Gas sensor array based on carbon-based thin-film transistor for selective detection of indoor harmful gases

The identification of indoor harmful gases is imperative due to their significant threats to human health and safety. To achieve accurate identification, an effective strategy of constructing a sensor array combined with the pattern recognition algorithm is employed. Carbon-based thin-film transistors are selected as the sensor array unit, with semiconductor carbon nanotubes (CNTs) within the TFT channels modified with different metals (Au, Cu and Ti) for selective responses to NH₃, H₂S and HCHO, respectively. For accurate gas species identification, an identification mode that combines linear discriminant analysis algorithms and logistic regression classifier is developed. The test results demonstrate that by preprocessing the sensor array’s sensing data with the LDA algorithm and subsequently employing the LR classifier for identification, a 100% recognition rate can be achieved for three target gases (NH₃, H₂S and HCHO). This work provides significant guidance for future applications of chip-level gas sensors in the realms of the Internet of Things and Artificial Intelligence.