A Computational Mechanistic Study of Amidation of Quinoline N-Oxide

Feb 29, 2016 - Yang-Yang Xing , Jian-Biao Liu , Ying-Ying Tian , Chuan-Zhi Sun , Fang Huang , and De-Zhan Chen. The Journal of Physical Chemistry A ...
1 downloads 0 Views 3MB Size
Research Article pubs.acs.org/acscatalysis

A Computational Mechanistic Study of Amidation of Quinoline N‑Oxide: The Relative Stability of Amido Insertion Intermediates Determines the Regioselectivity Jian-Biao Liu,* Xie-Huang Sheng, Chuan-Zhi Sun, Fang Huang, and De-Zhan Chen* College of Chemistry, Chemical Engineering and Materials Science, Collaborative Innovation Center of Functionalized Probes for Chemical Imaging in Universities of Shandong, Shandong Normal University, Jinan 250014, P. R. China S Supporting Information *

ABSTRACT: The origin of site selectivity of quinoline Noxide substrate in Ir(III)-catalyzed amidation with tosyl azide was investigated computationally. The reaction proceeds exclusively at the C8 position, instead of the C2 position, which has been reported previously in many other reactions. C2-Amidation is kinetically impossible under the reaction condition according to our calculations, with high apparent activation energy up to 51.1 kcal/mol. The high energetic span is caused by the deep-lying 5-membered amido insertion intermediate, in which a strong stabilization effect was observed due to nN → π*CN delocalization. For C8-amidation, however, the 6-membered counterpart is relatively unstable, making the activation energy only about half the value of C2-amidation. Meanwhile, denitrogenation is found to be turnoverlimiting in the reaction. The oxidation state changes of the Ir center during the stepwise C−N bond formation were investigated, and a considerably higher effective oxidation state was found in the Ir−nitrenoid intermediate. The ineffective RhIII catalyst was also studied. In comparison with the results of the IrIII catalyst, the RhIII catalyst features higher energy profiles and higher apparent activation energies. A dual role of acetic acid additive participating both in the C−H activation and protodemetalation was also demonstrated. KEYWORDS: quinoline N-oxide, site selectivity, C−H activation, metal effect, activation energy, oxidation state possess relatively strong acidity.7 Moreover, controlling the site selectivity on the carbocyclic ring (C5−C8 positions) remains a challenge,4 which is noteworthy in view of the fact that the corresponding functionalized quinolines have important utilities in various areas.8 An alternative approach is employing a directing group that brings metal catalyst to the desired C−H bond. Guided by this concept, the group of Chang recently reported Ir(III)-catalyzed C8-amidation of quinoline N-oxide with organic azide under mild conditions, using the N-oxide moiety as directing group (see Scheme 1).9 This reaction is highly regioselective, and no regioisomeric product P2 was formed. The N-oxide moiety is fascinating from a green chemistry perspective because the 8-amidated product P1 can be readily converted to 8-aminoquinoline by deoxygenation.10 A possible reaction pathway involving 5-membered metallacycle intermediate formed by C8−H activation was proposed by the authors for the C8-amidation reaction, on the basis of kinetic study, isolated intermediate, and previous literature. However, details of the reaction mechanism, especially the

1. INTRODUCTION Quinolines are among the most versatile structural motifs widely utilized in medicinal1 and materials chemistry.2 Thus, development of efficient synthetic procedures for their functionalization is of high importance.3 Although substitution reactions are useful for synthesizing functionalized quinolines, electrophilic substitution reactions occur preferentially at the carbocyclic ring rather than the heterocyclic ring.4 On the other hand, transition-metal-catalyzed C−H bond functionalization is an attractive strategy due to high functional group tolerance, and moreover, significant advances have been made.5 However, it is still difficult to control the position of activated C−H bond in substrates such as quinolines, which have more than one reactive site. Therefore, development of site-selective C−H functionalization of quinolines is highly desirable. Several methods have been reported in recent years with respect to the functionalization at the C2 position of quinolines or quinoline N-oxides such as alkylation, alkenylation, alkynylation, arylation, acetoxylation, and amidation.6 The high C2-selectivity is due to intrinsic reactivity of CN bond and coordination effect between N or NO and metal catalyst. In contrast, only a few examples have been reported for functionalization at C3 and C4 positions4 where the protons © XXXX American Chemical Society

Received: December 23, 2015 Revised: February 19, 2016

2452

DOI: 10.1021/acscatal.5b02938 ACS Catal. 2016, 6, 2452−2461

Research Article

ACS Catalysis Scheme 1. Selective C8-Amidation of Quinoline N-Oxide

original paper of SMD,15 solvation free energies were calculated at M05-2X/BS2 level in 1,2-DCE with SMD based on the M06/BS1-opimized geometries. BS2 represents a mixed basis set of ECP60MDF_AVDZ for Ir, ECP28MDF_AVDZ for Rh and 6-31G(d) for other atoms. To get more accurate electronic energies, single-point energy calculations were performed for all the M06/BS1-optimized structures at M06/BS3 level. BS3 represents a mixed basis set of ECP60MDF_AVTZ for Ir, ECP28MDF_AVTZ for Rh, and the 6-311++G(2d,2p) basis sets for other atoms. It should be emphasized that precise representation of experimental conditions such as solvent and additive effects is still a challenging problem for computational studies, especially for polarized systems. The difficulties of obtaining accurate free energies in solution were discussed in a combined experimental and theoretical study of Plata and Singleton on the Morita− Baylis−Hillman reaction17 and also in Harvey’s theoretical study of polar reactions in solution.18 The uncertainties in free energies can qualitatively affect the discussion. Several examples in the literature showed that the inclusion of the entire entropy contributions may lead to unreasonably high activation energies.19 A pragmatic approach is that entropies should be halved, to account for the loss of translational and rotational degrees of freedom in solution.20 However, such entropy correction is unsatisfactory theoretically, and according to the results of Plata and Singleton, the large errors in free energies were attributed to the errors of enthalpy, not the entropy. For the amidation reaction studied herein, the barrier height is found to be overestimated, given that the reaction proceeds at 50 °C for 12 h. Although the entropy correction mentioned above seems “arbitrary”, it was found that this approach improved agreement between experiment and theory for the reaction discussed here. Therefore, 50% of the originally computed translational and rotational entropy contributions were included to calculate the Gibbs free energy in solution. The standard state correction due to the difference between the gas-phase standard state (323.15 K and 1 atm) and the solution-phase standard state of 1 M equals RTln26.5, where R is the gas constant and T is the temperature. The energy of N2 was calculated without the solvation energy due to the positive value in 1,2-DCE. Based on previous reports, the counterion NTf2− has a negligible influence on the overall Gibbs free-energy surface.13 Therefore, the weakly coordinating counterion was not included in present studies. All calculations were performed using Gaussian 09 package.21 Natural bond order (NBO) calculations were performed by GenNBO 5.0 program22 using the wave function got from M06/BS1 level in 1,2-DCE with the SMD model on the selected systems. The 3D structures were prepared using CYLView.23

factors controlling the site-selective amidation have not been fully explored yet. According to the study of Shibata on C8 alkenylation of quinoline N-oxide using a cationic rhodium(I) catalyst, C−H bond cleavage occurs at both the C2 and C8 positions.11 However, C8−H cleavage is favored due to formation of more stable 5-membered rhodacycle, while C2− H cleavage leads to an unstable 4-membered intermediate. Larionov recently reported a combined experimental and computational study on the regioselective C8-arylation of quinoline N-oxide by palladium(II) catalyst.12 The observed regioselectivity was rationalized on the basis of the relative stability of the 5- and 4-membered palladacycles formed by C8−H and C2−H bond activation, respectively. For the reaction shown in Scheme 1, it is of interest to perform a comparative investigation on amidation of quinoline N-oxide at both the experimentally observed C8 position and the unobserved C2 position, which would be helpful to understand the directing effect of N-oxide and the reactivity differences between different C−H bonds in quinoline N-oxide. Additionally, Cp*-based RhIII counterpart was found to be ineffective for the amidation shown in Scheme 1. According to the analogous metal-catalyzed C−H amination of benzamides with organic azides, both the Cp*-based IrIII and RhIII systems have catalytic activities, albeit with different reaction rate and efficiency.13 The difference in catalytic activity between the two systems was attributed to the intrinsically strong relativistic effects of iridium. Thus, considering the notable dependency of amidation reaction on the metal center, the Rh(III)-catalyzed amidation should be also investigated. Herein we present computational studies on the selective amidation of quinoline N-oxide, focusing on the origin of the regioselectivity and the different catalytic activities of IrIII and RhIII systems.

2. COMPUTATIONAL METHODS Geometry optimizations and frequency calculations were performed at the M06 level14 of density functional theory (DFT) in 1,2-dichloroethane (1,2-DCE) using the SMD solvation model15 with default convergence criteria. The M06 method has been used to study Cp*-based group 9 metalcatalyzed amination and yields reasonable results that are in good agreement with experimental observations.13 The relativistic effects were approximated by a Hamiltonian with spin-averaged scalar-relativistic (SR) effective core potentials (ECP) for Ir and Rh atoms from the Stuttgart/Cologne group (ECP60MDF for Ir and ECP28MDF for Rh).16 The contracted Gaussian ECP-adapted basis sets ECP60MDF_AVDZ for Ir, ECP28MDF_AVDZ for Rh and 6-31++G(d,p) basis sets for other atoms were used (the mixed basis set is named as BS1). The results of frequency calculations were examined to confirm each structure is a local minimum (no imaginary frequency) or a transition state (only one imaginary frequency). Thermodynamic corrections at 323.15 K and 1 atm for all structures in 1,2-DCE were obtained by harmonic frequency calculations at the M06/BS1 level. Considering the good performance of M05-2X/6-31G(d) among the various methods tested in the

3. RESULTS AND DISCUSSION The reaction mechanism was extended to incorporate both the C8-amidation and C2-amidation of quinoline N-oxide, as shown in Scheme 2, on the basis of Chang’s reports24 and our 2453

DOI: 10.1021/acscatal.5b02938 ACS Catal. 2016, 6, 2452−2461

Research Article

ACS Catalysis Scheme 2. Proposed Catalytic Cycle of the C8- and C2-Amidation of Quinoline N-Oxide

calculations. Amidation at C8 or C2 position proceeds via similar catalytic cycle that comprises the following steps: (1) formation of active catalyst and subsequent acetate-assisted cyclometalation with quinoline N-oxide leading to the formation of iridacycle, (2) coordination of azide and subsequent denitrogenation to generate metal−nitrenoid species, (3) migratory insertion of nitrenoid moiety into Ir− CQ bond, and lastly, (4) protodemetalation assisted by acetic acid resulting in formation of final product and regeneration of catalyst. The details of these processes will be discussed in the following sections. As shown in Scheme 2, the mechanism reveals the dual role of participating acetic acid additive, that is, in generation of iridacycle and in protodemetalation at the final stage of catalytic cycle to release the amidated product, which was validated recently by experimental mechanistic studies.25 3.1. Structure of Catalytically Active Species and Formation of Metallacycle. Under reaction conditions, when 4 equiv of silver salt (AgNTf2) was added to solution of dimeric iridium precursor [Cp*IrCl2]2 in the presence of acetic acid additive (15 equiv), dicationic [Cp*Ir]2+ catalyst was assumed to be generated in situ at first. However, there are still vacant coordination sites remaining at the metal center, allowing further occupations of various ligands in the solution. The corresponding reaction can be written as [Cp*Ir]2 + + An− + Bm − ⇌ [Cp*Ir(A)(B)]2 − n − m

Table 1. Calculated Gibbs Free Energy Changes (in kcal/ mol) for eq 1

m−

An−

Bm−

ΔG

1 2 3 4 5 6 7 8 9 10

κ2-OAc− Q HOAc NTf2− κ2-OAc− κ2-OAc− κ2-OAc− κ2-OAc− Q Q

/ / / / κ1-OAc− Q HOAc NTf2− Q NTf2−

−34.5 −0.7 8.3 −6.4 −70.2 −57.1 −47.8 −46.3 −31.8 −24.4

[Cp*Ir(κ2-OAc)(κ1-OAc)] compared with that of [Cp*Ir(κ2OAc)(Q)]+. However, there is a relatively small driving force for additional binding of the other two ligands (HOAc and NTf2−). Finally, on the basis of experimental results25 and our calculated free energies of formation, the [Cp*Ir(κ2-OAc)]+ is chosen as the catalytically active species and also the starting point of the whole catalytic cycle.27 Coordination of Q to the active species 1 by O atom forms complex 2. The subsequent C8−H as well as C2−H bond cleavages proceed through the inner-sphere acetate-assisted concerted metalation−deprotonation (CMD) mechanism,28 generating, respectively, the 5-membered complexes 3 and 4membered 3′. The free-energy profile for the cyclometalation reaction is shown in Figure 1, together with structures of CMD transition states. For comparison, we also performed computational studies on the amidation reaction catalyzed by [Cp*Rh(κ2-OAc)]+. Their results are discussed and compared with that of [Cp*Ir(κ2-OAc)]+ catalyst. As shown in Figure 1, compared with the results of C2−H cleavage, the C8−H cleavage aided by HOAc is both thermodynamically and kinetically more feasible, with an attainable activation free energy (23.5 kcal/mol; TS23

(1)



where A and B represent competing OAc , HOAc, NTf2−, and Q ligands. It should be noted that the counterion NTf2− can also be connected to the unsaturated metal center by Ncoordination.26 The calculated Gibbs free-energy changes for eq 1 are listed in Table 1. Among the four ligands, OAc− is the energetically preferred one throughout. Additional monocoordinations of OAc− and Q to [Cp*Ir(κ2-OAc)]+ result in stable [Cp*Ir(κ2-OAc)(κ1-OAc)] and [Cp*Ir(κ2-OAc)(Q)]+ complexes, respectively. A more exergonic value is observed for n−

entry

2454

DOI: 10.1021/acscatal.5b02938 ACS Catal. 2016, 6, 2452−2461

Research Article

ACS Catalysis Scheme 3. Resonance Structures of TsN3

analysis of the reaction by Nγ coordination was performed.13,31 As shown in Figure 2 (right structure), rotation of the weak M−Nα bond in complexes 4 and 4′ may result in different conformers. We have optimized all the possible configurations, and their relative free energies are compared.32 As expected, the calculated results reveal that the free-energy differences are small (within 3.3 kcal/mol). Meanwhile, denitrogenations of these conformers occur with similar energy barriers,33 so only the pathways leading to the intermediates that facilitate the following nitrenoid insertion processes are considered. As shown in Figure 3, the process of formation of complex 4 (4′) is exergonic. From the resulting complex, the reaction

Figure 1. Free-energy profiles for the acetate-assisted cyclometalation. Red and blue profiles represent Ir(III)-catalyzed reactions at C8 and C2 positons, respectively. Values in parentheses refer to the results of RhIII catalyst, and bond lengths are given in Å. Energies are relative to [Cp*M(κ2-OAc)]+ + Q + TsN3, and are mass balanced (M = Ir, Rh; similarly hereinafter).

− 2) for IrIII catalyst. The larger driving force may be attributed to the intrinsic stability of 5-membered isomer 3, which is 12.1 kcal/mol more stable than the 4-membered isomer 3′. The results of RhIII catalyst are analogous to that of IrIII catalyst. 3.2. Denitrogenation of the Intermediate Formed by Metallacycle with Azide. Organic azides are among the most promising and environmentally friendly aminating reagents,29 because the only byproduct is nitrogen. Due to their important applications in amination, both the geometric and electronic structures of organic azides have been deeply investigated.30 According to the experimental and computational study of FSO2N3 and CF3SO2N3 by Willner, the two kinds of sulfonyl compound are found to prefer synperiplanar configurations between SO and N3 due to the predominant anomeric interaction of nσ(N) → σ*(S−O).30b The molecular structure of tosyl azide (TsN3) was optimized and shown in Figure 2. Different from the results of sulfonyl compounds described above, TsN3 prefers a staggered conformation that is sterically favored (see the left structure in Figure 2). Similar to other organic azides, TsN3 has two predominant resonance structures as shown in Scheme 3. The coordinatively unsaturated metallacycles 3 as well as 3′ provide enough space for further coordination of TsN3 by either Nα or Nγ atom to form complexes 4 and 4′. Previous theoretical studies have exclusively addressed the Nα coordination, so no further

Figure 3. Free-energy profiles for the denitrogenation process. Red and blue profiles represent Ir(III)-catalyzed reactions at C8 and C2 positons, respectively. Values in parentheses refer to the results of RhIII catalyst.

proceeds via Nα−Nβ cleavage to generate nitrenoid, which occurs with an energy barrier of 20.7 (21.6) kcal/mol for reaction at C8 (C2) position (relative to 4 and 4′) for IrIII catalyst. The optimized structures of 4 and 4′, transition states TS45 and TS45′, and intermediates 5 and 5′ are shown in Figure 4. Compared with Nα−Nβ bond length in the separated TsN3, there is only a slight bond elongation (ca. 0.01 Å) between Nα and Nβ in the bound azide 4 (4′). However, Nβ− Nγ bond lengths in the unbound and bound azides are almost unchanged and relatively short, suggesting the predominance of the left resonance structure in Scheme 3. In both cases shown in Figure 4, the transition states are approximately halfway through N2 elimination, as judged by comparison of IrN bond lengths in the complexes, indicating a strengthening of IrN bond. In TS45 (TS45′), Nβ−Nγ bond distance is also relatively shorter, whereas the Nα−Nβ bond is longer than Nβ− Nγ by about 0.5 Å, and N3 units have lost the linearity. To investigate the changes of bonding from 4 to 5, the orbital evolution during N2 loss was analyzed and shown in Scheme 4. Because all the stationary points in step 2 of C2amidation pathway lie above C8-amidation pathway, only the latter is considered herein. In 4, the interaction between metal

Figure 2. Optimized structure of TsN3 (left) and the possible conformation of complex 4 formed by metallacycle with TsN3. 2455

DOI: 10.1021/acscatal.5b02938 ACS Catal. 2016, 6, 2452−2461

Research Article

ACS Catalysis

Figure 4. Optimized structures of intermediates and transition states involved in the denitrogenation process, with selected bond distances (in Å) and bond angles. Values in parentheses refer to the results of RhIII catalyst.

(TS45 − 4), and the process is exergonic by 30.4 kcal/mol (4 − 5). A comparison of the values for RhIII catalyst with 24.8 kcal/mol barrier and 20.0 kcal/mol exergonicity suggests IrIII catalyst makes this step thermodynamically more favorable by 10.4 kcal/mol, which can be attributed to the stronger IrN bond in iridium−nitrenoid compared with RhN bond in rhodium−nitrenoid. The bonding difference is effectively revealed by the calculated MN Wiberg bond index (WBI; 1.19 for Ir and 1.06 for Rh). 3.3. Amido Insertion Leading to C−N Bond Formation. The metal nitrenoids 5 and 5′ formed in step 2 subsequently undergo amido insertion reactions via the threecentered transition states TS56 and TS56′, respectively (Figure 5). The nitrenoid insertions into Ir−C8 and Ir−C2 bonds are exergonic by 37.3 and 67.0 kcal/mol and proceed with small energy barriers of 7.1 and 7.6 kcal/mol, respectively (Figure 6). The strong exergonic character of this step is due to the generation of much stronger C−N bond in 6 (6′) than IrN bond in 5 (5′).35 As shown in Figure 6, the nitrenoid insertion product 6′ is lower in energy by up to 23.6 kcal/mol than 6 (M = Ir). To obtain further insight into the energy difference between 6 and 6′, we performed NBO analysis,36 in which the stabilization energy E(2) for each donor NBO(i) and acceptor NBO(j) associated with delocalization i → j defined as

Scheme 4. Orbital Changes during the Generation of Metal Nitrenoid Species

and nitrogen is weak and has no covalent character, which results in little bond length changes in the bound azide as mentioned above. In TS45, the dπ-backdonation takes place from dxy into N-px vacant orbital. The formed IrN bond has double bond character with donation from N-py to Ir-dx2‑y2 and backdonation from Ir-dxy to N-px orbital.34 The corresponding important orbital interactions in complex 5 are illustrated in Scheme 5. As indicated in Figure 3, for IrIII catalyst, formation of nitrenoid species 5 occurs with a barrier of 20.7 kcal/mol

⎡ F 2 ⎤ (i , j) ⎥ E(2) = ΔEij(2) = qi⎢ ⎢⎣ εj − εi ⎥⎦

Scheme 5. Schematic Representation of Important Orbital Interactions in 5

(2)

where q is the donor orbital occupancy, ε and F are, respectively, diagonal and off-diagonal elements of NBO Fock matrix. A larger E(2) value means a greater donating tendency from the donor to acceptor orbital and thus a greater extent of conjugation. In the case of 6′ (M = Ir), besides the vicinal interactions within the quinoline N-oxide framework, a stronger interaction between the lone pair of N in substituent group −NTs and the antibonding of vicinal CN in the ring was observed. As shown in Figure 7, this type of nN → π*CN delocalization results in a stabilization energy of 73.3 kcal/mol. 2456

DOI: 10.1021/acscatal.5b02938 ACS Catal. 2016, 6, 2452−2461

Research Article

ACS Catalysis

Figure 7. Important NBO donor−acceptor interactions in 6 and 6′ (M = Ir) and the corresponding NBOs (isovalue = 0.05).

the other hand, the calculated natural population analysis (NPA) charges suggested the concerted process was redoxneutral, while the oxidation state of the metal center during the stepwise pathway changed during the course of C−N bond formation. It has been demonstrated that the high-valent nitrenoid intermediates are critical for C−N bond formation,39 whereas experimental evidence of high-valent intermediates in the amidation of quinoline N-oxide is still lacking. Computationally, the changes of metal valency can be revealed by NPA results. The calculated population changes of intermediates involved in the stepwise pathway of IrIII catalyst are listed in Table 2.

Figure 5. Optimized structures of TS56, TS56′, 6, and 6′ with selected bond lengths (in Å). Values in parentheses refer to the results of RhIII catalyst.

Table 2. Population Changes of the Intermediates Involved in C−N Bond Formationa 4 Ir NαTs Q Cp* NβNγ a

Figure 6. Free-energy profiles for the amido insertion. Red and blue profiles represent Ir(III)-catalyzed reactions at C8 and C2 positons, respectively. Values in parentheses refer to the results of RhIII catalyst.

0 0 0 0 0

(76.18) (88.18) (75.22) (74.73) (13.69)

TS45

5

TS56

6

−0.16 0.28 −0.25 −0.06 0.19

−0.33 0.50 −0.28 −0.21 0.31

−0.18 0.43 −0.38 −0.19 0.31

−0.10 0.53 −0.53 −0.21 0.31

A positive value represents an increase in the population (vice versa).

From 4 to 5, the electron population of NβNγ increases by 0.31e. Electron populations of Q and Cp* decrease by 0.28e and 0.21e, respectively, to compensate the decrease in Ir atomic population during departure of NβNγ. Consistent with the population changes, Ir atomic population decreases by 0.33e, and the population of NTs increases by 0.50e. From 5 to 6, further charge transfers from Q are observed, and the populations of both Ir and NTs increase. The population changes of Ir center reveal the changes of oxidation state during C−N bond formation, which can be further explained by the changes of electron configuration, as shown in Scheme 6. In 4, the valence d6 orbitals are split as (dxy)2(dxz, dyz)4(dx2‑y2)0(dz2)0. In 5, backdonation from Ir-dxy to N-px orbital occurs. Meanwhile, the Ir-dxy orbital becomes unoccupied, and the Ir center takes d4 electron configuration. The N2 departure from 4 implies an oxidation addition process with concomitant formation of IrV metal center and NTs2− dianion. The resulting

The strong stabilization effect makes N in −NTs coplanar with quinoline N-oxide ring. While in 6 (M = Ir), the energy of the accepting orbital π*CC is much higher than that of π*CN in 6′, making the stabilization due to nN → π*CC is only about one-sixth of the value in 6′. The calculated WBIs of C2−NTs in 6′ and C8−NTs in 6 are 1.21 and 1.09, respectively. The stronger C2−NTs is also well reflected on the bond lengths (1.37 Å for C2−NTs and 1.41 Å for C8−NTs; see Figure 5). 3.4. Oxidation State Changes of Ir Center during the C−N Bond Formation. In Chang’s mechanistic studies on rhodium-catalyzed C−H amination of arenes, they also checked the possibility of a concerted pathway for the formation of nitrenoid insertion product, starting directly from the azide coordinated metallacycle.37 The plausible concerted pathway is found to be unfavorable due to the higher energy barrier.38 On 2457

DOI: 10.1021/acscatal.5b02938 ACS Catal. 2016, 6, 2452−2461

Research Article

ACS Catalysis Scheme 6. Electron Configuration Changes during the C−N Bond Formation

5 is thermodynamically unstable and subsequently undergoes the reductive amido insertion. In 6, the electron configuration of Ir changes back to d6. 3.5. Protodemetalation Assisted by Acetic Acid To Release the Product and Regenerate the Catalyst. From intermediates 6 and 6′, the final protodemetalation in the catalytic cycle proceeds via coordination of HOAc to the metal center. HOAc may coordinate with the metal amido species from two opposite directions, resulting in two different pathways that were found to have similar free-energy profiles. For clarity, we only discuss pathway b, and the results of pathway a are listed in Figure S2 in the Supporting Information. The HOAc-coordinated complexes 7 and 7′ then undergo CMD processes that involve hydrogen transfer from HOAc to the amido nitrogen. Release of aminated products from 8 and 8′ occur as the final step, regenerating the catalytically active species 1. Free-energy profiles for the protodemetalation are shown in Figure 8. For IrIII catalyst, protodemetalation of 6

Figure 9. Optimized structures of TS78b, TS78b′, P1b, and P2b with selected bond lengths (in Å). Values in parentheses refer to the results of RhIII catalyst.

Figure 10. Free-energy profiles of the full catalytic cycle for IrIII and RhIII catalysts.

shows the free-energy profiles of the full catalytic cycle for C8 and C2-amination of quinoline N-oxide with both IrIII and RhIII catalysts. According to the energetic span model introduced by Kozuch and Shaik, the energetic span (δE) that serves as the apparent activation energy of catalytic cycle is defined by the energy difference between turnover-frequency(TOF)-determining transition state (TDTS) and the TOF-determining intermediate (TDI), and the reaction driving force.41 For IrIII catalyst, TS45 and TS45′ are the TDTSs for functionalization at C8 and C2 positions, respectively. The TDIs for C8 and C2-amination are Q-coordinated complex 2 and HOAc-coordinated complex 7b′ respectively. The apparent activation energy for C8-amination is 26.6 kcal/mol (TS45 − 2) and the value is 51.1 kcal/mol (TS45′ − 7b′ + ΔGr; ΔGr = −71.2 kcal/mol) for C2-amination. By comparing the energetic spans of the two free-energy surfaces (ΔΔG‡ = 24.5 kcal/mol), the C8 selectivity is in good agreement with experiment, in which no isomeric product P2b was formed and the yield of P1b is up to 92%. Based on the apparent activation energy, the rate constant k, which is actually a simple expression of the TOF, can be calculated by

Figure 8. Free-energy profiles for the protodemetalation of pathway b. Red and blue profiles represent Ir(III)-catalyzed reactions at C8 and C2 positons, respectively. Values in parentheses refer to the results of RhIII catalyst.

readily occurs with a lower energy barrier of 4.5 kcal/mol, while the barrier for 6′ is higher (18.1 kcal/mol). The calculated energy barriers agree well with the experiment, in which only product P1b was observed.40 The optimized structures of transition states and products are shown in Figure 9. 3.6. Free-Energy Profiles of the Full Catalytic Cycle and Comparison with Experimental Results. Figure 10

kBT −δE / RT e (3) h The calculated rate constant of Ir(III)-catalyzed C8amination is 4.1 × 10−4 min−1. Because the rate constant is k=

2458

DOI: 10.1021/acscatal.5b02938 ACS Catal. 2016, 6, 2452−2461

Research Article

ACS Catalysis not available in Chang’s paper,9 the value can only be estimated from the reported reaction profile and is about 6.4 × 10−4 min−1, assuming the mechanism is the same with that of a similar amination reaction.37 Unexpectedly, our calculated rate constant agrees well with the derived value. In addition, the kinetic isotope effects (KIE) were investigated by Chang, and only small values were measured from the Ir(III)-catalyzed reactions involving Q and Q-d7.9 The experimental observation is fully consistent with the red profile illustrated in Figure 10 that indicates no significant KIE would be measured since the “turnover-limiting step”42 does not involve the C−H bond cleavage. It is also of interest to calculate the KIE based on the theoretical results, which will provide further verification of the proposed reaction mechanism. For the sake of simplicity, the kinetic scheme can be represented as KIE1

KIE 2

rate−1

rate2

2 HooooI 3 ⎯⎯⎯⎯→ 5

C8-amidation, the analogous transition state TS45 remains TDTS, while intermediate 2 becomes TDI. The apparent activation energy is calculated to be 26.6 kcal/mol. Our calculated results provide a reasonable explanation for the experiments in which a high yield of P1b was obtained, while no isomeric product P2b was formed. Furthermore, denitrogenation step is the turnover-limiting step instead of the experimentally proposed protodemetalation step. For comparison, the ineffective RhIII catalyst was also investigated to explore the metal effects on the reaction. All the stationary points in the free enegy profiles of RhIII catalyst lie above that of IrIII catalyst and higher apparent activation energies are found for both the C8 and C2-amination, explaining the minimal catalytic activity observed experimentally. Additionally, oxidation state changes of Ir center during the stepwise C−N bond formation were investigated. In 4, the interaction between TsN3 and metal catalyst is weak, and the electron configuration is d6 (IrIII). During the departure of N2, oxidation addition occurs, and backdonation from Ir-dxy to N-px orbital takes place in 5. Hence, a considerably higher effective oxidation state should be assigned to the metal center, being closer to IrV than to the formal oxidation number IrIII. After reductive amido insertion, the electron configuration of Ir changes back to d6. Dual role of acetic acid additive was also demonstrated computationally: generating iridacycle in the C− H activation step via an inner-sphere HOAc-assisted CMD pathway, and as a proton source to release the amidated product in the final stage of the reaction. Overall, our calculated results indicate the relative stability of amido insertion intermediates greatly affects the TOF of catalytic cycle. Not surprisingly, compared with the results of C2-amidation, C8-amidation pathway is both kinetically and thermodynamically more favorable in the first half of the catalytic cycle, due to the stability of 5-membered metallacycle formed from C−H bond activation. However, the dominant factor controlling the site selectivity is the relative stability of later amido insertion intermediates, resulting in different activation energies. Finally, our results suggest that the relative stability of intermediates is critical for predicting and understanding the site selectivity of metal-catalyzed C−H functionalization of quinolines and their derivatives.

(4)

since intermediate 3 undergoes irreversible step of the reaction mechanism. Then the experimentally observed kinetic isotope effect will be determined by KIEobs =

KIE 2 + KIE1·Cf 1 + Cf

where Cf = rate2/rate−1. calculated according to

17,43

(5)

The kinetic isotope effect is



kH/kD = eΔΔG / kBT = e(ΔZPEI −ΔZPETS)/ kBT

(6)

where ZPE represents zero-point vibrational energy and ΔΔG‡ equals the difference between the ΔZPE (= ZPEH − ZPED) in the intermediate I and transition state TS. The calculated KIE1 and KIE2 are 4.54 and 1.05, respectively, and the final KIEobs is 1.08, in accord with the measured kinetic isotope effects (kH/kD = 1.1−1.2). Compared with the results of IrIII catalyst, the RhIII catalyst features higher energy profiles for both the C8 and C2amination. The TDTS and TDI of the two catalysts are identical. The final apparent activation energies for Rh(III)catalyzed C8 and C2-amination are 35.5 and 55.3 kcal/mol, respectively, again excluding the possibility of amidation at C2 position. The calculated activation energy of C8-amination is consistent with the experiment that reveals the common used RhIII catalyst is ineffective for amidation and the yield of P1b is less than 1%.



ASSOCIATED CONTENT

S Supporting Information *

4. CONCLUSIONS The mechanism of Ir(III)-catalyzed amidation of quinoline Noxide with tosyl azide was studied by DFT calculations. The free-energy surfaces of amidation at both the C8 and C2 positions were investigated and compared. The two reactions proceed via similar catalytic cycle comprising four main steps: (i) C−H activation at either C8 or C2, (ii) denitrogenation, (iii) amido insertion, and (iv) protodemetalation. One remarkable difference between the two free-energy surfaces is the energy difference between amido insertion intermediates 6 and 6′. A stronger delocalization effect arising from nN → π*CN was found in 6′, while the effect decreases significantly in 6 because of the higher energy of the accepting π*CC. For C2-amidation, TS45′ and 7b′ are the TDTS and TDI, respectively. The calculated apparent activation energy is up to 51.1 kcal/mol, indicating the kinetic impossibility of amidation at C2 position under the reaction condition. For

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscatal.5b02938. Geometries and relative energies of different configurations of complexes 4 and 4′; free-energy profiles for the protodemetalation of pathway a; and relative Gibbs free energies, relative enthalpies, and Cartesian coordinates of optimized structures (PDF)



AUTHOR INFORMATION

Corresponding Authors

*E-mail for J.-B. L.: [email protected]. *E-mail for D.-Z. C.: [email protected]. Notes

The authors declare no competing financial interest. 2459

DOI: 10.1021/acscatal.5b02938 ACS Catal. 2016, 6, 2452−2461

Research Article

ACS Catalysis



(f) Makida, Y.; Ohmiya, H.; Sawamura, M. Chem. - Asian J. 2011, 6, 410−414. (g) Hitomi, Y.; Hiramatsu, K.; Arakawa, K.; Takeyasu, T.; Hata, M.; Kodera, M. Dalton Trans. 2013, 42, 12878−12882. (9) Hwang, H.; Kim, J.; Jeong, J.; Chang, S. J. Am. Chem. Soc. 2014, 136, 10770−10776. (10) Jeong, J.; Lee, D.; Chang, S. Chem. Commun. 2015, 51, 7035− 7038. (11) Shibata, T.; Matsuo, Y. Adv. Synth. Catal. 2014, 356, 1516− 1520. (12) Stephens, D. E.; Lakey-Beitia, J.; Atesin, A. C.; Ateşin, T. A.; Chavez, G.; Arman, H. D.; Larionov, O. V. ACS Catal. 2015, 5, 167− 175. (13) Figg, T. M.; Park, S.; Park, J.; Chang, S.; Musaev, D. G. Organometallics 2014, 33, 4076−4085. (14) (a) Zhao, Y.; Truhlar, D. G. J. Chem. Phys. 2006, 125, 194101. (b) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215−241. (15) For the introduction of solvation model based on density (SMD) and the performance of tested methods, see: Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B 2009, 113, 6378−6396. (16) (a) Peterson, K. A.; Figgen, D.; Dolg, M.; Stoll, H. J. Chem. Phys. 2007, 126, 124101. (b) Figgen, D.; Peterson, K. A.; Dolg, M.; Stoll, H. J. Chem. Phys. 2009, 130, 164108. (17) Plata, R. E.; Singleton, D. A. J. Am. Chem. Soc. 2015, 137, 3811− 3826. (18) Harvey, J. N. Faraday Discuss. 2010, 145, 487−505. (19) For recent examples, see: (a) Parmelee, S. R.; Mazzacano, T. J.; Zhu, Y.; Mankad, N. P.; Keith, J. A. ACS Catal. 2015, 5, 3689−3699. (b) Ertem, M. Z.; Himeda, Y.; Fujita, E.; Muckerman, J. T. ACS Catal. 2016, 6, 600−609. (20) (a) Wang, T.; Liang, Y.; Yu, Z.-X. J. Am. Chem. Soc. 2011, 133, 13762−13763. (b) Kua, J.; Krizner, H. E.; De Haan, D. O. J. Phys. Chem. A 2011, 115, 1667−1675. (c) Andrada, D. M.; Granados, A. M.; Solà, M.; Fernández, I. Organometallics 2011, 30, 466−476. (d) Li, H.; Lu, G.; Jiang, J.; Huang, F.; Wang, Z.-X. Organometallics 2011, 30, 2349−2363. (e) Huang, F.; Lu, G.; Zhao, L.; Li, H.; Wang, Z.-X. J. Am. Chem. Soc. 2010, 132, 12388−12396. (f) Andrada, D. M.; JimenezHalla, J. O. C.; Solà, M. J. Org. Chem. 2010, 75, 5821−5836. (g) Liang, Y.; Liu, S.; Xia, Y.; Li, Y.; Yu, Z.-X. Chem. - Eur. J. 2008, 14, 4361− 4373. (h) Yu, Z.-X.; Houk, K. N. J. Am. Chem. Soc. 2003, 125, 13825− 13830. (i) Strajbl, M.; Sham, Y. Y.; Villà, J.; Chu, Z. T.; Warshel, A. J. Phys. Chem. B 2000, 104, 4578−4584. (j) Hermans, J.; Wang, L. J. Am. Chem. Soc. 1997, 119, 2707−2714. (k) Abraham, M. H. J. Am. Chem. Soc. 1981, 103, 6742−6744. (l) Wertz, D. H. J. Am. Chem. Soc. 1980, 102, 5316−5322. (21) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. J. A.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09, Revision D.01; Gaussian, Inc.: Wallingford, CT, 2009. (22) Glendening, E. D.; Badenhoop, J. K.; Reed, A. E.; Carpenter, J. E.; Bohmann, J. A.; Morales, C. M.; Weinhold, F. NBO 5.0; Theoretical Chemistry Institute, University of Wisconsin, Madison: Madison, WI, 2001. (23) Legault, C. Y. CYLview, 1.0b; Université de Sherbrooke: Canada, 2009; http://www.cylview.org. (24) For a recent review of transition-metal-catalyzed C−N bond forming reactions using organic azides as the nitrogen source, see: Shin, K.; Kim, H.; Chang, S. Acc. Chem. Res. 2015, 48, 1040−1052.

ACKNOWLEDGMENTS This work was financially supported by the National Natural Science Foundation of China (NSFC No. 21375082) of China. J.-B. L. thanks Y.-H. Qiu for helpful discussion.



REFERENCES

(1) (a) Michael, J. P. Nat. Prod. Rep. 2008, 25, 166−187. (b) Eicher, T.; Hauptmann, S.; Speicher, A. The Chemistry of Heterocycles: Structures, Reactions, Synthesis, and Applications, 3rd ed.; Wiley-VCH: Weinheim, Germany, 2013; pp 386−405. (2) (a) Hughes, G.; Bryce, M. R. J. Mater. Chem. 2005, 15, 94−107. (b) Kimyonok, A.; Wang, X. Y.; Weck, M. J. Macromol. Sci., Polym. Rev. 2006, 46, 47−77. (3) (a) Mongin, F.; Quéguiner, G. Tetrahedron 2001, 57, 4059− 4090. (b) Mphahlele, M. J.; Lesenyeho, L. G. J. Heterocycl. Chem. 2013, 50, 1−16. (4) For a recent review of transition-metal-catalyzed site-selective C− H functionalization of quinolines beyond C2 selectivity, see: Iwai, T.; Sawamura, M. ACS Catal. 2015, 5, 5031−5040. (5) Selected reviews on C−H activations: (a) Kakiuchi, F.; Chatani, N. Adv. Synth. Catal. 2003, 345, 1077−1101. (b) Fagnou, K.; Lautens, M. Chem. Rev. 2003, 103, 169−196. (c) Seregin, I. V.; Gevorgyan, V. Chem. Soc. Rev. 2007, 36, 1173−1193. (d) Daugulis, O.; Do, H.; Shabashov, D. Acc. Chem. Res. 2009, 42, 1074−1086. (e) Lyons, T. W.; Sanford, M. S. Chem. Rev. 2010, 110, 1147−1169. (f) Cho, S. H.; Kim, J. Y.; Kwak, J.; Chang, S. Chem. Soc. Rev. 2011, 40, 5068−5083. (g) McMurray, L.; O’Hara, F.; Gaunt, M. J. Chem. Soc. Rev. 2011, 40, 1885−1898. (h) Yeung, C. S.; Dong, V. M. Chem. Rev. 2011, 111, 1215−1292. (i) Ackermann, L. Chem. Rev. 2011, 111, 1315−1345. (j) Li, B.-J.; Shi, Z.-J. Chem. Soc. Rev. 2012, 41, 5588−5598. (k) Colby, D. A.; Tsai, A. S.; Bergman, R. G.; Ellman, J. A. Acc. Chem. Res. 2012, 45, 814−825. (l) Song, G.-Y.; Wang, F.; Li, X.-W. Chem. Soc. Rev. 2012, 41, 3651−3678. (m) Kuhl, N.; Hopkinson, M. N.; WencelDelord, J.; Glorius, F. Angew. Chem., Int. Ed. 2012, 51, 10236−10254. (n) Engle, K. M.; Mei, T. S.; Wasa, M.; Yu, J.-Q. Acc. Chem. Res. 2012, 45, 788−802. (o) Arockiam, P. B.; Bruneau, C.; Dixneuf, P. H. Chem. Rev. 2012, 112, 5879−5918. (p) Yamaguchi, J.; Yamaguchi, A. D.; Itami, K. Angew. Chem., Int. Ed. 2012, 51, 8960−9009. (6) For reactions of quinolines, see: (a) Berman, A. M.; Bergman, R. G.; Ellman, J. A. J. Org. Chem. 2010, 75, 7863−7868. (b) Wen, P.; Li, Y.; Zhou, K.; Ma, C.; Lan, X.; Ma, C.; Huang, G. Adv. Synth. Catal. 2012, 354, 2135−2140. (c) Ren, X.; Wen, P.; Shi, X.; Wang, Y.; Li, J.; Yang, S.; Yan, H.; Huang, G. Org. Lett. 2013, 15, 5194−5197. For reactions of quinoline N-oxides, see: (d) Kanyiva, K. S.; Nakao, Y.; Hiyama, T. Angew. Chem., Int. Ed. 2007, 46, 8872−8874. (e) Cho, S. H.; Hwang, S. J.; Chang, S. J. Am. Chem. Soc. 2008, 130, 9254−9256. (f) Wu, J.; Cui, X.; Chen, L.; Jiang, G.; Wu, Y. J. Am. Chem. Soc. 2009, 131, 13888−13889. (g) Campeau, L.; Stuart, D. R.; Leclerc, J.; Bertrand-Laperle, M.; Villemure, E.; Sun, H.; Lasserre, S.; Guimond, N.; Lecavallier, M.; Fagnou, K. J. Am. Chem. Soc. 2009, 131, 3291− 3306. (h) Araki, Y.; Kobayashi, K.; Yonemoto, M.; Kondo, Y. Org. Biomol. Chem. 2011, 9, 78−80. (i) Ryu, J.; Cho, S. H.; Chang, S. Angew. Chem., Int. Ed. 2012, 51, 3677−3681. (j) Wu, Z.; Pi, C.; Cui, X.; Bai, J.; Wu, Y. Adv. Synth. Catal. 2013, 355, 1971−1976. (k) Chen, X.; Zhu, C.; Cui, X.; Wu, Y. Chem. Commun. 2013, 49, 6900−6902. (l) Li, G.; Jia, C.; Sun, K. Org. Lett. 2013, 15, 5198−5201. (7) Shen, K.; Fu, Y.; Li, J.-N.; Liu, L.; Guo, Q.-X. Tetrahedron 2007, 63, 1568−1576. (8) (a) Frederickson, C. J.; Kasarskis, E. J.; Ringo, D.; Frederickson, R. E. J. Neurosci. Methods 1987, 20, 91−103. (b) Harkins, S. B.; Peters, J. C. Organometallics 2002, 21, 1753−1755. (c) Chen, B.; Dodge, M. E.; Tang, W.; Lu, J.; Ma, Z.; Fan, C.; Wei, S.; Hao, W.; Kilgore, J.; Williams, N. S.; Roth, M. G.; Amatruda, J. F.; Chen, C.; Lum, L. Nat. Chem. Biol. 2009, 5, 100−107. (d) Yakushchenko, I. K.; Kaplunov, M. G.; Krasnikova, S. S.; Roshchupkina, O. S.; Pivovarov, A. P. Russ. J. Coord. Chem. 2009, 35, 312−316. (e) Yoon, J.; Wilson, S. A.; Jang, Y. K.; Seo, M. S.; Nehru, K.; Hedman, B.; Hodgson, K. O.; Bill, E.; Solomon, E. I.; Nam, W. Angew. Chem., Int. Ed. 2009, 48, 1257−1260. 2460

DOI: 10.1021/acscatal.5b02938 ACS Catal. 2016, 6, 2452−2461

Research Article

ACS Catalysis (25) Gwon, D.; Park, S.; Chang, S. Tetrahedron 2015, 71, 4504− 4511. (26) Stricker, M.; Oelkers, B.; Rosenau, C. P.; Sundermeyer, J. Chem. - Eur. J. 2013, 19, 1042−1057. (27) Although Cp*Ir(OAc)2 is calculated to be more stable than [Cp*Ir(OAc)]+, the latter was used as the catalytically active species based on the following reasons: (i) in ref 25, Chang investigated systematically the role of acid additive in the Ir-catalyzed direct C(sp2)−H amidation, the mono-cationic species [Cp*Ir(OAc)] (NTf2) is mainly responsible for the C−H bond activation; (ii) considering there should be one vacant coordination site left for further coordination of the substrate, it’s not necessary to use the Cp*Ir(OAc)2 as the starting point because energy barriers for the C(sp2)−H activation in presence of two and one coordinated OAc− are similar: Xing, Z.; Huang, F.; Sun, C.; Zhao, X.; Liu, J.-B.; Chen, D. Inorg. Chem. 2015, 54, 3958−3969. (28) (a) Sperger, T.; Sanhueza, I. A.; Kalvet, I.; Schoenebeck, F. Chem. Rev. 2015, 115, 9532−9586. (b) Balcells, D.; Clot, E.; Eisenstein, O. Chem. Rev. 2010, 110, 749−823. The CMD mechanism equals ambiphilic metal ligand activation (AMLA) via 6-membered transition states, see: (c) Boutadla, Y.; Davies, D. L.; Macgregor, S. A.; Poblador-Bahamonde, A. I. Dalton Trans. 2009, 5820−5831. (29) Intrieri, D.; Zardi, P.; Caselli, A.; Gallo, E. Chem. Commun. 2014, 50, 11440−11453. (30) (a) Karaghiosoff, K.; Klapötke, T. M.; Krumm, B.; Nöth, H.; Schütt, T.; Suter, M. Inorg. Chem. 2002, 41, 170−179. (b) Zeng, X.; Gerken, M.; Beckers, H.; Willner, H. J. Phys. Chem. A 2010, 114, 7624−7630. (c) Kubicki, J.; Luk, H. L.; Zhang, Y.; Vyas, S.; Peng, H.; Hadad, C. M.; Platz, M. S. J. Am. Chem. Soc. 2012, 134, 7036−7044. (31) (a) Boren, B. C.; Narayan, S.; Rasmussen, L. K.; Zhang, L.; Zhao, H.; Lin, Z.; Jia, G.; Fokin, V. V. J. Am. Chem. Soc. 2008, 130, 8923−8930. (b) Lamberti, M.; Fortman, G. C.; Poater, A.; Broggi, J.; Slawin, A. M. Z.; Cavallo, L.; Nolan, S. P. Organometallics 2012, 31, 756−767. (c) Manca, G.; Gallo, E.; Intrieri, D.; Mealli, C. ACS Catal. 2014, 4, 823−832. (32) For calculated relative free energies, see Table S1 and S2 in the Supporting Information. (33) See Figure S1 in the Supporting Information. (34) The single Ir−C bond arised from donation from C-pz to Ir-5dz2 is not discussed throughout for the purpose of clarity. (35) According to energy decomposition analyses of heterometallabenzenes by Solà, the C−N bond is considerably stronger than M−N (M = Rh, Ir) bond by ca. 70 kcal/mol: El-Hamdi, M.; El Bakouri El Farri, O.; Salvador, P.; Abdelouahid, B. A.; El Begrani, M. S.; Poater, J.; Solà, M. Organometallics 2013, 32, 4892−4903. (36) Glendening, E. D.; Landis, C. R.; Weinhold, F. WIREs Comput. Mol. Sci. 2012, 2, 1−42. (37) Park, S. H.; Kwak, J.; Shin, K.; Ryu, J.; Park, Y.; Chang, S. J. Am. Chem. Soc. 2014, 136, 2492−2502. (38) Our attempts to locate the plausible concerted transition state failed. According to a similar concerted transition state reported in ref 37, the energy barrier of the concerted pathway is 23.3 kcal/mol higher than that of the stepwise pathway. (39) (a) Lau, W.; Huffman, J. C.; Kochi, J. K. Organometallics 1982, 1, 155−169. (b) Koo, K.; Hillhouse, G. L. Organometallics 1995, 14, 4421−4423. (c) Koo, K.; Hillhouse, G. L. Organometallics 1996, 15, 2669−2671. (d) Lin, B. L.; Clough, C. R.; Hillhouse, G. L. J. Am. Chem. Soc. 2002, 124, 2890−2891. (e) Muñiz, K. Angew. Chem., Int. Ed. 2009, 48, 9412−9423. (40) The optimized geometries of P1b and P1a are very similar to each other. (41) (a) Kozuch, S.; Shaik, S. Acc. Chem. Res. 2011, 44, 101−110. (b) Kozuch, S. WIREs Comput. Mol. Sci. 2012, 2, 795−815. (42) For the definition of “turnover-limiting step”, see: Bosnich, B. Acc. Chem. Res. 1998, 31, 667−674. (43) Simmons, E. M.; Hartwig, J. F. Angew. Chem., Int. Ed. 2012, 51, 3066−3072.

2461

DOI: 10.1021/acscatal.5b02938 ACS Catal. 2016, 6, 2452−2461