No abstract

the contact width, and L is the nanotube length. These quantities are dif®cult to measure independently and calculations are modeldependent. As a guide, we apply a JKR model for the contact of cylinders 25 and ®nd a contact width of 3 nm for a tube radius of 13.5 nm. Our measurements of 0.006 N m -1 for the friction force per unit length is then consistent with a shear stress of 2 3 10 6 Pa. This can be compared with a value of 5 3 10 6 Pa, as inferred from AFM tip/graphite measurements 16 . To compare rolling and sliding in a single tube, we can calculate the force (4 nN for L 590 nm) and that would be needed to slide tube B, which in fact rolls. Finally, we note that the area under the lateral force trace is a direct measure of energy loss in rolling. For tube B, we measure an energy loss of 8 6 3 3 10 2 16 J per revolution. The sliding energy loss expected for this distance (85 nm) can be calculated using the frictional force of 4 nN, yielding 3 3 10 2 16 J.When we compare our lateral force measurements for sliding and rolling cases, we ®nd that the stick peaks in rolling are higher than the lateral force needed to sustain sliding, and that the energy cost for rolling is larger than that of the sliding cases. Why should the nanotubes roll? We speculate that, owing to the size and surface features of the rolling nanotubes, a stick peak for sliding in side-on pushing might exist that is larger than the threshold for rolling. Atomic-scale substrate interactions may also play a roll as we have observed this characteristic rolling only on graphite. Rolling behaviour has been accompanied by a preferential, threefold, inplane orientation that indicates intimate nanotube/graphite contact, and perhaps lattice registry. Rolling may occur only when both the nanotube and the underlying graphite have long-range order. In these cases that there may be a barrier for sliding which is larger than that for rolling and may preclude the direct measurement of sliding friction 7 . M

We present a combined study by scanning tunneling microscopy and atomistic simulations of the emission of dissociated dislocation loops by nanoindentation on a (001) fcc surface. The latter consist of two stacking-fault ribbons bounded by Shockley partials and a stair-rod dislocation. These dissociated loops, which intersect the surface, are shown to originate from loops of interstitial character emitted along the <110> directions and are usually located at hundreds of angstroms away from the indentation point. Simulations reproduce the nucleation and glide of these dislocation loops.

We find that less than 0.01 monolayer of S can enhance surface self-diffusion on Cu(111) by several orders of magnitude. The measured dependence of two-dimensional island decay rates on S coverage (theta(S)) is consistent with the proposal that Cu3S3 clusters are responsible for the enhancement. Unexpectedly, the decay and ripening are diffusion limited with very low and very high theta(S) but not for intermediate theta(S). To explain this result we propose that surface mass transport in the intermediate region is limited by the rate of reaction to form Cu3S3 clusters on the terraces.

The oldest known magnetic material, magnetite, is of current interest for use in spintronics as a thin film. An open question is how thin can magnetite films be and still retain the robust ferrimagnetism required for many applications. We have grown 1-nm-thick magnetite crystals and characterized them in situ by electron and photoelectron microscopies including selected-area x-ray circular dichroism. Well-defined magnetic patterns are observed in individual nanocrystals up to at least 520 K, establishing the retention of ferrimagnetism in magnetite two unit cells thick.