IBM carbon transistor is latest step in nanotechnology race
Monday, May 20th 2002, 12:00 am
News On 6
NEW YORK (AP) _ In another step forward in atom-level ``nanoscale'' electronics research, IBM says it has developed ultra-fast transistors from atoms of carbon.
IBM was announcing Monday that researchers assembled the transistors from carbon nanotubes, long, tube-shaped carbon molecules that have been identified as one of a few promising substitutes for silicon.
Silicon, the main in ingredient in the semiconductors on which the computer industry depends, is expected to reach the end of its usefulness in a decade or so.
Phaedon Avouris, IBM's manager of nanoscale science and one of the researchers who developed the transistor, said the announcement was significant because the carbon transistors were created using existing techniques used for silicon-based transistors _ but the nanotube devices allowed higher-speed electron flow than current silicon transistors.
``I think we need another three years of research to see how much better we can do and to see if it's worth starting a development process,'' Avouris said. ``Right now we're testing the waters. Unless we can make the nanotube devices better than silicon, it's not worth it.''
To gain continual jumps in processing speed, silicon transistors and other electronic circuitry have already shrunk to the molecular level.
But the molecular structure of silicon, whose semiconducting properties lend transistors their switching ability, prevents it from being shaved much thinner.
Carbon nanotube molecules are but one of the materials identified as a promising silicon substitute.
At Harvard, researchers have already assembled nanoelectronic devices using ``silicon nanowires'' made of germanium, silicon, gallium arsenide or gallium nitride, said Stan Williams, Hewlett-Packard's director of quantum science research.
Williams said HP has embarked on a third path to the same goal, which bypasses semiconductors entirely, fashioning circuits of molecular switches and nano-scale metallic wires.
``Our approach allows you to get much smaller than any semiconductor can get,'' Williams said.
Williams said nanoscale components ought to begin showing up in electronic devices, such as electronic memory modules, in three to five years.
``The first use we'll see is in sensors, then in memory,'' Williams said. ``Then we'll see integrated packages with sensing and communications capabilities.''
For instance, tiny sensors could be mixed with epoxy and painted onto the surface of an airplane or bridge, to alert maintenance crews of structural stress, Williams said.
``Nanoelectronics is going to change the way we do everything, but it's going to take 10 years,'' Williams said. ``We're in the early days now.''
Avouris said IBM is more focused on proving that carbon nanotubes are a viable concept, rather than integrating them into electronic devices.
He said IBM hopes it can convert its existing silicon chip making facilities _ including the yet-unopened $2.5 billion plant in East Fishkill, N.Y. _ to carbon nanotube construction, if it proves viable.
Commercial viability is one of the biggest hurdles, said Roger Kay, a technology analyst with IDC. ``They might get this physically working in a decade or so, but it'll be another decade after that before it's cost-effective.''
South Korea's Samsung Electronics has already integrated nanotubes into functioning products _ using them as conducting filaments, rather than semiconductors. Samsung has demonstrated a flat-panel monitor that uses carbon nanotubes as light filaments.
Motorola and Lucent Technology are also researching nanoscale electronics, as are myriad university scientists, Williams said.