Structure-sensitive poisoning

The combined use of the modem tools of surface science should allow one to understand many fundamental questions in catalysis, at least for metals. These tools afford the experimentalist with an abundance of information on surface structure, surface composition, surface electronic structure, reaction mechanism, and reaction rate parameters for elementary steps. In combination they yield direct information on the effects of surface structure and composition on heterogeneous reactivity or, more accurately, surface reactivity. Consequently, the origin of well-known effects in catalysis such as structure sensitivity, selective poisoning, ligand and ensemble effects in alloy catalysis, catalytic promotion, chemical specificity, volcano effects, to name just a few, should be subject to study via surface science. In addition, mechanistic and kinetic studies can yield information helpful in unraveling results obtained in flow reactors under greatly different operating conditions. 

The rationalization of the structure sensitivity of a certain reaction might also be a problem. First, one has to answer the following question Is the structure sensitivity observed inherent for the reaction studied and therefore related to the varying exposure of different sites to the gas phase, or is the sensitivity induced by side reactions which themselves (and not the reaction followed) are structure sensitive The latter problem was first formulated by Katzer (227), who also demonstrated that an apparent structure sensitivity may be a consequence of a structure-sensitive self-poisoning (228). 

The same Pt/SiC catalysts were poisoned by CS2 the hydrogenation of (+)-apopinene was used as an indicator reaction (ref. 14). The amount of CS2 necessary to eliminate the hydrogenation activity permits calculation of the fraction of metal sites active in olefin hydrogenation. This fraction correlates with the number of step sites. Similarly, a good correlation is found between this fraction and the rate of methyloxirane transformation (Table 1). These results reveal that the structure-sensitivity is caused by the variation in the number of active sites, and the steps appear to be the active sites for the regioselective hydrogenation of methyloxirane. 

We have been able to identify two types of structural features of platinum surfaces that influence the catalytic surface reactions (a) atomic steps and kinks, i.e., sites of low metal coordination number, and (b) carbonaceous overlayers, ordered or disordered. The surface reaction may be sensitive to both or just one of these structural features or it may be totally insensitive to the surface structure, The dehydrogenation of cyclohexane to cyclohexene appears to be a structure-insensitive reaction. It takes place even on the Pt(l 11) crystal face, which has a very low density of steps, and proceeds even in the presence of a disordered overlayer. The dehydrogenation of cyclohexene to benzene is very structure sensitive. It requires the presence of atomic steps [i.e., does not occur on the Pt(l 11) crystal face] and an ordered overlayer (it is poisoned by disorder). Others have found the dehydrogenation of cyclohexane to benzene to be structure insensitive (42, 43) on dispersed-metal catalysts. On our catalyst, surfaces that contain steps, this is also true, but on the Pt(lll) catalyst surface, benzene formation is much slower. Dispersed particles of any size will always contain many steplike atoms of low coordination, and therefore the reaction will display structure insensitivity. Based on our findings, we may write a mechanism for these reactions by identifying the sequence of reaction steps ... 

Catalytic reactions at a metal surface involve a subtle and delicate balance of adsorption forces. Too weak an adsorption and the catalyst will have low activity, too strong and the surface becomes poisoned by adsorbed reactants or products. Consequently, quite small changes in the nature of a metal surface may result in significant variations in catalytic properties. Structure sensitivity is known to exist. There is good evidence that the selectivity and activity of a metal catalyst are affected by changes in structure and/or electronic properties. 

Is it known that the rate of hydrogenolysis reactions are extremely sensitive to effects of alloying, surface contamination, poisoning, etc. Consequently, in all cases where supported metals are used there must be concern as to whether apparent particle size effects are due to structure sensitivity or to some minor contamination effect. In the few cases where clean single crystal surfaces have been used there is evidence of a structure effect.338 However, the maximum change in activity between different crystal faces seems to be about a factor of 10. For Ni single crystals the (100) surface is more active than the (111) surface. A similar conclusion has been reached for oriented Ni powder samples.339... 

Nevertheless, two factors strongly influence the heat of sulfur chemisorption on metal surfaces relative coverage and crystallographic structure. Thus sulfur chemisorbs at high coordination sites and, as a result, a selective poisoning of structure-sensitive reactions, preferentially catalyzed by these sites, may occur. Such a simple geometrical model can be used to explain change in selectivities induced by sulfur adsorption. 

Acidic forms of zeolites are well suited as supports for metal functions which are employed for hydrogenation, since they can also withstand the presence of traces of sulfur compounds frequently found in feedstocks of petrochemical industry. It should be noted, however, that hydrogenation is a structure insensitive reaction so it will primarily depend upon the concentration of the accessible metal particles and the adsorption constant of the unsaturated hydrocarbon. This may offer an explanation as to why Pt catalysts, for example, are still active for hydrogenation, when their activity for dehydrocyclization or hydrogenolysis (i.e., for structure sensitive reactions) is completely lost (e.g., by poisoning). 

Primary structure sensitivity resulting from the effect of changing particle size on step and kink density appears therefore to be present here at short reaction times. Secondary structure sensitivity (including the effect of carbonaceous poisoning on the reaction rate) appears not to be present here. Thus Somorjai has reported that the dehydrogenation reaction of cyclohexane to cyclohexane is insensitive to both structural featureSt whereas the dehydrogenation of cyclohexene to benzene la very sensitive to the densities of atomic steps and kinks and the order of the carbonaceous overlayer on the platinum crystal surface. 

The electro-oxidation of H2/CO mixtures is a very complicated reaction with many variables, e.g., temperature, CO concentration, and pH. From a practical standpoint, the structure sensitivity is not so interesting, since at any realistic level of CO, e.g., >10 ppm CO, and temperature the pure-Pt surface is highly poisoned by adsorbed CO, and the electrode polarization is impractically large. It is, however, still of fundamental importance to know the structure sensitivity of the reaction on pure Pt in order to understand the properties of Pt-based alloy catalysts that are not so highly poisoned by the CO, i.e., so-called CO-tolerant catalysts. For our purposes here, we discuss only one characteristic measure of the structure sensitivity, shown in Figure 14. For a wider range of results, we refer the interested reader to [54] and references therein. Figure 14 shows the current for H2 oxidation at 50-mV... 

Catalytic experience tells us that frequently only a small fraction of the sites at the surface of the metal participate in the reaction. Recently, Krai measured the metal surface area of palladium catalysts supported on carbon, their specific activity for various hydrogenation reactions, and their poisoning by thiophene (44). By combining the classic technique of poisoning with the measurement of metal surface area. Krai was able to show that the Taylor ratio, i.e., the fraction of active sites in each particular reaction, changed from unity to about 10 . With a Taylor ratio of unity, the reaction would be called facile. Otherwise, we deal with a structure-sensitive or demanding reaction. 

Bond and Ponec,8 Ponec,38 and Katzer39 urge caution in this subject, mainly because they are aware of the number of variables at work in experimental demonstrations of structure-sensitivity which may not have been eliminated, including induced effects such as deactivation and self-poisoning. 

Deactivation may be structure-sensitive. Thus Lankhorst et al. (132) have shown that variations of TOF with d for hexane reforming on Pt were caused more by increases in carbon deposition with d than by decreases in the intrinsic rate of reaction with d. For model stepped-surface single crystals of platinum, Somoijai and Blakely (133) showed that carbon deposition was favored on terraces over edges and comers, so that the reforming reactions favored by the latter sites are poisoned less rapidly than those occurring mostly on the flat surfaces. Thus small particles are expected to be poisoned less rapidly than large ones, for this example. 

---