Industrial & Engineering Chemistry Research, Vol.53, No.49, 19117-19127, 2014
Kinetic Model of Catalytic Self-Condensation of Cyclohexanone over Amberlyst 15
The kinetics of heterogeneously catalyzed self-condensation of cyclohexanone using Amberlyst 15 as a catalyst has been determined from data obtained in a batch reactor. Temperature and catalyst concentration ranges used were 70-110 degrees C and 666 g(cat dry)center dot kg1, respectively. Runs were carried at both 5 bar of pressure, in order to avoid the evaporation of the water produced, and under vacuum conditions (0.4 bar) in order to remove the water produced. A negative impact of water on the reaction rate was found due to adsorption onto the catalyst surface and promotion of the reverse reaction. The main products of cyclohexanone self-condensation were dimers 2-(1-cyclohexenyl)cyclohexanone (D2) and 2-cyclohexylidenecyclohexanone (D3), this reaction being the first step in obtaining 2-phenylphenol. Undesired trimers were obtained significantly at the highest temperature used. Kinetic modeling was carried out, taking into account reactant and product concentrations (cyclohexanone, dimers, trimers, and water), catalyst concentration, and temperature. Both potential rate law and models based on LangmuirHinshelwood theory were tested. The Langmuir isotherm for water adsorption was also determined from experimental data. Strong water adsorption was found if water was not distillated while cyclohexanone adsorption was negligible and condensation products were weakly adsorbed onto the catalyst surface. Adsorption constants calculated for water and dimers were 0.017 and 3.63 X 10(-4) kg center dot mmol(-1), respectively. The activation energy obtained for trimer production (220.52 kJ center dot mol1) was higher than that corresponding to the dimer formation (68.46 kJ center dot mol(-1)) explaining the decrease in dimers at the higher temperature. The stability of the catalyst was analyzed in a continuous fixed bed reactor (FBR), finding that no deactivation took place in the 300 h used as time on stream. The kinetic model obtained in batch mode was also able to explain the data obtained in the FBR satisfactorily.