Design of novel sensing solutions based on coupled MEMS resonators

In MEMS sensors, the natural frequency of the structure is designed to be sensitive to the quantity of interest; it is monitored by placing the resonator in an oscillator loop and comparing the oscillation frequency to that of a reference oscillator. If the reference oscillator has the same sensitivity to environmental drifts as the one used for sensing, a differential measurement of the quantity of interest can be obtained. To avoid temperature gradients and the resulting measurement errors, the two oscillators must be placed as close to each other possible. Parasitic couplings (electrical, mechanical, electrostatic …) between the oscillators are inevitable and result in unwanted signal injection, and frequency pulling or locking. Alternatively, one may seek to take advantage of existing couplings, or even enforce them, and take advantage of the parametric sensitivity of coupled resonators to implement differential sensors, with two or more resonators in close proximity. Several approaches exist for coupling MEMS resonators: the coupling may be passive (e.g. electrostatic or mechanical) or active (enforced through a linear or nonlinear actuation). The resonators may be operated in open-loop or closed-loop, at large or small oscillation amplitudes. Several output metrics may also be used to convey the information one seeks to measure, with different sensitivities, and different requirements on the interface electronics. To this day, however, these many possibilities have only been superficially investigated, on a case by case basis, for lack of a comprehensive analysis framework. The purpose of this PhD thesis is to establish such a framework, use it to map the design space of coupled MEMS sensors, and propose, fabricate and test innovative solutions that take advantage of the redundant information available in the different sensing modes of coupled architectures.