Journal of Catalysis 226 (2004) 334–342 www.elsevier.com/locate/jcat Deactivation kinetics of Ag/Al 2 O 3 catalyst for ethylene epoxidation G. Boskovic 1 , N. Dropka ∗ , D. Wolf, A. Brückner, M. Baerns Institute for Applied Chemistry Berlin-Adlershof, Richard-Willstätter-Str.12, D-12489 Berlin, Germany Received 20 January 2004; revised 28 May 2004; accepted 2 June 2004 Available online 2 July 2004 Abstract Deactivation kinetics of a commercial Ag/Al 2 O 3 catalyst was investigated using accelerated deactivation tests in a Berty-type gradientless recycle reactor. Separability of catalytic reaction and deactivation kinetics were established by operation in a “deactivation compensation” mode, keeping temperature and oxygen concentration constant. It was shown that sintering is the main source for deactivation and that kinetics of deactivation can be described by using a general power-law equation with an order of deactivation equal to 1 with respect to the driving force (a − a ss ), a being the time-dependent activity and a ss the final steady-state activity after an extended period of catalyst operation. The long-time prediction of catalyst behavior suggested that an improved catalyst should work at a lower temperature to avoid sintering.  2004 Elsevier Inc. All rights reserved. Keywords: Deactivation; Kinetics; Ethylene epoxidation; Accelerated deactivation; Deactivation compensation 1. Introduction Deactivation of catalysts is one of the most investigated problems in industrial catalysis [1] due to its negative impact on both economy and environment. Catalyst improvement is often achieved through identification of sources for its de- activation, with the final goal of preventing deactivation and increasing long-term catalyst activity, selectivity, and stabil- ity. The development of Ag-based catalysts for ethylene epoxidation, which has been the subject of research for many years, is an example of how much effort and time needs to be put into catalyst improvement [2]. Over the years the selec- tivity of these catalysts improved from about 70% to more than 80% as reported in the patent literature [3]. Neverthe- less, stability is still an important issue since these catalysts deactivate under industrial operating conditions. Among several possible mechanisms for deactivation of Ag-based catalyst, sintering of the active metal is broadly accepted as, if not the single, but in any case, the most important one [4–6]. * Corresponding author. Fax: (49) 30-6392-4454. E-mail address: dropka@aca-berlin.de (N. Dropka). 1 On leave from University of Novi Sad, Faculty of Technology, 21000 Novi Sad, Serbia and Montenegro. For compensating the loss in catalyst activity under conditions in practice, reaction temperature is usually in- creased [7]. To prolong the period of high catalyst activity a kinetic model of deactivation predicting the optimal oper- ation mode would be helpful [8]. For this purpose suitable experimental kinetic data of catalyst deactivation must be available. To save time and manpower for long-term cata- lyst tests, accelerated deactivation on laboratory or bench scale is desirable [1,9,10]; they allow study of the process by broadening the range of reaction conditions [11–13], be- ing an option that is rather restricted in commercial and pilot plant reactors. Furthermore, such accelerated deactivation studies provide deactivated catalyst samples for further post- mortem catalyst analysis [1], in a much shorter time span than under regular operation. Benefits from such compre- hensive investigation are direct, by means of cost reduction, as well as indirect, through minimization of causes for cata- lyst deactivation once they have been learned [14]. Although reaction kinetics of Ag-based catalysts for eth- ylene oxidation was extensively investigated in the past [15– 25], rarely any attention has been paid to accelerated deacti- vation [26,27], and no data were reported related to kinetic modeling of deactivation. It is the goal of the present study to derive deactivation kinetics of a commercial Ag/Al 2 O 3 cata- lyst based on accelerated deactivation tests at the laboratory scale [28], and to validate the obtained model by applying 0021-9517/$ – see front matter  2004 Elsevier Inc. All rights reserved. doi:10.1016/j.jcat.2004.06.003