Near Edge X-Ray Absorption Fine Structure

Near-edge X-ray absorption fine structure (NEXAFS) spectroscopy probes electronic tran-sitions from a core level (e.g. from K-shell) to an unoccupied orbital or continuum state.

Detailed information can be found in literature [124–128]. Typically, strong and distinc-tive transitions to unoccupied molecular orbitals (MO) appear in the energy region from just below to about 50 eV above the core level ionization edge (e.g. K-edge).

During the measurement, the sample is irradiated with monochromatic X-rays. The energy of the X-rays is varied in the desired range. The dominant process is absorption, which results in a core hole and an excited electron. The electron can either be excited to the vacuum (emission as photoelectron), or to an unoccupied MO. The hole is subsequently filled with an electron, either radiatively by emission of a fluorescent photon or non-radiatively by emission of an Auger electron. The emission of a photon or electron is a direct measure for the existence of a core hole and thus for the preceding X-ray absorption.

A corresponding energy diagramm is illustrated in Figure 2.11.

Information on electronic Structure

In NEXAFS studies, the dependence of the photoabsorption cross section on the energy of the incident X-ray beam is investigated. The spectra are dominated by a step function, which results from the excitation of a core electron to the vacuum. Near the step, resonant transitions to unoccupied MO are superimposed. Such excitations occur if the energy of the X-ray matches the energy difference between the initial state and an unoccupied MO.

2.2 Experimental Methods

Figure 2.12: A schematic NEXAFS spectrum with ionization potential and resonant tran-sitions toπ^∗ and σ^∗ orbital (adapted from [124]).

Figure 2.13: Illustration of the polarization dependence of transition resonances toσ^∗ and π^∗ MO. The maximum amplitude of aσ^∗ orbital is along the bond axis and the maximum amplitude of the π^∗ orbital is along the normal of the bond direction. The electric field vector has to match the polarization of the MO in order to excite the transition (adapted from [124]).

In unsaturated hydrocarbons, excitations to the first unoccupied π^∗ MO appear below the ionization step while excitations to σ^∗ MO have higher energy than the step edge. A schematic illustration of a NEXAFS spectrum is shown in Figure 2.12.

Information on molecular Orientation

In addition to the information on the electronic states of molecules, NEXAFS can also provide information on their orientation. Molecular bonds and their MO are highly direc-tional. Therefore, the transition intensity of a K-shell spectrum depends not only on the energy, but also on the orientation of the electric field vector relative to the orientation of the MO.σ orbitals have a maximum amplitude along the bond axis whileπ orbitals have a maximum amplitude perpendicular to the bond axis. The polarization dependence of NEXAFS transitions is illustrated in Figure 2.13. In contrast, the excitation of electrons to the vacuum is independent from the polarization. Therefore, the intensity of the step

Measuring the dependence of the transition intensity on the polarization of the X-ray beam can be achieved in two ways. Either the geometry between the sample and the X-ray beam is varied with fixed beam polarization; or the beam is switched between s and p polarization while the sample–beam geometry is fixed at small incident angle. The latter method has two major advantages. On the one hand, exactly the same region of the surface is probed in both measurements, in contrast to angular-dependent investigations, where a large surface area is probed at small incident angles and a small are at large incident angles. On the other hand, the experiment are much less time-consuming. Typically, switching the polarization of the beam takes much less time than aligning the sample before each measurement.

3 Pd/Fe ₃ O ₄ Model Catalysts

Since several decades, surface science methods are successfully applied to study chemical reactions over single crystal surfaces, including stepped and defect-rich surfaces [11, 129, 130]. Studies on these systems give valuable information on the reactivity of different surface facets or specific sites. In this work, a Pd(111) single crystal was used to study the adsorption and intrinsic reactivity of hydrocarbons on (111)-facets. However, single crystals only poorly resemble industrial catalysts, which are highly complex materials with respect to their composition and structure. Generally, single crystals cannot reproduce the characteristic structural and electronic properties of supported catalysts, which are responsible for the kinetic effects that we discussed in chapter 2.1.2. To overcome this limitation, model supported catalysts have been developed two decades ago by several groups [10, 12–19, 131]. In model catalysts, distinct structural complexity is introduced in a well-defined manner to mimic properties of industrial catalysts. In contrast to indus-trial catalysts, model catalysts are fully accessible by surface science methods and their structure can be characterized in great detail.

A model catalyst is based on a well-defined thin metal oxide film that is grown onto a single crystal substrate. The oxide film acts as support for particles of the catalytically active metal. The underlying metal single crystal as a substrate ensures a macroscopically planar structure and good electric and thermal conductivity, which is required for many surface science techniques. There are two principle ways of preparing thin oxide films on top of a conducting substrate: either by oxidation of the single crystal substrate [132–135]

or by deposition of the metal onto the substrate and subsequent oxidation [136, 137]. In this study, we used the latter method to grow an iron oxide film on a Pt(111) substrate.

On the Fe₃O₄ film, Pd particles are deposited as the active phase of the catalyst. The preparation of the Pd/Fe₃O₄/Pt(111) model catalysts is schematically illustrated in Figure 3.1. It will be discussed in more detail in the following part of this chapter.

Figure 3.1: Preparation of the Pd/Fe₃O₄/Pt(111) model catalysts

Figure 3.2: (a) LEED diffraction pattern and (b) STM image of an Fe₃O₄ film; (c) model of the crystal structure of Fe₃O₄ (from [142])

3.1 Fe

O

film on Pt(111)

The preparation and growth of iron oxide on a Pt(111) single crystal substrate have been described in detail in literature [138–143]. First, a monolayer of FeO is grown on a clean Pt(111) single crystal by physical vapor deposition of iron from an electron beam evap-orator under UHV at 125 K and subsequent oxidation in 1·10⁻⁶ mbar O₂ at 990 K. To avoid sputtering damages to the surface, the same potential was applied to the sample as to the iron evaporant in the evaporator. Next, Fe₃O₄ is prepared on top of the FeO by cycles of 4-6 ML Fe deposition under UHV at room temperature, followed by oxidation in 1·10⁻⁶mbar O₂at 900 K. After six cycles, an approximately 10 nm thick film is obtained.

The morphology of the first Fe layer strongly influences the structure of the subsequently formed Fe₃O₄(111) film. Fe deposition at low temperature is required to ensure a complete wetting of the Pt(111) surface and thus to form a 2D FeO(111) layer. Fe₃O₄ grows on top of the FeO film first as three-dimensional islands. With increasing coverage, the islands grow and coalesce to form a closed and flat Fe₃O₄ film. The sharp spots in the low-energy electron diffraction (LEED) pattern in Figure 3.2a show a well-ordered long-range struc-ture. The scanning tunneling microscopy (STM) image in Figure 3.2b illustrates extended, atomically flat terraces, separated in height by steps of 5 ˚A or a multiple of that.

The Fe₃O₄ film has been characterized in detail by many techniques, e.g. photoelectron spectroscopy (PES) [144, 145], Auger electron spectroscopy (AES) [144], STM [141, 142, 146, 147], X-ray photoelectron diffraction (XPD) [146, 148], LEED [140, 143, 144, 149], and thermal desorption spectroscopy (TDS) [142, 150]. It was shown that a thin Fe₃O₄ film has the same crystal structure as the bulk phase of magnetite. This is an inverse spinel structure where the O²⁻ ions are ordered in an fcc-grid. The tetraedric vacancies are filled with Fe³⁺ ions and the octaedric sites are occupied by Fe²⁺ and Fe³⁺ ions (see Figure 3.2c). Nevertheless, the surface termination is still under discussion. Based on dynamic LEED and STM studies combined with theoretical calculations, Ritter and Weiss suggested a termination by tetraedrically coordinated Fe³⁺ ions [149]. In contrast, Lemire et al. concluded from TDS and IRAS measurements of adsorbed CO, combined with HREELS (high resolution electron energy loss spectroscopy) studies a model with termination by octahedrally coordinated Fe²⁺ ions [142].