Mechanistic Investigations of the Asymmetric Hydrogenation of Enamides with Neutral Bis(phosphine) Cobalt Precatalysts

.


INTRODUCTION
2][3][4] Alongside well-established precious-metal rhodium, iridium, and ruthenium catalysts, chiral bis(phosphine) cobalt complexes have emerged as a privileged class of catalysts for this transformation, generating highly enantioenriched alkane products from a range of prochiral alkenes (Scheme 1a).Welldefined precatalysts have been prepared across a range of cobalt oxidation states, including neutral bis(phosphine) cobalt(0), (I), and (II), 5,6 as well as cationic bis(phosphine) cobalt(I) derivatives. 7In-situ catalyst generation from a combination of bis(phosphine), readily available cobalt precursors, such as CoCl2•6H2O, and zinc enables a reliable method for catalyst discovery by high-throughput experimentation. 6Bis(phosphine) cobalt catalysts have demonstrated remarkable activity and functional group tolerance and the highly enantioselective hydrogenation of unfunctionalized olefins, 5 enamides, 6 carboxylic acids, [8][9][10] hydrazones, 11 and enynes 12 have all been reported.Notable examples include the efficient hydrogenation of pharmaceutically-relevant substrates such as dehydro-levetiracetam and the precursor to L-DOPA. 6,8][15][16][17][18] While bis(phosphine)cobalt(0) complexes are expected to be highly reducing, these catalysts exhibit high and often optimal activity and enantioselectivity in methanol and other protic solvents. 6Despite the synthetic utility and high enantioselectivity across a range of substrates, limited experimental evidence has been provided to understand the mechanism of action and origin of enantioselectivity with this class of catalysts.Open questions include: the resting state of the catalyst, the interaction of precatalysts with reactants (hydrogen and alkenes), the enantio-and rate-determining steps, and the role of cobalt redox cycling versus redox-neutral pathways.

Scheme 1. Mechanistic investigations of bis(phosphine) rhodium-and cobalt-catalyzed asymmetric enamide hydrogenation.
In contrast, the mechanism of the asymmetric hydrogenation of enamides with cationic bis(phosphine) rhodium(I) precatalysts has been extensively studied.Two major mechanistic pathways have been proposed, differentiated by the order of the reaction with respect to the reactants, enamide and hydrogen.In an unsaturated pathway, rhodium-enamide complex formation precedes the reaction with H2 (Scheme 1b, right), whereas in a dihydride pathway, rhodium-dihydride complex formation precedes the reaction with the enamide (Scheme 1b, left).Both unsaturated and dihydride pathways have been experimentally supported, with mechanistic distinctions depending on the identity of the bis(phosphine) ligand and the enamide substrate being used. 19In their landmark studies on the (DIPAMP)rhodium-catalyzed (DIPAMP = (R,R)ethylenebis[(2-methoxyphenyl)phenylphosphine]) asymmetric hydrogenation of prochiral enamides, Landis and Halpern, [20][21][22][23][24][25][26] established an unsaturated anti-"lock-and-key" pathway, where rhodium-enamide diastereomers are rapidly equilibrating in solution and enantioselection is a result of their relative rates of hydrogen activation, rather than the relative abundance of the diastereomers in solution.Specifically, Landis and Halpern found a pathway that fit the "major-minor" concept, where the minor diastereomer of the rhodium-enamide resting state led to the major enantiomer of the product, with diastereomer interconversion facilitated by formation of methanol solvento complexes (Scheme 1c, left).8][29][30][31][32][33][34][35][36] Here it is proposed that enantioselection occurs after enamide association to the cobalt complex, through partial dissociation and reassociation of the C=C bond and enantio-determining migratory insertion.0][11] In comparison to the reported studies on cationic rhodium(I) catalysis, neutral cobalt(0/II) complexes are paramagnetic.The featureless 1 H NMR spectra and the absence of any observable 31 P NMR signals due to the paramagnetism complicated their characterizations.This is in contrast to the rhodium catalysts, where diagnostic sets of 31 P NMR signals with fine splitting patterns have enabled the straightforward differentiation of rhodium diastereomers and other intermediates in solution.  To orcome this challenge, we have turned to a combination of UV-visible and EPR spectroscopies, as well as X-ray crystallography and DFT studies to characterize well-defined cobalt complexes and reaction intermediates.Here we describe mechanistic studies on the neutral, [(R,R)-( iPr DuPhos)Co]-catalyzed ((R,R)-iPr DuPhos = (+)-1,2-bis[(2R,5R)-2,5-diisopropylphospholano]benzene) asymmetric hydrogenation of simple prochiral enamides.Through organometallic studies, kinetic analysis, deuterium labeling experiments, and DFT calculations, our studies support an unsaturated pathway for [(R,R)-( iPr DuPhos)Co] complexes, with a cobalt-enamide resting state and a rate-determining oxidative hydride transfer from H2 to the enamide.Both experimental and computational evidence support the possibility of cobalt-redox cycling.
The solid-state structures of all three cobalt(enamide) complexes were determined by single-crystal X-ray diffraction (Figure 1) and established similar distorted square pyramidal geometries with two-point coordination of the substrate to cobalt at both the enamide oxygen and the C=C bond.Significant lengthening of the C=C bond was observed in all three structures (Table 1).For the square pyramidal, neutral cobalt(0) complexes, the alkene occupies the basal plane and the carbonyl oxygen occupies the axial position.By comparison, for the square-planar, cationic cobalt(I) 7 and rhodium(I) 24 complexes, both alkene and the carbonyl oxygen lie in the basal plane.This finding highlights the unique properties of asymmetric catalysis with cobalt(0) and cobalt(I) catalysts where stereo-interaction between the identical prochiral substrate and chiral ligand is altered by merely a oneelectron redox event at cobalt.While the unit cell of 1-MAA contained both the pro-(R) and pro-(S) diastereomers in a 1:1 ratio, the crystals of both 1-MAC and 1-4F MAC contained only the pro-(R) diastereomer.Interestingly, this is the opposite stereochemistry from the (S)-enantiomer obtained from catalytic hydrogenation of all three enamides with 1-COD.The paramagnetic benzene-d6 1 H NMR spectra exhibited broad, featureless resonances and accordingly, benzene-d6 solution magnetic moments each corresponded to an S=1/2 ground state (Table 1). 44The X-band EPR spectra for each complex in toluene or THF glass (10 K) exhibited a distinct rhombic signal consistent with an S = ½ cobalt complex exhibiting diagnostic hyperfine coupling of gx tensor to the 59 Co nucleus (I = 7/2, 100% natural abundance) (Figure 2).
Table 1.Magnetic data and C=C bond distances observed in cobalt(enamide) complexes.
It is important to note that observation of a particular diastereomer in the solid-state structure is not necessarily a reflection of the diastereomeric ratio of the bulk sample.Due to the uninformative nature of the NMR spectra for these cobalt(enamide) complexes and their high solubility in hydrocarbon solvents, further information about solution-state diastereomeric ratios were not obtained.However, as will be discussed later in this text, density functional theory calculations (PBE0-D3BJ, [IEFPCM(methanol)] performed on 1-MAA demonstrated only a 0.2 kcal/mol energy difference between the pro-(R) and pro-(S) diastereomers, supporting the hypothesis that they likely exist in nearly equimolar concentrations in solution and in bulk in the solid state.(1-MAC), and c) (R,R)-( iPr DuPhos)Co( 4F MAC) (1-4F MAC) at 30% probability ellipsoids with hydrogen atoms omitted for clarity, except for amide NH and alkenyl CH in the trisubstituted enamide.Observation of both the pro-(R) and pro-(S) diastereomers in the unit cell of 1-MAA and exclusively the pro-(R) diastereomer in the unit cells of 1-MAC and 1-4F MAC, despite the fact that the catalytic hydrogenations of all three substrates furnished the (S)-alkane in high enantiomeric excess, prompted further study into the relevance of these compounds to catalysis and reactivity with hydrogen.Addition of 4 atm of H2 to a MeOH solution of 1-MAA generated >95% conversion to MAA-H2 (MAA-H2 = N-acetylalanine methyl ester) in 80% enantiomeric excess (ee) (S), along with a new, unidentified paramagnetic cobalt compound (Scheme 3).Stoichiometric hydrogenation of 1-MAC with 4 atm of H2 in MeOH produced >95% conversion to MAC-H2 (MAC-H2 = N-acetylphenylalanine methyl ester) in 70% ee (S) along with the formation of a new paramagnetic cobalt product.Similarly, addition of 4 atm of H2 to a MeOH solution of 1-4F MAC gave >95% conversion and 70% ee (S).While these enantiomeric excesses are significantly lower than those obtained from the catalytic reactions and the exact ratio of cobalt-enamide diastereomers in solution is unknown, hydrogenation to primarily (S) product from a mixture of pro-(R) and pro-(S) diastereomers could indicate that cobaltenamide diastereomers are interconverting in solution, similar to the conclusions reported for cationic rhodium catalysts. 45ne possible explanation for the lower ees in the stoichiometric reaction is that diastereomer interconversion in the catalytic cycle is facilitated by exchange with free substrate in solution. 46ccordingly, addition of 10 equivalents of MAC to a benzene-d6/THF solution of isolated 1-4F MAC showed conversion to 1-MAC and liberation of free 4F MAC in a matter of minutes as observed by 1 H and 19 F NMR, demonstrating rapid exchange of 1.4668 (9)   pro-(R) pro-(S) bound enamide with free enamide in solution.Our laboratory has previously studied the kinetics of ligand exchange between 1-COD-d12 and free cyclooctadiene in solution, determining that exchange occurs through a dissociative pathway. 47The stoichiometric experiments described above demonstrated the competency of the bis(phosphine) cobaltenamide complexes to undergo productive reactions with H2, albeit to yield hydrogenated products with reduced enantiomeric excesses as compared to optimized catalytic conditions, rendering their relevance to catalysis inconclusive.Subsequently, catalytic experiments were performed to evaluate the competency of 1-enamide compounds as precatalysts.Each of the bis(phosphine) cobaltenamide complexes was evaluated as a precatalyst for the catalytic hydrogenation of its respective enamide.Catalytic hydrogenation of each enamide was performed with either 1-COD or its respective 1-enamide complex as precatalyst and gave nearly identical results in terms of both yield and enantioselectivity (Scheme 4).These results demonstrate that the bis(phosphine) cobalt-enamide complexes are competent precatalysts for asymmetric hydrogenation, providing access to the active catalytic cycle as efficiently as the previously reported precatalyst, 1-COD. 6However, additional experiments were required to determine whether the bis(phosphine) cobalt-enamide complex serves as an active catalyst species (unsaturated pathway) or simply a precatalyst that is converted to a Co(II)-dihydride (dihydride pathway, Scheme 3) or some other catalytically active species.

Synthesis, Characterization, and Evaluation of a Bis(phosphine) Cobalt(II)-Dihydride Bimetallic Complex.
To evaluate the possibility of a dihydride pathway in cobaltcatalyzed asymmetric enamide hydrogenation, the synthesis of neutral cobalt hydride complexes was targeted as these compounds are expected to be catalytic intermediates.The initial synthetic approach focused on the hydrogenation of bis(phosphine) cobalt(II) dialkyl complexes.Addition of 4 atm of H2 to a pentane solution of (R,R)-( iPr DuPhos)Co(CH2SiMe3)2 resulted in a rapid color change from orange to purple, furnishing a dark purple solid (Scheme 5).Cooling a saturated diethyl ether solution of the product at -35 ºC deposited single crystals suitable for X-ray diffraction.The solid-state structure (Figure 3) identified the product as the dinuclear cobalt complex with bridging hydrides, [(R,R)-( iPr DuPhos)Co]2(µ2-H)3(H) (2).The distance between the two cobalt atoms is 2.266(2) Å, consistent with a Co-Co bond. 48hile only three bridging hydrides were located on the Fourier difference map, the number of hydrides was determined with a Toepler pump experiment.A comproportionation reaction between two equivalents of (R,R)-(   THF was included as a cosolvent in these reactions as 2 was insoluble in MeOH.Analysis of the resulting product mixture by 1 H NMR spectroscopy demonstrated that 11% of the starting enamide (22% conversion as compared to cobalt) was converted to MAA-H2 in 55% ee favoring the (S) alkane.
Repeating this experiment in benzene-d6 did not result in detection of H2 gas as judged by 1 H NMR spectroscopy.
Similarly, treatment of 2 with four equivalents of MAC under the same conditions converted 38% of the starting enamide (76% conversion as compared to cobalt) to MAC-H2 in just 10% ee, again favoring the (S) alkane (Scheme 6).Treatment of 2 with four equivalents of 4F MAC under the same conditions also converted 26% of the starting enamide (52% conversion as compared to cobalt) to 4F MAC-H2 in just 22% ee, again favoring the (S) alkane (Scheme 6).Repeating this experiment in an open vial produced similar conversion, suggesting that enamide insertion into the dihydride occurs (rather than H2 reductive elimination, enamide coordination and reaction with stoichiometrically formed H2.Because the 1 H NMR spectrum was uninformative, EPR spectroscopy was used to determine the cobalt-containing product of enamide insertion into 2. Unsurprisingly, addition of 10 equivalents of 4F MAC with 2 in a MeOH/THF (1:1) solution for one hour generated an EPR spectrum that matched that for isolated 1-4F MAC.This result demonstrates that upon reduction of Co(II) to Co(0) and alkane release, free enamide coordinates to the Co(0) center.While these reactions demonstrate some competence of 2 towards alkene insertion, the relatively low ees obtained from these insertion reactions raised questions about the relevance of 2 to catalytic hydrogenation.

Scheme 6. Reactions of enamides with [(R,R)-( iPr DuPhos)Co]2(µ2-H)3(H).
The dicobalt tetrahydride, 2 was also evaluated as a precatalyst for the asymmetric hydrogenation of each enamide.Reaction of 4 atm of H2 with a 0. The slightly diminished enantioselectivities demonstrated in all three catalytic reactions as compared to those catalyzed by 1-COD or any of the 1-enamide complexes, suggest that 2 may perform a single-turnover insertion event with low selectivity before entering the active catalytic cycle.In-Situ Reaction Monitoring to Determine the Catalyst Resting State.Stoichiometric and catalytic experiments demonstrated that both cobalt-enamide and cobalt-dihydride complexes were synthetically accessible and were competent towards their respective stoichiometric reactions (hydrogenation or enamide insertion, albeit with lower enantiomeric excesses), and were effective precatalysts.Therefore, additional experiments were required to differentiate between potential pathways.Determination of the catalyst resting state is also critical for distinguishing these pathways.Due to the paramagnetism of the cobalt(0) and (II) complexes, a combination of experiments was required to determine the catalyst resting state (Scheme 7a) as the 1 H NMR spectra were broad and featureless.Due to the distinct absorbance features exhibited by the previously isolated cobalt precatalysts, reaction monitoring by UV/visible spectroscopy was performed (Scheme 7b).A gasadapted UV/visible cuvette was used for these measurements (see SI for details). 50,51With 3 mol% of 1-COD as the precatalyst, the hydrogenation of 0.10 mmol of 4F MAC was monitored under 1 atm of H2 at 23 ºC.Rapid consumption of 1-COD was observed, as was a new cobalt complex with an absorption maximum centered in the solvent region.This peak corresponded to the same spectroscopic feature measured for independently prepared 1-4F MAC, supporting the role of this compound as the catalyst resting state.Analogous experiments with 1-COD for the hydrogenation of MAA generated an absorption maximum in the solvent region, consistent with the spectroscopic signatures for independently prepared 1-MAA.
To provide additional support for these experiments, freezequench X-band EPR experiments were conducted on active catalytic mixtures (Scheme 7c).Specifically, a J. Youngadapted EPR tube was charged with a 1:1 THF/MeOH solution of 4F MAC and 1-COD (5 mol%).Addition of 4 atm of H2 was followed by several minutes of vigorous shaking to ensure thorough mixing of H2 and the contents were then frozen in liquid nitrogen.The X-band EPR spectrum of the frozen glass at 10 K exhibited a distinct rhombic feature with diagnostic 59 Co hyperfine coupling.Simulation of the experimental spectrum established that the major feature observed was identical to the EPR spectrum of authentic 1-4F MAC, confirming the ( iPr DuPhos)cobalt-enamide complex as the catalyst resting state.Importantly, repeating an identical freeze-quench EPR experiment with catalyst initiation from 2, rather than 1-COD, similarly demonstrated 1-4F MAC as the catalyst resting state.This result corroborates the conclusions of the previous section, that while 2 may be able to enter the active catalytic cycle through subsequent formation of 1enamide, 2 is not an on-cycle intermediate.Performing an identical experiment with MAC and 1-COD demonstrated that 1-MAC was the major feature also supporting the cobalt enamide compound as the catalyst resting state.Scheme 7. Catalyst resting state analysis using UV/visible and EPR spectroscopies.
Taken together, these experiments establish bis(phosphine) cobalt-enamide complexes as the catalyst resting states in the catalytic hydrogenation reactions with all three enamides used in this study.Cobalt-enamide complexes are expected to be intermediates along the unsaturated pathway (where cobaltenamide complexation occurs before hydrogen activation), but not along a dihydride pathway.Therefore, the observation of cobalt-enamide complex as the catalyst resting state suggests that a dihydride pathway may not be relevant in the catalytic hydrogenation reaction.Rate Law for the Bis(Phosphine) Cobalt-Catalyzed Asymmetric Hydrogenation of Enamides.To gain additional insight into the mechanism of cobalt-catalyzed asymmetric hydrogenation, the experimental rate law for the hydrogenation of 4F MAC was determined.Initial experiments were performed at relatively low (1-4 atm) H2 pressures in J-Young NMR tubes with constant gas mixing to avoid complications from mass transfer.was used as the precatalyst, modeling the rhodium(I) kinetic analysis described by Landis and Halpern,26 in an attempt to minimize complications from precatalyst activation.Variable time normalization analysis (VTNA) of reaction time courses as described by Burés were applied 53 in which the initial concentrations of 4F MAC, H2, and 1-4F MAC were systematically varied.Analysis of the resulting overlay plots established the rate law show in in eq 1 (Figure 4).rate To verify that this rate law is maintained at higher pressures of H2 that are used under more optimized conditions, reaction progress experiments were performed in a high-pressure reactor.Due to the higher gas pressure, precatalyst loadings were lowered and varied between 0.5 and 2 mol%, 4F MAC concentrations were varied between 2.1x10 -4 and 8.4x10 -4 M, and the H2 pressure was maintained at 500 psi.Each measurement was made in triplicate.Data points were obtained after quenching the hydrogenation after 5 minutes and filtration through alumina and quantification of conversion by 19 F and 1 H NMR spectroscopies.The data also supported the rate law presented in equation 1 with a positive dependence on catalyst loading and no dependence on substrate concentration.Identical reactions at 100 and 200 psi demonstrated a positive dependence on the concentration of H2, again in agreement with equation 1.
The rate law presented in eq 1 is consistent with a ratedetermining step involving activation of H2 by the cobalt catalyst.The pseudo-zero order dependence on enamide concentration (rather than an inverse order), in combination with the observed cobalt-enamide resting state, suggests enamide saturation kinetics rather than a mechanism where complete enamide dissociation is required prior to activation of H2.These results are therefore inconsistent with a dihydride pathway, in which cobalt reacts with H2 prior to enamide binding.On the other hand, these results could support an unsaturated pathway, in which formation of a cobalt-enamide complex occurs prior to rate-limiting H2 activation.These kinetic results are consistent with those reported by Landis and Halpern in their studies of rhodium(I)-catalyzed asymmetric alkene hydrogenation. 26H2 Pressure Effects.A key observation by Landis and Halpern in their work on the rhodium-catalyzed unsaturated hydrogenation pathway was an inverse dependence of pressure on product enantioselectivity arising from competition between diastereomer interconversion and H2 activation.Accordingly, H2 pressure dependence studies were performed on both the stoichiometric hydrogenation of 1enamide and the catalytic hydrogenation of enamides by 1-COD.
Figure 4. Overlay plots obtained from VTNA for the determination of the rate law for the hydrogenation of 4F MAC with H2 with 1-4F MAC as the precatalyst: (a) order in 4F MAC, (b) order in 1-4F MAC, (c) order in H2 pressure.
The hydrogenation of identical samples of isolated 1-4F MAC in MeOH was performed at 100, 200, and 400 psi in a highpressure reactor, and gave enantioselectivities of 83%, 80%, and 75% ee (S), respectively (Table 2).The observation of higher enantioselectivites at lower pressures is consistent with competitive rates of diastereomer interconversion and H2 activation.In contrast, catalytic experiments involving the hydrogenation of enamide by 1-COD (3 mol%) in MeOH showed little to no dependence of the enantioselectivity on the H2 pressure.Hydrogenation of MAA at 100, 200, 300, and 400 psi gave corresponding enantioselectivities of 95%, 95%, 94%, and 90% ee (S).Hydrogenation of MAC and 4F MAC gave consistent enantioselectivities of 96% and 95% ee (S), respectively, across a pressure range of 100-400 psi.These results suggest that under standard catalytic conditions with a large excess of substrate, the rate of diastereomer interconversion is significantly faster than the rate of H2 activation, such that the observed enantioselectivity exhibits little dependence on H2 pressure.Based on this model, performing the hydrogenation reaction at low substrate concentration and high H2 pressure should produce a lower enantioselectivity as compared to the standard catalytic reaction.Accordingly, hydrogenation of 4F MAC by 1-COD (20 mol%) at 400 psi of H2 resulted in diminished  Catalytic hydrogenations were also performed with hydrogen deuteride (HD) to probe directional specificity in the enamide hydrogenation.All three substrates were hydrogenated using 1-COD as the precatalyst in MeOH. 1 H and 2 H NMR spectroscopies were used to determine relative proton and deuterium incorporations in the and -positions.
Quantitative 13 C NMR spectroscopy and high-resolution mass spectrometry were used to determine the relative quantities of H2, HD, DH, and D2 products (see SI for details).For MAA, addition of 4 atm of HD to a solution of MAA (0.10 M) and 2 mol% of 1-COD at room temperature for 16 hours resulted in preferential deuterium incorporation into the over the position in a 1.45:1 ratio (Scheme 8c).To rule out the possibility that the preferential incorporation was the result of reversible alkyl hydride formation in methanol, the reaction was repeated in THF and similar 1.28:1 ratio for deuterium incorporation into the over -positions was observed, respectively.Identical experiments with MAC and 4F MAC with 1-COD in MeOH showed similar preference for the position (1.32:1 and 1.33:1 respectively).For comparison, the catalytic hydrogenation of MAA and MAC by [Rh(DIPHOS)(NBD)][BF4] was reported to give a 1.21:1 and 1.33:1 preference, respectively, for the -position in THF. 49ese results support an irreversible H2 addition step to the cobalt-enamide complex. 49

Mechanistic Interpretation of Experimental Results.
Taken together, the experimental results provide insights into the mechanism of cobalt-catalyzed asymmetric hydrogenation.Stoichiometric and catalytic reactions with both cobalt-enamide (1-enamide) and cobalt-dihydride (2) complexes demonstrated their catalytic competency and as such provide little distinction about whether a dihydride or an unsaturated pathway is operative.The observation of a cobalt-enamide resting state during catalytic hydrogenation suggests that enamide coordination to the cobalt occurs prior to H2 activation, favoring an unsaturated over a dihydride pathway.Reaction kinetic analyses support saturation in enamide, and a rate-determining step involving H2 activation by a cobalt-enamide complex.Again, this supports an unsaturated pathway, whereby formation of the cobaltenamide complex formation precedes H2 activation.enantioselectivity under stoichiometric, catalytic, and intermediate conditions where the substrate to catalyst ratio is changed, established that in the presence of large excesses of substrate, interconversion of cobalt-enamide diastereomers occurs at a rate that is much faster than H2 activation.Therefore, product enantioselectivity is a reflection not of the relative abundance of cobalt-enamide diastereomers in solution, but rather of their relative rates of reaction with H2, consistent with the anti-"lock-and-key" concept.These observations are also consistent with the observation of the "wrong" pro-(R) diastereomer in the unit cells of all three 1-enamide solid-state structures, although paramagnetism precludes determination of solution-state diastereomeric ratios.Deuterium labeling experiments demonstrate that the protons transferred to the product are derived from hydrogen gas, with a KIE similar to those observed previously in rhodium unsaturated pathways.Likewise, HD partitioning ratios are indicative of a 2,1-insertion step.Taken together, each of these results is consistent with an unsaturated pathway, while many results are inconsistent with a dihydride pathway.
It is important to note that other pathways, aside from the idealized limits of dihydride and unsaturated pathways are also possible.In contrast to the rapid equilibration of cobaltenamide diastereomers prior to H2 activation, it is possible that interchange between the pro-(R) and pro-(S) pathways occurs following activation of H2, such as has been proposed by Gridnev and Imamoto. 35In this case, enantiointerconversion occurs after H2 activation but before migratory insertion through reversible partial dissociation and reassociation of the enamide to the metal through the C=C bond.However, this mechanism is inconsistent with the observed dependence of cobalt-enamide hydrogenation enantioselectivity on hydrogen pressure.It is also possible that partial enamide dissociation takes place earlier in the catalytic cycle, such as in the H2 activation step.S94), in agreement with experimental findings.Thus, hydrogenation pathways begin from 1-MAA.
The computed structure of 1-MAA has the enamide coordinated to the cobalt by the C=C double bond and the oxygen of the amido group, as observed in the X-ray structure of 1-MAA (Figure 1).Interestingly, there is only a 0.2 kcal/mol energy difference between the pro-(R) and pro-(S) diastereomers of 1-MAA, suggesting that they will be formed in approximately equal quantities (Figure S96), which may explain the crystallographically observed diastereomeric mixture of 1:1 (Figure 1a).Calculations establish a high dissociation energy of 56.6 kcal/mol for the MAA substrate (Figure S95), indicating a strong Co-enamide interaction.Attempts to calculate structures with partial dissociation of the enamide in the presence of ligating methanol were unsuccessful, however, enamide coordination solely through the carbonyl groups leads to a state 23.8 kcal/mol higher in energy than 1-MAA, indicating that such structures are not relevant during the catalytic cycle.With 1-MAA as the starting point, three mechanistic pathways were considered: a classical redox Co(0/II) mechanism A, a Co(0/II) imine mechanism B, 55 and a nonredox metallacycle pathway C where cobalt exhibits the +2 oxidation state during the whole catalytic cycle. 56The DFT calculations were performed with an explicit MeOH molecule added to the model, as we have shown for related calculations on (R,R)-( Ph BPE)Co(MAA) that inclusion of a polar solvent molecule stabilizes the charges at the rate-limiting transition states and thus affects the final energies and enantioselectivities. 57,58 The first considered pathway is the redox-type mechanism A (Scheme 9).In the initial step of mechanism A, coordination of H2 to 1-MAA occurs through TS-H2 to form a 1-MAA-dihydrogen (-bound H2) intermediate.Importantly, H2 does not undergo oxidative addition to form a Co(II)-dihydride, as formation of the -complex is preferred by 21.5 kcal/mol over the dihydride (Figure S97).Subsequently, oxidative hydride transfer to the -carbon occurs to give a 1-alkyl-hydride intermediate (Figure 5).This step is found to be rate-Scheme 9. Calculated Co(0)/(II) redox cycling pathway for the hydrogenation of MAA (mechanism A) (barriers relative to 1-MAA, 298K, PBE0-D3BJ/IEFPCM(methanol)).
limiting for mechanism A with an overall barrier of 25.1 kcal/mol for the formation of the (S)-product and 27.9 kcal/mol for the formation of the (R)-product.In the last step, reductive elimination releases the product and regenerates the 1-MAA species.An overall barrier of 25.1 kcal/mol for the (S)-pathway would be feasible at the experimental temperature of 298 K (based on the discussion by Baik and coworkers, a reasonable conversion at 298 K requires a barrier around 25 kcal/mol or lower). 59Analysis of the optimized geometries shows that the preference for the (S)-TS-Hyd transition state can be ascribed to stronger interactions between cobalt and a carbonyl of the enamide, as well as more favorable C-H…O interactions, together with less steric clash with the i Pr substituents on the DuPhos ligand compared to (R)-TS-Hyd (Figure 5, bottom).The computed ee of 98.2% for MAA is in excellent agreement with the experimental ee of 98.5% (S).The computed energies and selectivities support redox-type mechanism A, which also agrees well with other experimental results (vide supra), including a diastereomeric cobaltenamide resting state and a first-order dependence on the H2 concentration.In addition, the results of the HD Preferential olefin insertion into the M-H versus M-D bond in the initial step during HD hydrogenation is wellprecedented in rhodium catalysis. 28,29,42,49With cobalt, the preferential deuterium incorporation in the -position also supports formation of 1-alkyl-hydride following oxidative hydride transfer.Two other mechanistic pathways beginning from 1-MAA were also computed but are ruled out (see SI for details).Both redox imine mechanism B (Figure S98) and non-redox metallacycle mechanism C (Figure S99) have overall computed barriers above 30 kcal/mol, inconsistent with the room temperature reaction conditions used experimentally.Additionally, both pathways would predict the wrong, (R) hand of the product as compared to the (S)-product obtained from the experimental catalytic reaction and redox mechanism A. These results are consistently reproduced by three different DFT protocols (SI, Table S7).Analaysis of the alternative substrate 4F MAC also supported feasible barriers for mechanism A (Table S8), alongside a computed H2/D2 KIE of 1.33 (Table S9), which is in good agreement with the experimentally observed value of 1.2(2) (Scheme 8).We note that the computational results for (R,R)-( iPr DuPhos)Co differ from the related (R,R)-( Ph BPE)Co complex, where mechanism C appears to be energetically accessible for hydrogenation of MAA. 57o summarize, computational results support that (R,R)-( iPr DuPhos)Co-catalyzed hydrogenation of MAA proceeds by an unsaturated Co(0)-Co(II) redox pathway (Scheme 9), which does not involve formation of a distinct Co-dihydride species.These conclusions are consistent with experimental results, particularly that 1-MAA is the lowest energy species along the catalytic cycle and that H2 addition to the substrate is rate-determining.Agreement is also found with the experimental enantiomeric excess and the results of experimental hydrogenation of MAA with HD.

CONCLUSIONS
Experimental and computational mechanistic investigations on the neutral [(R,R)-( iPr DuPhos)Co]-catalyzed hydrogenation of prochiral enamides have been performed.Independent syntheses established that both bis(phosphine) cobaltenamide and -hydride complexes are stable, isolable species, which could be used productively in both catalytic and stoichiometric reactions (with diminished enantiomeric excesses in the stoichiometric reactions).Catalyst resting state analysis using UV-visible and EPR spectroscopies revealed a cobalt-enamide resting state for all three representative enamides, supporting an unsaturated pathway for enamide hydrogenation, whereby reaction with enamide precedes H2 activation.Catalytic reaction kinetics in MeOH using VTNA techniques established a first order dependence on H2 and bis(phosphine) cobalt-enamide concentrations, with a zeroorder dependence on enamide substrate concentration.These results are consistent with reaction of H2 with a bis(phosphine) cobalt-enamide complex as the rate-determining step, similarly in agreement with an unsaturated pathway.Deuterium-labeling experiments demonstrated a homolytic pathway for H2 cleavage and H2 versus D2 kinetic isotope effects and HD partitioning experiments are consistent with irreversible and regioselective hydride transfer to the cobaltenamide complex.Computations using MAA as representative substrate support an unsaturated pathway with 1-MAA as the lowest energy intermediate and a ratedetermining oxidative hydride transfer from a -bonded 1-MAA-dihydrogen complex to the -position of the substrate, a key difference from the rhodium unsaturated pathway involving oxidative addition of H2.Both the computational results and deuterium labeling experiments demonstrate that cobalt redox-cycling is reasonable for enamide hydrogenation with [(R,R)-( iPr DuPhos)Co]-complexes.
Overall, an unsaturated pathway by Co(0/II) redox cycling has been elucidated for the cobalt-catalyzed asymmetric hydrogenation of enamides.Although the term "unsaturated" as a general mechanistic descriptor is used to signal substrate coordination prior to H2 activation, several distinctions are noted between the Co(0/II) and Rh(I/III) unsaturated mechanisms, including the charge, oxidation state and geometries of intermediates, mode of H2 activation, and the interactions responsible for enantioselection.Nonetheless, many fundamental properties in reactivity and selectivity are consistent between the reaction mechanisms, establishing shared guiding principles for catalytic reaction design.
Figure 1.Solid-state structures of a) (R,R)-( iPr DuPhos)Co(MAA) (1-MAA), b) (R,R)-( iPr DuPhos)Co(MAC) (1-MAC), and c) (R,R)-( iPr DuPhos)Co( 4F MAC) (1-4F MAC) at 30% probability ellipsoids with hydrogen atoms omitted for clarity, except for amide NH and alkenyl CH in the trisubstituted enamide.Observation of both the pro-(R) and pro-(S) diastereomers in the unit cell of 1-MAA and exclusively the pro-(R) diastereomer in the unit cells of 1-MAC and 1-4F MAC, despite the fact that the catalytic hydrogenations of all three substrates furnished the (S)-alkane in high enantiomeric excess, prompted further study into the relevance of these compounds to catalysis and reactivity with hydrogen.Addition of 4 atm of H2 to a MeOH solution of 1-MAA generated >95% conversion to MAA-H2 (MAA-H2 = N-acetylalanine methyl ester) in 80% enantiomeric excess (ee) (S), along with a new, unidentified paramagnetic cobalt compound (Scheme 3).Stoichiometric hydrogenation of 1-MAC with 4 atm of H2 in MeOH produced >95% conversion to MAC-H2 (MAC-H2 = N-acetylphenylalanine methyl ester) in 70% ee (S) along with the formation of a new paramagnetic cobalt product.Similarly, addition of 4 atm of H2 to a MeOH solution of 1-4F MAC gave >95% conversion and 70% ee (S).While these enantiomeric excesses are significantly lower than those obtained from the catalytic reactions and the exact ratio of cobalt-enamide diastereomers in solution is unknown, hydrogenation to primarily (S) product from a mixture of pro-(R) and pro-(S) diastereomers could indicate that cobaltenamide diastereomers are interconverting in solution, similar to the conclusions reported for cationic rhodium catalysts.45One possible explanation for the lower ees in the stoichiometric reaction is that diastereomer interconversion in the catalytic cycle is facilitated by exchange with free substrate in solution.46Accordingly, addition of 10 equivalents of MAC to a benzene-d6/THF solution of isolated 1-4F MAC showed conversion to 1-MAC and liberation of free 4F MAC in a matter of minutes as observed by 1 H and19 F NMR, demonstrating rapid exchange of

Figure 3 .
Figure 3. Solid-state structure of [(R,R)-( iPr DuPhos)Co]2(µ2-H)3(H) (2) at 30% probability ellipsoids with hydrogen atoms, except for the bridging cobalt-hydrides, omitted for clarity.Isolated 2 was evaluated for its reactivity towards enamide insertion.Addition of 4 equivalents of MAA to a 1:1 THF-MeOH solution of 2 after 24 hours at room temperature resulted in a color change from deep purple to pale brown.THF was included as a cosolvent in these reactions as 2 was insoluble in MeOH.Analysis of the resulting product mixture by 1 H NMR spectroscopy demonstrated that 11% of the starting enamide (22% conversion as compared to cobalt) was converted to MAA-H2 in 55% ee favoring the (S) alkane.Repeating this experiment in benzene-d6 did not result in detection of H2 gas as judged by 1 H NMR spectroscopy.Similarly, treatment of 2 with four equivalents of MAC under the same conditions converted 38% of the starting enamide (76% conversion as compared to cobalt) to MAC-H2 in just 10% ee, again favoring the (S) alkane (Scheme 6).Treatment of 2 with four equivalents of 4F MAC under the same conditions also converted 26% of the starting enamide (52% conversion as compared to cobalt) to 4F MAC-H2 in just 22% ee, again favoring the (S) alkane (Scheme 6).Repeating this experiment in an open vial produced similar conversion, suggesting that enamide insertion into the dihydride occurs (rather than H2 reductive elimination, enamide coordination and reaction with stoichiometrically formed H2.Because the 1 H NMR spectrum was uninformative, EPR spectroscopy was used to determine the cobalt-containing product of enamide insertion into 2. Unsurprisingly, addition of 10 equivalents of 4F MAC with 2 in a MeOH/THF (1:1) solution for one hour generated an EPR spectrum that matched that for isolated 1- -situ reaction monitoring reveals cobalt-enamide resting state.b) Reaction monitoring by UV-visible spectroscopy in MeOH shows formation of 1-4F MAC under catalytic hydrogenation conditions.c) Freeze-quench EPR spectroscopy shows formation of 1-4F MAC under catalytic hydrogenation conditions in THF/MeOH.
ee (S), as compared to the standard catalytic reaction.

Figure 5 .
Figure 5. Optimized pro-(S) and pro-(R) transition states for oxidative hydride transfer for (R,R)-( iPr DuPhos)Co-catalyzed hydrogenation of MAA by redox mechanism A, with a hydrogen-bonded MeOH (barriers relative to 1-MAA, 298K, PBE0-D3BJ,IEFPCM[methanol]).The computed energies and selectivities support redox-type mechanism A, which also agrees well with other experimental results (vide supra), including a diastereomeric cobaltenamide resting state and a first-order dependence on the H2 concentration.In addition, the results of the HD hydrogenation experiments are quantitatively reproduced by Mechanism A. Based on the difference in zero-point energies, cobalt should have a larger difference in bond strength and a larger KIE for Co-D versus Co-H, as compared to C-D versus C-H.Therefore, the predicted oxidative hydride transfer to the -position of 1-MAA-dihydrogen in the ratedetermining TS-Hyd should result in preferential transfer of H to the -carbon and D to the cobalt, ultimately resulting in higher incorporation of deuterium in the -position of the product.Calculations on TS-Hyd with replacement of one hydrogen by deuterium support a 1.17:1 preference for initial H transfer versus initial D transfer from HD to the C of MAA, in agreement with the experimental results.