Select your localized edition:

Close ×

More Ways to Connect

Discover one of our 28 local entrepreneurial communities »

Be the first to know as we launch in new countries and markets around the globe.

Interested in bringing MIT Technology Review to your local market?

MIT Technology ReviewMIT Technology Review - logo


10 Emerging Technologies

(Page 3 of 10)

Silicon Photonics
OPTOELECTRONICS Making the material of computer chips emit light could speed data flow. By Neil Savage

The Internet lives on beams of light. One hair-thin glass fiber can carry as much data as thousands of copper wires. But inside your computer, copper still rules. The advantages of light haven’t translated from long-distance connections on the Internet to the short jump between computer chips, in part because the lasers used in optical communications are made from exotic semiconductors incompatible with the standard processes for making silicon computer chips. As computers get faster and faster, they’re nearing the physical limit of copper’s ability to carry more information, and they’ll need something like the fiber-­optic network in order to keep improving at the rate we’ve come to expect.

Getting silicon to emit light could be the solution. A light signal’s frequency is much higher than an electrical signal’s, so it can carry thousands of times as much information. Light also overcomes another problem with electrical signals; as transistors get closer together, the electrical signals passing through them start to interfere with each other, like radio stations broadcasting at the same frequency. But turning silicon into a light emitter has proved an extraordinarily difficult challenge. The problem is rooted in an energy-level mismatch between silicon’s electrons and its positively charged “holes” (electron vacancies in its crystal structure): when an electron meets a hole, it’s more likely to release its excess energy as vibration than as light.

But last fall, a team at the University of California, Los Angeles, became the first to make a laser out of silicon. In February, ­Intel scientists upped the ante, reporting a silicon laser that put out a continuous instead of a pulsed beam, a necessity for data communications. “Once you identify the right piece of physics, everything falls into place,” says UCLA electrical-engineering professor Bahram Jalali, who made the first silicon laser.

The right piece of physics is the Raman effect. Some photons of light that pass through a material pick up energy from the natu­ral vibration of its atoms and change to another frequency. Jalali fires light from a nonsilicon laser into silicon. Because of the ­Raman effect, the photons emerge as a ­laser beam at a different frequency. This Raman laser is “a fundamental scientific breakthrough,” says Mario Paniccia, director of Intel’s Photonics Technology Lab, which is working to create the devices needed for optical communications in silicon. In addition to building a laser, he and his colleagues created a silicon modulator, which allows them to encode data onto a light beam by making it stronger or weaker. Paniccia’s group is working to more than double the speed at which it can modulate a beam. A multibillion-­dollar infrastructure is already in place for making silicon chips, so Intel believes silicon lasers will be a cost-effective way to raise the computing speed limit.

Photonics-based interconnects between chips should start to appear in about five years, researchers say. The ultimate goal is to enable light-wave communication between components on the same chip, which is several years further out. Philippe Fauchet, professor of optics at the University of Rochester, believes on-chip optical communications will require a silicon laser powered by electricity, which would be cheaper and less complicated than one that depends on an external laser. If such a laser can be built, it will mean that everything from supercomputers on opposite sides of the globe down to the tiniest transistors can talk to each other at the speed of light.

Next Page »