Electromechanical Properties of Multiwall Carbon Nanotubes

We examine electrical and coupled electromechanical properties of multiwall carbon nanotubes using transport measurements performed in-situ inside a high resolution transmission electron microscope (TEM). In one experiment, large electrical currents are passed through the nanotubes and the failure modes for nanotube "burnout" examined . In a second set of experiments, the electrical resistance between the ends of nanotubes is measured as the tubes are either "telescoped" or partially telescoped and then severely but reversibly mechanically kinked. Our experimental results have implications for nanotube quantum charge transport mechanisms, Multiwall carbon nanotubes (MWNTs) are comprised of concentric nanotube shells, each shell apparently "just fitting" inside the next, with an intertube spacing roughly equal to the van der Waals graphite interplane distance, 3.4A^. This geometrical constraint suggests that some of the nanotube shells are individually either metals or semiconductors^. The composite shell structure may have a complex electrical behavior, especially if charge is transported from one concentric tube shell to the next. In addition, defect structures can affect nanotube transport. Some previous experiments have suggested that singlewalled carbon nanotubes (SWNTs) are more likely to behave as ballistic transport channels than are MWNTs^. On the other hand, careful "mercury dipping" experiments on MWNTs have indicated electrical conductance quantization plateaus suggestive of ballistic transport, perhaps confined to only the outer nanotube shell^-6. Magnetic flux quantization experiments have received similar interpretations^. We here report on electrical conductance measurements performed on MWNTs placed inside a high resolution transmission electron microscope (TEM) fitted with a custom-made electro-mechanical manipulation stage (with x,y,z coarse and fine mechanical motion control). The stage allows an individual MWNT to be selected, bonded to electrodes in a two-probe configuration, and mechanically manipulated while the resistance is monitored and the nanotube is viewed under high microscope magnification. The same apparatus has been previously used to "sharpen and peel" individual MWNTs^ and to "telescope" the inner core tubes from the outer shell housing, thus forming low-friction linear bearings^. In the first set of experiments to be described here, a MWNT is contacted and the electrical current through it is steadily increased until the nanotube electrically and CP590, Nanonetwork Materials, edited by S. Saito et al. © 2001 American Institute of Physics 0-73 54-0032-6/017$ 18.00 107 Downloaded 25 Mar 2003 to 128.32.212.214. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp mechanically fails, i.e. "burns out". One of several different nanotube failure modes might be expected, depending on the details of the transport mechanism of the MWNT. For example, if the nanotube is indeed a ballistic conductor, then the electrical resistance is confined to the contacts. The contacts therefore are the hot spots for energy dissipation (Power = IV), and these might be expected to fail first (i.e. the contacts should "blow off). If, on the other hand, the MWNT is a dissipative conductor, with the dissipation more or less uniformly distributed along the length of the tube (but the thermal heat sinking largely confined to the end contacts with a minimal cooling contribution from black body radiation), then the nanotube should assume a well-known temperature profile^ with a maximum temperature realized at the half-way point along the tube. In this case tube failure would initiate exactly halfway between contacts, in a perhaps catastrophic fuse-like vaporization mode. Experimentally, neither of these failure modes is observed! Inevitably, as the current is increased past a critical value (typically of order 200 ^A) the nanotube "burns out" in a seemingly random location, at a position where even high-resolution TEM imaging (prior to failure) shows no evidence for any obvious nanotube defect structure. MWNTs are never observed to "blow off the contacts, nor to fail exactly in the middle of the tube. Upon failure, the nanotube appears to mechanically separate over a small region (more like a cut of the tube rather than a fuse-like meltdown), and the two independent leftover pieces of the tube (still attached to the independent electrodes) appear to remain largely intact. If one assumes that the nanotube fails at the most severe defect, then one might expect that the remaining tube portions are more "defect free" than the original tube. Hence, successive burnouts of the remaining tube segments might be used to "purify" a given MWNT, in the sense that the largest remaining defects are successively cut out. We have tested this hypothesis, and indeed the remaining tube segments always have significantly higher threshold currents than the parent tube. We now describe a second set of experiments, for which a MWNT is first sharpened and peeled^, thus exposing core tubes. The manipulator electrode is then spot-welded to the core tubes and these are telescoped out of the housing tubes°. During the telescoping, the electrical resistance between the "ends" of the tube is monitored. It should be stated at the outset that the "spot welding" method of electrical contact leaves some ambiguity as to which of the concentric core tubes are actually physically contacted (similar ambiguities exist for electrical contact to the housing tubes as well). Nevertheless, at the very least we may assume that the largest diameter core tube is well contacted, and similarly is the largest diameter housing tube. A schematic for the experimental contact and mechanical manipulation configuration is shown in Fig. 1 A. One possible outcome of such an experiment is that the matrix element for charge transfer between the largest diameter core tube and the smallest diameter housing tube depends linearly on the physical "overlap area" between the tubes. In this case the resistance between the ends of the telescoping MWNT might then behave as a sliding variable resistor (a nanotrimpot!). On the other hand, one could imagine other possibilities, such as oscillations in the resistance with telescoping (as transfer matrix element resonances are encountered, depending on the chirality differences beween