Photocatalytic Generation of Aminium Radical Cations for C─N Bond Formation

Abstract

Aminium radical cations have been extensively studied as electrophilic aminating species that readily participate in C─N bond forming processes with alkenes and arenes. However, their utility in synthesis has been limited, as their generation required unstable, reactive starting materials and harsh reaction conditions. Visible-light photoredox catalysis has emerged as a platform for the mild production of aminium radical cations from either unfunctionalized or N-functionalized amines. This Perspective covers recent synthetic methods that rely on the photocatalytic generation of aminium radical cations for C─N bond formation, specifically in the context of alkene hydroamination, arene C─H bond amination, and the mesolytic bond cleavage of alkoxyamines.

Graphical Abstract

1. Introduction

The development of technologies for the synthesis of amines remains an area of great interest to industrial and academic chemists alike due to their ubiquity in pharmaceuticals, natural products, and other biologically active small molecules.(1,2) When retrosynthetically evaluating amine targets, most chemists will first consider approaches that rely on forming C─N bonds from a nucleophilic amine and an electrophilic carbon center. In fact, Ingold regarded ammonia as a canonical nucleophile in seminal publications introducing the concepts of nucleophilicity and electrophilicity.(3,4) This pattern of intuitive chemical reactivity is ingrained at the earliest stages of chemical education, with many introductory organic chemistry textbooks beginning discussions of nucleophilicity with reactions of amines.

Given this context, it is perhaps unsurprising that nucleophilic, polar amination procedures such as reductive amination,(5) nucleophilic substitution (S_N2 and S_NAr),(6,7) amide coupling,(8) and transition-metal-mediated cross-coupling reactions (Buchwald–Hartwig, Ullmann–Goldberg, Chan–Lam, etc.)(9-11) are the dominant methodologies for the construction of C─N bonds (Figure 1). Despite the utility of these transformations, the electron density that inherently makes amines potent nucleophiles renders them less effective as intermediates in other valuable reaction classes. For example, the addition of a N─H σ bond across a C─C π bond (olefin hydroamination) has been studied for decades. This process is kinetically challenging in the absence of a catalyst due to electronic repulsion between the nonbonding lone pair of the amine and the π electrons of the alkene. Stated alternatively, the hydroamination of electron-rich (nucleophilic) alkenes with (nucleophilic) amines represents the union of two centers of the same polarity. Seebach has extensively reviewed this type of synthetic challenge, outlining the principle of reactivity umpolung.(12) In the context of alkene hydroamination, this reactivity umpolung has most frequently been achieved by electrophilic activation of the alkene component upon coordination with an electron-deficient transition metal.(13-15) Coordination polarizes the alkene π bond and places the greatest partial positive charge on the most substituted carbon, which results in Markovnikov addition products in subsequent reactions with amine nucleophiles. While the electrophilic activation of olefins has seen success in numerous contexts, weakly coordinating alkenes are often unable to compete with strongly Lewis basic amines for metal coordination sites.

Figure 1.

Strategies for amine synthesis.

Alternatively, the polarity of an amine could be reversed in two conceptually similar but practically distinct manners (Figure 2). In the first, the amine is prefunctionalized with an electronegative leaving group. In a subsequent operation, a transition-metal catalyst(16) or organometallic intermediate(17) can engage with this electrophilic amine and an alkene in order to generate an oxidized amino-functionalized product or a formal hydroamination product when a terminal hydride reductant is included.(18) In the second strategy for umpolung amination, an amine undergoes single-electron oxidation to its corresponding radical cation. Aminium radical cations (ARCs) are electrophilic radical aminating species that undergo facile addition to nucleophilic alkenes and arenes. While this elementary step has been known and studied for decades, ARCs derived from unfunctionalized amine precursors have only recently found utility as reactive intermediates in catalytic amination reactions.

Figure 2.

Activation modes for catalytic alkene hydroamination.

Recently, photoredox catalysis has emerged as a platform for the mild generation of aminium radical cations. This Perspective will recount the mechanistic insights that led to the development of a suite of photocatalytic amination technologies, including alkene hydroamination, (hetero)arene C─H bond amination/pyridination, and the mesolytic bond cleavage of alkoxyamines. We also highlight the distinct advantages of photoredox catalysis and ARCs as electrophilic species in comparison to traditional closed-shell, nucleophilic amination methods.

2. Generation and Reactivity of Aminium Radical Cations

The first invocation of a nonaromatic aminium radical cation in the literature was in 1950 with Wawzonek and Thelen’s proposed mechanism for the Hofmann–Löffler–Freytag cyclization(19) in their preparation of N-methylgranatanine.(20) This reaction was initiated by UV-light-mediated homolysis of a weak N─Cl bond to generate aminyl and chlorine radicals. Neutral aminyl radicals are nucleophilic, open-shell intermediates that add to electron-deficient alkenes more readily than electron-rich alkenes.(21-23) Protonation of the aminyl radical results in a reactivity umpolung and converts it to an electrophilic aminium radical cation. This is the active species that carries out the key intramolecular 1,5-hydrogen atom transfer (HAT) in the Hofmann–Löffler–Freytag reaction.(24)

To expand the synthetic utility of ARCs beyond unimolecular contexts, careful control over the channels of reactivity available to it is required (Figure 3). The lifetime of the radical cation is chiefly controlled by the rate of α-deprotonation, since oxidation of the amine significantly acidifies proximal C─H bonds.(25) For example, the pK_a value of the α-C─H bond of Et₃N drops to approximately 14.7 (MeCN) upon single-electron oxidation, enabling favorable deprotonation by a second equivalent of Et₃N (whose conjugate acid’s pK_avalue is approximately 19 in MeCN).(26-29) In specific substrates, the rate of α-deprotonation can be diminished through structural modification of the amine architecture to instead promote bimolecular HAT reactivity; for example, deprotonation in bridged bicyclic systems is kinetically slow. This has made DABCO and quinuclidinium radical cations useful catalytic mediators of C─H bond abstraction via intermolecular HAT.(30)

Figure 3.

Aminium radical cation generation and reactivity.

Beyond HAT and α-deprotonation, ARCs have also been shown to undergo C─N bond formation with alkenes and arenes. It is expected that aminium radicals, bearing a positive charge, would react in an electrophilic manner. Consistent with this expectation, Chow studied the polar effects of piperidinium radical cation addition to substituted styrenes and found the rate of addition correlated with σ constants, with a negative ρ value of −1.34.(31) Piperidinium radicals exhibit a much greater dependence on electron affinity than other neutral radical additions (ρ = −0.7 for the addition of a trichloromethyl radical to styrene) but significantly less than the protonation of styrene (ρ = −3.42), wherein a significant positive charge density develops at the benzylic position in the transition state.(31) Chow hypothesized that the rate acceleration bestowed by electron-donating groups reflects a favorable interaction between the π electrons of the styrene and the cationic ARC.(31) This stabilization stands in contrast to unfavorable ground-state amine/alkene interactions that need to be overcome in nucleophilic hydroamination reactions, highlighting the distinct advantage of hydroamination with ARCs.

In order to produce the most stable carbon-centered radical, aminium radical cations characteristically add to the less substituted carbon of an alkene; this anti-Markovnikov addition stands in stark contrast to the pattern of addition observed with transition-metal catalysis (vide supra). Minisci in particular had pioneered the use of ARCs in this area, generating the requisite radical cations from thermal or photolytic homolysis of N-haloamines or hydroxylamine derivatives in the presence of acid additives to furnish overall 1,2-difunctionalized alkenes.(32,33) Minisci has utilized similar ARC precursors for the homolytic C─H amination of electron-rich arenes.(33-36)

N-Haloamines and hydroxylamine derivatives as utilized by Minisci are among the most common ARC precursors, but N-nitrosamines, N-nitroamines, and 2-tetrazenes have all been found to undergo bond homolysis.(25) Alternatively, aminium radical cations can also arise directly from the unfunctionalized parent amine via single-electron oxidation by a suitable chemical or electrochemical oxidant. This mode of generation is problematic because α-amino radicals, generated via α-deprotonation, are themselves easily further oxidized to the iminium ion.(37) Finally, electronic excited states of a photosensitized molecule can be quenched by an amine through an electron-transfer process, generating an ARC and a reduced acceptor molecule. Early studies on photosensitized electron transfer focused primarily on the triplet excited states of carbonyls and aromatic hydrocarbons.(38) This quenching process typically resulted in the net hydrogenation of the electron acceptor and oxidation of the amine donor to the corresponding iminium ion.

In the late 1970s, Meyer and Whitten demonstrated that, in addition to the photoexcited states of ketones and polycyclic aromatics, excited-state Ru polypyridyl complexes can also be quenched by amines.(39-41)The benefits Ru and Ir complexes, as photosensitizers include their strong absorption in the visible-light region and relatively long excited-state lifetimes. Quenching of the excited-state complex through electron transfer converts the energy of a visible-light photon into chemical energy, forming a charge-separated ARC and a reduced metal center. Pac in 1981 was among the first to utilize these intermediates for synthetic purposes, using 1,4-dihydropyridine (DHP) as the quencher for photoexcited [Ru(bpy)₃]Cl₂ ([Ru-1]Cl₂).(42)Rapid deprotonation of DHP generates a stabilized, highly reducing radical incapable of accepting an electron back from the reduced-state Ru(bpy)₃⁺, therefore enabling the reduced metal complex to instead transfer an electron to an electron-deficient olefin. Throughout the 1980s Pac expanded the scope of this catalytic system to include electron-deficient alkenes bearing esters, ketones, arenes, and nitriles.(43-46)Other notable contributions to this field include publications from Kellogg, Tanaka, and Fukuzumi, who all utilized DHPs or dihydroacridine (AcrH₂) as stoichiometric electron donors in the reductions of phenacyl sulfonium salts, benzyl halides, and phenacyl halides.(47-49) This flurry of early research on the reactions of photoexcited-state Ru complexes elucidated a number of key concepts that were critical considerations in the future development of photoredox processes.

Following the absorption of a photon, an electron from the metal-centered t_2g orbitals of Ru(bpy)₃²⁺ is excited to a ligand-centered π* orbital via metal to ligand charge transfer (MLCT) (Figure 4).(51) The initially generated singlet MLCT state undergoes rapid intersystem crossing (ISC) to the lowest energy triplet MLCT state, which has a long-lived excited state (1100 ns for Ru(bpy)₃²⁺).(52) The charge-separated complex simultaneously possesses a newly unoccupied hole in the metal t_2g orbital and a high-energy ligand-centered electron, rendering the excited-state complex simultaneously a more powerful oxidant and a more powerful reductant than the ground-state Ru(bpy)₃²⁺.(53) In a reductive quenching event, the photocatalyst accepts an electron from a substrate, yielding the substrate radical cation and reduced metal center. Through alternative oxidative quenching, the photocatalyst passes an electron to the substrate, forming a substrate radical anion and an oxidized metal center. The resulting charge-separated substrate radical ion and reduced/oxidized metal center are significantly higher in energy than their constituent starting materials, making them synthetically valuable reactive species in a wide variety of subsequent transformations.

Figure 4.

Photoredox properties of Ru(bpy)₃²⁺ ([Ru-1]²⁺). *In all discussions in this Perspective, we report redox potentials relative to the Fc⁺/Fc redox couple in MeCN and use the conversion constants as detailed by Addison where necessary to convert from originally reported data against other reference electrodes.(50)

In any photocatalytic scheme that utilizes the substrate radical ion, forward bimolecular chemistry must compete with nonproductive back electron transfer (BET), which returns both intermediates to their original electronic ground states (Figure 5A).(41,54) BET processes are generally kinetically facile and benefit from significant thermodynamic driving forces.(55) While the rate of BET can be modulated through careful manipulation of the ligand architecture of the catalyst,(40) early synthetic reports opted to circumvent BET by programming a system wherein a secondary reaction converts either the metal or the substrate radical ion to a redox-inactive intermediate incapable of participating in the electron transfer.(56,57) For reductive quenching cycles with amines as substrates, DHP and AcrH₂ were the stoichiometric reductants of choice in the earliest photoredox procedures because deprotonation in these systems is facile and is presumably fast enough to compete with BET (Figure 5B). Understanding the kinetic competition between BET and the forward reactions of the aminium radical cation proved to be a critical feature in later developments of uni- and bimolecular hydroamination processes.

Figure 5.

(A) Back electron transfer and (B) deprotonation of DHP^•+.

Following an initially fruitful period of research into photoredox catalysis, the field laid largely dormant for two decades. This changed in 2008 with contemporaneous publications by the Yoon and MacMillan groups, who both used Ru(bpy)₃²⁺ as a visible-light photoredox catalyst for intramolecular [2 + 2] cycloaddition reactions of enones and the α-alkylation of aldehydes with alkyl bromides, respectively.(58,59) Soon thereafter, Stephenson disclosed a photocatalytic dehalogenation of haloalkanes which employed Hünig’s base as a stoichiometric reductant,(60) reactivity later expanded by Gagné to the photoreduction of glycosyl halides.(61,62) These publications initiated a renaissance in synthetic photoredox catalysis, which continues to this day. Amines have since gained traction as a versatile class of substrates for photoredox catalysis, moving beyond their early uses as simple reductants. Nucleophilic α-amino radicals are useful intermediates for synthesis in their own right and have been utilized in a variety of C─C bond forming processes, engaging with a wide range of electrophiles.(63-66) Other groups have used stoichiometric chemical oxidants in concert with amines to generate iminium ions under mild reaction conditions, enabling reactions with various nucleophiles.(26,67) While these strategies have provided significant advances in the manipulation of amine architectures,(68) they are both limited to functionalization of a carbon atom adjacent to the nitrogen. Given the value of C─N bond forming reactions and the well-established electrophilic reactivity of aminium radical cations, photocatalytic procedures that directly engage the nitrogen of amines for C─N bond construction had remained conspicuously absent, until 2012, when the first of a series of reports utilizing aminium radical cations for C─N bond formation was reported.

3. Intramolecular Reactions of Arylaminium Radical Cations with Olefins for C─N Bond Formation

Indoles are an important class of heterocycle found in many bioactive natural products(69) and pharmaceutical substances.(70) As a result, numerous methods have been developed for their syntheses, exploiting many possible disconnections.(71,72) Indole synthesis through cyclization of an aniline onto a tethered olefin using oxidative Pd(II) catalysis was first demonstrated by Hegedus in 1976(73–76) and utilized in their total synthesis of the (±)-clavicipitic acids.(77) More recently, Buchwald has also reported such reactivity,(78) and both studies note the orthogonality of the Pd(II)-mediated Wacker-type process(79)over Pd(0)-mediated oxidative addition of substrate-bound halides. However, stoichiometric quantities of either benzoquinone or Cu(OAc)₂ terminal oxidants are required in these processes, which take place at elevated temperatures of 75–100 °C. These methods above undertake activation of the substrate olefin toward nucleophilic attack by the amine, through π coordination with a Pd(II) complex.

A 2012 report from Maity and Zheng offered an alternative, umpolung mode of activation for the synthesis of N-aryl indoles through the photocatalytic generation of electrophilic diarylaminium radical cations and addition to a tethered styrene, in a net-oxidative aerobic process.(80) In seeking to circumvent the typical reaction pathways of ARCs (vide supra) and focus on C─N bond formation, the authors studied diarylamines—systems lacking α-protons and thus incapable of α-amino radical or iminium ion formation through α-deprotonation and further oxidation, respectively.

In their method, white-light irradiation of an acetonitrile solution of styryl-substituted diarylamine substrates, [Ru(bpz)₃](PF₆)₂ photocatalyst ([Ru-2](PF₆)₂) (Figure 6), and silica gel in an air atmosphere led to rapid and efficient indole formation at ambient temperature (Schemes 1 and 2). The inclusion of silica gel was found to be essential for obtaining high product yields and was proposed to improve the solubility of molecular oxygen.(81) Control experiments revealed the requirement for both light and Ru photocatalyst, and irradiation of the substrate instead in the presence of tetraphenylporphyrin (TPP)—a widely used photosensitizer for the generation of singlet oxygen—gave no product.

Figure 6.

Photocatalysts employed in the generation of anilinium radical cations.

Scheme 1.

Intramolecular Cyclization of Diarylaminium Radical Cations onto Styrenes for Oxidative Indole Synthesis^a,(80)

^aPMP = 4-methoxyphenyl.

Scheme 2.

Intramolecular Cyclization of Diarylaminium Radical Cations onto Styrenes, with 1,2-Migration of a β-Substituent^a,(80)

^aPMP = 4-methoxyphenyl.

Two substrate classes emerged in this work: 15 examples of β-monosubstituted styrenes (e.g., 1–6), which underwent aromatization to the 2-substituted indole in isolated yields of 55–83% (Scheme 1), and 6 examples of β,β-disubstituted styrenes (e.g., 7–9), which first underwent a 1,2-alkyl or aryl shift, with substituents displaying the expected migratory aptitude,(82) prior to aromatization to the 2,3-disubstituted indoles in isolated yields of 40–62% (Scheme 2). It is noteworthy that a cyclopropane substituent in the α-styryl position remained intact throughout this sequence (5), indicating that the oxidation of the radical intermediate following C─N bond formation outcompetes the potential cyclopropylbenzyl radical ring opening (the rate of ring opening of the cyclopropylbenzyl radical is independently measured as k = 2.7 × 10⁵/s).(83) Specifically, N-(4-alkoxyphenyl)-N-arylamines were found to be essential for the desired transformation and replacement of the alkoxyphenyl group with a simple phenyl moiety gave no product. Through CV experiments, the authors observe that this alkoxy substituent facilitates substrate oxidation by lowering the potential requirement by approximately 150 mV.

The proposed mechanism of this transformation (Scheme 3) proceeds via single-electron oxidation of the diarylamine substrate mediated by photoexcited [Ru-2](PF₆)₂ (E_1/2(Ru(II)*/Ru(I)) = +1.07 V vs Fc⁺/Fc in MeCN)(84,85) to efficiently generate the diarylaminium radical cation (e.g., for the diarylamine substrate leading to indole 1, E_p/2 = +0.37 V vs Fc⁺/Fc in MeCN). This then engages with the substrate styrene in a 5-endo-trig mode of cyclization, to form a distonic aminium radical cation—defined as a species with spatially separated charge and radical sites on the same molecule.(86) The authors propose that, under the aerobic conditions of the reaction, the reduced state of the photocatalyst (E_1/2(Ru(II)/Ru(I)) = −1.18 V vs Fc⁺/Fc in MeCN)(84,85) reacts with molecular oxygen to re-form the ground-state Ru(II) complex. Superoxide formed through this catalyst turnover step subsequently mediates deprotonation of the distonic aminium radical cation, before further oxidation of the intermediate benzylic radical to a carbocation. The second oxidation step could plausibly be mediated by any one of the photoexcited state Ru(II) catalyst, molecular oxygen, or superoxide, all of which have the requisite potential to drive oxidation of the benzylic product radical (e.g., the oxidation potential of a benzylic secondary radical was independently measured as E_p/2 = −0.01 V vs Fc⁺/Fc in MeCN and that of a benzylic tertiary radial as E_p/2 = −0.22 V vs Fc⁺/Fc in MeCN).(87) The observed 1,2-migration in β,β-disubstituted styrenes (Scheme 2) indicates carbocation formation, as opposed to a mechanism for aromatization involving trapping of the intermediate radical with dioxygen or hydroperoxyl radical, followed by base-mediated elimination.

Scheme 3.

Proposed Mechanism of Photocatalytic Oxidative Indole Synthesis Through the Generation of Diarylaminium Radical Cations^a,(80)

^aPMP = 4-methoxyphenyl.

This seminal work demonstrated the synthetic feasibility of C─N bond formation through aminium radical cation addition to olefins, within a system lacking α-protons and thus unable to undergo often-observed side reactions of these intermediates. This is also an early example of a net oxidative photoredox reaction driven solely by air as the terminal oxidant and highlights the benefits offered through the umpolung approach to substrate activation offered through photocatalysis, in comparison to oxidative Pd(II) methods.

In seeking to extend the utility of arylaminium radical cations generated through photocatalytic substrate activation for C─N bond formation, in 2014 we reported a method for the redox-neutral intramolecular anti-Markovnikov hydroamination of styrenes with aryl alkylamines, for the synthesis of N-aryl pyrrolidines (Scheme 4).(88) These valuable saturated heterocycles are common in drug development, with a 2014 analysis finding 37 US FDA approved small-molecule pharmaceuticals bearing this motif.(89) In this substrate class, α-amino radical and iminium ion formation are possible, in addition to back electron transfer; thus, C─N bond formation would need to compete with these pathways to enable a successful reaction outcome.

Scheme 4.

Intramolecular Anti-Markovnikov Hydroamination of Styrenes via the Photocatalytic Generation of N-Alkyl-N-arylaminium Radical Cations(88)

We hypothesized that single-electron oxidation of the substrate amine (e.g., for the substrate leading to pyrrolidine 10, E_p/2 = +0.43 V vs Fc⁺/Fc in MeCN)(88) mediated by the photoexcited state of the well-matched Ru photocatalyst [Ru(bpy)₃]Cl₂ ([Ru-1]Cl₂) (E_1/2(Ru(II)*/Ru(I)) = +0.39 V vs Fc⁺/Fc in MeCN)(90)would lead to aminium radical cation generation. Our desire was that an irreversible 5-exo-trig cyclization of the aminium radical cation onto the tethered styrene would then lead to efficient C─N bond formation, before reduction of the distonic radical cation mediated by the reduced-state Ru(I) complex (E_1/2(Ru(II)/Ru(I)) = −1.71 V vs Fc⁺/Fc in MeCN),(90) and proton transfer to the resultant carbanion from the solvent would yield the closed-shell pyrrolidine product. A typical neutral secondary benzylic radical has a reduction potential of E_p/2 = −1.80 V vs Fc⁺/Fc in MeCN(87)—this likely represents an upper bound due to the stabilizing nature of zwitterion formation upon reduction of the distonic radical cation.

In proof of concept experiments, white-light irradiation of a methanol solution of a model styrene-tethered alkylarylamine substrate and [Ru(bpy)₃]Cl₂ ([Ru-1]Cl₂) gave a 43% isolated yield of the desired N-arylpyrrolidine (10). In addition to the cyclization product, small amounts of a homocoupling product resulting from the dimerization of the benzylic radical prior to reduction and catalyst turnover were isolated. The observation of this radical–radical coupling side-product suggested that the planned reduction of the distonic radical cation following cyclization—via ET with the Ru(I) state of the photocatalyst—was slow. Radical–radical coupling would lead to an accumulation of the catalytically inactive reduced-state Ru(I) complex, which we believed was the cause of the poor reaction efficiency.

To improve the situation, we reasoned that a more reducing photocatalyst would increase the rate of electron transfer to the benzylic radical and thereby prevent radical dimerization and catalyst inactivation. Switching to the heteroleptic Ir(III) photocatalyst [Ir(ppy)₂(dtbbpy)]PF₆ ([Ir-1]PF₆) (Figure 6) and use of a blue LED light source under otherwise identical reaction conditions gave an 88% isolated yield of pyrrolidine 10and no dimerization side products. [Ir-1]PF₆ is still sufficiently oxidizing in its photoexcited state (E_1/2(Ir(III)*/Ir(II)) = +0.28 V vs Fc⁺/Fc in MeCN)(91) to generate the initial aminium radical cation, but its reduced state is approximately 180 mV more reducing that that of [Ru-1]Cl₂ (E_1/2(Ir(III)/Ir(II)) = −1.89 V vs Fc⁺/Fc in MeCN),(91) leading to more rapid reduction of the distonic radical cation to achieve catalyst turnover. Control experiments reveal the requirement for both light and photocatalyst for any reactivity, and the use of aprotic solvents led to a reduced yield.

In contrast to the above Zheng report in which an electron-rich N-aryl group was required for aminium radical cation generation, reactivity here was optimal when the N-aryl moiety carried electronically neutral (77–95% isolated yields) or electron-poor (73–91% isolated yields) substituents. When an electron-rich N-(4-anisyl) group was introduced, a less-efficient reaction resulted (54% isolated yield). This likely reflects reduced electrophilicity of the radical cation intermediate leading to a diminished reactivity for C─N bond formation. A range of aryl substituents and heteroarenes (e.g., 12) on the styrene component were tolerated with little effect on the reaction efficiency. In addition to 25 examples of 5-exo-trig cyclization for the formation of pyrrolidines in isolated yields of 54–95% (e.g., 10–13), six examples of 6-exo-trig cyclization to form piperidines, morpholines (14), and piperazines (15) were reported in isolated yields of 44–88%. This is notable, in that C─N bond formation still outcompetes the additional possible side reaction of 1,5-HAT from the activated allylic C─H bond to the transient aminium radical cation.(19,92,93) Attempts to use alkyl di- or trisubstituted olefins in the cyclization reaction failed, due to the inability of the reduced-state Ir(II) photocatalyst to directly reduce the alkyl radical following C─N bond formation (the reduction potential of the tert-butyl radical is independently measured as E_p/2 = −2.0 V vs Fc⁺/Fc in MeCN(87)—approximately 110 mV beyond the capability of this Ir(III) dye).

Luminescence quenching experiments demonstrated that electron transfer from the aniline to the photoexcited-state Ir(III) complex was kinetically facile (K_sv = 1200 M⁻¹). A Hammett analysis of the rates of hydroamination in a series of para-substituted styrenes gave a linear relationship with σ_ρ and a modestly negative ρ value of −0.56, indicating that C─N bond formation between the electrophilic aminium radical cation and the nucleophilic olefin was rate-limiting.(31) If reduction of the benzylic radical was rate-limiting, we would have expected a positive ρ value due to the measured trend in reduction potentials.(94) Notably, neutral aminyl radicals exhibit positive ρ values in their addition reactions to para-substituted styrenes, thus providing further evidence that the aminium radical remains protonated during C─N bond formation.(22)Conducting the reaction in CD₃OH resulted in no deuterium incorporation at the benzylic position, but CH₃OD instead resulted in complete deuteration. This is consistent with a termination pathway involving stepwise electron transfer (ET) and then proton transfer (PT) to the benzylic radical, as opposed to a bimolecular HAT pathway from solvent.(95)

These two reports demonstrated the kinetic competence of diaryl- and arylalkylaminium radical cations for intramolecular C─N bond formation with styrene acceptor olefins, overcoming or outcompeting deleterious side processes such as back electron transfer, α-deprotonation, iminium ion formation, and 1,5-HAT. They showed that valuable indole and pyrrolidine heterocycles can be efficiently prepared at ambient temperatures and in the former case can replace stoichiometric chemical oxidants with air. Further work is now aimed at going beyond the cyclization of arylamines onto styrenes and investigating the possibility of intermolecular couplings of alkyl amines and unactivated olefins.

4. Intermolecular Anti-Markovnikov Hydroamination of Unactivated Alkenes with Alkylamines

While indoles and pyrrolidines are both abundant scaffolds in bioactive small molecules, we hoped to expand the scope of our hydroamination procedure beyond simple cyclizations. Intermolecular hydroamination, however, presents a unique set of challenges for catalysis, specifically in regard to the regioselectivity and thermodynamics of olefin addition.

With respect to regioselectivity, an amine can add to a substituted alkene at either the more substituted carbon (Markovnikov selectivity) or the less substituted carbon (anti-Markovnikov selectivity). While transition-metal-catalyzed methods resulting in Markovnikov addition are common (vide supra), anti-Markovnikov hydroamination typically requires either activated alkenes (1,3-dienes,(96,97) styrenes,(98,99)or allenes(100)) or directing groups (allylic amines(101) or alcohols(102,103)). Collectively, these important advances highlight the dearth of methods for the intermolecular, anti-Markovnikov hydroamination of unactivated alkenes with alkylamines. Given the known propensity of aminium radical cations to undergo anti-Markovnikov addition to alkenes, we hoped that photocatalytic hydroamination could address this longstanding gap in the literature.

As outlined previously, there are also thermodynamic challenges associated with hydroamination; in addition to the significant activation barrier associated with the ground-state addition of an amine to an alkene, bimolecular hydroaminations with certain amine classes often lack a significant thermodynamic driving force. Hartwig and co-workers experimentally determined the thermodynamics of vinylarene hydroamination with anilines and found that, while modestly favorable enthalpically, the entropic penalty levied by the union of two independent starting materials into a single product rendered the overall process thermoneutral or endergonic for some substrates.(104) To evaluate the thermodynamics of nonconjugated alkene hydroamination with secondary alkylamines, our group calculated the addition of 16 to trisubstituted olefin 17 to be nearly thermoneutral at room temperature (ΔG° = −0.1 kcal/mol, CBS-QB3). The analogous hydroamination of tetrasubstituted alkene 18, however, is appreciably endergonic (ΔG° = +4.9 kcal/mol, CBS-QB3) (Scheme 5).(105)

Scheme 5.

Thermochemistry of Selected Hydroamination Reactions with Dialkylamines*(105)

It is not possible for a ground-state catalyst alone to drive a hydroamination reaction to completion when the desired product is energetically uphill relative to the constituent amine and alkene starting materials (Figure 7A). To address this challenge, one approach that groups have taken is reagent-driven, wherein the −H and −NR₂ components of a net hydroamination reaction are separated into two distinct, high-energy reactants (Figure 7B).(106-108) This strategy couples a conventional hydroamination reaction to the net reduction of an electrophilic aminating reagent (most frequently hydroxylamine esters or benzisoxazole) by a silane, which builds in the necessary driving force for an otherwise thermoneutral process.(109) Though these reagent-driven methods have significantly advanced the state of the art in bimolecular alkene hydroamination, they generate stoichiometric quantities of byproducts and explicitly preclude the direct use of simple, abundant amines. Alternatively, the energy of visible-light photons can provide the necessary driving force to enable chemistry that would be unfavorable in the ground state, in a manner similar to photosynthesis and solar fuel generation (Figure 7C).

Figure 7.

(A) conventional, (B) reagent-driven, and (C) light-driven approaches to olefin hydroamination.

Our group has utilized the last strategy to develop two separate intermolecular, anti-Markovnikov hydroamination reactions of unactivated alkenes. In the first reaction, secondary dialkylamines are used as the substrates, wherein a number of tertiary amine products are obtained in high yield despite being higher in energy than the constituent reaction components. The second reaction utilizes primary alkylamines as the substrate to generate secondary amine products, which notably do not undergo further hydroamination to the overalkylated tertiary amine. In both reactions, the linear anti-Markovnikov products are exclusively observed. The development of these procedures is described herein.

4.1. Intermolecular Anti-Markovnikov Hydroamination with Secondary Dialkylamines

To interrogate the possibility of an intermolecular procedure,(105) we began with our previously optimized reaction conditions for intramolecular aniline hydroamination (vide supra), but with separate N-methylaniline and styrene as the two reacting partners. Using blue-light irradiation for prolonged reaction times, we did not observe any conversion to product. We hypothesized that C─N bond formation may have been the problematic step, since we had previously established this as the turnover-limiting step in the intramolecular regime. The anilinium radical cation generated upon reaction with the Ir(III) photooxidant can add to styrene or accept an electron from the reduced-state Ir(II) complex; to promote productive bimolecular reactivity, it would be necessary to either slow down BET or speed up C─N bond formation (vide supra). Lusztyk has shown that the N-methylanilinium radical cation adds to 1,1-diphenylethylene with rate constants of less than 10⁶ M⁻¹ s⁻¹, which we inferred must not be fast enough to compete with BET. In the same work it was noted, however, that piperidinium radical cation adds to the same olefin with a rate constant 3 orders of magnitude greater (k = 1.1 × 10⁹ M⁻¹ s⁻¹) than that of the aniline.(110) We hypothesized that the faster kinetics of C─N bond formation with dialkylamines might be sufficient to outpace BET and facilitate successful bimolecular amination reactivity.

The first challenge that we faced in developing an intermolecular hydroamination was identifying a photocatalyst capable of mediating all of the requisite electron-transfer steps. While [Ir(ppy)₂(dtbbpy)]PF₆([Ir-1]PF₆) (E_1/2(Ir(III)*/Ir(II)) = +0.28 V vs Fc⁺/Fc in MeCN)(91) (Figure 6) was sufficiently oxidizing in its photoexcited state to convert anilines to their corresponding radical cations, it is unable to oxidize dialkylamines (e.g., for piperidine, E_p/2 = +0.56 V vs Fc⁺/Fc in MeCN).(111) This necessitated a switch to a more-strongly oxidizing photocatalyst; to satisfy this requirement, we turned to [Ir(dF(Me)ppy)₂(dtbbpy)]PF₆([Ir-2]PF₆) (E_1/2(Ir(III)*/Ir(II)) = +0.59 V vs Fc⁺/Fc in MeCN)(112) (Figure 8) as a potential catalyst. This photoexcited-state complex is able to oxidize a dialkylamine to its radical cation, but now after alkene addition it cannot reduce the resultant carbon-centered radical to its corresponding carbanion. In the case of addition to a styrene, a photocatalyst must have a total redox window of approximately 2.4 V to mediate both the oxidation to the ARC and the reduction of the benzylic radical and close a redox-neutral catalytic cycle. When terminal olefins are used as substrates, the secondary radicals produced are reduced at potentials approaching E_p/2 = −2.4 V vs Fc⁺/Fc in MeCN,(113) bringing the total redox window requirement to approximately 3 V. [Ir-2]PF₆ emits at λ_max = 512 nm, which means that its maximum thermodynamically favorable redox window is 2.42 V and nonconjugated alkenes were thus untenable as substrates.

Figure 8.

Ir photocatalysts employed in the generation of alkylaminium radical cations.

Faced with this challenge, we evaluated whether we could redesign the hydroamination reaction to decrease the redox potential requirements that any single photocatalyst needed to satisfy. Nicewicz, Studer, and our own group have all used thiophenols as catalysts for terminal HAT events in photocatalytic alkene hydrofunctionalization.(114-116) Thiophenols possess weak S─H bonds (approximately 79 kcal/mol)(117)and undergo rapid HAT to carbon-centered radicals (k ≈ 10⁸ M⁻¹ s⁻¹), to generate a C─H bond and a thiyl radical.(118,119) Critically, the reduction potential of this thiyl radical (E_1/2 = +0.07 V vs Fc⁺/Fc in MeCN)(120) is dramatically less than that of the carbon-centered radical and is well within the range of the photocatalyst. An additional benefit of including the thiol cocatalyst is that it can return any off-cycle intermediates arising from α-deprotonation or HAT back to their ground-state, closed-shell species through the same HAT/ET/PT cycle.

In accordance with these principles, we found that, under irradiation with blue LEDs, [Ir(dF(Me)ppy)₂(dtbbpy)]PF₆ ([Ir-2]PF₆) and 2,4,6-triisopropylbenzenethiol (TRIP thiol) were jointly able to mediate the hydroamination of unactivated alkenes with secondary dialkylamines (Scheme 6).(105) Beyond the intramolecular hydroamination of secondary alkylamines for the synthesis of N-alkylpyrrolidines and piperidines (7 examples, 65–90% yield), this procedure successfully worked for the intermolecular hydroamination of secondary alkylamines, providing acyclic 3° alkylamines (43 examples, 37–98% yield). Mono-, di-, tri-, and tetrasubstituted alkenes all were efficient substrates in the hydroamination procedure and provided the desired 3° amine in good to excellent yields (e.g., 21–24). In particular, 24 (calculated to be +4.8 kcal/mol uphill relative to its amine and alkene components and thereby inaccessible with ground-state catalysis) demonstrated the ability of light-driven photoredox catalysis to generate products that are less stable than their constituent starting materials. Our laboratory has since expanded this concept to light-driven isomerization, deracemization, and lignin depolymerization.(121-123)

Scheme 6.

Intermolecular Anti-Markovnikov Hydroamination of Unactivated Alkenes with Secondary Alkylamines(105)

A variety of cyclic amines could be employed in the reaction (e.g., 21–26), including a substrate possessing both a primary and secondary amine (26). The chemoselectivity in this substrate exclusively favored alkylation at the secondary amine, which we attributed to selective oxidation of the secondary amine. While secondary amines could undergo thermodynamically favorable ET with the excited state of the photocatalyst, oxidation of the primary amine was prohibitively endergonic (e.g., for isopropylamine, E_p/2 = +1.16 V vs Fc⁺/Fc in MeCN).(124) Acyclic amines were competent coupling partners, albeit with diminished efficiency in comparison to piperidine (e.g., 27, 65% yield).

Mechanistically, the reaction was proposed to initiate via single-electron transfer between the photoexcited state of [Ir(dF(Me)ppy)₂(dtbbpy)]PF₆ ([Ir-2]PF₆) (E_1/2(Ir(III)*/Ir(II)) = +0.59 V vs Fc⁺/Fc in MeCN)(112) and the secondary amine. The feasibility of this elementary step was supported by Stern–Volmer quenching studies, which demonstrated that piperidine efficiently quenches the excited state of [Ir-2]PF₆ (K_SV = 200 M⁻¹). The secondary dialkylaminium radical cation generated is able to undergo anti-Markovnikov addition to an alkene, generating a distonic radical cation. The carbon-centered radical is quenched by TRIP thiol, producing a closed-shell ammonium ion and thiyl radical. Reduction of the thiyl radical to thiolate mediated by the reduced-state Ir(II) complex and subsequent deprotonation of the ammonium ion furnish the desired hydroamination product and return all catalytic components to their respective ground states.

To probe a potential competing mechanistic pathway of alkene oxidation, we conducted luminescence quenching studies with a variety of alkenes: 1-hexene, cyclohexene, 2-methylhex-1-ene, tetramethylethylene, and dihydrofuran. In all cases, increasing the concentration of the alkene does not decrease the luminescence intensity of [Ir-2]PF₆, providing evidence that this hydroamination procedure does not proceed through an alkene radical cation intermediate and is therefore mechanistically complementary to Nicewicz’s hydrofunctionalization work.(115,125)

The tertiary amine products generated in the reaction can also quench [Ir-2]PF₆ (K_SV = 180 M⁻¹ for 22), potentially leading to deleterious side pathways. The generally high yields suggest that product oxidation does not lead to significant quantities of product decomposition. We hypothesized that the tertiary ARC generated can undergo two processes: either BET or α-deprotonation to the α-amino radical. The former case returns the ARC to the closed-shell amine product, and the latter intermediate can undergo HAT with the thiol cocatalyst to once again regenerate product.

4.2. Selective Monoalkylation of Primary Alkylamines with Unactivated Alkenes

Following the successful development of an alkene hydroamination protocol with secondary dialkylamines, we considered whether we could extend this reactivity to include primary alkylamines as substrates. As with secondary amines, intermolecular hydroamination procedures with primary amines are rare. Marks(126-128) and Hultzsch(129-131) have each reported a number of elegant hydroamination reactions using organolanthanide complexes as catalysts and primary amines as substrates, but for unactivated alkenes only Markovnikov selectivity has been achieved.

Oe, and Xiao and Wang have respectively disclosed Ru- and Fe-catalyzed anti-Markovnikov formal hydroaminations of allylic alcohols that proceed via a borrowing-hydrogen sequence of dehydrogenation/1,4-conjugate addition/hydrogenation.(102,103) In both cases the authors reported an extensive substrate scope with secondary amines, but with few examples of primary amines. Oe’s substrate scope only included a single example of a primary amine substrate that proceeded in low yield; the authors hypothesized that the lower reactivity of this substrate was due to the lower nucleophilicity of primary amines in comparison to secondary cyclic amines.(102) Xiao and Wang’s system displayed improved reactivity with primary amines, but in the presence of 4 equiv of the allylic alcohol they exclusively isolated the overalkylated tertiary amine product rather than the secondary amine.(103) Controlling between mono- and dialkylation of primary amines is a pervasive challenge for electrophilic alkylation reactions more broadly, as primary and secondary amines often react with electrophiles at comparable rates;(132) therefore, mixtures of secondary and tertiary amines are often observed.(133,134)

Undesired overalkylation of the secondary acyclic amine products generated in a photocatalytic hydroamination of primary amines was also a concern. Secondary amines are oxidized at lower potentials than primary amines (vide supra); thus, any photocatalyst capable of generating an ARC from a primary amine substrate would also be expected to oxidize the secondary amine product. While the yields were lower than those with cyclic amines, secondary acyclic amines (potential products of a primary hydroamination process) are precedented substrates for photocatalytic hydroamination, as demonstrated by the hydroamination of 2-methyl-1-hexene (28) with diethylamine (16) (Scheme 7). With [Ir-2]PF₆ as the photocatalyst, 65% of the tertiary amine product was obtained (Scheme 7A); however, if [Ir(dF(CF₃)ppy)₂(4,4′-d(CF₃)-bpy)]PF₆ ([Ir-3]PF₆) (Figure 8) was used, then no conversion was observed (Scheme 7B). This was surprising, since [Ir-3]PF₆ is a powerful oxidant in its photoexcited state (E_1/2(Ir(III)*/Ir(II)) = +1.27 V vs Fc⁺/Fc in MeCN)(135)—thus, on thermodynamic grounds, both catalysts were expected to undergo favorable electron transfer with 16 to generate a diethylaminium radical cation. Interestingly, when piperidine was instead used as the substrate with [Ir-3]PF₆ as the photocatalyst, some degree of reactivity was restored and 18% of the hydroamination product was obtained (Scheme 7C).

Scheme 7.

Divergent Reactivity with Photocatalysts [Ir-2]PF₆ and [Ir-3]PF₆

Lusztyk measured the rate constants for piperidium and diethylaminium radical cation addition to 2-methyl-1-butene, respectively, and found that the former (k = 1.8 × 10⁸ M⁻¹ s⁻¹) reacted nearly a full order of magnitude faster than the latter (k = 2.6 × 10⁷ M⁻¹ s⁻¹).(110) We hypothesized that the divergent outcomes for cyclic and acyclic amines may arise from differences in the kinetic competition between BET and C─N bond formation for the respective photocatalysts and ARCs. In this proposal, we speculate that because a product was observed with diethylamine and [Ir-2]PF₆, it can be presumed that the rate of C─N bond formation can kinetically compete with BET. When [Ir-3]PF₆ was instead used as the photocatalyst, the rate of C─N bond formation between diethylaminium radical cation and 28 was dominated by BET and therefore no product was formed. However, switching to an amine whose radical cation reacts more quickly with the olefin allows C─N bond formation to again, to some extent, compete with the rate of BET.

Considering these ideas more broadly, we reasoned that this kinetic framework may enable the development of a method for hydroamination with primary amines that avoids overalkylation. On the basis of its excited-state potential, it was expected that [Ir-3]PF₆ would be able to oxidize a primary amine to the corresponding radical cation (e.g., for isopropylamine, E_p/2 = +1.16 V vs Fc⁺/Fc in MeCN).(124) While it was unclear whether the primary ARC would undergo productive C─N bond formation, we inferred from the diethylamine results above that any acyclic secondary dialkylamine product that did form should be inert to further hydroamination.

With some modifications to the reaction conditions from the secondary amine hydroamination procedure, we found that [Ir(dF(CF₃)ppy)₂(4,4′-d(CF₃)-bpy)]PF₆ ([Ir-3]PF₆) and TRIP thiol could mediate the anti-Markovnikov hydroamination of a wide range of unactivated alkenes with cyclohexylamine (Scheme 8).(136)Notably, excellent selectivity for the secondary amine product was observed in all cases (vide infra). Mono-, di-, and trisubstituted alkenes all underwent hydroamination in moderate to good yields (e.g., 30–32). Silyl enol ethers, a particularly nucleophilic class of alkenes, were excellent substrates in the optimized procedure, providing 1,2-amino alcohol products following an acidic workup (e.g., 33, 80% yield). For the amine scope, it was found that primary amines bearing primary, secondary, and tertiary branching in the α-position underwent efficient hydroamination with methylenecyclopentane (e.g., 32, 34, 35).

Scheme 8.

Intermolecular, Anti-Markovnikov Hydroamination of Unactivated Alkenes with Primary Amines(136)

Mechanistically, the reaction was proposed to initiate via electron transfer between the photoexcited state of [Ir-3]PF₆ and cyclohexylamine to furnish a highly reactive primary ARC. Stern–Volmer fluorescence quenching studies again demonstrated that this step is kinetically feasible, with K_SV = 45 M⁻¹.(136) The following steps mirror those previously described for the secondary amine hydroamination: anti-Markovnikov alkene addition by the ARC, HAT to the resultant C-centered radical mediated by thiol, reduction of the thiyl radical by the Ir(II) state of the photocatalyst, and deprotonation of the closed-shell ammonium ion to furnish the desired secondary amine product and regenerate the thiol catalyst.

Critically, the product is largely inert to further hydroamination despite the excess alkene utilized in the reaction. Excellent selectivities were observed for the desired products vs overalkylated tertiary amines (≥20:1 for all cases, 38:1 for 31), which compares favorably to traditional polar alkylation methods, wherein the ratio of secondary amine to tertiary amine is frequently ≤10:1.(133,134) This is despite the fact that the secondary amine product also quenches the excited state of the photocatalyst even more efficiently than the primary amine starting materials (K_SV = 941 M⁻¹ for 31).(136)

We are confident that the unique regioselectivity of this intermolecular hydroamination in tandem with the excellent selectivities observed for monoalkylation vs dialkylation will make this a valuable method for chemists seeking to synthesize complex secondary amines.

5. Intermolecular C─H Amination of (Hetero)arenes with Alkylaminium Radical Cations Generated through Photocatalysis

Transition-metal catalysis has been used extensively for the construction of arylamines through cross-coupling of prefunctionalized arenes and amines; for example, Pd catalysis for the Buchwald–Hartwig amination of (pseudo)haloarenes,(137,138) Cu catalysis for the Ullmann–Goldberg coupling of (pseudo)haloarenes with nitrogen nucleophiles,(10) and Cu catalysis for the Chan-Lam oxidative amination of aryl boronic acids.(139) The first two processes exploit the innate polarity of the coupling partners, with prefunctionalization of the arene component required to polarize the substrate toward oxidative addition and to provide site selectivity. The last process instead achieves the cross-coupling of two nucleophilic partners through inclusion of a stoichiometric oxidant—often Cu(II) or O₂—but still leverages the innate nucleophilic reactivity of the amine and requires preactivation of the arene partner for site selectivity and reactivity.

An alternative approach seeks to functionalize unactivated arenes through C─H amination technologies. This manifold offers distinct advantages in step economy and process sustainability and has been developed extensively in recent years.(140,141) Many of these recent methods utilize transition-metal catalysts to promote a rate-limiting C─H bond cleavage step, via a concerted metalation–deprotonation (CMD) mechanism and the formation of an organometallic intermediate.(142) Coordination of an amine coupling partner to the metal center, followed by reductive elimination, yields a new C─N bond and reduced transition-metal complex. Inclusion of a compatible oxidant to return the metal to its higher oxidation state allows for catalyst turnover. Often, a directing group is employed to govern site selectivity and to reduce reaction molecularity through the chelation of the metal complex in the proximity of one specific C─H bond.(143-145)

In this section, we discuss a collection of complementary methods enabling the photocatalytic undirected C─H amination of arenes and heteroarenes through the intermediacy of electrophilic aminium radical cations. These are mechanistically distinct from typical transition-metal-catalyzed methods in that C─H bond cleavage is no longer rate limiting, opening up distinctly different synthetic opportunities. As highlighted above, Minisci first demonstrated that primary and secondary aminium radical cations readily add to arenes when chloramines undergo either thermal or metal-catalyzed homolysis in acidic solvents.(146) We briefly note here several modern, related approaches to arene C─H amination via ARC generation with ground-state redox mediators; however, these are currently limited to the introduction of ammonia(147-150) or piperazine(151) through the use of electrophilic aminating reagents such as activated hydroxylamines and Selectfluor,(152) respectively.

Examples of photocatalytic methods for the generation of alkylaminium radical cations for arene C─H amination can be broken down into two categories. (i) The first is an example requiring no preactivation of the amine, which runs under net oxidative photocatalytic conditions and is initiated via single-electron oxidation of a substrate amine mediated by a photocatalyst. To our knowledge, there currently exists a single report of this mode of reactivity from the Nicewicz group. (ii) The second is examples requiring the preactivation of the amine component but are redox neutral with respect to the photocatalytic redox cycle and are initiated via single-electron reduction of this activated substrate mediated by a photocatalyst, of which there are two recent reports from the Leonori group.

5.1. Oxidative Generation of Alkylaminium Radical Cations for Arene C─H Amination with Primary Amines

In 2017, Nicewicz and co-workers disclosed a strategy for functionalized aniline synthesis via the net oxidative, direct C─H amination of arenes with alkylaminium radical cations generated through photocatalytic activation of primary amines. In the presence of the acridinium photocatalyst [Acr-1]BF₄ or [Acr-2]BF₄(153,154) (Figure 9) and under an oxygen atmosphere, blue-light irradiation of dichloroethane solutions of a series of 27 amines and 28 arenes gave N-alkylaniline C─H amination products in 32–95% isolated yields (Scheme 9). Optimal performance was realized with acridinium photocatalysts bearing 2,7-dimethyl ([Acr-1]BF₄) or 3,6-di-tert-butyl ([Acr-2]BF₄) substituents, which offered greater stability toward nucleophiles and radicals in comparison to an unsubstituted parent acridinium photocatalyst.(155) In some cases, the addition of a TEMPO cocatalyst was found to improve yields of the C─H amination products. Fourteen examples of α-amino ester N-arylation were included, directly utilizing the amino ester hydrochloride salt as the substrate and running the reaction in a biphasic solvent system consisting of 1,2-DCE and pH 8 aqueous phosphate buffer solution. Remarkably, enantiopure α-amino ester and α-methylbenzylamine substrates underwent the desired N-arylation reaction without racemization, indicating that C─N bond formation is sufficiently rapid to compete with α-deprotonation of the aminium radical cation intermediate.

Figure 9.

Acridinium photocatalysts employed in the generation of primary alkylaminium radical cations.

Scheme 9.

Photocatalytic Arene and Heteroarene C─H Amination with Primary Alkylamines^a,(156)

^a*Indicates minor site of amination. †2:2:1 DCE:Arene:pH 8 buffer. ‡40% TEMPO additive.

Electron-neutral and electron-rich arenes were effective coupling partners under conditions with arene as the limiting reagent. Halogen functional groups were tolerated (e.g., 39 and 42), allowing for possible sequential C─H amination and transition-metal-catalyzed cross-coupling. Late-stage amination of complex pharmaceutical substrates was demonstrated (for example clofibrate, 42), and three examples of pyridine (e.g., 40) and indazole (e.g., 41) C─H amination were also presented. Through modification of the reaction conditions to include an arene as the cosolvent, valine methyl ester underwent N-arylation with both benzene and toluene, yielding 43 and 44 in 40% and 50% yields, respectively. The ortho/para regioselectivity of C─H amination of monosubstituted arenes was typically modest, but para selectivity could be controlled to some extent through steric shielding of the ortho positions in silylated phenol substrates, overriding the innate regioselectivity observed in anisole (for example, 37 compared to 36). The regioselectivity was often higher in disubstituted arenes (e.g., 39). Exclusive mono-N-arylation of primary amines was observed in all cases, and no mention of diaminated arene products was given. The method unfortunately did not extend to the C─H amination of arenes with secondary alkylamines.

The majority of the arene scope in this work featured sufficiently electron rich substrates that two mechanisms of C─H bond amination may be operating simultaneously in the presence of these highly oxidizing acridinium photocatalysts [Acr-1]BF₄ (E_1/2(Acr⁺*/Acr^•) = +1.71 V vs Fc⁺/Fc in MeCN) and [Acr-2]BF₄(E_1/2(Acr⁺*/Acr^•) = +1.77 V vs Fc⁺/Fc in MeCN).(155,156) (i) The first is the oxidative generation of an electrophilic aminium radical cation intermediate (for example, the oxidation potential of glycine ethyl ester is E_p/2 = +1.22 V vs Fc⁺/Fc in MeCN)(156) and reaction with the neutral, nucleophilic arene (Scheme 10A) (ii) The second is the oxidative generation of an electrophilic arene radical cation intermediate (for example, the oxidation potential of anisole is E_p/2 = +1.49 V vs Fc⁺/Fc in MeCN)(156) and reaction with the neutral, nucleophilic amine (Scheme 10B). In these two regimes, the identity of the electrophilic and nucleophilic reaction components are switched.

Scheme 10.

Two Mechanistic Regimes Operating for the Oxidative C─H Amination of Arenes with Primary Amines, Catalyzed by Acridinium Photocatalysts [Acr-1]BF₄ and [Acr-2]BF₄(155,156)

Photoexcitation of the acridinium catalyst produces an excited state which engages in single-electron transfer with a substrate (amine or arene) to generate the electrophilic substrate radical cation and a stabilized acridine radical. These intermediates could then either engage in productive forward chemistry or return to their respective ground states via back-electron transfer. The transient acridine radical engages with molecular oxygen in an ET step, regenerating the active photocatalyst and superoxide as a byproduct. The acridine radical generated from the photocatalyst [Acr-1] is reported to undergo electron transfer with molecular oxygen at rates of 2.0 × 10¹⁰ M⁻¹ s⁻¹.(157)

The transient electrophilic intermediate (either the aminium radical cation (Scheme 10A) or the arene radical cation (Scheme 10B) reversibly reacts with the neutral closed-shell nucleophile (the arene or the primary amine, respectively), to forge a new C─N bond and generate a delocalized cyclohexadienyl radical cation intermediate. Basic superoxide or hydroperoxide generated through catalyst turnover deprotonates this distonic radical cation, before further oxidation with aromatization yields the aniline product. Terminal oxidation is proposed to occur either via direct trapping of the arene radical with molecular oxygen, followed by base-mediated elimination,(158) or indirectly in a process mediated by either the TEMPO additive or in situ generated hydroperoxyl radical.(159) These engage in a HAT event with the transient arene radical to generate the product arene and TEMPO-H or hydrogen peroxide, respectively. TEMPO can be regenerated from TEMPO-H through HAT(160) or potentially multisite proton-coupled electron transfer (MS-PCET)(161,162) steps with any of molecular oxygen, superoxide, or hydroperoxyl radical. In comparison to closed-shell alkanes, the BDFEs of C─H bonds proximal to a carbon-centered radical are significantly reduced, thus permitting favorable HAT and the formation of an unsaturated product.(163,164)

Only in the cases of benzene and toluene amination did the authors conclude, on the basis of thermodynamic considerations, that an exclusive aminium radical cation mechanism operates, as these cannot feasibly be oxidized by the employed photocatalysts (e.g., the oxidation potential of benzene is E_p/2 = +2.10 V vs Fc⁺/Fc in MeCN(165) and that of toluene is E_p/2 = +1.88 V vs Fc⁺/Fc in MeCN(165)). Arene radical cations experience a drastic enhancement in the acidity of any benzylic protons in comparison to the closed-shell arene, as deprotonation leads to stabilized benzylic radicals (e.g., the calculated pK_a value of the benzylic C─H bond in a toluene radical cation is −13 in MeCN).(27) Thus, it is a remarkable feature of those reactions that can conceivably proceed through multiple mechanisms that amines engage with the potential arene radical cation intermediate as nucleophiles for C─N bond formation, as opposed to bases for benzylic deprotonation.

An interesting mechanistic feature of these C─H amination methods is that the product anilines are all oxidized at potentials substantially lower than those of the substrate amines and/or arenes. For example, the oxidation potential of aniline 36 is E_p/2 = +0.36 V vs Fc⁺/Fc in MeCN,(156) and it therefore is expected to quench the excited-state photocatalyst at significantly faster rates than do the substrates. This feature and the observations that secondary amine products of the primary ARC C─H amination did not re-engage in productive N-arylation with further arene substrate and that aniline products did not undergo a second C─H amination via an arene radical cation pathway are noteworthy.

This report built on earlier work from the Nicewicz group demonstrating (hetero)arene C─H amination with N-heterocycles and the ammonia surrogate ammonium carbamate.(155) Thirteen N-heterocyclic substrates were effective coupling partners, including pyrazoles, triazoles, and tetrazoles. These reactions proceeded under conditions similar to those described above but through an exclusive arene radical cation manifold (e.g., Scheme 10B), as the heteroarene component is unable to undergo thermodynamically favorable single-electron oxidation with these acridinium dyes (for example, the oxidation potential of pyrazole is E_p/2 = +1.89 V vs Fc⁺/Fc in MeCN(155) and that of 1,2,4-triazole is E_p/2 = +2.45 V vs Fc⁺/Fc in MeCN).(155) The group developed a predictive model for site selectivity in the C─H amination of complex aryl and heteroaryl substrates under these conditions.(166) We note a complementary report from the Lei group, achieving a similar photocatalytic arene C─H amination with N-heterocycles employing a Co(III) cocatalyst to mediate a dehydrogenative process under anaerobic conditions.(167)

5.2. Reductive Generation of Alkylaminium Radical Cations for Arene C─H Amination from N-Functionalized Amines

The Leonori group has developed an alternative set of methods for C─H amination of arenes. Their strategy relies upon preactivation of the amine component to then achieve C─H amination under redox-neutral conditions with respect to photocatalytic turnover, initiated via single-electron reduction. In comparison to the method reported by Nicewicz and co-workers above, couplings under this reductive manifold can be successfully promoted using only modestly reducing photocatalysts operating at significantly lower excited-state redox potentials. This offers better compatibility with highly functionalized and oxidatively sensitive substrates, at the cost of requiring stoichiometric quantities of an activating reagent and acid promoter.

In 2017, the Leonori group reported the C─H amination of arenes with preprepared secondary O-aryl hydroxylamines, which could be synthesized from the corresponding amines in three steps of (i) N-oxidation with benzoyl peroxide, (ii) saponification of the resulting benzoate ester, and (iii) final S_NAr reaction with 2,4-dinitrofluorobenzene.(168,169) Blue-light irradiation of acetonitrile solutions of these O-arylhydroxylamines with a wide variety of arene substrates, in the presence of [Ru(bpy)₃]Cl₂ ([Ru-1]Cl₂) photocatalyst (Figure 4) and a superstoichiometric quantity of perchloric acid under anaerobic conditions, led to the efficient synthesis of tertiary aniline products in reaction times of just 15 min (Scheme 11). The wide scope impressively highlights the mildness of the conditions, with 37 different electron-neutral and electron-rich arene partners in 34–91% isolated yields, including the late-stage C─H amination of complex biologically relevant substrates such as indomethacin (50) and strychnine (51), and 14 different hydroxylamine partners including cyclic and acyclic species and those derived from the pharmaceutical compounds fluoxetine (48) and donepezil (49). The regioselectivity in monosubstituted arenes was typically modest (e.g., 45) but improved markedly in differentially disubstituted (e.g., 46) and polycyclic aromatic (e.g., 47) substrates.

Scheme 11.

Photocatalytic C─H Amination of Arenes and Heteroarenes with O-Aryl Hydroxylamines^a,(168)

^a*Indicates minor site of amination.

The authors noted that since they were not starting from the ground-state amine, a stoichiometric quantity of an acid is required to protonate the intermediate nucleophilic aminyl radical to obtain the desired electrophilic aminium radical cation for a successful reaction with arenes. Variation of the acid pK_a in the reaction optimization revealed that perchloric acid (pK_a = −10 in H₂O) was optimal, and weaker acids (AcOH, pK_a = 4.7 in H₂O; TFA, pK_a = 0.2 in H₂O; pTsOH, pK_a = −2.8 in H₂O) were ineffective at promoting efficient product formation. As support for these experimental observations, it was found through NMR studies (CD₃CN) that only perchloric acid was found to be able to fully protonate the weakly basic substrate. This was interpreted as meaning that protonation of the hydroxylamine is required prior to electron transfer for successful C─N bond formation; thus, the electrophilic aminium radical cation results directly from N─O bond cleavage and is immediately able to engage with the arene substrate for bimolecular C─N bond formation. Electron transfer can still occur to the unprotonated hydroxylamine, albeit at a slower rate, but the resultant nucleophilic neutral aminyl radical cannot carry out C─H amination with π-nucleophilic arenes at meaningful rates without protonation. Since aminyl radicals are short-lived reactive species, adding the requirement for a bimolecular protonation event prior to bimolecular C─N bond formation apparently renders this two-step sequence unable to outcompete unproductive pathways.

A mechanistic proposal for this reaction involves single-electron reduction of the protonated O-arylhydroxylamine substrate mediated by photoexcited [Ru-1]Cl₂ (E_1/2(Ru(III)/Ru(II)*) = −1.19 V vs Fc⁺/Fc in MeCN)(90) (Scheme 12). The reduction potential of the neutral hydroxylamine substrate derived from piperidine was recorded as E_p/2 = −1.28 V vs Fc⁺/Fc in MeCN, which is slightly endergonic with respect to the Ru(II) photoreductant.(168) Upon protonation, this is calculated to increase significantly to E^red(calcd) = +1.32 V vs Fc⁺/Fc(168) and therefore rapidly engage in electron transfer with the photoexcited state reductant. Comparing this to the reduction potential of the powerful oxidant Selectfluor (E_p/2 = −0.05 V vs Fc⁺/Fc in MeCN)(170) reveals just how profound an effect on the oxidant ability simple protonation has on this class of group transfer reagents. Following electron transfer, the N─O bond undergoes irreversible heterolytic fragmentation to eject the aminium radical cation and phenoxide. The secondary ARC then reversibly engages with the arene, forming a distonic radical cation which can be deprotonated by the phenoxide. At this point, the Ru(III) state of the photocatalyst oxidizes the radical to yield the carbocation, closing the catalytic cycle and, after proton transfer, yielding the product. Alternatively, a radical chain process can be initiated, wherein the intermediate distonic radical cation following C─N bond formation engages in electron transfer with the protonated substrate to give the product and further aminium radical cation.

Scheme 12.

Multiple Mechanistic Pathways for Arene C─H Bond Amination Through Secondary Alkylaminium Radical Generation from Protonated O-Aryl Hydroxylamines(168)

Control reactions revealed a significant background reaction in the absence of visible-light irradiation (for example, 51% yield in the absence of irradiation, vs 61% yield with irradiation for a model substrate). The protonated O-arylhydroxylamine substrate is a sufficiently powerful oxidant (for example, that derived from piperidine, E^red(calcd) = +1.32 V vs Fc⁺/Fc)(168) to engage in electron transfer with the weakly reducing ground-state Ru(II) complex (E_1/2(Ru(III)/Ru(II)) = +0.91 V vs Fc⁺/Fc in MeCN),(90) to enable a less-efficient dark cycle to occur. A similar observation was made by Ritter and co-workers on the ability of this ground-state Ru(II) complex to mediate the SET reduction of Selectfluor in related arene C─H amination work.(151)A quantum yield determination experiment (Φ = 44) supports the hypothesis of these multiple mechanistic pathways for C─H amination, with a radical chain process likely being dominant.

As this method exploits photoreduction of an activated substrate to generate the aminium radical cation with a very favorable thermodynamic driving force, it is expected that the product inhibition of catalytic turnover is significantly diminished in comparison to the work from Nicewicz. In those reports (vide supra), substrate and product both quench the excited-state photocatalyst through single-electron oxidation, and the product is significantly easier to oxidize than the substrate and thus is expected to slow down further conversion as the product accumulates. Conversely, in this work, the oxidation potentials of the product anilines (for example, aniline 45E_p/2 = +0.39 V vs Fc⁺/Fc in MeCN)(168) approximately match that of the photoexcited-state oxidant (E_1/2(Ru(II)*/Ru(I)) = +0.39 V vs Fc⁺/Fc in MeCN);(90) therefore, undesired product aniline oxidation is likely slow relative to the very favorable generation of the alkylaminium radical cation via reduction of the protonated hydroxylamine substrate.

It is important to note in this class of prefunctionalized substrates that formation of the aminium radical cation intermediate is irreversible. After single-electron reduction and N─O bond fragmentation, the aminium radical cation cannot re-form the hydroxyamine but must engage with the arene partner at meaningful forward rates if successful product formation is to occur. This mode of reactivity is distinct from those involving the reversible formation of aminium radical cations from unfunctionalized amines as reported by Zheng,(80) Nicewicz(156) and ourselves,(88,105,136) insofar as if the rate of productive forward reaction is slow, the closed-shell amine substrate can re-form via BET or PT/HAT pathways and re-engage in the same catalytic cycle without losses in mass balance.

Recognizing that a limitation of this work is that the hydroxylamine substrates typically required three synthetic steps to access, the Leonori group in collaboration with scientists at AstraZeneca devised a method of one-pot substrate activation to enable the direct reaction of primary and secondary amines for aryl C─H bond amination.(171) The group discovered that N-chloroamines, formed through initial reaction of an amine with N-chlorosuccinimide (NCS) prior to irradiation with blue LEDs, in the presence of the same Ru(II) photocatalyst ([Ru-1]Cl₂) were competent ARC precursors to enable arene C─H amination under similar reaction conditions (Scheme 13). Whereas the scope of hydroxylamine precursors was limited to secondary amines, this method of one-pot activation allowed for both secondary amine (32 examples, 30–99% isolated yields) and primary amine reactivity (7 examples, 25–74% isolated yields). A broader functional group tolerance was demonstrated, with alkyl halide, alkyl alcohol, ester, and olefin functionalized amines being employed. The scope with respect to the arene component was similarly broadened (58 examples, 17–99% isolated yields), and for the first time, moderately electron poor arenes such as fluoro-, chloro-, and trifluoromethoxybenzenes were competent partners for photogenerated ARCs with good efficiency. A wide range of highly functionalized arene coupling partners underwent successful amination, including dextromethorphan (54), (+)-dihydroquinidine (55), and the tetrapeptide Ac-Phe-Gly-Leu-Pro-OMe (56).

Scheme 13.

Arene C─H Bond Amination with Alkylaminium Radical Cations Generated from N-Chloroammonium Salts^a,(171)

^a*Indicates minor site of amination.

Mechanistically, this reaction is proposed to proceed similarly to the photocatalytic pathway discussed previously. Protonation of the N-chloroamine with perchloric acid (e.g., for protonated N-chloropiperidinium, E_p/2 = +0.05 V vs Fc⁺/Fc in MeCN)(171) facilitates single-electron reduction mediated by photoexcited [Ru-1]Cl₂ (E_1/2(Ru(III)/Ru(II)*) = −1.19 V vs Fc⁺/Fc in MeCN).(90) Following electron transfer, N─Cl bond heterolytic fragmentation and elimination of chloride generates the transient ARC irreversibly, which then engages with the arene reaction partner. This Ru(II) dye is insufficiently reducing in its photoexcited state to engage in SET with the neutral N-chloroamine (e.g., for N-chloropiperidine, E_p/2 = −2.18 V vs Fc⁺/Fc in MeCN).(171) In addition to serving as a proton source, perchloric acid controlled the selectivity for the desired radical C─H bond amination against polar background C─H bond chlorination in electron-rich arenes. With O-arylhydroxylamines (vide supra), the protonated substrate was a sufficiently strong oxidant to engage with the weakly reducing ground-state Ru(II) catalyst, and hence substantial dark reactivity was observed. This was not the case with the N-chloroamine, where no dark reactivity was observed. The possibility of radical chain propagation remains but is not likely to be a major pathway, as revealed through quantum yield determination experiments (Φ(MeCN) = 0.28, Φ(HFIP) = 0.93).

Since this report, the one-pot N-functionalization of amines with NCS and subsequent photocatalytic reductive generation of aminium radical cations has been used to promote two examples of broadened anti-Markovnikov olefin difunctionalization beyond hydroamination. The Leonori group reported the anti-Markovnikov aminochlorination of unactivated olefins, which are versatile intermediates toward a number of other 1,2-amino functionalized products such as diamines, amino alcohols, and amino nitriles, via aziridinium formation and nucleophilic ring opening.(172) Yu and Liu have reported the anti-Markovnikov aminoarylation of tethered N-aryl acrylamides, where now the proposed transient aminium radical cation adds in a Giese-type fashion(63) to the activated olefin, prior to a 5-exo-trig cyclization of the resultant carbon-centered radical onto the pendant anilide. This pathway leads to aminated oxindoles after subsequent rearomatization of the arene.(173)

In summary, two modes of reactivity have emerged through these works that enable arene C─H bond amination via either oxidative or reductive aminium radical cation generation. Challenges remain in the ability to control or alter innate regioselectivity preferences, which often leads to product mixtures in the case of monosubstituted arenes. However, these strategies offer complementary reactivity to transition-metal-catalyzed methods of C─H bond amination proceeding through rate-determining C─H bond cleavage, while also offering distinct advantages in the mild reaction conditions under which they can be promoted, using visible-light energy to drive the process.

6. Intermolecular C─H Pyridination of (Hetero)Arenes with Pyridinium Radical Cations Generated through Photocatalysis

N-Functionalized pyridinium salts have emerged as important redox-active reagents with utility in photoredox chemistry for group transfer reactions.(174,175) The more established mode of reactivity is as a precursor to a radical of the attached N-functional group through SET-triggered homolysis of the reagent for carbon, aminyl, or alkoxy radical generation. However, more recently their utility for pyridinium radical cation generation through an SET-triggered heterolytic fragmentation pathway of pyridinium triflate and fluoride reagents has been established. We discuss here the arene C─H pyridination reactivity of pyridinium radical cations generated from these reagents under photocatalytic activation. To our knowledge there currently exist four reports of this reactivity from the Togni, Carreira, and Ritter research groups.

A 2018 study from the Togni group explored the utility of N-trifluoromethoxypyridinium reagents, prepared simply by reaction of pyridine N-oxides with the Togni reagent,(176) for radical C─H trifluoromethoxylation of arenes via reductive SET-triggered N─O bond homolysis and trifluoromethoxy radical generation.(177)However, in addition to these desired products, even under fully optimized conditions a minor amount of N-aryl pyridinium products resulting from SET-triggered N─O bond heterolysis, pyridinium radical cation formation, and addition to the arene substrate was observed (Scheme 14). This was the first observation of photocatalytic pyridinium radical cation generation from this class of reagents, and simultaneous follow-up studies from this group in collaboration with the Carreira group(178) and the Ritter group(179,180) looked to further develop this observation into a synthetically useful arene C─H pyridination reaction. These N-aryl pyridinium products serve as surrogates to primary anilines through aminolysis, reminiscent of Zincke salt reactivity.(181)

Scheme 14.

Observation of an Unexpected Arene C─H Pyridination Product During Studies on Arene C─H Trifluoromethoxylation(177)

In order to selectively favor a SET-triggered heterolytic fragmentation pathway over a conventional homolysis pathway, these groups investigated pyridinium triflate and pyridinium fluoride reagents, for which the leaving group anion stabilizing ability is significantly increased to compensate for the unfavorable pyridinium radical cation formation. In this context, pyridinium triflates were favored over fluorides, due to the more straightforward synthetic access through simple sulfonylation of pyridine N-oxides(182) and higher reactivity for C─H pyridination.

Blue-light irradiation of acetonitrile solutions of pyridinium triflate reagents 61 and 62 in the presence of arene substrates and [Ru(bpy)₃](PF₆)₂ ([Ru-1](PF₆)₂) photocatalyst led to efficient arene and heteroarene C─H pyridination at room temperature with reaction times of typically less than 1.5 h (Scheme 15). The N-arylpyridinium products of the reaction could be isolated (e.g., 63–65) or directly subjected to an aminolysis protocol to liberate neutral primary anilines (e.g., 66–70). Whereas the scope of arenes for C─H amination with alkylaminium radical cations was typically limited to electron-rich and electron-neutral examples, with some extension to moderately electron-poor arenes in the later Leonori report(171) (vide supra), the more highly electrophilic nature of the pyridinium radical cation in these works enable expanded reactivity with significantly electron poor arenes and heteroarenes (e.g., 64–66). As a tradeoff for this increased reactivity, however, these C─H pyridination reactions typically exhibit decreased regioselectivity, with mixtures of ortho, meta, and para products typically resulting.

Scheme 15.

Simultaneous Reports of (Hetero)arene C─H Pyridination through the Visible-Light-Mediated Generation of Pyridinium Radical Cations^a,(178-180)

^a*Indicates minor site of amination.

Togni and Carreira reported 18 examples of the isolation of the N-arylpyridinium products (20–89% isolated yields), in addition to 14 examples of their direct aminolysis without intermediate isolation (28–74% isolated yields).(178) In addition, the expanded reactivity of these N-arylpyridinium salts beyond aminolysis was explored (Scheme 16). For example, hydrogenation over platinum oxide liberated the neutral aryl piperidine (e.g., 71, 9 examples). Partial reduction to the 1,2- (72) or 1,4-dihydropyridine (73) was achieved with either sodium borohydride or sodium amalgam, respectively. The pyridinium ring was rendered sufficiently electrophilic to react with carbanionic nucleophiles such as methylmagnesium chloride (74) and the Ruppert–Prakash reagent (75), with C2 selectivity. The Ritter group reported 16 examples of C─H pyridination and in situ aminolysis (37–95% isolated yields), including the late-stage C─H pyridination of the electron-deficient arene ring of sitagliptin (70).(179) By utilizing 2- or 4-chloropyridinium reagents, this method was extended to arene C─H pyridonation, through a two-step sequence of C─H pyridination/hydrolysis, without isolation of the intermediate salt, with 2 examples of 2-pyridonation and 16 examples of 4-pyridonation presented (29–98% isolated yields) with a similar arene scope to the first reports.(180) We note here a related procedure from Kano yielding similar N-arylpyridinium salts though photocatalytic arene radical cation formation and trapping with pyridine, with potassium persulfate as a stoichiometric oxidant. Though this method demonstrates an ability of [Ru(bpy)₃](PF₆)₂ ([Ru-1](PF₆)₂) to engage in significantly endergonic electron transfer with these substrate, the scope favors electron-rich arenes due to the requirement for arene oxidation.(183)

Scheme 16.

Expanded Reactivity of Pyridium Salts(178)

Mechanistically, these reactions are proposed to be initiated via single-electron reduction of the N-pyridinium triflate reagents (61, E_p/2 = −0.30 V vs Fc⁺/Fc in MeCN;(178)62, E_p/2 = −0.52 V vs Fc⁺/Fc in MeCN)(179)mediated by photoexcited [Ru-1](PF₆)₂ (E_1/2(Ru(III)/Ru(II)*) = −1.19 V vs Fc⁺/Fc in MeCN)(90) (Scheme 15). This SET event then triggers fast, irreversible N─O bond heterolysis and generation of the pyridinium radical cation. DFT studies reveal that a heterolytic pathway is favored by more than 10 kcal/mol in comparison to a homolysis pathway, in avoiding formation of a highly destabilized triflyl radical.(178) The pyridinium radical cation rapidly engages in an irreversible addition to an arene substrate, before electron transfer and proton transfer steps mediate N-arylpyridinium formation. This second electron-transfer event is likely mediated by the Ru(III) state of the photocatalyst, thereby closing a redox-neutral catalytic cycle. Ritter presents an alternative scenario, where further pyridinium triflate reagent can also facilitate the second electron transfer step and initiate an inefficient radical chain pathway. These proposals are supported by Stern–Volmer quenching studies, spin-trapping of the intermediate pyridinium radical cation, and kinetic isotope effect measurements, showing that C─H cleavage is not rate-determining.(184)

In summary, what began as a chance observation of side reactivity in arene C─H trifluoromethoxylation has now enabled the development of collection protocols for arene C─H aminations with pyridines and pyridones proceeding via pyridinium radical cation intermediates. These offer superior reactivity to alkylaminium radical cations for the amination of electron-poor arenes in particular and can be useful surrogates for primary anilines and piperidines through aminolysis and hydrogenation, respectively. However, this improved reactivity is accompanied by diminished site selectivity, and multiple regioisomers are often obtained in the amination of monosubstituted arenes.

7. Intermolecular Amination of Carbocations Generated through the Mesolytic Cleavage of Alkoxyamine Radical Cations

While the major focus of this perspective article is on the reactivity of aminium and pyridinium radical cations for C─N bond formation through their direct reaction with alkenes and arenes, we highlight here a mechanistically distinct pathway for photocatalytic C─N bond formation. We have developed a class of alkoxyamines that, when they are converted to the corresponding aminium radical cations by a suitable photooxidant, generate reactive carbocationic intermediates via mesolytic cleavage of the substrate C─O bond. These in turn react with nucleophilic nitrogen-based coupling partners for C─N bond formation.

Mesolytic cleavage is defined as the scission of a bond in a radical ion whereby a neutral radical and an ion are formed as two separate products.(185) Radical ions experience a significant weakening of chemical bonds proximal to the site of electron transfer, and scission can occur readily when these bond strengths are brought close to zero.(186) We can determine the molecular features governing the efficiency of a hypothetical mesolytic cleavage reaction in a radical cation intermediate through the construction of a simple thermodynamic cycle(187-189) (Scheme 17). Through this analysis, we see that the difference in bond strengths between the neutral closed-shell substrate (BDFE_Sub) and the radical cation (BDFE_RadCat) is equal to the potential difference between the substrate/radical cation redox couple (RX^+•/RX) and the resultant carbocation/carbon-centered radical redox couple (R⁺/R^•). Since the reduction potential of the carbocation (R⁺) is decoupled from the identity of the dissociated radical fragment (X^•), we can deduce that the magnitude of the bond-weakening effect primarily depends upon the oxidation potential required to generate the radical cation and the strength of the scissile bond in the neutral substrate.

Scheme 17.

Thermochemistry of the Mesolytic Cleavage of a Hypothetical RX Radical Cation

In order to exploit this elementary step for synthetically useful carbocation generation, we set out to design a substrate class in which we could access a radical cation intermediate at a mild oxidation potential, for widespread compatibility with complex substrates and diverse nucleophiles.(189) In setting this requirement, through the above analysis we see that, for mesolytic cleavage to occur readily, the substrate must have an already weak bond in its neutral ground state. We chose to investigate TEMPO-derived alkoxyamines in this context, as this class of precursors fit both of these requirements—being oxidized at potentials (E_p/2 ≈ 0.70 V vs Fc⁺/Fc in MeCN)(189) less positive than those of many common nucleophiles and having weak ground-state C─O bonds as a result of the high stability of the TEMPO radical (the BDFE of the C─O bond in the TEMPO adduct of isopropylbenzene is 26 kcal/mol(190,191) in comparison to the C─O bond strength in the corresponding benzylic alcohol of 81 kcal/mol).(192) Though this process is heavily precedented in the field of nitroxide-mediated polymerization (NMP) for radical generation via reversible thermolysis,(193,194) we were aware of only one existing report from the Braslau group demonstrating carbocation generation from this precursor using cerium(IV) ammonium nitrate as a chemical oxidant for the terminal functionalization of polymers created through NMP.(195)

Subjecting a nitromethane solution of tetrahydronaphthyl alkoxyamines to blue-light irradiation in the presence of the [Ir(dF(CF₃)ppy)₂(5,5′-d(CF₃)bpy)]PF₆ ([Ir-4]PF₆) photocatalyst (E_1/2(Ir(III)*/Ir(II)) = +1.30 V vs Fc⁺/Fc in MeCN),(135) (Figure 10) a range of C, N, and O nucleophiles efficiently coupled with the benzylic carbocation resulting from mesolytic cleavage of the alkoxyamine radical cation intermediate formed upon single-electron oxidation of the substrate (Scheme 18).(189) Successful nitrogen nucleophiles included aniline (76), diphenylamine (77), methyl carbamate (79), 4-toluenesulfonamide (80), and trimethylsilyl azide (81). An indole underwent C3-alkylation, as opposed to N-alkylation (78). In order to close the catalytic cycle, the TEMPO radical is reduced to TEMPO-H through a proton-coupled electron transfer mechanism,(196) as direct reduction of the TEMPO radical to the corresponding anion (E_1/2 = −1.95 V vs Fc⁺/Fc in MeCN)(197) is significantly beyond the capability of the reduced-state Ir(II) complex (E_1/2(Ir(III)/Ir(II)) = −1.07 V vs Fc⁺/Fc in MeCN).(135) The method was limited to the generation of tertiary and benzylic carbocations, and attempts to extend to secondary alkyl carbocation generation failed due to the inability of the transient aminium radical cation to undergo bond scission in preference to back electron transfer.

Figure 10.

Photocatalysts employed for the generation of alkoxyamine radical cations and the synthesis of pyrroloindoline natural products.

Scheme 18.

C─N Bond Formation through the Mesolytic Cleavage of Alkoxyaminium Radical Cations(189)

We employed this reaction in the enantioselective total syntheses of the pyrroloindoline natural products (−)-psychotriasine (86), (−)-calycanthidine, and (−)-chimonanthine.(198) In an initial step, enantioenriched TEMPO-substituted pyrroloindoline 83 was prepared in 93% ee as a common intermediate toward these alkaloids through photocatalytic PCET activation(199) of N-Cbz tryptamine (82) and trapping of the C3-radical with TEMPO, mediated by the fac-[Ir(ppy)₃] photocatalyst ([Ir-5]) and a chiral BINOL derived phosphate Brønsted base cocatalyst (Scheme 19). This PCET process is understood to proceed through an Ir(IV)/Ir(III) pathway, via initial oxidative quenching of photoexcited [Ir-5] (E_1/2(Ir(IV)/Ir(III)*) = −1.73 V vs Fc⁺/Fc in MeCN) with the TIPS–EBX oxidant (E_p/2 = −1.13 V vs Fc⁺/Fc in MeCN).(198) Subsequently, through the aforementioned mesolytic cleavage reaction manifold, we could promote facile tertiary carbocation generation and trap with C- and N-nucleophiles with complete stereoretention, as expected on the basis of cis-/trans-bicyclo[3.3.0]octane ring strain considerations.(200) For the synthesis of (−)-psychotriasine (86),(201) mesolytic cleavage was carried out in the presence of 2-iodoaniline, giving the expected 3-(N-arylamino)pyrroloindoline product (85). Larock-type annulation(202,203) and global reduction gave the natural product in only four steps from N-Cbz tryptamine (82).

Scheme 19.

Application of Alkoxyaminium Radical Cation Mesolytic Cleavage and Subsequent C─N Bond Formation for the Total Synthesis of (−)-Psychotriasine (86)(198)

Since our report, the Crich and Kamigaito groups have used TEMPO-derived alkoxyamines for photocatalytic carbocation generation in the context of glycosylation(204) and cationic polymerization, respectively.(205)The Coote group too have advanced this methodology using electrochemistry to activate alkoxyamines toward bond cleavage. They demonstrated that TEMPO–Me serves as an electrochemical methylating reagent of aromatic and vinyl carboxylic acids.(206,207) Through theoretical and empirical investigation of solvent and electrolyte,(208,209) the group discovered that an S_N2-like reaction pathway can be promoted, allowing for methyl group transfer, which is distinct from the S_N1-like pathway leading to stabilized carbocations operative in our photochemical method.

8. Conclusions

The photocatalytic oxidation of amines and reduction of N-functionalized amines have emerged as mild methods for the generation of aminium radical cations. In comparison to closed-shell, nucleophilic amines, aminium radical cations offer complementary electrophilic reactivity and anti-Markovnikov regioselectivity in the addition to alkenes that make them valuable intermediates in synthesis. These species readily participate in C─N bond forming processes and have subsequently been utilized for applications including alkene hydroamination and arene C─H amination. Looking forward, we optimistically anticipate further advances in the development of practically useful synthetic operations that utilize aminium radical cations for the construction of complex amine architectures.