High-performance MEMS accelerometers are key components of advanced equipment, which have been widely utilized in many fields, e.g., aerospace navigation, geological exploration, and military applications. Due to their high sensitivity, large bandwidth, and digital output, MEMS resonant accelerometers have drawn much attention in recent years. However, there are still many unsolved problems for silicon-based resonant accelerometers, which hinder the expansion of their market share. In this thesis, based on the in-depth analysis of the current research status of silicon-based resonant accelerometers, we propose our topology design. The testing and integration technology of the sensor is deeply studied. Further, a new sensing mechanism based on synchronization is carried out. The main content of the thesis is as follows.

Firstly, the working principle of the MEMS resonant accelerometer was introduced, and the topology structures of the double-ended tuning fork, the supporting beam, and the micro-lever were determined. Seven key parameters of the resonant accelerometers were summarized. The relationships among each parameter and the structure dimensions were studied through the finite element method. After that, we formulated the performance requirements of the accelerometer, proposed two prototype designs, and determined their structure dimensions.

Secondly, after manufacturing of the initial prototype, it was integrated with a DIP-24 ceramic package. The open-loop testing circuits were built to characterize the amplitude and phase response of the tuning forks. The scale factors of the accelerometers were measured through tilting analysis, and they were found to coincide with the theoretical value. The closed-loop testing circuits base on self-exciting oscillation were built to analyze the zero-biased instability of the accelerometers. After that, a dynamic testing system was built, including a low-frequency shaking table, a standard accelerometer, and a laser vibrator. The dynamic performance of the sensor was thoroughly evaluated.

Thirdly, using the vacuum eutectic furnace, the device-level vacuum packaging technologies based on eutectic melting and adhesive bonding were achieved. The vacuum degree and the lifetime of the packaging were analyzed according to the quality factors variation of the tuning forks. Then, the temperature coefficient of the accelerometer was studied through a thermostat. The polynomial fitting method was utilized to compensate for the temperature drift, and the error sources were analyzed. In addition, we proposed the neural network fitting method which can greatly reduce the compensation error by 2~3 orders of magnitude.

In the last part, we studied two special mode coupling phenomena——mode localization and synchronization. The mode localization mechanism between the tuning fork and the microlever was studied through finite element simulation, and the simulated results were reproduced in the experiments. The scale factor nonlinearity caused by mode veering was analyzed experimentally, and its influence on the accuracy of the sensor was quantitatively determined. Besides, we proposed the basic concept and mechanism of synchronization between MEMS resonators. Based on the theoretical model, a new sensing mechanism enhanced by synchronization was introduced. Accelerometers based on electrostatic coupling synchronization and unidirectional electro-coupling synchronization were designed respectively. Through static characterization, it is proved that the unidirectional electric synchronization can improve the zero-biased stability of the accelerometer by 5 times.

Finally, the research work of this dissertation was summarized, and future research was discussed.