Machine learning techniques for predicting the adsorption capacity of Synergistic biochar Functionalization with Pyrrole-Sulfanilic acid copolymer in mercury and chromium remediation
Nesma A. Fekry,Mohamed E. Mahmoud,Nesma K. Kamel,Mohamed F. Amira
DOI: https://doi.org/10.1016/j.cej.2024.158322
IF: 15.1
2024-12-14
Chemical Engineering Journal
Abstract:A novel and sustainable nanobiosorbent was developed via immobilization of N-doped-biochar onto co-polymer matrix, specifically sulfanilic acid with pyrrole in presence of formaldehyde as a crosslinker. Firstly, N-doped-biochar was synthesized from pumpkin seeds in the presence of KHCO 3 as an activator and urea as a source of nitrogen. This product was reacted with pyrrole, sulfanilic acid as monomers and formaldehyde to generate N-PSB@Co-Poly(SU-For-Py) and thoroughly characterized using Fourier Transformation Infrared (FT-IR), scanning electron microscopy (SEM), transmission electron microscopy (TEM), energy dispersive X-ray (EDX), X-ray diffraction (XRD), X-ray photoelectron (XPS), and thermogravimetric analysis (TGA), confirming its structure morphology and elemental composition. N-PSB@Co-Poly(SU-For-Py) demonstrated significant efficacy in the removal processes of Cr(VI) (99.63 % at pH 2) and Hg(II) (95.66 % at pH 3) from aqueous solutions. The experimental results aligned with the pseudo -second-order kinetic model (R 2 = 0.991 and 0.999 for Cr(VI) and Hg(II), respectively) and established equilibrium condition after 50 min contact time for both contaminants. This research also highlights the successful recycling of the nanobiosorbent over 5 cycles, maintaining a commendable >50 % efficiency. In addition, the use of artificial intelligence (AI) models to investigate and evaluate the sorption efficiency of Cr (VI) and Hg(II) onto N-PSB@Co-Poly(SU-For-Py) was also explored. Accordingly, the two machine learning (ML) algorithms used in this prediction are based on the support vector regression (SVR) and random forest (RF). The SVR algorithm was qualified utilizing three kernels: Pearson VII universal kernel (PUK), radial basis function (RBF), and polynomial kernel (PolyK). The optimization of the algorithms was based on (i) the correlation coefficient (R) and (ii) several error functions (mean absolute error, root mean square error, Relative absolute error, and root relative squared error). The results revealed that the best-performing ML algorithms were SVR-RBF and SVR-PUK which achieved the highest value of correlation coefficient (0.9919 and 0.9989) and the lowest error for Cr(VI) and Hg(II), respectively. Therefore, the investigated N-PSB@Co-Poly(SU-For-Py) can be categorized as a promising sustainable, regenerable, and affordable nanobiosorbent for removing the toxic Cr(VI) and Hg(II) ions from water.
engineering, chemical, environmental