Organic Letters
Letter
Scheme 2. Pathway of Decarboxylative Arylation
Table 1. Parameters for Decarboxylative Arylation of
a
Phenylglyoxylic Acid with 4-Bromobenzonitrile
variations from conditions given
above
conversion of 1a
yield
(%)
entry
(%)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
none
>99
>99
0
20
5
27
0
2
78
56
15
0
7
0
90
46
0
<1
<1
<1
0
<1
69
51
9
TXO instead of Cl-TXO
BPHO instead of Cl-TXO
Eosin Y instead of Cl-TXO
Rose Bengal instead of Cl-TXO
fac-Ir(ppy)3 instead of Cl-TXO
no photoredox catalyst
no photoredox catalyst
using 10 mol % Cl-TXO
no H2O
no dtbbpy
no nickel
no Li2CO3
no light
b
c
0
0
0
91
= 1.468 V vs SCE in DMF based on electrochemical and
sunlight instead of CFL
>99
radical RCOCO2 and ketyl radical anion [Cl-TXO]•−.
•
a
•
1a (0.2 mmol, 1 equiv), 2a (0.4 mmol, 2 equiv), NiCl2·gylme (0.02
Unstable RCOCO2 extrudes CO2 to generate acyl radical
PhCO•. The reduced [Cl-TXO]•− transfers one electron to
LnNiIX generating LnNi0 with concomitant regeneration of
photoredox catalyst Cl-TXO [E(NiII/Ni0) = −1.20026 and
Ered(Cl-TXO) = −1.506 V vs SCE]. The oxidative addition of
arylhalide (ArX) to LnNi0(solvent) (A) gives LnNiII(Ar)X (B),
which in turn intercepts an acyl radical to generate the Ni(III)
species LnNiIII(Ar)(COR)X (C). Subsequent reductive elim-
ination provides ketone and Ni(I) complex LnNiI(X) (D),
completing the cycle.
mmol, 10 mol %), dtbbpy (0.024 mmol, 12 mol %), Li2CO3 (0.4
mmol, 2.0 equiv), H2O (3.0 mmol, 15 equiv), photocatalyst (20 mol
%), 6 mL of DMF, N2, irradiation under 45 W CFL for 24 h with
cooling by a fan, high-performance liquid chromatography conversion
b
c
and yield. Using 2 mol % fac-Ir(ppy)3. Under 365 nm.
mol %) and TEMPO (2 equiv) under 45 W CFL gave acyl-
TEMPO adduct 2,2,6,6-tetramethylpiperidin-1-yl benzoate
suggest that a SET process from 2a to Cl-TXO* is favorable
and confirmed the formation of an acyl radical intermediate.
No ketone product was observed in the absence of Cl-TXO
(entry 7). The irradiation of 1a and 2a in DMF without Cl-
TXO under 365 nm for 24 h afforded 3aa in <1% yield (entry
8). These conditions preclude triplet−triplet energy transfer
between Cl-TXO* and the Ni catalyst.27 Decreasing the Cl-
TXO loading resulted in a 69% yield of 3aa (entry 9). The
absence of H2O was also detrimental, with the yield of 3aa
decreasing to 51% (entry 10). Water presumably assists in
solubilizing Li2CO3 in DMF. Screening solvents identified
DMF as the most suitable choice (Table S1). Without dtbbpy
to stabilize the metal catalyst, the light-driven reaction was
greatly suppressed, giving 3aa in only 9% yield (entry 11). The
absence of nickel or base resulted in no coupling (entries 12
and 13). Likewise, no reaction occurred in the dark (entry 14).
The yield of 3aa reached 91% after sunlight irradiation for 8 h
[entry 15; the maximum power density was ∼4.81 mW cm−2
(Figure S3c)]. This cross-coupling reaction (see the
reaction of 1a (1.09 g, 6 mmol) and 2a (1.8 g, 12 mmol) to
give 3aa in 66% HPLC yield (0.820 g) under 2 × 45 W CFL
irradiation for 72 h and in 77% HPLC yield (0.956 g) under
To evaluate this hypothesis, we initiated our investigation
with the model reaction of 4-bromobenzonitrile (1a) with
phenylglyoxylic acid (2a) in the presence of 20 mol % Cl-
TXO, 10 mol % NiCl2·glyme, 12 mol % dtbbpy (dtbbpy =
4,4′-di-tert-butyl-2,2′-bipyridine), and 2.0 equiv of Li2CO3
under the irradiation of a 45 W compact fluorescent lamp
our delight, the formation of 4-benzoylbenzonitrile (3aa) was
achieved in 90% yield (Table 1, entry 1). The excited-state
redox potential of the photocatalyst clearly had a significant
influence on the decarboxylative arylation (entries 1−6). The
utilization of thioxanthen-9-one [TXO, E(TXO*/TXO•−) =
1.34 V vs SCE in DMF23e] instead of Cl-TXO drastically
decreased the efficacy of the cross-coupling, affording 3aa in a
more modest 46% yield (entry 2). Benzophenone [BPHO,
E(BPHO*/BPHO•−) = 1.149 V vs SCE in MeCN], Eosin Y
(E* = 1.18 V vs SCE), Rose Bengal (E* = 0.99 V vs SCE), and
fac-Ir(ppy)3 (E* = 0.31 V vs SCE)23e with lower excited-state
oxidation potentials did not perform well in this coupling
(entries 3−6, respectively). Stern−Volmer quenching studies
(Figure S4) revealed that 1a did not quench the emission of
excited-state Cl-TXO in DMF. By contrast, a solution of 2a
and Li2CO3 did gradually decrease the photoluminescence
intensity (λex = 374 nm) of Cl-TXO together with a
concomitant increase in 2-oxo-2-phenylacetate concentration
(Figure S5). Carbon dioxide was the byproduct identified by
gas chromatography (GC) analysis (see Figure S6). The
addition of 2.0 equiv of 2,2,6,6-tetramethylpiperidin-1-oxyl
(TEMPO) to the standard reaction mixtures gave an only 14%
yield of the desired product. Reaction of 2a with Cl-TXO (20
The scope of this decarboxylative arylation of α-keto acids
was explored using the optimized reaction conditions
described above. As shown in Scheme 3, aryl bromides
substituted with electron-withdrawing groups in the para and/
or meta positions (3aa−3ga) were isolated in good yields (61−
88%). The reaction of 2-bromobenzonitrile (1h), methyl 2-
B
Org. Lett. XXXX, XXX, XXX−XXX