Samsung Describes Road to 14nm

OAKLAND, Calif. – Just days after it was confirmed the Samsung Galaxy S6 uses an Exynos processor made in Samsung's 14nm FinFET process, a Samsung executive talked about the company's road to 14nm. The milestone was a huge one for the Korean giant, beating rival TSMC to become the second chip maker after Intel to ship a 14nm FinFET chip.

“I think competition is healthy, it pushes everyone to achieve their limits – either from a technology angle, execution, or the ability to deliver service levels,” Kelvin Low, senior director of foundry marketing for Samsung Semiconductor, told EE Times.
Low would not comment on production volumes, yields, or the number of devices using Samsung’s 14nm process. Instead, he outlined the applications using 14nm and described the foundry’s FinFET strategy.
There are multiple designs already and some have taped out, Low said. End application segments include “single hand-held mobile computing,” consumer applications, graphics, and compute/networking. “Because the technology has developed so much, the chip can fit into a lot of applications,” he noted.
The common thread between these segments is the need for long battery life and energy efficiency. FinFETs promise lower leakage and lower power, which is important in handsets such as the Galaxy S6 and for data centers that want to increase their GHz/Watt performance.
Samsung’s 14nm chips are in production in Korea and Austin, Texas, though Low would not comment where certain chips are made. EE Times recently published a blog that questioned where the company’s Exynos 7420 chips were made – a Global Foundries fab in Albany, NY was among the possible locations.
“I don’t think it really matters externally, I think it matters that this technology is in production. I think where it’s produced is kind of secondary,” Low said.
Samsung spent several years developing its 14nm technology and debating which process node it would invest in after 28nm. Low expects that 28nm will still be a popular process node for years to come because of its price.

Kelvin Low

“[Going to 14nm] wasn’t an easy decision…the decision was made probably three years back and it wasn’t clear whether 20nm or FinFET would be right decision,” he said. “What we could offer at 20nm was limited by semiconductor physics. 20nm was probably the last node that planar architecture can still be realized; even though [customers] could accept it, it wasn’t ideal.”
FinFETs, on the other hand, provided improved performance and power but were risky from a time-to-production perspective. The 3D architecture also required Samsung researchers to understand new physics around how the architecture works with parts of a SoC.
“The parasitics associated with a 3D device are totally different from the planar device. We had to rethink how to make the abstraction work together with the EDA components…and make it as design friendly as possible so designers don’t feel a huge change when transition from planar to 3D architecture,” Low added.
FinFETs are also more discrete because they are no longer based on the length and width of transistors. Samsung is developing an education system to help engineers optimize the number of fins from an IP perspective.
Samsung is also investing in 28nm FD-SOI to provide customers with a cheaper, fully depleted silicon-on-insulator architecture. The cost per transistor has increased in 14nm FinFETs and will continue to do so, Low said, so an alternative technology such as 28nm SOI is necessary.
“Large and small companies are expecting foundries to play a bigger role in providing a more complete solution. Not just process technology and PDK, but elements like having a library, standard cells, memory, IOs or even basic IPs like USB,” Low added. “Having these building blocks available for our customers early and accessible, they can focus their internal resource to develop other differentiating IPs.”
Samsung and its competitors may be more inclined to play a larger role as the manufacturing equipment needed for nodes smaller than 28nm becomes prohibitively expensive.
“Another trend is there used to be a number of second source foundries where they were trying very hard to match their technology to the primary source. Going down below 28 we see these tasks as next to impossible,” he said. “It’s only more recently that we’re comfortable to bring [FinFets] to high manufacturing volume.”
Samsung is working with IBM’s Albany Nanotech lab to develop new materials and device architectures. Low had no comment on the company’s roadmap for next-generation process nodes.

http://www.eetimes.com/document.asp?doc_id=1326369