Automation in competitive removal of toxic metal ions by fired and non-fired beads

Abstract

Researchers are investigating adsorption as a means of combating heavy metal contamination because it is both environmental friendly and economically viable. In the present work, adsorption with fired and non-fired beads was applied to remove copper, nickel and zinc from the contaminated water. The adsorption capacity of beads was modeled and predicted by Decision Tree and Random Forest regression algorithms. With the use of these algorithms, it is possible to estimate adsorption capacity with less time and minimum effort while simultaneously reducing the number of experiments required. It is an effective way to deal with environmental pollution. It was found that random forest regression showed better results than decision tree regression and fitted well with R2 in range of 0.92–0.96 and 0.92–0.95 and a mean square error ranging from 0.08 to 0.16 and 0.13–0.27 for fired and non-fired beads. However, in Decision Tree regression, R2 ranged from 0.83 to 0.85 and 0.82–0.91 and mean square error was in range of 0.57–0.72 and 0.19–1.16 for fired and non-fired beads, respectively. Copper, nickel and zinc had deviations of 0.05%, 0.01%, 0.11% and 0.03%, 0.002%, 0.04% for fired and non-fired beads between artificial neural network predicted and experimental values. The deviation was quite minimal, demonstrating that the Levenberg-Marquardt method was effective in predicting the output function of the experiment. The investigation of adsorption dynamics revealed that ternary ion adsorption on both type of beads was mixed diffusion and transfer controlled. Freundlich isotherm together with pseudo-second-order kinetic model best suited in experimental data, which revealed that adsorption of ternary ions on heterogeneous surfaces of the beads was governed by chemisorption. Fired and non-fired beads possess high adsorption capacity of 0.82, 5.26, 0.35 and 16.94, 41.66, 7.40 mg/g for Cu2+, Ni2+ and Zn2+ ions, respectively. The adsorption of metals on the surface of fired and non-fired beads was endothermic resulting in enhanced randomness at solid-liquid interface.