Mind-boggling miniaturisation

Pratap Ravindran

WHEN a Bell Labs research team announced on November 8 in Science that it had made a transistor from a single molecule - or, perhaps more accurately, observed single molecules acting as transistors, the Indian media, which had gone ape over the cloning of Dolly, the human genome mapping project and so on, just didn't figure out the significance of the announcement.

Did the media goof? It most certainly did - because the break-through reported by the team at Bell Labs, a facility of Lucent Technologies, signified the dawn of an era of what could be called flexible electronics.

According to Hendrik Schon, the head of the Bell Labs team, ``you might think about flexible electronics (as) some things which silicon cannot do.''

Further, the possibility of making molecules work as transistors, suggests that Moore's Law will soon be history.

But, before we deal with what this development means, let's back up a little so as to gain perspective.

We're basically talking nanotechnology here. Albert Einstein had worked out the size of a single sugar molecule from experimental data on the diffusion of sugar in water. According to Einstein, each molecule measured about a nanometer in diameter - that is, a billionth of a meter. You can't imagine anything that small? Turn to Scientific American which describes a nanometer thus: ''The width of ten hydrogen atoms laid side by side, it is one-thousandth the length of a typical bacterium, one-millionth the size of a pinhead, one-billionth the length of Michael Jordan's well- muscled legs....''

So that was how the nanometer came about.

Years later, in 1959 (when an adding machine wouldn't fit into your pocket), physicist Richard Feynman, in the course of an after-dinner talk on the limits of miniaturisation, came out with a proposition that was considered bizarre - atom-by-atom construction.

Well, things have changed and the notion is accepted as anything but bizarre now. In fact, it's right there on the cutting edge....

Advanced nanotechnologies, driven by findings in molecular biology, are basically looking at bottom-up construction in which molecular machines assemble molecular building blocks to generate products, including new molecular machines.

This sounds very confusing, but it's not. Look at it this way: A machine can be defined as a device that performs a task. It has a design and is constructed using a particular process. It performs its function in accordance with the information built into it. And it uses power. Seen this way, you'll inevitably come to the conclusion that nanoscale machines already exist in the form of the functional molecular components of living cells - protein molecules, for instance. After all, the definition of a machine does not imply human design and purpose.

If you follow up on the cue given by biology, you'll find that it is entirely in order to think of placing every atom in a particular place as an active component with no unnecessary atoms rattling around the place, messing up the works. A machine or system built using this process will not be a liquid or gas as no molecules would move randomly. Nor would it be a solid in which molecules are fixed in place. Instead, this new ''machine-phase matter'' would be characterised by the molecular movement hitherto seen only in liquids and gases - as also by the mechanical strength associated with solids!

Getting back to nanotechnology and computing, IBM and Hewlett-Packard already have substantial nano research programmes in place because they realise that the days of silicon electronics and top-down manufacture are numbered.

However, it needs to be noted that nanotechnology is not going to be here tomorrow. Or even the day after. That is partly because it straddles the world of individual atoms and molecules in which quantum mechanics (which is about as weird as weird can get) rules and the macroworld. Further, since its earliest days, researchers in nanosciences have been obsessed with the need to ensure that the technologies do not cause accidents and that they are not abused. This intense pre-occupation with safety has caused scientists working in the field to pause and think about the implications of the September 11 terrorist attacks on the US. In specific, they are worried about the use of nanoweapons by terrorists.

Thus, smalltimes.com quotes Christine Peterson, President of Foresight Institute, as saying that biological nanoweapons represent a threat ''potentially tougher and smarter'' than the dissemination of anthrax spores.

In an article, Nanotech's Dark Side Debated in the Aftershock of September 11, Small Times correspondent, John Carroll states: ''For years now, nanotechnology has surged forward based on open source research methods. In this emerging field, free and open discussion is a prized feature.....''

''But every new step towards commercialisation, as nanotech makes the transition from the drawing board to reality, brings the fledgling industry closer to the day when many believe an inevitable round of government regulations will be needed to prevent abuses. And as nanotechnology begins to deliver new and improved weapons, secrecy will eventually cloak important research work.''

That's the bad news. The good news is that the computer industry is not sitting on its hands, waiting for nanotech to emerge from a shroud of secrecy.

By way of illustration, the semiconductor industry is slap-bang in the middle of a major switch to the mass-production of chips with copper, as against aluminium, to deal with the consequences of Moore's Law which states that the number of transistors on a given semiconductor doubles roughly every 18 months. This doubling is made possible by continually shrinking transistors. Smaller transistors mean that chip designers can cram more of them on to a chip, increasing the number of features that can be added on. Further, smaller transistors mean that signals travel faster as they have less distances to cross.

But life is rarely that simple. Cutting the distance between transistors also means that you end up with smaller wires that carry less current and are more resistant to electricity. Intel tried coping with this problem by tinkering around with the aspect ratio of its aluminium interconnects - but knew that a change in the material used was inevitable. Silver and gold were ruled out because of cost considerations, which left copper. With the introduction of copper in the production of chips, the industry will also have to change the process of etching which is basically the process by which circuits are laid down on wafers. It is now checking out what designers call the ''dual damocene'' process in which the circuit pattern is ''dug'' into the wafer and copper is electroplated across the surface.

The metal left in the dug-out areas after etching form the circuit pattern. The grapevine is that Intel will come out with its first copper Pentium 4s in the fourth quarter of the current year at 2.2 GHz. And that's not all. Even as it is making its transition to copper chips, Intel's Microprocessor Research Lab is working on the technology for using fibre-optic strands to transport data inside computers.

Chip makers are under constant pressure to shrink chips even while adding on more transistors to boost performance.

And therein lies the rub. An increase in the number of transistors means a hike in the amount of electricity required to run a chip. More electricity running through a shrunken chip results in more signal interference and reduced battery life in notebooks.

Fibre-optic connections will use laser beams - and not electrical impulses, to transmit signals. This means that they'll require less power. And will be faster.

Copper wire permits data transfer rates in the region of 10 gigabits per second (gbps) and Intel believes that, with a switch to optical connections, they'll be able to achieve a transfer rate of 20 gbps in the next few years.

Meanwhile, don't forget that, back at the ranch, Bell Labs scientists are working on molecular transistors. These are expected to be the ultimate in miniaturisation - but don't bet on it. Schon of Bell Labs says he is sceptical about atomic and sub-atomic quantum transistors but adds: ''Some people have ideas about making chains of atoms and then maybe moving the atoms to change the conductance.

But I don't see how you can amplify signals with that. I don't know - maybe some people will come up with clever ideas.''

Prophetic words, indeed. About a fortnight after the Bell Labs announcement, Israeli scientists told the press that they had built a nanocale computer of DNA molecules which could run a billion operations per second, with an accuracy of 99.8 per cent.....

(pratap@thehindu.co.in)