Updated. Samsung rolls out 10nm finFET technology.
The 10nm finFET market is heating up in the foundry business amid the ongoing push to develop chips at advanced nodes.
Not long ago, Intel announced its 10nm finFET process, with plans to ramp up the technology in 2017. Then, TSMC recently introduced its 10nm process, with plans to move into production by the fourth quarter of 2016.
Now, Samsung Electronics said that it has commenced mass production of system-on-a-chip (SoC) products with its 10nm finFET technology. Another foundry vendor, GlobalFoundries, is skipping 10nm and moving directly to 7nm.
Samsung, meanwhile, said it is the world’s first company to ship 10nm finFETs, beating its rivals to the punch. “For 10nm, we are announcing that we have started production,” said Hong Hao, senior vice president of the foundry business at Samsung Semiconductor. “We are the first one to production at the 10nm node.”
Samsung, which has been talking about 10nm for some time, sees the technology as a long-running node. Intel has a similar view, but TSMC sees 10nm as a half-node. “We believe the 10nm node will be a long node,” Hao said.
Samsung’s 10nm technology, according to Hao, is substantial improvement over 14nm. Samsung’s new 10nm finFET process, dubbed 10LPE, has several advantages over its 14nm technology. Compared to 14nm, it has a 30% increase in area efficiency with 27% higher performance and 40% lower power consumption.
As expected, Samsung’s 10nm technology will make use of 193nm immersion and multiple patterning. More specifically, it will make use of triple-patterning to allow bi-directional routing for IC designers.
Following the introduction of Samsung’s first-generation 10nm process (10LPE), the company’s second-generation process (10LPP) is targeted for mass production in the second half of 2017. Process design kits (PDK) and IP design kits are currently available. SoCs with 10nm process technology are expected to become available throughout 2017.
What about 14nm and 7nm?
Amid the push towards 10nm, the finFET market is also heating up at 16nm/14nm. For some time, GlobalFoundries, Intel, Samsung and TSMC have been ramping 16nm/14nm finFETs.
The market will shortly have a new competitor—Taiwan’s United Microelectronics Corp. (UMC). Some years ago, UMC licensed finFET technology from IBM. “UMC’s 14nm development is progressing as planned, with a technical boost from the licensing of IBM’s technology,” according to officials from UMC. “14nm finFET is under qualification and will be ready in 2017. 14nm is now ready for lead customer design-in.”
Following 14nm and 10nm, the market is expected to migrate to 7nm. GlobalFoundries, Intel, Samsung and TSMC are working on 7nm. Both Intel and Samsung hope to insert EUV at 7nm. Samsung, for one, believes that 193nm immersion and multi-patterning are too complex and expensive at that node, prompting the need for EUV at 7nm.
In contrast, GlobalFoundries and TSMC plan to use 193nm immersion and multiple patterning at 7nm. EUV will not be ready in time for the 7nm node schedules at GlobalFoundries and TSMC.
Related Stories
Deeper Inside Intel
Top process execs shed light on 10nm and beyond, plans for longer time between nodes, and the future of EUV, finFETs and stacked die.
7nm Market Heats Up
Foundries roll 7nm processes, but 10nm is next
The Week In Review: Manufacturing (Sept. 16)
UMC readies 14nm finFET
 |
 |
 |
 |
 |
 |
 |  |
The 30% shrink refers to their regular SRAM scaling, which goes from 0.070 to 0.049µm². For comparison, Intel’s 14nm regular SRAM is 0.0588µm² with SADP.
I’ve heard people say Samsung has now leapfrogged Intel, but you could also see it that Samsung has needed 2 more years for something that is only a meager 1.2x better but significantly costlier (the triple patterning to do 2D routing). That gives a bit of a different perspective..