A practical and catalytic reductive olefin coupling.

A redox-economic method for the direct coupling of olefins that uses an inexpensive iron catalyst and a silane reducing agent is reported. Thus, unactivated olefins can be joined directly to electron-deficient olefins in both intra- and intermolecular settings to generate hindered bicyclic systems, vicinal quaternary centers, and even cyclopropanes in good yield. The reaction is not sensitive to oxygen or moisture and has been performed on gram-scale. Most importantly, it allows access to many compounds that would be difficult or perhaps impossible to access using other methods.

Two-phase terpene total synthesis has been demonstrated to be a useful strategy to access steroids, diterpenes, and sesquiterpenes.[1] The approach consists of a cyclase phase, where a lowly oxidized terpene skeleton is rapidly constructed from inexpensive starting materials, and an oxidase phase, where a series of site-selective oxidations furnishes the desired natural product. Our laboratory is interested in pursuing such a strategy toward the ent-kaurane family of diterpenoids,[1d] whose members have been utilized in traditional Chinese medicine for centuries as they possess anticancer and antibacterial activities.[2] Analysis of an archetypical ent-kaurane, rosthorin A (1, Figure 1A), through the two-phase paradigm led to 2 as a proposed cyclase-phase end point. The C4–C5 motif in 2 appears in numerous terpenes and is biosynthetically derived from a polyene cyclization where C4 and C5 formally are electrophilic and nucleophilic positions, respectively.[3] Polarity reversal[4] of this disconnection led to 3 and inspired a method for directly adding an olefin, one of the most ubiquitous functionalities in organic chemistry, to an enone. Here we describe the development of a simple catalytic method for forging this type of C–C bond and demonstrate its scope in both intra- and intermolecular modes, even in highly sterically demanding contexts.

Figure 1

(A) Retrosynthesis of rosthorin A. (B) Examples of C–C bond formation that directly utilize alkynes and olefins. (C) Recent hydrofunctionalizations of unactivated olefins inspire reductive olefin coupling. (D) Optimization of reaction conditions. Ratios by GC/MS.

The disconnection leading 2 to 3 is desirable from the standpoint of redox economy.[5] Some inspirational and relevant examples are depicted in Figure 1B, where alkynes and olefins are used directly (i.e., without functional group interconversions) to form C–C bonds through advances in Ni catalysis,[6] metathesis,[7] reductive aldol reactions,[8] and hydrogenative C–C bond formation.[9] Of particular relevance to us are the powerful olefin functionalization methods of Boger,[10] Carreira,[11] Mukaiyama,[12] and others[13−15] which showed that unactivated olefins can be converted to putative radical species that can be intercepted by heteroatom acceptors (Figure 1C). Only three categories of C–C bond formation using these methods are reported to date: hydrocyanation with TsCN,[10b,11f] hydroaldoximation and hydrocyanooximation with phenyl sulfonyl oximes,[11i] and cyclization of 1,6-dienes.[10b,15] With these promising precedents, initial studies toward a reductive cyclization[16] were carried out.

When a simplified variant of proposed intermediate 3, enone 4, was subjected to Boger’s Fe2(ox)3·6H2O and NaBH4 conditions,[10] the cis-fused decalin 7 was obtained along with both unreacted starting material and a small amount of allylic alcohol 5 (Figure 1D, entry 1). Although the trans-decalin was originally desired, the cis-decalin still provided an opportunity to develop a general C–C bond-forming reaction. Thus, optimization of the formation of cis-decalin 7 was pursued.

Since NaBH4 would limit the functional group tolerance of the reaction, milder reducing agents were explored. NaBH(OAc)3 eliminated the formation of 5 but resulted in the undesired formation of 6, presumably from a competitive vinylogous Prins cyclization where Fe2(ox)3·6H2O simply acts as a Lewis acid[17] (entry 2). This byproduct was exclusively formed when (TMS)3SiH and Et3SiH were used as the reducing agent (entries 3 and 4), whereas PhSiH3 produced cis-decalin 7 as the major product along with a small amount of 6 (entry 5).

Switching the Fe(III) salt to Fe(acac)3 gave a monophasic reaction medium.[10a] Under these conditions, use of Et3SiH as the reductant led to the recovery of starting material (entry 6), whereas PhSiH3 led to full conversion to the desired 7 (entry 7). When the amount of Fe(acac)3 was decreased to 30 mol%, 7 was still formed exclusively (entry 8) with 60% isolated yield (Table 1), demonstrating that this reductive olefin coupling could be run in a catalytic fashion. Although not essential, the use of ethylene glycol as a cosolvent in entry 8 prevented the formation of PhSi(OEt)3, a byproduct of the reaction that complicated purification.

Table 1

Scope of the Intramolecular Reductive Coupling

Used 30 mol% Fe(acac)3 in EtOH/(CH2OH)2 (5:1).

Used 20 mol% Fe(acac)3 in DCE/(CH2OH)2 (5:1).

Run on gram-scale with 2.0 equiv PhSiH3.

Used 1.5 equiv PhSiH3.

Other substrates were also found to be amenable to intramolecular coupling (Table 1), with the reaction succeeding in more sterically demanding contexts, such as the formation of vicinal all-carbon quaternary centers.[18] Thus, cyclizing benzyl geranate (8), N-benzylgeranamide (10), and citral (12) to cyclopentanes 9, 11, and 13 proceeded smoothly in 76, 93, and 79% yield, respectively. As the formation of 9 demonstrates, decreasing the amount of Fe(acac)3 to 20 mol% and using DCE as cosolvent instead of EtOH still resulted in cyclization. The cyclizations were found to be insensitive to oxygen and moisture and were generally complete in <1 h.

Highly strained ring systems could also be formed, as evidenced by the cyclization of the skipped diene α-ionone (14) to generate cyclopropane 15 in 97% yield. This cyclization could also be conducted easily on gram-scale, albeit with a small reduction in yield (81%). Although cyclopropanes have been previously formed by irradiating skipped dienes in specialized cases,[19] to our knowledge, this represents the first general method for such a transformation. To our surprise, even (+)-nootkatone (16) was cyclized in 64% yield to give highly congested bicyclo[2.2.1]heptane 17, in which one of the newly formed adjacent quaternary centers is neopentyl. Such ring systems would be difficult to construct otherwise.

The applicability of this method to intermolecular settings was next examined by evaluating the reaction of a series of “donor” olefins with 3 equiv of methyl vinyl ketone (MVK) as an “acceptor” (Table 2). Upon exposure to the standard conditions, 1-methylcyclohexene (18) proved to be a competent donor, with the corresponding adduct 19 isolated in 66% yield. Although homocoupling of the participating olefins was anticipated, such products were not observed. Furthermore, the newly forged bond was formed at the more substituted side of the donor olefin, in a fashion complementary to the hydroboration/radical conjugate addition developed by Renaud[20] and the hydrozirconation/conjugate addition sequence developed by Wipf.[21] Both of those methods result in the opposite regioselectivity, where bond formation occurs at the least substituted side of the donor olefin.

Table 2

Donor Scope of the Intermolecular Reductive Coupling

Used 2.5 equiv PhSiH3.

Used 100 mol% Fe(acac)3.

Tolerated functional groups include TBS ethers and tert-butyl carbamates, as demonstrated by the formation of 21 and 23 in 73 and 62% yield, respectively. The use of styrene (24) as a donor led to ketone 25 as the major product in 87% yield. Additionally, the methodology tolerated heteroaromatic functionalities, although in modest yields, with 3-vinylpyridine (26) and N-Boc-3-prenylindole (28) giving 27 and 29 in 42 and 34% yield, respectively.

A competitive process that was occasionally observed was the reduction of the donor olefin to its saturated counterpart. Thus, coupling 1-dodecene (30) to MVK resulted in the isolation of dodecane (not shown) in 30% yield along with the desired ketone 31 in 40% yield. The sterically encumbered estrone 3-methyl ether derivative 33, which bears two vicinal all-carbon quaternary centers, was isolated in 54% yield as a single diastereomer, with the reduced derivative of 32 (not shown) also isolated in 17% yield. However, MVK could be added to (+)-sclareolide derivative 34 from its least hindered face to produce 35 in 56% yield without reduction of the starting material.

The scope of the acceptor olefins was next probed, using 1-methylcyclohexene (18) or TBS ether 20 as the donor (Table 3). In addition to MVK, methyl acrylate, N,N-dimethylacrylamide, and acrylonitrile all served as efficient acceptors, providing 36–41 in yields of 64–81%. Notably, TBS ether 20 can be coupled with acrylonitrile on gram-scale to provide nitrile 41 with only a minor reduction in yield. Phenyl vinyl sulfone also participated in the reaction, although it required a catalyst loading of 40 mol% and gave moderate yields of 42. The reaction proceeded more smoothly using doubly activated acceptors, with dimethyl fumarate providing diester 43 in 90% yield.

Table 3

Acceptor Scope of the Intermolecular Reductive Coupling

Used 1.0 equiv donor and 3.0 equiv acceptor.

Used 40 mol% Fe(acac)3.

Run on gram-scale.

Used DCE/(CH2OH)2 (1:1).

Used 100 mol% Fe(acac)3.

Used 1.0 equiv donor and 1.1 equiv acceptor.

Cyclic acceptors generally gave slightly lower yields, with 2-furanone providing adduct 44 in 61% yield. Cyclic enones could also be used, with 2-cyclopentenone furnishing 45 and 46 in 65% and 60% yield, respectively. Substitution at the α-position of the acceptor olefin was also tolerated, although the α-methylated cyclopentanone 47 was obtained in a diminished 38% yield as a 2:1 mixture of diastereomers. Interestingly, the use of 2-cyclohexenone as an acceptor led to a significantly lower yield of 48 as compared to 45, even when a stoichiometric amount of Fe(acac)3 was used. Finally, a heteroaromatic acceptor, acridine, could be employed to generate 49 as a proof of principle for applying this chemistry in reductive Minisci-type functionalization.[22]

The mechanism of this reductive olefin coupling was probed through a variety of experiments (Figure 2). Fe(acac)3 was found to be necessary for the coupling, as its omission resulted in only recovery of starting material (Figure 2A). Interestingly, running the reaction under oxygen-free conditions in a Schlenk tube fully cyclized α-ionone (14) to cyclopropane 15, indicating that oxygen was not responsible for reoxidizing the iron catalyst. Furthermore, subjecting 14 to the reaction conditions using either CD3OD or CH3OD as the solvent led to the isolation of cyclopropane 50, bearing a sole deuterium atom adjacent to the ketone (Figure 2B). The lack of deuterium incorporation on the cyclohexane ring in both instances suggests that the H that adds across the unactivated olefin originates from PhSiH3. Additionally, exposing non-deuterated 15 to identical reaction conditions did not lead to any deuterium incorporation, showing that deuterium incorporation takes place during the course of the reaction and is not an artifact of hydrogen-deuterium exchange with the solvent.

Figure 2

(A) Control experiments. (B) Deuterium labeling studies. (C) Proposed mechanism of the reductive olefin coupling.

A mechanistic scenario consistent with these results is outlined in Figure 2C. Donor olefin 53 would abstract a hydrogen radical from Fe hydride[23]52—derived from Fe(III) species 51 and PhSiH3[24]—to generate reduced Fe species 55 and tertiary radical 54. Alternatively, 54 could be generated by hydrometalation of 53, followed by homolysis of the Fe–C bond (not shown).[11e] Conjugate addition of 54 into Michael acceptor 56,[25] followed by single-electron transfer with 55, would provide intermediate 58 and regenerate 51, which would re-enter the catalytic cycle. Protonation of 58 would give the coupled product 59.

In conclusion, a practical method for the reductive coupling of olefins that utilizes a readily available and inexpensive Fe source as a catalyst has been developed. This reaction is operationally simple and has been run on gram-scale. Furthermore, it builds molecular complexity rapidly, with most reactions reaching completion in <1 h. The coupling works in both intra- and intermolecular settings and can generate highly hindered bicyclic systems, cyclopropanes, and vicinal quaternary centers.[26] The ability of this transformation to directly employ olefins in C–C bond-forming events bodes well for future applications in a variety of contexts, including the cyclase phase of two-phase terpene synthesis.