Ethylidyne Formation from Ethylene over Pt(111): A Mechanistic Study

12190

J. Phys. Chem. C 2010, 114, 12190–12201

Ethylidyne Formation from Ethylene over Pt(111): A Mechanistic Study from First-Principle Calculations Zhi-Jian Zhao, Lyudmila V. Moskaleva,† Hristiyan A. Aleksandrov,‡ Duygu Basaran, and Notker Ro¨sch* Department Chemie and Catalysis Research Center, Technische UniVersita¨t Mu¨nchen, 85747 Garching, Germany ReceiVed: January 21, 2010; ReVised Manuscript ReceiVed: June 7, 2010

The conversion of ethylene to ethylidyne on Pt(111) has been studied using density functional periodic slab model calculations. Similar to our recent investigation of this reaction on Pd(111), we considered the following three mechanisms: (M1) ethylene f vinyl f ethylidene f ethylidyne; (M2) ethylene f vinyl f vinylidene f ethylidyne; (M3) ethylene f ethyl f ethylidene f ethylidyne. We systematically compared three coverages of the adsorbate, 1/3, 1/4, and 1/9. Our calculations show that the typical barriers of hydrogenationdehydrogenation reactions on Pt(111), 19-92 kJ mol-1, are slightly lower than those on Pd(111), 25-120 kJ mol-1. The barriers of direct 1,2-H shift reactions are much higher, above 160 kJ mol-1. The surface coverage notably affects the relative barriers of the reactions, by up to 30 kJ mol-1. Mechanisms M1 and M2 are expected to be competitive. As the barriers of the three elementary steps of mechanism M3 are lower or comparable to the rate-limiting barriers of the other two mechanisms, M3 could be operative when a sufficient concentration of surface hydrogen is present. However, at such conditions one expects the formation of ethane rather than that of ethylidene. On the basis of our calculated vibrational frequencies and reaction barriers, we suggest that an intermediate identified in recent vibrational spectroscopic studies of the title reaction is possibly not ethylidene but perhaps vinyl. 1. Introduction Reactions of hydrocarbons over noble metal catalysts comprise a dynamic and growing field of research since the birth of modern petroleum industry. Transformations of ethylene on the Pt(111) surface serve as a simple model system for hydrogenation-dehydrogenation reactions and thus recently are attracting renewed interest, as reflected in several experimental1–4 and theoretical4–12 studies. Ethylene is known to form two types of adsorption complexes on Pt(111).13–20 π-Adsorbed ethylene has been observed at very low temperatures or in coadsorbed systems,13–15 whereas a transformation to a di-σ bonded species16–18 begins upon heating above 52 K19 at low pressures of ethylene. After the system is further heated to ∼250 K, ethylene begins to convert to ethylidyne, CH3-C≡, which is the only stable surface species at room temperature under UHV conditions.13,21 An analogous decomposition of ethylene to ethylidyne has also been observed on other transition metal surfaces such as Rh(111),22 Pd(111),23 Ir(111),24 and Ru(0001).25 Stable and hardly removable ethylidyne deposits cover the metal surface during ethylene hydrogenation over Pt(111) or Pd(111). While they do not directly participate in the reaction mechanism,26,27 recent studies on Pd(111) provided evidence that ethylidyne species probably block the sites available for ethylene adsorption and reaction with hydrogen and thus could indirectly affect the reaction kinetics of hydrogenation.28 * To whom correspondence should be addressed. E-mail: [email protected]. † Present address: Institut fu¨r Angewandte and Physikalische Chemie, Universita¨t Bremen, 28359 Bremen, Germany. ‡ Permanent address: Faculty of Chemistry, University of Sofia, 1126 Sofia, Bulgaria.

Figure 1. Elementary reaction steps of ethylene conversion to ethylidyne over Pt group metals. Arrows pointing to the right, left, and vertically down represent hydrogenation steps, dehydrogenation steps, and 1,2 H-shifts, respectively.

Although the formation of ethylidyne as the major ethylenederived species at room temperature has been known for about 30 years,21 the mechanism of the transformation of ethylene to ethylidyne has long been debated. Several plausible intermediates, including ethyl (CH3CH2),29 vinyl (CH2CH),30 vinylidene (CH2C),31 and ethylidene (CH3CH)32,33 were proposed on the Pt(111) surface. The corresponding reaction pathways connecting ethylene and ethylidyne are shown in Figure 1. For easy comparison, we assigned the labels of the various elementary

10.1021/jp100612y  2010 American Chemical Society Published on Web 06/29/2010

Ethylidyne Formation from Ethylene over Pt(111)

J. Phys. Chem. C, Vol. 114, No. 28, 2010 12191

reactions consistent with our earlier publications on the analogous reaction mechanisms on Pd(111).34,35 Later experimental research36–38 provided evidence against nearly all of these intermediates so that only ethylidene remained as the most favored candidate to date; for an overview of pertinent experimental results, see refs 39 and 40. Zaera et al. carried out extensive kinetic and spectroscopic studies38–41 related to the thermal chemistry of adsorbed ethylene on Pt(111) and suggested a two-step mechanism of ethylidyne formation (pathway c-d, Figure 1): (1) ethylene converts to ethylidene through a direct 1,2-H shift reaction and (2) ethylidene dehydrogenates to ethylidyne. However, such a mechanism could tentatively fit available experimental results on this system only by invoking a fast pre-equilibrium between ethylene and ethylidene in conjunction with a slow second step at coverages close to saturation. Recent theoretical studies provided strong evidence against the two-step mechanism above. Anghel et al.42,43 located the transition state structures of several reactions related to ethylene on the Pt(110) surface using density functional calculations. They reported a very high activation energy, 223 kJ mol-1, of the direct 1,2-H shift reaction from ethylene to ethylidene while they determined activation energies of hydrogenation-dehydrogenation reactions related to ethylene of only 25-70 kJ mol-1. Our recent studies of ethylene conversion on Pd(111) using density functional theory (DFT) methods34,35 also yielded activation energies of such hydrogenation-dehydrogenation reactions at only 25-120 kJ mol-1, whereas the direct conversion of ethylene to ethylidene seemed unlikely on Pd(111) as that barrier is very high, ∼200 kJ mol-1. Despite the 1,2-H shift step being unlikely, at present, the two-step mechanism via ethylidene seems to be generally accepted.2 Thus, one of the important goals of the present theoretical study is to show on the basis of calculated barriers that experimental results can be interpreted in a different way and other mechanisms may take place. Instead, we considered more complex three-step pathways a-e-d and a-f-g (Figure 1) that avoid direct H-shift reactions. Both pathways appear plausible on Pd(111). A very recent experimental study44 measured the activation barrier of the formation of ethylidyne on Pd(111) in close agreement with our theoretical prediction34 and supported the formation of a vinyl intermediate during the first rate-limiting step. Theoretical studies of ethylene transformations on Pt(111) are limited to the characterization of adsorbed C2Hn species,4–12 whereas the reaction path of ethylidyne formation has not been explored. Due to the well-known close similarity of catalytic properties between Pd(111) and Pt(111) surfaces, one expects the same mechanisms to be operative on Pt(111). To provide a systematic picture of ethylene conversion on the Pt(111) surface, we report here a computational study of three alternative routes for the transformation of ethylene to ethylidyne, involving ethylidene, vinyl, vinylidene, and ethyl as intermediates. We applied a plane-wave based DFT method and periodic slab models. On the basis of calculated reaction profiles we will discuss the viability of the proposed routes, suggesting a new interpretation of the existing experimental data. Throughout this work, we systematically compare the results of our calculations at different adsorbate coverage.

form of the exchange-correlation functional PW91.47 The interaction between atomic cores and electrons was described by the projector augmented wave (PAW) method.48,49 In the structure optimizations, the Brillouin zone was sampled using a Monkhorst-Pack50 mesh of 5 × 5 × 1 k points with firstorder Methfessel-Paxton smearing (smearing width of 0.15 eV).51 Subsequently, the energies were refined in single-point fashion employing a 7 × 7 × 1 k point grid. The valence wave functions were expanded in a plane-wave basis with a cutoff energy of 400 eV. We modeled the ideal Pt(111) surface by periodic five-layer slabs repeated in a supercell geometry with at least 1 nm vacuum spacing between them. The three “bottom” layers of a slab were kept fixed at the theoretical bulk-terminated geometry (Pt-Pt ) 282 pm, which is a little larger than the experimental value, 277 pm,52 due to the well-known tendency of GGA methods to overestimate bond lengths) and the two “upper” layers of Pt atoms were allowed to relax during geometry optimizations, together with the adsorbate, until the force on each atom was less than 2 × 10-4 eV/pm. The adsorbates were bound to one side of the slab models. We used three different unit cells, (3 × 3)R30°, (2 × 2), and (3 × 3), corresponding to surface coverages 1/3, 1/4, and 1/9, respectively. We located transition states only at the lowest and highest coverages, 1/9 and 1/3. Note that the saturation coverage of ethylene and ethylidyne on the Pt(111) surface was experimentally determined at 1/4.53 We nevertheless took coverage 1/3, which is higher than the saturation coverage, into consideration as the high coverage limit to be able to track the coverage dependence of energetic properties. The binding energy (BE) of an adsorbate was determined from BE ) Ead + Esub - Ead/sub, where Ead/sub is the total energy of the slab model, covered with the adsorbate in the optimized geometry; Ead and Esub are the total energies of the adsorbate in the gas phase (ground state) and of the clean substrate, respectively. Calculations on gas-phase hydrocarbon species with open shells were carried out in spin-polarized fashion. With the above definition, positive values of BE imply a release of energy or a favorable interaction. Transition states (TSs) of the reactions were determined by applying the dimer method54 or the nudged elastic band (NEB) method.55,56 In the latter case we used eight images of the system to form a discrete approximation of the path between fixed end points. The TS structures obtained in this way were further refined until the forces on atomic centers dropped below 2 × 10-4 eV/pm. For each optimized transition state structure we checked with a normal-mode analysis that only a single imaginary frequency exists. The vibrational frequencies were obtained from a normalmode analysis where the elements of the Hessian were approximated as finite differences of analytical gradients, displacing each atomic center by 1.5 pm either way along each Cartesian direction. In these calculations we used a 5 × 5 × 1 k point grid and Fermi broadening57 with a width of 0.1 eV scheme; also, we increased the energy cutoff to 520 eV and tightened the SCF energy convergence to 10-6 eV. Checks with finer k point grids showed that the resulting frequency should have a numerical precision better than 2-3 cm-1 in the selected model.

2. Models and Computational Details

3. Results and Discussion

We carried out slab-model DFT calculations with the planewave based Vienna ab initio simulation package (VASP).45,46 We used the generalized-gradient approximation (GGA) in the

Figure 1 depicts the complex reaction network, reactions a-j, that connect ethylene with ethylidyne via elementary hydrogenation/dehydrogenation or direct 1,2-H shift steps and involve

12192

J. Phys. Chem. C, Vol. 114, No. 28, 2010

Zhao et al.

TABLE 1: Calculated Structural Parameters (pm) and Binding Energies BE (kJ mol-1) of Ethylene Adsorbed on Pt(111) in Di-σ and π Modes at Different Surface Coverage θa C-C

θ di-σ ethylene

π ethylene

1/3 1/4 1/9 exp. 1/3 1/4 1/9 exp.

b

149 149b, 148c, 148i, 148l, 149l 149b 149f,g 141b 141b, 141c, 140i 141b 141j

C-Pt b

211 211b, 211c, 210i, 214l, 212l 211b, 212b 217b 217b, 218b, 218c, 222i 219b, 220b

BE b

105 114b, 117c, 100d, 101e, 122i, 109l, 127l 121b 71d,h 60b 70b, 73c, 65e, 54i 84b 40 ( 10k

a Also shown are available experimental values. b This work. c Ref 7. d Ref 4. e Ref 11. f Ref 16. g Ref 17. h Ref 59. i Ref 10. j Ref 14. k Ref 20. l Ref 8.

four intermediates (vinyl, ethylidene, vinylidene, ethyl). In this section we will describe all pertinent adsorption complexes and transition states of reactions a-j and we will discuss the structural and energetic characteristics of these elementary reactions at various coverages. 3.1. Adsorption Complexes of H and C2Hn (n ) 2-5) Species on Pt(111). Ethylene. At low temperature, previous studies identified two adsorption modes, di-σ and π mode, of ethylene on Pt(111).13–19 In the former complex, the adsorbate is attached to two adjacent Pt atoms in η2 fashion via two σ bonds at a bridge site; in the latter complex ethylene is thought to coordinate to only one Pt atom via a π donor bond at a top site. Spectroscopic studies indicated that π-bound ethylene is stable only at very low temperatures. In the temperature range 100-240 K only di-σ ethylene is monitored on the surface under UHV conditions,13,18 whereas ethylidyne forms at higher temperatures. However, at a pressure of several hundred Torr or above, the relative surface population of the three ethylene-derived species might change to some extent. For example, studies of ethylene hydrogenation on Pt(111) revealed that dehydrogenation to ethylidyne is somewhat suppressed on a surface, which is exposed to ethylene and hydrogen gas at room temperature; then adsorbed ethylene coexists with ethylidyne on the surface.58 Both di-σ and π-forms of adsorbed ethylene were detected in comparable concentrations. These experiments pointed to the role of weakly adsorbed π-bonded ethylene species as the main reactant during the hydrogenation of ethylene. Several recent theoretical studies reported structures and binding energies of the two ethylene adsorption modes,4,7,8,10,11,14,16,17,20,59 see Table 1. All calculations used computational setups similar to that of the present work, periodic slab models of 2-4 layers and the exchange-correlation functional PW91. Computational studies corroborate the experimentally observed energetic preference for di-σ ethylene, which was calculated 36-68 kJ mol-1 more stable than π-adsorbed ethylene at 1/4 coverage (Table 1).7,10,11 In accord with earlier studies, our calculations render the di-σ mode more favorable by 37-45 kJ mol-1, depending on coverage. This energy difference is larger than that calculated between di-σ and π species on Pd(111), 12-19 kJ mol-1,34 pointing to a more pronounced preference of di-σ ethylene on Pt(111) as the binding energies of the π species on the two surfaces are similar. The calculated adsorption energy of the di-σ mode on Pt(111) increases from 105 to 121 kJ mol-1 with decreasing coverage (Table 1). Comparing values at 1/4 coverage, our value of 114 kJ mol-1 nicely falls into the range of previous theoretical studies, 100-127 kJ mol-1, of BE values for di-σ ethylene on Pt(111).4,7,8,10,11 Calculations on reasonably large cluster models5 yielded binding energies of similar magnitude, ∼130 kJ mol-1.

In summary, calculations predict the binding energy, 105-121 kJ mol-1, of di-σ ethylene on Pt(111) to be 30-40 kJ mol-1 larger than on Pd(111), see ref 34 and references therein. In contrast, temperature programmed desorption (TPD) measurements suggested an adsorption energy of di-σ ethylene on Pt(111), 71 kJ mol-1 (Table 1),4,59 that is comparable to the TPD derived binding energy of di-σ ethylene on Pd(111), 68 kJ mol-1.60 Stronger binding on Pt(111) is expected according to our calculations. It is not too surprising that the quoted experimental value on Pt(111) is somewhat lower than our computational result because the PW91 GGA functional, used in the present work, is known to overestimate systematically chemisorption energies by up to 50 kJ/mol.61 The calculated adsorption energy of π ethylene on Pt(111), 60-84 kJ mol-1 (Table 1), is also somewhat overestimated compared to the experimental heat of adsorption from reflection absorption infrared spectroscopy (RAIRS), 40 ( 10 kJ mol-1.20 The adsorption sites of ethylene species on Pt(111) and the geometries of the corresponding adsorption complexes are expected to be similar to those identified on the Pd(111) surface; therefore, it was assumed in this and other studies4,7,10,16 that di-σ ethylene should also occupy a bridge site on Pt(111). Surprisingly, a study using diffuse low-energy electron diffraction (LEED)62 yielded a structure where di-σ ethylene adsorbs above a hollow site on Pt(111) and the C-C bond is tilted by ∼22° with respect to the surface plane. However, thus far, this result was not confirmed by theory. Slab-model computational studies8–10 argued in favor of the bridge structure with the C-C bond parallel to the surface, showing a 50-63 kJ mol-1 stronger binding energy compared to the hollow-site complex at 0.25 ML coverage. A near-edge X-ray adsorption fine structure (NEXAFS) study16 did not give direct information on the adsorption site of ethylene molecule, but assumed the C-C bond parallel to the surface plane. The geometries of adsorbed ethylene from published slabmodel calculations4,7,8,10,11 and those of the present work agree within 1-5 pm, Table 1. The C-C bond of di-σ ethylene is close to the bridge position, parallel to the surface plane. The calculated C-C bond length, 149 pm, agrees with the NEXAFS value, 149 ( 4 pm.16 Thus, the C-C bond is significantly longer than that of ethylene in the gas phase, 134 pm; as it approaches the bond length of ethane, 154 pm, the calculated result indicates a rehybridization of the carbon centers toward sp3 type. Note that the calculated C-C bond length is 4-5 pm longer than that determined for di-σ ethylene adsorbed on Pd(111),34 in line with the ∼30 kJ mol-1 larger binding energy of the complex on Pt(111). The C-Pt bond in the di-σ mode was calculated at 211-212 pm, up to 2 pm shorter than in the analogous complex on Pd(111).34



TABLE 2: Optimized Geometriesa (pm) and Energy Characteristics (kJ mol-1) of Intermediates in the Transformation of Ethylene to Ethylidyne at Various Coverages θ vinyl ethyl vinylidene ethylidene ethylidyne ethane

θ

C-C

1/3 1/4 1/9 1/3 1/4 1/9 1/3 1/4 1/9 1/3 1/4 1/9 1/3 1/4 1/9 1/3 1/4 1/9

147 147 147 151 151 151 141 141 141 150 150 150 149 149 149 152 153 153

208, 207, 207, 209 210 210 199, 198, 198, 207, 207, 207, 201, 202, 202,

C-Pt

BEb

208, 210 207, 209 207, 209

301 318 324 186 191 195 415 434 441 373 382 390 575 584 593 7 8 6

199, 198, 198, 207 207 207 202, 203, 202,

210, 223 212, 221 212, 221

202 203 202

a A-B, distance between atoms A and B. b Binding energy (BE) of an adsorbate (see text).

In contrast, the C-C bond of π-adsorbed ethylene is calculated at only 141 pm, indicative of sp2 hybridized carbon centers. The calculated value again agrees very well with the NEXAFS result, 141 pm.15 The C-Pt bond, 217-220 pm, is 6-9 pm longer than the corresponding bond of the di-σ complex. The coverage was found to have little effect on the geometries; bond lengths vary at most 3 pm from 1/3 to 1/9 coverage. Ethylidyne. Early LEED studies21,63 showed ethylidyne to adsorb at 3-fold hollow sites of Pt(111), with the C-C bond perpendicular to the metal surface. A more recent LEED study64 supported by density functional calculations10 suggested a slight preference of the fcc over the hcp site. Here we report a structure of ethylidyne on the favored fcc site (Table 2, Figure 2). We calculated the C-C bond at 149 pm and the C-Pt at 201-203 pm in good agreement with earlier theoretical values.10 Neither distance changes with coverage. The calculated geometry agrees very well with the structure derived from a LEED study,64 which indicates a C-C bond of 149 ( 5 pm and C-Pt bond of 191 ( 5 pm. We calculated the binding energy of ethylidyne at the 3-fold fcc site on Pt(111) at 575-593 kJ mol-1 (Table 2), using the doublet state of ethylidyne in the gas phase as reference. The binding energy increases with decreasing the coverage. The geometries of ethylidyne on Pt(111) and Pd(111) are quite similar, whereas the adsorption energies on Pt(111) were calculated 40-44 kJ mol-1 larger than those on Pd(111).34 Calorimetric measurements65 showed that 174 ( 4 kJ/mol of heat are released in the dissociative adsorption of ethylene to form ethylidyne species and surface hydrogen atoms on Pt(111). Slightly lower values, 145-157 kJ mol-1,66,67 resulted from measurements on supported Pt samples. Our calculations predict an energy change of 118-160 kJ/mol for this reaction, where more exothermic process corresponds to lower coverage. Intermediates: Vinyl, Vinylidene, Ethyl, and Ethylidene. Previous computational studies addressed the adsorption of these potential reaction intermediates on Pt(111) and Pd(111).5,8,12 In general, these species bind in a way that allows them to saturate the coordination of the unsaturated carbon centers. Vinyl (CH2CH) prefers to bind in µ3η2 fashion over a 3-fold site on Pt(111) (Figure 2). The calculated binding energy of this species

Figure 2. Optimized structures of the stable C2Hn species on Pt(111) at 1/9 coverage. Selected distances are in pm.

increases, 301-324 kJ mol-1, as the coverage decreases from 1/3 to 1/9 coverage. The C-C bond length is 147 pm, only 2 pm shorter than that of di-σ ethylene, indicating rehybridization to sp3 carbon, while the C-Pt bonds, 207-210 pm, are 1-5 pm shorter than in the latter system (Tables 1 and 2). The dehydrogenation product of vinyl, vinylidene (CH2C), also prefers to bind over a 3-fold site. It also adsorbs in a µ3η2 fashion on Pt(111) (Figure 2), with its C-C axis notably tilted from the surface normal, so that the molecular π orbitals have an improved overlap with surface d-orbitals. The calculated C-C bond length, 141 pm, is 6 pm shorter than in vinyl, indicating a higher C-C bond order. The C-Pt distances from the methylene moiety to the surface, 221-223 pm, are notably longer than the three C-Pt distances of the second C atom, 198-212 pm. We calculated the binding energy of vinylidene

12194


Zhao et al.

TABLE 3: Optimized Geometriesa (pm) and Energy Characteristics (kJ mol-1) of Transition Statesb of Ethylene Transformation to Ethylidyne at Various Coverages θ

c

TSa

CH2CH2 f CH2CH+H

TSb

CH2CH f CH3C

TSc

CH2CH2 f CH3CH

TSd

CH3CH f CH3C+H

TSe

CH2CH+H f CH3CH

TSf

CH2CH f CH2C+H

TSg

CH2C+H f CH3C

TSh-σ

CH2CH2+H f CH3CH2

TSh-π

CH2CH2+H f CH3CH2

TSi

CH3CH2 f CH3CH+H

TSj

CH3CH2+H f CH3CH3

θ

C-C

1/3 1/9 1/3 1/9 1/3 1/9 1/3 1/9 1/3 1/9 1/3 1/9 1/3 1/9 1/3 1/9 1/3 1/9 1/3 1/9 1/3 1/9

148 148 140 158 141 139 152 151 149 149 142 142 140 142 150 150 139 140 152 151 151 151

C-Hc 160 155 124, 123, 125, 126, 140 137 150 156 148 144 181 165 153 154 190 180 153 152 148 150

H-Ptc 161 162

159 137 145 151 165 167 169, 160, 165 166 161 162 162 162 159 159 161 162 169, 168,

217 281

214 216

C-Pt 207, 208, 199, 199, 198 200 202, 203, 207, 206, 199, 199, 199, 197, 210, 209, 231 222 206, 207, 229 228

209, 209, 199, 199,

220 220 275, 332 208, 213

202, 203, 208, 206, 210, 210, 199, 197, 227 228

222 226 226 225 213, 215, 216, 216,

224 225

220 217 271 233

∆Ed

Ea

51 14 -55 -51 31 30 -44 -70 -20 17 32 -15 -87 -36 -15 23 -63 -23 33 6 -17 12

92 75 195 168 229 202 29 19 65 78 69 53 58 80 75 88 37 61 88 74 67 77

e

a A-B, distance between atoms A and B in the transition state. b For the designation of the reactions and transition states, see also Figure 1. Distance that characterizes the bond that is breaking/forming during the reaction. d Reaction energy. e Activation energy.

at 415-441 kJ mol-1, where the interaction strength again increases with decreasing coverage. Ethyl (CH3CH2) adsorbs at the top site of Pt(111) (Figure 2), with the binding energy calculated at 186-195 kJ mol-1 (Table 2). This interaction is much weaker than those of vinyl or vinylidene as ethyl forms only one C-Pt bond. The calculated C-C bond length, 151 pm, is slightly longer than in di-σ ethylene, 149 pm, while the C-Pt bond, 209-210 pm, is close to that of di-σ ethylene (Tables 1 and 2). Ethylidene (CH3CH) binds at bridge sites of Pt(111), in µ fashion (Figure 2). The binding energy of this species is calculated between 373 and 390 kJ mol-1 depending on the coverage (Table 2). The calculated C-C bond length is 150 pm, manifesting the single-bond character. The two C-Pt distances were determined at 207 pm (Table 2). Coadsorbate: Hydrogen. As required by reactant and product states of the elementary reactions to be discussed below, the structures of C2Hn species with coadsorbed H atoms have also been optimized. We placed such reacting H atoms at free 3-fold hollow sites near the C2Hn species. Coadsorption of H atoms was found to have little effect on the geometries of the adsorbates; bonds changed at most by 2 pm, except for vinylidene. At high coverage, the repulsive interactions between vinylidene and H are so strong that the CH2 fragment is significantly pushed away from the surface, with the C-Pt bond, 232 pm, elongated by 9 pm with respect to the vinylidene adsorbed by itself. Overall, the adsorption geometries of the hydrocarbon fragments considered are quite close to those calculated on Pd(111). At higher coverage, vinylidene is oriented more flat on Pt(111) than on Pd(111). We attribute this difference to the slightly larger atomic radius of Pt compared to Pd; the calculated Pt-Pt and Pd-Pd distances in the bulk are 282 and 280 pm, respectively. Therefore, repulsive interactions between adsorbates at higher coverage are weaker on Pt(111). In all cases considered, we calculated adsorption energies on Pt(111) up to 53 kJ mol-1 larger than the corresponding values on Pd(111).

3.2. Transition State Structures for Ethylene Conversion to Ethylidyne. In the following, we will discuss in detail the elementary reactions of Figure 1 (a-j). We will mainly refer to the results obtained for a low surface coverage, 1/9, because in most cases coverage was found to have little effect on structures. Transition states are labeled as “TSx”, where x is one of the labels a-j of the corresponding elementary reactions in Figure 1. The corresponding reaction energies and barrier heights are summarized in Table 3. Hydrogenation Reactions. Common to all reactions to be described in this subsection is the presence of coadsorbed H atom in the initial state, which occupies a separate adsorption site. Consequently, in the initial state the lateral repulsion between the coadsorbates is more prominent at higher coverage than at lower coverage as is reflected in the coverage-dependence of the reaction energies. In general, hydrogenation reactions have been calculated more exothermic, by 30-50 kJ mol-1, as the coverage increases from 1/9 to 1/3; for dehydrogenation reactions to be described below, the situation is reversed (Table 3). Concomitantly, barriers of dehydrogenation reactions have been calculated 10-20 kJ mol-1 higher at high coverage (1/3) than at low coverage (1/9). In contrast, hydrogenation barriers are typically lower at high coverage, by 10-25 kJ mol-1, than the corresponding barriers at low coverage. Ethylene to Ethyl. As stated above, at UHV conditions only the di-σ form of adsorbed ethylene was found to be stable on Pt(111). In the presence of hydrogen, however, also π ethylene exists on the surface in notable concentrations. Nevertheless, experiments on supported Pt catalyst suggested68 that, at least in the absence of hydrogen, only di-σ complexes can convert to ethylidyne upon annealing, while π ethylene remains unaffected. On the other hand, π ethylene is believed to be the key intermediate of ethylene hydrogenation.58 We located the TS structure of the reactions of both π and di-σ ethylene to ethyl. In either case, the approaching H atom was placed at the fcc site closest to the adsorbed ethylene moiety. We start with the discussion of the hydrogenation TS for di-σ ethylene (TSh-σ, Figure 3) where the reacting H passes



Figure 4. Optimized structures of the transition states of the elementary reactions b, e, and g at 1/3 coverage. Selected distances are in pm.

Figure 3. Optimized structures of the transition states of the elementary reactions a-j (see Figure 1) at 1/9 coverage. Selected distances are in pm.

through a structure intermediate between top and bridge coordination. The H-Pt distance is shortened from originally 205 pm (at an fcc site) to 162 pm in TSh-σ, whereas the newly formed C-H bond is 154 pm (Figure 3, Table 3). The activation energy is 75 kJ mol-1 at 1/3 coverage and 88 kJ mol-1 at 1/9 coverage; both values are close to the result, 77 kJ mol-1, of an earlier calculation at 1/4 coverage.11 The hydrogenation of the π-adsorbed species proceeds via a five-member-ring transition state (TSh-π, Figure 3, Table 3) that involves C and H atoms of a forming bond, the other C atom as well as two Pt atoms. In TSh-π the C-H distance of the forming bond is 180 pm, while the C-Pt bond is totally broken, stretching from 220 pm in the reactant state to 291 pm in the transition state. The corresponding activation energy is 37 kJ mol-1 at 1/3 coverage and 61 kJ mol-1 at 1/9 coverage (Table 3). Vinyl to Ethylidene. The attacking hydrogen was placed at a neighboring 3-fold site of vinyl, in front of the C atom of the CH2 group. In the transition state TSe (Figure 3, Table 3), the reacting H atom moves toward the top site in front of the CH2 group forming one short H-Pt contact, 160 pm; concurrently, the C-H distance of the bond to be formed decreases from 270 pm in the initial state to 156 pm in TSe. At the same time, the C-Pt bond of the CH2 group elongates by 16 pm to 225 pm. The geometry of TSe is slightly different at 1/3 coverage (Figure 4, Table 3) where the 3-fold hollow site in front of the CH2 group is in the immediate vicinity of another vinyl species. Therefore, the attacking H atom can approach the CH2 group only from the side, forming two asymmetric H-Pt bonds in TSe, 169 and 217 pm. The activation barrier was calculated at 65 kJ mol-1 at 1/3 coverage and 78 kJ mol-1 at 1/9 coverage. Vinylidene to Ethylidyne. Initially, at 1/9 coverage, the attacking hydrogen atom is located in a similar position as in reaction e. In the transition state TSg (Figure 3, Table 3), this H atom moves to a top site with the H-Pt distance of 162 pm. The CH2 group, originally occupying this top position, is pushed upward so that the corresponding C-Pt distance elongates. At 1/3 coverage (Figure 4), this value changes from 232 pm in the initial state to 271 pm in TSg (Table 3). Due to a less strong repulsion between neighboring hydrocarbons, the C-Pt distance changes less at 1/9 coverage, from 223 to 233 pm. The C-H bond to be formed is still rather long in TSg, 181 pm at 1/3

12196


coverage and 165 pm at 1/9 coverage. The activation energy of this elementary step changes concomitantly, from 58 kJ mol-1 at 1/3 coverage to 80 kJ mol-1 at 1/9 coverage (Table 3). Dehydrogenation Reactions. In selected dehydrogenation reactions, the coadsorbed H atom in the final state resides at a different 3-fold site on the surface than in the corresponding initial state described above. However, previous calculations69 showed the diffusion barrier of H atoms over the Pt(111) surface to be as low as 4 kJ mol-1. Therefore, H diffusion over Pt(111) is assumed to be fast; hence, we skipped the modeling H diffusion steps. Ethylene to Vinyl. In the transition state structure TSa (Figure 3, Table 3), one of the C atoms moves from the top site to the adjoining bridge site, forming the second C-Pt bond, 220 pm. The dissociating H atom migrates over the top of the Pt atom, which forms the longer C-Pt bond (Figure 3), the C-H distance being elongated to 155 pm and the H-Pt distance shortened to 162 pm (Table 3). In the final state, the dissociated H atom arrives at a 3-fold site. The activation energy of this elementary reaction is as high as 92 kJ mol-1 at 1/3 coverage and 75 kJ mol-1 at 1/9 coverage. Vinyl to Vinylidene. In the transition state TSf (Figure 3, Table 3), the reacting C atom forms a third C-Pt bond, 215 pm, at a 3-fold hollow site, whereas the other C atom moves upward and the corresponding C-Pt distance increases to 217 pm. The dissociating H passes through a top-bound state, with the C-H bond elongated to 144 pm, and the forming H-Pt bond at 166 pm. In the final state, the dissociated H atom lands at a 3-fold site near to adsorbed vinylidene. The dehydrogenation barrier of vinyl is 69 kJ mol-1 at 1/3 coverage and 53 kJ mol-1 at 1/9 coverage. Ethyl to Ethylidene. The TS structure of TSi (Figure 3, Table 3) shows ethyl bound at a bridge site, the same as the product ethylidene. The dissociating H moves from the CH2 moiety over a top site toward a neighboring 3-fold site whereby the C-H distance elongates to 152 pm and the H-Pt distance decreases to 162 pm. Simultaneously, the reactive C center creates a second C-Pt bond, 225 pm in TSi, at a neighboring bridge site. We obtained activation barriers of 88 kJ mol-1 at 1/3 coverage and 74 kJ mol-1 at 1/9 coverage for this reaction. Ethylidene to Ethylidyne. In the course of this reaction, which proceeds via TSd (Figure 3, Table 3), the dissociating atom of the CH group migrates over a top site toward a neighboring 3-fold site. At the same time, the reactive C atom, originally adsorbed at a bridge site, moves toward a 3-fold site. In TSd, the C-H bond elongates by 27 pm, to 137 pm, the C-C bond is nearly perpendicular to the surface, hence, very close to the structure of the product complex of ethylidyne. The activation energy is 29 kJ mol-1 at 1/3 coverage and 19 kJ mol-1 at 1/9 coverage. This low barrier is quite remarkable; it suggests the dehydrogenation of ethylidene to ethylidyne to be fast and irreversible. Our result concurs with the experimental observation70 that ethylidene can readily transform to ethylidyne, already at 150 K. Hydrogen Shift Reactions. Ethylene to Ethylidene. This 1,2-H shift reaction was proposed38–41 as the first step of ethylene conversion to ethylidyne. In the corresponding transition state TSc (Figure 3, Table 3), one C-H bond stretches by 16 pm, to 126 pm, and the reactive H atom moves to the area between the two C atoms. This H, on the way to forming a new C-H bond, is located at a C-H distance of 151 pm. Concomitantly the hydride-donating C center shifts to an adjacent bridge position, but in TSc it still occupies a top site with the C-Pt bond slightly shortened by 10 to 200 pm. The hydride-accepting

Zhao et al. C center detaches from the surface, directing the C-C bond toward an upright orientation. As a peculiarity of this transition state, the migrating H atom does not interact with the metal surface; thus, the reaction is not surface mediated. This is reflected in the very high activation energy, 229 kJ mol-1 at 1/3 coverage and 202 kJ mol-1 at 1/9 coverage. Vinyl to Ethylidyne. The reactant, the transition state structure TSb, and the product occupy the same 3-fold site (Figures 2 and 3). In TSb (Figure 3, Table 3), the weakened C-H bond is elongated to 123 pm, which is close to the corresponding distance in TSc, whereas the newly formed C-H bond is 137 pm. At the same time, the hydride-accepting carbon center moves upward so that the C-C axis bends toward the surface normal, especially strong at 1/3 coverage, where the C-Pt bond in question breaks completely (Figure 4). At 1/9 coverage, the hydride-donating C atom builds a third C-Pt bond at the 3-fold site, 208 pm, whereas at 1/3 coverage this distance is still very long, 275 pm. The other two C-Pt distances of this carbon reduce by 8 pm with respect to the reactant vinyl, to 199 pm (Figure 3, Table 3). As in TSc, the migrating H atom travels far above the surface and does not seem to interact with Pt atoms. Consequently, the activation energy of this reaction is also very high, 195 kJ mol-1 at 1/3 coverage and 168 kJ mol-1 at 1/9 coverage. 3.3. Three Mechanisms of Ethylene Conversion to Ethylidyne. In the previous section, we discussed the TS structures of the various elementary steps and their energetics. We now will use this information to analyze possible reaction scenarios which connect ethylene with ethylidyne on Pt(111). The activation energies of 1,2-H shift reactions were calculated at ∼170 kJ mol-1 or larger, thus, much higher than the barriers of hydrogenation-dehydrogenation reactions. Barriers of similar height for 1,2 H-shift reactions were also reported in computational studies on Pd(111),34 Pt(110),42,43 Pt(111),71 Pt(211),71 and on Fe(100).72 Therefore, in the following discussion we mainly consider hydrogenation/dehydrogenation reactions other than the isomerization steps b and c (Figure 1). In line with our analysis of the analogous reaction mechanisms on Pd(111),34,35 there are three conceivable pathways for ethylene conversion to ethylidyne: (M1) ethylene f vinyl f ethylidene f ethylidyne; (M2) ethylene f vinyl f vinylidene f ethylidyne; (M3) ethylene f ethyl f ethylidene f ethylidyne. Some of the elementary steps are shared between different mechanisms. The reaction landscapes of the three mechanisms are depicted in Figure 5. Reaction in the Absence of Hydrogen. At UHV conditions and without hydrogen coadsorbed along with ethylene, only mechanisms M1 and M2 seem plausible because mechanism M3 involves an addition of an H atom as the first step, which would require the presence of surface hydrogen in relatively high concentration. Thus, we first will focus on mechanisms M1 and M2 at various coverages. Both M1 and M2 share the first step, the dehydrogenation of ethylene to vinyl (reaction a). Further they split into pathways e-d and f-g, whose reaction landscapes are compared in Figure 6. Experimental evidence suggests the first step of the transformation of ethylene to ethylidyne likely to be rate-limiting as no stable intermediates were detected by conventional spectroscopic techniques.38 The present calculations predict a barrier of 75-92 kJ mol-1 for reaction a. This value seems very reasonable for a reaction that occurs at ∼250-300 K and it is also in good agreement with the reported apparent barrier, 58-77 kJ mol-1, from various experiments for the overall transformation to ethylidyne.73–77 Nevertheless, according to the



Figure 6. Reaction energy profiles (kJ mol-1) of vinyl conversion to ethylidyne over Pt(111) at 1/3 coverage (black) and 1/9 coverage (red): comparison of routes via ethylidene (e-d, energies with respect to η2µ3vinyl coadsorbed with H) and via vinylidene (f-g, energies with respect to η2µ3-vinyl).

Figure 5. Reaction energy profiles (kJ mol-1) of ethylene conversion to ethylidyne over Pt(111) via various mechanisms: (a) (M1) ethylene f vinyl f ethylidene f ethylidyne; (b) (M2) ethylene f vinyl f vinylidene f ethylidyne; (c) (M3) ethylene f ethyl f ethylidene f ethylidyne (see Figure 1), at 1/3 coverage (black) and 1/9 coverage (red). In (b) and (c), one state energy is represented as a broken line to indicate a change in reference when calculating relative energies. The relative energies to the left are calculated with respect to di-σ ethylene in (a) and (b) or di-σ ethylene+H in (c) on Pt(111) at the various coverages. The relative energies to the right are shifted by the energy difference of vinyl and vinyl+H in (b) or ethylidene and ethylidene+H in (c).

results of our calculations illustrated in Figure 5 only at high surface coverage (1/3) can one definitely conclude that the first step, reaction a, is rate-limiting, whereas at low coverage there are other reaction steps with comparable barrier heights. Comparison of the two reaction landscapes of routes a-e-d and a-f-g in Figure 5 suggests that both of them could operate and compete at room temperature. Mechanisms M1 and M2

proceed with similar activation barriers, of ∼80 kJ mol-1 at low (1/9) coverage via reactions e and g, respectively, and of 65-69 kJ mol-1 via reaction steps e and f at high (1/3) coverage. Moreover, at low (1/9) coverage, these barriers are even slightly higher than the barrier of Reaction a at this coverage, 75 kJ mol-1. According to our calculated energy profile, it is thus possible that both vinyl and vinylidene may accumulate on the surface upon annealing, just before the formation of ethylidyne begins. Interestingly, kinetic studies by Zaera et al.40 indicate that the rate-limiting step of the overall conversion evidently changes with coverage. They monitored the ratio of hydrogen to deuterium released in the dehydrogenation of trideuterioethylene. At saturation, enhancement in hydrogen was observed with respect to the ratio expected from stoichiometry, whereas enhancement in deuterium was found at low coverages. Zaera et al. argued that such an effect would support their proposed mechanism with an initial pre-equilibrium between ethylene and ethylidene and a rate-limiting H-dissociation step d at saturation, whereas at low coverage, according to their proposal, the isomerization step c would become rate-limiting. That would be consistent with the enhanced deuterium desorption because H involved in the rate-limiting step is consumed faster. However, it is true in general that the observed kinetic isotope effect is consistent with a rate-limiting dehydrogenation step at high coverage and a rate-limiting step other than dehydrogenation (hydrogenation or H-shift) at low coverage. Hence, this experimental evidence is consistent with both M1 and M2. Recall that Zaera et al.38 ruled out a vinyl intermediate as precursor of ethylidyne because vinyl species were shown to convert back to ethylene before producing ethylidyne on Pt(111). In light of the present theoretical results, another look at the findings of that TPD study seems advisible. The calculated reaction landscape at 1/9 coverage (Figure 6) roughly corresponds to the coverages of vinyl iodide as estimated in ref 38 (θ ) 0.12-0.13). Indeed, on the basis of the calculated reaction barriers, one expects vinyl to form initially on the surface; then, upon annealing, it first partially converts to vinylidene (Ea ) 53 kJ mol-1), after that to ethylene (Ea ) 61 kJ mol-1), and finally to ethylidyne, which can be formed either via vinyl and ethylidene (Ea ) 78 kJ mol-1) or directly from vinylidene (Ea ) 80 kJ mol-1). In qualitative agreement with these theoretical predictions, the experimentalists identified the formation of vinylidene at ∼130 K, ethylene presumably was formed in the region of 200 K, and ethylidyne appeared after heating above 300 K.38 Still these findings do not really rule out vinyl as an

12198


Zhao et al.

TABLE 4: Calculated Vibrational Frequenciesa of Di-σ Ethylene and Ethylidyne on Pt(111) for Two Coverages θ and Assignmentb of the Normal Modes di-σ ethylene θ ) 1/3 θ ) 1/9 3099 3076 3016 3009 1418 1400 1229 1089 1024 1013 929 776 614

3080 3057 2996 2986 1421 1394 1184 1045c 1048 980 934 789 641

ethylidyne

assignment CH2 νa CH2 νa CH2 νs CH2 νs CH2 δ CH2 δ CH2 F CC ν/CH2 ω CH2 ω CH2 ω/CC ν CH2 F CH2 τ CH2 τ

experiment 3000j 2903g, 2904f, 2909e, 2920i, 2940j 1414g, 1416e, 1420j, 1430i j

1230 1042e, 1047g, 1050i 980i, 990j, 993g, 995e

θ ) 1/3 θ ) 1/9 assignment

experiment

3018 3012 2938

3032 3030 2964

CH3 νa CH3 νa CH3 νs

2950i 2884h, 2885g, 2886f, 2887d,e, 2890i, 2900j

1407 1407 1326 1107 962 961

1403 1402 1330 1095 957 957

CH3 δa CH3 Fa CH3 δs CC ν CH3 Fs CH3 Fs

1420i 1338d, 1339h, 1340e,g, 1350i, 1360j 1115g, 1118h, 1124e, 1126d, 1130i,j 900j, 980i

a Only normal modes with calculated frequencies above ∼600 cm-1 are shown. b Notations used: ν, stretching; δ, bending; τ, twisting; ω, wagging; F, rocking; a, asymmetric; s, symmetric; ip, in-plane. c The admixture of the second mode is very weak. d Ref 80. e Ref 33. f Ref 32. g Ref 81. h Ref 82. i Ref 13. j Ref 83.

intermediate in the conversion of ethylene to ethylidyne because the activation barriers for vinyl conversion to ethylidyne either via mechanisms M1 or M2 are predicted to be approximately equal to the barrier for ethylene dehydrogenation to vinyl, 75-80 kJ mol-1. Therefore, the overall transformation, being thermodynamically favorable, is quite feasible at around room temperature. Reaction under Hydrogenation Conditions. A practically relevant situation, where the formation of ethylidyne from ethylene occurs as a side reaction, is ethylene hydrogenation. At variance with UHV conditions, another possible route of ethylidyne formation could take place in the presence of a sufficient amount of hydrogen on the surface. Such a mechanism (M3) involves an initial hydrogenation of ethylene to ethyl which subsequently dehydrogenates to ethylidene and ethylidyne along route h-i-d (Figure 1). M3 shares the last step, reaction d, with mechanism M1. Figure 5c displays the calculated reaction energy profiles for the conversion of ethylene, from either π- or di-σ-bound species, to ethylidene via ethyl (M3) over Pt(111) for 1/3 and 1/9 coverages. As suggested by an in situ SFG experiment,58 π-bound ethylene is likely to be the primary intermediate in ethylene hydrogenation. This is corroborated by our calculation, which predicts a notably lower activation energy for the hydrogenation of π-ethylene than of di-σ-ethylene. At the high coverage (1/3), the former barrier to ethyl is only 37 kJ mol-1, 38 kJ mol-1 (Table 3) lower than the corresponding barrier from di-σ-ethylene. Similarly, at 1/9 coverage, a low barrier, 61 kJ mol-1, is also calculated for the hydrogenation of π- ethylene. Accordingly, over the Pt(111) surface, the hydrogenation of π-ethylene should be much easier than that of di-σ-ethylene. Let us return to the discussion of ethylene dehydrogenation to ethylidyne, Figure 5c. At high coverage (1/3) the barrier heights for elementary reaction h from π- or di-σ-ethylene and for reaction i, 37, 75, and 88 kJ mol-1, respectively, are lower than the activation barrier of ethylene dehydrogenation to vinyl, reaction a, 92 kJ mol-1 (Table 3). Hence, at this coverage mechanism M3 is energetically more favorable than M1 or M2 according to the calculations; our recent study of analogous transformations on Pd(111) led to a similar conclusion.35 At low coverage (1/9), the calculated barrier heights for reactions h and i, 61-88 kJ mol-1, are comparable with the highest barriers of mechanisms M1 and M2, 78-80 kJ mol-1 (Table 3), which suggests that M3 may at least be equally probable.

However, three further issues need to be considered in the context of the route via ethyl as intermediate. First, the existence of π-ethylene on Pt(111) requires the presence of (external) ethylene in the gas phase20 or certain coadsorbates.13–15 Second, hydrogen needs to be available in significant concentration on the surface to render the first hydrogenation step feasible. For example, with time-resolved Fourier-transform infrared spectroscopy, Frei et al.78,79 observed ethyl via peaks at 2893 and 1200 cm-1 during ethylene hydrogenation over a Pt/Al2O3 catalyst; however, these peaks were absent when the reactant mixture did not contain hydrogen gas. Third, ethyl could be further hydrogenated to ethane if H atoms were present in excess on the surface. The barrier of the latter process is only 67 kJ mol-1, that is, 21 kJ mol-1 lower than the activation energy of ethyl dehydrogenation to ethylidene (reaction i) at 1/3 coverage (Table 3), whereas at 1/9 coverage, the activation energies of the hydrogenation and the dehydrogenation of ethyl are comparable, 77 and 74 kJ mol-1, respectively. Hence, in the presence of excess hydrogen, hydrogenation to ethane (which irreversibly desorbs from the surface) should strongly compete with the dehydrogenation to ethylidene and ethylidyne via M3. Remark on a Spectroscopically Identified Transient Intermediate. Recent spectroscopic studies on the conversion of di-σ ethylene at Pt(111), with sum frequency generation (SFG)32 and RAIRS33 methods, observed a peak at 2957-2960 cm-1 and assigned it to the asymmetric stretch νa(CH3) of the CH3 group in ethylidene. The latter study covered a wider spectral range; it further identified a peak at 1387 cm-1, which developed in parallel with the 2960 cm-1 feature and was assigned to the symmetric bend δs(CH3) of ethylidene. These rather weak IR signals from transient species were beyond the sensitivity limits of previous RAIRS experiments. To provide an interpretation of these IR spectra, we carried out normal modes analyses for the stable species ethylene and ethylidyne (Table 4), and the intermediates, ethylidene, vinyl, and vinylidene (Table 5). In most cases, the vibrational frequencies of the stable species calculated below 2000 cm-1 lie within the intervals spanned by values from various experiments,13,32,33,80–83 with a maximum deviation of 62 cm-1. The calculated harmonic C-H stretching frequencies are generally too high compared to the strongly anharmonic modes in experiment. Experimentally determined anharmonicity constants of such modes are 60-70 cm-1.84,85 DFT calculations report even larger anharmonic corrections, 100-160 cm-1, for



TABLE 5: Calculated Vibrational Frequenciesa of Relevant C2Hn Intermediates (n ) 2-4) on Pt(111) for Two Coverages θ and Assignmentb of the Normal Modes µ-ethylidene θ ) 1/3 θ ) 1/9

assignment

2985 3038c 2990c 2927

2989 2966 2945 2880

CH3 νa CH3 νa/CH ν CH ν/CH3 νa CH3 νs

1406 1404 1328 1244 1049

1402 1391 1336 1226 1040

CH3 δa CH3 δa CH3 δs CH δip/CH3 Fs CC ν

η2µ3-vinyl

η-vinyl θ ) 1/3 θ ) 1/9

assignment

3114 3049 3028

3113 3058c 3029c

CH2 νa CH ν/CH2 νs CH2 νs/CH ν

1547c 1354c 1206 948

1537 1352 1192 941

CC ν/CH2 δ CH2 δ/CC ν CH2 F/CH δip CH2 F/CH δip

θ ) 1/3 θ ) 1/9

assignment

η2µ3-vinylidene θ ) 1/3 θ ) 1/9 assignment

3016 3007 2949

3023 2995 2959

CH2 νa CH ν CH2 νs

3124 3033

3133 3043

CH2 νa CH2 νs

1386 1108 1064 1013

1381 1099 1060 994

CH2 δ 1415 CH δip/CC ν 1263 CC ν/CH2 ω/CH δip 980 CH2 F

1413 1256 973

CH2 δ CC ν CH2 F

a Only normal modes with calculated frequencies above ∼950 cm-1 are shown. b Notations used: ν, stretching; δ, bending; τ, twisting; ω, wagging; F, rocking; a, asymmetric; s, symmetric; ip, in-plane. c The admixture of the second mode is very weak.

C-H stretching frequencies of organic species.86 These findings prevent a straightforward comparison of experimental and calculated frequencies in the present case. Although the results of the present work support ethylidene as one of the plausible intermediates in the course of ethylidyne formation from ethylene, we find a comment in order. On the basis of our calculated potential energy landscapes, it seems quite unlikely that ethylidene be observed by spectroscopic measurements because the calculated activation energy of ethylidene dehydrogenation to ethylidyne is too low, 19-29 kJ mol-1. In other words, detection of this intermediate is highly unlikely because of its expected short lifetime.87,88 Instead, the calculated energetics make both vinyl and vinylidene plausible candidates for the observed spectral features as both form, directly or indirectly, at low coverage of ethylidyne with activation barriers of ∼80 kJ mol-1. Our calculated vibrational frequencies provide further evidence that not only ethylidene, but also vinyl and vinylidene adsorbed on Pt(111) exhibit IR active vibrational modes with frequencies close to the experimental value 1387 cm-1. Indeed, ethylidene has a calculated CH3 asymmetric deformation mode δa(CH3) at 1391 cm-1; η2µ3vinyl and vinylidene have CH2 scissoring modes δ(CH2) at 1380 and 1413 cm-1, respectively (Table 5). Here and in the following we refer to frequencies calculated at 1/9 coverage, but in Table 5 we also provide the frequencies at coverage 1/3. Previous experiments seem to have ruled out both vinyl and vinylidene as intermediates. A SFG study by Cremer et al.89 explored the thermal evolution of acetylene on Pt(111) and reported a spectrum which they assigned to vinylidene species with four peaks at 2837, 2878, 2924, 2981 cm-1 at 125 K, whereas a signal at 2960 cm-1, associated with ethylidene, appeared only at higher temperatures. The assignment of vinylidene spectra is strongly supported by the comparison with the spectrum of an Os cluster analog.90 Another work by Cremer et al.32 ruled out also vinyl as intermediate, on the basis of their recorded low-temperature IR spectrum of vinyl iodide on Pt(111). The three peaks observed for vinyl iodide, at 2995, 3033, and 3068 cm-1 at 132 K, were all too high to account for the feature at 2960 cm-1 that appeared with low intensity only upon annealing. Despite these arguments dismissing vinyl, we note the possibility that vinyl, initially formed upon dissociation of vinyl iodide at low temperatures, in fact coordinates in η1 fashion; this other adsorption mode of vinyl exhibits a double CdC bond and adsorbs at a top site forming only one C-Pt bond. Our calculations rendered η-vinyl 58 and 75 kJ/mol less stable compared with η2µ3-vinyl at 1/3 and 1/9 coverage, respectively. After heating, this species may

convert to the more stable η2µ3-vinyl, which could cause the signal at 2960 cm-1. With its more flat orientation along the surface, η2µ3-vinyl would not exhibit as intense a symmetric stretching mode νs(CH2) as the η-vinyl species. Indeed, a highresolution electron energy loss spectroscopy (HREELS) study of vinyl iodide on Pt(111) provides evidence that supports our analysis.91 That HREELS spectra of surface vinyl species, recorded at 105-250 K, feature a C-C stretching mode ν(CC) at 1600 cm-1 that is characteristic of a double CdC bond and is inconsistent with our calculated frequencies of η2µ3-vinyl. Inspection of Table 5 shows that of all investigated C2 intermediates η-vinyl gives the closest match, 1537 cm-1, with the experiment feature just mentioned. In contrast, for η2µ3vinyl species we did not calculate any vibrational mode close to 1600 cm-1, just two C-C stretching modes ν(CC), coupled both with CH2 wagging ω(CH2) and CH in-plane bending modes δip(CH) at 1060 cm-1, or coupled only with δip(CH) at 1099 cm-1 (Table 5). In summary, our results (Table 5) tentatively suggest that not only ethylidene, but also vinyl may be responsible for the signals observed in experiments at 1387 and 2960 cm-1.32,33 The calculated energy landscape of ethylene transformations on Pt(111) favors vinyl as observed intermediate due to a low barrier for ethylidene dehydrogenation to ethylidyne. Although the results of our calculations may not be sufficiently conclusive to clarify the assignment of these bands, further spectroscopic studies of this issue will be helpful. 4. Conclusions We studied the mechanism of ethylene conversion to ethylidyne over Pt(111) by means of periodic slab model density function calculations. We explored three possible pathways: (i) via vinyl and ethylidene (M1), (ii) via vinyl and vinylidene (M2), and (iii) via ethyl and ethylidene (M3), and we compared three surface coverages, 1/3, 1/4, and 1/9. All hydrogenation-dehydrogenation steps involved on Pt(111) have calculated barrier heights in the range 19-92 kJ mol-1, while this range was slightly higher on Pd(111), 25-120 kJ mol-1.34,35 The lower rate-limiting activation barriers on Pt(111) are in line with the experimentally observed faster reaction for this system.92 Similarly to our findings in the case of the analogous reaction over Pd(111), mechanisms M1 and M2 are likely at work over Pt(111), in the absence of coadsorbed hydrogen. Both these mechanisms involve the dehydrogenation of ethylene to vinyl as the first step, followed by hydrogenation to ethylidene and dehydrogenation to ethylidyne (M1) or by a

12200


further dehydrogenation to vinylidene and rehydrogenation to ethylidyne (M2). Whereas on Pd(111) the first step of this transformation likely is rate-limiting, with an activation barrier 100-117 kJ mol-1, this is not necessarily the case on Pt(111). On the basis of the barrier heights calculated in the present work only at the limit of high surface coverage (1/3) can one definitely conclude that the first step, with an activation barrier 92 kJ mol-1 is rate-limiting, whereas at low coverage other reaction steps exhibit comparable barriers, ∼80 kJ mol-1. Thus, the surface coverage notably affects the relative barriers of the reactions on Pt(111). At both coverages studied M1 and M2 likely compete on the surface. At variance with M1 and M2, mechanism M3 starts with adding a H atom to ethylene and thus requires a notable concentration of hydrogen on the surface. Analysis of the activation energies shows that the barrier heights of the three elementary reactions of M3 are lower or comparable to the ratelimiting barriers as M1 and M2. Therefore, if enough hydrogen is preadsorbed or generated on the surface, intermediate ethyl is more likely to fully hydrogenate to ethane. The activation energy of this reaction is comparable to or lower than the barrier for ethyl dehydrogenation to ethylidene (and eventually to ethylidyne). Furthermore, once ethane is produced, it likely desorbs irreversibly from the surface. The present analysis revealed that the barrier of ethylidene conversion to ethylidyne is remarkably low, only 19-29 kJ mol-1. Hence, one is lead to question the assignment of spectroscopic evidence32,33 to ethylidene as intermediate in the conversion of ethylene to ethylidyne conversion. With such a low barrier, ethylidene when formed would be quickly converted to ethylidyne and its spectroscopic detection seems unlikely.87,88 On the other hand, the calculated barriers indicate that vinyl or vinylidene could be accumulated on the surface due to the relatively slow conversion rate of these two species to ethylidene or ethylidyne. Thus, our findings call for further experimental clarification. Acknowledgment. We thank Benjami Martorell for discussions. This work was supported by Deutsche Forschungsgemeinschaft and Fonds der Chemischen Industrie (Germany). Supporting Information Available: Cartesian coordinates of the atoms in the optimized structures of the clean Pt(111) surface, the reactants, products and intermediates of Tables 1 and 2 as well as the transition states of Table 3. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Wilson, J.; Guo, H.; Morales, R.; Podgornov, E.; Lee, I.; Zaera, F. Phys. Chem. Chem. Phys. 2007, 9, 3830. (2) Somorjai, G. A.; Contreras, A. M.; Montano, M.; Rioux, R. M. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 10577. (3) Vincent, R. S.; Lindstedt, R. P.; Malik, N. A.; Reid, I. A. B.; Messenger, B. E. J. Catal. 2008, 260, 37. (4) Essen, J. M.; Haubrich, J.; Becker, C.; Wandelt, K. Surf. Sci. 2007, 601, 3472. (5) Nieminen, V.; Honkala, K.; Taskinen, A.; Murzin, D. Y. J. Phys. Chem. C 2008, 112, 6822. (6) Jacob, T.; Goddard III, W. A. J. Phys. Chem. B 2005, 109, 297. (7) Watwe, R. M.; Cortright, R. D.; Mavrikakis, M.; Nørskov, J. K.; Dumesic, J. A. J. Chem. Phys. 2001, 114, 4663. (8) Watson, G. W.; Wells, R. P. K.; Willock, D. J.; Hutchings, G. J. J. Phys. Chem. B 2000, 104, 6439. (9) Watwe, R. M.; Cortright, R. D.; Nørskov, J. K.; Dumesic, J. A. J. Phys. Chem. B 2000, 104, 2299. (10) Ge, Q.; King, D. A. J. Chem. Phys. 1999, 110, 4699. (11) Hirschl, R.; Eichler, A.; Hafner, J. J. Catal. 2004, 226, 273. (12) Kua, J.; Goddard III, W. A. J. Phys. Chem. B 1998, 102, 9492. (13) Steininger, H.; Ibach, H.; Lehwald, S. Surf. Sci. 1982, 117, 685.

Zhao et al. (14) Cassuto, A.; Mane, M.; Jupille, J. Surf. Sci. 1991, 249, 8. (15) Cassuto, A.; Kiss, J.; White, J. M. Surf. Sci. 1991, 255, 289. (16) Sto¨hr, J.; Sette, F.; Johnson, A. L. Phys. ReV. Lett. 1984, 53, 1684. (17) Koestner, R. J.; Sto¨hr, J.; Gland, J. L.; Horsley, J. A. Chem. Phys. Lett. 1984, 105, 332. (18) Cremer, P. S.; Somorjai, G. A. J. Chem. Soc., Faraday Trans. 1995, 91, 3671. (19) Hugenschmidt, M. B.; Dolle, P.; Jupille, J.; Cassuto, A. J. Vac. Sci. Technol., A 1989, 7, 3312. (20) Kubota, J.; Ichihara, S.; Kondo, J. N.; Domen, K.; Hirose, C. Surf. Sci. 1996, 357-358, 634. (21) Kesmodel, L. L.; Dubois, L. H.; Somorjai, G. A. J. Chem. Phys. 1979, 70, 2180. (22) Koestner, R. J.; Van Hove, M. A.; Somorjai, G. A. Surf. Sci. 1982, 121, 321. (23) Kesmodel, L. L.; Gates, J. A. Surf. Sci. 1981, 111, L747. (24) Marinova, T.; Kostov, K. L. Surf. Sci. 1987, 181, 573. (25) Hills, M. M.; Parmeter, J. E.; Mullins, C. B.; Weinberg, W. H. J. Am. Chem. Soc. 1986, 108, 3554. (26) Beebe, T. P.; Yates, J. T. J. Am. Chem. Soc. 1986, 108, 663. (27) Stacchiola, D.; Calaza, F.; Zheng, T.; Tysoe, W. T. J. Mol. Catal. A 2005, 228, 35. (28) Molero, H.; Stacchiola, D.; Tysoe, W. T. Catal. Lett. 2005, 101, 145. (29) Somorjai, G. A.; Van Hove, M. A.; Bent, B. E. J. Phys. Chem. 1988, 92, 973. (30) Zaera, F. J. Am. Chem. Soc. 1989, 111, 4240. (31) Ditlevsen, P. D.; Van Hove, M. A.; Somorjai, G. A. Surf. Sci. 1993, 292, 267. (32) Cremer, P.; Stanners, C.; Niemantsverdriet, J. W.; Shen, Y. R.; Somorjai, G. Surf. Sci. 1995, 328, 111. (33) Deng, R.; Herceg, E.; Trenary, M. Surf. Sci. 2004, 560, L195. (34) Moskaleva, L. V.; Chen, Z. X.; Aleksandrov, H. A.; Mohammed, A. B.; Sun, Q.; Ro¨sch, N. J. Phys. Chem. C 2009, 113, 2512. (35) Moskaleva, L. V.; Aleksandrov, H. A.; Basaran, D.; Zhao, Z.-J.; Ro¨sch, N. J. Phys. Chem. C 2009, 113, 15373. (36) Zaera, F. J. Am. Chem. Soc. 1989, 111, 8744. (37) Zaera, F. Surf. Sci. 1989, 219, 453. (38) Zaera, F.; Bernstein, N. J. Am. Chem. Soc. 1994, 116, 4881. (39) Zaera, F. Langmuir 1996, 12, 88. (40) Zaera, F.; French, C. R. J. Am. Chem. Soc. 1999, 121, 2236. ¨ fner, H. Surf. Sci. 1996, 368, 371. (41) Zaera, F.; Janssens, T. V. W.; O (42) Anghel, A. T.; Wales, D. J.; Jenkins, S. J.; King, D. A. Chem. Phys. Lett. 2005, 413, 289. (43) Anghel, A. T.; Wales, D. J.; Jenkins, S. J.; King, D. A. J. Chem. Phys. 2007, 126, 044710. (44) Stacchiola, D.; Tysoe, W. T. J. Phys. Chem. C 2009, 113, 8000. (45) Kresse, G.; Hafner, J. Phys. ReV. B 1994, 49, 14251. (46) Kresse, G.; Furthmu¨ller, J. Comput. Mater. Sci. 1996, 6, 15. (47) Perdew, J. P.; Wang, Y. Phys. ReV. B 1992, 45, 13244. (48) Blo¨chl, P. E. Phys. ReV. B 1994, 50, 17953. (49) Kresse, G.; Joubert, D. Phys. ReV. B 1999, 59, 1758. (50) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188. (51) Methfessel, M.; Paxton, A. T. Phys. ReV. B 1989, 40, 3616. (52) Handbook of Chemistry and Physics, 85th ed.; Lide, D. R., Ed.; CRC Press, Boca Raton, FL, 2004-2005; pp 12-21. (53) Griffiths, K.; Lennard, W. N.; Mitchell, I. V.; Norton, P. R.; Pirug, G.; Bonzel, H. P. Surf. Sci. Lett. 1993, 284, L389. (54) Henkelman, G.; Jońsson, H. J. Chem. Phys. 1999, 111, 7010. (55) Mills, G.; Jońsson, H.; Schenter, G. K. Surf. Sci. 1995, 324, 305. (56) Jońsson, H.; Mills, G.; Jacobsen, K. W. In Classical and Quantum Dynamics in Condensed Phase Simulations; Berne, B. J., Ciccotti, G., Coker, D. F., Eds.; World Scientific: Singapore, 1998; p 385. (57) Mermin, N. D. Phys. ReV. 1964, 137, A1441. (58) Cremer, P. S.; Su, X. C.; Shen, Y. R.; Somorjai, G. A. J. Am. Chem. Soc. 1996, 118, 2942. (59) Windham, R. G.; Bartram, M. E.; Koel, B. E. J. Phys. Chem. 1988, 92, 2862. (60) Stacchiola, D.; Azad, S.; Burkholder, L.; Tysoe, W. T. J. Phys. Chem. B. 2001, 105, 11233. (61) Hammer, B.; Hansen, L. B.; Nørskov, J. K. Phys. ReV. B 1999, 59, 7413. (62) Do¨ll, R.; Gerken, C. A.; Van Hove, M. A.; Somorjai, G. A. Surf. Sci. 1997, 374, 151. (63) Kesmodel, L. L.; Dubois, L. H.; Somorjai, G. A. Chem. Phys. Lett. 1978, 56, 267. (64) Starke, U.; Barbieri, A.; Materer, N.; Van Hove, M. A.; Somorjai, G. A. Surf. Sci. 1993, 286, 1. (65) Yeo, Y. Y.; Stuck, A.; Wartnaby, C. E.; King, D. A. Chem. Phys. Lett. 1996, 259, 28. (66) Natal-Santiago, M. A.; Podkolzin, S. G.; Cortright, R. D.; Dumesic, J. A. Catal. Lett. 1997, 45, 155.

Ethylidyne Formation from Ethylene over Pt(111) (67) Shen, J. Y.; Hill, J. M.; Watwe, R. M.; Spiewak, B. E.; Dumesic, J. A. J. Phys. Chem. B 1999, 103, 3923. (68) Mohsin, S. B.; Trenary, M.; Robota, H. J. J. Phys. Chem. 1988, 92, 5229. (69) Ford, D. C.; Xu, Y.; Mavrikakis, M. Surf. Sci. 2005, 587, 159. (70) Janssens, T. V. W.; Zaera, F. J. Phys. Chem. 1996, 100, 14118. (71) Chen, Y.; Vlachos, D. G. J. Phys. Chem. C 2010, 114, 4973. (72) Lo, J. M. H.; Ziegler, T. J. Phys. Chem. C 2007, 111, 13149. (73) Ogle, K. M.; Creighton, J. R.; Akhter, S.; White, J. M. Surf. Sci. 1986, 169, 246. (74) Mohsin, S. B.; Trenary, M.; Robota, H. J. Chem. Phys. Lett. 1989, 154, 511. (75) Erley, W.; Li, Y.; Land, D. P.; Hemminger, J. C. Surf. Sci. 1994, 301, 177. (76) Godbey, D.; Zaera, F.; Yeates, R.; Somorjai, G. A. Surf. Sci. 1986, 167, 150. ¨ fner, H.; Zaera, F. J. Phys. Chem. B 1997, 101, 396. (77) O (78) Ko, M. K.; Frei, H. J. Phys. Chem. B 2004, 108, 1805. (79) Wasylenko, W.; Frei, H. J. Phys. Chem. B. 2005, 109, 16873. (80) Deng, R. P.; Jones, J.; Trenary, M. J. Phys. Chem. C 2007, 111, 1459. (81) Fan, J.; Trenary, M. Langmuir 1994, 10, 3649. (82) Ibach, H.; Lehwald, S. J. Vac. Sci. Technol. 1978, 15, 407.

J. Phys. Chem. C, Vol. 114, No. 28, 2010 12201 (83) Malik, I. J.; Brubaker, M. E.; Mohsin, S. B.; Trenary, M. J. Chem. Phys. 1987, 87, 5554. (84) Dwivedi, Y.; Rai, S. B. Vib. Spectrosc. 2009, 49, 278. (85) Perez-Delgado, Y.; Lewis, E. K.; Moehnke, C. J.; Salazar, M. C.; Hernandez, A. J.; Manzanares, C. E. Mol. Phys. 2009, 107, 1367. (86) Alparone, A. Chem. Phys. 2006, 327, 127, and references therein. (87) Obtained with an estimate of the life time, τ(s) ) 10-13 × [(εε*)/ε]1-s, from RRK theory (see ref 88), where ε is the total internal energy available to the intermediate and ε* is the activation energy, s is the number of active vibrational modes. In the present case, take ε to be equal to the reverse barrier of reaction e, from ethylidene to vinyl. At least this energy should be available to the newly formed intermediate. For 1/9 coverage, one has ε g 60 kJ/mol, ε* ) 20 kJ/mol, s e 12. Therefore, τ < 1 ps. (88) Laidler, K. J. Chemical Kinetics; 3rd ed.; Harper and Row: New York, 1987; p 157. (89) Cremer, P. S.; Su, X. C.; Shen, Y. R; Somorjai, G. A. J. Phys. Chem. B 1997, 101, 6474. (90) Evans, J.; McNulty, G. S. J. Chem. Soc., Dalton Trans. 1983, 639. (91) Liu, Z.-M.; Zhou, X.-L.; Buchanan, D. A.; Kiss, J.; White, J. M. J. Am. Chem. Soc. 1992, 114, 2031. (92) Wang, L. P.; Tysoe, W. T.; Ormerod, R. M.; Lambert, R. M.; Hoffmann, H.; Zaera, F. J. Phys. Chem. 1990, 94, 4236.

JP100612Y

Ethylidyne Formation from Ethylene over Pt(111): A Mechanistic Study

Recommend Documents