δ-C–H Halogenation Reactions Enabled by a Nitrogen-Centered Radical Precursor

Abstract

Nitrogen-centered radicals are versatile synthetic intermediates, with the ability to undergo diverse reaction pathways such as hydrogen atom transfer (HAT), β-scission, and addition across unsaturated systems. A long-standing impediment to the wider adoption of these intermediates in synthesis has been the difficulty in their generation. Herein, we disclose a new hydrazonyl carboxylic acid precursor to nitrogen-centered radicals and its application towards remote C–H fluorination and chlorination reactions of sulfonyl-protected alkyl amines via 1,5-HAT.

Graphical abstract

The selective functionalization of C(sp³)–H bonds remains a considerable challenge in organic chemistry^1–3. While the directed functionalization of C(sp³)–H bonds β and γ to a directing group via cyclometallated intermediates from C–H activation processes has been extensively explored³, examples for C–H functionalization at the δ-position and beyond remain rare^4–6 due to the entropically-disfavored formation of larger metallacycles. Heteroatom-centered radicals provide a reliable, alternative means to achieve remote C(sp³)–H functionalization via the process of 1,5-HAT (Figure 1A). In this vein, a variety of different approaches to remote, radical C(sp³)–H functionalization have been disclosed^7–10. Among this class of reactions, the Hofmann-Löffler-Freytag (HLF) reaction is a canonical example^11,12, wherein remote carbon-centered radicals (generated via HAT from a nitrogen-centered radical cation) react to generate remotely halogenated products. Although the HLF reaction has inspired a range of impressive remote C(sp³)–H functionalization reactions based on 1,5-HAT from nitrogen-centered radicals^13–25, examples for the remote, radical C–H fluorination of amine derivatives are relatively rare. Given fluorine’s important roles in medicinal²⁶, agro-²⁷, and materials²⁸ chemistry, we were curious as to whether we could develop an efficient means for remote C(sp³)–H fluorination via 1,5 HAT using nitrogen-centered radicals.

Figure 1.

A: Radical HAT as a general means for remote C–H functionalization. B: Our prior development of an oxime precursor to alkoxy radicals for remote C–H halogenation reactions. C: Development of a new hydrazonyl carboxylic acid precursor enables the remote C–H halogenation of sulfonamides.

In a previous work²⁹, we disclosed that oximes derived from pyruvic acid were efficient precursors to alkoxy radicals and demonstrated their application to achieve both the remote C(sp³)–H fluorination and chlorination of alcohols (Figure 1B). We believed that the design of an analogous hydrazone precursor for the generation of nitrogen-centered radicals might enable the remote fluorination of amine derivatives. We surmised that single-electron oxidation of the deprotonated hydrazonyl carboxylic acid 1 (Figure 1C) by a photoredox catalyst^30–32 would induce decarboxylation and expulsion of acetonitrile to deliver a nitrogen-centered radical competent for 1,5-HAT. Therefore, we set out to explore the possibilities of such a system, adopting sulfonamides as our model substrates due to their widespread use in pharmaceutically relevant molecules^33,34, their high stability toward hydrolysis, and their proven ability to engage in HAT processes^35–37.

Prior reports for the remote, radical C(sp³)–H fluorination of sulfonamides are rare. In 2018, Leonori³⁸ used a novel hydroxylamine-based precursor to generate nitrogen-centered radicals for HAT reactions; however, only one example is presented for the remote fluorination of a sulfonamide, targeting a tertiary C–H bond. More recently, Chen³⁹ disclosed a δ-C–H fluorination reaction of p-methoxybenzenesulfonamides using a proton-coupled electron transfer (PCET) strategy^40,41 for nitrogen-centered radical generation; however, this reaction was only effective with electron-rich sulfonamides. Meanwhile, Muñiz³⁷ demonstrated a radical crossover strategy for the remote fluorination of tolylsulfonamides, using a combination of iodine and a hypervalent iodine oxidant to generate the initial nitrogen-centered radical, followed by oxidation of the remote carbon radical to a carbocation and trapping with nucleophilic fluoride. Notably, this method was applicable only to the functionalization of tertiary C–H bonds. We hoped that our approach would yield a more general strategy for the remote fluorination of both methylene and methine C–H bonds across a wide variety of sulfonamides.

We devised a three step sequence to prepare the desired hydrazonyl carboxylic acid precursor 1 from the corresponding sulfonamides (Figure 2), featuring N-amination of the sulfonamide⁴², its condensation with methyl pyruvate, followed by hydrolysis to furnish the free hydrazonyl carboxylic acid. This amination-condensation sequence proved scalable to 15 mmol and delivered the desired compound in up to 58% across 3 steps. We then began to explore conditions to achieve the desired remote C–H fluorination. Pleasingly, when we applied the previously optimal conditions for our oxime system²⁹ to substrate 1a, we were able to achieve the δ-fluorinated product in 64% LCMS yield using Selectfluor as the fluorine source^43–45; subsequent modifications to this protocol in terms of bases and photocatalysts could improve the yield of this fluorination reaction to 71% isolated yield.

Figure 2.

Preparation of a new hydrazonyl carboxylic acid precursor to nitrogen-centered radicals.

We were also curious to see whether our new hydrazonyl carboxylic acid was competent in remote C–H chlorination reaction. Therefore a variety of bases, photocatalysts, and solvents were explored (see supporting information), ultimately allowing us to accomplish the remote C–H chlorination of 1a in 50% isolated yield using ethyl trichloroacetate (ETCA) as the chlorinating reagent^29,46. While several examples of remote chlorination are known for sulfonamides, ranging from classic HLF conditions⁴⁷ to more modern photochemical methods, these methods all require the intermediacy of nitrogen–chlorine bonds, either preformed^48,49 or generated in situ^50,51. In contrast, our method circumvents the need to generate nitrogen–chlorine bonds and so may find use in scenarios where the generation or intermediacy of nitrogen–chlorine bonds is undesirable. Indeed, given the synthetic versatility of organochlorine compounds and the presence of organochlorine motifs among several thousand natural products⁵², new technologies and approaches for aliphatic C–H chlorination will undoubtedly remain relevant in streamlining the synthesis of such compounds.

We then turned to explore the scope for both of these reactions (Figure 3, Figure 4). We were pleased to find that the reactions were successful across a range of different carbon scaffolds bearing varied functional groups. A variety of methylene C–H bonds were amenable to halogenation in yields of up to 82% yield for fluorination (2k) and 64% for chlorination (3f). Unlike in the C–H fluorination reaction, we observed significant formation (up to 20% yield) of the unfunctionalized sulfonamide as a by-product in the C–H chlorination reaction, which may explain the overall tendency for lower yields in this reaction and suggests that quenching of the carbon radical by ETCA is less efficient relative to quenching by Selectfluor. Notably, C–H bonds on adamantyl (2g, 3g) and cyclobutyl (2h, 3h) frameworks were successfully halogenated, as were substrates bearing azide (2t) and phenol (2v, 3v) functional groups. Moreover, sulfonamides derived from biologically relevant amines, such as Tuamine (2j, 3j), Forthane (2k, 3k), and Celecoxib (2u, 3u) were amenable to these protocols. While minor amounts of 1,6-isomers were formed for many of these substrates (see supporting information), in the case of substrate 2e, products derived from 1,5 and 1,6-HAT processes were obtained in a 1:1 mixture, presumably due to the weaker BDE of the benzylic C–H bond^53,54. While tertiary C–H bonds could be successfully fluorinated in high yields (2r, 2s), the analogous tertiary C–H chlorination products were found to be unstable and could not be isolated (olefins derived from elimination pathways were observed in crude NMR spectra of these reactions). Aside from tosyl, a variety of other sulfonamide substitution patterns were tolerated in the C–H fluorination reaction (2n, 2o, 2p, 2q), offering a general route to the fluorination of a variety of different sulfonamides. Indeed, the remote fluorination of electron-deficient sulfonamides such as the nosyl group (2q) has not been demonstrated previously. We were also pleased to discover that our fluorination protocol was scalable, obtaining a 61% yield of remotely fluorinated sulfonamide 2a whenever 1.09 g of the hydrazonyl carboxylic acid precursor 1a was subjected to the reaction conditions.

Figure 3.

Scope of the δ-C–H fluorination reaction. Yields reported are isolated yields of the monofluorinated products, which may be mixtures of regioisomers. In all cases, except 2e (δ:ε=1:1), the δ-fluorinated product was the major regioisomer (>90%). For some substrates, the formation of δ,δ-difluorinated product was also observed. Substrate 1k was diastereomerically enriched (69:31 d.r.). See experimental section for further details.

Figure 4.

Scope of the δ-C–H chlorination reaction. Yields reported are isolated yields. 3f was obtained as a 72:28 mixture of δ:ε-chlorinated products. Product 3d proved unstable to purification and spontaneously cyclized to the corresponding pyrrolidine 3dʹ. Substrate 1k was diastereomerically enriched (69:31 d.r.). See experimental section for further details.

Following exploration of the substrate scope, we then turned our attention to exploring the mechanisms of these transformations and the properties of this new radical precursor. Cyclic voltammogram measurements for the cesium salt of the hydrazonyl carboxylic acid precursor 1a led to the observation of an irreversible oxidation peak at E_p = 1.82 V (200 mVs⁻¹ in MeCN/H O vs. Ag⁺/Ag), which would be comparable to the excited state oxidation potentials for both photocatalysts (see supporting information). Fluorescence quenching experiments revealed that the cesium salt of the hydrazonyl carboxylic acid radical precursor strongly quenched the excited states of both photocatalysts, while minimal quenching was observed for the respective halogenating reagents, suggesting that a reductive quenching cycle was in effect (Figure 5A). Further measurements of both the fluorination and chlorination reactions established the quantum yields as 14.6% for the fluorination reaction and less than 1% for the chlorination reaction (Figure 5B), indicating that although the C–H fluorination reaction was photochemically much more efficient than the C–H chlorination reaction, there was inconclusive evidence for a radical chain mechanism⁵⁵.

Figure 5.

A Proposed catalytic cycle for remote C–H halogenation reactions on the basis of cyclic voltammetry and fluorescence quenching studies. For X = F, Y = 1-(chloromethyl)-1,4-diazabicyclo[2.2.2]octane-1,4-diium ditetrafluoroborate. For X = Cl, Y= ethyl dichloroacetyl. B Quantum yield measurements for the two reactions.

In conclusion, we have herein reported new protocols for the δ-C–H fluorination and chlorination of a variety of different sulfonamides through the development of a new precursor to nitrogen-centered radicals based on pyruvic acid. The transformations use readily available electrophilic halogenation reagents and proceed in up to 82% and 64% yield respectively. Preparation of the new precursor was scalable up to 15 mmol and our remote C–H fluorination protocol proved scalable to over 1 g. Given that methods for the generation of nitrogen radicals are relatively scarce⁵⁹, we expect that this newly disclosed hydrazonyl carboxylic acid precursor will facilitate the development of a diverse range of new transformations, both by the development of other remote C–H functionalization reactions, as well as through the other patterns of reactivity available to nitrogen-centered radicals⁶⁰, such as β-scission processes or addition across unsaturated systems for N–C bond formation.