Adsorption modeling of tetracycline removal by multi-walled carbon nanotube functionalized with aspartic acid and poly-pyrrole using Bayesian optimized artificial neural network

Abstract

BackgroundThe high use of tetracycline (TC) for the prevention and treatment of diseases has led to the formation of unknown compounds and various metabolized products of TC in aquatic environments. It seems that compounds containing carbon nanotubes have a good performance as an adsorbent for TC removal. MethodsMulti-Walled Carbon Nanotubes doped with aspartic acid (Asp) and polypyrrole (PPy) were synthesized to adsorb TC. The adsorbent morphology and its corresponding physicochemical properties were characterized by FTIR, XRD, FESEM, EDS and BET analysis. The adsorption capacity was investigated as a function of adsorbent dosage, pH, temperature, contact time, and initial concentration. Significant findingsThe maximum adsorption capacity was 34.50 mg/g, for an adsorbent dosage of 0.005 g, pH = 5, contact time of 60 min, and the temperature of 25℃. Reusability studies demonstrate that the removal efficiency of adsorbent after seven cycles has been more than 70%. Also, An Artificial Neural Network was developed to predict the adsorption capacity of the adsorbent. ANN performance was enhanced using Bayesian Optimization as the hyperparameter tuning scheme and 11 hyperparameters of the network as the decision variables. The improved ANN performance demonstrated a remarkable ability to model the adsorption process of TC (RMSE=0.33 and R2=0.9996).