Artificial Neural Network (ANN) for Modelling Adsorption of Lead (Pb (II)) from Aqueous Solution

Abstract

This study investigated the efficiency of rice husk carbon (RHC) for lead (Pb (II)) adsorption. The developed RHC was characterized by CHNS analyser, FTIR and FESEM. BET surface area, micropore area, micropore volume and average pore diameter of RHC were 58.54 m2/g, 14.53 m2/g, 0.007209 mL/g, and 45.46 A, respectively. Batch adsorption experiments were conducted to assess the effect of initial pH, contact time, initial Pb (II) concentration and RHC dose on Pb (II) removal. A contact time of 120 min and a pH value of 5 were found as optimum for Pb (II) adsorption; maximum adsorption occurred at 8 g/L of RHC dose. Artificial neural network (ANN) was applied to model Pb (II) adsorption. For prediction of Pb (II) adsorption from aqueous solution by RHC, the smallest mean square error (MSE) and the largest coefficient of determination (R2) values were, respectively, obtained as 6.0053 and 0.988567 with Levenberg–Marquardt algorithm (LMA). Hence, it was selected as the best training algorithm. Traincgf and traincgp functions followed this function with a MSE of 6.1496 and 6.2967, respectively. Adsorption of Pb (II) by RHC followed pseudo-second-order kinetics. The experimental data were described well by both Langmuir and Freundlich isotherm models. Thermodynamics study revealed that Pb (II) adsorption by RHC was spontaneous and endothermic, and the system randomness increased during the whole process. Pb (II) adsorption capacity of RHC was compared with different adsorbents. As evidenced by its high adsorption capacity, RHC can be used as an effective adsorbent for Pb (II) removal from aqueous solutions and wastewaters.