An efficient approach to device parameter extraction for statistical IC modeling

A technique called Recursive Inverse Approximation (RIA) has been developed for parameter extraction for statistical IC modeling. High accuracy and efficiency are achieved by the proposed methodology. RIA combines the global optimization, parameter prediction, parameter correction, and accuracy checking. RIA fundamentally solves the accuracy problem in statistical IC parameter extraction. The proposed method is much faster than the optimization-based method. The technique was implemented in SMIC - a program for statistical modeling of integrated circuits.