A new pyridine-2,6-bis(oxazoline) for efficient and flexible lanthanide-based catalysts of enantioselective reactions with 3-alkenoyl-2-oxazolidinones

Article abstract of DOI:10.1002/chem.200401213

A new pyridine-2,6-bis(oxazoline) (4) has been easily synthesised from the reaction of (1S,2S)-2-amino-1-phenylpropane-1,3-diol (1) and dimethyl pyridine-2,6-dicarboximidate (2), followed by TIPS (TIPS = triisopropylsilyl) protection of the 4′-CH₂OH group. The catalysts derived from 4 and eight lanthanide(III) triflates have been tested over three reactions involving 3-acryloyl- and 3-crotonoyloxazolidinones (5a,b): the Diels-Alder (DA) reaction with cyclopentadiene, the 1,3-dipolar cycloaddition with diphenyl nitrone and the Mukaiyama-Michael reaction with 2-trimethylsilyloxyfuran. Several reactions exhibit very good enantioselectivity (ee > 90%), and the opposite enantiomers can be easily obtained simply by changing the cation. This specific feature of the ligand can be appreciated in the DA reaction of 5a, since the catalyst [Sc^III(4)] gives the adduct (2′S)-9a with 99% ee, whereas the catalyst [Y^III(4)] gives the opposite enantiomer with 95% ee. A rationale of the enantioselectivity is proposed on the basis of the NMR spectra of La-based complexes involving 4 and 5 as ligands.

Full text of DOI:10.1002/chem.200401213

A New Pyridine-2,6-bis(oxazoline) for Efficient and Flexible Lanthanide-
Based Catalysts of Enantioselective Reactions with 3-Alkenoyl-2-
oxazolidinones
Giovanni Desimoni,* Giuseppe Faita, Matilde Guala, Anna Laurenti, and
[
a]
Mariella Mella
Abstract: A new pyridine-2,6-bis(oxa-
zoline) (4) has been easily synthesised
from the reaction of (1S,2S)-2-amino-1-
phenylpropane-1,3-diol (1) and dimeth-
yl pyridine-2,6-dicarboximidate (2), fol-
lowed by TIPS (TIPS=triisopropylsil-
action with cyclopentadiene, the 1,3-di-
polar cycloaddition with diphenyl ni-
trone and the Mukaiyama–Michael re-
action with 2-trimethylsilyloxyfuran.
Several reactions exhibit very good
enantioselectivity (ee>90%), and the
opposite enantiomers can be easily ob-
tained simply by changing the cation.
This specific feature of the ligand can
be appreciated in the DA reaction of
III
5a, since the catalyst [Sc (4)] gives the
adduct (2’S)-9a with 99% ee, whereas
III
the catalyst [Y (4)] gives the opposite
yl) protection of the 4’-CH OH group.
enantiomer with 95% ee. A rationale
of the enantioselectivity is proposed on
the basis of the NMR spectra of La-
based complexes involving 4 and 5 as
ligands.
2
The catalysts derived from 4 and eight
lanthanide(iii) triflates have been
tested over three reactions involving 3-
acryloyl- and 3-crotonoyloxazolidi-
nones (5a,b): the Diels–Alder (DA) re-
Keywords: asymmetric synthesis ·
cations · enantioselectivity · lantha-
nides · N ligands
Introduction
covered (and probably never will), but pyridine-2,6-bis(oxa-
zolines) (pybox) provide efficient catalysts for such a variety
of processes that a search in this field is certainly attractive.
[1]
The formation of carbon–carbon bonds through asymmetric
catalytic processes has attracted a great deal of interest and
the art of organic synthesis has progressed impressively. In
general, the catalysts consist of a cation coordinating an op-
tically active organic ligand and the complex must have at
least one free binding site to coordinate one reagent, pro-
moting the formation of the reactive complex involved in
the catalytic cycle.
The ideal ligand should offer some advantages: it should
be easily prepared, cheap, resistant under the reaction con-
ditions and very selective for a large number of processes,
hence very flexible. The ideal ligand has not yet been dis-
The reaction between (1S,2S)-2-amino-1-phenylpropane-
1,3-diol (1) and dimethyl pyridine-2,6-dicarboximidate (2)
has several of the advantages described above, and if the hy-
droxy group involved in the oxazoline formation is in the 1-
position, the resulting pybox will have a phenyl group in po-
sition 5’, which has been found to be very useful in the de-
velopment of efficient catalysts of the Mukaiyama–Michael
[2]
[3,4]
(MM) and Diels–Alder (DA)
reactions of alkenoyl-1,3-
oxazolidinones. The residual CH OH groups in the 4’-posi-
2
tion could interfere in the formation of the reactive complex
that involves the nitrogen atoms of the pybox, the cation
and the reagent, but a suitable protection of the hydroxy
groups should both avoid this and increase the shielding
effect exerted by the substituent at the 4’-position.
[
a] Prof. G. Desimoni, Prof. G. Faita, Dr. M. Guala, Dr. A. Laurenti,
Prof. M. Mella
Dipartimento di Chimica Organica
Universitꢂ di Pavia
[5]
Following the protocol reported by Aggarwal for the
preparation of the analogous box from dimethylmalonyldini-
Viale Taramelli 10
[6]
[7]
trile and 1, Mꢀller and Bolꢁa and Reiser and co-workers
2
7100 Pavia (Italy)
synthesised the desired 5-aryl-4-hydroxymethylpybox in a
regioselective manner and in good yields starting from the
corresponding p-nitrophenyl and p-methylthiophenyl (thio-
micamine) analogues of 1 (Scheme 1). These ligands were
Fax : (+39)382-987-323
E-mail: desimoni@unipv.it
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
3816
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/chem.200401213
Chem. Eur. J. 2005, 11, 3816 – 3824

FULL PAPER
Table 1. Diels–Alder reactions between 5a and 6 at À508C in CH
2
2
Cl in
the presence of 10% mol of catalyst and 4 ꢅ molecular sieves.
[
a]
[b]
[c]
[c]
Entry
r
[ꢅ] Triflate t [h] [9a]/[10a]
9a ee%
10a ee%
1
2
3
4
5
6
7
8
0.870
0.977
0.985
1.015
1.019
1.066
1.126
1.160
Sc
16
48
24
48
48
48
24
16
96:4
90:10
91:9
95:5
90:10
91:9
93:7
94:6
99 (2’S) Re >95 (2’S) Re
24 (2’R) Si racemic
Lu
Yb
Ho
Y
Eu
Pr
83 (2’R) Si 55 (2’R) Si
91 (2’R) Si 60 (2’R) Si
95 (2’R) Si 67 (2’R) Si
90 (2’R) Si 50 (2’R) Si
86 (2’R) Si 24 (2’R) Si
80 (2’R) Si racemic
La
[
a] r=ionic radius. [b] Yields were all quantitative. [c] The configuration
is given in parentheses, followed by the face approach.
Scheme 1.
gives an excellent enantioselectivity (ee 99%); in fact this
catalyst gives one of the best results reported in the litera-
ture.
tested in the addition of diethylzinc to benzaldehyde and cy-
[7]
clohexenone and, after silylation with several trialkylsilyl
II
chlorides, in the intramolecular Cu -catalysed cyclopropana-
This ligand produces another remarkable result, since the
enantioselectivity can be simply driven by changing the
cation: the Sc-based catalyst gives (2’S)-9a in 99% ee, while
the Y-based catalyst furnishes the opposite enantiomer with
up to 95% ee. This specific property makes pybox 4 a
unique ligand, better by far than cis-4’,5’-diphenyl-pybox,
which gave a similar effect, but with a less pronounced shift
tion reaction, but the enantioselectivity was always disap-
pointing.
Despite these unpleasant precedents, a test of the cata-
lysts based on silylated pybox derived from 1 was planned.
[4]
Results
in the enantioselectivity.
The analogous DA reaction run on the crotonoyl deriva-
tive 5b requires a higher temperature to be complete within
1–2 days and is less endo selective than the cycloaddition in-
volving 5a: if Sc is excluded, products 9b and 10b are ob-
tained in a ratio varying from 2:1 to about 1:1. The enantio-
selectivity obtained with the Sc catalyst (Table 2, entry 1) is
excellent, since the eeꢄs of (2’S)-9b and 10b are 93 and
99%, respectively. The catalysts with any other lanthanide
revert the sense of the induction, but good results in terms
of enantioselectivity are observed only for the endo isomer
and for Ho, Y and Eu (Table 2, entries 4–6).
Pybox 4 was synthesised by refluxing 1 and 2 in chloroben-
zene to give 3, which was isolated and silylated with iPr SiCl
3
and imidazole (Scheme 1) to give 4 in good yield on a multi-
gram scale. Compound 4 was the ligand used for the forma-
tion of the complexes with eight lanthanide triflates, taking
into account the specific affinity of pybox for these trivalent
[1]
cations and the interesting effects on the enantioselectivity
exerted by such cations.
In the choice of the reactions to be tested, the well-known
bicoordination ability of 3-alkenoyl-1,3-oxazolidin-2-ones 5
with pybox and lanthanide cat-
[1]
ions was considered; therefore
the DA reaction of cyclopenta-
diene (6), the 1,3-dipolar cyclo-
addition (1,3-DC) with diphe-
nylnitrone (7) and the MM re-
action with 2-trimethylsilyloxy-
furan (8), with either 3-acryloyl-
or
3-crotonoyloxazolidinones
(
5a,b), were tested (Scheme 2).
The Diels–Alder (DA) reac-
tion: The DA reaction of 5a
and 6 is strongly endo selective
and the endo-9a/exo-10a ratio
was always more than 90:10
(
Table 1). The enantioselectivity
is strongly influenced by the
nature of the cation: Ho, Y and
Eu give very good enantioselec-
tivities (entries 4–6: ee in the
range 90–95%) and Sc (entry 1)
Scheme 2.
Chem. Eur. J. 2005, 11, 3816 – 3824
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3817

G. Desimoni et al.
Table 2. Diels–Alder reactions between 5b and 6 at 178C in CH
the presence of 10% mol of catalyst and 4 ꢅ molecular sieves.
2
Cl
2
in
purity suitable to determine its absolute configuration
through the X-ray structure of a derivative only by using
trans-4,5-diphenyl bis(oxazoline) as the chiral ligand.
[
a]
[b]
[b]
[9]
Entry Triflate t [h] [9b]/[10b]
9b ee%
10b ee%
1
2
3
4
5
6
7
8
Sc
24
48
48
48
48
48
48
16
92:8
93 (2’S,3’R) Re >99 (2’S,3’R) Re
30 (2’R,3’S) Si racemic
If the exo selectivity is remarkable, the ee of 12a obtained
with the lanthanide/4 complexes (Table 3) is unsatisfactory
and cannot compete with the ee obtained with the above-
mentioned box-based catalysts.
Lu
Yb
Ho
Y
Eu
Pr
70:30
63:37
45:55
55:45
68:32
71:29
74:26
64 (2’R,3’S) Si 35 (2’R,3’S) Si
90 (2’R,3’S) Si 56 (2’R,3’S) Si
90 (2’R,3’S) Si 54 (2’R,3’S) Si
90 (2’R,3’S) Si 49 (2’R,3’S) Si
87 (2’R,3’S) Si 30 (2’R,3’S) Si
81 (2’R,3’S) Si 18 (2’R,3’S) Si
Table 3. 1,3-Dipolar cycloaddition of 5a and 7 at À208C in CH
2 2
Cl in the
La
presence of 10% mol of catalyst and 4 ꢅ molecular sieves.
[
a] Yields were all quantitative. [b] The configuration is given in paren-
[
a]
[b]
[b]
Entry Triflate t [h] [11a]/[12a]
11a ee%
12a ee%
theses, followed by the face approach.
1
2
3
4
5
6
7
8
Sc
48
48
48
48
48
48
48
16
22:78
10:90
9:91
8:92
10:90
6:94
50 (3’S,4’R) Re 20 (3’R,4’R) Re
24 (3’S,4’R) Re 67 (3’R,4’R) Re
Lu
Yb
Ho
Y
Eu
Pr
5 (3’R,4’S) -
62 (3’R,4’R) Re
The behaviour of both dienophiles in terms of enantiose-
lectivity response to the change of the cation is similar. Sc
gives (S)-9 as the preferred enantiomer, then, with the in-
creasing of the ionic radius, the enantioselectivity is reverted
with a maximum occurring with the medium-sized cations
53 (3’R,4’S) Si racemic
53 (3’R,4’S) Si 7 (3’S,4’S) -
68 (3’R,4’S) Si 51 (3’S,4’S) Si
56 (3’R,4’S) Si 66 (3’S,4’S) Si
26 (3’R,4’S) Si 65 (3’S,4’S) Si
6:94
10:90
La
[
a] Yields were all quantitative. [b] The configuration is given in paren-
(
Ho, Y, Eu). If log[(S)-9/(R)-9] or log[(S)-10/(R)-10] are
theses, followed by the face approach.
plotted versus the lanthanide ionic radius (r), a two lines are
obtained that intercept each other at Y, as clearly evidenced
by the graph of Figure 1 in the case of 9a. This is the same
behaviour already observed in the asymmetric catalysis with
Again, the enantioselectivity largely depends upon the
cation: Lu and Yb (entries 2,3) give 12a with the (3’R,4’R)
configuration (67 and 62% ee respectively), whereas Pr and
La (entries 7,8) give the opposite enantiomer with compara-
ble ee.
[4,8]
other 4’-alkyl-substituted pybox ligands.
The reaction between the crotonoyloxazolidinone (5b)
and 7 was run at room temperature; the yields are about
quantitative and nearly equal amounts of endo-11b and exo-
1
2b are obtained (Table 4). The ee of endo-11b is very dis-
appointing with small cations, and only Pr and La (the
larger cations; entries 7,8) give the (3’R,4’S,5’R) enantiomer
with up to 64% ee. In the case of exo-12b, a preferential ap-
proach of nitrone to the Re-face of the coordinated dipolar-
ophile is observed with the smaller cations, and
(
3’R,4’R,5’S)-12b is the favoured stereoisomer (Sc, Lu and
Yb; Table 4, entries 1–3), while the favoured approach with
the larger cations is on the opposite Si-face (Table 4, en-
tries 4–8), and
3’S,4’S,5’R)-12b is obtained by using Pr (Table 4, entry 7).
The relationships between enantioselectivity and ionic
a very good selectivity (90% ee) of
(
radius are somewhat different from those observed in the
case of the DA reaction, with trends that depend upon the
dipolarophile.
Figure 1. Plot of log[(S)-9a/(R)-9a] versus the lanthanide ionic radius (r).
1,3-Dipolar cycloaddition (1,3-
DC): The 1,3-DC between 5a
2 2
Table 4. 1,3-Dipolar cycloaddition of 5b and 7 at 258C in CH Cl in the presence of 10% mol of catalyst and
and diphenylnitrone 7 was run 4 ꢅ molecular sieves.
at À208C, the yields are nearly
[a]
[b]
[b]
Entry
Triflate
t [h]
[11b]/[12b]
11b ee%
12b ee%
quantitative, all catalysts are
exo-selective and 12a is the
major product, an unusual be-
haviour in the pybox-catalysed
1
2
3
4
5
6
7
8
Sc
72
72
72
72
72
72
72
72
37:63
45:55
49:51
59:41
56:44
50:50
50:50
56:44
racemic
54 (3’R,4’R,5’S) Re
50 (3’R,4’R,5’S) Re
34 (3’R,4’R,5’S) Re
48 (3’S,4’S,5’R) Si
47 (3’S,4’S,5’R) Si
81 (3’S,4’S,5’R) Si
90 (3’S,4’S,5’R) Si
85 (3’S,4’S,5’R) Si
Lu
Yb
Ho
Y
Eu
Pr
8 (3’R,4’S,5’R) -
11 (3’R,4’S,5’R) Si
30 (3’R,4’S,5’R) Si
30 (3’R,4’S,5’R) Si
45 (3’R,4’S,5’R) Si
64 (3’R,4’S,5’R) Si
64 (3’R,4’S,5’R) Si
[
[
c]
d]
[1]
1,3-DC between 7 and 5a.
[e]
For box-based catalysts exo-se-
lectivity is much more usual;
for example, compound 12a
La
[a] Yields were quantitative unless stated otherwise. [b] The configuration is given in parentheses, followed by
was obtained with optical the face approach. [c] Yield: 94%. [d] Yield: 98%. [e] Yield: 93%.
3818
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 3816 – 3824

Lanthanide-Based Catalysts
FULL PAPER
The Mukaiyama–Michael (MM) reaction: The MM reaction
of 5a and 8 to give the enantiomers 13a and 14a is not easy,
and only a few asymmetric catalysts have been applied to
such a reaction. Two catalysts were first tested: the (R)-3,3’-
The Sc complex (entry 2) gives excellent results both in
terms of diastereo- (anti:syn=98:2) and enantioselectivity
(the ee of (S,S)-13b is 94%) and it can compete with the
best asymmetric catalysts, specific for the MM reaction of
5a and 8, the bis[(4’R,5’R)-diphenyl]pybox complexes of La,
bis(N,N-diethylaminomethyl)BINOL/Sc(OTf) , and [(S,S)-
3
[2]
tert-butyl-box/Cu(OTf) systems, both giving only moderate
Eu and Ce triflate.
2
[10]
enantioselectivity. Recently, after a careful optimisation of
the reaction conditions and by using 4 ꢅ molecular sieves
The enantioselectivity changes with the cation: the ee of
(S,S)-13b decreases from Sc to Yb (entries 2,5,6); with Ho
and the following lanthanides (entries 7–11), adduct (R,R)-
13b is obtained as the preferential enantiomer, with a maxi-
mum corresponding to the medium-sized lanthanide (Eu,
entry 9). Therefore, the relationship between enantioselec-
tivity and ionic radius becomes again a net broken line as
previously observed in the DA reaction.
(
MS) and pentafluorophenol as additives, the (R)-N,N’-
bis(2-quinolylmethylene)BINIM/Ni(ClO ) system has been
4
2
reported to give the S enantiomer in excellent yield (93%)
[11]
and enantioselectivity (with up to 91% ee).
Of the eight lanthanide complexes of 4 tested, seven de-
stroyed the reagents, and only the Sc-based complex was an
efficient catalyst when one equivalent of alcohol with poor
coordinating ability (trifluoroethanol (TFE), hexafluoroiso-
propanol (HFIP) or tert-butyl alcohol) and MS were used as
additives (Table 5). In the presence of TFE, HFIP or
tBuOH (Table 5, entries 5–7) the yields are quantitative and
the enantioselectivity of 13a is comparable with the best re-
sults reported in the literature.
Discussion
Some points deriving from the above-presented results re-
quire discussion and rationalisation:
1
2
) The enantioselectivity reversion with the change of the
cation taking the chiral ligand as constant.
) The different selectivity induced by the catalysts with
the change of the second reaction partner taking the co-
ordinating reagent as constant.
Table 5. Mukaiyama–Michael reaction of 5a and 7 at À508C in CH
2 2
Cl
in the presence of 10% mol of catalyst, 4 ꢅ molecular sieves (MS), and
[
a]
several additives.
Entry
[
b]
Triflate
Additives
ee%
1
2
3
4
5
6
7
Sc
Sc
Sc
Sc
Sc
Sc
Sc
–
MS
78 (S)
89 (S)
89 (S)
92 (S)
92 (S)
93 (S)
The formation of opposite enantiomers by simply chang-
[
c]
CF
MS/EtOH
MS/CF CH
MS/(CF CHOH
MS/tBuOH
3 2
CH OH
[
c]
ing the lanthanide cation in the pybox-based catalyst has al-
[4,8,12–14]
[c]
ready been observed
and in the DA reaction the rela-
3
2
OH
[c]
3
)
2
tionships between enantioselectivity and lanthanide ionic
radius evidenced two different typologies. When the pybox
had a phenyl group at the 4’-position on the oxazoline ring,
a linear relationship was obtained, but when the 4’-substitu-
ent was an alkyl group, two lines were obtained that inter-
cepted each other at one of the medium-sized cations, with
[c]
[
a] Yields were quantitative except for entry 1 for which no reaction oc-
curred. [b] The configuration is given in parentheses. [c] One equivalent.
The conditions required to catalyse the reaction of 5b and
were less selective and, in the presence of MS and one
[4,8]
8
the extremes being Sc and La.
This effect was interpreted
equivalent of TFE, the results reported in Table 6 (en-
tries 2,5–11) were obtained.
as the result of a change in the coordination of chiral ligand
and reagents (4 and 5 in this research) around the different
cations.
The relationship between enantioselectivity and lantha-
nide ionic radius described above (see Figure 1) suggests
that, in spite of its complex structure, ligand 4 largely be-
haves as a 4’-alkyl-substituted pybox, and that (keeping the
cation constant) 5a,b could generate the same reacting inter-
mediate in each catalytic process (DA, 1,3-DC and MM re-
action) with the attack of the different reagents (6–8) on the
same unshielded enantioface of the coordinated reagent.
This hypothesis was tested first with the predominant
endo products of the DA reaction by plotting log[(S)-9b/
(R)-9b] versus log[(S)-9a/(R)-9a] (Figure 2). When the
cation is kept constant, the linear relationship strongly sup-
ports a similar behaviour of the reacting intermediates in-
volving 5a and 5b versus 6, independent of the coordination
number (CN) of the complex.
Table 6. Mukaiyama–Michael reaction of 5b and 7 at À208C in CH
2 2
Cl
in the presence of 10% mol of catalyst, 4 ꢅ molecular sieves, and several
additives.
[
a]
[b]
[b]
Entry Triflate Additives [13b]/[14b]
13b ee%
14b ee%
1
2
3
4
5
6
7
8
9
1
1
Sc
Sc
Sc
Sc
Lu
Yb
Ho
Y
Eu
Pr
–
96:4
98:2
98:2
96:4
96:4
95:5
90:10
93:7
87:13
75:25
61:39
82 (S,S) Re
44 (R,S) Re
[
[
[
[
[
[
[
[
[
[
c]
94 (S,S) Re >90 (R,S) Re
94 (S,S) Re >90 (R,S) Re
94 (S,S) Re >95 (R,S) Re
54 (S,S) Re
46 (S,S) Re
42 (R,R) Si
26 (R,R) Si
70 (R,R) Si
53 (R,R) Si
33 (R,R) Si
d]
e]
c]
c]
c]
c]
c]
c]
c]
19 (S,R) Si
8 (S,R) –
3 (R,S) –
18 (R,S) Re
33 (R,S) Re
4 (R,S) –
0
1
La
20 (S,R) Si
[
a] Yields were all quantitative. [b] The configuration is given in paren-
theses, followed by the face approach. [c] With one equivalent of
CF CH OH. [d] With one equivalent of tBuOH. [e] With one equivalent
of (CF CHOH.
To compare the reacting intermediates involved in the
1,3-DC and MM reactions with those of the DA cycloaddi-
3
2
3 2
)
Chem. Eur. J. 2005, 11, 3816 – 3824
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3819

G. Desimoni et al.
gain information about the reactive intermediates involved
in the catalytic cycles of the above DA, 1,3-DC and MM re-
1
13
actions, possible tools are H and C NMR spectroscopic in-
vestigations and X-ray structure analysis of lanthanide
pybox complexes.
Starting from the X-ray structure of a pybox-based com-
plex, the substitution of some ligands (generally one or
more triflate ions, one or more water molecules) with the
coordinating reagent allows us to propose a model of a reac-
tive intermediate that is able to rationalise the stereochemi-
cal outcome of the reaction. There have been reports in the
literature of two crystal structures of Sc–pybox complexes;
[
(
Sc(Ph-pybox)(OTf) (H O)] and [Sc(Inda-pybox)(OTf)3-
3 2
[15,16]
H O)].
In both cases, Sc has a CN of seven: the three
2
nitrogen atoms of the pybox, one triflate ion, and one water
molecule define the equatorial plane of the complex, while
the two remaining triflate anions occupy the two residual
apical positions. In the case of La, the reported crystal struc-
ture of [La(trans-4’,5’-diPh-pybox)(OTf) (H O) ] shows that
Figure 2. Plot of log[(S)-9b/(R)-9b] versus log[(S)-9a/(R)-9a].
3
2
4
tion, the enantioselectivity of the major products obtained
from the reactions of 5b (12b for the 1,3-DC and 13b for
the MM addition) were plotted versus the enantioselectivity
of the endo product 9b. Only the correlation between DA
and MM data is linear (Figure 3), and this is a good experi-
mental proof that similar reacting intermediates are in-
volved in both the reactions. After the coordination of 4 and
La has a CN of nine, because besides the three nitrogen
atoms of the chiral ligand, two axial triflates and four water
molecules are bound to La.
[2]
To rationalise the relationship between cation and enan-
tioselectivity, a homogenous set of crystal structures involv-
ing pybox and different lanthanides is required. Unfortu-
nately, such a homogenous set of data is not available in the
literature, but recently the molecular structures of six
lanthanide(iii) complexes have been determined. The li-
gands around the cation were the bidentate and negatively
charged PhC(NAr) , the CH SiMe group, and a variable
5
b around the cation, the attack of either 6 (DA reaction)
or 8 (MM addition) is on the same enantioface of the coor-
dinated 5b, to produce the same stereochemical outcome.
2
2
17]
3
[
number of THF molecules; the crucial point is that the
CN clearly increases with the increasing of ionic radius: the
CN is five for the Sc-based complex, six for Lu, Y and Gd,
and seven for Nb and La.
III
The Sc -mediated processes are discussed first, since the
model is simple and because Sc gives the best enantioselec-
tive catalysts. The Sc complex of (4’R,5’R)-2,6-bis(4’-methyl-
5
’-phenyl-1’,3’-oxazolin-2’-yl)pyridine (15) was found to be a
highly efficient catalyst of the DA reaction between 5a and
[4]
6
, since (2’S)-9a was obtained with up to 97% ee; the Sc
complex of 4, even if its configuration is opposite to that of
1
5, gives the same (2’S)-9a with 99% ee (Table 1, entry 1).
Figure 3. Plot of log[(S)-13b/(R)-13b] of the anti-adduct of the Mukaiya-
ma–Michael versus log[(S)-9b/(R)-9b] of the endo-adduct of the Diels–
Alder.
A linear relationship between enantioselectivity and ionic
radius can be interpreted as the result of two competing
pathways, each involving a reactive complex, the specific
configuration of which induces the preferred attack to the
The models of the reacting complexes can be derived
III
from the Sc –pybox molecular structures, if two vicinal tri-
flates and the bound water are removed and substituted
with 5, through a complexation with the oxazolidinone CO
[4,8]
Re or the Si face of the complexed reagent.
A broken
[4,8,18]
linear relationship requires a more complicated model. To
group occupying the apical position of the complex.
3820
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 3816 – 3824

Lanthanide-Based Catalysts
FULL PAPER
When a methyl group is at the 4’-position (pybox 15), the Si
face of the coordinated dienophile in the reacting intermedi-
ate 18 is shielded by the trans-phenyl group in the 5’-posi-
tion; the increased steric demand of the substituent in the
4
’-position of pybox 4 becomes the crucial factor in deter-
mining the face selectivity. The favoured attack of either cy-
clopentadiene 6 or trimethylsilyloxyfuran (8) on the reacting
intermediate 19 will involve the Re face to give (2’S)-9a,b
and (2’S)-10a,b from the attack of 6 to the coordinated dien-
ophile (DA reaction), and (S,S)-13b from the attack of 8 in
the MM reaction, Figure 4.
The reacting complexes of the diamagnetic trivalent cat-
ions Sc, Y, Lu and La, were tentatively characterised by
NMR spectroscopy in CDCl , the common ligand being the
3
pybox 4, and 5a, 5b or 3-acetyl-1,3-oxazolidin-2-one (5c)
(
the last taken as an unreactive model of the dipolarophile
Figure 4. Assumed reactive intermediates for the reactions of 5 with both
used to test diphenylnitrone as potential auxiliary ligand)
being alternatively involved in the formation of the com-
plex.
The H NMR spectra with Sc show the selective forma-
tion of complexes involving pybox and the cation, a behav-
iour with some precedents in pybox–Sc-catalysed reac-
6
(Diels–Alder) and 8 (Mukaiyama–Michael), catalysed by the Sc(OTf)
3
complexes of pybox 15 (reactive intermediate 18) and 4 (reactive inter-
mediate 19).
1
III
[8,19]
tions,
with 5a–c not appreciably involved in the com-
The complexes of 5a, 5b and 5c have very similar NMR
spectra, with the oxazolidinone behaving as a bidentate
ligand through its carbonyl groups, 4 is always a tridentate
ligand, and each complex shows absorption of the triflate
carbon as a quartet at 119.4 ppm. When one equivalent of 7
is added to complex 20c, the nitrone is not involved in any
plexation. When nitrone is added to the sample containing
5
c, 4 and Sc, it is clearly coordinated to the cation.
1
The H NMR spectra of [Y(4)] complexes are poorly re-
solved, but show the unequivocal coordination of 4, 5c, and
1
13
(
when added) 7 to the cation. The H and C NMR spectra
of the [1:1:1] complex [Lu(4)-
5c)] are sufficiently resolved to
(
1
13
III
3
demonstrate the formation of a Table 7. H- and C NMR spectra of the La triflate complexes between 4 and 5a, 5b, or 5c in CHCl .
complex involving both ligands
and at least one triflate around
2
the cation (q at 119.4 ppm, J-
(
C,F)=316 Hz). When one
equivalent of 7 is added, the
broadening and shifting of the
methyne proton from 7.95 to
4
5a
dH, dC
5b
dH, dC
5c
dH, dC
20c
dH, dC
20a
dH, dC
20b
dH, dC
8.02 ppm, and the absence of
dH, dC
the carbon absorptions formerly
at 148.7 and 135.8 ppm, support H4
the formation of a new complex H4’
H3
8.24
7.93
4.40
5.76
–
–
–
–
–
–
–
–
–
–
–
–
–
8.23
8.17
4.77
6.00
4.50
4.18
2.12
–
8.25
8.19
4.74
5.96
4.55
4.21
6.95
6.42
5.85
8.24
8.18
4.74
5.96
4.55
4.21
6.62
7.06
1.83
H5’
–
–
–
that indeed retains the triflate
Ha
and in addition contains ni-
Hb
4.44
4.10
7.50
6.55
5.91
–
–
–
–
62.0
42.5
153.2
164.9
126.9
131.6
–
4.28
4.04
7.25
7.10
1.95
–
–
–
–
61.9
42.5
153.4
165.0
121.3
146.6
–
4.36
4.04
2.56
–
–
–
–
–
–
61.6
41.9
153.4
169.9
22.8
–
trone.
[a]
[a]
He
[
[
a]
a]
When one equivalent of La- Hf
Hg
[a]
–
–
(
OTf) is added to an equimolar
3
C2
C2’
C4’
147.0
162.8
76.9
84.9
–
–
–
–
–
144.3
166.6
74.0
88.4
62.5
42.9
–
–
23.5
–
144.3
144.3
mixture of 4 and either 5a, 5b
or 5c in CH Cl , by diluting the
166.5
74.1
88.1
63.0
43.4
–
–
–
–
166.6
74.1
88.2
63.0
43.6
–
–
–
–
2
2
solutions with pentane, three C5’
solid complexes (20a, 20b and Ca
Cb
2
0c respectively) can be isolat-
1
13
Cc
Cd
Ce
ed and their H and C NMR
spectra are reported in Table 7
with those of the uncomplexed Cf
reagents.
–
–
SO
3
CF
3
–
119.4
119.4
119.4
[a] Data at 508C.
Chem. Eur. J. 2005, 11, 3816 – 3824
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3821

G. Desimoni et al.
kind of coordination (methyne H is a narrow singlet at
7
1
.95 ppm, sharp carbon absorptions are observed at 148.7,
35.8 and 121.7 ppm). Further information on the structures
of 20a–c was derived from their HPLC-MS, which evi-
denced the presence of two triflate ions in each complex. In
conclusion, the spectroscopic investigations demonstrate
III
that La coordinates three nitrogen and two oxygen atoms,
belonging to 4 and 5 respectively, and two triflate ions
around the cation, while the addition of nitrone had no ap-
parent effects.
To test the relevance of the isolated La-based complexes
in the catalytic cycle, cyclopentadiene was added to samples
of 20a (cooled to À508C) and 20b (at 178C), with molecular
sieves, in CH Cl . After 16 h quantitative yields of adducts
2
2
were obtained and the stereochemical results (see Experi-
mental Section) were nearly superimposable to those report-
ed in Tables 1 and 2 (entries 8). Thus it seems reasonable to
assume that 20a,b do not differ from the reacting intermedi-
ate involved in the La-catalysed DA and MM reactions.
The La complexes give two further pieces of information.
There is a strong analogy between the NMR spectrum of
Figure 5. Assumed reactive intermediates for the reactions of 5 with both
6
(Diels–Alder) and 8 (Mukaiyama–Michael), catalysed by the La(OTf)
3
complexes of pybox 16 (reactive intermediate 21) and 4 (reactive inter-
mediate 22).
2
0b and that of a complex with the same composition, but
coordinated reagent and only a reduction in the enantiose-
lectivity will be observed.
[2]
with trans-4’,5’-diPh-pybox (16) as chiral ligand, even if the
latter has been registered in CD CN. The absorptions be-
longing to coordinated 5b are very similar; the crotonoyl
The nonhomogenous behaviour of the set of catalysts
with respect to 1,3-DC may also be rationalised from the
above NMR spectra. The enantioselectivity induced into the
DA and MM reactions by changing the lanthanides is simi-
lar because the second reagent is not involved in any reac-
tive complexes, while the coordination ability of nitrone 7
has been found to depend upon the cation. Hence the rela-
tionship between enantioselectivity and lanthanide catalyst
of the 1,3-DC reactions will not be comparable with that ob-
served in the DA and MM reactions.
3
1
protons in the H NMR spectra are well resolved at 50–
7
08C and the d values of the complex including 16 are
Conclusion
nearly identical to those reported in Table 7. The signals Ca
and Cb carbon atoms have nearly identical d values in their
C NMR spectra; other signals of 5b disappear upon com-
plexation. The d values of the triflate C atom signals are
identical.
The new pybox ligand 4 described in this paper can be
easily synthesised on a multigram scale, and, with lanthanide
triflates, gives efficient enantioselective catalysis of the DA
and the MM reactions of 5a,b. The catalyst is less enantiose-
lective in the 1,3-DC between diphenylnitrone 7 and 5, even
if some isoxazolidines are obtained with up to 90% ee.
These results make pybox 4 competitive with 2,6-bis-
[(4’R,5’R)-diphenyl-1,3-oxazolin-2’-yl]pyridine (16), one of
the more flexible pybox ligands suitable for the preparation
of catalysts with lanthanide cations. The new ligand has an
important advantage over 16, since the enantioselectivity is
a function of the lanthanide with formation of opposite
enantiomers simply by changing the cation. This specific
character can be appreciated in the DA reaction of 5a, since
13
Even if the configuration of pybox 16 is opposite to that
of 4, the sense of induction of the DA and the MM reactions
for both catalysts derive from the attack to the same Si face
[2,3]
of 5b.
Therefore, in order to rationalise the stereochemi-
cal outcome, it is necessary to assume that, in going from
pybox 16 to 4, the increased steric hindrance of the substitu-
ent in the 4’-position determines a deformation of the cata-
lyst geometry without any change in the cation coordination
sphere (see structures 21 and 22, Figure 5). In the reactive
intermediate 21, derived from pybox 16, the Re face of the
coordinated alkenoyloxazolidinone is shielded by the phenyl
group at the 5’-position of the ligand, while in the deformed
complex 22 the shielding effect is exerted by the bulky sub-
stituent in the 4’-position of pybox 4. In both cases, despite
the opposite configuration of the chiral ligands, the reactive
complex will favour an approach to the same Si face of the
III
the complex [Sc (4)] gives the adduct (2’S)-9a with 99% ee,
III
whereas the complex [Y (4)] gives the opposite enantiomer
with 95% ee.
3822
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 3816 – 3824

Lanthanide-Based Catalysts
FULL PAPER
(
1’S,2’S,4’S)- and (1’R,2’R,4’R)-9a respectively, largely dependant on
Experimental Section
small variations of the solvents, and were checked with reference mix-
tures.
General methods and materials: Melting points were determined by the
1
13
General procedure for the enantioselective Diels–Alder reaction between
capillary method and are uncorrected. H and C NMR spectra were re-
corded at 300 and 75 MHz, respectively. HPLC-MS spectra were carried
out by using an LCQ DECA ion trap mass spectrometer equipped with
electrospray ionisation (ESI) ion source and controlled by Xcalibur soft-
ware 1.1 (Thermo-Finnigan, San Jose, CA, USA). For sample injection
5
(
(
b and 6: Compound 5b (0.046 g, 0.30 mmol), the chiral ligand pybox 4
0.03 mmol), the lanthanide triflate (0.03 mmol) and molecular sieves
0.040 g) were added to anhydrous CH Cl (0.3 mL) at ambient tempera-
2
2
ture in a rubber-septum-sealed vial; the mixture was stirred 1 h and then
kept at a constant temperature of 178C. Compound 6 (100 mL, ca.
1
scribed for the reaction between 5a and 6, except the HPLC analysis was
performed on a Chiralpak AD column with hexane/2-propanol (98:2) as
eluant (0.5 mLmin ). The average retention times were 32 and 36 min
for (1’S,2’R,3’S,4’R)- and (1’R,2’S,3’R,4’S)-3-(3-methylbicyclo[2.2.1]hept-
À1
the instrument syringe pump was used at a flow rate of 5 mLmin . Sam-
ples were dissolved in CH CN. Experiments were carried out in positive
3
.5 mmol) was added and the reaction was worked up as previously de-
ion mode under constant instrumental conditions: source voltage 2.0 keV,
capillary voltage 37 V, sheet gas flow 29 (arbitrary units), capillary tem-
perature 2008C, tube lens voltage 0 V. Dichloromethane was the hydro-
carbon-stabilised Aldrich ACS grade, distilled from calcium hydride and
used immediately; lanthanide triflates were Aldrich ACS reagents; pow-
dered molecular sieves 4 ꢅ were obtained from Aldrich and heated
under vacuum at 3008C for 5 h and kept in sealed vials in a dryer.
À1
5
’-ene-2’-carbonyl)-2-oxazolidinone (10b) respectively, 34.5 and 40.5 min
for (1’S,2’S,3’R,4’R)- and (1’R,2’R,3’S,4’S)-(9b) respectively.
General procedure for the enantioselective 1,3-dipolar cycloaddition be-
(
1S,2S)-2-Amino-1-phenylpropane-1,3-diol (1) and 3-acetyl-1,3-oxazoli-
tween 5a and 7: Compound 5a (0.042 g, 0.30 mmol), pybox
(0.03 mmol), the lanthanide triflate (0.03 mmol) and molecular sieves
(about 0.040 g) were added to anhydrous CH Cl (0.3 mL) at ambient
4
din-2-one (5c) were Aldrich commercial products. Dimethyl pyridine-2,6-
dicarboximidate (2) was prepared from 2,6-dicyanopyridine following the
literature method. 3-Acryloyl- and 3-crotonoyl-1,3-oxazolidin-2-ones
2
2
[
6]
temperature in a rubber-septum-sealed vial, and the mixture was stirred
[
20,21]
(
5a,b) were prepared following the literature method.
for about 1 h. The mixture was then cooled at À208C and after about
1
0 min 7 (0.060 g, 0.30 mmol) was added and stirring was continued at
(
(
4’S,5’S)-2,6-Bis(4’-hydroxymethyl-5’-phenyl-1’,3’-oxazolin-2’-yl)pyridine
3): A mixture of 1 (3.34 g, 20 mmol) and 2 (1.93 g, 10 mmol) in 1,2-di-
À208C until TLC showed that all dipolarophile had reacted. The reaction
was quenched in water, extracted with CH Cl , dried, the mixture of ad-
2
2
chloroethane (20 mL) was heated under reflux for 24 h. After cooling for
a few hours at 58C, 3 precipitated as a white solid which was filtered,
dried (2.67 g, 67% yield), and directly used in the following reaction. A
sample was crystallised from ethyl acetate as soft white crystals. M.p.
ducts 11a and 12a was separated from pybox ligand by column chroma-
tography (silica gel, cyclohexane/ethyl acetate 80:20 was the eluant), and
subjected to HPLC analysis using a Chiralpack AD column with hexane/
À1
2
D
5
1
2-propanol (8:2) as eluant (1.0 mLmin ). The quality of 2-propanol was
2
10–2118C; [a] =+20.5 (c=0.2 in CHCl
3
); H NMR (300 MHz, CDCl
3
,
crucial for the separation and C. Erba solvent was the best. The average
retention times were 17 and 19 min for (3’S,4’S)- and (3’R,4’R)-3-[(2’,3’-
diphenylisoxazolidin-4’yl)carbonyl]-1,3-oxazolidin-2-one (12a) respective-
ly; 20.5 and 24.4 min for (3’R,4’S)- and (3’S,4’R)-11a respectively.
2
58C, TMS): d=3.84 (brs, 2H; -CHHOH), 4.31 (m, 4H; 4’-H,
3
-
CHHOH), 5.23 (brs, 2H; -OH), 5.79 (d, J(H,H)=8.3 Hz, 2H; 5’-H),
3
7
7
.3–7.5 (m, 10H; aromatic protons), 7.69 (t, J(H,H)=7.8 Hz, 1H; 4-H),
.92 ppm (d, J(H,H)=7.8 Hz, 2H; 3-H); C NMR (75 MHz, CDCl
3
13
3
):
d=62.7 (-CH
2
OH), 76.7 (4’-C), 83.3 (5’-C), 125.4 (3-C), 125.9, 128.4,
General procedure for the enantioselective 1,3-dipolar cycloaddition be-
tween 5b and 7: Compound 5b (0.046 g, 0.30 mmol), the chiral ligand
pybox 4 (0.03 mmol), the lanthanide triflate (0.03 mmol) and molecular
1
1
28.8, 137.0 (4-C), 146.3, 163.0 ppm; IR (Nujol): n˜ =3308 (OH),
653 cm ; elemental analysis calcd (%) for C25
À1
23 3 4
H N O (429.5): C 69.91,
H 5.40, N 9.79; found: C 70.05, H 5.33, N 10.00.
4’S,5’S)-2,6-Bis[4’-(triisopropylsilyl)oxymethyl-5’-phenyl-1’,3’-oxazolin-2’-
yl]pyridine (4): A suspension of 3 (2.14 g, 0.5 mmol), triisopropylsilyl
chloride (2.12 g, 1.1 mmol) and imidazole (2.04 g, 3 mmol) in CH Cl was
stirred overnight at RT. The solvent was evaporated and the residue was
subjected to chromatography over a short column of neutral Al with
cyclohexane/AcOEt 85:15 as eluant. Pure 4 (3.15 g, 85% yield) soon sep-
2 2
sieves (about 0.040 g) were added to anhydrous CH Cl (0.3 mL) at ambi-
ent temperature in a rubber-septum-sealed vial; the mixture was stirred
for 1 h and then kept at a constant temperature of 258C. After about
(
1
0 min 7 (0.060 g, 0.30 mmol) was added and stirring was continued until
2
2
TLC showed that all of the dipolarophile had reacted. The reaction was
worked up as previously described for the reaction between 5a and 7 and
HPLC analysis was performed on a Chiralpak AD column with hexane/
2
O
3
À1
2
5
2-propanol (90:10) as eluant (1.0 mLmin ). The average retention times
arated as a white solid. M.p. 111–1128C; [a] =+62.2 (c=1.0 in CHCl
3
);
D
1
were 17 and 18 min for (3’R,4’R,5’S)- and (3’S,4’S,5’R)-3-[(4’-methyl-2’,3’-
diphenylisoxazolidin-4’-yl)carbonyl]-1,3-oxazolidin-2-one (12b) respec-
tively; 31 and 43 min for (3’R,4’S,5’R)- and (3’S,4’R,5’S)-11b respectively.
3
H NMR (300 MHz, CDCl , 258C, TMS): d=1.07 (m, 42H; TIPS), 3.84
3
3
(
dd, J(H,H)=9.0, 9.8 Hz, 2H; -CHHO), 4.21 (dd, J(H,H)=4.1, 9.8 Hz,
3
2
7
8
H; -CHHO), 4.40 (m, 2H; 4’-H), 5.76 (d, J(H,H)=6.4 Hz, 2H; 5-H),
.3–7.4 (m, 10H; aromatic protons), 7.93 (t, J(H,H)=7.8 Hz, 1H; 4-H),
.24 ppm (d, J(H,H)=7.8 Hz, 2H; 3-H); C NMR (75 MHz, CDCl
3
General procedure for the enantioselective Mukaiyama–Michael reaction
3
13
):
between 5a and 8: Compound 5a (0.042 g, 0.30 mmol), pybox
(0.03 mmol), Sc(OTf) (0.015 g, 0.03 mmol) and molecular sieves (about
0.040 g) were added to anhydrous CH Cl (0.3 mL) at ambient tempera-
4
3
d=11.8 (CH TIPS), 17.9 (CH
3
TIPS), 65.5 (CH O), 76.9 (4’-C), 84.9 (5’-
2
3
C), 125.8, 127.9 (3-C), 128.4, 137.1, 140.8 (4-C), 147.0, 172.6 ppm; IR
2
2
À1
(
(
Nujol): n˜ =1642 cm ; elemental analysis calcd (%) for C43
H
63
N
3
O
4
Si
2
ture in a rubber-septum-sealed vial; the mixture was stirred for about
742.2): C 69.59, H 8.56, N 5.66; found: C 69.75, H 8.48, N 5.78.
1 h. The mixture was then cooled to À208C, the required amount
(
0.30 mmol) of the additive reported in Table 5 was added, and after
General procedure for the enantioselective Diels–Alder reaction between
a and 6: Compound 5a (0.042 g, 0.30 mmol), pybox 4 (0.03 mmol), the
lanthanide triflate (0.03 mmol) and molecular sieves (about 0.040 g) were
added to anhydrous CH Cl (0.3 mL) at ambient temperature in a
about 10 min, 8 (0.047 g=50 mL, 0.30 mmol) was added with a microsyr-
inge. Stirring was continued at À208C until TLC showed that all 5a re-
5
2 2
acted, the reaction was quenched in water, extracted with CH Cl and
2
2
dried; the mixture of enantiomers 13a and 14a was separated from the
pybox ligand by column chromatography (silica gel, cyclohexane/ethyl
acetate 50:50 was the eluant), and submitted to HPLC analysis using a
rubber-septum-sealed vial; the mixture was stirred 1 h and then cooled at
À508C. Compound 6 (100 mL, ca. 1.5 mmol) was added with a microsyr-
inge and stirring was continued at À508C until TLC showed that all of
the dienophile had reacted. The reaction was quenched in water, extract-
ed with CH Cl , dried, the mixture of adducts 9a and 10a was separated
2 2
from pybox 4 by column chromatography (silica gel, 30 cmL, 1,5 cm di-
Chiralpack AD column with hexane/2-propanol (1:1) as eluant
À1
(
1.0 mLmin ). The average retention times were 16 and 23 min for (S)-
3
-(2’,5’-dihydro-5’-oxo-2’-furyl)propanoyl-1,3-oxazolidin-2-one (13a) and
(
R)-14a respectively.
ameter, cyclohexane/ethyl acetate 75:25 was the eluant), and subjected to
HPLC analysis using a Chiralcel OD column with hexane/2-propanol
General procedure for the enantioselective Mukaiyama-Michael reaction
between 5b and 8: Compound 5b (0.046 g, 0.30 mmol), the pybox 4
(0.03 mmol), the lanthanide triflate (0.03 mmol) and molecular sieves
À1
(
9:1) as eluant (1.0 mLmin ). The average retention times were 20 and
2
1 min for (1’S,2’R,4’S)- and (1’R,2’S,4’R)-3-(bicyclo[2.2.1]hept-5’-ene-2’-
carbonyl)-2-oxazolidinone (10a) respectively, 22.5 and 25 min for
(about 0.040 g) were added to anhydrous CH
2
Cl
2
(0.3 mL) at ambient
Chem. Eur. J. 2005, 11, 3816 – 3824
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3823

G. Desimoni et al.
temperature in a rubber-septum-sealed vial; the mixture was stirred for
h and then cooled to À208C. The required amount (0.30 mmol) of the
additive reported in Table 6 was added, and after about 10 min, 8
0.047 g=50 mL, 0.30 mmol) was added with a microsyringe. The reaction
Acknowledgements
1
This work was supported by MIUR and the University of Pavia. Thanks
are due to CINMPIS for a fellowship to A.L.
(
was worked up as previously described for 5a and 8 and the HPLC anal-
ysis was performed on a Chiralpak AD column with hexane/2-propanol
À1
(
2
2:1) as eluant (1.0 mLmin ). The average retention times were 19 and
[
[
1] G. Desimoni, G. Faita, P. Quadrelli, Chem. Rev. 2003, 103, 3119–
154.
2] G. Desimoni, G. Faita, S. Filippone, M. Mella, M. G. Zampori, M.
7 min for (R,S)- and (S,R)-3-(2’,5’-dihydro-5’-oxo-2’-furyl)butanoyl-1,3-
oxazolidin-2-one (14b) respectively, 22 and 25 min for (R,R)- and (S,S)-
3b respectively.
Complexes of 4, 5a–c and lanthanum triflate (20a–c)—general proce-
dure: The pybox 4 (0.111 g, 0.15 mmol), La(OTf)
and 5a–c (0.15 mmol) were dissolved in CH Cl
night the small amount of undissolved solid was filtered through a pip-
ette, and pentane (about 15 mL) was added. After a few hours a white
solid separated (about 75% yield) that could be crystallised from ben-
zene/hexane. The H and C NMR spectra are reported in Table 7, addi-
tional analytical and spectroscopic data are listed below.
3
1
Zema, Tetrahedron 2001, 57, 10203–10212.
[3] G. Desimoni, G. Faita, M. Guala, C. Pratelli, Tetrahedron 2002, 58,
2929–2935.
[4] G. Desimoni, G. Faita, M. Guala, A. Laurenti, Eur. J. Org. Chem.
2004, 3057–3062.
[5] V. K. Aggarwal, L. Bell, M. P. Coogan, P. Jubault, J. Chem. Soc.
Perkin Trans. 1 1998, 2037–2042.
[6] P. Mꢀller, C. Bolꢁa, Helv. Chim. Acta 2001, 84, 1093–1111.
[7] M. Schinnerl, M. Seitz, A. Kaiser, O. Reiser, Org. Lett. 2001, 3,
4259–4262.
[8] G. Desimoni, G. Faita, M. Guala, C. Pratelli, J. Org. Chem. 2003, 68,
7862–7866.
[9] G. Desimoni, G. Faita, M. Mella, M. Boiocchi, Eur. J. Org. Chem.,
in press.
3
(0.088 g, 0.15 mmol)
(2.0 mL). After one
2
2
1
13
2
D
5
Complex 20a: M.p. 140–1428C; [a] =À22.4 (c=0.5 in CHCl
3
); IR
À1
(
(
Nujol): n˜ =1761 (C=O), 1653 cm ; MS (2.0 keV, ESI): m/z (%): 1319
+
+
5) [4+5a+La(OTf)
2
] , 1178 (100) [4+La(OTf)
LaN Si (1469.4): C 42.51, H 4.80, N 3.81;
found: C 42.31, H 5.02, N 3.77.
2
] ; elemental analysis
calcd (%) for C52
H
70
F
9
4
O
16
S
3
2
2
D
5
[10] H. Katajima, K. Ito, T. Katsuki, Tetrahedron 1997, 53, 17015–17028.
[11] H. Suga, T. Kitamura, A. Kakehi, T. Baba, Chem. Commun. 2004,
1414–1415.
Complex 20b: M.p. 148–1508C; [a] =À27.6 (c=0.5 in CHCl
3
); IR
À1
(
Nujol): n˜ =1757 (C=O), 1653 cm ; MS (2.0 keV, ESI): m/z (%): 1333
+
+
(
25) [4+5b+La(OTf)
2
] , 1178 (100) [4+La(OTf)
LaN Si (1482.3): C 42.91, H 4.89, N 3.78;
found: C 42.77, H 5.03, N 3.91.
2
] ; elemental analysis
[
[
12] S. E. Schaus, E. N. Jacobsen, Org. Lett. 2000, 2, 1001–1004.
13] H. C. Aspinall, J. L. Dwyer, N. Greeves, P. N. Smith, J. Alloys
Compd. 2000, 303–304, 173–177.
calcd (%) for C53
H
72
F
9
4
O
16
S
3
2
2
5
Complex 20c: M.p. 160–1628C; [a] =À24.2 (c=0.5 in CHCl
3
); IR
D
[
[
14] S. Furuzawa, H. Matsuzawa, K. Metoki, Synlett 2001, 709–711.
15] D. A. Evans, Z. K. Sweeney, T. Rovis, J. S. Tedrow, J. Am. Chem.
Soc. 2001, 123, 12095–12096.
À1
(
(
Nujol): n˜ =1758 (C=O), 1653 cm ; MS (2.0 keV, ESI): m/z (%): 1307
+
+
3) [4+5c+La(OTf)
2
] , 1178 (100) [4+La(OTf)
LaN Si (1456.2): C 42.03, H 4.84, N 3.84;
found: C 42.31, H 5.02, N 3.89.
Diels–Alder reaction between 20a and 6: Complex 20a (0.050 g,
.035 mmol) and molecular sieves (about 0.020 g) were added to anhy-
drous CH Cl (0.2 mL) at ambient temperature under the conditions of
the DA reaction of 5a described above. Cyclopentadiene (6, 0.015 mL)
was added at À508C and worked up as above (the chromatographic
column was a pipette). The HPLC analysis gave a ratio 9a/10a=90:10;
2’R)-9a 75% ee; 10a was nearly racemic.
Diels–Alder reaction between 20b and 6: Complex 20b (and molecular
sieves) in anhydrous CH Cl was reacted with 6 at 178C and worked up
as above. The HPLC analysis gave a ratio 9b/10b=70:30; (2’R,3’S)-9a
2
] ; elemental analysis
calcd (%) for C51
H
70
F
9
4
O
16
S
3
2
[
[
16] D. A. Evans, K. A. Scheidt, K. R. Fandrick, H. W. Lam, J. Wu, J.
Am. Chem. Soc. 2003, 125, 10780–10781.
17] S. Bambirra, M. W. Bouwkamp, A. Meetsma, B. Hessen, J. Am.
Chem. Soc. 2004, 126, 9182–9183.
0
2
2
[18] D. A. Evans, C. E. Masse, J. Wu, Org. Lett. 2002, 4, 3375–3378.
[19] H. C. Aspinal, Chem. Rev. 2002, 102, 1807–1850.
[20] D. A. Evans, K. T. Chapman, J. Bisaha, J. Am. Chem. Soc. 1988, 110,
1
238–1256.
(
[
21] K. Narasaka, N. Iwasawa, M. Inoue, T. Yamada, M. Nakashima, J.
Sugimori, J. Am. Chem. Soc. 1989, 111, 5340–5345.
2
2
Received: November 26, 2004
Published online: April 13, 2005
8
0% ee; (2’R,3’S)-10b ee was about 10%.
3824
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 3816 – 3824