**5.** Metallic presence in support pores.

in terms of the process design, the separation of products from a catalyst and its regeneration, it has usually suffered from a lack of understanding of the so‐called "active site." This fact has plausibly impeded the rational development of these systems. On the contrary, homogeneous catalysis can be designed in a more rational way as its properties are easily tuned via ligand designs, providing a substantial comprehension of elementary steps. Nevertheless, these processes have often required undertaking many technical advances to be competitive to heterogeneous catalysts. From the industrial point of view, this led in the final practice to the preference of heterogeneous from homogeneous catalysis. However, the development of enhanced heterogeneous catalysts has been hindered typically owing to their content of numerous variable active sites and their low concentration. Homogeneous catalysts, on the contrary, have been well‐defined systems that could be easily characterized and studied. Comparative studies of homogeneous and heterogeneous catalyses, providing a successful implementation of appropriate homogeneous models with molecular modeling, can yield a new insight into the complex of processes accompanying heterogeneous catalysis. In order to perceive the involved complex molecular events, it is also essential to construct a clear‐cut active site, test its catalytic performance, and assess the relationship between its structure and activity. Finally, the acquired picture can be utilized to design a new generation of catalysts applicable

248 New Advances in Hydrogenation Processes - Fundamentals and Applications

Typically, the determining characteristic of heterogeneous catalysis lays in the structural composition of its active sites and molecular structures, in the case of a metal catalyst and the reactant, respectively [1, 2]. In other words, the chemical and physical properties of the surface of the active sites and the molecule entering the reaction generate a generous number of parameters decisive for the catalytic mechanism. In order to predict the catalytic behavior, geometric and electronic structural properties have been traditionally studied [2], particularly based on "Theory of Electronic Effects" [2, 3]. The core of this study consists in the active site (catalyst side)‐reaction center (substrate side) whose behavior is determined by their interde‐ pendent electronic properties, with each component either donating or accepting electrons. Orbitals of both of the reaction players are thus directed by relatively strongly dependent interactions of their affinity and repelling forces. Last but not the least, geometric degree of freedom may have decisive impact on the surface complex (active site‐reactant interaction) [2, 4]. However, the above list of parameters is not complete as other potentially significant aspects could contribute to the problem: thermodynamics, reaction conditions, and hydrogen‐catalytic

Electronic effects and spatial geometry are reckoned to play the most prominent role in heterogeneously catalyzed processes. Catalytic properties, related particularly to the metal

**4.** Substrate effects (strong metal‐support interaction (SMSI), hydrogen spillover, and

**2.** Metallic surface organization as well as other components (as in surface alloys).

behavior, are listed below by their descending importance [2, 6–11]:

**1.** Form and constitution of metal particles.

"as in homogeneous catalysis."

surface interaction [5].

**3.** Surface complexes.

surface area).

The record of metals having been successfully applied to heterogeneously catalyzed reactions would be considerably extensive. Additionally, no general treatment modifying their proper‐ ties (electronics, morphology, etc.) to a desired performance in individual reactions is available. As a result, the selection criteria obviously depend on rather empirical decisions.

Transition metals represent an extraordinary group of elements finding a widespread utiliza‐ tion in the applied heterogeneous catalysis. From the point of view of the band theory [12], all of them contain positively charged ions having shells with partially occupied *d*‐bands, in which electrons freely fluctuate. Predictably, these form bonds with adjacent atoms using *dsp* orbitals with inerratic *d*‐function. As a result, the structure of the metal surface is nonuniform, which is further pronounced [12] by depositing the metal particles on an internal surface of a suitable high surface area support (active carbon, zeolites, etc.). Catalytic particles are rarely uniform as evidently determined by their X‐ray diffraction diagrams. Elemental metal crystallites are composed of crystallographic planes diverse in their surface coordination and architecture. There is a direct implication [6, 13] of this phenomenon in a variable degree of saturation of coordination sites on the metal surface. As a consequence, a different catalytic behavior is macroscopically observed [6, 14, 15].

Current methods of molecular modeling offer a wide array of applications even to heteroge‐ neous catalysis. However, it is still necessary to significantly simplify the studied systems in comparison with the reality. Although modeling requires many compromises, its use allows obtaining very interesting results, which provide information on the mechanism of catalytic reactions on the surface. Owing to the development of computer technology, computer modeling and simulation have been increasingly recognized as important tools for the study and development of catalytic systems. These tools have the potential to penetrate into the reaction mechanisms, can predict properties of catalysts not yet synthesized, and provide information obtained by various computational techniques, which together with experimental results procure a comprehensive picture of the system. Nevertheless, it is always essential to maintain a proficient collaboration of computational chemists with researchers. In addition, in order to validate the new modeling techniques, theoretical results must be sufficiently confirmed experimentally. Therefore, it is advisable to have a comparison of various experi‐ ments, such as kinetic studies of reaction rates, thermodynamic information on adsorption, and spectroscopic data on the level of molecular structure. The double‐feedback technique (dual‐feedback mode) seems to be the "most prolific" strategy for the utilization of modeling in catalytic research where experiments are employed to confirm the results of modeling and the modeling is used to explicate the experimental results, to design new experiments, and perhaps in the future also to replace experiments with theoretical screening of different catalysts and reaction conditions.

One option to simplify the heterogeneously catalyzed reactions for the application of molec‐ ular modeling is finding a suitable model, which first needs to be designed, then mathemati‐ cally and experimentally tested and if its quality is proved, the method is validated and the limits of its applicability are determined. In the case of heterogeneous platinum catalysts designated for hydrogenation reactions, well‐structurally defined compounds of (ole‐ fin)Pt(PPh3)2 seems to offer a suitable model of adsorbed surface complexes as the key struc‐ tures in the transformation of a reactant to products. These compounds are platinum organometallic compounds containing coordinated olefins, which in a sense resemble the morphology of the metal surface and chemisorptions occurring on it. Although these cova‐ lent compounds cannot catalyze homogeneous hydrogenation, such as the rhodium analog Rh(PPh3)3Cl, their structure very well approximates the adsorbed state on heterogeneous platinum catalysts. The model of organometallic compounds (olefin)Pt(PPh3)2 represents an attractive alternative for the application to molecular modeling in heterogeneous catalysis. By comparing values with a similar significance (adsorption *vs*. dissociation energy) or the differences in geometric data, very good qualitative (corresponding order of stabilities) and even quantitative (correlating *E*ads and *E*diss) correlations can be achieved.
