Kinetic model: Effect of the operation conditions on the selective glycerol oxidation with Au supported catalysts
Résumé
The liquid-phase oxidation of glycerol has been studied using Au nanoparticles supported on different carbon materials: activated carbon (AC), graphite (G) and carbon ribbon-type nanofibers (CNF-R) [1]. The influence of the operating conditions (e.g. oxygen pressure, the reaction temperature, the NaOH/glycerol molar ratio and the catalyst amount) on the catalytic activity and product selectivity was studied in detail. Results have showed that the product distribution for the liquid phase oxidation of glycerol is clearly dependent on the reaction conditions. The oxygen pressure was found to have a very small influence on the reaction rates, being the reaction of zero order respect to oxygen. The selectivity to glyceric acid was significantly enhanced on increasing both the oxygen pressure and reaction temperature, which simultaneously led to a decrease in both glycolic and tartronic acid selectivity. Glycerol conversion and glyceric acid selectivity increased when the NaOH/glycerol molar ratio was raised from 1 to 2 and it remained practically unchanged at higher values. The increase of the glyceric acid selectivity has been attributed to the formation/decomposition of hydrogen peroxide during the reaction. This fact is directly related to the formation of glycolic acid, attributed to C-C bond rupture. Therefore, an increase in the oxygen pressure minimized the formation of hydrogen peroxide and an increase in temperature or the amount of NaOH produced the decomposition of hydrogen peroxide (2H2O2 2H2O + O2), favoring the selectivity to glyceric acid. Moreover, a linear correlation was observed between glycerol conversion and the amount of catalyst (up to 0.5 g), indicating that the reaction was kinetically controlled. A kinetic model was proposed on the basis of the obtained reaction results.A VBA-Excel application was developed to evaluate the kinetic parameters [2]. The Bader-Deuflhard method was used to evaluate the set of ordinary differential equations, whereas the Marquardt-Levenberg algorithm was used in the nonlinear regression procedure. In order to identify the reaction that best suited our experimental results, two reaction networks were considered: the reaction networks proposed by N. Dimitratos [3] and S. Demirel et al. [4]. Dihydroxyacetone and glyceraldehyde were not included in the reaction scheme because these intermediates were not observed experimentally in the present work. In addition, the intermediate reaction (glyceric acid to hydroxypyruvic acid) was not included due to the low experimental concentration of the latter. Nevertheless, the intermediate reaction (glyceric acid to tartronic acid) and the influence of the NaOH concentration were included. Thus, good agreement between the experimental and predicted data was confirmed (Figure 1). The fitting allowed the partial reaction orders n and nbase to be fitted to 0 and 0.5, respectively. These results confirmed that the oxidative dehydrogenation mechanism itself is zero-order with respect to oxygen. The F-test showed that the regression could be considered to be meaningful, because the value of the Fc/Ftest ratio was greater than 1. However, the obtained results show that oxalic acid was mainly produced by the oxidation of glycolic acid and not by the direct oxidation of mesooxalic acid.