Optimized mechanical Design of capacitive micromachined Switch: a CAD-Based Neural Model

In this paper, the authors propose the neural modeling of optimized design for pull-in instability reduction of micromachined capacitive shunt radio frequency (RF) micro electro mechanical system (MEMS) switch. Prediction of effective dielectric constant and hence the critical collapse voltage that represents the bridge position instability for two typical bridge geometrics have been derived using artificial neural network (ANN). The effects of residual stress, length of center conductor and the gap between the bridge and center conductor of switch in lowering the driving voltage have been studied. Based on the neural model results, we have observed the reduction of 3 V in critical collapse voltage for an increase of 10 μm in length of center conductor. We have also noted the strong variation in voltage (reduction of 0.8 V for 1 MPa residual stress reduction) with respect to residual stress change. We achieved the reduction of 1.5 V in collapse voltage by reducing the gap between the bridge and the center conductor by 0.1 μm. Among the two structures considered, the structure with lower width of the center conductor proved as an optimum in achieving low critical collapse voltage. Further, the performance of trained neural network with the training datasets derived from MATLAB simulation has been evaluated in terms of convergence speed and mean square error.