/ 24 July 1998

Taking IT to the limit

A new superfast transistor is set to revolutionise computer chips, writes Michael Brooks

Researchers at Yale University have developed a transistor so sensitive that it can watch single electrons moving along a wire.

The presence of one electron in the transistor switches it on, and it switches on and off 1 000 times as fast as any other comparable device.

But the most interesting aspect of the breakthrough is the scope for further developments in this direction – there is almost none.

“Where we’re at right now is something like 40 times worse sensitivity than the intrinsic limit,” says Robert Schoelkopf, who made the breakthrough with his colleague Daniel Prober.

Once you are dealing with single particles, there is a fundamental limit to the measurements that can be done at this quantum level. You can’t, for instance, know both the position and momentum of a quantum object like an electron, because an electron is neither a particle nor a wave, but both simultaneously. A particle has well- defined position, a wave has well- defined momentum.

But if you want to count electrons passing along a wire, they need to be particle-like: you need to know where they are to know when they pass, and this is the essence of Schoelkopf’s remarkable device.

His transistor has a slightly different design to those used in computer chips, but computer manufacturers are aware that similar problems cloud their horizon. Their own transistors are shrinking to achieve better performance, and are expected to reach the quantum limit in about 2010.

“Their transistors will stop behaving like the ones they have today, and start behaving like the ones we’ve been working on,” says Schoelkopf. “That will change all the rules about how you have to design the circuits.”

Transistors switch a current on and off by applying a voltage to a “gate” just above a semiconductor channel for electrons. This voltage repels electrons, stopping current flow through the channel.

Single electron transistors (SETs), like Schoelkopf’s, have a break in the channel, with a tiny conducting island in the middle. This means that the electrons have to jump on to the island, and jump off it on the other side. Only one electron can be on the island – its repulsive force is enough to keep the others away.

But changing the voltage at the gate can reduce this effect enough for another electron to make the jump to the island; the original occupant is then forced to jump off on the other side. Detecting individual electrons in a wire is done by putting the wire where the gate should be. As the electrons flow through the wire, each one changes the electric field enough to push new electrons on to the island. Thanks to Schoelkopf’s improvements to SET technology, the effect of this movement can now be measured.

The electrons in the channel are made to interact with radio waves. As the electrons make a jump, the radio waves give them an extra push, speeding up the process. Because the waves then lose energy, it also provides a way to see what the transistor is doing.

“By measuring how many radio waves are bouncing around, we can detect the state of the transistor – whether it is on or off,” he says.

Currently the system only works at below -272o C, just half a degree above absolute zero. This makes it expensive and tricky to run, but that doesn’t bother Nasa, the Yale team’s backer, which sees the device providing the ultimate in sensitivity for the photodetectors used in high-performance telescopes and microscopes.