Pattern transfer by plasma etch has become an indispensable tool to fabricate semiconductor devices whose critical feature dimensions are less than 3.0 μm.Traditional wet chemistry techniques do not maintain linewidth fidelity (isotropic profile) and lose applicability for features whose aspect ratio (width:thickness) is below 3:1.Plasma processes must keep pace with the rapid introduction of innovative device designs whose packing density is continually increasing.Hence many etch process developments are based on empirical, rather than theoretical, wisdom.Often the desired etch results of anisotropy, selectivity, uniformity and cleanliness become competing requirements.

Many of today's circuit designs have structures, topography and feature dimensions so interrelated that difficulty is usually encountered during the etch process development.Many of the single layer materials that were used in yesterday's devices have become multiple material films in the current advanced designs, increasing the complexity of the etch process and confusing the interpretation of the process results.An increasingly important tool that can be used for successful etch development is the statistical fractional factorial experimental design.This paper will discuss a few examples of plasma process development that involve a rank ordering of the importance of competing etch results.