Intel Corp. (Nasdaq: INTC) thinks long-wavelength VCSELs (vertical-cavity surface-emitting lasers) are hot, judging by the investment it's made in Phoenix-based startup Lytek Corp. (see VCSEL Startup Gets $3.8M).

Lytek was founded in May 2000 by Arizona State University professor Yong-Hang Zhang. In June, it received seed funding of $1.5 million from Intel Capital, and yesterday it announced an initial closing of its B round at $3.8 million. Wasatch Venture Fund, an affiliate of Draper Fisher Jurvetson, led the round, with Intel Capital topping it off.

At the time of the seed funding, Lytek was the only VCSEL company backed by Intel, says Zhang. Since then, however, Intel Capital has also put cash into Nova Crystals Inc., a startup making what it calls "PICSELS" (photonic integrated circuit surface-emitting lasers -- see Nova Crystal's Little Secret). Zhang views this as confirmation of the direction his company's taking.

At short wavelengths, VCSELs have already displaced all other types of laser in datacom applications, thanks to the quality of light they emit and certain manufacturing advantages (see Laser Blazers). They are set to repeat the feat at 1310 nanometers, and possibly 1550nm, if the materials technology can be tamed.

Until recently, that's been a fairly big "if." Ideally, the active (light-emitting) region of the laser and its multilayer mirrors should be based on the same material system so that it is possible to "grow" the entire structure in a single process. But materials that are good at light emission are generally poor for making multilayer mirrors, and vice versa. Scientists are still searching for the perfect combination of materials properties to get high performance from their lasers.

Lytek says it has a solution for making 1310nm VCSELs. It's licensing a technology -- developed in Zhang's research group at Arizona State -- for creating quantum wells (efficient light-emitting structures) using antimonide materials.

According to Zhang, this is a new way of making 1310nm VCSELs. Other approaches include wafer bonding, quantum dots, and nitride-based materials (see Laser Blazers, page 7 for details). All approaches have their upsides and downsides.

The big advantage of using either nitrides or antimonides in the active region is that these materials are compatible with established 850nm VCSEL processing, says Zhang. In both cases, the mirrors are made from the well-known combination of gallium arsenide and aluminium gallium arsenide.

But at the end of the day, antimonides have the edge over nitrides, Zhang contends. That's because they have a lower threshold current. Any current that's put in below threshold, gets wasted as heat, he says, making the laser less efficient.

Quite a few companies are betting on nitrides, however. They include Cielo Communications Inc., Infineon Technologies AG (NYSE/Frankfurt: IFX), and Picolight Inc.. Cielo is the only company to have launched a product so far, but Picolight says it's not far behind (see Cielo Pushes the Limits on Lasers and Picolight Attacks Access).

Lytek hopes to get sample products ready and move into manufacturing by the end of 2002. It's planning to keep its operation small and streamlined by outsourcing everything, including the manufacture of the VCSEL chip itself. Only 10 to 15 employees will be needed to run the show.

Within the next few months, Lytek is planning to move into its own facility, which will have a clean room and molecular-beam epitaxy (MBE) equipment for development purposes.

— Pauline Rigby, Senior Editor, Light Reading
http://www.lightreading.com