A LEAP INTO THE MICROCOSM

The reason that many computer scientists are pursuing this goal is that the shrinking of Technology executives at the Intel Corp.,the worlds largest chip-maker,say they believe that by coupling more advanced photolithographic techniques with new kinds of materials and by changing the design of the transistor,it will be possible to continue to scale down to sizes as small as 5 nanometreseffectively taking the industry forward until the end of the next decade.

Silicon will probably continue longer than we expect, said Michael C. Mayberry,an Intel vice president and the director of the companys component research programme.

Both Intel and IBM are publicly committed to a new class of transistors known as FinFETs that may be used as early as the 22-nanometer technology generation beginning in 2011 or 2012. Named for a portion of the switch that resembles a fish fin,these transistors have the dual advantage of offering greater density because they are tipped vertically out of the plane of the silicon wafer,as well as better insulating properties,making it easier to control the switching from a 1 to a 0 state.

But sooner or later,new materials and new manufacturing processes will be necessary to keep making computer technology cheaper. In the long term,new switches might be based on magnetic,quantum or even nanomechanical switching principles. One possibility would be to use changes in the spin of an individual electron to represent a 1 or a 0.

In Ross laboratory at IBM,researchers are concentrating on more near-term technology. They are exploring the idea of constructing FinFET switches in a radical new process that breaks away from photo etching. It is a kind of nanofarming. Ross sprinkles gold particles as small as 10 nanometers in diameter on a substrate and then suffuses them in a silicon gas at a temperature of about 1,100 degrees Fahrenheit. This causes the particles to become supersaturated with silicon from the gas,which will then precipitate into a solid,forming a wire that grows vertically.

IBM is pressing aggressively to develop this technology,which could be available commercially by 2012,she said. At the same time she acknowledged that significant challenges remain in perfecting nanowire technology. The mushroom-shaped wires in her laboratory now look a little bit like bonsai trees. To offer the kind of switching performances chip-makers require,the researchers must learn to make them so that their surfaces are perfectly regular. Moreover,techniques must be developed to make them behave like semiconductors.

IBM is also exploring higher-risk ideas like DNA origami,a process developed by Paul W.K. Rothemund,a computer scientist at the California Institute of Technology. The technique involves creating arbitrary two- and three-dimensional shapes by controlling the folding of a long single strand of viral DNA with multiple smaller staple strands. It is possible to form everything from nanometer-scale triangles and squares to more elaborate shapes like smiley faces and a rough map of North America.

That could one day lead to an application in which such DNA shapes could be used to create a scaffolding just as wooden moulds are now used to create concrete structures. The DNA shapes,for example,could be used to more precisely locate the gold nanoparticles that would then be used to grow nanowires. The DNA would be used only to align the circuits and would be destroyed by the high temperatures used by the chip-making processes.

At Intel there is great interest in building FinFET switches but also in finding ways to integrate promising III-V materials on top of silicon as well as exploring materials like graphene and carbon nanotubes,from which the company has now made prototype switches as small as 1.5 nanometers in diameter,according to Mayberry. The new materials have properties like increased electron mobility that might make transistors that are smaller and faster than those that can be made with silicon. At that very small dimension you have the problem of how do you make the connection into the tube in the first place, he said.

Given all the challenges that each new chip-making technology faces,it is tempting to suggest that the smaller,faster,cheaper trend may indeed be on the brink of slowing if not halting. Then again,as Mayberry suggests,the industry has a way of surprising its sceptics.