By Marlene Cimons, National Science Foundation
The transistor has been the fundamental building block of electronics since the 1940s. It is, essentially, a switch activated by as little as one volt of electricity that powers virtually all electronic systems, especially today’s popular modern digital information technology devices.
Yet, transistors use vastly more voltage than necessary, wasting an extraordinary amount of energy. This trend is pushing technology to its limits, and will almost certainly prompt a slowdown in the pace of information services that we have come to expect--and demand.
It’s a real dilemma, since transistors don’t work well at less than one volt, which is almost a thousand times more than most systems really need.
“Information technology is a very big factor in modern civilization, and has become more and more a part of our lives,” says Eli Yablonovitch, an engineering professor in the department of electrical engineering and computer sciences at the University of California at Berkeley. “It continues to use an increasing amount of energy, and we’ve become spoiled. We want more downloads from servers, that do more information processing, and it’s using more energy than it used to in the past.
“Every time people go on a social network, the server is chugging away to give them what they want, and using a lot of energy,” he continues. “We’re using more and more all the time, and we’re coming up against a real functional limit. The improvements that we expect may not emerge.”
Unless, of course, scientists find a way to develop a more efficient switch that can replace the transistor. It would have to function as well or better than the transistor, but at a fraction of the voltage. This is what researchers are trying to do at the UC Berkeley-based Center for Energy Efficient Electronics Science, which Yablonovitch directs.
To be sure, it is a formidable challenge. But if researchers are successful, the payoff in energy savings could be enormous. Yablonovitch explains: “The electricity that is used is proportional to the voltage squared, so if you reduce the voltage by 1,000, you actually reduce the energy consumed by 1000-squared - or a million. We are aiming to reduce power consumption by one million.”
The Center for Energy Efficient Electronics Science, also known as E3S, is a National Science Foundation science and technology center with partners at the Massachusetts Institute of Technology, Stanford University, Contra Costa College, Los Angeles Trade Technical College and Tuskegee University. As part of the project, the center offers science and engineering programs to community colleges and high schools which serve minorities, giving students “real research experience to motivate them to excel in science and engineering,” Yablonovitch says.
NSF supports the program with $5 million annually over five years.
Center scientists are studying four types of designs that someday could make today’s transistors obsolete. These include a new kind of semi conductor that resembles a transistor, but can be activated with less voltage; a nano-mechanical switch, which works like a light switch but is very tiny, and could be activated with fewer than 10 millivolts; optical signaling with tiny light pulses, and; a magnetic switch that can be activated by minimal current.
Center scientists already have experimental versions for all four approaches, but none works well enough--yet.
“We’re making progress, but the challenge is very difficult,” Yablonovitch says. “Many scientists worry that it’s not even possible to lower the voltage, so we have a very big hill to climb. We believe it is possible. We have shown it in theory--there doesn’t seem to be any theoretical obstacle--but actual implementation will be very hard. First, we have to demonstrate it in practice, then as a manufacturable technology.”
Several large electronics companies are collaborating with the center. “The big electronics companies are behind us, and want to see this happen,” he says. “They are very interested in reducing the power consumption in information technology.”
In related research, center scientists recently determined that the dimensional shape of the so-called “P-N junction,” which combines two regions of a semiconductor, can take any of nine different distinctive geometries, but the best-performing one is two-dimensional P meeting two-dimensional N, “layer to layer, thin and flat and facing each other,” Yablonovitch says.
This is important for energy efficiency because the best-performing P-N junction is not the one currently used in semi-conductors, he says. The different geometrical combinations “all behave differently, and it looks like when both layers are very thin, the transistor actuation voltage would be at its minimum,” he says. “What we have discovered is that it looks like a flat layer of P and N will turn out to be the most perfect switch.”
Yablonovitch believes that development of a new switch will be well underway within ten years, possibly even in early use by then. “I’m confident we’ll get it, if not by myself or my team, we hope that scientists all over the world will follow our lead,” he says. “It’s so important, that ultimately we will find a way to make information technology more energy efficient.”