Tuan Anh Trinh,
Stanislav Cherempei,
Daniel S. Rampon and
Jennifer M. Schomaker
*
Department of Chemistry, University of Wisconsin-Madison, Madison, Wisconsin 53706, USA. E-mail: schomakerj@chem.wisc.edu
First published on 7th February 2025
We report a method for the site- and stereoselective intramolecular amination of electron-deficient heterobenzylic C–H bonds via silver-catalyzed nitrene transfer (NT). A silver complex supported by a tripodal piperidine-based ligand afforded excellent reactivity under mild conditions (up to 96% yield), site-selectivity (up to >20:
1), and diastereoselectivity (up to >20
:
1 dr) for the amination of heterobenzylic C–H bonds that reacted poorly with other metal-based catalysts for NT. Our catalyst proved highly amenable to substrates bearing diverse competing sites for functionalization, including complex molecules derived from pharmaceuticals and natural products. Ligand screening revealed the importance of scaffold rigidity, leading to the discovery of an analogous quinuclidine-based ligand that further improved the site-selectivity against tertiary C–H bonds in a handful of challenging substrates. The cyclic sulfamate products were readily converted into highly functionalized motifs containing nitrogen-based heterocycles and diverse functional groups. Mechanistic studies suggested a radical-based pathway displaying relatively low sensitivity towards the electronic profiles of the heterobenzylic C–H bond, which contributes to the excellent substrate tolerance of this method.
Arguably, one of the most efficient routes to introduce an amine functionality at a heterobenzylic bond is to convert the targeted C–H bond directly to a new C–N bond.4 While transition-metal catalyzed nitrene transfer (NT) is a powerful and well-established synthetic tool for aminating benzylic C–H bonds, an electron-poor N-heteroarene presents unique challenges. Firstly, the electrophilicity of a putative metal–nitrene species may render it polarity mismatched with an electron-deficient heterobenzylic C–H bond.5 A second challenge is the propensity of a nucleophilic nitrogen atom(s) of the heteroarene to serve as a ligand for the metal, leading to deactivation of the catalyst if the binding is irreversible. The heteroaromatic nitrogen atom(s) may also react with the electrophilic nitrene precursors and hypervalent iodine reagents that are commonly employed in NT reactions6,7 to generate undesirable N-imino heteroaromatic ylides8–10 and other pyridinium adducts.11 Unsurprisingly, precedents of NT into electron-deficient heterobenzylic C–H bonds are scarce (Scheme 1B). Che reported a single example of intramolecular amination into a benzylic C–H bond adjacent to a C3-substituted pyridine ring using an azide precursor and a catalytic iron-porphyrin/NHC complex; however, harsh conditions were required.12 Meggers described NT of a N-benzoyloxyurea-derived nitrene precursor into a (2-pyridyl)benzylic C–H bond using an iron–phenanthroline catalyst;13 however, no examples bearing the alkyl chain on C3 or C4 of the pyridine ring or other electron-poor heterocycles were reported. Zhang aminated electron-deficient non-heterobenzylic C–H bonds, where the metalloradical reactivity of a cobalt-porphyrin complex alleviates the polarity mismatch.14 Our group also disclosed one example of nitrene insertion into a heterobenzylic C–H bond using a silver/bis(oxazoline) catalyst with a carbamate precursor in moderate yield.15 Nonetheless, these examples do not reveal whether the highly site-selective amination of electron-poor heterobenzylic C–H bonds is achievable in the presence of inherently more reactive sites, including tertiary alkyl and other activated C–H bonds. On the other hand, oxygenations,16–25 hydrazination,26 halogenations,27–30 thiolations,27 and heteroalkylations31 at heterobenzylic C–H bonds have been reported, but rely on the polar activation of the resonance forms of 2- and 4-alkylpyridines. Meanwhile, strategies for reactions of 3-alkylpyridines are scarce, due to the lower acidity of their benzylic sites (Scheme 1C).30,32 These limitations underscore the need for reliable, alternative methods to directly transform a broad range of heterobenzylic C–H bonds into valuable functional groups, particularly amines.
Herein, we report a silver-catalyzed protocol for site-selective intramolecular aminations of electron-deficient heterobenzylic C–H bonds via NT with sulfamates (Scheme 1D). We hypothesized that sulfamate-derived nitrene precursors would exhibit lower sensitivity to electronic effects and higher sensitivity towards the bond dissociation energies (BDE) of both activated and unactivated C–H bonds, as compared to carbamates.33 This difference in the behavior of the nitrene precursor could drive selective reaction at the electronically disfavored, yet weaker, heterobenzylic C–H bonds. Our method tolerates diverse heterocycles with different substitution patterns and displays exceptional chemo- and stereoselectivities, with moderate-to-excellent site-selectivities against an array of highly reactive competing C–H bonds, overriding the reactivity preferences seen with traditional catalysts for nitrene transfer.
Entry | Ligand | Solvent | 2a yield | 3a yield | RSM |
---|---|---|---|---|---|
a Reactions were carried out on a 0.1 mmol scale unless otherwise noted. Yields and diastereomeric ratios were determined by 1H NMR analysis of crude product mixtures using 1,3,5-trimethoxybenzene as an internal standard. b 2 mmol scale, 0.2 M PhOMe. c Isolated yield. | |||||
1 | L1 | CH2Cl2 | 66% (syn![]() ![]() ![]() ![]() |
3% | 7% |
2 | L2 | CH2Cl2 | 82% (syn![]() ![]() ![]() ![]() |
4% | <1% |
3 | L3 | CH2Cl2 | 71% (syn![]() ![]() ![]() ![]() |
12% | <1% |
4 | L4 | CH2Cl2 | 54% (syn![]() ![]() ![]() ![]() |
30% | <1% |
5 | anti-L5 | CH2Cl2 | 86% (syn![]() ![]() ![]() ![]() |
3% | <1% |
6 | L6 | CH2Cl2 | 41% (syn![]() ![]() ![]() ![]() |
5% | 12% |
7 | anti-L5 | PhOMe | 96% (syn![]() ![]() ![]() ![]() |
<1% | 4% |
8 | anti-L5 | iPrOAc | 32% (syn![]() ![]() ![]() ![]() |
<1% | 60% |
9 | anti-L5 | PivCN | 61% (syn![]() ![]() ![]() ![]() |
<1% | 32% |
10 | syn-L5 | PhOMe | 95% (syn![]() ![]() ![]() ![]() |
<1% | <1% |
11b | syn-L5 | PhOMe | 79% (syn![]() ![]() ![]() ![]() |
<1% | <1% |
![]() |
To validate the exceptional reactivity of our optimal reaction conditions, results were benchmarked against selected catalysts commonly employed in NT reactions, particularly those utilizing sulfamate precursors. Treatment of 1a with previously reported rhodium-,37 copper-,38 iron-39 and manganese-catalyzed40 protocols resulted in poor yields of 2a/3a and significant amounts of recovered 1a (Table 2, (A)), suggesting that the corresponding catalysts were deactivated by substrate inhibition. The desired product 2a was obtained in either low site-selectivity (under Rh- and Cu-catalyzed conditions) or as a minor regioisomer (under Fe- and Mn-catalyzed conditions). Other reported NT protocols that employ sulfamoyl azides (with Zhang's cobalt-porphyrin catalyst)14 and ureas (with Meggers' iron–phenanthroline catalyst)13 also failed to provide the respective pyridine-containing products (Table 2, (B)), despite their excellent performances with related substrates as shown in Scheme 1B. Specifically, the former reaction presumably suffered catalyst deactivation, and the latter only resulted in decomposition of starting material.
With the optimized conditions in hand, we investigated the substrate scope of the reaction, starting with the heteroaromatic moiety (Table 3). Excellent yields and selectivities (>20:
1 rr, dr) were observed with a range of para-C6-electron-donating and withdrawing groups (2b–2f). Interestingly, benchmarking studies carried out with 2d and 2f using Rh, Cu, and Mn catalysts revealed that their reactivity was heavily dependent on the electronics of the heteroaromatic ring. The presence of a more electron-rich pyridine ring in 2d improves reactivity at the expense of mass balance, while the electron-poor pyridine ring of 2f erodes the desired site-selectivity in favor of the formation of 3f. These results highlight another advantage of our Ag-catalyzed protocol over other metals, as it is capable of achieving the efficient functionalizations of heterobenzylic C–H bonds with diverse electronic profiles.
A lack of steric shielding of the N-atom in substrates lacking C6 substitution on the pyridine ring, which is particularly problematic for the benchmarking catalysts, did not cause substrate inhibition of the Ag catalyst. The presence of a substituent at the meta-C5-position did not affect the reactivity (2g). Reactions also proceeded smoothly with ortho-C4-substitutions (2h–2i), despite the requirement for increased loadings of catalyst and oxidant to ensure complete conversion within a reasonable period of time. It is worth noting that for 2i, the reaction outcomes were still excellent in the presence of an electronically deactivating and sterically encumbered ortho-CF3 group. Conversely, the Rh and Mn catalysts favored the other regioisomer 3i, along with <10% yield of 2i, while the Cu catalyst gave lower yield and selectivities. The 4- and 2-alkylpyridines (2j–2k) were also aminated in excellent yields using the Ag catalyst; however, the 2-alkylpyridine 2k gave lower dr and required C6 substitution to improve the mass balance. The reaction also tolerated diverse electron-deficient heteroarenes of medicinal interest (pyrimidine, quinoline, pyrazolo[1,5-a]pyrimidine) (2l–2n) and pharmaceutical-derived substrates (2o–2p) to afford the corresponding products in good-to-excellent yields and selectivities.
The aliphatic side chain was varied to examine the site-selectivity in the presence of other types of competing C–H bonds (Table 4, top). The preference for heterobenzylic C–H amination was good-to-excellent against unactivated secondary C–H bonds of the same tether length, regardless of side-chain structures or statistical control (2q–2s). In some cases, decreased conversions and site-selectivities under standard conditions were overcome by lowering the reaction temperature (2q) or changing the solvent (2r). In the presence of competing tertiary C–H bonds of the same tether length, the desired site-selectivity decreased, but still favored the electron-poor C–H in most cases to give the corresponding products in moderate-to-excellent yields (2t–2aa). This contrasts to benchmarking studies employing other metal catalysts (1t–1v and 1x–1y). For instance, the use of Rh catalysis overwhelmingly favored amination at the tertiary sites with low yields (<10%) of the desired heterobenzylic amination. Under our Ag-catalyzed conditions, site-selectivity against non-cyclic tertiary C–H bonds was not affected by the position of branching in the alkyl chain and showed an increase from 1.9:
1 (2t and 2v) to 5.9
:
1 when a bulky, inductively withdrawing group is proximal to the tertiary C–H bond (2u). Notably, the formation of 2v was exclusively anti-stereoselective, as opposed to the syn-stereoselectivity in C3-branched products. Against tertiary C–H bonds in simple cyclic systems, the product distribution displayed a close correlation with the predicted BDEs.41 In these cases, the desired regioisomers were obtained in ratios of 2.1
:
1, 1
:
1.1, and 2.8
:
1 against cyclobutyl (2w), cyclopentyl (2x), and cyclohexyl (2y) tertiary C–H bonds, respectively. To our delight, the product ratios improved remarkably in more complex substrates. In the presence of a competing cyclohexyl tertiary C–H bond, 2z was almost exclusively obtained in 96% yield and excellent dr. Product 2aa was also obtained in good yield and site-selectivity (3.0
:
1) against a competing cyclopentyl tertiary C–H bond. On the other hand, the heterobenzylic amination was considerably less site-selective against simple allylic (2ab, 1
:
1.6 rr), propargylic (2ac, 1
:
1.6 rr), and benzylic (2ad, 1
:
3.3 rr) C–H bonds, which have similar bond strengths and are more electronically favored. Despite this persistent challenge, the desired products were still obtained in moderate yields that were significantly higher than those obtained with Rh2(esp)2. The reaction again proved to perform better with a complex substrate, as 2ae was obtained in good yield and site-selectivity (4.0
:
1) in the presence of a competing tertiary, allylic C–H bond. These results indicate that the Ag catalyst not only selects for weak C–H bonds, but also provides excellent steric discrimination when multiple sites of similar BDEs are present in the same molecule. Finally, products derived from a primary sulfamate (2af, 3 mmol scale) and a tertiary heterobenzylic C–H bond-containing precursor (2ag) were obtained in good yields.
a Isolated yields in parentheses. rr, dr > 20![]() ![]() |
---|
![]() |
To further improve the site-selectivity for amination of heterobenzylic against tertiary C–H bonds via catalyst control, another round of catalyst screening was conducted on substrate 1t under the standard reaction conditions (see Table S2 in the ESI† for further details). In terms of the silver salt, AgBF4 improved the yield of 2t from 56% to 64% and thus was used for subsequent ligand screening. We attributed this improvement to the further attenuation of electrophilicity of the silver–nitrene complex that results from a stronger electrostatic interaction between the cationic silver–ligand complex and the smaller BF4 anion, which in turn favors NT at the less electron-rich C–H bond. After screening a variety of tris(pyridyl)amine-type ligands, we found conformational rigidity had a critical impact on both the overall yield and product ratio. Interestingly, while the ligands with a non-cyclic scaffold (L1 and L2) resulted in a decreased 2t/3t ratio of 1.3:
1, simply removing the pendant methyl group from syn-L5 led to an even more significant loss of site-selectivity (Table S2†). Attempts to replace the methyl group with larger groups were unsuccessful; thus, we envisaged replacing the piperidine backbone with a quinuclidine scaffold (L7) to enhance the conformational rigidity of the ligand. L7 contains a (syn,syn,anti) relative stereochemistry of the pyridine arms, rendering its coordination geometry comparable to that of syn-L5 (Table 3, bottom). Despite slightly lower conversions as compared to those observed with AgOTf/syn-L5 at the same reaction times, AgBF4/L7 delivered comparable product ratios among the simple substrates, with a noticeable improvement in the reaction of 1x to give 2x as the major regioisomer. We were pleased to observe other striking improvements in site-selectivity when AgBF4/L7 was employed with more complex substrates; for example, the selectivity for 2aa and 2ae increased to 4.4
:
1 and 4.6
:
1, respectively (Table 3, bottom). We surmised that the decreased flexibility of the coordination sphere associated with L7 raises the energy barrier for nitrene insertion into the more sterically congested tertiary site to a larger extent as compared to reaction at the heterobenzylic site. However, attempts to facilitate steric discrimination via the installation of bulky substituents, such as mesityl, on the pyridine rings of syn-L5, reversed the site-selectivity for 2t to favor tertiary insertion in 3t (Table S2†). This reversal was attributed to intramolecular repulsion and displacement of at least one of the ligand arms of the Ag complex from the substrate, deactivating the heterobenzylic C–H bond towards NT. This result further underscores the importance of scaffold rigidity to balance steric discrimination and stabilization of metal–ligand interactions. To probe the effects of substrate–ligand non-covalent interactions on inducing the site-selectivity,42 screening of aromatic solvents across a wide range of dielectric constants was performed using AgBF4/syn-L5 as the catalyst (see Table S2 in the ESI† for further details). An inverse correlation between the dielectric constants and the observed site-selectivity ratios was observed. We speculate that the excellent reactivity and selectivity resulting from the use of anisole as a solvent are due to both its superior solubilizing effects and relative non-polarity compared to dichloromethane and other aromatic solvents. The latter factor presumably enhances the attractive π–π and dispersion interactions between the heterocyclic substrate and ligand that are essential to inducing the site-selectivity for amination at the heterobenzylic C–H bond.42
The heteroarene-containing sulfamates are convenient building blocks; examples of post-functionalizations are shown in Scheme 2. The nitrogen of 2af was protected with a Cbz group to afford 4 in 61% yield. This intermediate can readily be ring-opened with various nucleophiles, including water, azide, morpholine and fluoride, to obtain the corresponding 1,3-functionalized products (5–8) in moderate-to-excellent yields (Scheme 2A). Reactions of disubstituted cyclic sulfamate 2t required more forcing conditions; Cbz-protection followed by treatment with AcOK in DMSO at 80 °C furnished N-Cbz-1,3-aminoketone (9) in 27% yield over 2 steps (Scheme 2B). This sequence can potentially provide efficient access to all four diastereomers of the corresponding 1,3-aminoalcohol, depending on the absolute stereochemistry of the C–O bond before the aminative cyclization step and after reduction of the ketone resulting from the deprotonative ring-opening step.
To gain insight into the potential mechanistic differences of this chemistry compared to previous NT systems, a Hammett study was conducted with substrates containing competing heterobenzylic C–H bonds (1ah–1ao, Scheme 3A). A reasonably linear correlation (R2 = 0.93) between log(kR/kH) and the σ+ Hammett constants was noted, with a ρ-value of −0.49. This ρσ+-value is less negative than those of the benchmarked sulfamate-based NT reactions (Ag(L1)OTf:42 −0.69; Rh catalyst:43 −0.55; Fe catalyst:40 −1.12; Mn catalyst:40 −0.88) and several other systems employing diverse catalysts and nitrene precursors (between −1.49 and −0.66).44–51 This result suggests that the Ag-catalyzed reaction has a relatively low sensitivity to the electronic effects on the heterobenzylic C–H bond. To determine whether the reaction undergoes a concerted or stepwise mechanism, radical trapping experiments were conducted. After subjecting 1a to the standard reaction conditions in the presence of 1 equivalent of TEMPO, 1 equivalent of butylated hydroxytoluene (BHT) or 0.5 equivalent of 9,10-dihydroanthracene (DHA), respectively (Scheme 3B), substantial losses of yield and mass balances were noted, although detection and isolation of any trapped products proved challenging under the reaction conditions. Notably, these results contrast with our previous observations that addition of DHA does not affect the amination of non-heteroaromatic benzylic C–H bonds using Ag(L5)OTf.35 While these results imply a heterobenzylic radical intermediate results from hydrogen atom transfer (HAT) to the Ag-nitrene species, we did not rule out a possibility that the radical scavengers impact the outcome by reacting with the oxidant. To further probe the intermediacy and relative lifetime of a heterobenzylic radical, a stereoretention experiment was performed with 1ap, which contains an enantioenriched tertiary heterobenzylic C–H bond (90% ee) (Scheme 3C). After complete conversion, the corresponding product 2ap was obtained with a significant loss of ee (47% ee). The partial racemization indicates the reaction indeed proceeds via a stabilized heterobenzylic radical intermediate following the initial HAT step, which then racemizes at a non-negligible rate relative to that of a product-forming radical rebound (RR) step. This result differs from our mechanistic findings in previous Ag-catalyzed NT systems, where the reactions are proposed to undergo a stepwise radical pathway involving barrierless RR,42 evidenced by the observation of complete stereoretention.35 In a radical clock experiment (Scheme 3D), small amount(s) of cyclopropane-ring opening product(s) were detected when a cyclopropane moiety is adjacent to the heterobenzylic site, further corroborating the intermediacy of a heterobenzylic radical.
In conclusion, we have developed a highly chemo-, site-, and stereoselective method to aminate electron-deficient heterobenzylic C–H bonds via silver-catalyzed intramolecular nitrene transfer. Key to enabling this challenging transformation is the catalyst-controlled alleviation of polarity mismatch and circumvention of substrate inhibition. Optimization of ligand design revealed the importance of conformational rigidity to improving the site-selectivity against other reactive C–H bonds. This protocol demonstrates exceptional tolerance of several heteroaromatic compounds, including those derived from drugs and natural products that represent a coveted yet elusive chemical space for traditional metal catalysts.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4sc08757g |
This journal is © The Royal Society of Chemistry 2025 |