Theoretical Investigation of Water Gas Shift Reaction Catalyzed by

Feb 6, 2012 - We have investigated the mechanism of M(CO)5 (M = Fe, Ru, Os) ... The water gas shift reaction (WGSR; CO + H2O → H2 + CO2) has played ...
0 downloads 0 Views 3MB Size
Download PDF
Article pubs.acs.org/JPCA

Theoretical Investigation of Water Gas Shift Reaction Catalyzed by Iron Group Carbonyl Complexes M(CO)5 (M = Fe, Ru, Os) Yu Chen,*,†,‡ Fuli Zhang,†,‡ Chunming Xu,†,‡ Jinsen Gao,†,‡ Dong Zhai,†,‡ and Zhen Zhao*,†,§ †

State Key Laboratory of Heavy Oil Processing, ‡College of Chemical Engineering, and §College of Science, China University of Petroleum, Beijing 102249, China S Supporting Information *

ABSTRACT: We have investigated the mechanism of M(CO)5 (M = Fe, Ru, Os) catalyzed water gas shift reaction (WGSR) by using density functional theory and ab initio calculations. Our calculation results indicate that the whole reaction cycle consists of six steps: 1 → 2 → 3 → 4 → 5 → 6 → 2. In this stepwise mechanism the metals Fe, Ru, and Os behave generally in a similar way. However, crucial differences appear in steps 3 → 4 → 5 which involve dihydride M(H) 2 (CO) 3 COOH − (4′) and/or dihydrogen complex MH2(CO)3COOH− (4). The stability of the dihydrogen complexes becomes weaker down the iron group. The dihydrogen complex 4_Fe is only 11.1 kJ/mol less stable than its dihydride 4′_Fe at the B3LYP/ II(f)++//B3LYP/II(f) level. Due to very low energy barrier it is very easy to realize the transform from 4_Fe to 4′_Fe and vice versa, and thus for Fe there is no substantial difference to differentiate 4 and 4′ for the reaction cycle. The most possible key intermediate 4′_Ru is 38.2 kJ/mol more stable than 4_Ru. However, the barrier for the conversion 3_Ru → 4′_Ru is 23.8 kJ/mol higher than that for 3_Ru → 4_Ru. Additionally, 4′_Ru has to go through 4_Ru to complete dehydrogenation 4′_Ru → 5_Ru. The concerted mechanism 4′_Ru → 6_Ru, in which the CO group attacks ruthenium while H2 dissociates, can be excluded. In contrast to Fe and Ru, the dihydrogen complex of Os is too unstable to exist at the level of theory. Moreover, we predict Fe and Ru species are more favorable than Os species for the WGSR, because the energy barriers for the 4 → 5 processes of Fe and Ru are only 38.9 and 16.2 kJ/mol, respectively, whereas 140.5 kJ/mol is calculated for the conversion 4′ → 5 of Os, which is significantly higher. In general, the calculations are in good agreement with available experimental data. We hope that our work will be beneficial to the development and design of the WGSR catalyst with high performance.

■

INTRODUCTION The water gas shift reaction (WGSR; CO + H2O → H2 + CO2) has played an important role in the chemical industry field. It can be used to remove poisonous CO from reformate hydrogen gas for fuel cell applications,1 and to enrich the content of H2 in water gas (synthesis gas).2 As a matter of course, the relevant catalysts are of central importance. Because of their mild conditions, transition metal carbonyls as homogeneous catalysts have attracted great attention. One of the most extensively explored mechanisms, both experimentally3 and theoretically,4 is for Fe(CO)5 catalyzed WGSR. In 1993, based on experimental results, Sunderlin and Squires5 postulated a catalytic cycle. A series of species such as (CO) 4 FeCOOH − , (CO)4FeH−, Fe(CO)4, and (CO)4FeH2 was proposed therein as possible intermediates. The theoretical pioneering work from Torrent, Solà, and Frenking6 on the complete mechanism of Fe(CO)5 catalyzed WGSR inspired further continuous investigations. The most significant improvements recently done by Barrows7 and Rozanska and Vuilleumier8 provided deeper insights into the WGSR mechanism. Since then, a complete and detailed description of Fe(CO)5 catalyzed WGSR cycle seemed to have taken shape. © 2012 American Chemical Society

Nevertheless, we found that some other paths should also be considered. Due to the small conformation difference between reactants/products and transition states, one or more steps for certain processes relevant to the mechanism should proceed. Therefore, we revisited the most up-to-date mechanism from Rozanska and Vuilleumier,8 and we proposed a revised catalysis reaction cycle (see Scheme 1).9 Since that new proposal can better explain the whole mechanism of Fe(CO)5 catalyzed WGSR, it would be significant to extend our study from Fe(CO)5 to other iron group members, i.e., M(CO)5, where M stands for Ru and Os, to shed light on the mechanism of homogeneous WGSR catalyzed by transition metal carbonyl complexes.

■

COMPUTATIONAL DETAILS The geometries have been optimized at the B3LYP10 level. A small-core effective core potential (ECP) with a (441/2111/N1) split-valence basis set has been employed for the metals, where Received: May 23, 2011 Revised: February 4, 2012 Published: February 6, 2012 2529

dx.doi.org/10.1021/jp204776a | J. Phys. Chem. A 2012, 116, 2529−2535

The Journal of Physical Chemistry A

■

Scheme 1. New Proposed Fe(CO)5-Catalyzed WGSR Cycle,9 Which Can Be Extended to Ru and Os with Some Tiny Differences in Complex 4a

Article

RESULTS AND DISCUSSION

All the intermediates (1−6) included in the catalysis cycle proposed for M(CO)5 catalyzed WGSR (M = Fe, Ru, Os) were fully optimized at the B3LYP/II(f) level; their geometries are shown in Figure 1. Figure 2 shows the calculated structures of transition states for the conversion between those species. TSr/ p_M denotes the transition state structure, where r and p are numbers of the reactant and product species, respectively, and M represents the metal atom (Fe, Ru, Os). The reaction energy profile connecting reactants, transition states, and products is described in Figure 3. Next we will discuss the reaction mechanism step by step along the cycle of 1 → 2 → 3 → 4 → 5 → 6 → 2 (see Scheme 1). Step I. M(CO)5 (1) + OH− → M(CO)4COOH− (2). It is wellknown that under basic condition M(CO)5 can react with OH−, generating a key reactive intermediate carboxylic acid anion M(CO)4COOH−. It is exciting to see that our calculations reproduced all the geometric characters observed in the experiments for this reaction. First, let us take a look at the geometries of pentacarbonyl M(CO)5 (M = Fe, Ru, Os) (1). For Fe(CO)5 (1_Fe), the axial Fe−CO bonds (Fe−CO(ax)) are calculated to be 1.818 Å, which is longer than the equatorial bonds (Fe−CO(eq) = 1.805 Å). The agreement of these results with the most recent experimental measurement (Fe−CO(ax) = 1.811(2) Å, Fe−CO(eq) = 1.803(2) Å18) is excellent. For Ru(CO)5 (1_Ru) and Os(CO)5 (1_Os), experimental data showed that the axial Ru−CO(ax) distances (1.941(13) Å) are shorter than those of the equatorial bonds (1.961(13) Å), while the axial Os−CO(ax) bonds (1.982(20) Å) are longer than the equatorial Os−CO(eq) bonds (1.937(19) Å).19 Our calculation conducted at the B3LYP/II(f) level predicts Ru−CO(ax) = 1.974 Å, Ru− CO(eq) = 1.980 Å, Os−CO(ax) = 1.976 Å, and Os−CO(eq) = 1.961 Å. Obviously, these results demonstrate the same trend as the experiment. To understand the interesting situation of Ru− CO bond lengths in 1_Ru, we have also performed NBO20 analysis of the triad 1_Fe, 1_Ru, and 1_Os. Back-donations of Ru(CO)4 → L(π) in 1_Ru are calculated to be 0.252 e for axially coordinated and 0.271 e for equatorially coordinated ligand L (L = CO), respectively (see Table S1 in the Supporting Information). The corresponding values for 1_Fe and 1_Os are all larger than 0.3 e. These results indicate that 1_Ru has the weakest π-back-donation from transition metal fragment M(CO)4 to L, where M = Fe, Ru, and Os and L = axial or equatorial CO. Although roughly longer Ru−CO distances in 1_Ru in comparison with M−CO distances in 1_Fe and 1_Os could be interpreted in this way, a definite answer can still not be given to the relative bond lengths of axial and equatorial Ru−CO. Following one reviewer’s suggestion, we compared the stabilization energies between the CO(ax)/CO(eq) and metal atom. Size consistent MP2 calculations (Supporting Information, Table S3) also showed a behavior of Ru(CO)5 different from that of its cognates. The order of average stabilization energy of single CO is axial CO (256.9 kJ/mol) > equatorial CO (214.2 kJ/mol) for Ru(CO)5, while it is the other way in the cases of Fe(CO)5 and Os(CO)5. All short M−CO bonds can contribute more to the stabilization of the corresponding complex. From this standpoint, the difficulty of explaining the relative M−CO bond lengths seems to be overcome. By the way, the interaction between M and five COs (CO)5 makes Ru(CO)5

a

For Os there is only dihydride while Fe and Ru have both dihydride and dihydrogen complexes. 6 is virtually an isomer of 2. See text for details.

N = 4, 3, and 2 for Fe, Ru, and Os, respectively. This basis set was derived from the (55/5/5) minimal basis set optimized by Hay and Wadt.11 For other atoms such as C, O, and H we applied 6-31G(d,p) basis sets. This combination is the standard basis set II of Frenking's group.12 Our previous calculations13 have showed that this basis set can well reproduce the structures for a series of Fe(CO)4L complexes, where L stands for varieties of ligands. For a better description, an extra f-function14 was further augmented into metal (denoted as II(f) to make it explicit). Single point energy calculations of B3LYP/II(f)-optimized geometries were carried out at the B3LYP level with a larger basis set II(f)++ for all reaction steps involved in the reaction mechanism. Basis set II(f)++ is the same as II(f) plus the addition of an s diffuse function on H and a set of three sp diffuse functions on C and O atoms, i.e., 6-31++G(d,p). All energy values are given at the B3LYP/II(f)++//B3LYP/II(f) level with zero-point-energy (ZPE) corrections if not specified. For selected species, we have also carried out improved energy calculations at the B3LYP/II(f) optimized geometries by using coupled-cluster theory at the CCSD(T)15/II(f)++ level. The nature of the stationary points was examined by vibrational analysis. The intrinsic reaction coordinate calculation (IRC)16 was performed with all transition states to confirm whether these transition states connect to the right minima. All calculations have been carried out with the program package Gaussian 03.17 2530

dx.doi.org/10.1021/jp204776a | J. Phys. Chem. A 2012, 116, 2529−2535

The Journal of Physical Chemistry A

Article

Figure 1. Geometries of selected intermediate species included in the M(CO)5/OH− catalyzed WGSR (M = Fe, Ru, Os). Bond distances are in Å.

at the MP2/II level.21 These excellent calculation performances prove that the theoretical method applied in this study is reliable. Second, the calculation results demonstrate that metallocarboxylic acid M(CO)4COOH− (2) (shown in Figure 1) is the most

obviously different as well, 129−281 kJ/mol less stable than Fe(CO)5 and Os(CO)5. It is worthwhile to add a note that these relative axial and equatorial bond lengths were also verified in the previous study conducted 2531

dx.doi.org/10.1021/jp204776a | J. Phys. Chem. A 2012, 116, 2529−2535

The Journal of Physical Chemistry A

Article

Figure 2. Geometries of transition states included in the M(CO)5/OH− catalyzed WGSR (M = Fe, Ru, Os). Bond distances are in Å.

stable structure, where the COOH− unit occupies the axial position in the parent complex M(CO)5 (1), which can be ascribed to the negative charge of COOH−. The result is identical to our previous prediction.13b The coordination of COOH− has caused some changes of M−CO bond lengths in complex 2: both M−CO(ax) and M−CO(eq) bonds become shorter, while C−O(ax) and C−O(eq) bonds elongate slightly. The most interesting change is the relative bond lengths of Fe− COs in 2_Fe: Fe−CO(ax) bond is now shorter than Fe− CO(eq) bonds. In addition, the M−COOH− bond distances show a monotonic increase down the group, i.e., 2.051, 2.177, and 2.201 Å for the iron triad 2_Fe, 2_Ru, and 2_Os, respectively. This can be ascribed to the relative order of atom radii in the iron group (Fe, 1.56 < Ru, 1.78 ≤ Os, 1.85 Å)22 and to the well-known effect of lanthanide contraction. Last but not least, our calculation confirms that step I is a strong exothermal and no-barrier reaction for all three metals.

Regarding the reaction heat effect ΔH, there is no obvious difference between metals, i.e., Fe, −445.2; Ru, −456.7; and Os, −464.5 kJ/mol. These results coincide well with experimental data. Step II. M(CO)4COOH− (2) → M(CO)4H− (3) + CO2. By releasing a molecule of CO2, complex 2 transforms via TS2/3 into metal hydride anion M(CO)4H− (3). Although the energy barrier for that transformation is rather high, 123.0, 123.4, and 124.7 kJ/mol for Fe, Ru, and Os, respectively, it should be able to proceed because the exothermicity of the forgoing process 1 → 2 is very strong (more than 400 kJ/mol). Synchronously, the energy of products 3 is found to be 36.0 (3_Fe), 32.7 (3_Ru), and 39.7 (3_Os) kJ/mol lower than that of the reactants 2, respectively. Up to this step, no substantial difference is observed between these metal cognates in term of energetics. Like COOH− in metallocarboxylic acid M(CO)4COOH−, the negatively charged H− in complex 3 also locates in the axial 2532

dx.doi.org/10.1021/jp204776a | J. Phys. Chem. A 2012, 116, 2529−2535

The Journal of Physical Chemistry A

Article

energy than TS3/4_Ru. Hence, 4′_Ru is thermodynamically more stable than 4_Ru, but dynamically not favorable to generate due to the higher energy barrier. In addition, because the relative stability of 4′_Ru is higher, the next step, i.e., H2 releasing, will be more difficult for 4′_Ru to take place. In other words, the conversion of 4′_Ru → 5_Ru needs more energy than 4_Ru → 5_Ru, at least 38.2 kJ/mol more energy theoretically. For the sake of clarity, the process 3_Ru → 4′_Ru and the succeeding process 4′_Ru → 4_Ru are shown in place of 3_Ru → 4_Ru in Figure 3 (see also the following discussion about step IV below). As for Os, there is no ambiguousness because our calculation results reveal that there is no dihydrogen complex. That is, for osmium the only possible form within the frameworks of both 4 and 4′ is dihydride complex. The local minimum 4′_Os has C1 symmetry with an H−H distance of 2.179 Å, and its COOH − configuration is similar to 4_Fe and 4_Ru intermediates. Additional calculations suggest that the global minimum is 4′_Os_b, which is only 8.3 kJ/mol more stable than 4′_Os, indicating that an interconversion between 4′_Os and 4′_Os_b is feasible (see also Figure S2 in the Supporting Information). Herein the pathways connected with 4′_Os are given in Figure 3 for a comparison with the cases of Fe and Ru. In summary, which isomer is more stable, dihydrogen or dihydride complex, correlates with the metals. Down the iron group, dihydrogen complexes become less stable while dihydrides become dominative. This trend can be well explained by the property of d electrons in the outermost shell for these metals. The relative energy of d electrons involved in back-donation to the σ* orbital of the dihydrogen ligand increases down the iron group (5d > 4d > 3d). Besides, we can characterize the ability of that back-donation with charge decomposition analysis (CDA)23 of the complexes 4_Fe and 4_Ru (Table 1). The back-

Figure 3. Energy profile of gas-phase WGSR catalyzed by M(CO)5/ OH− (M = Fe, Ru, Os) in Scheme 1, calculated at the level of B3LYP/ II(f)++ including ZPE contributions at the level of B3LYP/II(f) (in kJ/mol). Only key intermediates are indicated.

position. In the meanwhile, one of the four CO ligands is trans to H− and the other three are bent toward H−. This situation is the same as in Fe(CO)4CN− or Fe(CO)4NC−.13b Due to the coordination of H− to M(CO)4, the M−CO bonds become obviously shorter than those in parent 1, and the axial M−CO bonds are longer than the equatorial ones. These trends are identical for all three metals. Step III. M(CO)4H− (3) + H2O → M(CO)3H2COOH− (4, Dihydrogen Complex) or M(CO)3(H)2COOH− (4′, Dihydride). For this step, we first take a look at the case of iron. Our previous CCSD(T)//B3LYP/II study13b without an extra f-function for Fe has shown that for the dihydrogen complex Fe(CO)4H2 the equatorial structure is preferred over the axial one, and the H2 group is η2 bound to the metal center. Not surprisingly, in the dihydrogen complex Fe(CO)3H2COOH (4_Fe), which is equivalent to replacing one CO with COOH− in Fe(CO)4H2, H2 also stays in the equatorial position. The H− H distance of 0.879 Å herein is comparable with that of the free hydrogen molecule (0.743 Å). Another feature here is that the relative energy difference between 4_Fe and its isomer dihydride Fe(CO)3(H)2COOH (4′_Fe) is ignorable: 4_Fe is calculated to be 11.1 kJ/mol less stable than 4′_Fe, and the conversion from 4_Fe into 4′_Fe is also easy because the corresponding TS4/4′_Fe is only 2.5 and 5.0 kJ/mol above 4_Fe at the B3LYP/II(f)++ and CCSD(T)/II(f)++ levels, respectively (see Figure S1 and Table S2 in the Supporting Information). The reversed process 4′_Fe → 4_Fe proceeds rather easily as well because the corresponding energy barrier is only 13.6 and 27.5 kJ/mol above 4′_Fe, respectively. Therefore, the interconversion between the two isomers 4_Fe and 4′_Fe is feasible, thermodynamically as well as kinetically. This result indicates that it makes no sense for the Fe system to further differentiate which of these two forms, 4_Fe or 4′_Fe, is the real participator in the process 3 → 4 → 5. In the case of Ru, the difference of relative energy between the dihydride (4′_Ru) and the dihydrogen complex (4_Ru) is magnified: 4′_Ru is calculated to be 38.2 kJ/mol more stable than its isomer 4_Ru. The energy barrier for the conversion from 4_Ru to 4′_Ru via TS4/4′_Ru is very low, only 1.7 and 4.9 kJ/mol at the B3LYP/II(f)++ and CCSD(T)/II(f)++ levels, respectively (see Figure S1 and Table S2 in the Supporting Information). Thus, 4′_Ru should be considered as a key intermediate for step III. However, the associated transition state TS3/4′_Ru (Figure 2) is 23.8 kJ/mol higher in

Table 1. CDA Result for Dihydrogen Complexes M−L (M = [Fe(CO)3COOH]− and [Ru(CO)3COOH]−, L = H2) species

da

bb

rc

Δd

Fe Ru

0.384 0.357

0.252 0.209

−0.230 −0.229

0.003 0.016

L → M σ-donation. bL ← M π-back-donation. cL ↔ M repulsive polarization. dResidual term Δ.

a

donation from metal fragment [M(CO)3COOH]− to H2 ligand for Ru (0.209 e) is weaker than that for Fe (0.252 e). This phenomenon was also observed for metals of group 6, in which the back-donations H2 ← M(CO)5 calculated at the MP2/II level reduced in complexes [M(CO)5]H224 with Cr (0.143 e) > Mo (0.105 e). For additional background information on metal− dihydrogen and σ-bond coordination, readers are referred to a comprehensive work reported by Kubas.25 Now, back to the whole pathway 3 → 4 or 3 → 4′, the situations become more complicated, structurally and energetically. The activation barriers for adding water to 3 are rather large. The smallest one among the three cognates is for Ru: 143.5 kJ/mol for 3 → 4. The corresponding values for Fe and Os are 164.0 and 166.5 kJ/mol, respectively. In this regard, Ru catalyst is kinetically a bit more preferable for step III of the WGSR. Considering the pathway 3 → 4′ of Ru, however, this preference in kinetics disappears. This is because the barrier height turns into 167.3 kJ/mol, which is in the same order as the cases of Fe and Os. More interesting is that for Fe and Ru endothermic reactions 3 → 4 with ΔH = 72.8 and 53.6 kJ/mol, 2533

dx.doi.org/10.1021/jp204776a | J. Phys. Chem. A 2012, 116, 2529−2535

The Journal of Physical Chemistry A

Article

character. Also, the reaction barrier of 140.5 kJ/mol is significantly higher than the corresponding values for Fe (4_Fe → 5_Fe; 38.9 kJ/mol) and Ru (4_Ru → 5_Ru; 16.7 kJ/mol). This result indicates that the dehydrogenation of dihydride complexes occurs in general with more difficulty than that of dihydrogen complexes, especially for the complexes of the iron group. Step V. M(CO)3COOH− (5) + CO → M(CO)4COOH− (6). Upon the approach of a new CO, two equatorial COs in 5_Fe tend to go away in a direction opposite to the invading group. During this process the configuration of the COOH− group stays unchanged. Also, a 17.3 kJ/mol energy barrier is easy to overcome for the process from 5_Fe via TS5/6_Fe to 6_Fe. For Ru, the reactant, TS, and product are all in Cs or Cs-like symmetry which indicates that the process proceeds, similar to the situation of Fe, in a plane. The reaction barrier for Ru is also very low, only 7.3 kJ/mol. As for Os, TS5/6_Os has a structure without any symmetry (C1), and the energy barrier (8.8 kJ/mol) is in the same order as that of Ru. It is obvious that there is no significant difference between Fe, Ru, and Os regarding the energy barrier in this step. Step VI. Isomerization of M(CO)4COOH− (6 → 2). The product of step V, 6, isomerizes into 2 in a way out-of-plane for the hydrogen atom of the COOH− group. The conversion from 6_Fe to 2_Fe with an energy barrier of 43.5 kJ/mol occurs rather easily. Likewise, 49.9 and 51.3 kJ/mol are obtained for the cases of Ru and Os, respectively, indicating that there is no ultimate difference between the three metals for step VI. By and large, our calculation results demonstrate that step IV is the key step in the proposed stepwise mechanism for WGSR catalyzed by M(CO)5. Down the iron group, the trend to form dihydride becomes stronger. The relative energy difference between dihydrogen complex and dihydride for Fe is very small. Ru’s dihydride turns out to be even more stable than its dihydrogen complex. For Os there is no dihydrogen complex at all; only dihydride exists at the theoretical level. In view of the barrier height (Figure 3), both Fe and Ru are preferable for the whole catalysis cycle while Os has the weakest catalytic effect. This conclusion is consistent with the experimental result; i.e., only Fe(CO)5 (1_Fe) and Ru(CO)5 (1_Ru) are proven to be possible metal carbonyl catalysts in the WGSR.6,26

respectively, are predicted, and for Os it is calculated to be an exothermic reaction 3 → 4′ with ΔH = −11.3 kJ/mol. That is, Os species is thermodynamically favorable for step III in the WGSR. Step IV. M(CO)3H2COOH− (4) or M(CO)3(H)2COOH− (4′) → M(CO)3COOH− (5) + H2. Our recent study9 with pentacarbonyl iron has demonstrated that the transition state TS4/5_Fe, as depicted in Figure 2, links directly to 4_Fe and 5_Fe (Figure 1). The dehydrogenation occurs in a single step while the COOH− subunit keeps its configuration unchanged. An energy barrier of 38.9 kJ/mol needs to be surmounted for this process. For the case of Ru, the more stable 4′_Ru should be able to remove two hydrogen atoms or one hydrogen molecule to produce 5_Ru. However, it is found by calculations that the direct dehydrogenation process is impossible to complete. In other words, the process of 4′_Ru → 4_Ru must occur before the dehydrogenation. Afterward, the less stable 4_Ru can easily convert into the likewise less stable 5_Ru. The fact that 5_Ru, depicted in Figure 1, is 8.1 kJ/mol higher in energy than the most stable isomer of 5_Ru_c (see Figure S3 in the Supporting Information) makes further process even more feasible. It should be pointed out that without considering ZPE contributions it is impossible to obtain a precise relative energy between TS4/5_Ru and the product (H2 plus 5_Ru) since the product is abnormally 0.6 and 0.0 kJ/mol less stable than TS4/ 5_Ru at the B3LYP/II(f) and B3LYP/II(f)++ levels, respectively. With ZPE corrections the calculation results at these two levels agree very well, namely, the product is now correctly below TS4/5_Ru on the potential energy surface with the latter being 4.9 and 5.5 kJ/mol less stable at the theoretical levels, respectively. These abnormal values of 0.6 and 0.0 kJ/mol mentioned above can also be attributed to the grid-based DFT calculations. Replacing default grid 75302 with ultrafine grid 99590, we can acquire that at the B3LYP/II(f)++ level the product 5_Ru + H2 lies 0.05 kJ/mol lower than TS4/5_Ru even without ZPE correction. Meanwhile, the transformation from 4_Ru to 5_Ru requires only 16.2 kJ/mol energy. That is, in terms of reaction barrier this step also proceeds easily. We have also explored the possibility of whether another alternative pathway from 4′_Ru to 6_Ru proceeds through a concerted mechanism, in which the CO group attacks ruthenium while two hydrogen atoms dissociate. Varieties of initial relative configurations between attacking CO group and 4′_Ru are designed to obtain a valid TS from 4′_Ru directly to 6_Ru. CO can approach 4′_Ru from the side either the same as or opposite to the departing H atoms. One located TS structure is illustrated in Figure S4 in the Supporting Information, which is similar to the transition state obtained by Barrows7 for the possible reaction CO + Fe(CO)4H2 → Fe(CO)5 + H2. IRC analysis shows that our TS, resembling TS4/5_Ru, actually connects CO + 4_Ru as reactant and CO + 5_Ru + H2 as product. This result indicates that two hydrogen atoms have to interact with each other to form a hydrogen molecule and then move away while incoming CO does not play any role. Therefore we can say that the possibility for a concerted mechanism of Ru can be excluded. In contrast to Fe and Ru, for which the dihydrogen complexes exist, osmium goes through step 4 → 5 only via dihydride form. The dihydride 4′_Os has a longer H−H bond distance (2.179 Å) than 4′_Ru (2.107 Å). The dehydrogenation process proceeds from 4′_Os to 5_Os via TS4′/5_Os. TS4′/ 5_Os is found to have a C1 symmetry and more product-like

■

CONCLUSIONS

We have investigated the mechanism of M(CO)5 catalyzed WGSR where M is an element of the iron group. The calculation results reveal that the difference in relative stability between dihydride and dihydrogen complex of Fe, Ru, and Os causes their final behavior deviation in catalysis. 1_Fe and 1_Ru can catalyze the WGSR while 1_Os does not have an effective catalysis function in the reaction cycle. Although the water coordination process is thermodynamically more favored for the case of Os (ΔH = −11.3 kJ/mol) over Fe (72.8 kJ/mol) and Ru (53.6 kJ/mol) in step III, the fundamental trend remains that WGS catalyzed by carbonyl complex of Fe or Ru is dynamically more favorable over Os. This is mainly due to the very high activation barrier for Os species in step IV. In comparison with the energy costs of 8.9 (Fe) and 16.7 kJ/mol (Ru), the barrier of 140.5 kJ/mol for Os species is too high to overcome. This theoretical prediction agrees excellently with available experimental results. 2534

dx.doi.org/10.1021/jp204776a | J. Phys. Chem. A 2012, 116, 2529−2535

The Journal of Physical Chemistry A

■

Article

J. Chem. Phys. 1987, 86, 7041−7050. (e) Pople, J. A.; Head-Gordon, M.; Raghavachari, K. J. Chem. Phys. 1987, 87, 5968−5975. (16) (a) Fukui, K. Acc. Chem. Res. 1981, 14, 363−368. (b) Gonzalez, C.; Schlegel, H. B. J. Chem. Phys. 1989, 90, 2154−2161. (c) Gonzalez, C.; Schlegel, H. B. J. Phys. Chem. 1990, 94, 5523−5527. (17) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Pople, J. A.; et al. Gaussian 03, revision E.01; Gaussian, Inc.: Wallingford, CT, 2004. See the Supporting Information for the full citation. (18) Braga, D.; Grepioni, F.; Orpen, A. G. Organometallics 1993, 12, 1481−1483. (19) (a) Huang, J.; Hedberg, K; Davis, H. B.; Pomeroy, R. K. Inorg. Chem. 1990, 29, 3923−3925. (b) Huang, J.; Hedberg, K.; Pomeroy, R. K. Organometallics 1988, 7, 2049−2053. (20) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88, 899−926. (21) Ehlers, A. W.; Frenking, G. Organometallics 1995, 14, 423−426. (22) (a) Dean, J. A. Lange’s Handbook of Chemistry, 15th ed.; McGraw-Hill: New York, 1999. (b) Kelly, A.; Groves, G. W. Crystallography and Crystal Defects; Longman: London, 1970. (23) (a) Dapprich, S.; Frenking, G. J. Phys. Chem. 1995, 99, 9352− 9362. (b) Dapprich, S.; Frenking, G. CDA 2.1; Marburg, 1994. The program is available via anonymous ftp server: ftp.chemie.uni-marburg. de/pub/cda. (24) Dapprich, S.; Frenking, G. Organometallics 1996, 15, 4547− 4551. (25) Kubas, G. J. J. Organomet. Chem. 2001, 635, 37−68. (26) (a) Gross, D. C.; Ford, P. C. Inorg. Chem. 1982, 21, 1702−1704. (b) Kang, H.; Mauldin, C. H.; Cole, T.; Slegeir, W.; Cann, K.; Petit, R. J. Am. Chem. Soc. 1977, 99, 8323−8325.

ASSOCIATED CONTENT

S Supporting Information *

The optimized geometries of isomers for Fe, Ru, and Os species (Figures S1−S4), result of the NBO analysis of 1 (Table S1), calculated total energies and relative energy at the CCSD(T)/II(f)++ and B3LYP/II(f)++ levels of related species (Table S2), stabilization energies of axial and equatorial COs for M(CO)5 where M = Fe, Ru, Os (Table S3), full citation of ref 17. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Tel.: +86-10-89731072. Fax: +86-10-69724721. E-mail: chenyu@ cup.edu.cn (Y.C.); [email protected] (Z.Z.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We would like to thank the National Natural Science Foundation of China (Nos. 20873180, 20525621, 21036008), the Scientific Research Foundation for the Returned Overseas Chinese Scholars by Ministry of Education, and the Program of the State Key Laboratory of Heavy Oil Processing (Nos. 20064, 2008-4) for financial support.

■

REFERENCES

(1) (a) Vielstich, W.; Lamm, A.; Gasteiger, H. A. Handbook of Fuel Cells: Fundamentals, Technology, Applications; Wiley: Chichester, U.K., 2003. (b) Oetjen, H. F.; Schmidt, V. M.; Stimming, U.; Trila, F. J. Electrochem. Soc. 1996, 143, 3838−3842. (2) (a) Thomas, J. M.; Thomas, W. J. Principles and Practice of Heterogeneous Catalysis; VCH: New York, 1997. (b) Catalyst Handbook, 2nd ed.; Twigg, M. V., Ed.; Wolfe: London, 1989. (3) (a) Kang, H.; Mauldin, C.; Cole, T.; Slegeir, W.; Pettit, R. J. Am. Chem. Soc. 1977, 99, 8323−8325. (b) Grice, N.; Kao, S. C.; Pettit, R. J. Am. Chem. Soc. 1979, 101, 1627−1628. (c) Lane, K. R.; Lee, R. E.; Sallans, L.; Squires, R. R. J. Am. Chem. Soc. 1984, 106, 5767−5772. (4) Torrent, M.; Solà, M.; Frenking, G. Chem. Rev. 2000, 100, 439− 493. (5) Sunderlin, L. S.; Squires, R. R. J. Am. Chem. Soc. 1993, 115, 337− 343. (6) Torrent, M.; Solà, M.; Frenking, G. Organometallics 1999, 18, 2801−2812. (7) Barrows, S. E. Inorg. Chem. 2004, 43, 8236−8238. (8) Rozanska, X.; Vuilleumier, R. Inorg. Chem. 2008, 47, 8635−8640. (9) Zhang, F.; Zhao, L.; Xu, C.; Chen, Y. Inorg. Chem. 2010, 49, 3278−3281. (10) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652. (b) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. 1988, B37, 785−789. (11) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299−310. (12) Frenking, G.; Antes, I.; Böhme, M.; Dapprich, S.; Ehlers, A. W.; Jonas, V.; Neuhaus, A.; Otto, M.; Stegmann, R.; Veldkamp, A.; Vyboishchikov, S. F. In Reviews in Computational Chemistry; Lipkowitz, K. B., Boyd, D. B., Eds; VCH: New York, 1996; Vol. 8, pp 63−144. (13) (a) Chen, Y.; Petz, W.; Frenking, G. Organometallics 2000, 19, 2698−2706. (b) Chen, Y.; Hartmann, M.; Frenking, G. Z. Anorg. Allg. Chem. 2001, 627, 985−998. (14) Ehlers, A. W.; Böhme, M.; Dapprich, S.; Gobbi, A.; Höllwarth, A.; Jonas, V.; Köhler, K. F.; Stegmann, R.; Veldkamp, A.; Frenking, G. Chem. Phys. Lett. 1993, 208, 111−114. (15) (a) Pople, J. A.; Krishnan, R.; Schlegel, H. B.; Binkley, J. S. Int. J. Quantum Chem. 1978, 14, 545−560. (b) Bartlett, R. J.; Purvis, G. D. Int. J. Quantum Chem. 1978, 14, 561−581. (c) Purvis, G. D.; Bartlett, R. J. J. Chem. Phys. 1982, 76, 1910−1918. (d) Purvis, G. D.; Bartlett, R. J. 2535

dx.doi.org/10.1021/jp204776a | J. Phys. Chem. A 2012, 116, 2529−2535