Dispersion Makes the Difference: Bisligated ... - ACS Publications

Dec 11, 2014 - Dispersion Makes the Difference: Bisligated Transition States Found for the Oxidative Addition of Pd(PtBu3)2 to Ar-OSO2R and Dispersion...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/Organometallics

Dispersion Makes the Difference: Bisligated Transition States Found for the Oxidative Addition of Pd(PtBu3)2 to Ar-OSO2R and DispersionControlled Chemoselectivity in Reactions with Pd[P(iPr)(tBu2)]2 Eirik Lyngvi,†,‡,§ Italo A. Sanhueza,†,‡,§ and Franziska Schoenebeck*,† †

Institute of Organic Chemistry, RWTH Aachen University, Landoltweg 1, 52056 Aachen, Germany Laboratory for Organic Chemistry, ETH Zürich, Vladimir-Prelog-Weg 3, 8093 Zürich, Switzerland

S Supporting Information *

ABSTRACT: The manipulation of the steric nature of ligands is a key design principle in organometallic reactivity. While general intuition assumes steric effects to be repulsive, recent reports counterintuitively suggested that highly crowded hydrocarbon molecules may be stabilized more strongly than their less bulky analogues as a consequence of dispersion interactions. With the objective of investigating the significance of such attractive intramolecular dispersion forces in organometallic catalysis, we herein studied the effect of dispersion on the accessible geometries and reactivities for two trialkylphosphine ligands of different sizes in Pd-catalyzed cross-coupling reactions: i.e., L = PtBu3 and its smaller analogue L = P(iPr)(tBu2). Those methods that account well for dispersion (e.g., ωB97XD, B3LYP-D3) allowed the first location of bisphosphine-ligated transition states for the oxidative addition of Pd0L2 to aromatic C−O bonds, involving the bulky and widely employed ligand L = PtBu3. DFT methods without dispersion gave rise to dissociation of one phosphine ligand in all cases examined. To probe whether dispersion may even be a reactivity-controlling factor, we also examined the favored site selectivity of the reaction of Pd0L2 with 4-chlorophenyl triflate, for which the selectivity has previously been shown to be dependent on the ligation state of the reactive palladium species. Various DFT methods (PBE, B3LYP, M06L) and basis sets and different solvent models (COSMO-RS, CPCM) were assessed. While for Pd(PtBu3)2 dispersion-free and dispersion-containing methods predicted the monophosphine pathway via PdL and reaction at C−Cl to be favored, striking differences were observed for Pd[P(iPr)(tBu2)]2. Dispersion-free DFT predicted C−OTf addition by Pd[P(iPr)(tBu2)]2 to be disfavored by ΔΔG⧧ ≈ 20 kcal/mol, despite being experimentally accessible. In stark contrast, the involvement of dispersion adequately described the selectivity. The attractive dispersion forces of the crowded trialkyl substituents are therefore a key controlling factor in the competition between mono- and bisligated pathways.

INTRODUCTION Desired reactivities and selectivities of transition-metalcatalyzed reactions are generally achieved through the manipulation of the steric and electronic properties of the employed phosphine ligands.1 In this context, the ideal ligand is frequently identified as a result of elaborate screening approaches. However, owing to enormous developments in computing power, methods, and software, the employment of computational tools in the prediction and design of ligands is gaining increasing significance.2,3 In particular, the implementation of dispersion in DFT has been a tremendous methodological advance.4 As a result, several experimental reactivity phenomena could at last be explained with computational means.5 These include the findings by Schreiner, Fokin, and co-workers that only dispersion-corrected DFT methods could explain their experiments, counterintuitively indicating that bulky tBu3 groups increased the overall stability of hydrocarbons via attractive tBu-intramolecular H- - -H dispersive interactions.6−8 In an organometallic context, Ahlquist © XXXX American Chemical Society

and Norrby found dispersion to be crucial in describing [Pd(PPh3)n] complexes,9 and Harvey and Fey suggested that dispersion-corrected DFT would perform better in the quantitative reproduction of kinetic data.10 However, several recent reports also indicated potential overestimations of dispersion.11 As part of our interests in studying and designing organometallic reactivities,12 we recently began to explore the role and accuracy of dispersion. We hypothesized that the computational evaluation of trialkylphosphine ligands may be affected by dispersion, i.e. accessible geometries and favored pathways might deviate as a consequence of the bulky hydrocarbon ligands potentially becoming “attractive” with dispersion rather than repulsive, as would otherwise be expected on the basis of steric interaction. Received: November 27, 2014


dx.doi.org/10.1021/om501199t | Organometallics XXXX, XXX, XXX−XXX



ligation state of the reactive Pd species (which in turn is influenced by the nature of the ligand).16 We initially examined whether the ligation-state-dependent selectivity would also hold if dispersion-corrected methods were utilized, as the previous examinations had not involved dispersion. Thus, we calculated the transition states for oxidative addition to C−Cl and C−OTf by the model complexes Pd(PMe3)n (with n = 1, 2), employing CPCM (THF) PBE0-D3/6-311+G(d)//ωB97XD/6-31G(d) (with LANL2DZ for Pd). In accord with dispersion-free DFT methods, preferential addition of the monoligated PdPMe3 to C−Cl (ΔΔG⧧ = 2.5 kcal/mol) and bisligated Pd(PMe3)2 to C−OTf (ΔΔG⧧ = 9.2 kcal/mol) was calculated, confirming the site selectivity probe to be valid (see also Figure S1 in the Supporting Information). To search for a bisligated transition state for the full ligand PtBu3, DFT calculations were performed with Gaussian 09,17 applying two different methods that account well for dispersion: i.e., ωB97XD and B3LYP-D3. Geometry optimizations were done with the basis sets 6-31G(d) and LANL2DZ for Pd or def2-SVP. For both methods we were able to locate transition states for the oxidative addition to the C−OTf bond that resembled two ligands at the reactive palladium center.18 This was successfully accomplished for both basis sets, 6-31G(d)/LANL2DZ and def2-SVP. A representative geometry is illustrated in Figure 1. The crucial geometric bond distances of the located TSs are also given.

One of the most widely employed and very bulky trialkylphosphine ligands is PtBu3.13 It exhibits a relatively large cone angle of 182°,14 and this ligand would hence constitute a stern test for the evaluation of attractive dispersion forces. All previous attempts15 to locate bisligated transition states (TSs) for the oxidative addition by [Pd0(PtBu3)2] with dispersion-free DFT methods have so far been unsuccessful: one phosphine ligand always dissociated and oxidative addition occurred instead via the monoligated palladium species [Pd0(PtBu3)].15 This was ascribed to steric repulsion of the bulky phosphine ligands. We herein investigated the influence of dispersion on the oxidative addition of the bisligated palladium species Pd0(PtBu3)2 and its smaller analogue Pd0[P(iPr)(tBu2)]2 to 4-chlorophenyl triflate (1). We chose 1 as a test system, as the site selectivity is dependent on the ligation state in this case.12a,13c,16 It is therefore an ideal probe to correlate the computational findings versus experimental data (Scheme 1). Scheme 1. Objective of the Study and Test System To Investigate the Role of Dispersion in the Ligation State Controlled Site Selectivity

Specifically, our objective was to test whether dispersion might allow us to locate a bisligated TS and, if so, to assess whether dispersion could become a selectivity-controlling factor in favoring the bisligated over the monoligated pathway due to attractive dispersion forces between the trialkyl substituents of the phosphine ligands.

RESULTS AND DISCUSSION Can Bisligated Oxidative Addition Transition States Be Located with Dispersion for the Very Bulky L = PtBu3? Previous experimental studies established that subjecting Pd2(dba)3/PtBu3 (1/1 Pd/P ratio) to 4-chlorophenyl triflate (1) under typical Suzuki cross-coupling conditions led to exclusive functionalization at the C−Cl site in THF at room temperature.12a,13c Calculations with “standard” DFT methods previously showed that this selectivity is consistent with the monoligated palladium complex Pd0PtBu3 as the reactive species.12a,16 The origin of selectivity was found to be predominantly due to the lower distortion energies associated with reaction at the C−Cl site. In contrast, bisligated Pd species added preferentially to the C−OTf site as a result of greater interaction.16 Importantly, previous analyses showed that the selectivity was not directly dependent on the nature of the ligand but instead on the

Figure 1. Crucial geometric features of bisligated TSs.

The transition states all featured a concerted C−O bond cleavage and C−Pd bond formation via a three-membered arrangement, in analogy to transition states previously calculated for oxidative additions in the absence of dispersion and with smaller ligands.12a,16 Thus, dispersion has no significant impact on the TS arrangement with respect to cleaving and forming bonds, although there are deviations in the precise extent of bond cleavage/formation. The ωB97XDoptimized TSs exhibit Pd−O and C−O distances shorter than those obtained from B3LYP-D3 optimization (Figure 1). Moreover, the geometries obtained from transition state optimizations with B3LYP-D3 show a slightly larger basis set dependence for the bonds highlighted in red in Figure 1 in comparison to those obtained with ωB97XD. Moreover, shorter Pd−P distances resulted with B3LYP-D3.19 B

dx.doi.org/10.1021/om501199t | Organometallics XXXX, XXX, XXX−XXX



The transition states optimized with ωB97XD featured relatively long palladium phosphine distances (∼3 Å) (see Figure 1, entries 1 and 2). We therefore set out to examine the extent of bonding interactions between the phosphine ligands and the palladium center. For two atoms to be considered bound to each other, the “quantum theory of atoms in molecules” (QTAIM) requires a unique bond path (bp) as well as a bond critical point (bcp).20 This approach has frequently been applied to study the extent and nature of bonding in organometallic complexes.21 We performed a QTAIM analysis for all transition states using Multiwfn.22 Bond critical points and bond paths between palladium and the two phosphorus atoms (assigned as PA and PB in Figure 1) were obtained for all transition states, suggesting that both phosphine ligands are interacting with the palladium center in the located transition states.23 Figure 2 shows the calculated contour map of the

Figure 3. Bisligated TSs for the oxidative addition of Pd(PtBu3)2 to Ph-OTf (top) and Ph-OTs (bottom), calculated with B3LYP-D3 (top) or ωB97XD (bottom) with 6-31G(d)/LANL2DZ (for Pd). Figure 2. QTAIM analysis of TS1bis‑OTf, located with dispersion corrected DFT (entry 1, Figure 1).

for the reaction with 1. The competing, lowest energy reaction pathways constitute the oxidative addition of bisligated Pd(PtBu3)2 to C−OTf versus the addition of monoligated PdPtBu3 to C−Cl.12a,16 For the monoligated pathway, dissociation of one phosphine ligand from the Pd0(PtBu3)2 resting state prior to oxidative addition is required. This can potentially occur by two different mechanisms: i.e., a dissociative or an associative pathway (see Scheme 2).10 The dissociative mechanism involves loss of one ligand (PtBu3) to generate the monoligated Pd0 complex (PdPtBu3) that subsequently undergoes oxidative addition via TSmono, while the associative pathway loses the ligand through displacement by an arene to give intermediate Int1 in a single step (Scheme 2). Our results show that calculations with dispersion (PBE0D3) favor the associative pathway, forming Int1 via TSass (Figure 4, top). In contrast, calculations without dispersion (PBE0) highly disfavor the associative pathway, and Int1 is preferentially formed by a dissociative mechanism (Figure 4, bottom).10 Importantly, no matter whether dispersion is included or not, the oxidative addition represents the highest energy (and therefore selectivity-determining) point for test system 1. Investigation of the Predicted Site Selectivity for the Oxidative Addition of PdL2 to 1 with L = PtBu3. Having established the favored reaction pathways, we subsequently evaluated the predicted selectivity (ΔΔG⧧) between C−Cl (TSmono‑Cl) versus C−OTf (TSbis‑OTf) addition, using a variety of methods and basis sets. Figure 5 illustrates the effect of dispersion (blue) versus no dispersion (green) for selected DFT methods on the predicted ΔΔG⧧ value for the 6-

electron density for TS1bis‑OTf (for entry 1, Figure 1) with located bcps (blue dots) and bps (brown lines). Contour maps for all other TSs (entries 2−4) are given in the Supporting Information (Figure S3). The topological properties for the crucial bonds were also calculated (see Table S1 in the Supporting Information). These data suggest that the bonding interactions are intermediate in character (ionic with some covalent character), which is typical for metal−ligand bonds.24−26 These data strongly support the bisligated nature of the located transition states. We also reoptimized all bisligated TSs with “dispersion-free” DFT methods, applying B3LYP and PBE0 with the split basis set 6-31G(d)/LANL2DZ (for Pd). In all cases dissociation of one phosphine ligand was observed.15d To investigate the generality in substitution pattern at the aromatic ring, we also searched for TSs derived from addition of Pd0(PtBu3)2 to aromatic C−(pseudo)halogen bonds. We were unable to locate bisphosphine transition states for aryl halides but successfully located TSs for addition to PhOTf (i.e., without the Cl in the 4-position as in 1) and also for an aryl tosylate (Ph-OTs; see Figure 3). Again, these TSs could only be located when methods with dispersion were employed, highlighting the importance of attractive dispersion forces to stabilize these geometries. Impact of Dispersion on the Favored Reaction Pathway for the Oxidative Addition of PdL2 to 1 with L = PtBu3. We next studied whether dispersion may also impact the favored ligation state and thus overall site selectivity C

dx.doi.org/10.1021/om501199t | Organometallics XXXX, XXX, XXX−XXX



Scheme 2. Proposed Pathways for Oxidative Addition

Figure 5. Predicted selectivity (ΔΔG⧧) for C−OTf versus C−Cl addition for L = PtBu3 for CPCM (THF) DFT/6-311+G(d) with LANL2DZ (for Pd), with (blue) and without (green) dispersion, in kcal/mol.

on the free energy differences, giving a drop of up to 30 kcal/ mol in predicted ΔΔG⧧. Considering the remarkably small energy difference between bis- and monoligated pathways at M06L (ΔΔG⧧ = 3 kcal/mol), it seems reasonable to assume that it might even be possible to force the reaction pathway into a bisligated one, if the conditions are tuned accordingly.41 However, in our previous experiments with test substrate 1 under standard Suzuki crosscoupling conditions in THF with Pd(PtBu3)2 (3 mol %) in the presence of excess phosphine ligand, we had only observed C− Cl functionalization, in accord with the reactivity of a monoligated Pd0PtBu3 as the catalytically active species.27 In contrast, the sterically slightly less bulky ligand P(iPr)(tBu2) gives a reversal in selectivity upon an increase of the number of equivalents of phosphine ligand (see below for further discussion).27 Does this mean that there is an overestimation of dispersion forces? To explore this further, we evaluated the influence of basis set and solvent model on the predicted selectivities. It is challenging to accurately account for dispersion in solution. 9 a, 28,29 The COSMO-RS solvation model was suggested to treat the latter interaction more adequately by accounting for solute−solvent dispersion effects through surface-proportional terms.30,28a Thus, we also considered the energy difference (ΔΔG⧧) between the monoligated and bisligated pathways with COSMO-RS. Figure 6 presents the results. The COSMO-RS solvation model predicts slightly larger ΔΔG⧧ values relative to the CPCM results but still significantly smaller selectivity differences (ΔΔG⧧) relative to “dispersion-free” DFT. We next investigated the effect of the basis set. The 6311+G(d) basis set had been used in Figures 5 and 6, and the calculated predictions could potentially be a result of basis set errors, especially basis set superposition errors (BSSE).31 If large, BSSEs could lead to an overstabilization of TS1bis‑OTF, hence affording artificially small ΔΔG⧧ values. We therefore also examined the effect of the basis sets def2-TZVP and the large def2-QZVP that have previously been suggested to give

Figure 4. Favored reaction pathways for oxidative addition with dispersion (D3-corrected) (top) and without (bottom), calculated at CPCM (THF) PBE0/6-311+G(d)//ωB97XD/6-31G(d) with LANL2DZ (for Pd). The dissociative pathway is shown in black and the associative pathway in blue. Free energies (ΔG) are given in kcal/ mol.

311+G(d) basis set and application of solvation corrections for THF (CPCM). Dispersion was found to have a dramatic effect D

dx.doi.org/10.1021/om501199t | Organometallics XXXX, XXX, XXX−XXX



consequence of artificial stabilization of the bisligated pathway due to BSSEs. With all data clearly indicating that a monoligated pathway is favored for oxidative addition, we also examined the predicted selectivity (ΔΔG⧧) for C−Cl vs C−OTf addition for the monoligated pathway arising from oxidative addition by PdPtBu3. Figure 8 gives the predicted ΔΔG⧧ values for TSmono‑Cl vs TSmono‑OTf, calculated with a variety of methods using the solvation model COSMO-RS (for THF).

Figure 6. Predicted selectivity (ΔΔG⧧) for C−OTf versus C−Cl addition for L = PtBu3 for COSMO-RS (THF) DFT/6-311+G(d) with LANL2DZ (for Pd), with (blue) and without (green) dispersion.

smaller BSSEs.32 Figure 7 gives a summary of the computed ΔΔG⧧ values for three different basis sets, comparing 6-31G(d) Figure 8. Predicted selectivity (ΔΔG⧧) for C−OTf versus C−Cl addition for L = PtBu3, with and without dispersion; calculated at COSMO-RS (THF) DFT/def2-QZVP.

The selectivity trends predicted by most methods (dispersion corrected or not) are in agreement with the experimentally observed selectivity, favoring oxidative addition to the C−Cl bond. However, the dispersion-corrected data show in some cases ΔΔG⧧ values that are too small to be 100% selective for C−Cl and for B3LYP-D3 and even a wrong selectivity prediction (Figure 8). Given that free energies were considered, errors associated with the frequency calculations may potentially play a role.32 Hence, we also compared the selectivity prediction on the basis of the electronic energies (ΔΔE⧧; Figure 9) between TSmono‑Cl

Figure 7. Evaluation of effect of basis set on the predicted selectivity (ΔΔG⧧ in kcal/mol) for C−OTf versus C−Cl addition relative to monoligated pathway for L = PtBu3, for COSMO-RS (THF) and selected DFT methods.

with def2-TZVP and def2-QZVP, using the solvation model COSMO-RS. The corresponding (and largely analogous) results for CPCM are presented in the Supporting Information (Figure S4). Calculations with the 6-31G(d) basis set predict relatively small ΔΔG⧧ values (6−10 kcal/mol), similar to the results obtained with 6-311+G(d) in Figure 6. Within the Ahlrich basis set series, the predicted ΔΔG⧧ values were slightly larger, approximately 11−14 kcal/mol with def2-TZVP, and increased by another 1 kcal/mol with the larger def2-QZVP basis set. This basis set dependence was found to be independent of the solvation model employed. These results are in line with our previous experimental observations: i.e., the bisligated pathway cannot likely be adopted experimentally. Thus, these calculations suggest that the small ΔΔG⧧ value predicted above (in Figure 5) may predominantly be a

Figure 9. Predicted selectivity (ΔΔE⧧) for C−OTf versus C−Cl addition for L = PtBu3, with and without dispersion, calculated at COSMO-RS (THF) DFT/def2-QZVP. E

dx.doi.org/10.1021/om501199t | Organometallics XXXX, XXX, XXX−XXX



and TSmono‑OTf, predicting results in better agreement with experiments. Standard (dispersion-free) DFT methods qualitatively predicted the selectivities correctly in all cases. Investigation of the Predicted Site Selectivity for the Oxidative Addition of PdL2 to 1 with L = P(iPr)(tBu2). We subsequently extended our studies to the slightly less bulky analogue P(iPr)(tBu2) (cone angle 175°).14 Our group recently established that for this ligand the ligation state of the reactive Pd species may be altered.27 With PdL2 as catalyst, exclusive reaction with C−Cl was obtained in Suzuki cross-coupling reactions with 4-chlorophenyl triflate (1) in THF at room temperature. However, addition of a 10-fold excess of L = P(iPr)(tBu2) relative to PdL2 led to a complete selectivity reversal with exclusive C−OTf functionalization, in accord with a Pd bisphosphine species being reactive. This system should therefore furnish as an ideal additional test system to further probe the role of dispersion as a selectivity-controlling factor. In contrast to the case for L = PtBu3, for the oxidative addition of PdL2 to 1 with L = P(iPr)(tBu2) we successfully located bisligated transition states for the oxidative addition to C−OTf and C−Cl with (ωB97XD) and without dispersion (PBE0). In addition, for this ligand, the predicted selectivity preference was C−Cl addition for the reaction of PdL and C− OTf addition for the reaction of PdL2. However, in terms of the relative energy preference of the monoligated versus the bisligated pathway, we once again saw a striking effect of dispersion. Figure 10 presents the calculated selectivity

ligation state of the reactive palladium species and hence the overall site selectivity. In conclusion, using DFT methods that account well for dispersion, we successfully located the first bisphosphine transition states for the oxidative addition of Pd0(PtBu3)2 to activated aromatic C−O bonds. These transition states cannot be located with standard DFT methods, indicating that attractive H- - -H interactions6,7 may also be of importance for bulky hydrocarbon substituents of phosphine ligands. Extension of studies to the slightly smaller ligand analogue P(iPr)(tBu2) revealed that the same attractive dispersion interactions are a selectivity-controlling factor in monophosphine- versus bisphosphine-ligated pathways, with strikingly different selectivity predictions observed, depending on whether dispersion was accounted for or not. Overall, this study manifests the crucial role of dispersion in organometallic reactivity. In the context of ligand design, such attractive dispersion forces of bulky aliphatic ligand substituents may therefore be considered as an additional guiding design principle, together with the well-established parameters to correlate steric and electronic influences of ligands, such as the Tolman cone angle θ,1,14a percent buried volume (%Vbur),33 and the CO stretching frequency of [Ni(CO)3(L)] complexes.14a


Figure 10. Predicted selectivity (ΔΔG⧧) for C−OTf versus C−Cl addition for L = P(iPr)(tBu2), with and without dispersion, calculated at COSMO-RS (THF) DFT/def2-QZVP.


All geometries were optimized in the gas phase using dispersioncontaining methods such as B3LYP34-D335 and ωB97XD36 with the basis sets 6-31G(d) and an effective core potential (ECP) for Pd (LANL2DZ37) or def2-SVP, unless stated otherwise. Frequency calculations were performed at the same level of theory and were used to verify the nature of the stationary point as either a transition state or minimum. Transition states were further verified by following the intrinsic reaction coordinate (IRC) pathways. Single-point energies have been calculated and compared using a range of methods, with or without dispersion corrections, and various basis sets, including 6311+G(d), def2-TZVP, and def2-QZVP. The solvent models COSMO-RS and CPCM have been used to describe the solvation influence of THF. COSMO-RS values were obtained with COSMOtherm.38 Input files for COSMOtherm were prepared with Orca39 at 298 K with BP86/TZVP single-point calculations. D3 corrections have been employed using the original damping function.35 The standard state was converted to 1 M in solution (+1.89 kcal/mol). DFT calculations were performed using Gaussian 09 revision A.02 or D.01.17 ωB97XD optimizations and frequency calculations were calculated using Gaussian 09 revision A.02. B3LYP-D3 optimizations and frequency calculations were obtained using Gaussian 09 revision D.01. Likewise, single-point energies were calculated with Gaussian 09 revision D.01. Computational figures were created with CYLview.40 S Supporting Information *

(ΔΔG⧧) between C−OTf (TSbis‑OTf) versus C−Cl (TSmono-Cl) addition. Dispersion-free DFT methods favor the pathway involving ligand dissociation and oxidative addition at C−Cl via a monoligated Pd species by 22−25 kcal/mol, which would not be in line with the observed experimental selectivity switch at a higher concentration of ligand. Inclusion of dispersion, on the other hand, gives a dramatic drop in ΔΔG⧧ of ∼0−4 kcal/mol, which would be within the expected range of a switchable site selectivity.41 Considering that this system also has significant conformational freedom, this selectivity range can be considered acceptable. Overall, these results indicate that attractive dispersion forces between the alkyl substituents of the phosphine ligands are crucial in controlling the favored

Text, figures, tables, and an xyz file giving computational information, additional structures, Cartesian coordinates of calculated species, and the full ref 17. This material is available free of charge via the Internet at http://pubs.acs.org.


Corresponding Author

*E-mail for F.S.: [email protected]. Author Contributions §

These authors contributed equally.


The authors declare no competing financial interest. F

dx.doi.org/10.1021/om501199t | Organometallics XXXX, XXX, XXX−XXX



77, 799. (c) Kobylianskii, I. J.; Widner, F. J.; Kräutler, B.; Chen, P. J. Am. Chem. Soc. 2013, 135, 13648. (12) (a) Proutiere, F.; Schoenebeck, F. Angew. Chem., Int. Ed. 2011, 50, 8192. (b) Bonney, K. J.; Proutiere, F.; Schoenebeck, F. Chem. Sci. 2013, 4, 4434. (c) Sanhueza, I. A.; Wagner, A. M.; Sanford, M. S.; Schoenebeck, F. Chem. Sci. 2013, 4, 2767. (d) Nielsen, M. C.; Lyngvi, E.; Schoenebeck, F. J. Am. Chem. Soc. 2013, 135, 1978. (e) Lyngvi, E.; Schoenebeck, F. Tetrahedron 2013, 69, 5715. (f) Proutiere, F.; Aufiero, M.; Schoenebeck, F. J. Am. Chem. Soc. 2012, 134, 606. (g) Aufiero, M.; Proutiere, F.; Schoenebeck, F. Angew. Chem., Int. Ed. 2012, 51, 7226. (h) Anstaett, P.; Schoenebeck, F. Chem. Eur. J. 2011, 17, 12340. (13) (a) Lommel, W.; Muenzel, H. (I.G. Farbenindustrie AG) Verfahren zur Herstellung von Phosphoniumverbindungen. Ger. Offen. DE 730638, January 15, 1943. (b) Nicholas, P. P. J. Org. Chem. 1987, 52, 5266. (c) Littke, A. F.; Dai, C.; Fu, G. C. J. Am. Chem. Soc. 2000, 122, 4020. (d) Dai, C.; Fu, G. C. J. Am. Chem. Soc. 2001, 123, 2719. (e) Yamamoto, T.; Nishiyama, M.; Koie, Y. Tetrahedron Lett. 1998, 39, 2367. (f) Littke, A. F.; Fu, G. C. Angew. Chem., Int. Ed. 1998, 37, 3387. (14) (a) Tolman, C. A. Chem. Rev. 1977, 77, 313. (b) Bilbrey, J. A.; Kazez, A. H.; Locklin, J.; Allen, W. D. J. Comput. Chem. 2013, 34, 1189. (15) (a) Xue, L.; Lin, Z. Chem. Soc. Rev. 2010, 39, 1692. (b) Jover, J.; Fey, N.; Lloyd-Jones, G. C.; Harvey, J. N. J. Mol. Catal. A 2010, 324, 39. (c) Kozuch, S.; Martin, J. M. L. ACS Catal. 2011, 1, 246. (d) Li, Z.; Fu, Y.; Guo, Q.-X.; Liu, L. Organometallics 2008, 27, 4043. (e) Lam, K. C.; Marder, T. B.; Lin, Z. Organometallics 2007, 26, 758. (f) Ahlquist, M.; Norrby, P.-O. Organometallics 2007, 26, 550. (g) Vikse, K.; Naka, T.; McIndoe, J. S.; Besora, M.; Maseras, F. ChemCatChem. 2013, 5, 3604. (16) Schoenebeck, F.; Houk, K. N. J. Am. Chem. Soc. 2010, 132, 2496. (17) Frisch, M. J.; et al. Gaussian09, Revision D.01 and A.01; Gaussian Inc., Wallingford, CT. See the Supporting information for the full reference and more information. (18) Geometries of bisligated transition states located with other methods are illustrated in the Supporting Information. (19) For a discussion of potential overbinding at B3LYP-D3, see: Schneebeli, S. T.; Bochevarov, A. D.; Friesner, R. A. J. Chem. Theory Comput. 2011, 7, 658. (20) Bader, R. F. W. Atoms in Molecules; Oxford Science Publications: Oxford, U.K., 1990. (21) Selected examples: (a) Andrejeva, A.; Gardner, A. M.; Graneek, J. B.; Plowright, R. J.; Breckenridge, W. H.; Wright, T. G. J. Phys. Chem. A 2013, 117, 13578. (b) César, V.; Misal Castro, L. C.; Dombray, T.; Sortais, J.-B.; Darcel, C.; Labat, S.; Miqueu, K.; Sotiropoulos, J.-M.; Brousses, R.; Lugan, N.; Lavigne, G. Organometallics 2013, 32, 4643. (c) Farrugia, L. J.; Evans, C.; Senn, H. M.; Hänninen, M. M.; Sillanpäa,̈ R. Organometallics 2012, 31, 2559. (d) Farrugia, L. J.; Evans, C.; Lentz, D.; Roemer, M. J. Am. Chem. Soc. 2009, 131, 1251. (22) Lu, T.; Chen, F. Multiwfn: A Multifunctional Wavefunction Analyzer. J. Comput. Chem. 2012, 33, 580. (23) While this approach can provide valuable information on the nature of an interaction, concerns have also been raised. For discussion, see: Foroutan-Nejad, C.; Shahbazian, S.; Marek, R. Chem. Eur. J. 2014, 20, 1. (24) (a) Varadwaj, P. R.; Cukrowski, I.; Marques, H. M. J. Phys. Chem. A 2008, 112, 10657. (b) Varadwaj, P. R.; Varadwaj, A.; Marques, H. M. J. Phys. Chem. A 2011, 115, 5592. (25) The calculated |Vb|/Gb ratios suggest that the Pd6−P7 bonding interaction has greater ionic character than the Pd6−P9 interaction. (26) In metal−ligand bonds, typical values for Hb are small and negative, while those for ∇2ρb are positive, with the ratio |Vb|/Gb being between 1 and 2. See ref 24. (27) Proutiere, F.; Lyngvi, E.; Aufiero, M.; Sanhueza, I. A.; Schoenebeck, F. Organometallics 2014, 33, 6879. (28) (a) Grimme, S. ChemPhysChem 2012, 13, 1407. (b) Jacobsen, H.; Cavallo, L. ChemPhysChem 2012, 13, 1405. (29) Yang, L.; Adam, C.; Nichol, G. S.; Cockroft, S. L. Nat. Chem. 2013, 5, 1006.

ACKNOWLEDGMENTS We thank the MIWF NRW, RWTH Aachen, the SNF and ETH Zürich for financial support. Calculations were performed on the ETH High Performance Cluster “Brutus”.


(1) (a) Handbook of Organopalladium Chemistry for Organic Synthesis; Negishi, E.-I., de Meijere, A., Eds.; Wiley: New York, 2002. Metal-Catalyzed Cross-Coupling Reactions, 2nd ed.; Diederich, F., de Meijere, A., Eds.; Wiley-VCH: Weinheim, Germany, 2004. (b) Hartwig, J. F. Organotransition Metal Chemistry-From Bonding to Catalysis; University Science Books: Sausalito, CA, 2010. (c) Seechurn, C. C. J.; Kitching, M. O.; Colacot, T. J.; Snieckus, V. Angew. Chem., Int. Ed. 2012, 51, 5062. (2) For a recent example of computational ligand design, see: Nielsen, M. C.; Bonney, K. J.; Schoenebeck, F. Angew. Chem., Int. Ed. 2014, 53, 5903. (3) For examples of combining experimental and computational mechanistic organometallic studies, see: (a) Mollar, C.; Besora, M.; Maseras, F.; Asensio, G.; Medio-Simon, M. Chem. Eur. J. 2010, 16, 13390. (b) Bedford, R. B.; Gower, N. J.; Haddow, M. F.; Harvey, J. N.; Nunn, J.; Okopie, R. A.; Sankey, R. F. Angew. Chem., Int. Ed. 2012, 51, 5435. (c) Sköld, C.; Kleimark, J.; Trejos, A.; Odell, L. R.; Lill, S. O. N.; Norrby, P.-O.; Larhed, M. Chem. Eur. J. 2012, 18, 4714. (d) Butts, C.; Filali, E.; Lloyd-Jones, G.; Norrby, P.-O.; Sale, D.; Schramm, Y. J. Am. Chem. Soc. 2009, 131, 9945. (e) For a review, see: Bonney, K. J.; Schoenebeck, F. Chem. Soc. Rev. 2014, 43, 6609. (4) (a) Wu, Q.; Yang, W. J. Chem. Phys. 2002, 116, 515. (b) Grimme, S. J. Comput. Chem. 2004, 25, 1463. (c) Morgado, C.; Vincent, M. A.; Hillier, I. H.; Shan, X. Phys. Chem. Chem. Phys. 2007, 9, 448. (d) Jurecka, P.; Cerny, J.; Hobza, P.; Salahub, D. R. J. Comput. Chem. 2007, 28, 555. (e) Zhao, Y.; Truhlar, D. G. J. Phys. Chem. A 2005, 109, 5656. (5) (a) Zhao, Y.; Truhlar, D. G. Org. Lett. 2007, 9, 1967. (b) Benitez, D.; Tkatchouk, E.; Goddard, W. A. Chem. Commun. 2008, 6194. (c) Jacobsen, H. Phys. Chem. Chem. Phys. 2009, 11, 7231. (d) Schwabe, T.; Grimme, S.; Djukic, J.-P. J. Am. Chem. Soc. 2009, 131, 14156. (e) Fokin, A. A.; Gerbig, D.; Schreiner, P. R. J. Am. Chem. Soc. 2011, 133, 20036. (f) Grimme, S. Chem. Eur. J. 2012, 18, 9955. (g) Arp, H.; Baumgartner, J.; Marschner, C. J. Am. Chem. Soc. 2012, 134, 6409. (h) Nilsson Lill, S. O.; Ryberg, P.; Rein, T.; Bennstrçm, E.; Norrby, P.O. Chem. Eur. J. 2012, 18, 1640. (i) Dieckmann, A.; Houk, K. N. J. Chem. Theory Comput. 2012, 8, 5064. (j) Ehrlich, S.; Bettinger, H. F.; Grimme, S. Angew. Chem., Int. Ed. 2013, 52, 10892. (k) Rekken, B. D.; Brown, T. M.; Fettinger, J. C.; Lips, F.; Tuononen, H. M.; Herber, R. H.; Power, P. P. J. Am. Chem. Soc. 2013, 135, 10134. (l) Zhu, J.; Liang, Y.; Wang, L.; Zheng, Z.-B.; Houk, K. N.; Tang, Y. J. Am. Chem. Soc. 2014, 136, 6900. (6) (a) Schreiner, P. R.; Chernish, L. V.; Gunchenko, P. A.; Tikhonchuk, E. Y.; Hausmann, H.; Serafin, M.; Schlecht, S.; Dahl, J. E. P.; Carlson, R. M. K.; Fokin, A. A. Nature 2011, 477, 308. (b) Fokin, A. A.; Chernish, L. V.; Gunchenko, P. A.; Tikhonchuk, E. Y.; Hausmann, H.; Serafin, M.; Dahl, J. E. P.; Carlson, R. M. K.; Schreiner, P. R. J. Am. Chem. Soc. 2012, 134, 13641. (7) Grimme, S.; Schreiner, P. R. Angew. Chem., Int. Ed. 2011, 50, 12639. (8) For the nature of CH···HC interactions, see: Danovich, D.; Shaik, S.; Neese, F.; Echeverría, J.; Aullón, G.; Alvarez, S. J. Chem. Theory Comput. 2013, 9, 1977. (9) Ahlquist, M. S. G.; Norrby, P.-O. Angew. Chem., Int. Ed. 2011, 50, 11794. (10) (a) McMullin, C. L.; Fey, N.; Harvey, J. N. Dalton Trans. 2014, 43, 13545. (b) McMullin, C. L.; Jover, J.; Harvey, J. N.; Fey, N. Dalton Trans. 2010, 39, 10833. (11) For examples, see: (a) Jacobsen, H.; Cavallo, L. ChemPhysChem 2012, 13, 562. (b) Ronnholm, P.; Nilsson Lill, S. O.; Grafenstein, J.; Norrby, P.-O.; Pettersson, M.; Hilmersson, G. ChemPlusChem 2012, G

dx.doi.org/10.1021/om501199t | Organometallics XXXX, XXX, XXX−XXX



(30) Klamt, A. J. Phys. Chem. 1995, 99, 2224. (31) (a) Jansen, H. B.; Ros, P. Chem. Phys. Lett. 1969, 3, 140. (b) Liu, B.; McLean, A. D. J. Chem. Phys. 1973, 59, 4557. (32) (a) Ribeiro, R. F.; Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B 2011, 115, 14556. (b) Grimme, S. Chem. Eur. J. 2012, 18, 9955. (33) (a) Hillier, A. C.; Sommer, W. J.; Yong, B. S.; Petersen, J. L.; Cavallo, L.; Nolan, S. P. Organometallics 2003, 22, 4322. (b) Poater, A.; Cosenza, B.; Correa, A.; Giudice, S.; Ragone, F.; Scarano, V.; Cavallo, L. Eur. J. Inorg. Chem. 2009, 1759. (c) Clavier, H.; Nolan, S. P. Chem. Commun. 2010, 46, 841. (34) (a) Kohn, W.; Becke, A. D.; Parr, R. G. J. Phys. Chem. 1996, 100, 12974. (b) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (35) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys. 2010, 132, 154104. (36) Chai, J.-D.; Head-Gordon, M. Phys. Chem. Chem. Phys. 2008, 10, 6615. (37) For discussion of the suitability of LANL2DZ as ECP for Pd, see for instance: Roy, L. E.; Roy, P. J.; Martin, R. L. J. Chem. Theory Comput. 2008, 4, 1029. (38) Eckert, F.; Klamt, A. COSMOtherm, Version C2.1, Release 01.11; COSMOlogic GmbH & Co. KG, Leverkusen, Germany, 2010. (39) Neese, F. ORCA, Version 2.9 (Revision 0); MPI for Bioinorganic Chemistry, Mülheim an der Ruhr, Germany, 2011. (40) Legault, C. Y. CYLview, 1.0b; Université de Sherbrooke, 2009 (http://www.cylview.org). (41) The energies are calculated relative to a 1 M standard state. The true concentrations in catalytic reactions are lower, however, so a greater energy difference between the pathways can be expected.

NOTE ADDED AFTER ASAP PUBLICATION In the version of this paper that was published on December 11, 2014, there were errors in some of the formulas in Figure 4. In the version of the paper that appears as of December 12, 2014, the formulas are now correct.


dx.doi.org/10.1021/om501199t | Organometallics XXXX, XXX, XXX−XXX