Efficient hydroformylation in dense carbon dioxide using phosphorus ligands without perfluoroalkyl substituents

Article abstract of DOI:10.1002/adsc.200900008

Rhodium catalysts modified with triphenylphosphine, triphenyl phosphite, and tris(2,4-ditrrt-butylphenyl) phosphite have been evaluated for their performance in the hydroformylation of 1octene using carbon dioxide as the solvent. It is demonstrated that these catalysts are very efficient for the hydroformylation in carbon dioxide, although they are not designed for use in this medium. In particular, the catalyst prepared in situ from dicarbonyl(2,4-pentanedione)rhodium(I) and tris(2,4-di-tert-butyl-phenyl) phosphite gave rise to an initial turnover frequency in excess of 3x 10 ⁴ mol_aldehyde mol_Rh h^-1. Such a reaction rate is unprecedented for hydroformylation in supercritical carbon dioxide-rich reaction mixtures.

Full text of DOI:10.1002/adsc.200900008

UPDATES
DOI: 10.1002/adsc.200900008
Efficient Hydroformylation in Dense Carbon Dioxide using
Phosphorus Ligands without Perfluoroalkyl Substituents
Ard C. J. Koeken,a,* Nieck E. Benes,b Leo J. P. van den Broeke,a
and Jos T. F. Keurentjesa
a
Process Development Group, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
Fax : (+31)-40-244-6104; e-mail: a.c.j.koeken@tue.nl
Membrane Technology Group, Faculty of Science and Technology, Institute of Mechanics, Processes and Control Twente
b
(IMPACT), University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands
Received: January 7, 2009; Published online: May 28, 2009
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/adsc.200900008.
Abstract: Rhodium catalysts modified with triphe-
nylphosphine, triphenyl phosphite, and tris(2,4-di-
tert-butylphenyl) phosphite have been evaluated for
their performance in the hydroformylation of 1-
octene using carbon dioxide as the solvent. It is
demonstrated that these catalysts are very efficient
for the hydroformylation in carbon dioxide, al-
though they are not designed for use in this
medium. In particular, the catalyst prepared in situ
from dicarbonyl(2,4-pentanedione)rhodium(I) and
tris(2,4-di-tert-butyl-phenyl) phosphite gave rise to
an initial turnover frequency in excess of 3ꢀ
104 molaldehyde molRh hÀ1. Such a reaction rate is un-
precedented for hydroformylation in supercritical
carbon dioxide-rich reaction mixtures.
improved mass transfer rates, an environmentally
benign character, convenient separation from reaction
products by depressurization, and the possibility to
create a monophasic reaction system.[2a,4] Further-
more, the ongoing development of monodentate and
multidentate phosphorus ligands has resulted in more
efficient, selective, and versatile hydroformylation cat-
alysts.[5]
In general, many well-known hydroformylation cat-
alysts have a poor solubility in carbon dioxide-rich re-
action mixtures.[4a] The catalyst solubility in carbon di-
oxide can be improved by attaching perfluoroalkyl
substituents on the phosphorus ligands of the cata-
lyst.[6] Because the cost of perfluoroalkyl reagent
starting materials increases with chain length,[7] the
development of inexpensive “non-fluorous” CO2-
philic ligands is receiving increasing attention.[8]
Other ways to increase the solubility of higher molec-
ular weight substances in carbon dioxide rich media
are based on the use of cosolvents[9] or auxiliary
agents like peracetylated b-cyclodextrins.[10] More-
over, for catalytic applications organic solvents ex-
panded with carbon dioxide seem to combine both
the advantages of organic solvents and carbon diox-
ide.[11]
Keywords: carbon dioxide; homogeneous catalysis;
hydroformylation; phosphorus ligands; rhodium;
supercritical fluids
Introduction
An important example of homogeneous catalysis car-
Examples of well-known hydroformylation catalysts
ried out on an industrial scale is the hydroformylation are rhodium complexes with phosphorus ligands like
reaction.[1] Generally, cobalt or rhodium catalysts are triphenylphosphine or triphenyl phosphite. For a total
used in the hydroformylation reaction, to convert an pressure below 30 MPa the use of triphenylphosphine
alkene, carbon monoxide, and hydrogen into an alde- and triphenyl phosphite for the hydroformylation of
hyde product with a high atom economy.[1]
1-alkenes in carbon dioxide has resulted in moderate
During recent years, the use of carbon dioxide as a turnover frequencies[12] (TOF) in the order of 100–
solvent for homogeneously catalyzed hydroformyla- 200 molaldehyde molRhÀ1 hÀ1.[8a,b,e] One of the highest
tion reactions[2] and as a means to influence reaction TOF values reported in the literature for the hydro-
selectivity or activity[3] has received considerable at- formylation of a 1-alkene with a triphenylphosphine-
tention. Advantages of using carbon dioxide as a sol- modified rhodium catalyst in carbon dioxide is
vent over traditional liquid (organic) solvents include 174 molaldehyde molRhÀ1 hÀ1 at 808C.[8a] Recently, Galia
1442
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1442 – 1450

Efficient Hydroformylation in Dense Carbon Dioxide using Phosphorus Ligands
UPDATES
et al. have discussed the use of triphenylphosphine for
the hydroformylation of 1-octene in carbon dioxide.
They have reported
a
TOF of 116 molaldehyde
molRhÀ1 hÀ1 at an initial total pressure in excess of
31 MPa and 608C.[8d] We have applied triphenylphos-
phine in the rhodium-catalyzed hydroformylation of
1-octene in carbon dioxide, which resulted in an initial
rate of 460 molaldehyde molRhÀ1 hÀ1 at 708C and a initial
total pressure of about 50 MPa.[13] It is noted that in
these examples the reaction conditions vary with re-
spect to the temperature, phase behaviour, and the
amount of rhodium, phosphorus ligand, and reactants
that have been used. This makes a direct comparison
of the different results for the TOF difficult. Howev-
er, it appears that the reaction conditions we have
used allow for an effective use of triphenylphosphine-
modified rhodium catalysts.
Figure 1. Pressure and temperature as a function of time
during the hydroformylation in carbon dioxide with L1 as
the ligand (Table 1, entry 1).
Besides modifying the catalytic complex it is also
possible to tune the properties of the supercritical Results and Discussion
fluid (SCF) to adjust the solubility and reactivity. The
density of an SCF is an important parameter with re- During the batch experiments the temperature and
spect to the solubility of large molecules. Usually, an pressure of the reaction mixture were monitored and
increase in fluid density results in a higher solubili- logged. In Figure 1 a typical pressure and temperature
ty.[6f] Increasing the temperature while keeping the profile is shown. These results were obtained for L1
fluid density constant results in a higher pressure and (vide infra, entry 1 in Table 1). In all cases, 1-octene
a significant increase in phosphorus ligand solubili- was injected at t=0 h to start the reaction. Before the
ty.[6f] The combination of high temperature and high hydroformylation reaction was started, the catalyst
pressure also facilitates the existence of a single phase was prepared in situ (t=À1.0 to 0 h) in a supercritical
reaction mixture.[14,15] Laboratory-scale experiments at mixture containing 14.4 mol LÀ1 carbon dioxide
pressures in excess of 30 MPa are feasible and have (Table 1), 1 mol LÀ1 carbon monoxide and 1 mol LÀ1
been reported,[16–18] but applying high pressure to im- hydrogen. For the phosphite ligands (L2 and L3) a
prove catalyst solubility in an SCF is not a common somewhat longer catalyst formation time was applied
strategy. This might be attributed to the perception (see Supporting Information). In Figure 1 it can be
that on an industrial scale high pressure processing is seen that conversion of 0.97 mol LÀ1 1-octene (0.105
inherently uneconomical or unpractical. Nevertheless, mol 1-octene in 0.108 L reactor volume), 1 mol LÀ1
there are examples of commercial large-scale process- carbon monoxide and 1 mol LÀ1 hydrogen results in a
es carried out at high pressure.[19]
decrease in pressure of 24 MPa. It is noted that about
The main objective of this work is to study the per- 10% of the change in pressure can be attributed to
formance of a rhodium catalyst employing ligands sampling.[18]
without perfluoroalkyl substituents for the hydrofor-
The period of time to allow for the in situ forma-
mylation of 1-octene in carbon dioxide using a combi- tion of the catalyst prior to reaction is an aspect in
nation of high pressure (initial pressureꢀ30 MPa) which our experimental procedure differs from most
and temperature (ꢀ708C). In particular, we have methods reported in the literature. Often the catalyst
studied the use of hydroformylation catalysts generat- precursors and the reactants are already mixed to-
ed in situ from triphenylphosphine (L1), triphenyl gether before the desired reaction conditions are
phosphite (L2), or the bulky phosphite tris(2,4-di-tert- reached. The advantage of the procedure we use is
butyl-phenyl) phosphite (L3) and the metal complex that possible influences of catalyst formation on the
[Rh(CO)2acac] (acac=2,4-pentanedione). For com- kinetics of the hydroformylation reaction are mini-
parison, results are presented for the CO2-philic tris- mized.
A
In order to compare the ligands L1 to L4 we used a
has been used for the hydroformylation in organic reaction temperature of 708C and an initial maximum
solvents[20] and CO2 expanded organic solvents.[21] To pressure of about 50 MPa as reference conditions. It
the best of our knowledge the application of the has been established that with these conditions a
bulky phosphite L3 for the hydroformylation in dense single phase reaction system is present throughout the
CO2 has not been reported before.
extent of the reaction.[13] In Scheme 1, the most prob-
able reaction pathways are depicted for the hydrofor-
mylation of 1-octene.
Adv. Synth. Catal. 2009, 351, 1442 – 1450
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1443

Ard C. J. Koeken et al.
UPDATES
Table 1. Summary of conditions and results.[a]
[b]
[c,d]
No. L
T
pmax
n1-octene
1-oct- L:Rh TOF1-octene
TOF[c,e] tr
[h] [%]
Conv.
Selectivity [%]
linear branched octene iso-
l:b[f]
[-]
[8C] [MPa] [mmol] ene:Rh
ald.
ald.
mers, octane
1
2
L1 70 50.2
L2 70 50.2
105
105
105
105
105
100
106
2030
1910
1940
1970
2000
3660
1970
4.2
5.1
4.0
4.1
-
832
(1190)[g]
819
A
1820
1.01 44
74
73
68
67
58
58
66
65
47
50
76
76
63
62
25
24
28
26
26
33
23
24
24
31
24
22
29
34
1
3
4
7
16
9
12
11
29
19
1
2
8
3
2.9
3.0
2.4
2.6
2.2
1.7
2.9
2.7
1.9
1.6
3.2
3.4
2.2
1.8
G
1890
37,400
7107
3120
1.04 83
3.11 99
3[i] L3 70 49.2
31,400 0.05 88
1.00 100
6710
4
5
L4 70 49.3
70 49.8
0.98 98
3.03 100
0.96 99
–
2090
A
(3600)[g] 2.96 99
6[j] L1 90 46.7
7[i,k] L3 70 25.7
99
4.2
2840
2640
1.00 70
2.98 93
37,000
34,100 0.04 67
1.05 100
[a]
General applied conditions: CO=109 mmol, H2 =109 mmol, Vreactor =0.108 L, [Rh(CO)2acac]=53 mmol (average). CO2
amount=68 g (average value for entries 1–5). The exact catalyst preparation conditions are shown in the pressure and
temperature profiles given in the Supporting Information. The results for the conversion, selectivity, and l:b are based on
the composition of the samples obtained by GC analysis.
[b]
[c]
Maximum pressure reached upon injection of 1-octene.
Initial turnover frequency (TOF); obtained from multiplying 1-octene:Rh with the slope of a linear fit through conver-
sion (TOF1-octene) or yield (TOFald) data up to a conversion of 60%. For entries 3, 4, and 7 the initial rates are based on
the composition of first sample and the time of sampling.
[d]
[e]
[f]
[mol1-octene molRhÀ1 hÀ1].
[molaldehyde molRhÀ1 hÀ1].
The linear to branched ratio, l:b, is calculated by dividing the yield of linear aldehyde by the yield of all branched alde-
hyde products.
Maximum turnover frequency observed during the course of the reaction. In the other cases the initial TOF is close to
the maximum TOF.
At the onset of the reaction the temperature increased from 70 to about 818C.
[Rh(CO)2acac]=27 mmol, CO2 amount=62 g.
CO2 amount=52 g.
[g]
[i]
[j]
[k]
In Figure 2 the concentration profiles for five differ-
ent situations are compared. In Figure 2a–d the pro-
files are given for the situation that one of the ligands,
L1 to L4, has been applied. In Figure 2e the results
are shown for the situation where no additional phos-
phorus ligand has been applied. In Table 1 an over-
view of the corresponding conditions and the main re-
sults are given.
The reaction using L1 as the modifying ligand pro-
ceeds with a high chemoselectivity for aldehydes with
nonanal and 2-methyloctanal as the main products
(Table 1, entry 1, and Figure 2a). After a reaction
time of one hour the selectivities for nonanal and 2-
methyloctanal were 74% and 25%, respectively. At
the moderate temperature of 708C this aldehyde
product distribution can be considered to be common
for the L1-modified rhodium-catalyzed hydroformyla-
tion of 1-octene.[22] Using L1 only traces of 2-ethyl-
heptanal (selectivity=0.1%) and 2-propylhexanal (se-
Scheme 1. Reaction scheme for the hydroformylation of 1-
octene. Following the coordination of 1-octene to the rhodi-
um complex, hydride insertion can take place to form either
the 1-octyl or 2-octyl rhodium intermediate. The main reac-
tion products nonanal and 2-methyloctanal, depicted in the
grey shaded area, are formed through the catalytic steps in-
lectivity=0.01%) were observed after three hours volving the 1-octyl and 2-octyl intermediates, respectively.
1444
asc.wiley-vch.de
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1442 – 1450

Efficient Hydroformylation in Dense Carbon Dioxide using Phosphorus Ligands
UPDATES
Figure 2. Kinetics of the rhodium catalyzed hydroformylation of 1-octene modified by L1 (a), L2 (b), L3 (c), L4 (d), or un-
~
*
modified (e). Concentration profiles are obtained by sampling and GC analysis. =1-octene, =internal octene isomers,
!
&
^
octane, =nonanal, =2-methyloctanal, =2-ethylheptanal, " =2-propylhexanal. Reaction conditions are given in Table 1
(entries 1 to 5).
reaction time (tr). It is noted that the amounts of reac- a TOF of 1820 molaldehyde molRhÀ1 hÀ1 was obtained,
tants were near stoichiometric. At a high aldehyde which was considerable higher than the TOF obtained
yield, and a correspondingly low carbon monoxide with L1 (Table 1, entry 2). Additionally, a lower selec-
and hydrogen concentration, isomerization of octene tivity for the linear product and a higher selectivity
can become the dominant reaction.[22] However, for for the octene isomers was obtained, as compared to
L1 the concentration of the octene isomers reached a the results obtained with L1 (Figure 2b, Table 1,
significant value only after two hours of reaction. Fur- entry 2). For the L2-modified catalyst traces of 2-eth-
thermore,
a
maximum reaction rate of 1190 ylheptanal (selectivity=0.2%) and 2-propylhexanal
molaldehyde molRhÀ1 hÀ1 is observed, which is unprece- (selectivity=0.01%) were observed after three hours
dented for the hydroformylation of 1-octene in dense reaction time. Finally, also for L2 the values for the
carbon dioxide (Table 1, entry 1).
TOF seem to be exceptionally high in comparison
The application of L2 results in a different type of with literature data on hydroformylation in supercriti-
kinetics, as compared to the application of L1. For L2 cal media.[8e]
Adv. Synth. Catal. 2009, 351, 1442 – 1450
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1445

Ard C. J. Koeken et al.
UPDATES
In Figure 2c the results obtained with L3 are depict- for 2-propylhexanal. Clearly, after approximately
ed. It is clear that the use of L3 results in a very high three hours reaction time the selectivity for the
hydroformylation rate in combination with an isomer- branched aldehydes was lower for the unmodified cat-
ization rate considerably higher than that observed alyst in comparison with the L3-modified catalyst.
for L2. The yield of aldehydes determined from GC
Assuming that the catalytic species are dissolved
(gas chromatography) analysis was equal to 74% and homogeneously dispersed, the differences in re-
within 3 min of reaction time (Table 1, entry 3, tr = sults obtained with the four different ligands can be
0.05 h). This corresponds to an initial TOF of explained in terms of electronic and steric ligand ef-
31,400 molaldehyde molRhÀ1 hÀ1 at the relatively mild ini- fects. It seems that for the hydroformylation in super-
tial reaction temperature of 708C. At this reaction critical CO2 rich media a similar reasoning holds as
time the selectivity for 2-methyloctanal was 26% and for the hydroformylation in organic solvents. The
traces of 2-ethylheptanal (selectivity=0.1%) and 2- electron density on the phosphorus of L2 is lower
propylhexanal (selectivity=0.01%) were observed. than the electron density on the phosphorus of L1,
The normalized concentration of octene isomers in- which usually results in a faster hydroformylation.[24,25]
creased to 16% at 11 min (0.18 h) and then decreased From Table 1 it follows that indeed L2 leads to a
to a value of 4% at tr =2.93 h. At a high 1-octene con- more active catalyst at a moderate ligand excess, as
version, the Rh catalyst modified with L3 also started compared to L1. L3 is a bulky phosphite and has tert-
to convert the internal octene isomers. At the end of butyl substituents at the ortho positions of the aryl
the reaction (tr =2.93 h) the yield of 2-methyloctanal rings. This results in a ligand with a cone angle[26] con-
was 31%, the yield of 2-ethylheptanal was 4%, and siderable larger than that of L2. For hydroformylation
the yield of 2-propylhexanal was 3%.The total alde- in organic solvents it has been demonstrated that the
hyde yield of 96% at the end of the reaction is quite use of this type of bulky ligand results in an extremely
remarkable, considering that only a small excess of active catalyst.[24] Under hydroformylation conditions
carbon monoxide and hydrogen is present at the start usually only one L3 can coordinate to rhodium.[25] L2
of the reaction. In particular, for the catalyst modified is less sterically demanding than L3 and therefore can
with L3 one would expect that at a low carbon mon- even completely replace carbon monoxide as a ligand.
oxide and hydrogen concentrations isomerization Using a large excess of L2 can result in a catalyst with
should be more prevalent than the hydroformylation a relatively low activity.[24]
of internal octenes.[23] Finally, it should be noted that
Our results obtained with L2 and L3, under super-
in the initial stage of the reaction the fluid tempera- critical conditions, reflect the difference in steric con-
ture rose from 70 to about 818C. Thus, the values for formation between L2 and L3. The catalyst based on
TOF given in Table 1 do not strictly represent the ac- a ligand closely related to L3, tris(2-tert-butyl-4-meth-
tivity of the catalyst at 708C.
ylphenyl) phosphite, has a well-known ability to hy-
The use of L4, Figure 2d, has resulted in a more droformylate substituted alkenes.[27] The decrease in
active catalyst than the application of L1, L2, or the concentration of internal octene isomers over time
situation where no ligand was used. The rhodium cat- and the yield of branched aldehydes obtained after
alyst based on L4 has the ability to hydroformylate in- three hours of reaction (Figure 2c) confirm this fea-
ternal octenes.[13a,18] However, when comparing Fig- ture. Trifluoromethyl substituents also have an elec-
ure 2c and Figure 2d, the use of L3 appeared to result tron-withdrawing effect and it is known that the appli-
in a faster hydroformylation of internal alkenes than cation of L4 results in a more active catalyst, as com-
the use of L4. For L4 after three hours reaction time pared to the use of L1.[6e,13a,28,29] Consequently, the dif-
selectivities of 0.9% and 0.2% for 2-ethylhept
2-propylhexanal were found, respectively.
ACHTUNGTRENNUNG
For the fifth case, where the unmodified Rh-cata- action rate data reported in the literature for related
lyst has been used (Figure 2e, Table 1, entry 5) the re- catalysts applied in organic solvents.
sults for the hydroformylation show the lowest selec-
Our reactor does not allow for a visual observation
tivity for nonanal and a significant tendency for iso- of the phase behaviour of the reaction mixture. For
merization of 1-octene. A maximum normalized con- L1 and L4 it can be readily assumed that the catalyst
centration of 28% of octene isomers at tr =0.96 h was is dissolved in the supercritical reaction mixture based
found (Figure 2e) and the selectivity for nonanal was on literature reports.[6b,e,f,g,8d] However, for L2 and L3
50% at tr =2.96 h (Table 1, entry 5). For the reaction no detailed studies for the solubility of these two li-
with the unmodified catalyst the hydroformylation of gands in an SCF have been reported. Sellin and Cole-
internal octenes appeared to be slower than for the Hamilton observed that ligand L2 and the catalyst
situation where the L3-modified catalyst has been ap- formed a separate liquid phase at 1008C and 22 MPa
plied. At tr =2.96 h the selectivity for the branched al- total pressure during Rh-catalyzed hydroformylation
dehydes (31% in total) is distributed as follows: 28% of 1-nonene in carbon dioxide.[8e] Furthermore, they
for 2-methyloctanal, 2% for 2-ethylheptanal, and 1% found that other insoluble metal complexes modified
1446
asc.wiley-vch.de
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1442 – 1450

Efficient Hydroformylation in Dense Carbon Dioxide using Phosphorus Ligands
UPDATES
with monodentate phosphorus ligands could catalyze
the hydroformylation although with moderate values
for the TOF.[8e] Typically, they observed values for the
TOF between 25 and 120 molaldehyde molRhÀ1 hÀ1. Com-
pared to the conditions used by Cole-Hamilton and
co-workers, we used a significantly higher CO2 densi-
ty. Based on literature dealing with the solubility of
phosphorus ligands and the corresponding catalysts it
is expected that under the conditions we have applied
the solubility of the rhodium catalyst modified with
L2 has been higher than under the conditions used by
Cole-Hamilton and co-workers.[6f,g]
As the result obtained with L3 was most remark-
able, we performed a preliminary solubility study. It
Figure 3. The pressure and temperature as a function of
time when using L3 for the hydroformylation of 1-octene
and a lower CO2 amount. The corresponding conditions and
results are given in Table 1, entry 7.
was determined that for pure L3, as ligand and not
the rhodium complex, the saturation concentration in
carbon dioxide at 20 MPa and 708C was about 2.5ꢀ
10À3 mol LÀ1. This implies that pure L3 could be com-
pletely soluble at the reaction conditions applied in
this study. In comparison with L2, L3 contains six ad- (Table 1, entry 6), which is unparalleled for this ligand
ditional tert-butyl substituents. There have been some when using carbon dioxide as a solvent. As a result of
investigations into whether alkylation of ligands can the excess of L1, isomerization is almost completely
lead to an improvement in ligand and catalyst solubil- suppressed and a selectivity of 76% for the linear al-
ity in CO2.[6d,8b,d] For the case of triarylphosphines dehyde is obtained.
Galia et al. concluded that attaching tert-butyl groups
Second, L3 was applied at a lower total pressure.
to L1 did not appear to affect the solubility.[8d] It, By using a smaller amount of CO2 a lower total pres-
therefore, remains unclear whether the rhodium cata- sure was achieved, while the amount of reactants and
lyst based on L2 and L3 dissolves in the supercritical catalyst precursors was kept constant (Table 1,
phase or that a biphasic system, consisting of an or- entry 7). The objectives of using a lower pressure
ganic catalyst rich phase and a carbon dioxide rich were three-fold. First, we wanted to work with an ini-
phase, is formed.
tial total pressure, which was in the range of pressures
The results for the TOF (Table 1, entries 1 to 4) in usually encountered in reports on hydroformylation
combination with the results for the reaction rate and in carbon dioxide. Second, we wanted to establish
the selectivity presented for related catalysts in litera- whether the use of L3 also would result in a high re-
ture suggest that the catalysts are dissolved at the action rate under these conditions. Third, we were in-
conditions we have applied.[8b,c,13a,23,29,30] On the other terested in the global phase behaviour under these
hand, it is known that efficient homogeneous catalysis conditions.
can be carried out using multiphase systems.[31] Deter-
In Figure 3 the pressure-time and temperature-time
mining the distribution of metal complexes between histories are given for the hydroformylation of 1-
the phases and establishing the nature of the catalytic octene using 52 g of CO2 (Table 1, entry 7). After ini-
species in these phases requires further study.
tiation of the reaction (at t=0 h) the pressure de-
In addition to the reference conditions of 708C and creased rapidly from 25.7 MPa to 12.8 MPa within
a maximum initial total pressure of 50 MPa (see 8 min. Again a very high initial TOF of 34,100
Table 1), we have explored two alternative sets of re- molaldehyde molRhÀ1 hÀ1 was obtained (Table 1, entry 7).
action conditions using ligand L1 and L3. These con- Notably, based on the similar initial TOFs observed
ditions and the corresponding results are given as en- using either a lower or higher pressure, it appears that
tries 6 and 7 in Table 1.
catalyst formation at a lower pressure proceeds just as
First, L1 was used with a phosphorus to rhodium effectively as when a higher pressure is used. Based
ratio close to 100 at a temperature of 908C (Table 1, on the total amount of 1-octene injected, 106 mmol,
entry 6). With an increase in temperature from 70 to into the reactor of volume 0.108 L a total concentra-
908C an increase in reaction rate can be expected. tion of about 1 mol LÀ1 can be expected when 1-
However, a higher concentration of L1 usually results octene and the reaction products are homogeneously
in a lower reaction rate in combination with a higher dissolved in a single phase supercritical medium. By
selectivity.[32] The maximum total pressure applied at GC analysis of the samples it was derived that the
908C was lower than for the experiments at 708C. sum of the concentrations of 1-octene and reaction
Despite the large excess of L1 an initial TOF of products at the bottom part of the reactor was about
2640 molaldehyde molRhÀ1 hÀ1
has
been
obtained 2 to 3 mol LÀ1 (see the Experimental Section for de-
Adv. Synth. Catal. 2009, 351, 1442 – 1450
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1447

Ard C. J. Koeken et al.
UPDATES
grade), n-heptane (Merck, analytical grade), the internal
standard n-decane (Aldrich, >99% purity) and the substan-
ces involved in the reaction, n-octane (Aldrich, >99%), 2-
octene (ABCR, mixture of E and Z, 98%) and nonanal
(Fluka, >95%) used for the GC analysis (gas chromatogra-
phy with flame ionization detection) were used as received.
tails on the sampling procedure). This implies that ap-
plying a smaller amount of CO2 has resulted in a two-
phase reaction mixture with an organic rich phase at
the bottom part of the reactor, assuming that the cata-
lyst completely dissolves and does not form a third
phase. For the other cases reported here (Table 1, en-
tries 1–6) a total concentration, sum of 1-octene and
products, close to 1 mol LÀ1 was derived by GC analy-
sis. Using a lower pressure and consequently a two-
phase reaction system (Table 1, entry 7) appeared to
have a positive effect on the selectivity in comparison
with the single phase condition (Table 1, entry 3). At
tr =1.05 h, the selectivity for the linear aldehyde was
62%.
Hydroformylation in Carbon Dioxide
Details on the reactor system, the general procedures ap-
plied during the hydroformylation experiments and the anal-
ysis have been reported in detail in refs.[13,18] Important
note: the risks of using of the very toxic flammable carbon
monoxide gas and the highly flammable hydrogen gas in
combination with the use of high pressure were extensively
assessed. A variety of safety measures were taken including
using detection for carbon monoxide and hydrogen, working
in a fume hood, using pressure relief devices, and using
equipment with a pressure rating 10 to 40 MPa above the
maximum allowable working pressure of the reactor. Blank
reaction runs were performed regularly. In the Supporting
Information the pressure and temperature profiles are given
for the experiments corresponding to entries 1–7 in Table 1.
The procedure for sampling the reactor contents started
with rinsing the contents of the tubing connecting the
sample volume and reactor with a small volume high-pres-
sure syringe pump. The sample could be taken from either
the top or the bottom part of the reactor. For the experi-
ments corresponding to entries 1–6 in Table 1 samples were
taken from the top part of the reactor. For the experiment
corresponding to entry 7 a two-phase reaction system was
anticipated, because a considerably smaller amount of
carbon dioxide was applied in comparison with the other ex-
periments. Therefore, sampling was done from the lower
part of the reactor in that case. After taking the high pres-
sure sample the content of the sample volume (0.192 mL)
was carefully bubbled through a vial with a solution of n-
decane in toluene and afterwards rinsed with additional tol-
uene solution to collect 1-octene and its reaction products
quantitatively. The amount of the toluene solution used,
containing 0.01 mol LÀ1 of n-decane as a standard, was de-
termined. By means of GC analysis the concentration of 1-
octene and the reaction products in the toluene solution
could be determined. Subsequently, the concentrations of 1-
octene and reaction products present in the sample volume
could be calculated.
Conclusions
In conclusion, the reaction conditions reported here,
allow for a more efficient use of the rhodium catalysts
modified with L1, L2 or L3, than the conditions re-
ported previously in other investigations into Rh-cata-
lyzed hydroformylation of long-chain alkenes in the
presence of carbon dioxide. The catalyst derived from
L3 has shown to be a very effective catalyst at differ-
ent carbon dioxide densities. Applying L3 resulted in
the highest TOF reported for homogeneous hydrofor-
mylation catalysts in supercritical CO2-rich systems.
The favourable results obtained with L1, L2, and L3
could indicate that the reaction conditions used here
could be suitable for application of other related com-
mercially available ligands. Therefore, the results re-
ported here can direct to further advancement of the
use of carbon dioxide rich reaction systems for hydro-
formylation and related reactions using ligands with-
out perfluoroalkyl substituents.
Experimental Section
Materials
Carbon dioxide, carbon monoxide, and hydrogen, grades
5.0, 4.7 and 5.0, respectively, were obtained from Hoekloos
(The Netherlands). Prior to use CO2 was passed over a
Messer Oxisorb filter to remove oxygen and moisture. 1-
Octene obtained from Aldrich, was passed over activated
alumina, dried with pre-treated molecular sieves 3 ꢂ (Al-
drich, 4–8 mesh), and stored under argon. The rhodium pre-
cursor, dicarbonyl(2,4-pentanedione)rhodium(I), ([Rh(CO)2
Determination of Solubility of L3
First, L3 (3.5 g) was loaded in the reactor, the reactor was
heated to 708C, and consequently carefully pressurized up
to 20 MPa with carbon dioxide. The system was allowed to
stabilize at a stirring rate of 140 rpm. Samples were taken
from the top of the reactor. The sample volume was depres-
surized and rinsed with n-heptane. The resulting solution of
L3 in heptane was weighed. Subsequently, the solution was
analyzed using a UV-vis spectrophotometer (Shimadzu UV-
2501 PC). The concentration of L3 in heptane was calculat-
ed by means of a calibration line obtained by measuring
heptane solutions with known L3 concentrations.
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
a white to light yellow solid and was supplied by Arkema
(Vlissingen, The Netherlands). Triphenylphosphine (white
to off-white powder, reagent plus, 99% purity), triphenyl
phosphite (colourless liquid, 97% purity), and tris(2,4-di-
tert-butyl-phenyl) phosphite (white powder, 98% purity)
were all obtained from Aldrich. All catalyst precursors were
stored under argon. The solvents toluene (Merck, analytical
1448
asc.wiley-vch.de
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1442 – 1450

Efficient Hydroformylation in Dense Carbon Dioxide using Phosphorus Ligands
UPDATES
[7] W. G. Hollis Jr. , M. G. Poferl, M. D. Wolter, P. A.
Deck, C. Slebodnick, J. Fluorine Chem. 2008, 129, 119–
124.
Acknowledgements
The Netherlands Organization for Scientific Research
(NWO) is acknowledged for the financial support (VIDI
STW.07055). We thank G. O. H. M. Puts for his assistance
with the solubility measurements.
[8] a) Y. Hu, W. Chen, L. Xu, J. Xiao, Organometallics
2001, 20, 3206–3208; b) M. F. Sellin, I. Bach, J. M.
Webster, F. Montilla, V. Rosa, T. Avilꢄs, M. Poliakoff,
D. J. Cole-Hamilton, J. Chem. Soc. Dalton Trans. 2002,
24, 4569–4576; c) C. D. Ablan, D. Sheppard, E. J.
Beckman, M. M. Olmstead, P. G. Jessop, Green Chem.
2005, 7, 590–594; d) A. Galia, A. Cipollina, G. Filardo,
O. Scialdone, M. Ferreira, E. Monflier, J. Supercrit.
Fluids 2008, 46, 63–70; e) M. F. Sellin, D. J. Cole-Ham-
ilton, J. Chem. Soc. Dalton Trans. 2000, 1681–1683.
[9] For example: a) J. M. Dobbs, J. M. Wong, R. J. Lahiere,
K. P. Johnston, Ind. Eng. Chem. Res. 1987, 26, 56–65;
b) J. Xiao, S. C. A. Nefkens, P. G. Jessop, T. Ikariya, R.
Noyori, Tetrahedron Lett. 1996, 37, 2813–2816.
[10] A. Galia, E. C. Navarre, O. Scialdone, M. Ferreira, G.
Filardo, S. Tilloy, E. Monflier, J. Phys. Chem. B 2007,
111, 2573–2578.
References
[1] C. D. Frohning, C. W. Kohlpaintner, H.-W. Bohnen, in:
Applied Homogeneous Catalysis, 2nd edn., Vol. 1,
(Eds.: B. Cornils, W. A. Herrmann), Wiley-VCH, Wein-
heim, 2002, pp 31–103.
[2] a) P. G. Jessop, W. Leitner, in: Chemical Synthesis
Using Supercritical Fluids, (Eds.: P. G. Jessop, W. Leit-
ner), Wiley-VCH, Weinheim, 1999, pp 1–36; b) W.
Leitner, Acc. Chem. Res. 2002, 35, 746–756; c) S. Bek-
tesevic, A. M. Kleman, A. E. Marteel-Parrish, M. A.
Abraham, J. Supercrit. Fluids 2006, 38, 232–241;
d) P. G. Jessop, J. Supercrit. Fluids 2006, 38, 211–231;
e) E. J. Beckman, J. Supercrit. Fluids 2004, 28, 121–
191; f) A. Stobrawe, P. Makarczyk, C. Maillet, J.-L.
Muller, W. Leitner, Angew. Chem. 2008, 120, 6776–
6779; Angew. Chem. Int. Ed. 2008, 47, 6674–6677.
[3] In several homogeneously and heterogeneously cata-
lyzed reactions the use of carbon dioxide to pressures
above 10 MPa results in a change in selectivity and ac-
tivity. For example: a) Y. Guo, A. Akgerman, Ind. Eng.
Chem. Res. 1997, 36, 4581–4585; b) C. A. G. Carter,
R. T. Baker, S. P. Nolan, W. Tumas, Chem. Commun.
2000, 347–348; c) K. Wittmann, W. Wisniewski, R.
Mynott, W. Leitner, C. Ludger Kranemann, T. Rische,
P. Eilbracht, S. Kluwer, J. Meine Ernsting, C. J. Elsevi-
er, Chem. Eur. J. 2001, 7, 4584–4589; d) H.-S. Phiong,
C. G. Cooper, A. A. Adesina, F. P. Lucien, J. Supercrit.
Fluids 2008, 46, 40–46.
[11] P. G. Jessop, B. Subramaniam, Chem. Rev. 2007, 107,
2666–2694.
[12] The efficiency of a homogeneous catalyst under reac-
tion conditions can be indicated by the turnover
number (TON) and the turnover frequency (TOF). By
most researchers in the field of homogeneous catalysis
it is common to define the TOF as the rate of product
formation divided by the total amount of metal com-
plex. See: a) D. F. Shriver, P. W. Atkins, C. H. Langford,
Inorganic Chemistry, 2nd edn., Oxford University
Press, Oxford, 1994, pp 709–747; b) R. Shirt, M. Gar-
land, D. W. T. Rippin, Anal. Chim. Acta 1998, 374, 67–
91.
[13] a) A. C. J. Koeken, M. C. A. van Vliet, L. J. P. van den
Broeke, B.-J. Deelman, J. T. F. Keurentjes, Adv. Synth.
Catal. 2006, 348, 1553–1559; b) A. C. J. Koeken,
M. C. A. van Vliet, L. J. P. van den Broeke, B.-J. Deel-
man, J. T. F. Keurentjes, Adv. Synth. Catal. 2008, 350,
179–188.
[14] J. Ke, B. Han, M. W. George, H. Yan, M. Poliakoff, J.
Am. Chem. Soc. 2001, 123, 3661–3670.
[15] T. Jiang, Z. Hou, B. Han, Liang Gao, Z. Liu, J. He, G.
[4] a) P. G. Jessop , T. Ikariya, R. Noyori, Chem. Rev. 1999,
99, 475–493; b) A. Baiker, Chem. Rev. 1999, 99, 453–
473.
Yang, Fluid Phase Equilib. 2004, 215, 85–89.
[5] a) M. Beller, B. Cornils, C. D. Frohning, C. W. Kohl-
[16] R. Fink, E. J. Beckman, in: Chemical Synthesis Using
Supercritical Fluids, (Eds.: P. G. Jessop, W. Leitner),
Wiley-VCH, Weinheim, 1999, pp 67–87.
[17] Example of a recent report on synthesis at a pressure
above 30 MPa: D. C. J. Waalboer, M. C. Schaapman,
F. L. van Delft, F. P. J. T. Rutjes, Angew. Chem. 2008,
120, 6678–6680; Angew. Chem. Int. Ed. 2008, 47, 6576–
6578.
paintner, J. Mol. Catal.
A
1995, 104, 17–85;
b) P. W. N. M. van Leeuwen, in Rhodium Catalyzed
Hydroformylation, (Eds.: P. W. N. M. van Leeuwen, C.
Claver), Kluwer Academic Publishers, Dordrecht, 2000,
pp 1–13.
[6] a) S. Kainz, D. Koch, W. Baumann, W. Leitner, Angew.
Chem. 1997, 109, 1699–1701; Angew. Chem. Int. Engl.
Ed. 1997, 15, 1628–1630; b) D. R. Palo, C. Erkey, Ind.
Eng. Chem. Res. 1998, 37, 4203–4206; c) D. Koch, W.
Leitner, J. Am. Chem. Soc. 1998, 120, 13398–13404;
d) A. M. Banet Osuna, W. Chen, E. G. Hope, R. D. W.
Kemmitt, D. R. Paige, A. M. Stuart, J. Xiao, L. Xu, J.
Chem. Soc. Dalton Trans. 2000, 22, 4052–4055; e) D. R.
Palo, C. Erkey, Organometallics 2000, 19, 81–86; f) K.-
D. Wagner, N. Dahmen, E. Dinjus, J. Chem. Eng. Data
2000, 45, 672–677; g) Y. Shimoyama, M. Sonoda, K.
Miyazaki, H. Higashi, Y. Iwai, Y. Arai, J. Supercrit.
Fluids 2008, 44, 266–272; h) C. T. Estorach, A. Orejꢃn,
A. M. Masdeu-Bultꢃ, Green Chem. 2008, 10, 545–552.
[18] A. C. J. Koeken, S. J. M. de Bakker, H. M. Costerus,
L. J. P. van den Broeke, B.-J. Deelman, J. T. F. Keur-
entjes, J. Supercrit. Fluids 2008, 46, 47–56.
[19] For example: a) free radical polymerization of ethene:
A. Rudin, The Elements of Polymer Science and Engi-
neering, 2nd edn., Academic Press, London, 1999,
pp 349–376; b) ammonia synthesis, see, for example:
R. A. van Santen, J. W. Niemantsverdriet, Chemical
Kinetics and Catalysis, Plenum Press, New York, 1995,
pp 9–10; c) A. Nielsen, H. Bohlbro, J. Am. Chem. Soc.
1952, 74, 963–966; d) cobalt-catalyzed hydroformyla-
Adv. Synth. Catal. 2009, 351, 1442 – 1450
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1449

Ard C. J. Koeken et al.
UPDATES
tion: P. W. N. M. van Leeuwen, Homogeneous Catalysis
Understanding the Art, Kluwer Academic Publishers,
Dordrecht, 2004, pp 125–138.
[26] C. A. Tolman, Chem. Rev. 1977, 77, 313–348.
[27] P. W. N. M. van Leeuwen, C. F. Roobeek, J. Organomet.
Chem. 1983, 258, 343–350.
[28] S.-I. Fujita, S. Fujisawa, B. M. Bhanage, M. Arai, Tetra-
hedron Lett. 2004, 45, 1307–1310.
[20] M. M. H. Lambers-Verstappen, J. G. de Vries, Adv.
Synth. Catal. 2003, 345, 478–482.
[21] H. Jin, B. Subramaniam, A. Ghosh, J. Tunge, AIChE J.
2006, 52, 2575–2581.
[29] W. R. Moser, C. J. Papile, D. A. Brannon, R. A.
Duwell, J. Mol. Catal. 1987, 41, 271–292.
[22] P. W. N. M. van Leeuwen, C. P. Casey, G. T. Whiteker,
in: Rhodium Catalyzed Hydroformylation, (Eds.:
P. W. N. M. van Leeuwen. C. Claver), Kluwer Academic
Publishers, Dordrecht, 2000, pp 63–105.
[30] B. Breit, W. Seiche, J. Am. Chem. Soc. 2003, 125, 6608–
6609.
[31] B. Cornils, W. A. Herrmann, I. T. Horvꢅth, W. Leitner,
S. Mecking, H. Olivier-Bourbigou, D. Vogt, in: Multi-
phase Homogeneous Catalysis, Vol. 1, (Eds.: B. Cornils,
W. A. Herrmann, I. T. Horvꢅth, W. Leitner, S. Mecking,
H. Olivier-Bourbigou, D. Vogt), Wiley-VCH, Wein-
heim, 2005, pp 3–23.
[32] P. Cavalieri d’Oro, L. Raimondi, G. Pagani, G. Montra-
si, G. Gregorio, A. Andreetta, Chim. Ind. 1980, 62,
572–579.
[23] A. van Rooy, E. N. Orij, P. C. J. Kamer, P. W. N. M. van
Leeuwen, Organometallics 1995, 14, 34–43.
[24] S. N. Poelsma, P. M. Maitlis, J. Organomet. Chem. 1993,
451, C15-Cl7.
[25] P. C. J. Kamer, J. N. H. Reek, P. W. N. M. van Leeuwen,
in: Rhodium Catalyzed Hydroformylation, (Eds.:
P. W. N. M. van Leeuwen. C. Claver), Kluwer Academic
Publishers, Dordrecht, 2000, pp 35–62.
1450
asc.wiley-vch.de
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 1442 – 1450