New chiral α-aminophosphine oxides and sulfides: An unprecedented rhodium-catalyzed ligand epimerization

Article abstract of DOI:10.1039/b100217l

New chiral α-aminophosphine oxide N,P(O) and sulfide N,P(S) ligands have been prepared in one-pot syntheses by addition of Ph₂PH to (S)-PhCH=NCH(Ph)CH₃, followed by oxidation with O₂ or S₈. Crystallization from

Full text of DOI:10.1039/b100217l

View Article Online / Journal Homepage / Table of Contents for this issue
New chiral a-aminophosphine oxides and sulÐdes: an unprecedented
rhodium-catalyzed ligand epimerization
Jacques Andrieu,* Jean-Michel Camus, Rinaldo Poli and Philippe Richard
L aboratoire de Synthese et dÏElectrosynthese Organometalliques, Faculte des Sciences ““GabrielÏÏ,
Universite de Bourgogne, 6 Boulevard Gabriel, 21000 Dijon, France.
E-mail: Jacques.Andrieu=u-bourgogne.fr
Received (in Strasbourg, France) 21st December 2000, Accepted 18th May 2001
First published as an Advance Article on the web 18th July 2001
New chiral a-aminophosphine oxide N,P(O) and sulÐde N,P(S) ligands have been prepared in one-pot syntheses by
addition of Ph PH to (S)-PhCH2NCH(Ph)CH , followed by oxidation with O or S . Crystallization from cold
2
3
2
8
methanol leads to the isolation of an enantiomerically pure single N,P(O) diastereomer and to a 1 : 1 mixture of the
two N,P(S) diastereomers. The coordination chemistry of these ligands with [RhCl(COD)] and [RhCl(CO) ] has
2
2 2
been investigated under argon and syngas. At high temperatures, a PÈC oxidative addition of the N,P(O) ligand
followed by imine elimination leads to several hydrido rhodium species. The complexes containing an N,P(S) ligand
undergo the same process at room temperature. Catalytic amounts of the rhodium complexes catalyze the
epimerization of the N,P(S) ligand under argon at room temperature, the dicarbonyl complex being 14 times more
active than the COD complex. The same catalyzed epimerization takes also place for the N,P(O) ligand, but much
more slowly, and only at 35 ¡C and under a syngas pressure. Possible mechanisms for this catalytic process are
discussed. Catalytic tests in styrene hydroformylation have shown high activities and regioselectivities, but no
enantioselectivity, for the N,P(O)-containing rhodium complexes.
The challenge of asymmetric hydroformylation is not only
related to chemoselectivity (hydroformylation vs. hydro-
genation) and regioselectivity (branched vs. normal aldehyde),
but also to enantioselectivity.1,2 The hydroformylation of
vinyl aromatics leads mainly to branched aldehydes, which
can be further oxidized to the corresponding acids. These
compounds are e†ective nonsteroidal analgesics.3,4 Rhodium
complexes containing chiral diphosphine ligands give higher
activity and regioselectivity vs. platinum complexes, with mod-
erate enantiomeric excesses (e.e.). However, by using diphos-
phite and phosphine-phosphite ligands, the enantioselectivity
has been improved to 90% and more than 94%, respec-
tively.5,6 Three reviews on asymmetric hydroformylation have
been published during the last few years.1,2,7 If chiral phos-
phorus ligands seem to be very attractive for this reaction,
most of them are air-sensitive and their preparation is rather
tedious. Recently described chiral thioureas are the Ðrst air-
stable ligands leading to a signiÐcant enantioselectivity (40%
e.e.), albeit with a low conversion of 7%.8 Since there is a
strong demand for air-stable and easily accessible chiral
ligands, we decided to develop new air-stable chiral amino-
phosphine oxide ligands [hereafter abbreviated as N,P(O)]
and related sulÐde ligands [N,P(S)]. In fact, ligands such as A
or B give not only greater catalytic activities in hydro-
formylation than their corresponding aminophosphines in
combination with rhodium, but also higher branched/normal
aldehyde ratios.9h11 It is notable that a set of optically pure
a-aminophosphine oxide ligands (PÈC*ÈN) of type A has
recently been obtained by catalytic enantioselective addition
of diphenylphosphine oxide to cyclic imines.12 To the best of
our knowledge, the use of such chiral a-aminophosphine
oxides and sulÐdes (of type A) in asymmetric hydro-
formylation has not so far been reported. We note, however,
that a good conversion (85%) with a high enantioselectivity
(84%) has recently been reported for the N,P(O) ligand C in
the Ru-catalyzed transfer hydrogenation of phenylisopropyl
ketone.13 In this paper, we present the synthesis of new chiral
a-aminophosphine oxides and sulÐdes with a secondary amine
function, and their use in asymmetric hydroformylation of
oleÐns. Our results show that these ligands are not good
chiral auxiliaries for the enantioselective branched hydro-
formylation and that they epimerize rapidly in the presence of
rhodium.
Results and discussion
1. Synthesis of the aminophosphine oxide and sulÐde ligands
In order to obtain optically pure ligands, we based our syn-
thetic approach on the two simple reactions in Scheme 1. The
Ðrst step, namely the addition of Ph PH to an imine, has
2
already been studied and recently reported by us. The addi-
tion proceeds to an equilibrium position that was found to be
strongly a†ected by the electron richness of the system.14,15
The second step consists of an oxidation reaction with either
oxygen or sulfur. We used an optically active starting imine
Scheme 1
DOI: 10.1039/b100217l
New J. Chem., 2001, 25, 1015È1023
1015
This journal is ( The Royal Society of Chemistry and the Centre National de la Recherche ScientiÐque 2001

View Article Online
compound, namely (S)-PhCH2NCH(Ph)CH , in order to
induce chirality at the central C atom and hopefully lead to a
single diastereomer or to two diastereomers that could be
easily separated by virtue of their di†erent solubility proper-
ties.
the oxidation process is irreversible. This irreversibility is
demonstrated by the stability of the pure (R,S) aminophos-
phine oxide (obtained by crystallization, see below) toward
3
epimerization in CDCl solution. In a preparative scale
3
process carried out in Et O as solvent, the air-stable amino-
2
Before realizing the one-pot preparation of the N,P(O) and
N,P(S) ligands, we wished to examine in more detail the pre-
viously communicated14 equilibrium between the starting
phosphine oxide ligand Ph P(O)CH(Ph)NHCH(Ph)CH , 2,
2
3
forms with a poor d.e. (syn) of 12% (Scheme 2). As diastereo-
mers should exhibit di†erent solubility properties, a separa-
tion by fractional crystallization was attempted. White needles
of the pure syn-2 diastereomer were obtained quantitatively
by a single crystallization from cold methanol. The 1H and
31PM1HN NMR analyses conÐrmed the presence of the single
syn-2 diastereomer. The (R,S) conÐguration of the chiral
carbon atoms is conÐrmed by the X-ray structural analysis
(see below).
imine and Ph PH mixture and the two diastereomers of the
2
intermediate aminophosphine 1 [eqn. (1)]. An NMR study in
CDCl showed that an equilibrium was achieved after 5 h at
3
room temperature.
When the oxidation reaction was carried out in
a CH Cl solution with sulfur instead of oxygen, the 1H and
(1)
2
2
31PM1HN NMR analyses of the crude product showed
complete conversion after one week. The irreversibility of this
transformation is shown, as for the oxide derivative, by the
stability of the crystallized sample toward epimerization
(vide
infra).
The
aminophosphine
sulÐde
ligand
Fig. 1 shows the time evolution of the diastereomers
Ph P(S)CH(Ph)NHCH(Ph)CH , 3, forms with a moderately
2
3
syn-1 (R,S) and anti-1 (S,S) from 1H NMR monitoring
high d.e. (anti) of 70% (Scheme 2). The di†erent d.e. in the
in CDCl . The absolute conÐgurations were determined
oxygen and sulfur oxidations must result from a kinetic
3
by analogy with both diastereomers of CH CH(Ph)-
control during the addition reaction of Ph P(E)H (E \ O or
3
NHCH(Ph)CH , which give signals for the benzylic and
3
2
S) to the imine. Indeed, the 31P NMR spectra of aliquots
methine hydrogens at higher Ðelds for the (S,S) diastereomer
withdrawn during both oxidation reactions never show the
than for the (R,S) one.16 This was explained by assuming a
zig-zag conformation together with anti and syn dispositions
of the aryl groups in the (S,S) and (R,S) diastereomers, respec-
tively.17 Hence, in the anti or (S,S) isomer each benzylic
hydrogen is subjected to the ring current e†ect of the non-
adjacent aryl group. Thus, the 1HNMR spectra (see
Experimental) di†er signiÐcantly, especially for the benzyl and
methine signals.16
It is interesting to note from Fig. 1 that both syn and anti
stereoisomers form at approximately the same rate at the
beginning of the reaction. However, the diastereomeric ratio
changes remarkably after ca. 15 min. Indeed, the anti dia-
stereomer continues to form whereas the amount of the syn
one decreases until an equilibrium is reached, which corre-
sponds to a diastereomeric excess (d.e.) of 62% in favor of the
anti a-P,N ligand. This means that the two imine enantiofaces
are equally accessible to the phosphine attack, but the
reversible PÈC bond formation equilibrium eventually puts
the process under thermodynamic control, to yield an equi-
presence of free Ph PH. Rather, signals corresponding to
2
Ph P(E)H (E \ O or S) are observed.
2
Unfortunately, attempts to separate the two diastereomers
of 3 by crystallization from cold methanol as described for the
analogous oxide ligand led to a 1 : 1 diastereomeric mixture in
29% isolated yield. Thus, the solution is completely depleted
of isomer syn-3 and pure anti-3 should remain in solution.
Isolation of the pure anti-3 ligand from the residual Ðltrate,
however, was prevented by traces of unreacted Ph P(S)H.
2
2. X-Ray structural analyses of ligands 2 and 3
Suitable crystals of syn-2 and syn/anti-3 (1 : 1) for the X-ray
di†raction analyses were obtained under identical conditions
(see Experimental). Nevertheless, quite surprisingly, the unit
cell of the N,P(O) ligand contains only the (R,S) diastereomer,
whereas the N,P(S) ligand leads instead to the co-
crystallization of the (R,S) and (S,S) isomers. The absolute
conÐguration observed for the aminophosphine oxide con-
Ðrms the assignment made earlier on the basis of NMR data.
Neither compound shows any inter- or intramolecular
NHÉ É ÉE bonding interaction. Consequently, both are present
as monomers in the solid state (see Fig. 2). Selected bond
lengths and angles are reported in Table 1. The PÈC(1) bond
lengths in the N,P(S) ligand [1.893(2) and 1.856(2) A for the
librium constant of K
above equilibrium, it is impossible to separate the two dia-
stereomers of 1.
\ 4.25. Clearly, because of the
anti@syn
After achieving equilibrium, the reaction mixture was then
slowly oxidized (by opening the NMR sample to air). The half
time of this oxidation reaction is estimated to be ca. 4 h and
Fig. 1 Kinetic and thermodynamic controls in the formation of
chiral a-P,N (1) ligands.
Scheme 2
1016
New J. Chem., 2001, 25, 1015È1023

View Article Online
indicates that it should be easier to cleave the PÈC(1) bond for
the N,P(S) ligand than for the aminophosphine oxide. More-
over, in contrast to the average PÈC bond length found in 19
other structures containing the Ph P(S)ÈC(sp3)H moiety in
2
the Cambridge Crystallographic Database, which is 1.83(1) A,
the PÈC(1) bond length in the syn-3 diastereomer is abnor-
mally longer. This further indicates that the PÈC bond in this
compound could be easier to cleave. These assumptions will
be conÐrmed in a later section concerning the rhodium-
catalyzed epimerization. It is also notable that this isomer
with the unusually long PÈC bond, therefore presumably
having a higher relative energy, corresponds indeed to the
minor diastereomer at equilibrium according to the NMR
analysis. It can be noted that the PÈC(1)ÈN angle is signiÐ-
cantly di†erent in the two diastereomers of 3 and that no clear
correlation exists between syn-2 and syn-3 for this parameter.
3. Interaction of ligands 2 and 3 with rhodium complexes
Before carrying out catalytic tests, the stoichiometric inter-
actions between the N,P(O) and N,P(S) ligands and the
rhodium precatalysts [RhCl(COD)] and [RhCl(CO) ] were
2
2 2
investigated. Stirring
a
2 : 1 mixture of syn-2 and
[RhCl(COD)] in CDCl solution for 15 min at room tem-
2
3
perature leads to a single product that, according to the NMR
analyses, has an uncoordinated P2O moiety (31P NMR
chemical shift identical to the free ligand), see Scheme 3. The
N-coordination is not clearly shown by the 1H NMR because
of overlap of the NH resonance with those of the coordinated
cyclooctadiene protons. However, this coordination can rea-
sonably be expected considering the ability of secondary
amines to split the chloride bridges in [RhCl(COD)] , as pre-
2
viously proposed (although without spectroscopic evidence)
for analogous a-aminophosphonate ligands.18 The above
solution was then heated to 80 ¡C for 15 min. After evapo-
ration of the solvent, the 1H and 31P NMR spectra in C D
6
6
show at least four di†erent hydrido rhodium complexes (D)
and the presence of free imine (Scheme 3). These products pre-
sumably result from a PÈC oxidative addition followed by
imine elimination. Note that the Rh(III)ÈPPh (O) moiety has
2
recently been identiÐed by X-ray di†raction analysis in dirho-
dium chloride complexes.19 At temperatures as low as 55 ¡C,
similar hydrido rhodium species begin to appear after 4 h in
very minor quantities. The same NMR resonances are gener-
ated in a control experiment involving the addition of
Ph P(O)H to the same rhodium substrate at 80 ¡C in C D .
2
6 6
The addition of compound syn-2 to [RhCl(CO) ] at room
2 2
Fig. 2 ORTEP views of compounds syn-2 and syn/anti-3. The two
diastereomers in the latter view are shown in their correct relative
orientation. Thermal ellipsoids are drawn at the 30% probability level
for syn-2 and 50% for syn/anti-8. For clarity, only relevant hydrogen
atoms are shown.
temperature, on the other hand, leads to a rhodium complex
containing a bidentate N,P(O) ligand, which is stable in solu-
tion for several hours. The bidentate coordination mode is
conÐrmed by a new broad peak of the phosphine oxide
moiety at 40.7 vs. 31.5 ppm for the free ligand and by the l
PJO
absorption band at 1184 vs. 1265 cm~1 for the free ligand. In
syn and anti diastereomers, respectively] are longer than those
found in its oxide counterpart [1.832(2) A]. This can be
directly correlated with the weaker electron-withdrawing e†ect
of the sulfur atom relative to the oxygen atom. This di†erence
Table 1 Selected bond distances (A) and angles (¡) for compounds
syn-2 and syn/anti-3
syn-2 (E \ O)
syn-3 (E \ S)
anti-3 (E \ S)
PÈE
1.4870(13)
1.832(2)
1.462(2)
116.66(8)
104.82(12)
170(1)
1.9642(9)
1.893(3)
1.452(3)
111.17(8)
111.32(16)
157(1)
1.9651(8)
1.856(2)
1.463(3)
112.69(8)
103.90(14)
163(1)
PÈC(1)
NÈC(1)
EÈPÈC(1)
PÈC(1)ÈN
HÈC(1)ÈPÈE
Scheme 3
New J. Chem., 2001, 25, 1015È1023
1017

View Article Online
addition, the NH resonance in the 1H NMR spectrum is also
4. Rhodium-catalyzed hydroformylation of styrene with
shifted to lower Ðeld by coordination. However, two IR
ligands syn-2 and 3
absorption bands in CDCl at 2086 and 2015 cm~1 indicate
3
The results of all catalytic studies are assembled in Table 2.
All runs were carried out at a temperature of 55 or 35 ¡C, and
at a total pressure of 400 or 500 psi for the 1 : 1 COÈH
mixture, in chloroform or in benzene as solvent. A control
catalytic run without ligand (entry 1) shows complete conver-
sion after 20 h with a b : 1 regioselectivity of 89 : 11. By using
the presence of two types of terminal carbonyl ligands. This
observation, combined with the broad signals in the 1H and
31P NMR spectra, suggest a dynamic isomerization exchange,
E in Scheme 3, similar to that reported for the complex
RhCl(CO)[j2-N,P(O)] with N,P(O) \ Ph P(O)CH NMe .9
2
2
2
2
When this reaction was carried out under a CO atmosphere
Eu(hfc) as a chemical shift reagent, no e.e. was observed, as
Mby preparing [RhCl(CO) ] in situ from [RhCl(COD)]
3
2 2
2
expected. In the presence of ligand syn-2 (run 2) the conver-
under CON, the same E complexes are obtained. However, a
new square planar compound also forms, in which the P2O
moiety is uncoordinated due to the hemilabile character of the
ligand under CO.
sion rate is Ðve times higher with the same regioselectivity, but
with zero enantioselectivity. The use of the less polar solvent
C H (run 3) decreases the activity. Unfortunately, the regio-
6
6
selectivity is moderate and the catalyst is still non-
enantioselective, even though benzene is known to be the best
solvent to favor high regioselectivities and to prevent aldehyde
epimerization.5,6 Slightly lower activities and zero enantio-
selectivity are also observed by working at a lower tem-
perature in chloroform (run 4), however the b : 1 ratio is
greatly increased under these conditions.
The stoichiometric reaction between the N,P(S) ligand M1 : 1
mixture of syn and anti diastereomers of 3N with
[RhCl(CO) ] Mor with [RhCl(COD)] under CON at room
2 2
2
temperature either in CDCl or C D led, after 30 min, to a
3
6 6
mixture of four rhodium complexes according to the 31P
NMR spectrum. The two major products are characterized by
31P singlets at 50.42 and 50.25 ppm in CDCl , that is close to
3
The promoter e†ect of ligand syn-2 is further demonstrated
by the following additional experiments. The conversion was
found to be linear with time (zero-order dependence on
substrate) and dependent of the ligand : metal ratio as illus-
trated in Fig. 3 for experiments carried out in chloroform at
35 ¡C. In all cases, the regioselectivity is greater than 95 : 5
and the enantioselectivity is zero. The highest activity corre-
the chemical shifts of the two diastereomers of the free ligand.
Integration shows that the d.e. has surprisingly changed from
0% in the free ligand to 70% in the Rh products. The N-
coordination mode is conÐrmed by the 1H NMR chemical
shift of the NH proton from 3.55 ppm for the free amine to
3.80 ppm in C D solution. Therefore, in contrast to the same
6
6
reaction of the N,P(O) ligand described above, ligand 3 binds
rhodium in a j1-N fashion. This is surprising because the
reaction of the very similar ligand Ph PCH P(S)Ph with
2
2
2
[RhCl(CO) ] was reported to lead to RhCl(CO)[j2-
2 2
Ph PCH P(S)Ph ] in which the 31P NMR resonance for the
P(S)Ph moiety shifts to 52 ppm from the free ligand value of
2
2
2
2
40 ppm.20,21
The other two reaction products exhibit 31P NMR doublets
at 114 (with J \ 179 Hz) and at 71 ppm (with J \ 147
PRh
PRh
Hz), showing the direct interaction of the rhodium center with
phosphorus rather than with sulfur. These signals can likely
be assigned to two types of RhÈP species that we have not
been able to fully characterize, likely formed by a PÈC oxida-
tive addition process. These chemical shifts and the high RhÈP
coupling constants may be consistent with terminal or bridg-
ing (l : P,S) modes for the P(S)Ph ligand, both of which have
2
previously been established.22,23 At this point, we conclude
that rhodium complexes containing the N,P(S) ligand 3 are
unstable in solution, even at room temperature. Finally, since
the complexes containing the N,P(O) ligand syn-2 are unstable
only at high temperatures and since low temperatures are
known to decrease the conversion rate but at the same time
favor higher enantioselectivities in hydroformylation reac-
tions,2,6,24 we decided to examine the catalytic activity of Rh
complexes containing the ligand syn-2 at relatively low tem-
peratures (O55 ¡C).
Fig. 3 Conversion as a function of time for the hydroformylation of
styrene at various syn-2 : Rh ratios: 0.5 : 1 (squares), 1 : 1 (triangles),
2 : 1 (circles), 10 : 1 (inverted triangles). Reaction conditions: 0.025
mmol [RhCl(COD)] , 30 ml of CHCl , T \ 35 ¡C, P(CO : H
2
3
2
1 : 1) \ 500 psi. The conversion and regioselectivity (95 to 97% for all
experiments) were determined by 1H NMR spectroscopy.
Table 2 Hydroformylation of styrene catalyzed by [RhCl(COD)] and aminophosphine oxide (syn-2) or sulÐde (syn/anti-3) ligandsa
2
Entry
Solvent
Ligand
T /¡C
P/psi
% conversionb (time/h)
% b : lb
1
2
3
4
5d
6
7f
8
CHCl
3
È
2
55
55
55
35
35
55
55
35
400
400
400
500
500
400
400
500
4(3 h)
100(20 h)
89 : 11
89 : 11
85 : 15
96 : 4
96 : 4
95 : 5
93 : 7
96 : 4
CHCl
3
88(3 h)
16(3 h)
60(3 h)
35(3 h)
0(3 h)
0(3 h)
3(3 h)
100(4 h)
100(21 h)c
100(6 h)c
100(8 h)e
30(70 h)
30(67 h)
50(21 h)
C H
2
6
6
CHCl
CHCl
CHCl
CHCl
CHCl
2
3
3
3
3
3
2
3
3
Ph P(O)H
2
a Reaction conditions: [RhCl(COD)] (0.025 mmol), solvent (40 ml), CO ] H (1 : 1), styrene : Rh : ligand (167 : 1 : 1). b Determined by 1H NMR
2
2
spectroscopy, no ethylbenzene detected. c The e.e. at complete conversion is 0%. d Vinylnaphthalene used instead of styrene under similar
conditions. e The e.e. at complete conversion is 3%. f [Rh(COD) ](CF SO ) used instead of [RhCl(COD)] under similar conditions.
2
3
3
2
1018
New J. Chem., 2001, 25, 1015È1023

View Article Online
sponds to a ligand : metal ratio of 1 : 1, while it is much lower
for a ratio of 0.5 : 1, slighly lower for 2 : 1, and then it
decreases dramatically when a ratio of 10 : 1 is used. The TOF
for the experiment with 0.5 equiv. of ligand [10.2 mol
mol(Rh)~1 h~1] is approximately one half the TOF found
with 1 equiv. of ligand [21.6 mol mol(Rh)~1 h~1]. A possible
interpretation of these results is that only half the rhodium
forms the complex containing the aminophosphine oxide
ligand and subsequently the activity is decreased by a factor of
two, while higher ratios likely force the coordination of two or
more ligands to the rhodium center via equilibrium processes,
consequently decreasing the number of coordination sites and
leading to less active species.
Since ligand syn-2 promotes the activity of the rhodium
catalyst for the hydroformylation of styrene, other oleÐns have
also been tested under the same conditions. Methyl meth-
acrylate did not yield aldehydes after one day, and a conver-
sion of only 7% was achieved with a-methylstyrene after 70 h.
This catalytic system seems to be best suited for the hydro-
formylation of electron-deÐcient vinyl aromatics. Since, in
some cases, better e.e.s have been obtained with vinyl-
naphthalene instead of styrene,24,25 this oleÐn was also tested,
yielding results comparable to styrene. The activity is slightly
lower, the regioselectivity is about the same (entry 5), and the
e.e. is quite small (3%).
Fig. 4 Rhodium-catalyzed epimerization of [Ph P(S)CH(Ph)-
2
NHCH(Ph)Me]. Reaction conditions: ligand : Rh ratio \ 10 : 1,
C D (0.5 ml) at R.T. Squares: [RhCl(COD)] (0.05 mmol); triangles:
6
6
2
[RhCl(CO) ] (0.05 mmol).
2 2
(see X-ray structure analysis discussion in section 2), the lower
stability of the syn diastereomer may be the consequence of its
longer PÈC(1) bond length.
Under the same conditions, the chlorodicarbonylrhodium(I)
dimer exhibits a higher activity, leading to complete equili-
bration in less than one hour (see Fig. 4). To the best of our
knowledge, there is only one example of ligand epimerization
Mfor the bifunctional S,O ligand (R)-2-[(R)-phenylsulÐnyl]pro-
pionic acidN upon stereoselective and stoichiometric coordi-
nation to hemimetallocenic RhIII, IrIII, RuII or OsII
complexes.27 Here, we report the Ðrst case where such a
process is Rh-catalyzed. By applying the kinetic analysis of
reversible Ðrst-order reactions to our experimental measure-
ments [eqn. (2)],
Use of the N,P(S) ligand 3 with neutral or cationic rhodium
precursors gave dramatically lower activities (see entries 6 and
7 in Table 2), albeit with surprisingly high regioselectivities, by
analogy with the previous studies with the Ph P(E)CH NMe
2
2
2
(E \ S vs. O) ligands.9,10 The rapid decomposition of the Rh
complexes with ligand 3 mentioned in the previous section
may rationalize these results. In other cases, however, cationic
precursors displayed increased activities.8,26
k
t \ ln[%syn /(%syn [ %syn )]
(2)
obs
\ k
eq
t
eq
The last experiment in Table 2 (run 8) shows that the use of
with
the phosphine oxide Ph P(O)H leads to a poorer catalytic
2
performance relative to ligand syn-2 (entry 4). The reaction of
k
] k
and %syn \ 15%,
obs
obs`
obs~
eq
this phosphine oxide with the rhodium precursor, as shown in
the previous section, leads to the same products originating
from the thermal decomposition of the rhodium complex with
ligand syn-2 (form D in Scheme 3). This result further demon-
strates that the coordinated aminophosphine oxide is an effi-
cient ligand for styrene hydroformylation catalysis under
conditions where this decomposition does not occur.
a satisfactory Ðtting is obtained (see Fig. 5), giving a rate con-
stant k of 8.6(2) ] 10~3 (t \ 80 min, R2 \ 0.996) and of
obs
1@2
0.117(4) min~1 (t \ 5.9 min, R2 \ 0.994) for systems using
1@2
[RhCl(COD)] and [RhCl(CO) ] , respectively. This means
2
2 2
that the conversion rate with the latter catalyst is ca. 14 times
faster, which may be related to the greater electrophilic char-
acter of the rhodium center in this catalytic system.
The observed racemic mixture of the branched aldehyde
product could well result from an aldehyde racemization
reaction8 but, considering the relatively short reaction time,
we concluded that this cannot be the only reason to rational-
ize the absence of enantioselectivity. Thus, we wished to inves-
tigate the interaction between the rhodium center and ligands
2 and 3 in more detail.
From the values of the equilibration rate constant (k
obs`
), the
] k
) and the equilibrium constant (k
] /k
obs~
obs`
obs~
individual values of k
and k
may be derived. Experi-
obs`
obs~
ments with di†erent catalyst concentrations have not been
5. Rhodium-catalyzed epimerization of aminophosphine oxide
and sulÐde ligands
We have mentioned above that the aminophosphine sulÐde is
not efficient in hydroformylation. However, an NMR investi-
gation of the interaction between 3 and [RhCl(CO) ] sur-
2 2
prisingly showed that rhodium induces ligand epimerization.
Upon Rh coordination, the anti : syn ratio (as determined by
31P NMR integration) changes from 50 : 50 to 85 : 15 after ca.
30 min. The Ðnal ratio of the two coordinated ligand dia-
stereomers corresponds to the ratio of the two free diastereo-
mers before crystallization (see Scheme 3). Since all signal
intensities do not further change after several hours (including
those of the other by-products, see section 3), all compounds
are considered to be in a thermodynamic equilibrium situ-
ation. We were then stimulated to interact ligand 3 with a
catalytic amount of [RhCl(COD)] . Complete equilibration
2
was achieved after 8 h instead of 30 min (see Fig. 4), to yield
again the same anti : syn ratio of 85 : 15. As mentioned earlier
Fig. 5 Fitting of the kinetic data from Fig. 4 by eqn. (2).
New J. Chem., 2001, 25, 1015È1023
1019

View Article Online
carried out. However, under the reasonable assumption that
both direct and reverse reactions are Ðrst-order in the catalyst
(a catalyst inÑuence on the rate is shown by the shorter equili-
bration time when the rhodium complex was used
used under the hydroformylation conditions, leading to an
expected much faster epimerization. Indeed, a run of syn-
2 : Rh \ 1 : 1 under catalytic conditions (Table 2, entry 4) in
the absence of styrene was stopped after 3 h and the recovered
catalyst, upon treatment with tmeda, gave free 2 with a
syn : anti ratio of 59 : 41 (18% d.e.). It is then reasonable to
propose that the failure to observe any e.e. for this catalytic
hydroformylation may be related to the ligand epimerization
under syngas conditions.
stoichiometrically), namely k
\ k
[Rh], the second-
`@~
order rate constants k and k for the [RhCl(COD)] cata-
obs`@~
~
`
2
lyzed process are 7.31(3) ] 10~2 and 1.29(3) ] 10~2 1 mol~1
min~1, while those for the [RhCl(CO) ] catalyzed process
2 2
are 0.995(6) and 0.175(6) 1 mol~1 min~1, respectively (see
Scheme 4).
In the case of the N,P(O) ligand, no epimerization could be
6. Mechanistic considerations of the rhodium-catalyzed
epimerization
detected by using [RhCl(COD)] , even after 24 h. However, a
2
slow epimerization was observed when the same experiment
was conducted under a carbon monoxide pressure and even
faster under a syngas pressure (see Fig. 6). The increased rate
in the presence of hydrogen suggests that a hydride species
formed in situ catalyzes this epimerization process more effi-
ciently. This state of a†airs is analogous to that described for
As the observed rhodium-catalyzed ligand epimerization
appears to be an unprecedented phenomenon, we have con-
sidered its possible mechanism. Some suggestions may be for-
mulated on the basis of previously established reactivity.
Under appropriate conditions C ÈH, C ÈH or PÈC oxida-
ar
sp3
a
rhodium-catalyzed reversible PÈC bond cleavage of
tive additions involving PPh or other functionalized phos-
3
PPh .28,29 The more difficult epimerization of the N,P(O)
phines take place with Co(I), Rh(I) or Ir(I),28h30 for example.
3
ligand is probably the consequence of the stronger PÈC bond
Two possible catalytic pathways have therefore been con-
sidered, involving reversible CÈH or PÈC bond cleavages as
shown in Schemes 5 and 6, respectively.
(see X-ray structure analysis discussion), resulting from the
stronger electron-withdrawing e†ect of the oxygen atom vs.
the sulfur atom.
For the N,P(O) epimerization, the best activities were
observed under CO at high pressures, where the complex
Furthermore, Fig. 6 shows a faster epimerization of the
oxide ligand 2 when using a greater metal : ligand ratio, as
also observed for the analogous sulÐde ligand 3. Because of
the more stringent epimerization conditions, further studies to
establish the equilibrium position were not carried out. The
d.e. obtained (12%) after 24 h with a 5 : 1 ligand : metal ratio
and under a syngas pressure is identical to the d.e. of the free
ligand before crystallization (see section 1). By comparison
with the results of the sulÐde ligand epimerization, this may
well correspond to a thermodynamic equilibrium.
probably exists under the monodentate form RhCl(CO) [j1-
2
N,P(O)] according to the stoichiometric studies (see section 3).
An equilibrium with the bidentate form, RhCl(CO)[j2-
N,P(O)], however, is established at low CO pressures. For the
N,P(S) epimerization, on the other hand, the monodentate
form RhClL [j1-N,P(S)] [L \ COD or (CO) ] seems
2
2
2
present for both catalysts. It seems likely that the bidentate
coordination is not necessary for this catalytic transformation.
For the CÈH pathway (Scheme 5), a reversible C ÈH oxi-
sp3
It should be noted that the conditions leading to the exten-
sive epimerization of ligand syn-2 (r.t., 24 h) are comparable to
the conditions used for the hydroformylation catalysis (35 ¡C,
3 h). In addition, an equimolar amount of Rh and ligand are
dative addition to the starting complex (R)-F would lead to
the formation of intermediate (R)-G, stabilized by coordi-
nation of the P2E function in a pseudo-allylic moiety. Since
the stereogenic carbon atom is blocked during this process, its
reversal could only lead back to the starting absolute conÐgu-
ration. However, epimerization could occur via an g3/g1
equilibration of the pseudo-allyl ligand to give intermediate H,
for which free rotation around the PÈE single bond is pos-
sible. An alternative possibility is a reversible EÈH reductive
elimination to generate intermediate I, for which an alterna-
Scheme 4
Fig. 6 Rhodium-catalyzed epimerization of syn-2. Reaction condi-
tions: [RhCl(COD)] (0.025 mmol), CHCl (30 ml), T \ 35 ¡C.
2
3
Circles: CO (P \ 400 psi), ligand : Rh ratio \ 10 : 1; triangles:
COÈH (1 : 1, P \ 400 psi), ligand : Rh ratio \ 10 : 1; squares:
2
COÈH (1 : 1, P \ 400 psi), ligand : Rh ratio \ 5 : 1.
Scheme 5
2
1020
New J. Chem., 2001, 25, 1015È1023

View Article Online
The development of enantioselective branched hydro-
formylation catalysts with this class of ligands requires struc-
tural modiÐcations that block or strongly disfavor these
epimerization and deactivation pathways. E†orts in this direc-
tion are currently being pursued in our laboratory. We also
note that all epimerization mechanisms (except via direct tau-
tomerization to intermediate I in Scheme 5) require an oxida-
tive addition process. In the absence of rhodium, both ligands
2 and 3 are conÐgurationally stable. Therefore, we are also
interested in applying these ligands to enantioselective cata-
lytic reactions with metals that cannot undergo oxidative
addition processes.
Experimental
Materials and methods
All reactions were carried out in Schlenk-type Ñasks under
argon atmosphere. Solvents were puriÐed and dried under
dinitrogen by conventional methods. The 1H, 31PM1HN and
13CM1HN NMR spectra were recorded at 200.13, 81.03 and
51.32 MHz, respectively, on a Bruker AC 200 instrument in
CDCl solution. All chemical shifts are given in ppm. The
3
infrared spectra were recorded in CH Cl or in CHCl solu-
2
2
3
tion with an IFS 66 PerkinÈElmer instrument. The a rotation
polarization angle was measured on a PerkinÈElmer 141 Pol-
arimeter. The elemental analyses were carried out on a Fisons
EA 1108 apparatus. The imine compound was prepared by
the slow addition of a stoichiometric amount of distilled benz-
aldehyde to
a dichloromethane solution of (S)-methyl-
Scheme 6
benzylamine at 0 ¡C, according to the literature.31 The water
formed was removed under vacuum at 120 ¡C for 3 h. The
tive N-bonded isomeric form is also possible. Reversal of all
steps leads to (S)-F. It is interesting to remark that the N-
bonded isomer of I is a tautomeric form of the starting com-
pound (R)-F. Therefore, transformation of (R)-F to (S)-F could
also be achieved via this intermediate I without the need for
the hydrogen atom to transit via the Rh center.
For the PÈC pathway (Scheme 6), a reversible PÈC oxida-
tive addition from (R)-F leads to the 16-electron intermediate
(R)-J, which may be capable of deinserting the imine to lead
to (R)-K. Inversion of conÐguration may be achieved at this
point by g2/g1 imine rearrangement through L, without the
need to decoordinate the imine. On the other hand, we have
shown that the imine is expelled from the N,P(O) complex at
high temperatures, with formation of a mixture of hydride
compounds (S)-methylbenzylamine, sulfur, [RhCl(COD)]
2
(COD \ 1,5-cyclooctadiene) and [RhCl(CO) ] were obtained
2 2
from commercial sources and used as received, while Ph PH
2
was prepared by the conventional method.32
Syntheses
Formation of (R,S)- and (S,S)-Ph PCH(Ph)NHCH(Ph)-
2
CH , 1. To a solution of (S)-PhCH2NCH(Ph)CH (0.060 g,
3
3
0.287 mmol) in CDCl (1 ml) was added Ph PH (50 ll, 0.287
3
2
mmol). The mixture was stirred for 12 h, yielding an equi-
librium of 1 with the starting materials. 1: 1H NMR (CDCl ):
3
d 7.85È6.87 (m, aromatics) and 2.00 (s, br, NH); for (R,S)-1:
4.70 [d, PCH, 2J(PH) \ 4.9 Hz], 3.74 [q, NCH, 2J(HH) \ 6.4
products (see section 3). As already stated above,
a
Hz], 1.36 [d, CH , 2J(HH) \ 6.4 Hz]; for (S,S)-1: 4.16 [d,
3
Rh(III)ÈPPh (O) fragment in dirhodium chloride complexes
PCH, 2J(PH) \ 5.4 Hz], 3.61 [q, NCH, 2J(HH) \ 6.4 Hz],
2
has recently been proved by X-ray di†raction analysis.19 This
1.28 [d, CH , 2J(HH) \ 6.4 Hz]. d.e. (anti/syn) \ 62%.
3
experimental observation is the only tenuous criterion that
31PM1HN NMR (CDCl ): d 1.49 (s) for (R,S)-1 and 0.89 (s) for
3
currently makes us favor the PÈC pathway.
(S,S)-1.
Air oxidation of (R,S)- and (S,S)-1. NMR monitoring. To a
solution of (S)-PhCH2NCH(Ph)CH (0.060 g, 0.287 mmol) in
CDCl (1 ml) was added Ph PH (50 ll, 0.287 mmol). After the
Conclusions and perspectives
3
3
2
We have described what appears to be the Ðrst example of an
interesting rhodium-catalyzed epimerization process. The
aminophosphine oxide ligand 2 is efficient and regioselective
in the rhodium-catalyzed hydroformylation of styrene, but is
not an asymmetry-inducing ligand. The epimerization of the
chiral center between the nitrogen and phosphorus atoms
under the catalytic hydroformylation conditions may be
related to this lack of enantioselective induction. The epimer-
ization occurs by a mechanism that appears to involve a
reversible PÈC bond cleavage. This observation shows that
this type of N,P(O) ligand, albeit sufficiently active and regio-
selective, cannot be used in asymmetric hydroformylation with
rhodium catalysts. Furthermore, the corresponding N,P(S)
ligand 3 generates a less efficient catalyst, because of an easier
PÈC oxidative addition and catalyst deactivation.
equilibrium of 1 with the starting materials was attained (3 h),
an air stream was directly introduced into the CDCl solution,
3
which was then stirred for 20 h. The 31PM1HN NMR spectrum
showed that the oxidation reaction was nearly complete,
yielding both diastereomers of 2 with a d.e. (syn/anti) of 12%.
The NMR data of syn-2 are given in the next section. anti-2:
1H NMR (CDCl ): d 7.45È6.88 (m, aromatics), 4.10 [d, PCH,
3
2J(PH) \ 9.9 Hz], 3.61 [q, NCH, 2J(HH) \ 6.9 Hz], 2.57 (s, b,
NH, exchange with D O), 1.24 [d, CH , 2J(HH) \ 6.9 Hz].
2
3
3
31PM1HN NMR (CDCl ): d 32.85 (s).
Synthesis of (R,S)-Ph P(O)CH(Ph)NHCH(Ph)CH , 2. To
2
3
a solution of (S)-PhCH2NCH(Ph)CH (0.940 g, 4.50 mmol) in
3
Et O (15 ml) was added Ph PH (0.8 ml, 4.50 mmol). After a
few seconds a white precipitate was formed. The suspension
2
2
New J. Chem., 2001, 25, 1015È1023
1021

View Article Online
was stirred for 5 h under an inert atmosphere. An air Ñow was
br, NH, exchange with D O), 1.45 [d, CH , 2J(HH) \ 6.7
2
3
then introduced into the Et O solution for 10 min in order to
Hz]. 31PM1HN NMR (C D ): d 50.68 (s), (CDCl ): 50.42 (s).
2
6
6
3
displace all inert gas, and stirring was continued for 20 h in air
13CM1HN NMR (CDCl ): d 145.01È127.06 (m, aromatics),
3
to complete the oxidation. The white powder was collected by
Ðltration (1.51 g), dried under vacuum and dissolved in hot
methanol (50 ml). After several days at [30 ¡C, white needles
were obtained and identiÐed as the pure (R,S) isomer of 2
(0.966 g, 56%). Single crystals of syn-2 were obtained by slow
di†usion of an EtOH solution of 2 onto the same volume of
60.37 [d, PCH, 1J(PC) \ 58.35 Hz], 55.21 [d, NCH,
3J(PC) \ 10.11 Hz], 22.84 (s, CH ). IR (CHCl ) l
\ 1103
3
3
P/S
cm~1. mp \ 115 ¡C. Elem. anal. calc. for C
H
NSP (MW
27 26
427.15): C 75.85, H 6.13, N 3.28, S 7.48; found: C 75.70, H
6.25, N 3.32, S 7.47%.
Catalytic runs
water. syn-2: 1H NMR (CDCl ): d 7.98È7.09 (m, aromatics),
3
4.62 [d, PCH, 2J(PH) \ 10.3 Hz], 3.64 [q, NCH,
The hydroformylation reactions were carried out in a 300 ml
stainless steel Parr autoclave equipped with a magnetic drive
and an internal glass vessel. The temperature was controlled
by a rigid heating mantle and by a single loop cooling coil.
The autoclave was purged three times under vacuum/argon
before introducing the catalytic solution (for the amounts
2J(HH) \ 6.8 Hz], 2.50 (s, b, NH, exchange with D O), 1.26
2
[d, CH , 2J(HH) \ 6.8 Hz]. 31PM1HN NMR (CDCl ): d 31.64
3
3
(s). 13CM1HN NMR (CDCl ): d 148.32È128.82 (m, C-aromatics),
3
60.40 [d, PCH, 1J(PC) \ 78.6 Hz], 55.25 [d, NCH,
3J(PC) \ 11.4 Hz], 21.95 (s, CH ). IR: l
\ 1270 (CH Cl ),
3
P/O
2 2
1265 cm~1 (CHCl ). mp \ 187 ¡C. [a] (CH Cl , 25 ¡C,
used, see footnote to Table 2). The 1 : 1 COÈH mixture was
3
D
2
2
c \ 3.27 g/100 ml) \ [63¡. Elem. anal. calc. for C
H
NOP:
2
prepared by mixing the pure gases in a 500 ml stainless steel
27 26
C 78.81, H 6.37, N 3.41; found: C 78.83, H 6.45, N 3.77%.
cylinder before introduction into the autoclave at the desired
pressure (see Table 2). The zero time for the kinetic runs (see
Fig. 3) was considered as the time at which the desired tem-
perature (Table 2) was reached inside the autoclave. In order
to maintain as constant temperature and pressure conditions
as possible during each kinetic run, only a few ml of the cata-
lytic solution mixture were carefully withdrawn each time
from the autoclave under pressure. The CHCl or C D
Synthesis of (R,S)- and (S,S)-Ph P(S)CH(Ph)NHCH(Ph)-
2
CH , 3. To a solution of (S)-PhCH2NCH(Ph)CH (0.739 g,
3
3
3.54 mmol) in CH Cl (10 ml) was added Ph PH (0.63 ml,
2
2
2
3.54 mmol). The mixture was stirred for 3 h under an inert
atmosphere, leading to a pale yellow solution. Molecular
sulfur (0.139 g, 4.34 mmol) was introduced into the solution
and stirring was continued for seven days at room tem-
perature. The solvent was then removed under vacuum
leaving an orange viscous oil. The NMR analysis of this crude
product showed both diastereomers of 3 with a d.e. (anti) of
70%, slightly contaminated by the imine and the
diphenylphosphine sulÐde. Crystallization from methanol at
[80 ¡C a†orded the pure ligand (1 : 1 diastereomeric mixture)
as a white powder (0.144 g, 29%). Single crystals were
obtained by slow di†usion of an ethanolic solution of 3 onto
the same volume of water. anti-3: 1H NMR (C D ): d 8.01È
3
6 6
solvent was then rotary evaporated at room temperature and
the yellowÈorange viscous residue was analyzed by 1H and
31P NMR spectroscopies in CDCl solution.
3
Crystal structures of syn-2 and syn/anti-3
Single crystals of syn-2 and syn/anti-3 were obtained by slow
di†usion of an ethanolic solution of 2 or 3 onto the same
volume of water. Intensity data were collected on a Nonius
Kappa CCD at 293 K for syn-2 and 110 K for syn/anti-3.
Both structures were solved by direct methods and reÐned by
full-matrix least-squares methods (SHELXL-97)33 with the aid
of the WINGX program.34 Non-hydrogen atoms were aniso-
tropically reÐned. With the exception of the NH hydrogen
atom (for both compounds), which was located in the Ðnal
di†erence Fourier maps and reÐned freely, all hydrogen atoms
were included in a riding model. Crystallographic data for
compounds syn-2 and syn/anti-3 are reported in Table 3.
6
6
6.69 (m, aromatics), 4.53 [d, PCH, 2J(PH) \ 11.7 Hz], 3.60 [q,
NCH, 2J(HH) \ 6.2 Hz], 3.23 (s, br, NH), 1.21 [d, CH ,
3
2J(PH) \ 6.5 Hz]. 31PM1HN NMR (C D ):
d
50.28 (s),
6
3
6
(CDCl ): 50.25 (s). 13CM1HN NMR (CDCl ): d 145.01È127.06
3
(m, aromatics), 59.16 [d, PCH, 1J(PC) \ 66.20 Hz], 55.00 [d,
NCH, 3J(PC) \ 14.8 Hz], 24.64 (s, CH ). syn-3: 1H NMR
3
(C D ):
d
8.01È6.69 (m, aromatics), 4.92 [d, PCH,
6
6
2J(PH) \ 9.1 Hz], 3.73 [q, NCH, 2J(HH) \ 6.2 Hz], 3.43 (s,
Table 3 Crystal data and structure reÐnement for syn-2 and syn/anti-3
syn-2
syn/anti-3
Formula
M
T /K
C
411.46
293(2)
H
NOP
C
427.52
110(2)
H
NSP
27 26
27 26
Crystal system
Orthorhombic
Monoclinic
Space group
P2 2 2
P2
1 1 1
1
a/A
b/A
c/A
a/¡
5.804(1)
9.0532(4)
16.911(3)
23.518(6)
90
90
90
2308.3(8)
4
0.710 73
0.137
8469
18.5975(9)
13.4262(8)
90
92.218(2)
90
2258.8(2)
8
0.710 73
0.228
14 691
b/¡
c/¡
k/AŽ
3
Z
j/A
k/mm~1
Collect. reÑect.
Indep. reÑect.
4099 [R(int) \ 0.025]
9440 [R(int) \ 0.033]
Obs. reÑect [I [ 2p(I)]
R (obs. reÑect.)
R (indep. reÑect.)
3776
7329
R a \ 0.036, wR b \ 0.086
R a \ 0.042, wR b \ 0.075
1
2
1
2
R a \ 0.041, wR b \ 0.089
R a \ 0.069, wR b \ 0.083
1
2
1
2
a R \ & p F o [ o F p/& o F o. b wR \ [&w(F 2 [ F 2)2/&[w(F 2)2]1@2 where w \ 1/[p2(F 2) ] (0.036P)2 ] 0.49P] for syn-2 and w \ 1/[p2(F 2)
1
o
c
o
2
o
c
o
o
o
] (0.031P)2 ] 0.0P] for syn/anti-3 where P \ [Max(F 2, 0) ] 2*F2]/3.
o
c
1022
New J. Chem., 2001, 25, 1015È1023

View Article Online
12 K. Yamakoshi, S. J. Harwood, M. Kanai and M. Shibasaki,
T etrahedron L ett., 1999, 40, 2565.
13 A. M. Maj, K. M. Pietrusiewicz, I. Suisse, F. Agbossou and A.
Mortreux, T etrahedron: Asymmetry, 1999, 10, 831.
14 J. Andrieu, J. Dietz, R. Poli and P. Richard, New J. Chem., 1999,
23, 581.
15 J. Andrieu, C. Baldoli, S. Maiorana, R. Poli and P. Richard, Eur.
J. Org. Chem., 1999, 3095.
16 G. Alvaro, C. Boga, D. Savoia and A. Umani-Ronchi, J. Chem.
Soc., Perkin T rans. 1, 1996, 875.
17 M. B. Eleveld, H. Hogeveen and E. P. Schudde, J. Org. Chem.,
1986, 51, 3635.
18 I. Le Gall, P. Laurent, E. Soulier, J. Y. Salaun and H. des
Abbayes, J. Organomet. Chem., 1998, 567, 13.
19 B. Patel, S. J. A. Pope and G. Reid, Polyhedron, 1998, 17, 2345.
20 M. J. Baker, M. F. Giles, A. G. Orpen, M. J. Taylor and R. J.
Watt, J. Chem. Soc., Chem. Commun., 1995, 197.
21 S. O. Grim and E. D. Walton, Inorg. Chem., 1980, 19, 1982.
22 M. T. Pinillos, M. P. Jarauta, D. Carmona, L. A. Oro, M. C.
Apreda, C. Foces-Foces and F. H. Cano, J. Chem. Soc., Dalton
T rans., 1989, 1987.
23 M. T. Pinillos, M. P. Jarauto, D. Carmona, L. A. Oro, M. C.
Apreda, C. Foces-Foces and F. H. Cano, J. Organomet. Chem.,
1988, 345, C13.
24 C. G. Arena, F. Nicolo, D. Drommi, G. Bruno and F. Faraone, J.
Chem. Soc., Chem. Commun., 1994, 2251.
CCDC reference numbers 164809 and 164810. See http://
www.rsc.org/suppdata/nj/b1/b100217l/ for crystallographic
data in CIF or other electronic format.
Acknowledgements
We are grateful to the Ministere de lÏEducation Nationale, de
la Recherche et de la Technologie (MENRT) for support and
for a PhD fellowship (to J.-M. C.), the Centre National de la
Recherche ScientiÐque (CNRS) and the Conseil Regional de
Bourgogne for support of this work and Mr Cedric Balan for
technical assistance.
References
1
2
3
4
5
6
7
8
9
S. Gladiali, J. C. Bayon and C. Claver, T etrahedron: Asymmetry,
1995, 6, 1453.
F. Agbossou, J.-F. Carpentier and A. Mortreux, Chem. Rev.,
1995, 95, 2485.
J. P. Rieu, A. Bouchere, H. Cousse and G. Mouzin, T etrahedron,
1986, 42, 4095.
25 G. Chelucci, M. Marchetti and B. Sechi, J. Mol. Catal. A: Chem.,
1997, 122, 111114.
H. R. Sonawane, N. S. Bellur, J. R. Ahuja and D. G. Ulkarni,
T etrahedron: Asymmetry, 1992, 3, 163.
26 S. Naili, I. Suisse, A. Mortreux, F. Agbossou, M. Ait Ali and A.
Karim, T etrahedron L ett., 2000, 41, 2867.
M. Dieguez, O. Pamies, A. Ruiz, S. Castillon and C. Claver,
Chem. Commun., 2000, 1607.
27 M. Otto, J. Parr and A. M. Z. Slawin, Organometallics, 1998, 17,
4527.
N. Sakai, S. Mano, K. Nozaki and H. Takaya, J. Am. Chem. Soc.,
1993, 115, 7033.
F. Fache, E. Schultz, M. L. Tommasino and M. Lemaire, Chem.
Rev., 2000, 100, 2159.
J. A. J. Breuzard, M. L. Tommasino, F. Touchard, M. Lemaire
and M. C. Bonnet, J. Mol. Catal. A: Chem., 2000, 156, 223.
C. Abu-Gnim and I. Amer, J. Chem. Soc., Chem. Commun., 1994,
115.
28 A. G. Abatjoglou, E. Billig and D. R. Bryant, Organometallics,
1984, 3, 923.
29 A. G. Abatjoglou and D. R. Bryant, Organometallics, 1984, 3, 932.
30 S. Sjovall, P. H. Svensson and C. Andersson, Organometallics,
1999, 18, 5412.
31 G. Reddelien, Ber Dtsch. Chem. Ges., 1913, 46, 2718.
32 W. Hewertson and H. R. Watson, J. Chem. Soc., 1962, 1490.
33 G. M. Sheldrick, SHELX97 (Includes SHELXS-97, SHELXL-97),
Release 97-2, Programs for Crystal Structure Analysis, University
of Gottingen, Germany, 1998.
10 C. Abu-Gnim and I. Amer, J. Organomet. Chem., 1996, 516, 235.
11 C. Basoli, C. Botteghi, M. A. Cabras, G. Chelucci and M.
Marchetti, J. Organomet. Chem., 1995, 488, C20.
34 L. J. Farrugia, J. Appl. Crystallogr., 1999, 32, 837.
New J. Chem., 2001, 25, 1015È1023
1023