Quentin Diduck thinks it's time for a revolution in transistor technology. In fact, the graduate student at the University of Rochester claims the new ballistic deflection transistor (BDT) he's working on is as different from a conventional transistor as an old electron tube.
As today's transistors edge closer to heat, electrical leakage, and other realworld limits, academic and corporate researchers are examining a variety of exotic barrier-breaking designs. Diduck, though, believes it's time to throw out the conventional transistor model and replace it with something entirely new.
Instead of flowing through a transistor like water, electrons in the ballistic design deflect and bounce in a manner that resembles a game of atomic billiards (see the figure). The device itself works kind of like a solid-state two-dimensional CRT, with electrons shot across a wide path.
"We use a high-mobility material system that creates a two-dimensional electron gas (2DEG)," says Diduck. "This material system has a large electron mean free path, on the order of 150 nm, even at room temperature." The approach lets electrons travel ballistically rather than in conventional drift/diffusion.
To create the nanoscale BDT prototype, Diduck, University of Rochester computer engineering professor Marc Feldman, and several other researchers formed the 2DEG material—a semiconductor sheet— into a cross shape with a triangle etched out at the center. "If one applies a potential across the top and bottom of the cross structure, we get electrons coming up from the bottom moving toward the top of the device, just like an electron gun," Diduck says.
Lateral gates on each side of the cross structure help guide the electrons to opposite sides of the triangle. "The triangle is an artificial scattering mechanism that alters a small change in momentum introduced by the gates into a large change in momentum and deflects the electron into either the right or the left channel depending upon the bias of the gates," Diduck says. "Since momentum and energy are directly related... we have a small change in energy producing a large change in energy or, in effect, gain."
"It's a transistor based on an entirely new set of principles," says Feldman. In fact, the research's ambitious goal is to create a very low-power transistor that can operate at terahertz-plus speeds at room temperature.
"Speculation is that this design will scale better than current CMOS designs and will enable a further reduction in circuit size," Diduck says.
Diduck believes the BDT, which could be commercially available in approximately two years, initially will cost about the same as conventional technologies and has the potential to become much less expensive down the road. He expects the device to find its first use as a fiberoptic network switch.
"The telecom industry isn't afraid of exotic materials and are the most likely to be early adopters," Diduck says. "The general idea would be to provide a product that would enable orders of magnitude-better performance, although we are still too early in this research to know for sure if this technology will outstrip the competition."