Structure–Activity Relationship To Screen Ni–Bisphosphine

Structure–Activity Relationship To Screen Ni–Bisphosphine Complexes for the Oxidative Coupling of CO2 and ... Publication Date (Web): March 7, 201...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/Organometallics

Structure−Activity Relationship To Screen Ni−Bisphosphine Complexes for the Oxidative Coupling of CO2 and Ethylene Miasser Al-Ghamdi,†,‡ Sai Vikrama Chaitanya Vummaleti,† Laura Falivene,† Farhan Ahmad Pasha,‡ Dirk J. Beetstra,‡ and Luigi Cavallo*,† †

KAUST Catalysis Center, Physical Sciences and Engineering Division, King Abdullah University of Science and Technology, Thuwal 23955-6900, Saudi Arabia ‡ SABIC Corporate Research and Development, Thuwal 23955-6900, Saudi Arabia S Supporting Information *

ABSTRACT: Density functional theory calculations have been used to investigate competition between inner- and outersphere reaction pathways in the oxidative coupling of CO2 and ethylene for a set of 12 Ni−bisphosphine complexes, in order to build a QSAR approach correlating catalyst structure to calculated energy barriers for CO2 activation. The ligands were selected to explore different substituents on the P atoms (cyclohexyl, phenyl, and tert-butyl) and different lengths of the tether connecting the P atoms, −(CH2)n− with n = 1−3. As expected, the conclusion is that the inner-sphere reaction pathway is favored with unhindered ligands, while the outer-sphere reaction pathway is favored with hindered ligands. To find a possible correlation with molecular descriptors, we started using the buried volume as a steric descriptor. A reasonable correlation could be found for the energy barrier along the inner-sphere pathway, while scarce correlation was found for the energy barrier along the outer-sphere pathway, indicating that the steric bulkiness of the ligand disfavors approach of CO2 to the metal center. Much stronger correlation between the ligand structure and the energy barrier along the inner-sphere pathway was achieved when the steric descriptor was augmented by an electronic descriptor, consisting of the partial charge on the Ni atom. The much better correlation suggests that bisphosphine ligands have a non-negligible electronic impact on the catalyst performance.



ethylene to form a nickelalactone and (2) β-H elimination from the nickelalactone to release acrylate species. Buntine and coworkers reported DFT calculations rationalizing Hoberg’s experimental findings. During addition to ethylene, CO2 was suggested to interact with the metal center.15 This reaction pathway was defined later as being inner sphere.5 Over the past three decades, several Ni(0)-based catalysts containing bisamine or bisphosphine ligands have been proposed for this reaction. Ni(0)−bisphosphine complexes comprising the 1,2-bis(di-tert-butylphosphino)ethane ligand, dtbpe, are one of the few systems capable of promoting the coupling of CO2 and ethylene to a stable metallalactone, and mechanistic information for this system is also available in the literature.8,9,16−18 In the case of Ni(0) complexes, formation of the nickelalactone B (see Scheme 1) is generally accepted to start from the stable Ni−ethylene complex A, which has been observed experimentally, followed by insertion of a CO2 molecule into the Ni−C bond in a single step. On the basis of quantum mechanics calculations, Plessow et al. proposed an alternative mechanism for metallalactone formation that competes with the conventional mechanism described in the literature.5 In this

INTRODUCTION

Utilization of CO2 as an economically convenient and abundantly available C1 building block in the production of commodity chemicals is a topic attracting strong interest in the chemical industry.1−11 Nevertheless, the chemical inertness of CO2 is preventing the rapid development of large-volume processes based on the transformation of CO2 into higher value chemicals.12 Among the possible transformations, the transition-metal-promoted coupling of CO2 with olefins to produce acrylate1−11 is regarded as one of the dream reactions in catalysis (see Chart 1). This chemistry needs an efficient and active catalyst which helps to reduce the high activation energy demand for CO2.13 In the 1980s, Hoberg and co-workers first reported the synthesis of acrylates from CO2 and ethylene using Ni(0)-based catalysts.14 The proposed catalytic cycle consists of two important steps: (1) oxidative coupling of CO2 and Chart 1. Acrylic Acid Formation via Coupling of Ethylene and Carbon Dioxide Mediated by a Transition-Metal (TM) Catalyst

Received: December 6, 2016

© XXXX American Chemical Society

A

DOI: 10.1021/acs.organomet.6b00905 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

investigate these points, in the present study we expanded the valuable work of Limbach5 and of Pidko24 by further investigating the role of steric bulkiness and electronic effects on the behavior of Ni(0)−bisphosphine-based catalysts by considering consistent bisphosphine ligand subsets with either cyclohexyl (L1−L3) or phenyl substituents (L4−L6) at the phosphorus atoms and varying lengths of the carbon bridge, −(CH2)n− with n = 1−3, in the ligand backbone (see Figure 1). For completeness and better connection with earlier work,5,24 we also included tert-butyl substituents (L7−L9). The ligands L1−L9 were selected to vary a single parameter at a time, the bridge length or the substituent on the P atoms. Further, to give more steric and electronic variety to the examined set of ligands, we included in the set a bis(dicyclohexylphosphino)ferrocene ligand (dcpf, L10), a bisphosphine ligand presenting an aromatic linker (L11), and the bis(di-tert-butylphosphine)dimethylxanthene ligand (L12) (see Figure 1). Since steric and electronic properties are important parameters for tuning the activity and selectivity of organometallic catalysts, we chose the proposed ligand set to have a systematic variation in the steric and electronic environment around the Ni metal center. As a steric descriptor we used % VBur, initially developed for N-heterocyclic carbenes and later used to characterize the steric bulkiness of a variety of other ligands as well.25,26 As an electronic descriptor we used the simplest descriptor, which is the Mulliken charge on the Ni atom. As mentioned above, we computed the free energy surface for the oxidative coupling of CO2 and ethylene reaction catalyzed by Ni-based complexes based on ligands L1−L12, according to the mechanism shown in Scheme 1. In the spirit of a QSAR approach, the computed free energy barriers will be correlated with the steric and electronic descriptors mentioned above. The overall goal is to develop a correlation protocol, on the basis of descriptors that can be in principle applied to basically any family of metal-based catalysts, for the coupling of CO2 and olefins. Of course, application of the protocol to another family of catalysts will require proper validation.

Scheme 1. General Scheme for the Competitive Oxidative Coupling of CO2 and Ethylene (Inner- and Outer-Sphere Pathways) for Ni(0)-Based Catalysts

alternative reaction pathway, denoted an outer-sphere mechanism, CO2 adds to ethylene away from the metal center (see Scheme 1). Work has already been done to determine the underlying reasons for the occurrence of a preferential mechanism (inner vs outer sphere) for a given ligand. Useful insights were achieved by considering a series of bisphosphine ligands containing tert-butyl substituents (dtbpm, dtbpe, dtbpp; see Figure 1),5 with varying lengths of the carbon bridge,



COMPUTATIONAL DETAILS

All the DFT calculations were performed at the GGA level with the Gaussian G0927 set of programs, using the BP86 functional of Becke and Perdew.28,29 The electronic configuration of the molecular systems was described with the split-valence plus one polarization function basis set of Ahlrichs for H, B, C, N, O, and P (SVP30 keyword in Gaussian). For Ni we used the small-core, quasi-relativistic Stuttgart− Dresden effective core potential, with an associated valence basis set31−33 (SDD keywords in G0927). Geometry optimizations were performed without symmetry constraints, and the characterization of the located stationary points was performed by analytical frequency calculations. For better energetics, energies were re-evaluated via single-point calculations on the BP86/SVP geometries with the tripleζ plus one polarization function basis set proposed by Ahlrichs34 (TZVP keyword in Gaussian) using the BP86 functional. Solvent effects were estimated with the polarizable continuum solvation model PCM35,36 using toluene as solvent. To this BP86/TZVP electronic energy in solvent, zero-point energy and thermal corrections were included from the gas-phase frequency calculations at the BP86/SVP level. Since the main scope of this work is not to provide exact values for the absolute energy barriers but to provide a fast protocol for screening several catalysts, we aimed at using a computationally fast GGA functional. The underlying assumption is that the comparison between several models calculated with the same computational approach provides reliable trends due to cancellation of systematic errors of the computational protocol. Our final choice was based on

Figure 1. Ni-bisphosphine ligands considered in this work.

−(CH2)n− with n = 1−3, in the ligand backbone.19−23 In this ligand series, increasing the ligand bite angle leads to greater steric bulk, making the inner-sphere mechanism less favorable over the outer-sphere mechanism. Specifically, the energy barrier along the inner-sphere pathway systematically increases in the order dtbpm < dtbpe < dtbpp, due to increased steric pressure on the transition state TSinner, while the energy barrier along the outer-sphere reaction pathway is much less affected, suggesting that the bite angle can act as a possible descriptor to predict the preferential mechanism for the oxidative coupling reaction.5 However, the bite angle is probably not a good descriptor when ligands with substituents having different steric requirements have to be compared. Further, it is unclear to what extent different substituents on the P atoms can influence the electronic situation at the metal, an effect that can also contribute to modulate reactivity. To better B

DOI: 10.1021/acs.organomet.6b00905 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics test calculations on system L8 which indicated that BP86 electronic energies offer reasonable agreement with the more accurate in vacuum random phase approximation energies discussed in ref 5 (see the Supporting Information for further details). The buried volume and the steric map calculations were performed with the SambVca. 2.0 package.37 The radius of the sphere around the metal center was set to 3.5 Å, while for the atoms we adopted the Bondi radii scaled by 1.17, and a mesh of 0.1 Å was used to scan the sphere for buried voxels.

Moving to the whole ligand set L1−L12, we systematically evaluated the influence of the length of the bridge of the ligand on the free energy activation barrier along the inner and outer sphere pathways, G⧧inner and G⧧outer, respectively. For each ligand subset, increasing the length of the bridge leads to a systematic increase in the calculated G⧧inner barrier, while there is no specific trend observed between the length of the carbon spacer and the calculated G⧧outer barriers (see Table 1). It is worth mentioning here that we reproduce the trend reported in the literature for the ligand series containing tert-butyl substituents, L7−L9, with the calculated G⧧inner barriers following the trend L7 (31.7 kcal/mol) < L8 (36.2 kcal/mol) < L9 (39.7 kcal/mol). Additionally, we observed a similar trend in the calculated G⧧inner barriers for the ligand series L1−L3, L1 (30.5 kcal/mol) < L2 (33.6 kcal/mol) < L3 (34.5 kcal/mol), and L4−L6, L4 (32.5 kcal/mol) < L5 (33.9 kcal/mol) < L6 (36.1 kcal/mol). Of course, in line with previous work each of these trends can be correlated to the ligand bite angle, with larger bite angles increasing the energy barrier for the oxidative coupling reaction along the inner-sphere reaction pathway. Overall, for the considered ligand set L1−L12 the Nibisphosphine complexes containing L1 and L12 present the lowest and highest G⧧inner barriers (30.5 and 42.3 kcal/mol, respectively), while L3 and L4 present the lowest and highest G⧧outer barriers, 27.8 and 34.6 kcal/mol, respectively. From a geometrical point of view, analysis of the transition state geometries (see Figure 3) indicates a clear difference in the calculated Ni−O(CO2) and Ni−C(CO2) distances in the TSinner transition state for ligand L1 (2.244 and 2.205 Å, respectively) and L12 (2.445 and 2.546 Å, respectively), while the length of the emerging C−C bonds is very similar for both ligands (≈1.80 Å). To this end, for the ligand set L1−L12, we investigated if there exists any correlation between the



RESULTS AND DISCUSSION Table 1 summarizes the energetics of the CO2 and ethylene oxidative coupling according to the reaction mechanisms given Table 1. Energetics of the Competitive Oxidative Coupling Reaction Mechanisms Given in Scheme 1 for Ligands L1− L12a

a

ligand

A

G⧧inner

G⧧outer

B

L1 L2 L3 L4 L5 L6 L7 L8 L9 L10 L11 L12

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

30.5 33.6 34.5 32.5 33.9 36.1 31.7 36.2 39.7 39.7 34.0 42.3

29.7 32.2 27.8 34.6 32.9 31.5 31.3 28.5 32.4 31.8 31.9 32.1

6.8 10.6 10.8 8.8 10.4 13.8 8.5 11.5 19.5 12.5 9.2 20.3

Free energies in solution are given in kcal/mol, relative to complex A.

in Scheme 1, as catalyzed by the Ni-based complexes bearing the ligand set L1−L12. Figure 2 presents the computed free

Figure 2. Computed free energy surface for the competitive oxidative coupling reaction mechanisms given in Scheme 1 for an Ni− bisphosphine complex containing the L1 ligand. Free energies in solution are given in kcal/mol (toluene as the solvent) relative to the Ni(L1)(ethylene) coordinated complex A.

energy surface for the investigated competitive reaction mechanisms for L1 as representative of the whole ligand set. The reaction starts from the Ni(L)(ethylene) coordinated complex A. From complex A, we explored both the inner- and outer-sphere reaction pathways for the oxidative coupling of CO2 and ethylene, leading to the formation of the nickelalactone intermediate B. Our results suggest that for ligand L1 the inner-sphere mechanism is competitive with the outersphere mechanism, with the calculated TSinner and TSouter transition states lying 30.5 and 29.7 kcal/mol above complex A, respectively. The nickelalactone B lies 6.8 kcal/mol above complex A for L1.

Figure 3. Geometry of the inner-sphere transition states for ligands L1 (a) and L12 (b) and outer-sphere transition states for ligands L3 (c) and L4 (d). Free energies in solution are given in kcal/mol, relative to complex A. Selected distances are given in Å. C

DOI: 10.1021/acs.organomet.6b00905 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics calculated G⧧inner barriers and the Ni−O(CO2) and Ni− C(CO2) distances. Interestingly, we observed strong correlation (R2 = 0.90) between the G⧧inner barriers and the Ni−C(CO2) distance, while a reasonably good correlation (R2 = 0.81) was found with the Ni−O(CO2) distance (see Figure S1 in the Supporting Information). This direct correlation between the Ni−C(CO2) and Ni−O(CO2) distances and the G⧧inner barriers is consistent with increased steric pressure from the ligand, which prevents optimal approach of CO2 to the Ni center. Of course, this kind of correlation is meaningless in the outersphere mechanism, since CO2 is away from the Ni center. Next, we compared the energy barrier along the inner- and outer-sphere reaction mechanisms by estimating the free energy difference between the inner- and outer-sphere free energy barriers (ΔG⧧in‑out = G⧧inner − G⧧outer; see Figure 4). A positive

Figure 5. Steric maps of the Ni(L)(ethylene) complex for ligands L3, L6, and L9. The complexes are oriented placing the metal center at the origin of the sphere, with the middle point between the two P atoms aligned along the z axis at negative z values and with one P atom in the xz plane, as indicated at the top for system L3. Hydrogens are omitted for clarity.

available to CO2 to approach the metal center within the innersphere reaction pathway. To capture the different steric bulkiness of the different ligands, we calculated the percent buried volume, %VBur, of the L ligand in the Ni(L)(ethylene) complex A for the considered ligand set L1−L12. Next, we investigated the capability of the steric %VBur descriptor to correlate the energy barrier along both the inner- and outer-sphere pathways (G⧧inner and G⧧outer) for the whole ligand set L1−L12. The reasonably good correlation we found between the %VBur and the calculated G⧧inner barriers (R2 = 0.81; see the Supporting Information), and the poor correlation we found between %VBur and the calculated G⧧outer barriers (R2 ≈ 0.10; see the Supporting Information) are consistent with the concept that the steric bulkiness of the ligand has an effect on the energetics of the inner-sphere reaction pathway, while the outer-sphere reaction pathway is scarcely affected. To check for the effect of dispersion interactions on the calculated G⧧inner and G⧧outer barriers, and their correlation with %VBur, we considered the M06 functional for single-point solvent calculations using PCM. Our results presented identical correlations for G⧧inner and G⧧outer barriers versus %VBur, indicating that the dispersion interactions have a negligible impact (for more details see the Supporting Information). Nevertheless, the relatively low correlation between %VBur and the G⧧ inner barrier indicates that steric effects alone are not enough to capture properly the behavior of the different ligands. This is not surprising, considering the different electronic properties of the substituents on the P atoms. Thus, we decided to also include a descriptor capable of capturing electronic effects due to the different ligands. As an electronic descriptor we used the Mulliken charge on the metal center, qNi. Using qNi along with %VBur in a two-parameter fitting of G⧧inner resulted in a strong correlation (R2 = 0.90), as is evident from Figure 6, where the

Figure 4. Calculated ΔG⧧in‑out with the ligand set L1−L12.

value of ΔG⧧in‑out indicates that the outer-sphere mechanism is preferred, while a negative value indicates that the inner-sphere mechanism is operative. It is evident from Figure 4 that for ligands with a small bite angle, such as L1, L4, and L7, the inner-sphere mechanism is either favored or it is competitive with the outer-sphere mechanism, while the outer-sphere mechanism is clearly favored with ligands having a large bite angle, such L2, L3, L5, L6, L8−L10, and L12. Finally, we investigated the underlying reasons for the observed mechanism of preference for the considered ligand set L1−L12, possibly arriving at the use of a common descriptor correlating the ligands presenting different substituent groups on the P atoms. This is a challenging task, due to the fact that the ligands in the considered set are different from both electronic and steric points of view, which can significantly influence the kinetics of the oxidative coupling step under investigation. Despite the fact that the bite angle can explain the trend in the calculated G⧧inner for a ligand subset with the same P substituents, comparison of the ligands between two subsets having the same carbon spacer but different steric bulk makes the bite angle a relatively poor descriptor. For instance, on consideration of the ligands with the three-carbon bridge −(CH2)3−, with very similar bite angles (L3, 104.0°; L6, 101.2°; L9, 104.9°), the calculated G⧧inner energy barriers are significantly different: i.e. L3, 34.5 kcal/mol; L6, 36.1 kcal/mol; L9, 39.7 kcal/mol. The different steric requirements of these three ligands are indeed evident by analysis of the steric maps of the ethylene complexes A, reported in Figure 5, with the bulky tert-butyl substituents of L9 creating a narrow catalytic pocket, while the steric map of L3 shows a much larger space D

DOI: 10.1021/acs.organomet.6b00905 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

indicating that the electronic properties of the substituents of the P atoms have the capability of modulating the aforementioned free energy barrier. In light of the variety in the electronic and steric properties of the examined ligands, we think this work is particularly promising in terms of correlating ligands having (i) different spacers between the P atoms, (ii) substituents on the P atoms with different steric requirements, and (iii) substituents on the P atoms with different electronic properties. Considering that CO2 coupling with olefins is a remarkable topic, for which a winning catalyst has not yet been developed, the QSAR protocol we have developed, on the basis of descriptors that can be essentially applied to any family of metal-based catalysts, could be used to screen different ligands. Of course, before application of the proposed protocol to a different metal and ligand combination, careful testing would be required.

Figure 6. Correlation between the calculated and the predicted G⧧inner barriers for the considered ligand set L1−L12.

calculated and predicted G⧧inner barriers are plotted against each other (for more details on regression analysis, see the Supporting Information). It is worth mentioning here that, for each ligand subset (L1−L3, L4−L6, or L7−L9), the negative charge on Ni metal is slightly more negative, on average by −0.08e, for the systems with the −(CH2)2− and −(CH2)3− bridges in comparison to the single-carbon methylene bridge system. This small variation in the charges results in a small improvement in the correlation of G⧧inner from R2 = 0.81, with the steric descriptor only, to R2 = 0.90 when both the electronic and the steric descriptors are used. Of course, this indicates that the trend in G⧧inner is dominated by steric factors. Further, with such small variation in the charge on the Ni atom, no specific trend was observed between the charge on the Ni atom and the calculated G⧧inner (or G⧧outer) barriers. To summarize, these observations suggest that the ligand sterics (measured by the buried volume, %VBur) and electronics (measured by the Mulliken partial charge on the metal, qNi) together can provide a better description and a better predicting power of the activation energy for the oxidative coupling along the inner-sphere reaction pathway, within a purely QSAR approach.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.organomet.6b00905. Correlation plots and regression analysis of all the species discussed in this work (PDF) Cartesian coordinates and energies of all the species discussed in this work (XYZ)



AUTHOR INFORMATION

Corresponding Author

*E-mail for L.C.: [email protected]. ORCID

Luigi Cavallo: 0000-0002-1398-338X Notes



The authors declare no competing financial interest.



CONCLUSIONS In summary, we have presented a theoretical study describing the competitive reaction mechanisms (inner and outer sphere) for the oxidative coupling of CO2 and ethylene for a set of 12 Ni−bisphosphine ligand compounds presenting cyclohexyl (L1−L3), phenyl (L4−L6), and tert-butyl substituents (L7− L9), while varying the length of the carbon bridge, −(CH2)n− with n = 1−3, in the ligand backbone, as well as other ligands including a ferrocene, an aromatic, and a xanthene bridge (L10−L12). In agreement with the literature we found that increasing steric hindrance disfavors the inner-sphere reaction pathway, while the outer-sphere reaction pathway is scarcely affected. To correlate this behavior to the different structures of the ligands, we evaluated the steric and electronic properties of the ligands. Differently from the literature, we used the buried volume steric descriptor in place of the bite angle commonly used with chelating ligands, since the latter is unable to capture properly variations in steric hindrance when different substituents are on the P atoms. For electronic effects we used the partial charge on the metal center. Using only the steric descriptor results allows for a reasonable correlation with the free energy barrier for the inner sphere reaction pathway, R2 = 0.80, indicating that steric effects are the dominating term composing the free energy barrier for CO2 coupling with olefins. Adding the electronic descriptor in a two-variable fitting procedure results in a better correlation, with R2 = 0.90,

ACKNOWLEDGMENTS M.A.-G and S.V.C.V. thank the SABIC for financial support. This research was supported by the King Abdullah University of Science and Technology.



REFERENCES

(1) Stieber, S. C. E.; Huguet, N.; Kageyama, T.; Jevtovikj, I.; Ariyananda, P.; Gordillo, A.; Schunk, S. A.; Rominger, F.; Hofmann, P.; Limbach, M. Chem. Commun. 2015, 51, 10907. (2) Manzini, S.; Huguet, N.; Trapp, O.; Schaub, T. Eur. J. Org. Chem. 2015, 2015, 7122. (3) Jang, Y. S.; Choi, Y. S.; Byun, H. S. Korean J. Chem. Eng. 2015, 32, 958. (4) Yoon, S. D.; Byun, H. S. Korean J. Chem. Eng. 2014, 31, 522. (5) Plessow, P. N.; Schafer, A.; Limbach, M.; Hofmann, P. Organometallics 2014, 33, 3657. (6) Jing, Y.; Wei, L.; Wang, Y. D.; Yu, Y. M. Microporous Mesoporous Mater. 2014, 183, 124. (7) Khan, A. L.; Klaysom, C.; Gahlaut, A.; Vankelecom, I. F. J. J. Membr. Sci. 2013, 436, 145. (8) Jin, D.; Schmeier, T. J.; Williard, P. G.; Hazari, N.; Bernskoetter, W. H. Organometallics 2013, 32, 2152. (9) Lejkowski, M. L.; Lindner, R.; Kageyama, T.; Bodizs, G. E.; Plessow, P. N.; Muller, I. B.; Schafer, A.; Rominger, F.; Hofmann, P.; Futter, C.; Schunk, S. A.; Limbach, M. Chem. - Eur. J. 2012, 18, 14017. (10) Beuermann, S.; Buback, M.; Schmaltz, C. Ind. Eng. Chem. Res. 1999, 38, 3338.

E

DOI: 10.1021/acs.organomet.6b00905 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics (11) Murata, K.; Matsuda, A. Bull. Chem. Soc. Jpn. 1980, 53, 214. (12) Gerngross, T. U.; Slater, S. C. Sci. Am. 2000, 283, 36. (13) Tardy, D. C. Chem. Phys. Lett. 1972, 17, 431. (14) Hoberg, H.; Peres, Y.; Kruger, C.; Tsay, Y. H. Angew. Chem., Int. Ed. Engl. 1987, 26, 771. (15) Graham, D. C.; Mitchell, C.; Bruce, M. I.; Metha, G. F.; Bowie, J. H.; Buntine, M. A. Organometallics 2007, 26, 6784. (16) Hendriksen, C.; Pidko, E. A.; Yang, G.; Schaffner, B.; Vogt, D. Chem. - Eur. J. 2014, 20, 12037. (17) Wolfe, J. M.; Bernskoetter, W. H. Dalton Trans. 2012, 41, 10763. (18) Yang, G.; Schaffner, B.; Blug, M.; Hensen, E. J. M.; Pidko, E. A. ChemCatChem 2014, 6, 800. (19) Langer, J.; Görls, H.; Westerhausen, M. Inorg. Chem. Commun. 2010, 13, 488. (20) Langer, J.; Walther, D.; Görls, H. J. Organomet. Chem. 2006, 691, 4874. (21) Langer, J.; Görls, H.; Gillies, G.; Walther, D. Z. Anorg. Allg. Chem. 2005, 631, 2719. (22) Langer, J.; Görls, H.; Fischer, R.; Walther, D. Organometallics 2005, 24, 272. (23) Langer, J.; Walther, D.; Malassa, A.; Westerhausen, M.; Görls, H. Eur. J. Inorg. Chem. 2010, 2010, 275. (24) Yang, G.; Schäffner, B.; Blug, M.; Hensen, E. J. M.; Pidko, E. A. ChemCatChem 2014, 6, 800. (25) Poater, A.; Cosenza, B.; Correa, A.; Giudice, S.; Ragone, F.; Scarano, V.; Cavallo, L. Eur. J. Inorg. Chem. 2009, 2009, 1759. (26) Falivene, L.; Credendino, R.; Poater, A.; Petta, A.; Serra, L.; Oliva, R.; Scarano, V.; Cavallo, L. Organometallics 2016, 35, 2286. (27) Gaussian 09; Gaussian, Inc., Wallingford, CT, USA, 2009. (28) Becke, A. D. J. Chem. Phys. 1993, 98, 1372. (29) Perdew, J. P.; Wang, Y. Phys. Rev. B: Condens. Matter Mater. Phys. 1992, 45, 13244. (30) Schafer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571. (31) Andrae, D.; Haussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor. Chim. Acta 1990, 77, 123. (32) Ehlers, A. W.; Frenking, G. J. Chem. Soc., Chem. Commun. 1993, 1709. (33) Hollwarth, A.; Bohme, M.; Dapprich, S.; Ehlers, A. W.; Gobbi, A.; Jonas, V.; Kohler, K. F.; Stegmann, R.; Veldkamp, A.; Frenking, G. Chem. Phys. Lett. 1993, 208, 237. (34) Ahlrichs, R.; Taylor, P. R. J. Chim. Phys. 1981, 78, 315. (35) Cramer, C. J.; Truhlar, D. G. Chem. Rev. 1999, 99, 2161. (36) Hadzi, D.; Orville-Thomas, W. J.; Tomasi, J. J. Mol. Struct. 1994, 120, R7. (37) Falivene, L.; Credendino, R.; Poater, A.; Petta, A.; Serra, L.; Oliva, R.; Scarano, V.; Cavallo, L. Organometallics 2016, 35, 2286.

F

DOI: 10.1021/acs.organomet.6b00905 Organometallics XXXX, XXX, XXX−XXX