MATERIALS SCIENCE AND ENGINEERING

ATOMISTIC STRUCTURE OF PRECIOUS-METAL CATALYST NANOPARTICLES FOR PEM FUEL CELLS

A fuel cell is a device for generating electricity directly and very efficiently, from gaseous or liquid fuel, without combustion. Fuel cells are similar to batteries in that they also deliver electrical power from a chemical reaction. But unlike batteries, which store the chemical reactants within, fuel cells operate with reactants that are stored externally and must be supplied. There are several types of fuel cells. Presently, the two most promising technologies are proton-exchange membrane fuel cells (PEMFCs) and solid-oxide fuel cells (SOFCs). In PEM fuel cells, the required chemical reactions, in which fuel molecules need to lose or accept electrons, are facilitated by the presence of an appropriate "catalyst" material. Usually, the catalyst consists of finely dispersed platinum nanoparticles.

The fundamental reactions in PEMFCs are (i) the oxygen reduction reaction (ORR), occurring at the cathode, and (ii)  the hydrogen oxidation reaction (HOR), occurring at the anode. A major impediment to the wide-spread deployment of PEMFCs is sluggish kinetics of the ORR. Typically, the ORR is several orders of magnitude slower than the HOR. It is believed that the rate limiting step is the first electron transfer of the ORR (4 electrons are transferred in the overall mechanism). In order to accelerate this step of the reaction, catalysts must be used. Presently, these usually consist of precious metal, e.g. platinum. Obtaining a technically useful power output requires a very high catalyst surface area. This is achieved by dispersing small platinum clusters on a globular, high-surface-area carbon support. Figure 1 is low-resolution TEM (transmission electron microscopy) image electron micrograph of such a catalyst, denoted as "Pt/C." Arrow 1 points to a Pt cluster with an apparent diameter of 5×10^-9 nm. Arrow 2 points at the carbon support.

Recent research suggests that by alloying Pt with other transition metals, such as Co, has a benefical effect on the catalytic efficiency. It is believed that the alloying elements modify the electronic structure to change the electrochemical potential at which Pt oxidizes, thereby giving a larger potential window for the adsorption of oxygen. The state of the art catalyst for optimizing the ORR are clusters of Pt alloyed with Co, and supported on very-high-surface-area globular carbon (Pt-Co/C). However, not much is known to date about the atomistic structure of the alloy clusters is not well known. In order to understand the physical background of the performance improvement, one would like to know the spatial distribution of the Co atoms in the clusters. It seems likely that either Co or Pt are enriched at the catalytically active surface. Furthermore, it is suspected that the proximity of the surface to all atoms in a small cluster induces stresses and corresponding strains, which in turn may influence the spatial distribution of the alloying elements because they differ in atomic volume.

Therefore, we are developing experimental methods for directly measuring the strain state of alloy clusters. This requires a highly precise assessment of the atom coordinates in the cluster. A very powerful tool for obtaining this information is HRTEM (high-resolution transmission electron microscopy). Figure 2 is a high-resolution electron image of a Pt–Co cluster, recorded with our Tecnai F30. The cluster is sharply facetted on {111} and {200} planes, corresponding to the truncated octahedron shown in the lower left. In principle, the bright and dark spots in Fig. 2 represent columns of atoms aligned parallel to the viewing direction. However, owing to fundamental limitations of the imaging technique and electron-optical aberrations, the position of these intensity minima and maxima may not precisely represent the true position of the atom columns, and they may not be a strict one-to-one correspondence of intensity maxima/minima and atom columns. In general, HRTEM images suffer from "delocalization." This means that the information arriving at one image point is not entirely from one object point (projection point in the exit surface of the specimen) but originates from an extended region. Conversely, the information from one object point is distributed over a certain region in the image.

In Fig. 2, these complications become apparent as one moves toward the corner of the cluster marked by the arrow. The arrow in Fig. 2 points to "ghost" lattice fringes that leach out into the vacuum. Without removing these artifacts generated by delocalize station, it is difficult to determine where the metal atom cluster ends, and a determination of atom coordinates will be highly unreliable. A powerful method for removing the artifacts generated by delocalization is exit-wave reconstruction from focus series [1,2]. This is actually a method of electron holography, by which the aberration-free complex-valued electron wave function at the exit surface of the specimen is reconstructed from a series of images recorded at different focal length settings of the objective lens. In addition to removing artifacts, this method delivers both the amplitude and the phase of the electron wave. Compared to an ordinary HRTEM image, i.e. an electron intensity distribution, this is twice as much information. Owing to the particular interaction between high-energy electrons and matter, the phase of the electron wave usually provides directly interpretable and highly precise information about the position of atom columns in the specimen. Such artifact-free images of the true atomistic structure of catalyst nanoparticles will lend themselves for analyzing subtle strain effects by digital image processing.

Fig. 1. Low resolution micrograph of a Pt/C catalyst used in a PEMFC. Arrow 1 points at a Pt particle, supported on globular carbon (arrow 2).

Fig. 2. High-resolution TEM image a Pt–Co catalyst nanoparticle. The regular pattern of spots represents atom columns, viewed in a <100> direction. The arrow points at "ghost fringes" leaching out into vacuum –an artifact of delocalization.