Uncertainty Quantification Study For A Comprehensive Electrostatic MEMS Switch Model

This work presents an uncertainty analysis of a comprehensive model for an electrostatic MEMS switch. The goal is to elucidate the effects of parameter variations on certain performance characteristics. A sufficiently detailed model of an electrostatically actuated beam is developed. This model accounts for various physical effects, including the electrostatic fringing field, finite length of electrodes, squeeze film damping, and contact between the beam and the dielectric layer. The performance characteristics of immediate interest are the static and dynamic pull-in voltages for switch. Using Latin Hypercube and other sampling methods, the model is evaluated to find these performances characteristics when variability in the model’s geometric and physical parameters is specified. Response surfaces of these results were constructed via Multivariate Adaptive Regression Splines (MARS). Using a Direct Simulation Monte Carlo (DSMC) technique on these response surfaces gives smooth PDF’s of the outputs. The relative variation in output due to each input is used to determine the critical parameters