Reduced Oxide Sites and Surface Corrugation Affecting the Reactivity

Apr 22, 2013 - Reduced Oxide Sites and Surface Corrugation Affecting the Reactivity, ... clusters, where Au atoms are depleted from the top surface up...
1 downloads 0 Views 3MB Size
Download PDF
Article pubs.acs.org/Langmuir

Reduced Oxide Sites and Surface Corrugation Affecting the Reactivity, Thermal Stability, and Selectivity of Supported Au−Pd Bimetallic Clusters on SiO2/Si(100) Elad Gross,† Elishama Sorek,‡ Arumugam Murugadoss,‡ and Micha Asscher*,‡ †

Department of Chemistry, University of CaliforniaBerkeley, Berkeley, California 94720, United States Institute of Chemistry, The Hebrew University of Jerusalem, Jerusalem 91904, Israel

‡

ABSTRACT: The morphology and surface elemental composition of Au−Pd bimetallic nanoclusters are reported to be sensitive to and affected by reduced silicon defect sites and structural corrugation on SiO2/Si(100), generated by argon ion sputtering under ultrahigh vacuum (UHV) conditions. Metastable structures of the bimetallic clusters, where Au atoms are depleted from the top surface upon annealing, are stabilized by the interaction with the reduced silica sites, as indicated from CO temperature programmed desorption (TPD) titration measurements. Acetylene conversion to ethylene and benzene has been studied as a probe reaction, revealing the modification of selectivity and reactivity enhancement in addition to improved thermal stability on substrates rich in reduced-silica sites. These observations suggest that these unique sites play an important role in anchoring thermodynamically metastable conformations of supported Au−Pd bimetallic catalysts and dictate their high-temperature activity.

1. INTRODUCTION Engineering the metallic cluster−support interaction as a tool for tuning catalyst reactivity and selectivity is both a current and future goal of great importance. Metal nanoclusters supported on thin oxide films are widely used as a model for industrial catalysts. It was demonstrated that for certain structuresensitive and cluster-size-dependent reactions such as CO oxidation over Au/TiO2(110), maximum catalytic reactivity has been observed for gold clusters with a diameter of 3 nm.1 This phenomenon has been correlated with electronic structure modifications and the density of low-coordination surface atoms.1−5 Upon exposure to reactants at high pressures or annealing to elevated temperature, the metallic clusters tend to diffuse and aggregate on smooth supports, often resulting in a significant decrease in their catalytic reactivity.1,6−8 In contrast, on disordered, corrugated, and defect-rich oxide substrates, the metallic clusters are often partially negatively charged and typically more thermally stable.9−12 Charging of high electron affinity metallic clusters (such as Au) has also been suggested to occur in certain cases on very thin and smooth oxide films.13 Extended single crystals and gold particles decorated with palladium atoms have shown enhanced catalytic activity toward acetylene conversion to benzene compared to pure predominantly (111)-oriented palladium clusters when calibrated per palladium atom. This was demonstrated for a variety of reactions, in particular, for the case of acetylene hydrogenation14,15 and acetylene trimerization to benzene.16−19 This reaction is considered to be structure-sensitive, with the Pd(111) facet the most active among extended palladium surfaces. A critical ensemble size of Pd atoms has been © 2013 American Chemical Society

postulated as necessary for efficient benzene formation on small-cluster bimetallic catalysts.20−22 Electronic modifications of supported metal clusters have often been assigned to charge transfer from substrateunsaturated defect sites to the metallic clusters, a process that may enhance the reactivity of adsorbed molecules.3,22 We have previously reported on the morphology and density of clusters deposited on a silica surface via the buffer layer assisted growth (BLAG) method.23 In addition, the chemistry of monometallic and bimetallic Pd and Au nanoclusters with respect to acetylene hydrogenation to ethylene and its trimerization to benzene has been described.24,25 It was demonstrated that Pd−Au alloy clusters are more active than the pure Pd or Au clusters and are highly selective, promoting low-temperature acetylene hydrogenation to ethylene at a significantly higher rate relative to the trimerization reaction to form benzene. Bimetallic nanoclusters have traditionally been grown via a liquid-phase reduction−precipitation procedure on top of a high-surface-area dispersed oxide support, followed by hightemperature calcinations as an important element of the catalyst preparation. In the case of Au−Pd bimetallic clusters, extensive work by Hutchings and co-workers has shown the unique reactivity of such clusters toward hydrogen peroxide formation6 as well as alcohol oxidation catalytic reactions.26 The hydrogenation of alkenes, as studied by Louis et al., has been demonstrated as an efficient and selective catalytic reaction on highly dispersed, supported Au−Pd clusters deposited by a Received: January 2, 2013 Revised: March 28, 2013 Published: April 22, 2013 6025

dx.doi.org/10.1021/la400005d | Langmuir 2013, 29, 6025−6031

Langmuir

Article

3.1. Au−Pd Cluster Preparation and Characterization. Au−Pd BLAG clusters can be prepared by the following three procedures: (Type A) Two consecutive BLAG cycles are employed where Au atoms are evaporated first toward the ASW buffer layer and then deposited on the substrate. In the second cycle, Pd atoms were evaporated on top of a new ASW buffer layer. Between these two cycles, the sample was annealed to 170 K in order to ensure water layer removal. Both in-situ (AES and CO-TPD) and ex-situ (TEM, SEM, and XPS) characterization measurements of the particles have indicated that mostly separated Au and Pd particles were generated this way without any indication of alloy formation.24 (Type B) The second method involves Au and Pd consecutive evaporation on the same buffer layer, followed by annealing to 170 K. This preparation method has led to partial alloying of Au and Pd atoms upon annealing to 300 K and exposure to air, as indicated by TEM and XRD measurements.23−25 (Type C) The formation of Au−Pd alloy nanoclusters was accomplished by the third preparation method in which the two metals were simultaneously evaporated toward the ASW buffer layer. CO-TPD measurements have indicated not only that the cluster composition is Au−Pd alloy but also that the cluster surface is composed of the Au−Pd alloy. Type C cluster morphology and elemental analysis (determined by EDS) are demonstrated in Figure 1. The cluster morphology and density

codeposition−precipitation method on top of SiO2, TiO2, and Al2O3.27,28 This approach has been performed for decades as a realistic, industrially relevant way to study simple heterogeneous catalytic reactions. This methodology, however, suffers from the inherent difficulty of obtaining precise atomic-level understanding and control of the metal−support nature of the interaction. Model studies over planar, extended, and well-characterized support surfaces in an ultra-high-vacuum environment has demonstrated the capability to provide the necessary tool box to address some of the fundamental questions regarding the nature of metal and bimetallic−support interactions at the atomic and molecular levels.29−33 In this work, bimetallic Pd−Au clusters deposited as vapor under UHV conditions over condensed water as a buffer layer to be supported on corrugated, defect-rich planar silica substrates have been studied as a platform to explore the concept of sintering resistant model catalysts. These defects were analyzed by XPS to contain reduced silica sites. We found a strong stabilizing effect of these unique sites, keeping the reactivity of the bimetallic alloy nanoclusters above 500 K. In the absence of such sites, reactivity terminates below 250 K. Changes in selectivity toward ethylene formation is somewhat enhanced by these defect sites, attributed to the stabilization of elemental composition with depleted gold atoms at the surface of the bimetallic clusters.

2. EXPERIMENTAL SECTION The main aspects of the experimental UHV environment and overall setup were described elsewhere in detail.23,24 The buffer layer assisted growth (BLAG) method to deposit metal clusters has been employed with amorphous solid water as the buffer. Au and Pd clusters were grown on top of the native SiO2/Si(100) oxide substrate. These metallic atoms were evaporated from a resistively heated tungsten filament wrapped around Au or Pd (99.99% pure) wires, employing an in situ quartz microbalance (QMB) to determine the deposition rates (1 ± 0.1 Å/min). Upon evaporation of the ASW layer near 170 K, metallic clusters softly land on the substrate, depositing rough, nonannealed clusters on the support. Temperature-programmed reaction (TPR) measurements were performed following the adsorption of 3L (1 L = 10−6 Torr·s) isotopically labeled 13C2H2 at 120 K using a quadruple mass spectrometer (QMS) in order to follow the depletion of parent acetylene molecules while simultaneously monitoring the formation of the two main reaction products, ethylene and benzene. CO-TPD spectra were acquired following the exposure of the sample at 120 K to a saturation value of 30 L CO. Lowtemperature CO-TPD measurements were performed in another chamber capable of cooling the sample to 40 K. The clusters’ structure and elemental composition over a standard amorphous carbon (a-C) sample holder were determined using (ex-situ, room-temperature) HR-TEM measurements (Tecnai F20 G2, nominal line resolution 0.1 nm) with elemental analysis capability via energy-dispersive spectroscopy (EDS) measurements. XPS spectra were acquired with an AxisUltra (Kratos) instrument equipped with an Ar+ gun for sputter cleaning and depth profile analysis. The Auger spectrometer within our UHV system (LK Technologies) enabled in-vacuum, in-situ surface elemental characterization of the metallic nanoclusters before and following annealing or reactivity tests.

Figure 1. (a) Scanning TEM image of a single Pd−Au alloy cluster grown via the BLAG method after coevaporation of 2 Å Au and Pd. (b) EDS elemental analysis of the gold and palladium distribution across the red line marked along a type C alloy nanocluster.

are directly correlated to the metal dosage.25a At an accumulated metal deposition of more than 1 Å of gold and palladium, complex ramified clusters are formed (Figure 1), and below this amount, clusters tend to be mostly rounded and hemispherical.24,25b Here we have studied the correlation between the surface composition of nanoclusters and their chemical reactivity and selectivity while modifying the oxide support properties. We have previously demonstrated that Au− Pd alloy clusters effectively catalyze acetylene trimerization to form benzene. When compared to separate Au and Pd particles and the physical mixture of Au and Pd particles, the alloyed clustes enhanced benzene and ethylene formation by at least 5fold (calibrated per Pd surface atoms). CO-TPD measurements (a sensitive top-surface titration method) have indicated that the active Au−Pd alloy clusters formed by the BLAG method have a typical surface composition of 1:1 gold to palladium. In contrast, the surface composition of clusters that are prepared by direct evaporation of the metals on the SiO2 substrate (without a buffer layer) typically consisted of Pd and Au ensembles, displaying a lower reactivity. Pure gold clusters were found to be totally inert with respect to acetylene decomposition and overall reactivity to form ethylene and benzene.

3. RESULTS AND DISCUSSION Metallic nanoclusters that are deposited via the ASW-BLAG method, originally demonstrated for rare gas atoms as a buffer material,34 result in nonannealed, structurally metastable clusters as long as the substrate is kept at low temperature (here 120 K). 6026

dx.doi.org/10.1021/la400005d | Langmuir 2013, 29, 6025−6031

Langmuir

Article

sintering-resistant properties of our model catalyst, we have modified the nature and density of surface imperfections of both a-C and SiO2/Si(100) substrates. The surface defects were generated by Ar+ ions, striking the entire surface at a range of kinetic energies up to 3.0 keV prior to metallic cluster deposition. Annealing the BLAG clusters, prepared on top of the sputtered, defect-rich a-C substrate, from 300 to 600K has led to slower and smaller changes in the cluster density and morphology (Figure 2c,d, respectively). Similarly, the fraction of the substrate covered by metallic clusters has dropped only slightly as a result of the annealing process, from 37 to 31%, corresponding to a density change in the clusters from 10.6 × 1011 to 9.8 × 1011 cm−2 upon annealing. Although not specifically shown, we assume structurally similar behavior on silica. See further discussion in section 3.5. The importance of sputter-induced defect sites formed on the native SiO2/Si(100) substrate has to be identified with respect to its electronic properties in order to better understand the nature of the modifications and restructuring that subsequently occurs in the supported Au−Pd alloy clusters. We have sputtered a native SiO2/Si(100) sample and in situ characterized it by XPS in a different dedicated UHV chamber. When we struck the silicon oxide surface (Ar+ current of 750 ± 250 nA for 180 s) by increasing the kinetic energy of the argon ions from 1.4 to 3.0 keV, the Si 2p XPS signal intensity, assigned to SiO2 (103.2 eV), decreased and its position has slightly shifted to lower energies. The shift in the silica peak is accompanied by increases in the Si 2p3/2 (99.2 eV) and Si 2p1/2 (99.8 eV) signals (Figure 3). Standard XPS analysis of these spectra using the relevant Gaussian silicon peaks at their respective contributions (amplitude) indicates that high-kineticenergy Ar+ ions have led to the development of a partially reduced SiOx signal that is dominated by Si+ and Si2+ oxidation states (red curve in Figure 3).36 A similar effect was obtained when the sputtering has been extended at lower ion energies for

The properties of these metallic particles have enabled us to investigate the reactivity of the BLAG Au−Pd alloy clusters toward acetylene hydrogenation and trimerization as a case study for better understanding the support effects on the elemental composition of the clusters and their thermal stability and chemical reactivity, as will be discussed below. As was mentioned above, Pd−Au alloy clusters were prepared in this study by the simultaneous evaporation of 2 Å Pd and 2 Å Au on top of a 30 ML ASW buffer layer adsorbed on the SiO2/Si(100) substrate at 120 K and on amorphous carbon (a-C) substrates. Clusters grown on a standard a-C TEM sample holder were annealed to 300 and 600 K in the UHV chamber and subsequently ex-situ imaged by HR-TEM (Figure 2a,b, respectively). Following substrate annealing to

Figure 2. HR-TEM images of Au−Pd BLAG-grown bimetallic clusters at 300 K (a) and after annealing to 600 K (b). The clusters were prepared on a-C substrates. The same preparation as in images a and b but over sputtered a-C substrates: 300 (c) and 600 K (d) annealing temperatures. The HR-TEM images were taken ex situ at 300 K.

600 K, the clusters aggregated, resulting in a 20% decrease in their density from 6.2 × 1011 to 5.0 × 1011 cm−2. The effective metal-covered area has also dropped from 25 to 16%, and the branched elongated clusters transformed into round, hemispherical ones. It is important to mention that the smooth a-C and SiO2 substrates were previously compared with respect to their interaction with metallic nanoclusters such as Au, Cu, Pd, Ag, and Pt. The metallic clusters seem to respond in a similar way with respect to the saturation of their density at elevated substrate temperature on both substrates.35 Because all our reactivity measurements were performed over the SiO2 substrate, we assume that the detailed morphological studies based on TEM over a-C are relevant to the silica substrate as far as clusters’ structural aspects are concerned. This similarity has been verified in a previous study in our group by means of AFM measurements on the SiO2 substrate, as discussed in ref 24. 3.2. Characterization of the SiO2 Substrate. To investigate the thermal stability and consequently the

Figure 3. XPS spectra of Si-2p in SiO2 (blue) and reduced SiOx (red) signals following Ar+ sputtering at the indicated beam kinetic energy. Inset: SiO2 (blue) and SiOx (red) XPS peak areas as a function of the argon ions’ kinetic energy. 6027

dx.doi.org/10.1021/la400005d | Langmuir 2013, 29, 6025−6031

Langmuir

Article

In our measurements, we could not detect CO desorption from extended gold surfaces because our minimum surface temperature was 120 K. An analysis of data presented in Figure 4 suggests that Au−Pd alloy clusters on the sputtered, corrugated with reduced silica sites are more thermally stable and their surface elemental composition is less affected by the annealing process than on the smooth substrate. CO desorption from single Au and Pd atoms (peak desorption temperatures of 180 and 280K, respectively) could not be detected from clusters supported on the smooth, defect-free SiO2 after annealing to 300K (Figure 4c). In contrast, CO desorption from these sites became observable when the clusters were supported on the defect-rich, corrugated SiO2 surfaces even after annealing to 600 K. In particular, desorption from the low-temperature range of 150−280 K remains detectable (Figure 4d). Selectively integrating the area under each CO-TPD spectrum that corresponds to desorption from Pd atop sites and single Au atoms (120K to 320K temperature range) provides information on Pd−Au alloy effective coverage, namely, the fraction of cluster surface sites that are composed of a 1:1 gold to palladium elemental ratio, considered to be a perfect stoichiometric alloy (Figure 4b). These data demonstrate that when annealed to 200 K and above, the surface area of Pd−Au alloy clusters supported on the defect-rich SiO2 contains about four times more alloy-character ensemble composition than clusters supported on the smooth SiO2 surface. In addition, CO-TPD uptake measurements from alloy clusters deposited on defect-rich substrates have doubled compared to that obtained from the smooth surface. Moreover, according to the TPD spectra, the Pd atoms seem to have segregated more to the clusters surface in the case of the defectrich substrate. This is indicated from the significantly increased CO-TPD signal near 250 K associated with CO adsorbed on atop Pd atoms, relative to the peak at 170 K that reflects CO adsorbed on predominantly single gold atoms. We may conclude, therefore, that not only the clusters’ thermal stability but also their surface elemental composition has been affected by modifying the substrate properties. To probe the clusters’ surface further under different conditions, the changes in their elemental composition were characterized in situ by AES measurements (Figure 5). Pd/Si and Pd/Au AES signal ratios obtained from clusters prepared on the smooth SiO2 surface (black line in Figure 5a,b), following annealing to 600 K, have decreased to 2 (Figure 5a) and 20% (Figure 5b) of their initial values, respectively. These results support the CO-TPD data that clarify the presence of the two simultaneous phenomena that take place upon

a longer period of time (not shown). The amplitude of the SiOx signal gradually increases with beam energy (Figure 3, inset), and the main drop in the SiO2 XPS signal starts near 2.5 keV. 3.3. Support Effects on Clusters’ Surface Elemental Composition. A central question arises as to whether these sputter-induced reduced silica sites have affected and modified the metallic clusters’ surface elemental composition and structure, in addition to thermal stabilization. To address this issue, we have performed CO-TPD and AES measurements. Unlike other characterization methods, CO-TPD is sensitive only to the surface structure and its chemical composition; therefore, it actually titrates the presence of surface atoms. CO molecules desorb from the clean, smooth, sputtered SiO2 substrates below 110 K (not shown, see ref 37); therefore, all changes in the CO-TPD spectra above this temperature are necessarily due to modifications in the properties of alloy clusters. One should note, however, that this statement is correct only under the assumption that the low-coverage, lowtemperature CO adsorption−desorption processes do not affect or modify the clusters’ probed surfaces. On the basis of literature data, this assumption is considered to be reasonable. The TPD data suggest that two phenomena simultaneously take place upon annealing the smooth-SiO2-supported Au−Pd clusters: (a) aggregation, resulting in a decrease in the total exposed metallic surface atoms (more 3D clusters); (b) structural modifications within the alloy clusters, due to preferential gold atoms segregating toward the cluster surface (Figure 4c,d), which is well correlated with the lower surface free energy of gold (0.64 eV/atom for gold38 and 0.86 eV/atom for palladium39). It should be noted that CO-TPD from extended gold surfaces (centered around 100 K) is detectable provided the surface temperature can be cooled to 80 K.

Figure 4. (a) CO-TPD following the exposure of 30 L of CO at 120 K over Pd−Au clusters prepared on smooth and sputtered SiO2 substrates at the indicated argon ion beam energies. (b) Normalized, exposed bimetallic surface alloy sites deduced from relative CO-TPD uptake measurements obtained from alloy clusters supported on the (c) smooth (native oxide, black) and (d) sputtered, defect-rich, reduced silicon substrates (in red), plotted as a function of preannealing temperature, in the range of 150−600 K. (c) CO-TPD uptake following alloy clusters preanneal on smooth, native SiO2/ Si(100) substrates. (d) CO-TPD uptake following clusters preanneal on sputtered (5 min at 2.5 KeV) SiO2/Si(100) substrates.

Figure 5. (a) Pd/Si AES peak area ratio vs annealing temperature. Pd−Au alloy clusters were prepared on the smooth, native (black), and sputtered (red) SiO2 substrates by the simultaneous evaporation of 2 Å Pd and 2 Å Au on 30 ML H2O with subsequent annealing to different temperatures. (b) Pd/Au AES signal area ratios, plotted as function of the annealing temperature as in plot a. 6028

dx.doi.org/10.1021/la400005d | Langmuir 2013, 29, 6025−6031

Langmuir

Article

annealing. The first and dominant change is the aggregation of Pd−Au clusters following substrate annealing, as clearly demonstrated by the significant decrease in the Pd/Si AES signal ratio, namely, larger and more 3D clusters gradually form upon annealing. (See Figure 2b, revealing the behavior of the clusters on the a-C TEM substrate.) The second change is in the clusters’ surface composition, where the annealing process on the smooth silica substrate results in the segregation and diffusion of Au atoms to the cluster’s surface, resulting in a smaller Pd/Au AES ratio. In contrast to the smooth silica, the preparation of the Au−Pd alloy clusters on the sputtered substrate significantly affected and modified the clusters’ elemental composition as seen in Figure 5a,b (red lines). It has led to slower changes in both ratios of the Pd/Si and Pd/Au AES signals as a function of the annealing temperature. In this case, the Pd/Si and Pd/Au AES signal decreased to 50 and 75% of their initial values even as the sample was annealed up to 600 K, retaining a significant fraction of the alloy surface composition. Clusters prepared at 120 K on the smooth SiO2 retained a surface composition of 1:1 Pd to Au. In contrast, those clusters that have been deposited on the reduced silica surfaces at 120 K were stabilized at a Pd-rich surface composition that arises from the surface depletion of gold, leading to a Pd/Au ratio of 2:1 on the 2.5 keV sputtered surface (Figure 5a, red curve). This behavior opposes the thermodynamically more favorable composition, having mostly the lower-surface-energy atoms (gold) on the surface. 3.4. Silica Substrates’ Effect on the Reactivity of the Au−Pd Alloy Clusters. A sensitive probe of the clusters’ thermal stability and structural integrity is its chemical reactivity. Ethylene and benzene formation reactions have been studied using mass spectrometry following the adsorption of 3 L 13C2H2 on cold (120 K) BLAG Pd−Au alloy clusters prepared on native and sputtered SiO2/Si(100) substrates. The basic low-temperature (150−250 K) acetylene conversion mechanism requires the initial dissociation of a fraction of the adsorbed parent acetylene molecules. This leaves hydrogen adatoms coadsorbed with acetylene on the surface, followed by the hydrogenation reaction of neighboring acetylene molecules to form ethylene (eqs 1 and 2) and benzene (eq 3): 13

130 − 150 K

Figure 6. (a) Benzene formation yield following the adsorption of 3 L of 13C2H2 at 120 K on preannealed bimetallic Pd−Au clusters at the indicated substrate temperatures. The clusters were prepared on smooth native SiO2/Si(100) (black) and on sputtered substrates (red). (b) Benzene and ethylene relative yield as a function of argon ions kinetic energy at 120 K without preannealing.

deposited and grown on sputtered, defect-rich, reduced silica substrates were significantly more thermally stable. Although following annealing to 160 K the reactivity has dropped by 50%, a sizable level of reactivity was maintained even after annealing to 700 K (Figure 6a, red line). In previous reports, we have shown that a specific ratio of Pd/Au within a cluster optimizes the acetylene trimerization. It was demonstrated that the maximum reactivity toward benzene formation was achieved at a Pd/Au ratio of 1:1, obtained in crystalline bimetallic alloy clusters. The decrease in this ratio due to partial gold surface segregation resulted in a coupled decrease in the benzene formation rate. A similar but slower decrease in the ethylene formation rate was also measured. This has led to a small increase in overall selectivity toward ethylene formation upon annealing. The reason for the different response to annealing of the acetylene trimerization and hydrogenation reactions is their different sensitivities to the presence of Pd atoms. Although a minimum ensemble of Pd atoms is needed for benzene formation, even a single Pd atom surrounded by gold can catalyze the acetylene decomposition followed by the hydrogenation of neighboring acetylene to ethylene. Consequently, a decrease in the number of Pd surface atoms had a stronger effect on slowing down benzene formation rates than on ethylene formation. This was nicely correlated to the AES signal that revealed carbon peak growth (not shown) following the formation of ethylene even in cases where no benzene was formed at all. The carbon formation is due to the further decomposition of acetylene fragments, which is the source of hydrogen for ethylene formation (eq 1). The effect of surface carbon accumulation is therefore probably both in overall deactivation and enhancing the preference and selectivity toward ethylene formation. A more quantitative account of

heat

C2H 2(ad) ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ 13C2H(ad) + H(ad) ⎯⎯⎯→ 213C(ad) + 2H(ad)

(1) 150 − 250 K

heat

2H(ad) + 13C2H 2(ad) ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ 13C2H4(ad) ⎯⎯⎯→ 13C2H4(gas) 150 − 250 K

heat

313C2H 2(ad) ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ 13C6H6(ad) ⎯⎯⎯→ 13C6H6(gas)

(2) (3)

The competing reaction that has to take place in the same temperature range is the combination of three acetylene monomers, the trimerization reaction, to form benzene (eq 3). Clearly, it is a less probable reaction because an ensemble of several Pd atoms is necessary to catalyze it. Nevertheless, we have shown in our previous studies that Au−Pd alloy clusters have enhanced the selectivity of ethylene formation over benzene.24,25a Annealing the clusters prepared on the smooth, native silicon oxide to 160 K, prior to acetylene adsorption, has already led to a decrease in the benzene formation yield, with complete quenching upon annealing to 600 K (Figure 6a, black line). The decrease in reactivity has been correlated with the reduced active surface area and with the gold surface segregation and therefore with diminishing active surface sites on the clusters upon annealing, as discussed above. In contrast, clusters 6029

dx.doi.org/10.1021/la400005d | Langmuir 2013, 29, 6025−6031

Langmuir

Article

4. CONCLUSIONS We have demonstrated that the thermal stability and chemical reactivity of Pd−Au alloy nanoclusters prepared via a lowtemperature buffer-layer-assisted growth procedure are enhanced and their chemical selectivity modified by the presence of structural corrugation that is rich in reduced silica surface sites. The most important observation has been that thermodynamically metastable bimetallic elemental surface compositions were shown to be stabilized by these unique sites. Consequently, the selectivity in the conversion of acetylene to ethylene and benzene could be modified by a thermally stable low density of gold surface atoms. Cluster sintering and gold surface segregation became more activated; therefore, the overall reactivity could be maintained at a much higher temperature. Although demonstrated specifically for Au−Pd bimetallic clusters for the case of an acetylene conversion reaction to ethylene and benzene, we speculate that the implications of our results could affect other oxide-supported bimetallic catalysts.

surface carbon was outside the scope of this study. Particles that were supported on the highly defected SiO2 substrate show better thermal stability both against aggregation and surface atom rearrangements within a cluster. In this case, decreases of about 50 and 25% were observed for the Pd/Si and Pd/Au AES ratios, respectively. One can therefore estimate that there is about equal importance for whole cluster sintering as for intracluster gold atom surface segregation in leading to the above AES signal decay. When annealed to 600 K, at least 25% of the clusters’ surface contained Pd atoms; therefore, the benzene and ethylene formation were cut to about 30% of their initial rates (Figure 6a). In addition to the improved thermal stability, the sputtered, corrugated surface with reduced silica sites (no annealing) has also accelerated the ethylene formation rate compared to that of the defect-free silica, whereas the benzene formation yield has increased as well but at a slower pace. Before and following sputtering at 2.8 keV for 10 min, the ethylene to benzene formation rate ratio is 15:1 and 22:1, respectively (Figure 6b). These results may seem to contradict our conclusions from the CO-TPD titration data presented in Figure 4a. However, one should conclude that the enriched Pd surface composition of clusters on the defect-rich substrates apparently enhances acetylene decomposition faster and at lower temperatures than its trimerization. As a result, both ethylene and benzene formation rates simultaneously increase, but ethylene formation is accelerated more than benzene formation. A relevant previous study has shown that Pd atoms tend to segregate to the surface of Au/Pd nanoparticles upon adsorption of gas molecules such as carbon monoxide.27 Our results add the effect of the cluster−substrate interaction as an additional trigger for the surface segregation of Pd in the case of Au−Pd bimetallic clusters. 3.5. Amorphous Carbon and Silica Substrates: Morphology versus Electronic Effects on Reactivity. In this work, we have correlated structural and morphology modifications of bimetallic Au−Pd clusters observed on top of amorphous carbon (a-C) substrates as a result of defect formation using energetic Ar+ ion sputtering to similar modifications induced on such clusters deposited on SiO2/ Si(100). On the a-C surface, the effect of energetic ion sputtering is primarily corrugation and surface roughness formation. New measurable electronic states are not expected to be generated. These surface defects expose carbon atoms that are not fully coordinated with their neighbors or have more reactive dangling bonds compared to bulk, fully coordinated atoms. In contrast, silica surfaces treated in a similar way by energetic ions stabilize new electronic states that are characterized as reduced silica sites via XPS and Auger measurements. This change is in addition to morphology and structural modifications. The question arises whether the significant changes observed in reactivity and overall behavior of the Au−Pd bimetallic clusters over sputtered silica can be associated solely with the new electronic states (reduced silica sites). On the basis of the observed structural and thermal stability of the clusters against diffusion, aggregation and sintering on the sputtered, defectrich a-C substrate (Figure 2), and similar overall behavior expected on the silica surface,35 we conclude that electronic effects indeed dominate in the case of silica surfaces, and the role of pure morphology/structural modifications on cluster reactivity can most probably be neglected.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Partial support by an ACS-PRF grant, the US-Israel Binational Science Foundation, the Israel Ministry of Science, and the Israel Science Foundation is acknowledged. E.G. is grateful for a generous fellowship given by The Eshkol Foundation, administered by the Israel Ministry of Science. A.M. acknowledges the Lady Davis Foundation for a postdoctoral fellowship.

■

REFERENCES

(1) Valden, M.; Lai, X.; Goodman, D. W. Onset of Catalytic Activity of Gold Clusters on Titania with the Appearance of Nonmetallic Properties. Science 1998, 281, 1647. (2) Hashmi, A. S. K.; Hutchings, G. J. Gold Catalysis. Angew. Chem., Int. Ed. 2006, 45, 7896. (3) Yoon, B.; Häkkinen, H.; Landman, U.; Wörz, A. S.; Antonietti, J. M.; Abbet, S.; Judal, K.; Heiz, U. Charging Effects on Bonding and Catalyzed Oxidation of CO on Au8 Clusters on MgO. Science 2005, 307, 403. (4) Janssens, T. V. W.; Clausen, B. S.; Hvolbaek, B.; Falsig, H.; Christensen, C. H.; Bligaard, T.; Nørskov, J. K. Insights into the Reactivity of Supported Au Nanoparticles: Combining Theory and Experiments. Top. Catal. 2007, 44, 15. (5) Baumer, M.; Freund, H. J. Metal Deposits on Well-Ordered Oxide Films. Prog. Surf. Sci. 1999, 61, 127. (6) Herzing, A. A.; Kiely, C. J.; Carley, A. F.; Landon, P.; Hutchings, G. J. Identification of Active Gold Nanoclusters on Iron Oxide Supports for CO Oxidation. Science 2008, 321, 1331. (7) Yang, F.; Chen, M. S.; Goodman, D. W. Sintering of Au Particles Supported on TiO2(110) during CO Oxidation. J. Phys. Chem. C 2009, 113, 254. (8) Datye, A. K.; Xu, Q.; Kharas, K. C.; McCarty, J. M. Particle Size Distributions in Heterogeneous Catalysts: What Do They Tell Us about the Sintering Mechanism? Catal. Today 2006, 11, 59. (9) Wallace, W. T.; Min, B. K.; Goodman, D. W. The Nucleation, Growth, And Stability of Oxide-Supported Metal Clusters. Top. Catal. 2005, 34, 17. (10) Parker, S. C.; Campbell, C. T. Reactivity and Sintering Kinetics of Au/TiO2(110) Model Catalysts: Particle Size Effects. Top. Catal. 2007, 44, 3. 6030

dx.doi.org/10.1021/la400005d | Langmuir 2013, 29, 6025−6031

Langmuir

Article

(11) Wahlström, E.; Lopez, N.; Schaub, R.; Thostrup, P.; Rønnau, A.; Africh, C.; Lægsgaard, E.; Nørskov, E. J. K.; Besenbacher, F. Bonding of Gold Nanoclusters to Oxygen Vacancies on Rutile TiO2(110). Phys. Rev. Lett. 2003, 90, 26101. (12) Pacchioni, G. Oxygen Vacancy: The Invisible Agent on Oxide Surfaces. Chem. Phys. Chem. 2003, 4, 1041. (13) Pacchioni, G.; Giordano, L.; Baistrocchi, M. Charging of Metal Atoms on Ultrathin MgO/Mo(100) Films. Phys. Rev. Lett. 2005, 94, 226104. (14) Sarkany, A.; Geszti, O.; Safran, G. Preparation of Pd Shell−Au Core/SiO2 Catalyst and Catalytic Activity for Acetylene Hydrogenation. Appl. Catal., A 2008, 350, 157. (15) Choudhary, T. V.; Sivadinarayana, C.; Datye, A. K.; Kumar, D.; Goodman, D. W. Acetylene Hydrogenation on Au-Based Catalysts. Catal. Lett. 2003, 86, 1. (16) Tysoe, W. T.; Nyberg, G. L.; Lambert, R. M. Photoelectron Spectroscopy and Heterogeneous Catalysis: Benzene and Ethylene from Acetylene on Palladium (111). Surf. Sci. 1983, 135, 128. (17) Baddeley, C. J.; Ormerod, R. M.; Stephenson, A. W.; Lambert, R. M. Surface Structure and Reactivity in the Cyclization of Acetylene to Benzene with Pd Overlayers and Pd/Au Surface Alloys on Au{111}. J. Phys. Chem. 1995, 99, 5146. (18) Baddeley, C. J.; Tikhov, M.; Hardacre, C.; Lomas, J. R.; Lambert, R. M. Ensemble Effects in the Coupling of Acetylene to Benzene on a Bimetallic Surface: A Study with Pd{111}/Au. J. Phys. Chem. 1996, 100, 2189. (19) Storm, J.; Lambert, R. M.; Memmel, N.; Onsgaard, J.; Taglauer, E. Catalysis with Monometallic and Bimetallic Films: Trimerization of Acetylene to Benzene on Pd/Ru(001) and (Pd+Au)/Ru(001). Surf. Sci. 1999, 436, 259. (20) Holmblad, P. M.; Rainer, D. R.; Goodman, D. W. Particle Size Effects in the Acetylene Cyclotrimerization on Planar Model Al2O3 Thin Film Supported Pd Clusters. J. Phys. Chem. B 1997, 101, 8883. (21) Gentle, T. M.; Muetterties, E. L. Acetylene, Ethylene, and Arene Chemistry of Palladium Surfaces. J. Phys. Chem. 1983, 87, 2469. (22) Abbet, S.; Sanchez, A.; Heiz, U.; Schneider, W. D. Tuning the Selectivity of Acetylene Polymerization Atom by Atom. J. Catal. 2001, 198, 122. (23) Gross, E.; Horowitz, Y.; Asscher, M. Water as Buffer Material for Gold Nanocluster Growth. Langmuir 2005, 21, 8892. (24) Gross, E.; Popov, I.; Asscher, M. Chemical Reactivity of Pd−Au Bimetallic Nanoclusters Grown via Amorphous Solid Water as Buffer Layer. J. Phys. Chem. C 2009, 113, 18341. (25) (a) Gross, E.; Asscher, M. Structure−Reactivity Correlations in Pd−Au Bimetallic Nanoclusters. Langmuir 2010, 26, 16226. (b) Gross, E.; Asscher, M. Size to Density Coupling of Supported Metallic Clusters. Phys. Chem. Chem. Phys. 2009, 11, 710−716. (26) Meenakshisundaram, S.; Nowicka, E.; Miedziak, P. J.; Brett, G. L.; Jenkins, R. L.; Dimitratos, N.; Taylor, S. H.; Knight, D. W.; Bethell, D.; Hutchings, G. J. Oxidation of Alcohols Using Supported Gold and Gold−Palladium Nanoparticles. Faraday Discuss. 2010, 145, 341. (27) Hugon, A.; Delannoy, L.; Krafft, J.-M.; Louis, C. Selective Hydrogenation of 1,3-Butadiene in the Presence of an Excess of Alkenes over Supported Bimetallic Gold−Palladium Catalysts. J. Phys. Chem. C 2010, 114, 10823. (28) Kolli, N. E.; Delannoy, L.; Louis, C. Bimetallic Au−Pd Catalysts for Selective Hydrogenation of Butadiene: Influence of the Preparation Method on Catalytic Properties. J. Catal. 2010, 297, 79. (29) Somorgai, G. A. Introduction to Surface Science and Catalysis; Wiley: New York, 1994. (30) Goodman, D. W. Model Studies in Catalysis Using Surface Science Probes. Chem. Rev. 1995, 95, 523. (31) Campbell, C. T. Ultrathin Metal Films and Particles on Oxide Surfaces: Structural, Electronic and Chemisorptive Properties. Surf. Sci. Rep. 1997, 27, 1. (32) Freund H.-J., Goodman D. W. Handbook of Heterogeneous Catalysis, 2nd ed.; Wiley-VCH: Weinheim, Germany, 2008; Vol. 3, Chapter 4.5, pp 1309−1338.

(33) Freund, H.-J. Adsorption of Gases on Complex Solid Surfaces. Angew. Chem., Int. Ed. Engl. 1997, 36, 452. (34) Huang, L.; Chey, S. J.; Weaver, J. H. Buffer-Layer-Assisted Growth of Nanocrystals: Ag-Xe-Si(111). Phys. Rev. Lett. 1998, 80, 4095. (35) Schmidt, A.; Eggers, H.; Herwig, K.; Anton, R. Comparative Investigation of the Nucleation and Growth of fcc-Metal Particles (Rh, Ir, Ni, Pd, Pt, Cu, Ag, Au) on Amorphous Carbon and SiO2 Substrates during Vapor Deposition at Elevated Temperatures. Surf. Sci. 1996, 349, 301. (36) Tyson, W. R.; Miller, W. A. Surface Free Energies of Solid Metals: Estimation from Liquid Surface Tension Measurements. Surf. Sci. 1977, 62, 267. (37) Gross, E.; Lundwall, E. M.; Goodman, D. W.; Asscher, M. Gold Nanoclusters Deposited on SiO2 via Water as Buffer Layer: CO-IRAS and TPD Characterization. J. Phys. Chem. C 2007, 111, 16197. (38) Mezey, L. Z.; Giber, J. The Surface Free Energies of Solid Chemical Elements: Calculation from Internal Free Enthalpies of Atomization. Jpn. J. Appl. Phys. 1982, 21, 1569. (39) Bekkay, T.; Sacher, E.; Yelon, A. Surface Reaction during the Argon Ion Sputter Cleaning of Surface Oxidized Crystalline Silicon (111). Surf. Sci. 1989, 207, L377.

6031

dx.doi.org/10.1021/la400005d | Langmuir 2013, 29, 6025−6031