Using fuzzy inference system to predict Pb (II) removal from aqueous solutions by magnetic Fe3O4/H2SO4-activated Myrtus Communis leaves carbon nanocomposite

Abstract

In this research study, a magnetic nanocomposite consisting of the Fe3O4 nanoparticles immobilized on Myrtus Communis-derived activated carbon (MM-AC) was synthesized and then, characterized by FE-SEM and FT-IR analytical methods. The results showed that the sizes of the Fe3O4 nanoparticles were about 54 nm, and the changes in the intensities of the major peaks were associated with the binding process. The adsorption efficiency of the MM-AC was evaluated for Pb (II) removal from aqueous solutions. The effective parameters such as pH, adsorbent dosage, contact time and initial metal ion concentration were optimized to reach maximum Pb (II) removal efficiency (%). The equilibrium amount of Pb (II) adsorbed onto the MM-AC suggested that the removal of Pb (II) followed Langmuir model. The kinetic studies on the removal of Pb (II) revealed that the adsorption process obeyed pseudo-second-order kinetic model. The maximum Pb (II) removal efficiency by the MM-AC was obtained at pH = 5. The adsorption capacity of Pb (II) onto the MM-AC changed from 88.65 to 480.90 mg/g by increasing the initial concentration of Pb (II) in the range of 100–400 mg/L. The comparison of maximum monolayer adsorption capacity of the MM-AC with other adsorbents reported in the literatures for removal of Pb (II) indicated that the MM-AC had better removal efficiency. In order to predict Pb (II) removal efficiency, a methodology based on fuzzy inference system (FIS) including multiple inputs and one output was developed. Four input variables namely pH, contact time (min), adsorbent dosage (g), and initial concentration of Pb (II) were fuzzified using an artificial intelligence-based approach. A Mamdani-type of fuzzy inference system was applied to implement a total of 18 rules in IF-THEN format along with a fuzzy subset consisting of a combination of Triangular and Trapezoidal membership functions in eight levels. The max-min method was employed as fuzzy inference operator, while defuzzification process was conducted using the center of gravity (COG, centroid) method. The achieved coefficient of determination value (R2> 0.99) confirmed the excellent accuracy of fuzzy logic model as a trustworthy prediction tool for Pb (II) removal efficiency. The overall results suggested that the developed material can be employed as an efficient adsorbent for Pb (II) removal from polluted aqueous solutions on a full-scale operation.