Catalysis on small particles

Supported metal particles are widely used as catalysts to activate, enhance or modify gaseous reactions, and there is a profound interest in understanding correlations of activity and selectivity with characteristics like size, morphology, shape, composition, segregation in the case of alloys, as well as the role of the substrate. Surface physics and electron microscopy provide powerful tools for such investigations.

Our in-situ TEM allows us to perform reaction experiments on small particles just after production by vapour deposition, without breaking the vacuum. Thus, contamination is avoided, which would otherwise be caused by transfer procedures. As model systems, we are studying interactions of reactive gases, e.g. O₂, H₂, CO, with Pd and Pd alloys with Au, Pt, Ag, Ni and others on substrates like a-C, graphite, or SiO₂.

Catalytic oxidation of a-C
As an example, exposure of Pd particles on carbon substrates (a-C or HOPG) to a partial pressure of oxygen of 5x10^-6 mbar at temperatures above 400 °C causes oxidation of the support. This is explained by dissociative adsorption of oxygen molecules on the metal, and reaction of the oxygen monomers with the carbon at the interface. Gasification of the substrate can be followed in the TEM, as the particles become mobile upon "digging" holes or grooves.

Fig. 1: Catalytic oxidation of a-C by Pd particles under oxygen at 5x10^-6 mbar at 560 °C. The particles had also been deposited under oxygen at a somewhat lower temperature, which causes strong faceting, and the production of rectangular holes in the substrate.

In situ analysis of the kinetics for individual particles revealed that the oxidation rate is proportional to the oxygen partial pressure, as well as to the contact area. From the temperature dependence, an activation energy of 65 kJ/mole was determined, with tendency to further decrease at above 500 °C.

Alloying of catalyst particles is of particular interest as to affect the activity and selectivity, or to reduce contamination ("poisoning"), for example. However, the structure and stability of alloy particles is a major issue. As an example, addition of Ag to Pd particles results in a drastic decrease of the activity of oxidation of the a-C support film, which is much stronger than is expected from the bulk composition. Silver itself is inactive in this respect. Thus, it is reasonable that this is caused by segregation of Ag to the particle surface, driven by the lower surface energy of Ag than of Pd. A similar effect has been found for Pd-Au particles.

The case of Pd-Ni particles is more complicated, in that at low Ni contents, the activity of oxidation of the substrate is again reduced, but in addition, the Ni is oxidized, leading to decomposition of the alloy into NiO precipitates, and pure Pd. At greater Ni contents, dissolution of carbon renders the particles inactive.

Catalytic conversion of a-C to graphite

Pure Ni particles are not oxidised under these conditions. Rather, Ni reacts with the a-C substrate at temperatures above 600 °C and converts it into graphite. This occurs under vacuum as well as at elevated partial pressures of 5x10^-6mbar of oxygen, carbon monoxide or hydrogen, but only the latter seems to somewhat enhance the reaction. A typical reaction can be followed in the video sequence in fig. 2.

Fig. 2: Catalytic graphitisation of a-C by Ni particles at 670 °C under 5x10^-6 mbar of hydrogen. Note the coalescence of the spreading Ni with some small, originally inactive particles.
Field of view is 520 x 420 nm².

The reaction starts with nucleation of a graphene layer on the surface of the Ni particle. Subsequent graphene layers grow at the Ni/graphite interface. The thus thickening graphite shell is rather stable, and expels the Ni core, which starts to spread on the a-C substrate. During spreading, the Ni continues to graphitise the substrate at the interface. It seems that the driving force is the gain of energy upon formation of ordered graphene layers, and the adhesion of the metal to the reaction front. Propagation in a stop-and-go fashion leads to a rugged morphology of the polycrystalline graphite, as is illustrated in the perspective view in fig. 3.

Fig. 3: Perspective view onto a thin film of a-C, which has been partly graphitised by Ni. Walls and hillocks of graphite layers are formed at the edges of the propagating Ni patches.
Some inactive Ni particles are also seen.

It is clear that these processes involve rapid diffusion of both carbon atoms to, and Ni atoms away from the interface. So far, no evidence has been found for diffusion of C atoms through bulk Ni.
The reason, why some Ni particles are initially inactive is not quite clear, but seems to be caused by the formation of a graphitic/carbidic diffusion barrierat the metal/substrate interface. Interestingly, this may already occur during vapour deposition of the Ni particles at temperatures above 400 oC, and primarily affects smaller particles. Eventually, a propagating Ni front attacks such a barrier, and the originally inactive particle is activated. An example is shown in the time-lapse animation in fig. 4.

Fig. 4: An originally inactive, faceted Ni crystallite is activated by contact with a propagating Ni front, allowing carbon atoms to diffuse on the Ni surface and to feed the nucleation of graphite layers. Subsequently, part of the Ni flows out and is incorporated in the spreading Ni.
The rest of the Ni core is cut off, probably because diffusion was not fast enough.
Field of view is 500 x 370 nm².

Sometimes, the whole Ni core leaves its encapsulation before this is completely filled with graphite. This is probably be caused by the cohesive force of the metal, while propagating on the a-C substrate. Such a case is illustrated in fig. 5.

Fig. 5: High resolution TEM image of a graphite shell abandoned by a Ni core. The arrow indicates the outflow channel of the Ni. Graphite lattice fringes are also visible on the plane substrate. Image courtesy A. Kornowski, Institut für Physikalische Chemie, Universität Hamburg.

Last modified: 07/06/2005