Aqueous Pb(II) Removal Using ZIF-60: Adsorption Studies, Response Surface Methodology and Machine Learning Predictions

Abstract

Zeolitic imidazolate frameworks (ZIFs) are increasingly gaining attention in many application fields due to their outstanding porosity and thermal stability, among other exceptional characteristics. However, in the domain of water purification via adsorption, scientists have mainly focused on ZIF-8 and, to a lesser extent, ZIF-67. The performance of other ZIFs as water decontaminants is yet to be explored. Hence, this study applied ZIF-60 for the removal of lead from aqueous solutions; this is the first time ZIF-60 has been used in any water treatment adsorption study. The synthesized ZIF-60 was subjected to characterization using FTIR, XRD and TGA. A multivariate approach was used to investigate the effect of adsorption parameters on lead removal and the findings revealed that ZIF-60 dose and lead concentration are the most significant factors affecting the response (i.e., lead removal efficiency). Further, response surface methodology-based regression models were generated. To further explore the adsorption performance of ZIF-60 in removing lead from contaminated water samples, adsorption kinetics, isotherm and thermodynamic investigations were conducted. The findings revealed that the obtained data were well-fitted by the Avrami and pseudo-first-order kinetic models, suggesting that the process is complex. The maximum adsorption capacity (qmax) was predicted to be 1905 mg/g. Thermodynamic studies revealed an endothermic and spontaneous adsorption process. Finally, the experimental data were aggregated and used for machine learning predictions using several algorithms. The model generated by the random forest algorithm proved to be the most effective on the basis of its significant correlation coefficient and minimal root mean square error (RMSE).

Recent Progress on the Adsorption of Heavy Metal Ions Pb(II) and Cu(II) from Wastewater

“…The adsorption method is easily operational and low-cost, with a low residue of heavy metal ions [ 32 ]. According to the nature of the interaction between the adsorbent surface and the adsorbate, adsorption can be divided into physical adsorption and chemical adsorption, the cause of which can be divided into dispersion force, dipole interaction, quadrupole interaction, electrostatic force, charge transfer interaction, surface modification and pore adsorption.…”

“…The poisoning mechanism of Pb(II) in the human body is to inhibit the enzyme containing sulfhydryl in cells, which harms the biochemical and physiological functions of human body, which could cause problems in early childhood neurodevelopment [ 29 , 30 ], behavioral disorders [ 31 ], a drop in IQ, etc. [ 32 ]. It can cause devastating damage not only to brain development, but also to kidney function [ 33 , 34 ], histopathological changes and oxidative damage [ 35 , 36 ].…”

An Up-to-Date Review on the Remediation of Dyes and Phenolic Compounds from Wastewaters Using Enzymes Immobilized on Emerging and Nanostructured Materials: Promises and Challenges

“…MOFs are a type of framework structure composed of metal ions or clusters coordinated with organic ligands [124]. The synthesis of these frameworks involves the combination of metals and organic skeletons, resulting in highly porous structures with tunable and ultrahigh porosity, structural flexibility, chemical and thermal stability, large surface areas, and multiple functional sites [225,249,250]. In addition, MOFs can be modified by synthetic methods that enable the introduction of specific functional groups into their organic bridging ligands, allowing for the customization of the framework to meet specific requirements or applications [251].…”

Machine Learning for Heavy Metal Removal from Water: Recent Advances and Challenges

“…According to the literature, various ML algorithms have been applied to model HM-adsorption by solid adsorbents, such as artificial neural networks (ANN), 33 tree-based methods, 34 and supporting vector regression. 35 The most widely used ML algorithm is ANN (Figure 2a). 36 ANN has different types of layers, including input, hidden, and output layers.…”