SAN JOSE, Calif. – It’s a clash of the titans in the foundry industry between Globalfoundries Inc. and Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC).
The silicon foundry rivals are separately expanding their process offerings and services for customers, but they have different strategies to knock each other off from their respective perches.
For example, Globalfoundries has a high-k/metal-gate only offering for the gate stack at 28-nm. In contrast, TSMC is offering both a high-k/metal-gate and polysilicon gate-stack options at the node. TSMC is pushing hard for 450-mm fabs. Globalfoundries and other members of IBM Corp.’s ''fab club’’ have been less vocal about 450-mm.
Globalfoundries, TSMC, and, of course, Intel Corp., also have slightly different strategies in lithography, which is a big factor in scaling to bring products to market. At the SPIE event this week, Intel elaborated on its lithography strategy. Intel sees extreme ultraviolet (EUV) as its primary option for next-generation lithography (NGL). Maskless remains an option at Intel.
Globalfoundries is also high on EUV, and has suddenly warmed up to maskless lithography. TSMC is more bullish on maskless for NGL, but it is also warming up to EUV. In optical, TSMC has mainly relied on ASML Holding NV as its sole lithography vendor, while Globalfoundries prefers a dual-source strategy with ASML and Nikon Corp.
Globalfoundries, which was the former manufacturing arm of Advanced Micro Devices Inc., was recently spun off into a foundry company. Globalfoundries has been using 193-nm immersion since the 45-nm node and the foundry upstart believes it can push the technology down to at least 20-nm. (Intel started 193-nm immersion at 32-nm.)
Unlike TSMC, Globalfoundries is a big proponent of EUV lithography. Last year, the foundry company ordered a full-blown EUV production machinefrom ASML, bypassing ASML’s pre-production tool.
Globalfoundries plans to offer customers two options in lithography at the 20-nm node, said Harry Levinson, Senior Fellow at Globalfoundries and manager of strategic lithography at the company.
In the first option, customers will be able to use 193-nm immersion with double patterning at 20-nm, Levinson said. In the second option, customers will have access to EUV for use in pattering, he told EE Times.
The double-patterning option is a slower and more expensive solution. So Globalfoundries wants to give customers the option of EUV, which, in theory, is expected to lower the cost-of-ownership in lithography. The lack of power sources, resists and masks remain a concern, however.
As reported, Globalfoundries will install an EUV tool from ASML in the second half of 2012 and will be using EUV in volume production by 2014 or 2015. With risk production set to begin in the second half of 2012, the company is expected to deliver 22- and 20-nm technology to customers for product introduction in 2013. So until the EUV tool is delivered, Globalfoundries may resort to only offering one option at 20-nm: 193-nm immersion with double patterning.
Beyond 20-nm, Levinson did not elaborate on Globalfoundries’ lithography strategy. But surprisingly, the foundry company is beginning to explore the use of maskless lithography. Maskless is a possible solution for small-lot runs and ASICs.
''We’re starting to look at it,’’ he said. There are a range of maskless lithography vendors in the market, but to date, Globalfoundries has yet to find a solution that meets its requirements, he added.
Interestingly, no 16 nm node anymore. It became 15 and now 14 nm. It might even become 13 nm by production time. 20 nm is already fixed to double patterning with 193 nm immersion. It could extend one more node. Multiple patterning should be mainstream in 2012.
If TSMC has decided to use EUV at 16nm that means it's been proven cost effective in high volume. The negative of this is, the other technologies have not been proven, at least to TSMC's satisfaction. TSMC really knows high volume and their opinion is the way to bet.
its all okey but what about intel & samsung. It seems intel is not willing to be involved in EUV since its fully applicable either samsung. If they drag to spend money on EUV who will finance technology then? It seems double patterning is the only feasible solution till we reach below 20nm.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.