Light gets the bends in a photonic crystal.

Normally a straight arrow, light usually doesn't turn a corner. Now, researchers have worked out a novel way to make a beam change direction. This trick isn't done with mirrors, the usual way to send light on a new path, but with a photonic crystal, a structure that excludes light in a chosen range of wavelengths.

Computer simulations performed by John D. Joannopoulos and his colleagues at the Massachusetts Institute of Technology show that a specially constructed photonic crystal could steer light around a sharp bend with nearly perfect efficiency. In contrast, a simulated material with reflective properties similar to those of fiber-optic cables can guide only about 30 percent of the light, at best, around a corner. Their analysis appears in the Oct. 28 Physical Review Letters.

The calculations are modeled on a photonic crystal first built a few years ago (SN: 9/25/93, p. 199). The simulated device consists of a two-dimensional array of equally spaced gallium arsenide rods, whose spacing and size determine a range of wavelengths to which the crystal is opaque. Removing a row of rods forms an empty channel, or waveguide, through which light of the chosen wavelength can travel freely.

Moreover, turns and corners can be built into the waveguide. The MIT group's analysis shows that 100 percent of the light can swing around a slightly rounded corner and 98 percent can sweep around a 90#161# bend.

"I like the analysis and the way they presented it," says Costas Soukoulis, a physicist at Iowa State University in Ames, "but I'd like to see it done experimentally."

Waveguides made from photonic crystals would be promising components for electronic devices driven by light instead of electrons, such as optical computer chips. "It's a drive toward integration-making circuits as small as possible," Joannopoulos says.

Photonic crystals can also trap light in small areas, opening up the possibility of constructing miniature lasers.

The photonic crystal waveguide combines the best features of two existing technologies. Simple metal waveguides can direct microwave radiation around tight corners efficiently, but they don't work at visible or infrared wavelengths. On the other hand, waveguides made of insulating materials, like fiber-optic cables, carry visible light well but not around sharp corners.

One issue that will affect whether photonic crystals can be practical waveguides, Soukoulis says, is the effective velocity at which light moves around the bend-which might be quite slow. Because the light creates a standing wave by bouncing back and forth as it travels along the waveguide, it follows a circuitous, time-consuming path.

Joannopoulos and his group plan next to carry out computer simulations to predict what will happen in three dimensions and then to demonstrate the effect experimentally. Determining how the system would work in three dimensions is critical, Soukoulis says, because if the incident wave hits the channel at even a slight angle, most of it might go out of the plane of the crystal and not along the waveguide.