Helical microtubules of graphitic carbon

Sumio Iijima

NEC Corporation, Fundamental Research Laboratories, 34 Miyukigaoka, Tsukuba, Ibaraki 305, Japan

The synthesis of molecular carbon structures in the form of C₆₀ and other fullerenes¹ has stimulated intense interest in the structures accessible to graphitic carbon sheets. Here I report the preparation of a new type of finite carbon structure consisting of needle-like tubes. Produced using an are-discharge evaporation method similar to that used for fullerene synthesis, the needles grow at the negative end of the electrode used for the are discharge. Electron microscopy reveals that each needle comprises coaxial tubes of graphitic sheets, ranging in number from 2 up to about 50. On each tube the carbon-atom hexagons are arranged in a helical fashion about the needle axis. The helical pitch varies from needle to needle and from tube to tube within a single needle. It appears that this helical structure may aid the growth process. The formation of these needles, ranging from a few to a few tens of nanometres in diameter, suggests that engineering of carbon structures should be possible on scales considerably greater than those relevant to the fullerenes.

Solids of elemental carbon in the sp² bonding state can form a variety of graphitic structures. Graphite filaments can be produced, for instance, when amorphous carbon filaments formed by thermal decomposition of hydrocarbon species are subsequently graphitized by heat treatment^2,3. Graphite filaments can also grow directly from the vapour-phase deposition of carbon^4,5, which also produces soot and other novel structures such as the C₆₀ molecule^6–8.

Graphitic carbon needles, ranging from 4 to 30 nm in diameter and up to 1 mm in length, were grown on the negative end of the carbon electrode used in the d.c. are-discharge evaporation of carbon in an argon-filled vessel (100 torr). The gas pressure was much lower than that reported for the production of thicker graphite filaments⁵. The apparatus is very similar to that used for mass production of C₆₀ (ref. 9). The needles seem to grow plentifully on only certain regions of the electrode. The electrode on which carbon was deposited also contained polyhedral particles with spherical shell structures, which were 5–20 nm in diameter. The needle structures were examined by transmission electron microscopy (electron energies of 200 keV).

Figure 1 Electron micrographs of microtubules of graphitic carbon. Parallel dark lines correspond to the (002) lattice images of graphite. A cross-section of each tubule is illustrated. a, Tube consisting of five graphitic sheets, diameter 6.7 nm. b, Two-sheet tube, diameter 5.5 nm. c, Seven-sheet tube, diameter 6.5 nm, which has the smallest hollow diameter (2.2 nm). | high-resolution version |

Figure 2 Clinographic view of a possible structural model for a graphitic tubule. Each cylinder represents a coaxial closed layer of carbon hexagons. The meaning of the labels V and H is explained in the text. | high-resolution version |

High-resolution electron micrographs of typical needles show {002} lattice images of the graphite structure along the needle axes (Fig. 1). The appearance of the same number of lattice fringes from both sides of a needle suggests that it has a seamless and tubular structure. The thinnest needle, consisting of only two carbon-hexagon sheets (Fig. 1b), has an outer and inner tube, separated by a distance of 0.34 nm, which are 5.5 nm and 4.8 nm in diameter. The separation matches that in bulk graphite. Wall thicknesses of the tubules range from 2 to 50 sheets, but thicker tubules tend to be polygonized. This low dimensionality and cylindrical structure are extremely uncommon features in inorganic crystals, although cylindrical crystals such as serpentine¹⁰ do exist naturally.

The smallest tube observed was 2.2 nm in diameter and was the innermost tube in one of the needles (Fig. 1c). The diameter corresponds roughly to a ring of 30 carbon hexagons; this small diameter imposes strain on the planar bonds of the hexagons and this causes two neighbouring hexagons on the ring to meet at an angle of ~6°. For the C₆₀ molecule, the bending angle is 42°, which is much larger than for these tubes. The C–C bond energy calculated for the C₆₀ molecule is smaller than that of graphite¹¹, suggesting that bending the hexagons in C₆₀ lowers the bond energy. A similar effect of the bending on bonding energies might apply here. One of the key questions about the tubular structure is how the ABAB hexagonal stacking sequence found in graphite is relaxed, as it is impossible to retain this ideal graphite structure for coaxial tubes. There should be a shortage of 8–9 hexagons in going from one circumference of a tube to that inside it. Disordered graphitic stacking is known as turbostratic stacking, but no detailed accounts of stacking patterns in such structures have been reported. The argument here is also applicable to the spherical graphitic particles mentioned earlier⁶.

All the electron diffraction patterns (Fig. 3) taken from individual carbon needles are indexed by the {h0l} and {hk0} spots for hexagonal symmetry. The patterns always show strong (00l) spots when the needle axes are perpendicular to the [001] axis, supporting the idea of a coaxial arrangement of graphitic tubes. As shown in Fig. 2, two side portions of each tube (indicated by shading and labelled 'V') will be oriented so that the (002) planes satisfy the Bragg diffraction condition for the incident electron beam, and thus give (00l)-type spots. Individual (002) planes in these portions are directly image in Fig. 1.

The {hk0} patterns as a whole show mm2 mirror symmetry about the needle axis, and consist of multiple sets of {hk0} spots. For example, three sets of (hk0) spots seem to form ring patterns, only (100)- and (220)-type spots being seen (Fig. 3a). The diffraction pattern was obtained from a single tube consisting of seven sheets, confirmed by examining the electron microscope image. Referring again to Fig. 2, the top portion of the outermost tube, labelled ‘H’, and its counterpart on the bottom of the tube, give independently one set each of {hk0} patterns. If these two portions of the cylinder have the same orientation, they produce an identical {hk0} pattern. If three hexagon sheets on the tubes 1, 2 and 3 were oriented differently, they would give six different {hk0} spot patterns. Such a top-bottom effect is one of the requirements for the mirror symmetry. Another requirement is a helical arrangement of carbon hexagons on individual tubes, described further below.

Consider rotation of individual graphite sheets with respect to the needle axes. To explain the graphitic tube structure, the tube is cut at one side along the needle axis and unrolled. This is illustrated schematically in Fig. 4a. Fewer hexagons are drawn, than would form a real tube, but the essential needle geometry correctly represents one of our experimental diffraction patterns. A cylindrical tube can be formed by rolling up the hexagonal sheet about the filament axis (drawn by the heavy line) so as to superimpose hexagons labelled A and B at the top on A' and B' at the bottom. It will be found that the hatched hexagons are aligned perfectly to make a helix around the needle axis (see Fig. 4b). The helical arrangement of the hexagons is responsible for the mirror symmetry as observed in the experimental diffraction patterns. One complete spiral rotation leaves a pitch three hexagons in height at the tip of the needle. There are many possible pitches in the helix, depending on the orientation of the sheet with respect to the needle axis. In other words, the orientation of a hexagon sheet can be determined uniquely by referring to the needle axis. If hexagons A and B coincide with those labelled C and D, the needle axis will be along [010] and thus there will be no spiral rows of hexagons. This has, however, rarely been observed.

Figure 3 Electron diffraction patterns from individual microtubules of graphitic carbon. The patterns show mm2 symmetry, and are indexed by multiple superpositions of {h0l}-type reflections and {hk0} reflections of graphite crystal. The needle axes are horizontal. a, Superposition of three sets of {hk0} spots taken from a seven-sheet tubule. b, Superposition of four sets of {hk0} spots from a nine-sheet tubule. | high-resolution version |

Figure 4 a, Schematic diagram showing a helical arrangement of a graphitic carbon tubule, which is unrolled for the purposes of explanation. The tube axis is indicated by the heavy line and the hexagons labelled A and B, and A' and B', are superimposed to form the tube (for the significance of C and D, see text). b, The row of hatched hexagons forms a helix on the tube. The number of hexagons does not represent a real tube size, but the orientation is correct. c, A model of a scroll-type filament. | high-resolution version |

Because of the helical structure, the {hk0} spot patterns should always contain sets of even numbers of spots. Occasionally, sets of odd numbers of spots occur, such as the groups of three shown in Fig. 3a. These can be explained by frequent co-incidence of the top and bottom sheet orientations. Such an accidental coincidence will be increased for hexagonal symmetry of individual hexagon sheet when it is rolled. Consider the sets of four of {hk0} in Fig. 3b. Each of the sets of spots, which have equal intensity distributions, is rotated by ~6° about the (000) origin, or the needle axis. We take one set of (hk0) spots (indicated by arrows), corresponding to one of the hexagon sheets, as a reference whose [100] axis is rotated by 3° about the needle axis. The other three sets of spots are then rotated by 6°, 12° and 24° from the reference sheet. Considering the fact that the needle consists of only nine sheets and each set of {hk0} spots is generated from only three or four sheets, it is reasonable that every three or four sheets are rotated stepwise by 6° about the c axis. Any translational shift of the sheets cannot be detected in the electron diffraction patterns. A systematic change in sheet orientations was confirmed by (h0l)-type lattice image observations. The question of whether the systematic variation in the rotation angles in successive tubules acts to stabilize the structure remains to be answered.

The tips of the needles are usually closed by caps that are curved, polygonal or cone-shaped. The last of these have specific opening angles of about 19° or 40°, which can be rationalized in terms of the way that a perfect, continuous hexagonal network can close on itself.

According to Bacon's scroll model for tubular needle growth, needles could be formed by rolling up single carbon-hexagon sheets to form tubular filaments as illustrated in Fig. 4c. Such filaments should have edge overlaps on their surfaces. But I have observed no overlapping edges for the needles described here. Instead, I have observed concentric atomic steps around the needles (I have not confirmed that these are helical). On the basis of these new experimental findings on needle morphologies, I propose a new growth model for the tubular needles. That is, individual tubes themselves can have spiral growth steps at the tube ends (Fig. 4b). It is worth mentioning that the spiral growth steps, which are determined by individual hexagon sheets, will have a handedness. The growth mechanism seems to follow a screw dislocation model analogous to that developed for conventional crystals, but the helical structure is entirely different from the screw dislocation in the sense that the present crystals have a cylindrical lattice.

Received 27 August; accepted 21 October 1991.

1. Kroto, H. W., Heath, J. R., O'Brien, S. C., Curl, R. F. & Smalley, R. E. Nature 318, 162–163 (1985).
2. Oberlin, A. & M. Endo, J. Cryst. Growth 32, 335–349 (1976).
3. Speck, J. S., Endo, M. & Dresselhaus, M. S. J. Cryst. Growth 94, 834–848 (1989).
4. Tibbetts, G. G. J. Cryst. Growth 66, 632–638 (1984).
5. Bacon, R. J. appl. Phys. 31, 283–290 (1960).
6. Iijima, S. J. Cryst. Growth 50, 675–683 (1980).
7. Iijima, S. J. phys. Chem. 91, 3466–3467 (1987).
8. Kroto, H. W. Science 242, 1139–1145 (1988).
9. Krätschmer, W., Lamb, L. D., Fostiropoulos, K. & Huffman, D. R. Nature 347, 354–358 (1990).
10. Whittaker, E. J. W. Acta Cryst 21, 461–466 (1966).
11. Saito, S. & Oshiyama, A. Phys. Rev. Lett. 66, 2637–2640 (1991).

ACKNOWLEDGEMENTS. I thank Y. Ando for the carbon specimens.

| return to looking back index page |