Journal of the American Chemical Society
Communication
demonstrated the role of cobalt as the catalyst (entry 14, Table
1).
Table 1. Optimization of Reaction Conditions
The unexpected substituent effect on the reactivity of
acylating reagents shown above (entries 1−9, Table 1) is
remarkable. By simply changing the open-chain structures (as
in 3c−3e) to the 1,3,2-dioxophosphinanyl ring (as in 3a), the
reagent reactivity is drastically increased. Compared to the
open-chain structures, the 1,3,2-dioxophosphinanyl ring might
reduce the steric hindrance associated with the carbonyl group
and thus facilitates the attack of alkyl radicals. The gem-
dimethyl substitution in 3a may help to stabilize the chair
conformation of the heterocycle. This is evidenced by the X-
ray crystal structures of 3a and 3b in that the six-membered
heterocycle adopts a chair conformation in 3a but a half-chair
conformation in 3b. The CO and PO groups in 3b are
almost coplanar with a dihedral angle of −176°, indicating that
radical attack to the carbonyl from either face will encounter
the same steric hindrance. In contrast, the CO and PO
groups in 3a have a dihedral angle of −128°, resulting in the
blockade of one face of the carbonyl group but the clearance of
the other. As a consequence, the reaction with 3a exhibits
higher efficiency than that with 3b. These experiments are also
the first examples of intermolecular acylation with acylphosph-
onates. It should also be noted that, in the literature
reports,67−80 the proposed species responsible for acylation
of alkyl radicals are generally unstable organometallic
intermediates such as acyl-NiII intermediates. As a comparison,
acylphosphonates structurally similar to 3a are stable
crystalline solids that are easy to handle. Furthermore, they
can be readily prepared by reaction of acyl chlorides with
trialkyl phosphites in one step83,92,93 or in two steps by dialkyl
phosphite addition to aldehydes followed by Dess−Martin
oxidation in excellent yields (see SI for details). These
characteristics should encourage further applications of
acylphosphonates as radical acylating agents.
With the optimized conditions in hand, we examined the
scope of the method. As shown in Scheme 1, a variety of
terminal alkenes underwent hydroacylation reaction smoothly
leading to the synthesis of branched ketones 4b−4u in
satisfactory yields. Exclusive Markovnikov selectivity was
observed in all cases. Chemoselective hydroacylation was
observed in the case of 4u in which the isopentenyl moiety
remained intact. By lowering the catalyst loading to 5 mol %
and slightly increasing the amounts of TBHP and 3a, the
protocol was also applicable to internal alkenes as exemplified
by the synthesis of 4v−4x from the corresponding cyclo-
alkenes. Similarly, the reaction of methyl cyclopent-3-
enecarboxylate furnished the expected product 4y in 74%
yield. The presence of a wide range of functional groups was
well tolerated by the process. For example, alkyl (or aryl)
halides, amides, sulfonamides, esters, sulfonates, ethers, silyl
ethers, alkylphosphonates, nitriles, furans, and nitroalkanes all
proved to be compatible with the reaction. This excellent
functional group compatibility enabled the late stage
modification of complex molecules or drug derivatives. For
example, the reaction of alkene 1za containing an N-Boc-
protected dipeptide motif afforded product 4za in 52% yield.
Ketone 4zb having the sensitive diacetonefructose skeleton was
also achieved under the optimized conditions. Alkenes derived
from ibuprofen, hymecromone, nortropine, and estrone
produced the expected products 4zc−4zf in good yields.
Encouraged by the above results, we went on to test the
hydroacylation of alkene 1a with a number of acylphospho-
a
a
entry
variation from the “standard conditions”
yield (%)
1
2
3
4
5
6
7
8
none
78
52
19
21
28
8
17
11
10
17
trace
60
10
0
3b in place of 3a
3c in place of 3a
3d in place of 3a
3e in place of 3a
3f in place of 3a
3g in place of 3a
3h in place of 3a
3i in place of 3a
(MeO)3SiH in place of PhSiH3
Et3SiH in place of PhSiH3
without Selectfluor
without TBHP and Selectfluor
without 2
9
10
11
12
13
14
a
The reaction was carried out in 0.20 mmol scale in i-PrOH (6.0 mL).
Isolated yield based on 1a.
b
phenylsilane as the hydrogen source, tert-butyl hydroperoxide
(TBHP) as the oxidant, and 1-chloromethyl-4-fluorodiazo-
niabicyclo[2.2.2]octane bis(tetrafluoroborate) (Select-
fluor)90,91 as the additive, the reaction of 1a in isopropanol
at room temperature (rt) delivered the desired ketone 4a in
78% yield (entry 1, Table 1). No anti-Markovnikov hydro-
acylation product or chain-walking product10 could be
detected. Switching the acylating reagent from 3a to its analog
3b also provided ketone 4a in 52% yield (entry 2, Table 1). In
contrast, the use of dimethyl phosphonate 3c or diethyl
phosphonate 3d in place of 3a resulted in a very low (∼20%)
yield of 4a (entries 3 and 4, Table 1). A poor result was also
observed when 3a was replaced by monoacylphosphine oxide
3e or 3f (entries 5 and 6, Table 1). Acylphosphinates 3g and
3h and diacylphosphine oxide 3i proved to be incompetent
acylating reagents (entries 7−9, Table 1). Phenylsilane
exhibited much better performance than trimethoxysilane or
triethylsilane (entries 10 and 11, Table 1). A catalytic amount
of Selectfluor, an additive commonly used in cobalt-catalyzed
hydrofunctionalization reactions, helped to increase the
product yield (entry 12, Table 1). Other additives such as
N-fluoropyridinium salts were less efficient than Selectfluor in
promoting the transformation (see Table S2 in SI for details).
TBHP proved to be an oxidant superior to other peroxides or
hypervalent iodine reagents such as diacetoxyiodobenzene
(also see Table S2). Interestingly, the reaction also took place
in the absence of TBHP and Selectfluor, albeit in a low (10%)
yield (entry 13, Table 1), presumably because of the
contamination of trace O2.27 Finally, the control experiment
4956
J. Am. Chem. Soc. 2021, 143, 4955−4961