D. S. Bethune, C-H. Kiang*, M. S. de Vries, G. Gorman, R. Savoy, J. Vazquez, and R. Beyers

IBM Research Division Almaden Research Center 650 Harry Road San Jose, CA 95120-6099 *Affiliated with the Materials and Molecular Simulation Center, Beckman Institute California Institute of Technology Pasadena, CA 91125

Carbon exhibits a unique ability to form a wide range of structures. In an inert atmosphere it condenses to form hollow spheroidal fullerenes^1-4. Carbon deposited on the hot tip of the cathode of the arc-discharge apparatus used for bulk fullerene synthesis will form nested graphitic tubes and polyhedral particles^5-8. Electron irradiation of these nanotubes and polyhedra transforms them into nearly spherical carbon ‘onions’⁹. We now report that covaporizing carbon and cobalt in an arc generator leads to the formation of carbon nanotubes which all have very small diameters (about 1.2 nm) and walls only a single atomic layer thick. The tubes form a web-like deposit woven through the fullerene-containing soot, giving it a rubbery texture. The uniformity and single-layer structure of these nanotubes should make it possible to test their properties against theoretical predictions^10-13.

The initial aim of our experiments was to produce metallofullerenes and graphite-encapsulated nanocrystals of magnetic atoms. Electrodes were prepared by boring 4 mm diameter holes in 6 mm diameter graphite rods and filling them with mixtures of pure powdered metals (Fe, Ni or Co) and graphite. These filled anodes (~2 at % metal) were vaporized with a current of 95-105 amps in 100-500 torr of He in our arc fullerene generator. The results obtained with cobalt were unique.

When a Co-containing rod was used, what looked like spider webs formed in the chamber, draping between surfaces. The soot on the chamber walls was rubbery and could be peeled off in long strips. Normal fullerene soots (and those made with Fe or Ni containing rods) are crumbly. The soot and the web material were ferromagnetic. A transmission electron microscope (TEM) image of the web material (Figure 1) shows that the web consists of rounded soot particles a few tens of nanometres across, linked together by fine fibres. Individual threads can be traced for several micrometres. In some cases several fibres converge on a soot particle. Embedded within the soot particles are round cobalt clusters with diameters ranging from a few nanometres to roughly 20 nm. Both electron and X-ray diffraction patterns showed that these clusters are face-centered-cubic Co. This indicates that the clusters were rapidly quenched, since cobalt is normally hexagonal-close-packed below 400°C. Scanning electron microscope (SEM) images show that the rubbery soot deposits from the chamber walls contain thin fibres and soot particles similar to those in the web material, but with the particles in greater relative abundance. The carbon around the cobalt clusters consists partly of fullerenes, which can be extracted from the soot in typical amounts using toluene. Laser-desorption/laser-ionization mass spectrometry of the raw soot showed a CoC₆₀ peak, but this species was not found in a toluene extract of the soot.

A higher-magnification TEM image (Figure 2) reveals the structures underlying the fibre and web formation. Carbon nanotubes with single-atomic-layer walls and diameters of 1.2 ±0.1 nm are ubiquitous. The tubules apparently crossed, aggregated and tangled before being encased. Although the tubules are mostly coated with non-graphitic carbon, bare sections are also evident. Figure 3 (at still higher magnification) shows a bare nanotube with several round objects, comparable in size to fullerenes with 60-100 carbons, adhering to it. The circumference of the nanotubes would correspond to a belt of 15 or 16 edge-sharing hexagons with 0.142-nm sides.

Carbon fibres grow under diverse conditions^14-16. In the late 50's Bacon found that graphitic whiskers, 1-5 microns in diameter and centimeters in length, grow on the extremely hot cathode of a carbon arc run in high pressure argon¹⁷. Under similar conditions (but at lower pressures), tubular graphitic structures with 2-30 nm diameters and micrometre lengths form in the cathode deposits in an arc-fullerene generator^5-7. These nanotubes typically have walls 2-50 atomic layers thick.
On the other hand, in the presence of transition-metal catalyst particles, vapour-grown carbon fibre (VGCF) can be produced by pyrolysing a hydrocarbon/carrier-gas mixture at temperatures between 500°C and 1200°C^14-16. Yacaman et al.¹⁸ recently reported that some fibres produced by this method resemble the hollow graphitic tubes seen in fullerene-generator cathode deposits.

In contrast to these multilayered fibres and tubes, our cobalt-catalysed nanotubes have single-atomic-layer walls and a common diameter (~1.2 nm). They grow from carbon vapour (with no dissociation of hydrocarbon needed) at He pressures in the range 100-500 torr. Fullerenes form abundantly at the same time. Under the conditions we used, no fibre growth was observed using Fe, Ni or a 50:50 Ni:Cu mixture, all of which catalyse fibre growth in the presence of hot gaseous
hydrocarbons^19-20. We believe, therefore, that cobalt plays a special role in catalyzing the formation of these single-walled tubules, and suggest that a specific nucleation process may be responsible for their highly uniform diameter. For the moment the relationship between the nanotubes and the cobalt clusters is obscured by the encasing layer of carbon. The nanotubes are found in relatively cold regions of the chamber, co-condensed with (and mostly coated by) fullerene soot. It may be possible to control the amount of carbon that forms on the nanotubes by modifying the growth conditions, and the crystallinity of this carbon by post-annealing the coated nanotubes at high temperature. Such measures have been used to modify vapour-grown carbon fibres¹⁴, and could be important in attempting to exploit the uniformity of these vapour-grown nanotubes to develop new types of carbon fibres.

It may also be possible to isolate bulk quantities of bare, single-walled nanotubes. Such structures constitute a new type of all-carbon polymer. Theoretical calculations predict that they can be metallic or semiconducting, depending on their helical pitch^10-11. They might draw species into their interiors by capillary action, and they may be useful as catalytic containers, nanowires and solenoids²¹. The recent success of Ajayan et al.²² in filling nanotubes with lead supports some of these ideas. The availability of the single-walled carbon nanotubes reported here should permit characterization and further experiments.

Received 24 May; accepted 3 June 1993

1. Kroto, H.W., Heath, J.R., O’Brien, S.C., & Smalley, R.E. Nature 318, 162-163 (1985).

2. Krätschmer, W., Fostiropoulos, K. & Huffman, D.R. Chem. Phys. Lett. 170, 167-170 (1990).

3. Krätschmer, W., Lamb, L.D., Fostiropoulos, K. & Huffman, D.R. Nature 347, 354-358 (1990).

4. Meijer, G. & Bethune, D.S. J. Chem. Phys. 93, 7800-7802 (1990).

5. Iijima, S. Nature 354, 56-58 (1991).

6. Iijima, S., Ichihashi, T. & Ando, Y. Nature 356, 776-778 (1992).

7. Ebbesen, T.W. & Ajayan, P.M. Nature 358, 220-222 (1992).

8. Saito, Y., Yoshikawa, T., Inagaki, M., Tomita, M. & Hayashi, T. Chem. Phys. Lett. 204, 277-282 (1993).

9. Ugarte, D. Nature 359, 707-708 (1992).

10. Hamada, N., Sawada, S. & Oshiyama, A. Phys. Rev. Lett. 68, 1579-1581 (1992).

11. Mintmire, J.W., Dunlap, B.I. & White, C.T. Phys. Rev. Lett. 68, 631-634 (1992).

12. Saito, R., Fujita, M., Dresselhaus, G. & Dresselhaus, M.S. Phys. Rev. B46, 1804-1811 (1992).

13. Robertson, D.H., Brenner, D.W. & Mintmire, J.W. Phys. Rev. B45, 12592-12595 (1992).

14. Endo, M. Chemtech 18, 568-576 (1988).

15. Baker, R.T. Carbon 27, 315-323 (1989).

16. Tibbetts, G.G. J. Cryst. Growth 73, 431-438 (1985).

17. Bacon, R. J. Appl. Phys. 31, 283-290 (1960).

18. Jose-Yacaman, M., Miki-Yoshida, M., Rendon, L. & Santiesteban, J.G. Appl. Phys. Lett. 62, 657-659 (1993).

19. Baker, R.T. & Harris, P.S. in Chemistry and Physics of Carbon, Vol. 14, 83-165 (Marcel Dekker, New York, 1978).

20. Kim, M.S., Rodriguez, N.M. & Baker, R.T. J. Catal. 131, 60-73 (1991).

21. Pederson, M.R. & Broughton, J.Q. Phys. Rev. Lett. 69, 2689-2692 (1992).

22. Ajayan, P.M. & Iijima, S. Nature 361, 333-334 (1993).

We thank W. A. Goddard, III, R. D. Johnson, C. S. Yannoni, C. T. Rettner, and J. R. Salem for helpful discussions. CHK acknowledges partial support by the National Science Foundation and the Materials and Molecular Simulation Center (supported by DOE-AICD, Allied-Signal, BP America, Asahi Chemical, Asahi Glass, Chevron, BF Goodrich and Xerox).

Figure 1. TEM micrograph of web-like material showing strands of thread-like fibres and cobalt clusters (dark spots) embedded in carbon soot
particles.

Figure 2. TEM micrograph at higher magnification showing details of the web-like material. Running through the deposited non-graphitic
carbon are a single-walled nanotubes about 1.2 nm in diameter. Bare portions of these nanotubes are also evident. The dark spot in the upper-right corner is a cobalt cluster.

Figure 3. TEM micrograph of a bare section of a single-walled nanotube. The round objects adhering to the tube have diameters corresponding to fullerenes with 60-100 carbons.