Characterization of Silicon Waveguides For Non-Dispersive Infrared Gas Sensors

Carbon dioxide is an important gas for life on Earth. But as human activities have been expanding throughout modern history, the CO2 concentration in the atmosphere is increasing. High concentrations of carbon dioxide can lead to various consequences, such as climate change and poor air quality both indoors and outdoors. It is therefore of importance to detect this gas, in order to understand our environment, and to avoid health impacts that it may cause. Non-dispersive infrared sensors are widely used in CO2 sensing and are based on optical absorption technology. This thesis investigates the optical performance of suspended waveguides for non-dispersive infrared sensors, with regard to different material qualities, i.e. monocrystalline and polycrystalline silicon, and geometries of these waveguides. The waveguides that are studied in this thesis consist of splitters, and at the end of each splitter a grating coupler that projects the IR radiation perpendicularly from the plane of the chip. Measurements are conducted to evaluate the IR radiation propagation loss of the waveguides and their feasibility for sensing carbon dioxide. It has been found that longer waveguides suffer from high propagation losses. When comparing the polycrystalline silicon with monocrystalline silicon waveguides, it has been observed in the measurements that the IR radiation propagates better in monocrystalline silicon waveguides than in polycrystalline silicon because of their crystal structures. The measured propagation loss in polycrystalline silicon waveguides is less than the loss obtained for the monocrystalline silicon waveguides, although some intensities from the grating couplers are excluded in the calculations, due to low signal strength. It is also concluded that the studied waveguides are feasible for detecting carbon dioxide with a concentration of 1%. Further investigation regarding the feasibility of gas sensing using lower concentrations of CO2 would be interesting for future work