Best/worst case extraction and yield cost estimation using Monte Carlo statistical analysis

Methodology and simulation results from Monte Carlo sensitivity analysis based on discretization of the second-order Taylor expansion for electrical device characteristics are presented. The same dice conditions for both NMOS and PMOS were modeled. Non-degenerate distributions of electrical parameters were formed. Results of selected direct process/device simulation were mapped onto distributions obtained by the Monte Carlo analysis. The resultant distributions were used to establish yield cost associated with desired best/worst case electrical characteristics. The approach was tested for two technologies and showed excellent agreement with available experimental statistical data.