Organic Letters
Letter
derivatives in 54 and 20% yields, respectively. We thus
conducted these reactions on various para-substituted 7-
phenyl-1,3,5-cycloheptatrienes 1c−1h (R = Me, t-Bu, n-Bu,
OMe, Cl, and F) with nitrosobenzene 2a (X = H), yielding
expected products 5c−5h with a mixture of para and ortho
products in ratios of 2−2.5:1; herein, electron-donating
substituents on 1,3,5-cycloheptatrienes (X = Me, t-Bu, n-Bu,
and OMe) were more efficient substrates than their halo
analogues (X = Cl and F). The meta-substituted 7-phenyl-
1,3,5-cycloheptatrienes (R = Me, OMe, and Cl) were also
applicable to these reactions, further affording these aldehyde
derivatives 5i−5k satisfactorily with similar para/ortho ratios.
For ortho-substituted 7-phenyl-1,3,5- cycloheptatrienes 1l−1n,
their resulting para- or ortho-aryl-substituted benzaldehydes
5l−5n were also obtained in 40−46 and 20−33% yields,
respectively; herein, meta-methyl-derived product 5l was
produced in a small para/ortho ratio, 1.3:1. For para-(1′-
naphthyl)-1,3,5-cycloheptatriene 1o, its resulting product 5o
also has a small para/ortho ratio, 1.5:1. We also prepared two
7-(2′-thienyl)-1,3,5-cycloheptatrienes 1p and 1q (R′ = H and
Me), further affording the corresponding para- and ortho-(2′-
thienyl)benzaldehydes 5p and 5q in 50−58 and 25−29%
yields, respectively. We tested the reaction on 7-(2′-furanyl)-
substituted 1,3,5-cycloheptatriene 1r, affording a similar para/
ortho isomer with a ratio of 2.3:1. Finally, we prepared 7-alkyl-
1,3,5-cycloheptatrienes 1s and 1t, which unfortunately afforded
three regioisomers with comparable ratios (1.2:1:1). The
absence of meta isomers for aryl-substituted aldehydes 5b−5o
is due to a non-conjugated position between the aryl and
CN+(−O−)Ph groups in primary product 5a.
this process appears to be feasible because of a small kinetic
barrier.
Figure 1 shows an energy profile for our proposed
mechanism; in our DFT calculations, some radicals are
involved to minimize the feasibility that is supported by a
reasonable barrier, ca. 22.6 kcal/mol for transition structure
TS5. Treatment of cycloheptatriene 1a with gold-bound
nitroso species A to form gold-bound tropylium intermediate
Int-1 is slightly exothermic, with ΔH = −5 kcal/mol and a
kinetic barrier of ca. 12.8 kcal/mol. Species Int-1 readily forms
gold-bound 7-nitroxy-1,3,5-cycloheptatrienes C′, with ΔH =
−1.7 kcal/mol and a small barrier (+0.9 kcal/mol). A gold
exchange between nitrosobenzene 2a and intermediate C′
forms our postulated intermediate C and gold-bound nitroso
species A, which releases an enthalpy of 6.6 kcal/mol. For
species C, its 6-π-electrocyclization to generate intermediate D
appears to be feasible, with a barrier of 6.6 kcal/mol. We found
a low-energy profile for an aromatization of species D via a
hydrogen abstraction of this species by gold-bound nitroso
species A, yielding O-centered nitroxy radical E and N-
centered nitroxy radical Int-2; this endothermic process has an
activation energy of 14.9 kcal/mol and ΔH = +2.6 kcal/mol. A
gold exchange of Int-2 with nitrosobenzene generates gold-
bound nitroso A and O-centered nitroxy radical Int-3 that
subsequently reacts with radical E to form observed nitrone
product 3 and N-hydroxylanline 6. The last step has a barrier
of ca. 22.6 kcal/mol, which is the largest in the entire sequence.
In transition structure TS5, O-centered radical Int-3 abstracts a
cyclopropyl hydrogen of O-centered radical E to cleave its
cyclopropane ring, leading to a skeletal rearrangement.
As shown in Scheme 4, 1,3,5-cycloheptatrienes are readily
oxidized by Au-bound nitrosobenzene A to form tropylium B
We altered the synthetic procedure in Scheme 5 to treat
cycloheptatriene 1a, nitrosobenzene 2a, and enol ether 7a with
P(t-Bu)2(o-biphenyl)AuCl/AgNTf2 in one step; herein, we
obtained one 5-alkoxyisoxazolidine product 8a, given from
nitrosobenzene 2a and enol ether 7a via a [2 + 2 + 1]-
annulation (eq 6). We believe that cycloheptatriene 1a serves
as a hydrogen donor to enable this new annulation. We hence
replaced this triene with 1,4-cyclohexadiene to yield only
compound 8a, with L = t-BuXPhos being superior to P(t-
Bu)2(o-biphenyl) (eq 7); the resulting cis/trans isomers were
separable on a silica column. In contrast, the reaction of
nitrosobenzene 2a and enol ether 7a failed to form compound
Scheme 4. Proposed Path for a Tropylium/Benzylidene
Rearrangement
1
8a (eq 7). H nuclear magnetic resonance (NMR) data of
compound 8a were identical to those of authentic samples.12
The use of N-hydroxyaniline 6 in this sequence still delivered
compound 8a in 38% yield (cis/trans = 2.5:1) (eq 8). We also
tested various gold catalysts to optimize the yields of
compound 8a in various solvents; the results appear in Table
S1 of the Supporting Information.
and nitroxygold intermediate C, which are likely intermediates
for this oxidative rearrangement; the evidence is the presence
of para/ortho and para-, ortho-, or meta-substituted benzalde-
hydes when 7-aryl or 7-alkyl-cycloheptatrienes are the
substrates. Notably, a reported tropylium B → benzyl cation
B′ rearrangement15 is inappropriate in this solution system
because this process is feasible only in the gaseous phase.
Herein, we undertook density functional theory (DFT)
calculations to support a gold-catalyzed 7-nitroxy-cyclohepta-
triene (C) → N-benzylideneaniline oxide (3) rearrangement;
In Scheme 5, we next assessed the generality of such gold-
catalyzed [2 + 2 + 1]-annulations using various nitrosoarenes
5508
Org. Lett. 2021, 23, 5506−5511