Pennsylvania State University researchers have devised a technique for embedding an electronic junction directly into optical fiber, which potentially paves the way for more streamlined optical components.
The work represents the first time that electronic componentry has been directly built into fiber optics, said John Badding, a professor of chemistry who led the research.
Embedding high-speed electrical devices in the fiber has never been done before, Badding said. The Nature Photonics journal will publish a paper on the work, "Integration of GHz Bandwidth Semiconductor Devices inside Microstructured Optical Fibres," in an upcoming issue. The approach "provides a fundamentally new way of coupling an electronic device with the optical fiber, by building it right into the fiber adjacent to the light-guiding core," Badding said.
Converting light into electrical signals, and vice versa, is an operation vital for data communications. Data is moved about the globe mostly in the form of light traveling through fiber-optic cables. Yet, in order to be processed in a computer, the data must be in electronic form.
"You have light on the outside and electronics on the inside. At some point, those two things need to talk to each other," said John Ballato, a professor and director of Clemson University's Center for Optical Materials Science and Engineering Technologies. Clemson is conducting research similar to that of the Penn State researchers. "So what you are beginning to see are attempts to marry the optics and the electronics into a single form," said Ballato, who was not involved in the Penn State research.
Today, translating light and electronic signals requires a dedicated component, such as a silicon chip attached to the light source. Such components take up space and add a layer of complexity onto electro-optic system architectures. Also, because fiber is round and cylindrical and chips are flat, connecting the two is a challenge. A size disparity exists between the electronic and fiber components as well: The light-guiding pathways built into chips can be 100 times smaller than the optical fiber connected to the chip, causing a significant impedance mismatch that must be overcome.
The heart of Penn State's innovation is a new chemical procedure that involves depositing semiconducting materials layer by layer into tiny pores alongside a portion of the optical fibers, using a process called high-pressure chemical vapor deposition. "There was a lot of chemistry that went into making this," Badding said. The researchers didn't build an entire chip in the optical line that can convert photons into electrical impulses, which then can be further processed elsewhere. The junctions themselves are five to 10 microns wide, a few centimeters long, and can ingest data from frequencies as high as 3GHz on standard single-mode optical fibers.
Converting light into electronic signals directly in the optical fiber could lead to radical redesigns of current products, such as optical routers. It may even lead to products that would have been impossible to build otherwise. In much the same way the integrated circuit set the stage for the manufacturing of microprocessors, so too could hybrid electro-optical devices lead to entirely new technologies. "We're taking everything we've learned about integrated circuits, which are planar, and beginning to build analogous fiber-based devices, making them more integral to the fiber componentry that we use for communications," Ballato said.
Besides the research at Penn State and Clemson, work in this area is also being done at the Massachusetts Institute of Technology. Badding's approach is advantageous in that the layering process allows for a lot of control in how componentry can be embedded in the fiber, Ballato noted.
Badding admitted that the work is nascent. It doesn't address the complexities of converting electronic signals back into light, which would require additional semiconducting materials besides silicon. Nor was the work done on the multimode fiber now used in long-distance data communications. Nonetheless, it points a promising way to eventual commercialization, researchers note. "In my mind, the process is simple, and I can't see any reason it wouldn't be scalable as well," Badding said.
Sign up for Computerworld eNewsletters.