Investigation of SmI2-mediated cyclisation of δ-iodo-α,β-unsaturated esters by deuterium 2D NMR in oriented solvents

Article abstract of DOI:10.1016/S0957-4166(02)00259-8

We have investigated the mechanism of the SmI₂-mediated cyclisation of δ-iodo-α,β-unsaturated esters using proton-decoupled deuterium 1D and 2D NMR in weakly ordered polypeptide liquid crystal solvents. Analysis of the spectroscopic results dem

Full text of DOI:10.1016/S0957-4166(02)00259-8

TETRAHEDRON:
ASYMMETRY
Pergamon
Tetrahedron: Asymmetry 13 (2002) 1465–1475
Investigation of SmI2-mediated cyclisation of
d-iodo-a,b-unsaturated esters by deuterium 2D NMR in oriented
solvents
He´le`ne Villar,a Franc¸ois Guibe´,a,* Christie Aroulandab and Philippe Lesotb,*
aLaboratoire de Catalyse Mole´culaire, CNRS ESA 8075, ICMO, Baˆt. 420, Universite´ de Paris-Sud, 91405 Orsay, France
bLaboratoire de Chimie Structurale Organique, CNRS ESA 8074, ICMO, Baˆt. 410, Universite´ de Paris-Sud,
91405 Orsay, France
Received 18 April 2002; accepted 24 May 2002
Abstract—We have investigated the mechanism of the SmI2-mediated cyclisation of d-iodo-a,b-unsaturated esters using proton-
decoupled deuterium 1D and 2D NMR in weakly ordered polypeptide liquid crystal solvents. Analysis of the spectroscopic results
demonstrates that the cyclisation reaction takes place with concomitant and extensive racemisation at the g-position and the
reason for this racemisation is discussed. We also report an efficient 2D NMR strategy for distinguishing meso- from d- and
l-diastereoisomers based on the introduction of a CD2 probe in the molecules to be studied. This approach allows determination
of the diastereoisomeric and enantiomeric composition of a mixture. © 2002 Elsevier Science Ltd. All rights reserved.
1. Introduction
ordered polypeptide liquid crystals as an analytical tool.
In the course of this investigation, we have developed
an interesting and efficient method for distinguishing
meso- from d,l stereoisomers, and consequently for
safely assessing diastereoisomeric and enantiomeric
purity based on proton-decoupled deuterium (2H-{1H})
2D NMR in chiral and achiral anisotropic solvents.
It was recently reported that d-iodo- and d-bromo-a,b-
unsaturated esters readily cyclise, according to a 3-exo-
trig process, to cyclopropane compounds in the
presence of samarium diiodide and a proton source.1
This reaction was shown to be highly tolerant of vari-
ous substitution patterns in the starting homoallylic
halides. However, only racemic compounds were used
in this study and the important question—both from a
synthetic point of view and for a better understanding
of the reaction mechanism—as to whether a homochi-
ral g-substituted substrate would retain its enantiomeric
purity during the cyclisation process has not been
addressed. We present herein our investigations using
deuterium 1D and 2D NMR spectroscopy in weakly
2. Results and discussion
We chose the ‘anti’-(E)-benzyl 5-iodo-4-methylhex-2-
enoate 1 as a model substrate for cyclisation. Its cycli-
sation has already been studied in racemic form and it
was reported that the ratio of meso and d,l-cyclo-
propane products 2 and 3 was around 0.5 (Scheme 1).1
Scheme 1.
* Corresponding authors. E-mail: fraguibe@icmo.u-psud.fr (for the synthetic part); philesot@icmo.u-psud.fr (for the NMR part).
0957-4166/02/$ - see front matter © 2002 Elsevier Science Ltd. All rights reserved.
PII: S0957-4166(02)00259-8

1466
H. Villar et al. / Tetrahedron: Asymmetry 13 (2002) 1465–1475
Enantiomerically enriched benzyl (4R,5S)-5-iodo-4-
methylhex-(2E)-enoate was synthesised according to
Scheme 2. From diene 4, prepared by Wittig olefination
of tiglic aldehyde with carbobenzyloxymethylene phos-
butanol (3 equiv.) in THF (no HMPA added) at room
temperature for several hours. The reaction was then
quenched by adding dilute aqueous HCl and subjected
to standard extractive workup. 1H NMR analysis of the
crude products confirmed the already observed1 total
conversion of benzyl 5-iodo-4-methylhex-(2E)-enoate to
a ca. 1:2 mixture of meso-2 and d,l-3 cyclopropanes
(based on integration of the hydrogen signal next to the
carbobenzyloxy group in the isotropic proton
spectrum).
phorane,
(R,R)-2-[(E)-(benzyloxycarbonyl)vinyl]-2,3-
dimethyloxirane 5 was obtained in 95% yield and high
enantiomeric purity (94% e.e., as determined by HPLC)
by chemoselective and enantioselective epoxidation
with oxone in the presence of the fructose-derived
ketone 7 according to the procedure of Shi and co-
workers.2 Regio- and stereoselective reductive ring
opening of epoxide 5 with the ternary system3 Me2NH–
BH3/AcOH/Pd(PPh3)4 (5 mol%) gave homoallylic alco-
If spectral diastereodifferentiation based on the proton
resonance anisochrony was obtained in this example,
the determination of the enantiomeric composition of 3
was not possible. Note that NMR diastereodifferentia-
tion is not4 always observed in isotropic solvents even
when operating at very high magnetic fields. Experi-
mentally, neither chiral chromatographic approaches
nor conventional isotropic NMR methods involving the
use of chiral shift reagents or chiral solvents provided
hol
6 without any loss of enantiomeric purity
(e.e.=93% as determined by HPLC). Compound 6 was
then converted into enantioenriched (4R,5S)-5-iodo-4-
methylhex-(2E)-enoic acid benzyl ester 1 by reaction
with I2/PPh3/imidazole. The rather low yield obtained
in this last step is the result of competitive elimination
leading back to diene 4. Racemic 1 was obtained in a
similar way except that the racemic epoxide intermedi-
ate was prepared in 95% yield by chemoselective
monoepoxidation of 4 with one equivalent of MCPBA.
The two enantiomers of the iodo compound could not
be separated by chiral HPLC, but the fact that benzyl
5-iodo-4-methylhex-(2E)-enoate was obtained in
diastereoisomerically pure form (within 1H NMR detec-
tion limits, d.e. >95%) in the racemic as well as in the
enantioenriched series shows that the conversion of
alcohol to iodide takes place without any appreciable
epimerisation, presumably with clean inversion at the
hydroxyl bearing carbon atom.
successful results4,5 because
3
does not form
diastereoisomeric complexes or short-lived diastereoiso-
meric adducts in situ which give distinct NMR signals
on the basis of an isotropic chemical shift difference.
Chemical derivatisation of the molecule could be envis-
aged, but NMR results are not always guaranteed.
To circumvent these problems we have explored
another analytical strategy involving NMR in a weakly
ordered chiral polypeptide solvent.6 In this new
approach, the enantiomeric discrimination observed in
NMR spectra is due to selective orientational ordering
of two enantiomers within the oriented phase. This
effect depends both on geometrical and electrostatic
factors. Consequently, whatever their chemical func-
tionalities and their topology, two enantiomers will
The cyclisation of 1, either in racemic or enantiomeri-
cally enriched form, was achieved by exposure to
samarium diiodide (2.5 equiv.) in the presence of tert-
Scheme 2.

H. Villar et al. / Tetrahedron: Asymmetry 13 (2002) 1465–1475
1467
always have different orientational ordering in a chiral
liquid crystal (CLC). In almost all cases, this differen-
tial molecular ordering effect yields a doubling of the
spectra (one for each enantiomer) compared to that
observed with an achiral oriented solvent. This fact is
an obvious advantage compared to classical NMR
methods, which involve chiral solvents or chiral lan-
thanide chemical shift reagents, both approaches
requiring specific chemical functions in the analyte.
Notably, this selective ordering enables discrimination
between two diastereoisomers that have different
molecular ‘shapes’.7 Thus, we have shown that this
approach is a powerful and flexible tool, generally
providing better spectroscopic results than those
achieved using routine techniques.
Earlier investigations have shown that proton-decou-
pled deuterium (2H-{1H}) NMR of isotopically labelled
molecules was a very efficient means of enantiomeric
analysis, mainly for two reasons: firstly, the determina-
tion of the enantiomeric or diastereoisomeric composi-
tion by peak integration is easy and very accurate even
for highly enantioenriched mixtures, because we benefit
Figure 1. 61.4 MHz proton-decoupled deuterium 1D spec-
trum of the mixture I dissolved in the PBLG/CHCl3 phase
(sample 1) at 293 K. Here the d,l mixture is racemic. A
Gaussian filtering (GB=50%, LB=−2.5 Hz) was applied to
enhance the spectral resolution. DwiQ is the quadrupolar split-
ting for corresponding doublet. In the formula given at the
top of the figure, QDi is the deuterium quadrupole moment,
eqCꢀDi is the electric field gradient (EFG) along the CꢀDi
bond and SCꢀDi is the order parameter of the CꢀDi bond.
2
from an abundant spin. Secondly, the H-{1H} spec-
trum of two monodeuterated enantiomers oriented dif-
ferently in
a chiral anisotropic environment are
particularly simple to analyse because they usually con-
sist of two independent quadrupolar doublets (spin
I=1) centred on the corresponding la2niso which is very
close to li1sHo. The separation betweenHthe two compo-
nents is referred to as the quadrupolar splitting and
noted DwQ (see Fig. 1).6,8
2
Fig. 1a shows the H-{1H} spectrum of the mixture I
issued from cyclisation of racemic 1 and containing the
isotopically labelled meso-isomer and d,l-enantiomers
dissolved in the PBLG/CHCl3 phase (sample 1). Disre-
garding the signal (the most shielded component) from
residual CDCl3, twelve intense peaks, corresponding to
six quadrupolar doublets approximately centred on the
same chemical shift, are visible (note that some weak
deuterated impurity peaks are also observed in the
spectrum, but do not interfere with the other doublets).
To introduce deuterons into compounds meso-2 and
d,l-3 in one step without racemisation, the crude mix-
ture of cyclopropane esters was directly reduced with
lithium aluminium deuteride (LiAlD4) to yield a mix-
ture of cyclopropane dideuteroalcohols (Scheme 3).
This first mixture is referred to as mixture I. It could be
argued here that the synthesis of monodeuterated
derivatives could simplify subsequent analysis of the
2H-{1H} NMR spectrum in the oriented phase. How-
ever, this remark is partly false because the
monodeuteration of compounds would be chemically
more delicate and would not permit assignment of
meso- from d,l-diastereoisomers in the mixture as we
will see below.
To decipher this spectrum, we must keep three points in
mind: firstly, three distinct dideuterated species, a single
meso-isomer and two enantiomers, exist in the mixture
(see Fig. 2); secondly, two deuterons of a CD2 group in
a chiral molecule are diastereotopic, and are thus spec-
troscopically non-equivalent, as in isotropic solvents.
Consequently, we expect to observe two pairs of
quadrupolar doublets (if the chiral discrimination of the
derivatives 13 and ent-13 oriented in a CLC occurs),
each of them centred on the chemical shift of the
corresponding diastereotopic deuteron;8,9 thirdly, enan-
tiotopic nuclei in Cs symmetry molecules, such as the
The same reasons also preclude the investigation of the
mixture using natural abundance deuterium NMR,8
even if this approach was a priori much more attractive
because no further step for introducing a deuterium
probe was required.
Scheme 3.

1468
H. Villar et al. / Tetrahedron: Asymmetry 13 (2002) 1465–1475
>66%). In both cases, the meso/d,l assignment of dou-
blets as well as the d.e. measurement are possible. Other
spectral situations exist but they are not so informative
compared with the previous ones. Thus the presence of
three pairs of quadrupolar doublets with variable area
2
on the H-{1H} spectrum implies that d.e. and e.e. are
both not null, but neither the assignment of doublets
nor the composition of the mixture are possible. When
six doublets of equal area are observed, we can deduce
that the e.e. is null while the meso/d,l ratio is 0.5.
However, pairing doublets can be extremely difficult if
2
the H chemical shift of the meso and d,l isomers is too
small, and in all cases the assignment of the three pairs
of doublets to the corresponding isomers is precluded
from analysis of the 1D spectrum.
2
This last situation is observed in the H-{1H} spectrum
shown in Fig. 1. This result is therefore consistent with
the fact that the cyclisation of rac-1 leads to a 1:2
mixture of cyclopropanes 2 and 3 as already estimated
using isotropic NMR, but no further information can
be extracted from the spectrum. The differences in peak
intensity, which could theoretically be expected to be of
the same intensity, mainly arises from non resolved
2H–2H dipolar and scalar couplings between gem-
deuterons participating to the linewidth of resonances.
The presence of non-zero 2H–2H geminal couplings
leads to the possibility of correlating the two quadrupo-
lar doublets corresponding to the deuterons of the same
isomer, as is schematically presented in Fig. 3. This is
the reason why natural abundance deuterium NMR
could not be successfully used in this respect. Indeed
the very low natural abundance of deuterium nuclei
(0.015%) excludes the possibility of detecting the 2H–2H
Figure 2. meso (12) and active forms (13 and ent-13) of
isotopically labelled cyclopropane derivatives studied. The
enantiotopic deuterons in the meso compound are referred to
as M1 and M2. The diastereotopic deuterons in chiral com-
pounds are referred to as A1 and A2 in the A enantiomer, and
B1 and B2 in the B enantiomer. In this scheme, deuterons A1
(A2) and B2 (B1) are mirror images. Note that the numbering
A,B and 1,2 is chosen arbitrarily.
pro-R/pro-S deuterons in the meso-compound 12, are
also magnetically non-equivalent in a CLC (see Fig.
2).10 Consequently, two distinct quadrupolar doublets
(one for each enantiotopic deuteron) centred on the
same chemical shift are expected to be observed if the
spectral separation is large enough. In conclusion, in a
CLC, three pairs of quadrupolar doublets should be
visible if no overlap occurs fortuitously. Although the
diastereotopic and enantiotopic deuterons are theoreti-
cally anisochronous, in this example, the chemical shift
differences between them are too weak (<3 Hz) to
provide an unambiguous assignment of the various
quadrupolar splittings.
2
scalar and dipolar couplings, and hence the H-{1H}
spectra consist of a superimposition of independent
quadrupolar doublets which cannot be correlated.8
In order to detect the correlation peaks between
quadrupolar doublets, we recorded the deuterium
COSY 2D spectrum of the mixture. Fig. 4a shows the
deuterium COSY-45 2D map of the mixture in the
PBLG system recorded at 293 K. In this 2D contour
plot, we observe three types of deuterium signal: the
resonances of the deuterium spectrum appearing on the
main diagonal, the autocorrelation peaks appearing
displaced from the other diagonal by the chemical
shifts8 and the cross-correlation peaks between quadru-
The very small differences in chemical shifts for the
2
three stereoisomers yield a symmetrical H-{1H} spec-
trum, from which the identification of the two compo-
nents for each quadrupolar splitting is rather trivial,
but the distinction between the meso- and d,l-
diastereoisomers is not.
2
polar doublets sharing the same H-2H spin-spin cou-
pling. An important point to emphasise here is that
cross-correlation peaks are observable even when the
dipolar and scalar couplings are not resolved in the 1D
spectrum.11 We have employed a read pulse of 45° (and
not 90°) because for such an angle the autocorrelation
peaks are visible,11 and consequently we can draw
correlation squares on the 2D spectrum as for any
spin-1/2 homonuclear COSY-90 experiment. The use of
a 90° read pulse provides a similar spectrum, but the
diagonal peaks are simply no longer visible.11 The
analysis of the spectral information is therefore for-
mally identical to that performed on spin-1/2 COSY
spectra because any row or column of the square
exhibits the four components belonging to each pair of
The relative intensity of each pair of doublets (which
depends hopefully on the e.e. and d.e. of the mixture)
can afford information to allow the doublet assign-
ment. For instance, if the mixture is only diastereoiso-
merically enriched in meso compound (d.e. >33%), four
quadrupolar doublets over six possible of weak and
equal intensity (associated with the d,l enantiomers) will
be visible in the spectrum. In contrast, two small dou-
blets of weak and equal intensity (associated with the
meso compound) will be observed when the mixture is
only diastereoisomerically enriched in d,l species (d.e.

H. Villar et al. / Tetrahedron: Asymmetry 13 (2002) 1465–1475
1469
Figure 3. Spectroscopic principle for grouping of the various quadrupolar doublets observed in a mixture made of dideuterated,
meso-isomer and A,B enantiomeric pair in the chiral and achiral oriented phase. Coupling correlations are only expected to be
observed between doublets originating from deuterons belonging to the same stereoisomer. The magnitude of quadrupolar
M
M
A +B
2
A
B
splittings in the achiral phase is equal to ꢀDwMQ +M ꢀ=ꢀ(DwQ 1+DwQ 2)/2ꢀ for the meso, and ꢀDwQ ꢀ=ꢀ(DwQ1+DwQ2)/2ꢀ and
1
2
1
A
B
ꢀDwQA +B ꢀ=ꢀ(DwQ2+DwQ1)/2ꢀ for the active forms.
2
1
criminations, one quadrupolar doublet for the meso-
compound and two correlated quadrupolar doublets
for both enantiomers should be observed in the 2H-
{1H} spectrum if no overlap occurs. Indeed as the two
doublets associated with the d- and l-isomers in the
achiral phase originate from two pairs of two correlated
doublets in the CLC, a correlation peak between them
is also expected to be seen in the 2D spectrum (see Fig.
3). Consequently, the assignment of the meso-isomer
and the chiral pair becomes possible. This second step
is therefore crucial in this respect.
correlated doublets. This allows determination of the
pairs of quadrupolar doublets associated with the three
isomers. We can note that the doublets associated with
the deuterated impurities (labelled by an asterisk) are
also correlated by a dipolar coupling, and therefore
correspond to a CD2 probe.
If the information extracted from the COSY spectrum
in the CLC allows the differentiation between signals
for the three isomers, stereochemical assignment
through this method is however impossible and to
perform an eventual stereochemical attribution, we
need a further property that exists for the meso isomer
but not for the enantiomers. The answer is schemati-
cally given in Fig. 3. Indeed we can see that enan-
tiomeric discrimination involves two different species
while enantiotopic discrimination involves a single com-
pound. Thus, it becomes possible to distinguish meso-
from d,l-diastereoisomers (but not each enantiomer).
For this purpose we used an achiral polypeptide liquid
crystal, containing a racemic mixture of PBLG and
Fig. 4b shows the proton-decoupled deuterium COSY-
45 2D map of the mixture in the PBG system at 293 K.
The 2D spectrum was recorded using the same experi-
mental conditions as previously. As expected, the six
quadrupolar doublets in the CLC collapse now into
three doublets in the achiral phase and a unique corre-
lation peak is now observed between the inner and the
outer doublets. The meso/d,l enantiomers attribution is
therefore trivial in this second 2D spectrum.
PBDG (poly-g-benzyl-D-glutamate). Such solutions
denoted PBG in the following are also nematic but not
chiral. Consequently, solute molecules exchange rapidly
between the PBLG and PBDG vicinities, resulting in
identical average magnetic interactions for two enan-
tiomers or two enantiotopic elements.12 Due to the
elimination of both enantiomeric and enantiotopic dis-
The next step of the analysis consists of assigning the
pairs of doublets in the 2D spectrum recorded in the
chiral phase (see Fig. 4a). For this purpose, we must
keep in mind that the magnitude of quadrupolar split-
tings determined in the achiral phase corresponds in
principle to the algebraic average of the splittings mea-

1470
H. Villar et al. / Tetrahedron: Asymmetry 13 (2002) 1465–1475
M
M
2
ꢀDwMQ +M ꢀ=ꢀ(DwQ +DwQ )/2ꢀ
(1)
1
2
1
where DwMQ and DwMQ are the quadrupolar splittings of
enantiotopic deuterons. Similar relationships exist for
the enantiomeric pairs as we can write,
1
2
A
B
ꢀDwAQ +B ꢀ=ꢀ(DwQ +DwQ )/2ꢀ
(2)
1
2
1
2
and
ꢀDwAQ +B ꢀ=ꢀ(DwQ +DwQ )/2ꢀ
(3)
A
B
2
1
2
1
where DwAQ , DwQ , DwQ and DwBQ are the quadrupolar
splittings of diastereotopic deuterons in the enantiomers
noted A and B. Contrary to Eqs. (1)–(3) both involve
two quadrupolar splittings associated with each of
enantiomers and corresponding with diastereotopic
deuterons related by a plane of symmetry (see Fig. 2).
A
2
B
1
1
2
From these relationships we considered and calculated
all possible combinations of signs and experimental
quadrupolar splittings in the chiral phase, and then
retained the best fit, corresponding to the minimum
error between the experimental and calculated values.
The corresponding results obtained are collected in
Table 1 and the final assignment of doublets is shown
in Fig. 4a. The differences between the experimental
splitting and the best set of calculated values vary
between 10 and 17%. Although the relative discrepan-
cies are not negligible, we can safely retain this set
because all other solutions provide larger errors, and
therefore can be excluded. The discrepancies between
experimental and expected values presumably arise
from solvent effects in the racemic phase and the non-
strictly identical preparation of oriented solvent. Fur-
ther investigations are currently underway to
investigate these factors.
To answer the initial problem addressed at the begin-
ning of this work, we analysed a new sample (referred
to as mixture II) formed by cyclisation of enantioen-
riched (e.e.=93%) (4R,5S)-iodoester 1. Note that, in
the absence of racemisation, such a mixture is expected
to lead to meso-12, the (S,S)-enantiomer 13 and its
antipode ent-13 with a 93% e.e. in favour of 13.
Figure 4. (a) 61.4 MHz proton-decoupled deuterium COSY-
45 spectrum of mixture I (sample 1) recorded at 293 K in the
PBLG/CHCl3 phase. The 2D spectrum was acquired as a 256
(t1)×700 (t2) data matrix with 64 scans per t1 increments and
identical spectral widths in the two dimensions. The matrix
was zerofilled to 1 k (t1)×2 k (t2) data points prior to the
double Fourier transform. Sine-bell and square sine-bell filter-
ing were used in the F1 and F2 dimensions to enhance the
resolution of the magnitude spectrum. In F1 and F2 are
displayed the 1D spectrum. (b) Same experiment recorded in
the PBG/CHCl3 phase (sample 2). The quadrupolar doublets
due to each enantiotopic deuteron in the meso compound are
labelled M1 and M2. Those due to each diastereotopic
deuterons in each of the enantiomers are labelled A1, A2 and
B1, B2. Note that in each case the attribution of 1 and 2 is
arbitrary. Only the two quadrupolar doublets associated with
diastereotopic deuterons are correlated to each other.
Fig. 5a reports the 2H-{1H} spectrum of mixture II
expected to be enantioenriched. Here the spectrum
exhibits five quadrupolars doublets (over six possible).
This situation corresponds to the overlapping of two
quadrupolar doublets. Some impurities (labelled with
an asterisk) are also detected, but they do not interfere
with the analysis of other doublets. As before, the
analysis of doublets is not straightforward, and we
therefore applied the two-step strategy described previ-
ously. The peak assignment found is reported in Fig.
5a. As we expected, this assignment is similar to that
obtained for the mixture I even if the magnitude of
some quadrupolar splittings varied slightly. In particu-
lar we can note that the doublets associated with the
deuterons noted M1 and B1 are now fortuitously identi-
cal. These variations indicate that the orientational
ordering of solutes in the phase have slightly changed
compared to that of the mixture I. The difference in the
sample composition (see Table 2) and the nature of
sured in the chiral solvent. The relationship between
quadrupolar splittings measured in the chiral and the
achiral phase for the meso-compound is therefore,

H. Villar et al. / Tetrahedron: Asymmetry 13 (2002) 1465–1475
1471
Table 1. Assignment of quadrupolar doublets of the mixture I dissolved in the PBLG phase
ꢀDwAQ +B ꢀ (Hz)
ꢀDwMQ +M ꢀ (Hz)
ꢀDwAQ +B ꢀ (Hz)
ꢀDwQꢀ (Hz)
Error (%)
1
2
1
2
2
1
a
b
Enantiomeric pairs
ꢀDwQꢀ (Hz)
A1/B2
M1/M2
A2/B1
1791/12292
8192/17992
8692/20592
70
8591
14492
16192
17
10
10
130
145
a Experimental values measured in the PBLG phase.
b Experimental values measured in the PBG phase.
enables us to propose a chemical mechanism explaining
the racemisation process. Concerning the mechanism of
this SmI2-mediated ring closure, a radical anionic
sequence has already been proposed1 in which the
homoallylic radical A formed by an initial monoelec-
tronic reduction of the halide substrate rapidly cyclises
according to a 3-exo-trig process to the a-carbalkoxy-
substituted cyclopropylcarbinyl radical B (Scheme 4). A
second monoelectronic reduction of this electrophilic
radical followed by trapping of the resulting samarium
enolate C by the proton source ultimately leads to the
reaction product D. A similar sequence, involving 5-
and 6-exo-trig cyclisations at the radical stage had
previously already been proposed by Molander and
Harris13–15 for the related ring closures of z- or h-iodo-
a,b-unsaturated esters or amides to cyclopentanes or
cyclohexanes.
The main difference between these two reactions is that,
contrary to 5- and 6-exo-trig radical cyclisations, the
cyclisation of the homoallylic radical to cyclopropyl-
alkyl radical is a reversible process.16 According to the
kinetic measurements (in hydrocarbon solvents) and the
Arrhenius function for ring opening and ring closure of
carbalkoxysubstituted radicals species
B
and
A
Figure 5. (a) 61.4 MHz proton-decoupled deuterium spec-
trum of the mixture II dissolved in the PBLG/CHCl3 phase
(sample 3) at 293 K. No filtering was applied to enhance the
spectral appearance. (b) Deconvoluted spectrum. The d.e. and
e.e. values are 35% 5% and 20% 5%, respectively.
obtained by Beckwith and Bowry17 both reactions are
very rapid processes (kc=1.6×107 and k−c=1.2×107 s−1
at 353 K). More recent data presented by Newcomb
and co-workers18 obtained in THF reported that the
rate of ring-opening had been in fact underestimated,
and should be raised by a factor of approximately 50
for which the change of solvent from cyclohexane to
THF is only partially responsible. We are well aware
that the presence of samarium species could also
change these values appreciably. In any case, the exten-
sive racemisation at the g-carbon centre observed in the
present study shows that, under the conditions used (if
the reaction does follow the mechanism of Scheme 4),
the ring-opening of cyclopropyl radical B is more rapid
than its reduction to C by a second molecule of SmI2.
deuterated impurities, which seems to be different from
those contained in the mixture I, is responsible for this
effect.
The diastereoisomeric and enantiomeric composition
can now be determined using either an average for a
series of peak integrations in the spectrum or values
calculated from deconvoluted 1D spectrum (Fig. 5b).
The d.e. and e.e. values have been evaluated to 35%
5% and 20% 5%. Results from both integration meth-
ods are in full agreement. The d.e. value is therefore
consistent with that evaluated in the isotropic solvent
(meso/d,l ratio=0.5) and was not expected to change
compared to that of the mixture I. In contrast, the
enantiomeric excess of the mixture II is rather low, and
indicates a marked loss in enantiopurity.
As already stated,1 two carbanionic cyclisation pro-
cesses, in place of the radical mechanism, may be
envisaged to account for the cyclisation (Scheme 5). In
the first one, the cyclisation would involve an
intramolecular Michael-type addition of an organosa-
marium intermediate formed at the carbonꢀiodine
bond. In the second, a radical anion formed by one
electron reduction of the electron-withdrawn p-system
would displace the iodide atom by SN2 intramolecular
attack. In relation to these possibilities, the electro-
chemically-induced cyclisation of d-mesyloxy-a,b-
enoate 819 and SmI2-mediated reductive dimerisation of
The results given by the 2H-{1H} NMR in oriented
solvent unambiguously show that the SmI2-mediated
cyclisation of homochiral g-substituted d-halo-a,b-
unsaturated esters is not racemisation-free. This now

1472
H. Villar et al. / Tetrahedron: Asymmetry 13 (2002) 1465–1475
Table 2. Compositions of liquid-crystalline NMR samples investigated
Sample
Solute mixture Polymeric
solvent
DPa
PBLG/PBDG
Co-solvent
Solute (mg)b
Polymer (mg)b Co-solvent
(mg)b
% of polymer
in weight
1
2
3
4
I
I
II
II
PBLG
PBG
PBLG
PBG
562
562/914
562
CHCl3
CHCl3
CHCl3
CHCl3
12
12
20
20
100
51+51
99
400
402
450
451
19.5
19.7
17.4
17.5
562/914
51+50
a DP: Degree of polymerisation of polypeptide used.
b The accuracy on the weighting is 1 mg.
Scheme 4.
Scheme 5.
bis-enoate ester 920 leading to cyclopropane compounds
10 and 11, respectively, have been described in the
literature.
liquid crystals have clearly demonstrated that the cycli-
sation reaction is accompanied by extensive racemisa-
tion, probably as a result of faster ring reopening of
a-carbalkoxy cyclopropylcarbinyl radicals compared to
their reduction to a carbanionic species.
Returning to the present study, it should be noted that
the racemisation observed does not permit us to rule
out these two possibilities. Indeed, since SmI2 is a one
electron reductant, both possibilities entail the transient
formation of either an uncyclised (homoallylic) or a
cyclopropylcarbinyl radical, and therefore the possibil-
ity of an equilibration process between these two spe-
cies during the course of the reaction. However,
following Molander and Harris13 and Bennett and co-
workers,14a strong arguments against these carbanionic
alternatives have already been given.1 Moreover, the
fact that d-tosyloxy-a,b-enoates were found to be very
poor substrates21 for cyclisation in comparison with
their d-bromo analogues, would indicate that the radi-
cal anion mechanism is strongly disfavoured.
In the course of this work, we have developed an
efficient 2D NMR strategy for distinguishing meso-iso-
mers from d,l-diastereoisomers based on the introduc-
tion of a CD2 probe. This work emphasises the value
and analytical potential of NMR in weakly oriented
chiral media to the determination of enantiomeric and
diastereoisomeric compositions and to finding solutions
for applied stereochemical problems. The development
of such tools is of paramount importance considering
the difficulties involved in the control and evaluation of
the enantiomeric composition of chiral mixtures.
4. Experimental
4.1. Synthetic procedures
3. Conclusion
1
We have investigated the mechanism of the SmI2-medi-
ated cyclisation of l-halo-a,b-unsaturated esters. From
a synthetic point of view and for a better understanding
of the reaction mechanism it was of importance to
assess whether a homochiral g-substituted substrate
would retain its configuration during the cyclisation
process. The analysis of spectroscopic results using
proton decoupling deuterium NMR in weakly oriented
4.1.1. General. Routine H and 13C NMR spectra in
isotropic solvents were recorded on a Bruker AC 250 or
a Bruker AC 200 apparatus. Chemical shifts are
reported in l (ppm) referenced to the residual CHCl3
1H signal at 7.26 ppm and CDCl3 13C signal at 77.0
ppm. Optical rotations were measured with a Perkin-
Elmer 241 polarimeter. HPLC analyses were performed
on a Spectra-Physics P100-UV100 HPLC system with

H. Villar et al. / Tetrahedron: Asymmetry 13 (2002) 1465–1475
1473
detection at u=254 nm and equipped with a chiral
column Chiralcel OD-H from Daicel Chemical Indus-
tries Ltd. Mixtures of hexane and isopropanol were
used for elution. GC–MS analyses were carried out
on a OK1 DP 125 gas chromatograph equipped with
a CPSil quartz capillary column and connected to a
Riber Mag R10-10 mass detector. High resolution
mass spectra (HRMS) and electrospray mass spectra
were obtained on a Finnigan-MAT-95-S spectrometer.
All solvents used in the reactions were distilled under
were dissolved in dichloromethane (90 mL). A solu-
tion of rac-2-[(E)-(benzyloxycarbonyl)vinyl]-2,3-
dimethyloxirane (1.5 g, 6.46 mmol) and acetic acid
(1.13 mL, 1.95 mmol) in dichloromethane (10 mL)
was then rapidly added by syringe to the Schlenk
tube. After 15 min stirring at room temperature, total
conversion of the epoxide to homoallylic alcohol was
achieved (tlc analysis). The solvent was evaporated
and the crude residue was directly submitted to
column chromatography (silica gel, 80:20 cyclohexane/
ethyl acetate) to give (4R*,5R*)-4-methyl-5-hydroxy-
hex-(2E)-enoic acid benzyl ester as a colourless oil
argon and over
a
drying agent (P2O5 for
dichloromethane and benzophenone sodium for THF
and diethylether).
1
(1.36 g, 95%). H NMR (CDCl3, 200 MHz): l 7.42–
7.27 (m, 5H); 7.08–6.97 (dd, 3J=16.0 and 8.0 Hz,
4.1.2. 2-[(E)-(Benzyloxycarbonyl)vinyl]-2,3-dimethyloxi-
rane. rac-2-[(E)-(Benzyloxycarbonyl)vinyl]-2,3-dimethyl-
oxirane, rac-5. Benzyl (2E,4E)-4-methylhexa-2,4-
dienoate (4.54 g, 21 mmol) and 3-chloroperoxyben-
zoic acid (25–30% water content, 5.86 g, ca. 1.2
equiv.) were dissolved in chloroform (50 mL). The
reaction mixture was stirred for 8 h at room tempera-
ture and then extracted with aqueous 1N sodium
thiosulfate and 0.5N sodium hydroxide until pH ca.
9. The aqueous phases were extracted with
dichloromethane. The organic phases were dried over
MgSO4, filtered, evaporated and the residue was
purified by flash chromatography (silica gel, cyclohex-
ane/ethyl acetate 80:20) to afford a colourless oil
3
1H); 5.98–5.89 (dd, J=15.5 and 1.0 Hz, 1H); 5.19 (s,
2H); 3.83–3.71 (app. quint, 3J=6.5 Hz, 1H); 3.37
(broad s, 1H, OH); 2.36–2.25 (m, 1H); 1.18–1.15 (d,
3
3J=6.5 Hz, 1H); 1.11–1.08 (d, J=7.0 Hz). 13C NMR
(CDCl3): l 166.1; 151.3; 135.9; 128.5; 128.2, 121.3;
70.4; 66.1; 43.66; 20.35; 14.32. Electrospray (ESI, pos-
itive mode): 234.1 (M+, 18%); 252.1 (M+NH4 ).
+
Enantioenriched (4R,5R)-4-methyl-5-hydroxyhex-(2E)-
enoic acid benzyl ester 6 was obtained in a similar
manner, starting from (R,R)-2-[(E)-(benzyloxycar-
bonyl)vinyl]-2,3-dimethyloxirane. [h]2D0=+18.7 (c 0.41,
CH2Cl2). E.e.=93% (HPLC on Chiralcel ODH, hex-
ane/isopropanol 98:2).
1
(4.63 g, 95%). H NMR (CDCl3, 200 MHz): l 7.41–
7.34 (m, 5H); 6.85–6.79 (d, 1H, 3J=16.0 Hz, 1H);
4.1.5. (E)-5-Iodo-4-methyl-hex-2-enoic acid benzyl
ester. (4R*,5S*)-5-Iodo-4-methylhex-(2E)-enoic acid
benzyl ester, rac-1. To a magnetically stirred solution
of triphenylphosphine (1.12 g, 4.27 mmol) in
dichloromethane (5 mL) were successively added a
solution of I2 (1.1 g, 4.33 mmol) in dichloromethane
(2.5 mL) and a solution of imidazole (291 mg, 4.27
mmol) in dichloromethane (2.5 mL). An insoluble salt
formed in the medium. A solution of rac-(4R*,5R*)-
5-hydroxy-4-methylhex-(2E)-enoic acid benzyl ester
rac-6 1.00 g (4.27 mmol) in dichloromethane (2.5 mL)
was then added. The reaction mixture was stirred for
a few hours at room temperature while the conver-
sion of the alcohol was monitored by tlc analysis. As
soon as the reaction was complete (ca. 3 h), the reac-
tion mixture was concentrated on a Rotovapor and
the residue was resolved by column chromatography
(to be carried out preferentially in the dark) to afford
the iodide rac-1 as an oil (543 mg, 37% yield) and
diene 4 (370 mg) (silica gel, 80:20 cyclohexane/ethyl
acetate, Rf=0.4 (diene) and 0.5 (iodide)). (4R*,5S*)-
5-Iodo-4-methylhex-(2E)-enoic acid benzyl ester, rac-
6.12–6.06 (d, 3J=16.0 Hz, 1H); 5.20 (s, 2H); 3.03–
3
2.96 (q, J=4.5 Hz, 1H); 1.44 (s, 3H); 1.39–1.36 (d,
3J=5.45 Hz, 3H). 13C NMR (CDCl3): l 165.7; 150.4;
135.9; 128.4; 128.1; 121.0; 66.2; 61.4; 58.2; 14.8; 13.8.
HRMS (EI): calcd for C14H16O3 [M+] 232.1099, found
232.1107.
4.1.3.
(R,R)-2-[(E)-(Benzyloxycarbonyl)vinyl]-2,3-di-
methyloxirane 5. The epoxidation reaction was carried
out as described by Shi and co-workers2 by simulta-
neous addition via two syringe pumps of an aqueous
solution of oxone (2.3 g, 3.75 mmol) in aqueous
EDTA (4×10−4 M, 20 mL) and a solution of K2CO3
(2.3 g, 16.6 mmol) in water (20 mL) to a magnetically
stirred
buffered
biphasic
mixture
of
2:1
dimethoxyethane/acetonitrile (50 mL) and 33 mL of
aqueous solution containing Na2B2O4·10H2O (0.05 M)
and EDTA (4×10−4 M), with the diene (721 mg, 3.3
mmol), ketone 7 (215 mg, 0.83 mmol, 0.25 equiv.)
and tetrabutylammonium hydrogen sulfate (50 mg,
0.15 mmol). After column chromatography, the epox-
ide was obtained as an oil (727 mg, 95%). Its enan-
tiomeric purity (e.e.=94%) was measured by chiral
HPLC analysis (Chiralcel ODH, hexane/isopropanol
98:2). [h]2D0=+17.4 (c 0.41, CH2Cl2). HRMS (EI):
calcd for C14H16O3 [M+] 232.1099, found 232.1103.
1
1: H NMR (CDCl3, 250 MHz): l 7.42–7.28 (m, 5H);
6.98–6.89 (dd, 3J=15.5 and 8.0 Hz, 1H); 5.98–5.91
3
4
(dd, J=15.5 Hz, J=1.0 Hz, 1H); 5.22 (s, 2H); 4.26–
4.2 (dq, 3J=3.0 and 7 Hz, 1H); 2.22–2.15 (m, 1H);
1.88–1.86 (d, 3J=6.5 Hz, 3H); 1.17–1.14 (d, 3J=7.0
Hz, 3H). 13C NMR (CDCl3): l 165.9; 146.9; 135.6;
128.5; 128.1; 123.6; 66.2; 42.7; 22.2; 21.7; 19.0. GC/
MS (EI): 216 (M+−HI, 100%). Similarly, enantiomeri-
cally enriched (4R,5S)-5-iodo-4-methylhex-(2E)-enoic
acid benzyl ester 1 was prepared from (4R,5S)-5-
hydroxy-4-methylhex-(2E)-enoic acid benzyl ester.
4.1.4. (E)-4-Methyl-5-hydroxyhex-2-enoic acid benzyl
ester.
(4R*,5R*)-4-Methyl-5-hydroxyhex-(2E)-enoic
acid benzyl ester, rac-6. The reaction was completed
under an argon atmosphere: dimethylamineborane
complex (380 mg, 6.45 mmol) and palladium (tetra-
kis)triphenylphosphine (5 mol%, 373 mg, 0.325 mmol)

1474
H. Villar et al. / Tetrahedron: Asymmetry 13 (2002) 1465–1475
4.1.6. SmI2-induced cyclisation of racemic (4R*,5S*) and
enantiomerically enriched (4R,5S)-5-iodo-4-methylhex-
(2E)-enoic acid benzyl ester. 0.1N solution of SmI2 in
THF. In a graduated Schlenk tube under an argon
atmosphere, 1,2-diiodoethane (2.82 g, 10 mmol) was
diluted to a volume of 100 mL in anhydrous and
degassed THF. This solution was slowly transferred,
through a cannula into a second Schlenk tube contain-
ing metallic samarium (1.8 g, 1.19 mmol) and a stirring
bar. The yellow suspension was magnetically stirred at
room temperature for 2 h to give a deep blue solution.
This solution was stored for not more than a week
under argon and in the dark.
4.2. NMR spectroscopy in oriented solvents
The chiral and non chiral liquid crystalline NMR sam-
ples were prepared using the homopolypeptide (ca. 100
mg, commercially available from Sigma), the deuterated
solutes (ca. 10 mg) and organic co-solvent (ca. 400 mg)
directly weighed into a 5 mm o.d. NMR tube. The
exact compositions of each oriented NMR sample are
given in Table 2. All NMR tubes were sealed to avoid
solvent evaporation and centrifuged back and forth
until an optically homogeneous birefringent phase was
obtained. Various experimental details on the method
can be found in Ref. 6a. The 2H-{1H} 1D and 2D
experiments were performed at 9.4 T on a Bruker DRX
400 high-resolution spectrometer equipped with an
inverse multinuclear probe (BBI) operating at 61.4
MHz for deuterium and with a standard variable tem-
perature unit (BVT 3000). In order to remove the
proton–deuterium scalar and dipolar couplings, the
protons were broadband decoupled using WALTZ-16
composite pulse sequence. Other experimental NMR
parameters or details are given in the figure captions.
Cyclisation: Either (4R*,5S*)- or enantiomerically
enriched (4R,5S)-5-iodo-4-methylhex-(2E)-enoic acid
benzyl ester (200 mg, 0.58 mmol) together with tert-
butanol (165 mL, 1.75 mmol), were diluted with THF (2
mL). The resulting solution was added dropwise over a
period of 1 min and under stirring in a Schlenk tube
(argon atmosphere) containing a solution of SmI2 in
THF (0.1 M, 14.5 mL, 2.5 equiv.). After 2 h, during
which the deep blue colour of the SmI2 solution pro-
gressively discharged, the reaction mixture was
quenched with 0.1N HCl and extracted with diethyl
ether. The organic phase was washed with aqueous
sodium thiosulfate solution and dried over magnesium
sulfate.
References
1. David, H.; Afonso, C.; Bonin, M.; Doisneau, G.;
Guillerez, M.-G.; Guibe´, F. Tetrahedron Lett. 1999, 40,
8557.
2. Frohn, M.; Dalkiewicz, M.; Tu, Y.; Wang, Z.-X.; Shi, Y.
NMR and HRMS characterisation of cyclised propanes:
The crude products were directly purified by column
chromatography (silica gel, 80:20 cyclohexane/ethyl
acetate) to give 100 mg (80% overall yield) of a ca. 1:2
mixture of unseparated meso 2-(trans-2-trans-3-
dimethylcyclopropyl)acetic acid benzyl ester 2 and d,l-
2-(cis-2-trans-3-dimethylcyclopropyl)acetic acid benzyl
ester [3+ent-3].
J. Org. Chem. 1998, 63, 2948.
3. David, H.; Dupuis, L.; Guillerez, M.-G.; Guibe´, F. Tet-
rahedron Lett. 2000, 41, 3335.
4. Eliel, A. L.; Wilen, A. H. Stereochemistry of Organic
Compounds; John Wiley and Sons: New York, 1994; p.
493.
5. Kainosho, M.; Ajisaka, K.; Pirkle, W.; Beare, S. D. J.
Am. Chem. Soc. 1972, 94, 5924.
6. (a) Sarfati, M.; Lesot, P.; Merlet, D.; Courtieu, J. Chem.
Commun. 2000, 2069 and references cited therein; (b)
Aroulanda, C.; Sarfati, M.; Courtieu, J.; Lesot, P. Enan-
tiomer 2001, 6, 281.
1H NMR (CDCl3, 400 MHz): peak assignments to
respectively the d,l- and the meso-isomers were estab-
lished using COSY 2D proton NMR experiments d,l-2-
(cis-2-trans-3-dimethyl-cyclopropyl)acetic acid benzyl
ester 3: l 7.33–7.27 (m, 5H); 5.15 (s, 2H); 2.33 (d,
3J=7.35 Hz, 2H); 1.01–0.96 (two d, 3J= 6.45 Hz, 3H+3
H); 0.76–0.71 (m, 1H); 0.59–0.54 (m, 1H); 0.20–0.16 (m,
1H). meso-2-(cis-2-trans-3-Dimethylcyclopropyl)acetic
acid benzyl ester 2: 7.33–7.27 (m, 5H); 5.13 (s, 2H); 2.24
7. Meddour, A.; Canlet, C.; Blanco, L.; Courtieu, J. Angew.
Chem., Int. Ed. 1999, 38, 2391.
8. (a) Lesot, P. D.; Merlet, D. A.; Loewenstein, A. J.;
Courtieu, J. Tetrahedron: Asymmetry 1998, 9, 1871; (b)
Merlet, D.; Ancian, B.; Courtieu, J.; Lesot, P. J. Am.
Chem. Soc. 1999, 121, 5249.
9. Meddour, A.; Canet, I.; Loewenstein, A.; Pe´chine´, J.-M.;
Courtieu, J. J. Am. Chem. Soc. 1995, 116, 9652.
10. (a) Merlet, D.; Emsley, J. W.; Lesot, P.; Courtieu, J. J.
Chem. Phys. 1999, 111, 6890; (b) Aroulanda, C.; Merlet,
D.; Courtieu, J.; Lesot, P. J. Am. Chem. Soc. 2001, 123,
12059.
11. Theoretical considerations on spin-1 COSY 2D experi-
ments will be published elsewhere.
12. Canlet, C.; Merlet, D.; Lesot, P.; Meddour, A.; Loewen-
stein, A.; Courtieu, J. Tetrahedron: Asymmetry 2000, 11,
1911.
3
(d, J=7.15 Hz, 2H); 0.99 (s, 6H); 0.59-0.54 (m, 2H);
0.39–0.33 (m, 1H). Electrospray (ESI, positive mode) 2
and 3: 219.3 (23%, MH+); 236.3 (34%, M+NH4 ); 241.3
+
(75%, M++Na); 242.3.
4.1.7. Reduction of esters to deuterated alcohols. The
crude (unchromatographed) products of cyclisation
(theoretical amount: 0.58 mmol) were dissolved in THF
(1 mL) and added to a solution of LiAlD4 (20 mg, 0.48
mmol) in THF (1 mL) at 0°C. After stirring for 1 h the
reaction mixture was hydrolysed with 1N aqueous HCl
and extracted several times with ethyl acetate. The
organic phases were washed with water, dried over
MgSO4 and the solvent was evaporated. 1H and 2H
NMR analyses were carried out directly on the crude
residue.
13. Molander, G. A.; Harris, C. R. J. Org. Chem. 1997, 62,
7418.
14. See also: (a) Bennett, S. M.; Biboutou, R. K.; Zhou, Z.;
Pion, R. Tetrahedron 1998, 54, 4761; (b) see Ref. 15c, pp.
767–768.

H. Villar et al. / Tetrahedron: Asymmetry 13 (2002) 1465–1475
1475
15. For recent reviews concerning sequenced reactions with
samarium iodide, see: (a) Molander, G. A.; Harris, C. R.
Tetrahedron, 1998, 54, 3321; (b) Molander, G. A.; Harris,
C. R. Chem. Rev. 1996, 96, 307; (c) Krief, A.; Laval, A.-M.
Chem. Rev. 1999, 99, 745–777.
16. Nonhebel, D. C. Chem. Soc. Rev. 1993, 347.
17. Beckwith, A. L. J.; Bowry, V. W. J. Am. Chem. Soc. 1994,
116, 2710.
18. Horner, J. H.; Tanaka, N.; Newcomb, M. J. J. Am. Chem.
Soc. 1998, 120, 10379.
19. Gassman, P. G.; Lee, C. Tetrahedron Lett. 1989, 30, 2175.
20. Inanaga, J.; Handa, Y.; Tabuchi, T.; Otsubo, K.;
Yamaguchi, M.; Hanamoto, T. Tetrahedron Lett. 1991, 32,
6557.
21. David, H. The`se de doctorat en Sciences, Universite´
Paris-Sud, Orsay, 1999.