Primary Alcohols via Nickel Pentacarboxycyclopentadienyl Diamide–Catalyzed Hydrosilylation of Terminal Epoxides

Abstract

The efficient and regioselective hydrosilylation of epoxides co-catalyzed by a pentacarboxycyclopentadienyl (PCCP) diamide nickel complex and Lewis acid is reported. This method allows for the reductive opening of terminal, monosubstituted epoxides to form unbranched, primary alcohols. A range of substrates including both terminal and non-terminal epoxides are shown to work, and a mechanistic rationale is provided. This work represents the first use of a PCCP derivative as a ligand for transition metal catalysis.

Graphical Abstract

Epoxide-opening reactions offer a versatile means to access a variety of functional group patterns.¹ Among the myriad possibilities, one simple yet underdeveloped transformation is the conversion of epoxides to alcohols via catalytic reductive epoxide opening, ² especially of terminal epoxides 1 to form primary alcohols 2 (Figure 1a).³ Coupled with epoxidation,⁴ such a reaction could serve as an attractive alternative to the traditional hydroboration/oxidation approach for the anti-Markovnikov hydration of alkenes 3.⁵ Unfortunately, catalytic reductive epoxide opening has proven to be very difficult, particularly in regard to the regioselective reaction of terminal, non-styrenyl epoxides. Hydrogenolysis has been successfully applied to epoxide reduction, but these methods generally suffer from limited scope, are selective towards the Markovnikov product, and require forcing conditions.⁶ Methods which favor the formation of Anti-Markovnikov epoxides are generally limited in scope. Recently, Norton reported the generation of anti-Markovnikov alcohols via radical-based reduction of epoxides,⁷ although the method did not extend to terminal, mono-substituted epoxides. Beller reported major advances in this area with an iron-catalyzed hydrogenation of terminal epoxides with anti-Markovnikov selectivity and a cobalt-catalyzed regioselective hydrogenation of epoxides.⁸ Although the yields were generally high, these reactions required high pressure and elevated temperatures. Similarly, Warner reported an efficient isomerization-transfer hydrogenation sequence for the efficient generation of anti-Markovnikov alcohols.⁹ Thus, the development of direct regioselective epoxide reductions that occur at ambient temperature and pressure with broad substrate applicability remains a synthetic challenge.¹⁰

Figure 1.

(a) Conversion of terminal alkenes to primary alcohols. (b) PCCP 4 and PCCP-diamide 5 metal complexes.

In this regard, a number of reports have focused on catalytic hydrosilylation¹¹ as a milder approach to epoxide reduction, but none have achieved a generalized method for the regioselective reduction of terminal, monosubstituted epoxides.¹² To address this problem, we reasoned that increasing the Lewis acidity of a metal complex through new ligand development might be beneficial for both reactivity and regioselectivity.

One class of ligands that has a long and eminent history is the cyclopentadienyl (CP) anions, which participate in a variety of binding modes, most famously the η⁵-mode canonically found in ferrocene.¹³ In addition to the parent compound, substituted-CP ligands continue to offer useful new tools for reaction design.¹⁴ On the other hand, one distinct category of CP ligands that has not been exploited for catalysis is the pentacarboxycyclopentadienes (PCCPs) 4, despite their established ability to ligate metals via a variety of different binding modes (Figure 1b).¹⁵ Our group has exploited the highly electron-deficient nature, synthetic accessibility, and broad structural diversity¹⁶ of PCCPs in the development of a novel class of Brønsted acid catalysts.¹⁷ Furthermore, the multiple Lewis basic sites on the PCCP framework offer the possibility of cooperative activation via interaction with Lewis acidic additives, which we speculated might be particularly useful for reactions with epoxide substrates.¹⁸ Here, we show that derivatives of these intriguing materials, PCCP diamides 5, can also be useful as ligands for transition metal catalysis by offering a platform for the catalytic, regioselective hydrosilylation of terminal epoxides.

For this work, we first examined the reaction of 1-decene oxide 6 (Table 1). Reactions were conducted by mixing 5 mol% Ni(OAc)₂, 5 mol% ligand, 10 mol% KO-t-Bu, and 1.1 equiv Ph₂SiH₂, followed by addition of the epoxide. Our initial attempt to use pentacarbomethoxycyclopentadiene L1 as ligand was unproductive (entry 1). On the other hand, the use of PCCP diamide ligands led to appreciable product formation. Aniline-derived ligand L2 enabled 36% conversion over 16 h at 80 °C (entry 2), albeit with a 1:2 regioselectivity in favor of the secondary product 8. Interestingly, reducing the temperature led to a reversal of selectivity (entries 2–5); at room temperature, the ratio of products 7:8 was 85:15 (entry 5). Unfortunately, modifications to the phenyl substituents of L2 were universally detrimental.¹⁹ Meanwhile, of the solvents screened, acetonitrile proved optimal for both reactivity and selectivity (entries 5-7). To increase reactivity, we found that the addition of up to 15 mol% BF₃•OEt₂ dramatically increased the reaction rate, albeit with a small decrease in regioselectivity (entries 8-10).¹⁷ Importantly, the BF₃•OEt₂ on its own did not result in conversion of the epoxide (entry 11).²⁰ Utilizing acetylacetone as the ligand in place of the PCCP ligand also resulted in only trace conversion of the epoxide (entry 12). Finally, we found that ligand L3 provided higher levels of selectivity than L2 while still retaining high reactivity (entry 13).

Table 1.

Optimization for Ni-PCCP catalyzed hydrosilylative opening of 1-decene oxide.^a

entry	ligand	solvent	temp (°C)	additive	conv(%)	selectivity (7:8)
1	L1	MeCN	80	–	0	–
2	L2	MeCN	80	–	36	39:61
3	L2	MeCN	60	–	30	48:52
4	L2	MeCN	40	–	23	84:16
5	L2	MeCN	rt	–	11	85:15
6	L2	PhMe	rt	–	7	62:40
7	L2	CH2Cl2	rt	–	8	80:20
8	L2	MeCN	rt	5% BF3 •OEt2	67	76:24
9	L2	MeCN	rt	10% BF3 •OEt2	100b	75:25
10	L2	MeCN	rt	15% BF3 •OEt2	100c	78:22
11	–	MeCN	rt	15% BF3 •OEt2	0d	–
12	–	MeCN	rt	15% BF3 •OEt2	<5e	–
13	L3	MeCN	rt	15% BF3 •OEt2	100	85:15

^aReaction Conditions: 5 mol% Ni(OAc)_2, 5 mol% PCCP ligand, 10 mol% KO-t-Bu, 1.1 eq silane at room temperature for 16 h. Conversion was determined by ¹H NMR against mesitylene internal standard.

^b43% conversion after 4 h.

^c65% conversion after 4 h.

^dReaction was performed without Ni(OAc)₂ or KO-t-Bu.

^e5 mol% acetylacetone used in place of PCCP ligand

We next examined the scope of this method (Table 2). In addition to 1-decanol (9, entry 1), 1-hexanol was also formed in good yield (10, entry 2). Branching near the epoxide was tolerated (entries 3 and 4), although a small drop in regioselectivity was observed when the branch point was next to the site of reduction (entry 4). 3-Phenylpropene oxide gave rise to 3-phenylpropanol (13) with good yield and selectivity (entry 5). In terms of functional group compatibility, we found that halogens and ethers were tolerated (entries 6-8), with minimal effect on regioselectivity. Strongly electron-withdrawing phenylsulfonyl and phosphine oxide groups led to a single regioisomer of product (17 and 18, entries 9 and 10). Notably, alkenyl and alkynyl substrates furnished products 19 and 20 with no hydrosilylation across the π-bonds (entries 11 and 12). However, in the case of substrate 21, in which the alkene was in closer proximity to the epoxide, hydrosilylation across the alkene was observed.²¹ Additionally, we note that reactions with substrates bearing sensitive (e.g. 22) or significantly Lewis basic (23-26) functionality were not viable substrates for this catalyst system.

Table 2.

Scope studies for hydrosilylation of mono-substituted terminal epoxides.^a

^aReaction Conditions: 5 mol% Ni(OAc)_2, 5 mol% L3 ligand, 10 mol% KO-t-Bu, 1.1 eq silane at room temperature for 16 h. Yields determined on purified products of major isomer. Selectivity determined by ¹H NMR of crude reaction mixture. Hydrolysis Conditions: 15% NaOH (aq) and MeOH at room temperature for 8 h.

^bYields determined by ¹H NMR versus mesitylene as an internal standard.

We also examined the ability of this catalyst system to reduce other types of epoxides (Table 3) In these cases, the more synthetically accessible ligand L2 was suitable. As shown a variety of 1,1-disubstituted (entries 1 and 2), cyclic disubstituted (entries 3-5), trisubstituted (entry 6), and tetrasubstituted (entry 7) epoxides reacted with high efficiency and complete regioselectivity. Styrenyl epoxides led to the anti-Markovnikov products 34–36 exclusively in each case (entries 8-10). Finally, both ester and amide functionality were compatible with this chemistry (entries 11 and 12), albeit with lower efficiencies.

Table 3.

Scope studies for hydrosilylation of non-terminal and 1,1-disubstituted epoxides.^a

^aReaction Conditions: 5 mol% Ni(OAc)_2, 5 mol% L2 ligand, 10 mol% KO-t-Bu, 15 mol% BF₃•OEt₂, 1.1 eq silane at room temperature for 16h. Yields determined on purified products. In all cases, a single regioisomer was detected by ¹H NMR on crude reaction mixtures. Hydrolysis Conditions: 15% NaOH (aq) and MeOH at room temperature for 8h.

^bYield determined by ¹H NMR versus mesitylene internal standard.

^cTBAF deprotection used for desilylation.

Our proposed mechanism invokes the mechanism proposed by Gade for the nickel-catalyzed hydrosilylation of styrenic epoxides.¹² The mechanistic rationale involves the initial reaction of the PCCP ligand (L2 or L3), Ni(OAc)₂, two equivalents of KO-t-Bu, and an equivalent of silane to generate the active nickel hydride intermediate 41 (Figure 2).²² We speculate that coordination of 41 to the epoxide 1 (cf. 42) followed by insertion of the hydride leads to the nickel alkoxide 43 in the regioselectivity determining step, which then engages another equivalent of silane to furnish product 44 and regenerate the complex 41. We speculate that the electron-deficient metal center activates the epoxide towards insertion by the nickel-hydride. We propose that BF₃•OEt₂ increases the Lewis acidity of the complex by coordinating to one of the ligand carboxyl oxygens, e.g. 41.¹⁷ This interaction helps to weaken the internal C–O bond of the epoxide and thereby improve the reaction rate while maintaining the regioselectivity. Further, we speculate that the regioselectivity arises from the sterics surrounding the metal center where a smaller ligand promotes hydride addition at the more sterically hindered site.

Figure 2.

Proposed mechanistic rationale for Ni-PCCP diamide / BF₃•OEt₂ catalyzed hydrosilylation of epoxides.

In addition to the in-situ generation of the catalyst, the presumed nickel-alkoxide species 40b was isolated to assess its reactivity directly. Ligand complexation of Ni(OAc)₂ and L2 followed by reaction with KO-t-Bu resulted in the observed precipitation of KOAc and a paramagnetic solid, presumed to be 40b. A comparison of the infrared spectra of 39b and 40b revealed a new stretch at 1006 cm⁻¹, consistent with the C-O stretching frequency of previously reported nickel alkoxide species.²³ Due to the generation of paramagnetic intermediates, we were unable to characterize the catalyst by NMR and all attempts to grow single crystals for x-ray crystallography failed. Previous attempts to characterize nickel-PCCP complexes by single x-ray crystallography have also been unsucsessful.¹⁵ In addition, 40b was found to be active for the reduction of 1-decene oxide under the standard conditions (97% conversion after 16 h to furnish a 77:23 mixture of 1-decanol and 2-decanol). Because the Ni-PCCP complex was found to be viable after removal of the KOAc byproduct, we propose that the role of BF₃•OEt₂ is to not to sequester the acetate byproduct of ligation.

Alternatively, we considered that the product might also arise by way of Meinwald rearrangement (cf. 45→46) followed by carbonyl hydrosilylation (Figure 2b),²⁴ although we viewed this possibility as less likely since catalytic BF₃•OEt₂ itself did not affect such a rearrangement (eq. 1), which has been reported previously.^20,25 In further evidence against this mechanism, when the reactions of terminal epoxide 49 was conducted with PhSiD₂ (Figure 2c), the only observed product was 2-d-18. The absence of the regioisomeric product 1-d-18 is consistent with the pathway shown in Figure 2a.

In conclusion, we have developed a method for the regioselective reduction of terminal, monosubstituted epoxides. This effort was enabled by the use of PCCP-diamides as a new class of ligand, which in conjunction with a co-catalytic Lewis acid allowed for the generation of a highly active nickel complex.