UMC Seeks to Shed Image as ‘Fast Follower’

By Mark LaPedus, SemiMD senior editor

Looking to shed its image as a “fast follower,” United Microelectronics Corp. (UMC) has stepped up its capital spending efforts and outlined its new 28nm and 20nm process technology roadmap.

UMC's Sun

UMC’s initial 28nm process is ramping based on a polysilicon gate-stack. The company plans to move into “risk production” with a 28nm, high-k/metal-gate technology in the second half of 2012, said Shih-Wei Sun, chief executive of Taiwan foundry vendor UMC.

Then, UMC will move to 20nm, which is slated for “pilot production” by mid-2013, Sun said during a conference call this week. At 20nm, UMC hopes to close the gap against its rivals in GlobalFoundries, Samsung and TSMC.

The definitions of “risk” and “pilot” production vary from company-to-company and it’s unclear which foundry vendor is actually ahead at 20nm. Recently, TSMC chairman and CEO Morris Chang said that company’s 20nm risk production will begin in the second half of 2013.

Regarding 20nm, a spokesman for GlobalFoundries said: “Risk production will start in 2H 2012, followed by volume production in 2013. Of course, we are already running test chips with both customer and partner IP. For example, we recently announced a 20nm tape out in partnership with ARM.”

A spokeswoman for Samsung Electronics added: “We expect to have 20nm qualified and ready for production in the second half of 2012. Ramp depends on our customers.”

Meanwhile, in another surprise, UMC this week posted mixed results and raised its capital spending to $2 billion in 2012, up from $1.6 billion in 2011. Barclays Capital projected that UMC would spend $1 billion in 2012.

The moves prompted analysts to wonder if UMC is renewing its plans to become more of a leading-edge foundry instead of its current strategy of being a “fast follower.” UMC may have no choice but to step up its efforts in the foundry business — and for good reason: It may be a matter of survival for the company.

“There is a shift in the industry landscape,” said UMC’s Sun, but “we have not changed our strategy.”

As before, UMC faces several challenges. First, the question is whether or not the company can execute on its plans. Second, can it keep up with its rivals? Third, like all foundries, UMC is struggling to get its high-k/metal-gate (HKMG) process working.

The foundries in general have “major yield problems in ramping 28nm HKMG,” said Handel Jones, an analyst with International Business Strategies Inc. (IBS), at a recent event. The “problems are DFM-related: physical design, reticles, and process parameters variations. It is becoming increasingly important to have close feedback between three disciplines of design, lithography, and wafer processing.”

New foundry landscape

Meanwhile, at one time, the “Big 4” in the pure-play silicon foundry market were Chartered, SMIC, TSMC, and UMC. Today, TSMC remains the clear leader in the foundry business. The parent of GlobalFoundries Inc. recently acquired Chartered, which has been melded into GlobalFoundries.

China’s Semiconductor Manufacturing International Corp. (SMIC) has fallen behind and spilling red ink again, while UMC is holding on to its position as the world’s second largest foundry – by a thread (See rankings chart below).

Two foundry vendors, GlobalFoundries and Samsung, have emerged and are looking to knock UMC off its perch. In fact, Samsung is stepping up its capital spending in 2012.

UMC is a solid foundry company, but based on its current strategy, the Taiwan company looks vulnerable and could lose its spot in the rankings. At one time, UMC attempted to keep up with its neighbor TSMC and compete in the leading-edge foundry business.

But as the costs and risks have escalated in chip manufacturing, UMC in recent times took a step back and decided to become a “fast follower” in the process technology race. While UMC is still behind its rivals in the process race, the company now appears to be taking some steps to become what HSBC analyst Steve Pelayo called a “faster follower.”

For some time, UMC has been shipping a 40nm process, which will represent 15 percent of its sales by year’s end. It is also ramping up its initial 28nm process based on a polysilicon gate stack. A 28nm process with high-k will move into “risk production” in the second half of 2012, with production due in 2013. In total, 28nm will represent some 5 percent of its sales by year’s end, Sun said.

The two processes share the same tool sets, which can be converted from 40nm to 28nm. To play catch-up — and become a “faster follower” — the game plan is to convert more capacity from 40nm to its high-margin 28nm process, depending on business conditions. “28nm is our higher priority,” Sun said. “We would like to drive as much 28nm as we can.”

There are some hurdles. The problem with high-k/metal-gate among the foundries is the reliability of the technology. “Everyone is having the same challenges,” he said.

Over time, however, 28nm will represent a “longer-lasting node in the industry,” Sun said. On average, a 28nm process with high-k is 40 percent more expensive than 40nm. And 20nm is a staggering 50 percent more expensive than 28nm, he said. “If you look at 20nm, it will require double patterning. That’s very expensive,” he said.

UMC is working on 20nm, based on its second-generation, gate-last high-k technology. Two undisclosed customers are working with UMC at 20nm. One customer will go into “pilot production” by mid-2013, while the other in the second half of 2013, according to UMC.

GlobalFoundries, Samsung and TSMC are moving full speed ahead in developing 20nm technology, based on a planar process. At 20nm, Intel is the clear leader in technology. “Intel is ramping 22nm finFETs wafer volumes, but with some delays,” IBS’ Jones said.

“22/20nm planar CMOS will have a range of technology challenges. 20nm planar technology will be more difficult to optimize the parametric yields than 28nm, due to leakage-related DFM problems. It requires a very tight control over process parameters and close link with design flows,” Jones said.

The foundries vow they will move to finFETs at 14nm. “FinFET technology will be another major hurdle from design as well as a process perspective,” the analyst said. Intel plans to implement 14nm with finFETs by Q4 2013, giving it a “lead over competitors by (up to) 30 months.”

Share and Enjoy:
  • Print
  • Digg
  • StumbleUpon
  • del.icio.us
  • Facebook
  • Yahoo! Buzz
  • Twitter
  • Google Bookmarks
  • LinkedIn


Tags: , , ,

Comments

Leave a Reply