The UV photodesorption of molecular oxygen from a reduced TiO2(110) single-crystal surface was investigated as a function of photon excitation energy, substrate temperature, and preannealing conditions. A pump-delayed-probe method using pulsed lasers for UV excitation (pump) and VUV ionization (probe) were used in conjunction with time-of-flight mass spectrometry to measure velocity distributions of the desorbing O2 molecules. The measured velocity distributions exhibit three distinct features, two of which are attributed to prompt desorption resulting in “fast” velocity distributions and one “slow” channel whose average kinetic energy tracks the surface temperature. The latter is assigned to trapping-desorption of photoexcited O2* which are trapped in the physisorption well prior to thermal desorption. The velocity distributions show no dependence on photon energy over the range studied (3.45−4.16 eV), consistent with a substrate-mediated, hole-capture desorption mechanism. The observed prompt desorption channels have mean translational energies of ∼0.14 and ∼0.50 eV and are attributed to the photodesorption of two distinct initial states of chemisorbed oxygen. The identities of the chemisorbed initial states associated with oxygen vacancy or interstitial defect sites are discussed in light of previous experimental and theoretical studies of oxygen on reduced TiO2(110) surfaces.

The nature of the promotional effect of Fe addition to Rh/TiO2 and Rh/CeO2 catalysts for CO hydrogenation was investigated using FT-IR spectroscopy in an ultrahigh vacuum compatible transmission IR cell. CO adsorption experiments on Rh and FeRh showed vibrational signatures characteristic of linear and bridge bound CO on Rh0 as well as geminal-dicarbonyl species associated with Rh+. Compared to TiO2, the CeO2-supported catalysts show increased dispersion, reflected by decreased particle size, and a lower signal for linear versus geminal-dicarbonyl bonded CO. The absorption frequencies for CO on Rh/CeO2 are also redshifted relative to Rh/TiO2, which results from a weaker Rh–CO interaction, likely due to the increased reducibility of the CeO2 support. Upon addition of Fe, a new spectral feature is observed and attributed to CO bound to Rh in close contact with Fe, likely as a surface alloy. CO hydrogenation on (Fe)Rh catalysts on both supports was also studied. Compared to bare Rh, Fe containing catalysts promote formate and methoxy species on the surface at lower temperature (180 °C), which suggests an enhancement in methanol selectivity by Fe addition. At higher temperatures (220 °C), the spectral features appear similar, further confirming the role of Fe as a disrupter of large Rh0 crystallites and regulator of CO dissociation and CH4 formation.

In this Perspective, we review recent studies of size-selected cluster deposition for catalysis applications performed at the U.S. DOE National Laboratories, with emphasis on work at Argonne National Laboratory (ANL) and Brookhaven National Laboratory (BNL). The focus is on the preparation of model supported catalysts in which the number of atoms in the deposited clusters is precisely controlled using a combination of gas-phase cluster ion sources, mass spectrometry, and soft-landing techniques. This approach is particularly effective for investigations of small nanoclusters, 0.5–2 nm (<200 atoms), where the rapid evolution of the atomic and electronic structure makes it essential to have precise control over cluster size. Cluster deposition allows for independent control of cluster size, coverage, and stoichiometry (e.g., the metal-to-oxygen ratio in an oxide cluster) and can be used to deposit on any substrate without constraints of nucleation and growth. Examples are presented for metal, metal oxide, and metal sulfide cluster deposition on a variety of supports (metals, oxides, carbon/diamond) where the reactivity, cluster–support electronic interactions, and cluster stability and morphology are investigated. Both UHV and in situ/operando studies are presented that also make use of surface-sensitive X-ray characterization tools from synchrotron radiation facilities. Novel applications of cluster deposition to electrochemistry and batteries are also presented. This review also highlights the application of modern ab initio electronic structure calculations (density functional theory), which can essentially model the exact experimental system used in the laboratory (i.e., cluster and support) to provide insight on atomic and electronic structure, reaction energetics, and mechanisms. As amply demonstrated in this review, the powerful combination of atomically precise cluster deposition and theory is able to address fundamental aspects of size-effects, cluster–support interactions, and reaction mechanisms of cluster materials that are central to how catalysts function. The insight gained from such studies can be used to further the development of novel nanostructured catalysts with high activity and selectivity.Keywords: cluster deposition; heterogeneous catalysis; metal oxide; molybdenum sulfide; size-selected; transition metal; work function

Size-selected niobium oxide nanoclusters (Nb3O5, Nb3O7, Nb4O7, and Nb4O10) were deposited at room temperature onto a Cu(111) surface and a thin film of Cu2O on Cu(111), and their interfacial electronic interactions and reactivity toward water dissociation were examined. These clusters were specifically chosen to elucidate the effects of the oxidation state of the metal centers; Nb3O5 and Nb4O7 are the reduced counterparts of Nb3O7 and Nb4O10, respectively. From two-photon photoemission spectroscopy (2PPE) measurements, we found that the work function increases upon cluster adsorption in all cases, indicating a negative interfacial dipole moment with the positive end pointing into the surface. The amount of increase was greater for the clusters with more metal centers and higher oxidation state. Further analysis with DFT calculations of the clusters on Cu(111) indicated that the reduced clusters donate electrons to the substrate, indicating that the intrinsic cluster dipole moment makes a larger contribution to the overall interfacial dipole moment than charge transfer. X-ray photoelectron spectroscopy (XPS) measurements showed that the Nb atoms of Nb3O7 and Nb4O10 are primarily Nb5+ on Cu(111), while for the reduced Nb3O5 and Nb4O7 clusters, a mixture of oxidation states was observed on Cu(111). Temperature-programmed desorption (TPD) experiments with D2O showed that water dissociation occurred on all systems except for the oxidized Nb3O7 and Nb4O10 clusters on the Cu2O film. A comparison of our XPS and TPD results suggests that Nb5+ cations associated with Nb═O terminal groups act as Lewis acid sites which are key for water binding and subsequent dissociation. TPD measurements of 2-propanol dehydration also show that the clusters active toward water dissociation are indeed acidic. DFT calculations of water dissociation on Nb3O7 support our TPD results, but the use of bulk Cu2O(111) as a model for the Cu2O film merits future scrutiny in terms of interfacial charge transfer. The combination of our experimental and theoretical results suggests that both Lewis acidity and metal reducibility are important for water dissociation.

Two-photon photoemission spectroscopy (2PPE) was employed to investigate the electronic interactions at the interface of size-selected metal oxide clusters (Mo3O9, W3O9, Ti3O6, Mo3O6, W3O6, and Ti5O10) and a Cu(111) surface. The cluster–Cu interactions were probed by work function shifts measured by 2PPE as a function of local cluster coverage. For all the clusters studied, the work functions shifted to higher energies after cluster deposition, indicating negative interfacial dipole moments pointing toward the surface. The magnitudes of the derived interfacial dipoles are found to be in the order Mo3O9 ≈ W3O9 > W3O6 ≈ Mo3O6 > Ti5O10 > Ti3O6. DFT calculations of the electrostatic potentials at the interface and Bader charge analyses were used to assess the relative contributions of electron transfer and the structure-dependent cluster dipole moment to the observed work function shifts (ΔΦ). For the fully oxidized Mo3O9 and W3O9 clusters (+6 cation oxidation states), DFT calculations indicate that electron transfer from the Cu(111) support to the cluster is the dominant contribution. The smaller interfacial dipole moments for the Mo3O6 and W3O6 clusters are qualitatively consistent with the decreased ability of the reduced cations (+4 oxidation state) to accommodate charge from the Cu surface. The DFT calculations also predict small changes in ΔΦ for the titania clusters on Cu(111) but in the opposite direction of that observed experimentally. In the case of the Ti5O10/Cu(111) surface, this result is due to the net balance of cluster dipole and electron transfer contributions that have opposite signs. Overall, the results presented in this study show that a combination of coverage-dependent work function measurements and DFT calculations can be a powerful tool to investigate the electronic interactions, especially electron transfer, at the metal oxide–metal interface.

The photooxidation of ethanol and 2-propanol was studied under UHV conditions on a single crystal TiO2(110) surface using a combination of temperature programmed desorption (TPD) and pump–probe laser ionization techniques. Previous studies of these reactions have shown that the first step involves photocatalytic dehydrogenation to either an acetaldehyde or acetone intermediate. In this work, we show that when adsorbed alcohols are irradiated with UV light in the presence of molecular oxygen, methyl radicals are ejected from the surface. Furthermore, it is shown that these radicals possess kinetic energy distributions which are remarkably similar to those measured for the photooxidation of acetaldehyde and acetone. This result suggests that methyl radicals are produced during a second photocatalytic step which involves photooxidation of the aldehyde/ketone intermediates.

•Alloy oxidation varies depending on treatment.•O2 treatment results in Ga2O3 domains, conforming to previous work.•NO2 treatment at 700 K results in mixed CoGa surface oxide film.It has been shown that a Ga2O3 film forms on the surface of CoGa alloy crystals when exposed to oxygen (Pan, 2001 and Vlad, 2010). In this work we report the results of the characterization of surface oxides on CoGa(100) using X-ray photoelectron spectroscopy (XPS), low energy electron diffraction (LEED), and ion scattering spectroscopy (ISS). The oxides were synthesized using either O2 or NO2 as the oxidant at 300 K or in excess of 700 K. ISS scans showed that cobalt was always present in the top surface layer regardless of oxidation conditions. XPS showed that depending on the oxidant and the temperature, the composition of the oxide films vary depending on oxidation treatment, with some oxides being nearly all Ga2O3 and ordered with a sharp LEED pattern consisting of (2 × 1) domains rotated by 90º and others being Co–Ga mixed oxides that gave no diffraction pattern.

The UV photooxidation of 2-butanone on TiO2(110) was studied using pump–probe laser methods and time-of-flight (TOF) mass spectrometry to identify the gas-phase photoproducts and probe the dynamics of the photofragmentation process. A unique aspect of this work is the use of coherent VUV radiation for single-photon ionization detection of gas-phase products, which significantly reduces the amount of parent ion fragmentation as compared to electron impact used in previous studies. The pump–probe product mass spectra showed ions at mass 15 (CH3+) and mass 29 (C2H5+), which are associated with the primary α-carbon bond cleavage of the adsorbed butanone–oxygen complex, as well other C2Hx+ (x = 2–4) fragments, which could originate from ethyl radical secondary surface chemistry or dissociative ionization. Using two different VUV probe energies, it was possible to show that the fragment ions at mass 27 (C2H3+) and mass 28 (C2H3+) are not due to secondary reactions of ethyl radicals on the surface, but rather from dissociative ionization of the ethyl radical parent ion (mass 29). Another photoproduct at mass 26 (C2H2+) peak is also observed, but its pump–probe delay dependence indicates that it is not associated with nascent ethyl radicals. Pump-delayed-probe measurements were also used to obtain translational energy distributions for the methyl and ethyl radical products, both which can be empirically fit to “fast” and “slow” components. The ethyl radical energy distribution is dominated by the “slow” channel, whereas the methyl radical has a much larger contribution from “fast” fragments. The assignment of the C2Hx (x = 3, 4) fragments to ethyl (C2H5) dissociative ionization was also confirmed by showing that all three products have the same translational energy distributions. The origin of the “fast” and “slow” fragmentation channels for both methyl and ethyl ejection is discussed in terms of analogous neutral and ionic fragmentation processes in the gas phase. Finally, we consider the possible energetic and dynamical origins of the higher yield of ethyl radical products as compared to that for methyl radicals.

In this work, we report on product energy distributions for methyl radicals produced by UV photooxidation of a set of structurally related carbonyl molecules, R(CO)CH3 (R = H, CH3, C2H5, C6H5), adsorbed on a TiO2(110) surface. Specifically, laser pump–probe techniques were used to measure the translational energy distributions of methyl radicals resulting from α-carbon bond cleavage induced by photoexcited charge carriers at the TiO2 surface. Photoreaction requires the presence of co-adsorbed oxygen and/or background oxygen during UV laser (pump) exposure, which is consistent with the formation of a photoactive oxygen complex, i.e., η2-bonded diolate species (R(COO)CH3). The methyl translational energy distributions were found to be bimodal for all molecules studied, with “slow” and “fast” dissociation channels. The “fast” methyl channel is attributed to prompt fragmentation of the diolate species following charge transfer at the TiO2 surface. The average translational energies of the “fast” methyl channels are found to vary with R-substituent and correlate with the mass of the remaining surface fragments, RCOx (x =1 or 2). By comparison, the average energies of the “slow” methyl channels do not show any obvious correlation with R-substituent. The apparent correlation of the “fast” methyl translation energies with surface fragment mass is consistent with a simple two-body fragmentation event isolated on the diolate molecule with little coupling to the surface. These results also suggest that the total available energy for methyl fragmentation does not vary significantly with changes in R-substituent and is representative of exit barriers leading to “fast” methyl fragments.

The UV photooxidation of acetone on a reduced TiO2(110) surface was investigated using a combination of photodesorption and thermal desorption measurements and pump–probe laser detection of gas-phase products. In agreement with earlier studies, acetone adsorbed on TiO2 does not undergo a UV photoreaction unless codosed with molecular oxygen. The only gas-phase photoproducts are methyl radicals originating from fragmentation of the active acetone surface species and photodesorbed molecular oxygen. Postirradiation TPD measurements show that acetate is the primary surface product remaining after photooxidation. The dependence of the methyl radical formation rate on oxygen and thermal pretreatment of the TiO2 surface is consistent with the formation of an acetone–oxygen (diolate) complex involving adsorbed acetone and oxygen adatoms. Pump-delayed-probe laser techniques were used to measure the velocity and translational energy distributions of methyl radicals resulting from fragmentation of the acetone diolate. The observed translational energy distributions are well described by empirical fits involving two components with average energies of 0.19 eV (“fast”) and 0.03 eV (“slow”). The latter are found to be insensitive to surface temperature or preannealing conditions, suggesting that the “fast” and “slow” components represent different final states of methyl radicals originating from fragmentation of a single photoactive species. The methyl kinetic energy distributions were also found to be independent of UV pump energy which is consistent with a substrate-induced process involving thermalized charge carriers, electrons or holes, which transfer to the acetone diolate to induce fragmentation. The results are discussed in terms of probable substrate-induced photoreaction mechanisms and analogous molecular photofragmentation processes.

The electronic structure of thiophene adsorbed on Au(111) has been investigated by two-photon photoemission (2PPE) spectroscopy and density functional theory (DFT) calculations. The dominant interfacial feature observed in the 2PPE spectra is a nondispersive unoccupied state whose width and energy (referenced to the Fermi level) decrease with increasing coverage. We assign this feature to a thiophene LUMO-derived state of mixed sulfur and carbon p−π character. DFT calculations indicate that the experimentally observed decrease in width of this state with increasing coverage is a result of weakening of the thiophene−Au(111) interaction. Increasing the molecular density forces the thiophene plane to tilt away from the surface, rotating the molecular orbitals away from an orientation more favorable for S−Au and π−Au hybridization. We attribute the ∼0.2 eV shift of the LUMO toward the Fermi level to stabilization of the transient anion due to the increasing effect of charge-induced polarization of the neighboring thiophene molecules with increasing coverage.

A combination of experiment and density functional theory was used to investigate the energetics of CO adsorption onto several small MxSy+ (M = Mo, W; x/y = 2/6, 3/7, 5/7, 6/8) clusters as a probe of their atomic and electronic structure. Experimentally, tandem mass spectrometry was used to measure the relative yields of MxSy+(CO)n cluster adducts formed by collisions between a beam of mass-selected MxSy+ cluster ions and CO molecules in a high-pressure collision cell (hexapole ion guide). The most probable MxSy+(CO)n adducts observed are those with n ≤ x, that is, only one CO molecule bound to each metal site. The notable exception is the M5S7+ cluster, for which the n = 6 adduct is found to have nearly the same intensity as the n = x = 5 adduct. Density fuctional calculations were used to search for the lowest energy structures of the bare MxSy+ clusters and to obtain their relative stability for sequential CO binding. The calculated trends in CO binding energies were then compared to the experimental adduct distributions for assigning the ground-state structures. In this way, it was possible to distinguish between two nearly isoenergetic ground-state isomers for the M2S6+ and M3S7+ clusters, as only one isomer gave a calculated CO stabilization energy trend that was consistent with the experimental data. Similar comparisons of predicted and observed CO adsorption trends also provide evidence for assigning the ground-state structures of the M5S7+ and M6S8+ clusters. The latter contain metallic cores with most of the sulfur atoms bonded along the edges or in the faces of the metal core structure. The n = 6 and 7 adducts of M5S7+ are predicted to be more stable than the n = x = 5 adduct, but only the n = 6 adduct is observed experimentally. The DFT calculations show that the n = 7 adduct undergoes internal bond breaking whereas the n = 6 framework is stable, albeit highly distorted. For the M6S8+ cluster, the calculations predict that the two lowest energy isomers can bind more than six CO molecules without fragmentation; however, the apparent binding energy drops significantly for adducts with n > 6. In general, the ability of these small MxSy+ clusters to bind more CO molecules than the number of metal atoms is a balance between the gain in CO adsorption energy versus the strain introduced into the cluster structure caused by CO crowding, the consequences of which can be fragmentation of the MxSy+(CO)n cluster adduct (n > x).

Mass-selected cluster deposition was used to investigate the chemical and thermal properties of the Mo 4S 6 cluster deposited onto a Au(111) substrate. Auger spectroscopy and (13)CO thermal desorption measurements demonstrate that the clusters behave independently up to coverages of ∼0.15 ML, while at higher coverages, cluster crowding or island formation results in no net increase in Mo-atom adsorption sites. DFT calculations show that CO binding on the Mo-atom top site is strongly preferred over the side sites, with scaled binding energies in reasonable agreement with the experimentally derived binding energy of 0.7 eV. DFT calculations predict that the total adsorption energy for sequential addition of two CO molecules (top and side site) is nearly additive, whereas the addition of a third CO to another empty Mo side site is less stable. The latter is attributed to repulsive intercluster interactions and is consistent with the experimentally estimated sticking coefficient of 0.4 ± 0.1. In contrast to CO, we were unable to detect any adsorption of NH 3 onto the deposited cluster. The DFT calculations confirm these observations by predicting a very small NH 3 adsorption energy to the Mo 4S 6/Au(111) supported cluster. The difference in adsorbate binding (CO, NH 3) between the gas-phase and supported cluster highlights the role of the Au(111) substrate in modifying the electronic structure and chemical behavior of the supported cluster. Annealing of the Mo 4S 6/Au(111) surface above ∼500 K was found to significantly reduce the CO uptake of the supported clusters. These data are consistent with diffusion of intact clusters along the Au surface and the formation of 2D islands. Because of the unique stoichiometry of the as-deposited Mo 4S 6 clusters, aggregates formed by cluster diffusion are expected to exhibit distinctly different chemical behavior compared to near-stoichiometric MoS x ( x ≈ 2) platelet nanoclusters or amorphous thin films.

In this work, we report on product energy distributions for methyl radicals produced by UV photooxidation of a set of structurally related carbonyl molecules, R(CO)CH3 (R = H, CH3, C2H5, C6H5), adsorbed on a TiO2(110) surface. Specifically, laser pump–probe techniques were used to measure the translational energy distributions of methyl radicals resulting from α-carbon bond cleavage induced by photoexcited charge carriers at the TiO2 surface. Photoreaction requires the presence of co-adsorbed oxygen and/or background oxygen during UV laser (pump) exposure, which is consistent with the formation of a photoactive oxygen complex, i.e., η2-bonded diolate species (R(COO)CH3). The methyl translational energy distributions were found to be bimodal for all molecules studied, with “slow” and “fast” dissociation channels. The “fast” methyl channel is attributed to prompt fragmentation of the diolate species following charge transfer at the TiO2 surface. The average translational energies of the “fast” methyl channels are found to vary with R-substituent and correlate with the mass of the remaining surface fragments, RCOx (x =1 or 2). By comparison, the average energies of the “slow” methyl channels do not show any obvious correlation with R-substituent. The apparent correlation of the “fast” methyl translation energies with surface fragment mass is consistent with a simple two-body fragmentation event isolated on the diolate molecule with little coupling to the surface. These results also suggest that the total available energy for methyl fragmentation does not vary significantly with changes in R-substituent and is representative of exit barriers leading to “fast” methyl fragments.

The photooxidation of ethanol and 2-propanol was studied under UHV conditions on a single crystal TiO2(110) surface using a combination of temperature programmed desorption (TPD) and pump–probe laser ionization techniques. Previous studies of these reactions have shown that the first step involves photocatalytic dehydrogenation to either an acetaldehyde or acetone intermediate. In this work, we show that when adsorbed alcohols are irradiated with UV light in the presence of molecular oxygen, methyl radicals are ejected from the surface. Furthermore, it is shown that these radicals possess kinetic energy distributions which are remarkably similar to those measured for the photooxidation of acetaldehyde and acetone. This result suggests that methyl radicals are produced during a second photocatalytic step which involves photooxidation of the aldehyde/ketone intermediates.