Silica Hybrid Gel Catalysts Containing Group(VIII) Transition Metal Complexes: Preparation, Structural, and Catalytic Properties in the Synthesis of N,N-Dimethylformamide and Methyl Formate from Supercritical Carbon Dioxide

Article abstract of DOI:10.1006/jcat.1998.2151

Sol-gel derived hybrid materials containing group(VIII) transition metal complexes in a porous silica network were prepared and tested for the catalytic synthesis of N,N-dimethylformamide (dmf) and methyl formate (mf) from supercritical carbon dioxide, hydrogen, and dimethylamine and methanol/triethylamine, respectively. Bifunctional silylether phosphines X = Ph₂P(CH₂)₂Si(OEt)₃ and Y = Me₂P(CH₂)₂Si(OEt)₃ were used as ligands for the preparation of complexes of the type RuCl₂X₃, RuCl₂Y₃, MClX₃ (M = Ir, Rh), and MCl₂X₂ (M = Pt, Pd). The silylether complexes were anchored in a silica matrix by co-condensation with Si(OEt)₄. The textural and structural properties of the hybrid gels were characterized by means of ³¹P and ²⁹Si NMR spectroscopy, extended X-ray absorption fine structure (EXAFS), X-ray diffraction (XRD), transmission electron microscopy (TEM), and nitrogen and argon physisorption. ³¹P NMR spectroscopy confirmed nondestructive immobilization of the Ru, Pt, Pd, Rh, and Ir complexes. TEM and XRD analyses proved the homogeneity and noncrystallinity of the materials. The degree of condensation of the gels and the molecular mixing of the components has been investigated by solid state ²⁹Si NMR spectroscopy. Textural characterization showed that all gels were micro- to mesoporous. EXAFS measurements indicated no metal-metal interactions, confirming that the organometallic complexes were immobilized as monomers. From all catalysts silica matrix stabilized ruthenium complexes exhibited highest activities at 100% selectivity in dmf synthesis from CO₂, H₂, and dimethylamine. The corresponding turnover frequency (TOF) of 1860 h^-1 exceeded those reported so far for heterogeneous catalysts by a factor of 600. In methyl formate synthesis, TOFs up to 115 h^-1 were reached.

Full text of DOI:10.1006/jcat.1998.2151

JOURNAL OF CATALYSIS 178, 284–298 (1998)
ARTICLE NO. CA982151
Silica Hybrid Gel Catalysts Containing Group(VIII) Transition Metal
Complexes: Preparation, Structural, and Catalytic Properties
in the Synthesis of N,N-Dimethylformamide and Methyl
Formate from Supercritical Carbon Dioxide
,1
ꢀ
ꢀ
ꢀ
Oliver Kro¨cher, Rene´ A. Ko¨ppel, Michael Fro¨ba,† and Alfons Baiker
ꢀ
Laboratory of Technical Chemistry, Swiss Federal Institute of Technology, ETH Zentrum, CH-8092 Zu¨rich, Switzerland;
and †Institute for Inorganic and Applied Chemistry, Universita¨t Hamburg, D-20146 Hamburg, Germany
Received February 16, 1998; revised May 12, 1998; accepted May 12, 1998
of N,N-dimethylformamide (dmf) and methyl formate (mf)
from carbon dioxide has gained considerable attention as
an interesting alternative route for producing these basic
chemicals (6–20). Both reactions are exothermic but, de-
pending on the reaction conditions, neutral or slightly exer-
gonic due to the unfavorable entropy change. To reach high
product yields, high pressures and relatively low tempera-
tures have to be applied. In order to stabilize the formic
acid, which is regarded as a crucial intermediate, the ad-
dition of a basic additive is beneficial. In mf synthesis the
stabilizing additive is a tertiary amine such as triethylamine.
In dmf synthesis dimethylamine acts as a basic additive and
as reactant, forming the product on reacting with the formic
acid intermediate. To reach high reaction rates, the use of
a catalyst is crucial for this type of reactions.
Various group(VIII) based transition metal complexes
have been tested as catalysts in the synthesis of dmf and
mf from CO2 in liquid solvents (1, 6–14). Rhodium, irid-
ium, and ruthenium based systems with phosphine ligands
proved to be most effective for dmf synthesis (1, 8, 10).
However, overall catalytic activity remained low. Similar to
dmf synthesis, phosphine complexes of group(VIII) transi-
tion metals with basic co-catalysts were found to be active
homogeneous catalysts for the formation of methyl formate
from CO2, H2, and methanol in the presence of organic sol-
vents (1, 9, 13).
A substantially more efficient reaction system for dmf
and mf synthesis, not requiring any additional solvent, was
recently reported by Jessop et al. (6, 15–17), making use of
a supercritical phase of carbon dioxide (scCO2) and hydro-
gen, in which the [RuCl2(P(Me)3)4] catalyst is dissolved. In
this case scCO2 serves as both a reactant and a solvent. The
high reaction rate was attributed to favorable mass transfer
effects, high solubility of H2, and weak solvation of the cata-
lyst compared to liquid solvents (16, 21). The application of
complexes soluble in scCO2 was reported to be crucial for
achieving high activities in this reaction system.
Sol-gel derived hybrid materials containing group(VIII) tran-
sition metal complexes in a porous silica network were prepared
and tested for the catalytic synthesis of N,N-dimethylformamide
(dmf ) and methyl formate (mf ) from supercritical carbon dioxide,
hydrogen, and dimethylamine and methanol/triethylamine, respec-
tively. Bifunctional silylether phosphines X = Ph2P(CH2)2Si(OEt)3
and Y = Me2P(CH2)2Si(OEt)3 were used as ligands for the prepa-
ration of complexes of the type RuCl2X3, RuCl2Y3, MClX3 (M = Ir,
Rh), and MCl2X2 (M = Pt, Pd). The silylether complexes were an-
chored in a silica matrix by co-condensation with Si(OEt)4. The
textural and structural properties of the hybrid gels were char-
acterized by means of 31P and 29Si NMR spectroscopy, extended
X-ray absorption fine structure (EXAFS), X-ray diffraction (XRD),
transmission electron microscopy (TEM), and nitrogen and argon
physisorption. 31P NMR spectroscopy confirmed nondestructive
immobilization of the Ru, Pt, Pd, Rh, and Ir complexes. TEM and
XRD analyses proved the homogeneity and noncrystallinity of the
materials. The degree of condensation of the gels and the molecu-
lar mixing of the components has been investigated by solid state
29Si NMR spectroscopy. Textural characterization showed that all
gels were micro- to mesoporous. EXAFS measurements indicated
no metal–metal interactions, confirming that the organometallic
complexes were immobilized as monomers. From all catalysts silica
matrix stabilized ruthenium complexes exhibited highest activities
at 100% selectivity in dmf synthesis from CO2, H2, and dimethy-
lamine. The corresponding turnover frequency (TOF) of 1860 hꢁ1
exceeded those reported so far for heterogeneous catalysts by a fac-
tor of 600. In methyl formate synthesis, TOFs up to 115 hꢁ1 were
c
reached.
ꢂ 1998 Academic Press
INTRODUCTION
Based on its large-scale availability, CO2 represents a
suitable starting material for the production of valuable fu-
els and chemicals (1–7). In the past decades the synthesis
1 To whom correspondence should be addressed. E-mail: baiker@tech.
chem.ethz.ch.
284
0021-9517/98 $25.00
c
Copyright ꢂ 1998 by Academic Press
All rights of reproduction in any form reserved.

SILICA HYBRID GEL CATALYSTS
285
However, the use of homogeneous catalysts has often these functionalized silica materials and their catalytic be-
limited technical utility on a large scale, owing to difficul- havior in the synthesis of dmf and mf.
ties in catalyst separation and recovery. On the other hand,
known heterogeneous catalysts exhibited only low activity
for the synthesis of dmf (11, 22).
EXPERIMENTAL
Materials
An interesting design strategy is the heterogenization of
homogeneous transition metal complexes, resulting in ma-
terials which combine the activity, selectivity, and accessi-
bility of specially tailored metal centers in homogeneous
catalysts with the technical advantages of heterogeneous
catalysts. Early attempts to immobilize catalytically active
complexes were directed to anchor complexes to organic
polymers or inorganic supports (23–26). Among the sup-
ports used, silica has gained much attention due to the ease
of functionalization through its surface OH-groups and its
rigid structure. To surface-attach complexes, mostly bifunc-
tional silylether phosphines were bonded with the phos-
phine site to the metal center and the silylether site to free
hydroxy groups on the surface of silica. Inherent problems
with catalysts obtained by this route are leaching of the ac-
tive component under operating conditions and low surface
loading (25, 26).
In the past few years, anchoring of catalytically active
transition metal complexes via organic groups within oxide
networks (mainly silica) by a modified sol-gel process has
received considerable attention as a method to overcome
these problems (25–29). Silica gel-like metal-containing
materials with highly active tailored metal centers can
be produced by simultaneous cocondensation of tran-
sition metal silylether complexes with tetraalkoxysilane
according to:
All preparations were carried out under dry, oxygen-free
argon using Schlenk techniques unless stated otherwise.
Solvents were used in spectrophotometric purity, dried over
molecular sieve, degassed, and stored under argon. Vinyl-
triethoxysilane was supplied from Alfa and diphenylphos-
phine from Strem Chemicals. All other compounds were
supplied from Fluka. Dimethylphoshine was prepared as
described in the literature (30).
Silylether phosphine ligands. (Trisethoxysilyl)ethyldi-
phenylphosphine (1) was prepared by photochemical re-
action of vinyltriethoxysilane (20.2 g, 106 mmol) and
diphenylphosphine (19.8 g, 106 mmol) for 4 days and puri-
fied by distillation (1 Torr, 174ꢃC) to give a colorless liquid
(31), (1, 38.7 g, 97% ): 1H NMR (CDCl3) ꢀ 7.2–7.7 (m, 10 H,
ArH), 3.80 [q, J = 7.0 Hz, 6 H, OCH2CH3], 2.13 (m, 2 H,
CH2P), 1.21 [t, 9 H, J = 7.0 Hz, OCH2CH3], 0.72 (m, 2 H,
SiCH2); 31P{ H} NMR (CDCl3) ꢀ ꢁ8.4.
1
(Trisethoxysilyl)ethyldimethylphosphine (2) was pre-
pared by the reaction of vinyltriethoxysilane (13.1 g,
68.7 mmol) and dimethylphosphine (4.27 g, 68.7 mmol) ac-
cording to literature (31). Fractional distillation gave a col-
orless liquid (0.8 Torr, 61–63ꢃC), (2, 16.0 g, 92% ): 1H NMR
(CDCl3) ꢀ 3.83 [q, J = 7.0 Hz, 6 H, OCH2CH3], 1.41 (m,
2 H, CH2P), 1.23 [t, J = 7.0 Hz, 9 H, OCH2CH3], 0.99
2
[d, J(HP) = 1.8 Hz, 6 H, PCH3], 0.69 (m, 2 H, SiCH2);
1
13C{ H} NMR (CDCl3) ꢀ 58.6, 23.9 (d, 1J(CP) = 10.5 Hz, 1 C,
n Si(OEt)4 + m R-Si(OEt)3 + (4n + 3m) H2O
CH2P), 18.2, 13.4 [d, 1J(CP) = 13.5 Hz, 2 C, P(CH3)2], 5.5 [d,
H+
→
1J(CSi) = 9.2 Hz, 1 C, SiCH2]. 31P NMR (CDCl3): ꢀ = ꢁ44.1.
n Si(OH)4 + m R-Si(OH)3 + (4n + 3m) EtOH [1]
Hybrid gel precursors. Structural formula of the dif-
ferent precursor complexes are summarized in Scheme 1.
[2] RuCl2{PPh2(CH2)2Si(OEt)3}3 (3) was prepared according
to the literature (32). RuCl3 ꢄ xH2O (0.60 g, 2.30 mmol)
n Si(OH)4 + m R-Si(OH)3
H+
→
(SiO2)n(R-SiO2)m + (2n + 3/2m) H2O.
The catalysts produced are often as active as their homo- was dissolved in methanol (75 ml) in a Schlenk tube and
geneous analogues because organometallic complexes are 1 (5.18 g, 13.8 mmol) was added under stirring. Sodium
incorporated unchanged within a porous matrix which borohydride was added in small portions until the color
allows accessing the catalytically active centers. Moreover, changed to brown. After further stirring for 15 h methanol
the immobilized complexes are generally more stable to- was evaporated (15 h, 60ꢃC, 0.5 Torr). The solid residue was
wards oxygen and water, compared to their free analogues shaken 30 min with CH2Cl2 (50 ml) and the insoluble salts
and show negligible leaching.
subsequently filtered off over Celite. Finally the CH2Cl2
The aim of the present work was to tailor sol-gel de- was evaporated at 50ꢃC (15 h, 0.5 Torr) and the oily residue
rived catalysts for the synthesis of N,N-dimethylformamide extracted two times with pentane, (3, 5.93 g, 89% ): Anal.
and methyl formate from carbon dioxide, hydrogen, and di- Calcd for C60H87O9Cl2P3RuSi3: C, 55.37; H, 6.74. Found: C,
methylamine or methanol/triethylamine, respectively, with- 55.08; H, 6.46.
out additional solvents. In a preliminary study (18) we have
RuCl2{PMe2(CH2)2Si(OEt)3}3 (4). The preparation was
recently shown that sol-gel derived hybrid catalysts exhibit carried out analogously to the preparation of 3 using 2 as a
interesting properties for the synthesis of dmf. Here, we ligand. The color of the solution changed to yellow–brown
report the preparation and structural characterization of after the addition of sodium borohydride, (4, 2.37 g, 68% ).

¨
KROCHER ET AL.
286
SCHEME 1. Structural formula and denotations of group(VIII)-metal silylether complexes used as precursors for hybrid gel catalysts. Note that
structural formula of complexes [3] and [4] have not been fully confirmed (see text).
IrCl{PPh2(CH2)2Si(OEt)3}3 (5). [IrCl(COD)]2 (COD = solution was formed which was worked up as described for
1,5-cyclooctadiene) (0.491 g, 0.731 mmol) was dissolved in 7, (8, 3.10 g, 94% ): 1H NMR (CDCl3) ꢀ 7.2–7.5, 7.6–7.7 (2m,
toluene (10 ml) in a Schlenk tube and 1 (1.92 g, 5.10 mmol) 20 H, ArH), 3.71 (q, J = 7.0 Hz, 12 H), 2.47 (m, 4 H, CH2P),
1
in toluene (10 ml) was added dropwise under stirring. The 1.13 (t, J = 7.0 Hz, 18 H), 0.79 (m, 4 H, SiCH2). 31P{ H}
solution was stirred for 3 h, during which the color changed NMR (CDCl3) ꢀ 32.3 (cis), 22.0 (trans). Anal. Calcd for
from red to orange. Toluene wasevaporated (60ꢃC, 0.1Torr) C40H58Cl2O6P2PdSi2: C, 51.64; H, 6.28; Cl, 7.62. Found: C,
and the viscous residue was dissolved in hexane (25 ml). Af- 51.57; H, 6.25; Cl, 7.92.
ter cooling to ꢁ20ꢃC a red-orange oil was obtained which
Sol-gel derived catalysts. For all acid-catalyzed sol-gel
was dried at 60ꢃC and 0.1 Torr. The mother liquor was
preparations a 1.5l flat-bottom three-neck flask equipped
1
concentrated and cooled again, (5, 1.61 g, 81% ): 31P{ H}
with an argon inlet and a bubble gauge was used. Scheme 2
illustrates the main features of the incorporation of the
NMR (C6D6) ꢀ ꢁ1.8 [t, 2J(PP) = 14.1 Hz, 1 P], ꢁ13.0
[d, 2J(PP) = 13.6 Hz, 2 P]. Anal. Calcd for C60H87ClO9P3Ir
Si: C, 53.10; H, 6.46; P, 6.85. Found: C, 53.27; H, 6.66; P, 7.14.
RhCl{PPh2(CH2)2Si(OEt)3}3 (6) was prepared analog-
ously to 5, except that [RhCl(COD)]2 (0.542 g, 1.10 mmol)
and 1 (2.60 g, 7.17 mmol) was used. The color changed
from red-orange to red during the reaction, (6, 1.99 g,
1
2
71% ): 31P{ H} NMR (C6D6) ꢀ 46.9 [dt, J(PP) = 39.5 Hz,
1J(PRh) = 187.9 Hz, 1 P], 30.6 [dd, 2J(PP) = 39.4 Hz,
1J(PRh) = 139.2 Hz, 2 P]. Anal. Calcd for C60H87ClO9
P3RhSi: C, 56.84; H, 6.92; P, 7.33; Rh, 8.12. Found: C, 57.12;
H, 6.76; P, 7.34; Rh, 8.27.
cis-PtCl2{PPh2(CH2)2Si(OEt)3}2 (7). To a suspension of
PtCl2(COD) (0.997 g, 2.66 mmol) in toluene (20 ml) 1
(2.10 g, 5.58 mmol) was added dropwise. A yellow-greenish
solution was formed which was concentrated to 5 ml. After
the addition of pentane a white precipitate fell out which
was washed thoroughly with pentane, (7, 2.39 g, 88% ): 1H
NMR (CDCl3) ꢀ 7.1–7.5 (m, 20 H, ArH), 3.64 (q, J = 7.0 Hz,
12 H), 2.40 (m, CH2P, 4 H), 1.08 (t, J = 7.0 Hz, 18 H), 0.72
1
(m, 4 H, SiCH2). 31P{ H} NMR (CDCl3) ꢀ 11.8 [pseudo-t,
1J(PPt) = 3655 Hz, 2P].
cis-/trans-PdCl2{PPh2(CH2)2Si(OEt)3}2 (8).PdCl2(COD)
(1.01 g, 3.54 mmol) was suspended in toluene (25 ml) and
SCHEME 2. Reaction scheme illustrating the incorporation of RuCl2
1 (2.70 g, 7.17 mmol) was added dropwise. A clear orange {PMe2(CH2)2Si(OEt)3}3 into the silica matrix using a sol-gel process.

SILICA HYBRID GEL CATALYSTS
287
metal silylether complexes at the example of RuMe com- a sweep frequency of 162.0 MHz and using 85% ortho-
plex(Scheme 1). Appropriate amountsoftetraethoxysilane phosphoric acid as external standard. 31P NMR spectra
(TEOS) and the complexes were dissolved under argon in of solids, applying the cross-polarization magic-angle-
such an amount of acetone that no precipitation occurred spinning (CP/MAS) technique, were measured at a sweep
after addition of the acid. The amount of acid was calculated frequency of 162.0 MHz, using 85% orthophosphoric acid
to get a sevenfold excess of the stoichiometric equivalent of as external standard. Single-pulse-excitation (SPE) and
water necessary for the complete hydrolysis of the ethoxysi- CP/MAS 29Si NMR measurements were made with a
lyl groups. 2.8 N acetic acid or orthophosphoric acid was Bruker AMX 400 at 79.5 MHz, applying TMS as an internal
used. The flask was stoppered and the solution was moder- standard. The sample size was about 500 mg for solid state
ately stirred for 24 h. Then a continuous flow of argon was NMR spectroscopy. Magic-angle-spinning was routinely
passed through the flask for 3 days. During this time acetone carried out at 4 kHz spinning rate. The number of scans
evaporated off and glassy colored lumps remained, which was 2048 for CP/MAS measurements and at minimum 64
had the same color as the precursor complex. The solid was for SPE experiments. The repetition time for CP/MAS ex-
kept for further 4 days in an open beaker. The lumps were periments was about 4 s and 200 s for SPE. Chemical shifts
crushed with a mortar and washed three times with water are reported in ꢀ-units. T1ꢁH was measured via 29Si with
(20 ml) and three times with acetone (20 ml). The solid was a separate experiment described by Schaefer et al. (33).
dried for 6 h at room temperature under vacuum (0.1 Torr). Peak deconvolution and evaluation of the spectra were ac-
In a typical experiment 7 (0.713 g, 0.70 mmol) and tetra- complished with the standard software 1D WINNMR from
ethoxysilane (7.29 g, 7.82 ml, 35.0 mmol, 50 eq.) were Bruker. Measurements were performed at 300 K.
dissolved in acetone (80 ml) and orthophosphoric acid
For transmission electron microscopic (TEM) measure-
(18.2 ml) was added. Yield: 2.83 g of white RuMe-1. Thro- ments, samples were suspended in abs. ethanol, treated with
ughout the article a set of acronyms (Table 1) is used for ultrasound and applied to a carbon carrier foil. All investi-
the individual hybrid gels.
gations were carried out with a Philips CM30ST operated
at 300 kV (LaB6, Ko-Ap 3, d = 50 ꢂm, single tilt holder).
X-ray powder diffraction (XRD) patterns were mea-
sured on a Siemens 2/2 D5000 powder X-ray diffractome-
ter. The diffractograms were recorded with CuKꢃ radiation
over a 22 range of 10ꢃ to 80ꢃ using a position-sensitive de-
tector with Ni-filter.
Physicochemical Characterization
Elemental analyses of the prepared compounds were car-
ried out on a LECO CHN-900. Phosphorus contents were
determined with Pb(ClO4)2, and chlorine, by titration with
AgClO4 solution. Metal contents were determined by ICP-
AES.
The specific surface areas (SBET) and mean pore diame-
For obtaining 1H NMR spectra, a Bruker DPX 300 instru-
ment was used (sweep frequency 300 MHz, internal stan-
dard TMS). Multiplicities are abbreviated by s = singlet,
d = doublet, t = triplet, and q = quadruplet. 31P NMR spec-
tra of liquids were recorded on a Bruker AMX 400 at
¯
ters (dN2) were determined from nitrogen physisorption at
77 K using a Micromeritics ASAP 2010 instrument. Prior to
measurement, the samples were degassed to 0.1 Pa at 403 K.
SBET were calculated in a relative pressure range 0.05–0.2
assuminga cross-sectionalarea of0.162nm2 for the nitrogen
TABLE 1
Textural Properties of Hybrid Gel Catalysts
Physisorption: N2
¯
Physisorption: Ar
Metal
b
d
Catalyst
Complex
Complex : TEOS
Acid
cont./wt% SBET/m2 gꢁ1 a d(N2)/nm
Smicro/m2 gꢁ1 c d(Ar)/nm
¯
Si
RuPh-1
RuPh-2
—
0 : 100
1 :50
1 :20
1 :50
1 :20
1 : 20
1 : 50
1 : 50
1 :50
1 : 50
H3PO4
H3PO4
H3PO4
H3PO4
0
880
460
200
510
<5
3.2
2.8
3.4
2.3
—
3.4
2.1
2.4
4.1
2.3
—
—
130
290
—
—
—
2.1
2.0
—
2.1
1.8
2.0
—
RuCl2{PPh2(CH2)2Si(OEt)3}
RuCl2{PPh2(CH2)2Si(OEt)3}33
3
3
4
4
4
5
6
7
8
1.7
2.8
1.9
3.7
3.9
3.7
2.0
4.8
2.5
RuMe-1 RuCl2{PMe2(CH2)2Si(OEt)3}
RuMe-2 RuCl2{PMe2(CH2)2Si(OEt)3}3
RuMe-3 RuCl2{PMe2(CH2)2Si(OEt)3}33
H3PO4
MeCO2H
MeCO2H
MeCO2H
H3PO4
70
30
Ir
Rh
Pt
IrCl{PPh2(CH2)2Si(OEt)3}3,
RhCl{PPh2(CH2)2Si(OEt)3}
PtCl2{PPh2(CH2)2Si(OEt)3}23
PdCl2{PPh2(CH2)2Si(OEt)3}2
480
800
770
430
310
360
—
Pd
MeCO2H
210
1.8
a BET surface area.
b Mean pore diameter.
c Mircropore surface area determined from argon physisorption.
d Mean pore diameter from argon physisorption.

¨
KROCHER ET AL.
288
molecule. The pore size distributions were calculated apply- tivity detector. Gaseous compounds were separated on a
ing the Barrett–Joyner–Halenda (BJH) method (34) to the Porapak QS column (5 m, 1/8 in. OD, 80–100 mesh) and
desorption branch of the isotherms (35). The occurrence of liquid products on a Supelco SPB-1 fused silica capillary
microporosity could be estimated from t-plot constructions column (60 m, 0.53 mm ID, 5 ꢂm film).
(36). Argon physisorption measurements were carried out
Hydrogen, dimethylamine, and carbon dioxide were
at 77 K on the same instrument to determine the micropore purchased from Pangas, methanol (analytic purity) from
¯
surface area (Smicro) and the mean pore diameter (dAr), ap- Riedel-de-Hae¨n, and triethylamine (anhydrous, purum)
plying Dubinin–Astakhov calculations (37). from Siegfried Company. Prior to all tests the reactor was
The extended X-ray absorption fine structure (EXAFS) carefullycleaned with ethanol. Blank experimentswere car-
spectra were recorded at the storage ring DORIS III ried out regularly to confirm that no catalytically active ma-
(HASYLAB@DESY, Hamburg, Germany) at the terial remained in the reactor after cleaning. For further
¨
ROMOII beam line which was equipped with a Si (311) experimental details see Ref. (41).
double-crystal monochromator. In order to reduce contri-
RESULTS
butions from higher harmonics, the second monochromator
crystal was detuned to 50% of the maximum intensity. The
slit before the monochromator, as well as that in front
of the first ion chamber, was set to 1 mm height. The
Hybrid Gel Precursors
For the immobilization of organometallic complexes in
a silica matrix silylether complexes of group(VIII) trans-
ition metals have been prepared. Ligand exchange of
chloro COD complexes of Pd, Pt, Rh, and Ir with Ph2P
(CH2)2Si(OEt)3 produced the corresponding hybrid gel
precursors 5, 6, 7, and 8 (Table 1) in good yields. The cis-
configuration of 7 was confirmed by the 1J(195Pt, 31P) value
of 3655 Hz, which is typical for phosphines standing trans
to chlorine in complexes of this type (42, 43), and by com-
parison with the NMR data of the related compound cis-
PtCl2(PPh2Et)2 (44). The reaction of PdCl2(COD) with
Ph2P(CH2)2Si(OEt)3 gave 8 with a cis/trans ratio of 23 : 77.
The typical down-field shift of the cis-configuration of phos-
phine groups in this type of compounds allowed us to assign
the peaks in the 31P NMR spectra (42–46).
In case of the ruthenium complexes, only partial lig-
and exchange could be achieved when starting from
[RuCl2(COD)]n. Hence, a synthetic route from ruthe-
nium chloride and Ph2P(CH2)2Si(OEt)3 (1) or Me2P(CH2)2
Si(OEt)3 (2), respectively, with NaBH4 as reductant was
chosen. In contrast to the well defined precursors 5, 6, 7,
and 8, characterization of the Ru precursors by NMR spec-
troscopy was troublesome. In all cases five or more signal
groups were discerned in the 31P NMR spectra, making as-
signment of the peaks difficult. In contrast to the utiliza-
tion of simple phosphine ligands, the use of bifunctional
silylether ligands prevented the formation and isolation of
pure compounds. The main reasons were: (i) slightly al-
tering reaction conditions led to various products which
differed in the number of ligands or in the degree of dimer-
ization, (ii) fluctual behavior of the ligands, (iii) cleavage
of ligands in solution, and (iv) rearrangement reactions
(47).
spectra of the LIII absorption edges (Ir LIII and Pt LIII
)
were calibrated against the first inflection point of the Pt
LIII edge (11564 eV) of a Pt foil which was measured simul-
taneously. In case of the metal K edges (Ru, Rh, and Pd)
the spectra were calibrated against the first inflection point
of the Pd K edge (24,350 eV) of a Pd metal foil. All spectra
were recorded in a transmission mode at room temperature
using a continuous scan technique (QEXAFS) (38). For
the LIII edges the counting time per data point was 0.7 s
linked with a distance between data points of 2 eV, whereas
the K edges were recorded with 0.5–0.6 s per data point
and distances of 3–4 eV. All samples were prepared as
pressed polyethylene pellets in which the metal content
was adjusted to yield an absorption jump 1ꢂ of about
0.3–0.5 at the respective absorption edge. For further data
analysis the program WinXas (39) was used. Theoreti-
cal calculations of the phase shifts and back scattering
amplitudes were calculated using FEFF6.01 (40).
Catalytic Tests
Catalytic tests for the synthesis of dmf and mf were car-
ried out in a 500 ml stainless-steel autoclave agitated by
a straight 6-blade turbine. In a typical experiment for dmf
synthesis the autoclave containing 5ꢅ10ꢁ5 mol catalyst was
flushed with carbon dioxide, charged with liquid dimethy-
lamine (0.71 mol), pressurized with hydrogen (4 MPa), and
heated up to 373 K at a stirring rate of 300 minꢁ1. After
thermal equilibration hydrogen was pressed into the re-
actor (8.5 MPa) and then carbon dioxide (13.0 MPa) was
added to start the reaction. The reaction time was typically
15 h. After the reaction, the autoclave was cooled rapidly,
vented, and opened. The catalysts were filtered off using a
syringe filter (0.2 ꢂm). For the production of methyl for-
mate, methanol and triethylamine was charged instead of
dimethylamine.
Sol-Gel Derived Catalysts
An overview on the hybrid gel catalysts prepared is given
Product analysis was performed using a gas chromato- in Table 1. Among various methods tested for sol-gel pro-
graph (Hewlett Packard) equipped with a thermal conduc- cessing of the silylether complexes, a procedure similar to

SILICA HYBRID GEL CATALYSTS
289
TABLE 2
Comparison of 31P NMR Data of the Prepared Transition-Metal Complexes
in Solution and in the Solid Silica Matrix
Solution 31P NMR
31P CP MAS NMR
Complex
ꢀ/ppm
ꢀ/ppm
IrCl{PPh2(CH2)2Si(OEt)3}3
ꢁ13.0, ꢁ1.8a
30.6, 46.9b
11.8b
cis: 32.3b
trans: 22.0
ꢁ17.3, 4.9
39.2
10.3
cis: 30.8
trans: 21.3
RhCl{PPh2(CH2)2Si(OEt)3}3
cis-PtCl2{PPh2(CH2)2Si(OEt)3}2
cis-/trans-PdCl2{PPh2(CH2)2Si(OEt)3}2
a Solution spectrum in C6D6.
b Solution spectrum in CDCl3.
that reported by Schubert et al. (27) with modified gela- of the geometry about the metal center is reported in pre-
tion conditions adjusted to the precursors used gave best vious investigations (46). Based on these results no clear
results. The molar ratio metal complex: TEOS was varied assignment can be made. The two most feasible explana-
from 1 : 20 to 1 : 50. High silica amounts were chosen for tions are the distortion of the true concentrations by the
good fixation of the complexes within a stable matrix and CP-technique rather than a real change in geometry, or al-
to minimize negative interactions of neighboring complex ternatively, reversing the highfield/downfield shifts of the
molecules. Lower molecular ratios than 1 : 20 resulted in cis and trans structures in the solid matrix.
leaching of the catalyst, as indicated by colored washing liq-
The structure of the silica network of catalysts RuMe-1
uids. The metal contents of the catalysts, calculated on the and RuMe-3 (Table 1) was investigated by 29Si NMR mea-
basis of metal used for preparations and amount of sample surements. In Fig. 2 the SPE-spectrum of catalyst RuMe-
obtained after drying, varied from 1.7–9.2 wt% .
3 is shown as an example. Two groups of peaks are ob-
The resulting raw materials were glassy and clear, indi- served in the spectrum, corresponding to silicon species
cating a macroscopically homogeneous distribution of the with four oxygen neighbors (Q-type species) and with three
complexes. The acids could be washed out completely af- oxygen neighbors (T-type species), respectively. Q-type
ter gelation. For the RuMe-catalysts no gelation procedure species originate from TEOS, whereas T-type species are
leading to clear gels could be found. Precipitation always derived from the silylether complex. After deconvolution
occurred after addition of any kind of acid, base, or water. of the peaks, silicon atoms with different degrees of con-
In contrast to the partly air sensitive complex precursors, densation can be distinguished, corresponding to Si centers
all hybrid gel catalysts could be stored under ambient at- bridged via -O- linkages to one (T1), two (Q2, T2), three
mosphere after the sol-gel process.
(Q3, T3), and four (Q4) nearest neighbors. Typical isomer
Structural Properties of Hybrid Gels
NMR-spectroscopy. The 31P NMR spectra of the com-
plex precursors in solution were compared with CP/MAS
measurements of the prepared catalysts to confirm the un-
destructed incorporation of the complexes (Table 2). The
solid state NMR spectra were poorly resolved compared
to the liquid NMR spectra. The half width at full intensity
of the central component was about 15–20 ppm for the in-
corporated palladium complex and even broader for the
other compounds. Note that differences in the position of
the peaks are possibly due to the deviation of the liquid ma-
trices from the solid silica matrices. Moreover, coalescence
of peaks resolved in the solution spectra is observed for the
solid spectra (Rh-complex).
Upon immobilization of the palladium complex the ge-
ometry about the metal center apparently changed from
predominantly trans in the solution spectrum to predomi-
nantly cis in the solid state spectrum (Fig. 1). No conversion
FIG. 1. Comparison of the 31P NMR spectrum of PdCl2{PPh2(CH2)2
Si(OEt)3}2 in solution (a) with the 31P CP/MAS NMR spectrum of this
complex co-condensed with tetraethoxysilane (b).

¨
KROCHER ET AL.
290
The relaxation parameters TSiH and T1ꢁH determine dy-
namic properties of the silicon species, providing additional
information about the localenvironment ofthe different sil-
icon species. For the application of Eq. [3] these parameters
have to be determined. The T1ꢁH value describes the loss of
proton magnetization in the rotating frame and, thus, de-
pends on the mobility of the observed species on the NMR
time scale. T1ꢁH corresponds to the averaged relaxation of
the protons within 1–2 nm, due to the exchange of spin
energy within this range (48). Thus, different T1ꢁH values
indicate separate units of at least 1–2 nm size. Species with
the same T1ꢁH value should be bound to each other and
should be homogeneously distributed.
The proton relaxation time in the rotating frame T1ꢁH was
measured via 29Si with a separate experiment described by
Schaefer et al. (33). Investigating catalyst RuMe-1, a mono-
exponential decay was found for the T3 species, resulting
in a constant T1ꢁH value of 7.0 ms (Fig. 3). On the con-
trary, a multi-exponential decay could be observed for the
Q4 and Q3 silicon species. The shape of the curve can be ex-
pressed by the sum of two or three exponential components,
each characteristic of a different type of Q4 and Q3 type sili-
con species. Assuming three exponential components, three
spatial regions of Q4 species with T1ꢁH relaxation times of
5.8, 8.8, and 13.3 ms, and three regions of Q3 species with
relaxation times of 4.9, 10.0, and 14.0 ms could be distin-
guished, thus, making the application of Eq. [3], requiring
constant parameters, not reliable.
FIG. 2. 29Si NMR spectrum (SPE) of a hybrid-gel catalyst (RuMe-3)
prepared from RuCl2{PMe2(CH2)2Si(OEt)3}3. Peaks are divided into two
groups, referring to silicon species derived from tetraethoxysilane (Q-
species, right side) and from silylether ligands, belonging to ruthenium
species (T-species, left side). Peaks are assigned to corresponding sili-
con species. Deconvoluted peaks used for integrating of the different sili-
con species are shown. The simulated spectrum (dotted line), which is the
sum of the deconvoluted peaks, illustrates the good match of simulated
and measured spectra.
shift values were ꢁ48.5/ꢁ58.5/ꢁ67.5 ppm for the T1/T2/T3
signals, and ꢁ91.5/ꢁ101.5/ꢁ110 ppm for the Q2/Q3/Q4 sig-
nals. For catalyst RuMe-3, prepared with acetic acid and
a complex:TEOS ratio of 1 : 20, the respective fractions
of the different structural units were 46% Q4 {Si(OSi)4},
34% Q3 {(HO)Si(OSi)3}, 4.6% Q2 {(HO)2Si(OSi)2}, 7.4%
T3 {RSi(OSi)3}, 5.4% T2 {R(HO)Si(OSi)2}, and 2% T1
{R(HO)2SiOSi}. Catalyst RuMe-1, prepared with ortho-
phosphoric acid and a complex, TEOS ratio of 1 : 50, con-
sisted of 55% Q4, 30% Q3, 2.7% Q2, 6.4% T3, 4.8% T2, and
1.8% T1 species, indicating that a smaller part of the gel
was uncondensed. The usage of a weaker acid with catalyst
RuMe-3 decreased the ratio of completely condensed Q4
species to Q3 species from Q4/Q3 = 1.83 (catalyst RuMe-1)
to Q4/Q3 = 1.35.
Extended X-ray absorption fine structure (EXAFS). To
gain further information on the structure and state of the
organometallic complexes after their immobilization in a
silica matrix X-ray absorption spectroscopic measurements
were carried out at the absorption edge of the respective
transition metal. For determining the short-range order
around the metal atom within the different hybrid gels
the EXAFS was analyzed for at least one representative
of each group (Fig. 4). Figure 5 illustrates the evaluation
of the respective EXAFS spectra ꢄ(k) ꢄ k3 of the iridium
Further important information about the gel structure
was obtained from 29Si CP/MAS NMR measurements.
Performing a cross-polarization (CP) experiment under
Hartmann–Hahn conditions enabled much shorter mea-
surement times in comparison to the 29Si NMR investiga-
tions described above. With this technique, polarization of
hydrogen is transferred to the silicons in a T2-type process
(spin–spin process), requiring only short contact times (TC
in the millisecond range). The CP transfer can be repeated
quickly and, thus, allows fast data accumulation (33).
Various contact times TC lead to different peak integrals
in the spectra; thus, direct quantification of the different
silicon species is not possible. The observed signal inten-
sities I(TC) depend on the real amount of silyl species I0,
according to
FIG. 3. Determination of T1ꢁH parameter for a hybrid-gel catalyst
(RuMe-1) prepared from RuCl2{PMe2(CH2)2Si(OEt)3}3.
I (TC ) = I0/(1 ꢁ TSiH /T1%H )(eꢁT /T ꢁ eꢁT /TSiH ). [3]
C
1%H
C

SILICA HYBRID GEL CATALYSTS
291
FIG. 4. Extended X-ray absorption fine structures of representative hybrid-gel catalysts. (For other properties cf. Table 1).
containing catalyst (Ir) by Fourier transformation (modi- in hybrid gel catalysts containing Rh, Ir, Pd, and Pt are the
fied radial distribution functions) and fitting with theoreti- same as in the precursor complexes, giving additional evi-
cal amplitude and phase functions. The Fourier transforms dence for their undestructed immobilization. For the RuPh
of the EXAFS spectra of the catalysts RuPh-2, prepared catalysts (ruthenium with phenyl phosphine ligands) the
from RuCl2{PPh2(CH2)2Si(OEt)3}3, and RuMe-2, prepared predicted coordination numbers are also found within the
from RuCl2-{PMe2(CH2)2Si(OEt)3}3, are depicted in Fig. 6. accuracy of the method, confirming the proposed structure
All hybrid gels show in general only one coordination shell of the ruthenium complexes in solution and in the solid
except for Pt, which exhibits a small additional peak at matrix. Note that for these samples tendentiously negative
˚
approximately 4 A caused by weak Pt–Pt interactions or deviations from the theoretical coordination number are
multiple scattering effects within the shell of next-nearest found. In contrast to the catalysts containing phenyl phos-
neighbors. For all other catalysts no metal–metal interac- phine ligands the calculated coordination number for cata-
tion was found, thus excluding dimerization or oligomer- lyst RuMe-2, made from RuCl2{PMe2(CH2)2Si(OEt)3}3,
ization reactions during the sol-gel process.
was 1 less than in the proposed precursor structure, indi-
In one of the fits reasonable structural parameters could cating that only two phosphine ligands are bonded to each
be obtained by fitting the first shell with just one type of ruthenium center.
ligand (P or Cl or O). Best results were achieved by refin-
ing a M-P and a M-Cl shell simultaneously in a two-shell fit
Transmission electron microscopy (TEM). For a more
(see Table 3). The number of metal–phosphine interactions detailed morphological and structural characterization, the

¨
KROCHER ET AL.
292
TABLE 3
Refined Structure Parameters for the Metal-Phosphorus (M-P)
and Metal-Chlorine (M-Cl) Bonds within the First Coordination
Shell of the Immobilized Metal Complexes
M-P
M-Cl
2
2
2
2
˚
˚
˚
˚
1ꢅ /A
Catalyst
N
R/A
1ꢅ /A
N
R/A
RuPh-1
RuPh-2
RuMe-2
Ir
Rh
Pt
2.83
2.75
2.06
2.96
2.75
2.02
1.95
2.269
2.273
2.191
2.267
2.225
2.227
2.245
0.0052
0.0047
ꢁ.0035
0.0047
0.0021
0.0020
ꢁ0.0022
1.88
1.89
1.72
0.95
1.15
1.92
1.94
2.382
2.389
2.299
2.348
2.346
2.330
2.356
0.0042
0.0037
ꢁ0.0037
0.0012
ꢁ0.0005
0.002
ꢁ0.0021
Pd
Note. N = coordination number, R = bond length, 1ꢅ2 = Debye–
FIG. 5. The evaluation of the EXAFS information is shown in the
Waller factor.
example for the iridium-containing catalyst (Ir). The extracted edge infor-
mation (top) is Fourier transformed (bottom left), after which the first shell
is back-transformed (bottom right, straight line) and fitted with theoretical
amplitude and phase functions (dotted line).
porated complexes on the surface area and the pore size
distribution of the final hybrid gels. According to IUPAC
classification (49), the catalysts showed, in general, a type-
IV physisorption isotherm with a more or less pronounced
type-I isotherm character, indicating mesoporous solids
with a strong contribution of microporosity. Corresponding
t-plot constructions confirmed the presence or absence
catalysts RuPh-1, RuPh-2, RuMe-2, and RuMe-3 were in-
vestigated by TEM. All samples were completely amor-
phous, as evidenced by the representative TEM image of
sample RuPh-2 shown in Fig. 7. Neither crystalline moi-
eties of silica nor any indication for the presence of ruthe-
nium particles were seen, corroborating that ruthenium was
finely distributed within the silica matrix. Moreover, elec-
tron diffraction yielded no distinct reflections, proving the
noncrystallinity of the catalyst.
X-ray diffraction (XRD). Figure 8 shows the X-ray dif-
fraction patterns of the samples Rh, Ir, Pd, Pt, RuPh-2,
RuMe-1, and RuMe-3. All diffractograms were similar, ex-
hibiting only broad bands but no sharp reflections, corrob-
orating the amorphicity of the materials.
Nitrogen and argon physisorption. Nitrogen physisorp-
tion experiments were carried out to investigate the influ-
ence of the structure and the concentration of the incor-
FIG. 6. Fourier transforms of the EXAFS of ruthenium containing
catalystswith phenylphosphine (RuPh-2) and methylphosphine (RuMe-2)
ligands, respectively.
FIG. 7. Transmission electron micrograph of a hybrid-gel catalyst pre-
pared from RuCl2{PPh2(CH2)2Si(OEt)3}3.

SILICA HYBRID GEL CATALYSTS
293
or -H4 desorption hysteresis was observed, indicating dif-
ferent degrees of mesoporosity. From the data listed in
Table 1 it can be seen that SBET was highest for pure sil-
ica (880 m2 ꢄ gꢁ1). The surface area of the catalysts pre-
pared with a complex : TEOS ratio of 1 : 50 varied within
a range of 430 to 800 m2 ꢄ gꢁ1. For the ruthenium catalysts
the structure of the immobilized complex and the amount
of complex used had a pronounced influence on the
surface area. Higher metal loadings resulted in signifi-
cantly lower surface areas. Note that catalysts made from
RuCl2{PMe2(CH2)2Si(OEt)3}3 with a complex : TEOS ra-
tio of 1 : 20 afforded markedly different BET surface areas,
because precipitation occurred during the preparation of
gels. When the preparations were repeated, SBET could be
reproduced within ꢆ5% . The contribution of the microp-
ore surface area, determined by argon physisorption, is also
listed in Table 1 for selected samples.
FIG. 8. X-ray diffraction patterns of hybrid-gel catalysts containing
different group(VIII metals.
of microporosity. No desorption hysteresis (pure type-I
isotherm) and, therefore, predominantly microporous char-
acter of the solids was observed for catalysts Ir and Pd.
Sample Pt revealed pronounced type-H2 desorption hys-
teresis, whereas for the other samples only weak type-H2
Typical pore size distributions of the catalysts are shown
in Fig. 9. Pure silica and the platinum-containing catalyst
exhibited only mesoporosity with distinct maxima between
FIG. 9. Textural properties of sol-gel immobilized group(VIII) transition metal complexes. Differential BJH pore size distributions {dV/dlog(D)}
derived from the desorption branches of the physisorption isotherms.

¨
KROCHER ET AL.
294
3.5 and 4.8 nm. Similar maxima could also be found for uced trace amounts of trimethylamine as a by-product.
the samples Rh, RuPh-2, and RuMe-3, but in these cases Trimethylamine can be formed as a thermodynamically fa-
with a significant contribution of micropores. The other ma- vored disproportionation product from the dimethylamine
terials exhibited transition from micro- to mesoporosity, reactant or by overreduction of dmf. No by-products could
discernible from the constantly sloping curve of the BJH- be detected in the gas phase, showing that neither CO for-
plot (e.g. catalysts Ir, Pd, and RuMe-1). Catalyst RuPh-2 mation by the reverse water–gas shift reaction nor methane
showed an exceptional BJH-plot with a slightly rising curve synthesis occurred under the conditions used. At low con-
for larger pores. Thus, a combination of micro-, meso-, and versions a white precipitate was obtained after pressure
macropores is found for this catalyst. Corresponding mean release, which melted within approximately one hour un-
pore diameters of the catalysts are listed in Table 1.
der degassing. GC analysis of this product showed signals
characteristic for dimethylamine and CO2, respectively,
indicative of a carbamate compound.
Catalytic Properties of Hybrid Gels
Results of the catalytic tests are summarized in Table 4.
The data provide a qualitative ranking of the catalytic be-
havior of the different hybrid gels. Pure silica (Si) was tested
as a reference and exhibited no catalytic activity. Compar-
ing the different transition metal catalysts reveals that the
activity, expressed as turnover frequency (TOF), decreases
in the order Ru > Ir > Pt,Pd > Rh. The rhodium catalyst
(Rh) exhibited only low activity and decomposed under
reaction conditions by producing black metallic deposits
on the wall of the autoclave. Although the iridium catalyst
(Ir) changed its color from beige to white during reaction,
indicating structural alteration, its activity remained high.
Regarding the influence of the silylether ligand of the
ruthenium complex, the use of methylphosphine ligands
(RuMe) resulted in more active catalysts, as compared
to phenylphosphine ligands (RuPh), independently of the
preparation conditions. However, a different effect of metal
The hybrid gel catalysts were tested for the solvent-free
synthesis of N,N-dimethylformamide and methyl formate,
respectively, from carbon dioxide. Under the experimental
conditions applied, the CO2/H2 mixture was in the super-
critical region, whereas due to their low solubility in scCO2
(50, 51), dma (methanol/triethylamine), and the products
dmf (mf) and water were in the liquid state (17). High
swelling of the liquid dma phase by dissolved CO2 was re-
ported by Jessop et al. (17), thus providing high local con-
centrations of reactants at the surface of the catalysts, which
are in the liquid phase at the bottom of the reactor.
Synthesis of N,N-dimethylformamide. All catalysts
were found to be stable under reaction conditions, except
Rh and Ir, whose color changed during reaction. GC ana-
lysis of the liquid phase indicated the production of dmf
with 100% selectivity in all cases, except for Pt, which prod-
TABLE 4
Catalytic Properties of Sol-Gel Derived Hybrid Gels in the Synthesis of N,N-Dimethylformamide (dmf)
from Carbon Dioxide, Hydrogen, and Dimethylamine
dmf
Catalyst
Complex
Complex : TEOS
Acid
TONa
TOF/hꢁ1 b
Yield/%
Selectivity/%
Si
—
0 : 100
1 :50
1 :20
1 :50
1 :50
1 :20
1 : 20
1 : 50
1 : 50
1 :50
1 : 50
H3PO4
H3PO4
H3PO4
H3PO4
H3PO4
H3PO4
MeCO2H
MeCO2He
MeCO2He
H3PO4
0
3230
4420
13670
110800
11800
11870
2900
0
220
290
900
1860
790
800
190
35
0
25
35
94
82
84
92
23
4
0
100
100
100
100
100
100
100
100
98d
RuPh-1
RuPh-2
RuMe-1
RuMe-1c
RuMe-2
RuMe-3
Ir
Rh
Pt
Pd
RuCl2{PPh2(CH2)2Si(OEt)3}
RuCl2{PPh2(CH2)2Si(OEt)3}33
RuCl2{PMe2(CH2)2Si(OEt)3}
RuCl2{PMe2(CH2)2Si(OEt)3}3
RuCl2{PMe2(CH2)2Si(OEt)3}3
RuCl2{PMe2(CH2)2Si(OEt)3}33
IrCl{PPh2(CH2)2Si(OEt)3}3
RhCl{PPh2(CH2)2Si(OEt)3}
PtCl2{PPh2(CH2)2Si(OEt)3}23
PdCl2{PPh2(CH2)2Si(OEt)3}2
530
1490
1410
100
90
10
11
MeCO2H
100
Note. Reaction conditions: 500-ml stainless-steel autoclave; n[Me2NH] = 0.71 mol; p[H2] = 8.5 MPa; p[CO2] = 13.0 MPa; T = 373 K; t = 15 h; stirring
rate = 300 minꢁ1; n[catalyst] = 5 ꢅ 10ꢁ5 mol (expressed as amount of group(VIII) metal).
a TON = mol dmf ꢄ (mol complex)ꢁ1
.
b TOF = TON ꢄ hꢁ1
.
c
n[Me2NH] = 6.4 mol , T = 406 K, t = 60 h, stirring rate = 500 minꢁ1, n[catalyst] = 1.8 ꢄ 10ꢁ5 mol. Due to the pressure drop in the reactor, H2 and
CO2 had to be recharged regularely.
d By-product trimethylamine.
e Change of active species under reaction conditions.

SILICA HYBRID GEL CATALYSTS
295
loading on activity was observed for the two ligands. When increasing the reaction time to 64 h, the TON
Whereas the turnover frequency of the ruthenium catalyst was higher but the TOF decreased to about 50 hꢁ1. The
with phenylphosphine ligands increased with higher metal maximum in activity was found at 373 K. For lower and
loading, the opposite behavior was found for the methyl- higher temperatures the rate of the reaction decreased,
phosphine ligand. The use of either phosphoric acid or which is in good agreement with the observation of Jessop
acetic acid in the sol-gel process had no significant effect on et al. (17), who found an optimum temperature of 353 K
the catalytic performance of the ruthenium-containing hy- for the homogeneously catalyzed reaction.
brid gels. With a ruthenium catalyst containing methylphos-
The utilization of an additional base (triethylamine) was
phine ligands (RuMe-1), a maximum turnover number crucial to reach high yields of methyl formate. Without base
(TON) of 110,800 at 100% selectivity was attained under methanol conversions were low. However, a molar ratio of
specific conditions (Table 4).
methanol : base of 10 : 1 was enough to reach high activities
The color of the ruthenium catalysts after catalytic test- in methyl formate synthesis. Addition of ꢆ -alumina as an
ing was significantly brighter than prior to testing. Leach- acidic catalyst for the esterification, instead of base led only
ing of active species from the catalysts could be neglected, to low yields of methyl formate.
since the liquid product exhibited no catalytic activity in a
DISCUSSION
subsequent activity test. Moreover, using the same ruthe-
nium catalyst for three consecutive experiments resulted in
a marked increase of catalytic activity for the second run,
Structural Properties of Hybrid Gel Catalysts
which remained stable in the third experiment. This behav-
ior can presumably be attributed to the formation of an ac-
tive hydrido intermediate formed by hydrogenolysis of the
ruthenium-chlorine bond under reaction conditions. The
good stabilizing effect of the sol-gel process on homoge-
neous complexes was supported by the long lasting activity
of the tested complexes. Fresh, as well as used, catalysts con-
taining hydrido species, which were used for catalytic test-
ing after storage under air for one year, showed no loss of
activity. This contrasts the findings with homogeneous ana-
logues, where the reaction mixtures rapidly changed their
color or precipitates formed, indicating decomposition of
the complexes.
For the immobilization of sensitive transition metal com-
plexes mild gelation conditions are required to prevent de-
struction of the complexes. A modified method of Schubert
et al. (27), utilizing mild acidic conditions at room tempera-
ture was found to be most suitable for our complexes. Age-
ing at the end of the sol-gel process was necessary to obtain
stable gels, which do not undergo leaching upon washing
with water or acetone. During ageing the degree of con-
densation of the network increases by polycondensation of
part of the remaining free hydroxyl groups.
The immobilization of the metal complexes was followed
using 31P NMR spectroscopy. Because of the anisotropy,
which cannot be averaged out in the solid phase, solid state
Synthesis of methyl formate. Catalyst RuMe-1, the spectra were always less resolved than the corresponding
most active catalyst in dmf production, was also tested in liquid spectra. Moreover, the characteristic peaks were al-
the synthesis of methyl formate (Table 5). In a series of ways broader than the expected intrinsic minimum peak
nonoptimized experiments methyl formate was produced width of solid state spectra. This shift dispersion originates,
very efficiently at a selectivity of 100% . A maximum TOF if the environments for the immobilized complexes are
of 115 hꢁ1 was reached, which is more than two times quite different and disordered, as anticipated from the gela-
better than the best reported homogeneous result (17). tion procedure. For the platinum-containing catalyst the
TABLE 5
Catalytic Properties of the Hybrid-Gel RuMe-1 Derived from RuCl2{PMe2(CH2)2Si(OEt)3}3 in the Synthesis
of Methyl Formate from Carbon Dioxide, Hydrogen, and Methanol
n[Catalyst]/ꢂmola
n[Methanol]/mol
n[Triethylamine]/mol
t/h
T/K
TONb
TOF/hꢁ1 c
17.2
14.1
17.7
17.6
17.3
0.74
0.74
0.75
0.99
0.74
0.074
0.074
0.074
0.074
1.0 ꢄ 10ꢁ3 d
15
15
15
64
15
353
393
373
373
373
790
1270
1740
3180
75
53
85
115
50
5
Note. Reaction conditions: 500-ml stainless-steel autoclave; p[H2] = 8.5 MPa; p[CO2] = 13.0 MPa; stirring rate = 300 minꢁ1
.
a n[catalyst] expressed as amount of ruthenium complex.
b TON = mol methyl formate ꢄ (mol complex)ꢁ1
.
c TOF = TON ꢄ hꢁ1
.
d Additive = ꢆ -Al2O3.

¨
KROCHER ET AL.
296
peaks were narrower than in the other cases, indicating a periments should be preferred for a reliable quantification
more uniform incorporation of the complexes.
of different species.
A rough picture of the gel structure can be drawn from
the NMR data available. The gels consist of ruthenium
species and silicon species, which are incompletely con-
densed. The great part of incompletely condensed sili-
con species (Q3, Q2, T2, T1) indicates a highly disordered
structure. Large amounts of Q4 species, not detectable by
CP/MAS measurements, indicate areas of bulk-Q4-species,
more than four single bonds distant from the next proton.
All ruthenium species are equally bound to their environ-
ment due to the constant T1ꢁH value for the T3 species,
indicating a homogeneous distribution of ruthenium and
silicon species. On the contrary, the ruthenium species are
surrounded by at least three different Q4 and Q3 species,
with T1ꢁH values lower, higher, and in the same range than
for the T3 species. Higher T1ꢁH values indicate spatial re-
gions of bulk-Q4-species, related to pure amorphous silica.
This observation is in good agreement with the high amount
of invisible silica, more than four bonds distant from next
proton.
Regarding the surface area and porosity of the different
hybrid gels, the data listed in Table 1 and shown in Fig. 9
for similarly prepared samples varied over a wide range,
indicating a pronounced influence of the structures of the
immobilized group(VIII) metal complexes on the textural
properties of the gels. Possible reasons are the different size
and geometry of the complexes and, thus, a disturbance of
the network structure in different ways. A correlation be-
tween the average pore diameter, which varied in the range
1.9–5.3 nm, and the preparation conditions is not dis-
cernible in general. For the Ru containing gels lower surface
areas are found for higher complex loadings. This observa-
tion is in agreement with findings of Schubert et al. (27),
who investigated the interrelation between complex con-
centration and specific surface area for a cobalt complex
immobilized by a similar method. Possible explanations for
this behavior are (i) easier relaxation of the polymeric net-
work by the introduction of the complexes, (ii) the influence
of the steric demand of the complexes on particle growth, or
(iii) electronic influences of the alkyl chain in the silylether
ligand on its reactivity in the sol-gel process (27). The small
surface areas, found for RuMe-2 and RuMe-3, are due to
precipitation, which occurred when the acid was added
for gelation, rendering control of the gelation conditions
difficult.
The transparency of the obtained glass-like colored solids
indicated a macroscopically homogeneous distribution of
the complexes in the silica matrix. TEM investigations
of the finely powdered gels corroborated the complete
homogeneity and the amorphous structure of the materials.
EXAFS measurements proved the unchanged immobiliza-
tion of the nonruthenium precursors. For the ruthenium
catalysts containing phenyl phosphine ligands (RuPh),
the proposed penta-coordinated structure of the ligand
sphere was also confirmed. The small negative deviations
from the theoretical coordination numbers in these cata-
lysts can be attributed to minor ligand cleavage in solution
during the sol-gel process. James and Markham observed
for a similar complex that, in a 1 N solution of RuCl2(PPh3)3
in benzene, 80% of the complex is dissociated (47, 52). The
substantial deviation of 1 in the coordination sphere for
catalyst RuMe-2 from the proposed precursor structure
RuCl2{PMe2(CH2)2Si(OEt)3}3 revealed that only two
phosphine ligands are bound to each ruthenium center,
indicating the dissociation of a phosphine ligand during
immobilization (47). Based on EXAFS measurements, all
complexes in the hybrid gels are monomers. Dimerization
and oligomerization reactions of the precursors during
the sol-gel process can be excluded, since no metal–metal
interactions are found (except for Pt). In contrast to the
heterogenization of silylether metal complexes by the
sol-gel method used which results in a homogeneous
distribution of the metal moieties throughout the material,
clustering of the metal is observed for the classical sol-gel
route. Lopez et al. used RuCl3 in a sol-gel process for
the production of highly dispersed ruthenium in silica
(53). The corresponding TEM micrographs showed small
particles of ruthenium within a silica support.
The highly disordered structure of the gels was corrobo-
rated by 29Si SPE NMR measurements. The direct detection
of the 29Si nuclei enabled the reliable quantification of the
different silicon species by integration of the deconvoluted
peaks. The accuracy of this method is estimated to be about
ꢆ5% (29). However, the slow relaxation of the 29Si nuclei
resulted in long repetition times of about 200 s, making SPE
measurements very time consuming.
The alternative quantification of the different silicon
species by CP measurements was difficult due to the great
part of silica in the catalyst. In this silica, moieties of con-
densed Q4 species, silicon atoms which are more than four
single bonds distant from the next proton, make nearly no
contribution to the peak intensity, due to the low cross-
polarization efficiency (29). Deviations of more than 20%
were obtained for the Q4 species, when results of CP/MAS
measurements and SPE experiments were compared. In
Catalytic Properties of Hybrid Gel Catalysts
Synthesis of N,N-dimethylformamide. Among all sol-
gel derived hybrid materials, ruthenium-containing silica
gels showed highest activity in dmf synthesis from CO2, H2,
and dma. No correlation between surface area and cata-
the case of high amounts of silica without contribution to lytic activity of the catalysts could be observed. Catalysts
signal intensity and species with different mobility, SPE ex- RuMe-2 with a surface area of <5 m2 ꢄ gꢁ1 and RuMe-3

SILICA HYBRID GEL CATALYSTS
297
with 70 m2 ꢄ gꢁ1 exhibited comparable catalytic activity to small amounts of methyl formate were found after addition
catalyst RuMe-1 with 510 m2 ꢄ gꢁ1
.
of ꢆ -alumina as an acidic esterification catalyst, suggesting
The white solids or highly viscous liquids obtained after a more pronounced effect of the stabilization of the formic
catalytic tests at low conversions in dmf synthesis can be acid after the first reaction step than the acceleration of the
best described as Dimcarb (54), an addition compound subsequent esterification (17).
of CO2 with dimethylamine of variable stoichiometry.
Dimethylammonium N,N-dimethylcarbamate, the precise again be explained by the favorable liquid phase reaction
2 : 1 complex, has a melting range between 30 and 45ꢃC.
design. Comparable to the swelling of amines, the volume
The high reaction rates in methyl formate synthesis can
The elucidation of the phase behavior inside the reac- of the methanol phase increases at elevated pressures (17),
tor is of crucial importance to explain the high efficiency resulting in high concentrations of hydrogen in the liquid
of the described reaction design. Under similar reaction methanol/triethylamine phase at the catalysts surface.
conditions Jessop et al. (17) observed high swelling of the
liquid phase, if a supercritical phase is present. Leitner also
stated that high pressures are required for swelling of the
CONCLUSIONS
liquid phase, but the gases do not have to be in the super-
Sol-gel derived hybrid materials have been used as het-
critical state for reaching this behavior (55). This means for
erogeneous catalysts for the solvent-free synthesis of N,N-
our system that high local concentration of dma, CO2, and
dimethylformamide and methyl formate utilizing supercrit-
H2 are provided at the catalyst surface. On the contrary,
ical CO2 as both solvent and reactant. Highly active and
only small amounts of amine were found in the gas phase
stable catalysts were prepared by incorporation of specially
(17), indicating that the liquid phase can favorably be used
tailored complexes of Ru, Rh, Ir, Pd, and Pt with bifunc-
as the reaction phase. In a study with homogeneous ruthe-
tional silylether phosphine ligands within a porous silica
nium complexes we could confirm the high efficiency of this
network applying the sol-gel method.
reaction design also at subcritical conditions (19).
The ruthenium-containing hybrid gel catalysts proved to
Upon catalytic testing the color of the catalysts became
be 100% selective and most active for dmf synthesis, af-
brighter, probably indicating the formation of colorless hy-
fording turnover numbers up to 110,800 with correspond-
drido complexes. This assumption is supported by the even
ing TOFs of 1860 hꢁ1. This turnover frequency exceeds
higher activity of the catalysts in a second and third run. Hy-
those previously achieved with heterogeneous catalysts
drido complexes are known to be active catalysts in homo-
more than 600 times. The ruthenium catalysts also selec-
geneous hydrogenations (32). They are simply formed from
tively catalyze the synthesis of methyl formate.
ruthenium dichloro phosphine complexes by hydrogena-
Textural investigations indicated that all hybrid gels were
tion in the presence of organic bases (32, 56). For similar
micro-to mesoporous, except the platinum-containingcata-
ruthenium chloro phosphine complexes used in the homo-
lyst, which was predominantly mesoporous. TEM and XRD
geneous synthesis of dmf and methyl formate the mono-
analysis proved the homogeneity and noncrystallinity of the
hydrido complex could be identified as an intermediate
materials. Valuable information concerning the bulk prop-
(17, 19, 57).
erties were obtained by solid state 29Si NMR spectroscopy,
Synthesis of methyl formate. Investigation of the methyl
formate synthesis at different temperatures revealed a max-
imum TOF at 393 K. The formation of formic acid in a first
reaction step is reversible, resulting in increasing decom-
position to the reactants hydrogen and carbon dioxide at
higher temperatures. In contrast, the subsequent esterifi-
cation requires higher temperatures to proceed efficiently.
This antagonistic effect results in the observed optimum
temperature, enabling a sufficiently effective esterification
of formic acid to methyl formate at relatively slow decom-
position rates. The hybrid catalysts seem to meet these re-
quirements, since the optimum temperature is shifted to
higher values.
revealing significant parts of incompletely condensed sili-
con species, indicative for a highly disordered structure of
the gels. EXAFS investigations indicated that the transi-
tion metal complexes were immobilized as isolated entities.
Nondestructive incorporation of the silylether complexes
was corroborated by 31P NMR and EXAFS.
ACKNOWLEDGMENT
Financial support of this work by the Bundesamt fu¨r Energiewirtschaft
(Project 21180) is gratefully acknowledged.
The addition of triethylamine as base is crucial for this
reaction to obtain substantial amounts of methyl formate.
10 mol% of base promoted the reaction sufficiently. The
base acts as a thermodynamic sink in the first step of the
reaction, which stabilizes the intermediate formic acid un-
der the severe conditions inside the reactor (58, 59). Only
REFERENCES
1. Behr, A., in “Catalysis in C1 Chemistry” (W. Keim, Ed.), p. 169.
D. Reidel, Dordrecht, 1983.
2. Behr, A., Angew. Chem., Int. Ed. Engl. 27, 661 (1988).
3. Chinchen, G. C., Denny, P. J., Jennings, J. R., Spencer, M. S., and
Waugh, K. C., Appl. Catal. 36, 1 (1988).

¨
KROCHER ET AL.
298
4. Halmann, M., “Chemical Fixation of Carbon Dioxide: Methods for
30. Parshall, G. W., Inorg. Synth. 11, 157 (1968).
Recycling CO2 into Useful Products,” CRC Press, Boca Raton, FL, 31. Niebergall, H., Makromol. Chem. 52, 218 (1962).
1993.
32. Harmon, R. E., Gupta, S. K., and Brown, D. J., Chem. Rev. 73, 21
5. Baiker, A., and Koeppel, R. A., in “Proceedings, International Con-
ference on Carbon Dioxide Utilization, Bari, Italy, 1993,” p. 295.
6. Jessop, P. G., Ikariya, T., and Noyori, R., Chem. Rev. 95, 259 (1995).
7. Leitner, W., Angew. Chem., Int. Ed. Engl. 34, 2207 (1995).
8. Haynes, P., Slaugh, L. H., and Kohnle, J. F., Tetrahedron Lett., 365
(1970).
9. Inoue, Y., Sasaki, Y., and Hashimoto, H., J. Chem. Soc., Chem. Comm.,
718 (1975).
10. Kiso, Y., and Saeki, K., Japan, Kokai 77 36,617, 1977.
11. Kudo, K., Phala, H., Sugita, N., and Takezaki, Y., Chem. Lett., 1495
(1977).
(1973).
33. Schaefer, J., Stejskal, E. O., and Buchdahl, R., Macromolecules 10, 384
(1977).
34. Barrett, E. P., Joyner, L. G., and Halenda, P. P., J. Am. Chem. Soc. 73,
37 (1951).
35. Broekoff, J. C. P., Stud. Surf. Sci. Catal. 3, 663 (1979).
36. Lippens, B. C., and DeBoer, J. H., J. Catal. 4, 319 (1965).
37. Dubinin, M., and Astakhov, V., Adv. Chem. Ser. 102, 69 (1971).
38. Frahm, R., Nucl. Instrum. Meth. Phys. Res. A 270, 578 (1988).
39. Ressler, T., J. de Physique IV C 2, 269 (1997).
40. Rehr, J. J., Mustre de Leon, J., Zabinski, S. I., and Albers, R. C., J. Am.
Chem. Soc. 113, 5135 (1991).
41. Kro¨cher, O., Ko¨ppel, R. A., and Baiker, A., J. Mol. Catal., in press.
42. Bemi, L., Clark, H. C., Davies, J. A., Fyfe, C. A., and Wasylishen, R. E.,
J. Am. Chem. Soc. 104, 438 (1982).
12. Phala, H., Kudo, K., Mori, S., and Sugita, N., Bull. Inst. Res., Kyoto
Univ. 59, 88 (1981).
13. Phala, H., Kudo, K., Mori, S., and Sugita, N., Bull. Inst. Res., Kyoto
Univ. 63, 63 (1985).
14. Schreiner, S., Yu, J. Y., and Vaska, L., Inorg. Chim. Acta 147, 139
(1988).
15. Jessop, P. G., Ikariya, T., and Noyori, R., J. Am. Chem. Soc. 116, 8851
(1994).
43. Pregosin, P. S., and Kunz, R. W., “31P and 13C NMR of Transition
Metal Phosphine Complexes.” Springer-Verlag, Berlin, 1979.
44. Grim, S. O., Keiter, R. L., and McFarlane, W., Inorg. Chem. 6, 1133
(1967).
16. Jessop, P. G., Hsiao, Y., Ikariya, T., and Noyori, R., J. Chem. Soc.,
Chem. Commun., 707 (1995).
17. Jessop, P. G., Hsiao, Y., Ikariya, T., and Noyori, R., J. Am. Chem. Soc.
118, 344 (1996).
18. Kro¨cher, O., Ko¨ppel, R. A., and Baiker, A., J. Chem. Soc., Chem.
Commun., 1497 (1996).
19. Kro¨cher, O., Ko¨ppel, R. A., and Baiker, A., J. Chem. Soc., Chem.
Commun., 453 (1997).
20. Kro¨cher, O., Ko¨ppel, R. A., and Baiker, A., Chimia 51, 48 (1997).
21. Jessop, P. G., Ikariya, T., and Noyori, R., Nature 368, 231 (1994).
22. Kro¨cher, O., Ko¨ppel, R. A., and Baiker, A., in “Process Technology
Proceedings” (Ph. Rudolf von Rohr and Ch. Trepp, Eds.), Vol. 12,
p. 91. Elsevier, Amsterdam, 1996.
23. Murrell, L. L., in “Advanced Materials in Catalysis” (J. J. Burton and
R. L. Garten, Eds.), p. 235. Academic Press, New York, 1977.
24. Whitehurst, D. D., Chemtech., 44 (1980).
25. Deschler, U., Kleinschmitt, P., and Panster, P., Angew. Chem. 98, 237
(1986).
26. Schubert, U., New J. Chem. 18, 1049 (1994).
27. Schubert, U., Rose, K., and Schmidt, H., J. Non-Cryst. Solids 105, 165
(1988).
28. Lindner, E., Bader, A., and Mayer, H. A., Z. Anorg. Allg. Chem. 598,
235 (1991).
45. Grim, S. O., and Keiter, R. L., Inorg. Chim. Acta 4, 56 (1970).
46. Komoroski, R. A., Magistro, A. J., and Nicholas, P. P., Inorg. Chem.
25, 3917 (1986).
47. Armit, P. W., Boyd, A. S. F., and Stephenson, T. A., J. Chem. Soc.,
Dalton Trans., 1663 (1975).
48. Voelkel, R., Angew. Chem. 100, 1525 (1988).
49. Sing, K. S. W., Everett, D. H., Haul, R. A. W., Moscou, L., Pierotti,
R. A., Rouque´rol, J., and Siemieniewska, T., Pure and Appl. Chem.
57, 603 (1985).
50. Francis, A. W., J. Phys. Chem. 58, 1099 (1954).
51. Bartle, K. D., Clifford, A. A., Jafar, S. A., and Shilstone, G. F., J. Phys.
Chem. Ref. Data 20, 728 (1991).
52. James, B. R., and Markham, L. D., Inorg. Chem. 13, 97 (1974).
53. Lopez, T., Herrera, L., Gomez, R., Zou, W., Robinson, K., and
Gonzalez, R. D., J. Catal. 136, 621 (1992).
54. Schroth, W., Andersch, J., Scha¨dler, H.-D., and Spitzner, R., Chem.
Ztg. 113, 261 (1989).
55. Leitner, W., Dinjus, E., and Gassner, F., J. Organomet. Chem. 475, 257
(1994).
56. James, B. R., Inorg. Chim. Acta, 73 (1970).
57. Whittlesey, M. K., Perutz, R. N., and Moore, M. H., Organometallics
15, 5166 (1996).
58. Graf, E., and Leitner, W., J. Chem. Soc., Chem. Commun., 623 (1992).
29. Lindner, E., Kemmler, M., Mayer, H. A., and Wegner, P., J. Am. 59. Gassner, F., and Leitner, W., J. Chem. Soc., Chem. Commun., 1465
Chem. Soc. 116, 348 (1994). (1993).