Pilgrim Beart, Cambridge 2003
Nanotechnology is technology capable of sensing and acting at the scale of individual molecules and has many exciting potential applications. At this scale, an important concept is self-assembly. Molecules floating in solution exhibit "selective binding". As they bump around randomly, the affinity of their various atoms causes one molecule to bond to another only in specific combinations and relative orientations, and thus to "self-assemble" into more complex structures. Millions of such bonds can occur independently every second, making self-assembly a highly parallel bottom-up process.
Nature's guides this process using DNA - a uniquely stable and "programmable" material - and we can now easily synthesise arbitrary DNA sequences. In biology DNA is part of a fantastically complex system which we are only beginning to understand, but in the bare environment of the lab we can constrain DNA to act only in simple ways that we do understand. We can thus move from the realm of complex biological science to the realm of straightforward pragmatic engineering - the realm of the Industrial Revolution and of Moore's Law.
There is an analogy with how we make integrated circuits. Nature allows many semiconductors connected in almost infinite variety. However to keep things simple we have chosen to constrain our choice to just one semiconductor (doped silicon), in only two dimensions (as a wafer), patterned with just a few basic kinds of structure (transistors), connected into a just few regular blocks (e.g. NAND gates) and generally all synchronised by a single global clock. This intentional narrowing of the possibilities has allowed the development of a very powerful tool-chain which automates the design process. A human team writes a high-level specification which is then translated, largely automatically, into a circuit of tens of millions of transistors - far more than could ever be patterned by hand - and then automatically fabricated by the millions.
Recent academic work has shown that it is possible to write DNA sequences that self-assemble into square two-dimensional "DNA tiles". These tiles are simple engineerable unit structures, analogous perhaps to the transistors of today's integrated circuits. The edges of these tiles then themselves exhibit selective binding affinity to particular edges of other tiles, causing the tiles to self-assemble into larger "mesoscale" structures. These structures have the potential to act as the scaffolding for nanotechnological applications, for example to pattern the surface of hard disk drives, or to manufacture integrated circuits 100x smaller than is possible with today's lithographic techniques. Many of the scarcely-understood complexities of biology - such as protein-folding - are simply ignored because they never happen in this simplified world.
Today nanotechnology is at the stage of Shockley's first transistor - hand-engineered lab gadgets - but imagine the future. An engineer writes a high-level description of a desired structure. Software works out how this structure can be made to self-assemble and then directs the sequencing of the low-level DNA tiles, which then assemble into the desired structure. It's just like writing a software program, only the result is physical instead of virtual - we will have a physical compiler. DNA tiles aren't the only way forward but they are the first step down a very exciting path.