Artificial Neural Network Approach for Modeling Cobalt Extraction from Biological and Water Samples by Magnetic Nanoparticles

In this study, an artificial neural network (ANN) model was used to predict the extraction efficiency of cobalt from biological and water samples by magnetic nanoparticles based on batch solid-phase extraction and inductively coupled plasma-optical emission spectrometry (ICP-OES). The effect of operational parameters, including solution pH, amounts of the complexing agent (1-(2-pyridylazo)-2-naphthol) and nanoparticles, and extraction time was studied. The parameters were optimized for the maximum extraction of cobalt ions. The optimum conditions were as follows: initial pH 11.0, contents of complexing agent and nanoparticles 0.75 mg/l and 125 mg, respectively, and extraction time 12.5 min. After backpropagation (BP) training, the ANN model was able to predict the extraction efficiency of cobalt ions with a tangent sigmoid transfer function (tansig) at a hidden layer with 15 neurons and a linear transfer function (purelin) at an output layer. The Levenberg-Marquardt algorithm (LMA) was found as the best of 11 BP algorithms with a minimum mean squared error (MSE) of 0.009895. The linear regression between the corresponding targets and the network outputs was shown to be satisfactory with a correlation coefficient (R-2) of 0.978. Under optimum conditions, the detection limit (LOD) of this method was 7.0 ng/l, and the relative standard deviation (RSD%) was 2.1% (n = 10, c = 10 mu g/l). The method of magnetic nanoparticles based on batch solidphase extraction was applied to the separation, pre-concentration, and determination of cobalt both in biological and water samples and in a certified reference material.