Transistor Could Spark Current of Change

Scientists are fine-tuning a new transistor that cranks out computations 10 times faster than existing computer technology.

The transistor, under development by federal scientists at Sandia National Laboratories, could benefit everything from computers and cell phones to satellites and toxic-materials sensors.

“If you can integrate this with conventional silicon processing, it would mean cheaper, faster, smaller, better,” said Paul R. Berger, a University of Delaware associate professor who has reviewed the federal team’s work.

Millions of transistors are hooked together with wires on the surface of a single microchip. The Sandia transistors are about the size of a traditional one--a square about 1/15th the width of a hair--but only half as many are needed.

“It has a chance to revolutionize electronics,” said Jerry Simmons, leader of the five-person team that spent about four years and $1 million developing the transistor.

In a common silicon transistor, a gate opens, sending electrons flowing down a duct. In the Sandia transistors, electrons “tunnel” from one semiconductor layer through an extremely thin barrier to another semiconductor layer. The rapid process allows electrons to avoid impurities along the path that slow traditional transistors.

The Sandia transistor is expected to perform 1 trillion computations per second.

In a few years, another federal laboratory in northern New Mexico expects to have the world’s fastest supercomputer, capable of performing 3 trillion to 5 trillion calculations a second--using traditional transistors.

That supercomputer already is projected to be about 30 times faster than the fastest supercomputer previously in use at Los Alamos National Laboratory--and 100 million times faster than the typical home or office computer.

If the Sandia transistor lives up to its promise, it could eventually allow supercomputers to operate 10 times faster than those with traditional transistors, Simmons said.

Scientists have worked with “tunneling” transistors since the late 1980s but have not been able to consistently mass-produce them. Sandia fixed the problem by stacking the semiconductors vertically instead of horizontally.

The new transistors, which apply quantum physics, are still five to 10 years from commercial use, partly because microchips must be redesigned to utilize them.

Two U.S. companies, Raytheon Co. and Motorola Corp., and half a dozen Japanese companies also are competing to develop transistors using quantum principles.

“I think we’ll see chips made that include quantum effect. Whether this transistor will be the winner remains to be seen,” says Alan Seabaugh, a scientist at Raytheon.

The success of the new-generation transistors depends on how well they will work when linked together in multiples and how reliable they will be when mass-produced, Seabaugh said.

The Sandia scientists still have work to do. The device works best at minus 321 degrees Fahrenheit. But Simmons expects Sandia transistors to work at room temperature by next year.