Talking about a revolution

29 Aug 2002

Unlike many people, Eli Yablonovitch does not believe that his invention - the photonic crystal - should be used to make all-optical circuits. He tells Nadya Anscombe why his new company will integrate electronics and optical components on a chip.

When Eli Yablonovitch invented the photonic crystal, many people did not realize that a revolution was about to start. After a slow beginning and much scepticism from the research community, the field of photonic crystals, or photonic band-gap structures, is now one of the most exciting and fast-growing fields in optoelectronics.

Researchers across the globe are now designing, redesigning, modelling and perfecting those tiny structures that Yablonovitch first described in a paper in 1986. Back then, making a photonic crystal involved drilling holes in dielectric plates and using trial and error to find the right structure.

Yablonovitch admits that many mistakes were made before finding a structure that worked. "My group eventually made the first successful photonic band-gap crystal in 1991 using a variant of the diamond structure," he said. This structure is now known as yablonovite.

Today the photonic crystal promises to revolutionize optoelectronics by enabling nanoscopic lasers, exotic fibres, ultra-efficient LEDs and photonic integrated circuits. Yablonovitch is particularly interested in photonic integrated circuits and has set up a company to commercialize them.

Called Luxtera, the firm has received $7m in first-round funding and plans to integrate optical and electronic elements onto a chip for dense wavelength-division multiplexing applications.

Size matters

Luxtera's technology is based on photonic-crystal research carried out by Yablonovitch, at the University of California at Los Angeles, US, and microlaser research carried out by Axel Scherer from the California Institute of Technology, Pasadena, US. Scherer's group has used photonic crystals to make some of the smallest ever lasers with volumes of only 0.03 µm³.

The company will face many challenges before it can put products on the market, but Yablonovitch is confident that this will happen "sooner than most people think". He says that the main challenge is coupling light into the circuit because "an entire nanophotonic integrated circuit with all of its components will fit inside the core of a singlemode fibre".

He said: "We need a way to mode-match from the giant core of a singlemode fibre to the tiny waveguides in a high-index-contrast photonic circuit."

In his nanophotonic integrated-circuit design not all of the components are made from photonic crystals and not all of them are optical. "An optical transistor is something that has been sought for years, but I think that it may be a misguided goal," said Yablonovitch. "It is easy to make an electronic transistor - why not use that in conjunction with nanophotonic components?

"The vision that I have involves using nanophotonics for transmission and electronic components for logic operations. There are many of us that have given up on the optical transistor - we are satisfied with the conventional CMOS device."

Yablonovitch also thinks that photonic crystals are not necessarily the best way to bend light round corners - a subject that many scientists are working on. "Using a photonic crystal is not the only way. A 45° mirror will easily bend light through 90°."

His vision of an integrated optoelectronic circuit can be implemented using 2D photonic-crystal structures and he does not see the need for 3D designs. He said: "There is still no convenient way of making 3D photonic crystals. It seems we can achieve most of what is needed by using 2D thin-film photonic crystals. While 3D photonic crystals make interesting research, integrated circuits can be made using 2D structures."

The old way is best

And while many groups around the world are coming up with novel manufacturing methods, Yablonovitch believes that the established technique of lithography is the answer. "Photolightography can do it now. It is a low-cost major industrial process so why do anything else?"

This is why Luxtera will be collaborating with major semiconductor firms to make its products. "I am optimistic; I don't think that [commercialization] will take a long time. The technology has caught up with the idea. You can already make subwavelength structures using lithography and manufacture photonic-crystal circuits in any semiconductor fab."

He points out that he is not the only person with this opinion. "UK start-up Meso Photonics is also optimistic about the commercialization of photonic crystals." Yablonovitch is also keeping his eye on Crystal Fibre in Denmark and Blaze Photonics in the UK, which are commercializing photonic-crystal fibre. "These two optical-fibre companies are interesting and exciting," he said. "They are the most advanced in the commercialization process."

So how does it feel to see so many groups across the world developing an idea that started in his laboratory? "I feel gratified, although it took much longer than I expected. When I published my first article, I expected everyone to see my idea and say it is obvious. But it took a while to convince people."