Hypercoordinate ketone adducts of electrophilic π3-H2SiRR′ ligands on ruthenium as key intermediates for efficient and robust catalytic hydrosilation

Article abstract of DOI:10.1021/ja509073c

The electrophilic π³-H₂SiRR′ σ-complexes [PhBP^Ph₃]RuH(π³-H₂SiRR′) (RR′ = MePh, 1a; Ph₂, 1b; [PhBP^Ph₃]^- = [PhB(CH₂PPh₂)

Full text of DOI:10.1021/ja509073c

Article
pubs.acs.org/JACS
3
Hypercoordinate Ketone Adducts of Electrophilic η ‑H SiRR′ Ligands
2
on Ruthenium as Key Intermediates for Efficient and Robust Catalytic
Hydrosilation
Mark C. Lipke and T. Don Tilley*
Department of Chemistry, University of California, Berkeley, Berkeley, California 94720-1460, United States
*
S Supporting Information
3
ABSTRACT: The electrophilic η -H SiRR′ σ-complexes
2
Ph
3
[
[
(
PhBP ]RuH(η -H SiRR′) (RR′ = MePh, 1a; Ph , 1b;
3
2
2
Ph − −
PhBP
]
= [PhB(CH PPh ) ] ) are efficient catalysts
3
2 2 3
0.01−2.5 mol % loading) for the hydrosilation of ketones
with PhMeSiH , Ph SiH , or EtMe SiH. An alkoxy complex
2
2
2
2
Ph
31
1
[
PhBP ]Ru−OCHPh (4b) was observed (by P{ H} NMR
3
2
spectroscopy) as the catalyst resting state during hydrosilation
of benzophenone with EtMe SiH. A different catalyst resting
2
Ph
2
state was observed for reactions using PhMeSiH or Ph SiH , and was identified as a silane σ-complex [PhBP ]RuH[η -H−
2
2
2
3
SiRR′(OCHPh )] (RR′ = MePh, 5a; Ph , 5b) using variable temperature multinuclear NMR spectroscopy (−80 to 20 °C). The
2
2
1
hydrosilation of benzophenone with PhMeSiH and 1a was examined by H NMR spectroscopy at −18 °C (in CD Cl ), and this
2
2
2
revealed that either 1a, 5a, or both 1a and 5a could be observed as resting states of the catalytic cycle, depending on the initial
PhMeSiH ]:[benzophenone] ratio. Kinetic studies revealed two possible expressions for the rate of product formation,
[
2
depending on which catalyst resting state was present (rate = k [PhMeSiH ][5a] and rate = k′ [benzophenone][1a]).
obs
2
obs
Computational methods (DFT, b3pw91, 6-31G(d,p)/LANL2DZ) were used to determine a model catalytic cycle for the
hydrosilation of acetone with PhMeSiH . A key step in this mechanism involves coordination of acetone to the silicon center of
2
1
a-DFT, which leads to insertion of the carbonyl group into an Si−H bond (that is part of a Ru−HSi 3c−2e bond). This
generates an intermediate analogous to 5a (5a-i-DFT), and the final product is displaced from 5a-i-DFT by an associative
process involving PhMeSiH2.
5
INTRODUCTION
B). For these mechanistic proposals, attack of the ketone
substrate at silicon results in heterolytic cleavage of the
coordinated Si−H bond. This transfers a silyl cation to the
■
Transition metal catalyzed carbonyl hydrosilation reactions are
useful for the reduction of ketones, aldehydes, and esters under
+
1
ketone substrate to produce an [R Si−OCR ] species,
3
2
mild conditions. There is considerable interest in under-
standing the mechanisms of these transformations, and a variety
of catalytic cycles have been proposed.
mechanistic proposals feature the attack of a ketone substrate at
an electrophilic silicon center in the coordination sphere of the
transition metal catalyst (Scheme 1). For example, it has been
proposed that cationic rhodium complexes might activate
secondary silanes to generate a silylene dihydride complex as a
which can then accept a hydride from the metal. Experimental
and computational investigations have presented support for
2
−5
Recently developed
5
this type of mechanism, and related pathways might be
important in hydrosilations of other substrates (e.g., nitriles,
amides, pyridines), which exhibit high selectivities for specific
7a,b
7c
products (e.g., N-silylimines,
N-silyl amines, N-silyl-1,4-
7d
dihydropyridines, respectively). Considering the apparent
variety of possible carbonyl hydrosilation mechanisms involving
electrophilic silicon species, and the general lack of detailed
mechanistic understanding, it is important to define specific
pathways of this type in more detail. A deeper understanding of
these mechanisms could provide insight into the selectivities of
hydrosilation reactions (e.g., enantioselectivity or 1,2- vs 1,4-
key intermediate responsible for binding the ketone substrate
4
(
path A). Investigation of this mechanistic hypothesis has been
somewhat hampered by the fact that silylene complexes have
not been isolated or otherwise clearly detected for the relevant,
1d,4
rhodium-based catalysts.
other transition metals exhibit high reactivities toward carbonyl
compounds, and this includes examples of catalytic ketone
Isolated silylene complexes of
3
,4
regioselectivity for α,β-unsaturated ketones), as well as aid in
the development of new hydrosilation reactions and improved
catalysts.
6
hydrosilation reactions. However, detailed mechanistic inves-
tigations are lacking for these catalytic systems, and thus the
silylene complexes have not been confirmed as participants in
the catalytic cycles for these reactions.
3
With these factors in mind, we examined electrophilic η -
Ph
3
H SiRR′ σ-complexes of the type [PhBP ]RuH(η -H SiRR′)
2
3
2
Cationic iridium and ruthenium R Si−H σ-complexes have
3
also been suggested to participate as key electrophilic
intermediates in catalytic ketone hydrosilation reactions (path
Received: September 3, 2014
©
XXXX American Chemical Society
A
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Scheme 1. Involvement of Electrophilic Silicon in Transition
Metal Catalyzed Hydrosilation Reactions
Table 1. Hydosilation of Benzophenone Catalyzed by 1a
(
RR′ = PhMe, 1a; Ph , 1b; Scheme 1, path C) as possible
a
b
1
2
At 23 °C in C D . Determined by H NMR using a C Me internal
6 6 6 6
c d
catalysts for ketone hydrosilation. This was particularly
interesting in light of the discovery that 1a,b react with Lewis
bases to form unusual adducts of the type [PhBP 3]Ru(μ-
standard. Generated in situ from PhMeSiH and 2. Neat PhMeSiH
as solvent. At 60 °C in C D . At 23 °C in CD Cl . PhMe-
(Ph HCO)SiH substrate generated in situ as an intermediate in the
2
reaction of PhMeSiH with 2 equiv of benzophenone.
2
2
e
f
g
6 6 2 2
Ph
8
H) SiRR′(base), and that adducts of this type are important
2
3
intermediates for a stoichiometric hydrosilation reaction
9
involving XylNC (Xyl = 2,6-Me C H ) and 1a,b. Indeed, as
shown here, complexes 1a,b are efficient catalysts for the
hydrosilation of electron-rich and electron-poor ketones. These
are the first examples of catalytic transformations involving η -
2
6
3
the yellow color had faded substantially (back to pale yellow)
and the H NMR spectrum of the mixture revealed quantitative
formation of the expected 1,2-hydrosilation product PhMeH-
1
3
Si−OCHPh (Table 1, entry 1). Identical results were
2
H SiRR′ σ-complexes, and a detailed mechanistic investigation
2
obtained when 1a was generated in situ from PhMeSiH and
2
3
Ph
was undertaken to examine the role of the η -H SiRR′ ligands
2
{[PhBP ]Ru(μ-Cl)} (2) prior to addition of benzophenone
3
2
in this catalysis. These studies reveal that the binding of a
ketone to the silicon center of 1a,b activates the ketone toward
insertion into a partially activated (coordinated) Si−H bond.
Notably, this mechanism exhibits key features of both the
silylene mechanism (i.e., involvement of an electrophilic silicon
(
Table 1, entry 2). Very low catalyst loadings were effective for
the hydrosilation of benzophenone with PhMeSiH . With 0.01
mol % of 1a in neat PhMeSiH , 81% conversion of
benzophenone was observed after 3 h, and quantitative yield
was achieved after 24 h (Table 1, entry 3). At an even lower
catalyst loading (0.001 mol % 1a), a 46% yield was achieved
after 48 h (Table 1, entry 4). These results (TOF = 45 s and
TON = 4.6 × 10 ) demonstrate that 1a is a highly efficient
catalyst for the hydrosilation of benzophenone.
2
2
4
center derived from the activation of two Si−H bonds) and
−1
the σ-silane mechanism (i.e., electrophilic Si−H σ-complexes as
5
4
key intermediates). Additionally, the involvement of a six-
Ph
10
coordinate silicon intermediate (e.g., [PhBP ]Ru(μ-H) Si-
3
3
3
(
RR′)←OCR″R‴) is a unique feature of the η -H SiRR′
2
The catalytic hydrosilation of benzophenone was less
mechanism for the hydrosilation of ketones.
efficient with Ph SiH (Table 1, entries 5 and 6), and did not
proceed at all with Pr SiH (entries 7 and 8). For the latter
reaction, the addition of Pr SiH to 1a initially resulted in the
displacement of PhMeSiH (by H NMR spectroscopy) and
2
2
i
2
2
RESULTS AND DISCUSSION
Hydrosilation of Ketones Catalyzed by 1a. The catalytic
hydrosilation of benzophenones with a variety of silanes was
examined for initial screening of the catalytic activity of the η -
i
■
2 2
1
2
the observation of a new Ru−H resonance that is consistent
3
Ph
3
i
1
with the formation of [PhBP ]RuH(η -H Si Pr ) ( H δ −7.45
3
2
2
Ph
8
H SiRR′ complex 1a (eq 1, Table 1). Addition of
2
ppm) as the major [PhBP ]Ru species (>90%) in solution.
3
After heating to 60 °C, this species had entirely converted into
the η -cyclohexadienyl complex [PhBP ]Ru(η -C D H) (3-
d , observed by H and P{ H} NMR spectroscopy). Thus,
5
Ph
5
3
6
6
1
31
1
11
6
it appears that for this example, catalyst deactivation by
3
i
formation of 3-d is faster than reaction of the η -H Si Pr
6
2
2
complex with benzophenone.
The hydrosilation of benzophenone was also ineffective
under these conditions when using a relatively small tertiary
benzophenone (in benzene-d ) to a pale yellow solution of
silane substrate (EtMe SiH, Table 1, entry 9). In this case, the
6
2
1
31
1
PhMeSiH (1.1 equiv) and 1 mol % 1a (in benzene-d ) resulted
formation of 3-d was evident (by H and P{ H} NMR
2
6
6
in an immediate change in color to golden yellow. After 15 min,
spectroscopy) after addition of 1a to the reactants in benzene-
B
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
d . The rapid formation of 3-d may be due to the lower
poor (4,4′-difluoro, entry 6), and thus the electronic properties
of the benzophenone substrates do not appear to substantially
affect the rate or yield of catalytic hydrosilation.
Acetophenone was also effective as a substrate, but required a
higher catalyst loading to achieve full conversion (Table 2,
entries 10 and 11), due to deactivation of the catalyst by
6
6
Ph
2
stability of [PhBP ]RuH(η -H−SiMe Et) complexes, relative
3
2
3
to the η -H SiRR′ complexes involving secondary silanes.
2
However, the hydrosilation reaction with EtMe SiH proceeds
2
in CD Cl to give the hydrosilation product in quantitative
2
2
yield (Table 1, entry 10). Similarly, 2 equiv of benzophenone
undergo quantitative hydrosilation with PhMeSiH to form
formation of 3-d . At full conversion of acetopheonone, a 90%
yield of the hydrosilation product was observed by H NMR
2
6
1
PhMeSi(OCHPh ) when using CD Cl as the solvent (Table
2
2
2
2
1
, entry 11), whereas the reaction stops at the formation of
spectroscopy, along with at least one minor organic side
product that was not identified. Cyclopentanone was also
converted to more than one product (Table 2, entry 13), and in
this case the selectivity for the hydrosilation product was even
lower (65% yield at full conversion of cyclopentanone). The 1-
cyclopentenyl silyl ether C H O−SiHMePh was identified as
PhMeHSi−OCHPh when C D is the solvent. Notably, for
hydrosilation using EtMe SiH, the reaction solution was orange
2
6
6
2
in color rather than the golden yellow color consistently
observed for reactions utilizing Ph SiH or PhMeSiH . This
2
2
2
observation suggested that different resting states might be
present for hydrosilation using tertiary silanes or secondary
silanes, and this possibility was confirmed by NMR spectros-
5
7
the other product formed (35% yield), and thus activation of
the α-C−H bonds of the ketone competes with hydrosilation in
this case. However, this was not observed for cyclopropyl
phenyl ketone, which underwent hydrosilation with similar
efficiency to that of the benzophenones (Table 2, entry 12).
Fluorinated acetophenones were also effective substrates, but
the reactions proceeded slower than with benzophenones or
acetophenone. With 1,2,3,4,5-pentafluoroacetophenone, the
reaction required a much longer time (24 h) and higher
catalyst loading (2.5 mol %) to achieve a high yield of the
hydrosilation product (Table 2, entries 14 and 15). Catalytic
hydrosilation of α,α,α-trifluoroacetophenone was even less
efficient, and required heating to 80 °C for hydrosilation at an
appreciable rate (Table 2, entry 16). The lower reactivity with
1
31
1
copy. The H and P{ H} NMR spectra of the reaction
solutions indicate that the resting state for the catalytic cycle is
Ph
an alkoxide complex [PhBP ]Ru−OCHPh (4b) when
3
2
EtMe SiH is the silane substrate, and an Si−H σ-complex of the
2
Ph
2
type [PhBP ]Ru(H)[η -H−SiRR′(OCHPh )] (RR′ = MePh,
3
2
5
a; RR′ = Ph , 5b) for the secondary silane substrates (see
2
mechanistic investigation below for complete discussion of
isolation and characterization of 4b, and in situ observation and
characterization of 5a,b).
Several additional ketones were examined as substrates for
catalytic hydrosilation with PhMeSiH , using 1a as the catalyst
2
(eq 2, Table 2). A variety of 4-substituted and 4,4′-disubstituted
these electron-deficient substrates is consistent with the
3
possible role of the electrophilic η -H SiRR′ ligand in catalysis,
2
since binding of these substrates to silicon would be less
favorable than for more nucleophilic ketones.
Mechanistic Investigations with Stoichiometric Re-
actions. The carbonyl hydrosilation reactions described above
3
are the first examples of catalytic activity involving η -H SiRR′
2
benzophenones underwent quantitative hydrosilation within 20
8
complexes (1a,b). Thus, it was of interest to examine what role
min at room temperature (1.1 equiv of PhMeSiH and 1 mol %
3
2
the η -H SiRR′ ligand might have in the mechanism of these
2
1
a in benzene-d ). The substituted benzophenones ranged
6
catalytic hydrosilations. We previously observed the ability of
from electron rich (4-methyoxy, Table 2, entry 5) to electron
XylNC to undergo 1,1-insertion into an Si−H bond of the
Ph
adducts [PhBP ]Ru(μ-H) Si(RR′)←CNXyl (RR′ = MePh,
3
3
Table 2. Hydosilation of Ketones Using PhMeSiH
Catalyzed by 1a
9
2
1a(CNXyl); RR′ = Ph , 1b(CNXyl)), and it seems possible
that a similar reaction step might play a key role in the 1,2-
a
2
b
hydrosilation of ketones (Scheme 2). However, in 1a,b-
entry
ketone (RR′CO)
p-C H F, Ph
1a (mol %)
time (h)
yield (%)
(
CNXyl), the C≡N π* orbital is well positioned to accept a
1
2
3
4
5
6
7
8
9
1
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.5
0.25
0.5
0.75
24
100
100
100
100
100
100
100
100
100
44
6
4
hydride, and this may not be true for the CO π* orbital in
p-C H Cl, Ph
1
6
4
p-C H Br, Ph
1
6
4
Scheme 2. 1,1- and 1,2-Insertions into the Si−H Bond of
p-C H Me, Ph
1
6
4
Ph
[
PhBP ]Ru(μ-H) Si(RR′)←(Substrate)
p-C H OMe, Ph
1
3
3
6
4
p-C H F, p-C H F
1
6
4
6
4
p-C H Cl, p-C H Cl
1
6
4
6
4
p-C H Br, p-C H Br
1
6
4
6
4
p-C H Me, p-C H Me
1
6
4
6
4
10
11
12
13
14
15
16
Ph, Me
1
Ph, Me
2.5
1
90
Ph, cyclopropyl
cyclopentanone
C F , Me
100
c
d
1
65 /35
1
69
90
75
6
5
C F , Me
2.5
2.5
24
6
5
e
Ph, CF3
24
a
Room temperature in C D . Determined by 1H NMR using a
b
6
6
c
d
C Me internal standard. Cyclopentyl silyl ether product Cyclo-
pentenyl silyl ether product Heated to 80 °C in C D .
6
6
e
6
6
C
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
the ketone adducts [PhBPPh ]Ru(μ-H) Si(RR′)←OCR″R‴
Complex 4b was isolated in analytically pure form following
3
3
Ph
-
(
OCR″R‴ = ketone substrate; RR′ = MePh, 1a(ketone);
an independent synthesis involving treatment of {[PhBP 3]
Ru(μ-Cl)} (2) with 2 equiv of the diphenylmethoxide salt
RR′ = Ph , 1b(ketone)). In these latter species, the ketone is
2
2
2
31
1
expected to bind via an sp hybridized lone pair on oxygen, and
NaOCHPh2 (eq 3). The P{ H} NMR spectrum of 4b
this positions the CO π* orbital such that it is orthogonal to
the nearest hydride ligand (Scheme 2). Thus, the ketone must
rotate away from its ideal bonding geometry in order for the
carbonyl group to insert into the Si−H bond of 1a,b(ketone),
and this requirement could prevent such a mechanism from
being active for the hydrosilation of ketones. Additionally, a
3
mechanism that does not involve an η -H SiRR′ complex must
2
confirmed that it is the same species observed as the possible
be possible, as is evident from the hydrosilation reactions using
1
resting state for catalysis with EtMe SiH. The H NMR
2
EtMe SiH. Thus, additional information on possible hydro-
2
spectrum of 4b (in C D ) displayed a broad resonance for the
6
6
silation mechanisms was sought, and the hydrosilation of
benzophenone was chosen for detailed examination since this
substrate cleanly provides the hydrosilation product in high
yields using PhMeSiH , Ph SiH , or EtMe SiH.
alkoxide C−H bond, and this resonance was observed at
slightly different chemical shifts depending on the concen-
1
tration of the solution ( H δ 6.52 ppm at 70 mM; δ 6.56 ppm at
2
2
2
2
1
0 mM), suggesting that 4b might exist in solution as a
During the hydrosilation with EtMe SiH, it was possible to
2
observe a new 3 P{ H} NMR resonance (δ 78.45 ppm in
1
1
monomer−dimer equilibrium. Complex 4a was previously
found to be monomeric in the solid state (by single crystal X-
ray diffraction analysis), and it has now been confirmed that 4a
CD Cl ) near that previously reported for the tert-butoxy
2
2
9
Ph
t
31
1
complex [PhBP ]RuO Bu ( P{ H} δ 80.44 ppm in CD Cl ,
3
2
2
9
is also primarily monomeric in solution as evident from a
molecular weight determination using the Signer method
4
a). This latter species reacts with EtMe SiH in benzene-d to
2 6
11
t
quantitatively form EtMe Si−O Bu and 3-d . Thus, it
2
6
(
Expected MW = 859.74 g/mol; found MW = 952 ± 95 g/mol
seemed possible that the observed intermediate species might
12
Ph
in Et O). The solution molecular weight of 4b could not be
2
be a related diphenylmethoxy complex [PhBP ]Ru−OCHPh2
4b), and that reaction of this species with EtMe SiH forms the
3
determined by this method since 4b exhibits partial
(
2
1
decomposition within 3 h in solution (by H NMR spectros-
hydrosilation product EtMe Si−OCHPh (Scheme 3). This
2
2
31
1
copy), but the similarity of its P{ H} NMR data to that of 4a
suggests that 4b is also primarily monomeric in solution.
Treatment of an orange solution of 4b (in C D ) with
Scheme 3. Possible Catalytic Cycle for Hydrosilation of
6
6
Benzophenone Using EtMe SiH
2
EtMe SiH (12 equiv) results in a fading of the orange color to
2
yellow within 1 min, and quantitative formation of 3-d and
6
1
31
1
EtMe Si−OCHPh (by H and P{ H} NMR spectroscopy,
2
2
eq 4). This establishes that 4b can react with EtMe SiH rapidly
2
enough to account for the observed catalytic hydrosilation of
benzophenone with EtMe SiH.
2
It seemed possible that the hydrosilation of benzophenone
with secondary silanes PhMeSiH and Ph SiH proceeds by a
2
2
2
mechanism similar to that proposed for EtMe SiH. However,
2
1
31
1
the H and P{ H} NMR spectra of the reaction solutions
indicated that 4b was not the catalyst resting state with these
1
secondary silanes as substrates. Instead, the H NMR spectra of
the reaction solutions (in C D ) displayed a new Ru−H
6
6
1
1
resonance ( H δ −6.03 ppm using PhMeSiH ; H δ −5.60 ppm
using Ph SiH ), but the P{ H} NMR spectra of the reaction
2
31
1
2
2
solutions contained no observable resonances. These hydride
complexes were also formed by stoichiometric reaction of 1a,b
should also generate [PhBPPh ]Ru−H, or a reactive adduct of
with benzophenone (1 equiv in C
D or CD Cl ), and are
3
6
6 2 2
Ph
this hydride species with a weakly coordinating ligand (e.g.,
identified as silane σ-complexes of the type [PhBP
3
]Ru(H)-
, 5b; eq
2
Ph
2
[
PhBP ]Ru(H)(L), L = solvent, product, EtMe SiH).
(η -H−SiRR′OCHPh
2
) (RR′ = MePh, 5a; RR′ = Ph
3
2
Insertion of benzophenone into the Ru−H bond would
regenerate 4b, thus allowing for catalytic turnover. Benzene
can compete with benzophenone for insertion into the Ru−H
bond, and rapid formation of the catalytically inactive species 3-
d would explain the lack of catalysis when using C D as the
5). For the stoichiometric reactions, complexes 5a,b were
formed in high yield (ca. 90% by H NMR spectroscopy), and
1
small amounts (ca. 10%) of RR′HSi−OCHPh
, 3-d (in
2
6
1
C D
6 6
), and 4b were also observed as products (by H and
3
1
1
P{ H} NMR spectroscopy). Complexes 5a,b are unstable to
6
6
6
solvent and EtMe SiH as the silane.
decomposition or reaction with benzophenone (see below),
2
D
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
and these processes may be responsible for formation of the
minor products that were observed.
Complex 5a was also formed by treatment of [PhBPPh3]Ru-
O Bu) (4a) with the isolated hydrosilation product
t
(
PhMeHSi−OCHPh (4 equiv) in CD Cl (eq 6; note that
2
2
2
only 2 equiv of alkoxysilane are consumed, but that an excess
was used to increase the reaction rate). In this reaction, one
equiv of PhMeHSi−OCHPh is consumed by reaction with
2
t
1
4
a to form PhMe(Ph HCO)Si−O Bu (observed by
H
2
Ph
NMR spectroscopy) and [PhBP ]RuH, which is then trapped
3
by PhMeHSi−OCHPh to form 5a. Since this route to 5a
2
starts with the fully formed hydrosilation product PhMeHSi−
OCHPh , it supports the possibility that 5a,b are Si−H σ-
2
complexes of the hydrosilation products bound to the
Ph
[
PhBP 3]RuH fragment. Complexes 5a,b were unstable to
decomposition (see below), and this prevented isolation of
these potential intermediates.
Additional evidence for the identity of 5a,b was obtained by
multinuclear NMR experiments. The presence of Ru−HSi
2
9
1
bonding in 5a,b is evident from Si− H J-coupling observed in
the 29Si-filtered H and Si− H HMBC NMR spectra of
1
29
1
1
29
samples generated in situ in CD Cl ( H δ −6.39 ppm, Si δ 22
2
2
Figure 1. Upfield 1H NMR region (−5.7 to −7.7 ppm) for 5a
1
29
ppm, J = 50 Hz, 5a; H δ −5.97 ppm, Si δ 20 ppm, J
=
SiH
SiH
1
1
3
collected at −80 C. (a) H NMR spectrum. Note that the integral for
5
1 Hz, 5b). The observed J values are time-averaged for
1
31
SiH
the most upfield resonance was arbitrarily chosen as 1 H. (b) H{ P}
NMR spectrum. (c) Si− H HMBC NMR spectrum.
two hydride ligands, as is evident from integration of the Ru−H
29
1
1
resonances in the H NMR spectra for 5a,b. The presence of
only two hydride ligands suggests that the third hydride ligand
of 1a,b had transferred to benzophenone, and this was
spectroscopy) consists of two closely overlapping resonances of
1
1
31
supported by the H NMR spectra of 5a,b (in CD Cl ),
equal height ( H{ P} δ −6.10, −6.14 ppm, Figure 1b). The
2
2
1
which display resonances for the methine C−H of the
Ru−H resonance observed at −6.54 ppm (by H NMR
1
−
OCHPh2 group ( H δ 6.17 ppm, 5a; 6.40 ppm, 5b).
spectroscopy) also appears to correspond to two different Ru−
Additionally, this C−H signal exhibits weak J-coupling to the
H resonances that happen to overlap. This is evident from the
2
9
1
29
29
1
Si resonance (J < 3 Hz, by H− Si HMBC NMR), and
Si− H HMBC NMR spectrum (−80 °C, Figure 1c), which
SiH
2
9
this is consistent with the 3-bond coupling for an Si−OC−H
displays this Ru−H resonance as coupled to two different Si
29 1
1
4
1
moiety. This C−H resonance is not observed in the H NMR
NMR resonances ( Si δ 14, 22 ppm; H δ −6.54 ppm). Note
that the coupling of an apparent Ru−H resonance to two
Ph
3
2
spectrum for a deuterium labeled sample [PhBP ]Ru(D)(η -
Ph
-
29
DSiMePh(OCDPh )) (5a-d ) prepared from [PhBP 3]
different Si resonances could also indicate the presence of a
2
3
3
RuD(η -D SiMePh) (1a-d ), and this provides confirmation
ruthenium complex possessing two inequivalent silicon centers,
but this possibility is ruled out by the high yield for the
formation of 5a from 1a, which possesses only one silicon. A
total of three 29Si NMR resonances were observed and each
2
3
that this C−H bond is derived from one of the hydrides of 1a.
Additional information about the identity of 5a was obtained
from variable temperature NMR experiments on a solution of
1
29
5
a (prepared in situ in CD Cl ). Notably, in the H NMR
couples to a different pair of Ru−H resonances ( Si δ 14 ppm,
2
2
1
29
1
spectra, the Ru−H resonance exhibits coalescence at −30 °C
H δ −6.54, −7.38 ppm; Si δ 22 ppm, H δ −6.54, −6.93
29 1
1
and at −80 °C there are four Ru−H resonances observed ( H δ
ppm; Si δ 30 ppm; H δ −6.10, −6.14 ppm). These data are
consistent with the presence of three different isomers of 5a
that each feature two inequivalent Ru−HSi linkages.
−
6.12, −6.54, −6.93, −7.38 ppm, Figure 1a). These four
observed resonances were determined to actually correspond to
The three observable conformational isomers of [PhBP ]-
Ph
six total Ru−H resonances, some of which are overlapping. The
3
1
31
2
H{ P} NMR spectrum (−80 °C) revealed that the apparent
Ru(H)[η -H−SiMePh(OCHPh )] (5a) may arise from differ-
ent rotational conformations of the SiMePh(OCHPh ) group
2
1
Ru−H resonance observed at −6.12 ppm (by H NMR
2
E
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
5a-i, 5a-ii, 5a-iii, eq 7). Computational model isomers of 5a-i
Article
(
the ruthenium center. As a result, these NMR data provide
strong support for the identity of 5a even though this species
and 5a-ii (with the OCHPh group replaced by OCHMe )
2
2
could not be isolated and fully characterized.
1
A second Si−CH resonance in the H NMR spectrum of 5a
3
(
δ −0.59 ppm) integrates in a 3:1 ratio with another Ru−H
1
resonance ( H δ −6.93 ppm), and these two resonances exhibit
29
29
1
coupling to the same Si nucleus in the Si− H HMBC NMR
2
9
spectrum ( Si δ 22 ppm, Figure 1c, see Supporting
Information for expanded spectrum that includes the Si−CH
3
were examined by DFT geometry optimization calculations and
resonance). Note that for each of the two observed Si−CH
3
1
are predicted to be very similar in energy (5a-i-DFT and 5a-ii-
resonances ( H δ −0.59, −2.73 ppm), there should be two
corresponding Ru−H resonances (i.e., a total of four Ru−H
resonances for the two isomers), and that the overlapping Ru−
15
DFT, ΔGDFT = −0.66 kcal/mol, Figure 2). The higher energy
1
H resonances observed at −6.54 ppm (by H NMR
spectroscopy) integrate appropriately to correspond to the
1
remaining two expected Ru−H resonances (i.e., the H NMR
signal at −6.54 ppm integrates as equal to the sum of the
1
integrals for the H NMR resonances at −6.93 and −7.38 ppm,
Figure 1a). The Si−CH resonance expected for the third
3
1
isomer of 5a could not be located in the H NMR spectrum,
and this might be due to overlap of this methyl signal with
1
other resonances in the 0.7−1.8 ppm region of the H NMR
spectrum.
Complexes 5a,b are unstable in solution and decompose
1
31
1
within 12 h (by H and P{ H} NMR spectroscopy) to
provide the hydrosilation products RR′HSi−OCHPh in
2
quantitative yield (Scheme 4). Complex 3-d was the major
6
Scheme 4. Comparison of the Reactivity of 5a and 5b
Figure 2. DFT models for isomers of the intermediate 5a. Note that
i
the OCHPh group was truncated to O Pr. (a) 5a-i-DFT. (b) 5a-ii-
2
DFT. Note that the agostic C−H → Ru interaction and the relatively
weak Ru−H → Si interaction are indicated by narrow lines.
isomer (5a-i-DFT) features two similar Ru−HSi interactions
1.64, 1.63 Å), while these
Ru−H
(d
= 1.89, 2.01 Å; d
Si−H
interactions are highly unsymmetrical for the other isomer
dSi−H = 1.71, 2.27 Å; dRu−H 1.70, 1.63 Å; 5a-ii-DFT). This
latter isomer also features an agostic interaction between
ruthenium and a C−H bond of the Si−CH group (d
(
=
Ru−H
3
2
.14 Å, 5a-ii-DFT).
organometallic product of 5a,b decomposition in C D , and
6
6
Notably, for 5a-ii-DFT, the corresponding isomer of 5a (5a-
this indicates that loss of the silane product generates a reactive
1
Ph
ii, eq 7) could be identified by the presence of an upfield H
NMR resonance (at −80 °C) that displays coupling to an
[PhBP ]Ru−H species that then undergoes addition to the
3
C D solvent. Notably, complexes 5a,b react with excess
6
6
1
3
13
1
upfield C NMR resonance in the C− H HSQC spectrum
benzophenone (4 equiv, Scheme 4) to give distinctly different
products. Treatment of 5a with benzophenone results in nearly
1
13
1
(
H δ −2.73 ppm, C δ −50.3 ppm). This H NMR resonance
integrates in a 3:1 ratio with the Ru−H peak observed at −7.38
complete consumption of 5a after 1 h and formation of 3-d in
6
1
ppm (by H NMR spectroscopy, −80 °C), and this is
95% yield, along with the dialkoxysilane hydrosilation product
1
consistent with the agostic C−H bond being part of a Si−
PhMeSi(OCHPh ) (by H NMR spectroscopy). This latter
2
2
CH group, with rapid exchange between all three C−H
product appears to result from reaction of the initial
hydrosilation product PhMeHSiOCHPh2 with 4b (formed
3
positions. Notably, the agostic C−H interaction completes an
1
8 electron count for the ruthenium center, and this would
after displacement of PhMeHSiOCHPh by benzophenone).
2
seem to preclude the presence of additional ligands bound to
Consistent with this possibility, the reaction of 5b with
F
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
benzophenone resulted in the formation of 4b and the
concentrations of 0.15−0.20 M) to a solution of 1a and
1
hydrosilation product Ph HSiOCHPh after 4 h (by
H
PhMeSiH . Over the course of the reaction, the Ru−H
2
2
2
NMR spectroscopy). Thus, it appears that the hydrosilation
product may be displaced from ruthenium by an associative
mechanism involving the binding of benzophenone to 5a,b
prior to dissociation of the product. The subsequent formation
of 4b is consistent with a hydrosilation mechanism that is
analogous to that for tertiary silanes, in which the C−H bond
forming step involves insertion of the carbonyl group into a
reactive Ru−H bond to generate an alkoxy complex as a key
intermediate.
resonance for 1a moves steadily downfield, and returns to its
normal value upon complete consumption of benzophenone.
Additionally, with a much larger loading of benzophenone
(0.70 M) the Ru−H resonance was observed even further
1
upfield ( H δ −8.08 ppm). This apparent dependence of the
Ru−H chemical shift of 1a on the concentration of
benzophenone suggests that this resonance might result from
a rapid interconversion of 1a and its benzophenone adduct
Ph
[PhBP ]Ru(μ-H) Si(MePh)←OCPh (1a(OCPh )),
3
3
2
2
Additional experiments suggest that the hydrosilation
mechanism for secondary silanes is distinct from that of
tertiary silanes. First, complex 5a reacts much more rapidly with
eq 8). Note that a similar concentration-dependent perturba-
PhMeSiH2 (2 equiv, <1 min, Scheme 4) than with
3
benzophenone, and this regenerates the η -H SiMePh complex
2
1
a in quantitative yield. Furthermore, stoichiometric reactions
of 4b with secondary silanes (Scheme 5) rule out the possibility
Scheme 5. Reactions of 4b with Secondary Silanes
tion of the Ru−H resonances of 1a has previously been
observed as a consequence of equilibration between 1a and a
8
weakly associated THF adduct, 1a(THF) (eq 8). Unfortu-
nately, 1a(OCPh ) could not be more clearly identified due
2
to the weak binding of benzophenone to 1a, and the rapid
conversion of 1a(OCPh ) to 5a.
2
Interestingly, 1a and 5a were simultaneously observed (by
H NMR spectroscopy) during catalytic reactions monitored at
1
−
18 °C. The ratio of the concentrations of these ruthenium
species was dependent on the relative concentrations of the
substrates PhMeSiH and benzophenone. With a large initial
2
excess of benzophenone (5 equiv relative to PhMeSiH ), the
2
initial concentration of 5a was ca. 6 times larger than that of 1a
1
(
determined by H NMR spectroscopy). Conversely, an initial
excess of PhMeSiH2 (5 equiv relative to benzophenone)
resulted in observation of a larger concentration of 1a (ca. 5
times more 1a than 5a). Thus, the ratio [1a]:[5a] appears to be
that 4b is an intermediate in the catalytic hydrosilation
reactions using PhMeSiH or Ph SiH . Treatment of 4b with
2
2
2
PhMeSiH (2 equiv in C D ) results in an immediate color
directly proportional to the [PhMeSiH
Further support for this possibility was obtained by monitoring
catalytic reactions starting with other initial ratios of the
2
]:[OCPh
2
] ratio.
2
6
6
change from orange to yellow and formation of the expected
silyl ether in quantitative yield. Complex 3-d was the major
6
organometallic product (75%) and 1a was observed as a minor
product (25%). The treatment of 4b with Ph SiH (5 equiv)
substrates (e.g., initial [PhMeSiH
2
]:[Ph
2
CO] ratios of 17:10,
10:11, and 10:15, see Supporting Information). However, the
low concentrations of each ruthenium species and the
broadness of the Ru−H resonances prevented precise
integration of the Ru−H resonances for 1a and 5a. Thus, the
exact quantitative dependence of the ratio [1a]:[5a] on the
concentration of the reactants could not be definitively
established.
2
2
provided similar results (90% yield of Ph HSiOCHPh and 3-
2
2
d6), except that 5b was formed as the minor product (10%)
rather than 1b. Thus, reactions of secondary silanes with 4b
represent an effective route for the formation of hydrosilation
products, but these results demonstrate that this pathway would
result in rapid deactivation of the catalyst in C D by the
6
6
formation of 3-d . Since the rapid formation of 3-d was not
Given the dependence of the [5a]:[1a] ratio on the ratio of
6
6
observed under catalytic conditions for most substrates, an
alternate mechanism must be considered to account for the
high turnover numbers for catalysis with secondary silanes.
Mechanistic Investigation by Kinetic analyses. The
hydrosilation of benzophenone using PhMeSiH (in CD Cl )
[Ph
2
CO] to [PhMeSiH
2
], the kinetics of the reaction were
1
examined by H NMR measurements (at −18 °C) using 5:1
and 1:5 initial ratios of the reactants (Figure 3). Under
conditions of excess benzophenone, complex 5a was the major
Ph
Ph
[PhBP
]Ru species observed (85−100% of [PhBP
3
]Ru by
3
2
2
2
1
1
was monitored by H NMR spectroscopy at −18 °C using 1a as
the catalyst. At this temperature, both 5a and 1a were observed
by H NMR spectroscopy) within 1 min of mixing the
H NMR spectroscopy), and the reaction exhibited first order
dependence on the concentration of the limiting reactant
1
(
PhMeSiH (Figure 3a). Under these conditions, the reaction
2
reactants, whereas only 5a was observed as a resting state for
was also determined to exhibit first order dependence on the
concentration of 5a (Figure 3a), and thus an overall second-
order rate law was determined (eq 9). In contrast, under
1
catalytic reactions examined at 23 °C. Interestingly, the H
NMR resonance for the Ru−H groups of 1a (δ −7.26 ppm in
CD Cl at −18 °C) shifts slightly upfield (to ca. δ −7.5 ppm)
conditions of excess PhMeSiH , the Ru−H resonance for 1a
2
2
2
Ph
immediately after the addition of benzophenone (typical
was observed to account for >85% of [PhBP ]Ru present (by
G
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 6. Proposed Catalytic Cycle for Hydrosilation of
Benzophenone with PhMeSiH and 1a as a Catalyst
2
high concentrations of benzophenone relative to PhMeSiH2,
the 1,2-insertion step appears to occur rapidly enough that the
rate limiting step becomes displacement of the product silane
Figure 3. Kinetic data from catalytic hydrosilation reactions. (a) Plots
of −ln([PhMeSiH ]/[PhMeSiH ] ) versus time for 3 different catalyst
(
(
PhMeHSi−OCHPh ) from 5a by the reactant silane
2
2
2 0
PhMeSiH ) to regenerate 1a, and the rate law becomes that
loadings and excess benzophenone (5 equiv). (b) Plots of −ln-
2
of eq 9. The rate dependence on k′ [5a][PhMeSiH ] (eq 9)
(
[benzophenone]/[benzophenone] ) versus time for 3 different
obs
2
0
catalyst loadings and an excess of PhMeSiH (5 equiv).
indicates that the binding of PhMeSiH2 to 5a assists in
2
displacement of the product from ruthenium, and this
associative pathway for product/silane exchange avoids the
1H NMR spectroscopy), and the reaction was found to exhibit
first order dependence on the concentration of benzophenone
and 1a (second order overall, Figure 3b and eq 10). Note that
eqs 9 and 10 could conceivably result from pseudo-first-order
simplifications of a third-order rate law involving both
substrates (i.e., d[P]/dt = k″ [PhMeSiH ][Ph CO][Ru]),
Ph
formation of the free 14 electron [PhBP ]Ru−H species that
3
could be responsible for undesired side reactions (e.g.,
formation of 3-d in C D ). Thus, this mechanism explains
6
6
6
why catalysis occurs for the secondary silanes in C D , but not
6
6
for tertiary silanes, which cannot lead to the formation of
intermediates analogous to 5a,b.
obs
2
2
but this possibility was eliminated by an additional experiment
using nearly equal amounts of each substrate (initial
Computational Investigation of the Catalytic Cycle.
The proposed catalytic cycle was examined by DFT calculations
15
[
PhMeSiH ]:[Ph CO] ratio of 10:11; see Supporting
on the hydrosilation of acetone with PhMeSiH (Scheme 7).
2
2
2
Information).
A DFT model of the complete structure of 1a (1a-DFT,
G1a‑DFT = 0) was used as the starting point for the catalytic
cycle. The binding of acetone to the silicon center of 1a-DFT
to form the adduct 1a-ace-DFT is predicted to be endergonic,
d[P]
dt
=
k [PhMeSiH ][5a]
obs 2
(9)
but only by a small amount (ΔG
= +5.1 kcal/mol).
1
a‑aceDFT
d[P]
dt
1
=
k′obs[Ph CO][1a]
This is consistent with the experimental observation that the H
NMR chemical shift for the Ru−H resonance of 1a exhibits
small changes depending on the concentration of benzophe-
none, which suggests an equilibrium between 1a and a weakly
2
(10)
The observation of two apparent catalyst resting states (1a
and 5a) and the determination of two distinct rate laws (eqs 9
and 10) indicate that the catalytic cycle includes two steps that
are similar in energy, such that either step can be rate limiting
depending on the ratios of the reactants used. Thus, the two
rate laws that were determined provide information about two
distinct steps in the catalytic cycle, and these rate laws are
consistent with the catalytic cycle depicted in Scheme 6. This
mechanism starts with the binding of benzophenone to the
bound benzophenone adduct 1a(OCPh ). For 1a-ace-DFT,
2
2
3
the Si−OC angle is wider than expected for an sp or sp
hybridized oxygen (Si−OC angle = 138.19°, Figure 4a). In
8,9
contrast, other Lewis bases (i.e., DMAP, PMe , XylNC) bind
3
to the silicon center of 1a,b with a more ideal geometry about
the donor atom (for example, 1b(DMAP) exhibits Si−N−C
2
angles of 120.2(2)° and 123.8(2)° for the sp hybridized donor
Ph
3
8
silicon center of [PhBP ]RuH(η -H SiMePh) (1a) to form
nitrogen). The wide Si−O−C angle for 1a-ace-DFT might be
3
2
the adduct 1a(OCPh ). This facilitates a 1,2-insertion of the
due to unfavorable steric interactions between the acetone
2
Ph
−
carbonyl group into the Si−H portion of an Ru−HSi 3c−2e
bond, to form the observed intermediate 5a. If the 1,2-insertion
step is rate-limiting for the catalytic cycle, then applying the
steady state approximation to the concentration of 1a(O
methyl groups and the [PhBP ] ligand. Additionally, one of
3
the Si−H distances in 1a-ace-DFT is relatively long (d
2.20, 1.83, 1.91 Å), whereas related base adducts exhibit three
=
Si−H
Ru−HSi interactions with roughly equivalent bond dis-
8
,9
CPh ) provides the rate law depicted in eq 10. However, at
tances. This difference may be due to weak binding of the
2
H
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 7. Catalytic Cycle Determined by DFT Calculations for the Hydrosilation of Acetone Using 1a
still indicates the presence of a weak Ru−H → Si interaction to
17,13b
provide silicon with a coordination number of 6.
A transition state for C−H bond formation was determined
to have an energy barrier that is readily accessible at room
temperature (ΔG = 20.2 kcal/mol). At this transition state,
TS2
all of the Si−H distances have increased (d
= 2.33, 2.06,
Si−H
1
.88 Å, Figure 4b), with the longest Si−H distance
corresponding to the hydride that is transferred to the carbonyl
group. Thus, at the transition state, the Si−H interaction
associated with the migrating hydrogen is nearly completely
1
7,13b
broken,
and this transition state resembles those expected
for transfer of a terminal metal hydride to a ketone that is
4,6
bound to a silylene ligand. The silicon center has moved
Ph
away from the central axis of the [PhBP ]Ru moiety (B−
3
RuSi angle = 166.6°, TS ; 173.2°, 1a-ace-DFT), and the
2
OCMe group has rotated so that oxygen appears to be
2
bound to silicon through the CO π-bond rather than via an
oxygen lone pair (Si−OC−C dihedral angles = 85.64°,
1
16.0°, TS ). This positions the carbonyl group for accepting
2
the hydride, but the Ru−H distance is not significantly
elongated at the transition state (d = 1.63 Å, 1a-ace-
Ru−H
DFT; dRu−H = 1.70 Å, TS ), and the C−H distance has
2
decreased considerably but is still fairly long (3.44 Å, 1a-ace-
DFT; dC−H = 1.99 Å, TS ). Thus, it appears that the energetic
2
cost for H-migration derives primarily from the repositioning of
acetone to accept the hydride, and that flexibility of the [(μ-
H) SiMePh(OCMe )] moiety allows this to occur with a
Figure 4. (a) Structure of the model acetone adduct 1a-ace-DFT
determined by DFT structure optimization. Note that the com-
paratively weak Ru−H → Si interaction is indicated with a narrow
bond line. (b) Transition state for the C−H bond forming step of the
3
2
fairly low barrier. The transfer of the hydride ligand occurs
without a significant perturbation to the other two Ru−H−Si
interactions, and this results in the ready formation of 5a-i-
DFT. Thus, activation of the ketone at silicon ultimately
accounts for the experimentally observed formation of 5a,b as
key intermediates in the hydrosilation of benzophenone with
catalytic cycle (TS ). Bonds that are breaking or forming are depicted
2
with narrow lines. Note that for both structures, the non-hydridic
hydrogens have been omitted for clarity.
ketone relative to stronger Lewis bases, such that a stronger
PhMeSiH
A transition state for the addition of PhMeSiH to 5a-i-DFT
2
was found to be associated with an energy barrier
2 2 2
or Ph SiH .
3
resemblance to the initial η -H SiMePh complex is preserved in
2
16
the acetone adduct. Note that the longest Si−H distance for
1
a-ace-DFT is too long to correspond to a σ-H−Si ligand, but
(ΔΔGTS3−5a‑i‑DFT = +24.1 kcal/mol) that is higher than that
I
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
for the hydride transfer, but only by a small amount (3.9 kcal/
Interestingly, experimental and computational results in-
dicate that an entirely different mechanism is responsible for
hydrosilation with secondary silanes. These reactions proceed
mol). Interestingly, the addition of PhMeSiH to 5a-i-DFT
2
1
forms an η -H−SiHMePh complex (Ru−H−Si angle = 169.8°,
a-PhMeSiH -DFT), which minimizes steric crowding for this
5
via binding of the ketone substrate to the electrophilic silicon
2
3
intermediate. A transition state for dissociation of the product
center of the η -H SiRR′ ligand in 1a,b, followed by transfer of
2
from 5a-PhMeSiH -DFT could not be located after several
a hydride to the ketone. These hydrosilation reactions are the
2
attempts at transition state optimization calculations starting
from slightly different initial geometries. Instead, a transition
state was located for product dissociation from a slightly higher
energy diastereomer of 5a-PhMeSiH -DFT (5a-PhMeSiH ′-
first catalytic reactions to be identified as involving electrophilic
3
η -H
SiRR′ ligands, and it is interesting that these Si−H σ-
2
complexes mediate hydrosilation by a mechanism that differs
considerably from the mechanisms proposed for electrophilic
2
2
1
2
DFT, ΔΔG
= +2.4 kcal/mol),
and this transition state is associated with a very low barrier
= +0.4 kcal/mol). Note that a
η - and η -H−SiR
3
σ-complexes, whereby the coordinated Si−
5
a′‑PhMeSiH2′‑DFT−5a′‑PhMeSiH2‑DFT
H bond is cleaved upon attack of the ketone substrate at
silicon.
5
,7b
(
ΔΔG
Instead, the mechanism for the catalysts 1a,b more
TS4‑5a′ ‑PhMeSiH2′‑DFT
transition state analogous to TS , but leading to the formation
closely resembles catalytic cycles that propose the involvement
3
of electrophilic silylene complexes that form distinct L MH-
of 5a-PhMeSiH2′-DFT (TS ), was not calculated since it
n
3′
(
SiRR′←ketone) intermediates prior to the C−H bond
could be readily estimated that this transition state would be no
4,6
forming step.
more than 2.4 kcal/mol higher in energy than TS , and that
3
The catalytic cycle for 1a,b uniquely features a hyper-
coordinate silicon intermediate (i.e., 1a,b(OCRR′ )), and
this intermediate leads to insertion of the carbonyl group into a
highly polarized Si−H bond that is part of an Ru−H → Si
interaction. This step is similar to a 1,2-insertion step that has
previously been proposed for the stoichiometric hydrosilation
TS3 would be very similar in geometry to TS . After
′
3
dissociation of the hydrosilation product, rearrangement of
Ph
2
[
PhBP ]RuH(η -HSiHMePh) should occur very rapidly to
3
3
regenerate the η -H SiMePh complex 1a, and this step was not
2
examined computationally. Note that the two highest energy
barriers in the computationally determined catalytic cycle are
associated with reaction steps that are consistent with the two
experimentally determined rate laws for the hydrosilation of
−
21
of ketones with hydridosilicate anions (e.g., [(RO) SiH] ) or
4
for for catalytic hydrosilation reactions involving M[SiR H-
2
3
(
OCR′R″)] intermediates. However, the latter mechanistic
benzophenone with PhMeSiH (eqs 9 and 10). Thus, the
2
proposal has received little experimental or theoretical
accuracy of the computationally determined catalytic cycle is
bolstered by its consistency with experimental observations
made for the catalytic ketone hydrosilation reactions.
4a
support. For the present system, the 1,2-insertion step leads
to the formation of intermediates 5a,b, in which the product is
bound to ruthenium as a σ-silane ligand. This protects the
reactive Ru−H bond from detrimental side reactions (e.g.,
addition to benzene) until a new silane substrate displaces the
product. Thus, the η -H SiRR′ ligands play a crucial role in
CONCLUSION
■
3
3
The electrophilic η -H SiRR′ σ-silane complexes 1a,b are
effective hydrosilation catalysts for a variety of ketone
substrates. Complex 1a was a particularly efficient catalyst for
2
2
activating the ketone substrate, and this has the remarkable
effect of selecting for a hydrosilation pathway that is more
robust than that available to tertiary silanes. Future work will
focus on investigating the activation of additional unsaturated
substrates by the silicon center of 1a,b, as well as further
examining the more general role of how Ru−HSi
interactions might be useful for guiding the catalytic cycle of
hydrosilation reactions.
the hydrosilation of benzophenone with PhMeSiH , which
2
could be accomplished with high turnover rates and overall
turnover numbers. Interestingly, detailed mechanistic inves-
tigations revealed that there are two distinct catalytic cycles
available for these carbonyl hydrosilation reactions depending
on the silane substrate that is used (i.e., secondary or tertiary
silanes), but that the two pathways provide similar reaction
rates (i.e., within 1 order of magnitude at room temperature).
EXPERIMENTAL DETAILS
■
For the hydrosilation of benzophenone with EtMe SiH, the
2
General Considerations. All manipulations of air sensitive
compounds were conducted under a nitrogen atmosphere using
standard Schlenk techniques or using a nitrogen atmosphere glovebox.
Proteo solvents were dried using a JC Meyer solvent drying system,
and deutero solvents were vacuum transferred from appropriate drying
agents (NaK for C D and CaH for CD Cl ). Silanes were purchased
mechanism involves insertion of benzophenone into the Ru−H
bond of a highly reactive ruthenium hydride species to generate
Ph
[
PhBP ]Ru−OCHPh (4b), which was observed as the
3
2
3
1
1
resting state of the catalytic cycle (by P{ H} NMR
spectroscopy). Complex 4b was isolated and this species was
6
6
2
2
2
8
22
from commercial sources and used as received. Complexes 1a,b, 2,
found to react rapidly with EtMe SiH to form Et-
2
9
and 4a were prepared as previously reported. Diphenylmethanol was
Me SiOCHPh and regenerate the reactive ruthenium hydride
23
2
2
prepared based on a published procedure, and was deprotonated
species. This hydrosilation mechanism is analogous to the
using NaH (in THF) to form Na[OCHPh ], which was isolated as a
1
8
2
Chalk−Harrod mechanism for catalytic olefin hydrosilations,
white powder after evaporating the solvent under vacuum.
and similar mechanisms have previously been proposed for
catalytic carbonyl hydrosilation reactions involving tertiary
silanes. However, mechanistic studies on several hydrosilation
catalysts have revealed that this mechanism is often a minor
hydrosilation pathway, and that other, more active catalytic
NMR spectra were recorded on Bruker spectrometers at room
temperature unless otherwise noted. Spectra were referenced internally
1
9
1
by the residual proton signal relative to tetramethylsilane for H NMR,
solvent peaks for 1 C{ H} NMR, external 85% H PO for P{ H}
3
1
31
1
3
4
NMR, and tetramethylsilane for 2 Si− H HMBC experiments. The J
9
1
SiH
20
values for Ru−HSi resonances were determined by examining
cycles are responsible for the majority of catalysis. Thus, it is
1
31
satellite signals near the main Ru−H resonance in H{ P} NMR
notable that 4b reacts with EtMe SiH rapidly enough to
29
1
31
2
spectra or by the Ru−H resonances displayed in Si-filtered H{ P}
confirm that the Chalk-Harrod type pathway is primarily
responsible for the hydrosilation of benzophenone with
NMR experiments. Hydrosilation products were identified by
1
13
1
29
1
comparison of multinuclear NMR data ( H, C{ H}, and Si− H
HMBC NMR) to those previously reported for identical or closely
EtMe SiH when using 1a as a precatalyst.
2
J
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
1
related silyl ethers, and by GC−MS. Elemental analyses were
from the solution. An initial H NMR spectrum of the solution was
performed by the University of California, Berkeley College of
collected at −18 °C (temperature was calibrated by an external
standard of 4% MeOH in methanol-d ). In this initial H NMR
1
Chemistry Microanalytical Facility.
4
Ph
[
PhBP ]Ru−OCHPh (4b). Complex 2 (62 mg, 0.038 mmol) and
spectrum, the Ru−H resonance for 1a is displayed as a sharp signal
3
2
Na[OCHPh ] (16 mg, 0.078 mmol) were dissolved in 4 mL of THF
and the concentration of 1a was quantified by integration of the Ru−H
2
and the resulting red solution was stirred for 40 min. After this time,
the solution was evaporated under vacuum and the resulting solid was
resonance relative to the resonance for C Me . The sample was then
6 6
i
chilled in a dry ice/ PrOH bath before briskly shaking the NMR tube
to dissolve benzophenone into the solution. The sample was
immediately transferred to the NMR probe cooled to −18 °C and
extracted with Et O (3 mL) to give a brownish-red solution, which was
2
filtered and cooled to −35 °C. After 9 days, a brown crystalline
precipitate had formed and the solution was a lighter, purer red color.
The supernatant was removed by pipet and evaporated under vacuum
to provide 4b as an analytically pure, red-orange foam (44 mg, 66%).
Anal. Calcd for C H OBP Ru (969.853): C, 71.83; H, 5.40. Found:
1
allowed 1 min to equilibrate in temperature before collection of H
NMR spectra at 12 s intervals.
Computational Details. All calculations were performed using the
Gaussian ’09 suite of programs in the molecular graphics and
computing facility of the College of Chemistry, University of
California, Berkeley. Calculations were performed using the
B3PW91 hybrid functional with the 6-31G(d,p) basis set for all
5
8
52
3
1
C, 72.16; H, 5.03. H NMR (C D , 600 MHz): δ 8.18 (d, J = 7.1 Hz, 2
6
6
H), 7.75 (m, 6 H), 7.48 (br, 12 H), 7.28 (t, J = 7.3 Hz, 4 H), 7.11 (t, J
7.3 Hz, 1 H), 7.02 (t, J = 7.5 Hz, 2 H), 7.73−7.63 (m, 18 H), 6.56 (1
=
13
1
H, RuOCHPh ), 1.61 (br, 6 H, BCH P). C{ H} NMR (C D ,
main-group elements and the LANL 2DZ basis set for ruthenium. The
2
2
6
6
Ph
1
50.893 MHz): δ 148.61, 139.22, 138.63, 132.86, 132.44, 132.08,
full [PhBP
3
]Ru fragment and SiMePh fragment were used for all
1
30.57, 128.96, 128.90, 127.72, 127.47, 125.18, 84.37 (Ru−O
calculations. Vibrational frequencies were calculated for all converged
structures and confirm that these structures are transition states (one
imaginary frequency determined) or lie on minima (no imaginary
frequencies were determined). Energies for all species are free energies
31
1
CPh H), 13.91 (br, BCH P). P { H} NMR (C D , 161.967 MHz): δ
2
2
6
6
7
9.0.
In Situ Preparation and Observation of 5a,b. Complexes 5a,b
determined relative to 1a-DFT + acetone-DFT + PhMeSiH -DFT.
were prepared by the addition of benzophenone (1 equiv) to solutions
2
Only half the entropic contributions to the free energy differences that
were determined by DFT calculations were used, which was done as
an approximate correction for the determination of dilute gas-phase
free energies by DFT calculations rather than solution-state free
energies. This results in ca. an 8 kcal/mol decrease in the relative
energy of transition states or intermediates (in comparison to the gas
phase values) formed in bimolecular processes.
of 1a,b in C D or CD Cl . Upon addition of benzophenone, the pale
yellow color of 1a,b darkens to amber yellow. The solutions were
6
6
2
2
examined by 1H, 1H{ P}, P{ H}, and Si− H HMBC NMR
31
31
1
29
1
spectroscopy. Note that 1a, 3-d , and 4b were observed as minor
6
2
4
31
1
species in solutions of 5a,b prepared in this manner. The P{ H}
NMR signals for 5a,b could not be observed for samples at room
temperature, presumably as a result of broadening due to conforma-
tional changes in solution. At lower temperatures, several new P{ H}
NMR signals were observed and are consistent with the presence of
several conformational isomers of 5a. Note that the sample of 5a used
for low temperature NMR spectroscopy contained larger impurities of
31
1
ASSOCIATED CONTENT
Supporting Information
Additional experimental and computational details, and NMR
spectra. This material is available free of charge via the Internet
at http://pubs.acs.org.
■
*
S
4
b (ca. 30% of [Ru] present) and PhMeSi(OCHPh ) than typical for
2 2
samples of 5a prepared in situ, but that this did not interfere with
obtaining key low temperature NMR data for 5a. See Figure 1 and the
Supporting Information for low temperature NMR spectra. Distin-
AUTHOR INFORMATION
Corresponding Author
tdtilley@berkeley.edu
guishing room temperature NMR data are tabulated here.
■
1
5
a. H NMR (CD Cl 500 MHz) δ 7.75 (J < 3 Hz, 2 H, Si−Ph),
2
2
SiH
6
(
.16 (JSiH < 3 Hz, 1 H, SiOCPh H), 1.44 (br, 6 H, BCH P), −0.56
2
2
29
1
3 H, Si−CH ), −6.39 (m, J = 50 Hz, 2 H, Ru−H). Si− H HMBC
3
SiH
Notes
29
NMR: Si δ 22 ppm.
The authors declare no competing financial interest.
1
5
b. H NMR (CD Cl 600 MHz) δ 6.39 (J < 3 Hz, 1 H,
2 2 SiH
SiOCPh H), 1.40 (br, 6H, BCH P), −5.97 (m, J = 51 Hz, 2 H,
2
2
SiH
ACKNOWLEDGMENTS
29
1
29
Ru−H). Si− H HMBC NMR: Si δ 20 ppm.
■
Representative Procedure for Catalytic Hydrosilation Re-
This work was funded by the National Science Foundation
under Grant No. CHE-1265674. The molecular graphics and
computational facility (College of Chemistry, University of
California, Berkeley) is supported by the National Science
Foundation under Grant No. CHE-0840505. We thank Allegra
Liberman-Martin for discussion of kinetics data.
actions. Benzophenone (20 mg, 0.11 mmol) and PhMeSiH (13.5−
2
1
6.0 mg, 0.11−0.13 mmol) were dissolved in C D (0.6 mL) with
6
6
1
C Me as an internal standard. A H NMR spectrum of the mixture
6
6
was collected prior to adding 1a,b in C D (0.1 mL). The addition of
6
6
1
1
a produces an amber yellow solution that was examined by H NMR
spectroscopy within 15 min. It was noted that fading of the amber
yellow color to a pale yellow or colorless solution appeared to coincide
1
REFERENCES
with complete consumption of benzophenone (determined by H
■
NMR). The product was isolated by diluting the reaction solution with
hexanes (1 mL), passing this solution through a plug of silica, and
(1) (a) Ojima, I.; Li, Z.; Zhu, J. The Chemistry of Organic Silicon
Compounds; Wiley: Avon, 1998; Chapter 29. (b) Rappoport, Z.;
Apeloig, Y. The Chemistry of Organic Silicon Compounds; Wiley:
Chichester, U.K., 1998. (c) Roy, A. K. Adv. Organomet. Chem. 2007,
evaporating the solvent. This provided the product in good purity
1
(
≥95%) judged by H NMR spectroscopy. See Supporting
Information for NMR spectral data and GC/MS data for all isolated
55, 1−59. (d) Riener, K.; Ho
̈
̈
gerl, M. P.; Gigler, P.; Kuhn, F. E. ACS
products.
Catal. 2012, 2, 613−621.
Representative Procedure for Low Temperature Reaction
Monitoring and Kinetics Data Collection. Complex 1a (2.7 mg,
(2) (a) Ojima, I.; Nihonyanagi, M.; Kogure, T.; Kumagai, M.;
Horiuchi, S.; Nakatsugawa, K. J. Organomet. Chem. 1975, 94, 449.
(b) Ojima, I.; Kogure, T.; Kumagai, M.; Horiuchi, S.; Sato, T. J.
Organomet. Chem. 1976, 122, 83−97.
(3) Zheng, G. Z.; Chan, T. H. Organometallics 1995, 14, 70−79.
(4) (a) Schneider, N.; Finger, M.; Haferkemper, C.; Bellemin-
Laponnaz, S.; Hofmann, P.; Gade, L. H. Angew. Chem., Int. Ed. 2009,
48, 1609−1613. (b) Schneider, N.; Finger, M.; Haferkemper, C.;
Bellemin-Laponnaz, S.; Hofmann, P.; Gade, L. H. Chem.Eur. J.
0
.003 mmol) and PhMeSiH (11 mg, 0.09 mmol) were dissolved in
2
0
.6 mL of a stock solution of CD Cl containing C Me as an internal
2
2
6
6
standard. This solution was transferred to a J-Young NMR tube, which
was then charged with a small plastic tube that was packed with solid
benzophenone, and the NMR tube was sealed with a threaded Teflon
stopper. The plastic inset with benzophenone fits snuggly at the top of
the NMR tube, thus keeping the benzophenone substrate separate
K
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
2
009, 15, 11515−11529. (c) Gigler, P.; Bechlars, B.; Ku
Chem. Soc. 2011, 133, 1589−1596.
5) (a) Gutsulyak, D. M.; Vyboishchikov, S. F.; Nikonov, G. I. J. Am.
̈
hn, F. E. J. Am.
(
Chem. Soc. 2010, 132, 5950−5951. (b) Park, S.; Brookhart, M.
Organometallics 2010, 29, 6057−6064. (c) Yang, Y.-F.; Chung, L. W.;
Zhang, X.; Houk, K. N.; Wu, Y.-D. J. Org. Chem. 2014, 79, 8856−8864.
(
6) (a) Watanabe, T.; Hashimoto, H.; Tobita, H. J. Am. Chem. Soc.
2
006, 128, 2176−2177. (b) Ochiai, M.; Hashimoto, H.; Tobita, H.
Dalton Trans. 2009, 1812−1814. (c) Ochiai, M.; Hashimoto, H.;
Tobita, H. Organometallics 2012, 31, 527−530. (d) Mitchell, G. P.;
Tilley, T. D. J. Am. Chem. Soc. 1997, 119, 11236−11243. (e) Klei, S.
R.; Tilley, T. D.; Bergman, R. G. Organometallics 2002, 21, 4648−
4661. (f) Calimano, E.; Tilley, T. D. Organometallics 2010, 29, 1680−
1692. (g) Fasulo, M.; Tilley, T. D. Organometallics 2012, 31, 5049−
5057.
(
7) (a) Gutsulyak, D. V.; Nikonov, G. I. Angew. Chem., Int. Ed. 2010,
4
9, 7553−7556. (b) Boone, C.; Korobkov, I.; Nikonov, G. I. ACS
Catal. 2013, 3, 2336−2340. (c) Cheng, C.; Brookhart, M. J. Am. Chem.
Soc. 2012, 134, 11304−11307. (d) Gutsulyak, D. V.; van der Est, A.;
Nikonov, G. I. Angew. Chem., Int. Ed. 2011, 50, 1384−1387.
(
8) Lipke, M. C.; Tilley, T. D. J. Am. Chem. Soc. 2011, 133, 16374−
1
6377.
9) Lipke, M. C.; Tilley, T. D. J. Am. Chem. Soc. 2013, 135, 10298−
0301.
10) For recent examples of ketone hydrosilation catalysts that
exhibit high activity see reference 5b and (a) Yang, J.; Tilley, T. D.
Angew. Chem. Int., Ed. 2010, 49, 10186−10188. (b) Gade, L. H.; Cesar,
V.; Bellemin-Lapponaz, S. Angew. Chem., Int. Ed. 2004, 43, 1014−
017. (c) Ruddy, a. J.; Kelly, C. M.; Crawford, S. M.; Wheaton, C. A.;
Sydora, O. L.; Small, B. L.; Stradiotto, M.; Turculet, L. Organometallics
013, 32, 5581−5588. (d) Chakraborty, S.; Blacque, O.; Fox, T.;
Berke, H. ACS Catal. 2013, 3, 2208−2217.
11) Lipke, M.c.; Neumeyer, F.; Tilley, T. D. J. Am. Chem. Soc. 2014,
36, 6092−6102.
(
1
(
́
1
2
(
1
(
12) Zoellner, R. W. J. Chem. Educ. 1990, 67, 714−715. Note that
errors provided for the molecular weight measurement on 4a are based
on those typical of this method, rather than derived from multiple
repetitions of the experiment.
(
13) (a) Nikonov, G. I. Adv. Organomet. Chem. 1995, 53, 217−309.
b) Lachaize, S.; Sabo-Etienne, S. Eur. J. Inorg. Chem. 2006, 2115−
127.
14) Liepins,
Chem. 1985, 23, 485−486.
15) All DFT calculations were carried out with Gaussian 09 using
the B3PW91 functional and 6-31G(d,p)/LANL 2DZ basis sets.
16) Grumbine, S. K.; Straus, D. A.; Tilley, T. D. Polyhedron 1995,
4, 127−148.
17) Delpech, F.; Sabo-Etienne, S.; Daran, J.-C.; Chaudret, B.;
Hussein, K.; Marsden, C. J.; Barthelat, J.-C. J. Am. Chem. Soc. 1999,
(
2
(
̌ ̌
E.; Birgele, I.; Tomsons, P.; Lukevics, E. Magn. Reson.
(
(
1
(
1
(
(
21, 6668−6682.
18) Chalk, A. J.; Harrod, J. F. J. Am. Chem. Soc. 1965, 87, 16−21.
19) (a) Chakroborty, S.; Krause, J. A.; Guam, H. Organometallics
2
009, 28, 582−586. (b) Peterson, E.; Khalimon, A. Y.; Simionescu, R.;
Kuzmina, L. G.; Howard, J. A. K.; Nikonov, G. I. J. Am. Chem. Soc.
009, 131, 908−909. (c) Issenhuth, J.-T.; Notter, F.-P.; Dagorne, S.;
Dedieu, A.; Bellemin-Laponnaz, S. Eur. J. Inorg. Chem. 2010, 4, 529−
41.
20) (a) Du, G.; Fanwick, P. E.; Abu-Omar, M. M. J. Am. Chem. Soc.
007, 129, 5180−5187. (b) Shirobokov, O. G.; Kuzmina, L. G.;
Nikonov, G. I. J. Am. Chem. Soc. 2011, 133, 6487−6489.
21) Corriu, R. J. P.; Guerin, C.; Henner, B.; Wang, Q.
2
5
(
2
(
Organometallics 1991, 10, 2297−2303.
(
(
22) Betley, T. A.; Peters, J. C. Inorg. Chem. 2003, 42, 5074−5084.
23) Wiles, C.; Watts, P.; Haswell, S. J. Tetrahedron Lett. 2006, 47,
5
(
1
261−5264.
24) Wang, T.; Liang, Y.; Yu, Z.-X. J. Am. Chem. Soc. 2011, 133,
3762−13763.
L
dx.doi.org/10.1021/ja509073c | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX