Evolvulus alsinoides plant mediated synthesis of Ag2O nanoparticles for the removal of Cr(VI) ions from aqueous solution: modeling of experimental data using artificial neural network. (June 2022)

• Record Type: Journal Article
• Authors: Kavitha, B.; Deepa, R.; Sivakumar, S.
• Is Part Of: Materials today sustainability. Volume 18(2022)
• Publication Date: 2022-06
• Subjects: Mean squared error -- Silver oxide -- Isotherm model -- Thermodynamics -- Levenberg-Marquardt algorithm; Materials science -- Environmental aspects -- Periodicals; Sustainable engineering -- Periodicals; 620.11
• DOI: 10.1016/j.mtsust.2022.100124
• Languages: English
• ISSNs: 2589-2347

Abstract

Abstract: The objective of this research is to evaluate the applicability of Ag2 O nanoparticles prepared using the Evolvulus alsinoides plant extract by green method for the removal of Cr(VI) ions from aqueous solution. The adsorbents were characterized by several instrumental techniques, whereas the adsorption experiments were performed in a laboratory batch process using diphenylcarbazide method. The X-ray diffraction pattern proved that the synthesized nanoparticles were face-centered cubic phase structure with an average crystalline size of 47 nm. The optimum conditions for Cr(VI)ions adsorption were found to be pH 2, contact period of 60 min, and an initial Cr(VI) ions concentration of 100 mg/L. The isotherm model was fitted with Langmuir adsorption isotherm for both adsorbents. According to the study, Cr(VI) ions have a maximum adsorption capacity of 99.05 mg/g. The pseudo-second-order model and the kinetic data were in good agreement. The effect of temperature and thermodynamics confirmed that adsorption was spontaneous, feasible, and endothermic in nature. In a competitive environment, the ion selectivity was also tested in the presence of a variety of competing ions, and Cr(VI) ions were preferentially removed. Artificial neural network modeling of Cr(VI) ions adsorption was used to obtain the best values of the variables for highest removal efficiency. With a lowest mean squared error of 0.0056, the Levenberg–Marquardt algorithm was judged to be the best. Higher R 2 and lower mean squared error values were obtained with the ideal artificial neural network architecture with a 6-12-1 topology. The linear regression between the network outputs and the generated targets was found to be good, with a correlation coefficient of around 0.995, and the experimental data was best matched to the artificial neural network. Graphical abstract: Image 1