July 14, 2011 – Wednesday at SEMICON West 2011 was a lithography-rich day, with the Sokudo Lithography Breakfast and some interesting keynote speech comments.
| SEMICON West 2011 |
| Day 1: SOI vs. FinFET, ReRAM vs. 3D NAND, and lots of video data |
| Day 2: A lithography-rich day |
| Day 3: Advancement in LED, exec perspectives, solar observations |
The always-anticipated litho breakfast did not disappoint. Mark Kelling of GlobalFoundries described his company’s approach to the transition from 22 to 20nm, and what this will involve: moving poly pitch from 100 to 80nm, and metal pitch from 80 to 64nm. GlobalFoundries will adopt a unique double-patterning process that utilizes a negative-tone develop resist process, he noted, but did not reveal the materials involved.
The two remaining advanced optical lithography suppliers, Nikon and ASML, both offered roadmap presentations that outlined their expectations for the last generations of 193 and the advent of EUV technology into production. Nikon noted that its upcoming 621D optical system will have a remarkable 2nm/3s overlay spec. These advanced systems will feature scanning tuning exposure control to be able to extend immersion 193nm performance. On the EUV side, their EUV-1 development tool will have an NA of 0.25, with the production-oriented EUV-HVM boosting this to NA=0.4.
For its part, ASML laid out an EUV roadmap going from the developmental NXE3100, also with a 0.25 NA, to the 3300B, which will have specified throughput of 125 wafers/hour, assuming the availability of a source that can deliver 15mJ of energy to the wafer surface. ASML currently has three EUV tools at customer sites running wafers, with a fourth being installed, and plans in place to ship 10 units in 2012. Interestingly, the company has not yet made a final decision on its source supplier.
Another angle on post-optical lithography came from Mapper Lithography, which is developing a multi-beam e-beam direct-write system. Using excellent presentation graphics, Bert Jan Kampherbeek explained that Mapper’s initial machine, the 1.1, will have throughput of 1WPH at 25nm, when shipped to French research agency Leti and Taiwanese foundry TSMC next year. The company plans to use clustering of multiple beam sources to boost throughput to 10WPH and ultimately 100WPH in time for the 16nm process node. Mapper’s value proposition will hinge on throughput per unit investment; they hope to sell machines for