by David Bradley
While some may scoff at the thought of creating sub-microscopic machines, nature has been doing it quite successfully for some time to say the least. Scientists are now turning to software to help them design their own nanomachines.
Everyone is by now familiar with the tiny, futuristic world of nanotechnology. The media is full of countless pictures of electron micrographs showing the wonders of this new field. There is "Nanoman", sub-microscopic writing, a minute atlas and even a buckyball abacus. The protagonists of nanotechnology are rubbing their hands in anticipation as scientists near the ultimate goal of a universal molecular assembler that can be programmed to build sub-microscopic machines - fulfilling the molecular prophecies of nanotech pioneer K Eric Drexler.
Like other nanotechnology practitioners, Drexler is based in California, at the Institute for Molecular Manufacturing in Palo Alto. Although these machines are still a phenomenon of the virtual world, he believes that sub-microscopic devices created from the manipulation of a very small number of atoms will someday change the world we live in. He took his case to Al Gore's Commerce, Science and Transportation Subcommittee on Science, Technology and Space on June 26 1992.
Drexler's dream is of molecular machines that assemble themselves, make copies and then build other nanoscale machines, with various functionalities for everything from attacking individual diseased cells to creating a pork chop from kitchen waste by way of a molecular computer.
The techniques used to build Nanoman and the bucky abacus, which involve manipulating individual atoms with macroscopic devices such as microscopes, will likely not provide the ultimate method of creating nanodevices, says Bruce Damer, a virtual reality expert at DigitalSpace Corporation in Santa Cruz, California. "I feel that successful nanotech will definitely not follow the 'hand-tool' approach of scanning transmission microscopes, but will emulate natural processes, such as deposition structure building - like the formations in Yellowstone Park. Another might be trapped flow reactions as found in living cells, where detectors work with membranes and carriers to bring reactants to a location," he says.
Hand tools are too clumsy to be an efficient means of creating nanodevices, despite the advances in control that have been made in the last ten years. Many chemists favour supramolecular chemistry. Julius Rebek Jr, now at the Scripps Research Institute in La Jolla, California, is designing small molecules that self replicate. Nobel prize winner Jean-Marie Lehn, at the Louis Pasteur University in Strasbourg, France, has developed numerous sensor molecules and chemical models that emulate enzymes, receptors and channels. Fraser Stoddart, now at UCLA, has built molecular scale switching devices. These and many other chemists are taking the first tentative steps towards molecular machines from the vantage point of the chemistry laboratory. Tying together these threads might eventually lead to working devices that can control, manipulate and process other atoms and molecules.
While the supramolecular chemists rely on intellectual design based on chemical principles, those collaborating with the computer scientists are taking a more adventurous route. Drexler's original idea was to build a computer model of an "assembler" that would construct the bearings, gears, tunnels and vessels of a working device, for example. The model would reveal whether the atoms needed for such a structure will fit together to form a stable "supermolecule" on-screen. If they do then it is just a quantum leap to building the structure in the test-tube. According to Ralph Merkle of Xerox parc in Palo Alto California,
"Present capabilities preclude the manufacture of any but the most rudimentary molecular structures. The computer-based design and modelling of molecular machines is, however, quite feasible with present technology". More to the point, he says, modelling is a cheap and easy way to explore the diversity of putative molecular machines, allowing dead ends to be avoided.
Damer has developed such a modelling tool for nanotechnologists. Dubbed AMOEBA after the single-celled, archetypal natural nanodevice, the tool has brought together concepts from biological modelling, control systems, artificial life and agent software.
The software's main purpose is to take researchers one step on from designing single nanocomponents, allowing them to look at how molecular feedstocks might be sorted and processed, and how end products may be accumulated by a nanomachine. AMOEBA uses a toolkit of simple components including tubes, vesicles, detectors and functional end nodes to make a virtual device that Damer refers to as a "molecular flow machine". The program's output uses generic tokens to represent either chemical, electrical or photonic signals, so a processing device constructed with AMOEBA could be as applicable to making pork chops as to designing high density logic circuits from molecules.
According to Damer, AMOEBA allows potential devices to be built in a virtual environment so that their optimal construction can be determined - following Merkle's plan - in the long term enabling the construction of a model of a complete assembler. AMOEBA, Damer says, was designed with advice from oil company Chevron, which is interested in nanoscale sensors placed in-line inside pipelines and reaction chambers for monitoring product composition and processing. "Simple detectors that can measure fuel mixtures and octane would be worth billions to this industry," says Damer.
Damer says that AMOEBA was, at the time of writing, being recast in the Java language. "It will be used to drive virtual environments in the new version of VRML 2.0," he explained. The engine has been dubbed nerves and is a simple, high-performance flow net. Damer is enthusiastic that once the software enters the public domain in a 3-D visualization environment it should become a powerful tool for nanotechnology research.
Programmer Carol Shaw - wife of Xerox's Merkle - has also designed a nanotechnology modelling program. This allows researchers to ascertain what might be feasible once an assembler has been built. Her program, Molecular Assembly Sequence Software (mass) generates a sequence of chemical reaction steps that could be used by an assembler. The steps follow a process akin to chipping away at a piece of diamond at the atomic level to sculpt a tough device such as a gear. "You can step through using the software and see how parts of the so-called diamond 100 and diamond 110 surfaces might be generated [to create the gear]," explains Shaw.
mass makes the big assumption that molecular tools are available to perform necessary operations such as site-specific addition (deposition) or removal (abstraction) of hydrogen atoms and insertion of carbon atoms, etc. to build the devices. The program starts with a model of the desired target structure, in pdb format, and aims to determine how to disassemble the structure. "Since the automation of the assembly/disassembly process hasn't been implemented, however, it would be laborious to create the assembly sequence for an entire object," adds Shaw.
The idea is analogous to retrosynthetic analysis used by chemists to work out what simple chemical building blocks to use in order to synthesize a more complex molecule. An important feature of mass is that it allows application of the tools in a way that is not chemically reasonable. This can be either a good thing or a bad thing. Initially, it means that separate analysis is required to verify that intermediate structures are stable. However, freeing the designer from the constraints of known chemistry can encourage more creativity.
Shaw's program is currently only available for the Apple Mac but is freely downloadable from the Web. Shaw admits she has not investigated the possibility of collaborations. "A lot of people download the mass software, but I don't have information on who they are, or how much they actually use it," she says.
Computer scientist and electrical engineer William Ware of Xionics Document Technologies in Massachusetts USA feels that although nanotechnology is still in its infancy, most mainstream scientists are not taking it as seriously as they should. Ware has designed a package independently from his work at Xionics, called NanoCAD which might help change scientists' perceptions. He confesses that at the moment NanoCAD is more of a hobby than a serious research tool: "I am hoping that over time, I will be able to develop NanoCAD into the kind of tool that would be useful to somebody doing real nanotechnology design." From the software standpoint, Ware sees the design cycle as involving three steps: define a structure, simulate the structure in action, and view the results. He points out that because of current computing limitations, simulation and viewing are probably only possible separately. With more powerful computing it might be possible to simulate on the fly and watch parts move in real-time. Defining the structure seems likely still to remain a separate step, however.
Ware's philosophy was to release the program and its source as freely as possible. "My reasoning is that we need to initiate wide public discussion of nanotechnology as soon as possible, and that NanoCAD could be a useful tool in that discussion with educational value."
Bobby Sumpter, Donald W. Noid and Robert E. Tuzun at Oak Ridge National Laboratory in Tennessee are attempting to explain some fundamental features of nanodevices from the chemical physics perspective. Using simulation techniques, they have found that it is the geometry of nanostructures - rather than purely chemical properties - that controls mechanical behaviour. Turning nanomodels into working reality might not be as limited by chemistry as it would seem.
The team is incorporating its simulation expertise into a new nanotechnology package in collaboration with the designer of another cad package y; by Geoff Leach of the Department of Computer Science, RMIT, Melbourne Australia).
The team points to rapidly developing technologies they believe will be among the first to benefit from nanotechnology, such as hydrogen storage for electric vehicle batteries and sensors relying on understanding and controlling fluid dynamics at progressively smaller sizes.
Assuming the likes of NanoCAD and Crystal Sketchpad evolve successfully they might allow researchers to do more nanotechnology modelling relatively easily. This, Ware maintains, is the real prerequisite for almost any path to working nanotechnology. He points out that such software could also have the secondary effect of familiarizing researchers other than theoretical chemists with the maths and algorithms involved in molecular modelling, and perhaps imparting to them an intuitive feel for how molecules work.
Al Globus of the NASA Ames Research Center in Moffett Field, California, is using various packages to implement molecular modelling of nanotechnology. He is using software such as Roger Sayle's public domain RasMol and Xmol, both well-known visualization packages for rendering relatively small molecules into their 3-D shape, together with proprietary molecular mechanics programs such as MSI's Cerius 2, for modelling. He and Jie Han also at NASA have also created a package called Nanodesign.
Nanodesign is designed specifically to handle fullerene and models of carbon nanotubes - buckyballs and buckytubes. Globus believes fullerenes are just right for making gears and other nanocomponents in the relatively near term. He and his team are ascertaining how teeth might be added to create cogs, wheels and gears onscreen and have even modelled a pump!
NASA established its computational nanotechnology program last year with the aim of simulating a hypothetical programmable molecular machine that could replicate itself and build other products - Drexler's putative assembler. Globus is cautious of making predictions about when nanotechnology will happen, "I have no idea what the first devices will be - except that they will probably be computer components," he says. "What gets done in our lifetime depends on the amount of effort expended, which depends on public support for government work and perceived profitability for commercial work," he adds.
While he is reserved about prophesizing, he is confident of his designs. "The unique properties of fullerenes, in dimension and topology, allow one to design various nanodevices and molecular machinery parts, such as a carbon nanotube-based gear," he explains. Shafts take the form of multi-walled carbon nanotubes and gear teeth are benzyne molecules bonded onto the nanotube. Globus says that his team's "extensive quantum chemical calculations and molecular simulations" support the feasibility of chemically synthesizing these devices, and he is confident of the development of nanotechnology. "If our group does really well we might see useful products in twenty years. The timescale is very sensitive to the level of effort, which is quite low at the moment - but growing rapidly."
Although most of the work in this field is being performed in the usa, there is some activity in Europe. Philippe Van Nedervelde is busy starting up a nanotech company with the working name Eutactix. "I have recently landed a contract with NASA Ames Research Center (Al Globus to be specific) for producing NanoDesign as a basic VRML-based on-line nanosystems cad-cam package. This is the beginning of a promising collaboration with NASA, but there are no results yet." Nedervelde is also completing his grooming as the spokesperson and representative of the Foresight Institute (Drexler's setup) in Europe.
From the computer blueprint components it is not a huge leap to imagine a molecular version of Babbage's Difference Engine. Super powerful and high-speed laptop Babbage machines might be just the tools needed to model the assembler-like structures containing billions and billions of atoms to bring nanotechnology out of science fiction.
For those who still doubt nanotechnology, consider. It is only just over forty years since the discovery of DNA yet we can manipulate and express genes and control the growth and development of organisms. Nanotechnology is based only on molecular-scale systems - like those that evolved in life, but under our control.
The models will only ever be useful if the ability to create them for real is achieved, however. "Science works primarily on what can be tested with existing experimental technique," says Globus. "As our laboratory techniques get better, mainstream science is getting more interested."
Damer points to the fact that the supramolecular chemists striving to understand and emulate nature may be taking the right tack. "I think the focus of Drexler, Merkle et al is concentrated too much on the building of structures. Nature is less about cell walls than the complex dance of molecules passing through and over those walls." Ware echoes this notion. "Viruses piggy-back on the replicative capabilities of cells; perhaps we can do the same. Nanotechnology may start out looking like present-day genetic engineering that evolves in a direction of increasingly flexible designs," he says. "We now make bacteria that produce insulin; we might one day make bacteria that produce gears or molecular logic gates or memory cells."
Come the curtain call, it will be a fusion of computing and chemistry that brings real drama to the virtual stage.
Previously in Elemental Discoveries:
Green silicon production
P2P for scientists
Women in science
Academic poaching of researchers
Permanent implantable contact lenses
Profile of ETH Zurich