Artificial Neural Network Approach Modeling for Sorption of Cobalt from Aqueous Solution Using Modified Maghemite Nanoparticles

Abstract

This research defines the utilization of the artificial neural network (ANN) for demonstrating the sorption percentage of cobalt from an aqueous solution using modified maghemite nanoparticles. The effect of operating conditions such as temperature (°C), initial cobalt concentration, initial pH, contact time (min), and sorbent mass (g) are focused on the best conditions for maximum cobalt ions removal. Prepared nanoparticles were described using X-ray diffraction (XRD), scanning electron microscopy (SEM), X-ray fluorescence (XRF), and Fourier transform infrared spectroscopy (FTIR) measurements. The Langmuir model had been adequately matched with the experimental equilibrium data. Kinetic data demonstrate that the pseudo-second-order and intraparticle diffusion models regulate the kinetic processes of sorption. An ANN model was developed using 25 data sets for training, five data sets for validation, and 10 data sets for testing by a single-layer feedforward back-propagation network with 20 neurons to get a minimum mean square error (MSE). A tanh-sigmoid was used as the activation function for input and purelin for output layers. The high correlation coefficient (R2)=1 for trained data; R2=0.998 for tested data; MSE=3.78×10−28 of the trained data; and MSE=6.0513×10−9 for tested data between the model, and the experimental data revealed that the model could forecast the release of cobalt from an aqueous solution using modified maghemite nanoparticles efficiently.