Developing the novel model for kinetics and mass transfer in sun-light-driven photodegradation of malachite green over Bi2O2CO3/CuBi2O4 (50:50) flower-like nanophotocatalyst: Machine and deep-learning algorithms

Abstract

The flower-like Bi2O2CO3-CuBi2O4 (50:50) nanocomposite was fabricated using the sono-hydrothermal procedure, including the nanosheets as Bi2O2CO3 and spherical clusters as CuBi2O4), and was employed for photodegradation of malachite green with the high-concentration under the visible light region that the high elimination percentage of nearly 91.6% was achieved in 180 min. Various analyses were carried out using XRD, DRS, BET-BJH, FESEM, TEM, and pHpzc techniques over Bi2O2CO3-CuBi2O4 (50:50). Furthermore, for the modeling of the photodegradation process, simplifying the complexity of the photodegradation process, and the influence of operating parameters on the process, advanced intelligent techniques, including deep learning algorithms and machine learning algorithms such as Gaussian Process Regression (GPR) and Support Vector Machines (SVM) were employed that was achieved remarkable accuracy with coefficients of determination (R2) exceeding 0.99. In these models, the initial concentration of MG, pH solution, photocatalyst loading, and process time served as inputs, while the conversion of MG was the output. Subsequently, based on the AOP fundamentals (active species and radicals), a novel kinetic model was developed, and the optimization of kinetic parameters for our novel kinetic model using the Genetic algorithm was conducted (R2 =0.9862). In continuation, to explore other aspects and kinetic parameters for a comprehensive understanding of the complexity of the process, the power polynomial model (R2 =0.9446) and previously developed kinetic models were investigated using the Genetic algorithm. Besides, a novel modelling approach was introduced, employing a combination of Computational Fluid Dynamics (CFD) and Genetic Algorithm (GA) techniques. This innovative method was utilized to accurately estimate the physical properties of synthetic nanophotocatalysts without the need for traditional laboratory analysis based on the transport phenomena. Furthermore, with this model, we successfully have determined mass transfer parameters in both the adsorption and photodegradation processes. Moreover, this model enabled the determination of the surface reaction rate of the nanophotocatalyst −Rdye=(8.9874e−8)Cpr;r=Rp1.0381 [ug/(cm2.s)], contributing to a thorough comprehension of photocatalytic activity with high accuracy.