By Jack Mason
Small Times Correspondent

LAKE GEORGE, N.Y., Sept. 12, 2002 -- Nanotechnology and nanomaterials will be critical to solving the semiconductor industry's formidable challenges, said John Kelly, IBM senior vice president and a keynote speaker at the Albany Symposium on Global Nanotechnology, which began Wednesday.

One sign of nanotech's rising role in the chip-making business was obvious in the event's new name. Last year's meeting was called the Albany Symposium on Global Semiconductor Issues.

IBM's Kelly keynote address focused on the great challenges the semiconductor industry will have to surmount to continue growing.

He noted that as chip components continue to shrink toward the nanoscale and silicon devices become more complex, creating the "masks" or templates used to etch them via photolithography will become prohibitively expensive.

Such masks currently cost upwards of a million dollars per set, Kelly said, and will become much more costly as the industry moves toward the shorter wavelengths and dimensions of extreme ultraviolet, or soft X-ray, lithography.

Kelly said IBM is looking toward a "maskless" future in which circuitry patterns as small as 10 nanometers are made not with light rays, but through molecular self-assembly. "We're relying on nanotechnology to create materials that will self-select to produce features at the critical dimensions our technology needs to move forward," he said.

While photolithography will not hit the wall in the near future, Kelly surveyed three techniques that IBM researchers were investigating to create nanoscale patterns on a wafer surface without photolithography.

One approach entailed embossing a nanotextured pattern on silicon that could serve as a kind of pegboard for attaching components. Another used a polymer layer to create a crystalline template on the chip surface. In a third approach, an array of 50-nanometer silicon pillars could serve as attachment points for 10 nanometer capacitors or other components.

Kelly also said he thought carbon nanotubes would find their way into semiconductor devices sooner rather than later. "When I was here last year, I said that carbon nanotubes were 10 to 20 years off," he said. "The progress the world has made in only a year has really surprised me."

He cited research at IBM and elsewhere that has advanced from producing single nanotube transistors to building working circuits in less than a year.

Nanotubes could also be essential for solving the problem of current leakage that traditional transistors encounter at the nanoscale, Kelly said. "Last year nanotubes were interesting, far-off research. Now I think we're going to see increased activity around nanotube devices because of their performance and density characteristics, but also because they get around the current leakage issues."

Even the copper wires that interconnect different parts of a chip are running into new challenges as they get smaller. "Surface roughness, grain structure and impurities can affect how well they conduct," he said. Insulating copper wires from silicon is also becoming a much larger challenge as chip dimensions shrink.

"As we move from 130 to 90 to 65 nanometer (architectures), we'll have to throw more and more nanomaterial capabilities at these issue if we want to solve these very difficult problems," Kelly said.