Hydrocarboxylation of olefins using an amphiphilic palladium catalyst, activity and recycling properties. NMR identification of some reaction intermediates

Article abstract of DOI:10.1016/j.molcata.2006.06.036

The high solubility in acidic solutions of N-bis(N′,N′-diethyl-2-aminoethyl)-4-aminomethylphenyl-diphen ylphosphine (N3P) make it a suitable candidate for study and comparison to the more commonly studied trisulfonated triphenylphosphine (TPPTS) ligand in the palladium catalysed aqueous hydrocarboxylation reaction. The catalyst employing N3P shows an inverted regioselectivity compared to the TPPTS system. Non-coordinating anions give the best results in terms of activity and stability of the catalyst. Due to N3P amphiphilic character and contrary to sulfonated phosphines reaction, it is possible to recycle the catalyst, both by extracting the substrate and by extracting the catalyst into an organic solvent. The hydrocarboxylation of styrene, 1-octene and 4-penteneoic acid demonstrates that the reaction rate is strongly dependent on the solubility of the substrates. Using the water-soluble 3-buten-1-ol as substrate, two palladium zerovalent complexes, two palladium hydrides, one acyl and one alkyl complexes were identified by means of NMR and IR.

Full text of DOI:10.1016/j.molcata.2006.06.036

Journal of Molecular Catalysis A: Chemical 259 (2006) 231–237
Hydrocarboxylation of olefins using an amphiphilic palladium catalyst,
activity and recycling properties
NMR identification of some reaction intermediates
Magnus Karlsson, Adriana Ionescu∗, Carlaxel Andersson
Department of Organic Chemistry, Chemical Centre, Lund University, P.O. Box 124, S-22100 Lund, Sweden
Received 16 December 2005; received in revised form 23 May 2006; accepted 7 June 2006
Available online 31 July 2006
Abstract
The high solubility in acidic solutions of N-bis(Nꢀ,Nꢀ-diethyl-2-aminoethyl)-4-aminomethylphenyl-diphenylphosphine (N3P) make it a suitable
candidate for study and comparison to the more commonly studied trisulfonated triphenylphosphine (TPPTS) ligand in the palladium catalysed
aqueous hydrocarboxylation reaction. The catalyst employing N3P shows an inverted regioselectivity compared to the TPPTS system. Non-
coordinating anions give the best results in terms of activity and stability of the catalyst. Due to N3P amphiphilic character and contrary to
sulfonated phosphines reaction, it is possible to recycle the catalyst, both by extracting the substrate and by extracting the catalyst into an organic
solvent. The hydrocarboxylation of styrene, 1-octene and 4-penteneoic acid demonstrates that the reaction rate is strongly dependent on the
solubility of the substrates. Using the water-soluble 3-buten-1-ol as substrate, two palladium zerovalent complexes, two palladium hydrides, one
acyl and one alkyl complexes were identified by means of NMR and IR.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Amphiphilic phosphine; Aqueous phase; Hydrocarboxylation; Palladium; Recycling
1. Introduction
We have previously published a report on the synthesis and
use of the amphiphilic phosphine N3P, N-bis(Nꢀ,Nꢀ-diethyl-2-
aminoethyl)-4-aminomethylphenyl-diphenylphosphine (Fig. 2)
[6]. The acidic conditions required in the hydrocarboxylation
reaction and the high solubility of N3P and its complexes in
acidic aqueous solutions make N3P a suitable candidate for
study and comparison to the more commonly studied trisul-
fonated triphenylphosphine (TPPTS) ligand in the palladium
catalysed aqueous hydrocarboxylation reaction. Due to its
amphiphilic character, and contrary to sulfonated phosphines,
it is possible to extract excess N3P and its complexes into
an organic phase. This enables biphasic separation since
the acid product is highly water-soluble. Compared to the
most commonly used ligand for this reaction, trisulfonated
triphenylphosphine (TPPTS), N3P is sterically less demanding
and more electron rich, factors which might influence both the
activity and the regioselectivity of the catalyst.
The synthesis of carboxylic acids from olefins, carbon
monoxide and water using a palladium catalyst, i.e., the hydro-
carboxylation reaction (Fig. 1), has gained considerable atten-
tion during the last years, and the field has been reviewed several
times during the last years [1]. The recent interest is due partly to
the finding that the reaction can be conducted in neat water or in
aqueous biphasic systems, employing sulfonated phosphines as
ligands [2–4] and the use of a water-soluble catalyst is advanta-
geous since water is one of the reactants. Water-soluble catalysts
also facilitate separation of the product from the reaction mix-
ture, as well as catalyst recovery.
The hydrocarboxylation reaction is used industrially in the
production of the analgesic and anti-inflammatory drug ibupro-
fen, (2-(4-isobutylphenyl) propionic acid), where the reaction is
one of the steps in the synthesis [5].
In this paper, results from palladium catalysed hydrocarboxy-
lation of styrene, 1-octene and 4-penteneoic acid, employing
N3P as the ligand are presented. In addition, the recycling prop-
erties of the catalyst system and NMR identification of some
∗
Corresponding author. Tel.: +46 46 222 03 09; fax: +46 46 222 44 39.
E-mail address: adriana.ionescu@inorg.lu.se (A. Ionescu).
1381-1169/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2006.06.036

232
M. Karlsson et al. / Journal of Molecular Catalysis A: Chemical 259 (2006) 231–237
until complete dissolution. The solution was then transferred
to the autoclave and the substrate added. In the experiments
using styrene as the reacting olefin, p-tert-butyl catechol (6 mg,
0.036 mmol) was added as polymerisation inhibitor. The auto-
clave was closed and pressurised (25 bar) and vented three times
before the heating and stirring was started. After having reached
the set temperature the pressure was adjusted to the set value.
The autoclave was cooled to room temperature and vented after
reaction completion. The reaction mixture was then transferred
to a Schlenk tube and extracted with 4 × 3 ml diethyl ether. The
autoclave was rinsed with 2 × 3 ml diethyl ether. The composi-
tion of the combined organic phase was analysed by GLC. In
the recycling experiments the aqueous phase from the initial run
containing the catalyst was transferred to the autoclave and the
pH was adjusted to the selected value, followed by addition of
fresh substrate and polymerisation inhibitor before starting the
reaction according to the above procedure.
Fig. 1. The hydrocarboxylation reaction.
The hydrocarboxylation reaction with 4-penteneoic acid fol-
lowed the above protocol but the work-up was different. After
completed reaction, the reaction mixture was transferred to a
Schlenk tube and cooled in an ice bath, after which the pH
was set to ≈12 by slow addition of aqueous NaOH. The result-
ing turbid solution was extracted with 3 × 3 ml toluene, leaving
a colourless aqueous phase containing the sodium salt of the
carboxylic acid products and unreacted 4-penteneoic acid. The
water was removed under vacuum and the residue was exam-
Fig. 2. N3P.
key intermediates in the catalytic cycle, using the water-soluble
3-buten-1-ol as substrate, are also described.
2. Experimental
2.1. General procedures
1
ined by H NMR to determine yield and regioselectivity. The
All chemicals of the highest grade purity were purchased
from commercial sources and used as received, except N3P [6]
and PdCl2(PhCN)2 [7], which were synthesised according to
literature procedures. All solvents used in the manipulation of
N3P and the catalysts were repeatedly degassed prior to use.
Standard Schlenk technique was used in the manipulation of the
catalysts and isolated compounds.
combined toluene phases, containing catalyst and excess ligand,
were extracted with 3 × 5 ml water containing methanesulfonic
acid. The pH in the final combined aqueous phase was adjusted
to 1.8 using additional amounts of methanesulfonic acid, and
then transferred to the autoclave and the reaction started again
following the above procedure.
N3P·(HCl)x was prepared by dissolving N3P in diethyl ether,
and bubbling HCl (g) through the solution until precipitation of
the salt occurred. Filtration afforded a white solid, which on
analysis showed to contain varying amounts of HCl.
The NMR spectra were recorded at 21 ◦C, on a Varian
Unity 300 MHz spectrometer with an observation frequency of
121 MHz for 31P, using 85% H3PO4 as external standard, and
of 300 MHz for 1H. The high pressure NMR experiments were
performed using a sapphire tube assembly [8].
The IR spectra were recorded on a Nicolette FT-IR spectrom-
eter as Nujol mulls.
The hydrocarboxylation reactions were performed in a 50 ml
Autoclave Engineers HastaloyTM autoclave equipped with a
sampling valve and a temperature and stirring speed control
device.
2.3. Attempts to identify and isolate reaction intermediates
using 3-buten-1-ol as the reacting alkene
2.3.1. Reaction sequence 1
PdCl2(PhCN)2 and N3P·(HCl)x (P/Pd = 10) were charged
into an autoclave and dissolved in D2O (CPd = 10 mM) and the
resulting solution was then subjected to CO (g) at 50 bar and
50 ◦C for 15 min. After depressurisation, two equivalents of 3-
butene-1-ol were added under CO. The immediate addition of
THF provided a yellow precipitate which was analysed by NMR
and IR. The stability of the precipitated complex was monitored
by observing the peaks in 31P NMR at 80 ◦C for 1 h after which
the organic products were identified by IR.
2.3.2. Reaction sequence 2
PdCl2(PhCN)2 and N3P·(HCl)x (P/Pd = 5) were charge into
a Schlenk tube and dissolved in D2O (CPd = 5 mM). CO (g) was
bubbled through the resulting solution at room temperature for
15 min and then two equivalents of 3-butene-1-ol were added
while keeping the reaction mixture under a CO atmosphere. An
aliquot of the solution was transferred to a HP NMR-tube and
monitored by observing the 31P spectrum first at 1 bar then at
50 bar CO. A 4 h after start of the experiments THF was added to
2.2. Hydrocarboxylation reactions
In a typical experiment involving styrene or 1-octene, N3P
(70 mg, 0.14 mmol) and two equivalents of acid were dissolved
in 15 ml degassed water. Additional acid was added to set the
pH to 1.8, after which the palladium complex PdCl2(PhCN)2
(6 mg, 0.016 mmol) was added and the mixture was stirred

M. Karlsson et al. / Journal of Molecular Catalysis A: Chemical 259 (2006) 231–237
233
the remaining solution in the Schlenk tube, furnishing a yellow
precipitate. The precipitate was separated and analysed by NMR
and IR. The stability of the precipitated solid was monitored by
31P NMR at 80 ◦C over a period of 1 h after which the organic
products were identified by GC.
plex 1 is reduced under CO to a mixture of zerovalent complexes
viz. Pd(N3P)3 2 and Pd(N3P)4 3, where complex 3 is predom-
inant. The broad line-widths observed, compared to the nar-
row line-width (5.5 Hz) reported for Pd(TPPTS)3·9H2O and the
observation of two separate peaks is probably a reflection of the
smaller cone angle of N3P relative that of TPPTS, which enables
the formation of the tetra-coordinated complex 3 and ligand
interchange between the two complexes. The peaks get sharper
and shift a little up-field (23.6 and 21.9 ppm) at 0 ◦C (Fig. 3a),
but the line-width is still broad (179 Hz). The same spectrum
taken under different conditions (P/Pd = 5, CPd = 5 mM, 1 bar
CO, see Section 2.3.2) exhibits only one very broad resonance
at 22.2 ppm (Fig. 3b) corresponding to the complexes 2 and 3
shifted closer together and in comparable amounts. This demon-
strates that the CO pressure is not an important factor in the
reduction step and that the ligand concentration controls the
equilibrium (Eq. (1)) between the two zerovalent complexes 2
and 3:
3. Results and discussion
3.1. NMR studies—identification of intermediates
The gross mechanistic features of the hydrocarboxylation
reaction in the case of the TPPTS-based catalyst system have
been verified by NMR identification of the key intermediates [9].
Initially, we were interested in whether the same catalytic cycle
also applies for the N3P based catalyst, i.e., if similar intermedi-
ates could be identified. The precursor complex PdCl2(PhCN)2
dissolved smoothly in a D2O solution of the hydrochloride salt of
N3P·(HCl)x (P/Pd = 10, see Section 2.3.1) to furnish a solution
with showed three resonances in its {H} 31P NMR spectrum: one
singlet resonance at −4.5 ppm, straightforwardly assignable to
the free ligand, a triplet at 34.2 ppm, and a doublet at 30.2 ppm
(2JP–P 13.7 Hz). Theshiftsandcouplingconstantofthelattertwo
resonances correspond well to those observed for cationic com-
plexes of the general composition [PdCl(L)3]+ Cl−, 1, which has
been shown to form under similar conditions using PdCl2 and
triphenylphosphine (TPP) [10] or TPPTS [11]. The room tem-
perature spectrum run after subjecting the solution to CO (g)
at 50 bar pressure and 50 ◦C for 15 min showed no peaks corre-
sponding to complex 1 instead it exhibited two broad resonances
centred at 22.7 ppm and 24.8 ppm. The up-field shift agrees with
that of the TPPTS complex Pd(TPPTS)3·9H2O [12], hence com-
Pd(N3P)3 + N3P ↔ Pd(N3P)4
(1)
We have not been able to verify the formation of a hydride
1
by either H or 31P NMR at the pH at which dissolution of
N3P·(HCl)x brings about. However, when the CO promoted
reduction was carried out as described earlier [9], i.e., using
a 60% (v/v) aqueous trifluoroacetic acid as the solvent, proto-
nation of complexes 2 and 3 occurred virtually quantitatively,
yielding two different hydrides which are stable enough to be
1
observed by H and 31P NMR. Based on the observed NMR
data, one of these hydride complexes is structurally identical to
that earlier observed in the TPPTS based system, i.e., a planar
or planaroid cationic palladium species with three phosphorus
Fig. 3. 31P {H} NMR spectra of the zerovalent complexes Pd(N3P)3 2 and Pd(N3P)4 3: (a) P/Pd = 10; 0 ◦C; (b) P/Pd = 5; 25 ◦C.

234
M. Karlsson et al. / Journal of Molecular Catalysis A: Chemical 259 (2006) 231–237
atoms and one hydride, with a peak at −8.5 ppm consisting of
a doublet of triplets (2JH–P(cis) 13.5 Hz (trip), 2JH–P(trans) 175 Hz
The yellow solid isolated after letting complex 4 be con-
verted to complex 5 showed neither no peak in the range of
CO stretching vibrations nor did the GC analysis revealed any
carbonylated product (acid or lactone) after thermal decompo-
sition. The only organic product observed was butyraldehyde
which is a result of isomerization of 3-buten-1-ol. All our exper-
imental evidences point to that complex 4 is an acyl complex,
possibly the iso-acyl 4a, since the hydrocarboxylation product is
␣-methyl-␥-valerolactone. 4a is the kinetically controlled com-
plex which slowly converts to the thermodynamically controlled
alkylcomplex5. Theratherhighstabilityofcomplex 5isastrong
indication that it is a ring-closed complex with the oxygen atom
of the alkene coordinating to the central palladium atom (Fig. 4).
1
(d)). In the { H} 31P NMR spectrum, this resonance corresponds
to a triplet at 19.7 ppm and a doublet at 26.3 ppm with 2JP–P of
37 Hz. Without decoupling these resonances were observed as
one doublet of triplets with a 2JH–P(trans) coupling of 175 Hz and
a broad resonance where the expected coupling of 13.5 Hz was
lost in the line width.
The second hydride gave a pentet at −7.8 ppm with a 2JH–P
of 39 Hz, consistent with a coupling between the hydride and
1
four magnetically equivalent phosphorus atoms. In the { H} 31P
NMR spectrum this hydride gave a singlet at 21.5 ppm, which
in the decoupled mode transformed into a doublet with 2JH–P of
39 Hz. These NMR observations are consistent with a dimeric
specieswithahydrideandacarbonmonoxidemoleculeasbridg-
ing ligands and with a tetrahedral coordination around the two
palladium atoms, [Pd2(␮-H)(␮-CO)(N3P)4]+. A dimer of sim-
ilar constitution is known from similar experiments employing
triphenylphosphine as the ligand [13], but was reported not to
appear using TPPTS [9].
3.2. Catalytic screening experiments
Initially we tried to establish suitable conditions (palladium
precursor, Bro¨nstedt acid, temperature and molar ratio Pd/P) for
carrying out the hydrocarboxylation reaction using N3P as a lig-
and. PdCl2 is commonly used as palladium source in in situ gen-
erationoftheactivecatalyst, butPdCl2(PhCN)2, whichisreadily
available and more reactive, offered certain advantages—the for-
mation of the catalyst precursor [PdCl(N3P)2]+Cl− proceeded
much faster, and PdCl2 often left small amounts of insoluble
material. Since no negative effect of the liberated benzonitrile
was observed, all the experiments were carried out using the
benzonitrile complex.
There have been differing opinions as to the influence of the
anion of the Bro¨nstedt acid on the hydrocarboxylation reaction
and an example, demonstrating that coordinating anions e.g.
Cl−, Br− or I− lower the initial activity but raise the long-term
activity, can be found [15]. Results from other investigations [3]
show that coordinating anions decompose the hydride as a result
of higher stability of cationic versus neutral hydrides [16]. We
have studied four different acids in the hydrocarboxylation of
styrene. From the results (Table 1) of the initial screening of the
N3P based catalyst using these acids, it is clear that Bro¨nstedt
acids with non-coordinating anions give the best results in terms
of conversion (%), and that methanesulfonic acid is the best. The
trend in activity resembles the one found in the carbonylation of
1-(4-isobutylphenyl) ethanol [3], as in both cases organic sul-
fonic acids provide the best yield. With fluoroboric acid, the
solubility of the N3P based catalyst posed a problem. Although
the catalyst employing this acid appeared fully dissolved at tem-
peratures above 70 ◦C, at room temperature the solubility is
considerably lower, making the solution turbid. This restricted
solubility makes catalyst recycling difficult and unreliable.
The solution containing the mixture of the zerovalent com-
plexes 2 and 3 reacted smoothly with 3-butene-1-ol to furnish
a mixture of two new complexes viz. complex 4, with a singlet
resonance at 20.7 ppm and another complex, 5, with a singlet
resonance at 21.1 ppm in the 31P NMR spectrum. The rela-
tive proportion of these two complexes varied over time—the
peak corresponding to complex 4 dominated direct after addition
of the alkene but, in time, the peak corresponding to complex
5 gradually increased in intensity at the expense of the peak
corresponding to complex 4. At a CO pressure of 50 bar the
conversion is complete within 3 h and the inter-conversion can-
not be reversed at even higher CO pressure. The constitution of
these complexes cannot be deduced from the 31P NMR spec-
trum but serendipitously, we found that the complexes can be
precipitated from the aqueous solution by addition of THF or
dioxane, although not in analytically pure form. The solid iso-
lated direct after addition of 3-buten-1-ol proved to be complex 4
with a singlet peak in the 31P spectrum at 20.7 ppm. The IR spec-
trum with a strong vibration at 1677 cm−1, which is in the range
of CO stretching frequencies of previously isolated palladium
acyl complexes [14], indicates that it contains an acyl complex.
The singlet in 31P NMR leaves two alternative structures for
this acyl complex 4, as given in Fig. 4 (4a: iso-acyl; 4b: n-acyl
complexes). By thermal decomposition of the aqueous solution
containing complex 4, only the zerovalent complex 2 is visible in
31P NMR, and the IR analysis showed the presence of ␣-methyl-
␥-butyrolactone (1754 cm−1) and butyraldehyde (1720 cm−1).
Fig. 4. Proposed structures for acyl complex 4 and alkyl complex 5 (X = Cl−, D2O, CO; P = phosphine).

M. Karlsson et al. / Journal of Molecular Catalysis A: Chemical 259 (2006) 231–237
235
Table 1
Influence of acid and olefin in the hydrocarboxylation reaction
Substrate
Acid
Conversion (%)
n/i
TSA
HBF4
HCl
76
38
26
92
3.4
3.7
3.0
2.6
Styrene
MeSO3H
1-Octene
MeSO3H
30
2.4
General reaction conditions: PCO = 50 bar; T = 100 ◦C; precursor: PdCl2-
(PhCN)2 (0.016 mmol); ligand: N3P (0.14 mmol); styrene/Pd: 250; 1-octene/Pd:
100; polymerisation inhibitor for styrene: p-tert-butyl catechol (0.036 mmol);
pH 1.8; solvent: H2O; total volume: 15 ml; reaction time: 5 h.
Fig. 5. Recycling of the catalyst in the hydrocarboxylation of styrene.
The solubility of the catalyst is high in hydrochloric acid. The
substrate conversion in hydrochloric acid, however, is low, as is
the stability of the catalyst, and inevitably formation of various
amounts of palladium black occurred. The low conversion can
probably be attributed to decreased stability of the palladium
hydridespeciesduetothechlorideionspresent. Thelowstability
of the complexes in hydrochloric acid, yielding palladium black
are in contrast to previous observations regarding the TPPTS-
based catalyst [15], for which it was claimed that chloride ions
prevented the catalyst from decomposing.
The rate of product formation is strongly temperature depen-
dent, and the higher the temperature the higher the reaction rate.
However, the thermal stability of the catalyst system sets an
upper limit, and 100 ◦C was found to be the highest temperature
at which the catalyst, employing N3P as ligand, is stable, i. e.
no metal particle formation was detectable at 50 bar CO. Thus,
the conditions used for all the catalytic experiments were set to
100 ◦C and 50 bar CO.
We have noted spectacular colour changes in the recycling
procedure. During the extractions the colour of the aqueous solu-
tion changed from bright yellow to deep purple. This colour
change was accompanied by a release of carbon monoxide from
the over-saturated solution, but the solution regained its original
yellow colour in the following catalytic run. Attempts to employ
31P NMR to monitor which type of species gave rise to the
purple colour were unsuccessful. However, palladium-carbonyl
clusters involving phosphines are known to show strong colours
[17].
Styrene is known to be susceptible to polymerisation under
high temperature conditions, which is why a radical scavenger,
p-tert-butylcatechol, wasused. Withoutthisadditivepolystyrene
was always formed in visible amounts. Even in the experi-
ments employing styrene, reported here, which all included
p-tert-butylcatechol, we cannot exclude the formation of small
amounts of visually undetectable polystyrene.
The phosphine to palladium ratio influences the stability of
the catalyst, and more than stoichiometric amounts of the lig-
and are needed to prevent catalyst decomposition, although in
the ratio range of 5–12, it has only minor effects on the regios-
electivity. A ratio of 9, which was used throughout this study,
provided a stable catalyst.
3.4. Regioselectivity
The linear to branched (n/i) ratio of the phenylpropionic acids
formed was constant in the recycling experiments, but varied
between 2.6 and 3.7 depending upon the acid used in the reac-
tion (Table 1). This might be related to solubility differences.
The highest n/i ratio was found for the least soluble catalysts, so
it is possible that aggregation does occur to some extent, even
at higher temperature, and that this affects the regioselectivity
of the catalyst. More importantly, the observed regioselectiv-
ity is considerably different from that previously observed for
TPPTS based catalyst, which gave a n/i ratio of 0.8 [2]. From
the standpoint of product value, the observed high selectivity
toward the linear isomer is a clear disadvantage—the branched
isomer is the one related to the ibuprofen class of compounds,
and thus the most valuable one. From the standpoint of a mecha-
nistic understanding of the origin of regioselectivity our finding
is, however, of interest. The electronic and steric properties of
N3P are very similar to those of triphenylphosphine (TPP) [6].
It therefore differs from the bulkier and more electron deficient
trisulfonatedtriphenylphosphine(TPPTS), andthesedifferences
are obviously enough to totally change the regiochemistry. Con-
sidering the regio-control, there are two different possibilities;
the steric properties of the spectator ligands most likely affect
3.3. Recycling experiments
In the hydrocarboxylation of styrene, the catalyst employing
methanesulfonic acid could be recycled at least four times with-
out loss of activity. In fact, a slight increase in activity occurred
between the first and the second run (Fig. 5). This increase can
be explained by either a longer activation time of the catalyst
precursor 1 used in the first run, compared to the activation time
of the resting state catalyst in consecutive runs, or that the chlo-
ride ions present in the precursor are lost in the recycling step,
and this leads to a more active catalyst. Although the activity
is retained during consecutive runs, we cannot exclude that cat-
alyst decay is occurring to some extent. The retained activity
then could be apparent due to the low substrate solubility (vide
infra). However, in the recycling experiments, there was no for-
mation of palladium black, nor did we observe any colouring
of the ether extraction phases indicative of the loss of metal
complex.

236
M. Karlsson et al. / Journal of Molecular Catalysis A: Chemical 259 (2006) 231–237
Table 2
the stability of the palladium alkyl complex formed by insertion
of the olefin in the Pd–H bond, bulkier ligands promote for-
mation of the linear alkyl complex. Based on steric arguments
alone, the N3P ligand would give predominantly the branched
isomer. On the other hand, if the insertion of the olefin in the
Pd–H bonds yields two isomeric Pd-alkyls (linear and branched)
which are in rapid equilibrium and which insert CO at differ-
ent rates, there would be mainly electronic rather than steric
control. Unfortunately, since the two spectator ligands N3P and
TPPTS differ with respect to both steric bulk and donor strength,
they do not provide a clear-cut answer as to the cause of the
regio-control.
Hydrocarboxylation of penteneoic acid
Substrate (pentenoic acid)
Conversion (%)
n/i
Run (1)
Run (2)
94
91
3.1
3.1
General reaction conditions: PCO = 50 bar; T = 100 ◦C; precursor: PdCl2-
(PhCN)2 (0.016 mmol); ligand: N3P (0.14 mmol); substrate/Pd: 560; pH 1.8;
acid: MeSO3H; solvent: H2O; total volume: 15 ml; reaction time: 1 h.
of the catalyst into toluene represents a reverse procedure, in
that the aqueous phase contains the product, for which any
appropriate method of work-up can be used to recover the
product.
3.5. Mass-transfer
The results using 4-pentenoic acid, displayed in Table 2,
demonstrate that the reverse recycling procedure can be applied
with only a minor loss in activity. Pentenoic acid reaches 94%
conversion in one hour while styrene needs five hours for 92%
conversion. This, again, is a strong indication that the solubil-
ity of styrene limits its rate of reaction. The linear to branched
ratio of 3.1 is slightly higher than the selectivity in the 1-octene
reaction. Coordination of the carbonyl group on 4-pentenoic
acid to the palladium atom in the alkyl intermediates would
lead to seven or six-membered ring structures for the linear and
branched alkyl complexes, respectively. Based on the higher sta-
bility of a six-membered ring, the branched isomer is expected,
providing carbonyl interaction occurs. The preferential forma-
tion of the linear isomer observed, however, indicates that the
carboxylic group on the substrate plays no role in determining
the regioselectivity.
A slow rate of phase boundary mass transfer frequently
causes problems with the reaction rate in biphasic reaction mix-
tures, and this is also the case when applying catalyst based
on the N3P ligand in the hydrocarboxylation reaction. This is
evident by comparing the conversions for 1-octene and styrene
under the same reaction conditions. Styrene, having a reported
solubility of 1.1 mmol/l in water [2], is much more water-soluble
than 1-octene and has a conversion of 92% while 1-octene of
only 30% (Table 1). Still, it is probably the solubility of styrene
in water which determines its reaction rate. Additionally, we
observed that 1-octene slowly isomerises under the reaction
conditions, and the product mixture contained approximately
10% of 2-ethyl heptanoic acid. The influence of phase bound-
ary mass transfer on the reaction rate also became evident by
comparing reaction rates for both styrene and 1-octene in neat
form or dissolved in toluene—a much higher reaction rate was
found by adding the reactant olefins in undiluted form directly
to the aqueous reaction mixture. Even so, the reaction rate is
most likely limited by the low water solubility of these two
substrates.
Besides the main product, adipic acid, 8% 2-methyl-1,5-
pentanedioic acid was also observed as a result of isomerization
of 4-pentenoic acid.
4. Conclusions
The importance of mass-transfer rates is also nicely
demonstrated in the reaction of 4-penteneoic acid. Initially,
this substrate was chosen to evaluate the possibility of using
a “reverse” recycling procedure to recover and reuse a cat-
alytic system based on an amphiphilic phosphine ligand.
4-Penteneoic acid represents a class of alkenes highly soluble
in water, and its product in the hydrocarboxylation reaction,
adipic acid, is even more water-soluble. This, of course,
makes the separation of catalyst and product, and hence the
catalyst recycling, cumbersome. Although in the case of
penteneoic acid it is possible to separate the substrate and the
product from the catalyst by repeated extractions, this method
poses the risk of introducing air into the system, ultimately
leading to oxidation of phosphine ligands. One can also easily
find substrates/products with solubility so close to that of
the amphiphilic catalysts that separation becomes virtually
impossible. In such a case, the amphiphilic catalyst presents a
clear advantage over the mere water-soluble ones (e.g. catalysts
employing sulfonated phosphines), since an amphiphilic cata-
lyst can be transferred into an organic phase while keeping the
reactants/products in a basified aqueous phase. Compared to the
normal recycling procedure for amphiphilic catalysts, extraction
The hydrocarboxylation of styrene, 1-octene and 4-pente-
neoic acid using a palladium catalyst employing the amphiphilic
phosphine N3P demonstrates that the reaction rate is strongly
dependent on the solubility of the substrates.
Using N3P in the hydrocarboxylation reaction it is possible
to recycle the catalyst, both by extracting the substrate and by
extracting the catalyst into an organic solvent. This is a clear
advantage compared to the more commonly used ligand, trisul-
fonated triphenylphosphine (TPPTS). The catalyst employing
N3P also shows an inverted regioselectivity compared to the
TPPTS system.
A whole range of reaction intermediates, from zerovalent
complexes to alkyl and acyl complexes were identified by means
of NMR.
Acknowledgements
Financial support from TFR (the Swedish Research Council
for Engineering Sciences) and SSF (Swedish Foundation for
Strategic research) is gratefully acknowledged.

M. Karlsson et al. / Journal of Molecular Catalysis A: Chemical 259 (2006) 231–237
237
References
[8] A. Cussanelli, U. Frey, D.T. Richens, A.E. Merbach, J. Am. Chem. Soc.
118 (1996) 5265.
[9] G. Verspui, I.I. Moiseev, R.A. Sheldon, J. Organomet. Chem. 586 (1999)
196.
[10] S. Yamazaki, Polyhedron 4 (1985) 1915.
[11] G. Papadogianakis, J.A. Peters, L. Maat, R.A. Sheldon, J. Chem. Soc.
Chem. Commun. (1995) 1105.
[12] G. Papadogianakis, L. Maat, R.A. Sheldon, Inorg. Synth. 32 (1998) 25;
C. Binkowski, J. Cabou, H. Bricout, F. Hapiot, E. Monflier, J. Mol. Catal.
215 (2004) 23;
[1] F. Bertoux, E. Monflier, Y. Castanet, A. Mortreux, J. Mol. Catal. 143 (1–3)
(1999) 11;
W.A. Hermann, SyntheticMethodsofOrganometallicandInorganicChem-
istry: Catalysis, vol. 10, Thieme, Stuttgart, Germany, 2002, p. 42;
I. del Rio, C. Claver, P.W.N.M. van Leeuwen, Eur. J. Inorg. Chem. (2001)
2719.
[2] F. Bertoux, S. Tilloy, E. Monflier, Y. Castanet, A. Mortreux, J. Mol. Catal.
138 (1) (1999) 53.
F. Monteil, P. Kalck, J. Organomet. Chem. 482 (1994) 45.
[13] V.N. Zudin, V.D. Chinakov, V.A. Nekipelov, V.A. Likholobov, Y.I. Yer-
makov, J. Organomet. Chem. 289 (1985) 425.
[3] G. Verspui, G. Papadogianakis, R.A. Sheldon, Catal. Today 42 (1998) 449;
G. Verspui, J. Feiken, G. Papadogianakis, R.A. Sheldon, J. Mol. Catal. 146
(1–2) (1999) 299.
[14] K. Suzuki, M. Nishida, Bull. Chem. Soc. Jpn. 46 (1973) 2887;
P.E. Carrou, R.F. Heck, J. Am. Chem. Soc. 98 (14) (1976) 4115;
Cavinato, G:, L. Toniolo, A. Vavasori, J. Mol. Catal. 219 (2) (2004) 233.
[15] F. Bertoux, E. Monflier, Y. Castanet, A. Mortreux, J. Mol. Catal. 143 (1–3)
(1999) 23.
[4] M.S. Goedheijt, J.N.H. Reek, P.C.J. Kamer, P.W.N.M. van Leeuwen, Chem.
Commun. (1998) 2431.
[5] V. Elango, K.G. Davenport, M.A. Murphy, G.N. Mott, E.G. Zey, B.L. Smith,
G.L. Moss, European Patent Appl. EP 0,400,892 (1990).
[6] M. Karlsson, M. Johansson, C. Andersson, J. Chem Soc. Dalton Trans.
(1999) 4187.
[16] V.V. Grushin, Chem. Rev. 96 (1996) 2011.
[17] K. Kudo, M. Hidai, Y. Uchida, J. Organomet. Chem. 33 (1971) 393.
[7] J.R. Doyle, P.E. Slade, H.B. Jonassen, Inorg. Synth. 6 (1960) 218.