Feature Extraction From Impedance Spectrum of Au/ZrO2 Nanotube/Zr-Based MIM Sensor for Selective Discrimination of VOCs: A Potential Approach for E-Nose Application

This report proposes a new method to extract features from the impedance spectrum of zirconium oxide (ZrO2) nanotube-based single sensor to detect and distinguish multiple volatile organic compounds (VOCs). The process of electrochemical anodization was utilized to synthesize a metal-insulator–metal (MIM) structured sensor consisting of ~8.5-
$\mu $
m-long ZrO2 nanotubes (Au/ZrO2 nanotube/Zr). The impedance analysis was implemented in order to examine variations in capacitance and resistance at distinct points of contact in the presence of different VOCs. The structural interdependence was determined via the utilization of circuit modeling. The parameter values that were modeled were utilized as feature vectors in the construction of a feature matrix. The feature matrix underwent principal component analysis (PCA) and linear discriminant analysis (LDA) for processing. The combined variance of the initial two components in the PCA and LDA was 88.75% and 99.71%, correspondingly. The discrimination of acetone, benzene, formaldehyde, methanol, and toluene/xylene was successfully accomplished independent of their concentration and 96.67% of accuracy was obtained using support vector machine (SVM). Moreover, the model that has been formulated integrates a distinctive technique for extracting features with a solo and unique sensor that has the ability to differentiate various VOCs.