Investigation of the hydroformylation of ethylene in liquid carbon dioxide

Article abstract of DOI:10.1016/S0022-328X(02)01217-2

In situ, high-pressure NMR was used to investigate the hydroformylation reaction of ethylene in liquid CO₂ using Rh(CO)₂acac as the catalyst precursor and (p-CF₃C₆H₄)₃P, tris(p-trifluoromethylphenyl)phosphine, as the ligand under different thermodynamic conditions (T = 10-23 C, P_CO = P_H2 = 10-15 bar, P_C2H4 = 10-15 bar, P_CO₂ = 207 bar). ¹H-NMR was used to monitor the reaction progress of the hydroformylation of ethylene with a rhodium catalyst under select conditions of temperature and CO₂ solvent pressure. Potential resting states of the rhodium catalyst were investigated by ³¹P(¹H)-NMR. This is the first description of a rhodium catalyzed hydroformylation reaction in liquid CO₂ monitored in situ by high pressure NMR.

Full text of DOI:10.1016/S0022-328X(02)01217-2

Journal of Organometallic Chemistry 650 (2002) 249Á257
/
www.elsevier.com/locate/jorganchem
Investigation of the hydroformylation of ethylene in liquid carbon
dioxide
Clement R. Yonker *, John C. Linehan
Fundamental Science Division, Pacific Northwest National Laboratory, Richland, WA 99352, USA
Received 16 July 2001; received in revised form 28 January 2002; accepted 29 January 2002
Abstract
In situ, high-pressure NMR was used to investigate the hydroformylation reaction of ethylene in liquid CO2 using Rh(CO)2acac
as the catalyst precursor and (p-CF3C6H4)3P, tris(p-trifluoromethylphenyl)phosphine, as the ligand under different thermodynamic
conditions (Tꢀ
/
10Á
/
23 8C, PCO
ꢀ
/
PH
ꢀ
/
10Á
/
15 bar, PC H
ꢀ
/
10Á
/
15 bar, PCO
ꢀ
207 bar). 1H-NMR was used to monitor the
/
reaction progress of the hydroformylation of ethylene w2ith a rhodium catalys2t under select conditions of temperature and CO2
solvent pressure. Potential resting states of the rhodium catalyst were investigated by 31P{1H}-NMR. This is the first description of a
rhodium catalyzed hydroformylation reaction in liquid CO2 monitored in situ by high pressure NMR. # 2002 Elsevier Science B.V.
All rights reserved.
2
4
Keywords: High pressure NMR; Carbon dioxide; Hydroformylation; Rhodium; Ethylene
1. Introduction
ro-substituted phosphines which have enhanced solubi-
lity in CO2. These hydroformylation investigations in
supercritical CO2 have all involved the use of a liquid
alkene substrate for the experimental studies. This limits
Supercritical fluids, as solvents for homogeneous
catalytic reactions, have gained more attention over
the past 10 years. Numerous groups have focused on the
use of supercritical CO2 as a solvent for hydrogenation,
the reaction temperature range to ꢀ31 8C and above
/
for a single fluid homogeneous phase. The current
experimental effort involves the use of gaseous ethylene
dissolved in CO2, which allows the direct determination
of lower temperature reaction rates and the species
present in liquid CO2. Another factor limiting experi-
mental study of supercritical fluid reaction systems
involves the use of high-pressure spectroscopic techni-
ques. Most experimental studies of hydroformylation
reactions in both liquid and supercritical fluid solvents
have involved the use of high-pressure FTIR spectro-
scopy, due to the high sensitivity of the technique. NMR
can give quite detailed molecular information regarding
the structure of reactants, products and intermediates,
but this can be restricted by the need for concentrations
[1Á
/
5] carbonÃ
/
carbon bond formation, [6,7] hydrofor-
mylation, [8Á
/
17] and olefin metathesis [18]. Supercritical
CO2 has some advantages over standard solvents, which
include the ability to control solvent effects through
changes in pressure and temperature, and high diffusiv-
ity/low viscosity. In addition, the high solubility of
gases, in particular H2, make supercritical CO2 an
attractive solvent. CO2 has also been touted as an
environmentally benign solvent for separations, extrac-
tions and chemical synthesis. A disadvantage of using
CO2 as a reaction solvent is the limited solubility of
ligands and transition metalÁphosphine complexes.
/
Leitner et al. [13,14] and Palo and Erkey, [17] among
others, have investigated the catalytic behavior of
rhodium catalysts synthesized using ‘CO2-philic’ fluo-
in the range of 10Á
pressure NMR studies of hydroformylation reactions
have been limited. Rathke et al. [8Á10] have focused on
/100 mM. Therefore, in situ high-
/
cobalt-carbonyl catalysts, studying the linear to
branched ratio of the product aldehyde under super-
critical conditions using their toroid coil high-pressure
NMR cell. Leitner et al. [19] have used a single-crystal
* Corresponding author. Tel.: ꢁ1-509-372-4748; fax: ꢁ1-509-375-
6660.
E-mail address: clem.yonker@pnl.gov (C.R. Yonker).
0022-328X/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 0 2 2 - 3 2 8 X ( 0 2 ) 0 1 2 1 7 - 2

250
C.R. Yonker, J.C. Linehan / Journal of Organometallic Chemistry 650 (2002) 249Á257
/
sapphire NMR vessel to investigate the hydroformyla-
tion of hexene in supercritical CO2. Other experimental
efforts in supercritical fluids have used NMR as an off-
line monitor for batch reaction processes, [14,17] or used
high-pressure NMR to study hydroformylation reac-
tubing for connection to the SS316 gas manifold system
connected to an ISCO pump.
2.1.1. Safety considerations
tions in liquid hydrocarbon solvents [20Á24].
/
This cell design was hydrostatically tested at room
temperature (r.t.) to failure at 1000 bar and beyond with
the time for failure being determined. From this
information the mean time between failure at 500 bar
could be determined. However, unlike metal vessels, it is
impossible to specify a maximum working pressure or
time limit for the PEEK NMR cell. As with all high-
pressure experiments, laboratory personnel should not
be directly exposed to the pressurized vessel. Use of the
PEEK NMR cell should follow established protocols for
length of time exposed to pressure and temperature, and
determining the radial dimension for any sign of
polymer stress.
The experimental effort reported here details the
hydroformylation reaction of ethylene using the (p-
CF3C6H4)3P ligand with the rhodium catalyst precursor
Rh(CO)2acac in liquid CO2 at 10 and 23 8C. The
subsequent reactions of HRh(CO)((p-CF3C6H4)3P)3
during hydroformylation were investigated by high-
pressure NMR in a stepwise manner, first C2H4 (12.5
bar) was mixed with CO2 and secondly, a small amount
of C2H4 (3 bar) in an excess of H2ÁCO was mixed with
/
CO2 in an attempt to characterize the complexes formed
during the reaction. Further investigations of the
fundamental physicochemical behavior of Rh catalysts
and hydroformylation reactions in sub- and supercritical
fluids are needed for the development of commercial
processes based on these solvents.
2.2. General procedure for the rhodium catalyzed
hydroformylation of ethylene in liquid CO2
2. Experimental
In a typical experiment, the PEEK NMR cell was
charged with 4.9 mg (19 mmoles) of Rh(CO)2acac and
The gases H2ÁCO (1:1), ethylene (99.99%), and CO2
/
26.5 mg (57 mmoles) of (p-CF3C6H4)3P for a ꢀ
ratio of ligand to Rh and then evacuated. The maximum
concentration in the PEEK NMR cell was ꢀ32 mM for
the rhodium catalyst. Next, ꢀ25 bar of H2ÁCO (1:1)
/3:1 mole
(SFC Grade) were purchased from Scott Specialty Gases
and used without further purification. The compounds
Rh(CO)2acac and (p-CF3C6H4)3P were purchased from
Strem and used as received. All spectra were acquired on
a Varian (VXR-300) 300 MHz pulsed NMR spectro-
meter with a 7.04 T superconducting magnet and
/
/
/
(300 mmoles of each) was added to the PEEK cell,
followed by adjusting the total pressure to 207 bar with
CO2. This solution diffusionally mixed and reacted at
1
chemical shifts (d) are reported in ppm. The H-NMR
spectra are externally referenced to Me4Si and the 31P-
NMR spectra to phosphoric acid. Pressure was mea-
50 8C for ꢀ60 min. Following this time period the
/
sample was cooled, depressurized, and evacuated over
night to remove any volatile side products (e.g. acet-
ylacetone). The PEEK NMR cell containing the Rh
catalyst was then placed in the 10 mm NMR probe in
sured using a strain gauge with a precision of 9
/
10 psi.
Temperature was controlled to 90.1 K using the
/
nitrogen gas bath controller on the NMR spectrometer.
the magnet, pressurized to 20 bar with H2ÁCO (1:1),
/
followed by ethylene to a total pressure of 30 bar.
Therefore, the partial pressures of the three gases were
approximately equal for this set of conditions. The total
pressure in the NMR cell was adjusted to a final
pressure of 207 bar with CO2. The hydroformylation
reaction kinetics were followed by 1H-NMR spectra
recorded at 10 or 15 min intervals at 10 and 23 8C. The
relative number of moles of ethylene, propanal, and H2
were determined from the integration of the respective
NMR signals. Concurrently, 31P{1H}-NMR spectra
were recorded to determine which phosphorus-contain-
ing species were present during the reaction. After the
hydroformylation reaction was completed, the cell could
be evacuated and recharged with the reaction gases.
Resting states of the rhodium complex were determined
by charging the PEEK NMR cell with only ethylene
2.1. Construction of the PEEK NMR cell
The polyether ether ketone (PEEK) NMR cell is
based on a similar design published by Wallen et al. [25]
(Fig. 1). The current PEEK cell design uses a cone and
taper high-pressure sealing surface within the cap
instead of an o-ring seal. The cap is made from carbon
fiber reinforced PEEK and the cell body from PEEK.
The dimensions of the cell are 3 mm ID, 10 mm OD, and
ꢀ6.5 cm long. Connection to the cap is with a PEEK
/
finger-tight fitting, using PEEK 1/16 in. OD by.01 in. ID
Fig. 1. Schematic of PEEK NMR cell.
mixed into CO2 or a large excess of H2ÁCO in CO2.
/

C.R. Yonker, J.C. Linehan / Journal of Organometallic Chemistry 650 (2002) 249Á
/257
251
3. Results and discussion
The investigation of the rhodium catalyzed hydro-
formylation reaction in liquid CO2 allows the further
investigation of CO2 as a reaction medium for catalytic
processes. NMR facilitates the investigation of reaction
intermediates and can be used to follow the course of
reaction as a function of pressure and temperature. The
rhodium catalyst precursor, Rh(CO)2acac, and the
phosphine ligand, (p-CF3C6H4)3P, were chosen for
study due to their solubility in CO2 and the ease of
forming the rhodium catalyst in situ in the PEEK NMR
cell. In this manner the catalyst system can be synthe-
sized in the PEEK cell and either kept under vacuum or
dissolved in liquid CO2 until initiating the hydroformy-
lation reaction.
Fig. 2. (A) 31P{1H}-NMR spectrum, upon removal of the syngas,
after reaction of Rh(CO)2acac, (p-CF3C6H4)3P, and syngas at 10.0 8C
and 223 bar in CO2, (B) 31P{1H}-NMR spectrum of the reaction
between Rh(CO)2acac and (p-CF3C6H4)3P with 10 bar COÁH2 in CO2
/
at 22.8 8C and 223 bar total pressure, and (C) 31P{1H}-NMR
spectrum of the reaction between Rh(CO)2acac and (p-CF3C6H4)3P
Scheme 1 shows the reaction steps generally accepted
for the hydroformylation reaction of alkenes in liquids
and supercritical fluids using a rhodium catalyst. The
with 10 bar COÁH2 in CO2 at 40.0 8C and 223 bar total pressure.
/
reaction intermediates IIIÁVII have not been identified
/
volatile products, then the precipitated solid was re-
dissolved in CO2 at 10.0 8C. The only phosphorus-
containing species observed under these conditions in
the CO2 solution was HRh(CO)((p-CF3C6H4)3P)3. As
expected for a 3:1 PR3:Rh mole ratio, there was no free
phosphine ligand present in this solution. The minor
experimentally [26]. Not included in Scheme 1, for
clarity, are the different stereoisomers of the rhodium
species. The 1H-NMR spectrum of the ligand, (p-
CF3C6H4)3P, shows two nonequivalent sets of protons
on the benzene ring. The 31P{1H}-NMR spectrum of the
ligand, (p-CF3C6H4)3P, dissolved in pure CO2 shows a
peak at ꢀ/24.0 ppm is due to a small amount of the
sharp singlet at ꢀ
/
ꢂ5.0 ppm.
/
phoshine oxide, (p-CF3C6H4)3PÄ
/
O [27]. Fig. 2(B), is the
Fig. 2(A) shows the 31P{1H}-NMR spectrum of
HRh(CO)((p-CF3C6H4)3P)3, prepared in situ, dissolved
in CO2 at 10.0 8C and 223 bar. The catalyst formation
reaction was completed with 20 bar of syngas in CO2 for
31P{1H}-NMR spectrum of both HRh(CO)((p-
CF3C6H4)3P)3 and HRh(CO)2((p-CF3C6H4)3P)2 formed
in situ in the PEEK NMR cell from a mixture of the
ligand, Rh(CO)2acac and H2ÁCO dissolved in CO2 at
/
ꢀ
/
1 h at 50 8C in the PEEK NMR cell. After this time,
22.8 8C and 223 bar total pressure (3:1 PR3:Rh mole
the NMR cell was evacuated for ꢀ24 h to remove any
/
ratio). Fig. 2(C) is the 31P{1H}-NMR spectrum of both
HRh(CO)((p-CF3C6H4)3P)3
CF3C6H4)3P)2 formed in situ in the PEEK NMR cell
from a mixture of the ligand, Rh(CO)2acac and H2ÁCO
and
HRh(CO)2((p-
/
dissolved in CO2 at 40.0 8C and 223 bar total pressure
(3:1 PR3:Rh mole ratio). The free phosphine, (p-
CF3C6H4)3P, can also be seen in Fig. 2(B) and (C)
over a chemical shift region of d ꢀ
/
ꢂ
/
3.0 to ꢂ5.5 at
/
these temperatures, consistent with loss of phosphine
from the HRh(CO)((p-CF3C6H4)3P)3 to form
HRh(CO)2((p-CF3C6H4)3P)2. The broadening in the
31P signals in Fig. 2(C) as the temperature increases is
due to chemical exchange, similar to that reported by
Horvath et al. [28] for the water soluble HRh(CO)[P(m-
C6H4SO3Na)3]3 system.
Bianchini et al. [20] reports for HRh(CO)(PPh3)3 a
doublet centered at d 41.3 (toluene-d8) in the 31P{1H}-
NMR spectrum with a one-bond rhodiumÁ
ous coupling constant of 153.9 Hz. On pressurizing the
solution with H2ÁCO at room temperature, the authors
report the formation of several products which includes
the carbonylation of HRh(CO)(PPh3)3 to
HRh(CO)2(PPh3)2. The 31P{1H}-NMR spectrum (tolu-
/
phosphor-
/
Scheme 1. Accepted mechanism for the hydroformylation of ethylene
catalyzed by a Rh(I) complex [26].

252
C.R. Yonker, J.C. Linehan / Journal of Organometallic Chemistry 650 (2002) 249Á
/257
1
ene-d8) has a signal at d 37.8 (d, JPÃRh
which is assigned as the HRh(CO)2(PPh3)2 complex, and
a signal at d ꢂ4.8 (s), which corresponds to free PPh3.
For HRh(CO)((p-CF3C6H4)3P)3, Palo and Erkey [17]
report a doublet centered at d 40.2 (1JPÃRh
156 Hz)
and a signal for the free ligand at d ꢂ
6.3 in the 31P{1H}-
ꢀ
/
138.7 Hz),
our NMR investigations to examine this discrepancy.
We found that a yellow precipitate formed, and the FT-
IR (Nujol) of the solid is consistent with a Rh(0) dimer,
of the form Rh2(m-CO)2(CO)2ꢁx((p-CF3C6H4)3P)4ꢂx
/
ꢀ
/
(xꢀ0, 1, or 2). Visual inspection for the Rh(0) dimer in
/
/
the PEEK NMR cell was not possible.
After the in situ formation of the rhodium catalyst,
the high pressure NMR cell was evacuated and the solid
precipitate was re-dissolved in CO2 at 10 8C and 223
bar. In CO2, with no syngas present, the sole rhodium
NMR spectrum (CDCl3). Based on the NMR data
described above in toluene-d8 and CDCl3 for the
HRh(CO)2(PPh3)2 and HRh(CO)((p-CF3C6H4)3P)3
complexes, the doublet at d 40.5 (1JPÃRh
ꢀ
/
156 Hz) in
Fig. 2 is assigned as HRh(CO)((p-CF3C6H4)3P)3 and the
doublet at d 37.6 (1JPÃRh
145 Hz) in Fig. 2(B) and (C)
containing
species
should
be
HRh(CO)((p-
1
ꢀ
/
CF3C6H4)3P)3. The H-NMR spectrum of the hydride
region of HRh(CO)((p-CF3C6H4)3P)3 is shown in Fig. 3.
The hydride proton of HRh(CO)((p-CF3C6H4)3P)3
is assigned as HRh(CO)2((p-CF3C6H4)3P)2. Table 1 lists
the 31P{1H}-NMR chemical shift values and the one-
bond phosphorusÁ
complexes and other structures shown in Scheme 1
observed in situ in liquid CO2.
Our observations are in contrast to Horvath’s work in
toluene-d8 and Leitner’s work in scCO2 [19,28]. We see
/
rhodium coupling constants for these
appears as a quartet, which is centered at d ꢂ10.2
/
1
(see Table 1). The H spectrum shown in Fig. 3 was
obtained under the same conditions as Fig. 2(A), with
no syngas present in the sample. The one-bond
rhodiumÁ
solved. The two-bond phosphorousÁ
constant of ꢀ14 Hz is consistent with a hydride in the
/
proton coupling constant could not be re-
both
species,
HRh(CO)((p-CF3C6H4)3P)3
and
/
proton coupling
HRh(CO)2((p-CF3C6H4)3P)2 present in CO2 with syn-
gas, whereas Horvath only reports HRh(CO)2((p-
CF3C6H4)3P)2 after HRh(CO)((p-CF3C6H4)3P)3 is ex-
posed to syngas and Leitner reports only HRh(CO)2(3-
H2F6-(R,S)-BINAPHOS)2 in scCO2. In fact, ours is the
first observation of a significant amount of the mono-
/
apical position cis to the three equatorial phosphines in
a trigonal bipyramidal geometry [29].
In the CO2 solution with syngas present, the signal in
1
the H-NMR for the hydrides corresponding to struc-
tures I and II (see Scheme 1) observed at 10 8C and
23 8C were extremely broad due to chemical exchange.
This made it difficult to directly identify the hydride of
carbonyl
rhodium
monomer,
HRh(CO)((p-
CF3C6H4)3P)3 observed in situ under syngas. We believe
these differences may be due to solvent effects. In all
previous cases, the in situ spectroscopic study was
performed in the presence of a hydrocarbon solvent
while CO2 appears to yield a different ratio of the
organometallic products. In addition, Leitner’s observa-
tions were carried out in the presence of a small amount
of THF and with a bidendate ligand in which the
chelating binding mode is the most stable. We do not
observe in the NMR spectra the Rh(0) dimers at these
1
HRh(CO)2((p-CF3C6H4)3P)2 from the H spectrum.
Fig. 4(A) shows the 1H-NMR spectrum of
HRh(CO)((p-CF3C6H4)3P)3 after exposure to 12.5 bar
of ethylene in 207 bar CO2 at 0 8C. The ethylene (d ꢀ
5.5) and ligand (d 7Á8) signals are not shown for clarity.
The two signals at chemical shifts of 2 ppm and ꢀ1.4
ppm are unbound acetylacetone from the Rh precursor
molecule. The signal at ꢀ0.9 ppm in Fig. 4(A) could be
/
/
/
/
due to ethylene bound to rhodium, such as in structure
IV in Scheme 1. The solubility of the proposed ethylene
complex, HRh(CO)((p-CF3C6H4)3P)2(h2-C2H4), ap-
pears to be greater than HRh(CO)((p-CF3C6H4)3P)3 in
pure CO2 at 0 8C. The extreme upfield shift (0.9 ppm)
for the signal corresponding to the bound ethylene of
HRh(CO)((p-CF3C6H4)3P)2(h2-C2H4) is similar to our
temperatures
(10 8C
and
23 8C),
Rh2(m-
0, 1, or 2),
CO)2(CO)2ꢁx((p-CF3C6H4)3P)4ꢂx (xꢀ
/
reported by Bianchini et al. [20] at ꢂ20 8C with the
/
PPh3 ligand using 31P{1H}-NMR, and by others using
high pressure IR in hydrocarbon solvents. We con-
ducted in situ IR studies under identical conditions to
Table 1
Properties of the different rhodium complexes with the (p-CF3C6H4)3P ligand as determined by high-pressure NMR
Complex
structure
Free ligand 31P{1H}-NMR
(ppm)
Rh complexes 31P{1H}-NMR
(ppm, d)
1JPÃRh (Hz)
Rh complexes, hydride signal 1H-NMR
(ppm)
a
I
ꢂ5.06
ꢂ6.3
ꢂ5.06
40.5
40.2
37.6
36.5
32.0
ꢀ156
156
ꢂ10.2
ꢂ9.9
b
I
II
IV
ꢀ145
ꢀ197
ꢀ82
VIII
a
Structures are shown in Scheme 1.
From ref [17] in CDCl3.
b

C.R. Yonker, J.C. Linehan / Journal of Organometallic Chemistry 650 (2002) 249Á
/257
253
Fig. 3. 1H-NMR spectrum, upon removal of the syngas, after reaction of Rh(CO)2acac, (p-CF3C6H4)3P, and syngas at 10.0 8C and 223 bar in CO2.
earlier high-pressure NMR photolysis work using
MeCpMn(CO)2(h2-C2H4) and CpRe(CO)2(h2-C2H4) in
investigations are necessary to confirm the structure of
the species assigned as the ethylene complex. In a similar
manner, the 31P{1H}-NMR spectrum of the proposed
HRh(CO)((p-CF3C6H4)3P)2(h2-C2H4) product shows a
supercritical ethylene and CO2Á
/C2H4 mixtures [30,31].
In these cases, the h2-C2H4 was reported at ꢀ
/
2.5 ppm
for MeCpMn(CO)2(h2-C2H4) in ethylene at 0 8C and
274 bar, and at 2.07 ppm for the photolysis product
different one-bond phosphorusÁrhodium coupling con-
/
stant compared with HRh(CO)((p-CF3C6H4)3P)3. Fig. 5
shows the in situ 31P{1H}-NMR spectrum of the
reaction mixture under the same conditions as shown
1
CpRe(CO)2(h2-C2H4) in liquid C6D6. The H chemical
shifts are density (pressure) dependent. We have ob-
served pressure dependent 1H chemical shift changes
greater than 1 ppm [30]. Therefore, the assignment of
in Fig. 4. The doublet at ꢀ
/
36.5 ppm (1JPÃRh
ꢀ197 Hz)
/
is assigned to the HRh(CO)((p-CF3C6H4)3P)2(h2-C2H4)
complex (see Table 1). The doublet centered at 40.5 ppm
the peak at ꢀ
/
0.9 ppm to the h2-C2H4 adduct is
plausible. The molecular symmetry change on the
addition of the ethylene to the rhodium complex was
observed in the hydride splitting of HRh(CO)((p-
CF3C6H4)3P)2(h2-C2H4). The quartet shown in Fig. 3
for HRh(CO)((p-CF3C6H4)3P)3 becomes a multiplet of
either a triplet or a doublet of doublets. Further
(1JPÃRh
CF3C6H4)3P)3.
ꢀ155 Hz) corresponds to HRh(CO)((p-
/
The acyl complexes (structures VII and VIII in
Scheme 1) were investigated by exposing HRh(CO)((p-
CF3C6H4)3P)3 to an excess amount of H2ÁCO (1:1) in
/
the presence of a small partial pressure of ethylene. The
1
1
Fig. 4. (A) H-NMR spectrum of HRh(CO)((p-CF3C6H4)3P)3 and ethylene (12.5 bar) in CO2 at 207 bar and 0 8C and (B) H-NMR spectrum of
HRh(CO)((p-CF3C6H4)3P)3 in CO2 at 207 bar and 0 8C, before exposure to ethylene. The peaks labeled by * are due to the enol form of
acetylacetone from the Rh precursor molecule.

254
C.R. Yonker, J.C. Linehan / Journal of Organometallic Chemistry 650 (2002) 249Á257
/
Fig. 5. 31P{1H}-NMR spectrum of a mixture containing HRh(CO)((p-CF3C6H4)3P)2(h2-C2H4) and HRh(CO)((p-CF3C6H4)3P)3 in CO2 at 0 8C
and 207 bar. This spectrum was obtained after exposure of HRh(CO)((p-CF3C6H4)3P)3 to ꢀ10 bar of ethylene.
resulting 31P{1H}-NMR spectrum of such an experi-
ment is shown in Fig. 6. In this experiment,
HRh(CO)((p-CF3C6H4)3P)3 was exposed to 3 bar of
ture of this degradation product remains to be eluci-
dated. It is unlikely that the degradation product is the
phosphine oxide, which appears at ꢀ21 ppm in C6D6 in
/
ethylene and ꢀ
/
138 bar of H2Á/CO (1:1) in a total final
the 31P{1H}-NMR spectrum. The signal at d 31.5 is
actually a doublet with part of the signal being obscured
by the much larger signal at 32.1 ppm [33]. The one-
bond phosphorus-rhodium coupling constant for this
pressure of 207 bar in CO2. The H2Á/CO (1:1) gas
mixture acted as an anti-solvent for the Rh complex
causing precipitation. Therefore, the sample mixture in
the PEEK NMR cell was evacuated to ꢀ/20 bar and
doublet is ꢀ
/
8294 Hz. In the hydroformylation of 1-
/
then re-pressurized with CO2 to a final pressure of 207
bar. In this manner, the Rh complexes demonstrate
enough solubility to be observed by 31P{1H}-NMR.
In Fig. 6(A), the doublet centered at d 40 is due to
hexene in toluene-d8 reported by Bianchini et al., [20] a
signal in the 31P{1H}-NMR spectrum at 27.5 ppm (d,
1JPÃRh
ꢀ
/
78.0 Hz) was assigned as the acylÁRh complex
/
Rh(CO(CH2)5CH3)(CO)2(PPh3)2. Considering the dif-
ferent phosphorus ligand and solvent system for the
experiment shown in Fig. 6, and the similar one-bond
HRh(CO)((p-CF3C6H4)3P)3 (1JPÃRh
ꢀ155 Hz). The
/
large singlet signal at d 32.1 is postulated to be a
degradation product of the (p-CF3C6H4)3P as it no
longer shows coupling to rhodium. Various mechanisms
of catalyst deactivation through the degradation of the
phosphorus ligand have been proposed [32]. The struc-
phosphorusÁrhodium coupling constant, the signal at
/
1
31.5 ppm (d, JPÃRh
due to the acylÁRh complex shown as structure VIII in
Scheme 1 (see Table 1). The PEEK NMR cell containing
ꢀ82 Hz) in Fig. 6(A) is most likely
/
/
Fig. 6. (A) 31P{1H}-NMR spectrum of the acylÁ
/
Rh complex after exposure to 3 bar ethylene, ꢀ138 bar H2 ÁCO (1:1), and 66 bar CO2 for 10 min,
/
then depressurized and repressurized to 207bar CO2 at 50 8C. (B) 31P{1H}-NMR spectrum of the remaining products from the reaction described
above after evacuation for 12 h (in CO2 at 207 bar and 50 8C).

C.R. Yonker, J.C. Linehan / Journal of Organometallic Chemistry 650 (2002) 249Á
/257
255
the Rh complex mixture was depressurized and evac-
uated for ꢀ12 h. On re-pressurizing with CO2 to a
pressure of 207 bar, the spectrum in Fig. 6(B) was
obtained, in which the proposed acylÁRh complex is no
For the two temperatures of 10 and 23 8C, the peak
areas were normalized by the number of protons giving
rise to the resonance signal. The normalized peak area,
which is proportional to the number of moles, can then
be plotted for the hydroformylation reaction as a
function of time. These data are plotted in Fig. 8, for
the conditions of 3:1 ligand:Rh mole ratio and approxi-
mately equal pressures for the three gases (ethylene, H2
/
/
longer observed in the system (because of the residual
1
acetylacetone interference, the H-NMR was not useful
for spectral interpretation of the acyl complex).
The potential application of high-pressure NMR to
the hydroformylation reaction of ethylene with a Rh
catalyst in CO2 not only lies with the identification of
potential intermediate complexes, but the kinetics of the
reaction process can be directly followed through time-
resolved NMR acquisition during the reaction. In the
temperature range of 10 and 23 8C, the hydroformyla-
tion kinetics are slow enough such that the product
formation can be followed by recording the NMR
spectra. The equation governing the global reaction
rate could be determined from a detailed series of
experiments covering a range of substrate and reactant
concentrations as published previously for CO2 in a
and CO) of ꢀ12.5 bar at the two temperatures. At both
/
10 and 23 8C, the propanal formation is linear as a
function of time. This relationship allows the turn-over
frequency for the formation of the aldehyde ([mol
aldehyde][mol Rh]ꢂ1 hꢂ1) to be determined for these
two temperatures over a large reaction time. These
observed frequencies are ꢀ
/
5 [mol aldehyde][mol Rh]ꢂ1
hꢂ1 and ꢀ34 [mol aldehyde][mol Rh]ꢂ1 hꢂ1 at 10 and
/
23 8C, respectively. The catalyst species present during
the hydroformylation reaction were monitored by
31P{1H}-NMR, only structures I and II were seen in
equilibrium in solution with a small amount of oxidized
ligand present at longer times.
batch reaction process [15Á17]. The experiments de-
/
tailed in the following section are a subset of a wider
range of experimental conditions proposed for investi-
gation at a future date. The rate of the hydroformyla-
tion reaction of ethylene in CO2 with a rhodium catalyst
was determined by measuring the peak areas of the
reactants (ethylene and H2) and the product (propanal)
4. Conclusions
In situ high-pressure NMR investigations of homo-
genous catalytic reactions in liquid and supercritical
CO2 are useful in identifying reaction intermediates and
determining observed reaction rates. The investigation
of the rhodium catalyzed hydroformylation reaction of
ethylene in liquid CO2 demonstrates the advantages of
1
in the H-NMR spectrum as a function of time.
A typical experimental result is shown in Fig. 7. The
spectrum is a snapshot of the hydroformylation reaction
140 min after the initiation of the reaction. The
experimental mole ratio of ligand:Rh is 3:1, the mole
such
a spectroscopic investigation. The proposed
HRhCO((p-CF3C6H4)3P)2(h2-C2H4) intermediate was
identified directly in situ for the first time. This
preliminary set of experiments demonstrates that high-
pressure NMR can be used to observe in situ the
reaction in supercritical CO2 solvent between highly
ratio of C2H4:Rh is ꢀ
/
18:1, with 20 bar of H2ÁCO (1:1),
/
and an ethylene pressure of 10 bar in CO2 at a total final
pressure of 207 bar and 10 8C. As can be seen in Fig. 7,
the different species in solution are identified and the
area of the signals can be determined.
Fig. 7. 1H-NMR spectrum during the hydroformylation reaction of ethylene after 140 min. Reaction conditions are 3:1 ligand:Rh mole ratio, 20 bar
H2 ÁCO (1:1), 10 bar ethylene, final pressure 207 bar CO2 at 10 8C. The peaks labeled by * are due to acetylacetone from the Rh precursor molecule.
/

256
C.R. Yonker, J.C. Linehan / Journal of Organometallic Chemistry 650 (2002) 249Á257
/
1
Fig. 8. Normalized peak area versus reaction time for the H-NMR spectra during the hydroformylation reaction. Ethylene (k, I), H2
and Propanal (m, j) are shown at 10 8C (circle symbols, solid lines) and 23 8C (square symbols, dashed lines).
[6] D.K. Morita, D.R. Pesiri, S.A. David, W.H. Glaze, W. Tumas,
soluble gaseous reactants and an organometallic cata-
lyst. This technique is useful for determining reaction
intermediates and describing the turn-over frequency of
Chem. Commun. (1998) 1397.
[7] M.A. Carroll, A.B. Holmes, Chem. Commun. (1998) 1395.
[8] J.W. Rathke, R.J. Klinger, United States Patent, 5,198,589,1993.
[9] J.W. Rathke, R.J. Klinger, T.R. Krause, Organometallics 11
(1992) 585.
reactions in sub- and supercritical fluids. A more
detailed investigation of the hydroformylation system
reported here is in progress. It is expected that as
supercritical fluids extend themselves into the realm of
reaction solvents for large-scale or commercial pro-
cesses, further effort will be needed to identify the
fundamental reaction mechanism under such solvent
conditions.
[10] J.W. Rathke, R.J. Klinger, T.R. Krause, Organometallics 10
(1991) 1350.
[11] Y. Guo, A. Akgerman, Ind. Eng. Chem. Res. 36 (1997) 4581.
[12] Y. Guo, A. Akgerman, J. Supercrit. Fluids 15 (1999) 63.
[13] S. Kainz, D. Koch, W. Baumann, W. Leitner, Angew. Chem. Int.
Ed. Engl. 36 (1997) 1628.
[14] D. Koch, W. Leitner, J. Am. Chem. Soc. 120 (1998) 13398.
[15] D.R. Palo, C. Erkey, Ind. Eng. Chem. Res. 37 (1998) 4203.
[16] D.R. Palo, C. Erkey, Ind. Eng. Chem. Res. 38 (1999) 3786.
[17] D.R. Palo, C. Erkey, Organometallics 19 (2000) 81.
[18] A. Furstner, D. Koch, K. Langemann, W. Leitner, C. Six, Angew.
¨
Acknowledgements
Chem. Int. Ed. Engl. 36 (1997) 2466.
[19] G. Francio, K. Wittmann, W. Leitner, J. Organomet. Chem. 621
(2001) 130.
The authors would like to thank Dr. A. Getty for her
helpful comments regarding this manuscript. Work at
the Pacific Northwest National Laboratory (PNNL)
was supported by the Office of Science, Office of Basic
Energy Sciences, Chemical Sciences Division of the US
Department of Energy, under Contract DE-AC076RLO
1830.
[20] C. Bianchini, H.M. Lee, A. Meli, F. Vizza, Organometallics 19
(2000) 849.
[21] C. Bianchini, P. Barbaro, G. Scapacci, F. Zanobini, Organome-
tallics 19 (2000) 2450.
[22] S.C. van der Slot, P.C.J. Kamer, P.W.N.M. van Leeuwen, J.A.
Iggo, B.T. Heaton, Organometallics 20 (2001) 430.
[23] F.R. Bergman, J.M. Ernsting, F. Muller, M.D.K. Boele, L.A. van
¨
der Veen, C.J. Elesevier, J. Organomet. Chem. 592 (1999) 306.
[24] A. Castellanos-Pa´ez, S. Castillo´n, C. Claver, P.W.N.M. van
Leeuwen, W.G.J. de Lange, Organometallics 17 (1998) 2543.
[25] S.L. Wallen, L.K. Schoenbachler, E.D. Dawson, M.A. Blatch-
ford, Anal. Chem. 72 (2000) 4230.
References
[26] P.W.N.M. van Leeuwen, C.P. Casey, G.T. Whiteker, in:
P.W.N.M. van Leuwen, C. Claver (Eds.), Rhodium Catalyzed
Hydroformylation Chapter 4, Kluwer Academic Publishers,
Dordrecht, 2000, p. 64.
[1] P.G. Jessop, Y. Hsiao, T. Ikariya, R. Noyori, J. Am. Chem. Soc.
118 (1996) 344.
[2] P.G. Jessop, T. Ikariya, R. Noyori, Organometallics 14 (1995)
1510.
[27] The (p-CF3C6H4)3P phosphine was oxidized by the reaction
with excess t-butylperoxide in benzene. Gentle heating was
necessary to complete the reaction. The product was observed
by 31P-NMR.
[3] P.G. Jessop, T. Ikariya, R. Noyori, Science 269 (1995) 1065.
[4] M.J. Burk, S. Feng, M.F. Gross, W. Tumas, J. Am. Chem. Soc.
117 (1995) 8278.
[5] P.G. Jessop, T. Ikariya, R. Noyori, Nature 368 (1994) 231.

C.R. Yonker, J.C. Linehan / Journal of Organometallic Chemistry 650 (2002) 249Á
/257
257
[28] I.T. Horvath, R.V. Kastrup, A.A. Oswald, E.J. Mozeleski, Catal.
Lett. 2 (1989) 85.
[31] J.C. Linehan, C.R. Yonker, J.T. Bays, S.T. Autrey, T.E.
Bitterwolf, S. Gallagher, J. Am. Chem. Soc. 120 (1998) 5826.
[29] L.A. van der Veen, M.D.K. Boele, F.R. Bregman, P.C.J. Kramer,
P.W.N.M. van Leeuwen, K. Goubitz, J. Fraanje, H. Schenk, C.
Bo, J. Am. Chem. Soc. 120 (1998) 11616.
[32] A.M. Trzeciak, J.J. Zio´ lkowski, Coord. Chem. Rev. 190Á
/
192
(1999) 883.
[33] On expansion of this region of the 31P{1H}-NMR spectrum, the
second peak of the doublet can be discerned on the shoulder of the
larger peak at 32.1 ppm.
[30] J.C. Linehan, S.L. Wallen, C.R. Yonker, T.E. Bitterwolf, J.T.
Bays, J. Am. Chem. Soc. 119 (1997) 10170.