Effects of chlorine on ethylene epoxidation over silver

To study the effect of chlorine on ethylene epoxidation over silver powder, three types of experiments were used: steady-state-kinetic, chlorine-step-input and isotopic-oxygen-switch experiments. To get reliable and reproducible kinetic data, we pretreat the silver catalyst with at least six oxidation/reduction cycles, and use high reactant flow rate--300 cc/min. Steady-state experiments show that the reaction orders of reactants for both carbon dioxide and ethylene oxide formations decrease with increasing reactant concentration. The order in O$\sb2$ goes from 1 to 0.5 and in C$\sb2$H$\sb4$ from 0.5 to 1. For a fresh catalyst, selectivity is 39% under reactant-sufficient conditions and decreases to 25% under reactant-deficient conditions. Selectivity increases as flow rate and temperature increases. Chlorine-step-input experiments show that selectivity increases and activity decreases with increasing chlorine. Activation energies for carbon dioxide (24 kcal/gmol) and ethylene oxide (19 kcal/gmol) do not change with added chlorine, indicating no change of the rate-determining step. With long-term treatment of chlorine, activity and selectivity reach steady states, resulting from equilibrium of chlorine adsorption and desorption rates over silver. Selectivity at high ethylene composition is low because of an increasing removal rate of chlorine by ethylene, confirmed by XPS measurements. The kinetic experiments after oxidation and reduction treatment of the chlorine treated catalysts show that chlorine penetrates into the subsurface of silver. This finding is also supported by XPS measurements. The subsurface chlorine reappears on the silver surface when surface oxygen exists. Isotopic-oxygen-switch experiments show that there is no measurable reversibly-adsorbed oxygen, but that there is appreciable irreversibly-adsorbed oxygen responsible for epoxidation reaction. A model is proposed to account for the statistical product distribution of oxygen isotopes and for carbon dioxide scrambling experiments. Quantitative analysis of the data shows that the sizes of an irreversibly-adsorbed-oxygen pool and a carbonate-oxygen pool are 10% and 4% of a monolayer, respectively, for a fresh catalyst. The size of the inactive oxygen pool, which does not participate in ethylene epoxidation reaction but exchanges oxygen with an active oxygen pool, is about 10 times that of an active oxygen pool. All three oxygen pools increase with chlorine addition