Studio e progettazione di sensori di gas a base di ossido di stagno = Study and design of tin oxide-based gas sensors

Considering the importance of monitoring toxic and hazardous gases in the air due to their impact on the environment and human health, it is crucial to have gas sensors capable of detecting even very low concentrations with good response time, recovery time, selectivity, sensitivity, reproducibility, and long-term stability. In this work, tin oxide nanowires (characterized by a high surface-to-volume ratio) are investigated as gas sensors. Tin oxide is chosen thanks to its sensitivity and capability of chemically interact with gases like NO2 which is one of the most critical gases; it can change its electrical properties based on the concentration of the gas. Its adsorption performance, good dispersion, and chemical stability at a low cost make it an excellent candidate for the creation of metal oxide-based gas sensors. The chemical and electronic properties of tin oxide are discussed to better understand the mechanism that dominates the detection process. The device model, consisting of two aluminium pads and the sensitive nanowire that connect them, is simulated using Sentaurus software with proprietary code. The entire fabrication process is simulated, including the deposition of the various materials, the etching and so on. The created model geometry is then meshed for FEM-based electrical characterization, and it is characterized for various applied voltages. Initial simulations provide a first result of what can be expected from the next experimental measures. In the Fondazione Bruno Kessler (FBK) laboratories in Trento, sensors have been fabricated on four different wafers (varying in nanowire thickness), each containing 49 devices with the same parameters. After fabrication, these devices are tested in clean air and at room temperature to check for good contacts between the metal pads. The current behaviour is compared with data extrapolated from Sentaurus, revealing some important discrepancies. This comparison allows for adjusting the concentration of arsenic doping, which simulates the presence of oxygen vacancies in tin oxide. These vacancies are responsible for its semiconductor behaviour. The used methodology establishes a relation between doping and vacancies, which is valuable for future simulations before starting technological processes. The sensors can then be tested under gas conditions at higher temperatures. In this scenario, a significant connection to the simulation domain is found: the presence of gas increases the resistance of the nanowires, as well as a specific density of trap levels in the electrical simulation. Both the density of trap states and their distance from the conduction band can be adjusted, fitting the two conditions. In conclusion, a relation between doping concentration and oxygen vacancies has been established. For future projects, where sensors are tested under gas conditions, also a correspondence between gas concentration and trap levels can be determined. Moreover, it should be conducted also long-term testing to evaluate the stability and reliability of sensors to identify and mitigate potential issues of degradation of materials and sensor performance over time. In addition, for future works it should be an opportunity to fix the wrong simulation behaviour that is probably caused by some junction phenomena between metal and oxide which have not been considered.