Comparative investigation of artificial neural network and response surface approach in the optimization of indium recovery from discarded LCD screen with the presence of ionic liquids

This paper aimed to predict the solvent extraction conditions to maximize indium recovery from discarded LCD screens. Two approaches, including the response surface methodology (RSM) and the artificial neural network (ANN), were utilized to predict the efficiency of indium recovery. The main parameters, such as aqueous phase acidity (A), indium concentration (B), ionic liquid concentration (C), and aqueous to organic phase ratio (D), were investigated by using CyphosIL 101 diluted in kerosene as the organic phase. The experimental results were used to train a multilayer perceptron for creating an ANN model with the structure of one, eight, and three for input, hidden, and output layers, respectively. The optimum conditions by the RSM approach to provide the maximum efficiency of indium recovery were. 4 mol/L A, 197.79 ppm of B, 0.009 mol/L of C, and 1.58 mol/L of D. By contrast, the ANN approaches illustrated the optimal A, B, C and D equal to 4.2, 194.32, 0.0085, and 1.52 %, respectively. Two statistical approaches described the satisfactory data, and the superior data was obtained with the ANN approaches. The use of two ionic liquids verified the indium recovery from the discarded LCD screen, and 99.7 % of indium ions were separated and extracted into the stripping solution.