Feedforward Artificial Neural Network-Based Model for Predicting the Removal of Phenolic Compounds from Water by...

Area of Science:

• Green Chemistry
• Materials Science
• Environmental Science

Background:

• Deep eutectic solvents (DESs) are increasingly vital in green chemistry.
• Functionalization of multi-walled carbon nanotubes (CNTs) is explored for novel adsorbent development.
• Wastewater treatment requires efficient methods for removing persistent organic pollutants like 2,4-dichlorophenol (2,4-DCP).

Purpose of the Study:

• To synthesize and characterize novel adsorbents using DES-functionalized CNTs for 2,4-DCP removal.
• To optimize adsorption conditions (pH, dosage, contact time) for maximum 2,4-DCP removal.
• To apply a feedforward backpropagation neural network (FBPNN) for predicting adsorption capacity and validating experimental findings.

Main Methods:

• Synthesis of DES-functionalized CNTs as adsorbents.
• Batch adsorption experiments to study the effect of pH, adsorbent dosage, and contact time.
• Application of pseudo-second-order kinetic model to analyze adsorption rate.
• Development and validation of an FBPNN model to predict adsorption capacity.

Main Results:

• DES-functionalized CNTs demonstrated high efficiency in removing 2,4-DCP from aqueous solutions.
• Adsorption kinetics were accurately described by the pseudo-second-order model.
• The FBPNN model achieved high prediction accuracy (R² = 0.99) with a low mean square error (5.01 × 10⁻⁵).
• Sensitivity analysis confirmed the model's robustness in predicting the impact of experimental parameters.

Conclusions:

• DES-functionalized CNTs are promising adsorbents for 2,4-DCP removal.
• The pseudo-second-order model effectively describes the adsorption mechanism.
• FBPNN is a reliable tool for predicting adsorption capacity and optimizing wastewater treatment processes.