Transistor design goes back to the power of empty space

Vacuums in electronics are popular things, and not just among guitarists who like to seek out amplifiers that use glowing valves. Even now, there is nothing to beat them when you need to handle very high voltages and high power levels, such as the 100W or more used by large communications satellites or the kilowatts of large-scale radar systems. At the lower end of that scale, silicon-based technologies have eaten into the market for vacuum electronics. That looks to be gaining pace now that materials such as silicon carbide and gallium nitride have become much easier to work with. Those two materials both demonstrate larger electron bandgaps than that of silicon, which makes it possible to build transistors with breakdown voltages measured in kilovolts.

Despite the advances made by SiC and GaN materials, Massachusetts Institute of Technology (MIT) researcher Winston Chern argued in a presentation at the International Electron Device Meeting (IEDM) last week: “Vacuum electronics are just fundamentally better. As long as your packaging doesn’t break down, the vacuum can withstand these high voltages. It doesn’t require any materials development: you don’t have to figure out how to grow thicker stacks of materials.”

The trouble with vacuum electronics is one of size, from thermionic valves to the magnetrons used for microwave-frequency amplification. Speed is also an issue as the main mechanism for releasing electrons into high vacuum is heat. The time it takes the heat to dissipate in the cathode makes it hard to switch the device off and on quickly, which is a drawback for high-power RF circuitry.

“Our goal is to build a vacuum transistor with the benefits of vacuum but with the footprint and cost of a semiconductor component,” Chern said of the work being carried by him and his colleagues at MIT and a couple of other institutions in Cambridge and nearby Boston.

The packaging technology has yet to be worked out, though the team is confident that one can be found. High vacuum is already a feature of some micromachined components, such as the accelerometers and gyroscopes found in today’s smartphones and tablets. So far the experiments have been performed in a larger vacuum chamber. This has let them get to a proof of concept that demonstrates a more responsive method for pushing electrons across a vacuum barrier.

Instead of a heated electrode, the cathode is an array of tiny silicon spikes which, when a voltage is applied, cause the electrons to tunnel through the substrate and into the vacuum before being pulled into a charged anode. The result is something that looks a little bit like a valve in cross-section but which behaves much more like the metal-oxide-semiconductor field-effect transistor (MOSFET) that now dominates electronics design. “The field emission array and the MOSFET look slightly different and operate with slightly different characteristics, but both have linear and saturations. These devices behave somewhat similarly to semiconductor transistors and they can be packaged to look like a MOSFET, making them a good replacement for high-voltage and high-power applications,” Chern claimed.

With no materials compositions to tweak in the vacuum channel, the main method to change device characteristics such as breakdown voltage lies in the distance between anode and cathode. There is a finite limit to vacuum’s resistance to voltage but that is at a pretty high level. “20MV per centimetre is when things start to tunnel into vacuum,” Chern said.

There are other limitations to a device like this when it is miniaturised in the same way as a conventional power transistor. Heat capacity is one: in the current design, 30A per square centimetre is how much the device needs to handle at the bottom.

Another factor is the electrical capacitance in the spikes, which seems small measured individually, just 50 attofarads or so. “We should be able to get to 1GHz operation without optimisation but we should be able to get one or two orders of magnitude beyond that,” Chern said: that upper limit would push the devices up into the operating range of magnetrons. “We think these devices could operate very quickly once you optimise their design. Vacuum transistors really shine when you look at the power/frequency tradeoff.”

The next step is work on increasing the electric field the devices can operate in and make the vacuum packaging work reliability, especially under the intense heat the transistors will have to endure. “But with all that we could make a very promising device,” Chern claimed.