Enhancing the development-rate model for optimum simulation capability in the subhalf-micron regime

Abstract

Methods used in the extraction of lithographic modelling parameters for simulation packages such as PROLITHJ2' are examined. The results reveal hitherto unconsidered aspects of the development process which, when implemented in the simulations, give improved agreement with practical results with regard to characteristics such as resolution, depth-of-focus, linearity and dense/isolated bias.These refinements, which are particularly influential in the sub-half-micron regime, include the variation in photoresist dissolution properties as a function of depth into the resist film and also a small but powerful development "notch" which is observed in the development rate vs. PAC concentration curve as it approaches the minimum dissolution rate.This work therefore shows that current development models may not be adequate for some applications, and that great care must be taken in deriving and using the correct set of parameters for any one situation.

Notch Model for Photoresist Dissolution

“…Some recent experimentally derived R(m) behavior is not well fit by either model [9]. Figure 1 shows the best fit of both the Mack and the enhanced Mack development rate models to one set of data from Ref.…”

“…Recent experimental development rate data has shown that current models do not give satisfactory agreement with the data for some high resolution i-line photoresists [9]. In particular, a "notch" exhibited in the dissolution rate versus photoactive compound (PAC) concentration curve had dramatic influence on the lithographic response of several photoresists but was not well fit by current models.…”

“…In the region around a 0.5 PAC concentration the data exhibits a "notch" behavior where the actual development rate drops very quickly to a value much less than that predicted by either model. It was previously shown that this sudden drop in rate, resembling a "notch" in the otherwise slowly varying behavior, is critical to resist performance [9]. It seems likely that today's high contrast i-line resists have been engineered to produce this notch by default, since it produces good lithographic performance.…”

Enhancing the development rate model in optical lithography simulation of ultrathick resist films for applications such as MEMS and LIGA

“…For consistency, in all cases the depth measured was that of the middle space of the group of three. 4. RESULTS.…”

Assessment of different simplified resist models

“…At the moment, these models are very specific, but hardly give reliable predictions on critical dimensions (CD) of final patterned features simultaneously with accurate resist profiles. Moreover, the techniques for measurements and parameter extraction are often very delicate, and incorrect input parameters is another source of simulation to experiment mismatch [3,4] . Finally, if lithographic simulations consist in manually changing some parameters in a way that the match between real and simulated data improves, the results can become very subjective.…”