Copper-free asymmetric allylic alkylation with grignard reagents

Article abstract of DOI:10.1002/anie.201000577

(Chemical Equation Presented) Open wide and say AAA: The copper-free asymmetric allylic alkylation reaction of Crignard reagents, catalyzed by N-heter-ocyclic carbenes, is reported for allyl bromide derivatives. This reaction offers good enantioselectivit

Full text of DOI:10.1002/anie.201000577

Communications
DOI: 10.1002/anie.201000577
Asymmetric Catalysis
Copper-Free Asymmetric Allylic Alkylation with Grignard Reagents**
Olivier Jackowski and Alexandre Alexakis*
During the past decade, the copper-catalyzed enantioselective
allylic alkylation reaction has been extensively studied owing
to its powerful potential to deliver a variety of enantiopure
compounds. By using Grignard, diorganozinc, or triorganoa-
luminum reagents, the regio- and enantioselectivity of the
ꢀ
C C bond-formation step is controlled using different types
of ligands, such as phosphites, phosphoramidites, ferrocenes,
and peptides.[1] The use of ligands that are based on an
N-heterocyclic carbene (NHC) framework, first used in the
asymmetric copper-catalyzed conjugate addition reaction,[2]
was more recently introduced for the asymmetric allylic
alkylation (AAA) reaction with great success. The groups of
Okamoto,[3] Hong,[4] and Tomioka[5] used Grignard reagents
as their organometallic reagent, whereas Hoveyda and co-
workers[6] reported the use of diorganozinc and triorgano-
aluminum reagents. Most NHC ligands that have been
reported thus far are C2 symmetric; the Hoveyda group has
developed a series of bidentate ligands wherein a free
hydroxy group forms an intermediate alkoxy copper species
(Figure 1).
Since our first report on the copper-catalyzed AAA
reaction,[7] we have focused on Grignard reagents. In contrast
to diorganozinc reagents, Grignard reagents are commercially
Figure 1. NHC ligands used in Asymmetric Allylic Alkylation (AAA)
reactions.
available, or very easily prepared, and a variety of alkyl
groups can be added using these reagents. Another advantage
of Grignard reagents is their ability to deprotonate the
imidazolium salt,[8] without requiring a preformed copper or
silver carbene. Our NHC ligands kept the diphenyl imidazo-
line core, with a mesityl group on one nitrogen atom, and a
more flexible benzylic group, rather than a directly linked
aromatic group, on the second nitrogen atom. Thus, several
NHC ligands were synthesized based on this design
(Figure 2). During our work, Uchida and Katsuki went on
the same design with ligand L3, which appeared to be very
good for the copper-catalyzed asymmetric conjugate addi-
tion.[9]
During the course of these studies on copper-catalyzed
AAA reactions with Grignard reagents and the alkoxy NHC
ligands, we observed that the same result was obtained with or
without the copper catalyst (Scheme 1).
[*] Dr. O. Jackowski, Prof. Dr. A. Alexakis
Department of Organic Chemistry
University of Geneva
Figure 2. NHCs used in this study. Bn=benzyl, Mes=mesityl.
Quai Ernest Ansermet 30, 1211 Geneva (Switzerland)
Fax: (+41)22-379-32-15
E-mail: alexandre.alexakis@unige.ch
A preliminary investigation was performed using an ethyl
Grignard reagent on cinnamyl bromide, with and without
copper (Scheme 1). Surprisingly, the results were identical for
the two reactions. Such an observation had already been
reported by Lee and Hoveyda on a specific substrate and
specific Grignard reagents.[10] Very recently, they also
[**] We thank the Swiss National Research Foundation (Grant No.
200020-126663) and COSTaction D40 (SER contract No. C07.0097)
for financial support.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201000577.
3346
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 3346 –3350

Angewandte
Chemie
Table 1: Optimization of experimental conditions.[a]
Entry
Solvent
T [8C]
Conversion [%][b]
g/a[b]
ee [%][c]
1
2
3
4
5
6
7
8
Et2O
MTBE
THF
CH2Cl2
toluene
Et2O[d]
Et2O
ꢀ15
ꢀ15
ꢀ15
ꢀ15
ꢀ15
ꢀ15
0
ꢀ30
ꢀ15
ꢀ15
ꢀ15
ꢀ15
>99
53
40
47
40
>99
>99
98
76
94
76:24
75:25
50:50
10:90
50:50
77:23
70:30
67:33
70:30
50:50
–
85
72
42
40
67
84
77
86
83
84
–
Scheme 1. The NHC-catalyzed AAA reaction of a Grignard reagent.
CuTC=copper thiophene carboxylate.
Et2O
Et2O[e]
Et2O[f]
Et2O[g]
Et2O[h]
reported the AAA reactions of cinnamyl phosphate deriva-
tives without copper using organozinc and organoaluminum
reagents.[11] However, they insisted that Grignard reagents did
not work well for this reaction, even on cinnamyl halides. In
view of our results, we decided to explore this new reactivity
of NHC–Grignard-reagent complexes for the copper-free
AAA reaction.
First, the substrate scope, and particularly the leaving
group, was explored (Scheme 2). The highest enantioselectiv-
ities were obtained with a good leaving group, but the
9
10
11
12
ꢁ1
>99
76:24
82
[a] Reaction time was 1 hour in all cases. [b] Determined by 1H NMR
spectroscopy. [c] Determined by SFC on a chiral stationary phase.
[d] With CuTC (1 mol%). [e] 0.125m. [f] 0.5m. [g] No ligand. [h] 4 mol%
ligand. SFC=supercritical fluid chromatography, MTBE=methyl tert-
butyl ether.
stopped the reaction (Table 1, entry 11). We also checked that
the regio- and enantioselectivities were the same at the
beginning and at the end of the reaction. In conclusion, the
reaction works well in diethyl ether at ꢀ158C with 1 mol% of
ligand L1 and the reaction times can even be decreased to
10 minutes.
With these optimal conditions in hand, several ligands
were tested to improve the regio- and enantioselectivities
(Table 2).
Scheme 2. Leaving group effect on the AAA reaction.
Table 2: Ligand screening.
regioselectivity followed the opposite trend, the best result
was obtained with cinnamyl phosphate. The acetate was not a
successful leaving group, since direct attack on the carbonyl
group occurred. Therefore, the conditions were optimized
with cinnamyl bromide as the best compromise. The solvents,
the temperature, the catalyst loading, and concentration of
the reaction mixture were also evaluated (Table 1). The
reaction time was kept to one hour to better investigate the
reactivity.
First, lowering the temperature to ꢀ158C improved the
enantioselectivity without affecting the regioselectivity
(Table 1, entries 1 and 7). However, at ꢀ308C, a small erosion
of the regioselectivity was observed (Table 1, entry 8). The
solvent played an important role: a moderately coordinating
solvent was required (Table 1, entries 1 and 2), but both THF
(Table 1, entry 3) and a noncoordinating solvent (Table 1,
entries 4 and 5) afforded poor regio- and enantioselectivities.
The effect of concentration was also addressed: dilution
slowed down the reaction without any noticeable change in
selectivity (Table 1, entry 9), whereas increased concentration
decreased the regioselectivity (Table 1, entry 10). Higher
catalyst loading (4%, instead of 1%) was clearly not
important (Table 1, entry 12), but the absence of an NHC
Entry
L
Conversion [%][a]
g/a[a]
ee [%][b]
1
2
3
4
5
6
7
8
9
L1
L2
L3
L4
L5
L6
L7
L8
L9
L10
ꢂ99
ꢂ99
70
80
98
76:24
70:30
70:30
75:25
70:30
72:28
67:33
67:33
69:31
9:91
85
80
57
53
51
61
63
53
36
34
98
ꢂ99
ꢂ99
99
ꢂ99
10
1
[a] Determined by H NMR spectroscopy. [b] Determined by SFC on a
chiral stationary phase.
The counter ion (Cl versus BF4) had no significant effect
on the reaction (Table 2, entries 1 and 2). However, the
regiochemistry (around 7:3) seems to be directed by the
mesityl group on the ligand. Ligand L10, which contained a
benzyl group instead of a mesityl group, gave a reversal of the
regioselectivity in favor of the a product (Table 2, entry 10).
Angew. Chem. Int. Ed. 2010, 49, 3346 –3350
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3347

Communications
On the other hand, the enantioselectivity seems to be
controlled by the substituent on the other nitrogen atom of
the NHC. The simple benzyl motif in ligand L4 afforded lower
enantiomeric excess values, which showed the necessity of the
phenolic hydroxy group (Table 2, entry 4). The homologated
naphthol moiety L8 did not improve the enantioselectivity
(Table 2, entry 8). The number of carbon atoms between the
nitrogen and the oxygen atoms seems important, as the
carbene reported by Hoveyda et al. (L9), which contained a
phenylphenol moiety, gave the lowest asymmetric induction
(Table 2, entry 9). Similarly low enantioselectivities were
observed with L6 and L7, which both had four carbon
atoms between the substituents (Table 2, entries 6 and 7). The
steric hindrance around the hydroxy group also had an
important effect on the enantiomeric excess (Table 2,
entry 3), as the carbene reported by Uchida and Katsuki
(L3) gave lower ee values. Finally, the best result (85% ee)
was obtained with ligand L1 (Table 2, entry 1), which
contained three carbon atoms between the nitrogen and the
hydroxyl groups; this chain length seems to be essential for
the enantioselectivity.
Scheme 3. Tentative catalytic cycle.
Next, the scope of the reaction was screened with different
Grignard reagents on cinnamyl bromide (Table 3). The
selectivities observed seemed to be independent of the
Grignard reagent. Interaction with the substrate leads to
complex C, which adopts a pseudo-chair conformation that
offers an explanation for the observed regio- and enantiose-
lectivities. The product is then released and intermediate A is
regenerated to begin another cycle.
Table 3: Grignard reagent screen.
It should be kept in mind that, in the absence of an NHC,
the reaction afforded < 5% of substitution product. The
electron donation by the carbene to the Lewis acidic
magnesium atom should enhance the nucleophilicity of the
R group of the Grignard, which explain its higher reactivity.[12]
On the other hand, the hydroxy group could maintain a
tighter transition state that allows a better enantioselectivity
than L4, which lacks this hydroxy group.
Entry
R
Yield [%]
g/a[a]
ee [%][b]
1
2
3
4
5
6
7
8
9
Me
Et
nBu
iBu
88
82
83
86
97
97
90
93
88
58
83:17
76:24
77:23
84:16
79:21
88:12
82:18
69:31
68:32
73:27
70 (R)
85 (R)
85 (R)
82 (R)
80 (R)
86 (R)
86 (R)
82 (R)
84 (R)
68 (S)
tBuO(CH2)4
Ph(CH2)2
Me2C CH(CH2)2
iPr
Cy
tBu
With these encouraging results in hand, EtMgBr and
Ph(CH2)2MgBr were tested on different cinnamyl bromide
derivatives and on aliphatic substrates (Table 4).
=
With cinnamyl derivatives, the regioselectivities obtained
were independent of the aryl substituent. However, the
enantiomeric excess values increased (up to 91% ee) with
increasing steric hindrance (Table 4, entry 5) or with a more
electron-withdrawing 4-CF3C6H4 group (Table 4, entry 4). In
contrast, an electron-donating group, such as 4-MeOC6H4,
decreased the enantiomeric excess (Table 4, entry 3). The
reaction of an aryl Grignard reagent (Table 4, entry 7) only
afforded moderate results with lower reactivity, no regiose-
lectivity, and only 50% ee. In all cases involving aliphatic
substrates, the g product was obtained with high selectivities,
ranging from 86:14 to 95:5. Both the regioselectivity and
enantioselectivity followed a counter-intuitive trend, being
better with the bulkiest substituent (tBu group; Table 4,
entry 14). This is a reversal of the trend followed by copper-
catalyzed reactions, thus showing again the complementar-
ities of the two methods.
10
1
[a] Determined by H NMR spectroscopy. [b] Determined by SFC or GC
on a chiral stationary phase. Cy=cyclohexyl.
Grignard reagent used. Apart from the methyl Grignard
reagent (Table 3, entry 1), which is known for its low
reactivity and selectivity, all of the primary Grignard reagents
(Table 3, entries 2–7) offered quite good regioselectivities
(76:24 to 88:12) and good enantioselectivities (up to 86% ee).
The most remarkable result was the reaction with tBuMgBr
(Table 3, entry 9). The analogous copper-catalyzed reaction
gave a lower ee value,[1] thus showing that the copper-free
reaction may be complementary to the copper-catalyzed
established procedures.
The observed reactivity can be tentatively explained by
the following proposed mechanism (Scheme 3): First, the
imidazolidinium salt L1 is doubly deprotonated by two
molecules of the Grignard reagent to form intermediate A;
Complex B is then generated from the reaction of A with the
The procedure was extended to the formation of stereo-
genic quaternary carbon centers from their corresponding
trisubstituted allylic substrates, as a mixture of E and Z
compounds (Table 5). The reaction proceeded much faster
3348
www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 3346 –3350

Angewandte
Chemie
Table 4: Cinnamyl derivatives and aliphatic substrate scope.
9). Again, copper-catalyzed reactions have never afforded
such high g selectivities using Grignard reagents.[1]
In conclusion, a copper-free allylic alkylation reaction
with Grignard reagents was performed with up to 91% ee.
The enantioselectivity (around 85% ee) is independent of
substrate and Grignard reagent. The formation of stereogenic
quaternary centers is completely regioselective with good
enantiomeric excesses (up to 89% ee) observed with aromatic
substrates. This methodology seems complementary to the
copper-catalyzed analogue. Further work is underway to
improve the enantioselectivity with new NHCs.
Entry
R1
R2
Yield [%]
g/a[a]
76:24
ee [%][b]
1
2
3
4
5
6
7
8
9
11
12
13
14
Ph
Et
Et
Et
Et
Et
Et
82
87
59
80
88
84
66[c]
94
94
98
72
99[c]
75
85 (R)
88 (R)
73 (R)
90 (R)
91 (R)
87 (R)
50 (S)
33 (R)
54 (S)
61 (R)
78 (R)
84 (R)
80 (R)
4-MeC6H4
4-MeOC6H4
4-CF3C6H4
1-Naphthyl
2-Naphthyl
4-CF3C6H4
Me
Ph(CH2)2
BrCH2
Cy
77:23
81:19
76:24
71:29
74:26
50:50
86:14
87:13
>98:2
88:12
88:12
95:5
Ph
Ph(CH2)2
Et
Ph(CH2)2
Et
Et
Experimental Section
Cinnamyl bromide (0.5 mmol) and L1 (1 mol%) were suspended in
dry Et2O (2 mL) in a flame-dried Schlenk flask, under a nitrogen
atmosphere, and cooled to ꢀ158C. EtMgBr (3m in Et2O; 0.3 mL,
1.8 equiv) was added dropwise over 4 minutes. After conversion was
complete, the mixture is quenched by addition of a saturated solution
of NH4Cl (2 mL) and stirred at room temperature for 15 minutes. The
aqueous layer was separated and extracted with Et2O (3 ꢀ 3 mL). The
combined organic fractions were dried over Na2SO4, filtered, and
concentrated in vacuo. The residue was purified by flash column
chromatography and gave a mixture of g and a products (76:24) and
tBu
tBu
Ph(CH2)2
1
[a] Determined by H NMR spectroscopy. [b] Determined by GC or SFC
on a chiral stationary phase. [c] Conversion [%].
Table 5: Stereogenic quaternary carbon center formation.
1
85% ee (R; Table 1, entry 1). H NMR (300 MHz, CDCl3): d = 7.40–
7.26 (m, 5H), 6.49 (d, a), 6.33 (m, a), 6.06 (m, g), 5.13 (m, g ’), 3.24 (q,
J = 7.4 Hz, g), 2.29 (q, J = 7.0 Hz, a), 1.86 (m, g), 1.59 (m, a), 1.07 (t,
3H, J = 7.32 Hz, a), 0.98 ppm (t, 3H, J = 7.32 Hz, g). 13C NMR
(75 MHz, CDCl3): d = 144.4 (g), 142.2 (g), 130.9 (a), 129.9 (a), 128.4
(g), 128.3 (a), 127.6 (a), 126.7 (g), 126.1 (g), 125.9 (a), 114.0 (g), 51.7
(g), 35.1 (a), 28.3 (g), 22.5 (a), 13.7 (a), 12.1 ppm (g).
Entry R1
R2
Yield [%][a] g/a[b]
ee [%][c]
[d]
1
2
3
4
5
6
7
8
4-MeOC6H4
Et
Et
Et
Et
77
70
62:38
0
Ph[d]
>98:2 89 (R)
>98:2 39 (R)
>98:2 72 (R)
>98:2 84 (R)
>98:2 86 (R)
>98:2 64 (R)
>98:2 73 (R)
20
½aꢃD ¼ꢀ30 degcm3 gꢀ1 dmꢀ1 (c = 0.6 gcmꢀ3 CHCl3) for the mixture.
Ph[e]
27[g]
100[g]
The ee values were measured by SFC on a chiral stationary phase
(chiralcel OJ column, 2 mLminꢀ1, 200 bar, MeOH, 1% for 2 min then
2% minꢀ1): 5.56 min (R), 6.11 min (S).
Ph[f]
Ph[d]
Me2C CH(CH2)2 94
Et
Ph(CH2)2
Ph(CH2)2
=
[h]
4-CF3C6H4
55
86
65
Cy[i]
Received: January 31, 2010
Published online: April 6, 2010
tBu
[a] From the E substrate. [b] Determined by 1H NMR spectroscopy.
[c] Determined by GC or SFC on a chiral stationary phase. [d] Mixture
of E/Z isomers (8:2). [e] Cinnamyl chloride derivative as a mixture of
E/Z isomers (8:2). [f] Cinnamyl iodide derivative as a mixture of
E/Z isomers (75:25). [g] Conversion from the mixture [%]. [h] Mixture
of E/Z isomers (9:1). [i] Mixture of E/Z isomers (8:2) and ee value was
determined after epoxidation.
Keywords: allylic substitution · asymmetric catalysis ·
.
carbene ligands · Grignard reagents
[1] For reviews, see: a) A. Alexakis, C. Malan, L. Lea, K. Tissot-
Croset, D. Polet, C. Falciola, Chimia 2006, 60, 124 – 130; b) C. A.
Falciola, A. Alexakis, Eur. J. Org. Chem. 2008, 3765 – 3780; c) A.
Alexakis, J. E. Bꢁckvall, N. Krause, O. Pꢂmies, M. Diꢃguez,
Chem. Rev. 2008, 108, 2796 – 2823; d) S. R. Harutyunyan, T.
den Hartog, K. Geurts, A. J. Minnaard, B. L. Feringa, Chem.
Rev. 2008, 108, 2824 – 2852.
[2] a) F. Guillen, C. L. Winn, A. Alexakis, Tetrahedron: Asymmetry
2001, 12, 2083; b) J. Pytkowicz, S. Roland, P. Mangeney,
Tetrahedron: Asymmetry 2001, 12, 2087; c) A. Alexakis, C. L.
Winn, F. Guillen, J. Pytkowicz, S. Roland, P. Mangeney, Adv.
Synth. Catal. 2003, 345, 345 – 348.
[3] a) S. Tominaga, Y. Oi, T. Kato, D. K. An, S. Okamoto,
Tetrahedron Lett. 2004, 45, 5585 – 5588; b) S. Okamoto, S.
Tominaga, N. Saino, K. Kase, K. Shimoda, J. Organomet.
Chem. 2005, 690, 6001 – 6007.
[4] a) H. Seo, D. Hirsch-Weil, K. A. Abboud, S. Hong, J. Org. Chem.
2008, 73, 1983 – 1986; b) D. Hirsch-Weil, D. R. Snead, S. Inagaki,
H. Seo, K. A. Abboud, S. Hong, Chem. Commun. 2009, 2475 –
2477.
with the E substrate, and could be stopped with minimum
conversion of the Z substrate.[13] In fact, the Z isomer
decreased the enantiomeric excess, presumably because,
with the same facial selectivity, it affords the opposite
enantiomer.
The trend in regioselectivity observed above with the
bulkiest substrates is also followed here. Almost complete
g regioselectivity was observed and was independent of the
substrate. An exception is the substrate with an electron-
donating group, which is associated to a dramatic loss of
enantioselectivity (Table 5, entry 1), probably through an SN1
mechanistic pathway. The stereoselectivity was also quite
good, up to 89%, using aromatic compounds (Table 5,
entries 2, 4–6). The aliphatic substrates offered a slight
decrease in the enantiomeric excess (Table 5, entries 7 and
Angew. Chem. Int. Ed. 2010, 49, 3346 –3350
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3349

Communications
[5] K. B. Selim, Y. Matsumoto, K. Yamada, K. Tomioka, Angew.
[8] a) D. Martin, S. Kehrli, M. dꢄAugustin, H. Clavier, M. Mauduit,
A. Alexakis, J. Am. Chem. Soc. 2006, 128, 8416; b) H. Hꢃnon, M.
Mauduit, A. Alexakis, Angew. Chem. 2008, 120, 9262 – 9264;
Angew. Chem. Int. Ed. 2008, 47, 9122 – 9124.
[9] T. Uchida, T. Katsuki, Tetrahedron Lett. 2009, 50, 4741 – 4743.
[10] Y. Lee, A. H. Hoveyda, J. Am. Chem. Soc. 2006, 128, 15604 –
15605.
[11] Y. Lee, B. Li, A. H. Hoveyda, J. Am. Chem. Soc. 2009, 131,
11625 – 11633.
[12] S. E. Denmark, G. L. Beutner Angew. Chem. 2008, 120, 1584–
1663; Angew. Chem. Int. Ed. 2008, 47, 1560 – 1638; Angew.
Chem. Int. Ed. 2008, 47, 1560 – 1638.
Chem. 2009, 121, 8889 – 8891; Angew. Chem. Int. Ed. 2009, 48,
8733 – 8735.
[6] a) A. O. Larsen, W. Leu, C. N. Oberhuber, J. E. Campbell, A. H.
Hoveyda, J. Am. Chem. Soc. 2004, 126, 11130 – 11131; b) J. J.
Van Veldhuizen, J. E. Campbell, R. E. Giudici, A. H. Hoveyda,
J. Am. Chem. Soc. 2005, 127, 6877 – 6882; c) 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;
d) D. G. Gillingham, A. H. Hoveyda, Angew. Chem. 2007, 119,
3934 – 3938; Angew. Chem. Int. Ed. 2007, 46, 3860 – 3864; e) Y.
Lee, K. Akiyama, D. G. Gillingham, M. K. Brown, A. H.
Hoveyda, J. Am. Chem. Soc. 2008, 130, 446 – 447.
[13] For further details, see the Supporting Information.
[7] A. Alexakis, C. Malan, L. Lea, C. Benhaim, X. Fournioux,
Synlett 2001, 927 – 930.
3350
www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 3346 –3350