Palladium model catalysts

Abstract

This thesis is concerned with the oxidation and reduction of palladium and the adsorption and oxidation of carbon monoxide on palladium. Both subjects are closely related to each other by the background studies needed for the research on these two topics. Actually, the background studies constitute the main part of this thesis.

We used palladium in the form of a (111) single crystal, a polycrystalline foil and in the form of particles ranging in size from 1.5 to 13 nm supported on SiO2/Si(100). Some experiments have been performed on a silica supported palladium catalyst. The SiO2/Si(100) supported particles, the model catalysts, are the most interesting samples as they are chemically identical to silica supported palladium catalysts. Structure sensitivity and particle size effects are difficult to study on `real' catalysts, but our model catalysts give the opportunity to investigate these effects in detail. The characteristics and preparation of the model catalysts are described in the chapters 1 and 2.

We mainly used AES and XPS to determine the particle size of the samples. With XPS it is even possible to determine the thickness of an oxide skin on the particles. The particle size and the thickness of the oxide skin is calculated from the intensity ratios of the substrate peaks and the Pd0 and Pd2+ peaks. This is certainly not straightforward. Chapter 3 deals with the mathematics we have developed to make this possible.

Palladium oxidises not easily; it is a truly noble metal. Temperatures above 570 K (300 C) and atmospheric pressure is needed to form bulk oxide. At the low pressure conditions used in the chapters 5 and 8, PdO formation is limited to the surface layer only. Thermodynamics, however, are not the reason for this `noble' character: PdO is the stable phase even at very low oxygen pressures and temperatures up to 1100 K. The noble character is related to the crystal structures of palladium and palladium oxide. The crystal structure of the metal and the oxide are very incompatible, making the growth of an oxide layer on the metal difficult.

At low pressures oxygen adsorption is limited to the very top layers and only at elevated temperatures nucleation of a surface oxide occurs (chapter 5). To form the surface oxide oxygen atoms have to be present in the surface layer of the single crystal or the particles. On the close packed surface of Pd(111) migration of oxygen atoms into the surface layer is difficult compared to the more open surface of the particles. As a result, surface oxide is more easily formed on the particles than on Pd(111). However, we observed no particle size effect in the interaction of oxygen with particles ranging in size form 1.5 to 9.5 nm (chapter 8).

At atmospheric pressures bulk oxide is formed in a layer by layer fashion. The reconstruction needed for the formation of each oxide layer inside the particle results in an activation energy of at least 100 kJ/mol. PdO is formed stoichiometrically and no large amount of excess oxygen is present, i.e. oxygen dissolved in the metal or oxide (chapter 4 and 6). The atmospheric oxidation experiments were performed in the VG-XPS on two samples only; one with 5 nm particles and one with 8 nm particles. There were no significant differences between the oxidation of the 5 and 8 nm particles (chapter 4).

The thermal reduction of palladium oxide layers and particles was investigated with XPS (chapter 4). Palladium oxide decomposes in vacuum when the temperature exceeds 450 K, but we performed the experiments at a temperature of 770 K (500 C) to obtain a convenient decomposition rate. The rate of the reduction is linearly proportional to the surface area; the rate limiting step is the formation of molecular oxygen on the surface. During the reduction a metallic core grows inside the particle. The last monolayer of oxide, however, has a much lower decomposition rate than bulk or core oxide. The smallest particles used for the reduction experiments had a diameter of 3.5 nm. In such particles about 40 percent of the atoms is located at the surface. As a result, PdO particles of this size behave very much like surface oxide. Larger PdO particles resemble bulk PdO and lose their oxide easily up to the last monolayer.

With ellipsometry the oxidation or reduction rate can be monitored continuously and reducing gases like hydrogen or carbon monoxide can be used easily. Unfortunately, it was not possible to oxidise the samples in the Ellipsometry system, making this system useless for these experiments.

One limitation of the Ellipsometry system (and many other systems with only AES and no XPS) was solved during the research of this thesis: the ability to analyse the presence of PdO with AES quantitatively. In chapter 6 we show that the palladium M45N1N23 transition, which is almost absent in palladium metal, becomes quite intense in palladium oxide. The intensity of the transition is linearly proportional to the amount of palladium oxide. This remarkable increase of the intensity of the palladium M45N1N23 transition during oxidation has never been reported before.

The adsorption and oxidation of carbon monoxide was studied with ellipsometry. This optical technique makes it possible to monitor processes on the surface under reaction conditions without the possibility of beam damage or interaction. We used ellipsometry to measure the carbon monoxide and oxygen coverage simultaneously during the exposure of the gases. The production of CO2 was measured with a simple mass spectrometer. The experiments were performed on a Pd(111) single crystal and on palladium model catalysts with particles ranging size from 1.5 to 9.5 nm.

The carbon monoxide adsorption experiments are described in chapter 7. The saturation coverage of CO was 0.5 ML on all samples, including the single crystal, with respect to the number of surface atoms. The initial isosteric heat of adsorption was 148 +/- 5 kJ/mol on all samples. The heat of adsorption decreased with an increasing CO coverage. The decrease of the heat of adsorption with an increasing CO coverage occured in a similar way on all samples. Clearly the adsorption of CO is structure insensitive.

Also with the oxidation of carbon monoxide, described in chapter 8, particle size effects are absent. However, there is a difference between the reaction on supported particles and on Pd(111). At the applied conditions p(O2), p(CO)<=10^(-3) Pa, T<=540 K) oxygen adsorbs mainly on the surface of Pd(111), with a saturation coverage of 0.25 ML. Diffusion of oxygen into the surface layer and the subsequent nucleation of surface oxide occurs only very slowly on Pd(111) at these conditions (chapter 5). So on Pd(111) we observe the typical Langmuir-Hinshelwood reaction between oxygen and carbon monoxide. Since CO inhibits the adsorption of oxygen, the reaction rate is governed by the CO partial pressure. On the particles, however, the oxygen diffuses directly into the surface layer(s) up to a coverage of about 1 ML and surface oxide formation occurs already at temperatures of 400 K. At high CO coverages oxygen is still able to adsorb, probably due to the open surface compared to the single crystal. The oxygen present in the surface layers is quite unreactive and remains present even when p(CO) >> p(O2). The carbon monoxide seems to react only with the surface oxide causing short oscillations when the CO pressure is changed. We never observed sustained oscillations. Of course, the question arises whether this behaviour is typical for particles or that the behaviour of the particles is just an addition of low index plane properties. Oscillations in the reaction of carbon monoxide with oxygen are well known for Pd(110).

Since there are still a lot of `basic' problems unsolved of reactions on palladium single crystals, ellipsometric investigations on palladium single crystals should be continued, in addition to the research on model catalysts. Furthermore, atmospheric reactions should be performed as well. Ellipsometry gives the opportunity to close the infamous pressure gap between surface science and catalysis.