UCLA

Silicon Microelectromechanical (MEMS) resonators are being developed for a wide variety of applications including frequency reference applications, positioning systems (gyroscopes) force sensors (AFM) and energy harvesters. In these applications energy dissipation greatly influences device performance. For example in a frequency filter the dissipation will determine the bandwidth of the filter. Many applications require dissipation to be minimized and all applications require accurate characterization of dissipation. In recent years advanced modeling techniques for some energy loss mechanisms (e.g., thermoelastic dissipation) have been introduced to predict resonator performance based on fundamental physics. The resonator can lose energy through a variety of energy pathways including air damping, losses through the anchor, surface dissipation, resistive damping and thermoelastic dissipation (TED). As modeling techniques improve more and more dissipation mechanisms can be predicted a proiri, saving significant cost in fabrication trial and error. TED, air damping and resistive damping have accurate models, however significant work remains to develop accurate general models for anchor loss and surface dissipation.

This work provides a dual approach to MEMS resonator design. For resonators limited by TED, or any loss mechanism that can currently be modeled, this work leverages a new bio inspired design optimization approach called binary particle swarm optimization (BPSO) used to optimize energy dissipation in MEMS resonators. BPSO produces mask ready designs that minimize damping. This approach was used to optimize low TED resonators and resulted in a measured 33% improvement over the previous intuitive design approach. Secondly in order to address the lack of an accurate general model for anchor loss this work introduces a novel anchor loss modeling approach independent of resonator frequency and shown accurate across 2 orders of magnitude in frequency. The main goal of this work is to encourage the MEMS community to move away from a trial and error fabrication approach and leverage modern modeling and optimization techniques to design high performance resonators.