Organic Letters
Letter
yields (6k, 6l), while ortho- led to slightly diminished product
formation (6m). Electron-donating groups at C-4 such as
methyl and methoxy (6o−q) were found to be effective.
Gratifyingly, electron-deficient groups such as trifluoromethyl
at C-3 (6r) and cyano (6s) at C-4 (6m) were also well
tolerated. C-2 substituted heteroaromatic furan (6n) and
thiophene (6o) also led to the rearranged product under the
optimized reaction conditions in good yields. α-Substituted
benzyl groups were tested next for their thermal [1,3]
migration. The rearrangements with tertiary migrating groups
proceeded at a faster rate than the secondary benzyl to furnish
products in good overall yields with low to moderate
diastereoselectivities. The 1-phenylethyl group migrated with
a 79% yield and 1.7:1 diastereomeric ratio, and 1-β-
naphthylethyl rearranged to 67% product with 2.3:1 diaster-
eoselectivity. Interestingly, the 1-phenylcyclopropyl ethyl
group migrated efficiently (yield 75%, dr = 2:1) without any
ring-opening product.11 The low diastereoselectivities in these
products might be a combination of poor d.r. of in situ formed
[1,3] precursors and dissociative nature of the rearrangement.
A 1,1-bisphenylmethane (6y) also migrated efficiently to yield
71% of the rearranged product. Alkyl groups such as methyl,
ethyl, and isopropyl as the migrating group remained
unreactive under the optimized reaction conditions, and
heating at higher temperature led to a complex reaction
mixture, presumably via multiple air oxidation paths.12
Scheme 1. Mechanistic Experiments
With the establishment of a broad substrate scope for the
thermal [1,3]-rearrangement protocol, we next conducted
mechanistic studies on this facile migration. First, a radical trap
reaction with TEMPO was conducted to obtain direct
evidence for a bis-radical path. A 1 equiv amount of TEMPO
did not alter the product formation significantly, although we
detected a TEMPO trapped benzyl radical by HRMS analysis
of the crude reaction mixture. Increasing the TEMPO to 5
equiv led to a lowering of yield for the [1,3] product (58%)
along with 18% isolated TEMPO trapped product (Scheme
1A). Next, we carried out a crossover experiment with 3a and
3i (Scheme 1B), which mainly resulted in intramolecular
products 6a and 6i along with ∼10% total cross-products 6b
and 6h. The partial trapping of radicals via TEMPO and a
minor amount of crossover product formation indicate a
solvent cage recombination of bis-radical intermediates. To
further distinguish between radical versus ionic paths, we
compared the rate of rearrangement for the 4-cyanobenzyl
migrating group to the parent benzyl group (Scheme 1C). The
4-cyano substituent is expected to reduce the rate of the ionic
reaction13 while accelerating the radical reaction path.14 The
thermal rearrangement (Scheme 2A). The fact that our
substrates rearrange effectively at a lower temperature indicates
further stabilization via solvent on the captodatively stable
radical III. To estimate the relative solvent effects, we first
determined the ΔG’s in solvents with an increasing dielectric
constant for these three α-keto radicals. The calculations show
a considerably higher magnitude of stabilization (5.0 vs 1.8 vs
1.3 kcal/mol from gas-phase to DMSO) for radical III over
radical II and I. The origin of this solvent stabilization was
attributed to their polarization which was examined via spin
densities for radical II and III in different solvents (Scheme
2B) using the SMD solvation model.9,16 The computations
show that the reduction in spin density at radical carbon in a
higher polarity solvent was greater in magnitude for our system
(III) compared to the α-ester radical II. Conversely, the spin
density on electron donor nitrogen in III increased
significantly, while no significant change was observed on α-
carbons of II. Both the spin density distribution and ΔG
calculation in different solvents shows our captodatively stable
radical III was more polarizable than a typical α-keto radical
and exerted higher stabilization in polar solvents. The solvent
effect was tested experimentally, which showed a 1.6-fold rate
enhancement in DMSO compared to toluene at 40 °C. We
observed significant spin density on sulfur (0.12), which is
indicative of its superior effect on radical III stabilization and
success over other carboxylic acid protecting groups tested.
Finally, we demonstrated the removal of the 2-amino-
thiophenol from the [1,3] rearranged products. Several
reported S,N-acetal deprotection reagents were unsuccessful,
but AgNO3 at 60 °C afforded the desired products 8 in 55−
64% yield.17 Optimization with various silver salts led to
cleaner deprotection with AgBF4 in MeCN/H2O (3:1) at 60
°C for 2 h to the diketone products with various C-2
1
kinetic experiments via H NMR monitoring in DMSO-d6 at
40 °C resulted in a 3.5 times faster reaction with the 4-cyano
substrate, further supporting the radical mechanism (see SI for
details).
To compare the relative energy required for thermal radical
pair [1,3]-rearrangements, we calculated the ΔG associated
with the bond dissociation step for Claisen,1 Shiina,4g and our
system using DFT (ωB97XD/6-31+G(d,p)) (Scheme 2).9,15
We chose the migrating group as benzyl for all three systems to
correlate the effect of α-ketyl radical stability (I, II, and III) on
the bond dissociation step. The gas-phase ΔG for the
homolytic cleavage of the Claisen system is 34.3 kcal/mol,
consistent with the very high temperature for its rearrange-
ment. For the fully substituted Shiina system, the ΔG is 12.9
kcal/mol, while for our 2-aminothiophenol protected system
the ΔG is 13.0 kcal/mol, in line with their unusually facile
892
Org. Lett. 2021, 23, 890−895