Issue Date: January 26, 2004
Drexler and Smalley
Although I by no means claim to be an expert in nanotechnology, the exchange between K. Eric Drexler and Richard E. Smalley strikes me as a case of two people talking past each other (C&EN, Dec. 1, 2003, page 37). Drexler appears to be writing metaphorically about a sort of nanoscale production line, and he seems optimistic that the state of the art will advance to the point where the metaphor can become reality. Smalley observes correctly, but perhaps not entirely relevantly, that present knowledge indicates no way in which it would be possible to put Drexler's ideas into practice. He then in effect challenges Drexler to explain how he is going to set about making his scheme work. Drexler cannot respond to this challenge, because he simply doesn't know. If he knew, he would be building such systems instead of writing about them. 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. I am unaware of any fundamental laws of physics that mitigate against Drexler's concepts.
As an aside, I find it interesting that at least some of the issues addressed in this exchange were anticipated half a century ago by the science-fiction writer Philip K. Dick in his story "Second Variety."
The commentary on Drexler and Smalley's debate from a number of sources makes it clear that your publication has made a substantial contribution to the field of nanotechnology. Perhaps, now that the debate has shifted from magical impossible inventions to actual chemistry, real investigation of the concepts will be possible.
The subfield of molecular manufacturing has not received appropriate attention: Denials cannot replace scientific criticism, and the latter has been notably lacking. This remains the case today. Many of Smalley's statements about the limitations of underwater chemistry were more applicable to the current bio-nano focus of the National Nanotechnology Initiative than to anything Drexler has proposed.
Smalley's discussion of the limitations of "enzymelike" chemistry was instructive. He appears to base his argument on the assertion that enzymes cannot work without water. However, this is incorrect, as almost two decades of published results demonstrate. This underscores the fact that more investigation will be needed to determine the limitations and capabilities of mechanically guided surface chemistry.
During the past decade, detailed proposals for the architecture and technology of molecular manufacturing systems have been developed. Studies can be done today, both in simulation and in the lab, of reactions that could be useful for forming covalent solids or other useful chemicals by direct programmable deposition.
I must strongly disagree with Smalley's recommendation for protecting "our children" from fear. The best antidote to fear is knowledge, not denial. For example, the fact that modern plans for molecular manufacturing include nothing resembling a "gray goo nanobot" is comforting and should be publicized. Hasty arguments against the possibility of mechanically guided surface chemistry--which has already been demonstrated--will not, in the long run, be either comforting or productive. If accepted, these arguments could lead to bad policy that cannot respond to real risks and benefits.
I encourage you to lead the call for further studies of the science and technology involved in mechanically guided chemistry. It would not require much work to develop initial estimates of its capability as a fabrication technology. Indeed, much work has already been done in "Nanosystems" (1992) and subsequent papers. No significant error has yet been found in this work.
We need to start addressing the theory of molecular manufacturing directly with scientific studies. Unless a currently unknown problem is found in the theory, development of a general-purpose manufacturing capability could be quite rapid. This would not allow time for last-minute policy-making. Several technical studies which support this point are available on our website, www.crnano.org.
I was fascinated with the exchange of letters between Smalley and Drexler on the limits of nanotechnology (or the lack thereof). It occurred to me that, besides the difficulties to the practical realization of molecular manufacturing discussed by Smalley, there is an additional one: the problem of scale.
Drexler's statement that "scaling down moving parts by a factor of a million multiplies their frequency of operation by the same factor" suggests that he was targeting a machine capable of completing 1 million reactions per second. So, assuming that there is such a machine, how long would it take to produce 1 mole of reactants? A back-of-an-envelope estimate gives 100 million years. While even the million reactions per second is hard to imagine being practical (issues with delivering the reactants and removing the products at the same breathtaking rate appear major stumbling blocks to me), even a millionfold speedup would require 100 years to produce 1 mole!
Furthermore, I also wonder (but it would take more than a back-of-an-envelope calculation to elaborate) if molecular manufacturing does not violate the second law of thermodynamics. Moving molecules one by one sounds very much like an enhanced Maxwell's demon.
New York City
Although the Smalley/Drexler exchange highlights apparently deep physical science theoretical differences, it also raises interesting science and technology policy issues. Experiment and engineering demonstrations will eventually prove which scientist is right. However, the policy question is, How ought society make its investment in nanotechnology given the sharp disagreement between visions of what is possible? What are the benefits, costs, risks, and options? How should society invest?
Although Nobel Laureate Smalley may think nanobot assemblers are nonsense, from an educated layperson's perspective, it appears reasonable that there are reputable scientists and engineers (including the late Nobel Laureate Richard Feynman) who believe mechano-synthesis, "machine-phase" chemistry is possible. A prudent course of action then is conceptualizing the social investment in molecular manufacturing as a real option. Real options are similar to financial options in that one purchases the right, not the obligation, to take further action, such as the right (not obligation) to buy a stock at a particular price.
In this case, society effectively buys an option to continue research and risk-management strategies with regard to molecular manufacturing, recognizing that, at some future time and given more information, it may allow that option to expire and not pursue future research. Or it may exercise the option and buy more R&D if results are promising. In either case, there is a value to flexibility that is explicitly accounted. Perhaps most important, this flexibility is more than just economic; it provides time for the many stakeholders to discuss and reflect on the broader social and ethical implications of nanotechnology.
University Park, Pa.
It is sad to see a greatly revered scientist like Smalley being so obtuse on a subject of such importance.
I have been an associate of the Foresight Institute for many years and have had many opportunities to debate at length with Drexler, Ralph Merkle, and others. I do not agree with them on all points, and I believe there are some valid criticisms to be made of their positions. I don't think Merkle, in particular, takes entropy quite seriously enough when he envisions flawless nanotech systems. But the existence of what I sometimes call "the dirt problem" does not in any way mean that molecular nanotechnology is inherently impossible, any more than it means that it is impossible to build electronic systems with millions of transistors. Other nanotech researchers, notably Stan Williams' group at Hewlett-Packard, understand the importance of living with defects and have already demonstrated computer architectures that can tolerate a huge defect density.
Biology provides an incontrovertible "existence proof" that some form of molecular manufacturing is possible, and I have previously argued in the chapter "Paths to Nanotechnology" (Krummenacker, Markus, & James Lewis, editors. "Prospects in Nanotechnology," New York: John Wiley & Sons, 1995) that the earliest assembler-like technology is most likely to arise in a biotechnological context. Smalley accepts that something like enzymes or ribosomes can do precise chemistry. But he then seems unable to imagine that it will be possible to transcend biology's aqueous heritage and do the same kinds of things in other solvents or in vacuum, saying, "If it is ... non-water-based ... then there is a vast area of chemistry that has eluded us for centuries." Why? There is plenty of chemistry we already understand that can be done without water, and in fact much that must be.
It is also fatuous of him to demand that Drexler solve the implementation details of future technologies. These will come in time, with much work and investment; if we knew all the answers already, we would be using nanotech now! And since it is unlikely that nanotech will be the best and cheapest way to build everything, it is pointless to list a handful of specific things that might be difficult for it. Neither a potter's kiln nor a modern silicon fab can produce a feather, but that does not make kilns or fabs useless or impossible.
Howard A. Landman
Fort Collins, Colo.
I read with interest the exchange between Smalley and Drexler on the subject of nanomechanical fabrication. One thing that struck me was the differentiation claimed by Drexler between a chemical system and a mechanical one, since, in the limit, the two are one in the same thing; namely, a chemical system. The only difference is the constraints imposed on certain vibrational and rotational states. In the "mechanical" framework, there are clearly fewer states allowed in the design of the molecular assembly that could deviate from productive movement along the desired reaction pathway.
The best way to think of it is an enzyme-substrate complex with two fairly discrete sets of partitioned vibrational/rotational states. One of these sets is associated with movements along the reaction pathway (the pushing together of two molecular entities), and the other is a set of states with relatively low-frequency rigid atomic arrangements that act as a container/positioner for orienting reagents and a thermal bath for energy transfer during the "mechanical event." I fear the coupling of these two sets of states will not be sufficient to allow control of the thermal repercussions of the push, and an intervening medium will be necessary. Water works very well for this in biological systems. So, nanomachines in some kind of medium will probably be essential. Then, of course, one will have to get those interfering atoms out of the way--a lot of complexity to control in a very small volume. Could it be that we are pushing against an entropic limit?
Muir of Ord, Ross-shire, Scotland
One significant error needs to be corrected in Smalley's argument. Enzymes can function in nonaqueous environments. Omission of this well-documented catalytic capability significantly weakens Smalley's argument regarding his perceived limitations of "molecular manufacturing." 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. Both would be well served by further study.
The "Point-Counterpoint" illustrates the fundamental difference between the way engineers and chemists (and biologists) visualize problems and their solutions. The engineer tends to oversimplify into basic, highly controllable unit operations that will always operate within specifications given adequate controls. Meanwhile, the chemist (and biologist) overly acknowledge the inherent and partially undefined complexities of reactions, believing things are always more complicated than they appear. The topic may change, but such differences in view will always exist, with the scientific reality lying somewhere in the middle.
Being well versed in nanofabrication, I cannot believe Drexler's mechanosynthesis is conceivable under conditions where chemical binding is unfavorable. Smalley is correct in his thesis against a mechano "chemistry" assembler.
A computer programmer might underestimate the power of chemical binding and instability, even on the nanoscale. A chemist cannot ignore this, as Smalley thoughtfully argues.
DNA is a natural embodiment of an assembler adhering to chemical laws. Analogs to DNA's activity will prove difficult to reproduce in another chemical system, and the search for this may not be worth the effort. The consequences of unknown activity levels of an assembler in another chemical repertoire might be dangerous. Genetic engineering is a more fruitful area of research than mechano-assembly.
Nanotechnology research should not be deferred, since most investigations do not involve assemblers, but rather modified conventional chemistry. Prudence, not undue worrying, is warranted in large-scale new technology deployment. Typically, new materials are tested for hazards in industrial application. If testing is overlooked, the results can be undesirable. Safety is indicated even in science leading to industrial innovation.
Different structures of carbon have vastly different degrees of hazard or benign effect. Comparing diamond to carbon soot, and asbestos fibers to macroscopic silicate glasses, is instructive. Asbestos is analogous to nanotubes and comparable in dimensions, and the hazards of carbon soot and asbestos are well known. Contrast this with inert macroscopic diamonds and many bulk silicate glasses.
Wide-scale deployment of new nanomaterials prudently warrants toxicity testing to mitigate risks of another asbestos problem. Nanotubes should be used with caution, comparable to what should have been done with asbestos. Was asbestos worth using in construction in large quantities? Probably not. Are there applications that might warrant the use of asbestos--probably. Asbestos is not as toxic as feared when properly applied. Airborne particle management is safe practice, not folly. In manufacturing, specific methods of preparation and application of asbestos minimize risk, as does protective gear.
Will nanotubes have broad applications that are worth their use? I cannot tell, but it is folly to use carbon nanotubes in consumer clothing for something as trivial as stain protection without testing. Safety characterization should be done for large-scale preparation of nanotubes and for manufacturing processes incorporating them in cloth fiber. Protective gear may be warranted. Asbestos is a fine example of why this is good practice.
Commonsense safety is not the final arbiter of research trends, but is often worth considering. Researchers should be unhindered by cautions warranted in large-scale deployments, but they should apply responsibility to framing proper testing prior to such deployment. History teaches us of past mistakes.
Nanotechnology outside the domain of a mechanochemistry assembler is a fertile field of research and must be pursued with vigor of intellectual vision and creativity, yet deployed with responsible caution where analogous examples indicate we might well have reason to be prudent.
The future is what one makes of it and how one capitalizes on opportunities in new fields. Excessive unwarranted worry should not be the overriding concern.
- Chemical & Engineering News
- ISSN 0009-2347
- Copyright © American Chemical Society