Technological Landscapes: Dreams of an Optical Computer

The technological landscape consists of both ascendant and broken dreams. One such dream is that of higher density, lowest cost, fastest-ever optical computers manufactured with all-silicon components.

So, when Professor Lorenzo Pavesi’s nanosilicon photonics team at the University of Trento, Italy coupled the optical emission from a tiny dot of silicon to a silicon waveguide, they emblazoned a trail in the landscape that moved the breakthrough computer dream one step closer to reality.

Due to Pavesi’s multiple contributions to nanosilicon photonics he was welcomed as a guest lecturer by the Boston Chapter of the IEEE Photonic Society to summarize his latest achievements. The lecture was held in the cafeteria of MIT’s Lincoln Laboratory in Lexington on the evening of Dec. 8, 2011.

As Pavesi pointed out in his captivating lecture, the arc of densely packaged conventional computer chips has followed an upward trajectory over the past 40 years that is hundreds of times ahead of the technology for optically-driven computer chips. This grossly uneven status introduces a two-fold problem. First, there is the insufficient progress in the overall optical computer domain. Second, there is the dilemma that conventional computer chip technology is running out of room for growth. As a result, the promise of a breakthrough computer still lies within the all-silicon optical computer dream. Pavesi has feverishly pursued this concept over the past decade.

It is a concept that was born nearly 50 years ago.

It was the winter of 1962. My job at Lincoln Laboratory was to find material best suited for use in the construction of an emergent optical device termed a gallium-arsenide (GaAs) injection laser. The laser was the next step after the discovery of GaAs light emitting diodes (LEDs) in the years prior.

The level of expectation was palpable as the laboratory attempted to be the first to announce the discovery of the first GaAs laser and thereby beat the GE group in Schenectady/Syracuse, NY and the IBM team in Yorktown Heights, NY. All three teams published within days of each other in the peer-reviewed Applied Physics Letters. Simultaneously, laser technologists began to envision an ultrafast optical computer that would be based on GaAs lasers optically coupled to silicon (Si) computer chips.

Unfortunately, the dream remains unfulfilled 48 years later, even though there has been a monumental effort world-wide to achieve this ultimate machine. The delay has to do with the most fundamental nature of semiconductors – their ability or inability to produce optical radiation as opposed to simply heating up the material. It turns out that material such as GaAs converts electrical energy into optical energy at a rate far in excess of materials such as silicon. In fact, in 1962 it was inconceivable that silicon could be developed as an effective emitter of enough useful radiation required for the dream machine.

Moreover, what was unknown then and remains true today is that coupling the GaAs devices to Si chips would constitute an immense and costly challenge. The industry prescribed a dream machine in which all the elements could be manufactured from silicon and reality had limited progress to a hybrid of several nearly incompatible materials.

And then serendipity, the constant companion of research, entered at center-stage.

As Si chip technology created smaller and smaller-sized devices, new properties of silicon began to arise due to quantum effects that were previously over-ridden by other properties inherent in larger sized devices. To everyone’s delight, when the silicon devices were manufactured in the microscopic range, they generated useable visible and near infrared radiation and reopened the gates for the all-silicon optical computer.

These tiny chunks of silicon are referred to as quantum dots and are approximately one to five nanometers in diameter. This is really small. It’s the size of just a few thousand atoms. Most importantly, just as the development of LEDs created a new world of display systems, the discovery of useful levels of optical radiation from quantum Si-dots has reenergized dream-machine design.

Professor Lorenzo Pavesi’s tireless efforts to realize the optical computer dream was recognized by the Italian government when he received the title of Cavaliere (Knight) for scientific merit. He was elected the distinguished speaker of the IEEE-Photonics Society for the years 2010 and 2011. Lincoln Laboratory continues work in this area with innovative programs exemplified by their integrated photonics initiative with MIT for advanced communications systems.