In heterogeneous catalysis research, extended single-crystal surfaces of catalytic metals have been largely used to study fundamental aspects of, e.g., gas adsorption and desorption. To bridge the "material gap" between such model catalysts and industrial catalysts (generally consisting of small catalytic metal particles supported on some "large area support" ), a new class of model catalysts have developed: supported metal particles grown on a flat support. In this thesis such a model catalyst system - Pd supported on single-crystal MgO - is studied, from fabrication to reactions, with the aim to investigate the relation between catalyst structure and catalyst performance.

For growth studies, Pd films with nominal thicknesses, t_f, varied between 0.5 and 200 nm have been grown on polished MgO(100) wafers, kept at substrate temperatures,T_s, of 100, 300 and 600°C. The depositions have been made in high vacuum, primarily by electron-beam evaporation, but for comparison, sputtering has also been used. Microstructure, strain and orientation of the Pd films have been studied using acombination of x-ray diffraction (XRD), transmission electron microscopy (TEM) and atomic force microscopy (AFM). The method of grazing incidence XRD (GIXRD) using a standard laboratory x-ray source has been explored and shown to give usefulin formation even about the thinnest films.

The catalytic performance has been studied in an ultra-high vacuum chamber using desorption mass spectrometry. Both carbon monoxide and deuterium have been used in oxidation measurements of oxygen precovered samples, since CO and D₂ give complementary information about the samples. The titration experiments have been performed at film temperatures, T_f, of 100, 200 and 300°C, after oxygen exposures varied from 0.2 to 150 L. Three films consisting of separate Pd particles of varying size supported on MgO(100), and three continuous Pd films, grown on MgO(100), MgO(110) and MgO(111), respectively, have been studied.

The Pd films are shown to grow with a strong epitaxial preference on MgO(100),with Pd<100>//MgO<100>. In a few cases, also four 111 -oriented Pd domains, rotated 90° with respect to each other, are detected using XRD. The growth morphology shows an initial three-dimensional particle growth, for all T_s values. Both TEM and AFM observations show that for T_s ≥ 300°C, the Pd particles are faceted. For Pd grown on MgO(100), the main facets are (111) surfaces on the particle sides and a (100) top surface, sometimes with a small portion of (100) and (110) surfaces existing on the particle sides. When the particles through consecutive coalescence reach a certain size(~0.3 μm at 600°C), the required mass transport to reassume faceting is too large to be achieved during the short time the samples are kept at the high deposition temperature. For T_s = 300 and 600°C, no large differences are seen between evaporated and sputtered samples with t_f ≥ 2.5 nm. For the films with t_f = 0.5 nm, the Pd particles are larger and less strained in the case of sputtering, suggesting that the energetic species present during sputtering increases the apparent surface diffusivity of the Pd adatoms during growth.

Titration measurements on the continuous Pd films show that for such transient conditions, the time-dependent CO₂ desorption rate is much more sensitive to structure and temperature than the D₂O desorption rate. The initial CO reaction probability,however, is always ≥ 0.5 for T_f ≥ 200°C , whereas the initial D₂ reaction probability varies with oxygen preexposure and between the different films. For most measurements on the continuous films, the CO and D₂ reaction probabilities are single-valued functions of oxygen coverage for all studied T_f. It is shown that large oxygen exposures of the 200 nm thick Pd(100) film create an oxygen-covered surface, which efficiently blocks hydrogen dissociation, but on which the CO reaction probability is close to unity.

For the particle films, titration measurements show that both the MgO support and the different Pd facets have to be taken into account to explain the behaviour of the desorption rates. 0₂ and D₂ dissociate on the Pd particles and spill over to the MgO, whereas CO adsorbs and reacts both on the Pd and on the oxide. The D₂O desorption rate is not a single-valued function of oxygen coverage and shows larger differences between the different samples than the CO₂ desorption rate. This is explained by considering the blocking of gas dissociation/adsorption caused by oxygen adsorbed on Pd surfaces. Comparing the Pd(100) and Pd(111) facets, oxygen shows a larger difference in blocking efficiency for hydrogen adsorption than for carbon monoxide adsorption. The larger the oxygen preexposure, the larger the probability that the D₂ molecules dissociate primarilyon the Pd(111) surfaces.