2013-10-15

Co.Labs

What Comes After Silicon In The Race For Faster Chips?

Some of our future computers may ditch electrons in favor of photons.



We've written about several "post-silicon" chip technologies recently at FastCo.Labs. You may remember clever breakthroughs like the carbon nanotube transistor and the nanotube computer. But even with clever innovations like these, the emphasis is still on electrons flowing through semiconductors. That's something a creation from Singapore may change--because it points the way toward all-optical chips.

The innovation comes from a team at the A*STAR Institute of Microelectronics, who have created something that at first blush sounds simple: A nano-scale optical switch.

Though we're great at shooting vast amounts of data down optical fibers at light speed, the data is encoded, decoded, and processed by shuffling through the pedestrian electrical circuits of a traditional silicon computer. The idea of getting rid of the silicon altogether is seductive, because an all-optical or even part-optical chip has a number of potential benefits including lower power consumption and high data bandwidths. But we haven't been able to translate our fiber optical innovations into the sort of tech needed for optical chips yet because devices like MEMS light switches simply flip too slowly from one state to the other.

Not the new "silicon ring" NEMS device, though--versus microsecond response times from traditional MEMS mirrors, the switch reacts in 50 nanoseconds.

It works in a surprising way. A tiny, flexible silicon ring is micromachined into a thin layer of silicon between two microscopic waveguides for laser light. When incoming light passes by the ring through one waveguide, the electromagnetic field of the light itself creates resonant electrical waves in the ring. This, thanks to the laws of physics, sucks energy out of the incoming light beam...effectively stopping it from passing.

To allow the beam to pass, or to turn the switch "on," another low power light beam is sent down the other waveguide. This of course causes its own resonant response in the ring, causing it to bend slightly. The bend changes the characteristics of the ring so that it won't resonate at the same frequency as the first light beam...and so that beam then passes on.

If you imagine lots of these optical NEMS switches stacked up, with the output light of one beam being used to trigger the state of another NEMS switch further down the circuit, you can see how it would be possible to build up complex circuitry. Similarly NEMS optical switches could simplify the interface between traditional electronic and optical data processing tech.

What's the practical upshot? Not much right now, though the team is already working on actually making optical circuits out of the switches. But it does mean that in the not-too-remote future you may have access to data processing power at supercomputer speeds on your desktop.

[Image: Flickr user Jenna]