Sintering under reaction conditions

We have proposed a novel method, a periodic pulse reaction technique, to accelerate sintering under reaction conditions for the rapid estimation of catalyst life (refs. l,2j. Although the sintering is frequently accelerated by thermal treatment, the sintering behavior in the reaction atmosphere may be different from thermal sintering (ref. 3), and, furthermore, the sintering sometimes proceeds at much lower temperature in the reaction atmosphere. Thus, for the rapid estimation of catalyst life, h is necessary 10 accelerate the sintering under reaction conditions. 

In the periodic pulse technique, hydrocarbons and O2 are fed alternately and repeatedly over the catalyst bed to subject the catalyst to separated and forced oxidation-reduciion cycle. This redox cycle is expected to promote sintering under reaction conditions, leading to the rapid estimation of catalyst life. The method was applied to the oxidation of C2H4 on CuO/SiCb and Ag/Sft>2 catalysis, mid the deactivation in the periodic pulse run was found to be, at most ten times as List as that in the continuous flow run (refs. 1,2). 

SMSI is also thought to affect methanation catalysts (normally transition metal or noble metals supported on alumina), which are used in the producton of substitute natural gas (SNG). In general, heating in H2 causes sintering on alumina and silica supports and heating in O2 or steam can cause dispersion and particle coalescence at 200 °C (Rukenstein and Lee 1984,1987, Nakayama et al 1984). The data have been based on ex situ EM studies. Here EM methods, especially under dynamic reaction conditions, can provide a wealth of new insights into metal-support interactions under reaction conditions. 

In situ ETEM permits direct probing of particle sintering mechanisms and the effect of gas environments on supported metal-particle catalysts under reaction conditions. Here we present some examples of metals supported on non-wetting or irreducible ceramic supports, such as alumina and silica. The experiments are important in understanding metal-support interactions on irreducibe ceramics. 

The decrease in activity of heterogeneous Wacker catalysts in the oxidation of 1-butene is caused by two processes. The catalyst, based on PdS04 deposited on a vanadium oxide redox layer on a high surface area support material, is reduced under reaction conditions, which leads to an initial drop in activity. When the steady-state activity is reached a further deactivation is observed which is caused by sintering of the vanadium oxide layer. This sintering is very pronounced for 7-alumina-supported catalysts. In titania (anatase)-supported catalysts deactivation is less due to the fact that the vanadium oxide layer is stabilized by the titania support. After the initial decrease, the activity remains stable for more than 700 h. 

The results of the above characterization studies indicate that also in titania-supported catalysts the vanadium oxide layer slightly sinters. Since the vanadium oxide dispersion strongly effects the activity of the catalyst [16], it is likely that this sintering process is causing the deactivation observed in Fig. 3. The TPR and TPD results show that also some carbonaceous deposits are formed under reaction conditions, but these deposits are only present in low concentrations and, therefore, not likely to cause the deactivation of the catalyst. 

Nonetheless, characterization of catalysts in reactive atmospheres by XRD is a powerful method for obtaining the basic information needed to determine structure-activity correlations. Many phenomena of structural deactivation, either by sintering or recrystallization, are accessible by XRD under reaction conditions. The practice of approximating reacting atmospheres by simple-to-handle proxies (hydrogen for hydrocarbons or dry gases instead of steam-loaded feeds) has to be abandoned. 

In addition to palladium, the catalysts used commercially always contain alkali salts, preferably potassium acetate. Additional activators include gold, cadmium, platinum, rhodium, barium, while supports such as silica, alumina, aluminosilicates or carbon are used. The catalysts remain in operation for several years but undergo deactivation. The drop in activity is due to a gradual sintering of the palladium particles which causes the catalytically active area to decrease progressively. Under reaction conditions potassium acetate is slowly lost from the catalyst and must continuously be replaced. 

It is necessary to determine rj(e) under reaction conditions, and a life test should be included in any catalyst development effort. The data from this test will allow r] to be fitted as a function of time on stream, 6. Equations 10.35 and 10.36 can obviously be used to model deactivation processes other than site sintering, and ko can be regarded as an empirical constant with units of reciprocal time. 

There are two well-established models commonly used to explain sintering at the fimdamental level. They are based on the migration and coalescence of whole crystallites in one case [1-3] and of adatoms [4-6] in the other. The actual process of sintering under given conditions is very likely a combination of adatom and crystallite migration on the surface of the support, with gas-phase transport being possible in some cases. The relevance of each route depends on the particular metal-support pair, as well as on the atmosphere and temperature employed during pretreatment and reaction. 

Importantly, this study also revealed size-dependent particle sintering under opemndo conditions. Specifically, particles with sizes corresponding to the maximum of catalytic activity and smaller particles had been affected because of direction-specific migration on the Ti02(l 10) surface, which seemed particularly pronounced at the start of the catalytic test. It was concluded that a model based on reaction-heat-induced migration is too simple to explain observed results. 

As stated in section 12.4, it is usual to mount the sensing elements in a diffusion head of the type shown in Figure 12.4a. In such an assembly the characteristic plateau in the signal as a function of temperature (Figure 12.5a) is reached when the rate of reaction becomes controlled by the rate of diffusion of gas through the sinter. Under these conditions, the signal is directly dependent on gas concentration and independent of both element temperature and oxygen concentration. 

The need for Ertl s approach becomes evident once a sample of an industrial catalyst is put under closer scrutiny. Figure 5.20 shows a high-activity catalyst with a rather large specific surface area comprised of nanometer-sized active particles. Under reaction conditions, these are reduced into metallic iron, covered by a submonolayer of potassium (and oxygen), which acts as an electronic promoter. The configuration of active particles is stabilized against sintering by a framework... 

---