CuH-Catalyzed Enantioselective 1,2-Reductions of α,β-Unsaturated Ketones

Abstract

The first study describing a general technology for arriving at valued nonracemic allylic alcohols using asymmetric ligand-accelerated catalysis by copper hydride is described.

Asymmetric copper hydride chemistry has become an especially powerful tool for controlling chirality in a variety of substrate types.¹ Most notably, nonracemically ligated CuH can be used to direct remarkably selective hydride delivery to the β-site in a variety of Michael acceptors (Scheme 1, path A). In the absence of extended conjugation, asymmetric 1,2-additions of CuH are now known for aromatic ketones,² diaryl³ and heteroaromatic ketones,⁴ and imines.⁵ Redirecting the natural tendency for copper complexes away from additions in a 1,4-sense can be challenging. The potential to alter, in the achiral manifold, such regioselectivity toward the 1,2-mode by a “subtle interplay of steric and electronic factors” of the phosphine ligand on copper was recognized years ago by Stryker.⁶ Overcoming the inherent mechanistic preference for initial d-π*-complexation associated, e.g., with Cu(I)-olefin soft-soft interactions in α,β-unsaturated ketones, remains an unsolved problem notwithstanding the synthetic potential of the resulting nonracemic allylic alcohols (Scheme 1, path B). While isolated examples of copper-catalyzed enantioselective 1,2-reductions of enones exist,⁷ any semblance of a general asymmetric protocol resulting from the correlation of substrate substitution pattern with ligand biases and/or tuning of reaction conditions for this important transformation is still lacking. Herein, we describe new methodology for the enantioselective CuH-catalyzed 1,2-reduction of α-substituted unsaturated ketones leading to secondary allylic alcohols (Scheme 1).

Scheme 1.

Pathways for addition of CuH to unsaturated ketones

As illustrated in Table 1, optimization studies using enone 1 revealed that (1) 1,2-addition to arrive at cinnamyl alcohol 2 is strongly favored over conjugate addition; (2) ee’s on the order of 90% could be achieved; (3) ligands in both the SEGPHOS⁸ and BIPHEP⁹ series give similar levels of induction; (4) diethoxymethylsilane (DEMS) as the stoichiometric source of hydride¹⁰ gives the best ee’s; (5) Et₂O is the solvent of choice; (6) reactions should be run at −25 °C for optimal conversion and enantioselectivity; (7) the sense of induction is such that (L2)CuH¹¹ produces the S-allylic alcohol, while (L3b)CuH leads to the enantiomeric product.

Table 1.

Selected optimization conditions for regio- and stereo-controlled 1,2-reductions (see SI for full details)^a

Entry	Ligand	Solv.	T (°C)	Yield of 2 (%)b	ee of 2 (%)c
1	L1	THF	rt	90	50 (S)
2	L2	THF	rt	78	75 (S)
3	L2	THF	−25	87	86 (S)
4	L2	Et2O	−25	83 (98)d	91 (S)
5e	L2	Et2O	−35	n.d.	n.d.
6	L3a	Et2O	−25	96	89 (R)
7	L3b	Et2O	−25	95	91 (R)
8	L3c	Et2O	−25	99	90 (S)
9f	BDP	THF	rt	-	-

^aPerformed on a 0.1 mmol scale in 0.3 mL solvent.

^bBy ¹H NMR using Ph₃CH as internal standard.

^cBy chiral HPLC. Absolute stereochemistry was determined by comparing optical rotation to that of the known compound.

^dIsolated yield (0.25 mmol scale).

^eLow conversion after prolonged reaction time.

^f1,2-/1,4-ratio = 1:7, 60% isolated yield of 1,4-reduced enone.

Several additional examples of acyclic and cyclic enones can be found in Table 2. α’-Substitution with an alkyl group other than methyl in 1 leads to the desired product 3 in high ee using L3b, while α-substitution with residues including ethyl and n-pentyl (4 and 5) gives consistently high yields and ee’s of 1,2-addition products with one or both ligand systems.¹² Modified educts with either α-phenyl (6) or α-bromo (7), likewise, lead to 1,2-adducts, albeit in somewhat lower ee’s. Replacing the β-phenyl group in 1 with an alkyl moiety (as in 8) did not alter the outcome of the reaction.

Table 2.

CuH cat. asymmetric 1,2-reductions of α-substituted enones^a

^aReactions were carried out on 0.25 mmol scale in 0.5 mL Et₂O. Isolated yields after column chromatography are given in parentheses. Ee’s were determined by chiral HPLC or GC analyses. Stereochemistry shown was determined by analogy to 2 (see Table 1).

^bAbsolute stereochemistry determined by comparing optical rotations with known compounds.

^cSee text.

^dSee SI.

The impact of variation in substituents on a β-aryl ring in an educt was also investigated. Electron-donating as well as electron-withdrawing groups were tolerated and gave secondary allylic alcohols 9-14 in high yields and good ee’s. Surprisingly, a strong electron-withdrawing group (e.g. a nitro group) led to significant amounts of the corresponding 1,4-reduced product when L2 was used (see SI), whereas L3b gave the desired alcohol 13 with excellent regio- and stereocontrol.¹²

Various cyclic arrays (15-17) fit into the anticipated pattern of regio- and enantio-control using (DTBM-SEGPHOS)CuH. The mild conditions involved allow for isolation of a nonracemic cyclohexenol 17 bearing a cross-coupling partner vinyl triflate without losses due to ring fragmentation observed with harsher reducing agents.¹³ While treatment of (R)-pulegone with catalytic [(R)-L2]CuH gave the highly favored anticipated cis-product (93%; 99:1 dr), CuH complexed by ent-L2 led predominantly to the less common trans isomer 18 (88%; 4:1 dr).¹⁴

The influence exerted by an α-substituent is further highlighted by the case of exocyclic olefin-containing enone 19. Notwithstanding full accessibility of CuH to the β-site, delivery of hydride takes place in a 1,2-fashion, giving allylic alcohol 20 in 78% ee (Scheme 2).

Scheme 2.

(L3b)CuH catalyzed 1,2-addition to a β,β-unsubstituted enone

The potential for a ligated CuH complex to induce asymmetry in two distinct functional groups within the same pot is illustrated in Scheme 3. Simultaneous exposure of enone 1 and enoate 21 (1:1 ratio) to conditions first favoring enone 1,2-reduction gave 2, with <5% conjugate reduction of 1 being observed. Without isolation, addition of t-BuOH (1.1 equiv), as originally reported by Stryker,^6,15 was used to enhance the rate of catalyst regeneration. The presence of this additive, along with added silane (1.1 equiv), led to asymmetric 1,4-reduction of 21 to ester 22, both processes taking place in high isolated yields and excellent ee’s.

Scheme 3.

One reagent, two reactions: 1-pot asymmetric 1,2-reduction of an enone and 1,4-reduction of an enoate

In summary, regioselectivity in reactions of non-racemicaly ligated, in situ-generated CuH can be dramatically shifted to favor asymmetric 1,2-over normally observed 1,4-reductions of α,β-unsaturated ketones. This powerful methodology affords high yields and ee’s of resulting allylic alcohols of defined olefin geometries and central chirality.