Computing with molecules
This article first appeared in Personal Computer World magazine, November 1999.
A team of researchers want to build a computer with the combined power of 100 high-end workstations. Not such a surprising ambition, you might think, given the rapidly increasing power of silicon technology -- but there's a twist. Their machine will be the size of a grain of sand. The scientists, from Hewlett-Packard and the University of California at Los Angeles, want to synthesise a computer chemically. The key to their work is a strange substance called rotaxane, a single molecule of which can be made to function as an on-off switch -- the basis of all logic operations.
Rotaxanes are a class of organic molecule complexes which comprise a molecular ring threaded onto a central linear molecule, like a bead on a wire. The ring can slide freely along the central molecule, and extra "blocking molecules" at each end stop it falling off. The team created a microscopically thin layer of rotaxanes, and sandwiched them between a pair of electrodes. Each electrode is etched with contact points, such that a rotaxane molecule forms a bridge between corresponding contact points on each electrode. Normally, electrons can travel across the rotaxane from one electrode to the other -- so the switch is closed. Applying a control votage, however, breaks the rotaxane's structure, opening the switch and preventing electrons from crossing.
The team went further, and connected groups of switches together, demonstrating the molecules could perform the logical operations on which all computing is based. The importance of this research is that for the first time computing elements have been created by chemistry, instead of conventional photolithography. And it's heady stuff. As Rice University researcher James Tour says: "a single molecular computer could conceivably have more transistors than all of the transistors in all of the computers in the world today".
But because the molecules are synthesised using conventional chemical reactions, there's a fundamental problem. Reactions are rarely 100% successful, so how can the computer possibly work if some of its molecules aren't well-formed? The team already has the answer to that one. A previous project created an experimental computer architecture called "teramac", which was built from hundreds of conventional silicon chips, some of which were known in advance to be faulty. The lattice of chips could be interconnected in a huge number of ways, using switching and routing mechanisms similar to those in a telephone exchange. Control software, runing on chips known to be good, locates buggy chips and ensures that they're avoided. Although riddled with duff chips, the teramac worked impressively well.
Another problem is how to reliably connect groups of rotaxane molecules together, because even the thinnest kinds of conventional wires are enormous at molecular scales. One promising possibility is to use "nanotubes" -- tubes just a few atoms in circumference, whose walls are made of linked carbon atoms. The technology already exists to make these "molecular wires", which can be grown to lengths a million times their diameters.
The field of molecular electronics -- moletronics -- is growing fast, and while researchers are keeping their feet on the ground for now, the ideas are flowing thick and fast: computers as intelligent molecular gels; processors small enough for Fantastic Voyage-style exploration and mapping of living tissue; computers invisibly woven into the fibres of your clothes. We'll have molecular data storage that never runs out of space, and seamless integration of machine and life at the molecular level.
As Hewlett-Packard researcher Phil Kuekes put it, "eventually computers are going to be so small, you won't be aware of them". Until they crash, perhaps.
Toby Howard teaches at the University of Manchester.