Utilization of a double-cross-linked amino-functionalized three-dimensional graphene networks as a monolithic adsorbent for methyl orange removal: Equilibrium, kinetics, thermodynamics and artificial neural network modeling

Herein, it is aimed to develop a high-performance monolithic adsorbent to be utilized in methyl orange (MO) adsorption. Therefore, amino-functionalized three-dimensional graphene networks (3D-GNf) fulfilling the requirements of reusability and high capacity have been fabricated via hydrothermal self-assembly approach followed by a double-crosslinking strategy. The potential utilization of 3D-GNf as an adsorbent for removal MO has been assessed using both batch-adsorption studies and an artificial neural network (ANN) approach. Graphene oxide sheets have been amino-functionalized and cross-linked, by ethylenediamine (EDA) during hydrothermal treatment, following the glutaraldehyde has used as a double-crosslinking agent to facilitate the crosslinking of architecture. The successful fabrication of 3D-GNf has been confirmed by field-emission scanning electron microscopy (FESEM), Fourier transform infrared (FT-IR), Raman and X-ray photoelectron spectroscopy (XPS). Moreover, N2 adsorption/desorption isotherms have revealed the high specific surface area (1015 m2 g-1) with high pore volume (1.054 cm3 g-1) and hierarchical porous structure of 3D-GNf. The effect of initial concentration, contact time, and temperature on adsorption capacity have been thoroughly studied, and the kinetics, isotherms, and thermodynamics of MO adsorption have been modelled. The MO adsorption has been well defined by the pseudo-second-order kinetic model and Langmuir isotherm model with a monolayer adsorption capacity of 270.27 mg g-1 at 25 degrees C. The thermodynamic findings have revealed MO adsorption has occurred spontaneously with an endothermic process. The Levenberg-Marquardt backpropagation algorithm has been implemented to train the ANN model, which has used the activation functions of tansig and purelin functions at the hidden and output layers, respectively. An optimum ANN model with high-performance metrics (coefficient of determination, R2 = 0.9995; mean squared error, MSE = 0.0008) composed of three hidden layers with 5 neurons in each layer was constructed to forecast MO adsorption. The findings have shown that experimental results are consistent with ANN-based data, implying that the suggested ANN model may be used to forecast cationic dye adsorption.