Reactivity of CO on Sulfur-Passivated Graphene-Supported Palladium

Article pubs.acs.org/JPCC

Reactivity of CO on Sulfur-Passivated Graphene-Supported Palladium Nanocluster Arrays Fabian Düll, Florian Spaẗ h, Philipp Bachmann, Udo Bauer, Hans-Peter Steinrück, and Christian Papp* Lehrstuhl für Physikalische Chemie II, Universität Erlangen-Nürnberg, Egerlandstr. 3, 91058 Erlangen, Germany S Supporting Information *

ABSTRACT: Nanocluster arrays on graphene are suitable model systems for catalysis. We use this model system to study sulfur poisoning, which is a major deactivation process of heterogeneous catalysts. Using high-resolution X-ray photoelectron spectroscopy, we investigated the adsorption and desorption of CO on sulfurpoisoned graphene-supported palladium nanoparticles. We find that sulfur blocks CO adsorption sites, with hollow sites being most affected. From the unchanged desorption temperatures, we conclude that the Pd−CO bond strength is not altered by coadsorbed sulfur. The degree of site blocking compared to the amount of sulfur indicates that the palladium nanoparticle surface is dominated by (111) and (100) facets.

■

INTRODUCTION Palladium/graphene nanocomposites1 are catalyst materials with a wide area of applications like oxidation of alcohols2 and formic acid, C−C cross coupling,3,4 and hydrogen sensing.5 Although the synthetic approach in these studies, using exfoliated graphene and wet-chemical dispersion of Pd particles, is easily scalable, the produced systems are not well-defined and do not allow for investigating the fundamental properties of Pd nanoclusters on graphene. By use of the lattice mismatch of Rh(111) and graphene, an intrinsic Moiré pattern of graphene is produced.6 Its template effect enables the growth of regular arrays of nanoclusters with a narrow size distribution.7−11 They can be used as model systems to investigate chemical reactions and the related kinetics.12−14 This model system is close to the applied catalysts and might serve to bridge the materials gap to flat single crystals and regularly stepped crystals. The surfaces of the Pd nanoparticles typically are a mixture of differently orientated facets, steps, and kinks. They have a narrow size distribution and a rather innocent support, while real catalysts consist of differently sized metal particles on often complex support materials; the latter often have a crucial influence on the catalytic activity. To characterize model systems, CO is often used as a probe molecule. On Pd(111),15 CO adsorbs mainly on threefold hollow and on-top sites, while bridge sites are preferred on Pd(110)16 and Pd(100).17 Stepped surfaces, for example, Pd(211),18 Pd(210),19 Pd(311),19 and Pd(510),20 offer additional yet defined complexity by introducing different step adsorption sites. For palladium, the influence of steps on CO adsorption depends on the crystallographic orientation of the surface: While on Pd(211) CO adsorbs preferentially at step sites for low coverages,21 terrace adsorption sites are preferred on Pd(510).20 Pd(211) has (100)-like steps and (111)-like terraces, while Pd(510) has (110)-like steps and (100)-like terraces. The above-mentioned behavior thus © XXXX American Chemical Society

indicates a preference of (100) adsorption sites over (111) and (110) adsorption sites for CO on palladium. Heterogeneous catalysts suffer from five kinds of deactivation processes: poisoning, coking, sintering, solid-state transformation, and loss of active material by volatilization and abrasion.22−24 A major cause for the deactivation of palladium catalysts25 is poisoning by sulfur.26,27 Catalyst poisons work via three mechanisms: physical blocking of the active surface area, changes in the electronic or geometric surface structure, or hindering of the diffusion of other adsorbates. Real catalysts often show a loss of activity in the treatment of exhaust gases from burning of natural gas28 and gasoline.29 These fuels contain gases like SO2 and H2S as a sulfur source that leads to partial deactivation of palladium.26,30 In other applications, the modification of the catalyst by sulfur is not considered a problem but is even welcome. For example, sulfur enhances the selectivity of the conversion of butadiene to butenes on Pd/ SiO2.31 Because of the wide use of palladium as a catalyst in large-scale applications, a detailed understanding of the poisoning mechanism of sulfur is of high interest. Since CO is a common probe molecule in surface science, the influence of sulfur on the adsorption and desorption of CO was studied on Pd foils32 and Pd(100) single crystals.33 A coverage-dependent poisoning mechanism was suggested.34 While for low sulfur coverages CO−S interactions prevent the formation of high-density CO structures, direct blocking of CO adsorption sites occurs for high sulfur coverages. With sulfur binding to fourfold hollow sites on Pd(100), as deduced by LEED I(V) (low-energy electron diffraction intensity analysis),17,35 a quarter of a monolayer of sulfur completely blocks Received: November 15, 2016 Revised: December 28, 2016 Published: December 28, 2016 A

DOI: 10.1021/acs.jpcc.6b11480 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 1. Adsorption of 2 × 10−9 mbar of CO on 0.53 ML of Pd/graphene/Rh(111) at 150 K for a sulfur precoverage of θS,rel = 0.14 ML: (a) selected C 1s spectra during adsorption up to exposures of 1.2 L (blue spectrum); (b) fit of C 1s spectrum at 1.2 L CO; (c) difference spectrum of (b) with graphene and carbide signal subtracted; (d) selected difference spectra collected during adsorption, with the corresponding CO coverages denoted in the figure.

resolution of 150 meV, 450 eV for the Pd 3d5/2 region with 260 meV resolution, and 260 eV for the S 2p region with 200 meV resolution. All spectra were referenced to the Fermi level. After subtraction of a linear background, the spectra were fitted with asymmetric Doniach-Sunjic functions convoluted with a Gaussian function.40 All S 2p doublets exhibit an S 2p3/2:S 2p1/2 ratio of 2 and a binding energy difference of 1.19 eV. CO coverages were calibrated by assuming a coverage of θCO,rel = 0.75 ML as CO saturation on sulfur-free Pd particles at 150 K (1 ML (monolayer) is defined as 1 CO molecule per surface atom). This value corresponds to the saturation coverage of CO on Pd(111),41 Pd(100),42 and Pd(110)16 single crystals at this temperature. The absolute sulfur coverage was calibrated using a c(2 × 2) overlayer of S with a coverage of 0.5 ML on a Pd(100) single crystal.43 To gain relative sulfur coverages θS,rel, the surface of the particles was determined by comparing the integrals of the CO-saturated nanoparticles and 0.75 ML CO on Rh(111) forming a (2 × 2) structure.44 This comparison yields a value of θCO,abs = 0.27 ML; that is, the number of surface atoms of the particles is 36% of the number of surface atoms of the Rh(111) substrate. Consequently, a measured absolute sulfur coverage of θS,tot = 0.051 ML results in a relative coverage of θS,rel = 0.14 ML on the particles; this calculation is based on the reasonable assumption that sulfur cannot bind to graphene. Graphene was grown by chemical vapor deposition of 2 × 10−8 ethene at 920 K.6 Pd was deposited on the sample via electron beam evaporation at 150 K and Pd coverages were calibrated via the QCM. To gain a stable shape of the nanoparticles, they were saturated with CO at 150 K and consecutively heated to 500 K.12 This is necessary, as the first CO adsorption/desorption cycle leads to particle sintering.

the Pd(100) surface for CO. Recent DFT (density functional theory) calculations support this hypothesis, as they show a weakening of the Pd−CO bond when one Pd atom is shared by a bridge-bonded CO and sulfur, while bridge sites between two Pd atoms both bonded to sulfur are blocked for CO.36 In this work, we report on the influence of sulfur concerning the adsorption and desorption behavior of CO on graphenesupported Pd nanocluster arrays by synchrotron-based highresolution X-ray photoelectron spectroscopy (HR-XPS37). We investigate the effects of sulfur on the different CO adsorption sites as well as the interaction between sulfur and CO.

■

EXPERIMENTAL SECTION The experiments were performed at the beamlines U49/2PGM1 and U49/2-PGM2 of BESSY II at Helmholtz-Zentrum Berlin in a transportable UHV apparatus38 consisting of two chambers. A hemispherical electron analyzer (Omicron EA 125 U7 HR), a quadrupole mass spectrometer, and a three-stage supersonic molecular beam are mounted to the analysis chamber, enabling pressures of up to 1 × 10−5 mbar on the sample surface. LEED optics, a sputter gun, an electron beam evaporator for Pd deposition, and a quartz crystal microbalance (QCM) for monitoring the deposition rate are attached to the preparation chamber. The sample temperature can be adjusted between 100 and 1400 K by liquid nitrogen cooling and resistive heating. In the back of the sample, an additional tungsten filament is mounted to perform XPS during a linear heating ramp (0.5 K/s) without disturbing electric and magnetic fields that go along with resistive heating. These experiments are referred to as temperature-programmed XPS (TP-XPS).39 All XP spectra were recorded in normal emission. The excitation energies were 380 eV for the C 1s region with a B


Article


Figure 2. Quantitative analysis obtained from fitting C 1s spectra measured during CO adsorption at a pressure of 2 × 10−9 mbar at 150 K and successive desorption of CO on sulfur-free and sulfur-poisoned Pd nanoparticles on graphene/Rh(111): (a) total CO coverage and (c) individual CO coverages (COA and COB) during adsorption as obtained by peak fitting (see, e.g., Figure 1c); (b) total CO coverage and (b) individual CO coverages during TPXPS of CO, β = 0.5 K/s.

After the first desorption experiment with CO, the number of step sites greatly decreases, while the terrace-like facets of the nanoparticles grow. Further changes are not observed in the following CO adsorption/desorption cycles. Sulfur poisoning of the particles was achieved by adsorbing SO2 at 150 K and heating the sample to 500 K. For the experiments, 2 × 10−9 mbar CO was dosed at 150 K. CO exposures are given in Langmuir (1 L = 1.33 × 10−6 mbar s).

free nanoclusters.12 A satisfying fit is achieved for all spectra by quantifying the data with these two CO species at 285.99 and 286.11 eV. As the surface of the Pd nanoparticles is considered to consist of different steps and facets,11 the assignment of the CO species is not straightforward but can be done by considering the binding energies found on single-crystalline surfaces.15,46 From this comparison, the species at 285.99 eV (COA) is assigned to CO adsorbed at Pd steps and at terrace hollow sites, while the species at 286.11 eV (COB) is assigned to CO bonded to terrace bridge and possibly top sites. Similar measurements have been performed for different sulfur coverages, that is, θS,rel = 0.00, 0.04, 0.10, and 0.14 ML. The corresponding quantitative analyses are shown in Figure 2a (total CO coverage) and Figure 2c (individual species). The measurements performed with different amounts of sulfur show that the total amount of adsorbed CO at saturation is reduced by the preadsorbed sulfur, from 0.75 ML for the sulfur-free particles to 0.42 ML for θS,rel = 0.14 ML, that is, to 56%. The analysis further shows that not only the amount of CO decreases with increasing amount of adorbed sulfur but also the relative site occupation changes. The COA species (full symbols) shows a maximum for all experiments; this maximum is reached at a lower exposure and a lower coverage of CO for increasing amounts of sulfur, which is attributed to the fact that less free highly coordinated adsorption sites remain in the presence of sulfur. At exposures above 0.3−0.4 L, the COB species (open symbols) grows, and simultaneously also a rearrangement (depopulation) of already adsorbed CO from step/hollow sites occurs. Both CO species saturate at 0.7−0.8 L, with the saturation coverage decreasing with increasing amount of coadsorbed sulfur. This effect is more distinct for the step/hollow species (COA), with a reduction by 55% (that is, from 0.375 to 0.17 ML) than for COB at bridge/top sites with a

■

RESULTS AND DISCUSSION We investigated the influence of co-adsorbed sulfur on the adsorption and desorption behavior of CO on Pd nanoclusters arrays. Selected spectra after CO adsorption at 150 K from the experiment with the highest amount of sulfur of θS,rel = 0.14 ML are shown in Figure 1. The thick black spectrum shows the pristine Pd particles after poisoning with sulfur. The blue spectrum was recorded after saturation of the particles with CO. Both spectra show the two characteristic signals of graphene, at 285.08 and 284.53 eV, resulting from different interaction strengths of the carbon atoms in the Moiré pattern with the Rh(111) substrate.45 After adsorption of CO, the graphene signals are slightly damped (blue spectrum in Figure 1a). Upon increasing the CO exposure, a CO-related peak evolves at ∼286 eV. For a quantitative analysis, the spectra are fitted. Figure 1b shows the resulting fit, with the slightly damped graphene signal (orange), a small carbide contribution at 283.83 eV (magenta) and two CO-related peaks, COA at 285.99 eV (green) and COB at 286.11 eV (blue); the latter components (after subtraction of all other components) are shown in Figure 1c. In Figure 1d, the development of the CO-related feature is shown for increasing exposure; evidently, at low coverage first the peak at 285.99 eV develops and only at higher coverage the second peak at 286.11 eV grows in. This behavior is similar to what is found on sulfurC


Article


and increases with the amount of adsorbed sulfur and therefore is assigned as Pd−S peak. With these peaks, the CO adsorption experiment and also the thermal evolution (see below) for all different S precoverages can be consistently fitted. The three Pd−CO species are a consequence of the complex shape of the Pd particles, with a large variety of possible adsorption sites. To understand this complex behavior, we first discuss the situations on single-crystal Pd surfaces. By comparison of the relative binding energies (with respect to the bulk signal) to that for single crystals,32 reasonable assumptions on the peak positions and thus on the CO adsorption sites can be made. To describe the different possible binding conformations, three species denoted as α, β, and γ are necessary. One CO molecule bonded to one Pd atom leads to a shift of ∼1 eV toward higher binding energy; for example, a shift of +1 eV is expected for a Pd atom with one CO bonded on top or with two CO molecules bonded in a bridge configuration. This shift is, for example, found for one on-top CO on Pd(111);15 also two bridge CO molecules on Pd(100) result in a shift of +1.00 eV.46 On Pd(111), shifts of +0.51 eV were observed for one bridge CO, and +0.70 eV for two hollow CO molecules.15 On Pd(100), a shift of +0.55 eV was found for one bridge CO.46 One CO bonded on a threefold hollow site leads to a shift of +0.36 eV, that is, a configuration corresponding to one-third of a CO molecule per Pd atom. In our measurements, for Pd−CO γ at 336.16 eV, a shift of +1.04 eV is found, which fits one on-top CO on Pd(111) and two bridge CO molecules on Pd(100) single crystals. Pd−CO β at 335.78 eV is shifted by +0.66 eV and thus is assigned to Pd atoms that are bonded to one CO molecule on a bridge site or two CO molecules bonded on hollow sites, with nominal shifts of 0.51 and 0.70 eV on Pd(111),15 and 0.55 eV on Pd(100).46 A shift of +0.36 eV is found for the Pd−CO α species at 335.48 eV and is assigned to hollow-bonded CO. In Figure 4a,c, the quantitative analysis of the adsorption experiments is depicted for different sulfur precoverages. As for the C 1s region, the overall behavior of the different Pd species during adsorption and desorption of CO shown in Figure 4 is not affected by sulfur. The main effect is a decrease of the coverage for all three Pd−CO species (Figure 4a). At low exposures, the Pd−CO α species (one Pd−CO hollow) increase fast and reach their maximum before 0.1 L exposure. With increasing amount of sulfur, the maximum shifts to lower CO exposures and coverages. This effect is explained by sulfur atoms adsorbing on hollow sites, and blocking these sites for CO. Therefore, saturation of the hollow sites occurs at lower exposures for S-precovered particles. For 0.14 ML S, the maximum of Pd−CO α signal is decreased by 35% as compared to the S-free particles. At higher CO exposures, Pd−CO α decreases and is first replaced by Pd−CO β and later also by Pd−CO γ. This decrease occurs at lower exposures for higher θS. This is because there are less isolated hollow sites for CO available and consequently interactions with neighboring CO molecules will happen at lower exposures. Note that the Pd− CO α signals do not start at zero, as there is already some CO adsorbed from the background during the cooldown of the sample before the start of the experiment. The Pd−CO β signal increases right from the beginning, and also shows a maximum directly before the disappearance of the Pd−CO α. This maximum is also decreased in intensity by sulfur, that is, 23% for 0.14 ML S. The Pd−CO β saturates around 0.7 L. The Pd− CO γ species appears at 0.1 L exposure and increases until saturation around 0.7 L. At CO saturation, the majority of Pd

reduction by 31% (that is, from 0.375 to 0.26 ML), for 0.14 ML S (black symbols). This behavior can be understood considering the fact that sulfur prefers to bind to highly coordinated sites, like step or hollow sites. We find that the total amount of CO is reduced from 0.75 to 0.43 ML (Figure 1a). Considering a maximal sulfur coverage of 3/7 ≈ 0.43 ML on Pd(111)47,48 and a linear decrease of the CO coverage with θS, 0.14 ML S should reduce the adsorption sites for CO by 33%, that is, from 0.75 to 0.50 ML. On the other hand, on Pd(100), total blocking of CO adsorption is achieved already for 0.25 ML S;33 here, 0.14 ML should reduce the adsorption sites by 56%, that is, from 0.75 to only 0.32 ML. The fact that our result of θCO,rel = 0.42 ML falls between these values is an indication that the surface of the nanoparticles is comprised of a mixture of (111) and (100) facets. This is also supported by a linear extrapolation that leads to a sulfur coverage of 0.36 ± 0.03 ML that would lead to a complete blocking of all adsorption sites for CO. Again this is between the results expected for (111) and (100) facets. To further study the adsorption of CO on S-poisoned Pd particles, we discuss the XP spectra of the Pd 3d region. As was shown for single-crystal surfaces,15 these spectra, together with the information deduced above, allow for a deep understanding of the CO adsorption behavior. In Figure 3, selected Pd 3d5/2 spectra measured during CO adsorption on a sulfur-precovered surface (θS,rel = 0.14 ML) at 150 K are shown.

Figure 3. Selected Pd 3d5/2 spectra measured during exposure to 2 × 10−9 mbar of CO on 0.53 ML of Pd/graphene/Rh(111) at 150 K, θS,rel = 0.14 ML. The corresponding CO coverages are denoted in the figure.

The adsorption of CO leads to strong changes of the Pd 3d signal: During CO adsorption, the signal shifts to higher binding energies, as a result of the formation of Pd−CO bonds, and only a shoulder remains at 335.12 eV. This behavior is similar to the behavior observed on single-crystalline surfaces and sulfur-free particles.12 Accordingly, peak assignment and fitting is performed analogously, using the same parameters; the corresponding fits and fit parameters are shown in the Supporting Information, in Figure S1 and Table S1. The peak at 335.12 eV is assigned to the Pd bulk, as it is not affected by the CO adsorption except a signal decrease due to damping and possibly photoelectron diffraction induced by the CO adsorbed on the clusters. Two peaks Surface 1 and Surface 2 (334.80 and 335.58 eV) account for unoccupied Pd surface atoms, and three peaks for Pd-CO α, Pd-CO β, and Pd-CO γ (335.48, 335.78, and 336.16 eV, respectively) represent Pd atoms with different interactions with CO. An additional peak is observed at 335.44 eV; it is not found for sulfur-free clusters D


Article


Figure 4. Quantitative analysis of fits of Pd 3d5/2 spectra during adsorption and desorption of CO on Pd/graphene/Rh(111) for sulfur-free and poisoned particles with different sulfur loadings: (a, c) adsorption of CO at a pressure of 2 × 10−9 mbar at 150 K; (b, d) TPXPS, β = 0.5 K/s. The results are split into (a, b) CO-related and (c, d) non-CO-related species for better visibility.

atoms is bonded to one CO on top or two bridge-bonded CO molecules, while the minority is bonded to one bridge CO or two hollow COs. Both the Pd−CO β and the Pd−CO γ are only affected by sulfur in the form of an overall decreased intensity. Thereby, the influence of sulfur on the Pd−CO γ is stronger with a loss of 36% compared to a loss of 27% for Pd− CO β for 0.14 ML S. Thus, sulfur seems to block neighboring bridge positions especially effective for CO. Parallel to the increase of the Pd− CO species, the free Pd surface decreases and vanishes around 0.45 L (Figure 4c). Also, the bulk signal is damped fast in the beginning until 0.2 L CO and then more slowly until saturation. The smaller intensity of the bulk signal for the S-covered clusters as compared to the S-free clusters (which were measured in a previous study12) is partly due to a somewhat lower total Pd coverage in the present study and damping by the adsorbed S atoms; in addition, some changes in cluster shape cannot be ruled out. Furthermore, as the cluster shape is considered as rather flat with a height of about 6.6 Å and a diameter of about 3 nm,11 an increase in the particle thickness would only slightly affect the surface area. The height of the Pd−S signal is constant over the whole experiment, but it shifts from 335.44 to 335.69 eV due to lateral interactions with coadsorbed CO. Next, we discuss the changes in the S 2p core level spectra upon CO adsorption on the poisoned palladium nanoclusters; see Figure 5. For the sulfur precovered, CO-free particles, one sulfur species is found at 161.82 eV (S 2p3/2) that we assign to sulfur bonded in hollow or step sites. During CO adsorption, a second species occurs at 161.55 eV (S 2p3/2), shifting toward 161.45 eV for high exposures. This binding energy difference is explained by adsorbate−adsorbate interactions between sulfur and CO, which possibly induce a site change of part of the sulfur atoms. Interestingly, the quantitative analysis in Figure 6 shows that, independent of the sulfur amount, the distribution

Figure 5. (a) Selected S 2p spectra measured during adsorption of 2 × 10−9 mbar of CO on 0.53 ML of Pd/graphene/Rh(111) at 150 K, θS,rel = 0.14 ML. The positions of the fitted peaks of (b) are marked as lines. (b) Fit at CO saturation, θCO,rel = 0.42 ML.

of both sulfur species is always one to one at CO saturation. On the basis of this 1:1 ratio, we propose that due to co-adsorbed CO S is displaced toward a neighboring S atom, thereby forming a S pair at different adsorption sites. The driving force for this compression is the creation of addional adsorption sites for CO, and also explains the site change despite the fact that on the clean surface sulfur is known to be immobile at temperatures below 180 K.49 In the literature, differences of E


Article


Figure 6. Quantitative analysis of the distribution of S hollow and S−CO species for different amounts of preadsorbed sulfur during (a) exposure to 2 × 10−9 mbar of CO on 0.53 ML of Pd/graphene/Rh(111) at 150 and (b) during TPXPS, β = 0.5 K/s.

species increases and then stays constant until 350 K, before it decreases again, while the hollow species increases throughout the entire heating ramp. At 480 K, the sulfur species that is shifted due to the interactions with CO has not completely vanished, which is explained by temperature-induced broadening of the spectra. After cooling down of the sample, the original S 2p spectrum is observed. Therefore, the CO adsorption−desorption cycle is reversible. Finally, we want to draw some general conclusion from the results obtained from binding energy shifts in the C 1s and Pd 3d regions. At first, we shortly discuss the adsorption on the clean particles: Although one has to bear in mind that the results cannot be directly correlated with each other, there are similar phenomena in all three core levels observable. In both, the C 1s and the Pd 3d, the dominant Pd−CO species depends on the CO coverage. In the Pd 3d spectra, the Pd−CO α species dominates for low coverages, the Pd−CO β species for medium coverages, and the Pd−CO γ species for high exposures; the counterpart in the C 1s spectra are the step/ hollow COA species, which dominate for low and medium coverages. This fits together, as Pd−CO α is assigned to a Pd atom with one hollow-bonded CO, while Pd−CO β is a Pd atom with two hollow-bonded CO molecules or one bridgebonded CO. The increasing amount of COB in the C 1s for high coverages fits the increase of Pd−CO β and later Pd−CO γ with the latter being assigned to two bridge-bonded CO molecules or one on-top CO. Next, we describe the effects on the site occupation with increasing S coverage. While a simple decrease of all CO species is observed for low CO coverages, for high CO coverages also the ratio of the CO species depends on the sulfur coverage. On sulfur-free particles COA (step and hollow sites and COB (bridge and maybe top sites) have a 1:1 ratio, while with increasing amounts of sulfur, COB becomes the major species. In the Pd 3d region the dominant Pd−CO γ is more affected by S blocking than the Pd−CO β in absolute and relative terms. Both observations are explained with sulfur binding to hollow sites. The decrease of hollow-bonded CO is straightforward as these sites are directly blocked. Furthermore, one hollow-bonded sulfur atom also disables the three (on (111) facets) or four (on (100) facets) Pd atoms surrounding the hollow site to bind one on-top CO or two bridge CO molecules, while they are still capable of binding one bridge CO. As no change of the desorption temperatures of CO was found, the Pd−CO bond strength seems to not be influenced by co-adsorbed sulfur, which is in contrast to DFT calculations.36,51 This shows that sulfur blocks adsorption sites only physically for CO and no strong electronic effects are observed.

0.9−1.0 eV between hollow- and bridge-bonded sulfur on palladium50 have been reported for pure sulfur layers. The smaller difference of 0.4−0.5 eV seen here could be either due to lateral interaction or due to the fact that the two adsorption sites are very similar. For smaller amounts of sulfur, the conversion from the hollow species to the CO interaction species happens for lower exposures and thus CO coverages, as there is less CO needed to interact with all adsorbed sulfur atoms. Interestingly, the fraction of the sulfur in contact with CO is higher than the fraction of hollow sulfur for medium CO exposures with a maximum between 0.4 and 0.5 L and with a more distinct effect for smaller amounts of sulfur. Note that we observe a small damping of about 20% of the sulfur signal by the CO. To complete our investigations, we studied the thermal evolution of the adsorbed CO layers. When heating the sample, the behavior observed during adsorption is reversed. The quantitative analysis of the C 1s spectra from the TPXPS in Figure 2b (total CO coverage) and Figure 2d (individual species) shows that two processes occur in parallel between 200 and 300 K. The COB species decreases by desorption and we also find a rearrangement to step/hollow sites, which is the reversal of the observations during the adsorption. At 300 K, the COA species shows a maximum before desorbing at higher temperatures. The whole desorption process is finished at 475 K and is not influenced by the preadsorbed sulfur since all experiments show the same behavior at the same temperatures, independent of θS. In the Pd 3d region, shown in Figure 4b,d, first small changes start to occur already at 150 K, indicating that these spectra are somewhat more sensitive to small structural changes in the adsorbed layer. Pd−CO β increases at the cost of the Pd−CO γ until 230 K. While the latter keeps decreasing, the Pd−CO β signal is constant up to 280 K and the Pd−CO α forms again. Also, some free surface is found above 230 K. Between 180 and 280 K, the Pd−S peak shifts back to 335.44 eV. At 330 K, all Pd−CO γ is gone, while Pd− CO β is decreasing and the Pd−CO α is increasing. The latter has a maximum, whose temperature depends on the amount of co-adsorbed sulfur. For sulfur-free particles, it is found around 400 K, buts shifts to 375 K for 0.14 ML S. After these maxima, only desorption of CO occurs, which is completed at 480 K, again independent of sulfur precoverage. The bulk species and the free surface species gain back their initial intensities. When the quantitative evaluation of S 2p spectra in Figure 6 is analyzed, we also find that desorption of CO overall leads to an inversion of the effects observed during the adsorption. As for the adsorption, these effects are stronger for smaller amounts of sulfur. Between 200 and 250 K, the CO interaction F


Article


■

(6) Gotterbarm, K.; Zhao, W.; Höfert, O.; Gleichweit, C.; Papp, C.; Steinrück, H.-P. Growth and Oxidation of Graphene on Rh(111). Phys. Chem. Chem. Phys. 2013, 15, 19625−19631. (7) N’Diaye, A. T.; Gerber, T.; Busse, C.; Myslivecek, J.; Coraux, J.; Michely, T. A Versatile Fabrication Method for Cluster Superlattices. New J. Phys. 2009, 11, 103045. (8) Donner, K.; Jakob, P. Structural Properties and Site Specific Interactions of Pt with the graphene/Ru(0001) Moiré Overlayer. J. Chem. Phys. 2009, 131, 164701. (9) Pan, Y.; Gao, M.; Huang, L.; Liu, F.; Gao, H. J. Directed SelfAssembly of Monodispersed Platinum Nanoclusters on Graphene Moiré Template. Appl. Phys. Lett. 2009, 95, 093106. (10) Sicot, M.; Bouvron, S.; Zander, O.; Rüdiger, U.; Dedkov, Y. S.; Fonin, M. Nucleation and Growth of Nickel Nanoclusters on Graphene Moiré on Rh(111). Appl. Phys. Lett. 2010, 96, 093115. (11) Gotterbarm, K.; Steiner, C.; Bronnbauer, C.; Bauer, U.; Steinrück, H.-P.; Maier, S.; Papp, C. Graphene-Templated Growth of Pd Nanoclusters. J. Phys. Chem. C 2014, 118, 15934−15939. (12) Gotterbarm, K.; Bronnbauer, C.; Bauer, U.; Papp, C.; Steinrück, H.-P. Graphene-Supported Pd Nanoclusters Probed by Carbon Monoxide Adsorption. J. Phys. Chem. C 2014, 118, 25097−25103. (13) Gotterbarm, K.; Späth, F.; Bauer, U.; Bronnbauer, C.; Steinrück, H.-P.; Papp, C. Reactivity of Graphene-Supported Pt Nanocluster Arrays. ACS Catal. 2015, 5, 2397−2403. (14) Gotterbarm, K.; Späth, F.; Bauer, U.; Steinrück, H.-P.; Papp, C. Adsorption and Reaction of SO2 on Graphene-Supported Pt Nanoclusters. Top. Catal. 2015, 58, 573−579. (15) Surnev, S.; Sock, M.; Ramsey, M. G.; Netzer, F. P.; Wiklund, M.; Borg, M.; Andersen, J. N. CO Adsorption on Pd(111): A HighResolution Core Level Photoemission and Electron Energy Loss Spectroscopy Study. Surf. Sci. 2000, 470, 171−185. (16) Ramsey, M. G.; Leisenberger, F. P.; Netzer, F. P.; Roberts, A. J.; Raval, R. Surf. Sci. 1997, 385, 207−215. (17) Kolthoff, D.; Jürgens, D.; Schwennicke, C.; Pfnür, H. The Local Geometry of Chalcogen Atoms on Pd(100): The Low Coverage Phases of O and S. Surf. Sci. 1996, 365, 374−382. (18) Orita, H.; Itoh, N.; Inada, Y. A Comparison of CO Adsorption on Pt(211), Ni(211), and Pd(211) Surfaces Using Density Functional Theory. Surf. Sci. 2004, 571, 161−172. (19) Conrad, H.; Ertl, G.; Koch, J.; Latta, E. E. Adsorption of CO on Pd Single Crystal Surfaces. Surf. Sci. 1974, 43, 462−480. (20) Svensson, K.; Rickardsson, I.; Nyberg, C.; Andersson, S. Step Site Adsorption and Ordering of CO on Ni(510) and Pd(510). Surf. Sci. 1996, 366, 140−148. (21) Ramsier, R. D.; Lee, K. W.; Yates, J. T. CO Adsorption on Stepped Pd (112): Studies by Thermal and Electron Stimulated Desorption. Surf. Sci. 1995, 322, 243−255. (22) Forzatti, P.; Lietti, L. Catalyst Deactivation. Catal. Today 1999, 52, 165−181. (23) Moulijn, J. A.; van Diepen, A. E.; Kapteijn, F. Catalyst Deactivation: Is It Predictable?: What to Do? Appl. Catal., A 2001, 212, 3−16. (24) Bartholomew, C. H. Mechanisms of Catalyst Deactivation. Appl. Catal., A 2001, 212, 17−60. (25) Albers, P.; Pietsch, J.; Parker, S. F. Poisoning and Deactivation of Palladium Catalysts. J. Mol. Catal. A: Chem. 2001, 173, 275−286. (26) Yu, T.-C.; Shaw, H. The Effect of Sulfur Poisoning on Methane Oxidation over Palladium Supported on γ-Alumina Catalysts. Appl. Catal., B 1998, 18, 105−114. (27) Meeyoo, V.; Lee, J. H.; Trimm, D. L.; Cant, N. W. Hydrogen Sulphide Emission Control by Combined Adsorption and Catalytic Combustion. Catal. Today 1998, 44, 67−72. (28) Mowery, D. L.; Graboski, M. S.; Ohno, T. R.; McCormick, R. L. Deactivation of PdO−Al2O3 Oxidation Catalyst in Lean-Burn Natural Gas Engine Exhaust: Aged Catalyst Characterization and Studies of Poisoning by H2O and SO2. Appl. Catal., B 1999, 21, 157−169. (29) Beck, D. D.; Sommers, J. W. Impact of Sulfur on Three-Way Catalysts: Comparison of Commercially Produced Pd and Pt-Rh Monoliths. Stud. Surf. Sci. Catal. 1995, 96, 721−748.

SUMMARY We investigated the influence of sulfur on the adsorption and desorption behavior of CO on graphene-supported palladium nanoparticles by HR-XPS. Sulfur adsorbs at highly coordinated sites and blocks these sites for CO. Other adsorption sites are blocked to a minor degree. As the desorption temperatures of CO are independent of the sulfur amount, the Pd−CO bond strength seems to not be effected by sulfur. Thus, blocking of adsorption sites on palladium by sulfur happens via steric hindrance, while electronic effects seem not to play an observable role. The degree of site blocking is between that of Pd(111) and Pd(100) surfaces, which indicates the nanocluster surface consists of a mixture of (111) and (100) facets. Interestingly, co-adsorbed CO has an influence on the preadsorbed sulfur; that is, we find indication of a displacement of S atoms such that S pairs are formed on the surface, creating more space to adsorb CO on the nanoparticles.

■

ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.6b11480. The used fitting parameters of the C 1s, Pd 3d5/2, and S 2p core levels, and exemplary fits of the Pd 3d5/2 as well as Pd 3d5/2 spectra before CO adsorption (PDF).

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Hans-Peter Steinrück: 0000-0003-1347-8962 Christian Papp: 0000-0002-1733-4387 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank H.Z.B. for the allocation of synchrotron radiation beamtime and BESSY II staff for support during beamtime. This work was supported by SFB 953 “Synthetic Carbon Allotropes” and the Cluster of Excellence “Engineering of Advanced Materials” in collaboration with Clariant Produkte (Deutschland) GmbH.

■

REFERENCES

(1) Blanita, G.; Lazar, M. D. Review of Graphene-Supported Metal Nanoparticles as New and Efficient Heterogeneous Catalysts. Micro Nanosyst. 2013, 5, 138−146. (2) Singh, R. N.; Awasthi, R. Graphene Support for Enhanced Electrocatalytic Activity of Pd for Alcohol Oxidation. Catal. Sci. Technol. 2011, 1, 778−783. (3) Siamaki, A. R.; Khder, A. E. R. S.; Abdelsayed, V.; El-Shall, M. S.; Gupton, B. F. Microwave-Assisted Synthesis of Palladium Nanoparticles Supported on Graphene: A Highly Active and Recyclable Catalyst for Carbon-Carbon Cross-Coupling Reactions. J. Catal. 2011, 279, 1−11. (4) Li, Y.; Fan, X.; Qi, J.; Ji, J.; Wang, S.; Zhang, G.; Zhang, F. Palladium Nanoparticle-Graphene Hybrids as Active Catalysts for the Suzuki Reaction. Nano Res. 2010, 3, 429−437. (5) Johnson, J. L.; Behnam, A.; Pearton, S. J.; Ural, A. Hydrogen Sensing Using Pd-Functionalized Multi-Layer Graphene Nanoribbon Networks. Adv. Mater. 2010, 22, 4877−4880. G


Article

The Journal of Physical Chemistry C (30) Lampert, J. K.; Kazi, M. S.; Farrauto, R. J. Palladium Catalyst Performance for Methane Emissions Abatement from Lean Burn Natural Gas Vehicles. Appl. Catal., B 1997, 14, 211−223. (31) Rodríguez, J. C.; Santamaría, J.; Monzón, A. Hydrogenation of 1,3-Butadiene on Pd/SiO2 in the Presence of H2S Deactivation and Reactivation of the Catalyst. Appl. Catal., A 1997, 165, 147−157. (32) Matsumoto, Y.; Onishi, T.; Tamaru, K. Effects of Sulphur on a Palladium Surface on the Adsorption of Carbon Monoxide and the Adsorption and Decomposition of Nitric Oxide. J. Chem. Soc., Faraday Trans. 1 1980, 76, 1116−1121. (33) Yamada, T.; Runsheng, Z.; Iwasawa, Y.; Tamaru, K. AdsorptionDesorption Kinetics of Carbon Monoxide on a Palladium (100) Single Crystal Surface Modified with Sulfur. Surf. Sci. 1988, 205, 82−99. (34) Jorgensen, S. W.; Madix, R. J. Steric and Electronic Effects of Sulfur on CO Adsorbed on Pd (100). Surf. Sci. 1985, 163, 19−38. (35) Berndt, W.; Hora, R.; Scheffler, M. A Leed Analysis of c(2 x 2)S on Pd(100). Surf. Sci. 1982, 117, 188−195. (36) Alfonso, D. R. Influence of Sulfur Poisoning on CO Adsorption on Pd(100). Top. Catal. 2012, 55, 267−279. (37) Papp, C.; Steinrück, H. P. In Situ High-Resolution X-Ray Photoelectron Spectroscopy - Fundamental Insights in Surface Reactions. Surf. Sci. Rep. 2013, 68, 446−487. (38) Denecke, R.; Kinne, M.; Whelan, C. M.; Steinrück, H.-P. In-Situ Core-Level Photoelectron Spectroscopy of Adsorbates on Surfaces Involving a Molecular Beam-General Setup and First Experiments. Surf. Rev. Lett. 2002, 9, 797−801. (39) Kinne, M.; Fuhrmann, T.; Whelan, C. M.; Zhu, J. F.; Pantförder, J.; Probst, M.; Held, G.; Denecke, R.; Steinrück, H. P. Kinetic Parameters of CO Adsorbed on Pt(111) Studied by in Situ High Resolution X-Ray Photoelectron Spectroscopy. J. Chem. Phys. 2002, 117, 10852−10859. (40) Doniach, S.; Sunjic, M. Many-Electron Singularity in X-Ray Photoemission and X-Ray Line Spectra from Metals. J. Phys. C: Solid State Phys. 1970, 3, 285−291. (41) Guo, X.; Yates, J. T. Dependence of Effective Desorption Kinetic Parameters on Surface Coverage and Adsorption Temperature: CO on Pd(111). J. Chem. Phys. 1989, 90, 6761−6766. (42) Uvdal, P.; Karlsson, P. A.; Nyberg, C.; Andersson, S.; Richardson, N. V. On the Structure of Dense CO Overlayers. Surf. Sci. 1988, 202, 167−182. (43) Maca, F.; Scheffler, M.; Berndt, W. The Adsorption of Sulphur on Pd(111) I. A LEED Analysis of the Adsorbate Structure. Surf. Sci. 1985, 160, 467−474. (44) Thiel, P. A.; Williams, E. D.; Yates, J. T.; Weinberg, W. H. The Chemisorption of CO on Rh(111). Surf. Sci. 1979, 84, 54−64. (45) Preobrajenski, A. B.; Ng, M. L.; Vinogradov, A. S.; Mårtensson, N. Controlling Graphene Corrugation on Lattice-Mismatched Substrates. Phys. Rev. B: Condens. Matter Mater. Phys. 2008, 78, 073401. (46) Andersen, J.; Qvarford, M.; Nyholm, R.; Sorensen, S.; Wigren, C. Surface Core-Level Shifts as a Probe of the Local Overlayer Structure: CO on Pd(100). Phys. Rev. Lett. 1991, 67, 2822−2825. (47) Grillo, M. E.; Stampfl, C.; Berndt, W. Low-Energy ElectronDiffraction Analysis of the (√7 × √7)R19.1 -S Adsorbate Structure on the Pd(111) Surface. Surf. Sci. 1994, 317, 84−98. (48) Speller, S.; Rauch, T.; Bömermann, J.; Borrmann, P.; Heiland, W. Surface Structures of S on Pd(111). Surf. Sci. 1999, 441, 107−116. (49) Streber, R.; Papp, C.; Lorenz, M. P. A.; Bayer, A.; Wickert, S.; Schöppke, M.; Denecke, R.; Steinrück, H.-P. Site Blocking and CO/ sulfur Site Exchange Processes on Stepped Pt Surfaces. J. Phys.: Condens. Matter 2009, 21, 134018. (50) Rodriguez, J. A.; Chaturvedi, S.; Jirsak, T. The Bonding of Sulfur to Pd Surfaces: Photoemission and Molecular−orbital Studies. Chem. Phys. Lett. 1998, 296, 421−428. (51) Alfonso, D. R. First-Principles Study of Sulfur Overlayers on Pd(111) Surface. Surf. Sci. 2005, 596, 229−241.

H


Reactivity of CO on Sulfur-Passivated Graphene-Supported Palladium

Recommend Documents