IBM Corp., Microelectronics DivisionArmonk, N.Y.

The way IBM sees it, we're all about to enter the copper age. For years, aluminum interconnects were standard fare on semiconductors. These tiny wires bridge the individual transistors to create a functioning processor. But, unfortunately, aluminum isn't a terrific conductor of electricity, and it imposes fundamental speed limits. As researchers began to push aluminum chip design to its logical limit, it became clear that additional breakthroughs were required to produce the next -- and more powerful -- generation of semiconductors. In September 1997 IBM announced that researchers had finally found a way to use copper to build semiconductors. The breakthrough means that chip designers can now construct far more powerful processors that take up less space and consume less power. Adding to the appeal, copper costs less than aluminum, making it a perfect match for an array of price-sensitive products, including PCs. IBM began shipping the copper chips in September. "It is another step in the continuing miniaturization of the semiconductor," states Bijan Davari, vice president of Advanced Logic Technology in IBM's Microelectronics Division. "Without an additional scaling down of CMOS [complementary metal-oxide semiconductor] technology, the growth and capabilities of computers would slow. CMOS has been the driving force for huge gains in computing power over the last 20 years," he points out. To be sure, it has become more difficult to build the narrower wires needed to boost performance -- the conventional method for producing gains. "This is a way to overcome that problem by using other technology to boost performance." Developing and perfecting the copper interconnects was no simple task. IBM first set out to incorporate copper into chips more than 10 years ago. Since then, several semiconductor manufacturers have begun to use copper in a few specific components -- though nobody had been able to use the material as a general substitute for aluminum. IBM's research finally paid off a couple of years ago when scientists found a way to solve a nagging problem: how to isolate the copper and keep it away from the silicon located only a micron or so away. By enclosing all copper in a protective sheath, which keeps the material away from the silicon structure and prevents oxidation, the chip becomes far more efficient with no risk of failure. "If the copper comes into contact with the silicon the entire chip will be destroyed," Davari explains. "It's essential to have physical barriers that can control high temperature, humidity, and voltage." But that was a complicated task. "When we began the project a decade ago, it wasn't at all obvious that such a design was possible," he states. Already, the copper technology is making inroads in IBM products. It's being used in IBM's Power PC 740/750, and will appear in all future versions of the Power PC. It's also being added to Big Blue's line of mainframe and midrange computers, including the S/390, RS/6000 and AS/400. The initial version of the technology was designed to operate at 400 MHz, though the copper can work with any future processor, regardless of the speed. But the impact doesn't stop there. Davari says that the technology is completely transparent across technology platforms and devices. Indeed, other semiconductor companies -- including Texas Instruments, Advanced Micro Devices, and Motorola -- have efforts underway to design and build systems based on the copper technology. Apple Computer has acknowledged that it will begin using 400-MHz copper PowerPC chips in the Mac early in 1999. Davari believes that copper chips, which typically provide a performance boost of 15% to 40% over comparable aluminum chips, could become nearly ubiquitous over the next few years. While IBM has protected much of the technology patents, the fact that IBM has made much of the design and engineering information available to others in the semiconductor industry will ensure that the technology enjoys widespread use. "Copper helps eliminate yet another fundamental bottleneck. It's only a matter of time before it's a standard technology," says Davari.