Nickel-Catalyzed Coupling of Alkenes, Aldehydes, and Silyl Triflates

Abstract

A full account of two recently developed nickel-catalyzed coupling reactions of alkenes, aldehydes and silyl triflates is presented. These reactions provide either allylic alcohol or homoallylic alcohol derivatives selectively, depending on the ligand employed. These processes are believed to be mechanistically distinct from Lewis acid-catalyzed carbonyl-ene reactions, and several lines of evidence supporting this hypothesis are discussed.

Introduction

Alkenes are one of the most versatile, utilized, and readily available classes of functional groups. Simple alpha olefins are produced in megaton scale each year industrially, highlighting the importance of these organic feedstocks.¹ Several indispensable transformations utilize olefins, such as Ziegler-Natta polymerization,² the Heck reaction,^3a–3e Wacker oxidation,^3a hydroformylation,^3a hydrometallation,^3a alkene cross-metathesis,⁴ epoxidation⁵ and dihydroxylation.⁵

The nickel-catalyzed coupling of alkenes, aldehydes, and silyl triflates that we recently developed is the first example of a transition metal-catalyzed coupling of simple, unactivated olefins and aldehydes that provides allylic alcohol derivatives.⁶ With careful choice of the supporting ligand on nickel, this coupling reaction can also selectively provide homoallylic alcohol derivatives that are generally not accessible using Lewis acid-catalyzed carbonyl-ene reactions (eq 1).⁷

(1)

Transition metal-catalyzed intermolecular reductive and alkylative coupling reactions have emerged as useful methods for the preparation of alcohol and amine derivatives. Nickel, palladium, rhodium and ruthenium catalysts have been found to be particularly effective in the intermolecular coupling of alkynes, 1,3-enynes, 1,3-dienes, allenes, enoate esters, enones, and enals with aldehydes, ketones, epoxides, glyoxylate esters, and imines.^8–11 A variety of reducing agents have been used in these reductive couplings, such as triethylborane, organozinc reagents, organosilanes, and molecular hydrogen. As yet, however, simple, unactivated alkenes such as ethylene and 1-octene have not been reported to undergo analogous catalytic reductive coupling reactions.

As a part of our program directed toward developing C–C bond forming reactions of “off-the-shelf”, simple starting materials, we became very interested in catalytic alkene–aldehyde coupling processes. Intramolecular versions of this transformation have been reported, such as transition metal-catalyzed cyclizations of enals and enones. For example, the titanium-catalyzed intramolecular reductive cyclization of enals and enones was first reported by Buchwald and Crowe.^12a,12b Recently Ogoshi has demonstrated a nickel-catalyzed cyclization of enones.^12c α,ω-Enals also undergo cyclization by way of a radical process¹³ and also in a Lewis acid-catalyzed carbonyl-ene reaction.⁷

Intermolecular coupling of unactivated alkenes and aldehydes is commonly mediated by a transition metal (stoichiometric) or accomplished by way of a carbonyl-ene reaction.^7,15 An interesting process developed by Worpel combines an alkene and an aldehyde through a silver-catalyzed silylene transfer reaction.¹⁶

Formation of an oxametallacycle through the coupling of simple alkenes and ketones has been observed with several transition metals such as titanium, zirconium and rhodium.¹⁴ These studies suggested that transition metal-catalyzed, intermolecular coupling of alkenes and aldehydes would be feasible under the appropriate conditions. Ogoshi recently observed that Lewis acids such as a silyl triflate and trimethylaluminum facilitated the formation of an oxanickellacycle through cyclization of α,ω-enals and α,ω-enones.¹⁷ We proposed that if the intermolecular coupling of an alkene and an aldehyde occurred, the nickel alkyl bond could undergo a β-hydride elimination, followed by the removal of triflic acid from nickel to regenerate the nickel catalyst. This mechanistic framework also resembles that in the Heck reaction, a very important example of a catalytic coupling of an alkene and an electrophile. ^3a–3e

The carbonyl-ene reaction has historically been the most direct method to combine simple alkenes and carbonyl compounds to provide homoallylic alcohol products. Recent efforts in this area have focused on asymmetric induction through the use of Lewis acids and chiral ligands, such as (bisoxazoline)CuX₂, (pybox)ScX₃, and (BINAP)TiX₂ complexes.^18a–18f Typically the alkenes that are employed in intermolecular carbonyl-ene reactions are 1,1-disubstituted and trisubstituted olefins. With respect to the carbonyl component, electron-deficient enophiles such as glyoxylates, glyoxamides, and chloral are generally more efficient than simple aromatic and aliphatic aldehydes. In fact, since the report of the carbonyl-ene reaction in 1943^7a there have been only a few scattered examples of intermolecular ene reactions between monosubstituted alkenes and simple aromatic and aliphatic aldehydes.¹⁹

To summarize, Lewis acid-catalyzed carbonyl-ene reactions are typically not feasible for the most readily available alkene and aldehyde building blocks. One of the nickel-catalyzed coupling of alkenes and aldehydes described herein is thus complementary in scope to the carbonyl-ene reaction; monosubstituted alkenes couple with simple aldehydes to provide a carbonyl-ene-type product in high yield.

Results

Our investigations commenced with the simplest olefin, ethylene, and benzaldehyde. After a brief examination of phosphorous-based additives, we found that a combination of Ni(cod)₂, tris-(o-methoxyphenyl)-phosphine ((o-anisyl)₃P), triethylamine, and triethylsilyl triflate (Et₃SiOTf) promoted the coupling of ethylene with a variety of aldehydes. In all cases a triethylsilylether of an allylic alcohol is obtained in good to excellent yield (Table 1), providing ready access to a class of allylic alcohol derivatives that have been used in cross metathesis reactions, for example.⁴

Table 1.

Nickel-Catalyzed Coupling of Ethylene, Aldehydes, and Silyl Triflates

entry	R (aldehyde)	R3SiOTf	product	isolated yield (%)
1	Ph	Et3SiOTf	1a	82 (65) b
2	p-tolyl	Et3SiOTf	1b	88 (65) b
3	o-tolyl	Et3SiOTf	1c	93 (64) b
4	p-anisyl	Et3SiOTf	1d	95 (65) c
5	2-naphthyl	Et3SiOTf	1e	95 (83) b
6	2-naphthyl	Me3SiOTf	1f	60
7	2-naphthyl	t-BuMe2SiOTf	1g	67
8		Et3SiOTf	1h	80
9	2-furyl	Et3SiOTf	1i	38
10 f		Et3SiOTf	1j	25
11 f		Et3SiOTf	1k	34
12	piv	Et3SiOTf	1l	70
13		Et3SiOTf	1m	81 (40) c, d
14	cyclohexyl	Et3SiOTf	1n	25 d (34) d, e

^aStandard procedure: Ni(cod)₂ (20 mol%) and (o-anisyl)₃P (40 mol%) were dissolved in 2.5 mL toluene under argon. Ethylene (balloon, 1atm) was substituted for argon. Triethylamine (600 mol%), the aldehyde (100 mol%, 0.5 mmol), and Et₃SiOTf (175 mol%) were added. The reaction mixture was stirred 6–18 h at 23 °C.

^b(o-anisyl)₃P was replaced by Cy₂PhP.

^c(o-anisyl)₃P was replaced by Ph₃P.

^dYields determined by ¹H NMR using DMF as a standard.

^eConducted under 2 atm of ethylene.

^fStirred at room temperature for 30 h.

Under 1 atm of ethylene, simple aromatic aldehydes such as benzaldehyde and p-tolylaldehyde undergo efficient coupling (entries 1–2). Ortho substitution on the aromatic aldehyde does not appear to deter the coupling process (entry 3), and notably, acid-sensitive heteroaromatic aldehydes such as 1-methyl-2-indolecarboxaldehyde (entry 8) and 2-furaldehyde (entry 9) are tolerated, even in the presence of Lewis acidic silyl triflates. As an additional advantage, other common silyl triflates can be used in the coupling reaction, providing orthogonal protection of the hydroxyl group when necessary (entries 5–7).

Remarkably, sterically demanding tertiary aliphatic aldehydes such as pivaldehyde and 2,2-dimethyl-3-oxo-propionic acid methyl ester couple with ethylene with the same efficacy as benzaldehyde (entries 12 and 13). Enolizable aldehydes are not appropriate substrates in this system, however, since they react rapidly with the silyl triflate and triethylamine to form alkenyl silyl ethers. The coupling of ethylene with cyclohexanecarboxaldehyde is fast enough, however, that a significant amount of coupling product is observed and can be isolated (entry 14).

Tris-(o-methoxyphenyl)-phosphine is the ligand of choice for the ethylene–aldehyde coupling. Other phosphines such as dicyclohexylphenylphosphine and triphenylphosphine provide lower yield under the same reaction conditions (entries 1–5, 13).

An interesting electronic effect is observed in these coupling reactions. Electron-rich aromatic aldehydes are more efficient substrates than electron-poor aromatic aldehydes. Among the four para-substituted aromatic aldehydes examined, electron-donating para-substituents (–Me and –OMe) improve the yield of the coupling reaction (entries 2 and 4). Electron-withdrawing para-substituents (–CF₃ and –CO₂Me) suffer from incomplete conversion, even after prolonged reaction time (entries 10–11). In such cases, products resulting from a pinnacol coupling are observed but are not observed in any other example.

The encouraging results in these ethylene–aldehyde coupling reactions prompted us to examine the scope of the alkenes in detail. Unlike ethylene, 1-octene can afford more than one possible coupling product depending on where the new carbon–carbon bond is formed. The examination of a series of ligands revealed several interesting observations regarding ligand-dependent regioselectivity.

Ligand Effect

Under similar reaction conditions as the ethylene–aldehyde cases, 1-octene and benzaldehyde undergo coupling in the presence of Ni(cod)₂, a ligand, triethylamine, and Et₃SiOTf. Two distinct types of coupling products are typically observed, namely a 1,2-disubstituted allylic alcohol product (A) and a homoallylic alcohol product (H). Different classes of phosphine ligands favor one or the other coupling products, as summarized in Tables 2 and 3 and Figure 1.

Table 2.

Ligand-Dependent Regioselectivity — Electron-Rich Phosphines ^a

entry	ligand	cone angle b	νCOb	yield (2b′) c	yield (2b) c	ratio (2b′:2b) d	combined yield (2b′+2b)
1	(n-Bu)3P	132	1915	11%	27%	29:71	38%
2	(n-Oct)3P	107		7%	14%	23:77	21%
3	(i-Pr)3P	160	1915	11%	2%	85:15	13%
4	Cyp3P			10%	3%	78:22	13%
5	Cy3P	170	1915	13%	3%	81:19	16%
6	(t-Bu)3P	182		7%	2%	78:22	9%
7	Cy2PhP	162	1917	37%	16%	70:30	53%
8 e	Cy2(o-tol)P	181		30%	5%	86:14	35%
9	Cy2(o-Ph-Ph)P			32%	22%	60:40	54%
10 f	Cy2FcP			14%	2%	88:12	16%

^aStandard procedure: Ni(cod)₂ (20 mol%) and a ligand (40 mol%) were dissolved in 1.5 mL toluene. Alkene (500 mol%), triethylamine (600 mol%), the aldehyde (100 mol%, 0.25 mmol) and Et₃SiOTf (175 mol%) were added. The reaction mixture was stirred 18 h at 23 °C.

^bSee ref 20.

^cYields determined by ¹H NMR using DMF as a standard.

^dRatio was determined by ¹H NMR of the crude reaction mixture.

^e48 h reaction time.

^f1250 mol% alkene was used.

Table 3.

Ligand Dependent Regioselectivity — Electron-Poor Phosphines ^a

entry	ligand	cone angle b	νCOb	combined yield c (2b+2b′)	ratio (2b:2b′) d	E:Z (2b) d
1 e	Cy2PhP	162	1917	73	29:71	n.d.
2	CyPh2P	153	1917	84	75:25	91:9
3	(o-anisyl)3P	194	1919	70	83:17	80:20
4	FcPh2P	173		70	88:12	81:19
5	(p-tol)3P	145	1920	78	92:8	67:33
6	Ph3P	145	1922	73	92:8	67:33
7	(p-F-Ph)3P	145	1924	74	92:8	57:43
8	(EtO)Ph2P	133	1926	81	95:5	75:25
9	(p-CF3-Ph)3P	145	1929	44	>95:5	69:31
10	(EtO)2PhP	121	1932	20	>95:5	89:11
11	(PhO)3P	128	1951	<5	n.d	n.d.

^aStandard procedure: Ni(cod)₂ (20 mol%) and a ligand (40 mol%) were dissolved in 2.5 mL toluene. Alkene (1 mL), triethylamine (600 mol%), the aldehyde (100 mol%, 0.5 mmol) and Et₃SiOTf (175 mol%) were added. The reaction mixture was stirred 18 h at 23 °C.

^bSee ref 20.

^cYields were determined by ¹H NMR using DMF as a standard.

^dRatio was determined by ¹H NMR after the products were treated with TBAF.

^e48 h reaction time.

Figure 1.

Plot of H:A ratio against the σ-electron donating ability of various phosphines.²⁰

The ratio of the allylic to the homoallylic products is opposite for trialkylphosphines in which all alkyl groups are linear (entries 1 and 2), relative to those in which the three alkyl groups are branched (entries 3–5) or tertiary (entry 6). Among six trialkylphosphines with very similar electron donating abilities,^20a tri-n-butylphosphine, the smallest of the trialkylphosphines examined, favors the homoallylic alcohol product while tricyclohexylphosphine and tri-t-butylphosphine, the largest among these, favor the 1,2-disubstituted allylic product. However, these sterically demanding ligands are not nearly as effective, affording the coupling products in low yield.

Notably, replacing one of the alkyl substituent of the tricyclohexylphosphine with a phenyl ring dramatically improves the yield (53% vs 16%; Table 2, entries 5 and 7) with a slightly diminished A:H ratio. Other aryldicyclohexylphosphines also display a similar yield enhancement (entries 8 and 9, as compared to entries 4–6). The bulky and electron-rich dicyclohexylferrocenylphosphine, however, seems to be more closely related to tri-t-butylphosphine, (poor yield for both the allylic and homoallylic alcohol products, entry 10).²¹ All of the sterically demanding dicyclohexylarylderivatives examined favor the allylic alcohol product, and dicyclohexylphenylphosphine is the optimal ligand in terms of yield and selectivity.

The pronounced ligand effects prompted us to examine other organophosphorus ligands (Table 3). Among the four triarylphosphine ligands with a similar cone angle but different para-substituents (entries 5–7 and 9),^20b,20c tris-(p-trifluoromethyl-phenyl)-phosphine, the least electron-rich ligand of the four, has the highest H:A ratio (entry 9) whereas tri-p-tolylphosphine, the most σ-electron donating among these four ligands, has the lowest H:A ratio (entry 5).

These data suggest that a higher H:A ratio can be achieved by decreasing the electron donating ability of the phosphine ligand. In accord with this hypothesis, ethyldiphenylphosphinite ((EtO)Ph₂P) further improves the H:A ratio in the case of 1-octene and benzaldehyde (entry 8, 95:5). The hypothesis becomes even more convincing when the H:A ratio is plotted against the σ-electron donating ability of various phosphines in Table 3 (Figure 1).²⁰ Very electron deficient phosphonites and phosphites are not effective ligands, however (entries 10 and 11). It should be noted that the cone angle of the ligands also affects the observed H:A ratio. Many of the less electron-rich ligands that favor the homoallylic product in Table 3 are also among the smaller ligands.

Based on the results of this study, we surmised that the coupling product ratio is determined by a combined effect of the electron donating ability and the cone angle of the phosphine ligands. High H:A ratios can be achieved by using less electron-rich phosphines with small cone angle such as (EtO)Ph₂P while high A:H ratios can be obtained by using electron-rich phosphines with a large cone angle such as Cy₂PhP.²²

Effects of the Base

Tertiary amines are the optimal bases for the nickel-catalyzed coupling of alkenes and aldehydes. Among different types of amine bases examined in ethylene couplings, only tertiary amines provide >20% yield of coupling products (Table 4, entries 1 and 3). Amines that likely are able to interact with nickel to a greater degree, such as pyridine (Table 4, entry 5 and Table 6, entry 6), are not effective. No coupling products are detected when inorganic bases are used in place of triethylamine (Table 4, entries 6–8).

Table 4.

Effect of Bases in the Ethylene–Benzaldehyde Coupling ^a

entry	base	yield b
1	Et3N	77
2	Et2NH	3
3	N-methylpyrrolidine	36
4	proton sponge	10
5	pyridine	12
6 c	K3PO4	<5
7	K3CO3	<5
8	Cs2CO3	<5

^aStandard procedure: Ni(cod)₂ (20 mol%) and (o-anisyl)₃P (40 mol%) were dissolved in 2.5 mL toluene under argon. Ethylene (balloon, 1atm) was substituted for argon. A base (600 mol%), benzaldehyde (100 mol%, 0.5 mmol), and Et₃SiOTf (175 mol%) were added. The reaction mixture was stirred 18 h at 23 °C.

^bYields were determined by ¹H NMR using DMF as a standard.

^cBenzaldehyde was replaced by 2-naphthaldehyde and the reaction was run at 0.25 mmol scale.

Table 6.

Effect of Bases in the 1-Octene–Benzaldehyde Coupling (Ph₃P)

entry	base	combined yield (2b+2b′) b	ratio (2b:2b′) b	ratio (E/Z) c (2b)
1	Et3N	64%	92:8	87:13
2	Cy2NMe	35%	>95:5	71:29
3	N-methylmorpholine	25%	94:6	69:31
4	N-methylpiperidine	<5%	n.d.	n.d.
5	N-methylpyrrolidine	<5%	n.d.	n.d.
6	DMAP	<5%	n.d.	n.d.

^aStandard procedure: Ni(cod)₂ (20 mol%) and Ph₃P (40 mol%) were dissolved in 2.5 mL toluene. 1-octene (1 mL), a base (600 mol%), benzaldehyde (100 mol%, 0.5 mmol), and Et₃SiOTf (175 mol%) were added. The reaction mixture was stirred 18 h at 23 °C.

^bYields and ratios were determined by ¹H NMR using DMF as a standard.

^cRatio was determined from the desilylated product by ¹H NMR.

Tertiary amines were further examined in the 1-octene coupling reaction (Tables 5–6), and triethylamine was consistently superior to other tertiary amines (Table 5, 1–4 and Table 6, 1–5). Tertiary amines smaller or larger than triethylamine compromised the yield of the coupling reaction (Table 5, entries 2–4).

Table 5.

Effect of Bases in the 1-Octene–Benzaldehyde Coupling (Cy₂PhP)

entry	base	combined yield (2b′+2b) b	ratio (2b′:2b) c
1	Et3N	61%	78:22
2	Et(i-Pr)2N	10%	60:40
3	Cy2NMe	20%	60:40
4	N-methylpyrrolidine	7%	71:29
5	2,6-lutadine	12%	58:42

^aStandard procedure: Ni(cod)₂ (20 mol%) and Cy₂PhP (40 mol%) were dissolved in 1.5 mL alkene. A base (600 mol%), benzaldehyde (100 mol%, 0.1 mmol), and Et₃SiOTf (100 mol%) were added. The reaction mixture was stirred 18 h at 23 °C.

^bYields were determined by ¹H NMR using DMF as a standard.

^cRatio was determined by ₁H NMR of the crude reaction mixture.

The nature of the tertiary amines in determining the yield deserves further comment. It appears that a balance of the nucleophilicity, basicity, and steric bulk of the amine base is required for the coupling reaction to occur efficiently. Amines can compete with the phosphorus ligand, alkene, and aldehyde for a coordination site on nickel. A more nucleophilic (σ-electron donating) or smaller amine might hinder the coordination of any of the other required components to the nickel catalyst. For instance, the less nucleophilic N-methylmorpholine (Table 6, entry 3) provides a better yield than N-methylpiperidine (Table 6, entry 4).

In summary, triethylamine is the best base for the nickel-catalyzed coupling of alkenes and aldehydes, probably because of a combination of low coordinating ability and appropriate basicity.

Source of Nickel

Since the precatalyst Ni(cod)₂ has two chelating diene ligands (1,5-cyclo-octadiene), other nickel(II) precatalysts without alkene ligands were examined, including Ni(Ph₃P)₄, Ni(acac)₂/Ph₃P/DIBAL-H, Ni(Ph₃P)₂Cl₂/n-BuLi, and Ni(Ph₃P)₂Br₂/n-BuLi (Table 7). Only the Ni(acac)₂/Ph₃P/DIBAL-H system is as efficient as Ni(cod)₂/Ph₃P (entry 3). Ni(Ph₃P)₄ is saturated with phosphine ligand, and perhaps alkene coordination to the nickel catalyst is thus inhibited. Similarly, Ni(cod)₂/Cy₂PhP is more efficient than Ni(acac)₂/Cy₂PhP/DIBAL-H and Ni(Cy₂PhP)₂Cl₂/n-BuLi (Table 8).²³ Therefore Ni(cod)₂ was used in all subsequent investigations.

Table 7.

Examination of Other Nickel Precatalysts ^a

entry	catalyst	yield (2b) b	yield (2b′) b	ratio (2b:2b′) c	combined yield b (2b+2b′)
1	Ni(cod)2/Ph3P	78%	6%	93:7	84%
2	Ni(Ph3P)4	41%	3%	93:7	44%
3	Ni(acac)2/Ph3P/DIBAL-H	72%	5%	93:7	77%
4	Ni(Ph3P)2Cl2/n-BuLi	32%	2%	94:6	34%
5	Ni(Ph3P)2Br2/n-BuLi	<5%	<5%	n.d.	<5%

^aStandard procedure: The nickel precatalyst system (20 mol%) was dissolved in 2.5 mL toluene. The alkene (1 mL), triethylamine (600 mol%), the aldehyde (100 mol%, 0.5 mmol), and Et₃SiOTf (175 mol%) were added. The reaction mixture was stirred 48 h at 23 °C.

^bYields were determined by ¹H NMR using DMF as a standard.

^cRatios were determined by ¹H NMR of the crude reaction mixture.

Table 8.

Examination of Other Nickel Precatalysts ^a

entry	catalyst	yield (2b′) b	yield (2b) b	ratio (2b′:2b) c	combined yield b (2b′+2b)
1	Ni(cod)2/Cy2PhP	52%	21%	71:29	73%
2 d	Ni(cod)2/Cy2PhP	53%	20%	73:27	73%
3	Ni(acac)2/Cy2PhP/DIBAL-H	39%	16%	71:29	55%
4	Ni(Cy2PhP)2Cl2/n-BuLi	21%	7%	75:25	28%

^aStandard procedure: The nickel precatalyst system (20 mol%) was dissolved in toluene. The alkene (500 mol%), triethylamine (600 mol%), the aldehyde (100 mol%), and Et₃SiOTf (175 mol%) were added. The reaction mixture was stirred 48 h at 23 °C.

^bYields were determined by ¹H NMR using DMF as a standard.

^cRatios were determined by ¹H NMR of the crude reaction mixture.

^d1 mL 1-octene was used.

Substrate Scope

Applying the results of the studies of ligand and base effects, the substrate scope of the nickel-catalyzed coupling of alkenes, aldehydes, and silyl triflates was next examined (Tables 9–12). In general, ethylene and monosubstituted alkenes are superior substrates in this coupling reaction, while 1,1-disubstituted alkenes and acyclic 1,2-disubstituted (cis or trans) alkenes are significantly less reactive. Trisubstituted alkenes do not react under the standard reaction conditions.

Table 9.

Preparation of Homoallylic Alcohol Products from Nickel-Catalyzed Alkene–Aldehyde Couplings ^a

entry	alkene	aldehyde	major product (2)	yield (%)(2:2′) b,c	E:Z (2) b
1 d		PhCHO	2a	73 (89:11)	n.a
2		PhCHO	2b	85 (95:5)	75:25
3 e				72 (>95:5)	75:25
4 e		p-anisaldehyde	2c	85 (>95:5)	75:25
5 e		p- Cl(C6H4)CHO	2d	37 (>95:5)	74:26
6 f		2-naphthaldehyde	2e	88 (>95:5)	70:30
7		1-methyl-2-indole- carboxaldehyde	2f	56 (>95:5)	83:17
8 f		t-BuCHO	2g	64 (>95:5)	78:22
9		PhCHO	2h	86 (92:8)	>95:5
10		o-anisaldehyde m-anisaldehyde p-anisaldehyde	2i (ortho) 2i (meta) 2i (para)	78 (92:8) 98 (92:8) 99 (92:8)	>95:5 >95:5 >95:5
11 f,g		p-anisaldehyde	2i (para)	98 (92:8)	>95:5
12 f		2-naphthaldehyde	2j	88 (95:5)	>95:5
13		1-methyl-2-indole-carboxaldehyde	2k	57 (>95:5)	>95:5
14		t-BuCHO	2l	65 (>95:5)	78:22
15		p-anisaldehyde	2m	91 (92:8)	69:31
16		PhCHO	2n	82 (>95:5)	81:19
17			2o	95 (86:14) h	n.a.
18			2p	99 (75:25) h	n.a.
19			3q	14 (>95:5) i	n.a.

^aStandard procedure: (entries 1–8, 15–18): To a solution of Ni(cod)2 (0.1 mmol) and EtOPPh₂ (0.2 mmol) in toluene (2.5 mL) at 23 °C under Ar were added the alkene (0.5 mL), triethylamine (3.0 mmol), the aldehyde (0.5 mmol), and Et₃SiOTf (0.875 mmol). The mixture was stirred 48 h at room temperature and purified by chromatography (SiO₂). Entries 9–14: Ph₃P was used in place of EtOPPh₂.

^bYields were determined by ¹H NMR using DMF as a standard.

^cSee Supporting Information for structures of the minor products (2a′-2p′).

^dPropene (1 atm) was used in place of Ar.

^eReaction time 18 h.

^fReaction temperature 35 °C.

^gFivefold larger reaction scale.

^hratio of 2:3.

ⁱratio of 3: (2q+2q′).

Table 12.

Coupling of Oxygen-Containing Alkenes with Aldehydes ^a

entry	R1 (alkene)	R2 (aldehyde)	ligand	major product	yield (%) b	ratio (4:4′) b	ratio (E:Z) c
1		Ph	Cy2PhP	4d	<5	n.d.	-
			(EtO)Ph2P	4d′	<5	n.d.	n.d.
2 e		Ph	Cy2PhP	4e	21	n.d.	-
			(EtO)Ph2P	4e′	<5	n.d.	n.d.
3		o-anisyl	Cy2PhP	4f	44 d	73:27	-
4 e		o-anisyl	(EtO)Ph2P	4g	66	7:93	50:50

^aStandard procedure: To a solution of Ni(cod)₂ (0.1 mmol) and ligand (0.2 mmol) in toluene (2.5 mL) at 23 °C under Ar were added the alkene (2.5 mmol), triethylamine (3.0 mmol), the aldehyde (0.5 mmol), and Et₃SiOTf (0.875 mmol). The mixture was stirred 48 h at room temperature and purified by chromatography (SiO₂).

^bDetermined by ¹H NMR of the crude reaction mixture using DMF as a standard.

^cThe ratio was determined by ¹H NMR of the mixture of E and Z homoallylic alcohols after the silyl group of the coupling product was removed by TBAF.

^dIsolated yield.

^e1.5 mmol alkene was used.

The Ni–EtOPh₂P system efficiently catalyzes the coupling of monosubstituted alkenes and simple aromatic aldehydes such as benzaldehyde (Table 9). While the couplings of ethylene with most aldehydes usually take less than 8 h to reach completion, those involving monosubstituted alkenes typically require more than 18 h (entries 2–3). Nevertheless, with EtOPh₂P as the ligand, nickel catalyzes the coupling of several monosubstituted alkenes and aldehydes in excellent yield. The reaction is also highly regioselective and E/Z selective, favoring an E-homoallylic alcohol product.

Aromatic aldehydes (Table 9, entries 2, 6, 9, 12), heteroaromatic aldehydes (entries 7 and 13), and sterically demanding aldehydes (entries 8 and 14) are excellent coupling partners with monosubstituted alkenes, affording an E-homoallylic alcohol derivative as the major product and an allylic alcohol derivative as the minor product, with a selectivity >95:5 in most cases. Monosubstituted aromatic aldehydes of all substitution patterns are tolerated (ortho-, meta-, and para-, entry 10). Aldehydes with an electron donating substituent in the para position (p-MeO–, entry 4) are more reactive than aldehydes with an electron withdrawing group in the same position (Cl–, entry 5), consistent with the observation in the ethylene–aldehyde couplings. The product derived from p-chlorobenzaldehyde can be elaborated further by way of a cross-coupling reaction.

Allylbenzene is an excellent substrate, and the homoallylic products are useful styrene derivatives. The E-isomer is observed exclusively. Oligomerization of the coupling product is not observed, as evidenced by the excellent yield of the coupling reactions (entries 9–14).

Linear monosubstituted olefins such as propene and 1-octene are not the only terminal olefins that can participate in this nickel-catalyzed reaction (entries 1, 2, 9, 15). Alkenes with substitution at the homoallylic position couple with benzaldehyde in similar regioselectivity and E/Z selectivity as in the case of 1-octene (entry 16, as compared to entry 2).

Alkenes with substituents at the allylic position, on the other hand, afford different results. A homoallylic alcohol derivative is still the major coupling product in the coupling of 2-methyl-butene (entry 17) and vinylcyclohexane (entry 18) with benzaldehyde. However, the minor product is an E-1,3-disubstituted allylic alcohol, rather than the usual 1,2-disubstituted allylic alcohol obtained from the coupling of unbranched alkenes (eqs 2–3). The coupling of 3,3-dimethyl-butene and benzaldehyde yields exclusively 1,3-disubstituted allylic alcohol product (eq 4). This observation maybe important in understanding the mechanism of these transformations, which is discussed in more detail in the discussion section below.

(2)

(3)

(4)

Allylic, rather than homoallylic alcohol derivatives can be prepared by the nickel-catalyzed coupling of alkenes and aldehydes simply by substituting Cy₂PhP for EtOPh₂P (Table 10). Hence, propene couples with naphthaldehyde to provide the allylic alcohol product in good yield and with the highest selectivity (Table 10, entry 1). In contrast to the Ni–EtOPh₂P system, the homoallylic alcohol is the minor product in this case.

Table 10.

Preparation of Allylic Alcohol Products from Nickel-Catalyzed Alkene–Aldehyde Couplings^a

entry	R1 (alkene)	R2 (aldehyde)	R3SiOTf	major product	yield (%) b (2+2′)	ratio (2′:2) c
1 d	Me	naphthyl	Et3SiOTf	2r′	82	84:16
2	n-hexyl	Ph	Et3SiOTf	2b′	70	71:29
3 d	Me	p-anisyl	Et3SiOTf	2s′	95	82:18
4	Ph		Et3SiOTf	2k′	56 e	80:20
5	isobutyl	Ph	Et3SiOTf	2n′	62	71:29
6		Ph	Et3SiOTf	2t′	72	71:29
7	c-hexyl	Ph	Et3SiOTf	2p′	5 f	n.d.

^aStandard procedure: Ni(cod)₂ (20 mol%), Cy₂PhP (40 mol%) were dissolved in 2.5 mL toluene. Excess alkene, triethylamine (600 mol%), the aldehyde (100 mol%, 0.5mmol), and Et₃SiOTf (175 mol%) were added. The reaction mixture was stirred 18 h at 23 °C.

^bUnless specified, isolated yield of all coupling products.

^cRatios were determined by ¹H NMR of the crude reaction mixture.

^d1 atm propene (balloon) was used and naphthaldehyde (100 mol%) was mixed with Ni(cod)₂ and Cy₂PhP before the addition of toluene.

^eYields were determined by ¹H NMR using DMF as a standard.

^fIsolated yield of the allylic product 2p′.

Once again, aromatic aldehydes and heteroaromatic aldehydes couple with straight chain monosubstituted alkenes in good yield (entries 1–2, 4). Electron donating p-anisaldehyde is, as before, more reactive than benzaldehyde (entries 1 and 3). Therefore, based on all the data that we gathered so far, it seems to be the trend that generally electron-donating aldehydes are more reactive than electron-poor aldehydes.

There are, however, some differences in the substrate scope of the alkene in the Ni–Cy₂PhP system relative to that of the Ni–EtOPh₂P system. While branching at the homoallylic position of the alkene does not affect the coupling efficiency (entry 5), branching at the allylic position significantly attenuates the yield of the allylic alcohol product (entry 7). For example, vinylcyclohexane has a dramatically lower A:H ratio (entry 7), and the homoallylic alcohol and a 1,3-disubstituted allylic alcohol are the major products.

In a competition study, benzaldehyde undergoes coupling with a monosubstituted alkene selectively in the presence of a trisubstituted alkene (entry 6); the trisubstituted double bond is stable to the reaction conditions. Carbocyclization is not observed, nor do we observe any isomerization of the trisubstituted double bond in the coupling product. This result enables the use of a trisubstituted double bond as a masked version of other functional groups.

Given that heteraromatic aldehydes are competent substrates in these coupling reactions, we became interested in the effect of heteroatoms on the alkene. N-allylphthalimide, N-homoallylphthalimide and N-homoallyloxazolidinone undergo coupling in both the Ni–Cy₂PhP and Ni–EtOPh₂P systems (Table 11, entries 1–3). In particular, the coupling of N-allylphthalimide and benzaldehyde in the Ni–EtOPh₂P system affords an enamine that appears to be stable to the coupling conditions. In contrast, allylbenzoate and homoallylbenzoate esters are much less efficient (Table 12, entries 1–4). A small amount of the allylic product is detected only with homoallylbenzoate (entry 2). When the benzoate group is further away from the terminal double bond, a better yield of the desired coupling product is observed (entry 3). These findings suggest an interaction of the heteroatoms on the alkenes to the nickel catalyst. We propose that since the oxygen on the phthalimide is less nucleophilic, it does not bind to the nickel as tightly as the benzoate oxygen. Therefore the coupling of N-allylphthalimide occurs more efficiently than allylbenzoate (Table 11, entry 1 and Table 12, entry 1). As the benzoate becomes further away from the double bond, the benzoate is less likely to coordinate to the nickel catalyst, and the reactivity of the alkene is restored (Table 12, entry 3). The silyl ether-tethered alkene (entry 4) does not experience the heteroatom attenuation effect, likely for the same reason as the benzoate ester in entry 3.

Table 11.

Coupling of Nitrogen-Containing Alkenes with Aldehydes ^a

entry	R1 (alkene)	R2 (aldehyde)	ligand	major product	yield (%) b	ratio (4:4′) b	ratio (E:Z) c
1		Ph	Cy2PhP	4a	67	74:26	-
			(EtO)Ph2P	4a′	43	12:88	60:40
2		o-anisyl	Cy2PhP	4b	54	71:29	-
			Ph3P	4b′	76	<5:95	83:17
3		Ph	Cy2PhP	4c	60	83:17	-
			(EtO)Ph2P	4c′	28	10:90	n.d.

^aStandard procedure: To a solution of Ni(cod)₂ (0.1 mmol) and ligand (0.2 mmol) in toluene (2.5 mL) at 23 °C under Ar were added the alkene (1.5 mmol), triethylamine (3.0 mmol), the aldehyde (0.5 mmol), and Et₃SiOTf (0.875 mmol). The mixture was stirred 48 h at room temperature and purified by chromatography (SiO₂).

^bDetermined by ¹H NMR of the crude reaction mixture using DMF as a standard.

^cThe ratio was determined by ¹H NMR of the mixture of E and Z homoallylic alcohols after the silyl group of the coupling product was removed by TBAF.

Discussion

General Mechanistic Framework

We believe that the nickel species that catalyzes the alkene–aldehyde coupling reactions above is not functioning simply as a Lewis acid. We propose that the coupling reaction proceeds through the formation of oxanickellacycle from a nickel(0) complex (Scheme 1). A syn β-hydride elimination would afford the coupling product and a nickel–hydride species, analogous to a Heck reaction.^3a–c Finally, base-promoted reductive elimination of the nickel–hydride intermediate could regenerate the nickel(0) catalyst. Note that a base-mediated β-elimination of the oxanickellacycle via an E2-like mechanism cannot be completely ruled out. Based on our observations and in analogy to the Heck reaction,^3a–c we believe the nickel–hydride pathway is operative (see below).

Scheme 1.

Ligand Effects

The interactions of nickel with the ligand, alkene, and aldehyde govern the assembly of the oxanickellacycle, and the oxanickellacycle in turn determines the product distribution. Scheme 2 summarizes the factors that control the product ratio in the alkene–aldehyde coupling reactions. In general, use of large phosphines favors the allylic alcohol product (A) (e.g., the coupling of 1-octene with benzaldehyde with Cy₂PhP as ligand yielded allylic alcohol as the major product.) The use of small phosphines (Bu₃P, (EtO)Ph₂P, Ph₃P, etc) on the other hand affords the homoallylic alcohol (H) as the major product.

Scheme 2.

Size of Coupling Partners

The substituents on the alkene and aldehyde also affect the ratio of the coupling products. The alkene substituents can either be closer to the ligand or the aldehyde substituent in the oxanickellacycle. Allylic alcohol product A is obtained in a significant amount when the alkene has no branching at the allylic position. On the other hand, branching at the allylic position does not affect the coupling process when a small ligand, such as (EtO)Ph₂P, is used, and homoallylic allylic alcohol H is formed in good yield. 3,3-dimethyl-1-butene, a sterically demanding monosubstituted alkene with no allylic hydrogen, provides 1,3-disubstituted allylic alcohol A′ as the sole product.

A large substituent on aldehyde favors the production of homoallylic alcohol. Less than 5% allylic alcohol product is observed when propene or 1-octene is coupled with pivaldehyde with Cy₂PhP as the ligand. In the following sections, we propose a detailed model consistent with all of these observations.

We begin by examining oxanickellacycle 1 in more detail (Scheme 3). The β-hydrogen of the oxanickellacycle 1 is not aligned with the C–Ni bond. Since β-hydride elimination generally occurs in the syn orientation, The –OSiEt₃ group must dissociate from nickel to allow bond rotation such that the β-hydrogen can align with C–Ni bond. At this stage, β-hydride elimination occurs and allylic product A is formed. The larger the phosphine ligand relative to the aldehyde substituent, the more likely oxanickellacycle 1 dominates because the alkene substituent would thus avoid severe steric repulsion with this ligand. The data shown in Table 2 support this proposal; the A:H ratio increases with the cone angle of the trialkylphosphine.

Scheme 3.

Oxanickellacycle 2 accounts for the formation of homoallylic alcohol H and allylic alcohol A′. Examination of oxanickellacycle 2 reveals that although the β-hydrogen in the oxanickellacycle (H_endo) is not aligned with the C–Ni bond, there are β-hydrogens outside the oxanickellacycle (H_exo) that are appropriately poised for β-hydride elimination once a free coordination site is available (Scheme 4).^3e The preferred conformation would align R² of the alkene trans to the C–C bond of the oxanickelacycle 2. Dissociation of one of the ligand on nickel provides a free coordination site for the syn β-hydride elimination to occur and provides the E-homoallylic alcohol H.

Scheme 4.

In order for the unusual allylic alcohol (A′) to form, the β-hydrogens in the oxanickellacycle (H_endo) must be eliminated instead of the exo-β-hydrogen (H_exo). Such a process requires dissociation of –OSiEt₃ and maybe favored when the exo-β-hydrogen is not aligned with the C–Ni bond, or when there is no exo-β-hydrogen (Scheme 5).

Scheme 5.

The coupling of vinylcyclohexane and benzaldehyde serves as a good example to illustrate the formation of 1,3-disubstituted allylic alcohol A′ (Scheme 5). Neither R¹ nor R² of vinylcyclohexane is a hydrogen atom, and hence the allylic position is very sterically encumbered. The usual allylic alcohol product A is not favored because the large substituent of vinylcyclohexane will not be accommodated next to the aldehyde substituent (R) in oxanickellacycle 1 (Scheme 3) due to severe steric repulsion.

Experimental data support this theory: The coupling of vinylcyclohexane with benzaldehyde using Cy₂PhP as ligand yields only 5% of the allylic alcohol product A (Table 10, entry 7, as compared with other unbranched alkenes in Table 10, entries 1–6). Using a smaller ligand, such as (EtO)Ph₂P, the large substituents in vinylcyclohexane can be accommodated by being closer to the ligand than to the aldehyde substituent, favoring oxanickellacycle 2 (Scheme 5). The exo-β-hydrogen of the oxanickellacycle, when aligned to with C–Ni bond, induces an unfavorable steric interaction between the cyclohexyl group and the C–C bond of the oxanickellacycle. Therefore the rate of β-hydride elimination from the exo-β-hydrogen decreases, and that of the endo-β-hydrogen increases, resulting in a greater amount of the unusual E-allylic product A′. The E-double bond geometry of A′ is obtained by minimizing steric repulsion during the β-H elimination step.

Alkenes without an allylic hydrogen cannot afford homoallylic alcohol products in the nickel-catalyzed alkene–aldehyde coupling reaction. For example, 3,3-dimethyl-1-butene, with no allylic hydrogen, couples with benzaldehyde to give exclusively E-1,3-disubstituted allylic alcohol product (A′).²⁴ Also, it appears that the steric bulk of the tert-butyl group renders formation of oxanickellacycle 1 extremely difficult, eliminating the possibility of affording 1,1-disubstituted allylic alcohol product A.

The proposed mechanistic framework is also supported by Ogoshi’s observation that cyclization of an α,ω-enal to form an oxanickellacycle is facilitated by the presence of a silyl triflate.¹⁷ A control experiment confirms that without silyl triflate, no coupling product is observed.²²

The evidence for the β-hydride elimination as the next step is the observation of isomerization and dimerization (hydrovinylation) of the starting olefins, which suggests the presence of a nickel–hydride (Ni–H) species, likely formed by a β-hydride elimination.^3d The requirement of a base in this catalyst system also supports the presence of a Ni–H species. A β-hydride elimination and subsequent base-assisted removal of triflic acid (reductive elimination) from the Ni–H species regenerates the Ni(0) catalyst (Scheme 1) and may also minimize side reactions by suppressing the presence of the Ni–H species.

We do not believe the direct precursor to the oxanickellacycle in this coupling reaction is a cationic nickel (II) species. Ni²⁺, Pd²⁺, and Pt²⁺ catalysts have been reported to be effective Lewis acids for carbonyl-ene reactions.^18g–i The nickel-catalyzed coupling of alkenes, aldehydes, and silyl triflates affords carbonyl-ene-type products in good yield, but the substrate scope is entirely different from that of a Lewis acid-catalyzed carbonyl-ene reaction. While the three cationic group 10 transition metal catalysts are effective in the carbonyl-ene reaction of the more nucleophilic alkenes such as 1,1-disubstituted alkenes and the more electrophilic aldehydes such as glyoxylate esters, they do not promote the coupling of monosubstituted alkenes and simple aldehydes.

The nickel catalyst system has the opposite alkene and aldehyde substrate scopes relative to those of the carbonyl-ene reaction. The nickel–phosphine catalyst selectively reacts with monosubstituted olefins, and we observe that electron rich aldehydes, such as p-anisaldehyde, consistently provide better yield than benzaldehyde and electron-deficient aldehydes. Although this coupling reaction readily provides homoallylic alcohol products corresponding to a carbonyl-ene reaction, it is more likely that the oxanickellacycle precursor is a Ni(0) species, and probably not just a Lewis acid catalyst.

To illustrate the difference between the Ni(0)–phosphine system and a Lewis acid system, β-citronellene and benzaldehyde were coupled under two conditions; using a classical Lewis acid and the Ni–EtOPh₂P conditions.6b As expected, the Lewis acid-catalyzed reaction reacts at the more nucleophilic trisubstituted double bond. For the Ni–EtOPh₂P system, however, the monosubstituted double bond reacts preferentially because it is the kinetically more accessible double bond. These observations are also in accord with many palladium-catalyzed reactions of alkenes (such as Wacker oxidation and alkene hydroamination), in that a monosubstituted double bond is usually more reactive than a more substituted double bond.^3a

The difference in substrate scope between the nickel-catalyzed alkene–aldehyde coupling and the carbonyl-ene reaction is further illustrated by competition experiments between a monosubstituted alkene and a 1,1-disubstituted alkene (Scheme 7).^25–26 Equal amounts of allylbenzene and methylenecyclohexane were included in the otherwise standard coupling conditions. The coupling reaction was highly selective; 92% of all of the coupling products detected are derived from allylbenzene. The presence of methylenecyclohexane does not change the H:A ratio of the coupling products of allylbenzene (as compared to Table 9, entry 10). A similar trend is observed between 1-octene and methylenecyclohexane (as compare to Table 9, entry 4), but the presence of excess methylenecyclohexane in the reaction mixture seems to lower the yield of the coupling reaction. This lower efficiency might be due to competition for a coordination site on nickel between monosubstituted alkenes and methylenecyclohexane.

Scheme 7.

Competition Experiments Between Monosubstituted and 1,1-Disubstituted Alkenes ^25–26

Further evidence that supports the notion that the nickel-catalyzed coupling of alkenes, aldehydes, and silyl triflates does not involve a carbonyl-ene reaction mechanism is that ethylene, with no allylic hydrogen, also participates in this coupling reaction using the same Ni–phosphine catalyst system.

Common side reactions in these nickel-catalyzed reactions are the dimerization (hydrovinylation)²⁷ and isomerization²⁸ of the starting olefin. One explanation for the requirement of excess alkenes in this coupling reaction is that the terminal alkene is isomerized to an internal alkene and that this new internal alkene is not reactive in the coupling process. While isomerization of olefin is common in the coupling reaction, hydrovinylation of olefins is observed in small amounts only when the alkene–aldehyde coupling process is not efficient. The presence of a base in the coupling reaction may keep the Ni-H concentration to a minimum, thus suppressing some of these side reactions.

Summary

The nickel-catalyzed coupling of alkenes, aldehydes, and silyl triflates represents a new alternative to both allylmetal reagents and alkenylmetal reagents (Scheme 8). The parent allylmetal reagent and vinylmetal reagent can now be replaced by propene and ethylene, respectively, using the nickel-catalyzed processes as described herein. The preparation of a terminal, monosubstituted alkene is generally more straightforward than that of the allylmetal species such as those shown in Scheme 8.

Scheme 8.

The transformation in this nickel-catalyzed alkene–aldehyde coupling reaction is, in effect, a C–H functionalization reaction of the alkene, involving addition to an aldehyde. Mechanistically, an entirely different process likely occurs, rather than oxidative addition into a C–H bond that would be expected to have a relatively high energy activation barrier.

Unlike the related transition metal-catalyzed reductive coupling reactions developed by our group and others,⁸ the nickel-catalyzed coupling of alkenes, aldehydes, and silyl triflates described in this work is not an overall reductive process (Scheme 9). Thus, the coupling of an alkene and an aldehyde, in theory, does not require a third component to form the allylic or homoallylic alcohol derivatives. However, both alkenes and aldehydes are generally unreactive toward each other. Thus, activation of either or both components is necessary. The Lewis acid-catalyzed carbonyl-ene reaction serves as a good example. Intermolecular carbonyl-ene reaction between monosubstituted alkene and unactivated aldehydes such as acetaldehyde is not a practical method under thermal conditions. The presence of a Lewis acid, however, allows the coupling to proceed at room temperature. The Lewis acidic nature of silyl triflate in the nickel-catalyzed alkene–aldehyde coupling reaction likely plays a similar role, providing sufficient activation of the electrophile for the nickel catalyst to promote the coupling reaction.

Scheme 9.

The two classes of the nickel-catalyzed coupling reactions of alkene, aldehyde, and silyl triflate presented here represent unique, non-reductive coupling processes that allow the preparation of derivatives of allylic alcohols or homoallylic alcohols from readily available olefins. The selectivity for these two products is highly ligand dependent, and high selectivity in either direction is possible. These coupling reactions are mechanistically different from Lewis acid-catalyzed carbonyl-ene reactions, and conceptually, alkenes serve as substitutes for both allylmetal reagents and alkenylmetal reagents.

Scheme 6.