Appl. Phys. A 69, 269–274 (1999) / Digital Object Identifier (DOI) 10.1007/s003399900116

Applied Physics A Materials Science & Processing  Springer-Verlag 1999

Towards processing of carbon nanotubes for technical applications G.S. Duesberg1,2 , J. Muster1 , H.J. Byrne3 , S. Roth1 , M. Burghard1 1 Max-Planck-Institut für Festkörperforschung, Heisenbergstr. 1, 70569 Stuttgart, Germany 2 Physics Dept., Trinity College Dublin, Dublin 2, Ireland 3 School of Physics, Dublin Institute of Technology, Kevin Street, Dublin 8, Ireland

Received: 17 May 1999/Accepted: 18 May 1999/Published online: 29 July 1999

Abstract. Production methods for carbon nanotubes are now well established and allow their synthesis on a scale of grams per day. For many potential applications of this unique material, its purification still remains a crucial problem. In this article various purification methods for single- and multi-wall carbon nanotubes are reviewed. These methods are compared in terms of their capacity, efficiency, and effects on the tubes. In addition, the use of Raman spectroscopy for monitoring the chromatographic purification of single-wall nanotubes is described. PACS: 78.30.Na; 61.16.Ch; 81.05.Tp; 81.20.Ym Carbon nanotubes have generated great interest since their discovery in 1991 [1]. Applications including electronic devices [2, 3], tools in nanotechnology [4], and hydrogen storage [5] have been proposed. In a standard arc-discharge experiment up to 500 mg of deposit with approximately one third multi-wall nanotubes (MWNTs) can be obtained [6]. In the optimised process for single-wall nanotubes (SWNTs) the daily production reaches 2–3 g of raw material, containing approximately 70% SWNTs [7, 8]. The laser ablation method [9] yields up to 20 g of raw material per day containing up to 50% SWNTs [10]. These methods however, yield an inhomogeneous raw product, consisting of tubes which are diverse in diameter and length. In addition, carbon species such as fullerenes, polyhedra, and other graphitised carbon structures as well as amorphous carbon are found. The different carbon species are closely entangled, occasionally even via chemical bonds. In the case of SWNTs, ropes consisting of up to 100 individual tubes are formed. In addition, metal cluster impurities may be attached to the ends of ropes. They originate from the catalyst (usually Ni, Co, Fe), which are necessary for the formation of the single-layer structures. The metal clusters are usually covered by amorphous carbon and interconnect the SWNTs in webs. Hence, in order to exploit the enormous potential for technical applications, effective purification of this material becomes necessary. Since NTs are insoluble in any solvent, it is hard to directly apply physical purification methods such as filtration

and sedimentation. On the other hand, a chemical purification method, sensitive to the only small differences in reactivity of the different carbon species, would be required. To quantify the purification process, mainly microscopic techniques such as scanning or transmission electron microscopy (SEM, TEM) have been used, by which the various carbon species can be identified. It has to be noted, that SEM, in contrast to TEM, is not able to resolve carbon nanoparticles and thin coatings of amorphous carbon. In general, microscopic techniques investigate only a small part of the sample. Therefore, the volume fraction of each carbon species has to be estimated by averaging over a number of sample images and no exact guideline for this procedure exists. The same applies to scanning probe microscopies, which are even more time consuming. Nevertheless, for imaging very thin ropes or individual SWNTs they are superior to electron microscopy [11]. In summary, the evaluation of NT purity by microscopic techniques partly depends on the subjective guidelines of the experimentalist. Alternatively, purification processes may be monitored in bulk by spectroscopic investigations. For example, Raman spectroscopy is well suited for investigation of NTs dispersed in an aqueous medium and is able to distinguish between different carbon species [12, 13]. The present article summarises the recent developments in purification of NTs. We contrast the differences between destructive and non-destructive purification methods for both, MWNTs and SWNTs. Chromatographic purification, which yields high purity samples of well-separated NTs, is presented in more detail. Examples of characterisation of thus purified SWNTs by scanning force microscopy (SFM) and Raman microscopy are given. 1 MWNT purification 1.1 Oxidation MWNTs, obtained by the arc-discharge method, are usually well graphitised and have a lower density of defects or strained sites (including pentagons) than most of the other carbon species in the raw sample. Consequently, several

270

methods have been aimed at selectively oxidising the carbon impurities, since an attack should preferentially take place on strained sites or defects. During gas phase oxidation, the soot is exposed to air or pure oxygen at high temperatures [14, 15], whereas for liquid phase oxidation, the raw material is attacked by strong oxidising agents such as HNO3 , KMnO4 , or OsO4 in water [16–18]. Small nanoparticles are completely decomposed [19] during the oxidative treatment. The MWNTs are also attacked, but due to their size relatively large tubes survive a strong oxidation. The oxidation of tubes preferentially takes place at the tips, but also along the tube body. As a result, thinned MWNTs with open caps and chemical functions, such as hydroxyl, carbonyl, or carboxylic (−OH, =C=O, −COOH) on the outer shell, [6, 20] are obtained. The opening of the tubes is attributed to the presence of defects and pentagons, located at the tube tips [21]. Recent investigations have revealed that the sensitivity of the outer shell to oxidation depends on curvature, faceting, and helicity [22].Therefore, not only the defect density of the carbon compounds, but many other factors determine their oxidation rates. This could explain the different results obtained by different authors. For example, Rodewald and co-workers performed a detailed thermogravimetric inspection during gas phase oxidation, as well as intensive TEM studies of soot material purified with different oxidising agents in water (liquid phase oxidation). Their studies could not reveal any useful difference in the oxidation rates of MWNTs and of the other carbon species [23]. In addition, large graphitic clusters, which occur relatively often in the raw material, could not be completely removed by oxidation for yields as low as 1%. 1.2 Filtration and flocculation Bonard et al. have introduced a non-destructive separation method for MWNT soot [24]. The MWNTs are dispersed with the aid of ultrasound in an aqueous surfactant solution and become incorporated into micelles. This incorporation is reversible and does not alter the tubes chemically. Subsequent centrifugation removes a large fraction of macroscopic particles and amorphous carbon. After two successive filtrations through track-etched membranes during strong sonication, 90% MWNTs are obtained. In addition, a length separation of the suspended MWNTs can be achieved by flocculation. By raising the surfactant concentration in the suspension the micelles are forced to aggregate, leading to a length-dependent precipitation of the MWNTs, which starts with the longest. 1.3 Chromatography Size exclusion chromatography (SEC) has been successfully employed to purify micellar suspended MWNTs [25]. In this method, the raw material is dispersed with an ultrasonic tip in an aqueous 1 mass % sodiumdodecylsulfate (SDS) solution. The suspension is eluted through a column filled with controlled pore glass. During elution through the highly porous matrix, the diffusion of large particles, such as NTs, into the pores is hindered. Therefore a separation by size takes place and various fractions of pure and length-selected MWNTs in aqueous suspension are collected. In a typical experiment, 50 mg raw soot can be purified in four hours with

Fig. 1. TEM image of chromatographically purified MWNTs. The tubes are well separated and have a narrow length distribution

overall material loss of maximally 10%. In Fig. 1 a TEM image of chromatographically purified MWNTs adsorbed on a carbon-coated grid is shown. As well as their high purity, the MWNTs are well separated and fall into a narrow length range. On closer inspection, it is recognised that the tube tips are undamaged. SFM and SEM investigations of MWNTs adsorbed on modified silica surfaces revealed length-separated MWNTs in the different chromatographic fractions (see also Sect. 2.4). 2 SWNT purification 2.1 Oxidation To remove the amorphous carbon and catalytic particles many researchers heat SWNT raw material in nitric acid [26–28]. The employed reaction times, temperatures, and concentrations of HNO3 vary considerably, but similar results are obtained by the different groups. Vaccarini et al., for example, stir the raw material for 24 h in 2.5 mol/l HNO3 at 100 ◦ C [28]. The acid-treated SWNTs and ropes of SWNTs are covered by a debris, which consists of decomposition products [27]. Due to the presence of carboxylic acid groups, the debris can be washed away with a basic aqueous solution, leaving 25%–50% of the starting material. Higher purity can be achieved by an additional filtration step. For this purpose, the washed material is suspended in water with the aid of a non-ionic surfactant. To prevent the blocking of pores, cross-flow filtration systems are employed. In order to remove the surfactant residues subsequent to filtration, the tubes are annealed under nitrogen at 350 ◦ C for 24 h. This is followed by a treatment for 24 h at 1600 ◦ C to remove the remaining metal particles [29]. For 10 g of raw material, 1 g of very pure (> 90%) SWNTs are obtained by this method.

271

Tohji et al. introduced the hydrothermally initiated dynamic extraction (HIDE) treatment. This method consists of disintegrating the raw material for 12 h in water at 100 ◦ C, followed by an oxidation step in air [30, 31]. Afterwards, hydrochloric acid is added to dissolve the now unshielded catalytic particles. In this way, 95% pure SWNTs are obtained. However, an overall yield of merely 2% is reported. Oxidative methods have the advantage that metal particles and amorphous carbon are almost completely removed. Moreover, they are cheap and capable of purifying large quantities of material. However, there are several indications that the SWNTs are at least partially attacked during the acid treatment. The material loss during oxidation is larger than the content of impurities, which means that some SWNTs are destroyed. In this respect, Durjardin et al. have proposed that the outer SWNTs of a bundle are decomposed and form the debris, which protects the inner tubes from further oxidation [27]. The sensitivity of SWNTs towards an oxidative degradation is underlined by the fact that lower onset temperatures and higher oxidation rates are found compared to MWNTs [22]. This behaviour is explained by the high strain of the graphene layers due to the small diameter of the SWNTs. An important question arises as to the extent of damage to the SWNTs, in the purified sample. After the acid treatment and washing step, SWNTs can be dispersed in water (especially upon ultrasonic agitation), which is not the case for pristine material. HNO3 -treated SWNTs have also been used as a starting material for templates. After a second oxidation step and the adsorption of a polyelectrolyte layer, cluster particles could be attached onto the surface-modified tube [32]. These observations indicate the presence of functionalised SWNTs in the purified samples, which are expected to have considerably different properties to those of undamaged tubes. The functional groups on the tube surface can be regarded as defects which may distort the conjugated π-system. Raman investigations on oxidised SWNTs provide evidence for such a modification of the SWNTs. In Fig. 2, Raman spectra [33] of SWNT material before and after treatment in concentrated ni-

tric acid for 3 h at 130 ◦ C are shown. The spectrum of the pristine material shows a split G-line, as expected for SWNTs [34]. According to Pimenta et al. [35] the pronounced modes around 1540 cm−1 originate from metallic SWNTs, which are resonantly enhanced for excitation with an energy of ≈ 1.9 eV (as is the case in this experiment). However, in the spectrum of the oxidised material the low energy modes of the G-line around 1540 cm−1 are not visible and the remaining modes are upshifted. These changes upon oxidation indicate the loss of metallic character due to functionalisation or a selective decomposition of metallic SWNTs. The additional contribution to the large broad peak of disordered carbon around 1350 cm−1 (D-line) in the spectrum of the oxidised SWNTs, might originate from disordered or carboxylated carbon, which is formed upon oxidation. For comparison, the spectrum of the debris exhibits a broad peak around 1350 cm−1 and no radial breathing modes (RBM) indicating the completed decomposition of SWNTs. Although nitric-acid-treated samples might also contain intact SWNTs, the extent of chemical modification in oxidised samples should be carefully evaluated in future experiments. Vacuum annealing at 1200 ◦ C is suggested to heal the defects, but so far it has not been proven that the treatment completely reconstructs the SWNTs [26]. 2.2 Centrifugation As mentioned above, it is possible to disperse SWNT networks with ultrasonic agitation in aqueous surfactant solutions. Bandow et al. dispersed the raw material in an aqueous 0.1 mass % solution of benzalkonium chloride and concentrated the SWNTs by multiple centrifugations [36]. In this way the SWNT content of the sample is raised from 3%–5% to 40%–70%. SWNTs dispersed with the aid of ultrasonic agitation in an aqueous 1 mass % solution of SDS can be purified by a single centrifugation for 30 min. Most of the material is collected in the sediment, but very pure tubes remain in low concentration in the supernatant. This simple method supplies purified SWNTs suitable for adsorption experiments [37] or the fabrication of Langmuir–Blodgett films [38]. 2.3 Filtration

Fig. 2. Raman spectra (excitation at (λ = 632.8 nm) of SWNT raw material, SWNTs after oxidation, and the debris. For the oxidised SWNTs the lowenergy modes (around 1540 cm−1 ) of the G-line disappeared and the highenergy mode is upshifted to 1590 cm−1 . The debris exhibits a pronounced D-line, but has no RBM, which is characteristic of SWNTs

In general, filtration of SWNTs suffers from blocking of the filter pores. To overcome this problem, Shelimov et al. filtrate SWNTs suspended in methanol during agitation with an ultrasonic tip [39]. After 2.5–6 h of filtration, SWNTs with more than 90% purity are found on the filter membrane. There is, however, some evidence that the tubes are cut and damaged during the intensive and long sonication. Bandow et al. first dissolve the fullerenes with CS2 before the material is suspended in an aqueous surfactant solution [40]. A microfiltration cell allows filtration under stirring and overpressure. The filtration step is repeated three times, with a final yield of 90% pure SWNTs. The authors noted that this method is only appropriate for soot material containing already high amounts (> 50%) of SWNTs.

272

2.4 Chromatography Chromatography has also been successfully used to purify SWNTs samples [41]. In four hours, 10 mg of raw material can be purified by using a 2.5 cm × 46 cm column. Highly pure and length-selected SWNTs in aqueous suspension are obtained. For characterisation, the different fractions were deposited on surface-modified silicon wafers. In Fig. 3, a SFM image of SWNTs on a wafer is shown together with the section analysis along the two black lines. The AFM profiles show that the height of the objects ranges from 1–4 nm. It is assumed that objects smaller than 2 nm are individual SWNTs, whereas larger objects are small ropes of SWNTs. Detailed investigations showed that approximately one half of the objects are individual SWNTs, demonstrating that chromatography supplies well-separated SWNTs. Furthermore, a statistical evaluation of tube dimensions in subsequent fractions revealed a significant difference in the average length of the SWNTs/ropes. As a representative example, the average length was determined to 1.3 µm, 1 µm, and 0.6 µm for the fractions 2, 3, and 5, respectively. Raman spectra of the various liquid fractions can be recorded to characterise the purification process not only on a microscopic scale, but also in bulk [42]. The D-line around 1350 cm−1 and the G-line, centred at 1580 cm−1 , is associated with the carbon impurities and the SWNTs in a sample, respectively [39, 42]. Their relative intensities provide a measure for purity, although it is not possible to use this as an absolute value, because the intensities depend on many different factors, for example resonant enhancement effects. In Fig. 4, the spectra of four different fractions of the same chromatographic separation are shown. Although the intensity of the D-line is nearly constant, the intensities of the

Fig. 4. Raman spectra of four different liquid fractions of a chromatographic separation. The intensity of the G-line, which originates from the SWNTs, decreases with fraction number, whereas the D-line, related to impurities, is nearly constant

G-line clearly decrease with increasing fraction number, indicating a lower SWNT content. In Fig. 5, the peak area ratio A1580 /A1350 is plotted for the fractions of a chromatographic separation. The change in this ratio clearly proves the successful purification, and provides a simple way to monitor the process. SWNTs purified by chromatography facilitate the investigation of the properties of individual tubes. For example, Raman microscopy could be performed on individual SWNTs adsorbed onto substrates for surface enhanced Raman spectroscopy (SERS) [43]. The coverage of the substrates with chromatographically purified SWNTs can be controlled by the adsorption time, so that on average only a few are present

Fig. 3. AFM image of chromatographically purified SWNTs adsorbed on a surface-modified silica substrate. The section analysis along the black lines reveals that the objects are individual SWNTs or small ropes (d < 4 nm) of SWNTs

273

even a single mode in the low-frequency region of the RBM. In Fig. 7, three spectra, recorded on three different spots of a SERS substrate, are displayed. Each spectrum exhibits only one narrow peak in the low-frequency region, as recognised in the inset of Fig. 7 (the feature at 181 cm−1 is an instrumental artefact). RBMs that could be fitted by single Lorentzians with a linewidth (FWHM) smaller than 10 cm−1 are supposed to originate from individual SWNT or a rope of SWNTs with uniform diameter. This FWHM is in the range of the expected natural linewidth of the RBM [44]. Thus, chromatographic purification combined with SERS allows Raman spectroscopic characterisation of individual tubes. Fig. 5. Ratio between the peak area of the D-line and G-line of the raw sample and the fractions of a chromatographic separation

in an area of 1 µm2 . The SEM image in Fig. 6 shows an example of such spatially separated individual SWNTs or small ropes of SWNTs absorbed on a rough silver surface. Using a Raman microscope with a laser spot of approximately 1 µm in diameter, these substrates allow the recording of a spectrum of only a few objects. In contrast to bulk samples, the spectra exhibit distinct features at the G-line and only a few or

Fig. 6. SEM image of SWNTs on a substrate for SERS investigations. A few tubular structures can be recognised on the rough silver surface, indicating that only a few objects are in the area of the incident laser spot (d ≈ 1 µm) of the Raman microscope

3 Conclusions Compared to other fields of NT research, for example synthesis or electrical transport investigations, relatively low progress has been achieved in NT purification during the last year. Until now, no purification method that fulfils all requirements for technical processing of NTs is available. However, in many cases one of the various purification methods satisfies the requirements of the experimentalist. Oxidative methods have the advantage that they are cheap and easy to apply. Yields are relatively low, especially in the case of MWNTs, but very pure samples have been reported. The major disadvantage of such chemical treatments is that the tubes are chemically altered. For some applications, the introduction of surface functions may be tolerated or even required (for example for composite materials), but especially in the case of SWNTs, one must be aware that the properties of the NTs might be changed. Incorporation of NTs into micelles is reversible and does not chemically alter the NTs, opening the possibility to apply physical sorting methods. Large-scale methods such as filtration and sedimentation can be easily performed in good yields. On the other hand, these methods are time consuming because a lot of steps are involved, and even then purity of the tubes is often not sufficient. Chromatography has proven to provide length-selected NTs of high purity, but so far the material throughput is low. To summarise, purification still remains the bottle neck for technical applications, especially where large amounts of material are required. Acknowledgements. We are grateful to P. Redlich (Max-Planck-Institut fuer Metallforschung, Stuttgart) for valuable discussions, as well as C. Journet, (Univ. Montpellier), and W. Maser (Instituto de Carboquimica, Zaragossa) for supplying us with SWNT samples. This work was supported by the NAMITECH contract ERBFMRX-CT96-0067 (DG12 - MIHT).

References

Fig. 7. Three Raman spectra recorded on different spots of a SERS substrate. The spectra exhibit a well-resolved G-line and only one narrow mode, the RBM region (see inset)

1. S. Iijima: Nature 354, 56 (1991) 2. W.A. de Heer, J.-M. Bonard, K. Fauth, A. Châtelain, L. Forr´o, D. Ugarte: Adv. Mater. 9, 87 (1997) 3. S.J. Tans, M.H. Devoret, H. Dai, A. Thess, R.E. Smalley, L.J. Geerligs, C. Dekker: Nature 386, 474 (1997) 4. H. Dai, J.H. Hafner, A.G. Rinzler, D.T. Colbert, R.E. Smalley: Nature 384, 147 (1996) 5. A.C. Dillon, K.M. Jones, T.A. Bekkedahl, C.H. Kiang, D.S. Bethune, M.J. Heben: Nature 386, 377 (1997) 6. H. Hiura, T.W. Ebbesen, K. Tanigaki: Adv. Mater 7, 275 (1995) 7. T.W. Ebbesen, P.M. Ajayan: Nature 358, 220 (1992) 8. C. Journet, W.K. Maser, P. Bernier, A. Loiseau, M.L. de la Chapelle, S. Lefrant, P. Deniard, R. Lee, J.E. Fischer: Nature 388, 756 (1997)

274 9. A. Thess, H. Dai, R.E. Smalley: Science 273, 483 (1996) 10. A.G. Rinzler, J. Liu, H. Dai, P. Nikolev, C.B. Huffman, F.J. RodriguezMacias, P.J. Boul, A.H. Lu, D. Heymann, D.T. Colbert, R.S. Lee, J.E. Fischer, A.M. Rao, P.C. Eklund, R.E. Smalley: Appl. Phys. A 67, 29 (1998) 11. J. Muster, M. Burghard, S. Roth, G.S. Duesberg, E. Hern´andez, A. Rubio: J. Vac. Sci. Technol. B 16, 2796 (1998) 12. M.S. Dresselhaus, G. Dresselhaus, P.C. Eklund: Science of Fullerenes and Carbon Nanotubes (Academic Press, New York 1996) 13. J.M. Holden, P. Zhou, X.-X. Bi, P.C. Eklund, S. Bandow, R.A. Jishi, K.D. Chowdhury, G. Dresselhaus, M.S. Dresselhaus: Chem. Phys. Lett. 220, 186 (1994) 14. L.S.K. Pang, J.D. Saxby, S.P. Chatfield: J. Phys. Chem. 97, 6941 (1993) 15. S.C. Tsang, P.J.F. Harris, M.L.H. Green: Nature 362, 520 (1993) 16. S.C. Tsang, Y.K. Chen, P.J.F. Harris, M.L.H. Green: Nature 372, 159 (1994) 17. B.C. Satishkumar, A. Govindaraj, J. Mofokeng, G.N. Subbanna, C.N.R. Rao: J. Phys. B 29, 4925 (1996) 18. K.C. Hwang: J. Chem. Soc., Chem. Commun. 173 (1995) 19. R. Satio, G. Dresselhaus, M.S. Dresselhaus: Physical Properties of Carbon Nanotubes (Imperial College Press, London 1998) 20. T.W. Ebbesen, H. Hiura, M.E. Bisher, M.M.J. Treacy, J.L. ShreeveKeyer, R.C. Haushaiter: Adv. Mater. 8, 155 (1996) 21. P.M. Ajayan, T.W. Ebbesen, T. Ichihashi, S. Iijima, K. Tannigaki, H. Hiura: Nature 362, 522 (1993) 22. N. Yao, V. Lordi, S.X.C. Ma, E. Durjardin, A. Krishnan, M.M.J. Treacy, T.W. Ebbesen: J. Mater. Res. 13, 2432 (1998) 23. D. Rodewald, P. Redlich: internal report, Max-Planck-Institute für Metallforschung, 1998 24. J.-M. Bonard, T. Stora, J.-P. Salvetat, F. Maier, T. Stöckli, C. Duschl, L. Forro, W.A. de Heer, A. Chatelain: Adv. Mater. 9, 827 (1997) 25. G.S. Duesberg, M. Burghard, J. Muster, G. Philipp, S. Roth: Chem. Commun. 3, 435 (1998) 26. J. Liu, A.G. Rinzler, H. Dai, J.H. Hafner, R. Kelley Bradley, P.J. Boul, A. Lu, T. Iverson, K. Shelimov, C.B. Huffman, F. Rodriguez-Macias, Y.-S. Shon, T.R. Lee, D.T. Colbert, R.E. Smalley: Science 280, 1253 (1998)

27. E. Durjardin, T.W. Ebbesen, A. Krishnan, M.M.J. Treacy: Adv. Mater. 10, 611 (1998) 28. L. Vaccarini, C. Goze, R. Aznar, V. Micholet, C. Journet, P. Bernier: Synt. Met. 103, 2492 (1999) 29. L. Vaccarini: private communication 30. K. Tohji, T. Goto, H. Takahashi, Y. Shinoda, N. Shimizu, B. Jeyadevan, I. Matsuoka, Y. Saito, A. Kasuya, T. Ohsuna, K. Hiraga, Y. Nishina: Nature 383, 679 (1996) 31. K. Tohji, H. Takahashi, Y. Shinoda, N. Shimizu, B. Jeyadevan, I. Matsuoka, Y. Saito, A. Kasuya, S. Ito, Y. Nishina: J. Phys. Chem. B 101, 1974 (1997) 32. J. Muster, G.S. Duesberg, S. Roth, M. Burghard: this issue 33. All Raman spectra reported were measured using a ISA Labram 1B microscope at λ = 632.8 nm 34. A. Kasuya, Y. Sasaki, Y. Saito, K. Tohji, Y. Nishina: Phys. Rev. Lett. 72, 4434 (1997) 35. M. Pimenta, A. Marucci, S.D.M. Brown, M.J. Matthews, A. Rao, P.C. Eklund, R.E. Smalley, G. Dresselhaus, M.S. Dresselhaus: J. Mater. Res. 13, 2396 (1998) 36. S. Bandow, S. Asaka, X. Zhao, Y. Ando: Appl. Phys. A 67, 23 (1998) 37. M. Burghard, G.S. Duesberg, G. Philipp, J. Muster, S. Roth: Adv. Mater. 10, 584 (1998) 38. V. Krstic, G.S. Duesberg, J. Muster, M. Burghard, S. Roth: Chem. Mater. 10, 2338 (1998) 39. K.B. Shelimov, R.O. Rinat, A.G. Rinzler, C.B. Huffmann, R.E. Smalley: Chem. Phys. Lett. 282, 429 (1998) 40. S. Bandow, A.M. Rao, K.A. Williams, A. Thess, R.E. Smalley, P.C. Eklund: J. Phys. Chem. B 101, 8839 (1997) 41. G.S. Duesberg, J. Muster. V. Krstic, M. Burghard, S. Roth: Appl. Phys. A 67, 117 (1998) 42. G.S. Duesberg, W. Blau, H.S. Byrne, S. Muster, M. Burghard, S. Roth: Synt. Met. 103, 2484 (1999) 43. M. Lamy de la Chapelle, S. Lefrant, C. Journet, W. Maser, P. Bernier: Carbon 36, 705 (1998) 44. G.S. Duesberg, W. Blau, H.S. Byrne, S. Muster, M. Burghard, S. Roth: unpublished 45. S. Bandow, S. Asaka, Y. Saito, A.M. Rao, L. Grigorian, E. Richter, P.C. Eklund: Phys. Rev. Lett. 80, 3779 (1998)