For more than half a century, progress in the production of silicon-based electronic components has made possible smaller, more powerful, more reliable, and cheaper products. But conventional microelectronic technology is beginning to reach its limits. Companies and academic researchers believe photonics will light the way to the next generation of microprocessor, communication, and sensing systems.
Nothing is faster than the speed of light. "That's really what it comes down to," says Stefan F. Preble, assistant professor of microsystems engineering in RIT's Kate Gleason College of Engineering and leader of RIT's Nanophotonics Group. "Current computer technology is limited by how quickly electrons can move. So there's definitely a motivation to use light for dealing with information."
Preble's group is developing ultra-small devices and systems that leverage the high bandwidth, low power, low latency, and sensitivity that are possible with light. They are focusing on integrating nanophotonic devices on a silicon CMOS platform in order to take advantage of the advanced fabrication techniques used in the microelectronics industry. The goal is to directly integrate photonics with current electronic devices.
Although the tremendous potential of nanophotonics has generated much interest, development of the technology for practical use in computers and other products is in the very early stages. "Right now we're really in the prehistorics of what we can do with photonics. At this point, it's possible only to have on order of maybe 1,000 components integrated on a chip," says Preble. "We're trying to develop the technologies, the architectures to scale that up to millions, so that eventually we could have these incredibly powerful computers."
A key to realizing this goal is developing technologies to control the characteristics of light. The basic building block for this is known as the ring resonator. "It is the photonic equivalent of the electronic transistor. While it will likely never replace the electronic transistor, it enables us to process information at the speed and bandwidths of light," says Preble. "With these ring resonators, we can directly turn the light on and off, which allows digital information (in the form of ones and zeros) to be transmitted."
However, ring resonators can do much more, Preble says. "We need to be able to manipulate the speed, frequency, and direction of light."
Preble's work is supported by two of the U.S. Defense Department's most prestigious honors, the Defense Advanced Research Projects Agency (DARPA) Young Faculty Award and the Air Force Office of Scientific Research (AFSOR) Young Investigators Research Program Award. He has also received funding from the National Science Foundation, the Semiconductor Research Corporation, and from RIT. Preble, who joined the RIT faculty in 2007, received MS and Ph.D. degrees from Cornell University and a BS in electrical engineering from RIT (2002).
"Dr. Preble is one of the top young researchers in the country working in the "elds of nanophotonics and microsystems," says Bruce Smith, professor and director of the RIT microsystems engineering program. "His accomplishments in these highly competitive and prestigious programs bears this out."
The Nanophotonics Group includes four RIT Ph.D. candidates, one postdoctoral researcher, and several undergraduate students.
The Air Force and the DARPA awards are focused on developing devices and systems to control all of the properties of light. The research is built on Preble's discovery of a new type of wavelength converter.
"Wavelength converters are needed to perform wavelength division multiplexing, where optical signals at different wavelengths are transmitted simultaneously. However, to date, wavelength converters have used schemes that are fundamentally non-linear, making their integration on a microelectronic chip challenging," says Preble.
His device uses a linear effect where the wavelength of light trapped in a ring resonator is changed by dynamically tuning the resonance of the cavity. Preble explains: "You can think of a guitar string as the acoustic equivalent of an optical cavity. By changing the length of a guitar string, you can create a rich set of sounds. We can do exactly the same thing to light by 'changing the length' of optical cavities." An article detailing this work was featured on the May 2007 cover of Nature Photonics.
"However, we can do much more than just change the wavelength of light by dynamically changing a resonator," says Preble. the research efforts in his AFOSR and DARPA programs envision not only changing the wavelength of light but developing a whole suite of tools for processing optical signals, such as optical memory/delays, optical isolators, and signal generators. "Dynamic nanophotonics enables unprecedented control of the properties of light, from its color to its speed."
The group has already made significant progress in achieving this control. Recently they demonstrated the slowing of light by a factor of (300, which is four times better than the best previous effort. "So that's a really huge improvement of the technology," says Preble. An article describing this work, "Controlled Storage of Light in Silicon Cavities," was published in the Feb. 1, 2010, issue of Optics Express.
Central to creating these dynamic photonic devices is a high-performance electro-optic modulator. the device uses a high bit-rate electronic signal to modulate a continuous light signal. Previous silicon modulators all suffer from a trade-off in speed, size, power, voltage, absorption, and temperature stability. The RIT group is working on a new type of silicon electro-optic modulator that utilizes a non-linear effect known as the DC Kerr Effect.
While this effect is small in crystalline silicon, Preble's group has shown that it can be quite large in a different form of silicon: nanocrystals. Preble anticipates that the modulator would represent a dramatic advance in performance and would also have many applications beyond electro-optic signal conversion, ultimately enabling high-performance on-chip wavelength-division multiplexed information processing systems.
The DARPA and AFOSR projects are working toward advanced functionalities that may make it into future products. At the same time, the RIT group, in collaboration with the University of Rochester and the University of New Mexico, is developing devices and architectures to improve the performance of the multicore computer systems so pervasive on our desktops today. the project, sponsored by the NSF and the Semiconductor Research Corp., aims to utilize photonics to relay information between the cores.
In current-day multicore systems, communication between the cores is handled electronically. Simply moving the information in this manner consumes a huge amount of power. the researchers are developing a multicore chip architecture that employs a hybrid electronic-photonic network to achieve low power, low latency, high bandwidth, and reliable interconnections among cores. Because power is in proportion to frequency, the electronic devices need to operate at a moderate 5 gigabits/second; however, the photonic links enable efficient operation at much higher data rates (around 40 gigabits/second).
"The multidimensional learning spans optics and lasers, device physics, electronic circuits, fabrication and process technologies, computer and network architecture, and error control coding," says Paul Ampadu, director of the University of Rochester's Embedded Integrated System-on-Chip Research Group.
"Stefan has always believed in the hybrid photonic/electronic approach to constructing reliable, high-speed, lowenergy multicore systems that exploit the unique characteristics of each technology," says Ampadu. "Working with Stefan and his team, one is constantly pushed to seek non-trivial groundbreaking solutions. We truly have a great team with a common vision; indeed, we so enjoy working together that we probably would've undertaken this project even without external funding. We can only look forward to many more of such exciting future collaborative ventures."
The team has already exceeded the initial accomplishment targets.
"This is is still in the realm of fundamental academic research," Preble says. "But this project is closest to commercialization. It could conceivably be in our computer chips by 2015 or so."
The RIT group is now looking at pushing the limits of light by utilizing the unique properties of individual photons-their quantum behavior-to realize quantum optical computers. "Quantum computers would really revolutionize our world," says Preble. "They would represent a fundamental divergence from current technology: Unlike conventional bits in today's computers-which represent information as zeros and ones-quantum bits (qubits) can have multiple values at the same time.
A quantum optical system consists of three basic elements: the photon source, quantum circuit, and detector. Past efforts to create quantum optic devices have been limited by use of large-scale components, such as beam splitters and polarizers, which were not integrated on a single chip. The goal of the RIT project is to integrate all of the components on a chip, creating a system that is low powered, compact, and robust. In addition, it will be rapidly and intelligently reconfigurable by making use of closely integrated CMOS electronics. The initial work is being funded by the university.
"I really feel quite strongly about this as a direction," says Preble. "I think we can make a huge impact on integrating the entire system on a chip."