Out of the Lab: The Hole Thing
One project at Sandia National Laboratories in the U.S. promises to yield optical waveguides that can guide light around tight bends with zero losses. Currently, optical waveguides rely on small differences in refractive index to channel light. To get light around a corner with no losses requires a bend radius of several centimeters in some cases -- not good news for making compact optical circuits. Sandia's technology has huge potential for overcoming this limitation.
In a separate but related development, scientists at Kyoto University in Japan have made the integrated equivalent of an optical add-drop multiplexer -- though at the moment it only performs the drop function. The device is based on microcavities, which select a specific frequency of light from a waveguide (as a fiber Bragg grating does) and spew it out of the top surface. But unlike Bragg gratings, which need to be several centimeters in length, the microcavity is less than 1 micrometer wide.
The number of manufacturers working on optical integration is an indicator of how important a direction this is considered to be. However, most of these companies -- for example Bookham Technologies Ltd. and the newest, Sparkolor Corp. (see Sparkolor Plays Catch Up) -- take components made in separate processes and then glue them onto a substrate. Intense Photonics Ltd. and a few others are attempting to make monolithic chips (see Scotland Spawns Component Startups), but they're at a very early stage, and the number of components they can integrate is small.
Making large numbers of components at the same time on the same chip will require a radically different approach. In that sense, the integrated optics industry is at the same stage that the electronics industry was 30 years ago, when people thought that six transistors was the most they would ever squeeze onto a chip.
Ready for more? Take a deep breath:
Both developments use a concept called a "photonic crystal". This involves micromachining arrays of very small holes into a planar wafer. The diameter of each hole (they're usually circular) is actually smaller than the wavelength of light used for telecommunications. In this realm the physics gets complicated and all sorts of weird things start to happen.
For a start, it's possible to make a so-called "perfect mirror." This is formed from a regular hexagonal array of holes, giving the appearance of a honeycomb. In addition, the holes must be identical in size. In practice, the holes are quite shallow because they are drilled into the surface of a thin wafer, but this doesn't stop them from working. The perfect mirror can reflect light incident from any angle in the plane of the wafer (unlike the edge of a standard waveguide, the properties of which change with angle) for a selected band of wavelengths.
Take two perfect mirrors, put them side by side with a narrow gap in between, and the result is a waveguide. Unlike today's waveguides, which leak out light when angles get tight, a photonic crystal waveguide can guide light around 90° bends with zero losses -- at least that's the theory. But making and measuring these things has proven fairly challenging.
To make this idea work in reality, the light has to be confined in the vertical direction; otherwise it gets radiated out of the top of the wafer. That's because the photonic crystal is a two-dimensional structure -- it controls light in the plane of the wafer, but can't stop it from escaping out of the top or bottom.
Shawn Lin and his colleagues at Sandia are the first to create a waveguide using photonic crystals and prove that it can transmit light of 1.5 micrometers wavelength with no losses. They used oxide layers above and below the waveguide to prevent light from leaking out of the top or bottom of the wafer. (The work is reported in the October 26 issue of the journal Nature and was co-authored by researchers from Massachussetts Institute of Technology).
Sandia's results follow hard on the heels of a significant development at Kyoto University. Susumi Noda and his colleagues have been working with microcavities, which are defects in the otherwise perfect pattern of the perfect mirror. Removing or changing the dimensions of a single hole in the hexagonal array creates the microcavity, which can capture and store light of a specific wavelength. Noda has used this idea to make an ultrasmall integrated device that can drop individual wavelength channels. The work is reported in the October 5 issue of Nature.
Noda's device comprises a photonic crystal waveguide adjacent to two microcavities. Light coming through the waveguide can be selectively tapped off by a microcavity, which then emits the light from the top surface of the wafer. All that's needed now is a way to funnel this light into an optical fiber and, hey presto, it's a two-channel optical add-drop multiplexer (OADM) measuring less than 10 micrometers in length. The frequency spacing for adding or dropping channels in such a device can be controlled at will, Noda claims in his research paper.
"This is very interesting as a demonstrator, though it's still quite far from being a commercial device," says Gareth Parry, professor of optoelectronics at Imperial College in London. Parry's research group is working on photonic crystal devices with Marconi Caswell Ltd., the R&D arm of Marconi Ltd. Parry, who has visited Noda in Japan and seen his OADM device, points out that there are still problems coupling light into the waveguide.
Clearly, the results from Kyoto need to be combined with the results from Sandia, and then two key pieces of a jigsaw puzzle will drop into place. Is this the beginning of true optical systems on a chip?
-- Pauline Rigby, senior editor, Light Reading http://www.lightreading.com