Machine learning assisted electrochemical sensing technology for the detection of nitrate ions in PM2.5

Abstract

This study investigates a graphene-modified glassy carbon electrode for detecting nitrate ions in particulate matter (PM2.5). The electrode configuration consists of a glassy carbon base electrode drop-coated with chemically-reduced graphene (CRGN), followed by an optimized nitrate ion-selective membrane (NO3-ISM) layer. Electrochemical impedance spectroscopy demonstrates graphene’s ability to enhance electron transfer by mitigating water layer formation. The nitrate sensor exhibits a linear potentiometric response from 0.1 mM to 0.1 M nitrate and theoretical Nernstian sensitivity. However, chloride interferences are observed. To improve selectivity, a support vector machine model utilizing particle swarm optimization (PSO-SVM) is developed. This machine learning approach shows significantly higher predictive accuracy than univariate calibration, even in mixed nitrate-chloride samples. The combination of the graphene-modified sensor and PSO-SVM algorithm provides an optimized nitrate detection strategy, advancing technologies for air quality monitoring.