This thesis presents the development of wafer level vacuum packaging processes using Au-Si eutectic and glass frit bonding contributing to the improvement of packaging concepts for a variety of MEMS devices. In the first phase of this research, micromachined resonators and pirani vacuum gauges are designed for the evaluation of the vacuum package performance. These designs are verified using MATLAB and Coventorware finite element modeling tool. Designed resonators and pirani vacuum gauges and previously developed gyroscopes with lateral feedthroughs are fabricated with a newly developed Silicon-On-Glass (SOG) process. In addition to these, a process for the fabrication of similar devices with vertical feedthroughs is initiated for achieving simplified packaging process and lower parasitic capacitances. Cap wafers for both types of devices with lateral and vertical feedthroughs are designed and fabricated. The optimization of Au-Si eutectic bonding is carried out on both planar and non-planar surfaces. The bonding quality is evaluated using the deflection test, which is based on the deflection of a thinned diaphragm due to the pressure difference between inside and outside the package. A 100% yield bonding on planar surfaces is achieved at 390&ordm / C with a v holding time and bond force of 60 min and 1500 N, respectively. On the other hand, bonding on surfaces where 0.15&mu / m feedthrough lines exist can be done at 420&ordm / C with a 100% yield using same holding time and bond force. Furthermore, glass frit bonding on glass wafers with lateral feedthroughs is performed at temperatures between 435-450&ordm / C using different holding periods and bond forces. The yield is varied from %33 to %99.4 depending on the process parameters. The fabricated devices are wafer level vacuum packaged using the optimized glass frit and Au-Si eutectic bonding recipes. The performances of wafer level packages are evaluated using the integrated gyroscopes, resonators, and pirani vacuum gauges. Pressures ranging from 10 mTorr to 60 mTorr and 0.1 Torr to 0.7 Torr are observed in the glass frit packages, satisfying the requirements of various MEMS devices in the literature. It is also optically verified that Au-Si eutectic packages result in vacuum cavities, and further study is needed to quantify the vacuum level with vacuum sensors based on the resonating structures and pirani vacuum gauges.

PureFize Technologies AB develops and manufactures a broadband ultraviolet (UVC) light technology device that is mercury-free and based on nanotechnology, using the principle of field emission. The light source is made of Ti and glass, which are hermetically bonded, using a low-temperature glass frit, at elevated temperatures. The bonding procedure will induce stresses in the device originating from the mismatch of the coefficient of thermal expansion (CTE) between the materials. Brittle materials, as glass, withstands tensile stresses poorly. Therefore, the stress magnitude and distribution needs to be understood.  This work develops a quality inspection method for the glass bond and internal stresses, as well as stress simulations of the device, to be used in production at the company. The glass bond width and the internal stresses in the device were classified and analyzed by light optical microscopy and by polarised light optical microscopy. The optical analysis was followed by pressure tests of the devices using a chamber that allowed for pressurized air up to 7 bar. In parallell with the experimental work, stress and deformation simulations of the device using the finite element method (FEM) was made. Data collected from the inspections and pressure tests were compiled and analyzed, showing clear connections between the glass bond quality and the device's ability to withstand external pressure. A narrow glass bond could withstand external pressure poorly, whereas a wide glass bond could withstand external pressure well. Correlations could be made both between the glass bond appearance and the stress patterns, as well as between the FEM simulations and the stress patterns in the device. It is clear that the stresses induced in the device after bonding originates from the CTE mismatch of the bonded components when cooling it from the bonding temperature to room temperature. The pressure testing method proved to be an efficient way of verifying the maximum pressure capacity of the devices.  The knowledge from this thesis can be used when further investigating induced stresses from glas frit bonding.