Copyright 1995 by Xerox, all rights reserved.
Further information on nanotechnology is available.
This article is available on the web at URL: http://www.zyvex.com/nanotech/nanocritics.html.
Addendum added in March 1996: Scientific American had a news story in the April 1996 issue titled Trends in Nanotechnology that quoted Jones' book review and not only omitted any reference to this rebuttal (despite the fact they had been told about it) but misleadingly said "The nanoists' response to this fusillade is simple: read Drexler's technical tome Nanosystems, ..." As their article quoted posts to sci.nanotech critical of nanotechnology it is difficult to understand the reasons for this convenient omission.
A detailed critique of the Scientific American news story is available at http://www.foresight.org/SciAmDebate/SciAmResponse.html
While nanocritics were relatively common several years ago, they have become increasingly rare; there has been some concern they might become an endangered species, unable to survive in the modern world. Most nanocritics tend to be rather shy and retiring, fearing (quite rightly) that they will appear foolish or ignorant. And very few indeed will commit their words to public print: this would, after all, force them to either defend their beliefs or confess to an amused world that they got it wrong.
This makes the recent sighting, in print, of an avowed nanocritic all the more exciting.
Nestled among 18 other book reviews, "Technical boundless optimism"  by David E. H. Jones (a.k.a. Daedalus) proclaims the doom of nanotechnology and an end to all this foolishness about molecular machines and atomically precise products!
Discrediting the impious upstarts is obviously of some importance both to Jones and to Nature, as Jones' review is the longest of the 18 and pays only token attention to the book being reviewed ("Nano!" by Ed Regis). Unfortunately, the credibility of his assault is undercut by his ignorance of the field. It's clear that he hasn't even read Nanosystems: Molecular Machinery, Manufacturing and Computation, by K. Eric Drexler. As Nanosystems was published in 1992, is widely acknowledged as the standard technical reference on nanotechnology, it is now 1995, and as "Nano!" refers to Nanosystems frequently; this required considerable effort. Jones seems to have missed most of the rest of the field as well. This is difficult as there are several books on the subject, a great many papers, a net news discussion group, several web sites, and a series of conferences held every other year that began in 1989 and which have drawn together hundreds of researchers from around the world. The fourth conference in this series, the Fourth Foresight Conference on Molecular Nanotechnology, will be held November 9-11, 1995, in Palo Alto California. Everyone interested in the development of molecular nanotechnology should attend (the URL for the conference web page is: http://www.zyvex.com/nanotech/nano4.html). The aspiring nanocritic (at least, those who want to get their facts straight) should attend as well.
And now to the unglamorous but necessary task of going over Jones' comments point by point.
Jones first paragraph starts with a little unsubstantiated innuendo. In particular, he compares the low cost promised by molecular manufacturing with the "too cheap to meter" claims made about nuclear power. The section in Nanosystems on the estimated manufacturing cost of products made by nanotechnology concludes that costs in the range of 10 to 50 cents per kilogram are reasonable. This is in section 14.5.6, starting on page 432, and includes an analysis of the assumptions that go into the cost analysis. While this is a remarkably low cost considering the quality, complexity and performance of the products being produced, it is neither too cheap to meter nor even significantly different from the cost of existing agricultural products (wheat, corn, etc.). The discussion explicitly excludes other costs, such as taxation, licensing, etc., which are likely to be significant but which are difficult to estimate based on technical information.
The second paragraph points out that nanotechnology is entirely different from microtechnology. In this, we are in agreement. Microtechnology, based on lithographic methods used for the mass production of semiconductor wafers, appears to be fundamentally unable to deliver the capabilities of nanotechnology.
The third paragraph has some basically irrelevant innuendo about Drexler and the development of nanotechnology. The most notable point is that Jones incorrectly gives Drexler's affiliation as MIT. Drexler completed his Ph.D. at MIT, but is now Chairman of the Foresight Institute and a Research Fellow of the Institute for Molecular Manufacturing, both on the west coast of the United States rather than the east coast. One of the points in the book Jones' was supposedly reviewing is that Drexler hasn't been at MIT for many years.
The fourth paragraph discusses Jones' emotional and subjective reaction to "Nano!"
The fifth paragraph has a simplistic and moderately inaccurate description of nanotechnology. One notable innuendo is Jones' claim that the exact specification of an assembler "...cannot yet be given," implying that such a specification is somehow a mysterious and impossible goal. Providing a complete and fully detailed design for an assembler is within the current state of the art. It is not conceptually difficult: it simply requires the appropriate resources. Various papers discussing the design issues and alternatives have already been published, and more are in process [3, 4, 5, 6, 7, 8, 9]. One paper available on the web is Self replicating systems and molecular manufacturing.
The sixth paragraph mentions a few potential negative aspects of the development of nanotechnology. Again, no substantive content. For a discussion of potential risks relevant to the use of self-replicating systems in manufacturing, see.
The seventh paragraph argues that nanotechnology cannot solve all problems "Even if it worked..." This is correct in general (though I express no opinion here on the specific examples he presents). I do not recall the claim being advanced in any published work by anyone with even a reasonable knowledge of nanotechnology that it can solve all problems. It can reduce costs, increase quality, improve performance, reduce weight, etc. for a broad range of manufactured goods. Many new products will become feasible or economical. These changes will likely be beneficial to many people for many reasons. They will also be applicable to weapon systems. Depending on how those weapon systems are deployed and used, the results could be either beneficial or detrimental.
Later in the seventh paragraph Jones begins to discuss issues of some substance. He says that "Nano!" provides five arguments in support of the feasibility of nanotechnology and begins to attack them. Why Jones thinks that a general book focusing primarily on the human aspect of the researchers in this field is a fit target for technical attack is unclear. It is even less clear why he doesn't consider the standard technical book in the field: Nanosystems. Perhaps he hasn't read it.
Interestingly, one of the arguments in favor of nanotechnology that Jones sites is that Feynman thought it would work. Jones makes no further comment on this. In 1959 Feynman said:
The principles of physics, as far as I can see, do not speak against the possibility of maneuvering things atom by atom. It is not an attempt to violate any laws; it is something, in principle, that can be done; but, in practice, it has not been done because we are too big.Quite a few people besides Feynman think the development of nanotechnology "...cannot be avoided." Perhaps Jones will review "There's Plenty of Room at the Bottom" and explain Feynman's errors to us.
Ultimately, we can do chemical synthesis. A chemist comes to us and says, "Look, I want a molecule that has the atoms arranged thus and so; make me that molecule." The chemist does a mysterious thing when he wants to make a molecule. He sees that it has got that ring, so he mixes this and that, and he shakes it, and he fiddles around. And, at the end of a difficult process, he usually does succeed in synthesizing what he wants. By the time I get my devices working, so that we can do it by physics, he will have figured out how to synthesize absolutely anything, so that this will really be useless.
But it is interesting that it would be, in principle, possible (I think) for a physicist to synthesize any chemical substance that the chemist writes down. Give the orders and the physicist synthesizes it. How? Put the atoms down where the chemist says, and so you make the substance. The problems of chemistry and biology can be greatly helped if our ability to see what we are doing, and to do things on an atomic level, is ultimately developed - - a development which I think cannot be avoided.
The eighth, ninth and tenth paragraphs are an attack on small machine tools that work by cutting away material. He appears to be arguing that a tool made of atoms cannot apply forces to an object made of atoms without itself suffering damage, and therefore this approach cannot be made to work. As the proposed chemical reactions for the mechanosynthesis of diamondoid structures have been presented both in Nanosystems and in articles by others[3, 4, 7, 12], it is unclear why he presents and attacks this straw man. I presume it is because he is literally unaware of the relevant literature.
One relatively short introduction to the proposals for the mechanosynthesis of diamondoid structures is Molecular manufacturing: adding positional control to chemical synthesis.
The eleventh paragraph argues that chemistry is limited to certain traditional approaches that involve manipulating molecules in bulk. "It [chemistry] tries to set up molecular encounters that, by their own nature, their random thermodynamic shuffling towards the most accessible local energy minimum and global entropy maximum, will form the desired product. A great many substances and material forms can be made this way, but an infinitely greater number cannot." Jones is quite correct in claiming that traditional chemistry is able to synthesize but a tiny fraction of the feasible molecular structures. However, attributing to chemistry as a whole the limitations of the specific methods to which he imagines chemists must restrict themselves is a more difficult position. He clearly views the adoption of positional control as being somehow forbidden. Many chemists will no doubt be surprised to find they are forbidden from using positional control, nor is there any support for this remarkable concept.
Imagine, for a moment, that we tried to build a radio using the methods outlined by Jones (and commonly used in chemistry today). We'd take the parts of the radio and put them into a bag and shake. Then we'd take out an assembled radio. This is a hard way to build a radio! This approach is common in chemistry not because of any fundamental requirement of physics, but, as Feynman put it, "...because we are too big." If we develop finer tools able to position finer parts, there are no fundamental obstacles that prevent us from positioning individual atoms and molecules and utilizing this control to build structures that are molecular both in their size and precision.
The twelfth paragraph carries Jones' views on the limits of chemistry further: "...the nanotechnologists do not seem to realize the chemical obstacles in their path. To break a chunk of raw material into its component atoms needs a lot of energy -- at least the latent heat of vaporization. And the single atoms when you have them cannot just be picked out and pushed around like so many marbles."
As the "nanotechnologists" in question routinely use molecular mechanics, molecular dynamics, semiempirical and ab initio quantum chemistry to model the behavior of both proposed molecular devices and the reactions to be used in their synthesis [3, 4, 6, 7, 12], it is difficult to understand the basis for Jones' claim.
He then attempts to argue that the demonstrated experimental ability to manipulate individual atoms, as in the manipulation of Xenon atoms on a nickel surface, is somehow impossible if we are dealing with atoms that form bonds. Unfortunately, the ab initio quantum chemistry analysis of (for example) the hydrogen abstraction tool  is not mentioned, nor the molecular dynamics modeling of that reaction at room temperature on a diamond (111) surface, nor is there any hint that Jones is even aware of the theoretical work that has gone into the analysis of reactions appropriate for the synthesis of diamond using positionally controlled molecular tools.
Why manipulation of atoms held together by relatively weak forces such as van der Waals forces is feasible while manipulation of atoms held together by stronger bonds is inherently infeasible, is not made clear in his argument.
Jones argues that the latent heat of vaporization makes the synthesis of (among other things) diamondoid structures infeasible. There is a substantial literature on the chemical reactions involved in the synthesis of diamond by chemical vapor deposition (CVD)[14, 15, 16, 17, 18]. Molecular tools that use reactions similar to those seen in the CVD growth of diamond, but which are positionally controlled, should be quite able to synthesize complex diamondoid structures[3, 4, 6, 7, 8, 12]. The proposed mechanosynthetic reactions involve the use of radicals, carbenes and applied mechanical force. These are the same kind of reactions involved in the growth of diamond films today. These mechanisms alone can produce more than sufficient energy to synthesize or break down a remarkably wide range of materials.
The fundamental point, however, is not that Jones disagrees with the proposed reactions, but that he is blissfully ignorant that such proposals even exist.
Later in the paragraph he argues that single atoms are reactive and will combine with whatever solvent they happen to find themselves in. He is probably confusing the claim that nanotechnology is "atomically precise" with the method of synthesis. Proteins are atomically precise, yet we do not make proteins by picking up and putting down individual atoms. Instead, proteins are made from clusters of atoms called amino acids. While some synthetic methods might involve the manipulation of individual atoms by appropriate tools ( the hydrogen abstraction tool discussed earlier removes a single selected hydrogen atom from a diamondoid surface), other reactions will involve small clusters of atoms (or indeed larger molecular components). This distinction is discussed in several places, including chapter eight and the footnote on page 9 of Nanosystems.
The concept that reactions must inherently take place in some solvent is unnecessarily limiting. The use of vacuum as an environment has a number of advantages, chief among them being that highly reactive structures will not react with it. Jones is still bound by the idea that the only permissible chemical reactions are ones that take place in a solvent where random encounters among molecules are unavoidable. Positional control allows highly reactive molecular components to be positioned, and encounters among them to be controlled.
Jones then observes that spontaneous reconstruction of the work piece should be avoided. This is generally correct (unless the reconstruction is one that is desired and has been planned for). He seems to imply that reconstruction of the work piece is some sort of fundamental problem. As most of the work piece can be a simple, hydrogen terminated structure the problems of reconstruction would seem to be highly limited. Only a single small region of the work piece need be operated on at any given time, hence it is not necessary to use the kind of reactive surfaces that might pose problems of reconstruction. The single small region might indeed be highly reactive. There are many molecular structures that are highly reactive yet stable in vacuum, so this constraint is not much of a problem.
Lest my words be construed as forbidding large highly reactive surfaces, it is worth pointing out that the non-hydrogenated diamond (110) surface is stable and the reconstruction of the diamond (100) surface is simple and predictable.
Again, however, the fundamental point is not that Jones objects to the proposed methods: he doesn't even know there are proposed methods.
In paragraph 13, he argues that biological systems do things in clever ways that nanotechnologists can't. In particular, he says "The genetic code does not specify the whole structure of an organism atom by atom. It provides a recipe that, if followed without too many blunders, produces a functioning creature capable of handing on much the same recipe." The original proposal for an assembler by Drexler includes an onboard computer. This computer would hold a description of the assembler. This description would not be encoded as a series of X,Y,Z coordinates for each atom but would instead apply an appropriate data compression scheme to the series of positional steps required to synthesize the structure (see page 426 et sequitur and page 434 et sequitur of Nanosystems). As the assembler design involves repetitious subunits, a series of subunits could be encoded in little more room than required to encode a single subunit. The precise steps for the construction of any particular instantiation of a subunit could be computed when the assembly of that subunit was initiated. Various other data compression methods are available as well. It appears that not only is Jones ignorant of the original proposal for an assembler, he is also ignorant of the work on self replicating systems[3, 5, 9, 19, 20] -- initiated by von Neumann's work on self replicating automata -- and much of computer science as well. The concept that assemblers are infeasible because data compression is impossible is stunning in its simplicity.
This, of course, entirely leaves aside the broadcast architecture[3, 9]. The broadcast architecture has no biological analog and appears to be strongly advantageous for manufacturing purposes. Perhaps Jones is unaware of this proposal.
Basically, in the broadcast architecture the assembler carries no on board blueprints (no "DNA") but instead receives and interprets instructions that are broadcast to it from a central source. Clearly, this eliminates the concern that the "blueprints" would be too large for the assembler to store.
On to paragraph 14. Jones argues that algorithms that would allow a large number of assemblers to cooperate on the construction of a complex object are infeasible. Jones previous efforts in algorithm design have been less than impressive. Although biological systems, using rather simple signaling mechanisms and algorithms, are able to produce structures as complex as the brain; Jones would have us believe that machines made by humans are forever barred from doing as well. Possibly we are unable to duplicate the "vital force" of living systems.
Nanosystems proposes the use of convergent assembly (chapter 14) for the construction of large systems. This eliminates the need to coordinate large numbers of autonomous assemblers, instead building up larger and larger components in a series of steps. Smaller components are assembled by small positional devices, and in their turn are assembled into larger components by larger positional devices. This is not to say that algorithms for coordinating the behavior of a large number of assemblers in solution are either infeasible or terribly difficult to design. One such algorithm is described briefly at the end of these comments.
Paragraph 15 is further unsubstantiated opinions on the part of Jones about the nature of self replicating systems. He argues that self replicating systems made from "unsubtle" materials such as diamond are infeasible. No doubt they lack vital force. No evidence is advanced in support of this remarkable claim. Again, Jones would be well advised to read the literature on self replicating systems. One good source is the 1980 NASA study on Advanced Automation for Space Missions[19, chapter 5], which has over 100 pages discussing self replicating systems.
Paragraph 16 is simply a rearrangement and restatement of his previous errors. He views small positional devices with astonishment but wisely (for once) refrains from overtly claiming that they are impossible. He does, however, say he doesn't believe that the "terabytes of brute-force coding" required to specify an assembler atom by atom is possible. Ignorant of both the design proposals for assemblers and the data compression methods that can reasonably be applied in the encoding process, it's not surprising that his estimates are unsupported. A rather generous estimate of the number of atoms in an assembler is one billion; Jones estimate of "terabytes" takes multiple kilobytes to encode the position of each atom. This is much more than required to list the X, Y and Z coordinates of each atom, which is itself a remarkably large representation. He must be employing a data expansion algorithm of considerable effectiveness. Perhaps he should seek a position consulting with Microsoft on their next release of Windows.
Paragraph 17 is remarkable in that Jones freely admits that the uncertainty principle is not a bar to the design of molecular machines. As a consequence, I will not give my standard discourse on the subject.
Jones compensates for this by asserting in paragraph 18 that Brownian motion and entropy raise "fundamental" questions. Indeed, he boldly claims that "... this damns their entire project." It appears that somewhere he obtained the impression that nanotechnological devices will have an infinite number of atoms in perfect structures with no defects. As chapter 6 of Nanosystems is dedicated to "Transitions, errors and damage" it's unclear where he got this idea.
In point of fact, the claim in Nanosystems is that error rates for properly designed molecular machines of about 10^9 cubic nanometers in size (many billions of atoms) can be made as low as several percent per year (page 158). This estimate is derived from a consideration of thermal, photochemical, radiation, and other damage mechanisms. The dominant error mechanism that appears difficult to substantially reduce is damage caused by background radiation. Nanosystems thus assumes that background radiation damage is the limiting factor in the error rate for properly designed molecular machines, and estimates this error rate based on published data about radiation damage. While the resulting error rate is remarkable by today's standards, it is certainly not zero. Note that the definition of an "error" is "any atom out of place." We add another item to Jones' reading list.....
Paragraph 19, in a section titled "Demon assemblers," discusses Maxwell's Demon. Paragraph 19 and 20 make no claims about assemblers. Paragraph 21 asserts that assemblers "look suspiciously like Maxwell's Demons." Why this is the case is not made clear. Evidently building atomically precise structures must violate the second law of thermodynamics. Existing atomically precise structures (such as proteins and the letters "IBM" spelled in Xenon atoms on a nickel surface) are given a special dispensation. Ah, the mysteries of thermodynamics!
Paragraph 21, the closing paragraph of the piece, degenerates into a series of questions that must be "properly formulated and answered" before nanotechnology can be taken seriously.
As these questions (and a great many others) have been systematically addressed in Nanosystems, it might be useful if we let Jones do his homework before inquiring whether he thinks they have been "properly formulated and answered."
A brief discussion of positional uncertainty caused by thermal noise and one method of dealing with it is available on the web.
The reader should understand that Jones is the creme de la creme of the nanocritics. He published a whole two pages in a (sometimes) respectable journal. He put his views before the scientific community for critical review and evaluation. Which he got.
I would ask future critics to at least read Nanosystems before laying claim to a deep and profound understanding of the foolishness of nanotechnology. It would make my life much easier. They might go farther and try and catch up with the current literature in the area. Or even attend one of the Foresight Conferences on Molecular Nanotechnology. An easier strategy would be to post their profound insights to "sci.nanotech." They might find out more quickly and easily how little they know.
Nanosystems was published in 1992 and is readily available in paperback versions for $25.00. Over 10,000 copies have been published. Various reading groups and courses have used it as a textbook. It provides a convenient source for the various technical issues and subjects that must be considered if we are to evaluate the feasibility of nanotechnology. No major errors in its methods or conclusions have been found. Skeptical statements about the feasibility of nanotechnology should be viewed with some skepticism unless the skeptic has at least read Nanosystems and is willing to point out the statements to which he objects. If critics are unable to do at least this, then either (a) they haven't read Nanosystems or (b) they don't object to any statement in it. Either way, it's hard to take such "critics" seriously.
The New York Times published an article in 1920 explaining that flight to the moon was impossible because there wasn't any air for a rocket to push against. Nature now joins the club with Jones' article.
Outline of an algorithm for coordinating the activities of multiple assemblers floating in solution.
Note that a variety of other approaches for the assembly of large objects from small objects are feasible. Convergent assembly is a particularly attractive approach.
I will assume we start with a vat of an appropriate liquid with some large number of independent assemblers. The assemblers have the ability to receive broadcast acoustic signals at some reasonable data rate (a megabit per second is technically unchallenging and should suffice). An analysis of acoustic transmission schemes can be found on page 472 of Nanosystems. We will assume that a single assembler, call it the "seed," has been designated and has coordinates 0,0,0. Any assembler that comes into contact with the seed is given coordinates from the seed based on its location with respect to the seed. For simplicity, we will assume that each assembler is mechanically constructed so that it can bind to six adjacent assemblers that are located on the faces of a cube centered around it. It can transmit information to or receive information from these assemblers. When two assemblers connect, they can either remain connected or disconnect again. Normally, two assemblers will disconnect. If, however, an assembler connects to the seed it will remain connected and will be given its coordinates from the seed. This coordinate will let the newly joined assembler determine its distance from the seed. At any given point in time, the cluster of assemblers will have a given diameter. Assemblers that are farther from the seed than the diameter will disconnect. Those that are closer than the diameter will remain connected. This will limit the growth of the cluster of assemblers, and will insure that most of the cluster is filled in before the size of the cluster is increased. The current diameter is increased and broadcast periodically to all the assemblers. Note that a variety of distance metrics are feasible, so the resulting "sphere" of assemblers can be any of a variety of shapes.
This algorithm will permit the growth of a cluster of assemblers, and will assign coordinates to each assembler. Broadcast transmissions can then address individual assemblers or arbitrary groups of assemblers. The obvious limitations of this algorithm are growth rate (each layer, one assembler thick or on the order of one micron thick, will take some fixed period of time to grow) and the handling of errors.
The most pernicious errors are caused by erroneous coordinates that are spread throughout the cluster. Local voting algorithms, however, can reduce the probability of this sort of failure to an acceptable level. A simple voting algorithm would have each assembler adopt as its own coordinate the coordinate agreed to by a majority of the assemblers in the vicinity. An assembler in disagreement with those in its vicinity would be isolated.
Once a sufficiently large cluster has been formed, the broadcast information can be used to specify which groups of assemblers are to build what substructures. As each assembler knows its own coordinates, it could listen to the appropriate part of the broadcast and ignore the rest. (Anyone familiar with the SIMD architecture will recognize the approach). Provided that the structural complexity of the object is not excessive, the relatively slow broadcast information will be sufficient to describe it. At a megabit per second, 3.6 gigabits of data can be broadcast in an hour, so the structure would have to be very complex before this approach would fail. As noted before, it is not necessary to select the stupidest and most obtuse encoding with which to describe the object.
If the growth rate of the cluster is too slow, multi-cluster algorithms which grow several clusters simultaneously can be adopted. Such algorithms would then have to merge clusters, just as the algorithm described was able to merge individual assemblers. Hierarchical algorithms that merge clusters of clusters, clusters of clusters of clusters, and so forth are feasible and should improve growth rates.
A variety of other algorithms are plausible. Again, it should be emphasized that the proposal described above is simple and can be improved upon. It is intended merely to point out feasibility.
1. "Technical boundless optimism," by David E. H. Jones, Nature, Vol. 374, No. 6525, April 27 1995, pages 835-837.
2. "Nano: the emerging science of nanotechnology;" by Ed Regis; Little, Brown; 1995.
3. "Nanosystems: molecular machinery, manufacturing, and computation," by K. Eric Drexler, Wiley&Sons, 1992.
4. "Theoretical studies of a hydrogen abstraction tool for nanotechnology," by C. B. Musgrave, J. K. Perry, R. C. Merkle, and W. A. Goddard III, Nanotechnology (1991) 2 187-195.
5. "Self Replicating Systems and Molecular Manufacturing," by R. C. Merkle (1992) Journal of the British Interplanetary Society, Vol 45, pp. 407-413.
6. "Computational Nanotechnology," by R. C. Merkle (1991), Nanotechnology, 2, pp. 134-141.
7. "Molecular Manufacturing: Adding Positional Control to Chemical Synthesis," by R. C. Merkle (1993), Chemical Design Automation News, 8, No. 9&10, page 1.
8. "Design-ahead for nanotechnology", by R. C. Merkle; in Prospects in Nanotechnology, by Markus Krummenacker and James Lewis, Wiley 1995.
9. "Self replicating systems and low cost manufacturing," by R. C. Merkle; in The Ultimate Limits of Fabrication and Measurement, M.E. Welland, J.K. Gimzewski, eds.; Kluwer, Dordrecht, 1994
10. "Risk Assessment," by Ralph C. Merkle; in Nanotechnology: Research and Perspectives, edited by B. C. Crandall and James Lewis, MIT press 1992.
11. "There's Plenty of Room at the Bottom," a talk by Richard Feynman at an annual meeting of the American Physical Society given on December 29, 1959. Reprinted in Caltech's Engineering and Science, February 1960, pages 22-36.
12. "Surface patterning by atomically-controlled chemical forces: molecular dynamics simulations," by Susan B. Sinnott, Richard J. Colton, Carter T. White, and Donald W. Brenner, Surface Science 316 (1994) pages L1055 to L1060.
13. "Positioning single atoms with a scanning tunnelling microscope," D. M. Eigler and E. K. Schweizer (1990), Nature 344, 524-526.
14. "Diamond Chemical Vapor Deposition," F. G. Celii and J. E. Butler (1991)Annu Rev. Phys. Chem. 42 pp. 643-684.
15. "Detailed surface and gas-phase chemical kinetics of diamond deposition," M. Frenklach and H. Wang (1991-I), Physical Review B, 43, pp. 1520-1545.
16. "Growth mechanism of vapor-deposited diamond," by Michael Frenklach and Karl E. Spear, J. Mater. Res. 3 (1), Jan/Feb 1988, pages 133-140.
17. "Chemical vapour deposition of diamond," J.C. Angus, A. Argoitia, R. Gat, Z. Li, M. Sunkara, L. Wang and Y. Wang (1993), Phil. Trans. R. Soc. Lond. A, 342, pp. 195-208.
18. "Thin film diamond growth mechanisms," J. E. Butler and R. Woodin (1993), Phil. Trans. R. Soc. Lond. A, 342 pp. 209-224.
19. "Advanced Automation for Space Missions," R. A. Freitas and W. P. Gilbreath (1980), National Technical Information Service N83- 15348.
20. "Self-Replicating Systems - A Systems Engineering Approach," G. von Tiesenhausen and W. A. Darbro (1980), NASA technical memorandum TM-78304, Marshall Space Flight Center, Alabama.
21. "Theory of Self-Reproducing Automata," by John von Neumann, edited and completed by Arthur W. Burks, University of Illinois Press, 1966.
This page is part of the nanotechnology web site.