Interface 98 focuses on realistic lithography

01/01/1999

Interface `98 focuses on realistic lithography

The 35th Interface Symposium on "best practices" in microlithography showed the effects of the semiconductor slowdown. The quality of the presentations was as high as ever, all of them on some aspect of optical lithography or its plausible extensions. The show convened in San Diego, CA, in November under the sponsorship of Olin Microelectronic Materials.

Masaru Sasago of Matsushita Electro-nics Corp. gave the keynote talk on "Lithography solutions for the sub-130-nm generations." According to Sasago, optical lithography will progress "forever" toward shorter wavelength projection schemes with 193-nm ArF laser illumination being followed by 157-nm F2 lasers, then 126-nm Ar2 emission, and ultimately 13-nm EUV. The $300 million ASET consortium in Japan is pursuing this agenda, with $85 million assigned to develop a synchrotron-based EUV system. Opinion in Japan differs from that in the US as to the insertion date of e-beam and other nonoptical lithographies, according to Sasago.

Papers reviewed progress on DUV photoresist systems. Uzodinma Okorananyanwu of AMD presented a paper on progress on 193-nm technologies, which reported 27 different formulations of single- and bi-layer resists being developed for 193-nm. According to Okorananyanwu, there are at least 80 variations of alicyclic, acrylate, hybrid, and Si-containing bilayer formulations. While the industry seems to be converging on a few systems with similar properties, Okorananyanwu reported that none of them is entirely adequate for 130-nm production, many showing inadequate etch-resistance, excessive line-edge roughness, and outgassing. More hopefully, Francis Houlihan of Lucent Technologies described a material strategy combining low-volatility photoacid with norbornene-maleic-anhydride polymer and a photodecomposible base to achieve etch resistance 45% better than i-line resists. Houlihan predicted that this system may be useful as a crossover resist for 248-nm, 193-nm, and SCALPEL exposure.

Other papers described some aspect of photomask technology or optical proximity correction (OPC), which are becoming critical for sub-250-nm patterning. Using different methodologies, three papers estimated that reticle CD variations now consume half of the total error budget for state-of-the-art production. According to Cherie Goodenough, Keeho Kim, and John Lilygren of Cypress Semiconductor, mask errors constrain the successful methods of OPC. For example, simpler hammerhead serifs give better intrafield CD uniformity, whereas mouse-ear serifs give better exposure latitude, but worse uniformity. Wolf Staud of Applied Materials averred that "Linewidth errors are the real killer defects." Another author suggested that such defects could only be overcome with strong phase-shifting masks, not just OPC.

Olivier Toublan of France Telecom/ CNET described a statistical method using electrical linewidth measurements (ELMs) to characterize CD variation and proximity effect across the projection lens field. The variation and the effect of mask CD variation proved larger for steppers than for scanners and larger away from best focus. Two other European papers described using ELMs for characterizing resist/BARC systems in a production environment and for use as a general metrology for the 180-nm generation. While the need for improved CD statistics may someday overcome metrologists` misgivings about ELM, for now there are probably too many open questions.

In the coming era of deep sub-micron (DSM) lithography, everything will depend on everything else. For example, Mircea Dusa of National Semiconductor and seven colleagues described a new linearity metric that incorporates CD variation with exposure. They were able to show that the traditional distinction between dense- and isolated-feature resists breaks down when combined with OPC. The new 193-nm scanner reviewed by Anne-Marie Goethals of IMEC requires OPC and annular illumination for adequate common process windows. Somehow, all of the options for all of the process steps must be made to work together to achieve economical yield. Such is the task over the next several years. - M.D.L.