IBM's announcement that it has developed a new, copper-based computer chip technology lit up headlines on CNN and in the major American daily newspapers Monday. To be sure, the company's technology, CMOS 7S, will eventually lead to smaller, faster, and cheaper chips.
But it won't give IBM much of a market-making edge, for leading chipmakers like Intel and AMD are also working on copper technology R&D, and expect to substitute more efficient copper for aluminum as a substrate on many of their chips.
The primary research for the IBM copper developments originated with the industry research consortium, Sematech, in 1988. The Semiconductor Research Consortium, another industry venture, also participated in early research.
"Research really got off the ground in 1989 when we started to get the funding," said Shyam Murarka, a researcher at the Rensselaer Polytechnic Institute in Troy, New York, who has worked with Sematech and SRC on the copper problem. "All of the industry has supported the work. IBM has gone with it. But I would expect that AMD and others will follow within six months to one year."
Competitors, therefore, greeted the IBM news coolly. "The IBM technology is a good accomplishment," admitted Adam Grossberg, a spokesman for Intel. "But the industry has known for years that it would have to move to copper."
Though research continues, the industry buzz is that copper-based chips - which could cost 20 percent less than today's chips and increase performance by 30 percent - are still extremely difficult to manufacture. "You have to use a buffer between the copper and silicon. Moreover, there is not enough equipment to place copper-based chips into high-volume manufacturing," said Grossberg.
The industry transition to copper should take at least five or six years, when circuit sizes get down to .13 microns, from their present size of .25 microns.
"In the next few years, anyone in the silicon volume-manufacturing industry, to be competitive, will have to incorporate copper in their business," said Scott Allen, a spokesman for Advanced Micro Devices. AMD, in fact, has had copper R&D as part of its technology development road map for quite some time, and has developed its own manufacturing process for copper chips.
Last month, moreover, Sematech announced that it had developed the first copper metalized wafers at its facility in Austin, Texas.
IBM's technology, CMOS 7S, or complementary metal oxide semiconductor manufacturing process, does, however, boast several technical accomplishments of its own. First, it integrates six layers of metal on a single chip. And CMOS 7S is the first technology to use a production copper metalization process.
Next, the technology lets circuitry on a chip be built to sizes of 0.20 microns, among the smallest in the industry. This is more than 500 times thinner than a human hair. CMOS 7S transistors also boast a 0.12 micron effective channel length, the shortest publicly reported in the industry. A channel length is the distance that current traverses through a transistor in semiconductor circuits; a shorter channel length results in stronger performance. The technology will also allow for up to 12 million gates on a single chip - between 150 million and 200 million transistors.
And lastly, transistors based on CMOS 7S technology operate at just 1.8 volts, creating an opportunity for low-power operation devices.
The early thinking is that devices powered by copper chips will be used in personal digital assistants, network computers, and even Internet-ready TVs. "You'll see faster computers on your desktop. You'll see better, faster access to the Internet. You will see handheld devices that are faster and cheaper, and the batteries in them will last longer," says John Kelly, vice president of technology for IBM.
One analyst, Sherman Whipple, a partner in the research firm of Whipple, Sargent & Associates, based in Hingham, Massachusetts, is even more imaginative: He believes that a copper-based 686-class chip could power a Web set-top box, filling a niche for consumers who don't desire a PC.
But others are more realistic. "I can't imagine these applications will appear soon," says Anne Powell, a London-based research analyst at Datapro, a division of the Gartner Group. "Don't expect one in a present under your Christmas tree this year."
At least one leading manufacturer doesn't think it will jump on the copper bandwagon right away. Melody Wolfe, a spokeswoman for TI, said that the Richardson, Texas-based company is spending money on copper R&D through an alliance with Sematech. But for the company's main chip product, digital signal processors, copper will likely remain "too expensive" as a replacement for aluminum, Wolfe said.
