t's not clear what that new microprocessor technology will be, but some candidates are starting to emerge.
"We're at the stage now where we're sorting out the options," George Scalise, president of the Semiconductor Industry Association, said during a panel discussion Friday that gathered representatives of some of the industry's largest companies with their partners in the academic world.
It generally takes the computer industry about 10 years to move a technology from the laboratory to commercial applications, so the successor to current silicon microprocessors probably already has been shown to work - somewhere - on a small scale.
"We have all of these alternatives," said John Kelly, a senior vice president at IBM Corp.'s research division. "If we only had one, I'd be extremely worried."
One promising avenue is "spintronics," which looks at the "spin" status of electrons rather than their charge. The research is not quite as exotic as it sounds: Spin is related to magnetism and already is exploited by some memory chips.
Logic gates - the basic building blocks of microprocessors - that use spin technology can be made out of the same materials as current chips but consume very little power.
Increased power consumption - and the waste heat it generates - is a showstopping problem for current technology, said Professor Alain Kaloyeros of the College of Nanoscale Science and Engineering. The latest chips produce more heat per unit of area than old-fashioned steam irons do, he said.
"Given that we're not in the business of making steam irons, and we don't want to use our laptops to iron our pants," it is imperative to find an alternative, Kaloyeros said.
One possible logic gate using nanotechnology - similar to current technology in that it uses electric charge - comprises a single metal atom suspended between two carbon-based molecules. Moving the atom one way or the other either blocks or lets through an electric charge.
Another application of nanotechnology, the manipulation of matter on an extremely small scale, is the carbon nanotube, essentially a sheet of carbon atoms rolled up into a tube. Researchers have been able for a few years to make logic gates out of nanotubes in different configurations.
It's not clear, however, whether it's possible to cluster enough nanotube logic gates to make a processor. The technology needed to accurately arrange millions of them would bear little resemblance to the semiconductor manufacturing of today, and it does not exist.
Other technologies that have shown promise - but which Friday's panelists thought would not be ready to take over from silicon within 10 years - include quantum computing, which seeks to take advantage of the seemingly bizarre way electrons interact with one another at the very smallest scales of matter.
"We do not think quantum computing is doable in our lifetime," Kelly said.
Another technology unlikely to be commercialized in the near future is photonic processing, where the computation is performed by particles of light rather than electrons. However, light is likely to become a dominant way to send information between chips, Kelly said.
Whatever new technology comes to dominate, it's likely that it will stand the semiconductor industry on its head. Companies like RCA that in 1950 were making vacuum tubes, the first electronic logic gate, did not evolve to become the ones making microprocessors in 1970. Intel Corp., Texas Instruments Inc. and others appeared to fill that need.
Paolo Gargini, director of technology strategy at Intel, noted that it's quite possible that whatever new technology is adopted will be easier to manufacture without the huge, multibillion dollar plants his company needs.
IBM earlier this year announced that its chip plants were going to start using a "self-assembling" process, in which material forms a nanoscale structure by itself, rather than being shaped with expensive lithography.
"If indeed these new options turn out to be easier ... maybe we'll have another type of industry," Gargini said.
Source Physorg
Share this Post with other
No comments:
Post a Comment