Ionic phosphine ligands with cobaltocenium backbone: Novel ligands for the highly selective, biphasic, rhodium-catalyzed hydroformylation of 1-octene in ionic liquids

Article abstract of DOI:10.1021/om000183y

The use of electron-poor phosphine-substituted cobaltocenium salts as ligands for the biphasic hydroformylation in ionic liquids has been investigated. Using improved oxidation methods, 1,1′-bis(diphenylphosphino)cobaltocenium nitrate, 1,1′-bis(diphenylphosphino)cobaltocenium hexafluorophosphate, and 1,1′-bis[1-methyl(1-diphenylphosphino)ethyl]cobaltocenium hexafluorophosphate have been synthesized. 1,1′-Bis(diphenylphosphinocobaltocenium hexafluorophosphate in particular proved to be a very suitable ligand for the biphasic hydroformylation of 1-octene in 1-butyl-3-methylimidazolium hexafluorophosphate (BMIM PF₆), enabling high catalyst activity, high selectivity to the n-product, and no detectable catalyst leaching. In contrast to aqueous biphasic systems, the ionic liquid BMIM PF₆ provides for the rhodium catalyst a low-coordinating medium with limited but sufficient solubility for 1-octene to allow high reaction rates.

Full text of DOI:10.1021/om000183y

3818
Organometallics 2000, 19, 3818-3823
Ion ic P h osp h in e Liga n d s w ith Coba ltocen iu m Ba ck bon e:
Novel Liga n d s for th e High ly Selective, Bip h a sic,
Rh od iu m -Ca ta lyzed Hyd r ofor m yla tion of 1-Octen e in
Ion ic Liqu id s†
Claudia C. Brasse, Ulli Englert, and Albrecht Salzer*
Institut fu¨r Anorganische Chemie der RWTH Aachen, Prof. Pirlet Strasse 1,
D 52056 Aachen, Germany
Horst Waffenschmidt and Peter Wasserscheid*
Institut fu¨r Technische Chemie und Petrolchemie der RWTH-Aachen,
Worringer Weg 1, D 52074 Aachen
Received February 28, 2000
The use of electron-poor phosphine-substituted cobaltocenium salts as ligands for the
biphasic hydroformylation in ionic liquids has been investigated. Using improved oxidation
methods, 1,1′-bis(diphenylphosphino)cobaltocenium nitrate, 1,1′-bis(diphenylphosphino)-
cobaltocenium hexafluorophosphate, and 1,1′-bis[1-methyl(1-diphenylphosphino)ethyl]co-
baltocenium hexafluorophosphate have been synthesized. 1,1′-Bis(diphenylphosphinocobal-
tocenium hexafluorophosphate in particular proved to be a very suitable ligand for the
biphasic hydroformylation of 1-octene in 1-butyl-3-methylimidazolium hexafluorophosphate
(BMIM PF6), enabling high catalyst activity, high selectivity to the n-product, and no
detectable catalyst leaching. In contrast to aqueous biphasic systems, the ionic liquid BMIM
PF6 provides for the rhodium catalyst a low-coordinating medium with limited but sufficient
solubility for 1-octene to allow high reaction rates.
In tr od u ction
known as ionic liquids (for general reviews about ionic
liquids see ref 4). These authors described in detail the
biphasic hydroformylation of 1-pentene with [Rh-
(CO)2acac]/triarylphosphine in, for example, 1-butyl-3-
methylimidazolium hexafluorophosphate (BMIM PF6).5
However, with none of the tested ligands was it possible
to combine high activity, complete retention of the
catalyst in the ionic liquid, and high selectivity for the
desired linear hydroformylation product. The use of
PPh3 resulted in significant leaching of the Rh catalyst
out of the ionic liquid layer. This could be suppressed
by the application of sulfonated triaryl phosphine
ligands, but a major decrease in catalyst activity was
found with these ligands. All ligands used in Chauvin’s
work showed poor selectivity to the desired linear
hydroformylation product (n/iso-ratio between 2 and 4).
Obviously, the Rh-catalyzed, biphasic hydroformylation
of higher olefins in ionic liquids requires the use of
ligand systems that are specifically designed for this
application.
Metallocene derivatives have found widespread use
as ligands. Phosphine ligands based on substituted
ferrocenes are even used industrially in enantioselective
catalytic hydrogenations.1 Compared to the isoelectronic
ferrocenes, cobaltocenium cations show reduced reactiv-
ity on the cyclopentadienyl rings due to their electronic
properties, thus being inert to electrophilic and radical
attack and all but the strongest oxidants.2 Owing to
their ionic character, they show good solubility in polar
solvents. These remarkable properties stimulated our
interest in synthesizing functionalized cobaltocenium
salts and in testing these as ligands in biphasic catalytic
reactions.
Biphasic catalysis is a well-established method for
effective catalyst separation and recycling. In the case
of Rh-catalyzed hydroformylation reactions this prin-
ciple was technically realized in the Ruhrchemie-
Rhoˆne-Poulenc process, where water is used as catalyst
phase.3 Unfortunately, this process is limited to C2-C5
olefins due to the low water solubility of higher olefins.
As an alternative polar medium for biphasic hydro-
formylation, Chauvin et al. suggested novel solvents
In the present article, we describe synthesis and
characterization of ionic ligands with a cobaltocenium
backbone and their successful application in the highly
selective, biphasic, rhodium-catalyzed hydroformylation
of 1-octene using ionic liquids as catalyst solvent. New
† Dedicated to Prof. Willi Keim on the occasion of his 65th birthday.
(1) Bader, R. R.; Baumeister, P.; Blaser, H. U. Chimia 1996, 50,
99.
(2) Sheats J . E. J . Organomet. Chem. Lib. 1977, 7, 461.
(3) (a) Kuntz, E. G. French Patent 2314910, Rhone-Poulenc, 1975.
(b) Kuntz, E. G. CHEMTECH 1987 570. (c) Cornils, B.; Herrmann,
W. A. Aqueous-Phase Organometallic Catalysis; Wiley-VCH: Wein-
heim, 1998.
(4) (a) Wasserscheid, P.; Keim, W. Angew. Chem., accepted for
publication. (b) Welton, T. Chem. Rev. 1999, 99, 2071. (c) Seddon, K.
R. J . Chem. Technol. Biotechnol. 1997, 68, 351.
(5) (a) Chauvin, Y.; Mussmann, L.; Olivier, H. Angew. Chem. 1995,
107, 2941. (b) Chauvin, Y.; Olivier, H.; Mussmann, L. EP 0776 880
A1, 1996.
10.1021/om000183y CCC: $19.00 © 2000 American Chemical Society
Publication on Web 08/15/2000

Phosphine Ligands with Cobaltocenium Backbone
Organometallics, Vol. 19, No. 19, 2000 3819
Ta ble 1. Cr ysta l Da ta , Da ta Collection
P a r a m eter s, a n d Con ver gen ce Resu lts for 1
Sch em e 1
formula
fw
system
space group (no.)
C34H28CoNO3P2
619.49
monoclinic
P21/c (14)
13.073(3)
9.705(6)
11.637(3)
101.20(2)
1448(1)
4
a, Å
b, Å
c, Å
â, deg
U, Å3
Z
d
calc, g cm-3
µ, cm-1
max, deg
1.42
7.34
30
0.901
θ
min transmission
max transmission
cryst dimens, mm3
no. of reflns
0.972
0.58 × 0.15 × 0.04
4226
indep obs reflns I > 1.0 σ(I)
1393
141
0.080
0.069
no vars
Ra
Rw
b
res electron dens, e Å-3
0.45
a
b
2
R ) ∑||Fo| - |Fc||/∑|Fo|. Rw ) [∑w(|Fo| - |Fc|)2/∑w|Fo| ]1/2. w-1
and optimized syntheses of 1,1′-bis(diphenylphosphino)-
cobaltocenium nitrate (1) and 1,1′-bis(diphenylphosphi-
no)cobaltocenium hexafluorophosphate (2) are pre-
sented. The preparation of the new compound 1,1′-bis[1-
methyl(1-diphenylphosphino)ethylcobaltocenium hexa-
fluorophosphate (4) is also described.
) σ2(Fo).
Resu lts a n d Discu ssion
Syn th esis of 1,1′-Bis(d ip h en ylp h osp h in o)coba l-
tocen iu m Nitr a te (1). Initially, we developed a pro-
cedure for the synthesis of the new complex 1,1′-
bis(diphenylphosphino)cobaltocenium nitrate (1). Our
special interest in 1 was based on the idea to use this
compound as ligand in the biphasic hydroformylation
in aqueous systems.
At first, we synthesized 1,1′-bis(diphenylphosphino)-
cobaltocene (3) following an improved procedure similar
to the methods given by Davison6 and DuBois.7 The
substituted cyclopentadiene C5H5PPh2 is obtained quan-
titatively from NaCp, TlCp, or LiCp. Subsequent quan-
titative deprotonation with n-BuLi generates the anionic
ligand, and in situ reaction with Co(acac)2 affords
complex 3 in an overall yield of 59% (see Scheme 1).
The latter method avoids the use in the deprotonation
step of highly toxic thallium reagents, which have been
used previously due to their smooth reactivity and high
stability.
1,1′-Bis(diphenylphosphino)cobaltocenium nitrate (1)
could be generated directly by the oxidation of 3 with
HNO3 in propionic anhydride. The compound was
characterized by single-crystal X-ray diffraction analy-
sis. Suitable crystals could be obtained by crystallization
from hot water. Both cobaltocenium cations and nitrate
anions are located on crystallographic inversion centers;
this implies a staggered conformation of the cation with
phosphine substituents in trans position similar to 37
and orientational disorder for the anion. Crystal data,
data collection parameters, and convergence results
concerning the structure determination of 1 are pre-
F igu r e 1. PLATON plot8 of the 1,1′-bis(diphenylphosphi-
no)cobaltocenium cation. Ellipsoids are scaled to enclose
30% probability. Selected bond distances [Å] and angles
[deg]: Co-C(average) 2.038(3), Co-ring 1.644(3), P-C1
1.832(5), P-C10 1.836(6), P-C20 1.813(6); C1-P-C10
101.4(2), C1-P-C20 100.8(3), C10-P - C20 102.4(3).
sented in Table 1. The PLATON plot of 1 along with
selected intramolecular distances and angles is given
in Figure 1.
Im p r oved Syn th esis of 1,1′-Bis(d ip h en ylp h os-
p h in o)coba ltocen iu m Hexa flu or op h osp h a te (2).
Davison et al. had described a method to prepare 1,1′-
bis(diphenylphosphino)cobaltocenium hexafluorophos-
phate (2) from 3 by in situ oxidation with oxygen in
acetic acid. However, using this method, they obtained
2 in only 20% yield.6 Obviously, their oxidation method
favored the oxidation of 3 to the related phosphine
oxides. We were able to show by 31P NMR that 2
prepared in this manner indeed contained large amounts
of phosphine oxides. Additionally, Davison et al. used a
purification method by chromatography over silica. As
the polarity of 2 leads to strong interactions with silica,
we suggest high losses of the product during this
procedure.
Recently, J ohnson et al. reported 2 (which they
isolated in 5% yield) being an unstable compound, that
decomposes even in dry solvents to form a black mud.9
(6) Rudie, A. W.; Lichtenberg, D. W.; Katcher, M. L.; Davison, A.
Inorg. Chem. 1978, 17, 7 (10), 2859.
(7) DuBois, D. L.; Eigenbrot C. W., J r.; Miedaner A.; Smart J . C.;
Haltiwanger, R. C. Organometallics 1986, 5, 1405.
(8) Spek, A. L. PLATON-94; University of Utrecht: The Nether-
lands, 1994.

3820 Organometallics, Vol. 19, No. 19, 2000
Brasse et al.
Sch em e 2
Sch em e 3
In contrast, we observed 2 to be a yellow powder that
is easy to handle and indefinitely stable and storable
under nitrogen.
ylfulvene, which proceeds regioselectively according to
a known method for the synthesis of alkyl-substituted
cyclopentadienides described by Coville et al.12 and gen-
erates directly the desired monosubstituted lithio[1-
methyl-(1-diphenylphosphino)ethyl]cyclopentadienide.
Parallel to our work, Erker et al. published a similar
procedure for the synthesis of phosphinomethylenecy-
clopentadienyl ligands.13
As shown in Scheme 3, 1,1′-bis[1-methyl(1-diphenyl-
phosphino)ethyl]cobaltocenium chloride is obtained by
the in situ reaction of lithium diphenylphosphide and
6,6-dimethylfulvene followed by the reaction with CoCl2
and subsequent oxidation by C2Cl6. Using this proce-
dure, a mixture of the phosphine and the related phos-
phine oxide was obtained in 83% yield. After anion ex-
change with NH4PF6, 4 could be easily isolated in pure
form as the PF6- salt, making use of the different solu-
bility of phosphine and phosphine oxide in methanol.
Ion ic Liqu id s. First screening experiments already
revealed that the water solubility of the phosphinoco-
baltocenium salts 2 and 4 was not high enough to allow
biphasic catalytic reactions in aqueous systems. Even
with the nitrate salt 1, almost complete leaching of the
Rh catalyst into the organic layer was observed after
the hydroformylation of 1-hexene in a biphasic mixture
of water and 1-hexene.
Using a new procedure, we were able to obtain 2 in
high yields free of phosphine oxides. This was achieved
by applying to diphenylphosphinocobaltocene 3 the
smooth anaerobic oxidation method developed by Her-
berich et al.10 (see Scheme 2). The reaction of 3 with
C2Cl6 generates the oily 1,1′-bis(diphenylphosphino)-
cobaltocenium chloride. The byproduct C2Cl4 can be
easily removed by evaporation. Anion exchange leads
to stable and solid 2. Using this method, we were able
to synthesize 2 without the formation of major amounts
of byproducts.
Aiming to apply 2 as chelating ligand in catalytic
reactions, we extended the synthesis to a preparative
scale (20 g). Even on this large scale an overall yield of
63% based on CoCl2 could be achieved.
Interestingly, it is not possible to synthesize 2 in
analogy with the procedure used for the preparation of
1. The highly exothermic reaction of hexafluorophos-
phoric acid with propionic anhydride is very difficult to
control, and the formation of phosphine oxides becomes
a major problem. Additionally, it is not easy to separate
2 from these byproducts. Conventional methods to
reduce organic PV-oxides to PIII with trichlorosilane in
benzene or toluene11 are not applicable in this case due
to the insolubility of the cobaltocenium salts in nonpolar
solvents. Attempts to use dichloromethane instead were
also not successful.
Syn th esis of 1,1′-Bis[1-m eth yl(1-d ip h en ylp h os-
p h in o)eth yl]coba ltocen iu m Hexa flu or op h osp h a te
(3). The introduction of an alkyl bridge between the
cyclopentadienyl ligand and phosphorus is interesting
for two reasons. On one hand, the electronic influence
of the electron-withdrawing cobaltocenium fragment on
the phosphine is considerably weakened. On the other
hand, the introduction of the alkyl bridge enhances the
structural flexibility and its ability to adapt to different
coordination polyhedra.
Alternatively, we chose the ionic liquid 1-butyl-3-
methylimidazolium hexafluorophosphate (BMIM PF6)
as standard solvent for our investigations of the biphasic
hydroformylation of 1-octene. Carrying the same anion
as the ionic catalyst complexes under investigation, this
ionic liquid was particular suitable for our investiga-
tions. BMIM PF6 was prepared using standard methods
described by Fuller et al.14 and de Souza et al.15 or
purchased from Solvent Innovation GmbH, Ko¨ln (http://
www.solvent-innovation.de).
Ca ta lysis. The catalytic systems were prepared in
situ by mixing [Rh(CO)2acac] with the ligand in BMIM
PF6 at room temperature. The results obtained in the
Our procedure to synthesize cobaltocene derivatives
with this ligand uses in its initial step the nucleophilic
addition of lithium diphenylphosphide to 6,6′-dimeth-
(12) Coville, N. J .; du Plooy, K. E.; Pickl, W. Coord. Chem. Rev. 1992,
116, 1.
(13) (a) Erker, G.; Bosch, B. E.; Fro¨hlich R.; Meyer, O. Organome-
tallics 1997, 16, 5449. (b) Bosch, B. E.; Erker G.; Fro¨hlich, R. Inorg.
Chim. Acta 1998, 270, 446.
(14) Fuller, J .; Carlin R. T.; De Long H. C.; Haworth, D. J . Chem.
Soc., Chem. Commun. 1994, 299.
(15) Suarez, P. A. Z.; Dullius, J . E. L.; Einloft, S.; De Souza, R. F.;
Dupont, J . Polyhedron 1996, 15, 1217.
(9) Shepard, D. S.; J ohnson, B. F. G.; Harrison, A.; Parsons, S.;
Smidt, Yellowo¨ess, L. J .; Reed, S. J . Organomet. Chem. 1998, 563, 113.
(10) Herberich, G. E.; Greiss, G. J . Organomet. Chem. 1971, 27, 113.
(11) Yeung, L. L.; Lam, W. W. L.; Haynes, R. K. Tetrahedron Lett.
1996, 37, 4729.

Phosphine Ligands with Cobaltocenium Backbone
Organometallics, Vol. 19, No. 19, 2000 3821
Ta ble 2. Com p a r ison of Differ en t P h osp h in e
Liga n d s in th e Rh -Ca ta lyzed Hyd r ofor m yla tion of
1-Octen e in 1-Bu tyl-3-m eth ylim id a zoliu m
Hexa flu or op h osp h a te (100 °C, 10 ba r syn ga s
p r essu r e)a
they attribute to their ability to allow back-bonding from
the catalytically active metal atom.
The same explanation may help to understand the
much lower activity and selectivity found with ligand 4
(comparison of entries c and e, Table 2). Obviously, the
alkyl bridge between the cyclopentadienyl and the
phosphorus atom of the ligand considerably weakens the
electronic influence of the central cobalt atom on the
phosphorus atom. The resulting electronic and steric
properties of 4 seem more to resemble a bidentate
diarylalkylphosphine. Consequently, selectivity and
activity of 4 are in the same range as found with dppe.
It has to be pointed out that only with the phosphi-
nocobaltocenium ligands 2 and 4 does the reaction take
place exclusively in the ionic liquid phase (clear and
colorless organic layer). The quantitative rhodium analy-
sis of the organic layer (by AAS and ICP after evapora-
tion of the organics and treatment with concentrated
HNO3) revealed that the amount of rhodium leaching
into the organic layer is less than 0.2%. Moreover, no
rhodium or cobalt deposition was found at the reactor
walls after reaction using these ligands.
entry
ligand
TOF/h-1
n/iso
S (n-Ald)b/%
a
b
c
d
e
f
PPh3
TPPTS
dppe
dppf
4
426
98
35
828
66
810
2.6
2.6
3.0
3.8
2.6
72
72
75
79
73
94
2
16.2
a
Conditions: ligand/Rh ) 2, CO/H2 ) 1:1, t ) 1 h, T ) 100 °C,
b
p ) 10 bar, 1-octene/Rh ) 1000, 5 mL of BMIM PF6. S (n-Ald) )
selectivity to n-nonanal in the product.
biphasic hydroformylation of 1-octene using different
ligands are summarized in Table 2. To evaluate the
properties of the ionic diphosphine ligands with cobal-
tocenium backbone, we compared the results obtained
with the ligands 2 and 4 with those obtained with PPh3,
two common neutral bidentate ligands, and with NaT-
PPTS as standard anionic ligand.
It is noteworthy that only with ligand 2 was a clear
enhancement of the selectivity to the linear hydro-
formylation product observed (Table 2, entry f). With
all other ligands used in our experiments, we found
n/iso-ratios in the range between 2 and 4. While this is
in quite good accordance with known results in the case
of PPh3 (Table 2, entry a), dppe (Table 2, entry c), and
dppf (in comparison with the monophasic hydroformy-
lation, taking into account slightly different reaction
conditions16) and also in good accordance with reported
results in the case of NaTPPTS (Table 2, entry b; by
comparison with the biphasic hydroformylation of 1-pen-
tene in BMIM PF65), it is more remarkable for the
bidentate cobaltocenium ligand 4. All hydroformylation
experiments showed low hydrogenation activity (<2%)
and some isomerization activity with nonconverted
1-octene and iso-octenes being typically on the same
order of magnitude after reaction. However, in all
experiments only minor amounts of 3- and 4-nonanales
(<1%) were found.
With regard to the ligand influence on the reaction
selectivity, a direct comparison of the results obtained
with the isoelectronic and isostructural metallocene
ligands dppf and 2 is of particular interest (comparison
of entries d and f in Table 2). Considering the high
structural similarity of both ligands, their different
influence on the reaction’s selectivity has to be at-
tributed to electronic reasons. The electron density at
the phosphorus atoms is significantly lower in the case
of ligand 2 due to the electron-withdrawing effect of the
formal cobalt(III) central atom in the ligand. This
interpretation is supported by former work from Casey
et al.17 and Moser et al.18 These groups described
positive effects of ligands with electron-poor phosphorus
atoms in selective hydroformylation reactions which
With all other ligands, a significant leaching of the
catalyst into the organic layer is observed (deep yellow
color of the organic phase). In the latter case, the
catalyst is active in both phases, which makes a clear
interpretation of solvent effects on the catalyst’s reactiv-
ity difficult.
In the case of complete retention of the catalyst in
the ionic liquid (entries e, f in Table 2), an easy catalyst
separation by decantation was possible. Moreover, we
could principally demonstrate that the recovered ionic
catalyst solution can be reused. The recycling experi-
ment showed the same activity and selectivity as in the
original run.
In this context, the high activity obtained with ligand
2 in the biphasic hydroformylation of 1-octene in BMIM
PF6 is very interesting. It clearly indicates the special
usefulness of this ionic liquid for the biphasic hydro-
formylation of 1-octene. BMIM PF6 provides a polar but
weakly coordinating medium for the rhodium catalyst.
Moreover, this ionic liquid shows very favorable solubil-
ity properties for a biphasic reaction with 1-octene.
While the rate of the biphasic hydroformylation of
1-octene in aqueous systems is known to be heavily
limited by the low solubility of 1-octene in water (about
0.0001 mol % at 25 °C19), BMIM PF6 combines limited
but sufficient solubility for 1-octene (about 2.5 mol %
at 25 °C) to allow high rates even in a biphasic reaction.
Con clu sion a n d Ou tlook
1,1′-Disubstituted cobaltocenium salts bearing phos-
phine donors have been synthesized using new and
improved methods and were successfully used as biden-
tate chelating ligands in the biphasic, rhodium-cata-
lyzed hydroformylation of 1-octene in 1-butyl-3-meth-
ylimidazolium hexafluorophosphate (BMIM PF6).
For the first time, 1,1′-bis(diphenylphosphino)cobal-
tocenium nitrate (1) was synthesized by direct oxidation
of 1,1-bis(diphenylphosphino)cobaltocene with nitric
acid. 1,1′-Bis(diphenylphosphino)cobaltocenium hexaflu-
(16) (a) Casey, C. P.; Whiteker, G. T.; Melville, M. G.; Petrovich, L.
M.; Gavney J . A.; Powell, D. R. J . Am. Chem. Soc., 1992, 114, 5535.
(b) Unruh, J . D.; Christenson, J . R J . Mol. Catal. 1982, 14, 19.
(17) Casey, C. P.; Paulsen, E. L.; Beuttenmueller, E. W.; Proft, B.
R.; Petrovich, L. M.; Matter, B. A.; Powell, D. R. J . Am. Chem. Soc.
1997, 119, 11817.
(18) Moser, W. R.; Papile, C. J .; Brannon, D. A.; Duwell, R. A. J .
Mol. Catal. 1987, 41, 271.
(19) Cornils, B., Herrmann, W. A., Eds. Aqueous-Phase Organome-
tallic Catalysis; Wiley-VCH: Weinheim, 1998.

3822 Organometallics, Vol. 19, No. 19, 2000
Brasse et al.
105.72 (d, J PC ) 3.3 Hz, C1 in C5H4P) ppm. 31P{1H} NMR (THF-
d8): δ -16.26 ppm.
orophosphate (2) and 1,1′-bis[1-methyl(1-diphenylphos-
phino)ethyl]cobaltocenium hexafluorophosphate (4) could
be synthesized in good yields using a mild oxidation
method with hexachlorethane as oxidizing agent.
Compound 2, in particular, proved to be a very
suitable ligand for the biphasic hydroformylation in
ionic liquids combining high solubility in BMIM PF6
with favorable electronic properties. The latter are
ascribed to a decrease of electronic density on the
phosphine substituents compared with the analogue
ferrocene derivative caused by the influence of the
central cobalt atom.
In contrast with aqueous biphasic systems, the ionic
liquid BMIM PF6 provides a weakly coordinating me-
dium for the rhodium with limited but sufficient solubil-
ity for 1-octene to allow high reaction rates. Using ligand
2 in BMIM PF6, it was possible to realize a biphasic
hydroformylation with high catalytic activity, high
selectivity to the desired n-product (selectivity to n-
nonanal: 94%; n/iso-ratio: 16.2), and without detectable
catalyst leaching. Furthermore, the technical advantage
of catalyst separation and recycling by simple phase
separation contributes considerably to the benefit of our
new concept to immobilize catalyst complexes in ionic
liquids using cationic ligands.
1,1′-Bis(d ip h en ylp h osp h in o)coba ltocen e (3). A solution
of LiC5H4PPh2 in THF is freshly prepared and cooled to -30
°C. While Co(acac)2 is added in situ, the color of the solution
turns immediately from yellow to dark brown. The reaction is
stirred overnight, and the solvent is changed to toluene and
filtered over Celite. After adding hexane to the concentrated
filtrate, a precipitate is formed slowly during storage at -80
°C. The solid is isolated, washed with hexane, and dried in
vacuo (isolated yield: 59%).
1,1′-Bis(diph en ylph osph in o)cobaltocen iu m Nitr ate (1).
To a suspension of 0.2 g of bis(diphenylphosphino)cobaltocene
3 (0.36 mmol) in 3 mL of propionic anhydride was added 0.5
mL of concentrated nitric acid. The resulting yellow-orange
solution was added dropwise to ether, and an oil was formed.
The latter was washed with ether, dissolved in a very small
amount of CH2Cl2, and again added dropwise to ether. The
formation of a pale yellow precipitate was observed, which was
filtered, washed with ether, and dried in vacuo (0.13 g, 0.21
mmol, 58%). For further purification, 1 was recrystallized from
hot water to obtain suitable crystals for the crystal structure
1
determination. H NMR (CD3NO2): δ 7.8-7.35 (m, 20H, Ph);
5.85, 5.52 (m, 8H, Cp) ppm. 13C{1H} NMR (CD3NO2): δ 136.10
(s, Ph-ipso); 135.07 (d, J PC ) 20.7 Hz, Ph-ortho); 131.70 (s, Ph-
para); 130.58 (s, Ph-meta); 90.56 (s, C1 in C5H4P); 89.84 (s, C3
in C5H4P); 89.24 (s, C2 in C5H4P) ppm. IR (KBr): 1475 (w),
1431 (s), 1381 (vs), 1193, 1116, 747, 726 (m), 700, 562 (s) cm-1
.
Anal. Calcd for C34H28CoNO3P2: C, 65.92; H, 4.55. Found: C,
65.54; H, 4.46.
Exp er im en ta l Section
1,1′-Bis(d ip h en ylp h osp h in o)coba ltocen iu m Hexa flu o-
r op h osp h a te (2). LiC5H4PPh2 is prepared as follows: To a
solution of 21.2 g of LiCpPPh2 (83 mmol) in 200 mL of THF
was added 5.6 g of CoCl2 (43 mmol), producing a dark brown
solution. After stirring this solution overnight under reflux,
5.79 g (24 mmol, 1.22 equiv) of C2Cl6 was added. The resulting
solution was stirred at room temperature for another 10 min.
After evaporation of all volatile substances in vacuo, the
residue was dissolved in 100 mL of CH2Cl2 and filtered through
cotton wool and Celite to remove LiCl. The filtrate was dried
in vacuo to obtain an oily, brown residue. The oil was dissolved
in 150 mL of acetone, and 6.5 g of NH4PF6 (40 mmol) in acetone
was added. No precipitate was formed. The solution was added
dropwise to 500 mL of water, producing an oil. After extraction
with 400 mL of CH2Cl2, the solution was added dropwise to
ether. The solid was filtered off and washed with ether to
obtain 18.82 g (27 mmol, 63%) of crude 2. For further
purification the solid was dissolved in acetone and filtered over
alumina, eluted with acetone, and recrystallized from CH2-
Cl2/ether to obtain 15.07 g (21.5 mmol, 50%) of pure 2. 1H NMR
(CD3NO2): δ 7.47 (m, 20H, Ph); 5.83, 5.51 (m, 8H, Cp) ppm.
13C{1H} NMR (CD3NO2): δ 136.35 (d, Ph-ortho, 20.7 Hz); 133.0,
131.88 (Ph-meta/para); 91.0, 90.45 (Cp-C2-5) ppm. 31P{1H}
NMR (CD3NO2): δ -22.24 (br s, PR2R′); -145.13 (sept, 707.6
Hz, PF6) ppm. Anal. Calcd for C34H28CoF6P3: C, 58.17; H, 4.02;
P, 13.24; F, 16.24. Found: C, 57.57; H, 3.53; P, 13.0; F, 16.6.
IR (KBr): 3111; 1474; 1431;1381 (w); 831 (vs); 750; 697; 558
Gen er a l P r oced u r es. All reactions were carried out under
nitrogen atmosphere using standard Schlenk techniques.
Solvents were dried and deoxygenated by conventional meth-
ods prior to use. Alumina was deactivated with 5% deoxygen-
ated water. 1-Octene for the hydroformylation experiments
was used as supplied by Aldrich.
NMR spectra were recorded on a Varian Mercury 200 (200
MHz, 1H; 50 MHz, 13C; 81 MHz, 31P) at ambient temperature.
Chemical shifts (δ) are given in ppm relative to internal SiMe4
(1H and 13C spectra) or to external H3PO4 (31P spectra). Mass
spectra were obtained with a Finnegan MAT 95 spectrometer.
AAS measurements were obtained using a Perkin-Elmer 1100
spectrometer; ICP analysis were run on a Spectro Instrument
System, Spectroflame D.
The catalytic experiments were carried out in 75 mL
stainless steel autoclaves equipped with a dropping funnel and
a magnetic stirring bar. The hydroformylation products were
analyzed by gas chromatography using a Siemens Sichromat
2 with a 50 m Pona HP-FS column.
Syn t h eses.
Dip h en ylp h osp h in ocyclop en t a d ien yl-
lith iu m . Freshly prepared NaCp20 or LiCp, prepared from the
reaction of CpH with 1 equiv of n-BuLi at -78 °C, is suspended
in ether, and 1 equiv of ClPPh2 dissolved in ether is added at
-30 °C. The reaction mixture is stirred for 1.5 h at room
temperature, then filtered over Celite. The solvent is removed
in vacuo, and the residue is dissolved in THF and cooled to
-78 °C. One equivalent of n-BuLi (1.6 M in hexane) is added
dropwise. The mixture is stirred for 1.5 h at room temperature.
The product LiC5H4PPh2 can be isolated quantitatively from
the reaction mixture as a pale yellow powder in quantitative
(m); 440 (w) cm-1
.
1,1′-Bis[1-m et h yl-1-(d ip h en ylp h osp h in o)et h yl]cob a l-
tocen iu m Hexa flu or op h osp h a te (4). A 4.3 mL sample of
diphenylphosphine (25 mmol) was dissolved in 40 mL of
diethyl ether, and 15.6 mL of n-BuLi (1.6 M, 25 mmol) was
added dropwise at -60 °C. The reaction mixture was allowed
to warm to room temperature and stirred at this temperature
for 3 h. To the resulting solution of LiPPh2, 2.6 g of 6,6-
dimethylfulvene (25 mmol) was added at -10 °C. A small
amount of THF was added to prevent the formation of a
precipitate. The solution was then stirred for another 3 h at
room temperature. After cooling the reaction mixture to -40
°C, 1.9 g of CoCl2 (15 mmol) was added. The solution was
stirred overnight at room temperature.
1
yield or used in situ. H NMR (THF-d8): δ 7.32-7.28 (m, 8H,
Ph-ortho); 7.15-7.06 (m, 12H, Ph-meta/para); 5.9-5.8 (m, 8H,
C5H4 ppm. 13C{1H} NMR (THF-d8): δ 145.70 (d, J PC ) 12.1
Hz, Ph-ipso); 133.92 (d, J PC ) 18.6 Hz, Ph-ortho); 128.04 (d,
J PC ) 6 Hz, Ph-meta); 127.17 (s, Ph-para); 113.13 (d, J PC
)
20.8 Hz, C2 in C5H4P); 107.38 (d, J PC ) 8.8 Hz, C3 in C5H4P),
(20) Herrmann, W. A., Salzer, A., Eds.; Synthetic Methods of
Organometallic and Inorganic Chemistry; G. Thieme Verlag: Stuttgart,
1996; Vol. 1, p 51.

Phosphine Ligands with Cobaltocenium Backbone
Organometallics, Vol. 19, No. 19, 2000 3823
At -50 °C 2.2 g of hexachlorethane (9 mmol, 1.2 equiv) was
added, and a greenish-yellow precipitate was formed im-
mediately. The solid was filtered off, washed with diethyl ether
and hexane, and dried in vacuo. A total of 8.5 g of solid was
isolated in this way. The latter was identified by 31P NMR
spectroscopy to be a mixture of phosphine and phosphine
oxides.
The isolated solid was dissolved in acetone/water, and a
solution of 2.7 g of NH4PF6 (16.5 mmol) in water was added
at room temperature. The formation of a brownish-yellow solid
was observed. The flask was filled with water, and the solid
was filtered off. The isolated solid was washed with diethyl
ether to yield 4.39 g of 7 (5.6 mmol, 47%). For further
purification 3 was washed with 60 mL of methanol to obtain
pure 4 in the form of a shiny yellow solid. 1H NMR (CD3NO2):
δ 7.40 (m, 10H, Ph); 5.56, 5.10 (2m, 4H, Cp-CH); 1.50 (d, 6H,
3J PH ) 14.4 Hz, CH3) ppm. 13C{1H} NMR (CD3NO2): δ ) 136.2
determined by gas chromatography using di-n-butyl ether as
internal standard. Using 2 as ligand, the remaining catalytic
ionic solution could be reused at least one more time.
Cr ysta llogr a p h ic Da ta Collection a n d Str u ctu r e De-
ter m in a tion of 1. Pale yellow single crystals of 1 obtained
as described above were studied on an ENRAF Nonius CAD4
diffractometer equipped with an incident beam monochroma-
tor. Crystal data, data collection parameters, and convergence
results are compiled in Table 1. Before averaging over sym-
metry-related reflections a numerical absorption correction
based on Gaussian integration21 was applied. The structure
was solved by Patterson and Fourier difference methods and
refined on
F
with the local version of the SDP suite.22
Crystallographic data (excluding structure factors) for the
structure reported in this paper have been deposited with the
Cambridge Crystallographic Data Centre as supplementary
publication no. CCDC-143 975. Copies of the data can be
obtained free of charge on application to The Director, CCDC,
12 Union Road, Cambridge CB2 1EZ, U.K. (fax, int. +1223/
336-033; E-mail, teched@chemcrys.cam.ac.uk).
4
1
(d, J PC ) 22 Hz, Ph-meta); 134.7 (d, J PC ) 20.6 Hz, Ph-ipso);
131.31 (s, Ph-para); 129.70 (d, 3J PC ) 7.8 Hz, Ph-ortho); 119.21
1
(s, Cp-1); 83.08, 82.28 (Cp-CH); 37.1 (d, J PC ) 25.2 Hz, Cq-
(CH3)2); 28.1 (d, 2J PC ) 20.1 Hz, CH3) ppm. 31P{1H} NMR (CD3-
1
NO2): δ 30.5; -142.04 (sept, J PF) 705 Hz, PF6) ppm. MS
Ack n ow led gm en t. Financial support for H.W. and
P.W. by the European Community under the BRITE/
EuRam-project BE96-3745 is gratefully acknowledged.
A.S., C.B., and U.E. gratefully acknowledge financial
support from the Deutsche Forschungsgemeinschaft
(DFG) within the Collaborative Research Center “Asym-
metric Syntheses with Chemical and Biological Means”
and the Fonds der Chemischen Industrie.
(FAB): m/z (Irel): +VE: 641 [M+], -VE: 145 (100) [PF6]. Anal.
Calcd for C40H40CoF6P2: C, 60.92; H, 5.36. Found: C, 60.33;
H, 5.22.
Ca ta lytic Exp er im en ts. In a typical experiment, the ionic
catalyst system was prepared in situ by mixing 5.2 mg of [Rh-
(CO)2acac] (0.02 mmol) with 0.04 mmol of the ligand in 5 mL
of 1-butyl-3-methylimidazolium hexafluorophosphate (BMIM
PF6) for 1 h at room temperature. The ionic catalyst solution
was transferred to a 75 mL steel autoclave equipped with a
dropping funnel and a magnetic stirring bar. A 3.15 mL (20
mmol) sample of 1-octene were placed into the dropping funnel,
and the autoclave was heated to 100 °C and pressurized with
CO/H2 (1:1) to the reaction pressure (batch reaction mode). To
start the reaction, the dropping funnel was opened and the
autoclave was stirred at 1000 rpm. After 1 h reaction time
the autoclave was cooled and degassed. The biphasic reaction
mixture was separated by simple decantation of the upper
phase. The catalyst activity and the n/iso-selectivity were
Su p p or tin g In for m a tion Ava ila ble: Tables with crystal
data, atomic coordinates, anisotropic displacement parameters,
and listings of interatomic distances and angles. This material
is available free of charge via the Internet at http://pubs.acs.org.
OM000183Y
(21) Coppens, P.; Leiserowitz, L.; Rabinovich, D. Acta Crystallogr.
1965, 18, 1035.
(22) Sheldrick, G. M. SHELXL93, Program for Structure Refine-
ment; University of Go¨ttingen: Go¨ttingen, Germany, 1993.