DigitalSpace Corporation 343 Soquel Avenue, Suite 70, Santa Cruz CA USA 95062
email: contact our Webmaster
WWW: http://www.digitalspace.com
ABSTRACT
The Amoeba simulator is a software system designed to aid researchers in Molecular Nanotechnology to explore a variety of configurations of potential molecular machines. The simulator is aimed at the stage in molecular manufacturing and molecular machine operations in which molecular feedstocks must be transported, sorted, processed and end products accumulated. We term machines in this stage "molecular flow machines". A simple, functioning model of a molecular flow machine using staged rotors to sort multiple molecular feedstocks is illustrated. Features and limitations of the architecture are discussed. Some further potential modeling applications for molecular flow machines are covered, including chemical industry sensors and an artificial olfactory organ. It is hoped that Amoeba's visual modeling environment depicting these theoretical systems "in operation" will stimulate further support and development of the field. Collaboration to guide further work on the simulator is requested.
INTRODUCTION
As Ralph Merkle wrote recently in [Mer91], "we can plan the development of molecular manufacturing systems on a computer just as Boeing might 'build' and 'fly' a new plane on a computer before actually manufacturing it". Incented by this stated need, we have initiated an adaptation of our token processing environment to produce a simulator called Amoeba (in honor of one of natures' most effective molecular manufacturing systems). Amoeba has been designed to aid researchers in molecular nanotechnology to explore a variety of configurations of potential molecular machines.
The simulator is directed at the stage in molecular manufacturing and molecular machine operations in which molecular feedstocks must be transported, sorted, processed and end products accumulated. We term these types of devices "molecular flow machines" (MFMs). Amoeba is a work in progress and a great deal remains to be built before it could approach the usefulness of a design compiler. We will review where we feel the Amoeba capability fits into computational nanotechnology, the current features and limitations of its architecture, an early rotor sorting model running in Amoeba, and some further applications of MFMs which could be simulated using this tool. We will close with a request for collaboration.
BRIDGING THE INFORMATION PROCESSING GAP BETWEEN MOLECULAR CAD AND APPLICATION DESIGN
Molecular modeling and CAD tools are now being employed in molecular nanotechnology. One such tool, Crystal Clear by Geoff Leach of the Department of Computer Science, RMIT, Melbourne Australia, is designed to enable full atomic detail designs of diamondoid nanomechanical structures [Lea95]. Another, the Molecular Assembly Sequence Software (MASS) by Carol Shaw, can generate a sequence of reaction steps that might be used by an assembler to synthesize a desired diamondoid molecular structure[Sha95]. A third example which comes closer to direct construction of molecular structures is the Nanomanipulator, a virtual reality interface to a scanning tunneling microscope [TayRus94].
The use of these and other tools will ultimately generate a set of manufacturable structure designs, which in turn will be used to fabricate basic building blocks. When these building blocks are ready to be assembled into a final molecular flow system, a new information processing requirement will arise.
The flow of molecules through nanoscale channels to locations with receptors can be compared to the bytes flowing through a computer operating system's message layers to its functional objects. Therefore, the design of a molecular flow machine can ultimately be treated as an information processing problem. When the quantity of molecules flowing on channels and the number of sub-assemblies in a design grows very large, the information processing problem will grow, possibly to the level of complexity of an operating system or silicon simulation.
We believe that the information processing requirements for accurate modeling of non-trivial molecular flow and other nanoscale systems will be significant. By focusing on simulating simple, near identical sub-assemblies and assuming a simple, well-defined environment, we hope that the Amoeba system will be capable of climbing this information processing curve. This approach corresponds with the recommendations found in appendix A.4.4 of [Dre92].
Given collaboration and further development, we hope that the Amoeba simulator can serve as a bridge between basic structural building block simulation and modeling of fully designed molecular flow machines and other applications in molecular nanotechnology.
FEATURES AND LIMITATIONS OF THE ARCHITECTURE
Amoeba structures are basically finite state machines in which there are many parallel state changes (tokens) passing through the system and accumulating at certain states (vesicles).
Amoeba Parts and Operation
Amoeba structures are built up from a small number of simple parts. It is hoped that this will permit a good mapping to a manufacturable topology. These parts include tubes, vesicles, detectors and functions. Generic tokens, representing molecules, flow through tubes and accumulate in vesicles where they are exposed to one or more detectors. The removal of tokens from vesicles by detectors is analogous to molecules binding to receptors. Functions are the objects activated by token detection. Functions are generic algorithmic plug-ins which can simulate a variety of operations, including transport into another vesicle by a sorting rotor as described in chapter 13.2 of [Dre92], application onto reactive encounter mechanisms for molecular milling as discussed in chapter 13.2.3 of [Dre92] or switching and motion of gates or logic rods covered in chapter 12 of [Dre92].
Features of the Architecture:
The Amoeba architecture was optimized both for performance and flexibility. It was felt that the quantity of molecules passing through a non-trivial MFM will be large and the number of sub-assemblies will also be substantial. In order to provide real-time feedback to designers, the kernel would have to provide high throughput of token travel steps, token detections, function activation and processing. In addition, the flexibility to add whole new systemic functions and localized plug-in features was deemed essential, as the modeling requirements will likely change repeatedly. Some features of the architecture include:
Other Systemic Features for Possible Addition
A number of new systemic features could be developed:
Other Requirements Necessary to Move Toward a Design Compiler
The Amoeba system compiles a simple textual design language into a binary image for execution. The Amoeba language can be entered in any text editor and resulting structure designs can be cut and paste into files or electronic mail systems for distribution. It is planned to add a visual editor to allow direct creation and manipulation of structure designs in two or three dimensions. For Amoeba to qualify as a design compiler for molecular manufacturing systems, it would have to model reaction geometries, flow rates, fluid viscosity, and many more factors, as outlined in chapter 14.6.5 of [Dre92]. It is hoped that the further development of the operating kernel and plug-in functions will gradually move the simulator toward this goal.
A DEMONSTRATION MODEL: A MULTIPLE FEEDSTOCK ROTOR FILTER
A simple, functioning model using staged rotors to sort multiple molecular feedstocks is illustrated in Figure 1. Two feedstocks enter at A and B. Sorting rotors, represented by the horizontal tubes labeled with a rotor symbols, order the molecules in the feedstock through successive stages of filtering. Undesired molecules are drawn out of the in-line vesicles at each stage. The two streams converge and are accumulated as a mixture at point C. Functional objects at points D, E, and F could provide feedback to regulate the feedstock input rate or could draw off the mixture for processing.
Figure 1: a simulation of a molecular flow machine having two cascaded sets of sorting rotors
Figure 2: a close up view of a vesicle containing tokens (representing molecules) with some binding to sites on the sorting rotor and the draw-off detector.
Figure 2 shows detection of tokens, representing molecules binding at sites. The upper tube draws off undesired molecules. The lower tube, representing a sorting rotor, draws off a sequence of ordered molecules to place them in the next vesicle. Molecules are represented by colours. Specific behavior of the sorting rotor, including error rates is not modeled here.
FURTHER SUGGESTED APPLICATIONS FOR MODELING
The small scale of molecular flow machines might suggest that their economic value will come in the processing of small volumes of valuable materials. As a replacement thyroid gland, for example, a nanosystem would save a lifetime of health problems and costly medication in return for the throughput of a relatively small amount of fluids. Whereas in large scale bulk process industries, such as petroleum refining, nanosystems would have a difficult time competing on a nano-dollars per molecule basis. We suggest, however, that molecular flow machines have wide applications in bulk process industries in sensing, measurement and analysis.
Chemical Industry Sensors
In the chemical industry, it is a common practice to flow feedstocks into a reactor where they come into contact with catalysts. A regular three dimensional geometry in the reactor is used to encourage or discourage reactions. As it turns out, attempts to design catalysts to perform specific tasks have not been very successful. If catalysts could be constructed using nano-molecular building blocks, then they would have a major impact on the bulk chemical processing industries. In this case, a MFM might be designed to operate within its own feedstock, rather than having the feedstock pass through it on narrow channels.
A possibly more promising area which MFMs could play a role in bulk process industries may be as in-line sensing, measurement and analysis devices. In petroleum refining, samples of product are sent to off-line laboratories for testing of important properties such as viscosity or octane rating. Exceeding a target rating by even a few tenths of an octane can be costly while falling short is unacceptable to the consumer. Octane is actually tested by running a sample through an internal combustion engine. The use of neural network software processing the infrared spectrum of product to correlate to octane has not been successful.
The use of MFM devices affixed to the insides of pipes to measure octane or viscosity and feed directly back into a system controlling blending would be of significant interest to the bulk petrochemicals industry. Other MFM devices could simply detect what product is moving in a pipeline. Operators inadvertently feeding incorrect product into a pipeline can be very costly blunders and there is currently no way to detect these errors. The use of molecular flow systems in this application is equivalent to reproducing part of the company laboratory and distributing it throughout the refining process, similar to medical applications for distributing molecular mechanisms in the human body as described in [Mer95].
Artificial Olfactory Organ
An application of MFMs which emerges from biomedical research is an artificial olfactory organ. A sensitive "artificial nose" would have many valuable applications including: in the protection of the environment by early detection of toxic emissions, in medicine to detect airborne viruses, and in manufacturing, we might suggest that the proper operation of a cookie baking machine can be largely determined by the aroma wafting from the cookies.
Odors are detected in our noses in the olfactory epithelium by about 1000 different odor receptors. Odors are complex and must bind their component molecules to several receptors to be detected. Researchers believe that the mammalian brain, working through a set of correlators in the olfactory bulb called the glomeruli, determines the precise combination of receptors activated by an odor. To accomplish this, approximately ten million axons distribute signals from receptors to the glomeruli, which then communicate with higher centers in the brain [Axe95].
Even if we could reduce the complexity of natural olfactory processors by an order of magnitude, we would still be faced with a daunting engineering task to construct an artificial odor detector to match the capability of a mouse (let alone a bloodhound). It may be that systems which must rely on millions of receptors, channels and correlation processes will be made feasible only through molecular manufacturing.
We suggest that the above applications fit the basic design of molecular flow machines and may be worth modeling for their perceived relevance to industry.
A CALL FOR COLLABORATION
Work on the Amoeba simulation system could be greatly advanced by collaboration with one or more of the research groups active in the field. As our focus at DigitalSpace is Computer Science and our background is professional software development, we require more education in the fields which are relevant to molecular nanotechnology. In return, we can bring our knowledge of software architecture and its use in simulation to any collaborative effort. It is our goal within a collaboration to bring the Amoeba simulator up to a level which will make it useful for testing configurations of molecular flow machine designs and demonstrating the value of potential molecular nanotechnology applications to a wider audience.
ACKNOWLEDGMENTS
We would like to thank Cliff Detz of Chevron Corporation for his lucid description of possible applications of MFMs in the chemical industry.
REFERENCES
[Axe95] Axel R "The Molecular Logic of Smell" Scientific American, New York, Oct. 1995.
[Dre92] Drexler K E "Nanosystems: Molecular Machinery, Manufacturing, and Computation" John Wiley & Sons, New York, (1992).
[Lea95] Leach G "Crystal Clear: a molecular CAD tool". Contact: Geoff Leach, Department of Computer Science, Royal Melbourne Institute of Technology, GPO Box 2476V, Melbourne, VIC 3001, Australia.
[Mer91] Merkle R C "Computational Nanotechnology", Nanotechnology, 2 (1991), Institute of Physics Publishing, pp. 134-141.
[Mer92] Merkle R C "Self Replicating Systems and Molecular Manufacturing" Journal of the British Interplanetary Society, Volume 45 (1992) pp. 407-413.
[Mer95] Merkle R C from a draft for an invited talk at the Second Annual Conference on Anti-Aging Medicine & Biomedical Technology for the Year 2010, December 4-6 1994, Las Vegas, Nevada.
[Sha95] Shaw C from World Wide Web documents describing MASS available at the URL (October 1995): http://www.portal.com/~carols/mass.html
[TayRus94] Taylor, Russell, "The Nanomanipulator: A Virtual-Reality Interface to a Scanning Tunneling Microscope" Ph. D. Dissertation, University of North Carolina, Chapel Hill, TR94-030, May, 1994.