Even though many thin-film piezoelectric aluminum nitride (AlN) on silicon (Si) Lamb wave resonators have been proposed in the literature, little focus has been set on the modeling of the quality factors Q. Their Qs are associated with numerous dissipation sources, which are often difficult to separate from each other in experiments. Besides, the values of Q measured in experiments can largely deviate from sample-to-sample of the same design. In order to gain better insight into these issues, we have applied numerical models to estimate anchor losses and thermoelastic damping to a large set of AlN-on-Si resonators specifically designed to have significantly different Qs. The data set includes biconvex resonators of different curvatures (designed to reduce anchor losses), regular flat-edge resonators, and different electrode patterns. For the broad range of devices tested, we show that the computed values of Q agree well with the experimental data. In particular, the experimentally measured values of Q in regular flat-edge Lamb wave resonators are due to comparable contributions of the two types of losses.
This paper validates an innovative simulation tool for the prediction of gas damping occurring in MEMS working in near vacuum at frequencies in the range of 20-30 kHz. Three different families of test devices, representing standard building blocks of MEMS, have been designed, fabricated, tested, and simulated. A total of 292 structures belonging to 36 different geometries have been addressed in order to confirm the ability of the numerical model to capture the dependence on geometrical parameters and to provide a quantitative prediction. Tests have been operated both at variable pressures in a vacuum chamber and in the closed package, and demonstrate an accuracy in the order of 15%
We propose a non‐intrusive deep learning‐based reduced order model (DL‐ROM) capable of capturing the complex dynamics of mechanical systems showing inertia and geometric nonlinearities. In the first phase, a limited number of high fidelity snapshots are used to generate a POD‐Galerkin ROM which is subsequently exploited to generate the data, covering the whole parameter range, used in the training phase of the DL‐ROM. A convolutional autoencoder is employed to map the system response onto a low‐dimensional representation and, in parallel, to model the reduced nonlinear trial manifold. The system dynamics on the manifold is described by means of a deep feedforward neural network that is trained together with the autoencoder. The strategy is benchmarked against high fidelity solutions on a clamped‐clamped beam and on a real micromirror with softening response and multiplicity of solutions. By comparing the different computational costs, we discuss the impressive gain in performance and show that the DL‐ROM truly represents a real‐time tool which can be profitably and efficiently employed in complex system‐level simulation procedures for design and optimization purposes.
The properties of ferroelectric devices are strongly influenced, besides crystallographic features, by defects in the material. To study this effect, a fully coupled electromechanical phase-field model for 2D ferroelectric volume elements has been developed and implemented in a Finite Element code. Different kinds of defects were considered: holes, point charges and polarization pinning in single crystals, as well as grain boundaries in polycrystals, without and with additional dielectric interphase. The impact of the various types of defects on the domain configuration and the overall coercive field strength is discussed in detail. It can be seen that defects lead to nucleation of new domains. Compared to the energy barrier for switching in an ideal single crystal, the overall coercive field strength is significantly reduced towards realistic values as they are found in ferroelectric devices. Also the simulated hysteresis loops show a more realistic shape in the presence of defects.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.