A detailed NMR- and DFT-based study of the Sakurai-Hosomi-Yamamoto asymmetric allylation reaction

Article abstract of DOI:10.1021/jo400451u

A Lewis acid complex between benzaldehyde and the silver catalyst was detected by ³¹P NMR and shown to be the direct precursor to allylation within the Sakurai-Hosomi-Yamamoto reaction. Structural and thermochemical hybrid-DFT calculations indicated that benzaldehyde predominantly formed an η¹-σ-complex with the catalyst; however, two other competing conformers involving different coordination modes were found, including an activated μ²-bound complex. The differences in ³¹P NMR shifts upon complexation were calculated by the gauge-independent atomic orbital (GIAO-DFT) method for each conformer. The minimum energy conformer was found to correlate well with chemical shift trends observed experimentally, and an analysis of Mullikan charge populations revealed that the carbonyl carbon of the highest-energy conformer was the most electron-deficient. Furthermore, one minor and three major silicon intermediates were detected by ²⁹Si NMR and, with the aid of ¹H- ²⁹Si HSQC, were assigned by comparison with parent compounds and GIAO-DFT calculations. Finally, a tentative mechanism was proposed based on these findings.

Full text of DOI:10.1021/jo400451u

Article
pubs.acs.org/joc
A Detailed NMR- and DFT-Based Study of the Sakurai−Hosomi−
Yamamoto Asymmetric Allylation Reaction
David Hurem, Andrey G. Moiseev, Razvan Simionescu, and Travis Dudding*
Department of Chemistry, Brock University, 500 Glenridge Avenue, St. Catharines L2S 3A1, ON, Canada
S
* Supporting Information
ABSTRACT: A Lewis acid complex between benzaldehyde
and the silver catalyst was detected by 31P NMR and shown to
be the direct precursor to allylation within the Sakurai−
Hosomi−Yamamoto reaction. Structural and thermochemical
hybrid-DFT calculations indicated that benzaldehyde predom-
inantly formed an η1-σ-complex with the catalyst; however,
two other competing conformers involving different coordi-
nation modes were found, including an activated μ2-bound
complex. The differences in 31P NMR shifts upon complex-
ation were calculated by the gauge-independent atomic orbital
(GIAO−DFT) method for each conformer. The minimum
energy conformer was found to correlate well with chemical
shift trends observed experimentally, and an analysis of Mullikan charge populations revealed that the carbonyl carbon of the
highest-energy conformer was the most electron-deficient. Furthermore, one minor and three major silicon intermediates were
1
detected by 29Si NMR and, with the aid of H−29Si HSQC, were assigned by comparison with parent compounds and GIAO−
DFT calculations. Finally, a tentative mechanism was proposed based on these findings.
tion, reported by Yamamoto et al. in 1999, which employs (R)-
(1,1′-binaphthalene-2,2′-diyl)bis(diphenylphosphine) [(R)-
BINAP] as a chiral ligand, has emerged as a particularly
important asymmetric transformation as it affords high levels of
diastereo- and enantioselectivity and relies on the use of
commercially available chiral ligands, aldehydes, and less toxic
allylsilanes.3 As for the mechanistic features of this reaction, it is
noteworthy that Yamamoto and co-workers have provided the
field with considerable mechanistic insight into and accom-
panying rationale for the Sakurai−Hosomi−Yamamoto allyla-
tion reaction and a number of other related Ag(I)-catalyzed
processes.3b,4 For instance, this group has reported a
comprehensive 31P NMR-, X-ray crystallographic data-, and
UV−vis-supported mechanistic investigation of a (R)-
BINAP2·AgPF6- and (R)-BINAP2·AgOAc-catalyzed Mukaiyama
aldol reaction that intriguingly afforded products of opposite
absolute stereochemical configuration depending on which
Ag(I) complex was used.4 Certainly of greater relevance to the
underlying mechanistic facets of the (R)-BINAP·AgF-catalyzed
Sakurai−Hosomi−Yamamoto reaction have been 31P NMR
spectroscopic-based findings that revealed that the resting state
of the catalyst in this reaction existed as a rapidly equilibrating
temperature- and solvent-dependent mixture of Ag(I) species 2
and 3, whereas the dominant species 1 formed irreversibly.5
Although the prior mechanistic studies addressing the (R)-
BINAP·AgF-catalyzed Sakurai−Hosomi−Yamamoto reaction
INTRODUCTION
■
Since the pioneering disclosure of the Lewis acid catalyzed C−
C bond-forming addition of allylsilanes to carbonyls in 1976,
the Sakurai reaction has become a mainstay in the field of
synthesis for the construction of secondary and tertiary
homoallylic alcohols [Scheme 1(a)].1
Scheme 1. Sakurai−Hosomi−Yamamoto Allylation
In line with these ideas, ongoing efforts toward the
development of synthetic approaches to homoallylic alcohols
continue to be driven by their recurrence in medicinal
compounds, natural products, agrochemicals, and numerous
other chemicals.2 Consequently, as a part of these develop-
ments, the AgF-catalyzed Sakurai−Hosomi−Yamamoto allyla-
Received: March 1, 2013
Published: April 9, 2013
© 2013 American Chemical Society
4440
dx.doi.org/10.1021/jo400451u | J. Org. Chem. 2013, 78, 4440−4445

The Journal of Organic Chemistry
Article
Table 1. 31P NMR Data for the Catalytic Mixture with and without Benzaldehyde
a
b
c
AgF/(R)-BINAP, 250 K
AgF/(R)-BINAP benzaldehyde, 250 K
AgF/(R)-BINAP benzaldehyde, 193 K
d
JAg−P (Hz)
integral (%)
δP (ppm)
JAg−P (Hz)
integral (%)
δP (ppm)
JAg−P (Hz)
integral (%)
δP (ppm)
1
242
405
−
−
760
6.6
18.9
−
−
74.6
16.3
13.9
−
−
1.1
242
389
682
−
7.3
20.3
72.4
−
16.3
14.0
2.1
−
242
402
−
6.4
19.1
−
17.1
14.7
−
2
3′ + 3
3′
3
703
27.1
10.6
−
−
−
754
47.4
−1.6
a
b
Spectra were obtained from (R)-BINAP (0.1 equiv) and AgF (0.06 equiv) in methanol-d4 at 248.9 K. Spectra were obtained in the presence of
c
d
benzaldehyde (1.0 equiv) at 253.4 K. Spectra were obtained in the presence of benzaldehyde (1.0 equiv) at 192.9 K. Taken as an average between
107Ag−P and 109Ag−P coupling constants.
have provided insight into the resting state of the catalytic
Ag(I) complexes involved in this transformation, still lacking at
this time is a lucid mechanistic understanding of the Ag(I)
species generated from the trimethoxysilane and aldehyde
reagents used in this reaction. Importantly, this information
would provide valuable insight into the transient catalytic
species that serve as the direct progenitors to the high
enantioselection of this process. The conventional wisdom
regarding the mechanism of the Sakurai−Hosomi−Yamamoto
allylation holds that three major Ag(I) complexes are formed
under the reaction conditions that conform to (1) a dimeric
(R)-BINAP2·AgF complex that catalyzes a nonstereoselective
background allylation; (2) a (R)-BINAP·AgF complex that has
been thought to mediate an asymmetric (Type II) allylation via
the transient interplay of a metal-based η1-allyl intermediate,
such as (R)-BINAP2·Ag(CH2CHCH2); and (3) a (R)-
BINAP·(AgF)2 complex that induces enantioselection by way
of a Lewis acid promoted (Type I) allylation process [Scheme
1(b)].6 As such, working within the framework of 1H, 13C, and
31P NMR spectroscopic studies and high-level DFT calcu-
lations, we report herein a mechanistic study of the Sakurai−
Hosomi−Yamamoto reaction.
Figure 1. 31P NMR spectra of (R)-BINAP (0.06 equiv) and AgF (0.1
equiv) (a) in methanol-d4 at 248.9 K, (b) in the presence of
benzaldehyde (1.0 equiv) at 253.4 K, (c) in the presence of
benzaldehyde (1.0 equiv) at 192.9 K, and (d) in the presence of
benzaldehyde (1.0 equiv) and allyltrimethoxysilane (1.5 equiv) at
233.8 K.
RESULTS AND DISCUSSION
■
At the outset of this study, we investigated the specific catalytic
roles and structural nature of these entities through an analysis
of the 31P NMR spectra and JAg−P (that is, 31P, 109Ag, and
107Ag) coupling constants of a mixture of AgF (10 equiv) and
(R)-BINAP (6 equiv) cooled to the routinely used reaction
temperature of 248.9 K (−24.1 °C). More specifically, it was
apparent from this analysis that three distinct coordination
complexes were present in the catalytic mixture in a ratio of
6:19:75, which were assigned as 1 (d, 16 ppm), 2 (d, 14 ppm),
and 3 (d, 1 ppm), based upon the observed chemical shifts,
average Ag−P spin−spin coupling constants (J value),7 and the
previous findings of Yamamoto et al. [Table 1 and Figure
1(a)].5
0 to 8 ppm sharpened into a well-resolved pair of doublets
[Figure 1(c)]. The upfield doublet of this set at −1.56 ppm
(JAg−P = 756 Hz) was assigned as the unbound complex 3,
whereas the lower field doublet (3′) at 10.6 ppm (d, JAg−P
=
705 Hz) was thought to result from formation of a weighted
distribution of aldehyde-bound complexes [(R)-
BINAP·(AgF)2*(OC(H)Ph)n (n = 1 or 2)]. Consistent
with this last assignment, the integrated area of the doublet at
10.6 ppm was equivalent in magnitude to that of the (R)-
BINAP·(AgF)2 complex present before the addition of
benzaldehyde depicted in Figure 1(a) and Table 1.
Having found a temperature at which the signals assigned to
3 and 3′ were clearly resolved and to better understand the
kinetic profile and respective off-and-on exchange rate of
aldehyde and catalyst association, we progressively heated the
sample of interest from 193 to ∼250 K (−20 °C). It was
apparent from this analysis that the 31P NMR signals assigned
to 3 and 3′ coalesced at a temperature of ∼220 K (Tc) with a
respective exchange frequency (kr) of 3.2 × 103 s−1, where Δν
is the full width of the peak at half of its maximum height at Tc
(eq 1). From there, application of the Eyring equation results in
a computed activation energy (ΔG⧧) of ∼7.2 kcal/mol for the
formation of 3′ (eq 2).8
An excess of benzaldehyde (10 equiv) was then added to the
sample under study at 253.4 K to gauge the extent to which the
aldehyde component of this reaction interacted with the resting
state(s) of the catalyst(s). The result was the noticeable
broadening of the 31P signal of the (R)-BINAP·(AgF)2 complex
(3) at 1 ppm into a poorly resolved doublet spanning the
spectral region from 0 to 8 ppm (3 + 3′). In contrast, the
signals assigned to 1 and 2 effectively remained unchanged
[Figure 1(b)]. Aware of the dynamic nature of this catalytic
system, we then probed the temperature dependence of this
mixture by cooling the sample to 192.9 K (−80.1 °C),
whereupon the broad signal spanning the spectral region from
kr = 2.2Δν
(1)
4441
dx.doi.org/10.1021/jo400451u | J. Org. Chem. 2013, 78, 4440−4445

The Journal of Organic Chemistry
Article
⎛
⎜
⎝
⎞
⎟
⎠
krh
ΔG⧧ = RT ln
c
kT
(2)
c
Consistent with a host−guest-type interaction between
benzaldehyde and the catalytic complex was the observation
1
of an upfield singlet in both the acquired H and 13C spectra
(5.57 and 97.2 ppm, respectively) at 193 K upon the addition
of benzaldehyde. On the basis of 13C−1H HSQC data, these
signals were found to be directly correlated and so were
thought to correspond to the aldehyde hydrogen and carbon of
benzaldehyde in the silver complex (see Supporting Informa-
tion).
Figure 2. Global and competing minimal geometries, free energies of
complexation (ΔG), condensed to carbonyl carbon Mulliken charge
analysis for the catalytic complex calculated at the DFT B3LYP/
DGDZVP level of theory.
Despite strong support for benzaldehyde coordination, the
type of complexation between benzaldehyde and the catalyst
(3) was still not fully understood. It is well-known that the
majority of main group, early transition metal, and lanthanide
Lewis acids form σ-type complexes with carbonyl-containing
compounds,9a−c while electron-rich transition metals form π-
type complexes.9a,d,e Furthermore, σ-type complexation of
aldehydes with Lewis acids has been shown to produce a
through a less stable, albeit kinetically more reactive,
intermediate such as 6 instead of the thermodynamically
more stable minimum 5.12 Along these same lines, consistent
with 6 being the major catalytic complex governing stereo-
induction in this reaction, a visual inspection of these three
species revealed that the si-face of the bound aldehyde in
structures 4 and 5 was sterically unhindered and open to attack.
Alternatively, if allylation ensued from 6, the sense of
stereoinduction would be consistent with the experimentally
observed re-facial stereoselectivity because the si-face of the
aldehyde is shielded by the catalyst.
9c
1
downfield shift of H and 13C signals of the carbonyl group,
and π-type complexes resonated upfield in comparison with the
noncoordinated aldehyde.9d,e In this case, however, the
presence of the sterically encumbering (R)-BINAP ligand
makes it difficult to envision this reaction proceeding through a
π-complex. On the other hand, it could be argued that a strong
shielding effect induced by the aromatic system of the ligand
could very well account for the observed upfield shift of 1H and
13C signals of benzaldehyde in the 3′ complex due to
anisotropic effects.10 Thus, it would appear that the upfield-
shifted signals observed in this instance do not provide
definitive evidence for π-type complexation.
Having at that point been unable to discern by NMR
experiments the precise mode by which benzaldehyde was
bound in complex 3′, but with strong evidence supporting the
formation of a (R)-BINAP·(AgF)2*(OC(H)Ph)n complex,
we turned to DFT calculations to develop a better under-
standing of the structural features of 3′. To this end, σ- and π-
bound complexes of (R)-BINAP·(AgF)2*(OC(H)Ph)n
having one or two benzaldehyde ligands were optimized in
the gas phase at room temperature at the B3LYP/DGDZVP
(density Gauss double-ζ with polarization functions basis set)
level.11 Interestingly, upon optimization of those geometries
initiated from π-type complexes, it was immediately brought to
our attention that they inevitably converged to σ-type
complexes. Nevertheless, out of those conformers considered,
minima 4−6 were found to be the lowest in energy (Figure 2).
With respect to these conformers, present in 6 was a μ2-σ-
coordinated benzaldehyde, whereas the benzaldehyde in
structures 4 and 5 was η1-σ-coordinated.
The influence of temperature and solvent effects upon these
complexes was then considered by reoptimizing these
structures at 193, 212, and 249 K and room temperature
with the IEFPCM (integral equation formalism polarizable
continuum model) method to account for the solvent methanol
generally employed in practice. These calculations indicated
that aldehyde-bound complex 5 was the lowest energy structure
among 4−6 (see Figure 2 and Supporting Information).
However, we discerned from an analysis of the computed
Mulliken charges of these three complexes that the carbonyl
carbon of the bridged structure 6 was more electron-deficient
then those of 4 and 5, suggesting that allylation likely proceeds
However, at odds with the computed nonsymmetrical nature
of 4−6 was our experimental observation of a single 31P NMR
shift at 10.6 ppm that implied the two phosphorus atoms of the
catalyst-bound aldehyde adduct formed in this reaction were
chemically shift equivalent. In view of this discrepancy and to
clarify this point further, we once again turned to DFT
calculations carried out within the framework of GIAO−DFT
(B3LYP/DGDZVP) NMR theory to help elucidate the
structure of (R)-BINAP·(AgF)2*(OC(H)Ph) (4−6). From
these calculations we concluded that the 31P NMR resonances
of structure 5 were downfield from those of the parent complex
(R)-BINAP·(AgF)2 (that is, catalyst 3), whereas those of 4 and
6 were shifted upfield with respect to 3 (Table 2). Surprisingly,
the difference in chemical shift between the two phosphorus
atoms in 5 of 0.3 ppm was smaller than that of 4 and 6.
However, it should be noted that this minute difference
between the computed 31P chemical shifts would not be
apparent from our experimental NMR results, even if it were
the two phosphorus atoms would appear to be equivalent on
the NMR time scale if there was rapid ligand exchange between
the two silver atoms of 5. Interestingly, the shift difference
(Δδ) in 31P NMR between 3 and 5 was found to decrease with
temperature, whereas the inverse trend was observed for 4 and
6 in relation to 3. A small, upfield shift was observed in the
1
calculated H signal corresponding to the aldehydic hydrogen
of 5, but an inverse trend was observed for 4 and 6. All of the
carbonyl carbons of 4−6 were shifted downfield. The
differences between the calculated and experimental magni-
tudes of Δδ 31P NMR shifts are thought to arise from two
sources: the intrinsic conformational dynamics of the (R)-
BINAP·(AgF)2*(OC(H)Ph) complex that, because of the
demanding nature of these calculations, could not be accounted
for in silico and the inadequate treatment of the long-range
effects at the level of theory used.13 The latter could also
contribute to the inconsistencies between experimental and
computed 1H proton and 13C carbon shifts because the
inductive effect of the large aromatic system of the catalyst
could have a significant impact on the magnetic susceptibilities
4442
dx.doi.org/10.1021/jo400451u | J. Org. Chem. 2013, 78, 4440−4445

The Journal of Organic Chemistry
Article
a
Table 2. Computed Relative Chemical Shifts from GIAO−DFT Calculations
249 K Δδ
212 K Δδ
193 K Δδ
b
31P
c
c
1H
b
c
c
1H
b
31P
c
c
1H
13C
31P
13C
13C
4
5
6
−1.5
1.5
6.3
1.0
0.9
0.32
−0.25
0.06
−1.5
1.5
6.3
1.0
0.9
0.32
−0.25
0.06
−1.0
2.0
6.3
1.0
0.9
0.32
−0.25
0.06
−4.2
−4.2
−3.8
a
Chemical shifts were computed for modeled structures 4−6 by GIAO−DFT at the B3LYP/DGDZVP level from the difference of isotropic
b
shielding tensors from the reference structure by the following equation: Δδ = σ − σref. The differences of 31P shifts were found using structure 3 as
c
a reference. The differences of 13C and H shifts were found using the structure of benzaldehyde calculated at the same level of theory.
1
of neighboring nuclei. Moreover, the binding of a second
benzaldehyde to either 4 or 5, which would more readily allow
for the generation of a symmetric complex, was associated with
an additional cost of 7.1 or 5.0 kcal/mol at 193 K, respectively,
and is therefore unlikely.
ysilane derivative that is strongly deshielded in this reaction
scenario.
From there we turned to GIAO−NMR DFT calculations to
aid in the assignment of the −83.5 ppm signal. Fluorinated
silanes were considered a likely candidate for the −84.5 ppm
signal, and as such fluorotrimethoxysilane (Si(OMe)3F) was
thought to be a reasonable starting point.16 The structure of
Si(OMe)3F was optimized at the (GIAO−DFT) B3LYP/6-
311G+(2d,p)//B3LYP/DGDZVP level of theory.17 The
structures of allyl-Si(OMe)3 and Si(OMe)4 were also optimized
in the same manner to corroborate our experimental results
with our computational methodology. Notably, there was a
strong correlation between our experimentally observed and
calculated chemical shifts both in trend and in magnitude
(Table 3). More specifically, the calculated shift for Si(OMe)3F
follows the same bias in relation to allyl-Si(OMe)3 and
Si(OMe)4 observed for the experimentally assigned shifts.
Privy to the above insight and in an attempt to gain a clearer
mechanistic understanding of how complex 6 facilitated
allylation, we monitored by NMR the subsequent effect of
added allyltrimethoxysilane (allyl-Si(OMe)3, 1.5 equiv) on the
cooled (T = 234 K) methanol-d4 solution of the catalyst (R)-
BINAP/AgF (10:6, 0.1 equiv) and benzaldehyde (1 equiv).
Readily apparent from the resulting 31P NMR spectrum was
that the signal assigned to 5 had markedly broadened, whereas
that assigned to 3 had completely disappeared. In view of this
dramatic change, the mixture was cooled further in an attempt
to resolve the broad overlapping signals of 1 and 2. This proved
to be ineffective but nevertheless did lend further support to the
ideas that a (R)-BINAP·(AgF)2*(OC(H)Ph)n complex such
as 6 formed under the reaction conditions, and that this
intermediate was a key mechanistic precursor to allylation.
To further expand upon this last idea, a subsequent in-depth
29Si NMR analysis revealed that besides three major silicon
species observed at −45.9 ppm (s), −77.3 ppm (s), and −84.9
ppm (m), one minor intermediate [−54.2 ppm (m)] was
formed at −80 °C in the reaction mixture. After the mixture
was warmed to −10 °C, the 29Si signals assigned to the major
and minor 29Si adducts of this mixture observed at −45.9 and
−54.2 ppm (m) disappeared, and the relative intensity of the
−77.3 ppm signal to the −84.9 ppm signal increased (see
Supporting Information). As for the identity of the silicon
species observed at −45.9 and −77.3 ppm, it was determined
by direct comparison with both an authentic sample prepared
in-house and reported 29Si NMR data that they corresponded
to allyltrimethoxysilane (−45.9 ppm) and tetramethoxysilane
(TMOS, −77.3 ppm).
Table 3. Computed 29Si Chemical Shifts Using GIAO−DFT
a
Calculations
DGDZVP
(ppm)
6-311G+(2d,p)
(ppm)
experiment
(ppm)
intermediate
allyl-Si(OMe)3
Si(OMe)4
−40.2
−58.3
−66.8
−56.0
−85.2
−91.9
−45.9
−77.3
−84.9
Si(OMe)3F
a
All 29Si NMR chemical shifts were computed using GIAO−DFT at
the B3LYP/DGDZVP and B3LYP/6-311G+(2d,p) levels on opti-
mized structures at the corresponding level of theory. Shifts are relative
to computed shifts of trimethylsilane at the corresponding level of
theory.
Taking into consideration the above results and building
upon related previously reported mechanistic work, we
tentatively propose the mechanistic cycle outlined in Scheme
2. The cycle begins with coordination of a single benzaldehyde
to 3 to generate a rapidly interconverting mixture of catalyst-
bound complexes such as 4, 5, and 6. Of these complexes, the
two-metal center bifurcated mode of Ag binding present in 6
leads to greater activation of the carbonyl carbon of
benzaldehyde, as indicated above by our Mulliken charge
analysis, likely making 6 the predominant species by which
carbon−carbon bond formation occurs. Another notable
feature of 6, as mentioned above, is the fact that the si-face
of benzaldehyde is well-shielded by the aromatic system of the
catalyst, whereas the re-face is sterically unhindered. Assuming
that allylation takes place at the less sterically shielded re-face of
benzaldehyde via a structural arrangement such as 6, the
reaction outcome would be consistent with the stereoselectivity
of the Sakurai−Hosomi−Yamamoto allylation observed ex-
perimentally when using (R)-BINAP. From 6, transallylation
ensues to generate a rapidly interconverting assembly of highly
Also in agreement with this assignment were 29Si−1H HSQC
experiments that revealed the three signals corresponding to
allylic and olefinic protons of the parent allyltrimethoxysilane
were strongly coupled to the silicon signal at −45.9 ppm.
Furthermore, all four of the observed silicon signals were found
to be strongly coupled to protons in the same spectral region as
those of the methoxysilane protons of the parent allylsilane
reagent at ∼3 ppm. In addition, the 29Si signals at −77.3 and
−84.9 ppm were not coupled to any other protons. As for the
minor silicon adduct, there was visible coupling to protons
having significant overlap with the allylic and olefinic protons of
the parent allylsilane. Although this made assignment of this
signal difficult, allyltrialkoxysilanes are known to form activated
silyl anions in the presence of halides;14 however, their 29Si
chemical shifts normally occur between −110 and −130 ppm.15
Consequently, we conjectured that this signal arises from the
transient formation of an activated, hypervalent allyltrimethox-
4443
dx.doi.org/10.1021/jo400451u | J. Org. Chem. 2013, 78, 4440−4445

The Journal of Organic Chemistry
Article
species from allyltrimethoxysilane both in the presence and in
the absence of benzaldehyde. Comparison to authentic samples
and GIAO−DFT calculated shifts suggest that these are TMOS
and fluorotrimethoxysilane. The results of this study provide
valuable insight into the catalytic species and their correspond-
ing intermediary ground state complexes, laying the ground-
work for computational modeling of allylation transition states
and the eventual complete elucidation of this mechanism.
Scheme 2. Proposed Catalytic Cycle
EXPERIMENTAL SECTION
NMR experiments were conducted on an instrument operating at 600
MHz. All reagents were used directly from the bottle without further
purification.
■
Representative NMR Methodology. To a dry NMR tube was
added 1.5 mL of a solution of (R)-BINAP (3.8 mg, 0.0061 mmol) and
AgF (1.5 mg, 0.012 mmol) in methanol-d4. The corresponding 31P,
13C, 1H, and 13C−1H HSQC NMR spectra were obtained at the
reported temperatures.
Benzaldehyde (13 μL, 0.12 mmol) was then added; the sample was
shaken, and the corresponding 31P, 13C, 1H, and 13C−1H HSQC NMR
spectra were obtained at the reported temperatures.
To the above mixture was added allyltrimethoxysilane (30 μL, 0.18
1
mmol); the sample was shaken, and the corresponding 31P, 13C, H,
29Si, 13C−1H HSQC, and 29Si−1H HSQC spectra were obtained at the
reported temperatures.
ASSOCIATED CONTENT
* Supporting Information
■
S
Thermochemical and structural data for calculations in the gas
phase at room temperature for the catalyst−benzaldehyde
complex; thermochemical, GIAO−DFT, and structural data for
calculations using the IEFPCM method for methanol at 193,
212, and 249 K for the catalyst−benzaldehyde complex;
thermochemical, GIAO−DFT, and structural data for silicon
intermediates; and NMR spectra. This material is available free
of charge via the Internet at http://pubs.acs.org.
reactive η1-allyl (7) and η3-allyl (8) species, which, depending
on their lifetimes, would allow for potential regiochemical
scrambling under the reaction conditions. As such, allylation of
benzaldehyde thereafter could foreseeably proceed from either
7 or 8; however carbon−carbon bond formation would take
place at the most substituted terminus of the allyl fragment,
which is in accord with the regioselectivity findings involving
crotylsilanes reported by Yamamoto et al.18 The closed
chairlike transition state (TS2) would help account for the
reported strong anti diastereoselectivity that occurs independ-
ently of the configuration of the starting crotyl- and allylsilanes
in these reactions.19 Alternatively, one could invoke formation
of the Ag−allyl species prior to benzaldehyde binding. In either
case, the following allylation would proceed with the same
regio-, enantio-, and diastereoselectivities. Further modeling of
transition state assemblies for the transallylation and allylation
processes is warranted to determine which of these pathways
are favored. To complete the catalytic cycle, we propose the
coordination of methanol to a silver center that catalyzes
proton transfer to generate the homoallylic alcohol 9 and sets
up for a ligand exchange between fluorotrimethoxysilane and
the silver−methoxy complex to regenerate the catalyst and
form TMOS.20 The exact mechanism of this proton transfer is
not well understood, and we cannot determine if methanol is
involved directly in the allylation transition state or if this
occurs in a stepwise manner after allylation.
AUTHOR INFORMATION
Corresponding Author
*tdudding@brocku.ca
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors thank Compute Canada for use of SHARC Net.
We are grateful to the Natural Sciences and Engineering
Research Council of Canada (NSERC) for funding of this
research.
REFERENCES
■
(1) (a) Hosomi, A.; Sakurai, H. Tetrahedron Lett. 1976, 17, 1295.
(b) Hosomi, A.; Sakurai, H. J. Am. Chem. Soc. 1977, 99, 1673.
(2) (a) Denmark, S. E.; Almstead, N. G. In Modern Carbonyl
Chemistry; Otera, J., Ed.; Wiley-VCH: Weinheim, Germany, 2000; p
299. (b) Chemler, S. R.; Roush, W. R. In Modern Carbonyl Chemistry;
Otera, J., Ed.; Wiley-VCH: Weinheim, Germany, 2000; pp 403.
(3) (a) Yanagisawa, A.; Kageyama, H.; Nakatsuka, Y.; Asakawa, K.;
Matsumoto, Y.; Yamamoto, H. Angew. Chem. 1999, 111, 3916; Angew.
Chem., Int. Ed. 1999, 38, 3701. (b) Yamamoto, H.; Wadamoto, M.
CONCLUSION
■
The results of our research suggest that three distinct
coordination complexes are formed in a mixture of AgF and
(R)-BINAP, only one of which forms a catalytic complex with
benzaldehyde. Computational results suggest that the lowest
energy conformer of this complex corresponds to an η1-bound
complex which by rotation can lead to a highly activated μ2-
bound complex. We have detected the formation of two silicon
Chem.−Asian J. 2007, 2, 692.
For a recent review on catalytic
enantioselective allylation, see: (c) Yus, M.; Gonzalez-Gomez, J. C.;
Foubelo, F. Chem. Rev. 2011, 111, 7774. For recent papers discussing
catalytic enantioselective carbonyl allylation reactions, see: (d) Haddad,
T. D.; Hirayama, L. C.; Singaram, B. J. Org. Chem. 2010, 75, 642.
4444
dx.doi.org/10.1021/jo400451u | J. Org. Chem. 2013, 78, 4440−4445

The Journal of Organic Chemistry
Article
(e) Hrdina, R.; Valterova, I.; Hodacova, J.; Cisarova, I.; Kotora. Adv.
Synth. Catal. 2007, 349, 822.
(4) Ohkouchi, M.; Masui, D.; Yamaguchi, M.; Yamagishi, T. J. Mol.
impact of methanol and the eventual fate of silicon as TMOS lead us
to believe that methanol is involved in the slow regeneration of the
catalyst and formation of TMOS from fluorotrimethoxysilane.
Catal. A: Chem. 2001, 170, 1.
(5) (a) Wadamoto, M.; Yamamoto, H. J. Am. Chem. Soc. 2005, 127,
14556. (b) For a detailed isolation and characterization of an AgOTf/
BINAP catalyst that afforded high levels of enantioselectivity in a
nitrosoaldol addition of allyltin reagents, see: Momiyama, M.;
Yamamoto, H. J. Am. Chem. Soc. 2004, 126, 5360. By scanning
metal:ligand ratios the authors were able to select for the exclusive
formation of one of three silver−BINAP complexes, allowing for
isolation and subsequent structural characterization of these complexes
by X-ray diffraction. The primary catalytic species that was found to be
responsible for the high levels of enantioselectivity observed in this
nitrosoaldol addition was structurally analogous to complex 3
proposed in our study.
(6) (a) Denmark, S. E.; Fu, J. Chem. Rev. 2003, 103, 2763.
(b) Denmark, S. E.; Beutner, G. L. Angew. Chem., Int. Ed. 2008, 47,
1560.
(7) (a) Muetterties, E. L.; Alegranti, C. W. J. Am. Chem. Soc. 1972,
84, 6386. (b) Goel, R. G.; Pilon, P. Inorg. Chem. 1978, 17, 2876.
(8) (a) Gutowsky, H. S.; Holme, C. H. J. Chem. Phys. 1956, 25, 1228.
(b) Evans, M. G.; Polanyi, M. Trans. Faraday Soc. 1935, 31, 875.
(9) (a) Gladysz, J. A.; Boone, B. J. Angew. Chem., Int. Ed. 1997, 36,
550. (b) Reetz, M. T.; Hullmann, M.; Massa, W.; Berger, S.;
̈
Radmacher, P.; Heymanns, P. J. Am. Chem. Soc. 1986, 108, 2405.
(c) Denmark, S. E.; Almstead, N. G. J. Am. Chem. Soc. 1993, 115, 3133.
(d) Peterson, E.; Khalimon, A. Y.; Simionescu, R.; Kuzmina, L. G.;
Howard, J. A. K.; Nikonov, G. I. J. Am. Chem. Soc. 2009, 131, 908.
́
(e) Fernandez, J. M.; Emerson, K.; Larsen, R. H.; Gladysz, J. A. J. Am.
Chem. Soc. 1986, 108, 8268.
(10) Pregosin, P. S.; Ruegger, H.; Salzmann, R.; Albinati, A.; Lianza,
F.; Kunz, R. W. Organometallics 1994, 13, 83.
(11) The DGDZVP basis set has been shown to provide accurate
vibrational frequencies for silver and other transition-metal complexes,
see: (a) Godbout, N.; Salahub, D. R.; Andzelm, J.; Wimmer, E. Can. J.
Chem. 1992, 72, 60. (b) Sosa, C.; Andzelm, J.; Elkin, B. C.; Wimmer,
E.; Dobbs, K. D.; Dixon, D. A. J. Phys. Chem. 1992, 96, 6630.
(12) Our previous studies of modeling transition states of analogous
indium catalysts suggested a bridged coordination mode of the oxygen
in the transition state. See: Mirabdolbaghi, R.; Dudding, T. Org. Lett.
2012, 14, 3748.
(13) Although the use of perturbation theory (MP2) was seen as a
possible resolution to this problem as it accounts for long range
interactions more accurately, see: Gauss, J. Phys. Chem. Chem. Phys.
1995, 99, 1001. Due to the associated computational cost of modeling
a system of this size, and given that the trends in computed
phosphorus shifts were consistent with experiment, the pursuit of
higher level calculations was deemed unnecessary.
(14) (a) Sato, K.; Kira, M.; Sakurai, H. J. Am. Chem. Soc. 1989, 111,
6429. (b) Cerveau, G.; Chuit, C.; Corriu, R. J. P.; Reye, C. J.
Organomet. Chem. 1987, 328, C17. (c) Hosomi, A.; Kohra, S.;
Tominaga, Y. J. Chem. Soc., Chem. Commun. 1987, 1517.
(15) Holmes, R. R. Chem. Rev. 1996, 96, 927.
(16) Wadamoto, M.; Ozasa, N.; Yanagisawa, A.; Yamamoto, H. J.
Org. Chem. 2003, 68, 5593.
(17) The standard methodology established by Cheeseman et al.
indicated that the most accurate chemical shifts can be obtained at the
B3LYP/6-311G+(2d,p) level for DFT calculations. See: Cheeseman, J.
R.; Trucks, G. W.; Keith, T. A.; Frisch, M. J. J. Chem. Phys. 1996, 104,
5497. The 31P shifts could not be calculated with this basis set due to
the presence of silver in the investigated structures.
(18) Wadamoto, M.; Naodovic, M.; Yamamoto, H. Eur. J. Org. Chem.
2009, 5132.
(19) Aoyama, N.; Hamada, T.; Manabe, K.; Kobayashi, S. J. Org.
Chem. 2003, 68, 7329.
(20) It is understood that the presence of methanol leads to a notable
increase in the overall yield of this reaction. See: Wadamoto, M.;
Yamamoto, H. J. Am. Chem. Soc. 2005, 127, 14556. The significant
4445
dx.doi.org/10.1021/jo400451u | J. Org. Chem. 2013, 78, 4440−4445