Highly sensitive chemoresistor based on Ag/TiO2/FTO sandwich structure for evaluation of component concentration in mixed ambient

•Memristor-based gas sensor for detecting methanol-ethanol mixtures.•Gas-sensitive performance in different resistive states evaluated by simulation.•The recovery time of the device under direct current bias was only 0.7 s.•Neural network algorithms were used to predict gas concentrations. The performance of membrane electrode assemblies (MEAs) in direct methanol fuel cells (DMFCs) can be greatly affected by ethanol as an impurity in methanol fuels. It is thus necessary to explore an effective method for accurately detecting the ethanol concentration in a mixed atmosphere of methanol and ethanol. Herein, we developed a chemoresistor based on Ag/TiO2/FTO sandwich structure and achieved accurate prediction for the concentration of each component in methanol-ethanol mixed atmosphere. First principles calculations were first performed to reveal the gas adsorption characteristics of TiO2 sensitive layer in different resistive state. It was indicated that methanol and ethanol molecules could be clearly distinguished by TiO2-based device in the high resistive state. In the gas-sensing test, the response value of Ag/TiO2/FTO chemoresistor in the high resistive state to 1 ppm methanol was 2.55, 14.8, 13.19, 15.04 and 17.34 times higher than that of the same concentration of ethanol, acetone, isopropanol, acetic acid and methyl formate, respectively. Meanwhile, when the device was in direct current mode after the response, an ultra-fast recovery characteristic (0.7 s) was observed due to the resistive switching function. Finally, the prediction of the component gas concentration in the methanol-ethanol mixed atmosphere was achieved by combining the optimized neural network algorithm. The present work brings new inspiration for future research on gas sensors.