Cooperative Metal + Ligand Oxidative Addition and ... - ACS Publications

Sep 19, 2017 - For the present case, massive changes in electronic structure do not incur massive energetic penalties. In conjunction with previous re...
0 downloads 12 Views 2MB Size
Article Cite This: Organometallics XXXX, XXX, XXX−XXX

Cooperative Metal + Ligand Oxidative Addition and σ‑Bond Metathesis: A DFT Study Kent G. Lopez,† Thomas R. Cundari,*,† and J. Brannon Gary*,†,‡ †

Department of Chemistry, University of North Texas Center for Advanced Scientific Computing and Modeling (CASCaM), Denton, Texas 76201, United States ‡ Department of Chemistry & Biochemistry, Stephen F. Austin State University, Nacogdoches, Texas 75962, United States S Supporting Information *

ABSTRACT: A computational study of the experimentally proposed mechanism of alkyne diboration by a PDICo complex yielded two fundamental catalytic steps that undergo remarkable electronic changes (PDI = bis(imino)pyridine). The reactions are envisaged via DFT (density functional theory) and MCSCF (multiconfiguration self-consistent field) simulations as (i) a cooperative metal + ligand oxidative addition and (ii) a σ-bond metathesis induced ligand to metal charge transfer. Analysis of the bonding of pertinent intermediates/TSs also yielded important insight that may be illuminating with regard to the larger field of green catalysis that seeks to ennoble base metals through synergy with potentially redox noninnocent (RNI) ligands. For the present case, massive changes in electronic structure do not incur massive energetic penalties. In conjunction with previous research, one may postulate that structural and energetic “fluidity” among several electronic states of RNI-M3d along the reaction coordinate is an essential signature of redox cooperativity and thus ennoblement.

1. INTRODUCTION There has been considerable interest in the organometallic community in “ennoblement” of base or Earth-abundant metals to mimic catalytic cycles that are well-established among platinum-group metals (PGMs).1 Among Earth-abundant catalysts, the 3d metals have been well studied, but many of the earlier 4d and 5d metals are also relatively plentiful. As is well-known, 3d metals in relation to the 4d and 5d metals show greater proclivity for 1e− vs 2e− transformations such as oxidative addition and reductive elimination. While 1e− transformations, notably radical processes, are routinely harnessed in biological catalysis, for many abiological catalyses they suffer from poor selectivity. A strategy to overcome the redox limitations of Earth-abundant metals3d, 4d or 5d entails the use of redox-noninnocent (RNI) supporting ligands. In a seminal paper, Heyduk et al. reported oxidative addition of Cl2 to a ZrIV complex.2 Through a variety of experimental methods this group showed that the two electrons needed to cleave the Cl−Cl bond and form two new Zr−Cl bonds come from the supporting ligand. A study by the same group described C−C reductive elimination to form biaryls from a formally ZrIVAr2 complex.3 Important theoretical treatments by Baik and co-workers4 corroborate the proposals of Heyduk et al. as to the nature of the two electrons originating from ligandbased orbitals. Calculations by the Baik group also reveal finer details vis-à-vis the nature of what are ostensibly intramolecular ligand-to-ligand electron transfers. The Heyduk group extended their work to nitrene transfer by formally d0 complexesZrIV and TaVto substrates such as isonitriles to synthesize carbodiimides.5 © XXXX American Chemical Society

These experimental and computational precedents led us to inquire the following. (1) Are cooperative metal + ligand 2e− transformations possible? (2) What are the electronic requirements for metal + ligand redox cooperativity? For the specific example of Earth-abundant 3d metal catalysts, one goal is to identify cases in which the central metal ion can source a single electron to (or drain 1e− from) a catalytically active site in synergy with a RNI ligand that will do likewise. Recent experimental work by Chirik and co-workers has convincingly answered question 1 in the affirmative.6 Through an impressive array of experimental techniques, supported by DFT, Darmon et al. showed that oxidative addition of a strained C−C bond of biphenylene is effected by bis(imino)pyridine FeII dinitrogen complexes to yield an FeIII biaryl complex. The two electrons needed to effect C−C activation are formally derived thusly: 1e− from the supporting ligand and 1e− from iron. More recently, the same group has disclosed interesting examples of catalytic boration (with Bpin-H)7 and diboration (with B2pin2)8 of alkynes by a cobalt complex with PDI supporting ligation (pin = pinacolato, PDI = bis(imino)pyridine). In the putative catalytic cycles, these researchers proposed oxidative addition of a B−H7 (or B−B8) bond to the cobalt center. These reactions seemed to be more strong candidates for a cooperative metal + ligand oxidative addition within a complete catalytic cycle. To this end, DFT calculationsaugmented by MCSCF (multiconfiguration selfconsistent field) analyseswere conducted of the diboration Received: September 19, 2017

A

DOI: 10.1021/acs.organomet.7b00715 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics catalytic intermediates and transition states to better understand the electronic and free energy transformations along the proposed catalytic cycle. The computations revealed (1) cooperative metal + ligand oxidative addition of the B−B bond of B2pin′2 and, perhaps even more surprisingly, (2) a competitive pathway in which σ-bond metathesis of a B−B bond across a Co−C bond induced an intramolecular LMCT (ligand to metal charge transfer) event.

Table 1. uB3LYP/6-31G(d,p) Calculated Relative Enthalpies and Free Energies (kcal/mol) of Low-Energy Spin States of PDICo(CCMe) (1)a

2. COMPUTATIONAL METHODS The Gaussian09 code9 was used for all simulations reported herein. The uB3LYP/6-31G(d,p) level of theory was employed. All pertinent spin statesincluding singletswere evaluated with unrestricted Kohn−Sham techniques. The SMD continuum solvation model was employed using THF as the solvent.10 Enthalpy and entropy corrections to the electronic energy assumed 1 atm and 298.15 K and were made with unscaled vibrational frequencies. All stationary points were characterized as minima or transition states via calculation of the correct number of imaginary frequencies, while transition states were confirmed via the calculation of the intrinsic reaction coordinates. MCSCF11 (multiconfiguration selfconsistent field) calculations were of the CASSCF (complete active space SCF) variety and employed a 12-orbital, 12electron active space using a variety of restricted and restrictedopen-shell Hartree−Fock initial guess wave functions.

a

multiplicity css

1 oss 1 3 5

ΔHrel

ΔGrel

DFT electronic description

+6.9 −1.0 0.0 +5.5

+8.3 +1.1 0.0 +3.3

(PDI)(LS-d8-CoI)(CCR−) (PDI•−)↓(LS-d7-CoII)↑(CCR−) (PDI•−)↑(LS-d7-CoII)↑(CCR−) (PDI•−)↑(HS-d7-CoII)↑↑↑(CCR−)

Abbreviations: oss, open-shell singlet; css, closed-shell singlet.

spin states) and 27° (the Npy−Co−Cacetylide bond angle is ∼178° for the triplet and quintet states of 1 but 151° for css1 and 170° for oss1). Test calculations on PDICo(CCMe) models with larger (iPr, Cy on imine nitrogens) alkyls indicate that Npy−Co−Cacetylide is affected significantly by the steric bulk of the imine C substituent, which must be kept in mind given the small models used in the computation of the reaction coordinate and MCSCF simulations. As noted above, the present DFT simulations slightly favor a triplet ground state for PDICo(CCR) (1) versus an openshell singlet. Indeed, the latter is predicted to be more stable via computation of the enthalpy. This is not surprising, given that in a thorough study by Knijnenburg et al. utilizing B3LYP methods the S-T splitting was found to be sensitive to ligand substituents and their effect on the crystal field splitting.12 Hartree−Fock results also propose a triplet ground state, albeit by a wider margin versus the singlet, ∼1 eV; the M06 functional predicts an S-T splitting similar to that of B3LYP, 9 kcal/mol. The best Lewis structure description suggested by analysis of the Mulliken-derived spin densities for the ground state of acetylide catalyst 1 is 3[(PDI•−)↑(LS-d7-CoII)↑(CCR−)]; the open-shell singlet is similar except that the two spins are antiferromagnetically coupled. The DFT description is corroborated by MCSCF calculations (CAS(12,12) active space) at the DFT-optimized geometry of 31. There is one metal-based (dz2) and one PDI Lπ (Figure 2) for two natural orbitals with occupation numbers of ∼1.0e−. As such, both DFT and MCSCF indicate redox noninnocence of the PDI supporting ligand in PDICo(CCMe), which echoes the conclusions for PDICoMe complexes by Bowman et al.13

3. RESULTS AND DISCUSSION 3.1. Electronic Structure of the PDICo(CCMe) Catalyst. A PDI model, where methyl substituents were utilized for both the imine carbons and imine nitrogens, was employed for computations described herein. Initial DFT geometry optimizations focused on the ground state, which was calculated to be a triplet (Figure 1) with an open-shell singlet

Figure 1. uB3LYP/6-31G(d,p)-optimized geometry of 3PDICo(C CMe). Bond lengths are given in Å and angles in deg. A superscript prefix numeral indicates the multiplicity. Color code: Co, salmon; N, magenta; C, yellow; H, blue.

(oss1), closed-shell singlet (css1), and quintet states close in energy (Table 1) and in geometry (see the Supporting Information for the geometries of higher energy spin states). Average absolute differences in bond lengths and bond angles among the spin states of 1 were small, ±0.02 Å and ±1° for bond lengths and bond angles, respectively, with maximum differences of ∼0.2 Å (longer Co−N bond lengths for higher

Figure 2. CAS(12,12)/6-31G(d) natural orbitals of 3PDICo(C CMe) with ∼1.0e− occupancy (isovalue 0.03). B

DOI: 10.1021/acs.organomet.7b00715 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

Figure 3. uB3LYP/6-31G(d,p)-calculated free energy (kcal/mol) profile for the catalytic diboration of a cobalt acetylide complex.

oxidative addition pathway via singlet intermediate 2 (Figure 4). The oxidative addition TS, (1-2)⧧, is predicted to be a

The singlet of 1, PDICo(CCMe), was indicated by initial uB3LYP/6-31G(d,p) simulations to be a closed-shell singlet, css 1. Tests of alternative initial guess wave function and wave function stability at the optimized minimum css1, however, found a solution lower in energy by ∼6 1/2 kcal/mol (ΔE) and thus closer to the 31 state. The same tests indicate the 31 wave function is stable. Geometry optimization with the stable wave function solution yielded oss1, with Npy−Co−Cacetylide = 170°, which is marginally lower than 31 in terms of electronic energy and enthalpy but slightly higher in free energy (Table 1). Most critically, both oss1 and 31 are indicated to be best described as (PDI•−)(LS-d7-CoII)(CCR−) by DFT, differing in spin coupling. The singlet state of complex 1 is revealed to be highly multiconfigurational via CAS(12,12)/6-31G(d) simulations, with two natural orbitals far removed from their ideal, single-configuration occupation numbers of 2e− and 0e−. The two most highly correlated natural orbitals have occupation numbers1.38e− and 0.62 e− (these are in- and out-of-phase combinations of PDI π and Co dπ orbitals)far removed from integer values. Tests with larger basis sets did not substantially change the orbital character and natural orbital occupation numbers. Complex 11 is thus revealed by the CAS simulations to have significant singlet biradical character, consistent with the analysis of Knijnenburg et al. for a larger set of PDICo-R complexes.12 3.2. Cooperative Oxidative Addition of pin′B-Bpin′ to PDICo(CCMe). The uB3LYP/6-31G(d,p)-calculated free energy profile for the mechanism proposed by Chirik and coworkers7,8 is shown in Figure 3. Test calculations with the more experimentally relevant cyclohexyl substituents on the minima in Figure 3 indicated, quite reasonably, that the latter, bulkier substituents uniformly push these stationary points higher in free energy by ∼6 kcal/mol on average relative to the starting material. Interestingly, two reaction sequences are computed to be competitive to go from the acetylide PDICo(CCMe) (1) to the singlet boryl-alkyne complex 1PDICo(Bpin′)(Bpin′-C CMe) (3). Note that here and throughout this study that a smaller model, B2pin′2, of the full B2pin2 was used in which the methyls were replaced by hydrogens. The first is a two-step,

Figure 4. uB3LYP/6-31G(d,p)-optimized singlet 2, the intermediate from oxidative addition of B2pin′2 to PDICo(CCMe) (1). Color code: Co, salmon; N, magenta; C, yellow; H, blue; O, red; B, pink. Bond lengths are given in Å and angles in deg.

singlet and is depicted in Figure 5. The free energy profile is flat in the vicinity of intermediate 2, including the transition states that follow and precede it (Figure 3). The competing pathway is a single-step [2 + 2] pathway via a triplet, σ-bond metathesis TS and is discussed in section 3.3. As noted above, oxidative addition of the B−B bond of B2pin′2 to 3PDICo(CCMe) yields the closed-shell (stable via stability testing of the wave function) singlet intermediate 2 (Figure 4). This stationary point has an unusual geometry for a low-spin, d6 six-coordinate complex, being far removed from the expected octahedral geometry with close B···B and B··· Cacetylide contacts of 2.12 and 1.89 Å, respectively. Distortion from octahedral coordination is ascribed to the Z-type nature of the Bpin′ ligands rather than unusual redox behavior by cobalt and/or PDI. To wit, replacing the Bpin′ ligands with typical Xtype ligands such as methyl and H gives the expected octahedral coordination upon geometry optimization. In addition, MCSCF calculations on diboryl complex 2 look C

DOI: 10.1021/acs.organomet.7b00715 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

metathesis mechanism for activation of H2 across the Co−Et bond.14 Hence, it is not surprising that only a few kilocalories per mole higher, ΔΔG⧧ ≈ 3 kcal/mol (Figure 3), than the metal + ligand cooperative oxidative addition TS just discussed is perhaps an even more remarkable transition state, (1-3)⧧. This transition state is predicted to be a triplet; the product of this σ-bond metathesis, the boryl-alkyne complex 3, is a closedshell singlet (stable via wave function testing). As noted above, 1 is best depicted as [(PDI•−)(LS-d7-CoII)(CCR−)]. Complex 3 is best described as a low-spin, square-pyramidal d8 complex: i.e., (PDI0)(LS-d8-CoI)(Bpin′)(η2-C,C-pin′B-C CR−). Thus, although σ-bond metathesis is an overall redoxneutral process, the B−B bond scission event has induced ligand to metal charge transfer. Subsequent steps to close the catalytic cycle involve transfer of a boryl ligand to the coordinated boryl-alkyne to give a 1,1diboryl-vinyl complex (34) (Figure 3). Complex 4 then undergoes a C−H bond activation via a four-centered TS arising from addition of a terminal alkyne C−H bond to the Co−Cvinyl bond of 4 to regenerate the catalyst, 1. Both steps are facile with calculated free energy barriers of 3 and 28 kcal/mol, respectiely (relative to the preceding minima). As such, the latter is commensurate with the barriers involved in B−B bond activation and is likely within the uncertainty of the DFT models employed.

Figure 5. uB3LYP/6-31G(d,p)-optimized singlet transition state, (12)⧧, for oxidative addition of B2pin′2 to 3PDICo(CCMe) (1). Color code: Co, salmon; N, magenta; C, yellow; H, blue; O, red; B, pink. Bond lengths in Å, angles in 0.

similar to those for the corresponding methyl hydride. Hence, 1 2 is viewed as a low-spin CoIII complex with PDI0(LS-d6CoIII)(Bpin′−)2((CCR−) formulation. As such, the overall B−B oxidative addition process may be viewed as [(PDI•−)(LS‐d7‐CoII)(CCR−)] + B2pin′2

oss,3

→ 1[PDI0(LS‐d6‐CoIII)(Bpin′− )2 (CCR−)]

4. SUMMARY, CONCLUSIONS, AND PROSPECTUS A computational study of the mechanism of alkyne diboration by a PDICo complex7,8 yielded two catalytic steps that engender remarkable electronic changes. Analysis of the bonding of pertinent intermediates/TSs yielded important insight that is illuminating with regard to catalysis research that seeks to ennoble Earth-abundant metals through synergy with potentially redox noninnocent ligands. The reactions are envisaged via DFT and MCSCF simulations as (i) a cooperative metal + ligand oxidative addition and (ii) a σbond metathesis induced ligand to metal charge transfer. In section 1, the following question was posed. What are the electronic requirements for metal + ligand redox cooperativity? It seems from the present work, in addition to the studies of Knijnenburg et al.12 as well as those of Ortuño and Cramer,15 that electron transfer between a 3d metal ion and a PDI ligand is quite facile during typical chemical reactions that comprise a catalytic cycle such as oxidative addition/reductive elimination or [2 + 2] addition. The work of the Heyduk, Baik, and Soper groups2−5,16 suggests that facile electronic reorganization is not limited to the specific example of Co ions or PDI. For the present case, massive changes in electronic structure do not incur massive energetic penalties. From the present work, in conjunction with previous research,12,17,18 one may postulate that structural and energetic “fluidity” among multiple electronic states of the RNI-M3d complex along the reaction coordinate is an essential signature of redox cooperativity and base-metal ennoblement. While energy matching of the metal-based 3d orbitals with (typically) Lπ and Lπ* orbitals has been appreciated in the literature of redox-noninnocent complexes,19,20 what is perhaps more critical is extensive orbital mixing among the relevant frontier molecular orbitals. Similar structural and energetic fluidity was seen in tetradentate diamide-diimine complexes of Ti and Cr.21,22 For those complexes, as here, considerable energetic and orbital mixing among the Lπ/π* and metal 3d orbitals along the reaction coordinates was implicated as being

Of the two electrons required at the catalytic active site to cleave the B−B bond and form two new M−B bonds, one electron comes from the CoII of 1 and one electron comes from the PDI radical anioncooperative metal + ligand oxidative addition. This result pertains whether the ground state is 31 or oss 1, as both are deemed to be [(PDI•−)(LS-d7-CoII)(C CR−)]. The oxidative addition transition state connecting 1 and 1 2 is an open-shell singlet via DFT. Thus, the calculations suggest that electron transfer to the catalytically active site from CoII and the RNI supporting ligand has not yet occurred before the transition state. A CAS(12,12) calculation of the oxidative addition TS reveals extensive mixing of the Co dπ and PDI π orbitals, yielding two natural orbitals far removed from integer occupation: (PDI π + Co dπ)1.59(PDI π − Co dπ)0.42 (Figure 6). The natural orbitals in Figure 6 highlight the extensive orbital mixing needed for 3d metal complexes to leverage redox noninnocence along a catalytic reaction coordinate. 3.3. σ-Bond Metathesis Induced Ligand to Metal Charge Transfer. In their study of olefin hydrogenation by a PDI-Co-Et complex, Knijnenburg et al. proposed a σ-bond

Figure 6. CAS(12,12)/6-31G(d) natural orbitals of optimized singlet transition state, (1-2)⧧, for oxidative the B−B bond of B2pin2 to 3PDICo(CCMe) to CoIII(Bpin′)2(CCMe). Occupancies are 1.59e− (left) (right), respectively (isovalue 0.03).

the DFTaddition of give 1PDIand 0.42e− D

DOI: 10.1021/acs.organomet.7b00715 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

(15) Ortuño, M. A.; Cramer, C. J. J. Phys. Chem. A 2017, 121, 5932− 5939. (16) Smith, A. L.; Hardcastle, K. I.; Soper, J. D. J. Am. Chem. Soc. 2010, 132, 14358−14360. (17) Williams, V. A.; Wolczanski, P. T.; Sutter, J.; Meyer, K.; Lobkovsky, E. B.; Cundari, T. R. Inorg. Chem. 2014, 53, 4459−4474. (18) Morris, W. D.; Wolczanski, P. T.; Sutter, J.; Meyer, K.; Cundari, T. R.; Lobkovsky, E. B. Inorg. Chem. 2014, 53, 7467−7484. (19) Frazier, B. A.; Wolczanski, P. T.; Lobkovsky, E. B.; Cundari, T. R. J. Am. Chem. Soc. 2009, 131, 3428−3429. (20) Frazier, B. A.; Williams, V. A.; Wolczanski, P. T.; Bart, S. C.; Meyer, K.; Cundari, T. R.; Lobkovsky, E. B. Inorg. Chem. 2013, 52, 3295−3312. (21) Heins, S. P.; Morris, W. D.; Wolczanski, P. T.; Lobkovsky, E. B.; Cundari, T. R. Angew. Chem., Int. Ed. 2015, 54, 14407−14411. (22) Heins, S. P.; Wolczanski, P. T.; Cundari, T. R.; MacMillan, S. N. Chem. Sci. 2017, 8, 3410−3418. (23) Olatunji-Ojo, O. A.; Cundari, T. R. Inorg. Chem. 2013, 52, 8106−8113. (24) Jacobs, B. P.; Wolczanski, P. T.; Jiang, Q.; Cundari, T. R.; MacMillan, S. N. J. Am. Chem. Soc. 2017, 139, 12145−12148. (25) Wiese, S.; McAfee, J. L.; Pahls, D. R.; McMullin, C. L.; Cundari, T. R.; Warren, T. H. J. Am. Chem. Soc. 2012, 134, 10114−10121.

essential to RNI behavior. Furthermore, previous research on group transfer by multiply bonded actor ligands (E) with potentially RNI supporting ligand complexes suggests that this proposal may be extended to RNI-M3dE complexes, an important class of intermediates/reagents for group transfer catalysis.21−25



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.organomet.7b00715. Cartesian coordinates of all calculated species (XYZ)



AUTHOR INFORMATION

Corresponding Authors

*E-mail for T.R.C.: [email protected]. *E-mail for J.B.G.: [email protected]. ORCID

Thomas R. Cundari: 0000-0003-1822-6473 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS K.G.L. acknowledges the NSF and UNT Chemistry for support through NSF-REU grant CHE-1461027. T.R.C. thanks the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences, and Biosciences for support of this research via grant DE-FG02-03ER15387 and the NSF for their support of computing facilities at UNT Chemistry via MRI grant CHE-1531468.



REFERENCES

(1) Chirik, P. J.; Wieghardt, K. Science 2010, 327, 794−795. (2) Blackmore, K. J.; Ziller, J. W.; Heyduk, A. F. Inorg. Chem. 2005, 44, 5559−5561. (3) Haneline, M. R.; Heyduk, A. F. J. Am. Chem. Soc. 2006, 128, 8410−8411. (4) Ketterer, N. A.; Fan, H.; Blackmore, K. J.; Yang, X.; Ziller, J. W.; Baik, M. H.; Heyduk, A. F. J. Am. Chem. Soc. 2008, 130, 4364−4374. (5) Heyduk, A. F.; Zarkesh, R. A.; Nguyen, A. I. Inorg. Chem. 2011, 50, 9849−9863. (6) Darmon, J. M.; Stieber, S. C. E.; Sylvester, K. T.; Fernández, I.; Lobkovsky, E.; Semproni, S. P.; Bill, E.; Wieghardt, K.; DeBeer, S.; Chirik, P. J. J. Am. Chem. Soc. 2012, 134, 17125−17137. (7) Obligacion, J. V.; Neely, J. M.; Yazdani, A. N.; Pappas, I.; Chirik, P. J. J. Am. Chem. Soc. 2015, 137, 5855−5858. (8) Krautwald, S.; Bezdek, M. J.; Chirik, P. J. J. Am. Chem. Soc. 2017, 139, 3868−3875. (9) Frisch, M. J., et al. Gaussian 09, Revision D.01; Gaussian Inc., Wallingford, CT, USA 2009. (10) Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B 2009, 113, 6378−6396. (11) Schmidt, M. W.; Gordon, M. S. Annu. Rev. Phys. Chem. 1998, 49, 233−266. (12) Knijnenburg, Q.; Hetterscheid, D.; Kooistra, T. M.; Budzelaar, P. H. Eur. J. Inorg. Chem. 2004, 2004, 1204−1211. (13) Bowman, A. C.; Milsmann, C.; Bill, E.; Lobkovsky, E.; Weyhermuller, T.; Wieghardt, K.; Chirik, P. J. Inorg. Chem. 2010, 49, 6110−6123. (14) Knijnenburg, Q.; Horton, A. D.; van der Heijden, H.; Kooistra, T. M.; Hetterscheid, D. G. H.; Smits, J. M. M.; de Bruin, B.; Budzelaar, P. H.M.; Gal, A. W. J. Mol. Catal. A: Chem. 2005, 232, 151−159. E

DOI: 10.1021/acs.organomet.7b00715 Organometallics XXXX, XXX, XXX−XXX