Globalfoundries intends to have 14nm finfet in volume manufacturing in 2014, the same timescale as Intel has for introducing 14nm finfet manufacturing.
In fact, GF’s 14nm process may use smaller transistors than Intel’s, said Chian because “Intel’s terminology doesn’t typically correlate with the terminology used by the foundry industry. For instance Intel’s 22nm in terms of the back-end metallisation is similar to the foundry industry’s 28nm. The design rules and pitch for Intel’s 22nm are very similar to those for foundries’ 28nm processes.”
On the Intel 22nm process the drawn gate length is actually 26nm.
GF’s 14nm process delivers a number of benefits over 20nm. “One benefit is that it reduces development time because people can easily migrate from 20nm to 14nm,” said Chian, “customers can use the same GDSII and swap the transistors.” GF calls it ‘fin-friendly migration.’
Fin-friendly migration benefits from the characteristic of GF’s 14nm process that it retains the 20nm interconnect while moving to 14nm transistors.
Other benefits of GF’s 14nm are that it will deliver 20-55% higher performance than 20nm depending on the operating voltage, or that it can deliver 40-60% better battery life with the same performance.
The GF 14nm technology uses rectangular-shaped fins. That could change when it moves into volue manufacturing when the exigencies of real-world production require compromises, but GF’s current expectation is that the rectangular shape will survive into manufacturing. Intel has had to go to a triangular-shaped fin which reduces the performance and the power-saving benfits of finfets.
At the front of the GF 14nm finfet node will be mobile SOCs. GF has been working with ARM since 2009 to optimise its processes for ARM-based SOCs. 14nm will be the first process at which Intel will put mobile SOCs to the front of the node.
By moving to 14nm finfet while keeping 20nm interconnect, GF has brought forward the intorduction of its 14nm process by one year.
Next year GF starts production on 20nm. GF’s 20nm SRAM, and various customer test chips on 20nm, are yielding “very well”, said Chian.