Journal of Physics and Chemistry of Solids 61 (2000) 10311036 www.elsevier.

nl/locate/jpcs

Large scale synthesis of single-wall carbon nanotubes by arc-discharge method

Zujin Shi a, Yongfu Lian a, Fu Hui Liao a, Xihuang Zhou a, Zhennan Gu a,*, Y. Zhang b, S. Iijima b, Hongdong Li c, Kwok To Yue c, Shu-Lin Zhang c
b a Department of Chemistry, Peking University, Beijing 100871, Peoples Republic of China Fundamental Research Laboratories, NEC Corporation, 34 Miyukigaoka Tsukuba, Ibaraki 305-8501, Japan c Department of Physics, Peking University, Beijing 100871, Peoples Republic of China

Abstract Using a graphite rod with a hole lled with the powder of a mixture of YNi alloy and graphite or calcium carbide and nickel as anode, single-wall carbon nanotubes (SWCNTs) with high yield were produced in quantity of tens of grams a day under the arc conditions of 40 60 A d.c. and helium pressure of 500 or 700 torr. The yield of SWCNTs can be quantitatively determined by thermogravimetric analysis (TGA) method. The morphology and structure of SWCNTs were observed by SEM, HREM and Raman spectroscopy. The results showed that the helium atmosphere strongly affects the yield of SWCNTs and the diameter distributions of SWCNTs are different when different catalyst was used, proving that the diameter of SWCNTs is dependent on the properties of the metal catalysts. Our results suggested a formation mechanism of SWCNTs and the roles played by nickel and yttrium or calcium atoms. 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Single-wall carbon nanotube; Production; Growth mechanism; Arc-discharge method

1. Introduction Single-wall carbon nanotubes (SWCNTs) have many potential applications including material as a good conductor [1], quantum wires [2], nano-devices [2] and eld-effect transistors [3], etc. Scientists everywhere are eager to nd a cheap way to produce large quantities of SWCNTs. Up to now, laser ablation [4,5] and d.c. arc-discharge method [6 8] are the most common ways to synthesize SWCNTs. With their amplied apparatus, Rinzler et al. could produce 20 g of 40 50 vol.% SWCNTs material in 48 h of continuous, largely unattended, operation [5]. Compared to the laser ablation method, the d.c. arc-discharge method can get more SWCNTs. Jounet et al. [6] obtained high yield of SWCNTs in the collar of the deposit with Y/Ni as catalyst. However, it is believed that a lot more products should be
This work was supported by the National Natural Science Foundation of China, no. 29671030. * Corresponding author. E-mail address: guzn@chemms.chem.pku.edu.cn (Z. Gu).

obtained in the soot. More recently, we synthesized SWCNTs in the cloth-like soot in quantity of grams with 40% purity by the arc-discharge method with YNi alloy as catalyst under 500 torr helium atmosphere. It was found that high helium pressure seems to favor the formation of SWCNTs [7]. Here, we report the large-scale production of SWCNTs under the arc conditions of 40 60 A d.c. and helium pressure of 500 or 700 torr by using a graphite rod with a hole lled with the powder of a mixture of YNi alloy and graphite or calcium carbide and nickel as anode. The yield of SWCNTs was quantitatively determined by thermogravimetric analysis (TGA) method. The morphology and structure of SWCNTs were observed by scanning electron microscope (SEM), high-resolution transmission electron microscope (HREM) and Raman spectroscopy. The results showed that the helium atmosphere strongly affects the yield of SWCNTs and the diameter distributions of SWCNTs are different when different catalyst was used, proving that the diameter of SWCNTs is dependent on the properties of the metal catalysts. Our results suggested a formation mechanism of SWCNTs and the roles played by nickel and yttrium or calcium atoms.

0022-3697/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S0022-369 7(99)00358-3

1032

Z. Shi et al. / Journal of Physics and Chemistry of Solids 61 (2000) 10311036

Fig. 1. TGA curves of the raw-soot produced with (a) YNi or (b) CaC2 Ni as catalyst under 700 torr of helium pressure and with (c) YNi as catalyst under 700 torr of helium pressure, in which the bold, dash and dot lines correspond to TG, DTG and Lorentzian t of DTG curve, respectively. (a) shows two stepwise weight-loss, amorphous carbon and SWCNTs standing 38.3% and 61.3%, respectively. (b) shows four stepwise weight-loss, amorphous carbon, SWCNTs, fullerenes and CNPs standing 55.9, 37.0, 5.3 and 1.8%, respectively. (c) shows two stepwise weight-loss, amorphous carbon and SWCNTs standing 52.1 and 38.4%, respectively.

Z. Shi et al. / Journal of Physics and Chemistry of Solids 61 (2000) 10311036

1033

2. Experimental The SWCNTs were produced by the d.c. arc-discharge method similar to the one reported earlier [7]. The anode was a f 6 150 mm 2 graphite rod (spectroscopy pure) with a f 4 100 mm 2 hole drilled and then lled with a powder mixture of YNi alloy (or CaC2/Ni) and graphite in metal/C atomic ratio of 310. An arc was generated between the anode and a sharp top graphite cathode at 40 A in 500 or 700 torr of helium static atmosphere. The gap distance of the electrodes was maintained at 10 mm by continuously translating the anode throughout the arc process. Later, cloth-like soot formed on the entire inner wall of the chamber. In general, it took 2 h to spend a 10 cm anode rod to yield 5 g of soot. If the composite rods were directly made from the mixture of graphite and catalyst powder, it was possible to obtain 60 g of soot a day. TGA was performed in a 951 Thermogravimetric Analyzer, DuPont Instruments. The conditions are 10C/ min ramp rate from room temperature to 900C at an air ow rate of 70 s/cm 3. The raw soot was characterized by SEM (S-4200, Hitachi), HREM (Topcon 002B, working at 200 kV) and Raman spectroscopy at room temperature (Renishaw System 1000, excited with 782 nm radiation).

3. Results and discussion 3.1. Thermogravimetric analysis of the raw-soot Fig. 1 shows the TGA graphs of the raw-soot produced

with YNi (Fig. 1a) and CaC2 Ni (Fig. 1b) as catalyst under 700 torr helium atmosphere. In each gure, the bold, dash and dot lines correspond to TG, DTG and Lorentzian t of DTG curve, respectively. From the gures, it can be observed that the initial burning temperature of the rawsoot is about 200C and there are two stepwise weightloss in the case of Fig. 1a but four stepwise in the case of Fig. 1b which can be assigned to amorphous carbon, SWCNTs, fullerenes and carbon nanoparticles (CNPs) successively from low temperature, according to literature [3,9] as well as our blank tests. For the former case, it is still unclear that the peaks indicating the weight-loss of fullerenes and CNPs in the third and fourth steps did not appear. The initial burning temperature of SWCNTs is close to that of the raw-soot, but the temperature with highest burning rate is higher than that of amorphous carbon. Therefore, 350C is an effective temperature to burn out amorphous carbon. It is well known that DTG is a common method for quantitative analysis, the content of each species corresponding to its peak area in DTG. Thus, it can be calculated that the yield of SWCNTs is 62% in the soot produced with YNi as catalyst and 37% in the soot produced with CaC2 Ni as catalyst neglecting the variation of metal weights, which are very close to the qualitative estimation with EM and Raman technique. This indicates that high yield of SWCNTs can be obtained using either YNi or CaC2 Ni as catalyst while the former is more effective. In the same manner, the yield of SWCNTs in the soot produced with YNi as catalyst under 500 torr helium atmosphere is 38% (Fig. 1c), which was close to the estimated result of EM and Raman technique [7]. This indicates that the yield of SWCNTs can

Fig. 2. The surface SEM image of raw cloth-like soot produced with YNi as catalyst under 700 torr of helium pressure, showing high density of SWCNT bundles which have a diameter of 2030 nm.

1034

Z. Shi et al. / Journal of Physics and Chemistry of Solids 61 (2000) 10311036

Fig. 3. HREM images showing: (a) a ower-like SWCNTs stacked of short SWCNTs in different directions; (b) individually grown SWCNTs with a diameter as large as 1.6 nm, a kink is indicated by a bold arrow. Encapsulation of C60 molecules (indicated by open arrows) in a SWCNT is also shown in (b); (c) a radial SWCNT bundles with a carbon nanoparticle core.

be dramatically elevated with the increment of helium pressure. 3.2. Morphology and structure of SWCNTs Fig. 2 shows a typical image of the cloth-like soot for the case of YNi alloy as catalyst under 700 torr helium pressure, where high density of SWCNT bundles can be found together with other particles on most of the surface

of the soot. The bundles generally have diameters of 20 30 nm and lengths longer than 15 mm. The exact length cannot be measured because no bundle was observed with both ends of bundle at the same time. The morphology of SWCNTs had been discussed in detail in our earlier report [12]. Short SWCNTs with an aspect ratio less than 20 can be formed into a ower-like SWCNTs (Fig. 3a) which is stacked by many short SWCNTs in different directions. Individually grown SWCNTs may have a diameter as large as 1.6 nm and sometimes contain kinks, as indicated by a bold arrow in Fig. 3b. In the same picture, we can see a row of closed shells inside a SWCNT (indicated by hollow arrows). The diameter of the shells is about 0.7 nm, nearer to that of C60 molecules. This observation is very similar to that using laser ablation samples by Smith et al. [11], where they interpret the closed shells as encapsulated C60 molecules in SWCNTs. The encapsulation of C60 in the nanotube implies an open-end growth scheme, which allows mass transportation into the nanotubes. Similarly, to the sea urchin reported by Subramoney et al. [13], the radial SWCNTs with a core of hollow multi-layer carbon nanoparticle can also be found in the sample (Fig. 3c), which is stacked by SWCNT bundles with different directions. In this gure, the fringes of nanoparticle are overlapped by SWCNT bundles, which implies that the particle is not the core for the growth of SWCNTs. The diameters of SWCNTs produced with the above two catalysts as seen from their TEM images are, surprisingly, very different. Most of those produced with YNi as catalyst have a diameter ranging from 1.3 to 1.4 nm (as shown in Fig. 4a), which is the same as those produced by the laserablation method with CoNi as catalyst. In this gure, a at (indicated by a open arrow) and a hemispherical (indicated by a bold arrow) top of SWCNT can be found. In contrast, those with Ca as catalyst have smaller diameters, ranging from 0.9 to 1.1 nm and contain very few SWCNTs with a diameter of 1.3 nm, as shown in Fig. 4b. The diameter difference can also be characterized by the

Fig. 4. HREM images of SWCNTs produced with (a) YNi and (b) CaC2 Ni as catalyst. Note that the SWCNTs produced with YNi as catalyst (a) have a larger diameter than those produced with CaC2 Ni as catalyst (b).

Z. Shi et al. / Journal of Physics and Chemistry of Solids 61 (2000) 10311036

1035

Fig. 5. Room temperature Raman spectra of SWCNTs: (a) produced with YNi and (b) CaC2 Ni as catalysts. Note that the vibrational frequencies of the radial-breathing mode of SWCNTs in (b) are obviously higher.

Raman spectra (Fig. 5). Comparing the two spectra in Fig. 5, it can be seen that the Raman features higher than 300 cm 1 are basically identical but they are very different in the low wavenumber area which contain radial-breathing mode of SWCNTs. Note that the highest vibrational frequency is 207.5 cm 1, corresponding to 1.13 nm diameter of SWCNTs according to the method proposed by Kurti et al. [9], in SWCNTs produced with YNi as catalyst, but is 228.2 cm 1 corresponding to 1.03 nm diameter of SWCNTs in those produced with CaC2 Ni as catalyst. The inset is the magnied gure in the low-frequency area. Many previous studies have reported that the frequency of the radial-breathing mode depends strongly on the diameter of SWCNTs, with lower frequencies corresponding to SWCNTs with larger diameters [9,10]. Therefore, SWCNTs with small diameters exist in the sample

produced with CaC2 Ni as catalyst, a result consistent with TEM images. 3.3. Organic solvent extracts of the raw-soot Typically, fullerenes and metallofullerenes are synthesized under a low pressure of helium atmosphere, less than 300 torr, in the arc-discharge method and the yield of metallofullerenes is very low [14]. In soot obtained above, however, there are rich metallofullerenes. The yield was estimated 10 times higher than that produced in low helium pressure but the yield of total organic solvent extractable substance was almost the same as in the original arc conditions [16]. Fig. 6 is a positive-ion MALDI-TOF massspectrum of CS2 extract from the soot produced with YNi as catalyst under 700 torr of helium atmosphere. It was

Fig. 6. The positive-ion MALDI-TOF mass-spectrum of CS2 extract, in which rich content yttrium metallofullerenes can be found.

1036

Z. Shi et al. / Journal of Physics and Chemistry of Solids 61 (2000) 10311036

proved that the yield of metallofullerenes strongly depends on the existence of nickel apart from high helium pressure. 3.4. Growth mechanism of SWCNTs From the above results and discussion, helium pressure is a very important factor for the high yield production of SWCNTs and metallofullerenes. No cloth-like soot was formed when the helium pressure was less than 300 torr and the yield of SWCNTs increases with the increase of helium pressure [7]. This conrms the suggestion proposed by Saito et al. [15], the vapor pressure of the catalyst metal being an important factor for the formation of SWCNTs. The vapor pressure of the catalyst metal decreases with the increase of helium pressure so that high pressure is benecial for the co-evaporation of metal and carbon. Up to now, the elements which can form metallofullerenes in arc-discharge method are rare earth elements and alkali earth metal. We think that these elements can also promote the formation of SWCNTs. We postulate that these elements are playing a key role in the growth process of SWCNTs because of their acetylide properties. These 2 metals can form chemical bonds with both C2 and other metal such as nickel, and also with the carbon nuclei of fullerenes which have conjugation p bonds. On the other hand, the encapsulation of C60 in the nanotube implies an open-end growth scheme, which allows mass transportation into the nanotubes [11,12]. Thus, in the growth process of nanotubes, the acetylide metal atom (or ion) will stay in the open cap end of the nanotube and at the same time draw nickel atoms (or ions) to the open area of the nanotube. As the growing nanotube moves out of the high temperature area, which supports nanotube growth, the open cap closes and the metal atoms (or ions) leave the end of the nanotube. In this way, the nickel element is the true catalyst but the acetylide metal only plays a linkage role in the formation of carbon nanotubes. This conjecture can reasonably account for the fact that the yield of SWCNTs can be greatly improved with a mixture of two metals as catalyst and that the different diameter distributions of SWCNTs can be obtained when different acetylide metal was used as catalyst.

different when different catalyst was used, proving that the diameter of SWCNTs is dependent on the properties of the metal catalysts. Our results suggested a formation mechanism of SWCNTs and the roles played by nickel and yttrium or calcium atoms.

Acknowledgements The authors would like to thank Miss Chaoying Wang of the Institute of Physics of the Chinese Academy of Sciences for SEM measurements.

References
[1] S.J. Tans, M.H. Devoret, H.J. Dai, et al., Individual singlewall carbon nanotubes as quantum wires, Nature 386 (1997) 474477. [2] P.G. Collions, A. Zettl, H. Bando, A. Thess, R.E. Smalley, Nanotube nanodevice, Science 278 (1997) 100103. [3] R. Martel, T. Schmidt, H.R. Shea, T. Hertel, et al., Single- and multi-wall carbon nanotube eld-effect transistors, Appl. Phys. Lett. 73 (1998) 24472449. [4] A. Thess, R. Lee, P. Nikolaev, et al., Crystalline ropes of metallic carbon nanotubes, Science 273 (1996) 483487. [5] A.G. Rinzler, J. Liu, H. Dai, P. Nikolaev, et al., Large-scale purication of single-wall carbon nanotubes: process, product, and characterization, Appl. Phys. A 67 (1998) 2937. [6] C. Journet, W.K. Maser, P. Bernier, A. Loiseau, et al., Largescale production of single-walled carbon nanotubes by the electric-arc technique, Nature 388 (1997) 756758. [7] Z.J. Shi, Y.F. Lian, X.H. Zhou, Z.N. Gu, et al., Massproduction of single-wall carbon nanotubes by arc discharge method, Carbon 37 (9) (1999) 14491453. [8] Y. Saito, Y. Tani, N. Miyagawa, et al., High yield of singlewall carbon nanotubes by arc discharge using RhPt mixed catalysts, Chem. Phys. Lett. 294 (1998) 593598. [9] J. Kurti, G. Kresse, H. Kuzmany, First-principles calculations of the radial breathing mode of single-wall carbon nanotubes, Phys. Rev. B 58 (1998) R8869R8872. [10] A.M. Rao, E. Richter, S. Bandow, et al., Diameter-selective Raman scattering from vibrational modes in carbon nanotubes, Science 275 (1997) 187191. [11] B.W. Smith, M. Monthioux, D.E. Luzzi, Encapsulated C60 in carbon nanotubes, Nature 396 (1998) 323324. [12] Y. Zhang, S. Iijima, Z.J. Shi, Z.N. Gu, Defects in arc discharged single-walled carbon nanotubes, Phil. Mag. Lett. 79 (1999) 473. [13] S. Subramoney, R.S. Rouff, D.C. Lorentd, et al., Radial singlelayer nanotubes, Nature 266 (1993) 637. [14] H. Shinohara, Endohedral rare earth metallofullerenes, Rare Earths 33 (1998) 7191. [15] Y. Saito, K. Kawabata, M. Okuda, Single-layered carbon nanotubes synthesized by catalytic assistance of rare-earths in a carbon arc, J. Phys. Chem. 99 (1995) 16 07616 079. [16] Y.F. Lian, Z.J. Shi, X.H. Zhou, X.R. He, Z.N. Gu, High-yield synthesis of endohedral metallofullerenes Y@C2n and Y2@C2n, Chin. Chem. Lett. 10 (1999) 425428.

4. Conclusion Large scale SWCNTs have been synthesized by using a graphite rod with a hole lled with the powder of a mixture of YNi alloy and graphite or calcium carbide and nickel as anode under helium pressure of 500 or 700 torr. The yield of SWCNTs was quantitatively determined by TGA. The morphology and structure of SWCNTs were observed by SEM, HREM and Raman spectroscopy. The results showed that the helium atmosphere strongly affects the yield of SWCNTs and the diameter distributions of SWCNTs are