Journal of the American Chemical Society
Article
yielding the desired product E1 in 78% yield, 91:9 er. In
consideration of the fact that B14 is prepared from bench-
stable benziodoxole (B13) with an excess amount of TMSN3
and is more easily to be handled, we explored the use of B13 as
the terminal oxidant under these reaction conditions. To our
delight, E1 was obtained with higher yield albeit a slight
decrease of enantioselectivity (90% yield, 89:11 er). Next, the
scope of this iron-catalyzed diazidation was examined. As
indicated in Table 2, an array of α,β-unsaturated ketones with
various substituents on the α-phenyl ring were all amenable in
diazidation (E2−E10, 75:25−91.5:8.5 er). Different hetero-
aromatic rings, such as 2,3-dihydrobenzofuryl-, piperonyl-, and
2-naphthyl-substituted enones were further tested, delivering
the desired products with good enantioselectivity (E11−E13,
83:17−91.5:8.5 er). Moreover, the acrylamides also performed
smoothly to furnish the target vicinal diazide E14 in 68% yield
with 71:29 er.
2.4. Synthetic Applications. To illustrate the potential
synthetic utility of this reaction, a gram-scale synthesis of chiral
azide D1 was performed. By treatment of 5 mmol of A1 in the
presence of L3-Pi1Ad/Fe(OTf)2 for 3 h, 1.06 g of the isolated
D1 was obtained with maintained enantioselectivity (Figure
2a). Chiral α-amino ketones,40,41 1,2-amino alcohols,42,43 and
vicinal diamines44,45 are important compounds and frequently
found in pharmaceuticals and medicinally relevant natural
products. Direct reduction of the product D1 by Pd/C with H2
provided the chiral amino ketone F1 (Figure 2a); reduction of
D29 and D30 afforded vicinal amino alcohols G1 and G2
(Table 1); and reduction E1 resulted in vicinal diamines H1
efficiently (Figure 2b). The absolute configuration of the
optically pure F1 was unambiguously determined to be S
according to X-ray crystallography. Furthermore, the copper-
catalyzed Huisgen cycloaddition was employed to transform
the chiral azides into the corresponding triazoles I1−I4 in
good yield without any erosion of the enantioselectivity
(Figure 2c).
We carried out the control experiments to probe into the
different reactivities of styrene,23,24 enone A1, and α,β-
for more details). It was found that in the presence of
Fe(OTf)2 styrene showed the highest reactivity, followed by
A28 and A1. Nevertheless, the corresponding product of
styrene dropped dramatically when chiral L3-Pi1Ad/Fe(OTf)2
was used. It implied that the catalytic species might be changed
upon the addition of chiral ligands. The styrene performed the
reaction principally via an intermolecular azidation pathway.
However, the introduction of tetradentate ligand with steric
hindrance generated a new octahedra species, which acted as a
strong Lewis acid to bond the α,β-unsaturated carbonyl
compounds. As a result, the gap between the SOMO of the
alkyl radical and LUMO of the electron deficient alkene might
be reduced,46 and the reactivity of enone or amide surpassed
the uncoordinated styrene.
Based on the above experimental evidence, we rationalized
the possible catalytic process for this iron-catalyzed asymmetric
carboazidation. As illustrated in Figure 3, the Lewis acid
2.5. Mechanistic Studies. To gain more insight of the
mechanism, some control experiments were conducted. First, a
radical-trapping experiment was performed with 2,2,6,6-
tetramethyl-1-piperidinyloxy (TEMPO) under standard con-
ditions with B8, and the desired product D34 was not
observed, while the alkyl-TEMPO adduct was isolated in 56%
yield based on B8 (Figure 2d). Then, the radical clock study
was run with either A29 or B15, and it was found that the ring-
opened product D39 and D40 were isolated in 65% and 24.5%
yield, separately (Figure 2e,f). These results strongly supported
that a radical addition process and benzylic radical were
involved in this reaction. EPR experiments using DMPO (5,5-
dimethyl-1-pyrroline-N-oxide) indicated the formation of
radicals via single-electron transfer, especially with the aid of
for more details).
Figure 3. Proposed catalytic cycle.
catalyst I, in situ generated from the tetradentate L3-Pi1Ad and
Fe(OTf)2, undergoes ligand exchange with TMSN3 to afford
L−Fe(II)−N3 species II. Togni’s reagent B2 oxidizes II via a
single-electron transfer (SET) process to generate the chiral
Fe(III)−N3 species III and CF3 radical. During this step, the
SET process may proceed prior or posterior to the
coordination of enone A1 to iron. Next, CF3 radical adds to
catalyst-bonded enone species III to form the trisubstituted
radical species, which coordinates to the metal center (IV).
Finally, the intermediate IV under goes the critical azido group
transfer, giving the product D1 and regenerating the catalyst.
2.6. DFT Calculations. To gain insights into the
mechanisms of azido transfer and the origin of the
enantioselectivity induced by the N,N′-dioxide ligand, we
carried out DFT calculations using L3-Pi1Ad as the ligand. The
discussion here is based on the data calculated on CPCM
(tetrahydrofuran), M06L/Def2-TZVP//M06L/6-31G(d,p),
SDD(Fe) level of theory. DFT calculations were performed
using Gaussian software (for more details, see the Supporting
Information). A benzylic radical bonded Fe(III) intermediate
IV was considered as a reactive species for the following
investigations (Figure 4). The six coordination sites of IV are
saturated by a tetradentate ligand, a carbonyl group of one
It was clear that the orange mixture of the L3-Pi1Ad/
Fe(OTf)2 complex in CH2Cl2 changed to a red solution upon
addition of TMSN3. IR analysis of the aforementioned mixture
showed that azido group absorption appeared at 2049 cm−1,
indicating a red shift in comparison to TMSN3 (2132 cm−1).
ESI-MS analysis of the mixture showed a clear peak at m/z
843.3279, which corresponded to [Fe3+ + L3-Pi1Ad + OTf− +
− +
N3 ] (see Figure S11 for more details). Those observation
confirmed the formation of Fe−N3 species and implied that
the ferrous species was readily oxidated into the ferric
intermediate.
11860
J. Am. Chem. Soc. 2021, 143, 11856−11863