A high-linearity capacitive MEMS accelerometer with trapezoidal comb fingers

To address the inherent trade-off between sensitivity and linearity in variable-gap comb-finger accelerometers, this paper proposes a high-linearity trapezoidal comb-finger structure. This structure improves linearity by increasing the gap along the comb-finger motion direction, while simultaneously enhancing the sensitive capacitance value through increased overlap area, thereby achieving improved linearity without sacrificing sensitivity. To accurately calculate the trapezoidal comb-finger capacitance, an analytical method based on "undulating-channel-to-parallel-plate mapping" is proposed, with calculation errors of less than 2% relative to finite element simulations. Theoretical analysis shows that the nonlinear coefficient of the trapezoidal comb-finger structure is only 64% of that of parallel-plate comb fingers. Accelerometers integrating trapezoidal comb fingers and parallel-plate comb fingers were fabricated separately using a three-layer SOI process, with both types sharing the same mechanical sensitivity, comb-finger gap, and readout circuit. Open-loop test results show that, compared to parallel-plate comb fingers, the trapezoidal comb fingers exhibit an 8.9% increase in sensing capacitance, a capacitive sensitivity of 1.816 pF/g (a relative increase of 1.4%), and a nonlinearity of 0.04% (a reduction of 73%). Furthermore, the improved accelerometer demonstrates excellent overall performance: a noise floor of 0.52 μg/√Hz @1 Hz and a bias instability of 19.65 μg. This study offers both theoretical value and promising engineering application prospects.