Solventless synthesis of propylene carbonate catalysed by chromium-salen complexes: Bridging homogeneous and heterogeneous catalysis

Article abstract of DOI:10.1016/j.molcata.2005.08.004

Various homogeneous chromium-salen complexes were synthesised and tested concerning their catalytic performance in the synthesis of propylene carbonate from carbon dioxide and propylene oxide. The most active complexes were immobilised on a silica support and tested in the same reaction. Different immobilisation methods were applied including anchoring of the complex through coordination with the metal, and covalent bonding via the ligand. Rates up to 300 mol_product molCr-1 h^-1 at 98% selectivity were achieved with the heterogeneous catalysts without the use of any additional solvent or co-catalyst. The coordinatively-bound complexes showed low stability and a very strong deactivation during reuse, whereas the corresponding covalently-bound complexes hardly showed any chromium leaching during catalyst recycling experiments. The catalysts were characterised by means of X-ray photoelectron spectroscopy (XPS), thermal analysis, inductively-coupled plasma optical emission spectrometry (ICP-OES) and diffuse reflectance IR spectroscopy (DRIFTS) in as-prepared state and after use.

Full text of DOI:10.1016/j.molcata.2005.08.004

Journal of Molecular Catalysis A: Chemical 242 (2005) 32–39
Solventless synthesis of propylene carbonate catalysed by
chromium–salen complexes: Bridging homogeneous
and heterogeneous catalysis
Michael Ramin, Fabian Jutz, Jan-Dierk Grunwaldt, Alfons Baiker∗
Department of Chemistry and Applied Biosciences, Swiss Federal Institute of Technology, ETH-Ho¨nggerberg, CH-8093 Zu¨rich, Switzerland
Received 6 June 2005; received in revised form 2 August 2005; accepted 2 August 2005
Available online 6 September 2005
Abstract
Various homogeneous chromium–salen complexes were synthesised and tested concerning their catalytic performance in the synthesis of
propylene carbonate from carbon dioxide and propylene oxide. The most active complexes were immobilised on a silica support and tested
in the same reaction. Different immobilisation methods were applied including anchoring of the complex through coordination with the
metal, and covalent bonding via the ligand. Rates up to 300 molproduct mol−Cr1 h−1 at 98% selectivity were achieved with the heterogeneous
catalysts without the use of any additional solvent or co-catalyst. The coordinatively-bound complexes showed low stability and a very strong
deactivation during reuse, whereas the corresponding covalently-bound complexes hardly showed any chromium leaching during catalyst
recycling experiments. The catalysts were characterised by means of X-ray photoelectron spectroscopy (XPS), thermal analysis, inductively-
coupled plasma optical emission spectrometry (ICP-OES) and diffuse reflectance IR spectroscopy (DRIFTS) in as-prepared state and after
use.
© 2005 Elsevier B.V. All rights reserved.
Keywords: Carbon dioxide fixation; Propylene carbonate; Propylene oxide; Carbon dioxide; Chromium–salen complexes; Immobilisation; Catalyst deactivation;
Infrared spectroscopy; Thermal analysis; XPS
1. Introduction
dioxide fixation [21–23]. Due to the use of chelate com-
plexes with high complexation ability, the immobilisation
of these complexes represents an attractive strategy towards
heterogeneous catalysts with improved stability. Different
immobilisation strategies have been reported in literature
[24], among them is coordination of the metal centre to
a modified support surface or covalently-bound inorganic
complexes.
In this work we aimed at the heterogeneous catalytic syn-
thesis of propylene carbonate in a process without the use of
co-solvents. In the first step we searched for suitable homo-
geneous catalysts that were also sufficiently active in the
absence of co-solvents. By variation of the structure of salen
ligands of the corresponding chromium–salen complexes, the
influence of these ligands on the catalytic behaviour was
investigated. In a further step the most promising homoge-
neous complexes were grafted onto a silica support using
different immobilisation methods. Finally, the resulting het-
The synthesis of propylene carbonate by the use of carbon
dioxide as C1-building block (Scheme 1) is a step towards
greener production of propylene carbonate. Toxic and haz-
ardous reactants, such as phosgene and carbon monoxide
[1–4] can be substituted by carbon dioxide which simulta-
neously serves as a solvent. Various homogeneous catalysts
were reported for this reaction [5–14] usually carried out in
the presence of a solvent. Heterogeneous catalysis would
be desirable due to easier product separation and because
the process can be simplified [15]. However, the activity
and stability of heterogeneous catalysts is still not satis-
fying [16–20]. Homogeneous chromium–salen complexes
were reported as active homogeneous catalysts for carbon
∗
Corresponding author. Tel.: +41 1 632 31 53; fax: +41 1 632 11 63.
E-mail address: baiker@chem.ethz.ch (A. Baiker).
1381-1169/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2005.08.004

M. Ramin et al. / Journal of Molecular Catalysis A: Chemical 242 (2005) 32–39
33
was added. The mixture was heated to 100 ◦C for 8 h, after-
wards quenched in cold water (75 ml), and a pH 2 was
adjusted with HCl (2 N). The water phase was extracted
three times with diethyl ether (20 ml), and finally the organic
phases were united. After removal of the solvent, the remain-
ing dark yellow oil was purified by chromatography (ethyl
acetate:hexane = 1:9). Yield was 11.67 g (28% based on 2-
Scheme 1. Synthesis of propylene carbonate from propylene oxide and car-
bon dioxide.
1
allylphenol). H NMR (500 MHz, CDCl3, 300 K): δ = 3.4
erogeneous catalysts were compared with respect to their
catalytic performance and their reusability.
(d, 2H, CH2CH CH2), 5.15 (m, 2H, CH CH2), 5.90–6.15
(m, 1H, CH CH2), 6.90–7.00 (m, 1H, ArH), 7.38–7.48 (m,
2H, ArH), 9.85 (s, 1H, CHO), 11.32 (s, 1H, OH); 13C NMR
(500 MHz, CDCl3, 300 K): 33.0, 116.5, 119.8, 131.2, 136.0,
137.5, 196.6.
2. Experimental
2.1. Synthesis of the homogeneous chromium–salen
complexes
2.2. Syntheses of the heterogeneous catalysts
2.2.1. Synthesis of catalyst 9
In general, the chemicals were used without any further
purification, and the liquids were of spectroscopic grade. All
products were analysed by NMR and elemental analysis.
According to Scheme 2, the syntheses of the complexes
1–8 (Scheme 3) were performed similarly as reported in
[25,26]. All reactants for the syntheses were commercially
available except 3-allylsalicylaldehyde, an essential reactant
of complex 8. It was synthesised as described in [27]. The
yield of the different supported complexes ranged from 42 to
86%. In the last step, chromium dichloride was added to the
ligand and oxidised in the presence of air.
(a) Modification of the silica
Silica (16.0 g; Fluka Silica Gel 60; BET surface area,
460 m2 g−1; mean pore size, 5.1 nm) was filled into
a three-neck flask. Under argon, triethylamine (1 ml)
in chloroform (100 ml) was given to the silica. After
30 min of stirring, 3-aminopropyltriethoxysilane (8.89 g,
40.2 mmol) in chloroform (100 ml) was slowly dropped
to the suspension. Afterwards, the mixture was refluxed
for 4 h at 60 ◦C. The silica was cleansed in Soxhlet
extractor with dichloromethane (200 ml) and dried under
vacuum. Yield of the modified silica denoted in the fol-
lowing as “silica–NH2” was 17.36 g.
2.1.1. Synthesis of 3-allylsalicylaldehyde
Tributylamine (4.35 g, 80 mmol) and SnCl4 (4.125 g,
20 mmol) were given to a solution of 2-allylphenol (25.05 g,
187 mmol) in toluene (500 ml). After stirring for 20 min at
ambient temperature, paraformaldehyde (9.05 g, 0.4 mmol)
(b) Anchoring of the Cr–salen complex
Dichloromethane (50 ml) was added to silica–NH2
(2.16 g) in a two-neck flask under argon. After addi-
tion of complex 4 (0.92 g, 1.8 mmol), the suspension was
Scheme 2. General route applied for the synthesis of the different chromium–salen complexes.
Scheme 3. Overview of the chromium–salen complexes used in homogeneous catalysis.

34
M. Ramin et al. / Journal of Molecular Catalysis A: Chemical 242 (2005) 32–39
refluxed for 3 h at 40 ◦C and stirred overnight at ambient
temperature. Yield was 2.48 g.
filtrated and dried under vacuum. Yield was 9.3 g of a
green powder.
(b) Variant 11b: immobilisation of the Cr complex
Complex 8 (3.00 g, 5.99 mmol) was dissolved in
toluene (50 ml) and propylene carbonate (2 ml). After
heating to 40 ◦C, triethoxysilane (4 ml) was added, fol-
lowed by a small amount of H2PtCl6. The solution
became maroon coloured. The mixture was stirred for
4 h at 40 ◦C and after cooling down to ambient tempera-
ture, stirred overnight.
2.2.2. Synthesis of catalyst 10
(a) Modification of the silica
Under argon, triethylamine (1 ml) was added to sil-
ica (15.5 g) and toluene (150 ml) in a three-neck flask.
Then 3-chloropropyltriethoxysilane (9.00 g, 37 mmol) in
toluene (15 ml) was slowly dropped to the mixture. The
slurry was heated for 4 h at 80 ◦C. Soxhlet extraction with
a mixture of ether and dichloromethane (1:1, 200 ml) was
carried out for 3 h. Yield was 20.64 g of white silica mod-
ified powder denoted “silica–Cl”.
In parallel, dried silica (5.81 g) was suspended in puri-
fied toluene (75 ml) and activated with triethylamine
(1 ml). After 1 h of stirring, the maroon-coloured solu-
tion was dropped into the suspension with the help of
a syringe. The mixture was stirred overnight. The yel-
low green catalyst was filtrated and washed with diethyl
ether. Yield was 6.83 g.
(b) Anchoring of the ligand via the aminopart
N,Nꢀ-Bis(2-hydroxybenzylidene)-3-aza-1,5-
pentanediamine dissolved in some toluene was
dropped to a two-neck flask filled with silica–Cl
(10.01 g) and toluene (180 ml). Thereafter, this mixture
was refluxed for 8 h at 110 ◦C. The yellow silica material
obtained was cleansed with Soxhlet extractor with
dichloromethane (200 ml) for 3 h until the solvent was
colourless. After drying under vacuum, ligand-modified
silica was obtained as a yellow powder (7.90 g).
(c) Complexation of the immobilised ligand
2.3. Carbon dioxide insertion reaction
The catalytic reactions were performed in a 250 ml
high pressure MEDIMEX HPM-P stainless-steel autoclave
equipped with an active heating and cooling system. The
quantity of carbon dioxide was measured by a balance (type
METTLER Toledo IP5 Multirange) on which the whole reac-
torsystemwasplaced. Thetemperatureprofilewascontrolled
by a EUROTHERM 900 EPC thermostat, the stirring was
done by a motor of type EMOD EEDF 56L/2A equipped with
CrCl2 (1.175 g, 9.61 mmol) was added to the immo-
bilised ligand (7.00 g) in toluene (150 ml). The sus-
pension was stirred for 4 h at ambient tempera-
ture under argon. Afterwards, silica was transferred
into Soxhlet extractor and washed for 2.5 h with
dichloromethane (200 ml). The silica was filtrated and
driedundervacuum. Yieldwas7.21 gofagreenishbrown
powder.
a magnetic-coupled gas stirrer. Stirring rate was 1000 min−1
.
Propylene oxide (FLUKA, spectroscopic grade) and the cat-
alyst were poured into the reactor, and the reactor was once
flushed with carbon dioxide (PANGAS, 99.995%). A defined
amount of carbon dioxide was dosed into the reactor with
the help of a compressor (NWA PM-101), and heating and
stirring were started. The reactor was heated to 140 ◦C for
30 min. After a certain reaction time the reactor was cooled
down to room temperature for 40 min. Afterwards, CO2 was
released by opening the outlet valve. This decompression
was carried out slowly over a 30 min period. To the reaction
mixture tert-butylbenzene was added as an internal stan-
dard, and the compounds were analysed by a gas chromato-
graph (HP-6890) equipped with a HP-FFAP capillary column
(30 m × 0.32 mm × 0.25 ␮m) and a flame ionization detec-
tor (FID). Products were identified by gas chromatography
(HP-6890) coupled with mass selective detection (HP-5973)
and reference chemicals. As co-products, only propylene and
dipropylene glycols were found.
2.2.3. Synthesis of catalyst 11
Two different preparation routes were used to synthesise
catalyst 11 depending on the sequence of the reaction steps.
(a) Variant 11a: immobilisation of the free ligand
Firstly, N,Nꢀ-Bis(2-hydroxy-3-allylbenzylidene)-1,2-
diphenylethylene-1,2-diamine (2.83 g, 5.65 mmol) was
dissolvedinamixtureofcyclohexane(10 ml)andtoluene
(10 ml). Then triethoxysilane (1.64 g, 10 mmol, 2 equiv-
alent) was added and the mixture was heated to 40 ◦C.
H2PtCl6 was previously dried at 150 ◦C for 4 h, and
finally a bit of H2PtCl6 was dissolved in propylene car-
bonate (20 ml). This catalytic solution was dropped to
the reaction mixture at 40 ◦C. After 1 h of stirring, dried
silica (10.9 g) was added, and the suspension was stirred
overnight at ambient temperature. The silica was sepa-
rated and washed with diethyl ether. 12.5 g of brownish
silica was obtained. Secondly, CrCl2 (0.75 g, 6.14 mmol)
was added to a suspension of the modified silica (10.00 g)
in tetrahydrofuran (100 ml). The suspension was stirred
for 4 h at ambient temperature under argon. Finally, the
silica was transferred into Soxhlet extractor and washed
for 2 h with dichloromethane (200 ml). The silica was
For the leaching and reuse experiments, the reaction mix-
ture was filtrated, and the catalyst washed, dried and reused
in the same way as described above.
2.3.1. Safety note
The experiments described in this paper involve the use
of high pressure and require equipment with an appropriate
pressure rating.

M. Ramin et al. / Journal of Molecular Catalysis A: Chemical 242 (2005) 32–39
35
Table 1
2.4. Analytical and physicochemical characterisation
Catalytic activity of the different catalysts (Schemes 3 and 4) in the reaction
of propylene oxide and carbon dioxide
The carbon content of the grafted ligands was analysed by
thermal analysis (TA). The samples were heated to 800 ◦C in
a stream of 20 vol.% oxygen in helium. A mass spectrometer
was used to detect the formation of carbon dioxide and water
(traces m/e = 44 and 18, respectively). When the combustion
finished, a defined pulse of carbon dioxide was injected into
the helium gas stream to calibrate the area of the m/e = 44
peak. This calibration allowed determining the carbon con-
tent of the sample.
Surface analysis of the immobilised catalysts (9–11) by
XPS was performed on a Leybold Heraeus LHS11 MCD
instrument using Mg K␣ (1253.6 eV) radiation. The appara-
tus is described in detail in [28]. The sample was pressed into
a sample holder, evacuated in a load lock to 10−6 mbar and
transferred to the analysis chamber (typical pressure <10−9
mbar). The peaks were energy-shifted to the binding energy
of the C(1s) peak to correct for the charging of the mate-
rial. The surface composition of the catalysts was determined
from the peak areas of C(1s), O(1s), N(1s), Si(2p), Cl(2p) and
Cr(2p) which were computed after subtraction of the Shirley
type background by empirically-derived cross-section factors
[29]. The relative error of the analysis was 5%.
Catalyst
Yielda (%)
TOFb (h−1
)
PO/catalyst ratio
1
2
3
4
5
6
7
8
9
10
11a
11b
a
26.6
47.3
47.1
62.0
17.9
24.4
18.4
12.3
41.0
29.7
0.9
90
150
150
170
60
90
60
50
1000
1000
1000
900
1000
1100
1000
1100
2500
2000
2400
2400
330
200
7
9.2
70
Reaction was carried out in 10 ml (140 mmol) of propylene oxide, cata-
lyst, carbon dioxide (480–550 mmol, 3.4–3.9 equivalent), reaction time was
3 h at 140 ◦C. Yield of propylene carbonate was based on propylene oxide,
the selectivity was always higher than 98%.
b
Turnover frequency [molproduct (molCr on catalyst h)−1].
diphenylethylenediamine (Scheme 3, catalyst 4), but the cat-
alytic activity towards propylene carbonate (Scheme 1) in
terms of the turnover frequency (TOF, molproduct mol−Cr1 h−1
)
of the other substituted diamines (catalysts 2 and 3) was quite
similar (150 h−1versus 170 h−1). Note that we assumed for
the calculation of the TOF that all Cr-complexes were active.
This is certainly a conservative estimate of the TOF since
some of the complexes may not contribute to the reaction
due to possible agglomeration or incomplete dissolution. The
higher conversion using catalyst 4 could be due to the higher
solubility of the aromatic rings in carbon dioxide.
Hence, diphenyldiaminoethane was used for further test-
ing of the catalysts prepared from different hydroxyben-
zaldehydes. As Table 1 shows that the alkyl chains in ortho
position to the hydroxy group decreased the activity due to
steric reasons, either by blocking access to the active cen-
tre or inhibiting the transition state. An alkyl chain in meta
position (catalyst 6) halved the activity, alkyl chains in ortho
position even decreased the activity to one-third of the non-
substituted complex. Catalyst 7 showed no further drop in the
activity compared to catalyst 5. So, additional substitution in
para position had no negative effect on the activity of the
complex. The most active catalyst prepared from hydroxy-
benzaldehyde was the catalyst derived from the unsubstituted
hydroxybenzaldehyde itself.
The homogeneous chromium–salen catalysts showed
good catalytic performance reaching a TOF of 170 h−1 at
a conversion of 70% and a selectivity higher than 98% in
the cycloaddition of CO2 to propylene oxide. It should be
stressed that this performance has been achieved without
any additional solvent or co-catalysts as applied in [21–23],
which has distinct advantages because the reactant can eas-
ily be separated from the product. The variation of the
diamine compound (catalysts 1–4) had a strong influence in
line with the results of Paddock and Nguyen [21], although
they used a Lewis-base and an additional solvent. In their
DRIFT spectra were recorded at room temperature in
a BRUKER Equinox 55 spectrometer, accumulating 100
scans at a resolution of 4 cm−1; the measured range was
4000–400 cm−1. The solid samples were diluted in potas-
sium bromide.
The chromium content was measured by inductively cou-
pled plasma optical emission spectroscopy (ICP-OES), per-
formed at ALAB AG (Urdorf, Switzerland).
3. Results and discussion
3.1. Syntheses of the free complexes
The preparation of the salen ligands and the corresponding
chromium complexes according to Scheme 2 was straightfor-
ward and thus, successful as described in Section 2. Partly,
the yields (42–86%) were somewhat lower than expected but
the main goal of this work was to obtain pure catalyst with
reasonable effort. Hence, intensive work-up of the complexes
was omitted at this stage. NMR of the ligands and element
analysis of the complexes confirmed the desired structure.
Strong coordination of solvent molecules to the chromium
centre usually hampered the removal of these molecules by
high vacuum, so the element analysis showed a slight differ-
ence in the calculated and measured values. This behaviour
is also reported in literature for similar compounds [21].
3.2. Catalytic activity of the homogeneous catalysts
In the first step, we varied the diamine part of the
ligand, and the most active component was aromatic

36
M. Ramin et al. / Journal of Molecular Catalysis A: Chemical 242 (2005) 32–39
Scheme 4. Immobilised chromium–salen complexes: 9 coordinated complex; 10 covalently-bound complex 3; 11 covalently-bound complex 8.
experiments also the diphenyl compound showed the high-
est activity, the difference to the cyclohexane compound
was however much higher than the one we found in this
work.
bulk silica (−58 ppm) and the rest by three oxygen atoms
(−66 ppm). Single bound silicon was not found.
Immobilisation via the amino part of the salen ligand could
be easiest achieved with complex 3 by modification of the
amino group and using hydroxybenzaldehyde as an alde-
hyde component (see Section 2). This led to one covalent
bond (catalyst 10). In contrast, modification of the aldehyde
part resulted in the immobilisation via two covalent bonds on
the silica surface (catalyst 11). In this case, complex 8 was
chosen, because the synthesis with a vinyl group in ortho
position is much simpler than the one in para position, and
the interest in the heterogeneous catalysts was triggered by
their stability.
The choice of the benzaldehyde component was also
important, though the substitution position seemed to have
a stronger influence on the activity than the structure of the
side chain itself. Substitution in ortho position of the hydroxy
group (catalysts 5, 7, 8) hampered the reaction much more
than substitution in meta position (catalyst 6). The influence
of a group in ortho position is much bigger than a corre-
sponding substitution in para position, though the double-
substituted catalyst 7 showed no additional decrease in the
activity compared to catalyst 5. Similar results were achieved
by Lu et al. [30] for cobalt salen complexes for the synthesis
of ethylene carbonate in the presence of additional solvents.
The additional group in para position also lowered the activ-
ity in this study by 25%. A group in ortho position decreased
the yield by 65% (catalysts 4 and 5). In conclusion, the impact
of substitution seems to be more related to steric factors than
to the solubility or the electronic properties of the different
ligands. In contrast, Darensbourg et al. [31] found an oppo-
site effect for the influence of groups on the phenolic ligand.
They used a different substrate and an additional co-catalyst,
so the solubility of the salen complexes was probably more
important than steric factors.
For characterisingthe immobilised catalysts, DRIFT spec-
tra were recorded (Fig. 1). The bands of the C N imine
vibration appeared at 1636 and 1613 cm−1 for the immo-
bilised ligand of catalyst 10 without chromium in accordance
with literature [26]. After addition of the chromium, the sig-
nals were shifted to 1619 and 1598 cm−1 [32]. The intensive
bands of catalyst 9 appeared at higher wavenumbers (1628
and 1602 cm−1); the bands of catalysts 11a and 11b appeared
at significantly lower wavenumbers. Catalyst 11b showed
bands at about 1620 and 1592 cm−1, whereas catalyst 11a
only showed a very weak, but broad signal at 1630 cm−1
.
This indicates a small total amount of C N bonds and in
addition, that both chromium–salen complexes and uncom-
plexed ligands are present. Obviously, the structure of the
salen complex was destroyed during the synthesis of the cat-
alyst. Hence, chromium was not coordinated to the grafted
ligand. The spectra of catalysts 9, 10 and 11b showed all
the typical shifted bands of the C N group. Consequently,
in this case the chromium was well immobilised in an intact
anchored complex. Additionally, with both catalysts 11a and
11b weak bands appeared between 1700 and 1670 cm−1
together with bands over 3000 cm−1 due to small amounts
of residual vinyl side chains.
TA-MS measurements were conducted to gain informa-
tion on the ligand loading for the immobilised catalysts
by determination of the carbonaceous compounds (Fig. 2).
Note that only the chromium-free immobilised ligands were
examined to avoid contamination of the TA equipment. The
3.3. Syntheses and characterisation of the
heterogeneous catalysts
Considering the results from the homogeneous catalysts,
corresponding salen complexes were immobilised using dif-
ferent grafting methods (Scheme 4). Catalyst 9 was prepared
by anchoring the metal to a modified silica surface by coor-
dination of the Cr3+-centre to an amine group. As there is no
special requirement for the complex structure, the most active
complex 4 was chosen. The modification of the silica surface
by reaction with different ethoxysilanes is well known in lit-
erature, and preliminary tests with 29Si NMR measurements
showed good feasibility of this synthesis. The NMR spectra
mainly showed silicon bound by two oxygen atoms to the

M. Ramin et al. / Journal of Molecular Catalysis A: Chemical 242 (2005) 32–39
37
Table 2
Surface analysis of the heterogeneous catalysts by XPS
Element (%)
9
10
11a
11b
C(1s)
O(1s)
N(1s)
Si(2p)
Cl(2p)
Cr(2p)
13.0
68.0
2.3
14.8
0.5
10.7
69.6
1.1
15.7
1.2
17.8
63.8
2.1
13.3
1.4
8.7
73.8
0.3
16.3
0.2
1.4
1.7
1.7
0.8
sterically-demanding ligand of catalyst 11a showed a carbon
loading of 1.0 mg C/g catalyst. The combustion of carbon
occurred in a narrow temperature range with a maximum at
478 ◦C. This is typical for species that were placed on the
same part of the surface. The small aminopropyl anchor of
catalyst 9 decomposed in a broad temperature range with
two different narrow maxima at 308 and 390 ◦C. The loading
(2.2 mg C/g) was higher than for the sterically-demanding
complex 11a probably because the molecule is smaller and
immobilisation is sterically less hindered. Finally, the high-
est carbon loading of the silica surface showed complex 10
(3.1 mg C/g), due to the sum of immobilised complex 3 and 3-
chloropropyltriethoxysilan. The curve of the mass loss of car-
bon as function of temperature looked like the superimposed
curves of the former two experiments. This could be traced
to the method of catalyst preparation. First, the silica surface
was modified by grafting of 3-chloropropyltriethoxysilane.
Afterwards, complex 3 reacted partly with the already immo-
bilisedprecursor. ThustheresultsoftheTA-MSindicatedthat
not all grafted precursors reacted with the salen complex.
The highest bulk chromium content (1.5%) quantified
by ICP-OES was found for catalyst 10. Catalyst 9 and the
two variants of catalyst 11 showed similar metal content
(about 0.6%). In contrast, the chromium surface concentra-
tion of catalyst 11a determined by XPS (Table 2) was higher
than the surface concentration of catalyst 11b. This supports
the observation from IR measurements that in catalyst 11a
chromium was not found as Cr–salen complex on the surface,
because XPS is surface sensitive and thus more sensitive for
uncomplexed chromium (in case of catalyst 11b the bulky
salen ligands can shield the chromium). XPS further showed
that in catalyst 11a both Cr and Cl are adsorbed on the sur-
face. As expected, catalyst 9 – grafted on a silica surface with
aminopropyl anchors – had the highest nitrogen content on
the surface. Also catalyst 11a showed a high nitrogen content,
whereas the nitrogen content of catalyst 11b was very low.
Thisimpliedahigherligandloadingofcatalyst11acompared
to 11b, but the ligands of catalyst 11a did not coordinate to
any chromium centre.
Fig. 1. DRIFT spectra of the immobilised catalysts (top). The labels of
the catalysts correspond to those indicated in Scheme 4. Different states of
catalyst 10 (bottom): (a) after one reaction cycle; (b) catalyst 10 as prepared;
and (c) immobilised salen ligand without chromium. The dotted line serves
as a guide for the eyes for better comparability.
3.4. Catalytic activity of the heterogeneous catalysts
Fig. 2. TA-MS measurements of (a) aminopropyl grafted on
silica—precursor of catalyst 9; (b) immobilised ligand of catalyst 10
without chromium; and (c) grafted ligand of catalyst 11a without
chromium. The samples were heated in 20 vol.% O2 in He. The solid black
lines are the signal of m/e = 44, corresponding to CO2, the grey dotted lines
are the signal of m/e = 18, due to H2O formation.
The activity in terms of TOF of the immobilised cata-
lysts (9, 10) for propylene carbonate formation (Scheme 1)
was higher than that of the free complexes. One reason for
this behaviour may be the site isolation of the immobilised

38
M. Ramin et al. / Journal of Molecular Catalysis A: Chemical 242 (2005) 32–39
complexes, whereas the homogeneous complexes may partly
agglomerate during reaction. Note that also the concentration
of the Cr-complexes in the reaction mixture was higher in
case of the homogeneous catalysts (Table 1). Homogeneous
catalysts with comparable catalytic activity were reported in
literature to be more active only in the presence of additional
solvents [21].
The simplest approach for anchoring the active complex
to a support was applied for the preparation of catalyst 9. The
fresh catalyst showed a very high TOF of 330 h−1. Also the
activity of catalyst 10 is high in comparison to the homoge-
neous catalysts. Catalyst 11a was nearly inactive, whereas
the activity of catalyst 11b prepared by a different procedure
was higher by one order of magnitude. This is perfectly in line
with the structural analysis of the complexes which showed
that only in the procedure where the complete complex was
grafted, the salen complex was intact in the resulting het-
erogeneous catalyst (catalyst 11b). The catalytic activity of
catalyst 11b was lower than that of catalyst 10 due to its side
chains in ortho position, as also observed for the homoge-
neous counterparts 8 and 3.
In conclusion, both catalysts 10 and 11b showed the same
dependence on structural influences as the homogeneous
ones, and even the catalytic activity was comparable for the
free and grafted complexes. Hence, the presented strategy to
immobilise the most active salen complexes was a reasonable
approach.
3.5. Reuse of the heterogeneous catalysts
Fig. 3. Chromium content (top) and catalytic activity (bottom) of the differ-
entimmobilisedcatalysts(Scheme4)afterreuseinthesynthesisofpropylene
carbonate (activity is expressed as TOF [molproduct (molCr on catalyst h)−1]).
An important criterion for heterogeneous catalysis is
reusability. Catalyst9showedastronglossofactivity(Fig. 3).
After one use, the yield dropped from 41 to 1.4%. Since
one-third of the chromium content remained on the catalyst,
not only leaching of the salen complex, but also deactiva-
tion of the remaining chromium species occurred. A similar
deactivation of coordinatively-bound catalysts was observed
for a styrene oxide system with additional solvent by Gar-
cia and co-workers [32,33], but no leaching was reported in
[34,35]. In conclusion, anchoring via the Cr metal atom was
not strong enough for the title reaction and led to leaching and
ligand exchange. This may be traced back to the rather dras-
tic conditions during the reaction or the strong propensity for
complexation of propylene oxide and its derivatives. In sum-
mary, this simple approach is not suitable and the observed
activity probably originates from homogeneous catalysis.
Significantly more stable were the covalently-bound cat-
alysts 10 and 11b (Fig. 3). However, this required a com-
promise in the catalytic activity, since either the amine or
benzaldehyde component had to be modified. A certain loss
of the chromium content was observed for catalyst 10 after
the first use. The high stability in subsequent runs indicates
that this may have resulted from adsorbed chromium that
was not removed at the work-up of the synthesis by wash-
ing under atmospheric conditions. Catalyst 10 was prepared
step by step on the silica surface and finally the chromium
chloride was added. This final step could have caused the
high adsorption of chromium. In contrast no strong “wash-
ing” effect was observed with catalyst 11b. In the latter case
the complete complex was grafted, the chromium content was
stable over the whole time. The observed loss in activity of
catalyst 11b was less pronounced than that of catalyst 10.
DRIFTS measurements of a used catalyst 10 showed
decreased signals at 1622 and 1600 cm−1 (Fig. 1). In accor-
dance with the loss of chromium content the signals of the
C N vibration became weaker. This indicated that the whole
complex and not only chromium was washed out during
reaction. Further, strong bands appeared in the DRIFTS of
the used catalyst 10 at 2977, 2935 and 2883 cm−1, and at
1794 and 1745 cm−1, corresponding to the adsorbed propy-
lene carbonate species. Strong adsorption of propylene oxide
and similar compounds was also reported in [36]. Hence,
polymerisation probably occurs as a side reaction during the
cycloaddition of carbon dioxide to propylene oxide leading
to some deactivation of the catalyst during reuse. This may
also explain why the activity of both catalysts 10 and 11b still
decreased while hardly any leaching was observed. Thus, the
decreasing “TOF” at stable chromium content was caused
by blockage or disintegration of an active chromium cen-
tre. Note that no deactivation was reported in literature for

M. Ramin et al. / Journal of Molecular Catalysis A: Chemical 242 (2005) 32–39
39
a related substrate, ethylene oxide [37,38], probably due to
less pronounced adsorption on the silica support.
[4] M. Aresta, E. Quaranta, Chemtech 27 (1997) 32.
[5] G. Rokicki, W. Kuran, Bull. Chem. Soc. Jpn. 57 (1984) 1662.
[6] G. Rokicki, W. Kuran, B. Pogorzelskamarciniak, Monatsh. Chem.
115 (1984) 205.
[7] M. Ratzenhofer, H. Kisch, Angew. Chem. Int. Ed. 19 (1980) 317.
[8] A. Baba, T. Nozaki, H. Matsuda, Bull. Chem. Soc. Jpn. 60 (1987)
1552.
[9] D.J. Darensbourg, S.J. Lewis, J.L. Rodgers, J.C. Yarbrough, Inorg.
Chem. 42 (2003) 581.
Therefore, the use of modified silica or other supports,
which are less polar together with the immobilisation strate-
gies shown in this study, may lead to catalysts with improved
stability for propylene carbonate synthesis from propylene
oxide and carbon dioxide. Research towards this aim is
presently pursued in our laboratory.
[10] H.S. Kim, J.J. Kim, B.G. Lee, O.S. Jung, H.G. Jang, S.O. Kang,
Angew. Chem. Int. Ed. 39 (2000) 4096.
[11] H.S. Kim, J.J. Kim, S.D. Lee, M.S. Lah, D. Moon, H.G. Jang,
Chem.-Eur. J. 9 (2003) 678.
[12] F.W. Li, C.G. Xia, L.W. Xu, W. Sun, G.X. Chen, Chem. Commun.
(2003) 2042.
[13] H. Kawanami, A. Sasaki, K. Matsui, Y. Ikushima, Chem. Commun.
(2003) 896.
[14] R.L. Paddock, Y. Hiyama, J.M. McKay, S.T. Nguyen, Tetrahedron
Lett. 45 (2004) 2023.
[15] A. Baiker, Appl. Organomet. Chem. 14 (2000) 751.
[16] S. Fujita, B.M. Bhanage, Y. Ikushima, M. Shirai, K. Torii, M. Arai,
Catal. Lett. 79 (2002) 95.
[17] H. Yasuda, L.N. He, T. Sakakura, J. Catal. 209 (2002) 547.
[18] K. Yamaguchi, K. Ebitani, T. Yoshida, H. Yoshida, K. Kaneda, J.
Am. Chem. Soc. 121 (1999) 4526.
[19] M. Tu, R.J. Davis, J. Catal. 199 (2001) 85.
[20] R. Srivastava, D. Srinivas, P. Ratnasamy, Catal. Lett. 89 (2003) 81.
[21] R.L. Paddock, S.T. Nguyen, J. Am. Chem. Soc. 123 (2001) 11498.
[22] D.J. Darensbourg, J.C. Yarbrough, J. Am. Chem. Soc. 124 (2002)
6335.
4. Conclusions
A series of chromium–salen complexes with different lig-
ands has been synthesised and tested for carbon dioxide
fixation in propylene oxide without additional solvents and
co-catalysts. The structure of the salen ligand had a strong
effect on the catalytic activity, whereas selectivity to propy-
lene carbonate was always high (>98%). The most active
complexes were immobilised on silica by coordinative or
covalent bonding.
The study shows that only covalently-bound complexes
exhibit reasonable stability under reaction conditions. Coor-
dinated complexes are too weakly anchored during propylene
oxide reaction with carbon dioxide under high pressure. No
advantage of double tethered complexes compared to com-
plexes anchored by single covalent bond was found. Instead,
the structure of the ligand had a pronounced influence on the
catalytic activity, similarly as observed with the correspond-
ing homogeneous catalysts.
[23] D.J. Darensbourg, R.M. Mackiewicz, J.L. Rodgers, A.L. Phelps,
Inorg. Chem. 43 (2004) 1831.
[24] M.H. Valkenberg, W.F. Holderich, Catal. Rev. 44 (2002) 321.
[25] F. Minutolo, D. Pini, A. Petri, P. Salvadori, Tetrahedron Asymmetry
7 (1996) 2293.
[26] R.G. Konsler, J. Karl, E.N. Jacobsen, J. Am. Chem. Soc. 120 (1998)
10780.
[27] S.K. Hwang, A. Juhasz, S.H. Yoon, N. Bodor, J. Med. Chem. 43
(2000) 1525.
Acknowledgements
[28] J.D. Grunwaldt, M.D. Wildberger, T. Mallat, A. Baiker, J. Catal. 177
(1998) 53.
[29] C.D. Wagner, L.E. Davis, M.V. Zeller, J.A. Taylor, R.H. Raymond,
L.H. Gale, Surf. Interface Anal. 3 (1981) 211.
[30] M.B. Lu, J.H. Xiu, R. He, K. Jin, L.M. Luo, X.J. Feng, Appl. Catal.
A 275 (2004) 73.
[31] D.J. Darensbourg, R.M. Mackiewicz, J.L. Rodgers, C.C. Fang, D.R.
Billodeaux, J.H. Reibenspies, Inorg. Chem. 43 (2004) 6024.
[32] C. Baleizao, B. Gigante, M.J. Sabater, H. Garcia, A. Corma, Appl.
Catal. A 228 (2002) 279.
[33] M. Alvaro, C. Baleizao, D. Das, E. Carbonell, H. Garcia, J. Catal.
228 (2004) 254.
[34] Y.P. Zhang, J.H. Fei, Y.M. Yu, X.M. Zheng, Catal. Lett. 93 (2004)
231.
[35] X.G. Zhou, X.Q. Yu, J.S. Huang, S.G. Li, L.S. Li, C.M. Che, Chem.
Commun. (1999) 1789.
[36] B.S. Uphade, T. Akita, T. Nakamura, M. Haruta, J. Catal. 209 (2002)
331.
[37] H.S. Kim, J.J. Kim, H.N. Kwon, M.J. Chung, B.G. Lee, H.G. Jang,
J. Catal. 205 (2002) 226.
¨
Financial support by the Bundesamt fur Energie (BFE)
is gratefully acknowledged. Thanks are due to Marek
Maciejewski for TA, Carsten Beck for ICP-OES measure-
ments, and Davide Ferri for the support during the DRIFTS
measurements.
References
[1] A.A.G. Shaikh, S. Sivaram, Chem. Rev. 96 (1996) 951.
[2] A. Behr, Angew. Chem. Int. Ed. 27 (1988) 661.
[3] H. Arakawa, M. Aresta, J.N. Armor, M.A. Barteau, E.J. Beckman,
A.T. Bell, J.E. Bercaw, C. Creutz, E. Dinjus, D.A. Dixon, K. Domen,
D.L. DuBois, J. Eckert, E. Fujita, D.H. Gibson, W.A. Goddard,
D.W. Goodman, J. Keller, G.J. Kubas, H.H. Kung, J.E. Lyons, L.E.
Manzer, T.J. Marks, K. Morokuma, K.M. Nicholas, R. Periana, L.
Que, J. Rostrup-Nielson, W.M.H. Sachtler, L.D. Schmidt, A. Sen,
G.A. Somorjai, P.C. Stair, B.R. Stults, W. Tumas, Chem. Rev. 101
(2001) 953.
[38] X.B. Lu, H. Wang, R. He, J. Mol. Catal. A: Chem. 186 (2002) 33.