"In a simple planar geometry, the structures can be used as filters; as mirrors to improve the performance of devices such as vertical-cavity, surface-emitting lasers; or as optical switches and shutters," said Dmitry Chigrin, a researcher at the University of Essen (Essen, Germany). "By rolling the structures into hollow fibers or tubes, the coatings can be used as inside walls of high-finesse waveguides and microcavities."

Chigrin's research group, working with research teams at the National Academy of Sciences and Belarusian State University (Minsk, Belarus), has both theoretically and experimentally verified a stacked-dielectric structure for optical frequencies. The reflective capabilities of dielectric stacks have simultaneously been discovered and explored by researchers at the Massachusetts Institute of Technology (Cambridge, Mass.), where an infrared version of a device has been demonstrated, as well as at the University of Bath (England).

Sergey Gaponenko, head of the research group at the Institute of Molecular and Atomic Physics of the Belarus National Academy of Sciences, considers the new structures important for both optics and optoelectronics. "This is an excellent example that new discoveries can be found in the very classical and canonical field of wave optics," he said.

The main advantage of the wide-angle reflectors, according to Gaponenko, is that they do not dissipate energy, whereas metallic mirrors both reflect and absorb light. "Metallic mirrors heat up and are destroyed when exposed to some high-power fluxes," he explained.

The new approach to building reflective surfaces is similar to Bragg reflectors, which similarly comprise dielectric stacks, in a technique that has become essential for confining light in the reflective cavities of microscopic diode lasers.

The new structures are composed of quarter-wavelength-thick layers with differing refractive indices. Light at the design wavelength enters the stack from above, at right angles to it. As it proceeds, it is reflected from each of the layer interfaces as a result of the refractive-index differential.

The reflection is reinforced by the optical-path differences among the reflected beams. Generated by the different distances that each beam travels before being reflected, as well as by phase changes introduced at some of the layer boundaries, the path differences cause the reflecting beamlets to interfere constructively. Thus, the beam as a whole is reflected.

The quarter-wavelength geometry of the structures is vital for the scheme to work properly, since the reflection drops off radically as soon as the incident angle is changed. That has meant that engineers have had limited options for applications where metal mirrors are not suitable but wide-angle reflection is necessary.

The Belarusian/German team designed its dielectric stack as a one-dimensional photonic crystal — a structure in which the propagation of particular electromagnetic waves is not allowed, because of forbidden bandgaps that are analogous to electron bandgaps in semiconductors. The bandgaps can be opened or closed based on the stack design. For light within the bandgaps, beams coming from any possible angle of incidence will be reflected.

According to Chigrin, the first photonic crystals produced in Minsk were accidents. "I was working with Andrei Lavrinenko at the Belarusian State University when, at the very beginning of 1998, we realized that we had found something new," he said. The colleagues were working with anisotropic one-dimensional photonic crystals and, occasionally, they would hit on the right parameters — just a large enough index contrast between the layers — to create the photonic-bandgap effect.

"I was really surprised to notice the direct evidence that such a lattice can reflect light at all angles," said Chigrin. "It was amazing and very confusing because the effect was long thought to be impossible. Almost any review of work in the field claimed it was impossible to achieve omnidirectional total reflection in low-dimensional dielectric structures. So, at first, we just checked the simplest case — a dielectric lattice of isotropic layers — and it worked."

Over the next few months, he said, the researchers probed to find holes in their theory, not believing that so straightforward an effect could have been overlooked for so long.

Crystal demos

"In May we presented our work to Sergey Gaponenko's group seminar," said Chigrin. "He had agreed with our theory and proposed to carry out an experiment to demonstrate it."

Several photonic crystals have subsequently been simulated and built. One 19-layer device has demonstrated total omnidirectional reflection in the low 600-nm (red) range.

According to Jonathan Dowling, senior research scientist at the Jet Propulsion Laboratory (Pasadena, Calif.), the experiment is not as straightforward to do as it sounds. "As a general rule, at shorter wavelengths, it is harder to find materials with a large index," he said. "Not just any two dielectrics will stick together to form robust layers of uniform and controllable size. Here, there should be a good match on the chemical properties, such as lattice structure and dimensions.

"Index considerations first limit the choice of dielectrics, and then the fabrication constraint is applied to the potential candidates."

As a rule, he said, "fabrication of photonic bandgap structures increases in difficulty as the wavelength decreases. So the Belarus group is to be congratulated on pushing the technology into the visible regime."