Sematech litho meeting: Good talk, no selections
01/01/1998
Sematech litho meeting: Good talk, no selections
An elite, worldwide group of researchers from semiconductor and equipment companies convened under SEMATECH auspices in Colorado Springs, CO, from Nov. 4-6 to discuss options for post-optical lithography. All major companies and nations were represented, including a delegation from the ASET consortium in Japan.While there was no real chance of anointing a single best technology for production of 100-nm (0.10-?m) CD chips in 2004 (and 70-nm CD chips in 2007), some had hoped that the five leading options would be narrowed to two or three. Instead it appears that individual companies and consortia mean to pursue at least four different directions for the next two years, with some believing that advanced ArF (193-nm) lithography may be capable of 100-nm resolution. To fulfill current roadmaps, development-level patterning tools will be needed 5-6 years early. In addition, resist challenges emerged as a key issue. Since resist companies do not recover R&D expenses for 7-10 years - and are still investing in ArF resists - post-optical resists may require special attention from R&D consortia.
SEMATECH director of lithography Karen Brown expressed satisfaction at the exchange of ideas that took place at the meeting, and the interaction between groups of researchers. "There was a lot of talk about how we can work together," she said, citing as an example the use of wafer-based masks by several of the technologies. "To use a wafer in an e-beam mask writer, you need a holder. Some groups have those, some don`t, but they shouldn`t have to invent them. That`s a small thing, but it`s important - it saves people, time, and energy."
Brown said there was a general desire for another conference, adding, "We will watch closely over the next few months and see what the trends are, and get a sense of the right timing." The follow-up meeting could be slated for 12 or 18 months from now, or in some other time frame.
Five leading options were discussed:
1. multiple-feature e-beam writing (with multiple low-voltage columns, using cell or block projection);
2. ion-beam lithography (supported by Siemens);
3. SCALPEL (promoted by Lucent Technologies);
4. EUV (Intel and the EUV-LLC); and
5. 1? x-ray (IBM and Mitsubishi).
Of these, 1? x-ray is the most highly developed, but SCALPEL reportedly received the most votes in an informal straw poll at the end of the meeting. Except for option 1, all require new and difficult mask technology.
Projected cost of ownership varied widely, from $50/level for ion projection to over $160/level for e-beam direct write. None approached the current cost of DUV lithography (less than $25/level), thus calling into question the economic assumptions of the SIA Roadmap.
According to Gene Feit of SEMATECH, plenary speaker at the Olin Interface `97 meeting in San Diego, November 10-11, only ion-beam lithography exposes resist economically today. The other four methods fall into two groups in terms of resist requirements. Low-energy e-beams and EUV do not penetrate the resist film, requiring top surface imaging (TSI) resists with less edge roughness than in present processes. The high-energy electrons of SCALPEL, cell projection e-beam, and x-ray photons penetrate the full resist stack, but deposit too little energy for economical exposure times.
Etch resistance and mechanical stability of resist films will be difficult for 70-nm CD processes. With a 3:1 aspect ratio, the resist film will be 200-nm thick, so even "volume" exposure will produce patterns similar to today`s TSI. New processes and formulations that protect the resist from capillary forces may be essential, even for 3:1 aspect ratios.
Another problem is resist outgassing into the vacuum required by post-optical methods. One 193-nm projection lens has reportedly already been damaged by outgassing at SEMATECH and returned to the vendor. IBM contemplates using a pellicle to protect the x-ray mask from vapors emitted by the nearby resist film during exposure. The delicate EUV optics will also require protection by a membrane, which will necessarily absorb radiation and lengthen exposure times. Nonconducting resist films deposited in charged-particle lithography systems cause image distorting effects familiar to maskmakers. Thus, resist stability seems to have emerged as an issue in all of the technologies.
Another topic of general concern was the mask. Each candidate (except for multifeature e-beam) requires a unique mask structure and process, some of them 1?, some membrane or stencil; EUV needs an elaborate substrate. Patterning, inspection, and repair were previously recognized as problems, but mask lifetime in the harsh exposure environment has emerged as an economic issue.
Other attendees seemed chastened by the difficulties revealed. No one discussed "winners." Most could point out how advocates of some other method had "lost," but they also seemed to be backing away from the choices advocated by their organizations. Several brought up the lesson of the Concorde and the Boeing 747 for the airline industry. There may not be a suitable successor to optical lithography for general-purpose production. If none appears, the semiconductor industry will survive, but it will be different! - M.D.L., P.N.D.