NGL's Critical Bridge: 157 nm

Ruth DeJule, Associate Editor -- Semiconductor International, 3/1/1999

The decision to limit funding to EUV and SCALPEL has been made by International SEMATECH's Executive Steering Council. Amid concerns about the technical risk and cost for all NGL technologies at and below the 70 nm node, International SEMATECH will focus its programs on solutions to technical issues, risks and minimizing costs of EUV and SCALPEL. Support for narrowing the technology focus is the anticipation of additional development time, the critical bridge being 157 nm lithography. While International SEMATECH continues to support a feasibility study for 157 nm, virtually all stepper manufacturers are looking into 157 nm technology.

A year ago, 157 nm technology was not on any industry roadmap. Lambda Physik had a light source, though industrial-grade and not up to lithography specifications, and CaF₂ of sufficient size and quality was not available. Today, technical breakthroughs have spurred companies like SVGL to commit to 157 nm development.

The most complex component of an exposure tool is the optical system. Material requirements for shorter wavelengths add a particularly difficult challenge. Fortunately, in 193 nm systems, CaF₂ optics can be limited to points of highest laser intensity; however, the entire 157 nm optical train must consist of CaF₂. Material requirements will depend in part on the particular type of optical system implemented; the choices are all refractive or catadioptic (combination of mirrors and refractive optics).

SVGL has demonstrated the ability to grow and make a beamsplitter cube 100 mm on a side for use in its 157 nm catadioptic lithography tool. (Source: SVGL)

In what appears to be an evolutionary path, 157 nm catadioptic optical systems are similar to current SVGL 193 nm tools, both being purged systems requiring a nitrogen environment, said James McClay, vice president, 157 nm technology at SVG Lithography (Wilton, Conn.). The primary difference is the need for a higher NA (>0.7) in 157 nm tools to support 90 nm design rules. To achieve this, McClay questioned the feasibility of using all-refractive systems because of the CaF₂ requirements. All things being equal, the same NA and field size, catadioptic optical systems are smaller in overall size and require smaller optical elements, McClay said. For example, due to the way the design is folded, the catadioptic design is approximately one third the size and weight of an equivalent all-refractive system, and lens elements may be half that of all-refractive optics in both size and numbers. While the catadioptic lens elements could range from 50 to 200 mm, all-refractive lens elements as large as 400 mm might be required depending on design. However, catadioptic's primary challenge is a beamsplitter that allows the optical beam to fold over itself.

The beamsplitter, the heart of a catadioptic optical system, is a cube, the size of which depends on the numerical aperture (NA); a larger NA requires a larger cube. For CaF₂, this is non-trivial. The handling and polishing of this thermally sensitive and easily chipped material has been formidable. Two years in development, SVG, in conjunction with Schott, recently demonstrated the feasibility of growing and processing this material for optical systems by successfully producing a 100 mm CaF₂ cube (see Figure). Though not large enough for a 0.7 NA system, it meets the specifications for a 0.45 NA, McClay said. Polishing, coating and reassembly have been successfully demonstrated on smaller CaF₂ cubes.

Progress is also being made in the development of a 157 nm source. Currently, Lambda Physik is working on F₂ eximer laser technology. Emitting at 157.629 and 157.523 nm, powers of 15 W and 1000 Hz have been achieved, numbers comparable to current 248 nm KrF and 193 nm ArF lasers. There are indications that Cymer (San Diego, Calif.) is also developing a 157 nm laser source.

Numerous projects have emerged, resulting in a 157 nm workshop held in February. In addition to optical material and sources, resists and photomasks are being investigated. SVGL anticipates offering a mini-scanner for reticle and resist development by the end of 2000.