Iron-mediated intermolecular N-group transfer chemistry with olefinic substrates

The dipyrrinato iron catalyst reacts with organic azides to generate a reactive, high-spin imido radical intermediate, distinct from nitrenoid or imido species commonly observed with low-spin transition metal complexes. The unique electronic structure of the putative group-transfer intermediate dictates the chemoselectivity for intermolecular nitrene transfer. The mechanism of nitrene group transfer was probed via amination and aziridination of para -substituted toluene and styrene substrates, respectively. The Hammett analysis of both catalytic amination and aziridination reactions indicate the rate of nitrene transfer is enhanced with functional groups capable of delocalizing spin. Intermolecular amination reactions with ole ﬁ nic substrates bearing allylic C – H bonds give rise to exclusive allylic amination with no apparent aziridination products. Amination of substrates containing terminal ole ﬁ ns give rise exclusively to allylic C – H bond abstraction, C – N recombination occurring at the terminal C with transposition of the double bond. A similar reaction is observed with cis - b -methylstyrene where exclusive amination of the allylic position is observed with isomerization of the ole ﬁ n to the trans - con ﬁ guration. The high levels of chemoselectivity are attributed to the high-spin electronic con ﬁ guration of the reactive imido radical intermediate, while the steric demands of the ligand enforce regioselective amination at the terminal position of linear a -ole ﬁ ns.


Introduction
The prevalence of nitrogen functionalities in bioactive molecules and materials has inspired the development of numerous techniques for direct C-H bond amination.Transition-metal catalyzed approaches 1,2 offer a unique opportunity to control the reactivity and selectivity of nitrene transfer.Careful modulation of metal identity and ligand design can drastically dictate the degree of bonding and electronic structure of the metalbound nitrene intermediates, ranging from simple oxidantmetal adducts, to singlet and triplet stabilized nitrenoids, to closed-and open-shell metal imidos (Fig. 1).In particular, metal stabilized nitrene transfer using noble metal catalysts, specically dirhodium, 3 have been successfully applied in the context of complex molecule synthesis.These well-behaved catalysts have been extensively studied and are believed to transfer functionality to both olens and C-H bonds via a singlet nitrenoid intermediate (Fig. 1b) in a concerted asynchronous mechanism. 4espite the success of the dirhodium class of catalysts, a renewed interest in late, rst-row transition metal nitrenoid catalysts, composed of copper, [5][6][7][8][9] cobalt, 10,11 manganese, [12][13][14][15] nickel, 16,17 and iron 12,18,19 has emerged.Their high d-electron count and compressed ligand elds create a unique opportunity for ancillary ligands to strongly inuence the bonding interactions between the nitrene moiety and the metal center.For instance, the binding of strong eld ligands destabilizes metalligand antibonding orbitals and promotes low-spin electronic congurations.Similar to their second-and third-row congeners, closed-shell rst-row transition metals will interact with nitrene precursors to generate either a metal-bound singlet nitrenoid (Fig. 1b) or a covalently-bound metal imido (Fig. 1c).Alternatively, weak eld ligand environments favor population of metal-ligand antibonding orbitals and high-spin electronic congurations.Communication between highly paramagnetic metal centers and N-group transfer reagents has the potential to translate radical character directly to the N-atom or along the metal-nitrogen bond vector (Fig. 1d or e).
Given the range of accessible metal-nitrene electronic congurations, it is not surprising that the established catalysts exhibit dramatically different reactivity, particularly when considering the functionalization of olenic substrates (Fig. 1).For instance, exposure of a macrocyclic tetracarbene-ligated Fe-complex exclusively catalyzes the aziridination of aliphatic olens, 19 while a phthalocyaninato Fe-complex facilitates preferential intramolecular allylic C-H amination, 18 and an Fe-porphyrin complex generates mixtures of both aziridine and allylic amine products. 12While each of these complexes is conned to a tetragonal eld by tetra-chelating macrocyclic ligands, the ligand eld strength, metal oxidation state, and nitrene source vary and are likely responsible for the changes in chemoselectivity.However, in each of these examples, the structure and electronic conguration of the reactive intermediate are only posited, as is the case for nearly all C-H amination catalysts.
Ideally, one would like to probe the inuence a given metal, ligand, and nitrene unit bears on reactivity and selectivity through the isolation of the reactive metal-nitrenoid intermediate.We previously reported 20 intermolecular amination and aziridination with aryl and alkyl azides using ferrous dipyrrinato complexes ( R L)FeCl(Et 2 O) [R ¼ 2,4,6-Ph 3 C 6 H 2 (Ar), 1-adamantyl (1), t Bu (2)] (Fig. 2).Isolation and characterization of the radical imido complex ( Ar L)FeCl(N(p-t BuC 6 H 4 )), wherein a highspin Fe(III) (S ¼ 5/2) is antiferromagnetically coupled to the imido radical (S ¼ -1/2) to give an overall high-spin ground state (S ¼ 2), suggests a similar electronic structure for the proposed iron-bound imido radical 3.Such an electronic structure places signicant radical character on both the Fe-N s and p bond vectors, facilitating both H-atom abstraction and radical recombination pathways.In this report, we have explored the implications of the reactive intermediate's unique electronic structure on both the reaction mechanism and chemoselectivity of intermolecular N-group transfer.We have established the catalyst's strong preference for allylic amination over aziridination with aliphatic olen substrates.Additionally, we report on the functionalization of a-olens to linear allylic amines with outstanding regioselectivity.

A. Benzylic amination and aziridination
We previously reported 20 that 1 catalyzes nitrogen group transfer to styrene and toluene with 1-azidoadamantane, resulting in the formation of 1-adamantyl-2-phenylazidridine (85% yield) and benzyladamantylamine (60% yield) respectively.In attempts to expand the substrate scope, a series of para-substituted toluenes and styrenes were subjected to 1 (5-10 mol%) and 1-azidoadamantane (Table 1).All substrates successfully underwent the desired intermolecular nitrogen group transfer, demonstrating that catalysis is amenable to substrates with both electron withdrawing and donating groups and ethereal C-H bonds (Table 1, entries 4 and 8).The yields for aziridination at room temperature are overall higher than that for amination (75-85% vs. 46-60%); this trend correlates with the lower propensity for the aziridine product to bind to the catalyst and retard the rate of catalysis.Amination yields can be improved upon heating to 60 C (entry 1) to encourage product dissociation from the electrophilic Fe center.
Styrene aziridination is amenable to a variety of organoazides.Exposure of 1 and styrene to 5 equivalents of tert-butyl azide (Table 1, entry 10) resulted in the corresponding aziridine in good yield (86%).The yield for 1-azidobutane drops signicantly (entry 11, 53%), likely due to imide decomposition pathways.In general, primary and secondary alkyl azides rapidly react with 1 to generate the imide intermediate, but are then prone to undergo either intramolecular C-H amination to yield N-heterocycles or a-hydrogen abstraction to yield linear imines 21 rather than intermolecular group transfer chemistry.Copper 9 and iridium 22 complexes also generate imines upon exposure to primary and secondary azides through nitrenoid intermediates.Aryl azides are also compatible with the aziridination reaction.Para-substitution of the aryl ring eliminates a bimolecular imido decomposition pathway, and noticeably improves the yield of aziridination (entry 12 vs.entry 13).Unfortunately application of either sulfonyl (entry 14) or silyl (entry 15) azides does not result in productive N-group transfer but rather catalyst decomposition through unproductive H-atom abstraction or ligand amination pathways, respectively.

B. Mechanism of benzylic amination and aziridination reactions
We previously reported that the mechanism of catalytic toluene amination and styrene aziridination proceeds via one-electron pathways, largely based on the non-classical KIE value (12.8) for toluene amination and the high-spin electronic structure of the putative intermediate 3. Following this initial report, we have conducted a series of labelling and Hammett studies to further corroborate a mechanistic proposal in which short-lived carboradical intermediates are formed en route to the functionalized products.
Exposure of 1 and 1-azidoadamantane to excess cis-b-deuterostyrene resulted in the isolation of cis-1-adamantyl-2-deutero-3-phenylaziridine, with no evidence of the trans isomer (Scheme 1).While isomerization of the cis-olen to a mixture of cis-and trans-aziridines is suggestive of a stepwise mechanism, lack of isomerization is less denitive.Either the reaction is occurring via a concerted mechanism or a two-step mechanism with radical recombination out-competing C-C bond rotation.The adamantyl units on both the dipyrrin and imido generate a very narrow cle in which the substrate can orient itself.The crowded environment likely impedes C-C bond rotation and renders radical recombination more facile.However, we cannot rule out a concerted aziridination mechanism with this experiment alone.
The electronic nature of the transition state was further studied by conducting a Hammett analysis of both the aziridination and amination reactions.The relative rates of catalytic aziridination and amination of a series of para-substituted styrenes and toluenes using 1 with one equivalent of 1-azidoadamantane were determined through pairwise competition experiments.Equimolar mixtures of styrene or toluene and a para-substituted styrene or toluene derivative, respectively, were treated with 1-azidoadamantane in the presence of a stoichiometric amount of the Fe complex 1.The ratios of the resulting aziridines and benzylamines were determined by GC/MS analysis and by 1 H NMR integration of the spectra taken on unpuried material.The results are summarized in Table 1.
In all cases, k R (k X /k H ) slightly exceeds one, suggesting that the rate of amination and aziridination is marginally enhanced with substitution by both electron donating and electron withdrawing groups.Plotting log k R versus s + parameters (Fig. 3a) results in a non-linear correlation between reaction rate and electron-donating ability of the para-substituent.Linear correlations are consistent with cationic charge stabilization in the transition state and typically observed in concerted amination chemistry with Rh 2 -carboxylate catalysts. 4The lack of correlation with our catalyst led us to consider the incorporation of a spin delocalization parameter, s c JJ , developed by Jiang and coworkers. 23The log k R vs. (s mb , s c JJ ) plot (Fig. 3b) gave rise to an appreciably more linear relationship (R 2 ¼ 0.99, 0.93 for amination and aziridination respectively).Multiple coefficient linear regression for the dual-parameter equation

C. Allylic amination
Next, the reactivity of a series of olens with 2 and 1-azidoadamantane was investigated (Table 1, entries 16-20) in order to determine the preference for aziridination over allylic C-H bond amination.Reaction of 2 with cis-b-methylstyrene (Table 1, entry 16) resulted in the exclusive formation of N-cinnamyladamantan-1-amine with complete isomerization to the trans olen.Cyclic alkenes cyclohexene and cyclooctene similarly resulted in allylic amination products in comparatively poorer yields (17% and 18%, respectively).In fact, the stoichiometric reaction of 2 with 1 equivalent of 1-azidoadamantane in cyclohexene resulted in only 12% of the allylic amine.While some azide remained unconsumed, a signicant amount of radically coupled bi-cyclohexenyl product was formed (as determined by GC/MS).Exposure of 2 and 1-azidoadamantane to linear, non-styrenyl terminal olens resulted in exclusive formation of the linear allylic amine products (Table 1, entry 19 and 20) as a mixture of cis and trans isomers.Stoichiometric application of 2 to 1-azidoadamantane at room temperature in excess 1-hexene resulted in the formation of the allylic amine N-adamantyl-N-hex-2enylamine in 77% yield, with the remaining mass balance belonging to 1-aminoadamantane and 1,2-diadamantyldiazene.Lowering the catalyst loading (20 mol%) and extending the reaction time to 24 hours resulted in only a slight increase in product formation (29%) and a signicant portion of unreacted azide remains (64%).As the reaction temperature is increased, product dissociation from the Fe center is encouraged, allowing for catalyst turnover and increased yields (29% at 25 C versus 43% at 60 C) but without reaching full conversion.At elevated temperatures (100 C), the azide is completely consumed, yet the yield of the desired N-adamantyl-N-hex-2-enylamine is diminished.We suspect that free nitrene forms via thermal decomposition of 1-azidoadamantane under these conditions, generating signicant amounts of diazene and 1-aminoadamantane.Additionally, several new products corresponding to N-group transfer to 1-hexene appear in the GC/MS trace at this temperature.Importantly, the linear allylic amine is not formed to any extent in the absence of catalyst 2, regardless of the reaction temperature.
Under optimized conditions, we explored the reactivity of a terminal, linear diene (entry 20).Again, primary amination products were obtained in moderate yield favoring the trans olen isomer.Diamine formation was not observed due to the large excess of diene substrate.No reaction is observed with substrates lacking allylic C-H bonds (e.g.tert-butylethylene).
While the method has synthetic limitations, the interesting chemo-and regioselectivity of linear allylic amination inspired us to investigate the mechanism of this transformation.Typically, intermolecular transition metal-catalyzed allylic amination does not proceed with exclusive olen transposition. 29owever, precedence for this transformation has been demonstrated previously using simple Fe II and Fe III halide salts 30 and Fe pthalocyanine complexes in the presence of phenylhydroxylamine. 31 Recently, a hetero-bimetallic Pd and Cr catalytic protocol has also been developed to affect a similar amination reaction. 27e have proposed two possible mechanistic pathways (Fig. 4).Both mechanisms begin with formation of the reactive imido radical intermediate 3, consisting of three bulky adamantyl units (two on the dipyrrinato ligand, one on the imido N).The olen then approaches the sterically crowded reaction centre trans to the Cl ligand.As such, heightened reactivity is observed with the least bulky, mono-substituted olen substrates (Table 1, entries 19 and 20).In pathway A (Fig. 4a), approach of the olen induces C-N bond formation at the primary carbon with homolytic cleavage of the p-bond.Next, the carboradical rapidly recombines with the Fe-bound amide to generate an aziridine product.Finally, under the reaction conditions, the aziridine is ring opened to the primary allylamine through either a heterolytic (as drawn) or homolytic pathway.In pathway B, the Fe-imido intermediate abstracts an H-atom from the olen substrate to generate an allylic radical, followed by radical recombination at the terminal carbon.
Analysis of the reaction with cis-b-methylstyrene was used to distinguish between pathways A and B (Fig. 4b).If pathway A were operative, initial aziridination would generate 1-adamantyl-2-methyl-3-phenylaziridine.In situ aziridine ring opening would then install the amino group at the benzylic carbon.If pathway B were operative, radical recombination would occur regioselectively at the less-hindered terminal position of the unsymmetrical allylic radical to generate N-cinnamyladamantan-1-amine.As discussed earlier, the linear allylic amine is the only product of the reaction, suggesting that pathway B is operative. 32This is an appealing mechanistic possibility, as it correlates well with our proposed mechanism for the benzylic amination of toluenes.

Conclusions
In line with previous studies of this system, we have determined that one-electron reaction pathways dominate intermolecular N-group transfer chemistry when mediated by the dipyrrinato ferrous complexes.As such, an innate preference for amination over aziridination has been established for a range of allylic C-H bond containing substrates.We attribute the extremely high levels of chemoselectivity to the high-spin electronic conguration of the reactive imido radical intermediate 3. Further, the steric demands of our ligand coupled with the bulky adamantyl imido fragment enforce regioselective amination at the terminal position of linear a-olens.The preservation of unsaturation following imido transfer furnishes products that are amenable to subsequent functionalization.

Fig. 1
Fig.1Available reactive intermediates and product distributions for transition metal-based nitrogen group transfer chemistry with olefinic substrates.

a
All reactions run with substrate as solvent.b Reaction with specied substrate and 1 (10 mol%) and AdN 3 at 23 C for 12 h.Yields determined by 1 H NMR, using ferrocene as an internal standard.c Ratio determined by 1 H NMR from single turnover competition experiment run with 1 and AdN 3 (1 equiv.) in 1 : 1 substituted : parent toluene or styrene.d Reaction run at 60 C. e Reaction run with 5 mol% 1. f Reaction with styrene and 1 (20 mol%) and specied azide at 23 C for 12 h.Yields determined by 1 H NMR, using ferrocene as an internal standard.g Yields were determined by GC/MS ( 1 H NMR). h Reaction with specied substrate and 2 (100 mol%) and AdN 3 at 23 C for 12 h.i Reaction with specied substrate and 2 (20 mol%) and AdN 3 at 23 C for 12 h.j Reaction with specied substrate and 2 (20 mol%) and AdN