Abstract for the Fourth Foresight Conference on Molecular Nanotechnology.


Donald W. Noid (a) Robert E. Tuzun (b) Bobby G. Sumpter

Chemical and Analytical Sciences Division
Oak Ridge National Laboratory
Oak Ridge, TN 37831-6197

The success of nanotechnology applications will depend in part on precise knowledge and control of the frictional behavior of moving parts and on the introduction of ordered motion (rotation, for example) in nanomachine components. Using novel molecular dynamics (MD) methods, we have simulated several model graphite nano-bearings consisting of two concentric carbon nanotubes (shaft and sleeve). The turning shaft was either instantaneously started or was torqued up to the desired rotational speeds. Frictional properties were size-, temperature-, and velocity-dependent. The presence of more than one bearing vibrational mode in some simulations created beat patterns in rotational speed that could possibly adversely affect bearing performance; placing a stretching tension on the bearing suppressed one of the modes and therefore the beats. Applying transverse loads to a rotating bearing also produced beats in rotational speed; because axial angular momentum was no longer conserved, shaft and sleeve beats patterns were different.

We have also simulated several model graphite nanometer scale laser-driven motors consisting of two concentric graphite cylinders as above but also with one positive and one negative electric charge attached to an end ring on the shaft; rotational motion of the shaft was induced by applying one or sometimes two oscillating laser fields. The shaft cycled between periods of rotational pendulum-like behavior and unidirectional rotation (motor-like behavior). The motor on and off times strongly depended on the motor size, field strength and frequency, and relative location of the attached positive and negative charges. In addition, the two-laser simulations showed much larger on times and more stable rotation than one-laser simulations. A mathematical model of the overall process was obtained by employing computational neural networks (CNNs). A CNN was able to "learn" the mapping from size, charge position, frequency and strength of the electric field to optimize the performance of the motor.

*Research sponsred by the Division of Materials Sciences, Office of Basic Energy Sciences, U.S. Department of Energy,
under contract DE-ACO5-84OR214OO with Lockheed Martin Energy Systems, Inc.