Parametric yield analysis on FEM based microsystem designs

As the importance of yield analysis in microsystems is increasing, designers can no longer neglect it in the design phase. The lack of reliable statistical methods to analyze yield of a design in a quick and efficient way has been a persistent problem until recently. One successful approach, worst-case based yield analysis, has not been completely exploited in the conventional design process of microsystems, making yield optimization of novel microsystem designs difficult. This paper proposes the integration of worst-case based yield analysis methods with one of the most conventional tools of microsystem design - FEM based design. By generating nonlinear response surface models from FEM simulations, efficient advanced worst-case based yield analysis methods can be utilized. We demonstrate the yield enhancement of FEM based microsystems with two different examples, a U-shaped thermal actuator and a capacitive pressure sensor.