Regioselective hydroacylation of 1,3-dienes by cobalt catalysis.

We describe a cobalt-catalyzed hydroacylation of 1,3-dienes with non-chelating aldehydes. Aromatic aldehydes provide 1,4-addition products as the major isomer, while aliphatic aldehydes favor 1,2-hydroacylation products. The kinetic profile supports an oxidative cyclization mechanism involving a cobaltacycle intermediate that undergoes transformation with high regio- and stereoselectivity.

A modern challenge in organic synthesis is the invention of methods that use catalysts derived from first-row transition metals.[1] A number of valuable olefin transformations, including hydrogenation,[2] hydroformylation,[3] and hydrovinylation,[4] have been achieved by cobalt catalysis. These breakthroughs highlight Co as an attractive and complementary alternative to Rh due to its relatively low cost and high abundance. Encouraged by this progress, we recently turned our attention to developing olefin hydroacylation[5−14] by cobalt catalysis.[15] Brookhart demonstrated the first and only previously known intermolecular Co-catalyzed hydroacylation (eq 1).[15a,15b]While promising, this strategy was limited to vinylsilanes, and aldehyde decarbonylation remained competitive. We imagined developing a Co(I)-catalyzed cross-coupling of various aldehydes and 1,3-dienes (eq 2). Our C–H bond functionalization would yield β,γ- and/or γ,δ-unsaturated ketones with high regioselectivity and excellent atom economy. Moreover, this proposed hydroacylation would afford regiocontrol distinct from current technologies, including the relevant Ru-catalyzed pathway developed independently by Krische[11c] and Ryu[11d] (eq 3).

Our proposal draws from a mechanism established by Hilt for the Co-catalyzed hydrovinylation of olefins to generate 1,4-dienes.[4] As shown in Scheme 1, a Co(II) precatalyst can be used to generate the Co(I) catalyst.[16] We hypothesize that Co(I) catalyst A would promote oxidative cyclization between aldehyde 1 and a 1,3-diene 2.[17] The resulting cobaltacycle B exists in two η1 forms: seven-membered cobaltacycle C and five-membered cobaltacycle C′.[4,18,19] Depending on the substrate and ligand, cobaltacycle C or C′ would undergo transformation by Path I or II, respectively. A β-hydride elimination generates intermediate D or D′, which can undergo reductive elimination to yield product 3 or 3′ and regenerate catalyst A. In Path I, formation of the cobaltacycle intermediate requires a cis-olefin geometry; thus, we propose this pathway would lead to the Z-isomer of 3 with high levels of stereocontrol.

Scheme 1

Proposed Cobalt-Catalyzed Hydroacylation of Dienes by Oxidative Cyclization

To test our hypothesis, we chose benzaldehyde 1a and isoprene 2a as model substrates. Control experiments confirm that these substrates are unreactive in the absence of either Co(II) salts or reducing agents.[20] We then examined the transformation using Co(II) (a catalyst precursor for hydrovinylation)[4] with various ligands and additives to generate the requisite Co(I) catalyst. A survey of 20 commercially available phosphines reveals 1,3-bis(diphenylphosphino)propane (dppp) is the most promising ligand. With Co(II)-dppp, the desired product is observed in 11% yield by using Zn/ZnI2, a known protocol for generating Co(I) (Table 1, entry 1).[4,16] By applying In as the reductant instead, we observe improved reactivity (21%, entry 2). Through a further survey of additives, we find that a combination of In/InBr3 gives the best result (57% yield, entries 3–5). Given the role of ZnI2 in hydrovinylation,[4] we assume InBr3 similarly promotes the formation of cationic Co(I) species. In the absence of these Lewis acids, the yield is diminished (13% yield, entry 6). In all cases, we observe the β,γ-unsaturated ketone in preference to the γ,δ-unsaturated ketone (up to 19:1 selectivity). In accordance with our proposed mechanism, the stereochemistry of the resulting trisubstituted olefin is Z, as confirmed by 2D NOESY analysis.

Table 1

Additives Effects for Hydroacylationa

entry	additives	yield (%)b	3aa/3aa′c	Z/E (3aa)c
1	Zn/ZnI2	11	12:1	10:1
2	In/ZnI2	21	14:1	16:1
3	In/InI3	45	17:1	13:1
4	In/InBr3	57	19:1	11:1
5	In/InCl3	28	14:1	17:1
6	In/—	13	13:1	18:1

1a (0.20 mmol), 2a (0.60 mmol), Co(dppp)I2 (5 mol%), In or Zn (20 mol%), MX (5 mol%), DCE (1 mL), 60 °C, 24 h.

Overall yield of 3aa and 3aa′, determined by 1H NMR or GC-FID with dimethyl terephthalate (0.05 mmol) as internal standard.

Determined by 1H NMR or GC-FID.

To tune the catalyst, we prepared six analogues of dppp by varying the substitution pattern on the aryl groups (L1–L6, Table 2). From this study, we find that L6 (where Ar = 3,4-(MeO)2C6H3) gives high yield, regioselectivity, and stereocontrol (87% isolated yield, >20:1 regioselectivity, and 19:1 Z/E, entry 8). With L6, the amount of In powder can be reduced to 10 mol% (entry 8). A trace amount of γ,δ-unsaturated ketone 4aa is observed as a minor product (3aa/4aa >20:1, entry 8). Our cobalt catalyst favors the 1,4-hydroacylation product 3aa over the 1,2-addition isomer 5aa which has been accessed with Ru(II) catalysis.[11c,11d] Thus, cobalt enables a rare type of hydroacylation that occurs across a conjugated π-system rather than a single π-bond.

Table 2

Ligand Effects of DPPP Analoguesa

entry	catalyst	yield (%)b	3aa/3aa′c	Z/E (3aa)c
1	Co(dppp)I2	57	19:1	11:1
2	Co(L1)I2	59	17:1	17:1
3	Co(L2)I2	63	19:1	13:1
4	Co(L3)I2	50	14:1	15:1
5	Co(L4)I2	83	>20:1	6:1
6	Co(L5)I2	95	>20:1	3:1
7	Co(L6)I2	83	>20:1	20:1
8d	Co(L6)I2	91 (87e)	>20:1	19:1

1a (0.20 mmol), 2a (0.60 mmol), Co(II) (5 mol%), In (20 mol%), InBr3 (5 mol%), DCE (1 mL), 60 °C, 24 h.

Overall yield of 3, determined by 1H NMR or GC-FID with dimethyl terephthalate (0.05 mmol) as internal standard.

Determined by 1H NMR or GC-FID.

In (10 mol%), DCE (0.5 mL).

Isolated yield.

With this protocol in hand, we explored the hydroacylation of diene 2a using 18 different aldehydes (Table 3). In general, good to high yields are obtained with various aromatic aldehydes 1 (60–97% yields, entries 1–13). In the case of electron-rich aryl aldehyde 1h, we observe a drop in stereoselectivity (2:1) due to competitive olefin isomerization. The use of a mixed solvent (1:1 DCE/toluene) presumably inhibits isomerization and allows isolation of the desired product in good yield and higher stereoselectivity (11:1, entry 8 vs 9). This catalyst promotes hydroacylation of heteroaromatic aldehyde 1k (94% yield, entry 12). Moreover, the hydroacylation of isoprene 2a with α,β-unsaturated aldehyde 1m provides a conjugated ketone in 86% yield and >20:1 regioselectivity (entry 14).

Table 3

Variation in the Aldehyde Scopea

entry	R in 1	yield (%)b	3/3′c	Z/E (3)c
1	Ph (1a)	87	>20:1	19:1
2	4-BrC6H4 (1b)	92	>20:1	>20:1
3	4-ClC6H4 (1c)	96	>20:1	>20:1
4	4-FC6H4 (1d)	60	>20:1	>20:1
5	4-CF3C6H4 (1e)	78	>20:1	>20:1
6	4-MeO2CC6H4 (1f)	88	>20:1	>20:1
7	4-MeC6H4 (1g)	90	>20:1	6:1
8	4-MeOC6H4 (1h)	97	>20:1	2:1
9d	4-MeOC6H4 (1h)	83	>20:1	11:1
10	3-BrC6H4 (1i)	64	>20:1	>20:1
11	3-ClC6H4 (1j)	89	>20:1	>20:1
12	2-furyl (1k)	94	>20:1	>20:1
13	2-naphthyl (1l)	94	>20:1	6:1
14	(E)-styryl (1m)	86	>20:1	6:1
15	Cy (1n)	17	2:1	n/a
16e	Cy (1n)	82	1:8	n/a
17e	n-C5H11 (1o)	78	1:5	n/a
18e	n-C6H13 (1p)	83	1:4	n/a
19e	n-C7H15 (1q)	91	1:5	n/a
20e	BnCH2 (1r)	74	1:4	n/a

1 (0.20 mmol), 2a (0.60 mmol), Co(L6)I2 (5 mol%), In (10 mol%), InBr3 (5 mol%), DCE (0.5 mL), 60 °C, 20–24 h.

Isolated yield of all isomers.

Determined by GC-FID or 1H NMR; trace amount of product 4 was observed as minor isomer (3/4 >20:1) for aldehydes 1a–m.

DCE/toluene (1:1, 0.5 mL) was used solvent.

1 (0.20 mmol), 2a (0.60 mmol), Co(dcpe)I2 (5 mol%), In (20 mol%), InBr3 (5 mol%), DCE/EtOAc (3:1, 0.5 mL), 50 °C, 20–24 h.

With catalyst Co(L6)I2, aliphatic aldehyde 1n is less reactive and undergoes hydroacylation with lower regioselectivity (Table 3, entry 15). Toward addressing this challenge, we investigated a range of parameters and found promising reactivity with a more electron-rich phosphine ligand, 1,2-bis(dicyclohexylphosphino)ethane (dcpe).[20] The resulting ketone products are obtained in good yields with a dramatic switch in regiocontrol (entry 15 vs 16). Generally, the 1,2-hydroacylation products 3′ are afforded as the major isomer with regioselectivities ranging from 4:1 up to 8:1 for aliphatic aldehydes 1n–r (entries 16–20). In these cases, we observe high C1-regioselectivity (C–C bond formation at 1-position of diene 1a) instead of C3-regioselectivity (eq 2 vs 3).

Next, we examined the scope of dienes (Table 4). For benzaldehyde, high yields (77–97%) and excellent regio- and stereoselectivities (>20:1 for both) are achieved for hydroacylation of 2-aryl-substituted butadienes, despite varying electronic and steric properties of substituents (entries 1–9). The catalyst loading can be reduced to 2 mol% for the coupling of benzaldehyde 1a and 2-phenylbutadiene (entry 2). The use of 2-cyclohexylbutadiene gives β,γ-unsaturated ketone 3aj with moderate stereoselectivity (6:1) and good yield (87%, entry 10). Hydroacylation of a 1,2-disubstituted diene occurs with moderate yield (entry 11),[21] while 2-phenylbutadiene 2b couples well with a variety of aldehydes (entries 12–17). Throughout these studies, we observe no aldehyde decarbonylation, the byproduct expected for hydroacylations involving C–H bond activation of non-chelating aldehydes.

Table 4

Variation of the Diene Partnera

1 (0.30 mmol), 2 (0.20 mmol), Co(L6)I2 (5 mol%), In (10 mol%), InBr3 (5 mol%), DCE (0.5 mL), 60 °C, 16 h. Only one isomer was observed, except for 3aj (E/Z 6:1). The geometry of 3ab and 3ak was assigned by NOESY spectra.

Co(L6)I2 (2 mol%).

Finally, we report mechanistic studies that further support our proposed oxidative cyclization mechanism, in preference to a C–H activation pathway.[15a,15b] First, a kinetic isotope effect (KIE, kH/kD) of 2.7 is observed from competition experiments between benzaldehyde 1a and d-1a (eq 8).[22] If hydroacylation occurs through a traditional C–H activation mechanism, either oxidative addition of Co(I) to the aldehyde or insertion of the diene into the Co–H will be the turnover-limiting step.[5d] However, we observe a zero-order dependence on both aldehyde and diene concentrations,[20,23] thus disfavoring a C–H activation pathway. The KIE is consistent with β-hydride or reductive elimination as the turnover-limiting step in Path I (Scheme 1).[22] Moreover, the zero-order dependence on both the aldehyde and the diene concentrations supports the possibility of metallacycle C or D as catalyst resting state (Scheme 1).

When the hydroacylation is performed with deuterio-benzaldehyde d-1a, the deuterium atom is incorporated at the 4-position of diene 2b without any detectable deuterium at other positions (eq 9). This result suggests that the hydroacylation with aromatic aldehydes proceeds through a 1,4-addition pathway (Path I, Scheme 1). For hydroacylation with aliphatic aldehyde d-1r, the deuterium atom is incorporated completely at the β- and δ-position of products d-3rb′ and d-3rb, respectively (eq 10). While further studies are warranted, our observations support a mechanistic proposal where Path I or II is favored (Scheme 1), depending on the properties of both the ligand and the substrate.

In contrast to the traditional mechanism of hydroacylation catalyzed by Rh(I) or Co(I), we propose an oxidative cyclization mechanism that avoids decarbonylation. Our catalyst promotes C1-regioselective hydroacylation of dienes and can be tuned to favor either 1,4- or 1,2-hydroacylation. Through the 1,4-hydroacylation pathway, we achieve the stereoselective synthesis of trisubstituted olefins, which are key building blocks that are challenging to access.[24] Ongoing efforts will focus on studying other catalysts for achieving higher selectivities and greater scope. Our study contributes to the emerging strategies available for hydroacylation via non-precious-metal catalysis.[13,15]