Adsorption of Congo red from aqueous solution onto KOH-activated biochar produced via pyrolysis of pine cone and modeling of the process using artificial neural network

Abstract

Most of dyes cause various environmental and health problems due to their toxic, mutagenic, and even carcinogenic properties. Therefore, several treatment methods are used to remove dyes from wastewater. Adsorption is one of the most preferred methods due to its easy application and high efficiency. The aim of this study is to prepare and characterize KOH-activated pine cone (APC) biochar and use it as adsorbent for removal of anionic diazo dye, Congo red (CR) from aqueous solution. The various operating parameters such as pH, contact time, temperature, initial dye concentration, and adsorbent dosage are optimized in batch adsorption system. Experimental results showed that the prepared APC biochar has a surface area of 1714.5 m(2)/g and was achieved 94.62% CR removal efficiency at an adsorbent dosage of 2 g/L. The Freundlich, Langmuir, and Temkin adsorption models were used for the mathematical description of the adsorption equilibrium. Experimental data showed the best compatibility with the Freundlich isotherm. Batch adsorption models, based on the assumption of the pseudo first-order, pseudo second-order, and intra particle diffusion mechanisms, were applied to examine the kinetics of the adsorption. Kinetic data fitted the pseudo second-order kinetic model. Calculated thermodynamic parameters indicated the spontaneous, endothermic, and the increased randomness nature of CR adsorption. Structural and morphological changes of APC biochar after adsorption process were determined by using Fourier-transform infrared spectroscopy (FT-IR) and scanning electron microscope (SEM) analysis. The prediction of the CR adsorption capacity of the APC biochar using artificial neural network (ANN) algorithm was modeled. For this purpose, many different ANN models have been developed. Among them, ANN10 gave the best results. According to ANN10 results, root-mean-squared error (RMSE), mean bias error (MBE), mean absolute error (MAE), and correlation coefficient (R-2) were estimated as 0.770, 0.310, 0.037, and 0.999, respectively. Consequently, the prediction results showed well agreement with experimental results.