Experimental analysis of the catalytic cycle of the borane-promoted imine reduction with hydrosilanes: Spectroscopic detection of unexpected intermediates and a refined mechanism

Article abstract of DOI:10.1021/ja409344w

The discovery of intermediates that had not been seen before in imine reduction involving borane-mediated Si-H bond activation provided new insight into the mechanism, eventually leading to a refined catalytic cycle that also bears relevance to asymmetric variants. The catalysis proceeds through an ion pair composed of a silyliminium ion and a borohydride that subsequently reacts to yield an N-silylated amine and the borane catalyst. The latter step is enantioselectivity-determining when using a chiral borane. It was now found that there are additional intermediates that profoundly influence the outcome of such enantioselective transformations. Significant amounts of the corresponding free amine and N-silylated enamine are present in equimolar ratio during the catalysis. The free amine emerges from a borohydride reduction of an iminium ion (protonated imine) that is, in turn, generated in the enamine formation step. The unexpected alternative pathway adds another enantioselectivity-determining hydride transfer to reactions employing chiral boranes. The experiments were done with an axially chiral borane that was introduced by us a few years ago, and the refined mechanistic picture helps to understand previously observed inconsistencies in the level of enantioinduction in reductions catalyzed by this borane. Our findings are general because the archetypical electron-deficient borane B(C₆F₅)₃ shows the same reaction pattern. This must have been overlooked in the past because B(C ₆F₅)₃ is substantially more reactive than our chiral borane with just one C₆F₅ group. Reactions with B(C₆F₅)₃ must be performed at low catalyst loading to allow for detection of these fundamental intermediates by NMR spectroscopy.

Full text of DOI:10.1021/ja409344w

Article
pubs.acs.org/JACS
Experimental Analysis of the Catalytic Cycle of the Borane-Promoted
Imine Reduction with Hydrosilanes: Spectroscopic Detection of
Unexpected Intermediates and a Refined Mechanism
Julia Hermeke, Marius Mewald, and Martin Oestreich*
Institut fur Chemie, Technische Universitat Berlin, Strasse des 17. Juni 115, 10623 Berlin, Germany
̈
̈
S
* Supporting Information
ABSTRACT: The discovery of intermediates that had not
been seen before in imine reduction involving borane-
mediated Si−H bond activation provided new insight into
the mechanism, eventually leading to a refined catalytic cycle
that also bears relevance to asymmetric variants. The catalysis
proceeds through an ion pair composed of a silyliminium ion
and a borohydride that subsequently reacts to yield an N-
silylated amine and the borane catalyst. The latter step is
enantioselectivity-determining when using a chiral borane. It
was now found that there are additional intermediates that
profoundly influence the outcome of such enantioselective transformations. Significant amounts of the corresponding free amine
and N-silylated enamine are present in equimolar ratio during the catalysis. The free amine emerges from a borohydride
reduction of an iminium ion (protonated imine) that is, in turn, generated in the enamine formation step. The unexpected
alternative pathway adds another enantioselectivity-determining hydride transfer to reactions employing chiral boranes. The
experiments were done with an axially chiral borane that was introduced by us a few years ago, and the refined mechanistic
picture helps to understand previously observed inconsistencies in the level of enantioinduction in reductions catalyzed by this
borane. Our findings are general because the archetypical electron-deficient borane B(C6F5)3 shows the same reaction pattern.
This must have been overlooked in the past because B(C6F5)3 is substantially more reactive than our chiral borane with just one
C6F5 group. Reactions with B(C6F5)3 must be performed at low catalyst loading to allow for detection of these fundamental
intermediates by NMR spectroscopy.
silylation7a using C6F5-substituted chiral boranes as catalysts.
INTRODUCTION
■
Reasons for this gap are manifold but are, inter alia, likely due
to the synthetic challenges to access such chiral boranes, the
discouraging levels of enantioselectivity, and also the lack of
mechanistic understanding.
Piers and co-workers prepared an α-pinene-derived chiral
borane 2 (Figure 1, left)9 that was later used by Chen and
Klankermayer in one example of an enantioselective imine
Fostered by a continuous demand for homogeneous catalysts
that promote H−H and also Si−H bond activation followed by
a reduction step, the unique reactivity of electron-deficient
boranes decorated with at least one C6FnH5−n group (n = 1−5
but not 0) is currently garnering considerable attention.1 Three
scenarios emerge when combining one of these potent Lewis
acids with Lewis bases: irreversible and reversible Lewis pair
formation depending on size and basicity of the donor but also
no Lewis pairing due to steric hindrance, a concept that is now
referred to as frustrated Lewis pair (FLP) chemistry.2 Both
reversible and FLP systems were shown to catalyze metal-free
hydrosilylation and hydrogenation, respectively.3−6
Almost 20 years ago, Piers and co-workers disclosed that
tris(pentafluorophenyl)borane [B(C6F5)3, 1] is an effective
catalyst for the hydrosilylation of various carbonyl compounds.3
Later, these authors extended their system to imine functions.4
1 and related members of this extraordinary class of
electrophilic boranes feature the ability to split not only Si−
H bonds (90 kcal/mol) but also the stronger H−H bond (108
kcal/mol) heterolytically, thereby even allowing for metal-free
hydrogenation.5,6 Despite recent major efforts in these areas,
there are just a handful of enantioselective variants of imine
reductions7,8 and an isolated example of carbonyl hydro-
Figure 1. Terpene-derived chiral borane (pre)catalysts for the
reduction of imines by hydrogenation8a,b and/or hydrosilylation.7a
Received: September 10, 2013
Published: November 1, 2013
© 2013 American Chemical Society
17537
dx.doi.org/10.1021/ja409344w | J. Am. Chem. Soc. 2013, 135, 17537−17546

Journal of the American Chemical Society
Article
reduction involving the activation of dihydrogen.6c The
enantiomeric excess was low (13% ee) but served as proof of
concept. Substantially higher enantiocontrol in hydrogenation
(≤83% ee)8b and hydrosilylation (≤87% ee)7a was achieved by
employing the camphor-based chiral borate 3 (Figure 1, right).
However, 3 alone would not produce appreciable enantiomeric
excesses, and the presence of a bulky phosphane to form a
preactivated FLP salt was necessary. These results are clearly a
major step forward but the exact role of the phosphane remains
to be clarified.
carbonyl oxygen atom by using a silicon-stereogenic silane.14 As
to imines, it is assumed that these pass through the same
mechanism (Scheme 2). Borane 1 activates the silane I
Scheme 2. Currently Accepted Mechanism of B(C6F5)3-
Catalyzed Imine Hydrosilylation
Just recently, Liu and Du accomplished an improved
asymmetric imine reduction (≤89% ee) utilizing the novel
axially chiral borane (S)-4 (Figure 2, left).8e The idea of using
followed by nucleophilic attack of the imine nitrogen atom of
III at the silicon atom of transient II, thereby forming ion pair
V (I→II→IV⧧→V). Hydride transfer from the borohydride
then closes the catalytic cycle and releases amine VI (V→VI).
We had also applied the silicon-stereogenic silane as a
stereochemical probe to this mechanism yet without success.13
In turn, by using both enantiomers of an axially chiral silane
reagent in combination with (S)-5·THF, we were able to
demonstrate through matched/mismatched pairs that the
reduction pathway of imine reductions proceeds as proposed.7b
As mentioned above, this imine hydrosilylation catalyzed by
(S)-5·THF yielded inconsistent results, indicating to us that the
current mechanistic picture requires further refinement. We
report here an experimental analysis of this process catalyzed by
chiral (S)-5·THF but also achiral B(C6F5)3 (1). The fact that 1
behaves identically to (S)-5·THF generalizes our findings with
implications for the related reduction with dihydrogen. By
using 1D and 2D multinuclear NMR techniques and deuterium
labeling, we were able to detect and fully assign unexpected
intermediates that also contribute to the interpretation of our
previous experimental findings.
Figure 2. Axially chiral borane catalysts for the reduction of imines by
hydrogenation (left) and hydrosilylation (right).
the binaphthyl motif was not new since Piers had reported a
binaphthyl system before10 (not shown), and we had already
made the rigidified congener (S)-5·THF11 (Figure 2, right) of
(S)-4. However, the beauty of that catalyst (S)-4 lies in the ease
of its accessibility compared to (S)-5·THF. It is prepared in situ
the same way as the terpene-derived boranes 2 and 3 (Figure
1), by alkene hydroboration with bis(pentafluorophenyl)borane
[HB(C6F5)2], known as Piers’s borane.12 Also, (S)-4 is more
reactive as it contains B(C6F5)2 rather than less Lewis acidic
B(C6F5) groups.
Our chiral borane (S)-5·THF was nevertheless sufficiently
Lewis acidic to slowly promote Si−H bond activation, and
application to the hydrosilylation of acetophenone-derived
imines immediately afforded promising results: 33% ee for
phenyl-substituted (E)-6a and 62% ee for benzyl-substituted
(E)-6b (Scheme 1).7b These values, for benzyl protection in
Scheme 1. Enantioselective Imine Hydrosilylations
Catalyzed by (S)-5·THF
RESULTS AND DISCUSSION
■
Imine Hydrosilylation Catalyzed by an Axially Chiral
Borane. Our investigation commenced with an NMR
spectroscopic analysis of the reduction of benzyl-substituted
imine (E)-6b with Me2PhSiH (7) as hydride source catalyzed
by our chiral borane (S)-5·THF. As mentioned above, we had
observed that the enantiomeric excess of (S)-8b was dependent
on catalyst loading and conversion. Also, the amount of THF
introduced with the catalyst seemed to influence enantioin-
duction as well as the reaction rate.15 Intrigued by this
observation, we monitored the reaction of (E)-6b and 7 in the
presence of various amounts of added THF by NMR
spectroscopy (0.80 and 5.3 equiv, Figure 3). NMR spectra
were recorded immediately after sample preparation and at
selected time intervals. With 0.80 equiv of THF, a small amount
of N-silylated enamine 9b was detected as an intermediate in
the 1H and 13C NMR spectra followed by quantitative
particular, were inconsistent over several runs though and were
found to be dependent on both conversion and the amount of
THF present. It was this observation that prompted us to
explore the mechanism in even more detail.7b,13
The reaction pathway of the hydrosilylation of carbonyls is
largely understood. Piers and co-workers revealed a counter-
intuitive three-step mechanism where the Si−H bond is initially
activated by B(C6F5)3 (1).3b Our laboratory later established
the details of the actual transfer of the silicon moiety onto the
17538
dx.doi.org/10.1021/ja409344w | J. Am. Chem. Soc. 2013, 135, 17537−17546

Journal of the American Chemical Society
Article
Figure 3. Effect of added THF: time-resolved NMR studies of the hydrosilylation of N-benzylated imine (E)-6b catalyzed by (S)-5·THF (THF
indicated by asterisk).
formation of N-silylated amine 10b (Figure 3, top and bottom
left). Assignment of 9b is based on its diagnostic resonance
signal at 102 ppm in the 13C NMR spectrum, characteristic for
a C(sp2) enamine carbon atom. The resolution of the vinylic
another intermediate in the 13C NMR spectrum, the free amine
8b without (!) the silicon group at the nitrogen atom (Figure 3,
bottom right); 8b was also visible in the 1H NMR spectrum but
its resonance signals overlapped with those of THF (broad at
high concentration) and N-silylated amine 10b (see the
Supporting Information).19
We explain the appearance of the free amine 8b at high THF
concentration (5.3 equiv) by an equilibrium between
protonated THF and the unreacted imine base (E)-6b. At
high concentration of protonated THF, that protonation
equilibrium will be more shifted to protonated (E)-6b, an
iminium ion that is eventually reduced to 8b by the
borohydride. Conversely, at low THF concentration (0.80
equiv), protonation of (E)-6b by the minor amounts of
protonated THF is less pronounced.
Time-resolved NMR studies also showed that 8b and 9b are
both formed directly at the beginning of the reaction and before
appearance of product 10b. Formation of free amine 8b by
hydrolysis of N-silylated amine 10b was excluded since no
disiloxane was formed and the water-sensitive catalyst
maintained its activity. However, the excess of THF had a
dramatic effect on the reaction rate, and full conversion was not
1
enamine protons in the H NMR spectrum required higher
magnetic field strength (see the Supporting Information for the
2
measurement at 700 MHz), but the JH−H coupling could still
not be resolved. Moreover, these resonance signals overlapped
with the resonance signal of the benzylic protons of the imine
(E)-6b. 2D NMR spectra further confirmed the existence of 9b
(see the Supporting Information).16 The same observation was
made when conducting this experiment without additional
THF; 9b was, however, present in a much smaller quantity (not
shown).
We reasoned that the enamine intermediate 9b emerges from
deprotonation of the silyliminum ion (cf. V in Scheme 2).
Hence, either THF or unreacted imine (E)-6b could act as
Brønsted bases in the deprotonation of V. Consequently, we
decided to probe this reaction at higher concentration of THF
(5.3 equiv). As anticipated, the presence of more THF led to an
increase in the quantity of N-silylated enamine 9b (see the
Supporting Information). Remarkably, we now found yet
17539
dx.doi.org/10.1021/ja409344w | J. Am. Chem. Soc. 2013, 135, 17537−17546

Journal of the American Chemical Society
Article
Figure 4. Time-resolved NMR studies of the hydrosilylation of N-phenylimine (E)-6a catalyzed by (S)-5·THF.
reached after 5 days. We explain this effect with the equilibrium
between the adduct (S)-5·THF (catalytically inactive) and the
free borane (S)-5 (catalytically active),11 and with more THF
present, this equilibrium is shifted toward the former inactive
form.
To verify whether these findings are applicable to related
acetophenone-derived imines, we changed the substrate from
benzyl-substituted imine (E)-6b to phenyl-substituted imine
(E)-6a. Time-resolved NMR measurements under otherwise
identical reactions conditions confirmed our assumption
(Figure 4). Both the corresponding free amine 8a and the N-
silylated enamine 9a were formed in substantial quantities.
Again, these intermediates were generated prior to the
formation of product 10a. This time, characteristic baseline-
separated resonance signals for the vinylic enamine protons
1
were detected at 4.74 and 5.18 ppm in the H NMR spectrum
(Figure 4, left), and the C(sp2) enamine carbon atom appeared
at a typical chemical shift of 111 ppm in the 13C NMR
spectrum (Figure 4, right). 2D NMR measurements further
supported this assignment (see the Supporting Information).16
Although no THF was added, resonance signals of the free
amine 8a could be observed in a significant quantity in both 1H
and 13C NMR spectra.20 More precisely, we discovered that the
unexpected intermediates, the free amine 8a and the N-silylated
Figure 5. Segment of the 1H NMR spectrum manifesting the
equimolar ratio of free amine 8a and N-silylated enamine 9a.
enamine 9a, are present in equimolar ratio during the catalysis
(Figure 5).
This decisive finding supports our previous hypothesis that
the enamine is formed through deprotonation of the initially
generated silyliminum ion (cf. V in Scheme 2) by imine (E)-6a.
The resulting protonated imine, i.e., iminium ion, would
subsequently be reduced by the borohydride. These events
(proton abstraction and hydride transfer) would not only
account for the formation of free amine 8a but also explain its
1:1 ratio with enamine 9a. The disappearance of the starting
material (E)-6a is followed by the consumption of 8a and 9a. It
is worthy of note that these intermediates were still present
after 8 days, although GLC analysis indicated 84% conversion
after 66 h. Only after 11 days, the intermediates were almost
fully consumed.
B(C6F5)3 (1) as catalyst (Figures 6 and 7). For this, we
performed the B(C6F5)3-catalyzed hydrosilylation of benzyl-
substituted imine (E)-6b with a catalyst loading of 2.0 mol %.
The amount of catalyst had to be lower than that employed in
the original work of Piers and co-workers4 to make the reaction
slow enough for it being monitored by NMR spectroscopy. The
results were similar to those obtained with (S)-5·THF as
catalyst. Before the appearance of product 10b, small quantities
of the free amine 8b and the N-silylated enamine 9b were
observed, and these had completely disappeared upon
completion of the reaction (Figure 6; for further information
on the analyzed data, see the Supporting Information).21 Not
surprisingly, the corresponding free amine 8a and enamine 9a
intermediates were also seen in equimolar ratio in the
hydrosilylation of phenyl-substituted imine (E)-6a (Figure 7;
B(C6F5)3-Catalyzed Imine Hydrosilylation. To explore
whether the formation of the observed intermediates is a
general feature of imine hydrosilylation catalyzed by electron-
deficient boranes, we examined the same reactions with
1
for the relevant segment of the H NMR spectrum, see the
17540
dx.doi.org/10.1021/ja409344w | J. Am. Chem. Soc. 2013, 135, 17537−17546

Journal of the American Chemical Society
Article
Figure 6. Time-resolved NMR studies of the B(C6F5)3-catalyzed hydrosilylation of N-benzylimine (E)-6b.
Figure 7. Time-resolved NMR studies of the B(C6F5)3-catalyzed hydrosilylation of N-phenylimine (E)-6a.
Supporting Information). We had to use extremely low catalyst
loadings of B(C6F5)3, and ∼0.1 mol % is just an approximation
due to the small scale on which these reactions were run.
Reaction times are therefore likely to vary. Again, significant
amounts of 8a and 9a were observed before product 10a
formed. As opposed to the catalysis with (S)-5·THF, full
conversion was achieved with B(C6F5)3 (1). It seems likely that
the formation of the free amine 8 and enamine 9 intermediates
had been overlooked in the past due to the fast reaction rate.
Experimental Verification of the Intermediates’
Product-Forming Ability. The fact that the free amine is
formed as an intermediate poses the question of whether and
how it converts into the N-silylated amine. As the free amine
forms in a quantity equimolar to the N-silylated enamine, we
decided to subject separately synthesized 8a and 9a16−18 to the
typical catalytic setups in the presence of an equimolar amount
of silane 7 (Scheme 3). To our delight, quantitative formation
of the N-silylated amine 10a was seen in the expected reactions
times (see the Supporting Information for time-resolved NMR
Scheme 3. Conversion of Separately Synthesized Free Amine
8a and N-Silylated Enamine 9a into N-Silylated Amine 10a
under the Typical Catalytic Setups
measurements). We anticipated that the silylation of the amine
nitrogen atom occurs in a borane-catalyzed dehydrogenative
Si−N coupling similar to the known Si−O coupling catalyzed
by 1.22 Such Si−N couplings are, however, unprecedented in
the literature but do indeed work.23 B(C6F5)3 (1) catalyzes the
17541
dx.doi.org/10.1021/ja409344w | J. Am. Chem. Soc. 2013, 135, 17537−17546

Journal of the American Chemical Society
Article
coupling of amine 8a and silane 7 at ambient temperature as
evident from the formation of N-silylated amine 10a and
dihydrogen (Scheme 4). Conversion is rather slow though, not
formation as well as the presence of stoichiometric amounts of
protons relative to the enamine formed.
The equimolar ratio between the free amines 8 and the N-
silylated enamines 9 (e.g., Figure 5 for (S)-5·THF as catalyst) is
an indication of imine (E)-6 acting as a Brønsted base. To
further substantiate our proposal, we decided to monitor the
intermolecular proton transfer from the imine α carbon atom to
the imine nitrogen atom of another imine by deuterium
labeling. For this, we prepared phenyl-substituted imine (E)-
Scheme 4. B(C6F5)3-Catalyzed Dehydrogenative Si−N
Coupling
6aCD with deuteration of the methyl group in the α position. A
3
deuteration grade sufficiently high for our purposes was
achieved by condensation of acetophenone-d3 and aniline-d7
(see the Supporting Information for additional details including
the mass spectrometric determination of the deuteration
grade). According to our hypothesis, the nitrogen atom of
the free amine must be deuterated, corresponding to the
CD3
formation of 8aND,CD . Conducting the reaction with (E)-6a
3
and chiral borane (S)-5·THF (3.0 mol %) instantly confirmed
our assumption (Figure 8, left). Time-resolved 2H NMR
measurements clarified the abstraction of a deuteron by the
reaching completion after 24 h (see the Supporting Information
for time-resolved NMR measurements). We attribute this
observation to the high stability of the intermediate ion pair
11a composed of a silylammonium ion and a borohydride. 11a
might be regarded as the dihydrogen adduct of an amine/
borane FLP, and these are reluctant to release dihydrogen at
room temperature.24 Instead, the backward reaction might
occur. Hence, we conclude that, applied to the mechanism of
the imine hydrosilylation, the lifetime of that acidic ammonium
ion is sufficiently long to protonate the N-silylated enamine,
thus producing the N-silylated amine concomitant with re-
formation of the silyliminium ion (cf. refined mechanism
depicted below in Scheme 7).
Experiments with Deuterium-Labeled Substrates. A
series of deuterium-labeling experiments at various positions of
the reactants was performed to gain further insight into the
amine/enamine formation pathway of the catalysis. To confirm
that the hydride of 7 is exclusively incorporated at the methine
carbon atom of the product, we used deuterated 7-d1 in the
hydrosilylation of benzyl-substituted (E)-6b (Scheme 5; for
nitrogen lone pair of (E)-6aCD . The broad N−D resonance
3
signal of the free amine 8aND,CD was detected at approximately
3
3.5 ppm. Also, no deuteration was seen at the methine carbon
atom of the free amine 8aND,CD and the N-silylated amine
3
10aCD , and that is in agreement with the above labeling
3
experiment (cf. (E)-6b→8bCD with 7-d1, Scheme 5). Replacing
chiral borane (S)-5·THF by B(C6F5)3 (1) as catalyst gave
identical results (Figure 8, right). The use of B(C6F5)3 (1)
corroborates that THF is not needed in the proton shuffle.
After we had established the reactants in the enamine-
forming deprotonation, we designed a crossover experiment to
verify its reversibility, i.e., the enamine reprotonation to yield
the corresponding iminium ion. For this, we subjected labeled
(E)-6aCX ,C D and non-labeled (E)-6aCH ,C H in approximately
equimolar ratio to the standard protocol with (S)-5·THF and
B(C6F5)3 (1) as catalysts, respectively (Scheme 6). Deuterium
3
6
5
3
6
5
incorporation at the methyl group of amine 8aCX ,C H emerging
3
6
5
from non-deuterated (E)-6aCH ,C H would then be solid
evidence for the assumed deprotonation/protonation equili-
brium; deuteration of the phenyl group at the nitrogen atom
(C6D5 versus C6H5) was used as a mass label to distinguish
3
6
5
Scheme 5. Fate of the Hydride at the Silicon Atom
amines 8aCX by mass spectrometry. We were delighted to find
3
that both crossover experiments produced 8aCX ,C H with
3
6
5
substantial deuteration at the methyl group and, at the same
time, 8aCX ,C D with a diminished deuteration grade. The
crossover is more pronounced with catalyst 1 although reaction
times are significantly shorter than those with (S)-5·THF. We
rationalize this counterintuitive observation on the basis of the
different reactivities of the borohydride intermediates. With
three C6F5 groups at the boron atom, we expect the
borohydride derived from 1 to be a weaker hydride donor
than that of (S)-5·THF with one C6F5 group. B(C6F5)3 (1) is,
however, superior in the Si−H bond activation step.
Consequently, formation of the silyliminium ion is very fast
with 1 but its borohydride reduction is slow, thereby allocating
more time for the deprotonation/protonation equilibrium. The
situation is reverse with (S)-5·THF. A scrambling experiment
in the absence of silane 7 excluded the possibility of a Lewis
acid-catalyzed exchange of the deuterium label between imines
3
6
5
further information on the analyzed data, see the Supporting
Information). As expected, deuterium incorporation was only
seen at the tertiary carbon atom of the amine 8bCD (after
hydrolysis); the methyl group that is involved in the enamine
formation remained intact, i.e., non-deuterated.
Another idea was to intercept the catalysis at the stage of the
enamine intermediate by the addition of deuterated methanol.
Partial deuteration in the α position of the CN group would
then be sound evidence for the existence of the enamine. To
(E)-6aCX ,C D and (E)-6aCH ,C H ; less than 10% deuterium
incorporation was found for the latter.
1
3
6
5
3
6
5
achieve that we monitored amine/enamine formation by H
NMR spectroscopy and added deuterated methanol before
product formation was seen. These experiments were, however,
not totally conclusive due to the low degree of amine/enamine
Refined Catalytic Cycle. The mechanism initially
proposed by Piers3b and further validated by us7b,14 is, in
17542
dx.doi.org/10.1021/ja409344w | J. Am. Chem. Soc. 2013, 135, 17537−17546

Journal of the American Chemical Society
Article
Figure 8. Time-resolved 2H NMR studies of the hydrosilylation of deuterium-labeled N-phenylimine (E)-6aCD catalyzed by either (S)-5·THF (left)
3
or B(C6F5)3 (1) (right): evidence for proton abstraction by unreacted (E)-6aCD
.
3
silyliminium ions by either of the isomeric imines (E)-III and
(Z)-III.17 The resulting N-silylated enamine VII features free
rotation around the C−N bond and, as a consequence,
reprotonation produces either of the iminium ions (Z)-V and
(E)-V. These are, of course, again susceptible to reduction by
the borohydride ((Z)-V or (E)-V→VI). Proton abstraction
from silyliminium ions V by imine III results in the formation
of iminium ions VIII, again as ion pairs with the borohydride
(these iminium ions are also formed in both isomeric forms
(E)-VIII and (Z)-VIII but we show just one out of four
possible deprotonation scenarios for the sake of clarity, (Z)-
V→(E)-VIII). (E)-VIII is subsequently reduced to the free
amine IX. That step releases the boron Lewis acid that is, in
turn, available for another Si−H bond activation with the amine
nitrogen atom of IX as a Lewis base (IX→X).23 The
ammonium ion X then serves as a Brønsted acid, protonating
enamine VII to reform the silyliminium ion V and to afford the
final product VI (cf. Scheme 4 and the associated discussion).
The refined mechanistic picture not only rationalizes the
unexpected formation of intermediates VII (the enamine) and
IX (the free amine) in equimolar ratio but also explains how IX
is subsequently transformed into VI (the N-silylated amine).
Effect on Asymmetric Transformations. The refined
mechanistic understanding now helps to define the challenge
associated with the enantioselective variant of the borane-
catalyzed hydrosilylation of imines. The enantioselectivity is
determined in the hydride transfer from the borohydride onto
the electrophilic carbon atom of an iminium ion intermediate.
It was believed that this would only involve the isomeric
silyliminium ions (Z)-V and (E)-V (Scheme 8, left), and their
ratio would largely depend on the isomeric purity of the
starting imine III.25 However, there is another enantioselectiv-
ity-determining hydride transfer in the competing deprotona-
Scheme 6. Crossover Experiments To Probe the
Silyliminium Ion Deprotonation/N-Silylated Enamine
Protonation Equilibrium
principle, still true but the new insight calls for an extension in
the case of imines with α-hydrogen atoms (Scheme 7). The
refined catalytic cycle also begins with the activation of silane I
by the boron Lewis acid with formation of the weak adduct II.
Nucleophilic attack of either imine (E)-III or imine (Z)-III at
the electrophilic silicon atom then yields the isomeric iminium
ions (Z)-V and (E)-V, respectively,25 as ion pairs with the
borohydride. The expected next step is the reduction of V by
the borohydride to afford product VI. The unexpected
alternative reaction pathway is the deprotonation of those
17543
dx.doi.org/10.1021/ja409344w | J. Am. Chem. Soc. 2013, 135, 17537−17546

Journal of the American Chemical Society
Article
Scheme 7. Refined Mechanism of the Hydrosilylation of Imines Catalyzed by Electron-Deficient Boron Lewis Acids (One out of
Four Possible Deprotonation/Protonation Scenarios Shown for the Sake of Clarity)
upon hydrolysis or alcoholysis, and these might be in part
reduced to the amine with unhydrolyzed borohydride.
Scheme 8. Enantioselectivity-Determining Hydride
Reduction of Two Pairs of Isomeric Iminium Ions
Contributing to the Global Enantiomeric Excess
CONCLUSION
■
We disclosed here a fundamental refinement of the catalytic
cycle of the borane-catalyzed imine hydrosilylation (Scheme 2
versus Scheme 7). The unexpected discovery of free amines and
N-silylated enamines as intermediates by room-temperature 1H
and 13C NMR spectroscopy paved the way for the elucidation
of an alternative reduction pathway. The catalysis was
previously believed to proceed exclusively through an ion pair
composed of a silyliminium ion and a borohydride that further
converts into an N-silylated amine and the borane catalyst
(Scheme 2). However, deprotonation of the silyliminium ion
by unreacted imine starting material was shown to be a
competing process. That proton abstraction produces an N-
silylated enamine (detected by NMR spectroscopy) and an
iminium ion (= protonated imine) with the borohydride as
counteranion. The borohydride reduction of that iminium ion
releases the free amine (detected by NMR spectroscopy) that is
transformed into the N-silylated amine in a few more steps that
also involve reprotonation of the N-silylated enamine (cf.
Scheme 7). The detection of these elusive but prominent
intermediates was previously thwarted by the high overall
catalytic activity of B(C6F5)3 (1), the conventional electron-
deficient borane utilized in these hydrosilylations. Our axially
chiral borane (S)-5·THF with just one C6F5 group is by far less
reactive than 1 in the Si−H bond activation. In turn, the
hydride donor strengths of the resulting borohydrides are likely
to be reverse. The reaction times are still significantly longer
tion/protonation pathway of the catalysis. The reduction of
isomeric iminium ions (E)-VIII and (Z)-VIII (Scheme 8, right)
also contributes to product formation, and it is more than likely
that VIII→IX (right) proceeds with different levels of
enantioinduction (even with opposite absolute configuration)
than V→VI (left). A total of two diastereomeric pairs of
iminium ions and eight stereoisomeric combinations of these
iminium ions with the chiral borohydride will produce a global
enantiomeric excess (after hydrolysis) that will be dependent
on the amount of enamine VII formed and, related to this, on
conversion. At incomplete conversion, remaining VII is an
additional issue. It transforms back into an iminium ion/imine
17544
dx.doi.org/10.1021/ja409344w | J. Am. Chem. Soc. 2013, 135, 17537−17546

Journal of the American Chemical Society
Article
(3) For B(C6F5)3-catalyzed CO hydrosilylation, see: (a) Parks, D.
J.; Piers, W. E. J. Am. Chem. Soc. 1996, 118, 9440−9441. (b) Parks, D.
J.; Blackwell, J. M.; Piers, W. E. J. Org. Chem. 2000, 65, 3090−3098.
(4) For B(C6F5)3-catalyzed CN hydrosilylation, see: Blackwell, J.
M.; Sonmor, E. R.; Scoccitti, T.; Piers, W. E. Org. Lett. 2000, 2, 3921−
3923.
with (S)-5·THF than with 1 (days versus hours), and that
initially helped us to observe those unexpected intermediates.
We were nevertheless able to demonstrate at low catalyst
loading that the B(C6F5)3 catalysis follows the same reaction
pattern.
The new mechanistic insight profoundly influences our
thinking about enantioselective variants using chiral boranes as
the source of asymmetric induction. There is an additional
enantioselectivity-determining hydride transfer step. The
absolute configuration and the level of enantioinduction will
be set in the borohydride reduction of the silyliminium ion
(expected path) and the iminium ion (unexpected path). With
this knowledge, we do now understand the dependence of the
enantiomeric excess on conversion (remaining amine/enamine
prior to hydrolysis) and on the presence of THF as a facilitator.
And, the chiral borohydride must differentiate the enantiotopic
faces of the diastereomeric pairs (E/Z) of (silyl)iminium ions
with the same sense of asymmetric induction but the efficieny
will clearly be different for each individual (silyl)iminium ion.
We would like to close with a remark on potential
implications of the present findings on asymmetric FLP
chemistry. H−H activation by FLPs or boranes alone is less
complicated because the competing pathways are degenerate in
the sense that both proceed through the same iminium ion
intermediate as an ion pair with the borohydride.
(5) We are not aware of any FLP- or borane-catalyzed CO
hydrogenation.
(6) For FLP- and borane-catalyzed CN hydrogenation, see:
(a) Chase, P. A.; Welch, G. C.; Jurca, T.; Stephan, D. W. Angew. Chem.,
Int. Ed. 2007, 46, 8050−8053. (b) Chase, P. A.; Jurca, T.; Stephan, D.
W. Chem. Commun. 2008, 1701−1703. (c) Chen, D.; Klankermayer, J.
Chem. Commun. 2008, 2130−2131. (d) Spies, P.; Schwendemann, S.;
Lange, S.; Kehr, G.; Frohlich, R.; Erker, G. Angew. Chem., Int. Ed. 2008,
̈
47, 7543−7546. (e) Sumerin, V.; Schulz, F.; Atsumi, M.; Wang, C.;
Nieger, M.; Leskela, M.; Repo, T.; Pyykko, P.; Rieger, B. J. Am. Chem.
̈
̈
Soc. 2008, 130, 14117−14119. (f) Rokob, T. A.; Hamza, A.; Stirling,
A.; Papai, I. J. Am. Chem. Soc. 2009, 131, 2029−2036. (g) Jiang, C.;
Blacque, O.; Berke, H. Chem. Commun. 2009, 5518−5520. (h) Eros,
G.; Mehdi, H.; Papai, I.; Rokob, T. A.; Kiraly, P.; Tarkanyi, G.; Soos, T.
́
̋
́
́
́
́
́
Angew. Chem., Int. Ed. 2010, 49, 6559−6563. (i) Zhao, L.; Li, H.; Lu,
G.; Huang, F.; Zhang, C.; Wang, Z.-X. Dalton Trans. 2011, 40, 1929−
1937. (j) Heiden, Z. M.; Stephan, D. W. Chem. Commun. 2011, 47,
5729−5731. For a brief review, see: (k) Paradies, J. Synlett 2013,
777−780.
(7) For FLP- or borane-catalyzed enantioselective CN hydro-
silylation, see: (a) Chen, D.; Leich, V.; Pan, F.; Klankermayer, J.
Chem.−Eur. J. 2012, 18, 5184−5187. (b) Mewald, M.; Oestreich, M.
Chem.−Eur. J. 2012, 18, 14079−14084.
ASSOCIATED CONTENT
■
(8) For FLP- or borane-catalyzed enantioselective CN hydro-
genation, see: (a) Reference 6c. (b) Chen, D.; Wang, Y.;
Klankermayer, J. Angew. Chem., Int. Ed. 2010, 49, 9475−9478.
S
* Supporting Information
Experimental details, characterization data, and 1H, 1H/1H
COSY, 2H, 13C, 1H/13C HMQC, 1H/13C HMBC, 1H/29Si
HMQC, and 29Si DEPT NMR spectra. This material is
available free of charge via the Internet at http://pubs.acs.org.
(c) Sumerin, V.; Chernichenko, K.; Nieger, M.; Leskela, M.; Rieger,
̈
B.; Repo, T. Adv. Synth. Catal. 2011, 353, 2093−2110. (d) Ghattas, G.;
Chen, D.; Pan, F.; Klankermayer, J. Dalton Trans. 2012, 9026−9028.
(e) Liu, Y.; Du, H. J. Am. Chem. Soc. 2013, 135, 6810−6813. For a
review, see: (f) Chen, D.; Klankermayer, J. In Frustrated Lewis Pairs II,
Expanding the Scope; Erker, G., Stephan, D. W., Eds.; Topics in
Current Chemistry 334; Springer: Heidelberg, Germany, 2013; pp 1−
26.
AUTHOR INFORMATION
Corresponding Author
martin.oestreich@tu-berlin.de
■
(9) Parks, D. J.; Piers, W. E.; Yap, G. P. A. Organometallics 1998, 17,
Notes
5492−5503.
The authors declare no competing financial interest.
(10) Morrison, D. J.; Piers, W. E.; Parvez, M. Synlett 2004, 2429−
2433.
ACKNOWLEDGMENTS
■
(11) Mewald, M.; Frohlich, R.; Oestreich, M. Chem.−Eur. J. 2011, 17,
̈
This research was supported by the Cluster of Excellence
Unifying Concepts in Catalysis of the Deutsche Forschungs-
gemeinschaft (EXC 314/2). M.O. is indebted to the Einstein
Foundation (Berlin) for an endowed professorship. We thank
Dr. Sebastian Kemper and Dr. Maria Schlangen (both TU
Berlin) for expert advice with the NMR and MS measurements,
respectively. We are grateful to Dr. Hendrik Klare (TU Berlin)
for valuable discussions.
9406−9414.
(12) (a) Parks, D. J.; Spence, R. E. v H.; Piers, W. E. Angew. Chem.,
Int. Ed. Engl. 1995, 34, 809−811. (b) Reference 9.
(13) Hog, D. T.; Oestreich, M. Eur. J. Org. Chem. 2009, 5047−5056.
(14) Rendler, S.; Oestreich, M. Angew. Chem., Int. Ed. 2008, 47,
5997−6000.
(15) The purification and isolation of (S)-5 required the formation of
the weak Lewis acid/base adduct (S)-5·THF.11
(16) N-Silylated enamines are a rare class of compounds, and we are
aware of just one report where the selective transformation of imines
into N-trimethylsilylated enamines were accomplished by mild
deprotonation of trimethylsilyliminium triflates with stoichiometrically
added amine base.17 At a late stage of this project, we were able to
apply methodology that was recently developed in our laboratory18 to
the preparation of the requisite enamines 9a and 9b. Our base-free
procedure originally transforms enolizable carbonyl compounds into
silyl enol ethers along with dihydrogen gas by dehydrogenative
coupling with silanes. Its application with a minor modification to
enolizable imines (E)-6a and (E)-6b with Me2PhSiH (1.0 equiv) yields
the otherwise difficult-to-make enamines 9a and 9b in decent purity.
The NMR spectroscopic characterization further confirms our earlier
assignments (see the Supporting Information for details): Hermeke, J.
REFERENCES
■
(1) For a critical review, see: (a) Piers, W. E.; Marwitz, A. J. V.;
Mercier, L. G. Inorg. Chem. 2011, 50, 12252−12262. For a recent
article providing a comprehensive list of references of synthetic
applications of B(C6F5)3-catalyzed Si−H bond activation, see:
(b) Robert, T.; Oestreich, M. Angew. Chem., Int. Ed. 2013, 52,
5216−5218.
(2) For state-of-the-art summaries, see: (a) Frustrated Lewis Pairs I,
Uncovering and Understanding; Erker, G., Stephan, D. W., Eds.; Topics
in Current Chemistry 332; Springer: Heidelberg, Germany, 2013. (b)
Frustrated Lewis Pairs II, Expanding the Scope; Erker, G., Stephan, D.
W., Eds.; Topics in Current Chemistry 334; Springer: Heidelberg,
Germany, 2013. For an authoritative review, see: (c) Stephan, D. W.;
Erker, G. Angew. Chem., Int. Ed. 2010, 49, 46−76.
Ph.D. Thesis, Technische Universitat Berlin, Germany, Projected
̈
2015.
17545
dx.doi.org/10.1021/ja409344w | J. Am. Chem. Soc. 2013, 135, 17537−17546

Journal of the American Chemical Society
Article
(17) Ahlbrecht, H.; Duber, E.-O. Synthesis 1980, 630−631.
̈
(18) Konigs, C. D. F.; Klare, H. F. T.; Ohki, Y.; Tatsumi, K.;
̈
Oestreich, M. Org. Lett. 2012, 14, 2842−2845.
(19) The analytic data of free amine 8b are in agreement with those
obtained for an independently prepared sample.
(20) The analytical data of free amine 8a are in agreement with those
obtained for an independently prepared sample.
(21) Experiments with added THF were not performed as it is ring-
opened and eventually deoxygenated by B(C6F5)3/silane combina-
tions: Gevorgyan, V.; Rubin, M.; Benson, S.; Liu, J.-X.; Yamamoto, Y.
J. Org. Chem. 2000, 65, 6179−6186. We detected silylated butan-1-ol:
Mewald, M. Ph.D. Thesis, Westfalische Wilhelms-Universitat Munster,
̈
̈
̈
Germany, 2012.
(22) Blackwell, J. M.; Foster, K. L.; Beck, V. H.; Piers, W. E. J. Org.
Chem. 1999, 64, 4887−4892.
(23) These preliminary results will be part of a forthcoming full
report on the B(C6F5)3-catalyzed dehydrogenative Si−N coupling.
(24) Sumerin, V.; Schulz, F.; Nieger, M.; Leskela, M.; Repo, T.;
̈
Rieger, B. Angew. Chem., Int. Ed. 2008, 47, 6001−6003.
(25) For an analysis of the E/Z isomerization of imines in the
presence of B(C6F5)3 (1), see: Blackwell, J. M.; Piers, W. E.; Parvez,
M.; McDonald, R. Organometallics 2002, 21, 1400−1407.
17546
dx.doi.org/10.1021/ja409344w | J. Am. Chem. Soc. 2013, 135, 17537−17546