| SAN JOSE, Calif. — Chartered Semiconductor, which is qualifying 90-nm products this quarter as part of its process alliance with IBM, expects pilot work to begin on 65 nm later this year with production runs beginning in 2006, its chief executive said. |
Chia Song Hwee said the timing between the 90- and 65-nm nodes is close, driven in part by Infineon Technologies AG, which, with Samsung, has joined IBM and Chartered to develop a 65-nm process.
"Infineon is driving the timeline of low power 65 nm because they have customers wanting it," he said. "They have product that's designed. So hitting the timeline is critical."
Hwee made his comments in an interview during a technology forum his company hosted here to discuss its 90-nm platform, which it began developing in late 2002 with IBM. The all-copper process includes up to nine layers of metal, low-k dielectric (k=2.9) and an optional triple well CMOS technology on bulk substrate.
The nominal process was qualified in May, and production qualification is expected to be completed this month, Hwee said. Low-power and high-performance processes are scheduled for qualification in the fourth quarter, he added.
The process, which was developed by a joint IBM-Chartered team at East Fishkill, N.Y., has largely been transferred to Chartered's Fab 7 in Singapore, according to David Vigar, IBM's director of technology development. That facility can accommodate up to 30,000 wafer outs a month in its 200,000 square-foot cleanroom.
While the commercial success of the joint venture remains to be seen, the deal was clearly a technology win for Chartered, which had struggled with profitability and direction in recent years against beefier Taiwanese rivals UMC and TSMC. Some observers suggest that the alliance actually led Infineon to bail out of a technology deal with UMC, in favor of the two for the 65-nm node.
IBM itself had some technology issues as well with UMC, particularly on low-k dielectrics.
"When we first talked to them [about 90 nm] we had a lot of doubts," Hwee said. "Why so many deals didn't work in the past and what do we need to do differently to make this be successful? Two years later, we proved that it could work."
"At the end of the day, I need IBM and IBM needs me," Hwee said.