Analysis of the dynamic features of metal oxide sensors in response to SPME fiber gas release

Abstract In this study a metal oxide sensor array is exposed to a time-dependent amount of gas inside the sensor chamber of negligible “dead” volume. Special parameters of the response kinetics are used for multi-parametric featuring of volatile organic compounds (VOCs). The composition of the atmosphere in the chamber varies due to the time-dependent release of the VOCs from a solid phase micro-extraction (SPME) fiber into the flow of synthetic air. Four types of volatile compounds, namely acetone, acetic acid, acetaldehyde and butyric acid, that are known being frequently emitted from infected wounds, are tested in this study. The explorative data analysis (EDA) of the features is performed for the sensor outputs obtained at different carrying gas flow rates and the VOC amounts. Influence of specific aspects of the SPME based sampling on the sensor outputs is estimated. It is demonstrated by the PCA results that the target compounds cannot be distinguished below 3–4 ppm if only the sensor outputs based on the signal magnitudes are used for the featuring of VOCs (static-compatible features). The dynamic features add significant information and allow a better discrimination of the volatile compounds. The classification of the target volatile compounds can additionally be improved by precise control of the VOCs expansion in the chamber in the dynamic exposure approach.