STANFORD, CA--The engine behind the formation of Silicon Valley--one of the most successful high-tech business environments in the world--is Stanford University. To understand this, take a stroll around the Stanford campus, where you will find the Hewlett Teaching Center, the Packard Electrical Engineering Building, the Varian Physics Building, and the WIlliam Gates Computer Science Building. The first three structures were named after alumni (or companies) who helped create the Valley, while the last was named after someone who benefited immensely from its fruits, and thus channeled $6 million back into the “engine.”
Famous for its focus on semiconductor chips, then software and the Internet, the Valley is also a hotspot for photonics. Helping to propel the innovation here too is Stanford, as evidenced by the variety of topics covered at the Stanford Photonics Research Center (SPRC) 2008 Annual Symposium: Global Impact of Photonics, held September 15–17, 2008 at the Hewlett Teaching Center. While roughly half the presenters hailed from universities and companies located around the U.S. and in other countries, the other half were Stanford professors and students.
Topics ranged from the esoteric (quantum science) to the everyday (car headlights). However, even presentations on exotic subjects had a practical slant. Photonic quantum-computing experiments have traditionally been extremely delicate setups based on, for example, individual atoms in Fabry-Perot cavities, but, as Hideo Mabuchi of Stanford noted in his presentation on quantum nonlinear dynamics, there is a push toward solid-state quantum computing elements such as those based on quantum dots. Joseph Kerckhoff of Stanford, in his talk on quantum error correction, spoke not only of solid-state systems but also of ways to make quantum memory retrieval less fragile, such efforts combining to “engineer robustness into ordinarily delicate systems,” in his words. So practical quantum computing, still years away, has begun to take shape.
Photonics in computing
In traditional electronic computing, boosting the speed and volume of data transmission is crucial to making better computers. Jeffrey Kash of IBM Research (Yorktown Heights, NY) outlined the progress being made in optical interconnects. The IBM Roadrunner 1 petaflop computer, now in operation, has all-fiber-optic rack-to-rack interconnects that drastically reduce cable bulk. Coming next, said Kash, are board-level optical interconnects; already built is the experimental Terabus link, with a 16-channel version transmitting 160 Gbit/s both ways, with the Terabus moving out of the lab and into systems in the next 5 to 6 years.
Ultimately, optical interconnects will become part of the chip itself, connecting processors to each other. Ray Beausoleil of Hewlett-Packard (Palo Alto, CA) described a global interconnect architecture that he felt could bring about an optical Moore’s Law, based on dense wavelength-division multiplexing (DWDM) and microring optical resonators mode-matched to waveguides, and relying on nanoimprint lithography for fabrication.
The explosive growth of the Internet is driving fiber-optic technology to ever-higher levels of performance. Loukas Paraschis, business-development manager at Cisco Systems (San Jose, CA), noted that by 2010, the average linked-in U.S. household will generate more data traffic than did the entire 1995 Internet backbone. He predicts a volume of 44 exabytes/month (1 exabyte equals 1018 bytes) of IP traffic by 2012; in response, WDM transport will need to evolve to 100 Gbit/s, but will have to do so without changes in the fiber-optic line or the EDFA (erbium-doped fiber amplifier) spacing. Paraschis sees convergence of WDM and IP, for example with WDM ports on routers.
Maurice O’Sullivan of Nortel Networks (Toronto, Ont., Canada) described some of the technology that will help boost fiber-optic transmission to 100 Gbit/s transmission rates, including an electronic dispersion-compensating modem, which contains a digital-signal-processing-instructed electric-field modulator and intensity detection, and which can “pre-compensate” for 3000 km of optical dispersion, meaning that the light signal enters the optical fiber with an amount of optical dispersion opposite to that in the fiber span.
Steering cars and mice
In the area of automotive photonics, Wende Zhang, a senior researcher at General Motors (Detroit, MI) began with a description of the DARPA Urban Challenge, in which a driverless car--equipped with vision systems and/or laser sensing systems, and guided by sophisticated algorithms--had to autonomously drive a 60 mile course while following all traffic rules. The photonic technology developed for the GM car, which won the DARPA challenge, can lead to practical and safe driver warning/assist systems (such as side blind-zone alerts), automatic lane centering, and even on-demand autonomous driving. Arne Stoscheck of Volkswagen (Wolfsburg, Germany) detailed the company’s research into all-LED headlights, leading to their introduction on the Audi R8. Future projects include intelligent high beams (selective light distribution) and laser fog lights, said Stoscheck.
In the medical arena, Viviana Gradinaru of Stanford described a technique for optical control of brain function. First, a form of DNA responsive to light is incorporated into a virus, which is then placed into a specific spot in the animal’s brain, perhaps when the animal is a mere embryo. The DNA is expressed in the brain; after the animal grows to maturity, a blue-emitting LED is fastened over a small hole in its skull. Gradinaru showed a video of a freely moving mouse that, upon photonic stimulation, ran around in startlingly repetitive circles until the LED was turned off. Ultimately, the researchers want to better understand neuropsychiatric diseases in humans and develop more-effective therapies.
Not all presentations were technical. Sven Strohband of Mohr Davidow Ventures (MDV; Menlo Park, CA), an early-stage venture-capital firm, talked about MDV’s movement into photonics, and what the company looks for in a potential venture. First, the potential product should have a large technical differentiation: “Not a 1.1X, but a 10X change,” said Strohband. Second, the development team should be balanced in not only fields of expertise, but personality types. Third, the technology should be past its preliminary development, or, as Strohband put it, “Less R and more D.”
Other areas covered at the SPRC symposium included advanced laser and nonlinear devices, nanophotonics, ultrafast physics and x-ray science, and optical clocks and applications.
