Molecularly Imprinted Polymer-Based Electronic Nose for Ultrasensitive, Selective Detection, and Concentration Estimation of VOC Mixtures

Abstract

This research introduces a groundbreaking electronic nose (E-Nose) that integrates advanced sensing materials with machine learning (ML). The sensing materials include molecularly imprinted polymers (MIPs) and multiwalled carbon nanotubes (MWCNTs), designed for enhanced performance. An optimized extreme learning machine (ELM) model enables highly selective detection and precise quantification of both individual and multiple volatile organic compounds (VOCs) within complex mixtures. Specifically, with transducers functionalized for specificity toward methanol, ethanol, butanol, and isopropanol, the proposed E-Nose achieved near-perfect estimation with an error of just 0.25% for individual VOCs and negligible error (0.75%-1.5%) for mixtures of two to four VOCs. The developed E-Nose demonstrated linear estimation of target VOC concentrations with high sensitivity and selectivity. Detection limits (DL) for all gases remained below safety thresholds, ensuring suitability for practical VOC sensing at room temperature (RT). Furthermore, the proposed E-Nose platform is adaptable and customizable for detecting and estimating the tested VOCs as well as other VOCs and gases, offering significant potential to revolutionize air quality monitoring.