Enantioselective copper-catalyzed allylic substitution reaction with aminohydroxyphosphine ligand

Article abstract of DOI:10.1002/adsc.200900054

A proline-derived aminohydroxyphosphine ligand induced excellent regio- and enantiose-lectivities (S_N2′: S_N2>94:6, 94-97% ee) in the copper-catalyzed substitution reaction of cinnamyl phosphates with Et₂Zn. A computational

Full text of DOI:10.1002/adsc.200900054

COMMUNICATIONS
DOI: 10.1002/adsc.200900054
Enantioselective Copper-Catalyzed Allylic Substitution Reaction
with Aminohydroxyphosphine Ligand
Naohiko Yoshikai,a Kotaro Miura,a and Eiichi Nakamuraa,*
a
Department of Chemistry, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
Fax : (+81)-3-5800-6889; e-mail: nakamura@chem.s.u-tokyo.ac.jp
Received: January 24, 2009; Revised: March 8, 2009; Published online: May 5, 2009
Dedicated to Prof. Armin de Meijere on the occasion of his 70th birthday.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/adsc.200900054.
Abstract: A proline-derived aminohydroxyphos-
phine ligand induced excellent regio- and enantiose-
lectivities (SN2’:SN2>94:6, 94–97% ee) in the
copper-catalyzed substitution reaction of cinnamyl
phosphates with Et2Zn. A computational modelling
study of the reaction pathway suggests that the dis-
placement of the leaving group takes place in an or-
dered, anti-fashion through simultaneous copper–
olefin and zinc–oxygen interactions, which is critical
for the enantioselection.
unsaturated carbonyl compound.[5] The multidentate
structure of the aminohydroxyphosphine ligand and
the modular character of the synthesis[8] readily allow
us to modify the ligand to make it suitable for other
types of asymmetric copper catalysis. We report here
that this ligand design is effective for the enantiose-
lective allylic substitution reaction. Thus, a proline-de-
rived ligand L5 was found to show excellent regio-
and enantioselectivities in the copper-catalyzed substi-
tution of allylic phosphates with Et2Zn. Theoretical
working models for the asymmetric induction are also
provided to account for the observed enantioselectivi-
ty.
À
Keywords: allylic substitution; C C bond forma-
tion; copper; organozinc reagents; phosphane li-
gands
We performed a preliminary examination of a vari-
ety of aminohydroxyphosphine ligands in the reaction
Chiral copper complex-catalyzed reactions of organo-
metallic reagents with electrophiles such as a,b-unsa-
turated carbonyl compounds and allylic halides and
pseudohalides provide powerful methods for the ste-
reoselective construction of carbon-carbon bonds.[1]
While the past decade has seen significant advances
in this area, the intrinsic mechanistic complexity of
these reactions involving multimetallic catalysis has
hampered an understanding of the origin of the enan-
tioselectivity and hence the rational design of the li-
gands and the catalysts.[2–6]
Through our theoretical mechanistic studies on or-
ganocopper transformations,[2] we suggested that an
effective chiral ligand would need to coordinate to
the nucleophilic copper atom on the one hand, and to
the Lewis acidic main group metal atom on the
other.[7] The validity of this suggestion was supported
by a tridentate aminohydroxyphosphine ligand L1
(Scheme 1) that effected the highly enantioselective
Scheme 1. Performance of aminohydroxyphosphine ligands
conjugate addition of a dialkylzinc to an acyclic a,b- in copper-catalyzed enantioselective allylic alkylation.
1014
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1014 – 1018

Enantioselective Copper-Catalyzed Allylic Substitution Reaction
COMMUNICATIONS
Table 1. Enantioselective allylic substitution of cinnamyl di-
Table 2. Enantioselective substitution of allylic phosphates
with Et2Zn.
ethyl phosphate with Et2Zn.[a]
Entry Cu salt
Solvent T [8C] SN2’:SN2[b] ee[c] [%]
1
2
3
4
5
6
7
8
Cu
Cu
Cu
Cu
Cu
Cu
CuOTf
Cu(OAc)2
CuI
U
THF
DME
Et2O
Toluene À25
CH2Cl2 À25
CH2Cl2 À40
CH2Cl2 À40
CH2Cl2 À40
CH2Cl2 À40
À25
À25
À25
87:13
88:12
82:18
81:19
86:14
88:12
88:12
88:12
93:7
93:7
92:8
91:9
98:2
40
44
55
46
60
64
64
64
65
68
82
88
97
28
Entry
R
SN2’:SN2[a]
ee[b] [%]
Yield[c] [%]
1
2
3
4
5
6
7
8
9
Ph
97:3
97:4
96:4
97:3
97:3
96:4
94:6
>99:1
95:5
97:3
96
95
97
96
96
96
94
84[d]
53
37
53
56
64
56
36
41
59
28
4-MeC6H4
4-NO2C6H4
4-ClC6H4
4-BrC6H4
4-CF3C6H4
2-Naphthyl
EtO2C
9
10
11
12
13
14
CuBr·SMe2 CH2Cl2 À40
CuCl
CH2Cl2 À40
CuCl2·2H2O CH2Cl2 À40
CuCl2·2H2O CH2Cl2 À78
c-C6H11
PhMe2Si
<10[e]
<10[e]
CuCN
CH2Cl2 À40
97:3
10
[a]
Reaction conditions: 5 mol% of Cu salt, 6 mol% of L5,
1.5 equiv. of Et2Zn, 20 h. Over 90% conversion of the
starting material was attained except for entry 13 (60%
conversion).
Determined by GC analysis.
Determined by chiral GC analysis.
[a]
Determined by GC analysis.
Determined by chiral GC or HPLC analysis.
Isolated yields unless otherwise noted.
Determined for an alcohol obtained by LiAlH4 reduction
of the product.
Estimated by GC analysis.
[b]
[c]
[d]
[b]
[c]
[e]
of cinnamyl diethyl phosphate and diethylzinc in THF
at À258C using Cu(OTf)2 (5 mol%) as a copper pre-
AHCTUNGTRENNUNG
catalyst (Scheme 1). For example, ligand L1, which the enantioselectivity up to 82% and 88%, respective-
gave excellent enantioselectivity in the conjugate ad- ly, with a regioselectivity of >90:10 (entries 11, 12).
dition reaction,[5,9] gave a poor enantioselectivity of By using CuCl2·2H2O at À788C, both the regio- and
11% ee. No asymmetric induction was observed with enantioselectivity reached to excellent levels
its diastereomer L2. A bulkier analogue L3 also gave (SN2’:SN2=98:2, 97% ee, entry 13). The use of CuCN
poor selectivity. A proline-derived ligand L5 gave resulted in high regioselectivity (97:3), but considera-
moderate enantioselectivity of 40% ee with moderate bly reduced the enantioselectivity (28% ee)
regioselectivity (87:13), while its diastereomer L4 did (entry 14), perhaps because of uniquely strong coordi-
not induce any enantioselectivity. Cinnamyl halides nation of the cyanide ion to the copper and zinc
were found to be poorly selective and cinnamyl ace- atoms.[10,11] On the basis of these results, we employed
tate unreactive.
CuCl2·2H2O as the copper source and CH2Cl2 as the
With these results in hand, we carefully optimized solvent in the following study.
the conditions for the ligand L5 as summarized in
Table 2 shows representative results of the enantio-
Table 1. Entries 1–5 show the effect of solvent on the selective substitution of allylic phosphates with Et2Zn.
regio- and the enantioselectivity at À258C. Among The reactions of the parent and substituted cinnamyl
ethereal solvents, Et2O gave better enantioselectivity phosphates gave uniformly high regio- and enantiose-
(55% ee) than the more strongly coordinating sol- lectivities (SN2’:SN2ꢀ96:4, ꢀ95% ee, entries 1–6). The
vents THF and DME (40–44% ee) (entries 1–3). The reactions were rather sluggish at À788C and gave
polar non-coordinating solvent CH2Cl2 gave even moderate yields (up to 64%) of substitution products.
better enantioselectivity (60% ee, entry 5), while non- The rest of the starting materials were entirely recov-
polar toluene gave only 46% ee (entry 4).
ered. Increased catalyst loading (up to 20%) and/or
Using CH2Cl2, the effect of the copper precatalyst further addition of Et2Zn (up to 3 equiv.) did not im-
was examined (entries 6–14). Copper salts having prove the yield very much. Me2Zn was entirely un-
oxygen counteranions (entries 6–8) gave identical re- reactive. A substrate bearing a 2-naphthyl group also
sults (SN2’:SN2 ratio of 88:12, 64% ee at À408C). gave excellent regio- and enantioselectivities
Iodide and bromide counteranions improved the re- (entry 7). An allylic phosphate bearing an ethoxycar-
gioselectivity (93:7), but had little effect on the enan- bonyl substituent resulted in an enantioselectivity at
tioselectivity (65–68% ee) (entries 9 and 10). In con- an acceptable level (84% ee) with perfect regioselec-
trast, the chloride counteranion proved to be benefi- tivity (>99:1), albeit in a low yield (entry 8).[12] Great-
cial. Thus, the use of CuCl and CuCl2·2H2O enhanced ly reduced reactivity and enantioselecitivity were ob-
Adv. Synth. Catal. 2009, 351, 1014 – 1018
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1015

Naohiko Yoshikai et al.
COMMUNICATIONS
Figure 1. Energy profile (ZPE corrected) for the reaction of
the Cu/Zn complex 1 (R=Me) and the allylic phosphate 2
(R1 =Ph, R2 =Me) at the level of B3LYP/SDD for Cu, Ahl-
richs SVP for Zn, and 6-31G(d) for the rest.
(3-Si to 4-Si via TS1-Si) is very facile (DE° =
4.8 kcalmolÀ1) and moderately exothermic (DE=
À10.3 kcalmolÀ1). The subsequent reductive elimina-
tion requires an activation energy as low as
6.9 kcalmolÀ1 and is highly exothermic. The calculated
energy profile does not give unambiguous information
on the turnover-limiting step of the catalysis; never-
Scheme 2. Proposed catalytic cycle.
served for aliphatic and silyl-substituted substrates theless, in light of the entropy cost of bimolecular as-
(entries 9 and 10).[13]
sociation, we presume that either the p-complexation
While little is known about the detailed reaction step or the oxidative addition step is responsible for
pathway of copper-catalyzed allylic alkylation using the slow turnover of the present catalysis.
organozinc reagents, we propose a catalytic cycle as
The low activation energy for the reductive elimi-
outlined in Scheme 2 for the formation of the major nation indicates that the enantioselectivity of the
À
enantiomer, on the basis of our previous mechanistic product is determined prior to the C C bond forma-
study on the allylic substitution of lithium organocup- tion, i.e., in the p-complexation or the oxidative addi-
rates.[10,14] First, the aminohydroxyphosphine ligand tion steps. In order to gain insights into the origin of
forms, upon deprotonation by the organozinc reagent, the asymmetric induction, we analyzed the p-complex
a copper/zinc bimetallic complex 1, where the phos- 3-Si and the oxidative addition TS TS1-Si as well as
phorus and the nitrogen atoms coordinate to the Cu their diastereomers (3-Re and TS1-Re) leading to the
and the Zn atoms, respectively, and the oxygen atom minor enantiomer (Figure 2).
bridges the two metals.[5] The complex 1 gives, by
An important common structural feature of 3-Si
copper–olefin and zinc–oxygen interactions with the and 3-Re is the location of the nucleophilic copper
allylic phosphate 2, a p-complex 3-Si. The complex atom and the Lewis acidic zinc atom. Thus, the
À
À
3-Si then undergoes oxidative addition (C O bond copper atom interacts with the olefinic moiety (Cu
cleavage) in a push-pull, anti manner to give an C: ca. 2.1 ꢁ) while the zinc atom is coordinated by
À
allylcopper
(III) intermediate 4-Si. The C C bond the Lewis basic P=O oxygen atom of the phosphate
À
forming reductive elimination of 4-Si gives a product- leaving group (Zn O: ca. 2.3 ꢁ). The alkoxide group
bound complex 5-Si, which regenerates the initial cat- in the ligand acts as a counteranion for the zinc atom
alytic species 1 through transmetalation with R2Zn.
À
(Zn O: ca. 1.9 ꢁ), and at the same time weakly coor-
À
A computational modelling study using DFT calcu- dinates to the copper atom (Cu O: ca. 2.6 ꢁ). In the
lations on the reaction of 1 (R=Me) and 2 (R1 =Ph, oxidative addition transitions states, one can find that
R2 =Me) along the catalytic cycle gave a reasonable C O bond cleavage (C O: ca. 1.9 ꢁ) is accompanied
À
À
energy profile (Figure 1).[15] The formation of 3-Si by tighter interactions between the zinc atom and the
À
from 1 and 2 is almost thermoneutral (DE=+0.3 kcal phosphate oxygen atom (Zn O: ca. 2.1 ꢁ) and be-
molÀ1), probably as the result of trade-off between tween the copper atom and the bridging oxygen atom
À
the energy gain through the copper–olefin and zinc– (Cu O: 2.3–2.4 ꢁ).
oxygen interactions and the energy loss due to the
We consider that the high degree of structural or-
structural deformation.[2] The oxidative addition step ganization of the p-complex and the oxidative addi-
1016
asc.wiley-vch.de
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1014 – 1018

Enantioselective Copper-Catalyzed Allylic Substitution Reaction
COMMUNICATIONS
Figure 2. Structures of diastereomeric p-complexes and oxidative addition transition states. Color code: green, copper; light
green, zinc; grey, carbon; white, hydrogen; red, oxygen; blue, nitrogen; pink, phosphorus. Hydrogen atoms on the ligand are
omitted for clarity. Values indicate bond lengths (ꢁ) and energies (kcalmolÀ1) relative to [1 + 2] (in parenthesis).
tion TS are responsible for the high enantioselectivity. asymmetric induction. The calculated structures of
The experimental enantioselectivity of 96% ee at the diastereomeric complexes and transitions states
À788C corresponds to a small energy difference of provide a pictorial illustration of the way that a multi-
1.5 kcalmolÀ1, which is rather difficult to reproduce at dentate ligand would hold many metal atoms together
the current level of theory. Nonetheless, both 3-Si and and exert cooperative effects.[3–5]
TS1-Si, leading to the major enantiomer, were calcu-
lated to be preferred to their diastereomers by
1.4 kcalmolÀ1 and by 0.5 kcalmolÀ1, respectively. It is
notable that we do not find any particularly severe
Experimental Section
steric interactions in the complexes and the transi-
tions states. For instance, the closest distances be-
Typical Procedure
tween the ligand and the phenyl group of the sub-
strate in 3-Si and 3-Re are 2.74 ꢁ and 4.41 ꢁ, respec- (27.0 mg, 0.06 mmol) in CH2Cl2 (0.5 mL) was stirred for 1 h
A solution of CuCl2·2H2O (8.5 mg, 0.05 mmol) and L5
at room temperature. The solution was then cooled to
À788C, and an allylic phosphate (0.5 mmol) was added
dropwise. After stirring for 15 min, Et2Zn (1.0M in hexane,
0.75 mL, 0.75 mmol) was added over 3 min, and the reaction
mixture was stirred for 3 d. The reaction was quenched by
MeOH (0.3 mL) and saturated Rochelle salt solution
(1 mL). After extraction with Et2O (1 mL ꢂ 3), the com-
bined organic layer was passed through a small plug of silica
gel and concentrated under reduced pressure. The crude
product was purified by silica gel chromatography (eluent:
tively. This subtlety of the selectivity control may be
the reason for the rather high sensitivity of the enan-
tioselectivity to the structure of the substrate
(Table 2). This model did not provide us any insights
on the low reactivity of Me2Zn as opposed to Et2Zn.
In summary, we have developed a proline-based
P,O,N tridentate ligand for the copper-catalyzed asym-
metric alkylation of allylic phosphates. The capability
of the ligand to hold Cu and Zn atoms together in
close proximity must be the driving force for the hexane) to obtain the desired allylation product.
Adv. Synth. Catal. 2009, 351, 1014 – 1018
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1017

Naohiko Yoshikai et al.
COMMUNICATIONS
Acknowledgements
[4] M. A. Kacprzynski, A. H. Hoveyda, J. Am. Chem. Soc.
2004, 126, 10676–10681.
[5] A. Hajra, N. Yoshikai, E. Nakamura, Org. Lett. 2006, 8,
4153–4155.
We thank the Ministry of Education, Culture, Sports, Science
and Technology of Japan for a Grant-in-Aid for Scientific
Research (#18105004 for E.N. and #19750026 for N.Y.), and
the Global COE Program for Chemistry Innovation. Compu-
tational time from the Research Center for Computational
Science, Okazaki National Research Institute, is also ac-
knowledged.
[6] S. R. Harutyunyan, F. Lꢆpez, W. R. Browne, A. Correa,
D. PeÇa, R. Badorrey, A. Meetsma, A. J. Minnaard,
B. L. Feringa, J. Am. Chem. Soc. 2006, 128, 9103–9118.
[7] E. Nakamura, M. Yamanaka, S. Mori, J. Am. Chem.
Soc. 2000, 122, 1826–1827.
[8] M. T. Reetz, Chem. Rev. 1999, 99, 1121–1162.
[9] J. M. Garcia, A. Gonzalez, B. G. Kardak, J. M. Odrio-
zola, M. Oiarbide, J. Razkin, C. Palomo, Chem. Eur. J.
2008, 14, 8768–8771.
References
[1] a) A. Alexakis, J.-E. Bꢃckvall, N. Krause, O. Pꢄmies,
M. Diꢅguez, Chem. Rev. 2008, 108, 2796–2823; b) S. R.
Harutyunyan, T. den Hartog, K. Geurts, A. J. Min-
naard, B. L. Feringa, Chem. Rev. 2008, 108, 2824–2852;
c) F. Lꢆpez, A. J. Minnaard, B. L. Feringa, Acc. Chem.
Res. 2007, 40, 179–188; d) H. Yorimitsu, K. Oshima,
Angew. Chem. 2005, 117, 4509–4513; Angew. Chem.
Int. Ed. 2005, 44, 4435–4439; e) A. H. Hoveyda, A. W.
Hird, M. A. Kacprzynski, Chem. Commun. 2004, 1779–
1785.
[2] E. Nakamura, S. Mori, Angew. Chem. 2000, 112, 3902–
3924; Angew. Chem. Int. Ed. 2000, 39, 3750–3771.
[3] Multimetallic Catalysis in Organic Synthesis, (Eds.: M.
Shibasaki, Y. Yamamoto), Wiley-VCH, Weinheim,
2004.
[10] N. Yoshikai, S.-L. Zhang, E. Nakamura, J. Am. Chem.
Soc. 2008, 130, 12862–12863.
[11] a) E. Nakamura, M. Yamanaka, J. Am. Chem. Soc.
1999, 121, 8941–8942; b) M. Yamanaka, E. Nakamura,
J. Am. Chem. Soc. 2005, 127, 4697–4706.
[12] K. E. Murphy, A. H. Hoveyda, J. Am. Chem. Soc. 2003,
125, 4690–4691.
[13] M. A. Kacprzynski, T. L. May, S. A. Kazane, A. H.
Hoveyda, Angew. Chem. 2007, 119, 4638–4642; Angew.
Chem. Int. Ed. 2007, 46, 4554–4558.
[14] M. Yamanaka, S. Kato, E. Nakamura, J. Am. Chem.
Soc. 2004, 126, 6287–6293.
[15] For details of the computational method, see Support-
ing Information.
1018
asc.wiley-vch.de
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1014 – 1018