Selective discrimination of hazardous gases using one single metal oxide resistive sensor

Monitoring of hazardous gases is nowadays very important, since the urbanized environment is more subject to this kind of pollutants. Therefore, a capillary network of small gas sensors capable to check the quality of the environment is necessary. Metal oxide gas nanosensors are small economic devices that can be easily integrated in any context, however they unfortunately lack of selectivity. We present an approach using hydrothermally grown nickel oxide nanowires working at different temperatures and creating a virtual sensors array, thus exploiting the thermal fingerprints (sensor response as a function of temperature) of the gases. Using only one nanostructured material (nickel oxide) and different machine learning techniques, the system can easily discriminate any of 7 harmful gases (C2H5OH, H-2, CO, LPG, CO2, NH3 and H2S, all of them reducing gases) with an accuracy of 100%. Furthermore, the nanosensor also evaluates the gas concentration with an average error lower than 15%. Our results show that, exploiting thermal fingerprints from a temperature gradient, single metal oxide resistive nanosensors can efficiently discriminate specific hazardous gases.