Identifying Binary Mixtures of Volatile Organic Compounds With Isomeric Components Using a Single Thermal Shock-Induced Generic SnO₂ Gas Sensor

Abstract

Identifying the components and their relative proportions in the binary mixtures of the structurally similar Volatile Organic Compounds (VOCs) are required for many applications. In this work, the binary vapor mixtures of VOCs with isomeric components at variable combined concentrations ranging from 200 to 2500 ppm are discriminated using a single generic tin oxide gas sensor. Sharp fluctuations in the pellet temperature bring in more valuable analyte nature-related information into the sensor's response patterns, which can be extracted and used for discriminating among complex mixtures. The classic staircase voltage waveform fails to produce the abrupt variations in the pellet temperature facilitated via the thermal shock induction method applied for the virtual array formation out of a single tin oxide chemiresistor. Linear discriminant analysis is applied after preprocessing and feature selection of the sensor response patterns. Two simple classifiers operating based on the Euclidean distance in the feature space are utilized for analyte classification. The recognitions of binary vapor mixtures of 1-propanol, 2-propanol, 1-butanol, and 2-butanol are demonstrated as examples. A total of 546 individual experiments are carried out to confirm the accuracy of the performance; all the binary test mixtures are identified correctly; with 2.3% of error in compositions determination. The experiments are satisfactorily repeated in the form of independent work sessions within a 1-month time interval to represent the repeatability of the response patterns. The thermal shock induction appears as a replacement for the sensor arrays which suffer from the multi-dimensional drift of the array components.