By Masataka Tsubo, WaferNews Contributing Editor, Japan
With 65nm production ramping, the focus at SEMICON Japan in December turned to cutting-edge technology for 45nm production and 32nm development, leaving lithography suppliers in the unenviable position of midwifing two new exposure technologies into production at once.
On the 45nm front, Nikon said it would deliver its first NA=1.07 193nm immersion system by January and ship more than ten immersion units this year. Its new model for 45nm production, with NA=1.3, is expected to ship by 4Q06. For the 32nm node, Nikon argues that EUV is the clear choice. “We offer the highest NA [immersion] tool and continue to lead the resolution race,” claimed Toshikazu Umatate, GM of development at Nikon Precision Equipment. “But we believe EUVL will be preferred to extending immersion with either ultrahigh NA or two-pass exposure.”
ASML had already shipped nine 193nm immersion systems by 3Q05 with about 10 more in backlog, and the company expects to ship 20-25 immersion systems in 2006. Initial systems used a lens with NA=0.93, but a NA=1.2 model was available by November. The NA=1.3 version is in the works. ASML is still planning an aggressive push into EUVL with alpha tools for IMEC and Albany NanoTech slated to ship in 2Q06, but the company is cautious not to oversell the technology. “We are more practical,” said Peter Jenkins, ASML’s VP of worldwide strategic marketing. “There are potentially three choices for 32nm node. It is the customer’s decision which one to take,” added ASML product manager Hans Bakker.
Canon, meanwhile, doesn’t plan to ship its first NA=1.3 immersion tool until 1Q07, when its entirely redesigned two-stage scanner body will be ready. The new system will use a modular design to shorten assembly time substantially. Canon opened an immersion lithography laboratory in December to help customers in their process development.
The company is also still weighing its choices for the 32nm node. “There are infrastructure problems to be solved to make EUV feasible, with the masks, photoresists, and laser sources,” said Shigeyuki Uzawa, senior GM of Canon’s research lab. “We have to consider non-EUV as well as EUV options,” he stated, noting that a decision will be needed by early 2007 to ensure beta tool deliveries in late 2008. Canon’s recent acquisition of Anelva and its vacuum technology expertise, though publicly said to be aimed at reducing production costs of its surface-conduction electron-emitter displays, could also aid the company’s EUV tool design, which also requires sophisticated vacuum technology.
Suppliers report progress on increasing the source power that will be needed for practical EUV lithography. Several companies have reported discharge produced plasma (DPP) sources using tin are now up to around 250-400W at the origin and around 50W at the intermediate focus where it counts, while the Jenoptic-Ushio joint venture Xtreme reports an 800W version that might get as much as 80W at the intermediate focus. Xe sources are currently producing only about 20W, and laser-produced plasmas less than 10W.
Cymer, however, continues to see more potential in laser sources. “We discontinued DPP development in 2004. Laser is scalable to increase power,” said Robert Akins, chairman and CEO. “We are working to develop a 100W source by 2007 and to deliver a high volume manufacturing source by 2009. The whole world will turn to EUVL once 100W is achieved.” — Masataka Tsubo, WaferNews Contributing Editor, Japan