Hydroformylation in perfluorinated solvents; improved selectivity, catalyst retention and product separation

Article abstract of DOI:10.1016/S0040-4020(02)00215-6

The hydroformylation of linear terminal alkenes using rhodium based catalysts under fluorous biphasic conditions in the presence and absence of toluene is reported. Using fluorinated ponytails to modify triarylphosphites and triarylphosphines, good selectivities and reactivities can be obtained, along with good retention of the catalyst and ligand within the fluorous phase. Using P(O-4-C₆H₄C₆F₁₃)₃ (P/Rh=3:1) as the ligand in toluene/perfluoro-1,3-dimethylcyclohexane, good results are obtained at 60°C, but decomposition of the catalyst and/or ligand occurs on increasing the temperature. More impressive results are obtained by omitting the toluene, with higher rates, better l/b ratios, and better retention of the catalyst and the phosphite within the perfluorocarbon solvent. Competing isomerisation restricts linear aldehyde selectivities to <76%. When P(4-C₆H₄C₆F₁₃)₃ is used as the ligand in the absence of toluene, even more impressive results can be obtained, with linear aldehyde selectivities up to 80.9%, high rates, and the retention of up to 99.95% of the rhodium and up to 96.7% of the phosphine within the fluorous phase. These results are compared with those of commercial systems for propene hydroformylation and with those previously reported in the literature for hydroformylation under fluorous biphasic conditions. Phase behaviour studies show that 1-octene is completely miscible with the fluorous solvent under the conditions used for the hydroformylation experiments, but that the product nonanal, phase separates.

Full text of DOI:10.1016/S0040-4020(02)00215-6

TETRAHEDRON
Pergamon
Tetrahedron 58 22002) 3901±3910
Hydroformylation in per¯uorinated solvents; improved
selectivity, catalyst retention and product separation
Douglas F. Foster,a David Gudmunsen,b Dave J. Adams,b Alison M. Stuart,b Eric G. Hope,b
David J. Cole-Hamilton,c,p Gary P. Schwarza and Peter Pogorzelecc
aCatalyst Evaluation and Optimisation Service ꢀCATS), School of Chemistry, University of St. Andrews, St. Andrews, Fife KY16 9ST,
Scotland, UK
bDepartment of Chemistry, University of Leicester, Leicester LE1 7RH, England, UK
cSchool of Chemistry, University of St. Andrews, St. Andrews, Fife KY16 9ST, Scotland, UK
Received 24 November 2001; revised 24 January 2002; accepted 31 January 2002
AbstractÐThe hydroformylation of linear terminal alkenes using rhodium based catalysts under ¯uorous biphasic conditions in the
presence and absence of toluene is reported. Using ¯uorinated ponytails to modify triarylphosphites and triarylphosphines, good selectivities
and reactivities can be obtained, along with good retention of the catalyst and ligand within the ¯uorous phase. Using P2O±4-C6H4C6F13)3
2P/Rhˆ3:1) as the ligand in toluene/per¯uoro-1,3-dimethylcyclohexane, good results are obtained at 608C, but decomposition of the catalyst
and/or ligand occurs on increasing the temperature. More impressive results are obtained by omitting the toluene, with higher rates, better l/b
ratios, and better retention of the catalyst and the phosphite within the per¯uorocarbon solvent. Competing isomerisation restricts linear
aldehyde selectivities to ,76%. When P24-C6H4C6F13)3 is used as the ligand in the absence of toluene, even more impressive results can be
obtained, with linear aldehyde selectivities up to 80.9%, high rates, and the retention of up to 99.95% of the rhodium and up to 96.7% of the
phosphine within the ¯uorous phase. These results are compared with those of commercial systems for propene hydroformylation and with
those previously reported in the literature for hydroformylation under ¯uorous biphasic conditions. Phase behaviour studies show that
1-octene is completely miscible with the ¯uorous solvent under the conditions used for the hydroformylation experiments, but that the
product nonanal, phase separates. q 2002 Elsevier Science Ltd. All rights reserved.
1. Introduction
There is considerable interest in replacing the cobalt based
catalysts with ones based on rhodium, since they are much
more active, work under much milder conditions and give
good l/b ratios. Such catalysts, modi®ed with PPh3, are used
commercially for the hydroformylation of short chain
alkenes 2C2±C5), but cannot be used for higher alkenes
because they decompose below the boiling point of the
product aldehyde and so the separation of the reaction
product from the catalyst is very dif®cult. One approach
to the resolution of this problem has been to render the
catalyst water soluble by sulphonating the ligand and to
carry out the reaction in a two phase aqueous±organic
system.5 This process has been commercialised for propene
and works well for chain lengths up to C7, but the very low
solubility of longer chain alkenes in the aqueous phase,
where the reaction takes place, means that the rates of
reaction are too low for a commercially viable process for
detergent aldehydes to be developed based on this approach.
To date, no rhodium based system has been commercialised
for the hydroformylation of long chain alkenes, although the
installation of a process involving Rh/PPh3 and low pressure
distillation of the products is currently in progress.6
Hydroformylation is an important industrial process used in
the production of some 6 M tonnes per year of aldehydes for
use in the manufacture of soaps, detergents and plasti-
cisers.1,2 One of the big challenges left in this area is to
develop rhodium based processes for the hydroformylation
of long chain alkenes 2C8±C24) to aldehydes, which are
important intermediates for plasticisers and detergents.
Current technology, which produces .1 M tonnes of
product per year is based on cobalt catalysts, often modi®ed
with tertiary phosphines.1,3 However, the products of these
cobalt catalysed reactions are usually alcohols rather than
aldehydes. These catalysts suffer from the disadvantages
that they require very forcing conditions 2close to 2008C
and 100 bar) and that catalyst recycling is dif®cult. Often
the percentage of the linear product produced in these
processes is low, although Shell have introduced a phos-
phine formally derived from addition of C20H42PH2 across
cycloocta-1,5-diene, which can give linear/branched 2l/b)
ratios as high as 10.4
Keywords: ¯uorous; biphasic; hydroformylation; rhodium.
Corresponding author. Tel.: 144-1334-463805; fax: 144-1334-463808;
e-mail: djc@st-and.ac.uk
p
Various new approaches to the development of rhodium
based catalysts for the hydroformylation of long chain
0040±4020/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S0040-4020202)00215-6

3902
D. F. Foster et al. / Tetrahedron 58 ꢀ2002) 3901±3910
alkenes are being explored, but amongst the most promising
is the use of a biphasic system based on ¯uorous and organic
in many ways, giving better activity, linear aldehyde selec-
tivity, and retention of both the ligand and the rhodium
complex in the ¯uorous phase. A preliminary communica-
tion of some of these results has been submitted11 and two
patents have been ®led.12
7,8
Â
solvents, originally introduced by Horvath.
These
per¯uorocarbon solvents ostensibly do not mix with organic
solvents. So, if a catalyst is designed to be selectively
soluble in the ¯uorous phase, any product that will be in
the organic phase can be separated by simple decantation.
2. Experimental
Â
The great beauty of the system that Horvath explored, which
gives it advantages over the aqueous biphasic approach, is
that one phase is formed under the hydroformylation
conditions, and phase separation occurs rapidly and cleanly
Gas chromatographic analyses of the product mixtures were
carried out on a Hewlett±Packard 5890 series gas chromato-
graph equipped with both a ¯ame ionisation detector
2GCFID) 2quantitative analyses) and a Hewlett±Packard
5890 series mass selective detector 2GCMS) for qualitative
analyses. The gas chromatograph was interfaced with a
Hewlett±Packard Chemstation for the determination of
peak areas by electronic integration. The GCMS method
employed a Supelcoe Meridian MDN-35 low polarity,
bonded crosslinked phase comprising of 235% phenyl)-
methylpolysiloxane fused silica capillary column
230 m£0.25 mm£0.25 mm ®lm thickness). The GCFID
method employed the more polar column: Supelcowax 10,
bonded carbowax 20 M poly2ethylene glycol) fused silica
capillary column 260 m£0.32 mm£1.0 mm ®lm thickness).
Â
on cooling. In his work, Horvath used rhodium com-
plexes of the ¯uorous-derivatised trialkylphosphine,
P2CH2CH2C6F13)3, and demonstrated that they could give
good activity for the hydroformylation of 1-decene with
good retention in the ¯uorous phase, such that the catalyst
could be recycled 9 times with only 4.2% loss of rhodium.
Leaching of the phosphine into the organic phase, although
not measured, was believed to be more extensive on the
basis of changes in the kinetics and selectivity of the reac-
Â
tion with catalyst reuse. The one problem with Horvath's
results is that the selectivity to the desired linear product 2l/b
ratio) was modest unless high concentrations of the expen-
sive ¯uorous modi®ed ligand were employed 2ligand
concentrations of 152 and 304 mmol dm23 at 1igand/Rh
ratios of 103 and 102, respectively, gave the best l/b ratios
at 6.3 and 7.8, respectively). In addition, alkene isomerisa-
tion was a competing reaction, consuming 8±9% of the
terminal alkene in many cases.
ICPMS analyses for Rh and P content in the organic phases
were carried out at the Chemical Analysis Group of Agilent
Technologies UK Ltd on an Agilent 7500a instrument. The
instrument was modi®ed for precise addition of oxygen into
the carrier gas to ensure complete organic sample decom-
position and a special plasma torch with narrow internal
diameter injector was used to reduce the effects of the
solvent vapour on the plasma. Platinum tipped cones were
employed together with a self-aspirating nebuliser and
solvent resistant seals. All measurements were carried out
in triplicate and compared with calibration curves obtained
using toluene solutions of known concentrations of Rh and
P, in the form of [Rh22,4-pentanedionato)2CO)2]
2[Rh2acac)2CO)2]) and PPh3, respectively.
Xiao has described an alternative system in which the
phosphine ligand is incorporated into a polymer by copoly-
merising styryldiphenylphosphine with 1H,1H,2H,2H-
per¯uorodecylacrylate.9 This proved to be active as a ligand
for hydroformylation under ¯uorous biphasic conditions
2hexane/toluene/per¯uoromethylcyclohexane 240:20:40%
v/v), with selectivities for 1-decene hydroformylation to
the linear aldehyde of up to 85%, but the rates are relatively
low 2turnover frequencies at 1008C,150 mol product
mol Rh21 h21). Experiments on catalyst leaching suggested
1 ppm Rh loss per hour, but complete loss of the per¯uoro-
methylcyclohexane into the organic phase by the end of the
third run. 70,000 catalyst turnovers were obtained with the
one batch of catalyst.
The per¯uorocarbon solvents, per¯uoromethylcyclohexane
and per¯uoro-1,3-dimethylcyclohexane 2Apollo) were
thoroughly dried 2by distillation from CaH2) and degassed
2N2 purge) before use. Toluene 2Ultra®ne) and the various
alkenes 2Aldrich) were dried and degassed by distillation
from Na under
a
N2 atmosphere. The complex
It is well known that triarylphosphine ligands can give much
better linear selectivity in rhodium based hydroformylation
reactions than trialkylphosphines,1,2,10 so we reasoned that
triarylphosphines bearing ¯uorous ponytails might be good
ligands for rhodium catalysed ¯uorous biphasic hydro-
formylation. We, therefore, tested them and also investi-
[Rh2acac)2CO)2] 2Strem) was used as received whilst the
¯uorous ligands were prepared by standard literature
procedures.13
2.1. Phase behaviour
Â
gated ¯uorous derivatised triarylphosphites. Horvath has
suggested that these phosphines may be less suitable
because their higher polarity might make them more soluble
in the organic phase than the trialkylphosphines.8 This may,
however, not be a serious problem provided that the
rhodium is retained in the ¯uorous phase, since the ligands
are expected to be thermally more stable than their rhodium
complexes and could be separated from the aldehyde
products by conventional distillation and recycled to the
reactor. We now report the results of these studies together
with results obtained in systems where the organic solvent is
omitted. Using just a per¯uorinated solvent proved superior
A high pressure NMR tube consisting of a 10 mm sapphire
tube and titanium head was charged with the liquids under
study.14,15 It was then pressurised with CO/H2 as required
and observed behind a screen. The tube was heated by
placing it in a beaker of silicon oil at known temperatures
and the phase behaviour observed. The sample was agitated
when necessary by gently shaking the tube.
Attempts to use HPIR in a CIR cell14,15 and to monitor
the concentration of toluene in the ¯uorous phase were
unsuccessful.

D. F. Foster et al. / Tetrahedron 58 ꢀ2002) 3901±3910
3903
Table 1. Phase behaviour of the per¯uoromethylcyclohexane/toluene
solvent system under typical hydroformylation conditions used in this
work 220 bar CO/H2 1:1)
2.2. Catalytic reactions
2.2.1. Typical run. An autoclave, ®tted with a substrate
injector containing 1-octene 21 cm3, 6.37 mmol), a mechan-
ical stirrer, a gas delivery system, an injection port and a
thermocouple was ¯ushed with CO/H2 21:1) to remove air.
Per¯uoro-1,3-dimethylcyclohexane 24 cm3) containing di-
carbonyl22,4-pentanedionato)rhodium2I) 2[Rh2acac)2CO)2],
0.01 mmol) and P2O±4-C6H4C6F13)3 20.03 mmol) was
added through the injection port against a stream of CO/
H2 using a syringe. The autoclave was pressurised with
CO/H2 21:l) to 20 bar and the pressure released. This ¯ush-
ing procedure was repeated twice more. After pressurising
to 16 bar the stirrer was started 21000 rpm) and the auto-
clave was heated to 608C for 45 mins. The 1-octene was
then added to the autoclave by forcing it in through the
substrate injector using a CO/H2 pressure of 20 bar. The
temperature, pressure in the autoclave and pressure in a
ballast vessel, from which gas was fed into the autoclave
through a mass ¯ow controller to keep the pressure within
the autoclave constant at 20 bar, were monitored and
recorded every 5 s.
Componenta of mixture 2cm3)
Phase behaviour
T
F
1-O
N
2.5
2
2.5
2
±
1
±
±
Partially miscible at 808C
Homogeneous single phase$ca.
608C
2
2
±
2
2
4
±
0.5
1
1
0.5
±
No apparent mixing up to 808C
Two layers persist
Homogeneous single phase$ca.
608C
No apparent mixing up to 808C
±
4
±
1
a
Toluene 2T), per¯uoromethylcyclohexane 2F), 1-octene 21-O), nonanal
2N).
Â
and con®rm that the mixtures described by Horvath are
monophasic under his conditions. More interestingly, we
discovered that 1-octene is fully miscible with per¯uoro-
methylcyclohexane at temperatures $ca. 608C under typical
hydroformylation conditions employed in this work 220 bar
CO/H2 l:l), but that nonanal phase separated even at 808C.
After 1 h, the stirrer was stopped and the autoclave was
allowed to cool. The gases were vented and the mixture
was syringed into a sample vial. The phases were separated
using a Pasteur pipette and then analysed. The colourless
organic phase was analysed for its organic components by
GCFID and GCMS, and for its rhodium and phosphorus
content by ICPMS. Samples of the as-obtained organic
phases were diluted with 9£ the volume of pure xylene
prior to the ICPMS analysis. The ¯uorous phase from
selected experiments were analysed for their organic
content by GCFID and GCMS. Kinetic data were obtained
from an analysis of the pressure drop in the ballast vessel
versus time.
3.2. Preliminary ligand screening
Table 2 lists the results obtained from the preliminary
screening of a variety of ligands in the rhodium catalysed
hydroformylation of l-hexene under ¯uorous biphasic
conditions using per¯uoro-1,3-dimethylcyclohexane and
toluene as the solvents together with a comparison with
PPh3 or P2OPh)3 under identical conditions and using
PPh3 in toluene alone. Throughout these preliminary screen-
ing experiments, the extent of rhodium leaching was not
measured, but visually, the retention into the ¯uorous
phase looked very good 2Fig. 1). These results show that
the reaction with PPh3 is almost unaffected by the presence
of the ¯uorous solvent, the kinetics are zero order in both
cases, although a reduced extent of alkene isomerisation
For reactions carried out in 50:50 toluene/¯uorous solvent
mixtures, 2 cm3 toluene replaced 2 cm3 of the ¯uorous
solvent so that the total amount of solvent remained
constant. Reactions were carried out at a variety of different
temperatures and loadings of rhodium precursor and/or
ligand. Most reactions were carried out once, but several
were carried out in duplicate or triplicate and show excellent
reproducibility for the catalytic results and reasonable to
good reproducibility for the leaching measurements. Table
2 contains results obtained during pre-screening of a number
of different ligands and Tables 3 and 4 contain the results
obtained while studying the ligands, P2O±4-C6H4C6F13)3,
and P24-C6H4C6F13)3, respectively, in more detail.
leads to an increased selectivity to the linear aldehyde.
Â
8
Horvath also reported a higher l/b ratio in the biphasic
mixture in the hydroformylation of l-decene, although in
his case the extent of isomerisation was very similar for
Â
the two solvent systems. Horvath found a slightly higher
rate in toluene, but our studies suggest a higher rate in the
biphasic system.
Excitingly the phosphites, P2O±3 or ±4-C6H4C6F13)3
showed much better l/b ratios and higher initial rates than
PPh3. These reactions were generally carried out at a [Rh] of
10 mmol dm23 2about l0±100£ those used by Horvath),
8
Â
and even at a ligand concentration of only 30 mmol dm23
,
3. Results and discussion
3.1. Phase behaviour
the l/b ratios of the aldehyde product were as high as 8.4. It
should be noted that l/b ratios depend on the total ligand
concentration not on the 1igand/rhodium ratio, and that,
because of the cost of the ligand, it is desirable to keep
the ligand concentration as low as possible. One experiment
was carried out using P2O±4-C6H4C6F13)3 with both the
ligand and rhodium concentration reduced by a factor of
Â
Horvath has reported that mixtures of alkene, toluene and
per¯uoromethylcyclohexane 220:40:40% by volume) are
monophasic above 1008C under the hydroformylation
conditions he used 210 bar CO/H2 1:l).8 In order to have
more information on the phase behaviour of our mixtures,
we examined visually solutions contained in a 10 mm
sapphire NMR tube. The results are presented in Table 1
Â
10 2to a similar rhodium loading to that used by Horvath,
but with a much lower ligand loading) but this gave, as
expected for the order of magnitude reduction in ligand
concentration, a lower l/b ratio 23.1). One disadvantage of

3904
D. F. Foster et al. / Tetrahedron 58 ꢀ2002) 3901±3910
Table 2. Hydroformylation of 1-hexene, using rhodium complexes of various phosphines and phosphites under ¯uorous biphasic conditions
Ligand
Conva 2%)
Isomb 2%)
Branchedb 2%)
Linearb 2%)
l/bc
Rate constd 2s21
)
TOF2I)e 2h21
)
2.5£1023g
3.2£1023g
6.3£1023
4.8£1023g
4.6£1023
1.8£1023
3.2£1023
2.3£1023
7.3£1023
3.1£1023
2.1£1023
900
1150
3600
1700
2650
10400
1850
1300
4200
1300
880
f
PPh3
PPh3
99.1
99.1
98.4
99.6
99.2
98.6
99.6
99.9
99.9
99.0
86.1l
6.1
1.8
10.6
8.0
14.6
9.3
10.7
3.9
25.1
24.0
18.2
23.7
9.0
21.8
11.7
15.4
53.6
10.8
64.5
67.9
73.3
69.6
67.9
75.6
67.5
77.2
80.6
45.5
74.2
10.4
2.7 22.2)
3.1 22.8)
3.8 22.4)
2.9 22.1)
8.4 23.2)
3.1 22.2)
6.6 23.4)
5.2 24.2)
0.9 20.8)
6.9 23.0)
0.2 20.2)
P24-C6H4C6F13)3
P2OPh)3
P2O±4-C6H4C6F13)3
P2O±4-C6H4C6F13)3
P2O±4-C6H4C6F13)3
P2O±3-C6H4C6F13)3
P2O±2-C6H4C6F13)3
P2O±4-C6H4C6F13)3
P2O±4-C6H4C6F13)3
h
i
0.8
14.0
24.9
j
k
A solution of [Rh2acac)2CO)2] 210.0 mmol dm23), phosphine/phosphite 230.0 mmol dm23) in a mixture of per¯uoro-1,3-dimethylcyclohexane 22.0 cm3),
toluene 22.0 cm3) and l-hexene 21.0 cm3, 8.0£1023 mol) at 708C and 20 bar CO/H2 21:1) for 1 h.
a
Conversion 2100%2% residual 1-hexene).
Percent of product by mole fraction. Isomˆsum 22- and 3-hexene); branchedˆsum 22-ethylpentanal12-methylhexanal); linearˆn-heptanal.
b
c
l/bˆratio of linear to branched aldehyde product. The ®gure in brackets refers to the ratio of n-heptanal to the sum 2branched aldehyde1isomerised alkene).
d
All reactions were found to be ®rst order in [1-hexene] to over 80% of 1-hexene consumption, unless denoted by footnote g.
TOF2I)ˆinitial turn over frequency. Moles 1-hexene converted to aldehyde per mole Rh per hour.
e
f
In toluene only 24 cm3).
g
Zero order rate constant, mol dm23 s21. Gas uptake consistent with `saturation' kinetics typically found when aryl phosphines are present at such low
concentrations 2zero order becoming ®rst order after some time).
h
[Rh2acac)2CO)2] 21.0 mmol dm23) and phosphite 23.0 mmol dm23).
8 bar CO/H2 21:1).
i
j
1-Nonene as substrate; Isomˆsum 22-, 3- and 4-nonene); branchedˆsum 22-butylhexanal12-propylheptanal12-ethyloctanal12-methylnonanal);
linearˆn-decanal.
k
2-Nonene as substrate, other assignments as footnote j.
Conversion 2100%2% residual 2-nonene).
l
these phosphite ligands was that signi®cantly higher
isomerisation was observed than when using PPh3. Electron
withdrawing groups 2¯uorous tails) attached to the aryl ring
of phosphites 2or phosphines) afford less basic ligands,
resulting in more facile CO dissociation in the catalytic
cycle and increased rates of hydroformylation. However,
and particularly with phosphites, the concomitant increase
in the electophilicity of the metal centre leads to increased
amounts of isomerisation which, in part, accounts for the
high l/b ratio observed for P2O±4-C6H4C6F13)3.16,17 Using
P2O±2-C6H4C6F13)3 as the ligand gave poor selectivity,
presumably because the ortho substituent results in signi®-
cant steric crowding on coordination.13b
Markovnikov, linear cycle) determines which cycle will
be followed, but the branched alkyl complex can either
undergo carbonylation to give the aldehyde or b-hydrogen
abstraction to give the isomerised alkene. Lack of anti-
Markovnikov selectivity in the ®rst step reduces the l/b
ratio, but lack of carbonylation selectivity in the second
leads to an increased l/b ratio at the expense of total
aldehyde production. A better indication of the success of
these hydroformylation reactions is the overall selectivity to
linear aldehyde, which is the desired product. On this
criterion, the phosphites are not signi®cantly better than
PPh3. The ¯uorinated phosphine, P24-C6H4C6F13)3, showed
®rst order kinetics and a high rate, but the selectivity under
these conditions was unexceptional.
The l/b ratio is determined in the catalytic cycle by two
effects. The sense of addition of the Rh±H bond across
the alkene 2Markovnikov, branched cycle, or anti-
One other experiment conducted in these screening runs
used 2-nonene as the substrate. Rhodium catalysts are
often not very active for the hydroformylation of internal
alkenes, but the catalyst based on P2O±4-C6H4C6F13)3 gave
a rate constant that was ca. 2/3 of that obtained for 1-nonene
and signi®cant quantities of linear aldehyde 2decanal) were
obtained.
Following these preliminary results, we were interested to
investigate the ligand, P2O±4-C6H4C6F13)3, in more detail
and under more realistic conditions. These optimisation
reactions were carried out using 1-octene as the substrate.
3.3. P%O±4-C6H4C6F13)3 as ligand with toluene/
per¯uoro-1,3-dimethylcyclohexane as solvent
The results obtained using P2O±4-C6H4C6F13)3 as the ligand
for the hydroformylation of 1-octene under different condi-
tions are shown in Table 3. For comparative purposes,
Figure 1. Retention of catalyst in ¯uorous phase 2orange) after hydrofor-
mylation below the colourless organic phase 2toluene/aldehyde).

Table 3. Hydroformylation of 1-octene at 20 bar CO/H2 21:1) employing the ¯uorinated phosphitc ligand, P2O±4-C6H4C6F13)3
Solventa
[Rh]b
[P]b
T
28C)
Time
2min)
Convc
2%)
Octaned
2%)
Isomd
2%)
Branchedd
2%)
n-Nonanald
2%)
l/be
Rate
constantf
TOF2I)g
2h21)
[Rh]/Aldh
[P]/Aldh
2mmol dm23
)
2mmol dm23
)
2mg mol2l
)
2mg mol2l
)
2s21
)
T/F
T/F
T/F
T/F
T/F
T/F
F
F
F
F
F
2.0
2.0
2.0
2.0
2.0
10.0
2.0
6.0
6.0
6.0
6.0
6.0
30.0
6.0
60
70
80
90
100
70
60
70
80
60
70
80
60
70
60
60
60
60
60
30
60
30
30
45
30
30
60
60
96.9
96.2
96.3
93.0
69.8
97.0
97.6
97.5
96.6
97.2
97.3
96.5
97.8
98.2
0.4
0.4
0.5
0.6
0.7
0.5
0.3
0.4
0.5
0.3
0.4
0.5
0.4
0.4
13.0
12.7
16.3
26.3
32.3
16.9
12.4
13.8
15.6
13.7
13.8
13.1
3.7
14.3
16.4
15.9
12.1
6.7
8.5
9.7
9.5
11.1
7.6
8.1
9.8
26.7
31.1
69.2
66.7
63.6
54.0
30.1
71.1
75.2
73.8
69.4
75.6
75.0
73.1
67.0
64.7
4.8 22.5)
4.1 22.3)
4.0 22.0)
4.5 21.4)
4.5 20.8)
8.4 22.8)
7.8 23.4)
7.8 23.2)
6.3 22.6)
9.9 23.6)
9.3 23.4)
7.5 23.2)
2.5 22.2)
2.1 22.0)
8.7£1024
1.4£1023
2.5£1023
2000
3200
5700
15.7 29.7)
62.9 238.9)
87.7 254.3)
88.0 254.5)
134.7 283.4)
24.0 23.0)
10.3 26.4)
10.2 26.3)
8.2 25.1)
31.8 221.8)
69.1 247.4)
87.8 260.1)
106.2 272.7)
129.9 289.0)
55.4 26.9)
i
±
i
±
4.5£1023
1.3£1023
3.5£1023
6.8£1023
1.9£1023
4.9£1023
1.1£1022
3.2£1023
2.4£1023k
2100
3000
8000
15600
870
2250
5000
1500
4400
14.0 29.6)
2.0
2.0
6.0
6.0
14.7 210.1)
17.1 211.7)
28.6 23.9)
43.7 26.0)
97.0 213.3)
10.0
10.0
10.0
10.0
2.0
30.0
30.0
30.0
30.0j
6.0j
20.6 22.6)
26.7 23.3)
53.0 26.6)
F
T
T
2.0
a
Reaction solvent. Either 4 cm3 of per¯uoro-1,3-dimethylcyclohexane 2F) or a mixture of 2 cm3 each of toluene and per¯uoro-1,3-dimethylcyclohexane 2T/F).
b
c
d
e
f
Concentration in mmol dm23 of [Rh2acac)2CO)2] [Rh] or P2O±4-C6H4C6F13)3 [P] in 5 cm3 reaction solution 2i.e. 4 cm3 of either F or 50/50 T/F11 cm3 1-octene). [1-octene]ˆ1274 mmol dm23 for all reactions.
Conversion 2100%2% residual 1-octene).
Percent of product by mole fraction. Isomˆsum 22-, 3- and 4-octene), generally .95% 2-octene; branchedˆsum 22-propylhexanal12-ethylheptanal12-methyloctanal), generally .95% 2-methyloctanal.
l/bˆratio of linear to branched aldehyde product. The ®gure in brackets refers to the ratio of n-nonanal to the sum 2branched aldehyde1isomerised alkene).
All reactions were found to be ®rst order in [1-octene] to over 80% of 1-octene consumption, unless denoted by footnote k.
TOF2I)ˆinitial turn over frequency. Moles 1-octene converted to aldehyde per mole Rh per hour.
g
h
Rhodium and phosphine loss to the organic product. Given as the concentration of Rh and Pby weight 2mg) per mole of aldchyde product. Figure in brackets is the % of Rh/Plost from the catalyst phase to the organic
product phase.
i
Kinetics of gas uptake con®rm a self catalysing decomposition of catalyst on injection of substrateÐhydrolysis or, more likely, reaction of aldehyde with phosphite 2see Section 3).
P2OPh)3 reactions 24 cm3 toluene11 cm3 1-octene).
j
k
Zero order rate constant 2mol dm23 s21).

3906
D. F. Foster et al. / Tetrahedron 58 ꢀ2002) 3901±3910
reaction temperature of 1008C. ICPMS was also used to
analyse the organic phase for phosphorus and this was
signi®cant at 608C 221.8%) again increasing on heating to
89% at 1008C. The increases observed on heating clearly
show that the catalyst and/or ligand are decomposing. If
they were stable, the amount of leaching would be expected
to be independent of temperature, since the separation is
carried out at the same temperature for each product
solution. Provided that the catalysts and ligand are stable
under the reaction conditions and the rhodium and phos-
phorus loadings are constant, the amount of rhodium and
phosphorus in the organic phase should be determined
solely by the partition coef®cients of the catalyst and the
ligand respectively between the two phases.
The high amounts of rhodium and phosphorus leaching in
the reactions carried out at 60±808C 2Table 3) initially
appears to be inconsistent with the classic ®rst order kinetics
observed at these lower temperatures 2Fig. 2). However, this
kinetic behaviour itself con®rms that catalyst decomposi-
tion, and concomitant leaching of rhodium and phosphorus,
only occurs when the majority of the 1-octene has been
converted to aldehyde; it takes only 26, 17 and 10 min for
75% of the 1-octene to hydroformylate at 60, 70 and 808C,
respectively 2Fig. 2), whereas the reactions were allowed to
proceed for 60 min in order to reach completion. The
evidence from the ®rst order kinetics, leaching data, and
the detection of HO±4-C6H4C6F13 in the GCMS of the
aldehyde product from all the reactions suggests that the
mode of decomposition of the catalysts involves the decom-
position of the phosphite ligand in the presence of the
product aldehyde. There are a number of ways in which
the aryl phosphite may decompose to afford HO±4-
C6H4C6F13 under the hydroformylation conditions
employed in the present study.l8,19 Simple thermal or
metal catalysed decomposition can be ruled out as its effect
would become apparent during the catalyst equilibration
period, 45±60 min, before injection of the 1-octene. Other
modes of decomposition of aryl phosphites in hydroformyl-
ation reactions include hydrolysis, nucleophilic attack on
aldehydes and oxidative cyclisations with aldehydes.18,19
Again, the simple ®rst order kinetics suggests that
hydrolysis cannot be taking place before the formation of
the aldehyde i.e. the presence of residual water in the reactor
cannot account for the decomposition. Aldol condensation
of the aldehyde product could lead to traces of water but we
have found no evidence for any traces of the expected Aldol
products from GCMS 2under known conditions for detec-
tion) of the reaction products, even for those reactions
carried out at 90 and 1008C where catalyst decomposition
is evident even at the start of aldehyde formation 2Fig. 2).
When we tried to carry out the hydroformylation reactions
with the same [Rh] concentration 22.0 mmol dm23) but with
a 15:1 ratio of phosphite to rhodium at 808C 2minimum
temperature to achieve signi®cant reaction), the gas uptake,
on injection of the substrate, very quickly began to deviate
from ®rst order behaviour and at the end of the reaction the
¯uorocarbon layer was, in contrast to the normal orange
catalyst solution colour, completely colourless and
apparently devoid of catalyst. The GCMS of the organic
layers in the reactions employing the high phosphite to
rhodium ratios not only contained high levels of HO±4-
C6H4C6F13 but also signi®cant amounts of aldehyde
Figure 2. ln[2l2conversion)/100] vs time plot for the hydroformylation of
1-octene with Rh/P2O±4-C6H4C6F13)3 in a 1:l toluene/C8F16 solvent
mixture under 20 bar CO/H2 21:1) with [Rh]ˆ2.0 mmol dm23, [P2O±4-
C6H4C6F13)3]ˆ6.0 mmol dm23 and [1-octene]ˆ1274 mmol dm23 at 2W)
kˆ2.5£1023 s2l; 2S) 908C; and 21) 1008C.
608C, kˆ8.7£1024 s2l
;
2K) 708C, kˆ1.4£1023 s2l
;
2X) 808C,
examples using toluene as solvent and P2OPh)3 or PPh3 as
the ligand are also included in Tables 3 and 4, respectively.
As expected the rate increased as the temperature was
increased from 60±808C, but above 808C, the kinetics of
the gas uptake suggested that the catalyst was decomposing
during the course of the reaction 2Fig. 2 and Table 3).
Incomplete conversion of the 1-octene was also obtained
at these higher temperatures. The l/b ratio dropped slightly
from 4.8±4.0 as the temperature was increased from
60±808C, mainly because of lower anti-Markovnikov
selectivity in the formation of the alkyl intermediates, as
indicated by a lower selectivity to nonanal. At the higher
temperatures, isomerisation seems to be a more important
pathway and this is most obvious when the catalyst is being
destroyed. At 608C, the organic phase appeared colourless,
but ICPMS analysis showed signi®cant leaching of rhodium
29.7%). This rhodium leaching increased up to 83.4% at a

Table 4. Hydroformylation of 1-octene in per¯uoromethylcyclohexane at 20 bar CO/H2 21:1) employing the ¯uorinated phosphine ligand, P24-C6H4C6F13)3
[Rh]a
[P]a
T
28C)
Time
2min)
Convb
2%)
Octanec
2%)
Isomc
2%)
Branchedc
2%)
n-Nonanalc
2%)
l/bd
Rate constante
2s21
TOF2I)f
2h21
[Rh]/Aldg
[P]/Aldg
2mmol dm23
)
2mmol dm23
)
)
)
2mg mol21
)
2mg mol21
)
2.0
2.0
2.0
1.0
2.0
2.0
2.0
10.0
10.0
10.0
6.0
6.0
6.0
3.0
20.0
20.0
20.0
30.0i
30.0i
30.0i
70
80
90
70
70
80
90
60
70
80
30
30
30
60
90
30
30
60
30
10
97.1
97.2
95.0
98.3
97.8
97.9
97.9
97.1
97.8
97.7
0.3
0.4
0.4
0.3
0.4
0.4
0.5
0.4
0.3
0.3
2.1
4.2
8.4
2.1
3.7
3.1
4.1
0.5
0.7
1.7
20.3
20.7
21.5
21.1
12.8
16.6
17.2
24.0
24.5
24.3
74.4
71.9
64.7
74.8
80.9
77.8
76.1
72.2
72.3
71.4
3.7 23.3)
3.5 22.9)
3.0 22.2)
3.6 23.2)
6.3 24.9)
4.7 23.9)
4.4 23.6)
3.0 23.0)
3.0 22.9)
2.9 22.8)
1.9£1023h
2.9£1023h
3.0£1023h
9.1£1024h
1.9£1023
5.9£1023
1.3£1022
1.1£1023h
3.0£1023h
7.0£1023h
3400
5300
5400
3300
4400
13500
29800
400
0.8 20.5)
2.3 21.4)
15.8 29.8)
1.9 22.3)
0.08 20.05)
0.38 20.23)
0.54 20.33)
9.1 26.2)
10.7 27.3)
15.5 210.6)
6.7 29.2)
15.9 23.3)
24.1 25.0)
18.6 23.8)
1100
2500
a
Concentration in mmol dm23 of [Rh2acac)2CO)2] [Rh] or P24-C6H4C6F13)3 [P] in 5 cm3 reaction medium 2i.e. 4 cm3 C7F1411 cm3 1-octene). [1-octene]ˆ1274 mmol dm23 for all reactions.
b
c
d
e
f
Conversion 2100%2% residual 1-octene).
Percent of product by mole fraction. Isomˆsum 22, -3- and 4-octene), generally .95% 2-octene; branchedˆsum 22-propylhexanal12-ethylheptanal12-methyloctanal), generally .98% 2methyloctanal.
l/bˆratio of linear to branched aldehyde product. The ®gure in brackets refers to the ratio of n-nonanal to the sum 2branched aldehyde1isomerised alkene).
All reactions were found to be ®rst order in [1-octene] to over 80% of 1-octene consumption, unless denoted by footnote h.
TOF21)ˆinitial turn over frequency. Moles 1-octene converted to aldehyde per mole Rh per hour.
g
Rhodium and phosphine loss to the organic product. Given as the concentration of Rh and Pby weight 2mg) per mole of aldehyde product. Figure in brackets is the % of Rh/Plost from the catalyst phase to the organic
product phase.
h
i
Zero order rate constant, mol dm23 s21 Gas uptake consistent with `saturation' kinetics typically found when aryl phosphines are present at such low concentrations 2zero order becoming ®rst order after some time).
PPh3, reactions 24 cm3 toluene11 cm3 1-octene).

3908
D. F. Foster et al. / Tetrahedron 58 ꢀ2002) 3901±3910
condensation products. Thus, we cannot rule out Aldol
condensation as the primary mode leading to phosphite
decomposition. However, the most important reaction
between a phosphite and an aldehyde is addition to give a
phosphonate.18,19 The reaction is autocatalytic since it is
acid catalysed and produces an acidic product. The
increased levels of acidic phosphonate together with the
extended reaction times 25 h at 808C) for these higher
phosphite to rhodium ratio reactions could also be
responsible for the observed Aldol condensation products.
We are continuing to investigate the exact mode of catalyst
decomposition.
except that the toluene was replaced by the same volume
of ¯uorocarbon solvent. The results are both surprising and
remarkable. In particular, the reaction rates are increased
signi®cantly and this increase is higher at 808C 2factor of
2.7) than at 608C 2factor of 1.5), and the l/b ratios are
dramatically improved, 7.8 at both 60 and 708C compared
with 4.8 and 4.1 at the same temperatures in the presence of
toluene. The reason for the higher rate and linear selectivity
in the absence of toluene are not clear. The difference in rate
may be related to any, or all, of the following: the order of
reaction with respect to the two gases, the differential solu-
bility of the gases, since this affects the partial pressure in
the gas phase and the solvation of the various complexes in
the different solvent systems. Since we have not measured
any of these parameters, we feel further speculation is
unwarranted. The extent of leaching of the rhodium 26%)
and of the phosphorus 29.6 and 10.1% at 60 and 708C
respectively) are greatly reduced and there is much less of
a temperature effect suggesting that the catalyst is more
stable in this system. This stabilisation is not suf®cient,
however, to allow the reaction to be successfully carried
out at higher temperatures. Using the higher catalyst and
ligand loadings, the l/b ratio can be pushed even higher
29.9) and both the rhodium 22.5%) and phosphorus 24%)
leaching are reduced in percentage terms, but not in absolute
terms.
Increasing the concentrations of rhodium and the ligand by a
factor of ®ve and carrying out the reaction at 708C leads to
an increase in rate, presumably because the retarding effect
of the excess ligand is exceeded by the increase in rate
caused by the increased [catalyst]. More dramatically, the
l/b ratio is greatly increased 2to 8.4), with the rhodium
leaching being only 3% 2cf. 38.9% under the same condi-
tions at the lower concentration). The phosphorus leaching
is also reduced 26.9% cf. 47.4% at the lower loading).
The selectivity to linear aldehyde using P2O±4-C6H4C6F13)3
in the mixed solvent system at 60±808C is similar to that
obtained when using P2OPh)3 in toluene 2Table 3).
However, comparison of the l/b ratios and yields of
isomerised alkene, clearly shows an increased tendency
for the branched alkyl intermediate to undergo b-hydrogen
abstraction when using P2O±4-C6H4C6F13)3 as the ligand
under biphasic conditions. As stated previously, this type
of behaviour is well documented for low pressure hydro-
formylations when electron withdrawing groups are substi-
tuted into the para-position of the aromatic ring of
phosphites.16
3.5. P%4-C6H4C6F13)3 as ligand
As a consequence of the rhodium and phosphorus leaching
and the detection of HO-4-C6H4C6F13 in the GCMS of the
aldehyde products from the hydroformylation studies with
P2O±4-C6H4C6F13)3, we decided to investigate further the
phosphine, P24-C6H4C6F13)3, which does not react with
aldehydes. For these experiments, only ¯uorous solvent
2no toluene) was used and this was per¯uoromethylcyclo-
hexane in place of per¯uoro-1,3-methylcyclohexane. The
change in ¯uorous solvent was made so as to allow direct
3.4. Reactions omitting toluene
Our studies on the phase behaviour of 1-octene and nonanal
with per¯uoro-1,3-dimethylcyclohexane in the presence and
absence of toluene showed that 1-octene is totally miscible
with the ¯uorous solvent at $608C under 20 bar CO/H2
21:l), but that nonanal is virtually immiscible. This led us
to consider whether there might be advantages in omitting
the toluene from the system, the most obvious being that a
separate distillation to remove the toluene from the product
would be unnecessary. It might also be possible to effect the
separation of the product from the ¯uorous phase at higher
temperatures, thus reducing processing costs associated
with heating and cooling the various components of the
mixture. The higher polarity of the aldehyde relative to
that of the toluene/aldehyde mixture and the relatively
smaller organic phase might also reduce the extent of the
interaction of the phosphite ligand with the aldehyde while
also reducing the level of leaching of rhodium into the
product. We, therefore, carried out hydroformylation
reactions of 1-octene in a system in which the only
solvent was the per¯uoro-1,3-dimethylcyclohexane
2Table 3). There have been three reports of ¯uorous biphasic
reactions being carried out in the absence of an organic
solvent.20±22
comparison of P24-C6H4C6F13)3 with the alkyl phosphine,
Â
8
P2CH2CH2C6F13)3, used by Horvath, without the added
complication of possible differences in the properties of
solvent. The change does, however, complicate direct
comparison of P24-C6H4C6F13)3 with P2O±4-C6H4C6F13)3.
The results for P24-C6H4C6F13)3 are collected in Table 4
together with, for comparative purposes, three examples
of toluene reactions using PPh3 as the ligand.
Under dilute conditions, [Rh] at 2.0 mmol dm23 and a phos-
phine/Rh ratio of 3:1, these reactions show the zero order
2saturation) kinetics expected for low concentrations of
phosphine.8 The l/b ratios are modest at ca. 3.5, but the
much lower levels of isomerised products, only slightly
higher than those obtained when using PPh3 in toluene
2Table 4), mean that the selectivity to the linear aldehyde
is almost as good as for the best reactions using the phos-
phite. Importantly, there is much less effect on the product
distribution from raising the temperature and both the
rhodium 20.5±2.5%) and phosphorus 26.2±9.2%) leaching
are low and vary little with the reaction temperature. This
suggests that the phosphine is stable under these conditions
although some increased leaching occurs for the reaction
carried out at 908C. Reducing the rhodium concentration
to 1 mmol dm23, whilst maintaining a phosphine/Rh ratio
Reactions were carried out under identical conditions,

D. F. Foster et al. / Tetrahedron 58 ꢀ2002) 3901±3910
3909
Table 5. Comparison of results obtained from the hydroformylation of 1-octene in per¯uoromethylcyclohexane using the rhodium catalyst containing P24-
C6H4C6F13)3, a literature precedent and with current industrial processes
b
This worka
UCCc
BASFc
Â
Horvath
[Rh] 2mmol dm23
)
2.0
20
1274
2 0
70
0.156
42.7
1012
10
100
2.7
286
±
15±18
90±95
2±4
1.8
150
[P] 2mmol dm23
)
[alkene] 2mmol dm23
Pressure 2bar)
Temperature 28C)
)
±
d
d
16
110
1.0
500
5.3
Rate 2mol dm23 h21
)
8.8
0.16
TOF 2h21
l/b
)
4400
6.3
3.7
80.9
0.08
3.3
1020
4.5
7.8
74.8
0.12
740
8.8
e
e
Isomerisation 2%)
Linear aldehyde 2%)
±
89.5
±
83.0
Rh loss 2mg Rh 2mol aldehyde)21
Ploss 2%)
)
±
±
±
±
f
f
f
f
Signi®cant
a
1-Octene as substrate, P24-C6H4C6F13)3 in per¯uoromethylcyclohexane.
Ref. 8, 1-decene as substrate, Rh/P2CH2CH2C6F13)3 in 50:50 vol% of toluene and per¯uoromethylcyclohexane.
Taken from Ref. 1, propene as substrate, Rh/PPh3 in condensed aldehydes. UCC process as run by Union Carbide, BASF process run by BASF.
Propene introduced in gas phase.
Propene cannot isomerise.
Data unavailable.
b
c
d
e
f
of 3:1, does not greatly affect the reaction 2both product
distribution and initial TOF are virtually identical), although
the reaction time is increased. However, increasing the
phosphine concentration to 20 mmol dm23 with the [Rh]
at 2 mmol dm23 210:1 ratio) considerably improves the l/b
ratio and increases the selectivity to linear aldehyde
280.9%). The reactions in the presence of this excess
phosphine become ®rst order, presumably because the
phosphine competes with the alkene for the rhodium and
some of it is thus trapped as trisphosphine species, but the
rate is still very acceptable 2see later) and, more importantly,
the rhodium leaching is dramatically reduced. In the
reaction carried out at 708C, 99.95% of the rhodium is
retained in the ¯uorous phase, together with 96.7% of the
ligand. The absolute amounts of phosphine in the organic
phase should be determined by the partition coef®cient
between the two phases and the total concentration of the
phosphine. The fact that this does not increase dramatically
when the phosphine concentration is increased by a factor of
3 may suggest that the solubility limit of the phosphine in
the aldehyde is being approached.
3.6. Comparisons with other systems
It is instructive to compare the performance of the catalysts
that we have been developing with those used commercially
and reported in the literature. P2OPh)3 and other phosphites
have been examined for their utility in rhodium catalysed
hydroformylation reactions, and indeed bulky phosphites
are amongst the best ligands known, with a bidentate
version having recently been commercialised by Union
Carbide.16 Generally, P2OPh)3 modi®ed catalysts give
higher rates under comparable conditions than those
containing PPh3, but the l/b selectivities are usually modest
2,2:l),16 unless the reactions are carried out under very low
CO/H2 pressures.17 Isomerisation is usually more signi®cant
with phosphites than with phosphines and we have
con®rmed that this is indeed the case under the conditions
we have been using 2Tables 2±4). We have recently shown
that high l/b ratios can be obtained when using P2OPh)3 as a
ligand for rhodium catalysed hydroformylation in ionic
liquids23 or in scCO2, but in the latter solvent this is
probably because the phosphite is insoluble and the catalysis
takes place in the solution of the rhodium complex in molten
P2OPh)3, where the local [Rh] and [P] are very high.24 We
have subsequently used this reaction to develop continuous
¯ow processes in ionic liquid±supercritical ¯uid biphasic
systems.23 Our results with P2O±4-C6H4C6F13)3 show
that this ligand, under ¯uorous biphasic conditions and
especially when no additional organic solvent is added,
gives an increased selectivity for the linear aldehyde relative
to conventional phosphites in organic solvents.
These results show that the omission of toluene from the
system allows the development of a process which is near-
ing the rigorous retention of rhodium in the system that
would be required for commercial application, whilst retain-
ing a high rate and a good selectivity to the linear aldehyde
product. Further re®nements of the ligand design should
allow a truly commercially viable system to be demon-
strated.
GCFID and GCMS analysis of the ¯uorous phases from the
later experiments 210 phosphine/1 rhodium) revealed, that
ca. 10% of the total organic product remains dissolved in the
¯uorous phase after phase separation. The relative propor-
tion of each product in the ¯uorous layer is similar to that of
the phase separated organic product and has been ignored in
the data collected in the tables. The still dissolved organic
product, including unreacted octene, could simply be
recirculated in a continuous ¯ow process where it would
remain at an equilibrium concentration.
For PPh3, which is currently used commercially for the
hydroformylation of propene and other short chain alkenes,
isomerisation is not usually a major problem and l/b ratios
can be high leading to aldehyde selectivities in the 80±90%
range.1 Table 5 presents a comparison between the best
results we have obtained 2in terms of linear selectivity and
rhodium retention into the ¯uorous phase) with those that
are obtained using two variants of the commercial process
for hydroformylation of propene. These show that, with
comparable [Rh] but only 10% of the phosphine loading

3910
D. F. Foster et al. / Tetrahedron 58 ꢀ2002) 3901±3910
2important because of the expense of the ¯uorous ligand)
and at a signi®cantly lower temperature, our system is some
®ve to eight times as productive as the commercial system,
due to a combination of the use of a ¯uorous modi®ed ligand
of lower basicity combined with the much lower phosphine
concentration employed, while giving a linear aldehyde
selectivity that is only slightly less at 80.9% compared
with 83% for the BASF process; the UCC process is
superior in this regard 289.5%). It should be remembered,
however, that propene cannot isomerise, so no loss of
substrate to isomerised alkenes can occur in the commercial
process. The reaction rate is, of course, related to the
[alkene] for these ®rst order reactions. The values quoted
References
1. Frohling, C. D.; Kohlpaintner, C. W. Applied Homogeneous
Catalysis with Organometallic Compounds; Cornils, B.,
Herrmann, W. A., Eds.; VCH: Weinheim, 1996; Vol. 1,
pp. 27±104.
2. Rhodium Catalysed Hydroformylation, van Leeuwen,
P. W. N. M., Claver, C., Eds.; Kluwer: Dordrecht, 2000.
3. Cornils, B. In New Synthesis with Carbon Monoxide; Falbe, J.,
Ed.; Springer: Berlin, 1980.
4. Shell International Research. Neth. Appl., 6,604,094, 1966.
5. Cornils, B. Applied Homogenous Catalysis with Organo-
metallic Compounds; Cornils, B., Herrmann, W. A., Eds.;
VCH: Weinheim, 1996; Vol. 1, pp. 577±600.
Â
in Table 5, for both our work and that of Horvath, are those
at an alkene loading corresponding to 20% by volume. This
would not be an unusual loading at which to operate a
continuous ¯ow reactor.
6. Chem. Engng. News, 1999, 2Oct 18) , 20.
Â
Â
8. Horvath, I. T.; Kiss, G.; Cook, R. A.; Bond, J. E.; Stevens,
Â
7. Horvath, I. T.; Rabai, J. Science 1994, 266, 72±75.
Â
P. A.; Rabai, J.; Mozeleski, E. J. J. Am. Chem. Soc. 1998, 120,
3133±3143.
Also included in Table 5 is a comparison of the best results
Â
reported in this paper with those obtained by Horvath under
9. Chen, W.; Xu, L.; Xiao, J. Chem. Commun. 2000, 839±840.
10. MacDougall, J. K.; Simpson, M. C.; Green, M. J.; Cole-
Hamilton, D. J. J. Chem. Soc., Dalton Trans. 1996, 1161±
1172.
the conditions where he was measuring catalyst leaching.8
His initial [Rh] is much lower than ours, but his [P] is much
higher and his [substrate] lower. Our l/b ratios, linear
selectivities and reaction rates even expressed as turnover
frequencies are much higher, whilst our rhodium leaching is
comparable. He did not measure the phosphine loss,
although notes that change in rate and l/b on recycling
suggests that some ligand leaching is occurring. The
dramatic changes observed seem to indicate that this
could be of the order of 3% per reaction. This is similar to
the level of phosphine leaching that we observe 23.3%).
11. Foster, D. F.; Adams, D. J.; Gudmunsen, D.; Stuart, A. M.;
Hope, E. G.; Cole-Hamilton, D. J. Chem. Commun. 2002,
722±723.
12. 2a) Cole-Hamilton, D. J.; Foster, D. F.; Gudmunsen, D.; Hope,
E. G.; Stuart, A. M. World Patent, WO 00/33956, 2000.
2b) Cole-Hamilton, D. J.; Foster, D. F.; Adams, D. J.;
Gudmunsen, D.; Hope, E. G.; Stuart, A. M. GB Patent
Application, 0126261.7, 2001.
13. 2a) Bhattacharyya, P.; Gudmunsen, D.; Hope, E. G.; Kemmitt,
R. D. W.; Paige, D. R.; Stuart, A. M. J. Chem. Soc., Perkin
Trans. 1 1997, 3609±3612. 2b) Adams, D. J.; Gudmunsen, D.;
Fawcett, J.; Hope, E. G.; Stuart, A. M. Tetrahedron, 2002, 58,
3827±3834.
4. Conclusions
We conclude that by using a system that consists of a
¯uorous solvent in which long chain alkenes are soluble
and a triarylphosphite or triarylphosphine carrying ¯uorous
ponytails on all three aryl groups, hydroformylation can be
successfully carried out in such a way that the product
aldehyde can be obtained in a substantially pure state as a
separate phase at the end of the reaction. Good l/b
selectivities and high rates are obtained even at low rhodium
loadings and with low P/Rh ratios 23±10). The best results,
which compare very well with those obtained using
commercial systems are obtained using Rh
14. Rankin, J.; Benyei, A. C.; Poole, A. D.; Cole-Hamilton, D. J.
J. Chem. Soc., Dalton Trans. 1999, 3771±3782.
15. Payne, M. J.; Cole-Hamilton, D. J. J. Chem. Soc., Dalton
Trans. 1997, 3167±3175.
16. Kamer, P. J. C.; Reek, J. H. N.; van Leeuwen, P. W. N. M. In
Rhodium Catalyzed Hydroformylation; van Leeuwen,
P. W. N. M., Claver, C., Eds.; Kluwer: Dordrecht, 2000;
pp. 35±62.
17. Trzeciak, A. M.; Ziolkowski, J. J. J. Mol. Catal. 1986, 34,
213±219.
18. van Leeuwen, P. W. N. M. In Rhodium Catalyzed Hydro-
formylation; van Leeuwen, P. W. N. M., Claver, C., Eds.;
Kluwer: Dordrecht, 2000; pp. 233±251.
22.0 mmol dm23)/P24-C6H4C6F13)3 220.0 mmol dm23
) at
708C and 20 bar CO/H2. This gives a linear aldehyde
selectivity of 80.9% and an initial productivity of
8.8 mol dm23 h2l. Rhodium leaching into the aldehyde
phase is limited to 0.05% [0.08 mg Rh 2mol aldehyde)2l]
of the rhodium charged and phosphorus leaching to 3.3%.
19. Billig, E.; Abatjoglou, A. G.; Bryant, D. R.; Murray, R. E.;
Maher, J. M. US Patent, 4,717,775, 1988.
20. de Wolf, E.; Speets, E. A.; Deelman, B. J.; van Koten, G.
Organometallics 2001, 20, 3686±3690.
Â
21. Juliette, J. J. J.; Horvath, I. T.; Gladysz, J. A. Angew. Chem.,
Int. Ed. Engl. 1997, 36, 1610±1612.
22. Xiang, J.; Toyoshima, S.; Orita, A.; Otera, J. Angew. Chem.,
Int. Ed. Engl. 2001, 40, 3670±3672.
Acknowledgements
23. Sellin, M. F.; Webb, P. B.; Cole-Hamilton, D. J. Chem.
Commun. 2001, 781±782.
We thank the Royal Society for Personal Fellowships 2E. G.
H. and A. M. S.), the EPSRC for a Postdoctoral Fellowship
2D. J. A.) and funding to use the Catalyst Evaluation and
Optimisation Service 2CATS). We are indebted to Ed
McCurdy, Agilent Technologies UK Ltd for carrying out
the ICPMS analyses.
24. Sellin, M. F.; Cole-Hamilton, D. J. J. Chem. Soc., Dalton
Trans. 2000, 11, 1681±1683.