Development of Thermally Stable and Environmentally Robust Microelectromechanical Systems (MEMS) Based Accelerometers
Abstract
Microelectromechanical systems (MEMS) inertial sensors are widely used in many motion sensing applications including acceleration and orientation. Many of these sensors employ a small resonating mass as the main sensing element. Various applications of MEMS resonating sensors require the sensor to endure temperature and dynamic environmental parameters variation, which can cause their resonance performance to degrade. This research focuses on developing understanding of improving thermal stability of MEMS resonating inertial sensor performance improvement and its characterization in dynamic environments. Thermal stability of the resonating structure can be achieved using doping-based passive compensation methods. Analytical and numerical modeling techniques are presented predicting n-type doped silicon temperature coefficients frequencies (TCFs) for the Lamé and extensional modes of vibration of MEMS resonators by modeling of single-crystal silicon. The model approach is based on an empirical exponential expression derived for the uniaxial deformation potential as a function of doping level using the analytical and experimental data available in the literature. It is used to calculate the stiffness variation and consequently to find frequency variation with temperature change. It is observed that frequency variation with temperature is closely dependent on the direction and the variation of the different doping levels. The model result for Lamé mode shows that the predicted silicon elastic constants are in agreement with previously reported experimental values with an average error for c11 as 0.58% and for c12 as 2.12% at the room temperature, while c44 does not alter with changing n-type doping. The resonance frequency with temperature dependence is calculated, showing that the first order temperature coefficient of frequency indicates zero values to achieve higher thermal stability at doping concentration level of 0.5 and 1 × 10e19 cm-3 for Lamé and extensional mode. Apart from thermal stability, MEMS accelerometer dynamic characterization is an important aspect for accurate and efficient use in different application environments. Variation in proof mass geometry of the resonator structure design has strong impacts on performance of devices directly. Various classical and novel resonator shapes were characterized with dynamic vibration situations. Data was taken in various parameter settings of low and high frequencies at various amplitudes of vibrations in portable shaker table, and then compared to get the conclusion about the sensor characterization parameters, dynamic behaviors, and noise study. Finally, MEMS accelerometers were tested in automobiles for a variety of applications as their uses in the automotive industry will require them to function in real dynamic environments. Test data was collected based on a comprehensive test plan such as engine started conditions in lab environment but not driving, and also during driving on roads, using a lab-built readout circuitry and commercial data processors in different locations on the vehicle. Later, the time, power spectral density (PSD) and fast Fourier transform (FFT) data are plotted to compare the accelerometer’s performances. The time and frequency domain results confirm the strength and frequencies of vehicle dynamic sources as well as the driving conditions of traffic and road surfaces as proof of the satisfactory performance of inhouse and commercial sensors. The frequencies of the vibrations for various dynamic sources of the vehicle were found, as expected for the tested vehicle model and make. The measured amplitudes are between 0.07 g and 2.5 g in both cases. The doping based thermal stability model, MEMS accelerometer dynamic characterization, and their applications in automotives presented in this work can further aid in achieving temperature stability of silicon MEMS resonators and in obtaining MEMS accelerometer characterization and signal noise parameters respectively.