Catalytic asymmetric cyano-ethoxycarbonylation reaction of aldehydes using a YLi3tris(binaphthoxide) (YLB) complex: Mechanism and roles of achiral additives

Article abstract of DOI:10.1021/ja042887v

Full details of a catalytic asymmetric cyano-ethoxycarbonylation reaction promoted by a heterobimetallic YLi₃tris(binaphthoxide) complex (YLB 1), especially mechanistic studies, are described. In the cyanation reaction of aldehydes with ethyl c

Full text of DOI:10.1021/ja042887v

Published on Web 02/18/2005
Catalytic Asymmetric Cyano-Ethoxycarbonylation Reaction of
Aldehydes using a YLi3Tris(binaphthoxide) (YLB) Complex:
Mechanism and Roles of Achiral Additives
Noriyuki Yamagiwa, Jun Tian, Shigeki Matsunaga,* and Masakatsu Shibasaki*
Contribution from the Graduate School of Pharmaceutical Sciences, The UniVersity of Tokyo,
Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
Received November 25, 2004; E-mail: mshibasa@mol.f.u-tokyo.ac.jp
Abstract: Full details of a catalytic asymmetric cyano-ethoxycarbonylation reaction promoted by a
heterobimetallic YLi3tris(binaphthoxide) complex (YLB 1), especially mechanistic studies, are described.
In the cyanation reaction of aldehydes with ethyl cyanoformate, three achiral additives, H2O, tris(2,6-
dimethoxyphenyl)phosphine oxide (3a), and BuLi, were required to achieve high reactivity and enantio-
selectivity (up to >99% yield and up to 98% ee). The roles of achiral additives and the reaction pathway
were investigated in detail. In situ IR analysis revealed that the initiation step to generate LiCN from H2O,
BuLi, and ethyl cyanoformate is rather slow. On the basis of mechanistic studies of the initiation step to
generate an active nucleophilic species, reaction conditions were optimized by using a catalytic amount of
acetone cyanohydrin as an initiator. Under the optimized conditions, the induction period decreased and
the reaction completed within 9 min using 5 mol % YLB at -78 °C. Catalyst loading was successfully
reduced to 1 mol %. Kinetic experiments and evaluation of the substituent effects of phosphine oxide
revealed that phosphine oxide had beneficial effects on both the reaction rate and the enantioselectivity.
The putative active species as well as the catalytic cycle of the reaction are also discussed.
Introduction
ate or O-phosphate. Although excellent enantioselectivity and
chemical yield were achieved in those catalytic asymmetric
The catalytic asymmetric cyanation reaction of carbonyl
compounds is one of the most powerful tools available to supply
useful chiral building blocks. Various catalyst systems have been
developed over the last two decades;1 (CH3)3SiCN and/or HCN
are most often used as cyanide sources to afford cyanohydrins
and their TMS ethers. The intrinsic instability of cyanohydrins
and their TMS ethers, however, is sometimes problematic for
further transformations. To avoid the problems, one-pot catalytic
asymmetric cyanation-O-protection reaction with a robust
protecting group was investigated recently by Deng,2 Belokon
and North,3,4 our group,5 and Na´jera and Saa´.6 Either ethyl
cyanoformate or diethyl cyanophosphonate was used as a
cyanide source, affording corresponding cyanohydrin O-carbon-
reactions,2-6 there is only limited information on the reaction
mechanisms based on detailed studies of asymmetric catalysis.4
Understanding the reaction mechanisms of the new variant in
asymmetric cyanation reactions, including the generation of
active species and catalytic cycle, is necessary for further rational
improvement of the reaction. In our preliminary report,5a
a
heterobimetallic YLi3tris(binaphthoxide) complex7 (YLB 1,
Figure 1) effectively promoted the reaction with the aid of three
achiral additives: H2O, tris(2,6-dimethoxyphenyl)phosphine
oxide (3a), and BuLi. Cyanohydrin O-carbonates were obtained
in high yield (up to >99% yield) and enantiomeric excess (up
to 98% ee) (Scheme 1). In most asymmetric reactions using
rare earth-alkali metal heterobimetallic complexes reported by
our group, the Lewis acid-Brønsted base cooperative function
(1) For recent reviews on asymmetric cyanation reaction, see: (a) Brunel, J.
M.; Holmes, I. P. Angew. Chem., Int. Ed. 2004, 43, 2752. (b) North, M.
Tetrahedron: Asymmetry 2003, 14, 147. (c) Gro¨ger, H. Chem.sEur. J.
2001, 7, 5246. (d) Gregory, R. J. H. Chem. ReV. 1999, 99, 3949.
(2) Tian, S.-K.; Deng, L. J. Am. Chem. Soc. 2001, 123, 6195.
(3) (a) Belokon, Y. N.; Blacker, A. J.; Clutterbuck, L. A.; North, M. Org.
Lett. 2003, 5, 4505. For related reactions, see also: (b) Belokon, Y. N.;
Gutnov, A. V.; Moskalenko, M. A.; Yashkina, L. V.; Lesovoy, D. E.;
Ikonnikov, N. S.; Larichev, V. S.; North, M. Chem. Commun. 2002, 244.
(c) Belokon, Y. N.; Carta, P.; Gutnov, A. V.; Maleev, V.; Moskalenko, M.
A.; Yashkina, L. V.; Ikonnikov, N. S.; Voskoboev, N. V.; Khrustalev, V.
N.; North, M. HelV. Chim. Acta 2002, 85, 3301. Same Ti-salen catalyst
was applicable in asymmetric cyanation reactions with (CH3)3SiCN: (d)
Belokon, Y. N.; Caveda-Cepas, S.; Green, B.; Ikonnikov, N. S.; Khrustalev,
V. N.; Larichev, V. S.; Moskalenko, M. A.; North, M.; Orizu, C.; Tararov,
V. I.; Tasinazzo, M.; Timofeeva, G. I.; Yashkina, L. V. J. Am. Chem. Soc.
1999, 121, 3968 and references therein.
(5) (a) Tian, J.; Yamagiwa, N.; Matsunaga, S.; Shibasaki, M. Angew. Chem.,
Int. Ed. 2002, 41, 3636. (b) Tian, J.; Yamagiwa, N.; Matsunaga, S.;
Shibasaki, M. Org. Lett. 2003, 5, 3021. With diethyl cyanophosphonate as
a cyanide source: (c) Abiko, Y.; Yamagiwa, N.; Sugita, M.; Tian, J.;
Matsunaga, S.; Shibasaki, M. Synlett 2004, 2434.
(6) (a) Casas, J.; Baeza, A.; Sansano, J. M.; Na´jera, C.; Saa´, J. M.
Tetrahedron: Asymmetry 2003, 14, 197. With diethyl cyanophosphonate
as a cyanide source: (b) Baeza, A.; Casas, J.; Na´jera, C.; Sansano, J. M.;
Saa´, J. M. Angew. Chem., Int. Ed. 2003, 42, 3143. Same Al catalyst was
applicable in asymmetric cyanation reactions with (CH3)3SiCN: (c) Casas,
J.; Na´jera, C.; Sansano, J. M.; Saa´, J. M. Org. Lett. 2002, 4, 2589.
(7) For reviews of asymmetric catalysis using rare earth-alkali metal hetero-
bimetallic complexes, see: (a) Shibasaki, M.; Yoshikawa, N. Chem. ReV.
2002, 102, 2187. (b) Shibasaki, M.; Sasai, H.; Arai, T. Angew. Chem., Int.
Ed. Engl. 1997, 36, 1236. For the preparation of YLB 1 complex and X-ray
structure, see: (c) Aspinall, H. C.; Dwyer, J. L. M.; Greeves, N.; Steiner,
A. Organometallics 1999, 18, 1366. (d) Aspinall, H. C.; Bickley, J. F. B.;
Dwyer, L. M.; Greeves, N.; Kelly, R. V.; Steiner, A. Organometallics 2000,
19, 5416.
(4) During preparation of this manuscript, North and Belokon reported
mechanistic studies concerning the structure of chiral Ti-salen catalyst in
solution using various cyanide sources: Belokon, Y. N.; Blacker, A. J.;
Carta, P.; Clutterbuck, L. A.; North, M. Tetrahedron 2004, 60, 10433.
9
10.1021/ja042887v CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 3413-3422
3413

A R T I C L E S
Yamagiwa et al.
Table 1. Optimization of Reaction Conditions with Achiral
Additives
additives
Ar3P
(mol %)
H2O
d
O
BuLi
time
(h)
Figure 1. Structures of YLi3tris(binaphthoxide) (YLB 1), ethyl cyano-
formate (2), and tris(2,6-dimethoxyphenyl)phosphine oxide 3a.
entry
(mol %)
(mol %)
T(
°
C)
% yield
% ee
1
2
3
4
5
6
7
8
9
10
0
0
10
20
30
30
30
30
0
0
0
0
0
0
0
0
0
0
10
-60
-60
-60
-60
-60
-60
-60
-78
-78
9
84
0
58
ND
64
83
88
89
92
90
94
7.5
8.5
2.5
2.5
2.5
1.5
2
Scheme 1. Catalytic Asymmetric Cyano-Ethoxycarbonylation
Reaction Promoted by a Mixture of YLB 1, H2O, BuLi, and Ar3P(O)
3a
3b 10
3b 10
3b 10
3b 10
3a 10
3a 10
3a 10
93
88
74
54
97
76
96
2
Results and Discussion
A. Development of a Catalytic Asymmetric Cyano-
Ethoxycarbonylation Reaction. Initial screening of both
cyanide and metal sources revealed that commercially available
ethyl cyanoformate 2 and (S)-YLB 1 prepared from Y(HMDS)3,
(S)-BINOL, and BuLi were promising candidates.11 With YLB
1 (10 mol %), H2O (10 mol %), and 2, cyanation of aldehyde
4a proceeded at -60 °C to afford 5a in 84% yield and 58% ee
(Table 1, entry 1). Interestingly, the reaction did not proceed
without the addition of H2O (entry 2). To fine tune the chiral
environment, various achiral additives were surveyed and
triarylphosphine oxide, Ar3P(O) 3, was determined to be
optimal.12 Addition of both H2O and Ph3P(O) 3b (10 mol %)
significantly improved both the reaction rate and the enanti-
oselectivity (entry 4). The highest selectivity was obtained by
adding 30 mol % H2O (entry 6). Ar3P(O) 3 was optimized, and
tris(2,6-dimethoxyphenyl)phosphine oxide 3a13 had the highest
reactivity and selectivity (entry 7). Moreover, by adding 10 mol
% BuLi, the reaction proceeded smoothly at -78 °C to afford
5a in 94% ee (entry 9). In the absence of BuLi, the reaction
rate was not sufficient at -78 °C (entry 8). The optimized
reaction conditions for cyanation, YLB 1 (10 mol %), H2O (30
mol %), BuLi (10 mol %), Ar3P(O) 3a (10 mol %), and 2 (1.2
equiv), were applicable to various aldehydes (Table 2). In all
cases, good yield and enantiomeric excess (87-98% ee) were
achieved using aromatic aldehydes (entries 1 and 2), R,â-
unsaturated aldehydes (entries 3-5), R-unsubstituted (entries 6
and 7), R-monosubstituted (entries 8 and 9), and R,R-disubsti-
tuted aliphatic aldehydes (entry 10). The short reaction time
(<3 h) for all aldehydes as well as broad substrate generality is
noteworthy.
had a key role in promoting the reactions.7a,b The active proton
of the nucleophiles is deprotonated by a Brønsted basic moiety
of the complex to form metalated nucleophiles. In the present
cyanation reaction, however, there is no active proton in the
cyanide source. The reaction mechanism, especially generation
of the active nucleophilic species, should be different from that
in previous reactions with heterobimetallic complexes. More-
over, in striking contrast to a catalytic asymmetric 1,4-addition
reaction of methoxylamine promoted by the same YLB 1
complex under strictly anhydrous conditions using desiccant,8
the asymmetric cyano-ethoxycarbonylation reaction required
H2O to achieve both high reactivity and enantioselectivity. In
our preliminary report,5a the roles of H2O and phosphine oxide
in the reaction were not clear. In a related catalytic asymmetric
cyanosilylation reaction, we reported the efficacy of Lewis
acid-Lewis base cooperative catalysts,9 which contained a
phosphine oxide unit in the chiral ligands as a Lewis basic
moiety to activate and control the orientation of the (CH3)3-
SiCN nucleophile. Interaction of phosphine oxide with the Si
of (CH3)3SiCN was proposed.9,10 Because a silyl reagent is not
involved in the asymmetric cyano-ethoxycarbonylation reaction,
the role of phosphine oxide should be different. In this article,
we report the full details of a catalytic asymmetric cyano-
ethoxycarbonylation reaction with YLB 1. Mechanistic studies
of the reaction are the main topic of this paper. The role of
achiral additives was investigated in detail. Both the initiation
step and the catalytic cycle of the reaction are discussed. On
the basis of information about the initiation step, reaction
conditions were improved by adding a catalytic amount of
acetone cyanohydrin as an initiator.
(11) Catalyst screening was performed without phosphine oxide and BuLi. Other
rare earth-alkali metal heterobimetallic complexes gave less satisfactory
results in terms of reaction rate and enantioselectivity. For example: Dy-
Li-BINOL, 4 h, y. quant, 45% ee; Yb-Li-BINOL, 4 h, y. quant, 31%
ee; Sc-Li-BINOL, 1 h, y. quant, 9% ee; Y-Na-BINOL, 0.5 h, y. quant,
0% ee; Y-K-BINOL, 5.5 h, y. quant, 8% ee.
(12) Screening of achiral additives was performed at -60 °C using aldehyde
4a and 10 mol % YLB 1. Other Lewis basic achiral additives gave less
satisfactory results. For example, (n-Bu)3P(O): 63% ee, (c-hexyl)3P(O):
69% ee, (t-Bu)3P(O): 63% ee, Ph3As(O): 78% ee, (PhO)3P(O): 70% ee,
NMO: 69% ee, Ph2P(O)CH2CH2P(O)Ph2: 82% ee, Ph3P: 71% ee.
(13) For the preparation of phosphine oxides, see: Kim, R. D.; Stevens, C. H.
Polyhedron 1994, 13, 727. For the procedure, see Supporting Information.
(8) (a) Yamagiwa, N.; Matsunaga, S.; Shibasaki, M. J. Am. Chem. Soc. 2003,
125, 16178. For discussion of the reaction mechanism using anhydrous
YLB as a catalyst, see: (b) Yamagiwa, N.; Matsunaga, S.; Shibasaki, M.
Angew. Chem., Int. Ed. 2004, 43, 4493.
(9) (a) Hamashima, Y.; Sawada, D.; Kanai, M.; Shibasaki, M. J. Am. Chem.
Soc. 1999, 121, 2641. For reviews, see also: (b) Shibasaki, M.; Kanai,
M.; Funabashi, K. Chem. Commun. 2002, 1989. (c) Shibasaki, M.; Kanai,
M. Chem. Pharm. Bull. 2001, 49, 511.
(10) For use of phosphine oxide to increase nucleophilicity of cyanide species
generated from TMSCN in the catalytic asymmetric cyanosilylation reaction
of aldehyde, see: Ryu, D. H.; Corey, E. J. J. Am. Chem. Soc. 2004, 126,
8106 and references therein.
9
3414 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005

Cyano-Ethoxycarbonylation Reaction of Aldehydes using YLB 1
A R T I C L E S
Table 2. Catalytic Asymmetric Cyano-Ethoxycarbonylation
Reaction of Various Aldehydes
entry
aldehyde (R)
product
time (h)
% yield
% ee
1
2
3
4
5
6
7
8
9
4a Ph
4b 1-naphthyl
4c (E)-CH3(CH2)2CHdCH
4d (E)-cyclo-C6H11-CHdCH
4e (E)-PhCHdCH
4f CH3(CH2)4
5a
5b
5c
5d
5e
5f
5g
5h
5i
2
2
3
3
3
3
2
2
2
3
96
97
>99
98
94
90
92
93
91
94
92
98
96
87
>99
93
4g CH3CH2
4h (CH3)2CH
79a
88a
97
4i
cyclo-C6H11
10
4j (CH3)3C
5j
93
a Product was volatile.
Under the optimized conditions, YLB required three achiral
additives, H2O, tris(2,6-dimethoxyphenyl)phosphine oxide 3a,
and BuLi, to achieve the best enantioselectivity and reaction
rate. The detailed catalytic cycle, mechanisms for the generation
of an active nucleophilic species, and the roles of the achiral
additives, however, were not clear. The following sections
describe mechanistic investigations to clarify these points.
B. Effects of H2O. Effects of the amount of H2O on reactivity
and enantioselectivity were examined using 10 mol % YLB 1,
10 mol % 3a, and 10 mol % BuLi in THF at -78 °C (Figure
2). Aldehyde 4h was used instead of 4a because enantioselec-
tivity with aldehyde 4h was more sensitive to the amount of
H2O compared with that of 4a. To clearly compare the difference
in enantioselectivity, both enantiomeric ratio (er: R-enantiomer/
S-enantiomer) and enantiomeric excess (ee: R-enantiomer -
S-enantiomer/R-enantiomer + S-enantiomer) are presented in
Figure 2. The enantiomeric ratio and chemical yield after 30
min were plotted against the amount of H2O (0-50 mol %). A
rather unexpected curve was observed for the enantiomeric ratio,
while the chemical yield was not affected (80-90% yield). To
explain the effects of H2O, the curve in Figure 2 was separated
into three sections (A, H2O, 0-10 mol %; B, H2O, 10-30 mol
%; and C, H2O, 30-50 mol %). In section A (H2O, 0-10 mol
%), the enantiomeric ratio was generally low, ranging from 2.5
to 11.8 (43-84% ee). The enantiomeric ratio increased gradually
depending on the amount of H2O in section B (H2O, 10-30
mol %), and the highest enantiomeric ratio was obtained with
20-30 mol % H2O. In section C (H2O, 30-50 mol %), too
much H2O had adverse effects on enantioselectivity. We
assumed that the effects of H2O were as follows. In section A,
all H2O was consumed by BuLi (10 mol %) to generate LiOH.
LiOH reacts with ethyl cyanoformate, and essentially no free
H2O remained in the reaction mixture. In sections B and C, an
excess amount of H2O remained in the reaction mixture and
coordinated with YLB, thereby affecting the chiral environment.
NMR spectroscopy revealed a reversible interaction between
YLB and H2O in solution. As shown in Figure 3a, NMR spectra
of anhydrous YLB in THF at -20 °C showed six sharp signals,
implying that the YLB complex has C3-symmetry in solution.
The signals gradually broadened, however, by adding H2O to
Figure 2. Water effects on enantioselectivity and chemical yield in the
catalytic asymmetric cyano-ethoxycarbonylation reaction of aldehyde 4h.
YLB (Figure 3b,c).14 In Figure 3d, Drierite (CaSO4), a neutral
desiccant, was added to the mixture of YLB and 3 equiv of
H2O. The sharp signals of YLB 1 were observed again. These
observations suggested that there is rapid equilibrium between
anhydrous YLB and the YLB-H2O complex (Figure 4). On
the other hand, Aspinall et al. reported the X-ray crystallographic
structure of both [Li(Et2O)]3[Eu(binol)3] (anhydrous EuLB) and
[Li(Et2O)]3[Eu(binol)3(H2O)] (aqueous EuLB).7c,d The anhy-
drous EuLB had nearly D3-symmetry, while aqueous EuLB had
bent C3-symmetry due to coordination of H2O with the europium
center. Aspinall et al. insisted that H2O would be effective for
finely tuning the chiral environment and enantioselectivity of
the catalyst. We assumed H2O would coordinate with the yttrium
center in solution and have effects similar to those observed
with the europium complex. Considering the coordination
number (CN ) 6 or 7) of rare earth-alkali metal heterobime-
tallic complexes observed in crystal structures,15 only one H2O
molecule would coordinate with the yttrium center.16 We
speculated that a slightly excess amount of H2O to YLB has
beneficial effects to make the equilibrium between YLB-H2O
(14) Free H2O in THF was not observed by 1H NMR, probably due to rapid
equilibrium between anhydrous YLB and the YLB-H2O complex.
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3415

A R T I C L E S
Yamagiwa et al.
Table 3. Base Effects on Catalytic Asymmetric
Cyano-Ethoxycarbonylation Reaction
entry
base
time (h)
% yield
% ee
1
2
3
4
5
6
7
none
none
BuLi
BuLi
LiHMDS
NaHMDS
KHMDS
0.5
2
0.5
2
2
2
17
76
95
96
94
92
69
94
90
96
94
93
90
83
2
Scheme 2. Catalytic Asymmetric Cyano-Ethoxycarbonylation
Reaction using LiCN as an Additive
Figure 3. NMR spectra of (S)-YLB (60 mM in d8-tetrahydrofuran) at -20
°C (a) without H2O, (b) with 60 mM H2O, (c) with 180 mM H2O, (d) with
180 mM H2O, and Drierite (CaSO4: 200 mg/mL of d8-tetrahydrofuran).
and potassium, as shown in entries 4-7, implying that lithium
metal was integrated into the active chiral species.
We hypothesized that the LiCN generated from LiOH and
ethyl cyanoformate was a key nucleophilic species. In the
racemic reaction using ethyl cyanoformate as a cyanide source,
the cyanide anion itself was proposed to be the key intermedi-
ate.17 In the catalytic asymmetric reaction, generated LiCN
should self-assemble with the chiral catalyst, otherwise a racemic
reaction would proceed. To examine our hypothesis, LiCN
prepared by a known procedure18 was added to the mixture of
YLB, H2O, and phosphine oxide 3a in the absence of BuLi. As
expected, the reaction proceeded smoothly, and the product was
obtained in 93% ee and quantitative yield (Scheme 2).19 The
similar enantioselectivity suggested that the same active species
was generated using LiCN instead of BuLi. On the other hand,
there was a difference in the reaction profile when reactions
(a) without BuLi, (b) with BuLi, and (c) with LiCN instead of
BuLi were monitored using in situ IR (Figure 5a-c). Aldehyde
4a was added to a mixture of YLB, H2O, Ar3PdO 3a, and
Figure 4. Supposed reversible interaction between YLB 1 and H2O in
solution.
and anhydrous YLB more favorable to the YLB-H2O complex;
30 mol % H2O gave the best result (Figure 4). In section C, too
much H2O slightly decreased enantioselectivity (with 50 mol
% H2O, er ) 36.9 and 94.7% ee), probably because too much
H2O promoted the racemic reaction pathway without YLB 1.
C. Effects of BuLi. BuLi affected reaction time rather than
enantioselectivity. As summarized in Table 3, the chemical yield
after 30 min in the absence of BuLi was only 17% (entry 1),
although the enantiomeric excess was high. The reaction did
not complete even after 2 h (entry 2, 76%). With BuLi, the
reaction completed within 30 min at -78 °C (entry 3). Similar
reactivity and enantioselectivity were obtained with LiHMDS
(entry 5), suggesting that the same active species was generated
with LiHMDS. Alkali metal affected both the enantioselectivity
and the reaction rate. Lithium gave better results than sodium
(17) For examples of racemic reactions, see: (a) Scholl, M.; Lim, C.-K.; Fu,
G. C. J. Org. Chem. 1995, 60, 6229 and references therein. (b) Okimoto,
M.; Chiba, T. Synthesis 1996, 1188. (c) Berthiaume, D.; Poirier, D.
Tetrahedron 2000, 56, 5995. For related racemic reaction using diethyl
cyanophosphonate as a cyanide source, see: (d) Harusawa, S.; Yoneda,
R.; Kurihara, T.; Hamada, Y.; Shioiri, T. Chem. Pharm. Bull. 1983, 31,
2932. (e) Kurihara, T.; Harusawa, S.; Yoneda, R. J. Synth. Org. Chem.
Jpn. 1988, 46, 1164 and references therein.
(15) Although rare earth metal complexes with a coordination number higher
than eight are well-known, maximum coordination number of rare earth
metals in rare earth-alkali metal-BINOL (1:3:3) heterobimetallic com-
plexes reported so far is seven, according to crystal structure analysis of
various complexes. See refs 7, 16 and references therein. See, also:
Aspinall, H. C. Chem. ReV. 2002, 102, 1807.
(16) Coordination of one molecule of H2O to a rare earth metal center was
observed when Pr, Nd, and Eu metals were used. See, ref 7. On the other
hand, in a related heterobimetallic Yb-Li-BINOL complex, Salvadori et
al. reported that the coordination number of the Yb metal center is six,
and that H2O does not coordinate with the Yb metal center. In the case of
YLB 1, the NMR experiment in Figure 3 implied the coordination of H2O
with the Y metal center. For analysis of the heterobimetallic Yb complex,
see: (a) Bari, L. D.; Lelli, M.; Pintacuda, G.; Pescitelli, G.; Marchetti, F.;
Salvadori, P. J. Am. Chem. Soc. 2003, 125, 5549. (b) Bari, L. D.; Lelli,
M.; Salvadori, P. Chem.sEur. J. 2004, 10, 4594.
(18) Livinghouse, T. Org. Synth. 1984, 60, 126. Although LiH was used to
generate LiCN from acetone cyanohydrin in the original report, we used
LiHMDS instead of LiH to obtain LiCN with high purity. [Procedure:
Caution! LiCN is highly toxic] To a stirred solution of acetone cyanohydrin
in THF at 0 °C was added 1 equiv of LiHMDS. The mixture was stirred
at 0 °C for 30 min, then solvent was evaporated under reduced pressure to
afford colorless powder in quantitative yield. Because LiCN is highly
hygroscopic, LiCN was handled under Ar and immediately used after
preparation.
(19) The beneficial effects of H2O on enantioselectivity were reconfirmed by
the control experiments excluding H2O. When the cyanation was performed
in the absence of H2O, enantiomeric excess decreased significantly. With
LiCN (10 mol %) in the absence of H2O (control experiment for Scheme
2), the results were 65% ee, 96% yield, after 2 h; with acetone cyanohydrin
(10 mol %) in the absence of H2O (control experiment for Scheme 3), the
results were 63% ee, 99% yield, after 2 h.
9
3416 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005

Cyano-Ethoxycarbonylation Reaction of Aldehydes using YLB 1
A R T I C L E S
Figure 5. Reaction profiles (a) without BuLi, (b) with BuLi, and (c) with
LiCN instead of BuLi.
additive (BuLi or LiCN), and the mixture was stirred for 5 min.
Then, ethyl cyanoformate 2 was added, and the carbonyl peak
of aldehyde 4a (1700 cm-1) was monitored (0-15 min). The
reaction proceeded smoothly with LiCN instead of BuLi (Figure
5c), while the initiation period was observed in Figure 5b. The
reaction profile had a sigmoid curve in Figure 5b. The initial
reaction rate at -78 °C was very slow in Figure 5a. The results
suggested that generation of the active species was very slow
in the absence of BuLi at -78 °C and was moderate even in
the presence of BuLi.
To investigate the generation of LiCN from LiOH and ethyl
cyanoformate in more detail, we performed in situ IR experi-
ments using a stoichiometric amount of the catalyst species and
ethyl cyanoformate. When 1 molar equiv of ethyl cyanoformate
2 was treated with a mixture of YLB (1 molar equiv), H2O (3
molar equiv), Ar3PdO 3a (1 molar equiv), and BuLi (1 molar
equiv) at -78 °C, the absorption of the carbonyl group of ethyl
cyanoformate (ν ) 1750 cm-1) gradually shifted to a new peak
(ν ) 1740 cm-1) (Figure 6). Because the two peaks (1740 and
1750 cm-1) overlapped, the profiles of the two peaks were
separated mathematically (Figure 6a).20 Unexpectedly, ethyl
cyanoformate was not completely consumed, although just 1
equiv of the cyanide source was used (Figure 6b). The result
Figure 6. Reaction profiles using a stoichiometric amount of catalyst.
implied a side reaction that consumed the LiOH. We assumed
that LiOH reacted with ethyl cyanoformate to generate LiCN
and ethyl hydrogen carbonate 6. Because 6 has an acidic proton,
6 would immediately react with LiOH to generate H2O and
lithium ethyl carbonate 7; 7 would be stable enough at -78 °C
in THF to avoid decomposition into CO2 and LiOEt, which
would react with H2O to regenerate LiOH. Thus, only one-half
of the LiOH would react with ethyl cyanoformate to generate
LiCN, and the other half would be consumed to generate lithium
salt 7. In fact, the absorption of 2 (ν ) 1750 cm-1) disappeared
completely within 15 min when an additional 1 molar equiv of
BuLi was added to the reaction mixture (Figure 6c, at 50 min).
The experiment shown in Figure 7 supported our assumption.
In Figure 7, 1 molar equiv of ethyl cyanoformate 2 was treated
with 50 mol % YLB, 150 mol % H2O, 50 mol % Ar3PdO 3a,
and 50 mol % BuLi at -78 °C for 1 h, followed by the addition
of 1 molar equiv of aldehyde 4h. The reaction was monitored
by gas chromatography. Chemical yield reached approximately
75%. The results suggested that only 25% of the ethyl
cyanoformate reacted to generate LiCN, and approximately 75%
of the ethyl cyanoformate remained when using 50 mol % LiOH.
If 7 decomposed smoothly at -78 °C to CO2 and LiOEt, the
chemical yield of product 5h should be less than 50% under
(20) ConcIRT AF, version 3.5.0.4 software (METTLER TOLEDO Autochem),
was used for mathematical treatment of IR spectra.
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3417

A R T I C L E S
Yamagiwa et al.
Scheme 3. Catalytic Asymmetric Cyano-Ethoxycarbonylation
Reaction using Acetone Cyanohydrin as a Precursor of LiCN
not suitable from a practical point of view due to its high toxicity
and hygroscopic properties. Acetone cyanohydrin is a suitable
precursor of LiCN, as shown in Scheme 3 and Figure 9. With
Figure 7. Reaction profile using 50 mol % YLB, 150 mol % H2O, 50 mol
% Ar3PdO 3a, and 50 mol % BuLi.
Figure 9. Reaction profile of catalytic asymmetric cyano-ethoxycarbony-
lation reaction with acetone cyanohydrin as an initiator.
10 mol % acetone cyanohydrin 8 as an initiator, the reaction
proceeded smoothly, and the product was obtained in the same
enantiomeric excess (93%) and yield (>99%) as that obtained
by using LiCN directly (Scheme 3).19 Because the reaction
proceeded too fast to evaluate the reaction rate with 10 mol %
catalyst loading, the reaction profile was monitored with 5 mol
% catalyst loading pretreated with 5 mol % acetone cyanohydrin
8. The reaction completed within 9 min at -78 °C (Figure 9).
In this case, the reaction profile did not have a sigmoid curve
and was similar to that observed with LiCN (see also Figure
5b,c).
Trials to reduce catalyst loading are summarized in Table 4.
In the absence of acetone cyanohydrin 8, the reaction with 1
mol % catalyst resulted in a less satisfactory reaction rate and
enantioselectivity compared with 10 mol % (entry 1) and 5 mol
% catalyst loading (entry 2). As shown in entries 3 and 4, the
reaction did not complete after 2 h (entry 3, 79% yield), and
product 5h was obtained in 96% yield after 9 h, but in only
90% ee (entry 4). On the other hand, in the presence of 1 mol
% 8, 1 mol % catalyst was sufficient to complete the reaction
within 2 h; 5h was obtained in 93% yield and 95% ee (entry
5).
Figure 8. Supposed initiation step to generate LiCN.
the conditions in Figure 7. The postulated initiation step to
generate LiCN is summarized in Figure 8. LiOH generated from
H2O and BuLi reacted with ethyl cyanoformate to afford LiCN.21
The reaction of LiOH with ethyl cyanoformate was rather slow
at -78 °C as detected by in situ IR (Figure 6). Approximately
one-half of the LiOH was consumed by the reaction with 6,
and decomposition of 7 was slow at -78 °C. Thus, only one-
half of the BuLi generated the active nucleophilic species.
D. Optimization of Reaction Conditions. As shown in the
in situ IR spectra in Figure 5a-c, the addition of a catalytic
amount of LiCN instead of BuLi was effective to avoid the
slow initiation step and to accelerate the reaction. Moreover,
by adding LiCN directly to the catalyst mixture, the undesired
pathway consuming LiOH was excluded, thus increasing the
amount of the active species. Use of LiCN itself is, however,
E. Effects of Phosphine Oxide. Triarylphosphine oxide was
essential to achieve high reactivity and enantioselectivity in the
present reaction.12 To evaluate substituent effects of the aromatic
ring, various triarylphosphine oxides (Figure 10) were synthe-
(21) In Table 1, entry 3, reaction proceeded without adding H2O at -60 °C.
We speculated that a part of lithium binaphthoxide of YLB might react
slowly with ethyl cyanoformate to produce active species. In the case
without BuLi, H2O would slowly attack 2 to generate LiCN (Table 1, entries
4-8).
9
3418 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005

Cyano-Ethoxycarbonylation Reaction of Aldehydes using YLB 1
Table 4. Trials to Reduce Catalyst Loading
A R T I C L E S
catalyst
8
time
(h)
entry
(
×
mol %)
(
×
mol %)
% yield
% ee
1
2
3
4
5
10
5
1
1
1
0
0
0
0
1
2
2
2
9
2
88
82
79
96
93
98
96
91
90
95
sized13 and used in an asymmetric cyano-ethoxycarbonylation
reaction of aldehyde 4h. The enantiomeric ratios with various
phosphine oxides are summarized in Figure 11. In the absence
of phosphine oxide, enantioselectivity was modest (er ) 8.7
and 79.0% ee). With 3a, the enantiomeric ratio was highest (er
) 88.8 and 97.8% ee). Phosphine oxides 3b-3e gave enantio-
selectivity much lower than that of 3a. The results with 3b-3e
suggested that electronic effects of substituents are not very
important. High enantioselectivity (er >50) was achieved when
Figure 12. Effects of amount of phosphine oxides 3a and 3b on
enantiomeric ratio.
using 3f, 3h, 3j, and 3k. Even one or two methoxy groups at
the ortho position were sufficient to achieve a high enantiomeric
ratio (3j, er ) 73.5; 3k, er ) 61.6). On the other hand, 3g and
3i with ortho-alkyl substituents were not effective, indicating
that the bulkiness of the substituents is not important. These
results suggested that chelating coordination with one of the
methoxy groups and phosphine oxide is important.
The ratio of phosphine oxide to the YLB complex was also
important to achieve optimal enantioselectivity (Figure 12).
When using 3a, just 1 equiv of 3a to YLB 1 produced optimal
enantioselectivity. Too much 3a gradually decreased enantio-
selectivity, probably due to acceleration of the undesired racemic
pathway without YLB. In the presence of excess 3a, 3a would
strongly coordinate with LiCN, and LiCN might dissociate from
YLB. On the other hand, the enantiomeric ratio of the product
was not significantly dependent on the amount of unsubstituted
phosphine oxide 3b, the coordination ability of which would
be weaker than chelating 3a.
Figure 10. Various substituted phosphine oxides.
To obtain further insight into the role of phosphine oxide,
initial rate kinetics were investigated using in situ IR. The results
are summarized in Figure 13a-f.22 In the presence of phosphine
oxide, there was first-order dependency on the catalyst mixture
(YLB:H2O:Ar3PdO:BuLi ) 1:3:1:1) (Figure 13a), zero-order
dependency on aldehyde (Figure 13b), and first-order depen-
dency on ethyl cyanoformate (Figure 13c). On the other hand,
in the absence of phosphine oxide, there was first-order
dependency on the catalyst mixture (YLB:H2O:BuLi ) 1:3:1)
(Figure 13d), zero-order dependency on the aldehyde (Figure
13e), and zero-order dependency on ethyl cyanoformate (Figure
13f). Phosphine oxide 3a changed the rate-determining step of
the reaction. In other words, phosphine oxide 3a accelerated a
step that was rate-determining without 3a, and another step in
the catalytic cycle became the new rate-determining step.
F. Active Catalyst Species and Catalytic Cycle. On the basis
of experimental results and discussion in sections B, C, and E
concerning H2O, BuLi, generation of LiCN, and phosphine
(22) In Figure 13, initial reaction rate is shown as relative rate. Rate with the
lowest concentration of variable factor is defined as standard (relative rate
) 1), and other rates are shown relative to the standard. In Figure 13a,c,
kinetics were measured without adding acetone cyanohydrin 8. For detailed
results of initial rate kinetics, see Supporting Information.
Figure 11. Effects of substituents on aromatic ring of phosphine oxides.
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3419

A R T I C L E S
Yamagiwa et al.
Figure 13. Initial rate kinetics (a-c) with phosphine oxide 3a, and (d-f) without 3a.
oxide, we hypothesize that YLB:H2O:LiCN:phosphine oxide 3a
) 1:1:1:1 would be an active catalyst species. The structure of
the proposed active catalyst species is shown in Figure 14a.
H2O would coordinate to the yttrium center and modify the
chiral environment of YLB. Phosphine oxide would coordinate
to the lithium cation of LiCN in a bidentate manner, thus
increasing the nucleophilicity of the cyanide anion23 as well as
affecting the enantioselectivity. Both LiCN and phosphine oxide
3a self-assembled with YLB, thus, the racemic pathway without
YLB was negligible. In fact, neither phosphine oxide 3a alone
nor a mixture of 3a and LiCN had good solubility in THF, even
at room temperature, while a mixture of YLB, 3a, and LiCN
easily dissolved in THF even at -78 °C. The difference in
solubility would assist self-assembly and suppress the racemic
pathway. ESI-MS spectra of YLB 1 alone showed four major
peaks: m/z ) 969.3 (YLB + Li+) as a base peak, m/z ) 1267.9
(YLB
+ BINOL-Li2 + ) 671.4 (YLi(bi-
Li+), m/z
naphthoxide)2 + Li+), and m/z ) 305.3 (BINOL-Li2 + Li+).
In the presence of phosphine oxide 3a, H2O, BuLi, and ethyl
cyanoformate, a new peak corresponding to YLB with two 3a
and lithium, m/z ) 1886.0 [YLB + (Ar3PdO 3a)2 + Li+],
(23) We assumed that Lewis basic phosphine oxide 3a would coordinate to the
lithium cation and increase electron density of the cyanide anion, thus
increasing nucleophilicity of the cyanide anion.
9
3420 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005

Cyano-Ethoxycarbonylation Reaction of Aldehydes using YLB 1
A R T I C L E S
Figure 14. (a) Structure of postulated active species, and (b) proposed
catalytic cycle of asymmetric cyano-ethoxycarbonylation reaction.
appeared and the peak corresponding to YLB (m/z ) 969.3)
disappeared.24 Although a peak corresponding to YLB:3a:Li+
) 1:1:1 was not observed under ESI-MS analysis conditions,
we thought the active species would contain only one 3a on
the basis of the experiments shown in Figure 12. On the other
hand, there was no peak corresponding to (Y + Li + 2
binaphtholate units). Although ESI-MS analysis does not
necessarily support the structure of an active species in the
catalytic cycle,25 the results at least suggested that self-assembly
of YLB with phosphine oxide 3a and LiCN occurred, and that
the framework of YLB (Y:Li:BINOL ) 1:3:3) would remain
unchanged in the reaction mixture with additives.26 Because the
coordination number of yttrium in YLB is assumed to be, at
most, 7, the coordination site of yttrium of YLB is occupied
with H2O and three BINOLs. Aldehyde would not coordinate
with the yttrium center. Aldehyde would coordinate with either
lithium metal or a proton of H2O coordinated with the yttrium
Figure 15. Profiles of IR spectra at -45 °C (condition A: a-1 and a-2)
and at -78 °C (condition B: b) after addition of aldehyde 4h to 9a.
Scheme 4. Two Possibilities to Induce Chiralitya
a Case 1: enantioselective addition of CN- to Aldehyde 4. Case 2:
dynamic kinetic resolution of rac-9.
center, although it is difficult to determine the exact coordination
site of aldehyde through spectroscopic analysis.
The postulated catalytic cycle of the reaction is shown in
Figure 14b. To explain the results of the initial rate kinetics in
Figure 13, the catalytic cycle should be constructed in three
steps (A-C in Figure 14b). Step A: aldehyde coordinates with
the active species (i) to afford (ii). Step B: cyanide attacks
aldehyde activated by the catalyst, affording the cyanohydrin-
catalyst complex (iii). Step C: cyanohydrin-catalyst complex
(iii) reacts with ethyl cyanoformate to afford the product and
to regenerate (i). In the absence of phosphine oxide 3a, the
reaction had zero-order dependency on the concentration of
aldehyde and ethyl cyanoformate and first-order dependency
on the catalyst mixture. Therefore, the rate-determining step
without 3a is thought to be step B. Phosphine oxide 3a increased
nucleophilicity of the cyanide anion, accelerating the rate of
step B. Thus, the rate-determining step with 3a changed to step
C.
(24) For ESI-MS spectra of YLB 1 under various conditions, see Supporting
Information.
(25) In the mechanistic studies of catalytic asymmetric reactions, the observed
predominant species is not always an active species. For example, see:
Landis, C.; Halpern, J. J. Am. Chem. Soc. 1987, 109, 1746.
(26) There is a controversy whether active species of rare earth-alkali metal
heterobimetallic complexes have two or three binaphtholates. Trials to
unequivocally determine the structures of active species of heterobimetallic
complexes in various asymmetric reactions are still ongoing. Structures of
active species would change depending on the metal and nucleophile used.
For a preliminary report from our group, see ref 8b. For other opinions,
see ref 16. In the present cyanation reaction, Y(HMDS)3:BuLi:BINOL )
1:3:3 complex (YLB 1) provided product 5a in 95% yield and 96% ee
within 30 min at -78 °C (Table 3, entry 3), while reaction rate with
Y(HMDS)3:BuLi:BINOL ) 1:1:2 mixture was very slow. Product was
obtained in only 20% yield after 2 h, albeit in high enantiomeric excess
(93%). On the other hand, Y(HMDS)3:BuLi:BINOL ) 0:2:1 mixture
promoted the cyanation reaction of 4a smoothly, and 5a was obtained in
93% yield after 2 h at -78 °C, but in only 2% ee. Thus, we speculate that
dissociation of Y:Li:BINOL ) 1:3:3 complex into Y:Li:BINOL ) 1:1:2
complex and Y:Li:BINOL ) 0:2:1 complex is less likely, and that Y:Li:
BINOL ) 1:3:3 would be the more probable framework of active species
in the present reaction. Further quantitative analysis to unequivocally
determine the structure of active species will be discussed elsewhere.
In the catalytic cycle in Figure 14b, there are two possibilities
to transfer chirality from the chiral catalyst to the product, as
shown in Scheme 4. One possibility is irreversible formation
of the chiral cyanohydrin intermediate 9, where enantioselective
addition of cyanide to aldehyde in step B would determine the
9
J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005 3421

A R T I C L E S
Yamagiwa et al.
Scheme 5. Evaluation of Reversibility of Cyanide Addition to
Aldehyde at -45 and -78 °C
the addition is reversible at -45 °C. Thus, the possibility of a
dynamic kinetic resolution pathway under the optimized reaction
conditions at -78 °C was excluded.28 In the present asymmetric
catalysis, the catalyst recognized the enantioface of aldehydes
in the cyanide addition step (step B in Figure 14b and Scheme
4).
In summary, we developed a catalytic enantioselective cyano-
ethoxycarbonylation reaction of aldehydes with ethyl cyano-
formate 2 using a heterobimetallic YLB 1 complex. Under
optimized conditions, achiral additives, H2O, tris(2,6-dimethox-
yphenyl)phosphine oxide 3a, and BuLi had key roles in
achieving high reactivity and enantioselectivity (up to >99%
yield and up to 98% ee). The mechanism of the present reaction
for generating the active nucleophilic species was different from
that in previously reported reactions using related heterobime-
tallic complexes.7a,b Detailed mechanistic studies revealed that
the initiation step to generate the active nucleophilic species is
rather slow under the preliminary optimized conditions. To
accelerate the initiation step, a catalytic amount of acetone
cyanohydrin 8 was effective as an initiator. In the presence of
8, the reaction completed within 9 min using 5 mol % catalyst
at -78 °C. With 8 as an initiator, catalyst loading was
successfully reduced to 1 mol %. The high reaction rate of the
present catalysis is noteworthy. Roles of achiral additives and
catalytic cycle of the reaction were also clarified through in
situ IR analysis, initial rate kinetics, and other mechanistic
experiments.
enantiomeric excess of the products. The other possibility is
dynamic kinetic resolution of racemic cyanohydrin intermediate
9. Because the reaction conditions are supposed to be somewhat
basic, step B in Figure 14b might be reversible. In this case,
enantioselectivity is induced during the reaction of the racemic
cyanohydrin-catalyst complex with ethyl cyanoformate (step
C). To determine the actual reaction pathway and enantiomer-
recognizing step of the reaction, reversibility of step B in Figure
14b was examined using in situ IR analysis. The experiments
are summarized in Scheme 5, and the results of in situ IR
analysis are summarized in Figure 15. When benzaldehyde 4a
was treated with 1 molar equiv of YLB, 3 molar equiv of H2O,
1 molar equiv of 3a, and 1 molar equiv of LiCN, IR absorption
of aldehyde 4a (ν ) 1700 cm-1) immediately disappeared,
suggesting generation of cyanohydrin intermediate 9a. Then, 1
molar equiv of aldehyde 4h was added to the reaction mixture.
The mixture was stirred at either -45 °C (condition A in
Scheme 5, Figure 15a-1,a-2) or -78 °C (condition B in Scheme
5 and Figure 15b), and the IR spectra profile was monitored.
Acknowledgment. We thank Grant-in-Aid for Specially
Promoted Research, Grant-in-Aid for Encouragements for
Young Scientists (B) (for S.M.) from JSPS and MEXT for
financial support.
Supporting Information Available: Experimental procedures,
detail data for kinetic studies, and ESI-MS. This material is
available free of charge via the Internet at http://pubs.acs.org.
As shown in Figure 15a, absorption of 4a (ν ) 1700 cm-1
)
JA042887V
was gradually regenerated at -45 °C, suggesting that the cyanide
group was reversibly transferred from 9a to 4h, affording 9h.
On the other hand, at -78 °C, the peak corresponding to 4a
was not detected (Figure 15b).27 These results indicated that
cyanide addition to the aldehyde is irreversible at -78 °C, while
(27) Because intensity of carbonyl absorbance of 4h was much weaker than
that of 4a, the y-axis scale in Figure 15b is different from that in Figure
15a.
(28) Kinetic dynamic resolution pathway was proposed by Deng et al. when
using chiral Lewis base catalyst in asymmetric cyano-ethoxycarbonylation
of ketones at -12 or -24 °C. See ref 2.
9
3422 J. AM. CHEM. SOC. VOL. 127, NO. 10, 2005