(OT) Chips again. Reaching 45 nm Gets More Interesting
Aaron Hand, Managing Editor -- Semiconductor International, 8/1/2003
reed-electronics.com
Lithography certainly took center stage at SEMICON West last month. Not that it isn't always an important discussion point, but you really couldn't go anywhere without hearing about the predictions regarding the various tricks for stretching 193 nm technology, the feasibility of 157 nm or extreme ultraviolet (EUV) lithography, or the darling of the day, immersion lithography.
It was a good venue to debate Intel's recent announcement that it may skip 157 nm lithography. Never mind that it was Intel that put 157 on the roadmap in the first place, in the late '90s. Intel's idea is that 193 nm lithography can be stretched through the 45 nm node, and EUV lithography can be ready by 32 nm. So are the tool and materials suppliers, who've spent billions of dollars to develop 157 nm technology, ready to throw it out the window? Well, not exactly. Despite Intel's considerable influence in the industry, the company's announcement is far from a concrete dictum. In fact, Intel hasn't even pulled out of 157 nm programs.
But if Intel does indeed refrain from relying on 157 nm lithography, there are factors that make that more plausible for the chip giant. For one, Intel's high-volume product mix means that it can afford to pay the $5M-$6M price tag that's likely to be attached to the extremely complex mask sets that will be associated with running 193 nm lithography at the 45 nm node. But what's particularly fishy is that none of the toolmakers sound nearly as confident as Intel that 193 is even doable at that node. The confidence difference comes in what exactly is considered to be the 45 nm node. In strict terms, a node is dictated by the DRAM half pitch of a technology. There's been some effort over the years to make the node names — 180 nm, 130 nm, 90 nm, 65 nm — actually match the half pitch number. But those numbers have been diverging again, and there's some considerable artistic license being taken.
At the 10th Annual Lithography Breakfast Forum, a keynote address from Intel made it clear that a 193 nm tool was considered the best bet for the 45 nm node — and this is a dry tool, not considering the much-talked-about immersion possibility. Focusing her talk on logic, Janice Golda, assistant director of Intel's Lithography Capital Equipment Development, presented a chart detailing various parameters of what Intel considers to be the 90, 65, 45 and 32 nm nodes. Although Intel's 90 nm node shows a half pitch of 110 nm (relatively close to the node name), the half pitch for the 45 nm node is listed at ~75 nm. This, Golda noted, could be done on a 193 nm tool with an NA of >0.9. "It's not all the process window we'd like, but it'll get us there," she said.
For many chipmakers, reaching what they consider to be the 45 nm node will likely call for immersion lithography — if the industry decides that is indeed the way to go. Even Intel, which is enthusiastically pushing EUV development for the 32 nm node, concedes that immersion 193 nm tools may be the necessary bridge to an eventual EUV use at 22 nm.
Although toolmakers don't plan to commit to this technology one way or another until around the end of the year, the comments seem generally positive at this point. International SEMATECH (Austin, Texas) sponsored another immersion lithography workshop on Friday before SEMICON West opened. The consensus of the almost 200 attendees was that the technique — which places water between the final imaging lens and the wafer to effectively lower the wavelength (see "Industry Takes a Hard Look at Immersion Lithography ," Semiconductor International , March 2003) — shows excellent promise, and has no major technical showstoppers in sight. At the moment, the industry is favoring a technique that provides local liquid delivery and removal, using pure water as the liquid medium for 193 nm lithography.
There are still questions to be worked out before the industry commits its resources to further development, but attitudes are upbeat about a technology that requires very little change in the existing infrastructure. It appears that immersion will require only minimal tweaking of resists, lenses, masks, etc. That's good news for a sector still facing substantial challenges with next-generation platforms (Table ). |
