A quadruple mass Microelectromechanical System (MEMS) Coriolis vibratory gyroscope has been re-engineered with the singular focus of minimizing nonlinear transduction mechanisms, thereby allowing for angle random walk (ARW) noise reduction when operating at amplitudes higher than 2 μm.

The redesign involved six primary steps: (i) the creation of an aspect-ratio independent deep reactive ion etch with minimal notching on 100 μm thick silicon-on-insulator device layer, (ii) the creation of micro-torr vacuum packaging capability, enabling operation at the thermoelastic dissipation limit of silicon, (iii) the redesign of Coriolis mass folded flexures and shuttle springs, (iv) the linearization of the antiphase coupler spring rate while maintaining parasitic modal separation, (v) the substitution of parallel plate transducers with linear combs, and (vi) the implementation of dedicated force-balanced electrostatic frequency tuners. Cross-axis stiffness is also reduced through folded-flexure moment balancing to further reduce ARW.

By balancing positive and negative Duffing frequency contributions, net fractional frequency nonlinearity was reduced to -20 ppm. The gyroscope presented in this research has achieved, a first reported of its kind, an ARW of 0.0005 °/√hr, with an uncompensated bias instability of 0.08 °/hr. These advancements hold promise for enhancing navigation and North-finding applications.

In tandem with gyroscope performance enhancements, vacuum packaging of ceramic chip carrier physics packages has achieved pressure levels below 1 micro-torr, a first in the field and remains state-of-the-art. Besides high-performance MEMS inertial sensors, ultrahigh vacuum packaging proves beneficial for chip scale atomic clocks, which require micro-torr vacuum levels to maintain fractional frequencies less than 10^-12.

Finally, an approach to tuning the quality factor mismatch between degenerate modes in as-fabricated gyroscopes has demonstrated a reduction in gyroscope bias instability. This tuning can be achieved by incorporating lead zirconate titanate into regions where the trade-off between mechanical Q, tuning Q, and bias instability reduction is balanced. Both modeling and empirical frequency data justify this approach, suggesting, for typical MEMS foundry Q mismatch of 7%, a 70× reduction in bias instability.