Andreas Schütze

This work studies poisoning by the cyclic siloxane octamethylcyclotetrasiloxane on a commercially available semiconductor gas sensor in TCO (temperature cycled operation). The data is evaluated using the Sauerwald-Baur model and the DSR method (differential surface reduction). The sensitivity towards several gases (volatile organic compounds, hydrogen and carbon monoxide) is evaluated and compared with a sensor in constant temperature operation mode. The physical and chemical processes on the sensitive layer as well as the resulting selectivity towards hydrogen are discussed. A feature is identified that can be derived from the Sauerwald-Baur model (the differential surface oxidation, DSO) and that quantitatively expresses the sensor condition regarding siloxane poisoning. With the help of this feature, a self-compensation of the sensor signal is demonstrated.

A gas mixing apparatus (GMA) can provide well-defined gas mixtures, which are necessary to calibrate gas sensors for quantitative measurements. This article presents a novel modular GMA and especially the representative measurements carried out to qualify and quantify its performance. New methods were developed to optimise the GMA’s settling speed and to enable self-monitoring. The GMA is able to provide up to 14 individual test gases, each in a large concentration range from sub-parts-per-billion to hundreds of parts-per-million. Additionally, the relative humidity can be set. The system is able to provide stable gas mixtures in approximately 6 min. The entire control software is realised in Python, which determines the control setpoints of all relevant system components based on the concentration input of the user. A major focus was placed on minimising the time of a gas mixture exchange to confirm design and additional software optimisation. Also, a number of techniques are shown for the analytical quantification of the system, including analysis of thermal desorption tubes by an accredited laboratory, showing maximal contaminations of e.g. only 4.18 ppb ethanol. Finally, self-monitoring of the system using photoionization detectors (PID) is demonstrated as an addition to flow readback for internal leakage detection.