Structure and Decomposition Pathways of Vinyl Acetate on Clean and

Dec 23, 2008 - The surface chemistry of vinyl acetate monomer (VAM) on clean and ... (VAM) from acetic acid, ethylene and oxygen was discovered...
0 downloads 0 Views 2MB Size
J. Phys. Chem. C 2009, 113, 971–978

971

Structure and Decomposition Pathways of Vinyl Acetate on Clean and Oxygen-Covered Pd(100) Zhenjun Li,† Florencia Calaza,† Craig Plaisance,‡ M. Neurock,‡ and Wilfred T. Tysoe*,† Department of Chemistry and Biochemistry, and Laboratory for Surface Studies, UniVersity of WisconsinsMilwaukee, Milwaukee, Wisconsin 53211, and Department of Chemical Engineering, UniVersity of Virginia, CharlottesVille, Virginia 22904-4741 ReceiVed: July 29, 2008; ReVised Manuscript ReceiVed: October 10, 2008

The surface chemistry of vinyl acetate monomer (VAM) on clean and oxygen-covered Pd(100) is explored experimentally in ultrahigh vacuum using reflection absorption infrared spectroscopy (RAIRS), X-ray photoelectron spectroscopy (XPS), and temperature-programmed desorption (TPD), combined with density functional theory (DFT) calculations. Both RAIRS and DFT calculations show that VAM adsorbs onto clean Pd(100) with its molecular plane close to parallel to the surface with the vinyl group bonded to a single palladium atom, which decomposes by a pathway initiated either by vinyl-acetate or vinyloxy-acetyl bond scission, where the former pathway predominates. VAM adsorbs on p(2 × 2)-O/Pd(100) and c(2 × 2)-O/ Pd(100) in a geometry in which the acetate group moves away from the surface in an adsorption site in which the vinyl group is located over two palladium atoms, where this geometry is induced by a repulsion between the acetate groups and adsorbed oxygen. Decomposition on oxygen-covered surfaces is initiated exclusively by vinyl-acetate bond scission presumably since, in this case, vinyloxy-acetyl bond scission is inhibited since the acetate group is remote from the surface. Introduction The palladium-catalyzed synthesis of vinyl acetate monomer (VAM) from acetic acid, ethylene and oxygen was discovered some thirty years ago,1 and the reaction has been shown to proceed on clean Pd(111) via insertion of adsorbed ethylene into surface acetate species2 to form an acetoxyethyl-palladium intermediate, which decomposes via a β-hydride elimination reaction to yield vinyl acetate.3,4 The surface chemistry of VAM has been previously investigated on Pd(111).5 However, the most active, model gold-palladium alloy catalysts are based on (100) alloys6-8 warranting the exploration of the chemistry of VAM on this surface. In addition, since coadsorbed oxygen affects the chemistry of acetic acid9-16 and ethylene17 on palladium surfaces, and since vinyl acetate synthesis includes oxygen as a reactant, the influence of coadsorbed oxygen on the surface chemistry of VAM is also explored. This work experimentally examines the adsorption mode, structure, and reaction pathways of vinyl acetate using reflection-absorption infrared spectroscopy (RAIRS), temperature-programmed desorption (TPD) and X-ray photoelectron spectroscopy (XPS), combined with density functional theory (DFT) calculations. Total combustion products, CO2 and water, are formed in the reaction to lower the selectivity and these can arise either from the combustion of ethylene,6,18,19 acetic acid,20 or vinyl acetate. Understanding the chemistry of vinyl acetate on Pd(100) will also provide insights into the latter chemistry.

to 1200 K, or cooled to 80 K by thermal contact with a liquidnitrogen-filled reservoir. Infrared spectra were collected using a Bruker Equinox infrared spectrometer and a liquid-nitrogencooled, mercury cadmium telluride detector. The complete light path was enclosed and purged with dry, CO2-free air. Data were typically collected for 1000 scans at 4 cm-1 resolution. Temperature-programmed desorption (TPD) data and X-ray photoelectron spectra (XPS) were collected in another ultrahigh vacuum chamber that has been described in detail elsewhere22 using a heating rate of 3 K/s for TPD, where desorbing species were detected using a Dichor quadrupole mass spectrometer placed in line of sight of the crystal. XPS data were collected with a Mg KR X-ray power of 250 W, at a pass energy of 50 eV. The binding energies were calibrated using the Pd 3d5/2 feature at 334.8 eV as a standard. The Pd(100) sample was cleaned using a standard procedure that consisted of heating to 1000 K in ∼4 × 10-8 Torr of oxygen and then annealing at 1200 K in vacuo to remove any remaining oxygen and the sample cleanliness was judged by Auger spectroscopy and oxygen TPD as described previously.22 Vinyl acetate (Aldrich, 99+%) and oxygen (Praxair, 99.9%) were transferred to glass bottles and attached to the gas-handling systems of the vacuum chambers, and the vinyl acetate was further purified by several freeze-pump-thaw cycles. The cleanliness of all reactants was monitored mass spectroscopically.

Experimental Methods Infrared data were collected using a system that has been described previously.21 The sample could be resistively heated

Theoretical Methods

* To whom correspondence should be addressed. Telephone:(414) 2295222. Fax: (414) 229-5036. E-mail: [email protected]. † University of WisconsinsMilwaukee. ‡ University of Virginia.

First-principles, periodic density functional theory (DFT) calculations were carried out using the Vienna ab initio simulation program (VASP).23 The wave function was constructed from plane waves with an energy cutoff of 396 eV. Vanderbilt, ultrasoft pseudopotentials24 were used to describe the sharp features of the wave functions in the core region. The

10.1021/jp806729c CCC: $40.75  2009 American Chemical Society Published on Web 12/23/2008

972 J. Phys. Chem. C, Vol. 113, No. 3, 2009

Figure 1. Infrared spectra of various exposures of vinyl acetate adsorbed on Pd(100) at 100 K (where the exposures are given in Langmuirs adjacent to the corresponding spectrum), and after heating to various temperatures (where the temperatures are marked adjacent to the corresponding spectrum).

Perdew-Wang 9125 form of the generalized gradient approximation (GGA) was used to model the gradient corrections to the correlation and exchange energies. The wave functions and electron density were converged to within 1 × 10-6 eV and the geometry was optimized until the force on each atom was less than 0.05 eV/Å. Using tighter convergence criteria for the forces did not lead to a change in the total energy greater than 0.01 eV. A 2 × 2 × 1 Γ-centered, k-point mesh was used to sample the first Brillouin zone and the second-order Methfessel-Paxton occupation scheme26 with smearing of 0.20 eV was used to determine the occupancy of each band. The bulk palladium lattice constant was optimized yielding a value of 3.96 Å, close to the experimental bulk palladium lattice constant of 3.89 Å. In order to gauge the effect of this small difference in lattice constant on adsorbate binding energies, we calculated the heat of adsorption of π- and di-σbonded ethylene on both lattices and the resulting binding energies varied by less than 0.02 eV between the two surfaces, The metal surface was modeled using a 4 × 4 unit cell comprised of four layers of atoms with 10 Å of vacuum, which separates the slab in the z-direction. The top two layers were allowed to relax within the geometry optimization. The latter two were held fixed at their bulk lattice positions. The frequencies were determined using a finite difference approximation of the derivatives of the forces with respect to the atomic coordinates. Each atomic coordinate was displaced by ( 0.01 Å to determine the finite difference derivatives. Results RAIRS, TPD, and XPS of Clean and Oxygen-covered Pd(100). Figure 1 displays a series of infrared spectra for VAM adsorbed on clean Pd(100) at 100 K, where the corresponding desorption spectra are displayed in Figure 2. Following low VAM exposures (of 0.3 and 0.8 L, 1 L ) 1 × 10-6 Torr s, where exposures are not corrected for the ionization gauge sensitivity) at 100 K, the spectrum displays features at ∼1758,

Li et al.

Figure 2. Temperature-programmed desorption data collected using a heating rate of 3 K/s for 1.5 L of vinyl acetate adsorbed on Pd(100) at 90 K monitored at various masses where the detected masses are indicated adjacent to the corresponding spectrum.

TABLE 1: Vibrational Frequencies of 1.5 L of Vinyl Acetate Adsorbed on Pd(100), Pd(100)-p(2 × 2)-O, and Pd(100)-c(2 × 2)-O at 100 K and Their Assignments experimental frequency/cm-1 VAM/Pd VAM/Pd(100)- VAM/Pd(100)(100) p(2 × 2)-O c(2 × 2)-O 1765 1649 1431 1367 1290 1225 1198 1142 1045, 1014 958 872

1762 1649 1431 1369 1292 1228 1196 1146 1045, 1022 959 881

1767 1649 1429 1371 1292 1234 1192 1148 1047, 1026 958 883

assignment CdO stretch CH2, CH3 plus CdC stretch CH3 bend CH3 bend C-C stretch C-C-O stretch H3C-C-O stretch C-O-C stretch CH3 rock HCdCH2 out-of-plane def. dCH2 rock

1431, 1367, 1217, 1142, and 1014 cm-1, with other less intense features. Additional peaks appear as the exposure increases to ∼1.5 L, and their vibrational frequencies and assignments are summarized in Table 1, and all of the features can be assigned to undistorted, molecular vinyl acetate. Heating the sample to ∼160 K results in the recovery of the spectrum found following low VAM exposures at ∼100 K, but with more intense peaks. Vinyl acetate desorbs at ∼137 K (Figure 2), and a state with a similar desorption temperature has been found previously for Pd(111)5 (at ∼140 K), which was assigned to the desorption of vinyl acetate multilayers, and a similar assignment is made here. The assignment of the 137 K peak to the desorption of molecular vinyl acetate is confirmed by comparing the relative intensities at various masses in Figure 2 with the mass spectrometer ionizer fragmentation pattern of vinyl acetate. The infrared spectrum found on heating to ∼160 K persists at ∼188 K, with a slight increase in the frequency of the 1217 cm-1 mode, which decreases substantially in intensity on heating to ∼240 K. Molecular vinyl acetate desorbs from the surface

Pathways of Vinyl Acetate

Figure 3. C 1s X-ray photoelectron spectrum for 3.0 L of vinyl acetate adsorbed on Pd(100) at 80 K and after annealing to various temperatures, where the temperatures are indicated adjacent to the corresponding spectrum.

at ∼251 K (Figure 2), again confirmed by the relative desorption intensities at various masses, and occurs at a slightly lower temperature than that found for Pd(111) (where it desorbs at ∼256 K5), so that the diminution in intensity of the infrared features found on heating to ∼188 K (Figure 1) is ascribed partially to desorption of vinyl acetate from the Pd(100) surface. The infrared spectrum formed on annealing to ∼160 K is thus assigned to vinyl acetate adsorbed directly onto the Pd(100) surface. However, a portion of the vinyl acetate thermally decomposes as evidenced by the appearance of a feature at ∼1404 cm-1 on heating to ∼240 K, and the presence of a feature at ∼1860 cm-1 due to carbon monoxide. The 1404 cm-1 feature decreases in intensity on warming to 290 K and disappears when the sample is heated to 340 K, and this is accompanied by the growth in intensity of the CO stretching mode. Vinyl acetate decomposition results in the desorption of acetic acid (29, 44, and 45 amu) at 287 K, methane (15 and 16 amu) at ∼320 K, and CO (28 amu), and CO2 (44 amu) desorption at ∼450 K (Figure 2). The corresponding C 1s XPS data are displayed in Figure 3. Since VAM contains four chemically distinct carbon atoms, it is difficult to unequivocally assign the C 1s profile at the resolution at which these data were collected. However, based on the spectra found for acetic acid,12,16 the peak at ∼288.9 eV binding energy (BE) obtained following adsorption at 80 K is assigned to the carboxylate carbon, and the methyl group is expected to have a binding energy of ∼285 eV and therefore forms part of the more intense profile at ∼284.8 eV BE. The C 1s signals due to the vinyl carbons will be midway between these two extremes. Heating to ∼160 K, to desorb any multilayer (see Figures 1 and 2), essentially maintains the C 1s profile, except for a shift in the 288.9 eV feature to ∼288.1 eV. This profile persists up to ∼215 K, indicative of the continued presence of predominantly VAM on the surface, in accord with the RAIRS data (Figure 1). Heating to ∼305 K produces a single feature at ∼285.4 eV, due to the formation of carbon monoxide, consistent with the appearance of a CO stretching mode at ∼1950 cm-1 (Figure 1).27 This peak disappears on heating to 457 K, consistent with the desorption of CO and CO2 from the

J. Phys. Chem. C, Vol. 113, No. 3, 2009 973

Figure 4. Infrared spectra of various exposures of vinyl acetate adsorbed on p(2 × 2)-O/Pd(100) at 100 K (where the exposures are given in Langmuirs adjacent to the corresponding spectrum), and after heating to various temperatures (where the temperatures are marked adjacent to the corresponding spectrum). Shown as an inset is a depiction of the p(2 × 2)-O/Pd(100) surface.

surface (Figure 2). Only a very small amount of carbon remains on the surface after heating to 550 K. The corresponding TPD, RAIRS, and XPS data for VAM adsorbed on p(2 × 2)-O/Pd(100) are presented in Figures 4-6 respectively. Initial VAM adsorption at 80 K (at an exposure of 0.6 L) produces features at 1431, 1369, 1196, 1146, 1045, and 1022 cm-1. Addition of further vinyl acetate (at an exposure of 1.5 L) leads to an increase in the number of features on the surface and these are summarized in Table 1 and assigned to adsorbed, molecular vinyl acetate. Again, multilayer VAM desorbs at ∼137 K (Figure 5) causing the reappearance of the peaks initially found following low exposures of the surface to vinyl acetate at 80 K when the sample is heated to ∼165 K and VAM desorption is confirmed from the relative intensities of the desorption profiles at various masses. The infrared features found on heating to ∼165 and 190 K are thus assigned to vinyl acetate adsorbed on the surface and the frequencies are summarized in Table 2. Note that the infrared spectrum, in particular the relative intensities of the features on the p(2 × 2)-O/Pd(100) surface, are different from those found on clean Pd(100) (Figure 1) indicating that coadsorbed oxygen affects the adsorption geometry of vinyl acetate. In particular, VAM on Pd(100) (Figure 1) produces a feature at 1758 cm-1 due to the CdO stretch and an intense peak at 1217 cm-1 due to the H3C-C-O stretching mode. In contrast, on the oxygen-covered surfaces, the CdO stretching mode is substantially weaker than on the clean surface and the most intense peaks are at ∼1196 cm-1 (H3C-C-O stretch) and 1137 cm-1 (C-O-C stretch) (Figures 4 and 7). These features persist on heating the sample to ∼190 K but disappear on heating to 245 K. However, little vinyl acetate desorbs from the surface since the 43 amu signal (the largest fragment of vinyl acetate) is weak above 200 K. Thus, a large proportion of the vinyl acetate decomposes to yield an intense feature at 1410 cm-1, also yielding a small CO peak at ∼1900 cm-1 (Figure 4). The presence of oxygen on the surface also affects the decomposition pathways compared to

974 J. Phys. Chem. C, Vol. 113, No. 3, 2009

Li et al. TABLE 2: Vibrational Frequencies of a Vinyl Acetate Monolayer Formed on Pd(100)-p(2 × 2)-O by Adsorbing VAM at 100 K and Heating to ∼165 K Compared with the Frequencies Calculated Using Density Function Theory and Their Assignments VAM/Pd(100)-p(2 × 2)-O experimental frequency/cm-1 1763 (weak) 1431 1369 1196 (intense) 1137 (intense) 1043 1015 951 900 860

Figure 5. Temperature-programmed desorption data collected using a heating rate of 3 K/s for 1.5 L of vinyl acetate adsorbed on p(2 × 2)-O/Pd(100) at 90 K monitored at various masses where the detected masses are indicated adjacent to the corresponding spectrum.

theoretical frequency/cm-1

assignment

1746 1472, 1454 1393, 1384 1314 1164 1140 1127 1064 1012, 1008 938 901 848

CdO stretch CH3 bend CH3 bend HCdCH2 twist H3C-C-O stretch HCdCH2 bend C-O-C stretch CH3 rock dCH2 wag HCdCH2 out-of-plane def. HCdCH2 out-of-plane def. dCH2 rock

presence of vinyl acetate on the surface as indicated by the RAIRS (Figure 4) and TPD (Figure 5) results. Heating to 225 and 263 K produces a doublet with relatively equally intense C 1s features at ∼286 and 283.8 eV BE. These disappear on heating to ∼340 K to produce a single feature at ∼284.6 eV binding energy assigned to the presence of CO,27 and this assignment is confirmed by the disappearance of this feature on heating to 505 K, due to the desorption of CO at 450 K (Figure 5). The 286 and 283.8 eV BE features thus appear to be associated with the presence of the intense 1410 cm-1 vibrational mode in RAIRS (Figure 7). Again, only a small residual carbon signal remains after heating to 505 K. The corresponding results for VAM on the c(2 × 2)-O/ Pd(100) surface are displayed in Figures 7 (RAIRS), 8 (TPD), and 9 (XPS). The chemistry on the surface with a higher oxygen coverage is similar to that found for p(2 × 2)-O/Pd(100), but

Figure 6. C 1s X-ray photoelectron spectrum for 3.0 L of vinyl acetate adsorbed on p(2 × 2)-O/Pd(100) at 80 K and after annealing to various temperatures, where the temperatures are indicated adjacent to the corresponding spectrum.

clean Pd(100) (Figure 5). Water desorption is found at ∼260 K, some methane desorption at 234 and 318 K, but substantially less than that from clean Pd(100) (Figure 2), as well as a peak due to carbon dioxide at 370 K. Finally CO desorbs at ∼450 K. The C 1s XPS results show peaks at 288.3 and 284.7 eV binding energies following adsorption at 80 K due to condensed vinyl acetate (Figure 6). Heating to ∼170 K shifts these peaks slightly to lower binding energies, consistent with the continued

Figure 7. Infrared spectra of various exposures of vinyl acetate adsorbed on c(2 × 2)-O/Pd(100) at 100 K (where the exposures are given in langmuirs adjacent to the corresponding spectrum) and after heating to various temperatures (where the temperatures are marked adjacent to the corresponding spectrum). Shown as an inset is a depiction of the c(2 × 2)-O/Pd(100) surface.

Pathways of Vinyl Acetate

J. Phys. Chem. C, Vol. 113, No. 3, 2009 975 TABLE 3: Vibrational Frequencies of a Vinyl Acetate Monolayer on Pd(100)-c(2 × 2)-O by Adsorbing VAM at 100 K and Heating to ∼165 K Compared with the Frequencies Calculated Using Density Function Theory and Their Assignments VAM/Pd(100)-c(2 × 2)-O

Figure 8. Temperature-programmed desorption data collected using a heating rate of 3 K/s for 1.5 L of vinyl acetate adsorbed on c(2 × 2)-O/Pd(100) at 90 K monitored at various masses where the detected masses are indicated adjacent to the corresponding spectrum.

Figure 9. C 1s X-ray photoelectron spectrum for 3.0 L of vinyl acetate adsorbed on c(2 × 2)-O/Pd(100) at 80 K and after annealing to various temperatures, where the temperatures are indicated adjacent to the corresponding spectrum.

with some notable differences. Low vinyl acetate exposures (∼0.6 L) yield an infrared spectrum with dominant features at 1192 and 1137 cm-1, similar to those found on the p(2 × 2)O/Pd(100) surface, along with other weaker peaks. Further VAM exposure produces a series of intense peaks due to multilayer adsorption that are also summarized in Table 1 along with their assignments. The VAM multilayer desorbs at ∼137 K (Figure 8) so that the infrared spectrum initially found at low exposures

experimental frequency/cm-1

theoretical frequency/cm-1

assignment

1765 (weak) 1433 1371 1192 (intense) 1137 (intense) 1040 1013 955 900-920

1746 1473, 1422, 1418 1374, 1352 1185, 1158 1137 1040 995 955, 953 907 839

CdO stretch CH3 bend CH3 bend H3C-C-O stretch C-O-C stretch CH3 rock H3CsCdO stretch dCH2 wag HCdCH2 out-of-plane def. dCH2 rock

is regained on heating to ∼165 K (Figure 7) and assigned to vinyl acetate adsorbed onto the surface and the frequencies and their assignments are summarized in Table 3. Again, note that this connotes a different geometry of vinyl acetate adsorbed on oxygen-covered Pd(100) from that found on the clean surface. An intense ∼1410 cm-1 feature forms on heating to ∼230 K and persists up to ∼340 K (Figure 7) and disappears completely on heating to ∼380 K (data not shown). In this case, a small amount of methane desorbs at ∼230 K, some water at ∼260 K and a sharp feature appears at ∼350 K containing fragments at all masses except 26, 27, 30, and 31 amu, and some CO desorbs at ∼460 K (Figure 8). The C 1s XP data for vinyl acetate on c(2 × 2)-O/Pd(100) are displayed in Figure 9. The spectrum exhibits features at 288.6 and 284.7 eV BE following adsorption at 80 K, assigned to molecular vinyl acetate. These shift slightly on heating to ∼170 K due to vinyl acetate adsorbed directly on the oxygen-covered surface. Heating to 250 and 300 K yields C 1s features at ∼286.8 and 284 eV binding energies, coincident with the appearance of the 1410 cm-1 peak detected in infrared spectroscopy (Figure 8). Heating to ∼380 K yields a feature at ∼286 eV binding energy due to CO on the surface27 and an additional peak at 283.8 eV that persists on heating to 550 K. This indicates that additional coadsorbed oxygen leads to a larger final carbon coverage, after heating to high temperatures, than on clean Pd(100). DFT Calculations of VAM on Clean and Oxygen-covered Pd(100). The structures on vinyl acetate and the vibrational frequencies were calculated for adsorption onto clean Pd(100), p(2 × 2)-O/Pd(100) and c(2 × 2)-O/Pd(100). Top and side views of vinyl acetate adsorbed onto the clean surface are displayed in Figure 10. This reveals that the molecule bonds predominantly via the vinyl group, but with the acetate group also oriented close to parallel to the surface in a structure that is very similar to that found on Pd(111).5 The structures formed both on the p(2 × 2)-O/Pd(100) and c(2 × 2)-O/Pd(100) surfaces are different from that on clean Pd(100) and top and side views of the calculated structures are depicted in Figures 11 and 12, respectively. In both cases, repulsion between the surface oxygen atoms and the acetate group cause it to move away from the surface, while the vinyl group is now located above two palladium atoms in a di-σ geometry, where the carbon-carbon double bond is parallel to the Pd(100) surface. In addition, the oxygen atom on the c(2 × 2) surface is considerably displaced from the 4-fold hollow site to allow the vinyl acetate to occupy two atop palladium sites. The corresponding calculated vibrational frequencies are summarized,

976 J. Phys. Chem. C, Vol. 113, No. 3, 2009

Li et al.

Figure 10. Top and sides views of the most stable structure of vinyl acetate adsorbed on Pd(100) calculated by density functional theory.

Figure 11. Top and sides views of the most stable structure of vinyl acetate adsorbed on p(2 × 2)-O/Pd(100) calculated by density functional theory.

along with the experimental values, in Table 2 for VAM on p(2 × 2)-O/Pd(100) and Table 3 for VAM on c(2 × 2)-O/ Pd(100).

geometry can be confirmed for VAM on Pd(100) using the surface selection rules in RAIRS.28 For example, the intensity of the CdO stretching mode (1758 cm-1) of the vinyl acetate monolayer is considerably attenuated on a surfaces heated to ∼188 K compared to that for the multilayer (at 1765 cm-1) indicating that the CdO bond is oriented close to parallel to the surface, while the CdC stretching mode (1649 cm-1) is completely absent, indicating that the vinyl CdC bond is also parallel to the surface. The methyl asymmetric deformation mode is relatively weaker in the monolayer (1431 cm-1) than for the multilayer, while the CH3 rocking mode (1013 cm-1) remains relatively intense, implying that the C-CH3 bond is oriented close to parallel to the surface. The intense 1245 cm-1 mode (H3C-C-O stretch) shifts to ∼1217 cm-1 so that the 1225 cm-1 mode is assigned to the multilayer, while the 1217 cm-1 mode is due to the vinyl acetate monolayer. However, the decomposition pathways for VAM on Pd(111) and Pd(100) appear to be somewhat different. On Pd(111), two distinct, initial thermal decomposition pathways are detected, both initiated by C-O bond scission.5 One pathway results in the formation of vinyloxy and acetyl species, while the second forms vinyl and acetate species (the latter pathway corresponding to the reverse of the proposed Moiseev pathway for the formation of vinyl acetate29,30). On Pd(111), the former pathway is evidenced by the appearance of infrared features at 1578 and 1087 cm-1, while the occurrence of the latter pathway is indicated by the presence of an intense feature at ∼1402 cm-1 due to the formation of η2-acetate species. There is no evidence

Discussion Geometry and Decomposition of Vinyl Acetate on Clean Pd(100). The infrared data indicate that multilayers of vinyl acetate desorb from Pd(100) at ∼137 K and is confirmed by comparing the VAM fragmentation pattern with the desorption intensities at various masses. This temperature is identical to that found on Pd(111)5 and also from both oxygen-covered Pd(100) surfaces (Figures 5 and 8). The resulting infrared spectrum of a monolayer of vinyl acetate formed on Pd(100) after annealing to ∼160 K (Figure 1) is characterized by its most intense feature at 1217 cm-1, with weaker peaks at 1758, 1431, 1397, and 1013 cm-1, with the presence of other, less distinct features. This spectrum is strikingly similar to the corresponding spectrum of a monolayer of vinyl acetate on Pd(111).5 In that case, the structure of vinyl acetate on a Pd(111) surface was calculated using DFT and yielded an experimental heat of adsorption in excellent agreement with the measured value.5 In this case, VAM adsorbed in a structure in which the plane of the molecule is oriented rather parallel to the surface, in particular with both the vinyl (CdC) and acetate (OCO) groups lying parallel to the surface and a similar structure is calculated on the Pd(100) surface (Figure 10). The similarity of the spectrum of vinyl acetate on Pd(100) to that measured on Pd(111) confirms a similar geometry on both surfaces. This

Pathways of Vinyl Acetate

Figure 12. Top and sides views of the most stable structure of vinyl acetate adsorbed on c(2 × 2)-O/Pd(100) calculated by density functional theory.

for any feature at 1087 cm-1, although there may be some weak intensity at ∼1575 cm-1. There is, however, a very clear feature that appears at 1404 cm-1, assigned to an η2-acetate species indicating that adsorbed VAM decomposes predominantly by cleavage of the vinyl-acetate bond on Pd(100). On Pd(111), the resulting vinyl group undergoes a facile reaction to form an ethylidyne species, which yields a sharp peak at 1333 cm-1, not seen on Pd(100). Vinyl species on Pd(111) hydrogenate to ethylene below 300 K,31 but no 28 amu intensity is observed that could be assigned to ethylene. This implies that the reactively formed vinyl species thermally decompose on Pd(100). Acetate species formed on clean Pd(100) from acetic acid rehydrogenate to acetic acid at ∼315 K in TPD, and also thermally decomposes to yield hydrogen, CO and CO2.9-16 The data in Figure 2 indicate that a portion of the adsorbed vinyl acetate desorbs at ∼251 K, and the 287 K acetic acid desorption feature is assigned to the rehydrogenation of η2-acetate species, based on the results found for acetic acid on Pd(100).16 The evolution of hydrogen, CO and CO2 is also found (Figure 2), as well as an intense methane desorption feature at ∼320 K. This is assumed to form by hydrogenation of adsorbed methyl species. This occurs on Pd(111) in two states at ∼190 and 304 K.32 It is apparent that at least a portion of the methane arises from hydrogenation of fragments formed by the decomposition of vinyl species although some carbon does remain on the surface after the sample has been heated to 550 K (Figure 3). The XPS results (Figure 3) are in accord with these conclusions, where the intensity of the C 1s profile decreases when

J. Phys. Chem. C, Vol. 113, No. 3, 2009 977 the sample is heated from 80 to 160 K due to multilayer vinyl acetate desorption. Heating to ∼264 K yields a spectrum very similar to that found for η2-acetate formed from acetic acid on Pd(100),16 and the C 1s features at 305 and 376 K BE, due to adsorbed CO,27 are consistent with the RAIRS (Figure 1) and TPD (Figure 2) results. Geometry and Decomposition of Vinyl Acetate on OxygenCovered Pd(100). The adsorption of oxygen on Pd(100) has been studied extensively33-35 and forms two distinct, ordered structures exhibiting p(2 × 2) and c(2 × 2) LEED patterns. Accordingly, the adsorption of VAM was studied on these p(2 × 2)-O/Pd(100) and c(2 × 2)-O/Pd(100) surfaces. The surface oxygen structures are depicted schematically as insets in Figures 4 and 7, respectively, where the oxygen occupies 4-fold hollow sites in both cases. The infrared data indicate that multilayers of vinyl acetate form on both oxygen-covered surface (see Table 1 and Figures 4 and 7) and desorb at ∼137 K (Figures 5 and 8), a temperature identical to that found on Pd(111)5 and Pd(100) (Figure 2). The intensities of the features in the infrared spectrum of vinyl acetate adsorbed on the oxygen-covered surfaces, however, are different from those found on clean palladium, while the vibrational frequencies are very similar and are summarized in Tables 2 and 3, along with the values calculated by density functional theory, and their assignments. On clean palladium, adsorbed VAM is characterized by an intense feature at ∼1217 cm-1 with less intense peaks at 1758, 1367, and 1013 cm-1 (Figure 1),5 as discussed above. The frequencies for VAM on the oxygen-covered surfaces are summarized in Tables 2 and 3 and are similar to those for the multilayer and for VAM adsorbed on the clean metal surfaces. This indicates that vinyl acetate is still present on the surface and this is in accord with the XPS data for VAM adsorbed on oxygen-covered surfaces (Figures 6 and 9). This suggests that vinyl acetate adsorbs with a different geometry on oxygen-covered surfaces than on clean palladium. The calculated geometries on both oxygen-covered surfaces are very similar to each other with the rehybridized vinyl group lying parallel to the surface, but with the acetate group moved away from the surface compared to clean Pd(100). In the spectra for VAM on oxygen-covered Pd(100) (Figures 4 and 7), the CdC mode is weak indicating that, based on the infrared surface selection rules,28 the vinyl group lies parallel to the surface. The intense features at 1196 cm-1 for VAM on p(2 × 2)-O/ Pd(100) and at 1192 cm-1 on c(2 × 2)-O/Pd(100) is assigned to a H3C-C-O mode and is shifted from 1217 cm-1 on the clean surface (Figure 1). The other intense feature at 1137 cm-1 for VAM on p(2 × 2)-O/Pd(100) and c(2 × 2)-O/Pd(100) is assigned to a C-O-C stretch. The C-O-C group lies close to parallel to the surface on clean Pd(100) (Figure 10) and is therefore much weaker than the 1217 cm-1 mode on clean Pd(100) (Figure 1). However, on the oxygen-covered surfaces it is oriented such that it has a significant component of the dipole moment perpendicular to the surface, thus enhancing its intensity (Figures 11 and 12). The presence of surface oxygen also affects the vinyl acetate decomposition pathway. Heating both VAM-covered surfaces to ∼290 K yields an intense feature at ∼1404 cm-1 due an η2acetate species (Figures 4 and 7), where more CO is formed on the p(2 × 2)-O/Pd(100) than the c(2 × 2)-O/Pd(100) surface (Figures 5 and 8). This suggests that, in both cases, vinyl acetate decomposes exclusively by cleavage of the C-O bond between the vinyl and acetate groups to yield adsorbed acetate and vinyl species. In the case of vinyl acetate on the oxygen-covered surfaces, the vinyloxy-acetyl bond is more remote from the

978 J. Phys. Chem. C, Vol. 113, No. 3, 2009 surface than the vinyl-acetate bond, rendering the decomposition pathway that is initiated by vinyloxy-acetyl bond scission much less probable. In addition, no vinyl acetate is found to desorb from the monolayer, as evidenced by the absence of any 43 amu intensity (the most intense mass spectrometer ionizer fragment of VAM) at ∼250 K (where it desorbs from Pd(111)5 and Pd(100), Figure 2). Thus, it appears that essentially all of the adsorbed vinyl acetate decomposes on both oxygen-covered surfaces. However, there is no evidence from the infrared data for the presence of the corresponding vinyl species on the surface indicating that this decomposes rapidly. Since the η2acetate species is stable on the surfaces to ∼300 K on p(2 × 2)-O/Pd(100) (Figure 4) and 340 K on the c(2 × 2)-O/Pd(100) surface (Figure 7), any products desorbing below these temperatures are assigned to vinyl decomposition products. This implies that the vinyl group decomposes to ultimately form methane at ∼234 K, water at ∼260 K on p(2 × 2)-O/Pd(100) (Figure 5) and CO, since some has formed prior to acetate decomposition (Figure 4). It may also deposit some residual carbon on the surface (Figure 6). Similar arguments can be made for the decomposition of vinyl acetate on c(2 × 2)-O/Pd(100) (Figure 7), where even more η2-acetate species are formed, but with no spectral evidence for the presence of surface vinyl species. In this case, the vinyl group appears to decompose to form methane at ∼230 K and water at ∼260 K, and some CO (Figure 7), but much less than on c(2 × 2)-O/Pd(100) (compare the ∼1900 cm-1 features due to adsorbed CO in Figures 4 and 7). The acetate species decomposes on p(2 × 2)-O/Pd(100) over a temperature range from 300 to 400 K to evolve methane at ∼318 and 370 K, CO2 at ∼370 K and hydrogen. In contrast, on c(2 × 2)-O/Pd(100) it decomposes to yields a number of species in a sharp feature at ∼350 K. Such sharp desorption states have been observed previously for acetate species on other surfaces and on oxygen-covered Pd(100).11-16 This arises on acetate-crowded surfaces, where the decomposition pathway, initiated by the plane of the acetate species tilting to allow the methyl group to access the surface,9,10 is inhibited by the neighboring acetate species. However, once the process starts, space is made available for subsequent acetate species to decompose very rapidly, yielding sharp desorption states.9-16 The C 1s XPS data (Figures 6 and 9) are in accord with these conclusions. The C 1s features at 286 and 283.7 eV BE that appear at 225 and 263 K on p(2 × 2)-O/Pd(100) (Figure 6) and at 250 and 300 K on c(2 × 2)-O/Pd(100) (Figure 9) have been assigned previously to η2-acetate species in accord with the respective infrared data (Figures 5 and 8). These decompose to leave CO on the surface, and heating above 500 K results in some carbon being deposited onto the surface, where the amount of surface carbon is larger for c(2 × 2)-O/Pd(100) (Figure 9) than for p(2 × 2)-O/Pd(100) (Figure 6). Conclusions Vinyl acetate adsorbs on clean Pd(100) with the molecular plane oriented close to parallel to the surface in a structure that is similar to that found on Pd(111). It decomposes predominantly by vinyl-acetate bond scission to produce acetate species, detected by infrared spectroscopy, with a small proportion

Li et al. decomposing by vinyloxy-acetyl bond scission. The geometry of vinyl acetate on oxygen-covered Pd(100) is completely different from that on the clean surface, where the acetate group moves away from the surface due to repulsive interactions with the adsorbed oxygen. In this case, the vinyl acetate decomposes exclusively by vinyl-acetate bond scission, an effect that is ascribed to the adsorption geometry on oxygen-covered Pd(100) in which the vinyloxy-acetyl bond is sufficiently remote from the surface that it cannot react. Acknowledgment. We gratefully acknowledge support of this work by the U.S. Department of Energy, Division of Chemical Sciences, Office of Basic Energy Sciences, under Grant No. DE-FG02-92ER14289. References and Notes (1) U.S. Patent number 3658888 (1972). (2) Stacchiola, D.; Calaza, F.; Burkholder, L.; Tysoe, W. T. J. Am. Chem. Soc. 2004, 126, 15384. (3) Stacchiola, D.; Calaza, F.; Burkholder, L.; Schwabacher, A. W.; Neurock, M.; Tysoe, W. T. Angew. Chem. 2005, 44, 4572. (4) Samonos, B.; Boutry, P.; Montarnal, R. J. Catal. 1971, 23, 19. (5) Calaza, F.; Stacchiola, D.; Neurock, M.; Tysoe, W. T. Surf. Sci. 2005, 598, 263. (6) Han, Y. F.; Kumar, D.; Sivadinayarana, C.; Goodman, D. W. J. Catal. 2004, 224, 60. (7) Chen, M. S.; Kumar, D.; Yi, C.-W.; Goodman, D. W. Science 2005, 310, 291. (8) Kumar, D.; Chen, M. S.; Goodman, D. W. Catal. Today 2007, 123, 77. (9) Neurock, M. J. Catal. 2003, 216, 73. (10) Hansen, E.; Neurock, M. J. Phys. Chem. B 2001, 105, 9218. (11) Bowker, M.; Morgan, C.; Couves, J. Surf. Sci. 2004, 555, 145. (12) Haley, R. D.; Tikhov, M. S.; Lambert, R. M. Catal. Lett. 2001, 76, 125. (13) Davis, J. L.; Barteau, M. Surf. Sci. 1991, 256, 50. (14) Madix, R. J.; Falconer, J. L.; Suszko, A. M. Surf. Sci. 1976, 54, 6. (15) Aas, N.; Bowker, M. J. Chem. Soc. Faraday Trans. 1993, 89, 1249. (16) Li, Z.; Gao, F.; Tysoe, W. T. Surf. Sci. 2008, 602, 416. (17) Sock, M.; Eichler, A.; Surnev, S.; Andersen, J. N.; Klotzer, B.; Hayek, K.; Ramsey, M. G.; Netzer, F. P. Surf. Sci. 2003, 545, 122. (18) Kragten, D. D.; van Santen, R. A.; Crawford, M. K.; Provine, W. D.; Lerou, J. J. Inorg. Chem. 1999, 38, 331. (19) Nakamura, S.; Yasui, T. J. Catal. 1970, 17, 366. (20) Crathorne, E. A.; MacGowan, D.; Mouris, S. R.; Rawlinson, A. P. J. Catal. 1994, 149, 54. (21) Wu, G.; Kaltchev, M.; Tysoe, W. T. Surf. ReV. Lett. 1999, 6, 13. (22) Kaltchev, M.; Tysoe, W. T. J. Catal. 2000, 196, 40. (23) Kresse, G.; Furthmu¨ller, J. Phys. ReV. B 1996, 54, 11169. (24) Vanderbilt, D. Phys. ReV. B 1990, 41, 7892. (25) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Singh, D. J.; Fiolhais, C. Phys. ReV. B 1992, 46, 6671. (26) Methfessel, M.; Paxton, A. T. Phys. ReV. B 1989, 40, 3616. (27) Borasio, M.; Rodriguez de la Fuente, O.; Rupprechter, G.; Freund, H.-J. J. Phys. Chem. B 2005, 109, 17791. (28) Greenler, R. G. J. Chem. Phys. 1996, 44, 310. (29) Moiseev, I. I.; Vargaftic, M. N.; Syrkin, Y. L. Dokl. Akad. Nauk. S.S.S.R. 1960, 133, 377. (30) Moiseev, I. I. In Catalytic Oxidations: Principles and Applications; Sheldon, R. A., Santen, R. A., Eds.; World Scientific: Singapore, 1995; p 203. (31) Azad, S.; Kaltchev, M.; Stacchiola, D.; Wu, G.; Tysoe, W. T. J. Phys. Chem. B 2000, 104, 3107. (32) Stacchiola, D.; Wang, Y.; Tysoe, W. T. Surf. Sci. 2003, 524, 173. (33) Stuve, E. M.; Madix, R. J.; Brundle, C. R. Surf. Sci. 1984, 146, 155. (34) Zheng, G.; Altman, E. I. Surf. Sci. 2002, 504, 253. (35) Simmons, G. W.; Wang, Y.-N.; Marcos, J.; Klier, K. J. Phys. Chem. 1991, 95, 4522.

JP806729C