Development of a highly α-regioselective metal-mediated allylation reaction in aqueous media: New mechanistic proposal for the origin of α-homoallylic alcohols

Article abstract of DOI:10.1021/ja029276s

This paper described a general method to obtain α-adduct homoallylic alcohols using indium, zinc, and tin in water. A new mechanism was proposed to account for the formation of these synthetically difficult-to-obtain molecules. Generally, this method can

Full text of DOI:10.1021/ja029276s

Published on Web 02/12/2003
Development of a Highly r-Regioselective Metal-Mediated
Allylation Reaction in Aqueous Media: New Mechanistic
Proposal for the Origin of r-Homoallylic Alcohols
Kui-Thong Tan,† Shu-Sin Chng,† Hin-Soon Cheng,‡ and Teck-Peng Loh*,†
Contribution from the Department of Chemistry, National UniVersity of Singapore,
3 Science DriVe 3, Singapore 117543, and Institute of Chemical and Engineering Sciences,
Ayer Rajah Crescent, Block 28, Unit #02-08, Singapore 139959
Received November 8, 2002 ; E-mail: chmlohtp@nus.edu.sg
Abstract: This paper described a general method to obtain R-adduct homoallylic alcohols using indium,
zinc, and tin in water. A new mechanism was proposed to account for the formation of these synthetically
difficult-to-obtain molecules. Generally, this method can be performed with a wide range of aldehydes and
allylic halides with just 6 equiv of water added, giving the R-adduct in high selectivities. To account for the
origin of the R-homoallylic alcohol, the reaction mechanism was carefully studied using 1H NMR, a crossover
experiment, and the inversion stereochemical studies of 22â γ-adduct homoallylic sterol to the 22R R-adduct
homoallylic sterol. From the results of mechanism studies, it is possible that two mechanism pathways
coexisted in the metal-mediated R-regioselective allylation. The metal salts formed from the metal-mediated
allylation can catalyze the γ-adduct to undergo a bond cleavage to generate the parent aldehyde in situ
followed by a concerted rearrangement, perhaps a retroene reaction followed by a 2-oxonia[3,3]-sigmatropic
rearrangement to furnish the R-adduct. The R-adduct can also be synthesized via the formation of an
oxonium ion intermediate between the γ-adduct and the unreacted aldehyde. The proposed mechanisms
were further supported by experimental findings from the addition of InBr3 to γ-adduct under similar
conditions.
Introduction
stereocenters can be assembled in a single step. This is highly
efficient in terms of atom economics,4 as all of the carbons form
Being important building blocks and versatile synthons,
homoallylic alcohols are highly featured in the organic syntheses
of many biological active molecules such as macrolides,
polyhydroxylated natural products, and polyether antibiotics.1
Among the existing means to construct these synthetically and
biologically important molecules, metal-mediated allylation2 is
one of the easiest and most convenient. Furthermore, with the
intensive and further development of catalytic stereo- and
enantioselective synthesis of homoallylic alcohol,3 one or two
the scaffold of the desired molecules.
However, many metal-mediated allylations involving allyl-
magnesium and allyllithium are often very difficult to handle
because the reactions have to be performed under strictly
anhydrous, oxygen-free, and low temperature conditions. The
metal-mediated Barbier type allylation reaction has provided
an alternative to overcome these problems by adding reagents
and metal together at room temperature using environmentally
benign solvents such as water and/or ethanol. Metals such as
indium,5 zinc,6 and tin7 are always used in typical metal-
mediated Barbier type allylations with carbonyl compounds and
allylic halides. The use of water as a solvent to carry out organic
† National University of Singapore.
‡ Institute of Chemical and Engineering Sciences.
(1) (a) Nicolaou, K. C.; Kim, D. W.; Baati, R. Angew. Chem., Int. Ed. 2002,
41, 3701. (b) Nicolaou, K. C.; Ninkovic, S.; Sarabia, F.; Vourloumis, D.;
He, Y.; Valllberg, H.; Finlay, M. R. V.; Yang, Z. J. Am. Chem. Soc. 1997,
119, 7974. (c) Bartlett, P. A. Tetrahedron 1980, 36, 3. (d) Paterson, I.;
Mansuri, M. M. Tetrahedron 1985, 41, 3569. (e) Germay, O.; Kumar, N.;
Thomas, E. J. Tetrahedron Lett. 2001, 42, 4969. (f) Romo, D.; Meyer, S.
D.; Johnson, D. D.; Schreiber, S. L. J. Am. Chem. Soc. 1993, 115, 9345.
(2) (a) Roush, W. R. In ComprehensiVe Organic Synthesis; Trost, B. M.,
Fleming, I., Heathcock, C. H. Pergamon Press: Oxford, 1991; Vol. 2, p 1.
(b) Yamamoto, Y.; Asao, N. Chem. ReV. 1993, 93, 2207.
(4) (a) Trost, B. M. Acc. Chem. Res. 2002, 35, 695. (b) Trost, B. M. Angew.
Chem., Int. Ed. Engl. 1995, 34, 259. (c) Trost, B. M. Science 1991, 254,
1471.
(5) For reviews and examples, see: (a) Li, C. J.; Chan, T. H. Organic Reactions
In Aqueous Media; John Wiley & Sons: New York, 1997. (b) Li, C. J.;
Chan, T. H. Tetrahedron 1999, 55, 11149. (c) Li, C. J.; Chan, T. H.
Tetrahedron Lett. 1991, 32, 7017. (d) Wang, R. B.; Lim, C. M.; Tan, C.
H.; Lim, B. K.; Sim, K. Y.; Loh, T. P. Tetrahedron: Asymmetry 1995, 6,
1825. (e) Paquette, L. A.; Mittzel, T. M. Tetrahedron Lett. 1995, 36, 6863.
(6) (a) Majee, A.; Das, A. R.; Ranu, B. C. Indian J. Chem., Sect. B 1998, 37,
731. (b) Killinger, T. A.; Boughton, N. A.; Wolinsky, J. J. Organomet.
Chem. 1977, 124, 131. (c) Luche, J. L.; Damiano, J. C. J. Am. Chem. Soc.
1980, 102, 7926. (d) Sprich, J. D.; Lewandos, G. S. Inorg. Chim. Acta
1983, 76, L241. (e) Patrier, C.; Luche, J. L. J. Org. Chem. 1985, 50, 910.
(f) Petrier, C.; Einhorn, J.; Luche, J. L. Tetrahedron Lett. 1985, 26, 1449.
(g) Luche, J. L. J. Organomet. Chem. 1987, 322, 177. (h) Wilson, S. R.;
Guazzaroni, M. E. J. Org. Chem. 1989, 54, 3087 (i) Shono, T.; Ishifune,
M.; Kashimura, S. Chem. Lett. 1990, 449. (j) Lu, W. S.; Chan, T. H. J.
Org. Chem. 2000, 65, 917.
(3) For some reviews and examples of stereoselective synthesis of homoallylic
alcohols, see: (a) Bode, J. W.; Gauthier, D. R.; Carreira, E. M. Chem.
Commun. 2001, 2560. (b) Yanagisawa, A.; Nakashima, H.; Ishiba, A.;
Yamamoto, H. J. Am. Chem. Soc. 1996, 118, 4723. (c) Roush, W. R.; Ando,
K.; Powers, D. B.; Palkowitz, A. D.; Halterman, R. L. J. Am. Chem. Soc.
1990, 112, 6339. (d) Faller, J. W.; Sams, D. W. I.; Liu, X. J. Am. Chem.
Soc. 1996, 118, 1217. (e) Yanagisawa, A. In ComprehensiVe Asymmetric
Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.; Springer-
Verlag: Berlin, 1999; Vol. 2, p 965. (f) Bach, T. Angew. Chem., Int. Ed.
Engl. 1994, 33, 417. (g) Hoveyda, A. H.; Morken, J. P. Angew. Chem.,
Int. Ed. Engl. 1996, 35, 1262.
9
2958
J. AM. CHEM. SOC. 2003, 125, 2958-2963
10.1021/ja029276s CCC: $25.00 © 2003 American Chemical Society

Development of a Metal-Mediated Allylation Reaction
A R T I C L E S
Scheme 1
Table 1. The Effect of Solvent on the Indium-Mediated Allylation
of Cyclohexanecarboxaldehyde with Crotyl Bromidea
d
entry
solvent
DMF
H2O
H2O
H2O
THF
CH2Cl2
THF/H2O
CH2Cl2/H2O
DMF/H2O
CH3CH2OH
neat
amount (equiv)
time (h)
yield % (R:γ)c
1b
2
6
72
72
24
72
72
24
72
24
72
72
24
65 (0:100)
90 (0:100)
85 (99:1)
87 (86:14)
20 (0:100)
reactions offers practical convenience as it alleviates the need
to handle flammable and anhydrous organic solvents. It also
simplifies the tedious protection-deprotection sequences for
molecules containing acidic protons, which leads to an increase
in overall synthetic efficiency.8
2 mL
6
6
6
6
6:6
6:6
6:6
6
3b
4
5
6
7
8
e
-
95 (0:100)
68 (>99:1)
80 (0:100)
97 (0:100)
Although this metal-mediated Barbier type allylation reaction
has been found to be highly regio- and stereoselective, one
severe limitation inherent in this reaction is the difficulty in
obtaining the R-adduct when allylic metals are employed. Apart
from some exceptions, the metal-mediated allylation of carbonyl
compounds with R-substituted allylic halide occurs regioselec-
tively at the γ-position (Scheme 1). Therefore, the direct
synthesis of R-adduct using allylic halide, metal, and carbonyl
compound in aqueous media has become one of the major
synthetic challenges to organic chemists.
In our previous studies, the reaction of aldehydes with
different allylic bromides and indium powder in water was
investigated.9 To our surprise, when the amount of water added
was decreased and the reaction time was lengthened, we
consistently detected a significant amount of the R-adduct.10
This interesting result prompted us to study the regioselectivity
of the reaction (R-adduct versus γ-adduct). After the successful
development and understanding of R-regioselective indium-
mediated allylation in water, we extended this reaction to
different metals. In this paper, we report a general strategy to
obtain R-homoallylic alcohols using three different metals,
indium, tin, and zinc, in water. Through detailed mechanism
studies described, a mechanism was proposed to account for
the formation of this R-adduct.
9b
10
11
c
-
a All reactions were performed with aldehyde (1 mmol), crotyl bromide
(1.2 mmol), and indium (1.5 mmol) at room temperature unless otherwise
noted. b The reactions were carried out at room temperature for 12 h,
followed by heating to 40 °C. c Determined by 1H NMR. d Combined yield.
e Neither the γ- nor the R-adduct was observed.
carboxyaldehyde with crotyl bromide under solvent-free condi-
tions (entry 11); however, neither the γ- nor the R-adduct was
acquired.11
It is important to note that the amount of water added is
crucial for the R-selectivity. The highest R-regioselectivity was
observed when the amount of water used was 6 equiv. When
the reaction was carried out in the presence of 12 equiv of water,
no R-adduct was detected even after the reaction was stirred
up to 72 h. Furthermore, regioselectivity increased significantly
when a higher temperature was applied (entry 3).
The results were pretty surprising, because R-homoallylic
alcohols can be obtained in very high selectivity and yield in
this condition. Previously reported results12 showed that R-ad-
ducts could be obtained directly only when very bulky aldehyde
and allylic halide were applied in the reaction. The addition of
only a small amount of water has completely altered the reaction
pathway and given virtually different regioisomers.
Results and Experiments
With the unveiling of water (6 equiv) as an important criterion
for obtaining the R-adduct, we extended the reaction to a variety
of substrates, and the results are as shown in Table 2. In all
cases, the R-adducts were obtained with high selectivities in
moderate to good yields.
Syntheses of r-Homoallylic Alcohols by Indium Metal. The
indium-mediated R-regioselective allylation was first explored
and tested with different solvents that were believed to give
high yield. Among the various solvents examined, the reactions
carried out in DMF, THF, ethanol, and water (2 mL) afforded
only the γ-homoallylic alcohols (Table 1, entries 1, 2, 5, and
10). The addition of equal amounts of water to THF or DMF
did not improve the regioselectivity. However, water (6 equiv)
and water/dichloromethane (6 equiv/6 equiv) distinguished
themselves by exhibiting excellent R-selectivity (entries 4 and
8). We also tried to perform the reaction of cyclohexane
It was found that the reaction of benzaldehyde with crotyl
bromide gave the R-regio isomer almost exclusively (entry 1).
Using 6 equiv of water as the solvent, we obtained a totally
different regioisomer as compared to previously published
results using water or other solvents. Furthermore, the reaction
of hydrocinnamaldehyde and hexanal with crotyl bromide also
gave very good R-regioselectivity (entries 3 and 4). The results
shown in Table 2 suggested that indium-mediated allylation can
give the homoallylic alcohol adducts regardless of the substit-
uents. Accordingly, an increase in the steric bulkiness of the
aldehyde and bromide gave no significant improvement in the
R-regioselectivity. When cinnamyl bromide was used, the
R-adducts were obtained with excellent selectivities and good
yields (entries 5-8). Even relatively unreactive ethyl 4-bromo
crotonate can be used to synthesize the desired R-adduct (entries
(7) For examples on Barbier-type tin allylation, see: (a) Nokami, J.; Otera, J.;
Sudo, T.; Okawara, R. Organometallics 1983, 2, 191. (b) Nokami, J.;
Wakabayashi, S.; Okawara, R. Chem. Lett. 1984, 869. (c) Zhou, J. Y.; Chen,
Z. G.; Wu, S. H. Chem. Commun. 1994, 2783.
(8) (a) Chan, T. H.; Li, C. J. Chem. Commun. 1992, 747. (b) Kim, E.; Gordon,
D. M.; Schmid, W.; Whitesides, G. M. J. Org. Chem. 1993, 58, 5500. (c)
Loh, T. P.; Cao, G. Q.; Pei, J. Tetrahedron Lett. 1998, 39, 1457. (d) Loh,
T. P.; Hu, Q. Y.; Vittal, J. J. Synlett 2000, 523. (e) Loh, T. P.; Lye, P. L.
Tetrahedron Lett. 2001, 42, 3511.
(9) (a) Loh, T. P.; Tan, K. T.; Yang, J. Y.; Xiang, C. L. Tetrahedron Lett.
2001, 42, 8701. (b) Loh, T. P.; Tan, K. T.; Hu, Q. Y. Tetrahedron Lett.
2001, 42, 8705.
(10) For some isolated cases that produce R-adducts from an indium-mediated
reaction, see: (a) Araki, S.; Ito, H.; Katsumura, N.; Butsugan, Y. J.
Organomet. Chem. 1989, 369, 291. (b) Araki, S.; Katsumura, N.; Butsugan,
Y. J. Organomet. Chem. 1991, 415, 7.
(11) For an indium-mediated reaction under neat condition, see: Yi, X. H.;
Heberman, J. X.; Li, C. J. Synth. Commun. 1998, 28, 2999.
(12) Isaac, M. B.; Chan, T. H. Tetrahedron Lett. 1995, 36, 8957.
9
J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003 2959

A R T I C L E S
Tan et al.
Table 2. Indium-Mediated Allylation of Aldehydes with Different
Table 3. Tin-Mediated Crotylation of Aldehydes
Bromidesa
entry
R
R
1
conditiona,b
yield (%)d
γ(syn:anti):R(E:Z)c
1
2
3
4
5
6
7
8e
Ph
n-C5H11
PhCH2CH2 Me
c-C6H11
t-Bu
n-C5H11
c-C6H11
Ph
Me
Me
A
A
A
A
A
B
B
B
83
81
82
78
42
60
54
80
45(50:50):55(75:25)
20(61:39):80(55:45)
29(37:63):71(53:47)
5(23:77):95(67:33)
1(-:-):99(83:17)
5(-:-):95(100:0)
1(-:-):99(100:0)
1(-:-):99(100:0)
c
entry
R
R
R
time (h)
yield % (R:γ)b (E/Z)b
1
2
1
2
3
4
5
6
7
8
Ph
Me
H
36
24
36
18
72
72
85
160
96
192
15
60 (99:1)(55/45)
85 (99:1)(70/30)
75 (98:2)(65/35)
67 (97:3)(55/45)
66 (98:2)(E)
73 (96:4)(98/2)
71 (99:1)(90/10)
50 (99:1)(95/5)
80 (80:20)(80/20)f
56 (85:15)(90/10)f
30 (95:5)
c-C6H11
n-C5H11
PhCH2CH2
Ph
c-C6H11
n-C5H11
PhCH2CH2
c-C6H11
Me
Me
Me
Me
Ph
Ph
Ph
H
H
H
H
H
H
H
H
H
Me
Me
Ph
Ph
Ph
Ph
a Condition A: The reactions were performed with aldehyde (1 mmol),
crotyl bromide (1.2 mmol), and tin (1.5 mmol) in water (0.108 mL) and
CH2Cl2 (0.385 mL) at room temperature for 48 h unless otherwise noted.
b Condition B: The reactions were performed with aldehyde (1 mmol),
cinnamyl bromide (1.2 mmol), and tin (1.5 mmol) in water (1 mL) at room
temperature for 24 h unless otherwise noted. c Determined by 1H NMR and
13C NMR. d Combined yield. e The reaction was stirred for 3 days with 3
mmol of cinnamyl bromide.
9e
10d,e
11e
CO2Et
CO2Et
Me
c-C6H6
Me
a All reactions were performed with aldehyde (1 mmol), bromide (1.2
mmol), and indium (1.5 mmol) with water (0.108 mL) at room temperature
for 12 h, followed by heating to 40 °C, unless otherwise noted. b Determined
by 1H and 13C NMR. c Total yield. d Because of the highly volatility of the
aldehyde, an excess of acetyl aldehyde was added. e Two equivalents of
water was used. f Z isomer was isolated as lactone.
Table 4. Zinc-Mediated Crotylation of Aldehydesa
9 and 10). However, the reaction of aldehyde with prenyl
bromide only afforded a moderately low yield (entry 11).13 In
short, this type of reaction can be applied generally irrespective
of the allylic bromides and aldehydes used.
To study the scope and limitations of the reaction, two other
metals, zinc and tin, which are known to work well in water,
were used to test the feasibility of the conditions. We used the
same condition as that used for indium to carry out the reaction.
With little surprise, zinc and tin also gave us similar results.
However, some modifications to the conditions were necessary
to give higher selectivity and yield.
entry
R
yield (%)c
γ(syn:anti):R(E:Z)b
1
2
3
4
5
t-Bu
34
55
66
58
50
55(10:90):45(62:38)
10(88:12):90(54:46)
5(32:68):95(73:27)
6(35:65):94(60:40)
3(50:50):97(77:23)
n-C5H11
c-C6H11
PhCH2CH2
Ph
a All reactions were performed with aldehyde (1 mmol), bromide (1.2
mmol), and zinc (1.5 mmol) in water (0.036 mL) and CH2Cl2 (0.128 mL)
at room temperature for 120 h. b Determined by H NMR and 13C NMR.
1
c Combined yield.
Tin- and Zinc-Mediated r-Regioselective Allylation. For
tin-mediated allylation, aliphatic aldehydes such as hexanal,
cyclohexanecarboxaldehyde, and trimethylacetaldehyde provided
much higher R-regioselectivities in moderate to good yields
(Table 3, entries 2, 4, and 5, respectively). Generally, an increase
in the steric bulkiness of the aldehyde from hexanal to
trimethylacetaldehyde resulted in significant improvements of
the R-regioselectivity. These results suggested that tin-mediated
allylation could also give R-homoallylic alcohol adducts regard-
less of the substituents on the aldehyde. Strikingly, when crotyl
bromide was replaced by cinnamyl bromide, no organic co-
solvent was required to run the reaction (entries 6-8).14 The
reaction could also be performed in aqueous media (1 mL). For
the case of cinnamyl bromide, the reaction did not exhibit any
significant selectivity due to steric bulkiness.
excellent R-regioselectivities were observed when a mixture of
water (2 equiv) and dichloromethane (2 equiv) was used as the
solvent. The results were summarized in Table 4.
Such interesting results displayed by the indium-, tin-, and
zinc-mediated allylation reactions prompted us to investigate
the mechanism in greater detail. To the best of our knowledge,
all of the reported procedures for highly R-regioselective metal-
mediated allylation reactions in water or an organic solvent are
usually coupled with additives, such as Lewis acids. Reported
examples include the addition of a stoichiometric amount of
AlCl3 to a Grignard reaction16 and the reaction of crotyl
tributylstannanes with aldehydes in the presence of a Lewis
acid.17 Furthermore, highly R-selective allylation can also be
achieved by the reaction with aldehydes of allylic barium,18
allylic cerium reagents,19 and Me3SiCl/NaI/H2O.20 There are also
a few special cases where the magnesium and indium-mediated
allylation reaction produces R-adducts. However, these isolated
cases provided insufficient data, and further investigation was
called for.
Our investigations using zinc revealed that allylic zinc is much
less reactive in aqueous media.15 For all of the aldehydes used,
none reached completion, which accounted for the moderate
yields obtained. In the zinc system, however, it was found that
(13) A small amount of THF (0.3 equiv) must be added to suppress the formation
of side product to afford 78% yield with an R to γ ratio of 70/30. This is
because the corresponding R-adduct undergoes an oxonium-ene cyclization
with aldehyde immediately once the R-adduct was synthesized. For
reference, see: Loh, T. P.; Hu, Q. Y.; Tan, K. T.; Cheng, H. S. Org. Lett.
2001, 3, 2669.
(16) Yamamoto, Y.; Maruyama, K. J. Org. Chem. 1983, 48, 307.
(17) (a) Keck, G. E.; Abbott, D. E.; Boden, E. P.; Enholm, E. J. Tetrahedron
Lett. 1984, 25, 3927. (b) Yamamoto, Y.; Maeda, N.; Maruyama, K. Chem.
Commun. 1983, 48, 1564. (c) Gambaro, A.; Gains, P.; Marton, D.; Peruzzo,
V.; Tagliavini, G. J. Organomet. Chem. 1982, 231, 307.
(18) (a) Yanagisawa, A.; Habaue, S.; Yamamoto, H. J. Am. Chem. Soc. 1991,
113, 8955. (b) Yanagisawa, A.; Habaue, S.; Yasue, K.; Yamamoto, H. J.
Am. Chem. Soc. 1994, 116, 6130.
(14) We also tried to add 6 equiv of water to the reaction; yet, a lower yield
was obtained. For example, reaction of hexanal aldehyde with cinnamyl
bromide in 6 equiv of water for 24 h afforded only 22% yield with no
change in regioselectivity.
(19) (a) Guo, B. S.; Doubleday, W.; Cohen, T. J. Am. Chem. Soc. 1984, 106,
4710. (b) Cohen, T.; Matz, J. R. J. Am. Chem. Soc. 1980, 102, 6900.
(20) Kanagawa, K.; Nishiyama, Y.; Ishii, Y. J. Org. Chem. 1992, 57, 6988.
(15) The zinc powder was used directly from the bottle purchased from Merck.
9
2960 J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003

Development of a Metal-Mediated Allylation Reaction
A R T I C L E S
Figure 1. 1H NMR study for indium-mediated allylation.
Scheme 2
Scheme 5. Crossover Experiment
Scheme 3
Scheme 4
By the fourth hour, the R-adduct started to appear, as can be
seen in the 1H NMR spectrum, the peaks around δ 5.5 and 3.3
ppm being assigned as the internal double bond and the
R-hydroxyl protons of the R-adduct. If the reaction was stirred
further, after 24 h, all of the γ-adducts were arranged to the
thermodynamically more stable R-adduct with complete con-
sumption of the aldehyde, as indicated by the disappearance of
the peaks at δ 5.8, 5.1, and 3.1 ppm.
1
A similar H NMR study (Figure 2) was carried out as well
with benzaldehyde and crotyl bromide using tin and zinc as
the metals. The H NMR spectra for tin-mediated allylation
Mechanism Study of Metal-Mediated r-Regioselective
Allylations. Three mechanisms have been proposed to account
for the high R-selectivities observed for the known systems.
They can generally be grouped into three categories: (i) The
external activator or Lewis acid is involved in the formation of
a six-membered transition state to afford the R-adduct directly
as shown in Scheme 2; (ii) direct attack of the allylic metal on
the R-position via a four-membered transition state (Scheme
3);21,22 and (iii) transmetalation of the allylic metal species with
its external activator or Lewis acid to generate a new allylic
species which undergoes a γ-attack to afford the R-adduct
(Scheme 4).
1H NMR Study for Indium- and Tin-Mediated Allylation.
To understand the origin of these linear homoallylic alcohol
adducts, the progress of the indium-mediated allylation of
cyclohexanecarboxaldehyde with crotyl bromide was monitored
by 1H NMR after workup at various intervals. The spectra were
taken at 2, 4, and 24 h, respectively (Figure 1).
1
worked up at 2, 4, 8, 24, and 48 h, respectively, were shown.
It was found that the reaction proceeded smoothly to give the
kinetically favored γ-homoallylic alcohol initially, which slowly
converted to the thermodynamic R-adduct. Similar observations
were obtained with the zinc system. These results of the NMR
studies suggested that the mechanisms for indium-, tin-, and
zinc-mediated allylation reactions are likely to be similar, if not
the same.
Crossover Experiment. A crossover experiment was con-
ducted using indium (Scheme 5). After the reaction, column
chromatography revealed the crossover products 1 and 2 in 10
and 11% yields, respectively. This proved that the rearrangement
of this γ-adduct to its isomer R-adduct is an intermolecular
process that may involve the cleavage of the γ-adduct to form
the aldehyde substituent and an allyl fragment.
Stereochemical Investigation of the Reaction. In addition
to the NMR studies and crossover experiment, the indium, tin,
and zinc systems were tested using a steroidal aldehyde to
investigate the stereochemistry of the reaction (Table 5). The
initial attempt made using indium-mediated allylation of the
steroidal aldehyde 3 with cinnamyl bromide afforded the 22â
γ-homoallylic alcohol 4 which rearranged to the 22R R-homo-
allylic alcohol 5 with complete inversion of stereochemistry (the
It was observed that the reaction proceeded very rapidly to
afford the kinetically favored γ-homoallylic alcohol, which
slowly converted to the thermodynamic R-homoallylic alcohol
(24 h).23 By the second hour, the γ-homoallylic alcohol adduct
was obtained as the sole product with no sign of any R-isomer.
(21) With hindered ketones such as di-tert-butyl ketone or di-iso-propyl ketone,
the R-adducts might be formed from fission of the C-C bond or directly
from the ketone and the crotyl Grignard via a four-center transition state,
see: Benkeser, R. A.; Siklosi, M. P.; Mozdzen, E. C. J. Am. Chem. Soc.
1978, 100, 2134.
(22) Besides obvious steric effects, regioselectivity in the carbanionic addition
is also governed by the relative electron densities and negative charge at
the R- and γ-position in the highest occupied molecular orbital (HOMO)
of the ambient allylic anion, see: Gompper, R.; Wagner, H. U. Angew.
Chem., Int. Ed. Engl. 1976, 15, 321.
(23) Process which converts readily available kinetic products to their less readily
accessible thermodynamic isomers, see: (a) Barbot, F.; Miginiac, P.
Tetrahedron Lett. 1975, 3829. (b) Gedye, R. N.; Arora, P.; Khalil, A. H.
Can. J. Chem. 1975, 53, 1943. (c) Tatsuta, K.; Tamura, T.; Mase, T.
Tetrahedron Lett. 1999, 40, 1925. For an example of a γ-adduct rearranging
to its R-adduct, see: (d) Hong, B. C.; Hong, J. H.; Tsai, Y. C. Angew.
Chem., Int. Ed. 1998, 37, 468.
9
J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003 2961

A R T I C L E S
Tan et al.
Figure 2. 1H NMR study for tin crotylation of benzaldehyde.
Table 5. Stereochemical Study of Metal-Mediated
R-Regioselectivity Allylation
Scheme 6. Proposed Mechanism for Rearrangement
yielda
(%)
SM recovered
(%)
entry metal
R
time (h)
4(syn:anti):5(E:Z)
1
In
Sn
Zn
Ph
Me
Me
48
96
120
54
50
40
30
5
20
1 (-: -):99 (99:1)
52(9:91):48(99:1)
78(35:65):22(99:1)
2a
3a
a Convergent yield.
relative stereochemistry was determined by single-crystal X-ray
diffraction analysis).24 In other words, the allyl fragment re-
attaches to the steroid in an anti-Cram manner, which excludes
the possibility of allyl anion re-addition to the aldehyde.
Unexpectedly, for the tin-mediated reaction, no R-allylation
adduct was observed in the reaction mixture even after 96 h. A
1H NMR study of the tin-steroidal system revealed that the
steroidal aldehyde was totally depleted after just 24 h. We
believed that the absence of the aldehyde might be the key to
the negative result. Indeed, using just a slight excess (0.1 mmol)
of steroidal aldehyde in the reaction, we successfully obtained
the desired R-homoallylic alcohol along with the γ-adducts.
Furthermore, the R-homoallylic alcohol was found with com-
plete inversion of stereochemistry at C22 with respect to the
γ-products. Such anti-Cram allylation R-adducts could not
possibly be obtained if these adducts were formed via the direct
attack of the allylic species on the aldehyde. The reaction of
steroidal aldehyde with crotyl bromide in the presence of zinc
metal also gave the desired R-homoallylic alcohol in a mixture
of products. Similarly, complete inversion of configuration at
C22 was observed.
studies, a new mechanism (Scheme 6) was proposed for this
metal-mediated allylation reaction in aqueous media. In op-
position to previously proposed mechanisms, our investigations
suggested that the reaction took place in two stages. The first
stage involves the metal-mediated allylation of the aldehyde to
form the γ-adduct 6. This initially formed the γ-adduct, which
then underwent a retroene reaction,25 followed by a 2-oxonia
[3,3]-sigmatropic rearrangement in the presence of the aldehyde
7 to give the R-adduct 14 in the second stage.26 The steps in
the second stage are likely to be facilitated by the presence of
residual metal salts generated from the first stage of the reaction.
In addition, the aldehyde required in the second stage could
either be unreacted aldehyde in the reaction27 (pathway B) or
Proposed Mechanism for Metal-Mediated Allylation. In
view of the similarity between the indium-, tin-, and zinc-
mediated allylation reactions, the mechanisms for the formation
of the R-homoallylic alcohols are believed to be the same. On
(25) For a recent review, see: Ripoll, J. L.; Valle´e, Y. Synthesis 1993, 659.
(26) An alternative mechanism for the formation of R-adduct via an intramo-
lecular [1,3]-sigmatropic rearrangement, contradicting the results from the
crossover experiment, which points towards an intermolecular rearrange-
ment. For reference, see: Wilson, S. T. Org. React. 1993, 43, 93.
1
the basis of the results of the H NMR and stereochemical
(24) The single-crystal X-ray diffraction analysis data have been published,
see: ref 10b and Loh, T. P.; Hu, Q. Y.; Ma, L. T. Org. Lett. 2002, 4, 2389.
9
2962 J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003

Development of a Metal-Mediated Allylation Reaction
A R T I C L E S
Conclusion
be aldehyde regenerated in situ from metal salt-catalyzed bond
cleavage of the γ-adduct (pathway A).28
In conclusion, a general R-regioselective metal-mediated
allylation reaction has been developed. We have demonstrated
that indium, tin, and zinc can be used to carry out this reaction,
giving moderate to good yields and selectivities. The mechanism
proposed is supported by the experimental results from 1H NMR,
crossover experiment, stereochemical studies using steroidal
aldehyde, and rearrangement investigations involving indium
metal salts. These studies have led to a new mechanistic proposal
for the origin of the high R-selectivity. This study showed that
the origin of the R-homoallylic alcohols from many of the
reported metal-mediated allylation reactions with or without
activators warrants further studies to clarify their mechanisms.
The proposed mechanism clearly illustrates the role of water
in the reaction. Besides being a good solvent for the metal-
mediated allylation of the aldehyde in the first stage, 6 equiv
of water is necessary in the second stage of the reaction for the
optimal formation of the crucial oxonium intermediate 10 and
the subsequent quenching of the rearranged intermediate 11.
The use of 12 equiv of water is believed to completely suppress
the formation of oxonium intermediate 10, causing the reaction
to truncate in the first stage and resulting only in the γ-adduct.
Because of the great importance of water in the second stage,
reactions carried out in other solvents did not give any desired
R-adducts (see Table 1, entries 1, 5, 6, and 10). In addition, the
use of water as a cosolvent with DMF and THF (entries 7 and
9) did not reverse the selectivity as it did with dichloromethane
(entry 8). It is thus believed that the DMF or THF actually
complexes strongly with indium salts essentially for the forma-
tion of intermediates 8-13, which then led again to the
truncation of the reaction in the first stage. The absence of the
second stage in the presence of DMF and THF is further
Acknowledgment. We thank the National University of
Singapore for generous financial support.
Supporting Information Available: Complete experimental
details, including characterization of all new compounds (PDF).
This material is available free of charge via the Internet at
http://pubs.acs.org.
29
supported by experimental findings from the addition of InBr3
to the γ-adduct under similar conditions.30
JA029276S
(27) (a) Nokami, J.; Yoshizane, K.; Matsuura, H.; Sumida, S. J. Am. Chem.
Soc. 1998, 120, 6609. (b) Nokami, J.; Anthony, L.; Sumida, S. Chem.-
Eur. J. 2000, 6, 2909. (c) Nokami, J.; Ohga, M.; Nakamoto, H.; Matsubara,
T.; Hussain, I.; Kataoka, K. J. Am. Chem. Soc. 2001, 123, 9168. (d) Loh,
T. P.; Hu, Q. Y.; Chok, Y. K.; Tan, K. T. Tetrahedron Lett. 2001, 42,
9277. (e) Loh, T. P.; Lee, C. L. K.; Tan, K. T. Org. Lett. 2002, 4, 2985.
(28) Loh, T. P.; Tan, K. T.; Hu, Q. Y. Angew. Chem., Int. Ed. 2001, 40, 2921.
(29) We found that In(I)Br can catalyze the rearrangement of γ-adducts to their
corresponding R-adducts too.
(30) We tried to effect the rearrangement of the γ-adduct (formed from
cyclohexane carboxyaldehyde and crotyl bromide) to its R-adduct by adding
0.5 equiv of InBr3 to the γ-adduct in various solvents, including 6 equiv
of water, DMF, and THF. Results revealed that after 72 h at room
temperature, only the mixture in 6 equiv of water showed signs of
rearrangement (γ: R ≈ 78:22), while those in DMF and THF showed no
progress. In fact, when 12 equiv of water was used, no rearrangement was
observed as well.
9
J. AM. CHEM. SOC. VOL. 125, NO. 10, 2003 2963