We studied the structural and morphological evolution of Rh clusters on an ordered ultrathin alumina film grown on NiAl(100) in annealing processes, under ultrahigh vacuum conditions and with various surface probe techniques. The Rh clusters, prepared on vapor deposition of Rh onto the alumina film at 300 K, had an fcc phase and grew in the (100) orientation; the annealing altered the cluster structure little—the lattice parameter decreased by a factor <2%—but the cluster morphology significantly. With elevated temperature, small clusters (diameter ≤1.5 nm) decreased little in size; in contrast, large clusters (diameter ≥2.0 nm) varied in a complex manner—their mean diameter decreased to about 1.5 nm on annealing to 450 K, despite their similar height, while it increased to above 2.0 nm at temperature ≥570 K. This atypical decrease in size was governed predominantly by energetics. Such a reduced size enhanced the total surface area as well as the reactivity of the clusters toward methanol decomposition, so increased the production of D2 (H2) and CO from decomposed methanol-d4 (or methanol). The result implies a higher temperature tolerance for Rh clusters on the alumina film and a practical approach to prepare small Rh clusters with high reactivity.

Pt and Rh nanoclusters, grown on deposition of Pt and Rh vapors onto graphene/Pt(111), show separate reactivity toward the decomposition of methanol-d4. The Pt (Rh) clusters had a mean diameter 2.0–3.5 nm (2.1–4.0 nm) and height 0.45–0.94 nm (0.41–0.9 nm) evolving with the coverage; they were structurally ordered, having an fcc phase and growing in (111) orientation, and had lattice constants similar to their bulk values. Methanol-d4 on the Pt clusters did not decompose but desorbed mostly, disparate from that on Pt(111) surface; the disparity arose as the adsorption energies of methanol-d4 on most surface sites of the Pt clusters became smaller than their single crystal counterpart. This size effect, nevertheless, did not apply on the Rh clusters, despite their similar atomic stacking; the Rh clusters showed a reactivity similar to that of the Rh(111) surface because the adsorption energies of methanol-d4 on both Rh clusters and Rh(111) are comparable. The distinct size dependence was rationalized through their electronic structures and charge distribution of Fukui function mapping. Our results suggest that reactive transition metals do not necessarily become more reactive while they are scaled down to nanoscale; their reactivity evolves with their size in a manner largely dependent on their electronic nature.

Annealing alters the support effect to promote the activity of alumina-supported Rh clusters in methanol dehydrogenation.

The size effect on the activity of a catalyst has been a focal issue since ideal catalysts were pursued, whereas that on the degradation of a catalyst, by reaction intermediates such as CO, is little discussed. We demonstrate that the dehydrogenation of methanol-d 4 on supported Rh nanoclusters precovered with CO (Rh −CO clusters) was obstructed, indicated by a decreased production of CO and D 2 ; the obstructive effect exhibits a remarkable dependence on the cluster size, with a minimum at a cluster diameter near 1.4 nm. The decreased production arose from a decreased reaction probability controlled by the increased activation energy for each dehydrogenation step (including formation of methoxy-d 3 ), adsorption energies of CO, and repulsion from the CO array on the Rh −CO surface. The effects of these factors in deactivating the clusters varied separately with the cluster size. Consequently, the size effect on the CO poisoning should be taken into account in engineering the cluster size to optimize the catalytic performance.