Hydroformylation of 1-Hexene in Supercritical Carbon Dioxide: Characterization, Activity, and Regioselectivity Studies

Article abstract of DOI:10.1021/es034544n

The hydroformylation of alkenes is a major commercial process used for the production of oxygenated organic compounds. When the hydroformylation reaction is performed using a homogeneous catalyst, an organic or aqueous solvent is employed, and a significant effort must be expended to recover the catalyst so it can be recycled. Development of a selective heterogeneous catalyst would allow simplification of the process design in an integrated system that minimizes waste generation. Recent studies have shown that supercritical carbon dioxide (scCO₂) as a reaction solvent offers optimal environmental performance and presents advantages for ease of product separation. In particular, we have considered the conversion of 1-hexene to heptanal using rhodium- and platinum-phosphine catalysts tethered to supports insoluble in scCO₂ to demonstrate the advantages and to understand the limitations of a solid-catalyzed process. One of the historical limitations of supported catalysts is the inability to control product regioselectivity. To address this concern, we have developed tethered catalysts with phosphinated silica and controlled pore size MCM-41 and MCM-20 supports that provide improved regioselectivity and conversion relative to their nonporous equivalents. Platinum catalysts supported on MCM-type supports were the most regioselective whereas the analogous rhodium catalysts were the most active for hydroformylation of 1-hexene in scCO₂.

Full text of DOI:10.1021/es034544n

Environ. Sci. Technol. 2003, 37, 5424-5431
Organic polymeric (4-7) and inorganic supports have
Hydroformylation of 1-Hexene in
Supercritical Carbon Dioxide:
Characterization, Activity, and
Regioselectivity Studies
been studied for heterogeneous catalysis (8-11). Inorganic
supports such as silica are more mechanically stable and
resistant against aging and solvent attack relative to their
polymeric analogues. However, the impregnation technique
that is often used to prepare silica-supported rhodium
catalysts has generally resulted in high levels of rhodium
leaching and poor activity and selectivity.
A N N E E . M A R T E E L , † T I M O T H Y T . T A C K , ‡
An alternative to impregnation is the use of sol-gel
condensation to prepare phosphinated silica supports (12-
15). The incorporation of the metal complex can then be
accomplished by stirring a mixture of the metal complex
and the phosphinated support in aqueous ethanol under
nitrogen. This synthesis enables a stronger coordination of
the metal complex to the phosphinated support, decreasing
considerably the leaching of rhodium. To improve the
selectivity of heterogeneous catalysts, the use of MCM
mesoporous molecular sieves as supports has been consid-
ered (11, 16). The MCM supports are attractive because they
provide easy separation of the catalyst from organic species
and high turnover numbers. The synthesis of micelle-
templated silicas based on the neutral templating method
(17) offers many advantages such as the simple washing for
separation and the easy recovery and reuse of the template,
which enhances the “greenness” of the manufacture of the
catalysts.
S E L M A B E K T E S E V I C , ‡
J U L I A N A . D A V I E S , †
M A R K R . M A S O N , † A N D
M A R T I N A . A B R A H A M * , ‡
Department of Chemistry and Department of Chemical and
Environmental Engineering, The University of Toledo,
Toledo, Ohio 43606
The hydroformylation of alkenes is a major commercial
process used for the production of oxygenated organic
compounds. When the hydroformylation reaction is performed
using a homogeneous catalyst, an organic or aqueous
solvent is employed, and a significant effort must be expended
to recover the catalyst so it can be recycled. Development
of a selective heterogeneous catalyst would allow
simplification of the process design in an integrated
system that minimizes waste generation. Recent studies
Although it has been well-documented that homogeneous
platinum/ tin systems are efficient for hydroformylation and
especially asymmetric hydroformylation, only a few studies
have been performed using heterogeneous platinum/ tin
systems (18, 19). The conversion of silica-supported platinum
catalysts was usually low with high rates of hydrogenation
and isomerization. Supported platinum catalysts have not
previously been studied as hydroformylation catalysts in
scCO2, aside from our recently communicated preliminary
results (20).
have shown that supercritical carbon dioxide (scCO ) as a
2
reaction solvent offers optimal environmental performance
and presents advantages for ease of product separation.
In particular, we have considered the conversion of 1-hexene
to heptanal using rhodium- and platinum-phosphine
catalysts tethered to supports insoluble in scCO to
2
demonstrate the advantages and to understand the limitations
of a solid-catalyzed process. One of the historical
Because of its environmental advantages, the use of scCO2
as a replacement for organic solvents in chemical reactions
has been widely investigated. Carbon dioxide is nontoxic,
does not form air-polluting daughter products, and ad-
ditionally, is cheap and plentiful. Work on hydroformylation
in scCO2 has been ongoing since Rathke et al. (21) first
reported on hydroformylation of propene by cobalt. Inves-
tigators have attempted to develop effective homogeneous
hydroformylation catalysts that are soluble in scCO2 and also
provide sufficient activity and selectivity to be commercially
viable (22-25). In nearly all cases, the activity was comparable
to that obtained in liquid-phase reactions, and reasonable
selectivity (typically L/ B ) 2-3) was obtained.
limitations of supported catalysts is the inability to control
product regioselectivity. To address this concern, we
have developed tethered catalysts with phosphinated silica
and controlled pore size MCM-41 and MCM-20 supports
that provide improved regioselectivity and conversion relative
to their nonporous equivalents. Platinum catalysts
supported on MCM-type supports were the most regio-
selective whereas the analogous rhodium catalysts were
the most active for hydroformylation of 1-hexene in scCO .
2
In the present work, the combination of the advantages
of using catalysts supported on inorganic materials prepared
via an environmentally preferable synthesis and the use of
the environmentally benign solvent scCO2 is studied. The
synthesis and characterization of silica- and MCM-supported
rhodium and platinum catalysts is described. The hydro-
formylation results are presented in terms of activity and
regio- and chemoselectivity to demonstrate the value of using
MCM supports for selective hydroformylation chemistry.
Introduction
Homogeneous hydroformylation reactions using rhodium-
phosphine catalysts in organic solvents have been extensively
studied and reviewed (1-3). It is well-known that an
important problem with homogeneous catalysts is the
difficulty to recover and recycle these catalysts. To eliminate
the need for the separation and recovery steps, heterogeneous
catalysis involving supported catalysts can be used. This
improves the environmental performance of the process by
eliminating the use of organic solvents and the energy
consumption associated with the recovery process.
Experimental Section
Phosphinated silica support 1a (14) and siliceous MCM-41
(support 1b) and MCM-20 (support 1c) were prepared as
previously reported (16). The preparation of phosphinated
MCM-41 and MCM-20 supports followed the procedure
described by Bemi et al. for modification of silica (26).
* Corresponding author phone: (419)530-7391; fax: (419)530-7392;
e-mail: martin.abraham@utoledo.edu.
† Department of Chemistry.
‡ Department of Chemical and Environmental Engineering.
9
5 4 2 4 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 23, 2003
10.1021/es034544n CCC: $25.00
2003 Am erican Chem ical Society
Published on Web 10/22/2003

TABLE 1. Summary of the Different Supports and Supported Metal Catalysts
support
support no.
metal precursor
catalyst no.
phosphinated silica
1a
1b
1c
1a
1b
1c
[RhCl(1,5-cod)]2
[RhCl(1,5-cod)]2
[RhCl(1,5-cod)]2
PtCl2{Ph2P(CH2)2Si(OEt)3}2
PtCl2{Ph2P(CH2)2Si(OEt)3}2
PtCl2{Ph2P(CH2)2Si(OEt)3}2
4a
4b
4c
7a
7b
7c
phosphinated MCM-41
phosphinated MCM-20
phosphinated silica
phosphinated MCM-41
phosphinated MCM-20
The rhodium precursor [RhCl(1,5-cod)]2 (cod ) cyclo-
octadiene) was synthesized as previously reported (27).
Rhodium catalysts supported on phosphinated silica supports
were prepared in our laboratory using procedures described
elsewhere (13, 14). Rhodium catalyst supported on MCM-41
(catalyst 4b) was prepared by stirring [RhCl(1,5-cod)]2 (0.05
g; 0.11 mmol) in 100 mL of ethanol for 15 min, after which
the phosphinated MCM-41 support 1b (1.00 g; 0.44 mmol of
P) was added, and the mixture was stirred for 24 h under
nitrogen. The product was isolated by filtration under
nitrogen, and the yellow catalyst was dried under vacuum
overnight. The phosphinated MCM-20-supported rhodium
catalyst 4c was synthesized with the phosphinated MCM-20
support 1c following the same procedure described for
catalyst 4b except that 81.6 mg of [RhCl(1,5-cod)]2 and 1.40
g of support 1c were used. The platinum precursor PtCl2-
{Ph2P(CH2)2Si(OEt)3}2 (26), phosphinated silica-supported
platinum catalyst 7a, and phosphinated MCM-41-supported
platinum catalyst 7b were prepared as previously reported
(20). Phosphinated MCM-20-supported platinum catalyst 7c
was prepared following the same procedure as that for catalyst
7b. All the catalysts have a metal to phosphorus molar ratio
of 1:2. Table 1 summarizes the different supports, metal
precursors, and catalysts used in this work.
FIGURE 1. CP/MAS 31P NMR spectra of phosphinated silica support
1a, MCM-41 support 1b, and MCM-20 support 1c.
Results and Discussion
Substantial catalyst characterization was conducted and is
described here. Following the discussion of the catalyst
materials, results obtained from the use of these catalysts for
the hydroformylation of 1-hexene are presented.
Silica (1a) and MCM-type (1b, 1c) supports, silica- and
MCM-supported rhodium catalysts (4a-4c), and supported
platinum catalysts (7a-7c) were characterized using CP/
MAS 13C and 31P NMR spectroscopy. All the catalysts were
characterized after hydroformylation using elemental analysis
and CP/ MAS 13C and 31P NMR spectroscopy and are named
“used” catalysts.
All NMR solution spectra were recorded on a Varian VXR-
400 spectrometer at 25 °C. Frequencies and standards were
as follows: 31P{1H} NMR: 161.9 MHz, external standard 85%
H3PO4/ D2O; 1H NMR: 400 MHz, standard tetramethylsilane
(TMS); 13C{1H} NMR: 100.4 MHz, standard TMS. The solid-
state CP/ MAS NMR spectra were recorded on a Chemagnetics
M200-S spectrometer operating at 4.7 T. The CP/ MAS NMR
spectra were all obtained using cross-polarization (CP), magic
angle spinning (MAS), and high-power proton decoupling.
Frequencies and standards: 13C NMR: 50.2 MHz, standard
hexamethylbenzene; 31P NMR: 80.8 MHz, standard triphen-
ylphosphine (PPh3). The uncertainties in chemical shifts and
coupling constants are estimated to be (1 ppm and (50 Hz,
respectively, for the CP/ MAS measurements. Errors in
reported line width are estimated to be 10%.
Catalyst Characterization. As the modified supports were
insoluble in common laboratory solvents, CP/ MAS 31P and
13C NMR spectroscopy were the preferred techniques for
characterization. The CP/ MAS 31P NMR spectra of phos-
phinated supports 1a-1c are shown in Figure 1 and revealed
resonances near 43.5 and -8 ppm; the asterisks indicate
spinning sidebands in the spectra. By comparison with the
solution 31P NMR of the ligand used in the preparation of the
supports, the resonance near -8 ppm was determined to be
a result of supported, uncoordinated phosphine. The pres-
ence of the deshielded resonance near 43.5 ppm was
indicative of a phosphorus(V) species as observed in the CP/
MAS 31P NMR spectrum of oxidized support 1a. Support 1b
exhibited a low-intensity resonance at 28 ppm that has not
been assigned, although we speculate that it may be due to
an additional phosphorus(V) species.
The CP/ MAS 13C NMR spectra of supports 1a-1c are
shown in Figure 2 and reveal several peaks: at 130 ppm
(C6H5-), centered at 60 ppm (-CH2-O-), and centered around
20 ppm (-CH2-Si- and -CH2-P-). The assignment of the
resonances reported in this work is in agreement with those
reported by Komoroski et al. (28).
The CP/ MAS 13C NMR spectrum of the rhodium precursor
[RhCl(1,5-cod)]2 revealed two resonances with the same
intensity: one centered at 30 ppm (allylic carbons) and the
other centered around 80 ppm (vinylic carbons). The
platinum precursor PtCl2{Ph2P(CH2)2Si(OEt)3}2 was charac-
terized as previously described (26).
Catalysts 4a-4c were characterized using CP/ MAS 31P
NMR spectroscopy before hydroformylation, as shown in
Figure 3. Two resonances around 40 and -10 ppm were
observed for catalyst 4a. The resonance at -10 ppm
accounted for the presence of uncoordinated phosphorus,
and the one at 40 ppm represented species with Rh-P
couplings and/ or the presence of oxidized phosphorus
Elemental analyses were carried out by Schwarzkopf
Microanalytical Laboratory (Woodside, NY). A duplicate
analysis was performed for each sample. Duplicate BET
surface area measurements were performed using a Mi-
cromeritics AutoChem 2910 volumetric adsorption analyzer.
The procedure for the hydroformylation reactions in scCO2
using supported rhodium catalysts has been previously
published (13-15). The hydroformylation reactions using
supported platinum catalysts were performed in a similar
manner except that the addition of a co-catalyst SnCl2‚2H2O
is necessary to obtain reasonable yields. A platinum to tin
molar ratio of 1:3.5 was used when performing hydrofor-
mylation with catalysts 7a-7c.
9
VOL. 37, NO. 23, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 5 4 2 5

TABLE 2. BET Surface Areas of Catalysts Synthesized in This
Project
BET surface
area (m2/g)
BET surface
area (m2/g)
catalyst
catalyst
4a
4b
4c
8.60; 9.16
642.1; 651.7
829.5; 856.8
7a
7b
7c
250.2; 252.4
715.6; 708.3
645.9; 654.2
phosphinated MCM-41 support was via attachment of the
ligand to the surface of preformed MCM-41, resulting in a
support in which all diphenylphosphino moieties were
accessible to the metal complex and the absence of unco-
ordinated phosphine.
In the CP/ MAS 13C NMR spectra of catalysts 4a-4c, the
ethene spacer is characterized by broad peaks around 30
ppm. The comparison between the CP/ MAS 13C NMR
spectrum of [RhCl(1,5-cod)]2 and the CP/ MAS 13C NMR
spectra of rhodium catalysts revealed that they exhibited the
features of both the cyclooctadiene (resonances at 30 and 80
ppm) and the support 1a (resonances at 130 and 60 ppm).
The CP/ MAS 13C NMR spectra of catalysts 4b and 4c revealed
that the majority of the features in the CP/ MAS 13C NMR
spectra of these catalysts corresponded to those of the
phosphinated MCM-41 support 1b.
FIGURE 2. CP/MAS 13C NMR spectra of phosphinated silica support
1a, MCM-41 support 1b, and MCM-20 support 1c.
The CP/ MAS 31P NMR spectra of catalysts 7a-7c all
exhibited a single resonance associated with J(195Pt-31P) of
1
3634 Hz (at 12 ppm), 3567 Hz (at 13 ppm), and 3701 Hz (at
12 ppm), respectively, as shown in Figure 4. The value of the
line width at half-height for catalyst 7a is 1010 Hz and is 969
Hz for catalyst 7b, which provides evidence of the distorted
environment of the immobilized platinum complexes. The
symmetry of the central resonances suggests that only one
type of phosphorus species is present. It has to be noted that
there was no phosphorus(V) species present in the CP/ MAS
31P NMR spectra of the immobilized platinum complexes.
Furthermore, the value of the coupling constant corre-
sponding to the mixture of cis- and trans-PtCl2{Ph2P(CH2)2-
Si(OEt)3}2 complexes on silica gel and on MCM-41 demon-
strates that the geometry of the complex PtCl2{Ph2P-
(CH2)2Si(OEt)3}2 switched to a cis geometry when the
platinum complex is anchored on the surface of these
supports. It is thought that a cis conformation would favor
the attachment of the phosphorus species to the support.
FIGURE 3. CP/MAS 31P NMR spectra of fresh rhodium catalysts
supported on silica (4a), MCM-41 (4b), and MCM-20 (4c).
species. In contrast, the CP/ MAS 31P NMR spectra of catalysts
4b and 4c before hydroformylation presented only a broad
resonance centered at 42 ppm. There is no uncoordinated
phosphorus evident in the CP/ MAS 31P NMR spectra of the
catalysts with MCM-41 supports. A possible explanation
might be the difference in the synthesis of supports on silica
and on MCM-41. The formation of silica by hydrolysis of
tetraethoxysilane and the condensation reaction with the
phosphine ligand were performed in one step, likely leading
to some diphenylphosphino moieties that were inaccessible
to the metal complex. In contrast, the preparation of the
The BET surface areas of all the different metal catalysts
are presented in Table 2. The comparison of the BET surface
FIGURE 4. CP/MAS 31P NMR spectra of platinum/tin catalysts supported on silica (7a), MCM-41 (7b), and MCM-20 (7c).
9
5 4 2 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 23, 2003

TABLE 3. Elemental Analysis of Catalysts after One
Hydroformylation Reactiona
catalyst
(molar metal:
P ratio ) 1:2)
wt % of
phosphorus
measd
wt % of
metal
measd
calcd molar
metal:P ratios
range
4a
used 4a
4b
used 4b
4c
used 4c
7a
used 7a
7b
used 7b
7c
2.91; 2.88
3.78; 4.59
3.40; 3.40
1.84; 1.81
0.60; *
1.81; 1.82
1.01; *
0.39-0.48
0.37-0.38
0.57-0.61
0.24
0.45-0.46
0.30
2.72; 2.77
0.95; 0.91
0.76; *
1.20; 1.21
1.01; *
0.90; 0.86
2.61;*
0.46-0.48
FIGURE 5. CP/MAS 13C NMR spectra of used rhodium catalyst
supported on MCM-41 (4b) as compared with the fresh material.
1.52; 2.13; 2.29 0.54; 0.71; 0.76 0.31-0.67
1.66; 1.69
1.62; *
1.62; 1.79
0.52; 0.63
4.91; 4.83
0.095; *
4.43; 4.36
3.44; 3.49
0.45-0.47
0.0093
0.39-0.43
0.867-1.065
Used catalysts 4a-4c were characterized by CP/ MAS 31
P
used 7c
and CP/ MAS 13C NMR spectroscopy. The CP/ MAS 31P NMR
spectrum revealed only one broad peak centered around 40
ppm, no different than observed for fresh catalyst. This
demonstrates that there was no uncoordinated supported
phosphine after hydroformylation. The broad resonance
suggests the presence of different species with Rh-P
couplings and/ or the presence of oxidized phosphorus
species.
The CP/ MAS 13C NMR spectra of fresh and used catalysts
4a had the same appearance except that the peak at 80 ppm
accounting for the presence of vinylic carbons of the
cyclooctadiene was absent in the CP/ MAS 13C NMR spectrum
of used catalyst 4a. This is as anticipated since cyclooctadiene
will be displaced from the rhodium during formation of the
catalytically active Rh(H)(CO)x(PPh2CH2CH2-)y species under
reaction conditions.
Although the CP/ MAS 13C NMR spectra of fresh and used
catalysts 4a were similar, the CP/ MAS 13C NMR spectra of
fresh and used catalysts 4b and 4c were significantly different
(see Figure 5). Used catalysts 4b and 4c presented an unusual
broad resonance between 10 and 50 ppm. This broad
resonance may be due to the presence of organic species
(such as heptanal) trapped in the pores of MCM-41 and MCM-
20, respectively. Solid-state CP/ MAS 13C NMR studies were
performed to characterize this broad peak. Catalyst 4b was
treated with heptanal, and a CP/ MAS 13C NMR spectrum
was acquired. The broad peaks between 10 and 50 ppm
observed in the CP/ MAS 13C NMR spectra of the used catalyst
4b and catalyst 4b treated with heptanal are identical, which
is consistent with the presence of heptanal trapped within
the pores of the used catalysts anchored on MCM-41. The
presence of a significant amount of heptanal product in the
used catalyst may partially account for the decreases in the
weight percent of phosphorus and metal relative to the fresh
catalyst. No attempts were made to desorb any materials
from used catalysts.
The CP/ MAS 31P NMR spectrum of used platinum-tin
catalysts 7a-7c provided confirmation of the results obtained
from the rhodium complex materials. The spectrum from
used catalyst 7a was substantially similar to that obtained
from the fresh material. However, the spectrum from used
catalysts 7b and 7c revealed a single broad resonance between
10 and 50 ppm, which is again consistent with the presence
of adsorbed heptanal in the catalyst pores. The observance
of the same peak broadening following reaction with the
rhodium complex and the platinum/ tin catalyst provides
further evidence that the peak broadening is a result of
interactions between the reactants (or products) and the
catalyst support.
a
An asterisk (*) indicates insufficient sam ple for duplicate analysis.
areas (SBET) of the different rhodium complexes anchored on
phosphinated silica, MCM-41, and MCM-20 shows the
following trend: SBET catalyst 4a < SBET catalyst 4b < SBET
catalyst 4c. The surface area of these catalysts represents the
total surface of the samples in contact with the external
environment and corresponds to the summation of the
internal surface of the pores plus the external surface of the
particles. As expected, catalysts prepared on supports with
small pores provided the largest surface area. The comparison
of the BET surface areas of catalysts 7a and 7b showed the
same trend already observed with the rhodium catalysts 4a-
4c. However, catalyst 7c presented a smaller BET surface
area than catalyst 7b, which does not follow the previous
trend.
Catalysts were also characterized by elemental analysis
before and after hydroformylation, and the results are shown
in Table 3. The elemental analysis of reused catalyst 4a
revealed a wt % P in the range 2.22-2.28 and a wt % Rh in
the range 2.49-3.04. The calculated molar Rh:P ratio range
was 0.33-0.41. The elemental analysis of used catalysts 4b
and 4c revealed a loss of more than half the content of
rhodium initially measured. The loss of phosphorus was
about 20% of the initial weight of phosphorus measured and
was less pronounced than the loss of rhodium. Catalyst 4c
anchored on MCM-20 lost less rhodium than catalyst 4b
anchored on MCM-41. The elemental analysis of used catalyst
7a shows that the weight percent of phosphorus actually
increased after hydroformylation whereas the weight percent
of Pt dramatically dropped by a factor of almost 4. The
increase in the wt % P after hydroformylation was also
observed with rhodium catalysts. Used catalyst 7b retained
approximately the same amount of wt % P but lost almost
all its Pt, resulting in a very low calculated molar Pt:P ratio
following reaction. On the contrary, used catalyst 7c had less
loss of phosphorus and platinum.
The large loss of rhodium and phosphorus is a concern
for reuse of the catalyst; however, this aspect was not
specifically addressed in the current research. Nevertheless,
previous data (13) indicated that there was no loss of activity
for repeated experiments using the same catalyst material,
for rhodium complex supported on silica materials. This result
is consistent with our current hypothesis that some unreacted
phosphine reagent was adsorbed onto the support, not
chemically tethered, which was simply extracted from the
support during the catalytic reaction. Similarly, rhodium
could have complexed to some of this “unattached” phos-
phine to form complexes that could also have been extracted
from the support under reaction conditions.
Reaction Analysis. Figure 6 shows the typical progress of
1-hexene hydroformylation catalyzed by a supported rho-
dium catalyst. The 1-hexene concentration decreases as the
hydroformylation products 2-ethylpentanal, 2-methylhexa-
9
VOL. 37, NO. 23, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 5 4 2 7

fluctuations due to variation of rhodium concentration in
experimental runs. The rhodium catalyst supported on
phosphinated MCM-20 provided a greater aldehyde yield
than that supported on MCM-41, which was in turn greater
than the aldehyde yield obtained when the catalyst was
supported on phosphinated silica. Since the surface area is
greatest for MCM-20-type catalysts, it is possible that
improved yield was simply due to higher surface area in the
support, which might lead to better distribution of the
rhodium catalyst. Measurement of rhodium “dispersion”
using the hydrogen uptake method indicates that dispersion
increases in the order of rhodium on phosphinated silica <
rhodium on MCM-41 < rhodium on MCM-20. Greater
availability of rhodium on the surface of the catalyst support
could provide the observed increase in the normalized
aldehyde yield.
FIGURE 6. Typical reaction concentration as a function of time for
rhodium catalysts on phosphinated silica and MCM-41 at 184 atm,
In addition to increased aldehyde yield, higher regiose-
lectivity was obtained by decreasing the average pore size of
the support. The rhodium supported on phosphinated MCM-
20 provided the highest regioselectivity. It is likely that the
small pore size of the MCM-20 support provides a reaction
cavity that is not desirable for formation of the branched
aldehyde, leading to the increased selectivity (29).
100 °C, [1-hexene]0 ) 0.097 M, [CO]0 ) [H ] ) 0.7 M.
2 0
Chemoselectivity, which was previously defined as the
ratio of aldehyde products to the isomerization products,
was also investigated and is shown in Figure 8c for the
rhodium catalysts. Initially, chemoselectivity was less than
0.5, indicating that more isomerization than hydroformyla-
tion products were obtained. However, the chemoselectivity
increased over time as the internal olefins were slowly
converted to aldehyde products. Since the internal olefin
cannot produce the linear aldehyde, the increasing chemose-
lectivity is accompanied by a decrease in the regioselectivity,
as more branched aldehydes are produced.
FIGURE 7. Reaction network demonstrating the pathways for the
hydroformylation of 1-hexene.
nal, and heptanal increase. The isomerization products, trans-
2-hexene and cis-2-hexene, eventually achieve a maximum
concentration, which declines as the reaction proceeds. The
hydrogenation product n-hexane behaves in the same
manner as the isomerization products. 1-Heptanol was never
observed in any of the experimental runs for any of the
catalysts evaluated. All samples taken prior to 30 min are
believed to reflect a brief nonhomogeneous condition in the
reactor due to the addition of 1-hexene and have been
observed previously (13, 14). Subsequent samples taken after
30 min demonstrate a smooth decline in the concentration
of 1-hexene.
In addition to rhodium catalysts on mesoporous supports,
platinum/ tin catalysts on mesoporous supports were also
investigated. Figure 9a compares the normalized yield, and
Figure 9b compares the regioselectivity of hydroformylation
of 1-hexene with platinum/ tin catalysts on phosphinated
silica, MCM-41, and MCM-20. The platinum/ tin on phos-
phinated MCM-20 support shows the highest normalized
yield, followed by the platinum/ tin catalyst on phosphinated
MCM-41, which is slightly higher than that of the phosphi-
nated silica-supported platinum/ tin system. It is believed
that catalysts anchored on mesoporous supports are more
active than the analogous catalysts anchored on silica
supports because of the increased surface area of the support.
Figure 7 provides a reaction pathway description for the
observed product profile, which has been used previously to
describe the hydroformylation of 1-hexene by supported
rhodium catalysts in scCO2. The reaction network shows that
hydroformylation of 1-hexene produces both heptanal (the
desired product), 2-methylhexanal, and isomerization prod-
ucts cis- and trans-2-hexene. The internal olefins can
subsequently be converted to 2-ethylpentanal and 2-meth-
ylhexanal but cannot directly produce the desired heptanal
product. Regioselectivity (or L/ B) is defined according to these
pathways as the formation of heptanal relative to the
formation of the branched aldehydes, and chemoselectivity
is defined as the formation of all aldehydes relative to the
isomerization products.
The regioselectivity of the platinum/ tin catalyst supported
on phosphinated MCM-20 was the highest observed in all
of the current experiments, followed by platinum on MCM-
41 and silica, respectively. This suggests that the formation
of linear aldehyde is enhanced relative to the formation of
branched aldehydes by the decreasing pore size of the
support. It is thought that the shape selectivity induced by
use of a smaller average pore size causes preferential
formation of the less bulky linear product.
As observed from Figures 8a and 9a, the yields obtained
from the rhodium catalysts were greater than that obtained
from the platinum/ tin catalysts by a full order of magnitude.
However, in both cases, catalysts prepared on supports with
smaller average pore size provided greater aldehyde yields.
Figures 8b and 9b show that platinum/ tin catalysts are
generally more regioselective, but that decreasing the average
pore size of the support always increased the regioselectivity
of the catalyst.
The effect of support pore size was investigated by
comparing rhodium catalysts supported on phosphinated
silica, MCM-41, and MCM-20. In all cases, the temperature
was maintained at 100 °C, the initial concentration of
1-hexene was 0.097 M, and the syngas ratio was 7:1 relative
to the substrate. Figure 8a compares the normalized yield,
and Figure 8b compares the regioselectivity of hydrofor-
mylation of 1-hexene with rhodium supported on phosphi-
nated silica, MCM-41, and MCM-20. The yields were
normalized to the mass of rhodium in order to eliminate
Figure 9c compares the chemoselectivity of platinum/ tin
catalyst on phosphinated silica, MCM-41, and MCM-20. In
the case of the platinum/ tin catalyst system, the initial
chemoselectivity was near 1 and did not change substantially
9
5 4 2 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 23, 2003

FIGURE 8. (a) Comparison of yield as a function of time for rhodium catalysts (normalized to mass of rhodium) on phosphinated silica,
MCM-41, and MCM-20 at 184 atm, 100 °C, [1-hexene]0 ) 0.097 M, [CO]0 ) [H ] ) 0.7 M. (b) Comparison of regioselectivity as a function
2 0
of time of rhodium catalysts on phosphinated silica, MCM-41, and MCM-20 at 184 atm, 100 °C, [1-hexene]0 ) 0.097 M, [CO]0 ) [H ] )
2 0
0.7 M. (c) Comparison of chemoselectivity as a function of time for rhodium catalysts on phosphinated silica, MCM-41, and MCM-20 at
184 atm, 100 °C, [1-hexene]0 ) 0.097 M, [CO]0 ) [H ] ) 0.7 M.
2 0
over time. Thus, less isomerization products are produced
from the platinum/ tin catalyst than were produced using
the rhodium-supported materials.
In addition, it is noted that the regioselectivity obtained
with the platinum/ tin catalyst system was essentially constant
throughout the reaction. On the basis of the pathways shown
9
VOL. 37, NO. 23, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 5 4 2 9

FIGURE 9. (a) Comparison of yield as a function of time for platinum/tin catalyst systems (normalized to the mass of platinum) on
phosphinated silica, MCM-41, and MCM-20 at 184 atm, 100 °C, [1-hexene]0 ) 0.097 M, [CO]0 ) [H ] ) 0.7 M. (b) Comparison of regioselectivity
2 0
as a function of time of platinum/tin catalyst systems on phosphinated silica, MCM-41, and MCM-20 at 184 atm, 100 °C, [1-hexene]0 )
0.097 M, [CO]0 ) [H ] ) 0.7 M. (c) Comparison of chemoselectivity as a function of time of platinum/tin catalyst systems on phosphinated
2 0
silica, MCM-41, and MCM-20 at 184 atm, 100 °C, [1-hexene]0 ) 0.097 M, [CO]0 ) [H ] ) 0.7 M.
2 0
9
5 4 3 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 23, 2003

(11) De Vos, D. E.; Dams, M.; Sels, B. F.; Jacobs, P. A. Chem. Rev.
2002, 102, 3615-3640.
previously, it is thus possible to conclude that the platinum/
tin catalyst is less effective in converting the internal olefin
to aldehyde product, just as we have already observed the
decreased activity for the platinum/ tin catalyst for hydro-
formylation of the terminal olefin relative to the rhodium
system. Thus, we conclude that the highest selectivity to the
desired linear aldehyde can be obtained through the use of
mesoporous MCM-20-supported platinum/ tin catalysts,
although rhodium catalyst supported on mesoporous MCM-
20 provides reasonable selectivity with the highest yield.
(12) Lu, Z.; Lindner, E.; Mayer, H. A. Chem. Rev. 2002, 102, 3543-
3578.
(13) Tadd, A. R.; Marteel, A. E.; Mason, M. R.; Davies, J. A.; Abraham,
M. A. Ind. Eng. Chem. Res. 2002, 41, 4514-4522.
(14) Tadd, A. R.; Marteel, A. E.; Mason, M. R.; Davies, J. A.; Abraham,
M. A. J. Supercrit. Fluids 2003, 25, 183-196.
(15) Hemminger, O.; Marteel, A. E.; Mason, M. R.; Davies, J. A.; Tadd,
A. R.; Abraham, M. A. Green Chem. 2002, 4, 507-512.
(16) Beck, J. S.; Vartuli, J. C.; Roth, W. J.; Leonowicz, M. E.; Kresge,
C. T.; Schmitt, K. D.; Chu, C. T. W.; Olson, D. H.; Sheppard, E.
W.; McCullen, S. B.; Higgins, J. B.; Schlenker, J. L. J. Am. Chem.
Soc. 1992, 114, 10834-10843.
Acknowledgments
Although the research described in this paper was supported
by the Environmental Protection Agency under Grant
R-82820601-0, it has not been subjected to the Agency’s review
and therefore does not necessarily reflect the views of the
Agency, thus no endorsement of this research is implied. We
also recognize the support of the PQ Corporation for donation
of sodium silicate and Dr. Yong-Wah Kim for assistance with
solid-state NMR spectroscopy.
(17) Tanev, P. T.; Pinnavaia, T. J. Science 1995, 267, 865-867.
(18) Homs, N.; Clos, N.; Muller, G.; Sales, J.; de la Piscina, P. R. J. Mol.
Catal. 1992, 74, 401-408.
(19) Guo, I.; Hanson, B. E.; Toth, I.; Davis, M. E. J. Mol. Catal. 1991,
70, 363-368.
(20) Marteel, A. E.; Davies, J. A.; Mason, M. R.; Tack, T.; Bektesevic,
S.; Abraham, M. A. Catal. Commun. 2003, 4, 309-314.
(21) Rathke, J. W.; Klinger, R. J.; Krause, T. R. Organometallics 1991,
10, 1350-1355.
(22) Palo, D. R.; Erkey, C. Organometallics 2000, 19, 81-86.
(23) Sellin, M. F.; Cole-Hamilton, D. J. J. Chem. Soc., Dalton Trans.
2000, 1681-1683.
(24) Sellin, M. F.; Bach, I.; Webster, J. M.; Montilla, F.; Rosa, V.; Aviles,
T.; Poliakoff, M.; Cole-Hamilton, D. J. J. Chem. Soc., Dalton Trans.
2002, 4569-4576.
Literature Cited
(1) Beller, M.; Cornils, B.; Frohning, C. D.; Kohlpaintner, C. W. J.
Mol. Catal. A 1995, 104, 17-85.
(2) Trzeciak, A. M.; Zio´ łkowski, J. J. Coord. Chem. Rev. 1999, 190-
192, 883-900.
(3) Van Leeuwen, P. W. N. M., Claver, C., Eds. Rhodium Catalyzed
Hydroformylation, 1st ed.; Kluwer Academic Publishers: Boston,
MA, 2000.
(4) Leadbeater, N. E.; Marco, M. Chem. Rev. 2002, 102, 3217-3274.
(5) McNamara, C. A.; Dixon, M. J.; Bradley, M. Chem. Rev. 2002,
102, 3275-3300.
(6) Nozaki, K.; Itoi, Y.; Shibahara, F.; Shirakawa, E.; Ohta, T.; Takaya,
H.; Hiyama, T. J. Am. Chem. Soc. 1998, 120, 4051-4052.
(7) Ajjou, A. N.; Alper, H. J. Am. Chem. Soc. 1998, 120, 1466-1468.
(8) Hanaoka, T.; Arakawa, H.; Matsuzaki, T.; Sugi, Y.; Kanno, K.;
Abe, Y. Catal. Today 2000, 58, 271-280.
(9) Song, C. E.; Lee, S. Chem. Rev. 2002, 102, 3495-3524.
(10) Wight, A. P.; Davis, M. E. Chem. Rev. 2002, 102, 3589-3614.
(25) Koch, D.; Leitner, W. J. Am. Chem. Soc. 1998, 120, 13398-13404.
(26) Bemi, L.; Clark, H. C.; Davies, J. A.; Fyfe, C. A.; Wasylishen, R.
E. J. Am. Chem. Soc. 1982, 104, 438-445.
(27) Giordano, G.; Crabtree, R. H. Inorg. Synth. 1990, 28, 88-90.
(28) Komoroski, R. A.; Magistro, A. J.; Nicholas, P. P. Inorg. Chem.
1986, 25, 3917-3925.
(29) Baiker, A. Chem. Rev. 1999, 99, 453-473.
Received for review June 2, 2003. Revised manuscript re-
ceived July 29, 2003. Accepted August 14, 2003.
ES034544N
9
VOL. 37, NO. 23, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 5 4 3 1