Photoinduced C(sp³)–H Chalcogenation of Amide Derivatives and Ethers via Ligand-to-Metal Charge-Transfer

Abstract

A photoinduced, iron(III) chloride-catalyzed C–H activation of N-methyl amides and ethers leads to the formation of C–S and C–Se bonds via a ligand-to-metal charge transfer (LMCT) process. This methodology converts secondary and tertiary amides, sulfonamides, and carbamates into the corresponding amido-N,S-acetal derivatives in good yields. Mechanistic work revealed that this transformation proceeds through a hydrogen atom transfer (HAT) involving chlorine radical intermediates.

Graphical Abstract

Amines and amides are quintessential moieties in organic synthesis. They are found in a wide variety of natural products, pharmaceuticals, peptides, and fine chemicals.¹ Similarly, sulfur-containing compounds represent over 20% of FDA-approved drugs, which is the third largest heteroatom after oxygen and nitrogen.² A well-known bioactive scaffold in sulfur-containing drugs is the amido-N,S-acetal moiety found in natural products and antibacterials, such as penicillin, penicillin derivatives, and fusaperazine A (Scheme 1A).³

Scheme 1. Molecules of Interest & Current Technologies^a.

^a(A) Selected bioactive compounds of interest. (B) Radical approaches to amido-N,S-acetals. (C) Proposed photoinduced method.

Traditional synthetic approaches to access these amido-N,S-acetals proceed through either the substitution reaction of halogenated amides at the amino carbon, or through the addition reaction to N-acyl imines by thiol nucleophiles. Unfortunately these prefunctionalized imines and halogenated intermediates are often unstable to moisture, not easily stored for prolonged periods of time, and require sophisticated reaction conditions.⁴ Direct C–H thiolation of amides represent a more attractive approach for late-stage diversification and functionalization, and bypasses the need for these unstable intermediates.

In the past decade, two radical-promoted site-selective C–H thiolations of amides at the amino carbon have been developed (Scheme 1B).⁵ Unfortunately, both these methods use peroxides at 120 °C to achieve the desired transformation, which poses safety concerns and limits scalability.⁶ In contrast, the development of a photoinduced methodology would enable such transformations to occur under mild conditions at room temperature and presumably without the use of unstable and potentially explosive peroxides. Among photoinduced transformations, ligand-to-metal charge transfer (LMCT) processes have recently gained increased attention⁷ because of the simplicity of the methods and the affordability of its catalysts, which often use earth abundant metals, such as copper and iron,⁸ as opposed to traditional iridium and ruthenium photoredox catalysts. C–H activation transformations promoted by LMCT processes have been shown to proceed through chlorine radical intermediates by Doyle, Rovis, Walsh, and Schelter among others.⁹ The high reactivity of the chlorine radical makes it an ideal intermediate to initiate hydrogen-atom-transfer (HAT) processes with all kinds of C(sp³)–H bonds¹⁰ from activated ethers and amides to unactivated alkanes.

Recently, our group has reported various site-selective C–H functionalization of amides and ethers under mild and photoinduced conditions,^11a–c as well as photoinduced cross-couplings that generate C–S and C–Se bonds.^11d Combining these interests to our interest in green photoinduced methodologies,¹² we envisioned that chlorine radicals generated via LMCT processes could enable site-selective C(sp³)–H bond activation of the amino carbon of amides, sulfonamides, and carbamates to form stabilized alkyl radical and then react with disulfide to give the desired amido-N,S-acetals derivatives. Herein, we present the first example of a photoinduced C–H sulfenylation of amides, amide derivatives, and ethers using affordable and earth abundant FeCl₃ via a LMCT process (Scheme 1C).

Our investigations began using diphenyl disulfide (1a) and N,N-dimethylacetamide (DMA) (2a) as model substrates (Table 1) in the presence of 10 mol % iron(III) chloride as catalyst under air and irradiation of 40 W 390 nm LED. An initial solvent screen comparing acetonitrile (CH₃CN), DMSO, and THF (entries 1–3), revealed that the transformation only proceeds in acetonitrile (81%) and THF (26%). It is worth noting that neat conditions using DMA as solvent (entry 4) did not afford the desired product. This observation further differentiates this method with other radical approaches that require neat conditions to achieve the desired transformation, which limits the scope of the reaction to liquid amides.⁵ The use of other catalysts and iron sources (entries 5–8) did not improve the transformation (for full catalyst screen see Supporting Information S7). Lowering catalyst loading to 5 mol % (entry 9) had a deleterious effect on yield (59%). Similarly, using inert argon atmosphere (entry 10) lowered the yield to 61%. Switching LED lamps from 390 to 427 nm reduced the yield to 49% (entry 11). Control experiments in the absence of light (entry 12) or absence of catalyst (entry 13) did not generate the desired product. Additional optimization experiments detailing catalyst loading, reagent equivalencies, light sources, metal sources, solvents, and inert atmospheres are described in the Supporting Information S6–S10.

Table 1.

Optimization of the Reaction and Its Conditions^a

entry	catalyst (mol %)	solvent (mL)	yield (%)b
1	FeCl3 (10%)	CH3CN (2.0)	81 (78)c
2	FeCl3 (10%)	THF (2.0)	26
3	FeCl3 (10%)	DMSO (2.0)	NR
4	FeCl3 (10%)	DMA (2.0)	trace
5	FeCl3·6H2O (10%)	CH3CN (2.0)	67
6	FeCl2·4H2O (10%)	CH3CN (2.0)	30
7	FeBr3 (10%)	CH3CN (2.0)	NR
8	CeCl3 (10%)	CH3CN (2.0)	NR
9	FeCl3 (5%)	CH3CN (2.0)	59
10d	FeCl3 (10%)	CH3CN (2.0)	61
11e	FeCl3 (10%)	CH3CN (2.0)	49
12f	FeCl3 (10%)	CH3CN (2.0)	NR
13		CH3CN (2.0)	NR

^aOptimal reaction conditions: 1a (0.2 mmol, 1 equiv), 2a (1.0 mmol, 5 equiv), solvent (2 mL), anhydrous FeCl₃ (10 mol %), 390 nm LED (40 W), room temperature around reaction flask was 35 °C (heating caused by the LED lamp), reaction flask capped under air, 15 h.

^b1HNMR yields using dibromomethane as internal standard.

^cIsolated yield.

^dUnder argon.

^e427 nm LED.

^fReaction performed in the dark.

With optimal conditions in hand, we continued our investigations exploring the scope of aliphatic amides derivatives (Scheme 2). Using N,N-dimethylformamide (DMF) as amide source afforded the desired product 4 as a single regioisomer in good yield (65%), which differs from previous reports^5a that obtain a mixture of regioisomers in 1:1 ratio. Linear amide substrates, such as N,N-dimethylbutyramide, N,N-dimethylpentanamide, N,N,3-trimethylbutanamide, and N,N-dimethylisobutyramide, afforded the desired corresponding products 5–8 in moderate to good yields (48–80%).

Scheme 2. Amides Scope^a.

^aReaction conditions: 1a (0.2 mmol, 1 equiv), 2 (1.0 mmol, 5 equiv), MeCN (2 mL), anhydrous FeCl₃ (10 mol %), room temperature around reaction flask was 35 °C (heating caused by the LED lamp.), 15 h. All yields are isolated. ^b1 mmol scale reaction (see Supporting Information, page S19). ^cTwenty-four hours.

It is noteworthy that linear amides bearing various functional groups, nitriles, chloride, trifluoromethyl, and cyclopentane, were well tolerated and afforded products 9–13 in moderate to excellent yields (30–90%). Tetramethylurea afforded product 14 in good yield (68%). Secondary amides were also suitable substrates affording product 15 in 60% yield. Unsymmetrical aliphatic amides display excellent regioselectivity toward the methyl group over the methylene carbon affording products 16 and 17 in 70% and 55% yields, respectively. Aromatic moieties were also tolerated giving product 18 in moderate yield (62%). Importantly, when using N-methylpyrrolidone (NMP) as an amide source, two regioisomers were obtained (19a and 19b) in 75% overall yield with two regioisomers in a 3:2 ratio in favor of the primary carbon. As expected, this transformation could also proceed efficiently with sulfonamides and Boc-protected amine motifs (20–23) in moderate to good yields (45–75%), which are important frameworks in drug discovery and total synthesis.¹³ Other medicinally relevant motifs such as piperidine and pyrrolidine were also tested.¹⁴ While Boc-protected pyrrolidine and piperidine were not suitable substrates for this transformation (24 and 26), their acetyl-protected and mesyl-protected counterparts were well tolerated and afforded products 25 and 27 in 72% and 65% yield, respectively. Finally, to highlight the late-stage functionalization potential of our approach, an ibuprofen amide derivative was investigated to get the desired product 28 in good yield (72%), opening the door to the diversification of other bioactive compounds containing carboxylic acid moieties that can be readily transformed into their amide counter parts.

Next, we turned our attention to explore the scope of disulfides and other dichalcogenides (Scheme 3). Diaryl disulfides bearing electron donating and electron withdrawing groups on in the para position (29–31) were well tolerated (71–85%) with the exception of the nitro group, which afforded product 32 in moderate yield (45%). Sterics at the ortho position were also tolerated in good yields giving products 33 and 34 in 88% and 72% yield, respectively. Heteroaromatic substrates were also compatible and afforded products 35 and 36 in good yields (60–68%). Beyond aromatic disulfides, benzylic and alkyl disulfides were also investigated and afforded products 37 and 38 in low to moderate yields (45–63%). The lower yield obtained using di-tert-butyl disulfide is probably due to steric hindrance. Finally, other commercially available aryl dichalcogenides were investigated because of their potential bioactivity¹⁵ and synthetic interest.¹⁶ Lower yields were obtained when using diphenyl diselenide, but products 39–42 were obtained in synthetically useful yields for late-stage diversification studies (48–66%). It is worth mentioning that unreacted diselenide reagent was recovered in good yields. Unfortunately, diphenyl ditelluride was unsuitable in this reaction and only trace amounts of product 43 were formed. Previous methods have used FeCl₃ to add selenides and tellurides to unsaturated systems¹⁷ but not to amides.

Scheme 3. Disulfides Scope^a.

^aReaction conditions: 1 (0.2 mmol, 1 equiv), 2a (1.0 mmol, 5 equiv), MeCN (2 mL), anhydrous FeCl₃ (10 mol %), 390 nm LED (30 W), room temperature around reaction flask was 35 °C (heating caused by the LED lamp), 15 h. All yields are isolated.

Further exploration of the substrate scope revealed that this methodology could be extended to ethers, both cyclic and acyclic (Scheme 4). Commonly used ethers, such as THF, 1,4-dioxane, and diethyl ethers, afforded desired products 44–46 in moderate to good yields (66–75%). Tetrahydrothiophene was also evaluated and generated product 47 in good yield (80%), suggesting that this procedure may be expanded to other activated C(sp³)–H bonds alpha to heteroatoms. Finally, 2-methyl-THF afforded the desired product forming two regioisomers 48a (minor) and 48b (major) (1:3 ratio) as an inseparable mixture in 55% yield. Regioisomer 48b was generated as a mixture of diastereomers (d.r. = 3:1).

Scheme 4. Ethers Scope^a.

^aReaction conditions: 1 (0.2 mmol, 1 equiv), ethers (2 mL), anhydrous FeCl₃ (10 mol %), 390 nm LED (30 W), room temperature around reaction flask was 35 °C (heating caused by the LED lamp), 15 h. ^bMeCN (2 mL), ether (5.0 equiv, 1 mmol). All yields are isolated.

To expand our understanding of the reaction mechanism, a series of control experiments were performed (Scheme 5A). As expected, the addition of radical trapping agents, such as 2,2,6,6-tetramethyl-peperidinylxoyl (TEMPO), only afforded trace amount of product 3. The use of other trapping agents 1,1-diphenylethylene (1,1-DPE), 1,4-dinitrobenzene (1,4-DNB), and butylated hydroxytoluene (BHT) also produced similar results. Importantly, DMA radical intermediates were successfully trapped as TEMPO and 1,1-DPE adducts (49, 50) and observed by GC-MS. The chlorine radical was trapped with 1,1-DPE (51), while the thiyl radical was trapped as a 1,1-DPE and BHT adducts (52, 53) (Scheme 5A).

Scheme 5. Verification of the Presence of Radicals and Proposed Mechanism^a.

^a(A) Radical trapping experiments and (B) proposed mechanism.

UV–vis experiments (Supporting Information, pages S11–S13) show an absorption band for FeCl₃ with an absorption maximum at λ = 361 nm that tails off as far as 500 nm. On the other hand, neither diphenyldisulfide (1a) nor DMA (2a) absorb past 375 nm, strongly suggesting that FeCl₃ in MeCN is the species capable of photoexcitation (Figure S1, page S11). Indeed, when FeCl₃ absorbance was evaluated in different solvents (Figure S2, page S12), MeCN showed the most enhanced absorbance in the visible spectrum. Lastly, comparing different metal catalysts shows that FeCl₃ had the strongest absorption, which may explain the lack of reactivity of the other metal chloride salts (Figures S3 and S4, pages S12–S13).

On the basis of these observations and reported literature,^9,18 we propose that this transformation proceeds via a LMCT process between FeCl₃ and acetonitrile as a ligand (Scheme 5B). Iron(III) complex A can undergo excitation to B under visible-light irradiation and subsequently generates highly reactive chlorine radical C and forms iron(II) species D. Chlorine radical intermediate C performs a HAT process with the C(sp³)–H bond alpha to nitrogen of the amide 2a to generate stabilized alkyl radical E. Reaction between radical intermediate E and diphenyl disulfide 1a forms the desired product and produces thiyl radical F. Radical intermediate F can then oxidize iron(II) D to regenerate iron catalyst A (see cyclic voltammetry experiments in Supporting Information, page S10) in the presence of HCl and forms thiol G, which can reform disulfide 1a in the presence of O₂ from our atmospheric conditions (Scheme 5B). Although thermodynamically unfavorable,^18d there is a possibility that O₂ plays a minor role in the single electron oxidation of Fe(II) species D into Fe(III) catalyst A.

This work presents the first photoinduced C–H sulfenylation of amides, amide derivatives, and ethers using affordable and earth abundant FeCl₃ via a LMCT process. Broad functional group compatibility with synthetically useful moieties has been observed using this protocol, such a Boc- and mesyl-protected amines, nitriles, halogens, and sulfonamides, all of which are of great interest to medicinal chemistry. The transformation proceeds with moderate to excellent yields with secondary and tertiary amides, providing a complementary approach to current thiolation of N-methyl amides and ethers.