, respectively) and the2

C:Ni:Co:FeS target in Ar (Sample 3) were similar to those Raman spectra of Samples 1–3 were obtained withof the standard SWNT material produced by the vapor- excitation at 488, 514.5 (Ar-ion laser), 532 and 1064 nmization of C:Ni:Co in Ar. Long interleaved bundles of (Nd:YAG laser) with the same conditions for all samples.SWNTs (|20–30% by volume, |20 nm diam.) were Fig. 4 shows the spectra for Samples 1 and 2 in the radialobserved. A high-resolution TEM analysis of free-standing breathing mode (RBM) range in comparison to the spectrananotubes in Samples 1–3 revealed that the addition of of the standard SWNT material (C:Ni:Co/Ar). The RBMboth H and FeS (Sample 1) had the largest effect on the patterns of the latter are relatively narrow and can be2

diameter distribution of SWNTs (Fig. 2). In that sample, assigned to the resonantly excited SWNTs with diametersnanotubes with large diameters from 2 to 5.6 nm (Fig. 3) from |1.2 to |1.5 nm [17,18]. The RBM patterns ofwith a maximum at |3.7 nm were observed. Typically, the Samples 1–3 are much more complicated and changehistograms shown in Fig. 2 are based on |100 tubes and dramatically upon changing the excitation wavelengthconsequently are subject to the associated statistical uncer- (Fig. 4, Sample 3 not shown). Compared to the standardtainty. An EDX element analyzer incorporated in the SWNTs, these RBM patterns are in general downshifted inelectron microscope did not detect sulfur (Samples 1 and frequency and show, in particular for Sample 13). Furthermore, we were not able to detect any spatial (C:Ni:Co:FeS/Ar:H ), poorly resolved features down to 502

21separation of Ni, Co and Fe on the scale of |10 nm. A cm (this lowest measured Raman shift would correspondrelatively large amount of graphitic carbon particles vs. to |4 nm nanotubes). Except when excited at 1064 nm, theamorphous carbon was noted for Sample 3. Finally, no RBM Raman signals of Samples 1–3 are substantially(free-standing) multiwall nanotubes were found in Samples weaker than those of the standard SWNTs. This behaviour1–3. is especially striking at l 5 532 nm. Finally, Samplesexc

1–3 were not as homogeneous as the standard material as3.2. Infrared spectroscopy judged from micro-Raman spectra (laser spot |2 mm).

The Raman data indicate that Samples 1–3 (particularlyNo distinct bands, which may be assigned to molecular Sample 1) contain many SWNTs with diameters much

vibrations, could be observed in the IR spectra of Samples larger than |1.5 nm. However, the capabilities of Raman1–3 (ground in a tungsten carbide mortar and pressed in spectroscopy for characterization of such nanotubes appearKBr pellets). This implies a low content of hydrocarbons to be rather restricted because of the (inherently) broadand organic sulfur in these SWNT materials. Most of the distributions of SWNTs with large diameters and thevaporized sulfur and the hydrogen which took part in expected strong decrease of the resonant enhancement ofchemical reactions in the vaporization /hot zone appear to Raman signals of SWNTs with increasing nanotube diam-have been converted into gaseous products. Indeed, CH , eter [17].4

C H , C H , and CO (due to oxygen impurity in gas /2 2 2 4

target) were found in the downstream gas flow (IR gas 3.4. Optical spectroscopycell).

Fig. 5 shows electron absorption spectra of thin filmsprepared by spraying suspensions of Samples 1 and 3 inacetone onto a hot quartz substrate [19]. The spectrum ofthe standard material (C:Ni:Co/Ar) shown for comparisoncorresponds well to previously reported data [8,19]. Theabsorption bands S , S and M were attributed to optical1 2

transitions between van Hove singularities of the density ofstates (DOS) in semiconducting (S , S ) and metallic (M)1 2

SWNTs. The spectra of Samples 1 and 3 are similar, butshifted to lower photon energies. For example, the S band1

is shifted from 0.75 to |0.68 eV. Furthermore, relativelybroad bands are observed for Sample 1. These resultssuggest that only nanotubes with relatively small diameters(,2 nm) effectively contribute to the band structure in theabsorption spectra. Sample 1 might contain a minorfraction of such tubes (bundles) which escaped the TEManalysis (Fig. 2). The van Hove singularities of the DOSprobably ‘blur out’ with increasing nanotube diameter andFig. 2. TEM diameter distributions of SWNTs produced by laserthe corresponding absorption coefficients for SWNTs withvaporization of the C:Ni:Co:FeS composite carbon target in Ar15large diameters (.2 nm) are small. However, quantitativevol.% H (Sample 1), the C:Ni:Co target in Ar15 vol.% H2 2

(Sample 2), and the C:Ni:Co:FeS target in Ar (Sample 3). understanding of the optical properties of different SWNT

420 S. Lebedkin et al. / Carbon 40 (2002) 417 –423

Fig. 3. TEM micrographs of SWNTs produced by laser vaporization of the C:Ni:Co:FeS composite carbon target in Ar15 vol.% H2

(Sample 1). Micrograph (a) shows the largest observed tube with a 5.6 nm diameter. This and other tubes were overcoated with relativelyfew particles attributed to amorphous carbon (micrograph (b)). The observed ends of nanotubes appeared to be free from metal catalystparticles.

S. Lebedkin et al. / Carbon 40 (2002) 417 –423 421

Fig. 4. Low-frequency Raman spectra of SWNT materials containing large diameter tubes (Samples 1 and 2) compared to the standardSWNT sample (C:Ni:Co/Ar). The Raman signals of Samples 1 and 2 excited at 488, 514 and particularly at 532 nm were relatively weak.

21The small peaks in these spectra between 50 and 150 cm are due to air / laser lines.

materials requires further theoretical and experimental of different SWNT samples and of graphite powder, forinvestigations. comparison. The residual weight at high temperatures is

due to metal oxides. The large residual weight value for3.5. Thermogravimetric analysis (TGA) Sample 1 (up to |50 wt.%) can be explained by a partial

loss of vaporized carbon due to the formation of gaseousFig. 6 shows relative weight changes during combustion hydrocarbons (CH and C H , see above). The residual4 2 2

weights for the standard SWNTs (C:Ni:Co/Ar) and Sample3 are consistent with the amounts of Ni, Co and Fe (1 at.%each) in the carbon targets. The standard SWNT material

Fig. 5. Optical absorption spectra of SWNT films on a quartzplate. The features S , S and M were attributed to electronic1 2

transitions between pairs of van Hove singularities in semicon- Fig. 6. TGA traces for different SWNT materials and graphiteducting and metallic SWNTs, respectively. The spectra are shifted powder, for comparison. Heating rate 58C/min in Ar:O (92:8,2

for clarity. The step-like change of absorption at |1.4 eV is an v/v) flowing at 100 sccm. The residual weights are due to metalartifact from the spectrophotometer. oxides.

422 S. Lebedkin et al. / Carbon 40 (2002) 417 –423

exhibits a one-step TGA curve with onset at |3508C. In Acknowledgementscontrast, combustion of Sample 1 proceeds in a multi-step

This work was supported by the Deutsche Forschungs-fashion with the last step starting at a temperature abovegemeinschaft under SFB 551 ‘Kohlenstoff aus der Gasph-6008C, which practically coincides with the beginning ofase: Elementarreaktionen, Strukturen, Werkstoffe’ and bythe combustion of graphite (Fig. 6). This behavior seemsthe BMBF project ‘Grundlagenuntersuchungen zur Wasser-to be consistent with the broad diameter distribution ofstoffspeichenmg in Nano-Carbonstrukturen’. S.L. thanksSWNTs in Sample 1 including a large fraction of

¨ ¨Prof. K.J. Huttinger (Inst. fur Chemische Technik, Uni-nanotubes with diameters $3 nm. The chemical properties¨versitat Karlsruhe) for helpful discussions and the gift ofof such large diameter nanotubes, e.g. stability with respect

mesophase carbon pitch. The authors thank Dr. C.to oxidation, are likely close to those of graphite.Adelhelm and H. Kaiser (Forschungszentrum Karlsruhe,IMF-I) for elemental analysis.

4. ConclusionsReferences

Our results show that the addition of H to the process2

[1] Guo T, Nikolaev P, Thess A, Colbert DT, Smalley RE.gas, sulfur (FeS) to the C:Ni:Co composite target or, inCatalytic growth of single walled nanotubes by laser vapor-particular, both H and FeS, dramatically influence the2ization. Chem Phys Lett 1995;243:49–54.growth of SWNTs. Instead of nanotubes having diameters

[2] Rinzler AG, Liu J, Dai H, Nikolaev P, Huffman CB,within a narrow range of |1.2–1.5 nm (C:Ni:Co/Ar),Rodriguez-Macias FJ et al. Large-scale purification of single-tubes with large diameters broadly distributed from |2 towall carbon nanotubes: process, product, and characteriza-

|5.6 nm (TEM) were obtained by the laser vaporization of tion. Appl Phys A 1998;67:29–37.a C:Ni:Co:FeS target in Ar:H (Sample 1). The mecha-2 [3] Journet C, Maser WK, Bernier P, Loiseau A, de la Chapellenism by which hydrogen and sulfur influence the growth of ML, Lefrant S et al. Large-scale production of single-wallednanotubes is not yet clear. However, it is evident that large carbon nanotubes by the electric arc technique. Naturediameter SWNTs grow efficiently in the presence of sulfur 1997;388:756–8.

[4] Dai H, Rinzler AG, Nikolaev P, Thess A, Colbert DT,and hydrogen and this is not only restricted to theSmalley RE. Single-wall nanotubes produced by metal-cata-conditions of catalytic pyrolysis of hydrocarbons as sug-lyzed disproportionation of carbon monoxide. Chem Physgested by Cheng et al. [6]. The positive, but complicated,Lett 1996;260:471–5.effect of sulfur and hydrogen on the production of carbon

[5] Kong J, Cassell AM, Dai H. Chemical vapor deposition ofnanofibers has been noted by many researchers and ismethane for single-walled carbon nanotubes. Chem Phys

assumed to be due to a modification of the catalytic Lett 1998;292:567–74.activity of metal nanoparticles and a suppression of the [6] Cheng HM, Li F, Sun X, Brown SDM, Pimenta MA,formation of carbides and graphitic overlayers on their Marucci A et al. Bulk morphology and diameter distributionsurface [20]. However, which factors determine the growth of single-walled carbon nanotubes synthesized by catalyticof large diameter SWNTs and what is the nature of the decomposition of hydrocarbons. Chem Phys Lett

1998;289:602–10.distinct synergetic effect upon addition of both sulfur and[7] Cassell AM, Raymakers JA, Kong J, Dai H. Large scalehydrogen (Fig. 2) remains to be understood.

CVD synthesis of single-walled carbon nanotubes. J PhysThe laser-produced material with a high content ofChem B 1999;103:6484–92.large-diameter nanotubes (Sample 1) is of interest for

[8] Jost O, Gorbunov AA, Pompe W, Pichler T, Friedlein R,comparative studies of the physical and chemical prop-Knupfer M et al. Diameter grouping in bulk samples of

erties of SWNTs. At present, the production of large single-walled carbon nanotubes from optical absorptionamounts of this material is hindered by a significantly spectroscopy. Appl Phys Lett 1999;75:217–9.lower 1064 nm vaporization rate of composite carbon [9] Kataura H, Kimura A, Ohtsuka Y, Suzuki S, Maniwa Y,targets in an atmosphere of Ar:H than observed for pure Hanyn T et al. Formation of thin single-wall carbon2

Ar (see Experimental). It appears that relatively large nanotubes by laser vaporization of Rh/Pd–graphite compo-site rod. Jpn J Appl Phys 1998;37:L616–8.product particles quickly form in the vaporization zone in

[10] Bethune DS, Kiang CH, de Vries MS, Gorman G, Savoy R,the presence of hydrogen, which then strongly scatter /Vasquez J et al. Cobalt-catalysed growth of carbon nanotubesabsorb the laser radiation at 1064 nm. Indeed, we observedwith single-atomic-layer walls. Nature 1993;363:605–7.relatively strong scattering of visible light (excited metal

[11] Iijima S, Ichihashi T. Single-shell carbon nanotubes of 1-nm*atom and C emission as well as thermal radiation) fromn diameter. Nature 1993;363:603–5.the laser vaporization plume under these conditions. This [12] Bethune DS. Adding metal to carbon: production andpeculiarity of the laser vaporization of carbon targets in the properties of metallofullerenes and single-layer nanotubes.presence of hydrogen and an improvement of the pro- In: Andreoni W, editor, The chemical physics of fullerenesduction yield for large-diameter SWNTs will be the subject 10 (and 5) years later, Kluwer Academic, 1996, pp. 165–81,of further experiments. and references cited therein.

S. Lebedkin et al. / Carbon 40 (2002) 417 –423 423

[13] Kiang CH. Growth of large-diameter single-walled carbon [17] Dresselhaus MS, Eklund PC. Phonons in carbon nanotubes.nanotubes. J Phys Chem A 2000;104:2454–6. Adv Phys 2000;49:705–814, and references cited therein.

[14] Alvarez L, Righi A, Guillard T, Rols S, Anglaret E, Laplaze [18] Bandow S, Asaka S, Saito Y, Rao AM, Grigorian L, RichterD et al. Resonant Raman study of the structure and electronic E et al. Effect of the growth temperature on the diameterproperties of single-wall carbon nanotubes. Chem Phys Lett distribution and chirality of single-wall carbon nanotubes.2000;316:186–90. Phys Rev Lett 1998;80:3779–82.

[15] Liu C, Cong HT, Li F, Tan PH, Cheng HM, Lu K et al. [19] Kataura H, Kumazawa Y, Maniwa Y, Umezu I, Suzuki S,Semi-continuous synthesis of singlewalled carbon nanotubes Ohtsuka Y et al. Optical properties of single-wall carbonby a hydrogen arc discharge method. Carbon 1999;37:1865– nanotubes. Synth Met 1999;103:2555–8.8. [20] Ci L, Wei J, Wei B, Liang J, Xu C, Wu D. Carbon nanofibers

¨[16] Braun M, Huttinger KJ. Sintering of powders of poly- and single-walled carbon nanotubes prepared by the floatingaromatic mesophase to high-strength isotropic carbons. 3. catalyst method. Carbon 2001;39:329–35, and referencesPowders based on iron-catalyzed mesophase synthesis. Car- cited therein.bon 1996;34:1473–91.