Elettra | Nanoscale Materials Laboratory, SuperESCA Oxygen adsorption limited to the uppermost atomic layer in in ultrasmall Pt clusters The reactivity of surface atoms in chemical reactions has been extensively investigated over the years alongside the concept of reaction site. It is well established indeed that the coordination of an atom significantly influences its reactivity. However, when dealing with sub-nanometric systems such as in the case of nanoclusters, the concept of surface atoms becomes ambiguous, and disentangling the contributions of non equivalent atoms in a reaction poses a significant challenge. Understanding the role of individual atoms in atomic aggregated formed by just a few atoms is particularly important for noble metals like Pt, which is widely used in electrochemistry and nanocatalysis, and whose oxidized clusters show a chemical activity that is 100-1000 times higher than individual Pt atoms. In this study, we demonstrate that combining high-resolution x-ray photoelectron spectroscopy with density functional theory it is possible to attribute the oxidation of size-selected Pt nanoclusters exclusively to their uppermost atomic layer. Pt nanoclusters with 12 and 13 atoms were generated by means of the Exact Number of Atoms in each Cluster (ENAC) cluster source, developed by the Nanoscale Materials Laboratory, and deposited at 40 K directly in situ on graphene epitaxially grown on Ir(111) thanks to the connection between ENAC and the experimental chamber of SuperESCA beamline. Fig.1 (a) shows the Pt 4f core level spectra of the two clusters after the deposition. The binding energies are similar for both clusters, but almost 1 eV higher than the one of metallic bulk Pt, confirming the distinct nature of clusters with respect to bulk materials. Upon oxidation, the spectra broaden towards higher BE with two new components arising. These two components fall between known Pt bulk oxides binding energies making direct attribution challenging. For a rigorous interpretation of these results, density functional theory calculations were employed to study metallic and oxidized nanoclusters. For all the simulations, Pt core levels were computed in the final state approximation. Positively charged Pt12 and Pt13 were first relaxed in gasphase; their resulting neutral geometries were placed in the graphene-Ir moirè cell with four different orientations, to find the ground state structures. It is worth noting that, upon nanoclusters adsorption, graphene undergoes pinning towards the metallic substrate. Computed core levels of as-deposited Pt nanoclusters turned out to be correlated with both the distance between Pt atoms and substrate and their Bader charge, i.e. the electronic charge assigned to each atom. To study the oxygen adsorption, simulations were performed with a different number m of O atoms placed homogeneously on the Pt clusters (m=2, 4, 6, 8), as well as the stoichiometric PtnOn configurations. At low O densities (m<=4), the two clusters behave similarly, while for higher O concentrations significant structural differences arise, with only Pt12 undergoing a strong rearrangement. A linear correlation has been found between computed core levels and coordination with O, with an average shift of 80 meV per O atom. Nevertheless, none of these configurations reproduce adequately the experimental spectra (see Fig. 1 b). Particularly, the only way to obtain a good agreement with the data is by assuming that part of the clusters are less oxidized (m=4) and part of them are strongly oxidized (m>12), which cannot be justified from a statistical point of view. Given these results, we then placed O atoms on Pt clusters in a non homogeneous way, either concentrated on the top of the clusters or on the bottom, close to graphene. Remarkably, these two arrangements lead to different simulated spectra, with only the configuration of O atoms placed at the top of the clusters returning a good fit to the data (Fig. 1 c). The optimized structures of these clusters show a PtO4 squared planar motif with the Pt atom at the center, a typical characteristic of Pt3O4 and PtO bulk oxides. To shed light onto this oxidation mechanism, we evaluated the interaction energy of an O2 molecule as a function of its distance from graphene. The resulting potential energy curve exhibits a physisorption minimum at 3.15 Å from the carbon layer, as displayed in Fig. 1 (d). This distance is located around the second atomic layer of Pt nanoclusters and, given that the diffusion of physisorbed O2 molecules is high, we can assume there is a high concentration of oxygen molecules at distances greater than 3.15 Å from graphene, which leads to the preferential oxidation of Pt atoms at the top of the clusters. Additionally, the enhanced reactivity of these top-layer atoms can be attributed to their slightly negative Bader charge, in contrast to the partially oxidized atoms at the cluster–graphene interface. Moreover, upon oxygen density increasing, one of the two topmost atoms in Pt12 moves towards the second atomic layer, being still accessible by O2 physisorbed molecules but, at the same time, lowering the coordination of the remaining topmost atom, with a consequent change in local density of states and Bader charge. Pt13 on the contrary does not undergo such a structural reconstruction, in agreement with previous results that found that Pt12 is more reactive than Pt13. Our results highlight that a mechanism resembling oxygen reverse spillover can occur on graphene, where O2 adsorption on Pt nanoclusters followed by photoinduced oxidation is promoted by graphene pinning and local variations in Bader charge at the cluster’s apex. Elettra Highlights 2024–2025 122
RkJQdWJsaXNoZXIy MjQ4NzI=