Main

We used the same apparatus and conditions as described in ref. 6, except for the anode. This was prepared by boring a 3-mm-diameter hole in a 6-mm-diameter graphite rod. We filled these holes with a mixture of cobalt metal powder (5% at. metal) and the inner black core of a cathode deposit. This deposit was prepared using a graphite anode, and consisted of carbon nanotubes and other amorphous carbonaceous materials. We used a high-resolution transmission electron microscope (HRTEM, JEM 200-cx, operated at 200 kV) to characterize the morphologies and microstructures of the products. The specimens for HRTEM analysis were prepared by dissolving the black deposit in ethanol. After ultrasonic treatment, a drop of the liquid was sprayed onto a holey carbon copper grid.

Figure 1 shows an HRTEM image of the carbon nanotubes found in our experiment. This shows a uniform intershell spacing of 0.34 nm. The open arrow in Fig. 1 indicates the innermost nanotube, with a diameter of about 0.5 nm. Although the entire tip is not clearly imaged, this innermost nanotube is closed at the end by a half circle, as indicated in Fig. 1. We suspect that it is closed by half a C36 cage. Theoretical results predict that a zigzag nanotube with this diameter might have an (m,n) value of (6,0) and a diameter of 0.47 nm or (7,0)(0.55 nm).

Figure 1: HRTEM image of two nanotubes with outer diameters of 6 nm and 16 nm, respectively.
figure 1

The hollow arrow points to the innermost tube in the right nanotube, which has a diameter of 0.5 nm. The other arrow points to the tip of this tube, which we believe to be closed with half a C36 cage. Scale bar, 1 nm.

It has been suggested that carbon nanotubes are built up from atoms or atomic ions7. If this is true, anodes filled with graphite or carbon nanotubes should give the same results. Our results indicate that in the discharge the nanotubes could grow from carbon fragments. We propose that the carbon fragments within the arc have two forms before they begin to join up: curved (nanotube-filled anodes) and flat (graphite-filled anodes). Curved fragments will require less energy to form carbon nanotubes than flat ones. Use of a catalyst — cobalt, in our case — will result in curved fragments forming carbon nanotubes of the smallest diameter (0.5 nm in our experiment).

C60 has quite different properties from C36: for example it is soluble in toluene, whereas C36 is not5, and the high curvature and increased strain energy of this smallest nanotube might lead to many unusual properties. These tubes are in the form of the innermost shell of multiwall nanotubes. An important question that remains is whether these small-diameter nanotubes could occur as single-wall nanotubes.