Highly Selective Hydrogenation of C═C Bonds Catalyzed by a Rhodium Hydride

Abstract

Under mild conditions (room temperature, 80 psi of H₂) Cp*Rh(2-(2-pyridyl)phenyl)H catalyzes the selective hydrogenation of the C═C bond in α,β-unsaturated carbonyl compounds, including natural product precursors with bulky substituents in the β position and substrates possessing an array of additional functional groups. It also catalyzes the hydrogenation of many isolated double bonds. Mechanistic studies reveal that no radical intermediates are involved, and the catalyst appears to be homogeneous, thereby affording important complementarity to existing protocols for similar hydrogenation processes.

Graphical Abstract

INTRODUCTION

The catalytic hydrogenation of unsaturated organic compounds is a transformation of widespread importance in both academia and industry, particularly as applied for the preparation of pharmaceuticals, fragrances, and other fine chemicals.¹ The 1,4-reduction of α,β-unsaturated carbonyl compounds has attracted particular attention, and much effort has been devoted to its development.² Hydrogen sources other than molecular hydrogen,³ including silicon hydrides,⁴ formates,⁵ and alcohols,⁶ are often efficient and have become widely used.^7,8 To date, the majority of methods possessing both high efficiency and selectivity have used transition metal promoters (Ir,⁶ Pd,⁹ Co,^10,11 Ni,¹² and others¹³) with a variety of supporting ligands,^14,5 though there have also been recent disclosures of transition-metal-free hydrogenation reagents involving Se powder,¹⁵ borane,¹⁶ and electrons.^17–19

Nevertheless, most 1,4-reductions still suffer from drawbacks, such as poor tolerance of sensitive functional groups and a lack of effectiveness with highly substituted, sterically encumbered substrates. Indeed, such steric constraints can sometimes lead to an undesired stereochemical outcome and/or prevent hydrogenation entirely. Such issues arose in the Snyder group’s recent total syntheses of the coccinellid alkaloids, including targets such as exochomine,²⁰ arborisidine,²¹ and chilocorine C.^22,23 For example, one of the final steps in the exochomine work required the selective 1,4-reduction of the hindered enone 1 to 2 (Scheme 1A). However, the dithiolane group, benzylic ketone, and acyl pyrrole all proved prone to reduction and/or side-product formation; the desired product was best obtained by reducing 1 with silanes in the presence of stoichiometric Mn(dpm)₃, a procedure adapted from those reported by Magnus²⁴ and Shenvi.²⁵ Given this, and other related, examples, a robust catalytic procedure for the 1,4-reduction of α,β-unsaturated carbonyl compounds using H₂ was viewed as highly desirable.

Scheme 1.

(A) A Challenging Substrate for 1,4-Hydrogenation, with Alternative Sites of Reaction; (B) Chirik Proposal for Mechanism of Action of (η⁵-C₅Me₅)Rh(ppy)H; (C) Homogeneous Hydrogenation of α,β-Unsaturated Carbonyl Compounds and Isolated Olefins Catalyzed by RhH-1

The Norton group initially attempted to employ the Cr [CpCr(CO)₃H] and Co [Co(dmgBF₂)₂L₂] (L = H₂O, THF, etc.) catalysts that they had used^26–30 for H· transfer from H₂ to solve this problem, but neither gave any 2 from 1. We then considered (η⁵-C₅Me₅)Rh(ppy)H (ppy = 2-(2-pyridyl)-phenyl), RhH-1, developed in the Norton laboratory and shown to be a fast hydride and hydrogen atom donor, but a relatively poor proton donor (Scheme 1B).^31,32 Related Cp*Rh systems have been shown to effectively catalyze arene and olefin hydrogenation.^33–35 Indeed, in 2019 the Chirik group found that RhH-1 can catalyze the hydrogenation to ammonia of amides,³⁶ nitrides,³⁷ and related ligands.³⁸ Very recently, the use of the same precatalyst for the hydrogenation of N-heteroarenes has been reported by the same group.³⁹ They have proposed that, upon heating or irradiation, the reductive elimination of 2-phenylpyridine from RhH-1 can lead to the formation of catalytically active multimetallic clusters (and eventually nanoparticles), under varied H₂ pressures (4–36 atm) at elevated temperatures (80–100 °C).

To our delight, we found that (η⁵-C₅Me₅)Rh(ppy)H (RhH-1) does indeed show activity for the hydrogenation of the C═C bonds of enones. Herein we describe a highly selective and mild procedure for catalyzing the C═C hydrogenation of α,β-unsaturated carbonyl compounds and isolated olefins (Scheme 1C) which works on an array of substrates with high chemoselectivity and functional group tolerance. We follow-up these studies of scope with mechanistic investigations which reveal that our catalyst appears to be behaving in a homogeneous, rather than heterogeneous, manner.

RESULTS AND DISCUSSION

As shown in Table 1, we selected chalcone 3 as a test enone to develop and optimize our rhodium-catalyzed 1,4-hydrogenation method. With 3 mol % RhH-1 and 80 psi of H₂ gas in MeOH (0.05 M) at 23 °C, the reaction took 24 h to reach completion, affording reduction product 4 in 94% isolated yield (Table 1, entry 1). Lowering the catalyst loading or the pressure of H₂ gas eroded the yield during the same time period (entries 2 and 3). Further, changing the catalyst to the benzo[h]quinoline derivative RhH-2 gave a slightly lower yield (entry 4), while the use of solvents other than MeOH also proved deleterious (entries 5–7). As shown by control experiments, both the H₂ gas and the rhodium promoter are essential (entries 8 and 9).

Table 1.

Optimization of the Reaction Conditions^a

^a3 (0.20 mmol), RhH-1 catalyst (3 mol %), H₂ (80 psi), MeOH (4 mL) at room temperature, 24 h.

^bNMR yields using CH₂Br₂ as internal standard.

^cIsolated yield.

As shown by the reaction scope in Table 2, the method displays excellent chemoselectivity with various α,β-unsaturated carbonyl compounds, which in all cases underwent 1,4-reduction exclusively to form the indicated products in high yields (with the colored bond marking the site of hydrogenation). As can be discerned, chalcones containing both electron-rich and electron-poor arenes are reduced appropriately, to 4–9, and the reduction of the precursor to 5 can be scaled up without compromising the overall yield. Related substrates containing aromatic heterocycles such as imidazole or thiophene also react smoothly, giving good yields of 10 and 11. Vinyl phenyl ketones with substituents at the α or β position also undergo 1,4-hydrogenation, making 12 and 13 in good yield. In addition, the vinyl methyl ketones in the substrates leading to 14–16 are selectively reduced in excellent yields, while the trisubstituted double bonds in 15 and 16 remain untouched. No 1,6-reduction product was detected along with 16. Pleasingly, the α,β-unsaturated esters, vinyl amide, and vinyl sulfones within products 17–22 all posed no problems even with steric hindrance at the α position (as in the precursor of 17). The substituted cyclic, α,β-unsaturated ester in 22 was not reduced.

Table 2.

Scope of the Rhodium-Catalyzed Hydrogenation of α,β-Unsaturated Carbonyl Compounds^a,b

^aUnder the conditions of entry 1 in Table 1.

^bIsolated yields, average of two independent runs.

^cUsing ⁱPrOH as solvent instead of MeOH.

^dGram scale (10 mmol reaction).

Critically, the scope of the Rh-catalyzed hydrogenation extends to dienes and to cyclic enones. Both of the conjugated double bonds leading to 23 and 24 were hydrogenated with high efficiency, giving these materials in almost quantitative yields. The C═C bonds of cyclic enones are also hydrogenated in 1,4-fashion (leading to 25–28). The cyclic enones found in products 29 and 30, possessing β ethoxy substituents, are not hydrogenated, although the isolated C═C bonds are. The chemoselectivity of the Rh-catalyzed hydrogenation is further illustrated by the fact that acetals (24 and 26), esters (22 and 25), aryl halides (21 and 23), and even unprotected alcohol (6 and 28) are well tolerated, leaving ample room for further derivatization, as desired.

Although we did not observe byproducts with 1,2-reduction for any of the substrates used in Table 2, we did find that the 1,4-reduction products (34–36) from α,β-unsaturated aldehydes (31–33) undergo slow, further 1,2-reduction to afford 37–39 (Table 3). We note that both aliphatic and aromatic substituents seem to be tolerated at the β position. Of particular interest, the hydrogenation of the intermediate aldehyde is considerably slower than the 1,4-hydrogenation of the initial enone, as judged by the reaction times required.

Table 3.

Sequential Reductions & Synthetic Applications^a,b

^aUnder the conditions of entry 1 in Table 1.

^bIsolated yields, average of two independent runs.

^cTriethylamine (1.0 equiv) was added.

With these initial results in hand, we then returned to the highly substituted substrates that had caused difficulty for the Snyder group in their exochomine synthesis (cf. Scheme 1A).²⁰ For example, Stryker’s reagent (H₆Cu₆L₆) had given a sluggish reaction, with the principal product being the result of 1,6-reduction across the pyrrole ring. A similar 1,6-reduction result was obtained after one-electron reduction by SmI₂; by contrast, catecholborane and DIBAL-H gave the 1,2-reduction product, while DIBAL-H with Cu(I), RedAl, Pd°/n-Bu₃SnH, and sulfonylhydrazides (NBSH) gave no reaction. Specifically, we tried to hydrogenate somewhat simpler predecessors of 40 and 41 with our rhodium catalyst RhH-1 under our optimal conditions, and found that selective 1,4-hydrogenation of the C═C bonds of these two enones could be achieved in the presence of a dithiolane and a pyrrole, providing both 40 and 41 in high yields and diastereoselectivities (Table 3). The presence of the tert-butyl substituents caused no issues in these transformations, and the addition of Et₃N did not suppress the formation of 40 (suggesting that the tertiary amine found in exochomine itself would not be problematic if executed on even more advanced intermediates).

In view of the hydrogenation of the isolated double bonds leading to 29 and 30, we have further explored the utility of RhH-1 as a catalyst for the hydrogenation of other such olefins (cf. Table 4). As shown, carbon–carbon double bonds with a variety of electronic and steric properties gave high yields of the hydrogenated products 42–49. Functional groups that are not affected under these conditions include a boronic ester, a brominated arene, an unprotected indole, and free alcohols (both primary and tertiary). The trisubstituted C═C bond that belongs to the allylic alcohol precursor of 49 was reduced, but the remote, trisubstituted double bond remained untouched; in this case, the hydroxyl group might be serving as a directing group. Complete reduction of diphenylacetylene to 50 was observed in 6 h, while the terminal alkyne in mestranol is hydrogenated to afford 51 in high yield (94%).

Table 4.

Hydrogenation of Isolated Olefins^a,b

^aUnder the conditions of entry 1 in Table 1.

^bIsolated yields, average of two independent runs.

^cReaction time is 6 h.

The observed chemoselectivity of our Rh-catalyzed hydrogenation reactions suggested to us that they may also be useful in late-stage reductions during the synthesis of fine chemicals and/or pharmaceutical agents. Indeed, we can carry out such reactions (Scheme 2): for example, cyproheptadine is exclusively hydrogenated at the less substituted double bond, producing a good yield of the reduced pharmaceutical 52; reduction of the trisubstituted olefin in brucine delivers 53 in 94% yield. The late-stage hydrogenation of levofloxacin was similarly achieved with ease to give 54 in high yield, along with decarboxylation of the β-keto acid that occurs with standard hydrogenation protocols.⁴⁰

Scheme 2.

Late-stage Hydrogenation of Pharmaceuticals

In view of the effectiveness of RhH-1 in catalyzing the hydrogenation of unactivated alkenes, we have attempted its use on the exomethylene of 55, a substrate whose stereochemical outcomes in catalytic hydrogenation have been investigated in detail by Shenvi.²⁵ As shown in Scheme 3, we obtained in THF mostly the cis-disposed reduction product 56 (which is also the predominant product with traditional catalysts).^41–43 We obtained a near-equimolar mixture of the epimers 56 and 57 when the reaction was conducted in MeOH. Under no conditions did our catalyst prefer to form the more stable product 57, one which Shenvi has found to be the kinetic product with Mn(dpm)₃ or Co(dpm)₂ as the catalyst.⁴⁴

Scheme 3.

Effect of Solvent on Exomethylene Reduction

The outcome of such hydrogenation reactions is determined by both substituent effects and by solvent. With β-pinene (58), for example, there is no hydrogenation in MeOH, but in THF the equatorial methyl product 59 is obtained in 56% yield along with 44% of the isomerization product 60. The isomerization surely arises from the reversibility of the olefin insertion.

MECHANISM

In order to probe the difference between the mechanism of enones and that of isolated olefins, we have compared the results of deuterium labeling experiments with the α,β-unsaturated ketone in 61 with the results of such experiments with the isolated terminal olefin in 63. The extent of label incorporation is shown in Scheme 4. When CD₃OD was used as solvent, 0.75 D was transferred to the α position of 62, while no deuterium was detected in 64. The protonation of a Rhenolate intermediate with methanol is faster than reductive elimination of the enolate ligand with the H on rhodium.⁴⁵ The H₂ gas supplies the H atom added to the β carbon of 61, and both of the H atoms added to 63; the solvent supplies only the H atom found at the other α carbon of enone substrate 61! Reaction of the same substrate 61 with D₂ instead of H₂ gas resulted in 90% deuteration of 65 with only 0.25 hydrogen at the β position. Given that both the H₂ gas and the MeOH are present in large excess relative to the substrate and the catalyst, there is little scrambling between the H₂/D₂ and the solvent during these experiments.^46,47

Scheme 4.

Deuterium Labelling Experiments

Mechanisms that explain the results in Scheme 4 are shown in Scheme 5, with substrates bearing enones on the left and those with isolated olefins on the right. The generation of the active catalyst probably begins with the reductive elimination of phenylpyridine from RhH-1 (as suggested by Chirik),³⁹ followed by the addition of H₂ to the Cp*Rh(I). The resulting Cp*RhH₂ has been drawn by the Maitlis,³³ Finke,³⁴ and Chirik³⁹ groups in catalytic cycles, but to our knowledge has never been isolated or identified.

Scheme 5.

Mechanistic Proposal

To determine if the generation of carbon-centered radicals by MHAT (metal hydrogen atom transfer) was involved in this RhH-1 catalyzed hydrogenation, we treated 71 and 74 with H₂ in the presence of RhH-1. Although RhH-1 has a low bond dissociation free energy (BDFE = 52.3 kcal/mol)³² no trace of the radical cyclization products 73 or 76 was observed (the predominant products are shown in Scheme 6, i.e. 72 and 75).

Scheme 6.

Evidence against MHAT and Radical Formation

Heterogeneous and homogeneous catalysis have been established by Finke and Chirik in related Cp*Rh systems,^{33–35,39,48–51} but it appears that our catalyst is homogeneous. All materials dissolved with our reactions being clear and red within 20 min. We observed no precipitates even after 24 h of reaction time, and the kinetic plot in Scheme 7 suggests a clean first-order transformation of the tested substrate (S15, it is the precursor to compound 15) to product with a rate constant of about 2.3 × 10⁻⁵ M⁻¹ s⁻¹. A sigmoidal curve is typical of metal-particle formation. Furthermore, we found the addition of excess mercury after 200 min did not change the observed rate constant, a result typical of a homogeneous catalyst. Finally, TEM analysis (see the Supporting Information section for full details) of the residue when solvent was removed after the reaction showed neither heterogeneous metal particles nor rhodium clusters.⁵²

Scheme 7.

Mercury Poisoning Study

CONCLUSION

In summary, (η⁵-C₅Me₅)Rh(2-(2-pyridyl)phenyl)H (RhH-1) catalyzes the selective 1,4-reduction of α,β-unsaturated carbonyl compounds—even ones with bulky substituents—under H₂. It also catalyzes the hydrogenation of many isolated alkenes under mild reaction conditions. The system shows excellent selectivity and functional group compatibility, and appears to operate mechanistically under conditions that are homogeneous. The rhodium catalyst is a significant complement to the existing toolbox for metal-catalyzed selective hydrogenations and offers an opportunity to overcome longstanding challenges in the hydrogenation of polyfunctionalized molecules.