Reliability of contrast and dissolution-rate-derived parameters as predictors of photoresist performance

摘要

The definition and use of contrast in photolithography has been adapted from the science and technology of photographic imaging media. Thus, a high contrast resists/process should allow useful high resolution images to be delineated even from a low contrast optical system (ie. one with low MTF) and hence provide improved resolution for diffracted limited imaging. Furthermore, it has been widely recognized and clearly documented, for example in ASTM F1059-87 that: 'The contrast value is a measure of process latitude of the photoresist'. Despite these assertions, there is a large body of evidence which suggests that contrast may not be a quantitative predictor of either resolution or process latitude, and in several cases opposite trends are reported. The authors of this paper set out to explore the utility of reliability of photoresist contrast (which we will refer to as 'gamma') as a predictor or resist performance including image profile, linearity and critical dimension control, both as a function of process variables such as film thickness, bake and develop conditions and as a function of resist composition. The influence of the precision of experimental measurements on the reliability of gamma determination is investigated. The recently reported variation of gamma with film thickness, which exhibits both a 'swing' and 'bulk' effect is confirmed. Experimental and computer simulation results show a 180$DGR out of phase relationship with Eo and an almost linear decrease in gamma (measured at E$-max$/ or E$-min$/) with increase in film thickness, which is consistent with published theory. However, the phase relationship may be perturbed for conditions in which gamma is a strong function of process variables such as develop time and post exposure bake. It is shown that gamma correlates reasonably well with resist profile but less so with critical dimension, especially when comparing resists with varying optical absorption properties, or systems which exhibit high surface inhibition. Many correlation may only be meaningful it gamma measurements are carried out with control in resist film thickness to within the same limits necessary for reproducible lithography. Computer simulation is used as an aid to gain further insight into the relationship between gamma and the slope Tan $phi of the steepest portion of the log (dissolution rate) vs. log (dose) curve. It is concluded that the use of gamma and Tan $phi in combination should provide a more reliable measure of resist/process performance.