| For 25 years, optical lithography has successfully evolved to remain the mainstay patterning technology for the semiconductor industry. Imaging at progressively shorter wavelengths with larger numerical apertures has provided a 30% reduction in minimum feature size approximately every three years. A question that naturally arises is: what will ultimately limit optical lithography as we pursue higher resolution by migrating deeper into the ultraviolet?; As a means for exploring lithographic imaging at short wavelengths and high numerical apertures, we have built scaled prototypes of projection systems based on the catadioptric Markle-Dyson design. This remarkably simple two-element configuration offers diffraction limited imaging over a substantial field of view and arbitrarily large apertures. Requiring only one refractive element, the design is well suited to investigating the deep ultraviolet where few materials are transmissive. Operating at a numerical aperture of 0.7 with 193 nm illumination from an ArF excimer laser, we have demonstrated lithographic optics with the highest nominal resolution to date. Reflective masks are required for the system and have been fabricated with minimum linewidths of 0.12 {dollar}mu{dollar}m by electron beam patterning of silicon on fused silica. To characterize the projection optics independent of resist processing nonlinearities, a scanned grating technique was devised and an associated model based on partially coherent imaging was developed. The experimental results quantitatively confirm that diffraction limited imaging is achieved over the designed field. We have also printed projection exposures in resist with features down to 0.15 {dollar}mu{dollar}m.; This provides a proof of concept for extending optical patterning to at least the 1 Gbit dynamic memory generation. Further gains are available through use of phase shifting masks and other optical enhancements. Due to depth of focus constraints, numerical apertures substantially higher than 0.7 are unlikely. Still shorter ultraviolet wavelengths are conceivable, although at present there exist no sources of sufficient brightness for production lithography below 193 nm, and materials limitations become increasingly onerous. |
