Despite those advances, the world is making increasing demands to come up with advances across a range of technologies from computers to lighting, more cheaply, quickly and better. “As the projected demand for computing goes up by orders of magnitude, we need to keep the cost and energy on the curve to be efficient, so I think that we’re going to need dramatically different technologies in the future, and also technologies for optical processing and switching,” according to Professor Alice White, chair of the Mechanical Engineering Department at Boston University.
One possibility to expand the capacity of optical fibers is to create doped fiber amplifiers in other wavelength ranges, similar to what erbium-doped fiber has done for the 1550nm band. One possibility is bismuth-doped fiber, which has broad luminescence at 1200 nanometers, suggested White, who previously served as Chief Scientist with Alcatel-Lucent Bell Labs. "These advances have to come at reasonable cost or they're simply not going to be able to continue to drive the economy the way that they have historically. The whole community is depending on materials and materials processing component designers to come up with schemes that are cost and energy efficient, so integration is going to be key,” she said.
MIT Professor Lionel Kimerling, who also is Director of the Microphotonics Center, predicted board level photonics will be integrated in multi-core processor computers by 2017. Researchers in Kimerling’s EMAT lab at MIT demonstrated the first germanium (Ge) laser monolithically integrated on silicon substrates in 2010. “What you really need to do it right is every core to talk to every other core. What you really need to do it right is every core to talk to every other core. If that’s possible then you can use our silicon microphotonics, just embed that into the microprocessor design, and we can create an optical power supply bus, we can drop different colors of light into our data waveguide, we can modulate those colors with the signals coming out of the processing node and we can send them to other cores; they can be detected, and then they can use the data.”
“You can essentially get 220 optical channels, without any need to tune, using a silicon waveguide,” Kimerling said.
Related Materials Day articles