In previous studies, the selection of optimal gas-sensing materials for detecting target gases mainly relied on their response value, but other indices, such as the recovery capability of materials, have usually been overlooked. Here, we propose a new method for evaluating sensor effectiveness that includes a broader range of performance indices. In this study, four gas sensors based on metal-oxide semiconductors (WO3, CeO2, In2O3, and SnO2) were used as examples, and their performance in the detection of four decomposition products of sulfur hexafluoride (SF6) was investigated. After gas-sensing experiments, values for working temperature, response value, and recovery capability were obtained. A multivariate evaluation method of mixing principal component analysis, information entropy, and variation coefficient was developed to calculate the weights of various indices, and the sensors' optimal working temperatures could be identified quantitatively. Using five variables (working temperature, response value, recovery capability, fluctuation rate, and detection limit), we continued to apply this multivariate evaluation method to calculate the weights and acquire comprehensive scores for the four sensors. Finally, these scores were used to identify the optimal materials for detecting SF6 decomposition products. This procedure has the potential for selecting the best sensors for other gases.
Keywords: SF6 decomposition product; gas sensor; information entropy; multivariate evaluation; principal component analysis; variation coefficient.