Mechanisms and Origins of Chemo- and Regioselectivities of Ru(II

May 12, 2017 - The decarboxylation barrier is very sensitive to the tether length, and only the seven-membered ring intermediate can selectively under...
2 downloads 0 Views 8MB Size
Download PDF
Article pubs.acs.org/JACS

Mechanisms and Origins of Chemo- and Regioselectivities of Ru(II)Catalyzed Decarboxylative C−H Alkenylation of Aryl Carboxylic Acids with Alkynes: A Computational Study Jing-Lu Yu,† Shuo-Qing Zhang,† and Xin Hong* Department of Chemistry, Zhejiang University, Hangzhou 310027, China S Supporting Information *

ABSTRACT: The mechanisms and chemo- and regioselectivities of Ru(II)-catalyzed decarboxylative C−H alkenylation of aryl carboxylic acids with alkynes were investigated with density functional theory (DFT) calculations. The catalytic cycle involves sequential carboxylate-directed C−H activation, alkyne insertion, decarboxylation and protonation. The facile tether-assisted decarboxylation step directs the intermediate toward the desired decarboxylative alkenylation, instead of typical annulation and double alkenylation pathways. The decarboxylation barrier is very sensitive to the tether length, and only the seven-membered ring intermediate can selectively undergo the designed decarboxylation, suggesting a tether-dependent chemoselectivity. This tether-dependent chemoselectivity also applies to the alkyl tethers. In addition, the polarity of solvent is found to control the chemoselectivity between the decarboxylative alkenylation and [4 + 2] annulation. Solvent with low polarity (toluene) favors the decarboxylation pathway, leading to the decarboxylative alkenylation. Solvent with high polarity (methanol) favors the ionic stepwise C−O reductive elimination pathway, leading to the [4 + 2] annulation. To understand the origins of regioselectivity with asymmetric alkynes, the distortion/interaction analysis was applied to the alkyne insertion transition states, and led to a predictive frontier molecular orbital model. The asymmetric alkynes selectively use the terminal with the larger HOMO orbital coefficient to form the C−C bond in the insertion step.

■

transformation,4−7 which limits the synthetic application of this methodology. Recently, Zhao et al.,8 Ackermann and co-workers,9 and Gooβen et al.10 independently reported an elegant design of Ru(II)-catalyzed decarboxylative C−H alkenylation of aryl carboxylic acids with alkynes (Scheme 1). Through a designed alkenyl-assisted decarboxylation, a perfect synergy between the C−H functionalization and decarboxylation is achieved. The desired alkenylation proceeds with exceptional chemo- and regioselectivities under mild conditions, and the possible annulation products are not competitive under these conditions (Scheme 1). Intrigued by the judicious balance between the C− H functionalization and decarboxylation, we used DFT calculations to study this alkenylation reaction. The calculations indicated a mechanism involving sequential C−H activation, alkyne insertion, decarboxylation, and protonation. The

INTRODUCTION Carboxylate, as a small and ubiquitous functional group, has remarkable potential as a directing group in C−H functionalization. It is widely available in organic compounds and feedstock chemicals. Being one of the central synthons, carboxylate can be directly generated from a variety of functional groups and derivatized further.1 These unique features allow the carboxylateassisted C−H functionalization of great synthetic value, and substantial advances have been achieved toward the ideal atomand step-economy for this transformation.2 Combining the carboxylate-assisted C−H functionalization and metal-mediated decarboxylation provides a useful strategy to utilize carboxylate as a removable directing group in C−H functionalization.3 This approach has been achieved through a tandem or sequential process via a variety of transition metal catalysts, including Pd,4 Rh,5 Ir,6 and Cu,7 but the intrinsic high barrier of decarboxylation usually requires harsh reaction conditions. High reaction temperature and stoichiometric silver or copper oxidants are generally involved to facilitate the desired © 2017 American Chemical Society

Received: January 21, 2017 Published: May 12, 2017 7224

DOI: 10.1021/jacs.7b00714 J. Am. Chem. Soc. 2017, 139, 7224−7243

Article

Journal of the American Chemical Society

correction of 4.3 kcal/mol is applied to the reaction free energy (i.e., a reaction from m- to n-components has an additional free energy correction for (n − m) × 4.3 kcal/mol). This approach has been validated through a number of computational and experimental studies. Yu and co-workers have found that the entropy corrections are overestimated by about half in several cycloaddition reactions.24 Wang and co-workers have discovered the improved description of free energy changes in a number of metal-catalyzed reactions using the empirical approach from Martin.25 In addition, to correct the Gibbs free energies under pressure of 1 atm to the standard state in solution (1 mol/L), a correction of RT ln(cs/cg) (about 1.9 kcal/mol) is added to energies of all species except carbon dioxide. cs is the standard molar concentration in solution (1 mol/L), cg is the standard molar concentration in gas phase (0.0446 mol/L), and R is the gas constant. To quantify the effects of dispersion interaction, we reoptimized the resting state, rate-limiting transition state, and key selectivitydetermining transition states with B3LYP-D3 (Becke−Johnson damping function)26 functional. The overlay of the two sets of optimized structures shows excellent convergence (Figure S1), suggesting that the dispersion interaction has limited effects on the structures of the key intermediate and transition states that are involved in this reaction. The convergence of the two sets of optimized structures is further validated by calculating the energies at the M06/6-311+G(d,p)-SDD-SMD(toluene) level of theory. The energy differences between the two sets of optimized structures are very small, with a maximum absolute deviation of 1.1 kcal/mol (Figure S1).

Scheme 1. Chemo- and Regioselectivities of [Ru(p-Cymene) (OAc)2]-Catalyzed Decarboxylative C−H Alkenylation of Aryl Carboxylic Acids with Alkynes11

■

RESULTS AND DISCUSSION 1. Reaction Mechanism. Based on the previous mechanistic studies on Ru(II)-catalyzed C−H functionalization of arenes,27

competition between the alkenylation and annulations are explored to reveal the origins of chemoselectivity, and a unique solvent control is elucidated. In addition, the regioselectivity of alkyne insertion is explained through distortion/interaction analysis and frontier molecular orbital model.

■

Scheme 2. Proposed Mechanisms of [Ru(p-Cymene) (OAc)2]-Catalyzed Decarboxylative C−H Alkenylation of Aryl Carboxylic Acids with Alkynes

COMPUTATIONAL METHODS

All DFT calculations were carried out using Gaussian 09 program.12 All geometry optimizations were performed with the B3LYP functional13 using the LANL2DZ basis set14 for ruthenium and the 6-31G(d) basis set for the other atoms. The vibrational frequencies were computed at the same level of theory as for the geometry optimizations to confirm whether each optimized structure is an energy minimum or a transition state, and to evaluate the zero-point vibrational energy (ZPVE) and thermal corrections at 298 K. The single-point energies and solvent effects were computed with the M06 functional15 using the SDD basis set16 for ruthenium and the 6-311+G(d,p) basis set for the other atoms, based on the gas-phase optimized structures. The solvation energies were evaluated by a self-consistent reaction field (SCRF) using the SMD implicit solvent model.17 The parameters for toluene are chosen to avoid the complicated scenario of the mixed solvents (1,4-dioxane/ mesitylene/heptane), experimentally toluene solvent gives moderate yields for the desired alkenylation product with similar selectivities.8 The results in implicit toluene model are also compared to those in 1,4dioxane, mesitylene, heptane and methanol. The detailed solvent effects are discussed later. Fragment distortion and interaction energies were computed at the M06/6-311+G(d,p)-SDD level using the B3LYP/631G(d)-LANL2DZ geometries in the gas phase. The HOMO orbital coefficient was calculated by using the NBO18 module of Gaussian 09, using Multiwfn.19 Extensive conformational searches were conducted for all intermediates and transition states, and the lowest energy conformers are shown in this work. The 3D diagrams of molecules were generated using CYLView.20 The overlay of structures and root-meansquare deviations are generated using VMD.21 Because the thermal corrections are based on the ideal gas model, this approach ignores the solvent suppression on the rotational and translational freedoms of solutes, resulting in overestimation of entropy contributions to the reaction free energies in solution.22 To correct the entropy change in solution, we applied an empirical approach proposed by Martin and co-workers,23 because there is currently no widely accepted quantum mechanics-based approach to correct entropy in solution. For each component change in a reaction at 298 K and 1 atm, a

the proposed mechanisms of the [Ru(p-cymene) (OAc)2]catalyzed decarboxylative C−H alkenylation of aryl carboxylic acids with alkynes are shown in Scheme 2. Starting with the LRu(carboxylate)2 intermediate 1, the initial aryl C−H activation generates the cyclometalated intermediate 2. Subsequent ligand exchange with alkyne gives intermediate 3, and alkyne insertion into the aryl-ruthenium bond produces the seven-membered ring alkenylruthenium intermediate 4. This intermediate undergoes the proposed alkenyl-assisted decarboxylation to generate the 7225

DOI: 10.1021/jacs.7b00714 J. Am. Chem. Soc. 2017, 139, 7224−7243

Article

Journal of the American Chemical Society

Figure 1. DFT-computed Gibbs free energy changes (in toluene) of the most favorable pathway of [Ru(p-cymene) (OAc)2]-catalyzed decarboxylative C−H alkenylation of benzoic acid with diphenylacetylene.

ruthenium throughout the whole catalytic cycle, and the substitution of p-cymene coordination with substrate or product is thermodynamically unfavorable (Scheme S1). In addition, all the ruthenium complexes that have been characterized in the related studies have the p-cymene coordination.8,28b We have also considered the possible pathway with reversed sequence of protonations, first on the aryl-ruthenium bond and then on the alkenyl-ruthenium bond. This pathway is less favorable than that via TS20 and TS23 (Figure S3). Based on the calculated free energy changes of the whole catalytic cycle, the resting state of the catalytic cycle is the LRu(OOCPh) 2 intermediate 9, and the rate-determining step is the alkyne insertion via TS13 with a 25.7 kcal/mol overall barrier. This overall barrier is in agreement with the experimental conditions (80 °C, 24 h). The alternative alkenylation pathway in which the decarboxylation occurs between the two protonation steps is quite unfavorable (red pathway in Scheme 2). It requires a 36.2 kcal/mol overall barrier due to the difficult decarboxylation step without the assistance of alkenyl tether (Figure S4). The tether effects on the decarboxylation barrier are discussed later. 2. Origins of Chemoselectivity. In addition to the alkenylation product, a number of annulation products can be generated through the seven-membered ring Ru(II) intermediate 16 (Scheme 3). From 16, a C−O reductive elimination generates the oxidative [4 + 2] annulation product 25. This [4 + 2] annulation, discovered by Ackerman and co-workers,28 dominates in the [Ru(p-cymene) (OAc)2]-catalyzed reaction between benzoic acids and internal alkynes with high-polar solvent and stoichiometric oxidant. Thus, the experimental condition plays an important role in determining the chemo-

ruthenacyclopentadiene intermediate 5 (blue pathway). From 5, two protonations with aryl carboxylic acid produce the alkenylation product and regenerate the intermediate 1. Alternatively, the seven-membered ring intermediate 4 can undergo a protonation to form the intermediate 6 (red pathway), which then undergoes the decarboxylation to generate the arylruthenium intermediate 7. Subsequent protonation produces the same alkenylation product and intermediate 1. We first studied the proposed catalytic cycles, using benzoic acid and diphenylacetylene as the model substrates. The free energy changes of the most favorable pathway leading to the alkenylation product are shown in Figure 1. The optimized structures of selected intermediates and transition states are shown in Figure 2. From the experimental catalyst precursor, 8, ligand exchanges with benzoic acids lead to the more stable LRu(OOCPh)2 intermediate 9. This intermediate undergoes a concerted metalation-deprotonation (CMD) step via TS10, generating the cyclometalated intermediate 11. The alternative CMD transition state using acetate as the intramolecular base is less favorable as compared to TS10 (Figure S2). From 11, subsequent ligand exchange with alkyne gives intermediate 12, and alkyne insertion through TS13 produces the alkenylruthenium intermediate 14. 14 undergoes a facile isomerization to generate the benzene-coordinated intermediate 16, and subsequent decarboxylation via TS17 leads to the ruthenacyclopentadiene intermediate 18. From 18, the first protonation of alkenyl-ruthenium bond via TS20 is fast and reversible, and the second protonation of aryl-ruthenium bond via TS23 irreversibly produces the alkenylation product 24 and regenerates the intermediate 9. The p-cymene ligand is coordinated to 7226

DOI: 10.1021/jacs.7b00714 J. Am. Chem. Soc. 2017, 139, 7224−7243

Article

Journal of the American Chemical Society

Figure 2. DFT-optimized structures of selected intermediates and transition states of the [Ru(p-cymene) (OAc)2]-catalyzed decarboxylative C−H alkenylation of benzoic acid with diphenylacetylene. Only the α-carbons of certain phenyl groups are shown for simplicity. L is the p-cymene ligand.

selectivity between the decarboxylative alkenylation and [4 + 2] annulation. Alternatively, 16 can undergo an alkyne insertion to give the nine-membered ring intermediate 26, and subsequent C−O reductive elimination generates the eight-membered ring [4 + 2 + 2] annulation product 27. To the best of our knowledge, this [4 + 2 + 2] annulation has not been observed in related studies. In addition, the decarboxylation of 16 can occur to generate the five-membered ring intermediate 18, and the secondary insertion of alkyne leads to the seven-membered ring intermediate, 28 or 29. Both 28 and 29 can undergo the C−C reductive elimination to produce the decarboxylative [2 + 2 + 2] annulation product 30. The intermediate 28 can also be generated from the decarboxylation of 26. According to Zhao’s experimental results,8 the decarboxylative [2 + 2 + 2] annulation pathway is confirmed by stoichiometric reaction between Ru(pcymene) (OBz)2 complex and diphenylacetylene, and the 1,2,3,4-tetrasubstituted naphthalenes are the major byproducts of the reactions with electron-deficient benzoic acids. 2.1. [4 + 2] Annulation. We first studied the competition between the alkenylation and [4 + 2] annulation of intermediate 16. The free energy changes of the competing pathways are shown in Figure 3. The decarboxylation of 16 is facile with a barrier of 13.8 kcal/mol, and the ruthenacyclopentadiene intermediate 18 is irreversibly protonated to produce the

Scheme 3. Annulation Pathways of the Seven-Membered Ring Ru(II) Intermediate 16

7227

DOI: 10.1021/jacs.7b00714 J. Am. Chem. Soc. 2017, 139, 7224−7243

Article

Journal of the American Chemical Society

Figure 3. DFT-computed Gibbs free energy changes (in toluene) of the decarboxylative alkenylation pathway and [4 + 2] annulation pathway. Gibbs free energies are compared to LRu(OAc)2 intermediate 8.

discussions). This allows the intermediate 16 to undergo the designed decarboxylative alkenylation, instead of the oxidative [4 + 2] annulation. Importantly, the polar nature of the ionic stepwise reductive elimination pathway suggests that the energy changes of this pathway should be sensitive to the polarity of the solvent environment. Experimentally, the chemoselectivity between the decarboxylative alkenylation and [4 + 2] annulation can be switched with solvent. In Ackerman’s report,28a,b the [4 + 2] annulation is the major reaction with polar solvent and stoichiometric oxidant (such as cupper salt28a and molecular oxygen28b). The formation of the lactone-coordinated Ru(0) complex (e.g., 33), is confirmed by stoichiometric reactions (without oxidant) and X-ray single crystal structural analysis.28b The solvent control of this chemoselectivity is discussed later. 2.2. [4 + 2 + 2] Annulation. We next explored the competition between the alkenylation and [4 + 2 + 2] annulation. The free energy changes of the two competing pathways are shown in Figure 4. From 16, the coordination of diphenylacetylene gives the intermediate 37, and 37 undergoes the alkyne insertion via TS38. Subsequent C−O reductive elimination through TS39 produces the product-coordinated complex 40. We have also considered an alternative pathway of the [4 + 2 + 2] annulation in which the alkyne inserts into the Ru−O bond of 37, and the overall barrier is 14.1 kcal/mol higher as compared to the

alkenylation product (blue pathway). Alternatively, the C−O reductive elimination of 16 has two possible pathways, a neutral concerted pathway (labeled in green) and an ionic stepwise pathway (labeled in red). The neutral concerted pathway of C− O reductive elimination occurs via a classic three-centered transition state TS31, generating the lactone-coordinated intermediate 32. 32 further isomerizes to the stable Ru(0) complex 33, making the [4 + 2] annulation irreversible. In the ionic stepwise reductive elimination pathway, intermediate 16 undergoes a heterolytic cleavage of Ru−O bond via TS34 to generate the zwitterionic intermediate 35.29 Subsequent C−O bond formation via TS36 irreversibly produces the stable lactone-coordinated Ru(0) complex 33.30 Comparing the three competing pathways, the determining transition states are TS17, TS31 and TS34. The decarboxylation transition state, TS17, is over 4 kcal/mol more favorable than the other two competing transition states, suggesting a high chemoselectivity toward the decarboxylative alkenylation. This calculated chemoselectivity is consistent with Zhao’s8 and Ackerman’s9a experimental observations that the [4 + 2] annulation is not observed in toluene solvent. Through the judicious design of the cyclometalated ruthenium intermediate 16, the tether-assisted decarboxylation barrier is only 13.8 kcal/ mol, much lower than a decarboxylation barrier without tether assistance (about 25 kcal/mol, see Table 2 and related 7228

DOI: 10.1021/jacs.7b00714 J. Am. Chem. Soc. 2017, 139, 7224−7243

Article

Journal of the American Chemical Society

Figure 4. DFT-computed Gibbs free energy changes (in toluene) of the competing steps that determine the chemoselectivity between decarboxylative alkenylation and [4 + 2 + 2] annulation. Gibbs free energies are compared to LRu(OAc)2 intermediate 8.

pathway involving the alkyne insertion into the Ru-vinyl bond via TS38. Related details are included in Figure S5. The competition between the decarboxylative alkenylation and [4 + 2 + 2] annulation is determined by TS17 and TS38. TS17 is 5.7 kcal/ mol more stable than TS38, which is consistent with the experimental results that the [4 + 2 + 2] annulation is not observed. Again, the exceptionally low barrier of decarboxylation prevents the seven-membered ring intermediate 16 from participating in the possible annulation pathways. 2.3. [2 + 2 + 2] Annulation. For the [2 + 2 + 2] annulation leading to the napthalene derivatives, there are two possible pathways depending on the sequence of alkyne insertion and decarboxylation (Scheme 3). If the seven-membered ring intermediate 16 first undergoes the alkyne insertion and then the decarboxylation to achieve the [2 + 2 + 2] annulation, this [2 + 2 + 2] annulation pathway is much less favorable as compared to the decarboxylative alkenylation pathway. As shown in Figure 5, the seven-membered ring intermediate 16 can undergo the alkyne insertion through TS38 to generate the nine-membered ring intermediate 26. Subsequent decarboxylation via TS41 irreversibly gives the intermediate 28, and C−C reductive elimination through TS42 produces the annulation product. Compared to the productive alkenylation pathway, this [2 + 2 + 2] annulation pathway is much less favorable by 9.5 kcal/mol due to the 29.1 kcal/mol decarboxylation barrier of 26. This highlights the sensitivity of the decarboxylation barrier toward the tether length. Extending the tether length significantly increases the decarboxylation barrier, making this [2 + 2 + 2] annulation pathway unfavorable.

The alternative [2 + 2 + 2] annulation pathway involves the decarboxylation of intermediate 16 and alkyne insertion of ruthenacyclopentadiene intermediate 18 (Scheme 3). The free energy changes of this [2 + 2 + 2] pathway and the competing alkenylation pathway are shown in Figure 6. Since both pathways involve the decarboxylation of intermediate 16, the mechanistic divergence occurs after the decarboxylation. As discussed above, intermediate 19 undergoes two sequential protonations to produce the alkenylation product, with the second protonation via TS23 determining the overall efficiency. Alternatively, 19 can undergo an alkyne insertion via TS45 to generate the sevenmembered ring intermediate 29,31 and subsequent C−C reductive elimination irreversibly generates the napthalenecoordinated complex 43. Through the C−C reductive elimination via TS46, the Ru(II) catalyst is reduced to Ru(0) and loses the catalytic activity of decarboxylative alkenylation. Thus, the [2 + 2 + 2] annulation is detrimental to the desired alkenylation. Comparing the alkenylation and the [2 + 2 + 2] annulation involving decarboxylation of 16, the alkenylation pathway is 3.0 kcal/mol more favorable than the [2 + 2 + 2] annulation pathway (TS23 vs TS45, Figure 6). This 3.0 kcal/mol difference suggests that the diphenylacetlyene favors the alkenylation product, as observed in experiments,8 but this chemoselectivity can potentially be switched with substitution. Experimentally, the electron-deficient aryl carboxylic acids lead to the undesired [2 + 2 + 2] annulation product, and the napthalene-coordinated complex, 43, is isolated and characterized by single crystal X-ray diffraction.8 The substituent effects are discussed later. 7229

DOI: 10.1021/jacs.7b00714 J. Am. Chem. Soc. 2017, 139, 7224−7243

Article

Journal of the American Chemical Society

Figure 5. DFT-computed Gibbs free energy changes (in toluene) of the competing steps that determine the chemoselectivity between decarboxylative alkenylation and [2 + 2 + 2] annulation (via alkyne insertion of 16). Gibbs free energies are compared to LRu(OAc)2 intermediate 8.

2.4. Double Alkenylation. Double functionalization is a common side reaction in transition metal-catalyzed C−H functionalizations. However, in all the three reports from Zhao,8 Ackermann,9 and Gooβen,10 the double alkenylation products are not formed. In order to understand this chemoselectivity, we studied the free energy changes of the double alkenylation pathway, and the results are shown in Figure 7. From the seven-membered ring intermediate 16, the benzoic acid coordination generates the intermediate 47, and 47 undergoes the protonation of alkenyl-ruthenium bond via TS48. This protonation requires a 20.9 kcal/mol barrier as compared to 16, and irreversibly generates the intermediate 49. Subsequently, the sequential C−H activation, alkyne insertion and isomerization steps produce the seven-membered ring intermediate 56, and 56 undergoes the facile decarboxylation and protonations to eventually generate the double alkenylation product 57. The full free energy diagram of the double alkenylation pathway is included in the Supporting Information (Figure S6). The competition between the tether-assisted decarboxylation via TS17 and the protonation via TS48 determines the degree of functionalization. TS17 is 7.1 kcal/ mol more favorable than TS48, suggesting that the tetherassisted decarboxylation significantly outcompetes the protonation, leading to the monoalkenylation product. This agrees well with the experimental observations that no double

alkenylation products are formed, and also highlights the importance of the “deciduous” effects10 of the carboxylate directing group in this reaction. In addition, the C−H activation and alkyne insertion barriers depend highly on the ortho-vinyl substituent. In the productive monoalkenylation pathway, the barriers for C−H activation and alkyne insertion are 24.0 kcal/mol (9 to TS10) and 25.7 kcal/ mol (9 to TS13), respectively. While in the double alkenylation pathway, the additional ortho-vinyl substituent significantly increases the corresponding barriers, leading a 29.0 kcal/mol barrier for C−H activation (49 to TS50) and a 29.1 kcal/mol barrier for the alkyne insertion (49 to TS53). These changes of barriers are due to the steric repulsions between the bulky vinyl substituent and the carboxylate group in TS50 and TS53. During C−H activation and alkyne insertion, both transition states require the carboxylate-substituted phenyl ring to be coplanar with the carboxylate group, leading to significant steric repulsions between the vinyl substituent and the carboxylate group. To confirm this rationale on steric repulsions, we analyzed the distortion energies of the phenyl carboxylate moieties in the transition states (TS50 and TS53) and the intermediate 49. The changes of distortion energies are about 3 kcal/mol, which are comparable to the changes of barriers (Figure S7). Therefore, the additional vinyl substituent increases the C−H activation and 7230

DOI: 10.1021/jacs.7b00714 J. Am. Chem. Soc. 2017, 139, 7224−7243

Article

Journal of the American Chemical Society

Figure 6. DFT-computed Gibbs free energy changes (in toluene) of the competing steps that determine the chemoselectivity between the decarboxylative alkenylation and [2 + 2 + 2] annulation (via decarboxylation of 16). Gibbs free energies are compared to LRu(OAc)2 intermediate 8.

Figure 7. DFT-computed Gibbs free energy changes (in toluene) of the competing steps that determine the chemoselectivity between the monoalkenylation and double alkenylation. Gibbs free energies are compared to LRu(OAc)2 intermediate 8.

alkyne insertion barriers in the double alkenylation pathway due to the steric repulsions with the carboxylate group. 3. Origins of Regioselectivity with Asymmetric Alkynes. The experimental studies with asymmetric alkynes showed that the alkyl substituent of alkyne prefers to be proximal

to the forming C−C bond in the alkenylation product, while the aryl substituent prefers to be distal to the forming C−C bond (Scheme 1).8 Our computations on the catalytic cycle suggested that the alkyne insertion step is irreversible (Figure 1), thus the insertion transition state determines the regioselectivity of alkyne 7231

DOI: 10.1021/jacs.7b00714 J. Am. Chem. Soc. 2017, 139, 7224−7243

Article

Journal of the American Chemical Society

Figure 8. Optimized structures, relative Gibbs free energies and electronic energies, and distortion/interaction analysis of the competing insertion transition states with 1-phenylpropyne. Energies are in kcal/mol. L is the p-cymene ligand.

Scheme 4. Interactions between HOMO of Alkyne and pOrbital of Ruthenium-Bounded Aryl Carbon in the Insertion Transition States

Table 1. Relative Free Energies of the Insertion Transition States with Asymmetric Alkynes, and HOMO π Orbital Coefficients of the Two Alkyne Carbons

insertion. To directly compare with the experimental regioselectivity of the reaction using para-methoxybenzoic acid and 1phenylpropyne, we calculated the two competing insertion transition states with the identical substrates, and the results are shown in Figure 8. TS58, with the phenyl substituent distal to the forming C−C bond, eventually leads to the alkenylation product with a terminal phenyl substituent. TS59 leads to the alkenylation product with a terminal methyl substituent. TS58 is 2.4 kcal/mol more favorable than TS59, which is in good agreement with the experimental regioselectivity (16:1 favoring the distal attack).8 Because no significant steric repulsions are present in the two competing transition states, we hypothesized that the interactions between the five-membered ring ruthenacycle complex and 1-phenylpropyne determines the regioselectivity of insertion. The distortion/interaction analysis32,33 is applied to the insertion transition states for further insights, and results are shown in Figure 8. Each transition state structure was separated to two distorted fragments, the ruthenacycle complex and 1phenylpropyne. For each fragment, the energy difference between the distorted structure in the transition state and the ground-state structure is the distortion energy (ΔEdist‑alkyne for 1phenylpropyne, ΔEdist‑cpx for the ruthenacycle complex). The interaction energy (ΔEint) is the difference between the activation energy (ΔE‡) and the total distortion energy (ΔEdist‑alkyne + ΔEdist‑cpx). Consistent with our hypothesis, the interaction energy controls the regioselectivity. TS58 has a ΔEint of −64.7 kcal/mol, and ΔEint of TS59 is −62.5 kcal/mol. This leads to the 3.4 kcal/mol difference of the electronic energy

a

Gibbs free energies in kcal/mol.

barrier, while the distortion energies of the two fragments are similar in the two transition states. Previous theoretical studies on transition metal-mediated alkyne insertion found that the asymmetric alkyne prefers to use the alkyne carbon with the larger HOMO π orbital coefficient to form the C−C bond for the stronger orbital interaction.34 Therefore, this suggests that the 1-phenylpropyne, which favors 7232

DOI: 10.1021/jacs.7b00714 J. Am. Chem. Soc. 2017, 139, 7224−7243

Article

Journal of the American Chemical Society

Figure 9. DFT-computed Gibbs free energy changes (in toluene) from [Ru(p-cymene) (OAc)2] to the ruthenacyclopentadiene intermediate with paramethoxybenzoic acid substrate.

the distal attack transition state, should have a larger HOMO coefficient on the methyl-substituted carbon (Scheme 4). To verify this model, we calculated the HOMO coefficients for a number of asymmetric alkynes, and compared with the calculated regioselectivity (results are shown in Table 1). Indeed, 1-phenylpropyne (entry 1) has a larger HOMO orbital coefficient on the methyl substituted carbon (27.1% for C1 and 14.8% for C2). The substituents on phenyl group do not alter the HOMO distribution significantly (entry 2 and 3), and the insertion still favors the distal attack. Similar trend is observed for the amino substituent (entry 4). For electron neutral and withdrawing substituents, a switch of HOMO distribution occurs, and we predicted a reversed regioselectivity for the H and CHO substituents (entry 5 and 6). To test the limit of this model, we also studied the tBu-substituted alkyne (entry 7), and the regioselectivity does not agree with the above rationale based on HOMO distribution. In this case, C1 and C2 have similar HOMO distributions (41.6% and 42.0%), while the proximal transition state is 4.7 kcal/mol less stable than the distal one, due to the steric repulsions between the bulky tBu group and the phenyl group of the ruthenacycle complex. To further validate this theoretical model, we calculated a number of asymmetric alkynes with experimental results (entries 8−10). For the internal alkyne with cyclopropyl and phenyl terminals (entry 8), the frontier molecular orbital rationale also applies, and the calculated regioselectivity agrees well with Ackermans’ experimental results.35 The computed regioselectivities of the two methoxymethyl-substituted arylacetylenes (entries 9−10) are also in a good agreement with Zhao’s experimental results.8 Our calculations together suggest that the substituents with limited steric repulsions follow the theoretical model based on

frontier molecular orbital interactions, alkyne carbon with a larger HOMO orbital coefficient has a stronger interaction with the aryl carbon during insertion, resulting in a larger interaction energy and a lower insertion barrier. 4. Effects of Substrate. 4.1. Para-Substitution of Benzoic Acid. Experimental results showed that the electronic property of the para-substituents of benzoic acid has profound effects on the decarboxylative alkenylation reaction.8 With electron-donating substituents, such as OMe group, the decarboxylative alkenylation of substituted benzoic acids occurs smoothly. With electron-withdrawing substituents, such as NO2 group, the yields of desired alkenylation product decrease significantly (