March 1999

Will the Real Nanotech Please Stand Up?

Continued from page 4

By David Rotman

Enter the Hype

it's somewhere around here that the science starts getting mixed up with science fiction. If you can make a nanowheel, why not a nanogear? A self-powered nanoboat? Why not build a nanorobot to move around the atoms for you?

And while you're at it, why not make nanorobots that can replicate themselves, making it possible to staff nanofactories capable of piecing together almost anything out of the basic building blocks of atoms? Welcome to molecular manufacturing, as preached by nanoevangelist Drexler. At the core of the Drexlerian vision is a gizmo called an "assembler." This hypothesized robotic apparatus would work by mechanically positioning atoms into virtually any configuration. If the chemistry between the atoms doesn't take, the assembler would apply a small mechanical force (Drexler and his followers call it mechanochemistry). Get billions of these assemblers to work in parallel to arrange all the atoms just right-well, then, you can build just about anything you can imagine.

There's just one problem: Few chemists, physicists or materials scientists see any evidence that this will be possible. Many believers in the Drexlerian vision are computer scientists who delight in simulating how it all will work. They produce elegant molecular models of nanogears and pumps but offer no clear plan for how to actually build such things.

Proponents of molecular manufacturing aren't deterred by the skepticism of their more mainstream colleagues-although they do concede that their vision will take decades to be realized. Theoretical calculations and computer modeling say it can be done, insists Ralph Merkle, a computer scientist at the Xerox Palo Alto Research Center and a director, with Drexler, of the Foresight Institute. In particular, Merkle defends the two key proposals that have drawn the most fire from other scientists: the suggestion of self-replicating assemblers, and positional control of atoms and molecules to do mechanochemistry.

In self-replication, a molecular computer would direct the construction of a nanorobotic arm to build another computer; this second computer then directs the construction of another tiny computer, and so on. Self-replication is a concept that has been kicking around in computer science for years, says Merkle, and logically it should work. The idea of positional control calls for the robotic arms to precisely place atoms and molecules in a way that they bond, forming whatever you want. As long as you don't violate any physical laws, Merkle says, this mechanical approach to chemistry makes sense.

But Drexler's critics point out that chemistry is a very complex process at the molecular level. To play the game of chemistry, says Smalley, means controlling atoms in three dimensions. At each reaction site, atoms feel the influence of a dozen or so neighboring atoms; to do mechanochemistry, you would need to control the motion of each one. For a nanorobot, that would be an inconceivably complicated juggling act. Other highly respected researchers simply dismiss Drexler's ideas out of hand. Says IBM's Eigler: "He has had no influence on what goes on in nanoscience. Based on what little I've seen, Drexler's ideas are nanofanciful notions that are not very meaningful."