Dynamic Kinetic Asymmetric Hydroacylation: Racemization by Soft Enolization

Abstract

We report a dynamic kinetic asymmetric transformation (DyKAT) of racemic aldehydes by Rh-catalyzed hydroacylation of acrylamides. This intermolecular hydroacylation generates 1,4-ketoamides with high enantio- and diastereoselectivity. DFT and experimental studies provide mechanistic insights and reveal an unexpected Rh-catalyzed pathway for aldehyde racemization. Our study represents a pioneering kinetic resolution by intermolecular hydroacylation and contributes to the growing field of stereoconvergent catalysis featuring C–C bond construction.

Introduction

Racemization drives diverse phenomena, from the origin of d-amino acids¹ to the notorious behavior of medicines like thalidomide² (Figure 1A). Chemists exploit this interconversion³ of enantiomers to develop efficient strategies for constructing chiral building blocks via stereoconvergent catalysis.⁴ This subset of enantioselective catalysis includes stereoablative transformations,⁵ dynamic kinetic resolution (DKR), and dynamic kinetic asymmetric transformations (DyKAT).⁶ Despite significant progress, inventing efficient chiral resolutions remains an important challenge. Compared to hydrogenations⁷ or alcohol acylations,⁸ DKR/DyKAT processes that construct C–C bonds remain rare.⁹ Given the broad range of transformations possible with carbonyl compounds, a deeper understanding of how to influence carbonyl racemization will unlock a suite of asymmetric transformations. Herein, we present a novel DyKAT featuring intermolecular hydroacylation and identify soft-enolization as an overlooked pathway for racemizing aldehydes in preference to ketones.

Figure 1.

Racemization’s central role in diverse phenomena. (a) Racemization of l-amino acids in vivo via enzymes called racemases leads to D-amino acids while racemization of drug molecules such as thalidomide can lead to inactivity or toxicity. (b) Facile racemization of starting material is the key to DKR (left) and DyKAT (right). (c) Various racemization mechanisms exist for kinetic resolutions of carbonyls but the use of soft enolization is often overlooked. (d) State-of-art dynamic asymmetric hydroacylations are limited to cyclizations while we proposed a DKR/DyKAT of intermolecular hydroacylation using soft enolization for aldehyde racemization, generating a 1,4-ketoamide with remote stereocontrol.

Dynamic resolutions require an enantioselective transformation coupled with a rapid racemization and thus, understanding how to influence racemization is critical for achieving high yield and selectivity (Figure 1B). Among the classic examples of DKR, Noyori’s ketone hydrogenation¹⁰ relies on tautomerization, a mechanism that dominates for highly acidic carbonyl motifs (e.g., β-keto compounds). Such tautomerizations can be catalyzed by acid^11a or base.^11b−11f In the context of C–C bond formations, Johnson achieved DKRs of α-keto esters via alkylations^12a and arylations^12b with tertiary amine bases. In addition to general acid- or base-catalyzed mechanisms, alternative pathways for carbonyl racemization have emerged by transition metals or organocatalysis.¹³ Most recently, Shi reported an arylation of racemic ketones, introducing a unique mechanism for ketone racemization involving oxidative addition and β-hydride elimination; however, this mechanism has not been investigated with computational studies.¹⁴ By density functional theory (DFT) and experiment, Jørgensen and Houk elucidated a mechanism for stereoconvergent aldehyde α-arylation: a chiral aminocatalyst promotes equilibration by dynamic Walden inversion.¹⁵ In another application of aminocatalysis, List demonstrated a DKR of racemic aldehydes via reductive amination where aniline catalyzes racemization via enamine/imine formation.¹⁶ Our lab extended this concept by developing an intramolecular DKR to synthesize cyclopentanones using Rh and adamantylamine.¹⁷ In line with List, we proposed a racemization by the amine through reversible condensation and tautomerization to the enamine. While reasonable, this mechanism warrants deeper investigation. It occurred to us that soft enolization could also be a viable pathway, where Rh acts as a Lewis acid¹⁸ to activate the carbonyl compound toward deprotonation by a mild base. Although soft enolization is a mild and useful strategy for enolate synthesis,¹⁹ it has yet to be harnessed as a racemization step in stereoconvergent catalysis (Figure 1C).

With this challenge in mind, we set out to develop a novel chiral resolution of racemic aldehydes and investigate aldehyde racemization pathways through both experimental studies and DFT. Among potential transformations for racemic aldehydes, we focused on hydroacylation,^20,21 where aldehyde C–H bonds are activated and added across a π-bond. While a promising C–H to C–C bond formation, stereoconvergent hydroacylations²² resolving aldehydes have been limited to intramolecular variants to give cyclic ketones. Subsequent to our report,¹⁷ Sato demonstrated cyclization with allenals²³ and alkynals^11a to form six- to eight-membered rings. However, by using chelating substrates and appropriate catalysts, many intermolecular hydroacylations have been achieved.²⁴ Among this precedent, we chose Tanaka’s intermolecular hydroacylation of acrylamides²⁵ as an attractive platform for DKR or DyKAT. Our goal was to develop a catalyst combo²⁶ (using Rh and amine catalysts²⁷) that would (1) enable rapid epimerization of the aldehyde starting materials, preferentially over ketone products, (2) achieve high enantio- and diastereoselectivity, and (3) overcome Rh catalyst deactivation resulting from amine coordination. If successful, this intermolecular C–C bond construction would resolve α-substituted aldehydes to generate 1,4-keto carbonyl motifs bearing remote (1,4-stereogenic) centers in a single asymmetric transformation (Figure 1D).

Results and Discussion

We chose a model transformation featuring the addition of racemic α-branched aldehyde 1a to acrylamide 2a (Table 1). Initial studies focused on cationic Rh catalysts on the basis of Tanaka’s report.²⁵ We identified a promising catalyst generated by hydrogenation of Rh(nbd)₂BF₄ (see the Supporting Information).²⁸ The achiral bisphosphine ligand, dppb (L1), provided ketoamide 3aa in 67% yield and 2:1 dr (entry 1) at 80 °C, suggesting a poor inherent diastereoselectivity. In our previous DKR,¹⁷ bulky 1-adamantylamine (1-AdNH₂) was key to both diastereo- and enantiocontrol, presumably facilitating chemoselective racemization of the aldehyde starting material over the ketone product (Figure 1D). We reason that these dual catalysts act as a “frustrated Lewis pair”²⁹ because the steric bulk of the amine cocatalyst hinders the coordination to the Rh catalyst. To develop an asymmetric version, we examined chiral bisphosphine ligands with 1-AdNH₂ as the cocatalyst. The (R)-DTBM-SegPhos ligand (L2) was effective for the intramolecular hydroacylation of alkyl aldehydes;¹⁷ however, it promoted no reactivity in the proposed intermolecular hydroacylation (entry 2). The (R,R)-QuinoxP* ligand (L3), reported by the Tanaka group,²⁵ was unsuccessful at 60 °C (see the Supporting Information), but at an elevated temperature (80 °C) furnished 3aa in 37% yield and 6:1 dr (entry 3). We then discovered that JoSPOphos (L4) gave 3aa in 92% yield and >99% ee, and enhanced diastereoselectivity (12:1 dr) (entry 8). This ligand has shown to be a promising ligand across different hydroacylations^17,30a,30b and other hydrofunctionalizations.^30c−30f Overall, the ligands demonstrated excellent reactivity in this reaction (L1 and L4) both have relatively large bite angles (≥97°). However, L3 outperformed L2 despite a smaller bite angle, indicating that there is no clear correlation between bite angle and reactivity. Omitting the amine led to a decreased dr (4:1) (entry 4). Other bulky amines were investigated and resulted in diminished yields and diastereoselectivities (entry 5–6).³¹ At lower temperatures, we observed 42% yield with 14:1 dr (entry 7). Concentrations of reagents also affected reactivity and diastereoselectivity, (see Table S2 entry 18), presumably because they influence rates of aldehyde racemization and of hydroacylation.³² The absolute configuration of the 1,4-stereocenters for 3aa was established as (R,R) by X-ray crystallography.

Table 1. Optimization for DKR by Intermolecular Hydroacylation^a.

^aReaction conditions: 1a (0.3 mmol, 1.5 equiv), 2a (0.2 mmol, 1.0 equiv), Rh(nbd)₂BF₄ (10 mol %), ligand (10 mol %), amine (10 mol %), DCE (0.4 mL), 60 °C, 24 h. Bite angles were calculated through DFT studies.

^bIsolated yields.

^cDiastereoselectivities (dr) were determined by ¹H NMR analysis of unpurified reaction mixture.

^dEnantioselectivities (ee) were determined by chiral SFC analysis.

With a protocol in hand, we examined aldehydes 1 bearing different α-substituents (Table 2). α-Aryl aldehydes with para-substituted methoxy and halogens underwent DKR smoothly; ketoamides 3ba–3da(33) can be prepared with moderate to high yields (65–94%) and stereoselectivities (10:1–13:1 dr, >99% ee). In addition, a strong electron-withdrawing trifluoromethyl-substituted α-aryl aldehyde was tolerated (3ea, 93% yield, 11:1 dr, >99% ee). Both electron-donating and -withdrawing groups could be incorporated at the meta-position to provide ketoamides 3fa (78% yield, 10:1 dr, >99% ee) and 3ga (92% yield, 11:1 dr, >99% ee). The introduction of a sterically hindered ortho-methyl substituent led to a drop in both reactivity (39% yield) and stereocontrol (5:1 dr, 97% ee) in forming ketoamide 3ha. Ketoamides 3ia bearing a naphthyl group (86% yield, 13:1 dr, >99% ee) and 3ja with an acetal moiety (82% yield, 8:1 dr, >99% ee) were both accessible. For the alkyl substitution, α-methyl can be replaced with α-ethyl (3ka, 91% yield, 9:1 dr, 99% ee). An α,α-dialkyl aldehyde, isobutylaldehyde (S1a), afforded S3aa in a comparable yield (76%, see the Supporting Information) but additional steric bulk showed decreased reactivity (3la, 64% yield, 98% ee), consistent with the trend observed for α-aryl substrates. The decreased diastereoselectivity (7:1 dr) results from the more comparable size between methyl and cyclohexyl. Overall, these results showcase the first DKR/DyKAT of aldehydes by intermolecular hydroacylation with high stereocontrol.

Table 2. Scope of Aldehydes^a.

^aReaction conditions: 1 (0.3 mmol, 1.5 equiv), 2a (0.2 mmol, 1.0 equiv), Rh(nbd)₂BF₄ (10 mol %), L4 (10 mol %), 1-AdNH₂ (10 mol %), DCE (0.4 mL), 60 °C, 24 h. Yields of isolated products are given. Diastereoselectivities (dr) were determined by ¹H NMR analysis of unpurified reaction mixture. Enantioselectivities (ee) were determined by chiral SFC or HPLC analysis.

^bDiastereoselectivity (dr) was determined by quantitative ¹³C NMR analysis of unpurified reaction mixture.

Next, we investigated the acrylamide scope (Table 3). The steric tolerance at the α-position of the acrylamide is wide, with no effect on reactivity and a negligible impact on stereoselectivity when changing from methyl to benzyl (3ab, 91% yield, 9:1 dr, >99% ee) or phenyl (3ac, 83% yield, 9:1 dr, >99% ee). An unsubstituted acrylamide 2d provided 3ad (45% yield, 79% ee). The diminished yield is presumably due to competitive polymerization.³⁴ In place of Tanaka’s diphenyl acrylamide substrate,²⁵ we tested acrylamides derived from morpholine (2e), N,N-dibenzylamine (2f) and aniline (2g). While 3ae was produced with 11:1 dr, selectivities for the less sterically hindered substrates were considerably lower (3af and 3ag, 8:1 and 6:1 dr). The Weinreb amide 3ah was prepared (65% yield, 9:1 dr, >99% ee). This amide provides a convenient handle for further elaboration.³⁵

Table 3. Scope of Acrylamides^a.

^aReaction conditions: 1a (0.3 mmol, 1.5 equiv), 2 (0.2 mmol, 1.0 equiv), Rh(nbd)₂BF₄ (10 mol %), L4 (10 mol %), 1-AdNH₂ (10 mol %), DCE (0.4 mL), 60 °C, 24 h. Yields of isolated products are given. Diastereoselectivities (dr) were determined by ¹H NMR analysis of unpurified reaction mixture. Enantioselectivities (ee) were determined by chiral SFC analysis.

To inspect the tolerance of different heterocycles, we conducted a compatibility test³⁶ using the model reaction in this DKR (Table 4). Reactivities and enantioselectivities were not significantly affected by addition of pyrrole (93% yield, 99% ee) or thiophene (96% yield, 99% ee), but the diastereocontrol was less effective (both 8:1 dr). Furan decreased the yield (42%), but stereoselectivity was maintained (10:1 dr, >99% ee). Pyridine completely inhibited the reactivity, presumably due to strong binding to the Rh catalyst. The presence of sterically hindered 2,6-lutidine was tolerated; 3aa was isolated in 76% yield with a comparable dr (10:1). Compared to other heterocycles, indole had the biggest impact on lowering diastereoselectivity (6:1 dr). The use of isopropanol led to a lower yield (52%), as the aldehyde 1a was consumed by a competitive pathway (based on nuclear magnetic resonance (NMR) spectroscopy of the unpurified mixture). While we were encouraged by the functional group compatibility of this method, we were surprised to find that incorporating heterocycles at the α-position of acrylamide 2 is challenging and warrants further studies (see the Supporting Information).

Table 4. Examining Functional Group and Heterocycle Compatibility^a.

^aReaction conditions: 1a (0.3 mmol, 1.5 equiv), 2a (0.2 mmol, 1.0 equiv), Rh(nbd)₂BF₄ (10 mol %), L4 (10 mol %), 1-AdNH₂ (10 mol %), additive (1.0 equiv), DCE (0.4 mL), 60 °C, 24 h. Yields of isolated products are given. Diastereoselectivities (dr) were determined by ¹H NMR analysis of unpurified reaction mixture. Enantioselectivities (ee) were determined by chiral SFC analysis.

On the basis of literature precedent,^{20a,20c,30a,37} we propose the following catalytic cycle for the formation of the major diastereomer (2R,5R)-3 (Figure 2). The active Rh catalyst I is formed upon coordination of L4 to the precatalyst, followed by chelation of acrylamide (2) to give complex II. This complex undergoes oxidative addition to the C–H bond of aldehyde (1) to generate Rh complex III, which is coordinatively saturated and less likely to undergo undesired decarbonylation. Subsequent migratory insertion leads to formation of hydrorhodation intermediate IV. Finally, reductive elimination results in the formation of the second stereocenter, which preferentially gives (2R,5R)-3 and regenerates the active catalyst.

Figure 2.

Proposed mechanism for Rh-catalyzed hydroacylation.

However, key questions remain regarding the fine details of this mechanism and the nature of the observed stereoselectivity. Namely, (1) what is the origin of the DKR? (2) how does the reaction achieve such high stereoselectivity? (3) what is the role of the Rh and 1-AdNH₂ amine catalysts? (4) what is the turnover limiting step? Seeking to answer these mechanistic questions, we performed a combination of experimental and computational studies.

For the computational portion, we applied DFT to analyze the title reaction – the coupling of aldehyde 1a and acrylamide 2a to yield diastereomer (2R,5R)-3a catalyzed by Rh-L4 (Table 1, entry 8). DFT calculations were performed at the B3LYP-D3/6-311+G(d,p) LANL2DZ (Rh, Fe) PCM (DCE)//B3LYP-D3/6-31G(d) LANL2DZ (Rh, Fe) level of theory,³⁸ as implemented in Gaussian 16.³⁹ Thermal corrections were determined using Grimme’s quasi-rigid rotor harmonic oscillator approximation at 60 °C.⁴⁰ Intrinsic reaction coordinate (IRC) calculations were performed to confirm that transition state structures (TSs) connected minima along the potential energy surface (PES). Conformational searching was conducted both manually and using CREST (Conformer and Rotamer Ensemble Sampling Tool). The conformational space was extensively sampled; for example, the search resulted in 58 unique structures for the migratory insertion transition structures.

Our computational study sought to understand the critical mechanistic aspects of this DyKAT. The PES for the lowest energy pathways leading to the formation of the four possible diastereomers for hydroacylation is shown in Figure 3. Following the formation of active catalyst I, bidentate coordination of 2a to Rh-L4 (I) leads to intermediate II, which oxidatively adds into the aldehydic C–H bond of (R)- or (S)-1a (TS_OX-R or TS_OX-S). The additional chelation around Rh by 2a, yields a lower energy oxidative addition TS compared to oxidative addition from the monodentate Rh-1a complex which does not involve the acrylamide (see the Supporting Information for more details). The oxidative addition step is facile – TS_OX-R and TS_OX-S are both low barrier steps (ΔG^‡ = 7.7 and 4.5 kcal/mol, respectively), leading to the reversible formation of acyl hydride Rh intermediate (III). From there, 1,2-migratory insertion of bound acrylamide 2a into the Rh-hydride bond occurs as a highly exergonic step, affording stable, cyclic Rh-acyl-alkyl intermediate IV. This hydrorhodation pathway with Rh inserting into the terminal β-position of 2a exhibits a free energy barrier of 29.6 kcal/mol relative to intermediate III for (R)-1a (TS_MI-R).⁴¹

Figure 3.

Potential energy surface depicting the relative barriers of oxidative addition, hydrorhodation and reductive elimination in this Rh-catalyzed intermolecular hydroacylation reaction. DFT calculations were performed at B3LYP-D3/6–311+G(d,p) LANL2DZ (Rh, Fe) PCM (DCE)//B3LYP-D3/6–31G(d) LANL2DZ (Rh, Fe) level of theory.

In accordance with experimental evidence of (R)-1a as the matched enantiomer for this catalytic system, TS_MI-R lies 2.7 kcal/mol lower in energy than hydrorhodation with (S)-1a (TS_MI-S). This difference is large enough that the yield of product would be capped at ∼50% yield if there was no isomerization of aldehyde. Hydrorhodation intermediate (IV) then undergoes reductive elimination of the acyl group with the reduced acrylamide to form the ketoamide major diastereomer, (2R,5R)-3aa (TS_RE-RR), which is energetically favored over that of the minor diastereomer (2R,5S)-3aa (TS_RE-RS) by 1.6 kcal/mol.

A careful analysis of the potential energy surface indicates the first stereocenter (α to the ketone) is set during the irreversible migratory insertion step, where the matched (R)-1a reacts faster than (S)-1a (98.5:1.5). However, the second stereocenter is set in the reductive elimination step, which ultimately determines the enantio- and diastereoselectivity of the product. The formation of each stereoisomer is the product of the rates of formation for the migratory insertion and reductive elimination steps, resulting in an enhanced enantioselectivity of >99% ee and a predicted dr of approximately 11.3:1, which is in excellent agreement with experiment (2R,5R-3aa yields >99% ee and 12:1 dr).⁴²

To evaluate the origin of selectivity induced during the formation of the first stereocenter, a distortion-interaction analysis⁴³ was performed on TS_MI-R and TS_MI-S, in which distortion energy is the energy required to distort reaction components from their ground state structures into TS conformations while interaction energy describes how the distorted catalyst and substrate fragments interact within the TS. Energy decomposition reveals that despite interaction energy favoring the TS leading to the minor enantiomer (TS_MI-S) by 10.1 kcal/mol (ΔΔE^‡), the TS leading to the major enantiomer (TS_MI-R) undergoes 13.5 kcal/mol less distortion to convert to TS-like geometry compared to TS_MI-S. As such, the enantioselectivity at the first stereocenter is distortion-controlled in this model. Similarly, an energy decomposition analysis on TS_RE-RR and TS_RE-RS reveals that the origin of diastereoselectivity at the second chiral center is also distortion controlled. This indicates that matched enantiomer (R)-1a fits into the catalytic pocket more favorably than (S)-1a, binding with minimal distortion comparatively (see the Supporting Information for details).

To probe the mechanism further, we performed deuterium labeling and kinetic isotope effect (KIE) studies. Reaction of deuterated aldehyde d-1a forms ketoamide d-3aa in 83% yield under standard conditions (Scheme 1A). Analysis of d-3aa (using deuterium NMR spectroscopy) reveals deuterium incorporation at the α-position of the amide, as well as 14% deuterium incorporation at the cis β-position. In addition, NMR analysis of the unpurified reaction mixture confirms H/D exchange and deuterium scrambling of unreacted d-1a at the α-position (see the Supporting Information), which supports the reversible oxidative addition determined from the calculated PES (Figure 3).

Scheme 1. Isotopic Labeling and KIE Experiments^,,

Isolated yields.

Due to the overlap of α-proton and trans β-proton, 83% represents the total percent of deuterium incorporation of both positions (orange); 14% refers to the percent of deuterium incorporation of the cis α-position (black).

Reaction conditions: 1a (0.15 mmol, 0.75 equiv), 2a (0.2 mmol, 1.0 equiv), Rh(nbd)₂BF₄ (10 mol %), L4 (10 mol %), 1-AdNH₂ (10 mol %), DCE (0.4 mL), 60 °C, 30 min.

Deuterium incorporation into the residual acrylamide 2a was also observed.^30a Although d-1a was not isolable, approximately 40% deuterium incorporation at the terminal position of the olefin was observed in recovered d-2a (Scheme 1A). Deuterium incorporation of both protons could result from geminal proton scrambling in hydroacylation;⁴⁴ however, this is unlikely due to the irreversibility of migratory insertion leading to IV (Figure 3). We explored the possibility of an alternative pathway involving a migratory insertion where Rh inserts into the α-position,^37d forming intermediate IV’(37e) followed by reversible β-hydride elimination and oxidative addition, which could exchange the originally labeled aldehydic deuterium (highlighted in yellow) to the acrylamide d-2a (Scheme 1B). Interestingly, the relative barrier for formation of IV’ is energetically feasible, lying 6.0 kcal/mol below the barrier for TS_MI-R. We conclude that this alternative hydrorhodation to IV’ is a nonproductive pathway that produces the acrylamide starting material and accounts for the scrambling observed in the labeling experiment.

Competitive ‘one-pot’ KIE experiments using a 1:1 mixture of 1a and d-1a (Scheme 1C) were conducted. A primary k_H/k_D of 2.0 ± 0.3⁴⁵ suggests that a hydrogen bond is involved in the first irreversible step for the aldehyde; however, the observed KIE of 2.0 is qualitatively difficult to interpret.⁴⁶ The predicted k_H/k_D for the relevant steps in the catalytic cycle are 3.6 for oxidative addition, 3.1 for migratory insertion, and 0.7 for reductive elimination. According to the free energy surface, migratory insertion represents the first irreversible step for 1a. However, the measured KIE of 2.0 is only partially expressed with respect to the predicted k_H/k_D for TS_MI-R of 3.1. Dampening of the experimental KIE is likely due to the scrambling of the label in d-1a caused by both the reversibility of oxidative addition and the alternative 1,2-migratory insertion pathway (Scheme 1B). This serves as a cautionary example of qualitatively interpreting KIEs when isotopic label scrambling occurs prior to the turnover-limiting step; lending to the importance of computational insights when evaluating complex reaction mechanisms. Given the reversibility of oxidative addition based on deuterium labeling results and free energy barrier heights, we thus propose that migratory insertion to IV is the first irreversible step in the reaction and is likely turnover limiting.

We then pursued to answer our questions on the racemization pathways to understand the roles of 1-AdNH₂ and Rh catalyst. Primary amines are known to isomerize α-branched aldehydes, such as 1a, through the formation of an enamine intermediate.⁴⁷ In our case, we also observed the enamine (when 1a is treated with 1-AdNH₂) by NMR spectroscopy which suggests the possibility of racemization through a condensation mechanism (Scheme 2). Additionally, racemization of enantioenriched 1a with only 1-AdNH₂ was also observed, albeit at a slow rate (see the Supporting Information for details).

Scheme 2. Enamine Formation Studies^,

Reaction conditions: 1a (0.15 mmol, 1.0 equiv), 1-AdNH₂ (7 mol % or 1.0 equiv), DCE (0.2 mL), 60 °C, 30 min.

Ratio determined by ¹H NMR analysis of a portion of unpurified reaction mixture. With 1.0 equiv of 1-AdNH₂, aldehyde: enamine = 1:1.7.

While the condensation pathway is well-known,⁴⁸ the role of Rh in this racemization has yet to be explored. As shown in Table 5, we examined the transformation in the absence of base (1-AdNH₂). In these experiments, the reactions were quenched by addition of NaBH₄ and we recovered unreacted 1a as the resulting alcohol at various time points. The ee was measured and compared to the expected value of the aldehyde in a theoretical case where there was no aldehyde racemization occurring (in other words, a simple and not dynamic kinetic resolution). At 5 min, minimal racemization of the starting aldehyde 1a was observed (observed 12% ee, vs expected 15% ee, entry 1). At the 1 h time point, aldehyde 1a was recovered in 38% ee (expected 68% ee, entry 2). At 4 h, ketoamide 3aa was formed in 96% NMR yield with 4:1 dr. In a kinetic resolution, this result would indicate a depletion of the matched enantiomer. However, instead of highly enantioenriched 1a (>99% ee), we recovered reduced 1a in 42% ee (entry 3). These surprising results indicate that isomerization of the aldehyde occurs by Rh, in the absence of the amine cocatalyst. This conclusion is further supported by an observed decrease in ee of enantioenriched 1a when subjected to active Rh catalyst without acrylamide (Scheme 3). Throughout the hydroacylation, (S)-1a is the major recovered enantiomer, which aligns with the proposal that (R)-1a is the matched substrate.

Table 5. Racemization Studies Under Reaction Conditions^a.

^aReaction conditions: 1a (0.3 mmol, 1.5 equiv), 2a (0.2 mmol, 1.0 equiv), Rh(nbd)₂BF₄ (10 mol %), L4 (10 mol %), DCE (0.4 mL), 60 °C, 5 min–4 h.

^b10 mol %.

^cYields and diastereoselectivities of 3aa were determined by ¹H NMR analysis of a portion of unpurified reaction mixture using 1,3,5-trimethoxybenzne as an internal standard. The other portion of reaction mixtures was quenched using NaBH₄ (>20 equiv) in MeOH.

^dTheoretical ee of recovered 1a for kinetic resolution (KR) was calculated based on equivalence of 1a and yields and dr of 3aa assuming there was no racemization on a highly selective transformation. [Theoretical (R)-1a = starting equiv – major diastereomer yield; Theoretical ee = (Theoretical (S)-1a – Theoretical (R)-1a)/(Theoretical (S)-1a + Theoretical (R)-1a) × 100%].

^e1a was recovered as the alcohol for determination of enantiomeric excess (ee), which was performed by SFC analysis on a chiral stationary phase.

Scheme 3. Racemization of Enantioenriched Aldehyde.

While this implicates an unexpected DyKAT pathway for racemization involving Rh, the addition of 1-AdNH₂ also has a strong influence on the diastereocontrol (Table 5 entry 4). At 1 h, the observed diastereoselectivity for the generation of ketoamide 3aa was higher in comparison to the experiment performed without amine cocatalyst (10:1 vs 8:1) and enantioenrichment of recovered 1a was lower (14% ee vs 38% ee). Although enamine formation could promote racemization in this reaction, we could also imagine amine acting as a simple base. It is worth noting that the hydroacylation proceeds less efficiently with the addition of amine (Table 5, 53% vs 70% NMR yield at 1 h); this difference may be due to Rh-amine coordination or the presence of water generated by imine condensation. Thus, the amine could play multiple roles in enhancing the relative rate of racemization over hydroacylation.

Previous reports indicate that Rh can participate in various mechanisms for racemization. In the DyKAT of chiral sulfoxides, we discovered that Rh accelerates racemization through a formal 2,3-sigmatropic rearrangement.^49a Andrieu and coworkers reported that Rh can racemize chiral aminophosphine oxides or sulfoxides by reversible P–C bond cleavage.^49b In this study, we envisioned that cationic Rh can act as a Lewis acid¹⁸ and promote tautomerization through formation of an η¹-Rh enolate^50a (Figure 4, Pathway A). Alternatively, Rh could racemize 1a through a transition-metal catalyzed β-hydride elimination and form an η³-Rh enolate (Figure 4, Pathway B),^50b,50c similar to the pathway recently proposed by Shi for the Ni-catalyzed epimerization of α-substituted ketones.¹⁴ However, this unique β-hydride elimination to form metalloenolates has no theoretical support. Thus, we set out to study possible racemization pathways of 1a involving Rh and 1-AdNH₂ utilizing DFT.

Figure 4.

Proposed mechanisms for Rh-catalyzed epimerization of 1a.

Our calculations of the various racemization pathways indicate that the DyKAT follows the Curtin–Hammett principle; the equilibration of starting material enantiomers (R)-1a and (S)-1a through Rh-catalyzed racemization pathways described (vide supra) are lower in energy relative to the proposed turnover-limiting migratory insertion step in the catalytic cycle (Figure 5, TS_MI-R leads to the major diastereomer of product). The calculated PES aligns with the experimentally observed racemization of recovered aldehyde 1a (Table 5), consistent with a DyKAT (i.e., the rate of racemization is greater than the rate of product formation).^4,6 Crucial to maintaining the high reaction selectivity, calculations also suggest that intermediate IV and ketoamide product 3aa are unlikely to epimerize through similar pathways; the preservation of the stereocenters is ideal for a DKR and thus enables the high enantio- and diastereoselectivity observed experimentally (see the Supporting Information).

Figure 5.

Free energy surface comparing the potential racemization pathways of (R)-1a and (S)-1a. Key transition state structures for deprotonation by 1-AdNH₂ (TS_DepS) and β-hydride elimination by Rh (TS_BHE-S) are highlighted. DFT calculations were performed at B3LYP-D3/6–311+G** LANL2DZ (Rh, Fe) PCM (DCE)//B3LYP-D3/6-31G* LANL2DZ (Rh, Fe) level of theory.

Direct deprotonation of the Lewis acid Rh complex via the 1-AdNH₂ external base (Pathway A) is the lowest energy pathway located for the racemization of aldehyde 1a (TS_DepS ΔG^‡ = 7.7 kcal/mol, and TS_DepR ΔG^‡ = 8.2 kcal/mol above prereactive Rh-1a complex, Int_PRC-R).⁵¹ These deprotonation TSs both result in a stable η¹-Rh enolate intermediate Int_EnolateA, from which they can overcome a shuttle back into the main catalytic cycle and undergo hydroacylation. Furthermore, the TSs calculated for the alternative β-hydride elimination mechanism (Pathway B), TS_BHE-S and TS_BHE-R, are slightly higher in energy than the deprotonation pathway and lie 8.9 and 10.8 kcal/mol, respectively, above Int_PRC-R. We also calculated transition structures for the racemization pathways that include a π-coordinated acrylamide (2a), which are slightly higher in energy for both Pathways A and B, due to the more sterically hindered catalyst environment (see the Supporting Information). As such, strategic use of the bulky 1-AdNH₂base permits racemization of1aprior to undergoing irreversible migratory-insertion after which epimerization ofIVand3aais higher in energy than reductive elimination and thus unlikely to occur. We conclude that although all explored pathways are accessible with respect to migratory insertion, soft enolization is the dominant mode for the racemization of starting material aldehyde (1a) prior to entering the catalytic cycle (Figure 6).

Figure 6.

Revised mechanism for Rh-catalyzed hydroacylation involving racemization of substrate 1 prior to oxidative addition and an unproductive, reversible α-migratory insertion to IV’.

Conclusion

In summary, we have developed a kinetic resolution using intermolecular hydroacylation by leveraging the chemoselective racemization of an aldehyde starting material over the ketone product with a bulky 1-AdNH₂ cocatalyst. Various ketoamides can be obtained with high control for remote 1,4- stereocenters. Mechanistic studies point to racemization pathways involving both Rh and amine and thus support a DyKAT. We expect the insights and strategies described herein, especially the use of soft enolization, will facilitate future stereoconvergent reactions that feature C–C bond construction.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.5c01753.

• Experimental procedures, computational details, and spectroscopic data for all new compounds (PDF)

Author Contributions

^# S.A.C. and J.M.W. contributed equally to this work.

The authors declare no competing financial interest.