We hypothesized that it would be benefi-
cial to have moderate steric hindrance on the
NBE—sufficient to inhibit the cyclopropanation
pathway and promote theb-carbon elimina-
tion but not to inhibit the reaction between the
alkenyl-norbornyl palladacycle and the elec-
trophile (avoiding side productsAandB). Also,
a bulky hydride source, particularly one that
could slowly release hydride, would be desir-
able to minimize early hydride terminations
(avoiding side productsCandD). To test the
hypothesis, a range of NBE cocatalysts were
surveyed, and bifunctional carbonate-based
reagents that contain an electrophilic amine
moiety ( 25 , 26 ) and a masked hydride ( 27 ) were
designed. Whena-tetralone ( 1 ) was used as
the model substrate, the desiredb-tetralone
product ( 3 ) was ultimately obtained in 88%
yield in one pot after careful evaluation of var-
ious reaction parameters (tables S1 to S3).
Through sequential addition of the reactants,
the alkenyl triflate intermediate ( 2 ) was formed
in situ and underwent the subsequent trans-
posed amination and hydrolysis without the
need for isolation or change of the reaction
vessel. The key was to identify triflation con-
ditions that are efficient and compatible with
the Pd/NBE catalysis. With triflic anhydride as
the triflation reagent, the use of Cs 2 CO 3 as the
base in a mixed solvent of toluene and 1,4-
dioxane proved to be optimal, affording alkenyl
triflate 2 in near-quantitative yield (table S1,
entry 1). Toluene alone afforded a lower con-
version (table S1, entry 2). Although the bulky
pyridine base was efficient for the triflate for-
mation, it was detrimental for the subsequent
amination step, likely as a result of compat-
ibility issues (table S1, entries 3 and 4). Clean
reactions and good overall efficiency were
also observed when using lithium tetrame-
thylpiperidide (LiTMP) as the base to form
the nonaflate intermediate that reacted smooth-
ly in the Pd/NBE catalysis step (table S1, entry
5). This offers an alternative one-pot condition
(see below).
Under the optimal triflation conditions, vari-
ous electrophilic amine sources were investigated
for thea-amination and ipso-hydrogenation
step. The common electrophilic morpholino
benzoate (R1), together with an exogenous
hydride source, produced 3 in only 24% yield
while forming a substantial amount of the
direct ipso-reduction (A-type) side product
(table S1, entry 6). Through tethering with a
bulky alcohol fragment, the readily accessi-
ble, morpholine-derived carbonatesR2to
R6produced much-improved yields and sup-
pressed the side reactions, with the reagent
derived from 2,4-dimethyl-3-pentanol (R4)
showing superior results. Carbonate reagents
derived from other amines (R7toR9) were
also tested; the piperidine-based one (R8) was
found to be optimal in this case, as it resulted
in full conversion and negligible side products
compared with the others (table S2). Control
experiments indicated that both the palladiumand the NBE are essential to this reaction, and
diminished yield was observed in the absence
of the pyridone additive (table S1, entries 7 to 9)
( 28 ). Among all of the structurally modified
NBEs ( 29 ) that were investigated (table S1,
entry 10), the azetidine-amide NBEN10
proved to be the most effective. A compa-
rable yield was observed when lowering the
N10loading (table S1, entry 11), showing
some catalytic activity. Simple NBEN1and
C1-substituted N4 generated substantial
cyclopropaneBand enamineD-types of side
products, consistent with our previous find-
ings ( 21 ). The NBEs with C5 and C6 disub-
stitution (N2andN3) and C2-ester groups
(N5) proved to be inferior for this trans-
formation. TheN-methyl (N-Me) amide NBE
N7only afforded moderate yield while form-
ing NBE-incorporated side products, and
the bulkier dimethyl and diethyl-amide NBEs
led to substantially decreased reactivity
with more side productAformation and
homocoupling. The high reactivity and se-
lectivity ofN10is more evident in the more
challenging substrates, such as nonconju-
gated alkenyl triflate18a(table S4). In ad-
dition, the Buchwald Ph-DavePhos ligand
(Ph, phenyl) was most efficient ( 21 ). Though
the slightly less electron-rich Ph-JohnPhos
(L2) produced somewhat lower yield, reg-
ular DavePhos (L3) and PPh 3 (L4) showed
much-diminished reactivity (table S1, en-
try 12) ( 30 ).
With the optimized conditions in hand,
we further examined the scope of this car-
bonyl 1,2-transposition reaction (Fig. 3). A
range ofa-tetralone derivatives bearing sub-
stituents on both aliphatic ( 5 and 6 ) and
aromatic ( 8 to 12 and 14 ) rings afforded
the desiredb-tetralones in good yields via
the one-pot protocol. Chromanone ( 6 ) also
reacted, albeit in somewhat lower yields, prob-
ably owing to the coordinative capability of the
heteroatoms. Besidesa-tetralones, substrates
derived from five-membered 1-indanone ( 15 ),
seven-membered 1-benzosuberone ( 16 ), and
a substituted benzocycloheptenone ( 17 ) also
delivered the desired deconjugated products.
This one-pot protocol was particularly bene-
ficial when the corresponding alkenyl triflate
intermediates were sensitive to air, such as the
precursor to 8. For thiochromanone ( 7 ) and
the substrate that contained a highly electron-
deficient nitroarene ( 13 ), much higher overall
yields were obtained when the alkenyl triflate
intermediates were purified. In addition, non-
conjugated ketones also proved to be suitable
substrates. Although the one-pot procedure
(using LiTMP as the base) provided the desired
carbonyl 1,2-shifted product (e.g., 18 ) in a useful
yield, the two-step procedure generally resulted
in higher overall efficiency. Cyclohexanones
bearing substituents at thegposition were
converted to the correspondingb-substituted736 5NOVEMBER2021•VOL 374 ISSUE 6568 science.orgSCIENCE
Fig. 2. Challenges and mechanistic considerations of the Pd/NBE-catalyzeda-amination of alkenyl
triflates.Cyclopropanation and premature reduction of the NBE-containing intermediates are major side
reactions. ANP, alkenyl-norbornyl palladacycle; R, alkyl group; X, X-type ligands; Y, a substituent on the NBE;
3-exo-trig, three-membered ring formation via exocyclic ring closure at sp^2 atoms.
RESEARCH | REPORTS