Machine-learning-aided identification of ethanol in humid air using zinc complex capped CsPbBr3 resistive sensors

Abstract

In this work, a facile room-temperature (RT) solution-processing strategy was developed to synthesize perovskite CsPbBr₃ quantum dots (QDs), which were then modified with Zn-based organic ligands for enhancing their ambient environment and moisture stability. The as-prepared perovskite QDs were well dispersed in a trimethylbenzene solvent with a small particle size of ∼9 nm. Afterwards, a sensor was fabricated by spin-coating perovskite QDs on an interdigital electrode, leading to a sensitivity of 0.025 towards 400 ppm ethanol gas as well as a response/recovery time of 3.9 s/3.6 s at RT. This excellent sensing performance is attributed to the high carrier mobility and adjustable crystal structure of perovskite CsPbBr₃ QDs. Theoretical calculation results verified that the sensitivity towards ethanol might result from the formation of chemical bonds between perovskite QDs and Zn-based ligands, leading to a change of molecular orbital energy levels. Furthermore, with a combination of machine-learning algorithms, accurate recognition of ethanol was realized in an ambient environment with high humidity. Our research might be helpful to construct an intelligent sensor for application under complex conditions, such as coal mines and chemical factories.