The development of an artificial neural network – genetic algorithm model (ANN-GA) for the adsorption and photocatalysis of methylene blue on a novel sulfur–nitrogen co-doped Fe(2)O(3) nanostructure surface

In this study, a new sulfur–nitrogen co-doped Fe(2)O(3) nanostructure was synthesized via a simple and efficient method and characterized via UV-Vis spectrophotometry, X-ray diffraction, field emission scanning electron microscopy, energy-dispersive X-ray spectroscopy, and Brunauer–Emmett–Teller surface area analysis. The as-synthesized nanoparticles showed high efficiency for the removal of methylene blue. The experimental conditions including the dose of the nanoparticle, the concentration of the dye, pH and the light dose were studied and optimized. The removal percentage was approximately 95% in a short time (5 min). A three-layer artificial neural network (ANN) model was proposed for predicting the efficiency of the dye removal. The network was trained using the obtained experimental data at optimum values. Some training functions were tested and their ability to predict different numbers of neurons was evaluated. The coefficient of determination (R-squared) and the mean squared error (MSE) were measured for comparison. In order to improve the accuracy of the prediction and to remove its dependency on the number of neurons, the ANN parameters were optimized using the genetic algorithm (GA). The final model results showed an acceptable agreement with experimental data. Furthermore, the relative importance of the dose of the nanoparticle, the concentration of the dye, and pH on the efficiency were obtained as 39%, 46%, and 15%, respectively. Moreover, interestingly, the obtained results showed that this newly synthesized nanoparticle has some photocatalytic properties with a band gap of 1.65 eV and therefore, it can be proposed as a low-cost visible light-driven photocatalyst for engineering applications.