A novel mixed dimer of a norephedrine-derived chiral lithium amide and 2-lithium-1-methylimidazole, and catalytic enantioselective deprotonation of cyclohexene oxide

Article abstract of DOI:10.1002/1521-3765(20011015)7:20<4368::AID-CHEM4368>3.0.CO;2-K

Improved stereoselectivity has been obtained by using 2-lithium-1-methylimidazole, 2, as a replacement for lithium diisopropylamide (LDA) as a bulk base in catalytic deprotonations. The chiral lithium amide 6 of (1R,2S)-N-methyl-1-phenyl-2-pyrrolidinylpropanamine, 5, has been found to deprotonate cyclohexene oxide 3 in the presence of compound 2 to yield (S)-cyclohex-2-en-1-ol, 4, in 96% ee. Compound 2 is a carbenoid species conveniently generated from nBuLi and 1-methylimidazole, 1. The base 2 has also been found to play a more intimate role in the deprotonation. Investigations by ¹H, ⁶Li and ¹³C NMR of the ⁶Li¹⁵N isotopologue 8 of 6 have shown that 6 is homodimeric in THF and that, in the presence of 2, it forms a novel heterodimer 10. This heterodimer is found to be the dominant reagent in the initial state, rather than the homodimer of 6. Computational investigations with PM3 and B3LYP/6-311+G(d,p) have shown possible structures of the heterodimers, as well as the role of THF and 1 in the solvation of the dimers. The results are in line with the NMR results. Favoured complexes in the equilibria between homo- and heterocomplexes are also reported.

Full text of DOI:10.1002/1521-3765(20011015)7:20<4368::AID-CHEM4368>3.0.CO;2-K

FULL PAPER
A Novel Mixed Dimer of a Norephedrine-Derived Chiral Lithium Amide and
2-Lithium-1-methylimidazole, and Catalytic Enantioselective Deprotonation
of Cyclohexene Oxide
Mohamed Amedjkouh, Daniel Pettersen, Sten O. Nilsson Lill, Öjvind Davidsson,
and Per Ahlberg*[a]
Abstract: Improved
stereoselectivity
1. The base 2 has also been found to play
a more intimate role in the deprotona-
tion. Investigations by 1H, 6Li and
heterodimer is found to be the dominant
reagent in the initial state, rather than
the homodimer of 6. Computational
investigations with PM3 and B3LYP/
6 ± 311 ‡ G(d,p) have shown possible
structures of the heterodimers, as well
as the role of THF and 1 in the solvation
of the dimers. The results are in line with
the NMR results. Favoured complexes in
the equilibria between homo- and
heterocomplexes are also reported.
has been obtained by using 2-lithium-1-
methylimidazole, 2, as a replacement for
lithium diisopropylamide (LDA) as a
bulk base in catalytic deprotonations.
The chiral lithium amide 6 of (1R,2S)-N-
methyl-1-phenyl-2-pyrrolidinylpropan-
amine, 5, has been found to deprotonate
cyclohexene oxide 3 in the presence of
compound 2 to yield (S)-cyclohex-2-en-
1-ol, 4, in 96% ee. Compound 2 is a
carbenoid species conveniently generat-
ed from nBuLi and 1-methylimidazole,
6
13C NMR of the Li/15N isotopologue 8
of 6 have shown that 6 is homodimeric in
THF and that, in the presence of 2, it
forms a novel heterodimer 10. This
Keywords: asymmetric catalysis
´
chiral lithium amide ´ heterodimer
´ multinuclear (1H, 6Li, 13C) NMR ´
semiempirical calculations
for this is presumably that the competitive, nonstereoselective
deprotonation by LDA yields some racemic product.
Introduction
Homochiral lithium amides are being developed and used for
highly stereoselective deprotonations of, for example, epox-
ides to yield enantiomers of allylic alcohols[1±3] for use in such
syntheses as those of biologically active compounds.[4±9] Much
remains to be known about the nature of these reagents in
solution. Some have been shown to aggregate to homodimers
and are thus the reagents in the deprotonations.[10±14] There
have been a number of attempts to run stereoselective
deprotonations under catalytic conditions.[15±17] The chiral
lithium amide has been used in catalytic quantities and the
less-reactive lithium diisopropyl amide (LDA) has been used
as a bulk base by which the chiral lithium amide is regenerated
from the chiral amine produced.[18±21] However, under these
conditions the enantioselectivity obtained is usually lower
than that obtained under noncatalytic conditions. The reason
To improve the degree of stereoselectivity in catalytic
deprotonations, access to bulk bases that have lower kinetic
basicity than LDA but comparable thermodynamic basicity is
required. In our search for such bases, we made use of
1-methylimidazole, 1, as a precursor. This compound under-
goes carbon deprotonation at the C2 position with, for
example, nBuLi to yield 2-lithium-1-methylimidazole, 2,
which is a carbenoid species (Scheme 1).[22, 23]
Since proton transfer to carbon is usually slower than to
more electronegative atoms, such as nitrogen, compound 2
was expected to show lower kinetic basicity than a nitrogen
base of similar thermodynamic basicity. In THF, we found by
NMR that 2 has a basicity comparable to LDA. As expected,
and in contrast to LDA, compound 2 did not measurably
cause any deprotonation of cyclohexene oxide 3 to yield
cyclohex-2-en-1-ol, 4, even after very long reaction times.
These properties indicated to us that 2 should be useful as a
bulk base in catalytic asymmetric deprotonations.
[a] Prof. P. Ahlberg, Dr. M. Amedjkouh, D. Pettersen, S. O. Nilsson Lill,
Dr. Ö. Davidsson
Organic Chemistry, Department of Chemistry
Göteborg University, 41296 Göteborg (Sweden)
Fax : (‡46)31-772-2908
In this paper, we also report on the advantage of compound
2 as a bulk base and on its more intimate role through the
formation of a reagent heterodimer consisting of a monomer
of 2 and a monomer of a homochiral lithium amide. The
E-mail: Per.Ahlberg@oc.chalmers.se
Supporting information for this article is available on the WWW under
http://wiley-vch.de/home/chemistry/ or from the author.
1
structure of the novel reagent has been elucidated from H,
4368
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0720-4368 $ 17.50+.50/0
Chem. Eur. J. 2001, 7, No. 20

4368±4377
the one with LDA. That the same ee was obtained as for the
noncatalytic reaction of 6 with 3 suggests that 2 was operating
as predicted. However, the use of the lithium carbenoid 2 in
equimolar amounts to 6 in THF to deprotonate 3 resulted, to
our surprise, in a deprotonation with increased enantioselec-
tivity. Compound (S)-4 was formed in 96% ee, and this
enantiomeric excess remained constant during the whole
reaction, as shown by chiral GC. This suggests that compound
2 is not functioning just as a bulk base, but is participating
intimately in the deprotonation reaction.
These findings prompted us to investigate the nature of
reagent 6 in the presence and absence of the lithium
carbenoid 2. For this purpose 7, the 15N isotopologue of 5,
and 8, the 6Li/15N isotopologue of 6, were prepared and
1
studied in [D8]THF by multinuclear NMR spectroscopy. H,
6Li and 13C NMR spectra were obtained at different temper-
atures, and selected 1H and 13C NMR chemical shifts are
shown in Table 1.
The 6Li NMR spectrum of 8 at 808C mainly showed two
triplets of equal intensity but with different splittings (Fig-
ure 1a). This shows that two different lithium atoms are
present and that each of them is coupled, with the same
coupling constant, to two labelled nitrogens (15N). The
coupling constants (J(6Li,15N)) for the two lithiums were
found to be 5.8 Hz and 3.8 Hz. These results suggest that 6 is
mainly present as homodimer 9 with nonequivalent lithium
atoms in THF, and that one of the lithiums is tricoordinated
while the other is tetracoordinated.[10] Possible structures of 9
that have different nitrogen configurations are shown in
Scheme 3.
Scheme 1. Structures of labelled and nonlabelled precursors and their
lithiated counterparts, shown together with homodimers 9 and 11 and the
heterodimer 10.
6Li and 13C NMR spectroscopy. Computational studies have
shown possible structures of the heterodimers as well as the
role of THF and 1 in the solvation of the dimers. Favoured
complexes in the equilibria between homo- and heterocom-
plexes are also reported. Improved enantioselectivity, ob-
tained with the new heterodimer 10, is also reported.
Computational studies by PM3 of comparable states of
dimers with and without specific THF solvation show that the
state containing 9b ´ THF has the lowest enthalpy and, thus,
may be the dimer isomer observed in solution.[24]
1
Table 1. Assigned H and 13C NMR chemical shifts at 808C in [D8]THF for selected
Results and Discussion
atoms in 1, 2, 7, 8 and 10.
Recently, the norephedrine-derived diamine 5 was
synthesised, and the corresponding lithium amide 6 was
found to deprotonate 3 in THF to yield (S)-4 in 93% ee
(Scheme 2).[24, 25]
1
2
7
8
10
d
d
d
d
d
H1 at C1
3.67
7.44
7.03
6.85
33.1
139
3.69
±
6.81
6.75
36.0
204
H2 at C2
H3 at C3
H4 at C4
C1
Scheme 2. Enantiomerically specific deprotonation of 3.
33.0
201
C2
C3
C4
C5
C6
C7
C8
C9
130
121
129
117
128
116
46.0
77.9
150
When this reaction was run under catalytic condi-
tions with a reaction mixture composed of 6 (0.02m),
LDA (0.2m) and 3 (0.1m) in THF at 208C, the ee
decreased to 22%. In contrast, the catalytic reaction
mixture composed of 6 (0.2m), 2 (0.2m) and 3 (0.1m)
gave (S)-4 in 93% ee, but the reaction was slower than
35.8
67.9
143
67.1
52.9
45.3
77.4
149
72.6
54.3
71.7
54.0
Chem. Eur. J. 2001, 7, No. 20
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0720-4369 $ 17.50+.50/0
4369

P. Ahlberg et al.
FULL PAPER
Scheme 3. Possible structures of homodimers of 6.
monomer of 6Li-labelled 2. In such a dimer, each 6Li is
1
coupled to only one 15N. Selected H and 13C NMR chemical
shifts of 10 are shown in Table 1. The two doublets shown in
=
1
Figure 1c coalesce at 558C (DG218 K ˆ 11.1 Æ 0.3 kcalmol ).
Computationally investigated structures of the heterodimer
10 are displayed in Scheme 4.
Evidently, the heterodimer 10 is more stable than both 11,
the homodimer of 2, and 9, the homodimer of 6. Addition of a
further 0.25 equiv of 1 only resulted in shifts of the doublets to
lower field, since no 8 is available for deprotonation and
heterodimer formation (Figure 1d). The shifts indicate that 1
is solvating the heterodimer. The addition of altogether two
equivalents of 1 shifted the doublets further (Figure 1e); now
the one with the larger coupling constant (J(6Li,15N) ˆ 5.6 Hz)
appeared at lower field than the one with the smaller coupling
constant (J(6Li,15N) ˆ 3.8 Hz).
Figure 1. 6Li NMR spectra obtained at 808C of solutions of 8 in [D8]THF
in the absence and presence of added 1: a) 0.27m 8; b) 0.25 equiv of 1
added; c) 0.50 equiv of 1 added; d) 0.75 equiv of 1 added; e) 2.0 equiv of 1
added.
Addition of 0.25 equivalents (equiv) of the imidazole 1 to a
solution of 8 (0.27m) in [D8]THF resulted in dramatic changes
6
in the Li NMR spectrum. The intensity of the two triplets
decreased and new signals, which overlapped the triplets,
appeared (Figure 1b). Further addition of 0.25 equiv of 1 gave
a spectrum that mainly consisted of two doublets of about
equal intensity (Figure 1c). A separate experiment had shown
This novel heterodimer was also prepared by starting with a
solution of 1 (0.27m) in [D8]THF. When an equivalent amount
6
of nBu[6Li] (about 10m) was added, the Li NMR mainly
6
6
that the Li NMR spectrum of the Li-lithiated 1, that is the
6Li isotopologue of 2, only shows a singlet in [D8]THF at
808C (cf. below). Thus, the two doublets formed upon
addition of the 0.5 equiv of 1 together with 1H and 13C NMR
spectra indicate that the lithium amide 8 had deprotonated
the methylimidazole 1 at the 2 carbon to yield the diamine 7,
and that a new type of isotopically labelled dimer 10 had
formed, which was composed of one monomer of 8 and one
displayed a singlet (Figure 2a).
1H and 13C NMR spectra showed that this signal is due to
the 6Li-labelled lithium carbenoid 2. After addition of
0.25 equiv of diamine 7, the 6Li NMR spectrum shown in
Figure 2b was obtained; besides the singlet originating from
6Li-labelled 2, it shows the two doublets emanating from the
labelled mixed dimer 10. The measured coupling constants
(J(6Li,15N)) are 5.5 Hz and 4.0 Hz. Evidently, the lithium
4370
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0720-4370 $ 17.50+.50/0
Chem. Eur. J. 2001, 7, No. 20

Novel Dimer for Enantioselective Catalysis
4368±4377
carbenoid 2 is a strong enough base to deprotonate the
diamine 5 and yield a monomer of lithium amide 6 as part of a
mixed dimer, together with a monomer of 2. The deprotona-
6
tion of 5 by 2 produces the 1-methylimidazole, 1. The Li
NMR chemical shifts indicate that both the heterodimer 10
and the remaining 2 are solvated by 1. The spectrum after
addition of 0.5 equiv of 7 in total displayed essentially only
two doublets (shifted); this showed that only the heterodimer
10 was present (Figure 2c).
The reactivity of 10 is similar to that of 6. Recently, it has
been shown that the rate-limiting transition state in the
deprotonation of 3 by 6 in THF is composed of a dimer of 6
and one molecule of 3.[24] Possibly, the 10-promoted elimi-
nation of 3 is making use of a rate-limiting transition state
built from the heterodimer 10 and one molecule of 3. The
scope and limitations of the above findings in asymmetric
deprotonations are currently being explored. Below, results of
PM3 and DFT (B3LYP/6 ± 311 ‡ G(d,p)) investigations of
possible solvated and nonsolvated hetero- and homodimeric
structures together with equilibrium enthalpies are discussed.
Computational investigations: Some of the PM3- and DFT-
calculated structures are shown in Figures 3 ± 6, below. Others
are found in the Supporting Information.
Of the unsolvated heterodimers (Scheme 4), 10a and 10c
were found to have the lowest, and almost identical, enthalpy.
These two isomers differ in the orientation of the imidazoloid
ring. In each isomer, the two lithium atoms are part, together
with the carbenoid carbon, the imine nitrogen of the
imidazoloid and the amide nitrogen, of a five-membered ring
in an envelope conformation. LiA bonds to both the amide
nitrogen and the carbenoid carbon, while LiB is bonded both
to the amide and imine nitrogens. LiB also has a long bond (ca.
2.6 Š) to the carbenoid carbon (Figure 3). In 10a, LiB is also
internally coordinated to the pyrrolidine nitrogen, while in
10c LiA is coordinated to the pyrrolidine nitrogen. Thus in
10a, one lithium is dicoordinated and the other tetracoordi-
nated, while in 10c both lithiums are tricoordinated. On
optimising 10a with DFT, the LiB carbenoid-carbon bond is
elongated from 2.6 Š to 2.7 Š. In contrast, the other bonds to
Li are shortened, for example the LiA amide-nitrogen bond is
shortened by 0.1 Š.
Scheme 4. Possible structures of heterodimers built from a monomer of 6
and a monomer of 2.
Upon specific monosolvation by THF at the PM3 level, the
solvent coordinates preferably to the lithium that is not
internally coordinated to the pyrrolidine nitrogen. In 10c ´
THF, the long lithium carbenoid-carbon bond is elongated
(ca. 2.8 Š), while in 10a ´ THF the corresponding C Li bond
is almost unchanged (ca. 2.6 Š). Compounds 10a ´ THF and
10c ´ THF are of similar energy. Specific monosolvation by
1-methylimidazole, 1, has a similar effect on the structure as
found for THF, although the solvation enthalpy is much larger
1
1
for 1 ( 21 kcalmol ) than for THF ( 7 kcalmol ). The
solvation enthalpies calculated by using B3LYP/6 ± 311 ‡
1
G(d,p)//PM3 are 14 and 10 kcalmol , respectively. Di-
solvation by THF yields only an additional solvation enthalpy
1
of about
2 kcalmol
by using PM3 compared with
6 kcalmol by DFT. PM3 is known[26±28] to underestimate
solvation enthalpies of ethereal solvents by 3 ± 4 kcalmol , so
taking the entropy of solvation[27, 28] (ca. 5 kcalmol ) into
1
Figure 2. 6Li NMR spectra obtained at 808C of solutions of 6Li-labelled
2 in [D8]THF in the absence and presence of added 7: a) 0.27m 6Li-labelled
2; b) 0.25 equiv of 7 added and c) 0.50 equiv of 7 added.
1
1
Chem. Eur. J. 2001, 7, No. 20
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0720-4371 $ 17.50+.50/0
4371

P. Ahlberg et al.
FULL PAPER
Figure 3. PM3- and DFT-calculated non- and monosolvated heterodimer structures. Hydrogens are omitted for clarity.
account as well predicts that the heterodimers are most likely
a mixture of both mono- and disolvated species in THF. Upon
disolvation, it is energetically favourable to break the
coordination to the pyrrolidine nitrogen (Figure 4). For
example, 10e ± 10g were found to have more negative
solvation enthalpies than structures that had pyrrolidine
coordination. Several structures were found to be of almost
equal enthalpy (cf. Figure 4 and Supporting Information). The
4372
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0720-4372 $ 17.50+.50/0
Chem. Eur. J. 2001, 7, No. 20

Novel Dimer for Enantioselective Catalysis
4368±4377
Figure 4. PM3-calculated THF-disolvated heterodimer structures. Hydrogens are omitted for clarity.
solvation enthalpy from a second molecule of 1 was calculated
to be 15 kcalmol . To evaluate the PM3-calculated solva-
tion enthalpies for 1, PM3 enthalpies were compared with
single-point energies calculated by using B3LYP/6 ± 311 ‡
G(d,p)//PM3 on the 1-mono- and -disolvated monomers of
1
CH3Li and NH2Li. It was found that PM3 overestimates the
1
solvation enthalpy of 1 by 2 ± 8 kcalmol . If an entropy of
1
solvation of about 5 kcalmol is assumed, the heterodimers
Chem. Eur. J. 2001, 7, No. 20
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0720-4373 $ 17.50+.50/0
4373

P. Ahlberg et al.
FULL PAPER
10 are expected to be essentially completely disolvated by 1 as
long as >2 equiv of 1 are present (Figure 5). DFT, on the
other hand, gives only a second solvation enthalpy of
3 kcalmol for 10e ´ 12. In the PM3-calculated disolvated
structures, the carbenoid carbon is coordinated to only one
lithium. Also, with two solvating molecules of 1, it was found
that breaking the pyrrolidine nitrogen coordination was
energetically favourable. DFT calculations on 10a ´ 12 and
10e ´ 12 showed that the isomer without pyrrolidine-nitrogen
1
1
coordination was 0.3 kcalmol more stable. 10e ´ 12 was
Figure 5. PM3-calculated heterodimer structures disolvated by 1. Hydrogens are omitted for clarity.
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001 0947-6539/01/0720-4374 $ 17.50+.50/0
4374
Chem. Eur. J. 2001, 7, No. 20

Novel Dimer for Enantioselective Catalysis
4368±4377
calculated to have the lowest enthalpy also at PM3 level. The
distance from the amine nitrogen to the nearest lithium in
10e ´ 12 was calculated to be 4.7 Š.
thiazol-2-ylidene-glycoldimethyl ether)2.[22] The calculated
energy difference between the isomers 11a and 11b is almost
1
equal at both DFT and PM3 levels (0.24 and 0.42 kcalmol ,
PM3 and B3LYP/6 ± 311 ‡ G(d,p) optimised structures of
11 are shown in Figure 6. At PM3 level, 11 involves bonding of
Li to both C and N within each monomer. At DFT level, the
Li carbenoid-carbon bond is elongated from 2.4 Š (at PM3
level) to about 2.7 Š. Other bonds to lithium are shorter at
DFT than at PM3 level. The calculated structure is similar to
the crystal structure of the homodimer (3-lithium-4-tert-butyl-
respectively).
In Scheme 5, some results of the studies of equilibria
between homodimers 9 and 11 and heterodimers 10 with
or without solvation by THF or 1-methylimidazole 1 are
shown. Relative energies have been calculated both at the
PM3//PM3 and at the B3LYP/6 ± 311 ‡ G(d,p)//PM3 level
of theory.
Figure 6. PM3- and DFT-calculated structures of homodimers of 6 and 2. The homodimer 9 is monosolvated by THF and disolvated by 1. Hydrogens are
1
omitted for clarity. 11a and 11b are isomers, and their relative energies are given in kcalmol
.
Chem. Eur. J. 2001, 7, No. 20 ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0720-4375 $ 17.50+.50/0
4375

P. Ahlberg et al.
FULL PAPER
Experimental Section
General: All syringes and glass vessels
used were dried overnight in a vacuum
oven (508C) before being transferred
into
a glovebox (Mecaplex GB80
equipped with a gas-purification sys-
tem that removes oxygen and mois-
ture) under a nitrogen atmosphere.
The typical moisture content was less
than 0.5 ppm. All handling of the
lithium compounds was carried out in
the glovebox with gas-tight syringes.
The solvent used, THF, was distilled
from sodium and benzophenone. The
rearrangements of 3 were performed
in a nitrogen atmosphere. The concen-
tration of the commercially available
nBuLi (ca. 2.5m in hexanes, Acros)
was determined by titration.[24]
[D8]THF was stored over molecular
sieves (4 Š) in the glovebox. Com-
1
Scheme 5. PM3- and DFT-calculated enthalpies in kcalmol of states in equilibrium involving dimers with
varying degrees of solvation. State K is used as a reference state.
pound 1 (Sigma ± Aldrich) was distilled over CaH2. nBu[6Li] was prepared
as previously described.[29] Methyl[15N]amine hydrochloride (98% ‡ iso-
topic and chemical purity) was purchased from Cambridge Isotope
Laboratories. All GC analyses were run on a chiral stationary-phase
column (CP-Chirasil-DEX CB, 25m, 0.32 mm) from Chrompack. The
column was held at 958C (injector 2258C, detector 2508C) with helium
(2 mLmin 1) as a carrier gas. tR(3) ˆ 3.25 min, tR((S)-4) ˆ 7.45 min, tR((R)-
4) ˆ 7.90 min.
Of all states in Scheme 5, state K, which involves solvation
of each Li in 10e by one molecule of 1, is the most stable one
and is used as a reference state. In the equilibrium between
the nonsolvated dimers, state E is favoured over A by
1.7 kcalmol 1 with PM3 and by 7.1 kcalmol 1 with DFT. THF
1
solvation of 11a and 9b makes state B 8.9/16.4 kcalmol
more stable than state A (Figure 6). The solvation energies
Synthesis of (1R,2S)-N-methyl-1-phenyl-2-pyrrolidinylpropan[15N]amine
7: A modification of a procedure described earlier for the synthesis of 5 was
used for the synthesis of 7.[24, 30] After purification by short-path distillation,
the precursor amino alcohol (2.05 g, 10 mmol) was dissolved in THF under
a nitrogen atmosphere in a flask equipped with septa and cooled to 08C in
an ice bath. Triethylamine (4.25 mL, 30 mmol) was added to the stirred
solution, and methanesulfonyl chloride (1.48 mL, 20 mmol) was added
dropwise with a syringe over 20 min while a yellow precipitate formed. The
ice bath was removed and the mixture was stirred for another 1.5 h at RT.
Triethylamine (7.1 mL, 50 mmol) and methyl[15N]amine hydrochloride (1 g,
15 mmol) were added sequentially to the reaction mixture. Upon addition
of water (3 mL), the remaining precipitate dissolved; this resulted in a clear
solution, which was stirred at RT for another 48 h. The aqueous phase was
extracted with diethyl ether (3 Â 40 mL), and the collected organic layers
were washed with NaHCO3 (10%, 2 Â 40 mL) and brine (40 mL). The
organic phase was dried over Na2SO4, and evaporation in vacuum gave a
brownish oil as residue. Distillation at reduced pressure through a Vigreux
column and chromatography (silica gel) starting with dichloromethane
(100%), then with a gradient up to dichloromethane/methanol (95:5)
yielded the amine 7 as an colourless oil. Yield: 1.43 g, 66% (>99% NMR,
>99.2% ee of (1R, 2S)-7, chiral GC); b.p. 65 ± 678C / 3 Â 10 2 mbar;
1H NMR (400 MHz, CDCl3, 258C, TMS): d ˆ 7.15 ± 7.30 (m, 5H; Ph), 3.85
(d, J ˆ 3.2 Hz, 1H; PhCHN), 2.54 ± 2.57 (m, 4H; NCH2), 2.34 (s, 3H; NMe),
2.26 ± 2.28 (m, J ˆ 3.2 Hz, 1H; CHN), 1.90 (brs, 1H; NH), 1.75 ± 1.83 (m,
obtained by solvation by imidazole 1 are larger than for
solvation by THF. Thus, state C is 18.9/4.6 kcalmol more
1
stable than B. The most stable state involving the 1-disolvated
1
homodimers 11a and 9a is 2.0/1.5 kcalmol less stable than
state K. State K is 14.9/3.2 kcalmol 1 more stable than G with
10a monosolvated by 1. With the assumption that PM3
overestimates 1 solvation and an estimated solvation entropy
1
of about 5 kcalmol , 10 is indicated to be preferably
disolvated by 1. State H with the THF-disolvated heterodimer
1
10c ´ (THF)2 is 26.7/1.0 kcalmol higher in enthalpy than K.
This enthalpy difference indicates that 1 will replace THF in
the solvation of the dimers. This prediction agrees with the
NMR observations, in which the lithium signals are shifted
upon addition of 1 to a THF solution of the heterodimer 10.
Conclusion
In summary, we found that the imidazoloid 2 is a strong
enough base to deprotonate the diamine 5 to yield 6, which
forms the mixed dimer 10 with the remaining 2. On the other
hand, interestingly, 6 is found to be able to deprotonate 1 to
yield 2 which forms the heterodimer 10 with the remaining 6.
Evidently, the thermodynamically most stable complex in
solution is the heterodimer 10, and this complex is thus the
new reagent in the enantioselective deprotonation of the
meso-epoxide 3 yielding an enhanced stereoselectivity (96%
ee).
Computational investigations by PM3 and B3LYP/6 ±
311 ‡ G(d,p) have shown possible structures for the hetero-
and homodimers, and the role of THFand 1 in the solvation of
the dimers, in line with the NMR results. Favoured complexes
in the equilibria between homo- and heterocomplexes have
also been calculated.
4H; CH2), 0.80 (d, J ˆ 6.4 Hz, 3H; Me); 13C NMR (125 MHz, [D8]THF
,
258C): d ˆ 143.3, 128.8, 128.7, 127.2 (Ph), 67.9 (CHN), 67.8 (CHN), 53.0
(NCH2), 35.8 (NHMe), 25.3 (CH2), 13.4 (Me).
NMR, general: All NMR experiments were performed in Wilmad tubes
(5 mm) fitted with a Wilmad/Omnifit Teflon valve assembly (OFV) and a
Teflon/Silicon septum. NMR spectra were recorded on a Varian Unity500
spectrometer equipped with a 5 mm 1H, 6Li, 13C triple-resonance probe
head, custom built by Nalorac. Measuring frequencies were 499.9 MHz
1
(1H), 125.7 MHz (13C) and 73.57 MHz (6Li). The H and 13C spectra were
referenced to signals from residual protons at C2, and the C2 carbon in the
solvent [D8]THF: d ˆ 1.73 and d ˆ 25.57, respectively. Lithium resonances
were referenced to external 6Li in [6Li]Cl (0.3m) in [D4]MeOH (d ˆ 0.0) in
a separate NMR tube. The typical 908 6Li pulse was 20 ms. The probe
temperature was calibrated by using a methanol thermometer.
Typical NMR experiment: Amine 7 (44 mL, 0.2 mmol) was added to
[D8]THF (700 mL) in an NMR tube. The lithium amide 8 was generated by
4376
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0720-4376 $ 17.50+.50/0
Chem. Eur. J. 2001, 7, No. 20

Novel Dimer for Enantioselective Catalysis
4368±4377
titration of
7
with nBu[6Li] (ca. 10m, 20 mL) while monitoring the
[16] M. Amadji, J. Vadecard, J.-C. Plaquevent, L. Duhamel, P. Duhamel, J.
Am. Chem. Soc. 1996, 118, 12483.
[17] T. Yamashita, D. Sato, T. Kiyoto, A. Kumar, K. Koga, Tetrahedron
1997, 53, 16987.
[18] M. Asami, T. Ishizaki, S. Inoue, Tetrahedron: Asymmetry 1994, 5, 793.
[19] M. Asami, T. Suga, K. Honda, S. Inoue, Tetrahedron Lett. 1997, 38,
6425.
disappearance of the benzylic proton of 7 at d ˆ 3.8 and the appearance
of the benzylic proton of 8 at d ˆ 4.0 at 808C. The 1-methylimidazole, 1,
(4 mL, 0.05 mmol, 0.25 equiv) was added, and the solution was allowed to
equilibrate for 15 minutes before spectra were recorded. Li spectra were
recorded with at ˆ 2 s, d1 ˆ 18 s and nt ˆ 32.
6
Typical rearrangement of 3: Amine 5 (4.4 mL, 0.02 mmol) and 1-methyl-
imidazole 1 (16 mL, 0.20 mmol) were dissolved in THF (880 mL) in a
reaction vessel in the glovebox. After the vessel was transferred from the
glovebox, nBuLi (89 mL, 2.47m in hexanes, 0.22 mmol) was added under
nitrogen. The yellow reaction solution was allowed to equilibrate at
20.00 Æ 0.058C for 10 minutes in a thermostat (Heto Birkerùd). The
reaction was started by adding cyclohexene oxide 3 (10 mL, 0.10 mmol) to
the reaction mixture. In order to follow the reaction, samples (50 mL) were
withdrawn from the reaction vessel at different intervals, and diethyl ether
(0.5 mL) was added. The solutions were quenched in saturated NH4Cl
(0.25 mL) and washed with brine (0.25 mL). The samples were analysed by
chiral gas chromatography. The reaction yield of 4 was determined by using
an internal standard relative to 4 at the end of the reaction as previously
described.[24]
[20] M. J. Södergren, P. G. Andersson, J. Am. Chem. Soc. 1998, 120, 10760.
[21] M. J. Södergren, S. K. Bertilsson, P. G. Andersson, J. Am. Chem. Soc.
2000, 122, 6610.
[22] C. Hilf, F. Bosold, K. Harms, J. C. W. Lohrenz, M. Marsch, M.
Schimeczek, G. Boche, Chem. Ber. 1997, 130, 1201.
[23] C. Hilf, F. Bosold, K. Harms, M. Marsch, G. Boche, Chem. Ber. 1997,
130, 1213.
Â
[24] D. Pettersen, M. Amedjkouh, S. O. Nilsson Lill, K. Dahlen, P.
Ahlberg, J. Chem. Soc. Perkin Trans. 2 2001, Published on the Web
27th of April.
[25] S. E. De Sousa, P. OꢁBrien, H. C. Steffens, Tetrahedron Lett. 1999, 40,
8423.
[26] S. O. Nilsson Lill, P. I. Arvidsson, P. Ahlberg, Tetrahedron: Asymmetry
1999, 10, 265.
[27] A. Abbotto, A. Streitwieser, P. v. R. Schleyer, J. Am. Chem. Soc. 1997,
119, 11255.
[28] E. Kaufmann, J. Gose, P. v. R. Schleyer, Organometallics 1989, 8, 2577.
[29] G. Hilmersson, Ö. Davidsson, Organometallics 1995, 14, 912.
[30] D. Zhao, C.-y. Chen, F. Xu, L. Tan, R. Tillyer, M. E. Pierce, J. R.
Moore, Org. Synth. 2000, 77, 12.
[31] E. Anders, R. Koch, P. Freunscht, J. Comput. Chem. 1993, 14, 1301.
[32] J. J. P. Stewart, J. Comput. Chem. 1989, 10, 209.
[33] W. J. Hehre, B. J. Deppmeier, A. J. Driessen, J. A. Johnson, P. E.
Klunzinger, L. Lou, J. Yu, J. Baker, J. E. Carpenter, R. W. Dixon, S. S.
Fielder, H. C. Johnson, S. D. Kahn, J. M. Leonard, W. J. Pietro,
Spartan v. 5.0.1, Wavefunction Inc., Irvine, CA, 1997.
[34] W. Huang, personal communication, 13. 5. 1998.
[35] G. I. Csonka, J. Comput. Chem. 1993, 14, 895.
Computational details: Geometries were optimised at the PM3 level of
theory.[31±33] In Spartan, the option HHON[34] was used to correct for
hydrogens in close contact.[35, 36] All geometries were characterised as
minima on the potential energy surface by use of the sign of the eigenvalues
of the force-constant matrix obtained from
a frequency calculation.
Selected structures were optimised by using B3LYP/6 ± 311 ‡ G(d,p)[37±40]
as implemented in Gaussian 98.[41] Reaction energies were calculated at
PM3//PM3, B3LYP/6 ± 311 ‡ G(d,p)//PM3, or B3LYP/6 ± 311 ‡ G(d,p)//
B3LYP/6 ± 311 ‡ G(d,p) levels of theory.
Acknowledgement
We are grateful to the Swedish National Research Council for support.
[36] G. I. Csonka, J. G. Angyan, THEOCHEM 1997, 393, 31.
[37] A. D. Becke, J. Chem. Phys. 1993, 98, 5648.
[38] C. Lee, W. Yang, R. G. Parr, Phys. Rev. B: Condens. Matter 1988, 37,
785.
[39] R. Krishnan, J. S. Binkley, R. Seeger, J. A. Pople, J. Chem. Phys. 1980,
72, 650.
[1] N. S. Simpkins, Pure Appl. Chem. 1996, 68, 691.
[2] D. M. Hodgson, A. R. Gibbs, G. P. Lee, Tetrahedron 1996, 52, 14361.
[3] P. OꢁBrien, J. Chem. Soc. Perkin Trans. 1 1998, 1439.
[4] T. Kasai, H. Watanabe, K. Mori, Bioorg. Med. Chem. 1993, 1, 67.
[5] M. Asami, S. Inoue, Tetrahedron 1995, 51, 11725.
[6] D. M. Hodgson, A. R. Gibbs, Synlett 1997, 657.
[7] D. M. Hodgson, J. Witherington, B. A. Moloney, J. Chem. Soc. Perkin
Trans. 1 1994, 3373.
[40] M. J. Frisch, J. A. Pople, J. S. Binkley, J. Chem. Phys. 1984, 80, 3265.
[41] Gaussian 98 (Revision A.7), M. J. Frisch, G. W. Trucks, H. B. Schlegel,
G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A.
Montgomery, R. E. Stratmann, J. C. Burant, S. Dapprich, J. M.
Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi,
V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo,
S. Clifford, J. Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui, K.
Morokuma, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B.
Foresman, J. Cioslowski, J. V. Ortiz, B. B. Stefanov, G. Liu, A.
Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R. L. Martin,
D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, C.
Gonzalez, M. Challacombe, P. M. W. Gill, B. G. Johnson, W. Chen,
M. W. Wong, J. L. Andres, M. Head-Gordon, E. S. Replogle, J. A.
Pople, Gaussian, Inc., Pittsburgh PA, 1998.
[8] M. Asami, J. Takahashi, S. Inoue, Tetrahedron: Asymmetry 1994, 5,
1649.
[9] D. Bhuniya, A. DattaGupta, V. K. Singh, J. Org. Chem. 1996, 61, 6108.
Ê
[10] G. Hilmersson, P. I. Arvidsson, Ö. Davidsson, M. Hakansson, Organo-
metallics 1997, 16, 3352.
Ê
[11] G. Hilmersson, P. I. Arvidsson, Ö. Davidsson, M. Hakansson, J. Am.
Chem. Soc. 1998, 120, 8143.
[12] P. I. Arvidsson, G. Hilmersson, P. Ahlberg, J. Am. Chem. Soc. 1999,
121, 1883.
[13] D. Sato, H. Kawasaki, I. Shimada, Y. Arata, K. Okamura, T. Date, K.
Koga, Tetrahedron 1997, 53, 7191.
[14] G. Hilmersson, Ö. Davidsson, J. Org. Chem. 1995, 60, 7660.
[15] J. P. Tierney, A. Alexakis, P. Mangeney, Tetrahedron: Asymmetry 1997,
8, 1019.
Received: April 5, 2001 [F3177]
Chem. Eur. J. 2001, 7, No. 20
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
0947-6539/01/0720-4377 $ 17.50+.50/0
4377