Classification of Volatile Organic Compounds Using a Novel High-Frequency Quartz Crystal Microbalance Sensor Array Based on Ethyl Cellulose Microstructures

Detecting volatile organic compounds (VOCs) is essential because they affect air quality and can serve as biomarkers for non-invasive early disease detection. VOCs can be identified using electronic noses combining cross-reactive sensors, pattern recognition, and classification algorithms. Here, a single-polymer 12-sensor array based on 30 MHz quartz crystal microbalances was developed using three ethyl cellulose (EC) microstructures: casting films (CFs), anti-solvent microparticles (ASMs), and electrospray microparticles (ESMs). The array detected ethanol, ethyl acetate, and heptane at ppm concentrations under controlled conditions of 20 °C and 20% relative humidity. Scanning electron micrographs exhibited average mesh-hole diameters of 1.13 ± 0.44 μm for CFs, and average particle diameters of 60 ± 15 nm and 1.1 ± 0.3 μm for ASMs and ESMs, respectively. FTIR spectra indicated O–H and H–O–H bending bands for ASMs and EC, whereas these bands were notably attenuated for ESMs, suggesting their affinities for polar and non-polar compounds, respectively. Mahalanobis distance and a support vector machine classifier accurately discriminated VOCs in principal component analysis space. Principal component regression and partial least squares regression predicted VOC concentrations with R2 > 0.99, achieving estimated limits of detection of 143.6, 57, and 42.1 ppm by PCR and 70.5, 32.8, and 26.5 ppm by PLSR for ethanol, ethyl acetate, and heptane, respectively.