LONDON – Foundry chipmaker United Microelectronics Corp. is sampling a customer with a 28-nm application processor amid indications that the roll out of the leading-edge process technology is going better than expected, according to analysts at financial services firm Nomura Equities Research.
The application processor is believed to be the OMAP5, which is being manufactured by UMC (Hsinchu, Taiwan) for Texas Instruments, the analysts said in a note to clients.
They predicted that small volumes of the chip would be produced in the second and third quarters of 2012, which puts UMC about four quarters behind its local rival Taiwan Semiconductor Manufacturing Co. Ltd.
However, the analysts added that this is two to three quarters earlier than they had expected UMC to produce 28-nm silicon and that this could be the result of "smoother-than-expected R&D and pilot run," and "not much delay from design and process re-work."
Both Globalfoundries and TSMC are reported to have had problems with the ramp up of their leading edge manufacturing process technologies, although TSMC has denied this, saying that yield improvement on its 28-nm processes is on course.
A spokesperson for UMC said that 28-nm revenue was expected to contribute 5 percent of revenue in 2012, with volume production of the 28HLP process, which is based on a polysilicon/silicon oxynitride gate stack, scheduled for 3Q12. The 28HPM process, which includes a gate-last high-k metal gate (HKMG) gate stack is due to start "risk production" in the second half of 2012, the spokesperson said.
The spokesperson added that UMC has more than 10 customers and tape-outs at 28-nm and that engagements are for both poly-SiON and HKMG variants of the process.
Not doing the HKMG certainly makes things easier. If there are a good number of customers that can live with the leakage then this could be good money for UMC. It all depends on how much volume they can get.
Interesting to note that UMC had "smoother-than-expected R&D and pilot run," is it because it took precaution while building its 28nm process or was it plain lucky ? I am sure it would have learnt some lessons from TSMC/GF's mistakes when implementing 28nm process.
Does TI not care about the leakage / power dissipation for their OMAP built by UMC with ye olde polysilicon/silicon oxynitride gate stack rather than HKMG like TSMC / GloFo would have at 28 nm LP ? Perhaps this is just a stop gap / get a foot in the door for a OEM ( A?) who can't wait till UMC gets its HKMG ready at 28 nm.
Well, their 28nm is a conventional gate stack process (poly Si gate with SiON gate oxide) and a far more sophisticated high-k dielectric and metal gate process which is the main stream for TSMC and Intel, IBM.
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.