Mechanistic Basis for the Iridium-Catalyzed Enantioselective Allylation of Alkenyl Boronates

Abstract

Iridium(phosphoramidite) complexes catalyze an enantio- and diastereoselective three-component coupling reaction of alkenyl boronic esters, organolithium reagents and secondary allylic carbonates. The reaction proceeds through an allylation-induced 1,2-metalate shift of the alkenyl boronate to form non-adjacent stereocenters. Mechanistic investigations outline the overall catalytic cycle and reveal trends in reactivity and selectivity. Analysis of relative stereochemistry in products derived from a variety of 1,1-disubtituted alkenyl boronates provides insight into the transition state of the addition and indicates a concerted pathway. Kinetic analysis of the reaction revealed the kinetic order dependance in boronate, catalyst, and both the slow- and fast-reacting enantiomer of allylic carbonate as well as the turnover-limiting step of the reaction. Determination of nucleophile-specific parameters N and s_N for alkenyl boronate complexes enabled comparison to other classes of nucleophiles. DFT calculations indicate the addition of the alkenyl boronate to the cationic Ir(π-allyl) intermediate and the 1,2-metalate shift occur in a concerted mechanism. The stereoselectivity is determined by ligand–substrate steric repulsions and dispersion interactions in the syn addition transition state. Hammett studies supported the computational results with regard to electronic trends observed with both aryl-derived alkenyl boronates and aryl carbonates.

Graphical Abstract

INTRODUCTION

Three-component coupling reactions allow for the rapid construction of densely functionalized compounds from readily available starting materials.^1,2,3,4,5 Furthermore, the ability to independently vary multiple reactants provides access to broad chemical diversity in an atom economical fashion. Enantioselective variants of three-component couplings allow for the concise preparation of complex molecular architectures in a stereoselective manner.^6,7 Within the field of three-component couplings, alkenyl boronic esters hold a privileged position.^8,9 The empty p orbital on boron allows for pre-complexation with organometallic reagents to form boronates, which can then engage in bimolecular reactions through a 1,2-metalate shift. The resulting organoboronic esters provide a highly functionalizable handle that can be utilized in an array of stereospecific transformations.¹⁰ Likewise, alkenyl boronic esters are widely available. Many are commercially available, and their derivatives typically display good air and moisture stability that allows for ease of preparation and isolation.^{11,12,13,14,15}

Alkenyl boronate complexes can participate in three-component couplings through two related mechanistic manifolds. They can engage with electrophiles in a concerted mechanism, wherein an external electrophilic activator promotes migration of the organic fragment onto the adjacent sp²-hybridized carbon (Scheme 1a).^{16,17,18,19,20} Alternatively, the olefin can act as a radical acceptor to generate the α-boryl radical. Subsequent single electron oxidation can generate a carbocation adjacent to the boronate, which is quenched through a 1,2-metalate shift.²¹

Scheme 1.

1,2-Metalate Rearrangements and Ir-Catalyzed Allylation Reactions

Stereoselective 1,2-metalate shifts of alkenyl boronate complexes frequently involve anti addition across the olefin in which the electrophile and the migrating group add to opposite faces of the olefin. This phenomenon was exemplified by Denmark’s Lewis base-catalyzed carbosulfenylation of alkenyl boronates (Scheme 1b).²² The generation of an enantioenriched thiiranium intermediate resulted in a diastereospecific 1,2-migration providing the 1,2-anti-sulfeno-functionalized product from the trans-1,2-alkenyl boronate. Additionally, π-acidic late transition metal complexes can induce a 1,2-metalate shift as showcased in Morken’s conjunctive cross coupling reaction.²³ Vinyl boronates engaged a Pd(aryl)⁺ intermediate using a chiral nonracemic ligand that adequately discriminated the enantiotopic faces of the vinyl group, thereby rendering the process highly enantioselective. Further investigations of the relative stereochemistry using deuterated alkenylboronate complexes revealed an anti addition of the arene group and the Pd(aryl)⁺ complex across the-olefin (Scheme 1c). In related studies, our group developed an enantioselective three-component coupling reaction of lithiated indoles, organoboronic esters and allylic acetates in the presence of a palladium catalyst to prepare indoline derivatives with adjacent stereocenters.^24,25 This reaction appeared to proceed via a stepwise allylation-migration process wherein the Pd(π-allyl) intermediate exerts high facial selectivity in the allylation event, followed by predominantly anti-1,2-migration with selectivity dictated by the new C3 stereocenter (Scheme 1d). Alternatively, the Aggarwal group reported an example of selective syn addition in the formation of β-fluoroboronic esters wherein the phenyl and F groups are added to the same face of the olefin (Scheme 1e).⁹

To extend our previous investigations of electron-rich indole boronates (Scheme 1d), we sought to determine if less nucleophilic alkenyl boronates could engage metal(π-allyl) intermediates via an allylation event. In this context, iridium-catalyzed allylic substitution reactions accommodate various nucleophiles and electrophiles to provide branched allylic products in good regio- and enantioselectivity.²⁶ More specifically, the Carreira group has described multiple iridium-catalyzed allylic substitution reactions using phosphoramidite ligand L1, which was developed in their laboratory.^27,28,29,30 They have shown that secondary allylic alcohols engage with 1,1-disubstituted olefins and allyl silanes^31,32 in highly enantio- and regioselective substitution reactions (Scheme 1f). During mechanistic investigations, the Carreira group found that both enantiomers of the racemic alcohol react to form the same ratio of diastereomeric iridium(π-allyl) intermediates through an acid-promoted oxidative addition.³³ They postulated that rapid isomerization of the two diastereomeric allyl complexes occurs to provide the same outcome when using either enantiomer of the allylic alcohol. The nucleophilic attack on these diastereomeric iridium(π-allyl) complexes frequently occurs with high enantioselectivity, but the origin of this selectivity is still poorly understood. In separate studies, the You group investigated a time-dependent enantiodivergent synthesis using racemic allylic carbonates (Scheme 1g).³⁴ They discovered that under two separate reaction conditions, one catalyst can be used to sequentially promote two kinetic resolution reactions. These protocols provided access to either enantiomer of product using the same enantiomer of a chiral catalyst.

Our initial studies on the iridium-catalyzed enantioselective allylation of alkenyl boronates examined a three-component coupling of alkenyl boronic esters, organolithium reagents and secondary allylic carbonates.³⁵ This transformation relied on a kinetic resolution of the racemic allylic carbonate wherein one enantiomer reacted preferentially to form the Ir(π-allyl) intermediate, which then engaged the alkenylboronate species to provide the desired coupling product (Scheme 1h). Initial mechanistic investigations demonstrated that ligand (S)-L1 controlled the stereochemical course of the reaction through effects on both the oxidative addition step and the alkenyl boronate addition step. Likewise, analysis of the relative stereochemistry of products derived from trisubstituted boronate 1a revealed a surprising conclusion that the reaction proceeded via syn addition of the migrating alkyl group and the allyl fragment across the olefin (Scheme 2). From these initial observations, we sought to further investigate the mode of addition for the 1,1-disubtitued alkenyl boronate complexes. Key mechanistic questions included: (1) Do all classes of alkenyl boronates react through a syn addition across the olefin? (2) What is the kinetic order in boronate, catalyst and allylic carbonate? (3) Does the 1,2-metalate rearrangement involve a concerted or stepwise mechanism? (4) What are the stereo-determining and turnover-limiting steps of the reaction? (5) What is the origin of relative and absolute stereocontrol?

Scheme 2.

Diastereoselectivity of Trisubstituted Alkenyl Boronates^a

^aReaction was conducted on a 1.0 mmol scale using 2-bromo-2H-chromene (1.0 mmol), s-BuLi (1.0 mmol), 2,4,4,5,5-pentamethyl-1,3,2-dioxaborolane (1.0 mmol), [Ir(COD)Cl]₂ (2.5 mol%), (S)-L1 (10 mol%), (±)-2a (2.1 mmol) and LiCl (1.0 mmol). 3b was isolated as a single diastereomer after subsequent hydroboration-oxidation was conducted on a 0.3 mmol scale using 3b (0.3 mmol), BH₃·DMS (0.45 mmol). The borane was oxidized to the corresponding diol using NaOH and H₂O₂.

RESULTS AND DISCUSSION

Relative configuration of the 3-component coupling.

We previously reported the X-ray crystal of 3a, which provided experimental evidence of a syn addition for trisubstituted alkenyl boronates.³⁵ Initial ¹H NMR data indicated that diastereomer 3b also arose from syn addition, but the data were not definitive. We discovered that hydroboration-oxidation of 3b gave crystalline diol 4, which was subjected to X-ray crystallographic analysis. The X-ray structure provided concrete evidence that syn addition is conserved across both diastereomers (Scheme 2). We concluded that trisubstituted boronates react with low facial selectivity on the alkenyl boronate but with high syn selectivity and enantioselectivity.

To gain insight into the relative configuration of products derived from 1,1-disubstituted alkenyl boronates, a variety of monodeuterated alkenyl boronic esters were synthesized (Scheme 3, 4) and subjected to the asymmetric allylation. To determine the relative stereochemistry of the resulting tertiary boronic esters, they were first oxidized to the tertiary alcohol and subsequently cyclized via an iodoetherification to provide the tetrahydrofuran derivatives (see Supporting Information for details). We first explored alkyl migrants onto aryl-substituted olefins under our optimized reaction conditions. The Z-d-boronate 1b featured methyl as the migrating fragment and delivered product 5a in 85% yield and 8:1 diastereoselectivity (Scheme 3a). After further derivatization to tetrahydrofuran 6a, irradiation of the C2 methylene hydrogen showed NOE signals corresponding to the phenyl substituents at both adjacent carbons. For confirmation, the E-isomer of 1b was also prepared as boronate 1c to provide 5b in similar yields and diastereoselectivity. Product 5b is the C2 epimer of 5a, which upon NOE analysis of 6b now showed correlations to the benzylic methine and the methyl group at adjacent carbons. Likewise, boronate 1d was prepared where phenyl now acts as the migrating group onto an alkyl-substituted olefin providing 7 in 87% yield with high diastereoselectivity (Scheme 3b). The isolated product 8 showed NOE correlations between the C2 methylene hydrogen and both the benzylic hydrogens and arene group at adjacent carbons, comparable to 6a (see Supporting Information for details). In all cases, the minor diastereomer could not be isolated due to fleeting quantities. From results of alkyl-aryl derived boronates (i.e., an aryl group present either as the migrating group or as an olefin substituent), we conclude that alkenyl boronate addition to the Ir(π-allyl) proceeds with high facial selectivity for the enantiotopic faces of the olefin (8:1 - >15:1) and a syn-selective addition of the migrating organic fragment and the allyl fragment across the olefin. The syn-selectivity is suggestive of a concerted pathway in these cases.

Scheme 3.

Stereochemical Analysis of 1,1-Dibsubtituted Alkyl-Aryl Derived Alkenyl Boronates^a

^aReactions were conducted in THF (0.2 M) on a 0.3 mmol scale. 1b, 1c, and 1d were preformed using organolithium (0.3 mmol), monodeuterated alkenyl boronic ester (0.3 mmol) and LiCI (0.3 mmol), Et₂O, 0 °C–r.t., 30 min then redissolved in THF. [lr(COD)CI]₂ (2.5 mol%), (S)-L1 (10 mol%) and carbonate (0.3 mmol) were added. Yields of the isolated and purified materials after oxidation by treatment of boronic ester with NaOH and H₂O₂. The dr was determined by ¹H NMR. The er of the major diastereomer was determined by HPLC analysis. Relative stereochemistry was determined by NOE analysis after cyclization via iodoetherification using K₂CO₃ (3.0 equiv) and iodine (2.0 equiv), CH₃CN, 0 °C-r.t., 2 h.

Scheme 4.

Stereochemical Analysis of 1,1-Dibsubstituted Alkyl-Alkyl Derived Alkenyl Boronates

^aReaction was conducted in THF (0.2 M) on a 1.0 mmol scale to isolate both diastereomers. 1e was preformed using methyllithium (1.0 mmol), monodeuterated alkenyl boronic ester (1.0 mmol) and LiCI (1.0 mmol), Et₂O, 0 °C–r.t., 30 min then redissolved in THF. [lr(COD)CI]₂ (2.5 mol%), (S)-L1 (10 mol%) and (S)-2a (1.0 mmol) were added. Yields of the isolated and purified materials after oxidation by treatment of boronic ester with NaOH and H₂O₂. The dr was determined by ¹H NMR. The er of the major diastereomer was determined by HPLC analysis. Relative stereochemistry was determined by NOE analysis after cyclization via iodoetherification.

We next explored the relative configuration of products derived from boronate 1e involving an alkyl migrant onto an alkyl-substituted boronate (Scheme 4). Under optimized reaction conditions using carbonate (S)-2a, the reaction provided product 9 in 72% yield as a complex distribution of diastereomers. After oxidation and subsequent cyclization (10), diastereomers 9a and 9b could be isolated as an inseparable 5:1 mixture of diastereomers. NOE analysis revealed the major product of this mixture (9a) was the result of a syn addition across the olefin. We were unable to determine relative stereochemistry for 9b due to substantial overlap on NMR, but have assigned it by inference (see below). Likewise, 9c and 9d could be isolated as an inseparable 2:1 mixture of diastereomers. Irradiation of the C2 methylene hydrogen of the tetrahydrofuran derived from 9c showed correlations between the hydrogen and both the phenyl and methyl substituents at adjacent carbons. Thus, 9c also results from syn addition across the olefin, but with opposite facial selectivity compared to 9a (see Supporting Information for details). Alternatively, 9d gave an unexpected result, wherein NOE analysis revealed the methylene hydrogen interacted with the butyl group and benzylic hydrogens on adjacent carbons. This evidence led to the surprising conclusion that this diastereomer is the result of similar facial selectivity in the allylation compared to 9a, but proceeded through an anti addition across the olefin to give the C2 epimer of 9c. From these results we speculated that 9b might be the result of similar facial selectivity compared to 9c, followed by an anti addition pathway to give the C2 epimer of 9a. Taken together, these results indicate that alkyl migration onto alkyl-substituted alkenyl boronates proceeds with modest facial selectivity with regard to the olefin and could involve a stepwise process resulting in moderate diastereoselectivity. All classes of substrates react with high enantioselectivity.

Kinetic analysis of the 3-component coupling.

After determination of relative configuration for a variety of alkenyl boronate combinations, we sought to understand what role each component was playing in the overall rate of the reaction. The set of standard conditions for kinetic analysis is shown in Scheme 5, in which we used boronate 1f, carbonate (S)-2b, [Ir(COD)Cl]₂ and ligand (S)-L1 to provide boronic ester 11. One equivalent of LiCl was included in the reactions because it was found to slightly increase yield and diastereoselectivity. We found over the course of optimizing reaction conditions for kinetic analysis that the reaction proceeded at room temperature and was even too fast to monitor conveniently by NMR or in situ FTIR using the reaction conditions previously disclosed. Diluting the reactions 10-fold (0.02 M) relative to the original reaction conditions allowed us to obtain kinetic profiles and extract initial rates for each reaction.

Scheme 5.

General Scheme for Kinetic Analysis of Three-Component Coupling Reaction

^aAII reactions for kinetic experiments were conducted in THF (0.02 M) on a 0.3 mmol scale. 1f was preformed using PhLi, isopropenyl boronic ester and LiCI, Et₂O, 0 °C–r.t., 30 min then redissolved in THF. [lr(COD)CI]₂ (x mol%), (S)-L1 (4x mol%) and (S)-2b were added. Conversion and dr were determined by crude NMR analysis. The er of the major diastereomer was determined by HPLC analysis.

Our previous work established that the reaction occurs via a kinetic resolution of racemic carbonate. Control experiments revealed that 1) the fast-reacting enantiomer (or “matched” enantiomer) reacted with high stereospecificity and 2) the slow reacting enantiomer (“mismatched”) generated the same major enantiomer product as the matched enantiomer, albeit at greatly reduced selectivity with e.r. ~ 2:1. These results indicated that the catalyst controls selectivity at both the oxidative addition step and the addition step. For our first kinetic experiments, we used enantioenriched carbonate (S)-2b to determine order dependance in boronate and catalyst without introducing a more complicated reaction network during these analyses. The reaction was monitored using in situ IR spectroscopy by following consumption of carbonate over the course of the reaction. Conversion to product was confirmed via crude ¹H NMR and ¹⁹F NMR analyses.

We first conducted a set of “same excess” experiments to probe catalyst robustness under relevant reaction conditions (Figure 1, top).³⁶ Monitoring these two kinetic profiles allowed us to analyze the same reaction from two different starting points. Entry 2 features the standard reaction conditions wherein both the concentration of boronate 1f ([ate]) and the matched carbonate (S)-2b ([mat]) were both 0.02 M. Entry 1 includes double the concentration of both substrates while maintaining the same concentration of catalyst ([Ir]_T). The initial substrate concentrations of entry 1 will equal entry 2 when the former reaction reaches 50% conversion. From that point onward, the reaction mixtures have the same substrate concentrations, but different concentrations of product and turnovers performed by the catalyst. After the two kinetic profiles are time adjusted, they show a clear overlap, indicating that there is no observable product inhibition or catalyst deactivation under optimized reaction conditions (Figure 1, bottom). Additionally, the fact that the first 20 catalyst turnovers of entry 2 mirror the second 20 catalyst turnovers of entry 1 supports the use of initial rates to study the overall kinetic profile.

Figure 1.

“Same excess” experiment to probe for catalyst robustness or product inhibition, (a) kinetic profiles for standard reaction conditions from Scheme 5 (blue) and 2x the concentration of boronate 1f ([B]) and same [xs] of matched carbonate (S)-2b ([mat]) (red), (b) time-adjusted kinetic profiles from part (a).

To quantitatively assess the order dependance on each component in the coupling reaction, we collected initial rate measurements for varying concentrations of boronate, matched carbonate and catalyst (Figure 2). We then determined kinetic order of each component by plotting the negative natural log of observed initial rate measurements (−ln(k_obs)) vs. the negative natural log of concentration. The slope of this linear fit was used to determine the apparent order in each component. We observed a first order kinetic dependence in concentration of boronate (slope = 1.16) (Figure 2, top). Likewise, a first order kinetic dependance was also observed in concentration of catalyst (slope = 0.95), which indicates that a single Ir is involved in the ground state and transition state of the reaction (Figure 2, middle). By contrast, reactions with different concentrations of matched carbonate (S)-2b showed clear overlap of the initial rates after normalization, suggesting a zero-order kinetic dependance (Figure 2, bottom). From these results we concluded that the turnover-limiting step involves both an iridium complex and the alkenyl boronate.

Figure 2.

Order dependance of components in three component coupling reaction. Initial rate measured by monitoring consumption of enantioenriched carbonate (S)-2b via in-situ FTIR. [ate] = 1f; [Ir] = [Ir(COD)Cl]₂/(S)-L1; [carb] = (S)-2b. (a) Left, initial rate profiles for varying concentrations of boronate 1f. Right, determination of apparent rate order of boronate. (b) Left, initial rate profiles for varying concentrations of total catalyst. Right, determination of apparent rate order of catalyst. (c) Left, initial rate profiles for varying concentrations of carbonate (S)-2b. Right, normalized overlay of the three initial rate profiles by taking concentration of carbonate at each point minus initial carbonate concentration for each separate experiment.

The first-order kinetic dependence on [ate] indicated that boronate was involved in the transition state, but it was uncertain if it was activating the carbonate for oxidative addition or engaging the Ir(π-allyl) intermediate in an allylation event. To distinguish between these two possibilities, we conducted a secondary kinetic isotope effect experiment using alkenyl boronate 1g and mono-deuterated alkenyl boronate 1h (Scheme 6). Boronate 1g provided product 12 in 90% yield in excellent diastereo- and enantioselectivity. Likewise, boronate 1h provided the monodeuterated product 13 in similar yields and stereoselectivity. Parallel kinetic experiments were executed in triplicate and performed alternating between the two substrates to ensure accurate measurements.³⁷ These experiments revealed an inverse secondary kinetic isotope effect (k_H/k_D = 0.85), providing strong evidence for a change of hybridization from sp² to sp³ on the alkenyl boronate at the transition state. We concluded from these results that the turnover-limiting step involves addition of the alkenyl boronate to the Ir(π-allyl)⁺ intermediate.

Scheme 6.

Secondary Kinetic Isotope Effect^a

^aAII reactions for secondary kinetic isotope effect experiments were conducted in THF (0.02 M) on a 0.3 mmol scale in triplicate. 1f and 1g were preformed using PhLi (0.3 mmol), alkenyl boronic ester (0.3 mmol) and LiCl (0.3 mmol), Et₂O, 0 °C–r.t., 30 min then redissolved in THF. [lr(COD)CI]₂ (2.5 mol%), (S)-L1 (10 mol%) and (S)-2b (0.3 mmol) were added. Conversion and dr were determined by crude NMR analysis. The er of the major diastereomer was determined by HPLC analysis.

We next explored the effect of using racemic carbonate on the reaction rate. When racemic carbonate (±)-2b was subjected to standard reaction conditions, the reaction was notably slower than when (S)-2b was used. This interesting result suggested that mismatched carbonate might inhibit the reaction and prompted us to investigate what role it was playing in the catalytic cycle. When studying intermediates in the iridium-catalyzed allylic substitution reaction with branched allylic alcohols, the Carreira group found that the iridium-bound olefin complex was the major species before introduction of triflic acid to promote the acid-catalyzed oxidative addition.²⁸ We hypothesized that the branched allylic carbonates might also behave in the same manner. We first subjected the ClIr[(S)-L1]₂ complex to either (S)-2b or (R)-2b separately and monitored them via ³¹P NMR. Both enantiomers bind to form two distinct diastereomeric iridium-olefin complexes.³⁸ From these results, we speculated that the mismatched carbonate might inhibit the reaction through a reversible binding event with the iridium complex to form an off-cycle species that competes with binding of the matched carbonate (Scheme 7a). To quantify this phenomenon, we subjected the reaction to scalemic carbonate at different ratios (R:S) and measured initial rates. These experiments showed clear evidence of inhibition by the mismatched carbonate. For example, at constant value of [mat], the reaction rate progressively slowed with a 1:0, 1:1, 1:1.5 and 1:3 ratio of [matched]:[mismatched]. Additionally, in the presence of mismatched carbonate, the reaction shows a positive kinetic dependence on matched carbonate in contrast to the results from kinetic experiments with (S)-2b alone (see Figure 2c). Thus, using a constant concentration of mismatched carbonate, the allylation is substantially faster with a 3:1 vs 1:1 ratio of [matched]:[mismatched].

Scheme 7.

Mechanistic Hypothesis and Rate Law in the Presence of Mismatched Enantiomer

(a) General reaction scheme of kinetic experiments using scalemic carbonate 2b. All reactions were conduted in THF(0.02 M). 1f was preformed using PhLi (0.3 mmol), isopropenyl boronic ester (0.3 mmol) and LiCI (0.3 mmol), Et₂O, 0 °C–r.t., 30 min then redissolved in THF. [lr(COD)CI]₂ (2.5 mol%), (S)-L1 (10 mol%) and varying ratios of (S)-2b and (R)-2b were added (0.6 mmol total), (b) Top, initial rate profiles for various scalemic concentrations of carbonate were assessed. Bottom, linear dependance of observed rate on the concentration of mismatched carbonate (R)-2b. (c) Proposed mechanistic hypothesis, rate law of reaction with racemic carbonate and ratio of binding constants for carbonate enantiomers.

A proposed catalytic cycle is shown in Scheme 7c. First, Ir(I) complex 14 can initially bind to either the matched or mismatched carbonate to provide two diastereomeric Ir(olefin) complexes. The mismatched carbonate can inhibit the reaction by binding reversibly to form off-cycle complex 15. Alternatively, the matched carbonate forms the active Ir(olefin) complex 16 that can undergo a reversible oxidative addition process to form Ir(π-allyl)⁺ intermediate 17. Intermediate 17 can then engage with the alkenyl boronate 18 to provide the desired coupling product 19 bound to Ir(I). The equilibrium between the iridium-bound coupling product and free Ir(I) is proposed to lie heavily in favor of the unbound complex (K_prod) as evidenced by the lack of observable product inhibition from the “same excess” experiment. Exchange of product with substrate allows the iridium complex 14 to re-enter the catalytic cycle. A derived rate law that is consistent with experimental results is shown in Scheme 7c. The expression captures the first-order kinetic dependence on total iridium and boronate under all conditions. In the absence of mismatched carbonate, the reaction rate is independent of [matched]. However, when both enantiomers of carbonate are present, the addition shows positive order with respect to matched carbonate concentration, while inhibition is observed with the mismatched carbonate (see Supporting Information for details).

The derived rate law predicts that inverse rate (1/k_obs) should be linearly related to the ratio [mis]/[mat] (Scheme 7c). Indeed, plotting 1/k_obs vs. the ratio of mismatched carbonate to matched carbonate ([mis]/[mat]) provided us with a linear relationship, validating the rate law (Scheme 7b). Moreover, the slope of this line could be used to determine the relative binding constants of the matched and mismatched carbonate ([K_mis]/[K_mat]) to the Ir catalyst. The data suggest that the mismatched carbonate binds ~5 times more tightly to the iridium complex than the matched carbonate, explaining the inhibitory effect of the mismatched carbonate. The kinetic data is bolstered by qualitative ³¹P NMR data of the distribution of iridium-olefin complexes which demonstrated preferential binding of the mismatched carbonate (see Supporting Information for details). These experiments lead to the conclusions that: (1) the major species observed is the iridium-bound olefin complex, rather than the Ir(π-allyl) species, and (2) the reaction relies on a reversible oxidative addition process such that boronate addition is turnover-limiting.

Nucleophilicity of alkenyl boronate.

The Aggarwal group has measured the reactivity of boronate complexes towards various electrophiles.^39,40 However, the general nucleophilicity of alkenyl boronates reacting at the β-carbon has not been quantified. We wanted to provide calibration on the reactivity of alkenyl boronates relative to other common nucleophiles. To explore the general nucleophilicity of alkenyl boronates we employed a well-established benzhydrylium method.^41,42,43 Boronate 1f was allowed to react with benzhydrylium ions 20a-20c to obtain the desired coupling product 21 (Figure 3a). These reactions proceeded with exclusive formation of the tertiary boronic ester and contained no observable byproduct formation resulting from phenyl addition or isopropenyl addition to the electrophile (see Supporting Information for details).

Figure 3.

Determination of nucleophile-specific parameters N and s_N of alkenyl boronate complex 1f. (a) Reactions of boronate 1f with benzhydrylium ions 20a-20c and correlation of second-order rate constants k₂ with electrophilicity parameters E (pyr - pyrrolidine, thq - tetrahydroquinoline). (b) Comparison of nucleophilicity paramters N (susceptibility parameters s_N in parentheses) of boronate 1f and other C- and N-based nucleophiles (site of reactivity indicated by arrow).

For kinetic analysis, the reactions were conducted under pseudo-first-order conditions with excess boronate relative to electrophile. The reactions were monitored using UV-vis spectroscopy at λ_max for each corresponding benzhydrylium ion (20a-20c). The k_obs values were obtained from the slope of the graph of ln([20]) vs. time, and runs were conducted in triplicate then averaged. The secondary rate constants k₂ were determined from dividing the average k_obs values by the concentration of boronate and are shown for each experiment. A plot of log(k₂) against the corresponding electrophilicity constants E provided a linear correlation. The susceptibility parameter s_N can be determined from the slope of this line (s_N = 0.57). Likewise, the nucleophilicity parameter N can be approximated from dividing the y-intercept by the susceptibility parameter s_N (N ~ 11).

The nucleophilicity parameter allows for a direct comparison to other classes of compounds (Figure 3b). Compared to other boronate complexes, boronate 1f is more nucleophilic and reacts exclusively at the β-carbon, indicated by the arrow in Figure 3b. This comparison is consistent with the fact that we did not observe transfer of phenyl or isopropenyl from 1f to 20, i.e., 1f acts as a π-nucleophile, not a phenyl or alkenyl nucleophile. For comparison, the boronate shows similar nucleophilic reactivity to morpholine-derived enamine, as well as saccharin, a nitrogen-based nucleophile. In relation to other carbon-based nucleophiles, 1f is more nucleophilic than allyl stannanes and silyl enol ethers, such as Danishefsky’s diene. Alternatively, 1f is less reactive that cyclic silyl ketene acetals and much less reactive than enolates and alkyl amines (not shown).⁴⁴ Of note, the alkenyl ate complex nucleophilic reactivity is within the range of classes of compounds that are known to be reactive in Ir(π-allyl) chemistry. Overall, these experiments offer the first analysis of general reactivity for this class of boronate complexes.

Computational studies.

We performed density functional theory (DFT) calculations to explore the mechanism of the Ir-catalyzed allylation of alkenyl boronates and the origin of enantio- and diastereoselectivities. The DFT calculations were performed at the M06/6–311+G(d,p)–SDD(Ir),SMD(THF)//B3LYP-D3(BJ)/6–31G(d)–LANL2DZ(Ir),SMD(THF) level of theory. The calculated reaction free energy profile of the Ir/(S)-L1-catalyzed reaction of the matched enantiomer of allylic carbonate (S)-2a and phenyl alkenyl boronate 1f is shown in Figure 4a. The binding of (S)-2a to the active Ir(I) catalyst supported by two phosphoramidite ligands (S)-L1 (22) leading to π-alkene complex 23 is exergonic by 1.4 kcal/mol. We also computed the binding of the mismatched enantiomer (R)-2a to 22, which is more exergonic (23’, ΔG = −4.0 kcal/mol), consistent with the experimental observation of inhibition by the mismatched enantiomer (see Scheme 7). The less favorable binding of the matched carbonate to Ir(I) relative to the mismatched carbonate reflects a steric clash between the phenyl ring of the ally carbonate and the phosphoramidite ligand observed in the former complex, but not the latter. However, the matched Ir[(S)-2a] complex orients the OBoc group away from the Ir(I) center and is therefore better aligned for oxidative addition (Figure S21).

Figure 4.

Computational study of the Ir-catalyzed enantioselective allylation of alkenyl boronates. All Gibbs free energies and enthalpies are with respect to 2a and 22.

From 23, oxidative addition (TS1) cleaves the C–O(Boc) bond and forms cationic π-allyl Ir^III complex 24. The addition of alkenyl boronate 1f to the π-allyl complex and the 1,2-metalate shift occur via a concerted transition state (TS2). In TS2, the C-C bond formation between the π-allyl and the alkenyl boronate is much more advanced than the Ph migration. This is evidenced by the long forming C–C bond between the carbon on the migrating Ph group and the alkenyl carbon (2.65 Å, Figure 4b). This concerted addition/migration step has a higher activation barrier than the oxidative addition (TS1), and thus is the rate- and stereoselectivity-determining step. From the resulting product π-complex 25, product dissociation to regenerate active catalyst 22 is slightly endergonic by 0.5 kcal/mol. The weaker binding of product compared to the binding of substrate (S)-2a is consistent with the experimental evidence against product inhibition (Figure 1).

We evaluated eight possible stereoisomers of the boronate addition transition state. These include additions involving two prochiral π-faces of the alkenyl boronate to the exo/endo stereoisomers of the π-allyl complex, as well as stereoisomers placing the migrating Ph group syn or anti to the forming C–C bond (named syn and anti addition, respectively). Four addition transition states are shown in Figure 4b and 4c, and the other four transition state isomers are shown in Figure S23. The most favorable addition transition state TS2 involving the syn addition of alkenyl boronate leads to the experimentally observed major diastereomeric product (S,S)-26. A less stable syn-addition transition state isomer TS3, in which the Ph migrates to the opposite π-face of the alkene on the alkenyl boronate, leads to the minor diastereomeric product (R,S)-27. The anti addition transition states TS4 and TS5 are both much higher in energy than the most favorable syn addition transition state TS2. The origin of stereoselectivity via the syn addition pathway can be rationalized by ligand steric effects and stabilizing dispersion interactions. In the most favorable syn addition transition state (TS2), the migrating Ph is placed in a less occupied quadrant and does not have steric clash with the phosphoramidite ligands (see the ligand steric contour plot in Figure S24). Additionally, TS2 is stabilized by a T-shaped π∙∙∙π interaction between the migratory Ph and the Ph on the π-allyl. In the less favorable syn addition transition state TS3, the migrating Ph is in a more occupied quadrant, and the ligand–substrate steric repulsion distorts the alkenyl boronate to a non-staggered conformation at the forming C–C bond with the π-allyl, evidenced by a small ∠H–C¹–C²–C³ dihedral angle of −27.0°. Consistently, we previously observed higher diastereoselectivity with aryl migrants (~10:1) compared to alkyl migrants (~5:1).³⁵ On the other hand, the anti addition transition states (TS4 and TS5) are destabilized because of the lack of stabilizing π∙∙∙π interaction between the migrating Ph and the Ph on the π-allyl. Moreover, in these anti addition transition states, the sterically demanding Bpin moiety is placed proximal to the phosphoramidite ligand. The ligand–substrate steric repulsions in TS4 and TS5 are evidenced by the short H∙∙∙H distance of 2.20 Å and 2.11 Å between the Me group on the Bpin and the alkenyl group on L1, respectively. Therefore, the ligand steric effects and dispersion interactions both contribute to the excellent stereoselectivity via the syn addition pathway.

Hammett analysis.

To examine the substituent effects on the enantioselective allylation of alkenyl boronates, the reaction of simple para-substituted aryl boronates was first carried out. A Hammett plot showed a linear relationship between relative rate and their corresponding σ_p values (Figure 5a). The rho value was calculated from the slope of the line (ρ = 0.89), leading to the surprising result that electron-poor aryl groups migrate faster than electron-rich aryl groups. DFT calculations of the rate-determining alkenyl boronate addition transition states with the same series of aryl boronate substrates yielded a good correlation (R² = 0.92) with the experimental activation free energies derived from measured rate constants. DFT-calculated transition state structures indicate that weakening of the B–C(aryl) bond is more advanced than the formation of the C–C(aryl). For example, in TS2, a long forming C–C(aryl) bond distance of 2.65 Å was observed (Figure 4b). Therefore, electron-rich aryl groups may strengthen the B–C(aryl) bond in the aryl boronate, and thus impede the migration. Additional ¹¹B NMR experiments have ruled out the possibility that electron-poor boronates stabilize the boronate in an equilibrium between the boronate and the boronic ester.

Figure 5.

Experimental and computational studies of substituent effects on the rate of the Ir-catalyzed enantioselective allylation of alkenyl boronates.

Furthermore, a variety of other boronate complexes were analyzed in kinetic analyses of the three-component coupling reaction. Alkyl derived boronates were ~2 times less reactive than the parent phenyl migrant. This observation is consistent with the trend that electron-rich aryl groups migrate more slowly than their electron-poor counterparts. Likewise, a variety of α-boryl styrene derivatives were investigated. Phenyl and 4-MeO-Phenyl derived α-boryl styrenes reacted at similar rates compared to the parent isopropenyl derived boronate. More electron poor variants such as the 4-CF₃ and 4-CN styrenes resulted in uninterpretable kinetic profiles due to competing pinacol allylation as the major species (see Supporting Information for details)

Similar substituent effects were investigated for para-substituted secondary allylic aryl carbonates, and Hammett analysis demonstrated a linear relationship between relative reaction rate and their σ⁺ values (Figure 5b). The rho value was calculated from the slope (ρ = −1.32) indicating a positive charge build up at the electrophilic carbon in the turnover-limiting step. The experimentally observed electronic effects of aryl carbonates were corroborated by DFT calculations, which provided a good correlation between the computed (ΔG^‡_DFT) and experimentally derived (ΔG^‡_exp) Gibbs free energies of activation (R² = 0.93). We postulated that the electron rich carbonates help stabilize the positive charge build up-at the transition state during addition. Alternatively, electron-donating groups could shift the equilibrium from the Ir^I(olefin) complex towards the Ir^III(π-allyl)]⁺ complex, providing a higher concentration of the reactive intermediate.

Conclusions.

In summary, we have provided key mechanistic insights into the iridium-catalyzed enantioselective allylation of alkenyl boronates. The three-component coupling reaction involves the use of organoboronic esters, organolithium reagents and secondary allylic aryl carbonates. Stereochemical analysis revealed that a variety of alkenyl boronate complexes display selective syn addition of the migrating organic group and allyl fragment across the olefin. Significant data include: (1) kinetic order dependance of boronate, catalyst and both enantiomers of allylic carbonate; (2) experimental evidence supporting the proposed mechanistic hypothesis and derived rate law; and (3) Hammett analysis of substituent effects on the transition state for both aryl alkenyl boronates and aryl carbonates; (4) determination of nucleophilic reactivity of alkenyl boronates relative to other classes of compounds; and (5) DFT calculations that explain and corroborate the kinetic analysis. Together, these results implicate the reversibility of the oxidative addition to allylic carbonates, which reveals that loss of CO₂ is not immediate. CO₂ may be lost slowly during the reaction or upon workup. The DFT studies highlight unanticipated aspects of the transition state such as attractive interactions between the migrating group and the ligand framework leading to syn addition as well as implicating substantial negative charge buildup on the migrating carbon. Efforts to exploit these discoveries are ongoing.