Comparison of the results of two methods of modeling the response surface and artificial neural network with experimental results of effective parameters in cobalt electrowinning

In this study, the electrowinning process was investigated in the production of metal cobalt from a low chloride solution concentration by both response surface methodology and artificial neural network. Initially, the solvent extraction process from two different feeds was performed to recover cobalt ions by dithiophosphinic acid to the organic phase and transferring it to the aqueous phase with HCl solution (1 M). Then, the metal production process was performed using the electrowinning method and examining the effective parameters such as time, pH of solution, ion concentration and current ampere. The optimization conditions for pH of aqueous solution, time, cobalt ion concentration, and ampere were obtained equal to 4, 40, 0.008 mol/L, 1 A, respectively. The results showed that the temperature of 15 ◦ C was more favorable for the process. Increasing the current ampere, time, and concentration of ions in the solution helps to increase the current efficiency. The acidity values ​​in the aqueous phase are essential for the electrowinning process and depend on the system’s conditions. The XRF and EDX analysis results showed that in optimum conditions, metal cobalt with purity above 99.8% was accumulated on the cathode surface. The modeling results of two methods showed that the artificial neural network is in better agreement with the experimental results than the response surface methodology.