Carver Mead and Lynn Conway pioneered a government/university/industry collaboration in the ’70s and ’80s that helped train a generation of engineers to design the chips inside the electronic devices we use today. Now the University of Washington is hoping to spark a new kind of alliance, inspired by Mead and Conway, to help scientists experiment with designs for faster, more energy-efficient optoelectronic chips that compute using both electrical impulses and photons of light.
The new program, which the UW is calling the Optoelectronics Systems Integration in Silicon, or OpSIS, is being unveiled at an event on campus today. The program, to be spearheaded by Michael Hochberg, a UW assistant professor of electrical engineering, is being funded in part by Santa Clara, CA-based chip heavyweight Intel (NASDAQ: [[ticker:INTC]]) and the U.S. Department of Defense. Mead, the legendary Caltech professor whose work was harnessed decades ago at an electronics collaborative called MOSIS, plans to personally attend today’s event. So does Intel’s chief technology officer Justin Rattner, and UW’s engineering dean, Matt O’Donnell.
This new effort at UW is supposed to enable scientists from academia, government and industry to experiment more cheaply with new circuits that integrate electronic and photonic technology, O’Donnell says. The underlying hope is that future laptops, smartphones, and other devices will eventually run faster and use less power if some on-chip communications and computations can be carried out using light, instead of just electrons. In the low quantities needed by today’s researchers, however, silicon wafers etched with both electronic and photonic circuits are extremely expensive.
Starting in 1981, the MOSIS project at the University of Southern California addressed a similar problem by making it possible for academic, government, and commercial researchers to combine designs for various integrated circuits onto a single “mask,” the screen providing the etching pattern for a silicon wafer. This allowed many organizations to share the costs of fabrication. UW wants to do the same thing for optoelectronic circuits. If it can become the place where scientists from around the world send in their designs for more integrated electronic/photonic chips, Seattle could turn into a central hub for a new movement toward faster computing devices that consume less power.
“Everybody is working toward this,” O’Donnell says. “It’s wonderful. If this flies, and OpSIS becomes the de facto standard for integrated photonic/electronic design, then we’re the leader. You will be able to point to us as the place this originated.”
The OpSIS project has so far gathered about $250,000 in support from Intel, according to university spokeswoman Hannah Hickey, plus some other in-kind support, O’Donnell says. The U.S. Air Force is chipping in another $500,000 a year for five years, Hochberg says. The chips themselves will be fabricated by BAE Systems. The OpSIS project is being established at what the UW is calling its Institute for Photonic Integration.
There’s no guarantee this will work in any practical way. The OpSIS program today has a half dozen experimental users, who are testing to see whether they can create the building blocks for integrated photonic and electronic chips.
The OpSIS project revolves around what it calls “shuttle runs” in which researchers cut costs by sharing silicon wafers between multiple projects. OpSIS leaders hope the strategy will reduce costs for any one team by more than 100-fold, Hochberg said in a statement. Protocols and rules on how to do this are still being worked out, although UW said it wants to make it clear enough that even non-specialists can make integrated electronic and photonic chips.
“You want a minimum of rules because people are going to use the technology in ways that you never imagined,” Mead said in a UW statement. “You want people to use it in ways that seem crazy.”
O’Donnell said the original MOSIS program was “incredibly important” to USC’s electrical engineering program for many years. If UW can do something similar with the new program, it could have a similar impact in the Northwest, he says.
Mead, now an affiliate faculty member at the UW, said in a statement that he’s “rooting for it.” He adds: “It’s a wonderful thing and it needs to happen. I might even use it for some of my own research.”
