The role of binding constants in the efficiency of chiral catalysts immobilized by electrostatic interactions: The case of azabis(oxazoline)-copper complexes

Article abstract of DOI:10.1002/chem.200305739

As shown by theoretical calculations, azabis(oxazoline)-copper complexes are considerably more stable than the analogous bis(oxazoline)-copper complexes. This enhanced stability allows them to be efficiently immobilized by means of electrostatic interactions to different anionic supports, such as clays and nafion-silica nanocomposites, without the loss of a ligand, as is observed for bis(oxazolines). As a result, enantioselectivities of around 90% ee are obtained in the cyclopropanation reaction between styrene and ethyl diazoacetate. Moreover, the solid catalyst is easily recoverable.

Full text of DOI:10.1002/chem.200305739

FULL PAPER
The Role of Binding Constants in the Efficiency of Chiral Catalysts
Immobilized by Electrostatic Interactions: The Case of Azabis(oxazoline)
Copper Complexes
Josÿ M. Fraile,[a] Josÿ I. GarcÌa,[a] Clara I. HerrerÌas,[a] Josÿ A. Mayoral,*[a]
Oliver Reiser,*[b] Antonio Socuÿllamos,[a] and Heiko Werner[b]
Abstract: As shown by theoretical calculations, azabis(oxazoline) copper com-
plexes are considerably more stable than the analogous bis(oxazoline) copper
complexes. This enhanced stability allows them to be efficiently immobilized by
means of electrostatic interactions to different anionic supports, such as clays and
nafion silica nanocomposites, without the loss of a ligand, as is observed for bi-
s(oxazolines). As a result, enantioselectivities of around 90% ee are obtained in
the cyclopropanation reaction between styrene and ethyl diazoacetate. Moreover,
the solid catalyst is easily recoverable.
Keywords: cyclopropanation
¥
enantioselectivity ¥ heterogeneous
catalysts ¥ immobilized catalysts ¥
N ligands
Introduction
although in spite of its simplicity, this method is not often
used.[2]
Asymmetric heterogeneous catalysts[1] are very attractive
because they have inherently practical advantages over ho-
mogeneous catalysts, and because the resultant product
enantioselectivities are at the levels required in the pharma-
ceutical or agrochemical industry. One of the most widely
used methods to prepare asymmetric heterogeneous cata-
lysts is to immobilize the homogeneous counterpart on a
solid support.[2] Although the attachment of a ligand to a
support by covalent bond formation is the most widely used
method of immobilization,[2,3] on many occasions the syn-
thetic steps required prove to be difficult. Furthermore,
binding the ligand covalently to the support may modify the
conformational preferences of the catalytic complex, and
thereby lead to changes in enantioselectivity.[4] Alternatively,
immobilization can be achieved by electrostatic interaction,
In asymmetric catalysis the binding constant of the com-
plex formed between the chiral ligand and the metal precur-
sor is generally not considered. Nevertheless, a weak com-
plex may lead to the presence of free metal precursor, and
consequently, non-enantioselective catalytic centres. In a ho-
mogeneous phase this limitation is frequently overcome by
adding an excess of chiral ligand. However, in a heterogene-
ous phase the situation is not so simple because of the phe-
nomenon of site isolation. Consequently, it is very difficult
to produce a local excess of chiral ligand on the solid. On
the other hand, it is possible to add the chiral ligand to the
liquid phase, but this strategy eliminates most of the advan-
tages of immobilization. As a result of this problem it is
much easier to find efficient chiral heterogeneous catalysts
for ligand-accelerated reactions.[5] Thus, a high binding con-
stant is more important in heterogeneous than in homogene-
ous asymmetric catalysis, and the best ligand in solution
may not be the best one in a heterogeneous phase. Herein
we illustrate this reasoning using bis(oxazoline) and azabi-
s(oxazoline) copper complexes that have been immobilized
by electrostatic interactions onto anionic supports.
[a] Dr. J. M. Fraile, Dr. J. I. GarcÌa, Dipl.-Chem. C. I. HerrerÌas,
Prof. Dr. J. A. Mayoral, Dipl.-Chem. A. Socuÿllamos
Departamento de QuÌmica Orgµnica
Instituto de Ciencia de Materiales de AragÛn, Facultad de Ciencias
Universidad de Zaragoza-C.S.I.C., 50009 Zaragoza (Spain)
Fax : (+34)976-762-077
In our work on the immobilization of bis(oxazoline)
copper complexes by electrostatic interactions, we have
found that clays[6] and nafion-based solids[7] are suitable sup-
ports,[8] but their efficiency depends on the nature of the
chiral bis(oxazoline) ligand. These divergences come from
the different binding constants of the bis(oxazoline) copper
complexes. For example, with 2,2’-isopropylidenebis[(4S)-4-
E-mail: mayoral@unizar.es
[b] Prof. Dr. O. Reiser, Dipl.-Chem. H. Werner
Institut f¸r Organische Chemie, Universit‰t Regensburg
Universit‰tstrasse 31, 93053 Regensburg (Germany)
Fax : (+49)941-4121
E-mail: Oliver.Reiser@chemie.uni-regensburg.de
2997
Chem. Eur. J. 2004, 10, 2997 3005
DOI: 10.1002/chem.200305739
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

FULL PAPER
tert-butyl-4,5-dihydro-1,3-oxazole] (1a) the corresponding
copper complex cannot be efficiently immobilized, and most
of the chiral ligand is lost during the cationic-exchange proc-
ess. Only when an excess of ligand is used in the reaction
medium are high enantioselectivities obtained in the cyclo-
Therefore, we turned our efforts towards the CuI com-
plexes, which have only one positive charge, and because of
their closed-shell electronic structure, do not contain any un-
paired electrons. Although this does not truly reflect the sit-
uation present in the exchange process, it should be noted
that the CuI complex is actually the true catalytic species.[12]
Moreover, it is also the species that is present when the loss
of a chiral ligand occurs by competitive coordination of by-
products. Thus, the higher affinity of the CuI complex
should be reflected in the attainment of higher enantioselec-
tivities when the catalyst is applied in multiple cycles. More-
over, as problems were not observed with the convergence
of the theoretical calculations for the CuI complexes investi-
gated, we will only discuss the results obtained for these.
To estimate the binding affinity of each ligand to cop-
per(i) ions, the following equilibrium was considered [Eq.
(1)].
propanation[9] of styrene with ethyl diazoacetate.[10] Unfortu-
nately, the ligand is lost during the washing of the catalyst.
The use of a catalyst in which the metal is more tightly
bound to the chiral ligand should overcome this problem.
Azabis(oxazolines)[11] 2 that contain an electron-donating
amino group in the central bridge were envisioned to be
more electron-rich ligands than bis(oxazolines) that have a
methylene group as the bridging element of the two oxazo-
line units.
In the current paper we confirm our hypotheses through
theoretical calculations, as well as by experimental data ob-
tained from the electrostatic immobilization of azabis(oxa-
zoline) copper complexes to different supports and the use
of the resultant immobilized catalysts in the asymmetric cy-
clopropanation reaction between styrene and ethyl diazoa-
cetate.
1 a þ 2 aÀCuI Ð 1 aÀCuI þ 2 a
ð1Þ
Initially, full geometrical optimizations of both the 1a CuI
and 2a CuI complexes at the Hartree Fock (HF) level and
a 6-31G(d) basis set [HF/6-31G(d)] were carried out. Some
geometrical features of the CuI complexes are shown in
Figure 1, and their energies are summarized in Table 1. Al-
Results and Discussion
Relative stability of bis(oxazoline) and azabis(oxazoline)
copper complexes: To confirm the hypothesis that copper(ii)
ions have a higher affinity for azabis(oxazoline) ligands than
bis(oxazoline) ligands, we tried to determine the binding
constants for both types of complexes by UV and microca-
lorimetric measurements. However, several practical prob-
lems in the course of our experiments did not allow us to
obtain meaningful results. Fortunately, theoretical calcula-
tions can give an accurate estimation of relative equilibrium
constant values. Initially, the CuII complexes of bis(oxazo-
lines) were considered as these complexes are used in the
exchange process for the preparation of immobilized cata-
lysts, in which ligand loss has been observed. However, two
main problems arise with the theoretical calculations of such
species. First, the interaction between a doubly charged
cation that has a highly localized positive charge and a
ligand will be greatly overestimated in a gas-phase calcula-
tion, particularly, as in solution this is not as pronounced.
The second complication arises from the open-shell charac-
ter of CuII complexes that have one unpaired electron, as
this may lead to spin contamination and self-consistent field
(SCF) convergence problems. In fact, when the CuII com-
plexes were considered in geometrical optimizations at the
UHF/6-31G(d) level, serious problems of convergence, both
in the SCF wavefunction and in the geometrical parameters,
were encountered.
Figure 1. HF/6-31G(d) structures of the 1a CuI and 2a CuI complexes
(some hydrogen atoms have been omitted for clarity).
though Hartree Fock theory does not account for correla-
tion energy, the equilibrium considered [Eq. (1)] represents
an isodesmic process. Therefore, relative energies calculated
at the HF level are expected to be reasonably accurate.[13]
From a geometrical point of view, both complexes are
À
very similar. Thus, the N Cu distances are almost the same
for both compounds (1.99 ä). The azabis(oxazoline) com-
plex is slightly unsymmetrical because of the conformation
of the methyl group in the methylamine bridge, while the
2998
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2004, 10, 2997 3005

The Role of Binding Constants in the Efficiency of Chiral Catalysts
2997 3005
Table 1. Results of theoretical calculations for the equilibria between bis(oxazoline) and azabis(oxazoline) copper(i) complexes in the presence and ab-
sence of a Cu coordinated ethylene molecule.
Theoretical level
Energy of the species (Hartrees)
DE [kcalmolÀ1
]
1
2 CuI
1 CuI
2
HF/6-31G(d)
À919.853477
À919.857294
À925.827030
À925.830096
À2535.346509
À2535.393826
À2542.684486
À2542.730153
À2558.380100
À2558.426551
À2565.956341
À2566.001154
À896.811063
À896.811500
À902.549634
À902.552750
+5.5
+6.0
+3.5
+4.0
IPCM/HF/6-31G(d)
B3LYP/6-31G(d)
IPCM/B3LYP/6-31G(d)
1
2 CuI C2H4
1 CuI C2H4
2
HF/6-31G(d)
À919.853477
À919.857294
À925.827030
À925.830096
À2613.404706
À2613.450238
À2621.323431
À2621.368062
À2636.434460
À2636.480074
À2644.594131
À2644.638507
À896.811063
À896.811500
À902.549634
À902.552750
+7.9
+7.8
+4.2
+4.3
IPCM/HF/6-31G(d)
B3LYP/6-31G(d)
IPCM/B3LYP/6-31G(d)
À
À
N Cu N angle is somewhat open in the bis(oxazoline) com-
plex. The latter undoubtedly arises because of the smaller
C2-C-C2’ angle (116.38) in comparison to the C2-N-C2’
angle (123.68), although it should be kept in mind that both
chelate complexes are almost completely planar.
However, to our surprise, the second-order perturbation
energy calculations revealed that the interaction between
the empty d orbital of the Cu atom and the lone pairs of the
oxazoline nitrogen atoms are actually stronger in the bis(ox-
azoline) 1a ligand (29.5 kcalmolÀ1 for the 1a CuI complex
in comparison to 28.7 kcalmolÀ1 for the 2a CuI complex).
Similarly, charge transfer from the ligand to the Cu atom is
marginally greater for the bis(oxazoline) ligand (0.040 elec-
tron for the 1a CuI complex in comparison to 0.032electron
for the 2a CuI complex), and therefore, indicates that the
azabis(oxazoline) ligand does not have a more electron-do-
nating ability.
The electronic effect of the bridging nitrogen atom is no-
ticeable on the atomic charges of the oxazoline nitrogen
atoms. Thus, natural population analysis (NPA) shows that
the net atomic charges of the oxazoline nitrogen atoms are
À0.705 electron for the 1a CuI complex and À0.747 electron
for the 2a CuI complex. Figure 2shows the calculated mo-
lecular electrostatic potentials for both complexes, and as
can be seen, the negative charge (light gray zone encircled
The calculated energy variation for this equilibrium is
5.5 kcalmolÀ1. This suggests that the azabis(oxazoline) CuI
complex is more stable, and thus, confirms our initial hy-
pothesis. However, calculations carried out in a vacuum may
not necessarily represent the behaviour observed in solution.
Therefore, polarity solvent effects were taken into account
using the isodensity polarizable surface continuum method
(IPCM)[14] and the dielectric constant of dichloromethane
(e=8.9), which was the solvent used in the cyclopropanation
reaction. The results (Table 1) show that solvation of analo-
gous species on either side of the equilibrium is very similar.
The solvation energy calculated for the bis(oxazoline) 1a
and azabis(oxazoline) 2a ligands was 2.4 and 2.5 kcalmolÀ1,
respectively. On the other hand, the solvation energy of the
corresponding copper complexes was determined to be 29.1
and 29.7 kcalmolÀ1, respectively, as is expected for charged
species. As a consequence, the calculated energy of the equi-
librium in solution (6.0 kcalmolÀ1) was found to be very sim-
ilar to that determined in a vacuum.
Since Hartree Fock calculations do not account for elec-
tronic correlation energy, a density functional theory (DFT)
method was then used on the B3LYP/6-31G(d) basis set
both in a vacuum and in dichloromethane.[14] As can be seen
(Table 1), the calculated energies of 3.5 and 4.0 kcalmolÀ1
for the equilibria in vacuum and solution, respectively, are
slightly lower than those obtained by the HF method. How-
ever, the general conclusion from analysis of both methods
is the same, namely that the binding constant is not effected
greatly by solvation and that it is clearly higher for the aza-
bis(oxazoline) copper complex than for the analogous bi-
s(oxazoline) copper complex.
To obtain a deeper insight into the origin of the differen-
tial coordination ability of both kinds of ligands, we carried
out a natural bond orbital (NBO) analysis on each of the
CuI complexes at the HF/6-31G(d) level. From first analysis,
the coordination preference of the azabis(oxazoline) ligand
could be ascribed to the electron-donating ability of the
bridging nitrogen atom, which would cause the oxazoline ni-
trogen atoms to have an enhanced electron-donor character.
Figure 2. Molecular electrostatic potential maps of the 1a CuI and 2a
CuI complexes. Dark gray indicates positive zones, while light gray indi-
cates negative zones. The negative zone around the coordinating nitrogen
atom is encircled.
2999
Chem. Eur. J. 2004, 10, 2997 3005
www.chemeurj.org
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

J. A. Mayoral, O. Reiser et al.
FULL PAPER
and marked by a white arrow in Figure 2) is more concen-
trated on the oxazoline nitrogen atoms in the 2a CuI com-
plex. From these results, it can be proposed that the differ-
ent coordinating ability of ligand 2a over ligand 1a possibly
arises from an enhanced electrostatic interaction between
the Cu cation and the ligand.
To corroborate this explanation, a natural energy decom-
position analysis (NEDA)[15] was carried out on both com-
plexes. This analysis evaluates the interaction energy be-
tween the CuI center and the ligand by breaking it up into
several terms. Thus, in agreement with previous analyses the
charge-transfer term was found to be more important
(À88.4 kcalmolÀ1) for ligand 1a than for ligand 2a
(À85.1 kcalmolÀ1). On the other hand, the reverse is true
for the electrostatic term (À141.2kcalmol À1 for the 1a CuI
complex in comparison to À143.5 kcalmolÀ1 for the 2a CuI
complex). Overall, the interaction energy is higher for the
2a CuI complex than for the 1a-C-uI complex (À80.2and
À78.1 kcalmolÀ1, respectively). This is in agreement with
previous calculations.
A question that concerns the model compounds used for
[16]
this theoretical study arises. As already mentioned,
the
CuI complexes considered are fourteen-electron complexes,
and consequently, are probably not the true catalytic inter-
mediates. Therefore, to test the correctness of the conclu-
sions reached, we repeated the calculations using the initial
complex in the catalytic cycle, namely the ligand CuI olefin
complex. To this end, we considered the following equilibri-
um [Eq. (2)].
Figure 3. HF/6-31G(d) structures of the 1a CuI ethylene and 2a CuI
ethylene complexes (some hydrogen atoms have been omitted for clari-
ty).
plex). Finally, the NEDA results showed that the charge-
transfer term is more important for the 2a CuI ethylene
complex (À111.9 kcalmolÀ1) than for the 1a CuI ethylene
complex (À110.3 kcalmolÀ1), but that above all, the electro-
static term is much more important for the former
(À148.3 kcalmolÀ1) than the latter (À142.6 kcalmolÀ1).
Overall, the calculated interaction energy is stronger for the
2a CuI ethylene complex (À96.0 kcalmolÀ1 in comparison
to À92.5 kcalmolÀ1 for the 1a CuI ethylene complex). This
is in agreement with previous calculations.
1 a þ 2 aÀCuIÀethylene Ð 1 aÀCuIÀethylene þ 2 a
ð2Þ
The same theoretical levels considered in the former
study were also used for these compounds. In particular, full
geometrical optimizations at the Hartree Fock level using
the 6-31G(d) basis set and single-point energy calculations
at the B3LYP/6-31G(d), IPCM/HF/6-31G(d), and IPCM/
B3LYP/6-31G(d) theoretical levels were undertaken. Some
geometrical features of the CuI complexes are shown in
Figure 3 and their energies are summarized in Table 1.
The same trends observed previously are also reproduced
for these complexes. In particular, the equilibrium is shifted
to the left, and indicates a preferential coordination of the
azabis(oxazoline) 1a ligand over the corresponding bis(oxa-
zoline) 2a ligand. Moreover, the energy differences are even
greater than those calculated for Equation (1).
The NBO analyses also agreed with the former observa-
tions, and NPA revealed that the coordinating nitrogen
atoms are more negatively charged in the 2a CuI ethylene
complex (À0.827 electron) than in the 1a CuI ethylene
complex (À0.774 electron). This indicates that the bridging
nitrogen atom does have an electron-donating role. In con-
trast, the Cu atom was found to have an almost identical
charge in both complexes (0.982electron in the 2a CuI eth-
ylene complex and 0.981 electron in the 1a CuI ethylene
complex). Moreover, the occupancy of the empty Cu orbital
In summary, azabis(oxazoline) CuI complexes are more
stable than the analogous bis(oxazoline) CuI complexes be-
cause of the presence of stronger electrostatic metal ligand
interactions in the former. These in turn result from the
electron-donating ability of the bridging nitrogen atom. This
fact has important consequences for the electrostatic immo-
bilization of these complexes, since overall the Cu atom
keeps most of its positive charge, and hence can establish an
efficient interaction with the negatively-charged support.
Control of enantioselectivity by competitive ligand complex-
ation: In principle, the higher metal ligand affinity the less
Lewis acidic the complex, and consequently, the less active
the catalyst. On the other hand, a weakly binding ligand
might not sufficiently activate a metal for catalysis. To
obtain reliable kinetic data for copper(i)-catalyzed diazoace-
tate cyclopropanations is problematic because the reaction
greatly depends upon the slow addition of reagent in order
to attain a low concentration of diazoacetate. Therefore, to
gain insight into the relative activities of bis(oxazoline) 1
À
implicated in the N Cu charge transfer is also very similar
for both complexes (0.128 electron in the 2a CuI ethylene
complex and 0.126 electron in the 1a CuI ethylene com-
3000
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2004, 10, 2997 3005

The Role of Binding Constants in the Efficiency of Chiral Catalysts
2997 3005
and azabis(oxazoline) 2 copper complexes, we conducted a
series of competition experiments (Table 2) in which cop-
per(i)-catalyzed cyclopropanations in homogeneous solution
were carried out in the presence of mixtures of 1a/2b and
1b/2a, respectively. Since the tert-butyl-substituted ligands
viously described,[11] but an alternative method was necessa-
ry to prepare ligand 2c (Scheme 1). (S)-2-Amino-4-phenyl-
4,5-dihydro-1,3-oxazole (5) and (S)-2-ethoxy-4-phenyl-4,5-
dihydro-1,3-oxazole (7), which are readily prepared from
(S)-phenylglycinol, were condensed in the presence of p-tol-
uenesulfonic acid (p-TSA) to
give 8. Subsequent methylation
of 8 yielded azabis(oxazoline)
2c, and the copper(ii) com-
plexes were then prepared in
dichloromethane using equimo-
lecular amounts of Cu(OTf)2
and either chiral ligands 1 or 2.
Two different types of sup-
port were used, namely a syn-
Table 2. CuI-catalyzed cyclopropanations in the presence of bis(oxazoline) 1 and/or azabis(oxazoline) 2
ligands.[a]
thetic clay (laponite) and
a
Entry
Ligand 1
Ligand 2
(1 + 2)/Cu[b]
Yield [%]
3:4[c]
ee of 3[d] [%]
ee of 4[d] [%]
nafion-like material prepared
by grafting (HO)3Si(CH2)3-
(CF2)2O(CF2)2SO3K onto silica[17]
(Figure 4). Similar nafion silica
hybrid materials have demon-
strated to be excellent for im-
mobilization of the 1c CuII
complex,[7] but have the disad-
vantage of low functionaliza-
tion. The new solid used in the
1[e]
2[e]
3
4
5
1a
1b
1.1
1.1
2.2
2.2
5.0
2.2
2.2
77
n.d.
82
85
23
73:27
69:31
73:27
62:38
89:11
62:38
66:34
99
49
85
56
54
77
61
97
45
75
42
41
67
47
2a
2b
2b
2a
2b
1a
1b
1a
6
7
73
76
[a] Reaction conditions: diazoester (1 mmol) in CH2Cl2 (7 mL), styrene (3 mmol) in CH2Cl2 (3 mL), Cu(OTf)2
(0.01 mmol, 1 mol%), room temperature. [b] Equimolar amounts of 1 and 2 were used. [c] Determined by GC
using a DB 1301 column. [d] Determined by GC using a CP-Chiralsil DEX CB. [e] Taken from reference [18],
five equivalents of styrene were used.
1a and 2a give substantially
higher enantioselectivities than
their
isopropyl-substituted
counterparts 1b and 2b, analy-
sis of the enantioselectivities
obtained with mixtures of 1a/
2b and 1b/2a should shed light
on the dominant ligand in the
catalysis under investigation.
Indeed, reactions performed
in the presence of both bis(oxa-
zoline) 1 and azabis(oxazoline)
2 ligands clearly seem to be do-
minated by the latter. The
Scheme 1. Synthesis of N,N-bis[(4S)-4-phenyl-4,5-dihydro-1,3-oxazol-2-yl]methylamine (2c).
enantioselectivities obtained (entries 5 7, Table 2) closely
resemble the values observed when azabis(oxazoline) 2
ligands are used alone (entries 3 and 4, Table 2). Since for-
mation of the Cu carbene intermediate, which constitutes
the rate-limiting step of the cyclopropanation,[16] is very fast,
the reaction is probably not under Curtin Hammett condi-
tions. As a result, the reaction preferably takes place
through the most stable complex rather than being depend-
ent upon the relative activity of the complexes present in
equilibrium. In turn, we can conclude that azabis(oxazoline)
2 appears to have a higher binding affinity to copper than
bis(oxazoline) 1.
Figure 4. Structure of the nafion silica support.
current investigations greatly improved the cationic-ex-
change capacity and decreased the apparent molecular
weight of the immobilized complex, thereby making it much
more practical to use. The exchange was carried out in
methanol, and the solids were thoroughly washed and then
dried under vacuum. The presence of the complexes was
confirmed by elemental analysis and IR spectroscopy. For
the laponite-based solids the copper content was in the
range of 0.1 0.2mmolg À1, and as shown by the XRD pat-
Exchange of CuII complexes: The higher stability of azabox
Cu complexes prompted us to immobilize them onto anionic
solid supports using the general methodology described in
previous papers.[7] Ligands 2a and 2b were prepared as pre-
3001
Chem. Eur. J. 2004, 10, 2997 3005
www.chemeurj.org
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

J. A. Mayoral, O. Reiser et al.
FULL PAPER
tern of oriented samples, expansion was not observed in the
basal spacing. These results show that 2 CuII complexes are
mostly exchanged on the external surface of the clay. Typical
copper contents for the nafion silica solids were in the
range 0.25 0.35 mmolgÀ1.
The prepared solids were then tested in the benchmark
cyclopropanation reaction of styrene and ethyl diazoacetate.
The reagents were used in equimolecular amounts to allow
the results to be compared with those obtained for the anal-
ogous solids prepared from bis(oxazoline) 1. The same
amount of catalyst was used each time, but depending on
the copper loading of the catalyst, had a range between 0.3
and 1%. The results for both types of ligands are summar-
ized in Table 3. When laponite was used as the support, tert-
copper(i) has for azabis(oxazoline) 2 in comparison to bi-
s(oxazoline) 1.
Bis(oxazoline) 1a, which contains tert-butyl groups, was
not the only compound that displayed problems in the cati-
onic exchange. Ligand 1b, which contains isopropyl groups,
showed the same limitations and afforded very poor enan-
tioselectivities even on a laponite support. In confirmation
of the general premise that azabis(oxazoline) Cu complexes
are more stable, the catalysts prepared with ligand 2b on
either laponite or nafion silica supports gave rise to enantio-
selectivities that were in good agreement with the results
obtained in homogenous solution.
In contrast to the bis(oxazoline) ligands that bear
branched alkyl groups, the phenyl-substituted ligand 1c did
not display stability problems
either in cationic exchange onto
the solid supports or in biphasic
Table 3. Results obtained for the CuII-catalyzed cyclopropanation reactions.[a]
systems in which ionic liquids
were used.[19] Therefore, the
phenyl-substituted azabis(oxa-
zoline) 2c was prepared and
tested for comparative purpos-
es. Once again, azabis(oxazo-
line) 2c gave significantly
Ligand
Support
Run
Yield [%]
36
trans/cis
% ee trans
% ee cis
better yields and selectivities
than bis(oxazoline) 1c. As ex-
pected, the enantioselectivities
for both 1c and 2c were very
similar, and were close to those
1a
Laponite
1
235
1
245
1
1
22
1
1
22 :38
1
64:36
69
64
58:4243
37
83
74
2a
Laponite
46
9
71:29
76
16
71:2 81
1a
2a
Nafion silica
Nafion silica
37
60:40
68:3288
66:34
51:49
62:38
42
18
30[b]
81
obtained
phase.
in
homogeneous
[b]
1
84
5
54
77
13
43
1b
2b
Laponite
Laponite
31
39
62 53
31
63:37
28
From these results it can be
concluded that, in general,
higher yields and selectivities
are obtained with azabis(oxazo-
line) catalysts that have been
immobilized onto solid supports
than with the corresponding bis-
(oxazoline) catalysts. However,
when the copper(ii) complex of
azabis(oxazoline) 2a was immo-
bilized onto nafion silica, the
mixture had to be heated at
6
4
2b
Nafion silica
63:37
55
47
49
51
47
22
1
1
55
1c
2c
Laponite
Laponite
61:39
64:36
24
38
49
[a] Reaction conditions: styrene (5 mmol), ethyl diazoacetate (5 mmol, slowly added in two portions), catalyst
(150 mg), CH2Cl2 (5 mL), room temperature. Yield and selectivities were determined by gas chromatography
(methyl silicone and cyclodex-B columns). 3R and 4R are the major isomers. [b] Heating at 508C for some mi-
nutes was necessary to reduce the CuII ions to CuI ions.
butyl-substituted bis(oxazoline) 1a, which was found to be
the most selective ligand for the asymmetric cyclopropana-
tion of styrene in homogeneous phases,[18] afforded moder-
ate results (up to 69% ee compared to 99% ee in homoge-
nous solution). On the other hand, the analogous azabis(ox-
azoline) 2a performed considerably better. Besides giving
rise to higher yields and improved trans/cis selectivity, the
major trans-cyclopropane (R)-3 was obtained in 83% ee.
The difference between the two ligands was even more pro-
nounced when nafion silica was used as the support. In this
case, bis(oxazoline) 1a gave very poor enantioselectivities
(ꢀ18% ee) because of ligand loss during the exchange proc-
ess, whereas azabis(oxazoline) 2a gave an ee (88%) very
close to the value described in homogeneous phase
(91% ee).[11] In agreement with our theoretical calculations,
these results clearly demonstrate the higher affinity that
508C for some time in order for a reaction to occur. This
problem arises because the CuII ions must firstly be reduced
to the CuI cation as this is actually the active species in the
cyclopropanation reaction.[12,16] Once the copper is reduced,
the cyclopropanation reaction is fast at room temperature.
As a result of these observations, we decided to use cop-
per(i) complexes directly in the exchange process.
Exchange of CuI complexes: In our initial studies on electro-
static immobilization copper(i) complexes had been tested,
but no apparent advantages over the copper(ii) species had
been observed.[6] Furthermore, copper(i) salts and complexes
are less stable in air and are prone to disproportionate; this
makes their handling more difficult. However, the difficul-
ties encountered with the in situ reduction of the nafion
silica immobilized 2a CuII complex made the direct immo-
3002
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2004, 10, 2997 3005

The Role of Binding Constants in the Efficiency of Chiral Catalysts
2997 3005
bilization of a copper(i) complex of 2a desirable. Initially,
synthesis of the complex using CuCl as the copper(i) source
and the exchange process were conducted under an atmos-
phere of argon. Surprisingly, although the resultant solid
was accidentally exposed to air, excellent results were still
obtained and indicated that it was possible to prepare im-
mobilized copper(i) complexes without the use of an inert
atmosphere. Thus, CuCl complexes of 2a were prepared and
then exchanged in the same way as the complexes obtained
from Cu(OTf)2. The results for the subsequent cyclopropa-
nation reactions using these catalysts are summarized in
Table 4.
moval of these byproducts should improve catalyst perform-
ance, and protocols to overcome this kind of limitation are
currently under development.
Conclusion
The affinity of a metal towards ligands that act as chiral pro-
moters plays a crucial role in the electrostatic immobiliza-
tion of chiral catalysts. Theoretical calculations indicate that
azabis(oxazoline) copper complexes are significantly more
stable than the analogous bis(oxazoline) copper complexes,
and that this is the case because
electrostatic metal ligand inter-
actions make much more im-
Table 4. Results obtained for the CuI-catalyzed cyclopropanation reactions.[a]
portant contributions in the
former. This hypothesis was
corroborated by the efficient
electrostatic immobilization of
the 2a CuI complex, and by its
successful application in the
asymmetric cyclopropanation of
Ligand
Support
Run
Yield [%]
40
trans/cis
% ee trans
% ee cis
styrene in contrast to the results
obtained with the analogous bi-
s(oxazoline) catalyst 1a CuI.
The correct choice of support
and copper precursor, namely
the use of nafion-like supported
species on silica and CuCl,
leads to results that are compa-
rable to the optimized results
obtained in the homogeneous
phase (up to 90% ee).[11] Al-
though partial loss of enantiose-
lectivity in the third cycle as a
result of competitive ligand de-
2a
Laponite
1
234
1
236
3
1
235
1
22 :38
69:31
81
58
63:37
61
66:34
89
63:37
57:43
54
63:37
45
35
2a
Nafion silica
60
90
83
63:37
81
51
35
49
40
54
35
51
2b
2b
Laponite
55:45
Nafion silica
33
62 53
54
46
5
[a] Reaction conditions: styrene (5 mmol), ethyl diazoacetate (5 mmol, slowly added in two portions), catalyst
(150 mg), CH2Cl2 (5 mL), room temperature. Yield and selectivities were determined by gas chromatography
(methyl silicone and cyclodex-B columns). (R)-3 and (R)-4 are the major isomers.
Only minor changes were observed for the catalysts ob-
tained from ligand 2b and CuCl in comparison to the analo-
gous catalysts obtained from Cu(OTf)2, and which are sub-
sequently reduced during the reaction. Yields were slightly
higher for the CuI complexes, as was the enantioselectivity
obtained for the cis isomers. The results obtained for the
2a copper(i) complex depended upon the support used.
Slightly lower yields and enantioselectivities were obtained
when laponite was used, and the catalyst became less effi-
cient upon recovery. However, on a nafion silica support,
the use of CuI complexes was completely, if only slightly, ad-
vantageous. The reaction takes place at room temperature
in 60% yield and 90% ee for the trans isomer, and only
one equivalent of styrene is required. Moreover, the catalyst
can be efficiently re-used once after recovery, whereby the
same enantioselectivities are obtained. Unfortunately, in a
third run a substantial loss of enantioselectivity was detect-
ed. We attributed this catalyst deactivation to the formation
of byproducts such as diethyl maleate and its polymers,
which are formed from the undesired dimerization and oli-
gomerization of diazoacetate. These in turn are able to form
copper complexes that act as non-chiral active sites for the
cyclopropanation reaction. Consequently, suppression or re-
complexation was observed, the results show that efficient
heterogeneous catalysts can be prepared by electrostatic im-
mobilization when the binding constant of the chiral com-
plex is high enough so as to make the complex stable against
the cationic-exchange process.
Experimental Section
Theoretical calculations: Full geometrical optimizations at the HF/6-
31G(d) theoretical level were carried out for the CuI complexes of 2,2’-
isopropylidenebis[(4S)-4-tert-butyl-4,5-dihydro-1,3-oxazole] and N,N-
bis[(4S)-4-tert-butyl-4,5-dihydro-1,3-oxazol-2-yl]methylamine, both in the
presence and absence of a Cu coordinated ethylene molecule, as well as
for different conformations of the free ligands. On the basis of these geo-
metries, single-point energy calculations through the B3LYP hybrid func-
tional were carried out using the same basis set and a DFT method.[20]
Solvation effects through single-point energy calculations were also taken
into account, both at the Hartree Fock and DFT levels, using the IPCM
continuum model.[14] These calculations were named IPCM/HF/6-31G(d)
and IPCM/B3LYP/6-31G(d). All the calculations were carried out with
the Gaussian 98 package.[21]
NBO calculations were carried out on the basis of the HF/6-31G(d)
wavefunctions using the NBO 5.0 program[22] as implemented in the
NWChem package.[23]
3003
Chem. Eur. J. 2004, 10, 2997 3005
www.chemeurj.org
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

J. A. Mayoral, O. Reiser et al.
FULL PAPER
Molecular electrostatic potential plots were generated with the
Titan 1.0.5 program[24] by single-point calculations at the HF/LACVP*
level using the geometries previously optimized at the HF/6-31G(d)
level. The LACVP* basis set uses the standard split-valence double-z 6-
31G(d) basis set for the light elements, and a Hay Wadt pseudopotential
for Cu.[25]
argon. After 20 min, methyl iodide (156 mL, 2.5 mmol) was added to the
resultant red solution, and the reaction mixture was stirred at room tem-
perature for 10 h. The solvent was then evaporated under reduced pres-
sure, the residue was dissolved in CH2Cl2 (5 mL), and the organic phase
was washed with saturated NaHCO3 solution (5 mL). The aqueous phase
was further extracted with CH2Cl2 (3î5 mL), the combined organic
phase was dried with MgSO4, and the solvent was evaporated under re-
duced pressure to give N,N-bis[(4S)-4-phenyl-4,5-dihydro-1,3-oxazol-2-
yl]methylamine in quantitative yield. M.p. 60 628C; [a]2D0 =À61.4 (c=1.0
in CH2Cl2); 1H NMR (CDCl3): d=7.39 7.25 (m, 10H), 5.20 (dd, J=7.5
and 9.3 Hz, 2H), 4.79 (dd, J=8.4 and 9.3 Hz, 2H), 4.25 (dd, J=7.5 and
8.4 Hz, 2H), 3.52ppm (s, 3H); 13C NMR (CDCl3): d=158.9, 142.5, 128.6,
127.5, 126.5, 76.3, 67.4, 37.3 ppm; MS (70 eV, EI): m/z (%): 320.8 (100)
[M]+ ; HRMS calcd for C19H19O2N3: 321.1475; found: 321.1477.
Synthesis of N,N-bis[(4S)-4-phenyl-4,5-dihydro-1,3-oxazol-2-yl]methyl-
amine (2c)
(S)-2-Amino-4-phenyl-4,5-dihydro-1,3-oxazole (5): Sodium cyanide
(5.72g, 33 mmol) was added in small portions to a solution of bromine
(5.27 g, 33 mmol) in methanol (40 mL) at 08C. A solution of (S)-phenyl-
glycinol (4.11 g, 30 mmol) in methanol (7 mL) was then added and the
mixture was stirred for 1 h. After treatment with ammonia (15 mL, 25%
w/w), most of the solvent was evaporated under reduced pressure. The
residue was dissolved in NaOH (20%) and was then extracted with ethyl
acetate (4î40 mL). The combined organic phases were dried with
MgSO4, the solvent was evaporated under reduced pressure, and the re-
maining phenylglycinol was removed by kugelrohr distillation (608C,
0.01 Torr). Yield: 90%; 1H NMR (CDCl3): d=7.29 (m, 5H), 5.12 (brs,
2H), 5.07 (dd, J=7.4 and 9.1 Hz, 1H), 4.59 (dd, J=7.8 and 9.1 Hz, 1H),
4.02ppm (dd, J=7.4 and 7.8 Hz, 1H); 13C NMR (CDCl3): d=162.0,
143.6, 128.5, 127.2, 126.3, 75.1, 67.4 ppm.
Laponite-immobilized catalysts: Laponite (375 mg) was dried under
vacuum for 24 h prior to use. The chiral ligand (0.11 mmol) and the
copper precursor (Cu(OTf)2 or CuCl, 0.11 mmol) were dissolved in a
minimum amount of anhydrous CH2Cl2 under an atmosphere of argon,
the resultant solution was stirred for 15 min and was then filtered
through a PTFE microfilter. The solvent was evaporated under vacuum,
the complex was redissolved in methanol (4 mL), laponite was added,
and the suspension was stirred for 24 h at room temperature. The solid
was filtered off, washed with methanol (10 mL) and CH2Cl2 (20 mL), and
was then dried under vacuum for 24 h.
(S)-4-Phenyl-4,5-dihydro-1,3-oxazol-2-one
66.8 mmol) was slowly added to anhydrous ethanol (147 mL), and after it
was completely dissolved, solution of (S)-phenylglycinol (10.89 g,
(6):
Sodium
(1.53 g,
Nafion silica immobilized catalysts: The sodium form of the support was
prepared by passing a 2m NaCl solution through a column of the acidic
form until the pH was neutral. The sodium form was then washed with
deionized water and dried under vacuum at 1408C for 4 h prior to use.
The complex (0.19 mmol) was prepared in CH2Cl2 as previously descri-
bed. Nafion silica (475 mg) was added to a solution of the complex in
methanol, and the suspension was stirred for 24 h at room temperature.
The resultant solid was filtered, washed, and dried as previously descri-
bed.
a
66.74 mmol) in anhydrous ethanol (100 mL) was added followed by di-
ethyl carbonate (8.65 g, 73.3 mmol). The mixture was heated under reflux
for 15 h, then it was cooled and concentrated under vacuum. The residue
was dissolved in CH2Cl2 (196 mL) and the solution was washed with satu-
rated NH4Cl (98 mL). The aqueous phase was then further extracted
with CH2Cl2 (2î98 mL), the combined organic phases were dried with
Na2SO4, and the solvent was evaporated under reduced pressure. The
product was purified by crystallization from diethyl ether. Yield: 9.26 g
1
Characterization of the catalysts: Copper analyses were carried out by
plasma-emission spectroscopy on a Perkin-Elmer Plasma 40 emission
spectrometer. Elemental analyses were carried out on a Perkin-Elmer
2400 elemental analyzer. Step-scanned X-ray diffraction patterns of ori-
ented samples were collected at room temperature from 38 in 2q up to
608 using a D-max Rigaku system that contains a rotating anode. The dif-
fractometer was operated at 40 kV and 80 mA, and the CuKa radiation
was selected using a graphite monochromator. Transmission FTIR spec-
tra of self-supported wafers evacuated (<10À4 Torr) at 508C were taken
with a Mattson Genesis Series FTIR.
(85%); H NMR (CDCl3): d=7.34 (m, 5H), 6.53 (brs, 1H), 4.95 (dd, J=
7.0 and 8.6 Hz, 1H), 4.71 (t, J=8.6 Hz, 1H), 4.15 ppm (dd, J=7.0 and
8.6 Hz, 1H); 13C NMR (CDCl3): d=160.0, 139.5, 129.1, 128.6, 125.9, 72.4,
56.3 ppm.
(S)-2-Ethoxy-4-phenyl-4,5-dihydro-1,3-oxazole (7): A solution of ethyl-
oxonium trifluoroborate (6.05 g, 31.9 mmol) in anhydrous CH2Cl2
(50 mL) was added dropwise to a solution of (S)-4-phenyl-4,5-dihydro-
1,3-oxazol-2-one (4 g, 24.5 mmol) in anhydrous CH2Cl2 (50 mL) at 08C
under an atmosphere of argon. The reaction was stirred at room temper-
ature overnight and then it was slowly poured over a cold, saturated
Na2CO3 solution (100 mL). The organic phase was separated and the
aqueous phase was extracted with CH2Cl2 (3î25 mL). The combined or-
ganic phases were dried with MgSO4, and the solvent was evaporated
under reduced pressure to give the product as a yellow oil. Yield: 97%;
1H NMR (CDCl3): d=7.31 (m, 5H), 5.13 (dd, J=7.6 and 9.5 Hz, 1H),
4.72(dd, J=8.2and 9.5 Hz, 1H), 4.37 (c, J=7.1 Hz, 2H), 4.17 (dd, J=7.6
and 8.2Hz, 1H), 1.39 ppm (t, J=7.1 Hz, 3H); 13C NMR (CDCl3): d=
163.8, 142.8, 128.7, 127.6, 126.4, 75.6, 67.1, 66.9, 14.4 ppm.
Cyclopropanation reactions: The solid catalyst (150 mg) was added to a
solution of styrene (520 mg, 5 mmol) and n-decane (100 mg, internal
standard) in anhydrous CH2Cl2 (5 mL). A solution of ethyl diazoacetate
(290 mg, 2.5 mmol) in anhydrous CH2Cl2 (0.5 mL) was then slowly added
(2h) to the reaction mixture with a syringe pump, and the reaction was
monitored by GC.[7] After complete conversion of diazoacetate, a second
portion was slowly added in the same manner. After the reaction was
complete (typically 24 h), the catalyst was filtered off and washed with
CH2Cl2 (5 mL), and a third portion of diazoacetate was then added to
this solution to confirm the loss of catalytic activity. The solid was subse-
quently washed with CH2Cl2, dried under vacuum, and reused under the
same conditions.
N,N-Bis[(4S)-4-phenyl-4,5-dihydro-1,3-oxazol-2-yl]amine (8):
A small
amount of p-toluensulfonic acid was added to a solution of (S)-2-amino-
4-phenyl-4,5-dihydro-1,3-oxazole (0.972g, 6.0 mmol) and ( S)-2-ethoxy-4-
phenyl-4,5-dihydro-1,3-oxazole (0.955 g, 5.0 mmol) in anhydrous toluene
under an atmosphere of argon, and the mixture was heated at 508C for
24 h. The solvent was evaporated under reduced pressure and the crude
product was purified by chromatography on silica to yield 8 as an oil.
Crystals of 8 were obtained by recrystallization from acetone. Yield:
537 mg (35%); Rf =0.17 (ethyl acetete/hexanes 9:1); m.p. 198 2018C;
[a]2D0 =+475.8 (c=1.0 in CHCl3); 1H NMR (300 MHz, CDCl3): d=7.39
7.25 (m, 10H), 5.13 (dd, J=7.3 and 9.3 Hz, 2H), 4.72 (dd, J=8.6 and
9.3 Hz, 2H), 4.18 ppm (dd, J=7.3 and 8.6 Hz, 2H); 13C NMR (CDCl3):
d=166.4, 141.3, 128.9, 128.2, 126.4, 73.6, 63.1 ppm; MS (CI, NH3): m/z:
308.3 [M+H]+ ; elemental analysis calcd (%) for C18H17O2N3: C 70.34, H
5.58, N 13.67; found: C 70.36, H 5.49, N 13.63.
Acknowledgements
This work was made possible by the generous financial support of the
CICYT (project PPQ2002-04012), the MCYT and DAAD (AcciÛn Inte-
grada HA2001-0096), and the DGA. C.I.H. is indebted to the MCYT for
a grant. Dr. Mark A. Harmer (Central Research and Development,
DuPont) and Laporte Adsorbents are greatly acknowledged for their
generous gift of nafion silica support and laponite, respectively.
N,N-Bis[(4S)-4-phenyl-4,5-dihydro-1,3-oxazol-2-yl]methylamine
nBuLi (344 mL, 1.5 N in hexane, 0.52mmol) was added to a solution of
N,N-bis[(4S)-4-phenyl-4,5-dihydro-1,3-oxazol-2-yl]amine (154 mg,
0.5 mmol) in anhydrous THF (5 mL) at À788C under an atmosphere of
(2c):
[1] H.-U. Blaser, B. Pugin in Chiral Reactions in Heterogeneous Cataly-
sis, (Eds.: G. Jannes, V. Dubois), Plenum Press, New York, 1995,
pp. 33 57.
3004
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2004, 10, 2997 3005

The Role of Binding Constants in the Efficiency of Chiral Catalysts
2997 3005
[2] Chiral Catalysts Immobilization and Recycling, (Eds.: D. E. De Vos,
I. F. J. Vankelecom, P. A. Jacobs), Wiley-VCH, Weinheim, 2000.
[3] a) N. E. Leadbeater, M. Marco, Chem. Rev. 2002, 102, 32 17; b) C. A.
McNamara, M. J. Dixon, M. Bradley, Chem. Rev. 2002, 102, 3275;
c) Q.-H. Fan, Y.-M. Li, A. S. C. Chan, Chem. Rev. 2002, 102, 3385;
d) C. E. Song, S.-G. Lee, Chem. Rev. 2002, 102, 3495.
[18] D. A. Evans, K. A. Woerpel, M. M. Hinman, M. M. Faul, J. Am.
Chem. Soc. 1991, 113, 726.
[19] J. M. Fraile, J. I. GarcÌa, C. I. HerrerÌas, J. A. Mayoral, D. Carriÿ, M.
Vaultier, Tetrahedron: Asymmetry 2001, 12, 1891.
[20] a) C. Lee, W. Yang, R. Parr, Phys. Rev. 1988, 37, 785; b) A. D.
Becke, J. Chem. Phys. 1993, 98, 5648.
[4] See for example: B. Altava, M. I. Burguete, J. M. Fraile, J. I. GarcÌa,
S. V. Luis, J. A. Mayoral, M. J. Vicent, Angew. Chem. 2000, 112,
1563; Angew. Chem. Int. Ed. Engl. 2000, 39, 1503.
[5] D. J. Berrisford, C. Bolm, K. B. Sharpless, Angew. Chem. 1995, 107,
1159; Angew. Chem. Int. Ed. Engl. 1995, 34, 1059.
[6] a) J. M. Fraile, J. I. GarcÌa, J. A. Mayoral, T. Tarnai, Tetrahedron:
Asymmetry 1997, 8, 2089; b) J. M. Fraile, J. I. GarcÌa, J. A. Mayoral,
T. Tarnai, Tetrahedron: Asymmetry 1998, 9, 3997.
[7] J. M. Fraile, J. I. GarcÌa, J. A. Mayoral, T. Tarnai, M. A. Harmer, J.
Catal. 1999, 186, 214.
[8] For immobilization of bis(oxazoline) copper complexes onto zeo-
lites following a similar procedure see for example: S. Taylor, J. Gul-
lick, P. McMorn, D. Bethell, P. C. Bulman-Page, F. Hancock, F.
King, G. J. Hutchings, J. Chem. Soc. Perkin Trans. 2 2001, 1714, and
references therein.
[9] For a recent review on stereoselective cyclopropanation reactions
see: H. Lebel, J.-F. Marcoux, C. Molinaro, A. B. Charette, Chem.
Rev. 2003, 103, 977.
[10] J. M. Fraile, J. I. GarcÌa, M. A. Harmer, C. I. HerrerÌas, J. A. Mayor-
al, O. Reiser, H. Werner, J. Mater. Chem. 2002, 12, 3290.
[11] M. Glos, O. Reiser, Org. Lett. 2000, 2, 2045.
[12] a) T. Aratani, Pure Appl. Chem. 1985, 57, 1839; b) R. G. Salomon,
J. K. Kochi, J. Am. Chem. Soc. 1973, 95, 300; c) H. Fritschi, U. Leu-
tenegger, A. Pfaltz, Helv. Chim. Acta 1988, 71, 1553.
[21] Gaussian 98 (Revision A.11), M. J. Frisch, G. W. Trucks, H. B. Schle-
gel, 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..
[22] NBO 5.0, E. D. Glendening, J. K. Badenhoop, A. E. Reed, J. E. Car-
penter, J. A. Bohmann, C. M. Morales, F. Weinhold, Theoretical
Chemistry Institute, University of Wisconsin, Madison, 2001.
[23] NWChem, A Computational Chemistry Package for Parallel Com-
puters, Version 4.1, R. J. Harrison, J. A. Nichols, T. P. Straatsma, M.
Dupuis, E. J. Bylaska, G. I. Fann, T. L. Windus, E. Apra, W. de Jong,
S. Hirata, M. T. Hackler, J. Anchell, D. Bernholdt, P. Borowski, T.
Clark, D. Clerc, H. Dachsel, M. Deegan, K. Dyall, D. Elwood, H.
Fruchtl, E. Glendening, M. Gutowski, K. Hirao, A. Hess, J. Jaffe, B.
Johnson, J. Ju, R. Kendall, R. Kobayashi, R. Kutteh, Z. Lin, R. Lit-
tlefield, X. Long, B. Meng, T. Nakajima, J. Nieplocha, S. Niu, M.
Rosing, G. Sandrone, M. Stave, H. Taylor, G. Thomas, ; J. van -
Lenthe, K. Wolinski, A. Wong, Z. Zhang, Pacific Northwest Nation-
al Laboratory, Richland, Washington 99352 90999, USA, 2002.
[24] B. J. Deppmeier, A. J. Driessen, W. J. Hehre, J. A. Johnson, P. E.
Klunzinger, M. Watanabe, J. Yu, Titan 1.0.5, Wavefunction, Inc.,
Irvine, 2000.
[13] W. J. Hehre, L. Radom, P. von R. Schleyer, J. A. Pople, AbInitio
Molecular Orbital Theory, Wiley, New York, 1986.
[14] J. B. Foresman, T. A. Keith, K. B. Wiberg, J. Snoonian, M. J. Frisch,
J. Phys. Chem. 1996, 100, 16098.
[15] a) E. D. Glendening, A. Streitwieser, J. Chem. Phys. 1994, 100, 2900;
b) E. D. Glendening, J. Am. Chem. Soc. 1996, 118, 2473; c) G. K.
Schenter, E. D. Glendening, J. Phys. Chem. 1996, 100, 17152.
[16] J. M. Fraile, J. I. GarcÌa, V. MartÌnez-Merino, J. A. Mayoral, L. Sal-
vatella, J. Am. Chem. Soc. 2001, 123, 7616.
[25] J. P. Hay, W. R. Wadt, J. Chem. Phys. 1985, 82, 299.
Received: November 21, 2003
Published online: April 28, 2004
[17] M. A. Harmer, Q. Sun, M. J. Michalczyk, Z. Yang, Chem. Commun.
1997, 1803.
3005
Chem. Eur. J. 2004, 10, 2997 3005
www.chemeurj.org
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim