Photoelectron Microscopy

In the photoelectron process, electrons are emitted from a sample exposed to light. For solids, the energy of the light must exceed the energy necessary to eject an electron into the vacuum. This minimum energy is called the work function. For most surfaces, this requires ultra violet light. Photoelectron emission is sensitive to a single adsorbed monolayer on a surface. Because of this it is possible to observe single monolayer films by photoelectron microscopy.

Images of surfaces are obtained by simply focusing photoelectrons onto a viewing screen. This is accomplished by using an electron lens similar to those used in conventional electron microscopy. Figure 1 is a schematic diagram showing the basic features of a simple photoelectron microscope with an electromagnetic lens. In practice, however, more complicated arrangements with multiple lenses are used (6, 7). The contrast in a photoelectron image depends upon the relative photoelectron yield. Areas with different work functions release different numbers of photoelectrons, which, in turn, produce a different intensity on a viewing screen.

Although atomic resolution has not been achieved in photoelectron microscopy, the great advantage of this technique is its ability to observe dynamic events in real time, such as reaction fronts (5) and chemical kinetic oscillations (8–10) in single monolayer adsorbed films. These can be recorded at ordinary video rates (0.04 s per frame) or photographed directly from the viewing screen. Still photographs, however, only poorly portray the dynamics occurring during catalytic reactions on surfaces. The studies reported here were performed on single crystal platinum surfaces under ultra high vacuum.

The Oxidation of Carbon Monoxide on Pt{100}

Under certain reaction conditions, fronts are seen during the catalytic oxidation of carbon monoxide on platinum. The conditions necessary are shown schematically in Figure 2. This reaction is self-poisoned by carbon monoxide. At lower temperatures and high partial pressure of carbon monoxide, the carbon monoxide effectively forms an impenetrable monolayer film on platinum which blocks the adsorption and dissociation of oxygen, and stops the reaction. This is represented in Figure 2(a). In order for the reaction to proceed, a hole must be formed in the carbon monoxide film to allow the adsorption of oxygen. One way to produce holes in the film is to increase the temperature to 500 K and thermally desorb some of the carbon monoxide Figure 2(b). On an ideal defect-free surface this would occur randomly throughout the film and no fronts would be formed. A second method to produce holes in the carbon monoxide film is to provide impurity sites on the platinum which do not adsorb carbon monoxide, but which adsorb oxygen.

Oxygen adsorbed on such sites can react with the side of the carbon monoxide film, which is vulnerable to attack, although the top of the film remains inert. This is shown schematically in Figure 2(c) and (d). Once initiated at an impurity site, the reaction front propagates out into the carbon monoxide film, removing it by forming carbon dioxide, which desorbs, and replacing it by adsorbed oxygen. An adsorbed monolayer of oxygen does not poison the reaction. By artificially adding such impurity sites on the platinum, the reaction can be made to proceed more efficiently at lower temperatures, or at higher partial pressures of carbon monoxide. In practice, such sites are often present in the form of impurity inclusions or dust particles.

A reaction front propagating through a monolayer of carbon monoxide adsorbed on Pt{100} at 350 K is shown as a photoelectron image in Figure 3. The front initiated at an impurity inclusion, seen in the centre of each frame. The monolayer film of oxygen, which forms behind the reaction front, appears dark relative to the carbon monoxide because it has a higher work function, 6.46 eV (11), compared to 6.11 eV of carbon monoxide (11). A deuterium lamp, which has a maximum photon energy of ~6.9 eV (180 nm), was used as the light source.

To initiate the front, oxygen at a pressure of l.0 × l0^–2 Pa was added to the vacuum chamber of the microscope. This is near to the maximum pressure at which observations can be made. The partial pressure of carbon monoxide was increased until the platinum surface was saturated with a monolayer of carbon monoxide. This required a partial pressure of carbon monoxide of only 2.4 × l0^–3 Pa. Impurity sites were also saturated with adsorbed carbon monoxide under these conditions, and the reaction was self-poisoned. When the partial pressure of carbon monoxide was lowered reaction fronts initiated at dust particles and inclusions.

The carbon monoxide, which is relatively weakly adsorbed on some impurities, is displaced by competitive adsorption of oxygen. Oxygen adsorbed on the impurity sites then reacts with the side of the neighbouring carbon monoxide film.

The reaction-diffusion fronts illustrated in Figure 3 show an anisotropy. This is caused by step bunches on the surface which hinder diffusion. Step bunches, which are groups of atomic steps, are often observed on platinum surfaces after sputtering (12), thermal etching or catalytic etching (13). They also arise from glide lamellae or slip bands (14) which form on platinum when it is thermally stressed. Figure 4 shows step bunches on a Pt{100} single crystal surface. To make them visible in the photoelectron image, they were decorated with carbon (15). This was removed before the catalytic reaction, by heating in oxygen. The directions of the step bunches varied in different areas of the crystal.

The reaction-diffusion fronts travel more slowly over the steps compared to travel along them. Surface steps can strongly bind reactants and form activation barriers for diffusion over them (16, 17). Diffusion over steps is impeded and therefore reaction-diffusion fronts travel with a lower velocity in directions over steps, relative to directions parallel to steps.

The Nitric Oxide + Hydrogen Reaction on Pt{100}

In excess hydrogen, platinum readily catalyses the formation of ammonia from NO (18). This is an unwanted side reaction in automobile air pollution control catalysts, on which nitric oxide in the presence of hydrogen is preferentially converted into nitrogen and water (18). In fact, according to Shelef the unwanted production of ammonia was one of the most difficult problems to overcome in the development of automotive catalytic converters, and rhodium was eventually substituted for some of the platinum (18).

Reaction-diffusion fronts are seen in the reaction between nitric oxide and hydrogen on Pt{100} under certain reaction conditions, see Figures 5–7. The dark areas in these are covered with a single monolayer of nitric oxide, work function 5.82 eV (19). The bright areas are believed to be ammonia or a precursor of ammonia (that is NH or NH₂) based upon its work function (5 eV). This work function rules out the bright areas being pure hydrogen. The production of ammonia was also detected in the gas phase by mass spectroscopy, in agreement with studies by Madden and Imbihl (20).

The fronts of Figure 5 were initiated as follows: The sample was kept at 415 K and the hydrogen pressure adjusted to 1.7 × l0^–5 Pa. The surface was then saturated with nitric oxide by increasing its partial pressure to l × l0^–2 Pa. Upon reducing the partial pressure of nitric oxide to l × l0^–4 Pa, the fronts initiated at dust particles after an induction time of a few seconds. The fronts propagated out into the nitric oxide monolayer at a speed of 9 μ m/s in directions along steps and about one third as fast across steps.

Anisotropy of the fronts is more pronounced at lower temperatures. Figure 6 shows a front initiated at a dust particle at 410 K and then cooled to 375 K. The long streamers follow step bunches. Anisotropy increased at lower temperatures because diffusion over the steps becomes more difficult.

The collision of two fronts is shown in Figure 7; each initiated at a dust particle at 410 K. The sample was cooled to 380 K to reduce the speed of the fronts and to allow still photography. The impurity particles apparently initiate the reaction by providing empty sites which are necessary for nitric oxide to decompose.

Oxidation of Graphite on Pt{100}

Graphite and other carbonaceous materials are troublesome contaminants which block catalytic reactions on platinum. Often carbon must be removed periodically from platinum surfaces by combustion with oxygen. The reaction fronts formed during the oxidation of graphite on Pt{100} are shown in Figure 8. The light areas are covered with carbon, while the dark areas are covered with oxygen and show where the graphite was removed. The surface of graphite, which typically grows with its close-packed basal plane parallel to the platinum surface, can be quite inert towards oxygen and oxygen has difficulty in penetrating the graphite film. The graphite film was not a single monolayer, and can grow to become quite thick. Oxygen penetrates the film at dust particles and reacts with the side of the film. Once a hole is formed in the film, the graphite is rapidly removed as the reaction fronts propagate away from the initiation sites. The holes in the graphite film are circular. Lack of anisotropy occurs because graphite does not diffuse. These studies suggest that graphite may be more efficiently removed from platinum surfaces if particles, which readily adsorb oxygen, are intentionally added to the surface.

Conclusions

Many catalytic reactions on platinum are limited by the availability of unoccupied surface sites. A key to the production of more efficient catalysts, for reactions which exhibit self-poisoning, appears to lie in the introduction of empty surface sites through the addition of impurities which only weakly adsorb the reactants that cause self-poisoning. Inert films which normally cause self-poisoning are removed by reaction fronts which initiate at holes in the film at impurity sites.