Dynamic Phase Separation in Supported Pd–Au Catalysts - The

Jan 14, 2015 - SiO2-supported Pd–Au catalysts with Pd/Au molar ratios varying from 0.8 to 7.0 were used as catalysts for vinyl acetate synthesis und...
2 downloads 15 Views 484KB Size
Article pubs.acs.org/JPCC

Dynamic Phase Separation in Supported Pd−Au Catalysts Stefanie Simson, Andreas Jentys, and Johannes A. Lercher* Department Chemie and Catalysis Research Center, Technische Universität München, Lichtenbergstrasse 4, 85748 Garching, Germany S Supporting Information *

ABSTRACT: SiO2-supported Pd−Au catalysts with Pd/Au molar ratios varying from 0.8 to 7.0 were used as catalysts for vinyl acetate synthesis under industrial conditions. Continued operation of the bimetallic catalysts at 150 °C led to the formation of the Pd1Au1 phase in the particles, with the remaining Pd atoms forming Pd nanoparticles by leaching of Pd as acetate. The presence of these phases was monitored by X-ray absorption spectroscopy (XAS) of the used catalysts. Temperature-resolved in situ XRD of the reduced samples in an inert atmosphere confirmed the phase separation into a Pd-rich phase and a Au-rich phase above 160 °C. CO adsorption and XRD of the catalysts used at 180 °C showed that phase separation also took place during vinyl acetate synthesis. The pronounced temperature dependence of the morphology and surface composition of the bimetallic Pd−Au catalysts determines the selectivity; the activity; and, in particular, the stability during vinyl acetate synthesis.



INTRODUCTION Bimetallic noble-metal catalysts are key to many catalytic processes, including Pd-catalyzed oxidation or acetoxylation of ethylene to vinyl acetate, which is formally an oxidative esterification CH3COOH + C2H4 +

Pd1Au3; and, at lower temperatures, an equimolar Pd1Au1 phase (see Figure 1).7 Zhu and Hou8 investigated the role of mechanical stress in the chemical ordering of Pd−Au nanoalloys for 309- and 561-

1 O2 2

→ CH3COOCHCH 2 + H 2O

(1)

and is industrially performed on SiO2 or SiO2/Al2O3 supported bimetallic Pd/Au catalysts.1 Little consensus exists on the details of the reaction mechanism and active species, despite over 30 years of ongoing research.2 The reaction was reported to be highly structuresensitive; specifically, the catalytic properties were found to depend markedly on the local structure and chemical composition of the catalyst.3 The addition of the catalytically inactive Au to the Pd-based catalyst was reported to cause a pronounced increase in activity and stability.4,5 This behavior was linked to the presence of isolated Pd atoms in a specific configuration. The formation of these ensembles was claimed to depend subtly on the reaction conditions, and it was speculated that the structure is dynamically formed and disintegrated during the reaction.6 The Pd−Au phase diagram shows three ordered solid phases with compositions of Pd3Au1; © XXXX American Chemical Society

Figure 1. Pd−Au phase diagram reprinted from Okamoto and Massalski.7 Received: October 6, 2014 Revised: December 9, 2014

A

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Au ratios of 6.0, 2.0, and 1.1 are shown as examples in this article. Characterization. Atomic Absorption Spectroscopy (AAS). The concentrations of Pd, Au, and K were determined by atomic absorption spectroscopy using a UNICAM 939 AA spectrometer. Typically, 30 mg of the sample wass dissolved in a mixture of HF and HNO3/HCl. The samples were dried before dissolution at 473 K. IR Spectroscopy of Adsorbed CO. The samples were prepared as self-supporting wafers with a density of approximately 10 mg/cm−2. The samples were first activated in a vacuum (better than 1.0 × 10−7 mbar) at 300 °C for 1 h at a heating rate of 5 °C/min, cooled to room temperature, and subsequently reduced in H2 at 300 °C at a heating rate of 5 °C/ min. After reduction, the samples were outgassed at 300 °C for 30 min. After the pretreatment, the samples were cooled in 10 mbar He (added for enhanced thermal conductivity) to −150 °C, and an IR spectrum of each sample was taken. Subsequently, the cell was evacuated, and 1 mbar CO was introduced. To increase thermal conductivity, 10 mbar He was added, and spectra were collected at −150 °C. The IR spectra were recorded on a Vertex 70 spectrometer from Bruker Optics with a resolution of 4 cm−1 and an accumulation of 128 interferograms. The spectra were background-corrected, and the contributions of CO on the SiO2 support and on residual cations were subtracted. The (difference) spectra were deconvoluted using GRAMS/AI 9.0 R2 spectroscopy software constraining the peak positions and values for the half-width at half-height within a range of 4 cm−1 for the samples before reaction. For samples after reaction, the band positions of the CO adsorbed were allowed to vary within a range of 20 cm−1 to account for the peak broadening and peak shifts originating from the presence of residual K. X-ray Powder Diffraction. X-ray powder diffraction measurements were conducted on a Philips X’Pert Pro System using Cu Kα radiation with a wavelength of 0.154 nm (45 kV and 40 mA). The experiments were carried out with a rotating sample holder in a 2θ range of 5−70° with a step size of 0.019°/s. Temperature-dependent in situ measurements were performed in an Anton Paar HTK 1200 oven under flowing H2 or He at a heating rate of 3 °C/min and a cooling rate of 2 °C/ min. To prevent the formation of palladium hydrides, the H2 gas stream was changed to He at the maximum temperature. The deconvolution of the diffraction profiles was performed using HighScore Plus 3.0a software. X-ray Absorption Spectroscopy. X-ray absorption spectra of aged samples were collected at HASYLAB, DESY, Hamburg, Germany, on beamlines X1 and C. The storage-ring energy was 4.5 GeV, and the current decay during a typical fill was from 110 to 90 mA. Temperature-dependent in situ X-ray absorption spectroscopy measurements were performed at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France, on the BM26 (DUBBLE) Dutch-Belgian Beamline. The samples were prepared as self-supporting wafers, having a total absorbance of 2.0 for the Au LIII edge and 1.5 for the Pd K edge. The samples were flushed with He inside the sample chamber and cooled to liquid-N2 temperature for ex situ measurements. In situ measurements were conducted with the same temperature program as used for in situ XRD measurements. For extended X-ray absorption fine structure (EXAFS) analysis, the scattering contributions in the pre- and postedge background were removed from the X-ray absorption spectra by a third-order polynomial spline function. The oscillations

atom clusters. Three stable phases were identified, in good agreement with the Pd−Au phase diagram. Similar ordered phases were reported by Atanasov and Hou9 for 923-atom clusters. For these cuboctahedral particles, the formation of ordered phases was found to be caused by the negative mixing enthalpy, as the temperature-induced stress leading to disordering did not overcome the impact of the mixing enthalpy. Because the stability of these two phases is temperaturesensitive, a pronounced temperature dependence of the selectivity for vinyl acetate synthesis can be expected if phase formation is important. The catalyst activity in vinyl acetate synthesis varies in a complex pattern. Typically, the catalyst activity increases initially during the first 20−30 h time on stream and then exhibits a very slow deactivation over the next 110 h of reaction time.10 To stabilize the catalyst over longer time periods, it is essential to understand the structural properties of the catalyst. With respect to activity, it has been stated that the controlled growth of the metallic particles to sizes of about 2−3 nm is required to obtain highly active catalysts.11 Particles smaller than 1 nm are considered to be too small to create the catalytically active ensembles of two Pd monomers in the direct neighborhood. Under the reaction conditions, the surface of the catalysts is covered with several monolayers of acetic acid12 and water, the latter being formed in the reaction (see eq 1). The concentration of adsorbed acetic acid increases significantly with the introduction of potassium acetate (CH3COOK) added as a promoter, which significantly enhances the adsorption of acetic acid by forming a dimeric species with acetic acid.13 In a previous study,10 we showed that a bimetallic Pd1Au1 phase and a highly dispersed monometallic Pd phase in the form of palladium acetate was present under working conditions, and that both are only formed when the catalyst is exposed to the reactants. In contrast, thermal treatment under a reducing or inert atmosphere led to a different structure of the bimetallic particles. We attributed the restructuring of the bimetallic particles under working conditions to the formation of palladium acetate, predominantly on the larger Pd assemblies, followed by the subsequent dissolution of Pd from the bimetallic PdxAuy in the form of acetates.10 Based on this hypothesis, the present study addresses the temperature dependence of the restructuring process under a reactive atmosphere. A combination of bulk-sensitive methods, such as X-ray absorption spectroscopy (XAS) and X-ray diffraction (XRD), with the (surface-sensitive) sorption of CO was applied to obtain a detailed description of the (bi)metallic particle structure. By applying either a reactive or a nonreactive atmosphere over a wide temperature range, the formation of mono and bimetallic phases was studied qualitatively and quantitatively.



EXPERIMENTAL METHODS Synthesis. Pd−Au/SiO2 catalysts were prepared by incipient wetness impregnation with an aqueous solution of HAuCl3 and PdCl2 according to the method of ref 14. The synthesis procedure includes precipitation and washing steps to remove chloride from the catalyst. The series of catalysts was prepared with a constant total metal loading of 3 wt %. Flamesynthesized SiO2 (WACKER HDK) with a specific surface area of 200 m2/g was used as the support. After being freeze-dried, the Pd−Au/SiO2 catalysts were reduced in flowing H2 at 300 °C for 1 h at a heating rate of 3 °C/min. The samples with Pd/ B

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C Table 1. EXAFS Fit Results for Pd/Au Samples before and after Reaction at 150 °C Au−Pd

a

Au−Au

Pd−Au

Pd−Pd

Pd/Au molar ratio

before or after reactiona

N

r (nm)

N

r (nm)

N

r (nm)

N

r (nm)

1.1 1.1 2.0 2.0 6.0 6.0

BR AR BR AR BR AR

4.0 6.5 4.5 7.4 5.4 7.2

0.277 0.278 0.277 0.277 0.276 0.277

7.8 4.6 7.3 3.9 6.5 3.8

0.279 0.279 0.280 0.278 0.280 0.279

4.5 5.9 2.2 4.1 1.1 1.2

0.277 0.278 0.277 0.277 0.276 0.277

4.7 4.5 7.9 5.9 9.3 8.8

0.275 0.281 0.275 0.280 0.275 0.281

BR, before reaction; AR, after reaction.

Figure 2. Particles modeled with different concentrations of ordered Pd1Au1 phase (Au, orange spheres; Pd, blue spheres): (a) 0%, (b) 25%, (c) 75%, and (d) 100% ordered Pd1Au1 phase. Expected values of (i) NPdAu/NPdPd and (ii) NAuPd/NAuAu and concentrations of ordered Pd1Au1 in the particle.

were weighted by k2 and fitted in k space within the limits k = 2.1−12 Å−1 for the Pd K edge and k = 2.8−12 Å−1 for the Au LIII edge. XAS data processing and EXAFS analysis were performed using IFEFFIT15 with the Horae package16 (Athena and Artemis). The values for So2 were derived from EXAFS data of references with known coordination numbers and were fixed during analysis (So2 was found to be 1.0 for Au and for Pd). The multiple-edge fitting in k space was carried out using the following constraints17 NAuPd = NPdAux Pd /xAu

(1)

rAuPd = rPdAu

(2)

σAuPd = σPdAu

(3)

contribute to the amplitude of the EXAFS function, the third cumulant C3 was taken into account for temperaturedependent measurements.21 After having determined the amplitude-dependent parameters, the third cumulants C3, together with r and ΔE0, were refined, whereas all other parameters were held constant. ΔE0 was allowed to vary without constraints, because ΔE0 accounts for an overall phase shift between the experimental and calculated spectra.22 Modeling of the Local Environment of the Atoms in the Particles. To support the analysis of the properties of the metal particles using the first-shell coordination numbers from EXAFS, various potential metal particles with structures ranging from fully ordered phases to fully random ensembles were generated based on the fcc lattice structure using Accelrys Material Studio 6.0, and the (first-shell) coordination numbers for the Pd and Au atoms in these clusters were calculated based on the particle geometry. The particle size used for bimetallic particles was 2 nm (corresponding to 1985 metal atoms). The highly dispersed Pd phase was modeled by adding monometallic particles with a size of 0.5 nm (43 atoms). The particle sizes were chosen based on transmission electron microscopy (TEM) measurements of catalysts after reaction.

The interatomic distances (r), variances (Debye−Waller factor σ), and coordination numbers (N) were first determined using the correlated Einstein model18,19 [applying the ein(T,θ) function in Artemis], to account for the temperature dependency of the Debye−Waller factor. It was shown that neglecting the anharmonic term at elevated temperatures leads to a nonphysical decrease in interatomic distances.20 Because odd cumulants contribute to the phase whereas even ones C

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 3. Bimetallic particles modeled with an additional monometallic Pd phase (Au, orange spheres; Pd, blue spheres): (a) ordered Pd1Au1 phase with 55% dispersed Pd, (b) random Pd1Au1 phase with 55% dispersed Pd. Correlations of (i) NPdAu/NPdPd and (ii) Ntotal with the fraction of dispersed Pd for fully random and fully ordered Pd1Au1: (□) 100% random asnd (◊) 100% ordered bimetallic particles. The fraction of dispersed Pd was calculated based on the total content of Pd in the sample.

Aging under Industrial Reaction Conditions. Aging under industrial conditions was performed in a 6-fold reactor setup at temperatures of 150 and 180 °C. The gas composition consisted of 60 vol % C2H4, 13 vol % AcOH, 4.5 vol % O2, and N2 balance. The weight of the catalysts was selected in such a way that the amount of Pd was constant in the catalyst bed independent of the Pd/Au ratio present in the samples. SiC was used as an inert diluent to ensure an isothermal temperature distribution over the catalyst bed. The diluent-to-catalyst ratio was 10:1. The product stream was analyzed using a GC 2014 gas chromatograph from Shimadzu, equipped with a Haysep Q and a molecular sieve column and a thermal conductivity detector (TCD).

structure models of ordered and randomly arranged bimetallic particles together with different fractions of monometallic dispersed Pd particles were generated, and the average coordination numbers of the atoms were calculated. The correlation between the coordination numbers and the degree of ordering within the bimetallic particle is shown in Figure 2. For the ratios between the coordination numbers NAuPd/NAuAu and NPdAu/NPdPd, a value of 2 was obtained for particles with the Pd1Au1 phase (full ordering), whereas for a fully random arrangement of the Pd and Au atoms, both ratios were 1. Note that, in the presence of an additional monometallic phase, the correlations shown in Figure 2 can be applied only for the metal atoms incorporated in the bimetallic particles. For homogeneously distributed bimetallic particles, the sums of NPdAu + NPdPd and NAuPd + NAuAu should be equal, which was not observed in the EXAFS analysis (compare Table 1). Thus, the fraction of the monometallic phase was estimated from the deviation of these ratios for the metal atoms being present in both phases. In the case of the bimetallic PdAu particles studied here, the differences in the coordination numbers of the Pd and Au neighbors confirm that Au is present only in the bimetallic particles, whereas Pd forms an additional monometallic phase. Depending on the degree of ordering in the bimetallic phase, the additional Pd phase has a different effect on the coordination numbers. Figure 3 shows the ratio of NPdAu/ NPdPd as a function of the concentration of the dispersed monometallic Pd phase present. For an increasing fraction of dispersed Pd phase present in the catalysts, both the ratio NPdAu/NPdPd and the quantity Ntotal (i.e., NPdAu + NPdPd) decrease (see Figure 3). Applying the correlations presented in Figures 2 and 3 to the results of the EXAFS analysis, the fraction of dispersed Pd and the type of bimetallic phase were determined by the following steps.



RESULTS Particle Morphology. The results of the EXAFS analyses for the bimetallic Pd−Au particles before and after vinyl acetate synthesis at 150 °C are summarized in Table 1. (The complete results of the EXAFS analysis and examples of the fitted data are documented in the Supporting Information.) As discussed in ref 10, an NAuPd/NAuAu ratio of close to 2.0 was observed after reaction, which indicates the presence of an ordered Pd1Au1 phase. However, the NPdAu/NPdPd ratio before reaction was significantly smaller than 2.0 and varied with the Pd/Au molar ratio in the catalyst. This difference is attributed to the formation of a monometallic Pd phase during reaction by leaching of Pd from the Pd-rich bimetallic particles, which continued until an equimolar ratio of Pd and Au remained in the bimetallic particles. To support the hypothesis that bimetallic particles with a composition of Pd1Au1 together with a monometallic Pd phase (in the form of palladium acetate under the reaction conditions) were formed during vinyl acetate synthesis, D

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C From the NAuPd/NAuAu ratio, which is around 2 for all samples after reaction, the presence of Au in an ordered 1:1 phase was deduced. Using the ratio NPdAu/NPdPd, which includes contributions from the bimetallic and monometallic particles, in the second step, the fraction of isolated Pd particles can be estimated. The results are summarized in Table 2 and Table 2. Fraction of Dispersed Pda Calculated on the Basis of the Chemical Composition and the Correlations Found by Particle Modeling Pd/Au molar ratio

calculated from chemical compositionb

calculated from correlationsc

1.1 2.0 6.0

0.04 0.39 0.71

0.12 0.40 0.62

a

With respect to the total Pd concentration. bCalculated according to the expression fraction of dispersed Pd = [c(Pd total) − c(Pd in Pd1Au1 phase)]/c(Pd total). cCalculated as the average value of the fractions determined according to the correlation based on NPdAu/ NPdPd and NPd total found by particle modeling.

Figure 4. TTemperature dependent XRD patterns for the reflection at the 111 plane of bimetallic particles for (left) Pd/Au 1.1, (center) Pd/ Au 2.0 and (right) Pd/Au 6.0 at the temperature and atmosphere indicated. 2θ = 38.3° represents pure Au phases, and 2θ = 40.1° represents pure Pd phases.

compared to the concentration of dispersed Pd phase calculated on the basis of the chemical composition of the respective sample and the presence of a Pd1Au1 phase. Whereas the values agree well for the samples with a higher Pd/Au ratio, the Pd/Au 1.1 sample shows a slightly higher concentration of Pd nanoparticles than expected from the chemical composition, demonstrating the limitations of the combined EXAFS and simulation approach. Temperature-Dependent Phase Separation Followed by X-ray Diffraction. The temperature-dependent formation of bimetallic PdAu particles was investigated by in situ XRD. The experiments were carried out in H2, because preliminary tests showed that surface PdO, formed during storage of the catalysts in air, strongly restricts the mobility of the metal atoms within the particles. To avoid palladium hydride formation at lower temperatures (Pd−H was found to be stable only below 150 °C at 1 bar; see Supporting Information), in situ XRD data measured in reducing atmosphere are shown in the Supporting Information), H2 was replaced by He before the sample was cooled to room temperature after the reduction. Under reducing conditions, a pronounced phase separation toward a Pd-rich Pd3Au1 phase and a Au-rich Pd1Au3 phase was found as the temperature was increased to 300 °C, whereas during cooling to 170 °C in He, ordered phases were formed according to the Pd−Au phase diagram.7 The changes in the XRD pattern during temperature treatment for Pd/Au 1.1, Pd/ Au 2.0, and Pd/Au 6.0 are shown Figure 4. The phase segregation during the heating period started at approximately 160 °C, and the degree of phase separation was more pronounced for Pd-rich samples (Figure 4; example temperature-dependent XRD patterns are shown for Pd/Au 1.1, 2.0, and 6.0). The compositions of the alloy phases at the end of the temperature treatment are summarized in Table 3. Temperature-Dependent Phase Separation Followed by X-ray Absorption Spectroscopy. X-ray absorption spectroscopy was applied to verify the temperature-dependent phase separation observed by XRD. The samples were subjected to the same temperature treatment and gas atmosphere as for the temperature-dependent X-ray diffraction studies. As the phase separation at elevated temperatures was more pronounced for samples with high concentrations of Pd, a

Table 3. Alloy Compositions Determined by XRD of Samples Aged by Thermal Treatmenta Pd/Au molar ratio

PdxAuy phases presentb

1.1 2.0 6.0

Pd0.17Au0.83, Pd0.51Au0.49, Pd0.81Au0.19 Pd0.12Au0.88, Pd0.52Au0.48, Pd0.80Au0.20 Pd0.20Au0.80, Pd0.48Au0.59, Pd0.81Au0.19

a

Compositions obtained by deconvolution of the XRD profiles after the thermal treatment. bPhases present after thermal treatment in a reducing atmosphere (with He flushing to remove Pd−H species).

bimetallic sample with a Pd/Au molar ratio of approximately 5 was studied by in situ XAS in detail. Figure 5 shows the normalized X-ray absorption near-edge structure (XANES) spectra at the Au LIII edge. Because p-to-d electronic transitions are observed at the Au LIII edge, the electron density in the d states is reflected by the intensity of the white line. As a reference, the XANES spectrum of a monometallic Au/SiO2 sample is presented in orange. All XANES spectra measured on the bimetallic Pd−Au/SiO2 samples showed higher white line intensities than for Au/SiO2, which indicates an electron deficiency of Au. Note that s-to-p transitions are observed at the Pd K edge; therefore, only minor effects on the presence of Au on the electron density in the d states of Pd were observed in the XANES spectra at this edge. EXAFS analyses on the Au LIII edge and the Pd K edge were used to further characterize the temperature-dependent reconstruction in bimetallic PdAu particles. Table 4 summarizes the structural parameters at the different temperatures. Coordination numbers and interatomic distances were calculated from EXAFS analysis. The temperature dependence of the EXAFS signal was taken into account using the correlated Einstein model and including the third cumulant C3 in the fitting procedure. The total coordination number of Pd was smaller than the total coordination number of Au. Both Pd and Au and all bimetallic PdAu phases exhibit cubic face-centered lattices, for which the coordination number is 6 for the atoms at edges, 9 E

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 5. Normalized XANES spectra in He at 588 K at the Au LIII edge of Au/SiO2 and Pd−Au/SiO2 during temperature treatment: orange line, Au/SiO2; dark blue line, Pd−Au/SiO2 (50 °C); medium blue line, Pd−Au/SiO2 (100 °C); light blue line, Pd−Au/SiO2 (150 °C); red line, Pd−Au/ SiO2 (250 °C); dark red line, Pd−Au/SiO2 (300 °C).

Table 4. Results of the EXAFS Analysis of PdAu/SiO2 at Different Temperatures during Temperature-Dependent EXAFS Measurements Au−Pd

Au−Au

Pd−Au

Pd−Pd

temperature (°C)

N

r (nm)

N

r (nm)

N

r (nm)

N

r (nm)

50 100 150 200 250 300

4.9 4.6 4.7 4.6 4.4 4.4

0.280 0.281 0.278 0.278 0.277 0.277

4.8 4.2 4.9 5.9 6.0 5.9

0.286 0.286 0.284 0.284 0.284 0.285

0.99 0.91 0.94 0.92 0.88 0.89

0.280 0.281 0.278 0.278 0.277 0.277

5.7 5.7 5.5 5.7 5.9 5.7

0.282 0.281 0.277 0.276 0.277 0.276

Figure 6. Ratios of the coordination numbers at the (a) Au edge (NAuPd/NAuAu) and (b) Pd edge (NPdAu/NPdPd) as functions of temperature. Orange and blue lines are added as guides for the eye.

large excess of Pd, the values on the Pd edge are less affected; however, a slight change in average coordination numbers for temperatures above 150 °C was also observed (Figure 6b). Because the elevated temperatures were taken into account in the fitting procedure, the interatomic distances were accurate at higher temperatures. Figure 7 shows the temperature dependence of the interatomic distances. The Au−Au distance deviates only slightly from the bulk distance, and changes are within the measurement and fitting errors. The Pd−Pd distances are greater at lower temperatures (50 and 100 °C), which is attributed to the formation of palladium hydrides. The effect is less pronounced for the mixed distances, Pd−Au and

for atoms on the (111) planes, and 12 for atoms in the interior of the particle. Thus, low coordination numbers of Pd can originate from the presence of small monometallic particles or from a surface enrichment within a bimetallic particle, whereas higher total coordination numbers of Pd suggest a preferred location in the core of the bimetallic particle. Two temperature regimes with respect to coordination numbers were observed. The values for NAuAu in the low temperature range (≤150 °C) were lower than those at elevated temperatures. NAuPd, in contrast, was higher at lower temperatures and decreased with increasing temperature. The NAuPd/NAuAu ratio is illustrated in Figure 6a. Because of the F

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 8. XRD profiles of the 111 reflection of bimetallic particles after 120 h on stream at (left) 150 and (right) 180 °C for Pd/Au 1.1, Pd/ Au 2.0, and Pd/Au 6.0.

Figure 7. Temperature dependences of the (a) Au−Au, (b) Pd−Pd, and (c) Pd−Au (or equivalently Au−Pd) interatomic distances.

compositions found after 120 h on stream at reaction temperatures of 150 and 180 °C, which were calculated according to Vegard’s law.26 Please note that alloy compositions of samples aged at 180 °C were affected by the slow cooling rate (necessary to purge the AcOH from the system), which caused a re-formation of the Pd1Au1 phase that is stable only at lower temperatures. In general, the full width at half-maximum height of the reflections was higher for Pd/Au 1.1, whereas the XRD patterns of the Pd-rich samples Pd/Au 2.0 and 6.0 showed broader peaks for all three phases after reaction at 180 °C, (i.e., a Aurich phase, a highly Pd-rich phase, and a nearly equimolar phase). The reflections corresponding to the Pd-rich phase exhibit the highest full width at half-maximum for all samples investigated. The more Au is present in the sample, the higher the fraction of Au in the bimetallic phase after reaction at 150 °C. For samples aged at 180 °C, the Pd-rich and the Au-rich bimetallic phases showed lower Pd concentrations than samples after thermal treatment in an inert atmosphere (compare Tables 3 and 6). Table 6 summarizes the compositions obtained by fitting of the XRD profiles of sample after vinyl acetate formation at 150 and 180 °C. Low-Temperature CO Adsorption on Catalysts Reactively Aged at 150 and 180 °C. To evaluate the effect of the temperature during the reaction on the surface composition of the catalysts after exposure to the reactants, CO was adsorbed on catalysts aged at 180 °C. Previously, we showed that samples aged at 150 °C exhibit bimetallic particles with similar surface compositions and that Pd and Au are randomly distributed after reaction.10 As the samples were thermally equilibrated with the gas atmosphere under all experimental conditions, a change in the surface properties resulting from the heat released during the adsorption of CO27 can be excluded. The results for the samples aged at 150 and 180 °C are compared in Table 7. CO adsorbed on Au, randomly distributed and in close contact to Pd, is denoted as CO on “Au next to Pd” species. CO adsorbed on “Au next to Au” represents CO on Au atoms that are mainly surrounded by other Au atoms. The corresponding nomenclature is used to assign the bands for CO adsorbed on Pd. The assignments of the linear CO species on Au and Pd are supported by the calculations of Liu and Nørskov.28 The change in frequency of the CO stretching vibration reflects changes in the electron

Au−Pd (which were fixed to the same value during the analysis). The decomposition of the palladium hydride phase started at 150 °C, and the interatomic Pd−Pd distance approached the bulk distance of 0.273 nm. The decomposition of bulk palladium hydride was also observed by XRD, where a distinct shift toward higher 2θ values was observed for Pd/SiO2 in flowing H2 at low temperatures with respect to as-prepared sample, whereas at temperatures above 150 °C, the 2θ position was similar. In contrast, treatment in an inert atmosphere did not cause a shift in the 2θ position. Data on the in situ measurements of Pd/SiO2 are compiled in the Supporting Information. The Pd−Au distances decreased at 150 °C, which is again attributed to the decomposition of palladium hydride. Because Au does not form bulk hydrides, the contraction at this temperature was less pronounced for the mixed distances. The Debye−Waller factors σ2 obtained from EXAFS analysis increased linearly with increasing temperature above 100 K, which is in good agreement with previously published data, including studies of Au bulk materials.23−25 The temperature dependence of C3, as well as detailed results of the complete EXAFS analysis, are provided in the Supporting Information. Temperature-Dependent Phase Formation during Vinyl Acetate Monomer (VAM) Synthesis. After vinyl acetate synthesis at 150 °C, the bimetallic particles consisted of the Pd1Au1 phase with a degree of ordering close to 100%.10 To further investigate the temperature-dependent restructuring of the bimetallic particles during VAM synthesis, samples aged at 180 °C under the reaction atmosphere were analyzed. All samples showed three distinct XRD reflections after exposure to the reactant gases at 180 °C for 120 h, which are attributed to bimetallic phases with compositions close to the three stable ordered phases of PdAu alloys (see Table 5). The reflections for the 111 plane of the samples after reaction at 150 and 180 °C are shown in Figure 8. Table 6 summarizes the alloy Table 5. Alloy Compositions after 120 h on Stream for Reaction Temperatures of 150 and 180 °C Obtained by XRD Profile Fitting Pd/Au molar ratio

150 °C

180 °C

1.1 2.0 6.0

Pd0.42Au0.58 Pd0.47Au0.53 Pd0.51Au0.49

Pd0.02Au0.96, Pd0.45Au0.55, Pd0.79Au0.21 Pd0.08Au0.92, Pd0.51Au0.49, Pd0.67Au0.33 Pd0.06Au0.94, Pd0.45Au0.55, Pd0.89Au0.11 G

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Table 6. Ratios of the IR Intensities of CO Species Adsorbed on Pd and Au in Catalysts Reconstructed in Reactant Atmosphere at 150 and 180 °C Pd/Au molar ratio

reaction temperature (°C)

Au next to Pd/Au next to Aua

Pd next to Au/Pd next to Pd

linear/multifold on Pd

1.1

180 150 180 150 180 150

0.4 0.3 0.5 1.6 0.6 1.5

1.0 0.2 0.5 0.2 0.8 0.2

1.0 0.9 5.4 0.9 6.2 1.4

2.0 6.0 a

Minor overestimation due to overlapping of CO on “Au next to Pd” features and CO weakly adsorbed on the support.

close contact is concluded to result in an electron transfer from Au to Pd,28 leading to an increase in the occupation of the Pd d states and in an enhanced back-donation from the d-level electrons of Pd to the π* orbitals of CO. In the same way the back-donation for electronically altered Au to CO is reduced. These changes in back-donation resulted in a red-shift of the CO stretching vibration on Pd and a blue-shift on Au, the extent of both depend on the surface intermixing. The main difference between samples used at 150 and 180 °C is the ratio in intensity of linear to multifold bound CO on Pd, which is enhanced by a factor of up to 5 for samples aged at 180 °C. This indicates that after reaction at 180 °C the bimetallic surface was less intermixed compared to samples after reaction at 150 °C. Because multifold bound CO was only present on Pd, the effect was more pronounced for Pd-rich catalysts. Samples with low Pd/Au ratios, e.g., Pd/Au 1.1, exhibited high concentrations of Au-rich phases, for which multifold adsorption was not observed. The ratio of CO on “Au next to Pd”/“Au next to Au” was similar for all samples studied after aging at 150 and 180 °C. In general, samples aged at 180 °C showed lower ratios “Au next to Pd”/“Au next to Au” compared to samples used at 150 °C. This indicates that the surfaces were found to be more intermixed after reactions performed at 150 °C. Moreover, the overall intensity of CO adsorbed on Au compared to Pd decreased on samples aged at 180 °C, while the surface

Table 7. Concentrations of Pd and Au on a Catalyst before and after AcOH Treatment for 72 h sample

Pd concentration

Au concentration

catalyst before AcOH treatment catalyst after AcOH treatment AcOH solution after treatment

1.75 wt % 0.70 wt % 400 ppm

0.68 wt % 0.66 wt % 0.4 ppm

density of the Pd d-band states, which are attributed to an increasing number of Au atoms in contact with Pd. The IR spectra after adsorption of CO at −150 °C on the sample with a Pd/Au molar ratio of 6.0 aged at 150 and 180 °C are compared in Figure 9 (for the other samples, see the Supporting Information). The bands at 2156−2185 cm−1 are attributed to the interaction of CO with terminal hydroxyl groups on the support and with cations (e.g., K+). Additionally, a band at 2135 cm−1 was observed, which is assigned to CO weakly bound to SiO2.38 CO linearly adsorbed on “Pd next to Pd” and “Pd next to Au” can be observed in the ranges of 2090 and 2040 cm−1. Multifold adsorbed CO on Pd is found below 2000 cm−1; the intensity of these bands depends on the degree of surface mixing.29 CO linearly adsorbed on “Au next to Pd” and “Au next to Au” is located between 2130 and 2094 cm−1. The shift in wavenumbers with respect to CO on pure Pd (2094 cm−1) and Au (2098 cm−1) depends on the mutual electronic influence between the metallic components. The

Figure 9. IR spectra of CO adsorbed at −150 °C and 1 mbar partial pressure on PdAu/SiO2 with a Pd/Au molar ratio of 6.0: after 120 h time on stream at (a) 150 and (b) 180 °C. Yellow and orange lines represent CO on Au (orange shifted to higher wavenumbers by interaction with Pd); red lines represent CO on Pd (lighter red-shifted to lower wavenumbers by interaction with Au). Black lines represent measured data sets. H

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

toward the equimolar Pd1Au1 phase. The reconstruction in reactant atmosphere is based on the formation of palladium acetate species followed by a partial leaching into the liquid surface layer consisting of AcOH and water. In contrast, under an inert atmosphere, the temperature-dependent reconstruction does not involve leaching, but the reconstruction took place within the bimetallic particle. This temperature-dependent phase separation within the particle was directly observed by in situ XAS. The XANES spectra at the Au LIII edge indicated a lower electron density for Au in the bimetallic PdAu/SiO2 sample than in Au/SiO2. This implies an increasing electron deficiency in the Au d band through alloying with Pd. The white line heights show that temperature has only a slight effect on the electronic state of Au. In situ XRD showed that a temperature increase above 160 °C led to a phase separation toward the Pd-rich Pd3Au1 and the Au-rich Pd1Au3 phases. Because XAS data average over the volume, clusters with a larger number of Au atoms are weighted higher than small Au clusters.20 Thus, the Au-rich phase has a more pronounced influence for white line intensities and coordination numbers. Therefore, the slight decrease in white line intensity for higher temperatures can be assigned to the formation of the Au-rich phase, which is less alloyed and exhibits a lower electron deficiency in the d band of Au. The phase separation at temperatures above 150 °C was also observed from the variations in coordination numbers, where the Au−Au coordination number increased and simultaneously the Au−Pd coordination number deceased. Taking into account that the sample studied was Pd-rich (Pd/Au molar ratio ≈ 5), we conclude that the formation of a Pd3Au1 phase is favored. However, because bimetallic particles with various compositions were present after synthesis, some particles are likely to be rich in Au, which leads to the additional formation of the Au-rich Pd1Au3 phase. Note that a homogeneous distribution of the elements in the particles would lead to the formation of the Pd-rich Pd3Au1 phase, in which Au−Au neighbors are nonexistent (NAuAu = 0). As discussed above, Au-rich regions have a pronounced impact on the overall coordination numbers. Changes in the coordination numbers of Pd were less noticeable than for Au, because the reconstruction in the mainly Pd-rich particles does not cause significant changes in the environment of the Pd atoms. The high concentration of the Pd3Au1 phase was also well observed in the interatomic distances of Pd−Au and Au− Pd at temperatures above 150 °C. The Pd−Au, or equivalently Au−Pd, distances were closer to the Pd−Pd distance, which is attributed to the low concentration of Au within the Pd3Au1 phase. Because only a low concentration of Au−Au neighbors is expected to exist in the bimetallic particles mainly consisting of an ordered Pd3Au1 phase, the lattice distortion induced by Au located within the Pd lattice occurs only to a minor extent. Metal Clusters Present after Reaction. Combining the analysis of the EXAFS data with modeling of the local arrangement of the atoms inside the bimetallic particle confirms the reconstruction of the bimetallic metal phase under reaction conditions at 150 °C. All samples showed the same ratio for the Au neighbors (NAuPd/NAuAu), approximately 2.0, which is concluded to be caused by bimetallic particles with ordered the Pd1Au1 morphology. Further restructuring was not observed as soon as the bimetallic particles reached a composition of Pd1Au1 because this composition is thermodynamically stable at temperatures below 150 °C.

concentration of Pd was significantly enhanced (compare the intensities of orange and yellow peaks with respect to red peaks in Figure 9a,b). Samples after reaction at 150 °C showed a similar surface composition for the Au-rich phase (represented by “Au next to Pd”/“Au next to Au”), independent of the Pd/ Au ratio applied. A similar behavior was found for samples after 180 °C, however, with lower “Au next to Pd”/“Au next to Au” ratios. The ratio of bands assigned to CO adsorbed on Pd (“Pd next to Au”/“Pd next to Pd”) was highly affected by the presence of residual K, which remained on the metal particles after reaction and caused a pronounced broadening of the bands of CO adsorbed on Pd.30,31 Thus, values of “Pd next to Au”/“Pd next to Pd” cannot be used as a direct measure for the degree of surface intermixing between Pd and Au. Less K was found on samples after reaction at 180 °C as the features representing CO on K+ (bands in the region 2156−2185 cm−1) are less pronounced in Figure 9b than in Figure 9a.32,33 An additional measure for the degree of alloying is the shift in the frequency of the bands of CO adsorbed on “Au next to Pd” and “Pd next to Au” compared to the monometallic samples. In general, the variations in the wavenumbers induced by electronic interactions were less pronounced for samples aged at 180 °C (i.e.,10 cm−1) compared to the ones aged at 150 °C (>15 cm−1), which indicates that the bimetallic surface is less intermixed for samples treated at elevated temperatures. The shifts found for Pd were not only affected by the interaction between Pd and Au, but also by the presence of residual K and, thus, cannot be directly assigned to a certain degree of surface intermixing. In general, the effect of residual K was more pronounced for samples with higher Pd contents, because these samples exhibit a higher fraction of dispersed Pd in close contact to K. Leaching of Pd from the Bimetallic Particles. To verify that Pd was leached from the bimetallic particles in the presence of acetic acid,10 a fully reduced catalyst (Pd/Au 5.0) was dispersed in acetic acid at 25 °C and stirred for 72 h. The AcOH solution and the catalyst before and after the AcOH treatment were analyzed using AAS; the results are compiled in Table 7. The acetic acid was removed by centrifugation and had a yellow-orange color. AAS analysis showed that only Pd was removed from the sample, whereas the concentration of Au found in the AcOH solution after the treatment was negligible and the deviations in Au concentration of the catalysts were within the detection limits.



DISCUSSION Temperature-Dependent Phase Separation in Bimetallic Particles. In situ XRD indicated that a pronounced phase separation toward a Pd-rich and a Au-rich phase occurs in He at approximately 160 °C. This phase separation was more pronounced for samples with high Pd concentrations. During cooling to room temperature, bimetallic Pd−Au phases with compositions close to 3:1, 1:1, and 1:3 were formed in accordance with the Pd−Au phase diagram.7 The less pronounced phase separation for Au-rich samples can be attributed to the higher Au concentration in the bimetallic particles, which limits the formation of the Pd-rich Pd3Au1 phase, and thus, the separation toward a Pd-rich phase is not as visible by XRD as for the Pd-rich samples. Before reaction, various phase compositions of the bimetallic particles were present. Under reaction conditions at 150 °C, all samples self-assemble during the first 20−30 h on stream I

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

The ratio “Au next to Pd”/“Au next to Au” was used as a measure for the degree of surface intermixing between Pd and Au. High values indicate high degrees of surface intermixing, whereas phase separation toward Pd- and Au-rich phases will lead to a decreasing concentration of Pd and Au atoms in close neighborhoods. The low values found for all samples after reaction at 180 °C clearly indicate a low degree of intermixing. The values for the ratio “Au next to Pd”/“Au next to Au” were similar for all bimetallic Pd/Au catalysts studied at a given reaction temperature, which implies that the Au-containing surfaces were similar after reaction. At a reaction temperature of 150 °C, only equimolar bimetallic particles were present. At 180 °C, similar degrees of intermixing independent of the Pd/ Au ratio were observed, which can be related to the formation of the Pd1Au3 phase as identified by XRD. At elevated reaction temperatures, an enhanced leaching of Pd from the bimetallic particles under the reaction conditions was observed, which explains the differences in deactivation behavior at different temperatures as reported, for example, by Brückner et al.37 However, because we could not follow the reconstruction under the reaction conditions in situ by XRD and XAS, we cannot fully exclude an influence of the atmosphere present during characterization.

The concentration of the dispersed Pd phase, obtained from the Pd-derived coordination numbers (see Figure 3), is in good agreement with those calculated on the basis of the chemical composition. Both XRD and EXAFS analyses support the hypothesis that Au is fully incorporated into the equimolar bimetallic phase. Restructuring by dissolution of Pd from the particles and the formation of palladium acetate was confirmed by treating a reduced catalyst in AcOH for 72 h. Approximately 60% of the Pd was dissolved, whereas Au was not removed from the reduced sample. A pronounced AcOH-induced segregation of Pd to the bimetallic surface was also suggested by Owens et al.34 in a study of the adsorption and thermal decomposition of acetic acid on Au deposited on the Pd(111) surface. Similarly, the segregation effect was reported for the adsorption of formic acid on PdAu/γ-Al2O3, for which the surface concentration of Pd increased after contact with 2.2 vol % gaseous formic acid.35,36 Temperature-Dependent Self-Organization during Vinyl Acetate Formation. It was previously shown by us10 that the initiation period observed for bimetallic Pd−Au/SiO2 samples is related to the dynamic self-organization of the bimetallic particles at reaction temperatures close to 150 °C.36 At higher temperatures, the bimetallic catalyst segregates toward a Pd-rich phase and a Au-rich phase, leading to Pdrich and Au-rich surfaces. The XRD profiles of samples aged at 180 °C showed the presence of three alloy phases after reaction, whereas for samples aged at 150 °C, only one distinct Pd1Au1 phase was observed. Temperature-dependent in situ XRD and in situ XAS studies indicated that the phase separation of Pd1Au1 started above 160 °C, which is in line with the fact that that the Pd1Au1 phase is not stable during reactions at 180 °C. However, all three phases were found after reaction, which indicates that the Pd1Au1 phase is preferentially formed by reactant-induced self-assembly during VAM synthesis. Both the Pd-rich and Au-rich phases were found to have lower concentrations of Pd compared with the respective sample after thermal treatment in an inert atmosphere. This is attributed to the leaching of Pd from the bimetallic particle by acetic acid present under the reaction conditions (through the formation of palladium acetate). Because of the higher concentration of Pd within the Pd-rich phase, the formation of palladium acetate is favored and leads to a depletion of Pd in the Pd3Au1 phase. For the Pd1Au3, already a small extent of Pd leaching leads to a pronounced decrease in the overall concentration of Pd within the Au-rich domain. CO adsorption also confirmed the phase separation at higher reaction temperatures. For samples treated at 180 °C, an up-to5-fold increase in the ratio between linear and bridged adsorbed CO was found, pointing to the formation of a Pd-rich phase and a Au-rich phase at higher temperatures. The surface of the Pd-rich phase was significantly less intermixed than the surface of an equimolar phase. Because multifold adsorption of CO on Pd is favored for large Pd ensembles, the high ratio of linear/ multifold CO indicates the formation of a Pd-rich phase at 180 °C, which must be accompanied by the formation of a Au-rich phase, resulting from the phase separation discussed above. The low ratios for CO adsorbed on Au in close proximity to Pd (“Au next to Pd”) and CO adsorbed on Au without Pd neighbors (“Au next to Au”) (see Table 6) for samples after reaction at 180 °C also support the assumption of phase separation at higher temperatures.



CONCLUSIONS



ASSOCIATED CONTENT

The changes in the composition of the metallic phases present in a series of bimetallic Pd−Au catalysts supported on amorphous SiO2 with various PdxAuy compositions induced by the reactant atmosphere were studied by in situ XAS, XRD, and low-temperature CO adsorption. Thermal treatment in H2−He atmosphere showed a pronounced phase separation toward Pd- and Au-rich phases above 160 °C. Slow cooling in an inert atmosphere led to the (re)formation of three bimetallic phases, namely, Pd3Au1, Pd1Au1, and Pd1Au3. Dynamic selfassembly under an atmosphere at 150 °C led to the formation of a Pd1Au1 phase, independent of the initial Pd/Au ratio present in the catalyst after synthesis. The remaining Pd formed a highly dispersed monometallic Pd phase. For samples aged at 180 °C under the reaction conditions, the presence of the Pd1Au1 phase was more pronounced. Elevated reaction temperature promoted the surface enrichment of Pd, which is hypothesized to lead to the strong temperature dependence of the deactivation rates observed in catalytic studies by selective leaching of Pd.

S Supporting Information *

Additional information about the EXAFS analysis, including examples of multiedge fits, additional IR spectra of CO adsorbed on bimetallic samples and temperature-resolved XRD patterns. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel.: 0049-89-28913540. Fax: (+)49-89-289-13544. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. J

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C Notes

(19) Vaccari, M.; Fornasini, P. Einstein and Debye Models for EXAFS Parallel and Perpendicular Mean-Square Relative Displacements. J. Synchrotron Radiat. 2006, 13, 321−325. (20) Frenkel, A. I. Applications of Extended X-ray Absorption FineStructure Spectroscopy to Studies of Bimetallic Nanoparticle Catalysts. Chem. Soc. Rev. 2012, 41, 8163−8178. (21) Bauer, M.; Bertagnolli, H. The Amplitude Reduction Factor and the Cumulant Expansion Method: Crucial Factors in the Structural Analysis of Alkoxide Precursors in Solution. J. Phys. Chem. B 2007, 111, 13756−13764. (22) Bus, E.; Miller, J. T.; Kropf, A. J.; Prins, R.; van Bokhoven, J. A. Analysis of in Situ EXAFS Data of Supported Metal Catalysts Using the Third and Fourth Cumulant. Phys. Chem. Chem. Phys. 2006, 8, 3248−3258. (23) Okube, M.; Yoshiasa, A.; Ohtaka, O.; Fukui, H.; Katayama, Y.; Utsumi, W. Anharmonicity of Gold under High-Pressure and HighTemperature. Solid State Commun. 2002, 121, 235−239. (24) Giulian, R.; Araujo, L. L.; Kluth, P.; Sprouster, D. J.; Schnohr, C. S.; Foran, G. J.; Ridgway, M. C. Temperature-Dependent EXAFS Analysis of Embedded Pt Nanocrystals. J. Phys. Condens. Mater. 2009, 21, 155302. (25) Joseph, B.; Iadecola, A.; Malavasi, L.; Saini, N. L. TemperatureDependent Local Structure of NdFeAsO1−xFx System Using Arsenic K-Edge Extended X-ray Absorption Fine Atructure. J. Phys. Condens. Mater. 2011, 23, 265701. (26) Vegard, L. Die Konstitution der Mischkristalle und die Raumfüllung der Atome. Z. Phys. 1921, 5, 17−26. (27) Wang, Y. G.; Yoon, Y.; Glezakou, V. A.; Li, J.; Rousseau, R. The Role of Reducible Oxide−Metal Cluster Charge Transfer in Catalytic Processes: New Insights on the Catalytic Mechanism of CO Oxidation on Au/TiO2 from ab Initio Molecular Dynamics. J. Am. Chem. Soc. 2013, 135, 10673−10683. (28) Liu, P.; Nørskov, J. K. Ligand and Ensemble Effects in Adsorption on Alloy Surfaces. Phys. Chem. Chem. Phys. 2001, 3, 3814− 3818. (29) Kugler, E. L.; Boudart, M. Ligand and Ensemble Effects in the Adsorption of Carbon Monoxide on Supported Palladium−Gold Alloys. J. Catal. 1979, 59, 201−210. (30) Hayden, B. E.; Robinson, A. W.; Tucker, P. M. An Infra-Red Study of CO Adsorption on Potassium-Doped Pt(110)-(1 × 2); the Long Range Potassium−CO Interaction. J. Electron. Spectrosc. 1987, 44, 297−304. (31) Angevaare, P. A. J. M.; Hendrickx, H. A. C. M.; Ponec, V. Coadsorption and Promoter Effect in the Adsorption of CO on Palladium Catalysts. J. Catal. 1988, 110, 11−17. (32) Montanari, T.; Castoldi, L.; Lietti, L.; Busca, G. Basic Catalysis and Catalysis Assisted by Basicity: FT-IR and TPD Characterization of Potassium-Doped Alumina. Appl. Catal. A: Gen. 2011, 400, 61−69. (33) Montanari, T.; Matarrese, R.; Artioli, N.; Busca, G. FT-IR Study of the Surface Redox States on Platinum−Potassium−Alumina Catalysts. Appl. Catal., B 2011, 105, 15−23. (34) Owens, T. G.; Jones, T. E.; Noakes, T. C. Q.; Bailey, P.; Baddeley, C. J. The Effects of Gold and Co-Adsorbed Carbon on the Adsorption and Thermal Decomposition of Acetic Acid on Pd{111}. J. Phys. Chem. B 2006, 110, 21152−21160. (35) Bulushev, D. A.; Beloshapkin, S.; Plyusnin, P. E.; Shubin, Y. V.; Bukhtiyarov, V. I.; Korenev, S. V.; Ross, J. R. H. Vapour Phase Formic Acid Decomposition over PdAu/γ-Al2O3 Catalysts: Effect of Composition of Metallic Particles. J. Catal. 2013, 299, 171−180. (36) Pohl, M.-M.; Radnik, J.; Schneider, M.; Bentrup, U.; Linke, D.; Brückner, A.; Ferguson, E. Bimetallic PdAu−KOac/SiO2 Catalysts for Vinyl Acetate Monomer (VAM) Synthesis: Insights into Deactivation under Industrial Conditions. J. Catal. 2009, 262, 314. (37) Smejkal, Q.; Linke, D.; Bentrup, U.; Pohl, M. M.; Berndt, H.; Baerns, M.; Brückner, A. Combining Accelerated Activity Tests and Catalyst Characterization: A Time-Saving Way to Study the Deactivation of Vinylacetate Catalysts. Appl. Catal. A: Gen. 2004, 268, 67−76.

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This project was financially supported by Wacker Chemie AG. Part of this research was carried out at the light source DORIS III at DESY, a member of HGF. We thank the HASYLAB staff for assistance in using beamlines X1 and C. Part of this work was performed on the DUBBLE beamline at the ESRF. We are grateful to the beamline team for their invaluable assistance. Xaver Hecht and Martin Neukamm are acknowledged for their experimental support.



REFERENCES

(1) Kumar, D.; Chen, M. S.; Goodman, D. W. Synthesis of Vinyl Acetate on Pd-Based Catalysts. Catal. Today 2007, 123, 77−85. (2) Nakamura, S.; Yasui, T. The Mechanism of the PalladiumCatalyzed Synthesis of Vinyl Acetate from Ethylene in a Heterogeneous Gas Reaction. J. Catal. 1970, 17, 366. (3) Chen, M. S.; Luo, K.; Wei, T.; Yan, Z.; Kumar, D.; Yi, C. W.; Goodman, D. W. The Nature of the Active Site for Vinyl Acetate Synthesis over Pd−Au. Catal. Today 2006, 117, 37. (4) Chen, M.; Kumar, D.; Yi, C.-W.; Goodman, D. W. The Promotional Effect of Gold in Catalysis by Palladium−Gold. Science 2005, 310, 291−293. (5) Han, Y. F.; Wang, J. H.; Kumar, D.; Yan, Z.; Goodman, D. W. A Kinetic Study of Vinyl Acetate Synthesis over Pd-Based Catalysts: Kinetics of Vinyl Acetate Synthesis over Pd−Au/SiO2 and Pd/SiO2 Catalysts. J. Catal. 2005, 232, 467. (6) Luo, K.; Wei, T.; Yi, C. W.; Axnanda, S.; Goodman, D. W. Preparation and Characterization of Silica Supported Au−Pd Model Catalysts. J. Phys. Chem. B 2005, 109, 23517. (7) Okamoto, H.; Massalski, T. The Au−Pd (Gold−Palladium) System. J. Phase Equilib. 1985, 6, 229−235. (8) Zhu, B.; Hou, M. On the Role of Mechanical Stress in the Chemical Ordering of Nanoalloys. Eur. Phys. J. D 2012, 66, 63. (9) Atanasov, I. S.; Hou, M. A Multi-Range Order Parameter for Binary Alloy Bulk Materials and Nanoparticles. Eur. Phys. J. D 2009, 52, 51−54. (10) Simson, S.; Jentys, A.; Lercher, J. A. Dynamic Self-Organization of Supported Pd/Au Catalysts During Vinyl Acetate Synthesis. J. Phys. Chem. C 2013, 117, 8161−8169. (11) Kalevaru, V. N.; Benhmid, A.; Radnik, J.; Pohl, M. M.; Bentrup, U.; Martin, A. Marked Influence of Support on the Catalytic Performance of PdSb Acetoxylation Catalysts: Effects of Pd Particle Size, Valence States, and Acidity Characteristics. J. Catal. 2007, 246, 399. (12) Crathorne, E. A.; Macgowan, D.; Morris, S. R.; Rawlinson, A. P. Application of Isotopic Transient Kinetics to Vinyl Acetate Catalysis. J. Catal. 1994, 149, 254−267. (13) Haynes, W. M. Handbook of Chemistry and Physics, 91st ed.; CRC Press: Boca Raton, FL, 2010−2011; p 2584. (14) Eberle, H.-J.; Heidenreich, R.; Weis, J. Verfahren zur Herstellung von Katalysatoren und deren Verwendung für die Gasphasenoxidation von Olefinen. International Patent Application WO2008071610 A2, 2008 (in German). (15) Newville, M. IFEFFIT: Interactive XAFS Analysis and FEFF Fitting. J. Synchrotron Radiat. 2001, 8, 322. (16) Ravel, B.; Newville, M. Athena, Artemis, Hephaestus: Data Analysis for X-ray Absorption Spectroscopy Using IFEFFIT. J. Synchrotron Radiat. 2005, 12, 537−541. (17) Via, G. H.; Drake, K. F.; Meitzner, G.; Lytle, F. W.; Sinfelt, J. H. Analysis of EXAFS Data on Bimetallic Clusters. Catal. Lett. 1990, 5, 25−33. (18) Araujo, L. L.; Kluth, P.; de M. Azevedo, G.; Ridgway, M. C. Vibrational Properties of Ge Nanocrystals Determined by EXAFS. Phys. Rev. B 2006, 74, 184102. K

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C (38) Stevenson, S. A.; Lisitsyn, A.; Knözinger, H. Adsorption of Carbon Monoxide on Manganese-promoted Rhodium/Silica Catalysts as Studied by Infrared Spectroscopy. J. Phys. Chem. 1990, 94, 1576− 1581.

L

DOI: 10.1021/jp510089s J. Phys. Chem. C XXXX, XXX, XXX−XXX