Artificial intelligence and regression analysis for Cd(II) ion biosorption from aqueous solution by Gossypium barbadense waste.

In this study, batch biosorption experiments were conducted to determine the removal efficiency of Cd(II) ion from aqueous solutions by Gossypium barbadense waste. The biosorbent was characterized by Fourier transform infrared spectroscopy (FTIR) and scanning electron microscopy (SEM) connected with energy dispersive X-ray (EDX). The sorption mechanism was described by complexation/chelation of Cd2+ with the functional groups of O-H, C=O, -COO-, and C-O, as well as, cation-exchange with Mg2+ and K+. At initial Cd(II) ion concentration (C o), 50 mg/L, the adsorption equilibrium of 89.2% was achieved after 15 min under the optimum experimental factors of pH 6.0, biosorbent dosage 10 g/L, and particle diameter 0.125-0.25 mm. Both Langmuir and Freundlich models fitted well to the sorption data, suggesting the co-existence of monolayer coverage along with heterogenous surface biosorption. Artificial neural network (ANN) with a structure of 5-10-1 was performed to predict the Cd(II) ion removal efficiency. The ANN model provided high fit (R 2 0.923) to the experimental data and indicated that C o was the most influential input. A pure-quadratic model was developed to determine the effects of experimental factors on Cd(II) ion removal efficiency, which indicated the limiting nature of pH and biosorbent dosage on Cd(II) adsorption. Based on the regression model (R 2 0.873), the optimum experimental factors were pH 7.61, biosorbent dosage 24.74 g/L, particle size 0.125-0.25 mm, and adsorption time 109.77 min, achieving Cd2+ removal of almost 100% at C o 50 mg/L.