We live in a world that relies on transistors. These little electronic components moderate the flow of electricity in everything from computers to cars, refrigerators to radios. Since being invented in 1947, transistors have revolutionized society. Now, six decades later, their impact is greater than ever, and electronics consumes an ever-growing fraction of society’s electricity.
Even as the number of devices grows larger, the technologies inside them have gotten smaller and smaller. Transistors are commonly now on the scale of a mere 14 nanometers in size. UC Berkeley Electrical Engineering professor Eli Yablonovitch notes that because we have so many more devices, we need to make them 10,000 times more efficient.
One problem, he says, is that transistors waste energy as heat. You can feel this heat when your cell phone—which packs thousands of transistors into a tiny silicon chip—heats up as you use it. Heat, Yablonovitch points out, is not a very useful form of energy. “Pure energy is what’s comes right out of the wall socket,” he says, noting that once it gets flowing through our wires and devices, it gets degraded into heat.
Engineers have made transistors more efficient by shrinking them. Now that they’re as small as molecules, another approach is needed. “The question now,” Yablonovitch asks, "can we make a more-sensitive, lower-voltage, switch to that will consume less energy? Can we replace the transistor?”
Efforts like those at the Kavli ENSI, focused on basic research, are what will lead to the required expansion of knowledge. There are four different research approaches to making this leap in efficiency:
One approach is making the “more sensitive switch” Yablonovich speaks of, a switch that can turn on and off with minuscule amounts of voltage. Right now, it takes about one volt to flip a transistor. But Yablonovich believes that it’s possible to create nanoswitches that will permit switching voltage on the order of 10 millivolts—just 1% of what’s required now.
“We’re just at the beginning of this,” he says, pointing out that engineers have had a 60-year run of making things smaller. “Just as we’ve reduced the dimensions of transistors from microns to nanometers—that’s three orders of magnitude—we’ll need a 60-year run to go from a volt to a millivolt.”
Yablonovitch’s approach is one of several that researchers are taking in the effort to increase efficiency. A second approach to low voltage switching is mechanical switches. Right now, the circuits in our computers might have a small current leakage all the time. Mechanical nanoswitches will cut off that leakage current more completely. They would likely be slower than transistors, but in the internet-of-things, where everyday objects relay data to each other, a super-fast response time isn’t always vital.
Yablonovitch describes an example of such a mechanical switch, which he calls a “squitch.” It’s like a nanosandwich: The “bread” is two bits of metal that will pass a current when they make contact. Between them is a molecule that will keep the metals from fusing together when the current passes through. When the switch is turned on, the bits of metal squish the molecule between them, getting close enough to pass the current. When the switch is turned off, the molecule is no longer squished, and the current flow stops.
A third avenue of research arises from thinking about transistors as communications devices. “The purpose of a transistor is to send a signal on a wire,” Yablonovich points out. “So it’s really a communication device and the medium is wires.” Yablonovitch is a pioneer in the research and development of silicon photonics, a form of optical communication. Currently, he says, transmitting one bit of information requires about 20,000 photons. He believes that number can ultimately be reduced to 20 photons.
The fourth approach to increasing efficiency, still in its very early stages, is nanomagnetic switching. This idea builds on recent improvements in our ability to use electrical current to efficiently switch the poles of a magnet. New discoveries have made it possible to turn that changed magnetic field back into current, completing a logic circuit. The process needs some refining to make it viable, and researchers are striving to find solutions.
“This type of research, if you’re doing it right, is risky,” says Yablonovitch. He emphasizes, that the point is to explore many avenues. Different approaches will be fruitful for different types of applications. And what we learn along the way will form the basis of new advances that we can’t foresee.
“Fifty years ago, when the transistor was young,” Yablonovitch says, “people had uses in mind. But no one expected we’d be walking around talking on mobile computers.” He notes that fundamental changes only come with new knowledge that can’t be predicted. “The most important thing for us to do is to enable revolutions for which we can’t yet imagine the outcome.”