Photocatalytic Anti-Markovnikov Hydroamination of Alkenes with Primary Heteroaryl Amines

Abstract

We report a light-driven method for the intermolecular anti-Markovnikov hydroamination of alkenes with primary heteroaryl amines. In this protocol, electron transfer between an amine substrate and an excited-state iridium photocatalyst affords an aminium radical cation (ARC) intermediate that undergoes C–N bond formation with a nucleophilic alkene. Integral to reaction success is the electronic character of the amine, wherein increasingly electron-deficient heteroaryl amines generate increasingly reactive ARCs. Counteranion-dependent reactivity is observed, and iridium triflate photocatalysts are employed in place of conventional iridium hexafluorophosphate complexes. This method exhibits broad functional group tolerance across 55 examples of N-alkylated products derived from pharmaceutically relevant heteroaryl amines.

Graphical Abstract

Heteroaryl amines are prevalent in small molecule therapeutics and are of particular interest in medicinal contexts due to their unique hydrogen-bonding properties and high aqueous solubility.¹ These motifs are frequently synthesized via nucleophilic aromatic substitution and cross-coupling reactions, wherein C(sp²)–N bond formation occurs between a nucleophilic alkyl amine and an electrophilic haloarene.² Conversely, C(sp³)–N bond formation using widely available heteroaryl amines remains underdeveloped. S_N2-type reactivity is limited by the attenuated nucleophilicity of resonance-stabilized amines,³ and heteroaryl amines have a propensity to unproductively bind metal centers in transition metal-catalyzed processes.⁴

Intermolecular alkene hydroamination offers an attractive and atom-economical approach to C(sp³)–N bond formation by making use of widely available olefin and amine feedstocks. Numerous catalytic methods have been developed to accommodate a range of amine classes,^4a,5 including recent advances made by Hartwig, Nelson, Marsden, and Verma using primary heteroaryl amines.^4,6 Still, existing protocols are limited in scope and either proceed with Markovnikov regioselectivity or employ conjugated alkenes. The intermolecular anti-Markovnikov hydroamination of unactivated alkenes with heteroaryl amines remains a challenge with no general solution.

Our laboratory has developed a suite of photoredox-catalyzed hydroamination methods that proceed via single electron oxidation of amines to generate electrophilic aminium radical cation intermediates (Figure 1A).⁷ These species are well known to participate in anti-Markovnikov C–N bond formation with nucleophilic alkenes⁸ and deliver alkylated amine products in the presence of a thiol hydrogen atom transfer (HAT) catalyst. In 2014, we reported the intramolecular hydroamination of styrene-tethered anilines, demonstrating the use of aryl amines as competent ARC precursors for C–N bond formation.^7a However, these ARCs failed to undergo intermolecular couplings. This is presumably due to slow C–N bond formation,⁹ which is precluded by favorable back electron transfer (BET) between the ARC and the reduced state of the photocatalyst (Figure 1B).¹⁰

Figure 1.

(A) Photocatalytic hydroamination via ARC intermediates. (B) Kinetics of intermolecular C–N bond formation.⁹ (C) Trends in oxidation potential for various amines.^7a,11 (D) This work: anti-Markovnikov hydroamination with primary heteroaryl amines.

In contrast, 1° and 2° alkyl amine-derived ARCs undergo rapid C–N bond formation and are competent in intermolecular hydroamination reactions.^7b,7c,9 We postulated that the improved reactivity of alkyl amines is due to the increased electrophilicity of aliphatic ARCs relative to aniline-derived ARCs. This trend is reflected by the higher potentials required for aliphatic ARC generation (Figure 1C).¹¹ Nocera and coworkers further elucidated the relationship between oxidation potential and ARC electrophilicity by establishing that 1° alkyl amines undergo selective monoalkylation in part due to the increased rate of C–N bond formation for 1° alkyl ARCs relative to 2° ARCs.¹² Building on this understanding of ARC reactivity, we hypothesized that ARCs generated from heteroaryl amines would undergo intermolecular addition to alkenes at rates competitive with BET given their high oxidation potentials (Figure 1C). Here, we report the generation of aryl ARC intermediates for the intermolecular anti-Markovnikov hydroamination of alkenes with 1° heteroaryl amines (Figure 1D). This method enables the construction of a range of pharmaceutically relevant scaffolds and further elucidates characteristic trends in ARC reactivity.

We began our study by investigating the hydroamination of ethylidene cyclohexane with electron-deficient 2-aminopyrimidine 1a (E_p/2 = 1.21 V vs Fc⁺/Fc in MeCN). Upon blue light irradiation of a dioxane solution of 1a and ethylidene cyclohexane in the presence of 3 mol% [Ir(dF(CF₃)ppy)₂(5,5’-dCF₃-bpy)]PF₆ photocatalyst ([Ir-A]PF₆) and 30 mol% 4-methoxythiophenol HAT cocatalyst (4-OMe-PhSH), we observed 57% yield of the desired product 2a (Table 1, entry 1). The more oxidizing [Ir(dCF₃(CF₃)ppy)₂(5,5’-dCF₃-bpy)]PF₆ ([Ir-B]PF₆) provided a similar yield of 60% (entry 2). Considering recent work exploring ion-pairing dynamics in photoredox chemistry¹³ alongside the broader importance of counterions in catalytic transformations,¹⁴ we assessed counteranion identity as an optimization parameter. Intriguingly, reactivity significantly improved in the presence of tetrabutylammonium triflate (entry 3). This discovery led us to prepare [Ir-B]OTf, which resulted in an increased yield of 89% (entry 4). The more weakly coordinating tetrakis(3,5-bis(trifluoromethyl)phenyl)borate (⁻BAr^F₄) salt [Ir-B]BAr^F₄ gave diminished yield (entry 5), corroborating counteranion-dependent reactivity. Alternative solvents such as trifluorotoluene and dichloromethane resulted in lower yields of 2a (entries 6, 7), as did lower equivalents of alkene (entry 8). Omission of light, thiol, or photocatalyst resulted in complete loss of reactivity (entries 9–11).

Table 1.

Reaction Optimization^a

^aReactions were performed on 0.1 mmol scale. Yields were determined via ¹H NMR spectroscopic analysis of the crude reaction mixtures relative to 1,3,5-trimethoxybenzene as an internal standard.

^b2.5 equiv. alkene employed.

^cNo irradiation.

^d0 mol% 4-OMe-PhSH.

^ePhotocatalyst redox potentials were estimated relative to Fc⁺/Fc in MeCN (see Section S12).

Having optimized conditions for hydroamination with 1a, we evaluated 2-aminopyrazine 1b and 2-aminopyridine 1c, which are oxidized at lower potentials (E_p/2 = 0.98 V and 0.83 V vs Fc⁺/Fc in MeCN for 1b and 1c, respectively). Although the milder oxidant [Ir-A]OTf outperformed [Ir-B]OTf, yields of 2b and 2c in dioxane were relatively low (entries 12, 15). Trifluorotoluene solvent enhanced reactivity for both amines (entries 13, 16), and [Ir-A]OTf provided similar or improved yields relative to [Ir-A]PF₆ (entries 13–14, 16–17). Thus, we identified [Ir-A]OTf and [Ir-B]OTf as effective catalysts, and dioxane and PhCF₃ are preferred solvents. We note that empirically, [Ir-A]OTf is optimal for amines with E_p/2 < 1.20 V vs Fc⁺/Fc in MeCN, and [Ir-B]OTf is optimal for amines with E_p/2 > 1.20 V. Finally, while this study uses independently synthesized iridium triflate complexes as catalysts, commercial [Ir-A]PF₆ and tetrabutylammonium triflate can be employed in conjunction to effect the hydroamination in comparable yields. Additional optimization experiments are provided in Section S2.

With effective reaction conditions in hand, we explored the amine scope of this method (Table 2). In addition to 2-aminopyridine 1c, a variety of (pseudo)halogenated derivatives readily underwent hydroamination to furnish alkylated products in good yields (2c, 3–8). Alkyne and boronic ester functional handles were well tolerated in the reaction (9, 10), and 2-Aminopyridines bearing various electron-withdrawing groups were successful substrates (11–15).

Table 2.

Amine Scope^a

^aYields are the average of two experiments and are for isolated material on 0.5 mmol scale unless otherwise noted.

^b[Ir-A]OTf photocatalyst, PhCF₃ solvent.

^c[Ir-B]OTf photocatalyst, dioxane solvent.

^d[Ir-A]OTf photocatalyst, dioxane solvent.

^e[Ir-B]OTf photocatalyst, PhCF₃ solvent.

^f2.5 equiv. alkene.

^gYield was determined via ¹H NMR spectroscopic analysis of the crude reaction mixture relative to an internal standard.

^hReaction was conducted on 0.2 mmol scale.

ⁱMethylene cyclopentane was employed as the alkene partner.

Although 2-aminopyridine was highly reactive, unsubstituted 3- and 4-aminopyridine gave lower yields of the desired products (<30% on 0.1 mmol scale), consistent with the correlation between ARC electrophilicity and amine oxidation potential (E_p/2 = 0.83 V, 0.54 V, and 0.64 V vs Fc⁺/Fc in MeCN for 2-, 3-, and 4-aminopyridine, respectively). We found that 3-aminopyridines bearing electron-withdrawing groups were reactive substrates (16–18). Additionally, we hypothesized that the high basicity of 4-aminopyridine results in the deprotonation of ARC intermediates by closed-shell amine (pK_aH = 17.6 and 14.5 in MeCN for the conjugate acids of 4- and 2-aminopyridine, respectively).¹⁵ This finds precedence in previous work, where ARC deprotonation by amine starting material was identified as a major nonproductive pathway in hydroamination reactions with alkyl amines.¹² Notably, 4-Aminopyridines bearing substituents adjacent to the pyridyl nitrogen atom afforded alkylated products in good yields (19, 20), presumably due to steric effects or diminished basicity.

In addition to aminopyridines, aminopyrimidines were a versatile substrate class. Various 2-aminopyrimidines furnished alkylated derivatives in good yields (2a, 21–24), and substituted 4- and 5-aminopyrimidines were also well tolerated (25–27). Other successful aminoheterocycles included a pyrazine, pyridazine, quinazoline, and thiadiazole (2b, 28–30). While anilines were often less reactive than heteroaryl amines, bis(3,5-trifluoromethyl) aniline provided 31 in 82% yield. Medicinally relevant amines also underwent efficient hydroamination, including the cores of B-Raf inhibitors encorafenib (32) and dabrafenib (33). Finally, upon subjecting a soluble adenosine derivative to the reaction conditions, 34 was delivered in 77% yield. As such, this method can be employed to access N⁶-alkylated adenosines, which are commonly explored in medicinal contexts as adenosine receptor agonists.¹⁶ Throughout the amine scope, the Markovnikov regioisomer was consistently observed in trace yield but could be separated from the desired product (see Section S5). Overalkylated tertiary amine products were not observed, consistent with the reduced electrophilicity of product-derived 2° ARCs.^7c,12 A compilation of unsuccessful and low-yielding amine substrates can be found in Table S7.

The scope of the alkene partner was next evaluated using 2-amino-5-trifluoromethylpyridine as a model amine (Table 3). Styrene and allyl trimethylsilane underwent hydroamination in good yields (35, 36), requiring only 1.5 equivalents of olefin. Several 1,1-disubstituted alkenes displayed high reactivity, including a protected homoallylic alcohol and a phenylalanine derivative (37, 38). Allylic alcohols and protected allylic amines were well tolerated but less reactive, likely due to inductive deactivation of the alkene π-system (39, 40). Aliphatic 1,2-disubstituted alkenes gave rise to alkylated products in high yields (41, 42). Using 1,5-cyclooctadiene as the alkene, 42 arose from a transannular intramolecular hydroamination following the initial bimolecular reaction. Trisubstituted alkenes bearing piperidine and piperazine motifs underwent hydroamination readily to afford 43 and 44. Feedstock terpenes were also reactive under our hydroamination protocol (45, 46). Notably, linalool underwent regioselective hydroamination to yield 45, consistent with our observation that unactivated monosubstituted alkenes are unreactive in this method (e.g., 6% yield using 5.0 equivalents of 1-octene). Sterically congested α-tertiary amine 47 was formed from tetramethylethylene in excellent yield.

Table 3.

Alkene Scope^a

^aYields are the average of two experiments and are for isolated material on 0.5 mmol scale unless otherwise noted.

^b1.5 equiv. alkene.

^c2.5 equiv. alkene.

^d5.0 equiv. alkene.

^ePhCF₃ solvent.

^f[Ir(dF(CF₃)ppy)₂(bpy)]OTf photocatalyst.

^gFrom the corresponding triisopropylsilyl enol ether.

Heteroatom-substituted alkenes were also successful substrates. Products 48 and 49 were prepared from the corresponding vinyl ethers in excellent yields. An enol phosphinate furnished 50 in 47% yield, and 51 was synthesized from an N-vinyl pyrazole in 83% yield. In some cases, 2-aminopyridine 1c was more reactive than the model 2-amino-5-trifluoromethylpyridine. An enamide and a dihydrofuran underwent hydroamination with 1c to yield 52 and 53 in 51% and 50% yield, respectively. Finally, a triisopropyl silyl enol ether was employed to synthesize the muscle relaxant fenyramidol (54) after desilylation of the hydroamination product. The 55 examples contained within this scope highlight the mild and robust nature of this protocol for anti-Markovnikov alkene hydroamination.

Aiming to elucidate the beneficial role of the triflate anion, we conducted a series of preliminary mechanistic experiments. Our mechanistic hypotheses were guided by the proposed catalytic cycle depicted in Figure 2A.^7,12 Following photocatalytic ARC generation, C–N bond formation furnishes a C-centered radical that undergoes HAT with a thiol cocatalyst. Subsequent electron and proton transfer deliver the hydroamination product and concomitantly turn over the thiol and iridium catalysts. Given the significant influence of ion-pairing dynamics on photoredox processes,¹³ we suspected that the triflate anion might either modify the photophysical properties of its associated iridium complex or modulate electron transfer kinetics between amine and excited-state photocatalyst.

Figure 2.

(A) Proposed catalytic cycle for hydroamination and possible origins of counterion dependence. (B) Kinetic profile of hydroamination and Stern–Volmer quenching studies. (C) Comparison of photocatalyst lifetimes and quenching rates.

The hydroamination of ethylidene cyclohexane with 2-aminopyrimidine 1a catalyzed by [Ir-B]X (X = ⁻OTf or ⁻PF₆) was chosen as a model system for mechanistic experiments due to the significant counterion dependence observed during reaction optimization. Time-course studies revealed that [Ir-B]OTf gave higher yields of 2a than [Ir-B]PF₆ across the entire course of reaction, indicating faster reactivity with the triflate catalyst (Figure 2B, left). The ground-state reduction potentials and photoluminescence spectra of [Ir-B]PF₆ and [Ir-B]OTf in CH₂Cl₂ were nearly indistinguishable (Figures S15 and S23). With these data, we estimated E_1/2(*Ir^III/Ir^II) values of 1.52 V and 1.51 V vs Fc⁺/Fc in CH₂Cl₂ for [Ir-B]PF₆ and [Ir-B]OTf, respectively.¹⁷ This negligible difference in excited-state redox potentials suggests that the thermodynamic favorability of photoinduced electron transfer does not account for the counteranion-dependent reactivity of 1a. Additionally, the absorption spectra of [Ir-B]PF₆ and [Ir-B]OTf exhibited no significant differences (Figure S19).

Nonetheless, Stern–Volmer quenching studies in dioxane revealed that 1a undergoes more efficient electron transfer with [*Ir-B]OTf than [*Ir-B]PF₆ (K_SV = 47 and 34 M⁻¹s⁻¹, respectively; Figure 2B, right). Time-correlated single photon counting lifetime measurements of [*Ir-B]PF₆ (τ = 1072 ns) and [*Ir-B]OTf (τ = 1335 ns) in dioxane demonstrated a longer-lived excited state for [Ir-B]OTf. Together, these data revealed comparable quenching rate constants between each photocatalyst and 1a (k_q = 3.1 × 10⁷ M⁻¹s⁻¹ for [Ir-B]PF₆ vs 3.5 × 10⁷ M⁻¹s⁻¹ for [Ir-B]OTf; Figure 2C). As such, 2-aminopyrimidine quenches [*Ir-B]OTf more efficiently than [*Ir-B]PF₆ primarily due to the increased excited-state lifetime of [Ir-B]OTf. This difference in quenching efficiency may contribute to the observed counterion-dependent reactivity. However, prior studies of ARC-based aminations have indicated reversible ARC formation relative to rate-determining C–N bond formation.^7,12 On this basis, the small variance in quenching efficiency is likely not the sole factor leading to the enhanced reactivity observed with [Ir-B]OTf. We speculate that a tighter ion pair between 1a^·+ and ⁻OTf relative to ⁻PF₆ could stabilize the ARC or inhibit nonproductive deprotonation.¹⁸ Ongoing studies are directed towards further elucidating the role of the counteranion.

In summary, we have developed a general protocol for anti-Markovnikov alkene hydroamination using 1° heteroaryl amines. This method addresses a long-standing synthetic challenge in intermolecular hydroamination and further demonstrates the significant role that counterions can play in photoredox catalysis. We anticipate that the reactivity demonstrated herein will provide new opportunities for heteroaryl amine functionalization and C–N bond formation with aminium radical cations.