Nano Letters

Foresight Institute has just mailed out copies of the January 26 Letters page of C&E News (normally available only to magazine subscribers). It contains eight letters on the recently published Drexler-Smalley debate, including one from me (Chris Phoenix). The editors made an even-handed and thoughtful selection, and I think it's worth reviewing the opinions.

The first letter, from Evan Appelman, was balanced: "However, it is one thing to say that at present we know of no way to accomplish something, and quite another to assert categorically that we will never be able to accomplish it."

The second letter, mine, called for further investigation while pointing out errors in Smalley's position. (You can read CRN's extended analysis of the debate here.)

The third letter, from Mihaly Mezei, was skeptical toward Drexler, talking about "the problem of scale": that a single mechanosynthetic fabricator couldn't rapidly make a significant quantity of product. I (Chris) was a bit surprised to see that Drexler did not mention in this debate that the proposed nanofactory would include great numbers of fabricators operating in parallel (all fastened down and plugged in, of course -- these are not the free-floating "assemblers" of two decades ago). Mihaly also suggested: "Moving molecules one by one sounds very much like an enhanced Maxwell's demon." But the point of Maxwell's demon is that it has no power source, and mechanochemical fabricators would certainly be powered.

The fourth letter, from Darryl Farber, acknowledged the sharp disagreement among scientists about what is possible, and suggested that we should stay flexible and consider policy while treating research as a risk-management strategy: take it a piece at a time until we see who's right.

The fifth letter was from a Senior Associate of the Foresight Institute, Howard Landman. Howard called Smalley "obtuse on a subject of such importance," backing up his strong words with a reference to his own published writing in Prospects in Nanotechnology, and making the point that lots of chemistry happens without water. He closed by saying that Smalley's demand for implementation details is "fatuous." The only place I would argue with this letter is Howard's invocation of entropy as a difficulty in making flawless systems. Crystals are existence proofs of flawless arrangements of atoms.

Ken Smith weighs in next, with a technical analysis of the difference between a chemical and mechanical system: the mechanical one is still chemical, it's just more constrained. He worries (again invoking entropy) that the constraints will prevent the energy from reactions being properly thermalized. As near as I can figure, he's thinking that a diamondoid system might not provide enough of a "shock absorber." But in fact, Drexler would prefer that the energy not be thermalized, since that's less efficient than directly converting mechanical force. If energy does have to be thermalized, any of several kinds of shock absorbers could be used, such as high-friction sliding surfaces or amorphous materials. Although I don't agree that this is a serious problem, I will acknowledge that Ken presented an objection I hadn't heard before, and made me think just a bit.

Alfred Boyle pointed out Smalley's error in claiming that enzymes can only work underwater. "Smalley attempts to cast Drexler as ignorant to the complexities of synthetic chemistry, when in fact Smalley is ignorant of the broad capabilities of biological catalysts." Alfred goes on to discuss the difference between engineers and chemists: "The engineer tends to oversimplify into basic, highly controllable unit operations .... the chemist (and biologist) overly acknowledge the inherent and partially undefined complexities of reactions .... scientific reality lying somewhere in the middle." I would put it a little differently: the engineer does not oversimplify his understanding; instead, he simplifies the actual design, finding unit operations that are in fact controllable and predictable. The engineer does not try to access all of science, but simply builds what works.

The final letter is from a nano-skeptic, Mark Wendman. "I cannot believe Drexler's mechanosynthesis is conceivable under conditions where chemical binding is unfavorable." Well, no one said it was! I suspect Mark has never read, for example, Ralph Merkle's "hydrocarbon metabolism" paper. This paper shows that with suitable application of mechanical force, a loop of favorable reactions can be created and used for mechanosynthesis. Mark also argues that it may not be worth trying to do better than DNA, and unknown chemistry might be dangerous. He then goes into a long discussion of prudent handling of nanomaterials.

In summary, it looks like Drexler didn't convince everyone, but the remaining concerns are usually either off-topic or easy to answer. Smalley came under heavy fire for his mistake about enzymes and his general narrow-mindedness. We may hope that these letters are representative, and that others will pay attention to the debate and realize that self-assured denial of the possiblity of mechanosynthetic nanomachines is no longer credible.

Chris Phoenix

Comments

[Kurt]

For the past two weeks, I have been conducting a direct market campaign for my instruments (an optical thin-film tester and a surface acoustic wave system). Going through all of the university websites, I noticed that almost every university that has a chemistry and materials science program are working on self-assembly and bio-memetic (wet) nanotechnology.

I find it very difficult to believe that we will not have a comprehensive "wet" nanotechnology capability within 20 years. The problem I have with "dry" nanotech is that the currently proposed schemes to not take into acount the use of brownian motion. Browning motion is inherent to all molecular level systems and is the reason why all of the proximinal probe manipulations to date have all be done at cryogenic temperatures, not very useful for real-world systems.

If someone comes up with a "dry" nanotech that incorporates brownian motion as part of the operating principle, I would be much more convinced of its feasibility than I am now.

Prof. Smalley does alot of hand-waving in his arguments against the possibility of "dry" nanotech. However, if you read through the hand-having, the core of his argument is the issue of brownian motion.

There is a professor Seeman at NYU who has come up with a list of 10 technical hurtles that msu be overcome for a feasible nanotech to be developed. Does anyone here have a list of these hurtles?

[Richard Jones]

Chris, a tedious technical comment. Your statement that "Crystals are existence proofs of flawless arrangements of atoms" isn't quite right; at finite temperatures the second law says you have a finite number of point defects - vacancies and interstitials - with the fraction of sites with defects scaling like exp(-ev/kT) where ev is effectively the energy of cohesion of the solid per atom. These point defects don't disrupt the perfection of the long-ranged order (at least in 3 dimensions - the story is different in 2). One might think they aren't relevant in MNT because the typical system sizes are small enough for the number of equilibrium defects to be very small. But if you start having strained structures the exponential dependence on binding energy might start to make defects a problem.

[michael vassar]

Actually Chris, you did hear the thermalization argument before, in a slightly different form, when you used it correctly against a proposal that I made. I then salvaged the proposal by proposing a method for phonon scattering and energy dispersal, but my later research showed that enabling technologies were still needed. The details are different in this case though, and thermalization doesn't present a problem.

Professor Seeman's 1st name is Nadrian
His e-mail is ned.seeman at nyu.edu but his web-page doesn't work.

[Chris Phoenix, CRN]

To Kurt, re Brownian motion:

I don't know where the rumor started that Drexler's designs don't use Brownian motion. They do.

Brownian motion is useful. For example, it can provide diffusive transport. It can retry reversible molecular matching operations. And it (or more precisely, heat) can, if I'm not mistaken, lower friction--not just by reducing viscosity but by providing a denser collection of accessible states.

Retrying molecular conformations is mentioned in sorting rotor binding (Nanosystems 13.1). Whole mechanical systems moving beneficially under thermal noise is invoked to allow a "stuck" sorting rotor to reverse itself by rotating backward under thermal noise (Nanosystems 13.2.1.d).

Thermal noise is also invoked in Nanosystems in the chapter on friction: 10.3.5 "Static friction" begins with the statement, "Where delta-curly-V-sub-barrier << kT, static friction is effectively zero." T, or temperature, is a parameter of thermal noise.

Diffusive transport appears to be less efficient than some alternatives; compare for example Nanomedicine 3.2.1 vs. 3.4.3. But it could certainly be used if anyone wanted to.

Chris

[Chris Phoenix, CRN]

To Richard, re entropy:

This is truly an exception that proves a rule. The back of my envelope says that C-C bond energy is 154 kJ/mol, or 256 zJ/bond, or 512 zJ/atom. That over kT300 is 124. e^-124 is 10^-54 defects per atom. So at room temperature, there'd be about one entropy-required defect in every 1.6x10^28 kg of diamond.

Chris

[Chris Phoenix, CRN]

Prof. Seeman has given me permission to post his "Top 10" list. Here is what he sent me:

"""""
These are the challenges for structural DNA nanotech, as I see them.

Those with two stars have seen major progress, those with one have seen a beginning.

None are completely resolved.

[1] TO EXTEND 2-D RESULTS TO 3-D WITH HIGH ORDER -- Crystallography; Nanoelectronics.

[2] TO INCORPORATE DNA DEVICES IN 2-D AND 3-D ARRAYS -- Nanorobotics. *

[3] TO INCORPORATE HETEROLOGOUS GUESTS IN LATTICES -- Nanoelectronics; Crystallography. **

[4] TO EXTEND ALGORITHMIC ASSEMBLY TO HIGHER DIMENSIONS -- Smart Materials; Computation. **

[5] TO ACHIEVE ASSEMBLIES WITH HIERARCHICAL CHARACTER -- Complex Materials. *

[6] TO ACHIEVE FUNCTIONAL SYSTEMS -- Active Materials; Sensor Systems. *

[7] TO INTERFACE WITH TOP-DOWN METHODS AND THE MACROSCOPIC WORLD -- Nanoelectronic Reality.

[8] TO INCORPORATE COMBINATORIAL APPROACHES IN THE DESIGN -- Diversity; Programmability. *

[9] TO PRODUCE SYSTEMS CAPABLE OF SELF-REPLICATION -- Economy; Evolvability. **

[10] TO ADVANCE FROM BIOKLEPTIC SYSTEMS TO BIOMIMETIC SYSTEMS -- Chemical Control.
"""""

[Janessa Ravenwood]

This is very useful, Chris. Maybe you could supply us with an annual "Project Progress Chart" for developing MNT with different graph lines for each of the necessary steps/phases/components necessary to pull the whole thing off. How about it?

[Chris Phoenix, CRN]

Just re-read this discussion. Quick comments:

Richard, you said strained structures might have a higher entropy-induced error rate, and I didn't really answer that--I just agreed with you that it wouldn't affect diamond. But a 10^-15 error rate per atom corresponds to a bond energy of just 140 zJ/atom. That's really small, and you could easily avoid building something that strained. (Drexler covered the failure rate of strained bonds in Nanosystems.) It's easy, and common, and pointless, to find examples of things that won't work. Unless those examples imply that nothing in the space will work--and I've never seen that happen in molecular manufacturing.

Janessa, I don't think we know enough yet to go into as much detail for (diamondoid) MNT as Seeman did for DNA. The best I can do is guesstimate, hopefully within an order of magnitude, how much it'll cost to develop it in the next N years. And of course the graph lines will be next to useless in forecasting, since their future progress will depend on both funding and new insights.

Chris