OXFORD, England — Photonic crystals that operate in the visible spectrum have been generated via a technique that is intrinsically fast and, if some materials issues can be resolved, relatively inexpensive. Researchers at the University of Oxford have used holography to create three-dimensional molds that, once inverted by being filled with high-index materials and the original structure scrapped, have a photonic bandgap useful for some optoelectronic devices.

Though the method has not yet been perfected, researchers are hopeful that holographic fabrication, with its inherent flexibility, will make the use of photonic crystals more practical.

The work was carried out in Oxford's physics and chemistry departments and led by Andrew Turberfield and Bob Denning.

"No other groups have used this method," Denning said. "The main competition is from self-assembly of spheres, or from patterned semiconductors in two dimensions. Self-assembly is slow, difficult to control and prone to the formation of random crystal defects. Two-dimensional patterning does not provide the full optical confinement offered by 3-D systems."

Richard De La Rue, of the Optoelectronics Research Group at the University of Glasgow, agreed that the work is potentially important. "The technique can be adapted to generate a range of crystal symmetries," he said, "and the beams involved in the holographic exposure process can be manipulated, in principle, to incorporate specific defect structures into the lattice, thus producing useful device-type features such as waveguide channels."

But both Denning and De La Rue warned there many problems to overcome before commercialization is possible. According to Denning, three main issues must be resolved.

First, he said, is the need to "eliminate sources of nonuniformity in the pattern formation." Next, the team must "work out a method for templating a suitable high-refractive-index filler, which must be reasonably optically homogeneous." Finally, he said, they must create methods for the controlled introduction of defects, in order to create microcavities and waveguides.

Photonic crystals are structures that restrict the propagation of particular wavelengths of light through destructive interference. These forbidden wavelengths, as in electronics, are defined as being within the bandgap of the structure and are often also known as photonic bandgap materials.

The destructive interference is achieved by creating what is effectively a thick diffraction grating with three-dimensional symmetry and a very high-refractive- index contrast ratio. The result is similar to the multilayer mirrors used in all areas of optoelectronics, but in three dimensions instead of one.

Though conceptually simple, photonic crystals have been difficult to fabricate for the visible spectrum because the feature sizes involved have to be on the order of the wavelengths of visible light if they are to work. Though IC microlithography can now reach the required resolution, the crystals produced are limited in depth, which restricts their effectiveness.

Other disadvantages of conventional optical techniques include having to build up the pattern layer by layer or even point by point, which leads to insufficient flexibility in lattice design.

The new technique does not have those restrictions. Instead of building up a series of patterns on a substrate, the entire three-dimensional crystal is recorded at once by interfering four coherent laser beams. By choosing the appropriate direction, intensity and polarization relationships among them, the structure of the resulting interference pattern can be manipulated as required.

In the Oxford experiments, four laser beams are generated from a single source and are then individually manipulated using half-wave plates. The geometry of the incoming beams makes it possible to define the crystal pattern very precisely: Researchers can choose, for instance, between face- or body-centered cubic structures, various fill factors and so on. The hologram itself is recorded using a single 6-nanosecond pulse that exposes a thick layer of photoresist.

Using a short pulse has two major advantages in this setup. First, it prevents mechanical stability from being a problem during exposure. Second, it means that the slow chemical changes taking place inside the resist do not have enough time to change the optical characteristics of the material while the light is propagating through it (so distorting its path).

After exposure, the resist is baked to complete the polymerization of the exposed material. The rest is then dissolved to reveal the crystal. Unfortunately, this structure alone is not enough to make the material into a photonic crystal: The refractive index of the photoresist is just 1.6, too low to be of use.

"The most obvious approach in seeking full photonic bandgap behavior," said De La Rue, "is to 'invert' the photonic crystal by filling in the spaces between the balls of photoresist produced in the holographic exposure and development process."

The Oxford group has demonstrated that an inverted structure can be formed in polycrystalline titanium dioxides. This approach was successful, but, De La Rue points out, "They were not the first people to demonstrate this type of inversion process; it was first done with self-organized crystals."

This last step has not yet been perfected: Considerable shrinkage distorts the shape of the final crystal with respect to the mold. However, other processes for transferring the crystal shape into a high-index material are available, and the Oxford researchers are currently pursuing these. Also, even the distorted titania structure does show the hallmarks of a photonic crystal, with diffraction patterns through it exhibiting characteristic "Kossel lines."

Denning said the inversion problem is probably the biggest obstacle to using holography to make practical devices. However, if it can be solved, the technology could lead to the fabrication of a whole new family of high-density all-optical integrated circuits and low-threshold sources.

"Three-dimensional photonic crystals will firstly be applied as novel forms of a tunable solid-state laser rod, excited by photo-pumping," said De La Rue. "This application does not require full photonic bandgap behavior, although that would be a useful and a fundamentally interesting enhancement."

This is crucial because, with a refractive index of 2.6, titania is just at the edge of being a photonic bandgap material for visible wavelengths. "The structures made by the Oxford group could become the preferred approach in this context," he said, "if they can be shown to produce better control and more reproducible behavior."