Production and characterization of biochar derived from oil palm wastes, and optimization for zinc adsorption

Today, using low cost materials such as agricultural wastes as an adsorbent for heavy metals removal has gained attention in water and waste water treatment. This research aims to produce biochar (a porous material with high carbon content and low density) from three different types of oil palm wastes via pyrolysis process in a lab scale fixed bed reactor. The raw feed stocks for the pyrolysis experiment include oil palm frond (OPF), oil palm empty fruit bunches (OPEFB), and oil palm Mesocarp fiber (OPMF). The synthesized biochars were then characterized for their physiochemical properties using CHNS elemental analysis, proximate analysis, scanning electron microscopy (SEM), BET surface area, and Fourier transform infrared spectroscopy (FTIR). The adsorption capacity of produced biochars for removing zinc from aqueous solution was investigated by performing batch adsorption experiments. The result of batch adsorption experiments showed that oil palm empty fruit bunches biochar (OPEFBB) had the best efficiency for zinc removal and therefore was chosen for further optimization study. The estimation and modeling capacities of two statistical tools; response surface methodology (RSM) and artificial neural networks (ANNs) in determining and optimizing the effect of pyrolysis conditions on percentage of yield and adsorption capacity of OPEFBB toward zinc removal were evaluated. The effect of three independent variables namely: highest treatment temperature (HTT), heating rate (HR) and residence time (RT) on OPEFBB percentage of yield and adsorption capacity were determined. A central composite design was utilized to determine the effect of these factors as well as the interaction of them on responses. Based on central composite design, two second order regression models were developed for OPEFBB adsorption capacity and percentage of yield. The optimum actual values for percentage of yield and adsorption capacity were 25.49% and 15.18mg/g, respectively, under the predicted conditions of 615°C for HTT, 8°C/min for HR, and 128 minute for RT. The input and output of the RSM design was used in artificial neural networks for training purpose. The incremental back propagation algorithm demonstrated the best results and which has been used as learning algorithm for ANN in combination with Genetic Algorithm in the optimization. The estimated production conditions to reach the optimum actual values of yield at 25.38% and adsorption capacity of 15.29mg/g were HTT of 625°C, HR of 9 ̊ C/min and RT of 130 min. In both RSM and ANN methods, percentage of yield and adsorption capacity of OPEFBB were mostly influenced by the highest treatment temperature (HTT) followed by heating rate (HR) and residence time (RT). The performance of RSM and ANNs were compared in terms of root mean square error (RMSE), coefficient of determination (R²), and absolute average deviation (AAD). The results demonstrated that both models fitted the experimental data well; however the predicted values confirmed that ANN outperformed RSM due to superiority of ANN model in capturing non linear behavior and better estimating capability rather than RSM. The batch adsorption experiments for removal of zinc by optimum product were carried out by determining the impact of solution pH, biochar dosage and heavy metal concentration on the adsorption process. The results suggest that solution pH is one of the most important factors influencing the adsorption capacity. At low pHs, the removal of zinc ions was low due to high concentration of protons in sorption media and competition of protons with zinc ions for binding sites. By increasing pH, the removal of zinc showed an upward trend and reached the maximum value at pH6. After that by rising pH, precipitation and hydroxyl formation occurred which masked the true adsorption. Biochar dosage and heavy metal concentration also influenced the removal of zinc and the optimum values were found to be 10 g/l and 80 mg/l respectively. Four adsorption isotherms namely: Langmuir, Freundlich, Dubinin–Radushkevich, and Temkin were applied for modeling the adsorption equilibrium data. Among them Langmuir isotherm could describe the adsorption data better by coefficient of determination of 0.9988 and the maximum adsorption capacity was at 19.27 mg/g. From Dubinin equation, ion exchange mechanism was found to be predominant mechanism in the adsorption of zinc by OPEFBB