Iron complexes of tetramine ligands catalyse allylic hydroxyamination via a nitroso–ene mechanism

Iron(II) complexes of the tetradentate amines tris(2-pyridylmethyl)amine (TPA) and N,N′-bis(2-pyridylmethyl)-N,N′-dimethylethane-1,2-diamine (BPMEN) are established catalysts of C–O bond formation, oxidising hydrocarbon substrates via hydroxylation, epoxidation and dihydroxylation pathways. Herein we report the capacity of these catalysts to promote C–N bond formation, via allylic amination of alkenes. The combination of N-Boc-hydroxylamine with either FeTPA (1 mol %) or FeBPMEN (10 mol %) converts cyclohexene to the allylic hydroxylamine (tert-butyl cyclohex-2-en-1-yl(hydroxy)carbamate) in moderate yields. Spectroscopic studies and trapping experiments suggest the reaction proceeds via a nitroso–ene mechanism, with involvement of a free N-Boc-nitroso intermediate. Asymmetric induction is not observed using the chiral tetramine ligand (+)-(2R,2′R)-1,1′-bis(2-pyridylmethyl)-2,2′-bipyrrolidine ((R,R′)-PDP).


Allylic amination reactions
As an extension of our previously reported iron-catalysed allylic oxidation of cyclohexene (7) [45][46][47], we wished to explore potential C-N bond formation at this position using iron catalysis. Combining cyclohexene (7, in excess) with N-Bochydroxylamine (8) as the nitrogen source and the iron complex FeTPA (4) or FeBPMEN (5) afforded a mixture of products: the allylic hydroxylamine 9 alongside the Fenton oxidation products alcohol 10 and ketone 11 [53], and a small amount of tertbutyl carbamate (12, Scheme 1). Initial reactions under an argon or air atmosphere returned product mixtures in the ratios shown in Table 1.  Table 1.
Under an argon atmosphere, the allylic hydroxylamine 9 was produced in ~10% yield with either ligand; performing the reaction open to air lifted the yield of the allylic hydroxyamination product 9 as high as 40%, but also substantially increased yields of 10 and 11 (Table 1).
Control experiments using just the metal salt or each of the ligands on their own returned trace quantities of product 9 and varying levels of Fenton-type pathways (Table S3, Supporting Information File 1), confirming that FeTPA (4) and FeBPMEN (5) are active agents in promoting allylic hydroxyamination of cyclohexene.
The effect of catalyst loading was screened under an air atmosphere, since initial results indicated that better yields of 9 are obtained under air than argon. Thus cyclohexene (0.7 mL, 7 mmol, 100 equiv) was added to a solution of catalyst 4 or 5 (1-20 mol %) and BocNHOH (70 μmol, 1 equiv) in CH 3 CN (Table S4, Supporting Information File 1). Lowering the catalyst loading of FeTPA from 10 to 5 mol % led to a small Table 1: Catalytic allylic amination of cyclohexene (Scheme 1). Reaction conditions: catalyst 4 or 5 (7 μmol) and cyclohexene (0.7 mL, 7 mmol) were dissolved in CH 3 CN (total volume 10 mL) and stirred at room temperature under air or argon atmosphere while BocNHOH (8, 70 μmol) was added, then stirring was continued overnight (18 h increase in the yield of allylic hydroxylamine 9 with a significant decrease in the appearance of allylic oxidation products 10 and 11. The amount of FeTPA (4) could be further lowered to 2 and 1 mol %, bringing further small increases in the yield of 9.
However increasing loading of FeTPA (4) to 20 mol % halts the amination reaction, returning only allylic oxidation products 10 and 11. In contrast, changing the catalyst loading of FeBPMEN (5) up or down from 10 mol % lowers yields of 9; at 1 mol % or 20 mol % loading of catalyst 5, increased levels of 10 and 11 are observed, but at 5 mol % catalyst 5, the yields of all three oxidation products are diminished. Clearly the competing hydroxyamination and Fenton reaction pathways are sensitive to the amount of catalyst used relative to BocNHOH; optimum catalyst loadings are 1 mol % for 4 and 10 mol % for 5.
Nicholas and Kalita have reported that the addition of hydrogen peroxide can improve yields in their copper-catalysed allylic amination reactions using BocNHOH [41]. Thus the addition of hydrogen peroxide (1:1 relative to BocNHOH) to reactions with 4 or 5 was investigated. Using a 1:1:1 ratio of cyclohexene:BocNHOH:H 2 O 2 with FeTPA (4) at 1 mol %, allylic hydroxylamine 9 was formed in only 4% yield, with the allylic oxidation products 9 and 10 predominant. This is not unexpected given the propensity of hydrogen peroxide to react directly with iron complexes to produce 10 and 11 via Fentontype pathways [47,53].
We have previously observed solvent-dependent behaviour by non-heme iron complexes when mediating oxidation of cyclohexene in methanol versus acetonitrile as solvent [46,47]. Using methanol as solvent in the allylic amination reactions with FeTPA (4, 1 mol %) and cyclohexene in excess, yields of allylic hydroxylamine 9 dropped: 9 was formed in 10% yield (vs 27% in acetonitrile), while yields of allylic oxidation products 10 and 11 were also lowered, to 25% and 38% respectively (vs 54% and 36% in acetonitrile). BocNH 2 (12) was not observed. Using FeBPMEN (5, 10 mol %) in methanol, 10 and 11 were formed but target compound 9 was not observed at all. Presumably with methanol as solvent, the oxidising power of iron:ligand system is partially redirected to oxidise the solvent.
With a view to improving the synthetic potential of this reaction, the transformation was attempted at a 1:1 ratio of BocNHOH to cyclohexene. Thus BocNHOH (70 μmol), FeTPA 4 (1 mol %) and cyclohexene (70 μmol) were combined in acetonitrile and stirred for 18 hours at room temperature, open to the air. Allylic hydroxylamine 9 was formed in 6% yield; allylic oxidation products 10 and 11 were each observed in ≤1%. Reaction at 2:1 BocNHOH:cyclohexene did not significantly improve the yield of allylic amine 9 (7%). Similar results were obtained using FeBPMEN (5, 10 mol %) as catalyst, which yielded small amounts of 9 (8%) and 10 (2%) but not ketone 11. In their work using copper(I) iodide to catalyse similar reactions, Iwasa et al. conducted reactions at much higher concentrations of hydroxylamine and alkene (0.5 mmol BocNHOH and 0.75 mmol alkene in a total reaction volume of 1 mL) [40]. With this in mind, the FeBPMEN (5) reaction was repeated at 10-fold higher concentration (i.e., 1:1 BocNHOH/cyclohexene in a total reaction volume of 1 mL). Under these conditions the yield of allylic amine 9 doubled relative to the more dilute 1:1 reaction, to 17%; 10 and 11 were not observed.

Reaction using a chiral catalyst
The chiral catalyst Fe(R,R′)-PDP (6) has been used previously to promote asymmetric C-H oxidation reactions [51]. With a view to developing an asymmetric iron-catalysed allylic hydroxyamination reaction, catalyst 6 was prepared and used to effect conversion of cyclohexene 7 to hydroxylamine 9. This reaction afforded 9 in 13% yield, but only as the racemate: analysis by chiral GC (CP-Chirasil-Dex CB column) revealed two peaks with equal peak areas (t R = 8.6 and 8.8 minutes); the same peaks in the same ratio were observed using a reference sample of racemic 9.

Scheme 2:
Proposed mechanism for hydroxyamination of cyclohexene (7) by FeTPA (4) and FeBPMEN (5): (a) iron-mediated oxidation of BocNHOH (8) by O 2 affords the nitroso-species 13, which (b) undergoes an ene reaction with the alkene substrate; (c) an alternative disproportionation reaction to convert 8 to 13 can occur without involvement of O 2 and also generates BocNH 2 (12), observed as a byproduct at low levels (Tables 1, S3 and S4, Supporting Information File 1). The mechanistic evidence gathered to date suggests the formation of a free nitroso species and 'off-metal' reaction with the alkene.
Several groups have recently reported efficient methods for the asymmetric hydroxyamination of carbonyl compounds using acylnitroso species generated in situ along with chiral oxazolines [16,17], N-oxides [19] or amines [18,20,21] as ligands or organocatalysts. In these nitrosoaldol contexts, the chiral agents induce asymmetry by virtue of their influence over the enolate reaction partner. Achieving asymmetric induction in the nitroso-ene reaction is a trickier proposition [14], although this has been demonstrated in an intramolecular context [22].

Mechanistic studies
Previous studies of iron-promoted allylic amination reactions with N-phenylhydroxylamine, and copper-catalysed reactions with N-Boc-hydroxylamine (8) return regio-and chemoselectivity profiles that are consistent with reaction via nitroso-ene mechanisms [35,36,38,41]. Thus we hypothesised that the hydroxyamination reactions mediated by FeTPA (4) and FeBPMEN (5)  Nitroso species can be trapped by hetero-Diels-Alder reaction with dienes [54,55], and detection of the resulting hetero-Diels-Alder adducts used to confirm the formation of free nitroso intermediates. Thus trapping experiments were conducted using isoprene (14) to investigate the formation of a nitroso species in this reaction. First a 1:1 mixture of cycloadducts 15 and 16 was synthesised as a reference sample using Kirby's conditions for the hetero-Diels-Alder reaction (N-Boc-hydroxylamine (8), isoprene (14) and sodium periodate, Scheme 3a) [54,55]. Then isoprene (14) and N-Boc-hydroxylamine (8)  ratio (Scheme 3b); the observation of 15 and 16 in this reaction confirms that a Boc-nitroso species 13 is formed in the FeTPA/ FeBPMEN-catalysed reaction. It is interesting to note that the nitroso-ene product is not generally observed under the Kirby conditions, as the allylic hydroxyamination product 17 is unstable in the highly oxidising environment rendered by sodium periodate [56]. Observation of this product in the FeTPA-and FeBPMEN-mediated reaction of isoprene indicates the relative mildness of these conditions for nitroso formation.
Nicholas et al. have reported a similar experiment to study the iron(II,III) chloride-catalysed reaction of N-phenylhydroxylamine with 2,3-dimethyl-1,3-butadiene, in which the Diels-Alder cylcoadduct was not observed, and only the allylic amination product was formed [36,38]. Conversely Jørgensen and Johannsen report that N-phenylhydroxylamine and 1,3cyclohexadiene in the presence of iron(II) phthalocyanine do form the Diels-Alder cycloadduct [35]. Although the outcomes with the different catalysts were contrasting, only one product was observed in each case. Cenini et al. have reported that both cycloadduct and ene product are formed in the ruthenium-catalysed reaction of nitrobenzene (an alternate route to a nitrosobenzene intermediate) with isoprene [57]. As mentioned above, isoprene produces two different regioisomers in the hetero-Diels-Alder reaction with nitroso compounds, yet Cenini et al. only detected one isomer. From this observation they concluded that their Ru-catalysed amination reaction and the Diels-Alder reactions were occurring 'on metal' without generation of a free nitroso species [57].
In contrast, the FeTPA-mediated reaction generates both BocNO cycloadducts of isoprene along with the amination product. Furthermore, the reaction with the chiral system 6 renders zero asymmetric induction. Thus we conclude that the Fe-TPA/BPMEN reaction involves the nitroso-ene reaction of a free nitroso intermediate.

Conclusion
FeTPA (4) and FeBPMEN (5) are established catalysts for the hydroxylation, dihydroxylation and epoxidation of hydrocarbon substrates [48,[58][59][60]. In this study we have shown that they can also catalyse the allylic hydroxyamination of alkenes with N-Boc-hydroxylamine. Mechanistic investigations suggest the involvement of a free nitroso species which undergoes a nitroso-ene reaction with the alkene. The intermediacy of a free nitroso species means that asymmetric induction is not observed in reactions with the chiral catalyst Fe(R,R′)-PDP (6).

Experimental General experimental
All commercially available reagents were used without purification unless otherwise specified. Solvents for extraction and chromatography were distilled before use. Solvents for reactions were freshly distilled immediately prior to use. Tetrahydrofuran (THF) was dried over sodium wire and benzophenone. Dichloromethane and acetonitrile were dried over calcium hydride. Acetonitrile was degassed using three freeze-thaw cycles when it was to be used in an atmosphere of argon. Methanol (MeOH) was dried over magnesium methoxide. Alkenes used in allylic amination reactions were passed through a micro-column of neutral alumina immediately before use. Water was purified using a Millipore purification system. Analytical thin-layer chromatography (TLC) was performed using preconditioned plates (Merck Kieselgel 60 F254 . Infrared spectra were recorded on a Bruker ALPHA FTIR spectrophotometer (ZnSe ATR). Gas chromatography was carried out on a Hewlett Packard 5890A and 5890 Series II Gas chromatographs with ChemStation software using HP1 (Crosslinked Methyl Silicone Gum) and CP-Chirasil-Dex CB columns, respectively. Both chromatographs were equipped with split/splitless capillary inlets and flame ionization detectors (FID). UV-vis spectra were recorded on a Varian Carey 4000 UV-vis spectrophotometer.

Hydroxyamination reactions
Acetonitrile was freshly distilled from calcium hydride; for reactions under argon (i.e., anaerobic conditions), the solvent was subjected to three freeze-thaw degassing cycles immediately before use. Stock solutions of iron complex (22.6 mmol L −1 ) and BocNHOH (8, 70 mmol L −1 ) in degassed acetonitrile were prepared under an atmosphere of argon. Acetonitrile (8.0 mL) was stirred under the required environment (argon or air) while iron complex stock solution (0.3 mL, 6.8 μmol) and cyclohexene (0.7 mL, 6.9 mmol) were added. Using a syringe pump, the BocNHOH stock solution (1.0 mL, 70 μmol) was added to the reaction mixture over 30 min. The reaction was stirred for 18 h after which time the solvent was removed in vacuo. The residue was dissolved in ethyl acetate and passed through a micro-column of silica to remove the iron complex. The sample was subjected to analysis by GC using n-decane as an internal standard and the single point internal standard method (Supporting Information File 1) [62,63]. Each reaction was performed in triplicate and data presented above are the average of the three runs.

Supporting Information
Experimental procedures and characterization data for synthesis of ligands and iron complexes plus preparative-scale turnover reactions; details of GC conditions for analysis of turnover reactions; turnover data for control experiments and investigation of catalyst loading; UV-vis and 1 H NMR spectra evincing the interaction of BocNHOH (8) with FeTPA (4).

Supporting Information File 1
Experimental procedures and characterization data, GC conditions, UV-vis and 1 H NMR spectra.