Ir and NHC Dual Chiral Synergetic Catalysis: Mechanism and Stereoselectivity in γ-Butyrolactone Formation

Abstract

Cooperative dual catalysis is a powerful strategy for achieving unique reactivity by combining catalysts with orthogonal modes of action. This approach allows for independent control of the absolute and relative stereochemistry of the product. Despite its potential utility, the combination of N-heterocyclic carbene (NHC) organocatalysis and transition metal catalysis has remained a formidable challenge as NHCs readily coordinate metal centers. This characteristic also makes it difficult to rationalize or predict the stereochemical outcomes of these reactions. Herein, we use quantum mechanical calculations to investigate formation of γ-butyrolactones from aldehydes and allyl cyclic carbonates by means of an NHC organocatalyst and an iridium catalyst. Stereoconvergent activation of the racemic allyl cyclic carbonate forms an Ir-π-allyl intermediate and activation of an unsaturated aldehyde forms an NHC enolate, the latter of which is rate-limiting. Union of the two fragments leads to stereodetermining C-C bond formation and ultimately ring closure to generate the product lactone. Notably, CO₂ loss occurs after formation of the C-C bond and Et₃NH⁺ plays a key role in stabilizing carboxylate intermediates and in facilitating proton transfer to form the NHC enolate. The computed pathways agree with the experimental findings in terms of the absolute configuration, the enantiomer excess, and the different diastereomers seen with the (R)-and (S)-spirophosphoramidite combined with the NHC catalyst. Calculations reveal the lowest energy pathway includes both an NHC ligand and a phosphoramidite ligand on the iridium center. However, the stereochemical features of this Ir-bound NHC were found to not contribute to the selectivity of the process.

Graphical Abstract

1. INTRODUCTION

The chirality of bioactive molecules is crucial to their physiological functions since their target macromolecules in living organisms are also chiral.¹ Currently more than half of the drugs in today’s market are chiral², with 20 of the 35 drugs approved by the FDA in 2020 being chiral.³ These drugs are either given as a pure enantiomer or as the racemic compound⁴ despite the fact that biological activities as well as toxicity and metabolism of each enantiomer can differ greatly, as seen in the cases of limonene and thalidomide.⁵ Accordingly, asymmetric catalysis to generate enantioenriched products from racemic or achiral substrates has been a significant area of interest in modern organic synthesis.⁶

Despite the remarkable advances in the catalytic enantiocontrol of single stereocenters, it remains a formidable challenge to control the absolute and relative stereochemistry of multiple chiral centers in a single reaction.⁷ Recently, cooperative dual catalysis, which combines two mechanistically orthogonal catalytic cycles, has emerged as an effective strategy to tackle this problem.⁸ The use of two independent chiral catalysts has enabled stereodivergent synthesis of a wide range of molecules by simply utilizing different enantiomeric pairs of catalysts.⁹ However, the ability to achieve high diastereoselectivity and enantioselectivity is not always straightforward, as it depends on the degree of catalyst control, substrate control, and interactions between two different catalysts. Overall, the complexity of the mechanisms of these transformations renders reaction design very difficult.

N-Heterocyclic carbene organocatalysis has been widely employed in cooperative dual catalyst systems by combining it with a range of catalysts including Lewis acids, Bronsted acids including chiral phosphoric acids, and others.^10,13 Asymmetric coupling of the NHC-derived nucleophiles with electrophilic π-allyl species generated by transition metal catalysts has also been accomplished. Glorius and coworkers recently demonstrated the enantio- and diastereodivergent synthesis of α,β–disubstituted γ-butyrolactones by employing cooperative dual catalysis using chiral Ir-phosphoramidite/NHC pairs (Scheme 1), one of the first examples of its kind.¹¹

Scheme 1.

cis-Selective formation of lactones from aldehydes and allyl cyclic carbonates using Ir-phosphoramidite and NHC dual chiral catalytic systems.

Although, previous mechanistic DFT studies of dual chiral catalytic reactions involving two metal catalysts or a metal catalyst with an organocatalyst have been reported,¹² there have been no reports calculating the complete mechanisms of the more complex scenarios of a metal catalyst with an N-heterocyclic carbene. Reactions such as these are effective in achieving diastereo/enantioselectivity,¹³ but they raise unique challenges in obtaining mechanistic understanding since the NHC organocatalyst can coordinate strongly to transition metals. Therefore, multiple possibilities must be considered in determining the exact function of the NHC ligand in the catalytic cycle. Understanding these transformations would aid in the development of new stereoselective dual catalyzed reactions.

To gain insight into the mechanism of this transformation (Scheme 1) and specifically to probe the generation and reaction of NHC-enolate nucleophiles and Ir-π-allyl electrophiles, we turned to computational methods. The origin of enantio- and diastereoselectivity is also investigated by modeling different chiral combinations of phosphoramidite ligands and NHC organocatalysts. To explore the stereoselectivity, we selected the combinations of (S)-spiro-phosphoramidite (P) ligand and NHC catalyst (N1 and N2) that exhibited the highest diastereoselectivity and enantioselectivity. The origin of stereodivergence was also explored computationally using (R)-spiro-phosphoramidite (P ent) ligand and NHC catalyst N2.

2. COMPUTATIONAL METHODS

Computations were performed using the Gaussian 16 C. 01 program.¹⁴ The geometries were optimized using the B3LYP-D3 hybrid density functional theory¹⁵ with Pople’s 6-31G** basis set for all atoms. The Stuttgart-Dresden pseudopotential (SDD) basis set consisting of an effective core potential (ECP) for 60 core electrons and a double-ζ quality valence basis set for 17 valence electrons was employed for the iridium atom.¹⁶ Conformational searches were manually performed by comparing the energetics of all possible orientations of the ligand and substrate about the iridium center. All minima and maxima were respectively characterized by the number of imaginary frequencies as 0 and 1. The transition states were verified has a unique imaginary frequency. Intrinsic reaction coordinate (IRC) calculations were carried out to confirm that transition states connecting the reactant and product.¹⁷ The effect of a continuum solvent in toluene, was evaluated using the implicit conductor-like polarizable continuum model (CPCM).¹⁸ The zero-point vibrational energy (ZPVE) and thermal and entropic corrections obtained at 298.15 K and 1 atm pressure derived from the gas-phase computations with B3LYP-D3/6-31G**,SDD(Ir) have been applied to the electronic energies as obtained from the single-point energy evaluations in the solvent phase using CPCM(toluene)/B3LYP-D3/6-311G**,SDD(Ir) to estimate the Gibbs free energies of solutes in the condensed phase. The discussions in the text are presented using CPCM(toluene)/B3LYP-D3/6-31G**,SDD(Ir)//B3LYP-D3/6-31G**,SDD(Ir), unless otherwise specified. Graphical renderings of the optimized geometries were obtained by using a graphical user interface: namely, CYLView.¹⁹

3. RESULTS AND DISCUSSION

We began our computational study of the mechanism of the NHC/iridium dual-catalyzed stereodivergent synthesis of α,β-di-substituted γ-butyrolactones by analyzing each step of the reaction in detail (Scheme 2) using quantum mechanical calculations. The general cycles of both the Ir- and NHC-catalysis are given in Scheme 2. Initially, the active NHC catalyst (NHC*, Scheme 2) is formed from the protonated NHC precursor by means of Et₃N base. The NHC catalyst then activates the α,β-unsaturated carbonyl compound to form the NHC-enolate. Simultaneously, activation of the allyl cyclic carbonate occurs by the active Ir catalyst generates an Ir-π-allyl complex. Following their generation, C-C bond formation between the Ir-π-allyl complex and NHC-enolate occurs. Finally, ring closure generates the product with subsequent catalyst regeneration. Owing to the complexity of the mechanism of this process, our analysis is broken down into five parts including (i) generation of the NHC-enolate from the aldehyde, (ii) formation of the active iridium catalyst, (iii) activation of the allyl cyclic carbonate to generate the Ir–π-allyl electrophile, (iv) the position of decarboxylation from cyclic lactones in the catalytic cycle, and finally (v) stereoselective C-C bond formation between the nucleophilic NHC enolate and the electrophilic Ir-π-allyl species. In section 2.6, we explore the origin of diastereo- and enantioselectivity in this transformation.

Scheme 2.

General mechanism of the Ir-spiro-phosphoramidite/NHC dual-catalyzed lactone formation.

3.1. Generation of the NHC-Enolate.

To begin, the formation of the NHC-enolate and corresponding the relative Gibbs free energy barriers for each step in this process (in the absence of the BF₄− counterion, cationic) are summarized in Figure 1 (see Figure S1 for the energetics of the neutral NHC-enolate formation in the presence of the BF₄− counterion which are largely the same). Since the barriers for the neutral and cationic systems are very similar (presence and absence of counterion), we will discuss only the cationic system for simplicity. First, the active NHC catalyst forms by reaction with triethylamine. Next, C-C bond formation between the α,β-unsaturated aldehyde and NHC catalyst occurs ([1–2]^‡), with an overall barrier of 1.6 kcal/mol to generate intermediate 2. Proton transfer converts intermediate 2 to the Breslow intermediate 3. Finally, concerted β-protonation/deprotonation of 3 leads to NHC enolate 5.

Figure 1.

Relative Gibbs free energy profile (kcal/mol) diagram and corresponding optimized transition states for the formation of NHC-enolate from the NHC and aldehyde calculated with CPCM(toluene)/B3LYP-D3/6-31G**//B3LYP-D3/6-31G**.

Et₃N was found to be intimately involved in a number of these steps, substantially lowering the barriers compared to when it is absent. For example, a concerted vicinal C to O proton transfer from 2 to 3 benefits from the presence of a Et₃NH+ that hydrogen bonds to the oxygen and a second Et₃N that removes the proton thereby lowering the barrier to 14.9 kcal/mol from 42.0 kcal/mol without the Et₃N.²⁰ Pre-coordination of 3 with Et₃NH⁺ leads to 4, which is uphill in energy by only 4.7 kcal/mol, compared to 27.0 kcal/mol with H⁺ alone (see Tables S1 – S2 in the Supporting Information). Finally, a concerted β-proton abstraction by Et₃N and deprotonation from OH to Et₃N proceeds via another low energy transition state [4–5]^‡ with a barrier of 5.7 kcal/mol. Thus, the Et₃NH⁺ plays a crucial role throughout the NHC-enolate formation.

3.2. Formation of Active Ir Catalyst (Ir(P)(NHC)Cl):

The Ir portion of the catalytic cycle was next explored, beginning with the identification of the active catalyst species. Previous DFT^12a and experimental²¹ studies from a different transformation involving and iridium allyl and a Cinchona enolate, found that the active catalyst is Ir(P)₂Cl where P is a BINOL phosphoramidite ligand containing an azepine. The azepine alkene coordinates the Ir and displaces (COD).²¹ For the reaction under consideration here, the NHC can also ligate the iridium as supported by HRMS observation of the resultant complex.¹¹ Consequently, we have computed the energy of different forms of the putative active catalyst both before and after binding the substrate as shown in Scheme 3 (See Figure S2 in SI for a conformational study of NHC ligand around Ir atom). From the [Ir(COD)Cl]₂ precatalyst 6a, the spiro phosphoramidite P can exchange with the COD ligand to form the monomer 6b (0.0 kcal/mol). The allyl cyclic carbonate can bind to 6b (prior to exchanging the P for the NHC ligand) to give the pentacoordinate Ir complex 6d (−14.2 kcal/mol). Alternately, 6b can exchange a spiro phosphoramidite P with an NHC to form the lower energy square planar complex 6c (−24.4 kcal/mol), which can in turn bind the substrate to give 6e (−19.1 kcal/mol).

Scheme 3.

Possible routes for the formation of Ir-NHC complex 1e with substrate. The relative Gibbs free energies are reported from CPCM(toluene)/B3LYP-D3/6-31G**/SDD(Ir)//B3LYP-D3/6-31G**/SDD(Ir).

3.3. Activation of the Allyl Cyclic Carbonate by Ir:

The formation of the Ir-π-allyl intermediate from the allyl cyclic carbonate can occur through direct or stepwise elimination of CO₂ (C-O bond breaking followed by CO₂ elimination, see Scheme S1 and Figure S3). Since direct elimination of CO₂ in a concerted manner is disfavored energetically based on our computations (barrier of 45.0 kcal/mol, see SI Figure S3 for full details), we explored a stepwise elimination of CO₂ via ring opening of allyl cyclic carbonate intermediate followed by subsequent CO₂ extrusion. The unassisted ring opening to allylic carbonate has a barrier of approximately 30.0 kcal/mol. Incorporation of Et₃NH⁺ (generated from NHC formation) reduces the barrier²² further to 13.8 kcal/mol due to interactions with the carbonate oxygen (Scheme 4, see SI Scheme S1).

Scheme 4.

Formation of Ir-allyl-carbonates from allyl cyclic carbonates using Ir-P-NHC catalyst in the presence of a Bronsted acid (protonated triethylamines = Et₃NH⁺) generated during the formation of the NHC catalyst in the presence of Et₃N calculated with CPCM(toluene)/B3LYP-D3/6-31G**/SDD(Ir)//B3LYP-D3/6-31G**/SDD(Ir). C2 is labeled and highlighted with a gray circle.

This stereoconvergent reaction occurs via a common π-allyl intermediate that has been observed experimentally in an X-ray crystal structure.²¹ This intermediate could form in one of two ways. One stereoisomer could lead to the major intermediate (Ir-endo-π-allyl) and one to the minor intermediate (Ir-exo-π-allyl) followed by interconversion of exo to endo through a π-σ-π slippage process (“endo” refers to the anti-orientation of the Ir-Cl and C2-H of the alkene or ally; see Figure S4). Since previous DFT^12a studies have shown that this type of interconversion is a higher energy process than reversible formation of the π-allyl, efforts focused on the latter possibility. Namely, the R- and S-enantiomers of the racemic substrate (6) could each form two different conformers (endo vs exo) leading to the four adducts for 7 shown in Scheme 4. Each of these could react independently with the endo versions of 7 leading to the endo version of intermediate 8 and the exo version of 7 leading to the exo version of intermediate 8 as shown in Scheme 4.

Calculations revealed that the first chiral recognition event in this reaction occurs between [Ir(Cl)(P)NHC] catalyst and the allyl cyclic carbonate (S vs R) en route to coordinated adducts 7. The ligand creates a shape-selective chiral pocket that limits the conformations that the substrate can adopt once bound to the iridium center. If the endo approach where the C2–H of the allyl cyclic carbonate is anti to the Ir–Cl occurs, the Si face of the alkene is exposed for later reaction with a nucleophile (see Table S3 for space-filling models of catalyst and substrate). The corresponding syn-orientation of the allyl C2–H and Ir–Cl (exo approach) leaves the Re face exposed for reaction with a nucleophile. Overall, endo coordination leads to lower energy coordination adducts of 6 and to lower energy pathways to the π-allyl intermediate 8 (Figure 2). Within the endo pathway, the barriers for breaking the C-O bond from the S-substrate (13.8 kcal/mol) and the (R)-substrate (13.6 kcal/mol) are similar. The interaction energies (kcal/mol)²³ between the allyl-cyclic carbonate and Ir-(S)-spiro-phosphoramidite are most favorable for the lowest energy transition states: (S)-endo (−9.3), (R)-endo (−7.0), (S)-exo (−5.8), (R)-exo (0.0). The ability of the chiral catalyst to undergo stabilizing interactions with the substrates appears to be the dominant factor in determining outcomes rather than distortion (see Table S4 in the SI for further energy decomposition analysis). While the reaction of the S-substrate is endothermic and that of the R-substrate is exothermic, both adducts ultimately convert to the same conformation of Ir-endo-π-allyl (i.e. 8) allowing accumulation of the single intermediate as observed experimentally. Thus, control of the configuration of the phenyl bearing carbon (i.e., β-carbon) in the product is determined in this step, which in turn, depends on the chirality of the phosphoramidite involved.

Figure 2.

Relative Gibbs free energy (kcal/mol) profile and corresponding transition state geometries of the formation of 8 from the racemic allyl cyclic carbonate 6 with Ir-(S)-spiro-phosphoramidite calculated with CPCM(toluene)/B3LYP-D3/6-31G**/SDD(Ir)//B3LYP-D3/6-31G**/SDD(Ir). The line-angle version shows transition states.

3.4. Determining Decarboxylation Point:

After forming the Ir-πallyl intermediate 8, this species will then react with the NHC-enolate intermediate, both interacting through their prochiral faces, leading to the creation of two chiral carbon centers. Theoretically, this C-C bond formation can happen between the NHC enolate and the Ir-π-allyl-carbonate prior to CO₂ elimination (path A, red, Scheme 5) or between the NHC enolate and Ir-π-allyl-alkoxide intermediate formed by CO₂ elimination (path B, blue, Scheme 5). Glorius and coworkers suggested that the reaction follows path B in which C-C bond formation occurs between the NHC-enolate and the Ir-π-allyl-alkoxide ion⁸ (after decarboxylation, see Scheme S2) based on HRMS data suggesting the presence of an Ir-π-allyl-alkoxide ion intermediate. However, it is possible that CO₂ loss might have occurred during the mass spectrometry measurement. Consequently, we have computed the energetics of the C-C bond formation for both pathways A and B (Scheme 5).

Scheme 5.

Reaction paths from Ir-π-allyl 8 to the lactone product 13.

Both pathways commence from Ir-π-allyl 8 and converge on 12 to yield the same final lactone product. An exhaustive conformational search was performed to locate the lowest energy states for each pathway. In path A (red), the NHC-enolate adds to the Ir-π-allyl 8 prior to extrusion of CO₂ to generate adduct 10. Decoordination of the Ir catalyst from the resultant π-bond generates 11. Decarboxylation at this stage generates alkoxide 12. The alkoxide moiety in 12 adds to the acyl-NHC and subsequent elimination of the NHC catalyst yields the product lactone 13. In path B (blue), decarboxylation occurs from Ir- π-allyl 8 to generate 15 which undergoes addition with the NHC-enolate to generate 17. Decoordination of the Ir catalyst from the resultant π-bond converges onto 12.

In both of these pathways, Et₃NH⁺ plays a crucial role in lowering the energy of the transition states through hydrogen bonding. The Et₃NH+ does so by interacting with different oxygen atoms of carbonate group, as shown in Figure 3/Scheme 5. Based on the energy profile of path A, it is favorable for C-C bond formation to occur from the Ir-π-allyl 8, prior to CO₂ elimination, as this process only has a barrier of 8.6 kcal/mol ([9–10]^‡, Figure 3). Comparatively, CO₂ elimination prior to C-C bond formation (pathway B, see Scheme S2 for full details on this path) is disfavored, as it has a higher barrier (20.7 kcal/mol from 8). The greater energy associated with the CO₂ elimination in path B arises from distortion in the allyl induced by the Ir-P-N1 portion (distortion energy = 17.5 kcal/mol, see Table S5 for full details). In contrast, the distortion in the CO₂ elimination from path A is much smaller (distortion energy = 9.5 kcal/mol). Further, C-C bond formation in path A to generate 10 is thermodynamically favorable while CO₂ extrusion in path B to form 15 is endergonic. Thus, we conclude that C-C bond formation occurs first between the Ir-π-allyl 8 and the NHC enolate followed by CO₂ elimination (path A, red, Figure 3).

Figure 3.

Relative Gibbs free energy profile diagram and corresponding transition states of the formation of γ-lactones from Ir-π-allyl 8. All energy values were calculated with CPCM(toluene)/B3LYP-D3/6-31G**/SDD(Ir)//B3LYP-D3/6-31G**/SDD(Ir).

3.5. Origin of Stereoselectivity:

With the CO₂ elimination ascertained, we then investigated the determinants that establish the two stereocenters found in of the final lactone product 13 (Scheme 5 and Figure 3). The NHC enolate first engages Ir-π-allyl 8 to generate a pre-reaction complex 9. Reaction from this complex via transition state [9–10]^‡ forms the two chiral centers in allylated product (10). The origin of stereoselectivity in this step is complex as two chiral catalysts are employed, the chiral NHC on the enolate portion and the chiral Ir-(S)-spiro-phosphoramidite (P) on the π-allyl portion. To understand the key control factors, two different NHC catalysts having the same configuration (N1 and N2 = 5R, 6S-NHC) were explored with Ir-(S)-spiro-phosphoramidite (P) in this stereodetermining step.

The computed relative energies of the C–C bond forming transition states across all the stereochemically distinct modes for all catalyst combinations are provided in Table 1 (see Table S6 for information on the choice of single point method). A conformational and configurational study of both P-N1 and P-N2 combinations was undertaken (Scheme S3 and Tables S7 – S8 in the SI). For each catalyst combination, the values are reference to the lowest energy transition state (0.0 kcal/mol). The addition of the Si-face of the NHC–enolate to the Si-face of the Ir-π-allyl species is lowest in energy for both catalyst combinations (P-N1 and P-N2). This conformation will lead to a final product with a (3R, 4S) configuration as was observed experimentally. Addition of the Re-face of the NHC-enolate to the Re-face of the Ir-π-allyl intermediate, which forms the enantiomeric product (3S, 4R), is 5.1–8.6 kcal/mol higher in energy, which leads to an enantiomeric excess of >99% in favor of the (3R,4S) product, again in accord with the experimental findings. The diastereomeric combinations Re-Si and Si-Re are 1.7–1.5 kcal/mol and 10.6–10.2 kcal/mol higher in energy respectively. Using a Boltzmann distribution, these values translate to 95:5 and 93:7 diastereomeric ratios for P-N1 and P-N2, respectively, which are a good match to the experimental values of >95:5 and 91:9.

Table 1.

Relative Gibbs Free Energies (kcal/mol) for the C–C Bond Forming Transition States between the Prochiral Faces of the Nucleophile (Nu) and the Electrophile (El). Obtained with CPCM(toluene)/B3LYP-D3/6-31G**/SDD(Ir)//B3LYP-D3/6-31G**/SDD(Ir) for the Catalyst Combinations (S)-Phosphoramidite (P), (R)-Phosphoramidite (P_ent) and NHC (N1/N2).

NHC-enolate (Nu)	Ir-allyl (El)	Product	Rel ΔG (kcal/mol) P-N1		Rel ΔG (kcal/mol) P-N2		Rel ΔG (kcal/mol) Pent-N2
Si	Si (endo)	3R, 4S	0.0		0.0		1.4
Re	Si (endo)	3S, 4S	1.7		1.5		9.5
Re	Re (exo)	3S, 4R	5.1		8.6		9.3
Si	Re (exo)	3R, 4R	10.6		10.2		0.0
			DFT	Expt	DFT	Expt	DFT	Expt
		dr	95:5	>95:5	93:7	91:9	9:91	24:76
		ee (%)	>99	>99	>99	>99	>99	>99

On the other hand, when the (R)-phosphoramidite (P_ent) is paired with N2, the diastereomeric product with a (3R, 4R) configuration is formed from addition of the Si-face of the NHC-enolate to the Re-face of the Ir-π-allyl intermediate. Analogously, the enantiomeric product with a configuration of (3S, 4S) is formed from addition of the Re-face of the NHC-enolate to the Si-face of the Ir-π-allyl intermediate. This Re-Si transition state is 9.5 kcal/mol higher in energy than the Si-Re transition state (Table 1), giving an enantiomeric excess of >99% in agreement with experimental observations.

The stereocontrolling transition states were analyzed in detail to gain insight into the origin of enantioselectivity in this transformation. Examination of the non-covalent interactions in the transition states through NCI analysis²⁴ and AIM analysis²⁵ is very complex due to the large size of the system leading to numerous interactions (see Figures S5 – S8 in the SI). Limiting the NCI analyses to the interfacial region (see Figure S6 in the SI) between the NHC enolate and the Ir-π-allyl revealed two main regions of interaction (Ar/Ph and carboxylate/triazolium) as shown in Figure 4.

Figure 4.

The stereocontrolling C-C bond forming transition states between the Ir-π-allyl and NHC-enolate for the P-N1, P-N2 and P_ent-N2 catalyst combinations showing the favorable interactions between the Ir-π-allyl and NHC-enolate components in the Si-Si case. See Figure S9 for electrostatic potential maps highlighting electrostatic interactions.

In the lowest energy Si-Si transition state for both P-N1 and P-N2, the phenyl group (blue) of the NHC-enolate is further away from the aryl group (red) of the NHC bound to the iridium center. In contrast, in the higher energy Re-Si and Re-Re transition states, the NHC-enolate phenyl group (blue) crowds the aryl group (red) of the Ir-NHC.²⁶

To further identify the contributing factors to the enantioselectivity, a distortion/interaction analysis was performed as described by Houk and Bickelhaupt¹⁵ (Figure 5, see SI Table S9). The interaction between the two reacting partners (i.e., Ir-π-allyl and NHC-enolate) is computed by comparing the energy of transition state structure to that of the two separate components. In the case of P-N1 (top of Figure 5), the Si-Si transition state has a strong interaction energy and a distortion energy that offsets about half of the interaction energy. For the Re-Si, both the distortion and interaction energies are significantly smaller with a greater effect upon the latter that leads to a less stabilized species. For the Re-Re, the interaction energy is similar to that of the Si-Si, but the distortion energy is substantially larger. Using identical C-C bond forming distances reveals similar trends (see SI Table S10).

Figure 5.

Distortion-interaction analysis (kcal/mol) of the C–C bond forming transition states with the P-N1 and P-N2 catalyst combinations [B3LYP-D3/6-31G**,SDD(Ir)].

In the case of P-N2, there are similar interactions amongst the transition states, leading to similar interaction energies (Figure 5, bottom). The difference in energy among the transition states arises mainly from the distortion energies. In particular, the distortion in the Ir-π-allyl fragment increases across the series (see SI Figure S10 for overlaid structures of the optimized ground states and the three lowest energy transition states showing the distortion).

An energy decomposition analysis (EDA) was undertaken for the interaction energies using the second-generation absolutely localized molecular orbitals²⁷ (ALMO-EDA) method implemented in Q-Chem 5.0²⁸ and described by Liu²⁹ [computed with the HF/6-311G(d,p)]. The resultant values (Table 2) further differentiated the transition states permitting specific interactions to be identified.

Table 2.

Energy decomposition analysis of the interaction energies for C–C bond forming transition states (energies in kcal/mol).³⁰

		Electrostatic	Polarization	Charge Transfer	Dispersion	Pauli
P-N1	Si-Si	−101.0	−18.2	−38.8	−52.5	132.8
	Re-Si	−82.6	−14.8	−28.2	−46.7	115.7
	Re-Re	−92.4	−17.5	−33.3	−43.4	110.3
	Si-Re	−65.1	−9.9	−22.5	−44.8	92.9
P-N2	Si-Si	−64.0	−10.4	−22.4	−41.5	88.7
	Re-Si	−68.0	−11.5	−23.4	−36.2	92.1
	Re-Re	−66.6	−11.0	−22.1	−37.4	86.9
	Si-Re	−65.6	−10.1	−25.0	−36.7	98.6

For both the P-N1 and P-N2 cases, the dispersion component is strongest for the Si-Si approach in the transition states (−52.5 and −41.5 kcal/mol, respectively and arising mainly from the mesityl portion (both aryl and methyl groups) of the NHC bound to the Ir interacting with the benzyl group of the NHC enolate. The indane portion of the NHC does not contribute significantly (see Figures S11 – S12 for further discussion of dispersion in P-N2).

A view of the P-N1 transition consistent with this difference in dispersion interactions is shown in Figure 6. Namely, the NHC enolate (tube model) has a larger surface area of interaction with the hydrophobic portions of the well-defined cleft of the Ir-π-allyl (space-filling model) of the Si-Si transition compared to the Re-Si and Re-Re transition states. For the P-N1 Si-Si approach, this benefit incurs significant steric interactions (Pauli term) relative to the other approaches indicating that the fit in the cleft is not optimal. For the P-N2, it appears that the complementarity is better as judged by a more favorable dispersion for Si-Si and a Pauli steric term that is close to the minimum for the system.

Figure 6.

Optimized geometries (space-filling model = Ir-π-allyl; stick model = NHC-enolate) of the stereocontrolling C-C bond formation transition states between Ir-π-allyl and NHC-enolate in the case of the P-N1 catalyst combinations.

The remaining three terms (electrostatic, polarization, charge transfer) relate to components involving charges. These components appear to be the primary differentiators of three approaches for P-N1 with the strongest interactions being afforded the Si-Si, the second strongest to the Re-Re, and the weakest to the Re-Si transition state. These interactions are in turn modulated by the conformations imposed by the chiral ligands which limits how the substrates can approach each other.

Within the structures two sets of features could be identified that could account for these interactions. The first of these interactions centers on the carboxylate of the π-allyl (Figure 4, Figure 5). In the Si-Si transition state, this carboxylate is perpendicular to the triazole of the NHC-enolate allowing the terminal negatively charged oxygen atoms of the carboxylate group to interact with the positively charged triazole ring. In the Re-Si transition state, the carboxylate is parallel to the triazole and the average distances between the oxygens and the triazole ring are larger. In the Re-Re transition state, only one of terminal oxygen atoms of the carboxylate is oriented toward the triazolium and the average distances are again larger. In the Si-Re transition state, this interaction is absent (see Figure S13).

The second of these interactions centers on the cationic allyl and the NHC-enolate oxygen (Figure 4, Figure 6). This interaction is present in both the Si-Si and Re-Re transition states, but is absent in the Re-Si transition state. All told, the Si-Si transition state benefits from optimal carboxylate-triazolium interactions and enolate-π-allyl interactions while the Re-Si transition only utilizes the former and the Si-Si transition has a much weaker version of the latter.

3.6. Origin of Stereodivergence:

To investigate the origin of stereodivergence, we compared the energetics of both (S) and (R)-phosphoramidites (P and P_ent) with one NHC catalyst (N2). Both the P-N2 and P_ent-N2 stereoselectivity results are given in Table 1. We used distortion-interaction analysis to first probe the origin of stereoselectivity of P_ent-N2 (Table 3, see Table S11 for energy decomposition analysis). The lowest energy (Si-Re) transition state has lower distortion energies from both the Ir-π-allyl electrophile (10.0) and NHC-enolate nucleophile (6.8) fragments. Similarly, it has a greater interaction energy (−67.4), leading to the most favorable overall energy (E_ASE = −50.6) compared to the diastereo- and enantiomeric pairs, Si-Si (−49.6) and Re-Re (−41.9) as shown in Table 3.

Table 3.

Distortion-Interaction analysis of P_ent-N2.

Nu-El	Distortion			Interaction	EASE
	Ir-allyl	NHC-enolate	Total
Si-Re	10.0	6.8	16.8	−67.4	−50.6
Si-Si	12.6	6.4	19.0	−68.6	−49.6
Re-Si	16.1	12.4	28.5	−70.4	−41.9
Re-Re	8.4	10.2	18.6	−59.8	−41.2

Furthermore, we have analyzed the optimized transition state geometries of the P_ent-N2 catalyst combination (Figure 7). The diastereomeric Si-Re (0.0) and Si-Si (1.4) transition states have similar non-covalent interactions. The lower energy Si-Re transition states which leads to the corresponding product with a (3R, 4R) configuration has electrostatic interactions between carboxylate oxygen atoms of the Ir-allyl and the triazole of the NHC-enolate (Figure 7). However, the higher energy Si-Si transition state displays the same electrostatic interaction between the carboxylate and triazolium ring as well as C-H^...p and C-H^...O interactions. The lower distortion energy in the Si-Re transition state leads to a greater overall stabilization (E_ASE = −50.6, Table 3). In the case of the enantiomeric pairs Si-Re (0.0) and Re-Si (9.5), the lower energy Si-Re transition state has a greater contribution in the form of electrostatic interaction between carboxylate and triazolium ring of the NHC-enolate than in the Re-Si geometry (see Table S12 for a comparison of the Si-Re TS for P_ent-N2 and P2-N2).

Figure 7.

Optimized geometries (space-filling model = Ir-π-allyl; stick model = NHC-enolate) of the stereocontrolling C-C bond formation transition states between Ir-π-allyl and NHC-enolate in the case of the P_ent-N2 catalyst combinations.

Overall based on structural analysis, NCI plots, AIM analysis, interaction/distortion analysis, and EDA, the excellent diastereo/enantioselectivity observed in this transformation arises from two chiral catalysts interacting through docking of the NHC enolate into a chiral pocket on the Ir-π-allyl adduct. The most favorable fit occurs in the Si-Re assembly which optimizes charged interactions between the approaching components.

CONCLUSION

Herein, we have provided detailed proposals for the mechanism and origin of stereoselectivity of a cooperative asymmetric catalysis between Ir-spiro-phosphoramidite and N-heterocyclic carbenes (NHC) for the reaction of allyl-cyclic carbonates and unsaturated aldehydes using dispersion-corrected density functional theory. In this study, we found a mechanistic route for the activation of aldehydes by N-heterocyclic carbenes (NHC) which gives the NHC-enolate nucleophile as well as the stereoconvergent formation of a major Ir-π-allyl intermediate (electrophile).

Figure 8 summarizes the lowest energy overall pathway showing first the NHC-enolate formation followed by the intersection of both catalytic cycles in which Ir-π-allyl and NHC-enolate undergo C-C bond formation. The cyclic carbonate binds to the Ir-active species (7) and undergoes C-O bond cleavage via [7–8]^‡ to give the Ir-allyl intermediate (8). This species 8 then reacts with the NHC-enolate through the C-C bond formation transition state [9–10]^‡ to form two chiral centers in the allylated C-C bond product (10). Next, removal of Ir-complex gives the allylated C-C bond product of NHC catalyst (11) followed by facile CO₂ elimination [11–12]^‡ (5.9 kcal/mol barrier) to form 12. Finally, ring closur occurs via [12–13]^‡ (barrier of 4.7 kcal/mol) during which the allylated lactone product is also released from the NHC catalyst.

Figure 8.

Relative Gibbs free energy profile diagram showing the mechanistic proposal of γ-lactone formation from aldehydes and allyl cyclic esters using NHC and Ir-phosphoramidites.

The main conclusions are:

• Formation of the NHC is rate-limiting while C-C bond formation stereodetermining. The energetic span (TDTS-TDI) of the allyl forming portion is only 10.7 kcal/mol (C-O bond breaking transition state). This energetic span is even lower than that of the NHC-enolate cycle (14.9 kcal/mol) as shown in the complete energy profile (Figure 8). Thus, the overall energetic span of the entire transition metal- and organocatalytic cycles is 14.9 kcal/mol (for the Breslow intermediate formation).

• Both carbonates converge onto one π-allyl intermediate. The (R) and (S)-cyclic carbonates with endo-coordination to the Ir-phosphoramidite lead to the formation of the lower energy Ir-π-endo-allyl-carbonate.

• In both catalytic cycles, Et₃N and Et₃NH⁺ (generated from the NHC-H and Et₃N) play a crucial role. For the NHC-enolate formation, the base and conjugate acid facilitate the protonation/deprotonation processes need to generate the enolate. For the Ir-π-allyl, Et₃NH⁺ acts as a Brønsted acid facilitating cleavage of the cyclic carbonate. The Et₃NH⁺ stabilizes the resultant carboxylate such that decarboxylation is deferred to after Ir decoordination and immediately prior to lactone formation. Within this decarboxylation transition state, inclusion of the Et₃NH⁺ reduces the barrier.

• Calculations reveal a lower energy pathway for CO₂ elimination occurring after C-C bond formation rather than before. The resultant carboxylate plays important roles in the stereochemical determining event.

• For the stereodetermining C-C bond formation stage, we have modeled two different (5R, 6S)-NHC catalysts with Ir-(S)-spiro-phosphoramidite (as well as Ir-(R)-spiro-phosphoramidite with NHC-N2). Computational predictions of both enantio- and diastereoselectivities match with experimental stereoselectivity values show casing the utility of electronic structure calculations for studying complex chiral catalysts (∼210 atoms) in asymmetric reactions.

• Analysis using several tools reveals the stereochemistry determining features which arise from multiple points of interaction encompassing both dispersion recognition, strain deformation, and charged interactions.

• The NHC plays two roles acting as a ligand on the Ir center and forming the NHC enolate. The primary interaction from the NHC ligand on the Ir with the incoming NHC enolate occurs from the N-Ar group on the NHC bound to the Ir center. However, the portion with the stereochemistry is oriented away from the incoming NHC enolate. To determine whether a synergistic interaction occurs between the chiral portion of the phosphoramidite and the chiral NHC bound to the Ir, calculations were undertaken with an achiral NHC ligand bound to the Ir (Scheme 6). For this combination, the energy differences between the transition states were actually larger than found experimentally (see Table 1). Examination of three-dimensional models revealed more contacts between the chiral NHC and the phosphoramidite (Figure 9). It appears that the relief of steric interactions stabilizes the Si-Si approach and that the chiral NHC on the Ir actually reduces enantioselectivity. Overall, the chiral NHC contributes to the stereoselectivity through the NHC-enolate and not as a ligand at the Ir center.

Scheme 6.

Effect of achiral NHC ligand on stereoselectivity step.

Figure 9.

Interactions of chiral and achiral NHC ligand in the C-C bond forming transition state [9–10]^‡ in Figure 7.

In closing, calculations provide the overall reaction profile and also reveal how the different asymmetric elements interact in this dual catalyst system. Even for large systems, insights can be gained into the how the multiple control elements operate and the key enantio-controlling interactions.