The advent of molecular assemblers will eventually make possible the construction of virtually any specified arrangement of atoms1. Present knowledge of chemistry and molecular mechanics forms a sufficient foundation for understanding the basic capabilities of assemblers and of some of their products. Among the consequences of assembler technology will be dramatic advances in computation, materials, and the production of aerospace hardware.
But will build tomorrow more than we can build today. New tools will
let us build better systems, including yet better tools. And it is not
futile to try to understand what may be built with tomorrow’s tools: the
laws of nature do not change as technology does, and many laws seem well
understood. With today’s science and a willingness to think about tomorrow’s
tools, it may be that we can see the shape of tomorrow’s engineering. A
key development will be the molecular assembler.
The idea of molecular machinery was latent in the earliest descriptions of molecules as mechanical systems. The kinetic theory of gases marked an early milestone; the recognition that biochemical processes rest on molecular machines marked another. Discussion of molecular assemblers and a technology of molecular machines, however, is apparently quite recent1.
Richard Feynman’s 1959 talk on miniaturization2 foreshadowed the concept of the assembler, though it lacked any concrete discussion of chemical bonding or molecular machinery. Feynman observed that "The principles of physics, as far as I can see, do not speak against the possibility of maneuvering things atom by atom," and went on to remark: "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." Thus, while Feynman did not describe the assembler, he was clearly looking in a similar direction.
The case for the assembler can be made most briefly through consideration
of existing biochemical machinery. Enzymes, ribosomes, and the genetic
system demonstrate that molecular machines can exist, that they can position
and manipulate reactive molecules, and that they can guide the breakage
and formation of chemical bonds in specific ways under programmable control
(see Table 1). These biochemical machines show lower
bounds to the capabilities of general molecular machines; they also indicate
that protein engineering3,4
offers a path to the development of novel molecular machinery.
The greatest limit to modern synthetic chemistry is its limited control of reaction sites. The use of even primitive assemblers to position reactive molecules will give far more flexible control of where chemical reactions occur on large molecules: this will permit the creation of molecular structures far more complex than any that synthetic chemists can now construct. Today chemists mix molecules in solution, letting them tumble randomly and react wherever they can; achieving site specificity becomes increasingly difficult as the size of the molecules–and hence the number of possible reaction sites–grows large. Assemblers will place chemists in a fundamentally different situation. An assembler will resemble an industrial robot arm, able to move a part to a specific position with respect to a workpiece. This positioning ability will let chemists make even the most indiscriminately reactive molecules bond specifically, to build up specific molecular structures.
Among these structures will be components for molecular machines. Thus, however the first assembler may be built, we can expect to see an iterative process in which assemblers are used to build better assemblers. The use of strong, rigid materials in assemblers will aid the application of mechanical energy to chemical reactions; components with suitable electrical properties will do likewise for the use of electrical energy. Together, these advances will make possible the synthesis of an extremely broad class of structures1, which may be expected to include virtually all molecular structures of engineering interest (some limits are discussed in ref. 5).The protein-engineering path to assemblers is already being followed3. Other paths seem possible: these include exploitation of non-protein supramolecular chemistry and the development of manipulators based on the positioning technology used in scanning tunneling microscopes6. For understanding the capabilities of future technology, however, the path to be followed is unimportant:
All paths lead ultimately to assemblers able to build systems to complex,
atomic specifications, including systems built from independently assembled
parts. It is thus of interest to consider what sorts of systems can be
built, given the assumption that essentially any well-bonded molecular
structure can be synthesized, and that these structures can be combined
to build complex systems.
Molecules do, of course, have unique properties from an engineering perspective. Atoms are like prefabricated parts, but atoms of a given type are utterly identical and not subject to wear. Machines made by bonding together these parts are utterly identical, if they are built with the same pattern of atoms and bonds. Accordingly, they show no wear or fatigue, no subtle changes over time: they are either whole, or broken.
Molecular machines suffer from unusual problems and failure mechanisms. Thermal noise has significant effects, making parts shake to an extent that varies inversely with their stiffness. Slim parts many atoms long suffer vibrations that become significant on an atomic scale. Radiation damage is a major concern: it causes bond breakage and device failure at a rate that gives hundred-million atom systems a half-life of a few centuries under ambient terrestrial conditions8. This half-life for damage to occur is inversely proportional to system volume. Accordingly, large systems of molecular machinery will require redundant, fault-tolerant design.
The scaling laws for mechanical systems are such that the frequencies associated with moving parts are inversely proportional to their linear dimensions, all else being equal. Scaling industrial robot arms down to the 0.1 micron scale of an assembler arm gives a ten million-fold increase in operations per second. Mechanical computers built with parts of subnanometer scale will have clock frequencies in the gigahertz range9,10.
Biological analogies suggest two means of powering molecular machinery:
chemical energy and light, the energy sources of animal and plant cells.
Design work indicates that electric motors can be scaled down to a size
of about 50 nanometers5, making
electrical power another alternative.
We are already familiar with self-replicating molecular machines: crabgrass
is an example. These systems show that molecular replicators can produce
products of incredible complexity in vast quantity and at negligible cost.
Once the formidable challenges of tooling and design have been surmounted,
replicating assemblers can do the same for products designed by engineers.
Large numbers of assemblers in coordination lend themselves to the construction of large structures made of nearly flawless materials. Properly arranged carbon atoms will make diamond fiber, which can serve as the basis for tough, atomically-tailored fiber composites. Diamond12 has a tensile strength of about 50 gigapascal, a modulus of about 1000 gigapascal, and a specific gravity of about 3.5. Again, once the formidable challenges of tooling and design have been surmounted, it seems that hardware based on these materials can become inexpensive–indeed, there seems no reason why such hardware should cost more than unmilled wood does today, since both will be carbon-rich products of self-replicating molecular machinery.
If these "large structures made of nearly flawless materials" are spacecraft, then space travel will become inexpensive. Strong materials will make possible high-performance vehicles, while simultaneously allowing broad margins of safety to increase reliability and reduce maintenance requirements. The energy cost of access to space is already low, and solar collectors (for example) built by replicating assemblers can make energy costs lower yet5. When reliable, low-maintenance, high-performance vehicles finally become readily available, the main costs of spaceflight will have disappeared.