Enantioselective mannich-type reactions using a novel chiral zirconium catalyst for the synthesis of optically active β-amino acid derivatives

Article abstract of DOI:10.1021/ja001642p

Catalytic enantioselective Mannich-type reactions of silyl enol ethers with aldimines have been successfully performed using a novel chiral zirconium catalyst prepared from zirconium(IV) tert-butoxide (Zr(OrBu)₄), 2 equiv of (R)-6,6'-dibromo-1,

Full text of DOI:10.1021/ja001642p

8180
J. Am. Chem. Soc. 2000, 122, 8180-8186
Enantioselective Mannich-Type Reactions Using a Novel Chiral
Zirconium Catalyst for the Synthesis of Optically Active â-Amino
Acid Derivatives
Haruro Ishitani, Masaharu Ueno, and Sh uj Kobayashi*
Contribution from the Graduate School of Pharmaceutical Sciences, The UniVersity of Tokyo, CREST,
Japan Science and Technology Corporation (JST), Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
ReceiVed May 12, 2000
Abstract: Catalytic enantioselective Mannich-type reactions of silyl enol ethers with aldimines have been
successfully performed using a novel chiral zirconium catalyst prepared from zirconium(IV) tert-butoxide
t
(
Zr(O Bu)4), 2 equiv of (R)-6,6′-dibromo-1,1′-bi-2-naphthol ((R)-6,6′-Br2BINOL), and N-methylimidazole. The
use of aldimines having N-substituted hydroxyphenyl moieties is essential for obtaining high selectivities, and
the N-substituted groups were converted to free amines using oxidative cleavage. Aldimines derived from
aromatic aldehydes as well as heterocyclic and aliphatic aldehydes reacted with silyl enol ethers smoothly to
afford the corresponding â-amino acid derivatives in high yields and high enantioselectivities. Several NMR
experiments have been conducted to clarify the structure of the chiral Zr catalyst and also the catalytic cycle
of this asymmetric reaction. Finally, a new BINOL derivative, (R)-6,6′-bis(trifluoromethyl)-1,1′-bi-2-naphthol
((R)-6,6′-(CF3)2BINOL), has been prepared. It was shown that the turnover of the catalyst using this novel
ligand was improved, and high levels of yields and selectivities were obtained in the presence of a small
amount of the zirconium catalyst.
Introduction
be difficult to perform. In 1997, we reported the first example
of truly catalytic enantioselective Mannich-type reactions of
aldimines with silyl enolates using a novel zirconium catalyst6
and now in this article describe full details including the general
idea, catalyst structure, reaction mechanism, and scope of these
reactions. A novel efficient catalyst which attains high turnover
of the catalyst in this reaction is also reported.
Asymmetric Mannich-type reactions of aldimines with enolate
components are one of the most important carbon-carbon bond-
forming reactions in organic synthesis. The reactions provide
useful routes for the synthesis of chiral â-amino ketones or esters
(Mannich bases), which are versatile chiral building blocks for
the synthesis of many nitrogen-containing biologically important
compounds including â-amino acids and â-lactams. While
1
several asymmetric Mannich-type reactions using stoichiometric
Results and Discussion
2
,3
amounts of chiral sources have already been reported, very
4
little is known concerning catalytic asymmetric versions. This
To achieve truly efficient catalytic enantioselective Mannich-
type reactions of aldimines with silyl enolates, it was thought
that the choice of metals of Lewis acid catalysts was one of the
is in contrast to the recent progress achieved in catalytic enantio-
selective aldol reactions of aldehydes with enolate components,
5
especially silyl enolates (Mukaiyama aldol reactions). While
chiral Lewis acids have played leading roles in these reactions,
it is assumed that most Lewis acids would be trapped by basic
nitrogen atoms of the starting materials and/or products in
Mannich-type reactions and that truly catalytic reactions would
(3) For enantioselectve Mannich-type reactions using stoichiometric
amounts of chiral sources: (a) Corey, E. J.; Decicco, C. P.; Newbold, R.
C. Tetrahedron Lett. 1991, 39, 5287. (b) Ishihara, K.; Miyata, M.; Hattori,
K.; Tada, T.; Yamamoto, H. J. Am. Chem. Soc. 1994, 116, 10520. (c) Enders,
D.; Ward, D.; Adam, J.; Raabe, G. Angew. Chem., Int. Ed. Engl. 1996, 35,
981. See also: Risch, N.; Esser, A. Liebigs Ann. Chem. 1992, 233.
(
1) Kleinman, E. F. In ComprehensiVe Organic Synthesis; Trost, B. M.,
Ed.; Pergamon Press: Oxford, 1991; Chapter 2.3, Vol. 2, p 893.
2) Diastereoselective asymmetric Mannich-type reactions. For ex-
ample: (a) Fujisawa, T.; Kooriyama, Y.; Shimizu, M. Tetrahedron Lett.
1
3
(4) Quite recently, several catalytic enantioselective Mannich-type reac-
tions have been reported: (a) Fujieda, H.; Kanai, M.; Kambara, T.; Iida,
A.; Tomioka, K. J. Am. Chem. Soc. 1997, 119, 2060. (b) Hagiwara, E.;
Fujii, A.; Sodeoka, M. J. Am. Chem. Soc. 1998, 120, 2474. (c) Ferraris, D.;
Young, B.; Dudding, T.; Lectka, T. J. Am. Chem. Soc. 1998, 120, 4548.
(d) Fujii, A.; Hagiwara, E.; Sodeoka, M. J. Am. Chem. Soc. 1999, 121,
5450. (e) Yamasaki, S.; Iida, T.; Shibasaki, M. Tetrahedron 1999, 55, 8857.
(f) Yamada, K.-I.; Harwood, S. J.; Gr o¨ ger, H.; Shibasaki, M. Angew. Chem.,
Int. Ed. Engl. 1999, 38, 3504. (g) Kambara, T.; Tomioka, K. Chem. Pharm.
Bull. 1999, 47, 720.
(5) For review articles of this topic: (a) Carreira, E. M. In ComprehensiVe
Asymmetric Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamoto Y., Eds.;
Springer: Heidelgerg, 1999. Vol. 3, p 998. (b) Mahrwald, R. Chem. ReV.
1999, 99, 1095. (c) Groger, H.; Vogel, E. M.; Shibasaki, M. Chem. Eur. J.
1998, 4, 1137. (d) Nelson, S. G. Tetrahedron: Asym. 1998, 9, 357. (e)
Bach, T. Angew. Chem., Int. Ed. Engl. 1994, 33, 417.
(
996, 37, 3881. (b) Arend, M. Risch, N. Angew. Chem., Int. Ed. Engl. 1995,
4, 2861. (c) Matsumura, Y. Tomita, T. Tetrahedron Lett. 1994, 35, 3737.
d) Page, P. C. B.; Allin, S. M.; Collington, E. W.; Carr, R. A. E. J. Org.
(
Chem. 1993, 58, 6902. (e) Frauenrath, H.; Arenz, T.; Raabe, G.; Zorn, M.
Angew. Chem., Int. Ed. Engl. 1993, 32, 83. (f) Nolen, E. G.; Aliocco, A.;
Broody, M.; Zuppa, A. Tetrahedron Lett. 1991, 32, 73. (g) Evans, D. A.;
Urpi, F.; Somers, T. C.; Clark, J. S.; Bilodeau, M. T. J. Am. Chem. Soc.
990, 112, 8215. (h) Katritzky, A. R.; Harris, P. A. Tetrahedron 1990, 46,
87. (i) Kunz, H.; Pfrengle, W. Angew. Chem., Int. Ed. Engl. 1989, 28,
067. (j) Oppolzer, W.; Moretti, R.; Thomi, S. Tetrahedron Lett. 1989, 30,
603. (k) Gennari, C.; Venturini, I.; Gislon, G.; Schimperna, G. Tetrahedron
1
9
1
5
Lett. 1987, 28, 227. (l) Broadley, K.; Davies, S. G. Tetrahedron Lett. 1984,
2
5, 1743. (m) Seebach, D.; Betschart, C.; Schiess, M. HelV. Chim. Acta
(6) Ishitani, H.; Ueno, M.; Kobayashi, S. J. Am. Chem. Soc. 1997, 119,
7153.
1
984, 67, 1593.
1
0.1021/ja001642p CCC: $19.00 © 2000 American Chemical Society
Published on Web 08/11/2000

Synthesis of Optically ActiVe â-Amino Acid DeriVatiVes
J. Am. Chem. Soc., Vol. 122, No. 34, 2000 8181
Table 1. Effect of Zirconium Compounds (Eq 1)a
Table 2. Effect of Several Reaction Parameters (Eq 1)
catalyst
catalysta
entry
precursor
ZrCl
Zr(OTf)
ligand
base
time/h yield/% ee/%
mol %
entry [Zr]b ligand additivec temp/°C time/h yield/% ee/%
b
n
n
c
c
1
2
3
4
5
4
4a
4a
4a
4a
BuLi
BuLi
30
16
30
30
30
trace
b
4
quant
quant
quant
quant
2
2
9
1
2
3
4
5
6
7
8
9
20
20
20
10
10
20
20
20
20
10
10
10
5
4a
4a
4a
4a
4a
4a
4a
4a
4b
4b
4c
4d
4b
4b
4b
DTBP
NMI
NMI
NMI
NMI
2-MI
DMI
DMI
NMI
NMI
NMI
NMI
NMI
DMI
NMI
0
0
-15
-15
-45
0
16
16
16
30
30
16
16
16
16
16
16
16
16
16
30
quant
quant
80
quant
quant
37
63
47
73
quant
98
30
45
70
49
58
51
68
70
90
92
93
87
95
91
86
i
b
Zr(O Pr)
4
t
b
Zr(O Bu)
4
4
t
c
Zr(O Bu)
4a
38
a
b
The reaction was carried out using 10 mol % of Zr at -45 °C. 10
c
mol %. 20 mol %.
0
-15
-15
-45
-45
-45
-45
-15
-45
most important keys. After careful examination of various
metals, we decided to use zirconium(IV) as a metal center.
Another difficulty is the rather flexible aldimine-chiral Lewis
acid complexes which often have several stable conformers
7
,8
10
1
1
1
1
1
2
3
90
69
quant
75
9
(including the E/Z isomers of aldimines), while aldehyde-chiral
4
5
2
10
d
Lewis acid complexes are believed to be rigid. Therefore, in
the addition to aldimines activated by chiral Lewis acids, many
transition states would exist to decrease selectivities. To address
these issues, we planned to utilize a bidentate chelation (vide
infra).11
15
a
b
t
. cDTBP, 2,6-di-tert-
[
Zr]/ligand/additive ) 1/2.2/1.2. Zr(O Bu)
butylpyridine; NMI, N-methylimidazole; 2-MI, 2-methylimidazole;
DMI, 1,2-dimethylimidazole. [Zr]/ligand/additive ) 1/2.2/3.
4
d
Chart 1
We selected the aldimine 1c, and first, the reaction with silyl
enolate 2a was investigated. The effect of zirconium compounds
is shown in Table 1. In the presence of a zirconium compound
prepared from zirconium(IV) chloride (ZrCl4) and lithiated (R)-
1
,1′-bi-2-naphthol ((R)-BINOL), the reaction of 1c with 2a
proceeded sluggishly (entry 1). The zirconium compound
derived from zirconium(IV) triflate (Zr(OTf)4)/lithiated (R)-
i
BINOL or zirconium(IV) isopropoxide (Zr(O Pr)4) or zirconium-
appeared after 20 min. We thought that this result indicated
formation of an oligomeric or a polymeric structure of the
catalyst which might lead to the low selectivities in the Mannich
reactions. To address this problem, several additives were
t
(IV) tert-butoxide (Zr(O Bu)4)/(R)-BINOL catalyzed the reac-
tion, but almost no chiral induction was obtained (entries 2-4).
On the other hand, moderate selectivity was observed when the
t
catalyst was prepared from Zr(O Bu)4 (10 mol %) and 2 equiv
1
4
examined (Table 2). It was found that imidazole derivatives
of (R)-BINOL (20 mol %) (entry 5).1
2,13
1
5
gave promising results. While 45% ee of the product was
obtained by using N-methylimidazole (NMI) as an additive at
0
°C, higher enantiomeric excess was observed when the
reaction was carried out at -15 °C (entries 2 and 3). It was
also revealed that 1,2-dimethylimidazole (DMI) was an effective
additive and that the corresponding product was obtained in
the same level of enantioselectivity, while the yield was
decreased. To improve the yield and enantioselectivity of the
reaction, several conditions were further examined. It was
exciting to find that the enantiomeric excess was dramatically
increased to 90% when the catalyst was prepared using (R)-
(1)
6
,6′-dibromo-BINOL (4b; 6,6′-Br2BINOL; see Chart 1).16 The
same levels of the yields and selectivities were obtained when
R)-6,6′-dichloro-BINOL (4c) or (R)-6,6′- difluoro-BINOL (4d)
(
While a homogeneous solution was formed by combining
Zr(O Bu)4 and (R)-BINOL in dichloromethane, precipitates
was employed (entries 11 and 12). It was revealed later that
the effect of the substituents at the 6,6′-positions of BINOL
was a key to obtain high levels of yields, selectivities, and,
particularly, turnover of the catalyst. These results will be
discussed in the last part of this paper. On the other hand, high
enantioselectivities were still obtained when the catalyst loading
t
(7) Rare earths are also promising candidates for the catalytic activation
of aldimines. See: (a) Kobayashi, S.; Araki, M.; Ishitani, H.; Nagayama,
S.; Hachiya, I. Synlett 1995, 233. (b) Kobayashi, S.; Araki, M.; Yasuda,
M. Tetrahedron Lett. 1995, 36, 5773.
(8) Different types of chiral zirconium catalysts that are effective in ring-
opening reactions of epoxides, Diels-Alder reactions, and polymerization,
etc., were reported: (a) Nugent, W. A. J. Am. Chem. Soc. 1992, 114, 2768.
b) Bedeschi, P.; Casolari, S.; Costa, A. L.; Tagliavini, E.; Umani-Ronchi,
A. Tetrahedron Lett. 1995, 36, 7897. (c) Hoveyda, A. H.; Morken, J. P.
Angew. Chem., Int. Ed. Engl. 1996, 35, 1262, and references therein.
(12) Catalytic enantioselective allylation of aldehydes using a zirconium/
BINOL (1/1) catalyst was reported: (a) Bedeshi, P.; Casolari, S.; Costa,
A. L.; Tagliavini, E.; Umani-Ronchi, A. Tetrahedron Lett. 1995, 36, 7897.
(b) Yu, C.-M.; Yoon, S.-K.; Choi, H.-S.; Baek, K. J. Chem. Soc., Chem.
Commun. 1997, 763.
(
(
9) (a) McCarty, C. G. The Chemistry of the Carbon-Nitrogen Double
t
Bond, Patai, S., Ed.; John Wiley & Sons: New York, 1970; Chapter 9. (b)
Biørgo, J.; Boyd, D. R.; Watson, C. G.; Jennings, W. B. J. Chem. Soc.,
Perkin Trans. 2 1974, 757.
(13) The catalyst was prepared without removing BuOH.
(14) About 20 additives were examined.
(15) We expected the imidazole derivatives coordinated the zirconium
to form a monomeric structure.
(
10) (a) Corey, E. J.; Rohde, J. J.; Fischer, A.; Azimioara, M. D.
Tetrahedron Lett. 1997, 38, 33. (b) Corey, E. J.; Rohde, J. J. Tetrahedron
Lett. 1997, 38, 38. (c) Corey, E. J.; Barnes-Seeman, D.; Lee, T. W.
Tetrahedron Lett. 1997, 38, 1699.
(16) For the use of 6,6′-dibromo-1,1′-bi-2-naphthol, see: (a) Terada, M.;
Motoyama, Y.; Mikami, K. Tetrahedron Lett. 1994, 35, 6693. (b) Sasai,
H.; Tokunaga, T.; Watanabe, S.; Suzuki, T.; Itoh, N.; Shibasaki, M. J. Org.
Chem. 1995, 60, 7388.
(11) Ishitani, H.; Kobayashi, S. Tetrahedron Lett. 1996, 37, 7357.

8182 J. Am. Chem. Soc., Vol. 122, No. 34, 2000
Ishitani et al.
Table 3. Catalytic Enantioselective Mannich-Type Reactions
Using Ciral Ligand 4b (Eq 2)a
Scheme 1. Conversion to â-Amino Ester
Scheme 2. Formation of a Novel Zirconium Catalyst
a
b
[
Zr]/4b/additive ) 1/2.2/1.2, -45 °C, 16 h. The aldimine prepared
from cyclohexanecarboxaldehyde and 2-amino-m-cresol was used.
When the reaction was carried out at -23 °C, 71% yield and 71% ee
were obtained. The aldimine prepared from isovaleraldehyde and
c
2
-amino-m-cresol was used.
m-cresol (entry 8). It is noteworthy that the use of 2-amino-m-
cresol was effective not only in stabilizing the aldimine but also
in obtaining high stereoselectivity. Namely, only poor chiral
induction was observed when the aldimine derived from
cyclohexanecarboxaldehyde and 2-aminophenol was used (less
than 10% ee); however, the enantiomeric excess was greatly
improved when the aldimine derived from 2-amino-m-cresol
was employed. The methyl group of 2-amino-m-cresol would
prevent E/Z isomerization of the aldimine-Lewis acid complex
due to the steric repulsion between the cyclohexyl and methyl
groups. Similarly, a high level of enantioselectivity was observed
in the reaction of isovaleraldehyde (entry 9).
(
2)
was reduced to 2-10 mol %. Although the yield was decreased
when 5 mol % of the chiral catalyst including NMI was used
at -45 °C (entry 13), both yield and enantioselectivity attained
a satisfactory level by using DMI instead of NMI at -15 °C
entry 14). It should be noted that the same high level of
enantiomeric excess was obtained when 2 mol % of the chiral
catalyst was employed (entry 15).
The N-substituent of the product was easily removed accord-
ing to Scheme 1. Thus, methylation of the phenolic OH of
product 3a using methyl iodide and potassium bicarbonate and
deprotection using cerium ammonium nitrate (CAN) gave the
1
7
(
1
9
â-amino ester. The absolute configuration assignment was
At this stage, several control experiments were performed.
n
i
18
t
made by comparison of the optical rotation of the amino ester
When Zr(O Pr)4 or Zr(O Pr)4 was used instead of Zr(O Bu)4
in the reaction of 1c with 2a under the same reaction conditions
shown in Table 2, entry 10, high yields and selectivities were
also obtained (Zr(O Pr)4 (10 mol %), 88% yield, 93% ee; Zr-
O Pr)4 (10 mol %), quantitative, 88% ee). On the other hand,
2
0
with that in the literature.
Structure of the Chiral Catalyst. NMR experiments were
n
performed to clarify the structure of the chiral zirconium catalyst.
i
t
(
The catalyst was prepared from 1 equiv of Zr(O Bu)4, 2 equiv
it was revealed that the hydroxyphenyl group in 1c was essential
to obtain high selectivity. While 1c was treated with 2a under
the conditions shown in Table 2, entry 10, to afford the
corresponding adduct (3c) quantitatively with 92% ee, almost
no chiral induction (<5% ee) was obtained when the aldimine
prepared from 1-naphthylaldehyde and aniline or 1-naphthyla-
ldehyde and 2-methoxyaniline was used under the same reaction
conditions.
Other aldimines and silyl enolates were tested, and the results
are summarized in Table 3. Not only aldimines derived from
aromatic aldehydes but also aldimines derived from heterocyclic
aldehydes worked well in this reaction, and good to high yields
and enantiomeric excesses were obtained. Similar high levels
of enantiomeric excess were also obtained when the silyl enol
ether 2b derived from S-ethyl thioacetate was used. While
aldimines derived from aliphatic aldehydes such as cyclohex-
anecarboxaldehyde were known to be unstable, they worked
efficiently when 2-aminophenol was substituted with 2-amino-
of (R)-6,6′-Br2BINOL, and 2 equiv of NMI in benzene-d6. After
stirring at room temperature for 1 h, the solvent was removed
1
13
and H and C NMR were measured in dichloromethane-d2.
1
13
21
The H and C NMR spectra of the isolated catalyst and (R)-
6,6′-Br2-BINOL indicated that the molar ratio of Zr, (R)-6,6′-
2
2
1
Br2BINOL, and NMI in the catalyst was 1:2:2. Both H and
C NMR spectra revealed that the four naphthalene rings of
1
3
the zirconium catalyst were equivalent. In addition, the 13C NMR
chemical shift of phenolic carbons appeared at 159.6 ppm as a
single peak, while the chemical shift of the phenolic carbon of
free (R)-6,6′-Br2BINOL appeared at 152.7 ppm. These peaks
indicated that Zr-O-Ar moieties existed in the zirconium
catalyst and that the catalyst had a highly symmetrical form.
From these spectra, the structure of the catalyst was tentatively
(
19) Kronenthal, D. R.; Han, C. Y.; Taylor, M. K. J. Org. Chem. 1982,
4
7, 2765.
(20) Kunz, H.; Schanzenbach, D. Angew. Chem., Int. Ed. Engl. 1989,
28, 1068.
(21) The same levels of reactivity and selectivity were observed when
(
17) The same level of enantiomeric excess was obtained when the
using the isolated catalyst. The reaction of 1c with 2a gave 3c quantitatively
in 90% ee (Cf. Table 3, entry 3.).
(22) See Supporting Information.
reaction was performed at -78 °C.
n
i
t
(18) Zr(O Pr)4 and Zr(O Pr)4 are less expensive than Zr(O Bu)4.

Synthesis of Optically ActiVe â-Amino Acid DeriVatiVes
J. Am. Chem. Soc., Vol. 122, No. 34, 2000 8183
Scheme 3. Assumed Catalytic Cycle of the Mannich-Type Reaction
Scheme 4. Crossover Experiments
determined to be 5a as shown in Scheme 2. Two (R)- 6,6′-Br2-
BINOLs locate at equatorial positions around the zirconium,
and two NMIs coordinate at the axial positions. The highly
symmetrical structure is one of the characteristics of this catalyst.
Catalytic Cycle. An assumed catalytic cycle of this Mannich-
type reaction is shown in Scheme 3. The key is assumed to be
isomerization of the zirconium catalyst from 5a to 5b when
aldimines coordinate the zirconium.23 As mentioned above,
highly symmetrical structure 5a was suggested from the NMR
experiments. On the other hand, the bidentate coordination of
aldimines to the zirconium was proved to be essential, since
almost no chiral induction was observed when the aldimines
derived from aniline or 2-methoxyaniline were used. For the
bidentate coordination, one of the four equatorial Zr-O bonds
had to flip to form an axial Zr-O bond. The zirconium catalyst
24-26
5b coordinates to aldimine 1 to form zirconium complex 6.
A silyl enol ether attacks the aldimine to produce 7 first, and
then 8, whose trimethylsilylated oxygen atom attacks the
zirconium center to afford the product (9 first and then 10) along
with regeneration of the catalyst 5b. The product was obtained
as a trimethylsilylated form without the acidic workup.
The catalytic cycle was supported by the following experi-
ments. We first performed silyl crossover experiments (Scheme
yldimethylsilyl enol ether of S-ethyl thioacetate (2d, 0.5 equiv).
The reaction was first monitored by GC/MS analysis; however,
it was revealed that the expected products (11a-d) were not
stable under the analysis conditions. We then decided to use
1
3
13
C NMR analysis. The C NMR spectra of 11a-d and that
of the reaction mixture revealed that four products (11a-d) were
observed in the mixture. In addition, it was confirmed that no
silyl scrambling in the starting materials and the products
occurred under the reaction conditions. These results indicated
that intermolecular silyl scrambling occurred during the Man-
nich-type reaction process and that an intermediate such as 7
would exist during the reaction course. Since species such as 7
are expected to catalyze achiral Mannich-type reactions28 and
2
7
4). In the presence of chiral zirconium catalyst 5 (10 mol %),
aldimine 1a (1.0 equiv) was treated with the trimethylsilyl enol
ether of S-tert-butyl thioacetate (2c, 0.5 equiv) and the eth-
(
(
23) A direct pathway from 5a to 6 may be possible.
24) We assume that this flipping leads to two isomeric structures. See
also: Ishitani, H.; Kitazawa, T.; Kobayashi, S. Tetrahedron Lett. 1999, 40,
161.
25) We performed NMR studies concerning the mixture of the zirconium
2
(
1
13
catalyst and an aldimine. H and C NMR spectra of the mixture were
almost the same as those of 5a at both room temperature and -30 °C. We
assume that coordination of NMI to the zirconium is stronger than that of
the aldimine to the zirconium
(27) For silyl crossover experiments in aldol reactions, see: (a) Mikami,
K.; Matsukawa, S. J. Am. Chem. Soc. 1994, 116, 4077. (b) Carreira, E. M.;
Singer, R. A. Tetrahedron Lett. 1994, 35, 4323. (c) Denmark, S. E.; Chen,
C.-T. Tetrahedron Lett. 1994, 35, 4327. (d) Evans, D. A.; Kozlowski, M.
C.; Murry, J. A.; Burgey, C. S.; Campos, K. R.; Connell, B. T.; Staples, R.
J. J. Am. Chem. Soc. 1999, 121, 669.
(26) For the structure of 6, another form where the imino phenol
deprotonates and one of the BINOL ligands protonates may be possible.

8184 J. Am. Chem. Soc., Vol. 122, No. 34, 2000
Ishitani et al.
Chart 2. 13C NMR Chemical Shifts (ppm)
Scheme 5. Preparation of 6,6′-CF3BINOL (4e)a
high enantioselectivity is obtained in the present zirconium-
catalyzed asymmetric Mannich-type reactions, the life of 7 is
assumed to be very short. It is noted that the silyl scrambling
occurred during such short period. For isolation, 11a-d were
converted to â-amino thioesters 3d and 12 by acidic workup.
In addition, formation of N-silylated adduct 9 was also
confirmed by C NMR analysis. While formation of 9 was not
detected in the reaction of 1a with 2b under the standard reaction
conditions, the N-triethylsilylated adduct was observed by NMR
analyses when the Mannich-type reaction of 1a with 1-ethylthio-
a
Conditions: (a) NaH, MOMCl/THF (98%). (b) BuLi, I
/NMP. (d) HCl/MeOH.
2
/THF-
Et
2
O. (c) CuCF
3
1
3
Table 4. Catalytic Enantioselective Mannich-Type Reactions
Using Chiral Ligand 4e (Eq 2)a
silyl
catalyst/
entry
imine; R1
Ph (1a)
p-Cl-Ph (1b)
Ph (1a)
enolate mol % product yield/% ee/%
1
-triethylsiloxyethene was performed under the same reaction
conditions. That is, two signals at 6.76 and 6.98 ppm were
1
2
3
4
5
6
2a
2a
2b
2b
2b
2b
2
2
2
0.5
2
3a
3b
3d
3d
3e
3i
quant
quant
quant
78
97
65
87
83
92
96
84
83
observed in the 13C NMR spectra of the Mannich-type adduct
prepared from 1a and 1-ethylthio-1-triethylsiloxyethene using
the chiral zirconium catalyst. The signals correspond to those
of the N-triethylsilylated adduct (Chart 2). These results indicate
that 9 is an intermediate from 8 to 10 and that the rate of the
silyl migration from 9 to 10 is influenced by steric bulkyness
of the silyl groups.
Ph (1a)
p-Cl-Ph (1b)
(1f)b
i-C H
4 9
5
a
b
[
Zr]/4e/additive ) 1/2.2/1.2-78 °C, 16 h. The aldimine prepared
from isovaleraldehyde and 2-amino-m-cresol was used.
As for the sense of enantioselection, we postulated stereo-
chemical model 13 shown in Figure 1. In 13, the re face was
shielded by one of the naphthyl rings, and silyl enolates were
expected to attack to the si face of aldimines.29
also turnover of the catalyst. It was assumed that Lewis acidity
of the zirconium was increased by introducing the electron-
withdrawing groups at the 6,6′-positions of BINOLs. As one
of the strongest electron-withdrawing groups, we chose the
trifluoromethyl group, and a novel BINOL derivative, (R)-6,6′-
bis(trifluoromethyl)-1,1′-bi-2-naphthol ((R)-6,6′-(CF3)2BINOL,
4
e), was designed. Synthesis of 6,6′-(CF3)2BINOL (4e) was
performed according to Scheme 5. 6,6′-Br2BINOL (3b) was
converted to its methoxymethyl (MOM) ether (14), whose
bromine groups at the 6,6′-positions were converted first to iodo
groups (15) using I2 and then to trifluoromethyl groups (16)
using CuCF3 in N-methylpyrrolidin-2-one (NMP).30 After
deprotection of the MOM groups, 6,6′-(CF3)2BINOL (4e) was
isolated as white crystals.
The reaction of 1a with 2b was performed using a zirconium
catalyst prepared from Zr(O Bu)4, 6,6′-(CF3)2BINOL (4e), and
Figure 1. Assumed stereochemical model.
t
NMI. It was revealed that in the presence of 2 mol % of the
chiral zirconium catalyst, 1a reacted with 2b in dichloromethane
at -78 °C to afford the corresponding Mannich-type adduct
quantitatively in 92% ee. It was exciting to note that the reaction
also proceeded smoothly in the presence of 0.5 mol % of the
catalyst and the enantiomeric excess of the adduct was 96%.
Other substrates were tested, and the results are summarized in
Table 4. In the reactions of aldimines derived from aromatic
aldehydes, the desired Mannich-type reactions proceeded cleanly
to afford the corresponding adducts in high yields with high
enantiomeric excesses in the presence of 2 mol % of the chiral
zirconium catalyst. Also in the reaction of the aldimine derived
from an aliphatic aldehyde, the yield and enantiomeric excess
were improved even though 5 mol % of the catalyst was loaded.
Conclusion. A novel chiral zirconium catalyst has been
developed and successfully used in enantioselective Mannich-
Turnover Improvement. While highly selective catalytic
asymmeric Mannich-type reactions have been developed, turn-
over of the Zr catalyst was less than 20 in most cases. When 1
mol % of Zr catalyst 5 was used in the reaction of aldimine 1c
with silyl enolate 2a, lower yield and selectivity were obtained.
To improve the low turnover of the catalyst, we searched for
more efficient ligands in this reaction. In the previous investiga-
tions, introduction of two halogen atoms at the 6,6′-positions
of BINOLs was effective to improve not only selectivities but
(
28) (a) Colvin, E. W.; McGarry, D.; Nugent, M. J. Tetrahedron 1988,
4, 4157. A similar species is known to catalyze aldol reactions of silyl
enol ethers with aldehydes. (b) Kobayashi, S.; Mukaiyama, T. Chem. Lett.
989, 297. (c) Kobayashi, S.; Uchiro, H.; Fujishita, Y.; Shiina, I.;
Mukaiyama, T. J. Am. Chem. Soc. 1991, 113, 4247. See also ref 27b.
29) Preliminary experiments suggested that the reaction proceeded via
4
1
(
acyclic transition states. When 1c was treated with (Z)-1-ethyldimethylsi-
loxy-1-methoxy-1-propene in the presence of the zirconium catalyst (10
mol %) in dichloromethane at -78 °C, the desired anti adduct was
predominantly obtained in good enantiomeric excess (syn/anti ) 20/80,
(30) (a) Carr, G. E.; Chambers, R. D.; Holmes, T. F. J. Chem. Soc.,
Perkin Trans. 1. 1988, 921. See also: (b) Burton, D. J.; Wiemers, D. M. J.
Am. Chem. Soc. 1985, 107, 5014. (c) Wiemers, D. M.; Burton, D. J. J. Am.
Chem. Soc. 1986, 108, 832.
7
8% ee (anti)). On the other hand, the same anti adduct was obtained
preferentially (syn/anti ) 16/84, 52% ee (anti) when (E)-1-ethyldimethyl-
siloxy-1-methoxy-1-propene was used under the same reaction conditions.

Synthesis of Optically ActiVe â-Amino Acid DeriVatiVes
J. Am. Chem. Soc., Vol. 122, No. 34, 2000 8185
dried over Na SO . Pure 14 was obtained after recrystallization from
type reactions of silyl enol ethers with aldimines. The catalyst
2
4
t
CH Cl -hexane almost quantitatively (95%): Mp 135-136 °C. IR
was prepared by mixing Zr(O Bu)4, a BINOL derivative, and
2
2
-
1 1
(
(
KBr) 2961, 2899, 1589, 1498, 1244, 1050, 1062, 1024 cm . H NMR
CDCl ) δ 3.16 (s, 6H), 4.98 (d, 2H, J ) 6.8 Hz), 5.09 (d, 2H, J ) 6.8
NMI, and its symmetrical structure was revealed by NMR
analyses. While it has been known that catalytic activation of
aldimines is difficult compared to that of aldehydes, the use of
the zirconium complex was shown to be the key to completing
the catalytic cycle of the Mannich-type reactions. In addition,
a novel BINOL derivative, (R)-(CF3)2BINOL, was prepared, and
a high level of turnover of this Mannich-type reaction has been
achieved. This new ligand will be applied to other BINOL-
based chiral catalysts to enable versatile catalytic asymmetric
synthesis.
3
Hz), 6.98 (d, 2H, J ) 9.0 Hz), 7.29 (dd, 2H, J ) 2.2, 9.1 Hz), 7.60 (d,
2
H, J ) 9.0 Hz), 7.87 (d, 2H, J ) 9.0 Hz), 8.04 (d, 2H, J ) 1.8 Hz).
13
C NMR (CDCl ) δ 55.9, 95.0, 118.0, 118.0, 120.7, 127.1, 128.7, 129.7,
3
1
2 4
29.9, 130.8, 132.4, 152.9. Anal. Calcd for C24H20Br O : C, 54.16; H,
3
.79. found: C, 54.07; H, 3.88.
(
R)-6,6′-Diiodo-2,2-di(methoxymethyl)oxy-1,1′-binaphthyl (15):
31
To a solution of (R)-6,6′-dibromo-2,2′-di(methoxymethyl)oxy-1,1′-
binaphthyl (14, 8.0 g, 15 mmol) in THF (80 mL) was added a hexane
solution of n-BuLi (1.6 M; 28 mL, 45 mmol) at -78 °C under argon.
The reaction mixture was stirred for 30 min, and then a solution of
iodine (11.4 g) in THF (10 mL) was added. The reaction mixture was
allowed to warm to room temperature over 12 h and quenched with
Experimental Section
Typical Experimental Procedure for Asymmetric Reaction Using
a Chiral Zirconium Catalyst. A typical experimental procedure is
water. The resulting mixture was treated with aqueous 10% NaHSO
to destroy excess iodine. After being stirred for 1 h, the organic layer
was washed with saturated aqueous NaHCO , water, and brine and dried
over Na SO . After evaporation of the solvents, the residue was purified
3
described for the reaction of aldimine 1c with ketene silyl acetal 2a.
3
t
To Zr(O Bu)
4
(0.04 mmol) in dichloromethane (0.25 mL) was added
2
4
6
,6′-dibromo-1,1′-bi-2-naphthol (4b, 0.088 mmol) in dichloromethane
by silica gel column chromatography (hexane/AcOEt ) 9/1) and
(
(
0.5 mL) and N-methylimidazole (0.048 mmol) in dichloromethane
0.25 mL) at room temperature. The mixture was stirred for 1 h at the
recrystallization from CH
2
Cl
2
-hexane to afford 15 (7.4 g (79%)): Mp
-1
1
23-124 °C. IR (KBr) 2954, 2897, 1577, 1493, 1241, 1149, 1022 cm .
same temperature and cooled to -45 °C. Dichloromethane solutions
0.75 mL) of 1c (0.8 mmol) and 2a (0.96 mmol) were successively
added. The mixture was stirred for 10 h, and saturated NaHCO was
1
H NMR (CDCl
H, J ) 6.8 Hz), 6.84 (d, 2H, J ) 8.8 Hz), 7.45 (dd, 2H, J ) 1.8, 8.8
Hz), 7.58 (d, 2H, J ) 9.0 Hz), 7.84 (d, 2H, J ) 9.2 Hz), 8.27 (d, 2H,
3
) δ 3.17 (s, 6H), 4.99 (d, 2H, J ) 6.8 Hz), 5.09 (d,
(
2
3
added to quench the reaction. The aqueous layer was extracted with
13
J ) 1.8 Hz). C NMR (CDCl
28.6, 131.5, 131.5, 134.9, 136.6, 153.1. Anal. Calcd for C24
C, 46.03; H, 3.44. Found: C, 46.19; H, 3.45.
R)-6,6′-Bis(trifluoromethyl)-2,2-dihydroxy-1,1′-binaphthyl (6-
CF BINOL, 4e). A mixture of sodium trifluoroacetate (435 mg, 3.2
3
) δ 56.0, 95.0, 117.8, 120.6, 127.1, 127.1,
dichloromethane, and the crude adduct was treated with THF-1 N HCl
1
20 2 4
H I O :
(10:1) at 0 °C for 30 min. Water was added, and the resulting mixture
was neutralized. The aqueous layer was extracted with ethyl acetate,
and the combined organic layer was dried. After filtration and
evaporation under reduced pressure, the crude product was chromato-
graphed on silica gel to give the desired adduct 3c in a quantitative
yield. The optical purity was determined by HPLC analysis using a
chiral column (see below). Several products were fully characterized,
and the optical purities were determined after methylation of phenolic
OH group.
(
(
3 2
)
mmol), copper(I) iodide (609 mg, 3.2 mmol), 15 (250 mg, 0.4 mmol)),
and N-methylpyrrolidin-2-one (8.0 mL) was stirred under argon at 160-
1
80 °C for 8 h. After cooling, the reaction mixture was diluted with
AcOEt and water and filtered on a Celite pad. The aqueous layer was
extracted with AcOEt, and the combined organic layer was washed
with water and brine and dried over Na
solvents, the residue was purified by flash column chromatography on
silica gel (hexane/CH Cl ) 1/1) and the product was dissolved in CH
Cl (1.0 mL). To this solution was added saturated HCl methanolic
solution (1.0 mL) at 0 °C for 30 min. The resulting solution was diluted
with water and extracted with CH Cl . The combined organic layer
was washed with water and aqueous saturated NaHCO and dried over
Na SO . After evaporation of the solvents, the residue was purified by
flash column chromatography on silica gel (hexane/Et O ) 2/1) to
). Mp 111-
2 4
SO . After evaporation of the
(
R)-Methyl 2,2′-dimethyl-3-(2-hydroxyphenyl)amino-3-phenyl-
2
4
propionate (3a): [R]
D
+1.4 (c 1.15, CHCl ) (87% ee). Mp 112.5-
3
2
2
2
-
1
14 °C. IR (KBr) 3401, 1709, 1611, 1514, 1453, 1391 cm-1. 1H NMR
2
(
CDCl ) δ 1.21 (s, 3H), 1.24 (s, 3H), 3.68 (s, 3H), 4.57 (s, 1H), 6.36-
3
13
6
4
1
.76 (m, 4H), 7.21-7.28 (m, 5H). C NMR (CDCl ) δ 19.9, 24.2,
3
2
2
7.3, 52.3, 64.3, 113.2, 114.1, 117.6, 120.8, 127.3, 127.9, 128.3, 135.6,
3
i
38.9, 144.0, 178.0. HPLC Daicel Chiralpak AD, hexane/ PrOH ) 9/1,
2
4
flow rate 1.0 mL/min: t
Calcd for C18
R
) 9.3 min (3R), t ) 16.0 min (3S). Anal.
R
2
H21NO
3
: C, 72.22; H, 7.07; N, 4.68. Found: C, 72.28;
22
afford 4e (140 mg (83%)): [R]
12 °C. IR (KBr) 3465, 3038, 1633, 1311, 1198, 1152 cm . H NMR
CDCl ) δ 5.14 (s, 2H), 7.19 (d, 2H, J ) 8.8 Hz), 7.49 (dd, 2H, J )
D
-36.1 (c 1.10, CHCl
3
+
H, 7.20; N, 4.62. HRMS Calcd for C18
99.1497.
Removal of the N-Protecting Group. 3a (0.4 mmol), a CH
acetone (1:5) solution (5 mL), and K CO (299 mg) were combined at
room temperature. After the mixture was stirred for 8 h, NH Cl (aq)
H21NO
3
(M ) 299.1522, found
-1 1
1
(
2
3
3
I-
1.8, 8.8 Hz), 7.50 (d, 2H, J ) 9.0 Hz), 8.11 (d, 2H, J ) 9.0 Hz), 8.22
(s, 2H). 13C NMR (CDCl ) δ 110.5, 119.3, 123.4 (q, J ) 3.1 Hz), 125.0,
2
3
3
126.1, 126.3 (q, J ) 4.4 Hz), 126.5, 128.3, 132.7, 134.9, 154.5. 19
4
F
was added to quench the reaction. After the usual workup, methyl 3-[(2′-
methoxyphenyl)amino]-2,2-dimethyl-3-phenylpropionate was obtained
quantitatively. Oxidative cleavage using CAN was performed according
NMR (CDCl ) δ -142.4 (s, relative to CF COOH). HPLC, Daicel
3
3
i
Chiralpak AD, hexane/
PrOH ) 9/1, flow rate 1.0 mL/min: t ) 10.2
R
+
min (R) tR ) 25.9 min (S). HRMS, Calcd for C H F O (M
22
12
6
2
)
1
9
to the literature method. The absolute configuration assignment was
made by comparison of the optical rotation of N-free amino ester with
422.0741, found 422.0740.
NMR Experiments of the Chiral Zirconium Catalyst. To a C
D
6 6
2
0
that in the literature.
R)-Methyl 3-amino-2,2-dimethyl-3-phenylpropionate: H NMR
CDCl ) δ 1.20 (s, 3H), 1.25 (s, 3H), 3.65 (s, 3H), 3.87 (br s, 1H),
(
0.5 mL) solution of 6-BrBINOL (0.24 mmol) and N-methylimidazole
1
(
t
6 6 4
(0.24 mmol) was added a C D (0.5 mL) solution of Zr(O Bu) (0.12
(
7
1
3
mmol) at room temperature. The resulting solution was stirred for 1 h
at this temperature, then evaporated and dried under reduced pressure
at 50 °C for 3 h. The sample for the NMR experiment was prepared
1
3
.23-7.35 (m, 5H). C NMR (CDCl
27.7, 127.9, 128.3, 176.9. (Hydrochloride) [R]
3
) δ 20.6, 24.1, 52.0, 55.6, 127.3,
2
6
D
+34.6 (c 0.17, 1 N
2
0
23
HCl) (lit. [R]
R)-6,6′-Dibromo-2,2′-di(methoxymethyl)oxy-1,1′-binaphthyl (14).
To a suspension of sodium hydride (60%; 6.8 g, 170 mmol) in THF
D
-32.8 (c 1.1, 1 N HCl)).
1
13
2
using CD Cl
2
as a solvent. H and C NMR (CD
2
Cl
2
) data are shown
(
in Supporting Information.
Silicon Crossover Experiment. To a CH Cl solution (0.5 mL) of
2 2
(100 mL) was added (R)-6,6′-dibromo-2,2′-dihydroxy-1,1′-binaphthyl
the chiral zirconium catalyst (0.033 mmol), which was prepared as
described above, aldimine 1a in CH Cl (0.5 mL) and silyl enol ether
c and 2d in CH Cl (0.5 mL) were successively added at -45 °C.
After 8 h, saturated aqueous NaHCO was added to quench the reaction.
(15.1 g, 34 mmol) in THF (200 mL). The resulting solution was stirred
2
2
at room temperature for 30 min, and then chloromethyl methyl ether
6.5 mL, 85 mmol) in THF (10 mL) was added. The mixture was stirred
for 2 h and was carefully quenched with MeOH and water. The aqueous
layer was extracted with Et O, and the combined organic layer was
washed with water, saturated aqueous NaHCO , and brine and then
2
2
2
(
3
2
(31) Ireland, R. E.; Mueller, R. H.; Willard, A. K. J. Am. Chem. Soc.
1976, 98, 2868.
3

8186 J. Am. Chem. Soc., Vol. 122, No. 34, 2000
Ishitani et al.
The organic layer was separated, and the aqueous layer was extracted
Education, Science, Sports, and Culture, Japan. M.U. thanks
the JSPS fellowship for Japanese Junior Scientists.
with CH Cl and dried over Na SO . After concentration, the crude
product was dissolved in CD Cl and used for the NMR experiment.
2
2
2
4
32,33
2
2
tHE usual workup after treatment of 1 N HCl in THF gave the
corresponding â-amino ester. 3d: 27%, 73% ee; 12: 28%, 64% ee.
Supporting Information Available: Experimental details
1
13
and H and C NMR spectra of the catalyst and the crossover
experiments and products (PDF). This material is available free
of charge via the Internet at hppt://pubs.acs.org.
Acknowledgment. This work was partially supported by a
Grant-in-Aid for Scientific Research from the Ministry of
(
32) Cox, P. J.; Wang, W.; Snieckus, V. Tetrahedron Lett. 1992, 33,
253.
33) The NMR spectra are shown in Supporting Information.
2
(
JA001642P