Chiral zirconium catalysts using multidentate BINOL derivatives for catalytic enantioselective Mannich-type reactions; ligand optimization and approaches to elucidation of the catalyst structure

Article abstract of DOI:10.1021/ja053524d

Catalytic enantioselective Mannich-type reactions of silicon enolates with aldimines were investigated using chiral zirconium catalysts prepared from Zr(O^tBu)₄, N-methylimidazole, and newly designed multidentate BINOL derivatives. Th

Full text of DOI:10.1021/ja053524d

Published on Web 10/18/2005
Chiral Zirconium Catalysts Using Multidentate BINOL
Derivatives for Catalytic Enantioselective Mannich-Type
Reactions; Ligand Optimization and Approaches to
Elucidation of the Catalyst Structure
Yoichi Ihori, Yasuhiro Yamashita, Haruro Ishitani, and Shuj Kobayashi*
Contribution from the Graduate School of Pharmaceutical Sciences, The UniVersity of Tokyo,
Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
Received May 30, 2005; E-mail: skobayas@mol.f.u-tokyo.ac.jp
Abstract: Catalytic enantioselective Mannich-type reactions of silicon enolates with aldimines were
investigated using chiral zirconium catalysts prepared from Zr(OtBu)4, N-methylimidazole, and newly
designed multidentate BINOL derivatives. These new multidentate BINOL ligands were designed on the
basis of an assumed transition state structure of a chiral zirconium catalyst derived from two molecules of
(R)-6,6′-Br2-BINOL. Not only tetradentate BINOL 4 but also tridentate BINOL derivatives were found to
be effective, and high enantioselectivities were attained. In a structural study of the most effective zirconium
complex prepared from tridentate ligand 6e, several NMR experiments and DFT calculations were carried
out. Consequently, the structure of an active catalyst and plausible mechanism of asymmetric induction
were elucidated.
nitrogen-containing natural products.5 We have already reported
catalytic asymmetric Mannich-type reactions of imines prepared
Introduction
Development of chiral catalysts for asymmetric reactions is
one of the most fundamental missions in organic synthesis.1
Among them, chiral Lewis acids are known to be useful catalysts
for several reactions, and many combinations of metals and
chiral ligands have been widely explored.2 BINOL derivatives
are often employed as chiral sources for catalysts and offer
promising possibilities for modification and derivatization;
accordingly many BINOL derivatives have been investigated.3
Recently, multidentate BINOL derivatives have also been
designed for the establishment of precise asymmetric environ-
ments thus achieving high enantioselectivity.4
from aldehydes and 2-aminophenol using a chiral zirconium
complex prepared from zirconium (IV) tert-butoxide and (R)-
6,6′-disubstituted-1,1′-bi-2-naphthols.6 It has also been shown
that these catalytic systems were potent synthetic tools for the
preparation of optically active â-amino acids and their deriva-
tives.7 For example, a synthesis of (2R,3S)-3-phenylisoserine‚
hydrochloride, which is a precursor of the C-13 side chain of
paclitaxel, was demonstrated using (S)-6,6′-Br2-BINOL as
shown in Scheme 1.8
In these reactions the chiral zirconium catalyst formed C2
symmetric structures consisting of 1 equiv of Zr and 2 equiv of
(R)- or (S)-6,6′-Br2-BINOL and more than 2 equiv of DMI or
NMI in which all oxygen atoms of the BINOLs were oriented
equatorial, as confirmed by NMR analyses. DMI or NMI plays
important roles for dissociation of the monometallic Zr-BINOL
complex from the oligomeric species and regulation of a
structure of the Zr complex. In fact, almost no asymmetric
Catalytic asymmetric reactions of imines are of great interest
because of their potential usefulness for the synthesis of
(1) (a) Ojima, I., Ed. Catalytic Asymmetric Synthesis, 2nd ed.; Wiley-VCH:
New York, 2000. (b) Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.
ComprehensiVe Asymmetric Catalysis; Springer: Heidelberg, 1999.
(2) Yamamoto, H., Ed. Lewis Acids in Organic Synthesis; Wiley-VCH:
Weinheim, 2000; Vol. 2.
(3) Recent review of modified BINOL ligands in asymmetric catalysis: (a)
Brunel, J. M. Chem. ReV. 2005, 105, 857. (b) Chen, Y.; Yekta, S.; Yudin,
A. K. Chem. ReV. 2003, 103, 3155.
(4) (a) Harada, S.; Handa, S.; Matsunaga, S.; Shibasaki, M. Angew. Chem.,
Int. Ed. 2005, 44, 4365. (b) Yoshida, T.; Morimoto, H.; Kumagai, N.;
Matsunaga, S.; Shibasaki, M. Angew. Chem., Int. Ed. 2005, 44, 3470. (c)
Matsunaga, S.; Yoshida, T.; Morimoto, H.; Kumagai, N.; Shibasaki, M. J.
Am. Chem. Soc. 2004, 126, 8777. (d) Guo, H.; Wang, X.; Ding, K.
Tetrahedron Lett. 2004, 45, 2009. (e) Harada, T.; Hiraoka, Y.; Kusukawa,
T.; Marutani, Y.; Matsui, S.; Nakatsugawa, M.; Kanda, K. Org. Lett. 2003,
5, 5059. (f) Majima, K.; Takita, R.; Okada, A.; Ohshima, T.; Shibasaki,
M. J. Am. Chem. Soc. 2003, 125, 15837. (g) Harada, S.; Kumagai, N.;
Kinoshita, T.; Matsunaga, S.; Shibasaki, M. J. Am. Chem. Soc. 2003, 125,
2582. (h) Kumagai, N.; Matsunaga, S.; Kinoshita, T.; Harada, S.; Okada,
S.; Sakamoto, S.; Yamaguchi, K.; Shibasaki, M. J. Am. Chem. Soc. 2003,
125, 2169. (i) Kumagai, N.; Matsunaga, S.; Yoshikawa, N.; Ohshima, T.;
Shibasaki, M. Org. Lett. 2001, 3, 1539. (j) Yoshikawa, N.; Shibasaki, M.
Tetrahedron 57, 2569. (k) Kim, Y. S.; Matsunaga, S.; Das, J.; Sekine, A.;
Ohshima, T.; Shibasaki, M. J. Am. Chem. Soc. 2000, 122, 6506. (l) Ishihara,
K.; Yamamoto, H. J. Am. Chem. Soc. 1994, 116, 1561.
(5) Reviews: (a) Kobayashi, S.; Ishitani, H. Chem. ReV. 1999, 99, 1069. (b)
Denmark, S. E.; Nicaise, O. J.-C. In ComprehensiVe Asymmetric Catalysis;
Jacobsen, E. N., Ed.; Springer, Berlin, 1999; p 923. (c) Arend, M.;
Westermann, B.; Risch, N. Angew. Chem., Int. Ed. 1998, 37, 1044. (d)
EnantioselectiVe Synthesis of â-Amino Acids; Juaristi, E., Ed.; VCH:
Weinheim, 1997. (e) Kleinman, E. F. In ComprehensiVe Organic Synthesis;
Trost, B. M., Ed.; Pergamon Press: Oxford, 1991; Chapter 2.3, Vol. 2, p
893.
(6) (a) Kobayashi, S.; Ueno, M. In ComprehensiVe Asymmetric Catalysis;
Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.; Springer: Heidelberg,
1999; supplement 1, p 143. For recent examples using silicon enolates,
see also: (b) Josephsohn, N. S.; Snapper, M. L.; Hoveyda, A. H. J. Am.
Chem. Soc. 2004, 126, 3734. (c) Akiyama, T.; Itoh, J.; Yokota, K.; Fuchibe,
K. Angew. Chem., Int. Ed. 2004, 43, 1566. (d) Kobayashi, S.; Matsubara,
R.; Nakamura, Y.; Kitagawa, H.; Sugiura, M. J. Am. Chem. Soc. 2003,
125, 2507. (e) Jaber, N.; Carre´e, F.; Fiaud, J.-C.; Collin, J. Tetrahedron:
Asymmetry 2004, 14, 2067. (f) Wenzel, A. G.; Jacobsen, E. N. J. Am. Chem.
Soc. 2002, 124, 12964.
9
15528
J. AM. CHEM. SOC. 2005, 127, 15528-15535
10.1021/ja053524d CCC: $30.25 © 2005 American Chemical Society

Catalytic Enantioselective Mannich-Type Reactions
A R T I C L E S
Scheme 1. Synthesis of (2R,3S)-3-Phenylisoserine‚Hydrochloride
Using a Chiral Zirconium Catalyst
induction was observed without DMI or NMI.5a It was assumed
that flipping of one of the four Zr-O bonds from an equatorial
to an apical position occurred when a bidentate imine coordi-
nated; a stereochemical model is shown in Figure 1. Further,
the enantioselectivity observed could be explained by assuming
a transition state derived from flipping A, in which the si-face
of the imine was shielded effectively by a naphthalene ring of
the BINOL; however, a transition state from flipping B would
lead to a less selective reaction. Therefore, we thought that
suppression of flipping B would be a way to improve the
enantioselectivity. The most promising strategy to control the
flipping is fixation of the two BINOLs by a spacer, that is,
design of a novel tetradentate BINOL derivative.
Herein we describe the design and optimization of multiden-
tate BINOL ligands, scope of the reactions, NMR experiments,
and DFT calculations on the catalyst. The assumed structure of
an active catalyst and a plausible mechanism of asymmetric
induction are also proposed.
Results and Discussion
Figure 1. Structure of a chiral zirconium complex of bis(R)-6,6-Br2-
BINOL and assumed stereochemical model.
We designed tetradentate (R)-(R)-bis-BINOL-methane (BBM),9
in which two BINOL molecules in the assumed flipped structure
(7) Recently, several direct asymmetric Mannich-type reactions have been
reported. For direct Mannich-type reactions, see reviews: (a) Co´rdova, A.
Acc. Chem. Res. 2004, 37, 102. For recent examples, see also: (b) Cobb,
A. J. L.; Shaw, D. M.; Longbottom, D. A.; Gold, J. B.; Ley, S. V. Org.
Biomol. Chem. 2005, 3, 84. (c) Notz, W.; Watanabe, S.; Chowdari, N. S.;
Zhong, G.; Betancort, J. M.; Tanaka, F.; Barbas, C. F., III AdV. Synth.
Catal. 2004, 346, 1131. (d) Ibrahem, I.; Casas, J.; Co´rdova, A. Angew.
Chem., Int. Ed. 2004, 43, 6528. (e) Zhuang, W.; Saaby, S.; Jørgensen, K.
A. Angew. Chem., Int. Ed. 2004, 43, 4476. (f) Matsunaga, S.; Yoshida, T.;
Morimoto, H.; Kumagai, N.; Shibasaki, M. J. Am. Chem. Soc. 2004, 126,
8777. (g) Hayashi, Y.; Tsuboi, W.; Ashimine, I.; Urushima, T.; Shoji, M.;
Sakai, K. Angew. Chem. 2003, 115, 3805; Angew. Chem., Int. Ed. 2003,
42, 3677. (h) Hayashi, Y.; Tsuboi, W.; Shoji, M.; Suzuki, N. J. Am. Chem.
Soc. 2003, 125, 11208. (i) Matsunaga, S.; Kumagai, N.; Harada, S.;
Shibasaki, M. J. Am. Chem. Soc. 2003, 125, 4712. (j) Watanabe, S.;
Co´rdova, A.; Tanaka, F.; Barbas, C. F., III. Org. Lett. 2002, 4, 4519. (k)
Co´rdova, A.; Barbas, C. F., III. Tetrahedron Lett. 2002, 43, 7749. (l)
Co´rdova, A.; Watanabe, S.; Tanaka, F.; Notz, W.; Barbas, C. F., III. J.
Am. Chem. Soc. 2002, 124, 1866. (m) Co´rdova, A.; Notz, W.; Zhong, G.;
Betancort, J. M.; Barbas, C. F., III. J. Am. Chem. Soc. 2002, 124, 1842.
(n) List, B.; Porjaliev, P.; Biller, W. T.; Martin, H. J. J. Am. Chem. Soc.
2002, 124, 827. (o) List, B. J. Am. Chem. Soc. 2000, 122, 9336. For
examples using modified esters, see: (p) Bernardi, L.; Gothelf, A. S.; Hazell,
R. G.; Jørgensen, K. A. J. Org. Chem. 2003, 68, 2583. (q) Marigo, M.;
Kjærsgaard, A.; Juhl, K.; Gathergood, N.; Jørgensen, K. A. Chem.sEur.
J. 2003, 9, 2359. (r) Uraguchi, D.; Terada, M. J. Am. Chem. Soc. 2004,
126, 5356 and references therein.
Figure 2. Design of a linked bis-BINOL ligand.
of the zirconium catalyst prepared from (R)-6,6′-Br2-BINOL
were connected with a C1 spacer (Figure 2). BBM and analogues
were synthesized as shown in Scheme 2.9a Ortho-lithiation of
(R)-MOM-BINOL was carried out using Snieckus’ method,10
and then the arylaldehyde derivatives were treated to afford the
(8) Kobayashi, S.; Ishitani, H.; Ueno, M. J. Am. Chem. Soc. 1998, 120, 431.
(9) (a) Ishitani, H.; Kitazawa, T.; Kobayashi, S. Tetrahedron Lett. 1999, 40,
2161. BBM was synthesized by Shibasaki’s group independently: (b) Vogl,
E. M.; Matsunaga, S.; Kanai, M.; Iida, T.; Shibasaki, M. Tetrahedron Lett.
1998, 39, 7917. (c) Matsunaga, S.; Das, J.; Roels, J.; Vogl, E. M.;
Yamamoto, N.; Iida, T.; Yamaguchi, K.; Shibasaki, M. J. Am. Chem. Soc.
2000, 122, 2252.
(10) Cox, P. J.; Wang, W.; Snieckus, V. Tetrahedron Lett. 1992, 33, 2253.
9
J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005 15529

A R T I C L E S
Ihori et al.
Scheme 2. Synthesis of Multidentate BINOL Derivativesa
Table 1. Investigation of BBM 4 Analogues
t
a Conditions: (a) BuLi/TMEDA/Et2O or BuLi/Et2O then ArCHO. (b)
NaBH4/TFA/CH2Cl2. (c) HCl/MeOH. (d) Et3SiH/BF3‚Et2O/CH2Cl2.
Scheme 3. Stereoselection of the Chiral Zirconium Catalysts
obtained in 82% yield with 94% ee when the reaction was
conducted in toluene at 0 °C. Further, when an O-methylated
BBM 5 was used as a chiral ligand in the Zr-catalyzed Mannich-
type reaction, the desired adduct was obtained in 67% yield
with 74% ee (Table 1, entry 2). This result revealed that all
hydroxyl groups of tetradentate ligand 4 might not participate
in the establishment of an effective asymmetric environment
and that tridentate BINOL ligands might be efficient. We then
synthesized and employed a simplified tridentate BINOL
derivative 6a in the Mannich-type reaction. The desired reaction
proceeded smoothly, and the product was obtained in 74% yield
with 69% ee (entry 3). When bidentate benzyl analogue 7 was
employed, the yield and the enantiomeric excess decreased
(entry 4).
corresponding diarylmethanols.11 Deprotection of the MOM
groups and reduction of the dibenzylic position gave multi-
dentate BINOL analogues.
We then performed the Mannich-type reaction of imine 1a
with ketene silyl acetal 2a using a Zr catalyst prepared from
BBM 4. The reaction proceeded but more slowly when
compared with that using the Zr catalyst prepared from (R)-
6,6′-Br2-BINOL. In addition, unexpectedly, (S)-3a was ob-
tained by using the catalyst derived from BBM, while (R)-3a
was obtained when (R)-6,6′-Br2-BINOL was employed.9a That
implies the structure of the catalyst formed by BBM 4 is
completely different from the proposed structure in Figure 2,
and the nucleophile attacked the opposite π-face of the imine
when the catalyst prepared from 4 was used (Scheme 3). These
unexpected results, in contrast to our initial transition state
model, led us to conduct further investigations to clarify the
origin of this unique phenomenon.
To clarify the structure of the chiral zirconium complex
prepared from Zr(OtBu)4, NMI, and BBM 4, several NMR
experiments were conducted. However, 1H and 13C NMR spectra
of the complex were very complicated. It was assumed that the
complexes of zirconium and the linked bis-BINOL derivative
formed oligomeric structures, in contrast to the structure of the
zirconium catalyst prepared from Zr(OtBu)4, 6,6′-Br2-BINOL,
and NMI.12 On the other hand, the Mannich-type adduct was
We then searched for more effective ligands for zirconium
catalysts by screening simplified tridentate BINOL derivatives.
It was assumed that sterically demanding groups at the ortho
position of the phenol moiety are necessary to achieve high
enantioselectivity in tridentate BINOL derivatives, since the
steric bulk of the phenol moiety is lower than that of linked
bis-BINOL 4. Moreover, electron-withdrawing substituents
might be effective to increase the Lewis acidity of the catalyst.
We have already reported that turnover numbers of related
catalytic zirconium systems were much improved by introduc-
tion of electron-withdrawing groups at the 6,6′-positions of the
binaphthyl ring systems due to enhanced Lewis acidity of the
metal.12 Based on these considerations, we prepared several
tridentate BINOL derivatives and used them as ligands in the
Zr-catalyzed Mannich-type reaction of aldimine 1a with ketene
silyl acetal 2a. The results are summarized in Table 2.
In the presence of 10 mol % of a zirconium catalyst, which
was prepared from Zr(OtBu)4, 1.5 equiv of (R)-6, and 1.2 equiv
of NMI, the desired reactions proceeded smoothly in toluene
at 0 °C in almost all cases. In these tridentate ligands,
(11) Diarylmethanols were also prepared by reactions of (R)-3-CHO-MOM-
BINOL with ArLi or ArMgX.
(12) Ishitani, H.; Ueno, M.; Kobayashi, S. J. Am. Chem. Soc. 2000, 122, 8180.
9
15530 J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005

Catalytic Enantioselective Mannich-Type Reactions
A R T I C L E S
Table 3. Catalytic Asymmetric Mannich-Type Reactions
Table 2. Screening of Tridentate BINOL Derivatives
entry
ligand
R1
R2
R3
R4
R5
yield/%
ee/%
1
2
3
4
5
6
7
8
9
6b
6c
6d
6e
6f
6g
6h
6i
CF3
Me
Et
i-Pr
t-Bu
Ph
H
H
i-Pr
i-Pr
H
H
H
H
H
H
CF3
H
H
H
H
H
H
H
H
H
Ph
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
92
85
87
92
80
79
80
74
46
93
74
78
84
87
30
30
75
54
15
86
6j
6k
OMe
H
H
OMe
10
H
introduction of an electron-withdrawing group was not so
effective for improving the reactivity (entries 1 and 7). On the
other hand, it was found that the R1 group had an important
influence on the enantioselectivity and that a suitably sized
substituent R1 was required. Precisely, the size of the R1
correlates with enantioselectivity for smaller groups (entries 2,
3, and 4), yet when R1 is too bulky enantioselectivity is
decreased (entries 5 and 6). The highest enantioselectivity was
obtained when 6e (R1 ) iPr) was employed (entry 4).13 Further,
the substituent of the methylene position was found to play an
important role on the enantioselectivity (6j vs 6k). While 6k
gave high enantioselectivity, 6j showed only 15% ee. In the
case of 6j, it seemed that the methoxy group of the benzylic
position was too close to the activated aldimine in a transition
state (vide infra).
Next, we investigated the Mannich-type reactions of other
aldimines with silicon enolates using chiral zirconium catalysts
prepared from linked bis-BINOL methane 4 and tridentate
BINOL derivative 6e. The results are summarized in Table 3.
Imines derived from aromatic aldehydes having electron-
donating and electron-withdrawing groups worked well to afford
the desired Mannich-type adducts and high enantioselectivities.
Similar high levels of enantioselectivity were obtained when
the silicon enolates 2b derived from ethyl isobutyrate and 2c
from S-ethyl thioacetate were used (entries 4 and 5). The catalyst
prepared from 6e showed much higher activity than that
prepared from 4, and the reactions proceeded at lower temper-
ature using tridentate ligand 6e (entry 1 vs 2). The imines
prepared from sterically hindered aldehydes such as o-tolyl and
1- or 2-naphthylaldehydes, and those prepared from heterocyclic
aldehydes also gave the products in high enantioselectivities.
Moreover, it was also found that the reactions were slightly
accelerated in the presence of a small amount of phenol (entries
a Benzene was used as a solvent instead of toluene. b PhOH (10 mol %)
was added to the catalyst.
7, 9, and 21). In these cases, phenol served as a proton source
in the reaction; no alkoxide exchange with phenol occurred on
the Zr catalyst.14
NMR Experiments. NMR experiments were performed to
identify the nature of chiral zirconium catalyst. The complexes
were prepared from 1 equiv of Zr(OtBu)4, 3 equiv of NMI and
1.0, 1.5, 2.0, and 3.0 equiv of 6e in benzene-d6. After being
1
stirred at room temperature for 1 h, H and 13C NMR were
measured. Independently, we prepared the same catalysts and
(13) Very recently we reported the chiral niobium catalyst using tridentate
BINOL derivative 6e for the Mannich-type reaction: Kobayashi, S.; Arai,
K.; Shimizu, H.; Ihori, Y.; Ishitani, H.; Yamashita, Y. Angew. Chem., Int.
Ed. 2005, 44, 761.
(14) The 13C NMR signal of the phenolic carbon of phenol appeared at 156.2
ppm in benzene-d6. When phenol was added to the Zr catalyst (Table 4,
entry 3), it appeared at 159.1 ppm.
9
J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005 15531

A R T I C L E S
Ihori et al.
Figure 3. 13C NMR spectra of chiral zirconium complexes.
conducted the reaction of 1a with 2a in benzene at room
temperature. The results are shown in Table 4 and Figure 3.
In the case of using 1.0 equiv of 6e, 1H and 13C NMR spectra
showed the generation of three components (A, B, and C), and
A; in addition, complex A was found to contain 1 equiv of NMI.
In the case of using 1.5 equiv of 6e, complex B was produced
predominantly, and the molar ratio of complexes A, B, C was
0.21:1.2:1.0. It was found that the complex B held 6e of two
descriptions and that 13C NMR signals of the phenolic carbons
appeared at 152.5, 158.0, 158.9, 159.3, 160.9, and 162.4 ppm.
These peaks indicated that five Zr-O-Ar moieties existed and
that one hydroxyl group of the sidearm remained free in the
complex B. On the other hand, when more than 2.0 equiv of
the ligand were used, complex A was not detected. The NMR
1
the molar ratio of these complexes was 3.4:1.2:1.0. Both H
and 13C NMR spectra revealed that complex A held a unique
6e. The 13C NMR signals of phenolic carbons appeared at 159.6,
159.8, and 160.5 ppm, while the signals of the phenolic carbons
of free 6e appeared at 151.2, 152.3, and 154.0 ppm. These peaks
indicated that three Zr-O-Ar moieties existed in the complex
9
15532 J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005

Catalytic Enantioselective Mannich-Type Reactions
A R T I C L E S
Table 4. NMR Experiments of the Chiral Zirconium Complex of 6e
entry
ligand (equiv)
A:B:C
yield/%a
ee/%a
1
1.0
1.0
1.0
1.5
2.0
2.0
3.0
3.4:1.2:1.0
0.38:1.2:1.0
0.26:1.2:1.0
0.21:1.2:1.0
0.0:1.4:1.0
0.0:1.4:1.0
0.0:1.3:1.0
94
92
2b
3c
4
88
92
95
91
91
93
5
6b
7
65
91
a The reactions of 1a with 2a were performed using the corresponding
catalyst. b 1 equiv of 1a was added. c 1 equiv of PhOH was added.
Figure 5. Correlation between ee of ligand 6e and ee of the product.
complexes scarcely influenced enantioselectivities (entries 1,
3, 4, 5, and 6). Only when 3 equiv of 6e were used did the
reaction rate decreased slightly (entry 7).
We also investigated on a nonlinear effect to further elucidate
the actual catalyst structure.16 The correlation between the ee
of the ligand 6e and the ee of the product was carefully
examined, and it was concluded that no nonlinear effect was
observed in the reaction of 1a with 2a, as shown in Figure 5.
The result may suggest that the dimeric complex (B or C) having
two ligands is not an active catalyst species.
We assumed that complex C was inert as an active catalyst
due to steric hindrance. It would be difficult for the complex C
to activate an imine via bidentate coordination, because complex
C could not flip like 6,6′-Br2-BINOL due to the bulky
substituent at the 3-position of BINOL. Since the structure of
the complex B is rather flexible regarding the methylene groups
at the benzylic positions, the remarkable effect of a methoxy
group at the benzylic positions shown in Table 2, entry 9 vs 10
may not be explained by assuming complex B as an active
catalyst. Moreover, it was also suggested that the complex A
was not an active catalyst, because high enantioselectivity was
obtained in the absence of complex A (Table 4, entries 5 and
7).
We proposed a plausible mechanism of the complex forma-
tion as shown in Scheme 4. Complexes A, B, and C were
assumed to be generated from common intermediate D.
Complex A was formed by an intramolecular reaction of D.
Complex B was produced by an intermolecular reaction between
two molecules of D. Further, complex C was formed by an
intermolecular reaction of D with free 6e. Consequently, we
considered that complex A was a precursor of an active catalyst
and that the formation of the active catalyst was accelerated by
addition of a proton source like imine 1a or phenol. It was
assumed that the active catalyst was generated from complex
A and existed as a transient species or in very small amounts
in the reaction system, which were not detected by our NMR
analyses. Complex B might serve as a precursor of A, thus,
also a precursor of the active catalyst.17 Complex C might be
inert as a precursor due to stable bidentate coordination of the
ligands.
Figure 4. Assumed structures of the chiral zirconium complexes A-C.
spectra revealed that the complex C held a unique 6e and that
13C NMR signals of phenolic carbons appeared at 151.9, 158.3,
and 161.2 ppm. These peaks indicated that two Zr-O-Ar
moieties existed in complex C. Complex C was considered to
be a 1:2 complex of Zr and 6e with a highly symmetrical
structure, although it was formed in the presence of a large
excess of ligand 6e. The assumed structures of complexes A-C
are shown in Figure 4.
When 1 equiv of imine 1a was added to the catalyst prepared
from 1 equiv of 6e, formation of a new complex consisting of
the catalyst and the imine was not observed in the NMR spectra.
However, the molar ratio of complexes A, B, and C changed
to 0.38:1.2:1.0 (Table 4, entry 1 vs 2); that is, the ratio of
complex A was decreased.15 A similar tendency was observed
when 1 equiv of phenol was added (entry 1 vs 3). When using
2 equiv of 6e, the molar ratio of complexes A, B, and C did
not change (entry 5 vs 6). It was assumed that a phenolic proton
promoted a transformation of complex A into a new species
(vide infra). It is interesting that, although Mannich-type
reactions proceeded using these catalysts, the ratio of these
(15) When 1H NMR of complexes (a) and (c) in Figure 3 were measured in the
presence of hexamethylbenzene as an internal standard, it was observed
that the amount of complex A was decreased alone. The amounts of
complex B and C did not change at all in these cases.
(16) (a) Kagan, H. B. Synlett 2001, 888. (b) Girard, C.; Kagan, H. B. Angew.
Chem., Int. Ed. 1998, 37, 2923.
(17) It was observed that a quantity of complex B decreased as the Mannich-
type reaction proceeded, confirmed by other 1H NMR experiments.
9
J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005 15533

A R T I C L E S
Ihori et al.
Scheme 4. Formation and Assumed Structures of the Chiral
Zirconium Complexes with 6e
Figure 6. Calculated structures and relative energies of complex A.
conformers with lower energies for complex A. Then, these
conformers were calculated by the DFT method as mentioned
above and were shown in Figure 6.
The zirconium atom possessed a trigonal bipyramidal struc-
ture in all conformers; it turned out that conformer A-1 had the
lowest energy among all of them. In complex A-1, two oxygen
atoms (Oa and Oc) were oriented in apical positions, while one
oxygen atom (Ob), tert-butoxide, and NMI were in equatorial
positions. However, since NMI was assumed to be eliminated
from this complex by steric repulsion, the structure of the
zirconium turns into a tetrahedron, which seemed to be
coordinated by other ligands easily. Considering these results,
the imine might coordinate to this tetrahedral zirconium complex
and to be activated toward a nucleophilic attack of the ketene
silyl acetal; namely, the tetrahedral zirconium complex (E)
seemed to be an active catalyst. A stable structure of the
tetrahedral complex E was also estimated by the DFT calculation
(Figure 7).
DFT Calculations. To obtain further structural information
about the active catalyst, we performed calculations on complex
A; all calculations were carried out using TITAN molecular
modeling package.18 DFT calculations were done using JAG-
UAR quantum chemistry software19 including TITAN. All
starting geometries for DFT calculations were searched for
beforehand by molecular mechanics (Monte Carlo method) using
Merck’s MMFF94 force field.20 The basis set used for geometry
optimization is LACVP** which is a basis set of double-ú
quality incorporating an effective core potential of Hay and
Wadt21 for the Zr and using the 6-31G** basis set for the other
atoms. We employed the B3LYP functional,22 Becke’s three-
parameter hybrid functional23 for the exchange-correlation
energy combined with the Lee-Yang-Parr correlation func-
tional.24 Molecular mechanics calculations extracted seven
It is noted that the calculated structure of 6e in the complex
E was similar to that in the complex A-1. Therefore, transfor-
mation of complex A to complex E and its reverse reaction
would occur smoothly. The calculated structure of complex E
also indicated that an imine could approach the zirconium atom
from the opposite side to the Oa-Zr bond due to the steric
hindrance. Based on these results, the assumed transition state
model is shown in Figure 8. The complex of the active catalyst
and the imine was similar to the structure shown in Scheme 5.
(18) TITAN, version 1.0; Wavefunction, Inc., Schro¨dinger, Inc., 1999.
(19) JAGUAR, version 3.5; Schro¨dinger, Inc., 1998.
(20) Halgren, T. A. J. Comput. Chem. 1996, 17, 490.
(21) (a) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270. (b) Wadt, W.
R.; Hay, P. J. J. Chem. Phys. 1985, 82, 284. (c) Hay, P. J.; Wadt, W. R.
J. Chem. Phys. 1985, 82, 299.
(22) Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J. J. Phys.
Chem. 1994, 98, 11623.
(23) (a) Becke, A. D. J. Chem. Phys. 1992, 96, 2155-2166. (b) Becke, A. D.
J. Chem. Phys. 1993, 98, 1372. (c) Becke, A. D. J. Chem. Phys. 1993, 98,
5648.
(24) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. 1988, B41, 785.
9
15534 J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005

Catalytic Enantioselective Mannich-Type Reactions
A R T I C L E S
substituent on the dibenzylic position of 6e gave a significant
effect on the enantioselection of the reaction. In the case using
6k, the enantioselectivity was not affected; however, a signifi-
cant decrease of enantiomeric excess was observed using 6j as
a ligand. These results can be clearly explained from the
transition state model in Figure 8.
Since the methoxy group of 6j has an apparent steric repulsion
with the imine in the model, the complex seems to be unstable.
In contrast, the methoxy group in the complex prepared from
6k has a lesser steric effect in the model. Linked bis-BINOL
methane derivatives 4 would also behave as tridentate BINOL
6e, and identical stereoselectivity was observed between them.
Figure 7. Calculated structures of tetrahedral complex E.
Conclusion
Asymmetric catalysts with chiral zirconium complexes pre-
pared from multidentate BINOL derivatives were investigated
in Mannich-type reactions of aldimines with silicon enolates.
An initial ligand design as the transition state mimic led to an
unexpected inversion of the absolute configuration of the
Mannich adducts. Optimized zirconium catalysts using multi-
dentate BINOL derivatives, linked bis-BINOL or tridentate
BINOL derivatives, showed high enantioselectivities. The
structure of the chiral zirconium complex prepared from
Zr(OtBu)4, a tridentate BINOL derivative and NMI, was
investigated, and more than three kinds of complexes were
observed by NMR analyses. DFT calculations revealed the stable
structure of complex A and predicted an active tetrahedral
species (complex E). A mechanism of the asymmetric induction
was proposed, and a clear explanation of the ligand effect was
obtained.
Figure 8. Assumed transition state model.
Scheme 5. Assumed Mechanism of the Exchange of NMI for
Imine
At the starting point of this study, the well-designed tetra-
dentate BINOL derivative, BBM, was thought to work well in
our assumed transition state of the Mannich-type reaction.
However, the nature of the zirconium complexes was beyond
our knowledge and gave us unexpected and interesting results.
We believe that this new finding could lead to new effective
design of chiral zirconium catalysts.
Acknowledgment. This work was partially supported by
CREST, SORT, ERATO, Japan Science Tecnology (JST), and
Grand-in-Aid for Scientific Research from Japan Society of the
Promotion of Sciences (JSPS). We thank Mr. Takayuki
Kitazawa for his contribution to the early stage of this work.
Unfortunately, the SCF convergence was unsuccessful for this
complex.
In the model, the re face of the imine is shielded from
nucleophiles by the phenoxy sidearm of 6e. As mentioned in
Table 2, entries 9 and 10, stereochemistry of a methoxy
Supporting Information Available: Experimental details and
physical data (PDF). This material is available free of charge
via the Internet at http://pubs.acs.org.
JA053524D
9
J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005 15535