“Nonlinearities and Noise in Micromechanical Resonators: From Understanding to Characterization and Design Tools”

In this work we consider several nonlinearity-based and/or noise-related phenomena that have been recently observed in micro-electromechanical vibratory systems. The main goals are to closely examine these phenomena, develop an understanding of their underlying physics, derive techniques for characterizing parameters in relevant mathematical models, and determine ways to improve the performance of specific classes of micro-electromechanical systems (MEMS) used in applications. The general perspective of this work is based on the fact that nonlinearity and noise represent integral parts of the models needed to describe the response of these systems, and the focus is on situations where these generally undesirable features can be utilized or accounted for in design. We consider three different, but related, topics in this general area. The first topic uses the slowly varying states in a rotating frame of reference where we analyze the stationary probability distribution of a nonlinear parametrically-driven resonator subjected to Poisson pulses and thermal noise. We show that Poisson pulses with low pulse rates, as compared with the resonator decay rate, cause a power-law divergence of the probability density at the resonator equilibrium in both the underdamped (overdamped) regimes, in which the response does (does not) spiral in the rotating frame. We have also found that the shape of the probability distribution away from the equilibrium position is qualitatively different for the overdamped and underdamped cases. In particular, in the overdamped regime, the form of the secondary singularity in the probability distribution depends strongly on the reference phase of the resonator response and the pulse modulation phase, while in the underdamped regime several singular peaks occur in the distribution, and their locations are determined by the resonator frequency and decay rate in the rotating frame. Finally, we show that even weak Gaussian noise smoothens out the singular peaks in the probability distribution. The theoretical results are successfully compared experimental results obtained from collaborators at the Hong Kong University of Science and Technology. Second, we discuss a time-domain technique for characterizing parameters for models that describe the response of a single vibrational mode of micromechanical resonators with symmetric restoring and damping forces. These parameters include coefficients of conservative and dissipative linear and nonlinear terms, as well as the strengths of various noise sources acting on the mode of interest. The method relies on measurements taken during a ringdown response, that is, free vibration, in which the nonlinearities result in an amplitude-dependent frequency and …