EUV is the most likely candidate for the next generation of IC lithography, said Luc Van den hove, president and CEO of IMEC, speaking at the International Electronics Forum 2010 in Dresden this morning.
"Scaling will continue for many years to come", said Van den hove, "for logic, for DRAM, for NAND – each with their specific layout requirements - allowing faster scaling for things like NAND which has these very regular structures."
He added: " I strongly believe that, for mass volume production, EUV has the strongest chance to be introduced. There are a lot of challenges in EUV but several of the key infrastructure challenges are being addressed, and the next wave of EUV scanner technology will be introduced in the second half of this year."
"Our partners are making devices on EV scanners. They’re already producing devices on the tools developed by our partners", added Van den hove, "the products we have seen in the last six months make us confident that we can introduce EUV. But it’s a challenge, for sure."
Van den hove dismissed the likelihood of multiple e-beam becoming the next production technology. "Multiple e-beam is also a candidate," he said, "but the amount of investment going into multiple e-beam is extremely small."
As scaling proceeds, there will be differences in transistor structures, materials and manufacturing. "Around 20nm is when we’ll consider 3D devices like FINFETs and multi-gate transistors – at 15nm they’ll be used for sure," said Van den hove, "beyond 15nm, we may need a materials breakthrough. We have to increase the mobility of the channels by introducing germanium or III-V materials – grown on top of silicon, which is the only way to make them effective."
Exploratory concepts being pursued in transistor design are: Quantum Well devices; Tunnel FET; Graphene FET; and Carbon Nanotubes.
For non-volatile memory, Van den hove is expecting a major change. "Cross-bar technologies will be the next generation of flash, or a replacement for the current flash technology", said Van den hove.
One of the start-ups specifically pursuing cross-bar technology for non-volatile memory is Silicon Valley’s Unity Semiconductor.
On the manufacturing side, Van den hove saw stacking technologies as providing a route to increasing complexity and density. He foresaw 3D DRAM and 3D NAND made by stacking, and 3D CPU made by stacking multi-core logic on top of memory transistors. Additionally: "We can easily integrate photonics on top of silicon technology", he said.
"3D is going to be a strong enabler of boosting complexity and increasing performance", said Van den hove, "but it’s not a replacement for 2D scaling. 2D scaling always gives this dramatic improvement."