Design, fabrication, and characterization of beam-supported aluminum nitride thin film bulk acoustic resonators

Micro-mechanical filters comprised of bulk acoustic resonators are being fabricated and studied as a solution to the demands for low power consumption, high functionality devices in the telecommunication industry. A novel, suspended thin Film Bulk Acoustic wave Resonator (SFBAR) has been fabricated using an aluminum nitride film sputtered directly on a <100> silicon substrate. The suspended membrane design uses thin beams to support, as well as electrically connect, the resonator. The SFBAR has been fabricated by combining both thin film processing and bulk silicon micro machining. The AlN was etched in an Inductively Coupled Plasma (ICP) chlorine etch, using titanium dioxide as the masking material. A silicon Deep Reactive Ion Etch (DRIE) was used to create an open ended air cavity with a novel circular shape. A representative resonator, designated here as sample W9HS8 resonator 10018, was characterized with a Quality Factor values at resonance and anti-resonance of 68 and 151, respectively. The sample also has an effective electromechanical coupling coefficient of 4.6% and is free of spurious resonances. The response of the resonator was representative of the majority of the resonators tested. The Quality Factor and the effective electromechanical coupling coefficient were characterized as a function of the number and the length of the support beams. The length of the support beams was found not to have any effect on the quality factor at resonance or the effective electromechanical coupling factor. However, longer support beams do facilitate better frequency pair response. Device performance varied with the number of support beams: 70% of the resonators tested show a higher Figure of Merit with eight support beams than with four support beams. A Butterworth-Van Dyke (BVD) lumped element circuit model was used to simulate the response of the SFBAR. The results from the BVD simulation match the experimental data and provide insight into the response of the SFBARs.