Photonic crystals head toward the marketplace

25 Oct 2002

Researchers developing photonic crystals are breaking free of academia to launch next-generation components and fibres into the commercial marketplace. Jeremy Mills assesses the work of the new-entrant suppliers and their prospects for survival.

From Opto & Laser Europe November 2002

The term "photonic crystals" conjures up images of something exciting and exotic, beautiful rather than practical, and closer to the world of diamond-cutters in Amsterdam than the practicalities of optical communication. Yet researchers in more than 100 academic institutions are currently investigating whether the technology could be exploited in fibre-optic networks. More significantly, 10 start-up companies have raised $110m (€112m) to create commercial products based on photonic crystals - with at least half of that money secured in the cash-constrained climate of the past 10 months.

Pioneers of the technology argue that photonic crystals will revolutionize the field of optics in its broadest sense, with applications ranging from optical memory and quantum computing to photonic integrated circuits smaller than the core of singlemode fibre. But others are more sceptical, dismissing the technology as just another laboratory curiosity that will produce nothing more than conference papers and slick presentations.

"People expected holography to change the world," said Jim West, a photonic-crystal researcher at US-based fibre manufacturer Corning. "But right now the only place you'll see holograms is on your credit card." What is now clear is that the pressure is growing for the start-ups to prove that their esoteric technologies can deliver practical products at the right price.

Back to the beginning Although the first academic papers on photonic crystals appeared 25 years ago, the current surge in activity can be traced back to pioneering studies by Eli Yablonovitch at the University of California in Los Angeles, US, in 1987. He speculated that a bandgap could be created for photons in the same way that a semiconductor possesses an electronic bandgap. Yablonovitch proved his theory in 1990 by showing that microwaves would not propagate in a photonic structure created by drilling a 3D array of millimetre-sized air holes in a dielectric material.

By 2000, the height of the telecoms boom, photonic-bandgap materials had become hot property and the first start-ups were spun out of university research groups. These early innovators knew that photonic structures could confine light extremely tightly, allowing radiation to be guided and bent around sharp corners with virtually no energy loss. Such attributes make photonic crystals ideal for creating miniature optical circuits and highly efficient optical fibres that avoid the need for costly regenerators.

"Our circuits measure 5 x 5 mm, about the same size as a 90º bend in an existing AWG [arrayed-waveguide grating]," said Mike Jackson, chief technology officer at Canadian start-up Galian Photonics. Greg Parker, chief technology officer at UK firm Mesophotonics, concurs: "We are looking at size reductions of three orders of magnitude to reproduce an AWG in a photonic crystal."

But coupling fibres into such small circuits is not an easy task, particularly since photonic crystals are easier to produce in materials with a high refractive index. Both Jackson and Parker are rather coy on this problem, citing proprietary techniques and patent applications as reasons for not disclosing more details. However, Parker points out that Mesophotonics is using low-index materials to minimize the mismatch losses with optical fibre.

US start-up NanoOpto is taking a different approach, exploiting sub-wavelength elements perpendicular to the incident light, in the same way as a thin-film filter. "We achieve a ± 20° alignment tolerance, which gives our customers flexibility in design," said Hubert Kostal, the company's vice-president of marketing. He suggests that a Fresnel lens could be used to couple a fibre into an in-plane circuit.

The price is right Meeting cost targets means getting the manufacturing right. In this respect, the various photonic-crystal firms are pursuing notably different strategies. Some are aiming to capitalize on existing manufacturing technologies, while others are developing their own production processes. In both cases, the most popular material is silicon-on-insulator (SOI), with the hole patterns etched in a thin layer of silicon on top of an insulating silica substrate. A few other start-ups are looking at new materials that could be processed more quickly and cheaply than SOI.

Galian, along with US start-ups Luxtera and Clarendon Photonics, claims that its SOI-based technology is compatible with standard semiconductor processing techniques, and has outsourced manufacturing to specialist chipmakers. This approach yields layer thicknesses and aspect ratios for the holes within established limits, although Galian's Jackson acknowledges that the process needs a few small tweaks.

The difficulty with this solution lies in designing photonic devices that can be made with processes defined for the semiconductor world. This is a big challenge, since the design of photonic crystals is already a complex matter that requires powerful finite-difference, time-domain modelling algorithms running on clustered computers.

Elsewhere, NanoOpto has decided to develop a proprietary process that is optimized for the production of photonic crystals. A mould is used to imprint the circuit design into a polymer resist layer, and then the pattern is created in the silicon/silica layers with anisotropic reactive-ion etching. The downside is that the company must invest in costly processing equipment at a time when capital is in short supply.

An even more risky strategy is to explore the potential of alternative materials. Silicon Valley start-up NeoPhotonics, for example, is creating photonic structures in polymers, while Micro Managed Photons of Denmark has opted for patterns in thin gold films on a glass substrate. These unproven technologies raise major concerns about reliability and manufacturability.

Tune in, see the light Right now, any new technology for optical communication must be capable of producing tunable (i.e. dynamic) devices. At first glance, this seems a tall order for photonic crystals. University research suggests that photonic crystals could be made tunable by filling the holes with polymers or liquid crystals, but Galian's Jackson says that customers are "gun-shy" of new materials.

Instead, he believes that tunability can be achieved with existing SOI-based materials, even though they only yield small changes in refractive index. The trick, says Jackson, is to create designs that are sufficiently sensitive to the index change. Meanwhile, Nobi Kambe, vice-president of market development at NeoPhotonics, sees the addition of microelectromechanical systems structures as the solution to tunability, though he does acknowledge that this would lead to greater manufacturing complexity.

Differences in design and manufacturing philosophies are to be expected, but most firms agree that the greatest impact of photonic-crystal technology will be seen in integration. Hubert Kostal at NanoOpto reckons the real strength of the technology lies in its "ability to create an incredible range of optical functionality". For example, he believes that an "optical add-drop multiplexer the size of a sugar cube" is well within reach.

Greg Parker at Mesophotonics agrees that integration is the way forward. He points out that, unlike other technologies, photonic crystals add functionality without needing extra manufacturing processes, just a different pattern of holes. For Parker, the integration story is strengthened by the possibility of creating photonic-crystal lasers and gain-blocks within the same material to allow lossless subsystems.

Galian's Jackson also stresses the nonlinear properties of photonic crystals, suggesting that the technology could be exploited for wavelength conversion and optical regeneration as well as for photonic circuits. For some, however, that is a step too far, transporting the technology back into the realms of science fantasy.

Despite all the component-level activity, the clearest opportunity for photonic-crystal technology lies in the world of optical fibre. Traditional fibre confines light via total internal reflection at the boundary between the high-refractive-index core and the surrounding lower-index cladding. It's the core material that determines the fibre's loss, and the industry is already close to reaching the theoretical limit for silica.

Photonic-crystal fibres promise to reduce losses by one or two orders of magnitude by exploiting the optical bandgap to confine light within an air core. If such fibres could be produced in the same volumes as today's standard singlemode products, they would make US coast-to-coast optical transport possible without expensive regenerators and with fewer fibre amplifiers. Add to this the unexpected benefits - air-cored fibres don't share the nonlinear characteristics of conventional fibre; they exhibit almost no polarization-mode dispersion; and they are less susceptible to chromatic dispersion - and it's not difficult to see why people are excited about the possibilities of the technology.

At Corning, research programmes on photonic-crystal fibres have been in place for some years. Despite this, senior research scientist Jim West is under no illusions about the manufacturing issues that must be addressed before air-cored fibres are ready for deployment. He is also sceptical about potential applications in dispersion compensation, as he believes that existing fibre-based solutions already do a good job.

Reality check The big question now is whether all this technological promise can be turned into commercial reality. Some photonic-crystal firms launched their first product samples this year, but most are simple single-function devices that do not yet exploit a photonic bandgap. Next year should see a wider range of optical functions and the first integrated photonic-crystal products, though it will be 2004 before production-ready solutions are ready to run. Jaymin Amin, the director of optical systems development at Corning, remains cautious: "I don't think we'll see photonic-crystal components in the next five years," he notes, pessimistically.

Photonic crystals: the basics

Introduce a crystal-lattice pattern of tiny air holes into a dielectric material and you will see some interesting effects not readily explained by classical physics. Photons within a particular wavelength range simply will not propagate through the otherwise transparent material. This effect, known as a photonic bandgap, is best understood by analogy with the physics of semiconductors, where the interaction of electrons with a crystal lattice produces allowed energy states and a prohibited bandgap region.

To produce a photonic bandgap, the air holes need to be separated by a distance roughly equal to the photon wavelength divided by the refractive index of the dielectric, which implies that a hole spacing of about 450 nm is needed to achieve a bandgap at 1550 nm in silicon. The width of the bandgap depends on the contrast in refractive index between the two dielectric materials in the lattice, with a larger contrast yielding a wider bandgap.

As with semiconductors, introducing defects into the crystal lattice - in this case holes of different diameter - produces a band structure. Thus a photonic-crystal waveguide can be produced by omitting rows of holes (see picture). Extremely tight confinement within the waveguide region makes it possible to bend the light around sharp corners with low energy loss, enabling very small optical circuits.

Most researchers' attention is currently focused on 2D photonic structures, such as waveguides and planar circuits. But 1D versions, such as fibre Bragg gratings, are already in widespread use, while 3D photonic crystals, such as stacked planar structures and self-assembled nanostructures, remain the province of blue-sky research.