What’s .03 microns long and can be turned on and off 10 billion times a second? It’s a new transistor that has the potential to keep Moore’s Law on the books for at least several more years.
Developed by Intel, the world’s smallest, fastest complementary metal oxide semiconductor transistor features internal structures that are as thin as three atomic layers?about 1/100,000th the thickness of a sheet of paper. “We looked to the biological world for our inspiration,” says Gerald Marcyk, director of the components research lab in Intel’s Hillsboro, Ore.-based technology and manufacturing group. “The transistor easily fits inside a string of DNA.”
Intel’s breakthrough comes just in time, since many scientists had been warning that Moore’s Law (the principle that says the maximum number of transistors on a chip will double every 18 to 24 months) was about to come to an abrupt halt. Electronic circuits can only get so tiny before they start overheating and behaving erratically, and many scientists believed that the practical limit to semiconductor miniaturization was in sight. But by using a group of undisclosed new materials, Intel was able to lower to a minimum the debilitating voltage leakage that normally occurs in increasingly miniscule circuits. To reduce the friction that generates heat in tiny circuits, Intel researchers dropped the transistor’s voltage down to .85 volts (Intel’s Pentium 4 runs at 1.7 volts).
The transistor represents a two-generation leap ahead of today’s available technology, says Marcyk. In about five years, the company plans to cram 400 million of the transistors onto a chip that will run at 10GHz. Since Intel’s current top-of-the-line Pentium 4 processor has 42 million transistors and runs at 1.5GHz, it would appear that Moore’s Law is more than safe for the foreseeable future.
Still, for all of its promise, Intel’s transistor marks only a temporary reprieve. It’s widely believed in scientific circles that circuits will get so small in about 10 years that designers won’t have enough atoms available to separate components. Beyond that point, a new technology will have to take over, perhaps photonic circuits or something not yet discovered. “We’ll need to assess our options at that time,” says Marcyk. “But for now we’ve bought ourselves some very valuable breathing room.”