Elettra | Nanoscale Materials Laboratory, SuperESCA Challenging the oxidation state paradigm at the sub-nanometric scale The oxidation state is a cornerstone of chemical theory, widely used to interpret, classify, and predict the reactivity of chemical compounds. First introduced by Antoine Lavoisier in the 18th century, it remains a fundamental concept across multiple scientific disciplines. In materials science, oxidation state is particularly crucial, as it dictates an element’s chemical, physical, and thermodynamic behavior, influencing bond formation in novel compounds. Yet, defining oxidation states at the atomic scale becomes increasingly challenging as matter is confined below the nanometer threshold, where bulk properties no longer dominate. From an experimental perspective, core-level photoelectron spectroscopy— commonly known as X-ray Photoelectron Spectroscopy (XPS)—is widely recognized for its ability to provide reliable spectral fingerprints for determining atomic oxidation states. Through the seminal work of Siegbahn, XPS has become an indispensable tool in materials science. Its high sensitivity to elemental composition and surface environments makes it particularly effective for assessing oxidation states, especially in systems characterized by significant interionic charge transfer phenomena. In this study, we investigate whether core-level photoelectron spectroscopy can reliably characterize oxidation states at the subnanometric scale, where atomic aggregates are composed of just a few atoms. Our analysis focuses on the in situ oxidation of tungsten nanoclusters, specifically W13 and W25, produced using the Exact Number of Atoms in each Cluster (ENAC) cluster source developed in the Nanoscale Materials Laboratory. These clusters were deposited at 40 K onto epitaxial graphene grown on Ru(0001) to suppress surface diffusion. To monitor oxidation, we exploited the intense photon flux available at the SuperESCA beamline, enabling real-time acquisition of W 4f core-level spectra during oxygen exposure. The spin–orbit splitted W 4f peaks progressively shift to higher binding energies with increasing O2 exposure, stabilizing after a few Langmuir, indicative of saturation in oxygen adsorption. However, the resulting spectral complexity, much greater than that observed for atomic or molecular adsorbates on solid surfaces, required a theoretical framework for interpretation. We therefore employed, in a preliminary approach, density functional theory calculations to support the analysis. We systematically introduced increasing amounts of oxygen atoms into the clusters (Fig. 1 a) for the W13 case), reaching a maximum W/O ratio near 1:3. Oxygen adsorption induces pronounced geometric distortion, notably increasing the average W–W distance from 3.74 Å (W13) to 5.79 Å (W13O39). For all configurations in Fig. 1 (a), W 4f core levels were calculated via DFT, including final-state effects, and were than grouped into six families of atoms, each defined by the number n of W–O bonds and were used to construct the spectral lineshape for each W-nO family to fit the experimental data (see Fig. 1 b). It is worth noting that there is a large variability of the average W-O bond lengths in the nanoclusters, which decrease as core level binding energies increase, as can be observed in Fig. 1 (c). To better understand the correlation between core levels and oxidation states at the nanoscale, we applied Pauling’s bond valence formalism, which relates the bond valence Si of atom i to bond lengths Rij via Si = ∑j sij = ∑j exp[(R0− Rij)/b], where R0 and b are bond-specific constants, Rij is the bond length, and sij the bond order. Using this method, we computed valences for all W atoms across oxygen coverages. The parameters R0 (1.9 Å) and b (0.33) correspond to standard W–O bonds in tungsten compounds. The calculated core levels plotted against valences are shown in Fig. 1 (d). The graph shows a strong linear correlation between core levels and valence with mean core levels, valences, and standard deviations for each W-nO family indicated by colored rectangles. Two key observations emerge: (i) for each atomic coordination, core level dispersion is significant, ranging from 0.24 eV to over 0.68 eV, especially for 4O coordination; and (ii) notable overlap occurs between 2O–3O and 4O–5O coordination families. This clearly indicates that even highresolution core level measurements cannot precisely determine valency or atomic oxidation state in small clusters. This variability stems from the fluctuating interatomic distances in nanoclusters composed of few atoms, whose arrangements differ substantially from bulk crystal structures. In bulk tungsten, despite identical oxygen coordination, W–O bond lengths are fixed and vary between compounds; this is not the case for nanostructures, thus challenging conventional oxidation state assignments. The difficulty to directly and accurately determine oxidation states by XPS in nanomaterials has important implications for catalyst chemical property assessments. Since valence states strongly influence reactivity, selectivity, and catalytic pathways in metal-oxides, recognizing the heterogeneity in interatomic distances and valences is critical—not only for tungsten oxides but broadly across nanocluster oxides. Conversely, this configurational diversity in subnanometric catalysts offers opportunities by providing varied local environments on the same metaloxide. Our findings advocate moving beyond the single-active site model toward a statistical interpretation linking nanocatalyst structure and performance. Elettra Highlights 2024–2025 120
RkJQdWJsaXNoZXIy MjQ4NzI=