An automated approach to wafer distribution analysis

A set of statistical analysis techniques applied to parametric on-wafer testing are presented. The suitability of these techniques within an automated environment is emphasised, as an aid tool to control process stability and quality. The analysis includes a data filtering step where a central data distribution is assumed. The proposed filtering scheme tends to minimize the kurtosis and skewness of the parameter distribution. Proposed spatial analysis techniques applied to filtered data include data winsorizing, and first and second order momenta calculation. These statistics treat wafer non-homogeneities with a minimum amount of data as a counterpart to the classic 2D contour or 3D plot.