There is no question that fiberoptics have revolutionized the telecommunications industry, but optical fibers are by no means "perfect." According to John Joannopoulos, the Francis Wright Davis professor of physics at Massachusetts Institute of Technology (MIT; Cambridge, MA), some fundamental problems persist (besides breakage). One involves the difference in the polarization of the light entering and the light exiting the fiber, which can cause problems when the light is coupled into polarization-dependent devices. Another deals with light propagation by total internal reflection, which prohibits light from traveling through sharp bends in the fiber without high scattering losses. Although this characteristic may be beneficial to someone who wants to "tap" the wire, it limits the scale of possible miniaturization for optical devices.
JThe red and yellow bands indicate guided modes confined to the coaxial region of the omniguide. The dashed lines indicate modes with less than 20% localization within the coaxial region. Light-blue regions correspond to modes for which light is allowed to propagate within the dielectric mirror, and darker regions correspond to modes for which light is forbidden to propagate within the mirror. Inset: the electromagnetic field for a light mode (m=0) traveling through an all-dielectric coaxial cable.
oannopoulos and colleagues from MIT's center for material science and physics department think these issues are resolvable with a new waveguide design that combines features of metallic coaxial cable and dielectric waveguides.1 The coaxial omniguide consists of a coaxial waveguiding region with a low index of refraction bounded by two cylindrical, dielectric, multilayer, and omnidirectional reflecting mirrors. With this configuration, light is not limited by total internal reflection and can be guided around very sharp bends with a radius of curvature as small as the wavelength of the light. The all-dielectric coaxial waveguide also has the potential to overcome problems of polarization rotation and pulse broadening when transmitting light. In theory, it can be designed to support a single fundamental mode that is similar to the TEM mode of metallic coaxial cable, with features including radial symmetry and a point of zero dispersion (see figure).
Until recently, a common assumption in the research community was that an all-dielectric coaxial waveguide could not be designed to support a TEM-like mode. The key to doing so in the MIT waveguide model is the mirror technology, which is based on the "perfect" mirrors developed at the university in 1998.2 Like conventional metallic mirrors, these mirrors reflect light from all angles and polarizations, but they do so with the low-loss characteristics of dielectric mirrors.
The basic composition of each perfect mirror is a periodic multilayer planar structure consisting of alternating layers of film with low and high indexes of refraction. This structure essentially is tunable to produce a range of frequencies at which incoming light from any direction with any polarization is reflected. In line with this, there is a frequency for each angle of incidence and each polarization for which the phase shift is identical to metal. This "tunability" led the MIT researchers to consider the possibility of replacing the metal in coaxial cable with omnidirectional mirrors.
With the new waveguide design come a variety of related concerns, one being the determination of a method to couple light into the coaxial cable. A possible solution may be to use an omniguide with a very thin core that increases gradually to match the core of the coaxial configuration. Because the electromagnetic mode of a laser source can have a TEM00 mode, Joannopoulos and colleagues theorize that there should be efficient coupling into the TEM-like mode of the coaxial omniguide. Commercialization efforts for the new waveguide are under way at OmniGuide Communications Inc. (Cambridge, MA).
Paula Noaker Powell
REFERENCES
- M. Ibanescu et al., Science 289, 415 (July 21, 2000).
- Y. Fink et al., Science 282, 1679 (Nov. 27, 1998).