Selective gas detection and quantification using a resistive sensor based on Pd-decorated soda-lime glass

Abstract

A commercial soda-lime glass slide was decorated with palladium nanoparticles by UV light irradiation. The response, limit of detection, response and recovery times of the resistive gas sensor obtained were investigated at different temperatures (300−500 °C) with four different gases (acetone, benzene, ethanol, and toluene). To overcome the main problem of this type of sensor (the lack of selectivity due to the one-dimensional output signal) a new approach was applied, which merges the sensor response values at different working temperatures. The responses obtained at five different temperatures (300−500 °C), combined into 5-dimensional points, were then analyzed using a support vector machine. After a calibration with a training dataset, the detection system was able to accurately classify (recognize the gas) and quantify (estimate its concentration) all tested gases. The results showed that this sensing system achieved perfect classification (100 %) and a good estimation of the concentration of tested gases (average error <19 % in the range 1−30 ppm). These performance demonstrate that with our approach (different temperatures and machine learning) a single resistive sensor made of glass can achieve true selectivity and good quantification, while remaining much simpler, smaller and cheaper than an electronic nose.