IBM Moves Into Integrated Optics
Under the deal, IBM's micro-electronics division will license its silicon oxynitride (SiON) process technology to Kymata in return for a 3 percent equity stake in the startup. The duo plan to jointly develop a range of prototype optical devices based on this technology, targeting Q1 2001 for first samples. In addition, IBM and Kymata plan to develop their own products using the SiON process.
The deal also enables IBM to buy wafers from Kymata under favorable terms. That's a key point for the giant chip maker because it gives it a leg up in this new marketplace, according to Ron Soicher, IBM Micro's director of technology strategy.
By combining forces in this way, IBM and Kymata can make a much wider range of optical devices than either could do with its own technology, according to Brendan Hyland, Kymata's president and CEO.
Until now, Kymata has been making optical chips using a process called flame hydrolysis of silica on silicon. Meanwhile, IBM has developed SiON in its Zurich Research Laboratory. The technologies are complimentary, says Hyland.
"If you look into the history of Kymata, you'll see that our core technology -- flame hydrolysis of silica on silicon -- was dual-sourced from BT [British Telecommunications PLC (NYSE: BTY)] and Glasgow University." Now, he says, the same strategy is being extended to embrace IBM's SiON. "There are advantages in dual-sourcing our technology. Experience has shown us there's rarely one right answer." Right now, IBM and Kymata are the only companies openly planning to commercialize SiON technology. Industry experts say that only one other company, Total Micro Products BV, has expertise in SiON, but it was recently acquired by Kymata (see Kymata Buys MEMS Maker).
So, what's so special about SiON? Basically, it's a material with a high refractive index that's used to form the core of optical waveguides. As a result, it is possible to produce optical circuits with tighter bend radii, leading to chips that are a factor of three or four times more compact than those produced in silica. This makes it possible to manufacture certain devices that it's not possible to make in silica. The drawback is that it requires new procedures in the manufacturing plant, for example, for attaching fibers to the chip.
Some of the devices on the shared roadmap include solid state tunable filters (see Tunable Filters Go Solid State) and variable optical attenuators, which are used to balance the power of all the channels in a DWDM system. Optical switches that exploit the thermo-optic effect are also a possible target. Soicher says that IBM Micro has already demonstrated a seven-channel gain equalizer and a basic switching element. In the future he reckons that IBM could make 16x16 and possibly 32x32 switches, putting it among the leaders in thermo-optic technology (see Optical Switching Fabric and Lynx Claims Optical Switch Advance).
-- Pauline Rigby, senior editor, Light Reading http://www.lightreading.com