SAN JOSE, Calif. Motorola Inc. has backed away from its work to commercialize gallium-arsenide-on-silicon wafer technology, quietly shuttering a subsidiary launched in 2001 to lead the effort.
Researchers at the unit, Thoughtbeam Inc., are apparently moving back into R&D mode, sources said. But it is not known whether a drop in the price of GaAs wafers, technical difficulties or a combination of the two forced the change.
Motorola formed Thoughtbeam in November 2001 to develop, sell and license its GaAs-on-silicon technology which arose from research work at Motorola Labs for use in high-speed wireless and optoelectronic devices. At the time it was said the technology would open the door to less-expensive optical communications, high-frequency radio devices and high-speed microprocessor-based systems all in demand due to the broadband communications bubble.
Motorola veteran Padmasree Warrior was named the general manager of Thoughtbeam. But on Jan. 1, she was promoted to the position of chief technology officer at Motorola, and the company's online announcement of the appointment made no mention of a replacement at Thoughtbeam. A Motorola public-relations person confirmed that Thoughtbeam was no longer in existence.
Motorola signed its first licensing agreement for the production of GaAs-on-silicon wafers with IQE plc. The U.K-based company received a cash injection in return for equity, and was contracted to ship evaluation wafers back to Thoughtbeam in the first half of 2002.
"We did supply GaAs-on-silicon wafers but by the time we did, the GaAs wafer price had dropped, so as to make them uncommercial," said a spokesman for IQE. "Motorola suggested we start to look at indium-phosphide-on-silicon and gallium-nitride-on-silicon."
One driver for GaAs-on-silicon wafers at least while demand for GaAs exceeded supply in 2001 was the prospect they could be made at lower cost and in larger sizes than conventional GaAs wafers, said analyst Stephen Entwistle, vice president of the strategic-technologies practice at market research firm Strategy Analytics Ltd. (Milton Keynes, England). Another was the possibility of integrating light-emitting GaAs on the same substrate as silicon, he said.
The IQE spokesman said the company has continued to research compound semiconductors-on-Si, but that at the moment there is no urgency to that effort due to the weak markets.
When asked about the status of Thoughtbeam, the IQE spokesman said, "The people are still around, but they are back in research. I think Thoughtbeam itself has been a victim of one or more of the many rounds of cuts at Motorola. There's been a lot of scaling down. We've been working with them in R&D, although I think it is not seen as a priority."
IQE made "small volumes of wafers with quality GaAs on top, capable of emitting light and for making power amplifiers for cell phones," the spokesman added. "But the development went on the back burner."
Technical difficulties as well as economic ones may have beset Thoughtbeam. Analyst Entwistle said it was possible that prior to the formation of Thoughtbeam, Motorola Labs had made power amps only in MESFET technology, not using heterojunction bipolar transistor technology. "There have been rumors they were having problems doing that making HBT circuits," he said. "Of course, mobile-phone demand is for HBT power amps."
The issue of constructing power amplifiers using MESFET vs. HBT transistors hinges on the quality of the GaAs-on-Si substrate that could be achieved.
"The defect density of the substrate has a strong impact on the yield of the devices. MESFETs are more tolerant of defects in the lattice than HBTs," said Entwistle. Entwistle said he believes the GaAs-on-Si technology could come back, once the price of gallium rises again, as it almost certainly will.
"Gallium is now about $500 per kilogram but has been as high as $1,500/kg," he said. "As gallium mining is controlled by a few suppliers, and the raw gallium cost is 25 percent of GaAs wafer cost, when the price rises again, you don't want to be shipping a 625-micron wafer-thickness depth of GaAs when a wafer only needs to contain gallium in the top few microns."
But those prospects also depend on whether defect densities have been, or can be, taken sufficiently low to allow HBT fabrication, he added.