InvestorsHub Logo
icon url

drjohn

02/19/05 5:03 AM

#15872 RE: Saturn V #15871

Nice explanation Saturn even a layperson like me can get a grasp on the importance of intel's optical silicon tech.; Thanks.
icon url

The Duke of URL

02/19/05 10:45 AM

#15874 RE: Saturn V #15871

Thanks.
icon url

chipguy

02/19/05 11:38 AM

#15875 RE: Saturn V #15871

Todays slow electrical interface has caused everyone to pack more components on one chip to bypass the slow I/O interface. Optical interface might change that and might bring back the world of large computing mainframes.

I don't think systems will ever de-integrate. There is
simply too big a cost savings in minimizing chip count.
Also optical chip to chip gives you cheaper bandwidth
in theory but does nothing for latency, perhaps even
hurts it a bit.
icon url

mike306oh

02/19/05 12:46 PM

#15876 RE: Saturn V #15871

Saturn,

The next phase in the chip development will most certainly be the Optoelectronics phase (what I call a hybrid chip). All electronics chips can’t deliver speed. So the all-microelectronics based chips will be history soon. Next phase will be photonics integrated chips (what I call all optical chips).

You are right. For now Intel is getting into the hybrid chip:
Giving the fact that Intel is a recent entrant into the photonics field, the first step toward a hybrid chip is very significant.

From one of last PR:

“Initially, they discovered increasing the light pump power beyond a certain point no longer increased amplification and eventually even decreased it. The reason was a physical process called "two-photon absorption," which occurs when two photons from the pump beam hit an atom at the same time and knock an electron away. These excess electrons build up over time and collect in the waveguide until they absorb so much light that amplification stops.

"Intel's breakthrough solution was to integrate a semiconductor structure, technically called a PIN (P-type - Intrinsic - N-type) device into the waveguide. When a voltage is applied to the PIN, it acts like a vacuum and removes most of the excess electrons from the light's path. The PIN device combined with the Raman effect produces a continuous laser beam."

"Fundamentally, we have demonstrated for the first time that standard silicon can be used to build devices that amplify light," said Dr. Mario Paniccia, director, Intel's Photonics Technology Lab. "The use of high-quality photonic devices has been limited because they are expensive to manufacture, assemble and package. This research is a major step toward bringing the benefits of low-cost, high-bandwidth silicon based optical devices to the mass market."

Even more significant to me is the fact that contrary to all we knew thus far about high speed Optoelectronics (that were supposed to be based on III-V compound semiconductors), the last announcement from Intel demonstrates that Silicon can be used; at least for two of the critical optical components. Of course, you’ll still need to develop and integrate all the other optical components, i.e. arrayed-waveguide gratings, with aplication to channel multiplexers, demultiplexers, optical switches, dispersion compensators, plus the passive components such as lenses, and prisms. So yes, you are right in the near future the chips will use hybrid optoelectronic circuits.

To me Intel announcement is very important not only because Si is still the “king,” (sad for me to recognize it– as a former developer and strong believer of/in III-V devices for the last 20+ years), but because the integration of optics and microelectronics using all-Si components will happen much faster. Just for instance think that the attenuation of channel wave-guide fabricated using Si is expected to be quasi identical with equivalent glass fibers. Similar channel waveguide fabricated using polymers can be expected to have at least an order of magnitude higher attenuation. So yes, it looks like all Si is the way to go.

Next logical phase will be all optical chips. If this can be made on Si, the development of these chips will not take as long as previously thought. As you know, previous designs were involving all those fancy stuff such as nonlinear optics, and were supposed to be made from various materials that are hard to manufacture and integrate.

Not only fast speed computers, but a whole range of other applications will push the development of all optical chips. Think about all the clear advantages that integrated optics have as compared to the traditional electronics. Using light as the carrier of information, which is unaffected to electromagnetic perturbations, cause this field to be one of the most viable solutions concerning the telecommunications bottleneck. In addition, the application of integrated optics in the sensor field offers a better response as compared to the transducers used nowadays: Its capability to resist harsh environments, the measurement without direct contact and the safety in explosive media cause this to be of huge interest for the industry, etc, etc.

Mike