Rhodium Catalyzed Asymmetric Hydroamination of Internal Alkynes

Feb 13, 2018 - In the early catalytic event, uptake of alkyne is predicted to take place only after the oxidative addition of the Rh(I) active catalys...
1 downloads 10 Views 3MB Size
Article Cite This: J. Org. Chem. 2018, 83, 2627−2639

pubs.acs.org/joc

Rhodium Catalyzed Asymmetric Hydroamination of Internal Alkynes with Indoline: Mechanism, Origin of Enantioselectivity, and Role of Additives C. Athira, Avtar Changotra, and Raghavan B. Sunoj* Department of Chemistry, Indian Institute of Technology Bombay, Powai, Mumbai 400076, India S Supporting Information *

ABSTRACT: A comprehensive mechanistic study on the title reaction by using DFT(B3LYP-D3) computational method is reported. Explicit consideration of mono- (m-xylylic) and dicarboxylic acid (phthalic) in the key transition states reveals active participation of the carboxylic acid, beginning with the generation of a monomeric Rh(I) active catalyst and in the ensuing catalytic steps. In the early catalytic event, uptake of alkyne is predicted to take place only after the oxidative addition of the Rh(I) active catalyst to the carboxylic acid. The hydrometalation of the alkyne bound to the Rh(III)−H intermediate then generates a Rh(III)−vinyl intermediate, which in turn converts to a Rh(III)−allyl species. The inclusion of m-xylylic acid results in a two-step pathway to Rh(III)−allyl species via Rh−allene intermediate. A number of weak noncovalent interactions (hydrogen bonding and C−H···π) between the catalyst and the substrates and that involving m-xylylic acid are found to have a direct impact on the regiochemical preference toward the branched product and the enantiocontrolling hydroamination step involving C−N bond formation leading to the major enantiomer (S-allylic amine). The chiral induction is enabled by cumulative effect of noncovalent interactions, which is an insight that could aid future developments of chiral ligands for asymmetric hydroamination.



INTRODUCTION The development of efficient methods for the construction of C−N bonds leading to various amines has been an active endeavor in chemical synthesis. Amines continue to remain an attractive synthetic target owing to their presence in biologically active compounds, ranging from naturally occurring alkaloids to manmade therapeutics.1 Along with the methodological improvements toward the synthesis of various types of amines, asymmetric catalysis for the construction of chiral amines has received notable attention. In asymmetric amination reactions, different forms of selectivity (regio and stereo) could be tuned by the use of chiral ligands, additives and so on.2 Allyl amines are valuable building blocks in organic synthesis, which can be subjected to further synthetic transformations.3 Hence, the installation of chiral allylic amine on various molecules is an attractive objective. Allylic substitution4 and hydroamination5 are two widely used methods for the synthesis of allylic amines. Introduction of nitrogen and hydrogen across a double or triple bond is known as hydroamination wherein the N−H bond of an amine adds across an alkene or alkyne to generate the amines, imines and enamines. Hydroamination of alkenes and alkynes is a relatively newer method for the construction of C−N bond.5 The major advantage of hydroamination over other methods is that it avoids the formation of byproducts and renders enhanced catalytic efficiency. The method offers optimal atom economy and works with a range of readily available and inexpensive starting materials. In transition metal © 2018 American Chemical Society

catalyzed hydroamination, gaining control over chemo-, regio-, and stereoselectivities in hydroamination are relatively harder. To ensure chemoselectivity in hydroamination reactions, the catalyst should preferentially coordinate to the C−C multiple bond without reacting with other functional groups, or engage in other competing reactions like oligomerization.6 A more difficult issue with alkynes is to gain control over Markovnikov or anti-Markovnikov selectivity as well as the overall conversion.7 For internal C−C multiple bonds, regioselectivity can be influenced by introducing electron withdrawing or bulky substituents at one end of the multiple bond.8 In intramolecular hydroaminations, the Markovnikov or anti-Markovnikov addition is mostly controlled by the preferential formation of five- or six-membered rings.9 While the selectivity issues can be addressed by suitable selection of substrates, ligands, and metal catalysts, the thermodynamic and kinetic concerns can largely be minimized by the use of carboxylic acid additives. The generally regarded role of such additives is to protonate the electron-rich alkyne and thereby promotes the nucleophilic attack of amine. Similarly, transition metal catalysts can also help in decreasing the electron density of the alkyne/alkene coupling partner resulting in further improvement in the propensity for nucleophilic attack. Particularly noteworthy are the efficient use of metal−alkynes for the synthesis of linear Received: December 3, 2017 Published: February 13, 2018 2627

DOI: 10.1021/acs.joc.7b03047 J. Org. Chem. 2018, 83, 2627−2639

Article

The Journal of Organic Chemistry allylic amines10 and the use of carboxylic acid additives for branched allylic esters.11 These methods, making use of metal− alkyne to metal−allyl conversion, have been effectively employed in asymmetric catalysis toward the construction of allylic C−C, C−S and C−O bonds.12 Very recently, Dong and co-workers offered an elegant demonstration of hydroamination of internal alkynes with indolines to generate allylic amine (Scheme 1).13 Fascinating

Besides the inherent synthetic advantages of this allylic hydroamination reaction, it serves equally well as an example from which one could seek answers to a few general and important mechanistic questions. It is obvious that a metal− alkyne complex is formed between the catalyst and the substrate in the initial phase of the reaction. Such a species could undergo (a) a stepwise metal−allyl isomerization reaction via a metal-allene intermediate,11,14 or (b) a concerted isomerization directly to a metal−allyl species.15 Molecular insights and energetic details on which one of these pathways is likely to be preferred are of timely relevance. In addition, we wish to explore the role of carboxylic acid additives in this hydroamination reaction, details of which are not readily known. The origin of high enantioselectivity is another aspect of our motivation. The molecular insights on these could help exploit the potential of this class of hydroamination reactions by way of suitable choice of reacting partners as well as the chiral phosphine ligand backbone. In the following sections, we present our key results of a detailed mechanistic study by using density functional theory computations.

Scheme 1. Branch Selective Allylic Hydroamination of Internal Alkynes with Indolines13



aspects of this method are (a) good enantioselectivity, (b) regiochemical preference toward the formation of chiral branched allylic amine over the commonly noted linear allylic amines under the reaction conditions,13b and (c) the role of carboxylic acid additives in the overall catalytic pathway.

COMPUTATIONAL METHODS

Geometries of stationary points involved in the potential energy surface (such as reactants, intermediates, transition states and products) were optimized in the gas phase at the B3LYP-D3/631G** level of theory by using Gaussian09 (Revision D.01) suite of quantum chemical programs.16 The hybrid density functional B3LYP-

Scheme 2. Catalytic Cycle Leading to the Formation of a Branched Allylic Amine by Hydroamination of Internal Alkynes in the Presence of m-Xylylic Acida

a

(CH3)2C6H3COOH, 3,5-dimethyl benzoic acid. 2628

DOI: 10.1021/acs.joc.7b03047 J. Org. Chem. 2018, 83, 2627−2639

Article

The Journal of Organic Chemistry

through a π-coordination can lead to a pentacoordinate species (3). In intermediate 3, the insertion of the coordinated alkyne to the Rh(III)−H bond can furnish a Rh(III)−vinyl intermediate (4) via transition state (3−4)‡. The intermediate 4 can then undergo a β-C−H activation via (4−5)‡ to generate an important Rh−allene complex (5). Protonation of the Rh− allene complex by the coordinated carboxylic acid through transition state (5−6a)‡ can then generate a Rh(III)−allyl intermediate (6a). It might also be possible that a concerted 1,2 hydrogen shift via (4−6a)‡ can offer a direct route from Rh(III)−vinyl (4) to Rh(III)−allyl intermediate (6a), without the involvement of a Rh−allene complex (5). The η3-Rh(III)− allyl intermediate 6a can change to an η2 Rh(III)−allyl intermediate via transition state (6a−6c)‡. In the final step, the nucleophilic indoline can attack the η 2 Rh−allyl intermediate 6c to generate the product P via (6c−P)‡. The overall catalytic cycle can be considered as consisting of two key events, (a) conversion of Rh−alkyne (3) to Rh−allyl intermediate (6a/6c), and (b) the enantiocontrolling step involving the nucleophilic attack of the indoline on the Rh− allyl species (6c) to generate a branched allylic amine (P).30 To begin with, we have considered a carboxylic acid promoted dissociation of the dimeric precatalyst [((S,S)-BDPP)2RhCl]2 (1A) to a monomeric tetracoordinate active catalyst ((S,S)BDPP)Rh(Cl)(PhCOOH) 1 (where BDPP is (2S,4S)-2,4bis(diphenylphosphino)pentane). Alternative possibilities, wherein a direct uptake of the alkyne by the dimeric precatalyst 1A or by a monomeric (BDPP)Rh(Cl) species, is found to be of higher energies as compared to the formation of 1.29b In the most preferred pathway, we note that the oxidative insertion of Rh(I) to the O−H bond of the carboxylic acid occurs prior to the coordination of the alkyne to the metal center. It is important to note that the formation of such Rh(III)−H intermediates before the alkyne coordination in related situations have been experimentally observed.14c,15c The relative Gibbs free energy of the transition state for the formation Rh(III)−H complex via transition state (1−2)‡ is found to be 4.6 kcal/mol.31 Next, in the hydrometalation step, the Rh−alkyne converts to a Rh−vinyl intermediate. This step can be regarded as an insertion of the alkyne to the Rh(III)−H bond of intermediate 3 to generate intermediate 4.31c Since the alkyne is unsymmetrically substituted, the alkyne insertion to the Rh(III)−H intermediate can take place either at C1 or at C2 carbon, as shown in Scheme 3. Insertion at C1 yields vinyl intermediate 4a whereas insertion at C2 results in 4b. The Gibbs free energies of the transition states are found to be quite comparable for insertion at C1 and C2 positions (Scheme 3). The elementary step barrier with respect to the Rh(III)−H intermediate 2 is about 17.9 kcal/mol.32

D317 were used in conjunction with the SDD basis set for Rh.18 The remaining elements were treated with Pople’s 6-31G** basis set.19 Frequency calculations were performed at the same level of theory for all stationary points to identify whether they represents a minima (zero imaginary frequency) or transition states (one imaginary frequency). All the transition states were identified by the visual inspection of one and only one imaginary frequency which corresponds to its reaction coordinate. We have done intrinsic reaction coordinate (IRC) calculations to verify whether transition states are connected to reactants and products at B3LYP-D3/6-31G**,Rh(SDD).20 The final geometry obtained from the IRC calculations on both sides of the first order saddle point were subsequently subjected to further geometry optimization by using “OPT = CALCFC” option. Solvent effects, in tetrahydrofuran (THF) as the continuum dielectric (ε = 7.39), were evaluated at the B3LYP-D3/6-311G**,SDD using the gas phase optimized geometries.21 Solvation energies were calculated by selfconsistent reaction field using the SMD solvation model.22 Thermal and entropic corrections were then applied to the single point energies to obtain the Gibbs free energies of various stationary points. The default electronic energies, which refer to an ideal gas (p = 1 atm) standard state, were corrected so as to convert the same to the standard state corresponding to a species in solution with a standard concentration of 1 mol/L using eq 1 shown below. The standard state correction of RT ln(24.5) were added to the electronic energies of the reactants, intermediates and transitions states computed at the SMD(THF)/B3LYP-D3/6-31G**,SDD(Rh) level of theory to obtain the Gibbs free energies. The notations ° and °′ in eq 1 represent the standard states respectively at 1 atm pressure and 1 mol/L, in the gas phase.23 The temperature of the reaction under investigation is 70 °C, and hence T is set equal to 343 K in eq 1. Numerical value for the correction factor for the above eq 1 is found to be 2.2 kcal/mol. The discussions are presented on the basis of the Gibbs free energies at the SMD ( T H F ) /B3LYP-D3/6-311G**,SDD(Rh)//B3LYP-D3/631G**,SDD(Rh) level of theory.24

ΔG°′ = ΔG + RT ln(Q °′|Q °)

(1)

Topological analyses of electron densities were carried out using Bader’s Atoms-in-Molecule formalism wherein bond critical paths and bond critical points were identified.25 An interacting pair of atoms is identified by the presence of a bond path and a bond critical point (bcp) along the bond paths. The value of electron density at such bcps can be regarded as proportional to the strength of the interaction. The shared electron density shows a minimum at the bcp. Analyses of noncovalent interactions were carried out using noncovalent interaction index (NCI) plots developed by Yang and co-workers.26 We have also used activation strain method (distortion-interaction model) to examine the energy difference between important transition states.27 In activation strain method, the activation energy (ΔE‡) is equated to the sum of distortions in the reactant geometries (ΔEd‡) as noted in the transition state structures in comparison to the respective undistorted ground state geometries and the interaction energy (ΔEi‡) between such distorted fragments in the transition state geometries. The activation barrier can be written as ΔE‡ = ΔEd‡ + ΔEi‡. We have analyzed the overall Gibbs free energy profile diagram using the energetic span model proposed by Shaik and Kozuch.28



Scheme 3. Insertion of Unsymmetrical Alkyne to Rh(III)−H Bond To Generate Rh(III)−Vinyl Intermediatea

RESULTS AND DISCUSSIONS A general mechanism for the formation of a branched allylic amine through hydroamination of alkyne by using an indoline is shown in Scheme 2. The precatalyst Rh bis-phosphine complex {Rh(BDPP)2Cl}2 1A can dissociate, by the action of carboxylic acid (a desirable additive in the reaction), to give a tetracoordinate Rh(I)-complex (1).29 This monomeric Rh(I) intermediate 1 can be considered as the catalytically active species initiating the catalytic cycle. In the next step, 1 can undergo an oxidative addition to the coordinated carboxylic acid to generate a Rh(III)−H intermediate (2) via transition state (1−2)‡. Next, the alkyne uptake by the Rh(III)−H

a

The relative Gibbs free energies (in kcal/mol) with respect to the separated reactants are given in parentheses. The barriers are computed with respect to the lower energy Rh(III)−H intermediate 2.33 ArCOOH represents m-xylylic acid here.

2629

DOI: 10.1021/acs.joc.7b03047 J. Org. Chem. 2018, 83, 2627−2639

Article

The Journal of Organic Chemistry

Figure 1. (i) A general representation of β-C−H bond activation in the conversion of the Rh(III)−vinyl (4a) to Rh−allene intermediate (5). The relative Gibbs free energies (in kcal/mol) of the transition states with respect to the separated reactants are given in parentheses. The notation ′ is used to denote the transition states with phthalic acid as the additive. (ii) Optimized geometries of the transition states involving phthalic acid showing important interatomic contacts (in Å) and the corresponding electron densities (ρ × 10−2) at the bond critical points as well as the angle of interaction (θ, in °) for various noncovalent interactions. Atom color codes; black = C, white = H, red = O, dark cyan = P, yellow= Cl.

Stepwise Pathway from Rh−Vinyl to Rh−Allyl Intermediate. The stepwise pathway is also termed as the allene pathway owing to the involvement of an allene intermediate en route to 6a.11 In the stepwise pathway, a βC−H activation at the methyl group of the Rh(III)−vinyl intermediate 4a generates a Rh−allene complex 5 first. The formation of the allene intermediate could be reversible (i.e., 5 to intermediate 4a, see Figure 5) which is in agreement with the experimental observations reported by Dong and coworkers.13 As shown in Figure 1, this ligand assisted β-C−H activation can either be performed by the Rh-bound carboxylate (via (4a−5)A‡) or through the participation of a second molecule of carboxylic acid which is not directly bound to the Rh center (via (4a−5)B‡). In view of the use of carboxylic acid as additives in this reaction, we envisaged the potential involvement of a second molecule of carboxylic acid in a

In the most preferred geometry of these transition states, the hydride ligand prefers to remain trans to the chloride ligand than to the phosphine.34 We note that consideration of different likely relative positioning between various ligands around the Rh center is vital toward identifying the most preferred transition state for every elementary step in the catalytic cycle. In particular, positioning of the alkyne with respect to the phosphine ligands is found to have a direct bearing on the metal to alkyne back-donation, and consequently on the energetics of the alkyne insertion to the Rh(III)−H bond.35 As described above, the generation of two likely Rh(III)−vinyl intermediates, 4a and 4b, led us to consider a concerted and a stepwise pathway for the conversion of the Rh(III)−vinyl to Rh(III)−allyl species (6a). A number of interesting details of this conversion are provided in the following section. 2630

DOI: 10.1021/acs.joc.7b03047 J. Org. Chem. 2018, 83, 2627−2639

Article

The Journal of Organic Chemistry

Scheme 4. Concerted β-C−H Activation and Concomitant Protonation Involved in the Conversion of the Rh(III)−Vinyl (4a/b) to the Rh(III)−Allyl Intermediate (6a/b)a

a

ArCOOH= 3,5-dimethyl benzoic acid (m-xylylic acid) or 1,2-benzene dicarboxylic acid (phthalic acid).

hydrogen bonding (g) is found to be of higher energy (Figure 1). The transition state (4a′−5)A‡ is lower in energy than (4a′−5)A1‡, perhaps due to more number of C−H···π interactions (r, s and t). Though the number of hydrogen bonding interactions are more in the case of (4a′−5)A1‡ than in (4a′−5)A‡, the C−H···π interactions (r, s and t) between the ligand and substrate appears to play a more crucial role in lowering the energy of the β-C−H activation transition state. The presence of these noncovalent interactions were further confirmed by using topological analysis of electron density within the atoms in molecule (AIM) formalism.38 The electron densities (ρ × 10−2) and angle of interaction (θ = C−H−x angle, where x is a dummy atom placed at the centroid of the π system) at various bond critical points are analyzed to get an approximate measure of the strength of weak noncovalent interactions. The hydrogen bonding contacts a, d, o and p as well as the C−H···π interactions r and s are found to be common in both (4a′−5)A‡ and (4a′−5)A1‡. The strength of the hydrogen bonding interactions (a, d and o) are relatively higher in the case of case of (4a′−5)A‡ than the corresponding ones in (4a′−5)A1‡.39 In addition, number of more prominent C−H···π interactions (r, s and t) are noticed in (4a′−5)A‡ than that in (4a′−5)A1‡ (r and s). On the other hand, in (4a′−5)B‡, the second carboxylic acid group is hydrogen bonded to the chloride ligand. This relay proton transfer transition state (4a′− 5)B‡ exhibits a rich number of noncovalent interactions such as hydrogen bonding (a−h), C−H···π (i, j and l), O−H···π (k) and CO···π interactions (m, carbonyl group of the phthalate interacts with the CC of the vinyl ligand). These additional interactions provided by the flexible arm of the Rh-bound phthalate as well as the due to the participation of a second molecule of phthalic acid appear to lower the energy of the relay proton transfer transition state (4a′−5)B‡. A cumulative effect of different types of noncovalent interactions is responsible for the lowering the energy of β-C−H activation transition state in the stepwise pathway. After the formation of Rh−allene complex 5, the coordinated carboxylic acid can now protonate the central carbon atom (C2) of 5 to generate Rh(III)−allyl intermediate (6a) via (5−6a)‡ transition state.

relay proton transfer mechanism. The Gibbs free energies, provided in Figure 1, suggest that in the case of β-C−H activation by the Rh-bound m-xylylate ((4a−5)A‡) is about 5.0 kcal/mol lower as compared to when a second molecule of xylylic acid is involved ((4a−5)B‡). The activation barrier for the β-C−H activation, calculated with respect to intermediate 4a, also exhibits the same trend.36 Interestingly, when phthalate is involved, both modes of β-C−H bond activation TSs (4a′− 5)A‡ and (4a′−5)B‡ are found to be of comparable energies.37 This prediction provides an important lead that the nature of the carboxylic acid additive involved in this mechanism might exert a significant impact on the energetics of hydroamination reaction. If the effect of carboxylic acid additives is confined only to the acidity of the concerned acid, the energetic changes as described in Figure 1, cannot be fully rationalized. Hence, we have considered mechanistic models that explicitly include carboxylic acid additives in the transition states and have carefully examined the factors that influence the energetic course of this reaction upon changing from m-xylylic to phthalic acid. It is expected that the interaction of m-xylylic acid (3,5dimethyl benzoic acid) acid with the reactants in the transition state would be different from that of phthalic acid (1,2dicarboxylic acid). Certain interesting molecular features became evident through the analysis of the transition state geometries. An important Cl···H hydrogen bonding (shown as g in Figure 1) between the second carboxylic acid group of the phthalate and Rh-bound chloride is noticed in (4a′−5)B‡.38 In an alternative geometry (not shown here), where this interaction is less prominent, (4a−5)A‡ again becomes energetically more preferred than (4a−5)B‡.38b This is an interesting indication that the hydrogen bonding pattern provided by the ortho-dicarboxylic acid can offer additional stabilization to the concerted β-C−H bond activation TS. It is further noticed that the site of hydrogen bonding is different in (4a′−5)A‡ (q) and (4a′−5)B‡ (g). In (4a′−5)A‡, the orthocarboxylic acid group of the phthalate ligand develops a weaker hydrogen bonding with the methylenic C−H of the vinyl ligand (q). The corresponding transition state (4a′−5)A1‡ with Cl···H 2631

DOI: 10.1021/acs.joc.7b03047 J. Org. Chem. 2018, 83, 2627−2639

Article

The Journal of Organic Chemistry

Figure 2. Optimized geometries of the transition states for the formation of Rh(III)−allyl (6) from Rh(III)−vinyl (4) intermediate via concerted βC−H activation and protonation with phthalic acid acts as the additive. Relative Gibbs free energies (kcal/mol) calculated with respect to the separated reactants is given in parentheses. The interatomic contacts (in Å) and the corresponding electron densities (ρ × 10−2) at the bond critical points as well as the angle of interaction (θ, in °) for various noncovalent interactions are provided in the table. Only selected hydrogen and carbon atoms are shown for improved clarity. Atom color codes; black = C, white = H, red = O, dark cyan = P, yellow= Cl.

The relative Gibbs free energy of (5−6a)‡ is found to be 11.5 kcal/mol and the corresponding activation barrier is 16.0 kcal/ mol (Figure 5).40 Concerted Pathway from Rh−Vinyl to Rh−Allyl Intermediate. Another possible pathway for the direct formation of Rh(III)−allyl intermediate 6a is through a concerted β-C−H bond activation and a concomitant protonation in 4a (Scheme 4). In the concerted pathway, an explicit participation of a second carboxylic acid molecule promotes the vital hydrogen shift in the Rh(III)−vinyl intermediate. A 1,2-hydrogen shift in intermediate 4a and a 1,3-hydrogen shift in the case of 4b can lead to the formation of the Rh(III)−allyl intermediate (6a/6b). Although similar kinds of 1,2 and 1,3 hydrogen shift mechanisms were invoked in alkene isomerization reactions,41 the application of this concept to Rh−vinyl to Rh−allyl conversion has not been reported until date. The Rh(III)−allyl intermediate can exist in an exo (6a) or endo (6b) configuration depending upon the mutual alignment

of the C2−H bond of the allyl ligand and Rh-bound chloride. If C2−H and Rh−Cl bonds are in the same direction it is exo (6a) and if those are in opposite directions it is called endo (6b). The preferred site of protonation in Rh(III)−vinyl intermediate 4a is at the C2 carbon of the vinyl moiety and in the case of 4b protonation occurs at the C1 carbon (Scheme 4).42 In the case of m-xylylic acid, such a concerted pathway is found to be higher energy as compared to the corresponding stepwise pathway (vide infra).43 Next, the effect of phthalic acid in the concerted pathway is examined. The transition state geometries and the corresponding relative Gibbs free energies for the phthalic acid assisted relay proton transfer are given in Figure 2. The (4a′−6b′)A‡ and (4b′−6a′)A‡ are the transition states respectively for 1,2 and 1,3 hydrogen shift in Rh(III)−vinyl intermediates 4a′ and 4b′. The corresponding Gibbs free energies of these transition states are respectively 10.6 kcal/mol ((4a′−6b′)A‡) and 8.7 kcal/mol ((4b′−6a′)A‡). One of the carboxylic acid groups of 2632

DOI: 10.1021/acs.joc.7b03047 J. Org. Chem. 2018, 83, 2627−2639

Article

The Journal of Organic Chemistry

Figure 3. Optimized geometries for the regioselective C−N bond formation transition states (6c−P)‡. Relative Gibbs free energies (kcal/mol) calculated with respect to the separated reactants are given in parentheses. The interatomic contacts (in Å) and the corresponding electron densities (ρ × 10−2) at the bond critical points as well as the angle of interaction (θ, in °) for various noncovalent interactions are given in the table. Atom color codes; black = C, white = H, red = O, dark cyan = P, green= Cl.

the dicarboxylic acid additive engages in a relay transfer in these transition states. Comparison of the transition states for the stepwise ((4a′−5)A‡ 10.7 kcal/mol, Figure 1) and concerted ((4b′−6a′)A‡ 8.7 kcal/mol, Figure 2) mechanisms indicates that the concerted β-C−H activation involving a 1,3 hydrogen shift mechanism is 2 kcal/mol lower. The corresponding activation barriers for the concerted 1,3 hydrogen shift is 29.0 kcal/mol while the barrier in the stepwise mechanism is 31.0 kcal/mol.43 The Gibbs free energies suggest a switch in mechanism from stepwise to concerted, when phthalic acid is involved in the critical transition state from Rh−vinyl 4 to Rh−allyl 6. It could be possible that the second carboxylic acid group of the phthalate ligand can actively participate in a relay proton transfer in the concerted pathway. It is noticed that such a mode further stabilizes the proton transfer transition state ((4a′−6b′)B1‡) by 13.6 kcal/mol and the corresponding activation barrier is 17.4 kcal/mol with respect to the preceding hydrogen bonded Rh(III)−vinyl intermediate (Figure 2). This pathway is considerably lower than the stepwise mechanism (the activation barrier for the allene pathway is 31.0 kcal/ mol).43

Next, we have examined various noncovalent interactions (hydrogen bonding and C−H···π interactions), which contribute to the energetic preference toward a stepwise, or a concreted pathway. As already noted in earlier that the flexible arm of the ortho carboxylic acid displays different types of hydrogen bonding interactions in the case of transition state (4a′−5)B‡. In transition states (4a′−6b′)A‡ and (4b′−6a′)A‡, the pattern of the hydrogen bonding is same (a to i), but a few such hydrogen bonding interactions (b, c, d, f, g and h) are found to be better in (4b′−6a′)A than in (4a′−6b′)A. These hydrogen bonding interactions include CO···HC or C O···HO contacts between the additives, ligand and allyl ligand (f, g and h) as well as the Cl···H hydrogen bonding interactions (b, c and d). The strength of remaining interactions (a, e and i) are moderately higher in (4a′−6b′)A‡. The number of C−H···π interactions are less in the case of (4b′−6a′)A‡ (r and s) than in (4a′−6b′)A‡ (r, s and t). It is evident from the AIM data given in Figure 2 that the strength of the two common C−H···π interactions (r, s) is higher in (4b′−6a′)A‡ than in (4a′−6b′)A‡.39 The transition state (4a′−6b′)B1‡, which is lower in energy than both (4b′−6a′)A‡ and (4a′− 6b′)A, is stabilized by more number of stronger hydrogen bonding interactions (a to p). The number and strength of C− 2633

DOI: 10.1021/acs.joc.7b03047 J. Org. Chem. 2018, 83, 2627−2639

Article

The Journal of Organic Chemistry

Figure 4. Optimized geometries and relative Gibbs free energies (in parentheses kcal/mol) for the stereocontrolling C−N bond forming transition states via (6c−P)‡. The interatomic contacts corresponding to the noncovalent interactions between the catalyst, substrates and additives and the reaction coordinates (left) are given in Å and the bond angles (θ) are reported in degrees (deg) within square brackets. The electron densities (ρ × 10−2) corresponding to these noncovalent interactions at the bond critical points are given in the table. Atom color codes; black = C, white = H, red = O, dark cyan = P, green= Cl.

H···π interaction is very less in (4a′−6b′)B1‡ (t) than in (4b′− 6a′)A‡ (r, s and t). The hydrogen bonding provided by the flexible ortho carboxylate arm of the phthalate ligand as well as the involvement of the second molecule of the phthalic acid are important in lowering the energy of the concerted β-C−H activation transition state. The C−N Bond Formation between Rh−Allyl and Indoline. The η3-Rh(III)−allyl intermediate (6a/6b), as described in the previous section, can isomerize to a lower energy η2-Rh(III)−allyl intermediate (6c) via (6a−6c)‡.42b,44 The most important step in the catalytic cycle is the stereoselective C−N bond formation between 6c and indoline. This C−N bond formation can occur either via an inner-sphere mechanism or through an outer-sphere mechanism, depending on whether the incoming nucleophile is coordinated to the metal center or not. In the inner-sphere pathway, both the electrophile and nucleophile are coordinated to the Rh center and the C−N bond formation occurs via a reductive elimination. In the outer-sphere pathway, the nucleophile is not coordinated to the Rh and proceed via a concerted pathway wherein N−H deprotonation and C−N bond formation occurs

simultaneously. Importantly, the energies associated with the inner-sphere C−N bond formation are found to be higher than the corresponding outer-sphere alternatives.45 In the preferred outer-sphere pathway, the nucleophile adds to the η2-Rh(III)− allyl intermediate (6c) in a concerted fashion via (6c−P)‡ wherein the N−H deprotonation and C−N bond formation takes place in a concomitant manner. The most preferred C−N bond formation transition states is identified through a fairly detailed study encompassing different geometries of the Rh− allyl species, particularly with respect to the chiral (S,S)-BDPP ligand backbone.46 It is therefore of importance to examine what factors contribute to the observed allylic amination step leading to the branched (P) over a potential alternative path to a linear (PL) product in the present reaction. Origin of Regioselectivity in the C−N Bond Formation. In addition to the preferred outer-sphere mode of the C−N bond formation, the site of attack of the nucleophilic indoline on the Rh-π-allyl intermediate is an inherently interesting aspect. When the indoline forms a bond with the phenyl bearing carbon of the Rh-π-allyl intermediate, it results in a branched product (P) whereas the addition to the 2634

DOI: 10.1021/acs.joc.7b03047 J. Org. Chem. 2018, 83, 2627−2639

Article

The Journal of Organic Chemistry

the corresponding addition to the re-face. More specifically, the addition to the si-syn face of the Rh−allyl intermediate gives the S product whereas the re-syn face addition yields the R isomer. Here, notation syn conveys that the C2−H bond of the Rh− allyl moiety and the phenyl group of the allyl ligand are in the same direction. The Gibbs free energy for the transition state (6c−P)S‡ that corresponds to the formation of the S enantiomer of the product is 4.8 kcal/mol lower than (6c− P)R‡ for the R isomer (Figure 4). Such an energy difference between the diasteomeric transition states corresponds to an enantiomeric excess of >99%, which is in good agreement with the experimental value (90% ee).13a As can be noticed from Figure 4, these two enantiocontrolling transition states differ in the relative positions of the carboxylate and the allyl ligands. In (6c−P)S‡ they are nearly trans to each other (O1−Rh−C6 = 145.14°), whereas in (6c−P)R‡ it is more toward a cis arrangement (O1−Rh−C6 = 98.37°). Now, it is important to identify the key factors contributing to the predicted enantioselectivity in the allylic C−N bond formation. The ability to increase in the bite angle of the phosphine ligands on the catalyst, as compared to that in the preceding intermediate, has earlier been suggested to improve the flexibility of the chiral cavity and hence its ability to hold the reacting partners in closer proximity.50 However, in the present case, bite angle (β) in the most preferred transition state (6c− P)S‡ is found to be lower than that in 6c.51 Under this circumstance, we have analyzed the role of weak noncovalent interactions in the stereocontrolling C−N bond formation transition states and how the differences in such interactions impact the relative energies of the diastereomeric transition states. The optimized geometries of the most preferred C−N bond formation transition states are provided in Figure 4. We have reoptimized the stereocontrolling transition states also using other functionals, such as ω-B97XD and M06-L, using the SMD(THF) continuum solvation model. At both the SMD(THF)/M06-L/6-31G**,SDD(Rh) and the SMD(THF)/ωB97XD/6-31G**,SDD(Rh) levels of theory, the key structural features such as bite angle and noncovalent interaction distances remained nearly the same.13d A series of noncovalent interactions have been identified and are summarized beside each transition state geometry. Herein, we focus on the major differences in the pattern of noncovalent interactions, particularly the ones operating between the catalyst and the substrate. More number of C−H···π interactions, between the allyl moiety of the substrate and phenyl groups of the chiral phosphine ligand (denoted as e, f, and g in the Figure 4), are noticed in the lower energy (6c−P)S‡ than in the higher energy (6c−P)R‡. The only common interaction f is found to be weaker while other two interactions (e and g) are absent in (6c−P)R‡. The electron density and angle corresponding to the allylic C−H···π interaction ( f) in transition state leading to the branched product (6c−P)S‡ (ρ = 1.14 × 10−2 and θf = 34.5°) indicate a stronger interaction (f) than in the higher energy analogue (6c−P)R‡ (ρ = 0.54 × 10−2 and θf= 45.3).39 Relatively more prominent and larger number of hydrogen bonding interactions are found in (6c−P)S‡. These interactions (a, b, c, and d) between the Rh-bound carboxylate and the phenyl rings of the catalyst backbone are found to be more efficient in (6c− P)S‡ than in (6c−P)R‡ (only a and d). The angle of O···H−C hydrogen bonding interaction is more linear in the case of (6c− P)S‡ (>150°) whereas in (6c−P)R‡ it is more bent ( 120° or H−C−x < 60°) is used. Here we report the C−H−x angle (θ). (b) Nishio, M.; Umezawa, Y.; Hirota, M.; Takeuchi, Y. Tetrahedron 1995, 51, 8665−8701. (c) Umezawa, Y.; Tsuboyama, S.; Takahashi, H.; Uzawa, J.; Nishio, M. Tetrahedron 1999, 55, 10047−10056. (d) Takahashi, H.; Tsuboyama, S.; Umezawa, Y.; Honda, K.; Nishio, M. Tetrahedron 2000, 56, 6185−6191. (e) A cutoff criteria for hydrogen bonding distance O···H−C (< 2.5 Å and angle θ > 90°) were also suggested. See: Buck, M.; Karplus, M. J. Phys. Chem. B 2001, 105, 11000−11015. (40) The detailed analysis of different possibilities and energetics for protonation of Rh(I)-allene complex by carboxylic acid additive is given in Supporting Information. See in Figure S8 and Table S8. (41) (a) Vasseur, A.; Bruffaerts, J.; Marek, I. Nat. Chem. 2016, 8, 209−219. (b) Knapp, S. M. M.; Shaner, E.; Kim, D.; Shopov, D. Y.; Tendler, J. A.; Pudalov, D. M.; Chianese, A. R. Organometallics 2014, 33, 473−484. (42) The detailed analysis of 1,2 hydrogen shift and 1,3 hydrogen shift in Rh(III)−vinyl intermediate is given in Supporting Information. See Figures S13−S14 and Tables S9−S10. (b) The energies of different isomers of the Rh(III)−allyl intermediates are given in Figure S17 and Table S11. (43) A Gibbs free energy profile diagram for the comparison of stepwise and concerted mechanism for the formation Rh(III)−allyl intermediate is given in Supporting Information Figures S15−S16. (44) A detailed analysis of the different types of the isomerization of Rh(III)−allyl intermediate is given in Supporting Information (Figures S18−S20 and Tables S12−S13). (45) A stepwise pathway wherein the N−H deprotonation of the indoline precedes the C−N bond formation is found to be of higher energy. More details regarding additional possibilities are given in Figures S21−S23 and Table S14 in the Supporting Information. (46) The orientation of the C2−H bond of the allyl moiety, towards and away from the phenyl group of the BDPP ligand, is examined. More details are provided in Figure S24 and Table S15 in the Supporting Information. (47) (a) Kennemur, J. L.; Kortman, G. D.; Hull, K. L. J. Am. Chem. Soc. 2016, 138, 11914−11919. (b) Daniels, D. S. B.; Jones, A. S.; Thompson, A. L.; Paton, R. S.; Anderson, E. S. Angew. Chem., Int. Ed. 2014, 53, 1915−1920. (48) Details of natural population analysis (NPA) are given in Figure S25 and Table S16 in the Supporting Information. (49) Details regarding the activation strain analysis are given in Figure S26 and Table S17 in the Supporting Information. (50) (a) Trost, B. M.; Van Vranken, D. L.; Bingel, C. J. Am. Chem. Soc. 1992, 114, 9327−9343. (b) Zhang, W.; Li, W.; Qin, S. Org. Biomol. Chem. 2012, 10, 597−604. (c) van Leeuwen, P. W. N. M.; Kamer, P. C. J.; Reek, J. N. H.; Dierkes, P. Chem. Rev. 2000, 100, 2741−2769. 2639

DOI: 10.1021/acs.joc.7b03047 J. Org. Chem. 2018, 83, 2627−2639