Method for semiconductor process optimization using functional representations of spatial variations and selectivity

The authors present a methodology for determining the optimal equipment settings for a processing step based on experiment designs and model-based optimization. The proposed method for semiconductor process optimization uses two-layered models. The first layer involves creating a spatial model-one for each film of interest-of the etch results. The second layer maps the coefficients of the spatial models to equipment settings. All this is done before any optimization scheme is employed. The process engineer then may optimize the etch process by maximizing coefficients which contribute to the desired maximum etch rate, while minimizing coefficients which contribute to nonuniformity. He also may minimize coefficients which represent undesired etches, and thus obtain etch selectivity. The results of a study of a plasma-assisted silicon nitride etch step are presented