Accurate SWASV detection of Cd(II) under the interference of Pb(II) by coupling support vector regression and feature stripping currents

Abstract

Abstract The accuracy of square wave anodic stripping voltammetry (SWASV) in detecting Cd(II) is severely disturbed by Pb(II). Compared with the complex microenvironments under Pb(II) interference, the information obtained by SWASV peak currents in conventional electrochemical analysis is too limited to accurately detect Cd(II). Thus, the extraction of SWASV signals that can obtain comprehensive information on Cd(II) and Pb(II) is necessary. This study is the first to propose feature stripping currents by coupling SWASV with chemometrics. The feature stripping currents not only focused on stripping peak currents but also considered other stripping currents and background currents. The detection performance of feature stripping currents was compared with that of stripping peak currents by modeling analysis. Support vector regression (SVR) and the partial least-square algorithms were employed to establish non-linear and linear detection models, respectively. Results showed the SVR model established using feature stripping currents had the best anti-interference capability against Pb(II) and the highest detection accuracy for Cd(II), with the root-mean-square errors of training and test datasets of 0.1877 μg/L and 0.1409 μg/L. Finally, the practicality of the method developed herein was verified by analyzing the interference of Pb(II) on Cd(II) in soil samples with a satisfactory recovery rate of 99.51%. This study provides a new solution to interference problems in SWASV analysis.