New hybrid bidentate ligands as precursors for smart catalysts

Article abstract of DOI:10.1002/chem.200500542

1-Phosphanorbornadiene derivatives were grafted onto various periodically organized mesoporous powders, including a new zirconia/silica mixed oxide synthesized by aerosol techniques. After complexation with the [Rh(CO) ₂]⁺ fragment, these materials were revealed to be more active in olefin hydrogenation than their homogeneous counterparts. The reasons for this higher activity are discussed in the light of theoretical modeling. Various surface treatments, such as esterification, drying, and functionalization with PhSi(OEt)₃, provided insights into the nature and mechanism of formation of the active species. Zirconia-based materials were found to be active in internal olefin hydroformylation. Investigation of the mechanism of this reaction shows that the isomerization step is catalyzed by the Lewis acidic support, whereas the hydroformylation step is driven by the rhodium catalyst. Dissociation of these two steps leads to enhancement of activity.

Full text of DOI:10.1002/chem.200500542

DOI: 10.1002/chem.200500542
New Hybrid Bidentate Ligands as Precursors for Smart Catalysts
Frédéric Goettmann,[a, b] Cédric Boissière,[b] David Grosso,[b] François Mercier,[a]
Pascal Le Floch,*[a] and Clément Sanchez*[b]
Abstract: 1-Phosphanorbornadiene de-
rivatives were grafted onto various pe-
riodically organized mesoporous pow-
ders, including a new zirconia/silica
mixed oxide synthesized by aerosol
techniques. After complexation with
the [Rh(CO)2]+ fragment, these mate-
rials were revealed to be more active
in olefin hydrogenation than their ho-
mogeneous counterparts. The reasons
for this higher activity are discussed in
the light of theoretical modeling. Vari-
ous surface treatments, such as esterifi-
cation, drying, and functionalization
with PhSi(OEt)3, provided insights into
the nature and mechanism of forma-
tion of the active species. Zirconia-
based materials were found to be
active in internal olefin hydroformyla-
tion. Investigation of the mechanism of
this reaction shows that the isomeriza-
tion step is catalyzed by the Lewis
acidic support, whereas the hydrofor-
mylation step is driven by the rhodium
catalyst. Dissociation of these two steps
leads to enhancement of activity.
Keywords:
hydroformylation
·
mesoporous materials · P ligands ·
rhodium · supported catalysts
Introduction
ery in the early 1990s of periodically organized mesoporous
materials[3] opened a new area in the field of supported ca-
talysis: due to their high specific surface and their well-con-
trolled porosity, these materials are suitable for immobiliz-
ing catalytic complexes, and have been shown to preserve
both the catalytic selectivity and activity, and to allow high
reaction rates. Moreover, significant efforts to improve their
synthesis by increasing access to the support and presenting
more open pore geometries made to fine control of the dif-
fusion of the reactants in the inorganic matrix, and thus ac-
cessibility of the catalytic centers, possible.[4]
Two well-known approaches to heterogenization are usu-
ally employed[5] (Figure 1): the first, homogeneous support-
ed catalysis, relies on well-defined homogeneous catalysts
tethered to a support by an anchoring moiety. To limit inter-
actions between the support and the metal fragment, the an-
choring moiety is positioned a long way from the metal
center. The second approach, heterogeneous molecular ca-
Although homogeneous catalysts are usually more active
and selective than heterogeneous ones (apart from metallic
nanoparticles), their commercial success lies a long way
behind.[1] Indeed, heterogeneous catalysts are easier to
handle and can give a more sustainable catalytic process
even if they often require more drastic conditions and so-
phisticated purification techniques. Consequently, much
work has been devoted to heterogenization of useful homo-
geneous catalytic systems, especially organometallic com-
plexes, by using various supports such as clays, polymers,
and ceramics.[2] Among these materials, the most widely
studied and employed are oxides, particularly silica. Discov-
[a] F. Goettmann, Dr. F. Mercier, Prof. P. Le Floch
Laboratoire “Hétéroéléments et Coordination”
UMR CNRS 7653 (DCPH), Département de Chimie
Ecole Polytechnique, 91128 Palaiseau Cedex (France)
Fax : (+33)169-334-570
E-mail: lefloch@poly.polytechnique.fr
[b] F. Goettmann, Dr. C. Boissière, Dr. D. Grosso, Prof. C. Sanchez
Laboratoire de Chimie de la Matière Condensée
Université Pierre et Marie Curie
4 place Jussieu, 75005 Paris (France)
Fax : (+33)144-274-769
E-mail: clems@ccr.jussieu.fr
Figure 1. Examples of supported homogeneous catalysts. Left: the an-
choring chain consists of a propylsiloxane moiety (ref. [34]). Right: an
SOMC catalyst, with the metal directly bonded to the surface (ref. [6]).
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
7416
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 7416 – 7426

FULL PAPER
talysis (or surface organometallic chemistry (SOMC)[6]), ex-
ploits the presence of hydroxyl moieties on the surface of
the support to bind an isolated metal center to a well-con-
trolled environment.
new HBL, namely a grafted phosphanorbornadiene phos-
phonate derivative. This material was tested in the hydroge-
nation of alkenes and also found an interesting application
in the hydroformylation of 2,3-dimethylbut-2-ene, in which
process the Lewis acidity of the wall was exploited to pro-
mote alkene isomerization before the rhodium-catalyzed hy-
droformylation. This new class of hybrid, periodically organ-
ized, mesoporous materials built on the surface of binary (or
tertiary) mesoporous oxides opens a host of possibilities for
the controlled design of smart polyfunctional catalysts.
We recently reported on the synthesis[7] of a new family of
mixed P–O grafted ligands that fill the gap between these
two approaches, namely hybrid bidentate ligands (HBLs). In
this approach, an organic scaf-
fold (the 1-phosphanorborna-
diene skeleton) bearing a heter-
oatom that can coordinate to a
transition metal is still present,
as in homogeneous supported
complexes, but an atom of the
Results
anchoring moiety, typically an
Synthesis of the anchorable phosphanorbornadiene species
oxygen atom, also binds the
metal center (Figure 2). Since
1-Phospha-2-triethoxysilyl-4,5-dimethyl-3,6-diphenylnorbor-
nadiene (2): Synthesis of 2 was achieved by a well-known
procedure that relies on the reactivity of 2H-phospholes
toward functional alkynes (e.g., 1) (Scheme 1).[9] Important-
the latter atom is also linked to
the surface, these parts of such
systems can be regarded as hy-
brids between classical support-
ed catalysts and SOMC-based
catalysts.
Figure 2. First example of an
HBL rhodium complex
(ref. [7]) in which the metal
center is coordinated by both a
purely organic and an inorgan-
ic arm.
The proximity of the complex
and the pore wall provides a
very rigid and well-defined environment for the metal
center. Therefore, such bifunctional systems are suitable for
studying the influence of pore size and wall effects (curva-
ture, functionalities) on the catalytic activity.[8]
Herein we report on the catalytic behavior of rhodium(i)-
based catalysts coordinated by both silica- and zirconia-sup-
ported HBLs. For the silica-based materials, both the mech-
anism of formation and the influence of the functionaliza-
tion of the wall on the catalytic activity were studied. The
synthesis of new zirconia-rich, mesoporous, mixed ZrO2/
SiO2 powders is described, allowing characterization of a
Scheme 1. Synthesis of anchorable phosphine 2.
ly, the [4+2] cycloaddition process was very regioselective,
due to the electronic properties of the alkoxysilyl moieties.
As explained previously,[7] 2 behaved as a bidentate ligand
toward rhodium(i) precursors and variable-temperature
1H NMR experiments on the complex [Rh(cod)2][PF6]
(cod=cyclooctadiene) showed that the three ethoxy moiet-
ies exchange rapidly in solution with a coalescence tempera-
ture of À208C.
Abstract in French: Des dérivés du 1-phosphanorbornadiène
ont été greffés sur différents supports mésoorganisés, dont de
nouveaux oxydes mixtes silice–zircone produits par aérosol.
Ces matériaux, après complexation à des précurseurs de rho-
dium(I), se sont révélés être des catalyseurs d’hydrogénation
des oléfines plus actifs que leurs équivalents homogènes. Les
causes possibles de cette augmentation d’activité sont passés à
l’aune d’une étude théorique. De plus divers traitements de
surface, comme l’estérification des hydroxyls de surface, le sé-
chage, et la fonctionnalisation par PhSi(OEt)3, permettent
une meilleure compréhension du mécanisme de formation
des espèces actives. Les matériaux à base de zircone se sont
avérés être actifs en hydroformylation des oléfines internes.
L’étude du mécanisme de cette catalyse montre que l’étape
d’isomérisation de l’oléfine est catalysée par la zircone alors
que l’hydroformylation l’est par le complexe de rhodium. Le
fait que ces deux étapes soient réalisées par des catalyseurs
différents explique la bonne activité observée.
1-Phospha-2-triethoxysilyl-4,5-dimethyl-3,6-diphenylnorbor-
nene (2’): In ligand 2, the protons around the siloxane
moiety are not numerous enough to provide efficient CP-
MAS 29Si NMR measurements once grafted. Therefore the
1-phosphanorbornene 2’ was synthesized (see Scheme 2), as-
suming that it would display the same grafting ability as 2.
(The liquid 29Si NMR signal had quite a low chemical shift
(À49.6) for an alkyltriethoxysilane.
1-Phospha-2-ethoxydimethylsilyl-4,5-dimethyl-3,6-diphenyl-
norbornadiene (4): Compound 4 was synthesized by the
same procedure as for 2 (Scheme 3), by using ethoxydime-
Chem. Eur. J. 2005, 11, 7416 – 7426
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7417

C. Sanchez, P. Le Floch et al.
thylammonium bromide (CTEAB) as surfactant. It had a
specific surface area of 750 m2 gÀ1, an average pore diameter
of 23 Å, and a cubic (SBA-1) pore structure organization.
The second was a standard SBA-15 type support[13] with sur-
face area 630 m2 gÀ1, pore diameter 60 Å; the pore structure
had a hexagonal (p6m) arrangement. The last one was an
MSU type of support, specific surface area 310 m2 gÀ1, aver-
age pore diameter 110 Å.[14]
Mixed zirconia/silica materials: Mesoporous ZrO2/SiO2
mixed oxides are of growing importance as they provide en-
hanced mechanical and chemical stability compared with
their pure zirconia counterparts.[15] This is the first reported
synthesis of such materials using standard aerosol tech-
niques. In this work, we focused on obtaining final material
with high zirconia loadings. Preparation of the sol has been
reported previously for synthesis of thin films.[16] The atomi-
zation sol was prepared by mixing a Zr-containing sol
(Sol A), molar ratio ZrCl4/EtOH/surfactant=1:40:c, and a
given volume of a Si-containing sol (Sol B), molar ratio
SiCl4/EtOH/surfactant=1:40:c, fixing the Zr/Si ratio. The
surfactant loading depended on the nature of the surfactant.
Water was added just before atomization in a ratio H2O/
(Zr+Si)=10:1. The aerosol was then generated using a com-
mercial atomizer with dry air as atomization gas. The result-
ing powders were labeled ZSx(S) (in which x=Si loading
[mol%], ranging from 0 to 30; (S)=surfactant; (S)=F for
F127 or (S)=C for cetyltrimethylamonium bromide
(CTAB)).
When using a Pluronic surfactant (F127), as for the syn-
thesis of thin layers, the transmission electronic micrographs
(TEMs) and the XRD pattern showed a monodisperse po-
rosity (2q=0.94 for ZS20F). However, upon calcination at
3008C the powders became dark brown and FT-IR spectra
revealed the formation of carbonates, resulting from incom-
plete oxidation of organics. At higher temperature (4508C),
the specific surface dropped dramatically (from 110 m2 gÀ1
to <5 for ZS20F, for example). The porosity of ZS10F and
ZS20F before calcination is shown in Figure 3.
This calcination step was avoided by using CTAB as a sur-
factant, which can be removed by ethanol washing after 24 h
of consolidation at 1308C. Using CTAB, we were able to
synthesize mixed oxide aerosols with x ranging from 0 to
30%. Specific surface areas of the powders obtained were
fairly high (230–470 m2 gÀ1; see Table 1) for zirconium-rich
metal oxide porous powders. Average pore sizes were rather
small, even with ZS20C, but this is in good agreement with
the well-established sintering of zirconia-rich materials
during consolidation.[16] Structural characteristics of the zir-
conia/silica mixed powders are presented in Table 1 and
TEMs of a pure zirconia powder, ZS10C, ZS20C, and ZS30C
are shown in Figure 4.
Scheme 2. Synthesis of model phosphine 2’.
Scheme 3. Synthesis of a methylated derivative (4) of 2.
thyl(phenylethynyl)silane (3), which was readily prepared by
reaction of one equivalent of chlorodimethyl(phenylethy-
nyl)silane[10] with one equivalent of ethanol in THF. Com-
pound 4 was fully characterized by NMR techniques.
1-Phospha-4,5-dimethyl-3,6-diphenylnorbornadienylphos-
phonic acid (5): The synthesis of phosphanorbornadiene 5
has already been reported.[11] It was also shown that it be-
haves as a bidentate ligand toward rhodium(i) precursors.
These complexes were used as water-soluble hydroformyla-
tion catalysts but exhibited a low activity.
1-Oxo-1-phospha-4,5-dimethyl-3,6-diphenylnorbornadienyl-
phosphonic acid (5’): This was synthesized by HNO3 oxida-
tion of 5 in methanol (Scheme 4). It was characterized by
NMR and IR techniques.
Scheme 4. Synthesis of the oxide of 5.
Synthesis of the supporting material
Synthesis of the grafted materials
Silica materials: The silica materials used have all been de-
scribed previously. The first, of the SBA-1 type,[12] was syn-
thesized under acidic conditions in the presence of cetyltrie-
Grafting on the silica supports: The same procedure was
used for the three supports. Ligand 2 (2’) was added to a
7418
www.chemeurj.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 7416 – 7426

Hybrid Bidentate Ligands
FULL PAPER
suspension of the corresponding support in toluene at room
temperature and then stirred at 908C for 12 h. The resulting
materials are labeled 2@SBA-1, 2’@SBA-1, 2@SBA-15, and
2@MSU. The ligand loading was checked by thermogravi-
metric analysis (TGA) and 14, 12, 6, and 8 wt%, respective-
ly, of the organic compound was grafted. The CP-MAS
1H NMR characterizations of
2@SBA-1 showed that there
were few ethoxy groups left on
the surface, indicating that the
grafting was almost complete.
This was confirmed by CP-MAS
29Si NMR experiments on 2’@
SBA-1, in order to obtain good
cross-polarization yields. The
chemical shift (d=À75 ppm)
was consistent with a species
which is triply bonded (T3) to
the surface, relative to its d=
À49.61 ppm shift before grafting. Though the CP experi-
ments are not quantitative, the CP-MAS 31P NMR spectrum
indicated that a large part of the phosphine remained unoxi-
dized (d=À8 ppm).
Figure 3. TEMs of ZS10F127 (top) and ZS20F127 (bottom), showing the po-
Grafting on the zirconia-rich supports: Although grafting of
phosphonic-based species on zirconia-rich materials should
be easier than for the silica equivalents, because of the small
pore size of ZS20C, we applied the same grafting procedure
as for 2@SBA-1. The grafting was followed by 31P NMR
spectroscopy of the solution, with Ph3P as internal standard.
After 12 h of heating detectable traces neither of 5 nor of 5’
were found in solutions corresponding to a 20 wt% loading.
The CP-MAS 31P NMR spectrum of 5@ZS20C showed two
large main peaks at d=3.5 and 54 ppm, which were ascribed
to the phosphonate moiety[17] and to the phosphine, respec-
tively. Interestingly, the second chemical shift (d=54 ppm)
is shifted significantly downfield relative to that recorded
for the free ligand (d=À9 ppm). Usually, such a downfield
shift indicates that the phosphorus atom has been oxidized.
However, the FT-IR spectra of 5@ZS20C and 5’@ZS20C are
distinct, providing clear evidence that 5@ZS20C is not in an
oxidized state. In particular the characteristic absorption
band of the P=O bond (at 1380 cmÀ1) is missing in the spec-
trum of 5@ZS20C. (As described previously for phospho-
rosity before calcination.
Table 1. Characteristics of the zirconia/silica mixed powders.
Silica
content [%] surface
area [m2 gÀ1
Specific
BJH average
pore diameter [Å] d spacing [nm]
XRD[a]
]
ZS0C
0
470
<10
<20
20
–
ZS10C 10
ZS20C 20
ZS30C 30
320
230
230
4.2
5.5
–
<10
[a] X-ray diffraction.
nates grafted on titania,[18] no phosphonate P=O bands are
À1
À À
visible around 1200 cm , whereas the P O Zr bands are
visible around 1100 cmÀ1 together with the signal of the sup-
port; Figure 5).
Another possible explanation for this downfield shift is
that the phosphine was protonated by the acidic zirconia
surface. However, parallel experiments led us definitely to
rule out this hypothesis. Indeed, even when it was dissolved
in aqueous 12M HCl, ligand 5 proved to be very difficult to
protonate and only a very weak signal (compared with that
of nonprotonated 5) was detected at d=8 ppm in the
31P NMR spectrum. This chemical shift is consistent with the
tabulated value. For example, in ClPH(Ph)3 the phosphorus
Figure 4. TEMs of a) pure zirconia powder; b)ZS10C; c) ZS20C;
d)ZS30C.
Chem. Eur. J. 2005, 11, 7416 – 7426
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7419

C. Sanchez, P. Le Floch et al.
yield (2+Ph)@MSUd. However, such a surface treatment
does not usually result in complete drying or in isolated OH
groups, as this requires more drastic conditions.[21] Heating
at higher temperature (1208C), however, led to total passi-
vation of the catalytic material.
Complexation of the solids: The grafted materials were sus-
pended in CH2Cl2 with a given amount of rhodium(i) precur-
sor, [Rh(cod)2][PF6] or [Rh(acac)(CO)2], and then washed
three times with MeOH, yielding the corresponding [Rh-
(ligand)]@solid compounds.
Figure 5. FT-IR spectra of 5@ZS20C and 5’@ZS20C. Note the strong peak
at 1380 cmÀ1 (arrow) characteristic of R3P=O vibrations.
[Rh(2)(cod)]@SBA-1: The CP-
MAS 31P NMR spectrum exhib-
ited a single peak at d=58, con-
sistent with the spectrum of 6
(the structure of which is illus-
trated here; the other ligands
on the rhodium center are
omitted). No signal correspond-
ing to the [PF6À] counter anion
could be detected.
atom resonates at d=0 ppm.[19] The third hypothesis is that
the phosphorus atom of the grafted ligand 5 can coordinate
with the zirconia surface. Interestingly, 31P NMR experi-
ments indicated that reaction of two equivalents of 5 with
ZrCl4 in EtOH led to the formation of a new peak at d=
61 ppm (À5 ppm for the phosphonate moiety). This down-
field chemical shift is consistent
with that recorded for 5@
ZS20C. Therefore, it seemed
reasonable to assume that 5@
ZS20C adopts the structure
shown here (the locations of
the phosphonate oxygen atoms
are not exact). The coordina-
[Rh(2)(CO)2]@SBA-1: The FT-IR spectrum of the material
obtained with 2@SBA-1 when using [Rh(acac)(CO)2] as a
precursor clearly indicated that a displacement of the acac
ligand occurred (the two CO ligands featuring two charac-
teristic bands at 2002 and 2074 cmÀ1). These data corrobo-
rate the assumption that the ligand also behaves as a biden-
tate P^OÀ chelate.
tion of the phosphorus atom
with the surface probably ex-
plains why no oxidation takes place.
[Rh(5)(CO)2]@ZS20C: For this support, only [Rh-
(acac)(CO)2] was used as a precursor. CP-MAS 31P NMR
spectroscopy showed that the peak at d=54 ppm in 5@
ZS20C shifted to d=58 ppm (the same chemical shift as for
[Rh(2)(cod)]@SBA-1). FT-IR spectroscopy showed that the
acac ligand was not eliminated, since it can also act as a
ligand for the zirconia surface.[22] However, the shift of the
two C=O stretching bands (1590 and 1538 cmÀ1 on the solid
instead of 1564 cmÀ1 and 1527 cmÀ1 in [Rh(acac)(CO)2])
clearly indicates that it has been displaced by the grafted
ligand 5 (5@ZS20C), which thus acts as a bidentate P^OÀ
chelate.
This assumption was reinforced by additional solid-state
NMR experiments. We supposed that if a coordination of
the phosphorus atom with the zirconia surface occurred, the
resulting bond would probably be weak because of the poor
affinity of tertiary phosphines for ZrIV centers. Thus dis-
placement reactions were attempted using alkynes which
can act as ligands to ZrIV centers through their p system.[20]
As expected, reaction of 5@ZS20C with an excess of phenyl-
acetylene for 12 h resulted in the appearance in solid-state
31P NMR spectrum of a new signal at d=À8 ppm, which
corresponds to the free phosphine moiety of 5.
Catalytic tests: The catalytic activity of the supported cata-
lysts that had been prepared was evaluated in two processes:
the hydrogenation of 1-hexene and the hydroformylation of
2,3-dimethylbut-2-ene. All the homogeneous complexes
were formed in situ whereas all the supported complexes
were synthesized before being used as catalysts.
Synthesis of surface-modified grafted material
(2+OMe)@SBA-1: In order to check the influence of the
functionalization of the surface on the catalytic activity, 2@
SBA-1 was esterified by suspending the material for 1 h at
458C in methanol. Modification of the surface was con-
firmed by a 1.5 wt% uptake in TGA.
Hydrogenation of 1-hexene: All these tests were at room
temperature under H2 (7 bar). The metal/substrate ratio was
set at 1:1000.
(2+Ph)@SBA-1, (2+Ph)@MSU: Similarly, 2@SBA-1 and
2@MSU were treated with PhSi(OEt)3 at 908C in toluene
for 12 h. This resulted in an uptake of 6 wt% for both sup-
ports. (2+Ph)@MSU was dried for 24 h in vacuo at 608C to
Homogeneous catalysis: All the experiments were per-
formed using MeOH as solvent. As complex [Rh(2)(cod)]-
7420
www.chemeurj.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 7416 – 7426

Hybrid Bidentate Ligands
FULL PAPER
Table 2. Turnover frequencies [cyclesminÀ1] for the catalytic hydrogena-
tion of 1-hexene.
[PF6] proved to be inactive in the hydrogenation of hexene,
complex [Rh(4)(cod)][PF6] was tested in order to avoid the
exchange of the ethoxy moieties. A very low activity was ob-
served, combined with a long activation period (around
40 min). The addition of 1 equiv of water to [Rh(4)(cod)]-
Entry
Catalyst
MeOH
Toluene
1
2
3
4
5
6
7
8
9
[Rh(2)(cod)]@SBA-1
([Rh(2)(cod)]+OMe)@SBA-1
[Rh(2)(cod)]@SBA-15
[Rh(2)(cod)]@MSU
[Rh(5)(CO)2]@ZS20C
([Rh(2)(cod)]+Ph)@SBA-1
([Rh(2)(cod)]+Ph)@MSU
([Rh(2)(cod)]+Ph)@MSUd
48
–
4
10
–
–
–
30
12
4.6
10.3
[a]
72
10
30
15
12
15
–
À
[PF6] resulted in the breaking of the Si O bond, yielding a
P^OÀ chelate (Scheme 5). Indeed, the 31P signal of the com-
[b]
([Rh(2)(cod)]+Ph)@MSUd
[a] Empty entries (–) correspond to conditions that were not tested.
[b] Catalytic activity after addition of water (Rh/H2O=1:1).
Hydroformylation of 2,3-dimethylbut-2-ene
Scheme 5. Formation of the active species in the hydrogenation of 1-
hexene with [Rh(4)(cod)]PF6.
Hydroformylation catalysis: To check whether the Lewis
acidic properties of the zirconia-rich wall of ZS20C could be
advantageous for difficult reactions, [Rh(5)(CO)2]@ZS20C
was tested in the hydroformylation of 2,3-dimethylbut-2-ene,
a substrate known to be rather reluctant to undergo hydro-
formylation. In this reaction, isomerization into 2-methyl-
3,3’-dimethylprop-1-ene takes place before the hydroformy-
lation, which yields 3,4-dimethylpentanal (Scheme 6).[23]
plex shifts from d=52.8 to 52.4 ppm, whereas the release of
1
À
ethanol is clearly visible in the H spectrum (complexed Si
OEt moieties feature a quadruplet of doublets at d=
3.5 ppm with J(H,H)=5 Hz and J(H,Rh)=1 Hz, whereas
after hydrolysis a quadruplet appears at d=3.7 ppm with J-
3
3
3
(H,H)=7 Hz) The neutral complex had the same activity as
the cationic one but no activation period was needed. The
hydrogen consumption in the two experiments is presented
in Figure 6.
Scheme 6. Principle of the 2,3-dimethylbut-2-ene hydroformylation.
The results obtained after 48 h in experiments with
0.1 mol% of catalyst at 1108C under H2/CO (1:1, 30 bar),
summarized in Table 3, demonstrate that the supporting ma-
terial itself is active in isomerization of the internal olefin
(entry 1). The homogeneous catalyst is about four times less
active than the supported one (entries 2 and 3).
Table 3. Hydroformylation of 2,3-dimethylbut-2-ene.
Catalyst
Isomerization Aldehyde
Linearity[c]
rate[a] [%]
selectivity[b] [%] [%]
Figure 6. Hydrogen consumption during the hydrogenation of 1-hexene
1
2
3
ZS20C
[Rh(5)(CO)2]
[Rh(5)(CO)2]@ZS20C
30
0
0
3
10
40
100
100
100
~
&
with [Rh(2’)(cod)][PF6]: without addition of water; with pre-hydroly-
sis.
[a] Total percentage of 2,3-dimethylbut-1-ene formed. [b] Total percent-
age of aldehyde formed. [c] 3,4-Dimethylpentanal formed as a percentage
of the total aldehydes formed.
Heterogeneous catalysis: The substrate/catalyst ratio (calcu-
lated on the basis of the theoretical rhodium loading) was
set at 1000:1. The dry catalytic powders were suspended in
the solution containing the substrate before pressurization.
The most interesting catalytic activities are summarized in
Table 2.
The solids behaved well during recycling, as previously
mentioned.[7] The powders only had to be filtered off and
could be re-used directly in a new catalytic batch. By this
procedure a maximum TON of more than 100000 catalytic
cycles for [Rh(2)(cod)]@SBA-1 was reached (without any
notable activity loss after 20 recycles).
Discussion
Nature of the active species: The activation period (40 min)
observed when the [Rh(2’)(cod)][PF6] complex was used as
the catalyst strongly suggests that the active species is prob-
ably not the cationic species. Indeed, as previously ex-
plained, hydrolysis of this complex clearly promoted the hy-
drogenation process. Though CP-MAS Si29 NMR experi-
mental data showed that 2@SBA-1 was triply bonded
through to oxygen atoms on the surface, NMR and IR re-
Chem. Eur. J. 2005, 11, 7416 – 7426
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7421

C. Sanchez, P. Le Floch et al.
sults clearly suggest that 2@SBA-1 behaves as an LX-type
À
ligand. Therefore, a Si O bond must open during or just
after the complex formation, giving rise to the active species.
This assumption is supported by the catalytic tests on
([Rh(2)(cod)]+Ph)@MSUd. Indeed, when a dried catalyst
was used in toluene as solvent, the activity was found to be
much lower (4.6 catalytic cycles per minute) than when
methanol was used as solvent (15 catalytic cycles per
minute). Interestingly, when a wet catalyst was used the ac-
tivity was found to be comparable whatever the solvent
used (see Table 2, entry 7). This observation can be rational-
À À
ized by considering that methanol is able to break Si O Si
Figure 7. Optimized structures of models of complexes I and II.
bonds (esterification of the surface).[24] This Si O bond
À
breaking may be facilitated by coordination of one oxygen
lone pair to the rhodium center (Scheme 7). Furthermore,
the catalytic activity was observed to increase (10.3 cycles
per minute, comparable with the wet catalyst) when water
was added to the batch containing the dried catalyst in tolu-
ene and the batch was reloaded (Table 2, entry 9).
Table 4. Comparison of the most interesting bond lengths [Å] and angles
[8] in the model complexes I and II.
I
II
2.189
1.947
1.662
1.830
P–Rh
O1–Rh
Si–O1
2.196
2.111
1.725
1.638
Si–O2
P-Rh-O1
C1-P-Rh-O1
47.5
À1.6
91.5
1.6
À
Si O bond is also noted (1.726 Å in I; 1.621 Å for an unco-
ordinated oxygen atom). Though these variations are rela-
tively small, they reinforce the hypothesis that upon coordi-
À
nation to Rh the Si O bonds are weakened. The natural
À
Scheme 7. Formation of the active species through breaking of an Si O
bond by H2O or an alcohol. The other ligands on rhodium have been
omitted.
charge at rhodium in both model complexes I and II is also
important. Coordination of the oxygen atom in a covalent
manner results in a small decrease in the positive charge at
rhodium (+0.23 in II; +0.28 in I) (see Figure 8). However,
this increase in the electron density on the Rh center is too
modest to explain the enhancement of the catalytic activity
on going from the cationic to the neutral complex.
To gain insight into the influence of the bonding in the
active species, DFT calculations were performed using the
Gaussian O3 package with the B3PW91 functional. For de-
tails of the combination of basis sets used, see the Experi-
mental Section. The model complexes I and II were opti-
mized. For simplification the phenyl moieties of the corre-
sponding complexes were replaced by hydrogen atoms and
the ethoxy moieties by methoxy fragments, and neither the
counter ion nor other ligands were considered in these cal-
culations. Complex I is a cationic 14VE (valence electron)
complex in which coordination of the rhodium center occurs
through the lone pair at phosphorus and one lone pair of
one oxygen atom of the Si(OMe)3 group. In the neutral
14 VE complex II, which models the proposed active spe-
cies, the ligand behaves as an LX ligand. The two optimized
structures are presented in Figure 7 and the most significant
bond lengths and bond angles are listed in Table 4.
Figure 8. NBO charge distribution in complexes I and II.
Thus the effects of the oxygen and the phosphorus atom
of the phosphanorbornadiene on possible ligands in trans
positions were also investigated. For this purpose, structures
III and IV, both featuring two coordinated ethylene ligands
at rhodium, were optimized at the same level of theory. The
C3–C4 ligand is trans to the oxygen ligand and the C4–C5
one trans to the phosphorus atom of the phosphanorborna-
diene. For a view of the optimized structures see Figure 9;
the most significant bond distances and angles are listed in
Table 5.
Interestingly, a natural bond analysis of the optimized
structures revealed that, in complex I, coordination of one
lone pair on the Rh center only slightly reduces the negative
charge at the coordinated oxygen atom (À0.86 in I; À0.90
for an uncoordinated oxygen atom). A lengthening of the
7422
www.chemeurj.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 7416 – 7426

Hybrid Bidentate Ligands
FULL PAPER
Though some electronic differences exist between I and II
and between III and IV, it is clear that all these data cannot
account definitely for the higher catalytic activity of com-
plex II since the differences are so subtle. A possible explan-
ation may reside in the strain provided by the metallacycle
Rh-P-Si-O in II and IV. Indeed, whereas the oxygen ligand
is covalently bonded in II, the Si-O-Me group in I probably
acts as a hemilabile ligand, thus transiently leaving a poorly
coordinated and probably highly reactive rhodium center.
Further calculations are currently in progress to confirm this
hypothesis.
Figure 9. Optimized structures of models of complexes III and IV.
Influence of the surface functionalization: One important
factor controlling the activity of a heterogeneous catalyst is
the accessibility of the catalytic sites. The turnover frequen-
cy of a support with small pores and high tortuosity such as
SBA-1, with a cubic Pm3n structure, is lower (48 cycles per
minute) than that of a more open one like SBA-15, of 2D
hexagonal structure (72 cycles per minute). However, acces-
sibility is directed not only by the pore geometry, but also
by the interaction between the reactants and the surface of
the support. The high hydrophilicity of mesoporous silica
gels synthesized under acidic conditions[25] accounts for the
rather low activity observed with [Rh(2)(cod)2]@SBA-1 in
toluene, which is quite a classic solvent for homogenous hy-
drogenation (Table 2, entry 1). When methanol is used as
solvent, esterification of the silica surface takes place,
making it more hydrophobic and thus increasing the activity.
This is emphasized by the fact that with a material pretreat-
ed with methanol, the catalysis conducted in toluene
(entry 2 in Table 2) is about two times faster. When a meso-
porous silica-based material treated with PhSi(OEt) is used
(entry 6 in Table 2), the influence of the affinity between the
surface and the solvent becomes even more pronounced as
the activity in toluene is two times greater than in methanol.
Table 5. Comparison of the most interesting bond lengths [Å] and angles
[8] in the model complexes III and IV.
III
IV
2.319
P–Rh
2.312
O1–Rh
Rh–C2
Rh–C3
C2–C3
H2-C2-C3-H3
Rh–C4
Rh–C5
C4–C5
H4-C4-C5-H5
2.290
2.122
2.121
1.399
153.32
2.247
2.241
1.367
162.948
2.045
2.126
2.127
1.399
153.05
2.224
2.221
1.371
164.59
Interestingly, it appears that the effect of the oxygen li-
gands, whatever the charge at the rhodium center, is similar.
The olefin that is trans to the oxygen is more strongly
bonded to the rhodium center, whereas the olefin trans to
the P atom is more weakly bonded. This is clearly apparent
À
from the Rh C bond lengths which are always shorter for
À
the C2 C3 coordinated olefin, reflecting a stronger p back-
bonding. Concomitantly, this shortening is accompanied by
À
a lengthening of the C C bond (for example, in III: from
À
À
1.399 Å for C2 C3 to 1.367 Å for C4 C5). Evidence for the
À
stronger p back-donation in the C2 C3 olefin is also provid-
Influence of the chemical nature of the wall: A zirconia-rich
material does not seem to alter the catalyst behavior for hy-
drogenation notably. As shown by DFT calculations, the
charge on the rhodium center is slightly higher with ligand 5
than with the hydrolyzed ligand 2. The mean pore diameter
is also slightly smaller in ZS20C than in SBA-1. Both factors
tend to make [Rh(2)(CO)2]@SBA-1 a much more active cat-
alyst. Surprisingly the difference between the latter catalyst
and [Rh(5)(CO)2]@ZS20C is quite small (48 minÀ1 versus
30 minÀ1). This could be explained by the good affinity of
carbon–carbon double bonds for Lewis acids. If the associa-
tion constant between the hexene molecules and the acidic
sites of the zirconia surface is high enough, the diffusion
mechanism can be seen as a two-dimensional rather than a
standard three-dimensional one. According to this model,
the substrate moves onto the surface rather than being in
the solution, and thus the probability of meeting the active
center is enhanced.
ed by a decrease in the dihedral angle H2-C2-C3-H3 com-
pared with H4-C4-C5-H5. However, a conclusion about the
respective effect of oxygen ligands in the two structures is
much more difficult to reach. The results of an NBO (natu-
ral bond orbital) analysis are presented in Figure 10. In con-
trast to compound II, covalent coordination of the oxygen
ligand increases the positive charge at rhodium. This indi-
À
cates a better electronic transfer to the C2 C3 olefinic
ligand, which is more negatively charged in IV than in III.
Homogeneous tandem catalysts have attracted great inter-
est in recent years, in particular for industrial applica-
tions:[26] indeed, they allow one-pot multistep reactions with-
Figure 10. NBO charge distribution in complexes III and IV.
Chem. Eur. J. 2005, 11, 7416 – 7426
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7423

C. Sanchez, P. Le Floch et al.
SDT6960 instrument, from RT to 10008C at 58CminÀ1
.
Solid-state
out the expensive and fastidious product separation steps.
The main drawback of a homogeneous approach is that in-
compatible catalysts, such as bases and acids, cannot be
used. Anchoring of such catalysts on a solid support circum-
vents these problems. The most important papers in this
field report the anchoring of two catalytic entities on the
same support[27] or the use of an inorganic solid which itself
has two different catalytic properties.[28] Bifunctionality in
the case of [Rh(5)(CO)2]@ZS20C relies both on the pres-
ence of a rhodium center acting as an hydroformylation cat-
alyst and on the Lewis acidic properties of the wall. The
latter property can be exploited in the isomerization of in-
ternal olefins to yield linear aldehydes. As is evident in the
hydroformylation of 2,3-dimethylbut-2-ene, zirconia alone is
able to isomerize alkenes whereas the nongrafted rhodium
complex does not yield any isomerization product and is in-
efficient as a catalyst. This comparison confirms that the iso-
merization process is the slow step of the catalytic cycle. In
contrast, [Rh(5)]@ZS20C is able to perform both transfor-
mations simultaneously. Though the resulting bifunctional
catalyst compares with other reported homogeneous sys-
tems,[23] the concept in itself is new.
31P NMR spectra were recorded on
a Brucker Avance 300 (7.6 T,
300 MHz for 1H, 121 MHz for 31P) at a spinning rate of 14 kHz. Butyl-
lithium in ether solution was purchased from Aldrich.
Triethoxy(phenylethynyl)silane (1): BuLi (6.25 mL, 1.6m solution in hex-
anes, 10 mmol) was added carefully to a solution of phenylacetylene
(1.02 g, 10 mmol) in THF (15 mL) at À708C with vigorous stirring. The
solution was allowed to warm slowly to room temperature. After 2 h the
dark brown reaction mixture was added slowly to a solution of chloro-
triethoxysilane (1.98 g, 10 mmol) in THF (15 mL) at 08C. After being al-
lowed to warm again to RT, the solvents were removed in vacuo. The
LiCl salts were precipitated with hexanes and filtered off. The hexanes
were removed in vacuo, then the product was purified by Kugelrohr dis-
tillation (1608C at 10 mm) yielding a colorless oil. Yield: 2.5 g (95%);
3
1H NMR (CDCl3, 300 MHz): d=1.3 (t, J(H,H)=7 Hz, 9H; CH3), 3.9 (q,
3J(H,H)=7 Hz, 6H; OCH2), 7.3 (m, 3H; CH of phenyl), 7.5 ppm (m,
2H; CparaH of phenyl); 13C NMR (CDCl3, 75.5 MHz): d=18.4 (s, CH3),
59.5 (s, OCH2), 85.4, 104.5 (2 s, C of the alkyne), 122.3 (s, Cipso of the
phenyl), 128.7–132.8 ppm (3s, C of the phenyl); MS (IE): m/z: 263 [M+].
1-Phospha-2-triethoxysilyl-4,5-dimethyl-3,6-diphenylnorbornadiene (2):
A
solution of 1 (1.32 g, 5 mmol) and 1-phenyl-3,4-dimethylphosphol
(0.96 g, 5 mmol) in xylene (5 mL) were heated together in a sealed tube
at 1408C for 4 h. The reaction mixture was filtered over celite and xylene
was removed in vacuo. The product was isolated as a colorless oil. Yield:
2 g (90%); 1H NMR (CDCl3, 300 MHz): d=0.9 (t, 3J(H,H)=7 Hz, 9H;
CH3), 1.3 (s, 3H; CH3), 1.9 (m, 2H; CH2), 2.0 (s, 3H; vinylic CH3), 3.5
(q, 3J(H,H)=7 Hz, 6H; OCH2), 7.0, 7.2–7.5 ppm (d, J(H,H)=7 Hz, 2H;
m, 8H; phenyls); 13C NMR (CDCl3, 75.5 MHz): d=16.3 (s, vinylic CH3),
18.5 (s, CH3 of the ethoxy moieties), 20.9 (s, CH3), 59.5 (s, OCH2), 67.3
(s, CH2), 73.9 (s, C3), 125.5–128.7 (4s, aromatic CH), 139.8 (d, 2J(C,P)=
21 Hz, C4), 140.5, 155.8 (2s, Cipso of the phenyls), 141.5 (d, 1J(C,P)=
44.5 Hz, C2), 151.2 ppm (d, 1J(C,P)=25.7 Hz, C5); 31P NMR (CDCl3,
Conclusion
We have shown here that hydride bidentate ligands (HBLs)
are new materials that are able to bond to transition metals
and to act as very efficient catalysts. The range of characteri-
zation techniques that can be used, including kinetic studies,
allows good identification both of the complexes formed
and of the active species. The synthesis of mixed zirconia/
silica powders with a periodically organized mesoporosity
by an aerosol technique provides a convenient method for
the elaboration of supports that present additional Lewis
acidic properties (compared with standard silica). Moreover,
the use of ligands such as phosphines or amines bearing
phosphonate or carboxylate moieties opens up a new branch
of HBLs with promising catalytic properties, as confirmed
by the behavior of [Rh(5)]@ZS20C. Indeed, multifunctional
catalysis can be driven very easily by a suitable combination
of the catalytic properties of both the grafted complex and
the wall.
121.5 MHz):
d=À8.3 ppm;
MS
(IE):
m/z:
453
[M++1].
(2’):
1-Phospha-2-triethoxysilyl-4,5-dimethyl-3,6-diphenylnorbornene
Triethoxysilane (1.64 g, 10 mmol) and PtCl2 (26 mg, 0.1 mmol) were
added to a solution of phenylacetylene (1.02 g, 10 mmol) in ethanol
(15 mL). The solution was stirred at 508C for 12 h. The solvent was re-
1
moved in vacuo. H NMR showed a total conversion of the alkyne to cis-
triethoxystyrylsilane and triethoxy(1-phenylvinyl)silane in a 2:1 ratio. 1-
Phenyl-3,4-dimethylphosphol (1.29 g, 6.6 mmol) was added to the reac-
tion products without solvent and the mixture was heated in a sealed
tube at 1408C for 4 h. 31P NMR showed that only one product was
formed. The unreacted triethoxy(1-phenylvinyl)silane was removed by
Kugelrohr distillation. The product was isolated as a colorless oil. Yield:
1
3
1.1 g (60%); H NMR (CDCl3, 300 MHz): d=1.0 (t, J(H,H)=8 Hz, 9H;
CH3), 1.2 (s, 3H; CH3), 1.3 (s, 3H; vinylic CH3), 1.5 (m, 2H; CH2), 3.0
À
(d, J(H,P)=7 Hz, 1H, Si CH), 3.6 (m, 6H; OCH2), 7.0, 7.2–7.5 ppm (d,
J(H,H)=7 Hz, 2H; m, 8H; phenyls); 13C NMR (CDCl3, 75.5 MHz): d =
16.5 (s, vinylic CH3), 18.3 (s, CH3 of the ethoxy moieties), 21.0 (s, CH3),
28.2 (d, 2J(C,P)=29 Hz, C1), 50.4 (s, CH2), 53.3 (s, C2), 59.0 (s, OCH2),
65.5 (s, C3), 126.5–129.5 (4s, aromatic CH), 139.2 (d, 2J(C,P)=20.4 Hz,
C4), 142.6, 151.3 (2s, Cipso of the phenyls), 143.4 ppm (d, 1J(C,P)=
21.9 Hz, C2); 31P NMR (CDCl3, 121.5 MHz): d=À14.9 ppm; MS (IE): m/
z: 455 [M++1].
Experimental Section
1-Phospha-2-ethoxydimethylsilyl-4,5-dimethyl-3,6-diphenylnorbornadiene
(4): A solution of ethoxydimethyl(phenylethynyl)silane (1.02 g, 5 mmol)
and 1-phenyl-3,4-dimethylphosphol(0.96 g, 5 mmol) in xylene (5 mL) was
heated in a sealed tube at 1408C for 4 h. The reaction mixture is filtered
on celite and the xylene was removed in vacuo. The product was isolated
as a colorless oil. Yield: 1.6 g (90%); 1H NMR (CDCl3, 300 MHz): d =
All reactions were performed routinely under an inert atmosphere of
argon or nitrogen by Schlenk techniques and using dry deoxygenated sol-
vents. Dry THF and hexanes were obtained by distillation from Na/ben-
zophenone. Dry dichloromethane was distilled on P2O5 and dry toluene
on metallic Na. NMR spectra were recorded on a Bruker AC-200 SY
spectrometer operating at 300.0 MHz for 1H, 75.5 MHz for 13C, and
121.5 MHz for 31P. Solvent peaks have been used as internal references
relative to Me4Si for 1H and 13C chemical shifts; 31P chemical shifts are
relative to an 85% H3PO4 external reference. Powder XRD spectra were
recorded on a Philipps PW 1830. BET analyses were recorded on a Mi-
cromeritics ASAP 2000. IR spectra in KBr were recorded on a Nicolet
0.0, 0.04 (s, 6H, Si CH3), 1.1 (t, 3J(H,H)=7 Hz, 9H; CH3), 1.4 (s, 3H;
À
CH3), 2.1 (m, 2H; CH2), 2.2 (s, 3H; vinylic CH3), 3.5 (q, 3J(H,H)=7 Hz,
2H; OCH2), 7.0, 7.2–7.5 ppm (d, J(H,H)=5 Hz, 2H; m, 8H; phenyls);
13
À
C NMR (CDCl3, 75.5 MHz): d =0.0 (s, Si CH3), 16.5 (s, vinylic CH3),
19.0 (s, CH3 of the ethoxy moieties), 20.9 (s, CH3), 59.0 (s, OCH2), 67.5
(s, CH2), 74.2 (s, C3), 126.7–128.8 (4s, aromatic CH), 132.5 (s, C2), 139.9
(d, J(C,P)=12.8 Hz, C4), 141.3, 156.8 (2s, Cipso of the phenyls), 151.0 (d,
2
Magna 550. Thermogravimetric analyses were carried out on
a
TA
7424
www.chemeurj.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 7416 – 7426

Hybrid Bidentate Ligands
FULL PAPER
1J(C,P)=25.7 Hz, C1), 156.7 ppm (d, 1J(C,P)=14.3 Hz, C5); 31P NMR
(CDCl3, 121.5 MHz): d=À12.1 ppm; MS (IE): m/z: 393 [M++1].
Catalytic tests
Complexation: The rhodium precursor (either [Rh(cod)2][PF6] or [Rh-
(acac)(CO)2], 1 equiv) was added to a suspension of the given material
(20 mg) in dichloromethane (5 mL) under argon. After 30 min of stirring,
the powder was centrifuged and washed with dichloromethane (5 mL).
1-Oxo-1-phospha-4,5-dimethyl-3,6-diphenylnorbornadienyl
phosphonic
acid (5’): Aqueous HNO3 (1 mL, 15m, 15 mmol) was added to 5 (370 mg,
1 mmol) in methanol. After 15 min the reaction mixture was neutralized
with aqueous NaHCO3 and extracted with CH2Cl2. The organic phase
was dried over MgSO4 and the solvent was removed in vacuo. The prod-
uct was isolated as white crystals. Yield: 200 mg (55%); 1H NMR
(CDCl3, 300 MHz): d=1.2 (s, 3H; CH3), 2.0 (s, 3H; vinylic CH3), 3.4 (m,
2H; CH2), 7.0, 7.1–7.5 ppm (s, 2H; m, 8H; phenyls); 13C NMR (CDCl3,
75.5 MHz): d=14.4 (s, vinylic CH3), 17.5 (s, CH3), 67.5 (d, 1J(C,P)=
37.8 Hz, CH2), 67.2 (d, 3J(C,P)=71.0, C3), 126.2–129.2 (aromatic CH),
130.0 (m, C2), 133.7 (m, C4), 151.0 (d, 1J(C,P)=25.7 Hz, C1), 159.3 ppm
(m, C5); 31P NMR (CDCl3, 121.5 MHz): d=10.3 (d, 2J(P,P)=31.6 Hz;
Hydrogenation: Hexene (0.84 g, 1 mmol) was dissolved in MeOH
(20 mL). The weight of the catalyst added corresponded to a theoretical
substrate/Rh molar ratio of 1:1000. The mixture was introduced to a
stainless steel autoclave fitted with a glass vessel, degassed, and pressur-
ized at 7 bar. The mixture was stirred vigorously at room temperature.
The yield was determined by GC.
Hydroformylation: 2,3-Dimethylbut-2-ene (0.84 g, 1 mmol) was dissolved
in toluene (20 mL). The weight of catalyst added corresponded to a theo-
retical substrate/Rh molar ratio of 1:1000. The mixture was introduced to
a stainless steel autoclave, degassed, and pressurized at 30 bar of H2/CO
(1:1) mixture. The mixture was stirred vigorously at 1108C for 48 h. The
PO3H2), 51.7 ppm (d, 2J(P,P)=31.6 Hz; P=O); MS (IE): m/z: 387
[M++1].
À
À
1
yield was determined by GC and confirmed by H NMR spectroscopy.
Silica supports: Structural characteristics of SBA-1, SBA-15, and MSU
are summarized in Table 6.
Computational details: All computations were performed using the
Gaussian 03 suite of programs and gradient-corrected density functional
theory[29] by using the B3PW91 functional.[30] Optimizations of I and II
were carried out using the 6-311+G(d) basis set for the atoms constitut-
À À
À
ing the P C Si O chelate motif and the 6-31G* basis set for all other
atoms. The basis set employed for the rhodium atom incorporates the
Hay and Wadt small-core relativistic effective core potential and double-
zeta valence basis set (441/2111/31/1).[31] The exponent for the metal f
function applied is 1.350.[32] The same basis sets were employed for the
calculations of structures III and IV, the 6-311+G(d) basis set being em-
ployed for the C and H atoms of the two coordinated ethylene ligands.
The charge distribution in optimized structures was calculated with the
NBO partitioning scheme.[33] Minima were characterized by having no
imaginary frequency.
Table 6. Structural characteristics of SBA-1, SBA-15, and MSU.
Powder
Specific surface
Average pore
diameter [Å]
XRD
(d spacing) [nm]
area [m2 gÀ1
]
SBA-1
SBA-15
MSU
730
540
310
28
80
100
3.4
9.2
–
SBA-1:[12] In a typical run, cetyltriethylammonium bromide (1.2 g) was
dissolved at room temperature in aqueous HCl solution (61 mL, 3.6m).
TEOS (5.2 g) was then added with vigorous stirring. After 1 h a white
powder began to be precipitated; after 3 h precipitation was complete.
The powder was filtered off and washed three times with distilled water.
After 48 h of consolidation at 1108C, the surfactant was removed by a
24 h ethanolic Soxhlet extraction. A 4 h calcination step at 4508C ensur-
ed complete removal of surfactant.
Acknowledgements
We are grateful to the Saint-Gobain Recherche Company for funding.
The CNRS, the Pierre et Marie Curie University, and the École Polytech-
nique are thanked for supporting this work. The IDRIS Center (Orsay,
Paris XI) is gratefully acknowledged for the allowance of computer time.
C.S. wants to thank the dog and the goldfish.
SBA-15 and MSU: SBA-15 was synthesized according to reference [13],
and synthesis of MSU was as described in reference [14].
Mixed zirconia/silica supports
Sol A: Cetyltrimethylammonium bromide (0.625 g) was dissolved in etha-
nol (18.2 g). Zirconium tetrachloride (2.33 g) was then added with vigo-
rous stirring.
[1] B. Cornils, W. A. Herrmann, J. Catal. 2003, 216, 23–31.
[2] J. A. Gladysz, Chem. Rev. 2002, 102, 3215–3216.
[3] J. S. Beck, J. C. Vartuli, W. J. Roth, M. E. Leonowicz, C. T. Kresge,
K. D. Shmitt, C. T. W. Chu, D. H. Olson, E. W. Sheppard, S. B. Mac-
Cullen, J. B. Higgins, J. L. Schlenker, J. Am. Chem. Soc. 1992, 114,
10834–10843.
Sol B: Cetyltrimethylammonium bromide (0.625 g) was dissolved in etha-
nol (18.2 g). Silicon tetrachloride (1.68 g) was then added slowly with vig-
orous stirring, in order to avoid evolution of HCl.
Mixture: Sols A and B were mixed in a volume ratio corresponding to
the desired Zr/Si molar ratio and then atomized.
[4] A. Corma, Chem. Rev. 1997, 97, 2373–2419.
[5] A. Choplin, F. Quignard, Coord. Chem. Rev. 1998, 178, 1679–1702.
[6] C. Copéret, M. Chabanas, R. Petroff Saint-Arroman, J.-M. Basset,
Angew. Chem. 2003, 115, 164–191; Angew. Chem. Int. Ed. 2003, 42,
156–181.
[7] F. Goettmann, D. Grosso, F. Mercier, F. Mathey, C. Sanchez, Chem.
Commun. 2004, 1240–1241.
[8] G. J. D. Soler-Illia, C. Sanchez, B. Lebeau, J. Patarin, Chem. Rev.
2002, 102, 4093–4138.
Grafting procedure
Silica material: In a typical run the appropriate silica powder (100 mg)
was suspended in toluene (20 mL) under argon. The phosphine to be
grafted (50 mg) was added. The solution was heated at 908C for 12 h
with vigorous stirring. The powder was filtered off and submitted to a
12 h ethanolic Soxhlet extraction, under nitrogen. The resulting powder
was dried under vacuum for 12 h.
Mixed zirconia/silica material: In a typical run the appropriate silica
powder (100 mg) was suspended in methanol (5 mL) under argon. The
phosphinic acid (20 mg) to be grafted and triphenylphosphine (5 mg)
were added. The solution was heated at 908C for 12 h with vigorous stir-
ring. The powder was filtered off and the grafting rate was checked by in-
tegration of the 31P NMR spectra of the solution (with PPh3 as an inter-
nal standard) and submitted to a 12 h ethanolic Soxhlet extraction, under
nitrogen. The resulting powder was dried under vacuum for 12 h.
[9] F. Mathey, F. Mercier, C. Charrier, J. Fischer, A. Mitschler, J. Am.
Chem. Soc. 1981, 103, 4595–4597.
[10] N. Mézailles, N. Maigrot, S. Hamon, L. Ricard, F. Mathey, P. Le
Floch, J. Org. Chem. 2001, 66, 1054–1056.
[11] S. Lelièvre, F. Mercier, F. Mathey, J. Org. Chem. 1996, 61, 3531–
3533.
[12] V. Goletto, V. Dary, F. Babonneau, Mater. Res. Soc. Symp. Proc.
1999, 576.
Chem. Eur. J. 2005, 11, 7416 – 7426
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7425

C. Sanchez, P. Le Floch et al.
[13] D. Y. Zhao, J. L. Feng, Q. S. Huo, N. Melosh, G. H. Fredrickson,
B. F. Chmelka, G. D. Stucky, Science 1998, 279, 548–552.
[14] C. Boissière, A. Larbot, E. Prouzet, Chem. Mater. 2000, 12, 1937–
1940.
Sol–Gel Sci. Technol. 2003, 26, 43–46; d) F. Gelman, J. Blum, D.
Avnir, Angew. Chem. 2001, 113, 3759–3761; Angew. Chem. Int. Ed.
2001, 40, 3647–3649; e) F. Gelman, J. Blum, D. Avnir, J. Am. Chem.
Soc. 2000, 122, 11999–12000.
[15] Z. G. Wu, Y. X. Zhao, Y. H. Liu, Microporous Mesoporous Mater.
2004, 68, 127–132.
[16] G. Soler-Illia, E. L. Crepaldi, D. Grosso, C. Sanchez, J. Mater. Chem.
2004, 14, 1879–1886.
[17] D. Carrière, M. Moreau, P. Barboux, J. P. Boilot, O. Spalla, Lang-
muir 2004, 20, 3449–3455.
[18] G. Guerrero, P. H. Mutin, A. Vioux, Chem. Mater. 2001, 13, 4367–
4373.
[28] a) M. D. Martinez-Ortiz, D. Tichit, P. Gonzalez, B. Coq, J. Mol.
Catal. A 2003, 201, 199–210; b) D. Tichit, B. Coq, S. Cerneaux, R.
Durand, Catal. Today 2002, 75, 197–202.
[29] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A.
Robb, J. R. Cheeseman, V. G. Zakrzewski, J. J. A. Montgomery,
R. E. Stratmann, J. C. Burant, S. Dapprich, J. M. Millam, A. D. Dan-
iels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi, V. Barone, M.
Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J.
Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui, K. Morokuma, D. K.
Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. Cioslow-
ski, J. V. Ortiz, A. G. Baboul, B. B. Stefanov, G. Liu, A. Liashenko,
P. Piskorz, I. Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T.
Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challa-
combe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, J. L.
Andres, C. Gonzalez, M. Head-Gordon, E. S. Replogle, J. A. Pople,
Gaussian 98, Revision A-11, Gaussian, Inc., Pittsburgh (PA), 1998.
[30] A. D. Becke, J. Chem. Phys. 1993, 98, 5648.
[31] P. J. Hay, W. R. Wadt, J. Chem. Phys. 1985, 82, 299–310.
[32] A. Ehlers, M. Bohme, S. Dapprich, A. Gobbi, A. Hollwarth, V.
Jonas, K. Kohler, R. Stegmenn, A. Veldkamp, G. Frenking, Chem.
Phys. Lett. 1993, 208, 111.
[33] A. E. Reed, L. A. Curtiss, F. Weinhold, Chem. Rev. 1988, 88, 899–
926.
[19] “31P Nuclear Magnetic Resonance”: M. M. Crutchfield, C. H.
Dungan, J. H. Letcher, V. Mark, J. R. van Wazer, Topics in Phos-
phorus Chemistry, Wiley-Interscience, New York, 1967.
[20] S. L. Buchwald, R. B. Nielsen, Chem. Rev. 1988, 88, 1047–1058.
[21] D. W. Sindorf, G. E. Maciel, J. Am. Chem. Soc. 1983, 105, 1487–
1493.
[22] C. Sanchez, F. Ribot, New J. Chem. 1994, 18, 1007–1047.
[23] a) B. Breit, R. Winde, T. Mackewitz, R. Paciello, K. Harms, Chem.
Eur. J. 2001, 7, 3106–3121; b) A. Moores, N. Mézailles, L. Ricard, P.
Le Floch, Organometallics 2005, 24, 508–513.
[24] G. C. Ossenkamp, J. H. Kemmit, J. H. Johnston, Langmuir 2002, 18,
5749–5754.
[25] C. J. Brinker, G. W. Scherer, Sol–Gel Science: The Physics and
Chemistry of Sol–Gel Processing, Academic Press, San Diego, 1990.
[26] a) D. E. Fogg, E. N. dos Santos, Coord. Chem. Rev. 2004, 248, 2365–
2379; b) J.-C. Wasilke, S. J. Obrey, R. T. Baker, G. C. Bazan, Chem.
Rev. 2005, 105, 1001–1020.
[34] P. van Leeuwen, A. J. Sandee, J. N. H. Reek, P. C. J. Kamer, J. Mol.
Catal. A 2002, 182, 107–123.
[27] a) F. Gelman, J. Blum, D. Avnir, New J. Chem. 2003, 27, 205–207;
b) F. Gelman, J. Blum, D. Avnir, J. Am. Chem. Soc. 2002, 124,
14460–14463; c) F. Gelman, J. Blum, H. Schumann, D. Avnir, J.
Received: May 16, 2005
Published online: October 14, 2005
7426
www.chemeurj.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 7416 – 7426