Organic Letters
Letter
the catalytic borylation of aryl fluorides.15 Herein we report the
cobalt-catalyzed C−F bond silylation of aryl fluorides using
cheap chlorosilanes as silicon sources. Importantly, this
method is compatible with various silicon sources and can
be operated under aerobic conditions. Mechanistic studies
support in situ formation of a Grignard reagent from the aryl
fluoride, which is captured by the electrophilic silicon source.
The formation of Grignard reagents or organometallic
compounds by activation of aryl C−F bonds for further
transformations is still a challenging task16 compared with
other aryl halides such as aryl chlorides or bromides.17
Accordingly, the reaction can be further applied to modular
transformations of aryl fluorides to various functional groups if
other electrophiles are used instead of the chlorosilanes.
We determined that a combination of the salt Co(acac)2 and
1,3-diketiminate ligand L1 in the presence of magnesium and
chlorotrimethylsilane can accomplish the successful conversion
of the model substrate 4-fluorotoluene into 4-trimethylsilylto-
luene at room temperature (99% yield; Table 1, entry 1).
other silanes, reductants, or solvents were used (entries 6−11).
Lower loadings of the catalyst, magnesium, and the
chlorosilane gave lower yields (entries 12−14). No desired
product was observed without the cobalt salt, the ligand, and
magnesium (entries 15−17). Gratifyingly, the reaction can be
conducted under ambient conditions using non-degassed and
nondried solvents (entry 18). However, the addition of water
decreases the yield (entry 19), which could be partially
recovered using an additional 1 equiv of magnesium (entry
20). Finally, 1 equiv of 9,10-dihydroanthracene, a common
radical scavenger,19 did not affect the reactivity (entry 21),
providing a piece of evidence against the radical mechanism.
The scope of the reactions of both aryl fluorides and
chlorosilanes to form aryl silanes is represented in Figure 2.
Although isolation of the model product 2a gave a diminished
yield because of its volatility (bp of 2a = 192.5 °C20), other
simple aryl fluorides (1b and 1c) underwent efficient silylation
to give 2b and 2c, respectively, under the reaction conditions.
A sterically hindered aryl fluoride as well as electron-rich
substrates gave good yields (2e−h). Notably, heterocycle-
containing aryl fluorides were also effective substrates for the
reaction (2i−k). The reaction has fair functional group
tolerance, giving reasonable yields with substrates bearing
neopentylglycolatoboron (2l), protected ketone (2m), and
amide (2n). The method could also be applied to drug
derivatives such as blonanserin (2o), an antipsychotic drug
used for the treatment of schizophrenia, and N-protected
paroxetine (2p).
Importantly, chlorosilanes other than chlorotrimethylsilane
can be also used for the reaction. Triethoxy(aryl)silanes (2q−
u), which are useful cross-coupling partners, were accessed
from the corresponding chlorosilanes and aryl fluorides in
reasonable yields. 1-Methyl-1-arylsiletanes21 (2v−y) and
dimethyl(naphthalen-1-yl)silane (2z) were also obtained in
reasonable yields using a slightly modified reaction procedure
applicable to the large-scale synthesis (1.0 g scale for 2c, 470
mg scale for 2t, and 290 mg scale for 2z). In addition, under
the standard conditions, an aryl bromide was successfully
converted in slightly reduced yield (68%), which indicates that
this transformation is also applicable to other aryl halides.
Moreover, the reaction with 4-chloro-4′-fluorobiphenyl
showed exclusive silylation at the C(sp2)−Cl bond over the
C(sp2)−F bond, which is in good agreement with the strengths
Although we have not yet carried out in-depth studies,
several observations give the contours of the reaction
mechanism. In the absence of chlorotrimethylsilane from the
optimized conditions, the aryl fluoride is not consumed.
Interestingly, when a minimal amount of chlorotrimethylsilane
(0.1 equiv) was used to initiate the reaction, we observed full
consumption of the aryl fluoride after 22 h, as confirmed by
a persistent intermediate that comes from the aryl fluoride is
formed during the reaction. In addition, chlorosilane was found
to be an efficient activator of zinc or magnesium powder.22
Therefore, we believe that a catalytic amount TMSCl is
nessessary to initiate the reaction, possibly to activate
magnesium. This reaction mixture after the full consumption
of the aryl fluoride was subjected to several different conditions
to investigate the reactivity of the intermediate resulting from
C−F bond activation (Figure 3a, (1) and (2)). First, addition
of chlorotrimethylsilane to the mixture gave the desired
Table 1. Effects of the Reaction Parameters
a
entry
variation from the “standard conditions”
none
CoCl2 instead of Co(acac)2
Col2 instead of Co(acac)2
L2 instead of L1
yield (%)
1
2
3
99
95
32
3
4
5
6
7
8
L3 instead of L1
<1
<1
<1
<1
<1
7
Si2Me6 instead of Me3SiCl
Et3SiH instead of Me3SiCl
Me3SiOTf instead of Me3SiCl
DME, toluene, or acetonitrile instead of THF
Na instead of Mg
9
10
11
12
13
14
15
16
17
18
19
20
21
Zn or Mn instead of Mg
5 mol % Co(acac)2 and 5 mol % L1
1 equiv of Mg
2 equiv of Me3SiCl
no Co(acac)2
<1
29
57
94
<1
<1
<1
99
34
60
99
no L1
no Mg
open to air, GR-grade THF
open to air, GR-grade THF + 0.28 equiv of H2O
19 + 1 additional equiv of Mg
1 equiv of 9,10-dihydroanthracene added
a
GC yields using dodecane as an internal standard.
Co(acac)2 is an optimal cobalt salt among others that were
tested (entries 1−3). The structure of ligand L1 is important
for the reactivity, and less than 5% yield of the desired product
was obtained with other 1,3-diketiminate ligands (entries 4 and
5). Further density functional theory (DFT) calculations on
the structural dependence of the ligands18 also supported the
experimental results, showing that L1 has a lower activation
barrier (∼2.4 kcal/mol) for the C−F bond activation than L2
sium, and THF solvent are crucial for the reactivity, and less
than 10% yield of the desired product was observed when
B
Org. Lett. XXXX, XXX, XXX−XXX