Hollow MEMS

This work describes the Master’s Thesis project carried out at the Technical University of Denmark. The goal of the project is to design, fabricate, simulate and characterize a hollow MEMS sensor able to detect individual circulating immuno cells, to study their mechanical properties and also to investigate their response during a specific treatment. Combined density, viscosity, mass spectroscopy and IR absorption spectroscopy can be performed by tracking changes in the sensor’s resonant behavior. Based on a previous work, the design is optimized in order to face the requirements of the current cells shape and dimension leading to the development of a detailed microfabrication process. Before implementing the process flow, a finite element method (FEM) analysis of the system is performed, with the goal of verifying that the chosen design would be able to measure the intended effects, thus predicting the resonant behaviour of the system. In particular, the resonant frequency is monitored during the flow of a cell through the structure in order to ensure a detectable resonance shift. The simulations are performed in different conditions showing an acceptable shift of the frequency of ∼ 1000 ppm when the cell moves from the beginning of the channel to the central point. Specifically, assuming a noise level below 100 ppm, an acceptable approximated signal to noise ratio (SNR) > 10 is obtained. At the end of the work, the fabricated structure is characterized in its resonance frequency performances. The first flexural mode is measured together with the Quality factor through the vibrometer, resulting in Quality factor values of the order of 10^3. In order to verify a sufficient frequency stability, Allan Deviation measurements are performed by employing a Phase-Locked Loop (PLL) control system to track the resonance frequency over time, confirming the theoretical behaviour of the curve and verifying that sensitivity is mainly limited by white noise at short integration times and frequency shift (random walk) towards longer integration times, with a minimum Allan Deviation of the order of ∼ 10e−7 ppm at variable integration time depending on the implemented structure, ranging from few seconds to some fraction of seconds to tens of seconds.