Robotic friction
From Wise Nano
Microscale mechanical systems built by lithography (MEMS, MicroElectroMechanical systems) generally have very high friction and wear between sliding parts. It is often thought that this will extend to nanoscale systems, and anything designed like a robot will be unable to move.
Lithography is extremely imprecise: a shape 100 nm wide may have edges that are wavy by several nm. It can't make a really flat surface. So sliding surfaces will always be pushed together at a few points rather than riding across each other smoothly. And because the surface is not tightly made, atoms can be knocked off or cross-bonded to the other surface, causing rapid wear and even more friction.
Plans for nanoscale robotics, such as Drexler's diamondoid robots, start from the assumption that surfaces will be precise. Built a few atoms at a time, with each atom placed and bonded according to plan, nanoscale surfaces are expected to be quite precise with their atoms strongly bonded in place. Unlike biomimetic machines, robotic machines would be relatively stiff. This means that the interactions between the surfaces can be engineered precisely.
Any surface made of atoms will be bumpy at the atomic level. If the atoms line up correspondingly between the surfaces, then the bumps will create a strong sticking force. However, if the atoms do not line up, for example because one of the surfaces is turned at an angle, then the bumps will cancel out--assuming the construction is stiff enough that the bumps don't rearrange to fit between each other and let the surfaces squish together. This canceling, called "superlubricity," has been demonstrated in graphite. Thus, either stiffness or flexibility can be used to achieve low stiction.
Dissipation while moving can also be low. The rule of thumb is: As long as the system always has time to relax, and never goes through a sudden state transition (which can happen from either excessive speed or bad design), then it will not experience much friction. It's worth noting that stiff systems relax faster than floppy systems, implying that they can move faster with less friction.
In the absence of extremely high forces (much higher than Van der Waals force), a well-designed surface will not rearrange and bond to another well-designed surface sliding over it.
It is a matter of controversy how easy it will be to design good bearing surfaces. Drexler and Merkle have designed several that work great in simulation, but the state of technology does not permit building and testing them yet. The demonstration of superlubricity, and also Zettl's work with freely sliding and rotating buckytubes, provides some reason for optimism.
