PHOTONICS APPLIED: INTEGRATED PHOTONICS: Can optical integration solve the computational bottleneck?

摘要

An ambitious DARPA award to Sun Microsystems and its partners to develop a chip-to-chip and intrachip communications platform offers a glimpse into the world of next-generation integrated photonics technology and how it aims to break the computational bottleneck. Truly integrated photonics--the combination of discrete optical components such as light sources, modulators, switches, filters, splitters, waveguides, and detectors on a single integrated platform--remains a "holy grail" in the photonics community. Despite advances by companies like Intel (Santa Clara, CA), an all-optical platform for high-speed communications and signal processing has yet to be realized (see www.laserfocusworld.com/articles/289409). Companies such as Enablence (Ottawa, ON, Canada), HP Laboratories (Palo Alto, CA), and IBM Research (Yorktown Heights, NY) are working with indium phosphide (InP) and other semiconductor and polymer materials to develop photonic integrated circuits to replace electronic interconnects (see www.laserfocusworld.com/articles/346710). But an ambitious Defense Advanced Research Projects Agency (DARPA) program provides an excellent framework for understanding the challenges that face developers of these next-generation integrated photonics architectures. Funds under the DARPA program, called Ultraperformance Nanophotonic Intrachip Communication Program (UNIC)--pronounced "unique," were awarded to Sun Microsystems (Santa Clara, CA) and its partners at Kotura (Monterey Park, CA), Luxtera (Carlsbad, CA), the University of California at San Diego (UCSD; La Jolla, CA), and Stanford University (Stanford, CA) to develop optical chip-to-chip and intrachip communications technology that will help meet the challenge of increasing computer performance to keep pace with tomorrow's computationally intensive applications.