Limonene-derived phosphines in the cobalt-catalysed hydroformylation of alkenes

Article abstract of DOI:10.1039/b310233e

Cobalt complexes involved in the hydroformylation of alkenes using (4R,S)-4,8-dimethyl-2-octadecyl-2-phosphabicyclo[3.3.1]nonane (LIM-18), which consists of a mixture of two diastereomers, have been studied. Complexation to cobalt has been used to separate or enrich the enantiomers so that spectroscopic parameters can be determined for complexes of the form [Co₂(CO)_8-n(LIM-18)_n] (n = 1 or 2) and [Co(CO)₃(LIM-18)₂][Co(CO)₄] containing the different diastereomers. An X-ray structure of a complex of the form [Co(CO)₃(LIM-18)]₂ shows it to contain the (4R) isomer, but since this was isolated from a mixed solution of diastereomers, this does not definitively identify whether the (4R) or (4S) isomer is the more strongly coordinating isomer. Experimental studies backed up with molecular modelling suggest that steric and electronic effects determine which isomer reacts preferentially in coordination, protonation and methylation reactions. The (4R) isomer is the more strongly coordinating and is metallated preferentially, although the (4S) isomer is the more basic. Variable temperature and high pressure NMR studies on the acyl complex, [Co(C(O)C₅H₉)(CO)₃(LIM-18)], suggest that product elimination may be accompanied by isomerisation to branched acyl species. Modified cobalt catalysed hydroformylation of reactions suggest that the two diastereomers of LIM-18 do not give greatly different rates or selectivities and that a long chain secondary alkyl substituent on the LIM gives lower linear selectivity and a faster rate than LIM-18. LIM-Bu^t gives a selectivity similar to that obtained using LIM-18.

Full text of DOI:10.1039/b310233e

View Article Online / Journal Homepage / Table of Contents for this issue
Limonene-derived phosphines in the cobalt-catalysed
hydroformylation of alkenes
Anastasios Polas,†a James D. E. T. Wilton-Ely,a Alexandra M. Z. Slawin,a Douglas F. Foster,†b
Petrus J. Steynberg,c Michael J. Greenc and David J. Cole-Hamilton*a
a School of Chemistry, University of St. Andrews, St. Andrews, Fife, Scotland, UK KY16 9ST.
E-mail: djc@st-and.ac.uk; Fax: ϩ44-(0)1334-463808; Tel: ϩ44-(0)1334-463805
b Catalyst Evaluation and Optimisation Service, School of Chemistry,
University of St. Andrews, St. Andrews, Fife, Scotland, UK KY16 9ST
c Research and Development, Sasol Technology (Pty) Ltd., P.O. Box 1, Sasolburg, 1974,
South Africa
Received 26th August 2003, Accepted 16th October 2003
First published as an Advance Article on the web 11th November 2003
Cobalt complexes involved in the hydroformylation of alkenes using (4R,S)-4,8-dimethyl-2-octadecyl-2-phospha-
bicyclo[3.3.1]nonane (LIM-18), which consists of a mixture of two diastereomers, have been studied. Complexation
to cobalt has been used to separate or enrich the enantiomers so that spectroscopic parameters can be determined for
complexes of the form [Co2(CO)8Ϫn(LIM-18)n] (n = 1 or 2) and [Co(CO)3(LIM-18)2][Co(CO)4] containing the different
diastereomers. An X-ray structure of a complex of the form [Co(CO)3(LIM-18)]2 shows it to contain the (4R) isomer,
but since this was isolated from a mixed solution of diastereomers, this does not definitively identify whether the (4R)
or (4S) isomer is the more strongly coordinating isomer. Experimental studies backed up with molecular modelling
suggest that steric and electronic effects determine which isomer reacts preferentially in coordination, protonation
and methylation reactions. The (4R) isomer is the more strongly coordinating and is metallated preferentially,
although the (4S) isomer is the more basic. Variable temperature and high pressure NMR studies on the acyl
complex, [Co(C(O)C5H9)(CO)3(LIM-18)], suggest that product elimination may be accompanied by isomerisation to
branched acyl species. Modified cobalt catalysed hydroformylation of reactions suggest that the two diastereomers of
LIM-18 do not give greatly different rates or selectivities and that a long chain secondary alkyl substituent on the
LIM gives lower linear selectivity and a faster rate than LIM-18. LIM-But gives a selectivity similar to that obtained
using LIM-18.
formed instead of aldehydes.3 Since alcohols are generally
Introduction
desired, this is an advantage, but the hydrogenation activity
The hydroformylation of alkenes is one of the great successes
means that some alkene is usually lost through hydrogenation
of industrial homogeneous catalysis. For short chain alkenes,
rhodium catalysts modified with triphenylphosphine are used
to alkane.
The greatest advantage cobalt catalysts have is their increased
selectivity for linear alcohols specifically when using mixtures
because of their high activity under mild conditions and their
good selectivity towards the linear product aldehyde.1,2 How-
of internal and terminal alkenes produced in the Shell Higher
ever, for long chain alkenes, which give long chain aldehydes
Olefin Process developed in the 1960s.4 Existing technology and
for use in the manufacture of plasticisers, soaps and detergents,
infrastructure for hydroformylation under high pressure and
these catalysts have not been used because they decompose at
temperature, combined with the stability of cobalt catalysts
the temperatures required for the separation of the product
towards poisons, allowing for lower grades of feedstock, has
from the catalyst by distillation. A pilot plant employing low
also contributed to the continued use of cobalt catalytic
temperature distillation is currently under commissioning at
systems by petrochemical giants such as Exxon, Shell, CdF
Sasol in South Africa, however.
Chimie, Nissan and BASF for the industrial formation of ‘fatty
alcohols’.
Compound 1 is derived from the addition of a primary phos-
phine across cycloocta-1,5-diene, but two regioisomers can be
formed and these are relatively difficult to separate.5 In this
paper, we report some studies of a cobalt system modified with
a phosphine formally derived in a similar way, but from
(R)-(ϩ)-limonene. Detailed results of the use of this ligand
in cobalt catalysed hydroformylation reactions have been
patented6 and a recent publication reports that the chain length
of the alkyl group on phosphorus has a marked effect on the
activity and selectivity of the hydroformylation reaction.7
Increasing the chain length increases the reaction rate, but
decreases the linear selectivity and extent of alkene hydro-
genation. By using high pressure spectroscopy and molecular
modelling, these changes have been attributed to steric effects
leading to a greater proportion of the cobalt catalyst being
unmodified (no coordination of P ligand) for the longer chain
alkyl phosphines. The unmodified system gives higher rates but
lower selectivities and less competing alkene hydrogenation.7
Because of the potential problems of catalyst decomposition
during the product separation, all commercial plants for long
chain aldehydes and alcohols currently use cobalt based tech-
nology.1 Some systems use [Co2(CO)8] as the catalyst precursor,
in which case the products are aldehydes, the selectivity to the
desired linear aldehyde is low and the reaction conditions are
very harsh (300 bar, 200 ЊC). In other plants, tertiary phos-
phines are used to modify the cobalt based catalysts, with the
result that better linear selectivities can be obtained (up to 10:1
linear:branched (l:b) ratio with ligands such as 1), much
milder conditions are applied (90 bar, 200 ЊC) and alcohols are
† Current address: Sasol Technology (UK) Ltd., Purdie Building,
North Haugh, St. Andrews, Fife, Scotland, UK KY16 9ST.
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 3
D a l t o n T r a n s . , 2 0 0 3 , 4 6 6 9 – 4 6 7 7
4669

View Article Online
Table 1 31P data for cobalt complexes of the two different diastereo-
Results and discussion
a
mers of LIM-18 in CDCl3
Ligand synthesis and separation of the diasteroisomers
Complex
31P{1H}NMR (CDCl3)
The sterically demanding ligand (4-R,S)-4,8-dimethyl-2-octa-
decyl-2-phospha-bicyclo[3.3.1]nonane (LIM-18) 2 (see Scheme
1) is derived from the procedure recently reported by Robertson
et al.8 where phosphine (PH3) is reacted with (R)-(ϩ)-limonene
in the presence of a radical initiator. The formed bicyclic
secondary phosphine 3 is then reacted with 1-octadecene and
the anti-Markovnikov radical addition ensues. The resulting
product is a 55:45 mixture of two diastereomers exhibiting
resonances at δ Ϫ45.2 and Ϫ51.7 ppm in the 31P NMR spectrum.
Separation of the two diastereomers by conventional means
(distillation, recrystallisation and chromatography) proved
both difficult and unsuccessful partly because of oxidation of
the phosphine and partly because the long C18 tail makes the
physical properties of the diastereomers very similar. However,
on reaction of an excess of the phosphine with [Co2(CO)8], one
diastereomer was found to complex preferentially. Surprisingly,
the diastereomer resonating at δ Ϫ45.2 ppm (LIM-18A) was
more strongly coordinating, even though the greater shielding
of the P atom in the other isomer δ Ϫ51.7 (LIM-18B) suggests
it is the more basic diastereomer. This is supported by a proton-
ation experiment in which the LIM-18B isomer reacts preferen-
tially. Molecular modelling suggests that it is the (4R) isomer
which resonates at δ Ϫ45.27 (see below). The preferential co-
ordination of LIM-A can be used to separate the diastereomers.
[Co(CO)3((4R)-LIM-18)2][Co(CO)4]
[Co(CO)3((4S)-LIM-18)2][Co(CO)4]
[Co(CO)3((4R,S)-LIM-18)2][Co(CO)4]
37.8
40.6
37.9 (d, 2JP–P = 65 Hz)
40.5 (d, 2JP–P = 65 Hz)
[Co(CO)3((4R)-LIM-18)]2
[Co(CO)3((4S)-LIM-18)]2
[Co(CO)3((4R,S)-LIM-18)]2
37.9
40.4
37.6 (d, 3JP–P = 13 Hz)
40.4 (d, 3JP–P = 13 Hz)
Co2(CO)7((4R)-LIM-18)
Co2(CO)7((4S)-LIM-18)
(4R)-LIM-18
(4S)-LIM-18
(4R)-LIM-18-oxide
(4S)-LIM-18-oxide
41.4
44.0
Ϫ45.2
Ϫ51.7
45.5
47.4
a The assignments to (4-R) and (4-S) assume that LIM-18A is (4-R) and
LIM-18B is (4-S) (see text). The same assignment was made in ref. 7.
complexes, both phosphines can be the same or different
diastereomers. If they are different, the P atoms are inequiv-
alent and couple to one another.
Attempted determination of the absolute stereochemistry of the
different diastereomers
It has previously been shown8 that the two diastereomers arise
from an indiscriminate H addition to the tertiary carbon radical
ؒ
of 4 following addition of H2P to the exocyclic double bond of
limonene (see Scheme 1) and this is confirmed by the observ-
ation that deprotonation of the primary phosphine, 3, with
BuLi gives the phosphide anion, again as a mixture of two
diastereomers (31P NMR δ Ϫ128.6 and Ϫ142.6 ppm), thus rul-
ing out the phosphorus atom as being the centre of chiral
ambiguity. Subsequent radical additions must occur in a highly
diastereoselective manner as only two out of a possible eight
diastereomers were observed.10 The high steric strain of the
bicyclic phosphine was also observed to effect a stereoselective
mode of attack in nucleophilic substitution reactions. Coupling
the phosphide of this species to a long chain alkyl bromide
again exhibited only two peaks (δ Ϫ36.1, Ϫ42.1 ppm) rather
than the four expected if attack at the phosphorous were not
diastereoselective. As all other atoms on the bicyclic ring are
fixed, it is the orientation of the methyl group at the C4 position
of LIM-18 which determines both the stereoselective mode of
attack and the ligating ability.
Scheme 1 Synthesis of ligand 2 from R-(ϩ)-limonene and PH3.
(i) PH3 ϩinitiator, (ii) 1-octadeceneϩinitiator.
Crystals of [Co(CO)3(LIM-18)]2 were grown from
a
Reaction of LIM-18 with [Co2(CO)8] in light petroleum leads
to the precipitation of an oil, identified by comparison of its
spectroscopic properties with those of the PBu3 analogue9 as
[Co(LIM-18)2(CO)3][Co(CO)4]. By employing a deficiency of
LIM-18 (LIM-18:Co = 2:6) in this reaction, the oil contains
68% [Co(LIM-18A)2(CO)3][Co(CO)4], 29% [Co(LIM-18A)-
(LIM-18B)(CO)3][Co(CO)4] and 3% [Co(LIM-18B)2(CO)3]-
[Co(CO)4]. Heating this oil in toluene converted it to predomin-
antly [Co2(LIM-18A)2(CO)6)]. Alternatively, by using an excess
of LIM-18 (LIM 18:Co = 6:4), removing the oil and eluting
the supernatant liquid through a silica column (100% CH2Cl2 to
remove the cobalt complexes formed and then 100% Et2O to
elute the ligand), a sample of ligand enriched in LIM-18B
(>95%) was isolated.
sample of the oil, [Co(CO)3(LIM-18)2][Co(CO)4] enriched in
LIM-18A, on standing under an atmosphere of nitrogen. 31P
NMR analysis of the resulting crystals embedded in a
waxy solid revealed a 68:29:3 mixture of the diastereomers
with [Co(CO)3(LIM-18A)]2 in excess. A crystal was selected
and found to be suitable for X-ray structure determination
(Figs. 1 and 2, Table 2). Numbering of the ligand follows
that of the phosphinic acid which was reported recently by
Robertson et al.8
The separation of the two diastereomers and their sub-
sequent reactions with [Co2(CO)8] allowed the definitive
assignment of resonances observed in the 31P NMR spectra of
the complexes shown in Table 1. This was essential as up to
twelve resonance peaks of varying intensity within 10 ppm can
be observed in the 31P NMR spectrum of some complexes made
with unresolved LIM-18 due to differing products containing
one or two phosphines. In the case of the bis-phosphine
Fig. 1 X-ray crystal structure and numbering scheme for the core of
[Co2(CO)6(LIM-18)2]. The C18H37 chains have been omitted for clarity.
D a l t o n T r a n s . , 2 0 0 3 , 4 6 6 9 – 4 6 7 7
4670

View Article Online
Table 2 Selected bond lengths (Å) and angles (Њ) for [Co(CO)3(PC28H55)]2
Co(1)–Co(2)
Co(1)–P(2)
Co(1)–C(61)
Co(1)–C(62)
Co(1)–C(63)
P(2)–C(1)
2.6687(12)
2.192(2)
1.776(8)
1.671(8)
1.752(7)
1.865(8)
1.677(11)
1.654(2)
C–O
1.115–1.233
2.201(2)
1.774(7)
1.651(10)
1.718(9)
1.859(7)
1.821(10)
1.885(2)
Co(2)–P(32)
Co(2)–C(64)
Co(2)–C(65)
Co(2)–C(66)
P(32)–C(31)
P(32)–C(33)
P(32)–C(42)
P(2)–C(3)
P(2)–C(12)
P(2)–Co(1)–Co(2)
P(2)–Co(1)–C(61)
P(2)–Co(1)–C(62)
P(2)–Co(1)–C(63)
Co(1)–P(2)–C(1)
Co(1)–P(2)–C(3)
Co(1)–P(2)–C(12)
C(1)–P(2)–C(3)
C(1)–P(2)–C(12)
C(3)–P(2)–C(12)
C(61)–Co(1)–C(62)
C(61)–Co(1)–C(63)
C(62)–Co(1)–C(63)
O–C–Co
177.87(7)
94.7(3)
95.8(3)
P(32)–Co(2)–Co(1)
P(32)–Co(2)–C(64)
P(32)–Co(2)–C(65)
P(32)–Co(2)–C(66)
Co(2)–P(32)–C(31)
Co(2)–P(32)–C(33)
Co(2)–P(32)–C(42)
C(31)–P(32)–C(33)
C(31)–P(32)–C(42)
C(33)–P(32)–C(42)
C(64)–Co(2)–C(65)
C(64)–Co(2)–C(66)
C(65)–Co(2)–C(66)
170.23(8)
96.6(3)
88.6(3)
94.7(2)
96.9(3)
124.9(3)
114.8(4)
115.12(11)
102.5(5)
105.3(3)
87.4(5)
114.2(4)
117.3(4)
126.1(4)
170–177
128.0(3)
109.0(3)
109.99(10)
101.4(4)
100.5(2)
105.9(4)
116.3(4)
114.5(4)
127.8(4)
panied by a waxy solid containing 68:29:3 [Co(CO)3(LIM-
18A)]2 :[Co2(CO)6(LIM-18A)(LIM-18B)]:[Co(CO)3(LIM-
18B)]2. Some crystals were obtained from a different mixture
containing essentially pure LIM-18B, but these did not diffract
sufficiently for a structure solution.
Because of the problem of identifying unambiguously which
diastereomer binds more readily to the cobalt and in order to
try to determine to what extent the preferential binding of
LIM-18A to cobalt depends upon steric or electronic effects, we
extended our molecular modelling studies on the two ligands,
their complexes, [HCo(CO)3(LIM-18)], their protonated forms
and the products of their reactions with methyl triflate, as well
as experimental studies of the protonation and methylation
reactions. One of us has already reported that molecular
modelling studies suggest that it is the (4R) isomer which
binds preferentially in [HCo(CO)3(LIM-18)] and in a variety of
other limonene derived phosphines bearing different alkyl
substituents.7
Fig. 2 Molecular structure and packing diagram for [Co2(CO)6(LIM-
18)2].
Interesting features in the solid state structure of the dimer
complex include alignment of the C18 hydrocarbon chains in the
same plane. The chains interpenetrate to form flat sheets when
packing (see Fig. 2). The structure of the complex is compar-
able with previous reported structures of [Co(CO)3L]2 where L
is a monodentate ligand11,12 although the ligands are not related
by a C2 axis (i.e. the PCoCoP backbone is not linear as is usu-
ally the case11,12). The need for the complex to adopt both a
staggered arrangement and the flat sheet packing of the long
hydrocarbon chains distorts the preferred linear alignment of
the Co(1)–Co(2)–P(32) bond angle to 170Њ (cf. Co(2)–Co(1)–
P(2), 178Њ). Also affected is a more acute C(3)–P(2)–C(12) angle
of 87.4Њ compared to ∼102Њ as observed on the P(32) ligand and
other phosphine complexes. It is also in these ligand bonds that
the greatest variance in atomic distance occurs (P(2)–C(12) is
0.23 Å shorter than P(32)–C(42)). These distortions probably
arise from packing forces in the solid state. There is certainly no
evidence for inequivalent P atoms in the 31P NMR spectrum.
Also, unlike the crystallisation of the phosphinic acid where the
two diastereomeric pairs were incorporated into the crystal
lattice,8 the single crystal is seen to be diastereomerically pure.
The C(5) atom was affixed with R configuration (from (R)-(ϩ)-
limonene), which also fits well with the predicted compu-
tational analysis of the diffraction pattern. All other chiral
atoms were assigned from this point so that the ligand bound in
this complex has the (1-R,4-R,5-R,8-S) configuration whilst the
other stable diastereomer is (1-R,4-S,5-R,8-S). It should be
noted that the R stereochemistry observed for the phosphorus
atom P(2)/P(32) in the complex corresponds to the S configur-
ation at P in the free phosphine. Also, atoms C8/C38 have S
stereochemistry, as in both diastereomers of the phosphinic
acid derived from the LIM skeleton.8
Experimentally, LIM-18A reacts preferentially with Meϩ and
[Co2(CO)8]. 31P NMR studies show that the ratio of equilibrium
constants for LIM-18A and LIM-18B is 1.4 for the reaction
with methyl triflate whilst that for complexation with [Co2-
(CO)8] is 7.0. Protonation of LIM-18A and LIM-18B with
p-toluenesulfonic acid leads to very broad resonances in the
31P NMR spectrum. This broadening effect increases with
the amount of acid added but disappears once all phosphine
has been protonated. This indicates that an exchange process
is operating between protonated and non-protonated forms.
Measuring the spectra at low temperature (0 ЊC) reduced the
broadening dramatically and allowed accurate integration of
the signals. It was observed that the LIM-18B isomer reacts
preferentially with the acid, giving a value of 0.12 for the ratio
of equilibrium constants for LIM-18A and LIM-18B. This
suggests that the P atom in LIM-18B is slightly more basic than
that in LIM-18A (an observation supported by the higher field
chemical shift)‡ and so preference is determined predominantly
by electronic effects. The much greater preference for co-
ordination of the less basic LIM-18A to cobalt suggests that
steric effects dominate in that case. The preference for LIM-
18A in the methylation experiment seems also to be directed
by steric considerations, though the much lower value of 1.4
(compared to 7.0 for Co complexation) for the ratio of equi-
librium constants indicates that steric factors play a lesser role.
As the crystals were grown from an 82:17 mixed LIM-
18A:LIM-18B diastereomeric complex solution, it is not
possible definitively to confirm which of these is present in the
crystalline complex, particularly as the crystal was accom-
‡ The P–H coupling constant in 3 (196 Hz) and the νCO values are
identical for both isomers, so are not helpful in determining relative
basicities of the two diasteriomers.
D a l t o n T r a n s . , 2 0 0 3 , 4 6 6 9 – 4 6 7 7
4671

View Article Online
Table 3 Calculated heats of reaction (4R) or (4S)-LIM-18 with
[HCo(CO)4] and of (4R) or (4S)-LIM-2 with Hϩ (from H3Oϩ)
LIM isomer
Reagent
∆Hreact/kJ molϪ1
(4-R)-LIM-18
(4-S)-LIM-18
(4-R)-LIM-2
(4-S)-LIM-2
(4-R)-LIM-2
(4-S)-LIM-2
[HCo(CO)4]
[HCo(CO)4]
Meϩ
Ϫ3.437
9.377
176.81
175.43
Ϫ300.83
Ϫ290.16
Meϩ
Hϩ
Hϩ
Based on calculated heats of reaction for the complexation
reaction of the (4R) and (4S) LIM-18 diastereomers with
[HCo(CO)4], the preferred isomer is (4R) (Table 3).7 On this
basis, we conclude that the (4R) isomer binds preferentially to
cobalt and hence that the crystal selected for the X-ray structure
determination contains the more strongly coordinating
diastereomer.
Scheme 2 Proposed mechanism of formation of branched acyl
intermediates during high pressure NMR studies.
(1:1, 75 bar). The spectrum, at ambient temperature (Fig. 3b),
shows the appearance of a small amount of the [Co(CO)3(LIM-
18B)]2 resonating at 40.6 ppm, which was probably formed
through oxidative addition of molecular hydrogen followed by
reductive elimination of the aldehyde. A new peak observed at
21.0 ppm is thought to be from a LIM-18B branched acyl com-
plex, 7, rather than the LIM-18B terminal alkyl complex 8, (see
Scheme 2) because (i) the acyl complex is more stable than the
alkyl and (ii) the alkyl complex would be expected to resonate
much further upfield with respect to the acyl complex on the
basis of relative chemical shifts from literature data for acyl/
alkyl complex pairs.1,2 On increasing the reaction temperature
to 60 ЊC, (Fig. 3c) the amount of terminal acyl complex, 5,
decreases, resonances from both diastereomers of [Co(CO)3-
(LIM-18)]2 are observed and the predominant species are
[Co2(CO)7(LIM-18)]. The resonance at 21 ppm (which shifts
downfield with increasing temperature) increases along with a
new species at 20.1 ppm. Assuming the species at 21 ppm is in
fact the LIM-18B branched acyl complex, 7, then this new
species is thought to be the diastereomeric LIM-18A branched
acyl complex, 7. Also observed is a broad peak at ∼26 ppm
attributed to the hydride species, [HCo(CO)3(LIM-18B)].7
Increasing the temperature to 80 ЊC (Fig. 3d), accelerates the
conversion of the terminal acyl species, 5, to [Co2(CO)7(LIM-
18)] and [Co(CO)3(LIM-18)]2. The resonance for the LIM-18B
branched acyl complex, 7, is believed to overlap with that from
the LIM-18A terminal acyl complex, 5, with all three peaks
from the various acyl species being of similar intensity. The
solution was then left to cool to room temperature and allowed
to equilibrate over 90 min. A subsequent spectrum (Fig. 3e)
shows almost complete conversion to [Co2(CO)7(LIM-18)] and
[Co(CO)3(LIM-18)]2. Traces of of hydride (26.9 ppm) and the
acyl complexes, 5 (20.6, 18.6 ppm), are also present along with
the branched acyl species, 7 (21.3, 19.5 ppm). The other small
peaks between 20.7 and 18.6 ppm may possibly be from
branched acyl isomers at the C3 position.
Infrared analysis of the reaction products supports the con-
clusions drawn from the NMR experiments. Prior to the addi-
tion of syngas, the cobalt acyl complex, 5, is clearly visible with
νCO at 2039 and 1969 and a characteristic acyl νC᎐O band of
medium intensity at 1679 cmϪ1. After the addition of syngas
and heating to 80 ЊC, the only stretches observed are those of
the monophosphine dimeric species [Co2(CO)7(LIM-18)] and
[Co(CO)3(LIM-18)]2.
If the species resonating at 21.2 and 19.5 ppm are the (4R,S)-
branched acyl complexes, 6, it implies that, in the presence
of H2/CO and at temperatures below 100 ЊC, the rate of hydro-
genolysis of the terminal acyl complex is slow enough to
allow for the interconversion to the alkyl complex followed by
β-hydride elimination to form the alkene complex, remigration
of H and remigration of the branched alkyl ligand onto CO, as
shown in Scheme 2. It is known that alkene isomerisation
in these complexes occurs at a much faster rate than hydro-
An attempt was also made to correlate the observed experi-
mental basicity differences (LIM-18A and LIM-18B) with
theoretical values. From the data in Table 3 it can be seen that
for the differences between the heats of reaction of the two
diasteromers in protonation and methylation reactions are
small relative to their absolute values, so meaningful conclu-
sions cannot be drawn. The absolute accuracy of DFT quan-
tum calculations is dependent on many factors: the choice of
Hamiltonian, basis set used, the type of reaction and the reac-
tion environment. The absolute energy errors are still large
(∼4–65 kJ molϪ1).13 There are examples where the absolute
energy accuracy (thermochemistry) is entering the < 4 kJ molϪ1
domain.14 Relative energy differences which are used in this
paper are more accurate, however, arguments based on small
energy differences relative to the absolute values (methylation
and protonation reactions, Table 3) should always be treated
cautiously. The energy values reported in Table 3 are total elec-
tronic energies at 0 K only. For the purpose of this theoretical
study, thermodynamic properties (∆G, ∆H and T∆S) of the
said reactions were not calculated. Solvation effects were also
not accounted for in the theoretical study and both of the latter
are expected to impact on the calculated reactions energies
specifically for the alkylation and protonation reations where
charged species are involved. The energy differences observed
for the complexation reactions are more significant and they
suggest that the (4R) isomer coordinates preferentially, so we
have assumed that LIM-18A is the (1-R,4-R,5-R,8-S) isomer.
High pressure NMR studies
In order to obtain spectroscopic parameters for other possible
intermediates that may be observed under catalytic conditions,
HP NMR studies were carried out on the hexanoyl complex,
[Co{C(O)(CH2)4CH3}(CO)3(LIM-18)] (5), which was prepared
from the reaction of an equimolar amount of LIM-18 with a
solution of the acyl(tetracarbonyl)cobalt complex preformed
using known procedures.15,16
31P NMR analysis in toluene under CO (1 bar) at room tem-
perature (Fig. 3a) showed two resonances at 20.6 and 18.5 ppm
which are assigned to the two diastereomers of the acyl species,
5 in Scheme 2. Also present were two sharp peaks at 44.7 and
42.1 ppm, which are assigned to the monophosphine substi-
tuted dimeric species [Co2(CO)7(LIM-18)] (6) which have been
prepared separately from the reaction of [Co(CO)3(LIM-18)]2
with [Co2(CO)8] (see Experimental section and Table 1). The
marginally different chemical shifts arise because of the differ-
ent solvent used for the HPNMR studies. When the reaction
was repeated using a 1:4 Co:L ratio, the [Co(CO)3(LIM-18)]2
dimers were preferentially formed with only a small amount of
the acyl species being observed and no [Co2(CO)7(LIM-18)].
The solution of terminal acyl complex, 5, was transferred to
a high pressure NMR cell, sealed and pressurised with syngas
D a l t o n T r a n s . , 2 0 0 3 , 4 6 6 9 – 4 6 7 7
4672

View Article Online
Fig. 3 31P NMR studies of [Co(hexanoyl)(CO)3(LIM-18)].
formylation, even under reaction conditions (85 bar, 180 ЊC).
Hydroformylation reactions
Although it seems unlikely that this isomerisation involves acyl
intermediates, it is highly probable that if the alkyl group does
migrate from CO to Co, the reformed acyl complex will have
the acyl group on an internal C atom.
Having separated the two diastereomers of LIM-18, it was of
interest to know whether they would differ in terms of activity
or selectivity in hydroformylation reactions. Hydroformylation
D a l t o n T r a n s . , 2 0 0 3 , 4 6 6 9 – 4 6 7 7
4673

View Article Online
Table 4 Cobalt/LIM catalysed hydroformylation of 1-octene at 85 bar H2 :CO (2:1)
Rate
Residual
n-Octane
(%)
Aldehyde
(%)
Unknownb
(%)
Branched
const./
Expt.
Ligand
t/h T/ЊCa alkene (%)
alcoholc (%)
1-Nonanol l:bd
hϪ1
1
2
3
4
LIM-18
LIM-18
LIM-18
LIM-18
6
6
7
6
6
6
174.8
174.5
170.0
172.0
175.5
171.0
1.3
0.6
2.8
1.9
0.9
1.1
7.7
7.2
6.6
6.4
6.7
7.6
0.4
0.4
0.7
0.6
0.3
0.8
10.5
4.2
4.2
4.3
3.4
2.8
18.5
24.9
19.6
21.1
27.9
31.2
61.7
63.0
66.1
65.8
60.8
56.6
3.3 (5.5) 0.84
2.5 (4.2) 0.85
3.4 (5.5) 0.59
3.1 (5.1) 0.69
2.2 (3.7) 0.86
1.8 (3.0) 1.06
5 (MD)e LIM-18B
6
LIM-
CH(C10H21)2
LIM-
7
7
173.0
0.5
7.3
0.5
3.6
31.4
56.7
1.8 (3.0) 1.05
CH(C10H21)2
LIM-But
LIM-But
PBu3
8
9
10
11
6
7
171.0
170.5
3.4
3.3
8.3
8.3
11.7
16.5
1.6
1.2
2.9
0.4
4.2
4.7
8.1
20.1
19.6
6.4
62.5
62.9
42.3
55.5
3.1 (5.2) Ϫf
3.2 (5.3) Ϫf
6.6 (9.7) 0.15
5.4 (8.6) 0.54
10 170.0 28.6
190.5 4.3
PBu3
7
13.0
10.3
a Temperature is 0.5 ЊC. b There is 1 major unknown product. The GCMS of this product is virtually identical to that of 1-nonanol but it is not 2-, 3-,
4- or 5-nonanol. However, it does appear to be some isomeric form of C9H20O. c Sum of 2-methyloctanol ϩ 2-ethylheptanol ϩ 2-propylhexanol.
d Ratio: 1-nonanol/sum of (2-methyloctanol, 2-ethylheptanol and 2-propylhexanol). Figure in brackets: 1-nonanol/2-methyloctanol. e Major
diastereomer of (4R)-LIM-18. f Consistent, but complex, kinetics of gas uptake.
of a 50% solution of 1-octene in decane was carried out under
CO–H2 (85 bar) at 170 ЊC using Co:P of 1:4. The major prod-
ucts were 1-nonanol and 2-methyloctanol. Minor products
included 2-ethylheptanol and 2-propylhexanol from hydro-
formylation of isomerised octenes, the hydrogenation product,
octane, (up to 8%) and trace amounts of aldehyde. In all
experiments an unknown isomeric form of C9H22O was
detected but not identified.
of steric effects is further demonstrated by a comparison with
reactions carried out using PBu3 as a ligand (Expts. 10 and 11).
The lower reaction rate and higher l:b ratio, as well as the
increase in alkane formation confirm that less unliganded
cobalt is present in this system, despite the anticipated poorer
electron donating properties of this small ligand.
A mixture of the two diastereomers, LIM-18A and B (55:44)
was used as ligand in experiments 1–4 (Table 4). Under the
reaction conditions, most of the cobalt is complexed by LIM-
18A. Averaging the results produces a l:b ratio of ∼3.1 with a
rate constant of 0.74 hϪ1. When the reaction was repeated using
a sample enriched in LIM-18B, (>96%, Expt. 5) the rate was
not greatly altered (0.86 hϪ1), whilst the l:b ratio was reduced
somewhat (2.2). These results suggest that there are no great
differences between the two diastereomers in promoting select-
ivity, but that the weaker coordinating power of LIM-18B
means that more cobalt complex containing no bound
phosphine is present. It has been shown that when larger con-
centrations of unliganded cobalt complex are present, these
reactions lead to higher rates and lower l:b ratios.7
Conclusion
Using the preferential coordination of one diastereomer of
LIM-18 to cobalt, we have succeeded in synthesising complexes
enriched in one or other of the diastereomers and hence in
assigning their spectra. On the basis of modelling and preferen-
tial reactivity, we conclude that the (4R) isomer, despite being
slightly less basic, binds more strongly to cobalt as steric con-
straints dominate. A complex containing the (4R) isomer has
been characterised crystallographically. The enriched LIM-B
(believed to be (4S)), shows similar activity and selectivity to
the mixture of isomers in hydroformylation reactions. The
important acyl intermediate reductively eliminates aldehyde
under CO/H2, but also appears to undergo competitive isomer-
isation of the alkyl chain. Bulky secondary alkyl groups on the
P atom of the LIM skeleton give catalytic results indicative of
less coordinated cobalt being present, but But on the P atom
appears to give slightly better selectivity than LIM-18, per-
haps suggesting that the primary C18 chain is bulkier than the
tertiary-butyl group.
When the primary C-18 alkyl chain was replaced by a
secondary CH(C10H21)2 alkyl chain (Expts. 6 and 7), the rate
was increased (1.06 hϪ1) and the selectivity reduced (∼1.8), as
expected for the much bulkier and presumably more weakly
coordinating phosphine.
Interestingly, hydroformylation using a bulky tertiary butyl
modified ligand (Expts. 8 and 9) demonstrated both a good l:b
ratio of ∼3.2 and an efficient rate, but also an increased rate
of alkene hydrogenation. These results suggest that less un-
liganded cobalt complex is present in this system, perhaps
because the But group, despite its tertiary centre attached to P,
is sterically less demanding than LIM-18 with its primary
centre, but much longer chain. Using the X-ray crystallographic
data, we have measured the Tolman cone angle for the two
ligands in [Co2(CO)6(LIM-18A)2], as well as the two ligands in a
ruthenium complex of LIM-18A. The four ligands have cone
angles of 140, 126, 139 and 126Њ. We have then replaced the
C18H37 chains in each ligand by an idealised But group and
remeasured the Tolman cone angles as 124, 116, 124 and 116Њ
respectively. The important observation is that in each case, the
cone angle for the But derivatised ligand is some 10–16Њ smaller
than that for Lim-18A itself confirming that the But derivatised
ligand is less sterically demanding. Although the gas uptake
when using LIM-But was observed to be similar to that of
experiments 6 and 7, the kinetics were not first order and hence
a true rate constant could not be determined. The importance
Experimental
Microanalyses were performed by the University of St.
Andrews Microanalytical Service. Infrared spectra were
obtained using a Nicolet Protégé 460 FT spectrometer with CsI
optics interfaced to a personal computer via the OMNIC oper-
ating system. 1H, 13C and 31P NMR spectra were recorded on a
Bruker AM 300 or a Varian Gemini 300 spectrometer with
broadband proton decoupling for 13C and 31P nuclei. 1H and 13
C
NMR spectra were referenced internally to deuterated solvents,
which were referenced relative to TMS at δ = 0. 31P NMR
spectra were referenced externally to phosphoric acid 85%
H3PO4. Coupling constants are given in Hertz.
Manipulations were carried out under dry oxygen free nitro-
gen or argon using Schlenk techniques or a drybox. All solvents
were freshly distilled and dried; petroleum ether (boiling range,
40–60 ЊC), tetrahydrofuran and diethylether were distilled over
sodium diphenylketyl, toluene over sodium, dichloromethane
and cyclohexane were distilled over calcium hydride. All gases
D a l t o n T r a n s . , 2 0 0 3 , 4 6 6 9 – 4 6 7 7
4674

View Article Online
were purchased from BOC and used without further
purification.
Enrichment of [Co(CO)3(LIM-18A)2][Co(CO)4]
Under a carbon monoxide atmosphere, [Co2(CO)8] (0.50 g, 1.46
mmol) was dissolved in light petroleum (40 mL) and cooled to
0 ЊC. With rapid stirring, a solution of LIM-18 (3.70 g, 8.77
mmol) dissolved in light petroleum (20 mL) was added. On
addition of the phosphine, carbon monoxide gas was liberated.
Stirring was continued for a further 30 min at 0 ЊC followed by
standing to allow the salt to separate. The organic layer was
decanted via cannula and the resulting deep red viscous salt
was washed with cold pentane (20 mL) three times and dried
in vacuo. 31P NMR analysis confirmed enrichment of the
[Co(CO)3(LIM-18A)2][Co(CO)4] salt with the major peak
(69%) at δ 37.78 ppm. IR (neat) ν(CO): 2062 (s), 2005 (vs), 1986
[Co2(CO)8] was purchased from Aldrich and recrystallised
from pentane before use. The ligands 4,8-dimethyl-2-phospha-
bicyclo[3.3.1]nonane and 4,8-dimethyl-2-octadecyl-2-phospha-
bicyclo[3.3.1]nonane were procured from Cytech Canada Inc.
and used without further purification.
2-(1-Decyl-undecyl)-4,8-dimethyl-2-phospha-bicyclo[3.3.1]
nonane [LIM-CH(C10H21)2]
An ethereal solution of decylmagnesium bromide formed by the
Grignard reaction of 1-bromodecane (10.0 g, 45.27 mmol) and
magnesium turnings (2.2 g), was filtered into a solution of undeca-
nal (7.70 g, 45.29 mmol) at 0 ЊC with stirring. Disappearance of
the aldehyde and formation of the secondary alcohol heneicosan-
11-ol was monitored by thin layer chromatography, staining with
potassium permanganate solution. On completion of the reac-
tion, the mixture was quenched with water then extracted, dried
(MgSO4) and recrystallised with dichloromethane. Regioselective
bromination of the isolated secondary alcohol was achieved
using an in situ preparation of PPh3Br2 in DMF. The dry alcohol
(1.00 g, 3.20 mmol) was stirred with PPh3 (0.92 g, 3.51 mmol) in
dry dimethylformamide (20 mL) under an argon atmosphere.
Bromine was then added dropwise over a 15 min period while the
flask was kept below 55 ЊC. The addition was stopped when two
drops persisted in giving an orange colour to the solution.
Removal of the solvent followed by recrystallization of the crude
product with dichloromethane afforded 11-bromoheneicosane in
sufficient purity for use in the next stage of the reaction. The
reaction of the secondary halo-alkane (0.106 g, 0.28 mmol) with
lithium 4,8-dimethyl-2-phosphido-bicyclo[3.3.1]nonane (Li[PC10-
H18]) (0.050 g, 0.28 mmol) (formed in situ from 4,8-dimethyl-2-
phospha-bicyclo[3.3.1]nonane (HPC10H18) and an equimolar
amount of BunLi in THF) was conducted at 0 ЊC and the mixture
was then left to stir overnight. The solvent was removed in vacuo
and the crude product redissolved in petroleum ether (30 mL).
Degassed water (40 mL) was added, the layers separated and the
organic layer dried using MgSO4. Concentration of the product
followed by Kugelrohr distillation (125 ЊC, 0.1 mm Hg) to remove
impurities yielded the product [LIM-CH(C10H21)2] as an air sensi-
tive viscous oil 31P NMR (CDCl3, 121.5 MHz) δ Ϫ36.1, Ϫ42.1
ppm. It was used without further purification.
(vs), 1948 (s), 1889 (vs) cmϪ1 Found: C, 65.44, H, 10.06%;
.
C63H110Co2O7P2 (Mw = 1159.38) requires C, 65.27, H, 9.56%.
On standing, this oil produced a few crystals, one of which
was used for X-ray analysis and shown to be [Co(CO)3((4R)-
LIM-18)]2.
Note: Subsequent reactions to form the Co/LIM-18 species
are identical using enriched LIM-18B or enriched [Co(CO)3-
(LIM-18A)2][Co(CO)4] and the complexes have equivalent
infrared stretching frequencies, hence only the reaction for
the LIM-18A isomer is detailed. The 31P NMR spectra of the
different complexes are collected in Table 1.
[Co(CO)3(LIM-18A)2]2
Under an argon atmosphere, [Co(CO)3(LIM-18A)2][Co(CO)4]
(2.05 g, 1.81 mmol) was dissolved in toluene (40 mL) and
heated to 90 ЊC for 2 h to effect decarbonylation of the ionic
salt. The solution was then cooled to room temperature and the
toluene solvent removed in vacuo to give a deep red viscous oil.
IR (hexane) ν(CO): 2030 (w), 1968 (s), 1949 (vs), 1924 (w), 1898
(sh, w) cmϪ1. Found: C, 65.20, H, 10.20%; C62H110Co2O6P2 (Mw
= 1131.37) requires C, 65.82, H, 9.80%. Attempts at obtaining
spectra at low temperature were frustrated by precipitation of
the compound.
[Co2(CO)7(LIM-18A)]
[Co2(CO)8] (0.62 g, 1.81 mmol) was dissolved in petroleum
ether at room temperature and added to a carbon monoxide
purged solution of the disubstituted dimer [Co(CO)3(LIM-
18A)]2 (2.05 g, 1.81 mmol) in petroleum ether (20 mL). Carbon
monoxide was then bubbled through the reaction mixture
which was monitored by infrared spectroscopy for the dis-
appearance of the stretch at 1949 cmϪ1 and the subsequent
appearance of stretches for the substituted monophosphine
dimer at 2077 cmϪ1 and 1992 cmϪ1. IR (hexane) ν(CO): 2077 (s),
2021 (s), 1992 (vs), 1953 cmϪ1 (m).
2-(t-butyl)-4,8-dimethyl-2-phospha-bicyclo[3.3.1]nonane [LIM-
But]
[LIM-But] was similarly prepared by the metathetical reaction
of tBuCl (2.59 g, 0.35 mmol) with lithium 4,8-dimethyl-2-phos-
phido-bicyclo[3.3.1]nonane, Li[PC10H18] (0.32 mmol). The
crude product was isolated as above and purified by Kugelrohr
distillation (85 ЊC, 0.1 mmHg) yielding the product LIM-But
which was used without further purification. 31P NMR (CDCl3,
121.5 MHz) δ Ϫ36.5, Ϫ42.5 ppm.
[HCo(CO)3(LIM-18)]
LIM-18 (20 mg, 0.047 mmol) dissolved in deuteriated benzene
(1 mL) was added to a stirred solution of [Co2(CO)8] (81 mg,
0.237 mmol) in d6-benzene (1 mL). The reaction mixture was
then transferred to a 40 mL stainless steel autoclave which was
then charged with syngas (H2 :CO, 1:1) to a pressure of 66 bar.
The autoclave was heated to 170 ЊC for 2 h with stirring and
then quickly cooled to room temperature using iced water. The
contents were immediately transferred under argon and charac-
terised by NMR and infrared spectroscopy. 31P NMR (CDCl3):
HCo(CO)3(LIM-18A) δ = 26.9, HCo(CO)3(LIM-18B) δ = 28.5
ppm. IR (C6D6) ν(CO): 2077 (s), 2021 (s), 1992 (vs), 1953 cmϪ1
(m).
Isolation of LIM-18B
Under a carbon monoxide atmosphere, [Co2(CO)8] (2.00 g, 5.85
mmol) was dissolved in light petroleum (40 mL) and cooled to
0 ЊC. With rapid stirring, a solution of LIM-18 (7.42 g, 17.58
mmol) dissolved in light petroleum (20 mL) was then added. On
addition of the phosphine, carbon monoxide gas was liberated.
Stirring was continued for a further 30 mins at 0 ЊC followed by
standing to allow the salt to separate. The supernatant liquid
was then transferred via cannula and reduced in vacuo. The
resulting red oil was dissolved in a minimum of dichloro-
methane and loaded onto a silica column under argon. Elution
with 100% dichloromethane removed a fast flowing red/brown
band close to the solvent front (dimeric cobalt species). Sub-
sequent elution with 100% diethyl ether and removal of solvent
liberated LIM-18B in 96% purity. 31P NMR: δ Ϫ51.8 ppm.
[CH3(CH2)4COCo(CO)3(LIM-18)]
The preparation of the acyl complex was similar to those found
in literature.10,11 Hexanoyl chloride (0.39 g, 2.92 mmol) was
added dropwise to an ethereal solution of Na[Co(CO)4]
D a l t o n T r a n s . , 2 0 0 3 , 4 6 6 9 – 4 6 7 7
4675

View Article Online
(formed in situ from Co2(CO)8 (0.50 g, 1.46 mmol) and 5%
NaHg amalgam stirred vigorously in THF) at 0 ЊC and left
to stir for 1–2 h until the characteristic tetracarbonyl band at
1890 cmϪ1 disappeared. The suspension was filtered and the
filtrate stirred with an equimolar amount of LIM-18 (1.23 g,
2.92 mmol) at room temperature for two hours. The solvent
was evaporated under reduced pressure to yield a red oil, which
was readily soluble in organic solvents. 31P NMR (d8-toluene):
[CH3(CH2)4COCo(CO)3(LIM-18A)] δ = 20.6, [CH3(CH2)4-
COCo(CO)3(LIM-18B)] δ = 18.5 ppm. IR (hexane) ν(CO): 2039
(m), 1969 (s), 1947 (s), 1679 (m) cmϪ1
These proportions give 1000 ppm w/w of Co in solution with a
1:4 Co:P ratio. Oct-1-ene:decane 50/50 by volume.
High pressure NMR studies
The complex [CH3(CH2)4COCo(CO)3(LIM-18)] (100 mg) was
dissolved in toluene containing d8-toluene and transferred
to a 10 mm sapphire NMR tube fitted with a pressure head.
The tube was pressurised with a mixture of CO/H2 1:1 (75 bar).
The 31P NMR spectrum was recorded at room temperature,
60 ЊC and finally at 80 ЊC at which temperature it was left to
equilibrate for 15 min. The sample was cooled back to room
temperature and a final spectrum was recorded after 90 min.
Measurements of equilibrium constant ratios for reactions of
LIM-18A and LIM-18B with Me؉, H؉ and [Co2(CO)8]
X-ray crystal and molecular structure of [Co2(CO)6((1-R,4-R,
5-R,8-S )-LIM)2]§
Quantities (ca. 100 mg) of LIM-18 were weighed out accurately
into Schlenk tubes in a glove box and then moved to a fume
cupboard. A 1:4 mixture of d6-acetone/acetone (1 mL) was
added and aliquots of CF3SO3Me corresponding to approx-
imately 20, 40, 60 and 80% were added under nitrogen. After
stirring for 30 min, the solutions were transferred to NMR
tubes under nitrogen. The 31P NMR spectra of the solutions
were recorded and the ratio of equilibrium constants (KA/KB)
was obtained for each solution from integration of the spectra,
using the following equation:
X-ray diffraction measurements were made with graphite-
monochromated Mo-Kα X-radiation (λ = 0.71073 Å) using a
Bruker SMART diffractometer, intensity data were collected
using 0.3Њ width ω steps accumulating area detector frames
spanning a hemisphere of reciprocal space (data were inte-
grated using the SAINT program). Data were corrected for
Lorentz, polarisation and long-term intensity fluctuations.
Absorption effects were corrected on the basis of multiple
equivalent reflections or by semi-empirical methods. Structures
were solved by direct methods and refined by full-matrix least-
squares against F 2 (SHELXTL). All hydrogen atoms were
assigned isotropic displacement parameters and were con-
strained to idealised geometries. The alkyl chains were refined
with fixed geometry. The alkyl chains and carbonyls were
refined isotropically. Refinements converged to residuals given
in Table 4. All calculations were made with SHELXTL.17
CCDC reference number 218353.
KA/KB = (AMeϩ)(B)/(BMeϩ)(A)
where A, B, AMeϩ and BMeϩ are the integrals for the
signals from LIM-18A, LIM-18B and their methylated forms
respectively. The use of this equation is based on the assump-
tion that the receptivities of the isomer pairs LIM-18A and
LIM-18B and LIM-18AMeϩ and LIM-18BMeϩ are the same,
but does not assume that the receptivities of other com-
ponents (e.g. LIM-18A and LIM-18AMeϩ) are necessarily the
same.
See http://www.rsc.org/suppdata/dt/b3/b310233e/ for crystal-
lographic data in CIF or other electronic format.
Similar procedures were used in the corresponding measure-
ments with [Co2(CO)8] (1:4 d6-benzene/toluene) and p-toluene-
sulfonic acid (in 1:4 d6-acetone/acetone), except that those for
the latter were carried out at 0 ЊC. The equation used for the
protonation calculation is analagous to that above (with AHϩ
rather than AMeϩ etc.). That used for the cobalt complexation
experiment is given below:
Molecular modelling
All geometry optimizations were performed with the DMol3
Density Functional Theory (DFT) code18–20 as implemented in
the MaterialsStudio (Version 2.1.5) program package of
Accelrys Inc.21 on SGI and Compaq Alpha workstations.
DMol3 uses numerical orbitals for the basis functions as
opposed to analytical functions (i.e. Gaussian type orbitals). In
this study the double numerical basis set (DNP) containing a
p-polarization function for H and d-polarization functions for
other atoms were used. The DNP basis set was chosen because
it is equivalent in quality and size to the Gaussian 6-31G**
split-valence double-zeta plus polarization basis set. The latter
is generally accepted as the standard basis set in Hartree Fock
(HF) quantum calculations. The generalized gradient approx-
imation (GGA) functional by Perdew and Wang (PW91)22 was
used for all geometry optimisations. The convergence criteria
for these optimisations consisted of threshold values of 2 ×
10Ϫ5 Ha, 0.00189 Ha ÅϪ1 and 0.00529 Å for energy, gradient
and displacement convergence, respectively, while a self consist-
ent field (SCF) density convergence threshold value of 1 × 10Ϫ6
was specified. All calculated reaction energies reported herein
are total electronic energies at 0 K of the respective geometries
after optimisation.
KA/KB = (CoA)(B)2/(CoB)(A)2
where A, B, CoA and CoB are the integrals for the signals
from LIM-18A, LIM-18B, and the complexes [Co(CO)3(LIM-
18A)]2 and [Co(CO)3(LIM-18B)]2, respectively.
Catalytic reactions
Kinetic studies were performed using a 30 mL reaction vessel
fitted with a substrate injection facility, overhead mechanical
stirrer, thermocouple, pressure gauge and a high pressure syn-
thesis gas reservoir or ballast vessel. In a typical reaction, a
clear, colourless pre-prepared decane solution of LIM-18 (1.25
mL, 0.253 mmol) was added to the autoclave followed by a
clear, dark red pre-prepared decane solution of [Co2(CO)8]
(1.25 mL, 0.032 mmol) under a CO/H2 (1:2) flush. The com-
bined solution was flushed with 3 cycles of ca. 20 bar of CO/H2
(1:2) while being stirred. After flushing was completed, the
autoclave was pressurised to ca. 60 bar, sealed and heated to ca.
170 ЊC. Once the temperature had stabilised and the catalyst
had formed (approximately 1 h), 1-octene (2.5 mL) was
injected. The pressure was increased to 85 bar and maintained
constant by feeding gas from the ballast vessel through a mass
flow controller. The pressure drop in the ballast vessel was elec-
tronically monitored and plotted. The reaction was run for 6 h
in total, at which point the gas uptake had virtually ceased.
In the case of LIM-18 and [HCo(CO)3(LIM-18)], an anneal-
ing dynamics protocol, involving 200 steps of dynamics, a time
step of 0.0005 ps, an initial temperature of 200 K, a midcycle
§ Crystal data for Co2(CO)6(LIM-18)2: Co2C62H110O6P2, M = 1131.35,
orthorhombic, P212121, a = 8.8894(3), b = 12.6610(3) c = 60.1359(3) Å,
V = 6768.2(3) Å3 Z = 4, µ(Mo-Kα) = 0.58 mmϪ1 T = 293 K, SMART
diffractometer, of 29519 reflections measured 9604 were independent
(Rint = 0.0763) to give R = 0.105 for 4474 observed reflections,
[I > 2σ(I )], Flack parameter = 0.10(5).
D a l t o n T r a n s . , 2 0 0 3 , 4 6 6 9 – 4 6 7 7
4676

View Article Online
3 J. L. van Winkle, S. Lorenzo, R. C. Morris and R. F. Mason, US
Patent, 1969, 3420898.
4 D. Vogt, in Applied Homogeneous Catalysis with Organometallic
Compounds, ed. B. Cornils and W. A. Herrmann, VCH, Weinheim,
1996, p. 245.
5 J. H. Downing, V. Gee and P. G. Pringle, Chem. Commun., 1997,
1527.
temperature of 2000 K and a temperature increment of 200 K,
was followed to yield the most stable conformers. The dynamics
were performed with the anneal molecular dynamics function-
ality incorporated in the Accelrys Cerius2 (Version 4.6)
program21 suite, using the universal force field (UFF)23–25 The
trigonal bipyrimidal arrangement of the Cobalt metal centre
was constrained during the MD runs due to the lack of
appropriate force field parameters for Co and the said
geometry.7
After each MD run the ten most stable conformers were sub-
jected to a full geometry optimisation with the Semi Empirical
(SE) PM3 method26 as incorporated in the Spartan ’02 for Unix
program package available from Wavefunction27 The con-
former with lowest SE/PM3 predicted total electronic energy
was used for consequent DFT geometry optimisation as
described above. This procedure was followed to ensure that
global minima structures for both ligand and complex struc-
tures were obtained. For the purpose of this comparative study
only the global energy minimum structures are important and
although the mentioned protocol cannot guarantee global
energy minima, it was deemed comprehensive enough to
ensure energy values in very close proximity to the required
global energy minima.
6 J. P. Steynberg, K. Govender and P. J. Steynberg, World Patent, WO,
2002, 14248.
7 C. Crause, L. Bennie, L. Damoense, C. L. Dwyer, C. Grove,
N. Grimmer, W. J. van Rensburg, M. M. Kirk, K. M. Mokheseng,
S. Otto and P. J. Steynberg, Dalton Trans., 2003, 2036.
8 A. Robertson, C. Bradaric, C. S. Frampton, J. McNulty and
A. Capretta, Tetrahedron Lett., 2001, 42, 2609.
9 K. W. Kramarz, R. J. Klingler, D. E. Fremgen and J. W. Rathke,
Catal. Today, 1999, 49, 339.
10 Although 16 diastereomers are possible when there are 4 stereogenic
centres, the use of (R)-(ϩ)-limonene reduces this number to
eight.
11 R. F. Bryan and A. R. Manning, J. Chem. Soc., Chem. Commun.,
1968, 1316.
12 J. A. Ibers, J. Organomet. Chem, 1968, 14, 423.
13 J. Andzelm, K. Glassford, N. Govind, G. Fitzgerald and K. Stark,
ACS, 2002, Orlando; A. Scheiner, J. Baker and J. Andzelm,
J. Comput. Chem., 1997, 18, 775.
14 F. Axe and J. Andzelm, J. Am. Chem. Soc., 1999, 121, 5396.
15 R. F. Heck and D. S. Breslow, J. Am. Chem. Soc., 1962, 84,
2499.
16 E. Lindner and M. Zipper, Chem. Ber., 1974, 107, 1444.
17 G. M. Sheldrick, SHELXTL, Bruker AXS, 1999.
18 B. Delley, J. Chem. Phys., 2000, 113, 7756 and references therein.
19 B. Delley, J. Phys. Chem., 1996, 100, 6107.
20 B. Delley, J. Chem. Phys., 1990, 92, 508.
Acknowledgements
We would like to thank Sasol Technology Ltd. for the generous
provision of funding for postdoctoral fellowships (to A. P. and
J. D. E. T. W.-E.) and materials.
21 http://www.accelrys.com.
22 J. P. Perdew, Phys. Rev. B, 1986, 33, 8822.
23 A. K. Rappe, K. S. Colwell and C. J. Casewit, Inorg. Chem., 1993, 32,
3438.
24 A. K. Rappe, C. J. Casewit, K. S. Colwell, W. A. Goddard and
W. M. Skiff, J. Am. Chem. Soc., 1992, 114, 10024.
25 C. J. Casewit, K. S. Colwell and A. K. Rappe, J. Am. Chem. Soc.,
1992, 114, 10035.
References
1 C. D. Frohning and C. W. Kohlpaintner, in Applied Homogeneous
Catalysis with Organometallic Compounds, ed. B. Cornils and
W. A. Herrmann, VCH, Weinheim, 1996, vol. 1, p. 27.
2 P. N. W. M. Van Leeuwen and C. Claver, Rhodium Catalysed
Hydroformylation, Kluwer, 2000.
26 J. J. P. Stewart, J. Comput. Chem., 1989, 10, 209.
27 http://www.wavefun.com.
D a l t o n T r a n s . , 2 0 0 3 , 4 6 6 9 – 4 6 7 7
4677