Building on its wide-bandwidth, wide-angle, high-reflectivity dielectric mirror, a research team at Massachusetts Institute of Technology has designed a dielectric waveguide with several advantages over optical fiber. It will be smaller, and simulations indicate that it will have lower loss, higher bandwidth and better polarization maintenance than standard, single-mode optical fiber. Results of the team's work also suggest that the design will support much tighter cable bends without significant propagation losses. A higher-order mode simulation of a new optical cable shows the supported electromagnetic propagation pattern. The cable could eliminate the need for amplifiers and repeaters in optical communications. Courtesy of OmniGuide Communications Inc. The dielectric cable is similar to radio frequency coaxial cable in cross section. The outer and inner metallic conductors of the coaxial cable, however, are replaced with dielectric reflectors that prohibit the propagation of light within a specific wavelength range. Because propagation is not supported within the dielectric materials, the reflectors completely reflect incident light of these wavelengths. And because confinement in the cable is the result of omnidirectional reflection rather than the total internal reflection of traditional fiber, there are no constraints on the angle of incidence. This enables the cable to support propagation even when it is tightly bent. A tight bending radius should make it possible to route optical signals around a structure as small as a microchip. The possible applications are being explored by OmniGuide Communications Inc. of Cambridge, which was created by members of the MIT group. Uri Kolodny, director of marketing, said OmniGuide's development efforts are focused on producing a simpler cable. The company's first product will be a replacement for single-mode optical fiber that is composed of a hollow-core cylindrical structure with its inner walls constructed of an omni-directional reflective stack. "Because the propagation is in air," Kolodny explained, "we are not limited by effects of the propagating medium." For example, laser damage at the fiber entrance is not a significant issue, and there are no nonlinear material effects, so much higher power densities can be launched. The individual wavelength channels can be more closely spaced because of the absence of four-wave mixing, Raman scattering and other nonlinear effects that are intrinsic to the propagating media of fiber. Also, dispersion is significantly reduced. Combined with the effects of reflective confinement, the bandwidth is increased from 90 nm in current erbium-doped fiber amplifier transmission to as high as 800 nm in the new cable. Development is expected to take 12 to 18 months.