A MEMS electro-mechanical co-optimization platform featuring freeform geometry optimization based on a genetic algorithm.

Abstract

This paper describes a novel, system-level design methodology based on a genetic algorithm (GA) using freeform geometries for microelectromechanical systems (MEMS) devices. The proposed method can concurrently design and co-optimize the electronic and mechanical parts of a MEMS device comprising freeform geometries to achieve a better system performance, i.e., a high sensitivity, a good system stability, and large fabrication tolerances. Also, the introduction of freeform geometries allows higher degrees of freedom in the design process, improving the diversity and potentially the performance of the MEMS devices. A MEMS accelerometer comprising a freeform mechanical motion preamplifier in a closed-loop control system is presented to demonstrate the effectiveness of the design approach. The optimization process shows the main figure-of-merit (FOM) is improved by 195%. In the mechanical component alone (open-loop system), the product of sensitivity and bandwidth has improved by 151%, with sensitivity increasing by 276%. For closed-loop performance, there is an improvement of 120% for the ratio of open and closed-loop displacements. The product of sensitivity and bandwidth is improved by 27% in the closed-loop system. Excellent immunities to fabrication errors and parameter mismatch are achieved. Experiments show that the displacement of the MEMS accelerometer in the closed-loop system decreased by 86% with 4.85 V feedback voltage compared with that in the open-loop system under a 1 g 100 Hz acceleration input. The static and dynamic nonlinearities in the closed-loop system are improved by 64% and 61%, respectively, compared with those in the open-loop system, in the ±1 g acceleration input range. Besides, the closed-loop system improves the cross-axis sensitivity by 18.43%, compared with that in the open-loop system. It is the first time a closed-loop system for a MEMS accelerometer comprising a mechanical motion preamplifier is successfully implemented experimentally.