Model Discrimination by Unsteady-State Operation: Application to the Reduction of NO with CO on Iron Oxide z H. RANDALL, R. DOEPPER and A. RENKEN* Institute of Chemical Engineering, Swiss Federal Institute of Technology, CH-1015 Lausanne, Switzerland The steady-state catalytic reduction of zyxwvutsrq NO with CO on iron oxide was described with three different models, which all fit the experimental data equally well. By applying the transient method to this reaction, it has been shown that for reduced catalysts without adsorbed CO, the N20 reduction step is well described qualitatively by two of the models, whereas only one of them can describe the NO reduction step. In the presence of adsorbed CO, however, none of these models can explain the observed transient behaviour. Therefore, a modification of the best model was proposed and tested via computer simulation, which showed good agreement with transient experiments. On a decrit la reduction catalytique en regime permanent du NO avec du CO sur de l’oxyde de fer zyxw au moyen de trois modeles qui montrent tous un bon accord avec les donnees experimentales. En appliquant la methode transitoire a cette reaction, il a ete montre que pour des catalyseurs reduits sans CO adsorbe, I’etape de reduction du N20 est bien decrite qualitativement par deux de ces modeles, tandis qu’un seul d’entre eux est capable de decrire I’etape de reduction du NO. Cependant, en presence de CO adsorbe, aucun des modeles ne peut expliquer le comportement transitoire observe. On propose donc une modification du meilleur modele, qui a ete testee par simulation sur ordinateur et qui concorde bien avec les experiences transitoires. Keywords: transient, model discrimination, NO reduction, iron oxide. n recent years transient experiments have become a com- I mon tool for investigating the mechanism of heteroge- neous catalysed reactions, mainly because kinetics measured under steady-state conditions lead to overall models unable to predict the dynamic behaviour of those systems. The steady-state behaviour can often be described by several dis- tinct models, contrary to the dynamic behaviour, which is much more model-sensitive. Transient experiments are, therefore, also a useful tool for model discrimination (Renken, 1993). In this work, the catalytic reduction of NO with CO on iron oxide was investigated both under station- ary and transient conditions. It is commonly accepted that, on most transition metal oxides, the NO-CO reaction partIy follows a redox mechanism according to which the catalyst is oxidised by NO and reduced by CO. In addition, part of the reaction occurs via an associative mechanism, involving adsorbed molecules (Gasan-zade and Alkhazov, 1990; Yaskevich et al., 1990; Musil and Pour, 1983; Glazneva, 1978; Sazonova et al., 1977). There is still disagreement between the different authors concerning the nature and the reactivity of the surface intermediates. On the basis of tran- sient experiments, this work discriminates between different models based on literature (Echigoya, 1980, Glaneva et al., 1978), which all predict the steady-state behaviour of the reaction. Experimental CATALYST PREPARATION AND CHARACTERISATION The catalyst support, consisting of SiO, particles (200-250 pm, Grace GmbH, Worms, Germany) was impregnated in an aqueous solution of Fe(N03), 1 km01/m3 (>99% pure, Fluka, Buchs, Switzerland). It was then dried *Author to whom correspondence should be addressed. E-mail address: renken@iqc.dc.epfl.ch zyxwvutsr hood zyx 1 1 Figure 1 zyxwvut - Schematic diagram of the experimental set-up. Feed: zy Ar, NO, CO, N,O, N2, CO,, He, H,. under vacuum at room temperature, and finally calcined at 650°C. The amount of Fe was determined by atomic absorp- tion: 10.4 mass% Fe. Such a catalyst has a specific area of 302 m2/g and a mean pore diameter of -6 nm. EXPERIMENTAL APPARATUS The schematic diagram of the experimental set-up is shown in Figure 1. It consists of two separate feed sections converging to a 4-way valve, a fixed-bed reactor and an ana- lytical section. Both feeds consist of pure gases (>99.99%, except NO: >99.9%, Carbagas, Lausanne, Switzerland). The carrier gas, zyxwv Ar, was further purified using a molecular sieve and an oxygen trap. The gas flows were controlled with mass flow controllers (Brooks Instrument B.V., Veenendaal, The Netherlands and Bronkhorst High-Tech B.V., Ruurlo, The Netherlands). The reaction was carried out in a glass fixed-bed reactor (internal diameter: zy 5 mm, length: 300 mm) heated in a temperature controlled electric oven. The temperature of the catalyst was measured with two K-type thermocouples (Philips AG, Dietikon, Switzerland) at two opposite axial points in the bed, and their values never differed by more than 1 “C. The upstream 586 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING, VOLUME 74, OCTOBER, 1996