Generation of MOS Model Parameters Covering Statistical Process Variations

This paper describes a methodology by which accurate MOSFET model parameter sets, covering statistical IC manufacturing fluctuations, can be generated. This procedure employs multivariate statistics to convert correlated device model parameters into a much smaller set of independent manufacturing process-related factors. Worst-case model parameter sets are constructed from the derived factors. Comparisons between measured device characteristic distributions and predicted worst-case device performance limits are utilized to evaluate the proposed technique.