Enhanced uptake of antimonite onto Fe-Zr oxide-loaded MXene: Mass transfer and machine learning data mining

This work proposed introducing nanoscale Fe-Zr oxides onto the layered-surfaces of MXene carriers to maintain well dispersion and anchoring stabilization. The results indicated that Fe-Zr oxide/MXene has a superior adsorption capacity of 209.1 mg/g at pH 7.0. The mass transfer of antimonite onto Fe-Zr oxide/MXene was investigated by numerical solving the homogeneous surface diffusion model. The external mass transfer coefficient ( k f ) and surface diffusion coefficient ( D S ) were determined to be 6.48 × 10 −4 and 1.86 × 10 −8  cm/min, respectively, and the latter was rate-limited. The 2-D profiles of Sb(III) diffusion within the adsorbents were visualized. Data mining and knowledge discovery using machine learning models indicated that the gradient boosting regression (GBR) model successfully predicted the test datasets. The operational conditions outweigh adsorbent properties in capacity prediction, and equilibrium concentration and adsorbent dose, are the most two important features. Using the trained GBR model, the Sb(III) adsorption capacities of MXene (Ti 3 C 2 ) and Fe-Zr/MXene were well predicted. This work provides insights on understanding the mass transfer mechanisms of Sb(III) in functionalized-MXene adsorbents and employing machine learning models for tuning the treatments for better achievements.