Supported homogeneous catalyst makes its own liquid phase

Article abstract of DOI:10.1016/j.jcat.2014.10.019

A catalyst designed for homogeneous catalysis is shown to generate its own liquid phase if deposited onto a support. In this way, a macroscopically heterogeneous catalyst generates a microscopically homogeneous catalytic environment by self-organization. 2,2′-((3,3′-di-tert-butyl-5,5′-dimethoxy-[1,1′-biphenyl]-2,2′-diyl)-bis(oxy))bis(4,4,5,5-tetraphenyl-1,3,2-dioxaphospholane) modified rhodium complexes molecularly adsorbed onto porous silica powder show surprisingly high activity and regioselectivity in the gas-phase hydroformylation of propene to butanal, with no sign of deactivation. Operando IR investigations combined with density functional theory calculations confirm a side reaction: the aldol condensation of the butanal products. These heavier by-products accumulate inside the pores of the catalytic material. IR and gas chromatography show a direct relation between formation of enones, products of the aldol condensation, performance, and stability of the catalytic system. This demonstrates that the aldol condensation products generated in situ act as a solvent providing an ideal environment to the impregnated homogeneous catalyst.

Full text of DOI:10.1016/j.jcat.2014.10.019

Journal of Catalysis 321 (2015) 32–38
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Supported homogeneous catalyst makes its own liquid phase
Andre Kaftan a, Andreas Schönweiz b, Ioannis Nikiforidis c, Wolfgang Hieringer c, Katrin M. Dyballa d,
Robert Franke d,e, Andreas Görling c, Jörg Libuda a,f, Peter Wasserscheid b,f, Mathias Laurin a, , Marco
⇑
Haumann b,
⇑
a Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), Lehrstuhl für Physikalische Chemie II, Egerlandstr. 3, 91058 Erlangen, Germany
b Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), Lehrstuhl für Chemische Reaktionstechnik, Egerlandstr. 3, 91058 Erlangen, Germany
c Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), Lehrstuhl für Theoretische Chemie, Egerlandstr. 3, 91058 Erlangen, Germany
d Evonik Industries AG, Paul-Baumann-Str. 1, 45772 Marl, Germany
e Ruhr-Universität Bochum, Lehrstuhl für Theoretische Chemie, Universitätsstr. 150, 44780 Bochum, Germany
f Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), Erlangen Catalysis Resource Center, Egerlandstr. 3, 91058 Erlangen, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
A catalyst designed for homogeneous catalysis is shown to generate its own liquid phase if deposited onto
a support. In this way, a macroscopically heterogeneous catalyst generates a microscopically homoge-
neous catalytic environment by self-organization.
Received 15 July 2014
Revised 28 October 2014
Accepted 31 October 2014
2,20-((3,30-di-tert-butyl-5,50-dimethoxy-[1,10-biphenyl]-2,20-diyl)-bis(oxy))bis(4,4,5,5-tetraphenyl-1,3,
2-dioxaphospholane) modified rhodium complexes molecularly adsorbed onto porous silica powder
show surprisingly high activity and regioselectivity in the gas-phase hydroformylation of propene to but-
anal, with no sign of deactivation. Operando IR investigations combined with density functional theory
calculations confirm a side reaction: the aldol condensation of the butanal products. These heavier by-
products accumulate inside the pores of the catalytic material. IR and gas chromatography show a direct
relation between formation of enones, products of the aldol condensation, performance, and stability of
the catalytic system. This demonstrates that the aldol condensation products generated in situ act as a
solvent providing an ideal environment to the impregnated homogeneous catalyst.
Keywords:
Hydroformylation
Supported homogeneous catalysis
Operando spectroscopy
Infrared spectroscopy
Density functional theory
Silicon dioxide
Rhodium
Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction
aqueous biphasic [4], micro-emulsions [5–7], ionic liquids [8,9],
and supported liquids [10–15], have been investigated over the
With an annual global production capacity of more than
10 million tons, hydroformylation is one of the largest applica-
tions of homogeneous catalysis in industry [1]. Alkenes are con-
verted with syngas into aldehydes in the presence of a transition
metal catalyst, with the linear isomer usually being the desired
product.
These aldehydes are valuable intermediates for the synthesis of
bulk chemicals such as alcohols, esters, amines, surfactants, and
plasticizers [2]. Modern catalysts for shorter alkenes in the C2 to
C8 range are predominantly based on rhodium modified by
sterically demanding phosphine or phosphite ligands [3]. Reaction
conditions are mild with 80–130 °C and 10–60 bar syngas pres-
sure. Due to the high price of rhodium, efficient catalyst recycling
is mandatory, and several immobilization strategies, for example
past decades in academia and industry.
We have investigated the hydroformylation using molecularly
adsorbed ligand-modified rhodium complexes Rh-BzP on silica
gel [16]. C2AC4 feeds were used as substrates and could be con-
verted with high yield and minimal deactivation over 1000 h time
on stream. Still, the catalyst performance was strongly dependent
on the support structure and alkene feed. Immediate activity and
selectivity were observed when using silica gel 100, a material
with an average pore diameter of 11 nm. However, an induction
period in which the catalyst gained activity within the first hours
of experiment was observed using a silica material with an average
pore size of 423 nm (TrisoporÒ 423), as shown in Fig. S1 (see the
SI).
These observations combined with post-run analysis of the
RhASiO2 catalyst suggest the following hypothesis: The smaller
pores of silica 100 enhance the capillary condensation of the
aldehydes formed. This quickly facilitates the consecutive aldol
condensation reaction that provides the ideal solvent environment
for the rhodium catalyst (Scheme 1a and Fig. 1). Indeed, the enone
⇑
Corresponding authors. Fax: +49 (0)9131 85 28867 (M. Laurin). Fax: +49
(0)9131 85 27421 (M. Haumann).
E-mail addresses: mathias.laurin@fau.de (M. Laurin), marco.haumann@fau.de
(M. Haumann), marco.haumann@fau.de (M. Haumann).
http://dx.doi.org/10.1016/j.jcat.2014.10.019
0021-9517/Ó 2014 Elsevier Inc. All rights reserved.

A. Kaftan et al. / Journal of Catalysis 321 (2015) 32–38
33
resulting from aldol condensation, also called heavies, is a excel-
2.3. Kinetic experiments
lent solvent for PPh3-modified (Ph = phenyl) rhodium in industrial
hydroformylation, namely the UCC liquid recycle process [17–20].
These enone formation processes, which can be catalyzed by the
acidic silica support, are much slower in the larger pores. However,
evidences for such a catalyst generating itself an ideal environment
require a direct investigation of the enone formation. Here, we
investigated the behavior of a RhASiO2 catalyst (Scheme 1b) in
the hydroformylation of propene with diffuse reflectance IR spec-
troscopy (DRIFTS) [21].
The experiments were conducted in a tubular reactor setup pre-
viously described [15,22]. The catalyst was placed inside the reac-
tor, and a pre-mixed gas feed was passed over the bed. Online
analysis was carried out using an Agilent 7890A GC with a 1-mL
sampling loop, equipped with a GS-GasPro capillary column (Agi-
lent Technologies, length 30 m, inner diameter 0.32 mm) and a
flame ionization detector (FID).
2.4. Operando experiments
The catalyst characterization was performed in a Bruker Vertex
80v IR spectrometer equipped with a homemade sample compart-
ment cover. This cover is equipped with all necessary feedthroughs
to lead the gas in and out of the reactor, and control and measure
the temperature while keeping the optical path completely evacu-
ated during the measurements. DRIFTS was performed in a praying
mantis using a high-temperature reaction chamber (HVC-DRP-4)
from Harrick. The reactor was modified to include a type K thermo-
couple measuring the temperature directly in the powder. Mass
flows and pressures were regulated using Bronkhorst mass flow
and pressure controllers. Prior to the start of the reaction, the cat-
alyst powder was heated under an Ar flow (5 mLs minꢁ1, 2 bar) at
80 °C for 3 h to desorb water and solvent residues. The experimen-
tal protocol is further detailed in the main text. IR spectra were
recorded with a spectral resolution of 2 cmꢁ1, 151 scans, and a
scanning speed of 40 kHz yielding an acquisition time of 60 s per
spectrum. Online gas chromatography was performed on an iden-
tical GC setup as in the kinetic experiments. The gas stream was
injected every 10 min into the column, which was kept at 200 °C
(see Fig. S1 for an excerpt of the chromatogram). All gases were
used without further purification (Ar, Linde, >99.9999%; propene,
Linde, >99.8%; H2, Linde, >99.999%; CO, Linde, >99.997%; He, Linde,
>99.996%; synthetic air, Linde, 80% N2 20% O2, hydrocarbon free;
N2, Linde, >99.999%).
2. Materials and methods
2.1. Chemicals
Rh(CO)2(acac), methanol, and dichloromethane (HPLC grade)
were purchased from Sigma Aldrich and used without further
purification. The BzP ligand 2,20-((3,30-di-tert-butyl-5,50-dime-
thoxy-biphenyl-2,20-diyl)bis(oxy))bis(4,4,5,5-tetraphenyl-1,3,2-
dioxaphospholane) was supplied by Evonik Industries AG, Marl,
Germany. Silica gel 100 (63–200
lm) was purchased from Merck,
m) from VitroBio GmbH. The SiO2
and TrisoporÒ 423 (100–200
l
materials were thermally pre-treated at 600 °C for 18 h before
usage in catalyst preparation. Propene (99.5%), carbon monoxide
(99.97%), and hydrogen (99.999%) were purchased from Linde
AG.
2.2. Catalyst preparation
All syntheses were carried out using standard Schlenk tech-
nique under Argon (99.999%). Rh(CO)2(acac) was dissolved in
dichloromethane and stirred for 5 min. The ligand was added in
fivefold molar excess (ligand/Rh = 5), dissolved in dichlorometh-
ane, respectively, and the resulting solution was stirred for
5 min. After addition of appropriate amount of calcined silica 100
or TrisoporÒ 423, the suspension was stirred for another 10 min.
The organic solvent was removed by using a rotatory evaporator,
and the dry powder was further dried under vacuum (1 ꢀ 10ꢁ4 bar)
over night.
2.5. Computational details
Density functional theory (DFT) [23] calculations were
carried out using the TURBOMOLE program package [24,25]. The
(a)
(b)
Scheme 1. (a) Reaction sequence of hydroformylation followed by aldol condensation and (b) equilibrium for diphosphite-modified rhodium complexes and structure of 2,20-
((3,30-di-tert-butyl-5,50-dimethoxy-biphenyl-2,20-diyl)bis(oxy))bis(4,4,5,5-tetraphenyl-1,3,2-dioxaphospholane) (abbrev. BzP).

34
A. Kaftan et al. / Journal of Catalysis 321 (2015) 32–38
Fig. 1. Illustration of the process of pore filling with aldehyde and aldol products during the catalytic hydroformylation reaction using ligand-modified Rh complexes in
porous silica supports.
gradient-corrected exchange-correlation functional due to Becke
2.5
1 stage 2
stage 3
[26] and Perdew [27] was used in combination with the resolu-
tion-of-the identity technique (RI). A triple-zeta-valence-polarized
(TZVP) basis set [28] together with the corresponding auxiliary
basis set [29] obtained from the TURBOMOLE basis set library was
used. All calculated molecules have been fully optimized in the
gas phase without any symmetry or other constrains. No imaginary
frequencies were obtained in the frequency calculations. The
energy difference between the Z- and E-enone was found to be
2.7 kcal molꢁ1 in favor of the E-enone. Visualization of the calcu-
lated frequencies was performed using Vibeplot [30].
2.0
1.5
1.0
0.5
tubular reactor
DRIFTS reactor
0.0
98
96
94
92
90
3
2
1
0
O
O
3. Results and discussion
O
3.1. Comparison between tubular reactor and DRIFTS reaction cell
0
20
40
60
80
100
120
140
160
The reaction chamber for the DRIFTS cell can be operated at a
maximum pressure of 2 bar while the usual hydroformylation con-
ditions are 10 bar. In order to compare the performance of the Rh-
SiO2 catalyst, low-pressure experiments were conducted in the
tubular reactor under close to identical conditions (with a resi-
dence time of 4.1 s in the DRIFTS reactor and 6 s in the tubular
reactor, details on the reactor designs are provided in SI). Propene
was used as substrate because it was more accurate to dose via
mass flow controllers compared to alkenes with a higher carbon
number. It was expected that the condensation of the produced
butanal and therefore the formation of the enone film occur on a
longer timescale under the conditions applied, making it easier
to study the transient behavior in both reactor types. Online GC
analysis provided conversion and selectivity over time as depicted
in Fig. 2.
Despite the different designs of the reactors, the results were
very similar over the tested 160 h on stream. In both experiments,
propene conversion started from 0% and increased rapidly over the
first 6 h. Conversion reached a steady state after approx. 30 h time
on stream for which 1.6% was determined in the operando reactor
and 2.1% in the tubular one. The selectivity to linear butanal
already reached a steady state after 12 h with values of 98% in
the operando and 97% in the tubular reactor. These values are
slightly lower than in previous experiments using the same Rh-
BzP catalyst species, presumably due to the lower pressure in the
present case. Three stages of operation are identified from the
gas chromatogram shown in Fig. 2: In the first 6 h, the catalyst
showed an activation period (stage 1), where both conversion
and n/iso-selectivity changed significantly. In the second period,
time on stream / h
Fig. 2. Conversion (top) and selectivity (bottom) measured online behind the
tubular reactor (open symbols) and operando in the IR reactor (full symbols). The
selectivity toward 2-ethylhex-2-enal was measured online within the first 110 h in
the tubular reactor experiment. Reaction conditions: mcat = 700 mg (for operando IR
reactor: 60 mg), mRh = 0.2 wt.%, L/Rh = 5, T = 80 °C, p = 2 bar, ppropene = 0.4 bar,
pH2 = pCO = 0.8 bar, residence time = 6 s (for operando IR reactor: 4.1 s).
between 6 and 30 h, these changes became less pronounced and
the selectivity reached its steady state value while the activity
was still slowly increasing (stage 2). Finally, the steady state where
both activity and n/iso-selectivity remained unchanged over time
on stream was reached after 30 h (stage 3). The selectivity toward
the aldol product was also low in the beginning and increased rap-
idly during the first 6 h on stream in the tubular reactor. After
reaching a maximum selectivity of about 3% after about 18 h, the
amount of enone in the gas phase decreased over the whole exper-
iment. In addition, the reaction gas flow was cut off after 115 h in
the tubular reactor experiment, and the catalyst was stored under
helium inert gas overnight. Upon restart of the reaction, the system
almost instantaneously showed the very same catalytic perfor-
mance as before. Here, no prolonged activation period was
observed, indicating that the catalyst material was already acti-
vated when the hydroformylation reaction continued.
3.2. Operando DRIFTS measurements
Operando DRIFTS complements this reactivity study and
enables a better understanding of the nature of the three stages

A. Kaftan et al. / Journal of Catalysis 321 (2015) 32–38
35
identified. Fig. 3 displays the first and last 6 h of the total experi-
ment over 218 h. The difference spectra display the changes with
respect to the initial state of the sample.
ones at 1000, 1450, 1600, and 3000 cmꢁ1 that are all attributed to
the presence of propene. Note that H2 cannot be observed with IR
but was nevertheless introduced with the other gases. The inten-
sity of the gas-phase signals stabilized after 60 min. IR bands then
developed at 1723, 2066, 2750–3000, and 3200–3600 cmꢁ1. The
Introduction of the reactants at 10 min was identified by the
rapid appearance of bands around 2100 cmꢁ1 (CO), and additional
products
catalyst products
218
216
214
212
p = 2 bar, Ar
T = 303 K, Ar
p = 2 bar
H :CO:propene
2
4
2
0
T = 353 K
water
propene
CO(g)
propene
H :CO:propene
2
4000
3000
2000
1000
wavenumber / cm–1
Fig. 3. Time series during the DRIFTS operando experiment shown as a grayscale surface plot. Experiments were conducted under a constant total pressure of 2 bar, a mixture
of H2:CO:propene (ratio 2:2:1) flowing at 2.5 mLs minꢁ1 total flow, at 80 °C. After 213 h, the reactor was cooled down to 30 °C under the same flow rate. At 216 h, the reactor
was purged with Ar (10 mLs minꢁ1).
(a)
(b)
2066
1723
0.2
0.5
2114
1670
2012
2040
1986
96
72
48
24
12
6
2
1
0.5
2200
2150
2100
2050
2000
1950 1800
1750
1700
1650
1600
wavenumber / cm–1
wavenumber / cm–1
(c)
(d)
0.3
O
2066
1.0
0.8
0.6
0.4
0.2
0
1723
0.2
0.1
0
2114
2012
O
1670
2040
1986
0
20
40
60
80
100
0
20
40
60
80
100
time on stream / h
time on stream / h
Fig. 4. Time-resolved operando DRIFTS spectra of the CO region acquired after selected times between 30 min and 96 h. The gas-phase signals of CO and propene were
subtracted according to the procedure described in supporting information. (a) 1950–2200 cmꢁ1 region; (b) 1600–1800 cmꢁ1 region; (c) amplitudes from a; and (d)
amplitudes from b. IR spectra were recorded for approximately 16 h a day for 5 days. The reaction continues in the blank regions in (c) and (d), but IR was not measured due to
practical limitations.

36
A. Kaftan et al. / Journal of Catalysis 321 (2015) 32–38
large contribution at 3200–3600 cmꢁ1 is attributed mostly to con-
densation of water on the liquid-nitrogen-cooled MCT detector.
The strong increase in the signal after 213 h, when the reactor
was cooled to 30 °C, suggests condensation of water in the pores
of the catalyst that can come from aldol condensation that liber-
ates water (Scheme 1a) and possibly also from wet propene or
CO feeds. After 216 h, the reactor was purged with 10 mLs minꢁ1
Ar, as can be seen from the disappearance of the gas-phase and
water signals. Interestingly, the bands previously observed at
1723, 2066, and 2750–3000 cmꢁ1 and another band at
2144 cmꢁ1 remain visible after removal of all reactants. The band
at 2066 cmꢁ1 is attributed to the most active absorption resulting
from the formation of the monomeric catalyst complexes [31], the
band at 1723 cmꢁ1 to C@O stretching vibrations in the products,
and the bands at 2750–3000 cmꢁ1 to CAH stretching in the reac-
tion products.
From this overview, it is clear that operando DRIFTS experi-
ments have the potential to give information on the activation of
the catalyst, the reactants, and the products in real time and under
gas-phase hydroformylation.
Fig. 4 shows selected spectra in the CO region, (a) from 2200 to
1950 cmꢁ1 and (b) from 1800 to 1600 cmꢁ1. In Fig. 4a, the band at
2114 cmꢁ1 and its shoulder expanding to 2050 cmꢁ1 readily
appeared upon introduction of CO in the reactor. The band reached
a maximum after about 6 h and then slightly decreased with time
(see Fig. 4c). The band already identified at 2066 cmꢁ1 increased
together with three other bands at 2040, 2012, and 1986 cmꢁ1
.
The evolution of these four bands over time was parallel through-
out. Their intensity increased during the first 20 h and reached
near saturation at above 60 h. These bands characterize the com-
plex isomers formed upon activation of the catalyst precursor by
ligand and syngas, shown in Scheme 1b [31].
The band at 2114 cmꢁ1 and its shoulder are consistent with iso-
lated gem-dicarbonyl Rh(CO)2 [32–34]. These Rh(CO)2 species were
formed as soon as CO was added and indicate the presence of
highly dispersed Rh that is probably the result of a small amount
of catalyst decomposed during the pre-treatment of the catalyst
(see experimental section). Despite the presence of these isolated
Rh atoms, no sign of sintering or formation of particles was
observed, which would display the characteristic linear-bound
CO at 2080 cmꢁ1 [34,35]. Fig. 4b shows the absorption band
already identified at 1723 cmꢁ1 and a less intense band at
1670 cmꢁ1. It appeared that the band at 1723 cmꢁ1 intensified first,
approaching steady state after 6 h. The band at 1670 cmꢁ1 was first
observed after 6 h and kept growing over the whole experiment.
This could indicate that this later band stemmed from a by-product
Fig. 5. Calculated IR spectra of (i) E-enone, (ii) Z-enone and (iii) n-butanal. A Gaussian broadening with a full width at half maximum (FWHM) of 8 cmꢁ1 was applied to all
calculated frequencies. A 5ꢀ magnification factor was applied to the fingerprint region (wavenumbers below 1500 cmꢁ1). Simplified visualization of the individual normal
modes of the products: (iv) E-enone, (v) Z-enone and (vi) n-butanal (raw output can be found in the SI). The thickness of the lines representing the individual modes is
proportional to the stretching amplitude of the respective mode. The phase of the displacement is reflected by different colors. (For interpretation of the references to color in
this figure legend, the reader is referred to the web version of this article.)

A. Kaftan et al. / Journal of Catalysis 321 (2015) 32–38
37
of the reaction. At this point, however, unambiguous assignment of
the absorption features of the products is not possible.
Stage 2. Once the concentration of product in the pores reached
equilibrium and most of the catalyst had been activated and dis-
solved, the enone by-product started forming, providing an even
better solvent for the catalyst. There, the activity increased while
the regioselectivity remained constant.
Stage 3. Aldol condensation occurred over the whole experi-
ment, but at a rather low rate under our experimental conditions.
This formation of enone coincided with the stabilization of the cat-
alytic properties observed in the kinetic experiments. Despite the
increasing amount of enone, we could not detect a decrease of
the butanal concentration in the pores of the catalyst or in the
gas phase as could be expected from filled pores.
For this reason, the spectra for the E-enone, Z-enone, and n-but-
anal were calculated using density functional theory (DFT). A zoom
on the 1600–1800 cmꢁ1 region is shown for comparison with
Fig. 4b. The E-enone and the Z-enone both feature an intense
absorption at 1692 or 1675 cmꢁ1 (resp.) and a less intense absorp-
tion at 1633 or 1626 cmꢁ1 (resp.). This pattern is qualitatively com-
parable to the experimental spectra acquired operando and to the
reference measurements (cf. Figs. 4b and S4). The frequencies
obtained are also within the accuracy expected from harmonic cal-
culations of free molecules compared to the real system. Fig. 5iv–vi
shows the visualization of the corresponding vibrational modes in
terms of changes of internal coordinates [30]. It appears that the
Acknowledgments
vibrations at higher frequencies
[
m55(E- and Z-enone) and
m
31(n-butanal)] are due to stretching of the CHO groups and the
This project was financially supported by the ‘‘Deutsche
Forschungsgemeinschaft’’ (DFG) within the Excellence Cluster
‘‘Engineering of Advanced Materials’’ in the framework of the
excellence initiative. A.S., K.M.D., R.F., P.W., and M.H. gratefully
acknowledge financial support from the Bundesministerium für
Bildung und Forschung (BMBF, German Federal Ministry of Educa-
tion and Research) under the HY-SILP project (No. 01RC1107A) and
its framework programme Research for Sustainable Development
(FONA).
vibrations at lower frequency [m54(E- and Z-enone)] to stretching
vibrations of the C@C enone double bond.
With the help of the DFT calculations, the band observed
experimentally at 1670 cmꢁ1 can be attributed unambiguously
to the enones, whereas the band observed at 1723 cmꢁ1 contains
contributions from both n-butanal and the aldol condensation
products. Comparison of the evolution of these signals during
the experiment (Fig. 4d) with the GC experiments (Fig. 2) displays
a remarkable agreement. Furthermore, the highest aldol conden-
sation rate coincides with the high butanal production rate over
the first 6 h. After stabilization of the butanal signal, the aldol
condensation rate decreases steadily without ever reaching a
steady state. We can therefore conclude that butanals are pro-
duced first up to an equilibrium concentration in the pores, fol-
lowed by formation of enone by-products at a lower rate. These
findings validate the pore-filling model illustrated in Fig. 2, in
which aldehyde and aldol products condense inside the porous
network of commonly used silica materials during the course of
the reaction.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2014.10.019.
References
[1] S. Naqvi, Process Economics Program Report 21E: Oxo Alcohols, SRI Consulting,
Menlo Park, CA, USA.
[2] R. Franke, D. Selent, A. Börner, Applied hydroformylation, Chem. Rev. 112
(2012) 5675–5732.
[3] P.W.N.M. van Leeuwen, C. Claver, Rhodium Catalyzed Hydroformylation,
Kluwer Academic Publishers, Dordrecht, 2001.
4. Conclusions
[4] C.W. Kohlpaintner, R.W. Fischer, B. Cornils, Aqueous biphasic catalysis:
ruhrchemie/Rhône-Poulenc oxo process, Appl. Catal. A 221 (2001) 219–225.
[5] M. Haumann, H. Koch, P. Hugo, R. Schomäcker, Hydroformylation of 1-
dodecene using Rh-TPPTS in a microemulsion, Appl. Catal. A 225 (2002) 239–
249.
[6] T. Hamerla, A. Rost, Y. Kasaka, R. Schomäcker, Hydroformylation of 1-dodecene
with water-soluble rhodium catalysts with bidentate ligands in multiphase
systems, ChemCatChem 5 (2013) 1854–1862.
[7] J. Potier, S. Menuel, E. Monflier, F. Hapiot, Synergetic effect of randomly
methylated b-cyclodextrin and a supramolecular hydrogel in Rh-catalyzed
hydroformylation of higher olefins, ACS Catal. 4 (2014) 2342–2346.
[8] M. Haumann, A. Riisager, Hydroformylation in room temperature ionic liquids
(RTILs): catalyst and process developments, Chem. Rev. 108 (2008) 1474–
1497.
[9] F. Jérôme, M. Ferreira, H. Bricout, S. Menuel, E. Monflier, S. Tilloy, Low melting
mixtures based on b-cyclodextrin derivatives and N,N0-dimethylurea as
solvents for sustainable catalytic processes, Green Chem. 16 (2014) 3876.
[10] H.L. Pelt, J.J.J.J. Brockhus, R.P.J. Verburg, J.J.F. Scholten, Hydroformylation of
alkenes with supported liquidphase rhodium catalysts: the influence of the
dissolution of the produced alkanals on catalytic performance, J. Mol. Catal. 31
(1985) 107–118.
[11] H.L. Pelt, N.A. de Munck, R.P.J. Verburg, J.J.J.J. Brockhus, J.J.F. Scholten,
Hydroformylation of butene-1 and butene-2 over rhodium-slp catalysts, as
compared with the hydroformylation of ethene and propene, J. Mol. Catal. 31
(1985) 371–383.
[12] H.L. Pelt, R.P.J. Verburg, J.J.F. Schölten, The kinetics of the hydroformylation of
butene-1 over rhodium-supported liquid-phase catalysts: the influence of
pore filling, J. Mol. Catal. 32 (1985) 77–90.
[13] H.L. Pelt, P.J. Gijsman, R.P.J. Verburg, J.J.F. Scholten, The thermal and chemical
stability limits of supported liquid phase rhodium catalysts in the
hydroformylation of propene, J. Mol. Catal. 33 (1985) 119–128.
[14] J.P. Arhancet, M.E. Davis, J.S. Merola, B.E. Hanson, Hydroformylation by
supported aqueous-phase catalysis: a new class of heterogeneous catalysts,
Nature 339 (1989) 454–455.
[15] R. Franke, A. Schönweiz, Supported ionic liquid phase (SILP) materials in
hydroformylation catalysis, in: R. Fehrmann, A. Riisager, M. Haumann (Eds.),
Supported Ionic Liquids, Wiley-VCH, Weinheim, 2013.
The combination of a 218-h-long operando DRIFTS experiment
and density functional calculations of the resulting IR spectra gave
clear evidence that the BzP-modified Rh complex generates a
liquid enone phase in the pores of the silica substrate on which
it is adsorbed. The environment generated by the catalyst was
shown to provide favorable reaction conditions leading to good
activity and high robustness of the hydroformylation catalyst.
The experiment was furthermore reproduced under the same con-
ditions in a tubular fixed-bed reactor, and comparison of the exper-
iments in both types of reactor leads to a very good agreement. The
experiments showed that operando IR spectroscopy is a perfectly
suited method to study a homogeneous catalytic reaction as it
allowed to simultaneously identify the reactants, the formation
of the active catalyst (ee and ea monomers), the formation of the
reaction products (aldehydes) and of the by-products (enones).
Three stages were identified in the reaction that can be summa-
rized as follows.
Stage 1. The activation of the catalyst, that is, formation of the
(ee) and (ea) monomers took place during the first 6 h of the exper-
iment, which was identified as stage 1 in the kinetic experiments
using GC analysis. In the same period, mainly n-butanal formed
inside the pores of the catalyst, thereby dissolving more and more
of the molecularly adsorbed rhodium complex. Accordingly, the
butanal formation rate increased with increasing quantity of active
catalyst. Furthermore, GC measurements show that the activity
and the n/iso-selectivity were lower in this first period than at
steady state.

38
A. Kaftan et al. / Journal of Catalysis 321 (2015) 32–38
[16] A. Schönweiz, J. Debuschewitz, S. Walter, R. Wölfel, H. Hahn, K.M. Dyballa, R.
Franke, M. Haumann, P. Wasserscheid, Ligand-modified rhodium catalysts on
porous silica in the continuous gas-phase hydroformylation of short-chain
alkenes-catalytic reaction in liquid-supported aldol products, ChemCatChem 5
(2013) 2955–2963.
[26] A.D. Becke, Density-functional exchange-energy approximation with correct
asymptotic behavior, Phys. Rev. A 38 (1988) 3098–3100.
[27] J. Perdew, Density-functional approximation for the correlation energy of the
inhomogeneous electron gas, Phys. Rev. B 33 (1986) 8822–8824.
[28] F. Weigend, R. Ahlrichs, Balanced basis sets of split valence, triple zeta valence
and quadruple zeta valence quality for H to Rn: design and assessment of
accuracy, PCCP 7 (2005) 3297.
[17] D.L. Bunning, M.A. Blessing, patent EP 188.246, 1986.
[18] K.D. Sorensen, patent EP 484.976, 1991.
[19] D.L. Bunning, patent EP 404.193, 1990.
[20] D.R. Bryant, patent EP 504.814, 1992.
[29] K. Eichkorn, O. Treutler, H. Öhm, M. Häser, R. Ahlrichs, Auxiliary basis sets to
approximate Coulomb potentials, Chem. Phys. Lett. 240 (1995) 283–290.
[30] M. Laurin, QVibeplot: a program to visualize molecular vibrations in two
dimensions, J. Chem. Educ. 90 (2013) 944–946.
[31] A. van Rooy, P.C.J. Kamer, P.W.N.M. van Leeuwen, K. Goubitz, J. Fraanje, N.
Veldman, A.L. Spek, Bulky diphosphite-modified rhodium catalysts:
hydroformylation and characterization, Organometallics 15 (1996) 835–847.
[32] S.M. McClure, M.J. Lundwall, D.W. Goodman, Surface chemistry special
feature: planar oxide supported rhodium nanoparticles as model catalysts,
Proc. Natl. Acad. Sci. USA 108 (2011) 931–936.
[21] M.M. Schubert, T.P. Häring, G. Bräth, H.A. Gasteiger, R.J. Behm, New DRIFTS cell
design for the simultaneous acquisition of IR spectra and kinetic data using on-
line product analysis, Appl. Spectrosc. 55 (2001) 1537–1543.
[22] C.P. Mehnert, R.A. Cook, N.C. Dispenziere, M. Afeworki, Supported ionic liquid
catalysis ꢁ a new concept for homogeneous hydroformylation catalysis, J. Am.
Chem. Soc. 124 (2002) 12932–12933.
[23] W. Koch, M.C. Holthausen, A Chemist’s Guide to Density Functional Theory,
second ed., Wiley-VCH, Weinheim; New York, 2001.
[24] R. Ahlrichs, M. Bär, M. Häser, H. Horn, C. Kölmel, Electronic structure
calculations on workstation computers: the program system turbomole,
Chem. Phys. Lett. 162 (1989) 165–169.
[33] M. Frank, R. Kühnemuth, M. Bäumer, H.-J. Freund, Vibrational spectroscopy of
CO adsorbed on supported ultra-small transition metal particles and single
metal atoms, Surf. Sci. 454–456 (2000) 968–973.
[25] R. Ahlrichs, M.K. Armbruster, R.A. Bachorz, M. Bär, H.-P. Baron, R.
Bauernschmitt, F.A. Bischof, S. Böcker, N. Crawford, TURBOMOLE V6.5 2013,
[34] J. Evans, B. Hayden, F. Mosselmans, A. Murray, Adsorbate induced phase
changes of rhodium on TiO2(110), Surf. Sci. 279 (1992) L159–L164.
[35] M. Frank, R. Kühnemuth, M. Bäumer, H.-J. Freund, Oxide-supported Rh particle
structure probed with carbon monoxide, Surf. Sci. 427–428 (1999) 288–293.
in: TURBOMOLE GmbH,
A Development of University of Karlsruhe and
Forschungszentrum Karlsruhe GmbH, since 1989.