Dithiourea ligands in the rhodium-catalyzed hydride-transfer reduction of ketones - A theoretical and experimental approach

Article abstract of DOI:10.1002/1099-0690(200104)2001:8<1589::aid-ejoc1589>3.0.co;2-p

Various dithioureas bearing an aromatic ring on their terminal nitrogen atoms have been synthesized. These have been tested in the asymmetric reduction of ketones as catalyzed by a rhodium complex. The influence of electron-withdrawing and electron-donating substituents on the aromatic rings on the reactivity and the enantiomeric excess (ee) has been assessed. The coordination modes of a model thiourea have been studied by Density Functional Theory (DFT) calculations. The electronic effects have also been analyzed and an interpretation of the variation in the enantiomeric excess, based on a supposed change in the coordination mode, is given.

Full text of DOI:10.1002/1099-0690(200104)2001:8<1589::aid-ejoc1589>3.0.co;2-p

FULL PAPER
Dithiourea Ligands in the Rhodium-Catalyzed Hydride-Transfer Reduction of
Ketones ؊ A Theoretical and Experimental Approach
Maud Bernard,[a] Franc¸oise Delbecq,[b] Fabienne Fache,[a] Philippe Sautet,*[c] and
Marc Lemaire*[a]
Keywords: Asymmetric catalysis / Rhodium / Thioureas / Ketone reduction / Theoretical calculations
Various dithioureas bearing an aromatic ring on their ter-
minal nitrogen atoms have been synthesized. These have
been assessed. The coordination modes of a model thiourea
have been studied by Density Functional Theory (DFT) cal-
been tested in the asymmetric reduction of ketones as cata- culations. The electronic effects have also been analyzed and
lyzed by a rhodium complex. The influence of electron-with-
an interpretation of the variation in the enantiomeric excess,
drawing and electron-donating substituents on the aromatic based on a supposed change in the coordination mode, is
rings on the reactivity and the enantiomeric excess (ee) has
given.
Introduction
aromatic substituents are better than aliphatic ones
(Scheme 1).
The successful use of thioureas as ligands for the asym-
metric metal-catalyzed hydride-transfer reduction of ke-
tones has recently been described.[1,2] To the best of our
knowledge, this family of ligands has seldom been used in
asymmetric catalysis and only little is known about the way
in which these ligands are coordinated to the metal center.
Thus, Brunner et al.[3] have reported the synthesis of a
chiral rhodium complex with thioamide ligands as well as
its characterization by means of an X-ray structure analysis.
Its analysis revealed a binuclear Rh complex, where the two
metal atoms are bound by the sulfur atoms of the two
thioamide ligands. The other coordination sites are occu-
pied by the cyclooctadiene (COD) ligand. Cauzzi et al.[4]
have described the structure of the complex [(COD)(N,NЈ-
diphenylthiourea)RhCl]. The X-ray structure analysis
showed that the thiourea ligand is also bound to the Rh
center through the sulfur atom. In the course of our studies
on the use of nitrogen ligands in asymmetric catalysis, we
have determined the structure of the catalytic complex in
the rhodium-catalyzed asymmetric hydride-transfer reduc-
tion of ketones with diamine ligands using a dual experi-
mental and theoretical approach.[5,6] We have now synthe-
sized dithioureas and have followed the same strategy to
elucidate the way in which they are coordinated to the rhod-
ium center. We have previously shown[2] that the nature of
the substituents on the nitrogen atom of the dithiourea li-
gand has a considerable influence on the reaction and that
Scheme 1. Enantioselective hydride-transfer reduction of aceto-
phenone using dithiourea ligands
In order to design a more efficient and selective ‘‘tailor-
made’’ catalyst, it is of great importance to know the type
of interaction involved between the metal center and the
thiourea ligands (Scheme 2).
Scheme 2. Various possible binding sites of the thiourea ligands
Both the sulfur and nitrogen atoms could be coordinated
to the metal center, as well as a CϭS double bond or one
of the CϭN double bonds of the mesomeric form. The thio-
urea functional group could be a two- or four-electron-do-
nating ligand. Moreover, when a hydrogen atom is attached
to one of the nitrogen atoms, it can be abstracted, de-
pending on the experimental conditions, which leads to a
type X ligand (one-electron donor). Hydrogen bonds may
also be involved, as is the case with a chlorine atom in
ref.[4a]
In this work, the effects of structural modifications of
both the ligand and the substrate on the enantiomeric ex-
cess (ee) and conversion have been studied. For this pur-
pose, we have synthesized and tested various aromatic thio-
ureas bearing either bulky groups, electron-withdrawing or -
donating groups. Various ketones with electron-with-
drawing or -donating substituents in the para position have
[a]
`
Laboratoire de Catalyse et Synthese Organique, UMR5622,
UCBL1, CPE,
43 boulevard du 11 novembre 1918, 69622 Villeurbanne
Cedex, France
E-mail: Marc.Lemaire@univ-lyon1.fr
Institut de Recherches sur la Catalyse, Groupe de Chimie
[b]
[c]
´
Theorique,
2 avenue Einstein, 69626 Villeurbanne, Cedex, France
´
Ecole Normale Superieure,
´
´
46 allee d’Italie, 69364 Lyon Cedex 07, France
E-mail: sautet@catalyse.univ-lyon1.fr
Eur. J. Org. Chem. 2001, 1589Ϫ1596
WILEY-VCH Verlag GmbH, 69451 Weinheim, 2001
1434Ϫ193X/01/0408Ϫ1589 $ 17.50ϩ.50/0
1589

M. Bernard, F. Delbecq, F. Fache, P. Sautet, M. Lemaire
FULL PAPER
also been used as substrates. This study has been After chromatography on SiO2, the pure thiourea 6 was ob-
complemented by theoretical calculations.
tained in low yield and was kept under an inert gas.
Compounds with electron-donating groups, which in-
crease the electron density of the ligands, were also synthe-
sized. This family of compounds was obtained in very good
isolated yields (78Ϫ87%).
Dithiourea Synthesis
The dithioureas were synthesized by the addition of two
isothiocyanate moieties to one (R,R)-N,NЈ-dimethyl-1,2-di-
phenylethylenediamine unit at room temperature in
CH2Cl2. The products were isolated by filtration after pre-
cipitation in pentane (Scheme 3 and Table 1).
Acetophenone Reduction
Effects of Phenyl Substitution
The various ligands shown in Table 1 were tested in the
reduction of acetophenone under identical reaction condi-
tions. These reaction conditions were essentially chosen for
their high reproducibility with regard to both conversion
and selectivity, even though they are not the optimal ones
in terms of selectivity.
Scheme 3. Dithiourea synthesis
Table 1. Various dithioureas synthesized according to Scheme 3
The results obtained with ligands bearing bulky substitu-
ents in the vicinity of the nitrogen atoms are reported in
Table 2. In both cases, (S)-phenylethanol was preferentially
obtained and the activity and selectivity were quite similar.
The enantiomeric excess is the most salient characteristic
for synthetic applications. Nevertheless, the (R) and (S) iso-
mers are formed through two competing reactions and
therefore the (S)/(R) ratio corresponds to the ratio of the
rate constants and should be more representative of the li-
gand effect. In the present case, the (S)/(R) values were al-
most the same. These results indicate that steric hindrance
around the nitrogen atoms bound to the aromatic rings has
only a slight influence on the reaction course. Thus, we as-
sume that the dithiourea is unlikely to complex the rhodium
ion through these nitrogen atoms.
Table 2. Reduction of acetophenone using dithioureas with bulky
substituents; conditions: see Exp. Sect.
Entry Ligand Conversion [%] ee [%] (configuration) (S)/(R)
1
2
1
2
97
92
63 (S)
66 (S)
4.4
4.9
The results obtained with dithioureas bearing electron-
withdrawing and -donating groups are reported in Table 3
The introduction of bulky substituents on the aromatic and 4, respectively. The Hammett (σ) coefficients are also
ring can be expected to influence the coordination pattern given since the electronic effects of the substituents on the
of the ligand and thereby modify the reactivity of the com- aromatic ring can be classified according to these values.
plex. Therefore, we synthesized thiourea 2, which was isol- Both the catalytic activity and the selectivity can be seen to
ated in its pure form without further purification in good decrease with the introduction of an electron-withdrawing
yield (71%). All attempts to obtain compound 3 led to a group if we compare dithioureas 4, 5, and 6 with dithiourea
mixture of several products that was difficult to separate. 1. The ee values obtained were very low (around 15%) and
Steric factors might explain these observations.
thus the (S)/(R) ratio decreased from 4.4 to 1.2. The activity
Electronic factors may also play an important role in the also decreased. Nevertheless, introduction of a CN group
asymmetric induction. In the case of electron-withdrawing (Table 3, Entry 3) with σ ϭ 0.70 led to a higher conversion
groups, products 4 and 5 were obtained in good to excellent than introduction of a CF3 group (Table 3, Entry 2) with a
isolated yields, but the synthesis of thiourea 6 led to a mix- σ of only 0.54. Hence, the activity does not follow the order
ture of mono- and dithioureas along with degradation given by σ. Several hypotheses may be proposed to explain
products. Compound 6 was found to be light-sensitive and the overall loss of activity and selectivity. Electron-with-
particular care was needed to prevent its decomposition. drawing groups decrease the electron density on the atoms
1590
Eur. J. Org. Chem. 2001, 1589Ϫ1596

Dithiourea Ligands in the Rhodium-Catalyzed Hydride-Transfer Reduction of Ketones
Table 3. Reduction of acetophenone using dithioureas with electron-withdrawing substituents; for conditions see Exp. Sect.
FULL PAPER
Entry
Ligand
Hammett coefficient σХ[a]
Conversion [%]
ee [%] (configuration)
(S)/(R)
1
2
3
4
1
4
5
6
0
97
60
85
42
63 (S)
15 (S)
12 (S)
14 (S)
4.4
1.4
1.2
1.3
0.54
0.70
0.78
[a]
From: J. March, Advanced Organic Chemistry, 4th ed., John Wiley and Sons, New York, 1992, p. 244.
Table 4. Reduction of acetophenone using dithioureas with electron-donating substituents; for conditions see Exp. Sect.
Entry
Ligand
Hammett coefficient σХ[a]
Conversion [%]
ee [%] (configuration)
(S)/(R)
1
2
3
4
5
1
9
0
97
93
96
97
99
63 (S)
70 (S)
71 (S)
75 (S)
65 (S)
4.4
5.7
5.8
6.9
4.7
Ϫ0.08[b]
Ϫ0.28
Ϫ
7
8
10
Ϫ0.63
[a]
[b]
From J. March, Advanced Organic Chemistry, 4th ed., John Wiley and Sons, New York, 1992, p. 244. Ϫ σp ϩ 2 ϫ σo ϭ Ϫ0.28 ϩ 2
ϫ 0.10.
bound to the Rh center and thus they may also decrease
the stability of the metal complex. The ligand could then
be replaced by other nonchiral ligands such as iPrOH.
Moreover, the hydride ion becomes less electronegative,
which decreases its interaction with the ketone (elec-
trophile).
Table 5. Reduction of para-trifluoromethylacetophenone with vari-
ous ligands; reaction time 50 h; for further conditions see Exp. Sect.
Entry Ligand Conversion [%] ee [%] (configuration) (S)/(R)
1
2
3
1
4
8
99
28
35
29 (S)
8 (S)
34 (S)
1.8
1.2
2.0
With electron-donating groups, the reverse effect was ob-
served. The activity remained almost constant compared to
that of ligand 1 and the (S)/(R) ratio increased from 4 to
up to 7 (Table 4, Entry 4). Nevertheless, with the dithiourea
10, which has a higher σ coefficient, virtually no effect was
noticed compared to 1. It is possible that the NMe2 group
attached to the aromatic ring competes with the other N
atoms or with the S atoms of the ligand in complexing the
metal center, which thus modifies the selectivity.
tivity decreased. Steric factors and hydrogen bonding with
the solvent may be invoked to account for this behavior.
Irrespective of the ligand, the reduction of para-trifluo-
roacetophenone occurred with lower selectivities. Neverthe-
less, the best selectivity was obtained with electron-donating
groups attached to the ligand (34% ee, Table 5, Entry 3) as
opposed to 29% ee. These results could be interpreted in
terms of the formation of a donorϪacceptor complex be-
tween the substrate and the ligand. To test this hypothesis,
Effects of Ketone Substitution
In view of the marked influence of the presence of elec- para-methoxyacetophenone was reduced (Table 6). The ac-
tron-withdrawing or -donating groups on the aromatic ring tivity decreased considerably with this substrate, which can
of the ligand, we also studied the influence of such groups be explained in terms of the less favorable hydride attack
on the aromatic ketone substrates. Thus, para-trifluorome- on the less electrophilic carbonyl group. The difference in
thyl- and -methoxyacetophenone were reduced under stand- selectivity was nevertheless significant as it was measured
ard conditions with various dithiourea ligands.
at low conversion. An electron-donating group attached to
An electron-withdrawing group attached to the aromatic the ligand led to a better selectivity, while an electron-with-
ring of the substrate increases the positive charge on the drawing one led to a lower selectivity. This result is incon-
carbon atom of the carbonyl group and thus facilitates the sistent with the formation of an intermediate
hydride transfer. With ligand 1, the experimental data are donorϪacceptor complex. On the contrary, the electron
consistent with this hypothesis since almost complete con- density on the thiourea appears to be a key factor in deter-
version was achieved within 50 h (Table 5, Entry 1) as op- mining both the selectivity and efficiency.
posed to after 70 h in the case of unsubstituted aceto-
In conclusion, the experimental approach to analyzing
phenone (Table 2, Entry 1). With ligands bearing either the enantioselective hydride-transfer catalyzed by (thio-
electron-withdrawing or -donating groups, however, the ac- urea)rhodium complexes gives rise to several hypotheses
Eur. J. Org. Chem. 2001, 1589Ϫ1596
1591

M. Bernard, F. Delbecq, F. Fache, P. Sautet, M. Lemaire
FULL PAPER
Tble 6. Reduction of para-methoxyacetophenone with various di- corrected functional[10] for correlation. For the Rh atom,
thioureas; reaction time 17 h
we used the relativistic effective core potential of Hay and
Wadt with the corresponding double ζ basis set.[11]
A
Entry
Ligand
Conversion [%]
ee [%][a]
Ratio[b]
pseudopotential was also used for the core electrons of the
C, N, and O atoms.[12] The corresponding valence basis set
was of the 4-1G type[13] with a d polarization function on
N (α ϭ 0.80), C (α ϭ 0.75), and O (α ϭ 0.85). For H atoms,
we used the Duning double ζ basis set and added a p polar-
ization function on the hydride (α ϭ 1.00).
1
2
3
1
4
8
6
3
4
54
43
68
3.3
2.2
4.3
[a]
Absolute configuration undetermined, ee measured by HPLC;
[b]
for other conditions see Exp. Sect. Ϫ Ratio of the two isomers.
All the structures were fully optimized using the gradient
technique. The dithioureas used experimentally were too
large to be considered in the theoretical study. Therefore,
we performed the calculations on the monothiourea 11, in
which the second moiety of the dithiourea is replaced by a
methyl group (Scheme 4). 11 was successively substituted
with the electron-withdrawing CN group and the electron-
donating OCH3 group.
concerning the structure/properties relationship. Firstly, the
presence of bulky substituents in the vicinity of the nitrogen
atom bound to the aromatic part of the thiourea has been
shown to have little or no effect on either the selectivity or
the reaction rate. Therefore, this nitrogen atom is unlikely to
be a binding site in the metal complex. Secondly, electron-
withdrawing groups attached to the aromatic substituents
have been shown to have a dramatic negative effect on both
the selectivity and efficiency of the catalytic system. On the
contrary, an electron-donating group has been found to
have a marked positive effect on the enantioselectivity of
the hydride transfer. These results are broadly consistent
with the Hammett coefficients, since substituents with a
positive σ value give a lower selectivity, while those with a
negative σ value give a higher selectivity. Nevertheless,
within each group of σ coefficients (positive or negative),
no relationship exists between the relative order and the ee
values (compare Entries 3 and 5 in Table 4). Studies on the
effects of substituents on the aromatic part of the substrate
clearly show that these results cannot be attributed to
donorϪacceptor interactions between the substrate and the
ligand. Instead, they are probably due to modification of
the electron density around the binding site of the ligand.
Clearly, thioureas represent potential new ligands for asym-
metric catalysis, but more information concerning the coor-
dination sites will be necessary in order to design better
ligands. Unfortunately, such information cannot be ob-
tained by direct observation because of the lack of X-ray
structures of the catalytic species. Consequently, we under-
took a theoretical study in the hope of gaining insight into
the problems that arose in the experimental study.
Scheme 4. The thioureas under study
In a previous study, we showed that the supposed hydride
complex intermediate of the carbonyl reduction retains one
COD ligand, which we modelled with two ethylene molec-
ules.[5] Thus, we studied the coordination modes of a thio-
urea towards the 16eϪ fragment RhH(NH3)(C2H4)2, where
the coordination sphere is completed by an NH3 ligand. To
save computational time, we still simplified the thiourea and
replaced the phenyl ring by a methyl group. Thus,
NHCH3ϪCSϪN(CH3)2 (12) (Scheme 4) was used as a
model to study the coordination modes.
Thiourea Coordination Modes
Three possibilities exist for the coordination of a thiourea
ligand to a metal ion, i.e. through one nitrogen atom (13),
through the sulfur atom (14), or through the πCS system
(15). These are shown in Scheme 5 in the case of thiourea
12. For 13 or 14, the main interaction is the donation of
the lone pair of N or S into a vacant orbital of the metal
ion. In the case of 15, the main interaction is a back-dona-
tion from the metal ion to the π*CS orbital, the πCS orbital
being too low in energy to significantly interact with the
metal ion. The CϭS bond gives rise to the same types of
coordination structures as the CϭO bond of aldehydes or
ketones (η1 and η2). Nevertheless, with S being less electro-
negative than O, the main orbitals, π, π*, and the lone pairs
are higher in energy and hence the relative stability order
of the η1 and η2 forms can be different for thioureas.
Theoretical Study
We first compared the possible coordination modes of
the thiourea functional group towards the rhodium center.
Thereafter, we studied the electronic effects of substituents
on the aromatic ring in relation to the previous experi-
mental results.
Computational Details
The calculations were based on Density Functional The-
ory (DFT) at the generalized gradient approximation
(GGA) level. They were performed with the Gaussian 94
program.[8] We used Becke’s 1988 gradient-corrected func-
Scheme 5. Various coordination modes of thiourea 12 towards rho-
tional[9] for exchange and PerdewϪWang’s 1991 gradient- dium
1592
Eur. J. Org. Chem. 2001, 1589Ϫ1596

Dithiourea Ligands in the Rhodium-Catalyzed Hydride-Transfer Reduction of Ketones
FULL PAPER
The geometry optimization of 13 did not give a stable
structure: The thiourea ligand did not remain bound to the
metal center and the square-planar 16eϪ ML4 complex
RhH(NH3)(C2H4)2 was obtained. This complex has been
described previously.[14] The geometry optimizations of 14
and 15 gave the structures shown in Figure 1. Both are tri-
gonal-bipyramidal complexes with an axial hydride, like
those studied previously.[5] In 14, the thiourea ligand is
bound to Rh through the lone pair of the sulfur atom. The
[4]
˚
RhϪS bond (2.53 A) is longer than that quoted in ref.
˚
(2.40 A). This may be due to the higher Rh coordination in
14 (ML5 complex vs. ML4 for the experimental complex),
which leads to a weaker bond. The CϭS double bond is
slightly elongated compared to that in the free thiourea
˚
(1.73 vs. 1.69 A), in agreement with the X-ray data. The Cϭ
S bond lies almost in the basal plane (C1SRhH dihedral
angle ϭ 66°). A small interaction with the π-CS system can
thus occur, which lengthens the CϪS bond. The thiourea
moiety remains planar (N1C1SH ϭ 5° and N2C1SH ϭ
Ϫ176°). Another structure, starting with the CϭS bond
parallel to RhH, was also optimized. The CϭS bond
formed an angle of 30° with the RhH bond and the RhS
˚
bond became longer (2.60 A). This second structure was
less stable than 14 by 5.8 kcal/mol.
In 15, the thiourea ligand is bound to Rh through both
its S and C atoms, in an η2 coordination mode similar to
that seen for olefins. As in 14, the RhϪS and RhϪC1 bonds
˚
are rather long (2.5 A). The CϪS bond is more elongated
˚
than that in 14 (1.76 A) because of the back-donation into
the vacant π*CS orbital. The amine substituents NH(CH3)
and N(CH3)2 are bent away from the rhodium center (N1
by 8.3° and N2 by 10.1°). As regards the relative stabilities,
14 is more stable than 15 by 9.0 kcal/mol, which indicates
that thiourea 12 prefers the η1 coordination mode over the
η2 mode, in agreement with the complex found experiment-
ally.[4] The adoption of this coordination mode can be ra-
tionalized by considering at the orbitals of thiourea 12, the
geometry of which was also optimized. The main compon-
ents and energies of these orbitals are shown in Table 7. The
Figure 1. Optimized geometries of rhodium complexes 14 and 15;
˚
lengths in A, angles in °
HOMO of 12 is a sulfur lone pair; thus, with the πCS orbital fer η2 coordination. As mentioned above, the interacting
being much lower in energy and the LUMO (π*CS) being orbitals are higher in energy for CϭS than for CϭO. This
rather high, the best coordination involves this lone pair, favors the interaction of the lone pairs on S with the metal
thereby giving 14. This is in contrast to ketones, which pre- ion, giving the η1 form, and simultaneously decreases the
Table 7. Most important orbitals of thioureas 11aϪc and 12; main components and energies in eV; differences compared with the orbitals
of 11a are given in parentheses
Orbital
Main components
Energy
11a
11b
11c
12
1
π*CS, pN1, pN2, π*Φ
π*Φ
Ϫ0.65
Ϫ1.27
Ϫ1.54
Ϫ4.52
Ϫ5.12
Ϫ5.41
Ϫ6.25
Ϫ6.89
Ϫ1.48 (Ϫ0.83)
Ϫ1.89
Ϫ0.46 (ϩ0.19)
Ϫ1.22
2
3 LUMO
4 HOMO
π*CS, pN2, π*Φ
p1S, ε pN1
Ϫ2.42 (Ϫ0.88)
Ϫ4.97 (Ϫ0.45)
Ϫ5.61 (Ϫ0.49)
Ϫ5.88 (Ϫ0.47)
Ϫ6.93
Ϫ1.36 (ϩ0.18)
Ϫ4.39 (ϩ0.13)
Ϫ4.93 (ϩ0.19)
Ϫ5.07 (ϩ0.34)
Ϫ6.22
Ϫ1.06
Ϫ4.36
Ϫ5.01
Ϫ5.73
5
6
7
8
p2S, pN1, pN2, πCS
pN1, pN2
πΦ
πCS, pN1, pN2
Ϫ7.27
Ϫ6.33
Ϫ7.99
Eur. J. Org. Chem. 2001, 1589Ϫ1596
1593

M. Bernard, F. Delbecq, F. Fache, P. Sautet, M. Lemaire
FULL PAPER
back-donation into π*CS, which reduces the propensity for
η2 coordination.
The nitrogen lone pairs appear mainly in orbital 6, which
is relatively low in energy. Consequently, the interaction of
the nitrogen atoms with the metal ion is not sufficiently
strong to ensure the coordination of the thiourea ligand.
For comparison, the HOMO of the N,NЈ-dimethyl-1,2-di-
methylethylenediamine ligand [NH(CH3)CH(CH3)]2, which
coordinates to Rh,[15] lies at Ϫ4.42 eV and corresponds to
an out-of-phase combination of the nitrogen lone pairs. The
effect of the electron-withdrawing CϭS double bond in 12
is to lower the energy of the lone pairs on N, thereby pre-
venting the coordination of the thiourea ligand through its
N atoms.
Substituent Effects on 11
In the preceding section, we have shown that the thiourea
coordination mode depends on the shape and the relative
position of the interacting orbitals. We will now focus on
how the orbitals of thiourea 11 are modified when substitu-
ents are attached to the phenyl ring. The optimized struc-
tures obtained for 11a؊c are given in Figure 2. The geo-
metry does not vary greatly when the substituent is
changed. Some points can be noted: The CϪS bond length
˚
(ca. 1.70 A) is in agreement with the experimental X-ray
results obtained for
a
phenyldithiourea.[4] The
phenylϪN1ϪC1 moiety is almost planar, with the dihedral
angle C3ϪC2ϪNϪC1 measuring Ϫ171.4, Ϫ179.6, and
Ϫ163.1° in 11a, 11b, and 11c, respectively. The four atoms
C1, S, N1, N2 are also coplanar (dihedral angle of 177.5° in
11a and 11b and of 177.6° in 11c) and these two planes
form an angle of 47Ϫ51° (the dihedral angle SϪC1ϪN1ϪC2
varies from Ϫ133.5 to Ϫ128.8°); this is in agreement with
the value found experimentally.[4] Atom N2 is less pyram-
idal than N1, with a C1ϪN2ϪC7ϪC8 dihedral angle of
164Ϫ166° as opposed to a C1ϪN1ϪC2ϪH dihedral angle
of 142Ϫ146°. The lone pair on each nitrogen atom is almost
parallel to the πCS system and that on N1 is also almost
parallel to the πphenyl system. The CN and OCH3 substitu-
ents lie in the phenyl plane [the dihedral angle
H3CϪOϪCϪC is 179.4°]. This allows conjugation of the
πCN system and the oxygen lone pair with the phenyl π
system. As a result, the thioureas 11a؊c are almost fully
conjugated and the orbitals are delocalized over the whole
molecule. Nevertheless, in each orbital one component pre-
vails over the others.
The shapes and energies of the orbitals vary when the
substituents are changed. This is shown in Table 7, where
the most important occupied and unoccupied orbitals are
considered. In the same table, the orbitals of thiourea 12
are also given. For each orbital, the main component is in
italics. The orbitals of the four thioureas are of the same
shape. Compared to the orbitals of 12, those of 11a are
˚
Figure 2. Optimized geometries of thioureas 11aϪc; lengths in A,
lowered in energy by the presence of the phenyl ring. Never- angles in °
1594
Eur. J. Org. Chem. 2001, 1589Ϫ1596

Dithiourea Ligands in the Rhodium-Catalyzed Hydride-Transfer Reduction of Ketones
FULL PAPER
theless, π*CS is still high and thus the coordination is most crease seen in the enantioselectivity. The influence of a
likely to take place through the S atom as seen for 12.
Compared to the orbitals of 11a, those of 11b are all
donor group is much smaller.
The actual ligand is a dithiourea, possessing two sulfur
stabilized by the influence of the electron-withdrawing CN atoms. Binding of this ligand to the metal ion through both
group. The most stabilized is the π*CS orbital, which, be- sulfur atoms in an η1 geometry would require the formation
sides this effect, is further stabilized by an in-phase mixing of a nine-membered ring, which is far from favorable. Much
with one phenyl π* orbital containing a contribution from stronger interaction could be expected with ligands al-
π*CN. Thus, the interaction of the lone pair on S will be lowing the formation of smaller rings, for example with a
weaker and, conversely, the back-donation into π*CS will dithiourea in which the nitrogen atoms are directly linked
become more important. Therefore, taking into account the to one another.
small energy difference between the model complexes 14
The results presented here are qualitative but give a good
and 15, the coordination of 11b to Rh can change from η1 interpretation of the experimental results, especially the
to η2. This may explain the dramatic decrease in ee ob- variation in the enantiomeric excess. They will be completed
served when the phenyl ring bears electron-withdrawing in the future by calculations on more realistic ligands, in
substituents.
particular on dithioureas.
On the contrary, all the orbitals of 11c are destabilized by
the electron-donating OCH3 group. Orbital 6 is even more
destabilized by an out-of-phase mixing with a phenyl π or-
bital containing a contribution from an oxygen lone pair.
However, the destabilizing effect of OCH3 is smaller than
the stabilizing effect of CN, and the overall effect is that the
orbitals of 11c are not very different from those of 11a.
Hence, the coordination mode must be the same, which ac-
counts for the fact that the reactivity of 11c is not very
different from that of 11a.
Thus, the electron-withdrawing CN group has a larger
effect than the electron-donating OCH3 group. This is
partly due to the fact that the conjugation of the entire mo-
lecule is greater for CN than it is for OCH3, as evidenced by
the values of the relevant C3ϪC2ϪN1ϪC1 dihedral angles
(Ϫ179.6 and Ϫ163.1°, respectively), which allows for better
electronic communication.
Experimental Section
General: All commercially available products were used without
further purification. Conversions and enantiomeric excesses (ee
values) were monitored by gas chromatography using a CYDEX B
1
chiral column. Ϫ H and 13C NMR spectra were recorded with a
Bruker-AM200 FT spectrometer with CDCl3 as the solvent. Chem-
ical shifts are reported in parts per million (ppm) and coupling
constants are reported in Hertz (Hz). Ϫ All reactions were per-
formed in screw-top V-Vials (Aldrich Z11,515Ϫ0). Ϫ (1R,2R)-(ϩ)-
N,NЈ-Dimethyl-1,2-diphenyl-1,2-ethanediamine was synthesized
according to the previously described method of Mangeney et al.[7]
Ligand 2: Isolated yield 71%; white powder, m.p. 265 °C; [α]2D5
ϭ
1
Ϫ380 (c ϭ 3.3, CHCl3). Ϫ H NMR: δ ϭ 2.22 (s, 6 H, Ar-CH3),
2.31 (s, 6 H, Ar-CH3), 3.20 (s, 6 H, CH3N), 7.13Ϫ7.31 (m, 12 H,
Ar-H), 7.62Ϫ7.66 (m, 4 H, Ar-H), 8.13 (s, 2 H, NH). Ϫ 13C NMR:
δ ϭ 18.5 (CH3), 18.6 (CH3), 33.5 (CH3N), 60.9 (CH), 127.9Ϫ128.7
(C-Ar), 136.2, 136.4, 137.1, 137.2 (4Cq), 182.6 (CS). Ϫ IR: ν˜ ϭ
3373, 2947, 2919, 1498, 1479, 1453, 1333, 1291, 1228, 1095, 778,
700 cmϪ1. Ϫ C34H38N4S2 (566.82): calcd. C 72.05, H 6.76, N 9.89,
S 11.29; found C 72.17, H 7.04, N 9.84, S 11.15.
Conclusion
Both the experimental and theoretical studies described
in this article indicate that the coordination of the dithio-
ureas takes place through the S atom. As a consequence,
thiourea should be considered as an S ligand rather than as
an N ligand.
Ligand 3: Isolated yield 43%; powder, m.p. 125 °C; [α]2D5 ϭ Ϫ455
(c ϭ 3.1, CHCl3). Ϫ 1H NMR: δ ϭ 3.19 (s, 6 H, CH3N), 7.25Ϫ7.65
(m, 20 H, Ar-H ϩ NCHPh), 7.76 (s, 2 H, NH). Ϫ 13C NMR: δ ϭ
34.7 (CH3N), 61.4 (CH), 124.7Ϫ125.7 (C-Ar), 125.8Ϫ127.5 (CF3),
128.3Ϫ130.1 (C-Ar), 135.2, 136.5 (2Cq), 142.7 (CNH), 182.3 (CS).
Ϫ IR: ν˜ ϭ 3423, 3300, 3062, 3031, 2934, 1615, 1526, 1455, 1323,
1231, 1166, 1121, 1067, 838, 700 cmϪ1. Ϫ C32H28F6N4S2 (646.71):
calcd. C 59.43, H 4.37, F 17.64, N 8.67, S 9.90; found C 57.75, H
4.40, F 16.05, N 7.86, S 7.52.
Two coordination modes are possible: η1 through the
lone pair on S and η2 through the CϭS π bond. In the case
of the simplest thiourea, the interaction through the lone
pair on S is 9 kcal/mol more stable since the π*CS orbital is
rather high in energy and cannot interact efficiently. Elec-
tron-withdrawing or -donating substituents attached to the
phenyl ring considerably modify the energy levels of the
lone pairs and of the π*CS orbital, despite the distance be-
tween these substituents and the CϭS bond. The conjuga-
tion of the entire molecule allows the acceptor or donor
effect to be propagated over this rather long distance.
Since the energy difference between the η1 and η2 coor-
dination modes is small, it is most probable that the consid-
erable stabilization of the π*CS orbital by electron-with-
drawing substituents will result in a change of the coordina-
tion mode of the molecule in favor of the η2 mode. This
Ligand 4: Isolated yield 96%; powder, m.p. 157 °C; [α]2D5 ϭ Ϫ655
(c ϭ 3.2, CHCl3). Ϫ 1H NMR: δ ϭ 3.22 (s, 6 H, CH3N), 7.21Ϫ7.44
(m, 8 H, Ar-H), 7.53Ϫ7.69 (m, 10 H, Ar-H), 7.76 (s, 2 H, NH),
7.77 (s, 2 H, NCHPh). Ϫ 13C NMR: δ ϭ 34.8 (CH3N), 60.9 (CH),
107.4 (CCN), 118.9 (CN), 124.1Ϫ132.6 (C-Ar), 136.4 (Cq), 144.1
(CNH), 181.8 (CS). Ϫ IR: ν˜ ϭ 3304, 3029, 2927, 2227, 1604, 1515,
1476, 1453, 1378, 1317, 1231, 1177, 1073, 834, 700 cmϪ1. Ϫ
C32H28N6S2 (560.73): calcd. C 68.55, H 5.04, N 15.00, S 11.41;
found C 68.96, H 5.30, N 14.18, S 8.68.
Ligand 5: Isolated yield 11%; powder, m.p. 192 °C; [α]2D5 ϭ Ϫ744
1
(c ϭ 3, CHCl3). Ϫ H NMR: δ ϭ 3.22 (s, 6 H, CH3N), 7.19Ϫ7.32
significant structural change could explain the strong de- (m, 6 H, Ar-H), 7.35Ϫ7.63 (m, 10 H, Ar-H ϩ NCHPh), 7.74 (s, 2
Eur. J. Org. Chem. 2001, 1589Ϫ1596 1595

M. Bernard, F. Delbecq, F. Fache, P. Sautet, M. Lemaire
FULL PAPER
H, NH), 8.12Ϫ8.19 (m, 4 H, Ar-H). Ϫ 13C NMR: δ ϭ 34.9 68.42, H 6.76, N 14.09, S 10.72; found C 67.51, H 6.77, N 13.84,
(CH3N), 61.5 (CH), 121.6Ϫ129.4 (C-Ar), 136.2, 144.2, 145.6 (3Cq), S 10.65.
182.0 (CS). Ϫ IR: ν˜ ϭ 3628, 3341, 3059, 2936, 1596, 1565, 1548,
Typical Procedure for the Reduction of Ketones: [RhCl(COD)]2
1509, 1318, 1236, 1111, 1078, 854, 706 cmϪ1. Ϫ FAB calcd. for
(1.6 mg, 6.36 ϫ 10Ϫ3 mmol Rh) and 3 equiv. of dithiourea were
C30H28N6O4S2 ϩ H (600.71) 601.1691724; found 601.1717000.
mixed in 2 mL of a solution of potassium tert-butoxide in 2-pro-
panol {12 ϫ 10Ϫ3 mmol LϪ1; [tBuOK]/[Rh] ϭ 4}. After stirring for
Ligand 6: Isolated yield 87%; powder, m.p. 128 °C; [α]2D5 ϭ Ϫ480
1.5 h at room temperature under argon, the ketone (0.12 mmol)
was added. The resulting mixture was stirred for a further 15 h
1
(c ϭ 3.2, CHCl3). Ϫ H NMR: δ ϭ 3.08 (s, 6 H, CH3N), 3.74 (s,
6 H, OCH3), 6.79Ϫ6.84 (m, 4 H, Ar-H), 7.11Ϫ7.20 (m, 12 H, Ar-
at room temperature and then heated to 70 °C. The reaction was
H ϩ NCHPh), 7.43Ϫ7.47 (m, 4 H, Ar-H), 7.79 (s, 2 H, NH). Ϫ
monitored by GC.
13C NMR: δ ϭ 34.2 (CH3N), 55.5 (OCH3), 61.2 (CH), 113.9 (C-
Ar), 127.9Ϫ129.2 (C-Ar), 132.6 (Cq), 137.1 (CNH), 157.9 (OCq),
[1]
183.1 (CS). Ϫ IR: ν˜ ϭ 3347, 3029, 2933, 2834, 1610, 1514, 1478,
1464, 1336, 1242, 1179, 1075, 1032, 829, 753, 701 cmϪ1. Ϫ
C32H34N4O2S2 (570.77): calcd. C 67.34, H 6.00, N 9.82, S 11.23;
found C 66.96, H 5.99, N 9.47, S 10.67.
F. Touchard, P. Gamez, F. Fache, M. Lemaire, Tetrahedron
Lett. 1997, 38, 2275.
F. Touchard, F. Fache, M. Lemaire, Tetrahedron: Asymmetry
1997, 8, 3319.
H. Brunner, J. Bügler, B. Nuber, Tetrahedron: Asymmetry 1996,
[2]
[3]
7, 3095.
Ligand 7: Isolated yield 80%; powder, m.p. 130 °C; [α]2D5 ϭ Ϫ343
(c ϭ 3, CHCl3). Ϫ H NMR: δ ϭ 3.16 (s, 6 H, CH3N), 3.72 (s, 6
[4] [4a]
D. Cauzzi, M. Lanfranchi, G. Marzolini, G. Predieri, A.
Tiripicchio, M. Costa, R. Zanoni, J. Organomet. Chem. 1995,
1
[4b]
H, OCH3), 3.80 (s, 6 H, OCH3), 7.02Ϫ7.65 (m, 18 H, Ar-H ϩ
NCHPh), 7.98 (s, 2 H, NH). Ϫ 13C NMR: δ ϭ 33.7 (CH3N), 55.5
(OCH3), 55.6 (OCH3), 60.6 (CH), 122.1 (Cq) 126.7Ϫ129.5 (C-Ar),
137.2 (Cq), 152.5 (OCq), 158.1 (OCq), 182.2 (CS). Ϫ IR: ν˜ ϭ 3393,
3345, 3036, 2958, 2939, 2835, 2120, 2048, 1614, 1517, 1454, 1342,
1313, 1283, 1208, 1158, 1034, 830, 701 cmϪ1. Ϫ C34H38N4O4S2
(630.82): calcd. C 64.74, H 6.07, N 8.88, S 10.16; found C 63.42,
H 5.95, N 8.60, S 10.03.
488, 115. Ϫ
D. Cauzzi, M. Costa, L. Gonsalvi, M. A. Pel-
linghelli, G. Predieri, A. Tiripicchio, R. Zanoni, J. Organomet.
Chem. 1997, 541, 377.
[5]
[6]
[7]
[8]
M. Bernard, V. Guiral, F. Delbecq, F. Fache, P. Sautet, M.
Lemaire, J. Am. Chem. Soc. 1998, 120, 1441.
C. Saluzzo, R. ter Halle, F. Touchard, F. Fache, E. Schulz, M.
Lemaire, J. Organomet. Chem. 2000, 603, 30.
P. Mangeney, T. Tajero, F. Grosjean, J. F. Normant, Synthesis
1988, 255.
M. J. Frisch, G. W. Trucks, H. B. Schlegel, P. M. W. Gill, B. G.
Johnson, M. A. Robb, J. R. Cheeseman, T. Keith, G. A.
Petersson, J. A. Montgomery, K. Raghavachari, M. A. Al-La-
ham, V. G. Zakrzewski, J. V. Ortiz, J. B. Foresman, J. Cioslow-
ski, B. B. Stefanov, A. Nanayakkara, M. Challacombe, C. Y.
Peng, P. Y. Ayala, W. Chen, M. W. Wong, J. L. Andres, E. S.
Replogle, R. Gomperts, R. L. Martin, D. J. Fox, J. S. Binkley,
D. J. Defrees, J. Baker, J. P. Stewart, M. Head-Gordon, C. Gon-
zalez, J. A. Pople, Gaussian 94 (Revision B.1), Gaussian, Inc.,
Pittsburgh PA, 1995.
Ligand 8: Isolated yield 78%; powder, m.p. 220 °C; [α]2D5 ϭ Ϫ490
1
(c ϭ 3.1, CHCl3). Ϫ H NMR: δ ϭ 3.18 (s, 6 H, CH3N), 3.76 (s,
12 H, OCH3), 3.81 (s, 6 H, OCH3), 6.69Ϫ7.53 (m, 16 H, ArH ϩ
NCHPh), 7.79 (s, 2 H, NH). Ϫ 13C NMR: δ ϭ 34.4 (CH3N), 56.2
(2 OCH3), 60.9 (OCH3), 61.1 (CH), 102.7 (C-Ar), 128.0Ϫ129.5 (C-
Ar), 135.2, 135.8, 136.8, 152.9 (4 Cq), 181.8 (CS). Ϫ IR: ν˜ ϭ 3336,
2939, 2839, 1599, 1533, 1509, 1481, 1465, 1433, 1337, 1313, 1233,
1132, 1077, 1006, 700 cmϪ1. Ϫ C36H42N4O6S2 (690.87): calcd. C
62.59, H 6.13, N 8.11, S 9.28; found C 62.56, H 5.88, N 7.99,
S 8.15.
[9]
A. D. Becke, Phys. Rev. A 1988, 38, 3098.
[10]
[11]
[12]
J. P. Perdew, Y. Wang, Phys. Rev. B 1992, 45, 13244.
P. J. Hay, W. R. Wadt, J. Chem. Phys. 1985, 85, 299.
Y. Bouteiller, C. Mijoule, M. Nizam, J.-C. Barthelat, J.-P. Dau-
dey, M. Pelissier, Mol. Phys. 1988, 65, 295.
Ligand 9: Isolated yield 79%; powder, m.p. 143 °C; [α]2D5 ϭ Ϫ433
[13]
[14]
[15]
Y. Bouteiller, C. Mijoule, M. Nizam, J.-C. Barthelat, J.-P. Dau-
dey, M. Pelissier, private communication.
V. Guiral, F. Delbecq, P. Sautet, Organometallics 2000, 19,
1589.
M. Bernard, F. Delbecq, F. Fache, P. Sautet, M. Lemaire, Or-
ganometallics, 2000, 19, 5715.
1
(c ϭ 3.2, CHCl3). Ϫ H NMR: δ ϭ 2.97 [s, 12 H, N(CH3)2], 3.16
(s, 6 H, CSϪCH3N), 6.74Ϫ7.59 (m, 20 H, Ar-H ϩ NCHPh), 7.91
(s, 2 H, NH). Ϫ 13C NMR: δ ϭ 34.1 (CH3NCS), 41.0 [N(CH3)2],
61.2 (CH), 125.6Ϫ129.7 (C-Ar), 137.2 (Cq), 183.2 (CS). Ϫ IR: ν˜ ϭ
3363, 3281, 3030, 2941, 2886, 2799, 1613, 1524, 1334, 1231, 1164,
1074, 944, 817, 735, 700 cmϪ1. Ϫ C34H40N6S2 (596.85): calcd. C
Received September 20, 2000
[O00492]
1596
Eur. J. Org. Chem. 2001, 1589Ϫ1596