Selectivity of rhodium-catalyzed hydroformylation of 1-octene during batch and semi-batch reaction using trifluoromethyl-substituted ligands

Article abstract of DOI:10.1002/adsc.200700423

The regioselectivity of catalysts generated in situ from dicarbonyl rhodium(I)(2,4-pentanedione) and trifluoromethyl-substituted triphenylphosphine ligands has been evaluated during the hydroformylation of 1-octene. The influence of batch or semi-batch operation, the solvent, and the number of trifluoromethyl substituents has been investigated. During batch operation in a supercritical carbon dioxide (CO₂)-rich system the differential n:iso ratio increases from approximately 4 to a value of 12-16 at about 90-95% conversion for the catalyst based on bis[3,5-bis(trifluoromethyl)phenyl] phenylphosphine. For semi-batch conditions using hexane a constant n:iso ratio is obtained over a broad conversion range. Batch hydroformylation in neat 1-octene is faster than in a supercritical CO₂-rich, one-phase system, with a similar overall selectivity as observed in the supercritical case. The results provide further directions for the development of ligands that are especially designed for the separation of homogeneous catalysts in continuously operated hydroformylation in ScCO₂.

Full text of DOI:10.1002/adsc.200700423

FULL PAPERS
DOI: 10.1002/adsc.200700423
Selectivity of Rhodium-Catalyzed Hydroformylation of 1-Octene
during Batch and Semi-Batch Reaction using Trifluoromethyl-
Substituted Ligands
a
b
a,
Ard C. J. Koeken, Michiel C. A. van Vliet, Leo J. P. van den Broeke, *
b
a
Berth-Jan Deelman, and Jos T. F. Keurentjes
Process Development Group, Eindhoven University of Technology, P. O. Box 513, 5600 MB Eindhoven, The Netherlands
a
Fax: (+31)-40-244-6104; e-mail: l.j.p.van.den.broeke@tue.nl
Arkema Vlissingen B. V., P. O. Box 70, 4380 AB Vlissingen, The Netherlands
b
Fax: (+31)-113-612-984
Received: August 28, 2007; Published online: December 20, 2007
Abstract: The regioselectivity of catalysts generated stant n:iso ratio is obtained over a broad conversion
in situ from dicarbonyl rhodium(I)(2,4-pentane- range. Batch hydroformylation in neat 1-octene is
dione) and trifluoromethyl-substituted triphenyl- faster than in a supercritical CO -rich, one-phase
2
phosphine ligands has been evaluated during the hy- system, with a similar overall selectivity as observed
droformylation of 1-octene. The influence of batch in the supercritical case. The results provide further
or semi-batch operation, the solvent, and the number directions for the development of ligands that are es-
of trifluoromethyl substituents has been investigated. pecially designed for the separation of homogeneous
During batch operation in a supercritical carbon di- catalysts in continuously operated hydroformylation
oxide (CO )-rich system the differential n:iso ratio in scCO2.
2
increases from approximately 4 to a value of 12–16
at about 90–95% conversion for the catalyst based Keywords: fluorinated ligands; homogeneous cataly-
on
bis[3,5-bis(trifluoromethyl)phenyl]phenylphos- sis; hydroformylation; organic solvents; supercritical
A
C
H
T
R
E
U
N
G
phine. For semi-batch conditions using hexane a con- carbon dioxide
Introduction
is the attachment of ligands to soluble or insoluble
macromolecular supports, which can be separated by
[
13,14]
The hydroformylation of alkenes is an important ex- filtration or decantation.
The development of, for
[
15]
[16]
ample of homogeneous catalysis on an industrial example, diphosphines, diphospites, diphosphines
[1]
[17]
[18]
scale. Organometallic complexes of rhodium have by self-assembly, and tetraphosphines has led to
proven to be the most active and selective catalysts in rhodium-catalyzed hydroformylation of long-chain al-
hydroformylation. In the hydroformylation of pro- kenes with a very good regioselectivity for the linear
pene a water-soluble rhodium catalyst is used, facili- aldehyde product.
tating an easy separation from the organic product
The advantages of the application of carbon dioxide
[1,2]
phase.
However, water-soluble catalysts have limit- as a solvent for homogeneously catalyzed reactions
ed application in the hydroformylation of long-chain include the possibility to create a one-phase supercrit-
alkenes as the low solubility of long-chain alkenes in ical reaction mixture in which phase boundaries are
[
2]
water will result in relatively low reaction rates. Re- absent, high diffusivity of the different species, and
[8]
search dealing with hydroformylation of long-chain high solubility of carbon monoxide and hydrogen.
alkenes, therefore, focuses on alternative solvents in The solubility of common homogeneous catalysts in
[19]
conjunction with the development of ligands and ad- supercritical carbon dioxide, scCO , is limited. This
2
ditives to make rhodium catalysts more active and se- can be seen as an advantage in of view of the possibil-
[
3]
lective and easier to separate from the product. Per- ities to separate the catalyst, because CO has been
2
[
4]
[5]
fluorinated solvents, ionic liquids, water in combi- applied as an anti-solvent in order to precipitate and
nation with a phase-transfer agent, PEGs, and su- recycle the catalyst. However, the changes in pres-
percritical fluids are regarded as promising alterna- sure required to apply CO in this manner are possi-
[
6]
[7]
[20]
[8]
2
tive solvents, which can facilitate catalyst bly not very energy efficient. Therefore, to employ
[9–12]
separation.
Another approach to catalyst recycling the promising solvent characteristics of CO , an alter-
2
Adv. Synth. Catal. 2008, 350, 179 – 188
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
179

Ard C. J. Koeken et al.
FULL PAPERS
Scheme 1. Ligands used in the rhodium-catalyzed hydroformylation: tris[3,5-bis(trifluoromethyl)phenyl]phosphine (I), bis-
A
H
R
U
G
A
C
H
T
R
E
U
N
G
[3,5-bis(trifluoromethyl)phenyl]phenylphosphine (II), tris(3-trifluoromethylphenyl)phosphine (III), triphenylphosphine (IV).
In the text, the catalytic complexes L/[Rh(CO) acac] (acac: 2,4-pentanedione) are indicated by the ligand only.
2
native separation method and a catalyst, which is
soluble in a CO -enriched supercritical mixture, have
2
to be developed. Membrane separation is regarded as
a promising method to separate catalysts from a su-
[
21]
percritical reaction mixture.
In order to influence
and improve the solubility of Wilkinson-type hydro-
formylation catalysts in CO -rich reaction environ-
2
ments, perfluorinated groups or “tails” can be at-
[
21–23]
tached to the triphenylphosphine ligands.
Besides
the improvement of catalyst solubility in scCO or flu-
2
orous systems, phosphine ligands with perfluoroalkyl
groups also significantly influence the hydroformyla-
[24–28]
tion catalysis.
Scheme 2. Reaction scheme for the hydroformylation of 1-
octene (1a), with the two main products nonanal (2a) and 2-
methyloctanal (2b). The side products are (E,Z)-2-octene
Here we present a detailed study about the applica-
tion of trifluoromethyl-substituted triphenylphos-
phines in the rhodium-catalyzed hydroformylation of
(
1b, 1c), (E,Z)-3-octene (1d, 1e), (E,Z)-4-octene (1f, 1g), 2-
1-octene. Results obtained under batch conditions in
ethylheptanal (2c), 2-propylhexanal (2d), and n-octane (3).
scCO2 are compared with results obtained under L is the phosphine ligand.
batch and semi-batch conditions in organic solvents,
which is relevant for further development of continu-
ous hydroformylation in scCO2.
ular, the use of II and III has been evaluated in four
situations: batch and semi-batch with hexane as sol-
Results and Discussion
vent, batch with scCO as solvent, and solventless
2
(
neat). The experimental conditions and main results
Previously it was shown that, for the rhodium-cata- for the 12 cases studied are summarized in Table 1. A
lyzed hydroformylation of 1-octene, the influence of high ligand to rhodium ratio, L:Rh=50:1, is applied,
the trifluoromethyl substituents on the phosphine on which approaches industrial conditions. An enhanced
the activity and overall selectivity is about the same regioselectivity towards the linear product can be ex-
[29]
[1,30]
when scCO , hexane or toluene are applied as sol- pected at higher ligand concentrations.
vent.
(
The differ-
In this comparison it was observed that the ential selectivity indicates to what extent 1-octene is
final) ratio of linear over branched aldehydes (n:iso converted into the desired products at some moment
2
[26]
ratio) was almost always comparable although the ex- in the batch reactor. For commercial hydroformyla-
perimental conditions, like pressure and concentration tion processes predominantly CSTRs or loop reactors
[
31]
of the syngas, were considerably different. Therefore, are being used, and for these types of reactor con-
a closer look is taken at the cumulative (or overall or figurations the overall selectivity is equal to the differ-
integral) selectivity and the differential (or intrinsic) ential selectivity.
selectivity in the application of in situ prepared cata-
lysts generated from [Rh(CO) acac] and four differ-
2
ent ligands. Three trifluoromethyl-substituted triphe- Concentration Profile
nylphosphine ligands and the triphenylphosphine
ligand, shown in Scheme 1, have been used in the hy- For III (Scheme 1) it has been observed that during
droformylation of 1-octene (1a) (Scheme 2). In partic- the reaction the aldehyde selectivity decreased be-
180
asc.wiley-vch.de
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 179 – 188

Selectivity of Rhodium-Catalyzed Hydroformylation of 1-Octene
FULL PAPERS
Table 1. Summary of conditions and results.
[a]
[b]
[c,d]
[c,e]
[f]
Entry Solvent
L
n1a
mmol]
nCO:nH2
A
H
R
U
G
TOF1a
TOFald
X
Sald
[%]
S2a
[%]
S
[%]
n:iso
[-]
max
1bꢀ1g
[
[MPa]
[%]
[
g]
1
2
3
4
5
6
7
8
9
1
1
1
hexane[ II 49.8
49.9:49.2
51.6:50.8
[h]
4.5
4.6
5.0
4.6
50.1
49.5
49.6
51.0
6.5
3.04
1.79
3.10
2.07
9.82
3.29
1.33
0.463
15.4
7.38
3.50
1.03
2.89
1.69
3.01
2.02
8.87
3.13
1.28
0.455
97.7
97.0
99.0
99.4
99.2
97.2
91.7
93.0
98.7
99.9
99.8
95.9
95.4
96.8
98.4
98.8
90.5
94.0
96.6
99.0
94.3
94.7
96.7
98.3
78.4
78.7
76.6
76.8
69.6
78.4
79.3
77.1
77.8
79.8
79.1
75.8
4.1
2.7
1.0
1.0
6.1
5.1
2.9
0.1
4.5
4.6
2.7
1.1
4.60
4.34
3.52
3.50
3.33
5.02
4.59
3.53
4.71
5.37
4.48
3.38
g]
hexane[ III 49.9
g]
hexane[ II 49.7
g]
hexane III 49.9
[h]
CO2
CO2
CO2
CO2
Neat
Neat
Neat
Neat
I
105
108:108
109:109
109:109
108:109
104:103
104:105
104:105
104:105
II 105
III 105
IV 107
[
k]
[l]
[l]
I
104
14.3
0
1
2
II 100
III 100
IV 100
6.5
6.5
6.5
6.79
3.16
1.00
[
a]
3
ꢀ1
General applied conditions: T=708C, S/C=2.0”10 mol molRh . For entries 1–4 and 9–12: Vreactor =0.103 L; for en-
1
a
[26]
tries 5–8: Vreactor =0.108 L. Results in entries 5 and 7 have also been reported in a previous communication.
Maximum pressure reached upon injection of 1a.
Obtained from multiplying S/C with the slope of a linear fit through conversion (TOF ) or yield (TOF ) data up to a
[
[
b]
c]
1
a
ald
conversion of 60% (40% for IV).
[
[
[
d]
e]
f]
3
ꢀ1 ꢀ1
[
[
10 mol molRh h ].
1a
3
ꢀ1 ꢀ1
10 mol
molRh h ].
aldehydes
Conversion (X), selectivity for aldehydes (S ), selectivity for nonanal (S ), selectivity for isomers of 1a (S1b–ä1g) and the
ald
2a
n:iso after 3 h of reaction.
Amount of hexane used for entry 1: 44.7 mL; entry 2: 42.8 mL; entry 3: 47.2 mL; entry 4: 43.9 mL.
[
[
[
[
g]
h]
k]
l]
Semi-batch, CO:H =1.0:1.0.
2
Results after 9 h reaction.
The TOFald was determined from the pressure change in the first 5 min of reaction. TOF1a and TOF2a are estimated based
on TOFald and the composition of the first sample taken after 12 min.
Cumulative n:iso Ratio
To investigate whether there is a change in regioselec-
tivity, in particular a change in n:iso ratio as a result
of the mode of operation, II and III have been ap-
plied in batch and semi-batch modes with respect to
syngas and using hexane as the solvent. In Figure 2 it
can be seen that during semi-batch operation a con-
stant n:iso ratio over a broad conversion range is ob-
served, while during batch operation the n:iso ratio
increases clearly with an increase in conversion. The
initial activity of III in hexane is similar for the batch
case (entry 2), the semi-batch case (entry 4) and for
the case where a lower syngas pressure has been ap-
[
26]
plied (as reported in ref. ). The same holds for II
even when taking into account the reaction in scCO2
Figure 1. Normalized concentration profile of the hydrofor-
mylation of 1a in CO using II. The reaction conditions are (entry 6). It appears that the finding that the reaction
2
given in Table 1, entry 6.
rate is only slightly dependent on the concentration of
CO/H (1:1) as reported by Cavalieri dꢁOro et al. for
2
the hydroformylation of propene with [HRhCO(IV) ]
cause of a slow buildup of 1a isomers. For ligand II as the pre-catalyst also holds for II and III.
3
[
26]
[32]
a similar behavior is observed. In Figure 1 it is seen
To study the influence of solvent on the selectivity
that after approximately 1 hour of reaction the con- in more detail, the n:iso ratio as a function of conver-
centration of 2a still increases slowly, while the con- sion has been obtained using I to IV in a one-phase
centrations of 2b to 2d remain constant, which implies supercritical system, shown in Figure 3a, or in neat
that the selectivity of the catalyst changes during the 1a, given in Figure 3b, both in batch. The results for
reaction.
the n:iso ratio obtained for the neat experiments
Adv. Synth. Catal. 2008, 350, 179 – 188
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
181

Ard C. J. Koeken et al.
FULL PAPERS
tion of 2b to 2d through hydroformylation of 1a iso-
mers. It is noted that for neat conditions the highest
values for the n:iso after three hours of reaction ratio
have been obtained. Additionally, the rhodium cata-
lysts modified with I, II, or III are very active in the
solventless hydroformylation of 1a.
The final outcome of the reaction in terms of n:iso
ratio observed for neat 1a is roughly the same when
compared to the reaction in scCO except for I (en-
2
tries 5 and 9). However, the n:iso as a function of
conversion for the neat experiments appears to in-
crease linearly, while in CO the n:iso ratio increases
2
in a non-linear manner. The large difference between
the observed overall n:iso ratio for I can possibly be
explained by the fact that the reaction is very fast and
Figure 2. The n:iso ratio as a function of conversion (X) for
the experiments with ligand II and III in hexane for the
batch and semi-batch case (entries 1 to 4 in Table 1).
solubility of H and CO is low in neat 1a and alde-
2
hydes. Possibly, as a result of mass transfer limitations
the concentration of CO in the liquid phase is at a
minimum and this favors the formation of linear alde-
[
33]
show more or less the same dependence on the con- hydes.
version as the results obtained in scCO . F orII and
2
III a clear increase is observed in the n:iso ratio as
the reaction progresses. The increase in n:iso ratio is Catalytic Activity versus Selectivity
more pronounced for the supercritical batch system
than for the two-phase hexane batch system. For IV Based on the TOFvalues (Table 1) the hydroformyla-
in CO , the n:iso ratio is more or less constant as a tion in neat 1a is about 1.5 to 2 times faster, as com-
2
function of the conversion up to a conversion of 80% pared to reaction rates obtained in the other solvents.
but then also starts to increase at higher conversion.
This can probably be attributed to a higher local con-
For I, which is the most active catalyst in this com- centration of catalyst in the liquid 1a phase in combi-
parison, a small increase in the n:iso ratio is observed nation with the highest concentration of 1a possible.
up to a conversion of approximately 90% when The activity of the rhodium complex clearly increases
scCO is used. At high conversion, a decrease in n:iso when a triarylphosphine is applied with a greater
2
ratio is observed. Initially, I gives rise to a substantial number of trifluoromethyl substituents. Both in CO2
amount of 1a isomerization. Moreover, I has suffi- and in neat 1a the initial TOFvalues observed for
cient activity towards the hydroformylation of inter- ligand increase in the order: I>II>III>IV. As can
nal alkenes (1b to 1g), and as a result the n:iso ratio be deduced from Table 1 by comparing TOF1a and
drops at a high 1a conversion because of the produc- TOF , the initial rate of isomerization and hydroge-
ald
Figure 3. a) Ratio of linear to branched product, n:iso, ratio as a function of conversion for the experiments with ligand I to
IV in the one-phase supercritical CO . b) n:iso ratio as a function of conversion for the experiments with ligand I to IV in
2
neat 1a. It should be noted that in particular for I to III in neat 1a a minimum amount of samples were taken to minimize
the effect of sampling on the reaction.
182
asc.wiley-vch.de
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 179 – 188

Selectivity of Rhodium-Catalyzed Hydroformylation of 1-Octene
FULL PAPERS
nation of 1a towards the isomers 1b to 1g and 3, re- order I<IV<II<III in scCO . Palo and Erkey also
2
spectively, also increases for the ligand in the order: tested I and III in the rhodium-catalyzed hydroformy-
I>II>III>IV. The regioselectivity in terms of n:iso lation of 1a in scCO , but they did not observe a sig-
2
ratio, however, increases in the order II>III>IꢁIV. nificant change in n:iso ratio with an increase in ba-
[24]
The difference in regioselectivity of I, II, and III ob- sicity of ligands (I: n:iso=3.02, III: n:iso=3.13).
served in CO becomes most pronounced at a conver- Probably the reaction conditions they applied (T=
2
sion above 60% (Figure 3a).
508C and L:Rh=3:1) can account for this. This clear
[
34]
It has been demonstrated by Moser et al. that the deviation of I from the expected dependence of the
(
hydroformylation and isomerization) activity and the catalytic behaviour on the number of electron-with-
regioselectivity in terms of n:iso ratio of the catalyst drawing substituents in the series I to IV, which ap-
generated from rhodium and a para-substituted triar- pears to be independent of the solvent has not been
ylphosphine increases with decreasing basicity of the established before.
phosphine ligand. The basicity of the phosphine
ligand decreases upon attaching trifuoromethyl sub-
[
24]
stituents on the aryl ring of a triarylphosphine.
Moreover, for monodentate triarylphosphines it is ex-
Differential n:iso Ratio
pected that electron-withdrawing groups induce a An increase in (cumulative or overall) n:iso ratio with
higher fraction of the diequatorial coordinated [HCO- an increase in conversion implies that as the reaction
A
C
H
T
R
E
U
N
G
(alkene)Rh
A
C
H
T
R
E
U
N
G
(PAr ) ] intermediate (with regard to the proceeds the catalyst converts 1a with an increasing
3 2
phosphines) responsible for a higher selectivity to- differential n:iso ratio. In Figure 4a and Figure 4b a
[
35,36]
wards the linear aldehyde.
Additionally, the elec- comparison is made between the differential n:iso
tron-withdrawing groups present in this diequatorial ratio, calculated from the cumulative n:iso ratio, ob-
intermediate should lead to an even better selectivity tained with II and III, respectively. Additionally, the
towards the linear aldehyde, which was demonstrated differential n:iso obtained with IV in CO is shown in
2
by Casey et al. for the diphosphines BISBI and T- Figure 4b. From this comparison the influence of the
[
35]
BDCP.
Indeed, a similar dependence as observed operating conditions, a one-phase supercritical batch
[34]
by Moser et al.
of the regioselectivity (n:iso) and system, a two-phase batch system, and a two-phase
activity on the basicity of the ligand is observed here semi-batch system, becomes clear. For all cases where
when the ligands II to IV are considered, see Table 1. an additional solvent is used, the differential n:iso
The results obtained with I, however, deviate from ratio is initially about 3.7 and increases continuously
[
34]
the behavior described by Moser et al., because for with increasing conversion.
I a lower regioselectivity is observed than for II (in For the batch case with II and III in CO and
2
neat 1a) and III (in CO and neat 1a). Furthermore, hexane the increase in differential n:iso becomes
2
in the application of II to IV in scCO the regioselec- more pronounced at a conversion of above 50%. For
2
tivity (n:iso) is not clearly coupled with the overall se- IV in CO the increase in differential ratio becomes
2
lectivity for 2a (n-aldehyde selectivity, S , in Table 1) more apparent above a conversion of 70%. The high-
2a
as observed by Casey et al. for the bidentate phos- est value for the differential n:iso ratio is obtained for
[35]
phines. The n-aldehyde selectivity increases in the II in scCO ; the differential n:iso ratio increases to a
2
Figure 4. a) Differential n:iso ratio as a function of conversion for II (entries 1, 3, 6, and 10). b) Differential n:iso ratio as a
function of conversion for III (entries 2, 4, 7, and 11). The dashed lines indicate the trend.
Adv. Synth. Catal. 2008, 350, 179 – 188
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
183

Ard C. J. Koeken et al.
FULL PAPERS
value of approximately 14–16. In hexane, the differen- crease in (differential) n:iso ratio (Figure 3a) is in
tial n:iso increases less sharply, the highest value ob- good agreement with the observations by Koch
[
39]
tained is 6–6.5. For the semi-batch experiment per- et al. The difference in regioselectivity can probably
formed in hexane the differential n:iso ratio is con- be attributed to the steric effects caused by the per-
stant up to a conversion of about 90% and then fluoroalkyl substituents. Finally, the catalytic activity
drops to a value of about 2.5, as a result of the hydro- we observed (entry 8 in Table 1) is close to the activi-
ꢀ
1
ꢀ1
formylation of internal octenes. For neat 1a and using ty of 430 molald molRh
h
reported by Koch et al. for
II the maximum differential n:iso ratio observed is tris[4-(1H,1H,2H,2H-perfluorooctyl)phenyl]phosphine
[
39]
about 6.0 at a conversion around 82%. For III a simi- at 658C.
lar behaviour is observed; the highest differential Based on the differential n:iso of entries 2 and 4 as
n:iso ratio obtained is about 8–9 and 5–6 for scCO2 a function of conversion the similar overall n:iso
[
26]
and hexane, respectively.
found previously
when applying III during batch
Screening of new hydroformylation catalysts is pre- operation in scCO [n:iso (3 h)=4.6] or during semi-
2
dominantly done at “low” pressures and semi-batch batch operation in hexane [n:iso (3 h)=4.8] can be
with respect to the gaseous reactants. Usually, during explained. The syngas pressure applied in the semi-
semi-batch hydroformylation, using rhodium catalysts batch experiment using hexane as the solvent was
in the presence of excess phosphine ligand, a constant 1MPa. This syngas pressure corresponds to the pres-
or a decrease in n:iso ratio has been observed for a sure at about 80% conversion in the case of entry 10.
[
32,37,38]
variety of solvents.
For hydroformylation with The value of 5 for the differential n:iso ratio observed
IV applying a lower CO partial pressure results in a at 80% conversion is close to the value of 4.8 for the
higher regioselectivity, that is, a higher n:iso ratio, at overall n:iso observed previously.
the cost of a lower chemoselectivity, that is, a higher
[
30,32,33]
degree of hydrogenation and/or isomerization.
For supercritical conditions it is more convenient to Differential Selectivity
work batch-wise. To keep the concentration of syngas
constant in a supercritical reaction mixture, during By plotting the differential selectivity towards nonan-
semi-batch operation, requires a dosing mechanism al, S , and towards the nonanal isomers, S2b–2d, as a
2
a
that delivers syngas at high pressures. Moreover, the function of the conversion, the origin for the increase
pressure of the supercritical mixture not only changes in the differential n:iso ratio is further clarified. In
as a result of the conversion of syngas but also as a Figure 5a the differential selectivity obtained with II
result of a different interaction between 1a and CO2 in CO and hexane, batch and semi-batch, are com-
2
as compared to the interaction between the aldehydes pared. For the supercritical system (entry 6), the se-
and CO . Remarkably, for the batch hydroformylation lectivity towards the linear aldehyde product is more
2
of 1a no change in selectivity during the reaction in or less constant up to a conversion of 90%, while
scCO2 has been reported before in the litera- S2b–2d decreases from 18.5% at 9% conversion to 4%
[24,39–42]
ture.
This is most likely a result of either the at approximately 95% conversion. For the batch hy-
application of a more than two-fold excess of syngas droformylation in hexane (entry 1) a similar but more
with respect to the alkene, or the use of a “low” moderate behavior is observed. For the semi-batch
ligand to rhodium ratio, which in most cases is not case (entry 3), S decreases and S increases slight-
2
a
2b–2d
higher than 10:1. Only in the case when tris[3- ly. Above 90% conversion S2a decreases by about
1H,1H,2H,2H-perfluorooctyl)phenyl] phosphite was 10%, and S2b–2d increases by about 20%, which is re-
used a clear increase in n:iso ratio as function of the flected in the lower differential n:iso ratio. For IV in
(
[
39]
conversion was observed.
CO , both S and S2b–2d remain constant, in the con-
2 2a
In the application of the triphenylphosphine ana- version range of 5% to about 70%, above 70% S2a
logues tris[4-(1H,1H,2H,2H-perfluorooctyl)phenyl]- increase to a value of about 80% and S2b–2d decreases
phosphine or tris[3-(1H,1H,2H,2H-perfluorooctyl)- to a value of about 15%.
phenyl]phosphine in the rhodium-catalyzed hydrofor-
For the catalyst III a similar but more moderate be-
[
39]
mylation of 1a in scCO Koch et al. observed a con- havior of S2a and S2b–2d as a function of conversion is
2
stant n:iso ratio of about 5.5. In this case, the elec- observed than for II as can be seen in Figure 5b. In
tron-withdrawing effect of the perfluoroalkyl group the batch experiments in CO and hexane S only in-
2
2a
on the phosphorus is counteracted by the ethyl creases up to values of 85% at about 90% conver-
spacers” between the perfluoroalkyl group and the sion. However, in the conversion range of 80 to 95%
“
aryl ring. The electron density on the phosphorus of the differential n:iso ratio observed for III is some-
these phosphines with perfluoroalkyl chains is similar what lower than for II, because S2b–2d for III does not
to that of triphenylphosphine. So, the constant (differ- decrease as much as observed for II. It seems that III
ential) n:iso ratio obtained with IV in scCO up to maintains a higher chemoselectivity towards alde-
2
70% conversion and the subsequent moderate in- hydes than II at low CO partial pressure. For the
184
asc.wiley-vch.de
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 179 – 188

Selectivity of Rhodium-Catalyzed Hydroformylation of 1-Octene
FULL PAPERS
Figure 5. a) Differential selectivity for 2a (nonanal), S , and the differential selectivity towards the branched aldehydes,
2
a
S2b–2d, 2-methyloctanal (2b), 2-ethylheptanal (2c) and 2-propylhexanal (2d) using II as a function of conversion for entries 1,
and 6. b) Differential selectivity using III as a function of conversion for entries 2, 4 and 7.
3
semi-batch case (entry 4), S2a and S2b–2d remain more Experimental Section
or less constant up to a conversion of about 90%.
Above 90% conversion a similar behaviour as for II Materials
is observed, S decreases to 70% while S increas-
es to about 30%.
2a
2b–2d
Ligands I, II and III were supplied by Arkema (Vlissingen).
Ligand IV was purchased from Aldrich. Dicarbonyl rhodi-
um(I) 2,4-pentanedione ([Rh(CO) acac]) was obtained from
2
Fluka. Ligands I, II, III and [Rh(CO) acac] were stored
under argon. 1a was purchased from Aldrich, passed over a
column of activated alumina (Brockmann I, Sigma-Aldrich),
2
Conclusions
By following the reaction in time it was observed that dried over molecular sieves and stored under argon. Hydro-
gen grade 5.0, carbon monoxide 4.7 and carbon dioxide
a low CO partial pressure induces a higher differen-
grade 5.0 were purchased from Hoekloos. Prior to use
carbon dioxide was passed over a Messer Oxisorb filter to
remove oxygen and moisture. Analytical grade hexane was
obtained from Merck and dried over molecular sieves and
stored under argon. Toluene (analytical grade, Merck) and
decane (99% purity, Aldrich) used to prepare the samples
for GC analysis were used as received.
tial n:iso ratio for II and III than expected based on
the overall n:iso ratio, especially in scCO . This ap-
2
pears to be caused by a slower hydroformylation of 1-
octene to the branched aldehyde. For IV in scCO2
this phenomenon also takes place but in a more mod-
erate manner at a lower CO concentration.
The hydroformylation and isomerization activity in-
creases with the number of trifluoromethyl substitu-
ents and decreasing basicity of the ligand in the
order: I<II<III<IV. When the catalysts II to IV are
considered, also the regioselectivity selectivity (n:iso)
increases with the number of electron-withdrawing
trifluoromethyl groups. From the various results it can
be derived that it could be attractive if the hydrofor-
mylation of 1a could be operated continuously using
II with a n:iso ratio of about 12–16. The preferred
Hydroformylation in ScCO2
The procedure for the hydroformylation in CO was de-
scribed in a previous communication.
2
[26]
Hydroformylation in Neat 1a
The conditions during catalyst pre-formation for the experi-
ments without solvent were the same as in the experiments
conditions are a one-phase CO enriched mixture at a with CO as the solvent. After the pre-formation time of ap-
2
2
low CO partial pressure (for entry 6 approximately proximately one hour the reactor was cooled rapidly to
0
.3–0.6MPa). As can be deduced from Figure 5a, this room temperature and the gases were carefully vented. The
catalyst and excess ligand remained in the reactor and were
will be at the expense of a lower overall aldehyde se-
lectivity (n+iso) as the selectivity towards isomeriza-
tion products 1b and 1c will be higher.
stored overnight under an argon atmosphere. After evacuat-
ing the reactor, carbon monoxide and hydrogen were fed to
the reactor up to 4.9MPa at room temperature after the stir-
rer speed was set to 1400 rpm. Then the reactor was heated
up to 708C. When the reactor temperature was stabilized,
The results obtained in scCO and the effect of dif-
2
ferent substitution patterns on the ligands provide di-
rections for further development of ligands that are
especially designed for the separation of homogene-
1
a was pressurized to a pressure just above reactor pressure
and subsequently the desired volume was injected into the
ous catalysts in continuously operated hydroformyla- reactor within 30 s. This marked the start of the reaction.
tion in scCO2.
The samples taken from the liquid phase were immediately
Adv. Synth. Catal. 2008, 350, 179 – 188
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
185

Ard C. J. Koeken et al.
FULL PAPERS
ꢀ1
diluted with a solution of 0.01 molL of decane in toluene
and cooled to room temperature.
½jŠ
n
Sj ¼
ð4Þ
f½1aŠ ꢀ½1aŠ g
n:0
n
Hydroformylation in Hexane
where j=1b–1g, 2a–2d and 3.
The overall yield for a product p was [Eq. (5)]:
The experimental procedure was started by conveying the
desired amounts of catalyst (on average 26.0 mmol
[
Rh(CO) acac] and 1.31 mmol phosphine ligand) into the re-
2
½
pŠ
n
actor. The reactor was closed and alternately evacuated and
rinsed with argon three times. The reactor was evacuated
and the desired amount of hexane was injected at room
temperature. Then carbon monoxide and hydrogen were fed
to the reactor up to 4.3MPa. The stirrer was turned at a rev-
olution rate of 1400 rpm and the pressure typically dropped
to a pressure of 3.6MPa. After stirring at 508C for half an
hour the reactor was heated and kept for another half hour
at 708C before injecting 1a in a similar manner described
Y ¼
ð5Þ
p
½1aŠn:0
where p=1b–1g, 2a–2d and 3.
The differential selectivity Sq,k for product q was deter-
mined by [Eq. (6)]:
Yq,samplekþ1 ꢀY
q,samplek
Sq,k
¼
ð6Þ
for the scCO and the neat experiments. In the semi-batch
2
Xsamplekþ1 ꢀX
samplek
experiments the pressure was kept constant at 4.7ꢂ0.2MPa.
Samples were taken from the liquid phase and treated in a
similar way as in the neat experiments.
with q=1b–1g, 2a–2d and 3. k runs from 1 to the total
number of samples taken during an experiment minus 1. Sq,k
was only evaluated for 2a and the sum of the branched alde-
hydes 2b to 2d.
Analysis
The samples were analyzed offline on the same day as the
reaction was conducted using a Fisons Instruments GC-FID
equipped with a Restek Rtx-5 column (fused silica, length
The differential n:iso ratio was calculated as follows [Eq.
(
7)]:
3
0 m, internal diameter 0.53 mm) with helium as the carrier
S2a,k
gas. Calibration was done for 1a, 1b, 1c, 3 and 2a, the sensi-
tivity coefficients for the other octene and aldehyde isomers
were taken to be equal to those of 1a and 2a, respectively.
differential n:iso ¼
ð7Þ
k
S
2b,k þ S2c,k þ S2d,k
The differential parameters were evaluated as a function of
conversion. The conversion values corresponding to differ-
ential parameter values were calculated as follows [Eq. (8)]:
Reaction Parameters
To obtain consistent concentration profiles for 1a and its re-
action products, each concentration obtained by GC analysis
was normalized to the sum of all obtained concentrations
ðXsamplekþ1 þ Xsamplek
Þ
[Eq. (1)]:
X ¼
k
ð8Þ
2
½
iŠ
½
iŠ ¼ P
ð1Þ The substrate to catalyst ratio S/C was calculated as follows
Eq. (9)]:
n
½
iŠ
[
where i=1a–1g, 2a–2d and 3, and the subscript n refers to
the normalized values. The activity and selectivity of the dif-
ferent catalytic complexes was expressed in one of the fol-
lowing parameters. The conversion, X, was given by [Eq.
S
C
m ꢃ MW
1
a
Rh
¼
ð9Þ
m
ꢃ MW
Rh
1a
(
2)]:
with m1a the mass of 1a injected, MW1a the molar mass of
a, and mRh and MWRh the mass conveyed to the reactor
1
and the molar mass of the rhodium precursor, respectively.
The turnover number based on the conversion of 1a,
TON , was calculated as follows [Eq. (10)]:
½
1aŠ ꢀ½1aŠ
n:0
n
X ¼
ð2Þ
½1aŠ
n:0
1
a
with [1a]n.0 =1.
The (overall or cumulative) n:iso ratio was calculated as
follows:
S
C
TON1a
¼
ꢃ X
ð10Þ
The “initial” turnover frequency based on either 1a, 2a, or
½
2aŠ
n : iso ¼
ð3Þ aldehydes (2a–2d also abbreviated as “ald”) was calculated
½
2bŠ þ ½2cŠ þ ½2dŠ
by multiplying n1a,0 (initial amount of 1a in mol) with the
slope of a line fitted through the conversion, Y2a or Yald data
The overall selectivity, S , towards a product i was defined as
points up to a conversion where there was a linear trend
(typically up to a conversion of 0.6).
j
[Eq. (4)]:
186
asc.wiley-vch.de
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 179 – 188

Selectivity of Rhodium-Catalyzed Hydroformylation of 1-Octene
FULL PAPERS
Acknowledgements
van Leeuwen, K. Goubitz, J. Fraanje, Organometallics
1
995, 14, 3081; c) H. Klein, R. Jackstell, K.-D. Wiese,
This work was supported by an EET grant from SenterNo-
vem (agency of the Dutch Ministry of Economic Affairs).
The contribution of Paul van Bavel to the experimental work
is gratefully acknowledged.
C. Borgmann, M. Beller, Angew. Chem. Int. Ed. 2001,
40, 3408.
[16] For example: E. Billig, A. G. Abatjoglou. D. R. Bryant,
(to Union Carbide Corp.), European Patent Appl. EP
0
.213639, 1987.
17] For example: a) B. Breit, W. Seiche, J. Am. Chem. Soc.
003, 125, 6608; b) B. Breit, W. Seiche, Angew. Chem.
[
2
References
Int. Ed. 2005, 44, 1640.
[
18] Y. Yan, X. Zhang, X. Zhanga, Adv. Synth. Catal. 2007,
349, 1582.
19] P. G. Jessop, T. Ikariya, R. Noyori, Chem. Rev. 1999, 99,
[
1] C. D. Frohning, C. W. Kohlpaintner, H.-W. Bohnen, in:
nd
Applied Homogeneous Catalysis, 2 edn., Vol. 1, (Eds.:
B. Cornils, W. A. Herrmann), Wiley-VCH, Weinheim,
[
[
[
4
75.
2
002, p 31.
20] M. Solinas, J. Jiang, O. Stelzer, W. Leitner, Angew.
Chem. Int. Ed. 2005, 44, 2291.
21] a) L. J. P. van den Broeke, E. L. V. Goetheer, A. W.
Verkerk, E. de Wolf, B.-J. Deelman, G. Van Koten,
J. T. F. Keurentjes, Angew. Chem. Int. Ed. 2001, 40,
[
2] a) C. W. Kohlpaintner, R. W. Fischer, B. Cornils, Appl.
Catal. A 2001, 221, 219; b) B. Cornils, in: Multiphase
Homogeneous Catalysis, Vol 1, (Eds.: B. Cornils, W. A.
Herrmann, I. T. Horvµth, W. Leitner, S. Mecking, H.
Olivier-Bourbigou, D. Vogt), Wiley-VCH, Weinheim,
4
473; b) E. L. V. Goetheer, A. W. Verkerk, L. J. P.
2
005, p 27.
[
[
3] D. J. Cole-Hamilton, Science 2003, 299, 1702.
4] a) I. T. Horvµth, G. Kiss, R. A. Cook, J. E. Bond, P. A.
van den Broeke, E. de Wolf, B.-J. Deelman, G. van Ko-
ten, J. T. F. Keurentjes, J. Catal. 2003, 219, 126.
Stevens, J. Rµbai, E. J. Mozeleski, J. Am. Chem. Soc.
[22] S. Kainz, D. Koch, W. Baumann, W. Leitner, Angew.
1
998, 120, 3133; b) B. Richter, A. L. Spek, G. van Ko-
Chem. Int. Ed. 1997, 15, 1628.
ten, B.-J. Deelman, J. Am. Chem. Soc. 2000, 122, 3945.
5] P. B. Webb, M. F. Sellin, T. E. Kunene, S. Williamson,
A. M. Z. Slawin, D. J. Cole-Hamilton, J. Am. Chem.
Soc. 2003, 125, 15577.
[23] D. R. Palo, C. Erkey, Ind. Eng. Chem. Res. 1998, 37,
4203.
[24] D. R. Palo, C. Erkey, Organometallics 2000, 19, 81.
[25] D. J. Adams, J. A. Bennett, D. J. Cole-Hamilton, E. G.
Hope, J. Hopewell, J. Kight, P. Pogorzelec, A. M.
Stuart, Dalton Trans. 2005, 24, 3862.
[26] 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.
[
[
[
[
6] K. Kunna, C. Mꢂller, J. Loos, D. Vogt, Angew. Chem.
Int. Ed. 2006, 45, 7289.
7] D. J. Heldebrant, P. G. Jessop, J. Am. Chem. Soc. 2003,
1
25, 5600.
8] a) P. G. Jessop, W. Leitner, (Eds.), Chemical Synthesis
UsingSupercritical Fluids , Wiley-VCH, Weinheim,
[
27] 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.
1
2
999, pp 1–13; b) E. J. Beckman, J. Supercrit. Fluids
004, 28, 121.
[
9] B. Cornils, W. A. Herrmann, I. T. Horvµth, W. Leitner,
S. Mecking, H. Olivier-Bourbigou, D. Vogt, in: Multi-
phase Homogeneous Catalysis, Vol. 1, (Eds.: B. Cornils,
W. A. Herrmann, I. T. Horvµth, W. Leitner, S. Mecking,
H. Olivier-Bourbigou, D. Vogt), Wiley-VCH, Wein-
heim, 2005, p 3.
[
[
28] S.-I. Fujita, S. Fujisawa, B. M. Bhanage, M. Arai, Tetra-
hedron Lett. 2004, 45, 1307.
29] Applying scCO as a solvent” implies that the reaction
2
mixture, reactants, products and CO , were in a super-
2
critical one-phase state.
[
[
30] P. W. N. M. van Leeuwen, in: Homogeneous Catalysis:
Understandingthe Art , Kluwer Academic Publishers,
Amsterdam, 2004, p 139.
[
[
[
10] R. A. Sheldon, Green Chem. 2005, 7, 267.
11] J. M. DeSimone, Science 2002, 297, 799.
12] D. E. Bergbreiter, S. D. Sung, Adv. Synth. Catal. 2006,
31] K. R. Westerterp, W. P. M. van Swaaij, A. A. C. M.
Beenackers, Chemical Reactor Design and Operation,
3
48, 1352.
13] a) A. Corma, H. Garcia, Adv. Synth. Catal. 2006, 348,
391; b) B. M. L. Dioos, I. F. J. Vankelecom, P. A.
[
nd
John Wiley & Sons, Chichester, 2 edn., 1984, p 83.
1
[
[
[
[
32] P. Cavalieri dꢁOro, L. Raimondi, G. Pagani, G. Montra-
si, G. Gregorio, A. Andreetta, Chim. Ind. 1980, 62, 572.
33] K. L. Olivier, F. B. Booth, Hydrocarbon Process. 1970,
Jacobs, Adv. Synth. Catal. 2006, 348, 1413.
[
14] For example: a) R. P. J. Bronger, J. P. Bermon, J. N. H.
Reek, P. C. J. Kamer, P. W. N. M. van Leeuwen, D. N.
Carter, P. Licence, M. Poliakoff, J. Mol. Catal. A 2004,
4
, 112.
34] W. R. Moser, C. J. Papile, D. A. Brannon, R. A.
Duwell, J. Mol. Catal. 1987, 41, 271.
35] C. P. Casey, E. L. Paulsen, E. W. Beuttenmueller, B. R.
Proft, L. M. Petrovich, B. A. Matter, D. R. Powell, J.
Am. Chem. Soc. 1997, 119, 11817.
2
24, 145; b) F. Shibahara, K. Nozaki, Tamejiro Hiyama,
J. Am. Chem. Soc. 2003, 125, 8555; c) Z. K. Lopez-Cas-
tillo, R. Flores, I. Kani, J. P. Fackler, A. Akgerman, Ind.
Eng. Chem. Res. 2003, 42, 3893; d) A. R. Tadd, A. Mar-
teel, M. R. Mason, J. A. Davies, M. A. Abraham, J. Su-
percrit. Fluids 2003, 25, 183.
[
36] P. W. N. M. van Leeuwen, in: Homogeneous Catalysis:
Understandingthe Art , Kluwer Academic Publishers,
Amsterdam, 2004, p 147.
[
15] For example: a) T. J. Devon, G. W. Philips, T. A. Puck-
ette, J. L. Stavinoha, J. J. Vanderbilt, (to Eastman
Kodak), U. S. Patent 4,694,109, 1987; b) M. Kranen-
burg, Y. E. M. van der Burgt, P. C. J. Kamer, P. W. N. M.
[37] S. S. Divekar, B. M. Bhanage, R. M. Deshpande, R. V.
Gholap, R. V. Chaudhari, J. Mol. Catal. 1994, 91, L1.
Adv. Synth. Catal. 2008, 350, 179 – 188
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
187

Ard C. J. Koeken et al.
FULL PAPERS
[
[
38] H. H. Y. Unveren, R. Schomꢃcker, Catal. Lett. 2006,
[40] D. R. Palo, C. Erkey, Ind. Eng. Chem. Res. 1999, 38,
2163.
[41] T. Davis, C. Erkey, Ind. Eng. Chem. Res. 2000, 39, 3671.
[42] M. Gimꢄnez Pedrꢅs, A. M. Masdeu-Bultꢅ, J. Bayardon,
D. Sinou, Catal. Lett. 2006, 107, 205.
1
10, 195.
39] D. Koch, W. Leitner, J. Am. Chem. Soc. 1998, 120,
3398.
1
188
asc.wiley-vch.de
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 179 – 188