Theoretical understanding on the selectivity of acrolein hydrogenation over silver surfaces: the non-Horiuti–Polanyi mechanism is the key

Kaili Wang abc and Bo Yang *a
aSchool of Physical Science and Technology, ShanghaiTech University, 393 Middle Huaxia Road, Shanghai 201210, China. E-mail: yangbo1@shanghaitech.edu.cn
bShanghai Institute of Organic Chemistry, Chinese Academy of Sciences, Shanghai 200032, China
cUniversity of Chinese Academy of Sciences, Beijing 100049, China

Received 25th July 2017 , Accepted 10th August 2017

First published on 11th August 2017


Chemoselective hydrogenation of α,β-unsaturated aldehydes to unsaturated alcohols is not only an important reaction in the chemical industry but also a good model system to understand the catalytic selectivity in heterogeneous catalysis. In the current work, the selectivity of partial hydrogenation of acrolein (C3H4O), the simplest α,β-unsaturated aldehyde, is investigated employing density functional theory (DFT) calculations. Two hydrogenation mechanisms, namely the Horiuti–Polanyi mechanism and the non-Horiuti–Polanyi mechanism, are employed to study the partial hydrogenation of acrolein over Ag(111), Ag(100), Ag(211) and Ag(111)-mono surfaces. It is found that the hydrogenation of C3H4O to C3H5O at the terminal carbon and oxygen atoms follows the non-Horiuti–Polanyi mechanism in which C3H4O reacts with hydrogen molecules directly over all the silver surfaces studied, whilst atomic hydrogen is the active hydrogen species for the hydrogenation of C3H5O to C3H6O. Subsequently, the selectivities between partial hydrogenation products, i.e. propenol, propanal and enol, over silver surfaces with different morphologies are compared by calculating the energy difference between the rate-determining transition states. We find that the selectivity of propenol formation increases with the coordination number of surface silver atoms, which is in good agreement with the trend of selectivities obtained experimentally. It is also interesting to find that the selectivity trend obtained based solely upon the Horiuti–Polanyi mechanism for the hydrogenation of C3H4O to C3H5O and C3H5O to C3H6O cannot explain the experimental results. In other words, the non-Horiuti–Polanyi mechanism is able to give a more reasonable explanation for the selectivity trend observed experimentally than the normally used Horiuti–Polanyi mechanism in heterogeneous catalysis. Our work highlights the significance of the non-Horiuti–Polanyi mechanism in understanding heterogeneous catalytic hydrogenation reactions.


1. Introduction

Selective hydrogenation of α,β-unsaturated aldehydes (UALs) to unsaturated alcohols (UOLs) is an important process in the pharmaceutical and fine chemical industries, and this process has been investigated extensively with both experimental and computational methods.1–5 Competing reaction pathways for UOL formation are the chemoselective reduction of the C[double bond, length as m-dash]C bond-forming saturated aldehydes (SALs) and the 1,4-hydrogenation of UALs to produce enols. In general, hydrogenation of acrolein, the simplest UAL, possesses the lowest selectivity for allyl alcohol production compared with the hydrogenation of other UALs, since adding substituents would destabilize the interaction between the C[double bond, length as m-dash]C bond and the surface, and therefore facilitate the formation of UOLs.6–8

Platinum group metals (PGMs), such as Pd, Pt and Ru, typically show poor performance for the selective hydrogenation of acrolein to propenol.7–10 For instance, the selectivity of propenol formation was found to be less than 2% over Pt catalysts.7,8 Based on density functional theory (DFT) calculations, Sautet's group showed that the adsorption geometry of the UAL or the desorption of different partial hydrogenation products over Pt(111) controls the selectivity.6,7,11,12 Recently, Tuokko et al. reported that the selectivities of acrolein hydrogenation on Pd(111) and Pt(111) are governed by steric effects of the surface adsorbates, employing DFT calculations.4 The authors found that, with increasing acrolein coverage, the hydrogenation of the C[double bond, length as m-dash]O bond becomes less favorable on both metals, and a new perspective was provided on understanding the selectivity issue in this field. Various methods have been employed to improve the selectivity of UOL formation over PGM catalysts.5,10,13 For example, a Pt@SnOx/SiO2 catalyst with a core–shell structure was developed recently showing significantly promoted selectivity for the hydrogenation of acrolein to propenol, and it was found that the synergistic effect between Pt and SnOx plays an important role in the selectivity enhancement.13

In contrast, group 11 transition metals, especially Au and Ag, possess much higher selectivity (around 40%) towards propenol formation for the hydrogenation of acrolein.14–19 Claus and co-workers investigated several factors governing the performance of supported gold catalysts for the selective hydrogenation of acrolein, including the catalyst particle size effect, metal–support interactions and surface modification effect. Experiments under identical reaction conditions revealed an increase of propenol selectivity from 15% to 35% over Au/ZrO2, when the mean particle size of gold was increased from 4.0 nm to 7.7 nm.20

Regarding supported silver-catalyzed hydrogenation of acrolein, Bron et al. employed a broad pressure range, from ∼2 mbar to 20 bar, to examine the effect of reactant partial pressure on selectivity, and found that the formation of propenol was favored at high pressures.16,17 It was also found experimentally that the selectivity towards desired UOLs was also enhanced when the Ag/SiO2 catalysts were pre-treated with oxygen,21 and the results were further supported by DFT calculations.22 Furthermore, the particle size effect on the selectivity of UAL hydrogenation was observed for a variety of supported silver catalysts.20,23,24 Claus' group investigated the structure sensitivity of crotonaldehyde hydrogenation over silver catalysts, and found that, when the silver particle size is in the range of 1.4 to 2.8 nm, the larger silver particles give higher selectivity to crotyl alcohol than the smaller ones.23 Recently, Meyer and co-workers prepared a series of silica-supported silver catalysts with different particle sizes ranging from 1 to 9 nm. They found that, at both high and low pressures, the obtained selectivities of propenol formation over the supported catalysts with the smallest particles are always lower than those with larger silver particle size.24

As one of the most important reactions in scientific research and industry, hydrogenation was found to follow different reaction mechanisms. The Horiuti–Polanyi mechanism, where hydrogen molecules dissociate initially followed by the addition of atomic hydrogen to the substrate consecutively, has been universally recognized as the explanation for hydrogenation reactions for several decades.25,26 On many active metals such as Pd and Pt, H2 dissociation has been found barrierless,27–30 hence, this process is by no means the obstacle for complete hydrogenation reactions. However, there is tangible evidence that sizable barriers for H2 dissociation on group 11 metals may exist. The reaction mechanism for the hydrogenation of acrolein over gold-based catalysts was investigated systematically using DFT calculations and it was found that a non-Horiuti–Polanyi mechanism, demonstrating a process where molecular hydrogen directly reacts with acrolein, was favored in some cases.19 Vilé et al., combining experiments and theoretical calculations, found that the partial hydrogenation of propyne over supported Ag nanoparticles also prefers this non-Horiuti–Polanyi mechanism, which was named as an associative mechanism in their work.31

In the current work, we attempt to understand the selectivity trend of acrolein partial hydrogenation over silver catalysts with different sizes as well as the pressure effects on the selectivity employing DFT calculations. Considering that smaller metal particles would contain more low-coordination number atoms on the surface,23 we will study the selectivity of partial hydrogenation of acrolein on Ag(111), Ag(100), Ag(211) and Ag(111)-mono surfaces with decreasing coordination numbers of the surface silver atoms. The Horiuti–Polanyi and non-Horiuti–Polanyi mechanisms are examined and discussed for the hydrogenation pathways over these surfaces. Moreover, the selectivity trend obtained from using different mechanisms will be compared with the experimental results reported previously.

2. Computational details

All the density functional theory (DFT) calculations were carried out with periodic slab models using the Vienna Ab initio Simulation Package (VASP) code.32–35 The generalized gradient approximation (GGA) was used with the Perdew–Burke–Ernzerhof (PBE) exchange–correlation functional.36 The interaction between atomic cores and electrons was employed with the projector augmented wave (PAW) method.37,38 It was found in the literature that, when the dispersion correction method D2 was used,39 the adsorption of acrolein on Ag(111) was overestimated and surface reconstruction would happen over Ag(100).40,41 Therefore, the dispersion correction developed by Tkatchenko and Scheffler (PBE-TS),42,43 which is widely employed in DFT calculations,44,45 was taken into account in the current work. The lattice parameters of silver obtained from PBE and PBE-TS calculations are 4.07 and 4.16 Å, respectively. An energy cutoff of 500 eV and the force threshold in structural optimization and transition state location below 0.05 eV per Å were employed. Furthermore, the transition states were located with a constrained minimization method,46–48 and all of the structures were verified by vibrational frequency analyses. The adsorption energies were defined as follows:
Ead = Etotal − (Eg + Eslab)
where Etotal is the energy of the whole system after adsorption, Eg is the energy of a molecule located in the gas phase, and Eslab is the energy of the slab.

Four silver surfaces exposing silver atoms with decreasing coordination numbers, Ag(111), Ag(100), Ag(211) and Ag(111)-mono, were considered, the structures of which are included in the ESI as Fig. S1. Ag(111)-mono was created by adding a mono-row of silver atoms on Ag(111), and is similar to those structures reported previously for gold.19,49,50 Ag(111) and Ag(111)-mono surfaces were modeled with four-layer slabs and 4 × 4 surface supercells. The adsorption was considered on one side of the slabs, the bottom two layers were fixed during optimizations, and 3 × 3 × 1 k-point grids were employed in the calculations of the Ag surfaces investigated. The parameters used for the optimization of the Ag(100) surface were identical to those for Ag(111) and Ag(111)-mono surfaces except that 3 × 3 supercells were used for Ag(100). Twelve-layer slabs and 4 × 1 supercells were established for Ag(211) and 3 × 2 × 1 k-point grids were used. In order to minimize the interaction between slabs, a vacuum of more than 13 Å was built.

3. Results and discussion

3.1 Adsorption of reactants and products

Scheme 1 shows the structures of the reactant, intermediates and products involved in the partial hydrogenation pathways of acrolein to produce propenol, enol and propanal. For clarity, the oxygen and carbon atoms of acrolein shown in Scheme 1 are numbered as O1, C2, C3 and C4. The desired product propenol can be produced via consecutive O1–C2 or C2–O1 hydrogen addition reaction routes, which are defined as R12 or R21, respectively. Moreover, the intermediates CH2CHCHOH, CH2CHCH2O, CH2CH2CHO and CH3CHCHO are named MS1, MS2, MS3 and MS4, respectively, as one can see from Scheme 1.
image file: c7cy01500c-s1.tif
Scheme 1 Structures of the reactant, intermediates and products in the acrolein hydrogenation pathways to produce propenol, enol and propanal. The oxygen and carbon atoms of acrolein are numbered as O1, C2, C3 and C4. MS1, MS2, MS3 and MS4 are the reaction intermediates CH2CHCHOH, CH2CHCH2O, CH2CH2CHO and CH3CHCHO, respectively. The reaction pathways not considered in the current work are shown as dotted arrows.

The adsorption energies of atomic hydrogen, acrolein, propanal, enol and propenol obtained with and without dispersion corrections are listed in Table 1, and it is obvious that the adsorption energies calculated with dispersion corrections are relatively stronger. One can see that the adsorption of atomic hydrogen is rather weak on all silver surfaces, even with dispersion corrections considered. The adsorption of acrolein, propanal, enol and propenol becomes weaker with increasing coordination number of silver. The most stable adsorption geometries of the reactants, products and intermediates over Ag(111), Ag(100), Ag(211) and Ag(111)-mono obtained with dispersion corrections considered are presented in Fig. 1. Those configurations obtained without dispersion corrections are shown in Fig. S2 of the ESI. One can see obviously that the structures optimized with and without dispersion corrections are similar. It should be mentioned that the coverage effect on the adsorption energies is not considered in the current work due to the following two reasons: i) the adsorption of the reactants, acrolein and hydrogen, over Ag surfaces is much weaker than that over Pd(111) and Pt(111) surfaces;4 ii) the reaction temperature for acrolein hydrogenation is as high as ∼500 K and the entropic effect would make the adsorption energies of acrolein weakened by ∼1.5 eV at this temperature, and the coverage of acrolein on the surface would therefore be substantially decreased.

Table 1 Adsorption energies (in eV) of atomic hydrogen, acrolein, propanal, enol and propenol over Ag(111), Ag(100), Ag(211) and Ag(111)-mono surfaces obtained from PBE and PBE-TS calculations. All the energies are ZPE-corrected
Ag(111) Ag(100) Ag(211) Ag(111)-mono
Hydrogen PBE 0.14 0.21 0.20 −0.03
PBE-TS 0.10 0.13 0.12 −0.03
Acrolein PBE −0.06 −0.08 −0.20 −0.45
PBE-TS −0.48 −0.59 −0.62 −0.74
Propanal PBE −0.07 −0.09 −0.15 −0.25
PBE-TS −0.46 −0.52 −0.72 −0.74
Enol PBE −0.12 −0.15 −0.21 −0.26
PBE-TS −0.64 −0.71 −0.74 −0.63
Propenol PBE −0.15 −0.18 −0.27 −0.38
PBE-TS −0.68 −0.77 −0.77 −0.76



image file: c7cy01500c-f1.tif
Fig. 1 Adsorption configurations of acrolein (C3H4O), MS1 (CH2CHCHOH), MS2 (CH2CHCH2O), MS3 (CH2CH2CHO), MS4 (CH3CHCHO), propanal (CH3CH2CHO), enol (CH3CHCHOH) and propenol (CH2CHCH2OH) over Ag(111), Ag(100), Ag(211) and Ag(111)-mono surfaces obtained from PBE-TS calculations. The oxygen, carbon, hydrogen and silver atoms are shown in red, grey, white and light blue, respectively, and this notation is used throughout the paper.

3.2 Hydrogenation of C3H4O to C3H5O

Two hydrogenation mechanisms, i.e. Horiuti–Polanyi and non-Horiuti–Polanyi mechanisms, are utilized to analyze the reaction pathways of C3H4O hydrogenation to produce C3H5O. The elementary reactions considered include the dissociation of H2 and the hydrogenation of C3H4O to C3H5O with atomic and molecular hydrogen. Therefore, the two mechanisms of C3H4O hydrogenation can be written as:
Horiuti–Polanyi pathway: C3H4O(g) + H2(g) → C3H4O(ad) + 2H(ad) → C3H5O(ad) + H(ad)

non-Horiuti–Polanyi pathway: C3H4O(g) + H2(g) → C3H4O(ad) + H2(g) → C3H5O(ad) + H(ad),
where (g) and (ad) are the gaseous and adsorption states, respectively.
3.2.1 The Horiuti–Polanyi mechanism. The activation and reaction energies, obtained from PBE and PBE-TS calculations, of the elementary reactions involved in the Horiuti–Polanyi pathway of C3H4O hydrogenation to C3H5O on different silver surfaces are listed in Table 2. One can see that the dissociation of hydrogen needs to overcome high barriers over low-index silver surfaces. Using the entropy data reported for gaseous H2[thin space (1/6-em)]51 and assuming that the entropy of transition states is negligible compared with the gaseous entropy, the dissociative adsorption free energy barriers of H2 over Ag(111), Ag(100), Ag(211) and Ag(111)-mono surfaces at 500 K are estimated to be 1.99, 1.82, 1.79 and 1.52 eV, respectively, with dispersion corrections included, and 2.09, 1.98, 1.95 and 1.59 eV, respectively, without dispersion corrections. Comparing these dissociation barriers with those of the subsequent surface hydrogenation reactions with atomic hydrogen, one can find that hydrogen dissociation possesses the highest barrier and should be rate-determining for the hydrogenation of C3H4O to C3H5O over all the Ag surfaces studied. The corresponding transition state structures for C3H4O hydrogenation to C3H5O with atomic hydrogen, optimized with the PBE-TS method, are presented in Fig. 2, along with the C–H and O–H distances at the corresponding transition states. Configurations obtained without dispersion corrections are shown in Fig. S3 of the ESI.
Table 2 Activation energies (Ea/eV) and reaction energies (ΔE/eV) of hydrogen dissociation and the hydrogenation of C3H4O with atomic and molecular hydrogen obtained from PBE and PBE-TS calculations. All the energies are ZPE-corrected
Ag(111) Ag(100) Ag(211) Ag(111)-mono
E a ΔE E a ΔE E a ΔE E a ΔE
H2 + 2* → 2H* PBE 1.33 0.28 1.22 0.43 1.19 0.39 0.83 −0.05
PBE-TS 1.23 0.21 1.06 0.27 1.03 0.23 0.76 −0.07
C3H4O* + H* → C3H4OH* + * PBE 0.70 0.00 0.68 −0.13 0.71 −0.16 0.83 −0.02
PBE-TS 0.58 −0.03 0.63 −0.17 0.69 −0.17 0.80 −0.03
C3H4O* + H* → CH2CHCH2O* + * PBE 0.73 −0.22 0.62 −0.09 0.63 −0.08 0.95 0.04
PBE-TS 0.69 −0.21 0.62 −0.05 0.64 −0.13 0.98 0.06
C3H4O* + H* → CH2CH2CHO* + * PBE 0.93 −0.01 0.92 −0.07 0.68 −0.20 0.87 −0.02
PBE-TS 0.91 −0.05 0.96 −0.02 0.72 −0.12 0.89 0.00
C3H4O* + H* → CH3CHCHO* + * PBE 0.48 −0.77 0.30 −0.97 0.27 −0.96 0.29 −0.77
PBE-TS 0.39 −0.72 0.30 −0.89 0.27 −0.91 0.30 −0.70
C3H4O* + H2 + * → C3H4OH* + H* PBE 0.66 0.28 0.38 0.29 0.50 0.24 0.30 −0.07
PBE-TS 0.48 0.18 0.22 0.10 0.37 0.06 0.27 −0.10
C3H4O* + H2 + * → CH2CHCH2O* + H* PBE 1.05 0.06 0.74 0.33 0.90 0.32 1.38 −0.01
PBE-TS 0.88 0.00 0.61 0.22 0.72 0.10 1.32 0.00
C3H4O* + H2 + * → CH2CH2CHO* + H* PBE 1.10 0.27 0.92 0.36 1.14 0.20 1.43 −0.07
PBE-TS 1.00 0.16 0.82 0.25 1.02 0.11 1.36 −0.07
C3H4O* + H2 + * → CH3CHCHO* + H* PBE 0.56 −0.49 0.41 −0.54 0.43 −0.57 0.45 −0.82
PBE-TS 0.45 −0.51 0.31 −0.62 0.38 −0.68 0.39 −0.77



image file: c7cy01500c-f2.tif
Fig. 2 Transition state structures for C3H4O hydrogenation with atomic and molecular hydrogen are obtained from PBE-TS calculations. O1–H, C2–H, C3–H and C4–H distances at the corresponding transition states are also shown here. O1, C2, C3 and C4 are defined in Scheme 1.
3.2.2 The non-Horiuti–Polanyi mechanism. An analogous study was also carried out on the hydrogenation of acrolein following the non-Horiuti–Polanyi mechanism. The activation energies of the elementary reactions over all the surfaces studied are also listed in Table 2. The corresponding transition state geometries obtained from PBE-TS calculations are shown in Fig. 2 and the structures obtained without dispersion corrections are shown in Fig. S3. From Table 2, one can see that, for the hydrogenation of C3H4O to C3H5O with molecular hydrogen, the barriers of MS2 and MS3 formation are much higher than those of MS1 and MS4 formation, indicating that the formation of MS1 and MS4 is favored over all the silver surfaces studied, which is very likely due to the steric effect and is in agreement with the literature.11,12,18,52–55 The activation energy differences between O1 and C4 hydrogenation on Ag(111), Ag(100), Ag(211) and Ag(111)-mono surfaces are small for both PBE and PBE-TS calculations. Regarding the transition state structures shown in Fig. 2, it is clearly demonstrated that the O1–H distances are the shortest among the O(C)–H distances at the transition states of hydrogenation over the same silver surface.
3.2.3 Comparison between two hydrogenation mechanisms. It is demonstrated above that hydrogen dissociation is rate-determining in the Horiuti–Polanyi pathways of C3H4O hydrogenation to C3H5O over all the surfaces studied, and for the non-Horiuti–Polanyi mechanism, the barriers of MS2 and MS3 formation are much higher than those of MS1 and MS4 formation. Therefore, we only compare the activation energies of hydrogen dissociation with those of MS1/MS4 formation with molecular hydrogen, to verify the preference of these two mechanisms in the system. The correlation between the energies of transition states (ETS-gas) and the adsorption energies of the adsorbates (Ead), all with respect to the energies of the gaseous reactants, is shown in Fig. 3, and an extended Brønsted–Evans–Polanyi (BEP)-type relationship can be established.19,48,53,56,57 One can see that, when dispersion correction is considered, the adsorption on all silver surfaces becomes stronger and the energies of transition states are relatively more stable than those obtained without dispersion corrections. More importantly, it is obvious that the ETS-gas values obtained following the Horiuti–Polanyi mechanism are higher than those following the non-Horiuti–Polanyi mechanism in all the cases considered, which is consistent with the literature that the non-Horiuti–Polanyi mechanism should be favored in the weak reactant adsorption energy region, e.g. over silver surfaces.19,31,58 It would be rather interesting to see similar experimental studies as reported by Vilé et al. to be carried out to provide evidence supporting this non-Horiuti–Polanyi mechanism for acrolein hydrogenation over Ag surfaces.31
image file: c7cy01500c-f3.tif
Fig. 3 Relationship between the energies of the transition states of C3H4O hydrogenation to C3H5O referred to the gas phase C3H4O + H2 and the adsorption energies of C3H4O following different mechanisms. HP mechanism, NHP mechanism-C and NHP mechanism-O are the Horiuti–Polanyi mechanism, terminal carbon atom hydrogenation following the non-Horiuti–Polanyi mechanism and terminal oxygen atom hydrogenation following the non-Horiuti–Polanyi mechanism, respectively. Solid labels represent the results calculated with dispersion corrections included, while the results obtained without dispersion corrections are shown as hollow labels.

3.3 Hydrogenation of C3H5O to C3H6O

In light of the facts that hydrogenation of C3H4O to C3H5O prefers the non-Horiuti–Polanyi mechanism and MS2/MS3 formation is more difficult than MS1 and MS4 formation, we only consider the hydrogenation of intermediates MS1 and MS4 to produce C3H6O in the current work. Therefore, the formation of propenol, propanal and enol is considered following R12, R43 and R14 (R41) routes, respectively. It is shown in Table 3 that the activation energies of the hydrogenation of C3H5O to C3H6O with atomic hydrogen, in general, increase with decreasing coordination number of silver. The corresponding transition state structures obtained from PBE-TS calculations are presented in Fig. 4 along with the O1–H, C2–H, C3–H and C4–H distances at the corresponding transition states. Fig. S4 shows the corresponding configurations calculated without dispersion corrections.
Table 3 Activation energies (Ea/eV) and reaction energies (ΔE/eV) of C3H5O hydrogenation with atomic hydrogen obtained from PBE and PBE-TS calculations. All the energies are ZPE-corrected
Ag(111) Ag(100) Ag(211) Ag(111)-mono
E a ΔE E a ΔE E a ΔE E a ΔE
C3H4OH* + H* → CH2CHCH2OH* + * PBE 0.32 −0.85 0.31 −0.87 0.56 −0.79 0.84 −0.34
PBE-TS 0.23 −0.88 0.33 −0.77 0.55 −0.70 0.73 −0.42
C3H4OH* + H* → CH3CHCHOH* + * PBE 0.48 −1.21 0.47 −1.23 0.63 −1.11 0.85 −0.60
PBE-TS 0.36 −1.22 0.48 −1.10 0.61 −1.06 0.72 −0.67
CH3CHCHO* + H* → CH3CHCHOH* + * PBE 0.70 −0.43 0.76 −0.39 0.95 −0.30 1.14 0.14
PBE-TS 0.56 −0.54 0.74 −0.38 0.99 −0.33 1.09 0.00
CH3CHCHO* + H* → CH3CH2CHO* + * PBE 0.55 −0.66 0.47 −0.61 0.74 −0.52 0.87 −0.13
PBE-TS 0.48 −0.65 0.50 −0.50 0.78 −0.60 0.76 −0.41



image file: c7cy01500c-f4.tif
Fig. 4 Transition state structures for C3H5O hydrogenation to propenol (R12), enol (R41 and R14) and propanal (R43) with atomic hydrogen obtained from PBE-TS calculations. O1–H, C2–H, C3–H and C4–H distances at the corresponding transition states are also shown here.

It should be mentioned that the non-Horiuti–Polanyi mechanism is not considered for the hydrogenation of C3H5O to C3H6O on the silver surfaces investigated. This is due to the fact that, after the hydrogenation of C3H4O with molecular hydrogen, one hydrogen atom would be produced along with C3H5O on the surface. Moreover, the further hydrogenation of C3H5O with this hydrogen atom is found to be favored over the recombination of the hydrogen atoms to hydrogen molecules or the hydrogenation with molecular hydrogen due to the large entropic effect of hydrogen molecules. As one can see from Tables 3 and S1 in the ESI, the recombination barriers of two hydrogen atoms are higher than the corresponding hydrogenation barriers of C3H5O with hydrogen atoms in most of the cases studied, except for the cases of MS4 hydrogenation to enol on Ag(211) and Ag(111)-mono surfaces. However, the activation barriers of MS4 hydrogenation with hydrogen molecules at 500 K over Ag(211) and Ag(111)-mono surfaces are calculated to be 1.19 and 1.14 eV, respectively, with PBE-TS, and those barriers change to 1.36 and 1.23 eV, respectively, when dispersion corrections are not considered. These barriers are much higher than the corresponding activation barriers of hydrogenation with hydrogen atoms. Therefore, the hydrogenation of C3H5O over these surfaces studied is considered favoring the atomic-hydrogenation mechanism. This is consistent with the results we reported very recently on the hydrogenation of acetylene over Ag(211).58

3.4 Reaction pathway of the hydrogenation of C3H4O to C3H6O

From all the barriers and reaction energies of the elementary steps listed in Tables 2 and 3, we find that the mechanism of C3H4O hydrogenation to C3H6O over all the surfaces can be summarized as: C3H4O(g) + H2(g) → C3H4O(ad) + H2(g) → C3H5O(ad) + H(ad) → C3H6O(ad) → C3H6O(g). The energy profiles of C3H4O hydrogenation to C3H6O over Ag(111), Ag(100), Ag(211) and Ag(111)-mono surfaces are therefore obtained and shown in Fig. 5 (PBE-TS) and S6 (PBE). The following three trends can be readily obtained: (i) hydrogen dissociation is rate-determining in the whole Horiuti–Polanyi pathways; (ii) MS4 is more stable than MS1 over all the silver surfaces investigated; (iii) propanal is the most stable among the three partial hydrogenation products whilst propenol possesses the highest energy both on the surface and in the gas phase. We will show later that these three trends observed are rather important while comparing the selectivity to different C3H6O species produced over silver surfaces.
image file: c7cy01500c-f5.tif
Fig. 5 Energy profiles of the preferred acrolein hydrogenation pathways to propenol (12), enol (14 and 41) and propanal (43) over Ag(111), Ag(100), Ag(211) and Ag(111)-mono surfaces obtained from PBE-TS calculations. The gas phase total energy of acrolein + H2 is defined as the reference energy.

3.5 Particle size and pressure effects on the selectivity

The selectivity of acrolein hydrogenation to propenol, enol and propanal can be obtained quantitatively by comparing the effective barriers (Eeffa) of product formation, which are estimated with the kinetic model reported previously.19,59,60 In this model, we assumed that the heterogeneous catalytic processes can be simplified into two steps, namely dissociative adsorption and associative desorption. By solving the microkinetic equations quantitatively with steady-state approximation, the effective barriers can be determined and used to estimate the reaction rates. From the energy profiles shown in Fig. 5 and S6, we found that the adsorption process (C3H4O(g) + H2(g) → C3H6O(ad)) should be rate-determining over all the surfaces for the hydrogenation of acrolein, due to the large entropic effect on the energies of reactants under reaction conditions. Therefore, the rate-determining initial states are the same for all the cases studied, and the effective barriers only depend on the energies of the rate-determining transition states (RDTS). The energies of RDTS referred to the gas phase C3H4O + H2 (ERDTS-gas) to produce propenol, enol and propanal obtained from PBE and PBE-TS calculations are listed in Table 4. One can find that the higher RDTS energies indicate lower formation rates. A similar approach has been widely used in the literature.61,62
Table 4 The energies of rate-determining transition states referred to the gas phase C3H4O + H2 (ERDTS-gas) to produce propanal, enol and propenol obtained from PBE and PBE-TS calculations. ZPE corrections are included. R41 and R43 routes possess the same energies and are shown in one column
R12 R14 R43(R41)
Ag(111) PBE 0.60 0.70 0.51
PBE-TS 0.00 0.05 −0.03
Ag(100) PBE 0.53 0.68 0.33
PBE-TS −0.16 −0.01 −0.28
Ag(211) PBE 0.55 0.67 0.24
PBE-TS −0.01 0.05 −0.24
Ag(111)-mono PBE 0.32 0.33 0.00
PBE-TS −0.11 −0.12 −0.35


Since the isomerization of enols to the corresponding SALs takes place readily and the RDTS energies of R43 are lower than those of R14 over all the surfaces,4,18 we simply take into account the ERDTS-gas leading to propanal (R43) and propenol (R12) for comparison. The trend of the energy differences between propanal (ERDTS-gas,al) and propenol (ERDTS-gas,ol) formation (ΔERDTS-gas = ERDTS-gas,alERDTS-gas,ol) obtained from PBE-TS calculations as a function of the coordination number of surface silver atoms is shown in Fig. 6. It should be mentioned that the trends of ΔERDTS-gas obtained from PBE and PBE-TS calculations are identical, as one can find from Table 4. Since a higher ΔERDTS-gas value indicates higher propenol selectivity, one can see that the selectivity of propenol formation is enhanced on the low-index surfaces.


image file: c7cy01500c-f6.tif
Fig. 6 Comparison between the selectivity of acrolein hydrogenation to propenol obtained from PBE-TS calculations and those experimental results reported previously. ΔERDTS-gas is defined as the difference between the energies of rate-determining transition states to produce propanal and propenol, and higher ΔERDTS-gas indicates higher propenol selectivity. The experimental selectivity values are obtained from ref. 24, and the reaction is under 473 K, 1 atm and 10% conversion.

In addition, we include the experimentally obtained selectivity of propenol formation, reported by Wei et al. over a series of silica-supported silver catalysts,24 as a function of particle size in Fig. 6. Since the fraction of metal atoms with different coordination numbers exhibits dependence on the particle size,22,63,64 we find that our calculation is consistent with the selectivity trend observed experimentally. It should be mentioned here that to directly relate our computational results with the experimental selectivities reported is rather difficult. This is due to several reasons, such as the accuracy of DFT calculations and the real catalyst particles which always contain metal atoms with different coordination numbers. Therefore, the trends obtained from density functional calculations typically provide more physical insights into the experiments than the exact energy values calculated. It is further found that there would be no difference between the propenol selectivity obtained based solely upon the Horiuti–Polanyi mechanism over different surfaces, since hydrogen dissociation is rate-determining over all the surfaces and the selectivity of product formation is determined by the stability of gaseous products that is identical over different catalyst surfaces, indicating that the non-Horiuti–Polanyi mechanism is able to give more reasonable explanations for the selectivity trend observed experimentally than the normally used Horiuti–Polanyi mechanism in the current case.

Furthermore, Claus' and Meyer's experiments demonstrated that the selectivity of acrolein hydrogenation to propenol increases with increasing reactant partial pressure. We believe that, at low partial pressure, it is highly possible for most of the acrolein to adsorb at the silver sites with lower coordination numbers and lower propenol selectivity, e.g. steps and kinks, due to the more negative adsorption energies obtained in the current work. However, these sites would be fully occupied with increasing pressure and the possibility of reactant adsorption at those high-coordination number sites over Ag(111) may increase according to the adsorption energy trend presented in Table 1.17,65,66 Since high-coordination number sites possess higher selectivity to produce propenol, the selectivity of propenol formation therefore increases with increasing reactant partial pressure.

4. Conclusion

In summary, we have investigated the selectivity of partial hydrogenation of acrolein to propenol over Ag(111), Ag(100), Ag(211) and Ag(111)-mono surfaces using dispersion-corrected and uncorrected DFT calculations. The following conclusions can be drawn:

(i) The inclusion of dispersion correction would give rise to stronger adsorption of the adsorbates and transition states. However, the trends observed and conclusions obtained using these two methods are identical.

(ii) For the hydrogenation of C3H4O to MS1 and MS4, the non-Horiuti–Polanyi mechanism is favored over all the surfaces studied due to the high-energy barriers of hydrogen dissociation.

(iii) Atomic hydrogen is the active species in the subsequent hydrogenation of C3H5O to C3H6O, since we find that the further hydrogenation of C3H5O with the hydrogen atom produced from the first step of acrolein hydrogenation is more favorable than the recombination of the hydrogen atoms or the hydrogenation with molecular hydrogen.

(iv) According to the energy profiles obtained from the above mechanistic understanding, the selectivity of acrolein partial hydrogenation to produce propenol and propanal can be estimated, and it is found that the selectivity to propenol increases with the coordination number of silver atoms over the surfaces, which is consistent with the experimental results reported.

(v) We further find that there would be no difference between the propenol selectivity over all the surfaces when only the Horiuti–Polanyi mechanism is considered for the two hydrogenation steps, since hydrogen dissociation is rate-determining over all the surfaces. In other words, in the current case, the non-Horiuti–Polanyi mechanism is able to give more reasonable explanations for the selectivity trend observed experimentally than the normally used Horiuti–Polanyi mechanism in heterogeneous catalysis.

The alternative reaction pathway reported, i.e. the non-Horiuti–Polanyi pathway, expands the mechanistic understanding on hydrogenation reactions over inert metal surfaces where hydrogen dissociation cannot happen readily.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work is financially supported by ShanghaiTech University, the National Science Foundation of China (21603142), the Shanghai Pujiang Program (16PJ1406800) and the Shanghai Young Eastern Scholar Program (QD2016049). We thank the Shanghai Supercomputer Center for computing time.

References

  1. F. Delbecq, Y. Li and D. Loffreda, J. Catal., 2016, 334, 68–78 CrossRef CAS.
  2. P. Gallezot and D. Richard, Catal. Rev.: Sci. Eng., 1998, 40, 81–126 CAS.
  3. C. J. Kliewer, M. Bieri and G. A. Somorjai, J. Am. Chem. Soc., 2009, 131, 9958–9966 CrossRef CAS PubMed.
  4. S. Tuokko, P. M. Pihko and K. Honkala, Angew. Chem., Int. Ed., 2016, 55, 1670–1674 CrossRef CAS PubMed.
  5. M. Zhao, K. Yuan, Y. Wang, G. Li, J. Guo, L. Gu, W. Hu, H. Zhao and Z. Tang, Nature, 2016, 539, 76–80 CrossRef CAS PubMed.
  6. F. Delbecq and P. Sautet, J. Catal., 1995, 152, 217–236 CrossRef CAS.
  7. F. Delbecq and P. Sautet, J. Catal., 2002, 211, 398–406 CrossRef CAS.
  8. T. B. L. W. Marinelli, S. Nabuurs and V. Ponec, J. Catal., 1995, 151, 431–438 CrossRef CAS.
  9. H. G. Manyar, B. Yang, H. Daly, H. Moor, S. McMonagle, Y. Tao, G. D. Yadav, A. Goguet, P. Hu and C. Hardacre, ChemCatChem, 2013, 5, 506–512 CrossRef CAS.
  10. K. H. Dostert, C. P. O'Brien, F. Ivars-Barcelo, S. Schauermann and H. J. Freund, J. Am. Chem. Soc., 2015, 137, 13496–13502 CrossRef CAS PubMed.
  11. D. Loffreda, F. Delbecq, F. Vigné and P. Sautet, Angew. Chem., Int. Ed., 2005, 44, 5279–5282 CrossRef CAS PubMed.
  12. D. Loffreda, F. Delbecq, F. Vigné and P. Sautet, J. Am. Chem. Soc., 2006, 128, 1316–1323 CrossRef CAS PubMed.
  13. X. Lan, T. Wang, X. Li, N. Huang and J. Wang, Catal. Sci. Technol., 2016, 6, 7703–7707 CAS.
  14. C. Mohr, H. Hofmeister, J. Radnik and P. Claus, J. Am. Chem. Soc., 2003, 125, 1905–1911 CrossRef CAS PubMed.
  15. P. Claus, H. Hofmeister and C. Mohr, Gold Bull., 2004, 37, 3–4 CrossRef.
  16. M. Bron, D. Teschner, A. Knopgericke, B. Steinhauer, A. Scheybal, M. Havecker, D. Wang, R. Fodisch, D. Honicke and A. Wootsch, J. Catal., 2005, 234, 37–47 CrossRef CAS.
  17. M. Bron, D. Teschner, A. Knop-Gericke, F. C. Jentoft, J. Krohnert, J. Hohmeyer, C. Volckmar, B. Steinhauer, R. Schlogl and P. Claus, Phys. Chem. Chem. Phys., 2007, 9, 3559–3569 RSC.
  18. B. Yang, D. Wang, X. Q. Gong and P. Hu, Phys. Chem. Chem. Phys., 2011, 13, 21146–21152 RSC.
  19. B. Yang, X. Q. Gong, H. F. Wang, X. M. Cao, J. J. Rooney and P. Hu, J. Am. Chem. Soc., 2013, 135, 15244–15250 CrossRef CAS PubMed.
  20. C. Mohr, H. Hofmeister and P. Claus, J. Catal., 2003, 213, 86–94 CrossRef CAS.
  21. M. Bron, D. Teschner, U. Wild, B. Steinhauer, A. Knop-Gericke, C. Volckmar, A. Wootsch, R. Schlögl and P. Claus, Appl. Catal., A, 2008, 341, 127–132 CrossRef CAS.
  22. A. B. M. Kok Hwa Lim, I. V. Yudanov, K. M. Neyman, M. Bron, P. Claus and N. Rosch, J. Phys. Chem. C, 2009, 113, 13231–13240 Search PubMed.
  23. P. Claus and H. Hofmeister, J. Phys. Chem. B, 1999, 103, 2766–2775 CrossRef CAS.
  24. H. Wei, C. Gomez, J. Liu, N. Guo, T. Wu, R. Lobo-Lapidus, C. L. Marshall, J. T. Miller and R. J. Meyer, J. Catal., 2013, 298, 18–26 CrossRef CAS.
  25. J. Horiuti and M. Polanyi, Nature, 1934, 134, 377 CrossRef.
  26. J. Horiuti and M. Polanyi, Nature, 1933, 132, 819 CrossRef.
  27. B. Yang, R. Burch, C. Hardacre, P. Hu and P. Hughes, J. Phys. Chem. C, 2014, 118, 1560–1567 CAS.
  28. B. Yang, R. Burch, C. Hardacre, P. Hu and P. Hughes, Surf. Sci., 2016, 646, 45–49 CrossRef CAS.
  29. P. P. Wu and B. Yang, Phys. Chem. Chem. Phys., 2016, 18, 21720–21729 RSC.
  30. K. Yang and B. Yang, Phys. Chem. Chem. Phys., 2017, 19, 18010–18017 RSC.
  31. G. Vilé, D. Baudouin, I. N. Remediakis, C. Copéret, N. López and J. Pérez-Ramírez, ChemCatChem, 2013, 5, 3750–3759 CrossRef.
  32. G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1993, 47, 558–561 CrossRef CAS.
  33. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50 CrossRef CAS.
  34. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186 CrossRef CAS.
  35. G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 49, 14251–14269 CrossRef CAS.
  36. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868 CrossRef CAS PubMed.
  37. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775 CrossRef CAS.
  38. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979 CrossRef.
  39. S. Grimme, J. Phys. Chem. C, 2006, 27, 1787–1799 CAS.
  40. R. M. Ferullo, M. M. Branda and F. Illas, Surf. Sci., 2013, 617, 175–182 CrossRef CAS.
  41. W. A. Al-Saidi, H. Feng and K. A. Fichthorn, Nano Lett., 2012, 12, 997–1001 CrossRef CAS PubMed.
  42. A. Tkatchenko, R. A. DiStasio, R. Car and M. Scheffler, Phys. Rev. Lett., 2012, 108, 236402 CrossRef PubMed.
  43. A. Tkatchenko and M. Scheffler, Phys. Rev. Lett., 2009, 102, 073005 CrossRef PubMed.
  44. D. Yuan, Z. Han, G. Czap, C.-l. Chiang, C. Xu, W. Ho and R. Wu, J. Phys. Chem. Lett., 2016, 7, 2228–2233 CrossRef CAS PubMed.
  45. L. Zibordi-Besse, P. Tereshchuk, A. S. Chaves and J. L. F. Da Silva, J. Phys. Chem. A, 2016, 120, 4231–4240 CrossRef CAS PubMed.
  46. A. Alavi, P. Hu, T. Deutsch, P. L. Silvestrelli and J. Hutter, Phys. Rev. Lett., 1998, 80, 3650–3653 CrossRef CAS.
  47. Z.-P. Liu and P. Hu, J. Am. Chem. Soc., 2003, 125, 1958–1967 CrossRef CAS PubMed.
  48. A. Michaelides, Z. P. Liu, C. J. Zhang, A. Alavi, D. A. King and P. Hu, J. Am. Chem. Soc., 2003, 125, 3704–3705 CrossRef CAS PubMed.
  49. X. Lv, G. Lu, Z.-Q. Wang, Z.-N. Xu and G.-C. Guo, ACS Catal., 2017, 4519–4526 CrossRef CAS.
  50. A. Corma, M. Boronat, S. Gonzalez and F. Illas, Chem. Commun., 2007, 3371–3373 RSC.
  51. NIST: http://webbook.nist.gov/chemistry/, 2016.
  52. S. Laref, F. Delbecq and D. Loffreda, J. Catal., 2009, 265, 35–42 CrossRef CAS.
  53. D. Loffreda, F. Delbecq, F. Vigne and P. Sautet, Angew. Chem., Int. Ed., 2009, 48, 8978–8980 CrossRef CAS PubMed.
  54. X.-M. Cao, R. Burch, C. Hardacre and P. Hu, Catal. Today, 2011, 165, 71–79 CrossRef CAS.
  55. R. H. Ouyang and D. E. Jiang, ACS Catal., 2015, 5, 6624–6629 CrossRef CAS.
  56. J. K. Nørskov, T. Bligaard, A. Logadottir, S. Bahn, L. B. Hansen, M. Bollinger, H. Bengaard, B. Hammer, Z. Sljivancanin, M. Mavrikakis, Y. Xu, S. Dahl and C. J. H. Jacobsen, J. Catal., 2002, 209, 275–278 CrossRef.
  57. R. A. V. Santen, M. Neurock and S. G. Shetty, Chem. Rev., 2010, 110, 2005–2048 CrossRef PubMed.
  58. B. Yang, R. Burch, C. Hardacre, P. Hu and P. Hughes, Catal. Sci. Technol., 2017, 7, 1508–1514 CAS.
  59. B. Yang, X.-M. Cao, X.-Q. Gong and P. Hu, Phys. Chem. Chem. Phys., 2012, 14, 3741–3745 RSC.
  60. B. Yang, R. Burch, C. Hardacre, G. Headdock and P. Hu, J. Catal., 2013, 305, 264–276 CrossRef CAS.
  61. J. Cheng, P. Hu, P. Ellis, S. French, G. Kelly and C. M. Lok, J. Catal., 2008, 257, 221–228 CrossRef CAS.
  62. A. J. Medford, J. Wellendorff, A. Vojvodic, F. Studt, F. Abild-Pedersen, K. W. Jacobsen, T. Bligaard and J. K. Nørskov, Science, 2014, 345, 197–200 CrossRef CAS PubMed.
  63. S. Cao, F. Tao, Y. Tang, Y. Li and J. Yu, Chem. Soc. Rev., 2016, 45, 4747–4765 RSC.
  64. P. M. Mortensen, J.-D. Grunwaldt, P. A. Jensen and A. D. Jensen, Catal. Today, 2016, 259, 277–284 CrossRef CAS.
  65. M. P. Andersson, F. Abild-Pedersen, I. N. Remediakis, T. Bligaard, G. Jones, J. Engbæk, O. Lytken, S. Horch, J. H. Nielsen, J. Sehested, J. R. Rostrup-Nielsen, J. K. Nørskov and I. Chorkendorff, J. Catal., 2008, 255, 6–19 CrossRef CAS.
  66. J. K. Nørskov, T. Bligaard, B. Hvolbaek, F. Abild-Pedersen, I. Chorkendorff and C. H. Christensen, Chem. Soc. Rev., 2008, 37, 2163–2171 RSC.

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c7cy01500c

This journal is © The Royal Society of Chemistry 2017