We present an epitaxially-encapsulated 2×2 mm 2 quad-mass gyroscope (QMG). Relative to the earlier QMG which measured 8×8 mm 2 and required an external vacuum package and getter [1], this device is 16x smaller in area and is vacuum-sealed at the wafer-level. Due to the device's small size, high quality factor (Q) and large oscillation amplitude are required to achieve low noise. However, the device's high Q (85,000) makes it highly sensitive to mechanical nonlinearity, resulting in amplitude-frequency dependence and instability of the oscillator loop at large amplitudes. To overcome these problems, we demonstrate electrostatic compensation of the mechanical nonlinearity, enabling 10x greater amplitude and therefore scale factor (SF). Together with closed-loop amplitude control and quadrature compensation, this enables angle-random walk of 0.42 mdeg/s/VHz, comparable to the best QMG published to date. Closed-loop amplitude control and quadrature null are used to achieve a bias instability of 1.6 deg/hr.
We present an assortment of geometrically parameterized disk resonator gyroscopes (DRGs) individually designed for resonant frequency (f) and quality factor (Q). This work demonstrates accurate FEM-based predictions of Q for fabricated devices, and geometric optimizations that significantly improve this key performance metric for the DRG. Measurements on suites of fabricated devices confirm that losses are dominated by thermoelastic dissipation (TED). Qted is shown to account for 64-80% of the total energy dissipation. The temperature dependence of Q is also used to confirm and quantify the presence of pressure damping in the fabricated devices. At frequencies near 300 kHz, we show that Q can be increased by a factor of 2.6 and demonstrate feasibility of geometric/topological optimization of the fabricated devices with the aid of finite element software.
We demonstrate how to identify regions of major thermo-elastic dissipation (TED) in MEMS resonators and reduce this energy loss by modifying the device geometry. To demonstrate this, various geometries of a disk resonating gyroscope (DRG) are used. Devices are fabricated and tested to show that the TED-limited quality factor (Q) can indeed be increased using geometric manipulation, as predicted by simulation.
15 July 2020 in Journal of Microelectromechanical Systems
In this paper, we observe amplitude-dependent nonlinear dissipation in experimental measurements of the quality factor (Q). The ringdown response is used to quantify the magnitude of the amplitude-dependent effect in both polysilicon and single crystal silicon (SCS). The Q is reduced in both materials, up to 15.6%, and has a stronger influence on the measurements of Q in polysilicon than in SCS. Since the measured limiting dissipation mechanism in these resonators is thermoelastic dissipation (TED), we suggest these results align with models of amplitude-dependent TED. [2020-0160]
12 March 2019 in Journal of Microelectromechanical Systems
Fatigue in thick (> 20 μm), epitaxially deposited polysilicon MEMS is characterized from 25°C and up to 250°C in an environment free of oxygen and humidity. This work is the first to report fatigue initiated in the native epi-polysilicon, free from any contributions due to oxide or environmental conditions. Fatigue was most prominent above stresses of 1.5 GPa at all temperatures. Resonant frequency shift is used to measure crack growth, and elevated temperatures increased the total frequency drift and frequency drift rates. Since the epi-polysilicon surfaces roughen due to grain boundary migration during fabrication, we conclude that epi-polysilicon fatigue can be attributed to a subcritical cracking mechanism that arises from surface roughness. Furthermore, the measured increases in the critical crack length before failure suggest that the fracture toughness increases at the elevated temperatures.