Superadsorbent Fe3O4-coated carbon black nanocomposite for separation of light rare earth elements from aqueous solution: GMDH-based Neural Network and sensitivity analysis

A series of nanocomposites adsorbents with different magnetite/carbon black ratios were synthesized by using the co-precipitation method and used for separation of LREEs (Ce, La, and Nd) from aqueous solution. The adsorption efficiency of nanocomposites is strongly dependent on both pH and the loading carbon on nanocomposite. The maximum adsorption capacity (370 mg/g) was reported by nanocomposite with 20% Fe3O4 and 80% carbon in pH 7 for LREE initial concentration of 250 ppm. Results revealed that the LREEs adsorption behavior of the optimal adsorbent fits well with Langmuir isotherm and pseudo-first-order kinetic model. Moreover, the average values of thermodynamic parameters suggest the endothermic and irreversible chemisorption mechanism. An empirical correlation was obtained by using GMDH (Group Method of Data Handling)based Neural Network to predict the adsorption kinetics of LREEs as a function of ion's electronegativity, molecular weight, and initial concentration. The results showed that the correlation can predict the experimental data mostly lower than 12.5% and it can predict the results of other researches with similar conditions with up to 25% from the experimental values. Finally, the results of sensitivity analysis revealed that the adsorption of LREEs is more sensitive to ions electronegativity and molecular weight at equilibrium conditions.

Automated summary

A series of nanocomposites adsorbents with different magnetite/carbon black ratios were synthesized for separating LREEs, showing high adsorption efficiency and a maximum adsorption capacity of 370 mg/g. The adsorption behavior fits well with Langmuir isotherm and pseudo-first-order kinetic model. Using a Neural Network to predict the adsorption kinetics of LREEs showed high accuracy and validation with other research studies.