Alexander Gaeta's heavy research into light technology promises to lead to speedy microchips that use streams of photons barreling down microscopic waveguides instead of electricity flowing through minute wires.
Gaeta, a professor of applied and engineering physics at Cornell University, and Michal Lipson, a Cornell electrical and computer engineering assistant professor, are leading a research team that's investigating the potential of photonic silicon chips. The team recently created a chip-based broadband light amplifier, a potentially key component in future photonic circuits. Academic labs worldwide are anxiously pursuing photonic silicon research, aiming to create devices that mix photonics and electronics on the same inexpensively manufactured chip (see photo).
The new broadband light amp chip utilizes a technique known as four-wave mixing in which the signal destined for amplification is "pumped" by another light source inside a waveguide measuring only 300 by 550 nanometers. The photons in the pump and signal beams are tightly confined within the waveguides, allowing for a transfer of energy between the two beams. The chip's waveguides are silicon channels surrounded by silicon dioxide.
Since silicon doesn't emit light very well, Gaeta and his researchers turned to an obscure phenomenon known as the Raman effect to amplify light within the chip. The Raman effect relies on the fact that when light is transmitted through matter, part of the light is scattered in random directions. A small amount of the scattered light has frequencies that are removed from the frequency of the incident beam by amounts identical to the vibration frequencies of the scattering system, explains Gaeta. This light is called Raman scattering. If the initial beam is highly intense and monochromatic, a threshold can be reached beyond which light at Raman frequencies is amplified, builds up strongly and begins showing the characteristics of a stimulated emission.
Previous attempts at using the Raman effect for silicon-based optical amplification were hampered by that fact that signals could only be produced at a single wavelength. Gaeta's team, however, was able to generate signals across several wavelengths. "We have demonstrated that a particular effect can amplify almost 30 wavelength channels at once," says Gaeta. The technique can also be used to create a duplicate signal at a different wavelength, allowing the technology to easily convert a signal from one wavelength to another. Although other researchers have already created four-wave mixing amplifiers using optical fibers, these devices are tens of meters long, hardly small enough for chip-based applications.
The research team tested its invention with infrared light generated near 1,555 nm, the wavelength used in most fiber-optic communications. At 28 nm, the researchers were able to achieve amplification over a range of 1,512 nm to 1,535 nm. Longer wavelengths provided even greater amplification in a range from 1,525 to 1,540 nm. Gaeta predicts that much better performance could be obtained by refining the process. "We're still kind of feeling our way around," he says.
Gaeta also believes that four-wave mixing applications that have already been demonstrated in optical fibers will now be possible in silicon. These applications include all-optical switching, optical signal regeneration and optical sources for quantum computing. He predicts that hybrid photonic/electronic circuits will first be used in data repeaters and routers, where several different wavelengths are sent over a single fiber at the same time.
Innovations like the chip-based light amp will prove to be pivotal in moving photonics from long-haul telecom networks to computer backplanes, says Gaeta. "As the need to transmit data between processors outstrips the capabilities of pure electronic technology, photonics will step in as the solution," he says.