Catalytic Regiodivergent Kinetic Resolution of Allylic Epoxides: A New Entry to Allylic and Homoallylic Alcohols with High Optical Purity

Article abstract of DOI:10.1021/jo035674h

A novel regiodivergent kinetic resolution of a series of allylic epoxides with alkylzinc reagents is described. Results demonstrate the potential of chiral copper-phosphoramidite catalysts for enantiomer differentiation of allylic epoxides, allowing a chiral catalyst-stereoregulated synthesis of cyclic allylic and homoallylic alcohols with high optical purities.

Full text of DOI:10.1021/jo035674h

Ca ta lytic Regiod iver gen t Kin etic Resolu tion of Allylic Ep oxid es: A
New En tr y to Allylic a n d Hom oa llylic Alcoh ols w ith High Op tica l
P u r ity
Mauro Pineschi,* Federica Del Moro, Paolo Crotti, Valeria Di Bussolo, and Franco Macchia
Dipartimento di Chimica Bioorganica e Biofarmacia, Universita` di Pisa,
Via Bonanno 33, 56126 Pisa, Italy
pineschi@farm.unipi.it
Received November 13, 2003
A novel regiodivergent kinetic resolution of a series of allylic epoxides with alkylzinc reagents is
described. Results demonstrate the potential of chiral copper-phosphoramidite catalysts for
enantiomer differentiation of allylic epoxides, allowing a chiral catalyst-stereoregulated synthesis
of cyclic allylic and homoallylic alcohols with high optical purities.
In tr od u ction
been reported.5 However, RKR-concerning catalytic pro-
cesses that form carbon-carbon bonds are very rare. The
only two examples are the intramolecular cyclopropana-
tion of racemic allylic diazoacetates catalyzed by chiral
rhodium complexes6 and the zirconocene-catalyzed ad-
dition of EtMgBr to racemic dihydrofurans.7 In the
context of RKR processes forming carbon-carbon bonds,
we recently discovered a highly stereocontrolled trans-
formation of a racemic mixture by an organometallic
reagent and a chiral catalyst to give separable regioiso-
meric products.8 Our method was based on the addition
of dialkylzinc reagents (R ) Me or Et) to racemic cyclic
allylic epoxides in the presence of copper complexes of
nonracemic phosphoramidite (R,R,R)-1 (Scheme 1).9
This chiral catalyst was able to discriminate between
the enantiomers of semirigid allylic epoxides to give
separable regioisomers. The reaction showed a striking
complementary enantiomer-dependent regioselectivity.
The allylic alcohols (derived from an SN2′ addition on the
faster-reacting enantiomer of the cyclic allylic epoxide)
and above all the homoallylic alcohols (derived from an
SN2 addition on the slower-reacting enantiomer) were
obtained with good to excellent ee’s. We now wish to give
a full account of this work.
The development of efficient asymmetric catalysts is
a rational, useful way to improve the preparation of chiral
compounds with high optical purities. Kinetic resolution
of a racemic mixture is a well-established method for the
preparation of optically active compounds.1 However, the
major drawback with this approach is that a maximum
of only half of the racemic starting material is converted
into nonracemic products. Parallel kinetic resolution
(PKR) is an interesting strategy recently introduced, in
which both enantiomers of a racemate can be converted
into different products.2 This conceptual variation often
requires the use of two different stoichiometric chiral
reagents in parallel.3 Conceptually similar to PKR is the
so-called regiodivergent kinetic resolution (RKR), which
is a process in which a single chiral catalyst or reagent
reacts with a racemic substrate to form regioisomers
possessing opposite configurations on the newly formed
stereogenic centers. To date, there are only a few ex-
amples that demonstrate this concept. For example, RKR
reactions under nonstoichiometric conditions have previ-
ously been described in the Bayer-Villiger oxidation of
racemic ketones by means of enzymatic methods4a-c or
chiral catalysts,4d,e and more recently, an interesting
allylic substitution of racemic 5-vinyloxazolidininones
with phthalimide and a chiral palladium catalyst has
(5) Cook, G. R.; Sankaranarayanan, S. Org. Lett. 2001, 3, 3531 and
pertinent references therein.
(6) Doyle, M. P.; Dyatkin, A. B.; Kalinin, A. V.; Ruppar, Q. A.;
Martin, S. F.; Spaller, M. R.; Liras, S. J . Am. Chem. Soc. 1995, 117,
11021. While this manuscript was in preparation, an interesting new
carbon-carbon bond forming PKR of 4-alkynals catalyzed by Rh(I)/
Tol-BINAP was reported; see: Tanaka, K.; Fu, G. C. J . Am. Chem.
Soc. 2003, 125, 8078.
(7) Visser, M. S.; Hoveyda, A. H. Tetrahedron 1995, 51, 4383.
(8) Bertozzi, F.; Crotti, P.; Macchia, F.; Pineschi, M.; Feringa, B. L.
Angew. Chem., Int. Ed. 2001, 40, 930.
(9) For the kinetic resolution of allylic epoxides with dialkylzinc
reagents and copper-phosphoramidite catalyst, see: (a) Badalassi, F.;
Crotti, P.; Macchia, F.; Pineschi, M.; Arnold, A.; Feringa, B. L.
Tetrahedron Lett. 1998, 39, 7795. (b) Bertozzi, F.; Crotti, P.; Feringa,
B. L.; Macchia, F.; Pineschi, M. Synthesis 2001, 483. For the desym-
metrization reaction of symmetrical allylic epoxides, see: (c) Bertozzi,
F.; Crotti, P.; Macchia, F.; Pineschi, M.; Arnold, A.; Feringa, B. L. Org.
Lett. 2000, 2, 933. (d) Bertozzi, F.; Crotti, P.; Del Moro, F.; Feringa, B.
L.; Macchia, F.; Pineschi, M. Chem. Commun. 2001, 2606.
(1) For a recent review on kinetic resolution, see: Keith, J . M.;
Larrow, J . F.; J acobsen, E. N. Adv. Synth. Catal. 2001, 1, 5-26 and
references therein.
(2) For reviews, see: (a) Dehli, J . R.; Gotor, V. Chem. Soc. Rev. 2002,
31, 365. (b) Eames, J . Angew. Chem., Int. Ed. 2000, 39, 885.
(3) (a) Vedejs, E.; Chen, X. J . Am. Chem. Soc. 1997, 119, 2584. (b)
Cardona, F.; Valenza, S.; Goti, A.; Brandi, A. Eur. J . Org. Chem. 1999,
1319. (c) Pedersen, T. M.; J ensen, J . F.; Humble, R. E.; Rein, T.; Tanner,
D.; Bodmann, K.; Reiser, O. Org. Lett. 2000, 2, 535.
(4) (a) Alphand, V.; Furstoss, R. J . Org. Chem. 1992, 57, 1306. (b)
Adger, B.; Bes, M. T.; Grogan, G.; McCague, R.; Pedragosa-Moreau,
S.; Roberts, S. M.; Villa, R.; Wan, P. W. H.; Willetts, A. J . J . Chem.
Soc., Chem. Commun. 1995, 1563. (c) Petit, F.; Furstoss, R. Tetrahe-
dron: Asymmetry 1993, 4, 1341. (d) Gusso, A.; Baccin, C.; Pinna, F.;
Strukul, G. Organometallics 1994, 13, 3442. (e) Bolm, C.; Schlingloff,
G. J . Chem. Soc., Chem. Commun. 1995, 1247.
10.1021/jo035674h CCC: $27.50 © 2004 American Chemical Society
Published on Web 02/13/2004
J . Org. Chem. 2004, 69, 2099-2105
2099

Pineschi et al.
TABLE 1. RKR of Cyclic Allylic Ep oxid es 2-5 (s-cis
Con for m a tion ) w ith Dia lk ylzin csa
SCHEME 1. P r elim in a r y Resu lts of th e Ca ta lytic
RKR of Cyclic Allylic Ep oxid es
eec (%)
entry
substrate
R
T (h)
SN2′/SN2b
SN2′
SN2
1
2
3
4
5
6
7
8
9
2
2
3
3
4
4
4
5
5
5
Et
Bu
Me
Et
Me
Et
5
5
3
3
3
2
2
18
18
18
70/30
67/33
60/40
76/24
50/50
53/47
50/50
62/38
66/34
86/14
40
45
64
90
91
>97
>98
>95
>90
>95
>98
>95
>95
34
Resu lts a n d Discu ssion
>90
>90
>90
76
68
56
To extend the scope and the synthetic utility of our new
RKR protocol, we examined several cyclic and aliphatic
vinyl oxiranes having different sizes and bearing different
substituents, to evaluate the extent of the chiral recogni-
tion exhibited by the chiral catalyst with these sub-
strates. As an example of semirigid cyclic allylic epoxides
having blocked s-cis conformations, allylic epoxides 2-5
were examined, and the results are summarized in Table
1. The reactions were carried out under our protocol in
order to obtain a complete conversion of the starting
material (reaction monitored by GC and/or TLC) by the
use of an excess of R2Zn in the presence of a catalytic
amount of Cu(OTf)2 (1.5 mol %) and chiral ligand 1 (3.0
mol %) (see the typical procedure in the Experimental
Section). It should be noted that the same reactions
carried out in the absence of the chiral ligand are sluggish
at low temperatures, indicating that these reactions are
typical examples of a ligand-accelerated reaction, ame-
nable to adjustment depending on the nature of the chiral
ligand. Phosphoramidite 1, derived from (R)-BINOL and
(R)-bisphenylethylamine,10 was the chosen ligand be-
cause it proved to be slightly superior to its diastereo-
isomer (S,R,R)-111 with respect to the extent of regiodi-
vergency and the enantioselectivity of the alcohol reaction
products. For these reasons, ligand (R,R,R)-1 was con-
sistently used throughout this work. The data obtained
by the use of ligand 1 confirmed the possibility of
Bu
Med
Etd
Bud,e
10
a
All reactions were performed in accordance with the typical
procedure reported in the Experimental Section (1.5 equiv of R2Zn,
0.015 equiv of Cu(OTf)2, and 0.03 equiv of ligand 1). Conversion
of >98%, unless stated otherwise. For isolated yields, see the
Supporting Information. Determined by 1H NMR examination
b
of the crude mixture. c Determined by GC on chiral stationary
d
phases (see the Supporting Information for further details). 3.0
equiv of Bu2Zn and Et2Zn and 4.0 equiv of Me2Zn were used.
e Conversion of 83%.
obtaining an efficient, highly enantioselective RKR pro-
cess with all the cyclic substrates 2-5, by the use of
commercially available dialkylzinc reagents (entries 1-10,
Table 1). The enantioselective introduction of an alkyl
chain onto 1,3-cyclopentadiene monoepoxide 2 is of
particular importance for the synthesis of chiral nonra-
cemic cyclopentanoids.12 For this purpose, the enantiose-
lective epoxidation of cyclopentadiene followed by alky-
lation could be used, as well. However, the moderate
enantioselectivity exhibited by the epoxidation step
represents a drawback for its effective synthetic utiliza-
tion.13
To the best of our knowledge, the enantioselective reac-
tion of 2 with hard nucleophiles has not been described
except for our seminal article which demonstrated the
possibility of obtaining the SN2′ adduct of type 6 with
good regio- and enantioselectivities when the reaction
was performed in accordance with a kinetic resolution
protocol.9a We now report that the reaction of (()-2 with
an excess of dialkylzinc reagents in the presence of the
copper-phosphoramidite catalyst afforded a separable
mixture of the corresponding regioisomeric alcohols 6a ,b
(10) Feringa, B. L.; Pineschi, M.; Arnold, L. A.; Imbos, R.; de Vries,
A. H. M. Angew. Chem., Int. Ed. 1997, 36, 2620. Chiral ligand 1 was
recently used in the iridium-catalyzed allylic etherification and ami-
nation of achiral allylic esters: (a) Ohmura, T.; Hartwig, J . F. J . Am.
Chem. Soc. 2002, 124, 15164. (b) Lopez, F.; Ohmura, T.; Hartwig, J .
F. J . Am. Chem. Soc. 2003, 125, 3426.
(11) (S,R,R)-Diastereoisomeric phosphoramidite, derived from (S)-
BINOL and (R)-bisphenylethylamine developed by Feringa et al., is
to date one of the most effective chiral ligands for the enantioselective
addition of dialkylzinc to cyclic enones and for other metal-catalyzed
asymmetric processes. For some recent examples, see: (a) Duursma,
A.; Minnaard, A. J .; Feringa, B. L. J . Am. Chem. Soc. 2003, 125, 3700
and references therein. (b) Alexakis, A.; Benhaim, C.; Rosset, S.;
Humam, M. J . Am. Chem. Soc. 2002, 124, 5262. (c) J ensen, J . F.;
Svendsen, B. Y.; la Cour, T. V.; Pedersen, H. L.; J ohannsen, M. J . Am.
Chem. Soc. 2002, 124, 4558. For a review, see: (d) Feringa, B. L. Acc.
Chem. Res. 2000, 33, 346.
(12) For a discussion about the installation of an alkyl chain onto
cyclopentanoid systems, see: Ito, M.; Matsuumi, M.; Murugesh, M.;
Kobayashi, Y. J . Org. Chem. 2001, 66, 5881.
(13) Chang, S.; Heid, R. M.; J acobsen, E. N. Tetrahedron Lett. 1994,
35, 669.
2100 J . Org. Chem., Vol. 69, No. 6, 2004

Regiodivergent Kinetic Resolution of Allylic Epoxides
SCHEME 2. Com p lem en ta r y
En a n tiom er -Dep en d en t Regioselectivity
and 7a ,b, the latter obtained with a high level of enan-
tioselectivity (>90% ee, entries 1 and 2). Similarly, the
addition of Me2Zn (1.5 equiv) to (()-3 catalyzed by
Cu(OTf)2/(R,R,R)-1 afforded a 60/40 mixture of 8a (SN2′
adduct, 64% ee) and 9a (SN2 adduct, >97% ee) (entry 3).8
The catalyzed addition of Et2Zn (1.5 equiv) to the same
racemate afforded a 76/24 mixture of regioisomers 8a
(34% ee) and 9a (>98% ee) (entry 4). The RKR process
was particularly efficient when 1,3-cycloheptadiene mo-
noepoxide 4 was employed (entries 5-7). With this
substrate, the regiodivergency was practically ideal and
regioisomeric alcohols 10a -c and 11a -c, having opposite
configurations at the hydroxyl group-bearing carbon,
were obtained in almost equal amounts and with a high
enantiomeric excess (>90% ee) with all the dialkylzincs
used. Evidently, with this substrate, the asymmetric
matching of the chiral ligand with the enantiomers of
the substrate is considerable. Moreover, with allylic
epoxide 4, the regiodivergent chiral recognition can be
maintained also using different solvents: the addition
of Me2Zn to compound 4 in the presence of a catalytic
amount of Cu(OTf)2 (1.5 mol %) and chiral ligand 1 (3.0
mol %) substantially afforded the same results in terms
of regio- and enantioselectivies when the reaction was
carried out in THF, Et2O, CH2Cl2, or AcOEt.14 The same
reaction carried out in CH3CN turned out to be sluggish
and not regiodivergent, because it afforded the corre-
sponding syn SN2 addition product as the major addition
product. The syn addition appears to occur without the
intervention of the chiral copper complex, via the coor-
dination of zinc to the oxygen of the epoxide and in-
tramolecular transfer of the Me group.15 The regiodiver-
gency obtained by the application of our protocol to 1,3-
cyclooctadiene monoepoxide 5 is not only remarkable (see
entries 8-10) but also synthetically useful.16 In fact,
considering that the enantiomerically enriched trans-2-
alkyl-3-cycloocten-1-ols obtained in this reaction can
easily be reduced by catalytic hydrogenation to the
corresponding saturated compounds, our protocol offers
a new easy route to enantiomerically pure trans-2-alkyl-
substituted cyclooctanols. These compounds cannot be
simply prepared by direct asymmetric alkylation, because
the asymmetric ring opening of relatively unreactive
cyclooctene oxide is still today a difficult challenge, and
our two-step approach is a viable alternative to obtain
enantiomerically pure cyclooctanols.17 It should be men-
tioned that all the reactions reported in Table 1 can be
performed also at 0 °C with only marginal effects on
yields and regio- and enantioselectivities, and this rep-
resents a significant practical advantage.18 On the basis
of all the data collected, it seems clear that a chiral
recognition occurs in situ between the chiral catalyst and
each enantiomer of the allylic epoxide used as the
substrate.
We had previously observed that the use of a racemic
chiral phosphoramidite catalyst suppresses the regiodi-
vergency and affords only the corresponding SN2′ addition
products, that is, the racemic allylic alcohols, from both
enantiomers of the epoxide.8 To obtain more definite
evidence that there is a complementary enantiomer-
dependent regioselectivity, enantiomerically pure epoxide
5 was treated with a copper catalyst derived from either
(S,S,S)-1 or (R,R,R)-1 (Scheme 2). In the presence of
(R,R,R)-1, epoxide (1R,2S)-5 reacted with Me2Zn to give
with complete regioselectivity the corresponding enan-
tiopure allylic alcohol (1R,4R)-12a, whereas when (S,S,S)-1
was used, the corresponding homoallylic alcohol (1R,2R)-
13a was obtained with a good selectivity.16 Thus, it is
clearly demonstrated that it is possible to control the
regioselectivity of the copper-catalyzed addition reaction
of dialkylzincs to an enantiomerically pure cyclic allylic
epoxide, simply by choosing the appropriate enantiomer
of phosphoramidite 1.
Also, semirigid allylic epoxides such as 14-17, having
a blocked s-trans conformation, were examined under our
reaction protocol (Table 2). Complete conversion of allylic
epoxides (()-14 and (()-15 took place in 3 h and gave,
after the usual workup and chromatographic purification,
the corresponding allylic and homoallylic alcohols with
high yields (entries 1-3, Table 2). Valuable new cyclo-
pentanoid systems 18 and 19 can be obtained in an
enantioenriched form through the use of allylic epoxide
14 (entry 1). Actually, it was possible to obtain allylic
alcohol 18 with a good enantioselectivity (72% ee),
whereas a precise determination of the enantioselectivity
of SN2 adduct 19 was not possible. As for the six-mem-
bered allylic epoxide 15, an accurate examination of its
chemical behavior under our protocol with Et2Zn showed
that regioisomeric products 20b and 21b derive from the
enantiomers of 15 in two clearly distinct phases. The first
one is very fast, proceeding with SN2′ regioselectivity
to yield 20b (15 min at -78 °C), while the second
slower one which provided 21b (3 h at -10 °C to arrive
at completion) exhibited a complementary SN2 regiose-
lectivity. It was shown that, after 15 min at -78 °C, the
remaining vinyloxirane 15 (62% conversion) was enan-
tiomerically pure (>98% ee) and it reacted with nearly
complete regioselectivity at the C-2 oxirane carbon with
complete anti stereoselectivity to give the final product
distribution (SN2′/SN2 ) 55/45, entry 3). The catalyzed
addition of Me2Zn followed an even more pronounced
regiodivergent behavior, affording, after complete conver-
(14) However, this is not always true for all the examined substrates.
For example, the use of THF with 1,3-cyclohexadiene monoepoxide 3
afforded a more complex reaction mixture containing also variable
amounts of syn adducts.
(15) For a recent example concerning the syn addition of organozinc
species to cyclic 1,3-diene monoepoxides, see: Xue, S.; Li, Y.; Ha, K.;
Yin, W.; Wang, M.; Guo, Q. Org. Lett. 2002, 4, 905.
(16) For the determination of absolute and relative configurations
of 4-alkylcyclooctanols, see: Del Moro, F.; Crotti, P.; Di Bussolo, V.;
Macchia, F.; Pineschi, M. Org. Lett. 2003, 5, 1971.
(17) For some recent reports, see: (a) J acobsen, E. N. Acc. Chem.
Res. 2000, 33, 421. (b) Denmark, S. E.; Barsanti, P. A.; Wong, K.-T.;
Stavenger, R. A. J . Org. Chem. 1998, 63, 2428. (c) Denmark, S. E.;
Wynn, T.; J ellerichs, B. G. Angew. Chem., Int. Ed. 2001, 40, 2225.
(18) Attempts to perform the reaction entirely at rt gave 3-cyclo-
heptenone as the major product and only trace amounts of addition
compounds.
J . Org. Chem, Vol. 69, No. 6, 2004 2101

Pineschi et al.
TABLE 2. RKR of Cyclic Allylic Ep oxid es (s-tr a n s
with Me2Zn (1.5 equiv) in the presence of racemic chiral
catalyst 1 gave homoallylic alcohol 25a as the main
reaction product (>90% of the crude reaction mixture).
Considering that, in an enantioselective RKR process, the
product obtained with the higher ee is commonly associ-
ated with the slower reaction, the above observations for
epoxide 17 tend to indicate that, under our reaction
protocol, the SN2′ addition can be, in some cases, the
slower pathway in the addition process.
Con for m a tion )a
At this point, we thought it interesting to check the
behavior of acyclic allylic epoxides such as 26-34 under
our reaction protocol (Table 3). The conformational
mobility typically associated with these compounds could
give a better insight into the reaction mechanism. Ste-
reoelectronic considerations require the epoxide and
vinylic carbon centers to assume a nearly coplanar
disposition in the course of the reaction, even if we
recently demonstrated that different arrangements are
also possible.16 In open noncyclic allylic epoxides, the
required coplanarity necessary for an effective interaction
between the π-orbitals and the breaking oxirane allylic
C-O bond can be matched in the corresponding s-cis and
s-trans conformers. Assuming a more favorable anti
stereoselective mode of attack by the nucleophile, the
conformational equilibrium between the s-trans and s-cis
forms swaps the face of the double bond presented to the
chiral catalyst, thus rendering an effective chiral recogni-
tion as a really challenging event. Several acyclic allylic
epoxides with different substitution patterns (type and
geometry of the double bond and/or of the epoxide) were
therefore synthesized and examined under our reaction
condition protocol.
The use of commercially available 1,3-butadiene mo-
noepoxide 26 under our RKR protocol gave mainly achiral
conjugate addition products 35a ,b with moderate E/Z
diastereoselectivities and only minor amounts (3-5%) of
SN2 addition products 36a ,b (entries 1 and 2, Table 3).
When isoprene monoepoxide 27 was used, a slightly
better E/Z diastereoselectivity (E isomer > 91%) was
obtained for SN2′ adducts 37a ,b. Moreover, it is also
significant that with isoprene monoepoxide it was pos-
sible to obtain greater amounts (12-15%) of the SN2
addition products, homoallylic alcohols 38a ,b, containing
quaternary stereocenters, with respect to the use of
butadiene monoepoxide under the same reaction condi-
tions.20 Allylic epoxide (E)-28, bearing a terminal oxirane
function, is of special interest because the double bond
is in conjugation with the aromatic ring. This structural
arrangement is known to favor the formation of the SN2
addition product, when related copper-mediated allylic
alkylation of cynnamylic substrates has been used.21
Contrary to this expectation, the Cu(II)/ligand 1-cata-
lyzed addition of Me2Zn to epoxide 28 gave an inseparable
65/35 mixture of alcohols (E)-39 (SN2′ adduct) and 40 (SN2
adduct),22 respectively, both obtained with very low ee’s
eec (%)
entry
substrate
R
T (h)
SN2′/SN2b
SN2′
SN2
1
2
3
4
5
6
7
14
15
15
16
17
17
17
Et
3
3
3
18
18
18
18
56/44
49/51
55/45
>95/<5
23/77
72
96
80
nd
92
99
Me
Et
Etd
Me
Et
13
>97
32
88
90
56/44
52/48
72
Bu
82
a-c See corresponding footnotes in Table 1. Conversion of 34%.
Unreacted allylic epoxide 16 was recovered with 2% ee. nd ) not
determined.
d
sion of (()-15, allylic alcohol 20a (96% ee) and homoallylic
alcohol 21a (92% ee) (entry 2).8 Considering that the
construction of a quaternary stereocenter by a catalytic
asymmetric reaction is certainly a topic of current inter-
est, we examined allylic epoxide 16, which contains a
butyl group at the C-2 oxirane carbon.19 Unfortunately,
epoxide 16 turned out to be particularly unreactive when
treated with an excess of Et2Zn under our usual reaction
protocol, and only SN2′ adduct 22 and unreacted allylic
epoxide were recovered, both with low enantioselectivi-
ties. No trace of the desired corresponding SN2 adduct
23 was observed (entry 4). The copper-catalyzed addition
of Et2Zn to the unprecedented cycloheptenyl allylic
epoxide 17 afforded, after complete conversion, the cor-
responding allylic and homoallylic alcohols 24b and 25b,
respectively, with high yields and enantioselectivities
(entry 6). An even more regiodivergent reaction was
obtained in the copper-catalyzed addition of Bu2Zn, giving
compounds 24c and 25c with high enantioselectivities
(entry 7). In general, in the catalyzed addition of dialkyl-
zinc reagents to allylic epoxides, the allylic alcohol (SN2′
adduct) is the main reaction product and there is a clear
tendency of Me2Zn, with respect to Et2Zn, to give in-
creased amounts of SN2 adducts for all the substrates
examined. In this sense, it is interesting to note the
regiodivergent behavior favoring the formation of ho-
moallylic alcohol 25a (SN2 adduct) exhibited by Me2Zn
in the reaction with epoxide 17 (entry 5). In this case,
the minor regioisomeric allylic alcohol 24a was obtained
with a very high enantioselectivity (>97% ee), and when
the reaction was carried out in accordance with a kinetic
resolution protocol (0.50 equiv of Me2Zn), 25a was the
main reaction product. Moreover, the reaction performed
(20) Regioisomeric alcohols 37a ,b and 38a ,b were obtained as an
inseparable mixture, and a precise determination of their enantiose-
lectivity was not possible. However, when the reaction was performed
in the presence of racemic chiral ligand 1, an increase in SN2′ adducts
37a ,b (95% of the crude mixture) was observed. This is a clear, even
if indirect, indication that a chiral recognition is present to some extent
for isoprene monoepoxide.
(19) For reviews on the enantioselective formation of quaternary
carbon stereocenters, see: (a) Corey, E. J .; Guzma´n-Pe´rez, A. Angew.
Chem., Int. Ed. 1998, 37, 388. (b) Christoffers, J .; Mann, A. Angew.
Chem., Int. Ed. 2001, 40, 4591.
(21) Tseng, C. C.; Paisley, S. D.; Goering, H. L. J . Org. Chem. 1986,
51, 2884.
(22) Lautens, M.; Hiebert, S.; Renaud, J .-L. J . Am. Chem. Soc. 2001,
123, 6834.
2102 J . Org. Chem., Vol. 69, No. 6, 2004

Regiodivergent Kinetic Resolution of Allylic Epoxides
TABLE 3. RKR of Con for m a tion a lly Mobile Allylic Ep oxid esa
eec (%)
entry
substrate
R
T (h)
SN2′/SN2b
E/ Zb (SN2′)
SN2′
SN2
1
2
3
4
5
6
7
8
9
10
11
12
13
26
26
27
27
28
29
30
31
32
33
34
34
34
Et
Bu
Et
Bu
Me
Et
5
5
5
5
4
97/3
95/5
82/18
85/15
93/7
85/15
87/13
65/35
68/32
75/25
66/34
85/15
90/10
53/47
55/45
57/43
nd
nd
6
91/9
>95/<5
>95/<5
>95/<5
>95/<5
87/13
2
4
6
15
2
Et
Et
Et
Et
5d
4
rac
52
ndf
10
94
90
89
97
ndf
nd
nd
92
95
18e
4e
3
nd
Meg
Etg
Bug
>95/<5
>95/<5
>95/<5
1
4
d
a-c See corresponding footnotes in Table 1. Conversion of 30%. e Reaction performed with 3.0 equiv of Et2Zn. f The enantioselectivity
g
(4%) was determined only of the starting epoxide 32 by HPLC (Daicel Chiralcel OD-H). Reaction performed entirely at -78 °C.
(entry 5). Similarly, the catalyzed addition of Et2Zn to
cis-epoxide 29, bearing a terminal double bond, afforded
a 68/32 mixture of regioisomeric alcohols (E)-41 and 42,
respectively, with poor stereoselectivities (entry 6). With
epoxides 28 and 29, the use of racemic chiral ligand
(S,S,S)(R,R,R)-1 afforded almost the same regioisomeric
ratio, thus indicating that a chiral recognition is not
operative with this substrate (data not shown in Table
3). Allylic cis-epoxide 30, possessing a Z configuration of
the double bond, turned out to be relatively unreactive
under our reaction conditions (30% conversion after 5 h
at 0 °C), and the corresponding ring-opened products (E)-
43 (SN2′ adduct) and (Z)-44 (SN2 adduct) were obtained
as racemates (entry 7). However, it is to be noted that
within SN2′ adduct 43, deriving from an allylic rear-
rangement, the new double bond had an E configuration,
whereas in SN2 adduct 44 the original Z double bond
configuration present in the starting epoxide was fully
maintained.23 The preservation of the Z geometry of the
starting allylic epoxide was also observed in homoallylic
alcohol 46 (SN2 adduct), deriving from the regiodivergent
catalyzed addition of Et2Zn to trans-epoxide 31.
with the incursion of an efficient RKR process absent in
the other acyclic substrates so far examined. In particu-
lar, the behavior of allylic epoxide 31 was markedly
different from that of the closely related epoxide 30, the
only differences being the cis geometry of the epoxide and
the presence of an ethereal functionality in the latter
compound. In sharp contrast with the good results
obtained with compound 31, the copper-phosphoramidite-
catalyzed reaction with Et2Zn of trans-allylic epoxide 32,24
bearing an E double bond, gave an 85/15 mixture of two
regioisomeric SN2′ adducts 47a ,b (as an 87/13 E/Z
mixture) and 48 (SN2 adduct), respectively (entry 9). In
this case, it was possible to obtain in a pure state also a
substantial amount of (Z)-SN2′ adduct 47b. Unfortu-
nately, it was not possible to determine the enantiose-
lectivity of alcohols 47a ,b and 48. However, an exami-
nation of the reaction mixture composition after ∼4 h at
0 °C revealed that the starting allylic epoxide 32 was
essentially racemic at this point (75% of conversion), thus
excluding the incursion of an effective RKR process. We
also explored the reactivity of epoxide 33, deriving from
geraniol, in view of the possibility of forming a quater-
nary center by means of an SN2′ reaction. Actually, the
addition of Et2Zn to epoxide 33 performed under our
protocol afforded the desired allylic alcohol 49 (SN2
adduct) with a good regioselectivity (90%) but unfortu-
nately with a low ee (10% ee, entry 10). The same
reaction afforded only a minor amount (∼10%) of regio-
In this case, we were delighted by the good enantiose-
lectivities found in the corresponding ring-opening prod-
ucts 45 (52% ee) and 46 (97% ee) (entry 8), in accordance
(23) For a study on the relationship of the double bond configuration
between reactants and products in the cross-coupling reactions of allylic
substrates with organocopper reagents, see: Underiner, T. L.; Paisley,
S. D.; Schmitter, J .; Leshesky, L.; Goering, H. L. J . Org. Chem. 1989,
54, 2369.
(24) Olofsson, B.; Somfai, P. J . Org. Chem. 2002, 67, 8574.
J . Org. Chem, Vol. 69, No. 6, 2004 2103

Pineschi et al.
SCHEME 3. P ostu la ted Mech a n ism for th e RKR
(On ly On e En a n tiom er of th e Allylic Ep oxid e Is
Sh ow n )
isomeric homoallylic alcohol 50 (ee not determined).
Much to our surprise, we were able to create a quaternary
stereocenter efficiently and with a high enantioselectivity
when 1-vinyl-1-cyclohexene oxide (34)25 was used as the
substrate for our RKR procedure. This conformationally
mobile substrate gave a very fast reaction with dialky-
lzincs in the presence of catalytic amounts of Cu(II)/
ligand 1, to the point that the reaction could be com-
pletely performed at -78 °C, giving an almost equimolar
mixture of (E)-allylic alcohols 51a -c26 and homoallylic
alcohols 52a ,c, both with high optical purities (89-95%
ee’s) and quantitative yields (entries 11-13 and the
Supporting Information). It should be noted that the SN2
addition pathway to 34 allows the obtainment of a
quaternary carbon stereocenter with a high enantiose-
lectivity.19 Like most of the semirigid substrates exam-
ined in Tables 1 and 2, the reaction of epoxide 34 with
0.50 equiv of R2Zn catalyzed by Cu(II)/(S,S,S)-1, accord-
ing to a kinetic resolution protocol, and with R2Zn (1.5
equiv, catalyzed by racemic 1) afforded compounds of type
(E)-51 with high levels of SN2′ regioselectivity (>90%).
The asymmetric matching of the ligand with acyclic
allylic epoxide substrates is somewhat puzzling and
substrate-dependent, and it is difficult to give a complete
and coherent rationalization. However, good levels of E
stereoselectivity were uniformly obtained for the SN2′
adducts when conformationally mobile allylic epoxides
26-34 were used. Even if the ground state conformers
are not necessarily the same, it is reasonable that the
corresponding s-trans conformers are the reactive ones.
Con sid er a tion s on th e Rea ction Mech a n ism . The
mechanism of a copper-catalyzed allylic alkylation has
not been fully established yet, and scant information is
available in regard to the transition state. This is mainly
due to the scarcity of direct methods to investigate the
reaction pathway and to the lack of any intermediate
that can be isolated. Most probably, the dialkylzinc
reagent reduces in situ the Cu(II) salt to the Cu(I) salt,
which is the true catalytic species. Moreover, it is gen-
erally admitted that an oxidative addition occurs to form
a Cu(III) intermediate27 and that the rate-determining
step of a cuprate conjugate addition is the last stage of
the reaction, that is, the reductive elimination process.28
Steric factors do not seem to play an important role in
the regiochemical outcome of the copper-catalyzed re-
giodivergent additions to the allylic epoxides examined
by us. In fact, unlike the related use of stoichiometric
cuprates,29 the presence of alkyl substituents in the allylic
position does not hinder SN2 addition when conforma-
tionally mobile allylic epoxides are used. On the contrary,
it seems that a substituent in this position leads to an
increase of the SN2 adduct (entries 3, 4, and 11-13, Table
3). We have now shown that the slower reaction is
usually, but not exclusively, the SN2 addition pathway,
and it is this reaction (i.e., the slower reaction) that is
invariably associated with the formation of the product
with the higher enantioselectivity. The maintenance of
the Z double bond geometry present in allylic epoxides
30 and 31 in the corresponding SN2 adducts (compounds
44 and 46, entries 7 and 8) supports the notion that the
configuration of the double bond is partly retained
throughout the reaction.23 As shown in Scheme 3 for the
reaction of one enantiomer of epoxide 31, taken as a
model for all examined noncyclic allylic epoxides 26-34,
this experimental evidence might be consistent with an
oxidative addition process leading directly to a Cu(III)-
π-allyl system of type A30 in equilibrium with the corre-
sponding regioisomeric σ-allyl-copper(III) species B and
C. The subsequent reductive elimination steps, K1 and
K2, from B and C, respectively, probably play a crucial
role in determining the stereo- and regioselectivities
observed (Scheme 3).28
In this framework, the fundamental importance of the
sense of chirality of the chiral ligand, when the RKR is
operative, should be noted. Probably, the fact that the
absolute configuration of the chiral ligand, having matched
and mismatched combinations with the enantiomers of
the allylic substrate, determines the regiochemical out-
come seems to indicate that the nonracemic phosphor-
amidite ligand plays a fundamental role in the accelera-
tion of the carbon-carbon bond formation during the
reductive elimination step.28 Moreover, the presence of
an olefin is required for the reaction to occur, since cyclo-
hexene oxide and also the more reactive styrene oxide
do not give the corresponding ring-opened products under
the reaction conditions commonly used. These data seem
to indicate that a kinetically controlled obtainment of the
corresponding σ-allyl-copper(III) species B and C in a
totally independent way, by means of a selective nucleo-
philic cuprate addition to each enantiomer of the allylic
(25) Annis, G. D.; Ley, S. V.; Self, C. R.; Sivaramakrishnan, R. J .
Chem. Soc., Perkin Trans. 1 1981, 270.
(26) The E configuration of the double bonds in compounds 51a -c
has been established by 1D ROESY spectra (see the Supporting
Information).
(27) For very recent experimental evidence supporting the incursion
of Cu(III) intermediates, see: Karlstro¨m, A. S. E.; Ba¨ckwall, J .-E.
Chem.sEur. J . 2001, 7, 1981 and references therein.
(28) For the importance of the reductive elimination step in a copper-
catalyzed conjugate addition, see: Nakamura, E.; Yamanaka, M.; Mori,
S. J . Am. Chem. Soc. 2000, 122, 1826.
(30) Due to the scarcity of direct methods to investigate the reaction
pathway, the obtainment of σ-allyl-copper(III) species B, in equilib-
rium with regioisomeric σ-allyl-copper(III) species C, through a π-allyl
complex or a suprafacial 1,3-sigmatropic shift cannot be ruled out.
However, the substantial preservation of the stereochemical integrity
of the original double bonds present in the SN2 adducts seems to
indicate that the reductive elimination step is faster than the eventual
incursion of syn/ anti isomerization processes.
(29) For a review, see: Marshall, J . A. Chem. Rev. 1989, 89, 1503
and pertinent references therein.
2104 J . Org. Chem., Vol. 69, No. 6, 2004

Regiodivergent Kinetic Resolution of Allylic Epoxides
epoxide, is not likely.12 It is difficult to imagine a scenario
in which a complementary enantiomer-dependent regio-
selectivity having the described peculiarities can be
obtained, without the incursion of any common interme-
diates.
anhydrous toluene (1.5 mL) was stirred at rt for 40 min. The
colorless solution was cooled to -78 °C, followed by subsequent
addition of a solution of 1 (110 mg, 1.0 mmol) in toluene (0.5
mL). After 5 min, Et2Zn (1.36 mL of a 1.1 M solution in toluene,
1.5 mmol) was added, and the stirred reaction mixture was
allowed to warm slowly up to 0 °C. After 3 h (>98% conver-
sion), the mixture was quenched with saturated aqueous
NH4Cl solution (5 mL). Extraction with Et2O (2 × 35 mL) and
evaporation of the dried (MgSO4) organic phase afforded a very
clean crude mixture consisting only of alcohols 10b and 11b
(>90% crude yield). The crude reaction mixture was subjected
to flash chromatography (SiO2) with 20% EtOAc in hexanes
to give, as the second eluting fractions, 56 mg of pure 10b
(40%), as a liquid. Rf ) 0.20. 1H NMR δ 5.43-5.77 (m, 2H),
4.29-4.49 (m, 1H), 1.18-2.32 (m, 9H), 0.90 (t, 3H, J ) 7.2
Hz). 13C NMR δ 136.5, 135.3, 70.9, 40.2, 36.3, 32.2, 29.1, 22.7,
12.3. Anal. Calcd for C9H16O: C, 77.09; H, 11.50. Found: C,
76.88; H, 11.39. The enantiomeric excess of 10b (>90%) was
determined by chiral GC (â-cyclodextrin column): isothermal
110 °C, 22.0 psig; tR 43.82 min (minor), tR 44.36 min (major).
The first eluting fractions of the flash chromatography afforded
35 mg of pure 11b (25%), as a liquid. Rf ) 0.26 (20% AcOEt in
hexanes). 1H NMR δ 5.81-5.97 (m, 1H), 5.40-5.52 (m, 1H),
3.47-3.61 (m, 1H), 1.19-2.39 (m, 9H), 0.93 (t, 3H, J ) 7.4
Hz). 13C NMR δ 133.7, 132.7, 71.7, 47.8, 39.0, 29.0, 25.0, 24.0,
11.7. Anal. Calcd for C9H16O: C, 77.09; H, 11.50. Found: C,
77.21; H, 11.33. The enantiomeric excess of 11b (>90%) was
determined by chiral GC (â-cyclodextrin column): isothermal
110 °C, 22.0 psig; tR 34.06 min (minor), tR 34.56 min (major).
In conclusion, we have shown that the outcome of the
enantioselective regiodivergent kinetic resolution of ra-
cemic allylic epoxides with dialkylzinc reagents is effec-
tive for a variety of cyclic substrates, and also for some
acyclic substrates. The most fascinating and unusual
aspect is certainly that the regioselectivity of the reaction
depends directly on the absolute configuration of the
chiral catalyst. This new process allows the obtainment
of several new allylic and homoallylic alcohols with good
to excellent enantioselectivities. In most cases, the allylic
and homoallylic alcohol reaction products can easily be
separated by chromatography on silica gel. This method
is also amenable for a novel obtainment of a quaternary
carbon stereocenter with a high enantioselectivity. These
facts, together with the ease of the reaction procedure,
give a good level of practicality to our method and the
possibility to access several allylic and homoallylic alco-
hols in an enantioenriched form. The mechanism of this
reaction and the exact nature of the copper species
involved in this highly efficient catalytic process still need
to be established. Nevertheless, the data obtained sup-
port the notion that the reductive elimination is the regio-
and stereodetermining step.
Ack n ow led gm en t. This work was supported by
M.I.U.R. (PRIN 2002) and by the University of Pisa.
P.C. gratefully acknowledges Merck for the generous
financial support deriving from the 2002 ADP Chem-
istry Award.
Exp er im en ta l Section
Ma ter ia ls a n d Meth od s. For general information, see refs
8 and 16. The E configuration of the double bonds of com-
pounds 51a -c was established by proton 1D ROESY spectra.
1D ROESY spectra were recorded on a spectrometer operating
at 599.69 MHz using a 5 mm broadband inverse probe with a
z-axis gradient, with a spectral width of 5400 Hz and a mixing
time of 0.6 s.
Typ ica l P r oced u r e: P r ep a r a tion of (1R,4R)-4-Eth yl-
2-cycloh ep ten -1-ol (10b) a n d of (1S,2S)-2-Eth yl-3-cyclo-
h ep ten -1-ol (11b). A solution of Cu(OTf)2 (5.40 mg, 0.015
mmol) and chiral ligand (R,R,R)-1 (16.2 mg, 0.03 mmol) in
Note Ad d ed a fter ASAP P ostin g. This article was
released ASAP on 2/13/2004. Two entries in Table 1 were
changed. The paper was reposted on 2/16/2004.
Su p p or tin g In for m a tion Ava ila ble: Text giving experi-
mental procedures and characterization data for all new
compounds. This material is available free of charge via the
Internet at http://pubs.acs.org.
J O035674H
J . Org. Chem, Vol. 69, No. 6, 2004 2105