Contents

Interference-Partitioned Multichannel Calibration of CuO–NiO–ZnO Chemiresistive Arrays for VOC Identification and Low-Concentration Reliability

Author(s): James Carter1, Daniel Thompson2
1Center for Chromatographic Research Midland Scientific University Canada
2Department of Environmental Analytical Chemistry Great Lakes University United States
James Carter
Center for Chromatographic Research Midland Scientific University Canada
Daniel Thompson
Department of Environmental Analytical Chemistry Great Lakes University United States

Abstract

Selectively reporting VOCs from miniaturised metal oxide sensors is challenging especially when the analyte is accompanied by interfering gases of chemical activity. The work investigates an array of three CuO-NiO-ZnO chemiresistors for selective reporting of acetone, toluene, ethanol, and chloroform from either single-gas, binary-mixture, or ternary-mixture environments. The central questions include whether a small array can sustain selective recognition and low-concentration accuracy as mixture complexity increases and whether analyte-specific error provides information that is obscured when considering classification values. The data set comprises 2,241,436 observations of the one-gas regime, 227,617 observations of the two-gas regime, and 131,120 observations of the three-gas regime. An inference-partitioned calibration method is used to interpret the oxide channel responses based on the response polarity, transient characterisation, mixture background, and error analysis as function of gas concentration. Classification is shown to be highly discriminative using K-nearest-neighbour (KNN) and random forest models at 99.03290% and 98.70436%, respectively, for the three-gas regime. Linear classifiers, however, exhibit degraded performance. While global quantitative prediction is successful, lower-concentration figures of merit decline due to the presence of ternary interferents. Chloroform was shown to be the most difficult analyte to quantify because of its high root mean square error, detection limit, and quantification limit. With an interference partitioned calibration method, the detection limit for chloroform was reduced from 0.02478 ppm to 0.02070 ppm, whereas the quantification limit was reduced from 0.08260 ppm to 0.06900 ppm. Concluding, miniaturised arrays of CuO-NiO-ZnO sensors can provide reliable recognition of multiple VOCs, yet selective reporting requires more than just class identity and should consider analyte-specific error, detection limits, and quantification limits, among other parameters.

Keywords: VOC sensing; chemiresistive sensor array; CuO; NiO; ZnO; metal-oxide sensors; mixture interference; breath-marker analysis; multivariate calibration; K-nearest-neighbour; Random Forest
Copyright © 2025 James Carter, Daniel Thompson. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.