ACS Catalysis
Research Article
a
conversion to 13a when using NEt3 and pyridine (entry 9 and
10) with 59 and 52%, respectively, and recovery of the starting
materials with K2CO3 (entry 11) after 18 h at 80 °C. These
results show that a slightly bulkier, aromatic organic base like
2,6-lutidine is most effective.29
Table 2. Optimization Reduction of Carbonyls with Silanes
2.3. Mechanistic Studies. We initiated our mechanistic
investigations by analyzing the catalytic reaction mixture, using
in situ-prepared 3·BArF, by NMR spectroscopy (Scheme 2a).
After 30 min, our standard reaction in CD3CN only shows the
starting material (H2SiMePh 12, −16.9 ppm) and product
(13a, 8.3 ppm) in the 29Si{1H} NMR spectrum. However, a
plethora of species are observed in the 31P{1H} NMR
F
spectrum, with the major species being 3·BAr at −108.5
ppm, along with P2Ph4 at −17.8 ppm, (E)-(1,2-diphenylvinyl)-
diphenylphosphine at 7.1 ppm,9 smaller amounts of (Z)-(1,2-
diphenylvinyl)diphenylphosphine at −7.4 ppm,30 and benz-
hydryl diphenylphosphinate at 36.5 ppm.31 While phosphines
are known to facilitate hydrofunctionalization of carbonyls,32
noteworthy is the lack of catalytic activity observed using 10
mol % HPPh2 or P2Ph4.
b
b
entry
catalyst (mol %)
temp.
13a
14a
F
1
2
3
4·BAr (10)
4·BAr (10)
4·BAr (10)
3·BAr (10)
3·BAr (10)
4·BAr (5) + 2,6-lutidine (5)
3·BAr (5) + 2,6-lutidine (5)
r.t.
28
84
58
21
16
16
15
75
F
80 °C
80 °C
80 °C
80 °C
80 °C
60 °C
60 °C
80 °C
80 °C
80 °C
80 °C
F
b
F
A stoichiometric reaction to form 3·BAr with subsequent
F
4
bc
,
addition of 2,6-lutidine and 12 results in two new species by
29Si{1H} NMR spectroscopy. The major species is identified as
Me2SiHOTf at 23 ppm, formed by the facile cleavage of the
silane phenyl group.33 Analysis of the mixture by 31P{1H}
NMR spectroscopy shows 3·BArF, P2Ph4, and vinyl phosphine
as per the catalytic reaction, along with HPPh2 with its
characteristic peak at −40 ppm (Scheme 2b). Finally, when
performing the same stoichiometric study in the presence of
benzophenone 11 and analyzing the mixture after 10 min by
29Si{1H} NMR spectroscopy, we observe 13a at 8.3 ppm but
no traces of silyl triflate (Me2SiHOTf or Me2PhSiOTf).33 The
latter mixture was also analyzed by 31P{1H} NMR spectros-
copy, which shows the presence of 3·BArF, vinyl phosphine, and
P2Ph4, but no traces of HPPh2 are visible at any point of the
analysis (Scheme 2c).
F
5
5
6
7
8
9
>99
>99
>99
>99
c
F
c
F
d
e
F
9·BAr (5) + 2,6-lutidine (5)
F
10·BAr (5) + 2,6-lutidine (5)
F
3·BAr (10) + NEt3 (10)
3·BAr (10) + pyridine (10)
3·BAr (10) + K2CO3 (10)
59
52
F
10
11
F
a
Reaction condition: (i) activation step R2P(O)H (5−10 mol %),
F
Tf2O (5−10 mol %), NaBAr (5−10 mol %), 1 (5−10 mol %),
CD3CN (0.41 M), 60 °C, and 30 min; (ii) benzophenone 11 (0.25
mmol), dimethylphenyl silane 12 (0.25 mmol), 18 h, and conversion.
b
c
d
e
Isolated pre-catalyst. 5 h. 1 h. >99% after 58 h.
When performing hydrosilylation to form 13a using 10 mol
species are used, such as 3·OTf or 4·OTf, only deoxygenation
occurs with full conversion into 14a; this is due to hydrogen
release from silane reacting with co-catalytic TfOH (formed
from Tf2O).28 Thus, in line with Miura’s recent studies on
phosphirenium-derived heterocycles,9,11 we also note that the
use of a base is crucial to quench TfOH released during pre-
catalyst formation and therefore increase the selectivity toward
comprehensive benchmarking and optimization were under-
taken demonstrating that not only does TfOH lead exclusively
to the deoxygenation product 14a but also that superior
hydrosilylation catalysis takes place using a phosphirenium pre-
catalyst compared to the equivalent reaction using TfOPPh2. A
total of 5 mol % 2,6-lutidine was added to 5 mol % in situ-
F
% 9·BAr pre-catalyst and 2,6-lutidine, we also observe the
formation of substituted vinyl phosphine [i.e., (E)-(1,2-bis(4-
(trifluoromethyl)phenyl)vinyl)diphenylphosphine] in situ. Hy-
F
drosilylation to form 13a using 10 mol % 10·BAr pre-catalyst
and 2,6-lutidine gives a complex array of peaks in the 31P NMR
spectrum, making identification difficult although likely
including vinyl phosphines.
The use of 20 mol % (E)-(1,2-diphenylvinyl)-
diphenylphosphine and 20 mol % 2,6-lutidine in a catalytic
reaction of 11 and 12 does not give an appreciable quantity of
13a after 18 h at 80 C, indicating that the vinyl phosphine is
not an active catalyst in the hydrosilylation reaction, and this is
further supported by computational studies.34
F
Next, we undertook kinetic studies using NMR spectro-
scopic monitoring. By pre-forming the catalyst in situ, we have
analyzed the order in substrates 11 and 12 by varying the
amount of the starting material using 10 mol % 3·BArF. The
data obtained suggest first-order dependence in 11, while an
inverse order is observed for 1235 (Figure 3). Product
inhibition is not observed, with similar reaction rates obtained
when adding 0.5 equiv of product 13a to the catalytic reaction
of 11 and 12 in a 1:1 mixture (5.32 × 10−4 vs 5.63 × 10−4
mmol/dm3 min). Therefore, the negative order in silane can be
attributed to silane involvement in side-product formation or
catalyst degradation. Comparable data are obtained when using
prepared pre-catalyst 4·BAr prior to addition of substrates 11
and 12, and the selective formation of 13a is observed (5 h, 80
°C, CD3CN, entry 5). Changing to 3·BArF, with 5 mol % 2,6-
lutidine, gives good reactivity at lower temperature (entry 6),
and further improvement is seen with the electron-poor
phosphirenium species 9·BArF, for which reaction with 5 mol %
2,6-lutidine leads to full conversion into 13a at 60 °C after only
F
1 h (entry 7). In contrast, electron-rich 10·BAr reacted at a
slower rate with full conversion to the hydrosilylation product
requiring 58 h at 80 °C (entry 8). Clearly the electronics of the
alkyne used to prepare the phosphirenium pre-catalyst play a
substantial role in catalytic competency. Screening of the base
F
F
with 10 mol % 3·BAr as the catalyst of choice shows lower
isolated pre-catalyst 3·BAr in the absence of 2,6-lutidine,
5455
ACS Catal. 2021, 11, 5452−5462