SOUE News Issue 2

Research-in-Progress Lecture: Ultra-Fast Electronics

One of the two research-in-progress lectures from the 2002 Jenkin Day, given by Chris Stevens - report by David Witt

Chris set out to look at ways in which one might improve significantly on the speeds of present electronic devices, which are, roughly:

300 GHz for GaAs/Si narrow-band analogue integrated circuits;

3 GHz for digital CMOS (perhaps extendable to 10 GHz one day).

Five possible future devices were considered, with most attention to the last:

  1. Intel THz transistors with buried 30 nm channels. These are hard to make and raise a severe problem in heat dissipation;

  2. The resonant tunnel transistor, good in theory but even harder to make;

  3. The single electron transistor, in which the presence or absence of a single electron on the gate is sufficient to change the logic state. The devices are therefore extremely small and it is a real challenge to make an integrated circuit out of them;

  4. Optical logic, which can switch at the amazing speed of 50 THz. But a single switching "circuit" is impossibly large, requiring a whole optical bench;

  5. Rapid Single Flux Quantum Logic.

This last is effectively a set of superconducting rings containing Josephson junctions. Each ring is of a size that can hold just one quantum of magnetic flux passing through it, about 2x10-15 Wb or Vs (=h/2e). Thus an applied voltage of only 2 mV can make the flux move to an adjacent ring in just 1 ps. Chris explained how standard logic gates could in principle be developed using this phenomenon. An experimental device using a niobium superconductor has already been made into a flip-flop that could operate at 750 GHz. Devices using high-temperature superconductors could potentially work even faster.

Chris pointed out that there was a serious problem in testing new devices that were significantly faster than anything out of which one might build test equipment. But the optical techniques (No. 4 above), though no substitute for electronic logic circuits, could be ideal for making the test equipment now needed. Lasers could produce light pulses shorter than 1 ps, and as an example of what might be done he described an interesting technique for generating very short pulse trains or digital signals by combining light beams which had been subject to small differential delays. The delays are got by passing the light through stacks of glass microscope slides, each of which generates a delay of a few ps compared with an air path of the same length. The various delayed signals are then combined with a simple lens.

In conclusion, Chris said that the prospects for much faster electronics were looking good. Major research investment was under way - but we would need to overcome the reluctance to use cryogenics!

<<   Previous article Contents Next article   >>