Analyzing the anodic stripping square wave voltammetry of heavy metal ions via machine learning: Information beyond a single voltammetric peak

The application of electrochemical analysis for the detection heavy metal ions (HMIs) faces tremendous challenges due to its poor reproducibility and selectivity. It is necessary to develop fingerprint-type electroanalytical approaches for the identification of the analyte. Herein, combining principal component analysis with support vector machine classifier, we classified successfully the electroanalytical signals of six heavy metals merely based on their voltammetric peak shapes, where the absolute values of the current and potential were not taken into consideration. The variation of the measurement parameter such as the scanning frequency did not affect much on the identification of the analyte. It was found that the addition of K+ and Cl- had little influence on the identification result; while altering the electrolyte concentration, it was difficult to identify accurately the analyte, showing that the influence of the electrolyte cannot be ignored even at high concentrations. The combination of electrochemistry and machine learning is expected to improve the selectivity for the detection of HMIs in complex water environments.