|
Silylated, dry-developed resists have demonstrated superior resolution, beyond that of their solution-developed resist counterparts. However, the implementation of plasma-developed resists in the manufacture of integrated circuit devices has not reached volume fabrication despite the fact that silylated resists minimize various imaging pitfalls including thickness variations over topography, limited depth of focus, thickness or swing curve effects and reflection4nduced exposure variations. This is exemplified with deep-UV lithography at 248 nm in which silylated resists have printed 200 urn features with a 0.53 NA exposure tool. Plans for implementing silylation technology are not imminent because of the conviction that solution-developed resists will meet the specifications for 250 urn design rule devices. Yet, will actual yields using solution-developed resists be adequate, will the processing and materials costs be acceptible and will the processes be extendable to much smaller features (0.2 pm) with or without optical tricks? For silylated resists, on the other hand, the issues are the availability of production worthy equipment, the complex nature of the processing, yields which may be adversely affected by particles produced during the plasma development process, process latitude, sensitivity and cost. For extreme UV lithography (EUVL) the silylated resist issues are somewhat different. This paper examines various aspects of silylated, plasma-developed resists to determine if they have the potential of resolving 100 nm features in thick (700 nm) resist films needed for device fabrication. The starting point is DUV performance. We have used a bi!ayer resist comprised of Shipley XP-8844 n hard-baked Shipley MP-1807 to resolve 200 nm lines and spaces (l/s) in 800 nm thickness films using 72 mJ/cm doses of 248 nm light, imaging at a Rayleigh k1 value of 0.43. It has been achieved because of advanced, Si-rich silylating reagents, control of plasma development conditions in a helicon-source etcher, control of thermal processing to minimize acid catalyst migration in the chemically amplified resist and control of flow during silylation. Applying the above k1 value to imaging at 13.5 nm with (NA) 0.08 predicts a 1/s resolution of 73 nm for EUVL. In practice this can't be achieved yet, but we report here on new developments in plasma etching and silylated resist flow control that are helping to reach this goal. Certain processing phenomena may limit resolution. In the imaging layer, migration of acid catalyst during and after exposure may blur the image. During and after silylation, swelling and flow may enlarge certain regions nonspecifically. Another factor is anisotropic etching of the planarizing layer such that no undercut of the masking layer occurs even upon * Present address: Dept. of Chemical Engineering, Mass. Inst. of Technology, Cambridge, MA 02139 ** Present address: Lawrence Livermore National Laboratory, P0 Box 808, Livermore, CA 94550 overetching. A final factor is absorption, which for EUVL, is almost independent of the resist, but is sensitive to wavelength. At 13 .5 nm the absorbance is so high that only surface imaging resist schemes are likely to give high resolution imaging in thick (0.70 pm) features. This paper provides an overview of developments in each of these areas that may limit silylated resist technology. Methods are described that minimize silylated resist flow, uniformly and reproducibly remove Si surface residue from exposed areas and reduce undercut during plasma developmentand overetch. Usin these methods we were able to obtain the first thick film images of sub 200 nm 1/s at reasonable EUV doses of 24 mJ/cm . The incremental understanding that has contributed to these results is discussed. Keywords: lithography, EUV, resists, silylated, plasma develop
|
