ETCHED-QUARTZ FABRICATION ISSUES FOR A 0.25 MU-M PHASE-SHIFTED DYNAMIC RANDOM-ACCESS MEMORY APPLICATION

A 0.25 mum pattern from a dynamic random access memory application is used to assess the impact of an etched-quartz fabrication process on the lithographic performance of an actual phase-shifted circuit design. A new measurement tool called the aerial image measurement system is used to quantify the effects of an isotropic wet etch following the quartz reactive-ion etching step on the printability of the phase-shifted pattern at two values of partial coherence. The presence of significant transmission and phase errors, in part a result of optical scattering from the edges and/or comers of the etched-quartz trench, reduced the potential benefits obtained from phase shifting. At the optimum postetch conditions for a partial coherence of 0.36, only 65% of the ideal depth-of-focus was obtained. Transmission errors were reduced from greater than 25% with no postetch treatment to less than 5% with wet-etch depths in excess of 1100 angstrom. However, large phase errors, between 20-degrees and 60-degrees, were present for all postetch conditions, and were strongly dependent on the partial coherence. Optimization of depth-of-focus required a nominal wet-etch depth of 1200 angstrom for sigma = 0.6, but only a 900 angstrom depth at sigma = 0.36. Large variations in the phase error as a function of the postetch treatment conditions were attributed to electromagnetic resonance phenomena between the sidewalls of the etched-quartz trench.