Organic Letters
Letter
reactions via C−OMe cleavage are limited, likely due to the high
activation energy barrier for breaking the C−OMe bond.13
Recently, our group published a cation radical-accelerated-
nucleophilic aromatic substitution (CRA-SNAr) of methoxy or
benzyloxy-substituted arenes using azoles, ammonia, and
trifluoroethano as nucleophiles.14 Cyanide has shown to be a
competent nucleophile for arene C−H functionalization,15 thus
we believed we could extend the CRA-SNAr methodology for
the synthesis of benzonitriles. Herein, we report the use of
cyanide as a nucleophile in C−C bond formation for
benzonitrile synthesis utilizing a highly oxidizing acridinium
photoredox catalyst (Scheme 1B).16
We began by adapting our conditions previously developed
for arene C−H cyanation using TMSCN as the cyanide
source.15 Using 3,4-dimethoxybenzonitrile as a model substrate,
treatment with a catalytic quantity of acridinium salt Mes-Acr-
Ph+ under irradiation with 455 nm LEDs yielded the desired C−
O cyanation product in 18% yield as a mixture of two
regioisomers (Table 1, entry 1). In comparison to TMSCN,
The scope of this transformation was then examined using the
optimized conditions (Figure 1). Simple mono- and disub-
stituted arenes were investigated first. Arenes bearing 1,2-
dimethoxy substituents (1−5) were found to be the highest
yielding of this group. 2-Chloro substituted anisole derivatives
(6−8) also gave moderate to good yields of the desired product.
A variety of 1,2-dimethoxybenzophenone derivatives (12−17)
performed well with yields ranging from 48 to 83%. To our
surprise, selective cyanation was observed for 3-bromoveratrole
(9) and 1,2,3-trimethoxybenzene (10) yielding the products as
single regioisomers.
Aryl ethers bearing para-pyridine and pyrimidine substituents
(19−22) also produced the expected adducts, albeit in lower
yields. More complex substrates such as the noradrenaline (23)
and letrozole derivatives (24) also gave the desired benzonitrile
products in low to moderate yields.
The mass balance of this C−O cyanation reaction is excellent
with crude reaction mixtures typically consisting of only product
and unreacted starting material as the main chemical
components. Neither the inclusion of additional equivalents of
cyanohydrin nor extended reaction times were effective in
converting the remaining starting material to product,
suggesting that catalyst decomposition may be responsible.
The mass of a cyanide−acridinium adduct was detected by
HRMS supporting catalyst decomposition by adventitious
a
Table 1. Optimization for Catalytic SNAr Cyanation
Having evaluated the scope of this transformation, we sought
to investigate the mechanism and origins of regioselectivity of
this ipso-functionalization using a combination of experimental
and computational data. Examining the initial rates of a variety of
varying benzophenone derivatives, no linear free energy
relationship was observed. However, these experiments did
35−40). This is likely caused by a thermal barrier for the release
of cyanide from acetone cyanohydrin.
We previously reported that in the presence of a terminal
oxidant, nucleophilic attack of the anisole cation radical will
occur at the para- or ortho-position, which requires a second
irreversible oxidation to afford the C−H amination product.18
Without a terminal oxidant, however, the reaction pathway is
steered toward the ipso-product. Intriguingly, substrates which
underwent ipso-C−O cyanation efficiently were not competent
substrates under C−H cyanation conditions and vice versa. We
then sought to demonstrate that the absence of oxygen would
not simply allow a substrate to undergo ipso-substitution and
that the chemoselectivity is determined by the electronics of the
arene in the cation radical and ground state. To exemplify this,
we turned to computational modeling.
Our group has previously demonstrated that natural
population analysis (NPA) can be employed to computationally
predict the regioselectivity of cation radical mediated arene C−
H functionalization reactions.19 Upon oxidation, the ortho-,
para- and ipso-carbons of anisole are activated by an increase in
positive charge density, allowing for nucleophilic addition at
these positions.19 The difference between the predicted NPA
values of each carbon of the cation radical and ground state
arene were largely accurate in modeling the regioselectivity
observed for photoredox catalyzed C−H aminations. We elected
to apply this method to understand the chemoselectivity of
arene C−O cyanation. Utilizing the B3LYP/6-31-G+(d) level of
theory, the NPA for the arene neutral state and cation radical
state were determined for a variety of substrates. Comparing
these values with experimental results, we found that the largest
entry
solvent
MeCN
MeCN
DCE
DCE
DCE/TFE (3:1)
DCE/TFE(3:1)
DCE/HFIP (3:1)
DCE/TFE (3:1)
DCE/TFE (3:1)
base
yield (A:B)
b
1
10% v/v pH 8 buffer
10% v/v pH 8 buffer
10% v/v pH 8 buffer
33% NaHCO3(aq)
33% NaHCO3(aq)
NaHCO3 (5.0 equiv)
NaHCO3 (5.0 equiv)
NaHCO3 (5.0 equiv)
NaHCO3 (5.0 equiv)
18% (2.6:1)
24% (2.4:1)
24% (2.4:1)
35% (2.2:1)
80% (2.4:1)
90% (2.4:1)
53% (2.5:1)
n/a
2
3
4
5
6
7
c
8
d
9
n/a
a
b
Reactions run at 0.1 M with respect to substrate. TMSCN (4.0
equiv) used as a cyanide source. The reaction was run without LEDS.
No photocatalyst.
c
d
acetone cyanohydrin was found to be higher yielding and a more
economic cyanide source (entries 2−7). Other organic cyanide
sources or cyanide salts failed to improve the reaction yields
common organic and inorganic bases identified solid or aqueous
sodium bicarbonate as the optimal base for the generation of free
Under these reaction conditions, the weak basicity of sodium
bicarbonate may improve reactivity by keeping the concen-
tration of the cyanide anion low, which helps to prevent
undesired degradation pathways via oxidation of the cyanide
anion by the photoredox catalyst.17 When examining solvents,
we found that the inclusion of 2,2,2-trifluoroethanol (TFE)
resulted in a significantly enhanced yield of the desired product
(Table 1, entries 5 and 6). In contrast, 1,1,1,3,3,3-hexafluor-
oisopropanol (HFIP) only marginally increased the yields
relative to those observed in DCE (entry 7). No reaction was
observed in the absence of irradiation or the acridinium
photoredox catalyst (entries 8 and 9, respectively).
B
Org. Lett. XXXX, XXX, XXX−XXX