NRI, which is expected to begin operations as early as next year, will reach annual funding levels of $100 million or more during the implementation phase, with engineers from industry working at several university-based centers.

The institute will create "a road map for addressing challenges, focus nano research and eliminate redundancies," John E. Kelly III, the senior vice president in charge of IBM Corp.'s microelectronics operations, told the SIA board of directors at a meeting earlier this month.

"Its mission, simply stated, would be to see to it that by the year 2020 — only 15 or so years from now — U.S. companies are the leaders in design, development and manufacturing of a new genre of nanoscale devices."

"The SIA members realize it will take 12 to 15 years before we can take these inventions to market," said Paolo Gargini, director of technology at Intel Corp., who has been among those leading the effort to form the NRI. "We need to kick off in 2005 if we want to have something ready to use by 2020."

Preparations for the NRI began several years ago when the SIA asked Jim Hutchby, director of device science at the Semiconductor Research Corp. (SRC; Research Triangle Park, N.C.), to lead a committee evaluating the various nanoelectronics devices.

The nine-member SRC committee studied seven types of emerging research logic devices: rapid single-flux quantum devices; drift electron transport devices based on nanotubes and nanowires; resonant-tunneling devices; single-electron transistors; molecular devices; quantum cellular automata; and spin transistors.

Much of the results of the SRC committee's study were published in the 2003 edition of the International Technology Roadmap for Semiconductors, in a 50-page chapter on emerging research devices.

For several of the logic types studied, Hutchby said density and power consumption metrics become less attractive when clock circuitry is added. At this point, CMOS "still appears to be the most optimal solution" for logic functions.

The NRI, IBM's Kelly said, would "get the United States ready for 2020" by drawing on the best talent in industry, government and academia. "It would bring science from the invention phase into a technology proof-of-concept phase through focused research."

Kelly said the NRI would be governed by a board of directors, including SIA member companies, universities and government agencies. It would have dedicated researchers, faculty, students and assignees.

The federal government's National Nanotechnology Initiative (NNI) now devotes perhaps one-fifth of its $900 million 2004 budget to semiconductor and electronics-related research, with a much larger percentage going to bio and medical-related research.

The SIA nanoelectronics study group had one meeting with an NNI delegation in October, and will hold another meeting in mid-December, to ensure that the NRI and the NNI coordinate activities to avoid duplication.

"This is one of the first times industry and government are running almost parallel paths," said Intel's Gargini, who serves as chairman of the ITRS and sits on the technology advisory boards of the SIA, the SRC and International Sematech, among others. "We have to make sure that what the NRI does is not in competition with what government is already funding."

The first phase of the NRI nanotechnology research initiative will be funded at a rate of $3 million to $5 million per year. Most of the work will be done at dozens of universities, searching for the device types that could complement or eventually supplant CMOS. The second phase, Gargini said, will aim at developing proof-of-concept devices, with the work taking place at perhaps two or three universities in different regions, each able to fab prototypes that could be integrated into ring oscillators and other test devices.

The proof-of-concept phase will be funded at roughly $30 million per year, with the final implementation phase supported with $100 million or more of funding, from both industry and government.

In the second phase, assignees from the U.S. semiconductor companies will go to the universities, in part to make sure that the NRI research projects are headed toward practical ends. "At times," Gargini said, "universities like very complicated approaches. The skill of industry is to come up with something low-cost and effective. We don't want the assignees to manage, but we want them to be part of the game."

Surprisingly, Gargini said carbon nanotube devices are not part of NRI's initial purview. CNTs may come in much earlier, incorporated into CMOS-like devices that have a source, drain and gate, and which depend on electron conduction. At the NRI, "we are looking beyond carbon nanotubes, beyond electron conduction," Gargini said. "With CNTs, charge is still flowing. We have a hunch that what we are looking for may not involve current flowing."