The rate-determining step in the rhodium-Xantphos-catalysed hydroformylation of 1-octene

Article abstract of DOI:10.1002/chem.200700727

The rate-determining step in the hydroformylation of 1-octene, catalysed by the rhodium-Xantphos catalyst system, was determined by using a combination of experimentally determined ¹H/²H and ¹²C/ ¹³C kinetic isotope effects and a theoretical approach. From the rates of hydroformylation and deuterioformylation, a small ¹H/ ²H isotope effect of 1.2 was determined for the hydride moiety of the rhodium catalyst. ¹²C/¹³C isotope effects of 1.012(1) and 1.012(3) for the α-carbon and β-carbon atoms of 1-octene were determined, respectively. Both quantum mechanics/molecular mechanics (QM/MM) and full quantum mechanics calculations were carried out on the key catalytic steps, for "real-world" ligand systems, to clarify whether alkene coordination or hydride migration is the rate-determining step. Our calculations (21.4kcal mol^-1) quantitatively reproduce the experimental energy barrier for CO dissociation (20.1 kcal mol^-1) starting at the (bisphosphane)RhH(CO)₂ resting state. The barrier for hydride migration lies 3.8 kcal mol^-1 higher than the barrier for CO dissociation (experimentally determined trend ～3 kcal mol^-1). The computed ¹H/²H and ¹²C/¹³C kinetic isotope effects corroborate the results of the energy analysis.

Full text of DOI:10.1002/chem.200700727

FULL PAPER
DOI: 10.1002/chem.200700727
The Rate-Determining Step in the Rhodium–Xantphos-Catalysed
Hydroformylation of 1-Octene
Erik Zuidema,[a] Laura Escorihuela,[b] Tanja Eichelsheim,[a] Jorge J. Carbó,[b]
Carles Bo,[b, c] Paul C. J. Kamer,[a] and Piet W. N. M. van Leeuwen*[a, c]
Abstract: The rate-determining step in
the hydroformylation of 1-octene, cata-
lysed by the rhodium–Xantphos cata-
lyst system, was determined by using a
combination of experimentally deter-
mined 1H/2H and 12C/13C kinetic iso-
tope effects and a theoretical approach.
From the rates of hydroformylation
determined, respectively. Both quan-
tum mechanics/molecular mechanics
(QM/MM) and full quantum mechanics
calculations were carried out on the
key catalytic steps, for “real-world”
ligand systems, to clarify whether
alkene coordination or hydride migra-
tion is the rate-determining step. Our
calculations (21.4 kcalmolꢀ1) quantita-
tively reproduce the experimental
energy barrier for CO dissociation
(20.1 kcalmolꢀ1) starting at the (bi-
sphosphane)RhH(CO)2 resting state.
The barrier for hydride migration lies
3.8 kcalmolꢀ1 higher than the barrier
for CO dissociation (experimentally
determined trend ~3 kcalmolꢀ1). The
computed 1H/2H and 12C/13C kinetic
isotope effects corroborate the results
of the energy analysis.
1
and deuterioformylation, a small H/2H
isotope effect of 1.2 was determined
for the hydride moiety of the rhodium
catalyst. 12C/13C isotope effects of
1.012(1) and 1.012(3) for the a-carbon
and b-carbon atoms of 1-octene were
Keywords:
density
functional
calculations
·
hydroformylation ·
isotope effects · phosphane ligands ·
reaction mechanisms
Introduction
be achieved; all atoms of the reactants are incorporated in
the final product. This atom efficiency is crucial, because the
production of these aldehydes exceeds eight million tons per
annum.[1]
The hydroformylation of alkenes is one of the most impor-
tant homogeneously catalysed reactions in industry. The ad-
dition of carbon monoxide and hydrogen to alkenes, cata-
lysed by homogeneous cobalt, rhodium or platinum cata-
lysts, leads to the clean formation of aldehydes. The alde-
hydes are used in the manufacture of detergent and plasti-
ciser alcohols and other butanal-derived products on a
multi-ton scale. In principle, complete atom efficiency can
For the industrial production of butanal through hydrofor-
mylation of propene, several catalyst systems have been de-
veloped (and patented) that show high activity and high re-
gioselectivity towards the formation of the desired linear al-
dehyde. These catalyst systems are based on rhodium, often
modified by wide-bite-angle bidentate phosphane ligands
based on the highly successful BISBI ligand backbone
(BISBI= 2,2’-bis((diphenylphosphino)methyl)-1,1’-biphen-
yl), the Xanthene family of backbones[2] or related backbone
structures.[3] Bulky wide-bite-angle diphosphite ligands have
also been applied successfully in rhodium-catalysed hydro-
formylation reactions (see Scheme 1). These electron-poor
ligands generate highly active catalyst systems that show
very high regioselectivity towards the formation of linear al-
dehydes, even in the hydroformylation of internal alkenes to
linear aldehydes.[4]
[a] Dr. E. Zuidema, T. Eichelsheim, Prof. Dr. P. C. J. Kamer,
Prof. Dr. P. W. N. M. van Leeuwen
Vanꢀt Hoff Institute for Molecular Sciences
Universiteit van Amsterdam, Nieuwe Achtergracht 166, 1018 WV
Amsterdam, (The Netherlands)
E-mail: pwnm@science.uva.nl
[b] L. Escorihuela, Dr. J. J. Carbó, Dr. C. Bo
Universitat Rovira i Virgili, Campus Sescelades
Marcel.lí Domingo s/n, 43007 Tarragona (Spain)
[c] Dr. C. Bo, Prof. Dr. P. W. N. M. van Leeuwen
The Institute of Chemical Research of Catalonia
Avinguda Països Catalans 16, 43007 Tarragona (Spain)
The generally accepted dissociative catalytic cycle for the
hydroformylation of alkenes catalysed by phosphane-modi-
fied rhodium catalysts (Scheme 2) was first proposed by Wil-
kinson and co-workers.[5] Starting from trigonal-bipyramidal
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
Chem. Eur. J. 2008, 14, 1843 – 1853
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1843

ing on the electronic and steric properties of the ligand, dif-
ferent reaction kinetics have been observed experimentally.
Generally, the observed reaction kinetics are simplified to
fit one of two extreme cases. For catalysts modified by elec-
tron-poor ligands such as bulky monophosphites and the un-
modified rhodium-carbonyl catalyst, the hydrogenolysis of
acyl species 6 controls the overall rate of hydroformyla-
tion.[8,9] Consequently, the observed rate of hydroformyla-
tion for these catalysts systems is independent of the alkene
concentration, the rate shows a negative order in carbon
monoxide concentration and it exhibits a first-order rate de-
pendency on the hydrogen concentration (often referred to
as Type II kinetics). The resting state of these catalysts is
the acyl complex 7, which has been observed spectroscopi-
cally.[8,10] In contrast, the kinetics for electron-rich ligands,
including most bidentate phosphane ligands, are consistent
with a rate-determining step early in the catalytic cycle
(Type I kinetics). For these ligands the resting state of the
catalyst is the penta-coordinate rhodium hydride species 1.
Previously, it has been shown that under the experimental
conditions, the rate of carbon monoxide dissociation from
complex 1 is two orders of magnitude higher than the over-
all rate of hydroformylation.[11,12] Therefore, either the rate
of alkene coordination or the rate of hydride migration gov-
erns the overall rate of hydroformylation, but which one is
actually rate-limiting remains unclear.
Scheme 1. Wide-bite-angle ligands applied in the rhodium-catalysed hy-
droformylation. From left to right: BISBI-based bidentate ligands, the
xanthene family of ligands, diaryl bicyclo
bite-angle diphosphite ligands.
A
Here we present an investigation into the rate-determin-
ing step in the hydroformylation of 1-octene, catalysed by
the rhodium–Xantphos catalyst system. In this study, we
Scheme 2. Dissociative hydroformylation catalytic cycle as proposed by
Wilkinson and co-workers.
1
used a combination of H/2H and 12C/13C isotope effects to
rhodium hydride species 1, coordinatively unsaturated spe-
cies 2 is generated by loss of either one phosphane ligand or
one carbon monoxide ligand. Alkene coordination yields 18-
electron complex 3, which can undergo hydride migration to
form another coordinatively unsaturated species 4. Carbon
monoxide (re)coordination, followed by insertion yields acyl
complex 6. This complex can either react with carbon mon-
oxide yielding complex 7 or it can react with H2, liberating
the aldehyde product and regenerating the rhodium hydride
species 2.
Under standard hydroformylation conditions, most reac-
tion steps discussed above are believed to be reversible with
the exception of the product-forming hydrogenolysis of spe-
cies 6. Furthermore, deuterioformylation experiments by
Lazzaroni and co-workers by using the unmodified rhodi-
um-carbonyl catalyst reveal that in the hydroformylation of
both 1-hexene and styrene by using this catalyst system, the
formation of alkyl complex 4 is essentially irreversible at
room temperature. However, the formation of branched
alkyl complexes becomes reversible at elevated tempera-
tures.[6] Similar studies on the hydroformylation of 1-hexene
by bisphosphane-modified rhodium catalysts by Casey and
co-workers show that also for these electron-rich catalyst
systems, the formation of alkyl complexes 4 is essentially ir-
reversible under mild hydroformylation conditions.[7]
discriminate between alkene coordination and hydride mi-
gration as the rate-limiting step in the hydroformylation of
1-octene. The effect of isotopic substitution on the hydride
moiety of the rhodium catalyst on the rate of hydroformyla-
tion was studied by comparing the relative rates of hydrofor-
mylation and deuterioformylation of 1-octene. Interestingly,
whilst deuterioformylation has been used extensively to
probe the reversibility of the formation of alkyl complex 4,
the relative rates of hydroformylation and deuterioformyla-
tion have not been compared previously. Since the addition
of H2 to complex 6 is not rate-limiting for these wide-bite-
angle bisphosphane catalyst systems (Type I kinetics), the
relative rates of hydroformylation and deuterioformylation
should allow a direct discrimination between a rate-limiting
alkene coordination and hydride migration reaction step in
the hydroformylation of 1-alkenes. For the hydride migra-
tion reaction, which involves transfer of a hydride moiety
from the metal centre to the coordinated alkene ligand, a
large primary kinetic isotope effect would be expected. In
contrast, the hydride moiety is not involved directly in the
coordination of 1-octene to the metal centre. Therefore, if
alkene coordination determines the overall rate of hydrofor-
mylation, the observed kinetic isotope effect would be ex-
pected to be small or absent.
1
To complement the H/2H kinetic isotope effect, 12C/13C
The relative rates of the individual reaction steps are in-
fluenced by the ligands employed in the reaction. Depend-
kinetic isotope effects caused by the isotopic substitution in
the alkene substrate were also investigated. If alkene coordi-
1844
www.chemeurj.org
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 1843 – 1853

Rhodium–Xantphos-Catalysed Hydroformylation of 1-Octene
FULL PAPER
nation to complex 2 is rate-limiting, a primary kinetic iso-
tope effect is expected at both the a-carbon and b-carbon
positions of 1-octene, because both atoms are involved in
the reaction similarly. On the other hand, if the hydride mi-
gration reaction is rate-determining, only one carbon atom
ꢀ
is involved in the formation of the new carbon hydrogen
bond. The other carbon atom is involved in the formation of
ꢀ
a rhodium carbon s-bond, but the vibrational energy of this
bond is much lower than the vibrational energy of the
ꢀ
carbon hydrogen bond. Combined with the high kinetic
preference for the formation of the linear aldehyde product
for the Xantphos catalyst system, this would yield a large
observed kinetic isotope effect at the b-position and a signif-
icantly smaller kinetic isotope effect at the a-position of the
alkene substrate.
Figure 1. Dependence of the rate of hydroformylation (measured as the
initial change in the intensity of the nC of the aldehyde product per
=
O
second per mmol catalyst) on the initial 1-octene concentration
octene]0, determined by high-pressure IR spectroscopy.
A
To explain the experimentally determined kinetic isotope
effects and to unravel the rate-determining step of the reac-
tion, DFT-based calculations on the initial steps of the hy-
droformylation reaction were performed on “real-world”
rhodium–bisphosphane catalyst systems. Based on the com-
bined experimental and the theoretical results, we were able
to determine unambiguously the rate-limiting step in the hy-
droformylation of 1-octene, catalysed by the rhodium–Xant-
phos catalyst system.
Results and Discussion
*
Figure 2. The rates of hydroformylation ( ) and deuterioformylation (&)
1H/2H kinetic isotope effect: We have studied the relative
rates of hydroformylation and deuterioformylation of 1-
octene by using the rhodium–Xantphos catalyst system. The
initial rate of aldehyde formation was determined by high-
pressure infrared spectroscopy. Although this approach does
not yield absolute rates of reaction, it does allow the com-
parison of different rates under identical reaction conditions.
Furthermore, it allows us to study the initial rate of reaction
in detail, revealing any incubation effects at the start of the
reaction. Finally, the resting state of the catalyst can be
monitored during the reaction. A change of the resting state
of the catalyst from, for instance, complex 1 to acyl complex
7 might indicate a change in reaction kinetics between the
hydroformylation and deuterioformylation of 1-octene.
Initially, the dependency of the rate of hydroformylation
on the concentration of 1-octene in the reaction mixture was
determined at 808C, at 20 bar of H2/CO (1:1) and a rhodium
concentration of 1.0 mm. The results are shown in Figure 1.
A clear first-order rate dependence is observed, consistent
with a rate-determining reaction step early in the catalytic
cycle (Type I kinetics). Incubation effects after the addition
of 1-octene were not observed. Subsequently, we studied the
relative rates of hydroformylation and deuterioformylation
at 808C and 20 bar of H2/CO or D2/CO at a rhodium con-
centration of 1.0 mm and an initial 1-octene concentration of
637 mm. An example of the graphs obtained from both the
hydroformylation and deuterioformylation of 1-octene is
shown in Figure 2. Clearly, for both hydroformylation and
deuterioformylation, the reaction starts immediately and no
of 1-octene, monitored by the increase in the nC adsorption of the alde-
hyde product by high-pressure infrared spectroscopy.
=
O
incubation effects are observed. Although the formation of
new carbon deuterium bonds in the aldehyde product
ꢀ
during deuterioformylation partially obscures the absorp-
tions of the carbonyl ligands of the catalyst, no changes in
the resting state were observed upon addition of alkene sub-
strate, neither in the hydroformylation nor in the deuterio-
formylation reaction. The formation of acyl species 7, which
have
a very characteristic infrared adsorption around
1675 cmꢀ1 for phosphane-modified catalyst systems,[13] was
not observed. Furthermore, GC analysis of the final prod-
ucts revealed no large differences between the regioselectiv-
ities of the hydroformylation and deuterioformylation reac-
tions.
The observed initial reaction rate for the hydroformyla-
tion of 1-octene was (17.6ꢁ1.1) a.u.minꢀ1 mmolRhꢀ1,
whereas the observed rate of deuterioformylation was
(14.4ꢁ1.8) a.u.minꢀ1 mmolRhꢀ1. This results in an observed
kinetic 1H/2H isotope effect of 1.22ꢁ0.11. These results
were verified by regular batch autoclave experiments at
608C. The results from these experiments are shown in
Table 1. From the turnover frequencies of hydroformylation
and deuterioformylation, a 1H/2H kinetic isotope effect of
1.18ꢁ0.02 was calculated, in good agreement with the ex-
periments performed in the high-pressure infrared cell.
Chem. Eur. J. 2008, 14, 1843 – 1853
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1845

P. W. N. M. van Leeuwen et al.
Table 1. Results of the hydro- and deuterioformylation of 1-octene.[a]
state involving both a- and b-carbon atoms of 1-octene in
the rate-limiting step of the hydroformylation reaction.
Based on the high regioselectivity of the catalyst towards
the formation of the linear aldehyde product (Table 1) and
the irreversibility of the hydride migration step under the
reaction conditions, there is a high kinetic preference for the
hydride moiety to migrate to the b-carbon atom of the coor-
dinated alkene ligand. Therefore, we would expect a large
kinetic isotope effect on this carbon nucleus and a smaller
kinetic isotope effect on the a-carbon nucleus of the alkene
substrate if the migration step were rate-determining. This
behavior was previously observed by Landis and co-workers
in the polymerisation of 1-hexene by using zirconocene cata-
lysts containing different counterions. The preferential 1,2-
insertion of the coordinated 1-hexene moiety into the grow-
ing polymer chain resulted in the kinetic isotope effect at
the b-carbon position of 1-hexene (1.02), which was larger
than that observed at the a-carbon position (1.01) for all
catalyst systems studied.
Experiment
TOF[b]
l/b
% Isomers[c]
% Nonanal[d]
H2/CO
D2/CO
25.8ꢁ0.2
21.8ꢁ0.4
51ꢁ1
54ꢁ2
3ꢁ1
3ꢁ1
95ꢁ1
95ꢁ1
[a] T=608C, pCO=pH2 =10 bar, [Rh]=1.0 mm, Rh:L:1-octene=
1:10:637. [b] Turnover frequency in (molaldehyde)(molrhodium)ꢀ1 hꢀ1
.
[c] Percentage of octene isomers in all products formed in the reaction.
[d] Percentage nonanal in all products formed in the reaction.
The observed kinetic isotope effect was small, suggesting
a rate-limiting step for the hydroformylation of 1-octene
which does not involve the transfer of the hydride moiety.
However, the magnitude of the observed kinetic isotope
effect greatly depends on the atoms involved in the reaction
and the geometry of the transition state. Whilst for many or-
1
ganic reactions primary H/2H kinetic isotope effects of 5–7
are observed, many examples exist where the observed pri-
mary 1H/2H kinetic isotope effect is unusually small (or
large).[14] Especially in reactions involving non-linear transi-
tion states, primary kinetic isotope effects can be much
smaller than expected. Indeed, previous theoretical studies
on the hydride migration step of the hydroformylation reac-
tion have shown that the geometry of the transition state is
far from linear.[15–20] Furthermore, because the hydride mi-
gration step in the hydroformylation reaction involves the
The observed 12C/13C kinetic isotope effects appear to be
consistent with a rate-determining step involving the sym-
metrical addition of 1-octene to square-planar complex 2. If
alkene coordination to complex 2 proceeds through a sym-
metrical transition state, both the a-carbon and b-carbon
atoms of the alkene are involved in the reaction similarly,
leading to identical kinetic isotope effects on the two carbon
nuclei. It should be noted that theoretical studies on the bi-
molecular alkene coordination step in the hydroformylation
cycle are scarce in comparison to studies on the unimolecu-
lar hydride migration reaction.[19,20] As a consequence, little
is known about the exact geometry of the transition state
for this process. A recent computational study on ethene co-
ordination to model rhodium complexes has suggested that
the geometry of the transition state might be less symmetri-
cal than expected.[20]
ꢀ
breaking of a rhodium hydride bond, which is significantly
ꢀ
different from a carbon hydrogen bond, interpretations
purely based on a single small H/2H kinetic isotope effect
1
can be misleading. A more detailed discussion is provided
below.
12C/13Ckinetic isotope effects : 12C/13C isotope effects on the
a-carbon and b-carbon atoms of 1-octene were determined
at natural abundance by using the NMR methodology pio-
neered by Singleton and co-workers.[21] To this end, two
samples of 1-octene were hydroformylated to high conver-
sion (99.0% and 98.6%) at 608C and 20 bar of H2/CO. The
remaining octenes were subsequently recovered from the re-
action mixture by column chromatography and distillation.
The 12C/13C kinetic isotope effects were determined by quan-
titative 13C{1H} NMR spectroscopy by comparing the isotop-
ic enrichment of the recovered samples against a commer-
cial sample of 1-octene. The changes in isotopic composition
of the a-carbon and b-carbon atoms in the recovered 1-
octene relative to the commercial sample (R/R0) were calcu-
lated by using the signal of the g-carbon atom of 1-octene as
an internal standard. The results are summarised in Table 2.
Within experimental error, the observed 12C/13C kinetic
isotope effects on the a- and b-carbon atoms of 1-octene are
identical. These results suggest a symmetrical transition
It should also be borne in mind that the kinetic isotope ef-
fects determined via this technique (i.e., at natural abun-
dance) are heavily biased towards the rate-determining step
at high substrate conversion. Because the rate of alkene ad-
dition is directly proportional to the alkene concentration, it
is possible that 1-octene addition to complex 2 is not the
slowest reaction step in the catalytic cycle at high substrate
concentration, but that it becomes rate-determining as the
concentration of 1-octene decreases during the reaction. Al-
1
though both the determined H/2H and 12C/13C kinetic iso-
tope effects seem to point towards a rate-determining step
involving the symmetrical addition of 1-octene to complex
2, a more detailed understanding of the transition states of
the two reaction steps is required to unambiguously assign
the rate-determining step in the rhodium–Xantphos-cata-
lysed hydroformylation reaction.
Table 2. Observed 12C/13C kinetic isotope effects on 1-octene.
Theoretical analysis of the key catalytic steps: To under-
stand the observed kinetic isotope effects and unravel the
rate-determining step in the hydroformylation reaction cata-
lysed by rhodium catalysts modified by Xantphos-type li-
gands, we theoretically examined the early stages of the cat-
Experiment Conversion (%)
R/R0
Cb
12C/13C KIE
Ca Cb
Ca
1
2
99.0
98.6
1.10(4) 1.08(3)
1.05(1) 1.053(5) 1.012(3) 1.012(1)
1.020(9) 1.016(6)
1846
www.chemeurj.org
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 1843 – 1853

Rhodium–Xantphos-Catalysed Hydroformylation of 1-Octene
FULL PAPER
alytic cycle: i) CO dissociation; ii) alkene coordination, and;
iii) hydride migration. To this end, hybrid quantum mechan-
ics/molecular mechanics (QM/MM) and pure quantum me-
chanics (QM) calculations were carried out, for the “real-
world” bisphosphane ligand and ethene as a model for al-
kenes. We focused our study on Thixantphos as a ligand
(Scheme 3), which represents an intermediate case of natu-
ral bite angle among the series of Xantphos ligands, and for
which the kinetic data for the CO dissociation and the over-
all reaction rates are available.[12]
Scheme 3. The Thixantphos ligand.
First, we performed a systematic search of the possible
isomers for species 1, 2, 3, and the key transition states at
QM/MM level. One of the main problems in theoretical
characterisation of catalytic cycles is the growing number of
coordination modes and ligand conformations to be consid-
ered. Although the Xantphos-type ligands are rather rigid,
exhibiting pseudo-Cs symmetry, it is possible to define sever-
al ligand conformations and coordination modes (Figure 3).
Two conformations of the non-planar Xanthene array of
fused cycles were previously defined based on their bending
about the central ring.[19] In the conformation designated
proximal (p), this bending places the oxygen atom of the
Xanthene ring closer to the rhodium than in the conforma-
tion designated distal (d). Additionally, we have also consid-
ered the possible rearrangements of phenyl moieties, taking
Figure 3. Different types of ligand conformers and complex isomers for
species 1, 2 and 3. Thixantphos conformation distal (d) and proximal (p);
and equatorial–equatorial (ee) and equatorial–axial (ea) geometrical
ligand arrangements ee1 and ee2, and ea1 and ea2.
ones for penta-coordinate species, whereas for unsaturated
tetra-coordinate species the situation with the oxygen of the
Xanthene ring being closer to the rhodium seems more
stable. The differences between lowest energy species for ee
and ea isomers are less than 2 kcalmolꢀ1. In square-planar
complexes 2, the cis coordination mode of bisphosphane is
lower in energy than the trans one by several kcalmolꢀ1. F i-
nally, the ee1 and ea1 ligand rearrangement are more stable
than their respective ee2 and ea2 geometrical isomers.
After this initial screening at the computationally afforda-
ble QM/MM level, we reoptimised the lowest energy iso-
mers of equatorial–equatorial and equatorial–axial paths by
means of full DFT calculations. Figure 4 summarises the cal-
culated energetic profiles. The values correspond to elec-
tronic energies, which will be used for the discussion unless
otherwise stated. The selection of Thixantphos ligand
system allows us to compare our computed data directly
with the previously reported experimental kinetic data.
Making use of Transition State Theory one can estimate ac-
tivation barriers from experimentally determined rate con-
stants. The rate constant for CO dissociation in the [(Thix-
antphos)RhH(CO)2] complex derived from 13CO exchange
measurements in solution is about 200 hꢀ1,[12] which leads to
an estimated activation barrier of 20.1 kcalmolꢀ1. Moreover,
based on experimental observations that the overall rate of
the hydroformylation of 1-octene is estimated to be two
orders of magnitude slower than the rate of CO dissocia-
the
conformation
of
the
crystal
structure
for
(Nixantphos)Rh(CO)H
AHCTREUNG
cies, and characterising several rotational conformations for
tetracoordinated complex 2. In penta-coordinate trigonal-bi-
pyramidal complexes (e.g., 1 and 3), the bisphosphane li-
gands can coordinate in an equatorial–equatorial (ee) or in
an equatorial–axial (ea) manner. Assuming an axial prefer-
ence of the hydride ligand and an equatorial coordination of
the alkene,[16] it is possible to define two geometrical isomers
for ee coordination by exchanging the relative positions of
hydride and carbonyl axial ligands (ee1 and ee2). Likewise,
for ea alkene isomers the relative positions of ethene and
equatorial carbonyl ligands leads to two different geometri-
cal isomers, ea1 and ea2 (see Figure 3). In the case of the
tetra-coordinate square-planar species 2, the bisphosphane
ligands can have a cis (c) or trans (t) rearrangement. A
more detailed description of the geometries and relative en-
ergies of these isomers is provided in the Supporting Infor-
mation. It is interesting to highlight some general qualitative
trends. For a given species, the Thixantphos conformations
lie within a relatively narrow interval of energies. In general,
the distal isomers seem to be favoured over the proximal
Chem. Eur. J. 2008, 14, 1843 – 1853
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1847

P. W. N. M. van Leeuwen et al.
to the tetra-coordinated trans complex 2t did not yield a
transition state, indicating that the process is barrierless (see
Supporting Information for details). In contrast, a transition
state (TS1ea) connecting the lowest energy isomers of the
equatorial–axial path (2c and 3ea) exists that is
1.2 kcalmolꢀ1 lower in energy than complex 2c. This means
that at our computational level the equatorial–axial path is
energetically favoured over the bis-equatorial one by at
least 3.1 kcalmolꢀ1, which is the energy difference between
cis and trans isomers of CO dissociation products. Neverthe-
less, the ethene coordination leads to an almost isoenergetic
situation of the 3ee and 3ea ethene complexes, which then
could easily interconvert via a Berry pseudorotation rear-
rangement with a small intrinsic barrier.[16] The negative ac-
tivation barrier for ethene coordination through the ea path
can be understood by the formation of a long-distance inter-
action h1-(H) ethene intermediate (3’) that stabilises the un-
saturated complex 2 by an energy of approximately 2 kcal
molꢀ1 (see Figure 4). Introducing free energy contributions,
the relative energy of TS1ea with respect to 2c increases to
+9.0 kcalmolꢀ1, which is a value similar to that from previ-
ous calculations. This increase of association energies in gas-
phase calculations is due to the changing number of parti-
cles. In a more realistic system including solvent molecules,
the entropic effects are expected to be smaller because of
concomitant solvent reorganisation.[22] Thus, gas-phase
Gibbs-free-energy calculations are expected to overestimate
the entropy contributions of bimolecular processes such as
ligand dissociation in solution. In fact, our computed Gibbs
free energy for CO dissociation is 9.6 kcalmolꢀ1, which is
significantly lower than the experimental value of 20.1 kcal
molꢀ1. Note that the latter experimental value is closer to
that computed without entropic contributions (21.4 kcal
molꢀ1), suggesting that for calculations of bimolecular pro-
cesses in vacuum the electronic energy differences are more
reliable than the Gibbs-free-energy differences that do not
take into account solvent rearrangement effects. Moreover,
upon introduction of free-energy corrections, the energy dif-
ference between the resting state 1 and the TS1ea decreases
from 20.2 to 18.6 kcalmolꢀ1. From this analysis of energetics,
two main conclusions can be stated. First, the coordination
of ethene to the catalyst species 2 is barrierless or occurs
with a small energy barrier. Second, the energy required to
reach the transition state for ethene coordination from the
resting state of the catalyst 1 is lower than the energy re-
quired for carbonyl dissociation. Thus, the energy analysis
suggests that the alkene coordination can be excluded as a
candidate for the rate-determining step in the rhodium–
Xantphos-catalysed hydroformylation of ethene.[23]
Figure 4. Potential energy profile (kcalmolꢀ1) for the early catalytic steps
of the ethene hydroformylation catalysed by [(Thixantphos)RhH(CO)2].
Relative energy values are computed at the full DFT level. Solid lines
correspond to the equatorial–equatorial (ee) path, while dashed lines cor-
respond to the equatorial–axial (ea) path.
tion,[12] the overall barrier of hydroformylation should be
around 3 kcalmolꢀ1 higher in energy than the barrier for
CO dissociation. The initial dissociation of CO from the
resting state [(Thixantphos)RhH(CO)2] (1) was found to be
barrierless.[19] Our computed values for this process are 28.7
and 21.4 kcalmolꢀ1 for the ee and ea pathways, respectively
(see Figure 4). The calculations reproduce experimental ob-
servations in that the dissociation of equatorial CO from the
ea species is significantly faster than that from the ee com-
plexes. Moreover, the calculated value for the ea pathway is
in good agreement with the experimentally determined
value.
The next step in the catalytic cycle consists of the coordi-
nation of the alkene substrate to form a penta-coordinate
alkene complex 3. The addition of ethene to 2t and 2c is
predicted to be exothermic by 17.0 and 9.1 kcalmolꢀ1, re-
spectively, leading to two isoenergetic ethene complexes 3ee
and 3ea. Because the rate-determining step could be con-
trolled by the reaction barriers from 2 to 3, we examined
this process in detail. Two theoretical studies on this step in
the hydroformylation reaction have been reported previous-
ly. Gleich and co-workers have combined static and dynamic
quantum-mechanical calculations to study alkene coordina-
tion in model phosphane complexes.[20] Their transition state
searches only converged in the cases with apparently suffi-
cient steric repulsion between the incoming substrate and
the metal complex. They computed energy barriers ranging
from ꢀ1.2 to +1.4 kcalmolꢀ1, increasing to the range from
+8.3 to +10.2 kcalmolꢀ1 after free energy corrections. For
“real-world” Xantphos systems, Landis and co-workers have
computed larger barriers at the QM/MM level, which span
from +4.0 before to +17.1 kcalmolꢀ1 after free energy cor-
rections.[19]
Finally, we have examined the hydride migration step.
During this process, the alkene moiety rotates out of the
equatorial plain of trigonal-bipyramidal complex 3 and sub-
sequently, the hydride is transferred from the metal centre
to the alkene moiety. The corresponding transition states
TS2ee and TS2ea exhibited similar structural features to
those in previous theoretical studies.[15,16,18–20,24] The barrier
for the pathway originating from the ee ethene complex,
Using full DFT calculations, we were unable to locate any
transition state for ethene coordination through the bis-
equatorial pathway. Even a relaxed scan of ethene approach
1848
www.chemeurj.org
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 1843 – 1853

Rhodium–Xantphos-Catalysed Hydroformylation of 1-Octene
FULL PAPER
Table 3. Comparison of the experimentally observed 1H/2H and 12C/13C
kinetic isotope effects (KIE) and the theoretical prediction for alkene co-
ordination and hydride migration overall processes.
3ee!TS2ee, was computed to be 13.5 kcalmolꢀ1, whereas
the barrier for the alternate pathway originating from the ea
ethene complex, 3ea!TS2ea, was computed to be 13.3 kcal
molꢀ1. In both cases, the calculated barriers are much lower
than the estimated barrier for the overall reaction rate. If
we consider the relative energy of the hydride migration
transition state TS2ee with respect to the resting state 1ee,
the overall energy barrier of +25.2 kcalmolꢀ1 becomes
larger than the CO dissociation barrier. Interestingly, assum-
ing a larger extent in error cancellation when we compare
differences between activation barriers, the TS2ee structure
is positioned 3.8 kcalmolꢀ1 above that of the CO dissocia-
tion product 2c, close to the estimated trend (3 kcalmolꢀ1)
obtained from experimental data.[12] These results indicate
that the energy barrier required for taking the catalyst from
its resting state 1 to the transition state for hydride migra-
tion (TS2) controls the overall activity of catalytic cycle.
Note that the direct comparison of the relative energies of
the two transition states TS1 and TS2 is justified as both
structures have the same number of atoms. Therefore the
calculated free-energy corrections for the two structures will
largely cancel. Furthermore, any negative entropy contribu-
tions in the alkene coordination step would lead to further
destabilisation of alkene complex 3 and the subsequent hy-
dride migration transition state TS2 relative to the alkene
coordination transition state TS1.
In fact, a closer inspection to the bibliography suggests
that our findings are in line with previous theoretical studies
on the full catalytic cycle employing model catalysts systems.
Decker and Cundari,[18] and Morokuma and co-workers,[16]
found that the transition state for hydride migration was the
highest-energy species of the complete catalytic cycle. More
recently, Gleich and Hutter[20] examined different model
phosphane-modified catalysts, and observed that when free-
energy corrections were considered, hydride migration was
the highest energy point, whereas when no free-energy cor-
rections were considered the CO dissociation product
became the highest one, but only by 1–4 kcalmolꢀ1 higher
than the hydride migration TS.
1H/2H KIE
12C/13C KIE
Cb
Ca
experiment
1.22ꢁ0.11
1.04
1.52
1.012(3)
1.018
1.010
1.012(1)
1.008
1.015
alkene coordination (1ee!TS1ea)
hydride migration (1ee!TS2ee)
termining transition states TS1ea and TS2ee for alkene coor-
dination and hydride migration, respectively.
As expected, our calculations predict a small secondary
1H/2H KIE of 1.04 for ethene coordination, whereas for hy-
dride migration the value is slightly larger (1.52). Thus, the
comparison of the computed values to the observed KIE
(1.22) does not allow elimination of either of the two pro-
cesses as the rate-determining step in the catalytic cycle.
1
Considering the large H/2H KIE (5–7) observed for many
1
proton-transfer organic reactions, the small H/2H KIE com-
puted for the hydride migration process (1.52) may be sur-
prising. However, one should consider some additional facts.
ꢀ
First, the stretching frequencies of rhodium hydride bonds
are in the order of 2000 cmꢀ1,[25] which is significantly lower
than values of around 3000 cmꢀ1 commonly observed for
ꢀ
carbon hydrogen bonds. For complex 1ee, we calculate a
value of 1987 cmꢀ1. By using the harmonic oscillator approx-
imation, the corresponding rhodium-deuteride vibrational
frequency was estimated to be 1424 cmꢀ1, in close agree-
ment with previously reported Rh–D stretching frequencies
calculated for model rhodium complexes.[12] From the calcu-
1
lated Rh–H and Rh–D stretching frequencies, a H/2H KIE
ꢀ
of 3.4 is predicted at 608C for breaking the rhodium hy-
dride bond, which is considerably smaller than the 1H/2H
ꢀ
KIE of 6.9 predicted for the breaking of a C H bond. Sec-
ꢀ
ondly, the previous approximation is only valid if the Rh H
bond is broken at the transition state, which is not the case
ꢀ
here. On going from 1ee to TS2ee, the Rh H bond length-
ens by only 0.05 Š, indicating an early transition state con-
ꢀ
taining a nearly unperturbed rhodium hydride bond. For
In summary, the energy analysis points out that the pro-
cess to reach the hydride migration TS from the resting
state governs the overall activity in rhodium–Xantphos-cata-
lysed hydroformylation of alkenes, in spite of the experi-
mental kinetic isotope effects that seem to suggest a rate-de-
termining step involving alkene coordination. To investigate
the apparent contradiction between our theoretical analysis
and the experimentally determined kinetic isotope effects
and unambiguously determine the rate-limiting step in this
process, we have also computed the kinetic isotope effects
in the initial steps of the hydroformylation reaction.
the transition state structure TS2ee, we identified a normal
mode of vibration with a frequency of 1891 cmꢀ1 corre-
sponding to the rhodium-hydride stretching vibration. From
this value, and by using the harmonic oscillator approxima-
1
tion, it is also possible to estimate a H/2H KIE associated
with the partial bond breaking at the transition state of 1.1.
This value is consistent with the computed 1H/2H KIE,
which is smaller than that initially expected for hydride mi-
gration. Thus, we conclude that the small 1H/2H KIE ob-
served experimentally for rhodium–Xantphos-catalysed hy-
droformylation of alkenes cannot preclude a rate-determin-
ing step involving hydride migration.
Theoretical prediction of kinetic isotope effects: H/2H and
The nearly identical 12C/13C-KIEs on the a-carbon and b-
carbon atom of the terminal alkene observed experimentally
suggest a symmetrical rate-determining transition state in-
volving both carbon atoms. Closer inspection of the geome-
try of the transition state for ethene coordination TS1ea re-
veals that the incoming ethene ligand attacks the square-
1
12C/13C kinetic isotope effects (KIEs) were investigated by
using full DFT calculations on real-world ligand systems.
Table 3 compares the experimentally observed KIEs with
those computed from free-energy differences between the
lowest ground state in energy 1ee and the possible rate-de-
Chem. Eur. J. 2008, 14, 1843 – 1853
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1849

P. W. N. M. van Leeuwen et al.
planar complex 2c perpendicularly relative to the H-Rh-P
axis, with a H-Rh-C=C dihedral angle of ꢀ90.58. More inter-
estingly, one of the two approaching carbon atoms interacts
more strongly with the rhodium centre than the other, re-
the non-substituted carbon atom. Thus, the nucleophilic in-
teraction directs the attack of the rhodium lone pair in an
orbital of dz2 character to the a-carbon atom, which leads to
a transition state with the a-carbon atom closer to the rhodi-
um centre. Consequently, if alkene coordination were to de-
termine the overall rate of hydroformylation, we would
expect different carbon KIE values for both olefinic carbon
atoms. Clearly, this is not in accordance with our experimen-
tal observations.
ꢀ
sulting in two significantly different Rh C distances of 3.106
and 3.400 Š. The carbon=carbon double bond slips over the
carbonyl-rhodium-phosphane axis, forming a Rh-C-C attack
angle of 90.98 (Figure 5). This asymmetric arrangement of
For hydride migration, on the other hand, the difference
between the computed 12C/13C KIE of the two olefinic
carbon atoms is smaller than for ethene coordination. As-
suming that the whole reaction goes through the major pro-
linear path, the values for a- and b-carbon atoms are 1.010
and 1.015, respectively, which are in good agreement with
the experimentally observed 12C/13C KIE values of 1.012(3)
and 1.012(1), respectively. As we have already discussed, the
transition state structure for hydride migration can be de-
Figure 5. Schematic representation of the main geometrical parameters
(a) and orbital interactions (b) in the transition state for alkene coordina-
tion TS1ea. Distances in Š and angles in deg.
ꢀ
scribed as early with the rhodium hydride bond only slight-
ly perturbed with respect to the reactant. This indicates that
most of the activation energy for hydride migration must be
attributed to the rotation of the alkene from the equatorial
in-plane arrangement in the reactant to a nearly perpendicu-
lar orientation in the transition state. Since this movement
involves both olefinic carbons, similar 12C/13C KIE values
are obtained for the a- and b-carbon atoms of the olefin, in
agreement with experimental observations.
the olefinic moiety at the transition state can be explained
by considering that the rhodium centre reacts as a nucleo-
phile attacking the empty p* orbital of the incoming alkene
moiety. The Khon–Sham orbital energies calculated for the
reactants 2c and ethene agree with previous hypotheses.[26–28]
The HOMO of the tetracoordinated d8 rhodium complex is
a Rh(dz2)-based orbital, which is closer in energy to the
A
empty ethene p* orbital than the ethene p orbital is to the
empty Rh acceptor orbital. Thus, contrary to the classical
description of the incoming ligand as the donor and the
metal as acceptor of electron density, the mechanism for
alkene coordination is best described as proceeding via nu-
cleophilic attack of complex 2c on the C=C bond of the
alkene substrate. Moreover, the interaction of the HOMO
on the metal and the LUMO on the ligand involves orbitals
of different symmetry, and therefore the ethene has to slip
Conclusion
The rate-determining step in the hydroformylation of 1-
octene, catalysed by the rhodium-Xantphos catalyst system,
was determined by a combination of experimentally ob-
served H/2H and 12C/13C kinetic isotope effects and a theo-
1
1
retical approach. Both the small H/2H kinetic isotope effect
of 1.2 observed for the hydride moiety of the rhodium cata-
lyst and the symmetrical 12C/13C kinetic isotope effects ob-
served for the a- and b-carbon atoms of 1-octene suggest
that addition of 1-octene to the square-planar complex 2 is
the rate-limiting step in the hydroformylation of 1-octene
catalysed by this catalyst system. Full DFT calculations on
the catalytic cycle by using a “real-world” ligand system,
however, suggest that the hydride migration is the rate-de-
termining transition state. The calculations quantitatively re-
produce the experimentally determined energy barrier for
carbonyl dissociation in the Thixantphos ligand system
(20.1 kcalmolꢀ1 experimentally versus 21.4 kcalmolꢀ1 theo-
retically). The individual barriers for the other two elemen-
tary steps of the catalytic cycle are consistent with Type I ki-
netics, that is, alkene coordination and hydride migration
show significantly lower values (1 and 13 kcalmolꢀ1, respec-
tively). If one considers the energy differences between the
lowest ground state in energy, the resting state 1, and the
transition states for alkene coordination and hydride migra-
tion, the barriers increase to values close to that determined
for carbonyl dissociation.
to allow the interaction between the occupied Rh(dz2) and
A
the empty p* orbitals, explaining the asymmetry of the tran-
sition state[26] (see Scheme 2). Analogous phenomena have
been previously proposed in processes involving electron-
rich transition-metal-based systems, such as, the alkene ex-
change at d10 palladium[28] and the addition of carbon mon-
oxide to unsaturated tetra-coordinate d8 Ir complexes.[27]
In accordance with geometrical parameters for TS1ea, the
calculated 12C/13C KIE values for the closest carbon to rho-
dium (1.018) is larger than for the other carbon atom
(1.008). Clearly, the use of the simplified ethene substrate
does not allow us to distinguish between a- and b-carbon
atoms. However, it is reasonable to assume that for a more
realistic terminal alkene, the ligand attacks the complex
preferably through one of the two olefinic carbons (a or b),
and in the preferred transition state the alkene interaction
with the metal is asymmetrical. It is known that the alkyl
substituents polarise the p* orbital of the double bond to-
wards the a-carbon atom,[29] exhibiting for propene larger
DFT coefficients of the p-perpendicular atomic orbital in
1850
www.chemeurj.org
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 1843 – 1853

Rhodium–Xantphos-Catalysed Hydroformylation of 1-Octene
FULL PAPER
core potential operator was used to represent the 28 innermost electrons
of the Rh atom, as well as the 10 innermost electrons of the P and S
atoms.[35] The basis set for Rh, P and S atoms was that associated with
the pseudopotential,[35] with a standard double-x LANL2DZ contrac-
tion,[32] and in the case of P was supplemented by a d shell.[36] The 6-
31G(d) basis set was used for the alkene carbons and the carbonyl
ligand,[37] the 6-31G(p) basis set was used for the hydride ligand and
alkene hydrogen atoms,[37] whereas the 6-31G basis set was used for the
other atoms.[37] Geometry optimisations were carried out without any
symmetry restrictions and all stationary points were optimised with ana-
lytical first derivatives. Transition states were characterised by single
imaginary frequency, the normal mode of which corresponded to the ex-
pected motion. Kinetic isotope effects were calculated by using statistical
mechanics employing harmonic vibrational frequencies obtained from
frequency calculations.[32] The initial energy analysis of the possible iso-
mers (see Supporting Information) was performed by means of the more
computationally affordable hybrid quantum mechanics/molecular me-
chanics (QM/MM) calculations, by using the ONIOM method[38] as im-
plemented in the GAUSSIAN 98 series of programs.[32] The QM region
Moreover, whilst the overall barrier for alkene coordina-
tion is lower than that for CO dissociation, the energy re-
quirement to undergo hydride migration from the resting
state of the catalyst is 3.8 kcalmolꢀ1 higher than the barrier
for CO dissociation, which is in good agreement with the
trend estimated from previous experimental data.
1
Computed H/2H and 12C/13C KIE values support the con-
clusions drawn from the energy analysis of the initial steps
in the catalytic cycle and also explain the experimentally de-
termined kinetic isotope effects. Due to the “early nature”
of the TS for hydride migration, containing a nearly unper-
turbed rhodium hydride bond, the computed 1H/2H KIE
ꢀ
values are smaller than expected, and therefore, the small
1H/2H KIE values observed experimentally cannot preclude
a rate-determining step involving hydride migration. On the
other hand, the asymmetric nature of the transition state for
alkene coordination leads to non-symmetric 12C/13C KIE
values for the a-carbon and b-carbon atoms of the alkene
substrate, which are not in accordance with our experimen-
tally determined values. The combined experimental and
theoretical results strongly suggest that the overall process
from the pentacoordinated rhodium hydride resting state
species 1 to hydride migration 3!4 governs the overall ac-
tivity in the rhodium-Xantphos-catalysed hydroformylation
of 1-octene. In this context, the term “rate-determining
step” can be misleading because it implies that a single slow
step in the catalytic cycle is responsible for the overall rate
of reaction. Although the overall rate of hydroformylation is
limited by the rate of hydride migration, both CO dissocia-
tion and alkene coordination contribute to the overall barri-
er. Therefore it is more accurate to say that the rate of hy-
droformylation is determined by a set of reactions ending in
hydride migration. Finally, we would like to note that during
revision of the manuscript Sparta and co-workers have pub-
lished a theoretical study on the activity of rhodium-cata-
lyzed hydroformylation leading to the same conclusion as us
for catalysts modified by electron-donating ligands.[30]
of the catalysts was [RhH(CO)(PH3)2], whereas the substrate ethene was
N
fully in the QM region. The QM level was the same as described above.
Molecular mechanics calculations used the UFF force field.[39]
Hydroformylation and deuterioformylation using high-pressure infrared
spectroscopy: The high-pressure infrared (HP-IR) spectroscopic experi-
ments were preformed in a stainless-steel (SS 316) 50-mL autoclave
equipped with IRTRAN windows (ZnS, transparent up to 700 cmꢀ1
,
10 mm id, optical path length=0.4 mm), a mechanical stirrer, a tempera-
ture controller, a pressure transducer and a separate 50-mL second cham-
ber. In a typical experiment the IR autoclave was filled with a solution of
[Rh(acac)(CO)2] (4 mg, 15 mmol) and Xantphos (93 mg, 160 mmol) in cy-
U
clohexane (10.0 mL), prepared under an inert atmosphere. The autoclave
was purged three times with a 1:1 mixture of H/CO or D2/CO (22 bar)
and was subsequently pressurised to 17 bar. The second chamber was
filled with a solution (5 mL) of 1-octene (0.5 mL, 1.0 mL, 1.5 mL and
2.0 mL) and decane (0.75 mL, 3.88 mmol) in cyclohexane. This chamber
was flushed three times with H2/CO or D2/CO (22 bar) and was subse-
quently pressurised to 22 bar. The autoclave was heated to 808C and was
allowed to stabilise for 1 h. Rapid-scan IR measurements were started
(120 min, 4 scans per spectrum, 10 s between spectra, resolution 2 cmꢀ1
)
immediately followed by the addition of the 1-octene solution to the cat-
alyst solution. Difference spectra were obtained by subtraction of the in-
frared spectrum taken after 1.077 minutes from the other spectra. The in-
crease in peak area of the nonanal aldehyde signal (area 1674–1811 cmꢀ1
over time was analysed by linear regression.
)
Hydroformylation and deuterioformylation of 1-octene: The hydroformy-
lation experiments were performed in a stainless-steel (SS 316) autoclave
(196 mL). The autoclave was stirred mechanically and was equipped with
a separate reservoir, a pressure transducer and a thermocouple. In a typi-
Experimental Section
cal experiment [Rh(acac)(CO)2] (5.2 mg, 20 mmol) and Xantphos
A
(120 mg, 0.2 mmol) were dissolved in toluene (15.0 mL) under inert at-
mosphere. The autoclave was placed under vacuum for 30 min and was
subsequently purged three times using H2/CO or D2/CO (15 bar). The
catalyst solution was introduced into the autoclave by using a syringe.
The autoclave was purged an additional three times and was subsequent-
ly pressurised to 15 bar and heated to 608C. The solution was stirred at
this temperature for 1 h. A mixture of 1-octene (2 mL, 12.74 mmol),
decane (internal standard, 1.0 mL, 5.17 mmol) and toluene (2 mL) was
introduced into the separate reservoir and the reservoir was purged three
times using H2/CO or D2/CO (20.0 bar). The substrate solution was intro-
duced into the autoclave by overpressure and the autoclave was pressur-
ised to a total pressure of 20.0 bar. The hydroformylation reaction was
stopped after a pressure drop of approximately 0.5 bar by using a solu-
tion of tributylphosphite. The autoclave was cooled rapidly by using an
ice-water bath and was depressurised. The conversion and product distri-
bution were determined by GC analysis of the reaction mixture.
Standard syringe techniques were applied for transfer of air-sensitive re-
agents and dry solvents. Commercially available chemicals were used
without further purification, unless stated otherwise. Cyclohexane was
purchased from Rathburn, distilled from over sodium wire and stored
under argon. Toluene was purchased from Rathburn and distilled from
over sodium pieces. Pentane was purchased from Biosolv and distilled
from solution containing sodium/benzophenone/triglyme. 1-Octene was
purchased from Sigma–Aldrich and was purified over neutral aluminium
oxide prior to use. Nonanal was purchased from Sigma–Aldrich. Alumi-
num oxide, 90 active neutral (0.063–0.2 mm), was purchased from Merck.
Silica gel 100 (0.2–0.5 mm) was purchased from Biosolv. Naphthalene
was purchased from Acros chemical company and [Rh(acac)(CO)2] was
A
purchased from Merck. Xantphos was synthesised according to a litera-
ture procedure.[31] NMR spectra were recorded on
500 MHz spectrometer.
a Varian Inova
Computational details: Full quantum mechanics calculations on the Thix-
antphos ligand system were performed with the GAUSSIAN 98 series of
programs[32] within the framework of the density functional theory
(DFT)[33] using the B3LYP functional.[34] A quasi-relativistic effective
12C/13CKinetic isotope effect: Hydroformylation of 1-octene : Hydrofor-
mylation experiments were performed in a stainless-steel (SS 316) auto-
clave (196 mL). The autoclave was equipped with a mechanical stirrer
Chem. Eur. J. 2008, 14, 1843 – 1853
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1851

P. W. N. M. van Leeuwen et al.
and was also equipped with a separate reservoir, a pressure transducer
and a thermocouple. In a typical experiment, [Rh(acac)(CO)2] (5.2 mg,
wen, E. B. Walczuk-Gusciora, N. E. Grimmer, P. C. J. Kamer (Sasol
Technology Proprietary Limited), WO 2005049537, 2005.
[3] W. Ahlers, M. Roeper, P. Hofmann, D. C. M. Warth, R. Paciello
(BASFAktiengesellschaft), WO 2001058589, 2001.
[4] a) E. Billig, A. G. Abatjoglou, D. R. Bryant, R. E. Murray, J. M.
Maher (Union Carbide Corp.), US pat. 4,599,206, 1986; b) E. Billig,
A. G. Abatjoglou, D. R. Bryant (Union Carbide Corp.), US Pat.
4,668,651; US Pat. 4748261, 1987.
[5] D. Evans, J. A. Osborn, G. Wilkinson, J. Chem. Soc. A 1968, 3133.
[6] a) R. Lazzaroni, G. Uccello-Barretta, M. Benetti, Organometallics
1989, 8, 2323; b) G. Uccello-Barretta, R. Lazzaroni, R. Settambolo,
P. Salvadori, J. Organomet. Chem. 1991, 417, 111; c) A. Raffaelli, S.
Pucci, R. Settambolo, G. Uccello-Barretta, R. Lazzaroni, Organome-
tallics 1991, 10, 3892.
[7] C. P. Casey, L. M. Petrovich, J. Am. Chem. Soc. 1995, 117, 6007.
[8] M. Garland, P. Pino, Organometallics 1991, 10, 1693.
[9] a) V. S. Nair, S. P. Mathew, R. V. Chaudhari, J. Organomet. Chem.
1999, 143, 99; b) J. Feng, M. Garland, Organometallics 1999, 18, 417.
[10] P. C. J. Kamer, A. Van Rooy, G. C. Schoemaker, P. W. N. M. Van
Leeuwen, Coord. Chem. Rev. 2004, 248, 2409.
[11] a) L. A. van der Veen, P. C. J. Kamer, P. W. N. M. Van Leeuwen, Or-
ganometallics 1999, 18, 4765; b) L. A. van der Veen, P. C. J. Kamer,
P. W. N. M. Van Leeuwen, Angew. Chem. 1999, 111, 349; Angew.
Chem. Int. Ed. 1999, 38, 336.
[12] L. A. van der Veen, P. H. Keeven, G. C. Schoemaker, J. N. H. Reek,
P. C. J. Kamer, P. W. N. M. Van Leeuwen, M. Lutz, A. L. Spek, Orga-
nometallics 2000, 19, 872.
AHCTREUNG
20 mmol) and Xantphos (120 mg, 210 mmol, 10 equiv) were placed in the
autoclave. The autoclave was closed, evacuated for 30 min and was subse-
quently purged by using a mixture of H2/CO (1:1, 20 bar). Naphthalene
(0.5 g, 3.9 mmol) was dissolved in purified 1-octene (20 mL, 127 mmol)
under inert atmosphere and this mixture was subsequently introduced
into the autoclave. The autoclave was purged three times with a mixture
of H2/CO (1:1, 20 bar) and pressurised to 20 bar. The reaction mixture
was heated to 608C. At regular intervals the autoclave was repressurised
to 20 bar. The pressure was not allowed to decrease below 18 bar. The re-
action was run to high conversion and was stopped by rapidly cooling the
autoclave in an ice/water bath, followed by depressurisation. After open-
ing the autoclave, the reaction mixture was collected and stored under
argon. Conversions were determined by gas chromatography.
Recovery of octenes from hydroformylation reaction mixtures: The re-
maining alkenes in the hydroformylation reaction mixture were isolated
by column chromatography over silica, using pentane as an eluent. The
collected fractions were analysed by gas chromatography. Fractions con-
taining 1-octene were combined and concentrated to 2–3 mL by fraction-
al distillation of the solvent. GC analysis showed no significant amounts
of octenes in the distilled solvent. To remove trace amounts of pentane
still present in the residue, dry CDCl3 (10 mL) was added to the mixture
and the resulting solution was concentrated a second time to 2–3 mL by
careful distillation. For comparison, a commercial sample of 1-octene
(0.3 mL, 1.9 mmol) in nonanal (5 mL) was subjected to the same purifica-
tion procedure.
Determination of the 12C/13Ckinetic isotope effect : Both the commercial
sample and the recovered octene mixtures were analysed by quantitative
13C{1H} NMR spectroscopy. NMR samples were prepared under argon,
placing 1.5 mL of the octene solution in a 5 mm NMR-tube. 13C{1H}
NMR spectra were collected at 125.69 MHz, by using inverse-gated-de-
coupling in order to minimise nuclear Overhauser effects. A spectral
width of 200.00 ppm was used. The T1 relaxation times were determined
for all samples using the inversion-recovery method. Based on these
measurements, a recycle delay of 108 s (at least 6”T1) between calibrated
p/2 pulses was chosen. An acquisition time of 7.00 s was used, collecting
175935 points. Integrations were determined by using a constant region
of ꢁ5 times the peak-width-at-half-height. A zero-order baseline correc-
tion was applied, but in no case was a first-order correction applied. A
total of four spectra were obtained for each sample.
[13] a) G. Yagupski, C. K. Brown, G. Wilkinson, Chem. Commun. 1969,
1244; b) J. Zhang, M. Poliakoff, M. W. George, Organometallics
2003, 22, 1612; c) M. Caporali, P. Frediani, A. Salvini, G. Laurenczy,
Inorg. Chim. Acta 2004, 357, 4537.
[14] F. H. Westheimer, Chem. Rev. 1961, 61, 265.
[15] a) N. Koga, S. Q. Jin, K. Morokuma, J. Am. Chem. Soc. 1988, 110,
3417; b) D. G. Musaev, T. Matsubara, A. M. Mebel, N. Koga, K. Mo-
rokuma, Pure Appl. Chem. 1995, 67, 257; c) W. R. Rocha, W. B. D.
Almeida, Int. J. Quantum Chem. 2000, 78, 42; d) G. Alagona, C.
Ghio, R. Lazzaroni, R. Settambolo, Organometallics 2001, 20, 5394;
e) R. Schmid, W. A. Herrmann, G. Frenking, Organometallics 1997,
16, 701; f) D. Gleich, R. Schmid, W. A. Herrmann, Organometallics
1998, 17, 4828; g) D. Gleich, R. Schmid, W. A. Herrmann, Organo-
metallics 1998, 17, 2141; h) D. Gleich, W. A. Herrmann, Organome-
tallics 1999, 18, 4354; i) J. J. Carbo, F. Maseras, C. Bo, P. W. N. M. van
Leeuwen, J. Am. Chem. Soc. 2001, 123, 7630; j) J. J. Carbó, A.
Lledós, D. Vogt, C. Bo, Chem. Eur. J. 2006, 12, 1457; k) E. Zuidema,
E. Daura-Oller, J. J. Carbó, C. Bo, P. W. N. M. van Leeuwen, Orga-
nometallics 2007, 26, 2234.
Acknowledgements
[16] T. Matsubara, N. Koga, Y. Ding, D. G. Musaev, K. Morokuma, Orga-
nometallics 1997, 16, 1065.
[17] S. A. Decker, T. R. Cundari, J. Organomet. Chem. 2001, 635, 132.
[18] S. A. Decker, T. R. Cundari, New J. Chem. 2002, 26, 129.
[19] C. R. Landis, J. Uddin, J. Chem. Soc. Dalton Trans. 2002, 729.
[20] D. Gleich, J. Hutter, Chem. Eur. J. 2004, 10, 2435.
[21] D. A. Singleton, A. A. Thomas, J. Am. Chem. Soc. 1995, 117, 9357.
[22] S. F. Vyboishchikov, M. Bühl, W. Thiel, Chem. Eur. J. 2002, 8, 3962.
[23] In the calculations, ethene was chosen as a model substrate to de-
crease the number of structural isomers caused by the distinct orien-
tations of the alkene moiety in the metal complexes. For real-life 1-
octene, steric repulsion might destabilise transition state TS1, result-
ing in comparable transition state energies for CO dissociation and
alkene coordination.
[24] S. A. Decker, T. R. Cundari, Organometallics 2001, 20, 2827.
[25] a) A. R. Sanger, J. Mol. Cat. 1977, 3, 221; b) A. R. Sanger, L. R.
Schallig, J. Mol. Cat. 1977, 3, 101.
[26] a) O. Eisenstein, R. Hoffmann, J. Am. Chem. Soc. 1980, 102, 6148;
b) O. Eisenstein, R. Hoffmann, J. Am. Chem. Soc. 1981, 103, 4308.
[27] F. Abu-Hasanayn, K. Krogh-Jespersen, A. S. Goldman, J. Am.
Chem. Soc. 1994, 116, 5979.
This work was supported by the Netherlands Organisation for Scientific
Research (NWO/CW). The authors are grateful for financial support
from the MEC of Spain (CTQ2005-06909-C02-02/BQU, CTQ2005-06909-
C02-01/BQU and Consolider Ingenio 2010 CSD2006-0003), from the
Generalitat de Catalunya (2005SGR00715 and 2005SGR00104), and
from the ICIQ Foundation. The SARA Centre For High Performance
Computing is acknowledged for use of their computational facilities.
[1] C. D. Frohning, C. W. Kohlpainter in Applied Homogeneous Cataly-
sis with Organometallic Compounds,Vol. 1 (Eds.: B. Cornils, W. A.
Herrman), VCH Publishers, New York, NY (USA), 1996.
[2] a) H. Bohnen, J. Herwig, A. Joerg, D. Hoff, S. Surm, P. W. N. M. van
Leeuwen, R. Bronger, O. Stelzer (Celanese Chemicals Europe
GmbH), DE 10225282, 2003; b) H. Bohnen, J. Herwig (Celanese
Chemicals Europe GmbH), WO 2002068371, 2002; c) H. Bohnen, J.
Herwig (Celanese Chemicals Europe GmbH, Germany), WO
2002068369, 2002; d) W. Ahlers, R. Paciello, D. Vogt, P. Hofmann
(BASFAktiengesellschaft), WO 2002083695, 2002; e) W. Ahlers, D.
Wiebelhaus, R. Paciello, M. Bartsch, R. Baumann, D. Vogt, A.
Hewat (BASFAktiengesellschaft), WO 2002022261, 2002; f) P.
Rainer, W. Ahlers, T. Mackewitz, R. Paciello, M. Volland (BASF
Aktiengesellschaft), WO 2005039762, 2005; g) P. W. N. M. van Leeu-
[28] B. V. Popp, J. L. Thorman, C. M. Morales, C. R. Landis, S. S. Stahl, J.
Am. Chem. Soc. 2004, 126, 14832.
1852
www.chemeurj.org
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 1843 – 1853

Rhodium–Xantphos-Catalysed Hydroformylation of 1-Octene
FULL PAPER
[29] a) K. N. Houk, J. Sims, R. E. Duke, Jr., R. W. Strozier, J. K. George,
J. Am. Chem. Soc. 1973, 95, 7287; b) D. P. Fox, P. J. Stang, Y. Ape-
loig, M. Karni, J. Am. Chem. Soc. 1986, 108, 750.
[30] M. Sparta, J. Børve, V. R. Jensen, J. Am. Chem. Soc. 2007, 129, 8487.
[31] M. Kranenburg, Y. E. M. van der Burgt, P. C. J. Kamer, P. W. N. M.
Van Leeuwen, K. Goubitz, J. Fraanje, Organometallics 1995, 14,
3081.
[34] a) C. Lee, W. Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785; b) A. D.
Becke, J. Chem. Phys. 1993, 98, 5648; c) P. J. Stephens, F. J. Devlin,
C. F. Chabalowski, M. J. Frisch,J. Phys. Chem. 1994, 98, 11623.
[35] P. J. Hay, W. R. Wadt, J. Chem. Phys. 1985, 82, 299.
[36] A. Hçllwarth, M. Bçhme, S. Dapprich, A. W. Ehlers, A. Gobbi, V.
Jonas, K. F. Kçhler, R. Stegmann, A. Veldkamp, G. Frenking,Chem.
Phys. Lett. 1993, 208, 237.
[37] a) M. M. Francl, W. J. Pietro, W. J. Hehre, J. S. Binkley, M. S.
Gordon, D. J. Defrees, J. A. Pople, J. Chem. Phys. 1982, 77, 3654;
b) W. J. Hehre, R. Ditchfield, J. A. Pople, J. Chem. Phys. 1972, 56,
2257; c) P. C. Hariharan, J. A. Pople, Theor. Chim. Acta 1973, 28,
213.
[38] a) F. Maseras, K. Morokuma, J. Comput. Chem. 1995, 16, 1170; b) F.
Maseras, Chem. Commun. 2000, 1821; c) S. Humbel, S. Sieber, K.
Morokuma, J. Chem. Phys. 1996, 105, 1959; d) M. Svensson, S.
Humbel, R. D. J. Froese, T. Matsubara, S. Sieber, K. Morokuma, J.
Phys. Chem. 1996, 100, 19357; e) M. Svensson, S. Humbel, K. Moro-
kuma, J. Chem. Phys. 1996, 105, 3654; f) S. Dapprich, I. Komꢂromi,
K. S. Byun, K. Morokuma, M. J. Frisch, J. Mol. Struct. 1999, 462, 1.
[39] A. K. Rappꢃ, C. J. Casewit, K. S. Colwell, W. A. Goddard III, W. M.
Skiff, J. Am. Chem. Soc. 1992, 114, 10024.
[32] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A.
Robb, J. R. Cheeseman, V. G. Zakrzewski, J. J. A. Montgomery,
R. E. Stratmann, J. C. Burant, S. Dapprich, J. M. Millam, A. D. Dan-
iels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi, V. Barone, M.
Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J.
Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui, K. Morokuma, D. K.
Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. Cioslow-
ski, J. V. Ortiz, A. G. Baboul, B. B. Stefanov, G. Liu, A. Liashenko,
P. Piskorz, I. Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T.
Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, C. Gonzalez,
M. Challacombe, P. M. W. Gill, B. G. Johnson, W. Chen, M. W.
Wong, J. L. Andres, M. Head-Gordon, E. S. Replogle, J. A. Pople,
Gaussian 98 (Revision A.7), Gaussian Inc., Pittsburg, PA, 1998.
[33] a) R. G. Parr, W. Yang, Density Functional Theory of Atoms and
Molecules, Oxford University Press, Oxford (UK), 1989, p. 352; b) T.
Ziegler, Chem. Rev. 1991, 91, 651.
Received: May 12, 2007
Revised: October 22, 2007
Published online: December 5, 2007
Chem. Eur. J. 2008, 14, 1843 – 1853
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1853