MACHINE LEARNING AND STATISTICAL EXPERIMENTAL DESIGN METHODS FOR MODELING OF FACILITATED ADSORPTION OF METHYLENE BLUE DYE: ISOTHERM, KINETIC AND DESIRABILITY FUNCTION (DF)

This study is focused on usage of zinc sulfide nanoparticle loaded on activated carbon for adsorption of methylene blue from aqueous solutions. The properties of ZnS - NPs-AC were identified by Xray diffraction; field emission scanning electron microscopy and Fourier transform infrared spectroscopy. Response surface methodologyand support vector machine have been used for the optimization and modeling of pH, adsorbent mass MB concentration, and sonication time to develop predictive models for simulation the efficiency of methylene blue adsorption based on experimental data set achieved in batch study.The obtained results using LS-SVM (nonlinear approaches) have a better performance than of the CCD for prediction of understudy dyes adsorption. The significant factors were optimized using central composite design (CCD) combined with a Desirability function (DF) approach. The optimal conditions based on response (removal of MB) (pH 7.0, 0.015 g of ZnS-NPs-AC, 20 mg L-1 MB after 3 min sonication) were found out. At this optimum condition, the removal percentage was obtained 98.02% by DF approach. The adsorption equilibrium was well described by the Langmuir isotherm model with maximum mono-layer adsorption capacity of 243.90 mg g-1. Kinetic and rate of MB adsorption follow pseudo second-order kinetic model.