An Organometallic Umpolung Approach for Iron-Mediated Propargylic C–H Etherification

Abstract

The synthesis of propargyl ethers is an efficient way to access a variety of pharmaceuticals and natural products. However, direct etherification of simple alkynes via propargylic C–H activation remains largely underreported, while propargylic substitution of prefunctionalized alkyne starting materials remains the dominant method for the synthesis of propargyl ethers. Herein, we report an organometallic umpolung approach for iron-mediated C–H propargylic etherification. The stepwise iron-mediated C–H activation followed by mild oxidative functionalization enables the first use of unadorned alkynes to preferentially couple with alcohols for the efficient synthesis of synthetically important propargylic ethers with excellent functional group compatibility, chemoselectivity and regioselectivity.

Etherification enabled by C–H functionalization, specifically the construction of C–O bonds α to π bonds, is an important transformation to access a variety of pharmaceuticals and natural products.^[1] While a wide range of C–H etherification reactions has been developed and utilized for the construction of benzylic^[2] and allylic^[3] ethers, analogous propargylic C–H etherification remains conspicuously unreported, with propargylic substitution being the dominant approach to access propargylic ethers.

Among all propargylic substitution reaction and their derivatives, the Nicholas reaction is among the most reliable and heavily employed stoichiometric organometallic transformations for the synthesis of alkynes carrying functional groups at the propargylic position.^[4] Given its excellent functional group tolerance, the Nicholas reaction has been widely accepted as a powerful tool in the synthesis of natural products and bioactive molecules.^[5] In addition to the Nicholas reaction, a multitude of other propargylic substitution processes have also been developed.^[6] These include Lewis or Brønsted acid catalyzed processes,^[7] as well as transition metal catalyzed cross coupling^[8] and allenylidene mediated propargylic substitution.^[9] However, despite substantial recent progress, including the development of enantioselective processes, only functionally simple alcohols have been applied in these propargyl etherification reactions. Moreover, reactions mediated by unbounded propargyl cations are highly substrate dependent and, in many cases, may give the isomeric allene derivative as a side product.^[10] Yet, all currently reported propargylic etherification reactions require the pre-installation a leaving group.

In contrast to substitution-based strategies, the C–H functionalization of unadorned alkynes at the propargylic position is significantly underexplored, compared to analogous allylic C–H etherification,^[11] in part due to the higher α-C–H bond dissociation energy (BDE) of alkynes (92.5 kcal mol⁻¹ for propyne) compared to analogous alkenes (88.2 kcal mol⁻¹ for propene).^[12] Previous strategies on the oxidative coupling between nucleophiles and alkynes generally require electronic activation of the propargylic C–H bond with additional heteroatoms or π-bonds at the α-position,^[13] or proceed through intramolecular C–H insertion of metal carbene/nitrene intermediates derived from nucleophilic precursors.^[14] However, since metal-oxo species that function as oxene equivalents are limited to the delivery of a single oxygen atom, the coupling between alkynes and oxygen nucleophiles remains a relatively limited process. Direct oxygenation products like ynones,^[15] propargyl alcohols,^[16] and propargyl peroxides ^[17] can be obtained under catalytic and stoichiometric oxidative conditions, while a limited range of propargyl esters can be synthesized via the Kharasch-Sosnovsky reaction.^[18]

However, to the best of our knowledge, the synthesis of propargyl ethers from the simple alkynes by C(sp³)–H functionalization has not been previously reported. The lack of such transformation is not surprising, given the inapplicability of formal oxene insertion processes and the difficulty of controlling processes that generate free propargylic cations, necessitating the development of a new mechanistic framework. Moreover, the development of a generally applicable process would also need to address the selective functionalization of the propargylic position over allylic and benzylic C–H bonds that may also be present in more structurally complex substrates.

Previously, our group has developed a strategy for the catalytic generation of propargylic anion equivalents from alkynes for redox-neutral C–H functionalization.^[19] In this strategy, electron-rich allenyliron intermediates (Scheme 1b, A) are readily generated by deprotonation of alkyne–[Fp*]⁺ (Fp* = Cp*Fe(CO)₂, Cp* = η⁵-C₅Me₅) complexes by amine bases. The allenyliron intermediates underwent reactivity with electrophiles to deliver propargylic functionalization products in high regioselectivity. Intriguingly, reports by Selegue, Lapinte, and Giering indicate that, upon one electron oxidation, allyl-, alkenyl- and alkynyl-iron species similarly undergo oxidative coupling reactions in a highly regiocontrolled manner.^[20] These studies indicate that cyclopentadienyliron-based organometallic species can undergo polarity reversal (umpolung) to act as electrophiles when subjected to one-electron oxidation.^[21] Additionally, Lapinte and Sponsler had reported structurally characterized vinyl- and dienyl-iron radical cations and dications as part of electrochemical investigations.^[22] Consequently, we speculated that our Fp*-derived allenyliron complexes could similarly be oxidized to organometallic radical cations that serve as propargyl cation synthons for an umpolung propargylic functionalization process (Scheme 1b, B). In particular, the coupling of these radical cationic species with alcohols would generate propargylic ethers, potentially with high functional group compatibility and unique regioselectivity and chemoselectivity. Based on these considerations, we report herein the successful development of such a process, one that could be applied to the preparation of propargylic ethers from various internal alkynes in the presence of a range of heterocycles and pharmaceutical derivatives.

Scheme 1.

State of art for propargylic ethers synthesis via C–O bond formation.

To begin our investigation, we measured the oxidation potential of allenyliron species 1a’ to be +0.65 V vs. ferrocene (E^o_Fc+/0 = +0.45 V vs. Ag/AgCl) in CH₂Cl₂. With the electrochemical data obtained, we considered Ag⁺ sources as strong yet functional group tolerant oxidants, which were previously used in the oxidation of iron half-metallocene complexes.^[23] We were pleased to find that oxidation of 1a’ with AgBF₄ in the presence of benzyl alcohol (2a) afforded benzyl ether 3a in 8% yield (Table 1, entry 1). Of solvents tested for the reaction (entries 2-6), 2-methyltetrahydrofuran (2-MeTHF) was found to be optimal (entry 6). Subsequently, several other common oxidants were examined (entries 7-13). While AgOTf was also effective (entry 7), an alternative organic oxidant thianthrenium tetrafluoroborate (TTBF₄) also afforded the product in significant yield (entry 12). No product was observed without the addition of oxidant (entry 14). It was found that the reaction occurred rapidly, reaching completion within 5 min (entry 15). To improve the operational simplicity of the propargylic ether synthesis, a one-pot protocol that uses alkyne 1a directly as the starting material and avoids the isolation of the allenyliron intermediate (1a’) was developed (entry 17), with use of the alcohol as the limiting reagent further increasing the yield (entry 18). Under otherwise successful reaction conditions, the replacement of the Cp* ligand on Fe with Cp resulted in no formation of 3a (entry 16).

Table 1.

Preparation of key organoiron species and reaction optimization.

Entry[a]	Solvent	Oxidant	Time	Yield (%)[b]
1	MeCN	AgBF4	15 h	8
2	DCM	AgBF4	15 h	0
3	MeNO2	AgBF4	15 h	29
4	1,4-Dioxane	AgBF4	15 h	10
5	THF	AgBF4	15 h	46
6	2-MeTHF	AgBF4	15 h	52
7	2-MeTHF	AgOTf	15 h	37
8	2-MeTHF	Ag2O	15 h	0
9	2-MeTHF	Cu(OTf)2	15 h	0
10	2-MeTHF	FcPF6	15 h	0
11	2-MeTHF	(NH4)2Ce(NO3)6	15 h	0
12	2-MeTHF	TTBF4	60 min	30
13[c]	2-MeTHF	(4-BrC6H4)3NSbCl6	60 min	<5
14	2-MeTHF	no oxidant	15 h	0
15	2-MeTHF	AgBF4	5 min	56
16[d]	2-MeTHF	AgBF4	5 min	0
17[e]	2-MeTHF	AgBF4	5 min	58
18[f]	2-MeTHF	AgBF4	5 min	74(71)[g]

^[a]Reaction conditions: allenyliron (0.1 mmol), BnOH (0.3 mmol), oxidant (0.2 mmol), solvent (1.0 mL), rt.

^[b]NMR yields.

^[c]Mixture of allenyliron and alcohol added to [(4-BrC₆H₄)₃N]SbCl₆ at −78°C and allowed to warm to rt over 30 min.

^[d]The allenyliron based on Fp (= CpFe(CO)₂) was used instead of the Fp* complex.

^[e]Reaction conditions: allenyliron (in situ from 0.11 mmol alkyne), BnOH (0.3 mmol), oxidant (0.2 mmol), solvent (1.0 mL), rt (see the Supporting Info GP-IV for detailed process).

^[f]Reaction conditions: allenyliron (in situ from 0.33 mmol alkyne), BnOH (0.1 mmol), oxidant (0.5 mmol), solvent (1.5 mL), rt (see the Supporting Info GP-I for detailed process).

^[g]Isolated yield.

DCM = dichloromethane. DMF = N,N-dimethylformamide. THF = tetrahydrofuran. 2-MeTHF = 2-methyltetrahydrofuran. FcPF₆ = ferrocenium hexafluorophosphate. TTBF₄ = thianthrenium tetrafluoroborate.

We then applied the optimized condition for reaction scope exploration using more heavily functionalized alkyne (1) and alcohol (2) coupling partners. The reaction demonstrated compatibility toward a wide-ranging variety of heterocycles including a thiazole (3b), a pyridine (3c), a quinoline (3d), a pyrimidine (3e), a pyrazole (3h), and others (3f, 3g, 3i-3k). Unsaturated functional groups including internal (3q) and terminal alkenes (3p), an alkyne (3r), an allene (3o) and a diazene (3s) were also well tolerated. This reaction could also be applied to the sterically demanding propargyl ethers (3l-3n). In addition, various nitrogen containing functional groups like a nitrile (3t), a nitro group (3u), an unprotected amine (3w), a squaramide (3z), a carboxamide (3aa) and an azide (3ad) were likewise tolerated, and these substrates reacted with chemoselectivity for propargyl ether formation. Highly polar functional groups, including a carboxylic acid (3v), and a phosphonate ester (3x), were also compatible. An aryl boronic ester (3y) was similarly unaffected by the reaction conditions. Fluorine containing functional groups including a trifluoromethylthio group (3ac), a pentafluorosulfanyl group (3ai), and a perfluorophenyl group (3an) could also be successfully employed. Several drug molecules or their derivatives (3ar-3au) were also successfully functionalized using this transformation.^[24] This protocol could also make use of water as nucleophile for the regioselective synthesis of propargylic alcohols by C–H hydroxylation (3am). Notably, this approach was successful for a substrate containing several other C–H bonds susceptible to hydrogen atom abstraction (3am), demonstrating the orthogonality of this this approach to methods based on hemolytic C–H cleavage.

Besides the excellent scope of alcohols that could be employed, this transformation was also compatible with alkynes bearing electron-rich (3ag) and electron-poor (3ah) aryl groups. However, yields tended to be highest for electron-neutral alkynes, while electron-rich alkynes tended to give more 1,3-enyne as the side product due to overoxidation (see the Supporting Information for a discussion), while the key radical cation intermediate derived from electron-poor alkynes are likely to be less stable.^[25] A span of different steric properties was also tolerated, ranging from a methyl-substituted phenylacetylene (3ae) to one bearing a bulky diphenylmethyl group (3l). Dialkylacetylenes (3ap, 3aq) could also be successfully employed to give propargylic ethers in moderate yields, with high regioselectivity observed for the more substituted product in the case of unsymmetrical substrates. An isolated enyne (3al) and an isolated diyne (3ak) could also be applied in this reaction to give propargylic ethers with exclusive chemo- and regioselectivity, again illustrating the distinct selectivity of this process over other methods mediated by H–atom abstraction. It bears mention that the ratio of alkyne and alcohol could also be reversed to accommodate the accessibility of each coupling partner. While lower yields were usually observed when alkyne was used as limiting reagent, the amount of metal and oxidant needed could be significantly reduced, allowing for improved ease of purification (see the Supporting Information). Finally, a modified protocol using thianthrenium tetrafluoroborate (TTBF₄) as the oxidant was also developed, allowing for a procedure that avoids the use of silver.

Compared to the Lewis acid promoted substitution of propargylic alcohols, this approach exhibited exceptional tolerance to substrates containing Lewis basic nitrogen atoms. For example, attempts to affect the substitution of 4-phenyl-3-butyn-2-ol with 3-pyridinemethanol (2c) using a range of Lewis acids (FeCl₃, Et₂O·BF₃, Sc(OTf)₃) that have been used to promote the formation of propargylic cation intermediates were uniformly unsuccessful, resulting in no conversion of the alcohol.^[26] However, the same alcohol was a fully competent coupling partner under our reaction conditions (3c, 77% yield). Moreover, in all cases examined, the current protocol gave propargylic ether product without concomitant formation of the regioisomeric allenylic ether (>20:1 in the crude material). These observations strongly suggest that C–O bond formation is not mediated by (free) carbocationic intermediates.^[27] Furthermore, a preliminary comparison with the Nicholas reaction using glycolic acid as the substrate (3av, 22% yield) indicated that this process exhibits distinct chemoselectivity compared to the classical cobalt-mediated process (see the Supporting Information). When the reaction was conducted on larger scale, it was found that, by quenching the reaction mixture with a solution of NaI in acetone, much of the iron reagent could be recovered chromatographically as the bench stable iodide complex, Fp*I. Similarly, when TTBF₄ was used as the oxidant, thianthrene could be recovered in near-quantitative yield (Scheme 3).

Scheme 3.

Scale-up and recovery of reagents.

Since the formation of allenyliron intermediate has been well-studied by our group and others,^{[19b, 19e, 28]} we focused our attention on clarifying the mechanism of the later umpolung C–O bond forming process. A series of mechanism experiments were performed to gain some evidence for the key proposed intermediate. First, 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO, 1.0 equiv) was added to the reaction mixture to intercept the proposed intermediate. The propargyl-TEMPO adduct 4 was isolated in 56% yield (Scheme 4a), while the propargylic ether product 3a was not observed. The TEMPO adduct 4 could also be obtained from 1a’ alone in the absence of alcohol. A further control experiment was conducted without the presence of AgBF₄. However, these conditions did not afford adduct 4, suggesting that formation of 4 required oxidation by silver(I). Moreover, treatment of 1a’ with TEMPO⁺ also resulted in no reaction, suggesting that role of silver(I) was to oxidize the allenyliron species rather than the TEMPO. Considered in conjunction with previous observations regarding the scope and the regiocontrol, these experiments suggest an intermediate with radical character is generated from the reaction of allenyliron species and AgBF₄.^[29] We propose that this intermediate is radical cation B, which is intercepted by the alcohol to give β-metalloradical C.^{[19d, 30]} Further oxidation then affords the observed product 3 and [Fp*]⁺ fragment (recovered as Fp*I).

Scheme 4.

Mechanistic experiments.

To summarize, we have reported an organometallic umpolung approach for iron-mediated propargylic C–H etherification capable of coupling unactivated alkynes and alcohols for the synthesis of highly functionalized propargyl ethers. The reaction could be performed under mild conditions under operationally simple reaction conditions and in short reaction times, and each of the two coupling partners could be extensively varied, leading to products of exceptional structural and functional group diversity. Preliminary mechanistic investigations support a reaction pathway involving the nucleophilic trapping of organometallic radical cation intermediates. Further applications of deprotonative propargylic C–H functionalization under oxidative conditions are under investigation in our laboratory and will be reported in due course.

Scheme 2.

Propargyl ether scope.^[a].

[a] Unless otherwise stated, all yields were obtained following GP-I: 1) alkyne (0.66 mmol), Fp*I (0.6 mmol), AgBF₄ (0.6 mmol), DCM (0.6 mL), rt, 15 h; 2) Et₃N (1.2 mmol), rt, 1 h; 3) alcohol (0.2 mmol), AgBF₄ (1.0 mmol), 2-MeTHF (2.4 mL), rt, 30 min. [b] Following GP-II: 1) alkyne (0.66 mmol), Fp*(thf)BF₄ (0.6 mmol), Et₂O·BF₃ (1.8 mmol), DCM (0.6 mL), rt, 15 h; 2) Et₃N (2.4 mmol), rt, 1 h; 3) alcohol (0.2 mmol), AgBF₄ (1.0 mmol), 2-MeTHF (2.4 mL), rt, 30 min. [c] Following GP-III: 1) alkyne (0.9 mmol), Fp*(thf)BF₄ (0.6 mmol), DCM (0.6 mL), rt, 15 h; 2) Et₃N (0.72 mmol), rt, 1 h; 3) alcohol (0.2 mmol), TTBF₄ (0.8 mmol), 2,6-di-t-butylpyridine (0.9 mmol), 1,2-DCE (2.4 mL), 0°C, 30 min. [d] 4.0 equiv Et₃N used in step 2. [e] 3.0 equiv Et₂O·BF₃ added in step 3. [f] alkyne as limiting reagent (see the Supporting Information).