Re: Smalley's Objection to MNT feasibility

Eugene Leitl (eugene@liposome.genebee.msu.su)
Tue, 9 Dec 1997 18:00:20 +0300 (MSK)


On Mon, 8 Dec 1997 DOUG.BAILEY@EY.COM wrote:

> His problem is with thermal resolution and can be found at
> http://cnst.rice.edu/NanoWelch.html

I've just skimmed the article. His objection is not with the thermal noise
(which is adressed by cooling down the device) or positional accuracy
(which is not addressed as easily, for I cannot make a dynamics analysis
as there is no definite manipulator structure yet) but a control problem.
Too many things moving in too many direction over a surrealistic potential
hypersurface while we have coarse handles attached only to a few of them.

"On a length scale of more than one nanometer, the mechanical robot
assembler metaphor envisioned by Drexler almost certainly will work, but
within the 1 nm3 volume surrounding the reaction site there is a subtlety
and complexity that is often not fully appreciated even by practicing
chemists. In any chemical reaction for all but the most trivial of
molecules, the detailed mechanism takes place on a potential energy
surface in an N-dimensional hyperspace, where N is of the order of 10-30
(three dimensions for each atom in the near vicinity of the reaction
site). In order for a robot assembler to be completely universal, it would
have to be able to control the detailed movement of not only the atom(s)
to be added or removed, but also the detailed motion of all the atoms
within a few bond lengths of the reaction site. Unless the motion of all
the atoms is controlled in this reaction nanospace, there is no way of
insuring that only the desired reaction product is obtained. Since the
manipulator "fingers" of the robot would have to be made of atoms as well,
there appears to be at least one fatal problem with the concept of a
universal assembler: there simply is not enough room inside the 1 nm3
reaction volume both for the atoms desired in the final structure and the
atomic fingers necessary to control their movement."

Merkle says it is not a problem at all, since he has computer models which
prove otherwise. Intuitively, he might be right -- I do not buy the
corrugatedness of the energetic landscape when it cames to rigidly held
radical tools.

Merkle's newest (afaik) browsable paper can be found

http://nano.xerox.com/nanotech/hydroCarbonMetabolism.html

Where I do see problems, is the aforementioned positional accuracy (+- 10
pm should be plenty, but +-100 pm is imo too much), sterical hindrance of
reaction sites on diamondoid terraces, high surface contaminant
concentration at cryogenic UHV conditions and associated radical
lifetimes, sterical hindrance at the tip tool (a hollow tip may be bist),
which would make multiple tip reactions impossible, etc. etc. The possibly
worst problem is the bootstrap problem. We can see Hy Brazeal from here,
right behind the mist bank, but the ocean between us is studded with shark
fins.

It is a nightmare of seemingly trivial details, but then God is in the
details as John C. uses to say. If one is unwilling to call these
constraints physical laws, one can call them technical difficulties.
Whatever one may call them, if they are severe enough, machine-phase
autoreplicators are dead meat. Or dead diamond. Whatever.

> I'd appreciate anyone's informed opinion on Dr. Smalley's assertions.

ciao,
'gene