Detection of copper, lead, cadmium and iron in wine using electronic tongue sensor system

An array of 10 potentiometric chemical sensors has been applied to the detection of total Fe, Cu, Pb and Cd content in digested wine. As digestion of organic matter of wine is necessary prior to the trace metal detection using potentiometric sensors, sample preparation procedures have been optimized. Different variants of wet and microwave digestion and dry ashing, 14 conditions in total, have been tested. Decomposition of organic matter was assessed using Fourier transform mid-infrared spectroscopy and total phenolic content. Dry ashing was found to be the most effective method of wine digestion. Measurements with sensors in individual solutions of Fe(III), Cu(II), Pb(II) and Cd(II) prepared on different backgrounds have shown that their detection limits were below typical concentration levels of these metals in wines and, in the case of Cu, Pb and Cd below maximum allowed concentrations. Detection of Fe in digested wine samples was possible using discrete iron-sensitive sensors with chalcogenide glass membranes with RMSEP of 0.05mmolL−1 in the concentration range from 0.0786 to 0.472mmolL−1. Low concentration levels of Cu, Pb and Cd in wine and cross-sensitivity of respective sensors resulted in the non-linearity of their responses, requiring back-propagation neural network for the calibration. Calibration models have been calculated using measurements in the model mixed solutions containing all three metals and a set of digested wine sample. RMSEP values for Cu, Pb and Cd were 3.9, 39 and 1.2μmolL−1 in model solutions and 2, 150 and 1μmolL−1 in digested wine samples. [Display omitted] •Detection of trace metals in wine using an array of potentiometric chemical sensors.•Optimized procedure of wine digestion for metal detection using sensors.•Detection of copper, lead and cadmium using an array of potentiometric sensors.•Detection of iron by direct potentiometry with selective sensor.•ANN calibration models performed better than PLS for metal determination.