Fatigue-Induced Failure of Polysilicon MEMS: Nonlinear Reduced-Order Modeling and Geometry Optimization of On-Chip Testing Device.

In the case of repeated loadings, the reliability of inertial microelectromechanical systems (MEMS) can be linked to failure processes occurring within the movable structure or at the anchors. In this work, possible debonding mechanisms taking place at the interface between the polycrystalline silicon film constituting the movable part of the device and the silicon dioxide at the anchor points are considered. In dealing with cyclic loadings possibly inducing fatigue failure, a strategy is proposed to optimize the geometry of an on-chip testing device designed to characterize the strength of the aforementioned interface. Dynamic analyses are carried out to assess the deformation mode of the device and maximize the stress field leading to interface debonding. To cope with the computational costs of numerical simulations within the structural optimization framework, a reduced-order modeling procedure for nonlinear systems is discussed, based on the direct parametrization of invariant manifolds (DPIM). The results are reported in terms of maximum stress intensification for varying geometry of the testing device and actuation frequency to demonstrate the accuracy and computational efficiency of the proposed methodology.