Features

This kind of work is also having an impact on computer hardware. IBM's Millipede project aims to produce a practical storage device that recreates in miniature the old punched cards of early computing. It uses arrays of tiny fingers to make minuscule depressions in a very thin film of polymer overlaid onto silicon. At the tip of each finger is a sharp point just a few nanometres across. To record a bit, it's pressed into the polymer surface and heated to about 400C in order to melt the plastic and create a lasting pit.

To read what's stored, the finger is again pressed against the polymer, but this time heated only to 300C. At this temperature, it can't melt the plastic, but its rate of heat loss can be measured. Fitting snugly into a depression, it loses heat more quickly than if it's touching a smooth surface with just its tip. To erase a recorded bit, the tip makes a rapid series of closely overlapping indentations around the existing one to flow the plastic evenly into the old depression. Over 100,000 write/erase cycles have been achieved like this.

Prototype Millipedes can already store data at a density of one trillion bits per square inch, 20 times more than any magnetic storage device can manage. The speed is impressive too - although the read/write rate of each individual nanotip is slow, thousands can be fitted onto a tiny chip, making up in parallelism for what they lack in intrinsic speed. Although still an experimental device, it's hoped that a

ADVERTISEMENT

commercial version might become available by 2005.

Devices like the Millipede and the molecular cascade are, of course, in no way part of the self-replicating grey-goo problem, but the underlying development work on MEMS and molecular computing is very relevant to nanorobots that might one day be able to build their own offspring. Combine a simple movable nanofinger with an on-board molecular computer that can communicate with other fingers around it, and you have the foundations for an invisible assembly line - one that could potentially replicate endless copies of itself.

Down the tubes

Eric Drexler, inventor of the word 'nanotechnology', is very much a supporter of its development. Even so, when he wrote his seminal book Engines of Creation, he included a chapter titled 'Engines of Destruction' and noted that, 'Replicators can be more potent than nuclear weapons: to devastate Earth with bombs would require masses of exotic hardware and rare isotopes, but to destroy all life with replicators would require only a single speck made of ordinary elements.'

Although grey goo is the most dramatic threat to emerge from nanotechnology research, there are many others that may be far more likely and immediate in their impact. One of the more worrying questions raised in the University of Toronto report asks if nanomaterials might be the next asbestos.

Nanotechnology is creating all sorts of exotic new materials, such as carbon nanotubes and buckyballs, many of which aren't found naturally on earth except in tiny quantities. What impact they may have on health or the environment is little understood and hasn't received much attention so far.

Mostly, these materials consist of huge molecules made up of large numbers of carbon atoms. In the case of buckyballs, they're shaped like hollow spheres or rugby balls. In nanotubes, the carbon atoms form hollow fibre-like molecules. Collectively, all these large carbon structures are known as fullerenes and they offer all sorts of useful electrical, chemical and mechanical properties.