NANO@HOME Project Proposal

Introduction

One of the major objections to nanoassembly, involving the throughput of a single atomic force microscope (AFM) tip, has recently been negated by work of Hong & Mirkin [5] at Northwestern through the use of parallel dip-pen nanolithography.  Work at Hewlett Packard [6] and IBM [7] clearly show that they too are thinking in terms of massively parallel reading and writing of nanoscale information.

In my work regarding the megascale limits of nanoscale construction [8], it became clear that the characteristics of molecular nanoassembly (atomic precision, ultrastrong materials, programmability, self-replication, etc.) largely eliminate the concerns in traditional manufacturing or economic development such as energy and materials costs, capital costs, return-on-investment requirements, possible competition, lack-of-infrastructure, etc.  One possible problem involves systems complexity, but we may hope that that problem will be solved by continued evolution of software based design tools, as has been the case in the semiconductor industry.  The most significant barriers to very large scale applications of nanotechnology are (a) the element abundances; and (b) the availability of designs.  The first we can do little about until we have tremendously more energy at our disposal, probably requiring the disassembly of the gas giant planets.  So the primary barrier we face is the problem of obtaining functional nanoscale machine designs.

One can consider proteins as nanoscale machines. Nature has produced millions of them, but the methods used require extremely long time scales. There are now groups working on de novo protein design, such as Richardson's Labs at Duke [9], the Hecht Lab at Princeton [10], and the Mayo group at Caltech [11] and others. At most there are probably only few dozen groups (a few hundred people) working in this area. It is possible to consider buckytubes a nanoscale part or a quantum dots or single electron transistor as nanoscale devices, but here again, with the possible exception of buckytubes, there are only a few groups working in these areas.  These examples are unable to fulfill the potential, such as medical nanobots [12], that molecular nanoassembly with diamondoid or sapphire materials enables.  So we may encounter a situation in the near future in which we have the ability to construct diamondoid nanoscale parts but we will have no designs for the parts, tools and more complex systems that might be constructed.  At the current time the author knows of only 3 complete molecular scale diamondoid designs that have been worked out by Eric Drexler [13]^{FN8}. That seems to be an extremely small number when compared with the thousands of pages of "parts" found in catalogs on the desks of the average mechanical or chemical engineer working today.

Consider the 200 billion dollar a year semiconductor industry that is built on only a handful of transistor designs, that are compiled into dozens of functional libraries, that are laid out in a variety of ways to produce thousands of varieties of integrated circuits.  It has taken us 40 years to get to the stage where teams of electrical engineers have the basic parts and libraries that allow them to design and test circuits with millions of transistors in only months. In contrast, full scale nanobots contain billions of atoms! Will we be in the position in 2015 of having the ability to do diamondoid molecular assembly but find ourselves without any of the benefits of nanotechnology because our nanoengineers will be in the position of an electrical engineer of 1970, having to physically cut and paste little strips of plastic to make the IC mask designs for single transistors one by one?

It seems clear, the amount of attention devoted to "The Lack of Designs Problem", is far too little.  Furthermore, it is doubtful whether this problem can be solved by throwing money or people at it.  The United States essentially did this in the early-mid 1980's when it was realized that the HIV epidemic was a problem.  Even with a great deal of money from NIH available, progress was slow for many years, in part because basic research needed to be done, but to a greater degree because of the lack of people sufficiently trained in virology, immunology, epidemiology and a variety of other medical and biotechnology specialties who could effectively and productively conduct the required research.  I call this the "You can't push on a string" phenomena.  Even though the National Nanotechnology Initiative has to a degree legitimized work in nanotechnology and may pull people into this research area, there are still others, such as the biophysicist Steven Block of Stanford who make it clear that, "nanotechnologists ought to distance themselves from the giggle factor" [14].  Statements like this are likely to discourage young engineers from training themselves in the technologies and methods that will be required for the rapid design of nanoscale parts in a dozen years.  Young engineers obtaining an education, may not have the experience or foresight to plan for careers that will develop in dozen years.  To mitigate this problem, "We must begin by being there".  We must assume diamondoid nanoassembly is feasible and focus our attention on the lack of designs.

How can we solve "The Lack of Designs Problem"?

Nano@HOME has several obvious benefits:

1. It "popularizes" nanotechnology research. That may translate into increased general interest and possibly support in areas such as research funding.  Nano-widget screen savers showing the "state-of-a-design" would buy more interest in nanotechnology than millions of dollars of advertising.
2. It functions as a bridge to interest people in the science & engineering.  Currently, it is very difficult for most people to do Science@HOME.  The fact that you could design molecular parts, then potentially assemble them, then send them off to a supercomputer lab where they would actually be run, lets individuals, even children, become actively involved in scientific development.  The trick of letting the computer do most of the hard work (design layout) and letting the people do the easy work (yes, no, good, bad, here why don't you try this...) allows people to approach this from a self-educating aspect.  Discussion communities such as mailing lists, newsgroups, chat rooms, etc. would develop and evolve, sharing goals, ideas, strategies, successes, failures, etc.
3. There are thousands of engineering students in schools around the world that currently have no nearby nanotechnology centers or access to nanoengineering courses.  Nano@HOME provides them with a way of learning about nanotechnology.
4. There may be individuals who have no interest in SETI@HOME or GIMPS because they doubt the practical application of any results that might be produced.  Such individuals might find Nano@HOME very appealing.
5. There are many software developers in the extended nanotechnology "aware" community who have no easy way to contribute to the development of nanotechnology (i.e., no AFMs@HOME).  However, they can look at code, debug it, enhance it and run it on the "big iron" they typically tend to have at their disposal.  Over a 10 year period, hundreds or thousands of programmers might pick up enough nanoengineering expertise to make them valuable contributors to the large development teams that nanobots and other complex nanomachines will require.
6. Possible approaches exist to use the project to increase support for the Foresight Institute, the Institute for Molecular Manufacturing and an Open Source Nanodesign Initiative [28].

Nano@HOME Licensing

The Concept for Nano@HOME

1. Derive mechanical engineering "templates" real world macro-scale parts: (rivets, screws, beams, levers, pipes, valves, wheels, axles, hinges, joints, springs, gears, drills, grippers, switches, drills, fans, motors, pumps, etc.) that have been shrunk to nanoscales^{FN4,FN9}.  This will require obtaining cooperation and possible participation from professional societies [21] and perhaps large corporations [22].  The shrunken templates are put into a database that distributes them as "work units".
2. Construct a program that randomly, possibly using rules developed in an evolutionary learning phase, fills the template space with atoms, presumably starting with carbon (diamondoid structure), then varying as necessary to meet the physical space and molecular bonding requirements dictated by the template.
3. Compare the resulting design with the template according to various success criteria, e.g. best fit to 3D coordinates, least atomic bond strain, ease of assembly, etc.  Mutate and select as necessary to evolve better solutions.
4. Submit the design to large scale molecular simulation groups, for example, those at Caltech [23] or LSU [24], who then "test" the part for strength, modes of failure, robustness, etc.
5. Possibly loop back to the mutation and evolution steps.
6. Distribute this software to large numbers of people who use it to evolve designs for various parts.
7. Install the resulting designs into the Nano-widget Library for use in larger, nanobot scale devices.

What is needed?

Programs such as HyperChem have been able to do molecular modeling for many years.  Unfortunately this is a commercial product.  DiamondCAD from Zyvex or Will Ware's  NanoCAD tool are available and could serve as a foundations for Nano@HOME.  Other public domain molecular modeling packages are probably available.

Methods for computer driven evolution of mechanical designs and even the design of mobile robots have been explored by the Brandeis DEMO group [26].

If nanodesigns were developed by random mutation and selection, they could then be gradually improved through the use of Genetic Programming techniques [27].

So, it would appear that the task required is to stitch together the pieces that have been outlined into working programs that can function collectively as Nano@HOME.  If for various reasons the existing software, methods, licenses, etc. are unavailable or inappropriate, then new pieces that perform the various tasks would be required.
 

Possible Action Items

• Foresight/IMM would need to determine their interest in participating in such a project.
• The corporations and academic research labs involved in the work outlined above should be contacted with regard to both their interest in participating and the availability of their code.
• A license agreement would need to be written for any code to be released and legal agreements executed to put any code to be released under the license [See a related discussion in 28].
• The professional organizations [21] need to be contacted about promoting the project.
• The companies [22] with large macroscale part databases need to be contacted and they need to be "sold" with regard to making available those databases.  Methods for scaling those part designs down to "protein" sizes developed  (AutoCAD format to PDB format???).
• A database in which to store and distribute designs needs to be developed.
• The central engine of atomic-fill-in, mutation, satisfaction-of-critera judgment, and design reporting needs to be developed and tested.

(Currently this may be unavailable).

• Related discussions @ Nanodot:
• Nano@HOME Proposal
• Open Source CAD Code for MEMS
• Nature Magazine: Internet Computing and the Emerging Grid, Ian Foster, (7 Dec. 2000)

1. The demise of Moore's Law has been predicted for decades, however as the references below show, the problems are recognized and progress continues.  Even as Muller and coworkers announce the limits of traditional gate architectures, other groups at Berkeley and  Bell Labs  announce novel structures that transcend traditional limits.  Predicting the demise of Moore's Law is likely to be difficult since there are large numbers of corporate executives and research scientists who do not want to be "on board" when the ship sinks.  While atomic assembly will eventually be necessary in the semiconductor industry (perhaps circa 2015-2020), the industry is unlikely to function as a driver for molecular nanotechnology development because of the huge investment and inertia associated with current production methods.
2. At the current time, the "ideal" computers for running massive atomic simulations do not exist.  The ideal architecture would be a cellular automata architecture with millions or billions nodes utilizing a processor-in-memory or IRAM architecture and an instruction set specifically designed for interatomic forces calculations.  Historically, the architectures that could have been useful were the machines built by nCUBE and Thinking Machines Corporation.  It is likely that IBM's forthcoming Blue Gene architecture (original proposal; current plans: here & here) will be very useful in running simulations of nanodesigns. Shortly after this proposal was first published, Wired News announced in their article "Protein Fiends Join the Fold", the Folding@Home project, to do distributed protein folding.  This is based on the Mithral Client-Server SDK.  This problem is difficult as the principle investigator,  Dr. Vijay Pande, indicated that it would take 10,000 computer days to model even 1 simple protein folding.  It will not be clear until publications become available for review, whether the internode communications bandwidth requirements have been minimized sufficiently to allow this to work effectively on highly distributed, slowly interconnected nodes (the basic Internet situation in Y2000).  If this problem has been solved successfully, it would suggest that both the design and simulation of nanomachines is feasible in @Home environments.
3. The author believes that patents themselves are not bad as they function to create a level playing field in which small companies with novel ideas may compete effectively against larger companies.  Where they may be broken is when they are used to deny access to technologies, as Jeremy Rifkin has attempted to do, or where they function to produce highly inflated prices, as is the case with tests for certain cancer genes and many drugs or when they are used to patent non-novel preexisting (but unknown) information as is the case with natural genetic sequences.  They are certainly broken and are presumably invalid in the case of patenting preexisting software methods.  With regard to nanoscale designs, it is highly unlikely that there will be single "best" design for a specific type of nanopart.  With millions of computers potentially doing the designs, there could be a variety of patented parts whose patent holders could be in competition with one another for the use of their designs (so the royalties charged should not become onerous).  Efforts by various nonprofit or governmental organizations to create public domain parts and tool kits would decrease the time periods during which patents could be used by the holders to charge high prices.
4. I assume that nanobots and nanodevices will consist of parts that must be assembled.  I do this because the only examples of nanoscale devices we currently have, bacteria and eukaryotic cells, clearly contain many individual parts.  However, if you have molecular nanoassembly, then perhaps you may build things as single complete structures (i.e. no post-manufacturing assembly required).  Is there any engineering principle that says elements of nanobots or nanodevices must be manufactured separately and assembled in some way?  Not to my knowledge.  It seems to me though that a total nanodevice manufacturing system would be much more complex to design than one that constructs or grows the simple parts mentioned.  For that reason, it seems likely that it will come much later in the evolution of nanodesign.
5. After the initial release of this document, additional groups involved in distributed computing were brought to the attention of the author.  They include The Beowulf Project and Distributed Science and The ProcessTree Network^TM.
6. Subsequent discussions at Nanodot, have revealed several additional sources for CAD programs or related tools, including: Intellisense (Software Products); Microcosm Technologies (Products such as MEMCAD and FlumeCad); Open Cascade (FAQ; Projects).
7. There is a gRobots Project @ www.grobots.org attempting to create a virtual environment in which self-replicating systems with evolving software may be tested safely.  This would appear to be an extension of Artificial Life concepts into the nanotechnology arena.  In the authors opinion, Nano@Home needs to precede the gRobots Project, because we have lots of examples of being able to build evolving software and software based self-replicating systems.  What we lack is the underlying components from which nanoscale systems will eventually be constructed.  Nano@Home would produce libraries of parts, which are then assembled into designs that would presumably be tested and potentially evolved in the type of safe environment the gRobots group envisions.
8. 24/03/2002: Subsequent research by the author revealed there more than 3 nanoscale parts.  They are however much simpler than those cited since they are primarily sleeve bearings.  See [29] Chapters 9 & 10 and [30, Sections 2.3.2-2.4.1, esp. pgs. 55, 62 & 63].  At most, there are perhaps a dozen human designed nanoscale parts.
9. 24/03/2002: There are examples of many kinds of parts in the McMaster-Carr Catalog.

The author would like to thank Scott T. Jenson and Eric Heien for constructive comments.

1. Public Statement by Jim Von Ehr, seen on (KING?) TV in Seattle by the author circa March 2000.
2. Conversations between author and Robert Freitas, Jr., circa May 2000.
3. Jon William Toigo, "Avoiding a Data Crunch", Scientific American, May, 2000.
4. Interagency Working Group on Nanoscience, Engineering and Technology (IWGN), "Nanotechnology Research Directions: IWGN Workshop Report", Ch. 1.7.7, February, 2000.  Worth noting is that the discussion of Moore's Law has been highly divisive, e.g.
• Service, R. F., "Can Chip Devices Keep Shrinking?", Science 274(5294):1834-1836 (December 13, 1996).
• Brenner, A. E., "Moore's Law",  Science 275(5306):1401-1404 (March 14, 1997).
• Packan, P. A., "Pushing the Limits", Science 285(5436):2079-2081 (Sept. 24, 1999).
• Schulz, M., "The end of the road for silicon?", Nature 399:729-730 (June 24, 1999).
• Muller, D. A., et. al., "The electronic structure at the atomic scale of ultrathin gate oxides", Nature 399:758-761 (June 24, 1999)
• Bell Labs Physical Sciences Research: 60 nm Nanotransistor
• The FinFET and Ultra-thin Body MOSFET by groups lead by Chenming Hu and Tsu-Jae King from Berkeley.
• Miller, R. D., "In Search of Low-k Dielectrics", Science 286(5439):421-423 (Oct. 15 1999).
• Vertical Transistors from Don Monroe and Jack Hergenrother at Bell Labs. [News Release Nov. 15, 1999]
• Bell Labs Ballistic Transistor Has Virtually Unimpeded Current Flow [News Release Dec. 6, 1999].
• Ziemelis, K.,"The future of microelectronics", , Nature 406(6799):1021 (August 31, 2000).
• Peercy, P. S., "The drive to miniaturization", Nature 406(6799):1023-1026 (August 31, 2000).
• Ito, T. & Okazaki, S., "Pushing the limits of lithography", Nature 406(6799):1027-1031 (August 31, 2000).
5. S. Hong and C. A. Mirkin, "A Nanoplotter with Both Parallel and Serial Writing Capabilities", Science 288(5472):1808-11 (9 Jun 2000).
6. "A Decade Away: Atomic Resolution Storage", Scientific American, May, 2000.
7. "'Punch Cards' of the Future", Scientific American, May, 2000.
8. R. J. Bradbury, Matrioshka Brains (megascale computers constructed using nanoscale technologies) (1998-2000).
9. The Richardson's Home Page (http://kinemage.biochem.duke.edu/).
10. The Michael H. Hecht Home Page (http://www.princeton.edu/~hecht/research.html).
11. The Mayo Group Home Page: (http://www.mayo.caltech.edu/).
12. The most completely described examples of these are from Robert A. Freitas, Jr., and include Respirocytes, documented in "A Mechanical Artificial Red Cell: Exploratory Design in Medical Nanotechnology": http://www.foresight.org/Nanomedicine/Respirocytes.html (1996); "Clottocytes: Artificial Mechanical Platelets," Foresight Update No. 41, 30 June 2000, pp. 9-11:  http://www.imm.org/Reports/Rep018.html; and Endotheliocytes in an unpublished manuscript "Vasculomobile Nanorobot Aggregates as an Alternative to Nanosubmarines in the Rapid Treatment and Repair of Arteriosclerotic Lesions".  Nanomedicine, Volume I contains functional and capability descriptions for a dozen or more different types, but more detailed designs await the completion of Volumes II and III.
13. Designs are @ The Institute for Molecular Manufacturing (www.imm.org): Neon Pump, Fin-Motion Controller, and Differential Gear.
14. Steven M. Block, "What is Nanotechnology?", Nanotechnology and Nanoscience: Shaping Biomedical Research, June 25-26, 2000 NIH.
15. H. P. Shuch, "Distributed Processing Goes Galactic", Proceedings of Trenton Computer Festival 2000, 6 May 2000.
16. SETI@HOME (http://setiathome.ssl.berkeley.edu/).
17. R. J. Bradbury, "Misconceptions Regarding SETI, Dyson Spheres and the Fermi Paradox", (July 2000).
18. Kardashev, N. S., "Transmission of Information by Extraterrestrial Civilizations", Astronomicheskii Zh. 41(2):282-285 (1964). [Soviet Astronomy 8(2):217-220 (1964).]
19. V. Vinge, "The Coming Technological Singularity: How to Survive in the Post-Human Era", Vision-21:Interdisciplinary Science and Engineering in the Era of Cyberspace, NASA-CP-10129, pp. 11-22, Conference held March 21-23 at NASA Lewis Research Center, Westlake, OH (1993).
20. See: Distributed Computing Projects (www.mersenne.org), esp. The Great Internet Mersenne Prime Search, aka GIMPS@HOME.
21. For example: The American Society of Mechanical Engineers, The National Society of Professional Engineers,  The American Chemical Society, The Materials Research Society, etc.
22. Included would be any companies with large numbers of mechanical engineers, e.g. Boeing, Lockheed Martin, Case Corporation, GM, Ford, etc.
23. See: Materials and Process Simulation Center @ CALTECH  (William Goddard's Group).
24. See: Concurrent Computing Laboratory for Materials Simulations @ LSU
25. See the 3D Simulator under New Inventions @ Lego Mindstorms (http://www.legomindstorms.com).
26. See the Brandeis DEMO (Dynamical and Evolutionary Machine Organization) Project, which has used computers to evolve designs for a Lego Crane, Bridge and Table.  Also the EvoCAD (intelligent CAD System); See Also: Evolutionary Design by Computers (Peter Bently, ed.), Morgan Kaufmann (1999).
27. John R. Koza, Genetic Programming: On the Programming of Computers by Means of Natural Selection, MIT Press (1992); John R. Koza, Genetic Programming II : Automatic Discovery of Reusable Programs (Complex Adaptive Systems), MIT Press (1994);  J. R. Koza, et. al., Genetic Programming III: Darwinian Invention and Problem Solving, Morgan Kaufmann , (1999); Wolfgang Banzhaf, et. al., Genetic Programming: An Introduction : On the Automatic Evolution of Computer Programs and Its Applications Morgan Kaufmann (1998); See also: www.genetic-programming.com.
28. Bryan Bruns has a draft paper "Open Sourcing Nanotechnology" [http://www.cm.ksc.co.th/~bruns/open_mnt.htm] (2000); There is a  related discussion at Nanodot.
29. Drexler, K. E., Nanosystems: Molecular Machinery, Manufacturing and Computation Wiley-Interscience (1992).
30. Freitas Jr., R. A., "Nanomedicine: Volume I", Landes Biosciences (Oct 1999).

• Kraulis, P., "MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures", J. Appl. Crystallogr. 24:946-950 (1991). [MOLSCRIPT Home]
• Bacon, D. & Anderson, W. F., "A fast algorithm for rendering space filling molecule pictures", J. Mol. Graph 6:219-220 (1988).