Cobalt‐Catalyzed Radical Ligand Transfer (RLT) Enables Remote‐Markovnikov Hydroamination of Alkenes Authors: Arman Khosravi, Ho Martin Yuen, Seth Fremin, Nuwan A. Pannilawithana, Nol Supasueb, Yu Zhang, Peng Liu, Ming‐Yu Ngai Published: 2026 ABSTRACT ABSTRACTAmino alcohols are prevalent in pharmaceuticals, agrochemicals, and functional materials, yet regioselective remote C─N bond formation from simple alkenes remains limited. We report a remote‐Markovnikov formal 1,3‐hydroamination of allyl carboxylates enabled by cooperative photoredox, cobalt, and Brønsted acid catalysis. The reaction proceeds via metal–hydride hydrogen atom transfer (MHAT) followed by 1,2‐radical acyloxy migration (1,2‐RAM) to generate a tertiary carbon‐centered radical at a site remote from the original alkene, affording protected tertiary alkyl amino alcohols across a broad substrate scope, with secondary variants also accessible. Experimental and computational studies support a C─N bond‐forming step best described as radical ligand transfer (RLT) at a Co(III)─N intermediate, rather than Co(IV)/SN2 or carbocationic pathways. This rare nitrogen‐transfer manifold accounts for the observed regioselectivity and catalyst‐influenced stereochemical outcomes. By enabling migration‐controlled 1,3‐hydroamination, this strategy complements conventional 1,2‐hydroamination and expands accessible amino alcohol chemical space. Introduction Nitrogen‐containing functional groups are ubiquitous across pharmaceuticals [1, 2, 3, 4, 5, 6, 7], agrochemicals [8, 9], and functional materials [10, 11, 12, 13, 14, 15]. Analyses of bioactive molecules consistently identify alkyl amines among the most frequently occurring functionalities [2, 3, 16], highlighting the ongoing demand for efficient and selective strategies to construct C─N bonds while maintaining broad functional‐group compatibility and structural flexibility (Figure 1A). Alkene hydroamination represents one of the most direct approaches to alkyl amines [17, 18], although achieving broad scope and high regioselectivity remains challenging. Brønsted acid catalysis enables Markovnikov‐selective addition through cationic intermediates [19, 20, 21, 22, 23] but protonation can reduce amine nucleophilicity [17]. Alkali‐metal‐mediated methods rely on strongly basic amide nucleophiles and are most effective with electronically activated alkenes [24, 25]. Transition‐metal and lanthanide catalysis have expanded the scope of hydroamination by enabling milder conditions [17, 26, 27, 28, 29, 30], anti‐Markovnikov selectivity [18, 31, 32, 33, 34, 35], and asymmetric variants [36, 37, 38, 39, 40, 41, 42, 43]. Despite these advances, most established methods proceed through direct 1,2‐addition, and migration‐enabled or remote C─N bond formation remains less explored (Figure 1A) [44, 45, 46]. Hydroamination of alkenes and importance of amino alcohols. Cobalt‐catalyzed hydrofunctionalization via metal‐hydride hydrogen‐atom transfer (MHAT) has emerged as a powerful alternative paradigm for alkene functionalization [47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63]. In this manifold, MHAT [64, 65, 66, 67, 68] converts an alkene into a carbon‐centered radical under mild conditions, enabling bond constructions that are difficult to access through ionic pathways. Despite increasing application of MHAT to hydroamination [53, 69, 70, 71, 72, 73], the nature of the C─N bond‐forming step remains mechanistically unsettled. Prevailing proposals invoke either ionization to a carbocation and subsequent capture [53] or the formation of a Co(IV)–alkyl intermediate followed by nucleophilic substitution (Figure 1B) [69, 70, 71, 72, 73, 74, 75, 76, 77, 78]. While formation of Co(IV)–alkyl species at primary and secondary carbon centers is well established, their relevance to sterically congested tertiary radicals is far less certain. Here we report a remote‐Markovnikov‐selective formal 1,3‐hydroamination of allyl carboxylates enabled by photoredox, cobalt, and Brønsted acid catalysis. MHAT‐initiated radical generation followed by 1,2‐radical acyloxy migration (1,2‐RAM) furnishes a tertiary radical at a site remote from the original alkene (Figure 1C). Supported by experimental and computational evidence, we propose that C─N bond formation proceeds through radical ligand transfer (RLT) [79] at a Co(III)─N intermediate. This pathway represents a rare and previously weakly supported mode of nitrogen transfer in Co‐Salen‐catalyzed MHAT hydroamination and accounts for both the regioselectivity and stereochemical outcomes observed. Notably, this process directly furnishes tertiary alkyl amines, a class of nitrogen motifs that remains challenging to access selectively through catalytic hydroamination, particularly at sterically congested centers [80, 81, 82, 83, 84]. From a synthetic perspective, this migration‐enabled hydroamination provides access to amino alcohol architectures complementary to those obtained through conventional 1,2‐hydroamination, expanding accessible amino alcohol chemical space useful for medicinal and functional molecule design (Figure 1D) [85, 86, 87, 88, 89, 90, 91, 92, 93]. We commenced our studies using 2‐phenylbut‐3‐en‐2‐yl benzoate (1a) and N‐methylaniline (2a) as a model substrate (Table 1). A systematic evaluation of reaction parameters revealed that combining 1a (1.00 equiv) and 2a (1.50 equiv), HX‐1 (10.0 mol%) with Co‐1 (2.00 mol%) and Ir(ppy)3 (1.00 mol%) in THF (0.100 M), under blue LED irradiation at 40 °C for 18 h, afforded the desired product 3a in 84% yield (entry 1). No product formation was observed without cobalt (entry 2). Substituting Co‐2 gave only 41% yield (entry 3), while Co‐3 afforded 63% (entry 4), suggests that ligand structure can significantly influence the efficiency of cobalt‐mediated steps in the catalytic cycle (e.g., MHAT and RLT). Ir(ppy)3 was found to be irreplaceable: without it (entry 5), or when substituted with PTH or 4CzIPN (entries 6–7), no product was observed, signifying that its distinct redox and excited‐state properties are required. In the absence of HX‐1, the reaction did not proceed, highlighting its essential function in the formation of CoIII─H intermediate (entry 8). Additionally, substituting HX‐1 with other HX surrogates (entries 9–11) suggested the essential roles of the counter ion for optimal reactivity [94]. Control experiments confirmed the photochemical nature of the process, as no conversion was observed in the absence of light (entry 12). While heating was not strictly required, omitting external heating resulted in a diminished yield of 50% (entry 13). Finally, exposure to air completely inhibited the reaction (entry 14), consistent with the involvement of oxygen‐sensitive catalytic intermediates. TABLE 1Selected optimization experiments a . EntryDeviation from the standard conditionsYield%1none842without Co‐1 <23 Co‐2 instead of Co‐1 414 Co‐3 instead of Co‐1 635without Ir(ppy)3 <26PTH instead of Ir(ppy)3 <274CzIPN instead of Ir(ppy)3 <28without HX‐1 <29 HX‐2 instead of HX‐1 7510 HX‐3 instead of HX‐1 7311 HX‐4 instead of HX‐1 7712without light<213without heat5014air<2 a See Supporting Information (SI) for experimental details. Reaction yields were determined by 1H‐NMR using CH2Br2 as an internal standard. Bz, benzoyl.John Wiley & Sons, Ltd. Selected optimization experiments a . See Supporting Information (SI) for experimental details. Reaction yields were determined by 1H‐NMR using CH2Br2 as an internal standard. Bz, benzoyl. With optimized conditions in hand, the scope of anilines was evaluated (Table 2). A wide range of anilines underwent efficient migration‐enabled functionalization to afford β‐acyloxy tertiary alkyl amines. N‐methylaniline delivered 3a in 84% yield. Ortho‐, meta‐, and para‐methyl‐substituted anilines (3b–3d) were readily converted (63%–84% yield). Para‐substituted anilines bearing electron‐donating and electron‐withdrawing groups, including tert‐butyl (3e), methoxy (3f), halogens (3g–3i), trifluoromethyl (3j), trifluoromethoxy (3k), and pentafluoroaniline (3l), all furnished the corresponding products. Broad functional‐group tolerance was observed, accommodating nitrile (3m), ester (3n), ketone (3o), phenyl (3p), biphenyl (3q), dibenzothiophene (3r), dibenzofuran (3s), and boronic ester (3t) substituents. Disubstituted anilines, including 2,4‐dimethyl (3u) and 3,5‐dimethyl (3v) derivatives, were effective, as were other highly substituted systems (3w–3ad). Naphthyl (3ae) and heteroaryl anilines (3af, 3ag) participated efficiently. Beyond N‐methylanilines, tetrahydroquinoline (3ah), N‐cyclohexylaniline (3ai), N‐cyclobutylaniline (3aj), diphenylamines (3ak, 3al), and dihydrodibenzazepine (3am) afforded the corresponding aminated products in moderate to excellent yields. Overall, anilines spanning a broad electronic and steric range furnished the corresponding β‐acyloxy alkyl amines in generally good to excellent yields (typically 60%–90%). Diastereoselectivity varied depending on substrate structure, ranging from modest to high levels (up to >20:1), consistent with substrate‐dependence in the radical C─N bond‐forming step. Gram‐scale synthesis of 3k proceeded smoothly, delivering the product in 78% isolated yield. TABLE 21,3‐Hydroamination of allyl carboxylates with anilines via photoredox/cobalt/Brønsted acid catalysis a . a See Supporting Information for experimental details. Standard reaction conditions: 1 (0.200 mmol, 1.00 equiv.), 2 (0.300 mmol, 1.50 equiv.), Co‐1 (2.00 mol%), Ir(ppy)3 (1.00 mol%), HX‐1 (10.0 mol%), THF (0.100 M), blue LED, 40 °C, 18 h.John Wiley & Sons, Ltd. 1,3‐Hydroamination of allyl carboxylates with anilines via photoredox/cobalt/Brønsted acid catalysis a . See Supporting Information for experimental details. Standard reaction conditions: 1 (0.200 mmol, 1.00 equiv.), 2 (0.300 mmol, 1.50 equiv.), Co‐1 (2.00 mol%), Ir(ppy)3 (1.00 mol%), HX‐1 (10.0 mol%), THF (0.100 M), blue LED, 40 °C, 18 h. The scope of allyl carboxylates was next examined (Table 3). Aryl‐substituted substrates bearing electron‐withdrawing groups at the ortho‐, meta‐, or para‐positions (3an–3ar) underwent efficient 1,3‐hydroamination, as did biphenyl (3as) and ester‐substituted aryl (3at) derivatives. Heteroaromatic ester‐substituted aryl allyl carboxylates (3au–3aw) and a meta‐methoxy‐substituted substrate (3ax) were also competent. Secondary allyl carboxylates required elevated temperatures but afforded the desired products in synthetically useful yields (3ay–3aae). Notably, modification of the migrating group to a 4‐methoxybenzoate (3aaf, 3aag) improved reaction efficiency. Evaluation of the migrating group revealed that both electron‐rich and electron‐deficient benzoates (3aah–3aaq), as well as naphthyl (3aar), 4‐methoxybenzoate (3aas), 3‐trifluoromethylbenzoate (3aat), cyclohexanecarboxylate (3aau), and thiophenyl carboxylate (3aav) substituents were viable. Across aryl and aliphatic allyl carboxylates, the reaction delivered the desired products in moderate to excellent yields (50%–96%). Additional substrate examples are provided in Figure S2. Single‐crystal x‐ray diffraction of 3aag unambiguously confirmed the product structure and its relative stereochemistry. TABLE 31,3‐Hydroamination of allyl carboxylates with anilines via photoredox/cobalt/Brønsted acid catalysis a . a See Supporting Information for experimental details. Standard reaction conditions: 1 (0.200 mmol, 1.00 equiv.), 2 (0.300 mmol, 1.50 equiv.), Co‐1 (2.00 mol%), Ir(ppy)3 (1.00 mol%), HX‐1 (10.0 mol%), THF (0.100 M), blue LED, 40 °C, 18 h. b 80 °C instead of 40 °C.John Wiley & Sons, Ltd. 1,3‐Hydroamination of allyl carboxylates with anilines via photoredox/cobalt/Brønsted acid catalysis a . See Supporting Information for experimental details. Standard reaction conditions: 1 (0.200 mmol, 1.00 equiv.), 2 (0.300 mmol, 1.50 equiv.), Co‐1 (2.00 mol%), Ir(ppy)3 (1.00 mol%), HX‐1 (10.0 mol%), THF (0.100 M), blue LED, 40 °C, 18 h. 80 °C instead of 40 °C. To demonstrate practicality, complex substrates were evaluated (Table 4). Allyl carboxylates or anilines derived from adapalene (4a), probenecid (4b), ciprofibrate (4c), ibuprofen (4d), gemfibrozil (4e), D‐valine (4f), febuxostat (4g), and oxaprozin (4h) underwent selective 1,3‐hydroamination. Natural‐product‐derived substrates bearing vitamin E (4i), cholestanol (4j), tigogenin (4k), and estrone (4l) motifs also furnished the corresponding products. Late‐stage functionalization of complex molecules proceeded with good efficiency, highlighting the functional‐group tolerance and synthetic practicality of the method. TABLE 41,3‐Hydroamination of complex molecules via photoredox/cobalt/Brønsted acid catalysis a . a See Supporting Information for experimental details. Standard reaction conditions: 1 (0.200 mmol, 1.00 equiv.), 2 (0.300 mmol, 1.50 equiv.), Co‐1 (2.00 mol%), Ir(ppy)3 (1.00 mol%), HX‐1 (10.0 mol%), THF (0.100 M), blue LED, 40 °C, 18 h.John Wiley & Sons, Ltd. 1,3‐Hydroamination of complex molecules via photoredox/cobalt/Brønsted acid catalysis a . See Supporting Information for experimental details. Standard reaction conditions: 1 (0.200 mmol, 1.00 equiv.), 2 (0.300 mmol, 1.50 equiv.), Co‐1 (2.00 mol%), Ir(ppy)3 (1.00 mol%), HX‐1 (10.0 mol%), THF (0.100 M), blue LED, 40 °C, 18 h. Mechanistic studies were performed to gain insight into the reaction mechanism (Figure 2). Addition of TEMPO to reaction mixture completely inhibited the product formation implying a radical mechanism (Figure 1A). To further probe the nature of 1,2‐acyloxy migration, benzyl thiol, a known radical H‐atom donor and nucleophile [95], was employed. The formation of reduced 5a, rather than thioether 5b, supported a radical‐mediated 1,2‐acyloxy migration (Figure 2B). Using a enantioenriched catalyst at −20°C, product 3k was obtained with 24% ee, consistent with involvement of the cobalt catalyst in the C─N bond‐forming step and its ability to impart stereochemical bias (Figure 2C and Figure S4). To probe the topology of the migration, 18O‐labeling experiments with 18O‐1a generated both 18O‐3k‐1 and 18O‐3k‐2 in a 1.7:1 ratio (Figure 2D), indicating that 1,2‐radical acyloxy migration proceeds through both [2,3]‐ and [1,2]‐transition states, with the former favored. Measured O18/O16 ratio reflects the rearrangement pathway rather than isotopic scrambling (see Supporting Information for more details). Deuterium‐labeling experiments resulted in 25% deuterium incorporation at the terminal position of the product (Figure 2E), indicating that aniline N─H bonds contribute to formation of the Co─H species in the catalytic process. To distinguish between intra‐ and intermolecular acyloxy migration pathways, crossover experiments using 1y and 1a were conducted under optimized conditions (Figure 2F). The exclusive formation of non‐crossover products, 3aak and 3a, supports an intramolecular acyloxy migration mechanism. Stern–Volmer quenching studies revealed that the excited photocatalyst Ir(ppy)3* is quenched more efficiently by Co‐1 than by HX‐1, 1a or N‐methylaniline (Figure 2G), consistent with initial reduction of Co(II) to Co(I) by photoexcited Ir(ppy)3*, followed by protonation of Co(I) by HX‐1 to generate Co(III)–H species [96]. Finally, both light on/off experiments and quantum yield measurements indicated that an extended radical chain mechanism is unlikely under standard reaction conditions (Figure 2H) [97]. Mechanistic studies, see Supporting Information for details. DFT calculations provide energetic support for a radical relay pathway consistent with the experimental observations (Figure 3A) [98, 99]. MHAT from [CoIII]–H to 1b proceeds via TS1 with Markovnikov selectivity (ΔG ‡ = 14.1 kcal/mol), generating secondary radical III and [CoII] (Figure 3B). Radical III undergoes 1,2‐RAM to form a more stable tertiary radical IV via a [2,3]‐shift ([2,3]‐TS2), favored over the [1,2]‐pathway ([1,2]‐TS3) by 0.2 kcal/mol (Figure 3, see also Table S9), consistent with 18O‐labeling studies. Oxidation of [CoII] by [IrIV(ppy)3]•+(NTf2 −) followed by coordination of aniline and collidine (X‐1) assisted deprotonation to form [CoIII]–NR2 is exergonic (ΔG = –12.1 kcal/mol). Competing pathways involving solvent cage collapse of radical pair [68] to V (19.0 kcal/mol) is computed to be less favorable under these conditions. Final NR2 transfer from [CoIII]–NR2 to IV proceeds through a RLT process (TS4, ΔG ‡ = 9.7 kcal/mol), affording the desired coupling product 3a in a highly exergonic step. (A). Computed reaction energy profile for the photoredox/cobalt/Brønsted acid–catalyzed formal 1,3‐hydroamination of allyl carboxylate 1a (M06/def2‐TZVP/SMD(THF)//B3LYP‐D3/6‐31G(d)−SDD). (B). 3D transition‐state structures for MHAT and NR2 transfer generated using CYLview [100]. (C). Activation barriers for the [2,3]‐ and [1,2]‐RAM pathways, with Gibbs free energies referenced to secondary radical III (M06‐2X/6‐311+G(d,p)/SMD(THF)//B3LYP‐D3/6–31G(d)). A proposed catalytic cycle consistent with experimental observations and literature precedents [96, 101, 102] is shown in Figure 4. Upon the irradiation by blue light, the excited‐state photocatalyst [IrIII]* reduces [CoII] to [CoI], which is protonated by HX‐1 to furnish [CoIII]–H along with [IrIV]NTf2 and collidine (X‐1) [96]. The resulting cobalt hydride engages allyl carboxylate 1 via MHAT through TS1, affording the β‐acyloxyalkyl radical III and [CoII] species. Radical III subsequently undergoes 1,2‐RAM through both TS2 and TS3 to generate the more stable radical IV [99]. Concurrently, oxidation of [CoII] by [IrIV]NTf2 forms [CoIII] intermediate, closing the photoredox cycle. Coordination of aniline to the resulting [CoIII] catalyst followed deprotonation regenerates HX‐1 and delivers the [CoIII]–NR2 species, which was detected in HR‐MS (Figure S22). The key C─N bond‐forming event is proposed to proceed via radical ligand transfer, wherein radical IV undergoes an RLT reaction with the [CoIII]–NR2 species through TS4 [ 79, 103, 104]. This step furnishes the remote 1,3‐hydroamination product 3 and regenerates [CoII] catalyst, completing the catalytic cycle. Proposed reaction mechanism of the photoredox/cobalt/Brønsted acid‐catalyzed 1,3‐hydroamination of allyl carboxylate. In summary, we establish a migration‐enabled strategy for remote C─N bond formation that expands both the reactivity and mechanistic understanding of cobalt‐salen‐catalyzed MHAT hydroamination. By merging MHAT with 1,2‐RAM, the method enables functionalization at positions inaccessible to conventional 1,2‐hydroamination, affording protected β‐acyloxy alkyl amines with broad functional‐group tolerance. Combined experimental and computational studies support a C─N bond‐forming step proceeding via RLT at a Co(III)─N intermediate, in contrast to commonly invoked Co(IV)/SN2 or carbocationic pathways. More broadly, this work highlights the viability of RLT processes at cobalt─N bonds in MHAT catalysis and provides a conceptual foundation for future radical hydrofunctionalization design. Author Contributions Arman Khosravi: conceptualization, investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, data curation, supervision. Ho Martin Yuen: methodology, writing – review and editing, data curation, visualization, investigation, formal analysis, software. Seth Fremin: investigation, data curation, methodology, writing – review and editing, software. Nuwan A. Pannilawithana: methodology, data curation, investigation, validation, visualization. Nol Supasueb: methodology, investigation, formal analysis, data curation, visualization. Yu Zhang: investigation, methodology, visualization, formal analysis, software. Peng Liu: investigation, methodology, software, data curation, project administration, writing – original draft, validation, writing – review and editing, resources, supervision, funding acquisition. Ming‐Yu Ngai: validation, methodology, conceptualization, investigation, funding acquisition, writing – original draft, data curation, supervision, resources, project administration, writing – review and editing. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Supporting File: anie73134‐sup‐0001‐SuppMat.pdf.